article
stringlengths 3
714k
| abstract
stringlengths 0
152k
| article_id
stringlengths 9
15
|
|---|---|---|
Introduction
quantum memory can store information in superposition states of a collection of two - level systems. optical ion trap by laser cooling has been prepared to construct quantum logic gates @xcite. in those systems, negative role played by quantum decoherence @xcite is quite significant. randomization of the quantum states produced by entanglement with environmental modes is inevitable in case of storage or processing of non - orthogonal states and environmental interaction allows leakage of some information to the environment @xcite. since it is practically impossible to disentangle the system from the environment, our main efforts are focussed on minimizing decoherence. in this attempt of decoherence minimization, zeno dynamics plays a very significant role @xcite. quantum zeno effect @xcite is depicted as the complete freezing of the decay dynamics due to frequent measurement. it has been shown previously that very frequent measurement of excited states can suppress the decoherence @xcite. in our understanding decoherence and zeno effect has got intrinsic reciprocal relationship between them. the argument behind this statement is as follows : whenever any disturbance in the form of measurement dominates the time evolution of the state of the system, the system is forced to evolve in a subspace of the total hilbert space @xcite. this subspace is called zeno subspace ". nonselective measurement causes the appearance of these subspaces. facchi et.al @xcite have shown that frequent nonselective measurement splits the total hilbert space into invariant quantum zeno subspaces, between which probability leakage is not possible. but probability is conserved within each subspace. so each of the subspace can be considered as an reduced isolated system. if the system undergoes very strong environmental interaction, due to extreme decoherence, these isolated subspaces may not be sustainable. so we can infer that the zeno effect characterized by a certain time scale (zeno time), gives a kind of lower limit to decoherence, below which the process of decoherence will be uncontrollable. the relation between these two phenomena is reciprocal in the sense that within the zeno subspace, due to it s isolated nature, it precludes environment induced decoherence. exploiting this relation, we will formulate the procedure to compare the respective time scales and come up with a certain transitional temperature, below which asymptotic minimization of state decoherence is possible. + the master equation for the density operator in position representation of a certain quantum system can be given as @xcite [1.1] = --(-)-(x - x)^2 where the first term on the right hand side is the usual commutator term of the von neumann equation. the second term represents dissipation with @xmath0 as the relaxation rate. the third and last term represent the fluctuations leading to random brownian effects. this term being proportional to @xmath1, though has little effect on the diagonal peaks of the density matrix, but affects the off - diagonal peaks considerably and causes them to decay. hence the effect of this last term leads to the destruction of quantum coherence. from equation ([1.1]) we can easily get that the decay rate of the off - diagonal peaks of the density matrix [1.2] = -(x - x)^2 = -_dec ^-1where [1.3] _ dec= is the time scale on which the quantum coherence disappears and is defined as decoherence time. from the solution of equation ([1.2]), one can easily get [1.4] (x, x,t)=(x, x,0) (-t/_dec) decoherence visibly supresses the interference between macroscopically different quantum states, which is precisely the very property that distinguishes quantum mechanics from it s classical counterpart from observational perspective. here we will consider tunneling in a bistable potential as a model system to develop the expression for decoherence time. as a physically realistic example we will consider a system of laser cooled trapped ion @xcite, where decoherence appears in the dynamics of hyperfine states. comparison between decoherence time and zeno time for this specific case will lead us to find the transitional temperature over which decoherence will dominate the whole process. +
Decoherence-dwell time ratio: condition to control decoherence
let us first concentrate on the calculation of the relaxation rate @xmath2 in presence of dissipative interaction. in a recent paper @xcite we have estimated the weak value of dwell time for a dissipative spin - half system using the same formalism. the approach that has been used here, was originally developed by caldirola and montaldi @xcite introducing a discrete time parameter (@xmath3) incorporating the properties of environment. the schrdinger difference equation in presence of environment induced dissipation is given by [2.01] h_i|=i it has been shown @xcite that this equation has retarded nature and so naturally implies the dissipative character of it s solution. the discrete time parameter (@xmath3) appears as some sort of relaxation time, incorporating the environment induced dissipation. to supplement this difference equation, we will show further that the time parameter (@xmath3) can be expressed as a function of the energy eigen - values of the quantum states. now as a consequence of the retarded nature of eqn ([2.01]), we can see that the ground state will also decay. so to stabilize the ground state, the schrdinger difference equation is scaled as [2.1] (h_i - h_0)|=i where @xmath4 and @xmath5 are the hamiltonian for i - th and ground state respectively. @xmath5 is introduced in the equation to stabilize the ground state @xcite. we expand @xmath6 in taylor series to get [2.2] (h_i - h_0)|=i setting the trial solution as @xmath7 and solving for @xmath8, we get [2.3] = (1+i(e_i - e_0)/) where @xmath9 and @xmath10 are the eigenvalues for the corresponding hamiltonians. expanding the logarithm upto third order, we find that the time evolution takes the form [2.4] so from ([2.4]) we find the decay rate as [2.5] = setting the final hamiltonian as @xmath11, we also find the unknown time parameter as @xcite [2.6] = so using the value of the time parameter @xmath3 in equation ([2.5]), we find the decay constant [2.7] = this is the decay constant for a quantum system decaying from initial to final energy eigenstate denoted by @xmath4 and @xmath11 respectively. we will substitute this decay (relaxation) constant in the expression of decoherence time given by equation ([1.3]). the interaction parameter was given by @xmath3, which was substituted by a function of the initial and final hamiltonians considering the time evolution dynamics. but to calculate the spatial shift (@xmath12), we have to consider the spatial dynamics of the system. here we will focus our attention on an asymmetric double well potential approximated as a two - state system with considering only the ground states of the wells separated by an asymmetry energy @xmath13. we construct our model on the demonstration of a quantum logic gate prepared by a trapped ion laser cooled to the zero point energy @xcite. in this particular case, the target qubit is spanned by two @xmath14 hyperfine ground states (@xmath15 states) of a single @xmath16 ion separated by @xmath17 ghz. we set these two energy levels as the ground states of the two wells of the double well potential separated by the asymmetry energy @xmath18. the control qubit @xmath19 is spanned by the first two states of trapped ion (@xmath20), which can be identified by the first two states of each well approximated as harmonic oscillators. these two states are separated by @xmath21 mhz.so the basic four eigenstates are given by @xmath22. let us now consider a quartic potential of the form @xcite [2.8] v(x)=m^2 x^2 where @xmath23 and @xmath24 are dimensionless constants. for the particular case of double well, @xmath25 and @xmath26. and @xmath26,width=264,height=188] the two potential minima can be found at [2.9] x_0=0 x_2=these two minima are separated by the barrier maxima situated at [2.10] x_1=the potential can be expressed in terms of dimensionless variable @xmath27 as [2.11] v()=m^2a^2 ^2[^2-a+b] now the dimensionless form of the hamiltonian is given by [2.12] k = where @xmath28 is the actual hamiltonian and the parameter @xmath29. here, the wells are approximated as harmonic oscillators. so we expand the potential given by equation ([2.11]) around the first minima @xmath30 to find [2.13] v()=^2b^2 the normalized ground state wave function for this approximated potential can be set as [2.14] () = () ^1/2where @xmath31. + let us consider the transition from left well to the right well. so here the initial position of the particle is @xmath32. so @xmath33. now [2.15] ll < x^2>=a^2<^2>= a^2 _ -^ ^2e^-^2d + = therefore the average of the square of the spatial shift [2.16] x^2=<x^2>= by using ([2.7]) and ([2.16]) in equation ([1.3]), we deduce the decoherence time as [2.17] _ dec= () now the asymmetry energy (@xmath13) can be expressed as [2.18] = v(x_0)-v(x_2) for a double well potential (@xmath34), from equation ([2.18])we get [2.19] = = where the width of the well is given by [2.20] w = x_2-x_0= therefore the decoherence time of equation ([2.17]) can be expressed as [2.21] _ dec = we will estimate the numerical value of decoherence time after deriving a certain relation between the decoherence time and zeno time under some approximation. + we have already mentioned that, quantum zeno effect is the slow down of quantum to classical transition due to frequent measurement. by definition this effect is something that slows down the process of decoherence. it is possible to control quantum to classical transition by frequent energy measurement. it has been shown @xcite that as a result of extremely frequent measurement, the system - reservoir coupling is eliminated and thus decoherence can be halted. zeno time is the time scale within which the quantum states are frozen, ie the decay is halted. + in previous works @xcite we have derived the weak value of dwell time for interacting (via dissipation) systems, which is given by [3.1] ll _ w^d= + () where @xmath35 is the measurement time. now in explaining hartman effect " @xcite, dwell time can be interpreted as the lifetime of the decaying state in the barrier region @xcite. if the interval between consecutive measurements (@xmath36) is significantly smaller than the zeno time (@xmath37), then the dynamics of the decay slows down or even asymptotically halted @xcite. given the condition [3.2] ^m^z we see that the lifetime (@xmath38) of the decaying system, the measurement time (@xmath36) and the zeno time obey the relation [3.3] ^z if we consider the interpretation of dwell time as the lifetime of the decaying state, then we can set @xmath39. now under the assumption [3.4] ^m using ([3.1]) and ([3.3]) we can get the zeno time @xcite as [3.5] ^z= referring to the equation ([3.5]), we find that the approximation ([3.4]) is nothing but [3.6] ^m ^z which is similar to the approximation ([3.2]). this condition can be fulfilled, if the initial and final energy states (with eigenvalues @xmath9 and @xmath40 respectively) of the decaying system are closely spaced. now comparing equation ([2.21]) and ([3.5]) we get [3.7] = the preservation of quantum coherence leads us to conclude that zeno time represents a certain lower limit of decoherence time, beyond which the system loses it s quantumness ". the reason behind this statement lies in the definition of zeno time itself. if the measurement is frequent enough that @xmath41, we observe the asymptotic halting of state decay, hence preserves the coherence. but if the decoherence time is shorter than the zeno time, the state will decay even within that estimated zeno interval. as a result we will not be able to preserve the quantum coherence even by frequent measurement. from equation ([3.7]) we find that imposition of this lower limit to the decoherence time leads us to a certain transitional temperature [3.8] t_tran= above this temperature the coherence of the system ca nt be preserved even by zeno dynamics. we have calculated this transitional temperature for our model system of trapped be atom. since the periodicity of the optical lattice is generally in the micrometer range and @xmath42 is about @xmath43 joules, we get that the transitional temperature is almost 200 microkelvin for be atom. hence above this temperature, decoherence will dominate the whole scenario. this is certainly an achievable temperature in case of raman cooling. in case of raman cooling for optical ion trap @xcite, the cooling temperature of na atom is found to be around 1 microkelvin temperature, which is 0.42 times of the photon recoil temperature @xmath44. since the recoil temperature is inversely proportional to the mass of the atom (@xmath45), for a lighter atom like be, even lower cooling temperature can be achieved. it can not be concluded that below the transitional temperature there will not be any loss of coherence. but at least we can predict from our calculation that decoherence should not dominate the scenario. as the decoherence time becomes larger than zeno time, quantumness " of the system can be preserved within the zeno interval. + let us now calculate the decoherence time for a cooling temperature of 5 microkelvin. consider the transition @xmath46. in this case the zeno time will be @xmath47 @xcite. subsequently, the decoherence time is found to be [3.9] _ dec^0^+0 ^ -= substituting the corresponding values of the parameters we get that the decoherence time is about 7 nanoseconds, whereas the zeno time is found to be 0.17 nanoseconds. similarly the timescales for other transitions can also be calculated.
Conclusion
robust quantum memories are essential to realize the potential advances in quantum computation. optical ion trap can be realized as a quantum storage device. but it is also essential to protect the information, which can be lost due to environmental * _ interaction _ * in the form of decoherence. so it is very important to control the decohering effect in order to build an effective ion trap quantum computer. in this work, we have dealt with the question that whether and under what condition environment induced decoherence can be minimized. as we have discussed that in our understanding, the intrinsic relation between decoherence and quantum zeno effect can be exploited in this aspect. frequent nonselective measurement forces the system to evolve in the reduced zeno subspaces, which can be considered as some quasi - isolated " system. if the zeno effect (characterized by it s corresponding timescale) is strong enough, so that the reduced subspaces remains quasi - isolated even under the influence of environmental interaction, effect of decoherence can be controlled. based on this theoretical understanding, we have calculated a certain transitional temperature, by comparing the decoherence and zeno timescales. it is clear from the above analysis that below this transitional temperature we can increase the decoherence time by controlling the parameters (@xmath48). hence we can minimize the decohering effect asymptotically, though it can never be eliminated completely. 99 c.monroe et.al. ; phys.rev.lett. * 75 *, 4714(1995). j.i.cirac and p. zoller ; phys.rev.lett. * 74 *, 4091(1995). w.h. zurek ; progress in mathematcal physics ; * 48 * ; 1 (2007). charles h. bennett ; physics today * 48 *, 10, 24 (1995) l. viola and s. lloyd ; phys. a * 58 *, 2733 (1998). franson, b.c. jacobs and t.b. pittman ; phys. a * 70 *, 062302 (2004). b.mishra and e.c.g.sudarshan ; j.math.phys. * 18 *, 756(1977). c.b.chiu, e.c.g.sudarshan and b.mishra ; phys.rev.d * 16 *, 520(1977). s.tasaki et.al. ; international journal of quantum chemistry * 98 *, 160(2004). p. facchi and saverio pascazio ; j. phys.. jpn. * 72 *, 30 (2003). s.bhattacharya and s.roy ; phys.rev.a * 85*,062119(2012). p. caldirola and m. montaldi ; nuovo cimento soc. b * 53 *, 291 (1979). mohsen razavy ; _ quantum theory of tunneling_(2003),world scientific publishing, p 255. p. facchi et.at ; phys. a * 71 *, 022302 (2005). s.bhattacharya and s.roy ; arxiv:1209.0279v2. t. e. hartman ; j. appl. * 33 *, 3427(1962). h. g. winful ; new j. phys. * 8 *, 101 (2006). h. g. winful ; phys. rep. * 436 *, 1 (2006). h. g. winful ; phys.. lett. * 91 *, 260401 (2003). p. facchi and s. pascazio ; j. phys. a * 41 *, 493001 (2008). h. j. lee et.al ; phys.rev.lett. * 76 *, 2658 (1996). | decoherence time has been calculated for an optical ion trap of be atoms in a bistable potential model.
comparison has been made between decoherence time and zeno time for double well potential as a special case.
zeno time is considered as a lower limit of decoherence time for sustainable quantum coherence.
equality of the respective timescales provides a certain transitional temperature, below which decoherence can be asymptotically minimized. *
pacs numbers : * 03.65.xp, 03.65.yz + * keywords : * dissipative quantum system, decoherence, zeno time. | 1301.1127 |
Introduction
a widely studied model for growth of porous films is ballistic deposition (bd) @xcite, in which the particles incide perperdicularly to the substrate and aggregate at the first contact with the deposit @xcite. bd was originally proposed to describe sedimentary rock formation @xcite and was extended to model thin film growth and related systems by considering other aggregation mechanisms, non - colimated particle flux, or polydispersivity of particle size @xcite. most works on the ballistic - like models address the scaling features of the outer surface of the deposits, particularly for the connections with kardar - parisi - zhang (kpz) @xcite roughening. some works also connect the surface growth dynamics and the bulk properties of the porous deposits @xcite. this is an essential step for proposing models of porous materials, which have a large variety of technological applications, frequently in the form of thin films @xcite. some ballistic - like models are in a class of competitive growth models in which uncorrelated deposition (ud) is obtained for a certain value of a parameter and, near that value, a crossovers in kinetic roughening is observed @xcite. in the simplest model, particle aggregation follows the bd (ud) rule with probability @xmath7 (@xmath8). it was already studied numerically @xcite and with scaling approaches @xcite. for small @xmath7, there is an enhancement of characteristic times of the correlated kinetics (@xmath9) by a factor @xmath10 and of the outer surface roughness by a factor @xmath11 ; for a recent discussion of this topic, see ref. these features extend to other ballistic - like models with crossovers to ud @xcite and are related to the lateral aggregation. in case of surface relaxation after aggregation, the exponents in those relations are larger, corresponding to longer crossovers for small @xmath7 @xcite. a renewed interest in these competitive models was recently observed, with a focus on the limitations of scaling relations or with an emphasis on the properties of porous media. @xcite discussed the deviations from the dominant scaling of surface roughness at low @xmath7, which is essential for a quantitative characterization of surface properties in a variety of growth conditions. the effect of relaxation after collision of incident and aggregated particles was considered in ref. @xcite, also with a focus on surface properties. refs. @xcite considered the effect of a stickness parameter on the aggregation of the incident particles, which may attach to neighboring particles located below the outer surface. simulations in @xmath1 dimensions produced deposits with porosity ranging from very small values to approximately @xmath12 and suggested non - kpz behavior in one of the models @xcite. the first aim of this paper is to study surface and bulk properties of the model proposed in ref. @xcite combining a systematic analysis of simulation data and a scaling approach for small values of the stickness parameter. from the extrapolation of saturation roughness and relaxation times, we show that the model has kpz exponents in @xmath1 dimensions. comparison of roughness distributions confirms kpz scaling in @xmath2 dimensions, thus ruling out the proposal of non - kpz exponents. in the limit of small stickness parameter @xmath0, the scaling approach shows that the crossover time and the roughness scale as @xmath4 and @xmath5, respectively, for all substrate dimensions. these results show a shortened crossover when compared to all previously studied models with an ud component @xcite, which is a consequence of subsurface aggregation. the same approach predicts porosity and pore height scaling as @xmath6 and @xmath5, respectively. these predictions will be confirmed numerically in @xmath1 and @xmath2 dimensions. the approach can be extended to the model introduced in ref. @xcite, with the same crossover exponents due to the similar subsurface aggregation conditions. the rest of this work is organized as follows. in sec. [model], we present the sticky particle deposition model. in sec. [roughness], we analyze the surface roughness scaling of simulated deposits in @xmath1 dimensions. in sec. [scaling], we present a scaling approach that relates surface properties to the stickness parameter, and confirm those predictions with numerical simulations. in sec. [porosity], we extend the scaling approach to relate the porosity and the average pore height with the stickness parameter, again with support from numerical simulations. in sec. [spd3d], we show that the main results extend to @xmath2 dimensions. in sec. [conclusion], we summarize our results and conclusions.
Sticky particle deposition model
in all models discussed in this paper, square particles of size @xmath13 are sequentially released on randomly chosen columns of a one - dimensional discretized substrate of lateral size @xmath14 and move in a direction perpendicular to the substrate. here, @xmath15 is the number of columns. the time interval for deposition of one layer of atoms (@xmath15 atoms) is @xmath16. thus, at time @xmath17, the number of deposited layers is @xmath18. the model proposed in ref. @xcite is hereafter called sticky particle deposition (spd). in each site of the trajectory of the incident particle, it interacts with particles in nearest neighbor (nn) sites at the same layer (same height above the deposit) and particles in next neareast neighbor (nnn) sites at the layer immediately below it. this interaction is represented by a probabilistic rule of aggregation at its current position. the probability of aggregation to each neighbor is @xmath19 where @xmath0 is the stickness parameter, @xmath20 is the distance between the centers of the particles (@xmath21 for nn, @xmath22 for nnn), and @xmath23 is an exponent related to the nature of the interaction. in ref. @xcite, the cases @xmath24 and @xmath25 were respectively called coulomb - type and van der walls - type interactions, with most results being presented for the former. here we will restrict the analysis to the case @xmath24, in which aggregation to nn and nnn have probabilities @xmath0 and @xmath26, respectively. fig. [modelspd] helps to understand the aggregation rules of the spd model and the differences from other ballistic - like models. we first recall the rules of bd and of the next nearest neighbor bd (bdnnn) model @xcite. in bd, aggregation occurs at the first contact with a nn occupied site : particle a at position 2, particle b at position 6 in fig. [modelspd]. in bdnnn, aggregation occurs at the first contact with a nnn occupied site : position 1 for particle a and position 5 for particle b in fig. [modelspd]. in both cases, the incident particle interacts only with the particles at the top of each column, which are highlighted in [modelspd]. the set of top particles is called the outer surface of the deposit. in the spd model, particle a may aggregate at lattice sites marked with circles labeled 1 to 4, and particle b may aggregate at lattice sites marked with circles labeled 5 to 7. first, consider the trajectory of particle a. in position 1, two aggregation trials are executed due to the interaction with two nnn occupied sites ; the probability of aggregation in each trial is @xmath26. if it does not aggregate there, it moves to the position labeled 2, in which three aggregation trials are executed : two for interactions with the nn at the same height (probability @xmath0 for each one) and one for interaction with the nnn in the layer below, at the left (probability @xmath26). if the particle does not aggregate at position 2, then it moves to position 3 and may aggregate there with probability @xmath0 due to the interaction with the nn at the left. if aggregation does not occur in position 3, the incident particle will permanently aggregate at position 4, which is the top of the incidence column. relaxation to neighboring columns is not allowed. now we consider the trajectory of particle b. in position 5, two aggregation trials are executed, each one with probability @xmath26 (due to interactions with two occupied nnn sites). if the particle does not aggregate there, it moves to position 6, in which three aggregation trials are executed : two for the interactions with the lateral nn (probability @xmath0 for each trial) and one for the interaction with the nnn in the layer below (probability @xmath26). if no aggregation trial is successful at position 6, the particle moves to position 7 and aggregates there. in contrast to other ballistic - like models (e. g. bd and bdnnn), the spd model allows subsurface aggregation. in fig. [modelspd], position 3 is an example of subsurface position : it is not allowed in bd, nor in bdnnn, nor in any model of solid - on - solid deposition (which prescribe aggregation at the top of each column). in all cases, note that the interaction of an incident particle with an aggregated one is possible in two steps : the first one when they are nnn (larger distance), the second one when they are nn (smaller distance). it represents two possibilities of aggregation in the ingoing part of the trajectory of the incident particle. if the aggregation trials are not accepted, then the incident particle moves to a lower position. in this situation, this particle is in an outgoing trajectory respectively to those aggregated particles. for this reason, no aggregation trial is executed with a nnn aggregated particle in the layer above the current position of the incident particle. for instance, when particle a is at position 3 (third layer of the deposit), we do not execute aggregation trials with the black nnn sites at the fourth layer in fig. [modelspd]. the spd model resembles the model introduced in ref. @xcite and the slippery bd model (sbd) proposed in ref. @xcite, both studied in three - dimensional deposits (the latter with line seeds perpendicular to a flat inactive surface). most of our simulation work is in @xmath1 dimensions, similarly to ref. @xcite, but in sec. [spd3d] we show that the main results are also valid in three - dimensional samples. for simplicity, in the following sections we consider unit values of the lattice constant and of the time of deposition of a layer : @xmath27, @xmath28.
Kinetic roughening
the outer surface roughness is defined as @xmath29 } ^{1/2 }, \label{defw}\]] where @xmath30 is the height of the top particle of each column, the overbars indicate spatial averages, and the angular brackets indicate configurational averages. in systems with normal (in opposition to anomalous) scaling, the roughness follows family - vicsek (fv) scaling @xcite as @xmath31 where @xmath32 is the roughness exponent, @xmath33 is a relaxation time, and @xmath34 is a scaling function. in long times (@xmath35), @xmath36, so that @xmath37 saturates as @xmath38 where @xmath39 is a model - dependent constant. the saturation time @xmath33 scales as @xmath40 where @xmath41 is the dynamic exponent and @xmath42 is another model - dependent constant. the roughness for @xmath43 scales as @xmath44 with @xmath45 and another model - dependent constant @xmath46. [figroughness]a shows the surface roughness evolution of the spd model for three values of @xmath0 in @xmath47. (red solid curve), @xmath48 (green dashed curve), and @xmath49 (blue dotted curve). the dashed line has slope @xmath50 of kpz scaling. (b) saturation roughness as a function of the lattice size for @xmath51 (red squares) and @xmath52 (green triangles). the dashed line has slope @xmath53 of kpz scaling., width=264] for short times, there is a crossover from an initial regime of rapid roughness increase to a second regime in which it increases slower. for small @xmath0, the first regime is mainly of ud and the slope of the @xmath54 plot is near @xmath53. for @xmath55, lateral aggregation is frequent, thus the roughness at short times is larger than that of ud (e. g. @xmath51 in fig. [figroughness]a). it is difficult to find a pure ud regime in this case and to estimate the crossover time with accuracy. after this transient, the growth regime begins, with apparent power law scaling of @xmath37 [eq. ([defbeta])]. it is difficult to distinguish the different curves for small @xmath0 in fig. [figroughness]a ; this will be explained by the scaling approach of sec. [scaling]. the slope of those curves are near @xmath50, suggesting kpz scaling. at long times, there is an increase in the saturation roughness as @xmath0 decreases. [figroughness]b shows the saturation roughness as a function of lattice size @xmath15 for two values of @xmath0. they seem to be consistent with the kpz exponent @xmath56. however, linear fits of those plots give slopes slightly smaller than @xmath57, similarly to what was found in ref. for this reason, a systematic extrapolation of those results is necessary to decide whether the roughness scaling is kpz or not. we proceed by using the same methods of refs. @xcite, in which roughness scaling of various ballistic - like models was studied. effective roughness exponents are defined as @xmath58}{\ln{2 } }. \label{defalphal}\]] assuming that the saturation roughness has scaling corrections as @xmath59 @xcite, where @xmath60 and @xmath61 are constants, we expect @xmath62, where @xmath63 is another constant. fig. [alpha] shows effective exponents as a function of @xmath64 for @xmath51 and @xmath48, respectively using @xmath65 and @xmath66. these values of @xmath67 provide the best linear fits of the @xmath68 data for each stickness parameter. the asymptotic (@xmath69) estimates from those fits are @xmath70 and @xmath71, respectively. for @xmath48 (red triangles) with @xmath66 and @xmath51 (blue squares) with @xmath72., width=264] we estimate the dynamical exponent @xmath41 using the method proposed in ref. for each lattice size @xmath15, a characteristic time @xmath73 is defined as @xmath74 with @xmath75. the fv relation ([fv]) shows that @xmath73 is proportional to @xmath33 for fixed @xmath76, thus @xmath77. effective dynamical exponents are defined as @xmath78}{\ln{2 } }. \label{defzl}\]] figs. [zl]a and [zl]b show @xmath79 for @xmath51 and @xmath48, respectively, obtained with @xmath80. in both cases, the exponents oscillate near @xmath81, suggesting that finite - size corrections are very small. for (a) @xmath48 and (b) @xmath51.,width=264] the estimates of @xmath32 and @xmath41 are in very good agreement with kpz exponents @xmath82 and @xmath83, which is a strong numerical evidence that this model is in the kpz class in @xmath1 dimensions. a universal scaling is expected in the spd model because there is no change in its symmetries as the stickness parameter changes. in other words, the corresponding hydrodynamic growth equation may have coefficients dependent on the parameter @xmath0, but the leading spatial derivatives will be the same @xcite. due to the lateral aggregation and consequent excess growth velocity, kpz scaling is expected for any @xmath3. previous works on ballistic - like models @xcite have already shown that systematic extrapolation of finite - size or finite - time data are necessary to avoid crossover effects. as highlighted in ref. @xcite, this is a consequence of the large fluctuations in height increments, typical of those models. crossovers and finite - size corrections probably are the reasons for the deviations from kpz scaling observed in ref. @xcite. this may also be inferred by comparison with finite - size bd data from ref. @xcite. ref. @xcite suggests @xmath84 for @xmath85, while ref. @xcite gives effective exponents @xmath86 for bd in the same range of @xmath15. moreover, the growth exponents @xmath87 in lattice sizes from @xmath88 to @xmath47 for @xmath85, shown in ref. @xcite, are very near the corresponding estimates for bd in ref. @xcite (considering minimum linear correlation coefficient @xmath89 in the growth region).
Scaling for small stickness parameter
for small @xmath0, lateral aggregation is unprobable, thus most particles aggregate at the top of the column of incidence. at short times, the roughness increases as @xcite @xmath90 after a crossover time @xmath91, kpz scaling appears. our first step is to relate @xmath91 to @xmath0, for @xmath92. a typical configuration of two neighboring columns in ud is illustrated in fig. [columns]. it has a height difference @xmath93 because their heights increase without correlations. if @xmath94 is large, then a new particle inciding at the right column in fig. [columns] may aggregate at a number of positions of order @xmath94, as indicated by the circles. this means that the number of aggregation trials is of order @xmath94 and the aggregation probability for each trial is of order @xmath0. thus, the probability of no lateral aggregation after those trials (i. e. aggregation at the top of the column) is @xmath95. the probability that some lateral aggregation occurs is, consequently, @xmath96 the latter approximation requires @xmath97, which will be confirmed below. the average time for a lateral aggregation event at a given column is @xmath98. lateral aggregation immediately creates correlations between the heights of neighboring columns, thus the crossover time is @xmath99. ([wrandom]) and ([plat]) at the crossover (@xmath100, @xmath101) give @xmath102 thus, height fluctuations at the crossover scale as @xmath103 and the crossover time scales as @xmath104 these results confirm that @xmath105, as the approximation in eq. ([plat]) requires. the amplitudes of the saturation roughness [eq. ([fv])] and of the relaxation time [eq. ([scalingttimes])] scale as @xmath106 and @xmath91, respectively. following the exponent convention introduced by horowitz and albano @xcite, we have @xmath107 and @xmath108 these results are valid in any spatial dimension because ud properties are not dimension - dependent. the scaling exponents in eqs. ([defdelta]) and ([defy]) differ from those obtained in other competitive models with ballistic - like aggregation with probability @xmath7 and ud with probability @xmath8 ; in those systems, @xmath109 and @xmath110 @xcite. in solid - on - solid models with crossovers from ud to correlated growth, the exponents are also different : @xmath111 and @xmath112. the shorter crossover of the spd model is due to subsurface aggregation, which provides a large number of opportunities (of order @xmath106) for lateral aggregation of the incident particle (fig. [columns]). on the other hand, the relation @xmath113 obtained in other competitive models is also obeyed here because it is solely related to ud scaling (see e. g. the discussion in ref. @xcite). we performed simulations of the spd in @xmath47 and small values of @xmath0, from @xmath52 to @xmath114, until the steady states (roughness saturation). the saturation roughness @xmath115 and the characteristic times @xmath73 were calculated following the same lines of sec. [roughness]. [wt]a and [wt]b show @xmath73 and @xmath115, respectively, as a function of the stickness parameter @xmath0. since they were measured for constant @xmath15, they are expected to scale as the amplitudes @xmath42 [eq. ([scalingttimes])] and @xmath39 [eq. ([wsat])], respectively. fits of the data for @xmath116 give exponents @xmath117 [eq. ([defy]) and fig. [wt]a] and @xmath118 [eq. ([defdelta]) and fig. [wt]b]. and (b) saturation roughness @xmath115 in size @xmath47 as a function of the stickness parameter. solid lines are fits of the data for @xmath119.,width=264] the estimate of @xmath120 is in good agreement with the theoretical prediction of eq. ([defy]). however, the estimate of @xmath121 shows a discrepancy of @xmath122 from the theoretical prediction of eq. ([defdelta]). note that the fits in figs. [wt]a and [wt]b considered @xmath123, which are not very small values of @xmath0, thus deviations are expected, particularly in the smaller exponent (@xmath121). unfortunately, it is very difficult to obtain accurate estimates for smaller values of @xmath0 because relaxation times become very large and roughness fluctuations also increase. using smaller system sizes is also inappropriate because it enhances crossover effects @xcite estimated the crossover times for @xmath124 and obtained @xmath125, which is significanly different from the theoretical prediction in eq. ([tc]). however, measuring reliable crossover times is a difficult task, as explained in sec. [roughness]. on the other hand, the same work shows that the saturation time for @xmath47 scales as @xmath126, which is in good agreement with our estimate. the scaling of the amplitude @xmath46 in eq. ([defbeta]) can be predicted along the same lines of refs. @xcite for other competitive models : @xmath127 for the spd model in @xmath1 dimensions, we obtain @xmath128. this very small exponent gives a very slow variation of @xmath46 with the stickness parameter. it explains the small distance between the curves for different values of @xmath0 in fig. [figroughness]a. in ref. @xcite, a model similar to the spd was introduced. the particles incide vertically and, at each site of its trajectory with a nn occupied site, it may aggregate with probability @xmath7. otherwise, the particle moves down one site. if no lateral aggregation occurs, the particle aggregates at the top of the column of incidence. in fig. [modelspd], particle a may aggregate to positions labeled 2, 3, and 4. in positions 2 and 3, aggregation trials have probability @xmath7. if the particle does not aggregate at one of those points, it moves to position 4 and aggregates there. particle b may aggregate at position 6 with probability @xmath7, otherwise it moves to position 7 and aggregates there. for small @xmath7, most lateral aggregation trials are rejected, thus ud dominates. large local height fluctuations appear, similarly to fig. [columns]. the increase of the local height difference @xmath94 and the probability of lateral aggregation @xmath129 are given by eqs. ([deltah]) and ([plat]), with @xmath0 replaced by @xmath7. thus, the same reasoning of sec. [approach] leads to the same scaling relations of the spd model with @xmath0 replaced by @xmath7. in the notation of ref. @xcite, exponents @xmath130 and @xmath131 are predicted by our scaling approach. the numerical estimates of that work, @xmath132 and @xmath133, differ from those predictions, probably because they were obtained by data collapse methods that do not account for scaling corrections.
Porosity and pore height scaling
for @xmath85, the samples have large porosity @xmath134. when @xmath0 decreases, @xmath135 decreases because ud creates no holes. [samplesspd] shows regions of some samples obtained with small stickness parameters. the porosity decrease is accompanied by the formation of longer pores extended in the vertical direction. this is a consequence of the increase of the height fluctuation @xmath106 before a lateral aggregation event [eq. ([deltahc])]. (in lattice units) of samples grown with stickness parameters (a) @xmath52, (b) @xmath136, (c) @xmath137, and (d) @xmath138.,width=264] the number of deposited layers necessary to attain a steady state value of @xmath135 is relatively small, typically of the order of @xmath91 in eq. ([tc]). this is expected because pores are narrow, even in pure bd, thus porosity depends only on short wavelength height fluctuations, which saturate at short times (in the absence of scaling anomaly @xcite). our scaling approach can be used to predict the dependence of the porosity @xmath135 and the average pore height on the parameter @xmath0, as follows. during the time interval @xmath91 between two lateral aggregation events, the number of particles deposited at the top of a given column is approximately @xmath91 (note that we are still using unit lattice constant and unit deposition time of a layer, @xmath27 and @xmath28). the size of a long pore produced by the lateral aggregation is @xmath106 [eq. ([deltahc])]. consequently, for small @xmath0, the porosity (pore volume divided by total volume) is expected to scale as @xmath139 this is valid in the limit of very small @xmath0, in which @xmath140. the small exponent in eq. ([pscaling]) explains why a large decrease of @xmath0 leads only to mild reduction of porosity. this is remarkably illustrated in fig. [samplesspd], in which @xmath0 varies three orders of magnitude, while the porosity decreases from @xmath141 to @xmath142, i. e. changes by a factor smaller than @xmath143. the porosity scaling in the spd also differs from other competitive models involving ballistic - type aggregation. examples are the bidisperse ballistic deposition @xcite and the bd - ud competitive model, in which @xmath144 (@xmath7 is the probability of the ballistic - like component). we simulated the spd in size @xmath47 for small values of @xmath0 in order to measure the porosity between times @xmath145 and @xmath146. in all cases, @xmath147 is much larger than the crossover time and @xmath148 is much smaller than the relaxation time @xmath33. [phi]a shows the porosity as a function of the stickness parameter. the linear fit for @xmath149 gives @xmath150, in excellent agreement with eq. ([pscaling]). although these values of @xmath0 are very small, the corresponding values of @xmath6 and of @xmath135 are not very small. thus, scaling corrections are particularly weak in this case. . (b) average pore height as a function of the stickness parameter. dashed lines have slopes @xmath151 (right) and @xmath152 (left).,width=264] for very small @xmath0, the pores are long and isolated, as illustrated in fig. [samplesspd]d. the average pore height is expected to scale as eq. ([deltahc]), because a pore is formed only when a lateral aggregation event occurs. however, for @xmath153 or larger, many pores occupy two or more neighboring columns. this can be observed in figs. [samplesspd]a, [samplesspd]b, and [samplesspd]c. here we define pore height as the vertical distance between the aggregation position and the top of the incidence column in any lateral aggregation event. its average value, @xmath154, is taken over all lateral aggregation events between @xmath147 and @xmath148 in @xmath155 different samples. for small @xmath135, pores are isolated, thus @xmath154 is a reliable approximation of the average pore height, and is expected to scale as eq. ([deltahc]). for @xmath135 not too small, some pores occupy two or more neighboring columns, and all these columns contribute to @xmath154 (each one had a lateral aggregation event). fig. [phi]b shows @xmath154 as a function of @xmath0. the slope of that log - log plot evolves from @xmath151 for @xmath156 to @xmath152 for @xmath157. the latter is @xmath158 smaller than the theoretically predicted value @xmath159 [eq. ([deltahc])], which indicates the presence of large scaling corrections. ref. @xcite measured the porosity of samples with @xmath160, with results in qualitative agreement with ours. however, the low porosity scaling was not addressed there. @xcite suggests that the porosity scales with @xmath7 (equivalent to @xmath0) and with the lattice size @xmath15. the latter is expected only as vanishing corrections, since porosity does not depend on long wavelength fluctuations. this explains the small (effective) exponents @xmath0 and @xmath13 obtained in that work. on the other hand, ref. @xcite estimates the long - time scaling on @xmath7 with exponent @xmath161, which is to be compared with the theoretical prediction @xmath50. the discrepancy is probably related to the use of data collapse methods.
Spd model in @xmath2 dimensions
the aim of this section is to show that the main features of the spd model in @xmath1 dimensions can be extended to @xmath2 dimensions, namely the kpz roughening of the outer surface and the porosity scaling derived by the superuniversal approach of sec. [scaling]. the aggregation rules of the spd model have to be extended in this case. first, nn interactions are considered in two substrate directions, with a total of four nn in the same height. secondly, nnn interactions appear with aggregated particles in the same height (four neighbors) and with particles at the level immediately below (four neighbors). roughness scaling of ballistic - like models usually show large corrections @xcite. an alternative to search for the universality class of a given model is the comparison of scaled roughness distributions of relatively small systems because the finite - size corrections of those quantities are much smaller @xcite. we simulated the spd model with @xmath48 in substrates of lateral size @xmath88 up to the steady state (roughness saturation). in this regime, the square roughness @xmath162 of several configurations is measured. @xmath163 is the probability density of the square roughness of a given configuration to lie in the range @xmath164 $]. this quantity is expected to scale as @xmath165 where @xmath166 is the rms fluctuation of @xmath167 and @xmath168 is a universal function @xcite. fig. [dist] shows the scaled roughness distribution of the spd model and the distribution of the restricted solid - on - solid (rsos) model @xcite in substrate size @xmath88. the latter is a well known representative of the kpz class and its roughness distributions have negligible finite - size effects @xcite. the excellent collapse of the curves in fig. [dist] is striking evidence that the spd model also belongs to the kpz class in @xmath2 dimensions. (squares) and of the rsos model (solid curve) in @xmath2 dimensions, with @xmath88., width=264] we also simulated the spd model in size @xmath47 for small values of @xmath0 and measured the porosity between times @xmath169 and @xmath170. we observe that the porosity is larger than in the @xmath1-dimensional samples for the same value of @xmath0. for instance, for @xmath48, the porosity exceeds @xmath171. this is a consequence of the larger number of interactions of the incident particle with nn and nnn in @xmath2 dimensions, which facilitates lateral aggregation. [phi3d] shows the porosity as a function of @xmath0, for low values of that parameter. the linear fit for @xmath149 gives @xmath172, which is also in excellent agreement with eq. ([pscaling]). this supports the extension of the scaling approach of sec. [scaling] to @xmath2 dimensions. dimensions. the solid line is a linear fit of the data for @xmath173., width=264] an important consequence of this scaling approach is to facilitate the design of samples with the desired values of porosity and elongate pores. however, one has to take care with the fluctuations in the value of @xmath135 in the first layers of the deposit, typically produced at @xmath174.
Conclusion
we studied surface and bulk properties of porous deposits produced by a model proposed in ref. @xcite, in substrates with one and two dimensions. the model shows a crossover from uncorrelated to correlated growth for small values of the stickness parameter @xmath0. in @xmath1 dimensions, a systematic analysis of simulation data for saturation roughness and relaxation times shows that the model belongs to the kpz class. finite - size corrections explain the previous claim of deviations from kpz scaling. in @xmath2 dimensions, kpz roughening is confirmed by comparison of roughness distributions. a scaling approach for small values of @xmath0 is proposed to relate the crossover time and the local height fluctuations with that parameter, respectively giving exponents @xmath175 and @xmath159. these results are consequence of the ud properties, thus they do not depend on the spatial dimension. numerical results confirm these predictions. the crossover exponents are smaller than those of other competitive models that consider aggregation only at the outer surface @xcite. the same approach predicts the porosity scaling as @xmath6, which is in good agreement with simulation results in @xmath1 and @xmath2 dimensions. this result is important for using the model to produce porous samples representative of real materials. this may also help to model samples with desired porosity and pore height, particularly for the possibility of controlling the scaling properties by changing the kinetics of subsurface aggregation. | we study surface and bulk properties of porous films produced by a model in which particles incide perpendicularly to a substrate, interact with deposited neighbors in its trajectory, and aggregate laterally with probability of order @xmath0 at each position.
the model generalizes ballistic - like models by allowing attachment to particles below the outer surface. for small values of @xmath0, a crossover from uncorrelated deposition (ud) to correlated growth
is observed.
simulations are performed in @xmath1 and @xmath2 dimensions.
extrapolation of effective exponents and comparison of roughness distributions confirm kardar - parisi - zhang roughening of the outer surface for @xmath3.
a scaling approach for small @xmath0 predicts crossover times as @xmath4 and local height fluctuations as @xmath5 at the crossover, independently of substrate dimension.
these relations are different from all previously studied models with crossovers from ud to correlated growth due to subsurface aggregation, which reduces scaling exponents.
the same approach predicts the porosity and average pore height scaling as @xmath6 and @xmath5, respectively, in good agreement with simulation results in @xmath1 and @xmath2 dimensions
. these results may be useful to modeling samples with desired porosity and long pores. | 1506.02756 |
Introduction
planetary nebulae (pns) have become increasingly important in extragalactic astronomy, for distance determinations via their luminosity function (lf) (*??? * ; *??? * ; *??? * ; *??? * ; *??? * ; *??? * and references therein), as kinematic tracers of the dark halos of galaxies @xcite, and as tracers for the distribution and kinematics of the diffuse stellar population in galaxy clusters @xcite. due to their strong narrow line emission at @xmath3 \lambda 5007 $], pns can be easily detected out to distances beyond @xmath4 with narrow - band photometry and slitless spectroscopy @xcite, and to @xmath5 with multi - slit imaging spectroscopy @xcite. moreover, they are observed in elliptical and spiral galaxies, making them an indispensible tool to support distances obtained by other methods (such as cepheids, surface brightness fluctuations, the tully - fisher relation, sne ia), and to measure the kinematics of stellar populations whose surface brightness is too faint for absorption line spectroscopy. for distance determination the planetary nebulae luminosity function (pnlf) is normally modeled as having a universal shape that depends only on the absolute bright magnitude cutoff @xmath6 : @xmath7 where @xmath8 is the number of pns with absolute magnitude @xmath9 @xcite. observationally, the cutoff magnitude @xmath6 has a quasi - universal value of @xmath10 with only a weak dependence on host galaxy metallicity expressed by the system s oxygen abundance, and which can be compensated for by a quadratic relation in @xmath11 $] @xcite. in practice, the pn magnitudes @xmath12, after correcting for the interstellar reddening, are fitted to the model pnlf of eq. [pnlfeqn] convolved with the photometric error profile, yielding a value of the distance modulus @xcite. the absence of any systematic variations in @xmath6 and the pnlf shape has been verified in galaxies with significant population gradients, and among galaxies of different morphologies within galaxy clusters / groups up to virgo (*??? * ; *??? * and references therein). this universality of the pnlf and the cutoff magnitude @xmath6 must be considered surprising, given that the pn luminosity in the @xmath3 \lambda 5007 $] line depends on the mass and metallicity of the central star, as well as on the electron gas temperature, optical thickness and dust extinction of the surrounding nebula. indeed, some current semi - analytic simulations of the pnlf seem to be at odds with the observational trends. @xcite indicate small possible dependencies of @xmath6 on the total size of the pn population, on the time elapsed since the last episode of star formation, and on how optically thin the pns are ; concluding, however, that only careful studies would detect such effects in the observed pnlf. in contrast, more recent pnlf simulations by @xcite contradict the observed narrow spread in @xmath6 and predict large variations of several magnitudes depending on a variety of realistic star formation and evolution scenarios. so is the pnlf truly quasi - universal and its cutoff magnitude nearly independent of population age and metallicity? pns are also important as test particles to study the kinematics and dark matter distribution in the halos of elliptical galaxies. since the pn population is expected to arise from the underlying galactic stellar distribution, their radial velocities can be used as effective kinematic tracers of the mass distribution. however, the required pn sample sizes are many 100s @xcite, or at least 100 or more in conjunction with absorption line spectroscopy, which has limited this application to only a few nearby galaxies @xcite. in recent simulations of disk galaxy mergers involving dark matter, stars, and gas, @xcite predict that the young stars formed in the merger have steeper density profiles and larger radial anisotropy than the old stars from the progenitor galaxies, and they argue that if the pns observed in elliptical galaxies were to correspond to the young population rather than to all stars in the simulations, their velocity dispersion profile would match the measured dispersion profiles of @xcite. so do pns really trace the stars and their kinematics in elliptical galaxies? different stellar populations may have, and in general would have, different phase - space distributions in the same galaxy potential. the simplest approach for dynamical modelling, taking the pn velocities as a random sampling of the stellar velocities, is however valid only when the pn population properties and their kinematics are uncorrelated. except in special cases this also requires that the pnlf is independent of the stellar population. vice - versa, if there existed differences in the pnlf or the bright cutoff magnitude for different stellar populations, they would best be identified by studying the correlations between pn magnitudes and kinematics or positions of these tracers, in a single galaxy where all pns are at the same distance. in this paper, we report on such a study in the elliptical galaxy ngc 4697, an excellent target for this purpose because of the large sample of pn velocities known from @xcite. our analysis shows the existence of distinct pn populations which differ in their kinematics, brightnesses, and spatial distributions. this suggests that the answer to both the questions posed above may be no in general, different stellar populations may have slightly different pnlfs, and the observed pn population in elliptical galaxies may not be a fair tracer of their stars. the paper is organised as follows : in [data] we review the properties and pn data of this galaxy and discuss the magnitude and velocity completeness of our sample. our statistical analysis of these data is given in [analysis] where we demonstrate the inhomogeneity of the sample in the space of velocities, magnitudes, and positions. [discn],[result] conclude our work, giving also a brief discussion of its implications.
Ngc 4697: properties and pn sample
ngc 4697 is a normal, almost edge - on e4 - 5 galaxy located along the virgo southern extension. from the multi - colour ccd photometry of @xcite, the effective radius is @xmath13, the mean ellipticity is 0.45, and the pa is constant, consistent with a near - axisymmetric luminosity distribution. isophotal analysis shows that this galaxy has a positive @xmath14 coefficient suggesting a disk - like structure within @xmath15 @xcite. from optical spectroscopy, its dominant stellar population has an age of @xmath16 gyr @xcite, consistent with the red mean b - v=0.91 colour. stellar absorption line kinematics along the major axis (pa @xmath17) of ngc 4697 have been reported by @xcite and @xcite ; these velocity data can be well described by dynamical models based on the luminous mass distribution only. @xcite detected and measured radial velocities of 531 pns extending out to @xmath18 in this galaxy, with observational errors of @xmath19. via dynamical analysis, they determined a constant mass to light ratio @xmath20 within @xmath21, which is consistent with a @xmath22 gyr old stellar population with a salpeter mass function and slightly super solar metallicity. x - ray observations with rosat @xcite show only small amounts of hot gas in the halo of this galaxy. using more recent _ chandra _ data, @xcite could resolve most of the x - ray emission into nonuniformly distributed x - ray binary point sources (xps), suggesting that ngc 4697 has lost most of its interstellar gas. though ngc 4697 does not show any signature of recent interactions, @xcite present evidence that the distribution of these x - ray sources is inconsistent with the optical morphology of ngc 4697, and propose that these sources are mostly high mass x - ray binaries (hmxbs) associated with young stellar populations related to fallback of material in tidal tails onto a relaxed merger remnant, or to shock - induced star formation along tidal tails. they estimate typical fallback times of such tidal tails to be much longer than the settling timescale of the remnant and expect similar results for other elliptical galaxies with populations of @xmath23 gyr age. for the work in this paper we use the pn sample presented by @xcite. after removing the possible contaminants and unclear detections, they report unambigous detection of 535 pns. however only 526 out of 535 pns have confirmed measurements of velocity _ and _ magnitude, and we use these in our analysis. in order to determine the pnlf, it is crucial to estimate the magnitude where pn detection incompleteness sets in. detectability of a pn varies with the background galaxy surface brightness ; for a statistically complete sample the surface number density of pns should be directly proportional to it. @xcite show that their pn sample is statistically complete down to @xmath24 magnitudes outside an elliptical region of semi - major axis @xmath25. in our analysis, we have thus defined two data sets : a _ complete sample _ (with pns brighter than @xmath26, outside the central ellipse of semi - major axis @xmath25), and a _ total sample _ (consisting of all pns with measured magnitude and radial velocity). the total number of pns in these data sets is 320 and 526, respectively. the systemic velocity (@xmath27) of ngc 4697, obtained by averaging the observed velocity of all 526 pns is @xmath28, which agrees with the values quoted in the literature (*??? * and references therein). the on - band filter configuration used to detect and measure velocities of these pns has a peak wavelength of @xmath29 , peak transmission of @xmath30, equivalent width of @xmath31 , and fwhm of @xmath32 @xcite. the fwhm corresponds to a velocity range of @xmath33 around the systemic velocity of ngc 4697. thus the filter transmission width is large enough to facilitate observations of pns with all velocities bound to ngc 4697, irrespective of their magnitude. indeed, even at magnitudes as faint as @xmath34 in the total sample, pns with large velocities @xmath35 are detected. thus the velocity coverage in both samples (total and complete) is not biased with respect to the pn magnitudes. the pn magnitudes were measured by @xcite from their undispersed images ; they are accurate to 0.1 and 0.2 mag for @xmath12 brighter and fainter than 26.5, with systematic effects below 2%. as a further test relevant for the present work, @xcite used the redundancy provided between their e and w fields : plotting magnitude differences between the two measurements (e and w) of pn candidates as a function of difference in distance from the center of the ccd, they found a scatter diagram without any evidence of correlation. @xcite estimated the errors in the pn velocities from calibration, image registration, spectrograph deformation and guiding errors to be @xmath36. the velocities of 165 pns were measured independently in the e and w field of @xcite ; these velocities agree within a standard deviation of @xmath37. in order to check whether a systematic difference between the velocities of bright and faint pns could be introduced by an asymmetric psf (a possibility suggested by k. freeman), we have superposed the psf s of three groups of 10 of the brightest pns, one selected at random, and two selected among those pns with the highest and lowest radial velocities. in the three cases we estimated the shift of the centroid of the entire psf with respect to the centroid of the upper part. the shifts were smaller than @xmath38, and in some cases they were in the opposite sense compared to the results discussed below.
The distribution of ngc 4697 pns in velocity, magnitude, and position
in this section, we search for stellar population effects in the kinematics of the pns in ngc 4697, by analysing the total and complete data sets with respect to their three observables : velocity, magnitude and position. for both data sets, we convert the observed pn radial velocities into co - rotating or counter - rotating velocities, as follows. with the galaxy center at the origin of the reference frame, and the x - axis oriented along the major axis (pa@xmath39 deg), the absorption line stellar - kinematic data predict positive line - of - sight mean velocity with respect to the galaxy systemic velocity, at slit positions towards the south - west of the center with x coordinate @xmath40, and vice versa. we denote this sense of rotation as _ co - rotating _, and the opposite sense as _ counter - rotating_. by definition, the major axis absorption line data is _ co - rotating_. after subtracting the systemic velocity from the pn radial velocities, we define reduced velocities @xmath41 and denote the pns with @xmath42 @xmath43 (@xmath44) 0 as co - rotating (counter - rotating). by definition, the major axis absorption line data have @xmath45. the resulting values of @xmath42 are displayed against the observed magnitudes in figure [mvplot]. even at magnitudes as faint as @xmath34 in the total sample, pns with large velocities @xmath35 are detected, showing that there is no kinematic bias at faint magnitudes. henceforth, unless stated otherwise, we will always use the complete sample for our analysis. figure [mvplot] shows that the complete pn sample appears to exhibit a correlation between magnitudes and kinematics, with faint pns showing more co - rotation than bright pns. we have performed several statistical tests to verify the significance and look for the origin of this correlation. table [tabpear] shows the results of pearson s r - test for correlated data. velocities of counter - rotating pns are strongly linearly correlated with their brightness, while those of co - rotating pns are almost independent of their magnitude distribution. further, we divided our sample into 3 equal magnitude bins each of size @xmath46, hereafter referred to as _ faintest _, _ intermediate _, and _ brightest pns _, and computed the mean reduced velocity and its dispersion in each of these bins along with the significance of their differences. as shown in table [tabvtftest], the _ faint _ and _ bright pn subsamples _ defined through these bins have different mean reduced velocity and dispersions at @xmath47 and @xmath48 confidence, respectively. in figure [6bins] we have plotted the cumulative velocity distribution of pns in the brightest and faintest magnitude bins. there is a visible excess of bright pns with counter rotating velocities. it is particularly evident from this figure that the brightest counter rotating pns display a velocity distribution that differs from the rest of the pns from the complete sample with high confidence. thus it is clear that the observed correlation between the pns kinematics and their magnitudes is compelling, and it arises because the faintest and brightest pns have significantly different velocity distributions. there appears to exist an additional component of bright, counter - rotating pns with respect to the overall sample..pearson s r - test for linear correlation of pn magnitudes with reduced velocity @xmath49, for the entire complete sample, the co - rotating, and the counter - rotating subsamples. values of r close to @xmath50 indicate a strong linear correlation ; values close to @xmath51 indicate little or no correlation. @xmath52 is the probability that two uncorrelated variables would give the r - coefficient as large as or larger than the measured one, for a normal distribution of r. small values of @xmath52 imply significant correlation. [cols="^,^,^",options="header ",] , orange dashed line), faintest pns (@xmath53, purple dashed - dotted line), and the entire complete sample (black solid line). the entire velocity range is shown in the top panel. in the middle panel the velocities are divided according to their sense of rotation. the bottom panel shows the cumulative distribution (shown from @xmath54 to @xmath55 values), normalised at @xmath56. the kolmogorov smirnov probabilities show that the brightest counter rotating pns have significantly different velocity distribution from the rest of the pns.,title="fig : "], orange dashed line), faintest pns (@xmath53, purple dashed - dotted line), and the entire complete sample (black solid line). the entire velocity range is shown in the top panel. in the middle panel the velocities are divided according to their sense of rotation. the bottom panel shows the cumulative distribution (shown from @xmath54 to @xmath55 values), normalised at @xmath56. the kolmogorov smirnov probabilities show that the brightest counter rotating pns have significantly different velocity distribution from the rest of the pns.,title="fig : "], orange dashed line), faintest pns (@xmath53, purple dashed - dotted line), and the entire complete sample (black solid line). the entire velocity range is shown in the top panel. in the middle panel the velocities are divided according to their sense of rotation. the bottom panel shows the cumulative distribution (shown from @xmath54 to @xmath55 values), normalised at @xmath56. the kolmogorov smirnov probabilities show that the brightest counter rotating pns have significantly different velocity distribution from the rest of the pns.,title="fig : "] if these correlations have a physical origin, they should also be manifest in the spatial distribution of these pns. thus we now enquire whether the pn kinematics and magnitudes depend on their spatial location in the galaxy. in figure [tspace] we plot the spatial locations of all the 526 pns in this galaxy. the central incompleteness ellipse is also displayed. pns brighter than @xmath57 (which is @xmath58 magnitude deeper than the brightest pn) are shown as filled blue squares (co - rotating) and filled red triangles (counter - rotating). inside the incompleteness ellipse, the distribution of the bright pns appears to be concentrated around an elliptical annulus. however, we did not find any kinematic evidence (like a rotation curve signature) relating these bright pns to the central stellar disk : either they are not physically related to the disk, or the evidence from the data is inconclusive. outside the incompleteness ellipse, the distribution of bright pns does not follow the surface brightness of the host galaxy : there is a significant left - right asymmetry, with more bright pns to the right side of the galaxy minor - axis (@xmath59) than to the left side (@xmath60). @xcite discuss at length the possible differences in their e (east) and w (west) fields, and are convinced that the maximum systematic errors in the measured photometry, positions and radial velocities are below @xmath61, @xmath62 and @xmath63, respectively. hence we conclude that the left - right asymmetry in number counts of bright pns is not affected by detection uncertainties. subsequently, we carried out several tests to check whether the brightest and faintest, or the co- and counter - rotating pns are distributed differently in the galaxy. it turns out that the radial pn distribution is independent of their sense of rotation. however, the distribution of pn distances from the galaxy mid - plane differs for co- and counter - rotating pns at a confidence level of @xmath64. are shown by the orange dashed line, the red dash - triple dotted line, and the purple dash - dotted line, respectively. all pns in the complete sample are shown by the solid black line. note the large increase in the distribution of bright pns and the moderate increase in the distribution of intermediate pns, relative to the faint pns, at @xmath65 (see text for details). the ks probability of the faintest and brightest pns being drawn from the same underlying distribution is only @xmath66, while the intermediate pns are still compatible with the faint pns (ks probability @xmath67).,title="fig : "] are shown by the orange dashed line, the red dash - triple dotted line, and the purple dash - dotted line, respectively. all pns in the complete sample are shown by the solid black line. note the large increase in the distribution of bright pns and the moderate increase in the distribution of intermediate pns, relative to the faint pns, at @xmath65 (see text for details). the ks probability of the faintest and brightest pns being drawn from the same underlying distribution is only @xmath66, while the intermediate pns are still compatible with the faint pns (ks probability @xmath67).,title="fig : "] . solid (dashed) histograms show the mean velocity (dispersion) and their errors in different angular bins. the red ragged lines show running averages of mean velocity. _ top : _ velocities of pns brighter than @xmath68 in the complete sample. bin sizes are chosen so as to have approximately constant number of pns (@xmath69) in each bin. the mean velocity along the major (minor) axis is positive (negative) on both sides of the center. the running average is also over @xmath70 pn velocities. the velocity dispersion is largest (smallest) on the minor (major) axis. _ middle : _ velocities of pns with @xmath71 in the same angular bins as in the top panel. the running average is over @xmath72 pns. _ bottom : _ velocities of pns with @xmath73 in the same angular bins as in the top panel. the faint pns show a rotation pattern as for the absorption line data but with a smaller peak velocity, and are consistent with a flat dispersion of @xmath74. the running average is over @xmath75 pns. the blue dotted line shows a sinusoidal fit. the kinematics of the intermediate brightness pns is intermediate between the faint and bright pns.,title="fig : "]. solid (dashed) histograms show the mean velocity (dispersion) and their errors in different angular bins. the red ragged lines show running averages of mean velocity. _ top : _ velocities of pns brighter than @xmath68 in the complete sample. bin sizes are chosen so as to have approximately constant number of pns (@xmath69) in each bin. the mean velocity along the major (minor) axis is positive (negative) on both sides of the center. the running average is also over @xmath70 pn velocities. the velocity dispersion is largest (smallest) on the minor (major) axis. _ middle : _ velocities of pns with @xmath71 in the same angular bins as in the top panel. the running average is over @xmath72 pns. _ bottom : _ velocities of pns with @xmath73 in the same angular bins as in the top panel. the faint pns show a rotation pattern as for the absorption line data but with a smaller peak velocity, and are consistent with a flat dispersion of @xmath74. the running average is over @xmath75 pns. the blue dotted line shows a sinusoidal fit. the kinematics of the intermediate brightness pns is intermediate between the faint and bright pns.,title="fig : "]. solid (dashed) histograms show the mean velocity (dispersion) and their errors in different angular bins. the red ragged lines show running averages of mean velocity. _ top : _ velocities of pns brighter than @xmath68 in the complete sample. bin sizes are chosen so as to have approximately constant number of pns (@xmath69) in each bin. the mean velocity along the major (minor) axis is positive (negative) on both sides of the center. the running average is also over @xmath70 pn velocities. the velocity dispersion is largest (smallest) on the minor (major) axis. _ middle : _ velocities of pns with @xmath71 in the same angular bins as in the top panel. the running average is over @xmath72 pns. _ bottom : _ velocities of pns with @xmath73 in the same angular bins as in the top panel. the faint pns show a rotation pattern as for the absorption line data but with a smaller peak velocity, and are consistent with a flat dispersion of @xmath74. the running average is over @xmath75 pns. the blue dotted line shows a sinusoidal fit. the kinematics of the intermediate brightness pns is intermediate between the faint and bright pns.,title="fig : "] the left - right asymmetry is confirmed by inspecting the azimuthal distribution of the faint and bright pns. in the literature we found a related analysis by @xcite who compared the azimuthal distribution of _ chandra _ x - ray point sources (xps) with the optical surface brightness of ngc 4697. we follow their pa convention, and plot the cumulative angular distributions of the bright, intermediate, and faint pns in our complete sample in figure [zezang]. for comparison, the right panel of fig. 2 from @xcite is also shown. the angular distribution of all pns in the complete sample has a shape somewhere in between that of the xps and that of the optical light. the brightest pns are in complete disagreement with either of these distributions ; they seem to be more concentrated in a narrow angular sector between @xmath76 (see fig. [tspace]), with only @xmath66 probability that the faint and bright pn subsamples are drawn from the same azimuthal distribution. at the same time, the radial distribution of the faint and bright subsamples are not significantly different (figure [radial]). the velocity distribution of the brightest pns is also correlated with their azimuthal distribution. in figure [vang] we plot the mean radial velocity and its dispersion in angular sectors containing approximately constant number of pns from the bright subsample, as well as an angular running average of their mean radial velocity. along both sides of the major - axis of the galaxy, the brightest pns have positive velocity, while showing a relatively low velocity dispersion, perhaps due to infall as suggested for the xps by @xcite. along the minor - axis they have a large dispersion with a mean velocity that is negative. this kinematics is compatible with neither the faint pn velocities nor the stellar absorption line data. on the other hand, the fainter pns show a regular azimuthal distribution in the mean line - of - sight radial velocity and velocity dispersion. note that their velocity dispersion is in the mean smaller than that of the bright sample. the kinematics of the intermediate luminosity pns is intermediate between those of the bright and faint subsamples. the intermediate luminosity pns thus contain significant contributions from both of the different populations that dominate the bright and faint bins, respectively. of the major axis for bright and faint / intermediate pns separated at @xmath77. blue circles show stellar absorption line spectroscopy (als) data from @xcite, red crosses are individual pn velocities, and grey diamonds are near neighbour running averages. the faint / intermediate sample is in approximate agreement with the als data, while the bright sample has a velocity consistent with zero within @xmath78 and positive velocities on both sides of the galaxy center.] furthermore, including the bright pns in deriving kinematics for the whole pn sample introduces signficant contamination effects. in figure [majkindb], we demonstrate this by plotting the mean pn velocities along a @xmath79 wide slit about the major - axis. the pns have been separated into faint / intermediate and bright sub - samples at @xmath77. inside the @xmath78, the mean velocities of the faint pns agree with stellar absorption line spectroscopic (als) data from @xcite, while the bright pns have much smaller mean velocities, consistent with @xmath80. in the outer parts, the velocity asymmetry in the bright pns leads to a positive mean velocity on both sides of the galaxy center. both the faint / intermediate sample and the entire pn sample also show some asymmetry : the outer mean velocities on the @xmath81 side reach zero, but not the negative values expected from reflecting the positive values at @xmath82. this confirms that also the intermediate / faint subsample contains a fraction of pns stemming from the out - of - equilibrium population traced by the bright pns. however, the streaming velocity of the majority of the faint pns does appear to decrease on both sides of the center. similarly contaminated results could be expected for the derived velocity dispersions. several important conclusions can be drawn from these figures. (i) the bright pns as defined in section [subpop] and fig. [mvplot] do not trace the azimuthal distribution of light in ngc 4697. (ii) they do not trace the fainter pns in their azimuthal kinematics ; thus, a large fraction of them must belong to a separate pn subcomponent originating from a separate stellar population. (iii) third, they are not in dynamical equilibrium in the gravitational potential of ngc 4697. (iv) because this subpopulation does not trace the stars, including its pn velocities into dynamical analysis of the galaxy will lead to significant errors in the results. probability. _ bottom : _ cumulative magnitude distributions of the main pn sample (@xmath83, blue dotted line) and the extreme counter - rotating sample (@xmath84, red dash - dotted line) defined in the text, along with that of the total sample (black solid line). the first two luminosity functions are different with 99.7% confidence ; thus the pnlf can not be universal.,title="fig : "] probability. _ bottom : _ cumulative magnitude distributions of the main pn sample (@xmath83, blue dotted line) and the extreme counter - rotating sample (@xmath84, red dash - dotted line) defined in the text, along with that of the total sample (black solid line). the first two luminosity functions are different with 99.7% confidence ; thus the pnlf can not be universal.,title="fig : "] what can we learn about the luminosity functions of the two kinematic components of the pn system in ngc 4697? the cumulative magnitude distributions of the co - rotating and counter - rotating subsamples in fig. [mvplot] have only @xmath67 ks probability of stemming from the same distribution (top panel of figure [ksmag]). however, the fact that there are both co - rotating and counter - rotating bright pns in the bright sample whose azimuthal distribution is unmixed, shows that counter - rotation is not a _ clean _ discriminator after all for the secondary population of pns in ngc 4697. thus the luminosity functions of the main and secondary pn populations in this galaxy may be a lot more different than this figure of @xmath67 would suggest. from fig. [vang] we estimate that the main population of pns in ngc 4697 has a mean radial velocity @xmath85, with a dispersion of @xmath86, so its reduced mean velocity is @xmath87. thus in the bottom panel of fig. [ksmag] we show the cumulative magnitude distribution of the pns in the velocity range @xmath88 and compare it with the cumulative distribution of the pns with @xmath84. these velocity ranges are dominated by the main and secondary pn populations in the sample, respectively. now the kolmogorov smirnov significance test shows that the magnitude distributions from these sections of fig. [mvplot] have only probability @xmath89 of being drawn from the same distribution. this result is strong enough to imply that the pnlf can not be universal the pnlf in ngc 4697 depends on a kinematic selection. in the following, we will use the velocity range @xmath90 as an approximate kinematic selection criterion for the main pn population in ngc 4697. the luminosity function of the strongly counter - rotating pns in fig. [mvplot] a first approximation to the luminosity function of the secondary pn population in ngc 4697 differs from that of the main pn distribution so defined in the sense that it contains more bright pns near the cut - off and fewer faint pns than the main population (see fig. [ksmag]). now an important question is : in what proportion do both populations contribute to the brightest pns, and do they have different cutoff magnitudes? . circled symbols denote counter - rotating pns ; many of these are kinematic outliers. the two grey lines denote the position angle of the minor axis. the distribution of blue diamonds is approximatively uniform in pa except near pa@xmath91 where there are 12 pns (crossed diamonds) instead of the expected 2 pns. _ bottom : _ radial velocity magnitude plane. the symbols are as in the top panel. 6/8 of the brightest pns are either kinematic outliers or are found in the overdense angular region., title="fig : "]. circled symbols denote counter - rotating pns ; many of these are kinematic outliers. the two grey lines denote the position angle of the minor axis. the distribution of blue diamonds is approximatively uniform in pa except near pa@xmath91 where there are 12 pns (crossed diamonds) instead of the expected 2 pns. _ bottom : _ radial velocity magnitude plane. the symbols are as in the top panel. 6/8 of the brightest pns are either kinematic outliers or are found in the overdense angular region., title="fig : "] to investigate this, we show in figure [vla] the velocities, magnitudes, and position angles of the entire bright pn subsample (see figs. [mvplot] and [vang]). the top panel shows (i) that 13/16 of the bright pns, whose radial velocities differ most from those of the faint population, are counterrotating. this explains the differences between the velocity distributions of co- and counter - rotating pns that first suggested more than one pn population in section [subpop]. (ii) also, even in the kinematically normal bright pns, there is a large overdensity (10/12) in the angular range pa@xmath91. the bottom panel shows in addition that many of the brightest pns are either kinematic outlyers or found in the angular overdensity (6/8). clearly, to arrive at a main population of ngc 4697 pns that is in dynamical equlibrium in the gravitational potential, we must remove the angular overdensity. then we are left with 16 pns in the range @xmath92 of a total bright subsample of 42 pns ; however, there is some freedom in the way the angular overdensity is removed. in the following, we explore two assumptions : (i) using all 12 pns in the angular overdensity, but weighting each one by @xmath93, and (ii) removing the brightest 10 of the 12 pns in the overdensity. the resulting luminosity functions for both cases are plotted in figure [clumfun] ; they differ only slightly. thus in the following we use the kinematic condition @xmath94 together with assumption (ii) above to ensure azimuthal uniformity of the bright pns, as an improved selection criterion for pns in the main population in ngc 4697. we keep all pns in the intermediate luminosity bin with @xmath92, because fig. [zezang] showed that their azimuthal asymmetry is not large. also, we have checked that the mean angular velocity distribution in this magnitude bin after applying the kinematic selection follows approximately the sinusoidal variation of the faint pns, and the velocity disperion is approximately constant. the resulting main population sample is identified by their brightness, distribution and kinematic properties. comparing the luminosity function in figure [clumfun] of this main pn population, with the luminosity function of all pns in ngc 4697, we see that the bright cutoff of the main population is shifted to fainter magnitudes. we note that the bright cutoff could be shifted further to fainter magnitudes if some of the kinematically normal and azimuthally uniform bright pns were also part of the secondary population, which we can not tell from the present data. we can ask now what is the effect, in practice, on the pnlf distance determination. the reduced sample for the main pn population in ngc 4697 has 214 objects. after binning these data into 0.2 mag intervals, we transform the apparent magnitudes @xmath12 into absolute magnitudes, adopting the extinction correction of 0.105 mag @xcite and assuming different distance moduli, and we compare the results with the pnlf simulations of @xcite. this is the same procedure used in @xcite for the pnlf distance determination. the comparison is shown in fig. we conclude that the pnlf distance modulus should be increased slightly from 30.1 (the earlier determination based on the full sample) to 30.2 or 30.25. for comparison, the @xmath95-fit to the same data (blue line in fig. [clumfun]) gives 30.22. this correction would bring the pnlf distance modulus in better (but not perfect) agreement with the surface brightness fluctuation (sbf) distance modulus (@xmath96) reported by @xcite. ) of 214 objects, binned into 0.2 mag intervals. the apparent magnitudes @xmath12 have been transformed into absolute magnitudes @xmath97 by adopting an extinction correction of 0.105 mag and distance moduli indicated in each plot. the three lines are pnlf simulations @xcite. for a distance modulus 30.15 the brightest pns are a bit too weak, therefore the distance must be increased. but 30.35 is clearly excessive. the best fit is for 30.2 or 30.25., title="fig : "]) of 214 objects, binned into 0.2 mag intervals. the apparent magnitudes @xmath12 have been transformed into absolute magnitudes @xmath97 by adopting an extinction correction of 0.105 mag and distance moduli indicated in each plot. the three lines are pnlf simulations @xcite. for a distance modulus 30.15 the brightest pns are a bit too weak, therefore the distance must be increased. but 30.35 is clearly excessive. the best fit is for 30.2 or 30.25., title="fig : "]
Discussion
bright pns play a significant role in the analysis of the last section. is it possible that the brightest pns in ngc 4697 are contaminated by compact hii regions such as those observed in @xcite? the recent observations of @xcite have shown that this can not be so. these authors took spectra of 13/42 pns in our bright subsample in ngc 4697 with fors2@vlt ; these have no detectable continuum and the line ratios of metal - rich pns. the same argument also shows that these bright pns can not be background emission line galaxies. moreover, ly alpha emission galaxies would come in at [oiii] magnitudes of m@xmath98 (see fig. 4 of *??? *), while the bright pns in ngc 4697 have m@xmath99. furthermore, one may wonder whether the bright pns in ngc 4697 might simply be foreground objects which could be closer to us and hence brighter than the true ngc 4697 pns on which they would be superposed. given that ngc 4697 is located in the southern extension of the virgo cluster, known to contain an intracluster population of pns @xcite, this possibility deserves to be considered. however, the following observational facts show that the bright pns in ngc 4679 are not intracluster pns (icpns). (1) ngc 4697 is in fact closer than the virgo cluster. the pnlf from @xcite places it at @xmath100 (m - m@xmath101), while the distance modulus from the pnlf of m87 is m - m@xmath102 @xcite. even if we discarded the entire brightest 0.3 mag of the ngc 4697 pns, ngc 4697 would still be at 75% of the distance of m87. (2) the velocity dispersion of the bright and unrelaxed pns in ngc 4697 is @xmath103, but varying azimuthally, while that of the fainter main population is @xmath104 (table [tabvtftest], figs. [mvplot], [zezang]). both are much smaller than the velocity dispersion of icpns in virgo @xcite. the radial distribution of the bright population outside the incompleteness ellipse is concentrated towards the galaxy center and ks compatible (92%) with the radial distribution of the fainter pns (see fig. [radial]). thus also the bright pns in ngc 4697 are bound to the galaxy. (3) the surface density of pns with m@xmath99 is 0.58 pns / arcmin@xmath105, while the mean surface density of virgo icpns is 0.02 pns / arcmin@xmath105, much smaller @xcite. this last argument also rules out significant contamination by chance superpositions of pns even closer than ngc 4697. we conclude that the bright pn population in ngc 4697 consists of genuine pns, and that it is dynamically bound to the ngc 4697 system. the irregular angular distribution and kinematics of these pns, by comparison with the fainter main population, show that they must belong to a separate stellar population not yet in dynamical equilibrium with ngc 4697. pn samples as large as the one for ngc 4697 are still the exception. yet to undertake the analysis described in section [analysis] it was crucial to work with a complete sample of some 300 pns. the pn.s project @xcite may lead to complete samples of similar size but as of now the typical sample sizes are @xmath106 @xcite and their magnitude distributions and completeness have not been studied. the only other elliptical galaxy with a comparable (even larger) sample is cen a @xcite. a detailed analysis of the distribution of cen a pns in the magnitude velocity plane is still pending, but @xcite analysed the pnlf in cen a as a function of radius, based on narrow - band surveys. they concluded that no population effect on the pnlf bright cut - off could be seen, suggesting that the 0.3 mag difference between their main and outer halo samples was due to filter transmission uncertainties. analysis of the large sample from @xcite will clarify whether the pns in cen a are consistent with one or more subpopulations. this will be particularly interesting because cen a is believed to be the remnant of a galaxy merger, so one might expect pns from both the older stellar populations of the progenitors as well as from the stars formed in the subsequent interaction between them. without large kinematic samples, searches for pn subpopulations must be based on the pn luminosity distributions. the work of jacoby, ciardullo and collaborators cited in the introduction has shown that the pnlf is remarkably uniform. however, there are exceptions : we recall that in the halo of m84 there exists a small population of overluminous pns whose cutoff is @xmath107 mag brighter than that of the main m84 population, but which appear nonetheless bound to the halo of m84 @xcite. they must therefore be intrinsically bright, due to some stellar population difference. in m87, the only overluminous pns projected onto the galaxy for which velocities have been measured @xcite, have very large relative velocities with respect to m87. this is most naturally explained if these pns have fallen into the deep potential well of m87 from far out in the cluster ; this would again imply an intrinsic population difference. however, a larger kinematic sample in m87 is required to be certain. the existence of the bright pn subpopulation in ngc 4697 implies some uncertainty in the pnlf cut - off luminosity for this galaxy. depending on whether or not the azimuthally symmetric brightest pns are part of the main population, the cut - off luminosity of the main pn population in ngc 4697 is fainter than that of the whole population by @xmath108 mag (figs. [clumfun], [rob]). while this is consistent with the m84 result, is it also consistent with the systematic studies of pnlf distances? @xcite have compared the pnlf and sbf distances, finding a distribution of residuals with a systematic offset by @xmath109 mag, which they suggested is due to internal extinction effects, and with a fwhm of @xmath110 mag. the offset has in the mean - time been reduced to @xmath111 mag, following a revision of the sbf distance scale by @xcite. the width of this distribution is consistent with their determination of the observational errors in both methods. however, the offset we have determined for ngc 4697 is also consistent with the distribution of residuals in @xcite. we have shown that a large fraction of the bright pns in ngc 4697 belong to a secondary, dynamically young stellar population that is not well - mixed in the gravitational potential of the galaxy. late infall of tidal structures @xcite or a merger with a smaller galaxy some time ago would be natural ways to add such an unmixed stellar component to ngc 4697. what physical population difference is correlated with this dynamical youth? @xcite show from their spectroscopic data for 13 bright pns that these have near - solar metallicities. of these 13 bright pns, 6 are inside the incompleteness ellipse, one has no measured velocity, and 6 belong to our secondary population. @xcite also use long - slit spectroscopy to show that the metallicity of the integrated stellar population within one effective radius has solar or higher metallicities. these observations make it unlikely that metallicity is the main factor responsible for the different magnitude distributions of the main and secondary pn populations in ngc 4697. thus the more likely driver would appear to be an age difference, as suggested by @xcite and as might generally be expected in an accretion event. based on their result that the distribution of x - ray point sources in ngc 4697 does not follow the stellar light, @xcite have argued that this is because these sources were formed several @xmath112 years ago in tidal tails that are now falling back onto the galaxy. note, however, that the integrated light in ngc 4697 shows no evidence of young stars with mean age @xmath113 gyr @xcite, so this younger component could not be luminous enough to contaminate the integrated light to the level measured. also note that the observed increase of extinction in the pn envelope with pn core mass more than compensates for the increase of core luminosity with core mass, for bright pne in local group galaxies @xcite, so that stars with ages below @xmath114 gyr may not reach the [oiii] luminosity at the pnlf cutoff. a secondary stellar population younger than @xmath114 gyr is therefore unlikely as well. recently, @xcite have argued that the brightest pns in the pnlf must have core masses of @xmath115, corresponding to main sequence masses of @xmath116. they argue further that for such high - mass objects to occur in elliptical galaxies, these early - type galaxies would either have to contain a small, smoothly distributed component of young (@xmath117 gyr age) stars, or more likely, that the bright pns in these systems have evolved from blue straggler stars created through binary evolution. their blue straggler model, due to the assumption of a fixed distribution of primary - to - secondary mass ratios for the initial binaries, predicts that older stellar populations produce fewer bright pns per unit luminosity, as is observed, because the number of binary stars in a stellar population that can coalesce to @xmath116 blue stragglers decreases with time. if correct, this blue straggler model could also explain how the secondary population we found in ngc 4697 can contain a large fraction of the brightest pns in this galaxy, provided that the stellar population corresponding to this secondary pn population is appreciably younger than the main stellar population, whose age is @xmath16 gyr from optical spectroscopy @xcite. at the same time, this secondary stellar population must not be so young to violate either the constraints from the optical colours or from the envelope absorption - pn core mass correlation, i.e., must be older than @xmath118 gyr. we can give an estimate for the effect of such an intermediate age population on the optical colours as follows. the secondary subpopulation traced by the bright and predominantly counterrotating pns may contain @xmath119 of all pns in the complete sample for ngc 4697. a stellar population as blue as the bulge of m31 has a luminosity - specific pn density per unit @xmath120 up to 5 times higher than the populations characteristic for old elliptical galaxies @xcite. thus the subpopulation corresponding to the secondary pn population in ngc 4697 would be expected to contain @xmath121 of the blue luminosity of ngc 4697, spread over a large fraction of at least the e image. to detect this we need deep and accurate photometry. the unmixed spatial and velocity distributions of the secondary pn population in ngc 4697 show that that this population is _ dynamically young _, i.e, has not had time to phase - mix and come to dynamical equilibrium in the gravitational potential of ngc 4697. it may well be associated with tidal structures that were formed in a merger or accretion event @xmath122 gyr ago, and that have now fallen back onto the galaxy, or be associated with a more recent merger / accretion with a red galaxy such as described in @xcite. in a universe in which structures form hierarchically, such secondary stellar populations might be quite common in ellipitical galaxies, but they would be difficult to see. the present work shows that studying their pn populations is one promising approach of looking for such secondary populations. however, large pn samples are required ; most existing pn studies of early - type galaxies do not have the statistics for such an investigation. moreover, in only a fraction of cases may there be enough asymmetry signal to detect with a few hundred pns.
Conclusions and implications
we have analysed the magnitudes, kinematics and positions of a complete sample of 320 pns in the elliptical galaxy ngc 4697 from @xcite. this data set is large enough for drawing statistically significant conclusions, and it does not suffer from detection incompletenesses in either magnitudes or radial velocities. we know of no systematic effects in the data that could explain our results. our main conclusions are : 1. bright and faint pns in ngc 4697 have significantly different radial velocity distributions. the mean velocities of the faint and bright subsamples (co - rotating and @xmath80, respectively) and their velocity dispersions are different, with 94% and 99.3% confidence. thus the pns in ngc 4697 do not constitute a single population that is a fair tracer of the distribution of all stars. 2. the luminosity functions of the extreme counter - rotating subsample (@xmath84) and of the main population (defined by @xmath123) are different with 99.7% confidence. the pnlf is therefore not universal. 3. based on this, we suggest that there exist (at least) two pn populations in this galaxy. the secondary pn population in ngc 4697 is prominent in a sub - sample of counter - rotating pns brighter than @xmath0. the luminosity function of the entire extreme counterrotating sample may be a first approximation to the luminosity function of the secondary population. the spatial distribution of bright pns with @xmath124 is different from that of the faint pns. the bright pns do not follow the azimuthal distribution of the optical light, show a left - right asymmetry, and have a positive mean radial velocity on both sides of the galaxy major axis, but zero velocity and larger dispersion on the minor axis. they are not in dynamical equilibrium in the potential of the galaxy. the fainter population has rotation properties more similar to the absorption line velocities, with azimuthally constant dispersion. 5. using both their kinematics and angular distribution, we can estimate a lower limit to the statistical fraction of bright pns in the secondary population. based on this we estimate that the bright cutoff of the main population is uncertain by @xmath125 mag. our results have two main implications for the use of pns in extragalactic astronomy. first, for distance determinations with the pnlf, it may be important to understand how uniform the pn populations in the target galaxies are. from our analysis in ngc 4697 we estimate that unrecognized subpopulations of pns in smaller samples than that in ngc 4697 may lead to variations of @xmath125 mag in the bright cutoff. this would correspond to distance errors of some @xmath126, which, although a minor effect, could be significant in some cases. it will be necessary to verify how frequently such subpopulations occur in elliptical galaxies. we recall that also in the halo of m84 there exists a small population of overluminous pns whose cutoff is @xmath107 mag brighter than that of the main m84 population, but which appear nonetheless bound to the halo of m84 @xcite. the second implication concerns the use of pns as tracers for the angular momentum and gravitational potentials of elliptical galaxies. our analysis has shown that in ngc 4697 the bright pns do not trace the distribution and kinematics of stars and are not in dynamical equilibrium in the gravitational potential of the galaxy. the fainter pns look more regularly distributed but may also contain a fraction of pns that belongs to this out - of - equilibrium population. clearly therefore, mass determinations based on pn kinematics will in future require careful study of the pn samples being used, not only to verify that these pns are in dynamical equilibrium, but also to test for different dynamical components. even if in equilibrium, a younger population of stars may be more flattened or have a steeper fall - off than the main body of the elliptical galaxy. if the pns from this population are indeed somewhat brighter than the main population, one can recognize such differences from their lower velocity dispersion or different radial density profile. however, deep observations and large pn samples will be required. we are grateful to m. arnaboldi, r. ciardullo and r. saglia for helpful discussions, and to k. freeman, g. jacoby, e. peng and a. romanowsky for helpful comments on the manuscript. ns and og thank the swiss nationalfonds for financial support under grant 200020 - 101766. rhm would like to acknowledge support by the u.s.national science foundation, under grant 0307489. | we have analysed the magnitudes, kinematics and positions of a complete sample of 320 pns in the elliptical galaxy ngc 4697. we show (i) that the pns in ngc 4697 do not constitute a single population that is a fair tracer of the distribution of all stars. the radial velocity distributions, mean velocities, and dispersions of bright and faint subsamples differ with high statistical confidence.
(ii) using the combined data for pns brighter than @xmath0, we have identified a subpopulation of pns which is azimuthally unmixed and kinematically peculiar, and which thus neither traces the distribution of all stars nor can it be in dynamical equilibrium with the galaxy potential.
(iii) the planetary nebula luminosity functions (pnlf) of two kinematic subsamples in ngc 4697 differ with 99.7% confidence, ruling out a universal pnlf.
we estimate that the inferred secondary pn population introduces an uncertainty in the bright cutoff magnitude of @xmath1 mag for this galaxy.
we argue that this secondary pn distribution may be associated with a younger, @xmath2 gyr old stellar population, perhaps formed in tidal structures that have now fallen back onto the galaxy, as has previously been suggested for the x - ray point sources in this galaxy, or coming from a more recent merger / accretion with a red galaxy.
the use of pns for extragalactic distance determinations is not necessarily compromised, but their use as dynamical tracers of dark halos will require deep observations and careful analysis of large pn samples. | astro-ph0510499 |
Introduction
microarray technology is an important tool to monitor gene - expression in bio - medical studies @xcite. a common experimental design is to compare two sets of samples with different phenotypes, e.g. diseased and normal tissue, with the goal of discovering differentially expressed genes @xcite. statistical testing procedures, such as such as the t - test and significance analysis of microarrays @xcite, have been extensively studied and widely used. subsequently, multiple testing corrections are usually applied @xcite. a comprehensive review of such approaches are presented in @xcite. differential expression analysis based on univariate statistical tests has several well - known limitations. first, due to the low sample size, high dimensionality and the noisy nature of microarray data, individual genes may not meet the threshold for statistical significance after a correction for multiple hypotheses testing @xcite. second, the lists of differentially expressed genes discovered from different studies on the same phenotype have little overlap @xcite. these limitations motivated the creation of gene set enrichment analysis gsea @xcite, which discovers collections of genes, for example, known biological pathways @xcite, that show moderate but coordinated differentiation. for example, subramanian and tamayo et al. @xcite report that the p53 hypoxial pathway contains many genes that show moderate differentiation between two lung cancer sample groups with different phenotypes. although the genes in the pathway are not individually significant after multiple hypothesis correction @xcite, the pathway is. for those familiar with gsea and its output, figure [fig : pnas2005toy] shows the gsea results for the p53 hypoxial pathway. gsea also has the advantages of better interpretability and better consistency between the results obtained by different studies on the same phenotype @xcite. ackermann and strimmer presented a comprehensive review of different gsea variations in @xcite. unfortunately, gsea and related techniques may be ineffective if many individual genes in a phenotype - related gene set have weak discriminative power. a potential solution to this problem is to search for combinations of genes that are highly differentiating even when individual genes are not. for this approach, the targets are groups of genes that show much stronger discriminative power when combined together @xcite. for example, figure [fig : gb2005toy1] illustrates one type of differentially expressed gene combination discovered in @xcite. the two genes have weak individual differentiation indicated by the overlapping class symbols on both the two axes. in contrast, these two genes are highly discriminative in a joint manner indicated by the different correlation structure in the two - dimensional plot, i.e. they are correlated along the blue and red dashed line respectively in the triangle and circle class. such a joint differentiation may indicate that the interaction of the two genes is associated with the phenotypes even though the two genes, individually, are not. figure [fig : gb2005toy2] illustrates another type of phenotype - associated gene combination discovered in @xcite, usually named differential coexpression @xcite, in which the correlation of the two genes are high in one class but much lower in the other class. as discussed in @xcite, existing multivariate tests such as hotelling s @xmath0 @xcite, dempster s t1 @xcite are not suitable to detect such ` complementary'gene combinations because they only screen for differences in the multivariate mean vectors, and thus will favor pairs that consist of genes with strong marginal effects by themselves but not the genes like the four in fig. [fig : gb2005toy12]. for clarification, we use differential gene combination search (denoted as dgcs) to refer to the multivariate data analyses that are designed to detect the complementarity of different genes, rather than those designed to model the correlation structure of different genes (such as hotelling s @xmath0 and dempster s t1 test). a variety of other dgcs measures for complementary gene combination search are proposed for gene pairs @xcite in addition to the two illustrated in figure [fig : gb2005toy12]. several measures are designed for higher - order gene combination beyond pairs @xcite. these approaches can provide biological insights beyond univariate gene analysis as shown in @xcite. the limitations of gsea and the capabilities of dgcs motivate a gsea approach using gene combinations in which the score of a gene set is based on both the scores of individual genes in the set and the scores from the gene combinations in which these genes participate. unfortunately, gene combination techniques have not been used with the gsea approach in any significant way because of two key challenges. 1. * finding a technique to reduce the vast number of gene combinations. * there are exponentially more gene combinations than individual genes, i.e. in addition to the @xmath1 univariate genes, there are @xmath2 gene - pairs, @xmath3 gene - triplets, etc. many variations of gsea are based on a ranked list of the @xmath1 individual genes as illustrated in figure [fig : pnas2005toy]. including combinations in the ranked list might work for size-2 combinations @xcite, but would not be feasible for handling gene combinations of larger sizes. furthermore, this explosion in the number of gene combinations negatively impacts false discovery rates. thus, by adding so many gene combinations, we run the risk that neither groups of genes nor individual genes will show statistically significant differentiation. * combining results from the heterogeneous measures used to score different size gene combinations. * furthermore, because a gene can be associated with the phenotype either as an univariate variable or together with other genes as a combination, the importance of a gene set should be based on both the univariate gene scores and the gene combination based scores of its set members. however, different measures have a different nature, scale and significance, and thus are not directly comparable (to be detailed in section [sec : score2pv]). indeed, differences exist even between gene combinations of the same measure but of different sizes. therefore, the challenge lies in how to design a framework to combine different measures (a univariate measure plus one or more dgcs measures) together within the gsea framework. to the best of our knowledge, no existing work has sufficiently addressed these two challenges, although recent work presented in @xcite have made initial efforts at adding gsea capabilities to gene combination search. more specifically, two approaches are proposed in @xcite to help the study of a specific type of size-2 differential combinations as illustrated in [fig : gb2005toy2]. the experiments in these two studies provide some evidence about the benefits of the integration. however, a more general framework is needed that can also handle other types of size-2 differential combinations as illustrated in figure [fig : gb2005toy1], higher order differential combinations (e.g. sdc@xcite and the n - statistic @xcite), and multiple types of differential combinations. * contributions : * in this paper, we propose a general framework to address the above challenges for the effective integration of dgcs and gsea. specific contributions are as follows : 1. * a gene - combination - to - gene score summarization procedure (procedure _ a _) that is designed to handle the exponentially increasing number of gene combinations. * first, for a given gene combination measure and a certain @xmath4, the score of a size-@xmath4 combination is partitioned into @xmath4 equal parts which are assigned to each of the @xmath4 genes in the combination. because each gene can participate in up to @xmath5 size-@xmath4 combinations, each gene will be assigned with a score from each of these combinations. secondly, an aggregation statistic, e.g., maximum absolute value is used to summarize the different scores for a gene. with such a procedure, scores for all the size-@xmath4 gene combinations are summarized to @xmath1 scores for @xmath1 genes. this procedure can effectively retain the @xmath6 length of the ranked list while handling gene combinations of size-@xmath4 (@xmath7). * a score - to - pvalue transformation and summarization procedure (procedure _ b _) that is designed to integrate the scores contributed (in procedure @xmath8) from different gene combination measures and from gene combinations of different sizes. * the transformation is based on p - values obtained from scores derived from phenotype permutations. such a transformation enables the comparison of scores from different measures (either univariate or gene combination measures) and scores from the gene combinations of different sizes. subsequently, among all the p - values of a gene, the best is used as an integrated score of statistical significance. * integration of the above two procedures with gsea * more specifically, after procedures _ a _ and _ b _, each gene has a single integrated score. unlike traditional univariate scores, these @xmath1 integrated scores are based on both the univariate statistic and the gene combination measures. for the type of gsea variations that depend on phenotype permutation test, @xmath9 lists of @xmath1 integrative scores are computed, one for the real class labels and the other for the @xmath10 permutations. for the type of gsea variations that are based on gene - set permutation test, only the list of integrated scores for the real class labels are needed. an independent matlab implementation of the proposed framework is available for download, which allows most existing gsea frameworks @xcite to directly utilize the proposed framework to handle gene combinations with almost - zero modification. * experimental results that illustrate the effectiveness of the proposed framework. * we integrated three gene combination measures and the gsea approach presented in @xcite and produced experimental results from four gene expression datasets. these results demonstrate that the integrative framework can discover gene sets that would have been missed without the consideration of gene combinations. this includes statistically significant gene sets with moderate differential gene combinations whose individual genes have very weak discriminative power. thus, a gene combination assisted gsea approach can improve traditional gsea approaches by discovering additional disease - associated gene sets. indeed, the integrative approach also improve traditional dgcs since most gene combinations are not statistically significant by themselves. furthermore, we also show that the biologically relevant gene sets discovered by the integrative framework share several common biological processes and improve the consistency of the results among the three lung cancer data sets. * overview : * the rest of the paper is organized as follow. in section [sec : measures], we describe three gene combination measures used in the following discussion and experiments. in section [sec : methods], we present the technical details of the two procedures of the general integrative framework. experimental design and results are presented in section [sec : exp], followed by conclusions and discussions in section [sec : discssion].
Differential gene combination measures
in this section, we describe three dgcs measures for use in the following discussion and experiments. let @xmath11 and @xmath12 be two phenotypic classes of samples of size @xmath13 and @xmath14 respectively. for each sample in @xmath8 and @xmath15, we have the expression value of @xmath1 genes @xmath16. first, we have the following two measures (denoted as @xmath17 and @xmath18) defined for a pair of genes as presented in @xcite : @xmath19 @xmath20 where @xmath21 and @xmath22 are two genes, and @xmath23 represents the correlation of @xmath21 and @xmath22 over the samples in set @xmath24. as discussed in @xcite, @xmath17 and @xmath18 can detect the joint differential expression of two genes as illustrated in figure [fig : gb2005toy1] and figure [fig : gb2005toy2] respectively. @xmath17 and @xmath18 are used as two representative measures for gene pairs. other options for gene combination measures for gene pairs have been investigated in @xcite. we use the subspace differential coexpression measure (denoted as @xmath25) proposed in @xcite as the representative for measures for size-@xmath4 gene combinations, where @xmath4 can be any integer (@xmath7). @xmath26 where @xmath27 is a set of genes such that @xmath28 and @xmath29. @xmath30 and @xmath31 respectively represent the fraction of samples in @xmath8 and @xmath15 over which the genes in @xmath27 are coexpressed. @xmath25 is a generalization of @xmath18 for detecting the differential coexpression of @xmath4 genes (@xmath7), i.e. the @xmath4 genes are highly coexpressed over many samples in one class but over far fewer samples in the other. other options for size - k combinations include the _ n - statistic _ @xcite, _ supmaxpair _ @xcite, etc. signal - to - noise ratio (denoted as @xmath32) is used as the representative of traditional univariate statistics as in @xcite. @xmath33 where @xmath34 and @xmath35 are the mean expression of @xmath21 in class @xmath8 and @xmath15 respectively, and @xmath36 and @xmath37 are the standard deviation of the expression of @xmath21 in class @xmath8 and @xmath15 respectively. many other univariate statistics can be found in @xcite. in this paper, these four measures are used as representatives of each category for the illustration of the proposed integrative framework. however, the framework is general enough to handle other measures from each of these categories.
Methods
in section [sec : intro], we motivated the integration of dgcs with gsea, discussed two challenges associated with this integration, and briefly described two main procedures in the proposed framework. in this section, we present the technical details of the two procedures and their integration with gsea. there are two steps in procedure @xmath8. in step @xmath38, for each dgcs measure and each size-@xmath4 gene combination, its score is divided into @xmath4 equal parts and assigned to each of the @xmath4 genes in the combination. in step @xmath39, the scores assigned to a gene from all the size - k combinations in which the gene participates are summarized into a single score by an aggregation functions such as @xmath40. note that, for most univariate statistic and dgcs measures which can be either positive or negative (e.g. the four measures described in section [sec : measures]), the maximum is taken over the absolute values of the scores, and the sign of the score with the highest absolute value is recorded for later use. other simple statistics such as mean or median, or sophisticated ones such as weighted summation @xcite can also be used. since the focus of this paper is the overall integrative framework, we use @xmath40 for simplicity. we provide a conceptual example of procedure @xmath8 for a gene @xmath41 with a certain dgcs measure @xmath17. this example considers gene combinations up to size @xmath42 for illustration purpose. the gene is associated with scores assigned from gene combinations of size 2, 3 and 4 (denoted as @xmath43, @xmath44 and @xmath45 respectively) in which @xmath41 participates. in step @xmath39, the scores from @xmath43, @xmath44 and @xmath45 are summarized by three maximum values, respectively. please refer to the appendix section for the illustration of this example. procedure @xmath8 serves as a general approach to summarize the @xmath46 scores of all the size-@xmath4 combinations into @xmath1 scores for the @xmath1 genes. if we want to integrate gsea with one dgcs measure and a specific size-@xmath4, procedure @xmath8 by itself can enable most existing variations of gsea to search, with almost - zero modification, for statistically significant gene sets with moderate but coordinated gene combinations of size-@xmath4. such a gsea approach can collectively consider the gene combinations affiliated with a gene set, and may provide better statistical power and better interpretability for dgcs, as will be shown in the experiments. the hypothesis tested when one dgcs measure, say @xmath47, is integrated with gsea (by procedure @xmath8) is that, whether a gene set includes significantly many genes with highly positive (or highly negative) combination - based scores measured by @xmath47. an extended hypothesis can be whether a gene set includes significantly many genes with highly positive (or highly negative) scores, either univariate or combination - based scores measured by different dgcs measures. the biological motivation of this extended hypothesis is that, a gene can be associated with the phenotype either as an univariate variable or together with other genes as a combination. to test this extended hypothesis, we design a second procedure (_ b _) that can integrate the scores of a gene from different measures. before describing the steps in this procedure. we first discuss in detail the challenges of integrating heterogeneous scores from different dgcs measures and combinations of different sizes. 1. * the different nature of different measures * : different measures are designed to capture different aspects of the discriminative power of a gene or a gene combination between the two phenotypic classes. signal - to - noise ratio (@xmath32), a univariate gene - level statistic, measures the difference between the means of the expression of a gene in the two classes. in contrast, @xmath18, a differential coexpression measure for a pair of genes describes the difference of the correlations of a gene - pair in the two classes. thus, for a gene, the score of itself measured by @xmath32 and the score assigned and summarized from the @xmath48 size-2 gene combinations measured by @xmath18 are not directly comparable. similarly, the scores of different dgcs measures can also have a different nature, e.g. @xmath17 and @xmath18 as illustrated in figure [fig : gb2005toy12]. * the different scales of different measures * : different measures also have different ranges of values. for example, the range of @xmath32, @xmath17, @xmath18, and @xmath25 are @xmath49 $], @xmath50 $], @xmath51 $] and @xmath52 $] respectively. thus, they are not directly comparable. * differences in significance between different measures * : even after we normalize the scores of different measures to a single range, say @xmath52 $], they are still not comparable because the scores of different measures have different statistical significance. for example, a normalized @xmath32 score of @xmath53 may be less significant than a normalized @xmath17 score of 0.5, if there are many genes with normalized @xmath32 score greater than @xmath53 in the permutation test @xcite, but very few genes with normalized @xmath17 score greater than @xmath54 in the permutation test. note that, such differences in statistical significance also exists between gene combinations of different sizes, even for the same measure. take the subspace differential coexpression measure @xmath25 as an example. a score of @xmath54 for a size-2 combination may not be as significant as a score of @xmath54 for a size-3 combination as discussed in @xcite. to handle the above heterogeneity, we propose a score - to - pvalue transformation and summarization procedure that can enable the comparison and integration of the scores of different measures and combinations of different sizes. there are three major steps in procedure @xmath15. in procedure @xmath15 (score - to - pvalue transformation) for gene @xmath21 and measure @xmath18.,scaledwidth=95.0%] consider a concrete example. for a gene @xmath21 and a measure @xmath18, procedure @xmath8 computes a single summarized score. in this step, the original phenotype class labels are permutated say @xmath56 times, and for each permutation, the same procedure @xmath8 is applied, and a corresponding score for @xmath21 and @xmath18 is computed. we denote the score of @xmath21 and @xmath18 summarized with the original label as @xmath57, where @xmath58 is the gene index, and @xmath59 indicates the measure and @xmath60 means it is the score based on the original label. similarly, we denote the scores computed in each of the permutation as @xmath61, where @xmath62. these @xmath63 scores are organized in the table on the left in figure [fig : tabmerge2measures]. the @xmath56 scores computed in the @xmath56 permutations can be considered as the null - distribution for gene @xmath21 and measure @xmath18, and a p - value can be estimated for @xmath57. specifically, if @xmath57 is positive, the p - value is the ratio of the number of scores that are greater or equal to @xmath57 and the number of scores that are positive. similarly if @xmath57 is negative, the p - value is computed as the ratio of the number of scores which are less or equal to @xmath57 and the number of scores which are negative. note that, such a score - to - pvalue transformation is done for both @xmath57 and each of @xmath61 (@xmath62), if the gsea approach to be integrated is based on phenotype permutation test @xcite. otherwise, only @xmath57 needs to be transformed to p - value and will be used by the gsea approaches that are based on gene - set permutation @xcite. in this paper, we illustrate the proposed framework using the gsea approach presented by subramanian and tamayo et al. @xcite which is based on phenotype permutation test. essentially, step @xmath55 transforms the heterogeneous scores of a gene measured by different measures into their corresponding significance values, which are comparable to each other although their original values are not. suppose that there are @xmath64 different measures to be integrated, one of which is a univariate statistic, and the others are different dgcs measures for which we consider combinations of sizes up to @xmath65. after step @xmath55, each gene has a p - value for the univariate measure and up to @xmath65 p - values for each size of gene combination for each measure. in step @xmath59, the best transformed p - value] p - value associated with a gene is selected as the integrated significance. essentially, procedure @xmath15 integrates the scores of different dgcs measures for a gene and the univariate statistic of the gene into a single p - value. such a statistical significance - based integration of heterogeneous scores enables the comparison and thus the ranking of all the @xmath1 genes. however, this ranked list does not maintain the original directionality of the integrated scores of each gene. in particular, most univariate statistics and dgcs measures (e.g. all the four measures described in section [sec : measures]) can be either positive or negative. such directionality information is lost in step @xmath55 and @xmath59 because the p - value is non - negative. next, we describe a third step to maintain the directionality in the integration. in the simple case, the measures to be integrated capture the same type of differentiation between the two phenotype classes, e.g. @xmath18 and @xmath25. suppose there are two genes @xmath21 and @xmath22, whose integrated p - values are transformed respectively from two scores measured by @xmath18 and @xmath25 in step @xmath59. the signs of these two scores are comparable to each other, because both @xmath18 and @xmath25 capture the change of coexpression of a combination of genes. thus, we simply use the signs of these two scores as the signs associated with the integrated p - values of @xmath21 and @xmath22. similarly, we associate a sign to all the @xmath1 integrated p - values. and these @xmath1 p - values with associated signs can be used to rank the @xmath1 genes based on their significance as well as their direction of differentiation, i.e. p - values associated with positive signs are ranked with descending significance, and afterwards, p - values associated with negative signs are ranked with increasing significance. in the other case, if the measures to be integrated capture different types of differentiation between the two phenotype classes, the directionality can not be fully maintained. for example, suppose there are two genes @xmath21 and @xmath22, whose integrated p - values are transformed respectively from two scores measured by @xmath32 and @xmath18 in step @xmath59. the signs of these two scores are not comparable, because @xmath32 captures the change of mean expression, and @xmath18 captures the change of coexpression of a combination of genes. specifically, up - regulation of @xmath21 can be associated with either high or low coexpression of another gene - combination in which @xmath22 participates. thus, it is not reasonable to follow the same strategy to associate signs to the @xmath1 integrated p - values. if we know the correspondence of the signs of different genes in advance, e.g. the up - regulation of @xmath41 is associated with the low coexpression of @xmath67 and @xmath68, then the signs can be maintained. however, because it is not realistic to assume such prior knowledge, we propose the following heuristic approach which has proved a workable solution for our initial experiments. specifically, since the focus of step @xmath15 is to integrate different dgcs measures in addition to the univariate statistic @xmath32, we considered @xmath32 as the base measure. for the integrated p - values that are transformed from scores measured by @xmath32 in step @xmath59 (say there are @xmath69 of them), we use the signs of these @xmath69 @xmath32 scores for the @xmath69 integrated p - values. for the signs of the other @xmath70 genes, we assign positive signs to all of them once and negative signs to all of them a second time. correspondingly, we have two ranked lists similar to the simple case described above. note that, if the directionality of differential measures can be preserved, the power of this approach will be enhanced. to deal with the situation where signs are not comparable, other approaches will be explored. from the above description of procedure @xmath8 and @xmath15, we know that, if only one dgcs measure is used in the gsea framework, only procedure @xmath8 is needed. if one or multiple dgcs measures are integrated together with the univariate statistic @xmath32 in the gsea framework, procedure @xmath15 is needed in addition. in the first case, the integrative framework outputs a ranked list of @xmath1 scores with associated signs for the original class label, and @xmath56 lists corresponding to the @xmath56 permutation tests. in the second case, we have two sets of @xmath63 lists respectively for the two rounds of maintaining directionality in step @xmath66 in procedure @xmath15. in either case, the @xmath63 ranked lists along with the appropriate parameter settings and specification of gene sets can be used to run gsea. the only modification to gsea is the elimination of the initial gsea step to generate the scores, simulated and actual, that measure the level of differentiation between genes across different phenotypes. the proposed integrative framework is implemented as a matlab function (available at http://vk.cs.umn.edu/icg/), independently from the gsea framework to be integrated in this paper @xcite. as summarized by ackermann and strimmer @xcite, hundreds of variations of gsea are being used by different research groups. this independently implemented integrative framework can be easily applied to other variations of gsea. in our experiments, in order to have a fair comparison, we transform the @xmath63 ranked lists into the exact sample distribution as the original lists corresponding to @xmath71gsea. specifically, we only use the ranking information in the @xmath63 integrated ranked lists and map to them to the values in the original lists based on @xmath71gsea. essentially, the values in the ranked list passed to the gsea framework are exactly the same among @xmath71gsea, @xmath72gsea, @xmath73gsea, @xmath74gsea and @xmath75gsea, while the only difference is that the @xmath1 genes have different ranks in the lists. such a mapping ensures that the additionally discovered gene sets are because of the integration of gene - combinations in addition to univariate statistic, rather than simply the different value distributions in the @xmath63 lists.
Results
in this section, we present the experimental design and results for the evaluation of the proposed integrative framework. we first provide a brief description of the data sets and parameters used in the experiments. second, we describe and discuss the comparative experiments to study whether the integration of dgcs and gsea (denoted as dgcs@xmath76gsea) improves both dgcs and gsea. the two major evaluation criteria are the statistical power to discover (additional) significant results, and the consistency of the results across different datasets for the study of the same phenotype classes. the four datasets used in the experiments are described as follows : 1. three lung cancer datasets respectively denoted as boston @xcite, michigan @xcite and standford @xcite : all the three data sets consist of gene - expression profiles in tumor samples from respectively 62, 86 and 24 patients with lung adenocarcinomas and provide clinical outcomes (classified as `` good '' or `` poor '' outcome). the two phenotypic classes in these three datasets are denoted as @xmath8 and @xmath24 as in @xcite. 2. a data set from the nci-60 collection of cancer cell lines for the study of p53 status @xcite (denoted as @xmath77 data set) : the mutational status of the p53 gene has been reported for 50 of the nci-60 cell lines, with 17 being classified as normal and 33 as carrying mutations in the gene. the two phenotypic classes in this dataset are denoted as @xmath78 and @xmath79 as in @xcite. all four datasets were downloaded from the gsea website@xcite, and were already preprocessed as described in the supplementary file of @xcite. for all four data sets, we use the gene sets from @xmath43 in msigdb@xmath80 as in @xcite, as well as the same parameters. we consider one univariate statistic (@xmath32), and three gene - combination measures (@xmath17, @xmath18 and @xmath25) in our experiments. these four measures are described in section [sec : measures]. @xmath17 and @xmath18 are defined only for size-2 combinations. for @xmath25, we considered gene - combinations of size-@xmath59 and size-@xmath66 for the illustration of concept..number of with fdr less than @xmath81 discovered from the four data sets by each combination measure [cols="<,<,<,<,<",options="header ",] in this section, we study whether the question (q1) of whether integration of dgcs and gsea can improve traditional dgcs. for this comparison, we consider the integration of dgcs and gsea as a gsea - assisted dgcs approach. we first apply the traditional dgcs approaches on the four datasets to find statistically significant gene - combinations. we denote the three dgcs approaches respectively with the names of the three measures, i.e. @xmath17, @xmath18 and @xmath25. second, we apply the integrative framework, in which gsea is integrated respectively with the three dgcs measures, to find statistically significant gene sets with moderate but coordinated differential gene - combinations. we denote the three instances of the integrative approach respectively as @xmath82gsea, @xmath83gsea and @xmath84gsea. then, we compare the results of @xmath17, @xmath18 and @xmath25, respectively with the results of @xmath82gsea, @xmath83gsea and @xmath84gsea. table 1 lists the number of statistically significant gene combinations discovered respectively by the three measures on each of the four datasets, with an fdr threshold of @xmath81. table 1 lists the number of statistically significant gene sets discovered by integrating gsea respectively with the three dgcs measures on each of the four datasets, also with the same fdr threshold of @xmath81. three major observations can be made by comparing the two tables : table 1 shows that, in most cases, traditional dgcs discovers very few (less than 3) statistically significant gene combinations (although @xmath18 and @xmath25 have @xmath85 and @xmath86 gene - combinations on the boston data set, none of them have fdr lower than @xmath87). in contrast, table 2 shows that the integration of gsea with the three combination measures discover multiple significant gene sets in most of the cases. this difference implies that the discovered statistically significant gene sets include many moderate but coordinated differential gene combinations, even though the combinations are not significant by themselves as shown in table 1. this comparison demonstrates that traditional dgcs, similar to univariate gene analysis, has limited statistical power, and dgcs@xmath76gsea can increase that power. we further compare dgcs and dgcs@xmath76gsea by studying the consistency of their results on the first three data sets that are all from lung cancer studies, as done in @xcite. for dgcs, @xmath17 discovered @xmath59 genes on michigan but nothing from boston and stanford ; @xmath18 discovered @xmath85 combinations on boston but only 1 and 2 from michigan and stanford, respectively, and there are no common ones between the 645, 1, and 2 gene combinations ; @xmath25 discovered @xmath86 genes on boston but only @xmath55 gene on michigan and nothing from stanford, and the 10 and 1 combinations do not overlap. the inconsistent results make the follow - up biological interpretation very difficult. in contrast, when the three dgcs measures are integrated with gsea, several consistent themes can be observed : (i) apoptosis related pathways (marked by @xmath8 in table 2) : @xmath82gsea discovered four gene sets on boston, three of which are known to be closely related to cancer and specifically to apoptosis, i.e. _ nfkbpathway _, _ st - gaq - pathway _ and _ tnf - pathway_. this apoptosis theme is shared by the gene sets discovered by @xmath17-gsea from michigan and stanford, i.e. _ monocyte - ad - pathway _, _ hivnefpathway _, _ deathpathway _ and _ caspasepathway_. these apoptosis related pathways are enriched with the lung cancer samples with good outcome, which makes sense biologically and also corresponds to the proliferation theme supported by the gene sets enriched with the samples with poor outcome as reported in @xcite. several other examples of the result consistency, as indicated by other superscripts in table 2, are in the technical report. this comparison demonstrates that traditional dgcs, like univariate gene analysis, has poor result consistency across the three lung cancer data sets, and dgcs@xmath76gsea can improve its consistency by integrating dgcs measures with gsea. the number of significant gene sets discovered by the three versions of gsea varies, i.e. @xmath82gsea and @xmath83gsea discovered a bit larger number of significant gene sets than @xmath84gsea. however, @xmath84gsea still discovered several gene sets that are not discovered by @xmath82gsea or @xmath83gsea, e.g. one gene set from the michigan data set and three from the stanford data set. this indicates that @xmath82gsea, @xmath83gsea and @xmath84gsea have complementary perspectives, i.e. different combination measures capture different aspects of the difference between the phenotype classes (recall the two types of combinations in figure [fig : gb2005toy12]). this also demonstrates the proposed framework is general enough to integrate any type of dgcs with gsea. in this section, we want to answer the question (q2) of whether the integration of dgcs and gsea can improve traditional gsea. for this comparison, we consider the integration of dgcs and gsea as a dgcs - assisted gsea approach. we design three sets of comparisons. firstly, we compare the traditional univariate - statistic based gsea (denoted as @xmath71gsea) with the integrative framework where one gene - combinations measure is used instead of @xmath32. specifically, we compare the gene sets discovered by @xmath71gsea with the gene sets discovered by @xmath82gsea, @xmath83gsea and @xmath84gsea. then, we compare @xmath71gsea with the integrative framework where one gene - combinations measure is used in addition to @xmath32, i.e. @xmath72gsea, @xmath73gsea and @xmath74gsea. furthermore, we also study the integration of multiple gene - combinations measure in addition to @xmath32, e.g. @xmath75gsea. figure [fig : bigtable12312009half] displays the statistically significant gene sets discovered with different (combinations of) measures respectively from the four datasets. an fdr threshold of @xmath81 is used as in @xcite for comparison purpose. the results presented in @xcite are exactly reproduced, i.e. the gene sets listed in the rows corresponding to @xmath71gsea. in each of these four figures, we consider the traditional univariate - statistic based gsea (@xmath88gsea) as the baseline, and compare it with the rows corresponding to @xmath89gsea, @xmath90gsea, @xmath91gsea, @xmath72gsea, @xmath73gsea, @xmath74gsea and @xmath92gsea. from these comparisons, the following observations can be made. gsea) is considered as the baseline. for the other rows, we only list a gene set if it is only discovered by the integrative approach (with bolded name), or it has a non - trivially decreased fdr when it is discovered by the integrative approach (with bolded fdr). please refer to the appendix section for the complete tables.] first, we compare the rows corresponding to @xmath89gsea, @xmath90gsea, @xmath91gsea with the rows corresponding to @xmath88gsea. we bolded the additional gene sets that are only discovered by @xmath89gsea, @xmath90gsea, @xmath91gsea. for example, with @xmath71gsea, no statistically significant gene sets have been enriched with class a in the boston data set. in contrast, @xmath82gsea discovered @xmath42 gene sets, three out of which (discussed in @xmath93) are related to apoptosis which is consistent with the results on michigan and stanford. on the michigan dataset, @xmath83gsea discovered a gene set _ beta - alanine - metabolism _ that is not discovered by @xmath71gsea. this gene set is related to the responses of hypoxia, which is consistent with the results on boston and stanford. it is worth noting that, although most studies did not report statistically significant gene sets on the stanford dataset due to the very small sample size, @xmath89gsea, @xmath90gsea, @xmath91gsea respectively discovered 4, 4 and 3 significant gene sets. these additional gene sets were discovered because the three dgcs measures capture different types of the differentiation between the two phenotype classes, compared to the traditional univariate differential expression - based gsea. second, we compare the rows corresponding to @xmath72gsea, @xmath73gsea, @xmath74gsea with the rows corresponding to @xmath88gsea. we bolded the additional gene sets that are only discovered by the integrative approach. for example, on the boston data set, @xmath32 based gsea discovered 8 gene sets. in addition, @xmath72gsea discovered the _ gene set, and @xmath73gsea discovered the _ p53-signaling _ gene set. both ubiquitin - proteasome pathway and p53-signaling pathway are well - known cancer - related pathways that are also specifically related to lung cancer @xcite. (additional examples are in the technical report.) the gene sets that are discovered by dgcs - assisted gsea but not by @xmath94-gsea illustrate the benefits of using dgcs to assist gsea. next, we also observed that integrating multiple dgcs measures can further discover statistically significant gene sets. for illustration purpose, we compare the rows corresponding to @xmath72gsea, @xmath73gsea, @xmath74gsea with the rows corresponding to @xmath75gsea. @xmath75gsea discovers the _ g2pathway _ gene set and the _ gsk3pathway _ gene set, respectively from the boston and the michigan dataset. neither of these two pathways are discovered by @xmath88gsea, @xmath72gsea, @xmath73gsea and @xmath74gsea. the curated gene set _ g2pathway _ contains the genes related to the g2/m transition, which is shown to be regulated by p53 @xcite, a well - known cancer - related gene. the curated gene set _ gsk3pathway _ is the signaling pathway of gsk-3-@xmath95, which has been shown to be related to different types of cancer@xcite. these two cancer - related pathways are discovered by @xmath75gsea but not by @xmath88gsea, @xmath72gsea, @xmath73gsea and @xmath74gsea. this indicates that different members of these two pathways are differential between the two phenotype groups in different manners, i.e. the differentiation of some genes is captured by @xmath32, some by @xmath17, some by @xmath18 and some by @xmath25. these two pathways can be discovered to be statistically significant only when these measures are used together in the integrative framework. this demonstrates the benefits of the proposed framework for integrating multiple dgcs measures with a univariate measure. it is worth noting that, the gene sets discovered by the integrative framework with multiple measures are not necessarily a superset of those discovered by integrating each individual measure with gsea since, when different dgcs measures are integrated with gsea, the null - hypotheses tested in the gsea framework are correspondingly different. the highlight of the integrative framework is that, additional gene sets can be discovered when different dgcs measures are used to assist the traditional univariate statistic - based gsea. in practice, these different versions of gsea should be used collectively. : even when a gene set is discovered both before and after a dgcs measure is integrated into the framework, we can observe several interesting cases where the fdr of a gene set becomes much lower after the integration. we bolded the fdrs that significantly decreased when they are discovered by the integrative approach. for example, @xmath75-gsea, in which @xmath32, @xmath17, @xmath18 and @xmath25 are integrated together, discovers _ p53hypoxialpathway _ with an much lower fdr of @xmath96, two - order lower than @xmath32-gsea. this example indicates that several members of _ p53hypoxialpathway _ have weak individual differentiation measured by @xmath32, but have more significant differentiation when they are measured by @xmath25. this and other similar examples demonstrates the benefits of the proposed framework for integrating multiple dgcs measures. as presented in @xcite, @xmath71gsea discovered @xmath97 and @xmath98 gene sets respectively from the boston and michigan data sets, and 5 of the 8 in boston and 6 of the 11 in michigan are common. the three unmatched gene sets that are discovered in boston but not in michigan are _ glut - down _, _ leu - down _ and _ cellcyclecheckpoint_. interestingly, the latter two are discovered from both the boston and the michigan data sets by @xmath72gsea. such observations suggest that dgcs - assisted gsea also provides new insights to the consistency between different data sets. because different combinations of measures are used in the integrative framework, additional issues of multiple hypothesis testing arise, even though multiple hypothesis testing has been addressed for each measure via the phenotype permutation test procedure in the gsea framework proposed in @xcite. to investigate this, we designed experiments with 4 of the @xmath99@xmath100 possibilities of integrations, i.e. @xmath72gsea, @xmath73gsea, @xmath74gsea and @xmath75gsea. even using a collective (meta - level) multiple hypothesis correction, many discovered gene sets would still be significant. for examples, @xmath75gsea discovers _ p53hypoxialpathway _ from the boston data set with a low fdr of @xmath96, and @xmath75gsea discovers _ deathpathway _ from the michigan data set with a lower fdr of @xmath101. we also did additional permutation tests, in which we generate random gene sets with the same sizes as the sets in msigdb @xmath43, and do the same set of experiments as shown in figure [fig : bigtable12312009half]. the fdr values of the random gene sets computed in the integrative framework are mostly insignificant (higher than @xmath81).
Discussion
in this paper we motivated the integration of differential gene - combination search and gene set enrichment analysis for bi - directional benefits on both them. we proposed a general integrative framework that can handle gene - combinations of different sizes (@xmath7) and different gene - combination measures in addition to an univariate statistic used in traditional gsea. the experimental results demonstrated that, on one hand, gsea - assisted dgcs has better statistical power and result consistency than traditional dgcs. on the other hand, dgcs - assisted gsea can discover additional statistically significant gene sets that are ignored by traditional gsea and further improve the result consistency of the traditional gsea. the proposed framework can be extended in several ways. different variations of gsea will be considered. along these lines, we note that the proposed integrative framework is general enough to integrate most existing variations of gsea approaches summarized in @xcite with minimal amount of modification. also, it should be possible to integrate dgcs and gene - subnetwork discovery. both gsea and gene - subnetwork discovery @xcite can discover collections of genes, either known gene sets @xcite or subnetworks in a molecular network (e.g. protein interaction network), that show moderate but coordinated differentiation. in this paper, we integrate dgcs and gsea as an illustration of the general framework for integrating scores from different gene - combination measures and gene - combinations of different sizes, in addition to the traditional univariate statistic, but the same framework also applies to the integration of dgcs and gene subnetwork discovery. another direction is the use of this framework for the analysis of (gwas) snp data, by following the methodology proposed in recently work on pathway / network based analysis of gwas datasets @xcite. finally, it may be possible to use constraints on gene - combinations to improve our framework. in procedure @xmath8, for each gene - combination measure and an integer @xmath4, the score of a gene is assigned from all the @xmath5 possible gene - combinations. a further extension of procedure @xmath8 is to only consider the gene combinations, in which the @xmath4 genes appear in a common gene set, e.g. a pathway. such gene - set - based constraints may better control false positive gene combinations and improve the statistical power of the whole integrative framework.
Appendix
due to the space limit, table 2 and figure [fig : bigtable12312009half] are both summarized from the four complete tables that are available at http://vk.cs.umn.edu / icg/. specifically, table 2 is a high - level summary of the number of gene sets discovered and the biological processes associated with each of the gene sets. in figure [fig : bigtable12312009half], we listed the complete results for @xmath71gsea (the baseline), while for the other rows, we only list a gene set if it is only discovered by the integrative approach (with bolded name), or it has a non - trivially decreased fdr when it is discovered by the integrative approach (with bolded fdr). g. fang, g. pandey, m. gupta, m. steinbach, and v. kumar. mining low - support discriminative patterns from dense and high - dimensional data. technical report 09 - 011, department of computer science, university of minnesota, 2009. k. wang, m. narayanan, h. zhong, m. tompa, e. e. schadt, and j. zhu. meta - analysis of inter - species liver co - expression networks elucidates traits associated with common human diseases., 5(12):e1000616, 2009. | gene set enrichment analysis (gsea) and its variations aim to discover collections of genes that show moderate but coordinated differences in expression.
however, such techniques may be ineffective if many individual genes in a phenotype - related gene set have weak discriminative power.
a potential solution is to search for combinations of genes that are highly differentiating even when individual genes are not.
although such techniques have been developed, these approaches have not been used with gsea to any significant degree because of the large number of potential gene combinations and the heterogeneity of measures that assess the differentiation provided by gene groups of different sizes. to integrate the search for differentiating gene combinations and gsea
, we propose a general framework with two key components : (a) a procedure that reduces the number of scores to be handled by gsea to the number of genes by summarizing the scores of the gene combinations involving a particular gene in a single score, and (b) a procedure to integrate the heterogeneous scores from combinations of different sizes and from different gene combination measures by mapping the scores to p - values.
experiments on four gene expression data sets demonstrate that the integration of gsea and gene combination search can enhance the power of traditional gsea by discovering gene sets that include genes with weak individual differentiation but strong joint discriminative power.
also, gene sets discovered by the integrative framework share several common biological processes and improve the consistency of the results among three lung cancer data sets. | 1101.3474 |
Introduction
jets are found in accreting systems throughout the visible universe. for stellar - mass accretors, the evolution of such jets occurs on human timescales, and can be probed by resolved monitoring observations. x - ray binaries are one such class of objects, in which two distinct types of jets are observed, with a clear connection between the x - ray state of the source and the observed radio emission. from the flat radio spectra seen in the hard x - ray state, the presence of steady, compact, partially self - absorbed outflows is inferred, which have been directly resolved in two black hole (bh) systems ((*??? * dhawan et al. 2000) ; (*??? * stirling et al. 2001)). brighter, optically - thin, relativistically - moving jets are associated with high - luminosity, soft x - ray states during outbursts ((*??? * mirabel & rodrguez 1994)). our current understanding of the duty cycles and disc - jet coupling in black hole x - ray binaries (bh xrbs) derives from a compilation of x - ray spectral and timing information, together with radio flux density monitoring and a limited set of high - resolution radio imaging. the current paradigm, or ` unified model'((*??? * fender et al. 2004)) suggests that the jet morphology and power correlate well with position in an x - ray hardness - intensity diagram (hid). steady, self - absorbed jets are inferred to exist in the very low luminosity quiescent state and the higher - luminosity hard state. as the x - ray intensity increases in the hard state, so too does the jet power, with the radio and x - ray luminosities following a non - linear correlation, @xmath0. at about @xmath1@xmath2, the x - ray spectrum begins to soften, and the jet speed increases as the inner disc radius moves inwards. below a certain x - ray hardness (the ` jet line'), the core jet switches off and internal shocks develop in the flow, which are observed as bright, relativistically - moving radio ejecta. the source may remain at high luminosity for several weeks, making repeated transitions back and forth across the jet line, before the x - ray luminosity eventually decreases to @xmath3 where the spectrum hardens (note the hysteresis effect compared to the higher - luminosity hard - to - soft transition), the core jet is re - established, and the source fades back into quiescence. of the many different classes of neutron star (ns) x - ray binaries, only the low - magnetic field systems have shown evidence for radio emission. these systems are divided by mass accretion rate into two main classes ; the z - sources and the atoll sources, each with distinct x - ray spectral and timing characteristics. the z - sources are consistently accreting at or close to the eddington luminosity, whereas the atolls are accreting at a somewhat lower level. the x - ray spectral and timing properties of atolls show many similarities to black hole systems, with distinct soft (so - called ` banana') and hard (` extreme island') x - ray states, making them the best sources to compare with black hole x - ray binary outbursts. to date, only a handful of atoll sources have been detected in the radio band during simultaneous radio / x - ray observations, showing them to be systematically fainter than the black hole sources at the same eddington - scaled x - ray luminosity (*??? * (migliari & fender 2006)). however, it appears that a similar correlation between radio and x - ray luminosities holds in the hard - state atoll sources, but with a lower normalization and a steeper power - law index ; @xmath4. at high x - ray luminosities, close to the eddington limit where the sources show z - type behavior, bright transient ejecta are thought to exist, just as in black hole systems, although there appears to be only mild suppression of radio emission in the atoll sources when they reach a soft x - ray state. a third class of interacting binaries where mass is transferred from a donor to a degenerate compact object has a white dwarf as the accretor. one class of such systems, the dwarf novae (a type of cataclysmic variable ; cv), also have accretion discs, which periodically develop disc instabilities, leading to a sudden increase in the accretion rate and causing short - lived outbursts. during such outbursts, which recur at intervals of several weeks, these systems brighten by several magnitudes in the optical band. despite similarities to xrbs, with accretion onto a compact object and outbursts triggered by disc instabilities, no jets have thus far been directly resolved in cvs. generalizing the hid to a ` disc - fraction luminosity diagram'(dfld) by plotting the optical flux of the system against the fraction of emission arising from the power - law spectral component (as opposed to disc emission) shows that outbursts of dwarf novae follow a very similar track to bh and ns systems. extending this analogy with the unified model then suggests that they should show flat - spectrum radio emission in the rise phase of an outburst, and resolved ejecta during the subsequent spectral softening. the former prediction was spectacularly confirmed during an outburst of the dwarf nova ss cyg ((*??? * krding et al. 2008)). the radio emission was highly variable, peaking at 1.1mjy, and coincident with the optical outburst. during the decay, the radio spectrum was slightly inverted, suggestive of a compact jet, as resolved in bh xrbs. however, the existence of such a jet can only be directly verified with high - resolution imaging.
The jet acceleration and collimation probe of transient x-ray binaries (jacpot xrb)
the jet acceleration and collimation probe of transient x - ray binaries (jacpot xrb) project aims to probe the similarities and differences in the jet launching process between these three different source classes by conducting intensive monitoring of an outburst of each of these different classes of accreting compact object. time - resolved, high angular resolution monitoring in the radio band using the vlba and (e)vla, plus wsrt and evn when available, allows us to directly image the evolution of the jets over the course of an outburst. simultaneous multi - wavelength coverage using _ swift _ bat, _ rxte _ pca, and _ maxi _ in the x - ray band, plus optical and infrared monitoring using the two faulkes telescopes, fancam and ctio (via the smarts consortium), then enables us to tie the jet behavior to the corresponding changes in the accretion flow. with such detailed, multi - wavelength monitoring, we aim to directly test the prevailing paradigm for black hole x - ray binary outbursts (*??? * (fender et al. 2004)) and, by comparing the jet behavior across the three different source classes, to ascertain the role played in jet formation by the depth of the gravitational potential well and the presence or absence of a stellar surface and stellar magnetic field. one outburst of each class of accreting compact source has now been observed as part of the jacpot xrb project, and data reduction and analysis are underway. monitoring campaigns on one further black hole and one neutron star system have been approved by nrao, and are awaiting triggering events. , and the crosses mark the position of the central binary.] we triggered our first black hole observing campaign on the 2009 outburst of the black hole candidate source h1743 - 322. the x - ray evolution of the outburst has already been analysed in detail by (*??? * motta et al. (2010)) and (*??? * chen et al. (2010)). we monitored the outburst with the vla, atca and vlba, tracking the evolution of the radio emission through the entire outburst. while the vlba observations were hampered by the strong scattering and lack of good calibrators in the direction of the galactic center, we detected both the compact core during the initial rising hard state, and also the launching of ejecta immediately following the transition to the soft intermediate state (fig. [fig : h1743_images]). the disc - jet coupling in this system follows the standard pattern outlined by (*??? * fender et al. (2004)), as shown in the hardness - intensity diagram (hid) in fig. [fig : h1743_hids]. we see flat - spectrum radio emission from a compact, unresolved core jet in the hard intermediate state (hims) at the beginning of the outburst. this core radio emission then quenches, giving rise to very faint, steep - spectrum emission as the source moves into the soft intermediate state (sims), following which we detect bright, optically - thin ejecta. the radio emission does not completely vanish during the high soft state (hss), but shows a modest increase in flux density once the source returns to the hims, before fading with time during the decay phase in the low hard state (lhs). we draw particular attention to the pre - flare ` quench'phase seen during the hard to soft transition. originally noted by (* fender et al. (2004)), this short - lived phase is often missed owing to limited temporal sampling in the radio band during outbursts of black hole x - ray binaries, but is surely key to understanding the nature of the transition from steady compact jets to bright, relativistically - moving ejecta. over the reverse transition, the reactivation of the compact jet occurs at a hardness ratio of @xmath5, and appears to support the assertion of (*??? * fender et al. (2009)) that the ` jet - line'is not vertical. the 2009 november outburst of aql x-1 was observed using the vla, vlba, e - evn and _ rxte _ pca, resulting in the first milliarcsecond - scale vlbi detection of the source. we obtained full spectral coverage of the outburst from the radio through infrared (fancam, smarts), optical (faulkes), ultraviolet (_ swift _ uvot) and x - ray bands. modeling of the full spectral energy distribution and its evolution is underway. the radio and x - ray lightcurves, plus the hardness - intensity diagram (hid) and color - color diagram (ccd) of the outburst are shown in fig. [fig : aqlx1_lcs]. while full details of the observations and analysis may be found in (* miller - jones et al. (2010)), we summarize the most salient points here. while hysteresis in the hid of aql x-1 is well known (* (maitra & bailyn 2004)), there have been few previously - reported radio detections (*??? * (tudose et al. 2009)). our monitoring campaign provided the most complete radio coverage to date of an entire outburst of aql x-1, finding the radio emission to be consistent with being activated by transitions from a hard spectral state to a soft state, and also by the reverse transition, just as seen in bh systems. a further similarity with black hole systems was the quenching of the radio emission above a certain x - ray luminosity (@xmath6% of the eddington luminosity) while in a soft spectral state. however, in contrast to the bh systems, the vlbi observations combined with radio spectral information showed no evidence for steep - spectrum, optically - thin ejecta after the hard - to - soft state transition. the radio emission was at all times consistent with a compact, partially self - absorbed jet as seen in the hard states of bh systems. this may suggest a fundamental difference in the jet formation mechanism between the two classes of source. ss cyg is the brightest dwarf nova, at a distance of only 166pc (*??? * (harrison et al. 1999)). it undergoes several outbursts each year, and in 2010 april, the aavso notified us that a new outburst of the source was beginning. we were able to get on source with evla within 24h, and followed up the ensuing detection with a full monitoring campaign using evla, wsrt, vlba, _ rxte _, _ swift _ and aavso. unfortunately the source is too bright for _ swift _ uvot to provide coverage in the ultraviolet. the x - ray and radio light curves of the outburst are shown in fig. [fig : sscyg_lcs]. the outburst proceeded in a very similar fashion to that monitored by (*??? * wheatley et al. (2003)), showing a sharp drop in the hard x - ray emission at the peak of the optical outburst, and a subsequent recovery as the optical emission faded. as seen by (*??? * krding et al. (2008)), the radio emission rose rapidly at the beginning of the outburst, showing a steep radio spectrum (@xmath7, with @xmath8) immediately after the peak, suggestive of optically - thin ejecta. during the decay phase later in the outburst, the observed flatter radio spectrum is more consistent with a partially self - absorbed compact jet. the source was detected by the vlba during 5 of the 6 epochs. these detections imply a brightness temperature in excess of @xmath9k. as argued by (*??? * krding et al. (2008)), together with the radio spectral constraints this suggests the presence of an unresolved jet, as seen in accreting neutron star and black hole systems. as an aside, the astrometric accuracy of the vlba observations allowed us to fit for the source proper motion over the 10 days of observation. assuming a distance of 166pc (*??? * (harrison et al. 1999)), we find a proper motion of @xmath10masy@xmath11 in r.a., and @xmath12masy@xmath11 in dec., fairly consistent with that given in ucac3 (*??? * (zacharias et al. 2010)), but significantly different from the proper motion derived by (*??? * harrison et al. (2000)), with implications for the accuracy of the hst astrometric measurements. a future revision of the distance could remove the discrepancy between predictions of the disc instability model and the observed outburst luminosity (*??? * (schreiber & lasota 2007)).
Summary
we have embarked on an ambitious project to test the disc - jet coupling in three different classes of accreting stellar - mass compact objects, using high - resolution vlbi radio observations coupled with multiwavelength monitoring data. using these data, we aim to (a) test the ` unified model'of (*??? * fender et al. (2004)) for the disc - jet coupling in black hole x - ray binaries, and (b) to determine the role played in jet formation by the depth of the gravitational potential well, the stellar surface and the stellar magnetic field. we have already observed one outburst from each of a black hole, a neutron star and a white dwarf binary system, and while preliminary analysis of the data largely confirms the existing paradigm, some interesting similarities and differences between the evolution of black hole and neutron star outbursts have been found.
Acknowledgements
we are very grateful to the nrao, _ rxte _ and _ swift _ schedulers for their flexibility and prompt responses which have made these observing campaigns feasible. we would also like to thank mickael coriat and stephane corbel for sharing their atca data on h1743322, and matthew templeton and the worldwide network of aavso observers who triggered and supported our observing campaign on ss cyg. | relationships between the x - ray and radio behavior of black hole x - ray binaries during outbursts have established a fundamental coupling between the accretion disks and radio jets in these systems.
i begin by reviewing the prevailing paradigm for this disk - jet coupling, also highlighting what we know about similarities and differences with neutron star and white dwarf binaries. until recently
, this paradigm had not been directly tested with dedicated high - angular resolution radio imaging over entire outbursts. moreover, such high - resolution monitoring campaigns had not previously targetted outbursts in which the compact object was either a neutron star or a white dwarf. to address this issue,
we have embarked on the jet acceleration and collimation probe of transient x - ray binaries (jacpot xrb) project, which aims to use high angular resolution observations to compare disk - jet coupling across the stellar mass scale, with the goal of probing the importance of the depth of the gravitational potential well, the stellar surface and the stellar magnetic field, on jet formation.
our team has recently concluded its first monitoring series, including (e)vla, vlba, x - ray, optical, and near - infrared observations of entire outbursts of the black hole candidate h1743 - 322, the neutron star system aquila x-1, and the white dwarf system ss cyg. here
i present preliminary results from this work, largely confirming the current paradigm, but highlighting some intriguing new behavior, and suggesting a possible difference in the jet formation process between neutron star and black hole systems. | 1010.3062 |
Introduction
the scale size of galaxies is one of the fundamental parameters to elucidate the history of galaxy formation and evolution. the change of size and stellar - mass relations over cosmic time would pose strong constraints on models of galaxy evolution. the observational relations between galaxy size and stellar mass have been studied in the local universe, based on the sloan digital sky survey (shen et al. 2003 ; bernardi et al. 2011). using rest - frame optical bands, which presumably trace the distribution of stellar mass in galaxies, many studies have investigated galaxy sizes at higher redshift as a function of stellar mass for massive galaxies (@xmath16 m@xmath7). for example, the relations for galaxies at @xmath17 were studied for late - type galaxies (barden et al. 2005) and for early - type galaxies (mcintosh et al. 2005 ; trujillo, ferreras, & de la rosa 2011). damjanov et al. (2009) and mancini et al. (2010) gave size - mass relations of massive galaxies (@xmath18 m@xmath7) at @xmath19. williams et al. (2010) studied the relation with large samples of galaxies at @xmath20. for higher redshifts to @xmath10, size - mass relations have been obtained for galaxies with @xmath21 m@xmath7 (e.g., franx et al. 2008 ; nagy et al. 2011 ; cassata et al. 2011). many studies have corroborated that massive galaxies at high redshifts were much smaller than local galaxies with comparable mass (e.g., daddi et al. 2005 ; trujillo et al. 2006, 2007 ; toft et al. 2007 ; zirm et al. 2007 ; cimatti et al. 2008 ; buitrago et al. 2008 ; van dokkum et al. 2008, 2009, 2010 ; akiyama et al. 2008 ; damjanov et al. 2009 ; carrasco et al. 2010 ; cassata et al. 2010 ; szomoru et al. 2010 ; van der wel 2011 ; cassata et al. 2011). at a fixed stellar mass, spheroidal galaxies were significantly more compact at high redshift and evolved with rapid increase of the effective radius by a factor @xmath224 or even larger from @xmath23 (e.g., buitrago et al. 2008 ; carrasco et al. 2010) and by a factor @xmath222 from @xmath11 (e.g., van der wel et al. 2008 ; trujillo et al. 2011). the finding of compact massive galaxies with a high stellar velocity dispersion also supports their existence (van dokkum 2009 ; van de sande 2011). it contrasts with the absence of such compact massive galaxies in the local universe, though several candidates have been found at @xmath24 (stockton et al. 2010) and in the local universe (trujillo et al. 2009 ; valentinuzzi et al. the findings demonstrate that massive galaxies have increased their size dramatically since @xmath25 in a different manner from the evolution of less massive galaxies. however, other studies have reached contradictory conclusions. there is significant disagreement between the results of different studies. barden et al. (2005) found weak or no evolution in the relation between stellar mass and effective disc size for galaxies with @xmath26 m@xmath7 since @xmath11. for early - type galaxies at @xmath11, mcintosh et al. (2005) showed that luminosity - size and stellar mass size relations evolve in a manner that is consistent with a completely passive evolution of the red early - type galaxy population. it is also shown that not all high - redshift early - type galaxies were compact and underwent dramatic size evolution (e.g., toft et al. 2007 ; zirm et al. 2007 ; saracco et al. 2009, 2011 ; mancini et al. 2010 ; stott et al, 2011). from the study of surface brightness in rest - frame @xmath27 and @xmath28 bands at @xmath29, ichikawa et al. (2010) gave another evidence for no conspicuous evolution in galaxy sizes. as many previous studies show, the size evolution of galaxies are still controversial. any systematic errors in the observation or analyses could bias results towards such a significant evolution (e.g., mancini et al. 2010 ; hopkins et a. 2009b ; bouwens et al. 2004). the origin of the discrepancy could be ascribed to redshift effects that more distant galaxies look more compact due to the difficulty of measuring envelopes at low surface density. the light from the outer portion of high - redshift galaxies is apt to be hidden in noise for low @xmath30 observations. as the consequence, effective radii and total luminosity (or stellar mass) would be underestimated. on the other hand, very deep observations (e.g., szomoru et al. 2010 ; cassata et al. 2010 ; law et al. 2012) or stacking methods to enhance the faint envelope of galaxies (e.g., zirm et al. 2007 ; van dokkum et al. 2008, 2010 ; van der wel et al. 2008) have claimed that it is not the case. how significantly have the sizes of less - massive normal galaxies evolved from the early universe to the current epoch? in this context, we look into the evolution of stellar - mass and size scaling relations on the basis of half- and 90 percent - light encircles, focusing on less massive galaxies (@xmath31 m@xmath7) at @xmath1, using a deep @xmath4-selected galaxy catalogue. we infer that the outer radius is more influenced by merging effect than star formation or central activity. in 2, we describe the catalogue we used, which is among the deepest in the @xmath4 band to date. the depth is crucial for studying galaxies of low - surface brightness or galaxies which are dimmed due to the cosmological expansion at high redshift to measure the radius at faint outskirt of galaxies. the data analysis and the result for the size and stellar - mass relation are detailed in 3 and 4. the results are discussed in 5. ichikawa et al. (2010) studied the evolution of surface brightness of galaxies at @xmath29 in rest - frame @xmath27 and @xmath28 bands with the same data as the present study. we will discuss the consistency of the present result with their study. throughout this paper, we assume @xmath32, @xmath33, and @xmath34 km s@xmath35 mpc@xmath35. we use the ab magnitude system (oke & gunn 1983 ; fukugita et al.
Data
we use the @xmath4-band selected catalogue of the moircs deep survey (mods) in the goods north region (kajisawa et al. 2009, hereafter k09 ; kajisawa et al. 2011, k11), which are based on our imaging observations in @xmath36 bands with moircs (suzuki et al. 2008) and archived data. four moircs pointings cover 70 percent of the goods north region (103 arcmin@xmath37, hereafter referred as ` wide'field). one of the four pointings, which includes hdf - n (williams et al. 1996), is the ultra - deep field of mods (28 arcmin@xmath37, ` deep'field). as the accuracy of background subtraction was highly demanded for the study of faint end of galaxies, the background was scrutinized and carefully subtracted (see k11 for the details). the surface brightness limit was extensively examined in k11. the typical @xmath38 surface brightness fluctuations in one arcsec diameter were found to be @xmath2227 mag arcsec@xmath39 and @xmath2227.5 mag arcsec@xmath39 in @xmath4 band for the wide and deep fields respectively. these are @xmath40 mag deeper than those of previous studies in @xmath4 band. the depth is crucial for the study of low surface brightness features at the outskirts of galaxies because of the strong dependence of cosmological dimming of surface brightness on redshift. for the total magnitude @xmath41 in @xmath4 band, we use mag_auto obtained by sextractor (bertin & arnouts 1996). the @xmath42-percent completeness of the catalogue is @xmath43 in the wide field and @xmath2226 mag in the deep field. we exclude fainter galaxies from the present analysis. the fwhms of the final stacked images are 0.46 arcsec for the deep image and 0.530.60 arcsec for the wide image. the numbers of galaxies are 3555 and 6063, respectively. to obtain the stellar mass of mods samples, k09 performed sed fitting of the multiband photometry (@xmath44, 3.6 @xmath45 m, 4.5 @xmath45 m, and 5.8 @xmath45 m) with population synthesis models. we adopt the results with galaxev templates (bruzual & charlot 2003) and the salpeter initial mass function (see k09 for more details). the stellar masses (@xmath0) are obtained from the best - fit stellar mass - to - luminosity ratio in @xmath4 band and scaled with the @xmath4-band flux. a detailed comparison between different mass estimators is given in k09. in the present analysis, the near - infrared data (3.6@xmath45 m) of spitzer / irac are available for most of the sample galaxies (96 percent), so that sed fitting is reasonably reliable for the photometry at rest-@xmath27 (@xmath46 m) out to redshift @xmath223. we assume that the size of the stellar system in galaxies is represented by the size measured on @xmath4-band images, which are the rest - frame optical to near - infrared wavelengths at @xmath29. the difference of the sizes between optical and infrared light is discussed in 4. in the present catalogue, 209 galaxies are identified as x - ray sources. among them, 61 sources are massive galaxies emitting hard x - rays with @xmath47 m@xmath7 at @xmath48 (yamada et al.. one would expect galaxies with agn emission to have smaller half - light radius. considering possible effects on the size and mass estimates, we discard x - ray sources in what follows.
Analyses
as the definition of galaxy size, most previous studies have used the scale length (@xmath49), obtained by fitting the srsic profile (srsic 1968) to galaxy images. the accuracy of the scale length strongly depends on image quality and depth of observations. for bright galaxies, reliable galaxy sizes can be measured from ground - based data (trujillo et al. 2006 ; franx et al. 2008 ; williams et al. 2010). on the other hand, it is crucial to take an accurate measure of the profile for small objects near the resolution limit. in fact, konishi et al. (2011) applied two dimensional fitting for the present sample to obtain srsic indexes with a single component for a morphological study of mods galaxies. however, they found that reliable fitting was successful only for comparatively massive galaxies (@xmath26 m@xmath7) at @xmath11 and very bright galaxies (@xmath50) at higher redshift. the resolution of fwhm @xmath51 arcsec like the present sample would not allow reliable fitting with srsic profile to fainter galaxies. srsic parameters are sometimes degenerated between @xmath52 and @xmath49, which depends on the surface - brightness limit of the images (e.g., mancini et al. 2010 ; stott et al. 2011), so that the reliable fitting would be difficult for faint galaxies. moreover high - redshift galaxies exhibit a wide range of disturbed morphologies (e.g., kajisawa & yamada 2001). therefore, the fitting for less massive galaxies at high redshift could also be strongly influenced by their more amorphous features in the shape. in addition, we must take into account that we are observing a mix of galaxy morphologies. as such, we define the half - light radius (@xmath53) as the radius of a circular aperture, which encircles half the @xmath4-band light emitted from galaxies. as we here focus not just on massive (or bright) galaxies, but also on less massive galaxies up to @xmath10, we prefer @xmath54 to @xmath49 for the study of the structural parameters. if a galaxy profile extends to infinity, @xmath49 encloses half the flux in the srsic profile. however, as galaxies are supposed to have an edge at several times the scale length, @xmath49 does not always mean half - light radius. therefore, in general, @xmath53 is smaller than @xmath49. in the same way, we define 90 percent - light radius @xmath55. the 90-percent radius will give us more information at outer region of galaxies, where the size would be more influenced by merging effect. @xmath53 and @xmath56 are obtained by sextractor with phot_fluxfrac=0.5 and 0.9 figure [fig - kvsr50] shows @xmath53 versus @xmath41 for all objects in mods. mods catalogue lists many spectroscopically confirmed stars, which are also plotted in fig. [fig - kvsr50]. as few spectroscopic data were available for stars fainter than @xmath57 in the present region, we examined the reliability of @xmath54 using artificial stars. the artefacts were convolved with a moffat point spread function (@xmath58 and fwhm=0.5 arcsec in the deep field and 0.6 arcsec in the wide field), and then buried in images as noisy as those of the deep and wide fields. sextractor was applied to the artefacts in the same manner as for the mods catalogue. the results are plotted in fig. [fig - kvsr50]. the location of stars gives a clear boundary of unresolved galaxies. we define galaxies smaller than @xmath59 in the wide field @xmath60 in the deep field as unresolved galaxies. we first study the reliability of the galaxy size. seeing and background noise strongly affect the observed @xmath54 and @xmath55, especially for faint or small galaxies. to examine the effect on the size estimate, we generated mock galaxies with a 1/4-law or exponential profile extending to 3 times scale length. the images are then convolved with a moffat point spread function (@xmath58 and fwhm=0.5 arcsec in the deep field and 0.6 arcsec in the wide field), and buried in the simulated noise image. the galaxies were randomly generated with various magnitudes and effective radii or scale lengths, and analyzed with sextractor in the same manner as for the mods galaxies. the results are shown in fig. [fig - mockr90r50]. seeing and noise seriously change the observed radii specifically for galaxies with a 1/4-law profile. the observed @xmath53 is significantly smaller than the intrinsic values in general. the effect is stronger for fainter 1/4-law galaxies, whereas the effect is much smaller for exponential galaxies. the apparent smaller size than the intrinsic size would be due to the fact that the faint and large envelope of 1/4-law galaxies is hidden in background noise. we should be aware of this effect if the observation is not deep enough for early - type galaxies (see also hopkins et al. 2009b ; mancini et al. the effect will be discussed in more detail in the following section. it is noted that williams et al. (2010) claimed no systematic effects on @xmath49 for @xmath61, using the shallower mods images. the radii, @xmath53 and @xmath56, for small galaxies are affected by smearing due to seeing effect and the effect depends on galaxy morphology and size. therefore, we obtained eq. 1 to correct the size effect by fitting the deviation of the size as shown in fig. [fig - mockr90r50] (dash line). however, it is hard to examine the shapes of the present small or high-@xmath28 galaxies, so that we do not correct the effect which depends on morphology. @xmath62 we adopt @xmath63 and 0.35 arcsec for @xmath53 and @xmath56, respectively. the size errors for mock galaxies are depicted in fig. [fig - comparionsize] as a function of observed magnitude. we should take into account that much smaller values are obtained for the size of 1/4-law galaxies. figure [fig - comparionmag] is the difference between intrinsic and observed magnitudes. at the faint limit, observed magnitudes are fainter by @xmath64 mag for 1/4-law galaxies, while the effect is much smaller for exponential galaxies. these results suggest that shallower observations tend to more underestimate size and flux (therefore, stellar mass). we discuss later the influence on the results due to the systematic errors. using photometric (or spectroscopic, if available) redshift data, @xmath53 and @xmath56 in arcsec are converted to the physical size in kpc (@xmath2 and @xmath3). the photo-@xmath28 accuracy of the present sample is reasonably good, @xmath65 (@xmath66) (k11), so that we expect no serious influence on our result due to the photo-@xmath28 error. we will confirm this later using galaxies with spectroscopic redshifts. the results for @xmath2 and @xmath3 are shown in figs. [fig - massvsr50] and [fig - massvsr90] as a function of the stellar mass of galaxies for different redshift bins. the unresolved galaxies are displayed in another figure (fig. [fig - r50reject]) for reference. the results of effective radius (@xmath49) for local galaxies by shen et al. (2003) are also shown in fig. [fig - massvsr50] for comparison. size - mass relations for star - forming galaxies are consistent with that of late - type galaxies by shen et al. (2003), although our definition of half - light radius is different from that of shen et al. it should be noted that it would be difficult to discuss the consistency for massive early - type galaxies because of the small number of such galaxies in our sample. since our image quality is not high enough for classifying the galaxies into morphological classes at high redshifts, we divided the samples into quiescent and star - forming galaxy groups using a two - color diagnostic plot of rest - frame @xmath67 and @xmath68 colors (williams et al. 2009). the adopted selection criteria for quiescent galaxies are as follows : @xmath69 where @xmath67 and @xmath68 in the rest frame were obtained with the sed model fit to galaxies. the offset, @xmath70, is 0.69, 0.59, and 0.49 for @xmath71, @xmath72, and @xmath73, respectively. for @xmath74 galaxies, we applied the offset for @xmath75. therefore the selection would be less reliable (see williams et al. 2009), though the number of such galaxies is very small. additional criteria of @xmath76 and @xmath77 are required for quiescent galaxies at all redshifts to prevent contamination from unobscured and dusty star - forming galaxies, respectively. the small differences in the slope and offset of the regression lines in different redshift bins in figs. [fig - massvsr50] and [fig - massvsr90] would imply a universal relation between the stellar mass and size of galaxies, irrelevant to redshift. in this context, we plot the stellar - mass surface density (smsd), which is defined as @xmath78, in a single redshift bin as a function of stellar mass for all galaxies at @xmath79 (figs. [fig - sd50all] and [fig - sd90all]). the figures are essentially equivalent to the size - mass distribution. however, we can easily identify very compact or low - surface brightness galaxies in the figures, if they exist. for example, galaxies as compact (or diffuse) as 3 times smaller (larger) in size than the average should be located at @xmath221.0 dex above (below) the average. such low surface brightness galaxies can be recognized at @xmath9 m@xmath7 in fig. [fig - sd90all]. in contrast, massive and compact galaxies are rare in the present sample. it is noted that most of massive galaxies with @xmath80 m@xmath7 should be resolved in all redshift bins of the deep field. we conjecture that very compact galaxies coalesce in the unresolved galaxies in fig. [fig - r50reject].
Results
since the strong correlations of @xmath2 and @xmath3 with @xmath0 in figs. [fig - massvsr50] and [fig - massvsr90] are suggested, we obtain the least square fit between the size and mass of the galaxies with a linear regression, @xmath81 where @xmath82 is 50 for half light radius or 90 for 90% light radius and @xmath83 is the radius at @xmath84 m@xmath7. @xmath85 is 10 for all and star - forming galaxies, while @xmath86 for quiescent galaxies because the quiescent galaxies with @xmath87 m@xmath7 were not observed at @xmath88 and their @xmath89 is statistically less robust due to the small numbers of the sample in narrow mass ranges at high redshits. we note that quiescent galaxies are located on average below the regression lines in all redshift bins, though the offset is smaller for @xmath3 than for @xmath2. the fact suggests that quiescent galaxies are more compact than star - forming galaxies at a give mass. the best - fit slopes (@xmath90 and @xmath91) and offsets (@xmath92 and @xmath93) of the regression analysis with mean errors for all, quiescent galaxies (qsg), and star - galaxies (sfg) are summarized in table 1. the dispersion, @xmath94, of the linear fit is listed in the sixth and last columns. as the exclusion of unresolved sources could biases the results towards larger radii, we obtained in table 2 the least - squares fit with the unresolved galaxies in fig.[fig - r50reject]. we depict the evolution of the slope and offset as a function of redshift in fig. [fig - a_evolution]. while the slope for the quiescent galaxies are a little steeper than those for star - forming galaxies, the figure indicates that the slopes remain within @xmath95 and offsets do not significantly change from @xmath10 to @xmath12, irrespective of the sample selection except the quiescent population at @xmath96 where the statistical error is very large. for smsd in figs. [fig - sd50all] and [fig - sd90all], we define the regression as @xmath97 where @xmath98 is smsd at @xmath99 m@xmath7. the results are shown in table 3. ccrcccrccc & & & @xmath2 & & & & @xmath3 & + redshift & sample & @xmath100 & @xmath90 & @xmath101 & @xmath94 & @xmath100 & @xmath91 & @xmath102 & @xmath94 + & & & & (kpc) & & & & (kpc) & + @xmath103 & all & 5259 & @xmath104 & @xmath105 & 0.130 & 5236 & @xmath106 & @xmath107 & 0.148 + & qsg & 408 & @xmath108 & @xmath109 & 0.110 & 408 & @xmath110 & @xmath111 & 0.099 + & sfg & 4851 & @xmath112 & @xmath113 & 0.129 & 4828 & @xmath114 & @xmath115 & 0.151 + + @xmath116 & all & 543 & @xmath117 & @xmath118 & 0.133 & 542 & @xmath119 & @xmath120 & 0.168 + & qsg & 28 & @xmath121 & @xmath122 & 0.098 & 28 & @xmath123 & @xmath124 & 0.100 + & sfg & 515 & @xmath125 & @xmath126 & 0.133 & 514 & @xmath127 & @xmath128 & 0.170 + @xmath129 & all & 717 & @xmath130 & @xmath131 & 0.119 & 714 & @xmath132 & @xmath133 & 0.124 + & qsg & 70 & @xmath134 & @xmath135 & 0.126 & 70 & @xmath136 & @xmath137 & 0.107 + & sfg & 647 & @xmath138 & @xmath139 & 0.113 & 644 & @xmath140 & @xmath141 & 0.124 + @xmath142 & all & 1083 & @xmath143 & @xmath144 & 0.123 & 1079 & @xmath145 & @xmath146 & 0.151 + & qsg & 134 & @xmath147 & @xmath148 & 0.104 & 134 & @xmath149 & @xmath150 & 0.095 + & sfg & 949 & @xmath151 & @xmath152 & 0.121 & 945 & @xmath153 & @xmath154 & 0.156 + @xmath155 & all & 888 & @xmath156 & @xmath157 & 0.130 & 885 & @xmath158 & @xmath159 & 0.124 + & qsg & 83 & @xmath160 & @xmath161 & 0.105 & 83 & @xmath162 & @xmath163 & 0.085 + & sfg & 805 & @xmath140 & @xmath164 & 0.128 & 802 & @xmath165 & @xmath166 & 0.125 + @xmath167 & all & 519 & @xmath168 & @xmath169 & 0.127 & 514 & @xmath170 & @xmath171 & 0.161 + & qsg & 32 & @xmath172 & @xmath173 & 0.089 & 32 & @xmath174 & @xmath175 & 0.079 + & sfg & 487 & @xmath127 & @xmath176 & 0.127 & 482 & @xmath177 & @xmath178 & 0.165 + @xmath179 & all & 555 & @xmath180 & @xmath181 & 0.128 & 554 & @xmath182 & @xmath183 & 0.128 + & qsg & 28 & @xmath184 & @xmath185 & 0.121 & 28 & @xmath186 & @xmath187 & 0.094 + & sfg & 527 & @xmath188 & @xmath189 & 0.124 & 526 & @xmath190 & @xmath191 & 0.128 + @xmath192 & all & 700 & @xmath193 & @xmath194 & 0.121 & 696 & @xmath195 & @xmath196 & 0.146 + & qsg & 27 & @xmath197 & @xmath198 & 0.109 & 27 & @xmath199 & @xmath200 & 0.097 + & sfg & 673 & @xmath201 & @xmath202 & 0.119 & 669 & @xmath203 & @xmath204 & 0.147 + @xmath205 & all & 254 & @xmath206 & @xmath207 & 0.130 & 252 & @xmath208 & @xmath209 & 0.166 + & qsg & 6 & @xmath210 & @xmath211 & 0.040 & 6 & @xmath212 & @xmath213 & 0.035 + & sfg & 248 & @xmath214 & @xmath215 & 0.129 & 246 & @xmath216 & @xmath217 & 0.167 + ccrcccrccc & & & @xmath2 & & & & @xmath3 & + redshift & sample & @xmath100 & @xmath90 & @xmath101 & @xmath94 & @xmath100 & @xmath91 & @xmath102 & @xmath94 + & & & & (kpc) & & & & (kpc) & + @xmath103 & all & 6532 & @xmath218 & @xmath219 & 0.161 & 6502 & @xmath220 & @xmath221 & 0.161 + & qsg & 445 & @xmath222 & @xmath223 & 0.116 & 445 & @xmath224 & @xmath225 & 0.108 + & sfg & 6087 & @xmath226 & @xmath227 & 0.162 & 6057 & @xmath228 & @xmath229 & 0.164 + + @xmath116 & all & 646 & @xmath230 & @xmath231 & 0.156 & 645 & @xmath232 & @xmath233 & 0.181 + & qsg & 31 & @xmath234 & @xmath235 & 0.129 & 31 & @xmath236 & @xmath237 & 0.124 + & sfg & 615 & @xmath238 & @xmath239 & 0.156 & 614 & @xmath232 & @xmath240 & 0.183 + @xmath129 & all & 858 & @xmath241 & @xmath242 & 0.136 & 855 & @xmath243 & @xmath244 & 0.134 + & qsg & 73 & @xmath245 & @xmath246 & 0.127 & 73 & @xmath247 & @xmath248 & 0.111 + & sfg & 785 & @xmath249 & @xmath250 & 0.133 & 782 & @xmath251 & @xmath252 & 0.135 + @xmath142 & all & 1313 & @xmath253 & @xmath254 & 0.155 & 1308 & @xmath255 & @xmath256 & 0.161 + & qsg & 141 & @xmath257 & @xmath258 & 0.102 & 141 & @xmath259 & @xmath260 & 0.094 + & sfg & 1172 & @xmath261 & @xmath262 & 0.157 & 1167 & @xmath263 & @xmath264 & 0.166 + @xmath155 & all & 1082 & @xmath265 & @xmath266 & 0.164 & 1077 & @xmath253 & @xmath267 & 0.146 + & qsg & 88 & @xmath268 & @xmath269 & 0.103 & 88 & @xmath270 & @xmath271 & 0.084 + & sfg & 994 & @xmath272 & @xmath273 & 0.164 & 989 & @xmath274 & @xmath264 & 0.148 + @xmath167 & all & 675 & @xmath275 & @xmath276 & 0.159 & 668 & @xmath277 & @xmath278 & 0.176 + & qsg & 35 & @xmath279 & @xmath280 & 0.111 & 35 & @xmath281 & @xmath282 & 0.091 + & sfg & 640 & @xmath283 & @xmath284 & 0.160 & 633 & @xmath285 & @xmath286 & 0.179 + @xmath179 & all & 685 & @xmath287 & @xmath288 & 0.174 & 684 & @xmath289 & @xmath290 & 0.139 + & qsg & 36 & @xmath291 & @xmath292 & 0.128 & 36 & @xmath293 & @xmath294 & 0.120 + & sfg & 649 & @xmath295 & @xmath296 & 0.171 & 648 & @xmath297 & @xmath298 & 0.138 + @xmath192 & all & 923 & @xmath299 & @xmath300 & 0.152 & 918 & @xmath301 & @xmath302 & 0.156 + & qsg & 32 & @xmath303 & @xmath304 & 0.103 & 32 & @xmath305 & @xmath306 & 0.094 + & sfg & 891 & @xmath307 & @xmath308 & 0.151 & 886 & @xmath309 & @xmath310 & 0.157 + @xmath205 & all & 350 & @xmath311 & @xmath312 & 0.150 & 347 & @xmath313 & @xmath314 & 0.175 + & qsg & 9 & @xmath315 & @xmath316 & 0.097 & 9 & @xmath317 & @xmath318 & 0.109 + & sfg & 341 & @xmath319 & @xmath320 & 0.149 & 338 & @xmath321 & @xmath322 & 0.176 + ccrccrccccc & & & @xmath2 & & & & @xmath3 & + selection@xmath323 & sample & n & @xmath324 & @xmath325 & @xmath94 & n & @xmath326 & @xmath327 & @xmath94 + & & & & (@xmath328 kpc@xmath39) & & & & (@xmath328 kpc@xmath39) & + @xmath329 & all & 5259 & @xmath330 & @xmath331 & 0.26 & 5236 & @xmath332 & @xmath333 & 0.30 + & qsg & 408 & @xmath334 & @xmath335 & 0.22 & 408 & @xmath336 & @xmath337 & 0.20 + & sfg & 4851 & @xmath338 & @xmath339 & 0.26 & 4828 & @xmath340 & @xmath341 & 0.30 + with unresolved galaxies & all & 6532 & @xmath342 & @xmath343 & 0.32 & 6502 & @xmath344 & @xmath345 & 0.32 + & qsg & 445 & @xmath346 & @xmath347 & 0.23 & 445 & @xmath336 & @xmath348 & 0.22 + & sfg & 6087 & @xmath349 & @xmath350 & 0.32 & 6057 & @xmath342 & @xmath351 & 0.33 + @xmath352 & & 3980 & @xmath353 & @xmath354 & 0.25 & 3972 & @xmath355 & @xmath333 & 0.28 + @xmath356 & & 2378 & @xmath357 & @xmath358 & 0.25 & 2378 & @xmath359 & @xmath333 & 0.23 + spectroscopic redshift & & 1610 & @xmath360 & @xmath361 & 0.24 & 1610 & @xmath362 & @xmath363 & 0.22 + the errors in stellar mass mainly originated from sed fitting and photometric errors of the observations. in the course of @xmath364 fitting to obtain the best sed model with various parameters (e.g., star - formation time scale, photometric redshift for galaxies with no spectroscopic redshift available, age, extinction, and metallicity), the probability distributions of stellar mass can be calculated, where the photometry and photometric - redshift errors are included. the error for the observed smsd is dominated by the error for the stellar mass. see more details for the error estimate in k09 and ichikawa et al. (2010) the size and magnitude errors in accordance with galaxy magnitude have been examined using mock galaxies (figs. [fig - comparionsize] and [fig - comparionmag]). if all galaxies have a shape of 1/4-law, the error would be significant. recalling the size error due to the morphology strongly depends on magnitude, we examine the effect by confining the sample to bright galaxies, where the systematic error becomes smaller. we selected the samples with the magnitude limits, @xmath365 (23), 25(24), 26(25) for the deep (wide) fields, then obtained the regression lines again. the results are compared in table 3. it should be noted that the section does not change the result. one would be concerned about the reliance on photometric redshift estimates for high redshift galaxies (e.g., mosleh et al. 2011). in order to examine the error, we selected only the galaxies with spectroscopic redshift available and compared the result with the photo-@xmath28 sample. although the spectroscopic samples are limited to lower redshift, the result does not change our conclusion (table 3). due to possible color gradients in galaxies, it would be best to use images of the same rest - frame band with comparable @xmath30 for measuring the size of galaxies at all redshifts. nevertheless, we used the radius of galaxies measured on @xmath4-band image, because it is deepest among the images we used and because it is the rest - frame optical or longer band at @xmath29. if the central region of galaxies are younger (bluer) than the outer region, the mass - weighted radius could be larger than the luminosity - weighted radius. if it is older (redder), the result would be vice versa. star forming galaxies sometimes show strong morphological variation between observed wavelengths. bond, gawiser & koekemoer (2011) reported that this was not generically accompanied by a large difference in half - light radius. barden et al. (2005) measured the disk scale lengths of local galaxies in various bands. the average size is about 10 percent larger in @xmath27 band than in @xmath4 band. on the other hand, macarthur & courteau (2003) showed the contrary result that the distribution of disc scale lengths was a decreasing trend with increasing wavelengths (see also cassata et al. 2010 for early - type galaxies at @xmath366). to investigate the effect of the color gradient, we compare the 90 present - light radius of galaxies measured on the nearest rest - frame @xmath27 band for each redshift bin. the sizes of the present samples were obtained on acs @xmath367, @xmath28, and moircs @xmath368, @xmath369 bands and compared with that of @xmath4 band in fig. [fig - r90vsacs]. we used acs images binned to 0.117 arcsec per pixel, keeping the original image resolution. [fig - r90vsacs] demonstrates that the sizes of acs in @xmath367 and @xmath28 bands are systematically @xmath370 percent smaller than that in @xmath4 band. the convolution of the acs @xmath367-band image with psf and seeing enhances low surface brightness details. if the acs images are convolved with a gaussian (fwhm=2 pixel, 0.234 arcsec), the difference is decreased to @xmath2215 percent. as the convolution enhances the galaxy edge of low surface brightness, it tends to give a larger galaxy size (see also fig. 7 of mancini et al. it should be noted that the @xmath368 and @xmath369 images gives larger image size for smaller galaxies. the size strongly depends on the depth and the seeing size. although we have found the universal relation between @xmath0 and smsd (or size), which does not depend strongly on redshift, we have good reason to expect offsets from the relation for some galaxy populations (e.g., massive compact galaxies) in certain mass and redshift ranges. the regression analysis could be more weighted on the numerous less massive populations. therefore, it is likely that there are small (but not highly significant) offsets between galaxy populations. the small offsets would account for the size evolution of such populations. using the @xmath0@xmath371 relation obtained for the galaxies at @xmath372 (tables 1 and 2) as a reference, we examined the deviation of the median size of galaxies in the mass and redshift bins from the reference. the evolution of the massive (@xmath47m@xmath7) quiescent galaxies, defined in a @xmath67 and @xmath68 diagram (williams et al. 2010), is also obtained. (we use median values to avoid unreasonable contributions from outliers with large deviation.) we show the results in fig. [fig - r_evolution], where unresolved galaxies are included. in addition, the median values of @xmath15, which represent a sort of _ compactness _ of galaxies, are depicted in the figure to see the evolution of the compactness as a function of redshift. we note that the result with resolved galaxies are in good agreement with that with unresolved galaxies.
Discussion
using the deepest @xmath4-band image, we obtained scaling relations between stellar - mass (@xmath0) and size (or equivalently smsd) as a function of redshift for galaxies at @xmath1 in a wide mass range of @xmath373 m@xmath7. we defined the radii encircling half- (@xmath2) and 90 percent - light (@xmath3) in a circular aperture. the depth was found to be crucial for proper measure of the size and mass for less massive galaxies dimmed due to cosmic expansion. we conclude that there is no strong evidence for the size evolution at a given mass over the redshift range, irrespective of galaxy populations (star - forming, quiescent). the size - mass relation (@xmath374) have a universal slope (@xmath375) and small offsets (@xmath8 percent). in other words, as galaxies grow in mass through star formation or merging process, their sizes evolve in such a way that galaxies in general move along the scaling relation. the universal relation demonstrates that the stellar mass in galaxies with the same stellar mass was built up on average in a similar manner over cosmic times. the trend is insensitive to the stellar populations. our result is comparable with the @xmath376 found by shen et al. (2003) for low - mass galaxies (@xmath377 m@xmath7). on the other hand, it is in contrast to the steeper relation for massive galaxies obtained by previous studies (e.g., shen 2003 ; bernardi et al. the scaling relations are much tighter (@xmath3780.17 dex) than those (@xmath3790.5 dex) of previous studies in the local universe (e.g., shen et al. 2003) or at high redshifts (e.g., franx et al. 2008). the weak growth of the galaxy size, irrelevant to galaxy mass, is in disagreement with the scenario that more massive galaxies rapidly changed their sizes. if massive high-@xmath28 galaxies are several times smaller than local galaxies with comparable mass, they should be located well above the stellar mass vs. smsd relations. several candidates of such compact galaxies with @xmath380 possibly coalesce as unresolved galaxies in figs. [fig - sd50all] and [fig - sd90all]. however, such galaxies are found to be few. our finding is in contrast with the previous results for quiescent early - type massive galaxies (e.g., zirm et al. 2007 ; williams et al. 2010). with regard to smsd, the average smsds in @xmath2 for our sample galaxies with @xmath26 m@xmath7 are found not to evolve with redshift (log @xmath381 8.70). the average smsd for all galaxies at @xmath382 is @xmath383, which is very consistent with the result, @xmath384, for disk galaxies with @xmath26 obtained by barden et al (2005), though our sample could be a mixture of disk and elliptical galaxies. however, due to small volume of the present study, the results could be subject to statistical uncertainties. the weak size evolution presented in the present study is reconciled with the surface brightness (sb) evolution. ichikawa et al. (2010) presented a universal linear correlation between the stellar mass and sb in rest - frame @xmath27 and @xmath28 bands for galaxies at @xmath1, using the same sample as the present study. the correlation has a nearly constant slope, independent of redshift and color of galaxies in the rest-@xmath28 frame. in contrast, sb shows a strong dependence on redshift for a given stellar mass. it evolves as @xmath385. the increase in the luminosity with redshift is estimated by using the expected luminosity evolution from a single burst at @xmath386 for massive galaxies (@xmath26) and constant star formation for blue samples (@xmath387) for less massive galaxies. the redshift dependence of sb evolution is well explained by the pure luminosity evolution of galaxies out to @xmath10 without the need of the size evolution, which supports the results for young early - type galaxies (e.g., saracco et al. 2011) and for disk galaxies at @xmath388 (barden et al. 2005). nevertheless, there are some populations having a small offset from the universal scaling relation. we see more or less evolution in the size and compactness for such populations as a function of redshift in fig. [fig - r_evolution]. the figure suggests a moderate increase of 3050 percent for @xmath2 and @xmath3 for less massive galaxies (@xmath9 m@xmath7) from @xmath10 to @xmath11, while the sizes remains unchanged or slightly decrease towards @xmath12. for massive galaxies (@xmath13 m@xmath7), the evolution is @xmath14 percent in @xmath3 from @xmath10 to @xmath12, though that in @xmath2 is weaker. @xmath3 evolved as @xmath389, it is noted that trujillo et al. (2006), using the ground - based @xmath4-band data comparable with our depth, but in a smaller region (6.3 arcmin@xmath37), concluded that there was no evidence for significant evolution for the size - mass relation and that there was small increase (29 percent) since @xmath390 for the most massive bin @xmath391 m@xmath7. in our sample, the average @xmath2 for the most massive bin of @xmath392@xmath393 m@xmath7 is larger by @xmath394 percent at @xmath12 than at @xmath390, which is in good agreement with the result of trujillo et al. (2006), whereas it is much weaker than those of other studies for galaxies of similar mass (e.g., franx et al. 2008 ; buitrago et al. 2008 ; van der wel et al. 2009 ; williams et al. 2010). using new deep wfc3 data, cassata et al. (2011) and law et al. (2012) have given new evidences of compact galaxies at high - z. although the data in @xmath369 band is shallower by 1 mag arcsec@xmath39 than our @xmath4-band data, their conclusion would be robust. it would be possible that the compact galaxies are mingled in unresolved galaxies of the present sample or that the difference of the definition for half - right radius (@xmath395 and @xmath53) gives the contradict result. the ratio, @xmath15, in fig. [fig - r_evolution] would give a clue to understanding the evolution of the compactness of galaxies. while middle and low mass galaxies shows no evolution in the compactness, the galaxies in the most massive bin are more compact at higher redshift. recalling weaker evolution in @xmath2 than in @xmath3, we infer that the compactness evolution is ascribed mainly to the expansion of the outer rim of massive galaxies. in other words, at a given mass, massive galaxies in the local universe are more influenced by mergers or star formation at the outer envelopes than those at high redshifts. it would be worth noting that the galaxies in the massive bin of the deep field are complete in sampling over the present redshift range and their @xmath2 and @xmath3 are well resolved in the present psf. however, the present observation samples a comparatively small volume and therefore, the results could be subject to statistical uncertainties for massive galaxies of low number density due to field variances. the size evolution is often used to advocate the merging processes in the hierarchical paradigm of galaxy formation and evolution. minor dry merger is a plausible mechanism for weak size evolution (e.g., guo & white 2008 ; bezanson et al. 2009 ; naab, johansson, & ostriker 2009 ; hopkins et al. 2010). in that the size evolution is stronger in massive galaxies, our findings are qualitatively in agreement with previous studies based on simulations (e.g., boylan - kolchin, ma, & quataert 2006 ; hopkins et al. naab et al. (2009) showed that minor mergers or the accretion of relatively low - mass satellites may be the main driver for the late evolution of sizes of massive early - type galaxies. somerville et al. (2008) showed based on a cdm model of disc formation with @xmath26 m@xmath7 that the average size of discs at fixed stellar mass was about 50 percent larger than that at @xmath10. the predicted evolution in the mean size at a fixed stellar mass since @xmath11 is about @xmath396 percent, which is comparable with our observations. the newly accreted small galaxies preferentially populate the outer region of massive galaxies. the small size growth of the present result is plausibly accounted for by minor mergers or the accretion of relatively low - mass satellites (e.g., naab et al. 2009 ; hopkins et al. naab et al. (2009) and fan et al. (2010) showed the fractional variation of the gravitational radius of the main galaxy after @xmath100 minor - merger events with @xmath374 as @xmath397 where @xmath398 and @xmath399 are final and initial galaxy sizes, @xmath400 is the fractional ratio of merging galaxy to the main galaxy. it takes @xmath221 (@xmath223) minor - merger events with @xmath401 to increase the radius by @xmath402 (@xmath403) percent for massive galaxies since @xmath10, provided that the merger mass ratio is 1:10. it would not be unreasonable that massive galaxies experienced such a small number of minor merging since @xmath4043 (e.g., bundy et al. 2009 ; lpez - sanjuan et al. 2011). deep observations will be important in constraining the exact amount (or lack thereof) and distribution of merging galaxies, and how galaxies built up with redshift. in this context, our finding demonstrates that minor mergers in massive system built up an envelope of lower surface density materials. deeper imaging observations in a wider filed with high spatial resolution and consistent analyses from the local universe to high redshift will give constraints on the compactness and the amount of low - surface brightness material at the outer envelope as a function of redshift to improve our understanding of galaxy formation.
Acknowledgments
this work has been supported in part by a grant - in - aid for scientific research (21244012) of the ministry of education, culture, sports, science and technology in japan. we thank ramsey lundock for careful reading of the manuscript. mods catalogue has been accomplished by moircs builders. we owe the present study to their dedicated efforts. 99 akiyama m., minowa y., kobayashi n., ohta k., ando m., and iwata i. 2008, apjs., 175, 1 barden a., et al. 2005, apj, 635, 959 bernardi m., roche n., shankar f., steth, r. k., 2011, mnras, 412, l6 bertin e., arnouts s. 1996, a&as, 117, 393 bezanson r., van dokkum p. g., tal t., marchesini d., kriek m., franx m., coppi p., apj., 697, 1290 bond n. a., gawiser e., koekemoer a. m., 2011, apj, 729, 48 bouwens r. j., illingworth g. d., blakeslee j. p., broadhurst t. j., franx m., 2004, apj, 611, l1 boylan - kolchin m., ma c - p., quataert e., 2006, mnras, 369, 1081 bruzual g., charlot s. 2003, mnras, 344, 1000 buitrago f., trujillo i., conselice c. j., bouwens r. j., dickinson m., yan h., 2008, apj, 687, l61 bundy k., fukugita m., ellis r. s., targett t. a., belli s., kodama t., apj, 697, 1369 carrasco e. r., conselice c. j., trujillo i., 2010, mnras, 405, 2253 cassata p., et al., 2010, apj, 714, l79 cassata p., et al., 2011, apj, 743, 96 cimatti a., et al. 2008, a&a, 482, 21 daddi e., et al. 2005, apj, 626, 680 damjanov i., et al., 2009, apj, 695, 101 fan l., lapi a., de zotti g., danese l, 2008, apj, 689, l101 fan l., lapi a., bressan a., bernardi m., de zotti g., danese l, 2010, apj, 718, 1460 fontana a., et al. 2009, a&a, 501, 15 franx m., van dokkum p. g., schreiber m. f., wuyts s., labb, toft s., 2008, apj, 688, 770 fukugita m., ichikawa t., gunn j. e., doi m., shimasaku k., schneider d. p. 1996, aj, 111, 1748 guo q. white s. d. m., 2008, mnras, 384, 2 hopkins p. f., hernquist, l., cox t. j., keres, d., wuyts s., 2009, apj, 691, 1424 hopkins p. f., bundy k., murray n., quataert e., lauer t. r., and ma, c - p., 2009, mnras, 398, 898 hopkins p. f., bundy k., hernquist l., wuyts s., cox t. j., mnras, 401, 1099 ichikawa t., kajisawa m., yamada t., akiyama m., yoshikawa t., onodera m., konishi m., 2010, apj., 709, 741 kajisawa m., et al., 2009, apj, 702, 1393 kajisawa m., et al., 2011, pasj, 63, s379 kajisawa m., yamada t., 2001, pasj, 53, 833 konishi m., et al., 2011, pasj, 63, s363 law, d. r., steidel, d. c., shapley, a. e., nagy, s. r., reddy, n. a., erb, d. k., 2012, apj, 745, 85 lpez - sanjuan c., et al. 2011, a&a, 530, a20 macarthur l. a., courteau s., 2003, apj, 582, 689. mancini c., et al., 2010, mnras, 401, 933 mcintosh d., et al., 2005, apj, 632, 191 mosleh m., williams r. j., franx m., kriek, m., 2011, apj, 727, 5 naab t., johansson p. h., ostriker j. p., 2009, apj, 699, l178 nagy s. r., law d. r., shapley a. e., steidel c. c., 2011, apj, 735, 19l oke j. b., gunn j. e. 1983, apj, 266, 713 saracco p., longhetti m., andreon s., 2011, mnras, 392, 718 saracco p., longhetti m., gargiulo a., 2011, mnras, 412, 2707 srsic j.- 1968, atlas de galaxias australes (cordoba : obs. astron.) somerville r. s., et al. 2008, apj, 672, 776 szomoru d., 2010, apj.,714, l244 stott j. p., collins c. a., burke c., hamilton - morris v., smith g. p., 2011, mnras, 414, 445 shen s., mo h. j., white s. d. m., blanton m. r., kauffmann g., voges w., brinkmann j., csabai i., 2003, mnras, 343, 978 stockton a., shih h - y., larson k., 2010, apj, 709, l58 suzuki r., et al. 2008, pasj, 60, 1347 toft s., et al., 2007, apj, 671, 285 trujillo i., et al. 2006, apj, 650, 18 trujillo i., conselice c. j., bundy k., cooper m. c., eisenhardt p., ellis r. s. 2007, mnras, 382, 109 trujillo i., cenarro a. j., de lorenzo - cceres a., vazdekis a.,, de la rosa i. g., cava a., apj., 692, l118 trujillo i., ferreras i., de la rosa i. g., 2011, mnras, 415, 3903 valentinuzzi t. et al. 2010, apj, 712, 1232 van der wel a., bell e. f., van den bosch f. c., gallazzi a, rix h - w., 2009, apj., 698, 1232 van der wel a., holden b. p., zirm a. w., franx m., rettura a., illingworth g. d., ford h. c., 2008, apj, 688, 48 van der wel a., et al., 2011, apj, 730, 38 van de sande j., et al. 2011, apj, 736, l9 van dokkum p. g., et al., 2008, apj, 677, l5 van dokkum p. g., kriek m., franx m., 2009, nat, 460, 717 van dokkum p. g., et al., 2010, apj, 709, 1018 williams r. e., et al. 1996, aj, 112, 1335 williams r. j., quadri r. f., franx m., van dokkum p., toft s., kriek m., labb i., 2010, apj, 713, 738 yamada t., et al., 2009, apj, 699, 1354 zirm a. w., et al., 2007, apj, 656, 66 | we present scaling relations between stellar - mass (@xmath0) and the size of galaxies at @xmath1 for half- (@xmath2) and 90 percent - light (@xmath3) radii, using a deep @xmath4-band selected catalogue taken with the subaru telescope and moircs in the goods - north region.
the logarithmic slope @xmath5 is independent of redshift in a wide mass range of @xmath6 m@xmath7, irrespective of galaxy populations (star - forming, quiescent). the offset change is @xmath8 percent.
provided that optical light in the rest frame traces the stellar mass of galaxies, the universal relation demonstrates that the stellar mass was built up in galaxies over their cosmic histories in a similar manner on average irrelevant to galaxy mass.
the small offset in each stellar mass bin from the universal relation shows weak size evolution at a given mass. there is a moderate increase of 3050 percent for @xmath2 and @xmath3 for less massive galaxies (@xmath9 m@xmath7) from @xmath10 to @xmath11,
while the sizes remains unchanged or slightly decrease towards @xmath12. for massive galaxies (@xmath13 m@xmath7), the evolution is @xmath14 % increase in @xmath3 from @xmath10 to @xmath12, though that in @xmath2 is weaker.
the evolution of compactness factor, @xmath15, which becomes smaller at lower redshift, is suggestive of minor merging effect in the outer envelope of massive galaxies.
[firstpage] galaxies : evolution galaxies : fundamental parameters galaxies : high - redshift infrared : galaxies. | 1202.1138 |
Introduction
thermopower, @xmath1, is an important transport coefficient that offers valuable information about the electronic structure, the scattering processes and the mechanisms of carrier - phonon coupling in a system. in the last few years there has been growing experimental interest in @xmath1 of single wall carbon nanotubes (swcnts). several groups have reported thermopower measurements on bulk swcnt materials (e.g., mats, fibers, films) @xcite and on individual swcnts @xcite. however, only modest progress has been made up to now in understanding the unique features of @xmath1 in these systems. interesting issues concerning the large positive thermopower (@xmath2 80 @xmath3v / k) in pristine samples @xcite, the change of sign of @xmath1 upon exposure to oxygen @xcite and the effect of carrier - phonon coupling @xcite on @xmath1 still remain open. @xmath1 consists of two additive contributions which are diffusion, @xmath4, and phonon - drag, @xmath0. @xmath4 is due to the carrier diffusion in the presence of a temperature gradient and for degenerate systems varies linearly with @xmath5 according to mott s expression. @xmath0 originates from the interchange of momentum between acoustic phonons and carriers via the carrier - phonon interaction. the first theoretical models for the study of the phonon drag in metals @xcite and semiconductors @xcite were developed half a century ago. more recently, extensive theoretical and experimental work has been carried out on @xmath0 of low - dimensional semiconductor structures @xcite. recent experiments on @xmath1 in p - doped swcnt films and fibers @xcite provided clear evidence for the presence of @xmath0 at @xmath6 k. on the theory level, however, there is still an ongoing discussion about the role of @xmath0 in measured thermopower @xcite. so far, the theoretical studies of @xmath0 are confined to metallic armchair (10,10) tubes @xcite. however, in perfect metallic tubes with mirror electron - hole symmetry both @xmath4 @xcite and @xmath0 @xcite are expected to be negligibly small compared to the experimental data, due to the competition between the opposite contributions of electrons and holes. we note that the accuracy of the existing theoretical models @xcite for @xmath0 in metallic tubes has been questioned recently by mahan @xcite. also, a recent theoretical work @xcite pointed out that thermopower vanishes in one - dimensional conductors with a linear energy dispersion (as in the case of metallic tubes) due to electron - hole symmetry. in this paper we propose a theoretical model for the phonon - drag thermopower in semiconducting swcnts that are characterized by a non - linear energy dispersion. (a brief discussion on the behavior of @xmath0 in this kind of nanotubes appears in [].) we suggest that the measured thermopower in doped samples is due to the contribution of degenerate semiconducting nanotubes. in our model @xmath0 originates from carrier - phonon intraband scattering within the first 1d subband. as we discuss below, the dominant contribution to @xmath0 is made by long - wavelength acoustic phonons that backscatter carriers across the fermi surface. in this case the carrier - phonon coupling is much weaker in metallic tubes than in semiconducting tubes @xcite and, consequently, @xmath0 is expected to be substantially larger in the latter ones. we note that upon chemical or electrostatic doping the fermi level can be pushed into the conduction or valence band and the degenerate semiconducting tubes can be considered as one - dimensional metals. therefore the terms metallicand semiconductingrefer only to the different electronic structure in the two types of tubes (see, for example, ref. there are two equivalent theoretical approaches to the problem of phonon drag@xcite. in the first approach phonons are perturbed in the presence of a weak temperature gradient @xmath7. non - equilibrium phonons transfer part of their momentum to carriers due to the carrier - phonon coupling. then the phonon - drag contribution to the thermoelectric current @xmath8 is calculated by solving the coupled boltzmann equations for carriers and acoustic phonons @xcite. the phonon - drag thermopower is readily obtained by @xmath9 where @xmath10 is the carrier conductivity. in the second approach carriers are accelerated isothermally in the presence of a weak electric field @xmath11 and impart some of their momentum to phonons due to the carrier - phonon coupling. then the resulting phonon heat current and the phonon - drag contribution to the peltier coefficient is calculated @xcite. this method of evaluating @xmath0 is referred as @xmath12-approach @xcite because it provides a direct estimation of the peltier coefficient. the equivalence of the above two approaches is secured by onsager s symmetry relation. in this paper we follow the second approach which is more general and it can be applied even in systems where carriers do not behave semiclassically @xcite. the paper is organized as follows. in sec.ii we introduce the theoretical model for the calculation of @xmath0 in the semiclassical transport regime. an explicit expression for @xmath0 is derived in sec.iib and in sec.iic we derive a simple approximate expression for @xmath0 for the case of a highly degenerate semiconducting tube. numerical results for @xmath0 as a function of temperature, tube radius and position of fermi level are presented in sec.iii. in the same section we discuss the effect of screening. in sec.iv we compare our theory with available experimental data for acid - doped bulk swcnt samples.
Theory
we assume that the nanotube is a long indefinitely thin cylinder of radius @xmath13 and length @xmath14. the nanotube axis is along the @xmath15direction. the carrier wave function is @xcite @xmath16 where, @xmath17 is the space vector, @xmath18 is the carrier wave vector along the axial direction, @xmath19 is the azimuthal angle and @xmath20 labels 1d orbital subbands associated with the carrier confinement along the circumference. we assume, that the fermi level, @xmath21, is located between the first and the second 1d subbands (i.e., only the ground subband is occupied). then, the carrier energy is @xmath22 where @xmath23 is the carrier effective mass and @xmath24 denotes the position of the first van hove singularity. in carbon nanotubes phonons also exhibit 1d character. the lattice displacement at a point @xmath17 is @xcite @xmath25 where, @xmath26 is the polarization vector, @xmath27 is the phonon wave vector in the axial direction and @xmath28 denotes the phonon modes associated with phonon confinement along the circumference. due to the conservation of angular momentum only the three low - energy acoustic modes with @xmath29 (the so - called twisting, stretching and breathing modes) contribute to the carrier - phonon intraband scattering. the phonon frequencies and polarization vectors have been calculated within the continuum model proposed by suzuura and ando @xcite. the carrier - phonon interaction in carbon nanotubes has been studied in several texts within the tight - binding approximation @xcite or a continuous elastic theory @xcite. here we follow the continuous model of suzuura and ando @xcite according to which the carrier - phonon coupling is described via the acoustic deformation potential @xmath30 where @xmath31 is the deformation potential constant. the deformation potential approximation provides a good description of the carrier interaction with long - wavelength acoustic phonons. the last term in eq. ([def]) accounts for the nonzero curvature of the nanotube @xcite. the twisting mode does not participate to carrier - phonon scattering via the deformation potential coupling. moreover, in the long - wavelength limit (@xmath32), which is the regime of our interest, the breathing mode is dispersionless and does not contribute to @xmath0. thus, in what follows we consider only the stretching mode which is characterized by a linear dispersion @xmath33 where, @xmath34 is the sound velocity. the phonon polarization vector, @xmath35, for this mode in the limit @xmath32 is @xmath36 where @xmath37 and @xmath38 is poisson s ratio. ignoring the terms proportional to @xmath39 the above expression becomes identical with the one derived by de martino _ et al_. @xcite. we assume a small electric field @xmath40 in the axial direction of the nanotube. the presence of @xmath40 creates a net flux of carriers along the axis of the tube which results in a momentum transfer to phonons through the carrier - phonon coupling. we calculate the resulting phonon heat flux @xmath41 and obtain the phonon - drag contribution to the transport coefficient @xmath42 to get @xmath0 we utilize the onsager s relation @xmath43 where @xmath10 is the carrier conductivity and @xmath5 the absolute temperature. the phonon heat flux is given by @xmath44 where @xmath45 is the phonon group velocity and @xmath46 is the first order perturbation of the phonon distribution function. the perturbation @xmath47 is determined by the steady - state boltzmann equation for phonons in the relaxation time approximation when @xmath48. namely, @xmath49 where @xmath50 is the phonon relaxation time associated with phonon - phonon collisions and phonon scattering by imperfections. for simplicity we have ignored the dependence of @xmath50 on @xmath27. @xmath51 is the rate of change of the phonon distribution function @xmath52 due to phonon scattering by carriers. it is written in the standard form @xmath53 where @xmath54 and @xmath55 are the spin and the valley degeneracies, respectively, @xmath56 is the carrier distribution function and @xmath57 are the transition rates at which the carrier in a state @xmath18 is promoted to a state @xmath58 by absorbing (emitting) one phonon with wave vector @xmath27. when the external field @xmath40 is weak eq. ([ph - c]) is linearized and is solved in terms of @xmath47. then we get @xmath59 where, @xmath60 is the phonon relaxation time associated with scattering by carriers given by @xmath61,\]] and @xmath62 is the average equilibrium rate of absorption of phonons with wave vector @xmath27. it is given by @xmath63 where @xmath64 + 1\}^{-1}$] (with @xmath65) is the fermi - dirac function and @xmath66 denotes the transition rate in equilibrium. assuming that phonon - phonon scattering and phonon scattering by impurities dominate over the phonon - carrier scattering (@xmath67), eqs.([boltzmann]) and ([n1nn]) give @xmath68 in the above equation @xmath69 is the first order perturbation of the carrier distribution function. it is worth noting that eq. ([n1]) can be regarded as a starting point for the calculation of @xmath0 in all the problems treated within the @xmath12-approach@xcite. now, by substituting the phonon perturbation into ([q]) we take for the heat flux @xmath70 to determine the perturbation of the carrier distribution function @xmath69 entering eq. ([heat]) we use the 1d steady - state boltzmann equation @xmath71 where @xmath72 is the carrier charge and the rhs of eq. ([boltzmann2]) is the rate of change of the carrier distribution function due to elastic collisions with static imperfections. in the relaxation time approximation this term is written as @xmath73 where @xmath74 is the carrier relaxation time. equation ([boltzmann2]) is linearized to give @xmath75 where, @xmath76 is the carrier group velocity. by substituting eq. ([f1]) into ([heat]) and making use of eqs.([m]), ([onsager]) and ([gamma]) we finally get @xmath77\nonumber\\ \times f^{0}_{k}(1-f^{0}_{k^{\prime}})p^{a0}_{q}(k, k^{\prime}).\end{aligned}\]] the above expression is equivalent to the expression derived by kubakaddi and butcher@xcite for a quantum wire coupled to 3d phonons. the authors in ref. [] followed a different approach than this described here. they followed bailyn s theory @xcite and they calculated the phonon - drag contribution to the thermoelectric current that originates from the carrier scattering with non - equilibrium phonons in the presence of a small temperature gradient across the wire. their calculation was based on the solution of the coupled equations for electrons and phonons. the transition rate @xmath66 is calculated by using fermi s golden rule. the lattice displacement for the stretching mode is written in second quantized form @xmath78 where @xmath79 and @xmath80 are the phonon creation and annihilation operators, respectively, @xmath81 is the nanotube surface area and @xmath82 is the mass density. for the stationary carrier states considered here one easily finds @xmath83 where, @xmath84^{-1}$] is the phonon distribution in equilibrium, @xmath85 is the square of the carrier - phonon matrix element for the deformation potential coupling and @xmath86 is the 1d static dielectric function. by utilizing eqs. ([def]) and ([eta]) the matrix elements @xmath85 in the limit @xmath32 are written as @xmath87 where @xmath88. we note that the @xmath27-dependence of @xmath85 is typical for the carrier interaction with longitudinal acoustic phonons via an isotropic deformation potential @xcite. a similar expression to the one we derive here is given in ref. the dielectric function for a 1d gas confined to the surface of the carbon nanotube is calculated within the random phase approximation @xcite. for the carrier wave functions considered here we obtain @xmath89 where @xmath90 and @xmath91 are the modified bessel functions of the first and the second kind, respectively, and @xmath92 is the background dielectric constant. @xmath93 is the standard factor that accounts for finite temperature effects on the static polarization function @xcite @xmath94}.\]] to obtain an explicit expression for @xmath0 we substitute eq. ([rate]) into ([sgs1]). then the summation over @xmath58 is readily carried out by replacing @xmath58 by @xmath95 as a consequence of the momentum conservation condition imposed by the kronecker symbol @xmath96. moreover, the summations over @xmath27 and @xmath18 are transformed to the integrals @xmath97 the presence of the @xmath98function in eq. ([rate]) allows the immediate evaluation of the @xmath18integration. we see by inspection that @xmath99 where, @xmath100. the minus and the plus signs correspond, respectively, to positive and negative @xmath27. now, after some algebra, we finally obtain @xmath101 where, @xmath102 is the phonon - mean - free path, @xmath103^{1/2}$] is the fermi wave number and @xmath104 is the product of the fermi occupation factors @xmath105\]] with @xmath106. in deriving eq. ([sgf]) we have ignored the energy dependence of the carrier relaxation time and in eq. ([sgs1]) we have replaced @xmath74 by its value at the fermi level, @xmath107. this is a good approximation when @xmath108 @xcite. moreover, we have replaced @xmath10 by @xmath109 where @xmath110 is the density of carriers per unit length. interestingly, @xmath0 becomes independent of the carrier relaxation time. at low @xmath5 and assuming that @xmath111 is a small quantity compared to @xmath21 the product @xmath104 is approximated by @xcite @xmath112 with @xmath106. the @xmath98-function can be written in the following form @xmath113.\]] we see that @xmath114 resonates at @xmath115 for positive @xmath27. when the expression ([appr3]) for @xmath104 is substituted into eq. ([sgf]) the integration over @xmath27 is carried out straightforwardly by using the condition @xmath115. we note that @xmath116 and consequently, stretching phonons with @xmath117 make the dominant contribution to @xmath0. equation ([sgf]) is now significantly simplified and is written in the convenient approximate form @xmath118 where @xmath119 is given by @xmath120 in the above equations, @xmath121 is the frequency of a stretching phonon with @xmath117 and @xmath122 is an approximate expression for the dielectric function. to obtain @xmath122 we replace @xmath27 by @xmath123 in the denominator and in the arguments of the modified bessel functions @xmath90 and @xmath91 in eq. ([dielectric]). the factor @xmath93 is replaced by the average @xmath124 that is given by the expression @xmath125 @xmath124 has been evaluated numerically for several values of @xmath126 and @xmath13 and we find that in the degenerate limit and when @xmath127 (where @xmath128 is the fermi temperature) the following expression provides a very good fit @xmath129\]] where, @xmath130, @xmath131, @xmath132 and @xmath133. at low @xmath5 the effect of screening is severe and unity can be neglected in eq.([dielectric]). in this case the @xmath5-dependence of the dielectric function is described by eq.([aver2]). at temperatures where @xmath134 the dielectric function shows a weak @xmath5-dependence. then @xmath0 follows the law @xmath135 this activated behavior is characteristic in 1d systems where the fermi surface consists of two discrete points @xmath136 @xcite. we note that when screening is ignored and the phonon mean - free path is constant the @xmath5-dependence of @xmath0 given by eq. ([appr1]) is similar to what predicted by scarola and mahan @xcite for an armchair (10,10) metallic swcnt due to interband electron scattering between the two linear bands. however, the absolute magnitude of @xmath0 in a metallic tube is expected to be much lower than that predicted in ref. [] due to the competing contributions of electrons and holes to the thermoelectric current.
Numerical results
we assume that the free carriers are holes and we examine the dependence of @xmath0 on temperature, the radius of the nanotube and the position of the fermi level with respect to the position of the first van hove singularity. the analysis is the same for the case of electrons with the only difference being the sign of @xmath0. the values for the material parameters used in the calculations are @xmath137, @xmath138 ev @xcite, @xmath139 @xcite, @xmath140 @xcite, @xmath141 kgr / m@xmath142 and @xmath143 km / s @xcite. the hole effective mass is taken to be @xmath144 where @xmath145 is the tube radius in nm @xcite. we assume that @xmath146 @xmath147. against temperature for a swcnt of radius 0.5 nm. results are shown for four values of @xmath148 : 60 mev(dotted line), 90 mev (dashed - dotted line), 120 mev (dashed line) and 150 mev (solid line) when screening is taken into account (a) and when screening is ignored (b). the phonon mean - free path is taken to be 1 @xmath3 m. the inset shows the ratio @xmath149 as a function of @xmath150.,height=340] in fig. 1a we show the @xmath0 evaluated from eq. ([sgf]) for a p - type swcnt of radius @xmath151 nm as a function of @xmath5. the solid, dashed, dashed - dotted and dotted lines correspond, respectively, to @xmath152, 120, 90 and 60 mev. we note that at temperatures where the carriers are non - degenerate we have taken into account the thermal broadening effects on @xmath10. to assure the accuracy of the approximate expression ([appr1]), in the inset of fig. 1 we show the ratio @xmath149 as a function of @xmath150 for @xmath153 mev. @xmath0 and @xmath154 are calculated from eqs. ([sgf]) and ([appr1]), respectively. calculations of @xmath155 for 90, 120 and 150 mev also fall on to the same curve. we can see that in the degenerate limit the approximate result agrees very well with the exact expression for @xmath0. finally, in fig. 1b @xmath0 is calculated in the absence of screening, @xmath156. it turns out that screening induces a strong suppression of @xmath0 by 1 - 2 orders of magnitude. inspection of eq. ([dielectric]) shows that screening effects become more severe as @xmath13 decreases. we note that in the absence of screening @xmath0 levels off at high @xmath5 in agreement with previous estimations in metallic swcnts @xcite. however, when screening is introduced @xmath0 shows a quasi - linear @xmath5-dependence at high @xmath5 due to the temperature dependence of the dielectric function. the dielectric function @xmath122 as a function of the inverse temperature for a swcnt with @xmath157 nm is shown in fig.2 in fig.3 we show the dependence of @xmath0 on the fermi level with respect to the position of the first van hove singularity for temperatures @xmath158 k. the shown structure is due to two competing mechanisms which are the suppression of the carrier - phonon scattering and the increase of @xmath159 as @xmath126 increases. the tube radius is 0.5 nm. finally, in fig.4 we present the calculated values of @xmath0 as a function of the nanotube radius. at temperatures higher than 100 k we find that @xmath0 follows a law close to @xmath160. at lower temperatures @xmath0 shows a weaker dependence on @xmath13 especially at large values of r.
Comparison with the experiment-discussion
so far there is no clear evidence about the phonon - drag effect in isolated swcnts. the most relevant experiments were performed by yu et al. @xcite in an individual swcnt at temperatures above 100 k. the observed thermopower showed a linear @xmath5-dependence which was attributed to the linear diffusion component and a constant phonon - drag component of about 6 @xmath3v / k without, however, excluding the possibility of an additional contact effect. according to our analysis in section iii, the phonon - drag thermopower at relatively high temperatures shows a quasi - linear @xmath5-dependence and this makes difficult the separation of the diffusion and the phonon - drag contributions. nevertheless, when the calculated values for @xmath0 shown in fig.1a are fitted by a linear function of @xmath5 we find that the intercepts vary from 1.4 to 7.4 @xmath3v / k when the position of the fermi level with respect to the first van hove singularity varies from 60 to 150 mev. these values are in agreement with the experimental estimation of @xmath0 in ref. we note that the intercepts depend linearly on the phonon mean - free path and vary approximately as @xmath161. vavro _ et al_. @xcite and zhou _ et al. _ @xcite have reported thermopower measurements in p - doped bulk swcnt samples in a wide temperature range (10 - 200 k) that show clearly the signature of phonon drag. normally, in bulk samples nanotubes are self - organized into long ropes, which contain a large number of nanotubes (tens to hundreds) @xcite, forming a 3d network of complex geometry. thermopower in these nanotube networks exhibits a very similar behavior as this of an individual nanotube described in section iii. we have recently proposed a simple argument based on a model of parallel conductors which suggests that in a network with homogeneous doping and with a narrow distribution of tube diameters the measured thermopower resembles that of an individual tube @xcite. the resistivity measurements in the samples under consideration showed weak coupling between metallic nanotubes @xcite and hence the contribution from metallic tubes to the total conductivity is neglected. we also recall that the contribution of metallic tubes in @xmath1 is expected to be small compared to this of semiconducting tubes. therefore, we can use the theory for isolate semiconducting swcnts developed here to interpret the data in @xcite. in fig.5 the circles are the measured thermopower for a bulk sample prepared by pulsed laser vaporization (plv) and doped with hno@xmath162@xcite. the tube radius is @xmath163 nm. at low temperatures (up to 100 k) we fit the data for the total thermopower, @xmath1, by the expression @xmath164 the first term is the approximate expression ([appr1]) for @xmath0 and the second term corresponds to the diffusion component @xmath4. the sample is highly degenerate and at temperatures up to 100 k eq.([appr1]) accurately describes @xmath0. the @xmath5-dependence introduced by the dielectric function is given by eq. ([aver2]). the values we obtained for the parameters @xmath126, @xmath165 and @xmath166 are shown in table i. the logarithmic term in @xmath4 secures an excellent fit to the measured thermopower at all temperatures up to 100 k. if this term is neglected the theoretical values for the total thermopower are significantly larger than the experimental ones at high temperatures. we speculate that the @xmath167 term in @xmath4 is due to 2d weak localization (wl) effects @xcite. if this speculation is valid we would also expect a signature of wl in the conductivity measurements. we note that the relative change in conductivity should be the same as in @xmath4 but with an opposite sign @xcite. interestingly, we find that at temperatures @xmath168 k the conductivity follows the law @xmath169 where the value of @xmath170 is shown in table i. we see that @xmath166 and @xmath170 agree to each other remarkably well. the origin of the 2d wl in these samples is not well understood. it is likely related to the individual rope although in this case a 1d localization behavior would be expected @xcite. however, the phase coherence length, in the samples we discuss here, is comparable to the diameter of the rope @xcite and the 2d limit might be approached. et al. _ @xcite have also observed a @xmath171 dependence of the conductance for an individual multiwall cnt at 0.1 - 100 k which was attributed to 2d wl. finally, we should remark that wl is expected to have a negligible effect on @xmath0 @xcite. now, by using the values for @xmath126 we obtained from the fitting of the thermopower data at low @xmath5 we calculate @xmath0 in the whole temperature range from 10 to 200 k by using the exact expression ([sgf]). the only remaining unknown is the value of the phonon - mean - free path @xmath172 which is determined from the experimental data when the diffusion contribution is subtracted. we find that @xmath173 nm. this value is consistent with the values 0.25 - 0.75 @xmath3 m reported recently for an individual swcnt @xcite. our estimation for the total thermopower is shown as solid line in fig.5. the dashed and the dashed - dotted lines correspond to the phonon - drag and the diffusion contributions, respectively. by following a similar procedure as this described above we have interpreted the thermopower data for another bulk sample prepared by high pressure decomposition of co (hipco) and doped with h@xmath174so@xmath175 @xcite. the tube radius varied from 0.4 - 0.7 nm. conductivity measurements for this sample (designated as hpr93c) appear in @xcite. the experimental data for the ratio @xmath176 are shown as squares in fig.6. the values for the fitting parameters @xmath126, @xmath165 and @xmath166 are shown in table i. we also present the value of @xmath170 for comparison. in the calculations the tube radius is taken to be the average @xmath177 nm while for the phonon - mean - free path we obtained the value @xmath178 nm. the calculated values for @xmath176 is shown as solid line in fig.6..the values for the parameters @xmath165, @xmath166 and @xmath126 obtained from the fit of the thermopower data @xcite for @xmath179 k by using eq. ([fit]). in the last column we show for comparison the values for @xmath170 obtained from the resistivity data @xcite in the range 10 - 100 k. [cols="^,^,^,^,^,^",options="header ",] concerning the consistency of the fitting parameters @xmath165 and @xmath126 we should make the following remarks. by using the values for @xmath126 shown in table i and a simple tight binding model for the estimation of the first van hove singularity (see, for example, ref. []) the values we get for @xmath21 are in good agreement with those determined from reflectivity and raman measurements @xcite. also, @xmath165 varies inversely with @xmath180 in agreement with mott s expression for @xmath4. moreover, the values we extract for @xmath126 support recent arguments according to which h@xmath174so@xmath175 is a stronger dopant than hno@xmath162 @xcite. namely, according to our estimation for the fermi wave numbers, the fermi level is shifted by 94 and 155 mev below the top of the valence band for the plv+hno@xmath162 and hipco+h@xmath174so@xmath175 samples, respectively. in order to show clearly the effect of phonon drag in fig.6 we have plotted the ratio @xmath176 as a function of @xmath5. the circles and the squares are the measured values for the samples plv+hno@xmath162 and hipco+h@xmath174so@xmath175, respectively. the dashed lines are the theoretical estimates for @xmath0 and the solid lines are the calculated values for the total thermopower. the peaks at @xmath181 are associated to phonon - drag thermopower. the shift between the experimental and the theoretical values for @xmath182 is due to the logarithmic term in @xmath4. the position of the peak moves towards to higher temperatures as doping increases. this dependence can be understood by maximizing the ratio @xmath183 using eq. ([appr1]). then we get the following dependence @xmath184 it is important to add that the exponential suppression of @xmath0 at low temperatures is unique for 1d systems. in higher dimensions @xmath0 exhibits a power - law @xmath5 dependence at low temperatures @xcite. the observed peak in @xmath176, which is ascribed to phonon drag, underlies the 1d character of thermopower. this adds another confirmation that thermopower in bulk carbon nanotube - based materials is a property of the individual tube rather than a property of the network.
Conclusions
in summary, we have presented a rigorous model for the calculation of the phonon - drag thermopower in degenerately doped semiconducting swcnts. by using the derived model we investigated the dependence of @xmath0 on temperature, tube radius and position of the fermi level. we found that @xmath0 decreases with the increase of the tube radius following approximately a @xmath161 law at high temperatures. in the degenerate limit, we derive a simple expression for @xmath0 which can be used as a probe for the estimation of the free carrier density in doped tubes. according to this expression @xmath0 shows an activated @xmath5 dependence at low temperatures. screening effects of the carrier - phonon coupling reduce the magnitude of @xmath0 severely and result to a quasi - linear @xmath5-dependence of phonon drag at high @xmath5. finally, we have compared our model with available data in acid - doped bulk samples @xcite and we found a very good agreement in a wide temperature range.
Acknowledgements
the author wishes to thank dr. k. papagelis for extensive and useful discussions and prof. r. fletcher for stimulating remarks. 99 l. grigorian, k.a. williams, s. fang, g.u. sumanasekera, a.l. loper, e.c. dickey, s.j. pennycook and p.c. eklund, phys. lett. * 80 *, 5560 (1998) ; l. grigorian, g.u. sumanasekera, a.l. loper, s. fang, j.l. allen and p.c. eklund, phys. b * 58 *, r4195 (1998). afonin, y.m. galperin and v.l. gurevich, zh. fiz * 87 *, 335 [sov jetp * 60 *, 194] (1984) ; m.j. kearney and p.n. butcher, j. phys. c : solid state phys. * 21 *, l265 (1988) ; c. castellani, c. di castro, m. grilli and g. strinati, phys. rev. b * 37 *, 6663 (1988). a. miele, r. fletcher, e. zaremba, y. feng, c.t. foxon and j.j. harris, phys. b * 58 *, 13181 (1998) ; c. rafael, r. fletcher, p.t. coleridge, y. feng and z.r. wasilewski, semicond. * 19 *, 1291 (2004). | a theoretical model for the calculation of the phonon - drag thermopower, @xmath0, in degenerately doped semiconducting single - wall carbon nanotubes (swcnts) is proposed.
detailed calculations of @xmath0 are performed as a function of temperature, tube radius and position of the fermi level.
we derive a simple analytical expression for @xmath0 that can be utilized to determine the free carrier density in doped nanotubes. at low temperatures
@xmath0 shows an activated behavior characteristic of the one - dimensional (1d) character of carriers.
screening effects are taken into account and it is found that they dramatically reduce the magnitude of @xmath0.
our results are compared with previous published experimental data in bulk p - doped swcnt materials.
excellent agreement is obtained in the temperature range 10 - 200 k for a consistent set of parameters.
this is a striking result in view of the complexity of these systems. | 0912.4700 |
Introduction
the group of about @xmath8 so - called central compact objects (ccos) in supernova remnants (snrs) are distinguished by their steady surface thermal x - ray flux, lack of surrounding pulsar wind nebula, and non - detection at any other wavelength @xcite. three ccos are known pulsars, with periods in the range @xmath9 s, and spin - down rates that provide an estimate of their surface dipole magnetic field strength, which falls in the range @xmath10 g @xcite, smaller than that of any other young neutron star (ns). this weak magnetic field is evidently the physical basis of the cco class. the homogeneous properties of the approximately seven remaining ccos that have not yet been seen to pulse suggest that they have similar or even weaker @xmath7-fields than the known cco pulsars, and a more uniform surface temperature. that ccos are found in snrs (of ages @xmath11 yr) in comparable numbers to other classes of nss implies that they must represent a significant fraction of ns births, probably greater than that of magnetars, for example, as only 45 galactic snrs are known to host magnetars @xcite. the subsequent evolution of ccos is a glaring unknown, their immediate descendants not being evident in any existing survey. ccos should persist as cooling nss, detectable in thermal x - rays, for @xmath12 years according to ns cooling curves @xcite. if some are also radio pulsars, that phase could last for @xmath13 years. while there are not yet enough ccos to know whether they are intrinsically radio - quiet, it is very unlikely that the huge expected population of cco descendants are _ all _ hiding simply due to unfavorable radio beaming. therefore, it is difficult to understand why the region of (@xmath14) space in which ccos are found, between the bulk of the ordinary radio pulsars and the recycled `` millisecond''pulsars in binary systems, is relatively empty. most of the pulsars in this sparse region (see figure [fig : ppdot]) are thought to be `` mildly recycled, '' having been spun up by accretion from a high - mass companion for a relatively short time before a second sn occurred. defined as having @xmath0 ms and @xmath1 g, mildly recycled pulsars include double ns systems, and single ones thought to be the disrupted recycled pulsars (drps, @xcite) ejected when the binary is unbound after the second sn. (these are in contrast to the millisecond pulsars, which have low - mass companions.) the drps have characteristic ages @xmath15 of @xmath16 yr. historically, it was thought that hardly any pulsars are born with @xmath17 g, so that all such pulsars must be recycled. but the discovery of young ccos in this region of parameter space invalidates that assumption. just as the @xmath18 yr characteristic age of a cco is meaningless, the possibility that _ any _ low @xmath7-field radio pulsar is much younger than its characteristic age may now be considered. lllrrclrccc j0609@xmath192130 & 06 09 58.89 & @xmath1921 30 02.8 & 56 & @xmath20 & 38.73 & 1.2 & 22 & @xmath21 & @xmath22 & 1 + j1038@xmath190032 & 10 38 26.93 & @xmath1900 32 43.6 & 29 & @xmath23 & 26.59 & 1.2 & 880 & @xmath24 & @xmath25 & 2 + j1320@xmath263512 & 13 20 12.68 & @xmath2635 12 26.0 & 458 & @xmath27 & 16.42 & 0.68 & 310 & @xmath28 & @xmath29 & 3 + j1333@xmath264449 & 13 33 44.83 & @xmath2644 49 26.2 & 346 & @xmath30 & 44.3 & 1.4 & 410 & @xmath31 & @xmath32 & 4 + j1339@xmath264712 & 13 39 56.59 & @xmath2647 12 05.5 & 137 & @xmath33 & 39.9 & 1.2 & 310 & @xmath34 & @xmath35 & 4 + j1355@xmath266206 & 13 55 21.34 & @xmath2662 06 20.1 & 277 & @xmath36 & 547 & 8.3 & 22 & @xmath37 & @xmath38 & 5 + j1548@xmath264821 & 15 48 23.26 & @xmath2648 21 49.7 & 146 & @xmath39 & 126.0 & 4.4 & 360 & @xmath40 & @xmath41 & 5 + j1611@xmath265847 & 16 11 51.31 & @xmath2658 47 42.3 & 355 & @xmath42 & 79.9 & 1.7 & 160 & @xmath43 & @xmath44 & 6 + j1753@xmath261914 & 17 53 35.17 & @xmath2619 14 58 & 63 & @xmath45 & 105.3 & 2.2 & 130 & @xmath46 & @xmath47 & 6 + j1816@xmath265643 & 18 16 36.46 & @xmath2656 43 42.1 & 218 & @xmath48 & 52.4 & 1.6 & 470 & @xmath49 & @xmath50 & 4 + j1821@xmath190155 & 18 21 38.88 & @xmath1901 55 22.0 & 34 & @xmath51 & 51.75 & 1.8 & 24 & @xmath52 & @xmath53 & 7 + b1952@xmath1929 & 19 54 22.55 & @xmath1929 23 17.3 & 427 & @xmath54 & 7.932 & 0.70 & 9 & @xmath55 & @xmath56 & 8 + j2007@xmath192722 & 20 07 15.83 & @xmath1927 22 47.91 & 24 & @xmath57 & 127.0 & 5.4 & 250 & @xmath58 & @xmath59 & 9 + j2235@xmath191506 & 22 35 43.70 & @xmath1915 06 49.1 & 60 & @xmath60 & 18.09 & 1.1 & 630 & @xmath61 & @xmath62 & 10 [tab : drps] -@xmath63 diagram, including magnetars (blue crosses), inss (magenta asterisks), ccos (filled red stars) and drps (open blue stars). black dots are isolated pulsars and circled dots are pulsars in binaries. (pulsars in globular clusters are excluded as their period derivatives are not entirely intrinsic.) dashed lines of constant characteristic age and magnetic field are indicated.] the majority of ccos may have magnetic fields even weaker than those of the known cco pulsars, and may fall among the drps in (@xmath14) space. once the snr associated with a cco has dissipated, it would be difficult to distinguish an `` orphaned cco '' from a drp by timing alone if some ccos are radio pulsars. thermal x - ray emission, however, would allow a recently orphaned cco to be recognized as such up to @xmath64 yr. thermal emission from the cooling ns is the diagnostic that would distinguish an evolving cco from an old drp, whose negligible rotation - powered x - ray emission, thermal or non - thermal, would be orders of magnitude weaker. in this paper, we report an x - ray search for orphaned ccos from among the population drps, whose timing parameters are expected to be comparable. in section [sec : observations] we describe the new and archival observations of the drps. section [sec : results] gives the resulting upper limits on their temperatures and luminosities. in section [sec : discussion] we discuss the implication of these results for the possible evolutionary tracks of ccos.
Observations
our targets selected for x - ray observations are the 12 radio pulsars classified as drps by @xcite, plus the recently discovered psr j2007 + 2722 @xcite. these comprise all but one of the isolated pulsars in the galactic disk with magnetic field strength @xmath65 g and spin period @xmath0 ms listed in the atnf catalog (*??? * v1.46). their properties are listed in table [tab : drps]. the latest drp, psr j1821 + 0155 @xcite, the @xmath66 member of the class, was discovered too recently to be included in our x - ray sample. for ten of these objects not already observed in x - rays, we obtained 3.5 ks observations to search for point - like emission at their known (subarcsecond) radio locations. we justified this short observing time based on its ability to detect thermal emission from a cooling ns younger than @xmath67 yr, while thermal or nonthermal emission from a @xmath68 yr old drp would be many orders of magnitude less. detailed calculations of the detection limits on temperature and luminosity from these observations are presented below. we also analyzed 5 ks archival exposures on psr j0609@xmath192130 and psr b1952@xmath1929, and tabulate our prior results for psr j2007@xmath192722 @xcite. all observations were taken with the advanced ccd imaging spectrometer (acis, *??? *), operating in timed / faint exposure mode, with the targets placed on the back - illuminated ccd. acis has @xmath69 pixels, comparable to the on - axis point - spread function. the nominal acis pointing uncertainty is a radius of @xmath70. all data reduction and analysis was performed with the interactive analysis of observation software (ciao, *??? *) version 4.5, using the calibration database (caldb) v4.1.3. the background rates for these observations showed no evidence of flaring behavior, and the full exposure time was retained for analysis in each case. cccccl j0609@xmath192130 & 12687 & 4.99 & @xmath71 & @xmath72 & @xmath73 + j1038@xmath190032 & 13801 & 3.50 & @xmath74 & @xmath75 & @xmath73 + j1320@xmath263512 & 13797 & 3.42 & @xmath76 & @xmath77 & @xmath78 + j1333@xmath264449 & 13800 & 3.42 & @xmath79 & @xmath80 & @xmath81 + j1339@xmath264712 & 13799 & 3.42 & @xmath71 & @xmath82 & @xmath83 + j1355@xmath266206 & 13806 & 3.42 & @xmath84 & @xmath85 & @xmath86 + j1548@xmath264821 & 13805 & 3.42 & @xmath87 & @xmath88 & @xmath89 + j1611@xmath265847 & 13802 & 3.41 & @xmath90 & @xmath91 & @xmath92 + j1753@xmath261914 & 13803 & 3.42 & @xmath93 & @xmath94 & @xmath95 + j1816@xmath265643 & 13804 & 3.42 & @xmath96 & @xmath97 & @xmath98 + b1952@xmath1929 & 12684 & 4.99 & @xmath99 & @xmath100 & @xmath101 + j2007@xmath192722&6438,7254,8492&94.04 & @xmath87 & @xmath102 & @xmath103 + j2235@xmath191506 & 13798 & 3.42 & @xmath104 & @xmath105 & @xmath106 [tab : limits]
Results
figure [fig : images] presents thumbnail images in the 0.310 kev band centered around the radio coordinates of each drp, excluding psr j2007@xmath192722 reported elsewhere @xcite. in these @xmath107 sub - images, no pixel that is not definitely associated with a significant source contains more than two counts in the @xmath108 kev energy band. examination of each image shows no evidence for a source at the radio location within twice the nominal @xmath109 pointing uncertainty. in fact, no counts are detected in an adopted aperture of radius @xmath110 at the position of any target. this is not unexpected given the mean background rate of @xmath111 counts s@xmath112 pixel@xmath112, uniform across the 12 observations. for this rate, the mean number of counts in a 3.5 ks observation is 0.073 in a @xmath110 radius circle. there is a @xmath113 probability of detecting no counts in that aperture for a single observation, and only a @xmath114 chance of getting one or more counts in any of 12 observations. in no case are the coordinates of the nearest detected x - ray source consistent with the radio location, the closest being @xmath115 from psr j1816@xmath265643. with no evidence of any photon at the location of each drp, we calculate an upper limit on the thermal flux from an assumed cooling ns of radius @xmath116 km, to match the radius used to derive the theoretical cooling curves discussed in section [sec : discussion]. as photon counts follow the poisson distribution, the probability of having gotten zero photons is 0.0023 when the expected number of photons from a source is six. therefore the @xmath117 confidence (@xmath118) upper limit on the source flux is that which would predict six counts. we determine the blackbody temperature required for a fiducial source to produce six counts plus background in the detector by convolving an absorbed blackbody spectrum through the acis spectral response and computing the total counts in the @xmath108 kev bandpass generated for each observation. the blackbody flux normalization is fixed by the ratio @xmath119 for each target, where @xmath120 is the distance derived using the ne2001 galactic free electron density model of @xcite. an absorbing column density @xmath121 is estimated from the dispersion measure (dm) assuming a rule - of - thumb @xmath122, i.e., @xmath123 dm @xcite. table [tab : limits] presents the upper limits computed in this way on the blackbody temperature and bolometric luminosity of each pulsar, quantities measured at infinity. these generally correspond to @xmath124 in the range @xmath125 k and log @xmath126=31.8 - 32.8 $]. the two outliers are psr j1355@xmath266206 and psr j1548@xmath264821, which are less constrained because of their large distances and @xmath121. the uncertainties on these upper limits are dominated by systematic errors involving the dm derived distances and column densities. dm distance can have fractional uncertainty of 25% or larger (e.g., *??? the neutral column density estimated using a typical ionized fraction involves another uncertain assumption. furthermore, an error on @xmath127 amplifies the error on the temperature measurement, which comes from the low - energy end of the acis - s instrument response, around @xmath128 kev, where the detector sensitivity falls off rapidly and is poorly calibrated. unfortunately, these effects are difficult to quantify. we repeat, for completeness, that we would not expect to detect any of the drps if they are old, rotation - powered nss with spin - down power @xmath129. for comparison, we can use the dozen old pulsars whose x - ray detections were compiled by @xcite. these typically have @xmath130 with a scatter of a factor of 10. the same x - ray efficiency for the drps would produce @xmath131 erg s@xmath112, which is orders of magnitude below our upper limits.
Discussion
the upper limits on temperature and luminosity of each drp can be compared with standard (minimal) ns cooling curves, (e.g., *??? *) to place a lower limit on its age. these limits depend strongly on uncertain variables such as the critical temperature for superfluid neutron pairing, and the composition of the ns envelope, which is why there can not be a unique age limit for each entry in table [tab : limits]. roughly speaking, a luminosity limit of log @xmath132=32.8 $] requires an age @xmath133 yr for heavy element envelopes and @xmath134 yr for light elements, while log @xmath132=31.8 $] implies that @xmath135 yr (light) or @xmath136 yr (heavy). the cooling curves for light and heavy element envelopes cross over in this range of luminosities. the upper limits on temperatures and luminosities for the drps (with the possible exception of psr j1355@xmath266206) are smaller than those of all ccos but one. in no case does a drp overlap in possible age with the snr ages of the known ccos, which are @xmath11 yr. the dozen drps fail to qualify as evolved ccos in the age range that is, roughly speaking, 10 times the ages of the known ccos, where we expect their descendants to be 10 times as numerous. the meaning of these x - ray non - detections of drps for the evolution of ccos depends on the volume sampled by the surveys that discovered both populations, and their relative completeness. both are difficult to evaluate ; however, the volumes appear to be at least comparable. the @xmath2 ccos are found in snrs up to a maximum distance of @xmath137 kpc, and the drps appear to have a similar distribution of distance and galactic coordinates. therefore, the absence of radio pulsar counterparts of orphaned ccos appears to be real, at least in the range of magnetic field strengths which define the drps. @xcite noted that roughly four of the drps so defined could actually be interlopers from the population of normal pulsars, as extrapolated from the statistics of studies such as @xcite. however, as we argued previously, it may not be possible to make such a distinction. in any case, it would not change our conclusions regarding the fate of ccos, that there are no known radio pulsars with @xmath1 g that are their immediate, @xmath138 yr old, descendants, where we would expect to find @xmath139 orphans. another clue to the age of drps should be their distribution of heights @xmath140 above the galactic plane as listed in table [tab : drps]. however, as discussed by @xcite, these heights are smaller than one would expect for the average ns kick velocity of @xmath141 km s@xmath112 @xcite, which makes it difficult to use @xmath140 as an indicator of age for drps. at this velocity a ns would travel only 270 pc in @xmath142 yr, implying that x - ray detected orphaned ccos could have a similar @xmath140 height as the drps, which are thought to be much older. since they are old, the small scale height of the drps still requires an explanation. @xcite propose that the first sn in the parent binary was of a different type that would give little or no kick to the system, perhaps an electron - capture sn. a priori, one might not have expected dprs to be orphaned ccos. as it is, there are not enough drps compared to double ns systems according to standard evolutionary models that link them @xcite. any drp that is reassigned to a different population would only exacerbate this shortage. still, the evolutionary fate of ccos remains unknown after this survey. one possible solution is that radio luminosity is a declining function of spin - down power. if so, radio surveys could be grossly incomplete in detecting such low @xmath129 pulsars even though they are on the active side of the radio pulsar death line. there is good evidence that ordinary radio pulsars behave this way, with @xmath143 @xcite, because there is no pileup in the number of pulsars near the death line. however, it is not clear that this effect alone could explain the absence of orphaned ccos, because there are in fact many radio pulsars with lower @xmath129 than the cco pulsars. such an effect may also apply to the seven discovered, radio - quiet isolated neutron stars (inss, *??? *) which, however, have strong magnetic fields @xcite, but are close to the radio pulsar death line. the inss (fig. [fig : ppdot]) are a good analogy to our problem in that they _ are _ plausibly the descendants of the magnetars, following a fast epoch of magnetic field decay around @xmath144 yr @xcite. it is likely that the inss are kept hot for longer than ccos by their continuing magnetic field decay for up to @xmath64 yr @xcite, which could account for their abundance relative to the elusive orphaned ccos. it may be difficult to detect and/or recognize orphaned ccos if they cool faster than ordinary nss. one effect that can accelerate cooling is an accreted light - element envelope, which has higher heat conductivity than an iron surface @xcite. however, this effect actually makes ccos hotter than bare nss for their first @xmath145 yr, after which their temperatures plummet. therefore, the prediction that cco descendants should be detectable in soft x - rays remains robust. another plausible home for orphaned ccos would be among the radio pulsars with magnetic fields comparable to or higher than those of the cco pulsars. one theory for ccos postulates that they are born with a canonical ns magnetic field of @xmath146 g that was largely buried by fall - back of a small amount of supernova ejecta, @xmath147, during the hours and days after the explosion. the buried field will diffuse back to the surface on a time scale that is highly dependent on the amount of mass accreted @xcite, after which the ccos will join the bulk of the population of ordinary pulsars. for accretion of @xmath148, the regrowth of the surface field is largely complete after @xmath149 yr, but if @xmath150 is accreted, then the diffusion time could be millions of years. such a scenario addresses the absence of ccos descendants ; they turn into ordinary pulsars. it also has the advantage of not requiring yet another class of ns to exist that would only exacerbate the apparent excess of pulsars with respect to the galactic core - collapse supernova rate, a problem emphasized by @xcite. furthermore, magnetic field growth has long been considered a reason why measured pulsar braking indices are all less than the dipole value of 3. in this picture, ccos represent one extreme in the evolution of surface magnetic field, and almost any radio pulsar _ might _ be a former cco. finally, an intrinsically strong crustal magnetic field appears to be necessary to explain the existence of the thermal hot spots that enable us to detect pulsations from ccos in the first place (see discussion in @xcite). for the first @xmath67 yr, rapid field growth can only move a cco vertically upward in the @xmath151 diagram. such movement is difficult to detect directly using ccos, because it would require measuring the braking index or observing the change of the dipole magnetic field spectroscopically, neither of which is likely to be possible if the relevant time scale is @xmath152 yr. however, during their first @xmath145 yr, orphaned ccos in this scenario should still have periods of @xmath153 s and could have magnetic fields in the range @xmath154 g. a search of all 159 isolated radio pulsars in this range for thermal x - ray emission from such `` old '' pulsars would provide a promising avenue for finding orphaned ccos. of these, two are known x - ray sources, the faint (@xmath155 erg s@xmath112) nearby radio pulsars psr b1451@xmath2668 and psr b0950 + 08. x - rays from these sources are attributed to a combination of heated polar caps and non - thermal (magnetospheric) emission @xcite. if further x - ray surveys of radio pulsars fail to find any orphaned ccos, then it will be difficult to escape the conclusion that they are intrinsically radio quiet.
Conclusions
following the discovery that ccos have weaker magnetic fields than any other young pulsar, it became apparent that their descendants were not obviously present in radio or x - ray surveys. if their magnetic fields at birth are intrinsic, and do not change with time, then the region around the ccos in the @xmath156 diagram of radio pulsars should be densely populated with all of their descendants, unless they are radio quiet. the fact that this area is quite sparsely populated led us to survey a large fraction of the available radio candidates in x - rays, those which were previously understood to be mildly recycled pulsars. the `` smoking pulsar '' evidence of an orphaned cco should be an x - ray hot ns that could be detected, in a short observation, at an age up to @xmath145 yr, which is much younger than the characteristic ages of the targeted drps but much older than the known ccos. only upper limits on their thermal x - ray luminosities were found, in the range log @xmath3=31.8 - 32.8 $], which implies cooling ages @xmath4 yr. up to the age limits implied by the x - ray non - detections, there should be @xmath157 cco descendants in the volume sampled by radio pulsar surveys. since none have been found among radio pulsars with @xmath1 g, the next step should be to search for young, cooling nss among the radio pulsars with larger @xmath7-fields, comparable to or even larger than that of the cco, with @xmath158 g. an especially interesting possibility is that ccos have intrinsically strong @xmath7-fields that were promptly buried by a small amount of supernova debris, but will grow back to `` normal '' strength in @xmath144 yr. if such descendants of ccos are found in thermal x - rays among the ordinary radio pulsar population, it would help solve problems about their surface thermal patterns in addition to their evolution. otherwise, if the orphaned ccos are truly radio silent for some unknown reason, they could still be found in more sensitive all - sky surveys in soft x - rays, by analog with the (evidently more luminous) inss that were discovered this way. we thank dr. reinhard prix for discussion during the proposal phase. this investigation is based on observations obtained with the observatory. financial support was provided by award go2 - 13097x issued by the x - ray observatory center, which is operated by the smithsonian astrophysical observatory for and on behalf of nasa under contract nas8 - 03060. | we present a x - ray survey of the disrupted recycled pulsars (drps), isolated radio pulsars with @xmath0 ms and @xmath1 g. these observations were motivated as a search for the immediate descendants of the @xmath2 central compact objects (ccos) in supernova remnants, three of which have similar timing and magnetic properties as the drps, but are bright, thermal x - ray sources consistent with minimal neutron star cooling curves. since none of the dprs were detected, there is no evidence that they are `` orphaned '' ccos, neutron stars whose supernova remnants has dissipated.
upper limits on their thermal x - ray luminosities are in the range log @xmath3=31.8 - 32.8 $], which implies cooling ages @xmath4 yr, roughly 10 times the ages of the @xmath5 known ccos in a similar volume of the galaxy.
the order of a hundred cco descendants that could be detected by this method are thus either intrinsically radio quiet, or occupy a different region of (@xmath6) parameter space from the drps.
this motivates a new x - ray search for orphaned ccos among radio pulsars with larger @xmath7-fields, which could verify the theory that their fields are buried by fall - back of supernova ejecta, but quickly regrow to join the normal pulsar population. | 1307.2146 |
Introduction
recent progress in our understanding of faint galaxy data made possible by the combination of _ hubble space telescope _ (_ hst _) deep imaging and ground - based spectroscopy has dramatically increased our knowledge of the evolution of the stellar birthrate in optically - selected galaxies from the present - epoch up to @xmath9 @xcite, @xcite, @xcite. the explosion in the quantity of information available on the high - redshift universe at optical wavelengths has been complemented by the measurement of the far - ir / sub - mm background by dirbe and firas onboard the _ cobe _ satellite @xcite, @xcite, @xcite, by the detection of distant ultraluminous sub - mm sources with the scuba camera @xcite, @xcite, and by theoretical progress made in understanding how intergalactic gas follows the dynamics dictated by dark matter halos until radiative, hydrodynamic, and star formation processes take over @xcite, @xcite, @xcite. the ir data have revealed the ` optically - hidden'side of galaxy formation, and shown that a significant fraction of the energy released by stellar nucleosynthesis is re - emitted as thermal radiation by dust @xcite, @xcite. the underlying goal of all these efforts is to understand the growth of cosmic structures, the internal properties of galaxies and their evolution, and ultimately to map the star formation history of the universe from the end of the cosmic ` dark age'to the present epoch. in this talk i will focus on the galaxy number - apparent magnitude relation and its first moment, the integrated galaxy contribution to the extragalactic background light (ebl). the logarithmic slope of the differential galaxy counts is a remarkably simple cosmological probe of the history of stellar birth in galaxies, as it must drop below 0.4 to yield a finite value for the ebl. the recently released _ hubble deep field - south _ (hdf - s) images, together with other existing _ hst _ and ground - based observations, provide a unique dataset to estimate the spectrum and amplitude of the optical ebl from discrete sources. together with the far - ir / sub - mm background, the optical ebl is an indicator of the total luminosity of the universe, as the cumulative emission from young and evolved galactic systems, as well as from active galactic nuclei (agns), is recorded in this background. as such it provides, for a given initial mass function, a quantitative estimate of the baryonic mass that has been processed by stars throughout cosmic history. unless otherwise stated, an einstein - de sitter (eds) cosmology (@xmath10, @xmath11) with @xmath12 will be adopted in this talk. all magnitudes will be given in the ab system. the work presented here has been done in collaboration with l. pozzetti.
Galaxy counts
the hdf - s dataset includes deep near - ir nicmos images and the deepest observation ever made with the stis 50ccd filterless imaging mode. the galaxy sample used here was extracted from version 1 of the hdf - s catalog on ftp://archive. stsci.edu / pub / hdf_south / version1/. at near - ir wavelengths (in the f110w, f160w, and f222 m bandpasses, corresponding to the @xmath13, @xmath14, and @xmath15 filters), it consists of 425 objects detected in the @xmath16 image, over a field of @xmath17. the 50ccd (corresponding roughly to a @xmath18 filter) stis catalog consists of 674 objects detected again over a field of the same size. galaxy counts as a function of ab magnitudes. the sources of the data points are given in the text. note the decrease of the logarithmic slope @xmath19 at faint magnitudes. _ right : _ extragalactic background light per magnitude bin, @xmath20, as a function of @xmath21 (_ filled circles _), @xmath22 (_ open circles _), @xmath23 (_ filled pentagons _), @xmath24 (_ open squares _), @xmath13 (_ filled triangles _), @xmath14 (_ open triangles _), and @xmath15 (_ filled squares _) magnitudes. for clarity, the @xmath25 measurements have been multiplied by a factor of 2, 6, 15, 50, 150, and 600, respectively @xcite., scaledwidth=90.0%] figure 1 shows the hdf - n and -s galaxy counts compiled directly from the catalogs, as a function of ab isophotal magnitudes in the @xmath1 bandpasses for all galaxies with signal - to - noise ratio @xmath26 within the band. no correction for detection completeness have been made. a compilation of existing _ hst _ and ground - based data is also shown @xcite, @xcite. all magnitudes have been corrected to the ab system, while the second order colour corrections for the differences in the filter effective wavelengths have not been applied to the ground - based data (for the typical colours of galaxies in the hdf these corrections are less than 0.1 mag). the hdf optical counts agree well with previous surveys, to within @xmath27 in the magnitude range @xmath28. one should note, however, that different algorithms used for ` growing'the photometry beyond the outer isophotes of galaxies can significantly change the magnitude of faint galaxies. according to @xcite, roughly 50% of the flux from resolved galaxies with @xmath29 mag lie outside the standard - sized apertures used by photometric packages. an extragalactic sky pedestal created by the overlapping wings of resolved galaxies may also contribute significantly to the sky level, and would be undetectable except by absolute surface photometry @xcite. also, at faint magnitude levels, distant objects which are brighter than the nominal depth of the catalog may be missed due to the @xmath30 dimming factor. all these systematic errors are inherent in _ hst _ faint - galaxy photometry ; as a result, our estimate of the integrated flux from resolved galaxies will typically be too low, and must be strictly considered as a _ lower limit_.
The extragalactic background light
the contribution of known galaxies to the optical ebl can be calculated directly by integrating the emitted flux times the differential number counts down to the detection threshold. i have used the compilation of ground - based, _ hst _, and hdf data shown in figure 1 to compute the integrated flux at @xmath31 m. in all seven bands, the slope of the differential number - magnitude relation is flatter than 0.4 above @xmath32 (25) at near - ir (optical) wavelengths, and this flattening appears to be more pronounced at the shorter wavelengths. the leveling off of the counts is clearly seen in figure 1, where the function @xmath20 is plotted against apparent magnitude in all bands. while counts having a logarithmic slope @xmath33 continue to add to the ebl at the faintest magnitudes, it appears that the hdf survey has achieved the sensitivity to capture the bulk of the near - ultraviolet, optical, and near - ir extragalactic light from discrete sources. the flattening at faint apparent magnitudes can not be due to the reddening of distant sources as their lyman break gets redshifted into the blue passband, since the fraction of lyman - break galaxies at (say) @xmath34 is only of order 10% @xcite. moreover, an absorption - induced loss of sources can not explain the similar change of slope in the @xmath35 and @xmath15 bands. while this suggests that the surface density of optically luminous galaxies is leveling off beyond @xmath36, one should worry about the possibility of a significant amount of light being missed at faint magnitudes. bands (_ filled dots _), together with the firas 1255000 @xmath37 m (_ solid and dashed lines _) and dirbe 140 and 240 @xmath37 m (_ filled squares _) detections @xcite, @xcite. the _ empty squares _ show the dirbe points after correction for wim dust emission @xcite. also plotted (_ filled triangle _) is a foca - uv point at 2000 @xcite, and a tentative detection at 3.5 @xmath37 m (_ empty dot _) from _ cobe_/dirbe observations @xcite. the empty pentagons at 3000, 5500, and 8000 are recent detections from absolute photometry @xcite. upper limits are from @xcite, the lower limit from @xcite. note that the values obtained by integrating the brightness of resolved galaxies are strict lower limits to the ebl intensity.] the spectrum of the optical ebl is shown in figure 2, together with the recent results from _ cobe_. the value derived by integrating the galaxy counts down to very faint magnitude levels (because of the flattening of the number - magnitude relation most of the contribution to the optical ebl comes from relatively bright galaxies) implies a lower limit to the ebl intensity in the 0.22.2 @xmath37 m interval of @xmath38. including the tentative detection at 3.5 @xmath37 m by @xcite would boost @xmath39 to @xmath40. recent estimates of the optical ebl at 3000, 5500, and 8000 from absolute surface photometry by @xcite lie between a factor of two to three higher than the integrated light from galaxy counts. applying this correction factor to the range 30008000 gives an optical ebl intensity in excess of @xmath41 in the interval 0.23.5 @xmath37 cobe_/firas @xcite measurements yield @xmath42 in the 1252000 @xmath37 m range. when combined with the dirbe @xcite, @xcite, @xcite points at 140 and 240 @xmath37 m, one gets a far - ir background intensity of @xmath43. the residual emission in the 3.5 to 140 @xmath37 m region is poorly known, but it is likely to exceed @xmath44 @xcite. a ` best - guess'estimate to the total ebl intensity observed today is then @xmath45 in the rest of my talk, i will adopt a reference value for the background light associated with star formation activity over the entire history of the universe of @xmath46.
Ebl from quasar activity
a direct estimate of the contribution of quasars to the ebl depends on the poorly known bolometric correction and the possible existence of a distant population of dusty agns with strong intrinsic absorption, as invoked in many models for the x - ray background. these type ii qsos, while undetected at optical wavelengths, could contribute significantly to the far - ir background. it is in principle possible to bypass some of the above uncertainties by weighing the local mass density of black holes remnants @xcite. recent dynamical evidence indicates that supermassive black holes reside at the center of most nearby galaxies. the available data (see fig. 3) show a correlation (but with a large scatter) between bulge and black hole mass, with @xmath47 as a best - fit @xcite. the mass density in old spheroidal populations today is estimated to be @xmath48 @xcite, implying a mean mass density of quasar remnants today @xmath49 since the observed energy density from all quasars is equal to the emitted energy divided by the average quasar redshift, the total contribution to the ebl from accretion onto black holes is @xmath50 (@xmath51), where @xmath52 is the efficiency of accreted mass to radiation conversion (in units of 5%). therefore, unless dust - obscured accretion onto supermassive black holes is a very efficient process (@xmath53), a population of quasars peaking at @xmath54 is expected to make a contribution to the brightness of the night sky not exceeding 1020% @xcite, @xcite. . the symbols denote different galaxy types : _ empty circles _ (e), _ filled squares _ (s0), _ filled circles _ (sab), and _ empty squares _ (sbc - scd).]
The stellar mass density today
with the help of some simple stellar population synthesis tools we can now set a lower limit to the total stellar mass density that produced the observed sky brightness and constrain the cosmic history of star birth in galaxies. one of the most serious uncertainties in this calculation has always been the lower cutoff (usually treated as a free parameter) of the initial mass function (imf). observations of m subdwarfs stars with the _ hst _ have recently shed some light on this issue, showing that the imf in the galactic disk can be represented analytically over the mass range @xmath55 (here @xmath56 is in solar units) by @xmath57 (@xcite, hereafter gbf). for @xmath58 this mass distribution agrees well with a salpeter function. observations of normal galactic star - forming regions also show some convergence in the basic form of the imf at intermediate and high masses, a power - law slope that is consistent with the salpeter value @xcite. in the following i will use a ` universal'imf with the gbf form for @xmath59, matched to a salpeter slope for @xmath60 ; the mass integral of this function is 0.6 times that obtained extrapolating a salpeter function down to @xmath61. is quite small for a ` typical'imf. the use of the gbf mass function at low masses instead of salpeter leaves then the total radiated luminosity of the stellar population virtually unaffected.] , metallicity @xmath62 (_ solid line _) and @xmath63 (_ dotted line _), and a gbf@xmath5salpeter imf (see text for details). _ right : _ ebl observed at earth from the instantaneous formation at redshift @xmath64 of a stellar population having the same imf, solar metallicity, and mass density @xmath65 and 0.00075, as a function of @xmath64. _ solid curves : _ eds universe with @xmath51. _ dashed curves : _ @xmath66-dominated universe with @xmath67, @xmath68, and @xmath69., scaledwidth=90.0%] as shown in figure 4, the _ bolometric _ luminosity as a function of age @xmath70 of a simple stellar population (a single generation of coeval, chemically homogeneous stars having total mass @xmath71, solar metallicity, and the above imf) can be well approximated by @xmath72 (cf @xcite). over a timescale of 13 gyr (the age of the universe for an eds cosmology with @xmath51), about 1.3 mev per stellar baryon are radiated away. this number depends only weakly on the assumed metallicity of stars. in a stellar system with arbitrary star formation rate per comoving cosmological volume, @xmath73, and formation epoch @xmath74, the comoving bolometric emissivity at time @xmath75 is given by the convolution integral @xmath76 the total background light observed at earth (@xmath77) is @xmath78 where the factor @xmath79 at the denominator is lost to cosmic expansion when converting from observed to radiated (comoving) luminosity density. to set a lower limit to the present - day mass density, @xmath7, of processed gas @xmath5 stars (in units of the critical density @xmath80), consider now a scenario where all stars are formed instantaneously at redshift @xmath64. the background light that would be observed at earth from such an event is shown in figure 4 as a function of @xmath64 for @xmath81 (corresponding to 9, 6.5, and 4 percent of the nucleosynthetic baryon density, @xmath82 @xcite), and two different cosmologies. a couple of points are worth noting here : (1) the time evolution of the luminosity radiated by a simple stellar population makes the dependence of the observed ebl from @xmath64 much shallower than the @xmath83 lost to cosmic expansion, as the energy output from stars is spread over their respective lifetimes ; and (2) in order to generate an ebl at a level of @xmath6, one requires @xmath84 (for an eds universe with @xmath51), hence a mean mass - to - blue light ratio today of @xmath85 for a present - day blue luminosity density of @xmath86 @xcite. as shown in figure 4, the dependence of these estimates on the cosmological model is rather weak. with the adopted imf, about 30% of this mass will be returned to the interstellar medium in @xmath87 yr after intermediate - mass stars eject their envelopes and massive stars explode as supernovae. this ` return fraction ', @xmath88, becomes 50% after about 10 gyr. a visible mass density at the level of the above lower limit, @xmath89, while able to explain the measured sky brightness, requires all the stars that give origin to the observed light to have formed at @xmath90, and is, as such, rather implausible. a more realistic scenario appears to be one where the star formation density evolves as @xmath91 this model fits reasonably well all measurements of the uv - continuum and h@xmath92 luminosity densities from the present - epoch to @xmath93 after an extinction correction of @xmath94 mag (@xmath95 mag) is applied to the data @xcite, and produce a total ebl of the right magnitude (@xmath96). since about half of the present - day stars are formed at @xmath97 (hence their contribution to the ebl is redshifted away), the resulting visible mass density is @xmath98 (@xmath99). note that this estimate ignores the recycling of returned gas into new stars. the observed ebl therefore requires that between 7% and 16% of the nucleosynthetic baryons are today in the forms of stars, processed gas, and their remnants. according to the most recent census of cosmic baryons, the mass density in stars and their remnants observed today is @xmath100 @xcite, corresponding to a mean visible mass - to - blue light ratio of @xmath101 (@xmath51) (about 70% of this mass is found in old spheroidal populations). while this is about a factor of 2.5 smaller than the visible mass density predicted by equation ([eq : mplot]), efficient recycling of ejected material into new star formation would tend to reduce the apparent discrepancy in the budget. alternatively, the gas returned by stars may be ejected into the intergalactic medium. with an imf - averaged yield of returned metals of @xmath102,^{-1}$], the stellar yields @xmath103 of @xcite, and a gbf@xmath5salpeter imf.] the predicted mean metallicity at the present epoch is @xmath104, in good agreement with the values inferred from cluster abundances @xcite. | i review the constraints imposed by the observed extragalactic background light (ebl) on the history of the stellar birthrate in galaxies. at faint magnitudes,
the logarithmic slope of the galaxy counts is flatter than @xmath0 in all seven @xmath1 optical bandpasses of the _ hubble deep field - south _ imaging survey.
the integration of the number counts provides a lower limit to the surface brightness of the optical extragalactic sky of @xmath2, comparable to the intensity of the far - ir background from _ cobe _ data.
if the initial mass function has a salpeter slope with a lower mass cutoff consistent with observations of m subdwarf disk stars, a lower limit of @xmath3 (at hubble constant @xmath4) is derived for the visible (processed gas @xmath5 stars) mass density needed to generate an extragalactic background light (ebl) at a level of @xmath6.
the current ` best - guess'estimate to @xmath7 is @xmath8, about 16% of the nucleosynthetic baryon density.
the contribution of quasar activity to the observed ebl is unlikely to exceed 20%. | astro-ph9907268 |
Introduction
the creation and motion of large numbers of crystal lattice dislocations is the most fundamental feature of crystal plasticity. during the last half century, the physical properties of individual dislocations and their interactions with localised obstacles have been studied extensively. on the other hand, the complex collective dynamics of strongly interacting many - dislocation systems is still far from being understood. fortunately, everyday plastic deformation processes very often proceed orders of magnitude slower than the typical relaxation times of the underlying dislocation system. these conditions often permit to study the problem in a quasistatic approximation @xcite. beyond the quasistatic limit, however, much less work has been devoted to studying the dynamics of collective dislocation motions which lead to the formation of metastable configurations, and to transitions between such configurations in driven dislocation systems. however, such collective motions are crucial for understanding rapid dislocation processes which not only occur in shock loading but, in the form of dislocation avalanches, are a generic feature of the dynamics of driven dislocation systems @xcite. the first studies of dynamic relaxation processes in dislocation systems were performed by miguel et al. with the protocol of applying a constant external shear stress to well relaxed dislocation configurations @xcite. the ensuing creep relaxation was numerically shown to follow andrade s law stemming from the underlying intermittent and correlated motion of dislocation structures. the connection between the mesoscopic and macroscopic features of the process was, however, not analysed in detail. another direction was taken by the present authors who conducted systematic studies of the relaxation dynamics of initially random configurations of straight dislocations. this is an important issue since the elastic energy density @xmath0 of a random dislocation system of density @xmath1 is known to diverge with the logarithm of system size @xmath2, @xmath3 @xcite where @xmath4 is the modulus of the dislocation burgers vector. in a well - relaxed dislocation arrangement, on the other hand, the same quantity scales like @xmath5, i.e., the screening length corresponds to the mean dislocation spacing @xcite. as the mean square stress is proportional to the elastic energy density, this screening also removes a logarithmic divergence of the width of the internal stress probability distribution @xcite, and of the x - ray line width @xcite. numerical experience showed that, at least in single slip geometries, the relaxation processes that lead to screened dislocation arrangements exhibit slow, power law characteristics for quantities such as the elastic energy or the average dislocation velocity @xcite. a model was proposed which relates the power - law relaxation dynamics to the gradual extinction of initial dislocation density fluctuations @xcite. the present paper presents a comprehensive numerical investigation which allows to check in detail the model predictions and complements the earlier work by extending the investigation to multiple slip geometries and to dislocation systems of non - zero net burgers vector, and by studying the influence of an external driving stress on the relaxation process. the paper is organised as follows. in the problem is defined and technical details of the simulations are presented. unfolds a scaling model of the relaxation process from a chemical analogy and uses this model to predict the evolution of simulation measurables. then gives a detailed comparison between model predictions and numerical results. the results are discussed and conclusions are drawn in. an auxiliary calculation of the elastic energy of a random dislocation wall is presented in the appendix.
Problem definition and simulations
consider a system of @xmath6 straight edge dislocations running parallel to the @xmath7 axis of a cartesian coordinate system. let all dislocations have a common burgers vector pointing along the @xmath8 axis (a so - called single slip geometry), @xmath9, where @xmath10 is the sign of the @xmath11th dislocation. assuming overdamped glide motion with a dislocation velocity @xmath12 that is proportional to the local resolved shear stress, and zero dislocation mobility in the climb direction, the equation of motion of dislocation @xmath11 piercing the @xmath13 plane at @xmath14 can be written as @xmath15, \qquad \tau_{\mathrm{ind}}(\bi{r }) = g b \frac{x (x^{2}-y^{2})}{(x^{2}+y^{2})^{2}},\]] where @xmath16 denotes the dislocation glide mobility, @xmath17 $] where @xmath18 is the shear modulus and @xmath19 is poisson s ratio of the embedding isotropic crystal, @xmath20 denotes the resolved shear stress field induced by a positive dislocation located at the origin @xcite, and @xmath21 is a constant externally applied resolved shear stress. it is useful to introduce natural coordinates at this point which will be denoted by an apostrophe (@xmath22) in the following. measuring length in units of the average dislocation dislocation distance @xmath23 (where @xmath1 denotes the total dislocation density of dislocations including both signs and, in multiple slip geometries, including all slip systems), stress @xmath24 in units of @xmath25, and plastic strain @xmath26 in units of @xmath27 leads to the relations @xmath28 where @xmath29 is the elastic energy difference between two states of the system (energy per unit dislocation length). in natural coordinates takes the form @xmath30, \cr & \tau'_{\mathrm{ind}}(\bi{r }') = \frac{x'(x'^{2}-y'^{2})}{(x'^{2}+y'^{2})^{2 } } = \frac{\cos(\varphi) \cos(2\varphi)}{r'}, } \]] where @xmath31 denotes the angle between the @xmath8 axis and @xmath32. to study dislocation relaxation, a large number of discrete dislocation dynamics simulations have been performed. equations of motion were solved with the 4.5th order runge kutta fehlberg method. periodic boundary conditions were applied to a square simulation area with edges parallel to the slip planes, following the method used in @xcite. to avoid overly small timesteps during the final stages of approach of narrow dipoles (pairs of dislocations of opposite signs), a small number of extremely narrow dipoles were excluded from the solution of and forced to move as if they were isolated from the rest of the system. this is justified as the far - field stresses of the dislocations in a narrow dipole cancel, while their pair interaction diverges when the dislocation dislocation distance approaches zero. as a consequence, the dynamics of narrow dipoles is effectively uncoupled from the rest of the dislocation system. we do not allow for annihilation of narrow dipoles, which is a process that is governed by the atomistics of the dislocation cores. this implies that we consider dislocation spacings to be large in comparison with the dipole annihilation distance, which is believed to be of the order of one nanometer. as typical dislocation densities in highly dislocated crystals are of the order of @xmath33 m@xmath34, i.e. the average dislocation spacings are of the order of a hundred nanometers, this is not a severe restriction. a first set of simulations was started from random configurations of equal numbers of positive and negative dislocations. the number of simulated dislocations @xmath6 (which defines the system size @xmath35 since @xmath36) varied between @xmath37 and @xmath38. it is well known that the flow stress of single - glide dislocation systems is around @xmath39 in natural units @xcite. to allow the dislocation systems to reach mechanical equilibrium at the end of the relaxation, we restricted the applied external stresses to levels below the flow stress, using stresses between @xmath40 and @xmath41 natural units. as seen in, individual simulations showed strong avalanche - like activity during relaxation, as previously observed in @xcite. to reveal scaling properties of the relaxation process, the evolution of global parameters such as stored energy, mean absolute dislocation velocity, mean square velocity, and mean strain rate was averaged over @xmath42 to @xmath43 simulations starting from different random initial configurations. this ensemble averaging resulted in smooth ensemble averaged graphs as seen in. in addition to the relaxation of ` neutral'arrangements of dislocations moving on a single slip system aligned parallel to the edge of a square simulation area, we considered two variants of this basic setting (these simulations were only performed at zero external stress) : i) to elucidate the influence of different implementations of the boundary conditions, we performed single slip simulations with the simulation box oriented at an angle @xmath31 to the slip planes. this does not affect the short - range dislocation dislocation interactions but modifies the stress field created by the periodic images. ii) to study the influence of a net burgers vector on the relaxation process, we investigated the limiting case of fully polarised dislocation systems (dislocations of one sign only) of sizes @xmath44. to investigate the differences between single and multiple slip geometries, we also performed simulations in which dislocations of multiple slip systems were present. in these simulations we considered sets of equally populated slip systems, each containing edge dislocations with a zero net burgers vector that were initially distributed at random. the methodology of the multiple slip simulations was identical to the single slip simulations described above, with the following differences : i) for each dislocation in the system, the complete elastic stress tensor was computed, again assuming periodic boundary conditions in a square simulation area and using the method of @xcite. the forces acting on the dislocations were calculated from the stress tensor components using the peach koehler formula @xcite. ii) in addition to narrow dipoles, which were treated in the same way as described above, pairs of attracting dislocations on intersecting slip planes needed to be treated separately. such dislocation pairs react with each other and form a reaction product (` dislocation lock') with burgers vector equal to the net burgers vector of the constituent dislocations. in real crystals, both the mobility of dislocation locks and the stress required for separating them into the constituent dislocations depend on their atomic core structure. for simplicity, we assumed all dislocation locks to be immobile and to possess infinite separation stress. new dislocations joining an existing lock were assumed to annihilate with the constituent dislocation of the opposite burgers vector if such a dislocation was present. if a dislocation lock converted to an ordinary dislocation through such an annihilation event, it was assumed to become mobile again. to avoid overly small timesteps during the final stages of dislocation lock formation or reaction with a new dislocation (again due to diverging elastic interactions between the constituent dislocations) the moving constituent dislocations were pinned when their relative distance decreased below a small predefined reaction radius which mimics the core extension of the dislocation lock. for simplicity, the stress field of a lock was calculated as the superposition of the stress fields of the pinned constituent dislocations. although the positions of the constituent dislocations are scattered over a small region of the size of the reaction radius, their net stress field is a good approximation of the stress field of the lock at larger distances. finally, we note that this stress field is of long range character since the net burgers vector of a dislocation lock is not zero.
Theory
in this section a simple scaling argument is used to establish the asymptotic kinetics of a bimolecular combination reaction. we then adapt the fundamental ideas behind this argument to the relaxation of dislocation systems as described in the previous section. based on the adapted model, predictions are made for the evolution of several physical quantities which can be directly obtained from the simulations described in. these predictions are compared in detail to the numerical results in. consider the direct combination reaction @xmath45 starting from a random, balanced configuration of the two reactants @xmath46 where @xmath47 denotes the spatial average of the concentration field @xmath48. note that the initial concentrations of molecules @xmath49 and @xmath50 are equal only in an average sense because of the thermally induced random positions of the individual molecules. these concentration fluctuations need to be taken into account when modelling the reaction kinetics. for situations where long - distance transport of particles occurs by free brownian motion, a simple scaling argument which captures the essential physics was given by ovchinnikov @xcite (which the reader is advised to consult for further details). broadly speaking, this involves the subsequent dominance of two consecutive mechanisms directly corresponding to the different length scales inherent in the system : i) the reaction of those molecules that do not need long distance motion to find a reaction partner, followed by ii) the long range brownian motion and subsequent reaction of excess reactants remaining as a result of initial concentration fluctuations. as discussed in @xcite, the first stage of reaction is controlled by a bimolecular reaction rate @xmath51 which depends on the short - range interactions between the reaction partners. this stage can be discussed in the classical approximation @xmath52 solving for initial conditions yields @xmath53 because of thermal concentration fluctuations, not all of the molecules @xmath49 and @xmath50 can be consumed during stage 1. spatial fluctuations of the reactant concentrations lead to local excess of molecules of one type. as these excess molecules can not find a local partner, they need to migrate via long range brownian motion, leading to a second kinetic stage controlled by long - range diffusion. to characterise the concentration fluctuations in question we observe that, on scales larger than the range of the ` contact interactions'which govern the first stage of the reaction kinetics, the positions of molecules are statistically independent. as a consequence, the initial numbers of molecules @xmath49 or @xmath50 in a sufficiently large volume @xmath54 are poisson distributed. this implies that the mean numbers of excess molecules fulfil the relations @xcite @xmath55 where @xmath56 denotes spatial averaging over a large statistically homogeneous system or, equivalently, ensemble averaging over a large number of statistically independent and equivalent realizations. to further discuss the reaction kinetics during stage 2, we confine ourselves to the limiting case of an infinite bimolecular reaction rate @xmath57. this choice does not affect the generality of the discussion, it only affects the moment of the crossover between stage 1 and stage 2 kinetics. we introduce the concentration difference @xmath58 as @xmath59 in a hypothetical initial state before the reaction has been ` switched on'at @xmath60, @xmath61 and @xmath62 are statistically independent and @xmath58 has the initial statistical properties @xcite @xmath63 to proceed, we note that for @xmath57 molecules @xmath49 and @xmath50 can not coexist for @xmath64. therefore, @xmath58 gives a complete characterisation of the concentration map for @xmath65 : for @xmath66, @xmath67 and @xmath68 and for @xmath69, @xmath70 and @xmath71. because there exists no other physical length scale in the system, the size of the regions characterised by @xmath72 and @xmath73 is determined by the diffusion length @xmath74 (for simplicity, the same diffusion constant @xmath75 is assumed for both kinds of reactants). consider now the volume referring to the diffusion length at time @xmath76, @xmath77. one can suppose that for lengths larger than @xmath78 the fluctuations of @xmath58 are still not affected by diffusion ; therefore, one can write that @xmath79 in some regions, @xmath80 and in others, @xmath81 ; hence, on average @xmath82 meaning a slower kinetics than in stage 1. (for a more detailed derivation see @xcite). note that is only applicable within certain time limits @xmath83 $]. for instance, for @xmath57, @xmath84 is equal to the time the diffusion length @xmath78 needs to exceed the average intermolecular distance. @xmath85 is determined by the time needed by @xmath78 to reach the system size @xmath86, independent of the value of @xmath51. now we consider the relaxation of initially random dislocation configurations with the same number of positive and negative dislocations (@xmath87) following the equations of motion. although not immediately evident, this relaxation process has strong phenomenological similarities to the kinetics of the chemical reaction described in. to elucidate the analogy, the relaxation process will be envisaged as a gradual screening of the long range elastic stress fields of individual dislocations through the formation of dislocation dislocation correlations @xcite. we first envisage ` neutral'arrangements where dislocations of both signs are present in equal amounts. in a screened dislocation arrangement, the excess of one sign over the other has been eliminated on scales above a few dislocation spacings. any dislocation arrangement where excess dislocations are completely eliminated can be envisaged as an assembly of dipoles where each dislocation has exactly one partner of opposite sign within a distance of the order of @xmath23, and this picture will be used in the following argument. hence, we envisage the relaxation process as the gradual formation of a large number of dislocation dipoles consisting of dislocations of opposite signs, i.e., as a bimolecular reaction process analogous to the chemical reaction. the principal difference between the two processes lies in the dynamics of individual particles : for the dislocation system, dislocation glide motion is driven by dislocation dislocation interactions which scale like @xmath88, whereas in case of the chemical reaction we are dealing with diffusive brownian motion of the reactants. despite this difference, we may again construct a two - stage model for the dislocation relaxation process : i) in stage 1 adjacent opposite sign dislocations form dipoles ; ii) in stage 2 initial fluctuations in the excess dislocation density gradually die out from shorter towards longer length scales as excess dislocations which did not find a dipole partner in stage 1 undergo long range glide motion. the process terminates once the length scale on which fluctuations have been eliminated reaches the system size. to understand the time evolution of the system, we again consider the evolution of the typical length scale @xmath78 below which density fluctuations have already died out. to this end we consider that i) similar to the chemical case, areas of size @xmath89 typically contain mobile excess dislocations (which have still not ` reacted'into dipoles) with only one or the other sign and that ii) dislocation dipoles give rise only to short range stress fields with a @xmath90 decay. as the typical distance between opposite sign dislocations which try to find each other is proportional to @xmath91, the typical driving stress towards dipole formation scales as @xmath92. (recall that variables with an apostrophe (@xmath22) are measured in natural units.) the dislocation velocity is proportional to @xmath93, and therefore the characteristic time for eliminating excess dislocations on scale @xmath94 scales as @xmath95. hence we find that @xmath96 incidentally, this result is very similar to the evolution law of @xmath78 for brownian motion, @xmath97. by supposing that the mobile dislocations inherit the initial concentration fluctuations we find that at time @xmath98, regions of size @xmath99 contain about @xmath94 excess dislocations of one or the other sign, and @xmath100 dislocations in total (see the chemical model in). thus, the fraction of non - paired dislocations is estimated to decrease in time as @xmath101 following the chemical model in, it is straightforward to predict the time interval @xmath102 $] during which the above argument is expected to hold. the start time @xmath103 is characterised by @xmath2 reaching the dislocation dislocation distance @xmath23 (@xmath104 in natural units) and the process is finished when @xmath2 reaches the system size @xmath86. with this leads to @xmath105 in natural coordinates where @xmath6 denotes the total number of dislocations in the system. the fraction of ` non - paired'dislocations is not a convenient quantity for comparing the scaling model with dislocation dynamics simulations, as the definition of ` dislocation pairs'in a multipolar dislocation arrangement may be ambiguous. instead, we consider the evolution of the @xmath106th moment @xmath107 of the dislocation velocity and of the excess elastic energy, both of which can be determined from the simulations in a straightforward manner. to obtain scaling estimates for the relaxation of these quantities, we assume that all dislocation dipoles are at rest and only the excess dislocations move. furthermore, we assume that the motion of excess dislocations is not hindered by the dipoles already formed (this can be rationalised with the short range of the dipole stress field). the velocity of excess dislocations scales as @xmath108, leading to @xmath109 where and have been used. the dynamics of dislocations is assumed to be overdamped. hence, the work that is done by the internal stresses in driving the system is completely dissipated : the amount of dissipated energy exactly matches the reduction in elastic energy. the energy dissipated per unit time by a moving dislocation @xmath11 scales like @xmath110 (the peach koehler force acting on the dislocation is proportional to the stress), and consequently the time evolution of @xmath111 can be expressed as @xmath112 since the motion is overdamped, the dislocation velocity is proportional to the acting stress. in natural units and for @xmath113 $], the ensemble averaged elastic energy thus evolves like @xmath114 where was used to estimate the time evolution of the second velocity moment. the scaling argument in the previous section is based on the formation of dipoles consisting of edge dislocations of opposite signs. at first glance, such an argument seems to be completely inapplicable to systems where only dislocations of the same sign are present. dipole formation is clearly impossible in such systems. instead, a most conspicuous feature in the relaxation of single - sign edge dislocation systems is the formation of walls containing many dislocations of the same sign that are aligned in the direction perpendicular to the slip plane @xcite. accordingly, theoretical arguments have focused on parameters characterising the ` condensation'of dislocations into walls @xcite. however, even though dislocation wall formation is a most conspicuous feature, wall formation alone can _ not _ produce a screened dislocation arrangement. the authors of @xcite evaluate the driving force for wall formation by assuming periodic spacing of dislocations along a wall and using classical results found e.g. in @xcite. if we follow this line of reasoning and note that the energy of a dislocation in a wall decreases with decreasing dislocation spacing, the minimum - energy structure for the system at hand would be a single system - spanning wall. however, for an initially random dislocation system the @xmath115 positions are independent random variables and it is not easy to see how a periodic arrangement could form in the absence of dislocation climb. we calculate the energy of a random wall in the appendix and show that forming such a wall does not produce any energy reduction with respect to the initial random 2d arrangement. how then can an arrangement of dislocations of the same sign be screened? the answer was provided by wilkens @xcite who demonstrated that a screened dislocation arrangement can be constructed by eliminating dislocation density fluctuations above a certain scale. to this end, he proposed a construction where the crystal cross section is tiled into a grid of cells of size @xmath116, and the same number of dislocations is randomly distributed within each cell. this construction, which eliminates all density fluctuations on scales above the cell size, leads to an arrangement where the screening radius coincides with the cell size, @xmath117. taking the wilkens construction as a well - screened reference state offers a surprising outlook on the relaxation of initially random systems of same - sign dislocations. with respect to this reference state, the initial random arrangement contains density fluctuations on all scales which may be either positive (@xmath118, positive excess) or negative (@xmath119, negative excess). to achieve screening, dislocations must migrate from regions of positive to regions of negative excess, and this process is governed by the long - range stress fields associated with the presence of (positive or negative) excess dislocations. in other words, the kinetics of the process follows from exactly the same scaling argument as used in the previous section : we are dealing with the stress - driven elimination of excess dislocation densities, with the only difference that the excess is now not of positive over negative dislocations, but of the local dislocation density over the average one. if the above argument is correct, the relaxation kinetics of same - sign dislocation systems should be characterised by a slow power - law stage which has the same characteristics as the relaxation of neutral dislocation systems as discussed in the previous sections. we demonstrate in the next section that this is indeed the case.
Comparison with simulation results
the numerically determined evolution of the elastic energy is displayed in. the zero value of the energy was chosen to correspond to the final relaxed state of the system. as seen on the figure, the evolution of the elastic energy @xmath120 per dislocation can be fitted satisfactorily with the prediction in, @xmath121. power - law relaxation occurs from times @xmath122 onwards, in good agreement with the model prediction @xmath123 in. the final equilibrium value of the elastic energy is not a priori known but was fitted to the data such as to achieve a maximum extension of the linear scaling regime. unfortunately this precludes determination of the second critical time @xmath124 from these results. the presence or absence of an external stress below the flow stress of the relaxed dislocation system seems to have negligible influence on the evolution of the elastic energy. note that it is possible to collapse the curves for different system sizes @xmath125 by normalising the graphs with @xmath126, as done in. this observation is consistent with the fact that the elastic energy of the initial random dislocation system is of the order of @xmath127 while the energy of the final relaxed state is of the order of @xmath128 @xcite. for estimating the value of @xmath129 we used that the range of dislocation pair correlations in mechanical equilibrium is of the order of the mean dislocation dislocation distance @xmath23 @xcite. therefore, the elastic energy difference per dislocation between the initial and final states is of the order of @xmath130 despite this relation connecting the initial and final states of the system, the numerical finding that the @xmath131 curves for different system sizes can be collapsed on their entire course by normalising them with @xmath126 is not trivial, as the agreement extends also to the relaxation kinetics and characteristic crossover time @xmath132. and external stress values @xmath133.] the numerically calculated evolution of the mean square velocity @xmath134 is displayed in for zero applied stress and different system sizes. due to the connection between the mean square velocity and the elastic energy of the system, it is not surprising that similar statements apply here as for the evolution of the elastic energy. as seen in the figure, the model prediction @xmath135 in fits the data well from @xmath122. due to the fact that @xmath136 is proportional to the time derivative of the elastic energy, its graphs are much noisier than those obtained for the energy, preventing again the detection of the supposed upper critical time @xmath124. as for the energy, size effects can be scaled out with a normalisation factor @xmath137 which is a direct consequence of and. a final analogy to the evolution of the elastic energy is that external stresses have only negligible influence on the evolution of the mean square velocity. for this reason, simulations with non - zero external stresses were omitted from. for different system sizes @xmath6 at zero external stress.] in the evolution of the mean absolute velocity @xmath138 can be seen for different system sizes. again, a power law time dependence can be observed from @xmath122 although an exponent @xmath139 gives a better fit than the theoretically predicted @xmath140 expected according to equation. one may argue that slowly moving dislocation dipoles play a bigger role in this case, as their small velocities contribute more strongly to @xmath141 than to @xmath142. therefore, the gradually increasing number of slowly moving dislocations might be responsible for the reduced relaxation exponent. what makes this figure very interesting is the possibility to estimate values of @xmath124. it was found that @xmath143 gives a good approximation, in line with the model prediction @xmath144 in. it was also observed that normalisation with @xmath145 collapses the graphs referring to different system sizes @xmath125 in the region of small @xmath146. this is consistent with the relations for the elastic energy and the mean square velocity. finally, as in case of the energy and the mean square velocity, the evolution of the mean absolute velocity is not changed by the presence of external stresses below the macroscopic flow stress. for different system sizes @xmath6 at zero external stress] another numerically measurable quantity is the plastic strain rate, defined as @xmath147 in the following the evolution of @xmath148 is studied for applied shear stresses @xmath149 below the macroscopic flow stress for the present dislocation geometry. from the data of miguel and co - workers @xcite, this is estimated to be @xmath150 in natural units. as it was demonstrated in, external stresses in this range do not appreciably change the evolution of the elastic energy. we study the relaxation of the strain rate mainly in order to assess, by comparing with the work of miguel et al., the relevance of different initial conditions on the creep behaviour of dislocation systems. shows the numerically determined evolution of the plastic strain rate @xmath148 for different system sizes and external stress values. as can be seen, the plastic strain rate scales roughly in proportion with the external stress. the relaxation does not follow any discernible power law but is roughly exponential. this is in marked contrast with the findings of miguel et al. @xcite who for well relaxed initial configurations demonstrate an andrade - type power - law decay, @xmath151. the discrepancy points to the crucial importance of initial conditions for relaxation processes in dislocation systems a factor which is also borne out by the history dependence of creep relaxation processes that was demonstrated by miguel et al. @xcite. normalised with the external stress @xmath133 for different system sizes @xmath6 and external stress values @xmath133.] in this section we investigate the influence of different ways of implementing the periodic boundary conditions by tilting the angle between the edges of the simulation box and the trace of the slip planes. simulations with different tilt angles @xmath31 are physically equivalent except for the spatial arrangement of the periodic images of each dislocation (see). this arrangement affects the dislocation dislocation interactions on scales comparable to the simulation box size. also, the interaction energy of each dislocation with its periodic images affects the initial elastic energy of the system, which is smallest for @xmath152 and has a maximum for @xmath153. (left) and @xmath153 (right). the latter configuration has a higher elastic energy as the nearest neighbours of each dislocation are in an energetically unfavourable configuration.,scaledwidth=80.0%] the influence of simulation box orientation on the relaxation process is illustrated in. the absolute values of the squared velocity (or equivalently the energy dissipation rate) are higher for @xmath153 than for @xmath154. however, both curves differ only by a constant factor (the ratio of the initial excess energies), while the dynamics of the relaxation processes is otherwise identical. in systems with a single slip geometry with the simulation box oriented at different angles @xmath31 to the slip planes.] for systems of dislocations of the same sign we have numerically evaluated the evolution of the mean square velocity @xmath155 (or, equivalently, of the energy dissipation rate) and of the mean absolute velocity @xmath138. all calculations were performed at zero external stress since any applied stress would induce a sustained drift motion of the dislocation arrangement. results are shown in and together with the fit functions obtained for neutral dislocation arrangements (see and). it is evident that the relaxation of same - sign dislocation systems follows the same scaling laws that have been observed for systems containing equal numbers of dislocations of both signs. this provides strong support for our basic conjecture that relaxation is governed by the stress - driven elimination of excess dislocations in a process that progresses from small to large scales. the processes occurring on short scales, on the other hand, are evidently different for the two systems (dipole formation vs. formation of walls). this is reflected by the fact that the relaxation process in the single - sign dislocation systems shows an initial size dependence which is not present in neutral dislocation systems (see and for comparison). for different system sizes @xmath6 for dislocations of the same sign.] for different system sizes @xmath6 for dislocations of the same sign.] compares the relaxation of dislocation systems in single, double and triple slip. as seen in the figure, at long times the relaxation in multiple slip geometries accelerates in comparison with the relaxation in single slip. this is consistent with the idea that in multiple slip geometries relaxation processes proceed through the formation of dislocation locks and the annihilation of mobile dislocations at these locks. the long range stress fields associated with dislocation locks and the removal of dislocations accelerate the relaxation as seen in the figure. one of the main assumptions behind our scaling model, namely that the motion of mobile dislocations is governed mainly by their mutual interaction, no longer holds in multiple slip geometries. therefore, the scaling model can not be applied to these situations. indeed, as seen in, no power law relaxation regime can be detected in the multiple slip simulations. in dislocation systems with single and multiple slip geometries. in the figure, @xmath1 means the total density of dislocations on all slip systems. the last two curves were shifted downwards to help comparison.]
Discussion and conclusions
the present paper discusses the relaxation of initially random arrangements of straight, parallel edge dislocations. following a phenomenological analogy with the kinetics of bimolecular reactions @xcite the relaxation process can be divided into three consecutive stages. stage 1 is characterised by rapid rearrangements of neighbouring dislocations, leading to the formation of dipoles, multipoles, and/or short wall segments. stage 2 hosts the gradual extinction of initial fluctuations in the burgers vector density on ever increasing length scales through the long range transport of excess dislocations. this stage gives rise to characteristic power - law relaxation dynamics. the relaxation process terminates in stage 3 when the characteristic fluctuation length reaches the system size. our considerations focus on the power - law relaxation dynamics in stage 2. in section 3, we formulated a scaling theory for this process by making the analogy with a bimolecular reaction. to this end, we considered a highly simplified picture where pair interactions between positive and negative excess dislocations lead to long - range dislocation transport resulting in the formation of dislocation dipoles. this led to predictions for the evolution of the elastic energy and the first two moments of the dislocation velocity. these predictions were then compared to ensemble averages of discrete dislocation dynamics simulations, and convincing agreement was found for single slip geometries. for multiple slip geometries, however, the persistent long range stress fields of dislocation locks accelerate the relaxation process, to which the scaling model can no longer be applied. the actual dislocation processes in many - dislocation systems are much more complex than the simplified picture underlying our scaling arguments. instead of long - range transport of excess dislocations and dipole formation, we see complex rearrangements resulting in dislocation dipoles, multipoles, and walls. in addition, it is well known that dislocations have a propensity to form large - scale heterogeneous patterns consisting of dislocation - rich and dislocation depleted regions. in the following we briefly discuss how these complex static and dynamic features fit into the idealised picture we used in the previous sections. we have developed our scaling argument for dipole formation which can be considered a bimolecular reaction between positive and negative dislocations. however, actual dislocation arrangements are much more complex. it is therefore important to emphasise that the core of the argument is the elimination of large - scale fluctuations in the excess burgers vector density, and _ not _ the resulting arrangement of nearby dislocations. for the long - time asymptotics of the relaxation process, which is governed by the elimination of fluctuations on larger and larger scales, the small - scale arrangement of dislocations is virtually irrelevant at least as long as the local features (dipoles, multipoles, short walls and combinations of all these) do not give rise to long - range stresses. according to the present argument, a well - screened dislocation arrangement is one in which burgers vector fluctuations have been eliminated on all scales. if we are dealing with a neutral dislocation system (equal numbers of positive and negative dislocations) this means that the net burgers vector is zero in each small volume, i.e., for each dislocation we can find exactly one partner of opposite sign nearby. this motivates the dipolar picture even if the actual dislocation arrangements may be more complex. an alternative mechanism for creating well - screened dislocation arrangements is the formation of system - spanning walls of dislocations of the same sign. periodic arrangement of edge dislocations of one sign into a wall perpendicular to the glide plane removes the logarithmic divergence of the dislocation energy and introduces a screening length that is proportional to the dislocation spacing along the wall @xcite. walls are conspicuous both in simulations and in many experimentally observed dislocation microstructures. at first glance, wall formation mechanism seems to be completely at odds with the mechanism discussed in the present paper : formation of walls of same - sign dislocations increases, rather than reduces, the burgers vector density fluctuations. however, a closer investigation reveals that wall formation by itself is not a screening mechanism at all. forming a wall of randomly spaced dislocations does not reduce the energy in comparison with a random 2d dislocation arrangement (see appendix). instead, the screening effect is contingent on the equal spacing of dislocations, i.e. on suppressing fluctuations of the burgers vector density along the wall direction. if we start from a random dislocation arrangement this is not easy to obtain : either the dislocations must have climb degrees of freedom (which we do not consider in the present study), or dislocation motions that lead to the formation of multiple walls must be correlated over large distances such as to ensure that each wall collects only those dislocations that fit into an evenly spaced pattern. in the latter case, we are again dealing with the suppression of burgers vector density fluctuations on all scales above the wall spacing, and the long time asymptotics of this process is expected to obey our scaling theory. this is confirmed by the simulations, which however also demonstrate that the short - time behaviour is different for neutral dislocation systems where the local arrangement of dislocations is characterised by dipolar and multipolar patterns, and for single - sign dislocation systems where the local arrangement of dislocations is characterised by walls (compare and). our scaling argument considers the stress - driven long - range transport of excess dislocations. the picture underlying the argument is schematically shown in (top) : a positive excess dislocation at a is attracted by a negative excess dislocation at b and the two recombine by long - range motion which is not affected by the stress field of the dislocations in between a and b. the figure also indicates that this idealisation may not be feasible when we are dealing with multipolar arrangements rather than isolated narrow dipoles : in that case the mutual interaction of the excess dislocation may be much weaker than their interaction with other dislocations ` on the way '. as a consequence, recombination is much more likely to occur by a collective rearrangement as shown in (bottom). how does this affect our scaling argument? the total driving force for the process is the same in both cases. however, a collective rearrangement on scale @xmath2 is likely to involve @xmath156 dislocations [@xmath157 in (bottom)]. hence the driving force per dislocation is reduced by a factor of the order of @xmath6. however, the same is true for the characteristic distance that has to be covered by each dislocation (@xmath2 in the case of direct transport, @xmath158 in the case of collective rearrangement). as we assume that the dislocation velocity is proportional to the driving force, it follows that the characteristic time scale for eliminating the excess dislocation is the same in both cases, and our scaling argument remains valid. it is a well known phenomenon that dislocation microstructures forming during plastic deformation form heterogeneous patterns consisting of regions of high and low dislocation density, with characteristic lengths that are large in comparison with the dislocation spacing. on a conceptual level, possible mechanisms underlying this patterning were discussed by nabarro @xcite who pointed out that it may be energetically favourable to ` segregate'the dislocation microstructure into areas of high and low density : if we are dealing with a well - screened dislocation arrangement, the energy density scales like @xmath159. in this case it can be easily shown that it is energetically favourable to increase @xmath1 in some regions and decrease it proportionally in others. such ` energetically driven'dislocation patterning could be a reason for the formation of dislocation - dense and dislocation - depleted regions that is observed in many experiments. while this mechanism is not covered by the present model, it is not at variance with our considerations : in a neutral dislocation arrangement, the formation of dislocation - dense and dislocation - depleted regions might occur without disturbing the burgers vector balance. we note, however, that in our simulations large - scale dislocation patterning is not observed either because it does indeed not occur in single slip, or because the dislocation numbers in our simulations might be too small. in conclusion we discuss the relevance of the processes discussed in the present paper for real - world systems. the relaxation of a random dislocation system has no direct counterpart in real deformation experiments, since it is impossible to ` prepare'such a random system in the first place. our analysis of the screening of same - sign dislocation systems is, however, of general importance for understanding real dislocation patterns since it demonstrates that wall formation, though a conspicuous feature, can by itself not account for screening. this observation points to the importance of investigating long - range correlations between dislocation positions both within the walls and across different walls, and offers ample scope for future investigations. the investigated processes are of significant importance for discrete dislocation dynamics simulations of plasticity as the slow nature of the relaxation makes it difficult to obtain well - defined and energetically stable initial configurations. our comparison of strain - rate relaxation experiments with the results of miguel et al. demonstrates that the collective behaviour of dislocation systems may depend significantly on initial conditions. the analysis of this dependence is still in its infancy, yet understanding it is indispensable for carrying out dislocation plasticity simulations in a controlled and well - defined manner. financial support of the hungarian scientific research fund (otka) under contract no. k 67778, of the european community s human potential programme under contract nos. mrtn - ct-2003 - 504634 [sizedepen] and nmp3-ct-2006 - 017105 [digimat] and of nest pathfinder programme trigs under contract nest-2005-path - com-043386 are gratefully acknowledged.
Appendix: calculation of the elastic energy of a random dislocation wall
we consider a wall of infinite height running along the plane @xmath160 in an isotropic material. edge dislocations of burgers vector @xmath161 are distributed randomly along the wall with average linear density @xmath162. the geometry corresponds to a plane - strain situation, hence the elastic energy density can be written as @xmath163. \label{edens}\]] the total energy of the system is obtained by integrating over the system volume, @xmath164 where we have used that, for an infinite system, the second and third terms on the right - hand side do not contribute to the total energy. this can be shown as follows : for plane - strain deformation, the stresses can be written as derivatives of the airy stress function @xmath16, @xmath165, @xmath166 and @xmath167. hence, @xmath168 { \rm d}^2 r = \int_v [\partial_x\partial_y\chi \partial_x \partial_y \chi - \partial_x^2\chi \partial_y^2 \chi] { \rm d}^2 r. \label{airy}\]] partially integrating the second term in the integral on the right - hand side with respect to @xmath8 and @xmath115 shows that this integral contributes only surface terms to the total energy. these terms are negligible in the infinite - system limit. the ensemble - averaged stress at any point is given by summing over the stress fields of the individual dislocations in the wall and averaging over the different realizations of the random dislocation positions : @xmath169 where @xmath170 is the @xmath171 component of the stress created at @xmath172 by the @xmath106th dislocation. the elastic energy of the system depends on the averages of products @xmath173 where @xmath174 $]. in evaluating these averages we use that the @xmath115 coordinates of the individual dislocations are independent random variables : @xmath175. \label{indep}\end{aligned}\]] we now make the following observations : * the average stresses @xmath176 and their products @xmath177 depend on the @xmath8 coordinate only. * the average single - dislocation stresses @xmath178 and @xmath179 become zero in the limit @xmath180, since these stresses are antisymmetric functions of the @xmath115 coordinate. the same is true for the average total stresses @xmath181 and @xmath182. with these observations and using we can write the system energy as @xmath183, \label{efinal}\end{aligned}\]] where the second step follows by interchanging the averaging and the integration. @xmath2 is the system size (tending to infinity) which, in the absence of any other screening mechanism, delimits the divergence of the dislocation self - energy. for periodic boundary conditions, as used in our simulations, @xmath2 must be understood as the size of the periodic simulation box which in this case defines the screening length for an otherwise uncorrelated dislocation arrangement. @xmath184 is the total number of dislocations in the system. it follows from that the energy per dislocation is equal to the energy of a single unscreened dislocation and, hence, equals the energy in a completely random 2d arrangement. in other words, the arrangement of dislocations of the same sign in a random wall does (with the possible exception of surface terms that are negligible in the infinite system limit) not produce any reduction of the total energy. accordingly, the thermodynamic driving force towards forming such a wall is zero. zaiser m, _ statistical modelling of dislocation systems _, 2001 _ mat. a * 309310 * 30415 groma i, csikor f f and zaiser m, _ spatial correlations and higher - order gradient terms in a continuum description of dislocation dynamics _, 2003 _ acta mater. _ * 51 * 127181 zaiser m, _ scale invariance in plastic flow of crystalline solids _, 2006 _ adv. * 55 * 185245 miguel m c, vespignani a, zaiser m and zapperi s, _ dislocation jamming and andrade creep _, 2002 165501 miguel m c, moretti p, zaiser m and zapperi s, _ statistical dynamics of dislocations in simple models of plastic deformation : phase transitions and related phenomena _, 2005 _ mat _ a * 400401 * 1918 miguel m c, laurson l and alava m, _ material yielding and irreversible deformation mediated by dislocation motion _, 2008 _ eur. phys. b * 64 * 44350 wilkens m, _ das spannungsfeld einer anordnung von regellos verteilten versetzungen _, 1967 _ _ * 15 * 14127 wilkens m, _ das mittlere spannungsquadrat @xmath185 begrenzt regellos verteilter versetzungen in einem zylinderfrmigen krper _, 1969 _ 11559 zaiser m, miguel m - c and groma i, _ statistical dynamics of dislocation systems : the influence of dislocation dislocation correlations _, 2001 b * 64 * 224102 zaiser m and seeger a, _ long - range internal stresses, dislocation patterning and work hardening in crystal plasticity _, 2002, _ dislocations in solids vol. 11 _, ed f r n nabarro and m s duesbery (elsevier) csikor f f and groma i, _ probability distribution of internal stress in relaxed dislocation systems _, 2004 b * 70 * 064106 krivoglaz m a 1969 _ theory of x - ray and thermal neutron scattering by real crystals _ (new york : plenum) csikor f f, kocsis b, bak b and groma i, _ numerical characterisation of the relaxation of dislocation systems _, 2005 _ mat. eng. _ a * 400401 * 2147 csikor f f and zaiser m, _ scaling and glassy dynamics in the relaxation of dislocation systems _, 2006 _ proc. conf. on statistical mechanics of plasticity and related instabilities (29 august2 september 2005 bangalore) _ ed m zaiser _ et al _ (proceedings of science) 058 hirth j p and lothe j 1982 _ theory of dislocations _ (new york : wiley - interscience) bak b, groma i, gyrgyi g and zimnyi g, _ dislocation patterning : the role of climb in meso - scale simulations _, 2006 _ comp. * 38 * 228 ovchinnikov a a and zeldovich ya b, _ role of density fluctuations in bimolecular reaction kinetics _, 1978 _ chem. _ * 28 * 2158 groma i, gyrgyi g and kocsis b, _ debye screening of dislocations _, 2006 165503 thomson r, koslowski m and lesar r, _ energetics and noise in dislocation patterning _, 2006 b * 73 * 024104 nabarro f r n, _ complementary models of dislocation patterning _, 2000 _ phil. mag. _ a * 80 * 75964 | we study the relaxation dynamics of systems of straight, parallel crystal dislocations, starting from initially random and uncorrelated positions of the individual dislocations. a scaling model of the relaxation process
is constructed by considering the gradual extinction of the initial density fluctuations present in the system.
the model is validated by ensemble simulations of the discrete dynamics of dislocations.
convincing agreement is found for systems of edge dislocations in single slip irrespective of the net burgers vector of the dislocation system.
it is also demonstrated that the model does not work in multiple slip geometries.
_ keywords _ : defects (theory), fluctuations (theory), plasticity (theory) | 0811.1192 |
Introduction
in the ctf ii drive beam gun, cs - te photocathodes are used to produce a pulse train of 48 electron bunches, each 10ps long and with a charge of up to 10nc @xcite. in ctf, the main limit to lifetime is the available laser power, which requires a minimal quantum efficiency (qe) of 1.5% to produce the nominal charge. although cs - te photocathodes are widely used, a complete understanding, especially of their aging process, is still lacking. spectra of the qe against exciting photons may help to understand the phenomenon.
Measurements of qe against photon energy
according to spicer @xcite, the spectra of the quantum efficiency (qe) of semiconductors with respect to the energy of the exciting photons (@xmath0) can be described as : @xmath1 where @xmath2 is the threshold energy for photoemission, c@xmath3 and c@xmath4 are constants. to measure the spectral response of photocathodes, wavelengths from the near uv throughout the visible are necessary. to attain these, an * o*ptical * p*arametrical * o*scillator was built @xcite. a frequency - tripled nd : yag laser pumps a betabarium borate (bbo) crystal in a double - pass configuration, as shown in fig.[fig : opo]. the emerging signal - beam, with wavelengths between 409 nm and 710 nm, is frequency doubled in two bbo crystals. the wavelengths obtained are between 210 nm and 340 nm. the idler - beam delivers wavelengths between 710 nm and @xmath5 nm. the measurements of the spectral response of photocathodes were made in the dc - gun of the photoemission lab at cern @xcite, at a field strength of about 8 mv / m. spectra were taken shortly after the evaporation of the cathode materials onto the copper cathode plug, as well as after use in the ctf ii rf - gun @xcite at fields of typically 100 mv / m. to be able to interpret the spectra in terms of spicer s theory, it was necessary to split the data into 2 groups, one at `` low photon energy '' and one at high photon energy , see fig.[fig : cath87]. then, the data can be fitted well with two independent curves, following eq.([eq : spicer]), which give two threshold energies. for a typical fresh cs - te cathode, the high energy threshold is 3.5ev, the low one is 1.7ev, as shown in fig.[fig : cath87], upper curve. this might be a hint that two photo - emissive phases of cs - te on copper exist. several explanations are possible : the copper might migrate into the cs - te, creating energy levels in the band gap ; or possibly not only cs@xmath4te, but also other cs - te compounds might form on the surface and these might give rise to photoemission at low photon energy. a hint to this might be that the ratio of evaporated atoms of each element is not corresponding to cs@xmath4te, see below. after use, we found that not only the complete spectrum shifted towards lower quantum efficiency, but also that the photoemission threshold for high qe increased to 4.1ev, which is shown in fig.[fig : cath87], lower curve. one might expect that the photocathode is poisoned by the residual gas, preventing low - energy electrons from escaping. however, because typical storage lifetimes are of the order of months, the effect must be connected to either the laser light, or the electrical field. we also produced a cs - te cathode on a thin gold film of 100 nm thickness. as shown in fig.[fig : cath120], the shoulder in the low energy response disappeared. it is difficult to fit a curve for the spicer model to the low energy data. the high " photoemission threshold is at 3.5ev. at the moment, this cathode is in use in the ctf ii gun and will be remeasured in the future. in terms of lifetime, this cathode is comparable to the best cs - te cathodes, as it has already operated for 20 days in the rf - gun. as a new material presented first in @xcite, we tested rubidium - telluride. we took spectra of qe before and after use in the ctf ii gun, as for cs - te. remarkably, with this material, there was no shift in the photoemission threshold towards higher energies, but only a global shift in qe, see fig.[fig : rb2te]. this might be due to the lower affinity of rubidium to the residual gas. detailed investigations are necessary to clarify this. long lifetimes for cs - te cathodes are achieved only when they are held under uhv (@xmath6 mbar). other photocathode materials like k - sb - cs are immunized against gases like oxygen by evaporating thin films of csbr onto them @xcite. therefore, we evaporated a csbr film of 2 nm thickness onto the cs - te. fig.[fig : csbr] shows the spectrum before the csbr film (square points) and after it (round points). the qe at 266 nm dropped from 4.3% to 1.2%. in addition, the photoemission threshold was shifted from 3.9ev to 4.1ev. a long - term storage test showed no significant difference between uncoated and coated cathodes. more investigations will determine the usefulness of these protective layers. in order to increase the sensitivity of the on - line qe measurement during evaporation of the photocathodes, we monitored the process with light at a wavelength of 320 nm. we did not see any significant improvement in sensitivity, notably in the high qe region.
Stoichiometric ratio
film thicknesses are measured during the evaporation process by a quartz oscillator @xcite. typical thicknesses for high quantum efficiencies at @xmath7 nm are 10 nm of tellurium and around 15 nm of cesium. this results in a ratio of the number of atoms of each species of @xmath8, far from the stoichiometric ratio of 0.5 for cs@xmath4te. it is known that tellurium interacts strongly with copper @xcite, so that not all of the evaporated tellurium is available for a compound with subsequently evaporated cesium. therefore, we used also mo and au as substrate material. however, the ratio between the constituents necessary for optimum qe, did not change significantly. another reason might be that instead of cs@xmath4te, cs@xmath4te@xmath9 is catalytically produced on the surface. this compound, as well as some others, was found to be stable @xcite.
Lifetime in ctf ii
lifetime in ctf depends on parameters like maximum field strength on the cathode, vacuum and especially extracted charge. typically, a cathode is removed from the gun, if the qe falls below 1.5%. as shown in fig.[fig : lifetime], lifetime does not depend on the initial qe ; a cathode having an initial qe of 15% (round points) lasted as long as one with 5% (triangles).
High charge test
as shown in table[tab1], the average current produced in ctf ii is nearly a factor 10000 lower than what is required for the clic drive beam. a test to produce 1mc is under preparation in the photoemission laboratory at cern. the exact reproduction of the clic pulse structure would require the clic laser, which is still in the design stage .comparison of cathode relevant parameter [cols="^,^,^,^",options="header ",] [tab1] in a collaboration between rutherford appleton laboratory and cern. a test which is compatible with our current installation is the production of 1ma of dc current, which requires a uv laser power of 300mw at the cathode. for this test, we will illuminate the cathode with pulses of 100ns to 150ns pulse length, at repetition rates between 1khz and 6khz. as table[tab1] shows, this is a factor 1000 more average current than in ctf ii, and also demonstrates the basic ability of the cathodes to produce the ctf 3 drive beam (i=26@xmath10a). clic is still a factor 75 away. we are currently searching for ways to produce higher charges as well.
Conclusion
measurements of qe against photon energy are routinely made after production and after use of photocathodes. we have demonstrated that both low energy and high energy responses agree well with spicer s theory. a gold buffer layer reduces the low energy response of cs - te cathodes. more work is needed to understand the measurements of the stoichiometric ratio of cs - te. coating with 2 nm csbr significantly decreased the quantum efficiency, without improving the storage lifetime. for the high - charge drive beam of clic, it is still necessary to demonstrate the capabilities of cs - te, for which first tests will be done soon. e. chevallay, j. durand, s. hutchins, g. suberlucq, m. wrgel, photocathodes tested in the dc gun of the cern photoemission laboratory ", nuclear instruments methods in physics research section a, vol. 340, (1994) 146 - 156, cern clic note 203 e. shefer, a. breskin, r. chechik, a. buzulutskov, b.k. singh, m. prager, coated photocathodes for visible photon imaging with gaseous photomultipliers ", nuclear instruments methods in physics research section a, vol.433, no.1 - 2, (1999) 502 - 506 | for short, high - intensity electron bunches, alkali - tellurides have proved to be a reliable photo - cathode material.
measurements of lifetimes in an rf gun of the clic test facility ii at field strengths greater than 100 mv / m are presented. before and after using them in this gun, the spectral response of
the cs - te and rb - te cathodes were determined with the help of an optical parametric oscillator.
the behaviour of both materials can be described by spicer s 3-step model. whereas during the use the threshold for photo - emission in cs - te
was shifted to higher photon energies, that of rb - te did not change.
our latest investigations on the stoichiometric ratio of the components are shown.
the preparation of the photo - cathodes was monitored with 320 nm wavelength light, with the aim of improving the measurement sensitivity.
the latest results on the protection of cs - te cathode surfaces with csbr against pollution are summarized.
new investigations on high mean current production are presented. | physics0008061 |
Introduction
the interest in the feas - based superconductors@xcite is ongoing after six years of extensive research as still no consensus has been achieved concerning the superconducting character and pairing mechanism. lifeas is special amongst the many feas - based superconductors, as superconductivity appears in the parent compound at elevated temperatures without doping or application of pressure. this particularity of lifeas most likely arises from its electronic structure with strongly reduced nesting between electron and hole fermi - surface sheets as it was first deduced from angle - resolved photoemission spectroscopy (arpes) @xcite. in the 1111 and 122 families (named after their stoichiometry) the fermi nesting conditions are excellent stabilizing a spin density wave (sdw), which has to be suppressed by doping@xcite or the application of pressure@xcite in order to reach the superconducting state. lifeas does not exhibit any structural transition nor a magnetically ordered phase.@xcite theoretical calculations@xcite explain this fact by its poor fermi nesting properties and unusually shallow hole pockets around the @xmath2 point, which is in agreement with arpes experiments.@xcite the flat top of the hole pockets implies a large density of states around the @xmath2 point and in combination with small - momentum scattering vectors within the inner hole pocket this would favor ferromagnetic fluctuations and a triplet pairing mechanism.@xcite the symmetry of the order parameter has been a controversial subject, several reports using arpes, quasiparticle interference (qpi) or theoretical approaches favor an @xmath3 wave,@xcite while there is also support for a @xmath4-wave state.@xcite although the calculations in ref. support an @xmath3 wave state driven by collinear antiferromagnetic fluctuations, the authors state that ferromagnetic fluctuations stemming from the small hole pocket at the @xmath2 point may dominate at higher energies and/or at higher temperatures. in our previous work@xcite we have established the energy and temperature dependence of an antiferromagnetic excitation located at an incommensurate position @xmath5 resembling magnetic correlations in electron doped bafe@xmath1as@xmath1. similar results were obtained by wang et al. @xcite the origin of the magnetic signal has been interpreted as scattering between the electron pockets centered around the @xmath6 point and either the outer@xcite or the inner@xcite hole pockets around the zone center.in this work we present a comprehensive inelastic neutron scattering (ins) study using different cold and thermal triple - axis spectrometres and a time - of - flight instrument devoted to extend the characterization of the incommensurate antiferromagnetic fluctuations in single - crystalline lifeas. we present the inelastic scattered neutron intensity in absolute units using two different techniques leading to perfectly agreeing results. the magnetic fluctuations have been investigated up to energy transfers of 80 mev and spin - space anisotropies have been studied by polarized neutrons with longitudinal polarization analysis (lpa). furthermore, we have investigated @xmath7 in a broad @xmath8-@xmath9 range to search for any ferromagnetic fluctuation at elevated temperatures and energy transfers.
Experimental
the same single crystal sample as in ref. has been used for all the experiments presented here. the normalization to an absolute intensity scale has been done with data obtained at the thermal triple - axis spectrometer 1 t (laboratoire lon brillouin, saclay), which was used with a pyrolytic graphite (pg) monochromator and a pg analyzer. the final neutron energy was fixed at @xmath10 mev. the in20 spectrometer (institut laue - langevin, grenoble) was used with the flatcone multianalyzer in order to record @xmath11-maps with different @xmath12 values at different temperatures and energy transfers. in20 has also been used in the polarized mode using polarizing heusler (111) crystals as a monochromator and an analyzer. for the lpa a set of helmholtz coils was used to guide and orient the neutron polarization. lpa offers the possibility of distinguishing between nuclear and magnetic scattering and it furthermore allows the separation of the two magnetic components perpendicular to the scattering vector. generally, nuclear scattering is a non - spin - flip (nsf) process regardless of the initial neutron polarization state. only magnetic components perpendicular to the scattering vector (@xmath13 by definition) are accessible in a neutron experiment. the components perpendicular to the polarization axis (@xmath14 being in the scattering plane and @xmath15 being the perpendicular axis of the spectrometer) contribute to the spin - flip (sf) channel, while those parallel to the axis of polarization scatter into the nsf channel. the puma spectrometer (frm - ii, garching) was used with a pg monochromator and a pg analyzer with a fixed final neutron energy of @xmath10 mev. high energy transfers were measured at the time - of flight spectrometer maps (rutherford - appleton laboratory, didcot). the incident beam energies were @xmath16 and 100 mev with @xmath17 parallel to the @xmath18 axis. the measured intensities were normalized to absolute units by using a vanadium standard (with 30% error).
Results and discussion
in order to express the dynamic susceptibility of lifeas in absolute units data taken on the time - of - flight spectrometer maps and triple - axis spectrometer data from the 1 t instrument were used yielding perfect agreement. the time - of - flight data can be normalized by comparison with incoherent scattering from a vanadium sample and with the sample mass. this procedure is well - established at the maps instrument and described in large detail in reference.@xcite in contrast the normalization of triple - axis data is more complex as the resolution function and the beam profile are more structured. here we follow the most common way to normalize the magnetic scattering by comparison with phonon measurements on the same sample. this method, furthermore, excludes mistakes arising from impurity phases. the scattering potential of the sample is discussed in terms of the double - differential cross section @xmath19 with @xmath20 the final energy. in any ins experiment this entity is folded with the resolution and transmittance function of the instrument. we use the reslib programs@xcite to quantitatively analyse the scattering intensities. in our experiment a neutron monitor between the monochromator and the sample is used to scale the detector counts into the entity counts per given monitor (note that this monitor is corrected for higher order contaminations). the calculation splits the instrumental effects in the finite gaussian resolution and a transmittance term. the intrinsic double differential cross section is first folded with the gaussian resolution in the four - dimensional space consisting of q - space and energy, see eq. [eq : ddcs]. here @xmath21 is the resolution matrix according to the popovici approximation @xcite and @xmath22 the four - dimensional difference vector consisting of the q - space coordinates and energy, see eq. [eq : delta]. @xmath23 \right) } \label{eq : ddcs } \\ { \bf \delta } : = & (q'_x - q_x, q'_y - q_y, q'_z - q_z,\omega'-\omega) \label{eq : delta}\end{aligned}\]] in order to calculate the intensity in the detector one has to multiply the folded double cross section with a normalization factor @xmath24 describing amongst others the efficiency of the secondary spectrometer and the resolution function normalization @xmath25. in contrast to the reslib manual we do not include the @xmath26 factor to @xmath27 but follow the common practice keeping this factor in the double - differential cross section.@xcite @xmath28 for known resolution and transmission functions the study of a predictable signal allows one to determine the scale factor @xmath18 describing amongst others the effective sample size. the transformed double - differential cross section @xmath29 thus contains the intrinsic scattering strength of the system combined with the spectrometer properties. we use the scattering by an acoustic phonon for normalization. the single - phonon cross section is given by eq. [eq : crosssec] (refs. ) where @xmath30 is the bose population factor for neutron energy loss. @xmath31 denotes the dynamical structure factor of the particular phonon mode at this scattering vector, which can be calculated with the help of a lattice - dynamical model, see eq. [eq : fdyn]. the @xmath32 function in eq. [eq : fdyn2] is approximated in the calculation by a lorentzian profile with finite half width. the symbols in eqs. [eq : crosssec]-[eq : fdyn2] follow the same convention as in refs. . @xmath33 for an acoustic phonon close to the brillouin - zone center one may further simplify the calculation as all atoms in the primitive cell are parallel polarized with components @xmath34 (here @xmath35 and @xmath36 denote the individual and total masses, respectively). the dynamic structure factor then corresponds to that of the nuclear bragg reflection multiplied by the length of the scattering vector, @xmath37, the inverse square root of the total mass and by the cosine of the angle between scattering vector and phonon polarization, @xmath38. the latter factor is close to one in a reasonably chosen scan. @xmath39 } \label{eq : fdyn2}\]] the double differential cross section of the phonon scattering is obtained by subtracting the (refined) background from the raw data and then by dividing by the (refined) scale factor. the phonon dispersion is described by a simple linear relation, @xmath40. fitting the phonon cross - section with its intensity prefactors to the raw data using the reslib code yields a scale factor of 13.1(8) and a constant background of 2 counts per monitor. the raw data can therefore be converted into an absolute scale that still contains the resolution functions of the instrument, see the right axis of ordinate in fig. [fig : phonon]. (color online) raw data showing the transversal @xmath41-scan across the (220) phonon at @xmath42 k an energy transfer of 4.5 mev measured at the 1 t spectrometer. the scattered intensity is given in counts/(monitor 25000) on the left ordinate and in absolute cross section values on the right ordinate for the folded cross section @xmath43., scaledwidth=48.0%] in order to evaluate the magnetic signal we start with the autocorrelation of the spin fourier coefficients @xmath44, here @xmath45 denote the space indices, @xmath46 the neutron gyromagnetic factor, @xmath47 the electron charge, @xmath48 the electron mass, @xmath18 the speed of light, @xmath49 the magnetic form factor at the scattering vector, @xmath50 the land factor and @xmath51 the kronecker symbol. note that the second line of eq. [eq : ddcsmag] has the unit of an inverse energy (ev@xmath52). @xmath53 with the fluctuation dissipation theorem one may transform the cross section to the imaginary part of the generalized dynamic susceptibility, which we assume here to be isotropic in spin space. @xmath54\cdot 2\cdot \chi '' ({ \bf q},\omega) \nonumber \\ \label{eq : ddcsmag2}\end{aligned}\]] a susceptibility can be given in various units creating considerable confusion but here the unit problem drops out due to the term bohr - magneton, @xmath55, squared in the denominator. the natural microscopic unit to discus thevsusceptibility is thus @xmath56 per formula unit, which we will use in the following. deducing the absolute scale of the cross section of the magnetic fluctuation is now obtained in the same way as in the phonon case by subtracting the background and by dividing by the scale factor obtained from the phonon fit. however, in order to fit the data and deduce the background a model is needed to describe the generalized susceptibility @xmath57. we assume a superposition of single relaxor functions (eq. [eq : relaxor]) in energy with a lorentzian q - dependence centered at the four positions @xmath58=(0.5@xmath59@xmath60,0.5@xmath61@xmath60,0) and (0.5@xmath59@xmath62,0.5@xmath61@xmath62,0). we take only the in - plane components of @xmath63 into account. @xmath60 (@xmath62) is the incommensurability of 0.057(3) r.l.u. [0.17(2) r.l.u.] (see sec. [sec : tas] for a detailed description of the two signals) and the hwhm was refined to @xmath64=0.042(9) r.l.u [@xmath65=0.07(3)], which yields the best agreement with the experimental data. @xmath66 in a first step, the constant background (11 counts per monitor 25000 @xmath67 20 s) was determined and subtracted from the raw data which was then divided by the scale factor deduced from the phonon fit yielding the transformed double - differential cross section @xmath43. from this one may obtain a susceptibility folded with the instrument resolution and transmission by dividing by all the intensity prefactors in eq. [eq : ddcsmag2] ; this result is shown in fig. 2 on the right coordinate axis. the intrinsic strength and shape of @xmath57, however, can only be obtained by fitting the model, eq. [eq : relaxor], to the raw data (a value of 10.9 mev has been used for @xmath2 obtained by the single - relaxor fit to the data presented in sec. [sec : results].b). thereby we obtain @xmath68=7.4(8) and at the outer and inner incommensurate positions, respectively, see fig. [fig : absfluct]. (color online) incommensurate antiferromagnetic fluctuations at @xmath42 k and at an energy transfer of 5 mev measured at the 1 t spectrometer. the ordinates are given in absolute units of cross section (left) or of the generalized susceptibility (right). the inset shows the raw magnetic data before subtraction of the background and the division by the scale factor.,scaledwidth=48.0%] the incommensurate fluctuation has been reinvestigated at the thermal triple - axis spectrometer puma (frm - ii, garching). [fig : puma] shows transverse q - scans across the (0.5 0.5 0) position at different energy transfer clearly documenting a complex q - shape of the magnetic response. the incommensurate magnetic correlations exhibit at least an asymmetric profile with pronounced shoulders towards larger incommensurability (compared to the (0.5,0.5,@xmath69) center). therefore, the data have been described with two pairs of symmetrical gaussian functions on a constant background. note that parts of the data are contaminated by phonon scattering towards lower energy transfer for which these data points are not shown. the resulting fit curves [(red) solid lines]) show a very good agreement with the raw data, while the dashed (dash - dotted) curves indicate the contribution of the signal at @xmath70@xmath71(0.43 0.57 0) [@xmath72@xmath71(0.35 0.65 0)]. angle resolved photoemission spectroscopy (arpes) experiments @xcite have revealed the fermi surface to consist of two similarly sized electron - like sheets around the @xmath73 point and hole - like sheets around the @xmath2 point. in ref. the authors have identified the ins signal to be connected to scattering between the outer hole pocket and the electron pockets by using a simple tight - binding fit to the arpes data, while involvement of the inner hole pocket was concluded in ref. . deeper understanding of the nesting signal requires the analysis of the orbital character of the various fermi surface sheets which essentially arise from the @xmath74 @xmath75, @xmath76 and @xmath77 orbitals.@xcite there seems to be agreement that the outer hole pocket can be identified with @xmath77 orbital character which also contributes to the electron pockets. the @xmath77 states should result in two - dimensional bands, but @xmath75 and @xmath76 contributions yield considerable dispersion along the perpendicular directions and strong @xmath78 modulation of the fermi surfaces.@xcite if one associates the nesting magnetic correlations exclusively with @xmath77 orbitals it appears difficult to understand a split signal but an asymmetry or a shoulder can arise from a peculiar detail of the fermi surface shape that is not sufficiently well understood so far. sr@xmath1ruo@xmath79 is a well studied example with incommensurate magnetic correlations arising from fermi - surface nesting,@xcite and this material also exhibits an asymmetric magnetic response with a shoulder. quite recently four theoretical papers aimed to quantitatively model the variation of the superconducting gap on the fermi surface sheets arriving at contradictory results.@xcite the quantitative description of magnetic excitations by analyzing transitions between states with the same or different orbital character will help to arrive at a better understanding of the electronic structure of lifeas. the peak intensities of @xmath70 and @xmath72 have been followed as a function of energy above and below @xmath80. as it can be seen in fig. [fig : escanpuma] the main signal @xmath70 shows the same dependence as already reported in ref. with a crossover between the scattered intensity below and above @xmath80 at 4.5 mev and an increase of intensity above 7 - 8 mev. on the other hand the intensity at @xmath72 suggests a different behaviour in dependence on the energy transfer. the scattered intensity at 20 k stays above the one at 5 k up to an energy transfer of roughly 7 mev, above which the value @xmath81@xmath82@xmath83 becomes stronger than @xmath81@xmath84@xmath83. the different energy dependences of the signals at @xmath70 and @xmath72 strengthen the assumption of their independent origin and can be explained due to different gap values on different parts of the fermi surface. (color online) energy dependence of the ins scattering at @xmath70=(0.425 0.575 0) and @xmath72=(0.35 0.65 0) measured at the puma spectrometer. the dashed lines indicate where the scattered intensities of the normal [(red) circles)] and superconducting state [(blue) squares] cross.,scaledwidth=42.0%] the derived amplitude of the excitation at @xmath70 has been corrected for the monitor and the bose factor yielding the imaginary part of the generalized susceptibility which is shown in fig. [fig : relaxor]. the data have been fitted with single - relaxor functions. the data does not allow to state a clear tendency of the critical energy, however, a clear reduction of @xmath85 towards higher temperatures is observable. in addition the incommensurate magnetic correlations become strongly broadened at the temperature of only 100 k where the two peak structure has already changed into a broad plateau. (color online) imaginary part of the generalized susceptibility at 5 k, 20 k and 100 k as obtained by the amplitude from the fits to the data shown in fig. [fig : puma] and correction for the monitor and the bose factor. the solid lines represent fits by a single relaxor functions @xmath86,scaledwidth=42.0%] by using the triple - axis spectrometer in20 in combination with the flatcone multianalyzer, planar sections of the reciprocal space can be recorded by simple @xmath87 scans which are afterwards converted into @xmath8-space. as theory predicts that a ferromagnetic instability may dominate at higher temperatures and/or higher energies,@xcite maps of the reciprocal space have been recorded up to 150 k and an energy transfer of 40 mev focusing on the (100) and (110) positions. however, our obtained data does not give any hint for ferromagnetic fluctuations in lifeas. the in20 spectrometer has then been used with polarized neutrons whose polarization axis after the scattering process has been analyzed. the observation of the incommensurate signal in the sf channels proves its magnetic origin (note that the sf background has been subtracted according to the description in ref. ). the peak intensity at the point @xmath8=(0.43 0.57 0) has been measured as a function of the energy transfer for the sf@xmath88 and sf@xmath89 channels (fig. [fig : escan]). although only the sf@xmath88 channel has been measured with high statistics a slight spin - space anisotropy of the magnetic fluctuation is visible between 6 and 12 mev, where the out - of - plane fluctuation lies above the in - plane fluctuation similar to observations in electron doped bafe@xmath1as@xmath1 @xcite. however, the spin - space anisotropy in lifeas needs further experimental corroboration by measuring the other channels with better statistics. (color online) energy scan of the sf@xmath88 and sf@xmath89 intensities at @xmath8=(0.43 0.57 0) measured at the in20 spectrometer showing a local anisotropy of the magnetic fluctuation between 6 and 12 mev.,scaledwidth=42.0%] in order to reveal eventual weak ferromagnetic fluctuations a @xmath41-scan across @xmath8=(110) at @xmath90 k and @xmath91 mev has been carried out. all three sf channels revealed neutron counts similar to the sf background meaning that no significant magnetic scattering is present. due to the limitation of triple - axis spectrometers concerning the incident energy, high energy transfers have to be measured using a time - of - flight spectrometer. however, higher incident energies are at the cost of a loss in resolution. with the @xmath18 axis of the sample aligned along the incident beam one obtains a projection of @xmath7 along this axis after the measurement of a curved 3-dimensional hypersurface in the 4-dimensional manifold of reciprocal space. in the projection the @xmath12 component is an implicit variable which changes with energy tranfer, nevertheless being calculable, i.e. the obtained data is three - dimensional in @xmath92-space. in order to visualize the data the program mslice has been used which offers the possibility of averaging the data along a chosen axis to produce a slice or integrating along two axes to produce a cut. by measuring a standard vanadium sample with known mass the intensity can be normalized to an absolute scale in mb/(sr mev f.u.) by using the sample mass and molar mass. [fig : fluctuation](a) shows a slice of the @xmath93 plane which has been integrated between 10 mev and 30 mev for an incident beam energy of 55 mev. two peaks can be observed around the (0.5 0.5 0) position. by integrating the data perpendicular to the scan path indicated by the dashed line, one obtains the curve shown in fig. [fig : fluctuation](b) clearly revealing the incommensurability of the antiferromagnetic fluctuations. however, also the time - of - flight data indicates an asymmetric shape or an additional signal at larger incommensurability. two pairs of symmetrical gaussian functions on a constant background have been fitted to the data, from which the incommensurabilities @xmath94 and @xmath95 could be extracted. the value of @xmath60 is in good agreement with our previous results.@xcite note that the absolute intensity scale obtained by the renormalization to a vanadium standard (between 10 and 30 mev) is in very good agreement with our results shown in sec. [sec : absolute] (fig. [fig : absfluct], 5 mev) and also with a report on polycrystalline samples.@xcite (color online) (a) time - of - flight data showing @xmath7 in the @xmath93 plane integrated between 10 mev and 30 mev (@xmath96=55 mev, @xmath97=10 k. the incommensurate peaks are marked by a white ellipse. the black dashed indicates the cut along the [h -h 0] direction shown in (b). the fit of two pairs of symmetrical gaussian functions after the subtraction of a constant background (determined from the fit in the inset) yields the incommensurabilities @xmath94 and @xmath95.,scaledwidth=42.0%] in order to investigate the magnetic signal at higher energy transfers an incident neutron energy of 100 mev has been used. [fig : e100] shows @xmath93 slices of 10 mev thickness each. the magnetic fluctuation can be observed around the (0.5 0.5 0) point, but the loss in resolution becomes evident. however, there is a significant signal which can be separated from the background up to an energy transfer of 60 mev. for the slice in fig. [fig : e100](f) the signal is reduced to the background. due to the limited @xmath8-resolution in comparison to @xmath96=55 mev the incommensurability could not be investigated at higher energy transfers.
Conclusion
in summary we have extended our previous work concerning the characterization of the incommensurate antiferromagnetic fluctuations in lifeas. time - of - flight experiments show that the magnetic signal is observable up to energy transfers of 60 mev, while the incommensurability remains unchanged up to 30 mev (measurements of higher energy transfers were at the cost of resolution prohibiting a quantitative analysis of the incommensurability). longitudinal polarization analysis proved the magnetic origin of the observed signal and an eventual spin - space anisotropy between 8 and 10 mev could be deduced that resembles observation in other feas - based superconductors. the asymmetric shape of the incommensurate peak suggests the presence of two different signals which may correspond to scattering between the outer hole pocket and the inner electron pocket as well as between the outer hole pocket and the outer electron pocket. the different energy dependences of the peak intensities of @xmath70 and @xmath72 support the picture of two independent signals. furthermore, we have converted the intensity of the scattered neutrons into an absolute scale making it possible to compare the strength of the magnetic fluctuations in lifeas with those of related compounds. nearly optimally co - doped bafe@xmath1as@xmath1 yields a maximum value of roughly 7.5 mb/(sr mev fe) at the resonance feature (ref. ). by averaging the peak values of the triple - axis (fig. [fig : absfluct]) and time - of - flight data (fig. [fig : fluctuation]) we obtain 0.95 mb/(sr mev fe) rendering the low - temperature fluctuations in lifeas by a factor 8 weaker than the magnetic resonance in co - doped bafe@xmath1as@xmath1. this perfectly agrees with our earlier work,@xcite where we estimated the same ratio between the incommensurate fluctuations in lifeas and the commensurate resonance in ba(fe@xmath98co@xmath99)@xmath1as@xmath1 by normalizing the magnetic signal to the respective phonon signal. due to the incommensurability a factor of two is recovered for which the magnetic scattering per fe ion in lifeas is by roughly a factor four weaker than the respective scattering in nearly optimally co - doped bafe@xmath1as@xmath1 which must be reconciled with the fact that the superconducting transition temperature is only little reduced in lifeas. this work was supported by the deutsche forschungsgemeinschaft (dfg) through the priority programme spp1458 (grants no. be1749/13, bu887/15 - 1 and br2211/1 - 1). thanks the dfg for funding in the emmy noether programme (project 595/3 - 1). | we present an inelastic neutron scattering study on single - crystalline lifeas devoted to the characterization of the incommensurate antiferromagnetic fluctuations at @xmath0.
time - of - flight measurements show the presence of these magnetic fluctuations up to an energy transfer of 60 mev, while polarized neutrons in combination with longitudinal polarization analysis on a triple - axis spectrometer prove the pure magnetic origin of this signal.
the normalization of the magnetic scattering to an absolute scale yields that magnetic fluctuations in lifeas are by a factor eight weaker than the resonance signal in nearly optimally co - doped bafe@xmath1as@xmath1, although a factor two is recovered due to the split peaks owing to the incommensurability.
the longitudinal polarization analysis indicates weak spin space anisotropy with slightly stronger out - of - plane component between 6 and 12 mev.
furthermore, our data suggest a fine structure of the magnetic signal most likely arising from superposing nesting vectors. | 1407.0247 |
Introduction
interacting fermionic particles play a central role in the structure of matter and exist over a very broad range of energies, from extremely low temperature trapped atomic fermi gases, where @xmath10 k @xcite, to very high temperature primordial matter, like quark - gluon plasmas, where @xmath11 k @xcite. for all of these systems, the most intriguing physics is related to very strong interactions between fermionic particles, such as the strong coupling between electrons in high-@xmath8 superconductors and the strong interactions between neutrons in neutron matter. current many - body quantum theories face great challenges in solving problems for strongly interacting fermi systems, due to the lack of a small coupling parameter. for example, the critical temperature of a superfluid - normal fluid transition in a strongly interacting fermi gas has been controversial for many years. the critical temperature @xmath12 has been predicted to have values in the range between 0.15 and 0.35 by different theoretical methods @xcite. a complete understanding of the physics of strongly interacting systems can not yet be obtained from a theoretical point of view. there is a pressing need to investigate strongly interacting fermions experimentally. in recent years, based on progress in optical cooling and trapping of fermionic atoms, a clean and controllable strongly interacting fermi system, comprising a degenerate, strongly interacting fermi gas @xcite, is now of interest to the whole physics community. strongly interacting fermi gases are produced near a broad feshbach resonance @xcite, where the zero energy s - wave scattering length @xmath13 is large compared to the interparticle spacing, while the interparticle spacing is large compared to the range of the two - body interaction. in this regime, the system is known as a unitary fermi gas, where the properties are universal and independent of the details of the two - body scattering interaction @xcite. in contrast to other strongly interacting fermi systems, in atomic gases, the interactions, energy, and spin population can be precisely adjusted, enabling a variety of experiments for exploring this model system. intense studies of strongly interacting fermi gases have been implemented over the past several years from a variety of perspectives. some of the first experiments observed the expansion hydrodynamics of the strongly interacting cloud @xcite. evidence for superfluid hydrodynamics was first observed in collective modes @xcite. collective modes were later used to study the @xmath14 equation of state throughout the crossover regime @xcite. recently, measurements of sound velocity have also been used to explore the @xmath14 equation of state @xcite. below a feshbach resonance, fermionic atoms join to form stable molecules and molecular bose - einstein condensates @xcite. fermionic pair condensation has been observed by projection experiments using fast magnetic field sweeps @xcite. above resonance, strongly bound pairs have been probed by radio frequency and optical spectroscopy @xcite. phase separation has been observed in spin polarized samples @xcite. rotating fermi gases have revealed vortex lattices in the superfluid regime @xcite as well as irrotational flow in both the superfluid and normal fluid regimes @xcite. measurement of the thermodynamic properties of a strongly interacting fermi gas was first accomplished by adding a known energy to the gas, and then determining an empirical temperature that was calibrated using a pseudogap theory @xcite. recent model - independent measurements of the energy and entropy @xcite provide a very important piece of the puzzle, because they enable direct and precision tests that distinguish predictions from recent many - body theories, without invoking any specific theoretical model @xcite. one of the major challenges for the experiments in strongly interacting fermi gases is the lack of a precise model - independent thermometry. two widely - used thermometry methods are model - dependent, in that they rely on theoretical models for calibration. the first relies on adiabatic magnetic field sweeps between the molecular bec regime and the strongly interacting regime @xcite. subsequently, the temperature of the strongly interacting gas is estimated from the measured temperature in the bec regime using a theoretical model of the entropy @xcite. the second method, used by our group @xcite, is based on determining an empirical temperature from the cloud profiles that is calibrated by comparing the measured density distribution with a theoretical model for the density profiles. currently two model - independent thermometry methods have been reported for strongly - interacting gases. one is the technique employed by the mit group @xcite, which is only applicable to imbalanced mixtures of spin - up and spin - down atoms. that method is based on fitting the noninteracting edge for the majority spin after phase separation. another model - independent method is demonstrated in ref. @xcite, which is applicable to both balanced and imbalanced mixtures of spin - up and spin - down fermions. the energy @xmath1 and entropy @xmath2 are measured and then parameterized to determine a smooth curve @xmath4. then the temperature in both the superfluid and normal fluid regime is obtained from the fundamental thermodynamic relation @xmath5. in this paper, we will describe our model - independent thermodynamic experiments on a strongly interacting fermi gas of @xmath0li, which we have conducted at duke university. first, we will describe our measurements of both the total energy @xmath1 and the total entropy @xmath2 of a trapped strongly - interacting fermi gas tuned near a feshbach resonance. then, we determine the temperature @xmath5 after showing that the @xmath4 data are very well parameterized by using two different power laws that are joined with continuous @xmath1 and @xmath6 at a certain entropy @xmath7 that gives the best fit. to examine the sensitivity of the temperature to the form of the fit function, we employ two different fit functions that allow for a heat capacity jump or for a continuous heat capacity at @xmath7. we find that the @xmath6 values closely agree for both cases. we find a significant change in the scaling of @xmath1 with @xmath2 above and below @xmath7, in contrast to the behavior for an ideal fermi gas, where a single power - law well parameterizes @xmath4 over the same energy range. by interpreting @xmath7 as the critical entropy for a superfluid - normal fluid transition in the strongly interacting fermi gas, we estimate the critical energy @xmath15 and critical temperature @xmath8. both the model - independent @xmath4 data and the estimated critical parameters are compared with several recent many - body theories based on both analytic and quantum monte carlo methods. we also show how parameterizing the @xmath4 data provides experimental temperature calibrations, which helps to unify, in a model - independent way, the results obtained by several groups @xcite. first we relate the endpoint temperatures for adiabatic sweeps of the bias magnetic field between the strongly interacting and ideal noninteracting regimes, as used in the jila experiments to characterize the condensed pair fraction @xcite. this enables the ideal gas temperature observed for the onset of pair condensation @xcite to be related to the critical temperature of the strongly interacting fermi gas. the temperature obtained by parameterizing the strongly interacting gas data also calibrates the empirical temperature based on the cloud profiles, as used in our previous studies of the heat capacity @xcite. these temperature calibrations yield values of @xmath8 close to that estimated from our @xmath4 data. next, we discuss three different methods for determining the universal many - body parameter, @xmath16 @xcite, where @xmath17 is the energy per particle in a uniform strongly interacting fermi gas at @xmath14 in units of the energy per particle of an ideal fermi gas at the same density. first, we describe the measurement of the sound velocity at resonance and its relationship to @xmath16. then, we determine @xmath16 from the ground state energy @xmath18 of the trapped gas. here, @xmath18 is obtained by extrapolating the @xmath4 data to @xmath9, as suggested by hu et al. this avoids a systematic error in the sound velocity experiments arising from the unknown finite temperature. finally, to explore the systematic error arising from the measurement of the number of atoms, @xmath16 is determined in a number - independent manner from the ratio of the cloud sizes in the strongly and weakly interacting regimes. all three results are found to be in very good agreement with each other and with recent predictions. finally, we obtain three universal thermodynamic functions from the parameterized @xmath4 data, the energy @xmath19, heat capacity @xmath20, and global chemical potential @xmath21.
Experimental methods
our experiments begin with an optically - trapped highly degenerate, strongly interacting fermi gas of @xmath0li @xcite. a 50:50 mixture of the two lowest hyperfine states of @xmath0li atoms is confined in an ultrastable co@xmath22 laser trap with a bias magnetic field of 840 g, just above a broad feshbach resonance at @xmath23 g @xcite. at 840 g, the gas is cooled close to the ground state by lowering the trap depth @xmath24 @xcite. then @xmath24 is recompressed to a final trap depth of @xmath25k, which is much larger than the energy per particle of the gas, for the highest energies employed in the experiments. this suppresses evaporation during the time scale of the measurements. the shallow trap yields a low density that suppresses three body loss and heating. the low density also yields a weakly interacting sample when the bias magnetic field is swept to 1200 g, although the scattering length is @xmath26 bohr, as discussed in detail in [sec : entropy]. the shape of the trapping potential is that of a gaussian laser beam, with a transverse gaussian profile determined by the spot size and an axial lorentzian profile determined by the rayleigh length. to simplify the calculations of the ideal gas properties in subsequent sections, as well as the theoretical modelling, we take the trap potential to be approximated by a three dimensional gaussian profile, @xmath27 where @xmath28 is the @xmath29 width of trap for each direction. here, we take the zero of energy to be at @xmath30. when the cold atoms stay in the deepest portion of the optical trap, where @xmath31, the gaussian potential can be well approximated as a harmonic trap with transverse frequencies @xmath32, @xmath33 and axial frequency @xmath34, where @xmath35 here @xmath36 is the @xmath0li mass. at our final trap depth @xmath37, the measured transverse frequencies are @xmath38 hz and @xmath39 hz. the axial frequency is weakly magnetic field dependent since the total axial frequency has both an optical potential contribution @xmath40 determined by eq. [eq : trapfreqwelldepth] and a magnetic potential contribution arising from magnetic field curvature, @xmath41. the net axial frequency is then @xmath42. we find @xmath43 hz at 840 g and @xmath44 hz at 1200 g. the total number of atoms is @xmath45. the corresponding fermi energy @xmath46 and fermi temperature @xmath47 at the trap center for an ideal noninteracting harmonically trapped gas are @xmath48, where @xmath49. for our trap conditions, we obtain @xmath50k. using @xmath51, we can rewrite eq. [eq : udipole] as a symmetric effective potential, @xmath52 where @xmath53 is the scaled position vector. here, @xmath54 with @xmath55. to obtain the anharmonic corrections for the gaussian trap, we expand eq. [eq : reducedtruepotential] in a taylor series up to second order in @xmath56, @xmath57 model - independent energy measurement is based on a virial theorem, which for an ideal gas in a harmonic confining potential @xmath58 yields @xmath59. since the harmonic potential energy is proportional to the mean square cloud size, measurement of the cloud profile determines the total energy. remarkably, a trapped unitary fermi gas at a broad feshbach resonance obeys the same virial theorem as an ideal gas, although it contains superfluid pairs, noncondensed pairs, and unpaired atoms, all strongly interacting. this has been demonstrated both theoretically and experimentally @xcite. the virial theorem shows that the total energy of the gas at all temperatures can be measured from the cloud profile using @xmath60 where @xmath24 is the trapping potential and @xmath61 is the position vector.. [eq : energymeas] can be shown to be valid for any trapping potential @xmath24 and for any spin mixture, without assuming either the local density approximation or harmonic confinement @xcite. using eq. [eq : truepotentialquartic] in eq. [eq : energymeas] and keeping the lowest order anharmonic corrections, we obtain the energy per atom in terms of the axial mean square size, @xmath62. \label{eq : energygaussian}\]] here, we have used the local density approximation with a scalar pressure, which ensures that @xmath63. for the ground state, where the spatial profile is a zero temperature thomas - fermi profile, we have @xmath64. for energies @xmath65, where the spatial profile is approximately gaussian, we have @xmath66. since the anharmonic correction is small at low temperatures where the cloud size is small, we use the gaussian approximation over the whole range of energies explored in our experiments. for the conditions of our experiments, there is no evidence that the local density approximation breaks down for a 50:50 spin mixture. in this case, measurement of the mean square size in any one direction determines the total energy. from eq. [eq : energygaussian], we see that by simply measuring the axial mean square size @xmath67 at 840 g and measuring the axial trap frequency by parametric resonance, we actually measure @xmath68, the total energy per particle of the strongly interacting fermi gas at 840 g. this determines the total energy per particle in a model - independent way @xcite. the entropy @xmath2 of the strongly interacting gas at 840 g is determined by adiabatically sweeping the bias magnetic field from 840 g to 1200 g, where the gas is weakly interacting @xcite. the entropy @xmath69 of the weakly interacting gas is essentially the entropy of an ideal fermi gas in a harmonic trap, which can be calculated in terms of the mean square axial cloud size @xmath70 measured after the sweep. since the sweep is adiabatic, we have @xmath71 the adiabaticity of the magnetic field sweep is verified by employing a round - trip - sweep : the mean square size of the cloud at 840 g after a round - trip - sweep lasting 2s is found to be within 3% of mean square size of a cloud that remains at 840 g for a hold time of 2s. the nearly unchanged atom number and mean square size proves the sweep does not cause any significant atom loss or heating, which ensures entropy conservation for the sweep. the background heating rate is the same with and without the sweep and increases the mean square size by about 2% over 2s. the mean square size data are corrected by subtracting the increase arising from background heating over the 1 s sweep time @xcite. at 1200 g in our shallow trap, we have @xmath72, where the fermi wavevector @xmath73 and the s - wave scattering length @xmath74 bohr @xcite. we find that the gas is weakly interacting : for the lowest temperatures attained in our experiments, the gas at 1200 g is a normal fluid that we observe to expand ballistically. we have calculated the ground state mean square size at 1200 g in our gaussian trap, based on a mean - field theory, @xmath75 @xcite, which is close to that of an ideal harmonically trapped gas, @xmath76. here, @xmath77 is the mean square size corresponding to the fermi energy of an ideal noninteracting fermi gas at magnetic field @xmath78, which includes the magnetic field dependence of the axial trapping frequency : @xmath79. we expect that the entropy of the gas at 1200 g is close to that of an ideal gas, except for a mean field shift of the energy. we therefore assume that a reasonable approximation to the entropy is that of an ideal fermi gas, @xmath80, where @xmath81 is the ground state mean square size of an ideal fermi gas in the gaussian trapping potential of eq. [eq : reducedtruepotential]. here, we apply an elementary calculation based on integrating the density of states for the gaussian trap with the entropy per orbital @xmath82 $], where @xmath83 is the ideal fermi gas occupation number at temperature @xmath6 for an orbital of energy @xmath84. by calculating @xmath85 as a function of the _ difference _ between the finite temperature and ground state mean square cloud sizes, we reduce the error arising from the mean field shift at 1200 g, and ensure that @xmath86 for the ground state. the exact entropy of a weakly interacting gas @xmath69 at 1200 g, @xmath87, has been calculated using many - body theories @xcite for the gaussian potential of eq. [eq : reducedtruepotential]. in the experiments, we determine the value of @xmath88, where we take @xmath89, the value measured at our lowest energy at 1200 g by extrapolation to @xmath14 using the sommerfeld expansion for the spatial profile of an ideal gas. this result is close to the theoretical value, @xmath90. the entropy versus cloud size curve for an ideal noninteracting fermi gas and the exact value for a weakly interacting gas @xmath69 at 1200 g are plotted in fig. [fig : entropysizeoverlap]. we find that the entropies @xmath91 and @xmath92), agree within a few percent over most of the energy range we studied, except at the point of lowest measured energy, where they differ by 10%. the results show clearly that the shape of the entropy curve of a weakly interacting fermi gas is nearly identical to that of an ideal gas when the mean field shift of the ground state size is included by referring the mean square cloud size to that of the ground state. so we have to a good approximation, @xmath93 since the corrections to ideal gas behavior are small, the determination of @xmath94 by measuring the axial mean square size @xmath70 relative to the ground state provides an essentially model - independent estimate of the entropy of the strongly interacting gas. sound velocity measurements have been implemented for fermi gases that are nearly in the ground state, from the molecular bec regime to the weakly interacting fermi gas regime @xcite. a sound wave is excited in the sample by using a thin slice of green light that bisects the cigar - shaped cloud. the green light at 532 nm is blue detuned from the 671 nm transition in lithium, creating a knife that locally repels the atoms. the laser knife is pulsed on for 280 @xmath95s, much shorter than typical sound propagation times @xmath96 ms and excites a ripple in the density consisting of low density valleys and high density peaks. after excitation, the density ripple propagates outward along the axial direction @xmath97. after a variable amount of propagation time, we release the cloud, let it expand, and image destructively. in the strongly interacting regime, we use zero - temperature thomas - fermi profiles for a non - interacting fermi gas to fit the density profiles, and locate the positions of the density valley and peak. by recording the position of the density ripple versus the propagation time, the sound velocity is determined. a detailed discussion of potential sources of systematic error is given by joseph et al. @xcite. for a strongly interacting fermi gas in the unitary limit, the sound velocity @xmath98 at the trap center for the ground state is determined by the fermi velocity of an ideal gas at the trap center, @xmath99 and the universal constant @xmath16, @xmath100 a precision measurement of the sound speed therefore enables a determination of @xmath16 @xcite. as discussed below, the values of @xmath16 determined from the @xmath4 data and the sound velocity data are in very good agreement.
Comparison of thermodynamic data with theory
in the experiments, the raw data consists of the measured mean square cloud sizes at 840 g and after an adiabatic sweep of the magnetic field to 1200 g. using this data, we determine both the energy and the entropy of the strongly interacting gas. the data is then compared to several recent predictions. we begin by determining the axial mean square cloud sizes at 840 g and after the adiabatic sweep to 1200 g. since the atom number can vary between different runs by up to 20%, it is important to make the comparison independent of the atom number and trap parameters. for this purpose, the mean square sizes are given in units of @xmath77, as defined above. the measured mean square sizes are listed in table [tbl : energyentropy]. in the experiments, evaporative cooling is used to produce an atom cloud near the ground state. energy is controllably added by releasing the cloud and then recapturing it after a short time @xmath101 as described previously @xcite. for a series of different values of @xmath101, the energy at 840 g is directly measured from the axial cloud size according to eq. [eq : energygaussian]. then the same sequence is repeated, but the cloud size is measured after an adiabatic sweep to 1200 g. in each case, the systematic increase in mean square size arising from background heating rate is determined and subtracted. the total data comprise about 900 individual measurements of the cloud size at 840 g and 900 similar measurements of the cloud size after a sweep to 1200 g. to estimate the measurement error, we split the energy scale at 840 g into bins with a width of @xmath102. measured data points within the width of the energy bin are used to calculate the average measured values of the cloud sizes and the corresponding standard deviation at both 840 g and 1200 g. ''' '' & @xmath103&@xmath104&@xmath105&@xmath106&@xmath107&@xmath108 + ''' '' 1&0.568(4)&0.743(6)&0.548(4)&0.63(8)&0.91(23)&0.97(5) + ''' '' 2&0.612(5)&0.776(13)&0.589(5)&0.99(11)&1.18(22)&1.24(9) + ''' '' 3&0.661(5)&0.803(11)&0.634(5)&1.22(8)&1.36(20)&1.42(7) + ''' '' 4&0.697(9)&0.814(15)&0.667(8)&1.30(10)&1.43(18)&1.49(8) + ''' '' 5&0.74(1)&0.87(4)&0.71(1)&1.6(2)&1.72(18)&1.8(2) + ''' '' 6&0.79(1)&0.89(2)&0.75(1)&1.7(1)&1.79(15)&1.9(1) + ''' '' 7&0.83(1)&0.94(2)&0.79(1)&2.0(1)&2.03(16)&2.1(1) + ''' '' 8&0.89(2)&1.02(2)&0.84(2)&2.3(1)&2.32(18)&2.4(1) + ''' '' 9&0.91(1)&1.02(3)&0.86(1)&2.3(1)&2.31(16)&2.4(1) + ''' '' 10&0.97(1)&1.10(1)&0.91(1)&2.55(4)&2.57(17)&2.64(4) + ''' '' 11&1.01(1)&1.17(2)&0.94(1)&2.74(7)&2.75(19)&2.82(7) + ''' '' 12&1.05(1)&1.18(1)&0.98(1)&2.78(4)&2.80(17)&2.87(4) + ''' '' 13&1.10(1)&1.22(1)&1.03(1)&2.89(2)&2.90(15)&2.97(2) + ''' '' 14&1.25(2)&1.35(5)&1.15(1)&3.21(12)&3.20(14)&3.27(12) + ''' '' 15&1.28(1)&1.39(3)&1.18(1)&3.28(6)&3.28(14)&3.35(6) + ''' '' 16&1.44(2)&1.49(2)&1.31(2)&3.49(4)&3.48(9)&3.55(4) + ''' '' 17&1.53(2)&1.62(6)&1.39(2)&3.74(10)&3.73(11)&3.80(10) + ''' '' 18&1.58(1)&1.63(1)&1.42(1)&3.76(2)&3.74(8)&3.81(2) + ''' '' 19&1.70(2)&1.73(6)&1.52(1)&3.94(9)&3.92(7)&3.99(9) + ''' '' 20&1.83(5)&1.79(2)&1.62(4)&4.03(3)&4.01(1)&4.08(3) + ''' '' 21&1.93(3)&1.96(3)&1.70(2)&4.28(4)&4.26(6)&4.32(4) + ''' '' 22&2.11(5)&2.17(3)&1.83(4)&4.55(3)&4.53(7)&4.59(3) + the ratio of the mean square axial cloud size at 1200 g (measured after the sweep) to that at 840 g (measured prior to the sweep) is plotted in fig. [fig : sizeratio] as a function of the energy of a strongly interacting gas at 840 g. the ratio is @xmath109, since for an adiabatic sweep of the magnetic field from the strongly interacting regime to the weakly interacting regime, the total entropy in the system is conserved but the energy increases : the strongly interacting gas is more attractive than the weakly interacting gas. a similar method was used to measure the potential energy change in a fermi gas of @xmath110k, where the bias magnetic field was adiabatically swept between the strongly interacting regime at the feshbach resonance and a noninteracting regime above resonance @xcite. the resulting potential energy ratios are given as a function of the temperature of the noninteracting gas @xcite. in contrast, by exploiting the virial theorem which holds for the unitary gas, we determine both the energy and entropy of the strongly interacting gas, as described below. [fig : sizeconvertentropy] shows the entropy which is obtained from the mean square size at 1200 g @xmath111 as listed in table [tbl : energyentropy]. first, we find the mean square size relative to that of the ground state, @xmath112. we use the measured value @xmath113 for the lowest energy state that we obtained at 1200 g, as determined by extrapolation to @xmath14 using a sommerfeld expansion for the spatial profile of an ideal fermi gas. then we determine the entropy in the noninteracting ideal fermi gas approximation : @xmath114 $], where we have replaced @xmath81 by the ground state value at 1200 g. as discussed above, this method automatically ensures that @xmath9 corresponds to the measured ground state @xmath115 at 1200 g, and compensates for the mean field shift between the measured @xmath116 for a weakly interacting fermi gas and that calculated @xmath117 for an ideal fermi gas in our gaussian trapping potential. as shown in fig. [fig : entropysizeoverlap], the entropy obtained from a more precise many - body calculations is in close agreement with the ideal gas entropy calculated in the ideal gas approximation. the energy is determined from the cloud profiles at 840 g using eq. [eq : energygaussian]. finally, we generate the energy - entropy curve for a strongly interacting fermi gas, as shown in fig. [fig : energyentropy]. here, the energy @xmath1 measured from the mean square axial cloud size at 840 g is plotted as a function of the entropy @xmath2 measured at 1200 g after an adiabatic sweep of the magnetic field. we note that above @xmath118 (@xmath119) the @xmath4 data (blue dots) for the strongly interacting gas appear to merge smoothly to the ideal gas curve (dashed green). for an ideal fermi gas. for this figure, the ideal gas approximation to the entropy is used, @xmath106 of table [tbl : energyentropy]. [fig : energyentropy],width=384] in addition to the entropy calculated in the ideal gas approximation, table [tbl : energyentropy] also provides a more precise entropy @xmath120 versus the axial mean square cloud size. these results are obtained by hu et al. @xcite using a many - body calculation for @xmath121 at 1200 g in the gaussian trap of eq. [eq : reducedtruepotential]. perhaps the most important application of the energy - entropy measurements is to test strong coupling many - body theories and simulations. since the energy and entropy are obtained in absolute units without invoking any specific theoretical model, the data can be used to distinguish recent predictions for a trapped strongly interacting fermi gas. fig. [fig : comparetheory] shows how four different predictions compare to the measured energy and entropy data. these include a pseudogap theory @xcite, a combined luttinger - ward - de dominicis - martin (lw - ddm) variational formalism @xcite, a t - matrix calculation using a modified nozires and schmitt - rink (nsr) approximation @xcite, and a quantum monte carlo simulation @xcite. the most significant deviations appear to occur near the ground state, where the precise determination of the energy seems most difficult. the pseudogap theory predicts a ground state energy that is above the measured value while the prediction of ref. @xcite is somewhat low compared to the measurement. all of the different theories appear to converge at the higher energies.
Determining the temperature and critical parameters
the temperature @xmath6 is determined from the measured @xmath4 data using the fundamental relation, @xmath5. to implement this method, we need to parameterize the data to obtain a smooth differentiable curve. at low temperatures, one expects the energy to increase from the ground state according to a power law in @xmath6 and a corresponding power law in @xmath2, i.e., @xmath122. for a harmonically trapped ideal fermi gas, we have in the sommerfeld approximation an energy per particle in units of @xmath46 given by @xmath123. the corresponding entropy per particle in units of @xmath124 is @xmath125, so that @xmath126. we attempt to use a single power law to fit the @xmath127 curve for a noninteracting fermi gas in a gaussian trapping potential, with @xmath128, as in our experiments. the energy and entropy are calculated in the energy range @xmath129 and displayed as dots in fig. [fig : idealgasfit]. we find that a single power law @xmath130 fits the curve very well over this energy range. note that the power law exponent is @xmath131, close to the low temperature value. using the fit function, we can extract the reduced temperature @xmath132 as a function of @xmath85 and compare it to the theoretical reduced temperature @xmath133 at the same @xmath85. the results are shown as the green dashed line in fig. [fig : idealgasfit]. we see that the agreement is quite good except below @xmath134 and above @xmath135, where the deviation is @xmath136%. to improve the fit and to make a more precise determination of the temperature, we employ a fit function comprising two power laws that are joined at a certain entropy @xmath7, which gives the best fit. when used to fit the data for the strongly interacting fermi gas, we consider two types of fits that incorporate either a jump in heat capacity or a continuous heat capacity at @xmath7. in this way, we are able determine the sensitivity of the temperature and critical parameters to the form of the fit function. the two types of fits yield nearly identical temperatures, but different values of @xmath7 and hence of the critical parameters, as discussed below. we take the energy per particle @xmath1 in units of @xmath46 to be given in terms of the entropy per particle in units of @xmath124 in the form @xmath137 we constrain the values of @xmath138 and @xmath3 by demanding that energy and temperature be continuous at the joining point @xmath7 : @xmath139 by construction, the value of @xmath140 does not affect these constraints and is chosen in one of two ways. fixing @xmath141, the fit incorporates a heat capacity jump at @xmath7, which arises from the change in the power law exponents at @xmath7. alternatively, we choose @xmath140 so that the second derivative @xmath142 is continuous at @xmath7, making the heat capacity continuous. the final fit function has 5 independent parameters @xmath143, and takes the form @xmath144+e(s - s_c)^2 ; \,\,s\geq s_c. \label{eq : evsslowengtwopower}\end{aligned}\]] here, when @xmath140 is not constrained to be zero, it is given by @xmath145 fig. [fig : idealgasfit] shows the improved fit to the calculated energy versus entropy of a noninteracting fermi gas in a gaussian trap for @xmath128, using eq. [eq : evsslowengtwopower] with @xmath146, since the ideal gas has no heat capacity jump. in this case, both power law exponents @xmath147 and @xmath148 are close to @xmath149 as for the single power law fit. the temperature determined from the fit agrees very closely with the exact temperature, as shown in fig. [fig : idealgasfit] (red solid line). in contrast to the noninteracting case, we have found that the energy - entropy data of a strongly interacting fermi gas is not well fit by a single power law function @xcite. however, the two power - law function fits quite well, with a factor of two smaller value of @xmath150 than for the single power - law fit. here, we use @xmath151, where @xmath152 (@xmath153) is the fitted (data) value for the @xmath154 point, and @xmath155 is the corresponding the standard error. motivated by the good fits of the two power - law function to the ideal gas energy versus entropy curve and the good agreement between the fitted and exact temperature, we apply the two power - law fit function to the data for the strongly interacting fermi gas. ). for comparison, the dot - dashed green curve shows @xmath4 for an ideal fermi gas. for this figure, the ideal gas approximation to the entropy is used, @xmath106 of table [tbl : energyentropy]. [fig : energyentropyfit],width=384] fig. [fig : energyentropyfit] shows the fit (red solid curve) obtained with a heat capacity jump using eq. [eq : evsslowengtwopower] with @xmath141 and @xmath156, @xmath157, @xmath158, the ground state energy @xmath159, and the critical entropy @xmath160. also shown is the fit (blue dashed curve) with continuous heat capacity (@xmath146) and @xmath161, @xmath162, @xmath163, the ground state energy @xmath164, and the critical entropy @xmath165. the fit functions for the @xmath4 data for the strongly interacting fermi gas exhibit a significant change in the scaling of @xmath1 with @xmath2 below and above @xmath7. the dramatic change in the power law exponents for the strongly interacting gas suggests a transition in the thermodynamic properties. the power law exponent is @xmath166 above @xmath7, comparable to that obtained for the ideal gas, where @xmath167. the power law exponent below @xmath7 is @xmath168, which corresponds to the low temperature dependence @xmath169, close to that obtained in measurements of the heat capacity, where the observed power law was @xmath170 after the model - dependent calibration of the empirical temperature @xcite, see [sec : ttwiddle]. if we interpret @xmath7 as the critical entropy for a superfluid - normal fluid transition in the strongly interacting fermi gas, then we can estimate the critical energy @xmath15 and the critical temperature @xmath171. for the fits of eq. [eq : evsslowengtwopower] with a heat capacity jump @xmath172) or with continuous heat capacity (@xmath146), we obtain @xmath173 using the fit parameters in eq. [eq : criticalparam] yields critical parameters of the strongly interacting fermi gas, which are summarized in table [tbl : criticalparam]. the statistical error estimates are from the fit, and do not include systematic errors arising from the form of the fit function. ''' '' & @xmath174 & @xmath175 & @xmath176 + ''' '' expt @xmath4 fit@xmath177 & 2.2(1) & 0.83(2) & 0.21(1) + ''' '' expt @xmath4 fit@xmath178 & 1.6(3) & 0.70(5) & 0.185(15) + ''' '' heat capacity experiment@xmath179 & & 0.85 & 0.20 + ''' '' theory ref. @xcite & & & 0.30 + ''' '' theory ref. @xcite & & & 0.31 + ''' '' theory ref. @xcite & & & 0.27 + ''' '' theory ref. @xcite & & & 0.29 + ''' '' theory ref. @xcite & 2.15 & 0.82 & 0.27 + ''' '' theory ref. @xcite & 1.61(5) & 0.667(10) & 0.214(7) + we note that the fit function for @xmath180 previously used in ref. @xcite to determine the temperature was continuous in @xmath2 and @xmath1, but intentionally ignored the continuous temperature constraint in order to determine the entropy as a power of @xmath181 both above and below the joining energy @xmath15. as the continuous temperature constraint is a physical requirement, we consider the present estimate of the temperature @xmath6 to be more useful for temperature calibrations and for characterizing the physical properties of the gas than the estimate of ref. @xcite. in contrast to the temperature @xmath6, the estimate of @xmath8 depends on the value of the joining entropy @xmath7 that optimizes the fit and is more sensitive to the form of fit function than the temperature that is determined from the @xmath1 and @xmath2 data. for the fit function @xmath180 used in ref. @xcite, the temperatures determined by the fit function just above @xmath15, @xmath182, and below @xmath15, @xmath183, were different. an average of the slopes @xmath184 and @xmath185 was used to estimate the critical temperature. from those fits, the critical energy was found to be @xmath186, the critical entropy per particle was @xmath187. the estimated critical temperature obtained from the average was @xmath188, significantly higher than than the value @xmath189 obtained using eq. [eq : evsslowengtwopower], which incorporates continuous temperature. we are able to substantiate the critical temperature @xmath189 by using our data to experimentally calibrate the temperature scales in two other experiments. in [sec : onsetpaircond], we find that this value is in very good agreement with the estimate we obtain by calibrating the ideal gas temperature observed for the onset of pair condensation. nearly the same transition temperature is obtained in [sec : ttwiddle] by using the @xmath4 data to calibrate the empirical transition temperature measured in heat capacity experiments @xcite. table [tbl : criticalparam] compares the critical parameters estimated from the power - law fits to the @xmath4 data with the predictions for a trapped unitary fermi gas from several theoretical groups. we note that calculations for a uniform strongly interacting fermi gas at unitarity @xcite yield a lower critical temperature, @xmath190, than that of the trapped gas, where @xmath191 is the fermi temperature corresponding to the uniform density @xmath192. extrapolation of the uniform gas critical temperature to that of the trapped gas shows that the results are consistent @xcite. using the parameters from the fits and eq. [eq : evsslowengtwopower], the temperature of the strongly interacting fermi gas, in units of @xmath47 can be determined as a function of the entropy per particle, in units of @xmath124, @xmath193 here @xmath7 is given in table [tbl : criticalparam] from the fits to the @xmath4 data for the strongly interacting gas, eq. [eq : criticalparam] gives @xmath8. [fig : tempvsentropy] shows the temperature as a function of entropy according to eq. [eq : temperaturestrongint] for fits with a heat capacity jump and for continuous heat capacity.
Temperature calibrations
the estimates of the temperature of the strongly interacting fermi gas as a function of the entropy can be used to experimentally calibrate the temperatures measured in other experiments, without invoking any specific theoretical models. the jila group measures the pair condensate fraction in a strongly interacting fermi gas of @xmath110k as a function of the initial temperature @xmath194 in the noninteracting regime above the feshbach resonance @xcite. in these experiments, a downward adiabatic sweep of the bias magnetic field to resonance produces a strongly interacting sample. using our @xmath4 data, we relate the endpoint temperatures for adiabatic sweeps of the bias magnetic field between the ideal and strongly interacting fermi gas regimes. we therefore obtain the critical temperature for the onset of pair condensation in the strongly interacting fermi gas, and find very good agreement with our estimates based on entropy - energy measurement. in addition, we calibrate the empirical temperature based on the cloud profiles, which was employed in our previous measurements of the heat capacity @xcite. we relate the endpoint temperatures for an adiabatic sweep between the strongly interacting and ideal fermi gas regimes. [eq : temperaturestrongint] gives the temperature of the strongly interacting gas as a function of entropy, i.e., @xmath195. next, we calculate the entropy per particle @xmath196 for an ideal fermi gas in our gaussian trap, in units of @xmath124, with @xmath197 in units of @xmath47, as used in [sec : idealgas] to determine @xmath198. for an adiabatic sweep between the strongly interacting and ideal fermi gas regimes, where @xmath199, the temperature of the strongly interacting gas is related to that for the ideal fermi gas by @xmath200, \label{eq : adiabactictempcal}\]] which is shown in fig. [fig : tempsivni]. + for an adiabatic sweep from the ideal fermi gas regime to the strongly interacting fermi gas regime at low temperature @xmath201, the reduced temperature of the strongly interacting gas is greater than or equal to that of the ideal gas. this arises because the entropy of the strongly interacting gas scales as a higher power of the temperature than that of the ideal gas. in our present experiments, we could not take data at high enough temperatures to properly characterize the approach of the temperature to the ideal gas regime. above @xmath8, our @xmath4 data are obtained over a limited range of energies @xmath202 to avoid evaporation in our shallow trap. in this energy range, our data are reasonably well fit by a single power law. however, such a power law fit can not completely describe the higher temperature regime. we expect that the temperatures of the strongly interacting gas and ideal gas must start to merge in the region @xmath203, where the @xmath4 data for the strongly interacting gas nearly overlaps with the @xmath4 curve for an ideal gas, as shown in fig. [fig : energyentropy]. from fig. [fig : tempvsentropy], @xmath204 corresponds to @xmath205, approximately the place where the calibrations from the two different power law fits (for @xmath141 and @xmath146) begin to differ in fig. [fig : tempsivni]. we therefore expect that the single power law fit overestimates the temperature @xmath6 of the strongly interacting gas for @xmath206, yielding a trend away from ideal gas temperature, in contrast to the expected merging at high temperature. in ref. @xcite, projection experiments measure the ideal fermi gas temperature @xmath194 where pair condensation first appears. in those experiments, @xmath194 is estimated to be @xmath207 @xcite. from the calibration, [fig : tempsivni], we see that for @xmath208, the corresponding temperature of the strongly interacting gas is @xmath209 for both the red solid and blue dashed curves, which is almost the same as the ideal gas value. the critical temperature of the strongly interacting gas for the onset of pair condensation is then @xmath210, in very good agreement with the values @xmath211 and @xmath212 that we obtain from the two fits to the @xmath4 measurements. this substantiates the conjecture that the change in the power law behavior observed at @xmath8 in our experiments corresponds to the superfluid transition. in our previous study of the heat capacity, we determined an empirical temperature @xmath213 as a function of the total energy of the gas @xcite. the gas was initially cooled close to the ground state and a known energy was added by a release and recapture method. then a thomas - fermi profile for an ideal fermi gas was fit to the low temperature cloud profiles to determine the fermi radius. holding the fermi radius constant, the best fit to the cloud profiles at higher temperatures determined the effective reduced temperature, which is denoted @xmath214. the @xmath215 data @xcite was observed to scale as @xmath216 for @xmath217, while below @xmath218, the energy was found to scale as @xmath219. the transition point occurs at an energy @xmath220, which is close to the value @xmath221 obtained from power - law fit to the @xmath4 data for the fit with a heat capacity jump. assuming that @xmath222 corresponds to the superfluid - normal fluid transition, we can determine the corresponding value of @xmath12 for the strongly interacting gas. to calibrate the empirical temperature we start with @xmath215. then, as discussed in [sec : energyvstemp], eq. [eq : evsslowengtwopower] determines @xmath19 and hence @xmath223from the fits to our @xmath4 data. hence @xmath224 $], where @xmath225 is the reduced temperature of the strongly interacting gas and @xmath226 is the reduced energy. for simplicity, we give the analytic results obtained using the @xmath141 fit to the @xmath4 data, @xmath227 fig. [fig : tempsivni] shows the full calibration (green dashed curve). for comparison, the calibration obtained from the pseudogap theory of the cloud profiles gave @xmath228 for @xmath229, and @xmath230 above @xmath218. for @xmath222, we obtain from eq. [eq : tildetcalib] @xmath231 (see fig. [fig : tempsivni]), in good agreement with the value obtained for the onset of pair condensation and with the values @xmath232 and @xmath233 determined from the fits to the @xmath4 data.
Measuring the ground state energy
measurement of the ground state energy of a unitary fermi gas provides a stringent test of competing many - body theoretical predictions and is therefore of great interest. for a unitary fermi gas of uniform density in a 50 - 50 mixture of two spin states, the ground state energy per particle can be written as @xmath234 where @xmath235 is the local fermi energy corresponding to the density @xmath192. the ground state energy of the unitary fermi gas differs by a universal factor @xmath236 from that of an ideal fermi gas at the same density. the precise value of @xmath237 has been of particular interest in the context of neutron matter @xcite, and can be measured in unitary fermi gas experiments @xcite. the sound speed at temperatures near the ground state determines @xmath16 according to eq. [eq : soundres]. we have made precision measurements of the sound speed in a trapped fermi gas at the feshbach resonance @xcite. at 834 g, we vary the density by a factor of 30 to demonstrate universal scaling and obtain the value @xmath238. using eq. [eq : soundres] then yields @xmath239. note that the reference fermi velocity @xmath240 depends on the fermi energy of an ideal gas at the trap center and hence on both the trap frequencies and atom number (as @xmath241), which are carefully measured to minimize systematic errors @xcite. while the energy of the gas as measured from the mean square cloud size was close to the ground state value, the precise temperature of the gas was not determined. the universal parameter @xmath16 also can be determined by measuring the ground state energy @xmath18 of a harmonically trapped unitary fermi gas, which is given by @xmath242 our @xmath4 data enables a new determination of @xmath18 by extrapolating the measured energy @xmath4 to @xmath9. as pointed out by hui et al. @xcite, this method avoids a systematic error arising when the finite temperature is not determined in the measurements. from both of our fit functions below @xmath7, we obtain @xmath243. [eq : hobeta] yields @xmath244. this result is slightly more negative than that obtained in the sound speed experiments, which is reasonable since the sound speed measurements are done at finite temperature. both results are in very good agreement. one possible systematic error in these measurements arises from the determination of the atom number. the measurements of @xmath16 from the sound speed and from the energy - entropy measurements were done in different laboratories. the close agreement is gratifying, considering that the imaging systems that determine the atom number employed @xmath245-polarized light for the sound speed experiments, while the entropy - energy measurements used @xmath246-polarized light, for which the resonant optical cross section is a factor of two smaller than for @xmath245 polarization. to examine the systematic error arising from the atom number determination, we employ a third method to measure @xmath16 based on the measured ratio of the cloud size at 840 g and at 1200 g, which is number independent. the ratio of the ground state mean square sizes for the weakly and strongly interacting gases is predicted to be @xmath247 note that we obtain @xmath248 from a mean field calculation @xcite, in agreement with that obtained using a many - body calculation @xcite. our measurements for the ground state mean square size at 1200 g are accomplished by fitting a sommerfeld expansion of the axial density for an ideal fermi gas to the cloud profile @xcite. the fit determines the fermi radius @xmath249 and reduced temperature @xmath133, yielding @xmath250 for @xmath14, close to the predicted value of @xmath251. the ground state energy @xmath252 at 840 g from the entropy - energy experiments determines the ground state mean square size as @xmath253. hence, @xmath254. the corresponding @xmath255 from eq. [eq : groundstateratio]. since the mean square sizes are determined from the images and the ratio @xmath256 is number independent, this result shows that the systematic error arising from the number measurement is within the quoted error estimate. we also can determine @xmath16 by directly extrapolating to zero entropy the ratio of the axial mean square size of the weakly interacting fermi gas at 1200 g to that of strongly interacting gas at 840 g. when this is done, we obtain @xmath257, in very good agreement with the estimates based on the sound speed and ground state energy. finally, we can estimate the correction to the ground state energy arising from the finite scattering length at 840 g, @xmath258. for the trap conditions in the @xmath4 measurements, @xmath259, where @xmath260 is the wavevector for an ideal fermi gas at the trap center. to estimate the true unitary ground state energy at @xmath261, we first determine the leading order @xmath262 correction to the trapped atom density, where @xmath263 is the local fermi wavevector corresponding to the density @xmath192. the local chemical potential is estimated from ref. @xcite. using the notation of eq. [eq : reducedtruepotential] and a harmonic approximation, the corrected density yields @xmath264 $], where @xmath265 is the fermi radius for the unitary gas. according to the virial theorem (see eq. [eq : energymeas]), the mean square size and energy of the unitary gas are corrected by the same factor. the unitary ground state energy is then @xmath266 for @xmath267, we obtain @xmath268 and the value of @xmath269 obtained directly from @xmath270 is shifted to @xmath271. we also obtain the corrected value of @xmath272 in eq. [eq : groundstateratio] and @xmath273. table [tbl : beta] compares the values of @xmath16 obtained in our experiments to several recent predictions. note that the table does not include the finite @xmath274 correction for the @xmath4 measurement at 840 g described above. ''' '' & @xmath16 + ''' '' @xmath4 experiment & -0.59(2) + ''' '' sound velocity experiment & -0.565(15) + ''' '' cloud size ratio experiment & -0.61(2) + ''' '' ref. @xcite & -0.58(1) + ''' '' ref. @xcite & @xmath275 + ''' '' ref. @xcite & -0.599 + ''' '' ref. @xcite & -0.60(1) + ''' '' ref. @xcite & -0.646(4) +
Universal thermodynamic functions
using the @xmath4 data for the strongly interacting fermi gas and the temperature determined from the two power - law fits, we estimate several universal functions. first, we determine the dependence of the energy on temperature @xmath19 and the corresponding heat capacity, @xmath20. then we find the global chemical potential of the trapped gas as a function of the energy @xmath21. the energy is readily determined as a function of temperature using eq. [eq : evsslowengtwopower] for the case where there is a heat capacity jump and @xmath141, @xmath276;\,\,\,0\leq t\leq t_c\nonumber\\ e_>(t)&=&e_c+{\frac{s_ct_c}{d}}\left[\left(\frac{t}{t_c}\right)^{\frac{d}{d-1}}-1\right] ; \,\,\,t\geq t_c, \label{eq : energyvstemp}\end{aligned}\]] where the energy (temperature) is given in units of @xmath46 (@xmath47) and the critical energy @xmath15 is @xmath277 with @xmath18 the ground state energy. for the case with @xmath146, where the heat capacity is continuous, we determine the ordered pairs @xmath278 $] as a function of @xmath2 and plot @xmath19. fig. [fig : energyvstemp] shows the results using the best fits for both cases. of particular interest is the low temperature power law. for @xmath141, we obtain @xmath279 and @xmath280. since @xmath147 is near @xmath281, the energy relative to the ground state scales approximately as @xmath282. this is consistent with sound modes dominating the low energy excitations. however, one would expect instead that the free fermions on the edges of the trapped cloud would make an important contribution to the low energy excitations @xcite. over an extended range of @xmath201, the net entropy arising from the bose and fermi excitations has been predicted to scale as @xmath283, yielding an energy scaling @xcite as @xmath284. in this case, one would expect that @xmath285, i.e., @xmath286 in eq. [eq : energyvstemp], so that @xmath287. hence, the low energy power law exponents for the entropy should be between @xmath281 and @xmath288, which is barely distinguishable for our data. the heat capacity at constant trap depth @xmath289 is readily obtained from eq. [eq : energyvstemp] (where there is a heat capacity jump, since we have constrained @xmath141 in eq. [eq : evsslowengtwopower]). for this parameterization, @xmath290 where @xmath6 and @xmath8 are given in units of @xmath47, and @xmath7 is given in units of @xmath124. for the fit with a continuous heat capacity, we use @xmath195 to find @xmath291, and plot the ordered pairs @xmath292 $]. the heat capacity curves for both cases are shown in fig. [fig : cvst]. according to eq. [eq : heatcapacity], a jump in heat capacity occurs at @xmath7 : @xmath293 and @xmath294 differ when the power law exponents @xmath147 and @xmath148 are different. this is a consequence of the simple two power - law structure assumed for the fit function @xmath4 given by eq. [eq : evsslowengtwopower] for @xmath141, and can not be taken as proof of a true heat capacity jump. at present, the precise nature of the behavior near the critical temperature can not be determined from our data, and it remains an open question whether the data exhibits a heat capacity jump or a continuous heat capacity. the global chemical potential @xmath295 is readily determined from the fits to the @xmath4 data for a strongly interacting fermi gas, which obeys universal thermodynamics. the local energy density generally takes the form @xmath296, where @xmath297 is the local internal energy, which includes the kinetic energy and the interaction energy. here, @xmath192 is the local density, @xmath95 is the local chemical potential, @xmath298 is the pressure and @xmath299 is the total entropy per unit volume. the local chemical potential can be written as @xmath300, where @xmath24 is the trap potential. in the universal regime, where the local pressure depends only on the local density and temperature, we have @xmath301, as noted by ho @xcite. hence, @xmath302. integrating both sides over the trap volume and using @xmath303, where @xmath1 and @xmath304) are the total energy and average potential energy per particle, respectively, we obtain @xmath305 where @xmath2 is the entropy per particle. for simplicity, we assume harmonic confinement and use the virial theorem result, @xmath306, from eq. [eq : energymeas], which holds in the universal regime. then, eq. [eq : globalchempot1] yields the global chemical potential of a harmonically trapped fermi gas in the universal regime, @xmath307 by using the fit to the measured entropy - energy data to obtain the temperature @xmath5 from eq. [eq : evsslowengtwopower], the global chemical potential of a trapped unitary fermi gas can be calculated from eq. [eq : globalchempot]. for @xmath141, where the heat capacity has a jump, the simple power law fits above and below @xmath15 each yield a different linear dependence of @xmath295 on @xmath1, @xmath308 where @xmath309. we plot the chemical potential in fig. [fig : chemenergy]. the data points are obtained using eq. [eq : globalchempot] with the measured energy @xmath1 and entropy @xmath2 and the temperature determined from the fit to the @xmath4 data, using @xmath141 in eq. [eq : evsslowengtwopower], i.e., with a heat capacity jump. the solid red curve is given by eq. [eq : globalchempotvse]. we note that the low temperature data points in @xmath4 are best fit with the power law @xmath279, which is close to @xmath281. according to eq. [eq : globalchempotvse], this produces a slope near zero for @xmath310. since the power - law fit above @xmath15 gives @xmath311, the slope according to eq. [eq : globalchempotvse] changes from nearly zero for @xmath310 to negative for @xmath312. note that from eq. [eq : globalchempot], we obtain the slope @xmath313 since the entropy @xmath2 is continuous, we see that a jump in the heat capacity produces a corresponding jump in the slope of @xmath295 versus @xmath1. for comparison, fig. [fig : chemenergy] also shows the chemical potential obtained for @xmath146 in eq. [eq : evsslowengtwopower], where the heat capacity is continuous (blue dashed curve).
Conclusion
we have studied the thermodynamic properties of a strongly interacting fermi gas by measuring both the energy and the entropy. the model - independent data obtained in both the superfluid and the normal fluid regimes do not employ any specific theoretical calibrations, and therefore can be used as a benchmark to test the predictions from many - body theories and simulations. parameterizing the energy - entropy data determines the temperature of the strongly interacting fermi gas and also yields estimates of the critical parameters. we use the measured data to calibrate two different temperature scales that were employed in observations of the onset of pair condensation and in heat capacity studies. these calibrations yield critical temperatures in good agreement with the results estimated from our energy - entropy data. our data does not determine whether the heat capacity exhibits a jump or is continuous at the critical temperature. however, for a finite system with nonuniform density, the latter is most likely. considering that there is huge interest in determining the detailed behavior of the superfluid transition in a strongly interacting fermi gas @xcite, more precise determinations of the critical temperature, the heat capacity, and the chemical potential near the critical point, as well as the high temperature behavior and the approach to the ideal gas limit, will be important topics for future research.
Acknowledgements
this research has been supported by the physics divisions of the army research office and the national science foundation, the physics for exploration program of the national aeronautics and space administration, and the chemical sciences, geosciences and biosciences division of the office of basic energy sciences, office of science, u. s. department of energy. we thank willie ong for a careful reading of the manuscript. | strongly interacting fermi gases provide a clean and controllable laboratory system for modeling strong interparticle interactions between fermions in nature, from high temperature superconductors to neutron matter and quark - gluon plasmas.
model - independent thermodynamic measurements, which do not require theoretical models for calibrations, are very important for exploring this important system experimentally, as they enable direct tests of predictions based on the best current non - perturbative many - body theories. at duke university, we use all - optical methods to produce a strongly interacting fermi gas of spin-1/2-up and spin-1/2-down @xmath0li atoms that is magnetically tuned near a collisional (feshbach) resonance.
we conduct a series of measurements on the thermodynamic properties of this unique quantum gas, including the energy @xmath1, entropy @xmath2, and sound velocity @xmath3.
our model - independent measurements of @xmath1 and @xmath2 enable a precision study of the finite temperature thermodynamics.
the @xmath4 data are directly compared to several recent predictions.
the temperature in both the superfluid and normal fluid regime is obtained from the fundamental thermodynamic relation @xmath5 by parameterizing the @xmath4 data using two different power laws that are joined with continuous @xmath1 and @xmath6 at a certain entropy @xmath7, where the fit is optimized.
we observe a significant change in the scaling of @xmath1 with @xmath2 above and below @xmath7. taking the fitted value of @xmath7 as an estimate of the critical entropy for a superfluid - normal fluid phase transition in the strongly interacting fermi gas
, we estimate the critical parameters.
our @xmath4 data are also used to experimentally calibrate the endpoint temperatures obtained for adiabatic sweeps of the magnetic field between the ideal and strongly interacting regimes.
this enables the first experimental calibration of the temperature scale used in experiments on fermionic pair condensation, where the ideal fermi gas temperature is measured before sweeping the magnetic field to the strongly interacting regime.
our calibration shows that the ideal gas temperature measured for the onset of pair condensation corresponds closely to the critical temperature @xmath8 estimated in the strongly interacting regime from the fits to our @xmath4 data.
we also calibrate the empirical temperature employed in studies of the heat capacity and obtain nearly the same @xmath8.
we determine the ground state energy by three different methods, using sound velocity measurements, by extrapolating @xmath4 to @xmath9 and by measuring the ratio of the cloud sizes in the strongly and weakly interacting regimes.
the results are in very good agreement with recent predictions. finally, using universal thermodynamic relations, we estimate the chemical potential and heat capacity of the trapped gas from the @xmath4 data.
pacs numbers : 03.75.ss | 0811.1159 |
Introduction.
when one is dealing with classical field theories on a spacetime, the metric may appear as a given background field or it may be a genuine dynamic field satisfying the einstein equations. the latter theories are often generally covariant, with the spacetime diffeomorphism group as symmetry group, but the former often are considered to have only the isometry group of the metric as a symmetry group. however, @xcite (see also @xcite) indicated how theories with a background metric can be parametrized, that is, considered as theories that are fully covariant, if one introduces the diffeomorphisms themselves as dynamic fields. the goal of this paper is to develop this idea in the context of multisymplectic classical field theory and to make connections with stress - energy - momentum (`` sem '') tensors. as we shall see, the multimomenta conjugate to these new covariance fields form, to borrow a phrase from elasticity theory, the piola kirchhoff version of the sem tensor, and their euler lagrange equations are vacuously satisfied by virtue of the fact that the sem tensor is covariantly conserved. thus these fields have no physical content ; they serve only to provide an efficient way of parametrizing a field theory. nonetheless, the resulting generally covariant field theory has several attractive features, chief among which is that it is fully dynamic all fields satisfy euler lagrange equations. structurally, such theories are much simpler to analyze than ones with absolute objects or noncovariant elements. we emphasize that the results of this paper are for those field theories whose lagrangians are built from dynamic matter or other fields and a non - dynamic background metric. one of our motivations was to find a way to treat background fields and dynamic fields in a unified way in the context of the adjoint formalism. many of the ideas are applicable to a wider range of field theories, as @xcite already indicates, but in this paper we confine ourselves to this important class. the general case is presented in @xcite along with a more detailed discussion of parametrization theory and related topics.
The covariance construction.
suppose that we have a metric field theory in which the metric is an absolute object in the sense of @xcite. for instance, one might consider a dynamic electromagnetic field propagating on a schwarzschild spacetime. such a theory is not generally covariant, because the spacetime is fixed, and not all fields are on an equal footing, as the electromagnetic field is dynamic while the gravitational field is not. a somewhat different example is provided by nordstr@xmath0 m s theory of gravity (see 17.6 of @xcite), which is set against a minkowskian background. in this section we explain how to take such a system and construct from it an equivalent field theory that achieves the following goals : (i) : : the new field theory is generally covariant, and (ii) : : all fields in the new field theory are dynamic. this `` covariance construction '' is an extension and refinement of the parametrization procedure introduced by @xcite. [[setup.]] setup. + + + + + + as usual for a first order classical field theory, we start with a bundle @xmath1 whose sections, denoted @xmath2, are the fields under consideration. the dimension of @xmath3 is taken to be @xmath4, and we suppose that @xmath3 is oriented. let @xmath5 be a lagrangian density for this field theory, where @xmath6 is the first jet bundle of @xmath7 and @xmath8 is the space of top forms on @xmath3. loosely following the notation of @xcite or @xcite, we write coordinates for @xmath9 as @xmath10. in addition, in coordinates, we shall write @xmath11 evaluated on the first jet prolongation of a section @xmath2, the lagrangian becomes a function of @xmath12 ; we shall abbreviate this when convenient and simply write @xmath13. we assume that the fields @xmath2 are dynamic. [[example.]] example. + + + + + + + + we will intersperse the example of electromagnetism throughout the paper to illustrate our results. then @xmath14 is the cotangent bundle of 4-dimensional spacetime @xmath3, sections of which are electromagnetic potentials @xmath15. the corresponding lagrangian is written below. @xmath16 [[a - first - attempt - at - general - covariance.]] a first attempt at general covariance. + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + suppose that the spacetime @xmath3 comes equipped with a fixed, background metric @xmath17. the obvious first step in attaining general covariance is to allow @xmath17 to vary ; thus the metric will now be regarded as a genuine _ field _ @xmath18 on @xmath3. (when the metric is regarded as variable, we denote it by @xmath18, and when we want to revert to its fixed value we use @xmath17.) so we are led to view the lagrangian density as a map @xmath19 where @xmath20 is the bundle whose sections are lorentz metrics on @xmath3. we correspondingly write @xmath21 ; the semicolon is used to separate the dynamic from the nondynamic fields. (we emphasize that @xmath18 being variable does not mean that it is dynamic ; we discuss this point momentarily.) notice that we have tacitly assumed that the dependence of @xmath22 on the metric is pointwise that is, we have non - derivative coupling. (the more general case of derivative coupling will be considered in 5. in any event, we remark that derivatively - coupled theories are considered by many to be pathological.) [[example.-1]] example. + + + + + + + + the electromagnetic lagrangian density @xmath23 is @xmath24 where @xmath25 @xmath16 next, assume that the given lagrangian density @xmath22 has the following (eminently reasonable) covariance property for a diffeomorphism @xmath26 : @xmath27 where we assume that a way to lift the spacetime diffeomorphism @xmath28 to a bundle automorphism @xmath29 of @xmath14 has been chosen. [[example.-2]] example. + + + + + + + + for the electromagnetic 1-form potential @xmath15, we take the lift to be push - forward on the fiber, which makes it obvious that holds in this case. when condition holds, we say that the theory is generally covariant, i.e., the lagrangian density is @xmath30-equivariant. thus we have accomplished objective (i). however, the reader may well remark that this was ` too easy,'and would be quite right. the problem is that it is not clear how, or even _ if _, @xmath18 can now be made dynamic. certainly, @xmath18 can not be taken to be variational unless one adds a source term to the lagrangian density for @xmath18, for otherwise @xmath31 as the metric non - derivatively couples to the other fields. but what should this source term be? if @xmath18 is gravity, we could use the hilbert lagrangian, but otherwise this is unclear. [[the - covariance - field.]] the covariance field. + + + + + + + + + + + + + + + + + + + + + the solution to our problem requires more subtlety. we will sidestep both the issues of making @xmath17 variable, and then making @xmath18 dynamic, in one fell swoop as follows. we introduce an entirely new field, the `` covariance field '' into the theory. it will ` soak up'the arbitrariness in @xmath18, and will be dynamic. in this way we are able to generate a new generally covariant field theory, physically equivalent to the original one, in which all fields are dynamic. here is the construction. the key idea is to introduce a copy @xmath32 of spacetime into the fiber of the configuration bundle. consider (oriented) diffeomorphisms @xmath33, thought of as sections of the bundle @xmath34. we regard the diffeomorphisms @xmath35 as new fields, and correspondingly replace the configuration bundle by @xmath36. next, modify @xmath22 to get the new lagrangian @xmath37 defined on @xmath38 : @xmath39 thus, we obtain a modified field theory with the underlying bundle @xmath40. the general set up is shown in the figure below. -40pt the general set up for the introduction of covariance fields. let coordinates on @xmath41 be denoted @xmath42 and the associated jet coordinates be denoted @xmath43. then, writing @xmath44 and similarly for @xmath45, in coordinates equation reads @xmath46 where from the definition of pull - back @xmath47 we obtain @xmath48 from one verifies that the euler lagrange equations for the fields @xmath49 remain unchanged. [[example.-3]] example. + + + + + + + + for the electromagnetic field, our construction produces @xmath50 where @xmath51 is the jacobian of @xmath35 and @xmath52 @xmath16 we pause to point out the salient features of our construction. first, the fixed metric @xmath17 on spacetime is no longer regarded as living on @xmath3, but rather on the copy @xmath41 of @xmath3 in the fiber of the configuration bundle @xmath40. so @xmath17 is no longer considered to be a field it has been demoted to a mere geometric object on the fiber @xmath41. second, the variable metric @xmath18 on @xmath3 is identified with @xmath53, and thus acquires its variability from that of @xmath35. so @xmath18 as well is no longer a field per se, but simply an abbreviation for the quantity @xmath53. finally, we gain a field @xmath35 which we allow to be dynamic ; in the next subsection we will see that this imposes no restrictions on the theory at all. the first key observation is that the modified theory is indeed generally covariant. to this end, recall that, as was explained earlier, given @xmath54, there is assumed to be a lift @xmath55. for the trivial bundle @xmath56, we define @xmath57 the lagrangian density @xmath58 is @xmath59-equivariant, that is, @xmath60 this is an easy consequence of the definitions and, and the covariance assumption. indeed @xmath61 & = { \mathcal{l } } \left (j^1{\mspace{-1.5mu } } (\sigma_y(\phi))\, ; (\sigma^{-1 }) ^*(\eta^ * g)\right) \\[1.5ex] & = \sigma_*\!\left(\mathcal{l}(j^1 { \mspace{-1.5mu}}\phi)\, ; (\eta ^ * g))\right)\\[1.5ex] & = \sigma_*\!\left(\widetilde { \mathcal{l } } (j^1 { \mspace{-1.5mu}}\phi, j^1 { \mspace{-1.5mu}}\eta) \right).\end{aligned}\]] -24pt because of this property, we call @xmath35 the covariance field. [[example.-4]] example. + + + + + + + + from it is clear that the modified electromagnetic theory is generally covariant.
The dynamics of the covariance field.
next we will show something remarkable : the euler lagrange equation for the covariance field @xmath35 is vacuous. this is the main reason that, in the present context, we can introduce @xmath35 as a dynamic field with impunity, namely, its euler lagrange equation does not add any new information to, or impose any restrictions upon, the system. since, as we mentioned earlier, the euler lagrange equations for the fields @xmath49 remain unaltered, we see that _ the parametrized system is physically equivalent to the original system. _ first we compute the multimomenta conjugate to the field @xmath35 for the parametrized field theory with lagrangian @xmath37. recall that in multisymplectic field theory, the multimomenta conjugate to the multivelocities @xmath62 are defined by @xmath63 using the chain rule together with the relations and, we find that @xmath64 recall from @xcite that, as we have assumed that @xmath18 is the only nondynamic field, and does not derivatively couple to the other fields, the sem tensor density for the _ original _ system with lagrangian @xmath65 and metric @xmath18 is given by the hilbert formula : @xmath66 from we conclude that the multimomenta conjugate to to the covariance field @xmath35 are given by the piola - kirchhoff sem tensor density : @xmath67 this is a familiar object in elasticity theory. observe that @xmath68 is a two - point tensor density : it has one leg (@xmath69) in the spacetime @xmath41 in the fiber analogous to the spatial representation in elasticity theory, and the other leg (@xmath70) in the spacetime @xmath3 in the base analogous to the material representation. now we compute the euler lagrange equations for the @xmath71. these are : @xmath72 for @xmath73. expanding the derivatives via the chain rule and using the same type of calculation as in the derivation of to write the equations in terms of @xmath65 rather than @xmath74, the preceding equation becomes @xmath75 replacing @xmath76 by (half of) @xmath77, and differentiating using the product rule, we obtain @xmath78 for @xmath79. multiplying by the inverse matrix @xmath80 one gets @xmath81 for @xmath82. and now, we multiply by @xmath83, the inverse matrix of the jacobian @xmath84 @xmath85 for @xmath86. taking into account the symmetry @xmath87, the preceding equation becomes @xmath88 & - \ 2 \left(\frac{\partial \mathfrak t^{\mu \rho}}{\partial x^\mu}+\mathfrak t^{\mu \nu}\eta ^b{}_{,\mu \nu } \kappa^\rho{}_b \right) = 0.\end{aligned}\]] recalling the expression of the christoffel symbols of the metric @xmath17, namely, @xmath89 we obtain @xmath90 finally, recall how the christoffel symbols @xmath91 for @xmath17 and the symbols @xmath92 for @xmath93 are related : @xmath94 using this in gives @xmath95 for @xmath86, which is exactly the vanishing of the covariant divergence of the tensor _ density _ @xmath77. thus, we have proven the following basic result. the euler lagrange equations for the covariance field @xmath35 are that the covariant divergence of the sem tensor density @xmath96 is zero. it is known from proposition 5 in @xcite that the sem tensor is covariantly conserved when the metric @xmath18 is the _ only _ nondynamic field. thus, in our context, the equation @xmath97 is an identity, whence the euler lagrange equations for the covariance field @xmath35 are vacuously satisfied. consequently the covariance field has no physical import. we are free to suppose @xmath35 is dynamic, and so we have accomplished goal (ii) : we have constructed a new field theory in which _ all _ fields are dynamic.
The sem tensor.
it is interesting to compare the sem tensors for the original and parametrized systems. in @xcite the sem tensor density @xmath98 is defined in terms of fluxes of the multimomentum map @xmath99 associated to the action of the spacetime diffeomorphism group. we rapidly recount some of the basic ideas. consider the lift of an infinitesimal diffeomorphism @xmath100 to @xmath14 ; it can be expressed @xmath101 where we suppose that @xmath102 for some coefficients @xmath103. the largest value of @xmath104 for which one of the top coefficients @xmath105 is nonzero is the differential index of the field theory. we assume henceforth that the index @xmath106the most common and important case (e.g., when the fields are all tensor fields). in this context, theorem 1 along with remark 4 of @xcite shows that the sem tensor density @xmath107 for a lagrangian density @xmath108 is uniquely determined by @xmath109 for all vector fields @xmath110 on @xmath3 with compact support and all hypersurfaces @xmath111, where @xmath112 is the inclusion. the multimomentum map @xmath113 gives, roughly speaking, the flow of momentum and energy through spacetime ; according to the quoted theorem, the fluxes of this flow across hypersurfaces are realized via the sem tensor density. manipulation of (see formula (3.12) of @xcite) shows that @xmath107 is given by @xmath114 where the summation extends over _ all _ fields @xmath115. we apply this to the newly parametrized theory. note that if the index of the original theory is @xmath106, then that for the parametrized theory will be also. as well from we see that the lift of @xmath116 to @xmath56 is trivial : @xmath117 that is, there are no terms in the @xmath118 directions in @xmath119. thus the corresponding coefficients @xmath120 all vanish. the sem tensor for @xmath37 therefore reduces to @xmath121 on the other hand, @xmath122 and @xmath123 so that we can write @xmath124 but the first four terms on the rhs of this equation comprise the sem tensor density of the original theory since the @xmath125 do not derivatively couple to the @xmath49 (cf. (4.4) in @xcite). thus the sem tensor densities of the original and parametrized systems are related according to : @xmath126 but then @xmath127 on shell by the hilbert formula. therefore, we explicitly see that the sem tensor density for the fully covariant, fully dynamic modified theory vanishes. one can also obtain this result directly by applying the generalized hilbert formula (3.13) in @xcite to the parametrized theory, since it is fully dynamic. [[example.-5]] example. + + + + + + + + in the case of electromagnetism, one may compute directly from that @xmath127. one could also compute from that @xmath128 -32pt @xmath16
Derivative couplings.
here we briefly consider the situation, although perhaps exotic, when the metric derivatively couples to the other fields. for simplicity, however, we suppose the theory remains first order. so the lagrangian density is taken to be a map @xmath129 as before, modify @xmath22 to get the new lagrangian @xmath37 defined on @xmath130 : @xmath131 (since @xmath132 depends upon the first derivatives of @xmath35, @xmath133 will depend upon its second derivatives. thus, we obtain a modified _ second _ order field theory with the underlying bundle @xmath40.) the discussion proceeds as in the above, with only obvious changes. in particular, if @xmath134 is diff@xmath135-covariant, then so is @xmath136. as a simple illustration of a derivatively coupled theory, consider a vector meson with mass @xmath137. then @xmath14 is the tangent bundle of spacetime, and its sections @xmath138 are klein gordon vector fields. the lagrangian density is the map @xmath139 defined by @xmath140 where the semicolon denotes the covariant derivative with respect to @xmath18. our construction produces @xmath141 \nonumber \\[1.5ex] & \times \big [\phi^{{\mspace{1.5mu}}\rho}{}_{,\nu } + \big(\eta^h{}_{,\nu\xi } + \eta^p{}_{,\nu } { \mspace{1.5mu}}\eta^q{}_{,\xi } { \mspace{1.5mu}}\gamma ^h _ { pq } \big) \kappa^\rho{}_h{\mspace{1.5mu}}\phi^\xi\big] \nonumber \\[1.5ex] & - m^2 \phi^\sigma \phi^{{\mspace{1.5mu}}\rho } \bigg)\sqrt{-g}\, (\det \eta _ *) \, d^{{\mspace{1.5mu}}4}{\mspace{-1.5mu}}x\end{aligned}\]] where @xmath51 is the jacobian of @xmath35 and we have made use of. now we turn to the euler lagrange equations for the @xmath71 which, since @xmath136 is second order in the @xmath71, are : @xmath142 for @xmath73. the calculation of the lhs is similar to the previous one, but slightly more complicated. in any event, we find that @xmath35 satisfies the euler lagrange equations @xmath143 @xmath97, where now by the hilbert formula @xmath144.\]] thus for (first order) derivative couplings the covariance field remains vacuously dynamic. it is likely this will remain true for derivative couplings of arbitrary order, but we have not verified this as yet. we dedicate this paper to darryl holm on his 60@xmath145 birthday. we thank him for his interest in the ideas in this paper and for his many inspiring works over the years. mjg and jem thank the national science foundation for its occasional support of work of this sort. mcl was partially supported by dgsic (spain) under grant mtm2007 - 60017. | we give an exposition of the parametrization method of @xcite in the context of the multisymplectic approach to field theory, as presented in @xcite.
the purpose of the formalism developed herein is to make any classical field theory, containing a metric as a sole background field, generally covariant (that is, _ parametrized _, with the spacetime diffeomorphism group as a symmetry group) as well as fully dynamic.
this is accomplished by introducing certain `` covariance fields '' as genuine dynamic fields.
as we shall see, the multimomenta conjugate to these new fields form the piola
kirchhoff version of the stress - energy - momentum tensor field, and their euler
lagrange equations are vacuously satisfied.
thus, these fields have no additional physical content ; they serve only to provide an efficient means of parametrizing the theory.
our results are illustrated with two examples, namely an electromagnetic field and a klein gordon vector field, both on a background spacetime. | 0712.1883 |
Introduction
nowadays the family of iron pnictides is a well - established and important prototype system for unconventional high - temperature superconductivity. starting with the first famous compound @xcite in 2008, today several different sub - families with a wide structural variety are known. all different groups of iron pnictides share some common physical properties, such as their interesting and sometimes puzzling magnetic behavior. most compounds show a phase transition at low temperatures from a tetragonal to an orthorhombic crystal symmetry which is typically accompanied by the formation of long - range antiferromagnetic order.@xcite it is common believe that the suppression of these phase transitions for example by chemical substitution is crucial for the emergence of unconventional superconductivity.@xcite although it is obvious that an understanding of the magnetic fluctuations in the iron pnictides is mandatory to unveil the physics underlying the superconductivity, this task has proven to be more complex than anticipated.@xcite for example, there was discussion in the literature whether the magnetic moments are better described by an itinerant@xcite or a localized@xcite model and there is up to now no consensus concerning the role of correlation effects@xcite. furthermore, the magnitude of the magnetic moments is difficult to reproduce within density functional theory (dft) and it is known to be quite sensitive to computational parameters.@xcite one of the most important experimental tools to get insight into the electronic structure of the iron pnictides is angle - resolved photoemission spectroscopy (arpes). there are numerous publications on this topic, although it was shown that dft calculations have typically problems to reproduce all features of the arpes spectra correctly.@xcite this is often ascribed to strong correlation effects, although this question is still under discussion.@xcite another important difficulty which so far is often ignored is the connection between the magnetic phase of the iron pnictides and the resulting consequences for arpes. this is due to the formation of twinned crystals during the phase transition from tetragonal to orthorhombic and it results in mixed magnetic domains which are orthogonal to each other. macroscopic tools like arpes or transport measurements can so only see the averaged information, while information on the anisotropy is lost.@xcite this is a huge drawback considering a comprehensive study of the electronic structure in the iron pnictides, as it is known that the in - plane anisotropy plays a significant role.@xcite in experiment it is possible to effectively detwin the crystals by applying uniaxial stress during the measurement. this was already done successfully for the 122-prototype in the undoped and in the co - doped case. however, such measurements are connected with several technical difficulties and consequently they are rarely done.@xcite yet, to fully understand the electronic properties of the iron pnictide superconductors in a comprehensive way and to get a deeper insight concerning the influence of the in - plane anisotropy in the magnetic phase such studies are absolutely mandatory. although there is nowadays experimental data on detwinned crystals showing clearly the anisotropy in the fermi surface there is hardly any theoretical work focusing on this problem of magnetic anisotropy in arpes data. in this work this issue is addressed by a comprehensive dft study on the magnetic phase of and on the corresponding arpes spectra. the computational results can be directly compared to the available experimental arpes data on detwinned crystals.@xcite in order to deal with this complex situation the korringa - kohn - rostoker - green function (kkr - gf) approach is used, which was already shown to be indeed a very useful and accurate tool to deal with the iron pnictides.@xcite the impact of disorder due to substitution is dealt with by means of the coherent potential approximation (cpa), giving results fully compatible to supercell calculations and more reliable than those based on the virtual crystal approximation (vca).@xcite
Computational details
all calculations have been performed self - consistently and fully relativistically within the four component dirac formalism, using the munich spr - kkr program package.@xcite the orthorhombic, antiferromagnetic phase of is investigated in its experimentally observed stripe spin state using a full 4-fe unit cell. this implies antiferromagnetic chains along the @xmath1- and @xmath2-axes and ferromagnetic chains along the @xmath3-axis. the lattice parameters where chosen according to experimental x - ray data and the experimental as position @xmath4.@xcite to account for the influence of substitution in a linear interpolation for the lattice parameters with respect to the concentration @xmath0 is used based on available experimental data@xcite and vegard s law@xcite. more details on the procedure can be found in a previous publication.@xcite the treatment of disorder introduced by substitution is dealt with by means of the cpa. the basis set considered for a @xmath5 including @xmath6, @xmath7, @xmath8, @xmath9 and @xmath10 orbitals. for the electronic structure calculations the local density approximation (lda) exchange - correlation potential with the parameterization given by vosko, wilk and nusair was applied.@xcite the spectroscopical analysis is based on the fully relativistic one - step model of photoemission in its spin density matrix formulation. for more technical details on these calculations see ref.@xcite. the geometry of the spectroscopy setup was taken from experiment including a tilt of the sample around either the @xmath1 or @xmath3 axis. the incident light hit the sample under a constant polar angle @xmath11 and an azimuthal angle @xmath12 of either @xmath13 or @xmath14. these geometries are referred to as @xmath15 and @xmath16, meaning the direction of the incident light is either parallel to the antiferromagnetic or the ferromagnetic in - plane directions. the corresponding electrons were collected with an angle @xmath17 of @xmath18 or @xmath19 and a varying angle @xmath20 between @xmath21 and @xmath22. this geometry is in line to the experimental setup. if not indicated otherwise, an as - terminated surface was chosen. however, the question of surface termination will be discussed in more detail in the following.
Results and discussion
to describe the anisotropy of the iron pnictides in arpes calculations reasonably well one needs first to ensure that the spin - dependent potentials from the self - consistent field (scf) calculations are accurate enough. obviously, the magnetic ordering plays a significant role concerning the anisotropy of the electronic structure and hence the quality of the theoretical description of the arpes spectra is determined by the quality of the spin - dependent potentials. the most meaningful indication for a proper description of the magnetic state is good agreement with experimental data on the magnetic order. for the iron pnictides this is known to be a non - trivial task as the magnetic moments are often overestimated by dft.@xcite for the undoped mother compound a total magnetic moment of @xmath23 was obtained. experiment reports a total magnetic moment of approximately @xmath24 from neutron diffraction@xcite while mssbauer spectroscopy@xcite and @xmath25sr spectroscopy@xcite coherently give a value of around @xmath26. hence, the calculated total magnetic moment is found in good agreement with experiment and captures the proper order of magnitude accurately.@xcite more importantly, the cpa allows to evaluate the substitution dependent self - consistent evolution of the magnetic moments with increasing co concentration in. the corresponding results are shown in fig. [fig_magnmom], where the results for spin and orbital magnetic moments are given in an atom - resolved way. the total magnetic moment is calculated as substitutionally averaged sum over all contributions. in agreement with experiment the total magnetic moments shows a nearly linear decay until the long - range magnetic order disappears.@xcite in the calculations the critical co substitution for the disappearance of antiferromagnetic order occurs for @xmath27, which is in reasonably good agreement with the experimental value @xmath28.@xcite it should be mentioned that the results in fig. [fig_magnmom] are slightly improved with respect to experiment in comparison with our previous work@xcite due to the higher @xmath29 expansion used here. however, the trends in the magnetic moments and the resulting conclusions are the same. as the one - step model of photoemission fully accounts for matrix - elements as well as for surface effects the resulting spectra can be directly compared to experimental arpes data. as stressed before, it is extremely difficult to see the magnetic anisotropy correctly in experimental spectra because of the twinning of crystals. here reference is made especially to the work of yi _ et al._@xcite, who did remarkable measurements on detwinned single crystals of and by applying uniaxial stress to the crystals. similar results were obtained for example by kim _ et al._@xcite. in this context it is important to note that the brillouin zone (bz) of the magnetic 4-fe spin - density - wave (sdw) state is only half the size compared to the bz in the nonmagnetic 2-fe state. for that reason it is most appropriate to use in the following the notation for the 4-fe sdw bz where the information of @xmath30 and @xmath31 from the nonmagnetic bz is down - folded to one @xmath32 point.@xcite in fig. [fig_cal_afmain] the fermi surface around the @xmath32 point is shown in the sdw bz as calculated from the spin - dependent potentials for a photon energy of @xmath33ev. the overlay of black points corresponds to the experimentally measured bz, reproduced from the work of yi _ et al._@xcite. as can be seen, the agreement of the calculated fermi surface and the experimental data is remarkably good. characteristic are the bright intensity spots along the @xmath34-direction (i.e. along @xmath1), corresponding directly to the antiferromagnetic order along the @xmath1-axis and the bigger pedal - like structures along the @xmath35-direction (i.e. along @xmath3) which corresponds to the ferromagnetic order along the @xmath3-axis. (12,10.0) (2.20,1.2) in the 4-fe sdw bz for a photon energy of @xmath33ev. the overlay of black points is a reconstruction of the sdw bz from experimental arpes data, reproduced from the work of yi _ et al._@xcite.,title="fig : "] (1.2,1.35)(0,1.95)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (1.85,0.9)(1.95,0)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (0.2,4.0) (5.2,-0.1)@xmath36 $] (10.7,1.2) in the 4-fe sdw bz for a photon energy of @xmath33ev. the overlay of black points is a reconstruction of the sdw bz from experimental arpes data, reproduced from the work of yi _ et al._@xcite.,title="fig : "] (11.0,9.5)(0,0)max (11.0,1.0)(0,0)min it should be noted that in fig. [fig_cal_afmain] the intensity over two different light polarizations was averaged, namely for the direction of the incident light either parallel to the antiferromagnetic @xmath1-axis @xmath15 or parallel to the ferromagnetic @xmath3-axis @xmath16. all features of the electronic structure are visible for both polarizations of light. however, the intensity patterns vary notably with the polarization due to matrix element effects, indicating strong multiorbital character, just as seen in experiment.@xcite if not indicated otherwise this averaging will be applied in the following. for comparison the two contributions to the total fermi surface for @xmath33ev are shown polarization - resolved in fig. [fig_cal_afpol], for the incident light direction being either parallel to the @xmath3-axis @xmath16(fig. [fig_cal_afpol] (a)) or parallel to the @xmath1-axis @xmath15(fig. [fig_cal_afpol] (b)). (5.9,8) (1.10,1.2)-axis (a) or parallel to the antiferromagnetic @xmath1-axis (b)., title="fig : "] (0.2,1.35)(0,1.3)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (0.75,0.9)(1.3,0)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (-0.6,3.0) (2.7,-0.1)@xmath36 $] (0.65,7.0)(a) (2.75,7.0)@xmath16 (7.3,8) (1.20,1.2)-axis (a) or parallel to the antiferromagnetic @xmath1-axis (b)., title="fig : "] (0.85,0.9)(1.3,0)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (2.8,-0.1)@xmath36 $] (0.65,6.95)(b) (2.75,6.95)@xmath15 (7.0,1.2)-axis (a) or parallel to the antiferromagnetic @xmath1-axis (b)., title="fig : "] (7.3,6.95)(0,0)max (7.3,1.0)(0,0)min it can be seen that for @xmath16 the intensity of the bright spots along the @xmath1-axis is significantly enhanced while for @xmath15 the intensity around the inner circle of @xmath32 is enhanced. this polarization dependence is again in full agreement with the experimental findings.@xcite at this point it was shown that the detwinned, antiferromagnetic fermi surface obtained by the calculations agrees very well with experiment. one may also ask how the fermi surface of a twinned, antiferromagnetic crystal should look like and how does it differ from the nonmagnetic case. therefore, the calculated fermi surfaces for both cases are shown in fig. [fig_cal_twin] and compared with experimental arpes data@xcite of twinned crystals at @xmath37k (a) and @xmath38k (b). please note that the transition from a paramagnetic to an antiferromagnetic state occurs at around @xmath39k, accordingly the experimental data shown as overlay of black points corresponds to the nonmagnetic and the twinned antiferromagnetic state, respectively. the representation of the twinned fermi surface is based on a superposition of spectra obtained independently for antiferromagnetic states rotated by @xmath19 against each other which is supposed to be a good approximation for twinned crystals, see for example the work of tanatar _ et al._@xcite. (5.9,8) (1.10,1.2) against each other. the overlay of black points is reproduced from the experimental arpes data of yi _ et al._@xcite, title="fig : "] (0.2,1.35)(0,1.3)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (0.75,0.9)(1.3,0)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (-0.6,3.0) (2.7,-0.1)@xmath36 $] (0.65,7.0)(a) (2.1,7.0)nonmagnetic (7.3,8) (1.20,1.2) against each other. the overlay of black points is reproduced from the experimental arpes data of yi _ et al._@xcite, title="fig : "] (0.85,0.9)(1.3,0)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (2.8,-0.1)@xmath36 $] (0.65,6.95)(b) (2.75,6.95)twinned (7.0,1.2) against each other. the overlay of black points is reproduced from the experimental arpes data of yi _ et al._@xcite, title="fig : "] (7.3,6.95)(0,0)max (7.3,1.0)(0,0)min comparing the nonmagnetic and the twinned antiferromagnetic state with each other it is obvious that there is a significant difference in the shape of the fermi surface which is due to the underlying change of the electronic structure during the magnetic phase transition. however, the twinned fermi surface is in principle isotropic along @xmath34 and @xmath35 due to the fact that the in - plane anisotropy cancels almost completely for two magnetic domains that are rotated by @xmath19 against each other. this means that although some influence of the magnetic ordering can be seen for twinned crystals, it is not possible to deduce information about the important in - plane anisotropy from the corresponding spectra. this stresses again the importance of arpes measurements and calculations on detwinned crystals to investigate the magnetic structure correctly. to summarize, the agreement of the calculations with the experimental data is altogether quite well for the nonmagnetic as well as for the twinned magnetic state. (5.9,8.8) (1.1,1.2) dispersion as seen by arpes along the both in - plane real space axes @xmath1 (a) and @xmath3 (b). the black lines mark the photon energies where the alternation of @xmath40 and @xmath41 can be seen along @xmath42. notably, the vertical intensity stripes at @xmath43 in (a) seem almost independent on @xmath42, indicating some connection to a surface related phenomenon., title="fig : "] (0.4,1.95)(0,1.55)[l]2.5,3.0,3.5,4.0 (0.88,0.9)(1.28,0)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (-0.5,3.7) (2.8,-0.1)@xmath44 $] (0.65,8.3)(a) (2.85,8.2)@xmath1-axis (7.0,8.8) (1.1,1.2) dispersion as seen by arpes along the both in - plane real space axes @xmath1 (a) and @xmath3 (b). the black lines mark the photon energies where the alternation of @xmath40 and @xmath41 can be seen along @xmath42. notably, the vertical intensity stripes at @xmath43 in (a) seem almost independent on @xmath42, indicating some connection to a surface related phenomenon., title="fig : "] (0.88,0.9)(1.28,0)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (2.8,-0.1)@xmath45 $] (0.65,8.3)(b) (2.85,8.2)@xmath3-axis (6.9,1.2) dispersion as seen by arpes along the both in - plane real space axes @xmath1 (a) and @xmath3 (b). the black lines mark the photon energies where the alternation of @xmath40 and @xmath41 can be seen along @xmath42. notably, the vertical intensity stripes at @xmath43 in (a) seem almost independent on @xmath42, indicating some connection to a surface related phenomenon., title="fig : "] (7.25,8.2)(0,0)max (7.25,1.0)(0,0)min going back to the original study of detwinned antiferromagnetic crystals the @xmath42 dispersion is shown along the @xmath1- and @xmath3-axes in fig. [fig_cal_kz]. the difference between @xmath40 and @xmath41 manifests itself mainly by the alternating intensity distributions. we find @xmath40 for photon energies of @xmath4622ev and @xmath47ev respectively, while @xmath41 can be found at @xmath48. this is in good agreement with literature which reports @xmath41 at @xmath49ev and @xmath40 at @xmath50ev.@xcite the anisotropic features, namely the bright spots along @xmath34 and the pedals along @xmath35 seem quite independent on @xmath42, which agrees with the experimental reports on the detwinned crystals.@xcite for further discussion the fermi surfaces for @xmath51ev and @xmath47ev respectively are shown in fig. [fig_cal_3448]. the important aspect to note is that the anisotropic features are preserved independent on @xmath42, meaning they are preserved for @xmath40 as well as for @xmath41. however, the most striking anisotropy between the @xmath1- and @xmath3-directions seen in the @xmath42 dispersion are the almost vertical intensity lines along the @xmath1-axis for @xmath43. they are surprisingly robust concerning the @xmath42 dispersion, already indicating possible surface related phenomena, as will be discussed later in more detail. (5.9,8) (1.10,1.2)ev and @xmath47ev, corresponding to either @xmath41 (a) or to @xmath40 (b). it can be seen that the topology of interest, namely the anisotropic features are principally independent on @xmath42.,title="fig : "] (0.2,1.35)(0,1.3)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (0.75,0.9)(1.3,0)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (-0.6,3.0) (2.7,-0.1)@xmath36 $] (0.65,7.0)(a) (2.45,7.0)@xmath51ev (7.3,8) (1.20,1.2)ev and @xmath47ev, corresponding to either @xmath41 (a) or to @xmath40 (b). it can be seen that the topology of interest, namely the anisotropic features are principally independent on @xmath42.,title="fig : "] (0.85,0.9)(1.3,0)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (2.8,-0.1)@xmath36 $] (0.65,6.95)(b) (2.45,6.95)@xmath47ev (7.0,1.2)ev and @xmath47ev, corresponding to either @xmath41 (a) or to @xmath40 (b). it can be seen that the topology of interest, namely the anisotropic features are principally independent on @xmath42.,title="fig : "] (7.3,6.95)(0,0)max (7.3,1.0)(0,0)min (6.8,9.5) (1.1,1.2) (0.2,1.25)(0,0.96)[l]-0.4,-0.3,-0.2,-0.1, 0.0, 0.1, 0.2, 0.3 (1.05,0.9)(1.25,0)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (-0.4,3.5) (-0.50,8.9)@xmath52 (1.55,8.3)(a) (3.25,8.3)@xmath1-axis (6.8,8.8) (1.1,1.2) (0.12,1.25)(0,0.96)[l]-0.4,-0.3,-0.2,-0.1, 0.0, 0.1, 0.2, 0.3 (1.05,0.9)(1.25,0)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (1.55,8.3)(b) (2.7,8.3)nonmagnetic (7.8,8.8) (1.1,1.2) (0.12,1.25)(0,0.96)[l]-0.4,-0.3,-0.2,-0.1, 0.0, 0.1, 0.2, 0.3 (1.55,8.3)(c) (3.25,8.3)@xmath3-axis (7.3,1.2) (7.6,8.2)(0,0)max (7.6,1.0)(0,0)min + (6.8,9.1) (1.1,1.2) (0.2,1.25)(0,0.96)[l]-0.4,-0.3,-0.2,-0.1, 0.0, 0.1, 0.2, 0.3 (1.05,0.9)(1.25,0)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (-0.4,3.5) (3.2,-0.1)@xmath44 $] (-0.50,8.9)@xmath53 (1.55,8.3)(d) (3.25,8.3)@xmath1-axis (6.8,8.8) (1.1,1.2) (0.12,1.25)(0,0.96)[l]-0.4,-0.3,-0.2,-0.1, 0.0, 0.1, 0.2, 0.3 (1.05,0.9)(1.25,0)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (3.2,-0.1)@xmath45 $] (1.55,8.3)(e) (2.7,8.3)nonmagnetic (7.8,8.8) (1.1,1.2) (0.12,1.25)(0,0.96)[l]-0.4,-0.3,-0.2,-0.1, 0.0, 0.1, 0.2, 0.3 (1.05,0.9)(1.25,0)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (3.2,-0.1)@xmath45 $] (1.55,8.3)(f) (3.25,8.3)@xmath3-axis (7.3,1.2) (7.6,8.2)(0,0)max (7.6,1.0)(0,0)min to complete the study of the in - plane anisotropy in the undoped compound the spin - dependent bands are investigated polarization - dependent along the two in - plane directions @xmath1 and @xmath3 for @xmath33ev and for comparison the isotropic bands of the nonmagnetic case. anisotropies due to the orthorhombic lattice distortion for the nonmagnetic case are very small and have no significant influence, as shown also in earlier work.@xcite the nonmagnetic bands for the polarizations @xmath16 and @xmath15 are shown in fig. [fig_cal_bs00] (b) and (e), respectively. already for the nonmagnetic case it becomes obvious, that more information can be deduced for light with a polarization parallel to the ferromagnetic chains. for a perpendicular light polarization the intensity for some bands decrease so strongly that they practically seem to vanish. this is however not due to a vanishing of the bands but only due to the strong intensity variation, i.e. matrix element effects, as already mentioned before and as seen in experiment. for the spin - polarized band structure with antiferromagnetic order along the @xmath1-axis the corresponding cases for @xmath16 and @xmath15 are shown in fig. [fig_cal_bs00] (a) and (d), respectively. the green solid lines are guides to the eye which emphasize the two important anisotropic bands in these spectra. first of all, there is some significant reorientation of the bands compared to the nonmagnetic case. most striking is the appearance of a steep double - u shaped band which was not visible in the nonmagnetic case. this new appearance is most likely due to down - folding of the brillouin zone when going from the 2-fe cell to a magnetic 4-fe cell. the second important band is a pure hole pocket which is compared to the nonmagnetic case shifted to higher binding energies. it should be noted that the intensity of this band is extremely polarization dependent. it is the dominating band for @xmath16 while it is barely visible for @xmath16. comparing with the polarization dependent fermi surface in fig. [fig_cal_afpol] it is obvious that this band is also part of the bright intensity spots in the fermi surface along the @xmath1-direction and very characteristic for the anisotropy. it is also noteworthy that these two significant bands cross each other exactly at the fermi level. this crossing is also reported in experiment as can be seen from the black points in fig. [fig_cal_bs00] (a), which are reproduced from the experimental arpes data of yi _ et al._@xcite. thus, the experimental arpes data could be again well reproduced by the calculations. the situation for the magnetic bands with ferromagnetic order along the @xmath3-axis as shown in fig. [fig_cal_bs00] (c) and (f) is in many aspects similar to that for the bands along the @xmath1-direction. one can identify two prominent anisotropic bands, one with a double - u like shape which has a higher intensity for a light polarization of @xmath15, while the other band marked with the solid green line is the dominating one for light @xmath16. the important difference is that these bands do not touch each other as they are significantly shifted away in binding energy. note that also no crossing is reported in experiment.@xcite why these steep bands with the double - u shape can not be seen in experiment for the @xmath3-direction gets also clear : the responsible band is simply completely shifted above the fermi level. this observation can in principle be compared to the band splitting in ferromagnets. note that along the @xmath3-axis there is ferromagnetic coupling while along the @xmath1-axis the magnetic order is antiferromagnetic. thus, for one sees along the ferromagnetic chains a band splitting of approximately @xmath54ev for a magnetic moment around @xmath55. this is comparable for example to ni which shows a band splitting of approximately @xmath56ev for a moment of approximately @xmath57.@xcite consequently, for decreasing magnetic moments upon alloying one expects a reduced band splitting together with a continuous matching of the anisotropic bands. to investigate this issue in further detail one has to look at the evolution of the arpes band structure for increasing co substitution on the fe position which goes in hand with the reduction of the magnetic moments. (6.6,9.3) (1.1,1.2) (0.2,1.25)(0,0.96)[l]-0.4,-0.3,-0.2,-0.1, 0.0, 0.1, 0.2, 0.3 (1.05,0.9)(1.25,0)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (-0.4,3.5) (3.2,-0.1)@xmath44 $] (-0.50,8.9)*_a_-axis * (1.55,8.3)(a) (3.0,8.3)@xmath58 (6.6,8.8) (1.1,1.2) (0.25,1.25)(0,0.96)[l]-0.4,-0.3,-0.2,-0.1, 0.0, 0.1, 0.2, 0.3 (1.05,0.9)(1.25,0)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (3.2,-0.1)@xmath36 $] (1.55,8.3)(b) (3.0,8.3)@xmath59 (6.6,8.8) (1.1,1.2) (0.25,1.25)(0,0.96)[l]-0.4,-0.3,-0.2,-0.1, 0.0, 0.1, 0.2, 0.3 (1.05,0.9)(1.25,0)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (3.2,-0.1)@xmath36 $] (1.55,8.3)(c) (3.0,8.3)@xmath60 (7.0,8.8) (1.1,1.2) (0.25,1.25)(0,0.96)[l]-0.4,-0.3,-0.2,-0.1, 0.0, 0.1, 0.2, 0.3 (1.05,0.9)(1.25,0)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (3.2,-0.1)@xmath36 $] (1.55,8.3)(d) (3.0,8.3)@xmath61 (7.0,1.2) (7.3,8.2)(0,0)max (7.3,1.0)(0,0)min + (6.6,9.3) (1.1,1.2) (0.2,1.25)(0,0.96)[l]-0.4,-0.3,-0.2,-0.1, 0.0, 0.1, 0.2, 0.3 (1.05,0.9)(1.25,0)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (-0.4,3.5) (3.2,-0.1)@xmath45 $] (-0.50,8.9)*_b_-axis * (1.55,8.3)(e) (3.0,8.3)@xmath58 (6.6,8.8) (1.1,1.2) (0.25,1.25)(0,0.96)[l]-0.4,-0.3,-0.2,-0.1, 0.0, 0.1, 0.2, 0.3 (1.05,0.9)(1.25,0)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (3.2,-0.1)@xmath45 $] (1.55,8.3)(f) (3.0,8.3)@xmath59 (6.6,8.8) (1.1,1.2) (0.25,1.25)(0,0.96)[l]-0.4,-0.3,-0.2,-0.1, 0.0, 0.1, 0.2, 0.3 (1.05,0.9)(1.25,0)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (3.2,-0.1)@xmath45 $] (1.55,8.3)(g) (3.0,8.3)@xmath60 (7.0,8.8) (1.1,1.2) (0.25,1.25)(0,0.96)[l]-0.4,-0.3,-0.2,-0.1, 0.0, 0.1, 0.2, 0.3 (1.05,0.9)(1.25,0)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (3.2,-0.1)@xmath45 $] (1.55,8.3)(h) (3.0,8.3)@xmath61 (7.0,1.2) (7.3,8.2)(0,0)max (7.3,1.0)(0,0)min substitution of fe with co in is one of the common ways to induce superconductivity in by electron doping. the substitution does consequently diminish the strength of the antiferromagnetic coupling within this compound until the long - range magnetic order collapses and superconductivity emerges.@xcite as the strength of the magnetic order decreases with co doping, in experiment as well as in the calculations, it can be assumed that the strong in - plane anisotropy does also decrease. the breakdown of the long - range antiferromagnetic order in fig. [fig_magnmom] appears for a somewhat higher co concentrations than in experiment. thus, the breakdown of the anisotropy is expected at higher doping levels. this is true for a concentration of @xmath61 in which is the first substitution level where the magnetic order did completely vanish in the self - consistent calculation. considering the evolution of the magnetic moments in fig. [fig_magnmom] one can see that the initial magnetic moment decreased to approximately @xmath62, @xmath63 and @xmath64 of the original value for co substitutions of @xmath58, @xmath65 and @xmath66 respectively. to investigate the impact of alloying in detail, the arpes band structure for these concentrations including the already nonmagnetic @xmath61 for both directions @xmath1 and @xmath3 is presented in fig. [fig_cal_bsco]. the calculations are performed for @xmath33ev comparable to fig. [fig_cal_bs00] but they are only shown for a light polarization of @xmath16 because is was already clarified that the anisotropic bands can be best seen with this specific polarization. the black dashed lines in fig. [fig_cal_bsco] are shown for comparison and they correspond to the band position of the anisotropic bands in the undoped compound with the highest anisotropy, seen in fig. [fig_cal_bs00] (a) and (c). the green solid lines are guides to the eye to identify more easily the corresponding anisotropic bands for the specific co concentrations. the difference between black dashed lines and green solid lines is thus the change of the original anisotropy with increasing co substitution. for the case of the nonmagnetic with @xmath61 shown in fig. [fig_cal_bsco] (d) and (h) the anisotropy has completely vanished and the band structures coincide with each other. this could be expected from experiment and it is reproduced in the calculations. it should be noted again at this point, that the crystal lattice is still orthorhombic, however, the lattice anisotropy is indeed too weak to be visible in the band structure.@xcite some other interesting findings can be deduced from the evolution of the band structure upon co substitution. first of all the intensity of the double - u shaped band decreases continuously, however, it only completely disappears after the collapse of the long - range antiferromagnetic order. the change in anisotropy for the antiferromagnetic order along the @xmath1-axis is mostly characterized with the consequent shift of the hole - pocket to lower binding energies. this is also experimentally reported for a decrease in the magnetic coupling strength, either induced through co doping or increasing temperature.@xcite concerning this situation for the ferromagnetic order along the @xmath3-axis the most prominent feature is the shift of the double - u shaped band to lower binding energies. what can be seen in fig. [fig_cal_bs00] (e) to (h) is that the energy difference of these two main anisotropic bands does strongly and continuously decrease. this is in agreement with the assumption of a smaller band splitting for decreasing ferromagnetic coupling strength. in summary one can say that for the antiferromagnetic order along the @xmath1-axis mostly the hole - pocket changes while the double - u shaped band stays more or less constant. for the ferromagnetic order along the @xmath3-axis it is the other way round. the double - u shaped band undergoes the strongest change while the other band stays more or less unchanged in energy and shape. the final result is the same in both cases, a matching of the bands and a consequent isotropic in - plane band structure. this detailed analysis allows to follow the change from the strong in - plane anisotropy of the undoped compound to the isotropic behavior in the co substituted system in a continuous way with direct correspondence to arpes. thus, this approach based on kkr - cpa proves its advantages for investigating the iron pnictide superconductors at regions of interest which are difficult to evaluate by means of other band structure methods. c|c (6.2,9.3) (1.1,1.2) (0.2,1.15)(0,1.48)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (2.10,0.9)(1.45,0)[l]-0.2, 0.0, 0.2 (-0.5,2.8) (3.0,-0.0)@xmath44 $] (4.50,8.6)*as - terminated * (2.5,7.5)(a) (5.5,7.5)fermi surface (7.0,1.2) (7.1,7.2)(0,0)max (7.1,1.0)(0,0)min (7.5,8.8) (1.1,1.2) (2.10,0.9)(1.45,0)[l]-0.2, 0.0, 0.2 (3.0,-0.0)@xmath36 $] (7.0,1.2) (7.1,7.2)(0,0)max (7.1,1.0)(0,0)min & (6.2,9.3) (1.1,1.2) (0.2,1.15)(0,1.48)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (2.10,0.9)(1.45,0)[l]-0.2, 0.0, 0.2 (3.0,-0.0)@xmath44 $] (4.50,8.6)*ba - terminated * (2.5,7.5)(c) (5.5,7.5)fermi surface (7.0,1.2) (7.1,7.2)(0,0)max (7.1,1.0)(0,0)min (7.5,8.8) (1.1,1.2) (2.10,0.9)(1.45,0)[l]-0.2, 0.0, 0.2 (3.0,-0.0)@xmath36 $] (7.0,1.2) (7.1,7.2)(0,0)max (7.1,1.0)(0,0)min + (6.55,9.3) (1.1,1.2) (0.2,1.25)(0,0.96)[l]-0.4,-0.3,-0.2,-0.1, 0.0, 0.1, 0.2, 0.3 (0.9,0.9)(1.25,0)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (-0.4,3.5) (3.0,-0.0)@xmath44 $] (2.0,8.5)(b) (3.0,8.5)band structure along @xmath1-axis (7.05,1.2) (7.25,8.2)(0,0)max (7.25,1.1)(0,0)min (7.4,8.8) (1.1,1.2) (0.9,0.9)(1.25,0)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (3.0,-0.0)@xmath36 $] (7.0,1.2) (7.25,8.2)(0,0)max (7.22,1.0)(0,0)min & (6.55,9.3) (1.1,1.2) (0.2,1.25)(0,0.96)[l]-0.4,-0.3,-0.2,-0.1, 0.0, 0.1, 0.2, 0.3 (0.9,0.9)(1.25,0)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (3.0,-0.0)@xmath44 $] (2.0,8.5)(d) (3.0,8.5)band structure along @xmath1-axis (7.05,1.2) (7.25,8.2)(0,0)max (7.25,1.1)(0,0)min (7.4,8.8) (1.1,1.2) (0.9,0.9)(1.25,0)[l]-0.4,-0.2, 0.0, 0.2, 0.4 (3.0,-0.0)@xmath36 $] (7.0,1.2) (7.25,8.2)(0,0)max (7.22,1.0)(0,0)min using the one - step model of photoemission one can identify different surface states and can thus clarify the origin of surface bands. the reason for the occurrence of surface - states has long been developed in multiple scattering theory, which is the underlying basis of the spr - kkr method.@xcite the so - called determinant condition uses the reflection matrices of the bulk crystal @xmath67 and of the surface barrier potential @xmath68, which connects the inner potential of the bulk crystal with the vacuum level. the appearance of a surface state is given by the following condition : @xmath69 for better visualization we plot @xmath70 in the following. if this expression is bigger than approximately @xmath71 we speak about a surface state. for values between @xmath72 and @xmath71 the state is defined as a so - called surface resonance. for values below one has bulk states. more details can be found in the overview by braun and donath.@xcite the application of this determinant approach is demonstrated in fig. [fig_cal_det]. here one can see the fermi surfaces and the band structures along the @xmath1-axis for an as - terminated and a ba - terminated surface, respectively, together with the corresponding plot of @xmath70 on the right hand side of each picture. the bands are shown for @xmath33ev and they are averaged over the two light polarizations @xmath16 and @xmath15 in order to make all relevant contributions equally visible. it should be noted, that the determinant condition itself and without a high intensity in the corresponding band structure plot is only an indication for a surface state or a surface resonance. only if a high intensity in the @xmath70 plot coincides with a band in the band structure one can associate this band with a clear surface character. for example has the bright octagon shape of the determinant plot of the fermi surface in fig. [fig_cal_det] (a) not a corresponding counterpart in the band structure plot. the two high intensity spots along the @xmath1-axis which are equally visible in the fermi surface as well as in the determinant condition are in clear contrast to this behavior. thus, this feature has a surface related origin, more specifically a surface state as the intensity of @xmath70 is in the order of @xmath73. this surface state can be also identified in the band structure along the @xmath1-axis as shown in fig. [fig_cal_det] (b) where a strong intensity in the determinant plot coincides with the steep bands that cut the fermi level and which are part of the already discussed double - u shape. consequently, these bands can be identified as surface states. this is in accordance with the earlier findings for the @xmath42-dispersion in fig. [fig_cal_kz] (a) where the vertical intensities at @xmath43 were independent on @xmath42, indicating a connection to a surface related phenomenon. another verification for the surface related origin of these bands is shown in fig. [fig_cal_det] (c) and (d), where the corresponding fermi surface and band structure are shown for a ba - terminated surface. as already indicated before all other calculations presented in this paper are under the assumption of an as - terminated surface. obviously, the assumed surface termination has also an influence on surface related phenomena and surface states might be shifted significantly in energy. indeed, the surface states discussed for the as - terminated surface have completely vanished in the ba - terminated case. the corresponding high intensities are missing in the @xmath70 plots and in the band structure of fig. [fig_cal_det] (d) the characteristic steep bands from the as - terminated surface have also vanished. no intensities in the determinant plot coincide with features in the band structure and thus surface effects have been removed by the ba termination. overall, the fermi surface and the band structure have undergone significant changes for the altered surface termination. the characteristic anisotropic features of the fermi surface in fig. [fig_cal_det] (a) are hardly visible in the ba - terminated case in fig. [fig_cal_det] (c). it seems like the ba layer on top acts as some kind of damping layer which reduces the intensity and blurs the electronic states which are clearly visible in an as - terminated surface. in particular one has to note that the agreement with experimental arpes data is significantly better for an as - terminated surface compared to the ba - terminated one. especially the steep bands along the @xmath1-axis are seen in experiment@xcite and they could be successfully identified as surface states which are only visible for an as - terminated surface. this result can be used for conclusions on the most likely surface termination of. interestingly, the surface termination in this material is still not clear and under debate, although several experimental measurements and theoretical calculations exist.@xcite according to first principle calculations only three possibilities for the surface termination exist, namely a fully as - terminated or a fully ba - terminated surface as well as an as surface covered with half of the stoichiometric ba atoms.@xcite there are experimental scanning tunneling microscopy (stm) and low - energy electron - diffraction (leed) measurements which indicate a ba - terminated surface@xcite. however, there are also experimental stm + leed data which clearly favor an as - terminated surface@xcite. the arpes calculations clearly favor an as - terminated surface as it was shown that agreement with experiment is considerably better compared to the ba - terminated one. this can not rule out the possibility of some partial covering with few ba adatoms but one can state that every ba atoms on top muffles the electronic structure seen in experiment and that this structure is due to an as - terminated compound. so one would expect a more or less clean as - terminated surface as most probable surface termination for. additional covering with some ba atoms might be possible but also might have some degrading influence on the quality of the arpes measurement.
Summary
the munich spr - kkr package was used for self - consistent and fully relativistic calculations of orthorhombic in its experimentally observed stripe antiferromagnetic ground - state for @xmath74 up to @xmath61. the substitutional disorder induced by co on fe positions was dealt with on a cpa level which was earlier shown to be fully equal to a comprehensive supercell calculation.@xcite magnetic moments of @xmath23 for undoped were reproduced and additionally a reasonable magnetic behavior for increasing co substitution with a continuous decrease of the magnetic moments until a collapse of the antiferromagnetic order at 15% co concentration was reached. this is in good agreement with experimental behavior.@xcite concerning arpes most experimental data available is actually insufficient to talk about in - plane anisotropy due to twinning effects during the phase transition from the nonmagnetic tetragonal to the antiferromagnetic orthorhombic phase. a complicated detwinning process, typically with uniaxial stress on the single crystal, is necessary to gain anisotropic data of the electronic structure.@xcite referring to the available experimental data it was possible to reproduce the electronic structure of in very good agreement with experiment. the fermi surface shows all important anisotropic features, namely some bright spots of intensity along the antiferromagnetic order along the @xmath1-axis and more blurred pedals along the ferromagnetic order along the @xmath3-axis. also in agreement with experiment a strong dependence on the polarization of light was found, been either parallel to the ferromagnetic or to the antiferromagnetic order, indicating the strong multiorbital character. for comparison the fermi surface of the nonmagnetic phase as well as a hypothetical fermi surface for a twinned arpes measurement as a superposition of two antiferromagnetic cells rotated by @xmath19 to each other was shown. both were again in agreement with experiment. in addition to the anisotropic @xmath42 dispersion some focus was put on the anisotropic band structure along the @xmath1- and @xmath3-axes and it was compared to the nonmagnetic case. one could identify the important anisotropic bands and these could be interpreted in terms of band splitting for the ferromagnetic chains along the @xmath3-axis, principally comparable to typical ferromagnetic band splitting as observed for example in ni. in addition the evolution of these anisotropic bands for small steps of @xmath0 in was presented until the breakdown of long - range antiferromagnetic order. the decreasing band splitting and a continuous matching of the anisotropic bands could be reproduced in great detail and consistent with experimental findings. finally the so - called determinant condition @xmath70 was used to evaluate possible surface states of the band structure. it was possible to identify steep bands along the @xmath1-axis as surface states. these are at least partially responsible for the characteristic bright intensity spots in the electronic structure along the @xmath1-axis and can also be seen in experiment. interestingly, these surface states are only visible near the fermi level for an as - terminated surface. it was shown that a ba - terminated top - layer acts as some kind of damping which moves the surface states far away and blurs the electronic structure. significantly better agreement with experimental data is found for an as - terminated surface. this leads to the conclusion that an as - terminated surface would be most likely, an issue that is in fact experimentally not convincingly clarified.@xcite some ba adatoms might be still possible but one would expect a negative influence on the quality of the measurements. to conclude, this publication was successful in reproducing the strong in - plane anisotropy of and its behavior under substitution in very good agreement with experiment using arpes calculations. these calculations allow even predictions on possible surface terminations.
Acknowledgments
we acknowledge the financial support from the deutsche forschungsgemeinschaft dfg (projects for 1346) and from the bundesministerium fr bildung und forschung bmbf (project 05k13wma). we further thank for the support from centem plus (l01402). j. fink, s. thirupathaiah, r. ovsyannikov, h. a. drr, r. follath, y. huang, s. de jong, m. s. golden, y.- z. zhang, h. o. jeschke, r. valent, c. felser, s. dastjani farahani, m. rotter, and d. johrendt, phys. b * 79 *, 155118 (2009). l. x. yang, y. zhang, h. w. ou, j. f. zhao, d. w. shen, b. zhou, j. wei, f. chen, m. xu, c. he, y. chen, z. d. wang, x. f. wang, t. wu, g. wu, x. h. chen, m. arita, k. shimada, m. taniguchi, z. y. lu, t. xiang, and d. l. feng, phys. lett. * 102 *, 107002 (2009). h. gretarsson, a. lupascu, j. kim, d. casa, t. gog, w. wu, s. r. julian, z. j. xu, j. s. wen, g. d. gu, r. h. yuan, z. g. chen, n.- l. wang, s. khim, k. h. kim, m. ishikado, i. jarrige, s. shamoto, j.- h. chu, i. r. fisher, and y.- j. kim, phys. b * 84 *, 100509 (2011). p. vilmercati, a. fedorov, f. bondino, f. offi, g. panaccione, p. lacovig, l. simonelli, m. a. mcguire, a. s. m. sefat, d. mandrus, b. c. sales, t. egami, w. ku, and n. mannella, phys. b * 85 *, 220503 (2012). v. b. zabolotnyy, d. s. inosov, d. v. evtushinsky, a. koitzsch, a. a. kordyuk, g. l. sun, j. t. park, d. haug, v. hinkov, a. v. boris, c. t. lin, m. knupfer, a. n. yaresko, b. buchner, a. varykhalov, r. follath, and s. v. borisenko, nature * 457 *, 569 (2009). d. v. evtushinsky, d. s. inosov, v. b. zabolotnyy, a. koitzsch, m. knupfer, b. bchner, m. s. viazovska, g. l. sun, v. hinkov, a. v. boris, c. t. lin, b. keimer, a. varykhalov, a. a. kordyuk, and s. v. borisenko, phys. b * 79 *, 054517 (2009). d. v. evtushinsky, d. s. inosov, v. b. zabolotnyy, m. s. viazovska, r. khasanov, a. amato, h.- h. klauss, h. luetkens, c. niedermayer, g. l. sun, v. hinkov, c. t. lin, a. varykhalov, a. koitzsch, m. knupfer, b. bchner, a. a. kordyuk, and s. v. borisenko, new journal of physics * 11 *, 055069 (2009). d. v. evtushinsky, v. b. zabolotnyy, t. k. kim, a. a. kordyuk, a. n. yaresko, j. maletz, s. aswartham, s. wurmehl, a. v. boris, d. l. sun, c. t. lin, b. shen, h. h. wen, a. varykhalov, r. follath, b. bchner, and s. v. borisenko, phys. b * 89 *, 064514 (2014). m. yi, d. lu, j.- h. chu, j. g. analytis, a. p. sorini, a. f. kemper, b. moritz, s.- k. mo, r. g. moore, m. hashimoto, w.- s. lee, z. hussain, t. p. devereaux, i. r. fisher, and z.- x. shen, proc. ac. sci. us. * 108 *, 6878 (2011). a. a. aczel, e. baggio - saitovitch, s. l. budko, p. c. canfield, j. p. carlo, g. f. chen, p. dai, t. goko, w. z. hu, g. m. luke, j. l. luo, n. ni, d. r. sanchez - candela, f. f. tafti, n. l. wang, t. j. williams, w. yu, and y. j. uemura, phys. rev. b * 78 *, 214503 (2008). t. goko, a. a. aczel, e. baggio - saitovitch, s. l. budko, p. c. canfield, j. p. carlo, g. f. chen, p. dai, a. c. hamann, w. z. hu, h. kageyama, g. m. luke, j. l. luo, b. nachumi, n. ni, d. reznik, d. r. sanchez - candela, a. t. savici, k. j. sikes, n. l. wang, c. r. wiebe, t. j. williams, t. yamamoto, w. yu, and y. j. uemura, phys. b * 80 *, 024508 (2009). t. kondo, r. m. fernandes, r. khasanov, c. liu, a. d. palczewski, n. ni, m. shi, a. bostwick, e. rotenberg, j. schmalian, s. l. budko, p. c. canfield, and a. kaminski, phys. b * 81 *, 060507 (2010). v. b. nascimento, a. li, d. r. jayasundara, y. xuan, j. oneal, s. pan, t. y. chien, b. hu, x. b. he, g. li, a. s. sefat, m. a. mcguire, b. c. sales, d. mandrus, m. h. pan, j. zhang, r. jin, and e. w. plummer, phys. * 103 *, 076104 (2009). | by means of one - step model calculations the strong in - plane anisotropy seen in angle - resolved photoemission of the well - known iron pnictide prototype compounds and in their low - temperature antiferromagnetic phases is investigated.
the fully - relativistic calculations are based on the korringa - kohn - rostoker - green function approach combined with the coherent potential approximation alloy theory to account for the disorder induced by co substitution on fe sites in a reliable way.
the results of the calculations can be compared directly to experimental spectra of detwinned single crystals.
one finds very good agreement with experiment and can reveal all features of the electronic structure contributing to the in - plane anisotropy.
in particular the local density approximation can capture most of the correlation effects for the investigated system without the need for more advanced techniques.
in addition, the evolution of the anisotropy for increasing co concentration @xmath0 in can be tracked almost continuously.
the results are also used to discuss surface effects and it is possible to identify clear signatures to conclude about different types of surface termination. | 1602.05027 |
Bethe free energy and belief propagation
consider a generic factor model @xcite with a binary configurational space, @xmath12, @xmath13, which is factorized so that the probability @xmath14 to find the system in the state @xmath15 and the partition function @xmath16 are @xmath17 where @xmath9 labels non - negative and finite factor - functions @xmath18 with @xmath19 and @xmath20 represents a subset of @xmath21 variables. relations between factor functions (checks) and elementary discrete variables (bits), expressed as @xmath3 and @xmath4, can be conveniently represented in terms of the system - specific factor (tanner) graph. if @xmath3 we say that the bit and the check are neighbors. any spin (a - posteriori log - likelihood) correlation function can be calculated using the partition function, @xmath16, defined by eq. ([p1]). general expression for the factor functions of an ldpc code is @xmath22 let us now reproduce the derivation of the belief propagation equation based on the bethe free energy variational principle, following closely the description of @xcite. (see also the appendix of @xcite.) in this approach trial probability distributions, called beliefs, are introduced both for bits and checks @xmath23 and @xmath24, respectively, where @xmath25, @xmath26. a belief is defined for given configuration of the binary variables over the code. thus, a belief at a bit actually consists of two probabilities, @xmath27 and @xmath28, and we use a natural notation @xmath29. there are @xmath30 beliefs defined at a check, @xmath31 being the number of bits connected to the check, and we introduce vector notation @xmath32 where @xmath33 and @xmath12. beliefs satisfy the following inequality constraints @xmath34 the normalization constraints @xmath35 as well as the consistency (between bits and checks) constraints @xmath36 where @xmath37 stands for the set of @xmath38 with @xmath39, @xmath40. the bethe free energy is defined as a difference of the bethe self - energy and the bethe entropy, @xmath41 where @xmath42, @xmath33 and @xmath12. the entropy term for a bit enters eq. ([bethe]) with the coefficient @xmath43 to account for the right counting of the number of configurations for a bit : all entries for a bit (e.g. through the check term) should give @xmath44 in total. optimal configurations of beliefs are the ones that minimize the bethe free energy ([bethe]) subject to the constraints ([ineq],[norm],[cons]). introducing these constraints into the effective lagrangian through lagrange multiplier terms @xmath45 and looking for the extremum with respect to all possible beliefs leads to @xmath46, \nonumber\\ & & \!\!\!\!\!\! \frac{\delta l}{\delta b_i(\sigma_i) } = 0 \label{lbi } \\ & & \!\!\rightarrow\quad b_i(\sigma_i)=\exp\left[\frac{1}{q_i-1}\left(\gamma_i+ \sum\limits_{\alpha\ni i}\lambda_{i\alpha}(\sigma_i)\right)-1\right]. \nonumber\end{aligned}\]] substituting @xmath47 into eq.([lba],[lbi]) we arrive at @xmath48 where @xmath49 is used to indicate that we should use the normalization conditions ([norm]) to guarantee that the beliefs sum up to one. applying the consistency constraint ([cons]) to eqs. ([ba]), making summation over all spins but the given @xmath21, and also using eq. ([bi]) we derive the following bp equations @xmath50 the right hand side of eq. ([ba0]) rewritten for the ldpc case ([factor_ldpc]) becomes @xmath51 thus constructing @xmath52 for the ldpc case in two different ways (correspondent to left and right relations in eq. ([ba0])), equating the results and introducing the @xmath53 field @xmath54 one arrives at the following bp equations for the @xmath53 fields : @xmath55 iterative solution of this equation corresponding to eq. ([iter]) with @xmath11 is just a standard iterative bp (which can also be called sum - product) used for the decoding of an ldpc code. a simplified min - sum version of eq. ([iter]) is @xmath56 \min_{j\neq i}^{j\in\beta } \big| \eta^{(n)}_{j\beta } \big|+ \frac{1}{\delta}\sum\limits_{\beta\ni i}\eta_{i\beta}^{(n) }, \nonumber\end{aligned}\]]
Termination curve for standard iterative bp
to illustrate the standard bp iterative decoding, given by eqs. ([iter],[min - sum]) with @xmath57, we consider the example of the @xmath58 $] code of tanner @xcite performing over awgn channel channel characterized by the transition probability for a bit, @xmath59, where @xmath60 and @xmath61 are the input and output values at a bit and @xmath62 is the snr. launching a fixed codeword into the channel, emulating the channel by means of a standard monte - carlo simulations and then decoding the channel output constitutes our experimental " procedure. we analyze the probability distribution function of the iteration number @xmath0 at which the decoding terminates. the termination probability curve for the standard min - sum, described by eq. ([min - sum]) with @xmath11, is shown in fig. [tc123] for @xmath63. . notice that the probability of termination (successful decoding) without any iterations is always finite. few points on the right part of the plot correspond to the case when the decoding was not terminated even at the maximum number of iterations, @xmath64 (decoding fails to converge to a codeword). [tc123],width=288] the result of decoding is also verified at each iteration step for compliance with a codeword : iteration is terminated if a codeword is recovered. this termination strategy can still give an error, although the probability to confuse actual and a distant codewords is much less than the probability not to recover a codeword for many iterations. if one neglects the very low probability of the codewords confusion, then the probability of still having a failure after @xmath0 iterations is equal to the integral / sum over the termination curve from @xmath65 and up. note also that the probability that even infinite number of iterations will not result in a codeword can actually be finite. discussing fig. [tc123] one observes two distinct features of the termination probability curve. first, in all cases the curve reaches its maximum at some relatively small number of iterations. second, each curve crosses over to an algebraic - like decay which gets steeper with the snr increase. the emergence of an algebraically extended tail (that is a tail which does not decay fast) is not encouraging, as it suggests that increasing the number of iteration will not bring much of an improvement in the iterative procedure. it also motivates us to look for possibilities of accelerating convergence of the bp algorithm to a minimum of the bethe free energy. note also the wiggling of termination curves for @xmath66 near the crossover point (see fig. [tc123]). it is possibly related to the cycling of the bp dynamics (and thus the inability of bp to converge).
Relaxation to a minimum of the bethe free energy
the idea is to introduce relaxational dynamics (damping) in an auxiliary time, @xmath67, thus enforcing convergence to a minimum of the bethe free energy. one chooses @xmath29 as the main variational field and considers relaxing variational equations eqs. ([lbi]) according to @xmath68 while keeping the set of remaining variational equations eqs. ([norm],[lba],[cons]) intact. here positive parameters @xmath69 have the physical meaning of correlation / relaxation times. performing calculations, that are completely equivalent to the ones described in section [sec : bethe], we arrive at the following modified bp equations @xmath70 we are interested to approach (find) a solution of the original bp eq. ([bp1]). one assumes @xmath71, thus ignoring the second term under @xmath72 in eq. ([qeq]). the resulting continuous equation is @xmath73 eq. ([iter]) represents a simple discretized version of the eq. ([bp3]) where the correlation coefficients @xmath69 are chosen to make the coefficient in front of the second term on the left hand side of eq. ([bp3]) independent of the bit index, @xmath8. then the resulting time dependent coefficient can be rescaled to one by an appropriate choice of the temporal unit ; @xmath74 is the uniform discrete time, @xmath7 is positive integer, @xmath75 ; the left hand side (right hand side) of eq.([bp3]) is taken at @xmath76 (@xmath74) and the temporal derivative is discretized in a standard retarded way, @xmath77. this choice of relaxation coefficients and discretization, resulted in eq. ([iter]), was taken out of consideration in the final formula for simplicity, realizability at all positive @xmath10 and also its equivalence to the standard iterative bp at @xmath78.
Modified iterative bp: test of performance
we test the min - sum version ([min - sum]) of the modified iterative bp with the monte carlo simulations of the @xmath58 $] code at few values of snrs. the resulting termination probability curves are shown in fig. [tc] for @xmath79. the simulations show a shift of the probability curve maximum to the right (towards larger number of iterations) with the damping parameter decrease however once the maximum is achieved, the decay of the curve at a finite @xmath10 is faster with the number of iterations than in the standard bp case. the decay rate actually increases as @xmath10 decreases. (a), @xmath80 (b), and @xmath81 (c). [tc],width=288] (a), @xmath80 (b), and @xmath81 (c). [tc],width=288] (a), @xmath80 (b), and @xmath81 (c). [tc],width=288] we conclude that at the largest @xmath0 the performance of a modified iterative bp is strictly better. however to optimize the modified iterative bp, thus aiming at better performance than given by the standard iterative bp, one needs to account for the trade - off between decreasing @xmath10 leading to a faster decay of the termination probability curve at the largest @xmath0, but on the other side it comes with the price in the actual number of iteration necessary to achieve the asymptotic decay regime. the last point is illustrated by fig. [er], where the decoding error probability depends non - monotonically on @xmath10. one can also see that the modification of bp could improve the decoding performance ; e.g., at @xmath82 and maximally allowed @xmath83 (after which the decoding unconditionally stops) the decoding error probability is reduced by factor of about 40 by choosing @xmath84 (see the bottom curve at fig. [er](b)). for @xmath80 (a) and @xmath81 (b). different curves correspond to different maximally allowed @xmath0 : starting from @xmath85 (top curve) and increasing @xmath0 by factor of @xmath86 with each next lower curve. the points on the right correspond to the standard bp (@xmath87). [er],width=288] for @xmath80 (a) and @xmath81 (b). different curves correspond to different maximally allowed @xmath0 : starting from @xmath85 (top curve) and increasing @xmath0 by factor of @xmath86 with each next lower curve. the points on the right correspond to the standard bp (@xmath87). [er],width=288]
Conclusions and discussions
we presented a simple extension of the iterative bp which allows (with proper optimization in the @xmath88 plane) to guarantee not only an asymptotic convergence of bp to a local minimum of the bethe free energy but also a serious gain in decoding performance at finite @xmath0. in addition to their own utility, these results should also be useful for systematic improvement of the bp approximation. indeed, as it was recently shown in @xcite solution of the bp equation can be used to express the full partition function (or a - posteriori log - likehoods calculated within map) in terms of the so - called loop series, where each term is associated with a generalized loop on the factor graph. this loop calculus / series offers a remarkable opportunity for constructing a sequence of efficient approximate and systematically improvable algorithms. thus we anticipate that the improved iterative bp discussed in the present manuscript will become an important building block in this future approximate algorithm construction. we already mentioned in the introduction that our algorithm can be advantageous over other bp - based algorithms converging to a minimum of the bethe free energy mainly due to its simplicity and tunability. in particular, the concave - convex algorithms of @xcite, as well as related linear programming decoding algorithms @xcite, are formulated in terms of beliefs. on the contrary our modification of the iterative bp can be extensively simplified and stated in terms of the fewer number of @xmath89 fields each associated with an edge of the factor graph rather than with much bigger family of local code - words. thus in the case of a regular ldpc code with @xmath90 checks of the connectivity degree @xmath31 one finds that the number of variables taken at each step of the iterative procedure is @xmath91 and @xmath92 in our iterative scheme and in the concave - convex scheme respectively. having a tunable correlation parameter @xmath93 in the problem is also advantageous as it allows generalizations (e.g. by turning to a individual bit dependent relaxation rate). this flexibility is particularly desirable in the degenerate case with multiple minima of the bethe free energy, as it allows a painless implementation of annealing as well as other more sophisticated relaxation techniques speeding up and/or improving convergence. m. chertkov, v. chernyak, _ loop calculus in statistical physics and information science _, phys. e * 73 *, 065102(r) (2006) ; cond - mat/0601487. m. chertkov, v. chernyak, _ loop series for discrete statistical models on graphs _, j. stat. (2006) p06009, cond - mat/0603189. j. feldman, m. wainwright, d.r. karger, _ using linear programming to decode linear codes _, 2003 conference on information sciences and systems, the john hopkins university, march 12 - 14, 2003. | the decoding of low - density parity - check codes by the belief propagation (bp) algorithm is revisited.
we check the iterative algorithm for its convergence to a codeword (termination), we run monte carlo simulations to find the probability distribution function of the termination time, @xmath0. tested on an example @xmath1 $] code, this termination curve shows a maximum and an extended algebraic tail at the highest values of @xmath0. aiming to reduce the tail of the termination curve we consider a family of iterative algorithms modifying the standard bp by means of a simple relaxation.
the relaxation parameter controls the convergence of the modified bp algorithm to a minimum of the bethe free energy.
the improvement is experimentally demonstrated for additive - white - gaussian - noise channel in some range of the signal - to - noise ratios.
we also discuss the trade - off between the relaxation parameter of the improved iterative scheme and the number of iterations
. low - density parity - check (ldpc) codes @xcite are the best linear block error - correction codes known today @xcite.
in addition to being good codes, i.e. capable of decoding without errors in the thermodynamic limit of an infinitely long block length, these codes can also be decoded efficiently.
the main idea of belief propagation (bp) decoding is in approximating the actual graphical model, formulated for solving statistical inference maximum likelihood (ml) or maximum - a - posteriori (map) problems, by a tree - like structure without loops.
being efficient but suboptimal the bp algorithm fails on certain configurations of the channel noise when close to optimal (but inefficient) map decoding would be successful.
bp decoding allows a certain duality in interpretation.
first of all, and following the so - called bethe - free energy variational approach @xcite, bp can be understood as a set of equations for beliefs (bp - equations) solving a constrained minimization problem. on the other hand,
a more traditional approach is to interpret bp in terms of an iterative procedure so - called bp iterative algorithm @xcite.
being identical on a tree (as then bp equations are solved explicitly by iterations from leaves to the tree center) the two approaches are however distinct for a graphical problem with loops. in case of their convergence,
bp algorithms find a minimum of the bethe free energy @xcite, however in a general case convergence of the standard iterative bp is not guaranteed.
it is also understood that bp fails to converge primarily due to circling of messages in the process of iterations over the loopy graph. to enforce convergence of the iterative algorithm to a minimum of the bethe free energy some number of modifications of the standard iterative bp
were discussed in recent years.
the tree - based re - parametrization framework of @xcite suggests to limit communication on the loopy graph, cutting some edges in a dynamical fashion so that the undesirable effects of circles are suppressed.
another, so - called concave - convex procedure, introduced in @xcite and generalized in @xcite, suggests to decompose the bethe free energy into concave and convex parts thus splitting the iterations into two sequential sub - steps. noticing that convergence of the standard bp fails mainly due to overshooting of iterations
, we develop in this paper a tunable relaxation (damping) that cures the problem.
compared with the aforementioned alternative methods, this approach can be practically more advantageous due to its simplicity and tunability. in its simplest
form our modification of the bp iterative procedure is given by @xmath2 where latin and greek indexes stand for bits and checks and the bit / check relations, e.g. @xmath3, @xmath4 express the ldpc code considered ; @xmath5 is the channel noise - dependent value of log - likelihoods ; and @xmath6 is the message associated at the @xmath7-th iteration with the edge (of the respective tanner graph) connecting @xmath8-th bit and @xmath9-th check.
@xmath10 is a tunable parameter. by choosing a sufficiently small @xmath10 one can guarantee convergence of the iterative procedure to a minimum of the bethe free energy. on the other hand
@xmath11 corresponds exactly to the standard iterative bp. in the sequel
we derive and explain the modified iterative procedure ([iter]) in detail.
the manuscript is organized as follows.
we introduce the bethe free energy, the bp equation and the standard iterative bp in section [sec : bethe].
performance of standard iterative bp, analyzed with a termination curve, is discussed in section [sec : term].
section [sec : relax] describes continuous and sequentially discrete (iterative) versions of our relaxation method.
we discuss performance of the modified iterative scheme in section [sec : perf], where bit - error - rate and the termination curve for an ldpc code performed over additive - white - gaussian - noise (awgn) channel are discussed for a range of interesting values of the signal - to - noise - ratios (snr).
we also discuss here the trade - off between convergence and number of iterations aiming to find an optimal strategy for selection of the model s parameters.
the last section [sec : con] is reserved for conclusions and discussions. | cs0607112 |
Introduction
chemical evolution models (cem) @xcite were early developed to try to understand the origin of the radial gradients of abundances, observed in our galaxy (mwg). most numerical models in the literature, including the multiphase model used in this work, explain the existence of this radial gradient by the combined effects of a star formation rate (sfr) and an infall of gas which vary with galactocentric radius in the galaxy. a radial decrease of abundances has also been observed in most spiral galaxies @xcite although the shape of the radial distribution changes from galaxy to galaxy. among other global trends it is found that for isolated non - barred spirals the steepness of the radial gradient depends on morphological type, with later types showing steeper gradients @xcite, with other general galaxy properties as surface brightness and neutral and molecular gas fractions also playing a role @xcite. the radial gradient tends to be wiped out however for strongly barred galaxies which show flat abundance distributions @xcite. irregulars galaxies also show uniform abundances throughout @xcite. the abundance gradient pattern seems to show an on - off mode @xcite, being very steep for the latest spiral types and very flat for irregulars. all these considerations become clear when the gradient is measured in dex / kpc, but there are indications that suggest a gradient independent of galaxy type when it is measured in dex / scale length @xcite. in order to analyze the behaviour of the radial distribution of abundances and the value of the radial gradient from a theoretical point of view a large number of models is necessary. historically, cem aiming to reproduce radial abundance gradients have been, however, applied only to the mwg. actually, there is a lack of tools to determine the chemical evolutionary state of a particular galaxy, besides our works applying the multiphase models to spiral galaxies. the recent works by @xcite are valid for galaxies other than the mwg. their calculations use the angular momentum and rotation curves as model inputs keeping the star formation efficiency constant for all galaxies @xcite. this technique may not be flexible enough to validate the models against observational data. in fact, a comparison to see if these models reproduce the observed abundance radial distributions of particular galaxies has not been done. it is always possible to extract some information by using evolutionary synthesis models in comparison with spectro - photometric observations. this method, very useful for the study of elliptical galaxies, does not result equally successful in the case of spiral galaxies due to the difficulty of measuring the spectral indices, except for the bulges @xcite, from which ages and metallicities are obtained. furthermore, even when these measurements are done with confidence @xcite, in order to apply this technique to spiral galaxies, a combination of chemical evolution and evolutionary synthesis models is required to solve the uniqueness problem associated to the first ones and the age - metallicity degeneracy associated to the second ones @xcite. at present, the available options are either to use the classical closed box model or a galactic chemical evolution (gce) model. however, the closed box scenario is recognised to be inadequate to describe the evolution of most galaxies and in fact its application in many cases can yield misleading results @xcite. in particular, the fact of assuming that a system has a constant total mass with a monotonically decreasing star formation according to a schmidt law, prevents the reproduction of the observational characteristics of most galaxies. on the other hand, the evolution of a galaxy with present time properties different from the milky way will not necessarily be equal to that predicted by a gce model. realistic chemical evolution models adequate to describe different types of spiral and irregular galaxies are therefore clearly needed. the multiphase model, whose characteristics have been described in @xcite, has been applied and checked against observational constraints not only for the milky way galaxy @xcite, as it is commonly done, but also for a sample of spiral galaxies (discs and bulges) of different morphological types and total masses @xcite. the observed radial distributions of gas, oxygen abundances and star formation rate have been reproduced rather successfully and the observed correlations between abundance gradients and galaxy characteristics are also reproduced @xcite. this galaxy sample, which includes the best studied objects, is however small (only 11) and encompasses a restricted range of morphologies and masses. the application of the model can however be extended to a larger sample if an adequate parameter space is defined thus providing the required chemical evolution of different types of galaxies. the model uses as input parameters the collapse time scale to form the disc, which depends on the total mass of the galaxy, and the efficiencies to form molecular clouds and stars which we assume different from galaxy to galaxy. the radial distributions of total mass constitute the fundamental input of the multiphase model. they are easily computed when the rotation curves are available (moll & mrquez, in preparation). if this is not the case, some assumptions are necessary. in this work, we have used the universal rotation curve from (*??? * hereafter pss96) to calculate a large number of mass radial distributions representing theoretical protogalaxies or initial structures which will evolve to form the observed spiral discs or irregulars. the total mass of each simulated galaxy, besides having its own effect on the galaxy evolution, defines the characteristic collapse time - scale or gas infall rate onto the disc. regarding molecular cloud and star formation efficiencies, which will take values between 0 and 1, we have chosen 10 different sets of values for each radial distribution of total mass. we have computed how the chemical evolution proceeds for galaxies defined by the different parameter combinations. the final bi - parametric grid consists of 440 models simulating galaxies of 44 different total masses. this work represents an extension of our previous work that can help to understand the general trends observed in spiral and irregular galaxies concerning neutral and molecular gas distributions, abundance radial distributions etc. but, most importantly, by using these models we can predict the time evolution of a galaxy when only present time data are known. in section 2 we summarize the general characteristics of the multiphase chemical evolution model and the strategy of its application to the objects of the grid. the results are presented in section 3 including the time evolution of galaxies and some radial distributions for the present time. in section 4 we give the calibration of the grid models with the mwg and a restricted sample of well studied spiral galaxies and we discuss the model results in a global way. finally, the conclusions of this work are presented in section 5. the results obtained in this work will be available in electronic form at cds via anonymous ftp to cdsarc.u-strasbg.fr (130.79.128.5), via http://cdsweb.u-strasbg.fr/abstract.html, or at http://wwwae.ciemat.es/@xmath0mercedes.
The multiphase model
the model used in this work is a generalization of that developed for the solar neighborhood in @xcite and later applied to the whole mwg @xcite and other spiral galaxies @xcite. the enriched material proceeds from the restitution by dying stars, considering their nucleosynthesis, their initial mass function (imf) and hence the delayed restitution and their final fate, via a quiet evolution, or type i and ii supernova explosions. most recent works support the idea that the imf is practically universal in space and constant in time @xcite, showing only local differences. we have adopted the imf from @xcite, very similar to a scalo s law @xcite and in good agreement with the most recent data from @xcite, as can be seen in fig. [imf]. the original model has been modified in order to use metallicity dependent yields. nucleosynthesis yields for massive stars have been taken from @xcite. for low mass and intermediate stars we have used the set of yields from @xcite. for type i supernova explosion releases, the model w7 from @xcite, as revised by @xcite has been taken. in this model each galaxy is described as a two - zone system with a halo and a disk. it is assumed that the halo has a total mass which is initially in gas phase. the total mass, m, and its radial distribution, m(r), are calculated from the corresponding rotation curve derived from the universal rotation curve of pss96. these authors use a homogeneous sample of about 1100 optical and radio rotation curves to estimate their profile and amplitude which are analysed statistically. from this study they obtain an expression for the rotation velocity, v(r), as a function of the rotation velocity at the optical radius (the radius encompassing 83% of the total integrated light), @xmath1, the galaxy radius normalised to the optical one, @xmath2, and a parameter @xmath3, which represents the ratio of the galaxy luminosity to that of the mwg, @xmath4 : @xmath5 where @xmath6 these authors also found that @xmath1 in @xmath7, depends on @xmath8 as : @xmath9^{1/2 } } \label{vopt}\]] the same occurs with the optical radius, which depends on luminosity through the expression : @xmath10 kpc, which we will also use..galaxy characteristics dependent on the total mass. [cols=">,<,^,^,^,^,^,^ ",] in table [grid_m] we show the characteristics obtained with these equations for 44 different values of @xmath8. in column (1) we give the number of the radial distribution, defined by the value of @xmath8, given in column (2). the optical radius, @xmath11, and the virial radius, @xmath12, are given in columns (3) and (4) respectively. we have defined a characteristic radius for each galaxy, which we will use as our reference radius, as @xmath13. this radius, in kpc, is given in column (5). column (6) gives the rotation velocity, in @xmath14, reached at a radius @xmath15 kpc (see pss96 for details). the total mass of the galaxy, calculated with the classical expression @xmath16 @xcite, in units of 10@xmath17 m@xmath18, is given in column (7), and, finally, the characteristic collapse time scale, in gyr, which will be described below, is given in column (8). each galaxy is divided into concentric cylindrical regions 1 kpc wide. from the corresponding rotation curve we calculate the radial distributions of total mass, m(r), with the expression @xmath19. from these distributions of the total mass, we easily obtain the one corresponding to each of the cylinders, @xmath20. both distributions @xmath21 and @xmath20 are shown in fig. the total mass includes the dark matter component (dm) which, in principle, does not take part in the chemical evolution. however, the dm contribution seems to be negligible for the large massive galaxies, mostly in the regions where the chemical evolution is calculated. according to (*??? * ; *??? * and references therein) 75% of the spiral galaxies are well fitted without a dark matter halo and the failure to reproduce an other 20% is directly related to the existence of non - axisymmetric structures (bars or strong spiral arms). the gas computed following the previous section collapses to fall onto the equatorial plane forming the disc as a secondary structure. the gas infall from the halo is parametrised by @xmath22, where @xmath23 is the gas mass of the halo and @xmath24 is the infall rate, that is, the inverse of the collapse time scale @xmath25. the collapse time - scale for each galaxy depends on its total mass through the expression : @xmath26 @xcite, where @xmath27 is the total mass of the galaxy in 10@xmath28 (column 7 of table [grid_m]) and @xmath29 is its age, assumed to be 13.2 gyr in all cases. we calculate a characteristic collapse time scale @xmath30 for each galaxy from the ratio of its total mass, @xmath31, and the mwg, @xmath32 : @xmath33 where @xmath34 is the collapse time scale at the characteristic radial region for the mwg, @xmath35 kpc. this collapse time scale (@xmath36 gyr) was determined from the corresponding value for the solar neighborhood, @xmath37 gyr. this value for @xmath38 is very similar to that found in other standard galactic chemical evolution models @xcite and it is constrained by a large number of data, such as the [o / fe] _ vs _ [fe / h] relation for stars in the halo and the disc, the present infall rate, (0.7 m@xmath18pc@xmath39gyr@xmath40, see *??? *), or the g - dwarf metallicity distribution @xcite, all of them well reproduced in our model with this long scale to form the disc (see *??? we would like to emphasize that the characteristic collapse time - scale computed with equation [tau] is not a free - fall time (@xmath41 gyr for the mwg), being much longer than that since it has been calculated through the calibration with the solar neighbourhood collapse time scale @xmath38. the mass which does not fall onto the disc will remain in the halo, and yields a ratio @xmath42 for the baryonic component which is also in agreement with observations. the relative normalization of the halo, thick and thin disc surface mass densities @xcite gives an approximated proportion of 1:22:200, which implies that the halo surface mass density must be about 1/100 that of the disc component, also in agreement with the ratio obtained by the multiphase model. [tcoll] a) shows the characteristic time - scales _ vs _ the rotation velocity @xmath43 for our models. solid dots represent the values used in our previous models for individual spiral galaxies @xcite. an important consequence of the hypothesis linking the collapse time scale with the total mass, is that low mass galaxies take more time to form their discs, in apparent contradiction with the standard hierarchical picture of galaxy formation. this characteristic is, however, essential to reproduce most observational features of spiral and irregular galaxies (see also *??? in fact, recent self - consistent hydrodynamical simulations in the context of a cosmological model @xcite show that a large proportion of massive objects are formed at early times (high redshift) while the formation of less massive ones is more extended in time, thus simulating a modern version of the monolithic collapse scenario (*??? furthermore, this assumption is supported by various sets of observations @xcite which seem to demonstrate that a large proportion (@xmath44 %) of the massive galaxies formed their stars at @xmath45 while a much smaller proportion of the less massive ones have converted their baryon mass into stars at that redshift. it is evident that the collapse time - scale must vary with galactocentric radius. assuming that the total mass surface density follows the surface brightness exponential shape, the required collapse time - scale should also depend exponentially on radius, with a scale length @xmath46 (where @xmath47 is the corresponding scale length of the surface brightness radial distribution). thus, we assume : @xmath48 in principle, we might expect that @xmath49. however, in @xcite we have shown that such large scale length for the infall rate produces a final radial variation for the elemental abundances in disagreement with the observed distributions in spiral galaxies. (see *??? * for a wide discussion about the effect of this parameter on the radial distribution of abundances). on the other hand, we must bear in mind that the surface brightness distribution is the final result of the combination of both the collapse and the star formation processes, and therefore the collapse time - scale may have in principle a different dependence on radius than the surface brightness itself. for the sake of simplicity, we assume a scale length @xmath50, corresponding to half the scale length of the exponential disc @xmath51 given by pss96. the value of @xmath52 decreases with the mass of the galaxy, in agreement with observations (*??? * ; *??? * ; *??? * see their fig. 4). the radial dependence of the infall rate is not imposed _ a priori _ in our scenario, but it is consequence of the gravitational law and the total mass distribution in the protogalaxy. the physical meaning is clear : galaxies begin to form their inner regions before the outer ones in a classical inside - out scheme. this halo - disc connection is crucial for the understanding of the evolution of a galaxy from early times, the inside - out scenario being essential to reproduce the radial gradient of abundances @xcite. in fact, in a chemo - dynamical model @xcite, this scenario is produced naturally. the radial variation of the collapse time - scale for each galaxy, calculated with equation [tau_r], is shown in fig. [tcoll]b), where we also draw a solid line at 13 gyr, the assumed age of galaxies. if the collapse time - scale is larger than this value, there is not enough time for all the gas to fall onto the disc : only a small part of it has moved from the halo to the equatorial plane and the disc formation is not yet complete. this could correspond to the situation observed by @xcite, who have found an extended component of hi, different from the cold disc, located in the halo, rotating more slowly than the disc and with radial inward motion. the model computes the time evolution of each population which inhabits the galaxy. in the various regions of the disc or bulge and in the halo, which are treated separately, we allow for different phases of matter aggregation : diffuse gas (@xmath53), clouds (@xmath54, except in the halo), low - mass stars (@xmath55), high - mass stars (@xmath56), and stellar remnants.] the value for stellar mass range division is related to nucleosynthesis prescriptions : stars with masses lower than 4 m@xmath18 only produce light elements, and do not contribute to the interstellar medium enrichment., allows a very easy comparison of our resulting metallicity distribution with the observed one, based on g - dwarf low mass stars.] the mass in the different phases of each region changes by the following conversion processes, related to the star formation and death : 1. star formation by the gas - spontaneous fragmentation in the halo 2. cloud formation in the disc from diffuse gas 3. star formation in the disc from cloud - cloud collisions 4. induced star formation in the disc _ via _ massive star - cloud interactions 5. diffuse gas restitution from these cloud and star formation processes in the halo, devoid of molecular clouds, the star formation follows a schmidt law for the diffuse gas @xmath23 with a power @xmath57 and a proportionality factor @xmath58. in the disk, stars form in two steps : first, molecular clouds, @xmath59, form out of the diffuse gas, @xmath60, also by a schmidt law with @xmath57 and a proportionality factor called @xmath61. then, cloud - cloud collisions produce stars by a spontaneous process at a rate proportional to a parameter @xmath62. moreover, a stimulated star formation process, proportional to a parameter @xmath63, is assumed. an advantage of using our model is that it includes a more realistic star formation than classic chemical galactic evolution ones in which the sfr prescriptions are based on a schmidt law, depending on the total gas surface density. instead, the multiphase model assumes a star formation which takes place in two - steps : first, the formation of molecular clouds ; then the formation of stars. this simulates a power law for the gas density with an exponent @xmath64 and with a threshold gas density as shown by @xcite, and, more importantly, it allows the calculation of the two different gas phases present in the interstellar medium. in fact, the actual process of star formation, born by observations @xcite, is closer to our scenario with stars forming in regions where there are molecular clouds, than to the classical schmidt law which depends only on the total gas density. our assumed sfr implies that some feedback mechanisms are included naturally and are sufficient to simulate the actually observed process of creation of stars from the interstellar medium. the formed massive stars induce the creation of new ones. but, at the same time, these star formation processes also may destroy the diffuse or molecular clouds, thus preventing the total conversion of gas into stars and ejecting more gas once again into the ism point (v). in particular, massive stars destroy the molecular clouds that surround them, due to the sensitivity of molecular cloud condensation to the uv radiation @xcite. this mechanism restores gas to the ism, thus decreasing the star formation. both regulating process are included in our model, although neither heating or cooling mechanisms for the cloud components are included explicitly in our code. the complete set of equations is given in @xcite for each radial region. we summarize here only those related to the star formation processes in the halo, @xmath65, and in the disk, @xmath66 : @xmath67 where h and d indicate halo and disk. @xmath58, @xmath61, @xmath62 and @xmath63, besides the previously described @xmath25 are the parameters of the model. in the classical method of application of a chemical evolution model to a given region or galaxy, the input parameters are considered as free and chosen as the best ones in order to reproduce the selected observational constraints of the galaxy. in our models, however, not all these input parameters can be considered as free. the parameter @xmath24 is defined by the total mass radial distribution, as already explained. regarding @xmath58, @xmath61, @xmath62 and @xmath63, we have tried to reduce to a minimum their degree of freedom. their radial dependence may be estimated as (*??? * see) : @xmath68 where g is the universal gravitational constant, @xmath69 and @xmath70 are the halo and disk volume of each radial region,.] @xmath71 is the average cloud density and @xmath72 is the average mass of massive stars. the proportionality factors are called efficiencies that represent probabilities associated with these processes. in this way, the free parameters @xmath58, @xmath61, @xmath62 and @xmath63, proportionality constants in our star and cloud formation laws, but variable for each radial region, are computed through the efficiencies, which represent the efficiencies of star formation in the halo, @xmath73, cloud formation, @xmath74, cloud cloud collision, @xmath75, and interaction of massive stars with clouds, @xmath76, in the disc. only a given efficiency for each process must be selected for the whole galaxy, although the original parameters, and its corresponding processes, maintain their radial dependence. the term associated to the induced star formation describes a local process and, as a result, its coefficient @xmath76 is considered independent of the galaxy modelled and the location in it. the term @xmath73 is also assumed constant for all haloes. therefore, for all our 440 models, both efficiencies take the same values already used in our previous model for the mwg. only the other two efficiencies @xmath74 and @xmath75 are allowed to vary for each galaxy, being characteristic for each one of them. within each galaxy @xmath75 and @xmath74 are independent of the position. the fact that galaxies with the same gravitational potential or mass but different morphological type or appearance exist, implies that the evolution of a galaxy does not depend solely on gravitation, even if this may be the most important factor, but also on certain dynamical conditions. these conditions can not be taken into account, obviously, in a simple chemical evolution model, but may change the evolution of a galaxy, (mostly the star formation rate through temperature variations). we may consider them as included in our efficiencies to form molecular clouds and stars, @xmath74 and @xmath75, which are allowed to change from one galaxy to another. we assume that @xmath74 and @xmath75 vary between 0 and 1, according with its efficiency meaning. in principle both parameters should be allowed to vary independently from each other, what would increase very much the number of models to be calculated. furthermore, some of those models will not be physically possible. on the other hand, on the basis of our previous works, a trend between both efficiencies seems to exist, increasing or decreasing together for a given galaxy respect to the values appropriate for the mwg model. we have therefore studied the observational data related to molecular cloud and star formation from the available gas in order to check if a correlation between these two processes exists. to this aim we have used the data from @xcite which correspond to a large sample of galaxies and refer to atomic and molecular gas masses and to luminosities in the ir band and in h@xmath77 emission, two well known indicators of the star formation rate. with them, we have tried to establish if some correlation appears among our efficiencies, @xmath74 and @xmath75. since the modelled star formation rate have two steps to form stars, we will analyse the transformation of diffuse gas in molecular gas and of the conversion of the molecular clouds into stars. from equations (10) and (11), our parameters @xmath74 and @xmath75, efficiencies of transforming the diffuse gas in molecular clouds and forming stars out of them, can be written as : @xmath78 while from equations (7) and (8) we have : @xmath79 in these expressions the subscript @xmath80 has been omitted, since in all cases we refer to the disc. but the second term is much smaller than the first and therefore we can approximate @xmath81 we may approximate @xmath82, where @xmath83 is the mean time necessary to transform the diffuse gas in a molecular cloud, in units of @xmath84 : @xmath85 where @xmath86 and @xmath87 are the total mass in diffuse and molecular gas in units of @xmath88, and sfr is the star formation rate in units of @xmath89. therefore : @xmath90 sfr is usually estimated from the @xmath91 luminosity @xcite : @xmath92 and the volume of the disc is @xmath93 where r is in kpc and @xmath94 is in @xmath95. we have all data to estimate these efficiencies, except @xmath96. recent estimates @xcite for this cloud accumulation time scale give values several times the gravitational contraction scale (which is @xmath97 1 - 4 myr), that is @xmath98 myr, and probably smaller than 50 myr. the probable range is @xmath99 $] yr. we have given three possible values : @xmath100, @xmath101 and @xmath102 yr in order to take into account other possible slower modes @xcite. in fig. [ratio] we represent the ratio @xmath103 computed for @xmath104 as a function of galaxy morphological type. single galaxies are represented as @xmath105 while solid dots correspond to average values obtained by binning them into 11 types. a constant ratio for all types is consistent with the data points, shown by the solid line and giving a mean value @xmath106 with @xmath107. to 10.] taking into account the large range of variation of @xmath74 and @xmath108 (more than 5 orders of magnitude) for the whole set of data, it is surprising that the @xmath109 ratio takes values around 0.35 for all galaxies. for other values of @xmath96, this value changes but remains constant. they are shown in the graph as dotted lines at values 0.00 (@xmath110 myr) and 0.80 (@xmath111 myr). taking this into account, we have assumed a ratio @xmath109= 0.4 for our computed models. ccc n & @xmath112 & @xmath75 + 1 & 0.95 & 0.88 + 2 & 0.80 & 0.57 + 3 & 0.65 & 0.34 + 4 & 0.45 & 0.14 + 5 & 0.30 & 0.05 + 6 & 0.15 & 1.0e-2 + 7 & 0.075 & 1.5e-3 + 8 & 0.037 & 2.6e-4 + 9 & 0.017 & 3.7e-5 + 10 & 0.007 & 4.0e-6 + we have computed 10 models for each mass radial distribution, allowing @xmath74 to take values between 0 to 1 as given in table [efi]. for each one of these efficiencies,@xmath75 has been fixed according to the ratio @xmath113 and their values are also given in the table. each set of efficiencies has been labelled by a number n from 1 to 10. _ summarizing, only the characteristic collapse time - scale, depending on the total mass, and the set of efficiencies, denoted by number n, are varied from model to model. _ we thus obtain 440 different models which represent all possible combinations of the collapse time - scale with the values of n.
Results
the results corresponding to the mass of each region and phase, the star formation rate and the supernova rates, for the 440 computed models are shown in tables [phases], and [sfrs]. the elemental abundances for the discs are shown in table [abundances]. here we only give, as an example, the results of the model corresponding to radial distribution number 22 and @xmath114, with a rotation velocity of @xmath97 140 km.s@xmath40, for the first and last time steps of the evolution. the whole tables with the complete time evolution from 0 to 13 gyr, with a time step of 0.5 gyr, for the whole set of models will be available in electronic form at cds via anonymous ftp to cdsarc.u-strasbg.fr (130.79.128.5) or via http://cdsweb.u-strasbg.fr/abstract.html, or http:/wwwae.ciemat.es/@xmath115mercedes / grid. rrccccccc time & r & mtot & mdisc & mgas(hi) & mgas(h@xmath116) & mstars(m@xmath117) & mstars(m@xmath118) & mremnants + (gyr) & (kpc) & (10@xmath28) & (10@xmath28) & (10@xmath28) & (10@xmath28) & (10@xmath28) & (10@xmath28) & (10@xmath28) +... &... &... &... &... &... &... &... &... + 0.1 & 14. & 0.69e+01 & 0.47e04 & 0.47e04 & 0.89e08 & 0.40e17 & 0.31e18 & 0.21e19 + 0.1 & 12. & 0.69e+01 & 0.22e03 & 0.22e03 & 0.11e06 & 0.65e15 & 0.49e16 & 0.34e17 + 0.1 & 10. & 0.68e+01 & 0.10e02 & 0.10e02 & 0.14e05 & 0.13e12 & 0.93e14 & 0.65e15 + 0.1 & 8. & 0.67e+01 & 0.48e02 & 0.47e02 & 0.20e04 & 0.28e10 & 0.21e11 & 0.14e12 + 0.1 & 6. & 0.66e+01 & 0.22e01 & 0.22e01 & 0.31e03 & 0.75e08 & 0.56e09 & 0.39e10 + 0.1 & 4. & 0.58e+01 & 0.91e01 & 0.87e01 & 0.47e02 & 0.21e05 & 0.16e06 & 0.11e07 + 0.1 & 2. & 0.29e+01 & 0.21e+00 & 0.17e+00 & 0.39e01 & 0.23e03 & 0.17e04 & 0.13e05 +... &... &... &... &... &... &... &... &... + 13.2 & 14. & 0.69e+01 & 0.58e02 & 0.46e02 & 0.13e02 & 0.10e04 & 0.22e07 & 0.70e06 + 13.2 & 12. & 0.69e+01 & 0.27e01 & 0.16e01 & 0.10e01 & 0.89e03 & 0.16e05 & 0.67e04 + 13.2 & 10. & 0.68e+01 & 0.13e+00 & 0.54e01 & 0.43e01 & 0.28e01 & 0.31e04 & 0.25e02 + 13.2 & 8. & 0.67e+01 & 0.57e+00 & 0.14e+00 & 0.11e+00 & 0.29e+00 & 0.21e03 & 0.32e01 + 13.2 & 6. & 0.66e+01 & 0.23e+01 & 0.26e+00 & 0.19e+00 & 0.16e+01 & 0.79e03 & 0.21e+00 + 13.2 & 4. & 0.58e+01 & 0.49e+01 & 0.18e+00 & 0.18e+00 & 0.39e+01 & 0.88e03 & 0.65e+00 + 13.2 & 2. & 0.29e+01 & 0.29e+01 & 0.19e01 & 0.51e01 & 0.23e+01 & 0.95e04 & 0.50e+00 + in table [phases] we list in column (1) the time, in gyr and in column (2), the galactocentric distance, in kpc, in column (2). columns (3) to (9) give the mass, in units of 10@xmath119, in each region and phase : column (3) the total mass in each region ; column (4) the mass of the disc region ; column (5) the mass in the diffuse gas phase;column (6) the molecular gas ; column (7) the mass in low and intermediate mass stars ; column (8) the mass in massive stars and, finally, column (9) the mass in remnants. table [sfrs] gives, for each time step in gyr and radial distance in kpc, listed in columns (1), and (2) respectively, the star formation rate, in units of @xmath120, in the disc and the halo regions in columns (3) and (4) ; the supernova rates, for types ia, ib and and ii, for the disc in columns (5), (6) and (7) and for the halo in columns (8),(9) and (10) all of them in units of 100 yr@xmath40. the abundances in the disc for 14 elements are given in table [abundances] for each time step in gyr and radial distance in kpc (columns (1), and (2)).the abundances by mass of : h, d, @xmath121he, @xmath122he, @xmath123c, @xmath124c, n, o, ne, mg, si, s, ca, and fe are given in columns (3) to (16). the time evolution of several models is shown in the next figures. in each one, 4 panels are shown, corresponding to 4 different maximum rotation velocities and/or radial distribution of masses. we have selected the values corresponding to @xmath1250.03, 0.19, 1.0 and 2.5 corresponding to galaxies with rotation velocities of 48, 100, 200 and 290 @xmath126, respectively, and representing typical examples of spiral and/or irregular galaxies. for each panel we show the results for the 10 selected values of the efficiencies (or equivalently 10 rates of evolution) from @xmath127, corresponding to the highest efficiency values and hence the most evolved models, to @xmath128, with the smallest values and thus the least evolved ones. we have computed models for galactocentric radii up to @xmath129 with a step @xmath130 which depends on total galactic, being larger (up to 4 kpc) for the most massive modelled galaxies and smaller (only 1kpc) for the lowest mass ones. results are shown only for the region located at the galactocentric distance closest to the characteristic radius @xmath131 defined in table [grid_m]. [hit_rc] shows the time evolution of the diffuse gas density. in all cases, the diffuse gas surface density is seen to increase rapidly at early times and then declines slowly. the first abrupt increase is a consequence of the infall rate of gas from the halo and hence does not depend on the model efficiencies. the increase is faster for more massive galaxies. the later decline is due to gas consumption in the process of molecular gas formation and hence depends on the efficiency @xmath74, producing lower densities for the galaxies with higher efficiencies. fig.[h2t_rc] shows the time evolution of the molecular gas phase whose formation is delayed respect to that of the diffuse gas. it is clear that the maximum density is reached later than that corresponding to the diffuse gas density. for instance, for @xmath132, @xmath133 gyr for @xmath134 and 200 km s@xmath40, while it is 0.3 gyr for @xmath135 km s@xmath40 and reaches @xmath136 gyr when @xmath137 km s@xmath40. rrrrrrrrrr time & r & sfr(disc) & sfr(halo) & sn - ia (disc) & sn - ib (disc) & sn - ii (disc) & sn - ia (halo) & sn - ib (halo) & sn - ii(halo) + (gyr) & (kpc) & (m@xmath138) & (m@xmath138) & (@xmath139) & (@xmath139) & (@xmath139) & (@xmath139) & (@xmath139) & (@xmath139) +... &... &... &... &... &... &... &... &... &... + 0.1 & 14. & 0.26e15 & 0.51e01 & 0.17e17 & 0.14e18 & 0.42e14 & 0.45e02 & 0.11e01 & 0.15e+01 + 0.1 & 12. & 0.43e13 & 0.52e01 & 0.27e15 & 0.14e16 & 0.69e12 & 0.46e02 & 0.11e01 & 0.16e+01 + 0.1 & 10. & 0.82e11 & 0.54e01 & 0.52e13 & 0.25e14 & 0.13e09 & 0.47e02 & 0.12e01 & 0.16e+01 + 0.1 & 8. & 0.18e08 & 0.57e01 & 0.12e10 & 0.55e12 & 0.30e07 & 0.51e02 & 0.12e01 & 0.17e+01 + 0.1 & 6. & 0.49e06 & 0.63e01 & 0.31e08 & 0.15e09 & 0.80e05 & 0.55e02 & 0.14e01 & 0.19e+01 + 0.1 & 4. & 0.13e03 & 0.61e01 & 0.89e06 & 0.44e07 & 0.22e02 & 0.55e02 & 0.14e01 & 0.18e+01 + 0.1 & 2. & 0.14e01 & 0.28e01 & 0.11e03 & 0.60e05 & 0.24e+00 & 0.26e02 & 0.65e02 & 0.84e+00 +... &... &... &... &... &... &... &... &... &... + 13.2 & 14. & 0.52e05 & 0.45e01 & 0.13e05 & 0.56e05 & 0.16e03 & 0.18e01 & 0.54e01 & 0.14e+01 + 13.2 & 12. & 0.38e03 & 0.46e01 & 0.10e03 & 0.42e03 & 0.11e01 & 0.18e01 & 0.54e01 & 0.14e+01 + 13.2 & 10. & 0.74e02 & 0.46e01 & 0.23e02 & 0.86e02 & 0.22e+00 & 0.18e01 & 0.55e01 & 0.14e+01 + 13.2 & 8. & 0.50e01 & 0.44e01 & 0.18e01 & 0.59e01 & 0.15e+01 & 0.18e01 & 0.53e01 & 0.13e+01 + 13.2 & 6. & 0.19e+00 & 0.29e01 & 0.73e01 & 0.22e+00 & 0.56e+01 & 0.13e01 & 0.35e01 & 0.86e+00 + 13.2 & 4. & 0.21e+00 & 0.25e02 & 0.11e+00 & 0.25e+00 & 0.62e+01 & 0.24e02 & 0.31e02 & 0.74e01 + 13.2 & 2. & 0.22e01 & 0.17e07 & 0.29e01 & 0.28e01 & 0.68e+00 & 0.14e03 & 0.27e07 & 0.51e06 + fig. [sfrt_rc] shows the star formation rate history for the same cases as before. these histories result extraordinary different, even for equal efficiencies. taking into account that the radial region shown is equivalent in all galaxies, it indicates that the primary agent driving the time evolution of the sfr is the galaxy mass through the collapse time scale. on the other hand, the resulting star formation history in most models, is smooth, showing values larger than 10@xmath140 only at the early evolutionary phases of the most massive galaxies with very high efficiencies. to 220 mm landscape table 5 to go here. [abundances] the time evolution of the oxygen abundance relative to hydrogen, expressed as 12+log(o / h), for the models described above is shown in fig.[oht_rc]. present time oxygen abundances look very similar for all the galaxies with @xmath141. the models with low efficiencies, @xmath142, show very low abundances although always larger than 12+log(o / h)= 7, about the lowest observed oxygen abundance in galaxies. fig.[ofe_rc] shows the evolution of [o / fe] when [fe / h] increases. the usual relation for the mwg can be taken to be represented by the @xmath143 line from panel c), showing almost a _ plateau _ for low metallicities (@xmath144<1.5 $]) and then a decline toward the solar value. the _ plateau _ appears in the models corresponding to massive massive galaxies with high efficiencies @xcite. the less massive galaxies show a slow decline without _ plateau _ for low values of n, and a steeper decrease for higher values. thus, we may expect that [o / fe] in the less evolved galaxies will soon reach the solar value @xcite. we now analyse the present time results obtained with this bi - parametric grid of models. in order to do this, we start presenting only the present time radial distributions of gas, oxygen abundances and star formation rate which we will analyse in the following subsections. the radial distribution of gas and elemental abundances provide two basic model constraints since they are easily extracted from observations. in addition, in many cases, the star formation surface density and/or the surface brightness can also be derived. the radial distribution of atomic gas surface density for the present time is shown in fig. we can see that the atomic gas surface density shows a maximum somewhere along the disc, as it is usually observed (*??? * ; *??? * ; *??? * and see fig.[chequeo]). the value of this maximum depends on n : the models with the highest efficiencies have smaller gas quantities and their maximum densities are around 3 - 4 m@xmath145. for intermediate efficiencies (@xmath146), these maximum values rise to @xmath147 m@xmath145. for all efficiencies the radial distributions are very similar independently of their galactic mass, except for those corresponding to @xmath148 (@xmath149) which show much lower densities, except in the central region. in fact, a characteristic shown by all distributions is the similarity among models of the same efficiencies but different total mass, only scaled by their different optical radius. with the exception of the @xmath150 model, all the others show, for a same n, differences small enough as to simulate a dispersion of the data. for the less evolved theoretical galaxies (n @xmath151), the neutral gas distribution display a clear dependence on galactic mass. the maximum density values are always large, as expected, due to the small efficiencies to form molecular clouds, which do not allow the consumption of the diffuse gas, but this maximum density increases from @xmath152 for @xmath150 (@xmath153) to @xmath154 for @xmath155 (@xmath156) and up to @xmath157 for @xmath158 (@xmath159). the consequence of a shorter collapse time - scale for the more massive spirals is clearly seen : the maximum is located at radii further away from the centre due to the exhaustion of the diffuse gas in the inner disc which moves the star formation outside. the smaller the galaxy mass, the closer to the centre is the maximum of the distribution, which resembles an exponential, except for the inner region. in fact, a shift in the maximum appears in each panel. in some cases, however, the low values of the surface gas density are due to the fact that the gas did not have enough time to fall completely onto the equatorial disc. this effect is very clear for the model with @xmath150 which shows a very steep distribution with gas densities lower than 15 @xmath160. in this case the gas shows a radial distribution with a maximum at the centre. the low density out of this central region is not due to the creation of stars but to the long collapse time - scale which prevents the infall of the sufficient gas to be observed. besides the variations due to the differences in total mass, which correspond to different collapse time - scales, and because we have selected different efficiencies to form stars and molecular clouds, a same total mass may have produced discs in different evolutionary states. thus, a same @xmath161 may result in a disc of 7 - 8 kpc and atomic gas densities around 5 @xmath160, or a disc of only 5 kpc with a maximum density of 40 @xmath160 in the region of 2 kpc. in the same way, a galaxy with a large value of the total mass, as the one with @xmath162, may show a high gas mass density and a little disc for a high value of n or, on the contrary, be very evolved and therefore show no gas and a large stellar disc if the value of n is low. the first object (@xmath163) could correspond to the case of low surface brightness galaxies, while the last ones (@xmath164) could be identified as the typical high surface brightness spiral galaxies. an important success of the multiphase models has been the ability to reproduce the radial distributions for the atomic gas and the molecular gas separately, which is possible due to the assumed star formation prescription in two steps, thus allowing the formation of molecular clouds prior to the appearance of stars. another important consequence of this sfr law is that it takes into account feedback mechanisms, even negative. if molecular clouds form before stars, this implies a delay in the time of star formation. the molecular gas shows an evolution similar to that of the diffuse gas, but with a certain delay. this delay allows the maintenance of a radial distribution with an exponential shape, as usually observed, for a longer time, although in some evolved galaxies @xmath165 is also consumed in the most central regions, thus reproducing the so - called central _ hole _ of the molecular gas radial distribution, observed in some galaxies @xcite, the mwg being one of them (see fig6b in next section). this is seen in fig.[h2] for efficiencies corresponding to @xmath166, (depending on the total mass), for which the model lines turn over at the inner disc, which corresponds to the regions located at the border between bulge and disc. thus, the galaxies with the highest efficiencies (@xmath167) show a maximum in their radial distribution of @xmath168, which is always closer to the centre than that of the atomic gas distribution. galaxies with the lowest efficiencies (@xmath166) show larger surface densities of molecular than atomic gas because the efficiency to form stars from molecular clouds is smaller than the efficiency to form these clouds. therefore, the conversion of diffuse to molecular gas occurs more rapidly that the subsequent formation of stars. the total mass converted into stars forms out the stellar disc in each galaxy. these stellar discs are represented in fig.[profiles] as the stellar surface density radial distributions. these profiles may be compared with brightness surface radial distributions, after the appropriate conversions. the values of efficiencies have a small influence in the resulting stellar surface density distribution shape : the total mass of stars created is similar for all n @xmath169, although they are formed at different rates, that is the resulting stellar populations have different mean ages. for the most evolved cases, most stars were created very rapidly, while for the less evolved ones, stars formed later as average, as it is shown in fig. [sfrt_rc] for the characteristic radius region. therefore, the radial distributions of surface brightness would result very similar for galaxies of a given galactic total mass, but colors are expected to be different, redder for the galaxies with the highest efficiencies for the formation of stars and molecular clouds. a very interesting result is that the central value of the distribution is practically the same, around 100 m@xmath145, for all rotation curves and efficiencies, in agreement with freeman s law. only galaxies with the smallest efficiencies or the less massive discs show central densities smaller than this value. we can not compute a surface brightness only with these models, but assuming a ratio @xmath170 for the stellar populations, this implies a surface luminosity density of @xmath171 and scale lengths in agreement with observed generic trends. only in models for n@xmath172 the stellar discs show a different appearance : they look less massive, as corresponding to discs in the process of formation. this implies that the surface brightness is lower than for the other types for a similar characteristic total mass. in any case, we remind that all information related to photometric quantities, and this also applies to the disc scale lengths, which may be obtained from the star formation histories and enrichment relations of models, need the application of evolutionary synthesis models, what is out of the scope of this work. one of the most important results of this grid of models refers to the oxygen abundances, shown in fig. a radial gradient appears for most of models. this is due to the different evolutionary rates along radius : the inner regions evolve more rapidly that the outer ones, thus steepening the radial gradient very soon for most model galaxies. then, the radial gradient flattens for the more massive and/or most evolved (small n) galaxies due to the rapid evolution, even in the outer regions, which produces a large quantity of elements, with the oxygen abundance reaching a saturation level. this level is found to be around @xmath173 dex. observations in the inner disc of our galaxy support this statement @xcite. this result was already found in our previous works. moreover, the larger the mass of the galaxy, the faster the effect : a galaxy with @xmath174 km.s@xmath40 has a flat radial gradient for efficiencies corresponding to n@xmath175 1 or 2, while a galaxy with v@xmath176 km.s@xmath40, shows a flat distribution for @xmath177. the less massive galaxies maintain a steeper radial distribution of oxygen for almost all efficiencies, with very similar values of the gradients. nevertheless, for any galaxy mass, if @xmath178 the radial abundances distribution are flat. thus, the less evolved galaxies show no gradient, such it is observed in lsb galaxies @xcite, the intermediate ones show steep gradients, and the most evolved galaxies have, once again, flat abundance radial distributions. the largest values of radial gradients correspond to the intermediate evolutionary type galaxies, with the limiting n varying according to the total mass of the galaxy. the more massive galaxies only show a significant radial gradient if n@xmath179 while the less massive ones have a flat gradient only if @xmath180 with the rest having very pronounced radial gradients even for @xmath181. for the evolved galaxies, the characteristic efficiencies are high for all the disc, thus producing a high and early star formation in all radial regions. in this case, the oxygen abundance reaches very soon a saturation level, flattening the radial gradient developed at early times of the evolution. the characteristic oxygen abundance, measured at r@xmath182, is higher for the more massive galaxies and lower for the less massive ones. however, this correlation is not apparent when the central abundance is used, due to the existence of the saturation level in the oxygen abundance, which produces a flattening of the radial gradient in the inner disc, even for galaxies with intermediate efficiencies. actually, this saturation level would correspond to the integrated yield of oxygen for a single stellar population. as oxygen is ejected by the most massive stars of a generation of stars, it appears rapidly in the ism. it is therefore impossible to surpass this level of abundance. in fact, the oxygen abundance radial distribution shows sometimes a bad fit to observations in the central parts of the discs, which give frequently abundances larger than 9.10 dex. this absolute value of the oxygen abundance is not reached in any case by the models. all the computations performed within the multiphase approach reach a maximum @xmath183 dex which no model can exceed. we might think that uncertainties in the calculations of the stellar yields elements, still very dependent on the assumed evolution of stars, are a possible reason for this saturation level appears in theoretical models. however, low and intermediate mass stars, for which yields show large variations among the different groups (see *??? * and references therein), depending on the assumed hypothesis in the calculation of the stellar evolution in the latest stellar evolutionary phases, do not produce oxygen ; these calculations only affect to n and c abundances. on the other hand, the most important uncertainty for the production of oxygen in massive stars refers to the strength of stellar winds responsible for stellar mass loss. this mass loss implies less production of oxygen and a larger ejection of carbon. in our models we have used woosley & weaver s yields which do no include stellar winds. therefore, from this point of view, the modelled oxygen abundances must be considered upper limits. it should be recalled, however, that all oxygen data yielding values larger than 9.1 dex have been obtained from observations of hii regions where the electronic temperature could not be measured, and hence the oxygen abundances have been derived through empirical calibrations which are very uncertain in the high abundance regime. actually, the shape of the radial distribution of oxygen changes depending on the calibration used (e.g. *??? the suspicion that these abundances are overestimated at least by 0.2 dex is very reasonable @xcite. abundance estimates in hii regions historically considered as metal - rich, result to be almost solar once the electronic temperature has been finally measured @xcite. radial distributions of star formation rate surface density show an exponential shape in the outer disc, but a less clear one in the inner regions, where some models show a distribution flatter than that of the molecular gas, and even decreasing toward low values of the star formation rate in the centre. these sfr radial distributions are very similar to those estimated by @xcite from h@xmath77 fluxes, as we show in fig.[chequeo] for a few galaxies. the star formation rate assumed in our models do not produce bursts of massive stars in the low mass galaxies whatever the efficiencies. only the massive galaxies are able to keep a large quantity of gas in a small region, usually at the centre although sometimes at the inner disc regions. on the contrary, low mass galaxies collapse very slowly, and thus the star formation rate maintains a low level during the whole life of the galaxy. in fact, recent works suggest the same scenario for both low mass and low surface brightness galaxies @xcite in order to take into account the observed data. our resulting abundances and gas fractions for low mass galaxies seem to be in rough agreement with these findings @xcite, although our model results can not still be compared with photometric data, for which many more observational sets exist.
Discussion
the first application of any theoretical model is to check its validity for the mwg. a large set of observational data for the solar neighbourhood and the galactic disc exists and therefore the number of constraints is large compared to the number of free parameters of the computed models. the model for @xmath143, and total mass distribution number 28, corresponding to @xmath184 and maximum rotation velocity @xmath185 km s@xmath40, is the model more representative of the mwg. the results corresponding to this model are shown in figs. [mwg1] and [mwg2] together with the available data. [mwg1] shows the time evolution of the star formation history, panel a), and the age - metallicity relation, panel b), of the region located at r@xmath1758 kpc from the galactic centre, as compared to data for the solar neighbourhood. the observational trends are reproduced adequately, although the model predicted maximum of the star formation rate appears slightly displaced toward earlier times with respect to observations. [mwg2] shows the present time radial distributions of diffuse and molecular gas, mass of stars, star formation rate and oxygen and nitrogen abundances for the galactic disc. the diffuse gas radial distribution, panel a), is well reproduced in shape, although the maximum observed gas surface density is somewhat displaced to outer radii compared with the oldest data. it, however, fits well the most recent data obtained from @xcite for radii r@xmath186 kpc. regarding the molecular gas density, panel b), the distribution is quasi - exponential from @xmath187 kpc and decreases at the inner disc regions. taking into account that recent data give low densities at these inner regions, we consider that our model results may be adequate. the stellar mass distribution, panel c), is exponential in shape in agreement with the surface brightness distribution, and radial distributions, panels e) and f), reproduce those shown by the most recent data. the only feature which is not well reproduced by the model is the sfr radial distribution shown in panel d), which decreases toward the inner disc in apparent discrepancy with observations. the modelled star formation rate distribution has a maximum in @xmath188 kpc, while the observed one increases exponentially toward the galactic centre, or levels off at 3 - 4 kpc. on the other hand, the maximum of the atomic gas density is observed around 10 - 11 kpc, and the molecular gas density has its maximum observed at around 6 kpc. therefore, it results difficult to explain, from an observational point of view, how the star formation rate remains so high at the inner disc (inside the central 3 - 4 kpc), where both gas phases are already consumed. in fact, the recent data from @xcite show a decline for r@xmath189 kpc and a decreasing star formation rate for the inner disc regions is also observed in a large number of galaxies, as we will see in the next section. actually, the mwg radial distribution of the present star formation rate is still a matter of discussion @xcite since the detection of young sources, which are embedded in gas clouds, is difficult and might be affected by selection effects. data from pulsars, supernovae or ob star formation @xcite seem to indicate that the radial distribution of the sfr has a maximum around 5 kpc, decreasing toward both smaller and larger radii. on the other hand @xcite obtained a much flatter distribution from the observed cosmic rays and gamma radiation spectra. since a decrease at the inner disc is not unreasonable from the comparison with other spiral galaxy data, we consider that the model results for the mwg are acceptable. there is only a small sample of galaxies for which large observational data sets, including neutral and molecular gas distributions, exist. in what follows we show a comparison of our model results with the data corresponding to theses galaxies. lccccccccccc galaxy & t & type & d & vel@xmath190 & mass distr. & @xmath30 & r@xmath182 & n & @xmath191 & @xmath192 & @xmath193 + name & & class & (mpc) & (@xmath126) & number & (gyr) & (kpc) & & (gyr) & & + ngc 300 & 7 & scd / sd & 1.65 & 85 & 13 & 13.31 & 2.3 & 7 & 13.3 & 0.07 & 0.007 + ngc 598 & 6 & sc / scd & 0.84 & 110 & 21 & 8.13 & 2.9 & 6 & 10.3 & 0.05 & 0.005 + ngc 628 & 5 & sc & 11.4 & 220 & 31 & 3.43 & 7.7 & 5 & 3.28 & 0.25 & 0.01 + ngc 4535 & 5 & & 16.6 & 210 & 30 & 3.59 & 7.4 & 5 & 3.50 & 0.28 & 0.02 + ngc 6946 & 6 & scd & 7 & 180 & 25 & 4.94 & 6.2 & 6 & 4.26 & 0.18 & 0.02 + the characteristics and corresponding input parameters of this galaxy sample are given in table [sample]. for each galaxy, column (1), the morphological type index is given in column (2), while the classical hubble type is given in column (3). the adopted distance (taken from references following table [data], column (2) is in column (4) ; the maximum rotation velocity is given in column (5). the number of the radial distribution of total mass corresponding to the column (1) of table [grid_m] chosen to represent each galaxy is in columns (6). the characteristic collapse time and the characteristic radius corresponding to each distribution are given in columns (7) and (8). the number n chosen as the best one to reproduce the observations is in column (9). the last columns, (10) to (12) give, for comparison purposes, the collapse time - scale and efficiencies for molecular cloud and star formation used in our previous models @xcite for these same galaxies. the radial distributions of the different quantities : atomic and molecular gas densities, star formation rate and oxygen abundance, for these galaxies are shown in fig. [chequeo] together with the corresponding observational data, taken from the references given in table [data]. in the first row of panels of fig. [chequeo] we can clearly see that the radial distributions of neutral hydrogen are very well reproduced by the models. for each galaxy, the distribution shows a maximum along the disc, as predicted. also, for galaxies with similar total mass, such as ngc 4535 and ngc 6946, this maximum is higher for later morphological type. for galaxies with the same value of t, but different total mass, such as ngc 6946 and ngc 598, this maximum does not change its absolute value, but it is located at a different galactocentric distance, closer to the galaxy centre for the less massive one, due to the longer collapse time - scale which reflects on a slower evolution. we would like to point out that the agreement between model results and data is much improved when good quality data (usually the most recently published ones) are used. the selection of the best obtained distances improves extraordinarily this agreement, as well. the same applies to the rest of the panels in this figure. the radial distributions of molecular cloud surface density for each galaxy, except ngc 300 for which no data exist, are shown in the second row of fig. [chequeo]. the agreement between model results and observations is good for the outer discs which follow a quasi exponential distribution, but is not so good for the inner discs where the molecular hydrogen surface density is observed to continue increasing instead of turn over as predicted by the models. we might artificially decrease the efficiencies @xmath75 in these zones, instead to maintain them constant as we do, and thus recover the observed exponential shape. but in this case oxygen abundances will be smaller than observed in these same regions. on the other hand we should recall that the molecular masses are estimated from the co intensity through a calibration factor which depends on metallicity in a way which would produce smaller molecular gas densities than usually assumed at the inner galactic disc @xcite. taking into account that recent data yield low densities at these inner regions, we consider that our model results may represent adequately the reality. in any case, differences between models and data are larger than in the case of the diffuse gas, which is not unexpected, given the larger uncertainties involved in the derivation of molecular hydrogen masses. the radial distribution of the star formation rate for each galaxy is shown in the third row of fig [chequeo]. the agreement between model results and data is very good for the most massive galaxies for which both the maximum of the star formation rate and its location is well reproduced. for the less massive galaxies, the predicted central turnover of the distribution is not observed. in the last row of fig. [chequeo], the oxygen abundance radial distribution for each galaxy, except ngc 4535 for which no data exist, is shown. the oxygen radial gradient is reproduced in all cases and also the observed trend of steeper radial distributions larger radial gradients for the late type galaxies is well reproduced by the models. lccccc galaxy & d & h & h@xmath116 & sfr & [o / h] + ngc 300 & tu & rog, puc &... & deh & pag, chr, deh + ngc 598 & tp & new, cs & hey & hey, ken, rk & ka, vil + ngc 628 & mk & wev & al, nis, ry, ss & lr, mk, ry & br, mcc, zee + ngc 4535 & tp & cay, war & ky, nis, ry & ken, mk &... + ngc 6946 & we & car & ry, ty, wal & ken, mk, ry & br, mcc + al : @xcite ; br : @xcite ; car : @xcite ; chr : @xcite ; cay : @xcite ; cs : @xcite ; deg : @xcite ; deh : @xcite ; hey : @xcite ; ky : @xcite ; ken : @xcite ; ka : @xcite ; lr : @xcite ; mar : @xcite ; mcc : @xcite ; nn : @xcite ; new : @xcite ; pag : @xcite ; puc : @xcite ; rog : @xcite ; rk : @xcite ; ry : @xcite ; ss : @xcite ; ty : @xcite ; tu : @xcite ; tp : @xcite ; zee : @xcite ; vil : @xcite ; wal : @xcite ; war : @xcite ; wev : @xcite we would like to emphasize that the models shown in this section have not been computed specifically for each of the sample galaxies, as was done in our previous works. the models shown in fig.[chequeo] were selected among the 10 available ones for their corresponding rotation velocity following table[dis] of the grid. thus the good agreement found between model results and data demonstrate that our bi - parametric models are able to adequately reproduce real galaxies. the computed star formation rate reproduces the relation obtained by @xcite, when it is represented _ vs _ the total gas surface density as can be seen in fig. [ken]. in the two upper panels of the figure (@xmath194), we have over - plotted the data from @xcite and @xcite which are seen to fall in the locus defined by the models. models with @xmath142 seem to be out of the region where this data lie. however, due to their extremely low efficiencies, we may assume that these models do not simulate normal bright spiral galaxies, but other kind of objects with a low stellar content and high gas fractions more similar to dwarf and lsb galaxies. we have, therefore, taken data from @xcite on these kinds of objects and computed the densities for both quantities assuming an optical radius of 5 kpc for all of them. obviously, a change in this radial dimension would vary the final values of our estimates, but our hypothesis is probably valid within a factor of 2 (radii less than 10 kpc). under this assumption, we see that the points, shown in the third panel of fig.[ken], fall in the upper locus defined by the models. if the radius were smaller than assumed, the densities would be even higher, and the points would move in the figure following the direction given by the arrow. a second factor not included in these estimates, is the molecular gas content. there are some works suggesting that the molecular gas amounts to less than 10% in this kind of galaxies while other seems to indicate that the molecular mass may be as large as in the brightest massive spirals. in any case, the inclusion of this factor would move the points to the right. in both cases the data points would populate the region of the diagram occupied by the models. we therefore conclude that our models are able to reproduce the observed trend of sfr _ vs _ total gas surface density for different types of galaxies. the model computed abundance radial gradients for oxygen have been obtained by fitting a least squares straigh line to the oxygen abundances for radii @xmath195 kpc. the chosen radial range tries to eliminate the central region (where the oxygen abundance distribution flattens) and the outer region where there are no data. thus the calculated gradients may be compared more precisely with the corresponding observed ones. these radial gradients, measured as dex / kpc, are represented as a function of rotation velocity in fig. the relation between both quantities shows that the radial distributions are steeper for the models with lower rotation velocities if n@xmath196. the radial gradients for a given rotation velocity are larger in absolute value for increasing n, and tend to zero value for the most massive galaxies and low n. the models with @xmath197, however, tend to deviate from this function and show lower absolute values. that is, for a similar rotation curve, their radial distributions of abundances show a flattening as compared to those corresponding to the models with @xmath198. the rotation velocity at which the models deviate from the common locus depends on @xmath199 : for @xmath200 it occurs for @xmath201 while for @xmath202 almost all rotation velocity curves produce flat radial gradients of abundances. data from @xcite are over - plotted on this fig. we see that the modelled trend reproduce the observations, although a more profound analysis about this (and other) correlation will be performed in the future. in order to use this grid for a given galaxy for which observational data are known, we must first select the radial distribution of mass. this means that we must know the total mass, the maximum rotation velocity or, if none of them is available, the luminosity or the magnitude in the i band. according to this value we may choose the number of the distribution from table [grid_m]. then, 10 different models, corresponding to the 10 different efficiencies, are available. the standard procedure in chemical evolution would be to see which of them is able to reproduce with success known data such as elemental abundances or gas densities. if more than one observational constraint is available, some kind of minimum error or maximum probability technique may be used for the purpose of choosing the best model. once this selection is performed we may use the time evolution given by the chosen model to predict the star formation history, the age - metallicity relation, the stellar populations or any other quantity relative to the modelled galaxy. obviously, the largest the observational number of constraints, the smallest the uncertainty in the selection of the best model. the old and well known uniqueness problem of the chemical evolution suggest that more than one model may reproduce the data. actually, we have shown in @xcite that this problem reduces greatly when more than two observational constraints can be used. only 4% of the 500 models computed with with different input parameters could reproduce the observations relative to hi, oxygen abundance and sfr at the same time. therefore we are confident that models may provide the evolutionary history of a given galaxy within a reasonable accuracy. sometimes, however, these observational constraints are not available (or not with sufficient precision) to select the adequate model. in that case, some other method to choose the possible evolutionary track is necessary. our best models from previous works give us evidence that the efficiencies to form molecular clouds from atomic gas and stars from molecular gas, seem to depend on the galaxy morphological type. if this were the case, the selection of the best model might be easier. in fact, this dependence was already found in @xcite, where these authors quantified the efficiencies to form molecular clouds and the frequency of cloud - cloud collisions, finding that a variation of 10 in the parameters h and @xmath61 is needed when the hubble type changes from one stage to the next (sa to sb, etc..). in order to check if a relation between efficiencies and morphological type exists, we have plotted in fig. [emu1], panel a) the logarithmic efficiency @xmath74, computed following the equation (19) from section 2.3, as a function of t for the data of @xcite. there @xmath105 symbols correspond to the raw values while solid dots represent the averaged values obtained for 10 bins, one for each t. a clear decreasing correlation appears for @xmath203. values for @xmath204 fall above the trend but the number of points is small for what we assume that the final increase is probably spurious. based on our previous works, we expect a relation of the form : @xmath205. therefore, in panel b) we have plotted the logarithmic efficiency as a function of @xmath206 where a linear correlation is apparent. a least - squares fit gives, for @xmath203 : @xmath207 therefore : @xmath208 on the other hand, taking into account the relationship obtained for the ratio between both efficiencies, @xmath74 and @xmath75, this latter one may also be expressed by a similar function : @xmath209 where b would be 8, following our hypothesis from section 2. the result that efficiencies are related to galaxy morphological type may be useful when a model for a given galaxy must be chosen and the observational data are scarce. in that case, once the total mass distribution has been selected from the rotation curve or the luminosity for the galaxy, an initial choice can be made under the assumption that the galaxy evolution is represented by the model of efficiencies n with @xmath210. we warn, however, that given the large dispersion of the data around the averaged efficiency values in fig. [emu1], other values of n are also possible. this finding would imply that the features of spiral galaxies depend at least on two parameters : total mass and morphological type, as some studies about the characteristics of galaxies, as the seminal work by @xcite, show (see also *??? * ; *??? * ; *??? * among those more related with the chemical evolution of galaxies). it is also well known that the radial gradient, measured as dex @xmath211, observed in spiral galaxies depends on morphological type : late - type galaxies show steeper radial distributions of oxygen abundances than earlier ones, which show in some cases almost no gradient @xcite. at the same time, a correlation seems to exist between oxygen abundance and total mass surface density which, in turn, is related to morphological type. this correlation is stronger when the total mass of the galaxy, including the bulge, instead of the mass of the exponential disc alone, is used @xcite. it is evident from these works that a linear sequence with the mass can not be obtained, and therefore our models, being a bi - parametric grid, would be adequate to find the best model for each galaxy, although the existing relation between the morphological type of galaxies and their total mass does difficult to discriminate which of these features is the origin of the different evolution of galaxies. from the chemical evolution point of view, a recent work (moll & mrquez, in preparation), where the multiphase evolution model has been applied to a large sample (67) of spiral galaxies, seems to demonstrate that the total mass, through its influence on the collapse time scale, has a larger effect on the evolution of a galaxy and its final radial gradient of abundances than the selected values of the named efficiencies. thus, it seems as if galaxies would evolve mostly due to their total mass with some dispersion around the mean trend depending on hydrodynamical and environmental characteristics, which are taken into account in some way by the efficiencies values. the morphological type would then be the consequence of the different conditions of the intergalactic medium out of which the galaxy formed.
Summary and conclusions
we have calculated the chemical evolution of a wide set of theoretical galaxies characterised by their total mass, through the collapse time scale, and their efficiencies to form molecular clouds and massive stars from cloud - to - cloud collisions. with the selection of parameters and inputs described above, we have ran a total of 440 models, with 44 different rotation curves implying 44 values of total mass, characteristic collapse time - scale and disc radius and 10 sets of efficiencies for each one of them, implying 10 evolutionary rates for the star formation and gas consumption in the disc. for each model we have obtained the time evolution of the halo and the disc, and therefore the corresponding radial distributions for the relevant quantities (masses, abundances, star formation rate, etc...). the star formation history for each radial region (halo and disc, separately) and within each one, the mass in each phase of matter : diffuse gas, molecular gas, low - mass stars, massive stars and remnants, has been followed. besides that, we have obtained the abundances of 15 elements : h, d, he3, he4, c12, c13, n14, o16, ne, mg, si, ca, s, fe and neutron - rich nuclei, in all radial regions for both halo and disc. the results of our work can be summarised as follows : 1. the atomic gas shows a maximum in its radial distribution for all galaxies. this maximum is nearer to the galaxy centre in the low mass or less evolved galaxies than in the more evolved or massive galaxies, for which the maximum is along the disc and moving toward the outer zones. this behavior produces, in some cases, a central _ hole _ in the distribution of the diffuse gas. the diffuse gas radial distribution results to be a very strong constraint for selecting the best model out of the 10 computed ones with different efficiencies, corresponding to the total mass of a given spiral galaxy. 2. the oxygen abundance reaches a maximum level, as a consequence of a saturation effect which occurs earlier for the massive and more evolved galaxies. the less evolved galaxies do not reach this saturation level, except in the central region, and therefore show a steep radial gradient in their oxygen abundance. the less massive and less evolved galaxies have not yet developed a radial gradient and show flat radial distributions. this simulates an on - off effect : for @xmath200 a radial gradient appears if @xmath212 while at @xmath213 it only appears for @xmath214. this behavior is in agreement with observations and solves the apparent inconsistency shown by trends showing steep gradients for late type galaxies and flatter ones for the earliest ones, while, at the same time, most irregulars show no gradient at all and very uniform abundances. 3. the model calculated star formation rates reproduce the observed trend between the surface density of this quantity and the total gas density including, not only the massive normal galaxies data, but also the low surface brightness ones. actually, to our knowledge, there are not other chemical evolution models which compare predicted with observed radial distributions of diffuse and molecular gas, star formation rate and abundances for disc galaxies other than mwg. in this sense our models should be considered as an improvement over the standard ones. a study of the possible correlations among galaxy properties and different features obtained from the complete set of results is now possible, since a statistically significant number of theoretical models is available. this requires a deeper analysis which is out of the scope of this paper and will be done in the next future.
Acknowledgments
we thank an anonymous referee for many useful comments and suggestions that have greatly improved this paper. this work has been partially funded by the spanish ministerio de ciencia y tecnologa through project aya-2000 - 093. this work has made use of the nasa astrophysics data system, and the nasa / ipac extragalactic database (ned), which is operated by the jet propulsion laboratory, caltech, under contract with the national aeronautics and space administration. braun, j. m., 2001, dwarf galaxies and their environment, proceedings of the bonn / bochum - graduiertenkolleg international conference, germany, 23 - 27 january 2001, eds. : de boer k.s., dettmar r.- j., klein u., shaker verlag, p. 5 meyer, m. r., adams, f. c., hillebrandt, l. a., carpenter, j. m., & larson, r. b., 2000, in protostars & planets iv, eds. manning, v., boss, a. p., & russell, s. s. (tucson : the university of arizona press), p.121 rrrrrrrrrrrrrrrr time & r & h & d & @xmath121he & @xmath122he & @xmath123c & @xmath124c & n & o & ne & mg & si & s & ca & fe +... &... &... &... &... &... &... &... &... &... &... &... &... &... &... &... + 0.1 & 14. & 0.770 & 0.70e04 & 0.10e04 & 0.230 & 0.11e06 & 0.21e09 & 0.13e07 & 0.92e06 & 0.23e06 & 0.32e07 & 0.55e07 & 0.26e07 & 0.37e08 & 0.42e07 + 0.1 & 12. & 0.770 & 0.70e04 & 0.10e04 & 0.230 & 0.12e06 & 0.22e09 & 0.13e07 & 0.95e06 & 0.23e06 & 0.34e07 & 0.57e07 & 0.27e07 & 0.38e08 & 0.44e07 + 0.1 & 10. & 0.770 & 0.70e04 & 0.10e04 & 0.230 & 0.13e06 & 0.23e09 & 0.14e07 & 0.10e05 & 0.25e06 & 0.35e07 & 0.60e07 & 0.28e07 & 0.40e08 & 0.46e07 + 0.1 & 8. & 0.770 & 0.70e04 & 0.10e04 & 0.230 & 0.14e06 & 0.25e09 & 0.15e07 & 0.11e05 & 0.27e06 & 0.38e07 & 0.65e07 & 0.31e07 & 0.43e08 & 0.50e07 + 0.1 & 6. & 0.770 & 0.70e04 & 0.10e04 & 0.230 & 0.15e06 & 0.28e09 & 0.17e07 & 0.12e05 & 0.30e06 & 0.43e07 & 0.73e07 & 0.35e07 & 0.49e08 & 0.56e07 + 0.1 & 4. & 0.770 & 0.70e04 & 0.10e04 & 0.230 & 0.18e06 & 0.32e09 & 0.19e07 & 0.14e05 & 0.35e06 & 0.50e07 & 0.85e07 & 0.40e07 & 0.57e08 & 0.65e07 + 0.1 & 2. & 0.770 & 0.70e04 & 0.10e04 & 0.230 & 0.45e06 & 0.46e09 & 0.28e07 & 0.38e05 & 0.95e06 & 0.13e06 & 0.22e06 & 0.10e06 & 0.14e07 & 0.15e06 +... &... &... &... &... &... &... &... &... &... &... &... &... &... &... &... + 13.2 & 14. & 0.769 & 0.69e04 & 0.12e04 & 0.231 & 0.59e04 & 0.12e06 & 0.68e05 & 0.15e03 & 0.34e04 & 0.52e05 & 0.10e04 & 0.50e05 & 0.71e06 & 0.24e04 + 13.2 & 12. & 0.768 & 0.69e04 & 0.13e04 & 0.231 & 0.10e03 & 0.21e06 & 0.96e05 & 0.27e03 & 0.57e04 & 0.84e05 & 0.20e04 & 0.96e05 & 0.13e05 & 0.42e04 + 13.2 & 10. & 0.764 & 0.65e04 & 0.21e04 & 0.234 & 0.43e03 & 0.98e06 & 0.38e04 & 0.11e02 & 0.21e03 & 0.30e04 & 0.86e04 & 0.40e04 & 0.55e05 & 0.17e03 + 13.2 & 8. & 0.754 & 0.58e04 & 0.44e04 & 0.240 & 0.11e02 & 0.44e05 & 0.16e03 & 0.28e02 & 0.54e03 & 0.76e04 & 0.22e03 & 0.10e03 & 0.14e04 & 0.44e03 + 13.2 & 6. & 0.746 & 0.51e04 & 0.73e04 & 0.246 & 0.16e02 & 0.11e04 & 0.37e03 & 0.40e02 & 0.78e03 & 0.11e03 & 0.33e03 & 0.16e03 & 0.21e04 & 0.67e03 + 13.2 & 4. & 0.736 & 0.43e04 & 0.13e03 & 0.253 & 0.21e02 & 0.22e04 & 0.64e03 & 0.49e02 & 0.98e03 & 0.14e03 & 0.44e03 & 0.21e03 & 0.28e04 & 0.93e03 + 13.2 & 2. & 0.693 & 0.63e05 & 0.43e03 & 0.287 & 0.37e02 & 0.71e04 & 0.18e02 & 0.81e02 & 0.18e02 & 0.26e03 & 0.83e03 & 0.40e03 & 0.56e04 & 0.20e02 + | we present a generalization of the multiphase chemical evolution model applied to a wide set of theoretical galaxies with different masses and evolutionary rates.
this generalized set of models has been computed using the so - called _ universal rotation curve _ from @xcite to calculate the radial mass distribution of 44 _ theoretical _ protogalaxies.
this distribution is a fundamental input which, besides its own effect on the galaxy evolution, defines the characteristic collapse time - scale or gas infall rate onto the disc.
we have adopted 10 sets of values, between 0 and 1, for the molecular cloud and star formation efficiencies, as corresponding to their probability nature, for each one of the radial distributions of total mass.
thus, we have constructed a bi - parametric grid of models, depending on those efficiency sets and on the rotation velocity, whose results are valid in principle for any spiral or irregular galaxy.
the model results provide the time evolution of different regions of the disc and the halo along galactocentric distance, measured by the gas (atomic and molecular) and stellar masses, the star formation rate and chemical abundances of 14 elements, for a total of 440 models.
this grid may be used to estimate the evolution of a given galaxy for which only present time information such as radial distributions of elemental abundances, gas densities and/or star formation, which are the usual observational constraints of chemical evolution models is available.
[firstpage] galaxies : abundances galaxies : evolution galaxies : spirals galaxies : stellar content | astro-ph0501370 |
Introduction
swimming microorganisms are ubiquitous in nature, and have long been known to play important roles in marine life ecosystems, animal reproduction, and infectious diseases. in these processes, cell motility is crucial.@xcite at the small scales relevant to swimming cells, inertial forces are negligible, and locomotion is constrained by purcell s `` scallop '' theorem stating that any body deformation reversible in time yields zero net motion.@xcite fluid - based cellular motility relies therefore on non - time reversible deformation, for instance by propagating waves along cilia or flagella.@xcite among the various types of locomotion seen in nature, one commonly observed for bacteria is that of helical propulsion, where a flagellum (or a bundle of flagella) rotates as a helix, inducing forward propulsion. a typical example of an organism employing helical propulsion is the bacterium _ escherichia coli _ coli_).@xcite this bacterium alternates `` run '' and `` tumble '' periods : in the former, flagella are synchronized in a coherent bundle and propel the cell forward, whereas in the latter flagella are disorganized, changing the cell orientation and subsequent swimming direction. during run periods, when _ e. coli _ cells are isolated in a bulk flow, they swim in straight (noisy) lines. however, cell locomotion is strongly affected by nearby boundaries. swimming microorganisms often evolve in confined environments, be it by solid boundaries, free surfaces, or liquid interfaces. in some cases, confinement results from channel boundaries, for example along the mammalian female reproductive tract.@xcite surfaces can also be a key element in the microorganism function, as in the case of surface associated infection or biofilm formation.@xcite since such problems are dominated by viscous dissipation, long - range hydrodynamic interactions have been argued to play important roles, resulting in a significant alteration of the locomotion of microorganisms.@xcite over the past years, intensive theoretical, numerical and experimental work has helped uncover the kinematics and dynamics modifications of swimming properties by boundaries.@xcite for bacteria employing helical propulsion (such as _ e. coli _), two different effects induced by boundaries have been discovered and quantified. these organisms swim in the forward direction (cell body forward) and are being propelled from the back. they thus push on the surrounding fluid forward and backward, and such swimmers are referred to as `` pushers ''. in the presence of a nearby solid wall, _ e. coli _ tends to aggregate close to walls.@xcite this is in fact observed for any kind of pusher, not necessarily one exploiting helical propulsion.@xcite a second property, observed solely for helical swimmers, is a circular motion of the cells in a plane parallel to the surface. this was accounted for both experimentally and theoretically in the case of a solid wall @xcite and a free surface.@xcite notably, the circular motion occurs in an opposite direction in the presence of a solid wall (clockwise, cw, when viewed from inside the fluid) or a free surface (counterclockwise, ccw, see fig. [fig:1]). this change in rotation direction is qualitatively similar to the drag increase or decrease observed for the motion of a colloidal particle near a rigid wall and a free surface.@xcite indeed, a solid wall and a free surface induce opposite effects, no - slip for a rigid boundary vs. free slip in the case of a free interface. , and counter - clockwise motion at a free surface (right panel) [reprinted figure with permission from r. di leonardo, d. dellarciprete, l. angelani, and v. iebba, phys. rev. 106, 038101 (2011). copyright (2011) by the american physical society].] past experimental results have been explained theoretically considering newtonian fluids and perfect interfaces, meaning either a no - slip wall or a shear - free surface. theoretical models do predict a single circular direction, cw in the presence of a solid wall vs. ccw in the presence of a free surface, and are consistent with the results illustrated in fig. [fig:1]. however, recent experiments on _ e. coli _ swimming near glass plates and free surfaces show that the distinction in the direction of the circular motion is not straightforward, and both cw and ccw rotations are observed under seemingly similar experimental conditions.@xcite in the initial study of lemelle _ et al. _ (2010),@xcite only cw motion was observed above a glass plate, but both cw and ccw at a free surface, suggesting that particles and surfactants could alter the free slip boundary condition. this hypothesis was further investigated by changing the concentration of a particular polymer that can aggregate at a free surface.@xcite the authors confirmed this qualitative change of behavior, observing a clear dependence on the polymer concentration of the fraction of cells undergoing ccw motion. a similar change in rotation direction was recently highlighted experimentally at a solid wall, when the solution contains polymers.@xcite using a special surface treatment, the polymer concentration at the solid wall was modified, generating possible slip, and resulting in ccw motion. these recent experiments demonstrate that the presence of polymers or surfactants could have a dramatic effect on motility of nearby cells. in this paper we present a modeling approach to quantify the dynamics of swimming bacteria near complex interfaces. when polymers are present in the solution, their concentration close to surfaces is reduced due to higher shear and confinement.@xcite this wall depletion results in the formation of a thin fluid layer of lower viscosity at the wall, thereby modifying significantly the no - slip condition. on scales larger than this thin layer, the equivalent behavior at the wall is an apparent partial slip, characterized by its slip length @xmath0 ranging from @xmath1 nm to 10 @xmath2m.@xcite similarly, a liquid interface covered with surfactants acts as a thin two - dimensional fluid layer separating the liquid phases. this layer has its own rheological properties, and modifies the stress and velocity jumps between the two fluids.@xcite as a consequence, the presence of surfactants can affect significantly the boundary conditions and resulting flow.@xcite in the present work, we address the role of altered boundary conditions on swimming microorganisms, focusing on interface - induced reorientation, attraction vs. repulsion by the surface, and the impact on circular motion. using an analytical framework based on multipole expansions for describing the hydrodynamic interactions between a swimming microorganism and an interface, we show how complex interfaces affect hydrodynamic interactions, providing possible explanations to past experimental results. whereas interface alignment and attraction are seen to be universal properties, the direction of the circular motion turns out to strongly depends on the properties of the fluid, on the bacterium shape and, in some cases, the distance to the interface. in sec. [sec : modeling] we present the modeling approach used throughout the paper. we first introduce the different interfaces considered, and then our solution method quantifying the leading - order effect of hydrodynamic singularities. in sec. [sec : stokesletclosetoaboundary] we recall some existing results for the flow generated by a point force near boundaries and derive in particular the solution in the case of a surfactant - covered interface. sec. [sec : resultstruetrue] is devoted to the main results of the paper, quantifying the impact of complex boundary conditions on swimming bacteria, first on reorientation and attraction, and then on circular motion. we finally conclude in sec. [sec : ccl] while some of the technical details are given in appendices [sec : appb]-[sec : typicaldifferentialequationsandsolutions].
Modeling
throughout the paper, we use the word interface to refer to any boundary separating two phases. we denote phase 1 the fluid where the swimming bacterium is located, while the second phase can either be a solid, a liquid, or a gas. the presence of a nearby interface affects bacteria locomotion in different ways depending on its flow boundary conditions. in this work, we consider three types of interfaces, sketched in fig. [fig:2]a : (i) a clean interface, i.e. a fluid - fluid interface with no surfactant, (ii) a flat surface with a partial slip condition as a model for a polymer depletion layer, and (iii) a liquid interface covered with incompressible surfactants. in the case of flow generated by swimming microorganisms, the typical capillary number, scaling the flow - induced stress with surface tension, is very small, typically lower than @xmath3. as a consequence we will neglect any interfacial deformation induced by the microorganisms, and focus therefore on planar interfaces with normal @xmath4 (fig. [fig:2]a). the first type of interface we will consider is a clean interface between two fluids. this problem is characterized by the viscosity ratio between the two fluids, @xmath5 where the reference viscosity @xmath6 is the viscosity of the fluid where the bacterium is located. if the viscosity of the second fluid vanishes, @xmath7, the interface behaves as a free surface, whereas if the second fluid viscosity tends to infinity, we have @xmath8, and the interface becomes equivalent to a rigid wall. a clean liquid interface imposes impermeability and the continuity of tangential velocities and stresses.@xcite the corresponding boundary conditions written at @xmath9 are @xmath10 where @xmath11 is the symmetric rate of strain tensor, and where we have used the superscript @xmath12 to denote fluid @xmath12. in the case of a free surface (@xmath7), these equations reduce to a free slip condition, whereas in the limit of a rigid wall (@xmath13), the boundary imposes a no - slip condition, @xmath14. the second type of interfaces that we will consider models a rigid wall when the fluid contains polymers, and therefore is subjected to wall depletion. for simplicity, we focus solely on the modification of the boundary conditions and assume no change in the fluid rheology. the model problem will therefore become that of a bacterium swimming in a newtonian fluid close to a partial - slip boundary, meaning that the wall velocity is proportional to the normal velocity gradient. defining @xmath0 the slip length, the boundary conditions read @xmath15 where @xmath16 is the flow velocity in the plane of the interface. the last type of interface we are interested in characterizing are those covered with surfactants. this type of interface behaves as an impermeable, bi - dimensional fluid layer separating the two phases, and the bulk stress jump in the absence of surface forces is given by @xmath17,\end{aligned}\]] where @xmath18 is the surfactant concentration, @xmath19 the surface velocity, @xmath20 the in - plane gradient operator, and @xmath21 (resp. @xmath22) the bulk (resp. surface) stress tensor.@xcite in the present analysis, it is appropriate to focus on the stationary limit on the length scales of swimming bacteria. additionally, we assume that the concentration of surfactant on the surface is large enough so that the relative change in surface tension is of order 1. in that case, as the capillary number is very small (typically lower than @xmath3), the surfactant concentration can be considered to be uniform on the surface.@xcite under this assumption, the interface is incompressible, @xmath23, and the surface stress tensor reads @xmath24 where @xmath25 is the surface tension and @xmath26 the in - plane rate of strain tensor.@xcite the resulting boundary conditions at @xmath9 state continuity of in - plane velocity, no flow perpendicular to the surface, in - plane incompressibility, and balance between shear stresses from the outside flows, surface viscous stresses and marangoni stresses @xmath27 where @xmath28 is the non - dimensional surface viscosity, @xmath29 @xmath30 being a flow length scale, chosen here to be the bacterium distance to the wall.@xcite the parameter @xmath28 is often referred to as boussinesq number, comparing surface to bulk stresses @xcite. note that the condition for incompressibility prevents the limit of a clean interface to be reached by simply applying @xmath31, and only the case of a no - slip boundary can be recovered in the limit @xmath32. in order to analyze the effect of an interface on the swimming microorganism, we decompose the flow into two contributions, one being the flow generated by the same organism in the absence of interface, @xmath33, and a contribution due solely to the interface, @xmath34, so that @xmath35. the effect of the interface on the swimming microorganism is then determined using faxn s law, modeling the bacterium shape as that of a prolate ellipsoid swimming along an intrinsic direction @xmath36. noting @xmath37 the aspect ratio of the entire organism (i.e. cell body and flagellar bundle), the interface - induced velocity, @xmath38, and rotation rate, @xmath39, of the microorganism are given by @xmath40,\end{aligned}\]] where @xmath41 is the location of the bacterium @xcite (fig. [fig:2]b). the fact that the interface effect is evaluated only at the location of the swimmer is a key point here we do not seek to describe the entire flow field in the presence of the aforementioned interfaces, but only the flow at a specific location. the reynolds numbers associated with swimming microorganisms are very small.@xcite in this limit, the flow generated by a swimming microorganism satisfies the linear stokes equations,@xcite which allow for a multipole expansion of any flow.@xcite in the present work, we will represent the flow induced by a swimming microorganism using flow singularities. general flow singularities and notation are presented below while in sec. [sec : resultstruetrue] we will focus on the specific singularity model used for microorganisms using helical propulsion. flow singularities are derived from the green s function for stokes flows, @xmath42, which gives the flow at position @xmath43 generated by a point force @xmath44 located at @xmath41 and oriented along the direction @xmath45 as @xmath46 where @xmath47, and @xmath48. this solution is called the stokeslet, and higher - order singularities are then derived from this fundamental singularity.@xcite we introduce here the stokeslet dipole, @xmath49, and quadrupole, @xmath50, @xmath51 and the (potential) source dipole, @xmath52, and quadrupole, @xmath53, @xmath54 where the notation @xmath55 is used to denote a gradient taken with respect to the singularity location, @xmath41. an useful combination of stokes dipoles is the rotlet,@xcite which is the anti - symmetric part of a stokeslet dipole, and models the flow generated by a point torque @xmath56 = \frac{{{\bm{c}}}\times{{\bm{r}}}}{r^3},\end{aligned}\]] where @xmath57. the symmetric part of a stokeslet dipole is called a stresslet, @xmath58 $]. note that all these singularities are @xmath59-linear functions of their @xmath59 orientation vectors. a singularity oriented along arbitrary directions can thus be expressed as a combination of similar singularities along the different basis vectors. the mathematical method used for solving the problem of a point singularity located close to a boundary with specific conditions is blake s method, presented first for the problem of a stokeslet close to a rigid wall @xcite. the effect of the interface is mathematically equivalent to an additional flow generated by a system of hydrodynamic image singularities located on the other side of the interface. for complex boundary conditions, the image system is a spatial distribution of singularities. the problem can be significantly simplified by guessing part of the image system. noting @xmath60 the image system, the flow is decomposed as @xmath61 where @xmath33 is the bulk solution due to the singularity itself, @xmath62 is a guess in the image system, and @xmath63 and @xmath64 are the unknowns. for example, in the case of a stokeslet close to a free surface, a good guess for @xmath62 is a stokeslet symmetric with respect to the interface, which is enough to enforce both the impermeability condition and that of zero shear stress, resulting in @xmath65. the major advantage of this decomposition is that the forcing term due to the singularity in the flow equation is solved by the bulk flow term @xmath33. as a result, @xmath63 and @xmath64 satisfy stokes equations in the absence of forcing terms @xmath66 where @xmath67 is the velocity field and @xmath68 the pressure, @xmath69 being the fluid viscosity. that problem can be more easily solved in fourier space,@xcite using a two - dimensional (2d) fourier transform, defined as @xmath70 } = \frac{1}{2\pi}\int\int{f(x, y, z)e^{ik_1x+ik_2y}\mathrm{d}x\mathrm{d}y}.\end{aligned}\]] some useful fourier transforms are referenced in appendix [sec : appb]. the solution of stokes equations in a 2d fourier space is straightforward, and reads for @xmath71 and @xmath72 @xmath73 the unknown coefficients in this solution are determined using the continuity equation@xmath74 together with the relevant boundary conditions. once the solution is obtained for the stokeslet, it is then possible to derive the solution for higher - order singularities from the stokeslet solution. this is achieved by applying the same operator acting on the singularity position, @xmath75, where @xmath76 is the direction where the derivative is taken. note that this can be done more easily in fourier space, as @xmath77}=ik_1,\quad { \mathcal{f}\left[{{\bm{e}}_\mathrm{y}}.{\nabla_0}\right]}=ik_2, \quad { \mathcal{f}\left[{{\bm{e}}_\mathrm{z}}.{\nabla_0}\right]}={\frac{\partial } { \partial h}}.\end{aligned}\]] however, it is also possible to directly apply blake s method to any singularity, which can be more convenient if a good guess of @xmath62 is found.
Stokeslet close to a complex interface
as a singularity is a linear function of its orientation vectors, one only needs to solve the problem for the stokeslet along the directions parallel and perpendicular to the interface. the flow generated by higher - order singularities can then be derived from these projections. by symmetry reasons, all orientations in the plane parallel to the interface are equivalent, we therefore choose @xmath78 to be the singularity plane. we review below some solutions of the flow generated by parallel and perpendicular stokeslets close to a fluid boundary, and then use the method described in the previous section for determining the flow in the presence of a surface covered with incompressible surfactant, providing an alternative to the derivation by bawzdziewicz _ (1999).@xcite the solution in the case of slip is given in lauga and squires (2005)@xcite along the same lines. the case of parallel and perpendicular stokeslets were derived by blake in the case of a solid wall, a free surface, and a fluid - fluid interface.@xcite in the presence of a clean fluid - fluid interface, the image system is made of a finite number of point singularities located at the image position @xmath79 in fluid 2 (see notation in fig. [fig:2]b). the image system is composed of a stokeslet, a stokes dipole, and a source dipole, whose intensities vary with the viscosity ratio @xmath80 and the distance @xmath30 to the interface. for stokeslets parallel, @xmath81, and perpendicular to the surface, @xmath82, the image system is given by @xmath83 where we use @xmath84 in order to simplify the notations, and where the superscript @xmath85 implies that the singularity is located at the image position, @xmath79. these singularities give the flow in fluid 1, where the singularity is located. the flow in fluid 2 is given by a second set of singularities, given by @xmath86,\\ & \text{im}^{(2)}\{{{\bm{g_s}}}({{\bm{e}}_\mathrm{z}})\}= \frac{2 h}{1+{\lambda } } \big[{{\bm{g_{sd}}}}({{\bm{e}}_\mathrm{z}},{{\bm{e}}_\mathrm{z}})+h{{\bm{g_{d}}}}({{\bm{e}}_\mathrm{z}})\big].\end{aligned}\]] this second set of images is located at the singularity position @xmath41, in fluid 1 (see fig. [fig:2]b). the solutions are therefore obtained fully analytically, and higher - order solutions can be obtained by taking derivatives with respect to the singularity position (@xmath87). note that since the image strengths are functions of the singularity distance to the interface @xmath30, taking derivatives of the image system along @xmath4 generates additional singularities.@xcite in the presence of a solid wall, the second set of images vanishes since there is no flow in the solid region. this was the no - slip solution originally presented by blake.@xcite as we are interested in this work only on the effect of the interface on the swimming bacterium, we will focus only on the first set of images, eqs. and, giving the flow field in the region where the swimming microorganism is located. the no - slip and no - shear solutions have been used in the past to explain different swimming behaviors of bacteria observed experimentally, such as wall attraction, alignment, and circular motion.@xcite the particular case of no - slip boundary (@xmath88) is noteworthy, and will be used in the following as a reference solution @xmath89 in the presence of slip or surfactants, the image system is no longer a set of point singularities. the solution to the problem of a stokeslet close to a partial slip boundary was addressed by lauga and squires,@xcite using blake s method. the case of a surfactant - covered interface has been addressed by bawzdziewicz _ et al._,@xcite using a batchelor - type decomposition of the flow. in order to use the same formalism throughout our paper, we derive below the solution of a stokeslet close to a surface covered with incompressible surfactants using blake s method. we consider here the case of a stokeslet perpendicular and parallel to a surfactant - covered interface as defined in sec. [sec : interfaces]. without loss of generality, the stokeslet strength is taken to be 1. for this problem, we take @xmath62 to be the opposite stokeslet located at the image position @xmath79. this choice yields at @xmath9 for a perpendicular and parallel stokeslet respectively @xmath90({{\bm{e}}_\mathrm{z } }) = -\frac{h}{4\pi{\eta}_1}\frac{1}{r_h^3}(x{{\bm{e}}_\mathrm{x}}+y{{\bm{e}}_\mathrm{y}}),\quad \left[{\bm{u}}+{\bm{v}}\right] ({ { \bm{e}}_\mathrm{x}})= -\frac{h}{4\pi{\eta}_1}\frac{x}{r_h^3}{{\bm{e}}_\mathrm{z}},\end{aligned}\]] where @xmath91. the problem is then solved in fourier space. in the case of a perpendicular stokeslet, the boundary conditions read in fourier space @xmath92 for @xmath93 and 2, where index 1 (resp. 2) is associated with the @xmath94 (resp. @xmath95) coordinate. this set of equations is satisfied by @xmath96, which corresponds to a solid wall. we recover therefore a known result : because of surface incompressibility, the flow generated by a stokeslet perpendicular to an interface covered with incompressible surfactants is identical to that in the presence of a solid wall.@xcite in this case, the boundary conditions are given in fourier space by @xmath97 introducing the coefficients @xmath98, @xmath99, @xmath100 and @xmath101 from sec. [sec : solmeth], this system can be solved directly, leading to @xmath102, and @xmath103 it is interesting to note that the coefficients in the no - slip problem read @xmath104 hence, introducing @xmath105e^{-kz}/(8\pi{\eta}_1)$], the flow field @xmath63 can be written as @xmath106 where @xmath107 is the known solution of the problem in the presence of a no - slip boundary, @xmath108. after an inverse fourier transform of @xmath109, one can identify two differential equations satisfied by @xmath110 and @xmath111 in real space, which can be written as a single vectorial differential equation on @xmath112, @xmath113 where @xmath114 is a dipole along @xmath115 of rotlets along @xmath4, located at @xmath79. the rotlet dipole involved in this flow corresponds to two vertical counter - rotating point vortices, merging at @xmath79. note that this singularity can also be written in a way similar to that of ref.@xcite as @xmath116. the flow in the second fluid satisfies a similar problem, @xmath117 with the rotlet dipole being located at @xmath41. as a result, the total flow generated by a stokeslet parallel to an interface covered with surfactants can be written as @xmath118 where @xmath119 is the flow generated by a stokeslet along @xmath120 in the presence of a no - slip boundary. we recover therefore the result of ref.@xcite where the flow is the sum of the no - slip contribution and a `` surface - solenoidal '' flow, i.e. a 2d flow decaying in @xmath121. in order to obtain the exact expression for the flow at the position of the singularity, one still needs to integrate a third order differential equation, eq. . since our goal is not to derive the entire flow but only the effect of the boundary on the swimming microorganism, we can simplify in the following way. the singularity is located at @xmath122 and since the differential equations giving the flow are equations in @xmath121, it is possible to set @xmath122 before integrating. as a result, the additional flow induced on the singularity is given by @xmath123, where @xmath124 and @xmath125 the latter equation yields @xmath126, which could have been anticipated by symmetry. the former equation can be integrated analytically, as detailed in the next section.
Swimming microorganisms and interfaces
in this work, we model a swimming bacterium in the far field and thus assume that it is small compared to any flow length scale. in particular, the distance between the cell and the interface, @xmath30, must remain larger than the bacterium size. this far - field approach allows for a representation of a swimming microorganism as a combination of point singularities, while the organism size and shape play a role only through the aspect ratio @xmath37 involved in faxn s law, eq. . we focus here on microorganisms using helical propulsion, like _ e. coli _, as illustrated schematically in fig. [fig:3]a. such micro - swimmers are force- and torque - free, and can be considered as axisymmetric in average. we choose as a convention that the microorganism swims in the @xmath78 plane, along direction @xmath127. at leading order (spatial decay @xmath128), the far - field flow generated by a swimming bacterium is well captured by an axisymmetric stokeslet dipole, @xmath129, as sketched in fig. [fig:3]b.@xcite this singularity is force- and torque - free, and accounts for the spatial distribution of thrust and drag on a flagellated swimming cell.@xcite this representation is commonly used for describing far - field hydrodynamic interactions, both to quantify collective dynamics @xcite and wall effects.@xcite however, by symmetry, this leading - order singularity can not be responsible for the observed circular rotation along the normal to the interface since @xmath78 is a symmetry plane of the stokes dipole. it is thus necessary to add higher - order singularities decaying as @xmath130 to capture circular swimming. those singularities could be : (i) a stokes quadrupole, related to the length asymmetry between the flagella and the body, (ii) a potential source dipole accounting for the finite size of the bacterium, and (iii) a rotlet dipole capturing the counter - rotation between the flagella and the body. by symmetry, it is straightforward to see that the only singularity that could potentially lead to nonzero rotation along @xmath4 is the rotlet dipole @xmath131, illustrated in fig. [fig:3]c. we are interested in three effects induced by a nearby boundary : (a) reorientation in the swimming plane, given by rotation rate @xmath132 ; (b) attraction or repulsion by the wall, quantified by wall - induced velocity @xmath133 ; and (c) circular motion, measured by the out - of - plane rotation rate @xmath134. reorientation and attraction can both be accounted for by the leading - order stokes dipole, as analyzed in sec. in contrast, the circular swimming which is due to the rotlet dipole is addressed in sec. [no]. we consider here the leading order singularity representing a swimming microorganism, a stokes dipole, and focus on two effects. we first address the issue of reorientation in the @xmath78 plane induced by a boundary on a tilted micro - swimmer moving along the direction @xmath127. with the stable wall - induced orientations derived, @xmath135, we then address the attractive vs. repulsive nature of the interface. a stokes dipole tilted along the angle @xmath136 is a combination of parallel and perpendicular stokes dipoles as @xmath137 the wall - induced rotation rate is dependent on the orientation of the microorganism and the local strain rate, see eq. . in the following, the strength of the stokes dipole is taken to be 1, modeling a `` pusher '' swimmer as relevant to any flagellated bacteria moving cell body first (a puller corresponds to a negative strength). in the presence of a clean interface, the image system is the set of point singularities listed above, and the problem can be solved analytically, leading to the wall - induced rate @xmath138. \label{omcleansurf}\end{aligned}\]] notably, the sign of this rotation rate is given by @xmath139 as the term in the bracket is always positive. as a result, the interface will always tend to align the (pusher) stokes dipole in the direction parallel to the interface. this result was previously shown in the case of a solid wall,@xcite and is thus generalized here to any fluid - fluid interface. the stable orientation is therefore a microorganism swimming parallel to the interface, @xmath140. in that case, the induced velocity along the vertical axis reads @xmath141 we see thus that a swimming pusher will be attracted by a nearby liquid interface for any value of the viscosity ratio. in the limits of a free surface (@xmath142) and a solid wall (@xmath143), standard results are recovered.@xcite in the presence of a partial slip boundary, the flow due to the boundary is equivalent to that generated by a continuous distribution of singularities along the vertical axis on the other side of the surface.@xcite taking derivatives of higher - order solutions is not straightforward in real space, but is easily carried out in fourier space. it is interesting to note that @xmath144 is a symmetry plane for the first two singularities of the decomposition shown in eq. , namely @xmath145 and @xmath146. as a result, these terms do not contribute to any rotation rate through vorticity, only the stresslet does. however, all terms contribute to the total rotation rate, as the swimmer s orientation breaks the symmetry, acting on the strain contribution to the rotation rate, eq. . the three singularities necessary for describing a tilted stokes dipole are derived with a particular choice for the different velocity fields @xmath62, so that @xmath62 corresponds to the solution in the presence of a free surface. this limit is reached in the partial slip model when @xmath0 tends to infinity. for the stresslet @xmath147, we choose therefore @xmath148 ; the no - slip solution reads @xmath149 following the procedure outlined above we determine the solution in the presence of a partial slip boundary in fourier space @xmath150 similarly we have for the parallel stokes dipole @xmath145 @xcite @xmath151, \label{eq : sdpar}\end{aligned}\]] with the no - slip solution @xmath152. for the perpendicular stokes dipole @xmath146, we have @xmath153.\end{aligned}\]] the rotation rate of the cell along @xmath115 is then computed in fourier space for the three singularities, and the total rotation rate on a tilted stokes dipole close to a partial slip boundary finally reads @xmath154 evaluated at the singularity position, where index sdx stands for parallel stokes dipole, sdz for perpendicular stokes dipole, and ss for the stresslet contribution. the technical difficulty in this problem is the inverse fourier transform of the rotation rate and we refer to appendices [sec : appa] and [sec : typicaldifferentialequationsandsolutions] for the details related to the inversion problem and solution of the resulting differential equations. the final solution in real space is given analytically by @xmath155-f_3\!\!\left[\frac{h}{{\ell}}\right]-h_4\!\!\left[\frac{h}{{\ell}}\right]\right) + { \frac{\gamma^2 - 1}{\gamma^2 + 1}}{\sin^2 \!{\theta}}\nonumber\\ & \qquad\qquad \quad \quad + { \frac{\gamma^2 - 1}{\gamma^2 + 1}}\frac{h}{{\ell } } \bigg[-2{\sin^2 \!{\theta}}f_3\!\!\left[\frac{h}{{\ell}}\right] + (3{\cos^2 \!{\theta}}-2{\sin^2 \!{\theta}})h_4\!\!\left[\frac{h}{{\ell}}\right]\nonumber\\ & \qquad \qquad\qquad \qquad \quad+(1 + 3{\sin^2 \!{\theta } }) \left(f_4\!\!\left[\frac{h}{{\ell}}\right]-\frac 12f_5\!\!\left[\frac{h}{{\ell}}\right]+\frac 12g_4\!\!\left[\frac{h}{{\ell}}\right]-g_5\!\!\left[\frac{h}{{\ell}}\right]\right)\bigg]\bigg\},\end{aligned}\]] where the functions @xmath156, @xmath157 and @xmath158 derive from the exponential integral function of order @xmath59, @xmath159, as @xmath160 with @xmath161 , characterizing the alignment of the bacterium with the boundary, as a function of the orientation angle @xmath162 and for different values of the dimensionless slip length @xmath163 (0.1, 1, and 10, thin solid lines). bold dashed lines correspond to the limits of vanishing slip length (no - slip condition) and infinite slip length (no shear). (b) qualitative physical picture for interface alignment and attraction ; streamlines are thin dashed lines, thin arrows (dark blue online) show local effects while thick arrows (light red online) show global effect on the swimming microorganism.] the dependence of the rotation rate @xmath164 induced by a partial slip boundary on a tilted stokeslet dipole is shown in fig. [fig:4]a, as a function of the tilt angle @xmath162, for different values of the slip length. we see that the slip length, similarly to the viscosity ratio in the case of a clean liquid interface, does not modify qualitatively the reorientation dynamics and alignment rate with the boundary and a pusher micro - swimmer will always tend to align parallel to the boundary. the limit of infinite slip length, @xmath165 in eq. , leads naturally to the free surface limit. furthermore, using the asymptotic properties of the exponential integral functions, we have @xmath166 and @xmath167 when @xmath94 tends to infinity, which can then be used to recover quantitatively the reorientation dynamics in the no - slip limit. we then compute the velocity induced on a micro - swimmer parallel to the boundary, modeled by a parallel stokes dipole, eq. . the integration can be done in a similar way as for the rotation rate, and using integration by parts we get a simple expression for the vertical velocity as @xmath168 a partial slip boundary will always attract a pusher swimmer for any value of the slip length, generalizing therefore the known results for @xmath169 (no - slip) and @xmath170 (no - shear). we finally consider the case of a stokes dipole close to a liquid interface covered with incompressible surfactants. from the stokeslet solution presented in sec. [sec : surfactstokeslet], it is possible to derive higher - order solutions. using eq., the flow generated by a tilted stokeslet dipole is therefore given by @xmath171 with @xmath172 using the same methodology as that described in the previous section and appendix [sec : appa], we get that @xmath173 ^ 0 + { \cos^2 \!{\theta}}~\!\omega_{xx } + { \cos { \theta}}{\sin { \theta}}~\!\omega_{xz}$], where @xmath174 ^ 0 $] is the solution in the presence of a no - slip boundary, given by eq. with @xmath13, and @xmath175 (resp. @xmath176) is the contribution from the flow @xmath177 (resp. @xmath178). the resulting rotation rate is given by @xmath179 + { \frac{\gamma^2 - 1}{\gamma^2 + 1}}{\cos^2 \!{\theta}}\frac{3}{\beta } \int_1^\infty{\!\!\! \int_1^\infty{\frac{f_5(2bst)}{t^3s^4}\mathrm{d}s}\mathrm{d}t}\nonumber\\ & + \frac{1}{2\beta}f_3(2b)\left[1+{\frac{\gamma^2 - 1}{\gamma^2 + 1}}({\sin^2 \!{\theta}}-{\cos^2 \!{\theta}})\right]\bigg\},\end{aligned}\]] where @xmath180 and the functions @xmath156 are defined in eqs. -. for any value of the parameters and orientation angle, the sign of the rotation sign is that of the prefactor @xmath139. as a result, a surface covered with incompressible surfactants tends to always align a swimming microorganism with the surface, similarly to the no - slip and no - shear cases. furthermore, since the additional velocity field @xmath112 does not have a vertical component, we see immediately that the vertical velocity induced by the interface is the same as that in the presence of a no - slip boundary, and again a surface covered with surfactants will attract a pusher toward the interface. in summary, the leading - order singularity modeling a swimming microorganism allowed us to generalize results known for both a solid wall and a free surface. the interface, be it a clean fluid - fluid interface, a partial slip boundary, or surface covered with surfactants always induces alignment parallel to the nearest surface and attraction toward it. the only common boundary conditions to these three types of interfaces is impermeability, which strongly confines the fluid in the vertical direction. this confinement provides a qualitative argument for explaining alignment and attraction of pushers near any type of boundary, as sketched in fig. [fig:4]b. when tilted, a pusher will experience a higher vertical fluid force on the part of the cell closer to the interface, inducing a reorientation in the parallel direction. when swimming parallel to the interface, fluid is attracted towards the cell on its side, leading to attraction toward the surface. from a mathematical point of view, impermeability on the surface is enforced by a symmetric singularity as an image in the second fluid. at leading order it is therefore as if there were two symmetric micro - swimmers, which tend to align and attract each other when they are pushers.@xcite we turn now to a higher - order representation of a swimming bacterium in order to account for its circular motion near interfaces. given the results above concerning the leading - order effect of the interfaces, we will assume that the the microorganism is swimming parallel to it. the micro - swimmer is now modeled by a rotlet dipole along the swimming direction, @xmath120, which can be written as a combination of stokes quadrupoles, @xmath181.\end{aligned}\]] as a result, the flow generated by a rotlet dipole close to a boundary can be obtained from that of the stokeslet, one only needs to consider the parallel stokeslet along @xmath115 rather than @xmath120. in this section, in order to perform dimensional analysis, we write the strength of the rotlet dipole as @xmath182 (units of n.m@xmath183). in the case of a clean fluid - fluid interface, we start by considering the problem using dimensional analysis. in a far - field approach, the bacterium geometry plays a role only through the non - dimensional aspect ratio @xmath37 appearing in faxn s law. the dimensional quantities involved in the rotation rate are then the rotlet dipole strength @xmath182, the two viscosities @xmath6 and @xmath184, and the distance @xmath30 to the interface. a straightforward dimensional analysis then yields a rotation rate normal to the interface given by @xmath185 where we recall that @xmath186. the non - dimensional function @xmath187 needs to be determined analytically, but we can already see from eq. that the sign of @xmath188, and thus the direction of rotation of the circular motion, will be determined by a comparison between the viscosity ratio and the swimmer geometry, and will be independent of the distance to the interface @xmath30. moreover, within the context of our far - field approach, we observe that this rotation rate decays as @xmath189. the circular motion is therefore likely to occur very close to the interface. starting from the image system of a stokeslet in the presence of an interface, we derive the image system for a rotlet dipole along @xmath120, and get @xmath190,\end{aligned}\]] where @xmath191 is a stresslet dipole. recall that the limit of a free surface (resp. no - slip wall) is recovered in the limit of vanishing (resp. infinite) viscosity ratio.@xcite the structure of the image system in eq. reveals two roles played by the interface : (i) a kinematic role, through the impermeability condition, that forces the existence of an opposite rotlet dipole independent of the viscosity ratio, and (ii) a viscous component, related to the balance of tangential velocity and stress. interestingly, in the limit of a free surface (@xmath7), the viscous contribution vanishes, and the only remaining image singularity is the opposite rotlet dipole.@xcite the rotation rate induced by a clean interface is then computed from eq. using the flow from eq. , leading to the result @xmath192 in eq. we can identify the two different contributions which were apparent in the image system : (i) a kinematic contribution independent of the viscosity ratio, and (ii) a viscous contribution due to the rotlet and stresslet dipoles (a potential singularity has no vorticity), independent of the micro - swimmer geometry. re - writing eq., we get another form for the non - dimensional function @xmath193 in eq. as @xmath194 one can see from this expression that the sign inversion for the rotation rate, and thus the transition from ccw to cw circles, occurs for @xmath195 (see fig. [fig:5]). as expected from the dimensional analysis, we found a condition involving the swimmer geometry and the viscosity ratio for determining the rotation direction, regardless of the distance to the interface. for spherical swimmers (@xmath196), the transition is predicted to occur exactly at a viscosity ratio of 1, but as the body becomes elongated this threshold is significantly modified. we would therefore expect a clear differentiation in rotation depending on the swimmer shape, for an identical viscosity ratio., as a function of the viscosity ratio @xmath80 divided by the aspect ratio square @xmath197 (for a fixed value of @xmath198). circular motion is predicted to be cw when @xmath199 and ccw otherwise.] in the presence of a clean interface between two given fluids, only the bacterium shape determines the rotation direction, and not the distance to the interface. when both fluid have comparable viscosities, the dominant behavior should be that of the free surface limit, a counter - clockwise rotation, as soon as the bacterium is elongated. this is not consistent with experiments, where both cw and ccw rotations where observed at a free surface or a solid wall.@xcite this indicates a more complex role played by the interface, and motivates the next two sections. in the presence of a partial slip boundary, an additional length is introduced in the problem. a dimensional analysis similar to the one carried out above yields @xmath200 and thus the sign of the rotation rate should depend on the distance to the wall, as opposed to the case of a clean fluid interface. in order to analyize the effect of a partial slip boundary, we derive the solution using blake s method directly for a rotlet dipole. choosing @xmath201, we have in fourier space @xmath202 where @xmath203. following an analysis similar to that detailed in appendix [sec : appa], we then obtain the rotation rate induced on the micro - swimmer @xmath204 \nonumber\\ & -\frac{h}{{\ell}}\left[(\gamma^2 - 1)g_5\!\!\left(\frac{h}{{\ell}}\right)+(3\gamma^2 + 1)h_5\!\!\left(\frac{h}{{\ell}}\right)\right]\bigg\},\end{aligned}\]] with the functions @xmath156, @xmath157, and @xmath158 defined in eqs. -. from eq. , the known limits of a free surface (@xmath205) and solid wall (@xmath206) are easily recovered. , induced by a partial slip wall on a parallel rotlet dipole as a function of the normalized slip length, @xmath163, for three values of @xmath37 (1, 3, and 10) ; (b) critical value of @xmath163 at which the rotation sign changes as a function of @xmath37. the regions where the circular trajectories are cw and ccw are indicated on the figure.] we plot in fig. [fig:6](a) the perpendicular rotation rate, @xmath188, as a function of the normalized slip length, @xmath163, for different values of @xmath37. for small slip lengths compared to the distance of the cell to the wall, the effect of slip is negligible, and the standard result of a no - slip boundary is recovered (cw rotation). however, when the swimmer gets close enough to the wall, or when the slip length becomes large enough, the rotation rate changes sign and takes that due to a free surface (ccw rotation). since we argued above that cells are always attracted to interfaces, we thus expect ccw rotations to be observable in this case, consistently with the recent experimental results in the presence of slip - inducing polymers.@xcite in fig. [fig:6](b) we further plot the dependence of the critical dimensionless slip length at which the rotation changes sign on the aspect ratio of the cell. we find numerically that @xmath163 scales approximatively as @xmath207 at large values of the aspect ratio. in this last section we consider the case of an interface covered with incompressible surfactants. using dimensional analysis, we obtain that the rotation perpendicular to the surface is @xmath208 and therefore the distance to the wall, @xmath30, also plays a role as it is included in @xmath28, the non - dimensional surface viscosity, the flow generated by a rotlet dipole close to a surfactant - covered interface can be derived from the stokeslet solution, or computed directly using blake s method. we have @xmath209 where @xmath210 is the additional flow field generated by a stokeslet along @xmath115 (i.e. eq. with a rotation of @xmath211 along @xmath4). the rotation rate along the vertical axis can then be derived directly, leading to the analytical result @xmath212\bigg\},\end{aligned}\]] with the functions @xmath156 defined in eqs. -. , induced by an interface covered with incompressible surfactants on a parallel rotlet dipole : (a) contour values of @xmath188 for a spherical swimmer (@xmath213) and (b) for an ellipsoidal swimmer (@xmath214). the dashed line (red online) indicates the location where the rotation rate changes sign. regions of cw and ccw rotations are schematically shown.] the dependence of this rotation rate with the two non - dimensional viscosities, @xmath215 and @xmath216, is shown in fig. [fig:7]. for low cell aspect ratio, the sign of the circular rotation is similar to the no - slip case for a wide range of parameters (cw rotation), and thus opposite to the prediction in the case of clean interface. the sign of rotation is then changing at low @xmath28 (low surfactant concentration) and low viscosity ratio @xmath217. furthermore, as the aspect ratio of the cell @xmath37 increases, the region displaying ccw rotation becomes larger, allowing for both rotations to be potentially observed experimentally in cell populations of different sizes or on surfaces with fluctuations in surfactant concentration. comparing with other interfacial models, we get that when @xmath28 tends to infinity, the solid wall limit is recovered. furthermore, and despite the fact that the surfactant model is not supposed to recover exactly the clean - interface limit, we see that the clean - interface threshold for rotation inversion (@xmath218), corresponds to the order of magnitude of the vertical asymptote of the dashed line in fig. [fig:7]. , at which the rotation changes sign, as a function of @xmath217 for three values of @xmath37 (a), and as a function of @xmath37 for four values of @xmath217 (b). regions of cw and ccw rotations are indicated on the figure.] we then plot in fig. [fig:8] the critical value of @xmath28 at which there is a change in sign of the rotation rate, as a function of the viscosity ratio @xmath217 for different aspect ratios (fig. [fig:8]a), and as a function of @xmath37 for different @xmath217 (fig. [fig:8]b). for large cell aspect ratios, we find that @xmath28 scales as @xmath197, similarly to the scaling seen for @xmath217. this can be interpreted by noting that this problem is the same as that in the presence of a clean interface, but for the presence of a third fluid, and thus a third viscosity that needs to be compared with the viscosity where the swimming microorganism is located. hence, the criterion for a change of sign in the rotation direction is similar for both non - dimensional viscosities, @xmath219 and @xmath28. there is in general a wide range of surface viscosities, depending on multiple parameters, such as the type of surfactants or temperature.@xcite for instance, for a bacterium swimming at a typical distance of 1 @xmath2 m from the interface, @xmath28 can be of order @xmath220 when the interface is a monolayer at a water - air interface.@xcite as a result, cw rotation (as in the no - slip case) is more likely to be observed. in contrast, in the case of an amphiphilic bilayer,@xcite the values can be much smaller, @xmath221. as a result, there is likely a wide range of parameters where both cw and ccw rotation could be observed in a population of cells. recent experimental results on contaminated free surfaces showed that both cw and ccw circular motion could be seen.@xcite by considering the presence of surfactants on the surface, our analysis shows that the high viscosity fluid film at the interface could indeed alter the natural shear - free rotation direction and lead to cw motion.
Conclusion
in this paper we have used a far - field hydrodynamic approach to model the surface swimming of bacteria employing helical flagella. the motivation for this work was the discrepancy between theoretical predictions and experimental observations. specifically, theory predicts that near a rigid wall the cells should always display cw motion, whereas recent experiments where polymers were used to induce slip at the wall showed that rotation in the opposite direction was possible. similarly, cells should rotate in a ccw motion at a free surface whereas if surfactants are present experiments show that cw motion is also observed. to develop a model we have represented the helical swimmer as a superposition of hydrodynamic singularities and investigated its hydrodynamic interactions with three types of surfaces : a clean fluid - fluid interface, a rigid wall with a finite slip length, and an interface covered by incompressible surfactants. the leading - order singularity in the flow field of the cell is a stokes dipole (stresslet), characterized by a @xmath128 spatial decay. the interactions between that singularity and all three types of surfaces systematically lead to a reorientation of the swimming cells parallel to, and an attraction by, the surface. circular motion of the cells are due to wall effects on a higher - order singularity, namely a rotlet dipole, which decays spatially as @xmath130. in that case, the specific boundary conditions at the interface, together with the shape of the cell, play a crucial role in determining the direction of rotation of the cell, and transitions between cw and ccw are predicted to take place in similar experimental setups. our results indicate thus that the recent experimental finding on transition in rotation direction can be understood as the consequence of complex boundary conditions on the nature of hydrodynamic interactions between the swimming cells and the surfaces. the main assumption made in our paper is that we have only considered the leading - order hydrodynamics effects for all influences of the interfaces (attraction and rotation). this is, admittedly, a severe assumption which is expected to break down as soon as the cell is within about one body length from the interface. in order to obtain more quantitative predictions, one would then need to either include the effect of higher - order singularities, or resort to a fully computational approach. the advantage of our approach however is that it allows us to identify the fundamental physical process at play in setting the direction of rotation, and that it is expected to remain valid generically for all cells exploiting helical swimming. our findings could potentially be exploited in a numbers of ways, for example surface swimming could be used as a proxy for determining the rheological properties of the nearby interface or to selectively stir or sort individual cells from bacterial populations. we hope that our study will motivate further work along these directions.
Acknowledgements
the authors thank the department of mechanical and aerospace engineering at the university of california, san diego where this work was initiated.
Known fourier transforms
we reference here some useful fourier transforms @xmath222 } = \frac{1}{k}e^{-kz},\quad { \mathcal{f}\left[\frac{1}{r^3}\right] } = \frac{1}{z}e^{-kz},\quad { \mathcal{f}\left[\frac{x}{r^3}\right] } = \frac{ik_1}{k}e^{-kz}\end{aligned}\]] where @xmath223.
Reorientation of a stokes dipole close to a partial slip boundary
we present here the details of the derivation for the rotation rate, @xmath132, induced by a partial slip boundary on a tilted stokes dipole. the method is general, and the same procedure is applied throughout the paper for deriving the rotation rate induced by a complex interface. in fourier space, the rotation rate induced by a flow @xmath224 on a prolate spheroid of aspect ratio @xmath37, oriented along @xmath36, reads @xmath225 & = \frac{1}{2 } \begin{pmatrix } -ik_2{\tilde{u}}_z-{\frac{\partial { \tilde{u}}_2}{\partial z}}\\ { \frac{\partial { \tilde{u}}_1}{\partial z}}+ik_1{\tilde{u}}_z\\ ik_2{\tilde{u}}_1-ik_1{\tilde{u}}_2 \end{pmatrix } \nonumber\\ & \quad+ \frac{1}{2}{\frac{\gamma^2 - 1}{\gamma^2 + 1}}\begin{pmatrix } { \cos { \theta}}{\sin { \theta}}[ik_1{\tilde{u}}_2+ik_2{\tilde{u}}_1]-{\sin^2 \!{\theta}}\left[{\frac{\partial { \tilde{u}}_2}{\partial z}}-ik_2{\tilde{u}}_z\right]\\ [{ \sin^2 \!{\theta}}-{\cos^2 \!{\theta}}][{\frac{\partial { \tilde{u}}_1}{\partial z}}-ik_1{\tilde{u}}_z]-2{\cos { \theta}}{\sin { \theta}}\left[ik_1{\tilde{u}}_1+{\frac{\partial { \tilde{u}}_z}{\partial z}}\right]\\ { \cos { \theta}}{\sin { \theta}}[{\frac{\partial { \tilde{u}}_2}{\partial z}}-ik_2{\tilde{u}}_z]-{\cos^2 \!{\theta}}\left[ik_1{\tilde{u}}_2+ik_2{\tilde{u}}_1\right] \end{pmatrix}. \label{eq : om_gen}\end{aligned}\]] the tilted stokeslet dipole has three contribution, a stresslet, @xmath226, and two stokeslet dipoles, @xmath145 and @xmath146. we compute first the stresslet contribution. for a stresslet @xmath226, we have @xmath227, where @xmath228 the coefficients of the no - slip solution read @xmath229 we note @xmath230 $] and @xmath231 $]. from the expression of the rotation rate in eq. , we have @xmath232. from eq., we note that @xmath233, and thus @xmath234. this decomposition holds in real space,@xcite leading to @xmath235 the resulting differential equation reads @xmath236 we decompose then @xmath237 in two terms, @xmath238, so that @xmath239 this last equation can be rewritten as @xmath240 the first term, @xmath241, can be computed directly knowing the no - slip solution. however, the second term is not straightforward, and needs to be determined in fourier space, as well as @xmath242. we have then @xmath243 e^{-kz }, \\ { \tilde{\omega}}^2 & = -\frac{1}{8\pi{\eta}_1}\frac{{\ell}k_2 ^ 2 e^{-k(z+h)}}{({1+{\ell}k})({1 + 2{\ell}k})}\left[1+{\frac{\gamma^2 - 1}{\gamma^2 + 1}}\left({\sin^2 \!{\theta}}-{\cos^2 \!{\theta}}+ 2{\cos { \theta}}{\sin { \theta}}\frac{ik_1}{k}\right)\right], \label{eq : bug}\end{aligned}\]] these expressions can be inverted, as all contributions are known fourier transforms (see appendix [sec : appb]). we have therefore @xmath244, with @xmath245,\\ & \left(1-{\ell}{\frac{\partial } { \partial z}}\right)\left(1 - 2{\ell}{\frac{\partial } { \partial z}}\right)\omega^2 = \frac{{\ell}}{8\pi{\eta}_1 } \left[1+{\frac{\gamma^2 - 1}{\gamma^2 + 1}}({\sin^2 \!{\theta}}-{\cos^2 \!{\theta}})\right]{\frac{\partial } { \partial z}}\left(\frac{1}{{r}^3}-\frac{3y^2}{{r}^5}\right)\nonumber\\ & \qquad \qquad \qquad \qquad \qquad \qquad\quad + \frac{{\ell}}{8\pi{\eta}_1 } { \frac{\gamma^2 - 1}{\gamma^2 + 1}}\sin(2{\theta}){\frac{\partial ^2}{\partial y^2}}\left(\frac{x}{{r}^3}\right),\end{aligned}\]] where @xmath246, with @xmath124. since we are looking for the solution at @xmath122, we need to integrate the following equations @xmath247, \\ & \left(1 - 2{\ell}{\frac{\mathrm{d } } { \mathrm{d } { z}}}\right)^2\omega_{ss}^{11 } = -\frac{3{\ell}({\sin^2 \!{\theta}}-{\cos^2 \!{\theta}})}{2\pi{\eta}_1 } { \frac{\gamma^2 - 1}{\gamma^2 + 1}}\frac{1}{{z}^4}\left(1-\frac{4h}{{z}}\right),\\ & \left(1-{\ell}{\frac{\mathrm{d } } { \mathrm{d } { z}}}\right)\left(1 - 2{\ell}{\frac{\mathrm{d } } { \mathrm{d } { z}}}\right)\omega_{ss}^2 = -\frac{3{\ell}}{8\pi{\eta}_1 } \left[1+{\frac{\gamma^2 - 1}{\gamma^2 + 1}}({\sin^2 \!{\theta}}-{\cos^2 \!{\theta}})\right] \frac{1}{{z}^4}\cdot\end{aligned}\]] the final rotation rate due to @xmath63 reads @xmath248. for the parallel stokes dipole, @xmath145, we have @xmath249.\end{aligned}\]] the corresponding coefficients in fourier space are given by @xmath250 following the same procedure, we find that the rotation rate, @xmath132, induced by a parallel stokes dipole @xmath145 is given by @xmath251 where @xmath252 for the perpendicular stokes dipole, @xmath146, we have @xmath253.\end{aligned}\]] the rotation rate, @xmath132, induced by a perpendicular stokes dipole @xmath146 is given by @xmath254, with @xmath255
Typical differential equations and solutions
the differential equations in @xmath256 giving the rotation rate and velocities induced by the nearby boundary are of the three following types @xmath257 where @xmath59 is a positive integer. we keep here the coefficients corresponding to the partial slip case, however similar equations are obtained in the case of a surfactant - covered interface. knowing that the solution must vanish at infinity, the solutions for these equations read @xmath258 where @xmath159 is the exponential integral function of order @xmath59 defined in eq. . 45ifxundefined [1] ifx#1 ifnum [1] # 1firstoftwo secondoftwo ifx [1] # 1firstoftwo secondoftwo `` `` # 1''''@noop [0]secondoftwosanitize@url [0] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop _ _ (, , ) link:\doibase 10.1119/1.10903 [* *, ()] @noop * * () @noop _ _ (, ) @noop * *, () link:\doibase 10.1146/annurev.mi.49.100195.003431 [* *, ()] link:\doibase 10.1007/bf02461846 [* *, ()] link:\doibase 10.1146/annurev.micro.57.030502.091014 [* *, ()] link:\doibase 10.1073/pnas.1019079108 [* *, ()] link:\doibase 10.1098/rspa.2009.0520 [* *, ()] link:\doibase 10.1073/pnas.1202934109 [* *, ()] link:\doibase 10.1038/1981221a0 [* *, ()] link:\doibase 10.1103/physrevlett.101.038102 [* *, ()] link:\doibase 10.1017/s0022112008004953 [* *, ()] link:\doibase 10.1103/physreve.82.056309 [* *, ()] link:\doibase 10.1103/physreve.84.041932 [* *, ()] link:\doibase 10.1016/s0006 - 3495(90)82436-x [* *, ()] link:\doibase 10.1529/biophysj.105.069401 [* *, ()] link:\doibase 10.1128/jb.00397 - 10 [* *, ()] link:\doibase 10.1103/physrevlett.106.038101 [* *, ()] @noop _ _, butterworth - heinemann series in chemical engineering (, , ) link:\doibase 10.1039/c3sm51426a [* *, ()] \doibase http://dx.doi.org/10.1016/j.bpj.2013.05.026 [* *, ()] @noop * *, () link:\doibase 10.1080/15583724.2011.615961 [* *, ()] link:\doibase 10.1021/ma971339h [* *, ()] link:\doibase 10.1103/physrevlett.101.026101 [* *, ()] link:\doibase 10.1016/s0001 - 8686(00)00077 - 4 [* *, ()] @noop () link:\doibase 10.1146/annurev - chembioeng-061010 - 114202 [* *, ()] link:\doibase 10.1063/1.869875 [* *, ()] link:\doibase 10.1103/physreve.87.063010 [* *, ()] @noop * *, () link:\doibase 10.1063/1.857686 [* *, ()] link:\doibase 10.1017/s0022112075000614 [* *, ()] @noop * *, () @noop * *, () link:\doibase 10.1017/jfm.2012.101 [* *, ()] link:\doibase 10.1103/physrevlett.89.058101 [* *, ()] link:\doibase 10.1103/physrevlett.95.204501 [* *, ()] link:\doibase 10.1063/1.3041776 [* *, ()] link:\doibase 10.1146/annurev - fluid-121108 - 145434 [* *, ()] link:\doibase 10.1007/bf02353701 [* *, ()] link:\doibase 10.1021/la980465r [* *, ()] link:\doibase 10.1529/biophysj.105.062653 [* *, ()] | flagellated bacteria exploiting helical propulsion are known to swim along circular trajectories near surfaces.
fluid dynamics predicts this circular motion to be clockwise (cw) above a rigid surface (when viewed from inside the fluid) and counter - clockwise (ccw) below a free surface.
recent experimental investigations showed that complex physicochemical processes at the nearby surface could lead to a change in the direction of rotation, both at solid surfaces absorbing slip - inducing polymers and interfaces covered with surfactants.
motivated by these results, we use a far - field hydrodynamic model to predict the kinematics of swimming near three types of interfaces : clean fluid - fluid interface, slipping rigid wall, and a fluid interface covered by incompressible surfactants.
representing the helical swimmer by a superposition of hydrodynamic singularities, we first show that in all cases the surfaces reorient the swimmer parallel to the surface and attract it, both of which are a consequence of the stokes dipole component of the swimmer flow field.
we then show that circular motion is induced by a higher - order singularity, namely a rotlet dipole, and that its rotation direction (cw vs. ccw) is strongly affected by the boundary conditions at the interface and the bacteria shape.
our results suggest thus that the hydrodynamics of complex interfaces provide a mechanism to selectively stir bacteria. | 1406.4412 |
Introduction
the m31 globular cluster (gc) b158 (a.k.a. bo 158), named following the revised bologna catalogue v3.4 @xcite, contains a bright x - ray source that was discovered by the einstein observatory @xcite, and has shown up in every x - ray observation of the region since ; we call this x - ray source xb158. xb158 exhibited strong intensity modulation on a 10017@xmath050 s (@xmath12.78 hr) period during the 2002 january xmm - newton observation @xcite. @xcite found similar variation in the folded lightcurves from a 1991 june rosat observation and a 2000 june xmm - newton observation ; they found that the amplitude of modulation decreased with increasing source intensity. they found no such modulation in the 2001 june xmm - newton observation, setting a 2@xmath3 upper limit of 10% modulation. assuming that this represents the orbital period, @xcite found that this is probably a neutron star binary with a low mass donor with a separation @xmath510@xmath6 cm (i.e. a low mass xb - ray binary, lmxb). however, analysis of the unfolded 2000 june xmm - newton lightcurve revealed a single deep dip at the end of the observation, with no evidence for dips in the two previous orbital cycles @xcite. furthermore, @xcite analyzed 3 proprietary xmm - newton observations over 2004 july 1719, finding @xmath1100% dipping for one orbital cycle, and zero evidence for dips in other cycles ; we concluded that the disk was precessing, with dips only visible for some part of the super - orbital cycle. such behavior is associated with the `` superhump '' phenomenon that is observed in accreting binaries where the mass ratio is smaller than @xmath10.3 @xcite. superhumps were first identified in the superoutbursts of the su uma subclass of cataclysmic variables (accreting white dwarf binaries), manifesting as a periodic increase in the optical brightness on a period that is slightly longer than the orbital period @xcite. su umas are a subclass of dwarf novae with orbital periods @xmath72 hr, that exhibit occasional superoutbursts that last @xmath85 times as long as the normal outbursts @xcite. @xcite proposed that these superoutbursts are enhanced by a tidal instability that occurs when the outer disk crosses the 3:1 resonance with the secondary ; the additional tidal torque causes the disk to elongate and precess, and also greatly enhances the loss of angular momentum (and therefore the accretion rate). the disk precession is prograde in the rest frame, and the secondary repeats its motion with respect to the disk on the beat period between the orbital period and the precession period, a few percent longer than the orbital period. the secondary modulates the disk s viscous dissipation on this period, giving rise to the maxima in the optical lightcurve known as superhumps. some short period, persistently bright cvs exhibit permanent superhumps @xcite. cccccccccccc obs & @xmath9 & ra & dec & xrt exp & @xmath10/@xmath11 & counts + + 00032702001 & 0.0 & 00 43 00.81 & + 41 17 00.2 & 2234 & 10.0 & 57 + 00032702002 & 0.9 & 00 43 06.53 & + 41 16 54.6 & 2510 & 9.7 & 41 + 00032702003 & 2.0 & 00 43 03.73 & + 41 16 40.8 & 2299 & 9.5 & 13 + 00032702004 & 3.0 & 00 43 06.35 & + 41 16 51.3 & 2299 & 9.6 & 40 + 00032702005 & 3.1 & 00 43 13.10 & + 41 16 30.4 & 2489 & 9.2 & 43 + 00032702006 & 4.4 & 00 43 03.22 & + 41 13 58.5 & 2572 & 7.0 & 65 + 00032702007 & 5.4 & 00 43 10.44 & + 41 15 10.9 & 2511 & 7.9 & 51 + 00032702008 & 6.4 & 00 43 06.44 & + 41 16 06.4 & 2635 & 8.9 & 57 + 00032702009 & 7.3 & 00 43 11.97 & + 41 16 27.7 & 2476 & 9.1 & 29 + 00032702010 & 8.8 & 00 43 03.76 & + 41 16 15.5 & 2644 & 9.1 & 39 + 00032702011 & 9.5 & 00 43 08.40 & + 41 17 27.6 & 2405 & 10.2 & 53 + 00032702012 & 10.1 & 00 43 07.20 & + 41 16 29.8 & 2502 & 9.2 & 43 + 00032702013 & 11.2 & 00 43 03.63 & + 41 17 36.2 & 2514 & 10.5 & 48 + 00032702014 & 12.4 & 00 43 04.82 & + 41 17 00.3 & 2411 & 9.8 & 38 + 00032702015 & 13.2 & 00 43 07.68 & + 41 15 24.0 & 2490 & 8.2 & 11 + 00032702016 & 14.2 & 00 43 01.12 & + 41 15 11.3 & 2511 & 8.2 & 27 + 00032702017 & 15.1 & 00 43 06.35 & + 41 14 46.6 & 2273 & 7.6 & 54 + 00032702018 & 16.8 & 00 43 02.79 & + 41 17 00.9 & 2314 & 9.9 & 55 + 00032702019 & 17.1 & 00 43 08.59 & + 41 15 26.4 & 1992 & 8.2 & 45 + 00032702020 & 18.1 & 00 43 04.88 & + 41 17 31.5 & 1389 & 10.3 & 13 + 00032702021 & 19.4 & 00 43 03.24 & + 41 17 28.0 & 2690 & 10.3 & 18 + 00032702022 & 20.2 & 00 43 02.68 & + 41 17 26.6 & 2686 & 10.3 & 42 + 00032702023 & 21.1 & 00 43 03.07 & + 41 17 07.0 & 2821 & 10.0 & 65 + 00032702024 & 22.3 & 00 43 04.56 & + 41 16 19.5 & 2659 & 9.2 & 50 + 00032702025 & 23.2 & 00 43 10.49 & + 41 16 54.1 & 2711 & 9.6 & 46 + 00032702026 & 24.3 & 00 43 07.86 & + 41 15 05.1 & 2740 & 7.8 & 35 + 00032702027 & 25.1 & 00 43 06.42 & + 41 16 41.0 & 2424 & 9.5 & 18 + 00032702028 & 26.1 & 00 43 10.04 & + 41 16 54.2 & 2550 & 9.6 & 49 + 00032702029 & 27.2 & 00 43 10.52 & + 41 17 33.9 & 2524 & 10.2 & 59 + 00032702030 & 28.4 & 00 43 06.50 & + 41 17 25.9 & 2331 & 10.2 & 41 + 4u 1916@xmath12053 is a high inclination neutron star lmxb with an x - ray period of 50.00@xmath00.08 min and an optical period 50.458@xmath00.003 min @xcite. it exhibits periodic x - ray intensity dips due to absorption by material in the outer disk ; the amplitude of these dips varies over @xmath10 to @xmath1100% on a @xmath14 day period, the precession period of the disk @xcite. @xcite showed that ns lmxbs with orbital periods shorter than @xmath14.2 hr are likely to exhibit superhumps, and identified 4u 1916 - 053 as a persistent superhumping source. xb158 appears to be somewhat analogous to 4u 1916@xmath12053 @xcite. in @xcite we modeled the 2004 july 17 xmm - newton pn spectrum of xb158 with a blackbody and a power law, finding @xmath13 = 2.0@xmath00.2 kev, the photon index @xmath14 = 2.0@xmath00.3, @xmath4/dof = 19/19 and the 0.310 kev luminosity was 1.5@xmath00.6@xmath15 erg s@xmath16 ; fitting a single power law emission model yielded a photon index of 0.57@xmath00.09, which is harder than any black hole spectrum @xcite, meaning that the accretor is a neutron star. we conducted three dimensional smoothed particle hydrodynamical (sph) modeling, assuming a 1.4 @xmath17 ns, a 2.78 hr period, a total system mass of 1.8 @xmath17, and a luminosity @xmath180% of the eddington limit. as a result, we estimated the disk precession period to be 29@xmath01 times the orbital period, i.e. 81@xmath03 hr. in @xcite we examined chandra observations of @xmath130 m31 globular cluster x - ray sources spanning @xmath112 years (89 acis and 45 hrc), including xb158. we observed 0.310 kev luminosities @xmath1420@xmath18 erg s@xmath16, and proposed that this is due to varying accretion rates over the disk precession cycle. since we expected a super - orbital period for xb158 over time - scales of a few days, we obtained a series of 30 swift observations, each with 2.5 ks exposure and spaced @xmath11 day apart. in this work we present our analysis of these swift results, and find evidence for a super - orbital period of @xmath16 days
Observations and data analysis
we made 30 swift observations over 2013 february 8 march 9 (obsids 00327020010032702030, pi r. barnard). xb158 was one of two main targets for this survey, and the pointing was chosen to optimize the results from both targets. however, the actual pointings varied significantly over the 30 observations. a journal of swift observations is provided in table [journ] ; for each observation we give the time relative to the first observation, pointing, xrt exposure, and off - axis angle, along with the net source counts. for each observation, we placed circular regions around xb158 and a suitable background region. we obtained the net source counts using the `` counts in regions '' tool in the ds9 image viewer, and estimated the intensity by dividing the net counts by the exposure time. these data were obtained in order to determine the extent to which the varying off - axis angles affected our results. we also created spectra from the same extraction regions, created appropriate ancillary response files using xrtmkarf, and found the appropriate response file using the quzcif tool. none of the spectra were suitable for free spectral fitting, hence we obtained luminosity estimates by assuming a model obtained from previous observations. for chandra acis observations with @xmath2200 net source photons, the mean line - of - sight absorption (@xmath19) was 9@xmath04@xmath20 atom @xmath21 (@xmath4/dof = 7/11), and the mean power law index (@xmath14) was 0.52@xmath00.03 (@xmath4/dof = 5/11). this is consistent with the best absorbed power law fit to the 2004 july 17 xmm - newton observation (@xmath19 = 0.1, @xmath14 = 0.57@xmath00.09, @xmath4/dof = 30/24 *??? as we noted earlier, that xmm - newton spectrum was best described by a blackbody + power law model, but neither the chandra nor swift spectra were sufficient to constrain the two - component emission model. however, we were able to estimate the luminosity for each observation by assuming a fixed emission model, allowing only the normalization to vary. we fitted each spectrum using xspec 12.8.2b, fixing @xmath19 = 9@xmath20 atom @xmath21 and @xmath14 = 0.52, to find the normalization required to make absorbed model intensity 1.00 count s@xmath16. we then calculated the unabsorbed flux for this model, allowing us to convert from intensity to flux. multiplying the conversion by the background - subtracted intensity provided by xspec yielded the instrument - corrected, background - subtracted source flux. the luminosity was calculated from the flux assuming a distance of 780 kpc @xcite. we fitted the lightcurve with constant and sinusoidal components using the qdp program provided in heatools, performing a simple search for periodicity. @xcite created a periodogram suitable for unbinned data with a mean of zero that produces exactly equivalent results to such least - squares fitting of sinewaves, but also allows comparison of the best period with other periods. the likelihood that any peak in the periodogram is real is given by the false alarm probability (@xmath22), where a low value of @xmath22 indicates that the peak is likely to be significant. if the highest frequency to be probed is @xmath23 times higher than the lowest frequency, then the power, @xmath24, required for a false alarm probability @xmath22 is given by @xmath24 = @xmath25 $] ; for small @xmath22, @xmath26 @xcite.
Results
we present our 30 day, 0.310 kev swift xrt lightcurve of xb158 in figure [swiftlc]. we see that the 0.310 kev luminosity varied by a factor @xmath15 ; the luminosity dropped from @xmath12.3@xmath15 erg s@xmath16 to @xmath15@xmath18 erg s@xmath16 in @xmath12 days, assuming the mean chandra absorbed power law model. we note that these swift observations are non - contiguous, spacing the 2.5 ks observing time over several hours ; the low intensities (@xmath10.0050.025 count s@xmath16) meant that there was no appreciable variability within each observation. we find no evidence for a dependence of luminosity on off - axis angle ; the luminosities for the observations when xb158 have the largest and smallest off - axis angles have consistent values, while observations at an off - axis angle of 8.2@xmath11 resulted in a factor @xmath15 range in luminosity. the conversion factor for translating 1 count s@xmath16 into 0.310 kev flux ranged over 1.091.29@xmath27 erg @xmath21 count@xmath16 for all observations except the second one, where it was 1.79@xmath27 erg @xmath21 count@xmath16. this @xmath110% variation about the mean in instrumental correction is clearly not sufficient to account for the factor @xmath15 variation in luminosity. fitting our lightcurve with a constant intensity yielded a best fit luminosity of @xmath11.3@xmath15 erg s@xmath16, with @xmath4 = 181 for 29 degrees of freedom (dof). however, adding a sinusoidal variation component yielded a much improved fit, with @xmath4/dof = 43/26 ; for this model, the period is 5.65@xmath00.05 days, with an amplitude of 7.1@xmath00.6@xmath18 erg s@xmath16 around a mean luminosity of 1.43@xmath00.04@xmath15 erg s@xmath16, and a phase of 88.1@xmath00.7 degrees. all uncertainties in this work are quoted at the 1@xmath3 level. this sinusoidal variability yielded @xmath28 = 138 for 30 bins, with 3 extra free parameters ; f - testing showed that the probability for this improvement being due to chance was 3@xmath29, equivalent to a @xmath25@xmath3 detection. in figure [swift_per] we show our swift lightcurve folded on a 5.65 day period and fitted with the best fit sinusoid. we present our lomb - scargle periodogram for the 0.310 kev swift xrt luminosity lightcurve of xb158 in figure [ls], and also indicate the power required for false alarm probabilities @xmath22 = 0.5, 0.05, and 0.005 for reference. we tested 30 frequencies, and oversampled each frequency by a factor of 50 ; this resulted in a periodogram covering a wider range of periods than is interesting (up to @xmath11500 days), so we show only part of the periodogram here. the periodogram shows a single strong peak, with the maximum power of 10.4 at a period of 5.60 days, corresponding to @xmath22 = 0.0009 ; the range of periods which are detected at a @xmath23@xmath3 level is 5.45.8 days. while there is a small peak at the 3 day period, @xmath22 = 1 for this peak
Discussion and conclusions
xb158 is a high inclination x - ray binary associated with the m31 globular cluster b158. it exhibits deep intensity dips on a 2.78 hr period in some observations but not others, prompting @xcite to suggest that the disk is precessing, caused by the `` superhumping '' phenomenon observed in low mass ratio systems where the disk crosses the 3:1 resonance with the donor star. @xcite predicted a disk precession period of 29@xmath01 times the orbital period, i.e. 81@xmath03 hr, inspiring a month of daily monitoring of the m31 central region by swift. fitting a sinusoid to the lightcurve revealed a 5.65@xmath00.05 day super - orbital period (1@xmath3 uncertainties) ; the 0.310 kev luminosity varied by a factor @xmath15, which is consistent with the range of luminosities observed in the acis observations of our 13 + year chandra monitoring campaign @xcite. the peak of the lomb - scargle periodogram is at 5.6 days, consistent with that obtained from least squares fitting of the lightcurve with a sinusoid. the suggested super - orbital period is @xmath170% longer than predicted by our 3d sph simulations @xcite. none of the authors of the current paper are experts in sph ; however, j. r. murray stated in a private communication that the longer than expected super - orbital period is likely due to the mass ratio of the donor to the accretor being lower than assumed. the mass of a roche lobe filling main sequence star, @xmath30, may be approximated as @xmath30 @xmath31 0.11 @xmath32, where @xmath32 is the orbital period in hours @xcite, and we originally assumed a donor mass of @xmath10.30 @xmath17. for the accretor, we assumed a 1.4 @xmath33 neutron star. a power law emission model fit to the 2004 july 16 xmm - newton spectrum of b158 yielded a photon index of 0.57@xmath00.09 @xcite, which is considerably harder than any spectrum emitted by a black hole binary. however, some neutron stars have masses @xmath22@xmath17 (see e.g. *??? * ; *??? *), and it is possible that xb158 contains a particularly massive neutron star. we noted in @xcite that other xbs such as xb146 exhibited strong luminosity fluctuations between fairly consistent maxima and minima during our chandra monitoring observations, and suggested that this long - term behavior could be indicative of a short period / low mass ratio system. our new findings support this hypothesis.
Acknowledgments
we thank the anonymous referees for suggesting improvements to this work, in particular prompting more rigorous estimation of the super - orbital period. we thank the swift team for making this work possible. this work was funded by swift grant nnx13aj76 g. | the m31 globular cluster x - ray binary xb158 (a.k.a. bo 158) exhibits intensity dips on a 2.78 hr period in some observations, but not others.
the short period suggests a low mass ratio, and an asymmetric, precessing disk due to additional tidal torques from the donor star since the disk crosses the 3:1 resonance.
previous theoretical 3d smoothed particle hydrodynamical modeling suggested a super - orbital disk precession period 29@xmath01 times the orbital period, i.e. @xmath181@xmath03 hr. we conducted a swift monitoring campaign of 30 observations over @xmath11 month in order to search for evidence of such a super - orbital period. fitting the 0.310 kev swift xrt luminosity lightcurve with a sinusoid yielded a period of 5.65@xmath00.05 days, and a @xmath25@xmath3 improvement in @xmath4 over the best fit constant intensity model.
a lomb - scargle periodogram revealed that periods 5.45.8 days were detected at a @xmath23@xmath3 level, with a peak at 5.6 days.
we consider this strong evidence for a 5.65 day super - orbital period, @xmath170% longer than the predicted period.
the 0.310 kev luminosity varied by a factor @xmath15, consistent with variations seen in long - term monitoring from chandra.
we conclude that other x - ray binaries exhibiting similar long - term behaviour are likely to also be x - ray binaries with low mass ratios and super - orbital periods. | 1501.01978 |
Introduction
filaments are the most prominent features visible in the galaxy distribution. this finding dates back to a few papers in the seventies and eighties @xcite. subsequent work substantiates this (e.g. @xcite, @xcite, @xcite, @xcite, @xcite, @xcite, @xcite, @xcite) and shows the filaments to be statistically significant @xcite. it is now well accepted that galaxies are distributed in an interconnected network of clusters, sheets and filaments encircling voids. this complicated pattern is often referred to as the cosmic web. despite this progress, it still remains a challenge to quantify the cosmic web that is so distinctly visible in galaxy redshift surveys (eg. sdss dr5, @xcite). statistical measures like the void probability function @xcite, percolation analysis @xcite and the genus curve @xcite each quantifies a different aspect of the cosmic web. the minkowski functionals @xcite are very effective to quantify the shapes of individual structural elements like sheets or filaments. in @xmath5 dimensions there are @xmath10 minkowski functionals, namely the volume, surface area, integrated mean curvature and integrated gaussian curvature. @xcite introduce the shapefinders, essentially ratios of the minkowski functionals, as a very effective shape diagnostic. a @xmath6 dimensional version of shapefinders @xcite has been extensively used to quantify the filamentarity in the galaxy distribution (@xcite and references therein). centered on a galaxy located in the filament.,scaledwidth=40.0%] though the minkowski functionals and the shapefinders are very effective techniques to quantify the shapes of individual structural elements like sheets or filaments, it is very different when dealing with the cosmic web which is an interconnected network of filaments, sheets and clusters. for example consider a sheet connected to a filament as shown in figure [fig : exp1]. the minkowski functionals are global properties of the entire object the area is the sum of the areas of the sheet and the filament etc., and the fact that object is actually a combination of two different elements would be lost. it is necessary to quantify the local shape at different points in the object in order to determine that it actually is a combination of a sheet and a filament. in this paper we consider the `` local dimension '' as a means to quantify the local shape of the galaxy distribution at different positions along the cosmic web.
Method of analysis
we choose a particular galaxy as center and determine @xmath2 the number of other galaxies within a sphere of comoving radius @xmath3. this is done varying @xmath3. in the situation where a power law @xmath11 gives a good fit over the length - scales @xmath12, we identify @xmath0 as the local dimension in the neighbourhood of the center. the values @xmath13 and @xmath5 correspond to a filament, sheet and cluster respectively. it may be noted that the term `` cluster '' here denotes a three dimensional, volume filling structural element and is not to be confused with a `` cluster of galaxies ''. values of @xmath0 other than @xmath14 and @xmath5 are more difficult to interpret. for example, a galaxy distribution that is more diffuse than a filament but does not fill a plane would give a fractional value (fractal) in the range @xmath15. referring to figure [fig : exp1], we expect @xmath16 and @xmath17 when the center is located in the filament and the sheet respectively. this is provided that the center is well away from the intersection of the filament and the sheet. when the intersection lies within @xmath12 from the center, there will be a change in the slope of @xmath2 when it crosses the intersection. it is not possible to determine a local dimension at the centers where such a situation occurs. we perform this analysis using every galaxy in the sample as a center. in general it will be possible to determine a local dimension for only a fraction of the galaxies. it is expected that with a suitable choice of the @xmath3 range _ ie. _ @xmath18 and @xmath19, it will be possible to determine the local dimension for a substantial number of the centers. the value of the local dimension at different positions will indicate the location of the filaments, sheets and clusters and reveal how these are woven into the cosmic web. in this _ letter _ we test this idea and demonstrate its utility by applying it to simulations. we have used a particle - mesh (pm) n - body code to simulate the @xmath20 dark matter distribution. the simulations have @xmath21 particles on a @xmath21 mesh with grid spacing @xmath22. the simulations were carried out using a lcdm power spectrum with the parameters @xmath23. we have identified @xmath24 particles, randomly drawn from the simulation output, as galaxies. these have a mean interparticle separation of @xmath25, comparable to that in galaxy surveys. this simulated galaxy distribution was carried over to redshift space in the plane parallel approximation. the subsequent analysis to determine the local dimension was carried out using this simulated sample of @xmath24 galaxies. since the resolution of the simulation is about @xmath26, we ca nt choose @xmath18 to be less than that. the value of @xmath19 is determined by the limited box size. we have chosen the value of @xmath18 and @xmath19 to be @xmath6 and @xmath27 respectively. increasing @xmath19 causes a considerable drop in the number of centers for which the local dimension is defined. the analysis was carried out for @xmath28 different, independent realizations of the dark matter distribution. figure [fig : exp2] shows @xmath2 for three different centers chosen from a particular realization. the @xmath29 error at each data point is @xmath30 due to the poisson fluctuation. for each center we have determined the power law @xmath1 that provides the best fit to the data. the power law fit is accepted if the chi - square per degree of freedom satisfies @xmath31 and the value of @xmath0 is accepted as the local dimension corresponding to the particular center. the power law fit is rejected for larger values of @xmath32, and the local dimension is undetermined for the particular center. the criteria @xmath31 is chosen as a compromise between a good power law fit and a reasonably large number of centers for which @xmath0 can been determined. the number of centers for which @xmath0 can been determined falls if a more stringent criteria is imposed on @xmath32.
Results and conclusions
of a center with @xmath9 are shown as points. the galaxy distribution was converted to a set of 1s and 0s on a grid of spacing @xmath33. connected structures were identified using the friend - of - friend algorithm where adjacent 1s are identified as belonging to the same connected structure. only the two largest structures have been shown, both pass very close to the center and they appear connected in the figure. the @xmath0 values have been painted on these structures with red, blue and green representing @xmath4 and @xmath5 respectively. the @xmath0 value is undetermined for the parts of the structures shown in black.] the number of centers for which it is possible to determine a local dimension varies between @xmath34 to @xmath35 _ ie. _ less than @xmath36 variation across the @xmath28 different realizations. the @xmath0 values in the interval @xmath37, @xmath38 and @xmath39 have respectively been binned as @xmath4 and @xmath5. we have inspected the galaxy distribution in the vicinity of a few centers in order to visually identify the structures corresponding to the respective @xmath0 values. figure [fig : fil1n] shows the galaxy distribution within @xmath7 of a center which has local dimension @xmath40. we expect this center to be located in a filament. we have used the friend - of - friend algorithm with linking length @xmath41 to identify connected patterns in the galaxy distribution. we note that all the galaxies in this region connect up into a single structure if the linking length is doubled to @xmath42. using @xmath41 yields several disconnected structures of which we show only the largest two. both the structures pass close to the center, and appear connected in the figure. of the two structures, one with two well separated tentacles appears below the center, whereas the other with two nearly connected tentacles appears above it. the filamentary nature of these structures is quite evident. out of the total @xmath43 galaxies in these two structures, @xmath0 is undetermined for @xmath44. there are @xmath45 galaxies with @xmath46, all located in a contiguous region near the center, and @xmath47 and @xmath48 galaxies with @xmath49 and @xmath5 respectively spread out along the structure at locations quite different from the center. while it is relatively easy to visually identify the filament corresponding to the centers with @xmath50, we have not been able to visually identify the sheets and clusters corresponding to the centers with @xmath49 and @xmath5 shown in figure [fig : fil1n]. we note that a visual identification of the sheets and clusters corresponding to @xmath51 and @xmath5 respectively has been possible in other fields where the largest structure is predominantly sheetlike or clusterlike. it is quite apparent that the largest structure in figure [fig : fil1n] is predominantly filamentary. values in a box of size @xmath52 ^ 3 $], red, blue and green denote @xmath4 and @xmath5 respectively.] figure [fig : allshown] shows the distribution of @xmath0 values over a large region from one of the realizations. the distribution shows distinct patterns, there being regions with size ranging from a few @xmath53 to tens of @xmath54 where the @xmath0 value is constant. the centers with @xmath9 appear to have a more dense and compact distribution compared to the centers with @xmath51, whereas the centers with @xmath55 appear to have a rather diffuse distribution. figure [fig : hist1] separately quantifies two different aspects of the distribution of @xmath0 values, (a.) the fraction of centers with a particular @xmath0 value, and (b.) the fraction of volume with a particular @xmath0 value. for this analysis the @xmath0 values were divided into bins of width @xmath56 and the centers for which a @xmath0 value could not be determined were discarded. we find that the fraction of centers and the volume fraction both have very robust statistics with a variation of @xmath57 or less across the @xmath28 different realizations. we first consider the distribution of the fraction of centers with different @xmath0 values. the bin at @xmath58 contains the maximum number of centers, and the two bins at @xmath58 and @xmath6 together contain more than @xmath59 of the centers. this indicates that the maximum number of centers for which @xmath0 can be determined lie in sheets and filaments. assuming that this is representative of the entire matter distribution, we conclude that the bulk of the matter is contained in sheets and filaments, with the sheets dominating. we next consider the the fraction of volume occupied by a particular @xmath0 value. we have estimated this using a mesh with grid spacing @xmath60. each grid position was assigned the @xmath0 value of the centers within the corresponding grid cell. cells that contain centers with different values of @xmath0 were discarded. it was possible to assign a @xmath0 value to @xmath61 grid cells. we find that the distribution peaks at @xmath62, the peak is rather broad and more than @xmath59 of the volume for which it was possible to assign a @xmath0 value is occupied by values in the range @xmath6 to @xmath5. assuming that this result is valid for the entire volume, we conclude that the volume is predominantly filled with sheets and clusters. while the matter is mainly contained in sheets and filaments, the filaments do not occupy a very large fraction of the volume. this leads to a picture where the filaments have relatively higher matter densities as compared to clusters. to test this we separately consider the centers with @xmath0 values @xmath14 and @xmath5, and for each value we determine the fraction of centers as a function of the density environment. the density was calculated on a @xmath63 grid after smoothing with a gaussian filter of width @xmath64. we find (figure [fig : hist2]) that in comparison to the centers with @xmath9 and @xmath6, the centers with @xmath55 are more abundant in under - dense regions in preference to the over - dense regions. this is consistent with the picture where filaments and sheets located along the cosmic web contain the bulk of the matter and also most of the high density regions, whereas the centers with @xmath65 are predominantly located in under - dense regions away from the cosmic web, namely the voids. the results presented above are specific to the length - scale range @xmath66. it is quite possible that the relative abundance of clusters, sheets and filaments would be quite different if the same analysis were carried out over a considerably different range of length - scales, say for example @xmath67. here we have repeated the entire analysis using a smaller range of length - scales @xmath68 for which the results are also shown in figure [fig : hist1]. we find that the results are qualitatively similar to those obtained using @xmath66, indicating that our conclusions regarding the relative abundance of clusters, sheets and filaments are quite robust and are not very sensitive to small changes in the range over which @xmath0 is determined. this also indicates that the @xmath0 values determined using length - scales @xmath69 are not severely biased by the presence of other structural elements with @xmath70 of the one on which the center is located. in conclusion we note that the local dimension provides a robust method to quantify the shapes and probe the distribution of the different, interconnected structural elements that make up the cosmic web. we plan to apply this to the sdss and other galaxy surveys in future.
Acknowledgment
p.s. is thankful to biswajit pandey, kanan datta and prasun dutta for useful discussions. p.s. would like to acknowledge senior research fellow of university grants commission (ugc), india. for providing financial support. | it is now well accepted that the galaxies are distributed in filaments, sheets and clusters all of which form an interconnected network known as the cosmic web.
it is a big challenge to quantify the shapes of the interconnected structural elements that form this network.
tools like the minkowski functionals which use global properties, though well suited for an isolated object like a single sheet or filament, are not suited for an interconnected network of such objects.
we consider the local dimension @xmath0, defined through @xmath1, where @xmath2 is the galaxy number count within a sphere of comoving radius @xmath3 centered on a particular galaxy, as a tool to locally quantify the shape in the neigbourhood of different galaxies along the cosmic web.
we expect @xmath4 and @xmath5 for a galaxy located in a filament, sheet and cluster respectively. using lcdm n - body simulations we find that it is possible to determine @xmath0 through a power law fit to @xmath2 across the length - scales @xmath6 to @xmath7 for @xmath8 of the galaxies.
we have visually identified the filaments and sheets corresponding to many of the galaxies with @xmath9 and @xmath6 respectively. in several other situations
the structure responsible for the @xmath0 value could not be visually identified, either due to its being tenuous or due to other dominating structures in the vicinity.
we also show that the global distribution of the @xmath0 values can be used to visualize and interpret how the different structural elements are woven into the cosmic web.
methods : data analysis - galaxies : statistics - large - scale structure of universe | 0812.1800 |
Introduction
the discovery of the acceleration of the universe by the measurements of the luminosity - redshift relation for type ia supervonae @xcite, combined with the observations of the anisotropies in the cosmic microwave background (cmb) @xcite, confirmed that about 70% of the density of the universe is made up of an unknown form of energy. this component, usually referred to as dark energy ", is unclumped and smoothly distributed, with an exotic property that it has negative pressure to cause the expansion of the universe to accelerate. the simplest candidate of dark energy is a non - vanishing, positive cosmological constant @xcite. we have to, however, explain why it is non - zero but vanishingly small (@xmath0) : possibly we can resort to some yet unknown fundamental symmetry which makes the cosmological constant vanishingly small but non - zero, or we might invoke an anthropic consideration that the observed small value of the cosmological constant allows life and that is why we observe it @xcite) different vacua, and some of them may be suitable for some kind of intelligent observers like us.]. an alternative form of dark energy is a slowly rolling scalar field, called quintessence, which has not yet relaxed at its ground state @xcite. for recent years, there have been considerable developments in the dynamics of the quintessence field. for example, in the context of so - called tracker solutions @xcite, the simplest case arises for a potential of the form @xmath1, where @xmath2 is positive and @xmath3 is an adjustable constant. by choosing @xmath3 suitably, it is possible to make a transition from an early matter - dominated universe to a later quintessence - dominated universe, free from fine tuning of the initial conditions. another interesting possibility, @xmath4-essence, is to introduce a non - canonical kinetic term for the scalar field @xcite, which makes the evolution of the scalar field dependent on the background equation of state, explaining why dark energy is dominating now. anyway, apart from the details, the situation that a scalar field is slowly rolling down its potential is reminiscent of the primordial inflation @xcite, where a scalar field (the inflaton) provides the vacuum - like energy density (@xmath5) necessary for a phase of the accelerated expansion by slowly rolling down its flat potential. hence one may naturally raise a question of how we can couple the early accelerated expansion, the primordial inflation, and the current one together @xcite : is it possible to cause the acceleration we observe recently by the primordial inflation? in this paper, we are going to discuss this possibility using a simple model based on the hybrid inflation @xcite, which arises naturally in many string - inspired inflation models, in particular in potentials for moduli fields. this paper is organized as follow : in section [model], we present a simple model of dark energy and analyze its dynamics in detail, presenting several conditions which should be satisfied for our model to work properly. unlike the conventional lore that the true minimum of the quintessence potential is presumed to vanish, we find that positive, zero, or negative vacuum energy is possible as we choose a different set of parameters. in section [discussions], we discuss the variation of the equation of state @xmath6 which might be detected in future observations such as hetdex @xcite. also we briefly address the possibility of realizing our model in supersymmetric theories and obstacles to overcome. we briefly summarize this paper in section [summary].
A model of dark energy
in this section, we discuss a simple model of dark energy based on the hybrid model of inflation @xcite. perhaps the simplest way to combine the onset of present acceleration of the universe with the primordial inflation is to directly couple the inflaton with the quintessence field. in this case, however, first we should ensure that inflation lasts for a long enough time (at least 60 @xmath7-foldings) without being disturbed by the interaction with the quintessence field. moreover, the quintessence potential must remain extremely flat after the inflaton reaches its minimum and decays to reheat the universe. rather, it is more plausible to couple the quintessence field with some different field which plays no role during the inflaton rolls down its potential and only works to finish inflation : this is what the waterfall field in hybrid inflation model does. hence we can write the effective potential as @xmath8 where we take @xmath9 as the inflaton field, @xmath10 as the waterfall field, and @xmath11 as the quintessence field. as can be seen from the coupled terms above, we will experience two phase transitions, which distinguish different stages of the evolution of the universe. in the following subsections, we will discuss each stage in detail. the effective masses squared of @xmath10 and @xmath11 are @xmath12 and @xmath13, respectively. hence, for @xmath14, the only minimum of the effective potential, eq. ([potential]), is at @xmath15. also, with @xmath15, for @xmath16, the only minimum of the effective potential is at @xmath17. thus, at the early stage of the evolution of the universe, @xmath10 and @xmath11 are trapped at 0 while @xmath9 remains much larger than @xmath18 for a long time. this stage lasts until @xmath19, and at that moment we assume that the vacuum energy density @xmath20 is much larger than the potential energy density of the inflaton field @xmath21, so that the hubble parameter is given by @xmath22 where @xmath23 is the reduced planck mass. we can additionally assume that the vacuum energy @xmath24 is dominated by @xmath25, i.e., @xmath26 so that the subsequent evolutions of the universe after inflation, e.g., reheating, becomes identical to the usual hybrid model. this immediately gives @xmath27 now, combining eqs. ([h1stlast]) and ([psidomination]), the slow - roll condition for @xmath28 gives @xmath29 note that this gives another bound on @xmath28 as @xmath30, and the tightest bound on @xmath28 depends on the comparison between the prefactors of this expression and eq. ([mbound1]), i.e., @xmath31 and @xmath32. as discussed above, when @xmath9 becomes smaller than @xmath18, a phase transition with the symmetry breaking for @xmath10 occurs. to finish inflation as soon as @xmath9 reaches the critical value @xmath18, i.e., for the so - called waterfall ", we first require that the absolute value of the effective mass squared of @xmath10 be much larger than @xmath33. during inflation, using the slow - roll equation of motion @xmath34 and eqs. ([h1stlast]) and ([psidomination]), we can find that when the inflaton @xmath9 reaches @xmath18, it decreases by @xmath35 during the time interval @xmath36. at that time, the absolute value of the effective mass squared of @xmath10 is given by latexmath:[\[\label{psimass } is much greater than @xmath33 for @xmath38 also, we demand that the time scale for @xmath9 to roll down from @xmath18 to 0 be much shorter than @xmath36 : let us denote @xmath39 as the time @xmath9 takes as it moves from @xmath18 to 0. then, from @xmath40 where we have used eq. ([psimass]) and the slow - roll equation of @xmath9, we have @xmath41 which gives another condition @xmath42 note that the parameters with respect to @xmath9-@xmath10 transition are further constrained from the observed cobe amplitude of density perturbations : we obtain @xcite @xmath43 which looks similar to eqs. ([waterfall1]) and ([waterfall2]). combining eqs. ([waterfall1]), ([waterfall2]) and ([pertconst]), we can extract the valid range of the parameters @xmath28, @xmath44, @xmath31 and @xmath45 : e.g. if we take @xmath46, @xmath47 and @xmath48, all the conditions are satisfied. after this phase transition, the waterfall field @xmath10 oscillates at the minimum and decays so that the universe is reheated of @xmath11 to @xmath10, and in turn to the thermal bath, the oscillation of @xmath10 may affect the dynamics of @xmath11. however, we note that because of the negative sign of their interaction term in eq. ([potential]), the pattern of instability is quite different from the usual parametric resonance @xcite even though the interaction is strong enough : we will study this issue in more detail separately.]. with these conditions, as soon as @xmath9 reaches @xmath18, inflation ends within a hubble time, @xmath36. also note that when such a rapid phase transition occurs within observationally interesting range, still we can calculate the power spectrum and the spectral index for the density perturbations @xcite by generalizing the standard perturbative method @xcite : generally, the power spectrum shows a scale dependent oscillations after the phase transition @xcite. while @xmath10 is rolling down to @xmath49, the effective mass squared of @xmath11 might become negative then another phase transition occurs. for this to happen, we want @xmath50, unless no instability for @xmath11 would develop. this is equivalent to the condition is settled at @xmath51. when @xmath10 moves along @xmath52 direction, this is equivalent to the condition @xmath53 this means, combined with eq. ([psisigma]), @xmath44 is (much) bigger than @xmath54. we are grateful to andrei linde for pointing out this.] @xmath55 note that we can obtain the same condition by requiring that the minimum along @xmath11 direction, @xmath56 given by @xmath57 be real, i.e., @xmath58 once the above condition is satisfied and @xmath11 rolls down the effective potential, after all the fields are settled at @xmath59 where the potential becomes @xmath60 when @xmath61, eq. ([globalmin]) is greater than zero and it corresponds to a non - zero, positive vacuum energy. hence no matter @xmath11 evolves quickly or not, we are provided with the source of the current acceleration of the universe in this case, and this acceleration will last forever : the universe will eventually behave as a de sitter space. note that the possible maximum value of @xmath62, @xmath63, could be at most as large as the current critical density, @xmath64. however, an extreme fine tuning is required to match this value : if we restrict our interest to this case only, our model is no better than the conventional @xmath65cdm because the latter is simpler and hence preferable. anyway here we do not try to improve this situation. a study of alleviating this fine tuning problem is very important and interesting, but is outside the scope of the present paper. for the case @xmath66 eq. ([globalmin]) is exactly zero and no vacuum energy exists. this case corresponds to the usual quintessential inflation @xcite, where the cosmological constant @xmath65 is assumed to be zero due to some unknown symmetry, and the observed dark energy is supplied by e.g. another scalar field, here @xmath11. to be able to explain the acceleration of the universe, we require that still @xmath11 be rolling down so that the potential is non - zero. that is, @xmath11 should roll extremely slowly. note that although @xmath11 begins to evolve only after @xmath10 reaches @xmath67, for the most time during @xmath11 rolls down the effective potential to @xmath56, @xmath10 has already settled at @xmath51 within a time @xmath68 as soon as @xmath69. thus, we just set @xmath70 throughout the evolution of @xmath11. under this circumstance, when @xmath11 begins to roll, @xmath71 is given by @xmath72 for @xmath11 to slowly evolve, we require that @xmath73, the absolute value of the effective mass squared of @xmath11, be much smaller than @xmath33, from which we obtain a condition @xmath74 where we have used eq. ([v0=0condition]). indeed, the same bound could be found when we also require that @xmath11 roll very slowly so that for @xmath11 to move by an infinitesimal displacement @xmath75, it takes much longer time than @xmath36 : the effective potential is given by @xmath76 where we have used eq. ([v0=0condition]). hence, combining with the slow - roll equation @xmath77 we find @xmath78 i.e. @xmath79 for @xmath80. requiring @xmath81 gives @xmath82 which is the same as eq. ([v0=0con1]). recent observations find that the vacuum energy, or cosmological constant, is very small and positive, with its value being of order @xmath83. the case of negative cosmological constant is not observationally justified, which seems to support the claim that the true minimum of the quintessence potential is zero. the vacuum state with negative energy is, nevertheless, an interesting theoretical possibility, especially in string theories where anti de sitter solutions are popular. hence a negative potential may play an important role in cosmology motivated from string theory or m theory, such as cyclic universe model @xcite. we obtain @xmath84 when @xmath85 has a real solution and that this solution is smaller than @xmath56. if we take the limit @xmath86, the effective mass squared of @xmath11 is given by @xmath87 imposing the slow - roll condition, this gives @xmath88 the same as eqs. ([v0=0con1]) and ([v0=0con2]). since the dark energy observed now is positive definite, @xmath11 should be still evolving and not have crossed @xmath89 : thus we again require that @xmath11 roll down very slowly, from which we obtain the same conditions as the previous section, eqs. ([v0=0con1]) and ([v0=0con2]). however, unlike the cases before, the true vacuum energy is negative and after a (tremendously) long time the universe will collapse eventually @xcite, no matter how small the magnitude is.
Discussions
generally, for evolving dark energy models the equation of state @xmath6 is not a constant but a slowly evolving function. to provide an acceleration of the universe it is constrained to be smaller than @xmath90, usually taken to be @xmath91. recent combined observations of wmap and sdss constrains @xmath6 to be @xmath92, with uncertainties at the 20% level @xcite. for the first case discussed in the previous section where @xmath61, this is easily satisfied : @xmath6 is very slowly varying as @xmath11 rolls down its potential, finally becoming exactly @xmath92 at the absolute minimum @xmath62. when @xmath93, @xmath6 will be eventually 0 after @xmath11 settles at @xmath56. for the third case, however, it is not as trivial as the other cases. naively, we may take @xmath94 to be non - zero although very small, then we can write @xmath95 where @xmath96 stands for the kinetic energy, @xmath97. at first look, it seems that @xmath6 is divergent at @xmath98 and a discontinuity appears and moreover a phantom phase @xmath99 exists after that point. let us more closely see this. the simplest way is to solve the relevant equations numerically : we choose @xmath100 to solve for @xmath101 and @xmath102. in figure [fig_w], we plot the resulting equation of state @xmath6 where, as is shown, it oscillate. to understand this behavior first let us consider the expansion of the universe with such a negative potential @xcite : the universe keeps expanding until the moment @xmath103 where the scale factor reaches its largest value, and the universe shrinks afterwards. from the friedmann equation describing a flat universe @xmath104 which has no solution with @xmath105, we can see that to maintain a real value of @xmath71 no matter positive or negative, @xmath106 should remain positive. hence, when @xmath107 is negative, the kinetic energy must compensate this negative potential energy and consequently these energies oscillate out of phase : for example, at the negative bottom of the potential, the field is moving fastest so that we have maximum kinetic energy here so that @xmath6 becomes very large. this is the reason why the equation of state @xmath6 is oscillating, we do nt see such an oscillation.] and sometimes becomes greater than 1, the `` stiff '' fluid. we can easily find the relation with respect to the redshift @xmath108, by making use of the relation @xmath109 : by solving the equations @xmath110 where a prime denotes the derivative with respect to @xmath108, we can obtain @xmath6 in terms of @xmath108. a typical relation between @xmath6 and @xmath108 is shown in the right panel of figure [fig_w]. supersymmetry is believed to be the most promising candidate related to the fundamental problems in particle physics, such as the hierarchy problem and the gauge coupling unification. since we have not yet observed any supersymmetric partner of known particles, supersymmetry is broken. in the primordial universe where the energy scale is much higher than the present one, however, the rich structure of supersymmetric and supergravity theories should have played an important role. hence it is natural to try to implement inflationary scenario within supersymmetry. there has been an encouraging progress on the hybrid inflation scenario in the context of supersymmetric theories @xcite. for example, consider a simple superpotential @xcite @xmath111 where @xmath112 and @xmath113 are a pair of superfields in non - trivial representations of some gauge group under which @xmath9 is neutral. then, in a globally supersymmetric theory, the effective potential is given by @xmath114 here, the absolute supersymmetric minimum appears at @xmath115. however, for @xmath116, @xmath112 and @xmath113 obtain positive masses squared and hence are confined at the origin. by simply adding a mass term @xmath117 which softly breaks supersymmetry, we see that the hybrid inflation scenario is possible. if we take into account radiative corrections instead, the total effective potential including one - loop corrections is @xmath118\]] when @xmath119, and we can see that still inflation is possible. similarly, we can hope to implement our scenario within supersymmetry. the simplest possibility should be @xmath120-terms since, as we can see from eq. ([potential]), the coupled term of @xmath10 and @xmath11 has a negative sign, so that an instability for @xmath11 could be developed as @xmath10, initially confined at zero, rolls away from the origin. this is easily achieved by assuming that @xmath10 and @xmath11 are oppositely charged under some gauge symmetry. if it is a @xmath121 symmetry, we can write the @xmath120-term contribution as @xmath122 where we assumed that @xmath10 and @xmath11 have charges equal to @xmath92 and @xmath123 respectively, @xmath124 is the gauge coupling, and @xmath125 is a fayet - iliopoulos @xmath120-term. we may also include another gauge symmetry under which @xmath9 and @xmath10 are charged with the same sign, then @xmath120-term is given by @xmath126 where we have introduced another gauge coupling @xmath127 and fayet - iliopoulos term @xmath128. taking only eqs. ([dterm1]) and ([dterm2]) into account, the coupled terms are given by @xmath129 we can reproduce the same forms for @xmath130 and @xmath131 terms as eq. ([potential]) as long as @xmath132 is positive. this sheds some light on the realization of eq. ([potential]). however, this is far from a realistic possibility. first of all, we have not considered any @xmath133-term contributions. we do not want to couple @xmath9 and @xmath11 to guarantee their desired behaviors on the flat enough effective potential, but it seems not easy to achieve this through the contributions from @xmath133-term. more fundamentally, in globally supersymmetric theories the scalar potential is either positive or zero, which does not include the interesting case of @xmath84. also, it is thought that below the planck scale particle physics is described by an effective @xmath134 supergravity theory derived from string theory. hence, a more detailed analysis of our model and associated problems should be addressed in the context of supergravity : for example, the mass of the field should be very small, of order the present hubble parameter @xmath135, so that it is still rolling toward its true minimum. however, usually scalar fields acquire masses of order the gravitino mass @xmath136, much heavier than @xmath137. one way to evade this difficulty is to use pseudo nambu - goldstone bosons problem associated with generic scalar field potentials in supergravity.] @xcite, e.g., string / m theory axion, as @xmath11 field @xcite. we will leave such an analysis as a challenging future work.
Summary
we have investigated a simple dark energy model based on hybrid inflation. the quintessence field @xmath11 is coupled to the waterfall field @xmath10 so that as @xmath10 rolls towards @xmath138, @xmath11 begins to move along the effective potential provided that eq. ([psisigma]) is satisfied. a number of bounds on the parameters of the model are found, such as eqs. ([psidomination]), ([mbound1]), ([mpsibound1]), ([waterfall1]) and ([waterfall2]). an interesting point is that the true minimum of the effective potential, eq. ([globalmin]), depends on our choice of the parameters, allowing the vacuum state with positive, negative and zero energy. when it is negative, the universe will eventually collapse, showing an oscillation in the equation of state @xmath6 for dark energy which we can hope to detect in future observations. this model could be realized in supersymmetric theories via @xmath120-term contributions, but including @xmath133-term parts and supergravity effects makes this model not easy to be achieved. we thank kiwoon choi, salman habib, andrei linde and carlos muoz for helpful discussions and suggestions. we are also indebted to the anonymous referee for many important and invaluable comments. jg is grateful to the sf05 cosmology summer workshop, and sk to the summer institute 2005 for hospitalities when this work was in progress. this work was supported in part by the astrophysical research center for the structure and evolution of the cosmos funded by the korea science and engineering foundation and the korean ministry of science, the korea research foundation grant krf pbrg 2002 - 070-c00022, and the brain korea 21. p. j. e. peebles and b. ratra, _ astrophys. j. _ * 325 *, l17 (1988) ; b. ratra and p. j. e. peebles, _ phys. * d 37 *, 3406 (1988) ; r. r. caldwell, r. dave and p. j. steinhardt, _ phys. _ * 80 *, 1582 (1998) ` astro - ph/9708069 ` c. armendariz - picon, v. mukhanov and p. j. steinhardt, _ phys. lett. _ * 85 *, 4438 (2000) ` astro - ph/0004134 ` ; c. armendariz - picon, v. mukhanov and p. j. steinhardt, _ phys. * d 63 *, 103510 (2001) ` astro - ph/0006373 ` w. h. kinney and a. riotto, _ astrophys. _ * 10 *, 387 (1999) ` hep - ph/9704388 ` ; p. j. e. peebles and a. vilenkin, _ phys. rev. _ * d 59 *, 063505 (1999) ` astro - ph/9810509 ` ; m. peloso and f. rosati, _ j. high energy phys. _ * 12 *, 026 (1999) ` hep - ph/9908271 ` ; a. b. kaganovich, _ phys. * d 63 *, 025022 (2001) ` hep - th/0007144 ` ; g. huey and j. e. lidsey, _ phys. * b 514 *, 217 (2001) ` astro - ph/0104006 ` ; a. s. majumdar, _ phys. rev. _ * d 64 *, 083503 (2001) ` astro - ph/0105518 ` ; k. dimopoulos and j. w. f. valle, _ astropart. * 18 *, 287 (2002) ` astro - ph/0111417 ` ; n. j. nunes and e. j. copeland, _ phys. _ * d 66 *, 043524 (2002) ` astro - ph/0204115 ` ; e. i. guendelman and o. katz, _ class. _ * 20 *, 1715 (2003) ` gr - qc/0211095 ` ; k. dimopoulos, _ phys _ * d 68 *, 123506 (2003) ` astro - ph/0212264 ` ; m. giovannini, _ phys. rev. _ * d 67 *, 123512 (2003) ` hep - ph/0301264 ` ; t. biswas and p. jaikumar, _ phys. * d 70 *, 044011 (2004) ` hep - th/0310172 ` ; m. r. garousi, m. sami and s. tsujikawa, _ phys. _ * d 70 *, 043536 (2004) ` hep - th/0402075 ` ; m. sami and v. sahni, _ phys. * d 70 *, 083513 (2004) ` hep - th/0402086 ` ; t. biswas and p. jaikumar, _ j. high energy phys. _ * 08 *, 053 (2004) ` hep - th/0407063 ` ; a. gonzalez, t. matos and i. quiros, _ phys. _ * d 71 *, 084029 (2005) ` hep - th/0410069 ` ; s. barshay and g. kreyerhoff, _ mod. lett. _ * a 19 *, 2899 (2004) ` astro - ph/0410478 ` ; r. rosenfeld and j. a. frieman, _ j. cosmol. astropart. * 09 *, 003 (2005) ` astro - ph/0504191 ` ; x. zhai and y. zhao, _ nuovo cim. _ * 120 b *, 1007 (2005) ` gr - qc/0508069 ` ; x. zhai and y. zhao, _ chin. phys. _ * 15 *, 2465 (2006) ` astro - ph/0511512 ` j. choe, j.- o. gong and e. d. stewart, _ j. cosmol. phys. _ * 07 *, 012 (2004) ` hep - ph/0405155 ` ; j.- o. gong, _ class. * 21 *, 5555 (2004) ` gr - qc/0408039 ` ; m. joy, e. d. stewart, j.- o. gong and h.- c. lee, _ j. cosmol. astropart. phys. _ * 04 *, 012 (2005) ` astro - ph/0501659 ` e. j. copeland, a. r. liddle, d. h. lyth, e. d. stewart and d. wands, _ phys. * d 49 *, 6410 (1994) ` astro - ph/9401011 ` ; g. lazarides, r. k. schaefer and q. shafi, _ phys. rev. * d 56 *, 1324 (1997) ` hep - ph/9608256 ` s. barshay and g. kreyerhoff, _ mod. lett. _ * a19 *, 2899 (2004) ` astro - ph/0410478 ` _ erratum - ibid. _ * a20 *, 373 (2005) ; r. rosenfeld and j. a. frieman, _ j. cosmol. phys. _ * 09 *, 003 (2005) ` astro - ph/0504191 ` | the situation that a scalar field provides the source of the accelerated expansion of the universe while rolling down its potential is common in both the simple models of the primordial inflation and the quintessence - based dark energy models.
motivated by this point, we address the possibility of causing the current acceleration via the primordial inflation using a simple model based on hybrid inflation.
we trigger the onset of the motion of the quintessence field via the waterfall field, and find that the fate of the universe depends on the true vacuum energy determined by choosing the parameters.
we also briefly discuss the variation of the equation of state and the possible implementation of our scenario in supersymmetric theories.
kaist - th/2005 - 16 1.5 cm 0.5 cm 1.2 cm | astro-ph0510207 |
[sec:intro]introduction
graphene as the basis of a new generation of electronics@xcite has been the center of much attention in the last years, and devices based on nanostructured graphene have been put forward. the most generic form of nanostructured graphene is graphene nanoribbons (gnr),@xcite and other structures, such as graphene anti - dot lattices@xcite, can be viewed as networks of them. gnrs are potential candidates for molecular wires with tailored conductance properties. for graphene - based nanostructures the edges and their passivation, as well as defects inside the structure, can play crucial roles for the transport properties.@xcite however, characterization of edge passivation or structural / chemical defects is challenging especially after device fabrication. raman spectroscopy@xcite can give information about defects on large areas of the sample, while tip - enhanced raman spectroscopy (ters)@xcite have been used in combination with stm on gnrs. however, raman studies involve averages over larger areas (> 10 nm), and does not yield information about the impact of vibrations on transport. in that aspect inelastic electron tunneling spectroscopy (iets) serves as a way of performing non - destructive characterization yielding vibrational / phonon fingerprints of a range of defects. in order to interpret iets experiments, theoretical modeling of the inelastic signals in the electronic current due to electron - phonon (e - ph) scattering is needed. gnrs have been fabricated using different strategies including lithographic techniques,@xcite chemical synthesis,@xcite epitaxial growth@xcite, and longitudinal unzipping of carbon nanotubes.@xcite furthermore, several groups have succeeded in atomically precise bottom - up fabrication of armchair gnrs (agnr)@xcite, chiral gnrs,@xcite and agnr hetero - junctions@xcite grown on metal surfaces. experimentally, the vibrational properties have been investigated by raman spectroscopy and the electronic structure has been mapped out by stm, angle - resolved (two - photon) photo - emission and high - resolution electron energy loss spectroscopy.@xcite signatures of phonon excitation were observed by stm in the differential conductance spectroscopy performed at the zigzag termini state of agnrs adsorbed on au(111), and these signatures were shown to be sensitive to modifications in the local atomic geometry.@xcite agnrs have also been lifted up from the weakly bonding au(111) surface with the tip of a stm enabling measurements of the voltage - dependent conductance in suspended configurations.@xcite from the theoretical side density - functional theory (dft) has been used to investigate the stability of structural and chemical reconstructions of gnr edges,@xcite together with the transport and band - gap engineering.@xcite the vibrational properties and phonon band structure have been calculated with empirical potentials@xcite and dft.@xcite in addition, there have been theoretical predictions@xcite of the raman spectrum, in good agreement with experiments.@xcite for a finite agnr the role of zigzag termini states have been studied theoretically, comparing dft to the many - body hubbard model.@xcite inspired by the recent lifting experiments by koch, @xcite we here investigate theoretically the signals of e - ph scattering in the conductance of long gnrs between metal electrodes. our aim is two - fold. first, we want to address the role phonon scattering in the transport characteristics of pristine gnrs. second, we wish to compute detailed iets for different gnrs under varying charge carrier concentrations and explore how different types of realistic defects may modify the iets and thus possibly be directly probed in transport measurements. we focus on the two most generic edge types, namely armchair (agnr) and zigzag (zgnr), and pay attention to the effects of spin polarization in the latter case. in actual experiments the substrate or an applied gate potential control the fermi level @xmath0 in the ribbons. to address this variability we scan @xmath0 using a numerically effective scheme enabling fast calculations of the iets.@xcite we find that the agnr generally display two robust iets signals around @xmath1 and @xmath2 mv corresponding to the d- and g - modes of raman spectroscopy and that a dehydrogenated dimer at the edge should further leave a clear defect signal at around @xmath3 mv. for the zgnr we find that the spin polarization breaks the mirror symmetry around the middle of the ribbon resulting in iets signals from a range of modes around the d- and g - mode energies. for both agnr and zngr defects which break the planar symmetry of ribbons allows for contributions to the iets from out - of - plane phonon modes. the paper is organized as follows. first we discuss our atomistic model setup for the density functional and electron transport calculations, and outline the approach for the iets simulations. in sec. iii we present our results for pristine agnr and zgnr and relate their transport properties and iets to the band structures. in sec. iv we turn to the defected systems by considering realistic possibilities of defects in the edge passivation, backbone bonding motifs, and presence of adatoms. finally, a summary and our conclusions are presented in sec.
[sec:method]methods
we calculate the electronic and vibrational structure from dft using the academic codes siesta / transiesta.@xcite we employ the generalized gradient approximation (gga) for the exchange - correlation functional,@xcite a single - zeta polarized (szp) basis set for the carbon and hydrogen atoms, and use a cut - off energy of 400 - 500 ry for the real - space grid. these choices, balancing accuracy and computational cost, provide a good description to investigate trends and general behavior of the substantial number of systems considered in this work. the vibrational degrees of freedom, calculated by diagonalization of the dynamical matrix, and the e - ph couplings are extracted from finite differences as implemented in the inelastica code.@xcite the armchair and zigzag gnrs considered here are shown in fig. [fig : cleangnr]. we adopt the usual two - probe setup with the device region (@xmath4) coupled to left (@xmath5) and right (@xmath6) electrodes with all electronic matrix elements expressed in a local basis set. the primitive unit cell of the agnr (zgnr) consists of 18 (10) atoms and in our calculations this unit cell is repeated 10 (18) times in the transport direction to form the scattering regions illustrated in fig. [fig : cleangnr](a, e). the electrode couplings @xmath7 are included on the two first / last unit cells before folding onto @xmath4. in our treatment a subset of atoms in @xmath4 is allowed to vibrate. we fix this dynamical region, restricted by the condition that the e - ph couplings are fully included inside @xmath4, to the 4 and 6 central unit - cells for the agnr and zgnr, respectively. the corresponding e - ph couplings used to calculate the inelastic electron transport are thus expressed in the center 6 unit - cells for the agnr and 8 unit - cells for the zgnr. the convergence of our results with the size of the dynamical region is addressed below. we generally consider nanoribbons that are suspended between two metallic leads. in the case of the lifting experiments,@xcite these would correspond to the metal sample surface and the stm tip. we wish here to focus on the action inside the gnrs and put aside the possible complications due to the detailed electronic structure of the metals, and the metal - gnr interface in particular. to this end we introduce a simple model of the metal electrodes without substantial electronic features : we use semi - infinite gnrs with highly broadened states (effectively smearing out energy gaps). in practice this is done by adding a finite numerical imaginary part @xmath8 to the energy argument in the electrode recursion calculation.@xcite this scheme ensures that the phonon effects originate from the gnrs themselves and not from details of the metal - gnr interface, which is generally unknown in the stm experiments. the electronic band structures for the infinite ribbons, along with the transmission and density of states (dos) are shown for @xmath9 ev in fig. [fig : cleangnr](b, c, d) and fig. [fig : cleangnr](f, g, h) for agnr and zgnr, respectively. we note that the broadened transmission spectrum [fig. [fig : cleangnr](d)] is quite consistent with the experimentally reported differential conductance curves for agnr.@xcite the electronic states involved in the transport are shown in fig. [fig : channels] in terms of the transmission eigenchannels@xcite in the valence and conduction bands of the agnr and zgnr. their spatial symmetry play a significant role for the selection rules involved in the inelastic scattering as discussed later. in principle, the electronic structure should be evaluated at finite bias. however, without a detailed model of the connection to the metal electrodes (where an important part of the voltage drop will take place) and for sufficiently long systems (in which the electric field will be small), it is reasonable to use the zero - voltage electronic structure and to simply assume a symmetric voltage drop over the two identical, idealized device - electrode interfaces. more specifically, in the following we consider that the chemical potentials of the electrodes move according to the applied bias voltage @xmath10 and possibly an applied gate voltage @xmath11 (mimicking actual doping or electrostatic gating that modify the charge carrier concentration in an experimental setup) according to @xmath12. for a device strongly coupled to the electrodes, a coupling between the electron current @xmath13 and a phonon mode @xmath14 ideally shows up at zero temperature as a step discontinuity in the differential conductance when the inelastic phonon emission process becomes energetically allowed, that is, when the chemical potential difference exceeds the quantum of vibrational energy, @xmath15. thus, around the emission threshold the electronic states involved in the scattering process are those at @xmath16 and @xmath17. the iets signal, conventionally expressed as the ratio between the second and first derivatives of the current with respect to the voltage, @xmath18 is calculated by considering the e - ph coupling as the perturbation on the current, evaluated using the nonequilibrium green s functions (negf). in the so - called lowest order expansion (loe) the inelastic part of the differential conductance can be written as,@xcite @xmath19 where summation over the vibration index @xmath14 is assumed. @xmath20 and @xmath21 are the `` universal '' (system - independent) functions that depend on the applied bias @xmath10, phonon energy @xmath22 and the temperature @xmath23. assuming the electronic and phononic distribution functions are given by the fermi - dirac and bose - einstein distributions, respectively, their analytical expressions can be written as : @xmath24,\nonumber\end{aligned}\]] where @xmath25 is the conductance quantum, @xmath26 is the fermi - dirac function, @xmath27, and @xmath28 denotes the hilbert transform. the signal amplitudes @xmath29 and @xmath30 of the symmetric and antisymmetric signals in the differential conductance are even and odd in bias, respectively. for a symmetric structure the asymmetric signal vanishes in the wide - band approximation (loe - wba)@xcite. however, this is not guaranteed in the more general treatment employed here,@xcite where the energy dependence of the electronic structure is explicitly taken into account. the amplitudes @xmath29 and @xmath30 are expressed in terms of electronic structure quantities and e - ph couplings,@xcite @xmath31+{\rm i m } b_\lambda, \label{eq : gamma } \\ \kappa_\lambda = & 2{\rm re } b_\lambda, \label{eq : kappa}\end{aligned}\]] where @xmath32 is defined as @xmath33. \label{eq : ic4ag}\end{aligned}\]] in the above, @xmath34 denotes the e - ph coupling matrix for mode @xmath14, @xmath35 the retarded / advanced unperturbed green s functions, and @xmath36 the spectral density matrices for left / right moving states with the time - reversed version @xmath37. the purely electronic quantities are thus being evaluated at the chemical potentials of the left / right electrodes corresponding to the excitation threshold for each vibration. we compute @xmath34 with the finite - difference scheme of inelastica taking the vacuum energy as a common reference (in absence of real metal leads to pin the fermi energy).@xcite in the localized atomic basis set of siesta all the above quantities are matrices defined in the electronic space corresponding to region @xmath4. the second derivatives of the universal functions in eqs. ([eq : sym])-([eq : asym]) are sharply peaked around the phonon threshold. for this reason the coefficients @xmath38 and @xmath30 can be considered voltage - independent with their values computed exactly at the threshold. due to the computational efficiency of the loe scheme described above we are able to evaluate the iets on a fine grid of gate voltages @xmath11 spanning a large range of relevant values between valence and conduction bands of the gnrs.
Pristine graphene nanoribbons
now we first turn to the iets results of the two pristine (clean) ribbons, and in the following section to the impact of selected defects in the iets. as our main system we focus on the agnr systems directly relevant for the lifting experiments.@xcite the results for the zgnr are provided mainly as comparison and to look into the role of chirality and in particular effects rooted in spin polarization, and thus we now discuss these separately. as representative of the agnr class we have investigated a pristine agnr with a width of @xmath39 dimers (7-agnr) corresponding to a c - c edge distance of 7.5 (see fig. [fig : cleangnr]). it presents a direct semi - conducting band gap @xmath40 due to the lateral confinement and can be classified as a `` large - gap ribbons '' since @xmath41 is an integer in the relation @xmath42.@xcite we obtain @xmath43 ev at the present level of approximation (dft - gga and szp basis set), as seen from the electronic band structure shown in fig. [fig : cleangnr](b). this value is smaller than those estimated experimentally (@xmath44 2.3 - 2.6 ev for a flat agnr on au(111) @xcite and @xmath45 ev for an agnr suspended between surface and stm - tip@xcite) due to the underestimation of electron - electron interaction@xcite which plays an more important role in quasi one - dimensional gnrs compared to pristine graphene. dielectric screening from the substrate also influences significantly the actual gap size : a band gap of @xmath46 ev for a 7-agnr was found to be lowered to @xmath47 ev on a hexagonal boron - nitride (hbn) substrate using gw calculations,@xcite similar to the lowering calculated for a 7-agnr on au(111).@xcite in general we expect that underestimation of band gaps would mainly amount to a simple scaling the fermi level position within the gap. we first discuss the effect of the finite size of the dynamical region in our treatment. figure [fig : convclean](a) shows how the iets signals for the agnr (at fixed gate voltage @xmath48 v) vary as a function of the size of the dynamical region, ranging from 1 to 6 unit cells. for easy comparison, the data are normalized by the number of vibrating unit cells. as the signal amplitudes in this representation are roughly constant we conclude that the absolute iets simply scale linearly with the active e - ph coupling region. consequently, the magnitudes in iets may thus provide insight into the active scattering region in actual experiments. further, as we find that both iets amplitude and shape is well converged with 4 vibrating unit cells, we fix the dynamical region to this size in the following analysis. the computed iets signals for the agnr as a function of varying gate voltage are shown in fig. [fig : ietsclean](a) as a density plot. specific iets spectra at selected gate voltages are shown in fig. [fig : ietsclean](c) for both the intrinsic part (temperature broadening at @xmath49 k) as well as that one would observe employing the experimental lock - in technique (additional broadening due to a modulation voltage of @xmath50 mv). we find that for the agnr there are generally two well - defined iets signals appearing around @xmath1 and @xmath2 mev, corresponding to the d- (ring breathing) and g- (e@xmath51 phonon) modes, respectively, also observed in raman spectroscopy.@xcite. the d - signal also has a shoulder with a local maximum at @xmath52 mev with contributions from several modes. these three distinct features are indicated with vertical lines in fig. [fig : ietsclean](a, c). shifting @xmath0 inside the gap region with a relatively small gate voltage @xmath53 v does not affect the iets appreciably. however, when @xmath0 comes close to the conduction band of the agnr the signal increases by a factor of five and a small peak - dip feature appear similar to the one reported for gated benzene - dithiol molecular contacts.@xcite upon further gating into the conduction band the iets signals undergo a sign reversal (from peaks to dips) as the transmission increases beyond approximately @xmath54 for the involved channels.@xcite similar effects are also found by gating into the valence band of the agnr. we can easily identify the most important vibrational mode vectors @xmath55 for the iets from the two amplitudes @xmath56 and @xmath57 given in eqs. ([eq : gamma])-([eq : kappa]). these modes can further be analyzed in terms of the phonons in the infinite agnr. to do so we introduce the measure @xmath58 representing the overlap between modes in the finite dynamical cell and the phonon band modes weighted by the size of the iets signal, @xmath59 where @xmath60 is the phonon band mode indexed by @xmath61, and @xmath62 is the modes in a finite @xmath63 primitive cell long dynamical region index by @xmath14. the projections @xmath64 are depicted as widths of the phonon bands in fig. [fig : agnrphononband](a), where the red, green and blue colors refer to modes with frequencies in the ranges @xmath65 mev, @xmath66 mev, and @xmath67 mev, respectively. in total four bands contribute to the iets signal corresponding to the four signals seen in the intrinsic part of the iets spectrum in fig. [fig : ietsclean](c). the corresponding @xmath68-point phonon modes inside the primitive cell for the infinite ribbon are shown in fig. [fig : agnrphononband](b - e). we next turn to our results for the pristine zgnr shown in fig. [fig : cleangnr](e). it has a width of @xmath69 zigzag `` chains '' (4-zgnr) corresponding to a c - c edge distance of 7.26 . the breaking of sublattice symmetry for the zgnr and lack of pseudo - phase result in different selection rules for the matrix elements and difference in for example raman signals.@xcite the zgnr generally presents spin - polarized edge states exhibiting a small band gap at the dft level,@xcite in our case @xmath70 ev (we note that this gap disappears in simpler tight - binding descriptions@xcite or spin - degenerate dft calculations). the spin - polarized edge states play the major role for the conduction, see the spin - down eigenchannels visualized in fig. [fig : channels](e - h). since the edge states break the mirror symmetry with respect to the middle of the ribbon, there are fewer symmetry - forbidden inelastic transitions between the scattering states for the zgnr. thus, we expect a wider range of modes to contribute to the iets signal as compared to the agnr case. indeed this is in agreement with the findings shown in fig. [fig : convclean](b) and fig. [fig : ietsclean](b, d). the greater number of modes contributing to the iets for the zgnr results in broader signals with similar magnitudes as compared to the iets for agnr. as for the agnr case the iets signal is well converged with a dynamical region consisting of 6 vibrating unit cells [fig. [fig : convclean](b)]. for zgnrs the ring breathing is forbidden by symmetry, thus the iets is generally characterized by transverse and longitudinal modes. to explore the impact of spin - polarization on the zgnr - iets we compare in fig. [fig : nospin] the results from both spin - degenerate and spin - polarized calculations. without gate voltage (@xmath71 v) the iets display opposite signs due to the spin - induced gap. only a single peak contributes to the spin - degenerate iets while several peaks contribute to the spin - polarized iets. even if the zgnr is tuned by @xmath72v to become metallic and the two treatments then show the same overall sign in iets, the spin - polarized iets persists to show a much richer structure. this difference suggests that iets could be a way to indirectly observe spin - polarized edge states. projecting the modes contributing to the iets onto the phonon band modes further underlines how several bands with different symmetries contribute to the spin - polarized iets, while only a couple of bands contributes to the spin - degenerate iets, see fig. [fig : bandzgnr]. again we use eq. ([eq.weightf]) for this characterization, where the overlap for @xmath48 v corresponds to the red color and the overlap for @xmath72 v corresponds to the difference between the blue and red color in fig. [fig : bandzgnr], respectively. it is clear that spin - polarization permits more modes to contribute to the iets. in contrast to the spin - degenerate case, where the symmetric electronic states (with respect to the middle of the ribbon) only can couple to the symmetric vibration modes, the symmetry lowering of the electronic states by spin - polarization opens up also for scattering also via odd modes.
Defected graphene nanoribbons
in this section we address the modification and new signals in iets that arise due to various defects in the gnr. regardless of the fabrication method, defects will inevitable occur. for example, if the agnrs are synthesized from a precursor molecule, involving heating and dehydrogenation, as reported by cai @xcite and blankenburg, @xcite there is a chance that the reaction is incomplete and some of the c - c bonds between the precursor molecules do not form. also there is a chance that a part of the final agnr will have dehydrogenated edges or are passivated by two hydrogen atoms. finally, defects may be introduced on purpose by locally dosing a high current from the tip of a stm.@xcite (color online) electronic properties of the agnr structures shown in fig. [fig : xyzagnr]. the total transmission is shown with black lines. the ratio @xmath73, where @xmath74 is the transmission originating from the most transmitting eigenchannel is shown with green dashed lines (this ratio gives a lower bound to the number of contributing eigenchannels). the dos for the c atoms in the dynamical region is shown with red lines (offset by 3 units).] . in fig. [fig : xyzagnr] we show the structures of pristine agnr along with 8 different defect configurations which we have considered. these include four defects in the edge passivation as follows : a single edge side with an extra hydrogen atom [` 1h - edge `, fig. [fig : xyzagnr](b)], two edge sides with each an extra hydrogen atom [` 2h - edge `, fig. [fig : xyzagnr](c)], one hydrogen replaced by a fluorine atom [` 1f - edge `, fig. [fig : xyzagnr](d)], and a dehydrogenated edge with 4 hydrogen atoms removed from each side [` 8h - free `, fig. [fig : xyzagnr](e)]. we have also considered defects in the atomic structure in the form of one, two, or four broken c - c bonds [` 1c - broken `, ` 2c - broken `, ` 4c - broken `, fig. [fig : xyzagnr](f)-(h)] as well as a cu adatom on the agnr [` cu - adatom `, fig. [fig : xyzagnr](i)]. for all these systems the entire dynamical region was relaxed, i.e., the parts of the agnrs shown in fig. [fig : xyzagnr]. defects may influence the iets signal in two ways. first, a defect can have a direct impact by changing the vibrational degrees of freedom. in order for the change in the vibrational spectrum to give a signal in the iets, the new vibrations must couple to the current, and preferably have frequencies which do not coincide with ones already giving iets signals for the pristine ribbons. second, a defect can substantially change the electronic structure and thereby have an impact on the e - ph couplings associated with the active modes or even the transmission eigenchannels of the pristine ribbons, e.g., changing a peak in the iets to a dip (and vice versa) or enhancing asymmetric contributions via eq. ([eq : asym]). the electronic properties of the pristine agnr is shown in fig. [fig : transagnr](a). the carbon dos projected to the device region (red curve) reveals a gap as expected from the band structure [fig. [fig : cleangnr](b)], which is significantly broadened from the coupling to the metallic electrodes. the two valence and two conduction bands in the considered energy range naturally explain that the total transmission (black curve) is bound below a value of 2. further, the ratio @xmath75 (green dashed line), measuring the minimum number of contributing channels where @xmath74 is the transmission of the most transmitting eigenchannel, shows that both channels play a role for the transport, at least away from the edges of the direct band gap. measurements of shot noise may provide insights into this effective number of conductance eigenchannels.@xcite we can now discuss how the different defects modify the electronic properties. from fig. [fig : transagnr](b)-(i) we notice that not all defects change the elastic transmission, and furthermore, a change in elastic transmission needs not be unique for a specific defect. instead, iets may provide a additional fingerprint in the current that can be used to identify the type of defect. figure [fig : ietsdefectagnr] shows the computed iets as a function of gate voltage for the 8 different defects. as for the clean structure, the two peaks at @xmath1 and @xmath2 mev corresponding to the d- and g- raman modes are dominant for a range of gate values for all the structures. another feature, which is present in all the systems, is the appearance of several signals close to the band onsets. in the following subsections we discuss in more detail the transport characteristics with the different types of defects in agnrs. considering defects in the edge passivation [fig. [fig : xyzagnr](b - e)] the gap in the transmission is essentially unchanged [fig. [fig : transagnr](b - e)], except for the ` 1h - edge ` structure where a zero - energy resonance appears in the dos and transmission [fig. [fig : transagnr](b)]. this new peak can be attributed to tunneling via a mid - gap state which appears due to the local breaking of sub - lattice symmetry.@xcite thus, if a h atom is added to the neighboring c atom [` 2h - edge `, fig. [fig : xyzagnr](c)] the peak disappears [fig. [fig : transagnr](c)]. the addition of one or two h atoms on the same side also results in the closing of one transmission channel between the valence and conduction bands as shown in fig. [fig : transagnr](b, c). concerning the vibrational degrees of freedom, the addition of extra hydrogen to the edge results in new vibrational modes around 330 mev for ` 1h - edge ` and around 343 and 353 mev for ` 2h - edge `, clearly outside the bulk phonon band (ranging up to @xmath76 mev) of pristine agnr.@xcite comparing the iets in fig. [fig : ietsdefectagnr](a - c) we find that only ` 1h - edge ` gives a signal which differs significantly from the pristine case. figure [fig : ietsdefectagnr](k) shows specific iets for selected gate voltages for ` 1h - edge `. here, at @xmath77v (top green curve) we see how new signals appear at large voltages : for positive bias polarity two signals appear at 330 and 365 mev, respectively, while for negative bias polarity only an asymmetric signal around @xmath78 mev is present. the signal at 330 mev is due to vibrations of the @xmath79 [fig. [fig : armmodes](b)], while the signal at 365 mev [fig. [fig : armmodes](a)] is due to the h atom on the neighboring c atom. further, the amplitude of the signals around 169 and 196 mev is also found to depend on bias polarity. gating onto the zero - energy resonance for ` 1h - edge ` the iets signal [middle red curve in fig. [fig : ietsdefectagnr](k)] is dominated by large asymmetric signals for low energy vibrations due to the contribution from @xmath30 and eq. ([eq : asym]). we note that @xmath30 changes sign with bias polarity for this approximately left - right symmetric structure. this can be seen from the red iets curve in fig. [fig : ietsdefectagnr](k) which is roughly an odd function of the bias voltage. in close proximity of the zero - energy resonance a characteristic `` x - shape '' is observed in the gate - dependent iets, while away from it the signals approach that of the pristine agnr [fig. [fig : ietsdefectagnr](b)]. substituting a h atom with a f atom (` 1f - edge `) is seen to have virtually no effect in the iets of fig. [fig : ietsdefectagnr](d). this suggests that a significant change in the chemical composition directly involving the @xmath80-electronic system is required in order to obtain a signal although the vibrations are influenced by the heavier passivation. such a significant change in the passivation occurs for instance by removing four h atoms on each side (` 8h - free `), giving rise to four very narrow peaks in the dos around the conduction band, [fig. [fig : transagnr](e)]. these correspond to very localized dangling - bond states on the dehydrogenated dimers and therefore do not show up in the transmission. however, the dehydrogenated edges give rise to localized vibrations outside the range of the pristine vibrational spectrum.@xcite the in - phase vibration of the dehydrogenated c dimers at the armchair edges [fig. [fig : armmodes](f)] gives rise to an extra iets peak at 244 mev [fig. [fig : ietsdefectagnr](l)] matching the h - free mode measured by raman.@xcite we find that this signal is robust as it appears in the whole range of gate values. when gating into to the valence band a new signal appears around 43 mev [@xmath81 v in fig. [fig : ietsdefectagnr](e)] originating from a low energy edge vibration [fig. [fig : armmodes](g)]. the electronic transmission in gnrs is mediated by the carbon @xmath80 system. thus if a c - c bond fails to be formed during gnr synthesis or if it is broken again at a later stage, a large effect can be expected for the electronic conduction properties. this impact is indeed revealed in fig. [fig : transagnr](f - h). breaking one or two bonds results in the formation of two in - gap states which, broadened by the electrodes, make the gap appear smaller. the iets signals for the ` 1c - broken ` and ` 2c - broken ` in fig. [fig : ietsdefectagnr](f, g, m) have the same two signals at @xmath1 and @xmath2 mev as for the clean ribbon. however, the relative amplitudes are interchanged such that the `` d''-peak is now slightly more intense than the `` g''-peak. breaking four c - c bonds [` 4c - broken `, fig. [fig : xyzagnr](h)], resulting in constrictions of single c - c bonds, totally alter the dos which is now dominated by three sharp peaks as seen in fig. [fig : transagnr](h). the corresponding iets signals are shown in fig. [fig : ietsdefectagnr](h, n). in the proximity of the zero - energy resonance a broad range of signals at low vibrational energies appears (red curve in panel n) as well as a characteristic `` x - shape '' in the gate plot (panel h) similar to that of ` 1h - edge `. gating away from the resonance we observe two additional robust iets signals at @xmath82 and @xmath83 mev resulting from vibrations localized at the defect [fig. [fig : armmodes](d, e)]. transition metals are typically used for growth of graphene or as a substrate for the bottom - up synthesis of gnrs. thus it is of interest to consider the effect of adatoms of this type on gnrs. a cu adatom on graphene adsorbs preferentially in the on - top position.@xcite however, positioning cu such that it breaks the axial symmetry of our agnr, we find that it is most stable in a hollow site at the edge [` cu - adatom `, fig. [fig : xyzagnr](i)]. the dos and transmission in fig. [fig : transagnr](i) reveal a n - type doping effect shifting @xmath0 close to the conduction band while leaving the two transmission channels inside the gap relatively intact. for the pristine gnr the e - ph couplings of the out - of - plane vibrations are suppressed due to the symmetry of the @xmath80-orbitals. however, around the onset of the conduction band the iets signals in fig. [fig : ietsdefectagnr](i, o) is dominated by large asymmetric signals with significant contributions from out - of - plane phonons. these modes come into play due to breaking of the planar symmetry by the adatom. also note that by gating of @xmath0 within the gap these signatures of the adatom disappear, cf. the lower black curve in fig. [fig : ietsdefectagnr](o). let us next consider a series of defects for the zigzag graphene nanoribbon. in fig. [fig : structdefectzgnr] we show the atomic structures of pristine zgnr along with 8 different defect configurations. we consider the following defects in the edge - passivation : a single edge with an extra hydrogen [` 1h - edge `, fig. [fig : structdefectzgnr](b)], one hydrogen is replaced by either a f atom [` 1f - edge `, fig. [fig : structdefectzgnr](c)], an oh group [` 1oh - edge `, fig. [fig : structdefectzgnr](f)], or a @xmath84 group [` 1no_2-edge `, fig. [fig : structdefectzgnr](g)]. we also consider defects in the form of a cu adatom [` cu - adatom `, fig. [fig : structdefectzgnr](d)] or a li adatom [` li - adatom `, fig. [fig : structdefectzgnr](e)]. finally, we also study the effect of a structural defect in form of a 57 reconstruction [` r57 `, fig. [fig : structdefectzgnr](h)] and a substitutional defect where a c atom next to the edge is replaced by a si atom [` si - substitute `, fig. [fig : structdefectzgnr](i)]. for all these systems the entire dynamical region was relaxed, i.e., the parts of the zgnrs shown in fig. [fig : structdefectzgnr] using spin - polarized treatments. the spin degrees of freedom @xmath85 generalizes @xmath86 and @xmath87 [eqs.-] corresponding to two independent spin channels, which in general can have quite different amplitudes and even opposite sign. the observable iets would simply be the sum of these two components @xmath88. similar to the agnr case, the electronic properties in the device region with the different impurity configurations for the zgnr, now spin resolved, are summarized in fig. [fig : transdefectzgnr]. the iets of pristine zgnr was already discussed in sec. [subsec : cleanzgnr] and below we continue describing the iets fingerprints for the various defects. as commented above, the broader iets signals of pristine zgnr [figs. [fig : convclean]-[fig : ietsclean]] (as compared with agnr) can be understood from the breaking of the axial mirror symmetry and hence fewer symmetry - forbidden inelastic transitions. these broader signals may in general make the detection of defect signatures more difficult. for ` 1h - edge ` [fig. [fig : transdefectzgnr](b)] the iets resembles that of the pristine zgnr [fig. [fig : transdefectzgnr](a)] inside the gap. however, gating into the valance band [black curve in fig. [fig : transdefectzgnr](k)] the edge states start to extend into the middle of the ribbon, partially restoring mirror symmetry, and thus resulting in part of the pristine zgnr signals to disappear. here an extra signal appear due to edge - modes in the frequency range 194 to 199 mev with the most contributing mode at 196 mev as shown in [fig. [fig : zigmodes](a)]. the resulting iets signal can clearly be seen in the bottom curve in fig. [fig : ietsdefectzgnr](k). as for the agnr substituting a hydrogen with a fluorine atom has a very limited effect on the electronic properties and the iets signal. substituting a hydrogen with an oh group, according to fig. [fig : transdefectzgnr](f) and (o), have only a small effect on the spin down electrons, while it shrinks and add additional structure to the gap for the spin up electrons. for the spin up electrons there is a small peak inside the gap which gives rise to a large asymmetric iets signal around @xmath89v in fig. [fig : transdefectzgnr](o) lower curve, compared to the pristine case. the most contributing mode to the asymmetric iets signal is shown in fig. [fig : zigmodes](b). however, there is no clear signature of the oh group itself. in the same manner the substitution with a @xmath84 group removes the gap in the electronic properties without leaving any direct fingerprint of the @xmath84 group in the iets signal. as for the agnr we consider the effect of adatoms. for the cu adatom the transport gap shrinks for the spin up electrons while there is an in - gap peak for the spin down electrons, cf. [fig : transdefectzgnr](m). thus, for some gate values the iets signals reflect that the spin down electrons will back scatter while the spin up electrons will be forward scattered, and the observed signal is then the sum of these contributions. for a gate value of @xmath89 v, the iets signal is dominated by spin down electrons. due to the finite width of the in - gap peak, in the spin down transmission, the low frequency phonons (@xmath90 mev) give rise to back scattering while the high frequency phonons (@xmath91 mev) result in forward scattering. thus, the low and high energy signals have different signs as can be seen from fig. [fig : ietsdefectzgnr](l). interestingly, the low energy signal primarily consists of symmetric contributions from out - of - plane modes [fig. [fig : zigmodes](c)]. replacing the cu adatom with li, the transmission and dos, shown in fig. [fig : transdefectzgnr](e, n), reveals a spin dependent n - type doping effect, where @xmath0 is shifted the most for spin down. however, no in - gap peak is seen as for cu and the iets show no clear signature of the li atom. the formation of a ` r57 ` reconstruction results in peaks in the dos in the device region, just above @xmath0 for spin up [fig. [fig : transdefectzgnr](h)] and just below @xmath0 for spin down [fig. [fig : transdefectzgnr](q)]. the ` r57 ` breaks the symmetry both in the vibrational and electronic structure allowing for iets signals from a wider range of vibrations, resulting in broader peaks, as seen from fig. [fig : ietsdefectzgnr](h) and fig. [fig : ietsdefectzgnr](n). one of the contributing modes is localized at the border between the pentagon - ring and middle of the ribbon at @xmath92 [fig. [fig : zigmodes](d)]. this localized mode yield a relatively small signal compared to the other signals, however, contrary to the other modes the localized mode is not expected to be broadened if the coupling to phonons away from the dynamical region is taken into account. the breaking of symmetry in the electronic structure also give rise to difference signals for the two bias polarities. substituting a carbon with a silicon atom leads to an out - of - plane buckling, see fig. [fig : structdefectzgnr](i). however, both silicon and carbon have an @xmath93 electronic structure, and the electron transmission is basically similar to the pristine. on the other hand, the buckling give rise to low energy peaks in the iets signal originating from the e - ph coupling to the out - of - plane modes [fig. [fig : zigmodes](e)]. gating close to the band edge of the conduction band gives rise to different sign of the signals at low and high vibrational energies, as seen from the top curve in fig. [fig : ietsdefectzgnr](o).
Conclusions
in summary, we have investigated iets signals in symmetrically contacted armchair and zigzag graphene nanoribbons, considering both pristine as well as a selection of defected configurations under varying charge carrier conditions. for the clean agnr inelastic tunneling gives rise to two distinct peaks in the iets spectrum at @xmath1 mv and @xmath2 mv corresponding to the d - and g - modes of raman spectroscopy, respectively. by connecting the iets signals to the phonon band structure, we have clarified how only a single band contributes to the `` g - mode '' while three bands contribute to the broader `` d - mode ''. concerning defects in agnrs we have shown how some leave iets unchanged while others give clear signals. for instance, adding an extra hydrogen atom to a single edge side gives a clear signal for some gate values. this signal can be removed by adding another hydrogen atom to the neighboring edge side because the sub - lattice symmetry is restored. further, exchanging a single hydrogen atom with a fluorine atom in the passivation does not result in any change in both the elastic and inelastic tunneling. however removing 8 hydrogen atoms leaving part of the edge on each side without passivation, gives a clear robust signal throughout the investigated gate values. the signal, due to the vibration of the carbon dimers at the edge, has an energy around @xmath3 mev making it easy to detect since it is outside the vibrational spectrum of the pristine ribbon. breaking of one or two c - c bonds turns out to interchange the relative intensity of the `` g''- and `` d''-peaks. breaking 4 c - c bonds gives rise to signals caused by the defect tilted out of plane. lifting the symmetry of the @xmath80-electrons by adding a cu - adatom allows the out - of - plane modes to contribute. for the zgnr we find relatively broader iets signals especially in the absense of a large gate voltage (@xmath94 v). importantly, this is a consequence of the breaking of the axial mirror symmetry in the ribbon due to the presence of spin - polarized edges. thus, by comparing to spin - degenerate calculations, we suggest that iets can give an indirect proof of spin - polarization in zigzag ribbons. on the other hand, the broader iets features may make it difficult to identify the different defect signals reported in this paper. the presence of a ` r57`-reconstruction also broadens the iets by breaking both the electronic and vibrational symmetry. substituting a carbon atom with a silicon atom makes the ribbon buckle, breaking the planar symmetry, allowing the out - of - plane modes to contribute to the iets. this suggests that iets in principle could be used to gain information of the curvature of gnrs and other graphene - based structures. finally, as an outlook we note that here we presented calculations on long, symmetrically contacted systems where there is a significant overlap with both metallic electrodes. it would be interesting to extend such a study also to the asymmetric situation where a point tunnel contact is made to one end resembling, say, the coupling a stm tip.@xcite | inspired by recent experiments where electron transport was measured across graphene nanoribbons (gnr) suspended between a metal surface and the tip of a scanning tunneling microscope [koch _ et al_., nat.
nanotechnol.
* 7 *, 713 (2012)], we present detailed first - principles simulations of inelastic electron tunneling spectroscopy (iets) of long pristine and defected armchair and zigzag nanoribbons under a range of charge carrier conditions. for the armchair ribbons we find two robust iets signals around 169 and 196 mv corresponding to the d- and g - modes of raman spectroscopy as well as additional fingerprints due to various types of defects in the edge passivation. for
the zigzag ribbons we show that the spin state strongly influences the spectrum and thus propose iets as an indirect proof of spin polarization. | 1501.02266 |
Introduction
chemical master equations (cmes) are the accepted mathematical description of chemical systems in well - mixed conditions @xcite. these equations provide a mesoscopic description of chemical kinetics, interpolating between the microscopic regime of molecular dynamics and the macroscopic regime of rate equations (res). it has been shown that cmes are exact descriptions for any well - stirred and thermally equilibrated gas - phase chemical system @xcite. more recently it has been rigorously confirmed that their validity extends to chemical reactions in well - stirred dilute solutions @xcite. however, well before these rigorous demonstrations of the microscopic physical basis of the cme, scientists have employed these equations to probe the nature of mesoscopic chemical kinetics and in particular to understand how this may differ from kinetics on macroscopic length scales (see mcquarrie for a review @xcite of the literature up till 1967). we briefly review the cme formalism. consider a general chemical system consisting of a number @xmath4 of distinct chemical species interacting via @xmath5 elementary chemical reactions of the type : @xmath6 here @xmath7 is an index running from 1 to @xmath5, @xmath8 denotes chemical species @xmath9, @xmath10 and @xmath11 are the stoichiometric coefficients and @xmath12 is the macroscopic rate of reaction. if this system is well - mixed then its mesoscopic state is fully determined by the vector of the absolute number of molecules of each species, @xmath13, where @xmath14 is the number of molecules of the @xmath15species. the cme is then a time - evolution equation for the probability of the system being in a particular mesoscopic state @xcite : @xmath16 where @xmath2 is the volume of the compartment in which the reactions occur and @xmath17 is a step operator when it acts on some function of the absolute number of molecules, it gives back the same function but with @xmath14 replaced by @xmath18. the chemical reaction details are encapsulated in the stoichiometric matrix @xmath19 and in the microscopic rate functions @xmath20. the probability that the @xmath21 reaction occurs in the time interval @xmath22 is given by @xmath23. for elementary reactions, the microscopic rate function takes one of four different forms, depending on the order of the @xmath21 reaction : (i) a zeroth - order reaction by which a species is input into a compartment gives @xmath24 ; (ii) a first - order unimolecular reaction involving the decay of some species @xmath25 gives @xmath26 ; (iii) a second - order bimolecular reaction between two molecules of the same species @xmath25 gives @xmath27 ; (iii) a second - order bimolecular reaction between two molecules of different species, @xmath25 and @xmath28, gives @xmath29. the re description of the same system is much simpler. denoting the macroscopic concentration of species @xmath9 by @xmath30, the set of res describing the macroscopic kinetics of the reactive system represented by eq. (1) are given by : @xmath31 where @xmath32 is the vector of macroscopic concentrations and @xmath33 is the macroscopic rate function of the @xmath21 reaction which has the general mass - action form, @xmath34. res provide a continuous deterministic `` many molecule '' description of kinetics. this strongly contrasts with the cme description which constitutes a discrete, stochastic, `` any number of molecule '' description that is faithful to the underlying microscopic basis of chemical reactions. unfortunately, one of the main advantages of cmes over their re cousins, their discrete description, is also the source of their computational intractability. differential - difference equations, such as the cme @xcite, do not lend themselves easily to analysis. in contrast, there is a vast body of literature in engineering, mathematics and physics dealing with the analysis and solution of differential and partial differential equations. thus at an early stage, considerable effort was invested in obtaining a partial differential approximation of the cme. in the 1940 s, kramers @xcite and moyal @xcite developed a taylor series expansion of the cme ; by assuming that all terms with derivatives greater than two are negligible, one obtains the chemical fokker - planck equation (cfpe, @xcite), a second - order partial differential equation of the form : @xmath35 as gardiner mentions in his book @xcite, `` this procedure enjoyed wide popularity mainly because of the convenience and simplicity of the result '' and also because `` it is often simpler to use the fokker - planck equation than the master equation. '' a major and important difference between the cme and the cfpe is that @xmath14 is a positive integer for the cme while it is a real number for the cfpe. several authors have questioned the validity of the cfpe approximation. the approximation is obtained by a perfunctory truncation of the taylor expansion and hence it appears to be an uncontrolled and unjustified approximation of the cme. van kampen, in particular, was a leading and influential critic of the cfpe approximation. in the 1960 s and 70 s, he developed a systematic perturbative expansion of the cme in powers of the inverse square root of the system volume @xmath2 (the system - size expansion) and used it to show that to lowest order in the expansion, i.e. in the limit of large volumes the macroscopic limit, one obtains a fokker - planck equation which is of a different form than the cfpe @xcite. of particular concern is that van kampen s fokker - planck equation is linear whereas the cfpe is non - linear. note that by non - linear fokker - planck equation here we mean one such that its drift and diffusion coefficients are generally non - linear functions of the molecule numbers @xmath14 ; this convention is adopted since it is in mainstream use, for example see the book by van kampen @xcite. taking into account higher - order terms in the system - size expansion does not lead to the cfpe as well. however, interestingly, in the limit of large volumes, the cfpe does reduce to van kampen s linear fokker - planck equation @xcite. this led van kampen to conclude that any features arising from the non - linear character of the cfpe are spurious and not to be taken seriously @xcite. we note that the limit of large volumes in van kampen s system - size expansion is taken at fixed macroscopic concentrations and hence it corresponds to the limit of large molecule numbers @xcite. hence van kampen s conclusions can be equivalently stated as : the cfpe becomes a legitimate approximation of the cme in the limit of large molecular populations. a few studies at the time @xcite did suggest that the cfpe s validity extended beyond the linear regime. of particular importance is a result of horsthemke and brenig @xcite which motivated the present study. the authors considered a simple dimerization reaction @xmath36 whereby molecules of a monomer species @xmath37 are introduced in a compartment of volume @xmath2 and subsequently they bind to each other to form dimers @xmath38. assuming stationary conditions, the cme and cfpe are solved exactly. it is shown that the average concentration of monomers and the variance of the fluctuations from the two formalisms agree exactly to order @xmath39 and are respectively equal to @xmath40 and @xmath41, where @xmath42 is the macroscopic concentration obtained by solving the corresponding re in steady - state conditions. the same example can be found worked in van kampen s book @xcite wherein he shows that the linear noise approximation gives mean and variance equal to @xmath42 and @xmath41. as we mentioned before, a linearization of the cfpe will lead to the linear - noise approximation and hence from this example we can conclude that the non - linearity of the cfpe is non - spurious since it leads to a more accurate concentration estimate than that which is obtained from the linear - noise approximation. however one could argue that this higher accuracy is only particular to the dimerization example and not a general feature of the cfpe. because of this or other reasons, the results of hortshemke and brenig do not appear to have received the attention they deserved at the time and van kampen s conclusions about the cfpe were accepted, by and large, by the statistical physics community. approximately 40 years later after the inception of the system - size expansion, gillespie revived the question of the validity of the cfpe by deriving it without invoking truncation of the kramers - moyal expansion of the cme @xcite. to be precise, he derived the chemical langevin equation (cle) : @xmath43 where @xmath44 are temporally uncorrelated, independent gaussian white noises. this stochastic differential equation is exactly equivalent to the cfpe in the sense that its solution generates exact sample paths of the cfpe, eq. (4). essentially he showed that the cfpe approximation is valid provided two conditions are satisfied. a large number of molecules suffices to ensure that both conditions are satisfied however this is not a necessary condition. this suggests that there are regimes in which the particle numbers may not be very large and yet the cfpe may still provide a reasonably good approximation of the cme. however gillespie s derivation does not provide us with a means to estimate the accuracy of the cfpe for general chemical systems. questions regarding the validity and accuracy of the cfpe and cle are more important now than ever before. in the past decade, interest has virtually exploded in realistic stochastic simulations of biochemical reactions inside cells @xcite. the exact method of sampling the trajectories of the cme, the stochastic simulation algorithm @xcite, is computationally expensive and the cme is analytically intractable ; thus approximate methods such as the cfpe and the cle have come to the foreground as an alternative means to obtain numerical and theoretical insight into the functioning of intracellular biochemical networks @xcite. these networks are typically characterized by a large number of bimolecular reactions in which at least one of the species is present in very small molecule numbers @xcite, indeed the precise conditions in which the fidelity of the cfpe remains unclear. hence the question of the accuracy of the cfpe has nowadays become a practical one how much can we trust the conclusions derived from the cfpe or the corresponding cle? in this article, we derive formulas to estimate the relative error in the cfpe predictions of the mean concentrations and of the variance of the fluctuations about the mean. the results are valid for all monostable chemical reaction networks. as a byproduct of our derivation, we will also clarify the connection between the cfpe and van kampen s system - size expansion, in particular showing that the non - linear character of the cfpe is not completely spurious and that generally cfpe estimates are more accurate than those obtained from the linear fokker - planck equation. the article is organized as follows. in section ii, we use the multivariate system - size expansion to derive expressions for the mean concentrations and for the variance of the fluctuations as predicted by the cme accurate to order @xmath45. in section iii, we develop the system - size expansion of the cfpe and use it to derive expressions for the mean concentrations and for the variance of the fluctuations accurate to the same order as derived for the cme in section ii. in section iv, we use the results of the previous two sections to derive expressions for the relative error in the predictions of the cfpe. we also compare the predictions of the cfpe and the linear fokker - planck equation. these results are tested on two bimolecular reaction systems dimerization and an enzyme - catalyzed reaction in section v. we conclude by a discussion in section vi.
Perturbative expansion of the cme
we will now probe the mesoscopic description provided by the cme using the system - size expansion developed by van kampen @xcite. this method allows one to derive expressions for the mean concentrations and for the variance of the fluctuations about these concentrations, as predicted by the cme, accurate to the order of any desired power of the inverse square root of the volume. the only requirement for the expansion to hold is that the steady - state of the chemical system is asymptotically stable. for the applications that we are interested in, namely biochemical reactions in intracellular conditions, the number of molecules can be very small, in some cases just few tens of molecules of a given species per cell. we will derive equations accurate to @xmath45 this accuracy should be more than sufficient for the applications mentioned since terms of lower order, @xmath46, already imply corrections to the concentrations of the order of a single molecule in the compartment. to our knowledge this is the first time that the system - size expansion has been carried to this order for a general system of @xmath4 interacting chemical species. van kampen has treated a one species example to the same order in his book @xcite while elf and ehrenberg @xcite have derived the multivariate expansion to @xmath47. the starting point of the system - size expansion is to write the absolute number of molecules of species @xmath9 as : @xmath48 where @xmath30 is the macroscopic concentration of species @xmath9 as determined by the res. this has the effect of transforming all functions of @xmath14 in the cme into functions of @xmath49. the expansion of the cme proceeds by writing eq. (2) in terms of the new variables. details of this transformation can be found in @xcite ; here we will simply state the relevant results and use them for our present derivation. the variable change causes the probability distribution of molecular populations, @xmath50, to be transformed into the probability distribution of fluctuations, @xmath51, where @xmath52. the time derivative, the step operator and the microscopic rate function in the cme, read in the new variables : @xmath53 where @xmath54 note that in eq. (9) the microscopic rate function is expressed in terms of the macroscopic rate function. as we shall shortly see, this is convenient from a calculation point of view since the final expressions for the means and variances will be solely in terms of functions which appear in the res. note that the upper limit of the sum in eq. (9) is 2 because all reactions involve at most the interaction of two molecules and hence @xmath55 equals zero for @xmath56. although our analysis is specifically for elementary reactions, one can easily extend the approach to include `` elementary complex '' reactions @xcite. however we shall not pursue this here. substituting eqs. (7 - 9) in eq. (2) we get the following new form of the cme : @xmath57 note that terms proportional to @xmath58 do not appear in the expansion of the cme. this is because when one substitutes eqs. (7 - 9) in eq. (2), one equates terms of this order on both sides of the cme which simply gives us back the macroscopic res, eq. (3). to proceed further we need the explicit dependence of the right hand side of eq. (14) on the new variables @xmath49. this is obtained by substituting eqs. (10 - 13) in eq. (14) which leads to : @xmath59 note that in the above equation, we have used the einstein summation convention where all twice repeated indices are understood to be summed over 1 to @xmath4. the partial derivative @xmath60 denotes @xmath61. we have also used the following two convenient definitions : @xmath62 from eq. (3) it follows that @xmath63 and consequently @xmath64 represents the @xmath9-@xmath65 element of the jacobian matrix associated with the res of the system. note that eq. (15) to order @xmath66 is the linear fokker - planck equation which was mentioned in the introduction. the drift vector is linear in the @xmath67 variables while the diffusion tensor is independent of them. both depend on time via their own dependence on the macroscopic concentrations. this level of approximation is frequently called the linear - noise approximation, a nowadays popular means of estimating the size of the concentration fluctuations about the macroscopic concentrations @xcite. we are interested in the dynamics on mesoscopic length scales and hence we shall consider terms of higher order than @xmath66 in eq. (15). we now proceed to construct equations for the moments of the @xmath67 variables. we start by expanding @xmath51 as a series in powers of the inverse square root of the volume : @xmath68 from which it follows that the moments possess an equivalent expansion : @xmath69_j \omega^{-j/2},\]] where @xmath70_j = \int \epsilon_k \epsilon_m... \epsilon_r \ \pi_j(\vec{\epsilon},t) d\vec{\epsilon}.\]] the angled brackets denote the statistical average. some subtle points associated with the perturbative expansion in the probability density and with the physical interpretation of @xmath71_j$] are discussed in appendix a. the time - evolution equations for the moments are obtained as follows. one starts by substituting eq. (18) in eq. (15), multiplying the resulting equation on both sides by @xmath72 and integrating over @xmath73. equating terms of order @xmath74 on both sides of the equation gives the time - evolution equation for @xmath71_j$]. finally one constructs the time - evolution equation for the moments using eq. (19). as mentioned earlier, our aim is to determine the mean concentrations and the variance of the fluctuations about the means and hence we must relate the latter to the moments of the @xmath67 variables above. using eqs. (6) and (19), one can easily verify that the mean concentration of species @xmath9 and the variance of the fluctuations about it, accurate to order @xmath1 are respectively given by : @xmath75_j \omega^{-j/2 } + o(\omega^{-5/2 }), \\ \sigma_i^2 & = \biggl \langle \biggl(\frac{n_i}{\omega } \biggr)^2 \biggr \rangle - \biggl \langle \frac{n_i}{\omega } \biggr \rangle^2 = \omega^{-1 } (\langle \epsilon_i^2 \rangle - \langle \epsilon_i \rangle^2) \nonumber \\ & = \omega^{-1 } \biggr (\sum_{j=0}^{2 } [\epsilon_i^2] _ j \omega^{-j/2 } - \biggl(\sum_{j=0}^{1 } [\epsilon_i] _ j \omega^{-j/2 } \biggr)^2 - \omega^{-1 } [\epsilon_i] _ 0 [\epsilon_i] _ 2 \biggl) + o(\omega^{-5/2}).\end{aligned}\]] hence it is clear that to determine the mean and variance accurate to order @xmath1, we shall need to determine the first and second moments of the @xmath67 variables accurate to orders @xmath0 and @xmath39 respectively. we proceed by implementing the calculation recipe outlined just after eq. (20) to derive equations for the corrections to the second moments accurate to order @xmath39 : @xmath76_0 & = j_r^{w } [\epsilon_w \epsilon_k] _ 0 + (r \leftrightarrow k) + d_{rk }, \\ \frac{\partial}{\partial t } [\epsilon_r \epsilon_k] _ 1 & = j_r^{w } [\epsilon_w \epsilon_k] _ 1 + \frac{1}{2 } j_r^{wp } [\epsilon_w \epsilon_p \epsilon_k] _ 0 - \frac{1}{2 } j_r^{w(2) } \phi_w [\epsilon_k] _ 0 \nonumber \\ & + (r \leftrightarrow k) + j_{kr}^w [\epsilon_w] _ 0, \\ \frac{\partial}{\partial t } [\epsilon_r \epsilon_k] _ 2 & = j_r^{w } [\epsilon_w \epsilon_k] _ 2 + \frac{1}{2 } j_r^{wp } [\epsilon_w \epsilon_p \epsilon_k] _ 1 - \frac{1}{2 } j_r^{w(2) } \phi_w [\epsilon_k] _ 1 \nonumber \\ & - \frac{1}{2 } j_r^{w(2) } [\epsilon_w \epsilon_k] _ 0 + (r \leftrightarrow k) + j_{kr}^w [\epsilon_w] _ 1 + \frac{1}{2 } j_{rk}^{wm } [\epsilon_w \epsilon_m] _ 0 - \frac{1}{2 } j_{rk}^{w(2) } \phi_w.\end{aligned}\]] details of the calculations leading to the above equations are illustrated by a step - by - step derivation of eq. (25) in appendix b. note that the short - hand notation @xmath77 stands for all the expressions of the same form as the ones preceding the notation but with @xmath78 and @xmath79 interchanged. for example in eq. (24), @xmath77 stands for @xmath80_1 + \frac{1}{2 } j_k^{wp } [\epsilon_w \epsilon_p \epsilon_r] _ 0 - \frac{1}{2 } j_k^{w(2) } \phi_w [\epsilon_r] _ 0 $]. this notation will be used throughout the rest of the article since it enables the equations to be written in a compact way. the equation for @xmath81_0 $], eq. (23), is a lyapunov equation which can be solved either analytically (see for example @xcite) or else numerically, for example using the built in functions of matlab and mathematica. solution of the equation for @xmath81_1 $], eq. (24), requires the solutions of the equations for the first and third moments to order @xmath66 : @xmath82_0 & = j_r^{w } [\epsilon_w] _ 0, \\ \frac{\partial}{\partial t } [\epsilon_r \epsilon_k \epsilon_l] _ 0 & = j_l^{w } [\epsilon_w \epsilon_k \epsilon_r] _ 0 + (l \leftrightarrow k) + (k \leftrightarrow r) \nonumber \\ & + d_{rl } [\epsilon_k] _ 0 + (k \leftrightarrow l) + (r \leftrightarrow l).\end{aligned}\]] note that in eq. (27), @xmath83 stands for two expressions ; the first expression corresponds to the first term on the right hand side of eq. (27) with @xmath84 and @xmath79 interchanged and the second expression is the first expression just obtained with @xmath79 and @xmath78 interchanged. by a similar reasoning, it follows that @xmath85 in eq. (27) stands for @xmath86_0 + d_{lk } [\epsilon_r] _ 0 $]. note that in steady - state conditions, @xmath87_0 = [\epsilon_r \epsilon_k \epsilon_l] _ 0 = 0 $] and and consequently there is no correction to the second moments to @xmath46, i.e., @xmath81_1 = 0 $]. solution of the equation for @xmath81_2 $], eq. (25), requires the solutions of the corrections to the the first and third moments to order @xmath3 and the second and fourth moments to order @xmath88 : @xmath82_1 & = j_r^{w } [\epsilon_w] _ 1 + \frac{1}{2 } j_r^{wp } [\epsilon_w \epsilon_p] _ 0 - \frac{1}{2 } j_r^{w(2) } \phi_w, \\ \frac{\partial}{\partial t } [\epsilon_r \epsilon_k \epsilon_l] _ 1 & = j_l^{w } [\epsilon_w \epsilon_k \epsilon_r] _ 1 + \frac{1}{2 } j_l^{wp } [\epsilon_w \epsilon_p \epsilon_r \epsilon_k] _ 0 - \frac{1}{2 } j_l^{w(2) } \phi_w [\epsilon_r \epsilon_k] _ 0 \nonumber \\ & + (l \leftrightarrow k) + (k \leftrightarrow r) + d_{rl } [\epsilon_k] _ 1 + j_{rl}^w [\epsilon_w \epsilon_k] _ 0 + (k \leftrightarrow l) \nonumber \\ & + (r \leftrightarrow l) + d_{rkl }, \\ \frac{\partial}{\partial t } [\epsilon_r \epsilon_k \epsilon_l \epsilon_m] _ 0 & = j_r^{w } [\epsilon_w \epsilon_k \epsilon_l \epsilon_m] _ 0 + (r \leftrightarrow m) + (m \leftrightarrow k) + (k \leftrightarrow l) + d_{rm } [\epsilon_k \epsilon_l] _ 0 \nonumber \\ & + (m \leftrightarrow l) + (l \leftrightarrow k) + (r \leftrightarrow m) + (m \leftrightarrow l) + (k \leftrightarrow m).\end{aligned}\]] the procedure to obtain the second moments to order @xmath39 is now clear. one first solves eq. (23) to get @xmath81_0 $] ; then one solves eqs. (26 - 27) and substitutes in eq. (24) to get @xmath81_1 $] ; finally one solves eqs. (28 - 30) and substitutes these, together with the solution of eq. (23), in eq. (25) to get @xmath81_2 $]. the first moment equations and the corresponding equations for the mean concentrations can be obtained in an analogous manner as for the second moments. the equations for @xmath87_0 $] and @xmath87_1 $] have been already derived, eqs. (26) and (28), respectively. the equations for @xmath87_2 $] and @xmath87_3 $] are given by : @xmath82_2 & = j_r^{w } [\epsilon_w] _ 2 + \frac{1}{2 } j_r^{wp } [\epsilon_w \epsilon_p] _ 1 - \frac{1}{2 } j_r^{w(2) } [\epsilon_w] _ 0, \\ \frac{\partial}{\partial t } [\epsilon_r] _ 3 & = j_r^{w } [\epsilon_w] _ 3 + \frac{1}{2 } j_r^{wp } [\epsilon_w \epsilon_p] _ 2 - \frac{1}{2 } j_r^{w(2) } [\epsilon_w] _ 1.\end{aligned}\]] the procedure to obtain the first moments to order @xmath0 is now also clear. one first solves eq. (26) to get @xmath87_0 $] ; then one solves eq. (23) and substitutes in eq. (28) to obtain @xmath87_1 $] ; finally one uses the solutions already obtained when deriving the second moments to solve eqs. (31 - 32) for @xmath87_2 $] and @xmath87_3 $]. given the first and second moments accurate to @xmath0 and @xmath39, one finally determines the mean concentrations and the variance of the fluctuations about them accurate to @xmath1 from eqs. (21 - 22). although the procedure of obtaining the latter final expressions is fairly laborious, as we shall see in the next section, in order to obtain the leading order error in the predictions of the cfpe, it will only be necessary for us to solve very few of these equations explicitly.
Perturbative expansion of the cfpe
in this section we use the system - size expansion to derive expressions for the mean concentrations and the fluctuations about them, as predicted by the cfpe, accurate to @xmath45. to the best of our knowledge this is the first time that the expansion has been used on the cfpe although the method is similar in principle to the small - noise expansion of fokker - planck equations as presented by gardiner @xcite. the cfpe is obtained by truncating the kramer s moyal expansion to include at most second - order derivatives : @xmath89 note that the second step above, follows by taylor expanding the step operator. next we perform the system - size expansion on the cfpe, eq. (33), i.e., we make the variable transformation given by eq. (6) which transforms functions of @xmath14 into functions of the new variables @xmath49. the probability distribution @xmath50 is transformed into a new one @xmath90. note that the subscript @xmath91 will be used to distinguish quantities calculated using the cfpe from those previously calculated using the cme. the time derivative on the left hand side of the equation and the microscopic rate function @xmath92 transform as in the case of the cme and are given by eqs. (7) and (9) together with the definitions eqs. (11 - 13) and with @xmath51 replaced by @xmath90. the operators involving derivatives with respect to absolute particle number transform as follows : @xmath93 where the operators @xmath94 are as defined in eq. hence the cfpe in the new variables reads : @xmath95 note that whereas the transformation given by eq. (6) on the cme leads to an infinite series in powers of the inverse square root of the volume, eq. (14), the same transformation on the cfpe leads to a finite series with the highest order term being of order @xmath0 (this is only true for elementary reactions). the derivation of the equations for the time evolution of the moments of the @xmath67 variables proceeds in an exactly analogous manner as to that presented in detail in section ii. the probability distribution is written as a series in powers of the inverse square root of the volume, @xmath96 and the moments are then generally given by : @xmath97_{f, j } \omega^{-j/2},\]] where @xmath70_{f, j } = \int \epsilon_k \epsilon_m... \epsilon_r \ \pi_{f, j}(\vec{\epsilon},t) d\vec{\epsilon}.\]] the equations for the mean concentrations and the variance of the fluctuations about them are given by eqs. (21 - 22) with the subscript @xmath91 carried throughout. the time evolution equations for the corrections to the moments can be derived as before. although there is some repetition involved, we will state these equations in full so that the differences between them and those derived using the cme are very clear. the equations for the corrections to the second moments accurate to order @xmath39 are : @xmath76_{f,0 } & = j_r^{w } [\epsilon_w \epsilon_k] _ { f,0 } + (r \leftrightarrow k) + d_{rk }, \\ \frac{\partial}{\partial t } [\epsilon_r \epsilon_k] _ { f,1 } & = j_r^{w } [\epsilon_w \epsilon_k] _ { f,1 } + \frac{1}{2 } j_r^{wp } [\epsilon_w \epsilon_p \epsilon_k] _ { f,0 } - \frac{1}{2 } j_r^{w(2) } \phi_w [\epsilon_k] _ { f,0 } \nonumber \\ & + (r \leftrightarrow k) + j_{kr}^w [\epsilon_w] _ { f,0 }, \\ \frac{\partial}{\partial t } [\epsilon_r \epsilon_k] _ { f,2 } & = j_r^{w } [\epsilon_w \epsilon_k] _ { f,2 } + \frac{1}{2 } j_r^{wp } [\epsilon_w \epsilon_p \epsilon_k] _ { f,1 } - \frac{1}{2 } j_r^{w(2) } \phi_w [\epsilon_k] _ { f,1 } \nonumber \\ & - \frac{1}{2 } j_r^{w(2) } [\epsilon_w \epsilon_k] _ { f,0 } + (r \leftrightarrow k) + j_{kr}^w [\epsilon_w] _ { f,1 } + \frac{1}{2 } j_{rk}^{wm } [\epsilon_w \epsilon_m] _ { f,0 } - \frac{1}{2 } j_{rk}^{w(2) } \phi_w.\end{aligned}\]] note that these are the same as eqs (23 - 25) but with subscript @xmath91 ; the implicit reason for this is that only terms containing @xmath98 and @xmath99 contribute to the equations for the second moments and all such terms are equally present in eqs. (14) and (36). note also that eqs. (23) and (40) lead to the same solution, i.e., @xmath100_{0 } = [\epsilon_r \epsilon_k] _ { f,0}$]. the solution of @xmath100_{f,1}$] is dependent on the solutions of the time evolution equations for @xmath101_{f,0}$] and @xmath102_{f,0}$]. the equations for the latter are the same as eqs. (26 - 27) but with subscript @xmath91 ; this is since eq. (14) and (36) are equal to order @xmath66. it follows that @xmath103_{0 } = [\epsilon_r] _ { f,0}$] and @xmath102_{0 } = [\epsilon_r \epsilon_k \epsilon_l]_{f,0}$] from which we can conclude using eq. (41) that @xmath100_{1 } = [\epsilon_r \epsilon_k] _ { f,1}$]. however, as we now show, generally @xmath100_{2 } \ne [\epsilon_r \epsilon_k] _ { f,2}$]. the solution of @xmath100_{f,2}$] is dependent on the solutions of the time evolution equations for @xmath101_{f,1}$], @xmath102_{f,1}$] and @xmath104_{f,0}$] which are : @xmath82_{f,1 } & = j_r^{w } [\epsilon_w] _ { f,1 } + \frac{1}{2 } j_r^{wp } [\epsilon_w \epsilon_p] _ { f,0 } - \frac{1}{2 } j_r^{w(2) } \phi_w, \\ \frac{\partial}{\partial t } [\epsilon_r \epsilon_k \epsilon_l] _ { f,1 } & = j_l^{w } [\epsilon_w \epsilon_k \epsilon_r] _ { f,1 } + \frac{1}{2 } j_l^{wp } [\epsilon_w \epsilon_p \epsilon_r \epsilon_k] _ { f,0 } - \frac{1}{2 } j_l^{w(2) } \phi_w [\epsilon_r \epsilon_k] _ { f,0 } \nonumber \\ & + (l \leftrightarrow k) + (k \leftrightarrow r) + d_{rl } [\epsilon_k] _ { f,1 } + j_{rl}^w [\epsilon_w \epsilon_k] _ { f,0 } + (k \leftrightarrow l) \nonumber \\ & + (r \leftrightarrow l), \\ \frac{\partial}{\partial t } [\epsilon_r \epsilon_k \epsilon_l \epsilon_m] _ { f,0 } & = j_r^{w } [\epsilon_w \epsilon_k \epsilon_l \epsilon_m] _ { f,0 } + (r \leftrightarrow m) + (m \leftrightarrow k) + (k \leftrightarrow l) + d_{rm } [\epsilon_k \epsilon_l] _ { f,0 } \nonumber \\ & + (m \leftrightarrow l) + (l \leftrightarrow k) + (r \leftrightarrow m) + (m \leftrightarrow l) + (k \leftrightarrow m).\end{aligned}\]] equations (43) and (45) have the same form as eqs. (28) and (30) respectively. this combined with the fact that the right hand sides of eqs. (43) and (45) are functions of @xmath100_{f,0}$] and that @xmath100_{0 } = [\epsilon_r \epsilon_k] _ { f,0}$], implies that @xmath103_{1 } = [\epsilon_r]_{f,1}$] and @xmath105_{0 } = [\epsilon_r \epsilon_k \epsilon_l \epsilon_m]_{f,0}$]. however note that eq. (44) has one term missing compared to its counterpart eq. (29) and hence generally @xmath102_{1 } \ne [\epsilon_r \epsilon_k \epsilon_l] _ { f,1}$] from which it follows using eq. (42) that @xmath100_{2 } \ne [\epsilon_r \epsilon_k] _ { f,2}$]. the only remaining equations to be considered are those paralleling eqs. (31) and (32) for which we find : @xmath82_{f,2 } & = j_r^{w } [\epsilon_w] _ { f,2 } + \frac{1}{2 } j_r^{wp } [\epsilon_w \epsilon_p] _ { f,1 } - \frac{1}{2 } j_r^{w(2) } [\epsilon_w] _ { f,0 }, \\ \frac{\partial}{\partial t } [\epsilon_r] _ { f,3 } & = j_r^{w } [\epsilon_w] _ { f,3 } + \frac{1}{2 } j_r^{wp } [\epsilon_w \epsilon_p] _ { f,2 } - \frac{1}{2 } j_r^{w(2) } [\epsilon_w] _ { f,1}.\end{aligned}\]] by similar arguments to the above, these equations imply @xmath103_{2 } = [\epsilon_r]_{f,2}$] and @xmath103_{3 } \ne [\epsilon_r]_{f,3}$]. hence, in summary, we have obtained the following results : 1. @xmath101_{0 } = [\epsilon_r]_{f,0}$], @xmath100_{0 } = [\epsilon_r \epsilon_k] _ { f,0}$], @xmath102_{0 } = [\epsilon_r \epsilon_k \epsilon_l]_{f,0}$], @xmath105_{0 } = [\epsilon_r \epsilon_k \epsilon_l \epsilon_m]_{f,0}$] 2. @xmath103_{1 } = [\epsilon_r]_{f,1}$], @xmath100_{1 } = [\epsilon_r \epsilon_k] _ { f,1}$], @xmath102_{1 } \ne [\epsilon_r \epsilon_k \epsilon_l] _ { f,1}$] 3. @xmath103_{2 } = [\epsilon_r]_{f,2}$], @xmath100_{2 } \ne [\epsilon_r \epsilon_k] _ { f,2}$] 4. @xmath103_{3 } \ne [\epsilon_r]_{f,3}$] note that these results are not for the moments but for the corrections to the moments ; the real physical meaning of these results in terms of means and variances will be elucidated in the next section. using eqs. (32) and (47), eqs (25) and (42) and eqs. (29) and (44), we can respectively write down simple equations for the differences in the corrections to the first, second and third moments as predicted by the cfpe and the cme : @xmath106 where we have used the convenient definitions : @xmath107_{3 } - [\epsilon_r]_{f,3 }, \\ \delta_{rk } & = [\epsilon_r \epsilon_k] _ { 2 } - [\epsilon_r \epsilon_k] _ { f,2 }, \\ \delta_{rkl } & = [\epsilon_r \epsilon_k \epsilon_l]_{1 } - [\epsilon_r \epsilon_k \epsilon_l] _ { f,1}.\end{aligned}\]]
Comparison of the predictions of the cfpe and the cme
in this section we will use the results derived in the last section to obtain formulas for the absolute and relative errors (to leading order) in the cfpe predictions of the mean concentrations and the variance of the fluctuations. using these formulas we will be able to deduce the general conditions in which the differences between the cfpe and the cme are minimal. furthermore we will show that the cfpe is generally more accurate than the linear fokker - planck equation of van kampen and that the mean concentrations of the cfpe to order @xmath39 are precisely the same as those obtained from effective mesoscopic rate equations. we will now derive expressions for the leading order term of the absolute and relative errors made by the cfpe in predicting the mean concentrations and the variance of the fluctuations about the mean concentrations. we will also obtain an expression for the leading order term of the absolute error made by the cfpe in predicting the skewness of the probability distribution of the concentrations. the mean concentration predicted by the cme, @xmath108, is given by eq. (21) while the mean concentration predicted by the cfpe, @xmath109 is given by the same equation but with the subscript @xmath91 carried throughout. subtracting the two expressions and using the summary of results in section iii together with eq. (51) we get the absolute error in the cfpe concentration : @xmath110 the relative error follows easily : @xmath111 \biggl \langle \frac{n_i}{\omega } \biggr\rangle^{-1 } = \frac{\delta_i}{\phi_i } \omega^{-2 } + o(\omega^{-5/2}).\]] similarly, using eq. (22) and using the summary of results in section iii together with eq. (52) we find the absolute and relative errors in the variance of the fluctuations to respectively be given by : @xmath112 where @xmath113 is the variance in the concentration of species @xmath9 as estimated by the linear - noise approximation, i.e., @xmath114_{0 } - [\epsilon_i]_{0}^2)$]. hence the recipe for calculating the errors of the cfpe predictions is now complete. one first solves eqs. (48 - 50) and then substitutes their solution in eq. (54 - 57). note that this calculation recipe is valid for all times and not only in steady - state conditions. note also that since the denominator in eq. (57) is the linear - noise approximation estimate for the variance then the leading relative error term in the variance is proportional to @xmath39. in contrast the leading relative error term in the mean concentrations, eq. (55), is proportional to @xmath1. hence the cfpe s estimates of mean concentrations are generally expected to be more accurate than those of the variance of the fluctuations about the mean concentrations. finally we obtain the absolute error in the cfpe prediction of the skewness of the probability distribution of the concentration of species @xmath9. the skewness is defined as : @xmath115 the absolute error in the skewness is then @xmath116 where @xmath117 is the skewness predicted by the cfpe, i.e., eq. (58) with subscript @xmath91 throughout. as before, by using using eqs. (21 - 22) together with the summary of results in section iii and eq. (53) we get : @xmath118 we can now answer the question : which of the two, cfpe or linear fokker - planck equation, is the most accurate? we note that the linear fokker - planck equation (or equivalently the linear - noise approximation) is obtained by keeping only terms of order @xmath66 in eq. if we do the same on the expansion of the cfpe, i.e. eq. (36), then we also get the same linear fokker - planck equation. this equality implies that the cfpe becomes correct for large enough volumes or equivalently for large enough molecular populations. this result was known to van kampen and is discussed in the book by gardiner @xcite. within the linear - noise approximation, one can calculate the two quantities @xmath103_{0}$] and @xmath100_{0}$] using eqs. (26) and (23) respectively. the quantities @xmath103_{m}$] and @xmath100_{m}$] where @xmath119 are all zero in this approximation since the expansion has only terms to order @xmath66. now the initial condition for the cme is a delta function centered on the number of molecules as given by the res, i.e. at time @xmath120, the average number of molecules of the cme and the res agree and hence it follows that @xmath103_{0 } = 0 $] initially and for all times @xcite. these results together with eqs. (21) and (22), would seem to imply that within the linear - noise approximation, the mean concentrations are accurate to order @xmath3 while the variance is accurate to order @xmath39. however by considering terms of higher order than those leading to the linear - noise approximation, one arrives at the conclusion that actually the variance within this approximation is accurate to higher order than @xmath39. this can be deduced by noting that @xmath103_{0 } = 0 $] for all times implies @xmath121_{0 } = [\epsilon_r \epsilon_k] _ { 1}= 0 $] also for all times. hence it follows from eq. (22) that the variance in the linear - noise approximation is accurate to order @xmath0. now from eqs. (54) and (56), it is evident that generally the mean concentration and variance prediction of the cfpe are accurate to at least order @xmath0. hence the mean concentration prediction of the cfpe is more accurate than that which can be obtained from the linear fokker - planck equation. it is also clear that the higher accuracy comes from taking into account the non - linear character of the cfpe since the @xmath122 term in eq. (54) is obtained by considering terms in eqs. (14) and (36) of higher order than the linear - noise approximation. we can also derive an explicit equation for the mean concentrations predicted by the cfpe accurate to order @xmath39 : @xmath123_{f,0 } \omega^0 + \partial_t [\epsilon_i]_{f,1 } \omega^{-1/2 }) + o(\omega^{-3/2 }) \nonumber \\ & = \partial_t \phi_i + j_i^w \biggl(\biggl \langle \frac{n_i}{\omega } \biggr \rangle_f - \phi_i \biggr) + \frac{1}{2 } \omega^{-1 } (j_i^{wp } [\epsilon_w \epsilon_p]_{f,0 } - j_i^{w(2) } \phi_w) + o(\omega^{-3/2}).\end{aligned}\]] note that the first step proceeds by taking the time derivative of eq. (21) and the second step follows from using eqs. (26) and (43), bearing in mind that @xmath124_{f,0 } = [\epsilon_i]_{0}$]. hence the computation of the mean concentrations to this order requires only the solution of the res and of the lyapunov equation eq. note that eq. (60) is exactly the same as the effective rate equations recently derived by grima from the cme (eq. 60 is the same as eq. (22) together with eq. (24) in ref @xcite). consider the case where we have @xmath4 species interacting via @xmath5 elementary reactions of the equal - step type, i.e., in each individual reaction, either @xmath125 molecules of a species are generated or @xmath125 molecules are destroyed or no molecules are generated or destroyed. in such a case, the stoichiometric matrix elements are @xmath126 or @xmath127, where @xmath125 is a non - zero positive integer. three examples of equal - step reactions are : @xmath128{k_2 } 2x_1, x_1 \xrightarrow{k4 } \o \nonumber \\ & \o \xrightarrow{k_1 } 2 x_1 \xrightarrow{k_2 } \o \nonumber \\ & \o \xrightleftharpoons[k_2]{k_1 } x_1, \o \xrightleftharpoons[k_4]{k_3 } x_2, x_1+x_2 \xrightarrow{k_5 } \o \end{aligned}\]] the first reaction is autocatalytic where @xmath129 is some very abundant species whose number of molecules is considered constant ; this is a one - step, one species reaction scheme. the second reaction involves the burst input of two molecules and their dimerization, a two - step one species reaction scheme. the third reaction involves the production and degradation of two species and their bimolecular interaction ; this is a one - step, two species reaction scheme. for equal - step reactions, one species reaction schemes, the quantity @xmath130 evaluates to zero in steady - state conditions : @xmath131 note that in the last step, use was made of the steady - state condition : @xmath132. from eqs. (48 - 50), we can then deduce that @xmath133. hence it follows from eqs. (54) and (56) that the mean concentrations and the variance of fluctuations predicted by the cfpe for one species, equal - step reactions, are accurate to at least order @xmath1. this is impressive when one considers that the linear - noise approximation of the cme only leads to estimates accurate to order @xmath3 in the mean and order @xmath0 in the variance. these conclusions lend support to the results of an early investigation of the one species cfpe @xcite. however, this high accuracy of the cfpe is not generally true for multispecies equal - step reactions. for example, for the third reaction scheme in the examples considered above, one finds @xmath134 and @xmath135. the non - zero values of @xmath136 for some index values implies that the mean and variance predictions of the cfpe in this case are accurate to order @xmath0. in the previous subsection we have seen how @xmath137 is zero for one species, one - step reaction schemes and how this leads to a particularly high accuracy in the predictions of the cfpe. we now want to find the condition which forces @xmath138 for chemical reactions involving any number of species. consider the case where all reactions are reversible. since each reaction can be paired with its reverse, it follows that the formula for @xmath137 can then be written as : @xmath139,\end{aligned}\]] where the subscripts @xmath140 and @xmath141 indicates quantities evaluated for the forward and backward reactions respectively. the reversibility condition imposes @xmath142 and was used in deriving the last step. furthermore, a system of reversible reactions will always reach chemical equilibrium and in such conditions the system is characterized by detailed balance, i.e., @xmath143, the forward and reverse rates of each elementary reversible reaction balance @xcite. hence by eq. (63), @xmath138, in detailed balance conditions, and consequently by eqs. (48 - 50) and eqs. (54) and (56), the cfpe s predictions of mean and variance are accurate to order @xmath1. equilibrium conditions always imply detailed balance and hence our results suggest that the size of the differences between the predictions of the cfpe and the cme increase with how far is the system from equilibrium.
Applications
as a first application of our theory, we will estimate the relative errors in the cfpe predictions for a dimerization reaction. this is the simplest case of a bimolecular reaction mechanism. the main purpose of considering such a reaction is that both its cme and cfpe are exactly solvable and hence it provides us with a direct test of our expressions for the leading order error in the means and the variances as predicted by the cfpe. the set of reactions under study are : @xmath144 monomers, denoted as @xmath37, are pumped into some compartment at a rate @xmath145. pairs of monomers react with rate constant @xmath146 to form a dimer molecule, @xmath38. the concentration of dimers increases with time however the concentration of monomers becomes constant after a short time, i.e. the monomers reach a steady - state. since @xmath38 is not involved in the reaction, the mathematical description is solely in terms of the number of molecules of the monomers for the cme and cfpe and in terms of the monomer concentration for the re. the cme, eq. (2), for the dimerization reaction reads : @xmath147 multiplying the equation on both sides by @xmath148 and summing over @xmath149 from @xmath127 to infinity, we get the equivalent generating function equation : @xmath150 where @xmath151. this partial differential equation is solved in the steady - state with boundary conditions @xmath152 and @xmath153 @xcite leading to : @xmath154 where @xmath155, @xmath156 and @xmath157 is the modified bessel function of the first kind of order @xmath158. the mean concentration and variance of the concentration fluctuations about this mean according to the cme are then given by the following expressions : @xmath159 ^ 2 \biggr) \nonumber \\ & = \frac{\phi_1 ^ 2 [n_{ode } [i_1(4 n_{ode})]^2 - n_{ode } [i_0(4 n_{ode})]^2 + i_0(4 n_{ode }) i_1(4 n_{ode })] } { n_{ode } [i_1(4 n_{ode})]^2}.\end{aligned}\]] note that these expressions are obtained within an exact approach and are not approximations as the ones stemming from the system - size expansion of the cme. now we obtain expressions for the mean and variance using the cfpe approach. the cfpe, eq. (33), for the dimerization reaction reads : @xmath160 where @xmath161 and @xmath162. the exact stationary solution of this non - linear second order partial differential equation can be shown to be : @xmath163}{4 k_2 (n_1 - 1) n_1 + k_1 \omega^2 } \biggl (k_1 + k_2 \int_1^{n_1 } d \eta \exp \biggl [- \frac{3 k_1 \omega^2 \arctan h(\eta)}{2 \sqrt{k_2 } \sqrt{k_1 \omega^2 - k_2 } } + \eta \biggr] \biggr),\end{aligned}\]] where @xmath164. the constants @xmath165 and @xmath166 are to be determined by the boundary conditions and the normalization condition. the boundary conditions of the cfpe are @xmath167. note that the cfpe unlike the cme does not generally have a natural boundary at @xmath168 since the noise can sometimes drive the system to negative values of @xmath149 @xcite. note that this problem is also implicit in the stationary solution of the linear fokker - planck equation, a gaussian which is non - zero for negative particle numbers @xcite (see the end of this subsection for a further discussion of boundary conditions). the condition at @xmath169 fixes the value of @xmath166 while the condition at @xmath170 is automatically satisfied by the exponential pre - factor. the remaining constant @xmath165 is fixed by the normalization condition. since there is no closed form solution for the integral in eq. (71), @xmath165 has to be computed numerically ; once @xmath171 is determined, the mean and variance can be straightforwardly numerically computed as well. the exact relative error in the mean and variance predictions of the cfpe can now be computed. one first fixes the rate constants @xmath145 and @xmath146 and @xmath172. the normalization constant @xmath165 is found by numerical integration and from the ensuing steady - state probability distribution, one finds the mean, @xmath173, and variance @xmath174. the numerical error in the integration is essentially eliminated by performing the integration for a set of decreasing step size values and extrapolating to obtain the integral value at zero step size. using the same values of rate constants and @xmath172, one uses eqs. (68 - 69) to compute the mean and variance according to the cme. the exact relative errors in the mean and variance can then be found using @xmath175 and @xmath176, respectively. the exact absolute values of the relative errors in the cfpe predictions are shown by the red open circles in fig. 1 for parameter values @xmath177 and @xmath178. note that the relative error in the variance is larger than that in the mean. the errors increase with decreasing steady - state numbers of monomers. even for very small numbers, the errors are quite small. for example for a case in which the res predict 5 monomers in steady - state, the percentage relative errors in the mean and variance predictions of the cfpe are just @xmath179 and @xmath180 respectively. the high accuracy of the cfpe in low particle number conditions is indeed surprising since typically it has only been deemed accurate for systems characterized by large particle numbers. we can now test the accuracy of the theory developed in the previous sections by using it to obtain expressions for the approximate relative errors in the mean and variance and then compare these with the exact values as already obtained above. by inspection of the reaction scheme, eq. (64), it can be easily deduced that the stoichiometric matrix is @xmath181. from the definition of the macroscopic rate function vector (see introduction) it also follows that it is equal to @xmath182. this is all the information needed to calculate the estimates for the relative errors using our theory. the macroscopic concentration and the relevant entries of the d and j matrices evaluated at steady - state are then given by : @xmath183 these are substituted in eqs. (48 - 50) which are then evaluated at steady - state, leading to : @xmath184 the relative error in the mean concentration to leading order is then given by eq. (55) : @xmath185 where @xmath186 is the average number of monomers as predicted by the res. to compute the relative error in the variance we need to first estimate the variance to the linear - noise level of approximation. this is done by solving eq. (23) in steady - state : @xmath187_0 = -\frac{d_{11}}{2 j_1 ^ 1 } = \frac{3 k_1}{8 k_2 \phi}.\]] the variance is then @xmath188_0 $]. using the latter and eq. (76), it is found that eq. (57) evaluates to : @xmath189 the theoretical absolute values of the relative errors in the cfpe predictions, as given by eq. (78) and eq. (80), are shown by the solid blue lines in fig. 1 for parameter values @xmath177 and @xmath178. the theory is generally in very good agreement with the exact solution ; small discrepancies are only apparent in the error for the variance at molecule numbers less than approximately 5 monomers. the comparison has also been done for many other parameters values and as predicted by theory, in all cases, the graphs are the same as shown in fig. 1. we have also computed the exact errors by solving the cfpe with different boundary conditions. one could argue that constraints should be imposed on the cfpe such that it preserves the natural boundary of the cme at @xmath168. this can be fulfilled by requiring that the probability current of the cfpe vanishes at @xmath168 @xcite. in such a case the stationary solution of the cfpe has the form of eq. (71) with @xmath190 and @xmath165 is found by requiring that the solution is normalized on @xmath191. the exact errors computed with this new solution of the cfpe are practically indistinguishable from the previous solutions shown in fig. 1 except for a small discrepancy at @xmath192. the excellent agreement of our theoretical solution with both cfpe solutions is simply due to the fact that the probability of @xmath149 taking negative values in eq. (71) is very small, unless @xmath172 is also very small. as a second application, we consider the catalysis of a substrate species @xmath193 into a product species @xmath194 by an enzyme species via the michaelis - menten mechanism @xcite : @xmath195{k_0}c, \\ & c \xrightarrow{k_2 } e + p,\end{aligned}\]] where @xmath196 denotes the free enzyme, i.e. when it is not bound to substrate, and @xmath197 denotes the substrate - enzyme complex. we will denote substrate, complex and free enzyme as species 1, 2 and 3 respectively. note that the product species is missing from the kinetic description because it is a byproduct of the reaction and thus not involved in the reactions. the total enzyme concentration is a constant, @xmath198, since the enzyme is either bound to substrate or unbound. hence we effectively have a two variable system. the reaction system exhibits a steady - state in the concentrations of substrate and complex whenever the inequality @xmath199 is satisfied, i.e. when the rate at which substrate is pumped into the system is less than or equal to the maximum rate at which the enzyme can convert substrate to product. assuming such conditions, our aim is to calculate the relative errors in the mean and variance predictions of the cfpe, i.e. eqs (55) and (57) ; to achieve this, we will first need to solve eqs. (48)-(50), which we show in detail now. the stoichiometric matrix and the macroscopic rate function vector follow directly from their definitions (see introduction) : @xmath200 the rate equations and the @xmath201 and @xmath202 matrices follow by inserting the above in eq. (1), eq. (16) and eq. (17) to obtain : @xmath203 where @xmath204, @xmath205 is the michaelis - menten constant and @xmath206. note that @xmath207 is a measure of enzyme saturation since as the input rate of substrate, @xmath208, approaches the maximum rate at which the enzyme can catalyze the reaction, @xmath209, the proportion of enzyme in complex form increases accordingly, as can also be seen from eq. note also that @xmath210 is a measure of how far is the system from equilibrium. this is since if substrate binding would occur at equilibrium, i.e., @xmath211, then the relationship between the macroscopic concentrations would be @xmath212 while generally in steady - state conditions, i.e. @xmath213, the relationship between the macroscopic concentrations is @xmath214. both @xmath207 and @xmath210 are non - dimensional, positive fractions. substituting eqs. (84 - 87) in eqs. (48 - 50), setting the time derivative to zero and solving the resulting set of simultaneous equations we obtain : @xmath215 where @xmath216. the leading order term of the relative errors in the mean concentrations of substrate and complex, as predicted by the cfpe, are then given by substituting eq. (83) together with eq. (89) in eq. (55) : @xmath217 note that the relative error in the substrate concentration is always negative, i.e., the cfpe overestimates the mean substrate concentrations and it increases with the distance from equilibrium, @xmath210. there is no error in the cfpe estimate for enzyme concentration (at least to order @xmath1). to calculate the relative errors in the variance using eq. (57) we first need to compute the variance as estimated by the linear - noise approximation. this is obtained by solving eq. (23) using eq. (84) and eq. (88) : @xmath218 finally substituting the above two equations and eqs. (90 - 91) in eq. (57) we obtain the leading order term of the relative errors in the variance of the substrate and complex concentration fluctuations, as predicted by the cfpe : @xmath219 note that both relative errors are always positive implying that the cfpe underestimates the variance. we can now use the formulae given by eq. (92), eq. (94) and eq. (95) to estimate the relative error of the cfpe when modeling conditions typical of the intracellular environment. a principal characteristic of such an environment is that the number of molecules of some species can be quite small. a detailed protein abundance profiling of the escherichia coli cytosol by ishihama et al @xcite shows that the total number of enzyme molecules per cell approximately varies from a hundred to a few thousands. it is indeed in this limit of small numbers that it is frequently thought that the cfpe and the cle description are not very accurate. we quantitatively test this hypothesis using our formulae. we will first express our error formulae in terms of the average number of molecules of substrate and free enzyme as predicted by the res, i.e., @xmath220 and @xmath221. using eqs (83) we find that : @xmath222 substituting eq. (96) in eq. (92), eq. (94) and eq. (95) we get expressions for the errors in terms of @xmath223, @xmath224, @xmath207 and @xmath210. given fixed molecule numbers, @xmath223 and @xmath224, we can find the maximum error by varying @xmath207 and @xmath210 over their allowed range @xmath225 $]. repeating this procedure for various molecule numbers we can obtain simple two dimensional plots of the maximum error. the results for the maximum relative error in the predictions of the variance are shown in fig. the results verify that the predictions of the cfpe become increasingly accurate with increasing molecule numbers. they also show that the error incurred by using the cfpe for cases of small molecule numbers is very small : less than @xmath226 for a few tens of molecule numbers. it is noteworthy that this accuracy is far better than even that hypothesized by proponents of the cfpe @xcite. for example gillespie in his seminal paper on the derivation of the cle @xcite remarks in his conclusion that the cle (and hence the cfpe) approximation is probably not a good one when one models a system composed of three time - varying species with total molecular population of 2000 since it appears quite possible that the molecule number of at least one of the species becomes significantly small at some point in time. in contrast our theory seems to predict that the cfpe predictions will still be very accurate even when the molecule numbers are quite low. we have tested these predictions by numerically solving the cle for the michaelis - menten process using the euler - mayurama method to obtain the mean substrate concentrations and the variance of the substrate fluctuations about the means. the same were obtained from stochastic simulation algorithm simulations of the cme. the results are shown in fig. the parameters are chosen to be @xmath227, @xmath228, @xmath229, @xmath230, @xmath231 and @xmath232 since this gives conditions similar to those mentioned by gillespie above. the re solutions, eqs. (83), with the above parameters lead to @xmath233, @xmath234 and @xmath235 which, given a volume of @xmath232, would imply @xmath236, @xmath237 and @xmath238. the total molecular population of enzyme (free plus complex form) is 2500 molecules. each algorithm (euler - mayurama and stochastic simulation algorithm) was run 5 times leading to 5 independent estimates @xcite. note that even though the mean number of free enzyme molecules is considerably low, the predictions of the cfpe for both the mean and the variance agree (within sampling error) with those of the cme. for comparison we have also plotted the predictions of the linear - noise approximation (red lines) and of the mean concentration as predicted by the effective mesoscopic rate equation eq. (60) (blue line). the results clearly confirm that the cfpe is more accurate than the linear fokker planck equation associated with the linear - noise approximation and that indeed the mean concentrations of the cfpe are in excellent agreement with the effective mesoscopic rate equations derived in ref @xcite. the effective mesoscopic rate equation for the michaelis - menten reaction was first obtained in ref @xcite (see eq. (29) in the latter reference). for our set of parameters, the theoretical expressions, eqs. (92) and (94), evaluate to @xmath239 and @xmath240 ; these errors are so small that they are clearly masked by the sampling error inherent in the calculation of the mean and the variance from the long - time simulation trajectories. indeed, in agreement with our theory, from fig. 3 one can detect no significant difference between the cfpe and cme predictions. the numerical experiments were performed with various other parameter sets in all cases we could not detect any discrepancy between the cfpe and cme predictions within sampling error.
Discussion and conclusion
summarizing, in this article we have shown that (i) the mean and variance predictions of the cfpe are accurate to order @xmath0. since those of the linear fokker - planck equation are accurate to order @xmath3 for the mean and @xmath0 for the variance, it is clear that the cfpe is generally more accurate than the linear fokker - planck equation or equivalently the linear - noise approximation. (ii) for detailed balance conditions, the predictions of the cfpe are even more accurate, order @xmath1, i.e. in equilibrium or near equilibrium conditions the cfpe does an excellent job of approximating the cme. (iii) accuracy to such high order in inverse powers of the system volume implies that the cfpe estimates should be quite good even for small populations of molecules. our simulations for dimerization and enzyme - catalyzed reactions support these theoretical conclusions, with impressively good agreement down to an average of 5 molecules for the dimerization example. the cfpe s accuracy is indeed surprising given that it arises out of a naive truncation of the kramers - moyal expansion of the cme and that the cfpe can not be obtained from the systematic system - size expansion of the cme. only the linear fokker - planck equation can be derived from the latter expansion by considering terms of order @xmath66. this equation leads to mean and variance estimates which are accurate to orders @xmath3 and @xmath0. now if one wants more accurate estimates one needs to consider higher - order terms in the expansion. to get mean concentration estimates to order @xmath39 one needs to consider the term in the system - size expansion proportional to @xmath3 @xcite. to this order, one does not obtain the cfpe, rather one obtains a partial differential equation with a third - order derivative. however, it turns out that the mean calculated from this equation precisely agrees with that calculated from the cfpe to order @xmath39. if we even wanted to get more accurate means and variance, say both to order @xmath1, we need to consider terms in the system - size expansion to order @xmath0. this leads to a partial differential equation for the time evolution of the probability density function with derivatives as high as fifth order. once again this is not the cfpe. however under steady - state conditions obeying detailed balance, the estimates from this high - order differential equation and the cfpe exactly agree to order @xmath1. hence we have shown that though it is true that the cfpe does not arise out of the system - size expansion, nevertheless its predictions are better than those which can be obtained by considering only the first term of the expansion (the linear - noise approximation) as is conventional @xcite. it follows that the non - linear character of the cfpe is not completely spurious as originally suggested by van kampen @xcite. our study is the first one to our knowledge which systematically analyzes the validity of the non - linear multivariate cfpe and which derives approximate expressions for the size of the errors in the cfpe estimates previous studies @xcite have focused on the cfpe for unimolecular reactions and for unimolecular and bimolecular reactions involving one species @xcite. our analysis is based on the system - size expansion and thus has the same limitations, namely that it is only applicable for chemical systems which are `` asymptotically stable in the sense of lyapunov ''. this implies that from our analysis we can not draw any conclusions for bistable systems @xcite. within these constraints, the system - size expansion is a legitimate means of obtaining the moments of the cme accurate to any desired order @xcite. a few authors @xcite have expressed reservations regarding the accuracy of the expansion beyond the linear - noise level, their reasoning stemming from the fact that pawula s theorem @xcite states that a time - evolution equation for a probability density function with higher than second - order derivatives can not describe a stochastic process. however these misgivings are undue the higher - order partial differential equation stemming from the expansion truncated to some order is `` not an exact equation for a markov process that in some way approximates the original process ; rather it is an approximate equation for the exact p. '' @xcite. this statement of van kampen is generally true for any legitimate expansion of the cme, not only the system - size expansion ; for example risken and vollmer @xcite showed that taking into account higher - order derivatives than two in the kramers - moyal expansion of the cme also leads to more accurate solutions than if one just had to use the cfpe. the accuracy of the system - size expansion beyond the linear - noise approximation has also been verified by many recent studies @xcite, putting at rest any small doubts about its general validity. finally, the good agreement of our theoretical expressions for the errors with simulations is a clear indication of the soundness of our system - size expansion based approach. concluding our results offer theoretical and numerical support for gillespie s hypothesis @xcite regarding the validity of the cfpe in both mesoscopic and macroscopic systems. our formulas provide a simple means to estimate the error in the predictions of the cfpe and the associated cle and hence should be of wide applicability to both theoretical and numerical studies of stochastic chemistry.
Acknowledgments
r. g. acknowledges support by sulsa (scottish universities life science alliance).
Subtleties of the perturbative expansion in the probability density
by the normalization condition and the expansion of @xmath51 we have : @xmath241 equating powers of the volume we obtain : @xmath242 an analogous property has been discussed by gardiner in the different though related context of small noise expansions of the fokker - planck equation @xcite. the two properties above are useful in the computation of the integrals needed to arrive to eqs. (23 - 32) ; for more details see appendix b. it follows from eqs. (a2-a3) that only @xmath243 is a genuine probability density while the higher orders are negative in some regions of the @xmath244 space. from eq. (15), we see that to order @xmath66, the time - evolution of @xmath243 is given by a linear fokker - planck equation : @xmath245 which again verifies that @xmath243 is a probability density. however the time - evolution equations for @xmath246 where @xmath247, involve derivatives of order larger than two and hence by pawula s theorem @xcite @xmath246 can not be genuine probability density functions. the above arguments also imply that it is not correct to think of @xmath248_j$], where @xmath247, as genuine statistical moments ; rather they are best considered as placeholders or labels for the associated integrals @xmath249. in the main text we refer to them as corrections to the moments to order @xmath74. it is however important to bear in mind that though @xmath248_j$] are generally not true statistical moments, their linear superposition via eq. (19) is a genuine statistical moment. hence it is best to avoid associating any physical meaning to @xmath248_j$] and to simply regard them as a means to obtain the desired answer, i.e., @xmath250.
Detailed derivation of the time-evolution equations for @xmath81_2$]
the time - evolution equations are obtained by substituting eq. (18) in eq. (15), multiplying the resulting equation on both sides by @xmath251 and integrating over @xmath73. finally we equate terms of order @xmath39 on both sides of the equation to obtain the time - evolution equation for @xmath81_2 $]. the right hand side of the resulting equation simplifies by performing integration by parts ; there are 8 integrals which need such evaluation and we treat each one of them below. 1. @xmath252 d{\vec{\epsilon } } \nonumber \\ & = - j_{i}^w ([\epsilon_w \epsilon_k] _ 2 \delta_{i, r } + [\epsilon_w \epsilon_r] _ 2 \delta_{i, k }) \nonumber \\ & = - j_{r}^w [\epsilon_w \epsilon_k] _ 2 - j_{k}^w [\epsilon_w \epsilon_r] _ 2.\end{aligned}\]] + note that in eq. (15) we are summing over all twice repeated indices, which for the above integral are @xmath9 and @xmath253. use was made of this implicit summation on @xmath9 in the derivation of the last step. @xmath254 d{\vec{\epsilon } } \nonumber \\ & = d_{ip}(\delta_{p, r } \delta_{i, k } + \delta_{p, k } \delta_{i, r }) \int \pi_2 d{\vec{\epsilon } } = 0. \end{aligned}\]] + in the last step, we have made use of the fact that @xmath255, as shown in appendix a. 3. @xmath256 d{\vec{\epsilon } } \nonumber \\ & = - j_{i}^{wp } ([\epsilon_w \epsilon_p \epsilon_r] _ 1 \delta_{i, k } + [\epsilon_w \epsilon_p \epsilon_k] _ 1 \delta_{i, r }) \nonumber \\ & = - j_{k}^{wp } [\epsilon_w \epsilon_p \epsilon_r] _ 1 - j_{r}^{wp } [\epsilon_w \epsilon_p \epsilon_k] _ 1.\end{aligned}\]] 4. @xmath257 d{\vec{\epsilon } } \nonumber \\ & = - j_{i}^{w(2) } ([\epsilon_r] _ 1 \delta_{i, k } + [\epsilon_k] _ 1 \delta_{i, r }) \nonumber \\ & = - j_{k}^{w(2) } [\epsilon_r] _ 1 - j_{r}^{w(2) } [\epsilon_k]_1.\end{aligned}\]] 5. @xmath258 d{\vec{\epsilon } } \nonumber \\ & = j_{ip}^w [\epsilon_w] _ 1 (\delta_{p, r } \delta_{i, k } + \delta_{p, k } \delta_{i, r }) \nonumber \\ & = 2 j_{kr}^{w } [\epsilon_w] _ 1.\end{aligned}\]] + in obtaining the last step we have used the implicit summation over @xmath9 and @xmath125 and also the symmetrical property, @xmath259, which follows from the definitions given by eqs. (16 - 17). @xmath260 d{\vec{\epsilon } } \nonumber \\ & = - j_{i}^{w(2) } ([\epsilon_w \epsilon_k] _ 0 \delta_{i, r } + [\epsilon_w \epsilon_r] _ 0 \delta_{i, k }) \nonumber \\ & = - j_{r}^{w(2) } [\epsilon_w \epsilon_k] _ 0 - j_{k}^{w(2) } [\epsilon_w \epsilon_r] _ 0.\end{aligned}\]] 7. @xmath261 d{\vec{\epsilon } } \nonumber \\ & = j_{ip}^{wm } [\epsilon_w \epsilon_m] _ 0 (\delta_{p, r } \delta_{i, k } + \delta_{p, k } \delta_{i, r }) \nonumber \\ & = 2 j_{kr}^{wm } [\epsilon_w \epsilon_m] _ 0.\end{aligned}\]] + note that in the last step we have used the symmetrical property, @xmath262, which follows from the definitions given by eqs. (16 - 17). 8. @xmath263 d{\vec{\epsilon } } \nonumber \\ & = j_{ip}^{w(2)}(\delta_{p, r } \delta_{i, k } + \delta_{p, k } \delta_{i, r }) \int \pi_0 d{\vec{\epsilon } } = 2 j_{kr}^{w(2)}. \end{aligned}\]] + in the last step, we have made use of the fact that @xmath264, as shown in appendix a and the symmetry property used in the evaluation of the previous integral. ohphs d. t. gillespie, annu. phys. chem. * 58 *, 35 (2007) d. t. gillespie, physica a * 188 *, 404 (1992) d. t. gillespie, j. chem * 131 *, 164109 (2009) d. a. mcquarrie, j. appl. prob. * 4 *, 413 (1967) n. g. van kampen, _ stochastic processes in physics and chemistry _ (elsevier, 2007) h. kramers, physica * 7 *, 284 (1940) j. e. moyal, j. r. stat. soc. * 11 *, 151 (1949) c. w. gardiner, _ handbook of stochastic methods for physics, chemistry and the natural sciences _ (springer, 2004) n. g. van kampen, can. * 39 *, 551 (1961) n. g. van kampen, adv. * 34 *, 245 (1976) n. g. van kampen, the diffusion approximation for markov process, pp. 181 - 195. in : i. lamprecht and a. i. zotin, ed 1982. _ thermodynamics and kinetics of biological processes_. walter de gruyter and co., new york t. g. kurtz, j. chem. phys. * 50 *, 460 (1969) w. horsthemke and l. brenig, zeitschrift fuer physik b * 27 *, 341 (1977) d. t. gillespie, j. chem. phys. * 113 *, 297 (2000) t. e. turner, s. schnell and k. burrage, comp. biol. chem. * 28 *, 165 (2004) r. grima and s. schnell, essays in biochemistry * 45 *, 41 (2008) j. paulsson, phys. life revs. * 2 *, 157 (2005) t. c. meng, s. somani and p. dhar, in silico biology * 4 *, 293 (2004) d. t. gillespie, j. phys. chem. * 81 *, 2340 (1977) z. hou and h. xin, j. chem. phys. * 119 *, 11508 (2003) m. l. simpson, c. d. cox and g. s. sayler, j. theor. biol. * 229 *, 383 (2004) t. xiao, j. ma, z. hou and h. xin, new. j. phys. * 9 *, 403 (2007) j. wilkie and y. m. wong, chem. phys. * 353 *, 132 (2008) v. sotiropoulos et al, ieee / acm trans.. bioinform. * 6 *, 470 (2009) y. ishihama y. et al, bmc genomics * 9 *, 102 (2008) s. ghaemmaghami et al, nature * 425 *, 737 (2003) j. elf and m. ehrenberg, genome res. * 13 *, 2475 (2003) r. grima, j. chem. phys. * 133 *, 1 (2010) d. t. gillespie, j. chem. phys. * 72 *, 5363 (1980) j. keizer, _ statistical thermodynamics of nonequilibrium processes _ (springer, 1987) r. m. mazo, j. chem. phys. * 62 *, 4244 (1975) h. grabert, p. haenggi and i. oppenheim, physica * 117a *, 300 (1983) h. risken, _ the fokker - planck equation : methods of solution and applications _ (springer, 1996) a. cornish - bowden, _ fundamentals of enzyme kinetics _ (portland press, 1995) the estimates were calculated according to the following procedure. running the cle euler - mayurama algorithm once leads to a set of numbers @xmath265 where @xmath266 is the discretization time step of the algorithm. an estimate of the steady - state average concentration of species @xmath9 as predicted by the cfpe is then given by @xmath267 where @xmath268 is the maximum number of time steps. running the algorithm @xmath4 times leads to @xmath4 slightly different estimates of the mean of the cfpe ; this variation leads to statistical error in the numerical estimation of the mean concentrations and can be reduced by increasing the total number of time steps. the same arguments hold for the estimation of the variance. for the simulations leading to fig. 3 we chose @xmath269 and @xmath270. smaller @xmath266 and larger @xmath268 did not change the results. estimates of the concentrations and variances predicted by the cme can be obtained by sampling the trajectories of the stochastic simulation algorithm at intervals of time equal to the discretization time step of the cle. the means and variance are then obtained as for the cle. r. grima, bmc sys. biol. * 3 * : 101 (2009) d. t. gillespie, j. phys. a * 106 *, 5063 (2002) d. j. higham and r. khanin, open appl. j * 2 *, 59 (2008) r. f. pawula, phys. rev. * 162 *, 186 (1967) h. risken and h. d. vollmer, zeitschrift fuer physik b * 35 *, 313 (1979) r. grima, phys. letts. * 102 *, 218103 (2009) p. thomas, a.v. straube and r. grima, j. chem. * 133 *, 195101 (2010) c. cianci, f di patti, d. fanelli and l. barletti, arxiv:1104.5668v1 c. cianci, f di patti, d. fanelli, arxiv:1104.5570v1 | the chemical fokker - planck equation and the corresponding chemical langevin equation are commonly used approximations of the chemical master equation.
these equations are derived from an uncontrolled, second - order truncation of the kramers - moyal expansion of the chemical master equation and hence their accuracy remains to be clarified.
we use the system - size expansion to show that chemical fokker - planck estimates of the mean concentrations and of the variance of the concentration fluctuations about the mean are accurate to order @xmath0 for reaction systems which do not obey detailed balance and at least accurate to order @xmath1 for systems obeying detailed balance, where @xmath2 is the characteristic size of the system.
hence the chemical fokker - planck equation turns out to be more accurate than the linear - noise approximation of the chemical master equation (the linear fokker - planck equation) which leads to mean concentration estimates accurate to order @xmath3 and variance estimates accurate to order @xmath0.
this higher accuracy is particularly conspicuous for chemical systems realized in small volumes such as biochemical reactions inside cells.
a formula is also obtained for the approximate size of the relative errors in the concentration and variance predictions of the chemical fokker - planck equation, where the relative error is defined as the difference between the predictions of the chemical fokker - planck equation and the master equation divided by the prediction of the master equation. for dimerization and enzyme - catalyzed reactions,
the errors are typically less than few percent even when the steady - state is characterized by merely few tens of molecules. | 1106.4891 |
Introduction
the recent proliferation of smartphones and tablets has been seen as a key enabler for anywhere, anytime wireless communications. the rise of online services, such as facebook and youtube, significantly increases the frequency of users online activities. due to this continuously increasing demand for wireless access, a tremendous amount of data is circulating over today s wireless networks. this increase in demand is straining current cellular systems, thus requiring novel approaches for network design. in order to cope with this wireless capacity crunch, device - to - device (d2d) communication underlaid on cellular systems, has recently emerged as a promising technique that can significantly boost the performance of wireless networks @xcite. in d2d communication, user equipments (ues) transmit data signals to each other over a direct link instead of through the wireless infrastructure, i.e., the cellular network s evolved node bs (enbs). the key idea is to allow direct d2d communication over the licensed band and under the control of the cellular system s operator @xcite. recent studies have shown that the majority of traffic in cellular systems consists of downloading contents such as videos or mobile applications. usually, popular contents, such as certain youtube videos, are requested more frequently than others. as a result, enbs often end up serving different mobile users with the same contents using multiple duplicate transmissions. in this case, following the enb s first transmission of the content, such content is now locally accessible to others in the same area, if ues resource blocks (rbs) can be shared with others. newly arriving users that are within the transmission distance can receive the old " contents directly from those users through d2d communication. here, the enb only serves users that request new " content, which has never been downloaded. through this d2d communication, we can reduce considerable redundant requests to enb, so that the traffic burden of enb can be released. our main contribution is to propose a novel approach to d2d communication, which allows to exploit the social network characteristics so as to reduce the load on the cellular system. to achieve this goal, first, we propose an approach to establish a d2d subnetwork to maintain the data transmission successfully. as a d2d subnetwork is composed by individual users, the connectivity among users can be intermittent. however, the social relations in real world tend to be stable over time. such social ties can be utilized to achieve efficient data transmission in the d2d subnetwork. we name this social relation assisted data transmission wireless network by offline social network (offsn). second, we assess the amount of traffic that can be offloaded, i.e., with which probability can the requested contents be served locally. to analyze this problem, we study the probability that a certain content is selected. this probability is affected by both external (influence from media or friends) and internal (user s own interests) factors. while users interests are difficult to predict, the external influence which is based on users selections can be easily estimated. to this end, we define an online social network (onsn) that encompasses users within the offsn, which reflect users online social ties and influence to each other. in this paper, we adopt practical metrics to establish offsn, and the indian buffet process to model the influence in onsn. then we can get solutions for the previous two problems. latter we will integrate offsn and onsn to carry out the traffic offloading algorithm. further more, in order to evaluate the performance of our algorithm, we will set up the chernoff bound of the number of old contents user selects. to make the analysis more accurate, we also derive the approximated probability mass function (pmf) and cumulative distribution function (cdf) of it. from the numerical results, we show that under certain circumstances, our algorithm can reduce a considerable amount of of enb s traffic. our simulations based on the real traces also proved our analysis for the traffic offloading performance.
System model
consider a cellular network with one enb and multiple users. in this system, two network layers exist over which information is disseminated. the first layer is the onsn, the platform over which users acquire the links of contents from other users. once a link is accessed, the data package of contents must be transmitted to the ue through the actual physical network. taking advantage of the social ties, the offsn represents the physical layer network in which the requested contents of links to transmit. an illustration of this proposed model is shown in fig.[fig : onsn and offsn]. each active user in the onsn corresponds to the ue in the offsn. in the offsn, the first request of the content is served by the enb. subsequent users can thus be served by previous users who hold the content, if they are within the d2d communication distance. information dissemination in both onsn and offsn.,scaledwidth=20.0%] in the area covered by an enb, the density of users in public areas such as office buildings and commercial sites, is much higher than that in other locations such as sideways and open fields. indeed, the majority of the data transmissions occurs in those fixed places. in such high density locations, forming d2d networks as an offsn becomes a natural process. thus, we can distinguish two types of areas : highly dense areas such as office buildings, and white " areas such as open fields. in the former, we assume that d2d networks are formed based on users social relations. while in the latter, due to the low density, the users are served directly by the enb. the offsn is a reflection of the local users social ties. proper metrics need to be adopted to depict the degree of the connections among users. in @xcite, the authors identify that human mobility shows a very high degree of temporal and spatial regularity, and that each individual returns to a few highly frequented locations with a significant probability. thus, such social ties lead to higher probabilities to transmit data among users. the contact duration distribution between two users is assumed to be a continuous distribution, which has a positive value for all real values greater than zero. in addition, users encounter duration usually centers around a mean value. so we can adopt a @xmath0 distribution which is widely used in modeling the call durations @xcite to model the call duration between two users. to find the value for the two parameter @xmath1 and @xmath2, we need the mean and variance of the contact duration. as shown in fig. [fig : encounter history], given the contact duration @xmath3 and the number of encounters @xmath4 between ue @xmath5 and ue @xmath6 corresponding to user @xmath5 and user @xmath6 in the onsn, an estimate of the expected contact duration length @xmath7 and variance @xmath8 can be given by : @xmath9 @xmath10 contact history between ue i and ue j.,scaledwidth=11.6%] given the mean and variance of the contact duration, we can derive the contact duration distribution : @xmath11. thus, the probability density function (pdf) of contact duration will be given by : @xmath12 where @xmath13. then, we can calculate the probability of the contact durations that are qualified for data transmission. if the contact duration is not sufficient to complete a data package transmission, the communication session can not be carried out successfully. we adopt a closeness metric, @xmath14 to represent the probability of establishing a successful communication period between ue @xmath5 and ue @xmath6, which ranges from @xmath15 to @xmath16. the qualified contact duration is the complementary of the disqualified communication duration probability, so @xmath14 can be given by : @xmath17 where @xmath18 is the minimal contact duration required to successfully transmit one content data package, @xmath19 is the lower incomplete gamma function. hence, we can use the closeness metric @xmath14 to describe the communication probability between two ues, which can also be seen as the weight of the link between ue @xmath5 and ue @xmath6. then, a threshold @xmath20 can be defined to filter the boundary between different offsns and white " areas. to cluster users based on metrics such as closeness, we can adopt algorithms such as those used in social networks as in @xcite. then, with a properly chosen @xmath20, each pair in the offsn will have a strong direct neighboring relationship. onsn is the platform for content links to disseminate. we define the number of users in the onsn as @xmath21 which, in turn, corresponds to @xmath21 ues in the offsn. the total number of available contents in the onsn is denoted by @xmath22, @xmath23. given the large volume of content available online, @xmath24. @xmath25 represents the set of contents that have viewing histories and @xmath26 is the set of contents that do not have any. we adopt the indian buffet process (ibp) @xcite model which serves as a powerful tool for getting the content popularity distribution and predicting users selections. the ibp is a stochastic process in which each diner samples from an infinite selection of dishes on offer at a buffet. the first customer will select its preferred dishes according to a poisson distribution with parameter @xmath27. since all dishes are new to this customer, no external information exists so as to influence the selection. however, once the first customer completes the selection, the following customers will have some prior information about those dishes based on the first customer s feedback. customers learn from the previous selections to update their beliefs on the dishes and the probabilities with which they will choose the dishes. the behavior of content selection behavior in onsn is analogous to the dish selection in an ibp. if we view our onsn as an indian buffet, the online contents as dishes, and the users as customers, we can interpret the contents spreading process online by an ibp. so the probability distribution can be implemented from the ibp directly. one realization of indian buffet process.,scaledwidth=19.0%] in fig. [fig : ibp], we show one realization of an ibp. customers are labeled by ascending numbers in a sequence. the shaded block represents the @xmath28th user selected dish @xmath1. in ibp, the first customer selects each dish with equal probability of @xmath29, and ends up with the number of dishes following @xmath30 distribution. for subsequent customers @xmath31, the probability of also having dish @xmath1 already belonging to previous customers is @xmath32, where @xmath33 is the number of customers prior to @xmath28 with dish @xmath1 @xcite. repeating the same argument as the first customer, customer @xmath28 will also have @xmath34 new dishes not tasted by the previous customers following a @xmath35 distribution. the probabilities of selecting certain dishes act as the prior information @xmath36. for old " dishes which have been tasted before, @xmath37. for new" dishes which have not been sampled before, @xmath38. after user @xmath28 completes its selection, @xmath39 will be updated to @xmath40. this learning process is also illustrated in fig. [fig : ibp]. @xmath41 is the number of dishes that have not been sampled before user @xmath28 s selection session.
Proposed traffic offloading algorithm
in the previous section, we have introduced the basic model to formulate the subnetwork for d2d communication and predict users selection. in this section, we can integrate the two layer networks together and propose the traffic offloading algorithm based on the offsn and onsn models. as the inter - offsn interference of d2d communication can be restricted by methods such as power control @xcite@xcite, we can ignore the interference among different offsns. thus, we place an emphasis on the intra - offsn interference due to resource sharing between d2d and cellular communication. during the downlink period of d2d communication, ues will experience interference from other cellular and d2d communications as they share the same subchannels. thus, we can define the transmission rate of users served by the enb as @xmath42, and by d2d communication with co - channel interference as @xmath43, @xcite : @xmath44 @xmath45 where @xmath46, @xmath47 and @xmath48 are the transmit power of enb, d2d transmitter @xmath49 and @xmath50, respectively, @xmath51 is the channel response of the link between ue or enb @xmath5 and ue @xmath6, @xmath52 is the additive white gaussian noise (awgn) at the receivers, and @xmath53 represents the presence of interference from d2d to cellular communication, satisfying @xmath54, otherwise @xmath55. here, @xmath56, so @xmath57 represents the interference from the other d2d pairs that share spectrum resources with pair @xmath49. the transmission rate of the users that are only served by the enb without underlying d2d, thus also without co - channel interference is given by : @xmath58 in the proposed model, even though enb can offload traffic by d2d communication, controlling the switching over cellular and d2d communication causes extra data transmission. thus, there exists a certain cost such as control signals transmission and information feedback during the access process @xcite. therefore, for the enb that is serving a certain user @xmath28, we propose the following utility function : @xmath59r_d+m_n^0 r_c - m_n c_c,\]] where @xmath60 is the overhead cost for controlling the resource allocation process. we propose a novel and robust algorithm that can offload the traffic of enb without any sacrifice on the quality of service. the algorithm consists of multiple stages. in the first stage, the enb collects the encounter history between users to compute the closeness @xmath14. then, based on specific situation such as time and locations, enb chooses @xmath20 dynamically. by checking if @xmath61, i.e., if ue @xmath5 and ue @xmath6 satisfy the predefined closeness threshold, the enb can decide on whether to add this user into the offsn or not. by choosing a proper @xmath20 and power control, the interference among different offsns can be avoided. this process will continue until no more users can be further added to the enb s list. then, the users in the established offsn can construct a communication session with only intra - offsn interference. for websites that provide a portal to access content, such as facebook and youtube, the enb will assign a special tag. once a user visits such tagged websites, the enb will inspect whether the user is located in an offsn or a white " area. if the user is in a white " area, any requests of users will be served by the enb directly. if the user is located in an offsn, the enb will wait the user s future activities. by browsing online, current user can have the prior information @xmath62 of the content distribution in the onsn based on previous user s requests. as soon as the user requests data, the enb detects if there are any resources in the offsn, and then choose to set up d2d communication or not based on the feedback. for old content, the enb will send control signal to the ue @xmath5 that has the highest closeness @xmath63 with user @xmath28. then, ue @xmath28 and ue @xmath5 establish a d2d communication link. even if the d2d communication is setup successfully, the enb still waits until the data transmitting process finishes. if the d2d communication fails, the enb will revert back to serve the user directly. for new contents, the enb serves the user directly. after the selection is complete, the prior updates to the posterior @xmath64. the proposed d2d communication algorithm is summarized in algorithm @xmath16. * offsn and onsn generation * enb collects encounter information in cellular network find the closeness @xmath14 between two ues forming onsn by the users of corresponding ues * 2. user activity detection * * 3. service based on onsn activities *
Performance evaluation
to evaluate the traffic offloading performance of our algorithm, we derive a bound on the amount of traffic that can be offloaded. we show that this problem is equivalent to the amount of contents that have been downloaded. while those locally accessible contents is related to the number of total contents and new contents selected by the users. before we start to derive a closed - form expression on the number of old content, we first try to find its bound. here, we adopt the chernoff bound for analysis. let @xmath65 be a sequence of independent trials with @xmath66, @xmath67, and @xmath68 $]. then : for any @xmath69, there is a bound when @xmath70 $] @xmath71^\mu,\]] and a bound when @xmath72 @xmath73^\mu.\]] for the case of our model, the total number of contents user selects is @xmath74, the number of new contents @xmath75. then, the number of old contents @xmath76. hence, define the expected number of old contents @xmath77, @xmath78, with a chernoff bound of @xmath79^{\frac{n-1}{n}\alpha},\\ \end{split}\]] when @xmath80 $], and @xmath81^{\frac{n-1}{n}\alpha},\\ \end{split}\]] when @xmath82. as we can see, the number of old contents is the difference of two possion distribution which follows the skellam distribution. we can approximate pmf and cdf for the skellam distribution using the saddlepoint approximation. then, we have the approximated pmf and cdf for the number of old contents as follows : @xmath83 @xmath84, \end{split}\]] where @xmath85 and @xmath86. with the cdf function we can get the approximate number of old contents that each user select. thus, we can estimate the traffic that can be offloaded.
Simulation results and analysis
to evaluate the performance of our algorithm, we exploit a data set of sensor mote encounter records and corresponding social network data of a group of participants at university of st andrews by the crawdad team @xcite. in the first data set, they deployed @xmath87 t - mote invent devices over a period of @xmath88 days among @xmath87 users in the department of computer science building. this data set helps us to establish our physical layer offsn. in the second data set, they collected the participants facebook friend lists to generate a social network topology. with those information we can generate the corresponding onsn. then, we adopt the ibp to generate users selections online under the assumption that the size of content library is unbounded. we assume that the content selection process has already been performed for a number of times. thus, the enb can obtain the prior information of the content distribution. 0.3 0.3 0.3 as the d2d communication distance increases, the enb will have more possibilities for detecting available contents providers. as a result, the performance of traffic offloading will be better with a larger maximum distance. this assumption is shown in fig. [fig : traffic and distance]. in this figure, we can see that, increasing the maximum communication distance, yields a decrease in the enb s data rate and an increase in the amount of offloaded traffic. however, we note that, with the increase of the transmission distance, the associated ue costs (e.g., power consumption) will also increase. thus, the increase of d2d communication distance will provide additional benefits to the enb, but not for users. in fig. [fig : traffic and enb cost], we show the variation of the sum - rate at the enb as the cost for control signal varies. as the enb has to arrange the inter change process between cellular and d2d communication, necessary control information is needed. moreover, additional feedback signals are required for monitoring the d2d communication and checking its status. those costs will affect the traffic offloading performance of the system. in our simulation, we define the cost as the counteract to the gain in data rate from 5% to 50%. as we can see, increasing the cost on control signal, the offload traffic amount is decreased. in order to know the amount of traffic that can be offloaded, we specified a special case when @xmath89, then plot the chernoff bound and saddlepoint approximation of the cdf of the @xmath90th user s number of old contents in fig. [fig : bound]. as we can see, the approximated cdf lies between the chernoff bound. we have mentioned in the previous section, the mean value of the number of old contents is @xmath91. so there is a gap between the number of @xmath92 and @xmath93. then we simulate the @xmath90th user s selection and plot the empirical cdf. the simulated empirical cdf also lies between the upper and lower chernoff bound as we expected. in addition, the approximated cdf line is quite close to the simulated empirical cdf line, which proves our analysis in the previous section.
Conclusions
in this paper, we have proposed a novel approach for improving the performance of d2d communication underlaid over a cellular system, by exploiting the social ties and influence among individuals. we formed the offsn to divide the cellular network into several subnetworks for carrying out d2d communication with only intra - offsn interference. also we established the onsn to analyse the offsn users online activities. by modeling the influence among users on contents selection online using the indian buffet process, we have obtained the distribution of contents requests, and thus can get the probabilities of each contents to be requested. using our proposed algorithm, the traffic of enb has been reduced. simulation results based on real traces have shown that different parameters for enb and users will lead to different traffic offloading performances. c. xu, l. song, z. han, d. li, and b. jiao, resource allocation using a reverse iterative combinatorial auction for device - to - device underlay cellular networks, " _ ieee globe communication conference (globecom) _, anaheim, ca, dec. 3 - 7, 2012. n. golrezaei, a. f. molisch, and a. g. dimakis, base - station assisted device - to - device communication for high - throughput wireless video networks, " _ ieee international conference on communications (icc) _, pp. 7077 - 7081, ottawa, canada, jun. 10 - 15, 2012. f. li and j. wu, localcom : a community - based epidemic forwarding scheme in disruption - tolerant networks, " _ proc. ieee conference sensor, mesh and ad hoc communications and networks (secon) _, pp. 574 - 582, rome, italy, jun. 22 - 26, 2009. j. guo, f. liu, and z. zhu estimate the call duration distribution parameters in gsm system based on k - l divergence method, " _ ieee international conference on wireless communications, networking and mobile computing (wicom) _, pp. 2988 - 2991, shanghai, china, sep. 21 - 25, 2007. g. bigwood, d. rehunathan ma. bateman, t. henderson, and s. bhatti, exploiting self - reported social networks for routing in ubiquitous computing environments, " _ proceedings of ieee international conference on wireless and mobile computing, networking and communication (wimob 08) _, pp. 484 - 489, avignon, france, oct. 12 - 14, 2008. | device - to - device (d2d) communication has seen as a major technology to overcome the imminent wireless capacity crunch and to enable new application services. in this paper
, we propose a social - aware approach for optimizing d2d communication by exploiting two layers : the social network and the physical wireless layers.
first we formulate the physical layer d2d network according to users encounter histories.
subsequently, we propose an approach, based on the so - called indian buffet process, so as to model the distribution of contents in users online social networks. given the social relations collected by the evolved node b (enb)
, we jointly optimize the traffic offloading process in d2d communication.
in addition, we give the chernoff bound and approximated cumulative distribution function (cdf) of the offloaded traffic. in the simulation, we proved the effectiveness of the bound and cdf.
the numerical results based on real traces show that the proposed approach offload the traffic of enb s successfully. | 1510.04684 |
Introduction
elliptic curves underpin some of the most advanced mathematics currently being pursued, most notably (probably) the proof of fermat s last theorem by sir andrew wiles. recreational number theory is enjoyed by many people with * no * professional mathematics qualifications. the purpose of the present survey is to show how, even a small knowledge of elliptic curves and a computer, can help solve many problems. these problems are all reasonably simple to understand but can lead to some very interesting mathematics. we start with a simple * problem : * if possible, find rational numbers @xmath0 with @xmath1 where @xmath2 is a specified non - zero integer. the first stage is to analyze the problem. we will find that solutions are related to rational points on the elliptic curve @xmath3 the rational points form a group, often denoted @xmath4, which is finitely generated. this means that every element @xmath5 can be written @xmath6 where @xmath7 are points of infinite order, called generators, and @xmath8 is a point of finite order, called a torsion point. the subset of points of finite order is small and easily determined. for example, the above curve has @xmath9 as elements of @xmath8, together with the point at infinity which is the group identity. the torsion points often correspond to trivial solutions of the problem, and only rarely give a solution we seek. the quantity @xmath10 is called the * rank * of the elliptic curve. it can be zero, so that there are only torsion points, which usually means no non - trivial solutions. we can estimate the rank using the famous birch and swinnerton - dyer (bsd) conjecture and a moderate amount of computing. if the rank is computed to be greater than one, it is often reasonably easy to find at least one generator, though there are exceptions to this rule. the real computational problems are with curves where the rank is estimated to be exactly @xmath11. a by - product of the bsd calculations is an estimate of the height of the point - the higher the height, the more digits in the rational coordinates and so the point is harder to find (in general). there are two possible height normalizations available and i use the one described in silverman @xcite, which gives values half that of pari s current * ellheight * command. let me state at the outset that my primary interest is in computing * actual * numerical values for solutions of the many problems discussed rather than just proving a solution exists. to get anywhere in this subject, you need a computer and some software. to help with the algebra in the analysis of the problems, some form of symbolic algebra package would be very useful. i tend to use an old ms - dos version of derive but this falls down sometimes, so i have been known to use maxima (free) or even mathematica (not - free). more importantly, you need a package which does the number theory calculations on elliptic curves. the most powerful is magma, but this costs money. pari is a free alternative but does not have so many built - in capabilities. the vast bulk of my code is written using pari, since i only recently purchased magma as a retirement present to myself! if you have an old 32-bit machine, i would recommend ubasic for development work as it is very fast. i have tried to give the correct credit where appropriate. if you think i have failed in this task, please contact me. i have also tried to make each problem description self - contained so there is quite a lot of repetition between sub - sections. any mistakes are, of course, purely my fault. i would be extremely pleased to hear from readers with problems i have not covered.
Elliptic curves @xmath12
what is an * elliptic curve *? since i am aiming to make most of this report understandable to non - professionals, i will take a very simplistic approach, which will certainly appal some professionals. but you can get quite far in using elliptic curves, without worrying about topics such as cohomology or galois representations. there are several books that can be recommended on elliptic curves. at an introductory level silverman and tate @xcite is excellent. more advanced are the books by husemller @xcite, knapp @xcite and cassels @xcite. the two volumes by silverman @xcite @xcite have become the standard mathematical introduction. consider the simple cubic @xmath13 which has zeros at @xmath14. these are the only zeros, so the curve is non - negative for @xmath15 and @xmath16, but strictly negative for @xmath17 and @xmath18. make the very simple change of @xmath19 to @xmath20 giving @xmath21 which gives the curve in the following graph. firstly, this curve is symmetric about the x - axis, because if @xmath22 lies on the curve so does @xmath23. secondly, there are no points for @xmath17 or @xmath18, since @xmath24 in these intervals. this means that the curve has two disconnected components. the first for @xmath16 is a closed shape, usually called the `` egg '' - for obvious pictorial reasons. the second for @xmath15 is usually called the infinite component. this is all very nice and pretty, but it does not help us solve our problems. the problems we consider look for integer or rational solutions, and we can link these to finding rational points on the elliptic curve. a rational point is one where both coordinates are rational. there are @xmath25 obvious rational points on this curve, namely @xmath26, @xmath27, and @xmath28. a little simple search shows that @xmath29 gives @xmath30, so @xmath31 and @xmath32 are also rational points on the curve. let @xmath33 and @xmath34. the line joining @xmath35 to @xmath36 has equation @xmath37, and meets the curve where @xmath38 which gives the cubic equation @xmath39 this has at most @xmath25 real solutions, and we already know two of them, @xmath40 and @xmath29. we also know from basic algebra that the product of the three roots is @xmath41. thus the third point of intersection must be @xmath42 which gives @xmath43. so from two rational points we get a third rational point! thinking about this procudure, if @xmath35 and @xmath36 are any rational points, the line joining them will have rational coefficients. so the intersection with the curve will be a cubic with rational coefficients. the product of the three roots will be the negative of the constant in the cubic and so rational. since we know two of the roots are rational, the third must be rational. but the line has rational coefficients, so the y - coordinate must also be rational. so from two rational points we get a third rational point. this operation is thus closed on the set of rational points @xmath4. a closed binary operation is the first requirement for a group. with some further fiddling, we can define an `` addition '' operation on @xmath4, which gives a commutative group. this is discussed further in the appendices. the crucial point is that this group is finitely generated. this means that there are points @xmath44 such that @xmath45 where @xmath46 is any rational point, @xmath47 are integers and @xmath8 is an element of a small subset of @xmath4 which is called the torsion subgroup and which can be easily computed. thus, all we need to do to find rational points is to find @xmath48 and hence use these to find a solution to our original problem. the value of @xmath10 is called the * rank * of the elliptic curve, and it can be @xmath49. for example, @xmath50 has rank zero. the challenge is to compute these generator points. it can be very difficult, but that is what provides the challenge.
Introductory problem
let @xmath42, @xmath51 and @xmath52. then @xmath53 and @xmath54. it is a natural question to ask if this happens for other integers apart from @xmath55. @xmath56 the following lemma shows, however, that we are doomed to failure. * lemma. * the only integer solution is the one above. * we can assume, without loss of generality, that @xmath57 and so, either all of @xmath58 are positive or two of them are negative and one positive. suppose, firstly, that all the variables are positive and we can assume @xmath59. we have @xmath60 and @xmath61 if the right - hand - side is negative then @xmath62, contradicting @xmath63 being an integer. if @xmath64, the only value of @xmath19 giving a positive rhs is @xmath51, but this gives @xmath65. @xmath66 gives @xmath67 and @xmath68. if @xmath42, @xmath69, which only gives integer values when @xmath70, again for @xmath68. if we now assume two of @xmath58 are negative, let them be @xmath71 and @xmath19, with @xmath72. but implies @xmath73, a contradiction. thus, the only purely integer solution is the one at the start of the section. we now consider @xmath74. we have @xmath75, and, substituting, we have that @xmath76 must satisfy the quadratic equation @xmath77 for fixed @xmath2, we can generate rational values of @xmath71, and test if this quadratic factors into linear factors. for @xmath78, the following results were obtained in a simple search .equal sum and product results [cols= " <, >, >, > ",] (8,8) (0,0)(1,0)12 (0,0)(0,1)8 (0,8)(1,0)12 (12,0)(0,1)8 (0,0)(2,1)8 (8,4)(1,-1)4 (0,8)(2,-1)8 (8,4)(1,1)4 finally, for the @xmath79 configuration, consider the rectangle shown. we have @xmath80 with a solution to the @xmath79 configuration when @xmath81 and @xmath82. from @xmath83 and @xmath84 we have the quadratic equation @xmath85 which must have a discriminant being a rational square to give rational @xmath86. thus, there exists @xmath87, such that @xmath88 defining @xmath89 and @xmath90 gives the quartic @xmath91 using mordell s method, this can be transformed to the equivalent elliptic curve @xmath92 with @xmath93 there are, thus, two free parameters @xmath94 and @xmath95. for each possible pair, we try to find curves with rank greater than zero, and possible generators. then generate points on the curve, and values of @xmath96 and solve the quadratic for @xmath86. from these find @xmath97 and test if they are all positive and if @xmath82. no solution has been found, and, in fact, bremner and guy conjecture that a solution does not exist. i must admit that i agree with them. an interesting question is how small can @xmath98 be? it is easy to adapt the code for this. consider the @xmath99 magic square @xmath100 where we assume @xmath101. can we find such a square where the nine entries are distinct squares? so far, only a single square with @xmath102 square entries has been found, by andrew bremner @xcite @xcite. it is @xmath103 on the principle that one solution usually implies more than one, i have spent a lot of computer time looking for more solutions. they are all based on the elliptic curve approach described by bremner, starting from making @xmath55 entries square. to illustrate the approach, suppose we take the first and third rows to all be squares, so @xmath104 and we can, without loss of generality, consider all these variables to be rational. we have @xmath105 and @xmath106 if we think in terms of vectors, the first relation says that the lengths of @xmath107 are the same. thus each vector must just be a rotation of another, so @xmath108 and @xmath109 for some angles @xmath110 and @xmath111. these relations imply that the sines and cosines are rational, so @xmath112 and @xmath113 where @xmath114. these give @xmath115 in terms of @xmath116. substituting into the second relation gives a quadratic @xmath117 where @xmath118 is a fairly complicated rational function in @xmath119. if this quadratic is to give rational values for @xmath120, the discriminant @xmath121 must be a rational square. computing this, and clearing as many square terms as possible, we eventually arrive at the quartic in @xmath95, assuming @xmath96 has been selected, @xmath122 @xmath123 this quartic has a rational solution when @xmath124 and @xmath125. thus the quartic is birationally equivalent to an elliptic curve. after a reasonable amount of computer algebra, we find the relevant curve to be @xmath126 where @xmath127 with @xmath128 and @xmath129 and @xmath130. the inverse transformation is @xmath131 the elliptic curve has @xmath25 finite points of order @xmath132, at @xmath133, @xmath134 and @xmath135. there are also points of order @xmath136 when @xmath137. the torsion subgroup is thus @xmath138 or @xmath139. numerical testing suggests the former is always the case. if we take the denominator of the f - transformation, solve for @xmath140, and substitute into the elliptic curve we find a point @xmath141 which gives a rational point on the curve. adding to @xmath26 gives @xmath142 which has @xmath143, which suggests the curves have rank at least one. as a numerical example, @xmath144, gives @xmath145. the positive value gives @xmath146, but the negative value gives @xmath147. this gives @xmath148 and @xmath149 as possible values. from these, we can compute @xmath0 with @xmath150 @xmath151 with @xmath152 @xmath153 @xmath154 sadly, none of the other @xmath25 elements of the magic square are squares. we can easily perform all these calculations in pari, as well as considering the @xmath155 other distinct configurations of six squares. bremner s solution is found quickly in several configuration calculations, but no second solution has yet been discovered.
Acknowledgement
i would like to express my sincere thanks to andrew bremner for all his help, and his courtesy in answering lots of questions.
Basics of elliptic curves
the standard elliptic curve over @xmath156 can be written @xmath157 with @xmath158. firstly, we transform the equation of the curve to a simpler form. from @xmath159 we have @xmath160 so @xmath161 define @xmath162, so @xmath163 giving @xmath164 so that @xmath165 now, define @xmath166, giving @xmath167 and @xmath168 finally, set @xmath169 and @xmath170 leading to @xmath171 which we write as @xmath172 with @xmath101. this is the form we use for all the elliptic curves in this survey. it has the great advantage of symmetry about the u - axis, so, if @xmath173 is a point on the curve, so is @xmath174. we define addition of rational points in the usual chord - and - tangent way, see silverman and tate s @xcite excellent introduction to elliptic curves. if @xmath175 the line joining the points has equation @xmath176 and so substituting this into the equation gives a cubic in @xmath177 for the points where the line and curve meet. the constant of the cubic is @xmath178 which is rational, and also is @xmath179 where @xmath180 is the u - coordinate of the third intersection of the line and the curve. thus @xmath181 and the corresponding v - coordinate (call it @xmath140) is also rational. thus combining two rational points in this way gives a third rational point. this gives a closed binary operation on the set of rational points, which is often denoted by @xmath4. this is the first requirement for defining a group. to get the associative property needed for a group, we actually need to take the result of the operation to be the point @xmath182 and not the actual point of intersection @xmath183. we define the `` addition '' of rational points as @xmath184 with this binary operation, the set of rational points can be shown to be an abelian group. the identity (denoted by @xmath185) is the point at infinity, which is best approached by projectivising the curve, setting @xmath186 and @xmath187 to give @xmath188 and the point at infinity is @xmath189. this gives the inverse of @xmath173 as @xmath174. @xmath4 is a finitely generated group, and is isomorphic to @xmath190, where @xmath10 is called the rank of the curve and @xmath191 is called the * torsion subgroup * and is one of * @xmath192, * @xmath193. as proven by mazur @xcite, though first conjectured by beppo levi much earlier in the 20th century. this means that there exist @xmath10 rational points @xmath7 such that all rational points can be written @xmath194 where @xmath195 and @xmath196. the hard part in all the problems is finding @xmath197. if @xmath46 is a point of order @xmath132 then @xmath198, so @xmath199. the only way this can happen is if the v - coordinate of @xmath46 is zero, so that the u - coordinate is a zero of @xmath200. we want this root to be rational, but, since the coefficient of @xmath201 is @xmath11, any rational root will actually be an integer root. thus point of order @xmath132 in @xmath4 can only come from integer roots of @xmath202. there is a theorem of nagell and lutz, which states that for curves of the form, all rational torsion points have integer coordinates. there are two important points to note. this result does not always hold for curves of the form, and points of infinite order can have integer coordinates. if @xmath4 has a point of order @xmath132, at @xmath203 say, if we define @xmath204 the elliptic curve reduces to the very simple but important form @xmath205 with @xmath206. ten of the fifteen different torsion subgroup types have points of order @xmath132 so this form is a basic structure. suppose @xmath207 for a curve of this form. the point @xmath208 comes from the intersection of the tangent at @xmath46 with the curve. some simple algebra shows that the w - coordinate of @xmath209 is thus @xmath210 so must be a rational square and, by nagell - lutz, must be an integer square. this is an incredibly useful result.
Transforming quartics to elliptic curves
we now show how to transform the quartic @xmath211 with @xmath212, into an equivalent elliptic curve. we first consider the case when @xmath213, which is very common. if @xmath214 for some rational @xmath215, we substitute @xmath216 and @xmath217, giving @xmath218 thus, suppose @xmath219 we describe the method given by mordell on page @xmath220 of @xcite, with some minor modifications. we first get rid of the cubic term by making the standard substitution @xmath221 giving @xmath222 where @xmath223 we now get rid off the quartic term by defining @xmath224, where @xmath177 is a new variable and @xmath225 is a constant to be determined. this gives the quadratic in @xmath63 @xmath226 if @xmath71 were rational, then @xmath63 would be rational, so the discriminant of this quadratic would have to be a rational square. the discriminant is a cubic in @xmath177. we do not get a term in @xmath227 if we make @xmath228, giving @xmath229 and, if we substitute the formulae for @xmath230, and clear denominators we have @xmath231 with @xmath232 and @xmath233 if we set the denominator of to zero, we have @xmath234 which gives @xmath235 automatically giving a point on the curve. often, it will just be a torsion point, but, occasionally, we get the coordinates of a point of infinite order. note also, that if @xmath236, then @xmath237 so we always have a point of order @xmath132 from @xmath238. in this latter case, if we transfer the origin to this point of order @xmath132 and simplify we get @xmath239 with @xmath240 if @xmath241, we need a rational point @xmath173 lying on the quartic. let @xmath242, so that @xmath243 giving @xmath244 @xmath245 where the coefficient of @xmath246 is, in fact, @xmath247. define, @xmath248, and then @xmath249 giving @xmath250 @xmath251 we now, essentially, complete the square. we can write @xmath252 if we set @xmath253 and @xmath254 this gives @xmath255 and if we define @xmath256 we have @xmath257 multiply both sides by @xmath258, giving @xmath259 which, on defining @xmath260, gives @xmath261 define, @xmath262 and @xmath263 giving @xmath264 which is an elliptic curve. if we define @xmath265, we transform to the form @xmath266 all of the above are easy to program in any symbolic package. a variant of the above problem, that occurs often enough to be important, is when @xmath236, so the quartic is @xmath267 applying the above formulae gives an elliptic curve where the cubic right - hand - side has a linear factor, so the curve has a point of order @xmath132. transforming to make this point the origin, and doing a variable scaling gives @xmath268
Searching for rational points on @xmath269
in this section, we describe a very simple method for searching for rational points on @xmath270 where @xmath271. suppose @xmath272 where we set @xmath273, with @xmath274 and @xmath275, and @xmath276. then @xmath277 it is clear that @xmath278 iff @xmath279, so suppose both @xmath280. now @xmath281 can not divide @xmath282 since, if it did, that would give @xmath283 against @xmath274. thus @xmath284. similarly, we must have @xmath285, so @xmath286 which implies there must exist an integer @xmath287 with @xmath288 and @xmath289. thus @xmath290 and we can always write @xmath2 as @xmath291 where @xmath292 is squarefree. putting @xmath293 and @xmath294 into gives @xmath295 so that @xmath296 and @xmath297. thus @xmath298 and, since @xmath292 is squarefree by definition, @xmath299. thus there must exist an integer @xmath140 such that @xmath300. thus @xmath301 so that @xmath302, and, since @xmath303 we must have @xmath304. so we can determine possible values for @xmath292 from the squarefree factors of @xmath305. if @xmath306, there will be no negative @xmath292 values. now, @xmath307 and we can easily write @xmath308 where @xmath215 is squarefree. define @xmath309, @xmath310 and @xmath311, so that @xmath312 and @xmath313 thus we can look for solutions @xmath314 of and substitute into the right - hand - side of to see if we get a rational square. quite often @xmath315 and we can parameterize @xmath316 as @xmath317, @xmath318 and @xmath319, which makes for a very simple program. if @xmath320, we can find an initial non - trivial solution @xmath321 of and use standard quadric parameterisation to give @xmath322 where @xmath323. this gives @xmath324 this simple method provides lots of points of small heights. we can develop further by taking the numerator and denominator of this equation separately as @xmath325 @xmath326 where @xmath225 is squarefree. we try to satisfy both simultaneously. values of @xmath225 must divide the resultant of the two quadratics on the left - hand - sides of these two relations, so @xmath327 so there are only a computable finite set of possible values. suppose we find an initial solution @xmath328 to the first quadratic, then the line @xmath329 meets it again where @xmath330 substituting into the second quadratic, we must have @xmath331 with @xmath332 @xmath333 @xmath334 @xmath335 @xmath336 finding rational solutions to is a non - trivial problem. there is not enough space to describe the many tricks of the trade that should be employed to do this efficiently. this entire method has been the workhorse for finding hundreds of solutions.
Isogenies of elliptic curves
isogenies are a crucial tool in finding points with large heights by allowing us to consider curves which * might * have rational points with much smaller heights. the classic reference is the paper of vlu @xcite. in this case, points of order @xmath132 have coordinates @xmath338, and other torsion points occur in pairs @xmath339. if @xmath8 denotes the torsion subgroup, let @xmath340 be the subset consisting of all points of order @xmath132 together with * one * point from each of the pairs @xmath341. consider, first, the curve @xmath351 with just the torsion point @xmath26. then @xmath352, @xmath353 and @xmath354. thus equation is @xmath355 and setting @xmath356 gives the standard 2-isogenous curve @xmath357 suppose we find a point on the 2-isogenous curve @xmath359. solving the left - hand equation for @xmath360 will usually give a non - rational value for @xmath71. if, however, we form @xmath209 we find it has h - coordinate @xmath361 putting this into the left - hand equation we find @xmath362. for a curve of the form @xmath370, a point of inflexion would occur when @xmath371. this gives @xmath372 but there is not a point of inflexion at @xmath373 since @xmath95 is negative at this point. we can, however, get a simpler form for the 3-isogenous curve by using this point. | several problems which could be thought of as belonging to recreational mathematics are described.
they are all such that solutions to the problem depend on finding rational points on elliptic curves.
many of the problems considered lead to the search for points of very large height on the curves, which (as yet) have not been found. | 1610.03430 |
Introduction
in the current studies of ultracold quantum gases, a great deal of interest has been drawn to the study of bose - einstein condensates (becs) loaded into optical lattices (ols), i.e., spatially periodic potentials induced by the interference between counterpropagating laser beams book1,book2,morsch06. besides playing a crucial role in effectively tuning the interaction strength in the condensate, i.e., the ratio between the kinetic and interaction energies @xcite, ols offer an extremely useful tool for studies of the transition between the superfluid and mott - insulator states @xcite, and for the investigation of effects in the matter - wave dynamics due to the interplay between nonlinearity and the quasi - discreteness, which is induced by a deep lattice potential @xcite. the mean - field dynamics of the bec loaded into the ol is described by the cubic gross - pitaevskii equation (gpe) with the periodic potential book1,book2,morsch06. the respective bogoliubov s excitation spectrum features a band structure, similar to electronic bloch bands in solid state. if the ol potential is deep enough, the lowest - band dynamics may be approximated by the discrete nonlinear schrdinger (nls) equation trombettoni01. using this correspondence, the bec dynamics was studied in the framework of the nonlinear - lattice theory, see works trombettoni01,abdullaev01,alfimov02 and short review @xcite. the presence of the ol gives raise to energetic and dynamical instabilities, which have been predicted theoretically wu01,smerzi02,konotop02,wu03,menotti03,taylor03,kramer03,kramer05 and studied experimentally @xcite. an important application of the ols is their use for the creation and stabilization of matter - wave solitons. in particular, the periodic potential gives rise to localized gap solitons in the case of _ repulsive _ two - body interactions, as was predicted theoretically @xcite and demonstrated experimentally @xcite (with the attractive interactions, bright matter - wave solitons were created and observed in condensates of @xmath0li @xcite and @xmath1rb weiman atoms). more generally, the use of time- and space - modulated fields acting on atoms is a powerful tool for the control of soliton properties @xcite ; for instance, while the gpe without external potentials admits stable soliton solutions only in the 1d geometry sulem99,ablowitz04, ol potentials can stabilize solitons in any higher dimension @xcite. unlike 1d solitons, a necessary existence condition for their multidimensional counterparts, stabilized by means of ols, is that the soliton s norm must exceed a certain threshold value. another very useful tool frequently used in experiments with ultracold atomic gases is the control of the strength and sign of two - body interactions by means of an external magnetic field near the feshbach resonance @xcite. further, recent works proposed to exploit the possibility to control the strength of _ three - body _ interactions between atoms, independently from the control of the two - body collisions @xcite. one motivation for such studies is related to the possibility of creating new exotic strongly correlated phases in ultracold gases. indeed, quantum phases, such as topological ones or spin liquids, turn out to be ground - states of the hamiltonian including three- or multi - body - interaction terms, an example being fractional quantum - hall states described by pfaffian wave functions @xcite. in a recent work @xcite, a 1d bose gas with @xmath2-body attractive interactions was studied in the mean - field approximation, with the objective to create highly degenerate ground - states of hamiltonians including many - body terms. for the three - body interactions (@xmath3), the system is described by a _ gpe, i.e., the respective term in the energy density is proportional to @xmath4, where @xmath5 is the single - atom mean - field wave function (in the general case, a similar term is proportional to @xmath6). soliton solutions can be found for each @xmath2, but they represent the stable ground - state, with negative energy (which is defined as per eqs. ([e - functional]) and ([h0]), see below), only for @xmath7, being unstable excited states with positive energy at @xmath8. for @xmath3, soliton solutions are 1d counterparts of the well - known townes solitons @xcite, which play the role of the separatrix between collapsing and decaying localized states. the townes - like solitons with fixed norm (which is @xmath9, in the notation adopted below) exist only at a single critical value of the interaction strength, at which they feature the infinite degeneracy fersino08 : _ all _ the solitonic wave functions, @xmath10 ^{-1/2}$], with arbitrary width @xmath11 (see eqs. ([critical - value]) and (quintic-1) below), have _ zero energy _ but different values of the chemical potential, @xmath12. a relevant issue is how this infinite degeneracy is lifted by an external potential, especially by a periodic one corresponding to the ol @xcite. when the two - body interaction is present, the mean - field equation is the gpe with the cubic - quintic (cq) nonlinearity @xcite. as said above, it has been shown @xcite that it is possible to tune the strength of the two - body interactions independently from the three - body ones. in addition to that, in the framework of the effective gpe for the bec loaded into a nearly 1d ( cigar - shaped ") trap with tight transverse confinement, an effective attractive quintic term appears, in the absence of any three - atom interactions, as a manifestation of the residual deviation from the one - dimensionality @xcite. in any case, if the two - body interaction is repulsive while its three - body counterpart is attractive, soliton solutions to the cq gpe can be found in an exact analytical form (in the free space), but they feature an unstable eigenvalue in the bogoliubov - de gennes spectrum of small perturbations around them @xcite, while the instability of the townes - like solitons in the quintic equation is subexponential, being accounted for by a zero eigenvalue. the issue we address in this paper is the possibility to stabilize such solitons by means of the ol potential. previously, the stabilization of originally unstable solitons by means of the ol was considered, in the 2d @xcite and 1d @xcite settings alike, only for localized states of the townes type (recently, the stabilization of 2d solitons against the _ supercritical collapse _ by the ol was also demonstrated in the cq model in 2d, with both cubic and quintic terms being attractive @xcite). it was found that the ol with any value of its strength (i.e., with zero threshold) opens a _ stability window _ around the critical point corresponding to the townes solitons. in this work, we demonstrate that the ol opens a stability window for solitons in the cq model (with the repulsive cubic and attractive quintic terms) too, but only if the lattice strength exceeds a _ finite _ threshold value. apart from the context of bec, where the nonlinearity degree is related to the number of atoms simultaneously involved in the contact interaction, nls equations with the power - law and cq nonlinearities are also known as spatial - domain models of the light propagation in self - focusing media kivshar03 (for a brief overview of optical models based on the cq - nls equation, including references to experimental realizations, see recent works @xcite). in the case of the cubic nonlinearity (the kerr medium), effects of imprinted lattices on the transmission of light beams have been investigated both in local @xcite and nonlocal @xcite models. the paper is structured as follows. in section ii, we introduce the cq gpe corresponding to the mean - field description of the 1d bose gas with two - body repulsive and three - body attractive interactions. properties of the (unstable) soliton solutions to this equation are also recapitulated in section ii. in section iii, we use the variational approximation (va) (see ref. @xcite for a review) to discuss effects of the ol on the solitons. we introduce an appropriate ansatz and compute the corresponding energy. the limit of the vanishing two - body interaction is considered too and compared to previous results @xcite. in section iv, the stability region for the soliton solution in the presence of the repulsive two - body interaction and ol is determined and compared with numerical findings. the effect of an additional harmonic - trap potential is studied in section v, showing that the stability region depends on the matching between minima of the periodic potential and the location of the minimum of the harmonic trap. in section vi we present our conclusions.
The model
the quantum many - body hamiltonian for the 1d bose gas with @xmath2-body contact attractive interactions is @xmath13 ^{% \mathcal{n}}\left [\hat{\psi}(x)\right] ^{\mathcal{n}}\right\ }, \label{ham - contact}\]]where @xmath14 is the bosonic - field operator, @xmath15 is the nonlinearity strength and @xmath16is the single - particle hamiltonian, @xmath17 being the external potential. the case of @xmath7 in the homogeneous limit (@xmath18) corresponds to the integrable lieb - liniger model @xcite. for attractive interactions (@xmath15), its analytical solution was obtained by means of the bethe ansatz @xcite and for a large number of particles, @xmath19, the energy of the exact ground - state solution coincides with that obtained in the mean - field approximation calogero75. in the attractive lieb - liniger model, a finite ground - state energy per particle is provided by fixing product @xmath20 to a constant value @xcite, while for @xmath21 one has to set @xmath22 @xcite. in the heisenberg representation, the equation of motion for field @xmath23 is @xmath24 = \hat{h}_{0}\hat{\psi}-c\left (\hat{\psi}^{\dag } \right) ^{\mathcal{n% } -1}\left (\hat{\psi}\right) ^{\mathcal{n}-1}\hat{\psi}. \label{dyn}\]]the mean - field approximation reduces eq. ([dyn]) to the corresponding gpe with the power - law nonlinearity, @xmath25where the macroscopic wave function @xmath26 is normalized to the total number of atoms, @xmath19, and the nonlinearity degree is related to the order of the multi - body interactions, @xmath2 : @xmath27thus, the usual two - body interaction (@xmath7) corresponds to @xmath28, and the three - body interaction (@xmath3) to @xmath29. equation ([gnls]) conserves the energy, @xmath30 \psi (x), \label{e - functional}\]]which is the classical counterpart of quantum hamiltonian ([ham - contact]). in eq. ([h0]), @xmath17 is the external trapping potential, which typically includes a superposition of an harmonic magnetic trap and periodic ol potential, @xmath31, where the harmonic confining term is @xmath32. we take the periodic potential as @xmath33, where @xmath34 is proportional to the power of the laser beams which build the ol, and @xmath35, with @xmath36 ; here, @xmath37 is the wavelength of the beams, and @xmath38 the angle between them (the period of the lattice is @xmath39). parameter @xmath40 measures a mismatch between the minimum of the parabolic potential (at @xmath41) and the closest local minimum of the lattice potential : when @xmath42 (@xmath43) a minimum (maximum) of @xmath44 coincides with the minimum of @xmath45. in fact, except for section v, we consider the situation without the parabolic trap (i.e., @xmath46), therefore we set @xmath42 in this case. the time - independent power - law gpe corresponding to eq. ([gnls]) is (from now on, we use normalized units, with @xmath47 and @xmath48) @xmath49 \psi (x)=\mu \psi (x), \label{gnls - mu}\]]where @xmath50 is the chemical potential, @xmath51 and the norm of the wave function is @xmath9. in the free - space cubic model (@xmath28 and @xmath52), eq. ([gnls - mu]) is the integrable nls equation, whose multi - soliton solutions can be obtained by means of the inverse scattering method @xcite. the commonly known single - soliton nls solution is @xmath53where @xmath54 and @xmath55 is a real amplitude, the respective value of the chemical potential being @xmath56. for a general value of @xmath57, the integrability is lost even in the absence of the external potential @xcite ; nevertheless, the respective single - soliton solutions can be found in an explicit form ablowitz81,polyanin04. for the attractive three - body interactions (@xmath29), eq. ([gnls]) is the self - focusing quintic gpe, whose stationary version is @xmath58 \psi (x)=\mu \psi (x). \label{quintic - mu}\]]for @xmath59, if one fixes coefficient @xmath60 in front of the interaction term, the townes - like solitons exist for a particular value of the norm of the wave function @xcite. on the other hand, fixing the normalization of the wave function (recall that the norm is @xmath9 in our units) amounts, for @xmath61, to fixing a relation among the chemical potential and the interaction strength @xcite, so that for each @xmath60 it is possible to obtain a single soliton solution (although, as mentioned above, these solutions provide the ground - state in the infinite system only for @xmath62, i.e., for @xmath63). however, for @xmath29 (i.e., @xmath3) chemical potential @xmath50 remains indefinite, assuming arbitrary negative values, while the soliton solution of the form @xmath64satisfies the unitary normalization condition at a single (critical) value of the interaction strength @xcite, @xmath65at @xmath66, all solutions ([quintic-1]) share a common value of the energy, which is simply @xmath67 @xcite, as follows from eqs. ([e - functional]) and ([critical - value]). if the two - body interaction is added to the three - body attraction, the mean - field equation is the gpe with the cq nonlinearity, @xmath68 \psi (x)=\mu \psi (x). \label{cubic - quintic - mu}\]]as said above, we chiefly focus on the case of the _ repulsive _ two - body interactions, i.e., @xmath69. a family of exact soliton solutions to eq. ([cubic - quintic - mu]) with @xmath59 can be obtained in the exact form @xcite, which, for @xmath69, is @xmath70where @xmath71, and the maximum value of the density, at the soliton s center, is @xmath72a simple derivation of eq. ([cubic - quintic-1]) is presented in appendix a. obviously, for @xmath73 solution ([cubic - quintic-1]) reduces to townes - like soliton ([quintic-1]). imposing the above - mentioned normalization,@xmath74on solution ([cubic - quintic-1]), one arrives at relation@xmath75from where it follows that, for @xmath76, soliton solutions with @xmath77 satisfying normalization condition ([1]) exist for @xmath78. however, these solutions are unstable @xcite (in particular, because they do not satisfy the vakhitov - kolokolov stability criterion kolokolov73). in the following section we discuss how the ol can stabilize such localized solutions.
Variational approximation
both for @xmath79 and @xmath80 (@xmath7 and @xmath81), and for the gpe with the mixed cq nonlinearity, the presence of the periodic potential makes it necessary to resort to approximate methods for finding solitons. to this end, we use the va (variational approximation) @xcite based on the _ ansatz _ which yields exact soliton solution (quintic-1) of the quintic nls equation in the absence of the external potential : @xmath82here, width @xmath11 is the variational parameter to be determined by the minimization of the energy, while amplitude @xmath55 will be found from normalization condition ([1]). we expect that ansatz ([var]), which does not explicitly include the modulation of the wave function induced by the ol, may give a reasonable estimate of the soliton s energy for sufficiently small values of ol strength @xmath34 in eq. ([ol]), cf. the known result for the 2d equation with the cubic nonlinearity (@xmath79) and ol potential @xcite. in the case of the 3d gpe which includes the cubic term and harmonic trap, this approach leads to an estimate for the critical value of the number of atoms above which the condensate collapses, that was found to be in a reasonable agreement with results produced by the numerical solution of the gpe fetter95,ruprecht95. in 1d, the va based on the gaussian ansatz also provides for quite an accurate approximation to exact soliton solution ([cubic-1] @xcite. similar analyses carried out in the 1d model including the cubic term and ol @xcite have demonstrated that (unlike the 2d and 3d cases) the 1d soliton trapped in the ol potential does not have an existence threshold in terms of its norm (number of atoms). the energy to be minimized in the framework of the va is obtained by inserting ansatz ([var]) in the gpe energy functional given by eq. (e - functional). the kinetic and quintic - interaction energy terms in the functional both scale as @xmath83 ; then, the energy per particle computed from expression ([e - functional]) is @xmath84, \label{e3}\]]@xmath85where @xmath86 is defined in eq. ([critical - value]). for @xmath52 (without the ol), the scenario discussed in the previous section for the uniform cq gpe with the attractive three - body and repulsive two - body interactions is recovered, as energy ([e3]) reduces in that case to @xmath87for @xmath73, the energy is positive when @xmath88 (i.e., @xmath89, see eq. ([beta])) and vanishes at @xmath90 ; for @xmath91 (i.e., @xmath92) one obtains @xmath67, in agreement with the above - mentioned exact result showing the infinite degeneracy of soliton family ([quintic-1]), while for @xmath93 the energy is negative and diverges (to @xmath94) at @xmath95, signaling, in terms of the va, the onset of the collapse. with @xmath76, expression ([e3-homog]) does not give rise to any minimum of the energy, which agrees with the known fact of the instability of all the solitons in this case @xcite. a detailed study of minima of variational energy ([e3]) is presented in appendix b. in the following subsection, we consider the case of the self - focusing quintic gpe in the presence of the ol (@xmath96), while the discussion of the general case (@xmath97) is given in section iv. here we address the stability of localized variational mode ([var]), for different values the ol parameters, strength @xmath34 and wavenumber @xmath98, keeping @xmath73. the results of the analysis of minima of the variational energy ([e3]), presented in appendix b, can be summarized as follows (see also fig. [fig1]) : for @xmath99, the infinitely deep minimum of the energy is obtained at @xmath95, which corresponds to the collapse, as shown in fig. [fig1](a). for @xmath100, the collapse may be avoided, and three possibilities arise : there exists another special value, @xmath101, such that for every @xmath60 between @xmath102 and @xmath86 the energy has a minimum at @xmath103 and a maximum at @xmath104, while for @xmath105 the energy does not have a minimum at any finite value of @xmath11, see fig. [fig1](d). actually, two different situations should be distinguished for @xmath106 : there exists a specific value (refer to appendix b), @xmath107(with @xmath108) such that, for @xmath109, the energy has a _ global _ minimum at @xmath110 (which, thus, represents the _ ground - state _ of the boson gas in this situation), while, for @xmath111, the energy minimum at @xmath110 is a _ local _ one. in other words, taking into regard the fact that, as shown by eq. ([e3]), the energy - per - particle approaches value @xmath112 at large @xmath11, we conclude that, for @xmath113 (@xmath111), the energy satisfies inequality @xmath114 (@xmath115), as showed in figs. [fig1](b, c). obtained in the framework of the quintic gpe versus @xmath116 (in units of @xmath117) for @xmath118 (a) ; @xmath119 (b) ; @xmath120 (c) ; @xmath121 (d). in (a) the solid (dotted) line is the energy for @xmath93 (@xmath66) ; in (b)-(c), points of the energy minimum and maximum, @xmath122 and @xmath123, are indicated.] from the above analysis, we infer that for @xmath124 the ground - state is a delocalized one (although the metastable state, corresponding to the above - mentioned local energy minimum, exists for @xmath111), for @xmath119 the ground - state is represented by a finite - size soliton configuration (in agreement with ref. @xcite) and for @xmath93 it is collapsing. equation ([c_2_star]) shows that the width of the stability region depends on ratio @xmath125 : keeping fixed all other parameters, the decrease of the lattice spacing (i.e., the increase of @xmath98) leads to a reduction of the stability region. equation ([c_2_star]) also shows that for @xmath126 the va formally predicts @xmath127 : however, for @xmath128, the ground - state is delocalized and the variational ansatz ([var]) can not be used, as it does not take into account the modulation induced by the deep ol potential. in fig. [deloc-1], we plot the numerically found ground - state of the quintic gpe in a 1d box (@xmath129). it is seen that, with the increase of @xmath130, the configuration becomes broader, until a critical value is reached, as discussed in @xcite. in the inset of fig. [deloc-1] we plot the squared width @xmath131 of the numerically found ground - state @xmath132 versus @xmath60, which makes the delocalization transition evident : for @xmath124 the width @xmath133 is @xmath134, while around @xmath135 the width suddenly decreases. variational estimate ([c_2_star]) for the critical value @xmath136, as predicted by the va (see eq. ([c_2_star])), is displayed in fig. [comparison_quintic], together with numerical results. one observes a reasonable agreement between them, especially for small @xmath34, which is due both to the use of the more adequate ansatz ([var]), rather than a gaussian, and also because @xmath137 is found as the value at which the global (rather than local) minimum disappears. . solid lines, starting from the narrowest configuration, refer to @xmath138 (recall that @xmath139), and the dashed line refers to @xmath140. parameters are @xmath141, @xmath142 and @xmath143. inset : squared width @xmath144 of the ground - state as a function of @xmath60 (the dot - dashed line is a guide to the eye). critical value @xmath145 obtained from the numerical analysis is @xmath146, which should be compared with the corresponding value ([c_2_star]) predicted by the variational approximation, @xmath147.] as a function of @xmath148, according eq. ([c_2_star]) (for the quintic gpe), the dashed line corresponding to @xmath149. discrete symbols represent results obtained from the numerical solution of the quintic gpe. they designate the transition form the localized ground - state to the extended one (parameters are the same as in fig. ([deloc-1]). according to the variational approximation, the ground - state is delocalized (@xmath150) below the dotted line, and it collapses (@xmath151) for @xmath60 above the dashed line.]
The stability region for the condensate with competing two- and three-body interactions
the most interesting situation occurs when the two - body repulsive interaction (@xmath76) competes with the attractive three - body collisions (@xmath15). as said above, all solitons in the free space (@xmath52) are strongly unstable in this situation @xcite, and the possibility of their stabilization by the ol was not studied before. the analysis of variational energy ([e3]), presented in appendix b, yields the following results for this case. for @xmath93, the energy does not have a minimum at finite @xmath11, hence the ol can not stabilize the solitons in this case. if @xmath66, the energy has a global minimum at a finite value of @xmath11, when @xmath152for @xmath88, the energy features a global minimum at finite @xmath11 for @xmath153, where the modified critical value is @xmath154cf. definition ([c_2_star]) for @xmath155. the value @xmath137 depends upon @xmath156, vanishing for @xmath156 larger than the critical value @xmath157. this means that, to balance the destabilizing effect of the repulsive two - body interactions, the strength of the periodic potential, @xmath34, must _ exceed _ its own critical value, @xmath158otherwise, eq. ([c_2_star_g]) yields @xmath159, i.e., the ol can not stabilize the solitons. in fig. [deloc-2] we plot the numerically found ground - state of cq gpe ([cubic - quintic - mu]) for several values of @xmath34. it is seen that, at small @xmath34, the wave function @xmath5 remains delocalized, until a critical value is reached. in the inset of fig. [deloc-2] the squared width of the numerically generated ground - state is plotted versus @xmath34. in fig. [comparison_epsilon], we compare critical value @xmath160, as given by eq. ([epsilon_critical]), with numerical results : for small @xmath161, the predicted linear dependence of @xmath160 on @xmath161 is well corroborated by the numerical results, the relative error in the slope being @xmath162. in principle, the comparison between variational estimate ([epsilon_critical]) and numerical results might be further improved by choosing a variational wave function which, in the limit of @xmath52 (uniform space) would reproduce exact cq soliton ([cubic - quintic-1]). however, the calculations with such an ansatz are extremely cumbersome. . solid lines, starting from the narrowest wave function, refer to @xmath163, and the dashed line refers to @xmath164. parameters are @xmath165, @xmath166, @xmath142, @xmath167. inset : the squared width of the ground - state versus @xmath168 (the dot - dashed line is a guide to the eye). critical value @xmath169 obtained from the numerical data is @xmath170, which should be compared to the variational prediction given by eq. ([epsilon_critical]), which is @xmath171.] versus @xmath172 as given by eq. ([epsilon_critical]). symbols refer to results obtained from the numerical solution for the ground - state of the cubic - quintic gpe. they represent the delocalization transition. the parameters are the same as in fig. ([deloc-2]).]
The effect of the harmonic trap
in this section we aim to use the variational approximation based on ansatz ([var]) for examining the combined effect of the parabolic trapping potential acting along with an ol, i.e., we take eq. ([quintic - mu]) with external potential@xmath173cf. ([ol]), and disregard binary collisions (@xmath73). value @xmath174 (@xmath175) corresponds to the matching (largest mismatch) between the minimum of the harmonic potential and a local minimum of the lattice potential. the respective variational energy is obtained from (e - functional) with potential ([full - v]) : @xmath176. \label{e3-tot}\]] with @xmath177, the soliton is stable for @xmath88, and it collapses otherwise. with @xmath178, a richer behavior is predicted by the va. the system does stabilize for @xmath88, while, for @xmath93, the presence of the mismatched harmonic trap gives rise to a metastability region. since @xmath179 as @xmath95 and @xmath180 as @xmath90, one can encounter two possibilities : either @xmath181 is positive for all @xmath11 (and there are no energy minima), or equation @xmath182 has two roots, corresponding to a local minimum and a maximum. the equation for the value of @xmath11 at which energy ([e3-tot]) reaches the local minimum is @xmath183where @xmath184, and @xmath185@xmath186one can see that, for @xmath66 (i.e., @xmath92), eq. ([cond - trap]) does not have a nonvanishing solution if @xmath98 is smaller than a critical value, @xmath187while it has a nonvanishing solution for @xmath188. actually, for @xmath93 (i.e., @xmath189), eq. ([cond - trap]) with @xmath190 has two nonvanishing roots, one of which is a local minimum, while such roots do not exist for @xmath191. for @xmath188, the right - hand side of eq. ([cond - trap]) has a maximum value, which fixes the maximum value of @xmath192, i.e., the maximum value of @xmath60, which we refer to as @xmath193. then, for @xmath194, the variational energy does not have a local minimum. for @xmath195 there appears a finite metastability region, in terms of wavenumber @xmath98, as illustrated by fig. [q]. in other words, for fixed @xmath60, metastable states appear at large values of @xmath34. plane of the model (including the parabolic trap) the metastable region from the unstable one. the parameters are @xmath196, @xmath197, and @xmath198.]
Conclusions
in this work we have studied the effect of the ol (optical lattice) on the 1d bose gas with attractive three - body and repulsive two - body interactions, described by the gpe (gross - pitaevskii equation) with the cq (cubic - quintic) nonlinearity. actually, the effective quintic attractive term in the gpe may be induced by the residual deviation of the condensate, tightly trapped in a cigar - shaped confining potential, from the one - dimensionality (when the three body losses are negligible) @xcite or by three - body interaction terms between atoms according to recent proposals @xcite. in the absence of an external potential, soliton solutions to this equation with the cq nonlinearity are known in the exact form, but they all are strongly unstable. we have demonstrated that the ol opens a stability window for the solitons, provided that the ol strength, @xmath34, exceeds a finite minimum value. the size of the stability window depends on @xmath199, where @xmath98 is the ol s wavenumber. we have also considered effects of the additional harmonic trap, finding that, if the quintic nonlinearity is strong enough (@xmath99), a metastability region may arise, depending on the mismatch between minima of the periodic potential and harmonic trap.
Localized solutions of the cubic-quintic gross-pitaevskii equation
assuming that @xmath200 is real, we look for localized solutions to the cq nls equation, @xmath201with @xmath15 and @xmath69. interpreting @xmath202 as a formal time variable and @xmath203 as the coordinate of a particle, eq. ([cubic - quintic - mu - app]) formally corresponds to the newton s equation of motion of this particle, @xmath204where the effective mass is @xmath205, and the potential is @xmath206with an arbitrary additive constant chosen so as to have @xmath207. potential ([potential - app]) for @xmath77, which corresponds to normalizable solutions, is plotted in fig. [potential]. condition @xmath208 yields expression ([a - cubic - quintic]) for the soliton s amplitude. for @xmath209.] further, we make use of the conservation of the corresponding hamiltonian, @xmath210the boundary conditions for localized solutions, @xmath211, @xmath212, select @xmath213 in eq. ([energy - app]). taking into regard the fact that @xmath214 for @xmath215, and looking for solutions with @xmath216 at @xmath217, one obtains from here the soliton solution in an implicit form, @xmath218it further follows from eq. ([quadr - app-1]) that@xmath219with @xmath220. in eq. ([quadr - app-2]), we use notation @xmath221 and @xmath222. thus, from eq. (quadr - app-2) one obtains @xmath223 ^{2}+4a^{2}a^{4}% \mathcal{e}^{2}}. \label{quadr - app-3}\]]one can easily check that this expression yields @xmath224 for @xmath73, and that @xmath225, as it must be. finally, using relation @xmath226, one obtains eq. (cubic - quintic-1) from eq. ([quadr - app-3]), after a straightforward algebra.
The variational energy
in this appendix we aim to study minima of variational energy ([e3]). when @xmath73, one sees that, for @xmath93, the energy per particle tends to @xmath94 at @xmath95, and to @xmath112 at @xmath227. then, with regard to @xmath228, no local (metastable) minima exist, and variational wave function ([var]) is not the ground - state for any finite width. for @xmath66, one obtains the global minimum at @xmath229, which implies the collapse. for @xmath230, the situation is different : @xmath231 as @xmath232 (because @xmath89), and @xmath233 for @xmath234. then, it is necessary to find the value of @xmath192 at which derivative @xmath181 has two real zeros. introducing the parameter @xmath235with @xmath236 defined as per eq. ([beta]), one can write condition @xmath182 as @xmath237where @xmath238, as defined above. equation ([t - cond]) can be satisfied if @xmath239 is smaller than a maximum value, @xmath240, and it then has two roots, @xmath241 and @xmath242, which correspond, respectively to the minimum at @xmath110, and maximum at @xmath104 (see fig. [fig1]). for @xmath243, eq. ([t]) has no roots, hence the variational energy has no minima at finite values of the soliton s width, @xmath11. a plot of @xmath241 as a function of @xmath239 is presented in fig. [fig_theta], where the maximum value of @xmath241 is @xmath244. the energy minimum at @xmath241 is a global one if @xmath245 ; using eq. ([e3]), this condition reads @xmath246as one can see from fig. [fig_theta], condition ([condition - app]) is satisfied for @xmath247, where @xmath248 ; then, a global minimum exists only for @xmath249, while for @xmath250 the minimum is local, corresponding to a metastable state. using the value of @xmath137 and definition ([t]), one arrives at eq. ([c_2_star]). as a function of parameter @xmath239 (defined in eq. ([t - cond])) for @xmath73. the dashed line is the plot of function @xmath251 versus @xmath239. the maximum value of @xmath252 at @xmath253 is indicated.] for @xmath76 (recall it corresponds to the two - body repulsion), variational energy ([e3]) for @xmath93 does not have a minimum at finite values of @xmath11. however, for @xmath66 a finite minimum is possible. indeed, with definition of @xmath156 as per eq. ([g - cond - text]), condition @xmath182 can be written as @xmath254for @xmath255, eq. ([g - cond]) has two roots. by imposing the condition that the value of the energy at @xmath110 be smaller than @xmath112, one gets @xmath256. then, similar to the situation considered above, a global minimum exists only for @xmath257, while for @xmath258 the minimum is local. for @xmath88, condition @xmath259 reads @xmath260one can see that condition ([t - g - cond]) is satisfied for @xmath261, with @xmath262. then, for @xmath263, i.e., for @xmath34 small enough, the variational energy does not have a minimum. imposing the condition that the minimum is global leads to @xmath247, with @xmath264. then, for @xmath265, i.e. for @xmath34 smaller than a critical value, the variational energy can not have a _ global _ minimum at a finite value of @xmath11, i.e., localized states can not realize a global minimum. functions @xmath266 and @xmath267 are plotted in fig. [t - vs - g] ; in fig. [theta - max], we plot maximum value @xmath268 of @xmath241 for @xmath269, as a function of @xmath156. _ acknowledgement : _ we thank l. salasnich for useful comments and discussions. this work was partially supported by esf project instans and by miur projects quantum field theory and statistical mechanics in low dimensions ", and quantum noise in mesoscopic systems ". abdullaev, b. b. baizakov, s. a. darmanyan, v. v. konotop, and m. salerno, phys. a * 64 *, 043606 (2001) ; i. carusotto, d. embriaco, and g. c. la rocca, _ ibid_. * 65 *, (2002) 053611 ; b. b. baizakov, v. v. konotop, and m. salerno, j. phys. b * 35 *, (2002) 5105 ; e. a. ostrovskaya and y. s. kivshar, phys. * 90 *, (2003) 160407 ; opt. exp. * 12 *, (2004) 19 | we study the effect of an optical lattice (ol) on the ground - state properties of one - dimensional ultracold bosons with three - body attraction and two - body repulsion, which are described by a cubic - quintic gross - pitaevskii equation with a periodic potential. without the ol and with a vanishing two - body interaction term, soliton solutions of the townes type are possible only at a critical value of the three - body interaction strength, at which an infinite degeneracy of the ground - state occurs ; a repulsive two - body interaction makes such localized solutions unstable.
we show that the ol opens a stability window around the critical point when the strength of the periodic potential is above a critical threshold.
we also consider the effect of an external parabolic trap, studying how the stability of the solitons depends on matching between minima of the periodic potential and the minimum of the parabolic trap. | 0902.4734 |
Introduction
factor analysis is one of the most useful tools for modeling common dependence among multivariate outputs. suppose that we observe data @xmath0 that can be decomposed as @xmath1 where @xmath2 are unobservable common factors ; @xmath3 are corresponding factor loadings for variable @xmath4, and @xmath5 denotes the idiosyncratic component that can not be explained by the static common component. here, @xmath6 and @xmath7, respectively, denote the dimension and sample size of the data. model ([eq1.1]) has broad applications in the statistics literature. for instance, @xmath8 can be expression profiles or blood oxygenation level dependent (bold) measurements for the @xmath9th microarray, proteomic or fmri - image, whereas @xmath4 represents a gene or protein or a voxel. see, for example, @xcite. the separations between the common factors and idiosyncratic components are carried out by the low - rank plus sparsity decomposition. see, for example, @xcite. the factor model ([eq1.1]) has also been extensively studied in the econometric literature, in which @xmath10 is the vector of economic outputs at time @xmath9 or excessive returns for individual assets on day @xmath9. the unknown factors and loadings are typically estimated by the principal component analysis (pca) and the separations between the common factors and idiosyncratic components are characterized via static pervasiveness assumptions. see, for instance, @xcite among others. in this paper, we consider static factor model, which differs from the dynamic factor model [@xcite, @xcite (@xcite)]. the dynamic model allows more general infinite dimensional representations. for this type of model, the frequency domain pca [@xcite] was applied on the spectral density. the so - called _ dynamic pervasiveness _ condition also plays a crucial role in achieving consistent estimation of the spectral density. accurately estimating the loadings and unobserved factors are very important in statistical applications. in calculating the false - discovery proportion for large - scale hypothesis testing, one needs to adjust accurately the common dependence via subtracting it from the data in ([eq1.1]) [@xcite]. in financial applications, we would like to understand accurately how each individual stock depends on unobserved common factors in order to appreciate its relative performance and risks. in the aforementioned applications, dimensionality is much higher than sample - size. however, the existing asymptotic analysis shows that the consistent estimation of the parameters in model ([eq1.1]) requires a relatively large @xmath7. in particular, the individual loadings can be estimated no faster than @xmath11. but large sample sizes are not always available. even with the availability of `` big data, '' heterogeneity and other issues make direct applications of ([eq1.1]) with large @xmath7 infeasible. for instance, in financial applications, to pertain the stationarity in model ([eq1.1]) with time - invariant loading coefficients, a relatively short time series is often used. to make observed data less serially correlated, monthly returns are frequently used to reduce the serial correlations, yet a monthly data over three consecutive years contain merely 36 observations. to overcome the aforementioned problems, and when relevant covariates are available, it may be helpful to incorporate them into the model. let @xmath12 be a vector of @xmath13-dimensional covariates associated with the @xmath4th variables. in the seminal papers by @xcite and @xcite, the authors studied the following semi - parametric factor model : @xmath14 where loading coefficients in ([eq1.1]) are modeled as @xmath15 for some functions @xmath16. for instance, in health studies, @xmath17 can be individual characteristics (e.g., age, weight, clinical and genetic information) ; in financial applications @xmath17 can be a vector of firm - specific characteristics (market capitalization, price - earning ratio, etc.). the semiparametric model ([eq1.2]), however, can be restrictive in many cases, as it requires that the loading matrix be fully explained by the covariates. a natural relaxation is the following semiparametric model : @xmath18 where @xmath19 is the component of loading coefficient that can not be explained by the covariates @xmath17. let @xmath20. we assume that @xmath21 have mean zero, and are independent of @xmath22 and @xmath23. in other words, we impose the following factor structure : @xmath24 which reduces to model ([eq1.2]) when @xmath25 and model ([eq1.1]) when @xmath26. when @xmath17 genuinely explains a part of loading coefficients @xmath27, the variability of @xmath28 is smaller than that of @xmath27. hence, the coefficient @xmath19 can be more accurately estimated by using regression model ([eq1.3]), as long as the functions @xmath29 can be accurately estimated. let @xmath30 be the @xmath31 matrix of @xmath32, @xmath33 be the @xmath34 matrix of @xmath35, @xmath36 be the @xmath37 matrix of @xmath38, @xmath39 be the @xmath37 matrix of @xmath19 and @xmath40 be @xmath31 matrix of @xmath5. then model ([eq1.4]) can be written in a more compact matrix form : @xmath41 we treat the loadings @xmath36 and @xmath39 as realizations of random matrices throughout the paper. this model is also closely related to the _ supervised singular value decomposition _ model, recently studied by @xcite. the authors showed that the model is useful in studying the gene expression and single - nucleotide polymorphism (snp) data, and proposed an em algorithm for parameter estimation. we propose a projected - pca estimator for both the loading functions and factors. our estimator is constructed by first projecting @xmath30 onto the sieve space spanned by @xmath22, then applying pca to the projected data or fitted values. due to the approximate orthogonality condition of @xmath42, @xmath40 and @xmath39, the projection of @xmath30 is approximately @xmath43, as the smoothing projection suppresses the noise terms @xmath39 and @xmath40 substantially. therefore, applying pca to the projected data allows us to work directly on the sample covariance of @xmath43, which is @xmath44 under normalization conditions. this substantially improves the estimation accuracy, and also facilitates the theoretical analysis. in contrast, the traditional pca method for factor analysis [e.g., @xcite, @xcite] is no longer suitable in the current context. moreover, the idea of projected - pca is also potentially applicable to dynamic factor models of @xcite, by first projecting the data onto the covariate space. the asymptotic properties of the proposed estimators are carefully studied. we demonstrate that as long as the projection is genuine, the consistency of the proposed estimator for latent factors and loading matrices requires only @xmath45, and @xmath7 does not need to grow, which is attractive in the typical high - dimension - low - sample - size (hdlss) situations [e.g., @xcite]. in addition, if both @xmath6 and @xmath7 grow simultaneously, then with sufficiently smooth @xmath29, using the sieve approximation, the rate of convergence for the estimators is much faster than those of the existing results for model ([eq1.1]). typically, the loading functions can be estimated at a convergence rate @xmath46, and the factor can be estimated at @xmath47. throughout the paper, @xmath48 and @xmath49 are assumed to be constant and do not grow. let @xmath50 be a @xmath37 matrix of @xmath51. model ([eq1.3]) implies a decomposition of the loading matrix : @xmath52 where @xmath36 and @xmath39 are orthogonal loading components in the sense that @xmath53. we conduct two specification tests for the hypotheses : @xmath54 the first problem is about testing whether the observed covariates have explaining power on the loadings. if the null hypothesis is rejected, it gives us the theoretical basis to employ the projected - pca, as the projection is now genuine. our empirical study on the asset returns shows that firm market characteristics do have explanatory power on the factor loadings, which lends further support to our projected - pca method. the second tests whether covariates fully explain the loadings. our aforementioned empirical study also shows that model ([eq1.2]) used in the financial econometrics literature is inadequate and more generalized model ([eq1.5]) is necessary. as claimed earlier, even if @xmath55 does not hold, as long as @xmath56, the projected - pca can still consistently estimate the factors as @xmath45, and @xmath7 may or may not grow. our simulated experiments confirm that the estimation accuracy is gained more significantly for small @xmath7 s. this shows one of the benefits of using our projected - pca method over the traditional methods in the literature. in addition, as a further illustration of the benefits of using projected data, we apply the projected - pca to consistently estimate the number of factors, which is similar to those in @xcite and @xcite. different from these authors, our method applies to the projected data, and we demonstrate numerically that this can significantly improve the estimation accuracy. we focus on the case when the observed covariates are time - invariant. when @xmath7 is small, these covariates are approximately locally constant, so this assumption is reasonable in practice. on the other hand, there may exist individual characteristics that are time - variant [e.g., see @xcite]. we expect the conclusions in the current paper to still hold if some smoothness assumptions are added for the time varying components of the covariates. due to the space limit, we provide heuristic discussions on this case in the supplementary material of this paper [@xcite]. in addition, note that in the usual factor model, @xmath50 was assumed to be deterministic. in this paper, however, @xmath50 is mainly treated to be stochastic, and potentially depend on a set of covariates. but we would like to emphasize that the results presented in section [1541512515] under the framework of more general factor models hold regardless of whether @xmath50 is stochastic or deterministic. finally, while some financial applications are presented in this paper, the projected - pca is expected to be useful in broad areas of statistical applications [e.g., see @xcite for applications in gene expression data analysis]. throughout this paper, for a matrix @xmath57, let @xmath58 and @xmath59, @xmath60 denote its frobenius, spectral and max- norms. let @xmath61 and @xmath62 denote the minimum and maximum eigenvalues of a square matrix. for a vector @xmath63, let @xmath64 denote its euclidean norm. the rest of the paper is organized as follows. section [sec2] introduces the new projected - pca method and defines the corresponding estimators for the loadings and factors. sections [1541512515] and [s4] provide asymptotic analysis of the introduced estimators. section [sec5] introduces new specification tests for the orthogonal decomposition of the semiparametric loadings. section [sec6] concerns about estimating the number of factors. section [sec7] presents numerical results. finally, section [sec8] concludes. all the proofs are given in the and the supplementary material [@xcite].
Projected principal component analysis
in the high - dimensional factor model, let @xmath50 be the @xmath37 matrix of loadings. then the general model ([eq1.1]) can be written as @xmath65 suppose we additionally observe a set of covariates @xmath66. the basic idea of the projected - pca is to smooth the observations @xmath67 for each given day @xmath9 against its associated covariates. more specifically, let @xmath68 be the fitted value after regressing @xmath67 on @xmath69 for each given @xmath9. this results in a smooth or projected observation matrix @xmath70, which will also be denoted by @xmath71. the projected - pca then estimates the factors and loadings by running the pca based on the projected data @xmath70. here, we heuristically describe the idea of projected - pca ; rigorous analysis will be carried out afterward. let @xmath72 be a space spanned by @xmath73, which is orthogonal to the error matrix @xmath40. let @xmath74 denote the projection matrix onto @xmath72 [whose formal definition will be given in ([eq2.5]) below. at the population level, @xmath74 approximates the conditional expectation operator @xmath75, which satisfies @xmath76, then @xmath77 and @xmath78. hence, analyzing the projected data @xmath79 is an approximately noiseless problem, and the sample covariance has the following approximation : @xmath80 we now argue that @xmath33 and @xmath81 can be recovered from the projected data @xmath70 under some suitable normalization condition. the normalization conditions we impose are @xmath82 under this normalization, using ([eq2.1a]), @xmath83. we conclude that the columns of @xmath84 are approximately @xmath85 times the first @xmath86 eigenvectors of the @xmath87 matrix @xmath88. therefore, the projected - pca naturally defines a factor estimator @xmath89 using the first @xmath86 principal components of @xmath90. the projected loading matrix @xmath81 can also be recovered from the projected data @xmath71 in two (equivalent) ways. given @xmath33, from @xmath91, we see @xmath92. alternatively, consider the @xmath93 projected sample covariance : @xmath94 where @xmath95 is a remaining term depending on @xmath96. right multiplying @xmath81 and ignoring @xmath95, we obtain @xmath97. hence, the (normalized) columns of @xmath81 approximate the first @xmath86 eigenvectors of @xmath98, the @xmath93 sample covariance matrix based on the projected data. therefore, we can either estimate @xmath81 by @xmath99 given @xmath89, or by the leading eigenvectors of @xmath98. in fact, we shall see later that these two estimators are equivalent. if in addition, @xmath100, that is, the loading matrix belongs to the space @xmath72, then @xmath50 can also be recovered from the projected data. the above arguments are the fundament of the projected - pca, and provide the rationale of our estimators to be defined in section [sec2.3]. we shall make the above arguments rigorous by showing that the projected error @xmath101 is asymptotically negligible and, therefore, the idiosyncratic error term @xmath40 can be completely removed by the projection step. as one of the useful examples of forming the space @xmath102 and the projection operator, this paper considers model ([eq1.4]), where @xmath17 s and @xmath32 s are the only observable data, and @xmath103 are unknown nonparametric functions. the specific case ([eq1.2]) (with @xmath104) was used extensively in the financial studies by @xcite, @xcite and @xcite, with @xmath17 s being the observed `` market characteristic variables. '' we assume @xmath86 to be known for now. in section [sec6], we will propose a projected - eigenvalue - ratio method to consistently estimate @xmath86 when it is unknown. we assume that @xmath105 does not depend on @xmath9, which means the loadings represent the cross - sectional heterogeneity only. such a model specification is reasonable since in many applications using factor models, to pertain the stationarity of the time series, the analysis can be conducted within each fixed time window with either a fixed or slowly - growing @xmath7. through localization in time, it is not stringent to require the loadings be time - invariant. this also shows one of the attractive features of our asymptotic results : under mild conditions, our factor estimates are consistent even if @xmath7 is finite. to nonparametrically estimate @xmath105 without the curse of dimensionality when @xmath17 is multivariate, we assume @xmath29 to be additive : for each @xmath106, there are @xmath107 nonparametric functions such that @xmath108 each additive component of @xmath109 is estimated by the sieve method. define @xmath110 to be a set of basis functions (e.g., b - spline, fourier series, wavelets, polynomial series), which spans a dense linear space of the functional space for @xmath111. then for each @xmath112, @xmath113 here, @xmath114 are the sieve coefficients of the @xmath115th additive component of @xmath105, corresponding to the @xmath116th factor loading ; @xmath117 is a `` remaining function '' representing the approximation error ; @xmath118 denotes the number of sieve terms which grows slowly as @xmath45. the basic assumption for sieve approximation is that @xmath119 as @xmath120. we take the same basis functions in ([eq2.4]) purely for simplicity of notation. define, for each @xmath121 and for each @xmath122, @xmath123 then we can write @xmath124 let @xmath125 be a @xmath126 matrix of sieve coefficients, @xmath127 be a @xmath128 matrix of basis functions, and @xmath129 be @xmath37 matrix with the @xmath130th element @xmath131. then the matrix form of ([eq2.3]) and ([eq2.4]) is @xmath132 substituting this into ([eq1.5]), we write @xmath133 we see that the residual term consists of two parts : the sieve approximation error @xmath134 and the idiosyncratic @xmath40. furthermore, the random effect assumption on the coefficients @xmath39 makes it also behave like noise, and hence negligible when the projection operator @xmath74 is applied. based on the idea described in section [sec2.1], we propose a projected - pca method, where @xmath72 is the sieve space spanned by the basis functions of @xmath42, and @xmath74 is chosen as the projection matrix onto @xmath72, defined by the @xmath93 projection matrix @xmath135 the estimators of the model parameters in ([eq1.5]) are defined as follows. the columns of @xmath136 are defined as the eigenvectors corresponding to the first @xmath86 largest eigenvalues of the @xmath87 matrix @xmath137, and @xmath138 is the estimator of @xmath36. the intuition can be readily seen from the discussions in section [sec2.1], which also provides an alternative formulation of @xmath139 as follows : let @xmath140 be a @xmath141 diagonal matrix consisting of the largest @xmath86 eigenvalues of the @xmath93 matrix @xmath142. let @xmath143 be a @xmath37 matrix whose columns are the corresponding eigenvectors. according to the relation @xmath144 described in section [sec2.1], we can also estimate @xmath36 or @xmath81 by @xmath145 we shall show in lemma [la.1add] that this is equivalent to ([eq2.6]). therefore, unlike the traditional pca method for usual factor models [e.g., @xcite, @xcite], the projected - pca takes the principal components of the projected data @xmath71. the estimator is thus invariant to the rotation - transformations of the sieve bases. the estimation of the loading component @xmath39 that can not be explained by the covariates can be estimated as follows. with the estimated factors @xmath89, the least - squares estimator of loading matrix is @xmath146, by using ([eq2.1]) and ([eq2.2]). therefore, by ([eq1.5]), a natural estimator of @xmath147 is @xmath148 consider a panel data model with time - varying coefficients as follows : @xmath149 where @xmath17 is a @xmath13-dimensional vector of time - invariant regressors for individual @xmath4 ; @xmath150 denotes the unobservable random time effect ; @xmath5 is the regression error term. the regression coefficient @xmath151 is also assumed to be random and time - varying, but is common across the cross - sectional individuals. the semiparametric factor model admits ([eq2.8]) as a special case. note that ([eq2.8]) can be rewritten as @xmath152 with @xmath153 unobservable `` factors '' @xmath154 and `` loading '' @xmath155. the model ([eq1.4]) being considered, on the other hand, allows more general nonparametric loading functions.
Projected-pca in conventional factor models
let us first consider the asymptotic performance of the projected - pca in the conventional factor model : @xmath156 in the usual statistical applications for factor analysis, the latent factors are assumed to be serially independent, while in financial applications, the factors are often treated to be weakly dependent time series satisfying strong mixing conditions. we now demonstrate by a simple example that latent factors @xmath33 can be estimated at a faster rate of convergence by projected - pca than the conventional pca and that they can be consistently estimated even when sample size @xmath7 is finite. [ex3.1] to appreciate the intuition, let us consider a specific case in which @xmath157 so that model ([eq1.4]) reduces to @xmath158 assume that @xmath159 is so smooth that it is in fact a constant @xmath160 (otherwise, we can use a local constant approximation), where @xmath161. then the model reduces to @xmath162 the projection in this case is averaging over @xmath4, which yields @xmath163 where @xmath164, @xmath165 and @xmath166 denote the averages of their corresponding quantities over @xmath4. for the identification purpose, suppose @xmath167, and @xmath168. ignoring the last two terms, we obtain estimators @xmath169 these estimators are special cases of the projected - pca estimators. to see this, define @xmath170, and let @xmath171 be a @xmath6-dimensional column vector of ones. take a naive basis @xmath172 ; then the projected data matrix is in fact @xmath173. consider the @xmath87 matrix @xmath174, whose largest eigenvalue is @xmath175. from @xmath176 we have the first eigenvector of @xmath137 equals @xmath177. hence, the projected - pca estimator of factors is @xmath178. in addition, the projected - pca estimator of the loading vector @xmath179 is @xmath180 hence, the projected - pca - estimator of @xmath181 equals @xmath182. these estimators match with ([e3.2]). moreover, since the ignored two terms @xmath183 and @xmath184 are of order @xmath185, @xmath186 and @xmath187 converge whether or not @xmath7 is large. note that this simple example satisfies all the assumptions to be stated below, and @xmath188 and @xmath189 achieve the same rate of convergence as that of theorem [th4.1]. we shall present more details about this example in appendix g in the supplementary material [@xcite]. we now state the conditions and results formally in the more general factor model ([eq3.1]). recall that the projection matrix is defined as @xmath190 the following assumption is the key condition of the projected - pca. [ass3.1] there are positive constants @xmath191 and @xmath192 such that, with probability approaching one (as @xmath193), @xmath194 since the dimensions of @xmath195 and @xmath50 are, respectively, @xmath196 and @xmath37, assumption [ass3.1] requires @xmath197, which is reasonable since we assume @xmath86, the number of factors, to be fixed throughout the paper. assumption [ass3.1] is similar to the _ pervasive _ condition on the factor loadings [@xcite]. in our context, this condition requires the covariates @xmath42 have nonvanishing explaining power on the loading matrix, so that the projection matrix @xmath198 has spiked eigenvalues. note that it rules out the case when @xmath42 is completely unassociated with the loading matrix @xmath50 (e.g., when @xmath42 is pure noise). one of the typical examples that satisfies this assumption is the semiparametric factor model [model ([eq1.4])]. we shall study this specific type of factor model in section [s4], and prove assumption [ass3.1] in the supplementary material [@xcite]. note that @xmath33 and @xmath50 are not separately identified, because for any nonsingular @xmath199, @xmath200. therefore, we assume the following. [ass3.2] almost surely, @xmath201 and @xmath198 is a @xmath141 diagonal matrix with distinct entries. this condition corresponds to the pc1 condition of @xcite, which separately identifies the factors and loadings from their product @xmath202. it is often used in factor analysis for identification, and means that the columns of factors and loadings can be orthogonalized [also see @xcite]. [ass3.3] (i) there are @xmath203 and @xmath204 so that with probability approaching one (as @xmath193), @xmath205 \(ii) @xmath206. note that @xmath207 and @xmath208 is a vector of dimensionality @xmath209. thus, condition (i) can follow from the strong law of large numbers. for instance, @xmath22 are weakly correlated and in the population level @xmath210 is well - conditioned. in addition, this condition can be satisfied through proper normalizations of commonly used basis functions such as b - splines, wavelets, fourier basis, etc. in the general setup of this paper, we allow @xmath211 s to be cross - sectionally dependent and nonstationary. regularity conditions about weak dependence and stationarity are imposed only on @xmath212 as follows. we impose the strong mixing condition. let @xmath213 and @xmath214 denote the @xmath215-algebras generated by @xmath216 and @xmath217, respectively. define the mixing coefficient @xmath218 [ass3.4] (i) @xmath219 is strictly stationary. in addition, @xmath220 for all @xmath221 ; @xmath222 is independent of @xmath223. strong mixing : there exist @xmath224 such that for all @xmath225, @xmath226 weak dependence : there is @xmath227 so that @xmath228 exponential tail : there exist @xmath229 satisfying @xmath230 and @xmath231, such that for any @xmath232, @xmath122 and @xmath233, @xmath234 assumption [ass3.4] is standard, especially condition (iii) is commonly imposed for high - dimensional factor analysis [e.g., @xcite], which requires @xmath235 be weakly dependent both serially and cross - sectionally. it is often satisfied when the covariance matrix @xmath236 is sufficiently sparse under the strong mixing condition. we provide primitive conditions of condition (iii) in the supplementary material [@xcite]. formally, we have the following theorem : [th3.1] consider the conventional factor model ([eq3.1]) with assumptions [ass3.1][ass3.4]. the projected - pca estimators @xmath237 and @xmath238 defined in section [sec2.3] satisfy, as @xmath239 [@xmath240 may either grow simultaneously with @xmath6 satisfying @xmath241 or stay constant with @xmath242, @xmath243 to compare with the traditional pca method, the convergence rate for the estimated factors is improved for small @xmath7. in particular, the projected - pca does not require @xmath244, and also has a good rate of convergence for the loading matrix up to a projection transformation. hence, we have achieved a finite-@xmath7 consistency, which is particularly interesting in the `` high - dimensional - low - sample - size '' (hdlss) context, considered by @xcite. in contrast, the traditional pca method achieves a rate of convergence of @xmath245 for estimating factors, and @xmath246 for estimating loadings. see remarks [re4.1], [re4.2] below for additional details. let @xmath247 be the @xmath93 covariance matrix of @xmath248. convergence ([eq3.4add]) in theorem [th3.1] also describes the relationship between the leading eigenvectors of @xmath98 and those of @xmath249. to see this, let @xmath250 be the eigenvectors of @xmath249 corresponding to the first @xmath86 eigenvalues. under the _ pervasiveness condition _, @xmath251 can be approximated by @xmath50 multiplied by a positive definite matrix of transformation [@xcite]. in the context of projected - pca, by definition, @xmath252 ; here we recall that @xmath253 is a diagonal matrix consisting of the largest @xmath86 eigenvalues of @xmath98, and @xmath254 is a @xmath37 matrix whose columns are the corresponding eigenvectors. then ([eq3.4add]) immediately implies the following corollary, which complements the pca consistency in _ spiked covariance models _ [e.g., @xcite and @xcite]. [th3.2] under the conditions of theorem [th3.1], there is a @xmath141 positive definite matrix @xmath255, whose eigenvalues are bounded away from both zero and infinity, so that as @xmath193 [@xmath240 may either grow simultaneously with @xmath6 satisfying @xmath241 or stay constant with @xmath242, @xmath256
Projected-pca in semiparametric factor models
in the semiparametric factor model, it is assumed that @xmath257, where @xmath105 is a nonparametric smooth function for the observed covariates, and @xmath19 is the unobserved random loading component that is independent of @xmath17. hence, the model is written as @xmath258 in the matrix form, @xmath259 and @xmath36 does not vanish (pervasive condition ; see assumption [ass4.2] below). the estimators @xmath237 and @xmath238 are the projected - pca estimators as defined in section [sec2.3]. we now define the estimator of the nonparametric function @xmath29, @xmath260. in the matrix form, the projected data has the following sieve approximated representation : @xmath261 where @xmath262 is `` small '' because @xmath39 and @xmath40 are orthogonal to the function space spanned by @xmath42, and @xmath129 is the sieve approximation error. the sieve coefficient matrix @xmath263 can be estimated by least squares from the projected model ([eq4.1]) : ignore @xmath264, replace @xmath33 with @xmath237, and solve ([eq4.1]) to obtain @xmath265^{-1}\phi (\bx)'\by{\widehat\bf}.\]] we then estimate @xmath29 by @xmath266 where @xmath267 denotes the support of @xmath17. when @xmath268, @xmath36 can be understood as the projection of @xmath50 onto the sieve space spanned by @xmath42. hence, the following assumption is a specific version of assumptions [ass3.1] and [ass3.2] in the current context. [ass4.1] (i) almost surely, @xmath201 and @xmath269 is a @xmath141 diagonal matrix with distinct entries. \(ii) there are two positive constants @xmath191 and @xmath192 so that with probability approaching one (as @xmath193), @xmath270 in this section, we do not need to assume @xmath271 to be i.i.d. for the estimation purpose. cross - sectional weak dependence as in assumption [ass4.2](ii) below would be sufficient. the i.i.d. assumption will be only needed when we consider specification tests in section [sec5]. write @xmath272, and @xmath273 [ass4.2] (i) @xmath274 and @xmath22 is independent of @xmath275. \(ii) @xmath276, @xmath277 and @xmath278 the following set of conditions is concerned about the accuracy of the sieve approximation. [ass4.3] @xmath279, \(i) the loading component @xmath280 belongs to a hlder class @xmath281 defined by @xmath282 for some @xmath283 ; \(ii) the sieve coefficients @xmath284 satisfy for @xmath285, as @xmath286, @xmath287 where @xmath288 is the support of the @xmath115th element of @xmath17, and @xmath118 is the sieve dimension. \(iii) @xmath289. condition (ii) is satisfied by common basis. for example, when @xmath290 is polynomial basis or b - splines, condition (ii) is implied by condition (i) [see, e.g., @xcite and @xcite]. [th4.1] suppose @xmath241. under assumptions [ass3.3], [ass3.4], [ass4.1][ass4.3], as @xmath291, @xmath7 can be either divergent or bounded, we have that @xmath292 in addition, if @xmath244 simultaneously with @xmath6 and @xmath118, then @xmath293 the optimal @xmath294 simultaneously minimizes the convergence rates of the factors and nonparametric loading function @xmath29. it also satisfies the constraint @xmath295 as @xmath296. with @xmath297, we have @xmath298 and @xmath299 satisfies @xmath300 some remarks about these rates of convergence compared with those of the conventional factor analysis are in order. [re4.1]the rates of convergence for factors and nonparametric functions do not require @xmath244. when @xmath301, @xmath302 the rates still converge fast when @xmath6 is large, demonstrating the blessing of dimensionality. this is an attractive feature of the projected - pca in the hdlss context, as in many applications, the stationarity of a time series and the time - invariance assumption on the loadings hold only for a short period of time. in contrast, in the usual factor analysis, consistency is granted only when @xmath303. for example, according to @xcite (lemma c.1), the regular pca method has the following convergence rate : @xmath304 which is inconsistent when @xmath7 is bounded. [re4.2]when both @xmath6 and @xmath7 are large, the projected - pca estimates factors as well as the regular pca does, and achieves a faster rate of convergence for the estimated loadings when @xmath19 vanishes. in this case, @xmath305, the loading matrix is estimated by @xmath306, and @xmath307 in contrast, the regular pca method as in @xcite yields @xmath308 comparing these rates, we see that when @xmath29 s are sufficiently smooth (larger @xmath309), the rate of convergence for the estimated loadings is also improved.
Semiparametric specification test
the loading matrix always has the following orthogonal decomposition : @xmath310 where @xmath39 is interpreted as the loading component that can not be explained by @xmath42. we consider two types of specification tests : testing @xmath311, and @xmath312. the former tests whether the observed covariates have explaining powers on the loadings, while the latter tests whether the covariates fully explain the loadings. the former provides a diagnostic tool as to whether or not to employ the projected - pca ; the latter tests the adequacy of the semiparametric factor models in the literature. testing whether the observed covariates have explaining powers on the factor loadings can be formulated as the following null hypothesis : @xmath314 due to the approximate orthogonality of @xmath42 and @xmath39, we have @xmath315. hence, the null hypothesis is approximately equivalent to @xmath316 this motivates a statistic @xmath317 for a consistent loading estimator @xmath318. normalizing the test statistic by its asymptotic variance leads to the test statistic @xmath319 where the @xmath141 matrix @xmath320 is the weight matrix. the null hypothesis is rejected when @xmath321 is large. the projected - pca estimator is inappropriate under the null hypothesis as the projection is not genuine. we therefore use the least squares estimator @xmath322, leading to the test statistic @xmath323 here, we take @xmath324 as the traditional pca estimator : the columns of @xmath325 are the first @xmath86 eigenvectors of the @xmath87 data matrix @xmath326. connor, hagmann and linton (@xcite) applied the semiparametric factor model to analyzing financial returns, who assumed that @xmath328, that is, the loading matrix can be fully explained by the observed covariates. it is therefore natural to test the following null hypothesis of specification : @xmath329 recall that @xmath330 so that @xmath331. therefore, essentially the specification testing problem is equivalent to testing @xmath332 that is, we are testing whether the loading matrix in the factor model belongs to the space spanned by the observed covariates. a natural test statistic is thus based on the weighted quadratic form @xmath333 for some @xmath334 positive definite weight matrix @xmath335, where @xmath237 is the projected - pca estimator for factors and @xmath336. to control the size of the test, we take @xmath337, where @xmath338 is a diagonal covariance matrix of @xmath339 under @xmath340, assuming that @xmath341 are uncorrelated. we replace @xmath342 with its consistent estimator : let @xmath343. define @xmath344 then the operational test statistic is defined to be @xmath345 the null hypothesis is rejected for large values of @xmath346. for the testing purpose, we assume @xmath347 to be i.i.d., and let @xmath348 simultaneously. the following assumption regulates the relation between @xmath7 and @xmath6. [ass5.1] suppose (i) @xmath349 are independent and identically distributed ; @xmath350, and @xmath351 ; @xmath118 and @xmath309 satisfy : @xmath352, and @xmath353. condition (ii) requires a balance of the dimensionality and the sample size. on one hand, a relatively large sample size is desired [@xmath354 so that the effect of estimating @xmath342 is negligible asymptotically. on the other hand, as is common in high - dimensional factor analysis, a lower bound of the dimensionality is also required [condition @xmath350] to ensure that the factors are estimated accurately enough. such a required balance is common for high - dimensional factor analysis [e.g., @xcite, @xcite] and in the recent literature for pca [e.g., @xcite, @xcite]. the i.i.d. assumption of covariates @xmath17 in condition (i) can be relaxed with further distributional assumptions on @xmath356 (e.g., assuming @xmath356 to be gaussian). the conditions on @xmath118 in condition (iii) is consistent with those of the previous sections. we focus on the case when @xmath357 is gaussian, and show that under @xmath358, @xmath359 and under @xmath55 @xmath360 whose conditional distributions (given @xmath33) under the null are @xmath361 with degree of freedom, respectively, @xmath362 and @xmath363. we can derive their standardized limiting distribution as @xmath364. this is given in the following result. [th5.1] suppose assumptions [ass3.3], [ass3.4], [ass4.2], [ass5.1] hold. then under @xmath358, @xmath365 where @xmath48 and @xmath49. in addition, suppose assumptions [ass4.1] and [ass4.3] further hold, @xmath366 is i.i.d. @xmath367 with a diagonal covariance matrix @xmath338 whose elements are bounded away from zero and infinity. then under @xmath55, @xmath368 in practice, when a relatively small sieve dimension @xmath118 is used, one can instead use the upper @xmath369-quantile of the @xmath370 distribution for @xmath371. we require @xmath5 be independent across @xmath9, which ensures that the covariance matrix of the leading term @xmath372 to have a simple form @xmath373. this assumption can be relaxed to allow for weakly dependent @xmath374, but many autocovariance terms will be involved in the covariance matrix. one may regularize standard autocovariance matrix estimators such as @xcite and @xcite to account for the high dimensionality. moreover, we assume @xmath338 be diagonal to facilitate estimating @xmath342, which can also be weakened to allow for a nondiagonal but sparse @xmath338. regularization methods such as thresholding [@xcite] can then be employed, though they are expected to be more technically involved.
Estimating the number of factors from projected data
we now address the problem of estimating @xmath48 when it is unknown. once a consistent estimator of @xmath86 is obtained, all the results achieved carry over to the unknown @xmath86 case using a conditioning argument., then argue that the results still hold unconditionally as @xmath375.] in principle, many consistent estimators of @xmath86 can be employed, for example, @xcite, @xcite, @xcite, @xcite. more recently, @xcite and @xcite proposed to select the largest ratio of the adjacent eigenvalues of @xmath326, based on the fact that the @xmath86 largest eigenvalues of the sample covariance matrix grow as fast as @xmath6 increases, while the remaining eigenvalues either remain bounded or grow slowly. we extend ahn and horenstein s (@xcite) theory in two ways. first, when the loadings depend on the observable characteristics, it is more desirable to work on the projected data @xmath71. due to the orthogonality condition of @xmath40 and @xmath42, the projected data matrix is approximately equal to @xmath43. the projected matrix @xmath376 thus allows us to study the eigenvalues of the principal matrix component @xmath377, which directly connects with the strengths of those factors. since the nonvanishing eigenvalues of @xmath376 and @xmath378 are the same, we can work directly with the eigenvalues of the matrix @xmath379. second, we allow @xmath380. let @xmath381 denote the @xmath116th largest eigenvalue of the projected data matrix @xmath137. we assume @xmath382, which naturally holds if the sieve dimension @xmath118 slowly grows. the estimator is defined as @xmath383 the following assumption is similar to that of @xcite. recall that @xmath384 is a @xmath31 matrix of the idiosyncratic components, and @xmath385 denotes the @xmath386 covariance matrix of @xmath339. [ass6.1] the error matrix @xmath40 can be decomposed as @xmath387 where : the eigenvalues of @xmath338 are bounded away from zero and infinity, @xmath388 is a @xmath7 by @xmath7 positive semidefinite nonstochastic matrix, whose eigenvalues are bounded away from zero and infinity, @xmath389 is a @xmath31 stochastic matrix, where @xmath390 is independent in both @xmath4 and @xmath9, and @xmath391 are i.i.d. isotropic sub - gaussian vectors, that is, there is @xmath392, for all @xmath232, @xmath393 there are @xmath394, almost surely, @xmath395 this assumption allows the matrix @xmath40 to be both cross - sectionally and serially dependent. the @xmath87 matrix @xmath388 captures the serial dependence across @xmath9. in the special case of no - serial - dependence, the decomposition ([eq5.1]) is satisfied by taking @xmath396. in addition, we require @xmath339 to be sub - gaussian to apply random matrix theories of @xcite. for instance, when @xmath339 is @xmath397, for any @xmath398, @xmath399, and thus condition (iii) is satisfied. finally, the _ almost surely _ condition of (iv) seems somewhat strong, but is still satisfied by bounded basis functions (e.g., fourier basis). we show in the supplementary material [@xcite] that when @xmath338 is diagonal (@xmath5 is cross - sectionally independent), both the sub - gaussian assumption and condition (iv) can be relaxed. the following theorem is the main result of this section. [th6.1] under assumptions of theorem [th4.1] and assumption [ass6.1], as @xmath400, if @xmath118 satisfies @xmath401 and @xmath402 (@xmath118 may either grow or stay constant), we have @xmath403
Numerical studies
this section presents numerical results to demonstrate the performance of projected - pca method for estimating loadings and factors using both real data and simulated data. we collected stocks in s&p 500 index constituents from crsp which have complete daily closing prices from year 2005 through 2013, and their corresponding market capitalization and book value from compustat. there are @xmath404 stocks in our data set, whose daily excess returns were calculated. we considered four characteristics @xmath42 as in @xcite for each stock : size, value, momentum and volatility, which were calculated using the data before a certain data analyzing window so that characteristics are treated known. see @xcite for detailed descriptions of these characteristics. all four characteristics are standardized to have mean zero and unit variance. note that the construction makes their values independent of the current data. we fix the time window to be the first quarter of the year 2006, which contains @xmath405 observations. given the excess returns @xmath406 and characteristics @xmath17 as the input data and setting @xmath407, we fit loading functions @xmath408 for @xmath409 using the projected - pca method. the four additive components @xmath280 are fitted using the cubic spline in the r package `` gam '' with sieve dimension @xmath410. all the four loading functions for each factor are plotted in figure [fig : gcurves]. the contribution of each characteristic to each factor is quite nonlinear., @xmath411 from financial returns of 337 stocks in s&p 500 index. they are taken as the true functions in the simulation studies. in each panel (fixed @xmath115), the true and estimated curves for @xmath412 are plotted and compared. the solid, dashed and dotted red curves are the true curves corresponding to the first, second and third factors, respectively. the blue curves are their estimates from one simulation of the calibrated model with @xmath413, @xmath414.] we now treat the estimated functions @xmath280 as the true loading functions, and calibrate a model for simulations. the `` true model '' is calibrated as follows : take the estimated @xmath280 from the real data as the true loading functions. for each @xmath6, generate @xmath366 from @xmath415 where @xmath416 is diagonal and @xmath417 sparse. generate the diagonal elements of @xmath416 from gamma(@xmath418) with @xmath419, @xmath420 (calibrated from the real data), and generate the off - diagonal elements of @xmath417 from @xmath421 with @xmath422, @xmath423. then truncate @xmath417 by a threshold of correlation @xmath424 to produce a sparse matrix and make it positive definite by r package `` nearpd. '' generate @xmath425 from the i.i.d. gaussian distribution with mean @xmath426 and standard deviation @xmath427, calibrated with real data. generate @xmath428 from a stationary var model @xmath429 where @xmath430. the model parameters are calibrated with the market data and listed in table [table : calibfactor]. finally, generate @xmath431. here @xmath432 is a @xmath433 correlation matrix estimated from the real data. @lccd2.4d2.4c@ & + 0.9076 & 0.0049 & 0.0230 & -0.0371 & -0.1226 & @xmath434 + 0.0049 & 0.8737 & 0.0403 & -0.2339 & 0.1060 & @xmath435 + 0.0230 & 0.0403 & 0.9266 & 0.2803 & 0.0755 & @xmath436 + by projected - pca (p - pca, red solid) and traditional pca (dashed blue) and @xmath437, @xmath438 by p - pca over 500 repetitions. left panel : @xmath439, right panel : @xmath440.] and @xmath441 over 500 repetitions, by projected - pca (p - pca, solid red) and traditional pca (dashed blue).] we simulate the data from the calibrated model, and estimate the loadings and factors for @xmath442 and @xmath443 with @xmath6 varying from @xmath444 through @xmath445. the `` true '' and estimated loading curves are plotted in figure [fig : gcurves] to demonstrate the performance of projected - pca. note that the `` true '' loading curves in the simulation are taken from the estimates calibrated using the real data. the estimates based on simulated data capture the shape of the true curve, though we also notice slight biases at boundaries. but in general, projected - pca fits the model well. we also compare our method with the traditional pca method [e.g., @xcite]. the mean values of @xmath446, @xmath447, @xmath448 and @xmath449 are plotted in figures [fig : calibg] and [fig : calibf] where @xmath450 [see section [design2] for definitions of @xmath451 and @xmath452. the breakdown error for @xmath453 and @xmath39 are also depicted in figure [fig : calibg]. in comparison, projected - pca outperforms pca in estimating both factors and loadings including the nonparametric curves @xmath36 and random noise @xmath39. the estimation errors for @xmath36 of projected - pca decrease as the dimension increases, which is consistent with our asymptotic theory. and @xmath454 over 500 repetitions. p - pca, pca and sls, respectively, represent projected - pca, regular pca and sieve least squares with known factors : design 2. here, @xmath328, so @xmath455. upper two panels : @xmath6 grows with fixed @xmath7 ; bottom panels : @xmath7 grows with fixed @xmath6.] and @xmath456 by projected - pca (solid red) and pca (dashed blue) : design 2. upper two panels : @xmath6 grows with fixed @xmath7 ; bottom panels : @xmath7 grows with fixed @xmath6.] consider a different design with only one observed covariate and three factors. the three characteristic functions are @xmath457 with the characteristic @xmath458 being standard normal. generate @xmath459 from the stationary var(1) model, that is, @xmath460 where @xmath461. we consider @xmath462. we simulate the data for @xmath442 or @xmath443 and various @xmath6 ranging from @xmath444 to @xmath445. to ensure that the true factor and loading satisfy the identifiability conditions, we calculate a transformation matrix @xmath199 such that @xmath463, @xmath464 is diagonal. let the final true factors and loadings be @xmath465, @xmath466. for each @xmath6, we run the simulation for @xmath445 times. we estimate the loadings and factors using both projected - pca and pc. for projected - pca, as in our theorem, we choose @xmath467, with @xmath468 and @xmath469. to estimate the loading matrix, we also compare with a third method : sieve - least - squares (sls), assuming the factors are observable. in this case, the loading matrix is estimated by @xmath470, where @xmath471 is the true factor matrix of simulated data. the estimation error measured in max and standardized frobenius norms for both loadings and factors are reported in figures [fig : simpleg] and [fig : simplef]. the plots demonstrate the good performance of projected - pca in estimating both loadings and factors. in particular, it works well when we encounter small @xmath7 but a large @xmath6. in this design, @xmath328, so the accuracy of estimating @xmath472 is significantly improved by using the projected - pca. figure [fig : simplef] shows that the factors are also better estimated by projected - pca than the traditional one, particularly when @xmath7 is small. it is also clearly seen that when @xmath6 is fixed, the improvement on estimating factors is not significant as @xmath7 grows. this matches with our convergence results for the factor estimator. it is also interesting to compare projected - pca with sls (sieve least - squares with observed factors) in estimating the loadings, which corresponds to the cases of unobserved and observed factors. as we see from figure [fig : simpleg], when @xmath6 is small, the projected - pca is not as good as sls. but the two methods behave similarly as @xmath6 increases. this further confirms the theory and intuition that as the dimension becomes larger, the effects of estimating the unknown factors are negligible. we now demonstrate the effectiveness of estimating @xmath86 by the projected - pc s eigenvalue - ratio method. the data are simulated in the same way as in design 2. @xmath442 or @xmath443 and we took the values of @xmath6 ranging from @xmath444 to @xmath445. we compare our projected - pca based on the projected data matrix @xmath137 to the eigenvalue - ratio test (ah) of @xcite and @xcite, which works on the original data matrix @xmath326. . p - pca and ah, respectively, represent the methods of projected - pca and @xcite. left panel : mean ; right panel : standard deviation.] for each pair of @xmath473, we repeat the simulation for @xmath443 times and report the mean and standard deviation of the estimated number of factors in figure [fig : estimatek]. the projected - pca outperforms ah after projection, which significantly reduces the impact of idiosyncratic errors. when @xmath413, we can recover the number of factors almost all the time, especially for large dimensions (@xmath474). on the other hand, even when @xmath475, projected - pca still obtains a closer estimated number of factors. we test the loading specifications on the real data. we used the same data set as in section [sec7.1], consisting of excess returns from 2005 through 2013. the tests were conducted based on rolling windows, with the length of windows spanning from 10 days, a month, a quarter and half a year. for each fixed window - length (@xmath7), we computed the standardized test statistic of @xmath321 and @xmath476, and plotted them along the rolling windows respectively in figure [fig : testing]. in almost all cases, the number of factors is estimated to be one in various combinations of @xmath477. figure [fig : testing] suggests that the semiparametric factor model is strongly supported by the data. judging from the upper panel [testing @xmath478, we have very strong evidence of the existence of nonvanishing covariate effect, which demonstrates the dependence of the market beta s on the covariates @xmath42. in other words, the market beta s can be explained at least partially by the characteristics of assets. the results also provide the theoretical basis for using projected - pca to get more accurate estimation. from 2006/01/03 to 2012/11/30. the dotted lines are @xmath479.] in the bottom panel of figure [fig : testing] (testing @xmath480), we see for a majority of periods, the null hypothesis is rejected. in other words, the characteristics of assets can not fully explain the market beta as intuitively expected, and model ([eq1.2]) in the literature is inadequate. however, fully nonparametric loadings could be possible in certain time range mostly before financial crisis. during 20082010, the market s behavior had much more complexities, which causes more rejections of the null hypothesis. the null hypothesis @xmath328 is accepted more often since 2012. we also notice that larger @xmath7 tends to yield larger statistics in both tests, as the evidence against the null hypothesis is stronger with larger @xmath7. after all, the semiparametric model being considered provides flexible ways of modeling equity markets and understanding the nonparametric loading curves.
Conclusions
this paper proposes and studies a high - dimensional factor model with nonparametric loading functions that depend on a few observed covariate variables. this model is motivated by the fact that observed variables can explain partially the factor loadings. we propose a projected - pca to estimate the unknown factors, loadings, and number of factors. after projecting the response variable onto the sieve space spanned by the covariates, the projected - pca yields a significant improvement on the rates of convergence than the regular methods. in particular, consistency can be achieved without a diverging sample size, as long as the dimensionality grows. this demonstrates that the proposed method is useful in the typical hdlss situations. in addition, we propose new specification tests for the orthogonal decomposition of the loadings, which fill the gap of the testing literature for semiparametric factor models. our empirical findings show that firm characteristics can explain partially the factor loadings, which provide theoretical basis for employing projected - pca method. on the other hand, our empirical study also shows that the firm characteristics can not fully explain the factor loadings so that the proposed generalized factor model is more appropriate.
Proofs for section3
throughout the proofs, @xmath45 and @xmath7 may either grow simultaneously with @xmath6 or stay constant. for two matrices @xmath481 with fixed dimensions, and a sequence @xmath482, by writing @xmath483, we mean @xmath484. in the regular factor model @xmath485, let @xmath486 denote a @xmath141 diagonal matrix of the first @xmath86 eigenvalues of @xmath487. then by definition, @xmath488. let @xmath489. then @xmath490 where @xmath491 still by the equality ([ea.1add]), @xmath501. hence, this step is achieved by bounding @xmath502 for @xmath503. note that in this step, we shall not apply a simple inequality @xmath504, which is too crude. instead, with the help of the result @xmath505 achieved in step 1, sharper upper bounds for @xmath502 can be achieved. we do so in lemma b.2 in the supplementary material [@xcite]. consider the singular value decomposition : @xmath508, where @xmath509 is a @xmath93 orthogonal matrix, whose columns are the eigenvectors of @xmath98 ; @xmath510 is a @xmath511 matrix whose columns are the eigenvectors of @xmath512 ; @xmath513 is a @xmath31 rectangular diagonal matrix, with diagonal entries as the square roots of the nonzero eigenvalues of @xmath98. in addition, by definition, @xmath253 is a @xmath141 diagonal matrix consisting of the largest @xmath86 eigenvalues of @xmath98 ; @xmath254 is a @xmath37 matrix whose columns are the corresponding eigenvectors. the columns of @xmath514 are the eigenvectors of @xmath88, corresponding to the first @xmath86 eigenvalues. by assumption [ass3.3], @xmath528, @xmath529 hence, @xmath530 by lemma b.1 in the supplementary material [@xcite], @xmath531. similarly, @xmath532 using the inequality that for the @xmath116th eigenvalue, @xmath533, we have @xmath534, for @xmath260. hence, it suffices to prove that the first @xmath86 eigenvalues of @xmath320 are bounded away from both zero and infinity, which are also the first @xmath535 eigenvalues of @xmath536. this holds under the theorem s assumption (assumption [ass3.1]). thus, @xmath537, which also implies @xmath523. fan, j., liao, y. and mincheva, m. (2013). large covariance estimation by thresholding principal orthogonal complements (with discussion). _ journal of the royal statistical society, series b _ * 75 * 603680. | this paper introduces a projected principal component analysis (projected - pca), which employs principal component analysis to the projected (smoothed) data matrix onto a given linear space spanned by covariates.
when it applies to high - dimensional factor analysis, the projection removes noise components.
we show that the unobserved latent factors can be more accurately estimated than the conventional pca if the projection is genuine, or more precisely, when the factor loading matrices are related to the projected linear space.
when the dimensionality is large, the factors can be estimated accurately even when the sample size is finite.
we propose a flexible semiparametric factor model, which decomposes the factor loading matrix into the component that can be explained by subject - specific covariates and the orthogonal residual component.
the covariates effects on the factor loadings are further modeled by the additive model via sieve approximations. by using the newly proposed projected - pca,
the rates of convergence of the smooth factor loading matrices are obtained, which are much faster than those of the conventional factor analysis.
the convergence is achieved even when the sample size is finite and is particularly appealing in the high - dimension - low - sample - size situation.
this leads us to developing nonparametric tests on whether observed covariates have explaining powers on the loadings and whether they fully explain the loadings.
the proposed method is illustrated by both simulated data and the returns of the components of the s&p 500 index.
./style / arxiv - general.cfg , | 1406.3836 |
Introduction
evolutionary algorithms (ea) are a group of computational techniques, which employ the theory of natural selection to a population of individuals to generate better individuals. genetic programming (gp) is a paradigm of ea which uses hierarchical, tree structure, variable length representation to code solutions of a problem. gp can be used to intelligently search the solution space for finding the optimal solution of a problem. there are many gp tools (lil - gp, genexprotools, gplab, ecj, open beagle) @xcite developed by gp practitioners. however, none of these address the following demands of end user before applying them to solve symbolic regression problems : (i) ease of use and (ii) small learning curve. many of these tools are open source and available freely (lil - gp, ecj, open beagle, gplab for matlab) @xcite whereas the rest are available commercially (genexprotools) @xcite. many of these tools necessitate modification in source code in order to generate required experimental environment. determining the final solution, produced by these tools, demands translation of the output or digging the log files. due to these reasons, the interest of researchers and engineers in gp may get reduced. theses reasons motivated us to develop our own gp framework @xcite which uses the postfix notation for an individual representation. we have considered the following features which need to be supported by postfix - gp framework : (i) easy to extend, (ii) simple and quick procedure for the configuration and running of gp, (iii) a set of algorithm implementation for : (a) generating the initial population, (b) selection mechanisms, and (c) genetic operators, (iv) visualization of : (a) postfix - gp run analysis and (b) evolved solution with statistical measures, (v) one - step and multi - step prediction support, (vi) visualization of results for one - step and multi - step predictions, and (vii) storage and retrieval of evolved solutions to and from file. postfix - gp has been used in experimental work @xcite, @xcite, @xcite, @xcite. this paper presents the design and implementation of postfix - gp, an object oriented software framework for genetic programming. section [sec : introduction to gp] gives introduction to gp. section [sec : postfix - gp] presents the design of postfix - gp. moreover, the section also gives the implementation details and main features of postfix - gp. section [sec : case study] presents postfix - gp as a solution modeling tool by solving the benchmark symbolic regression problem. section [sec : featurecomparison] compares the features of postfix - gp with lil - gp @xcite, ecj @xcite, and jclec @xcite frameworks. this is followed by conclusions in section [sec : conclusions].
Introduction to genetic programming
standard gp @xcite employs a variable length, tree structure scheme for an individual representation. the tree can be used to represent logical expressions (if - then - else), boolean expressions (and, or, not) or algebraic expressions. the symbolic regression aims to find the functional relationship (mathematical expression) between given instances of inputs - outputs. gp can be used to perform symbolic regression (sr). when using gp for solving symbolic regression problems, the user need to specify the following items : (i) gp configuration parameters, (ii) terminal set and function set, (iii) fitness function. the main steps of standard gp @xcite are as follows : 1. random generation of an initial population of candidate solutions in tree form using the elements of function set and terminal set, selected by the modeler (user). 2. calculating fitness value of every individual of the population on the given training dataset (fitness cases) 3. selecting parents for mating based on the calculated fitness values, determined in previous step 4. applying sub - tree crossover and mutation (genetic operators) on selected parents for generating a new population of individuals. the process is repeated until the termination condition is fulfilled.
Postfix-gp
the important features of the proposed postfix - gp are categorized into : (i) training dataset, function set, and terminal set related, (ii) gp parameters related, (iii) test dataset prediction related, (iv) gp run analysis related, and (v) serialization and de - serialization of gp experiments and results. training dataset, function set, and terminal set related features : * loading of training dataset * loading of binary and unary functions * loading of constants gp parameters related : * selection of method to generate initial population * configuration of gp parameters like population size, number of generations, crossover rate, mutation rate * selection of crossover and mutation type * selection of selection scheme test dataset prediction related : * loading out - of - sample test dataset for one - step predictions * visualization of results of one - step predictions with statistical measures * loading out - of - sample test dataset for multi - step predictions * visualization of results of multi - step predictions with statistical measures gp run analysis related : * plotting best adjusted fitness vs number of generations * plotting average adjusted fitness vs number of generations * plotting solution size vs number of generations serialization and de - serialization of gp experiments and results : * serialization of gp parameters, function set, terminal set, and obtained solutions to a file * de - serialization of gp parameters, function set, terminal set, and obtained solutions from a file this section presents the implementation details of postfix - gp. this will be useful to the reader in understanding and customizing the proposed postfix - gp framework. for details related to postfix - gp solution representation scheme, refer @xcite, @xcite. postfix - gp framework is developed using microsoft.net framework @xcite on windows xp operating system. the zedgraph @xcite class library is used for plotting the charts. zedgraph is an open source graph library for.net platform. + the class diagram for postfix - gp is depicted in [fig : classdiagram]. the classes can be grouped into following categories : (i) representation (,), (ii) population (), (iii) crossover operator (,,,), (iv) mutation operator (,,), (v) selection schemes (,,,), (vi) gp parameters (), and (vii) statistical analysis of results (). [cols="^ ",] * lil - gp * : lil - gp @xcite is a gp toolkit implemented in c programming language. the toolkit is efficient and fast, as it is implemented in c. however, it is difficult to extend the toolkit compare to other object - oriented implementations of gp systems. lil - gp uses only tree structure for an individual representation and provides limited fitness measures. moreover, the toolkit does not provide graphical user interface to read an input (training) data file. the toolkit uses a parameter file to load gp parameters. the toolkit produces six reporting files (.sys,.gen,.prg,.bst,.his and.stt) that provide the statistical information of the gp run. there are many patches developed by different researchers to fix the bugs and to improve the functionality of the basic lil - gp. * ecj * : ecj @xcite is a java based framework for evolutionary computation and genetic programming. ecj is designed using the object - oriented concepts. classes of ecj framework are divided into four layers @xcite : (i) utility layer, (ii) basic and custom evolutionary computation layer, (iii) basic and custom genetic programming layer, and (iv) problem layer. as the framework is implemented in java, it is slower in speed. moreover, ecj uses a tree (and not an arrays) to represent an individual, which requires dynamic memory allocation. thus, the framework consumes large memory. ecj determines gp parameters from a parameter file. ecj determines classes to be loaded, the type of problem to be solved, the type of technique to use to solve the problem, and the way to report the statistical results of the run from the parameter file at the run time @xcite. this provides easy to extend functionality to the ecj. ecj stores statistical information of gp run in a text file. moreover, it provides flexibility to produce the extra output files through class customization but does not provide a gui to visualize this information. * jclec * : jclec @xcite is a java based framework for evolutionary computation and genetic programming. jclec @xcite is designed using the object - oriented concepts. the classes of jclec framework are divided in three layers : (i) system core, (ii) experiments runner (reads an ea script file, execute all indicated algorithms and produce report files), and (iii) genlab (a gui on the top of experiments runner and system core layers, provides functionality to edit the experiment files and to view the gp run results) @xcite. gp parameters can be set by the user either through genlab gui or through an xml (configuration) file. however, the structure of configuration files is not user friendly. the framework provides the gui to visualize statistical information of gp run. the jclec framework is easy to extend.
Conclusions
this paper presented the design and implementation of postfix - gp framework. the implementation details of postfix - gp, including an individual representation, different crossover operators, mutations, and selection mechanism were also presented. postfix - gp provides user interactive gui for performing different activities. the user can load training dataset, function set, and constants. the user can set the gp parameters through gui. the evolved solutions with their statistical measures can be visualized through gui. moreover, the user can also perform one - step and multi - step predictions using gui. the evolved solutions can be stored in binary format and can be retrieved later on. the user can also analyze postfix - gp run through gui. postfix - gp as a solution modeling tool was presented by solving symbolic regression problem. postfix - gp addresses the requirements of ease of use and small learning curve before utilizing it to solve the problems. it was developed to minimize the user s time required to set up and run gp experiments. c. gagn and m. parizeau, `` open beagle : a new c++ evolutionary computation framework, '' in _ gecco 2002 : proceedings of the genetic and evolutionary computation conference_.1em plus 0.5em minus 0.4em new york : morgan kaufmann publishers, 9 - 13 july 2002, p. 888. [online]. available : http://www.cs.ucl.ac.uk/staff/w.langdon/ftp/papers/gecco2002/gecco-2002-15.pdf v. dabhi and s. chaudhary, `` semantic sub - tree crossover operator for postfix genetic programming, '' in _ proceedings of seventh international conference on bio - inspired computing : theories and applications (bic - ta 2012) _, ser. advances in intelligent systems and computing, j. c. bansal, p. k. singh, k. deep, m. pant, and a. k. nagar, eds. 201.1em plus 0.5em minus 0.4emspringer india, 2013, pp. 391402. , `` time series modeling and prediction using postfix genetic programming, '' in _ advanced computing communication technologies (acct), 2014 fourth international conference on _, feb 2014, pp. 307314. x. li, c. zhou, w. xiao, and p. c. nelson, `` prefix gene expression programming, '' in _ late breaking paper at genetic and evolutionary computation conference (gecco2005) _, washington, d.c., usa, 25 - 29 jun. 2005, pp. 2531. j. h. holland, _ adaptation in natural and artificial systems : an introductory analysis with applications to biology, control and artificial intelligence_.1em plus 0.5em minus 0.4emcambridge, usa : mit press, 1992. | this paper describes postfix - gp system, postfix notation based genetic programming (gp), for solving symbolic regression problems.
it presents an object - oriented architecture of postfix - gp framework.
it assists the user in understanding of the implementation details of various components of postfix - gp.
postfix - gp provides graphical user interface which allows user to configure the experiment, to visualize evolved solutions, to analyze gp run, and to perform out - of - sample predictions.
the use of postfix - gp is demonstrated by solving the benchmark symbolic regression problem.
finally, features of postfix - gp framework are compared with that of other gp systems.
+ * _ keywords- _ * postfix genetic programming ; postfix - gp framework ; object oriented design ; gp software tool ; symbolic regression | 1507.01687 |
Introduction
several diagnostic protocols are usually adopted by dermatologists for analyzing and classifying skin lesions, such as the so - called _ abcd - rule _ of dermoscopy @xcite. due to the subjective nature of examination, the accuracy of diagnosis is highly dependent upon human vision and dermatologist s expertise. computerized dermoscopic image analysis systems, based on a consistent extraction and analysis of image features, do not have the limitation of this subjectivity. these systems involve the use of a computer as a second independent and objective diagnostic method, which can potentially be used for the pre - screening of patients performed by non - experienced operators. although computerized analysis techniques can not provide a definitive diagnosis, they can improve biopsy decision - making, which some observers feel is the most important use for dermoscopy @xcite. recently, numerous researches on this topic propose systems for the automated detection of malignant melanoma in skin lesions (e.g., @xcite). in our previous study on dermoscopic images @xcite, the segmentation of the skin area and the lesion area was achieved by a semi - automatic process based on otsu algorithm @xcite, supervised by a human operator. here, we propose a full automatic segmentation method consisting of three main steps : selection of the image roi, selection of the segmentation band, and segmentation. the paper is organized as follows. in section [proposedapproach] we describe the proposed algorithm, providing details of its main steps. in section [expres] we provide a thorough analysis of experimental results on the isic 217 dataset @xcite. conclusions are drawn in section [conclusioni].
Sdi algorithm
the block diagram of the segmentation algorithm proposed for dermoscopic images, named sdi algorithm, is shown in fig. [fig : overall]. the three main steps are described in the following. in order to achieve an easier and more accurate segmentation of the skin lesion, it is advisable to select the region of interest (roi), i.e., the subset if image pixels that belong to either the lesion or the skin. this region excludes image pixels belonging to (usually dark) areas of the image border and/or corners, as well as those belonging to hair, that will not be taken into account in the subsequent steps of the sdi algorithm. in the proposed approach, the value band of the image in the hsv color space is chosen in order to select dark image pixels ; these are excluded from the roi if they cover most of the border or the angle regions of the image. concerning hair, many highly accurate methods have been proposed in the literature @xcite. here, we adopted a bottom - hat filter in the red band of the rgb image. an example of the roi selection process is reported in fig. [fig : roiselection] for the isic 2017 test image no. 15544. here, we observe that the wide dark border on the left of the image, as well as the dark hair over the lesion, have properly been excluded from the roi mask. [cols="^,^ ",]
Conclusions
we proposed the sdi algorithm for dermoscopic image segmentation, consisting of three main steps : selection of the image roi, selection of the segmentation band, and segmentation. the reported analysis of experimental results achieved by the sdi algorithm on the isic 2017 dataset allowed us to highlight its pro s and con s. this leads us to conclude that, although some accurate results can be achieved, there is room for improvements in different directions, that we will go through in future investigations.
Acknowledgments
this research was supported by lab gtp project, funded by miur. w. stolz, a. riemann, a. b. cognetta, l. pillet, w. abmayr, d. holzel, p. bilek, f. nachbar, m. landthaler, and o. braun - falco, `` abcd rule of dermoscopy : a new practical method for early recognition of malignant melanoma, '' _ european journal of dermatology _, vol. 4, pp. 521527, 1994. m. burroni, r. corona, g. delleva, f. sera, r. bono, p. puddu, r. perotti, f. nobile, l. andreassi, and p. rubegni, `` melanoma computer aided diagnosis : reliability and feasibility study, '' _ clinical cancer research _, vol. 10, pp. 18811886, 2004. m. e. celebi, h. a. kingravi, b. uddin, h. iyatomi, y. a. aslandogan, w. v. stoecker, and r. h. moss, `` a methodological approach to the classification of dermoscopy images. '' _ computerized medical imaging and graphics _, vol. 31, no. 6, pp. 362373, september 2007. i. maglogiannis and c. n. doukas, `` overview of advanced computer vision systems for skin lesions characterization, '' _ ieee transactions on information technology in biomedicine _, vol. 13, no. 5, pp. 721733, 2009. v. cozza, m. r. guarracino, l. maddalena, and a. baroni, `` dynamic clustering detection through multi - valued descriptors of dermoscopic images, '' _ statistics in medicine _, 30, no. 20, pp. 25362550, 2011. [online]. available : http://dx.doi.org/10.1002/sim.4285 m. e. celebi, q. wen, h. iyatomi, k. shimizu, h. zhou, and g. schaefer, `` a state - of - the - art survey on lesion border detection in dermoscopy images, '' in _ dermoscopy image analysis _, m. e. celebi, t. mendonca, and j. s. marques, eds.1em plus 0.5em minus 0.4emcrc press, 2015, pp. | we propose an automatic algorithm, named sdi, for the segmentation of skin lesions in dermoscopic images, articulated into three main steps : selection of the image roi, selection of the segmentation band, and segmentation.
we present extensive experimental results achieved by the sdi algorithm on the lesion segmentation dataset made available for the isic 2017 challenge on skin lesion analysis towards melanoma detection, highlighting its advantages and disadvantages. | 1703.03186 |
Introduction
the methods used to calculate the energy and the lifetime of a resonance state are numerous @xcite and, in some cases, has been put forward over strong foundations @xcite. however, the analysis of the numerical results of a particular method when applied to a given problem is far from direct. the complex scaling (complex dilatation) method @xcite, when the hamiltonian @xmath5 allows its use, reveals a resonance state by the appearance of an isolated complex eigenvalue on the spectrum of the non - hermitian complex scaled hamiltonian, @xmath6 @xcite. of course in an actual implementation the rotation angle @xmath7 must be large enough to rotate the continuum part of the spectrum beyond the resonance s complex eigenvalue. moreover, since most calculations are performed using finite variational expansions it is necessary to study the numerical data to decide which result is the most accurate. to worsen things the variational basis sets usually depend on one (or more) non - linear parameter. for bound states the non - linear parameter is chosen in order to obtain the lowest variational eigenvalue. for resonance states things are not so simple since they are embedded in the continuum. the complex virial theorem together with some graphical methods @xcite allows to pick the best numerical solution of a given problem, which corresponds to the stabilized points in the @xmath7 trajectories @xcite. other methods to calculate the energy and lifetime of the resonance, based on the numerical solution of complex hamiltonians, also have to deal with the problem of which solutions (complex eigenvalues) are physically acceptable. for example, the popular complex absorbing potential method, which in many cases is easier to implement than the complex scaling method, produces the appearance of nonphysical complex energy stabilized points that must be removed in order to obtain only the physical resonances @xcite. the aforementioned issues explain, at some extent, why the methods based only in the use of real @xmath2 variational functions are often preferred to analyze resonance states. these techniques reduce the problem to the calculation of eigenvalues of real symmetric matrices @xcite. of course, these methods also have its own drawbacks. one of the main problems was recognized very early on (see, for example, the work by hol@xmath8ien and midtdal @xcite) : if the energy of an autoionizing state is obtained as an eigenvalue of a finite hamiltonian matrix, which are the convergence properties of these eigenvalues that lie in the continuum when the size of the hamiltonian matrix changes? but in order to obtain resonance - state energies it is possible to focus the analysis in a global property of the variational spectrum : the density of states (dos)@xcite, being unnecessary to answer this question. the availability of the dos allows to obtain the energy and lifetime of the resonance in a simple way, both quantities are obtained as least square fitting parameters, see for example @xcite. despite its simplicity, the determination of the resonance s energy and width based in the dos is far from complete. there is no a single procedure to asses both, the accuracy of the numerical findings and its convergence properties, or which values to pick between the several `` candidates '' that the method offers @xcite. recently, pont _ et al _ @xcite have used _ finite size scaling _ arguments @xcite to analyze the properties of the dos when the size of the hamiltonian changes. they presented numerical evidence about the critical behavior of the density of states in the region where a given hamiltonian has resonances. the critical behavior was signaled by a strong dependence of some features of the density of states with the basis - set size used to calculate it. the resonance energy and lifetime were obtained using the scaling properties of the density of states. however, the feasibility of the method to calculate the resonance lifetime laid on the availability of a known value of the lifetime, making the whole method dependent on results not provided by itself. the dos method relies on the possibility to calculate the ritz - variational eigenfunctions and eigenvalues for many different values of the non - linear parameter @xmath9 (see kar and ho @xcite). for each basis - set size, @xmath3, used, there are @xmath3 variational eigenvalues @xmath10. each one of these eigenvalues can be used, at least in principle, to compute a dos, @xmath11, resulting, each one of these dos in an approximate value for the energy, @xmath12, and width, @xmath13, of the resonance state of the problem. if the variational problem is solved for many different basis - set sizes, there is not a clear cut criterion to pick the `` better '' result from the plethora of possible values obtained. this issue will be addressed in section [model]. in this work, in order to obtain resonance energies and lifetimes, we calculate all the eigenvalues for different basis - set sizes, and present a recipe to select adequately certain values of @xmath3, and one eigenvalue for each @xmath3 elected, that is, we get a series of variational eigenvalues @xmath14. the recipe is based on some properties of the variational spectrum which are discussed in section [some - properties]. the properties seem to be fairly general, making the implementation of the recipe feasible for problems with several particles. actually, because we use scaling properties for large values of @xmath3, the applicability of the method for systems with more than three particles could be restricted because the difficulties to handle large basis sets. the set of approximate resonance energies, obtained from the density of states of a series of eigenvalues selected following the recipe, shows a very regular behaviour with the basis set size. this regular behaviour facilitates the use of finite size scaling arguments to analyze the results obtained, in particular the extrapolation of the data when @xmath15. the extrapolated values are the most accurate approximation for the parameters of the resonance state that we obtain with our method. this is the subject of section [recipe], where we present results for models of one and two particles. following the same prescription to choose particular solutions of the variational problem we obtain a set of approximate widths in section [golden - rule]. using the scaling function that characterizes the behaviour of the approximate energies as a guide, it is possible to find a very good approximation to the resonance width since, again, the data generated using our prescription seems to converge when @xmath4. finally, in section [discusion] we summarize and discuss our results.
Some properties of the variational spectrum
when one is dealing with the variational spectrum in the continuum region, some of its properties are not exploited to obtain more information about the presence of resonances, usually the focus of interest is the stabilization of the individual eigenvalues. the stabilization is achieved varying some non - linear variational parameter. if @xmath9 is the inverse characteristic decaying length of the variational basis functions, then the spectrum of the kinetic energy scales as @xmath16, moreover, for potentials that decay fast enough, the spectrum of the whole hamiltonian _ also _ scales as @xmath16 for large (or small) enough values of @xmath9 (see appendix). this is so, since the variational eigenfunctions are @xmath2 approximations to plane waves _ except _ when @xmath9 belongs to the stabilization region. when @xmath9 belong to the stabilization region of a given variational eigenvalue, say @xmath17, then @xmath18 (where @xmath0 is the resonance energy) and the variational eigenfunction @xmath19 has the localization length of the potential well. we intend to take advantage of the changes of the spectrum when @xmath9 goes from small to large enough values. the variational spectrum satisfies the hylleras - undheim theorem or variational theorem : if @xmath3 is the basis set size, and @xmath20 is the @xmath21 eigenvalue obtained with a variational basis set of size @xmath3, then @xmath22 actually, since the threshold of the continuum is an accumulation point, then for small enough values of @xmath9 and a given @xmath23 there is always a @xmath24 such that @xmath25 for the kinetic energy variational eigenvalues, and for fixed @xmath26, and @xmath27, if the ordering given by equation ([ordering]) holds for some @xmath9 then it is true for all @xmath9. of course this is not true for a hamiltonian with a non zero potential that support resonance states. so, we will take advantage of the variational eigenvalues such that for @xmath9 small enough satisfies equation ([ordering]) but, for @xmath9 large enough @xmath28 despite its simplicity, the arguments above give a complete prescription to pick a set of eigenstates that are particularly affected by the presence of a resonance. choose @xmath3 and @xmath29 arbitrary, and then look for the smaller values of @xmath30 and @xmath27 such that the two inequalities, equations. ([ordering]) and ([reverse]) are fulfilled. so far, all the examples analyzed by us show that if the inequalities are satisfied for some @xmath30 and @xmath27 then they are satisfied too by the eigenvalues @xmath31, for @xmath32.
Models and methods
to illustrate how our prescription works we used two different model hamiltonians. the first model, due to hellmann @xcite, is a one particle hamiltonian that models a @xmath33-electron atom. the second one is a two particle model that has been used to study the low energy and resonance states of two electrons confined in a semiconductor quantum dot @xcite. the details of the variational treatment of both models will be kept as concise as possible. the one particle model has been used before for the determination of critical nuclear charges for @xmath33-electron atoms @xcite, it also gives reasonable results for resonance states in atomic anions @xcite and continuum states @xcite. the interaction of a valence electron with the atomic core is modeled by a one - particle potential with two asymptotic behaviours. the potential behaves correctly in the regions where electron is far from the atomic core (@xmath34 electrons and the nucleus of charge @xmath35) and when it is near the nucleus. the hamiltonian, in scaled coordinates @xmath36, is @xmath37 where @xmath38 and @xmath39 is a range parameter that determines the transition between the asymptotic regimes, for distances near the nucleus @xmath40 and in the case @xmath41 the nucleus charge is screened by the @xmath34 localized electrons and @xmath42. another advantage of the potential comes from its analytical properties. in particular this potential is well behaved and the energy of the resonance states can be calculated using complex scaling methods. so, besides its simplicity, the model potential allows us to obtain the energy of the resonance by two independent methods and check our results. the two particle model that we considered describes two electrons interacting via the coulomb repulsion and confined by an external potential with spherical symmetry. we use a short - range potential suitable to apply the complex scaling method. the hamiltonian @xmath5 for the system is given by @xmath43 where @xmath44, @xmath45 the position operator of electron @xmath46 ; @xmath47 and @xmath48 determine the range and depth of the dot potential. after re - scaling with @xmath47, in atomic units, the hamiltonian of equation ([hamiltoniano]) can be written as @xmath49 where @xmath50. the variational spectrum of the two particle model, equation ([hamil2part]), and all the necessary algebraic details to obtain it, has been studied with great detail in reference @xcite so, until the end of this section, we discuss the variational solution of the one particle model given by equation ([hamil]). the discrete spectrum and the resonance states of the model given by equation ([hamil]) can be obtained approximately using a real @xmath51 truncated basis set @xmath52 to construct a @xmath53 hamiltonian matrix @xmath54. we use the rayleigh - ritz variational method to obtain the approximations @xmath55 @xmath56 for bound states this functions are variationally optimal. the functions @xmath57 are @xmath58 and @xmath59 are the associated laguerre polynomials of @xmath60 order and degree @xmath61. the non - linear parameter @xmath9 is used for eigenvalue stabilization in resonance analysis @xcite. note that @xmath9 plays a similar role that the finite size of the box in spherical box stabilization procedures @xcite, as stated by kar _ et. @xcite. resonance states are characterized by isolated complex eigenvalues, @xmath62, whose eigenfunctions are not square - integrable. these states are considered as quasi - bound states of energy @xmath0 and inverse lifetime @xmath1. for the hamiltonian equation ([hamil]), the resonance energies belong to the positive energy range @xcite. using the approximate solutions of hamiltonian ([hamil]) we analyze the dos method @xcite that has been used extensively to calculate the energy and lifetime of resonance states, in particular we intend to show that 1) the dos method provides a host of approximate values whose accuracy is hard to assess, and 2) if the dos method is supplemented by a new optimization rule, it results in a convergent series of approximate values for the energy and lifetime of resonance states. the dos method relies on the possibility to calculate the ritz - variational eigenfunctions and eigenvalues for many different values of the non - linear parameter @xmath9 (see kar and ho @xcite). the localized dos @xmath63 can be expressed as @xcite @xmath64 since we are dealing with a numerical approximation, we calculate the energies in a discretization @xmath65 of the continuous parameter @xmath9. in this approximation, equation ([densidad_sin_suma]) can be written as @xmath66 where @xmath67 is the @xmath30-th eigenvalue of the @xmath53 matrix hamiltonian with @xmath68 and @xmath39 fixed. in complex scaling methods the hamiltonian is dilated by a complex factor @xmath69. as was pointed out long time ago by moiseyev and coworkers @xcite, the role played by @xmath9 and @xmath70 are equivalent, in fact, our parameter @xmath9 corresponds to @xmath71. besides, the dos attains its maximum at optimal values of @xmath9 and @xmath0 that could be obtained with a self - adjoint hamiltonian without using complex scaling methods @xcite. so, locating the position of the resonance using the maximum of the dos is equivalent to the stabilization criterion used in complex dilation methods that requires the approximate fulfillment of the complex virial theorem @xcite. the values of @xmath72 and @xmath73 are obtained performing a nonlinear fitting of @xmath63, with a lorentzian function, @xmath74}.\]] one of the drawbacks of this method results evident : for each pair @xmath75 there are several @xmath11, and since each @xmath11 provides a value for @xmath76 and @xmath77 one has to choose which one is the best. kar and ho @xcite solve this problem fitting all the @xmath11 and keeping as the best values for @xmath0 and @xmath1 the fitting parameters with the smaller @xmath78 value. at least for their data the best fitting (the smaller @xmath78) usually corresponds to the larger @xmath30. this fact has a clear interpretation, if the numerical method approximates @xmath0 with some @xmath79, where @xmath3 is the basis set size of the variational method, a large @xmath30 means that the numerical method is able to provide a large number of approximate levels, and so the continuum of positive - energy states is `` better '' approximated. in a previous work @xcite we have shown that a very good approximation to the energy of the resonance state is obtained considering just the energy value where @xmath11 attains its maximum. we denote this value as @xmath76. figure [prefig1] shows the approximate resonance energy @xmath76 for different basis set size @xmath3, where @xmath29 is the index of the variational eigenvalue used to calculate the dos. we used the values @xmath80 and @xmath81 corresponding to the ones used before @xcite in the analysis of @xmath82 resonances. the figure [prefig1] also shows the value calculated using complex scaling. it is clear that the accuracy of all the values shown is rather good (all the values shown differ in less than 6@xmath83), and that larger values on @xmath84 provide better values for the resonance energy. these facts are well known from previous works, _ i.e. _ almost all methods to calculate the energy of the resonance give rather stable and accurate results for @xmath0. however, the practical importance of this fact is reduced : these are uncontrolled methods, so the accuracy of the values obtained from the dos can not be assessed (without a value independently obtained) and these values do not seem to converge to the value obtained using complex scaling when @xmath3 is increased and @xmath84 is kept fixed. there is another fact that potentially could render the whole method useless : for small or even moderate @xmath29, the values @xmath76 become _ unstable _ (see figure [prefig1]) when @xmath3 is large enough. this last point has been pointed previously @xcite. in the problem that we are considering is rather easy to obtain a large number of variational eigenvalues in the interval where the resonances are located, allowing us to calculate @xmath76 up to @xmath85, but this situation is far from common see, for example, references @xcite.
Scaling of the resonance energy
so far we have presented only results about the behaviour of the one particle hamiltonian, from now on we will discuss both models, equations ([hamil]) and [hamil2part]. it is known that the variational eigenvalues @xmath17 do not present crossings when they are calculated for some fixed values of @xmath3, _ i.e _ the variational spectrum is non - degenerate for any finite hamiltonian matrix. as a matter of fact the avoided crossings between successive eigenvalues in the variational spectrum are the watermark of a resonance. an interesting feature emerges when the variational spectrum for many different basis set sizes @xmath3 are plotted together versus the parameter @xmath9. besides the places where @xmath86 attains its minimum value, which correspond to the stabilization points, there are some gaps which correspond to crossings between eigenvalues obtained with different basis set sizes, see figure [prefig2]. moreover, the crossings corresponds to eigenvalues with different index @xmath29, and are the states that satisfy the inequalities equations ([ordering]),and ([reverse]). it is worth to remark that the main features shown by figure [prefig2] are independent of the number of particles of the hamiltonian and the particular values of the threshold of the continuum. figure [bundle2p] shows the behaviour of the variational eigenvalues obtained for the two particle hamiltonian equation ([hamil2part]). in this case the ionization threshold is not the asymptotic value of the potential, but it is given by the energy of the one particle ground state. the resonance state came from the two - particle ground state that becomes unstable and enters into the continuum of states when the quantum dots becomes `` too small '' to accommodate two electrons. for more details about the model, see reference @xcite. the left panel of figure [prefig3] shows the behaviour of the maximum value of the dos, @xmath87, for the one particle hamiltonian, obtained for different basis - set sizes and fixed @xmath29 (in this case @xmath85), and the @xmath88 obtained choosing a `` bundle '' of states that are linked by a crossing, these states have @xmath89 and @xmath90 respectively. from our numerical data, the maximum value of the dos scales with the basis - set size following two different prescriptions. for @xmath29 fixed, @xmath91, with @xmath92, while when the pair @xmath93 is chosen from the set of pairs that label a bundle of states @xmath94, with @xmath95. in particular, for @xmath85 we get that @xmath96, and @xmath97 when @xmath98. of course we can pick sets of states that are not related by a crossing. for instance, we also picked sets with a simple prescription as follows : choose a given initial pair @xmath99 and form a set of states with the states labeled by @xmath100 and so on. figure [prefig3](a) shows two examples obtained choosing @xmath101, @xmath102 and @xmath103 and @xmath104 both with @xmath105. quite interestingly, the data in figure [prefig3] show that the scaled maxima of the dos for a bundle and two different sets seem to converge to the _ same _ value when @xmath106, but only for the bundle the scaling function is @xmath107. the advantage obtained from picking those eigenvalues @xmath17 that belong to a given bundle is still more evident when the corresponding dos and @xmath0 are calculated. the right panel of figure [prefig3] shows the @xmath0 obtained from the dos whose maxima are shown in the left panel. it is rather evident that these values now seem to converge, besides, the extrapolation to @xmath4 results in a more accurate approximate value for @xmath0. in contradistinction, the values for @xmath0 corresponding to a fixed index @xmath29 (the values shown in the figure [prefig3] correspond to @xmath85) do not seem to converge anywhere close to the value obtained using complex rotation. figure [er2par] shows the resonance energies obtained from the bundles of states shown in figure [bundle2p] for the two - particle model. since the numerical solution of this model is more complicated than the solution of the one - particle model the number of approximate values is rather reduced. however, it seems that the data also supports a linear scaling of @xmath76 with @xmath108.
Scaling of the resonance width
many real algebra methods to calculate resonance energies use a golden rule - like formula to calculate the resonance width. in this section we will use the formula and stabilization procedure proposed by tucker and truhlar @xcite that we will describe briefly for completeness. this projection formula seems to work better for one - particle models. for two - particle models its utility has been questioned @xcite, so to analyze the width of the resonance states of the quantum dot model we fitted the corresponding dos using equation ([lorentz]). the method of tucker and truhlar @xcite is implemented by the following steps. choose a basis @xmath109 where @xmath9 is a non - linear parameter. diagonalize the hamiltonian using up to @xmath3 functions of the basis. look for the stabilization value @xmath110 and its corresponding eigenfunction @xmath111 which are founded for some value @xmath112. define the projector @xmath113 where @xmath114 is the normalized projection of @xmath111 onto the basis @xmath115 for any other @xmath9. diagonalize the hamiltonian @xmath116 in the basis @xmath115, again as a function of @xmath9, and find a value @xmath117 of @xmath9 such that @xmath118 where @xmath119 denotes eigenvalue @xmath84 of the projected hamiltonian for the scale factor @xmath117, and @xmath120 is the corresponding eigenfunction. with the previous definitions and quantities, the resonance width @xmath1 is given by @xmath121 where @xmath122.\]] despite some useful insights, the procedure sketched above does not determine all the intervening quantities, for instance there are many solutions to equation ([seg - estabili]) and, of course, the stabilization method provides several good candidates for @xmath111 and @xmath112. we are able to avoid some of the indeterminacies associated to the tucker and truhlar procedure using a bundle of states associated to a crossing, so @xmath111 and @xmath112 are given by any of the eigenfunctions associated to a bundle and @xmath123 comes from the stabilization procedure. then we construct projectors @xmath124 where @xmath19 is one of the variational eigenstates that belong to a bundle of states. with the projectors @xmath125 we construct hamiltonians @xmath126, and find the solutions to the problem @xmath127 since there is not an a priori criteria to choose one particular solution of equation ([tercera - estabili]) we show our numerical findings for several values of @xmath84. figure [prefig4] shows the behaviour of the resonance width calculated with equation ([golden2]), where we have used @xmath128 as @xmath111 and @xmath129, where @xmath130 and @xmath131. despite that the different sets corresponding to different values of @xmath84 do not converge to any definite value, for @xmath3 large enough all the sets scale as @xmath132, with @xmath133. since the resonance energy scales as @xmath107, at least when a bundle of states with a crossing is chosen to calculate approximations (see figure [prefig3]), we suggest that the right scaling for @xmath1 is given by @xmath134. of course for a given basis size, particular variational functions, stabilization procedures and so on, we can hardly expect to find a proper set of @xmath1 whose scaling law would be @xmath107. instead of this we propose that the data in the right panel can be fitted by @xmath135 then the best approximation for the resonance width is obtained fitting the curve and selecting the @xmath136 as @xmath137 for @xmath138 the closest value to one. as pointed in reference @xcite, the projection technique to calculate the width of a resonance can be implemented if a suitable form of the projection operator can be found. as this procedure is marred by several issues we used the dos method to obtain the approximate widths of a resonance state of the two particle model. figure [gama2par] shows the widths calculated associated to the energies shown in figure [er2par], the parameters of the hamiltonian are exactly the same. there is no obvious scaling function that allows the extrapolation of the data but, even for moderate values of @xmath3, it seems as the data converge to the value obtained using complex scaling.
Conclusions
in this work we analyzed the convergence properties of real @xmath51 basis - set methods to obtain resonance energies and lifetimes. the convergence of the energy with the basis - set size for bound states is well understood, the larger the basis set the better the results and these methods converge to the exact values for the basis - set size going to infinite (complete basis set). this idea is frequently applied to resonance states. the increase of the basis - set size in some commonly used methods does not improve the accuracy of the value obtained for the resonance energy @xmath0, as showed in figure [prefig1]. this undesirable behavior comes from the fact that the procedure is not variational as in the case of bound states. moreover, the exact resonance eigenfunction does not belong to the hilbert space expanded by the complete basis set. in this work we presented a prescription to pick a set or bundle of states that has linear convergence properties for small width resonances. this procedure is robust because the choice of different bundles results in very similar convergence curves and energy values. in fact, in the method described here, the pairs @xmath93 of the bundles play the role of a second stabilization parameter together with the variational parameter @xmath9. of a second we tested the method in others one and two particle systems and the general behavior of them is the same. the results are very good in all cases leading to an improvement in the calculation of the resonance energies. nevertheless we have to note that the method could no be applied in cases where two or more resonance energies lie very close because the overlapping bundles. the lifetime calculation is more subtle. the use of golden - rule - like formulas, as we applied here, always give several possible outcomes for the width @xmath139, corresponding to different pseudo - continuum states @xmath140. the projection technique, equation ([golden2]), is not the exception and it is not possible to select _ a priori _ which value of @xmath141 is the most accurate. the linear convergence of the dos with basis - set size suggests that the scaling in the lifetime value, in accordance with the energy scaling, should be linear. regrettably, the projection method gives discrete sets of values which can not be tuned to obtain an exact linear convergence. our recipe is to choose the set @xmath139 whose scaling is closest to the linear one, then the best estimation for the resonance width is obtained from extrapolation. many open questions remain on the analysis of the different convergence properties of resonance energy and lifetime. the method presented here to obtain the resonance energy from convergence properties works very well, but the appearance of bundles in the spectrum is not completely understood. even there is not a rigorous proof, the numerical evidence supports the idea that the behaviour of the systems studied here is quite general.
Appendix
in this appendix we give arguments that support our assumptions on the scaling of the eigenenergies with the basis - set parameter @xmath9. we present our argument for one body hamiltonians, but it is straightforward to generalize to more particles with pair interactions decaying fast enough at large distances. 1. let @xmath142 be an @xmath143 matrix with all its matrix elements having the form @xmath144, where @xmath145, then if @xmath146 the eigenvalues of @xmath142 scales with @xmath147 : @xmath148=0\;\rightarrow det[a(1)-\frac{\lambda(\eta)}{f(\eta)}\,i]=0 \rightarrow \lambda(1)=\frac{\lambda(\eta)}{f(\eta) } \,.\]] 2. let @xmath149 @xmath143 be symmetric matrices with @xmath150, and @xmath151 the eigenvalues of @xmath149 and @xmath152 respectively, in nondecreasing order, then, by the minimax principle @xcite consider a spherical one - particle potential with compact support, @xmath154 if @xmath155, and finite, @xmath156 (both conditions could be relaxed, but we adopt them for simplicity). let the basis - set functions be of the form @xmath157 with @xmath158 and @xmath159, then the coefficients take the form @xmath160. and then, by equation ([asa]), all the eigenvalues of the kinetic energy have the same scaling with @xmath16. we have to show that, in both limits, @xmath162 and @xmath163, for all the potential matrix elements hold @xmath164, and then, by the wielandt - hoffman theorem @xcite, the eigenenergies are a perturbation of the eigenvalues of the kinetic energy. a u hazi and h s taylor, phys. rev. a * 14 *, 2071 (1976). a ferrn, o osenda and p serra, phys. a * 79 *, 032509 (2009). h feschbach, ann. ny * 19 *, 627 (1962) n moiseyev, c corcoran, phys. rev. a * 20 *, 814 (1979). m e fisher, in _ critical phenomena _, proceedings of the 51st enrico fermi summer school, varenna, italy, m. s. green, ed. (academic press, new york 1971). s e tucker and d. g. truhlar, j.chem. * 86 *, 6251 (1987). f r manby and g doggetty, j. phys. b : at. mol.. phys. * 30 *, 3343 (1997) j. h. wilkinson, _ the algebraic eigenvalue problem _, oxford university press, london, (1965). | the resonance states of one- and two - particle hamiltonians are studied using variational expansions with real basis - set functions.
the resonance energies, @xmath0, and widths, @xmath1, are calculated using the density of states and an @xmath2 golden rule - like formula.
we present a recipe to select adequately some solutions of the variational problem.
the set of approximate energies obtained shows a very regular behaviour with the basis - set size, @xmath3.
indeed, these particular variational eigenvalues show a quite simple scaling behaviour and convergence when @xmath4. following the same prescription to choose particular solutions of the variational problem
we obtain a set of approximate widths. using the scaling function that characterizes the behaviour of the approximate energies as a guide
, it is possible to find a very good approximation to the actual value of the resonance width. | 1012.3177 |
Acknowledgments
we would like to thank d. jouan and zhang xiaofai for helpful discussions. this work is partly supported by the national natural science founction of china. 35 t. matsui and h. satz, phys. b178 *, 416 (1986). na38 collaboration, c. baglin et al., phys. lett. * b220 *, 471 (1989) ; * b251 *, 465 (1990). a. capella et al., phys. lett. * b206 *, 47 (1988) ; c. gerschel and j. huefner, phys. lett. * b207 *, 253 (1988). j. j. aubert et al., nucl. phys. * b213 *, 1 (1983). j.- peng et al., in proceedings of workshop on nuclear physics on the light cone, july 1988, world scientific publisher, p.65. d. kharzeev and h. satz. b366 *, 316 (1996). s. gavin and r. vogt. * b345 *, 104 (1990). r. vogt, s. j. brodsky and p. hoyer, nucl. phys. * b360 *, 67 (1991) ; r. vogt, nucl. phys. * a544 *, 615c (1992) ; w. q. chao and b. liu, z. phys. * c72 *, 291 (1996). d. kharzeev, h. satz, a. syamtomov and g. zinoviev, cern - th/96 - 72. na50 collaboration to appear in proceedings of quark matter96. r. l. anderson, slac - pub 1741 (1976). blaizot and j. y. ollitrault, phys. lett. * 77 *, 1703 (1996). wong, ornl - ctp 96/07 (1996). wong, introduction to high energy heavy - ion collisions ", world scientific publishing, singapore 1994, p.360. n. s. craigie, phys. rep. * 47 *, 1 (1978). b. andersson, g. gustafson, g. ingelman and t. sjstrand, phys. rep. * 97 *, 31 (1983). b. andersson, g. gustafson and b. nilsson - almqvist, nucl. b281 *, 289 (1987). b. andersson, g. gustafson and h. pi, z. phys. * c57 *, 485 (1993). g. gustafson, phys. lett. * b175 *, 453 (1986) ; g. gustafson and u. pettersson, nucl. phys. * b306 *, 746 (1988). b. andersson et al. z. phys. * c43 *, 625 (1989). l. lnnblad, `` ariadne version 4, a program for simulation of qcd cascades implementing the color dipole model '', desy 92 - 046 bengtsson and t. sjstrand, comput.. commun. * 46 *, 43 (1987). t. sjstrand, `` a manual to the lund monte carlo for jet fragmentation and @xmath71 physics : jetset version 7.3 '', available upon request to the author. g. a. schuler, cern - th 7170/94. * figure captions * + figure1 : @xmath0 cross sections divided by @xmath49 as a function of @xmath54. our results are compared with the experimental data from na51, na38 and na50 @xcite. + figure2 : @xmath0 cross sections divided by drell - yan cross sections as a function of @xmath72. our results are compared with the experimental data from na38 and na50 @xcite. + + @xmath73 gev / c & @xmath45&@xmath74 + @xmath75&@xmath76&@xmath77 + @xmath78&@xmath79&@xmath80 + @xmath81&@xmath82&@xmath83 + @xmath84&@xmath85&@xmath86 + @xmath87&@xmath88&@xmath89 + @xmath90&@xmath91&@xmath92 + | when the cross section of @xmath0 production is considered varying with the energy of the nucleon - nucleon interaction the production of @xmath0 in pa and aa collisions has been studied using fritiof model.
the calculation shows that the cross section of @xmath0 production per nucleon - nucleon collision " decreases with increasing mass number and centrality as a consequence of continuous energy loss of the projectile nucleons to the target nucleons in their successive binary nucleon - nucleon collisions.
we have compared our model predictions with the experimental data of @xmath0 production.
pacs number : 25.75.-q 3ex report - no : bthep - th-96 - 42 + dec.
1996 + tai an@xmath1, chao wei qin@xmath2 and yao xiao xia@xmath3 ttttt = tt = a) ccast (world lab.), p. o. box 8730 beijing, p.r.
china.
+ b) institute of high energy physics, academia sinica, + p. o. box 918(4 - 1), beijing, 100039, p.r. china.
+ suppression of @xmath0 production in high energy heavy ion collisions was proposed as an effective signature of qgp formation ten years ago @xcite.
the ensuing experimental data confirmed a significant suppression of @xmath0 production in both pa and aa collisions @xcite.
however, alternative explanations of the @xmath0 suppression exist based on the absorption of @xmath0 in nuclear matter @xcite.
the overall set of extensive data collected and analysed by na38 seems to support the absorption mechanism in p+b@xmath4, o+b@xmath4 and s+u collisions.
nevertheless a rather large absorption cross section (@xmath5 6.2 mb) has to be used in order to fit the experimental data, about three times larger than the total @xmath0-n cross section from the emc collaboration @xcite.
such a difference was already noted long time ago in @xcite.
recent calculations based on the colour octet model show that this phenomenological cross section could be understood by the absorption of pre - resonance states (@xmath6) in nuclear matter @xcite.
other sources of @xmath0 suppression in nuclear collisions, such as the interaction of @xmath0 particles with the produced mesons (called @xmath7) @xcite, gluon shadowing in nuclei, intrinsic charm component, energy degradation of produced @xmath8 pair, etc.
@xcite, have been introduced besides the absorption due to the @xmath0-n interaction to explain the data.
we in this paper propose a simple model to investigate @xmath0 suppression in pa and aa collisions focusing on the decrease of the cross section of @xmath0 production with increasing mass number and centrality due to the continuous energy loss of the projectile nucleons to the target nucleons in their successive binary nucleon - nucleon collisions, not on its later absorption in nuclear matter. from the calculations of this model,
we conclude that the absorption of @xmath0 particles in nuclear matter is only part of the sources of @xmath0 suppression seen in pa and aa collisions so far.
since the probability for a nucleon - nucleon collision leading to @xmath0 production is very small it is generally accepted that multiple @xmath0 production processes in pa and aa collisions can be neglected.
the studies of @xmath0 suppression based on the absorption mechanism actually assume that the probability of @xmath0 production per nucleon - nucleon collision
" is the same independent of the masses of the colliding nuclei at a given energy.
however, each binary nucleon - nucleon collision experienced by a projectile nucleon on its way out of the target in pa and aa collisions may not be the same if the projectile nucleon loses a fraction of its energy in each binary collision to the target nucleon such that the cms energy of each binary collision of this incoming nucleon with a target nucleon is different.
for the cross sections of the hard qcd parton - parton scatterings, which is responsible for @xmath0 production, increase with increasing energy it is conceivable that the probability of @xmath0 production would also depend on the energy of a nucleon - nucleon collision in a similar way.
it is confirmed both theoretically and experimentally that @xmath0 photoproduction cross section exhibits a strong threshold behaviour at the low energy region and then increases with energy at the higher energy region @xcite. in a participant - spectator model of nucleus - nucleus collisions, like fritiof
, each projectile nucleon may collide several times with the target nucleons.
if momentum transfers are assumed to take place in each binary nucleon - nucleon collision, then the cms energy of these binary collisions will decrease with time, thereby the probability to produce @xmath0 in each binary collision going down.
we in this paper have calculated how the cross section of @xmath0 production varies with increasing mass number of colliding nuclei and centrality using fritiof model.
the calculations show that @xmath0 production is suppressed in pa and aa collisions in comparison with the nucleon - nucleon collision, and the larger the mass number of colliding nuclei and the centrality, the greater the suppression, as a result that probability of finding a qcd hard process in a binary nucleon - nucleon decreases with increasing mass number of colliding nuclei and centrality. taking into account the @xmath0 absorption by @xmath0-n interaction our results are in good agreement with experimental data with the exception of the latest data of pb@xmath9pb collisions at 158 agev / c, which show a further @xmath0 suppression @xcite.
the absorption cross section @xmath10 needed to explain the data from p+b@xmath4, o+b@xmath4 and s+u collisions is about 1.4 mb in this paper, which is consistent with the experiments (emc collaboration @xcite gives a total @xmath0-n cross section 2.2 @xmath11 0.7 mb and the @xmath0-n quasi - elastic cross section is given in @xcite as 0.79 @xmath11 0.012 mb).
furthermore, our results also imply that some new mechanism of @xmath0 suppression seems to be needed particularly for understanding pb@xmath9pb data.
the authors in @xcite @xcite have attributed the further @xmath0 suppression in pb+pb data to the formation of a qgp state. as we have mentioned before the cross section of @xmath0 production will be different in each binary nucleon - nucleon collision in pa and aa collisions if the energy loss of the projectile nucleons to the target nucleons in the successive binary collisions is taken into account.
assume that @xmath12 is the mean cross section for the production of a @xmath0 particle in a binary nucleon - nucleon collision (here the average is done over all the binary collisions at an impact parameter @xmath13), then the total probability for producing a @xmath0 particle in the collisions of a + b at an impact parameter @xmath13 is the sum @xcite @xmath14^n[1-t(b)<\sigma^{nn}_{j/\psi } >] ^{ab - n }, \label{ff1}\]] where t(b) is the thickness function. because @xmath15 is a very small quantity (@xmath16 @xmath17 @xcite, @xmath18@xmath19 for central s+u collisions, for instance), the summation given by eq.([ff1]) is dominated by the first term with n=1.
the terms with n@xmath201 represent multiple @xmath0 production processes and shadowing corrections, which are very small and can be neglected.
then the probability for @xmath0 production in a+b collisions can be approximated to be @xmath21.
\label{ff2}\]] therefore, the cross section of @xmath0 production corresponding to a centrality bin, @xmath22, is given by the following formula @xmath23 where @xmath24 is the mean cross section of @xmath0 production `` per nucleon - nucleon collision '' within the centrality bin @xmath22.
we know that qcd hard scatterings (the gluon fusion and quark - antiquark annihilation) between partons are the main source of @xmath0 production.
let @xmath25 be the probability to have a hard scattering in a nucleon - nucleon collision and @xmath26 the probability to produce a @xmath0 from the hard scattering, then we can write out @xmath12 to be @xmath27 where @xmath28 is the total cross section of a nucleon - nucleon collision.
we have assumed that @xmath29 is approximately a constant in the energy span that we are concerning, so the product is the same for all the binary collisions. combining eq.([ff4]) with eq.([ff3]) and replacing @xmath30 by the ratio @xmath31 (@xmath32 is the number of binary collisions in an a+b collision and @xmath33 the number of binary collisions with a hard scattering.
both of them are the function of the impact parameter @xmath13) we finally obtain @xmath34 and for the minimum bias events we have @xmath35 we see from eq.([ff6]) that the dependence of the quantity @xmath36 on the masses of colliding nuclei or centrality is solely determined by how the mean probability of having a hard scattering in a binary nucleon - nucleon collision varies with the masses of colliding nuclei or centrality. before calculating @xmath37 in pa and aa collisions we will give a brief introduction of fritiof dynamics focusing on how a hard parton - parton scattering is distinguished from a soft one.
fritiof is a string model based on the concepts of the lund string model @xcite, which started from the modeling of inelastic hadron - hadron collisions and it has been successful in describing many experimental data from the low energies at the isr - regime all the way to the top sps energies @xcite @xcite.
this has been achieved by the introduction of a particular longitudinal momentum transfer scenario, gluon bremsstrahlung radiation (the dipole cascade model, dcm @xcite, and the soft radiation model, srm @xcite, this is implemented by the use of ariadne @xcite) as well as hard parton scattering (rutherford parton scattering, rps this is implemented by the pythia routines @xcite). in fritiof, during the collision two hadrons are excited due to longitudinal momentum transfers and/or a rps.
it is further assumed that there is no net color exchange between the hadrons.
the highly excited states will emit bremsstrahlung gluons according to the srm.
they are afterwards treated as excitations or the lund strings and the string states are allowed to decay into final state hadrons according to the lund prescription as implemented by jetset @xcite. in the fritiof model a hadron
is assumed to behave like a massless relativistic string (mrs) corresponding to a confined color force field of a vortex line character embedded in a type ii color superconducting vacuum.
a hadron - hadron collision is pictured as the multi - scatterings of the partons inside the two colliding hadrons.
this includes both the hard and the soft components depending on the four - momentum transfers @xmath38, or equivalently the transverse momentum transfers involved.
the soft part is described by a simple phenomenological model.
the hard scatterings can however be calculated from perturbative qcd, and correspond to the rutherford parton - parton scattering (rps). the divergence problem in rps
is handled by introducing the sudakov factor.
there will be color separation in the model, i.e. there will for each hadron be a color @xmath39 (a `` diquark '') continuing forward along the beam direction and a valence quark, a color @xmath40, moving in the opposite direction due to the longitudinal momentum transfer.
this will lead to bremsstrahlung of a dipole character.
a procedure therefore is adopted in fritiof that compares the `` hardness '' of the rutherford partons to that of the bremsstrahlung gluons.
the rps is accepted only if it is harder than the associated radiation. if the rps is `` drowned '', which is to say that it is softer than the radiation, then the rps is not acceptable and the collision proceeds as a purely soft collision. with this prescription
the rps spectrum is suppressed smoothly at small to medium transverse momentum region. for the hadron - nucleus and nucleus - nucleus collisions, the process has in the fritiof model been treated as a set of incoherent collisions on the nucleons.
thus a nucleon from the projectile interacts independently with the encountered target nucleons as it passes through the nucleus. the probability distribution for the number of inelastic collisions @xmath41 is taken from geometric calculations.
each of the sub - collisions is treated in the same way as an ordinary hadron - hadron collision, although the momentum transfers will again be additive and every encounter will make the projectile more excited.
if it interacts with @xmath41 nucleons in the target, @xmath42 excited string states will be formed as a result.
these string states will then independently emit associated bremsstrahlung radiation and then fragment into hadrons in the same way as individual strings.
this picture is supported by the fact that the global features of heavy ion collisions are satisfactorily explained by the collision geometry together with the independent hadron - hadron collisions. using fritiof it is straightforward to calculate the number of binary nucleon- nucleon collisions, @xmath32, and the number of the binary collisions with a hard scattering, @xmath43, in pa and aa collisions at a given impact parameter @xmath13, so that the mean probability to have a hard scattering in a binary nucleon - nucleon collision, @xmath30 = @xmath44, can be obtained.
since we are mainly interested in @xmath0 production in pa and aa collisions relative to that in the pp collision we do not need to know how a @xmath0 is actually formed from the hard scatterings in order to investigate the dependence of @xmath0 production cross sections on mass number and centrality.
we have calculated the quantity @xmath45 for various pa and aa collisions (and @xmath46 for different centrality bins in s+u and pb+pb collisions) at @xmath47=200 gev / c using fritiof. after determining @xmath29 by the data of the pp collision we plot our results of @xmath48 as a function of @xmath49 in figure 1 for the minimum bias events. for the cross section of @xmath0 production in different centrality bins we plot the results of @xmath50 as a function of @xmath51 in figure 2, where @xmath52 is the number of participants from the projectile and @xmath53 the number of participants from the target, since the drell - yan cross section in a given centrality bin is found in experiments proportional to an effective @xmath54 (= @xmath55).
in the same way, a constant has to be determined by the corresponding data of the pp collision.
the impact parameter bins are taken to be the same as those extracted by na38 and na50 @xcite.
we decided not to use the absorption length @xmath56 to be the longitudinal axis as used by na50 because @xmath56 calculated from the geometry model is not sensitive to the change of impact parameter for very central pb+pb collisions.
the results of our calculations show that the decrease of quantity, @xmath57 or @xmath50 (the cross section of @xmath0 production per nucleon - nucleon collision "), is due to the fact that the probability of the qcd hard scattering per binary nucleon - nucleon collision decreases with the increasing mass number and centrality.
when the absorption of @xmath0 by @xmath0-n interaction is also taken into account, i.e. the previous results are multipied by @xmath58 with @xmath59=0.14 n/@xmath60, @xmath61=1.4 mb and @xmath56 taken to be the same as those in @xcite, our model reproduces the data of @xmath0 suppression with the exception of the latest data from pb+pb collisions at 158 gev / c, which clearly show a further suppression.
one possibility which will bring about a further suppression of @xmath0 production is that @xmath62, the probability to produce a @xmath0 from a hard process, drops down suddenly under certain conditions.
this is equivalent to say that the @xmath8 produced from the hard processes can not form a bound state.
however at the moment our simple model can not estimate when this would happen. however, there are still other possible mechanisms of @xmath0 suppression, which are not included in our simple model.
the @xmath63 dependence of @xmath0 suppression is not investigated yet.
therefore, it is hard to make any conclusion now whether this further suppression in pb+pb collision is due to qgp formation.
it is known that there is no unique criterion to distinguish a hard process from a soft one in a nucleon - nucleon collision.
usually a @xmath64 is introduced to be the minimum transverse momentum of the produced partons from a hard process.
a dynamic criterion is applied in fritiof to chose a hard process by comparing the hardness of a rps parton with the hardness of the bremsstrahlung gluons as mentioned before. however
, the cross section of @xmath0 production should not depend on which criterion is actually used in the calculation.
we have thus calculated all the results of this paper using the conventional @xmath64 criterion in pythia (@xmath64= 1gev / c), just to check if our conclusion relies on the specific criterion in fritiof.
the calculations show that the results in these two cases are in agreement with each other.
we have also checked if the quantity @xmath65 is energy - independent as we have assumed.
a parametrization form of the @xmath0 cross section is given as @xcite @xmath66 where @xmath67 stands for the cms energy per nucleon.
therefore, if @xmath65 in eq.([ff6]) is energy - independent then we should have a ratio @xmath68 we have calculated @xmath45 and the ratio at various energies from @xmath47=60 gev / c to @xmath47= 450 gev / c for pp collisions (@xmath69= 1 for a pp collision) and the results are listed in tab.1, which show that this ratio in this energy region is not sensitive to the change of energy in comparison with @xmath70. but a threshold behaviour may exist at lower energies, which can be seen from the value at @xmath47=60 gev / c. | hep-ph9701207 |
Introduction
recent experiments and observations have already opened up what one can actually call the era of precision cosmology. it is now hoped that next years will see rather impressive advances leading not just to the determination of key cosmological parameters with an accuracy not even dreamed five years ago, but also to unprecedented scrutinies on fundamental aspects of particle and field theories. in particular, the connection between recent boomerang (de bernardis et al. 2000) and maxima (hanany et al. 2000) experiments with theory is twofold. they are linked through predictions from either inflationary (and perhaps cosmic string) models by using fundamental particle physics arguments (hu and white 1996) or from general more or less standard cosmological scenarios which may or may not include some new concepts such as quintessence (cadwell, dave and steinhardt 1998). the aim of this paper will concentrate at a particular aspect of cmb anisotropy measurements : the dependence of the position of the first doppler peak on the values of the relevant cosmological parameters within the realm of a quintessence model. kamionkowski, spergel and sugiyama (1994) originally derived the simple relation that the position of the first doppler peak @xmath2, where @xmath3, which has been the subject of some controversy. while frampton, ng and rohm (1998) have recently derived a similar dependence using a simple quintessential model, weinberg (2000) has argued that this formula is not even a crude approximation when @xmath4 is smaller than @xmath5. for a very recent discussion on quntessential model of cmb anisotropies see bond et al. (2000). assuming the holding of the cosmic triangle condition, we obtain in this paper that @xmath6 only depends on the topology of the universe, so confirming the original proposal by kamionkowski, spergel and sugiyama (1994).
Calculation of @xmath6
in order to derive an expression relating the position of the first doppler peak with the scalar field potential of quintessence models, we choose a general tracking model with time - dependent parameter @xmath7 for the state equation (zlatev, wang and steinhardt 1999). this kind of models can be related to particle physics and may solve the so - called cosmic coincidence problem (steinhardt 1997). one typically considers tracking quintessence fields @xmath8 for a ratra - peebles potential @xmath9 with a given constant parameter @xmath10 (ratra and peebles 1988). during its cosmic evolution, the equation of state parameter @xmath11 passes through several distinct regimes (brax, martin and riazuelo 2000), including first a kinetic regime lasting until @xmath12 with @xmath13, then the transition and potential regimes chracterized by @xmath14, to finally reach the proper tracking regime at @xmath15, where @xmath16 (with @xmath17 for radiation dominated universe and @xmath18 for matter dominated universe). on this regime, there exists a particular solution for the scalar field, @xmath19 (in which @xmath20 is the conformal time, with @xmath21 the scale factor) which is an attractor able to solve the cosmic coincidence problem. for @xmath22 a change of @xmath11 should then be expected when the universe becomes matter dominated once the the surface of last scattering is overcome. since the tracking regime is characterized by @xmath23 (@xmath24 being the sound velocity defined as @xmath25 (brax, martin and riazuelo 2000)), parameter @xmath11 becomes a constant given by @xmath26 when matter dominates. it has been recently argued that @xmath27 (balbi et al. 2001), so that if we choose for definiteness @xmath28, then @xmath29 for the vacuum quintessence field. in this case, the ratra - peebles potential becomes @xmath30 since the corresponding tracking solution, @xmath31, would then correspond to a constant parameter @xmath32 for the quintessence field, it should satisfy the constraint equation derived from the corresponding conservation laws and cosmological field equations (di pietro and demaret 1999, gonzlez - daz 2000). in terms of the cosmological parameters @xmath33, @xmath34, and the quintessence field potential @xmath35, this constraint equation can be written as (gonzlez - daz 2000) @xmath36 @xmath37 } + \omega_{\lambda}\left(\frac{v}{v_0}\right)^{(3\omega+4)/[3(\omega+1)]},\]] where @xmath38, the subscript 0 means current value and @xmath39. (2) corresponds to the generalized quintessence model with negative constant parameter for the state equation recently suggested (gonzlez - daz 2000). besides the contributions from the topological curvature, @xmath40, and gravitationally observable mass, @xmath41, this model distinguishes two essentially distinct contributions from vacuum energy : a varying cosmological term with positive energy density @xmath5 assumed to satisfy the conservation law @xmath42 at sufficiently small redshifts, and a quintessence negative energy density such that while @xmath43 (@xmath44 being the current value of the hubble constant) is always negative, @xmath45 is always positive, so satisfying the ford - roman s quantum interest conjecture (ford and roman 1999). adapting then the relation between the position of the first doppler peak @xmath6 and the cosmological parameters @xmath33 first derived by frampton, ng and rohm (1998)to our generalized model, we can finally obtain a general relation between @xmath6 and the quintessence potential @xmath35 of the form : @xmath46 @xmath47}\right]\right|_{z = z_r } s\left\{\frac{\sqrt{|\omega_k|}v_0'} { 3(\omega+1)v_0 } \int_{\phi(z=0)}^{\phi(z = z_r) } \frac{d\phi}{\left(\frac{v}{v_0}\right)^{(3\omega+1)/[6(\omega+1)]}}\right\ }, \]] where @xmath48 is the redshift at recombination, we have used the relation (di pietro and demaret 1999) @xmath49 and @xmath50 for @xmath51, @xmath52 for @xmath53 and @xmath54 for @xmath55. in the case of interest @xmath29, where we obtain both topological and dynamical acceleration, eqn. (3) reduces to @xmath56\right|_{z = z_r } s\left\{\frac{\sqrt{|\omega_k|}v_0'} { v_0 } \int_{\phi(z=0)}^{\phi(z = z_r)}d\phi\left(\frac{v}{v_0}\right)^{1/2}\right\ }, \]] and there is a solution to the constraint equation (2) given by @xmath57+\xi_0\right\}^{-1},\]] where @xmath58 @xmath59 in this case, the cosmological parameters must satisfy the relations @xmath60\geq 0\]] @xmath61 > 0.\]] together with the triangle equation @xmath62, eqns. (9) and (10) will restrict the values that the cosmological parameters may take on. on the other hand, we note from eqns. (4) and (6) that for large @xmath63 and @xmath64 one obtains @xmath65 and @xmath66 ; that is, while the potential (6) can be consistently interpreted as a generalization for smaller values of @xmath63 from the ratra - peebles potential, the attractor solution @xmath67 must necessarily change into @xmath68 as the universe becomes matter dominated. performing the derivative and integration in eqn. (5) after inserting solution (6), we finally obtain for the position of the first doppler peak when @xmath29 @xmath69 } \left.t\left\{\sqrt{\frac{|\omega_k|}{\omega_v\varphi_0 } } f\left[\varphi(z), r_0\right]\right|_{0}^{z_r}\right\ }, \]] where @xmath70 $] is the elliptic integral of the first kind (abramowitz and stegun 1965), @xmath71 @xmath72 @xmath73 and @xmath74 for @xmath51 and, unlike the function @xmath75, @xmath76 both for the closed and open cases. @xmath0 can run between two extreme values, such that @xmath77 a plot for the position of the first doppler peak, @xmath6, against the parameter @xmath0 is given in fig. it can be seen that the maximum value for @xmath6 is reached when @xmath6 approaches the value 190 at the flat case @xmath51, decreasing slowly therefrom as @xmath78 increases, a little more steeply for @xmath79 than for @xmath53. this seems to quite reasonably reproduce the results provided by boomerang, maxima and previous (netterfield et al. 1997, vollaek et al. 1997) experiments. although uncertainties about our results for @xmath29 must come from the possibility of having values of @xmath10 other than just unity (provided they are on the interval (0,2), or small deviation from time - independence of @xmath11 during cosmic evolution after recombination, they seems to strongly suggest a nearly flat topology for the universe. we note that cbm anisotropies and cosmic acceleration can be easily related in our model. to see this, let us consider solution (6) for the the flat case, i.e. @xmath80\right\}^{-1 }, \]] with @xmath81 this allows us to also construct a plot for the luminosity distance versus redshift by using the expression (gonzlez - daz 2000) @xmath82\right|_{0}^{z}\right\ }.\]] one can see readily that our flat solution in fact gives rise to a @xmath83 plot with a nearly straight line between @xmath84 and @xmath85 which appears to slightly accelerate thereafter, the full @xmath86 that corresponds to the @xmath63-interval of presently available type ia supernova observations, @xmath87 $], being around 12. this accelerating behaviour for the universe conforms quite well the data obtained from distant supernova ia (perlmutter et al. 1999, reiss et al. it is worth noticing that though our universe model is dynamically accelerating it still is topologically uniform as @xmath88 exactly vanishes in the flat case. as one separates from flatness one model produces topologically decelerating scenarios.
Conclusions
consistent solutions to the constraint equation for the quintessence potential have been obtained which all correspond to particular fixed sets of values for @xmath33, @xmath89, in such a way that the resulting value of the position of the first doppler peak @xmath6 becomes automatically fixed once just one of the parameters @xmath0, @xmath4 or @xmath90 is fixed, but does not depend on the ratio @xmath91. one can also conclude that @xmath6 acquires a maximum value which fits fairly well experimental results at @xmath51 (i.e. @xmath92). as the topology of the universe separates from flatness (@xmath93 if @xmath94, or @xmath95 if @xmath96), @xmath6 always decreases from its maximum value. since @xmath97, from the condition that @xmath98 be real it follows that the quintessence field @xmath8 is pure imaginary. this amounts to the interpretation that the classical solution is axionic, and hence the quintessence field can be taken as a genuine component of dark matter in such a way that one can define for the gravitationally observable mass of the universe the quantity @xmath99, with @xmath100. thus, one can always adjust our results to the currently favoured combination @xmath101, @xmath102 for any @xmath0. we finally remark that the results obtained in this paper refer only to the case of a quintessence parameter for state equation @xmath29 (i.e. @xmath28). it is expected that results even more adjusted to observation (i.e. closer to 200) can also be obtained within the model of this paper by slightly shifting @xmath11 towards more realistic smaller values approaching -0.8. abramowitz, m., and stegun, i.a., 1965, handbook of mathematical functions : dover, 1995 balbi, a., baccigalupi, c., matarrese, s., perrotta, f., and vittorio, n. 2001,, 547, l89 de bernardis, p. et al. 2000, nature, 404, 955 bond, j.r. 2000, astro - ph/0011379 brax, p., martin, j., and riazuelo, a. 2000, phys. d62, 103505 caldwell, r.r., dave, r., and steinhardt, p.j. 1998, phys. 80, 1582 di pietro, e., and demaret, j. 1999, gr - qc/9908071 ford, l.h., and roman, t.a. 1999, d60, 104018 frampton, p.h., ng, y.j., rohm, r.m. 1998, mod. a13, 2541 gonzlez - daz, p.f. 2000, phys. d62, 023513 hanany, s. et al. 2000,, 545, 1 hu, w., and white, m. 1996,, 471, 30 kamionkowski, k, spergel, d.n., and sugiyama, n. 1994,, 426, 57 netterfield, c.b. 1997,, 474, 47 perlmutter, s. et al. 1999,, 517, 565 ratra, b., and peebles, p.j.e. 1988, phys. d37, 3406 steinhardt, p.j. 1997, critical problems in physics, v.l. fitch and d.r. marlow, princeton : princeton university press weinberg, s. 2000, phys. d62, 127302 zlatev, l., wang, l., and steinhardt, p.j. 1999, lett. 82, 896 | by using a tracking quintessence model we obtain that the position of the first doppler peak in the spectrum of cmb anisotropies only depends on the topology of the universe, @xmath0, for any value of the ratio @xmath1, so that such a dependence is perfectly valid in the range suggested by supernova observations. | astro-ph0103194 |
Introduction
at the end of the last century, the astronomical observations of high redshift type ia supernovae (snia) indicated that our universe is not only expanding, but also accelerating, which conflicts with our deepest intuition of gravity. with some other observations, such as cosmic microwave background radiation (cmbr), baryon acoustic oscillations (bao) and large - scale structure (lss), physicists proposed a new standard cosmology model, @xmath0cdm, which introduces the cosmological constant back again. although this unknown energy component accounts for 73% of the energy density of the universe, the measured value is too small to be explained by any current fundamental theories.@xcite-@xcite if one tries to solve this trouble phenomenologically by setting the cosmological constant to a particular value, the so - called fine - tuning problem would be brought up, which is considered as a basic problem almost any cosmological model would encounter. a good model should restrict the fine - tuning as much as possible. in order to alleviate this problem, various alternative theories have been proposed and developed these years, such as dynamical dark energy, modified gravity theories and even inhomogeneous universes. recently, a new attempt, called torsion cosmology, has attracted researchers attention, which introduces dynamical torsion to mimic the contribution of the cosmological constant. it seems more natural to use a pure geometric quantity to account for the cosmic acceleration than to introduce an exotic energy component. torsion cosmology could be traced back to the 1970s, and the early work mainly focused on issues of early universe, such as avoiding singularity and the origin of inflation. in some recent work, researchers attempted to extend the investigation to the current evolution and found it might account for the cosmic acceleration. among these models, poincar gauge theory (pgt) cosmology is the one that has been investigated most widely. this model is based on pgt, which is inspired by the einstein special relativity and the localization of global poincar symmetry@xcite. et al_. made a comprehensive survey of torsion cosmology and developed the equations for all the pgt cases.@xcite based on goenner s work, nester and his collaborators@xcite found that the dynamical scalar torsion could be a possible reason for the accelerating expansion. et al_.@xcite extended the investigation to the late time evolution, which shows us the fate of our universe. besides pgt cosmology, there is another torsion cosmology, de sitter gauge theory (dsgt) cosmology, which can also be a possible explanation to the accelerating expansion. this cosmological model is based on the de sitter gauge theory, in which gravity is introduced as a gauge field from de sitter invariant special relativity (dssr), via the localization of de sitter symmetry.@xcite dssr is a special relativity theory of the de sitter space rather than the conventional minkowski spacetime, which is another maximally symmetric spacetime with an uniform scalar curvature @xmath1. and the full symmetry group of this space is de sitter group, which unifies the lorentz group and the translation group, putting the spacetime symmetry in an alternatively interesting way. but in the limit of @xmath2, the de sitter group could also degenerate to the poincar group. to localize such a global symmetry, de sitter symmetry, requires us to introduce certain gauge potentials which are found to represent the gravitational interaction. the gauge potential for de sitter gauge theory is the de sitter connecion, which combines lorentz connection and orthonormal tetrad, valued in @xmath3(1,4) algebra. the gravitational action of dsgt takes the form of yang - mills gauge theory. via variation of the action with repect to the the lorentz connection and orthonormal tetrad, one could attain the einstein - like equations and gauge - like equations, respectively. these equations comprise a set of complicated non - linear equations, which are difficult to tackle. nevertheless, if we apply them to the homogeneous and isotropic universe, these equations would be much more simpler and tractable. based on these equations, one could construct an alternative cosmological model with torsion. analogous to pgt, dsgt has also been applied to the cosmology recently to explain the accelerating expansion.@xcite our main motivation of this paper is to investigate (i)whether the cosmological model based on de sitter gauge theory could explain the cosmic acceleration ; (ii)where we are going, i.e., what is the fate of our universe ; (iii) the constraints of the parameters of model imposed by means of the comparison of observational data. by some analytical and numerical calculations, we found that, with a wide range of initial values, this model could account for the current status of the universe, an accelerating expanding, and the universe would enter an exponential expansion phase in the end. this paper is organized as follows : first, we summarize the de sitter gauge theory briefly in sec. [sec : de - sitter - gauge], and then show the cosmological model based on de sitter gauge theory in sec. [sec : cosm - evol - equat]. second, we rewrite these dynamical equations as an autonomous system and do some dynamical analysis and numerical discussions on this system in the sec. [sec : autonomous - system] and [sec : numer - demonstr]. next in the [sec : supern - data - fitt]th section, we compare the cosmological solutions to the snia data and constrain the parameters. last of all, we discuss and summarize the implications of our findings in section [sec : summary - conclusion].
De sitter gauge theory of gravitation
[supernovae data fitting]in dsgt, the de sitter connection is introduced as the gauge potential, which takes the form as @xmath4 -r^{-1}e^b_\mu & 0 \end{array } \right) \in \mathfrak{so}(1,4),\]] where @xmath5, @xmath6 and @xmath7, which combines the lorentz connection and the orthonormal tetrad are 4d coordinate indices, whereas the capital latin indices @xmath8 and the lowercase latin indices, @xmath9 denote 5d and 4d orthonormal tetrad indices, respectively.]. the associated field strength is the curvature of this connection, which is defined as @xmath10 -r^{-1}t^b_{~\mu\nu } & 0 \end{array } \right) \in \mathfrak{so}(1,4),\end{aligned}\]] where @xmath11, @xmath12 is the de sitter radius, and @xmath13 and @xmath14 are the curvature and torsion of lorentz - connection, @xmath15 which also satisfy the respective bianchi identities. the gauge - like action of gravitational fields in dsgt takes the form, @xcite @xmath16.\label{gym}\end{aligned}\]] here, @xmath17, @xmath18 is a dimensionless constant describing the self - interaction of the gauge field, @xmath19 is a dimensional coupling constant related to @xmath18 and @xmath12, and @xmath20 is the scalar curvature of the cartan connection. in order to be consistent with einstein - cartan theory, we take @xmath21 and @xmath22, where @xmath23. assuming that the matter is minimally coupled to gravitational fields, the total action of dsgt could be written as : @xmath24 where @xmath25 denotes the action of matter, namely the gravitational source. now we can obtain the field equations via variational principle with respect to @xmath26, @xmath27 \label{feq2}% & & \nabla_{\nu}f_{ab}^{~~\mu\nu}-r^{-2}\left(y^\mu_{~\,\lambda\nu } e_{ab}^{~~\lambda\nu}+y ^\nu_{~\, \lambda\nu } e_{ab}^{~~\mu\lambda } + 2t_{[a}^{~\mu\lambda } e_{b]\lambda}\right) = 16\pi gr^{-2}s^{\quad \mu}_{{\rm m}ab},%\end{aligned}\]] where@xmath28 represent the effective energy - momentum density and spin density of the source, respectively, and @xmath29 is the contorsion. it is worth noticing that the nabla operator in eq. ([feq1]) and ([feq2]) is the covariant derivative compatible with christoffel symbols \{@xmath30 } for coordinate indices, and lorentz connection @xmath31 for orthonormal tetrad indices. readers can be referred to ref.@xcite for more details on dsgt.
The cosmological evolution equations
since current observations favor a homogeneous, isotropic universe, we here work on a robertson - walker (rw) cosmological metric @xmath32.\]] for robertson - walker metric, the nonvanishing torsion tensor components are of the form, @xmath33 where @xmath34 denotes the vector piece of torsion, namely, in components, the trace of the torsion, and @xmath35 indicates the axial - vector piece of torsion, which corresponds in components to the totally antisymmetric part of torsion. @xmath34 and @xmath35 are both functions of time @xmath36, and their subscripts, + and -, denote the even and odd parities, respectively. the nonvanishing torsion 2-forms in this case are @xmath37 where @xmath38. according to the rw metric eq. and the torsion eq., the field equations could be reduced to @xmath39 \label{el-11}% & & \frac{\ddot a^2 } { a^2 } + \left(\dot t_+ + 2\frac{\dot a } a t_+ - 2\frac{\ddot a } a + \frac{6}{r^2}\right)\dot t_+ -\frac 1 4 \left(\dot t_- + 2 \frac { \dot a } a t_-\right)\dot t_- - t_+^4 + \frac 3 2 t_+^2 t_-^2 - \frac1 { 16 } t_-^4 \nonumber\\ & & \quad+ \frac { \dot a } a(4 t_+^2 - 3 t_-^2)t_+ - \left(5\frac{\dot a^2 } { a^2 } + 2 \frac k { a^2 } + \frac3 { r^2}\right) t_+^2 + \frac 1 2 \left(\frac 5 2\frac{\dot a^2 } { a^2 } + \frac k { a^2 } + \frac 3 { r^2}\right) t_-^2- 2\frac{\dot a } a \left(\frac{\ddot a } { a}- 2\frac{\dot a^2 } { a^2 } \right.\nonumber \\[0.2 cm] & & \quad \left. - 2 \frac k { a^2}- \frac6 { r^2}\right)t_+ - \frac 4 { r^2 } \frac{\ddot a } a -\frac{\dot a^2 } { a^2 } \left(\frac{\dot a^2}{a^2 } + 2\frac k { a^2}\right) + \frac2 { r^2 } -\frac{k^2}{a^4 } - \frac2 { r^2}\frac k { a^2 } + \frac6 { r^4 } = -\frac{16\pi g p}{r^2 }, \\[0.3 cm] \label{yang1 } % & & \ddot t_- + 3 \frac{\dot a } a \dot t_- + \left (\frac 1 2 t_-^2 - 6 t_+^2 + 12 \frac { \dot a } a t_+ + \frac{\ddot a } a - 5\frac{\dot a^2}{a^2 } - 2\frac k { a^2}+ \frac 6 { r^2}\right) t_-=0, \\[0.3 cm] \label{yang2}% & & \ddot t_+ + 3 \frac{\dot a } a \dot t_+ -\left (2 t_+^2 -\frac 3 2 t_-^2 - 6\frac{\dot a } a t_+ -\frac { \ddot a } a + 5 \frac { \dot a^2 } { a^2 } + 2 \frac k { a^2}- \frac 3 { r^2}\right) t_+ - \frac 3 2 \frac{\dot a } a t_-^2-\frac{\dddot a } a - \frac{\dot a\ddot a}{a^2 } \nonumber\\ & & \quad+ 2\frac { \dot a^3 } { a^3 } + 2\frac{\dot a } a \frac k { a^2 } = 0, % \end{aligned}\]] where eqs. and are the @xmath40 and @xmath41 component of einstein - like equations, respectively ; and eqs. and are 2 independent yang - like equations, which is derived from the @xmath42 and @xmath43 components of lorentz connection. the spin density of present time is generally thought be very small which could be neglected. therefore, we here assumed the spin density is zero. the bianchi identities ensure that the energy momentum tensor is conserved, which leads to the continuity equation : @xmath44 equation can also be derived from eqs. - , which means only four of eqs. - are independent. with the equation of state (eos) of matter content, these four equations comprise a complete system of equations for five variables, @xmath45. by some algebra and differential calculations, we could simplify these 5 equations as : @xmath46t_+,\\ \ddot{t}_- & = & -3h\dot{t}_- -\left[-\frac{15}{2}t_{+}^{2}+\frac{33h t_{+}}{2}-6h^{2}-\frac{3k}{a^2}+\frac{8}{r^{2}}+\frac{5}{4}t^{2}_{-}+\frac{3}{2}\dot{t}_+ \right.\nonumber\\ & & \left. + \frac{4\pi g}{3}(\rho+3p)\right]t_{-},\\ \dot{\rho}&=&-3h(\rho+p),\\ \label{eq : eos } w&=&\frac{p}{\rho }, \end{aligned}\]] where @xmath47 is the hubble parameter.
Autonomous system
if we rescale the variables and parameters as @xmath48 where @xmath49 is the hubble radius in natural units, these variables and parameters would be dimensionless. under this transformation, - remain unchanged expect for the terms including @xmath50 and @xmath51, which change into @xmath52 and @xmath53 respectively. the contribution of radiation and spatial curvature in current universe are so small that it could be neglected, so we here just consider the dust universe with spatial flatness, whose eos is equal to zero. by some further calculations, these equations could be transformed to a set of six one - order ordinary derivative equations, which forms a six - dimensional autonomous system, as follows, @xmath54t_+ + 6h\rho, \\[0.2 cm] \dot{t_{+}}&=&p,\\ \dot q&=&-3h q -\left(-\frac{15}{2}t_{+}^{2}+\frac{33h t_{+}}{2}-6h^{2}+\frac{8}{r^{2}}+\frac{5}{4}t^{2}_{-}+\frac{3}{2}p + \rho \right)t_{-},\\ \dot{t_{-}}&=&q,\\ \dot{\rho}&=&-3h\rho. \label{rho}\end{aligned}\]] for such an autonomous system, we can use the dynamics analysis to investigate its qualitative properties. critical points are some exact constant solutions in the autonomous system, which indicate the asymptotic behavior of evolution. for example, some solutions, such as heteroclinic orbits, connect two different critical points, and some others, such as homoclinic orbits, are a closed loop starting from and coming back to the same critical point. in the dynamics analysis of cosmology, the heteroclinic orbit is more interesting.@xcite thus, critical points could be treated as the basic tool in dynamics analysis, form which one could know the qualitative properties of the autonomous system. by some algebra calculation, we find all the 9 critical points (@xmath55) of this system, as shown in table 1. . [tab : critical - points]the critical points and their corresponding eigenvalues. the point 9 is not physically acceptable, for its negative energy density. [cols="<,<,<",options="header ",] based on the current observations, the present density of torsion in our universe is very small, so it is reasonable to assume that the initial values of all torsions and their first order derivative are zero at @xmath56. but their second order derivatives does not vanish yet, which would have a significant impact on the history and future of the evolution of our universe. in this case, the number of parameter and initial value is reduced, and the rest parameters and initial values are just @xmath12 and @xmath57. it is easy to find the best - fit of these 2 parameters, which are shown in table [tab : best - fit]. and the minimal @xmath58 is @xmath59, whereas the value for @xmath0cdm is @xmath60, with @xmath61. in fig. [fig : chi2-distribution] we show the @xmath58 distribution with respect to @xmath12 and @xmath52 compared to @xmath0cdm model, where the plane @xmath62 corresponds to the value of @xmath0cdm. furthermore, we plot the contours of some particular confidence levels, as shown in fig. [fig : contour]. from these figures, we could find that the evolution of our universe is insensitive to the initial value, which alleviate the fine - tuning problem. the @xmath58 distribution with respect to @xmath12 and @xmath52, compared to the @xmath0cdm, the plane @xmath63. here we assume that all the torsions and their first order derivatives vanish at present time.,width=642,height=340] the 68.3%, 95.4% and 99.7% @xmath58 confidence contours of dsgt with respect to @xmath12 and @xmath52, using the union 2 dataset. here we also assume that the current - time values for all the torsions and their first order derivatives are zero. the yellow point is the best - fit point., height=264]
Summary and conclusion
the astronomical observations imply that our universe is accelerating to a de sitter spacetime. this gives us a strong motive to consider the cosmic evolution based on the de sitter gauge theory instead of other gravity theories. the localization of de sitter symmetry requires us to introduce curvature and torsion. so in de sitter gauge theory, the torsion is an indispensable quantity, by which people tried to include the effect of spin density in gravity theory at first. but now this essential quantity might account for the acceleration of our universe, if we apply dsgt to cosmology. we found the cosmological equations for dust universe in dsgt could form an autonomous system by some transformations, where the evolution of the universe is described in terms of the orbits in phase space. therefore, by dynamics analysis to the dsgt cosmology, one could study the qualitative properties of this phase space. we found all 9 critical points, as shown in table [tab : critical - points]. we also analyzed the stabilities of these critical points, and found among these critical points there is only one positive attractor, which is stable. the positive attractor alleviates the fine - tuning problem and implies that the universe will expand exponentially in the end, whereas all other physical quantities will turn out to vanish. in this sense, dsgt cosmology looks more like the @xmath0cdm, than pgt cosmology. and we conducted some concrete numerical calculations of this the destiny of our model of the universe, which confirms conclusions from dynamics analysis. finally, in order to find the best - fit values and constraints of model parameters and initial conditions, we fitted them to the union 2 snia dataset. the maximum likelihood estimator here we used is the @xmath58 estimate. by minimizing the @xmath58, we found the best - fit parameters @xmath64 and the corresponding @xmath65, while the value for @xmath0cdm is @xmath60, with @xmath66. note that we here set all the initial values of torsions and their first - order derivatives to zero at @xmath67, since the contribution of torsion to the current universe is almost negligible. we also plotted the confidence contour fig. [fig : contour] with respect to @xmath12 and @xmath52, from which it is easy to see that the fine - tuning problem is alleviated and the evolution is not so sensitive to the initial values and model parameters. if we want to go deeper into cosmology based on de sitter gauge theory, there are a lot of work need to be done. we should fit this model to some other observations, like bao and lss etc, to constrain the parameters better. we also could study the perturbations in the early universe, and compare the results to cmbr data. these issues will considered in some upcoming papers.
Acknowledgments
this work is supported by srfdp under grant no. 200931271104 and shanghai natural science foundation, china, under grant no. 10zr1422000. t. padmanabhan, _ cosmological constant : the weight of the vacuum _, _ phys. * 380 * (2003) 235 - 320, [hep - th/0212290] ; x.- z. li and j.- g. o(n) phantom, a way to implement w @xmath68 -1 _ _ phys. * d69 * (2004) 107303, [hep - th/0303093]. hao and x.- z. li, _ phantom cosmic dynamics : tracking attractor and cosmic doomsday _, _ phys. _ * d70 * (2004) 043529, [astro - th/0309746] ; + c.- j. feng and x.- z. li, _ cardassian universe constrained by latest observations _ _ phys. lett. _ * b692 * (2010) 152 - 156, [arxiv:0912.4793]. f. w. hehl, _ four lectures on poincar gauge field theory _ in _ proc. of the 6th course of the school of cosmology and gravitation on spin, torsion, rotation, and supergravity _, eds. p. g. bergmann and v. de sabbata (new york : plenum) p5 + m. blagojevi, _ gravitation and gauge symmetries _, iop publishing, bristol, 2002. h. chen, f.- h. ho and j. m. nester, c.- h. wang, and h.- j. yo, _ cosmological dynamics with propagating lorentz connection modes of spin zero _, _ jcap _ * 0910 * (2009) 027, [arxiv:0908.3323] ; + p. baekler, f. w. hehl, and j. m. nester, _ poincare gauge theory of gravity : friedman cosmology with even and odd parity modes. analytic part _ rev. _ * d83 * (2011) 024001, [arxiv:1009.5112] ; + f.- h. ho and j. m. nester, _ poincar gauge theory with coupled even and odd parity dynamic spin-0 modes : dynamic equations for isotropic bianchi cosmologies _, [arxiv:1106.0711] ; + f.- h. ho and j. m. nester, _ poincar gauge theory with even and odd parity dynamic connection modes : isotropic bianchi cosmological models _, [arxiv:1105.5001]. li, c.- b. sun, and p. xi, _ torsion cosmolgical dynamics _, _ phys. _ * d79 * (2009) 027301, [arxiv:0903.3088] ; + x.- z. li, c.- b. sun, and p. xi, _ statefinder diagnostic in a torsion cosmology _, _ jcap _ * 0904 * (2009) 015, [arxiv:0903.4724] ; + x.- c. ao, x.- z. li, and p. xi _ analytical approach of late - time evolution in a torsion cosmology _ lett. _ * b694 * (2010) 186 - 190, [arxiv:101.4117]. guo, c.- g. huang, z. xu, and b. zhou, _ on beltrami model of de sitter spacetime _ _ * a19 * (2004) 1701 - 1710, [hep - th/0311156] ; + h.- y. guo, c.- g. huang, z. xu, and b. zhou, _ on special relativity with cosmological constant _, _ phys. lett.. _ * a331 * (2004) 1 - 7,[hep - th/043171] ; + h.- y. guo, c.- g. huang, z. xu, and b. zhou, _ three kinds of special relativity via inverse wick rotation _ * 22 * (2005) 2477 - 2480, [hep - th/0508094] ; + h.- y. guo, c.- g. huang, b. zhou, _ temperature at horizon in de sitter spacetime _, _ europhys. * 72 * (2005) 1045 - 1051. li, y.- b. zhao, and c.- b. sun, _ heteroclinic orbit and tracking attractor in cosmological model with a double exponential potential _, _ class. * 22 * (2005) 3759 - 3766, [astro - ph/050819] ; + j.- hao and x.- z. li, _ an attractor solution of phantom field _, _ phys. _ * d67 * (2003) 107303, [gr - qc/0302100]. | a new cosmological model based on the de sitter gauge theory (dsgt) is studied in this paper. by some transformations,
we find, in the dust universe, the cosmological equations of dsgt could form an autonomous system.
we conduct dynamics analysis to this system, and find 9 critical points, among which there exist one positive attractor and one negative attractor.
the positive attractor shows us that our universe will enter a exponential expansion phase in the end, which is similar to the conclusion of @xmath0cdm.
we also carry out some numerical calculations, which confirms the conclusion of dynamics analysis.
finally, we fit the model parameter and initial values to the union 2 snia dataset, present the confidence contour of parameters and obtain the best - fit values of parameters of dsgt. | 1111.1801 |
Introduction
let @xmath0 be a standard brownian motion. it @xcite defined the multiple stochastic integral of a function @xmath1, @xmath2 taking care to ensure that the diagonal sets, like @xmath3, do not contribute at all. for this reason the integral has very good properties and is easy to work with. however, for a function of the form @xmath4 we have that, in general, @xmath5 that means the it multiple integral does not behave like the integral with respect to a product measure. many years later, hu and meyer @xcite introduced (although they believed that this integral was already known @xcite, page 75) a multiple integral, @xmath6, which followed the ordinary rules of multiple integration. they called it the multiple stratonovich integral. furthermore, hu and meyer stated the relationship between the it and stratonovich integrals, the celebrated hu meyer formula, adding the contribution of the diagonals to the it integral : for a function @xmath7 symmetric with good properties, @xmath8}\frac{n!}{(n-2j)!j!2^j } i_{n-2j } \biggl(\int_{{\mathbb{r}}_+^{j}}f(\bolds\cdot, t_1,t_1,t_2,t_2,\ldots, t_{j},t_{j }) \,d t_{1}\cdots d t_{j } \biggr).\]] this formula is simple because the quadratic variation of the brownian motion is @xmath9, and the integral over coincidences of order three or superior are zero. following their ideas, sol and utzet @xcite proved a hu meyer formula for the poisson process. again, in that case, the formula is relatively simple because the variations of any order of the process can always be written in terms of the poisson process and @xmath9. from another point of view, engel @xcite, working with a general process with independent increments, related the (it) multiple stochastic integral with the theory of vector valued measures, and masani @xcite, using also vector valued measures and starting from the wiener s original ideas, developed both the it and stratonovich integrals (with respect to the brownian motion) and proved many profound results. the vector measures approach is no simple matter ; engel s work covers 82 pages, and masani s covers 160. an important and clarifying contribution was made by rota and wallstrom @xcite who used combinatorial techniques to show the features of the multiple stochastic integration. they did not really work with integrals, but with products of vector measures. however, the path towards a general theory of multiple stochastic integration had been laid. see also prez abreu @xcite for an interesting generalization to hilbert space valued random measures. further, vershik and tsilevich @xcite, in a more algebraic context, constructed a fock factorization for a lvy process, and some important subspaces can be described through rota and wallstrom concepts. we should also mention the very complete survey by peccati and taqqu @xcite in which a unified study of multiple integrals, moments, cumulants and diagram formulas, as well as applications to some new central limit theorems, is presented. it is worth remarking that rota and walstrom s @xcite combinatorial approach to multiple integration has been extended to the context of free probability in a very interesting and fertile field of research, started by anshelevich (see @xcite and the references therein). in fact, rota and walstrom s ideas fit very well with the combinatorics of free probability (see nica and speicher @xcite) and noncommutative lvy processes. our renewed interest in rota and walstrom s paper @xcite was motivated by anshelevich s work. in the present paper we use the powerful rota and wallstrom s @xcite combinatorial machinery to study the stratonovich integral (the integral with respect to the product random measure) with respect to a lvy processes with finite moments up to a convenient order. the key point is to understand how the product of stochastic measures works on the diagonal sets, and that leads to the diagonal measures defined by rota and wallstrom @xcite. for a lvy process those measures are related to the powers of the jumps of the process, and hence to a family of martingales introduced by nualart and schoutens @xcite, called teugels martingales, which offer excellent properties. specifically, these martingales have deterministic predictable quadratic variation and this makes it possible to easily construct an it multiple stochastic integral with respect to different integrators, which can be interpreted as an integral with respect to a random measure that gives zero mass to the diagonal sets. with all these ingredients we prove a general hu meyer formula. the paper uses arduous combinatorics because of our need to work with stochastic multiple integrals with respect to the different powers of the jumps of the process, and such integrals can be conveniently handled through the lattice of the partitions of a finite set. as in the brownian case (see, e.g., @xcite), there are alternative methods to construct a multiple stratonovich integral based on approximation procedures, and it is possible to relax the conditions on the integrator process by assuming more regularity on the integrand function. such regularity is usually expressed in terms of the existence of _ traces _ of the function in a convenient sense. the advantage of using lvy processes with finite moments lies in the fact that simple @xmath10 estimates for the multiple stochastic integral of simple functions can be obtained, and then the multiple stratonovich integral can be defined in an @xmath11 space with respect to a measure that controls the behavior of the functions on the diagonal sets. in this way, the problem of providing a manageable definition of the traces is avoided. we would like to comment that an impressive body of work on multiple stochastic integrals with respect to lvy processes has been done by kallenberg, kwapien, krakowiak, rosinski, szulga, woyczinski and many others (see @xcite and the references therein). however, their approach is very different from ours, and assumes different settings to those used in this work. for this reason, we have only used a few results by those authors. the paper is organized as follows. in section [sec2] we review some combinatorics concepts and the basics of the stochastic measures as vector valued measures. in section [random] we introduce the random measures induced by a lvy process, and we identify the diagonal measures in such a case. in section [measure] we study the relationship between the product and it measures of a set, and we obtain a hu meyer formula for measures. in section [integral] we define the multiple it stochastic integral and the multiple stratonovich integral and also prove the general hu meyer formula for integrals. in section [sec6], as particular cases, we deduce the classical hu meyer formulas for the brownian motion and for the poisson process. we also study the case where the lvy process is a subordinator, and prove that both the multiple it stochastic integral and the multiple stratonovich integral can be computed in a pathwise sense. finally, in order to make the paper lighter, some of the combinatorial results are included as an.
Preliminaries
we need some notation of the combinatorics of the partitions of a finite set ; for details we refer to stanley @xcite, chapter 3, or rota and wallstrom @xcite. let @xmath12 be a finite set. a partition of @xmath12 is a family @xmath13 of nonvoid subsets of @xmath12, pairwise disjoint, such that @xmath14. the elements @xmath15 are called the _ blocks _ of the partition. denote by @xmath16 the set of all partitions of @xmath12, and write @xmath17 for @xmath18. given @xmath19, we write @xmath20 if each block of @xmath21 is contained in some block of @xmath22 ; we then say that @xmath21 is a refinement of @xmath22. this relationship defines a partial order that is called the _ reversed refinement order _, and it makes @xmath16 a lattice. we write @xmath23, which is the minimal element, and @xmath24 the maximal one. we say that a partition @xmath25 is of type @xmath26 if @xmath22 has exactly @xmath27 blocks with 1 element, exactly @xmath28 blocks with 2 elements, and so on. in the same way, for @xmath29 and @xmath30, we say that the segment @xmath31 $] is of type @xmath32 if there are exactly @xmath27 blocks of @xmath22 in @xmath21 ; there are exactly @xmath28 blocks of @xmath22 that each one gives rise to 2 blocks of @xmath21, etc. necessarily, @xmath33 in that situation, the mbius function of @xmath31 $] is @xmath34 we use the mbius inversion formula, that in the context of the lattice of the partitions of a finite set, says that for two functions @xmath35, @xmath36 if and only if @xmath37 (see @xcite, proposition 3.7.2). as we commented in the, we will introduce two random measures on a @xmath38-dimensional space, and the diagonal sets will play an essential role. diagonal sets can be conveniently described through the partitions of the set @xmath39. we use the notation introduced by rota and wallstorm @xcite. let @xmath40 be an arbitrary set, and consider @xmath41. given @xmath42, we write @xmath43 if @xmath44 and @xmath45 belong to the same block of @xmath22. put @xmath46 and @xmath47 the sets @xmath48 are called _ diagonal sets_. note that @xmath49 and @xmath50. for example, for @xmath51 and @xmath52, we have @xmath53 and @xmath54 the sets corresponding to the minimal and maximal partitions are specially important @xmath55 and @xmath56 if @xmath57, then @xmath58 the above notation @xmath59 is coherent with the reversed refinement order @xmath60 in particular, @xmath61. let @xmath62 be a complete probability space. in this paper, a random measure @xmath63 on a measurable space @xmath64 is an @xmath10-valued @xmath21-additive vector measure, that means, a map @xmath65 such that for every sequence @xmath66, such that @xmath67, @xmath68 the @xmath21-additive vector measures defined on a @xmath21-field inherit some basic properties of the ordinary measures, but not all. so, for a sake of easy reference, we write here a uniqueness property translated to our setting. the proof is the same as the one for ordinary measures. [monotona] let @xmath63 and @xmath69 be two random measures on @xmath64, and consider a family of sets @xmath70 closed under finite intersection and such that @xmath71. then @xmath72 assume that the measurable space @xmath64 satisfies that for every set @xmath73 and every @xmath42, we have @xmath74. as rota and wallstrom @xcite point out, this condition is satisfied if @xmath40 is a polish space and @xmath75 its borel @xmath21-algebra. we extend the definition of _ good random measure _ introduced by rota and wallstorm @xcite to a family of measures ; specifically, we say that the random measures @xmath76 over a measurable space @xmath64 are _ jointly good random measures _ if the finite additive product vector _ measure _ @xmath77 defined on the product sets by @xmath78 can be extended to a (unique) @xmath21-additive random measure on @xmath79. this extension, obvious for ordinary measures, is in general not transferred to arbitrary vector measures (see engel @xcite, masani @xcite and kwapien and woyczynski @xcite). given a good random measure @xmath63 (in the sense that the @xmath38-fold product @xmath80 satisfies the above condition), the starting point of rota and wallstrom (@xcite, definition 1) is to consider new random measures given by the restriction over the diagonal sets ; specifically, for @xmath42 they define @xmath81}_\pi(c):=\phi^{\otimes n}(c_{\pi})\qquad \mbox{for } c\in\mathcal{s}^{\otimes n}.\]] the following definitions are the extension of these concepts to a family of random measures. let @xmath82 be jointly good random measures on @xmath64. for a partition @xmath42, define @xmath83 and @xmath84 in agreement with the notation in rota and wallstrom @xcite, when @xmath85, we simply write @xmath86 for @xmath87 and @xmath88}_\pi$] for the corresponding measure given in ([def - ito]). since @xmath89, then @xmath90, that is the product measure. the measure @xmath91 is called _ the it multiple stochastic measure _ relative to @xmath82. as the ordinary multiple it integral, the it multiple stochastic measure gives zero mass to every diagonal set different from @xmath92 : [0diagonals] let @xmath42 such that @xmath93. for every @xmath73, we have @xmath94 from ([incompatibilitat]) we have @xmath95. the basic result of rota and wallstrom @xcite, proposition 1, is transferred to this situation : [strato - ito1] @xmath96 and @xmath97 where @xmath98 is the mbius function defined in section [partitions]. the equality ([mesura - prod]) is deduced from ([union]) and the definitions ([def - producte]) and ([def - ito]). the equality ([mesura - ito]) follows from ([mesura - prod]) and the mebius inversion formula ([moebius]).
Random measures induced by a levy process
let @xmath99 \}$] be a lvy process, that is, @xmath100 has stationary and independent increments, is continuous in probability, is cadlag and @xmath101. in all the paper we assume that @xmath100 has moments of all orders ; however, if the interest is restricted to multiple integral up to order @xmath102, then it is enough to assume that the process has moments up to order @xmath103. denote the lvy measure of @xmath100 by @xmath104, and by @xmath105 the variance of its gaussian part. the existence of moments of @xmath106 of all orders implies that @xmath107 and @xmath108. write @xmath109,\nonumber\\[-8pt]\\[-8pt] k_2&= & \sigma^2+\int_{\mathbb{r } } x ^2 \nu(dx) \quad\mbox{and}\quad k_n=\int_{\mathbb{r } } x ^n \nu(dx)<\infty,\qquad n\ge3.\nonumber\end{aligned}\]] from now on, take @xmath110 $] and @xmath111)$]. the basic random measure @xmath112 that we consider is the measure induced by the process @xmath100 itself, defined on the intervals by @xmath113s, t])=x_t - x_s,\qquad 0\le s\le t\le t,\]] and extended to @xmath114)$]. the measure @xmath112 is an independently scattered random measure, that is, if @xmath115)$] are pairwise disjoint, then @xmath116 are independent. the random measures induced by the powers of the jumps of the process, @xmath117, are also used. consider the _ variations _ of the process @xmath100 (see meyer @xcite, page 319) @xmath118_t=\sum_{0<s\le t } (\delta x_s) ^2+\sigma^2t,\\ x ^{(n)}_t&=&\sum_{0<s\le t } (\delta x_s) ^n,\qquad n \ge3. \nonumber\end{aligned}\]] the processes @xmath119 are lvy processes such that @xmath120=k_n t\qquad \forall n\ge1.\]] so, the centered processes, @xmath121 are square integrable martingales, called _ teugels martingales _ (see nualart and schoutens @xcite), with predictable quadratic covariation @xmath122 we denote by @xmath123 the random measure induced by @xmath124, and for @xmath125, @xmath126 (we indistinctly use both @xmath127 and @xmath112). every @xmath123 is a independently scattered random measure. for @xmath128)$], @xmath129=k_{n+m } \int_{a\cap b}dt+ k_nk_m \int_{a } dt\int _ { b}dt.\]] we stress the following property, which is the basis of all the paper, and is a consequence of theorem 10.1.1 by kwapien and woyczynski @xcite. for every @xmath130, the random measures @xmath131 are jointly good random measures on @xmath132^n,\mathcal{b } ([0,t]^n)) $]. rota and wallstrom @xcite define the diagonal measure of order @xmath38 of @xmath112 as the random measure on @xmath133 $] given by @xmath134).\]] to identify the diagonal measures is a necessary step to study the stochastic multiple integral. in the case of a random measure generated by a lvy process we show that the diagonal measures are the measures generated by the variations of the process. [diagonal - simple] for every @xmath135)$] and @xmath136, @xmath137 where @xmath123 is the random measure induced by @xmath124. since both @xmath138 and @xmath123 are random measures, by proposition [monotona] it is enough to check the equality for @xmath139 $]. consider an increasing sequence of equidistributed partitions of @xmath140 $] with the mesh going to 0 ; for example, take @xmath141 and let @xmath142 to shorten the notation, write @xmath143 instead of @xmath144. consider the sets @xmath145^n\cup(t_1,t_2] ^n\cup\cdots\cup(t_{2^m-1},t] ^n.\]] random measures are sequentially continuous and @xmath146^n_{\widehat1}$], when @xmath147, so we have that @xmath148)=\lim_m \sum_{k=0}^{2^m-1 } (\phi ((t_k, t_{k+1 }])) ^n= \lim_m \sum_{k=0}^{2^m-1 } (x_{t_{k+1}}-x_{t_{k } }) ^n\]] in @xmath10. for @xmath149, @xmath150_t=\phi_2 ((0,t]) \qquad \mbox{in probability},\]] so the proposition is true in this case. for @xmath151, by it s formula, @xmath152 \nonumber\\ \label{equa } & & \qquad = n\int_0^t \biggl(\sum_{k=0}^{2^m-1 } (x_{s-}- x_{t_{k } }) ^{n-1}\mathbf{1 } _ { (t_{k},t_{k+1}]}(s) \biggr) \,dx_s\\ \label{equb } & & \qquad\quad{}+\pmatrix{n\cr2}\int_0^t \biggl(\sum_{k=0}^{2^m-1 } (x_{s-}- x_{t_{k } }) ^{n-2}\mathbf{1}_{(t_{k},t_{k+1}]}(s) \biggr) \,d[x, x]_s\\ \label{equc } & & \qquad\quad{}+\sum_{j=3}^n\sum_{k=0}^{2^m-1}\sum_{t_{k}<s\le t_{k+1 } } \pmatrix{n\cr j } (x_{s-}- x_{t_{k } }) ^{n - j } (\delta x_s) ^j.\end{aligned}\]] for @xmath153, the corresponding term in ([equc]) is @xmath154}(s) \biggr) \,dx^{(j)}_s.\]] hence, ([equa]), ([equb]) and ([equd]) have the same structure @xmath155 where @xmath156}(s)$] is a predictable process and @xmath157 is a semimartingale. since @xmath158 is left continuous, @xmath159 moreover, @xmath160 and the process @xmath161 \}$] is cadlag and adapted, and as a consequence, it is prelocally bounded (see pages 336 and 340 in dellacherie and meyer @xcite). by the dominated convergence theorem for stochastic integrals (dellacherie and meyer @xcite, theorem 14, page 338), @xmath162 finally, for @xmath163, the term in ([equc]) is @xmath164, and the proposition is proved. diagonal measures associated to a random measure of the form @xmath165 are needed. this is an extension of the previous proposition, and it is a key result for the sequel. [diagonal - diversos] let @xmath130, @xmath102, and @xmath135)$]. then @xmath166 as in the proof of the last proposition and with the same notation, it suffices to prove that for all @xmath167 @xmath168) \]] in probability. this convergence follows from proposition [diagonal - simple] by polarization.
The humeyer formula: measures
the hu meyer formula gives the relationship between the product measure @xmath169 and the it stochastic measures @xmath170. in this section we obtain this formula for measures and in the next one we extend it to the corresponding integrals. the idea of hu meyer formula is the following. given @xmath171^n)$], we can decompose @xmath172 so @xmath173 next step is to express each @xmath174 as a multiple it stochastic measure. for example, take @xmath175 and @xmath176. then, @xmath177 and we will prove that @xmath178 that is, both the product measure and the product set on the last two variables collapse to produce a diagonal measure, and since @xmath179, we get an it measure. to handle in general this property, we need some notation. given a partition @xmath180 with blocks @xmath15, we can order the blocks in agreement with the minimum element of each block. when necessary, we assume that the blocks have been ordered with that procedure, and we simply say that @xmath15 are ordered. in that situation, we write @xmath181 we start considering a set @xmath182, with @xmath135)$], and later we extend the hu meyer formula to an arbitrary set @xmath183^n)$]. [hu - meyer1] let @xmath135)$]. then @xmath184 to prove this theorem we need two lemmas the first one is an invariance - type property of product measures under permutations. we remember some standard notation. we denote by @xmath185 the set of permutations of @xmath186. consider @xmath187. 1. for a partition @xmath180 with blocks @xmath15, we write @xmath188 for the partition with blocks @xmath189. note that in general the blocks @xmath190 are not ordered, even when @xmath191 are. 2. for a vector @xmath192, we write @xmath193 given @xmath194, we put @xmath195 [permutacio] let @xmath187 and @xmath130. then for every @xmath196^n)$], @xmath197 and @xmath198 define the vector measure @xmath199 for @xmath200, we have that @xmath201 and it is clear that @xmath202 then, equality ([permut1]) follows from proposition [monotona]. to prove ([permut2]), first note that, by definition, the it stochastic measure satisfies @xmath203 moreover @xmath204. so it suffices to prove ([permut2]) for a set @xmath205. denote by @xmath206 the @xmath207-algebra trace of @xmath114^n)$] with @xmath133^n_{\widehat0}$], which is composed by all sets @xmath92, with @xmath208^n)$]. this @xmath21-algebra is generated (on @xmath133^n_{\widehat0}$]) by the family of rectangles @xmath209, with @xmath210 pairwise disjoint. by proposition [monotona], we only need to check ([permut2]) for this type of rectangle, and the property reduces to ([permut1]). the next lemma is an important step in proving theorem [hu - meyer1]. to have an insight into its meaning, consider the following example : let @xmath51 and @xmath211. with a slight abuse of notation, we can write @xmath212 by theorem [diagonal - diversos], @xmath213 however, if you consider @xmath214, even though @xmath215 and @xmath21 have the same number of blocks with 1 element and the same number of blocks with 2 elements (they have the same type), the computation of @xmath216 is not so straightforward. the lemma gives such computation. its proof demands some combinatorial results and it is transferred to appendix [prova - lema]. [lema - basic] let @xmath130, @xmath180 with blocks @xmath15 (ordered), and @xmath217)$]. then @xmath218 proof of theorem [hu - meyer1] by proposition [strato - ito1], @xmath219}(a^{n}).\]] so it suffices to prove that @xmath220}(a^{n})={\mathrm{st}}_{\widehat0}^{\overline\sigma}(a^{\ # \sigma}).\]] by the second statement in proposition [strato - ito1] we have @xmath221}(a^{n})=\sum_{\pi\in[\sigma,\widehat1]}\mu (\sigma, \pi)\phi ^{\otimes n}_\pi(a^n)=\sum_{\pi\in[\sigma,\widehat1]}\mu(\sigma, \pi) \phi^{\otimes n } (a^n_{\ge\pi }).\]] by lemma [lema - basic], @xmath222 let @xmath223 be the blocks of @xmath180 (ordered) and write @xmath224 the partition @xmath225 $], with blocks @xmath226, induces a unique partition of @xmath227, with blocks @xmath228 such that @xmath229 (see proposition [bijeccio] in the). hence, for @xmath230, @xmath231 thus, from ([hu - form3]) and lemma [lema - basic], @xmath232 by ([hu - form2]) and ([hu - form4]) using again the bijection between @xmath233 $] and @xmath234 stated in proposition [bijeccio] in the, and proposition [strato - ito1], we obtain @xmath235}(a^{n})&=&\sum_{\pi\in[\sigma,\widehat1]}\mu (\sigma, \pi) (\phi_{s_1}\otimes\cdots\otimes\phi_{s_m }) (a^m_{\ge\pi^*})\\ & = & \sum_{\rho\in\pi_{m}}\mu(\widehat0,\rho) (\phi_{s_1}\otimes\cdots\otimes\phi_{s_m }) (a^m_{\ge\rho }) = { \mathrm{st}}_{\widehat 0}^{\overline\sigma}(a^{\#\sigma}).\end{aligned}\]] in order to extend the hu meyer formula for a general set in @xmath236^n)$], we use a set function to express for an arbitrary set the contraction from @xmath237 to @xmath238. that is, given a partition @xmath180, with blocks @xmath223 ordered, we want to contract a set @xmath208^n)$] into a set of @xmath114^{\ # \sigma})$] according to the structure of the @xmath21-diagonal sets. with this purpose, define the function @xmath239^{\#\sigma } & \longrightarrow & [0,t]^n,\nonumber\\[-8pt]\\[-8pt] (x_1,\ldots, x_m) & \longrightarrow & (y_{1},\ldots, y_{n }), \nonumber\end{aligned}\]] where @xmath240, if @xmath241. for example, if @xmath51 and @xmath242, @xmath243 note that @xmath244 see appendix [q - sigma] for more details. [hu - meyer2] let @xmath208^n)$]. then @xmath245 we separate the proof in two steps. in the first one, we show that it is enough to prove the theorem for a rectangle of the form @xmath246 where @xmath247 are pairwise disjoint. in the second step we check formula ([hu - meyer2-form]) for those rectangles. _ first step. _ by proposition [monotona], it suffices to prove the theorem for a rectangle @xmath209. since every rectangle can be written as a disjoint union of rectangles such that every two components are either equal or disjoint, we consider one of this rectangles, @xmath200, where for every @xmath248, @xmath249 or @xmath250. now we show that the formula ([hu - meyer2-form]) applied to @xmath251 is invariant by permutations : specifically, we see that for any permutation @xmath187 @xmath252 the first equality is deduced from ([permut1]). for the second one, applying proposition [perm - delta](i), we have @xmath253 where @xmath254 is the permutation that gives the correct order of the blocks of @xmath188 (see the lines before proposition [perm - delta]). by lemma [permutacio] @xmath255 where the last equality is due to the fact that @xmath256 by the definition of @xmath257 [see ([barra2])]. finally, @xmath258 because we are adding over all the set @xmath259. _ second step. _ consider @xmath260 with @xmath261 pairwise disjoint and @xmath262. by theorem [hu - meyer1], @xmath263 where the last equality is due to the fact that @xmath264 and the definition of the it measure @xmath265. let @xmath266 be the partition with blocks @xmath267 there is a bijection between the elements @xmath180, with @xmath268, and @xmath269 such that @xmath270 where we use equality ([projector2]) in the. then, @xmath271 where the last equality is due to the fact that if @xmath272, then @xmath273 [see ([projector2])].
Multiple ito and stratonovich integral, and the corresponding humeyer formula
we extend theorem [hu - meyer2] to integrals with respect to the random measures involved. we first define an it - type multiple integral and an integral with respect to the product measure. we generalize the multiple it integral with respect to the brownian motion (it @xcite ; see also @xcite) to a multiple integral with respect to the lvy processes @xmath274. as we will prove, that integral can be interpreted as the integral with respect to the it stochastic measure. the ideas used to construct this integral are mainly it s ; however, the fact that these processes (in general) are not centered obstructs the classical isometry property, being substituted by an inequality. write @xmath275^n,\mathcal{b}([0,t]^n), (dt)^{\otimes n})$]. denote by @xmath276 the set of the so - called it - elementary functions, having the form @xmath277 where @xmath278)$] are pairwise disjoint, and @xmath279 is zero if two indices are equal. it is well known (see it @xcite) that @xmath276 is dense in @xmath280. consider @xmath281 and define the multiple it integral of @xmath282 with respect to @xmath274 by @xmath283 let @xmath281 and @xmath284. then @xmath285\le\alpha_{\mathbf{r}}\int _ { [0,t]^n}f^2(t_1,\ldots, t_n) \,dt_1\cdots dt_n,\]] where @xmath286 is a constant that depends on @xmath287 but not on @xmath282. the proof follows exactly the same steps as that of theorem 4.1 in engel @xcite. the key point is that the measures @xmath288 can be written as @xmath289 where @xmath290 is the centered and independently scattered random measure corresponding to @xmath291. the extension of the multiple it stochastic integral to @xmath280, stated below, is proved as in the brownian case (see it @xcite). [ito - ext - th] the map @xmath292 can be extended to a unique linear continuous map from @xmath280 to @xmath10. in particular, @xmath293 satisfies the inequality @xmath294\le\alpha_{\mathbf{r}}\int _ { [0,t]^n}f^2(t_{1},\ldots, t_{n }) \,dt_1\cdots dt_n.\]] as in the brownian case, it is useful to express the multiple integral in terms of iterated integrals of the form @xmath295 where @xmath296 is a permutation of @xmath186. this integral is properly defined for @xmath297. this can be checked using the decomposition of @xmath298 as a special semimartingale @xmath299, where, as we said in section [random], @xmath291 is a square integrable martingale with predictable quadratic variation @xmath300. the previous iterated integral then reduces to a linear combination of iterated integrals of type @xmath301 being @xmath302 either @xmath9 or @xmath303. hence, at each iteration, the integrability condition @xmath304<\infty\]] of a predictable process @xmath305 with respect to @xmath306 can be easily verified. next proposition gives the precise expression of the multiple integral as a sum of iterated integrals. since we are integrating with respect to different processes, we need to separate the space @xmath133^n$] into simplexs. let @xmath297. then @xmath307 where @xmath308, and the integrals on the right - hand side are interpreted as iterated integrals. by linearity and density arguments, it suffices to consider a function @xmath309 where @xmath310 $] are pairwise disjoint, and a computation gives the result. when @xmath311, we write @xmath312 instead of @xmath313 ; in that case, the multiple it integral enjoys nicer properties. [nicer] 1. let @xmath297. then @xmath314 where @xmath315 is the symmetrization of @xmath282 @xmath316 2. assume @xmath317=0 $]. for @xmath318, @xmath319=\delta_{n, m } k_2^n n!\int _ { [0,t]^n}\widetilde{f } \widetilde g \,d\mathbf{t},\]] where @xmath320, and 0 otherwise. 3. let @xmath297 be a symmetric function. then @xmath321 we now state the relationship between the it stochastic measure @xmath170 and the it multiple integral @xmath322. [stochastic - int] let @xmath208^n)$] and @xmath323. then @xmath324 by ([desigualtat - integral]) the map @xmath325 defines a vector measure on @xmath114^n)$]. on the left - hand side of ([ito - int - mes]), the it measure satisfies @xmath326 now, look at right - hand side of ([ito - int - mes]). for @xmath327 we have that @xmath328^n_\pi$]. for all @xmath93, @xmath329\le \alpha_n\int_{[0,t]^n } \mathbf{1}_{c_\pi}\,dt_1\cdots dt_n\le\alpha _ n\int _ { [0,t]^n } \mathbf{1}_{[0,t]^n_\pi } \,dt_1\cdots dt_n=0.\]] hence, @xmath330 from ([f-0]) and ([f-00]), it suffices to prove ([ito - int - mes]) for a set @xmath205. as in the proof of the second part of lemma [permutacio], this can be reduced to check that equality for a rectangle @xmath331\times\cdots\times(s_n, t_n]$], with the intervals pairwise disjoint. this follows from the fact that both sides of ([ito - int - mes]) are equal to @xmath332)\cdots\phi_{r_n}((s_n, t_n])$]. the property @xmath333 is lost when the integrators are different. however, from proposition [stochastic - int] and ([permut2]) we can deduce the following useful property : [perm - int] let @xmath297, and @xmath323, where @xmath334. consider @xmath187. then @xmath335 given a map @xmath336^n\to{\mathbb{r}}$], the integral with respect to the product measure @xmath169 is called the multiple stratonovich integral, and denoted by @xmath6. its basic property is that the integral of a product function factorizes @xmath337 where @xmath338 in order to construct this integral, we consider ordinary simple functions of the measurable space @xmath339^n,\mathcal{b}([0,t]^n))$]. specifically, denote by @xmath340 the set of functions with the form @xmath341 where @xmath342^n), i=1,\ldots, k$]. for such @xmath282, define the multiple stratonovich integral by @xmath343 the integral of a simple function does not depend on its representation, and it is linear. moreover, [hu - meyer - funct - prop] let @xmath344. then we have the hu meyer formula @xmath345 where the function @xmath346^{\#\sigma } \to[0,t]^n$] is introduced in ([qsigma]), @xmath347 is the vector whose components are the sizes of the ordered blocks of @xmath21, and @xmath348 is the multiple it integral of order @xmath349 with respect to the measures @xmath350. by linearity, it suffices to consider @xmath351, where @xmath208^n)$]. a generic term on the right - hand side of ([hu - meyer - funct1]) is @xmath352 and @xmath353 hence, by proposition [stochastic - int], @xmath354 and ([hu - meyer - funct1]) follows from theorem [hu - meyer2]. let @xmath180, with @xmath355, and denote by @xmath356 the image measure of the lebesgue measure @xmath357 by the function @xmath346^{m } \to[0,t]^n$]. the image measure theorem implies that for @xmath358^n\to{\mathbb{r}}$] measurable, positive or @xmath359-integrable, @xmath360^n } f(t_1,\ldots, t_n) \,d\lambda_\sigma(t_1,\ldots, t_n)= \int_{[0,t]^m } f(q_\sigma(t_1,\ldots, t_m)) \,dt_1 \cdots dt_m.\hspace*{-28pt}\]] define on @xmath114^n)$] the measure @xmath361 and write @xmath362 for @xmath363^n,\mathcal { b}([0,t]^n),\lambda_n)$]. in order to extend the multiple stratonovich integral we need the following inequality of norms : let @xmath344. then @xmath364\le c\int_{[0,t]^n}f^2\, d\lambda_n,\]] where @xmath251 is a constant. by ([hu - meyer - funct1]), ([desigualtat - integral]) and ([image]), @xmath365 & \le & c \sum_{\sigma\in\pi_n } { \mathbb{e}}\bigl [\bigl(i_{\#\sigma}^{\overline\sigma}(f\comp q_\sigma) \bigr)^2 \bigr]\\ & \le & c \sum_{\sigma\in\pi_n}\int_{[0,t]^{\#\sigma}}(f\comp q_\sigma)^2\, dt_1\cdots dt_{\#\sigma}\\ & = & c \sum_{\sigma\in\pi_n}\int_{[0,t]^{n}}f^2 \,d\lambda_\sigma\\ & = & c\int _ { [0,t]^{n}}f^2 \,d\lambda_n.\end{aligned}\]] the main result of the paper is the following theorem : the map @xmath366 can be extended to a unique linear continuous map from @xmath362 to @xmath10, and we have the hu meyer formula @xmath367 the extension of @xmath368 to a continuous map on @xmath362 is proved using a density argument and inequality ([desigualtat - strato]). to prove the hu meyer formula, let @xmath369 and @xmath370 such that @xmath371 in @xmath362. for every @xmath180, we have @xmath372 in @xmath373 ; hence, from theorem [ito - ext - th] the it integrals on the right - hand side of ([hu - meyer - funct]) converge, and the formula follows from proposition [hu - meyer - funct - prop]. 1. let @xmath374,dt)$]. then @xmath375 and @xmath376 this result is easily checked for simple functions @xmath377 and extended to the general case by a density argument. 2. in order to prove the hu meyer formula for @xmath368 it is enough to assume that the process @xmath100 has moments up to order @xmath103. 3. for @xmath378, the measure @xmath359 is singular with respect to the lebesgue measure on @xmath133^n$]. for example, for @xmath149 and @xmath379, let @xmath380\}$] be the diagonal of @xmath133 ^ 2 $]. then @xmath381 is concentrated in @xmath382, that has zero lebesgue measure, but @xmath381 is nonzero @xmath383 ^ 2}\mathbf{1}_{d}(s, t) \,d\lambda _ { \widehat 1}(s, t)=\int _ { [0,t]}\mathbf{1}_{d}(t, t) \,dt = t.\]] 4. as in the brownian case (see @xcite and the references therein), there are other procedures to construct the multiple stratonovich integral. the main difficulty in every approach is that the usual condition @xmath297 in it s theory is not sufficient to guarantee the multiple stratonovich integrability of @xmath282. the reason is that one needs to control the behavior of @xmath282 on the diagonal sets @xmath133_\sigma^n$] that have zero lebesgue measure when @xmath384. we solve this difficulty using the norm induced by the measure @xmath385, which seems to be appropriate for dealing with the diagonal sets, avoiding in this way the difficulty of a manageable definition of the _ traces_. when the function @xmath369 is symmetric, the hu meyer formula can be considerably simplified. we show that we can assume that symmetry on @xmath282 without loss of generality. let @xmath369. then @xmath386, where @xmath315 is the symmetrization of @xmath282 [see ([simetrizacio])]. the proof is straightforward for @xmath387, @xmath171^n)$], using lemma [permutacio]. by linearity the equality @xmath388 is extended to @xmath389, and by density to @xmath362. next we show the hu meyer formula for a symmetric function @xmath282. in general (for @xmath282 symmetric), the function @xmath390 is nonsymmetric, but as we will see in the proof of the next theorem, its multiple it integral depends only on the block structure of @xmath21 (the type of @xmath21). for example, with @xmath391, @xmath392 and @xmath393, we have that @xmath394 that is nonsymmetric. its integral is @xmath395 take @xmath396. then @xmath397 and @xmath398 where the last equality is due to proposition [perm - int]. we use the following notation : given nonnegative integers @xmath399 such that @xmath400, we write @xmath401=(\underbrace{1,\ldots,1}_{r_1},\underbrace { 2,\ldots, 2}_{r_2},\ldots).\]] note that this corresponds to @xmath402 when @xmath403 we also write @xmath404 for @xmath405, with @xmath21 the above partition. let @xmath406 be a symmetric function. then @xmath407 } (f\comp q_{r_1,\ldots, r_k }), \]] where the sum is extended over all nonnegative integers @xmath408 such that @xmath400, for @xmath409. let @xmath369 symmetric. for every @xmath180 and @xmath187, @xmath410 where ([equa]) is due to proposition [perm - delta](ii), the equality ([equb]) follows from the symmetry of @xmath282 and ([equc]) from proposition [perm - int] and the fact that @xmath257 gives the correct order of @xmath188 [see ([barra2])]. this implies that all the partitions that have the same number of blocks of 1 element, the same number with two elements, etc. (i.e., they have the same type) give the same it multiple integral in the hu meyer formula. to obtain ([hu - meyer - simetric]) it suffices to count the number of partitions of @xmath39 with @xmath27 blocks with 1 element, @xmath28 blocks with 2 elements@xmath411 blocks with @xmath412 elements, which is @xmath413 _ final remark. _ one may expect that by decomposing the lvy process into a sum of two independent processes, one with the small jumps and the other with the large ones, the assumption of the existence of moments could be avoided. however, this decomposition introduces dramatic changes to the context of the work, and such an extension is beyond the scope and purposes of the present paper.
Special cases
when @xmath414 is a standard brownian motion, @xmath415)=t \quad\mbox{and}\quad \phi_n=0,\qquad n\ge3.\vadjust{\goodbreak}\]] it follows that in the hu meyer formula only the partitions with all blocks of cardinality 1 or 2 give a contribution, and all the it integrals are a mixture of multiple stochastic brownian integrals and lebesgue integrals. we can organize the sum according the number of blocks of two elements. for a partition having @xmath45 blocks of @xmath416 elements, and @xmath406 symmetric, the multiple it integral is @xmath417}_{n - j}(f)\\ & & \qquad=\int_{[0,t]^{n - j}}f(s_1,\ldots, s_{n-2j},\\ & & \hspace*{84.4pt}t_1,t_1,\ldots, t_{j},t_{j }) \,dw_{s_1}\cdots dw_{s_{n-2j } } \,d t_{1}\cdots d t_{j}\\ & & \qquad = i_{n-2j } \biggl(\int_{[0,t]^{j}}f(\bolds\cdot, t_1,t_1,\ldots, t_{j},t_{j }) \,d t_{1}\cdots d t_{j } \biggr),\end{aligned}\]] where the last equality is due to a fubini - type theorem. therefore, @xmath418}\frac{n!}{(n-2j)!j!2^j}\nonumber\\[-8pt]\\[-8pt] & & \hspace*{17.6pt}{}\times i_{n-2j } \biggl(\int_{[0,t]^{j}}f(\bolds\cdot, t_1,t_1,t_2,t_2,\ldots, t_{j},t_{j})\, d t_{1}\cdots d t_{j } \biggr),\nonumber\end{aligned}\]] which is the classical hu meyer formula (see @xcite). on the other hand, in the measure @xmath385 only participate the measures @xmath359 corresponding to the partitions above mentioned. consider the measure @xmath419 on @xmath133 ^ 2 $], that is, for a positive or @xmath420 integrable function @xmath421, @xmath422 ^ 2}h(s, t) \,d\ell_2(s, t)=\int_{[0,t]}h(t, t)\,dt.\]] given the partition @xmath180, @xmath423 we have @xmath424 let @xmath425 be a standard poisson process with intensity 1, and consider the process @xmath426. for every @xmath102, @xmath427 and hence, a multiple it integral can be reduced to a linear combination of multiple integrals where all the integrators are @xmath428 or @xmath429. for @xmath406 symmetric, each integral @xmath430 } (f\comp q_{r_1,\ldots, r_k})$] in ([hu - meyer - simetric]) can be expressed in terms of the number of lebesgue integrals that appear @xmath431 } (f\comp q_{r_1,\ldots, r_k})\\ \hspace*{-5pt}&&\qquad=\sum_{j=0}^{r_2+\cdots+r_k}\hspace*{-0.9pt}i_{r_1+\cdots+r_k - j } \biggl (\int_{[0,t]^j } \biggl(\mathop{\sum_{l_1,\ldots, l_j = r_1 + 1}}_{\mathrm { different}\mbox { } } ^{r_1+\cdots+r_k}\hspace*{-0.9pt } (f\comp q_{r_1,\ldots, r_k})\\ \hspace*{-5pt}&&\hspace*{204.5pt}{}\times (t_1,\ldots, t_{r_1+\cdots+r_k }) \biggr)\,dt_{l_1}\cdots dt_{l_j } \biggr),\end{aligned}\]] and the hu meyer formula of sol and utzet @xcite can be deduced from this expression. a subordinator is a lvy process with increasing paths. an important example of a subordinator with moments of all orders is the gamma process, denoted by @xmath432, which is the lvy process corresponding to an exponential law of parameter 1. its lvy measure is @xmath433 the law of @xmath434 is gamma with mean @xmath9 and scale parameter equal to one. a gamma process can be represented as the sum of its jumps, that are all positive, @xmath435 the lvy measure of @xmath436 is (see schoutens @xcite) @xmath437 and the teugels martingales are @xmath438 in this case, unlike the brownian motion and the poisson process, the hu meyer formula does not simplify, due to the fact that the diagonal measures can not be expressed in a simple way in terms of, say, the process and a deterministic measure. however, for a gamma process, and in general, for a subordinator without drift (see below for the definition) with moments of all orders, both the multiple it and stratonovich integrals can be computed pathwise integrating with respect to an ordinary measure. this is a multivariate extension of the property that states that the stochastic integral and the pathwise lebesgue stieljes integral with respect to a semimartingale of bounded variation are equal ; such property was proved for the integral with respect to a lvy process of bounded variation by millar @xcite under weak conditions, and part of our proof follows his scheme. let @xmath439 be a subordinator. the lvy it representation of @xmath100 takes the form @xmath440 with @xmath441 (see sato @xcite, theorems 21.5 and 19.3). the number @xmath442 is called the _ drift _ of the subordinator, and we will assume that @xmath443. consider the sequence of stopping times @xmath444 with disjoint graphs that exhaust the jumps of @xmath445, and @xmath100 only has jumps on these times (see, e.g., dellacherie and meyer @xcite, theorem b, page xiii, for a construction of this sequence). denote by @xmath446 the set of @xmath38-tuples @xmath447, with @xmath448, and all entries different. for @xmath130, define a measure on @xmath133^n$] by @xmath449 where @xmath450 is a dirac measure at point @xmath451, with the convention that the sum is 0 if @xmath452. we have the following property : let @xmath439 be a subordinator without drift and with moments of all orders. with the preceding notation, for every @xmath297, @xmath453^n}f \,d m_{r_1,\ldots, r_n}\qquad \mbox{a.s.}\]] first, note two facts : \(a) @xmath454 is a finite measure @xmath455^n)&=&\sum_{(t_{i_1},\ldots, t_{i_n})\in j_n } (\delta x_{t_{i_1 } }) ^{r_1}\cdots(\delta x_{t_{i_n } }) ^{r_n}\\ & \le&\sum_{t_{i_1}\le t,\ldots, t_{i_n}\le t } (\delta x_{t_{i_1 } }) ^{r_1}\cdots(\delta x_{t_{i_n } }) ^{r_n}\\ & = & x_t^{(r_1)}\cdots x_t^{(r_n)}<\infty.\end{aligned}\]] \(b) if the intervals @xmath331,\ldots,(s_n, t_n]$] are pairwise disjoint, then ([path]) is true for @xmath456\times\cdots \times (s_n, t_n]}$]. the proof is straightforward. we separate the proof of the proposition in two steps. _ step _ 1. formula ([path]) is true for every map @xmath457^n\to { \mathbb{r}}$] @xmath458-measurable and bounded, where @xmath458 is the @xmath21-field on @xmath133^n$] generated by the rectangles @xmath331\times\cdots\times(s_n, t_n]$], with @xmath331,\ldots,(s_n, t_n]$] pairwise disjoint. to prove this claim we use a convenient monotone class theorem. denote by @xmath459 the family of functions that satisfy ([path]) ; it is a vector space such that : @xmath460. if @xmath461, @xmath462 for some constant @xmath463, and @xmath464, then @xmath465. to see (i), consider the dyadic partition of @xmath133 $] with mesh @xmath466, write @xmath467,\qquad j=1,\ldots, 2^k,\]] and define @xmath468 by the remark (b) at the beginning of the proof, @xmath469^n}f_k \,d m_{r_1,\ldots, r_n}.\]] moreover, @xmath470 out off the diagonal sets @xmath133^n_\sigma$], with @xmath471, and then @xmath470 a.e. with respect to the lebesgue measure, and in @xmath280. therefore, @xmath472 on the other hand, for every @xmath473, the measure @xmath454 does not charge on any of the above mentioned diagonal sets. thus, the convergence @xmath470 is also @xmath454-a.e. by the monotone convergence theorem, @xmath474^n}f_k \,d m_{r_1,\ldots, r_n}=\int_{[0,t]^n}1\, d m_{r_1,\ldots, r_n},\]] and (i) follows. point (ii) is deduced directly from the monotone convergence theorem and taking into account that under the conditions in (ii) we have @xmath475 in @xmath280. again by remark (b) above, the indicator of a set @xmath331\times \cdots\times(s_n, t_n]$], with @xmath331,\ldots,(s_n, t_n]$] pairwise disjoint, is in @xmath459, and this family of sets is closed by intersection. by the monotone class theorem, it follows that all bounded @xmath476-measurable functions are in @xmath459. _ step _ 2. extension of ([path]) to all @xmath297. first, note that @xmath458 is the @xmath21-field generated by the borelian sets @xmath477^n)$] such that @xmath478^n_{\widehat0}$]. then, given @xmath477^n)$], the indicator @xmath479^n_{\widehat0}}$] is @xmath458 measurable. let @xmath480, and assume @xmath481. there is a sequence of simple (and then bounded) functions such that @xmath482. define @xmath483^n_{\widehat0}}$], which is @xmath458 measurable, and @xmath484 a.e. with respect to the lebesgue measure. the convergence is also in @xmath280, and then @xmath485. on the other hand, @xmath484, @xmath454-a.e. so @xmath486^n}f_m^0 \,d m_{r_1,\ldots, r_n}= \int_{[0,t]^n}f \,d m_{r_1,\ldots, r_n}.\]] by step 1, we get the result. for a general @xmath297, decompose @xmath487. finally, for a subordinator without drift and with moments of all orders, the multiple stratonovich measure can be identified with the @xmath38-fold product measure of @xmath488. so for @xmath489, by definition, @xmath490^n}f \,d\phi^{\otimes n}.\]] using similar arguments as in the previous proposition, but easier, it is proved that @xmath490^n}f \,d\phi^{\otimes n}\qquad \forall f\in l^2(\lambda_n).\]] then, the hu meyer formula can be transferred to a pathwise context.
Appendix
fix a partition @xmath180, with blocks @xmath15. let @xmath491, with blocks @xmath492 ; each block @xmath493 is the union of some of the blocks @xmath15. hence, we can consider the partition @xmath227 that gives the relationship between the @xmath493 s and the @xmath494 s, that is, @xmath495 has blocks @xmath228 defined by @xmath496 [bijeccio] let @xmath180 with @xmath355. with the above notation, the map @xmath497&\longrightarrow & \pi_m,\\ \pi&\mapsto&\pi^*\end{aligned}\]] is a bijection and, for @xmath498 $], @xmath499 moreover, @xmath500 where @xmath501 is the mbius function on @xmath502. \1. for a subset @xmath505 we denote by @xmath506 the image of @xmath507 by @xmath508 @xmath509 given a partition @xmath180, with blocks @xmath223, let @xmath188 be the partition with blocks @xmath510 defined by @xmath511. note that in general the blocks @xmath190 are not ordered. the application @xmath512 is a bijection and for @xmath513, @xmath514 this last property is clear, because if @xmath515, and @xmath516, then @xmath517 \2. for a vector @xmath192, we write @xmath519 and the application @xmath520 determines a bijection on @xmath504, that we also denote by @xmath508. for a set @xmath194, we write @xmath521 in particular, for @xmath522, @xmath523 notice that if we look for the position of a particular set, say @xmath524, in @xmath525, we find it at place @xmath526 @xmath527 this last observation gives some light to the next property : \(i) let @xmath532. write @xmath533 therefore, @xmath534 the condition on the right is equivalent to @xmath535 [see ([equivalencia])]. so, returning to the @xmath536 s, @xmath537 call @xmath538 and @xmath539. we have @xmath540 hence, @xmath541. consider a partition @xmath180 with blocks @xmath15 (ordered). if the elements of each block are consecutive numbers, then, @xmath543 when @xmath21 does not fulfill the previous condition, the expression ([descomposicio]) is not valid. however, since we are interested in computing @xmath544, thanks to lemma [permutacio], we fortunately can permute both the set and the product measure to make things work. the next proposition is essential for this purpose. let @xmath223 be the blocks of @xmath21 ordered. if @xmath21 is such that @xmath553, by theorem [diagonal - diversos], @xmath554 for the general case, let @xmath508 be the permutation given by proposition [particio] and write @xmath555. by proposition [particio] (first), and @xmath556 (second), we have @xmath557 by lemma [permutacio] and the first part of the proof, @xmath558 where @xmath559. for every @xmath560, @xmath561 given a partition @xmath180, with blocks @xmath15 (ordered), the function @xmath405 [see ([qsigma])] is defined by @xmath562^m & \longrightarrow & [0,t]^n,\\ (x_1,\ldots, x_m) & \to & (y_{1},\ldots, y_{n}),\end{aligned}\]] where @xmath240, if @xmath241. this function is a bijection between @xmath133^m$] and @xmath133^n_{\ge\sigma}$], and it is borel measurable because @xmath563 given a partition @xmath180, with blocks (ordered) @xmath564 and a permutation @xmath187, as we commented, the blocks of @xmath188 in general are not ordered. it is convenient to consider the permutation @xmath565 that gives the correct order of the blocks of @xmath188, that means, @xmath566 if @xmath567 is the first block of @xmath188, @xmath568 if @xmath569 is the second block, and so on ; in other words, @xmath570 are the blocks of @xmath188 ordered. remember that we defined [see ([barra])] the @xmath349-dimensional vector @xmath571. | in the framework of vector measures and the combinatorial approach to stochastic multiple integral introduced by rota and wallstrom [_ ann.
probab. _ * 25 * (1997) 12571283], we present an it multiple integral and a stratonovich multiple integral with respect to a lvy process with finite moments up to a convenient order. in such a framework,
the stratonovich multiple integral is an integral with respect to a product random measure whereas the it multiple integral corresponds to integrate with respect to a random measure that gives zero mass to the diagonal sets.
a general hu meyer formula that gives the relationship between both integrals is proved.
as particular cases, the classical hu meyer formulas for the brownian motion and for the poisson process are deduced.
furthermore, a pathwise interpretation for the multiple integrals with respect to a subordinator is given.
, and . . | 0802.3112 |
Introduction
the _ kissing number _ @xmath6 is the highest number of equal non - overlapping spheres in @xmath7 that touch another sphere of the same size. in three dimensions the kissing number problem asks how many white billiard balls can _ kiss _ (i.e. touch) a black ball. the most symmetrical configuration, 12 balls around another, is achieved if the 12 balls are placed at positions corresponding to the vertices of a regular icosahedron concentric with the central ball. however, these 12 outer balls do not kiss one another and each may be moved freely. this space between the balls prompts the question : _ if you moved all of them to one side, would a 13th ball fit? _ this problem was the subject of the famous discussion between isaac newton and david gregory in 1694. most reports say that newton believed the answer was 12 balls, while gregory thought that 13 might be possible. this problem is often called the _ thirteen spheres problem_. the problem was finally solved by schtte and van der waerden in 1953 @xcite. a subsequent two - page sketch of an elegant proof was given by leech @xcite in 1956. leech s proof was presented in the first edition of the well - known book by aigner and ziegler @xcite ; the authors removed this chapter from the second edition because a complete proof would have to include too much spherical trigonometry. the thirteen spheres problem continues to be of interest, and new proofs have been published in the last several years by hsiang @xcite, maehara @xcite (this proof is based on leech s proof), brczky @xcite, anstreicher @xcite, and musin @xcite. note that for @xmath8, the kissing number problem is currently solved only for @xmath9 @xcite, and for @xmath10 @xcite (see @xcite for a beautiful exposition of this problem). if @xmath0 unit spheres kiss the unit sphere in @xmath7, then the set of kissing points is an arrangement on the central sphere such that the (euclidean) distance between any two points is at least 1. this observation allows us to state the kissing number problem in another way : _ how many points can be placed on the surface of @xmath11 so that the angular separation between any two points be at least @xmath12? _ it leads to an important generalization. a finite subset @xmath13 of @xmath11 is called a _ spherical @xmath14-code _ if for every pair @xmath15 of @xmath13 with @xmath16 its angular distance @xmath17 is at least @xmath14. let @xmath13 be a finite subset of @xmath18. denote @xmath19 the set @xmath13 is then a spherical @xmath20-code. denote by @xmath21 the largest angular separation @xmath20 with @xmath22 that can be attained in @xmath18, i.e. @xmath23 in other words, _ how are @xmath0 congruent, non - overlapping circles distributed on the sphere when the common radius of the circles has to be as large as possible? _ this question is also known as the problem of the `` inimical dictators '', namely _ where should @xmath0 dictators build their palaces on a planet so as to be as far away from each other as possible? _ the problem was first asked by the dutch botanist tammes @xcite (see (*??? * section 1.6 : problem 6)), while examining the distribution of openings on the pollen grains of different flowers. the tammes problem is presently solved for several values of @xmath0, namely for @xmath1 by l. fejes tth @xcite ; for @xmath2 by schtte and van der waerden @xcite ; for @xmath3 by danzer @xcite (for @xmath24 see also brczky @xcite) ; and for @xmath4 by robinson @xcite. the tammes problem for @xmath5 is of particular interest due to its relation to both the kissing problem and the kepler conjecture @xcite. actually, this problem is equivalent to _ the strong thirteen spheres problem _, which asks to find the maximum radius of and an arrangement for 13 equal size non - overlapping spheres in @xmath25 touching the unit sphere. it is clear that the equality @xmath26 implies @xmath27. brczky and szab @xcite proved that @xmath28. bachoc and vallentin @xcite have shown that @xmath29. we solved the tammes problem for @xmath5 in 2012 @xcite. we proved that + _ the arrangement @xmath30 of 13 points in @xmath31 is the best possible, the maximal arrangement is unique up to isometry, and @xmath32. _ in this paper, using very similar method we present a solution to the tammes problem for @xmath33. we note that there is an arrangement of 14 points on @xmath18 such that the distance between any two points of the arrangement is at least @xmath34 (see (*??? 4) and http://neilsloane.com/packings/dim3/pack.3.14.txt). this arrangement is shown in fig. [fig1]. and its contact graph @xmath35. @xmath36.,title="fig : "] and its contact graph @xmath35. @xmath36.,title="fig : "] the first upper bound @xmath37 was found in @xcite. actually, this value is the famous fejes tth bound @xmath38 for @xmath33. brczky and szab @xcite improved the fejes tth bound and proved that @xmath39. bachoc and vallentin @xcite using the sdp method have shown that @xmath40.
Main theorem
the arrangement @xmath41 of @xmath42 points in @xmath31 gives a solution of the tammes problem, moreover the maximal arrangement for @xmath33 is unique up to isometry and @xmath43. * contact graphs. * let @xmath13 be a finite set in @xmath44. the _ contact graph _ @xmath45 is the graph with vertices in @xmath13 and edges @xmath46 such that @xmath47. * shift of a single vertex. * we say that a vertex @xmath48 _ can be shifted _, if, in any open neighbourhood of @xmath49 there is a point @xmath50 such that @xmath51 where for a point @xmath52 and a finite set @xmath53 by @xmath54 we denote the minimum distance between @xmath55 and points in @xmath56. * danzer s flip. * danzer (*??? 1) defined the following flip. let @xmath57 be vertices of @xmath45 with @xmath58. we say that @xmath49 is flipped over @xmath59 if @xmath49 is replaced by its mirror image @xmath60 relative to the great circle @xmath59 (see fig. [fig3]). we say that this flip is _ danzer s flip _ if @xmath61. * irreducible contact graphs. * we say that a graph @xmath45 is _ irreducible _ provided it does not allow danzer s flip and no vertex in @xmath13 can be shifted. the concept of irreducible contact graphs was invented by schtte - van der waerden @xcite, fejes tth @xcite, and danzer @xcite. actually, in these papers as well as in our paper @xcite this concept has been used for solutions of the tammes problem with @xmath62. recently, we enumerated all of the irreducible contact graphs (with and without danzer s flip) for @xmath63 @xcite. * @xmath41 and @xmath35. * denote by @xmath41 the arrangement of 14 points in fig. 1. let @xmath64. it is not hard to see that the graph @xmath35 is irreducible. * maximal graphs @xmath65. * let @xmath13 be a subset of @xmath31 with @xmath66 and @xmath67. denote by @xmath65 the graph @xmath45. actually, this definition does not assume that @xmath65 is unique. we use this designation for some @xmath45 with @xmath67. * graphs @xmath68. * let us define five planar graphs @xmath69 (see fig. [eliminated]), where @xmath70, and @xmath71. note that @xmath72, is obtained from @xmath35 by removing certain edges..,title="fig : "] .,title="fig : "] .,title="fig : "] .,title="fig : "] .,title="fig : "] @xmath65 is isomorphic to @xmath68 with @xmath73 or @xmath74. @xmath65 is isomorphic to @xmath75 and @xmath76. it is clear that lemma 2 yields theorem 1. now our goal is to prove these lemmas.
Proof of lemma 1
here we give a sketch of our computer assisted proof. for more details see http://dcs.isa.ru/taras/tammes14/ . the combinatorial properties of @xmath77 have previously been considered [@xcite, (*??? * chap. vi), @xcite and @xcite]. in particular, for @xmath33 we have : 1. @xmath65 is a planar graph with 14 vertices ; 2. any vertex of @xmath65 is of degree @xmath78 or @xmath79 ; 3. any face of @xmath65 is a polygon with @xmath80 or @xmath81 vertices ; 4. if @xmath65 contains an isolated vertex @xmath82, then @xmath82 lies in a hexagonal face. moreover, a hexagonal face of @xmath65 can not contain two or more isolated vertices. [cor1] in our papers @xcite the main relations between these parameters were considered ((*??? * propositions 3.63.11), (*??? * proposition 4.11)). let us list those results here. let @xmath83 be an irreducible contact graph in @xmath31 with faces @xmath84. let @xmath85. denote by @xmath86, @xmath87 the set of its angles. here @xmath88 denotes the number of vertices of @xmath84. 1. @xmath89 for all @xmath90 and @xmath91. @xmath92 for all @xmath90 and @xmath91, where @xmath93 is the angle of the equilateral spherical triangle with side length @xmath94. @xmath95 for all vertices @xmath82 of @xmath96. here @xmath97 denotes the set of angles with the vertex in @xmath82. 4. if @xmath98 then @xmath84 is an equilateral triangle with angles @xmath99 5. in the case @xmath100, @xmath84 is a spherical rhombus and @xmath101, @xmath102. moreover, we have the equality : @xmath103 6. in the case @xmath104, @xmath84 is a convex equilateral spherical polygon with angles @xmath105. denote by @xmath106 vertices of @xmath84. the polygon @xmath84 is uniquely defined (up to isometry) by its @xmath107 angles and @xmath94. then functions @xmath108 and @xmath109, where @xmath110 and @xmath111 are also uniquely defined. it follows that + (a) @xmath110 for @xmath112 ; and + (b) @xmath113 for all @xmath114. 7. now consider the case when there is an isolated vertex inside @xmath84. (it is only if @xmath115.) define @xmath116 then @xmath117 our proof of lemma 1 consists of two parts : + (i) create the list @xmath118 of all graphs with 14 vertices that satisfy proposition 3.1 ; + (ii) using linear approximations and linear programming remove from the list @xmath118 all graphs that do not satisfy the known geometric properties of @xmath65 (proposition 3.2). to create @xmath118 we use the program _ plantri _ (see @xcite). this program is the isomorph - free generator of planar graphs, including triangulations, quadrangulations, and convex polytopes. (brinkmann and mckay s paper @xcite describes plantri s principles of operation, the basis for its efficiency, and recursive algorithms behind many of its capabilities.) the program generates about @xmath119 billion graphs in @xmath118, i.e. graphs that satisfy proposition 3.1. namely, @xmath118 contains @xmath120 graphs with triangular and quadrilateral faces ; @xmath121 with at least one pentagonal face and with triangular and quadrilaterals ; @xmath122 with at least one hexagonal face which do not contain isolated vertices. the list of graphs with one and more isolated vertices relies on graphs in @xmath123 with @xmath124 that contain at least @xmath125 hexagons. for instance, the list of graphs in @xmath118 with exactly one isolated vertex consists of @xmath126 graphs. however, this list may contain isomorphic graphs. let @xmath127 be a finite point set such that its contact graph @xmath45 is irreducible. properties (i)-(iv) are combinatorial properties of @xmath45. there are several geometric properties. note that all faces of @xmath45 are convex. since all edges of @xmath45 have the same length, @xmath20, all its faces are spherical equilateral convex polygon with number of vertices at most @xmath128. consider now a planar graph @xmath96 with given faces @xmath129 that satisfy corollary 2.1. we are going consider embeddings of this graph into @xmath31 as an irreducible contact graph @xmath45 for some @xmath127. any embedding of @xmath96 in @xmath31 is uniquely defined by the following list of parameters (variables) : + (i) the edge length @xmath94 ; + (ii) the set of all angles @xmath86, @xmath87 of faces @xmath84. here by @xmath88 we denote the number of vertices of @xmath84. let us consider a graph @xmath96 from @xmath118. we start from the level of approximation @xmath130. now using proposition 3.2 we write the linear equalities and inequalities below. \(a) from proposition 3.2(3) we have 14 linear equalities @xmath131 \(b) since @xmath132, from proposition 3.2(2) we have @xmath133 where @xmath134. \(c) for a quadrilateral @xmath135 with angles @xmath136 we have the equalities @xmath137 and inequalities @xmath138 \(d) for a quadrilateral @xmath135 we also have the linear inequalities @xmath139 these inequalities follow from proposition 3.2(5). we have @xmath140 if we consider the maximum and minimum of @xmath141 with @xmath142 $] and @xmath143, then we obtain these inequalities. so from these linear equalities and inequalities we can obtain maximum and minimum values for each variable. it gives us a domain @xmath144 which contains all solutions of this system if they there exist. if @xmath144 is empty, then we can remove @xmath96 from the list @xmath118. the first step, @xmath130, `` kills '' almost all graphs. after this first step all that remained were @xmath145 graphs without isolated vertices, @xmath146 graphs with one isolated vertex, and no graphs with two and more isolated vertices. next we consider @xmath147. in this step @xmath144 is divided into two domains and for both we can add the same linear constraints as we did when @xmath130. moreover, for this step we add new linear constraints for polygons with five and higher vertices. in this level we obtain the parameterdomain @xmath148. if this domain is empty, then @xmath96 can not be embedded to @xmath31 and it can be removed from @xmath118. actually, for @xmath149 we can repeat the previous step, divide @xmath148 into two domains and obtain additional constraints as we did when @xmath147 for both parts independently. we can repeat this procedure again and again. in fact, by increasing @xmath150 we increase the number of sub - cases. however, in practically every step some sub - cases vanish. we repeat this process for @xmath151 and obtain a chain of embedded domains : @xmath152 if this chain is ended by the empty set, then @xmath96 can be removed from @xmath118. in the case that a graph @xmath96 after @xmath153 steps still `` survives '', i. e. @xmath154, then it is checked by numerical methods, namely by the so called nonlinear `` solvers ''. (we used, in particular, ipopt.) if a solution there exists, then @xmath96 is declared as a graph that can be embedded, and if not, then @xmath96 is then removed from @xmath118. in @xcite are given some numerical details for this algorithm.
Proof of lemma 2
in this section we present a proof of lemma 2. actually, two approaches are considered here - geometric and analytic. both methods are rely on the geometric properties of @xmath65. the first method we already applied to prove lemma 2 in our solution of the tammes problem for @xmath5 @xcite. in this method by using the symmetries of a graph @xmath155 we find certain relations between the variables and using them we prove that @xmath156. the geometric method is elementary, but it is not trivial and is relatively tricky. for @xmath157, we found a proof that is based on the geometric approach. however, for the cases @xmath158, we could not find a simple geometric proof. for those cases we apply the analytic approach. the idea of the analytic approach is very similar to the connelly s `` stress matrix '' method @xcite. perhaps, this method is not as elementary and explicit as the geometric approach, however it works for all cases and can be applied with a computer assistant. * geometric approach : the case @xmath166. * let @xmath155. proposition 3.2(3,4,5) allows us to prove some equalities for variables @xmath167. namely, assume that two vertices, a and b, of @xmath96 are adjacent to two triangles and two quadrilaterals. then @xmath168. we denote the correspondent angles by @xmath169 and @xmath170 respectively. (here, as above, @xmath171 denote the angle of the equilateral triangle.) we have @xmath172 if additionally, @xmath173 and @xmath174 are the opposite vertices of a quadrilateral @xmath135 in @xmath96, then the equality @xmath175 (see proposition 3.2.5) implies the equality @xmath176. it is not hard to prove that if @xmath186 $], where @xmath187 is sufficiently small, then @xmath94 is uniquelly defined, i.e. there is a continuous function @xmath188 on @xmath189 $] such that @xmath190. since @xmath191, if @xmath192 is a solution of (4.1), then @xmath193 is also a solution. it implies that the function @xmath188 is even, i. e. @xmath194. we present this function in fig. 4. note that @xmath200. from fig. [dbyx] we can see that @xmath201 for @xmath202. therefore, @xmath203. it can be rigorously proved. indeed, @xmath204 which means that the function @xmath205 is monotonically decreasing. since all @xmath206, we obtain @xmath207. * analytic approach : the case @xmath214. * suppose that @xmath215 is a configuration of @xmath0 points in the sphere @xmath31 such that @xmath45 is the maximal graph. then points in @xmath13 can not to get closer together. so any slight motion of points in @xmath13 can not increase the minimal distance @xmath20. therefore, @xmath13 is an infinitesimally rigid configuration @xcite. we say that an @xmath216 symmetric matrix @xmath217 is the _ equilibrium stress _ matrix if to each pair of distinct vertices @xmath218 of @xmath219 we have @xmath220, @xmath221 when @xmath218 is not an edge of @xmath222, and for each @xmath90, the equilibrium equation @xmath223 holds. here @xmath224 is the unit tangent vector at the point @xmath225 to the great circle that passes through the points @xmath225 and @xmath226. actually, these conditions for the equilibrium stress matrix can be derived from the karush - kuhn - tucker conditions @xcite, where stresses @xmath227 correspond to lagrange multipliers and the equilibrium equation corresponds to the stationary condition. note that, the inequality @xmath220 holds because @xmath228 can not be increased. in fact, after computations we have approximation intervals for all parameters of @xmath233 therefore, we can compute intervals also for all @xmath234 and @xmath235, where @xmath236 let @xmath237 $] and @xmath238 $]. here @xmath239 and @xmath240 are sufficiently small numbers. thus, (4.3) implies @xmath241 j. leech, the problem of the thirteen spheres, math. gazette * 41 * (1956), 22 - 23. levenshtein, on bounds for packing in @xmath247-dimensional euclidean space, sov. dokl. * 20*(2), 1979, 417 - 421. | the tammes problem is to find the arrangement of @xmath0 points on a unit sphere which maximizes the minimum distance between any two points.
this problem is presently solved for several values of @xmath0, namely for @xmath1 by l. fejes tth (1943) ; for @xmath2 by schtte and van der waerden (1951) ; for @xmath3 by danzer (1963) and for @xmath4 by robinson (1961).
recently, we solved the tammes problem for @xmath5.
the optimal configuration of 14 points was conjectured more than 60 years ago. in the paper
, we give a solution of this long - standing open problem in geometry.
our computer - assisted proof relies on an enumeration of the irreducible contact graphs. | 1410.2536 |
[sec:level1]introduction
in 2003, belle reported the discovery of a charmoiniumlike neutral states x(3872) with mass=@xmath21 mev with width @xmath22 mev @xcite and latter confirmed by do @xcite, cdf @xcite and barbar @xcite. this discovery fed excitement in the charmonium spectroscopy because of unconventional properties of the state. x(3872) could not be explained through ordinary meson(@xmath23) and baryon (qqq) scheme. the conventional theories predicts complicated color neutral structures and search of such exotic structures are as old as quark model @xcite. after the discovery of x(3872), the large number of charge, neutral and vector states have been detected in various experiments, famous as the xyz states. recently, the charge bottomoniumlike resonances @xmath24 and @xmath25 have been reported by belle collaboration in the process @xmath26 and @xmath27 @xcite. moreover, a state reported by besiii collaboration in @xcite as @xmath28 in the @xmath29 reaction, again the besiii collaboration reported a state @xmath30 from invariant mass @xmath31 in the @xmath32 reaction @xcite, whereas the belle @xcite and cleo @xcite reconfirmed the status of the state. the sub structure of the all these states are still a open question, they might driven exotic structure like tetraquark, molecular or hybrid, expected as per theory of qcd, needs theoretical attention. + in the present study, we focus on the molecular structure, as meson - antimeson bound state, just like deuteron. the multiquark structures have been studied since long time @xcite. t@xmath33rnqvist, in @xcite predicted mesonic molecular structures, introduced as @xmath34 by using one pion exchange potential. with heavy flavour mesons, various authors predicted the bound state of @xmath0 and @xmath12 as a possible mesonic molecular structures as well as studied the possibilities of the @xmath4 and @xmath15 as vector - vector molecule @xcite, also it have been studied in various theoretical approaches like potential model @xcite, effective field theory @xcite, local hidden gauge approach @xcite etc.. + in the variational scheme, we have used the potential model approach to study the meson - antimeson bound system. for that, we have used the hellmann potential @xcite (superposition of the coulomb + yukawa potential) with one pion exchange potential (opep). we assume that the colour neutral states experience residual force due to confined gluon exchange between quarks of the hadrons (generally known as residual strong force), skyrme - like interaction. as mentioned by greenberg in ref. @xcite and also noted by shimizu in ref @xcite that this dispersive force (also called london force) or the attraction between colour singlet hadron comes from the virtual excitation of the colour octet dipole state of each hadron @xcite. indeed, long ago skyrme @xcite in 1959 and then guichon @xcite, in 2004 had remarked that the nucleon internal structure to the nuclear medium does play a crucial role in such strong effective force of the n - n interaction. in the study of the s - wave n - n scattering phase shift, in ref.@xcite, khadkikar and vijayakumar used the colour magnetic part of the fermi - breit unconfined one - gluon - exchange potential responsible for short range repulsion and sigma and pion are used for bulk n - n attraction. in this way, with such assumption of the interaction, the mass spectra of the dimesonic bound states are calculated. + for molecular binding, the ref.@xcite found that the quark exchange alone could not bind the system, led to include one pion exchange. the ref.@xcite mention some additional potential strength required with one pion exchange. whereas, the dynamics at very short distance led to complicated heavy boson exchange models as studied in @xcite. in all these studies @xcite, one common conclusion was extracted that the highly sensitive dependence of the results on the regularisation parameter. to avoid these dependency and complicated heavy boson exchange in this phenomenological study, we used the hellmann potential in accordance to delicate calculation of attraction and repulsion at short distance. the overall hellmann potential represents the residual strong interaction at short distance in flavour of the virtual excitation of the colour octet dipole state of each colour neutral states. the opep is included for long range behaviour of the strong force. the ope potential could be split into two parts (i) central term with sipn - isospin factor (ii) tensor part. we have analyse the effect of these two parts. by calculating the spin - isospin factor as in @xcite, we have found the symmetry braking in our results which was also discussed by t@xmath33rnqvist in @xcite, whereas, the tensor term to be found play a very crucial role implicit the necessity of it. in that way, bound state of @xmath0 is compared with the state x(3872) which also have been predicted as mesonic molecule by the authors of ref.@xcite whereas states @xmath24 and @xmath24 which are close to the @xmath12 and @xmath15 threshold. + to test the internal structure of the state, in general, one have to look for the decay pattern of the state. in ref. @xcite, the hadronic decays of the x(3872) have been studied in accordance to its decay mode sensitive to the short or long distance structure of the state. to test the compared states as dimesonic system, we have used the binding energy as input for decay calculation. we have adopt the formula developed by authors of ref. @xcite for the partial width sensitive to the long distance structure of the state, whereas, the formula for the decay mode sensitive to short distance structure of the state is taken from @xcite. in ref. @xcite, authors predicted existence of the neutral spin-2 (@xmath35=@xmath36) partner of x(3872), would be @xmath4 bound state, and in same way expected spin-2 partner of @xmath12, would be @xmath15 bound state, on the basis of heavy quark spin symmetry and calculated the hadronic and radiative decays. we have used formula of @xcite for radiative decay for predicated states. with calculated binding energy, the decay properties are in good agreement with @xcite. + the article is organize as follows, after the brief introduction we have discussed the hellmann potential and ope potential in sec - ii then in sec - iii we presents our theoretical approach for the calculation of the mass spectra as well as for decay width, in sec - iv we present our results for deuteron as per our model and generalize our approach. in sec - v and vi we presents our results of dimesonic states and finally we summarise our present work.
The hellmann and one pion exchange potential
two color neutral states formed the bound states, just like deuteron is the bound state of proton and neutron. in the case of the dimesonic states, the meson and antimeson are takeover as a constituents, forming a bound state. the interaction potential between two color neutral constituents taken as a phenomenological hellmann potential @xcite with one pion exchange potential(opep) @xcite. the colour neutral states experience residual force due to confined gluon exchange between quarks of the hadrons. the attraction between colour singlet hadron comes from the virtual excitation of the colour octet dipole state of each hadron. when two colour neutral states are come enough close to each other so the quark of the one colour neutral states can feel to quark of the other state creates manifestation of the energy. once the interaction become strong enough then this manifestation of the energy being the source for the creation for the quark anti quark pair. this residual force is analogous to the residual interaction between two electrical neutral atoms, experienced the van der waals force. the overall hellmann potential represents the residual strong interaction at short distance, where, these pseudopotential has the form, namely @xmath37 here, the constant @xmath38 is the residual strength of the strong running coupling constant between the two colour neutral states and b are the strength of the yukawa potential whereas @xmath39 is the relative separation between constituents. the value of the @xmath38 could be determined through the model, such as @xmath40 where @xmath41 and @xmath42 are constituent masses, m=2@xmath41 @xmath42/(@xmath41+@xmath42), @xmath43=1 gev, @xmath44=0.250 gev and @xmath45 is number of flavour @xcite. so, the coulomb interaction is constrained by the value of @xmath38. the constant b and c appeared in the yukawa potential of eq.(1) play very trivial role on overall characteristic of the hellmann potential. in the fig(3) (see appendix - a), one can observe that once the coulombic interaction is fixed, the variation of the constant b and c are inversely proportional to each other. as the value of the b increases, it increases the repulsive nature of the potential while the constant c smoothing the curve as well as it increases the strength of the yukawa part - inversely to b. to show such nature of the hellmann potential, in the fig(3), the graph have been plotted for various set of values of the b and c. for the bound state, the value of b can take both positive and negative values. with the negative values of the b the overall potential become attractive. for more detailed discussion on the hellmann potential, we suggest to the readers for ref.@xcite. in our case of the dimesonic bound states calculations, we assume that the hellmann potential at very short distance cares for the delicate cancellation of attraction and repulsion respectively which is mainly taken care by heavy boson exchange in the one boson exchange model. + whereas, the long range behaviour of the interaction part is accomplished by one pion exchange. the model carried the net potential as the hellmann potential plus ope potential plus relativistic correction. we introduced and discussed the relativistic correction after this section. for instance, to fix the model (or the potential) for deuteron which is widely believed to be have molecular structure and well studied in obe potential model, we fit the values of the constant @xmath46 and @xmath47 to get approximate binding energy of deuteron. then, to generalize the model to dimesonic system, we arrive on the relation between b and c in accordance of the mass of the system, + if @xmath48, where @xmath49 is the threshold mass of dimesonic state and @xmath41 mass of deuteron and n is integer number (n=1,2,3..), then @xmath50 here, @xmath46 and @xmath47 are the constant, fitted for deuteron binding energy. + the one pion exchange potential (opep) taken for long range interaction which is well studied for nn - interaction. the ope potential for nn - interaction takes the form @xcite @xmath51\end{aligned}\]] where @xmath52 is the nucleon - pion coupling constant, @xmath53 and @xmath54 are spin and isospin factors respectively while the @xmath55 and @xmath56 are defined as @xmath57 @xmath58 whereas the @xmath59 is the usual tensor operator expressed as @xmath60, is mainly responsible for long range tail of the potential and play very crucial role in the nn - interaction. the expression of opep in the eq.(4) is for the point like pion. while, in a more realistic picture where the pion itself has its own internal structure, it is natural to introduced the usual form factor due to the dressing of the quarks. by introducing this finite size effect @xcite, we have \(a) (b) (c) (d) @xmath61 @xmath58 thus, the function @xmath62 and @xmath63 with the finite size effect take the form @xmath64\]] @xmath65\]] now, one pion exchange potential for dimesonic system could be written as @xmath66 @xmath67\end{aligned}\]] such that, the opep becomes for dimesonic systems, @xmath68 where the @xmath69=0.69 is the meson - pion coupling constant, @xmath49 and @xmath70 are the average mass of two constituent of the dimesonic state and pion mass respectively, while @xmath71 is the pion form factor. the mesons and quark masses with their quantum numbers are taken from the listing of particle data group@xcite. the constituent meson - quark coupling constant may derive by using the goldberger - treiman relation on suitable estimates of known @xmath72 coupling constant. the relation between quark - boson and nucleon - boson coupling could expressed as @xcite @xmath73 the effective ope potential can be split into a central and tensor term proportional to @xmath74 @xmath75 and @xmath59 @xmath75 respectively. in ref.@xcite, thomas et.al. mentioned and discussed sign convention and detail calculation of spin - isospin factor. in ref.@xcite, authors showed the inconsistent sign convention adopted by ref.@xcite in the calculation of spin - isospin factor of @xmath76(@xmath36) and then derive and explain overall sign for determination of @xmath74 @xmath75. we are agreed with ref.@xcite and adopt the same. for total spin s, total isospin state i and charge conjugation parity c, the spin - isospin factor for central term are given by @xcite @xmath77 the matrix element of the tensor oerator for different spin state are real numbers and it is well discussed by t@xmath33rnqvist in ref.@xcite. for example, the matrix element of @xmath59 in the case of deuteron @xmath78 with such interaction the opep potential shown in eq(11), one need to solve the coupled channel schr@xmath33dinger equation. in the present study, we have focus only on the s - wave spectra of dimesonic state. as such, we have taken only the s - wave tensor contribution of particular spin state. the matrix element of such tensor operator of dimesonic states are given by @xmath79 the value of the form factor @xmath80 in opep affects the strength of the potential very drastically. the sensitivity of the @xmath80 have discussed by authors of ref.@xcite. in the fig(1), the nature of the opep in different spin - isospin state are shown for different values of @xmath80 (noted that only the s - wave contribution of the tensor interaction is considered). one can analysed from the plot, as the value of the @xmath80 increases the strength of the potential is also increases drastically. in ref.@xcite, t@xmath33rnqvist noted the value of @xmath80 fall in range 0.8 - 1.5 gev for fit the nn scattering data where as in case of mesonic molecule, specially, the heavy meson which are very small compared to the size of the nucleon, the larger value of @xmath80 is expected and the large value of @xmath80 increases the binding energy. ref.@xcite shows the results are very sensitive to @xmath80 and the binding energy no longer monotonically increases with @xmath80 with opep model. for instant, we fixed @xmath80=1.5 gev consistent with nn - scattering data and increase the dominance of the hellmann potential, such that we arrive at relation of the eq(2). + the factors @xmath74 @xmath75 and @xmath59 @xmath75 makes the opep spin and isospin state dependent. as per the literature @xcite, the most attractive channel is (i, s)=(0,0) while the channel (i, s)=(2,0) is the most repulsive. but, as in fig(1), we observed the channel (i, s)=(1,1) is most attractive while (i, s)=(0,1) noted as most repulsive one. the next order of the attractive channels are (i, s)=(0,0) and (i, s)=(1,2). whereas, the channels (i, s)=(0,1),(i, s)=(2,0) and (i, s)=(1,0) expected to be unbound. the observed change in the nature of the these channels may be due to only the s - wave contribution of the tensor interaction. it clearly indicates the dominant effect of tensor operator at long range part of the potential. the discussed effect of opep is also reflect in our results, tabulated in table - ii and table - iii..masses of the mesons (in mev)@xcite [cols="^,^,^,^,^,^,^,^ ",] there are large possibilities of dimesonic states with heavy light flavours, also shown in our calculated mass spectrum(see table - iii). we have predicted the mass and root mean square radius. for instant with @xmath76 dimesonic states, for the decay properties, we focus on these two states @xmath81 and @xmath82 as spin-2 partner of @xmath0 and @xmath12 molecules respectively and also expected as per heavy quark spin symmetry(hqss). the study of decay properties with mass spectra is very important to investigate their sub structure. we have calculated the hadronic and radiative decay of these states. the hadronic decay modes of @xmath81 @xmath83 by using the eq(20)(with @xmath84=0.993), the calculated partial decay width, we have @xmath85 similarly the decay width for @xmath82 with the same formula (with @xmath84=0.998) to be found @xmath86 to understand the electromagnetic branching fraction we need to understand the interaction with photon and s - wave mesons and their contribution due to light and heavy quarks @xcite. in addition to, radiative decay is more sensitive to long distance molecular structure. the radiative decay mode of @xmath81 expressed as @xmath87 thus the decay width for the state calculated as per eq(23) (with @xmath88 = 1 gev), we have @xmath89 the calculated partial decay widths are in good agreement with the results calculated in ref@xcite. the radiative decay for the state @xmath82 could also calculated similar to @xmath81. @xmath90 for this decay mode of @xmath82, in our model the radiative decay of @xmath82 to be found forbidden with suggestive value of @xmath88 = 1 gev. with very large value of @xmath88 (@xmath918), we have get the comparable results with ref. moreover, the radiative decay width for @xmath82 calculated in @xcite is @xmath91 10 ev, which is very small. with very large value of @xmath88, we have found @xmath91 14 ev. (see the table - iv, for the results of calculated partial widths with comparisons with others).
Summery
in summary, we have calculated the mass spectra of dimesonic states with heavy - light flavour mesons by using the hellmann and one pion exchange potential. the calculated binding energy of dimesonic states are found overestimated which is also expected in variational approach. we have analysed the change in the binding energy due to charge conjugation parity and isospin, agreed with ref.@xcite suggested isospin symmetry braking. we have also discussed the effect of tensor term in one pion exchange. we includes only the s - wave contribution of tensor term and observed shuffling of channels which is different from expected channels as in @xcite. it is remarkably pointing out the effect of the tensor term and its contribution in different processes, led us to conclude that it can not be ignore. whereas, the contribution from the relativistic correction in the potential to the binding energy is @xmath92 which is almost one third to our previous work @xcite. + in addition, we have calculated the decay properties of @xmath0, @xmath12, @xmath4 and @xmath15 (the formalism adopted from ref.@xcite), using our calculated binding energy as input where formalism mainly dependent on the masses. to enforced the our predictions, we have attempted the decay calculations as well to test the molecular structure of compared states with dimesonic states on the basis of decay modes which may have responsible for their long and short distance structure, as studied in the literature @xcite. we have found comparable results of decay properties (with our mass spectra) @xcite, support the prediction of @xmath0, @xmath12, @xmath4 and @xmath15 bound states as mesonic molecules. we support the molecular structure of the x(3872) as @xmath0 bound state with @xmath93 dominant decay mode. the long distance radiative decay may gives information of the mixing of charge to neutral channel of the bound state which lack in this study. we have predicted the mass spectra and decay properties of the possible spin-2 partner (@xmath94) of x(3872) in charm sector as well as spin-2 partner of @xmath12 in the bottom sector, as miguel et.al studied in @xcite, also expected in heavy quark spin - flavour symmetry. + apart form these states, we have suggest mass spectra of possible dimesonic states as mesonic molecules with @xmath95, @xmath96, @xmath97, @xmath98 and @xmath99 mesons. wei chen et.al in @xcite suggested open flavour tetraquark structure with having strange quark, with mass range 6.9 - 7.3 gev, with @xmath100 or @xmath101. in ref.@xcite, gui - jun ding suggest @xmath5 molecule. the masses of @xmath102 and @xmath76 dimesonic states with @xmath95, @xmath96, @xmath97 and @xmath98 mesons are fall in the same range of 7.1 - 7.3 gev. moreover, we have also predicted the masses of the dimesonic states having @xmath99 mesons as a constituent. we look forward to the experimental facilities and working groups to take more attention for searches of the possible dimesonic states as well as for the confirmations of our theoretical predictions. + * acknowledgements * d. p. rathaud would like to thanks prof. r. c. johnson for the useful discussion. a. k. rai acknowledge the financial support extended by d.s.t., government of india under serb fast track scheme sr / ftp /ps-152/2012.
Hellmann potential
\(a) (b) \(c) (d) \(e) (f) \(g) (h) \(i) (j)
Decay properties
\(i) for hadronic decay @xcite : one of the constituent meson decay into two mesons : @xmath103 + here, m is the mesonic molecule (m = d+b, where d and b are constituents) decaying into product mesons a, d, e in which one of the meson is the constituent of the dimesonic state, for example, @xmath104, here, @xmath105 or @xmath106 + from ref @xcite, on the bassis of the tree level approximation the amplitude is given by @xcite @xmath107 where g=0.69 the quarks meson coupling constant, @xmath108= 92.2 mev pion decay constant where as @xmath109=0.35 @xmath110 is the coupling constant to the dimesonic state to the charge or neutral channels @xcite. thus, decay formula reads @xmath111 where @xmath112 is the three momentum of the pion. while @xmath113 and @xmath114 are the the four momenta of ad and ac systems. since the amplitude is dependent on the invariant masses we have @xmath115=@xmath116 and @xmath117=@xmath118 of the final state ea and ad pairs respectively @xcite. here, @xmath119 is the k@xmath120llen function. for the known value of @xmath115 the range of the @xmath117 could be determined by the values of momentum @xmath121 is parallel or antiparallel to @xmath122 as here, @xmath129 is the dimesonic coupling constant, @xmath130 is the fine structure constant, @xmath131gev is the charm quark mass, @xmath132 @xmath133 (@xmath134 is the light quark mass), @xmath70 and @xmath135 is the pion and kaon masses, respectively. whereas, @xmath108 and @xmath136 are the pion and kaon decay constant, respectively. thus the partial radiative decay width given as for the known value of the @xmath138 the range of @xmath139 could be determined by the values of momentum @xmath122 is parallel or antiparallel to @xmath140 as @xmath141 here, @xmath142 @xmath143 and @xmath140 are the photon energy and momentum in the @xmath138 cm frame, respectively. whereas, the amplitude for the @xmath15 read as with @xmath145=@xmath146 for @xmath15 system and could be calculated according to eq.(b8). @xmath42=4.6 gev is mass of the bottom meson. for more detailed of the calculations and formalism of decay calculation we refer to ref.@xcite. al.(particle data group), chin. c * 38 *, 090001 (2014). choi et al. (belle collaboration), phys. lett. * 91 *, 262001 (2003. abazov et al.(d0 collaboration), phys. lett. * 93 *, 162002 (2003). d. acosta et al. (cdf collaboration), phys. lett. * 93 *, 072001 (2004). b. aubert et al. (babar collaboration), phys. d * 71 *, 071103 (2005). m. gell - mann, phys. * 8 *, 214 (1964). g. zweig, cern report no. th-401 ; report no. th-412,(1964). a. bonder et. al.(belle collaboration), phys. lett. * 108 *, 122001 (2012). m. ablikim et al. (besiii collaboration), phys. 111 *, 242001 (2013). m. ablikim et al. (besiii collaboration), phys. lett. * 111 *, 252001 (2013). z. q. liu et al. (belle collaboration), phys. * 111 *, 252002 (2013). t. xiao, s. dobbs, a. tomaradze, and k. k. seth, phys. lett. b * 727 *, 366 (2013). jaffe, phys. d. * 38 *, 195 (1977). jaffe, phys. d. * 15 *, 267 (1977). j. weinstein et. al., phys. d. * 27 *, 588 (1983) ; phys. d. * 41 *, 2236 (1990) ; phys. d. * 47 *, 911 (1993). nils a. t@xmath33rnqvist, phys. rev. lett. * 67 *, 557 (1991). nils a. t@xmath33rnqvist, z. phys. c * 61 *, 525 (1994). c. e. thomas and f. e. close, phys. rev. d*78 *, 034007 (2008). gui - jun ding, jia - feng liu and mu - lin yan, phys. d * 79 *, 054005 (2009). mahiko suzuki, phys. d * 72 *, 0114013 (2005). f. e. close and p. r. page, phys. b * 578 *, 119 (2004). ajay kumar rai and d. p. rathaud, eur. j c * 79 *, 412 (2015). a. k. rai et. al., nucl. phys. a * 782 *, 406 (2007). s. patel and p. c. vinodkumar, eur. j c * 76 *, 356 (2016) ; eur. j a * 50 *, 131 (2014). s. fleming et. rev. d*76 *, 034006 (2007). a. miguel et. al., eur. j c * 75 *, 547 (2015). feng - kun guo et. al., eur. j c * 74 *, 2885 (2014). r. molina and e. oset, phys. d * 80 *, 114013 (2009). h. hellmann, acta physicochim, urss * 44 *, 913 (1935) ; ibid. * 4 *, 225 (1936) ; ibid. * 4 *, 324 (1936). o. w. greenberg and h. j. lipkin, nule. a * 370 *, 349 (1981). k. shimizu, rep. phys. * 52 *, 1 - 56 (1989). t. h. r. skyrme, nucl. phys. * 9 *, 615 (1959);nucl. phys. * 31 *, 556 (1962). guichon and a.w. thomas, phys. lett. * 93 *, 132502 (2004). s. b. khadkikar and k. b. vijaykumar, phys. b * 254 *, 320 (1991). s. swanson, phys. lett. b * 588 *, 189 (2004). nils a. t@xmath33rnqvist, phys. lett. b * 590 *, 209 (2004). a. miguel et. al., phys. b * 755 *, 337 (2016). f. aceti, r. molina, and e. oset, phys. d * 86 *, 0113007 (2012). n.n. achasov and e.v. rogozina, jetp lett. * 100 *, 227 (2014). eric braaten and masaoki kusunoki, phys. rev. d*72 *, 054022 (2005). rui zhang et. al., phys. d*65 *, 096005 (2002). m. w. paris and v. r. pandhaipande, phys. c * 62 *, 015201 (2000). v. r. pandharipande, s. c. piper and r. schiavilla, notes on nuclear forces on light nuclei. http://minimafisica.biodec.com/member/k/schiavilla-notes. d. ebert et. phys. rev. d * 79 *, 114029 (2009). janusz adamowaski, phys. rev. a * 31 *, 43 (1985). d. o. riska and g. e. brown, nucl. phys. a * 679 *, 577 (2001). n. devlani et. al., eur. j. a * 48 *, 104 (2012) ; eur. j. a * 50 *, 154 (2014). y. koma, m. koma, and h. wittig, phys. lett 97, 122003 (2006). a. abulencia et. al.(cdf collaboration), phys. lett. * 98 *, 132002 (2007). yan - rui liu et.al., eur. phys. j c * 56 *, 63 (2008). n.n. achasov and e.v. rogozina, jetp lett. * 100 *, 227 (2014). p. del amo sanchez et al.(babar collaboration), phys. d * 82 *, 011101 (2010). kunihiko terasaki, yitp 14 - 94 (2014), arxiv:1411.7483v1 [hep - ph]. j. m. dias, f. aceti and e.oset, phys. rev. d*91 *, 076001 (2015). a. bonder et. al., phys. d*84 *, 054010 (2011). wei chen, t. g. steele and shi - lin zhu, phys. rev. d*89 *, 054037 (2014). yubing dong et. al., phys. d * 79 *, 094013 (2009). gui - jun ding, eur. j c * 64 *, 297 (2009). | in this work, we have calculated the mass spectra and decay properties of dimesonic (meson - antimeson) bound state in the variational scheme. the intermesonic interaction considered as the hellmann potential and one pion exchange potential.
the mass spectra of the @xmath0, @xmath1, @xmath2, @xmath3, @xmath4, @xmath5, @xmath6, @xmath7, @xmath8, @xmath9, @xmath10,@xmath11, @xmath12, @xmath13, @xmath14, @xmath15, @xmath16, @xmath17 etc.. are calculated.
the states x(3872), @xmath18, @xmath19 and @xmath20 are compared with @xmath0, @xmath4, @xmath12 and @xmath15 dimesonic bound states. to probe the molecular structure of the compared states
, we have calculated the decay properties sensitive to their long and short distance structure of hadorinc molecule, our results suggested the compared states driven the molecular structure. apart from these the other calculated mass spectra of dimesonoic states are predicted and for such bound states, the experimental search are suggested.
pacs numbers : : 12.39.jh, 12.39.pn, 13.25.jx,14.40.rt | 1608.03781 |
Introduction
coherent control of atomic and molecular dynamics using optical fields has attracted much attention, both theoretically and experimentally @xcite. thus far, most theoretical work has focused on the idealized case of isolated systems, where loss of quantum phase information due to decoherence, i.e. coupling to the environment, is ignored. such effects are, however, crucial to control in realistic systems, since loss of phase information results in loss of control. for this reason efforts to understand control in external environments @xcite @xcite and to compensate for the resultant decoherence (e.g., @xcite@xcite) are of great interest. there exist a number of basic interference schemes@xcite that embody the essence of coherent control. one is the @xmath0 vs. @xmath1 photon scenario where control results from interference between state excitation using @xmath0 and @xmath1 photons simultaneously. in this letter we provide an analytic solution for control in the two - level @xmath0 vs. @xmath1 photon control scenario in the presence of decoherence. for simplicity, we examine the 1 vs. 3 photon case, although the solutions obtained below apply equally well to the @xmath0 vs. @xmath1 photon case, with obvious changes in the input rabi frequencies and relative laser phases. in 1 vs. 3 photon control@xcite a continuous wave electromagnetic field composed of a superposition of a fundamental and third harmonic wave is incident on a system. by varying the relative phase and amplitude of the fundamental and the third harmonic one can alter the population of the state excited by the incident field. clearly, decoherence can be expected to diminish the 1 vs. 3 photon induced interference, and hence the control over excitation. although extensive theoretical @xcite - @xcite and experimental @xcite - @xcite studies have been carried out on the 1 vs. 3 photon coherent control scenario, there has been no serious examination of the stability of this control scheme in an external environment, barring a derivation of a simple analytical expression for the autoionization of a two - level atomic system for weak laser intensities, using the rate approximation @xcite. amongst the various possible influences of an environment on a system we focus on the loss of phase coherence, that is, dephasing. dephasing is expected to occur on a time scale more relevant to control, since the duration of control field can be on the order of a picosecond or less, wheras the typical time scale for energy transfer is considerably longer @xcite. in this paper we show that the 1 vs. 3 photon phase control scenario (which controls the population) in a two - level system, when coupled to an environment, reduces to the analytically soluble monochromatic field case, but with an effective rabi frequency that is determined by the relative phase and amplitudes of the two fields. sample results for control as a function of relative laser phase in the presence of dephasing are then provided. the possiblity of solving the off - resonance case is also noted.
1 + 3 photon control
consider a two - level bound system interacting with an continuous wave (cw) electromagnetic field and assume that the energy levels undergo random stark shifts without a change of state during collisions with an external bath, e.g., elastic collisions between atoms in a gas. the cw field @xmath2 is treated classically, and the ground and the excited energy eigenstates states, of energy @xmath3 and @xmath4 are denoted @xmath5 and @xmath6, respectively. in general, the system density operator @xmath7 obeys the liouville equation, @xmath8-{\mr}\rho. \label{liouville12l}\end{aligned}\]] here @xmath9, where the free atomic hamiltonian term is @xmath10 and the atom - field interaction term within the dipole approximation is @xmath11\]] with electric dipole operator @xmath12. the second term in eq. ([liouville12l]), @xmath13, is a dissipative term that can have a variety of nonequivalent forms associated with various master equations. below we assume simple exponential dephasing of the off - diagonal @xmath14. in the simplest 1 vs. 3 control scenario, a two - level system is subject to the linearly polarized laser field : @xmath15, \label{efield2l } \end{aligned}\]] where @xmath16 is the real time - independent amplitude and @xmath17 is the phase of the corresponding field, with @xmath18. here the subscripts @xmath19 denotes the fundamental and its third harmonic, and @xmath20 denotes the complex conjugate of the terms that precede it. the fields have frequencies @xmath21 and @xmath22, chosen so that the third - harmonic and the three fundamental photons are on resonance with the transition from the ground state @xmath5 to the excited state @xmath6. in the standard scenario @xcite, control is obtained by changing the relative phase and amplitudes of two fields, which results in the alteration of the degree of interference between the two pathways to the excited state. within the rotating - wave approximation, the slowly varying density - matrix elements of the states @xmath5 and @xmath6, @xmath23, (@xmath24) and @xmath25 obey the following set of equations : @xmath26\nonumber\\ \pd { \sigma_{22}}{t}&= & { \mrm{im } } [({ \mu}_{12}^{(3)}{\mathcal{e}}_f ^3 /\hbar+ { \mu}_{12}{\mathcal{e}}_h e^{i\phi } /\hbar)\sigma_{21 }], \label{density_matrix112l}\nonumber\\ \pd { \sigma_{21}}{t}&=&-\gamma_{{p } } \sigma_{21 } + \frac{i}{2 } ({ \mu}_{21}^{(3)}{\mathcal{e}}_f ^3 /\hbar+ { \mu}_{21}{\mathcal{e}}_h e^{-i\phi } /\hbar)(\sigma_{11}-\sigma_{22 }), \label{density_matrix212l}\end{aligned}\]] with @xmath27 here @xmath28 is the dephasing rate, @xmath29 is the frequency difference between levels @xmath30 and @xmath31 and @xmath32. the quantities @xmath33 and @xmath34 denote the one - photon matrix element for the harmonic field and the effective three - photon matrix element for the fundamental field for the @xmath35 transition. below, we use @xmath36 and @xmath37, omitting the subscripts for simplicity. the controllable relative phase is @xmath38. it is convenient to define the one- and three - photon rabi frequencies by @xmath39 and @xmath40 given in terms of their amplitudes and phases, by @xmath41 and @xmath42. note that, although @xmath43 and @xmath44 are real for a bound system, we derive all the equations under the assumption that they can be complex so that the analysis can be extended to complex matrix elements arising in transitions to the continuum. since @xmath45 and @xmath46 are real and positive, @xmath47 and @xmath48 are determined by @xmath43 and @xmath44 : @xmath49 to amalgamate these rabi frequencies and the relative laser phase of @xmath50, we define the effective rabi frequency @xmath51 : @xmath52 where @xmath51 is real and positive. here @xmath51 and @xmath53 are related to @xmath54 and @xmath55 as @xmath56/ [ { \cos{(\phi+\theta_h)}+\frac{|\omega_f|}{|\omega_h|}\cos{\theta_f}}],\end{aligned}\]] where @xmath57 it is worth noting some features of @xmath51 that are evident from eq. ([omegaeff3]). first, the total excitation probability obtained in lowest order perturbation theory for 1 vs. 3 photon phase control in a two - level system @xcite is proportional to @xmath51. hence, @xmath51 can be used to predict the controlled population, and its dependence on @xmath50, when the fields are weak. further (see below), @xmath51 plays a major role in determining the transient behavior of the excited state population for any field intensity in the absence or presence of dephasing. second, the interference term in eq. ([omegaeff3]) can be controlled by varying @xmath58, that is, by manipulating the relative phase @xmath50 of the two fields. since @xmath43 and @xmath44 are real in a bound system, possible values of @xmath59 are 0 and @xmath60. when @xmath59=0,, @xmath61, @xmath62. on the other hand, when @xmath63,, @xmath64, @xmath65. thus, opposite interference effects are observed depending on the signs of @xmath43 and @xmath44. third, @xmath66. if @xmath67, the smallest @xmath51 is not zero, and thus complete destructive interference, that is, zero excitation from the ground to the excited state, does not occur. rewriting eq. ([density_matrix212l]) in terms of @xmath51 and @xmath53, @xmath68 \\ \pd { \sigma_{22}}{t}&= & { \mrm{im } } [\omega_{\rm{eff } } e^{i\theta } \sigma_{21 }], \label{redensity_matrix112l}\\ \pd { \sigma_{21}}{t}&=&-\gamma_{{p } } \sigma_{21 } + \frac{i}{2 } [{ \omega_{\rm{eff}}}e^{-i\theta }] (\sigma_{11}-\sigma_{22 }). \label{redensity_matrix212l}\end{aligned}\]] and introducing @xmath69, @xmath70, and @xmath71, gives @xmath72 the resultant equations are now of standard form@xcite, but with @xmath51 replacing the rabi frequency of the single field case discussed in ref @xcite. note that the longtime steady - state solution to eqs. ([cut]) to ([cwt]) is found by setting @xmath730, giving @xmath740. this implies that, regardless of initial conditions and for sufficiently large time, pure dephasing leads to an equilibrium state with equal populations in the ground and the excited states and with no remaining coherence. substituting eq. ([cvt]) into eq. ([cwt]) gives a simple equation for @xmath75 : @xmath76 in the important case where initially the ground state is populated and the coherence is zero [i.e., @xmath77, @xmath78, the excited state population @xmath79 is given by @xmath80+\frac{1}{2}\,\, & { \mrm{for } } & \gamma_{\mrm{p } } < 2\omega_{\rm{eff } }, \label{oscillation } \\ \rho_{22}=\frac{[-\lambda_2 e^{\lambda_1 t}+\lambda_1 e^{\lambda_2 t }] } { 2(\lambda_2-\lambda_1)}+\frac{1}{2}\,\,\,\,\,\,\,\,\,\,\,\, & { \mrm{for } } & \gamma_{{p } } > 2\omega_{\rm{eff } }, \label{mono}\\ \rho_{22}=-\frac{e^{-\frac{\gamma_{{p } } t}{2}}}{2}(1+\frac{\gamma_{{p}}t}{2})+\frac{1}{2}\,\,\,\,\,\,\,\,\,\,\,\, & { \mrm{for } } & \gamma_{{p}}= 2\omega_{\rm{eff } }, \label{expo}\end{aligned}\]] where @xmath81, and @xmath82 $]. the general behavior of the solution is seen to be determined by relative size of the dephasing time and the period of the rabi oscillation. analogous analytic results can be obtained for @xmath83 which decays with rate @xmath84. if the external field is intense enough so that @xmath85, then @xmath86 shows oscillations that are exponentially damped with time. on the other hand, if dephasing dominates over the rabi oscillation, so that @xmath87 or @xmath88, @xmath89 increases monotonically. however, in all cases @xmath86 reaches a stationary value of 0.5 at long times and @xmath90 for short times. the behavior of the excited state population for several values of @xmath91 for a given value of @xmath51 (here chosen as 2@xmath92) is sketched in fig. [2lcwanal]. for @xmath93, the introduction of dephasing increases the period of the oscillation and causes the amplitudes to decay as @xmath94. although this is a cw laser field case, we can extract the result for the field being switched off at a specific time, i.e., a square pulsed laser which is on from @xmath95 to @xmath96, by examining the population at time @xmath97. (this assumes that there are no additional energy levels excited by the frequency breadth of the truncated cw source). significantly, one can end up with an increased @xmath89 even for a larger dephasing, depending on the pulse duration. for example, assume that we turn off the field at @xmath98=1. if there is no dephasing, then the excited state population at @xmath98 = 1 is 0 and is thus less than that of any of the other cases with dephasing. on the other hand, for @xmath99 (here @xmath100), there is no oscillation ; @xmath89 just increases monotonically towards 0.5, where the system reaches the steady - state slower with increasing dephasing. if we were to consider a pulse rather than a cw laser field for this relatively strong dephasing case, the excited state population would be expected to increase up to 0.5 with the increase in the pulse duration. typical behavior of @xmath89 and of the 1 vs. 3 photon phase control profile (i.e., @xmath89 as a function of generic phase control variable @xmath58) for several values of rabi frequencies and @xmath91 are shown in fig. [control2lt]. here we assume that the fields are abruptly turned off at the times indicated in the figure captions to produce a square pulse and the intensities are chosen so that @xmath101, to enhance the interference effects. the effective rabi frequency is then @xmath102. while @xmath103 leads to a complete constructive interference of the two transition amplitudes, @xmath104 leads to a complete destructive interference,, no excitation from the ground to the excited state. the typical control behavior seen in fig. [control2lt] depends upon the pulse duration, as well as upon @xmath51 and @xmath105. for example, when the field is weak and @xmath106, then @xmath89 is given by [eq ([oscillation])-([expo])] @xmath107~~{\mrm{for}}~~\gamma_{{p } } = 0 \label{precos}\]] @xmath108 hence, the system shows a @xmath109 rule ". [a similar rule obtains from eq. ([oscillation]) - eq. ([expo]) when @xmath110 and @xmath111, when @xmath112 and @xmath113 and @xmath114 are much less than one, and for @xmath115 when @xmath116.] note also that eq. ([precos]) predicts oscillatory behavior of @xmath89 as a function of @xmath51 at fixed @xmath98, as observed later below. the control profiles for small @xmath54 and @xmath117 (thin dashed lines in fig. [control2lt]) are then seen to be monotonically decreasing from the maximum excitation at @xmath103, to zero excitation at @xmath104, i.e. they follow the @xmath109 rule ". by contrast, for strong intensity (thin solid lines in fig. [control2lt]) in which there are many rabi cycles during the pulse, the control curve is not necessarily monotonic since the final excited populations are determined by the time at which the fields are turned off. introducing dephasing is seen to lead to a decreased range of control whose magnitude depends on the relative strength of the dephasing and on the effective rabi frequencies, according to eqs. ([oscillation]) - ([expo]). figure [control2lt] demonstrates that phase control profiles are strongly dependent on the pulse duration. for weak intensities, as the pulse duration increases, the degree of control improves and the control curve continues to approximately follow a @xmath118 law (e.g., eq. ([cos])). this behavior is seen both in the absence and in the presence of dephasing, although dephasing reduces the yield for a given pulse duration. in the strong field case the control profile varies strongly with pulse duration. in particular, with @xmath119, if the pulse duration is smaller than the oscillation period (= 1/2) of the @xmath120 case, then @xmath89 decreases with increasing @xmath58, as shown in fig. [control2lt](a). for the pulse duration greater than that period, the control profile no longer follows @xmath118 and the maximal yields start to appear at @xmath121, as shown in figs. [control2lt](b) to (d). in all strong intensity cases, the addition of dephasing results in a decay of @xmath89 with a rate of @xmath122 for a given @xmath51. thus the degree of the control worsens in the presence of dephasing as the pulse duration increases. note that the introduction of dephasing leads to a degree of control @xmath123 that converges to 0.5, where @xmath123 is defined as the difference between the maximum and minimum excited state populations. finally, we note that this treatment can be extended in two directions. the most obvious is to extend it to the general two - level @xmath0 photon + @xmath1 photon interference scenario@xcite where the structure of the problem is exactly the same as that of the 1 vs. 3 photon case. the equations above therefore hold, but with the one and three photon rabi frequencies and phases replaced by the @xmath0 and @xmath1 photon rabi frequencies and phases. the second is to consider the more general case that includes the equal detuning of both fields from the @xmath124 transition, i.e. @xmath125, and where the populations of levels @xmath6 and @xmath5 decay with the same rate @xmath126. ([redensity_matrix112l]) to ([redensity_matrix212l]) become : @xmath127, \label{gredensity_matrix112l}\\ \pd { \sigma_{22}}{t}&= & -\gamma_d(\sigma_{22}-\sigma_{2e }) + { \mrm{im } } [\omega_{\rm{eff}}e^{i\theta } \sigma_{21 }], \label{gredensity_matrix222l}\\ \pd { \sigma_{21}}{t}&= & -(\gamma_d+\gamma_{{p}}+i\delta)\sigma_{21 } + \frac{i}{2 } [{ \omega_{\rm{eff}}}e^{-i\theta }] (\sigma_{11}-\sigma_{22 }). \label{gredensity_matrix212l}\end{aligned}\]] here @xmath128 and @xmath129 are the steady - state values of @xmath130 and @xmath131, respectively, when @xmath132, and are introduced to allow for relaxation to equilibrium. in terms of @xmath133, @xmath134, and @xmath75, the above equations lead to @xmath135 where @xmath136, @xmath137 and @xmath138. note that these are then of the same form as the usual bloch equations for a monochromatic field. torrey gave detailed analytical solutions for these equations in the monochromatic field case @xcite and the same analytical solutions for the 1 vs. 3 photon phase control case can be used, where the single field @xmath139 considered by torrey is replaced by @xmath51. we do not pursue this direction in this letter.
Summary
in summary, we have obtained an analytic solution for @xmath0 vs. @xmath1 photon phase control of a two - level system in an environment described by a @xmath140 dephasing time, with @xmath141 and @xmath142 as a specific example. the results should serve as a prototype for understanding the results of @xmath0 vs. @xmath1 photon phase control in more complicated systems, such as controlled xenon ionization and ibr photodissociation@xcite. 100 p. brumer and m. shapiro, _ principles of the quantum control of molecular processes _ (wiley, new york, 2003). s. a. rice and m. zhao, _ optical control of molecular dynamics _ (wiley, new york, 2000). h. rabitz, r. de vivie - riedle, m. motzkus, and k. kompa, science * 288 *, 824 (2000). j. cao, c. j. bardeen, and k. r. wilson, j. chem. phys. * 113 *, 1898 (2000). m. demirplak and s. a. rice, j. chem. * 116 *, 8028 (2002). j. gong and s. a. rice, j. chem. phys. * 120 *, 3777 (2004). b. d. fainberg and v. a. gorbunov, j. chem. phys. * 117 *, 7222 (2002). m. a. nielsen and i. l. chuang, _ quantum computation and quantum information _ (cambridge university press, cambridge, 2000). m. shapiro and p. brumer,, 052308 (2002). p. zanardi and m. rasetti,, 3306 (1997). l. duan and g. guo,, 737 (1998). d. a. lidar, d. bacon, and k. b. whaley,, 4556 (1999). j. i. cirac and p. zoller,, 4091 (1995). c. j. myatt, b. e. king, q. a. turchette, c. a. sackett, d. kielpinski, w. m. itano, c. monroe, and d. j. wineland, nature * 403 *, 269 (2000). d. j. tannor, r. kosloff, and a. bartana, faraday discuss. * 113 *, 365 (1999) and references are therein. m. shapiro, j. w. hepburn, p. brumer, chem.. lett. * 149 *, 451 (1988) c. k. chan, p. brumer, and m. shapiro, j. chem. phys. * 94 *, 2688 (1991). d. petrosyan and p. lambropoulos, phys. lett. * 85 *, 1843 (2000). d. petrosyan and p. lambropoulos, phys. a * 63 *, 043417 (2001). j. c. camparo and p. lambropoulos,, 552 (1997). c. chen, y - y. yin and d.s. elliott, phys.. lett. * 64 *, 507 (1990). c. chen, y - y. yin and d.s. elliott, phys. lett. * 65 * 1737, (1990). x. wang, r. bersohn, k. takahashi, m. kawasaki, and h.l. kim, j. chem. * 105 *, 2992 (1996). karapanagioti, d. xenakis, d. charalambidis and c. fotakis, j. phys. b * 29 *, 3599 (1996). for a review see : r. j. gordon, l. zhu, and t. seideman, acc. 32 *, 1007 (1999). l. k. iwaki and d. l. dlott, j. phys. a. * 104 *, 9101 (2000). h. graener, r. zrl and m. hofmann, j. phys. b. * 101 *, 1745 (1997). h. c. torrey, phys. rev. * 76 *, 1059 (1949). l. allen and j. h. eberly, _ optical resonance and two - level atoms _ (wiley, new york, 1975). h. han and p. brumer (manuscript in preparation). h. han, ph.d. dissertation, university of toronto, 2004. excited state population as a function of time for various dephasing rates, @xmath91 shown inside the box for @xmath143. all the variables are in dimensionless units. data points are connected by straight lines as a guide. excited state population versus relative phase for two different @xmath144 : solid lines and dashed lines denote the case at @xmath145 and @xmath146, respectively. thin lines and thick lines denote the case at @xmath147 and @xmath148, respectively. fields are turned off at (a) @xmath149, (b) @xmath150, (c) @xmath151, and (d) @xmath152. note that the data in panel (d) is too widely spaced to produce the last @xmath153 maxima, whose exact locations can be predicted from eq. ([cos]). all the variables are in dimensionless units. data points are connected by straight lines as a guide. | decoherence effects on the traditional @xmath0 vs. @xmath1 photon coherent control of a two - level system are investigated, with 1 vs. 3 used as a specific example.
the problem reduces to that of a two - level system interacting with a single mode field, but with an effective rabi frequency that depends upon the fundamental and third harmonic fields.
the resultant analytic control solution is explored for a variety of parameters, with emphasis on the dependence of control on the relative phase of the lasers.
the generalization to off - resonant cases is noted. | quant-ph0502058 |
Introduction
neural networks (nn s) have become in the last years a very effective instrument for solving many difficult problems in the field of signal processing due to their properties like non - linear dynamics, adaptability, self - organization and high speed computational capability (see for example @xcite and the papers therein quoted). aim of this paper is to show the feasibility of the use of nn s to solve difficult problems of signal processing regarding the so called virgo project. gravitational waves (gw s) are travelling perturbations of the space - time predicted by the theory of general relativity, emitted when massive systems are accelerated. up to now, there is only an indirect evidence of their existence, obtained by the observations of the binary pulsar system psr 1913 + 16. moreover, the direct detection of gw s is not only a relevant test of general relativity, but the start of a new picture of the universe. in fact, gw s carry complementary information with respect to electromagnetic and optical waves, since the gw s are practically not absorbed by the matter. the aim of the virgo experiment is the direct detection of gravitational waves and, in joint operation with other similar detectors, to perform gravitational waves astronomical observations. in particular, the virgo project is designed for broadband detection from @xmath0 to @xmath1. the principle of the detector is shown in figure 1. a @xmath2 @xmath3 arm - length michelson interferometer with suspended mirrors (test masses) is used. the phase difference @xmath4 between the two arms is amplified using fabry - perot cavities of finesse @xmath5 in each arm. aiming for detection sensitivity of @xmath6, virgo is a very delicate experimental challenge because of the competition between various sources of noise and the very small expected signal. in fact, the interferometer will be tuned on the dark fringe, and then the signal to noise ratio will be mainly limited, in the above defined range of sensitivity, by residual seismic noise, thermal noise of the suspensions photon counting noise (shot noise). in figure 2 the overall sensitivity of the apparatus is shown. in this figure it is easy to see the contribution of the different noise sources to the global noise. in this context we use a multi - layer perceptron (mlp) nn with the back - propagation learning algorithm to model and identify the noise in the system, because we experimentally found that fir nn s and elman nn s did not work in a satisfying manner. both the fir @xcite and elman @xcite models proved to be very sensible to overfitting and were not stable. furthermore the elman network required a great number of hidden units, while the fir network required a great number of delay terms. instead, the mlp proved succesfull and easy to train because we used the bayesian learning paradigm.
Neural networks for time-domain system identification
nn s are massively parallel, distributed processing systems. they are composed of a large number of processing elements (called nodes, units or neurons) which operate in parallel. scalars (called weights) are associated to the connections between units and determine the strength of the connections themselves. computational capability is due to the connections between the units and to their collective behaviour. furthermore, information representation is distributed, i.e. no single unit is charged with specific information. nn s are well - known for their universal approximation capability @xcite. system identification consists in finding the input - output mapping of a dynamic system. a discrete - time multi - input multi - output (mimo) system (or a continuous - time sampled - data system) can be described with a set of non - linear difference equations of the form (_ input - state - output _ representation) : @xmath7 where @xmath8 is the state vector of the system, @xmath9 is the input vector and @xmath10 is the output vector. since we can not always access the state vector of the system, therefore we can use an input - output representation of the form : @xmath11 \label{eq : io}\end{aligned}\]] where @xmath12 and @xmath13 are the maximum lags of the input and output vectors, respectively, and @xmath14 is the pure time - delay (or dead - time) in the system. this is called an arx model (autoregressive with exogenous inputs,) and it can be shown that a wide class of discrete - time systems can be put in this form @xcite. to build a model of the form ([eq : io]), we must therefore obtain an estimation of the functional @xmath15 $], which generally is nonlinear. given a set of input - output pairs, a neural network can be built @xcite which approximates the desired functional @xmath16 $]. such a network has @xmath17 inputs and @xmath18 outputs (see figure 3). a difficulty in this approach arises from the fact that generally we do not have information about the model order (i.e. the maximum lags to take into account) unless we have some insight into the system physics. furthermore, the system is non - linear. recently @xcite a method has been proposed for determining the so - called _ embedding dimension _ of nonlinear dynamical systems, when the input - output pairs are affected by very low noise. furthermore, the lags can be determined by evaluating the _ average mutual information _ (ami)@xcite. such methodologies, although not always successful, can be nevertheless used as a starting point in model design.
Virgo
in the virgo data analysis, the most difficult problem is the gravitational signal extraction from the noise due to the intrinsic weakness of the gravitational waves, to the very poor signal - to - noise ratio and to their not well known expected templates. furthermore, the virgo detector is not yet operational, and the noise sources analyzed are purely theoretical models (often stationary noises), not based on experimental data. therefore, we expect a great difference between the theoretical noise models and the experimental ones. as a consequence, it is very important to study and to test algorithms for signal extraction that are not only very good in signal extraction from the theoretical noise, but also very adaptable to the real operational conditions of virgo. for this reason, we decided the following strategy for the study, the definition and the tests of algorithms for gravitational data analysis. the strategy consists of the following independent research lines. the first line starts from the definition of the expected theoretical noise models. then a signal is added to the virgo noise generated and the algorithm is used for the extraction of the signal of known and unknown shape from this noise at different levels of signal - to - noise ratio. this will allow us to make a number of data analysis controlled experiments to characterize the algorithms. the second line starts from the real measured environmental noise (acoustic, electromagnetic,...) and tries to identify the noise added to a theoretical signal. in this way we can test the same algorithms in a real case when the noise is not under control. using this strategy, at the end, when in a couple of years virgo will be ready for the first test of data analysis, the procedure will be moved to the real system, being sure to find small differences from theory and reality after having acquired a large experience in the field.
A neural network-based model of the virgo system
as we have seen in the introduction, the virgo interferometer can be characterized by a sensitivity curve, which expresses the capability of the system to filter undesired influences from the environment, and which could spoil the detection of gravitational waves (such a noise is generally called seismic noise). the sensitivity curve has the following expression : @xmath19 + s_{\nu } & \quad f\geq f_{\textrm{min } } \\ s(f_{\textrm{min } }) & \quad f < f_{\textrm{min } } \end{array } \right. \label{eq : exprsenscurve}\]] where : * @xmath20 * @xmath21 is the shot noise cut - off frequency * @xmath22 is the pendulum mode * @xmath23 is the mirror mode * @xmath24 is the shot noise the contribution @xmath25 of violin resonances is given by : @xmath26 ^{2}+\phi _ { i}^{2}}+\frac{1}{i^{4}}\frac{f_{i}^{(f)}}{f}\frac{c_{f}\phi _ { i}^{2}}{\left [\left (\frac{f}{if_{i}^{(f)}}\right) ^{2}-1\right] ^{2}+\phi _ { i}^{2 } } \label{eq : violinres}\]] where the different masses of close and far mirrors are taken into account : * @xmath27 * @xmath28 * @xmath29 * @xmath30 * @xmath31 note that we used a simplified curve for our simulations, in which we neglected the resonances (see figure 4). samples of the sensitivity curve @xmath32 can be obtained by evaluating the expression ([eq : exprsenscurve]) at a set of frequencies @xmath33, @xmath34 $]. the samples of the sensitivity curve allow us to obtain the system transfer function (in the frequency domain), @xmath35, such that : @xmath36 > from this, by means of an inverse discrete fourier transform, samples of the system transfer function (in the time domain) can be obtained. our aim is to build a model of the system transfer function ([eq : tranfun]). assuming that the interferometer input noise is a zero mean gaussian process, by feeding it to the system (i.e. filtering it through the system transfer function) we obtain a coloured noise. the so obtained white noise - colored noise pairs can then be used to train an mlp, as shown in figure 3. the first step in building an arx model is the model order determination. to determine suitable lags which describe the system dynamics, we used the ami criterion @xcite. this can be seen as a generalization of the autocorrelation function, used to determine lags in linear systems. a strong property of the ami statistic is that it takes into account the non - linearities in the system. usually, the lag is chosen as the first minimum of the ami function. the result is reported in figure 5, in which the first minimum is at 1. to find how many samples are necessary to unfold the (unknown) state - space of the model (the so called _ embedding dimension _ @xcite) we used the method of @xcite, the lipschitz decomposition. the result of the search is reported in figure 6. from the figure we can see that, starting from lag three, the order index decreases very slowly, and so we can derive that the width of the input window is at least three. in order to test the nn s capability in solving the problem, we chose a width of 5, both for input and output (i.e. @xmath37). in this way, we obtained a nn with a simple structure. furthermore, some preliminary experiments showed that the system dead - time is @xmath38 ; this gives the best description of the system dynamics. another fundamental issue is the nn complexity, i.e. the number of units in the hidden layers of the nn. usually the determination of the network complexity is critical because of the risks of overfitting. since the nn was trained following a bayesian framework, then overfitting was of no concern ; so we directed our search for a model with the minimum possible complexity. in our case, we found a hidden layer with 6 @xmath39 units is optimal. the bayesian learning framework (see @xcite and @xcite) allows the use of a _ distribution _ of nn s, that is, the model is a realization of a random vector whose components are the nn weights. the so obtained nn is the _ most probable _ given the data used to train it. this approach avoids the bias of the cross - validatory techniques commonly used in practice to reduce model overfitting @xcite. to allow for a smooth mapping which does not produce overfitting, several regularization parameters (also called _ hyperparameters _) embedded in the error criterion have been used : * one for each set of connections out of each input node, * one for the set of connections from hidden to output units, * one for each of the bias connections, * one for the data - contribution to the error criterion. usually, the hyperparameters of the first three kinds are called _ alphas _, while the last is called a _ beta_. the approach followed in the application of the bayesian framework is the `` exact integration '' scheme, where we sample from the analytical form of the distribution of the network weights. this can be done if we assume an analytic form for the prior hyperparameters distribution, in particular a distribution uniform on a logarithmic scale : @xmath40 this distribution is _ non - informative _, i.e. it embodies our complete lack of knowledge about the values the hyperparameters should take. the chosen model was trained using a sequence of little less than a million of patterns (we sampled the system at 4096hz for 240s) normalized to zero mean and unity variance with a _ whitening _ process. note that the input - output pairs were processed through discrete integrators to obtain pattern - target pairs, as shown in figure 3. the nn was then tested on a 120s long sequence. the nn was trained for @xmath41 epochs, with the hyperparameters being updated every @xmath42 epochs. a close look at the @xmath43 hyperparameters shows that all the inputs are relevant for the model (they are of the same magnitude ; note that this further confirms the pre - processing analysis). the @xmath44 hyperparameter shows that the data contribution to the error is very small (as we would expect, since the data are synthetic). the simulations were made using the matlab@xmath45 language, the netlab toolbox @xcite and other software designed by us. in figure 7, the psds of the target and the predicted time series are shown ; in the lower part of the figure is reported the psds of the prediction residuals. in figure 8, the psd of a 100hz sine wine added to the noise is shown, with the signal extracted by the network. as can be seen, the network recognizes the frequency of the sine wave with the maximum precision allowed by the residuals.
Conclusion
in this paper we have shown some preliminary tests on the use of nn s for signal processing in the virgo project. some observations can be elicited from the experimental results : * in evaluating the power spectral densities (psds), we made the hypothesis that the system is sampled at @xmath46. it is only a work hypothesis, but it shows how the network reproduces the system dynamics up to @xmath47. note that the psds are nearly the same also if we were near the nyquist frequency. * the psd of the residuals shows a nearly - white spectrum, which is index of the model goodness (see @xcite). - : : to increase the system model order and to test if there are significant differences in prediction ; - : : to test the models with a greater number of samples to obtain a better estimate of the system dynamics ; - : : to model the noise inside the system model to improve the system performance and to allow a multi - step ahead prediction (i.e. an output - error model) | in this paper a neural networks based approach is presented to identify the noise in the virgo context.
virgo is an experiment to detect gravitational waves by means of a laser interferometer.
preliminary results appear to be very promising for data analysis of realistic interferometer outputs. | astro-ph9906107 |
Introduction
the degree of linear polarization of sunlight scattered by an asteroid toward an observer depends on the phase - angle, namely the angle between the asteroid - sun and the asteroid - observer directions. the phase - polarization curves of all atmosphereless bodies of the solar system exhibit qualitatively similar trends, but their detailed features vary according to the specific properties (including primarily the geometric albedo) of individual surfaces. in the phase - angle range @xmath1, asteroids exhibit the so - called branch of _ negative polarization _, in which, in contrast to what is expected from simple single rayleigh - scattering or fresnel - reflection mechanisms, the plane of linear polarization turns out to be parallel to the plane of scattering (the plane including the sun, the target and the observer). the plane of linear polarization becomes perpendicular to the scattering plane, a situation commonly described as _ positive polarization _, at phase angle values larger than the so - called _ inversion angle _, which is generally around @xmath2. a few years ago, @xcite discovered a class of asteroids exhibiting peculiar phase - polarization curves, characterized by a very unusual extent of the negative polarization branch, with an inversion angle around @xmath3, much larger than the values commonly displayed by most objects. since the prototype of this class is the asteroid (234) barbara, these objects have been since then commonly known as _ barbarians_. only half a dozen barbarians are known today : (234) barbara, (172) baucis, (236) honoria, (387) aquitania, (679) pax, and (980) anacostia @xcite. the polarimetric properties of the barbarians are fairly unexpected. the observed large negative polarization branch is not predicted by theoretical models of light scattering, but in fairly special situations, including surfaces composed of very regularly - shaped particles (spheres, crystals) or surfaces having considerable microscopic optical inhomogeneity @xcite. although barbarians are certainly uncommon, they do exist, and the interpretation of their polarization features may lead to important advances in our understanding of both light - scattering phenomena, and of the origin and evolution of these objects. potential explanations range from peculiar surface composition and/or texture, to the possible presence of anomalous properties at macroscopic scales due the presence of large concavities associated with big impact craters @xcite. for instance, (234) barbara has a very long rotation period, which might be the effect of a big collision. @xcite suggested that (234) barbara could have a surface characterised by large - scale craters. this is confirmed by an analysis of still unpublished occultation data by one of us (pt). in terms of taxonomy based on spectro - photometric data, all known barbarians are classified as members of a few unusual classes, including @xmath4, @xmath5, and (in only one case) @xmath6. (234) barbara itself is an @xmath5 asteroid (here we use the taxonomic classification of *???. however, there are @xmath4-class asteroids which are normal " objects not exhibiting the barbarian properties. this fact seems to rule out a direct relationship between taxonomic class (based on the reflectance spectrum) and polarimetric properties. on the other hand, @xmath4, @xmath5 and @xmath6 classes are located, in a principal component analysis plane, along adjacent locations, which although non - overlapping, seem to represent some kind of continuous spectral alteration surrounding the most common @xmath7 class complex. the fact that the six known barbarians identified so far belong all to one of these three classes suggests that surface composition could be responsible for their polarimetric properties. even more important, two @xmath4-class barbarians, (387) aquitania and (980) anacostia, exhibit very similar reflectance spectra, both sharing the rare property of displaying the spectral signature of the spinel mineral @xcite. actually, it was exactly the fact that (980) anacostia was found to be a barbarian that led @xcite to observe polarimetrically (387) aquitania, and to discover that also this object shares the same polarimetric behaviour. spinel ([fe, mg]al@xmath8o@xmath9) is a mineral characterized by indistinct cleavage and conchoidal, or uneven fracture properties. in terms of optical properties, the mgal@xmath8o@xmath9 form of spinel has a fairly high refractive index (@xmath10), which becomes even higher in the spinel variety having a high iron content (hercynite) (@xmath11, i.e., much above the values characterizing the most common silicates present on asteroid surfaces, *??? spinel is an important component of calcium aluminum - rich inclusions (cai) found in all kinds of chondritic meteorites. cais are refractory compounds which are thought to be among the first minerals to have condensed in the proto - solar nebula. they are the oldest samples of solid matter known in our solar system, and they are used to establish the epoch of its formation @xcite. in terms of spectroscopic properties, spinel is characterized by the absence (or extreme weakness) of absorption bands around 1@xmath12 m, and by the presence of a strong absorption band at 2@xmath12 m. @xcite concluded that, to model the available near - ir spectra of spinel - rich asteroids, it is necessary to assume abundances of the order of up to 30% of cai material on the surface. this extreme abundance, which causes a high refractive index, might also be responsible for the anomalous polarization properties. such high cai abundances have never been found in meteorite on earth (so far, the richest cai abundance, found on cv3 meteorites, is about 10%). therefore, @xcite conclude that spinel - rich asteroids might be more ancient than any known sample in our meteorite collection, making them prime candidates for sample return " missions. many interesting questions are certainly open. which processes are involved in the onset of physical mechanisms which produce the barbarian polarimetric behaviour? are barbarians really among the oldest objects accreted in our solar system? if so, why are they the only one class of primitive objects being characterized by an anomalous polarimetric behaviour? why are they so rare? are they unusually weak against collisions and fragmentation? do space - weathering phenomena affect their polarimetric properties? are the taxonomic classifications of some @xmath4, @xmath5, @xmath6 objects possibly wrong (in such a way that all barbarians might be member of a unique class and not spread among three of them)? it is clear that an important pre - requisite to improve our understanding of these objects, and to draw from them some possible inferences about the origin, composition and subsequent evolution of the planetesimals orbiting the sun at the epoch of planetary formation, is to find new members of the barbarian class, but where to look for them? an aid to a barbarian search comes from the fact, recently confirmed by @xcite, that the spinel - bearing barbarian (980) anacostia belongs to a family of high - inclination asteroids. this family is named watsonia from its lowest - numbered member (729) watsonia. (980) anacostia belongs to a small grouping which is included in the watsonia family and merges with it at larger values of mutual distances between the members. in other words, (980) anacostia belongs to a distinct sub - group of the watsonia family, but the possible independence of this sub - group from the rest of the family is highly uncertain and questionable @xcite. another known barbarian, (387) aquitania, though not being a member of the watsonia family, is also located in close vicinity in the space of proper orbital elements (see section[results]). finally, a member of the watsonia family, asteroid (599) luisa, is not a known barbarian (no published polarimetric measurements are available for it), but it is known to be one of the few spinel - rich asteroids identified so far (thus sharing some common properties with both anacostia and aquitania). one should also note that the distribution of albedos of the members of the watsonia family observed at thermal ir wavelengths by the wise satellite @xcite is strongly peaked around values between @xmath13 and @xmath14 with only a few likely interlopers, suggesting that the family is not a statistical fluke web facility available at url http://mp3c.oca.eu/]. a possible common collisional origin of these asteroids opens new perspectives for the search of new barbarians and for the interpretation of their properties. asteroid families are the outcomes of fragmentation of single parent bodies disrupted by catastrophic collisions. therefore, if (980) anacostia, and possibly also (387) aquitania, were issued from the disruption of a common parent body exhibiting the rare properties which produce the barbarian polarization phenomenon, it would be natural to expect that also the other, still not observed members of the watsonia family should be found to be barbarians. moreover, among the watsonia family members at least one, (599) luisa, is a known spinel - rich asteroid, like anacostia and aquitania. finally, wise albedo values in the same range of those of watsonia family members have also been derived for most barbarian asteroids known so far, namely 234, 172, 236, 679 and 980.
New observations
ccrrcr@@xmath15lr@@xmath15lc date & time (ut) & & & phase angle & & & wise + & & (sec) & & & & & albedo + & & & & & + 2013 07 05 & 23:41 & 480 & 5492 & 18.79 & @xmath161.14 & 0.10 & 0.05 & 0.10 & @xmath17 + 2013 07 29 & 01:44 & 960 & & 18.31 & @xmath161.01 & 0.09 & 0.00 & 0.09 + 2013 06 03 & 09:40 & 2000 & 42365 & 23.30 & @xmath160.83 & 0.15 & @xmath160.06 & 0.15 & @xmath18 + 2013 08 03 & 09:08 & 960 & & 18.55 & @xmath161.73 & 0.12 & @xmath160.06 & 0.12 + 2013 07 12 & 23:50 & 960 & 56233 & 17.83 & @xmath161.07 & 0.16 & @xmath160.09 & 0.16 & @xmath19 + 2013 08 05 & 01:03 & 2200 & & 19.31 & @xmath161.09 & 0.12 & @xmath160.07 & 0.12 + 2013 07 30 & 00:38 & 2000 & 106059 & 18.30 & @xmath161.06 & 0.30 & @xmath160.15 & 0.31 + 2013 08 04 & 01:09 & 4000 & & 18.82 & @xmath160.94 & 0.14 & 0.00 & 0.14 + 2013 08 28 & 01:24 & 4800 & & 19.64 & @xmath160.84 & 0.20 & @xmath160.09 & 0.20 + 2013 07 06 & 01:33 & 1400 & 106061 & 20.21 & @xmath160.94 & 0.11 & 0.06 & 0.11 + 2013 08 09 & 02:46 & 4000 & & 23.97 & @xmath160.57 & 0.12 & @xmath160.05 & 0.12 + 2013 07 06 & 02:05 & 960 & 144854 & 21.30 & @xmath160.81 & 0.12 & 0.13 & 0.12 + 2013 08 05 & 02:06 & 4000 & & 24.19 & @xmath160.15 & 0.12 & @xmath160.13 & 0.13 + 2013 08 13 & 06:28 & 4000 & 236408 & 18.31 & @xmath160.97 & 0.15 & 0.22 & 0.15 & @xmath20 + 2013 07 07 & 02:40 & 3440 & 247356 & 19.97 & 0.10 & 0.15 & @xmath160.33 & 0.15 + 2013 04 17 & 08:58 & 4800 & 320971 & 23.78 & 0.11 & 0.31 & @xmath160.06 & 0.31 + 2013 06 03 & 07:08 & 3440 & & 20.53 & @xmath160.03 & 0.22 & @xmath160.06 & 0.21 + light) for the targets of the present investigation, compared with the polarization curves (in @xmath21 light) of the (234) barbara, and of (12) victoria, a large @xmath4-class asteroid which does not exhibit the barbarian behaviour. black symbols : the seven targets exhibiting the barbarian polarimetric behaviour ; red symbols : our two targets that display a `` normal '' polarimetric behaviour ; small blue symbols and blue curve : available data for (234) barbara @xcite, and the corresponding best - fit curve using the linear - exponential relation @xmath22 + c \cdot \alpha$], where @xmath23 is the phase angle in degrees ; dashed, green curve : the best - fit curve for the @xmath4-class asteroid (12) victoria (for the sake of clearness, individual observations of this asteroid are not shown).,width=283] our target list includes nine objects that are members of the watsonia family (not limited to the anacostia sub - group) according to @xcite. these objects are listed in table [tab_observations]. we note that a more recent family search whose results are publicly available at the astdys web site also identified a watsonia family, but using a more conservative value of the critical level of mutual distances between family members to be adopted to define the family. as a consequence, the astdys version of the watsonia family has a smaller membership list, and does not include our target objects 42365, 144854, 247356 and 320971. we observed our targets using the vlt fors2 instrument @xcite in imaging polarimetric mode, and obtained @xmath24 broadband linear polarization measurements in the @xmath25 special filter from april to september 2013. the choice of an @xmath25 filter (instead of the standard @xmath21 filter traditionally used in many asteroid polarimetric observations) was dictated by the need of improving the s / n ratio. polarimetric measurements were performed with the retarder wave - plate at all positions between 0 and 157.5, at 22.5 steps. for each observation, the exposure time cumulated over all exposures varied from 8min (for 5492) to 1h and 20min (for 320971). data were then treated as explained in @xcite, and our measurements are reported adopting as a reference direction the perpendicular to the great circle passing through the object and the sun. this way, @xmath26 represents the flux perpendicular to the plane sun - object - earth (the scattering plane) minus the flux parallel to that plane, divided by the sum of these fluxes. it is therefore identical to the parameter indicated as @xmath27 in many asteroid polarimetric studies. in these conditions, for symmetry reasons, @xmath28 values are expected to be zero, and inspection of their values confirms that this is the case for our observations. at small phase angles (@xmath29), all asteroids exhibit negative polarization, whereas at phase angles @xmath30, nearly all asteroids exhibit positive polarization. by contrast, at phase angle @xmath31, barbarians still exhibit a relatively strong negative polarization (@xmath32%). therefore, to identify barbarians, we decided to measure the polarization at phase angles @xmath33, and establish whether the measured polarization plane is found to be parallel or perpendicular to the scattering plane.
Results
the results of our polarimetric measurements are given in table [tab_observations] and shown in figure [figure1], in which a comparison is also made with the phase - polarization curve of (234) barbara and (12) victoria, a big, non - barbarian @xmath4-class asteroid. in the figure each target is represented by a different symbol, and some targets have more than one measurement. in the observed phase - angle range, seven of our targets show a polarization value @xmath32%, consistent with the value exhibited by (234) barbara. in fact, (234) barbara exhibits marginally higher polarization values (in absolute value) than our targets, possibly due to the different filter in which the observations were performed. the striking result of our observing campaign is that seven out of nine asteroids of our target list are barbarians. the two exceptions are 247356 and 320971. however, we know _ a priori _ that all family member lists are expected to include some fraction of random interlopers @xcite, then we conclude that both non - barbarian objects in our target list may be family interlopers. for instance, asteroid 320971 is listed as a watsonian member by @xcite, but it is not included among the watsonia members identified in the new astdys list of asteroid families. on the other hand, three other targets not included in the astdys member list, but present in the @xcite member list, are found to be barbarians. we conclude therefore that, apart from some details concerning family membership, the watsonia family is an important repository of barbarian asteroids. this is the immediate and most important result of our investigation. we also note that the number of watsonia barbarians identified in our observing campaign is larger than the whole sample of barbarians previously known. we also note that, except in case of 236408 and 247356 (only a single measurement each) the resulting @xmath34 values are always well consistent with zero.
Discussion and future work
our results confirm once again that asteroid polarimetry can provide an important contribution to asteroid taxonomy, as noted in the past by several authors (see, for instance, *??? several problems, however, are now open, and deserve further theoretical and observational efforts. the first issue to be addressed is the relationship between the barbarian polarimetric features and the unusual amount of spinel measured via spectroscopy in some of the known barbarians. in other words, the new barbarians that we have found in our investigation must be spectroscopically observed in the visible and near - ir wavelengths in order to check whether they exhibit the spinel features. we do believe that this will be the case, since we already know three objects that are both spinel - rich and barbarians (234 barbara, 387 aquitania and 980 anacostia), ad we know also that the watsonia family includes at least one other spinel - rich asteroid (599 luisa) @xcite, but we clearly need confirmation from new observations. similarly, we need to search for the barbarian polarimetric feature in other spinel - rich asteroids. in particular, we are interested not only in 599 luisa, but also in the henan family, which is known to include at least three spinel - rich asteroids @xcite. next, we need to understand why the barbarian and the anomalous spinel abundance phenomena are so rare. apparently, only a handful of barbarian parent bodies existed, and only their disruption into many fragments made it possible to identify today larger numbers of these objects. in principle, one might wonder whether the unusual properties displayed by these objects might also be in some (still obscure) way a consequence of the collisional events themselves. another open problem is the origin of (387) aquitania. this asteroid is not included in any family list, including both the most recent classification available at the astdys site, and the @xcite classification of high-@xmath35 families. yet, this object is quite close, in terms of orbital elements, to (980) anacostia and the rest of the watsonia family. the values of proper elements for (387) aquitania and (980) anacostia (available in the astdys database) are @xmath36, @xmath37, @xmath38, for (387) aquitania, and @xmath39, @xmath40, @xmath41 for (980) anacostia, where @xmath42 is the proper semi - major axis, @xmath43 the proper eccentricity, and @xmath35 is the proper inclination. the only one relevant difference, which prevents any family search to include them in an acceptable group, is the proper eccentricity. one should conclude that there is no relation between (387) aquitania and (980) anacostia, and that the similarity of their orbital semi - major axes and inclinations are just a coincidence. moreover, both asteroids display stable orbits in terms of characteristic lyapunov exponents, and any possible evolution due to non - gravitational forces, (e.g. the yarkovsky effect) seem also very unlikely, because both asteroids are fairly large, with sizes of the order of @xmath44 km according to thermal radiometry data. finally, the collision which could have produced a family including two fragments of this size had to be extremely energetic, and should have produced a huge family, of which there is no evidence today. on the other hand, aquitania and anacostia exhibit too many relevant similarities in their physical properties, which suggests that they may come from a unique parent body fragmented by a collision. a very tentative scenario is that, in a very early epoch, a violent collision destroyed a big parent body which was the progenitor of the barbarians we see today in this zone. most of the first - generation fragments produced by the event are now gone due to gravitational perturbations and yarkovsky - driven drift in semi - major axis. the current watsonia family could be the product of a more recent disruption of one of the original fragments of the first event. this region could still include several first - generation fragments, preferentially the biggest ones which, due to their size, experienced no or extremely weak yarkovsky evolution, like (387) aquitania and (980) anacostia. some gravitational perturbations could have also played some role. if one looks at the @xmath42 - @xmath43 or @xmath42 - @xmath45 plots making use of the facility available at the astdys site, it is easy to see that the watsonia family is crossed, in semi - major axis, by a couple of thin resonances, one of which is located close to the value of @xmath42 of both (387) aquitania and (980) anacostia. so, we can not rule out entirely the idea that some limited evolution in eccentricity of these objects could be due to these perturbation effects and the two objects might have been originally closer in the proper element space. we stress again that this kind of scenario is only eminently speculative and deserves further analysis before it can be accepted as a plausible one. we point out that a so large relative fraction of observed members of the watsonia family turning out to be barbarians has a further implications to be taken into account. if the parent body was differentiated, or had some gradient of composition as a function of depth, with the presence of cai material limited to some thin shell, then the spinel signature would be present only in a minority of the family members. this is in disagreement with what we observe, indicating that the anomalous composition of the parent body was not limited to some thin region. in conclusion, we are not looking simply at some kind of surface process (like alteration by space weathering or impacts), but at the properties displayed by asteroids originated from different depths inside a single parent body which was anomalously spinel - rich in most of its volume (this can be even more true if the watsonia family is a secondary event derived by a first - generation destruction of a larger body.) the issues of section [intro] are still open and deserve careful attention. the immediate result of our investigation is that we have now a much larger sample of barbarians than before at disposal for future physical investigations (including photometric, spectroscopic, and polarimetric campaigns.) at the same time, our results definitively confirm the family of watsonia as a sizeable group of barbarians having a common collisional origin, and open new perspectives for the development of models in order to try and sort out in a coherent scenario several pieces of observational evidence which appear now to form a complicated puzzle.
Acknowledgments
part of this work was supported by the cost action mp1104 _ polarization as a tool to study the solar system and beyond_. in particular, funding for short term scientific mission cost- stsm - mp1104 - 12750 of ac at the armagh observatory is kindly acknowledged. the work of b.n. has been supported by the ministry of education and science of serbia, under the project 176011. a careful review by k. muinonen helped to improve the paper. | barbarians, so named after the prototype of this class (234) barbara, are a rare class of asteroids exhibiting anomalous polarimetric properties.
their very distinctive feature is that they show negative polarization at relatively large phase - angles, where all `` normal '' asteroids show positive polarization.
the origin of the barbarian phenomenon is unclear, but it seems to be correlated with the presence of anomalous abundances of spinel, a mineral usually associated with the so - called calcium aluminum - rich inclusions (cais) on meteorites.
since cais are samples of the oldest solid matter identified in our solar system, barbarians are very interesting targets for investigations.
inspired by the fact that some of the few known barbarians are members of, or very close to the dynamical family of watsonia, we have checked whether this family is a major repository of barbarians, in order to obtain some hints about their possible collisional origin.
we have measured the linear polarization of a sample of nine asteroids which are members of the watsonia family within the phase - angle range @xmath0.
we found that seven of them exhibit the peculiar barbarian polarization signature, and we conclude that the watsonia family is a repository of barbarian asteroids.
the new barbarians identified in our analysis will be important to confirm the possible link between the barbarian phenomenon and the presence of spinel on the surface.
[firstpage] asteroids : general polarization. | 1401.1931 |
Introduction
the internet has become a near indispensable tool with both private individuals and organizations becoming increasingly dependent on internet - based software services, downloadable resources like books and movies, online shopping and banking, and even social networking sites. the issue of network security has become significant due to the prevalence of software with malicious or fraudulent intent. malware is the general term given to a broad range of software including viruses and worms designed to infiltrate a computer system without the owner s permission @xcite@xcite. cohen s conclusion in his 1987 paper that computer viruses are potentially a severe threat to computer systems @xcite is still valid in real networks today @xcite@xcite@xcite. current security systems do little to control the spread of malicious content throughout an entire network @xcite@xcite. most security systems are designed to protect a single computer unit. these properly protected units make up only a fraction of online computers. these highlight the necessity of examining the dynamics of the spread of malware in order to be able to develop proper control strategies. studies on the spread of malware in computer networks date back to the late 1980s @xcite and are generally based on the mathematical approach to the spread of diseases in biological populations. math models developed for spread of malware within a computer network such as the kephart - white model and other models adapted from it are based on the kermack - mckendrick model. these models have an implicit assumption that all nodes in the network are always available for `` contact '' @xcite@xcite. however, it is a basic limitation of malware that it can only be passed on to another computer if there is a path through which information can be passed @xcite, so the states of the nodes of the network whether they are online or offline have an effect on the dynamics of the spread. in this work, we model the spread of malware utilizing an ising system to represent an isolated computer network. the state of each node is a composite of its connection status and health. the spin state of a node defines its connection status to be either online or offline. connections are established with the premise that autonomous networks configure themselves @xcite. the health status describes whether a node has been infected or not, and infection can propagate only among online nodes. the ising model was originally intended for simulating the magnetic domains of ferromagnetic materials. its versatility has allowed it to be applied to other systems wherein the behavior of individuals are affected by their neighbors @xcite@xcite@xcite. it has been applied to networks and network - like systems @xcite such as neural networks @xcite@xcite, cooperation in social networks, and analysing trust in a peer - to - peer computer network @xcite.
The model
a computer network is modeled by an @xmath0 ising spin system. associated with each node is a spin @xmath1 corresponding to two possible states : @xmath2 for online and @xmath3 for offline. the local interaction energy is given by @xmath4 the interaction parameter, @xmath5, determines the degree and type of dependence of @xmath1 on its neighbors. the nearest neighbors or local neighborhood are defined according to the network topology and are usually von neumann or moore neighborhoods @xcite@xcite. summing up all local energies gives the total energy, @xmath6, of the system. global energy, @xmath6, is associated with network efficiency and more efficient networks are characterized by lower energies. note that while interaction energies are explicitly dependent on the nearest neighbors, the state of each node is implicitly dependent on the state of the entire system. a node will change its configuration provided that the new energy of the system is lower than the previous. if the resulting energy is higher, the new configuration is accepted with probability @xmath7 in the standard ising procedure, @xmath8 is the change in energy, @xmath9 is temperature, and @xmath10 is the boltzmann constant. here, @xmath9 relates to network traffic. to model the spread of infection, each node is assigned a health status separate from its spin. the health status is either infected or susceptible. every online susceptible has a probability @xmath11 of becoming infected, where @xmath12 offline nodes do not transmit or receive data. hence, they do not participate in the infection part. [[program - specifics]] program specifics + + + + + + + + + + + + + + + + + the computer network is a @xmath13 lattice. nearest neighbors are defined to be the four adjacent nodes. the interaction parameters are all set to @xmath14. eq.[generalising] becomes @xmath15 for the interaction energy calculations, circular boundary conditions are imposed. parameters are scaled such that @xmath16. initially, all nodes are offline (@xmath17). every time step, the entire system is swept in a left - to - right top - to - bottom fashion, evaluating each node for a possible change in state. the mean energy per node @xmath18 of each configuration is stored and averaged at the end of the run. the spread of infection begins with a single infective. at @xmath19, one node is selected at random and infected. as the infection spreads, the number of susceptibles, @xmath20, and infectives, @xmath21, for each time step are stored. because no means for removal of infection is provided, all nodes eventually become infected. it is at this time that the program is terminated.
Analysis of results
the model was tested for @xmath9-values ranging from @xmath22 to @xmath23. the infection curves of five trials were averaged for each @xmath9. the average infection curve was normalized by dividing it by the total number of nodes to get the fraction of infectives @xmath24. because it can no longer be assumed that nodes are always available for connection, a regular decay equation is used to model the fraction of infectives curve. a system with @xmath25 nodes has @xmath20 susceptibles and @xmath21 infectives at time @xmath26. within the time - frame @xmath27, the number of susceptibles being converted to infectives is @xmath28. as time passes, @xmath28 decreases as the population of susceptibles is exhausted. thus, the probability of conversion, given by @xmath29 decreases with time. in equation form, this is @xmath30 where @xmath31 is the decay constant. the solution to eq.[decayeq] is @xmath32 where @xmath33, the initial number of susceptibles, is just the total number of units in the system. using these, the expression for the number of infectives, @xmath21 may be written as @xmath34 this may be normalized to @xmath35 note that the actual rate of spread varies with time, and @xmath31 provides a measure of the average rate of spread. the fits were made using the unweighted levenberg - marquardt algorithm of gnuplot ver.4.2 @xcite initialized with @xmath36. for consistency, because some runs terminate very rapidly, we consider only the first 50 time - steps. during the first 50 iterations _ : the rate of spread of the infection increases with @xmath9. for the above graphs, the resulting decay constants are:@xmath37, @xmath38, @xmath39, @xmath40, @xmath41, and @xmath42, width=240] from fig.[allcompare], it appears that the spread of infection becomes faster as @xmath9 increases. for @xmath43 and @xmath44, the rates of spread are very slow, neither reaching @xmath45-infected at the last iteration. particularly, for @xmath43, no new infectives were produced. these low - traffic systems are not dynamic as nodes have a low probability of coming online from their initial offline state. the network is also very efficient, @xmath46 and @xmath47, which may be interpreted as information exchange being limited to necessary transactions. for this reason, there is little information exchange and hence a slow spread. for very high @xmath9, as in @xmath48 and @xmath23, the spread is rapid and nearly @xmath49 infection is reached. this suggests that very high traffic means a large volume of information exchange that leads to a faster spread of infection. the system is also inefficient at very high @xmath9, with @xmath50. it is worth mentioning that the average infection curves of @xmath51 and @xmath52 nearly coincide indicating rates of spread that are very similar. -dependence of rates _ : the increase in the rate of infection corresponds with the decrease in efficiency in the network. note that @xmath6-values are negative.,width=240] the observations are supported by the calculated decay constants. the calculated @xmath31 initially increases with traffic but is capped off at very high @xmath9 where it becomes constant. this behavior is similar to the saturation region in a traffic network where flux saturates at high densities. the saturation region indicates that information exchange is no longer freely flowing and that some kind of congestion has occurred @xcite. in fig. [exponents], there is an evident transition that occurs in both the average rate of spread and the efficiency of the network. at the `` congested '' region, the efficiency of the network is very low ; while at the `` free flow '' region, the efficiency of the network is comparatively high. congestion occurs because networks can only handle a limited amount of traffic in the form of data packets. when there is too much traffic, the network is forced to store or drop packets making it inefficient @xcite@xcite. an increase in packet loss with increasing data traffic is reflected by the decrease in efficiency at the congestion region. the congestion is most likely a result of the limited size of the network and the `` finite - size effect '' may be confirmed by testing a larger network@xcite.
Conclusion
our ising model approach accounts for the connection status of nodes in an infected network. unlike most epidemic models where all nodes are assumed to be always connected, the model allows malware propagation only among online nodes. we found that the rate of infection becomes faster in less efficient networks with higher data traffic and saturates as the network becomes congested. deepayan chakrabarti, yang wang, chenxi wang, jurij leskovec, and christos faloutsos., `` epidemic thresholds in real networks '', _ acm transactions on information and systems security _, vol. 10, no. 4, january 2008, article 13 jeffrey o. kephart, et al. `` biologically inspired defenses against computer viruses '' _ proceedings of the 14th international joint conference on artificial intelligence _ vol. 1, montreal, quebec, canada, 1995, pp.985 - 996 | we introduce an ising approach to study the spread of malware.
the ising spins up and down are used to represent two states online and offline of the nodes in the network.
malware is allowed to propagate amongst online nodes and the rate of propagation was found to increase with data traffic. for a more efficient network,
the spread of infection is much slower ; while for a congested network, infection spreads quickly.
computer networks ; computer viruses ; epidemiology | 1007.4938 |
Introduction
hera deeply inelastic scattering (dis) results on structure functions demonstrate a rapid bremsstrahlung growth of the gluon density at small x. when interpreted in the same framework as the parton model, this growth is predicted to saturate because the gluon occupation number in hadron wave functions saturate at a value maximally of order @xmath1 ; dynamically, nonlinear effects such as gluon recombination and screening by other gluons deplete the growth of the gluon distribution@xcite. gluon modes with @xmath2 are maximally occupied, where @xmath3 is a dynamically generated semi - hard scale called the saturation scale. for small @xmath4, @xmath5 is large enough that high occupancy states can be described by weak coupling classical effective theory@xcite. this color glass condensate description of high energy hadrons and nuclei is universal and has been tested in both dis and hadronic collisions. in particular, saturation based phenomenological predictions successfully describe recent lhc p+p data @xcite and predict possible geometrical scaling of transverse momentum distribution@xcite similar to the geometrical scaling observed previously in dis. the object common to dis and hadronic collisions is the dipole cross section @xmath6. in the cgc framework, the dipole cross section can be expressed in terms of expectation values of correlators of wilson lines representing the color fields of the target. the energy dependence of this quantity comes from renormalization group evolution but to get the realistic impact parameter dependence one has to rely on models involving parametrizations constrained by experimental data. in the large @xmath7 limit, the dipole cross section is related to the un - integrated gluon distribution inside hadron / nucleus as @xmath8^{2}. \label{eq : unint - gluon}\]] for hadron - hadron collisions, the inclusive gluon distribution which is @xmath9-factorizable into the products of un - integrated gluon distributions in the target and projectile is expressed as @xmath10
Saturation models of hera dis
two models of the dipole cross - section that have been extensively compared to hera data are the ip - sat @xcite and the b - cgc @xcite models. in the former the impact parameter dependence is introduced through a normalized gaussian profile function @xmath11 and in the latter through a scale @xmath12. for a detailed discussion of the parameters involved in these models and their values from fits to hera data, see ref. @xcite. the saturation scale in the fundamental representation for both the models can be calculated self consistently solving @xmath13=2(1-e^{-1/2})$]. the corresponding adjoint saturation scale @xmath14, relevant for hadronic collisions, is obtained by multiplying @xmath15 by 9/4. in the range @xmath16-@xmath17, the behaviour of @xmath14 (see fig.[fig : satscale] left) at @xmath18 can be approximated by a function of the form @xmath19 with @xmath20 for the b - cgc model and @xmath21 for the ip - sat model. [fig : multdist]
Results and discussion
multiparticle production in high energy hadronic collisions can be treated self consistently in the cgc approach. the glasma flux tube picture @xcite predicts @xcite that the n - particle correlation is generated by the negative binomial distribution @xmath22. it is characterized by two parameters, the mean multiplicity @xmath23 and @xmath24. at a given impact parameter of the collision, the mean multiplicity @xmath25 is obtained by integrating eq. [eq : ktfact1] over @xmath26. in the glasma picture, the parameter @xmath27 with @xmath28 @xcite. the quantity @xmath29 shown in fig.[fig : satscale] (right) is the number of flux tubes in the overlap area @xmath30 of two hadrons. convolving @xmath31 with the probability distribution @xmath32 for an inelastic collision at @xmath33-fig. [fig : multdist] (left)-one obtains @xcite the n - particle inclusive multiplicity distribution as shown in fig. [fig : multdist] (right). various kinematic variables exhibit scaling with the saturation scale@xcite. the mid - rapidity multiplicity density scales with functional forms like @xmath34 and @xmath35 whereas a linear functional form seem to provide very good fit to the energy dependence of @xmath36 as shown in fig.[fig : scaling][left]. these results are suggestive that @xmath37 is the only scale that controls the bulk particle multiplicity. in ref. @xcite it has been shown that @xmath26 spectra in @xmath38 collisions exhibit geometric scaling assuming a simple form of @xmath37. in our case we use a scaling variable @xmath39, where @xmath37 is directly calculated in the ip - sat model. as shown in fig.[fig : scaling][right], an approximate scaling below @xmath40 is observed for transverse momentum distribution in @xmath38 collision energy @xmath41 gev. going to lower energies we observe systematic deviations from the universal curve. + in summary, our description of multiplicity distribution successfully describes bulk lhc p+p data. in particular, we observe that the dominant contribution to multiplicity fluctuations is due to the intrinsic fluctuations of gluon produced from multiple glasma flux tubes rather than from the fluctuations in the sizes and distributions of hotspots. the @xmath26-spectra in p+p at high energies exhibits universal scaling as a function of @xmath39. the observed scaling indicates that particle production in this regime is dominantly from saturated gluonic matter characterized by one universal scale @xmath37. ridge like two particle correlation structures in @xmath42 in high multiplicity p+p collisions may provide more detailed insight into its properties @xcite. v. khachatryan _ et al. _ [cms collaboration], phys. lett. * 105 *, 022002 (2010). k. aamodt _ et al. _ [alice collaboration], eur. j. c * 68 *, 345 (2010). a. dumitru, k. dusling, f. gelis, j. jalilian - marian, t. lappi, r. venugopalan, arxiv:1009.5295 [hep - ph]. | dipole models based on various saturation scenarios provide reasonable fits to small - x dis inclusive, diffractive and exclusive data from hera.
proton un - integrated gluon distributions extracted from such fits are employed in a @xmath0-factorization framework to calculate inclusive gluon distributions at various energies.
the n - particle multiplicity distribution predicted in the glasma flux tube approach shows good agreement with data over a wide range of energies.
hadron inclusive transverse momentum distributions expressed in terms of the saturation scale demonstrate universal behavior over a wider kinematic range systematically with increasing center of mass energies.
saturation ; lhc p + p collision ; cgc ; deep inelastic scattering | 1101.5922 |
Introduction
the introduction of reflective optical elements revolutionized the design and imaging capabilities of telescopes @xcite. while the possibility of creating large diameter objectives was one of the primary aims, a significant advantage is that reflective optical elements provide a means of reducing or eliminating the spherical and chromatic aberrations that are inherent in lens based systems. here we explore the opposite extreme of microfabricated optics. microfabricated optical elements are key components for the development and integration of optics into a range of research and commercial areas @xcite. to date, the majority of the work in microphotonics has been in refractive elements, i.e. microlenses. however, in this regime microlenses typically have significant numerical apertures and surface curvatures, which introduce large aberrations. a number of groups have in recent years discussed the design and fabrication of concave micromirrors @xcite. these examinations are largely driven by two purposes ; for optical tweezing, and for integration into atom optics. spherical mirrors have been demonstrated to collect light from single ions @xcite. parabolic mirrors, similar to those described here but with larger length scales, have been also been used as highly efficient collectors of light from single ions @xcite, atoms @xcite, and point sources @xcite. in a similar manner they may be used to tightly focus light onto atomic samples, which has to date been shown with refractive optics @xcite. in this work we consider the use of reflective micro optical components for focusing light. we present the construction and optical characterization of parabolic reflectors with an open aperture of radius 10 @xmath0 m and measured focal lengths that range from 24 @xmath0 m to 36 @xmath0 m. detailed mapping of the focused intensity field is made possible by the development of a previously unreported adaptation of a confocal microscope that allows the illumination of the reflector with collimated light, while still maintaining the highly - desirable large - numerical - aperture confocal collection. using this device we obtain 3d data about the focal plane demonstrating diffraction limited focussing. we also discuss the application of the parabolic mirror for use in atomic physics and tweezing experiments.
Fabrication
the details of the fabrication of concave paraboloid structures through ion - beam milling are covered in @xcite. briefly, a focused ion - beam (gallium ions with typical currents of 50 - 300 pa and accelerating voltages of 30 kv) is used to precisely sculpt a silicon substrate with the required mirror profile, which is subsequently coated with gold to provide a highly - reflective coating. in focused ion - beam milling, controlling the dose of ions to which an area is exposed allows a region of the surface to be sputtered to a known depth, due to a linear relationship between depth and dose in silicon. the applied dose is a function of the beam current, the dwell time, and number of passes the beam makes over an area. by tracing a number of concentric discs of increasing radius, whilst linearly increasing the dose, a parabolic depression can be milled into the substrate. in principle this would create a stepped contour, however due to edge - effects of the milling process, as well as redistribution of etched material, a larger number of passes creates a smooth contour of the parabolic dish. further details of the construction and characterization can be found in @xcite, where an rms roughness of 4.0 nm was measured by afm over the range of the concave parabolic surface. due to the identical manufacturing process similar values are expected in this work.
Theory
the propagation of light using microscopic optical elements is a well represented topic in the literature. here we restrict our discussion to the behavior of light fields after wavelength scale apertures and curved surfaces. in their work, goldwin and hinds @xcite derive analytic results for spherical mirrors, which they further compare with numerical integration of maxwell s equations. meanwhile, bandi _ et. _ consider the propagation of light after a wavelength - scale aperture using a fresnel representation of the fields @xcite, which offers the possibility of adding focusing to the formalism. here, however, the reflected field was modeled using the angular spectrum method @xcite, which provides a mapping of an electric field from a particular plane into a secondary plane ; @xmath2 where @xmath3 are the cartesian spatial directions, @xmath4 is the wavevector in the @xmath5-direction, @xmath6 is the electric field, and @xmath7 is the 2d fourier spectrum of the electric field in the plane @xmath8. the result of this compact equation can be understood by first noting that the spatial spectrum of an electric field, @xmath9 can be translated from a plane @xmath10 to another plane @xmath8 using the helmholtz propagator, @xmath11 where the helmholtz propagator in reciprocal space is @xmath12, @xcite. we then note that the electric field can be calculated from its spatial spectrum in a plane are by the inverse fourier transform, @xmath13, @xmath14 these relations, eq. (2) and eq. (3) then clearly show the result of eq. (1), and can be used to calculate the electric field in an arbitrary plane, given that is is known in one plane. the essential details of calculating the final intensity profile using the angular spectrum method can be clearly seen in fig. [fig : theory_schematic](c) ; the initial electric field is fourier transformed, the helmholtz propagator is then applied, before the inverse fourier spectrum is taken. finally, the resulting intensity field is found from the modulus - squared of the electric field. example codes are available on request from the corresponding author. for this work, in the plane of @xmath10 a curvature is numerically added to the phase - fronts within the area of the aperture of the mirror. the curvature is expressed as @xmath15, where @xmath16 is the total wavevector, @xmath17, @xmath18 is the radial distance from the center of the mirror, @xmath19 is the mirror aperture radius, and @xmath20 is the focal length. this modification of the optical wavefront represents the position dependent optical path - length difference, due to the spatially varying propagation distance to the metal surface, fig. [fig : theory_schematic]. care must be taken to include all modes of the field, including the evanescent ones, in order to return the full field simulation. all simulations were performed in matlab with the focal length, @xmath20 and the position of the surface as free parameters for fitting to the measured data.
Probing
although transmission confocal imaging provides a powerful means of examining microlenses @xcite, reflection images of focusing elements require post analysis to interpret the data. in traditional reflection confocal microscopy the illuminating laser is focused onto the imaging plane at a point that is confocal with a pinhole in the imaging axis of the microscope. while this allows accurate probing of the surface it precludes imaging the focal plane of our concave mirrors directly. instead one may directly image about the center of curvature of the mirror and then infer the transfer function of the mirror assuming the point spread functions of the other optical elements are known. in this work we devise a scheme that allows interrogation of the focus while illuminating a mirror with a collimated laser beam, fig. [fig : confocal_schematic]. this arrangement means that the intensity distribution that would be formed in an optical tweezing experiment @xcite, for example, can be faithfully interrogated. probing the focal plane of the parabolic mirrors requires illumination with collimated light co axial with the mirror axis (orthogonal to the substrate surface). the working distance of the objective lens used (olympus uplansapo 40x/0.9@xmath21) is only 180 @xmath0 m, which severely limits introducing further optical elements before the object. to circumvent these restrictions a non - polarizing beam splitter (npbs) was introduced between the objective lens and the scanning column of a commercial confocal microscope (olympus fv1000 scan head with olympus ix81 inverted microscope). this enabled an external probe beam to be introduced through the back aperture of the objective lens using the npbs to combine the beam into the imaging axis, see fig. [fig : confocal_schematic]. the probe, a commercial solid - state 589 nm laser (based on sum - frequency generation using two lasing lines of an nd : yag laser), was focused, after fiber - coupling to clean up the spatial mode, at the back - focal plane of the objective lens. the specific probe wavelength was chosen for convenience (good detector efficiency and high optical power in a single transverse mode) - the numerical simulations were also conducted for this wavelength to account for wavelength - dependent diffraction effects. the collimation of the beam emitted was optimized by varying the axial position of the coupling lens. the npbs, optical fiber, and coupling lens were mounted in a holder with three axis adjustment, which allowed for overlapping of the optical axes. the emitted beam had a diameter of 1 mm, justifying the assumption of a spatially uniform intensity profile across the micromirror aperture (20 @xmath0 m diameter). the magnification of the combined optical system was calibrated by measuring the aperture diameter of the mirrors. in order to visually verify the surface quality of the mirrors, as well as to measure the aperture diameter of the mirrors the surface of the mirrors was also probed using conventional confocal microscopy @xcite, without removal of the additional optical elements. no aberrations were observed to have been introduced.
Results
m parabolic micromirror. a) shows data collected from the surface of the mirror substrate ; b) taken 25 @xmath0 m before the focus (@xmath22 m in fig [fig : data_vs_simulation]), c) taken at the position of the focus, and d) taken 25 @xmath0 m after the focus (@xmath23 m in fig [fig : data_vs_simulation]). the black scale bars in a), b), d) indicate a length of 10 @xmath0 m, which is the width of image c). the linear colormap for each image is normalized from white to black for minimum to maximum value respectively.,width=321] optical measurements were taken for four mirrors with different focal lengths, all with 10 @xmath0 m radii apertures. two dimensional scans were taken, with 77 nm pixel size, of the reflected intensity field in defined planes parallel to the surface of the substrate, fig. [fig : raw_data]. a series of such scans, separated by 1 @xmath0 m, then return a high - resolution, 3d map of the intensity field. the raw data were then radially averaged about the optical axis of the beam. the optical axis was independently found by fitting a 2d gaussian to individual scans near the focal plane and then linearly extrapolating the center of mass fits to regions where a simple 2d - gaussian fit was not valid, _ e.g. _, such as in fig. [fig : raw_data](b). (note, however, that about the focus, fig. [fig : raw_data](c), a 2d - gaussian gives an excellent fit to the data.) this step was taken to ensure any residual tilts in the optical system were accounted for. the averaged data were then combined, as shown in fig. [fig : data_vs_simulation](a), to visualize the cylindrically symmetric intensity field. the focal length was extracted from the distance between the surface and the peak of the intensity, as well as by the best fit of the numerical model, fig. [fig : data_vs_simulation](b). these data show excellent qualitative agreement as the detailed structure of the field away from the focus, which is emphasized by the logarithmic scale on the color axis. to further illustrate the agreement between experiment and theory the on - axis intensity profile of the data and the simulation are shown in fig. [fig : data_vs_simulation](c). m mirror, using a log - scale to highlight the detail in the intensity field away from the focus. the data in each figure are each normalized, with the same color - axis for both. the color - axis, [-4 0], corresponds to a range of four orders of magnitude in intensity. the measured (solid blue line) and simulated (dashed line) on - axis intensity profiles, c), and radial intensity profiles at the focus, d), show agreement between experiment and model. also included in d), as a red dotted line, is a gaussian fit to the simulated data.,width=453] quantitative results from the measurements are in table [table], below. the measured waists given are the minimum e@xmath24 radii returned from gaussian fits about the waist. the waists in the orthogonal directions were found to be equal and to occur in the same place, within measurement error. it can be seen in fig. [fig : data_vs_simulation](d) that a gaussian is an excellent fit to both simulation and experimental data near the focus. the diffraction limit was calculated as @xmath25, where @xmath26 is the numerical aperture, @xmath27 which was derived using the measured focal length @xcite. it is clear that there is good agreement between the measured minimum beam waists and the prediction from diffraction theory. .measured beam waists (e@xmath24 radius) and focal lengths for a range of parabolic mirrors with 10 @xmath0 m radius aperture. errors, where quoted, are one standard deviation. [cols="^,^,^,^,^ ",]
Discussion
the intensity patterns observed in these data suggest that the these mirrors would be favorable for trapping of small clouds of ultra - cold atoms, or even single atoms @xcite. for a total power of 1 w in the illuminating beam we estimate a peak intensity at the focus of @xmath28 w / m@xmath29 ; with a dipole trapping wavelength of @xmath30 nm this corresponds to a trap depth of 370 @xmath0k for rubidium atoms @xcite. the trap depth would scale at least with the square of the aperture of the mirrors (the power captured scales as the aperture diameter squared, while the peak intensity increases with the numerical aperture) meaning that deeper traps could readily be designed. two dimensional arrays of single trapped atoms could be formed from one laser beam illuminating a congruent 2d array of mirrors. a feature of this array is that the illumination of individual mirrors would then provide single atom addressability, but on the length scale of the mirror aperture, which is inherently larger than the single atom focus. m. the blue dotted line corresponds to the position of peak intensity, the spot of arago, after an open aperture of diameter equal to that of the simulated mirror.,width=377] although we have shown, in fig. [fig : data_vs_simulation], that the angular spectrum method yields agreement with the measured data in the regime probed here, further simulations show that focusing power of parabolic micro - mirrors is limited at longer focal lengths. in fig. [fig : focal_vs_simulation] the position of peak intensity as a function of mirror design focal length shows that diffraction from the aperture edge begins to dominate at focal lengths significantly longer than the aperture radius. in the limit of large focal length the location of the peak intensity tends towards that predicted for diffraction from the edges ; this is shown in fig. [fig : focal_vs_simulation] by the dotted line which is found from simulating diffraction from an aperture of radius equal the that of the mirrors. for the shortest focal length measured in this work, @xmath31 m, we derive a numerical aperture of @xmath32 for operation in water. as shown in @xcite by merenda _, stable 3d trapping is to be expected in this regime. further work will be conducted to demonstrate optical tweezing using these mirrors.
Conclusions
construction and diffraction limited focusing from parabolic micromirrors with radii of @xmath33 m (@xmath34) has been demonstrated. simulations performed using the angular spectrum method show clear agreement with data and reproduce detailed features evident in the measured 3d intensity field. the results offer promise for use of these micromirrors for experiments in optical trapping in both biological and atomic - physics experiments.
Acknowledgments
the authors are extremely grateful to alastair sinclair for initial project design and for discussions, to jonathan pritchard for help with simulations, and to john harris for assistance. pfg acknowledges the generous support of the royal society of edinburgh. | we report on the fabrication and diffraction - limited characterization of parabolic focusing micromirrors. sub micron
beam waists are measured for mirrors with 10-@xmath0 m radius aperture and measured fixed focal lengths in the range from 24 @xmath0 m to @xmath1 m.
optical characterization of the 3d intensity in the near field produced when the device is illuminated with collimated light is performed using a modified confocal microscope.
results are compared directly with angular spectrum simulations, yielding strong agreement between experiment and theory, and identifying the competition between diffraction and focusing in the regime probed.
99 h. c. king, _ the history of the telescope _
(dover, 2003).
l. novotny, and b. hecht, _ principles of nano - optics _ (cambridge university, 2012).
w. r. jamroz, r. kruzelecky, and e. i. haddad, _ applied microphotonics _
(taylor & francis, 2006).
r. vlkel, m. eisner, and k.j.
weible, `` miniaturized imaging systems, '' microelectron
. eng. * 67 *, 461 - 472 (2003).
m. trupke, e. a. hinds, s. eriksson, e. a. curtis,.
moktadir, e. kukharenka, and m. kraft `` microfabricated high - finesse optical cavity with open access and small volume, '' appl.
phys. lett. *
87 *, 211106 (2005).
f. merenda, j rohner, j.-
m.
fournier, and r.-
p.
salath `` miniaturized high - na focusing - mirror multiple optical tweezers, '' opt.
express * 15 *, 6075 - 6086 (2007).
f. merenda, m. grossenbacher, s. jeney, l. forr, and r.-
p.
salath `` three - dimensional force measurements in optical tweezers formed with high - na micromirrors, '' opt.
lett. * 34 *, 1063 - 1065 (2009).. y. s. ow, m. b. h. breese, and sara azimi `` fabrication of concave silicon micro - mirrors, '' opt.
express * 18 *, 14511 - 14518 (2010).
g. w. biedermann, f. m. benito, k. m. fortier, d. l. stick, t. k. loyd, p. d. d. schwindt, c. y. nakakura, r. l. jarecki jr., and m. g. blain `` ultrasmooth microfabricated mirrors for quantum information, '' appl.
. lett. * 97 *, 181110 (2010).
g. shu, n. kurz, m. dietrich, and b. blinov, `` efficient fluorescence collection from trapped ions with an integrated spherical mirror, '' phys.
rev.
a * 81 *, 042321 (2010).
j. merrill, c. volin, d. landgren, j. amini, k. wright, s. doret, c. pai, h. hayden, t. killian, d. faircloth, k. r. brown, a. w. harter, and r. e. slusher, `` demonstration of integrated microscale optics in surface - electrode ion traps, '' new j. phys. * 13 *, 103005 (2011).
r, maiwald, a, golla, m, fischer, m, bader, s, heugel, b, chalopin, m. sondermann, and g. leuchs `` collecting more than half the fluorescence photons from a single ion, '' phys.
. lett. * 86 * 043431 (2012).
m. fischer, m. bader, r, maiwald, a, golla, m. sondermann, and g. leuchs, `` efficient saturation of an ion in free space, '' appl.
phys.
b - lasers o. * 117 *, 797 - 801 (2014).
a. roy, a.b.
shen jing, and m.d barrett, `` the trapping and detection of single atoms using a spherical mirror, '' new j. phys.
* 14 *, 093007 (2012).
r. noek, c. knoernschild, j. migacz, t. kim, p. maunz, t. merrill, h. hayden, c. s. pai, and j. kim, `` multiscale optics for enhanced light collection from a point source, '' opt. lett. * 35 *, 2460 - 2462 (2010).
m. k. tey, z. chen, s. a. aljunid, b. chng, f. huber, g. maslennikov, and c. kurtsiefer, `` strong interaction between light and a single trapped atom without the need for a cavity, '' nat. phys.
* 4 *, 924 - 927 (2008).
m.t.
langridge, d.c.
cox, r.p.
webb, and v. stolojan, `` the fabrication of aspherical microlenses using focused ion - beam techniques, '' micron * 57 *, 56 - 66 (2014).
j. goldwin, and e. a. hinds, `` tight focusing of plane waves from micro - fabricated spherical mirrors, '' opt.
express * 16 *, 17808 - 17816 (2008).
t. n. bandi, v. g. minogin,, and s. nic chormaic, `` atom microtraps based on near - field fresnel diffraction, '' phys. rev. a * 78 *, 013410 (2008).
j. goodman, _ introduction to fourier optics _ (roberts, 2005).
e. gu, h. w. choi, c. liu, c. griffin, j. m. girkin, i. m. watson, m. d. dawson, g. mcconnell, and a. m. gurney, `` reflection - transmission confocal microscopy characterization of single - crystal diamond microlens arrays, '' appl.
. lett. * 84 *, 2754 - 2756 (2004).
a. ashkin, `` forces of a single - beam gradient laser trap on a dielectric sphere in the ray optics regime, '' biophys.
j. * 61 *, 569 - 582 (1992).
n. schlosser, g. reymond, i. protsenko, and p. grangier, `` sub - poissonian loading of single atoms in a microscopic dipole trap, '' nature * 411 *, 1024 - 1027 (2001).
r. grimm, m. weidemller, and y. b. ovhinnikov, `` optical dipole traps for neutral atoms, '' adv.
atom mol.
opt. phys. *
42 *, 95 - 170 (2000). | 1411.2851 |
Introduction
a natural and very general extension of the einstein - maxwell lagrangian yielding a general system of equations for a non - minimal coupling between the gravitational and electromagnetic fields, with non - linear terms, was set up and studied in @xcite. within this general theory, a special theory, worth of discussion, arises when one restricts the general lagrangian to a lagrangian that is einstein - hilbert in the gravity term, quadratic in the maxwell tensor, and the couplings between the electromagnetism and the metric are linear in the curvature terms. the motivations for setting up such a theory are phenomenological, see, e.g., @xcite for reviews and references.this theory has three coupling constants @xmath0, @xmath1 and @xmath2, which characterize the cross - terms in the lagrangian between the maxwell field @xmath10 and terms linear in the ricci scalar @xmath11, ricci tensor @xmath12, and riemann tensor @xmath13, respectively. the coupling constants @xmath0, @xmath1 and @xmath2 have units of area, and are a priori free parameters, which can acquire specific values in certain effective field theories. more specifically, the action functional of the non - minimal theory linear in the curvature is @xmath14 where @xmath15 is the determinant of the spacetime metric @xmath16, and the lagrangian of the theory is @xmath17 here @xmath18, @xmath19 being the gravitational constant and we are putting the velocity of light @xmath20 equal to one, @xmath21 is the maxwell tensor, with @xmath22 being the electromagnetic vector potential, and latin indexes are spacetime indexes, running from 0 to 3. the tensor @xmath23 is the non - minimal susceptibility tensor given by @xmath24 where @xmath0, @xmath1, and @xmath2 are the mentioned phenomenological parameters. the action and lagrangian ([action1])-([susceptibility2]) describe thus a three - parameter class of models, non - minimally coupled, and linear in the curvature @xcite. lagrangians of this type have been used and studied by several authors. the first and important example of a calculation of the three couplings was based on one - loop corrections to quantum electrodynamics in curved spacetime, a direct and non - phenomenological approach considered by drummond and hathrell @xcite. this model is effectively one - parameter since the coupling constants are connected by the relations @xmath25, @xmath26, @xmath27. the positive parameter @xmath3 appears naturally in the theory, and is constructed by using the fine structure constant @xmath28, and the compton wavelength of the electron @xmath29, @xmath30. in these models it is useful to define a radius @xmath31, an effective radius related to the non - minimal interaction, through @xmath32. thus, the corresponding effective radius for the non - minimal interaction in this case, is the drummond - hathrell radius @xmath33, given by @xmath34. in @xcite one also finds a quantum electrodynamics motivation for the use of generalized einstein - maxwell equations. phenomenological models, i.e., models based on external considerations to obtain the couplings, or parameters, @xmath0, @xmath1, and @xmath2, have also been considered. prasanna @xcite wanting to understand how the strong equivalence principle can be weakly violated in the context of a non - minimal modification of maxwell electrodynamics, has shown that @xmath35, @xmath3 a free parameter, is a good phenomenologically model. another type of requirement, one with mathematical and physical motivations, is to impose that the differential equations forming the non - minimal einstein - maxwell system are of second order (see, e.g., @xcite). for instance, in @xcite, by imposing a kaluza - klein reduction to four dimensions from a gauss - bonnet model in five dimensions, thus guaranteeing second order equations for the electric field potential @xmath36, and metric @xmath16, it was discussed a model in which @xmath37 and @xmath38, i.e., with @xmath5, @xmath6 and @xmath7. so the extra non - minimal term is a kind of gauss - bonnet term, and the model is called the gauss - bonnet model. yet another type of requirement, this time purely mathematical, was suggested in @xcite. the idea is connected with the symmetries of the non - minimal susceptibility tensor @xmath23 (see eq. ([susceptibility2])). for instance, one can recover the relations @xmath37 and @xmath38, used in @xcite, by the ansatz that the non - minimal susceptibility tensor @xmath39 is proportional to the double dual riemann tensor @xmath40, i.e., @xmath41, for some @xmath42 (see, @xcite for details and motivations). analogously, one can use the weyl tensor @xmath43 in the relation @xmath44, for some @xmath45, or the difference @xmath46 instead of @xmath40, to introduce some new linear relations between @xmath47, namely @xmath48 and @xmath49. yet another type of requirement is to choose the parameters so that one obtains exact solutions. as we will see this will lead to a model with @xmath37 and @xmath9, i.e., @xmath5, @xmath8, @xmath9. since this model is integrable we call it the integrable model. a subcase of this has additional interest and is called the fibonacci soliton. up to now we have a theory defined through eqs. ([action1])-([susceptibility2]), with each chosen set of values for the parameters @xmath0, @xmath1, and @xmath2, giving a model. we have seen that the reduction from three - parameter models to one - parameter models, specified by the one parameter @xmath3 and the relations between @xmath0, @xmath1 and @xmath2, happens in several instances, either through direct calculation, as in @xcite, or through phenomenological and other considerations, as in @xcite-@xcite or @xcite and here. this certainly simplifies the analysis, and we will consider this one - parameter type of models, in which @xmath0, @xmath1, and @xmath2, have a specified relation to the parameter @xmath3. for all these models, one can pick an effective radius @xmath50, as in the drummond - hathrell case, which gives the range of the non - minimal interaction between the gravitational and electric fields. of course, @xmath31 can be set to zero, in the case the world is pure einstein - maxwell, or otherwise can have a given specified value. the radius @xmath33 defined above is a candidate but in principle not the unique choice. thus, possible estimations of the parameter @xmath3, and so of @xmath31, from, for instance, astrophysical observations, are undoubtedly of interest (see, e.g., @xcite). now, after choosing a model, specified by @xmath3 and by the relations between @xmath0, @xmath1, and @xmath2, it is important to study exact solutions. exact solutions of the equations of non - minimal electrodynamics in non - linear gravitational wave backgrounds were obtained in @xcite@xcite, and a non - minimal bianchi - i cosmological solution was discussed in @xcite in this context. here we want to study charged black hole and other charged solutions of non - minimal models. the reissner - nordstrm solution is a standard solution in pure einstein - maxwell theory, with two horizons, an event and a cauchy horizon, and a timelike singularity at the center (see, e.g., @xcite). paradigmatic charged black hole solutions also appear in the framework of einstein theory minimally coupled to non - linear electromagnetic fields, as well as other matter fields. such solutions were found by bardeen and others @xcite-@xcite and the main feature is that they are regular, without singularities inside the horizon. within quartic gravity non - singular charged black hole solutions have also been found @xcite. since the non - minimal theory we are considering possesses new degrees of freedom, namely, the phenomenological parameter @xmath3 and its relations to @xmath0, @xmath1, and @xmath2, we believe that these allow to introduce new aspects to the problem of finding black hole and other solutions of each chosen model. three aspects can be mentioned. first, one wants to have a gauge in order to compare the new solutions. thus, we study the reissner - nordstrm solution, the trivial solution in this context, where @xmath51 with @xmath52, in order to understand the novel features, such as causal and singularity structure, of the new solutions. second, one should try to search for models exactly or quasi - exactly soluble. this requirement will take us, among the models cited above and the many other possible models, to two interesting models. they are, the gauss - bonnet type model where @xmath5, @xmath6 and @xmath7, and the integrable model where @xmath5, @xmath8, @xmath9. in both models we perform a detailed analysis. third, we vary @xmath3 within each of the two nontrivial models and in trying to consider non - minimal extensions of the reissner - nordstrm solution, we search for features that are similar or distinct from the two paradigmatic solutions, the reissner - nordstrm solution itself and the bardeen solutions. in our search for charged black hole solutions in non - minimal models we work in schwarzschild coordinates and impose certain requirements. the first requirement is connected with the electric field @xmath53. we demand it is a regular function on the interval @xmath54, the origin being also regular (i.e., the value @xmath55 is finite), and for large values of @xmath56 the electric field is coulombian, @xmath57, where @xmath58 is the electric charge. the second requirement is concerned with the metric functions @xmath16. these should take finite values at the center (@xmath59). horizons at @xmath60 are not excluded, and far away the solutions should be asymptotically flat. upon these conditions and solving the non - minimal equations in certain cases we will find electric charged solutions with one horizon only, thus causally distinct from the reissner - nordstrm. however, like reissner - nordstrm, the solutions have a singularity at the center, although here the singularity is spacelike instead, as the schwarzschild case, but conical, thus much milder. we have also found in one model a gravitational charged soliton, without horizons, where the fields are well behaved, apart from a mild conical singularity at the center. although this and other solutions with horizons are almost regular at the center, we have not obtained strictly non - singular black hole of the type found by bardeen and others @xcite-@xcite. the difference is based on two aspects. firstly, we consider spherically symmetric static solutions with @xmath61, in contrast to the schwarzschild, reissner - nordstrm, and minimal regular bardeen solutions. secondly, we assume that the values @xmath62 and @xmath63 are finite, but can differ from one. moreover, we admit, that @xmath62 can, in principle, be equal to zero. this means that the scalar curvature invariants for such a metric can take infinite values at the center, and the solution of the non - minimal einstein equations is not regular at the center in this general sense. however, in many cases the singularity is a conical one, so much milder than the nasty ones of schwarzschild and reissner - nordstrm. in summary, we find charged black hole solutions different in horizon structure from the reissner - nordstrm but similar to bardeen black holes, and although singular, in certain cases can be considered quasi - regular conical singularities, in - between schwarzschild and reissner - nordstrm types of singularities and the no singularities of bardeen. this paper is organized as follows. in section [genericalities], in particular in subsection [genericalities2], using the lagrangian formalism of the introduction, we establish a three - parameter non - minimal einstein - maxwell model. in subection [static], we set up static equations for studying black holes and reduce the three - parameter model to a one - parameter model. in section [solutions] we study specific spherically symmetric one - parameter solutions. in subsection [basics] we define the basic quantitities and basic variables. in subsection [rnmodel] we display the reissner - nordstrm solution as a preparation, where @xmath51 with @xmath52. in subsection [gbmodel] we analyze in detail a one - parameter model, the gauss - bonnet model, with @xmath5, @xmath6, @xmath7 (i.e., @xmath37 and @xmath38), using the known solution of the abel equation, the key equation of the model, and the dynamical system associated with this model. we display the charged black hole solutions, and we focus on a specific exact solution and its critical properties. in subsection [nonamemodel] we consider in detail an exactly integrable one - parameter model, the integrable model, with @xmath5, @xmath8, @xmath9 (i.e., @xmath37 and @xmath9). we examine a special sub - model, the fibonacci soliton, of this one - parameter model and present the corresponding exact solution. in subsection [table] we present, by means of a table, the summary of the results of the models studied. in section [conc] we conclude.
Non-minimal coupling, linear in the curvature, between gravity and electromagnetism: equations and reduction from three parameters to one parameter for static spherically symmetric systems
the variation of the lagrangian ([action1])-([susceptibility2]) with respect to the metric yields (see, @xcite for details), @xmath64 \,. \label{standardform}\]] the energy - momentum tensor of the pure electromagnetic field, @xmath65, is @xmath66 the definitions for the other three parts of the stress - energy tensor, @xmath67, @xmath68 and @xmath69, are @xmath70 @xmath71 - f^{ln}(r_{il}f_{kn } + r_{kl}f_{in }) - \nonumber\\ & & { - } r^{mn } f_{im } f_{kn } { - } \frac{1}{2 } \nabla^l \nabla_l (f_{in}f_{k}^{\ n }) { + } \frac{1}{2}\nabla_l \left [\nabla_i(f_{kn}f^{ln }) { + } \nabla_k(f_{in}f^{ln }) \right] \,, \label{part2}\end{aligned}\]] @xmath72 in addition, the non - minimal electrodynamics associated with the lagrangian ([action1])-([susceptibility2]) obeys the equation @xmath73 = 0 \,, \quad \nabla_k f^{*ik } = 0 \,, \label{m1}\]] where @xmath74 is the maxwell tensor and @xmath75 is dual to it. consider now these master equations in the case of a static spherically symmetric spacetime. using schwarzschild coordinates, the line element for a static spherically symmetric system can be put in the form @xmath76 where @xmath77 and @xmath78 are metric potentials that depend on the radial coordinate @xmath56 only. this form of the line element is useful as when @xmath77 and @xmath79 are simultaneously zero it signals the presence of an event horizon. assume also that the electromagnetic field inherits the static and spherical symmetries. then the potential four - vector of the electric field @xmath36 has the form @xmath80 from ([potentiala]) the maxwell tensor is equal to @xmath81, where a prime denotes the derivative with respect to @xmath56. to characterize the electric field, it is useful to introduce a new scalar quantity @xmath53 as @xmath82. then the electric field squared is @xmath83 from which one obtains in turn @xmath84. since the expressions @xmath79 and @xmath85 enter frequently in the master equations, it is sometimes convenient to use the functions @xmath86 and @xmath87 defined as @xmath88 and @xmath89. in summary, the functions @xmath90 and @xmath91 are alternatives to the functions @xmath92, @xmath93, and @xmath94. the maxwell equations ([m1]) with ([potentiala]) give only one non - trivial equation, namely, @xmath95^{\prime } = 0 \,, \label{eequ}\]] which can be integrated immediately to give @xmath96 \right.\]] @xmath97 where @xmath58 is a constant, to be associated with the central electrical charge of the solution. this equation gives the electric field of a central charge, corrected by the radial component of the dielectric permeability tensor, @xmath98. this component describes the vacuum screening effect on the central charge, due to the interaction of the vacuum with curvature, analogously to the screening of a charge by a non - homogeneous medium in a spherical cavity. supposing the spacetime to be asymptotically flat, i.e., @xmath99, one can see that ([1e]) yields asymptotically the coulomb law @xmath100, and the constant @xmath58 indeed coincides with the total electric charge of the object. the equations for the gravitational field ([standardform]) with ([t0])-([part3]) and the metric potentials ([metric1]) redefined as in ([sigman]) can be rewritten as a pair of equations for @xmath86 and @xmath87, respectively, @xmath101^{\prime}}{\kappa r^2 } = - \left (e^2\right)^{\prime \prime } n (q_1+q_2+q_3)\nonumber\\ + \left (e^2\right)^{\prime } \left[- \frac{1}{2}(q_1+q_2+q_3)\left (n^{\prime } + \frac{8n}{r}\right)+ \frac{n}{r}(2q_1+q_2) \right]\nonumber\\ + e^2 \left [\frac{1}{2 } + (q_1+q_2+q_3)\left (n^{\prime \prime } + 3 n^{\prime } \ \frac{\sigma^{\prime}}{\sigma } + 2n \frac{\sigma^{\prime \prime}}{\sigma } - \frac{n^{\prime}}{r } - 2 \frac{n}{r^2 } \right) \right.\nonumber\\ \left. + (2q_1+q_2)\left (2\frac{n^{\prime}}{r } + 2\frac{n}{r } \frac{\sigma^{\prime}}{\sigma } + \frac{n}{r^2 } \right) + q_1 \frac{(n-1)}{r^2 } \right]\,, \label{2e}\end{aligned}\]] @xmath102\nonumber\\ + e^2 \left[(q_1+q_2+q_3) \frac{2\sigma^{\prime } } { r \sigma } - \frac{2q_3}{r^2 } \right] \,.\label{3e}\end{aligned}\]] the first equation can be reduced to an equation for @xmath53 and @xmath86, by extracting the term @xmath103 from the second one. the second equation contains the unknown functions @xmath53 and @xmath87 only. thus, eqs. ([1e])-([3e]) form the key system of equations for the non - minimal einstein - maxwell model of a static spherically symmetric object. it is a system of three ordinary differential equations of second order for the three unknown functions, @xmath53, @xmath86 and @xmath87. the form of equations is not canonical. in principle, the electric field @xmath53 can be extracted explicitly from ([1e]) as a function of @xmath104, @xmath105, @xmath106, @xmath107, @xmath103 and @xmath56. inserting such @xmath53 into the eqs. ([2e])-([3e]), we obtain equations for @xmath86 and @xmath87 of fourth order in their derivatives. below we focus on models admitting solutions to eqs. ([1e])-([3e]), such that they can be represented by a series expansion regular at @xmath108, i.e., @xmath109 where @xmath110 symbolizes generically the functions @xmath53, @xmath86, and @xmath87. our purpose is to find solutions satisfying three conditions : first, the electric field @xmath53 should be a continuous function regular at @xmath108 (@xmath111) and also should be of coulombian form at @xmath112. second, the metric functions @xmath86 and @xmath113 should be regular at @xmath108. third, in terms of the functions @xmath93 and @xmath94 the asymptotic flatness requires that @xmath114, and @xmath115, @xmath116. so, essentially, one can put, @xmath117 and @xmath118. note that the regularity of the functions @xmath53, @xmath86, and @xmath87 at @xmath108 does not guarantee that the solution of the einstein - maxwell model is characterized by regular curvature invariants. for instance, when @xmath119 is finite but @xmath120, the model displays a conical singularity and the scalar invariants of the curvature tend to infinity as @xmath121. in considering solutions such that the fields are finite we try to be as close as possible to bardeen s idea of having black hole solutions without singularities, by finding a regular @xmath53 and putting the metric coefficients in the form @xmath122 and @xmath123, for some @xmath124 @xcite. as we will see it will turn out that this is not achieved, since although the electric and metric potentials are regular, the black hole solutions found here are singular at the center, where the curvature invariants blow up. notwithstanding, these solutions are very interesting. using the ansatz ([regular]) and the eqs. ([1e])-([3e]) one can couple the values @xmath55, @xmath119, @xmath125, and @xmath0, @xmath1, @xmath2. the relations are different for the cases @xmath126 and @xmath127, which we now analyze. _ (i) @xmath126 : _ when @xmath126, but @xmath128, one obtains from the system ([1e])-([3e]) that @xmath129, and the decomposition ([regular]) is valid at @xmath121, when the following conditions are satisfied, @xmath130= 2q_1 + q_2\,, \label{regularity2}\end{aligned}\]] for generic @xmath0, @xmath1, and @xmath2. there are two specific cases. when @xmath37, but both @xmath131 and @xmath132, then @xmath133 and @xmath134. when @xmath37 and @xmath135, simultaneously, then @xmath136, providing @xmath0 is negative, and @xmath119 is fixed by the relation @xmath137. as in the case @xmath127, see below, here the ricci scalar @xmath138 takes an infinite value at @xmath108. _ (ii) @xmath127 : _ when all three functions, @xmath53, @xmath86, and @xmath87 are regular at @xmath108, and @xmath87, appearing in the denominator of eqs. ([1e])-([3e]), does not vanish at @xmath108, one obtains from the system ([1e])-([3e]) the following set of equations @xmath139 = q \,, \quad e^2(0) \ 2 q_3 = 0 \,, \nonumber\\ n(0) \left [1 + \kappa e^2(0) (q_1 - q_2 - 2 q_3) \right] = 1 + q_1 \kappa e^2(0) \,, \label{regularity1}\end{aligned}\]] for generic @xmath0, @xmath1, and @xmath2. since the charge of the object, @xmath58, is considered to be non - vanishing, one obtains immediately from the first equation of the set that @xmath140 and @xmath120. thus, we infer, @xmath141 } \,, \quad q_2 = \frac{2 [n(0)-1] + \kappa e(0) q}{2 n(0) \kappa e^2(0) } \,, \quad q_3=0\,. \label{r2}\]] the relations ([regularity1]) give that the curvature invariants are infinite in the center @xmath108. for instance, when @xmath86 and @xmath87 are regular in the center and @xmath142, then the ricci scalar @xmath143 tends to infinity at @xmath121, since @xmath120, as well as generally @xmath144. now we want analyze the simplest cases of the system of eqs. ([1e])-([3e]). one sees that there is an immediate simplification when @xmath37, since second order derivatives and products of first order derivatives disappear from the equations. in such a case the system ([1e])-([3e]) reduces to @xmath145 @xmath146 @xmath101^{\prime}}{\kappa } = r \left (e^2\right)^{\prime } (2q_1+q_2) n + \nonumber\\ + e^2 \left [\frac{r^2}{2 } - q_1 + 2 r (2q_1+q_2)\left (n^{\prime } + n \frac{\sigma^{\prime}}{\sigma } \right) + n (3q_1 + q_2) \right]\,. \label{13e}\end{aligned}\]] moreover, the three - dimensional matrix, composed of the coefficients in front of the first derivatives @xmath147, @xmath105 and @xmath148, has rank two. this means that for @xmath37 the system ([11e])-([13e]) can be reduced to one algebraic equation and two differential equations of the first order. the corresponding algebraic equation is @xmath149 e - q = 0 \,, \label{cubic}\]] which in turn links two functions, @xmath53 and @xmath86. now from equation ([cubic]), one can consider three subcases, that emerge from the case @xmath37. first we consider briefly the trivial case in this context, @xmath150 i.e., the reissner - nordstrm limit, see subsection [rnmodel]. then we consider two interesting non - trivial cases : first, @xmath151, second, @xmath152, when @xmath151 it is easy to express @xmath86 in terms of @xmath53. in the subsection [gbmodel] we consider a specific model in this class, characterized by the supplementary condition @xmath153 this is the gauss - bonnet type model, which has been considered as an important model in @xcite, and for which we present an extended analysis with relevant new details. for the second case @xmath154 @xmath53 decouples from @xmath86 and we deal with a cubic equation for the determination of the electric field. we will consider such a model, the integrable model, in subsection [nonamemodel].
Solutions of the reduced three-parameter model: solutions of one-parameter models
one should first define three quantities, @xmath155, @xmath156, @xmath157, as follows @xmath158 now, the models we are going to discuss here are essentially one parametric, with @xmath0, @xmath1, and @xmath2 being a multiple of some parameter @xmath3. it is useful to introduce first a quantity @xmath31, given through @xmath159 with @xmath31 being a radius. from @xmath155, @xmath156, and @xmath31, one can then construct two independent dimensionless quantities, namely @xmath160 and @xmath161 the @xmath162 quantity gives the deviation from the standard reissner - nordstrm case, and @xmath163 fixes the ratio between the total mass to the total charge of the object. in addition, for what follows below, it is useful to write the equations of motion by defining two dimensionless variables, a normalized radius @xmath164 and a normalized electric field @xmath165, defined as follows @xmath166 also, the function @xmath167 can be defined in terms of another useful function @xmath168, i.e., @xmath169 the physical interpretation of @xmath168 is connected with the effective mass of the object, i.e., @xmath170, as we will see below. with these quantities defined we now discuss the reissner - nordstrm limit and the two new models. when the non - minimal parameters @xmath0, @xmath1, and @xmath2 are set to zero, i.e., @xmath171 and so @xmath172 as well, we can integrate immediately eqs. ([1e])-([3e]). in terms of the above functions, the system of key equations can be rewritten as @xmath173 @xmath174 @xmath175 the solutions to these equations are @xmath176 @xmath177 @xmath178 the function @xmath168 in ([nrn]) is here given by @xmath179 where @xmath163 is defined in ([k0]), and one can see through eq. ([2eqdimrn]) that @xmath180 obeys the equation @xmath181 of course one can transform to the original fields @xmath53, @xmath93, and @xmath94, giving @xmath182 @xmath183 @xmath184 which are the usual reissner - nordstrm functions. note, then, that the physical interpretation of @xmath168 given in ([yrn]) is connected with the effective mass of the object, i.e., @xmath170, which in turn is given by the definition @xmath185 thus, @xmath168 is related to the dimensionless effective mass, since @xmath186, with @xmath187, @xmath188 being the asymptotic mass of the object. we study in this context the usual static spacetimes with @xmath172, i.e., the schwarzschild and reissner - nordstrm spacetimes, which are special solutions of the full system of equations. these two solutions are heavily singular, both the metric and the kretschmann scalar diverge at @xmath108. so these solutions are outside the spirit of the solutions we want to find. they do not serve as models. however, they are of interest to set the nomenclature, and to have a gauge with which we can compare the solutions we find in the two models studied below. note that for these solutions @xmath189. in these cases the problem of searching for horizons is equivalent to finding their radial position @xmath190 through the solutions of @xmath191, where @xmath170 is the effective mass. in terms of dimensionless mass @xmath168, ([nyxrn]), or eq. ([yrn]), this condition can be written as @xmath192. for the schwarzschild metric @xmath193, where @xmath194 is defined in ([k0]). then, one obtains only one horizon at @xmath195, which is just the schwarzschild radius, @xmath196. for the reissner - nordstrm metric one has @xmath197, and the equation @xmath198 gives three different cases : (i) @xmath199 (i.e., @xmath200, or @xmath201 in the standard notation) : there are two solutions @xmath202, and @xmath203, corresponding to the outer and inner horizons, respectively, of a usual reissner - nordstrm black hole. (ii) @xmath204 (i.e., @xmath205, or @xmath206) : there is one horizon only, given by @xmath207, corresponding to an extremal black hole. (iii) @xmath208 (i.e., @xmath209, or @xmath210) : there is no solution to the equation @xmath192. the object is a naked singularity. consider now, in eqs. ([action1])-([susceptibility2]), the following specific one - parameter model @xmath211 for some parameter @xmath3. in this model the susceptibility tensor is proportional to the double - dual riemann tensor and is divergence - free @xcite, i.e., @xmath212, and @xmath213 moreover, the coupled einstein and electromagnetic equations are second order in the derivatives, which is the reason why this model is called a gauss - bonnet model. gauss - bonnet gravity in five and higher dimensions has the property that its equations are of second order, lovelock gravity being a generalization of it. actually, one can show that the model specified by ([modeladopted]) comes from kaluza - klein reduction to four dimensions from five a dimensional gauss - bonnet theory, i.e., einstein gravity plus a gauss - bonnet term @xcite. using ([modeladopted]), eqs. ([11e])-([13e]) convert, respectively, into @xmath214 @xmath215 @xmath216^{\prime } = \frac{1}{2}\kappa e^2 \left [r^2 + 2 q(1 - n) \right]\,. \label{13e1}\]] after appropriate redefinitions these equations agree with the ones discussed in @xcite. the correspondingly modified algebraic eq. ([cubic]) coincides with ([11e1]). we now consider two ways of analyzing this system of equations : first, we use a power series expansion, and second, we apply the formalism of dynamical systems. \(i) the abel equation 0.1 cm using from eq. ([11e1]) that @xmath217\,, \label{equane}\]] one obtains the abel equation (see, e.g., @xcite) for @xmath53 from ([13e1]), @xmath218 similarly, eliminating @xmath53 from eq. ([11e1]) one can transform eq. ([13e1]) into the abel equation for a function @xmath219, here defined as @xmath220 thus, eq. ([13e1]) is given by @xmath221 \left [r^2 + 2q \theta(r) \right] = \frac{\kappa q^2}{2 } \,. \label{equat}\]] clearly, the values @xmath55 and @xmath119 are related through @xmath222. searching for solutions @xmath86 regular at @xmath108, we have to consider @xmath55 to be non - vanishing, @xmath140. in such a case eq. ([equae]) yields @xmath223. this means that @xmath3 has to be positive in these solutions. thus, although we analyze all cases, we tend to focus in models with @xmath224. finally, a possible solution, regular at @xmath108, should be characterized by @xmath225. it is convenient to use the auxiliary quantities @xmath226, @xmath227, and @xmath162 used before (see eqs. ([rqpropersaid])-([aratio])), to write @xmath228 using eqs. ([11e1])-([13e1]), plus the asymptotic conditions @xmath117, @xmath118, as well as the condition that the electric field is asymptotically coulombian @xmath100, one can obtain the following formula for the asymptotic mass @xmath188, @xmath229 \right\ } = \lim_{r\to\infty } r(1-n) \,. \label{equam2}\]] 2.7 cm (ii) the solution of the abel equation for small @xmath56 0.1 cm in the vicinity of the point @xmath108 the solutions for @xmath53, @xmath86 and @xmath87 are assumed to have a polynomial form of the type given in eq. ([regular]). the decomposition of a regular solution @xmath53 with non - vanishing @xmath55 is @xmath230 \,, \label{small1}\]] and the corresponding @xmath86 with finite @xmath119 is given by @xmath231 for this solution the effective mass @xmath170 becomes equal to zero at @xmath108. moreover, @xmath232, when electric and non - minimal radii coincide, i.e., @xmath233, @xmath234. 0.4 cm (iii) power series expansion with respect to @xmath235 0.1 cm the decomposition of the electric field yields @xmath236 \,, \label{18}\]] where the @xmath237 are defined below. infinity is a regular point for @xmath86, thus, taking into account ([equam2]) one obtains the following decomposition of @xmath86 @xmath238 where again, the @xmath237 are given below. the function @xmath239 is equal to one in the schwarzschild and the reissner - nordstrm cases, but not in general. when @xmath240 the logarithm of this function can be represented by the decomposition @xmath241\,. \label{expansionofab}\end{aligned}\]] the @xmath237 coefficients can be taken from eq. ([17]), and starting from @xmath242 can be found by the recurrence formula @xmath243 with @xmath244 the decompositions ([18])-([expansionofab]) are regular at @xmath245 and absolutely converge in the interval @xmath246, where @xmath247. note that the terms @xmath248 and @xmath249 are the schwarzschild and reissner - nordstrm terms, respectively, and that the @xmath237 for @xmath250 are the terms that give the post reissner - nordstrm behavior. note also that for @xmath52, i.e., the reissner - nordstrm case (or the schwarzschild case when, further, @xmath251), the function @xmath252 in ([expansionofab]) is equal to one, as it should. for @xmath253, eqs. ([18])-([expansionofab]) give useful asymptotic formulas, showing that, for @xmath53 and @xmath86, the first post - reissner - nordstrm terms are of fifth order in @xmath254, and that the decomposition for @xmath255 starts with a term of fourth order. when necessary one should convert from @xmath106 and @xmath252 to @xmath78 and @xmath77. numerical calculations, see figs. [k=2sqrt2gaussbonnetmodel]-[k=1gaussbonnetmodel], confirm that the corresponding curves tend to the corresponding horizontal asymptotes, when @xmath56 goes to infinity, i.e., @xmath53 tends to zero, and @xmath256 and @xmath94 tend to 1. it follows from ([18]) that, for positive @xmath3, @xmath257, the curvature coupling effects on the electric field are analogous to a dielectric medium, since the asymptotic electric field effectively decreases. for negative @xmath3, @xmath258, there are no solutions with @xmath86 regular at @xmath108, and, since we are mostly interested in regular or quasi - regular solutions, we do not fully discuss this case. , @xmath6 and @xmath7, of gravitational electrically charged objects characterized by @xmath259, for @xmath260 (@xmath199) plots (a), (b), (c), and (d) depict the metric potentials @xmath261 and @xmath94, the function @xmath262, and the electric field @xmath263, respectively, as functions of @xmath264, for solutions with different values of the non - minimal quantity @xmath162. the reissner - nordstrm black hole has @xmath172, the curves of which are clearly shown in the plots. in this case the curves for @xmath261 and @xmath94 have two zeros representing the inner and outer horizons, and for @xmath253 they go to one, while the curve @xmath263 tends to zero, respectively, and when @xmath265 these curves tend to @xmath266 (the electric field is in a logarithmic scale). for @xmath267 and @xmath268 the black holes behave quite similarly as the case @xmath172, with two horizons, and the function @xmath263 tends to finite values as @xmath265. for @xmath269 the curve for @xmath261 tends to a finite negative value when @xmath121, and takes the value zero only once. on the other hand @xmath94 has two zeros one at the same point as @xmath261, the event horizon, and the other at @xmath108, signaling the presence of a singularity there. the function @xmath263 tends to finite values as @xmath265. for @xmath270 the curves @xmath256 and @xmath94 have one zero, and thus one horizon only, at the same @xmath56, and then tend to infinity as @xmath265, an analogous behavior to the schwarzschild black hole. the function @xmath271 tends to finite values as @xmath265. see text for more details.,width=648,height=480] , @xmath6 and @xmath7, of gravitational electrically charged objects characterized by @xmath259, for @xmath204 plots (a), (b), (c), and (d) depict the metric potentials @xmath261 and @xmath94, the function @xmath262, and the electric field @xmath263, respectively, as functions of @xmath264, for solutions with different values of the non - minimal quantity @xmath162. the extremal reissner - nordstrm black hole has @xmath172, the curves of which are clearly shown in the plots. in this case the curves for @xmath261 and @xmath94 have one double zero representing an extremal horizon, and for @xmath253 they go to one, while the curve @xmath263 tends to zero, respectively, and when @xmath265 these curves tend to @xmath266 (the electric field is in a logarithmic scale). for @xmath267 and @xmath268 the black holes behave quite similarly as the case @xmath172, with one horizon, and the function @xmath263 tends to finite values as @xmath265. for @xmath269 the curve for @xmath261 tends to a finite negative value when @xmath121, and takes the value zero only once. @xmath94 has two zeros one at the same point as @xmath261, signaling there is only one horizon, and the other at @xmath108, signaling the presence of a singularity there. the function @xmath263 tends to finite values as @xmath265. for @xmath270 the curves @xmath256 and @xmath94 have at the same @xmath56, one zero, and thus one horizon only, and then tend to infinity as @xmath265, an analogous behavior to the schwarzschild black hole. the function @xmath263 tends to finite values as @xmath265. see text for more details.,width=648,height=475] , @xmath6 and @xmath7, of gravitational electrically charged objects characterized by @xmath259, for @xmath272 (@xmath208) plots (a), (b), (c), and (d) depict the metric potentials @xmath261 and @xmath94, the function @xmath262, and the electric field @xmath263, respectively, as functions of @xmath264, for solutions with different values of the non - minimal quantity @xmath162. the reissner - nordstrm naked singularity has @xmath172, the curves of which are clearly shown in the plots. in this case the curves for @xmath261 and @xmath94 have no zeros, and for @xmath253 they go to one, while the curve @xmath263 tends to zero, respectively, and when @xmath265 these curves tend to @xmath266 (the electric field is in a logarithmic scale). for @xmath267 and @xmath268 the naked singularity behaves quite similarly as the case @xmath172, and the function and @xmath263 tends to finite values as @xmath265. for @xmath269 the curve for @xmath261 tends to a finite positive value as @xmath121, while @xmath94 has a zero at @xmath108, signaling the presence of a singularity there. the function @xmath263 tends to finite values as @xmath265. for @xmath270 the curves @xmath256 and @xmath94 have one zero at the same @xmath56, and thus one horizon only, and then tend to infinity as @xmath265, an analogous behavior to the schwarzschild black hole. thus, by tuning the non - minimal quantity @xmath162 one can turn a reissner - nordstrm naked singularity, which has @xmath172, into a black hole, when @xmath270. the function @xmath263 tends to finite values as @xmath265. see text for more details., width=648,height=489] we now study this model, specified through eq. ([modeladopted]), using a dynamical system analysis. 0.5 cm (i) first analysis : the plots and numerics 0.3 cm _ (a) key dynamic equation : _ in order to find the regular solutions @xmath53, @xmath86 and @xmath87 in the whole interval @xmath273 let us transform the master eqs. ([11e1]), ([12e11]) and ([13e1]) to the independent variable @xmath274, a dimensionless radius, and to the unknown dimensionless function @xmath168 given in eqs. ([nrnnew]) and ([yrn]), i.e., @xmath275\,. \label{ydef}\]] the physical interpretation of @xmath168 is connected with the so - called effective mass of the object, @xmath170, see eq. ([nyxrn]). putting these definitions into eq. ([13e1]), we obtain the following key equation @xmath276 this equation is indeed a key one, since using its solution, @xmath168, we can represent explicitly the electric field by @xmath277 the metric function @xmath167 by eq. ([nyxrn]), and @xmath278 by the integral form @xmath279 moreover, @xmath78 follows from @xmath280 and @xmath77 from @xmath281, i.e., @xmath282 \exp{\left\{2a \int^x_{\infty } \frac{d x'}{x'}\,z^2(x')\right\ } } \,. \label{newb}\]] note that the function @xmath168 is also a function of the quantity @xmath162, so in general should be written as @xmath283. 0.5 cm _ (b) three typical cases : _ in this problem there are two independent dimensionless quantities, constructed from @xmath155, @xmath156, and @xmath31, namely @xmath162 and @xmath163, see eqs. ([aratio]) and ([k0]). note that @xmath163, besides fixing the ratio between the total mass and the charge of the object, gives the value of the dimensionless mass @xmath168 at @xmath245, since @xmath284. taking into account the quantity @xmath163, and in conformity with the reissner - nordstrm solution, let us distinguish three different situations, (i) @xmath199, (ii) @xmath204, (iii) @xmath208, within each situation the quantity @xmath162 can vary from zero to infinity [k=2sqrt2gaussbonnetmodel]-[k=1gaussbonnetmodel] display typical cases in each situation, and fig. [yingausbonnetmodel] shows the behavior of @xmath168. in slightly more detail : (i) @xmath199 (i.e., @xmath200) : for @xmath199, we use @xmath285 as a typical value for the numerical analysis, see the plots in fig. [k=2sqrt2gaussbonnetmodel], (see fig. [k=2sqrt2gaussbonnetmodel]a for @xmath256, fig. [k=2sqrt2gaussbonnetmodel]b for @xmath94, fig. [k=2sqrt2gaussbonnetmodel]c for @xmath286, and fig. [k=2sqrt2gaussbonnetmodel]d for @xmath53). when @xmath172 this case gives the usual reissner - nordstrm black hole with two horizons. for other @xmath162s there are also black holes, some with different properties. (ii) @xmath204 (i.e., @xmath205) : for @xmath287 see the plots in fig. [k=2gaussbonnetmodel], (see fig. [k=2gaussbonnetmodel]a for @xmath256, fig. [k=2gaussbonnetmodel]b for @xmath94, fig. [k=2gaussbonnetmodel]c for @xmath286, and fig. [k=2gaussbonnetmodel]d for @xmath53). when @xmath172 this case gives the extreme reissner - nordstrm black hole with one horizon. for other @xmath162s there are also interesting solutions with black holes. (iii) @xmath208 (i.e., @xmath209) : for @xmath208, we use @xmath272 as a typical value for the numerical analysis, see the plots in fig. [k=1gaussbonnetmodel], (see fig. [k=1gaussbonnetmodel]a for @xmath256, fig. [k=1gaussbonnetmodel]b for @xmath94, fig. [k=1gaussbonnetmodel]c for @xmath286, and fig. [k=1gaussbonnetmodel]d for @xmath53). when @xmath172 this case gives a reissner - nordstrm naked singularity, a solution without horizons. for other @xmath162s there are also solutions. and @xmath283 for the non - minimal gauss - bonnet model, with @xmath288, @xmath6 and @xmath7, of gravitational electrically charged objects characterized by @xmath259 plots of the curves @xmath289 and @xmath283 are shown, with @xmath290. for @xmath291 (with @xmath292) the characteristic curve @xmath293 is not intercepted by the function @xmath283. for @xmath269 the intersection takes place at @xmath294. for @xmath268 the point of crossing floats along the characteristic curve, and the integral curve has two branches. finally, for @xmath267 there is no intersection of @xmath283 with the characteristic curve. in the plot @xmath289 it is shown explicitly the existence of a point obeying @xmath295 with positive @xmath296., width=648,height=453] 0.5 cm _ (c) scaling of the key equation : _ the key eq. ([key]) remains invariant after the following scale transformations : @xmath297 thus, the critical values of @xmath162 that one may eventually encounter when @xmath272, are also critical values that one can easily find for arbitrary @xmath163 using the formula @xmath298. 0.5 cm (ii) second analysis : critical properties of the family of the solutions 0.3 cm _ (a) about the mass function @xmath283 when @xmath294, @xmath289 - the critical value of the quantity @xmath162, @xmath296 : _ we now write explicitly that @xmath180 is a function of both @xmath164 and @xmath162, @xmath299 since this is important to our analysis. plots of this dimensionless mass function @xmath283 are displayed in fig. [yingausbonnetmodel]. in fig. [yingausbonnetmodel]a, @xmath283 is shown for several values of @xmath162, and in fig. [yingausbonnetmodel]b, a plot for @xmath289 as a function of @xmath162 is shown. a simple qualitative analysis shows that the mass function @xmath283 at the central point @xmath294, i.e. @xmath289, as a function of the quantity @xmath162, has to possess a zero. indeed, when @xmath172, @xmath300, (where, recall, @xmath301), corresponding thus to the reissner - nordstrm solution. when @xmath302, @xmath303, corresponding thus to the schwarzschild solution. note as well that when @xmath304, @xmath305. in addition, when @xmath306, @xmath307, a condition at infinity that holds for arbitrary @xmath162. in order to prove our assertion, that @xmath289 as a function of @xmath162 possesses a zero, consider then @xmath289 as a function of the quantity @xmath162 in the interval @xmath308. one can see, that @xmath309, and @xmath310. supposing that @xmath289 is continuous in such an interval, one can conclude that there exists at least one specific value of the @xmath162 quantity for which @xmath311, where @xmath296 is the value of @xmath162 for which @xmath311. [yingausbonnetmodel]b shows that, for @xmath272 the zero of the function @xmath312 happens, when @xmath313. but the most interesting fact is that the curve @xmath314 displays a discontinuity in the first derivative with respect to @xmath162 just at @xmath269. one can see explicitly a finite jump of the derivative at this point, @xmath269. nevertheless, the function @xmath315 itself is continuous at this point. that is why @xmath296 is a critical value of the quantity @xmath162. for other @xmath163s the critical values can be found using the scaling properties ([scaletransformations]), yielding @xmath316. for instance @xmath317 and, for the typical case we study, @xmath318. for negative @xmath162, @xmath319 asymptotically when @xmath320. 0.5 cm _ (b) the critical points of the associated autonomous two - dimensional dynamical system : _ the key eq. ([key]) can be put as an autonomous dynamical system @xmath321 where @xmath322 and @xmath323 is an auxiliary parameter. in eq. ([autonomoussystem]) there is one critical point at @xmath324 in the vicinity of this critical point the variables @xmath164 and @xmath180 are connected by the relation @xmath325, which means this point is a saddle point when @xmath162 is positive, and a center when @xmath162 is negative. if @xmath326 there are two separatrices @xmath327. the equation for @xmath328 can also be written as dynamical, @xmath329 \,, \label{21}\]] which is much more complicated than eq. ([autonomoussystem]) for @xmath168. nevertheless, if @xmath3 is positive (i.e., @xmath162 is positive), one can find the critical points immediately, @xmath330 in order to present the integral curves for the total interval of the auxiliary parameter @xmath323, we resort to numerical calculations. the results are presented in figs. [k=2sqrt2gaussbonnetmodel]-[k=1gaussbonnetmodel]. it is clear that for the critical @xmath269 the curve for @xmath331 tends to one when @xmath112, and takes a finite value @xmath332 at the center of the object, @xmath108. this critical curve is a separatrix between the curves having @xmath270 and those having @xmath333. the same type of behavior happens with the curves for @xmath94. one also has that for @xmath334 there exists a unique integral curve for @xmath168 with asymptotic value given by @xmath335, and for @xmath265 one has @xmath295. again, this curve behaves as a separatrix. similar reasoning goes to the curve @xmath53. in other words, the critical point @xmath336, given in eq. ([criticalpointsy1/a]), is just a saddle point, corresponding to the critical value @xmath296, obtained numerically, and it is the final point of the unique integral curve, which coincides with the separatrix of the function @xmath337 at small @xmath164. when @xmath258 and, thus, @xmath162 is negative, there are no regular or quasi - regular solutions to the master equations for the whole interval @xmath338. 0.5 cm _ (c) vertical asymptotes and the electric barrier : _ from eq. ([key]) one sees that, when @xmath339, the derivative @xmath340 becomes infinite and vertical asymptotes appear in the graph of @xmath168 versus @xmath164. at the point @xmath341, for which @xmath342, the electric field ([newe]) becomes infinite, and so these vertical asymptotes can be interpreted in terms of an electric barrier. when @xmath267, figs. [k=2sqrt2gaussbonnetmodel]d, [k=2gaussbonnetmodel]d, and [k=1gaussbonnetmodel]d show that for all three values of @xmath163 the curve for the electric field has a form of a finite barrier. indeed for @xmath253 the electric field @xmath53 tends to zero, then at some radius it reaches a maximum, the barrier height, and finally goes to a finite positive value when @xmath265. a charged test particle with sufficiently high energy can overcome this barrier, be trapped in the potential well and then oscillate inside. when @xmath172, more specifically, when @xmath343, the barrier height increases and tends towards the center @xmath265. for @xmath172 (see again figs. [k=2sqrt2gaussbonnetmodel]d, [k=2gaussbonnetmodel]d, and [k=1gaussbonnetmodel]d) one obtains the standard, reissner - nordstrm, behavior, @xmath344. this means that the electric barrier has become infinite, taking its maximum value (an infinite value) at the center @xmath108. in other words, the vertical asymptote for the electric field appears at @xmath108. finally, when @xmath326 this vertical asymptote and the position of the infinite electric barrier shift from the center towards positive @xmath56 values, then stop at some value of the quantity @xmath162, drift back again to smaller values of @xmath56 and, finally, the infinite electric barrier disappears at @xmath345, (with @xmath292). when @xmath269 the curve tends to the horizontal asymptote at @xmath121, there is yet no trap. when @xmath346, one can see a finite electric barrier with the corresponding traps, for all values of @xmath163. since the integrals in ([newab]) and ([newb]) diverge when @xmath53 is discontinuous, then in searching for solutions with regular functions @xmath53, @xmath256, and @xmath94, we should reject all the cases where vertical asymptotes appear. for @xmath347, numerical calculations show that vertical asymptotes do not appear. so we will mainly consider solutions for this range of the quantity @xmath162, i.e., @xmath345. (d) horizons : _ to analyze the @xmath345 case, with a non - singular electric field and possible non - singular metric potentials @xmath79 and @xmath77, we have studied previously, for comparison, the usual static spacetimes with @xmath172, i.e., the schwarzschild and reissner - nordstrm spacetimes, which are special solutions of eq. ([metric1]). one has for these that @xmath348. for @xmath345 one sees from ([expansionofab]) that @xmath349, in contrast to the @xmath172 case. nevertheless, as in the @xmath172 case, horizons are still given by the condition @xmath350, or @xmath192. we have analyzed this numerically. for @xmath270 the results are the following : (i) @xmath199 (i.e., @xmath200)) : a typical case is @xmath351, see fig. [k=2sqrt2gaussbonnetmodel]. one finds that for @xmath352, there is only one horizon. (ii) @xmath204 (i.e., @xmath205) : for @xmath287, see fig. [k=2gaussbonnetmodel]. one finds that for @xmath353, there is one horizon also. (iii) @xmath208 (i.e., @xmath209)) : a typical case is @xmath354, see fig. [k=1gaussbonnetmodel]. one finds that for @xmath355, there is only one horizon also. note that figs. [k=2sqrt2gaussbonnetmodel]-[k=1gaussbonnetmodel] show that when @xmath270, for all formal possibilities (@xmath199, @xmath204, @xmath208) the curves @xmath79 tend monotonically to minus infinity and cross the line @xmath356 only once. thus, for arbitrary @xmath270 the plots of @xmath79 are continuous, irregular at the center and characterized by one horizon. these solutions have thus an analogous behavior to the schwarzschild black hole. moreover, by tuning the non - minimal quantity @xmath162 one can turn a reissner - nordstrm naked singularity, with @xmath208 and @xmath172, into a black hole, when @xmath270. now, since the @xmath269 is a very special case, we discuss it in particular. 0.5 cm _ (e) the solution with @xmath269 : _ in the framework of the model under discussion, i.e., when @xmath288, @xmath6, and @xmath7, all the three functions, @xmath53, @xmath256, and @xmath94, are regular in the interval @xmath357 if and only if @xmath358 (recall @xmath359). this means that we deal with a one - parameter family of exact solutions, the arbitrary quantity being @xmath360, and all the curvature coupling constants, @xmath0, @xmath1 and @xmath2, being expressed explicitly via @xmath163. the critical @xmath3 corresponding to @xmath296 is thus, with the help of eqs. ([rq])-([k0]), given by, @xmath361. we now consider these solutions in more detail : (i) there are two different solutions, corresponding to the separatrices @xmath362 at small @xmath164. the physical solution, the solution that gives the appropriate limit when @xmath363 and has no jumps on the derivative of the characteristic curve @xmath364, is @xmath365. the plots of @xmath256, @xmath94 and @xmath53 are displayed in figs. [k=2sqrt2gaussbonnetmodel]-[k=1gaussbonnetmodel]. (ii) the solutions, which we discuss, are characterized by finite values at the center, and @xmath140, @xmath366, and @xmath367, but @xmath225 and @xmath368. (iii) for this model @xmath369, and there are two distinct critical radii ; first, the radius for which @xmath370 ; second, the radius for which @xmath371 which signals an infinite redshift surface, and an event horizon in the case of static spacetimes, such as the ones we are treating here. (iv) for @xmath372, one finds that such a solution exists when @xmath373, i.e., @xmath374 ; this is a necessary condition (see, e.g., ([small2])). moreover, within this case it is possible to have such a zero when @xmath375, i.e., @xmath376 or @xmath377. for @xmath378, one has, @xmath379 is positive and without zeros. (v) for the infinite redshift surface and event horizon, @xmath371, one finds that when @xmath269, for the critical quantity, the function @xmath94 is zero both at @xmath108 and at the radius for which @xmath372. (vi) the kretschmann scalar diverges at @xmath108, so although the metric functions are regular, spacetime is not, time stops. 0.5 cm _ (f) remark : _ some important aspects derived from qualitative and numerical analyses of this non - minimal model, with @xmath288, @xmath6 and @xmath7, can be found in @xcite. our results are in thorough concordance with this initial analysis. we have supplemented those aspects on several grounds, of which we stress briefly three novel details obtained here : (i) we have found a complete converging decomposition of the solution of the abel equation based on the recurrence formula ([191]). this gives us not only the asymptotic decompositions for @xmath112, but also the possibility to link the limiting formulas for @xmath265 and @xmath112 (see ([18])-([expansionofab])). (ii) we have formulated and discussed the problem of the infinite electric barrier, associated with the vertical asymptote appearing when @xmath380, thus completing physically and mathematically the analysis given in @xcite. (iii) we have found, first, the critical value of the non - minimal parameter @xmath3, namely, @xmath361, second, the scaling law of the critical parameter @xmath381 for different values of the asymptotic quantity @xmath382, namely, @xmath383, third, the significance of the choice for @xmath163, namely, @xmath199, @xmath204, and @xmath208, in the qualitative analysis. now, we consider the model with @xmath384 which, as we have seen, has the property that @xmath53 decouples from @xmath86 and we deal with a cubic equation for the determination of the electric field. since the basic feature of the model is that it is integrable, we call it the integrable model. for such a model the susceptibility tensor @xmath23 can be written in terms of the einstein tensor @xmath385 as follows, @xmath386, % \label{chie } $] where we have put @xmath387. thus, the model becomes one - parametric and we can introduce the dimensionfull quantities defined above, @xmath156, @xmath157, @xmath31, and the dimensionless quantities @xmath162 and @xmath163. we assume that @xmath157 inherits the sign of the charge @xmath58 and the quantity @xmath162 can be positive or negative depending on the sign of @xmath3. then, we introduce the two dimensionless variables, the normalized radius @xmath164 and the normalized electric field @xmath165, defined in ([dimssrn]). in terms of these, the system of key equations can be rewritten as @xmath388 @xmath389 = 1 - 2 z + (x^2+a) z^2 \,, \label{2eqdim}\]] @xmath390 clearly, eq. ([1eqdim]) for @xmath165 is the key equation for finding @xmath167 and @xmath278 from ([2eqdim]) and ([3eqdim]), respectively. if instead of @xmath167 one uses @xmath168 then eq. ([2eqdim]) gives an equation for @xmath391 of the type of eq. ([ydashrn]) or eq. ([key]), but more complicated, which for this analysis is not very illuminating. for this model it is better to start analyzing the electric field @xmath53, or its redefinition @xmath165. consider now eq. ([1eqdim]) in detail. ([1eqdim]) is a one - parameter algebraic equation of third order for the dimensionless electric field. below we denote its solution as @xmath392. the function @xmath392 can be generally presented by the well - known cardano formula, nevertheless we prefer to analyze qualitatively this one - parameter family of solutions. depending on the value of the quantity @xmath162 the solution @xmath392 can possess one or three real branches. the corresponding plots are presented in fig. [electricfieldinnonamemodel]. when @xmath172, one obtains, as it should, the coulombian solution @xmath393. the curves displaying @xmath392 for non - vanishing values of the quantity @xmath162 are more sophisticated. and @xmath9, the rescaled electric field @xmath392, @xmath394, as a function of the scaled radius @xmath164 @xmath395 of gravitational electrically charged objects characterized by @xmath396 (a) displays the solution @xmath392 when the non - minimal quantity @xmath162 satisfies the inequality @xmath397. there are three real branches of the solution @xmath392. nevertheless, only one of them is defined on the whole interval @xmath398. only one branch tends to the horizontal asymptote @xmath399 at @xmath363. (b) displays the solution @xmath392 in the case @xmath400. there are three real starting points @xmath401, @xmath402, @xmath403 for the three corresponding branches of the solution @xmath392. nevertheless, two branches are not continuous, only the third being defined on the whole interval @xmath404. (c) displays the solution @xmath392 for the important case @xmath405, the fibonacci soliton. (d) displays the solution @xmath392 in the case @xmath406. (e) displays the solution @xmath392 in the case @xmath172 which is a coulombian electric field. the curve is not continuous. (f) gives an example of the solution for negative @xmath162 with a vertical asymptote. when @xmath162 tends to zero remaining negative, the vertical asymptote shifts towards the line @xmath294, and the solution @xmath407 converts, finally, into the coulombian solution. at large values of @xmath164 the plots of the function @xmath408 tend to the coulombian curve @xmath409 for all values of the quantity @xmath162., width=648,height=648] 0.2 cm _ (i) @xmath326 : _ when @xmath162 is positive, the functions @xmath392 take finite values for all values of the quantity @xmath162, see the curves a, b, c, d in fig. [electricfieldinnonamemodel]. the initial values @xmath410 satisfy the cubic equation @xmath411. there is only one real solution of this cubic equation for @xmath326, if the discriminant @xmath412 is positive, i.e., when @xmath413, see box @xmath414 of fig. [electricfieldinnonamemodel] for details. this plot displays three real branches of the solution @xmath392, nevertheless, only one of them is defined on the whole interval @xmath398. two other branches are defined for @xmath415 only and contact at the point @xmath416. only one branch tends to the horizontal asymptote @xmath399 at @xmath363. when @xmath417, the discriminant @xmath418 is negative or equal to zero, which guarantees that there are three real starting points, @xmath401, @xmath402, @xmath403 for the three corresponding branches of the solution @xmath392, see the curves in boxes b, c, d of fig. [electricfieldinnonamemodel]. nevertheless, when @xmath419, two branches of the solution @xmath392 are not continuous, only the third being defined on the whole interval @xmath420, see box b of fig. [electricfieldinnonamemodel]. when @xmath421 all three branches are continuous and are defined on the whole interval @xmath422, one of them is asymptotically coulombian. there are three horizontal lines @xmath423, @xmath399 and @xmath424, which yield distinct ranges for the functions @xmath425, @xmath426, and @xmath427. clearly, the curve of coulombian type is in - between the separatrices @xmath399 and @xmath428. the model with critical @xmath162, call it @xmath296 again, is the one that has @xmath234, i.e., @xmath405.this model can be solved analytically, and we consider this case in detail below. finally, when @xmath162 tends to zero remaining positive, the starting points @xmath401 tend to minus infinity, and @xmath402, @xmath403 grow infinitely. clearly, at @xmath429 the branch @xmath426 is the only branch that remains visible at finite values of @xmath164, and is coulombian. 0.2 cm _ (ii) @xmath172 : _ when @xmath172, one obtains the coulombian solution @xmath393, see the curve in box @xmath430 of fig. [electricfieldinnonamemodel]. 0.2 cm _ (iii) @xmath267 : _ when @xmath162 is negative, the coefficient @xmath431 in the first term of eq. ([1eqdim]) vanishes at @xmath432 and the line @xmath433 is the vertical asymptote of the graph @xmath392, see the example of the curve for @xmath434 in box @xmath435 of fig. [electricfieldinnonamemodel]. when @xmath162 tends to zero remaining negative, the vertical asymptote shifts towards the line @xmath294, and the solution @xmath407 converts, finally, into the coulombian solution. at large values of @xmath164 the plots of the function @xmath408 tend to the coulombian curve @xmath409 for all values of the quantity @xmath162. thus, for @xmath436 the solutions @xmath392 are not regular at all in the range @xmath398. to study the gravitational part of the solution, we observe that the solution of ([2eqdim]) for @xmath106 can be presented in quadratures @xmath437 \right\ } \,, \label{31eqdim}\end{aligned}\]] where @xmath392 is supposed to be already found. the constant of integration, @xmath163, can clearly be related to the asymptotic mass of the object @xmath188. when @xmath438 and @xmath439, eq. ([31eqdim]) yields that @xmath440, as defined in ([k0]). on the other hand, searching for a solution @xmath441, which is finite at @xmath294, and taking into account that @xmath410 is finite, we should require that @xmath442 \,. \label{kzreq}\]] in this case the formula ([31eqdim]) transforms into @xmath443 \,, \label{431eqdim}\end{aligned}\]] providing @xmath444 since the electric field at the center satisfies the condition @xmath445 (see ([1eqdim])), then @xmath446 can be rewritten as @xmath447 in accordance with the first relation from ([regularity1]). thus, for the family of solutions with regular functions @xmath448 and @xmath449 the quantity @xmath163 is a function of the value @xmath410, i.e., depends on the quantity @xmath162 according to the formula ([kzreq]), @xmath450. also, from ([3eqdim]) for @xmath252 one finds @xmath451 \right\ } \,, \label{21eqdim}\]] with @xmath392 being found from ([1eqdim]). for the asymptotically coulombian branch we have to set @xmath452 and thus @xmath453. then, the function @xmath79 is taken from @xmath454 and the function @xmath77 is taken from @xmath78 and @xmath252, @xmath281. , @xmath8 and @xmath9, of gravitational electrically charged objects characterized by @xmath396 plots (a), (b), (c), and (d) depict the functions @xmath455, @xmath77, @xmath456 and @xmath457, respectively, as functions of @xmath264, for three typical values of the quantity @xmath162, when @xmath458. these functions are regular and take finite values at @xmath108. the solution with @xmath459 is a solution without horizons, since @xmath455 and @xmath77 are positive everywhere. indeed the @xmath405 solution is a soliton of the model, the fibonacci soliton. it has a mild conical singularity at the center, and is a solution without horizons.,width=648,height=486] in the reissner - nordstrm solution and in the previous discussed model, i.e., the gauss - bonnet model (see sections [rnmodel] and [gbmodel], respectively), we divided the solutions according to @xmath199, @xmath204, and @xmath208. here it is no more convenient to make such a division. the reason is that @xmath163 is not a free quantity here, rather @xmath450. since it is not a free quantity, we can not classify the models with respect to it, we can only calculate this quantity (numerically) after solving the problem as a whole. in the gauss - bonnet model of section [gbmodel] we could classify through @xmath163 because the equations for @xmath165 and @xmath167 can not be decoupled, thus, @xmath410 and @xmath446 are connected. this means that we can choose @xmath163 as a convenient quantity for the classification, the dependent quantity being @xmath410. here, in this model, the solution for @xmath165 satisfies a decoupled equation, the latter does not depend on @xmath167. thus solving the decoupled equation for @xmath392, we can classify the quantity @xmath410 as the independent one. then, we obtain @xmath441 and, as we see, @xmath446 depends on @xmath460, an so @xmath163 is a dependent quantity, @xmath450. an explicit example where @xmath461 is calculated is given below for the case @xmath405. so, as when discussing the electric field, we again divide the analysis into three cases, here with subcases. 0.2 cm _ (i) @xmath326 : _ we should further divide into three subcases : @xmath462 : the electric field is discontinuous or irregular at the center. since in this work we focus on regular electric fields everywhere, although interesting, we do not discuss these models here. @xmath234, i.e, @xmath405 : we plot in fig. [metricinnonnamemodel], boxes (a), (b), (c), (d), the functions @xmath79, @xmath77, @xmath463, and @xmath457, respectively. from the figure it is clear that @xmath464 is characterized by the absence of horizons. for @xmath405 the solution is regular, or better, quasi - regular, and is a soliton of the theory, the fibonacci soliton. due to its interest, the case @xmath465 will be solved next explicitly. @xmath466 : this case is interesting. in this case we plot in fig. [metricinnonnamemodel], boxes (a), (b), (c), (d), the functions @xmath79, @xmath77, @xmath463, and @xmath457, respectively, for two positive values of @xmath162 within this range, namely, @xmath467. from the figure it is clear that positive @xmath162 in this range gives one horizon, and the solution possesses a quasi - regular center. since the singularity at the center is a conical one, these black hole solutions can be considered as quasi - regular solutions, and thus are of great interest. note that extremal black holes have only one zero, which in turn is a double zero. so from the figure above, the non - minimal black holes do not characterize as extremal. rather, they are of the schwarzschild type, with one horizon, a spacelike singularity, but here, different from schwarzschild, the singularity is mild, it is a conical singularity. 0.2 cm _ (ii) @xmath468 : _ it is the reissner - nordstrm case. for this case the electric field is irregular at the center. 0.2 cm _ (iii) @xmath469 : _ for these cases the electric field is discontinuous or irregular at the center, like the reissner - nordstrm case, and we do not discuss these models here. let @xmath405. then the cubic equation ([1eqdim]) takes the form @xmath470 = 0 \,. \label{3to2}\]] one sees that eq. ([3to2]) splits into one linear equation and one quadratic equation. one branch of solutions of ([3to2]), the linear one, describes a constant electric field @xmath471, or, equivalently, @xmath472. this branch is of no great interest. another branch @xmath473 is given by the function @xmath474 $], which is bounded. the graph of this function starts from @xmath475 and tends asymptotically to the line @xmath476. the behavior of such electric field is not of coulombian type, and will be not discussed further. yet, there is a third branch. the branch @xmath477 given by the formula @xmath478 \,, \label{z3}\]] describes a coulombian type electric field. at @xmath363, one has @xmath479, or equivalently, @xmath57. the graph of this function starts from @xmath480. interesting to note that the starting points @xmath481 are associated to the well - known fibonacci series and the `` golden section '' @xmath482. for the coulombian type solution, the function @xmath483, which we write simply as @xmath167 when suitable, is regular at the center only if the constant @xmath163 satisfies ([kzreq]). the corresponding quadrature for @xmath167 is @xmath484 \,, \label{nsi}\]] where @xmath278 is given by ([si]). clearly, @xmath485 and @xmath486, so that @xmath487, and the relations ([regularity1]) are satisfied. the plot of the function @xmath167 for @xmath405 is shown in fig. [metricinnonnamemodel]. clearly, the function @xmath167 is positive in the interval @xmath488. for the coulombian type solution ([z3]) the function @xmath489, which we write simply as @xmath278 when suitable, is given by @xmath490 with @xmath491 being equal to @xmath492. then one finds @xmath79 from @xmath454 and @xmath493. the function @xmath494 is also positive, and @xmath495. thus, this solution is a solution without horizons and is regular. although the curvature scalars diverge, the singularity at the center is a mild one, it is a conical singularity. the asymptotic mass @xmath188 of the object defined as @xmath496 \right\ } \,, \label{mass1}\]] is represented in this case by the integral @xmath497 \,. \label{mass2}\]] numerical calculations give the value @xmath498 which yields in addition, @xmath499 this is thus a very interesting solution. it is a soliton, in the sense that is made of the very own fields of the theory, the gravitational and electric fields, it is a solution without horizons, and it is quasi - regular, with a conical singularity at the center. note also that the value @xmath405 can be regarded as a critical one. there are two reasons for this. first, it is clear from fig. [electricfieldinnonamemodel], that for the unique case @xmath405 there exists a bifurcation point, in which the two branches of the curve @xmath392, or @xmath449, intersect. when @xmath500, the coulombian branch of the electric field curve is discontinuous. when @xmath501, three continuous regular branches exist. the second reason is that at @xmath405 the corresponding curve on fig. [metricinnonnamemodel] plays a role of a separatrix ; when @xmath502, desirable curves, continuous and regular at the center, exist, otherwise they do not appear. in fig. [table] we present a table in which the results of the various models studied are summarized.
Conclusions
we have shown that the original non - minimal einstein - maxwell theory with three parameters @xmath0, @xmath1, and @xmath2, is reducible in natural different ways to a theory with one parameter @xmath3 only, in which the three parameters obey two relations between themselves. we have then studied two special models for static spherically symmetric solutions obeying the following requirements : the electric field @xmath53 is regular everywhere in the interval @xmath503, being coulombian far from the center. from the solutions of this class we extract the ones, for which the metric coefficients @xmath256 and @xmath94 are regular at the center @xmath108 and tend to one asymptotically as @xmath504. the first non - minimal model, the gauss - bonnet model (with @xmath5, @xmath6, @xmath7, @xmath3 free), displays charged black hole solutions with one horizon only, when the dimensionless non - minimal quantity @xmath162, with @xmath259 naturally appearing in the model, exceeds a critical value @xmath296, @xmath270. although the black hole is electrically charged the solutions found have one horizon only, and are similar in this connection to the schwarzschild solution. when @xmath333, the solutions are discontinuous in the interval @xmath503, or irregular at the center @xmath108. another main result in this model is that there exists a unique solution, the solution for the critical value @xmath505, which does not possess horizons and is characterized by regular fields @xmath53, @xmath261 and @xmath94, with @xmath506 and @xmath507, although the curvature invariants blow at the origin. the second model, the integrable model (with @xmath508, @xmath8, @xmath9, @xmath3 free), is also characterized by one critical value @xmath509 of the non - minimal quantity @xmath162. when @xmath267 or @xmath500, the solutions are irregular. when @xmath502 one obtains black holes with electric field regular everywhere and with only one horizon, like the schwarzschild solution. finally, when @xmath405, i.e., at the critical value of the quantity @xmath162, there exists a solitonic solution with a conical singularity at the center, but otherwise well behaved this solution can be called the fibonacci soliton, since the well - known @xmath510 number (@xmath511), associated with the golden section, appears naturally in the expressions for the central values of the electric field @xmath512 and the metric coefficients are also related to @xmath510, namely @xmath513 and @xmath514 $]. summing up, we can say that the non - minimal curvature induced interaction between the gravitational and electromagnetic fields provides an electric field, of static spherically symmetric charged objects, which is regular everywhere for different relations between the coupling constants @xmath0, @xmath1 and @xmath2. as for additional regularity of the metric coefficients @xmath94 and @xmath93, the non - minimal interaction can provide models which have very specific, critical, values for the coupling constants, in which the geometry has at most a conical, and thus mild, singularity. this is in line with problem posed by bardeen @xcite, where one should look for theories with regular black hole solutions. we have partially solved it within these models.
Acknowledgments
ab thanks the hospitality of centra / ist in lisbon, and a special grant from fct to invited scientists. this work was partially funded by fundao para a cincia e tecnologia (fct) - portugal, through project poci / fp/63943/2005, and by russian foundation for basic research through project rfbr 08 - 02 - 00325-a. | using a lagrangian formalism, a three - parameter non - minimal einstein - maxwell theory is established.
the three parameters, @xmath0, @xmath1 and @xmath2, characterize the cross - terms in the lagrangian, between the maxwell field and terms linear in the ricci scalar, ricci tensor, and riemann tensor, respectively.
static spherically symmetric equations are set up, and the three parameters are interrelated and chosen so that effectively the system reduces to a one parameter only, @xmath3. specific black hole and other type of one - parameter solutions are studied.
first, as a preparation, the reissner - nordstrm solution, with @xmath4, is displayed.
then, we seek for solutions in which the electric field is regular everywhere as well as asymptotically coulombian, and the metric potentials are regular at the center as well as asymptotically flat. in this context, the one - parameter model with @xmath5, @xmath6, @xmath7, called the gauss - bonnet model, is analyzed in detail.
the study is done through the solution of the abel equation (the key equation), and the dynamical system associated with the model.
there is extra focus on an exact solution of the model and its critical properties. finally,
an exactly integrable one - parameter model, with @xmath5, @xmath8, @xmath9, is considered also in detail.
a special sub - model, in which the fibonacci number appears naturally, of this one - parameter model is shown, and the corresponding exact solution is presented.
interestingly enough, it is a soliton of the theory, the fibonacci soliton, without horizons and with a mild conical singularity at the center. | 0712.4066 |
Introduction
it is widely believed that there are supermassive black holes at the centers of galaxies, and these are hypothesized to be the central engines for active galactic nuclei (agns) and gamma ray bursts (grbs). two main possibilities are considered as the energy source. one is the gravitational energy of accreting matter and the other is the rotational energy of the black hole or the accretion disk surrounding it. however, the details of the energy extraction process are not clear. it is also not understood well how the energy is converted into that of agns or grbs. blandford and znajek showed that the rotational energy of a rotating black hole can be extracted in the form of poynting flux along magnetic field lines penetrating the event horizon @xcite, which is known as the blandford - znajek (bz) mechanism. its efficiency depends on the rotational velocity of the black hole and the configuration of the magnetic field : the extraction of the rotational energy becomes more efficient the more magnetic field lines penetrates the event horizon and the more rapidly the black hole rotates. in the bz mechanism, poloidal magnetic fields which penetrate the event horizon play a crucial role for the energy extraction as well as for the formation of jets associated with agns. in fact, some numerical studies reported that poynting - dominated jets were produced @xcite. bick and janis showed that a magnetic field without an electric current is expelled from the event horizon of a maximally rotating black hole @xcite. this is analogous to the meissner effect in a superconductor. this effect for a rapidly rotating black hole would decrease the efficiency of the bz mechanism, though the larger rotational velocity of the black hole would increase the efficiency. in realistic astrophysical cases, however, there would be plasma around the black hole. how the meissner - like effect is affected by the existence of plasma is the main subject of this paper. we clarify the effect of an electric current on the meissner - like effect of an extreme black hole. komissarov and mckinney studied numerically the meissner - like effect of a kerr black hole @xcite. they carried out numerical simulations for a highly conductive magnetosphere until it almost reaches steady state, and there was no sign of the meissner - like effect in their numerical results. in this paper, we study how an electric current affects the meissner - like effect by solving a stationary problem analytically. since realistic situations are, in general, very complicated, it is difficult to model them. in order to reveal the essence of the plasma effect, we consider a very simple toy model : (i) we consider a stationary, axisymmetric force - free system of the electromagnetic field and plasma ; (ii) we consider a static spherically symmetric black hole spacetime with a degenerate horizon as a background spacetime rather than a rotating black hole. the degenerate horizon is the origin of the meissner - like effect in a vacuum black hole spacetime @xcite, and hence, by studying the electromagnetic field in this spacetime, we can see whether the meissner - like effect remains even in the case with an electric current. the spacetime considered in this paper is known as the reissner - nordstrm (rn) spacetime. by these assumptions, the basic equations reduce to only one quasi - linear elliptic equation for the magnetic flux function called the grad - shafranov (gs) equation @xcite. for the black hole spacetime, the gs equation has three regular singular points : one is at the event horizon, and the other two are at the inner and outer light surfaces on which the velocities of the magnetic field lines agree with the speed of light. for non - extreme cases, one boundary condition is imposed at each regular singular point so that the magnetic field is smooth everywhere. however, for a given electric current function, the obtained solution for the magnetic flux need not be @xmath0 but at most @xmath1 @xcite. although numerical @xmath0 solutions have been obtained by iteratively changing the functional form of the electric current @xcite, a mathematically rigorous proof for the existence of a @xmath0 solution has not yet been presented. furthermore, in the extreme case, two kinds of boundary condition must be imposed at once on the event horizon. we shall mention all these difficulties in solving the gs equation in iv. as will be shown in v, the monopole component is a unique configuration of the magnetic field on the event horizon if there is not an electric current. since there is no magnetic monopole in nature, this result implies the meissner - like effect of the extreme rn black hole. in order to study the electromagnetic field coupled to an electric current around an rn black hole, we use a perturbative method which includes two expansion parameters. one of these parameters corresponds to the rotational angular velocity of the magnetic fields. namely, we consider slow - rotating magnetic fields as was first considered by blandford and znajek @xcite. the other parameter is the ratio of the distance from the event horizon to the horizon radius, since we consider only the vicinity of the event horizon, which includes the inner light surface. although we can not take into account the outer light surface in our perturbative method, we can obtain approximate solutions sufficient to study the meissner - like effect with an electric current. this paper is organized as follows. in ii, we introduce the rn black hole as a background geometry. then we show the gs equation for the rn spacetime in iii ; the detailed derivation of the gs equation is given in appendices a and b. the regularity conditions for the gs equation and difficulties in solving this equation are described in detail in iv. using perturbative analyses, we study the cases with and without an electric current in v and vi, respectively. vii is devoted to summary and discussion. in appendix c, we show the relation between the kerr - schild coordinate system and the standard static coordinate system of the rn spacetime. in appendix d, we give a proof of a theorem on the magnetic field obtained by the present perturbative method. in this paper, we adopt the geometrized units, in which the newton s gravitational constant and the speed of light are unity, and the abstract index notation : small latin indices, excluding @xmath2 and @xmath3, indicate the type of tensor, whereas small greek indices, excluding @xmath4 and @xmath5, represent components with respect to the coordinate basis. the exceptional indices @xmath2, @xmath3, @xmath4, and @xmath5 denote the components of time, and the radial and azimuthal coordinates in the spherical polar coordinate system. the signature of the metric is diag@xmath6 $].
Background geometry
we consider a static and spherically symmetric spacetime of the following metric : @xmath7 with @xmath8 where we assume @xmath9. this spacetime is known as the rn spacetime. there are two horizons, which are determined by @xmath10 : @xmath11 and @xmath12 represent the radius of the event and cauchy horizons, respectively. the case of @xmath13 is called the extreme case.
Grad-shafranov equation
maxwell s equations are given by @xmath14 } & = & 0, \label{eq : maxwell - eq-1}\\ \nabla_bf^{ab } & = & 4\pi j^{a }, \label{eq : maxwell - eq}\end{aligned}\]] where @xmath15 is the field strength tensor of the electromagnetic field, @xmath16 is a current density, and @xmath17 is the covariant derivative. if the field strength tensor is expressed by using a 4-vector potential @xmath18 as @xmath19 then eq. ([eq : maxwell - eq-1]) is trivially satisfied, where @xmath20 is the ordinary derivative. as mentioned in @xmath21i, hereafter, we consider the axisymmetric and stationary electromagnetic field in the rn spacetime. in order to make the problem simple, we assume that the system field satisfies the force - free condition @xmath22 hereafter, we focus on the system of only eqs. ([eq : maxwell - eq]) and ([eq : ff - condition]). the formulation of a force - free electrodynamics field in the black hole spacetime was given by macdonald and thorne @xcite. our formulation is based on their work. in the case of a stationary, axisymmetric electromagnetic field, we can define the `` angular velocity '' of the magnetic field as @xmath23 the reason why @xmath24 can be regarded as the angular velocity of the magnetic field is described in appendix [derivation]. from eqs. ([eq : maxwell - eq]) and ([eq : ff - condition]), the gs equation is obtained as @xmath25 where a prime @xmath26 represents a derivative with respect to @xmath3, @xmath27 @xmath28 where @xmath29 and @xmath30 are the electric current and the magnetic flux through an axisymmetric polar cap, which are defined by eqs. ([eq : i - def]) and ([eq : psi - def]), respectively, and @xmath31 is defined by @xmath32 the derivation of the gs equation ([eq : gs - eq]) is given in appendices [derivation] and [i - psi]. in general, the gs equation for a black hole magnetosphere has three regular singular points : one is at the event horizon @xmath10, and the other two are at the light surfaces defined by @xmath33. a light surface is a timelike hypersurface on which the rotational speed of magnetic field lines are equal to the speed of light. the inner light surface given by a function @xmath34 has a spacelike section with spherical topology, whereas the outer one, @xmath35, has that of cylindrical topology. in the case of the kerr spacetime, the kerr - schild coordinate system is often adopted, for example, in numerical simulations (e.g., @xcite), since there is no coordinate singularity on the event horizon. therefore, we give the kerr - schild coordinate system for the rn spacetime in appendix [ks]. in appendix [ks], we show that the singular point on the event horizon in the gs equation appears even if we adopt kerr - schild coordinates. this is also true in the case of the kerr spacetime, though it is not shown in this paper (we will show it elsewhere). as long as a stationary magnetic field is considered, the singular point of the basic equation will appear on the event horizon, since there is no timelike killing vector field on or inside the event horizon, or in other words, the stationary configuration can not be realized inside the black hole.
Regularity conditions and boundary conditions
in this section, we consider only the case of @xmath36. then, by virtue of the symmetry of the background spacetime, without loss of generality, we may assume @xmath37. the case of @xmath38 will be treated as specific cases later. on the symmetry axis @xmath39, both @xmath30 and @xmath40 should vanish by their definitions. @xmath40 is a function of @xmath30, and thus, @xmath41 should vanish. in appendix [ks], we give the components of @xmath15 in the kerr - schild coordinate system @xmath42. from eq. ([eq : rt - comp]), we have @xmath43 where @xmath44 since the kerr - schild coordinate system is non - singular on the event horizon, @xmath45 must be finite there. thus, we have @xmath46. this leads to @xmath47 the above condition corresponds to the horizon boundary condition derived by znajek for the kerr black hole @xcite. it is seen from the gs equation ([eq : gs - eq]) that, in order that @xmath30, @xmath48 and @xmath49 are finite on the event horizon @xmath50, the following condition should be satisfied @xmath51 however, this condition is satisfied if @xmath30 satisfies the regularity condition ([eq : h - regularity]), and thus no additional constraint is imposed by this equation. in the extreme case, since the equation @xmath10 has a double root @xmath52, @xmath53 should vanish as well as @xmath54 so that @xmath45 is finite on the event horizon. these conditions imply @xmath55 on the event horizon @xmath56. we can see from the gs equation ([eq : gs - eq]) that, in order that @xmath30, @xmath48, and @xmath49 are finite on the event horizon for the extreme case, not only eq. ([horizon]) but also the following condition should be satisfied, @xmath57 this condition is satisfied if @xmath30 satisfies the regularity conditions ([eq : h - regularity-1]) and ([eq : h - regularity-2]), and thus no additional constraint is imposed by this equation. as mentioned, the light surfaces are singular points of the gs equation ([eq : gs - eq]). in the extreme case, the radial coordinates of the light surfaces, which are the roots of the equation @xmath33 in the domain @xmath58, are given by @xmath59 in order that @xmath49 and @xmath60 are finite on the light surfaces, @xmath61 must vanish there. this requirement leads to the following regularity conditions on the light surfaces : @xmath62 where @xmath63 , where @xmath64 and @xmath65., scaledwidth=80.0%] for simplicity, in this subsection, we assume that @xmath24 is constant. by virtue of this assumption, the conditions ([eq : neumann]) become neumann boundary conditions on the light surfaces. here, we consider the non - extreme case @xmath66. let us assume that the functional form of @xmath67 has already been determined before solving the gs equation. then, by solving the horizon regularity condition ([eq : h - regularity]), we obtain @xmath30 on the event horizon. since @xmath30 is the magnetic flux through the polar cap (see appendix b), @xmath30 should vanish at @xmath39. thus, there seems to be no freedom for setting a boundary condition for @xmath30 in solving eq. ([eq : h - regularity]), but this is not true. since @xmath39 is a regular singular point of eq. ([eq : h - regularity]), there still remains one degree of freedom for choosing a boundary value of the second - order derivative of @xmath30. hence, a dirichlet boundary condition for the gs equation is determined on the event horizon by the regularity condition ([eq : h - regularity]). by imposing this dirichlet boundary condition at @xmath50 and further neumann boundary conditions at @xmath68 ([eq : neumann]) and on the equatorial plane (e.g., the reflection - symmetric boundary condition @xmath69), a solution for the gs equation ([eq : gs - eq]) is uniquely determined in the domain @xmath70. by imposing two neumann boundary conditions ([eq : neumann]) on the two light surfaces @xmath71 and the equatorial plane @xmath72, and a further dirichlet boundary condition @xmath73 on the symmetry axis @xmath39, a solution for the gs equation is uniquely determined in the domain @xmath74. for the domain @xmath75, if we impose a boundary condition at @xmath30 for @xmath76, then we can obtain a solution for the gs equation. the gs equation can be solved for these three domains, @xmath77, @xmath74, and @xmath78, independently by the above procedure (see fig. [fg : non - extreme]). thus, for an arbitrary electric current @xmath67, the obtained solution for @xmath30 is, in general, not @xmath0 but at most @xmath1 at the boundaries @xmath71. it was first reported by contopoulos, kazanas, and fendt (ckf) that the continuity of the first - order derivative of @xmath30 at the light surface as well as the continuity of @xmath30 itself determines the functional form of the electric current @xmath67 in the case of the pulsar magnetosphere @xcite. they numerically obtained @xmath0 solutions for @xmath30 by an iterative method in which both @xmath30 and the functional form of @xmath67 are determined simultaneously. the ckf method was used by several authors to study the pulsar magnetosphere and they showed that this method also suitable for their case of interest @xcite. however, it should be noted that a mathematically rigorous proof for the existence of the @xmath0 solution for the gs equation has not yet been given. although there is at most one light surface in the case of the pulsar magnetosphere, there can be two light surfaces if @xmath24 is a non - vanishing constant in the case of the black hole magnetosphere (see fig. [fg : non - extreme]). thus, if we chose the functional form of @xmath67 such that @xmath30 is @xmath0 in the domain @xmath79, imposing an asymptotic boundary condition for @xmath76 does not guarantee the continuity of the derivative of @xmath30 at the outer light surface @xmath80. thus, in order to obtain a solution which is @xmath0 in the domain @xmath81, we need to solve the gs equation for the outermost domain @xmath82 as a cauchy problem with a boundary data for @xmath30 and the derivatives of @xmath30 on @xmath80. this implies that we can not impose the asymptotic boundary condition for @xmath83. in general, it is difficult to solve an elliptic - type differential equation numerically, such as the gs equation, as a cauchy problem due to the numerical instability. thus, in the case with two light surfaces, it is difficult to numerically obtain a solution for the gs equation in the outermost domain @xmath84. however, this might not be a serious problem, since we may understand whether the blandford - znajek mechanism works by studying only the domain @xmath85. uzdensky applied the ckf method to the magnetospheres of the schwarzschild black hole @xcite and of the kerr black hole @xcite, but he focused on only the cases in which there is only one light surface by virtue of a particular assumption on @xmath24 : uzdensky assumed that @xmath24 asymptotically decreases, and hence there is no outer light surface. thus, uzdensky succeeded in numerically obtaining global solutions without solving cauchy problems for the gs equation. , but for an extreme rn black hole @xmath86., scaledwidth=80.0%] here, we consider the extreme case @xmath86, which is the main case of interest in this paper. in this subsection, we also assume that @xmath24 is constant. the horizon regularity conditions ([eq : h - regularity-1]) and ([eq : h - regularity-2]) give boundary values of @xmath30 and the derivative of @xmath30. thus, in the extreme case, we must solve the gs equation ([eq : gs - eq]) as a cauchy problem even for the domain @xmath87 (see fig. [fg : extreme]). as mentioned, it is difficult to numerically solve the gs equation as a cauchy problem, and hence, it seems to be difficult to numerically obtain a solution for the physically important domain @xmath88 in the extreme case. further, even if we find a procedure for numerically solving the gs equation as a cauchy problem, the regularity condition on the inner light surface @xmath68 may not be satisfied for an arbitrary functional form of electric current @xmath67 : we must assume the functional form of @xmath67 to solve the gs equation as a cauchy problem, but, in general, the assumed electric current @xmath67 does not satisfy the regularity condition on the inner light surface. as a result, it seems to be impossible to obtain a solution for the gs equation numerically, which is finite on the inner light surface, in the extreme case. in this sense, the perturbative analytic approach discussed in vi is very important. we should note that even if we find analytically the electric current @xmath40 which guarantees the finiteness of @xmath30 and its derivative on the inner light surface, such a electric current @xmath40 might not guarantee the finiteness of both @xmath30 and its derivative on the outer light surface. this implies that either the force - free condition should break down near the outer light surface or the rotational velocity should decay far from the black hole so that the outer light surface does not exist, as in the situation studied by uzdensky.
Vacuum case
in this section, we consider the vacuum case @xmath89. here, we consider the case of @xmath38 on the event horizon, @xmath90. even if @xmath24 does not vanish except on the event horizon, it satisfies @xmath91 thus, on the event horizon, @xmath92 or @xmath93 should hold. in the former case, @xmath30 vanishes on the event horizon, implying that the magnetic flux does not penetrate the event horizon. in the latter case, from eq. ([eq : gs - eq]), in order that @xmath30, @xmath48, and @xmath49 are finite on the event horizon, the following equation should be satisfied : @xmath94 in the extreme case, since @xmath95 also holds on the event horizon @xmath56, we have @xmath96 the solution of the above equation which satisfies the regularity condition on the symmetry axis @xmath39 is @xmath97 where @xmath98 is an integration constant. the above solution implies that the magnetic field which can penetrate the event horizon is the only monopole component. by contrast to the extreme case, non - monopole components can penetrate the event horizon in the non - extreme case, since, in this case, eq. ([eq : h - regularity-0]) does not necessarily imply @xmath99. as a result, we can conclude that the meissner - like effect of the extreme black hole appears in the vacuum case. in the case that @xmath24 vanishes everywhere, we can obtain global solutions. the solutions which satisfy the regularity condition on the symmetry axis @xmath39 are written in the form @xmath100 where @xmath101 is the associated legendre function of the first kind with @xmath102. then, the gs equation ([eq : gs - eq]) becomes @xmath103 in the extreme case, in order that @xmath104, @xmath105, and @xmath106 are finite on the event horizon, @xmath104 of @xmath107 vanishes, and the only monopole component @xmath108 may remain. in this case, from the regularity condition on the event horizon ([eq : h - regularity]), we have @xmath109 for both extreme and non - extreme cases. thus, @xmath73 is a solution which satisfies the regularity condition on the symmetry axis @xmath39. the magnetic field does not penetrate the event horizon at all.
Case with an electric current
in this section, we focus on the extreme case @xmath52 and assume that the rotational velocity of the magnetic field @xmath24 is constant. in the case of @xmath38, it is seen from eq. ([eq : gs - eq]) that since @xmath110, the following condition should be satisfied on the event horizon : @xmath111 the above condition allows @xmath112 on the event horizon. however, since @xmath40 should vanish on the symmetry axis @xmath39, the allowed constant is zero. thus, in this case, the same argument as was used in the vacuum case discussed in the previous section is also true. the allowed configuration of the magnetic field on the horizon is only the monopole component ([eq : monopole]). hence, hereafter, we focus on the case of @xmath36. we are interested in the configuration of the magnetic field near the event horizon. in order to analyze the gs equation, we introduce the following dimensionless quantities : @xmath113 using these quantities, the gs equation ([eq : gs - eq]) becomes @xmath114 where @xmath115, \\ { \cal u}&=&2(y-1-\varepsilon^2y^4\sin^2\theta)\partial_y\psi, \\ { \cal w}&=&-\varepsilon^2 y^5\left[\sin2\theta \partial_\theta\psi -{\cal s}\left(\psi\right)\right].\end{aligned}\]] the regularity conditions on the event horizon ([eq : h - regularity-1]) and ([eq : h - regularity-2]) become @xmath116 the regularity conditions on the light surfaces ([eq : neumann]) become @xmath117 where @xmath118 here, we assume that @xmath119 can be written in the form of taylor series around the event horizon, @xmath120, as @xmath121 the coefficients @xmath122 are, in principle, determined by the gs equation ([gs]) with the regularity conditions ([hbo]), ([hbdo]), and ([ls]). using the expression ([taylor]), eqs. ([hbo]) and ([hbdo]) can be rewritten in the forms @xmath123 if we fix the functional form of @xmath124, we obtain @xmath125 and @xmath126 from the above equations, and further, we obtain @xmath122 of @xmath127 from eq. ([gs]) ; in order to get @xmath122 for @xmath128, we use an equation obtained by @xmath129-times differentiation of eq. ([gs]) with respect to @xmath130. for example, for @xmath131, by evaluating the gs equation ([gs]) on the event horizon @xmath120, we have @xmath132 \nonumber \\ & & = -2\cot\theta \left(\frac{d\psi^{(0)}}{d\theta}-\frac{{\cal s}(\psi^{(0)})}{\sin2\theta}\right) \left(\frac{1}{\sin^2\theta}+12\varepsilon^2\right) -\varepsilon^2\psi^{(1)}\left[2 -\frac{1}{2\sin^2\theta}\frac{d^2{\cal s}}{d\psi^2}(\psi^{(0)}){\psi^{(1)}}\right]. \label{gs-2 } \nonumber \\\end{aligned}\]] here, we should again note that @xmath124 can not be freely specified. the functional form of @xmath124 must be chosen such that the regularity condition ([ls]) on the inner light surface is satisfied. we consider slowly rotating magnetic fields, or in other words, we assume @xmath133. we rewrite the basic equations in the form of the power series with respect to @xmath134, and then we construct a solution of @xmath119 on the horizon, i.e., @xmath125, by perturbative procedures with respect to @xmath134. although, as mentioned, it seems to be impossible to determine the functional form of @xmath135 numerically, we can find it by this method. in order to construct a perturbative solution for @xmath122, we write @xmath136 further, we assume @xmath137 from eq. ([rls]), the location of the inner light surface is written as @xmath138 because @xmath139, we can express the quantities on the inner light surface by using the quantities on the event horizon. for example, @xmath140 at @xmath141 is written as @xmath142 since the main purpose of this study is to see the effect of an electric current on the configuration of the magnetic field on the event horizon, we focus on @xmath125. for this purpose, we rewrite eqs. ([hbd]) and ([gs-2]) in more appropriate forms as follows. by differentiating eq. ([hb]) with respect to @xmath4, we have @xmath143 from eq. ([hbd]), we have @xmath144 by subtracting eq. ([hb - diff]) from the above equation, we obtain @xmath145 it is easy to integrate the above equation, and we have @xmath146 where @xmath147 is an integration constant. in order to obtain @xmath126, we use eq. ([eq : psi1]) rather than eq. ([hbd]). by substituting eq. ([hb]) into eq. ([hb - diff]) and using eqs. ([cali - def]) and ([cals - def]), we have @xmath148 the above equation is equivalent to eq. ([horizon]). by substituting eq. ([horizon-2]) into the first term on the right - hand side of eq. ([gs-2]), we obtain @xmath149. \label{eq : psi0}\end{aligned}\]] we shall use the above equation rather than eq. ([gs-2]). here, we obtain the zeroth - order solutions for @xmath125. hereafter, the arguments of @xmath150 and @xmath151 are @xmath152 as long as we do not specify them. first of all, we write down the equations to obtain the zeroth - order solutions for @xmath125. from the lowest order of eqs. ([ls]), ([hb]), and ([eq : psi0]), we have @xmath153 where from eqs. ([cali - def]), ([cals - def]), and ([assumptions]), @xmath154 we can easily integrate eq. ([eq : psi00]) and obtain @xmath155 where @xmath156 is an integration constant. the above result implies that @xmath152 has only the monopole component. then, substituting eq. ([00-sol]) into eq. ([hb0]), we have @xmath157 where @xmath158 it is non - trivial whether the lowest order of the inner light surface regularity condition ([ls00]) is satisfied by @xmath152 and @xmath159 obtained above. from eqs. ([i1]) and ([s2]), we have @xmath160 it is easy to check that eqs. ([00-sol]) and ([s0-sol]) satisfy eq. ([ls00]). namely, we have obtained a small electric current which satisfies the lowest order of the inner light surface regularity condition. it is worthwhile to notice the meaning of the zeroth - order solutions for @xmath125, i.e., @xmath152. in the limit @xmath161, @xmath24 and @xmath40 become zero, whereas @xmath125 becomes @xmath152. since the case @xmath162 corresponds to the vacuum case, i.e., the case without an electric current, @xmath152 corresponds to the vacuum solution. as we showed in v, the vacuum solution has only the monopole component on the event horizon. ([00-sol]) is consistent with this fact. the small electric current @xmath163 can be regarded as a result of the slowly rotating monopole field @xmath152. next, we consider the correction of @xmath164 to the zeroth - order solution for @xmath125. hereafter, we assume @xmath165. in our perturbative method, the case of @xmath166 is quite different from the case @xmath167. if we choose @xmath166, we can obtain only the trivial solution @xmath168 using our perturbative method. we prove this statement in appendix [proof]. the equations determining @xmath169 are derived from eqs. ([ls]), ([hb]), and ([eq : psi0]) of @xmath164 : @xmath170 where @xmath171 we can see from eq. ([eq : psi01]) that @xmath169 is also the monopole solution. thus, the first - order correction merely adds a constant of @xmath164 to the integration constant of the zeroth - order solution. as a result, without loss of generality, we may assume for the first - order solutions that @xmath172 from eqs. ([hb1]), ([s3]), and the above equation, we have @xmath173 we should check the inner light surface regularity condition ([ls]). since @xmath174, eq. ([ls1]) becomes @xmath175 in order to estimate the above equation, we need @xmath176, which is the zeroth - order solution for @xmath126. we obtain @xmath176 from eq. ([eq : psi1]) of @xmath177 as @xmath178 where we assume that @xmath147 is the order of unity. substituting eq. ([00-sol]) into the above equation, we obtain @xmath179 by substituting eq. ([10-sol]) into eq. ([ls10]) and using the functional form of @xmath180 given by eq. ([s0-sol]), we can see that eq. ([ls10]) is satisfied. here it is worthwhile to notice that @xmath176, as well as @xmath152, necessarily corresponds to a vacuum solution. it is easy to check that eq. ([10-sol]) is consistent with eq. ([eq : non - monopole]). hereafter, we will make frequent use of eqs. ([ls00]) and ([ls10]) without giving an explicit reference. now, we consider the correction of @xmath181 to the zeroth - order solution for @xmath125. the equations determining @xmath182 can be obtained from eqs. ([ls]), ([hb]), and ([eq : psi0]) of @xmath181 : where @xmath184 in order to derive the above equations, we have used @xmath185 and @xmath186 obtained by substituting @xmath187 into eq. ([eq : psi1]). it should be noted that @xmath188 appears in eqs. ([ls02]) and ([eq : psi02]). in order to determine @xmath188, we use eq. ([horizon-2]) of @xmath181 : @xmath189=0.\]] by substituting eq. ([ls02]) into the above equation, we have @xmath190.\]] by subtracting the above equation from eq. ([eq : psi02]), we obtain the equation for @xmath188 as @xmath191 it is easily seen from eq. ([10-sol]) that @xmath192 is a particular solution for the above equation. thus, the general solution of the above equation is expressed by a linear combination of this particular solution and general solutions of the following homogeneous equation : @xmath193 the general solution of eq. ([homo - eq]) is given by @xmath194,\]] where @xmath195 and @xmath196 are arbitrary constants, and @xmath197 and @xmath198 are the associated legendre functions of the first and second kinds with @xmath199 and @xmath102, respectively. from the boundary condition at @xmath39, @xmath196 must vanish. hence, the most general solution of eq. ([20-eq]), which satisfies the boundary condition at @xmath39, is given by @xmath200 where we have used @xmath201, and @xmath202 is an arbitrary constant. here it is worthwhile to notice that the above result also corresponds to a vacuum solution. it is easy to check that eq. ([20-sol]), as well as @xmath152 and @xmath176, is consistent with eq. ([eq : non - monopole]). by using eq. ([s0-sol]) and substituting eq. ([10-sol]) into eq. ([eq : psi02]), we have @xmath203\sin^4\theta. \label{02-eq}\]] the above equation implies that, in general, the magnetic field on the event horizon includes non - monopole components of @xmath181. we can easily integrate this equation and obtain @xmath204, \label{02-sol}\]] where we have chosen the integration constant so that @xmath205, i.e., this correction consists of only non - monopole components. the functional form of @xmath206 is determined by using eq. ([hb2]) as @xmath207,\end{aligned}\]] where @xmath208 is defined by eq. ([xhat]). the solution with the corrections up to @xmath181 behaves near the inner light surface as @xmath209 although the solutions for @xmath210 for @xmath211 have been derived in the previous sections, we again show them with a slightly different parameterization : @xmath212\sin^2\theta, \\ \psi^{(1)}_0&=&c^{(0)}_0c'\sin^2\theta, \\ \psi^{(1)}_1&=&0, \\ \psi^{(2)}_0&=&\frac{1}{2}\left[3\left(c^{(0)}_0{c'}^2 -10c^{(0)}_2\right)\cos\theta+c^{(0)}_0c'\right]\sin^2\theta, \end{aligned}\]] where @xmath202 is given in this parameterization as @xmath213 by using the same parameterization as the above, the electric current is given by @xmath214,\end{aligned}\]] where @xmath208 is defined by eq. ([xhat]). we see that the arbitrary constants are only @xmath215 (@xmath216) and @xmath147. the reason this result takes this form is because if we choose @xmath119 and @xmath217 on the event horizon such that the regularity conditions ([hb]) and ([hbd]) are satisfied, then @xmath119 and @xmath135 are completely determined.
Summary and discussion
we studied a force - free magnetosphere in a static spherically symmetric black hole spacetime with a degenerate event horizon. we have found that if an electric current exists, higher multipole components of the magnetic field can be superposed upon the monopole component on the event horizon even if the two horizons degenerate into one horizon. this result is consistent with the numerical result given by komissarov and mckinney : they showed that the magnetic field lines of higher multipole components can penetrate an extreme kerr black hole if conductivity exists. the detailed geometrical structures of the extreme kerr black hole and the extreme reissner - nordstrm black hole are different from each other. however, since the degenerate structures of the horizons of these black holes are similar, the present results may be applicable to a certain extent for the extreme kerr black hole. if we require that there is no monopole component in the lowest - order configuration on the horizon, or equivalently, @xmath218, we obtain the trivial solution @xmath168, even though we take all - order corrections into account (see appendix [proof]). thus, the proposition in appendix d seems to imply that there is no non - trivial configuration without a monopole component on the event horizon of the extreme reissner - nordstrm black hole, even if an electric current exists. but this is not necessarily true. in order to see this fact, note that there is an exact monopole solution for the grad - shafranov equation ([eq : gs - eq]) : @xmath219 with the electric current @xmath220 where @xmath98 is an arbitrary constant. by contrast, the proposition in appendix d implies that, if @xmath218, there is no higher - order correction by which the configuration of the perturbative solution on the event horizon approaches to the monopole configuration in our perturbation scheme. in other words, our perturbative solution with vanishing @xmath152 can not approach to the above exact solution, even if we take into account all - order corrections. this fact suggests that even if an exact solution with a non - monopole configuration on the event horizon exists, the perturbative solution with vanishing @xmath152 can not approach to such a solution in our perturbation scheme. this possibility may arise from the assumption for the electric current ([assumptions]), which may be too strong, though the present analytic perturbation studies are impossible without this assumption. we would like to stress again that it is very difficult to obtain a stationary force - free magnetosphere by solving the grad - shafranov equation for the extreme black hole spacetime numerically. thus, we need to invoke analytic methods, as in the present study, or numerical techniques to follow the dynamical evolution of a force - free maxwell field until a stationary configuration is realized, as komissarov and mckinney used. as discussed in this paper, in the case that there are two light surfaces in the extreme reissner - nordstrm black hole spacetime, even though the magnetic field is regular, both on the event horizon and inner light surface, it will be singular on the outer light surface. if the angular velocity of the magnetic field is constant, two light surfaces necessarily exist. thus, if the dynamical evolution of a force - free maxwell field can be followed until it becomes stationary, then it is expected that the angular velocity decays far from the black hole so that the outer light surface does not exist. the extremity of charge or angular momentum changes the structure of boundary conditions for the grad - shafranov equation and seems to strongly affect global structures of the black hole magnetosphere. finally, we would like to suggest that analytic solutions obtained by this perturbation scheme becomes a benchmark for a numerical scheme to obtain solutions for stationary configurations of astrophysical magnetospheres, since our perturbation scheme is also suitable for non - extreme black hole cases. the authors would like to thank m. takahashi for useful lectures and discussions on black hole magnetospheres. is supported by a jsps grant - in - aid for creative scientific research no.
Derivation of grad-shafranov equation
a stationary, axisymmetric electromagnetic field implies @xmath221. then, from eq. ([eq : f - def]) and the force - free condition ([eq : ff - condition]), we have @xmath222. using these equations, the components of @xmath15 in the static coordinate system ([metric]) are written in the form @xmath223 where @xmath24 is defined by eq. ([eq : omegaf - def]), and @xmath224 is the determinant of the intrinsic metric of the spacelike hypersurface labeled by @xmath2. note that @xmath24 can be regarded as the angular velocity of the magnetic field. we consider an observer with an angular velocity @xmath225. his or her 4-velocity is given by @xmath226, where @xmath227 is a normalization factor. the electric field for this observer is given by @xmath228, and we can easily see from eq. ([eq : f - comp]) that @xmath229 vanishes. thus we may say that this observer is co - moving with the magnetic field, and the angular velocity of the magnetic field is @xmath24. substituting eq. ([eq : f - comp]) into the jacobi identity @xmath230}=0 $], we have @xmath231 the above equation implies that @xmath232, or equivalently, the equi-@xmath24 surface agrees with the equi-@xmath233 surface. thus we have @xmath234 using eq. ([eq : f - comp]), the maxwell equations imply the following equations : the @xmath2-component implies @xmath235 the @xmath5-component implies @xmath236 the @xmath3-component implies @xmath237 and the @xmath4-component implies @xmath238 where @xmath239 using eq. ([eq : f - comp]), the force - free condition implies @xmath240 substituting eq. ([eq : m - theta - comp]) to eq. ([eq : f - r - comp]), and substituting eq. ([eq : m - r - comp]) to eq. ([eq : f - theta - comp]), we have @xmath241 from the above equations, we have @xmath242 the above equations imply @xmath243 using the above equation, eqs. ([eq : f - r - comp - dash]) and ([eq : f - theta - comp - dash]) imply @xmath244 from eqs. ([eq : m - t - comp]) and ([eq : m - phi - comp]), we have @xmath245 \nonumber \\ & & = -4\pi\alpha\sqrt{\gamma } \left(j^\varphi - j^t\omega_{\rm f}\right), \label{eq : m - tphi - comp}\end{aligned}\]] where @xmath31 is defined by eq. ([eq : d - def]). noting that @xmath24 is a function of @xmath233 and substituting eq. ([eq : current]) into the right hand side of eq. ([eq : m - tphi - comp]), we have @xmath246 where @xmath247 and @xmath248 + \frac{r^2}{2}\frac{d{\cal b}^2}{da_\varphi}.\]] the above equation is called the grad - shafranov equation.
Electric current and magnetic flux
measured from the symmetry axis @xmath39 along the polar cap., scaledwidth=50.0%] here we introduce electric current @xmath40 and magnetic flux @xmath30 on a spacelike hypersurface labeled by @xmath2 which penetrate downward and upward an axisymmetric polar cap, respectively. these quantities were first introduced by macdonald and thorne @xcite and are related to @xmath249 and @xmath233 as follows. the polar cap is parameterized by @xmath250 and @xmath5, where @xmath250 is the proper length on the polar cap from @xmath39. the coordinates @xmath3 and @xmath4 on the polar cap are given as functions of @xmath250, i.e., @xmath251 and @xmath252 : by definition, @xmath253, and we assume that @xmath254 (see fig. [fg : polar - cap]). the orthonormal tangent vectors of the polar cap are @xmath255 then, the upward unit normal to the polar cap is @xmath256 we assume that the edge of the polar cap is @xmath257 and @xmath258. then, denoting the proper length @xmath250 at the edge by @xmath259, we have @xmath260d\ell \nonumber \\ & = & -\frac{1}{2}\int_0^{\ell_{\rm e}}\frac{d (\alpha b_\varphi)}{d\ell } d\ell = -\frac{1}{2}\alpha b_\varphi\biggl|_{(r,\theta)=(r_{\rm e},\theta_{\rm e }) }, \label{eq : i - def}\end{aligned}\]] where we have used eqs. ([eq : m - r - comp]) and ([eq : m - theta - comp]) in the second equality and assumed that @xmath261 from the regularity requirement. we can easily see from the above equation that the electric current through the polar cap is a function of the coordinate values of the edge, @xmath262. by a similar consideration to that for the electric current @xmath40, the magnetic flux @xmath30 can be written in the form @xmath263 d\ell = 2\pi\int_0^{\ell_{\rm e}}\frac{d a_\varphi}{d\ell } d\ell = 2\pi a_\varphi |_{(r,\theta)=(r_{\rm e},\theta_{\rm e }) }, \label{eq : psi - def}\end{aligned}\]] where @xmath264 (@xmath265) is the components of the skew tensor in the spacelike hypersurface labeled by @xmath2, and we have assumed @xmath266 from a regularity requirement. thus we have @xmath267 rewriting eq. ([eq : gs - equation]) using @xmath40 and @xmath30, the grad - shafranov equation eq. ([eq : gs - eq]) is obtained.
Relation between the kerr-schild and static coordinate systems
the line element of the reisner - nordstrm spacetime with the kerr - schild coordinate system @xmath42 is given by @xmath268 where @xmath269 and @xmath270. the relation between the kerr - schild coordinate system and the static one is given by @xmath271 from the above relation, we have @xmath272, @xmath273, and @xmath274, and @xmath275 using the above relations, we have @xmath276 by virtue of the stationary, axisymmetric nature of the electromagnetic field, we can easily see that the components of @xmath15 in the kerr - schild coordinate system are given as @xmath277. \label{eq : rt - comp } \\\end{aligned}\]] it should be noted that all the ordinary derivatives of the kerr - schild coordinates are equivalent to those of the static coordinates for the stationary axisymmetric field @xmath278, @xmath279 thus, even if the kerr - schild coordinate system is adopted, the equation for @xmath280 takes exactly the same form as eq. ([eq : gs - equation]).
Non-existence of non-monopole solution
* proposition * : within the perturbation scheme developed in this paper, the solution for @xmath125 with vanishing lowest - order solution @xmath152 is the only trivial solution @xmath168. 0.3 cm _ proof. _ we prove this by induction. we have already shown that, if we impose the non - existence of a monopole component for the lowest order of the perturbative solution, we obtain @xmath281 here, we assume that @xmath282 for @xmath283, or equivalently, @xmath284 then, we have @xmath285 where @xmath286 and we have used @xmath287, which is required from the regularity condition at @xmath39. substituting the above equation into eq. ([hb]), we obtain an equation of @xmath288, @xmath289 integrating the above equation, we have @xmath290^{-2\pi { \cal i}_0{}'},\]] where @xmath291 is an integration constant. the regularity condition implies @xmath292. hence, if @xmath291 does not vanish, @xmath293 must be satisfied, where @xmath294 is an arbitrary constant. in the neighborhood of @xmath295, @xmath296 and @xmath297. hence, if @xmath291 does not vanish, we have @xmath298 if we require the finiteness of @xmath299 at @xmath295, then @xmath291 must vanish, and, as a result, @xmath300 is obtained. q.e.d. r. d. blandford and r. l. znajek, mon. not. r. astron * 179 *, 433 (1977). j. c. mckinney and c. f. gammie, astrophys. j. * 611 *, 977 (2004). j. c. mckinney, astrophys. j. * 630 *, l5 (2005). j. f. hawley and j. h. krolik, astrophys. j. * 641 *, 103 (2006). j. bick and v. janis, mon. not. r. astron * 212 *, 899 (1985). s. s. komissarov and j. c. mckinney, mon. not. r. astron. lett. * 377 *, l49 (2007). j. bick, v. karas and t. ledvinka, astro - ph/0610841. d. macdonald and k. s. thorne, mon. not. r. astron. soc. * 198 *, 345 (1982). i. contopoulos, d. kazanas and c. fendt, astrophys. j. * 511 *, 351 (1999). j. ogura and y. kojima, prog.. phys. * 109 *, 619 (2003). d. a. uzdensky, astrophys. j. * 603 *, 652 (2004). d. a. uzdensky, astrophys. j. * 620 *, 889 (2005). a. gruzinov, phys.. lett. * 94 *, 021101 (2005). a. n. timokhin, mon. not. r. astron. soc. * 368 *, 1055 (2006). j. bick and l. dvork, phys. d * 22 *, 12 (1980). s. s. komissarov, mon. not. r. astron. soc. * 326 *, 41 (2001). s. nagataki, astrophys. j. * 704 *, 937 (2009). r. l. znajek, mon. not. r. astron. soc. * 179 *, 457 (1977). | it is known that the meissner - like effect is seen in a magnetosphere without an electric current in black hole spacetime : no non - monopole component of magnetic flux penetrates the event horizon if the black hole is extreme. in this paper, in order to see how an electric current affects the meissner - like effect, we study a force - free electromagnetic system in a static and spherically symmetric extreme black hole spacetime. by assuming that the rotational angular velocity of the magnetic field is very small, we construct a perturbative solution for the grad - shafranov equation, which is the basic equation to determine a stationary, axisymmetric electromagnetic field with a force - free electric current.
our perturbation analysis reveals that, if an electric current exists, higher multipole components may be superposed upon the monopole component on the event horizon, even if the black hole is extreme. | 1010.4104 |
Introduction
it is well known that quantum key distribution is one of the most interesting subjects in quantum information science, which was pioneered by c.bennett and g.brassard in 1984[1]. in the original paper of bennett, single photon communication was employed as implementation of quantum key distribution. however, despite that it is not essential in great idea of bennett, many researchers employed single photon communication scheme to realize bb84, b92[2]. because of the difficulties of single photon communication in practical sense, it was discussed whether one can realize a secure key distribution guaranteed by quantum nature based on light wave communication or not. in 1998, h.p.yuen and a.kim[3] proposed another scheme for key distribution based on communication theory(signal detection theory). this scheme corresponds to an implementation of secret key sharing which was information theoretically predicted by maurer[4], et al. however, yuen s idea was found independently from maurer s discussion. in the first paper of yuen - kim[3], they showed that if noises of eve(eavesdropper) and bob(receiver) are statistically independent, secure key distribution can be realized even if they are classical noises, in which they employed a modification of b92 protocol[2]. following yk s first paper, a simple experimental demonstration of yk protocol based on classical noise was reported[5], and recently yk scheme with 1 gbps and 10 km long fiber system based on quantum shot noise was demonstrated[6]. however, these schemes are not unconditional secure. that is, ability of signal detection of eve can be superior to that of bob. as a result, an interesting question arises `` is it possible to create a system with current technology that could provide a communication in which always bob s error probability is superior to that of eve? '' in proceedings paper of qcm and c 2002, yuen and his coworker reported that yk protocol can be unconditional secure, even if one uses conventional optical communication system[7]. this is interesting result for engineer, and will open a new trend of quantum cryptography. in this report, we simulate practical feature of yuen - kim protocol for quantum key distribution with unconditional secure, and propose a scheme to implement them using our former experimental setup[6].
Yuen-kim protocol
a fundamental concept of yuen - kim protocol follows the next remark. + * remark * : _ if there are statistically independent noises between eve and bob, there exist a secure key distribution based on communication. _ + they emphasized that the essential point of security of the key distribution is detectability of signals. this is quite different with the principle of bb-84, et al which are followed by no cloning theorem. that is, bb-84 and others employ a principle of disturbance of quantum states to give a guarantee of security, but yk protocol employs a principle of communication theory. it was clarified that this scheme can be realized as a modification of b-92. however, this scheme allows us use of classical noise, and it can not provide unconditional secure. then, yuen and his coworker showed that yk scheme is to be unconditional secure in which a fundamental theorem in quantum detection theory was used for his proof of security as follows. + * theorem * : (helstrom - holevo - yuen) + _ signals with non commuting density operators can not be distinguished without error. _ + so if we assign non commuting density operators for bit signals 1 and 0, then one can not distinguish without error. when the error is 1/2 based on quantum noise, there is no way to distinguish them. so we would like to make such a situation on process between alice and eve. to do so, a new version of yk scheme was given as follows : _ _ * the sender(alice) uses an explicit key(a short key:@xmath0, expanded into a long key:@xmath1 by use of a stream cipher) to modulate the parameters of a multimode coherent state. * state @xmath2 is prepared. bit encoding can be represented as follows : @xmath3 where @xmath4. * alice uses the running key @xmath1 to specify a basis from a set of m uniformly distributed two - mode coherent state. * the message @xmath5 is encoded as @xmath6. this mapping of the stream of bits is the key to be shared by alice and bob. because of his knowledge @xmath1, bob can demodulate from @xmath6 to @xmath5. here, let us introduce the original discussion on the security. the ciphering angle @xmath7 could have @xmath8 in general as discrete or continuous variable determined by distribution of keys. a ciphered two mode state may be @xmath9 the corresponding density operator for all possible choices of @xmath8 is @xmath10, where @xmath11 or @xmath12. the problem is to find the minimum error probability that eve can achieve in bit determination. to find the optimum detection process for discrimination between @xmath13 and @xmath14 is the problem of quantum detection theory. the solution is given by[8] @xmath15 as an example of encoding to create @xmath16 which is the error probability of eve, yuen et al suggested certain modulation scheme. in that case, closest values of a given @xmath8 can be associated with distinct bits from the bit at position @xmath8, and two closest neighboring states represent distinct bits which means a set of base state. in this scheme, they assumed that one chooses a set of basis state(keying state for 1 and 0) for bits without overlap. the error probability for density operators @xmath13 and @xmath14 becomes 1/2, when number of a set of basis state increases. asymptotic property of the error probability depends on the amplitude of coherent state[7][9].
Simplified modulation scheme in yk protocol
original scheme of yk protocol in the above can be realized by practical devices. to apply them to fiber communication system, we would like to realize them by intensity modulation / direct detection scheme. if one does not want to get perfect yk scheme, one can more simplify the implementation of yk protocol. from a fundamental principle in quantum detection theory, we can construct non - commuting density operators from sets based on non - orthogonal states when one does not allow overlap of the selection of a set of basis state for 1 and 0. on the other hand, when we allow overlap for selection of a set of basis state, one can use orthogonal state to construct the same density operators for 1 and 0. that is, @xmath17. however, in this case, unknown factor for eve is only an initial short key, and a stream of bits that eve observed is perfectly the same as those of alice and bob, though eve can not estimate the bits at that time. this gives still insecure situation. so, here, we employ a combination of non - orthogonality and overlap selection in order to reduce the number of basis sets. let us assume that the maximum amplitude is fixed as @xmath18. we divide it into 2 m. so we have m sets of basis state@xmath19. total set of basis state is given as shown in fig.1. each set of basis state is used for @xmath20, and @xmath21, depending on initial keys. @xmath22 so the density operators for 1 and 0 for eve are @xmath23 for the sets of @xmath24, @xmath25,@xmath26, let us assign 0 and 1 by the same way as eqs(4),(5). in this case, eve can not get key information, but she can try to know the information of quantum states used for bit transmission. so this is the problem for discrimination of 2 m pure states. the error probability is given by @xmath27 although we have many results for calculation of optimum detection problems[10][11][12], to solve this problem is still difficult at present time, because the set of states does not have complete symmetric structure. so we here give the lower bound and tight upper bound. the lower bound is given by the minimum error probability:@xmath28 for signal set @xmath29 which are neighboring states. it is given as follows : @xmath30 } \right)\]] the upper bound is given by applying square root measurement for 2 m pure states. the numerical properties are shown in fig.2-(a). thus if m increases, then her error for information on quantum states increases. in this case, pure guessing corresponds to @xmath31. the error probability of bob, however, is independent of the number of set of basis state, and it is given as follows : @xmath32 we emphasize that eve can not get key information in this stage, because the information for 1 and 0 are modulated by the way of eqs(4),(5). furthermore, this scheme can send 2 m bits by m sets of basis state. let us apply the original scheme such that m bits are sent by m sets of basis state. in this case, eve will try to get key information, so the density operators for eve become mixed states @xmath13, @xmath14 consisting of set of states which send 1 and 0, respectively. the numerical properties are shown in fig.2-(b). both schemes have almost same security, but the latter can only send m bits by m sets of basis state. in other word, the number of sets is reduced to 1/2 in the former scheme.
Primary design of experimental set
in implementing yk protocol by conventional fiber communication system, we use here our proposed system. figure 3 shows the experimental setup. the laser diode serves as 1.3@xmath33 m light source. a pattern generator provides a signal pulse string to send keys. a modulator which selects basis state follows a driver of laser diode. the selector gives selection of amplitude and assignment of 1 and 0, and is controlled by initial keys. the laser driver is driven by output signals of modulator. the optical divider corresponds to eve. the case 1 is a type of opaque ", and the case 2 is a type of translucent ". the channel consists of 10 km fiber and att. we can change the distance equivalently from 10 km to 200 km by att. the speed of pulse generator to drive laser diode is 311mbps, 622mbps, and 1.2gbps. the detector of bob is ingaas pin photo loaded by 50@xmath34 register, and it is connected to an error probability counter which can apply to 12 gbps. the dark current is 7@xmath35 and the minimum received power of our system is about -30 dbm. in this system, the problem for degree of security is only power advantage of eve which will be set in near transmitter(alice). when the eavesdropping is opaque, the error probability of bob increases drastically, and the error probability counter shows almost 1/2, which means that the error of eve is also 1/2. in this case, problem of communication distance is not so important. we can detect the existence of eve in any distance of channel. when the eavesdropping is translucent, eve has to take only few power(@xmath36 from the main stream of bits sequence in order to avoid the power level disturbance. in this case, the error of bob does not increase. as a result, alice and bob can not detect the existence of eve. the secure communication distance depends on the error probabilities of bob and eve. let @xmath37 be transparency of channel from alice to bob. the detectability for bob in this experiment setup depends on the signal distance(amplitude difference between two states as basis state) : @xmath38 for @xmath39, and that of eve depends on the signal distance : @xmath40. here we assume that @xmath41, and the total loss is 20db which corresponds to 100 km. since our receiver requires about -30 dbm, the transmitter is -10 dbm. when m increases, sufficiently the error of eve increases.
Conclusions
we examined a simulation of yk protocol based on intensity modulation / direct detection fiber communication system, and showed a design of implementation of secure system based on our experimental setup which was used to demonstrate the first version of implementation of yk protocol. we will soon report complete demonstration in experiment by the above system. | in this report, we simulate practical feature of yuen - kim protocol for quantum key distribution with unconditional secure. in order to demonstrate them experimentally by intensity modulation / direct detection(imdd) optical fiber communication system, we use simplified encoding scheme to guarantee security for key information(1 or 0). that is, pairwise m - ary intensity modulation scheme is employed.
furthermore, we give an experimental implementation of yk protocol based on imdd. | quant-ph0212050 |
Acknowledgements
b. s., v. s., a. b., and a. b. acknowledge support by the ministry of education, science, and technological development of the republic of serbia under projects on171017, iii43007, on171009, on174027 and ibec, and by daad - german academic and exchange service under project ibec. m. acknowledges support by the science and engineering research board, department of science and technology, government of india under project no. s. k. a. acknowledges support by the cnpq of brazil under project 303280/2014 - 0, and by the fapesp of brazil under project 2012/00451 - 0. numerical simulations were run on the paradox supercomputing facility at the scientific computing laboratory of the institute of physics belgrade, supported in part by the ministry of education, science, and technological development of the republic of serbia under project on171017. d. vudragovi, i. vidanovi, a. bala, p. muruganandam, s. k. adhikari, c programs for solving the time - dependent gross - pitaevskii equation in a fully anisotropic trap, comput.. commun. * 183 * (2012) 2021. r. k. kumar and p. muruganandam, j. phys. b : at. mol. * 45 * (2012) 215301 ; + l. e. young - s. and s. k. adhikari, phys. rev. a * 86 * (2012) 063611 ; + s. k. adhikari, j. phys. b : at. mol. opt. phys. * 45 * (2012) 235303 ; + i. vidanovi, n. j. van druten, and m. haque, new j. phys. * 15 * (2013) 035008 ; + s. balasubramanian, r. ramaswamy, and a. i. nicolin, rom. * 65 * (2013) 820 ; + l. e. young - s. and s. k. adhikari, phys. rev. a * 87 * (2013) 013618 ; + h. al - jibbouri, i. vidanovic, a. balaz, and a. pelster, j. phys. b : at. mol. opt. phys. * 46 * (2013) 065303 ; + x. antoine, w. bao, and c. besse, comput. * 184 * (2013) 2621 ; + b. nikoli, a. bala, and a. pelster, phys. a * 88 * (2013) 013624 ; + h. al - jibbouri and a. pelster, phys. a * 88 * (2013) 033621 ; + s. k. adhikari, phys. rev. a * 88 * (2013) 043603 ; + j. b. sudharsan, r. radha, and p. muruganandam, j. phys. b : at. mol. * 46 * (2013) 155302 ; + r. r. sakhel, a. r. sakhel, and h. b. ghassib, j. low temp * 173 * (2013) 177 ; + e. j. m. madarassy and v. t. toth, comput. * 184 * (2013) 1339 ; + r. k. kumar, p. muruganandam, and b. a. malomed, j. phys. b : at. mol. * 46 * (2013) 175302 ; + w. bao, q. tang, and z. xu, j. comput * 235 * (2013) 423 ; + a. i. nicolin, proc. ser. a - math * 14 * (2013) 35 ; + r. m. caplan, comput. * 184 * (2013) 1250 ; + s. k. adhikari, j. phys. b : at. mol. * 46 * (2013) 115301 ; + . marojevi, e. gkl, and c. lmmerzahl, comput. * 184 * (2013) 1920 ; + x. antoine and r. duboscq, comput. * 185 * (2014) 2969 ; + s. k. adhikari and l. e. young - s, j. phys. b : at. mol. * 47 * (2014) 015302 ; + k. manikandan, p. muruganandam, m. senthilvelan, and m. lakshmanan, phys. e * 90 * (2014) 062905 ; + s. k. adhikari, phys. a * 90 * (2014) 055601 ; + a. bala, r. paun, a. i. nicolin, s. balasubramanian, and r. ramaswamy, phys. rev. a * 89 * (2014) 023609 ; + s. k. adhikari, phys. rev. a * 89 * (2014) 013630 ; + j. luo, commun. nonlinear sci. * 19 * (2014) 3591 ; + s. k. adhikari, phys. rev. a * 89 * (2014) 043609 ; + k.- t. xi, j. li, and d.- shi, physica b * 436 * (2014) 149 ; + m. c. raportaru, j. jovanovski, b. jakimovski, d. jakimovski, and a. mishev, rom. j. phys. * 59 * (2014) 677 ; + s. gautam and s. k. adhikari, phys. a * 90 * (2014) 043619 ; + a. i. nicolin, a. bala, j. b. sudharsan, and r. radha, rom. * 59 * (2014) 204 ; + k. sakkaravarthi, t. kanna, m. vijayajayanthi, and m. lakshmanan, phys. e * 90 * (2014) 052912 ; + s. k. adhikari, j. phys. * 47 * (2014) 225304 ; + r. k. kumar and p. muruganandam, numerical studies on vortices in rotating dipolar bose - einstein condensates, proceedings of the 22nd international laser physics workshop, j. phys. conf. ser. * 497 * (2014) 012036 ; + a. i. nicolin and i. rata, density waves in dipolar bose - einstein condensates by means of symbolic computations, high - performance computing infrastructure for south east europe s research communities : results of the hp - see user forum 2012, in springer series : modeling and optimization in science and technologies * 2 * (2014) 15 ; + s. k. adhikari, phys. rev. a * 89 * (2014) 043615 ; + r. k. kumar and p. muruganandam, eur. j. d * 68 * (2014) 289 ; + j. b. sudharsan, r. radha, h. fabrelli, a. gammal, and b. a. malomed, phys. a * 92 * (2015) 053601 ; + s. k. adhikari, j. phys. b : at. mol. opt. phys. * 48 * (2015) 165303 ; + f. i. moxley iii, t. byrnes, b. ma, y. yan, and w. dai, j. comput. * 282 * (2015) 303 ; + s. k. adhikari, phys. e * 92 * (2015) 042926 ; + r. r. sakhel, a. r. sakhel, and h. b. ghassib, physica b * 478 * (2015) 68 ; + s. gautam and s. k. adhikari, phys. rev. a * 92 * (2015) 023616 ; + d. novoa, d. tommasini, and j. a. nvoa - lpez, phys. e * 91 * (2015) 012904 ; + s. gautam and s. k. adhikari, laser phys. * 12 * (2015) 045501 ; + k.- t. xi, j. li, and d.- shi, physica b * 459 * (2015) 6 ; + r. k. kumar, l. e. young - s., d. vudragovi, a. bala, p. muruganandam, and s. k. adhikari, comput.. commun. * 195 * (2015) 117 ; + s. gautam and s. k. adhikari, phys. rev. a * 91 * (2015) 013624 ; + a. i. nicolin, m. c. raportaru, and a. bala, rom. * 67 * (2015) 143 ; + s. gautam and s. k. adhikari, phys. rev. a * 91 * (2015) 063617 ; + e. j. m. madarassy and v. t. toth, phys. d * 91 * (2015) 044041. | we present hybrid openmp / mpi (open multi - processing / message passing interface) parallelized versions of earlier published c programs (d. vudragovi et al. (2012) @xcite) for calculating both stationary and non - stationary solutions of the time - dependent gross - pitaevskii (gp) equation in three spatial dimensions.
the gp equation describes the properties of dilute bose - einstein condensates at ultra - cold temperatures.
hybrid versions of programs use the same algorithms as the c ones, involving real- and imaginary - time propagation based on a split - step crank - nicolson method, but consider only a fully - anisotropic three - dimensional gp equation, where algorithmic complexity for large grid sizes necessitates parallelization in order to reduce execution time and/or memory requirements per node. since
distributed memory approach is required to address the latter, we combine mpi programing paradigm with existing openmp codes, thus creating fully flexible parallelism within a combined distributed / shared memory model, suitable for different modern computer architectures.
the two presented c / openmp / mpi programs for real- and imaginary - time propagation are optimized and accompanied by a customizable makefile.
we present typical scalability results for the provided openmp / mpi codes and demonstrate almost linear speedup until inter - process communication time starts to dominate over calculation time per iteration.
such a scalability study is necessary for large grid sizes in order to determine optimal number of mpi nodes and openmp threads per node.
bose - einstein condensate ; gross - pitaevskii equation ; split - step crank - nicolson scheme ; real- and imaginary - time propagation ; c program ; mpi ; openmp ; partial differential equation 02.60.lj ; 02.60.jh ; 02.60.cb ; 03.75.-b * new version program summary * + _ program title : _ gp - scl - hyb package, consisting of : (i) imagtime3d - hyb, (ii) realtime3d - hyb.
+ _ catalogue identifier : _
aedu_v3_0 + _ program summary url : _
http://cpc.cs.qub.ac.uk/summaries/aedu_v3_0.html + _ program obtainable from : _ cpc program library, queen s university of belfast, n. ireland.
+ _ licensing provisions : _ apache license 2.0 + _ no. of lines in distributed program, including test data, etc. : _ 26397.
+ _ no. of bytes in distributed program, including test data, etc. : _ 161195.
+ _ distribution format : _
tar.gz.
+ _ programming language : _ c / openmp / mpi.
+ _ computer : _ any modern computer with c language, openmp- and mpi - capable compiler installed.
+ _ operating system : _ linux, unix, mac os x, windows.
+ _ ram : _ total memory required to run programs with the supplied input files, distributed over the used mpi nodes : (i) 310 mb, (ii) 400 mb.
larger grid sizes require more memory, which scales with nx*ny*nz.
+ _ number of processors used : _ no limit, from one to all available cpu cores can used on all mpi nodes.
+ _ number of nodes used : _ no limit on the number of mpi nodes that can be used.
depending on the grid size of the physical problem and communication overheads, optimal number of mpi nodes and threads per node can be determined by a scalability study for a given hardware platform.
+ _ classification : _ 2.9, 4.3, 4.12. + _ catalogue identifier of previous version : _ aedu_v2_0.
+ _ journal reference of previous version : _ comput.
phys.
commun. 183 (2012) 2021.
+ _ does the new version supersede the previous version? : _ no.
+ _ nature of problem : _ these programs are designed to solve the time - dependent gross - pitaevskii (gp) nonlinear partial differential equation in three spatial dimensions in a fully anisotropic trap using a hybrid openmp / mpi parallelization approach. the gp equation describes the properties of a dilute trapped bose - einstein condensate.
+ _ solution method : _ the time - dependent gp equation is solved by the split - step crank - nicolson method using discretization in space and time
. the discretized equation is then solved by propagation, in either imaginary or real time, over small time steps.
the method yields solutions of stationary and/or non - stationary problems.
+ _ reasons for the new version : _ previous c @xcite and fortran @xcite programs are widely used within the ultracold atoms and nonlinear optics communities, as well as in various other fields @xcite.
this new version represents extension of the two previously openmp - parallelized programs (imagtime3d - th and realtime3d - th) for propagation in imaginary and real time in three spatial dimensions to a hybrid, fully distributed openmp / mpi programs (imagtime3d - hyb and realtime3d - hyb).
hybrid extensions of previous openmp codes enable interested researchers to numerically study bose - einstein condensates in much greater detail (i.e., with much finer resolution) than with openmp codes. in openmp (threaded) versions of programs, numbers of discretization points in x, y, and z
directions are bound by the total amount of available memory on a single computing node where the code is being executed.
new, hybrid versions of programs are not limited in this way, as large numbers of grid points in each spatial direction can be evenly distributed among the nodes of a cluster, effectively distributing required memory over many mpi nodes.
this is the first reason for development of hybrid versions of 3d codes.
the second reason for new versions is speedup in the execution of numerical simulations that can be gained by using multiple computing nodes with openmp / mpi codes.
+ _ summary of revisions : _ two c / openmp programs in three spatial dimensions from previous version @xcite of the codes (imagtime3d - th and realtime3d - th) are transofrmed and rewritten into a hybrid openmp / mpi programs and named imagtime3d - hyb and realtime3d - hyb.
the overall structure of two programs is identical.
the directory structure of the gp - scl - hyb package is extended compared to the previous version and now contains a folder scripts, where examples of scripts that can be used to run the programs on a typical mpi cluster are given.
the corresponding readme.txt file contains more details.
we have also included a makefile with tested and verified settings for most popular mpi compliers, including openmpi (open message passing interface) @xcite and mpich (message passing interface chameleon) @xcite.
transformation from pure openmp to a hybrid openmp / mpi approach has required that the array containing condensate wavefunction is distributed among mpi nodes of a computer cluster.
several data distribution models have been considered for this purpose, including block distribution and block cyclic distribution of data in a 2d matrix.
finally, we decided to distribute the wavefunction values across different nodes so that each node contains only one slice of the x - dimension data, while containing the complete corresponding y- and z - dimension data, as illustrated in fig.
[fig1].
this allows central functions of our numerical algorithm, calcluy, calcuz, and calcnu to be executed purely in parallel on different mpi nodes of a cluster, without any overhead or communication, as nodes contain all the information for y- and z - dimension data in the given x - sub - domain.
however, the problem arises when functions calclux, calcrms, and calcmuen need to be executed, as they also operate on the whole x - dimension data.
thus, the need for additional communication arises during the execution of the function calcrms, while in the case of fuctions calclux and calcmuen also the transposition of data between x- and y - dimensions is necessary, while data in z dimension have to stay contiguous.
transposition provides nodes with all the necessary x - dimension data to execute functions calclux and calcmuen.
however, this needs to be done in each iteration of numerical algorithm, thus necessarily increasing communication overhead of the simulation.
transposition algorithms that were considered where the ones that account for greatest common divisor (gcd) between number of nodes in columns (designated by n) and rows (designated by m) of a cluster configured as 2d mash of nodes @xcite.
two of such algorithms have been tested and tried for implementation : the case when gcd = 1 and the case when gcd > 1.
the trivial situation n = m = 1 is already covered by the previous, purely openmp programs, and therefore, without any loss of generality, we have considered only configurations with number of nodes in x - dimension satisfying n > 1. only the former algorithm (gcd = 1) was found to be sound in case where data matrix is not a 2d, but a 3d structure.
latter case was found to be too demanding implementation - wise, since mpi functions and data - types are bound to certain limitations.
therefore, the algorithm with m = 1 nodes in y - dimension was implemented, as depicted by the wavefunction data structure in fig.
[fig1].
implementation of the algorithm relies on a sliced distribution of data among the nodes, as explained in fig.
[fig2].
this successfully solves the problem of large ram consumption of 3d codes, which arises even for moderate grid sizes.
however, it does not solve the question of data transposition between the nodes. in order to implement the most effective (gcd = 1) transposition algorithm according to ref.
@xcite, we had to carry out block distribution of data within one data slice contained on a single node.
this block distribution of data was done implicitly, i.e., data on one node have been put in a single 1d array (psi) of contiguous memory, in which z - dimension has stride 1, y - dimension has stride nz, and x - dimension has stride ny*nz.
this is different from previous implementation of the programs, where the wavefunction was represented by an explicit 3d array.
this change was also introduced in order to more easily form user mpi datatypes, which allow for implicit block distribution of data, and represent 3d blocks of data within 1d data array.
these blocks are then swapped between nodes, effectively performing the transposition in x - y and y - x directions. together with transposition of blocks between the nodes
, the block data also have to be redistributed. to illustrate how this works,
let us consider example shown in fig.
[fig1](a), where one data block has size (nx / gisze)*(ny / gsize)*nz.
it represents one 3d data block, swapped between two nodes of a cluster (through one non - blocking mpi_isend and one mpi_ireceive operation), containing (nx / gsize)*(ny / gsize) 1d rods of contiguous nz data.
these rods themselves need to be transposed within the transposed block as well.
this means that two levels of transpositions need to be performed. at a single block level
, rods have to be transposed (as indicated in upper left corner of fig.
[fig1](a) for sending index type and in fig.
[fig1](b) for receiving index type).
second level is transposition of blocks between different nodes, which is depicted by blue arrows connecting different blocks in fig.
[fig1].
the above described transposition is applied whenever needed in the functions calclux and calcmuen, which require calculations to be done on the whole range of data in x - dimension. when performing renormalization of the wavefunction or calculation of its norm, root - mean - square radius, chemical potential, and energy, collective operations mpi_gather and mpi_bcast are also used.
figures [fig3] and [fig4] show the scalability results obtained for hybrid versions of programs for small and large grid sizes as a function of number of mpi nodes used.
the baseline for calculation of speedups in the execution time for small grid sizes are previous, purely openmp programs, while for large grid sizes, which can not fit onto a single node, the baseline are hybrid programs with minimal configuration runs on 8 nodes.
the figures also show efficacies, defined as percentages of measured speedups compared to the ideal ones.
we see that an excellent scalability (larger than 80% compared to the ideal one) can be obtained for up to 32 nodes.
the tests have been performed on a cluster with nodes containing 2 x 8-core sandy bridge xeon 2.6 ghz processors with 32 gb of ram and infiniband qdr (quad data rate, 40 gbps) interconnect.
we stress that the scalability depends greatly on the ratio between the calculation and communication time per iteration, and has to be studied for a particular type of processors and interconnect technology.
+ _ additional comments : _ this package consists of 2 programs, see program title above. both are hybrid, threaded and distributed (openmp / mpi parallelized). for the particular purpose of each program, see descriptions below. + _ running time : _ all running times given in descriptions below refer to programs compiled with openmpi / gcc compiler and executed on 8 to 32 nodes with 2 x 8-core sandy bridge xeon 2.6 ghz processors with 32 gb of ram and infiniband qdr interconnect. with the supplied input files for small grid sizes, running wallclock times of several minutes
are required on 8 to 10 mpi nodes.
+ _ special features : _ (1) since the condensate wavefunction data are distributed among the mpi nodes, when writing wavefunction output files each mpi process saves its data into a separate file, to avoid i / o issues. concatenating
the corresponding files from all mpi processes will created the complete wavefunction file.
(2) due to a known bug in openmpi up to version 1.8.4, allocation of memory for indexed datatype on a single node for large grids (such as 800x640x480) may fail.
the fix for this bug is already in 3c489ea branch and is fixed in openmpi as of version 1.8.5.
+ program summary (i) + _ program title : _
imagtime3d - hyb.
+ _ title of electronic files : _ imagtime3d - hyb.c, imagtime3d - hyb.h.
+ _ computer : _ any modern computer with c language, openmp- and mpi - capable compiler installed. + _ ram memory requirements : _ 300 mbytes of ram for a small grid size 240x200x160, and scales with nx*ny*nz.
this is total amount of memory needed, and is distributed over mpi nodes used for execution.
+ _ programming language used : _ c / openmp / mpi.
+ _ typical running time : _ few minutes with the supplied input files for a small grid size 240x200x160 on 8 nodes.
up to one hour for a large grid size 1920x1600x1280 on 32 nodes (1000 iterations).
+ _ nature of physical problem : _ this program is designed to solve the time - dependent gp nonlinear partial differential equation in three space dimensions with an anisotropic trap.
the gp equation describes the properties of a dilute trapped bose - einstein condensate.
+ _ method of solution : _ the time - dependent gp equation is solved by the split - step crank - nicolson method by discretizing in space and time. the discretized equation is then solved by propagation in imaginary time over small time steps.
the method yields solutions of stationary problems.
+ + program summary (ii) + _ program title : _
realtime3d - hyb.
+ _ title of electronic files : _ realtime3d - hyb.c, realtime3d - hyb.h.
+ _ computer : _ any modern computer with c language, openmp- and mpi - capable compiler installed. + _ ram memory requirements : _ 410 mbytes of ram for a small grid size 200x160x120, and scales with nx*ny*nz.
this is total amount of memory needed, and is distributed over mpi nodes used for execution.
+ _ programming language used : _ c / openmp / mpi.
+ _ typical running time : _ 10 - 15 minutes with the supplied input files for a small grid size 200x160x120 on 10 nodes.
up to one hour for a large grid size 1600x1280x960 on 32 nodes (1000 iterations).
+ _ nature of physical problem : _ this program is designed to solve the time - dependent gp nonlinear partial differential equation in three space dimensions with an anisotropic trap.
the gp equation describes the properties of a dilute trapped bose - einstein condensate.
+ _ method of solution : _ the time - dependent gp equation is solved by the split - step crank - nicolson method by discretizing in space and time. the discretized equation is then solved by propagation in real time over small time steps.
the method yields solutions of stationary and non - stationary problems. | 1601.04641 |
[sec:intro] introduction
the equilibrium behavior and the equipartition of energy between various degrees of freedom in nonlinear, nonintegrable discrete systems has attracted considerable interest since the seminal study of fermi, pasta and ulam @xcite. in hamiltonain systems with conserved number of excitations (waves) the maximum entropy principle suggests that in the final state of thermal equilibrium the statistics of the system is given by grand canonical gibbs distribution with the effective `` temperature '' and `` chemical potential '' @xcite. however unlike the conventional statistical mechanics the effective temperature of this grand - canonical distribution depends on the initial position in the phase space and for certain regions can become negative making the distribution non - normalizable. such regime is commonly attributed to the emergence of stable, localized, nonlinear structures corresponding to solitons in continuous systems @xcite and discrete breathers @xcite in discrete systems. from the point of view of the wave turbulence theory @xcite the resulting equilibrium distribution provides stationary rayleigh - jeans spectra @xcite. also thermalization of light in nonlinear multimode waveguides and cavities has recently attracted attention in the context of classical optical wave condensation @xcite. here we will study the phenomenon of thermalization in the context of light propagation in a system of coupled nonlinear optical waveguides but the results can have wider applicability beyond the scope of the nonlinear optics. when the individual waveguide modes are strongly localized the nonlinear propagation of light is most commonly modelled by the discrete nonlinear schrodinger equation (dnlse) @xcite. in fact most studies of thermalization in nonlinear _ discrete _ systems have concentrated on dnlse in one @xcite or two @xcite dimensions. thanks the plethora of the results in the field of `` dnlse thermalization '' the structure of the final equilibrium state and the thermodynamical conditions for the occurrence of discrete breathers are now well understood. among numerous discoveries in this area we would like to point reader s attention to the universal correlations in 1d systems of optical waveguides predicted in @xcite in the limit when the nonlinearity dominates over the linear coupling. in this limit the effective dimensionless temperature turns out to be a universal constant independent on system parameters (provided that the initial state is characterised by uniform intensities) and the same universality is also manifested in the shape of the field correlation function. in this paper we would like to focus on a much less studied model, namely, the thermalization of two coupled fields in the presence of the four - wave mixing (fwm) @xcite. in the context of nonlinear optics the situation corresponds to the propagation of polarized light in the birefringent material @xcite or mode interaction from different floquet - bloch bands @xcite. here we will concentrate on the first case, however the results presented here are quite general and can be applied to other nolinearly coupled systems. in order to give reference to the real - world units we use algaas as an common example of a material with cubic symmetry and fused silica as the corresponding example of isotropic crystal.
The model
the wave dynamics of the two orthogonally polarized fields is given by the following pair of coupled equations @xcite : [vector - dnlse] @xmath1 in the above equations, @xmath2 and @xmath3 are slowly varying filed envelopes of the te and tm polarized waves, @xmath4 is the polarization mode dispersion constant, @xmath5 is the vacuum wave vector, @xmath6 is the linear birefringence (@xmath7 for algaas), @xmath8 is the coupling constant (@xmath9 mm@xmath10), @xmath11 is the nonlinear coefficient, @xmath12 is the kerr coefficient (@xmath13 @xmath14/w for algaas), @xmath15 is the linear refractive index (@xmath16 for algaas). the dimensionless constants @xmath17 and @xmath18 represent the relative strength of self- and cross - phase modulation (spm and xpm). if one puts @xmath19 the system ([vector - dnlse]) breaks into two independent scalar dnlse equations. we can restrict ourselves to the case of positive coupling @xmath20 since the case of negative coupling can be recovered via a standard staggered transformation @xmath21, @xmath22. the change of sign in the nonlinearity can be also be compensated via more complicated transformation which involves staggering, complex conjugation and swapping : @xmath23, @xmath24. both transformations only affect the field correlation functions and phase distributions (and not e.g. the intensity distributions) in a controlled way and here without loss of generality we will also restrict ourselves to the case of positive nonlinearity. in this paper we assume periodic boundary conditions although in the thermodynamic limit @xmath25 this choice is not essential. note in passing that continuous analogues of system ([vector - dnlse]) were studied in @xcite with regards to pulse propagation in optical fibers. in any chosen nonlinear medium the dimensionless xpm and fwm constants, @xmath17 and @xmath18, are not independent and 3 possible cases of interest can be envisaged @xcite : 1. anisotropic cubic medium (e.g. algaas) : @xmath26. generic isotropic medium : @xmath27. 3. isotropic cubic medium (e.g. fused silica) : @xmath28, @xmath29. we will refer to cases (a) and (b),(c) as isotropic and anisotropic correspondingly. the system is hamiltonian with the hamiltonian : @xmath30 which is a natural conserved quantity in the system while the additional integral of motion is provided by the total pulse power (proportional to the sum of local intensities) @xmath31 additionally in the absence of the four - wave mixing (@xmath32) the individual powers in each polarization @xmath33 and @xmath34 are conserved. in the state of thermodynamic equilibrium the stationary field distribution @xmath35 maximizes the entropy @xmath36. however the nonlinear evolution always takes place on the shell @xmath37 and @xmath38 which introduces two constraints for the optimization problem. the solution then represents a grand canonical gibbs distribution (see e.g. @xcite) : @xmath39\]] where the lagrange multipliers @xmath40 and @xmath41 play the roles of the `` inverse temperature '' and `` chemical potential '' respectively while the normalizing factor @xmath42 has the familiar meaning of the partition function. the assumption of field thermalization implies that instead of averaging the dynamics of the system ([vector - dnlse]) over a long interval of @xmath43 one can compute the same averages via the equilibrium gibbs measure ([gibbs]). however from the point of view of the nonlinear optical waveguides it is impractical to use averaging over large distances since it requires optical waveguides that are way too long. instead the averaging can be understood as averaging over _ disordered initial conditions _ that can be be experimentally controlled @xcite. this brings about the notion of the thermalization distance @xcite @xmath44 after which the information about the initial state of the system is forgotten and the averaging over the initial conditions is equivalent to gibbs averaging. in our treatment we will assume (unless otherwise specified) that the initial amplitudes for both te and tm components are constant @xmath45, @xmath46 while the phases are uncorrelated uniformly distributed random variables. such assumption is not necessary but it simplifies the calculation of the ensemble averages of the initial hamiltonian and power. the knowledge of the partition function @xmath47 $] allows one to calculate an average energy per waveguide @xmath48 and the average intensity per waveguide @xmath49. on the other hand such averages must correspond to their initial values @xmath50, @xmath51 (averaged over the disordered initial conditions) since for each realization of the disorder these are conserved integrals of motion. indeed, from ([gibbs]) it follows that @xmath52 for each set of phase - averaged initial conditions (i.e. given pair @xmath53) the above two transcendental equations yield the effective inverse temperature @xmath40 and the chemical potential, @xmath41. as in the scalar case @xcite the necessary condition for the gibbs distribution ([gibbs]) to be normalizable is the positiveness of the temperature, @xmath54. thus the curve @xmath55 in the @xmath53 diagram represents a natural boundary between the conventional thermalization region (@xmath54) and the area where @xmath56 and the energy is localized in a form of discrete breathers that have been observed experimentally @xcite. it is convenient to introduce complex amplitudes and phases @xmath57, @xmath58 with @xmath59 being the phase difference between the two orthogonal components. the case @xmath60 (@xmath61-integer) corresponds to the linear polarization in @xmath62-the waveguide, @xmath63, @xmath64 describe circular polarization etc. this notation corresponds to the _ jones description _ of polarization (see e.g. @xcite). to calculate the partition function @xmath42 one needs to integrate the gibbs exponential in ([gibbs]) with the hamiltonian ([hamiltonian - vect]). in the new variables the integral takes the form : @xmath65 % \end{split}\]] where we have collected all the nonlinear coupling terms as well as chemical potential in the nonlinear interaction part @xmath66 with @xmath67 \\ & -\mu\,(a+b) \end{split } \label{hamiltonian - gamma}\]] while the linear waveguide coupling is given by the hamiltonian @xmath68. the gibbs distribution ([gibbs]) is normalizable when the partition function is finite. a close inspection of eq.([z - vector]) reveals that in order to achieve this not only the temperature must be positive @xmath54, but additionally the inequality @xmath69 this corresponds to possible thermalization for @xmath70 and for the anisotropic case @xmath71 this implies @xmath72. the borderline case @xmath73 (which includes the xpm without fwm @xmath74, @xmath32) generally requires special treatment and we will not consider it here. the non - normalizable property of the gibbs distribution indicates the emergence of the localized structures, i.e the genuine equilibrium state now consist of high - amplitude discrete breather (or several breathers if the system has not yet quite reached the equilibrium) interacting weakly with a small quasilinear background thermalized at infinite temperature, @xmath55 @xcite. so here the statistical mechanics provides us with a clue as to the regions in the parameter space where the localized structures can be observed in principle @xcite. in this paper we will only study the thermalization regime where the inequality ([thermalization - ineq]) is fulfilled and the temperature is positive so that gibbs distribution ([gibbs]) is always normalizable. we leave the analysis of the localized structures for future studies although we will comment on these in the following section. since it is impossible to evaluate the partition function @xmath42 in the closed form we will study 3 different limiting regimes which for the scalar case were already analysed in refs. these regimes are : i) low temperature limit @xmath75 ii) high temperature limit @xmath76 (which also serves as a borderline for emergence of localized structures) and iii) the anticontinuum (high intensity) regime when the effective nonlinearity parameter @xmath77 defined below in section [sec : nonl] is large, @xmath78. it turns out that doubling the amount of degrees of freedom as compared to the scalar case has a significant impact on the statistical properties of system, ([vector - dnlse]). we start our analysis with the first two regimes : @xmath75 and @xmath79.
The temperature boundaries and the thermalization region
the limit of @xmath75 corresponds to the configuration that minimizes the hamiltonian ([hamiltonian - vect]) subject to the given conserved pulse intensity per waveguide @xmath80. if we restrict our search to a solution with constant amplitudes, phase shifts and locked state of polarization (i.e. fixed @xmath81) the minimal configuration is achieved by the following field distribution : @xmath82, @xmath83 where the amplitudes @xmath84 and @xmath85 as well as chemical potential (i.e. the corresponding lagrange multiplier), @xmath41, are given by : @xmath86 this solution exists only for not too low intensities, i.e. for @xmath87 and provides a low bound for the energy per waveguide : @xmath88 in the limit of zero birefringence, xpm and fwm, @xmath89 we get the energy as a sum of energies of two identical scalar dnlses with the average intensity per waveguide @xmath90 - which corresponds to the result of ref.@xcite. in fig. [fig : h - p] (which is an analogue of fig.1 of ref.@xcite) we plot a phase diagram in the @xmath91 space for a specific case of 16 waveguides with @xmath74, @xmath92. both @xmath93 and @xmath80 have been rescaled to the dimensionless multiples of @xmath94 and @xmath95 respectively. one can see that eq.([hmin]) provides an excellent low boundary approximation even below the critical intensity for which the solution of ([min - amplitudes]) exists (which is @xmath96 for the chosen parameters). (color online) the different regions in @xmath91 parameter space. the ratio @xmath97 was assumed.,width=325] let us now turn to the opposite case of high temperatures @xmath79 assuming that the product @xmath98 remains finite. we will use the method similar to that of ref.@xcite applied earlier to the scalar case. in this limit one can neglect the linear coupling energy @xmath99 in the exponent of ([z - vector]) together with the birefringence contribution @xmath100. the partition function is then given by @xmath101 with @xmath102 \\ \times & \exp\left[-\frac{\beta\gamma}{2}\,\left(a^2+b^2 + 2\lambda_1 \, a \, b \right) -\beta\,|\mu| \,(a+b)\right]. \end{split}\]] next we assume that @xmath103 and write approximately @xmath104,\end{aligned}\]] where we have used the fact that the product @xmath105 is fixed while @xmath106 tends to zero. the answer is @xmath107 finally from eqs.([h - p]) we obtain in the leading approximation : @xmath108 the parabola @xmath109 provides the upper boundary @xmath55 for the thermalization region in the plane of parameters @xmath91 - see fig.[fig : h - p]. interestingly enough it does not depend on the 4-wave mixing constant @xmath18.
Nonlinearity dominated regime
in the following chapters we will adopt normalized units where the propagation distance is measured in units of coupling length @xmath110. we will also assume that the intensities of both te and tm components are @xmath111, @xmath112. we can also normalize the intensities of both components by the half of the initial intensity @xmath113 so that @xmath114. in the new dimensionless units one must simply substitute @xmath115 for @xmath116, @xmath117 for @xmath8 in the original coupled equations ([vector - dnlse]) and instead of the nonlinear coefficient @xmath118 one now has a dimensionless nonlinearity parameter @xmath119. since in most common nonlinear materials and waveguide geometries the ratio @xmath115 is in the order of unity the parameter @xmath77 indeed gauges the relative strength of nonlinearity @xcite. in this section we will be interested in the highly nonlinear regime, @xmath78. the motivation for this is twofold. firstly, this regime permits almost full analytical treatment which is always helpful when studying the general properties of any nonlinear system and secondly, in ref.@xcite it was shown that in the scalar case this limit corresponds to the reciprocal temperature @xmath40 that does not depend on the value of parameter @xmath77 and is a universal constant (in dimensionless units). the universality of the temperature in turn gives rise to a universal shape of the field correlation function. therefore it is interesting to see how this result changes in the vector case. one should expect that the dynamics and statistics in the vector case are much richer due to the doubled number of interacting degrees of freedom. here we will show that this is indeed the case. our main objects of interest will be the statistics of the intensity and polarization state of each waveguide as well as the distribution of the phase differences (i.e. phase gradients in the continuous limit) between the same components (e.g. te) of the adjacent waveguides together with the field correlation functions. it turns out that in the strongly nonlinear regime the statistics of the phase gradient is decoupled from those of the intensity and polarization. the latter are most conveniently described by using a popular alternative to the jones description of polarization, namely by introducing the four stokes parameters @xmath120 @xcite. they are related to the jones parameters via : @xmath121 the vector @xmath122 is called a stokes vector on a poincar sphere of radius @xmath123. the components of the stokes vector for each waveguide are related to the polarisation state of the waveguide while its magnitude provides the total intensity carried by both field components. for example, linear polarization corresponds to the equatorial plane @xmath124 while the circular clockwise and anticlockwise polarizations correspond to the north and south poles @xmath125. in what follows we will use both jones and stokes descriptions whichever is is more convenient. we will start our analysis with the anticontinuum limit when one can completely neglect both linear coupling and birefringence in eqs.([vector - dnlse]). as will be seen later this corresponds to the initial stages of evolution of field distribution towards the final state of equilibrium. the absence of linear waveguide coupling makes field dynamics in each waveguide independent from the rest. therefore instead of considering the system of coupled field equations ([vector - dnlse]) one can analyze the dynamics in a single waveguide. in particular the first three stokes parameters obey the following equations (the waveguide - index has been omitted for convenience) : @xmath127 this system is completely integrable and its various special cases and generalizations (like e.g. the inclusion of the birefringence terms @xmath128) have been extensively studied in literature both with connection to the nonlinear polarization dynamics @xcite in general and the dynamics of the polarization - locked vector solitons @xcite in particular. the complete integrability of system ([stokes - dyn]) is due to the existence of two integrals of motion of which the first is just the intensity, i.e. the zeroth stokes parameter, @xmath129, and the second, @xmath130, is related to the hamiltonian of the original system @xcite : @xmath131 in other words the dynamics of the system takes place on the intersection of the poincar sphere of radius @xmath129 and a hyperbolic (or elliptic) cylinder in the anisotropic case or a plane in the isotropic case given by the equation @xmath132. one can obtain an autonomous equation for @xmath133 which has the form of a duffing oscillator equation in the anisotropic case @xcite and harmonic oscillator equation in the isotropic case @xcite (see also appendix [sec : appendix]). the other two stokes parameters are recovered from eqs.([stokes - dyn]) while the phase of the te component can be determined by simple integration of the original field equations. since the system in the anticontinuum limit is completely integrable it does not thermalize, i.e. formula ([gibbs]) is inapplicable. instead one must use the exact solution for the stokes vector @xmath134 and average it directly over the initial conditions. as mentioned earlier we will assume that all waveguides initially have the same set of intensities, @xmath135, @xmath136, @xmath137 and random, independent phases. from the definition of the stokes parameters ([stokes]) it follows that that initial value @xmath138 is fixed while the points @xmath139 are uniformly distributed on a circle of radius @xmath140. in fig.[fig : anticont] we show the evolution of the marginal probability density functions (pdfs) @xmath141 and @xmath142 for the anisotropic and anisotropic cases obtained by averaging the solution of system ([stokes - dyn]) over the initial phase distribution. the parameters @xmath143 were chosen to correspond to the algaas compounds in the anisotropic case and fused silica for the isotropic case. one can see that the structure of the histograms is very sensitive to both the values of @xmath144-coefficients and the symmetry of the initial condition. for example when the initial intensity is equally distributed between the components, i.e. @xmath145, @xmath146 the marginal distribution @xmath147 shown in fig.[fig : anticont](a) is 4-modal. it is largely confined to the initial circle of radius @xmath148 and the four maxima are at the points @xmath149, @xmath150. at the same time the marginal pdf for the first stokes parameter @xmath133 is sharply centered at zero (fig.[fig : anticont](c)) i.e. for most initial realizations the symmetry between the components is preserved @xmath151. we have also checked this property for different runs with different distances, @xmath43 (not shown). this means that in the anticontinum limit for the anisotropic cubic crystal (like e.g. algaas) each given waveguide evolves into either linear or circular state of polarization - despite that no such preference existed in the initial conditions (the distribution was uniform on the circle). in the isotropic case one can observe that although the distribution of @xmath142 is still sharply peaked around zero (fig.[fig : anticont](c)) the distribution of the two remaining stokes parameters fig.[fig : anticont](b) is only bi - modal (and not 4-modal as in the anisotropic case) with the two maxima at @xmath149 corresponding to circular polarization only. the positions of the maxima in both cases can be easily explained by considering the two integrals of motion ([stokes - integrals]). as we have seen in both cases for most realizations one can neglect the value of @xmath133 and the pdf @xmath147 therefore remains confined to the circle of radius @xmath148 for all values of @xmath43. next if one looks at the distribution of the second integral of motion in ([stokes - integrals]), namely @xmath130, then for the initial values @xmath139 uniformly distributed on a circle of radius 2 one can see that the distribution of @xmath130 is bimodal with the two maxima at @xmath152 and @xmath153) in the anisotropic case and @xmath154 in the isotropic case. the intersection points of the pair of curves @xmath155 with the circle in the @xmath156 plane are exactly the four observed maxima in the anisotropic case and two in the isotropic one. in the case when initially the symmetry between the field components is broken (bottom row in fig.[fig : anticont]) the situation is much more complex owing to the multimodal features of the distribution of the parameter @xmath133 - see fig.[fig : anticont](f). each maximum of the @xmath142 gives rise to at least 2 maxima in the distribution @xmath141 producing complex crown - like shapes shown in fig.[fig : anticont](d),(e). each of the maxima now corresponds to a certain elliptically polarized state which varies from waveguide to waveguide. note also that the positions of the maxima remain fixed with the propagation distance @xmath43 while their widths experience weak periodic oscillations (not shown). the full theoretical analysis of these multimodal distributions requires further study which is beyond the scope of this paper. we have seen above that in the anticontinuum limit when the birefringence and the linear waveguide coupling can be neglected completely the system is integrable and instead of reaching thermal equilibrium it experiences periodic oscillations that can be averaged directly over the initial conditions to produce multimodal distribution for the established state of polarization. the question then arises as to what will happen if both the birefringence and the coupling are taken into account. one can expect that at the initial stages of evolution when the distance is less than the thermal length, @xmath44, the evolution of the system is close to anticontinuum limit and the state of polarization follows the statistics derived above for the anticontinuum limit. for @xmath157 however, the nonintegrability of the system becomes essential. the linear coupling between the waveguides leads to energy and power exchange between the waveguides so that eventually most of the initial peaks in the marginal pdf @xmath141 are destroyed and final thermal distribution sets in. this final distribution is characterized by the gibbs statistics ([gibbs]) with the partition function given by ([z - vector]). we wish to calculate the partition function and the probability distributions in this final state still assuming strong nonlinearity @xmath78 but now taking into account the linear coupling and birefringence as well. let us first assume that the initial conditions are such that the system thermalizes into a state with finite temperature, i.e. @xmath158 as @xmath159. we will shortly see that this is only possible for a very special choice of the initial conditions. then the dominating part of the exponent in ([z - vector]) is the nonlinear interaction term @xmath160. other terms (including the phase - dependent coupling energy) are relatively slow functions of the amplitudes and do not therefore contribute to the field intensity pdf and the partition function. however it is important to retain the chemical potential term since as we shall see below @xmath161 always. when calculating the partition function ([z - vector]) one can resort to the saddle point approximation in @xmath162 which implies that at large values of @xmath77 the main contribution to the integral comes from absolute minimum of @xmath163. this minimum is achieved for the values @xmath164 in other words the absolute minimum of the interaction hamiltonian subject to given total power is achieved by a circularly polarized state - the most symmetric of all. this minimum is also degenerate - for a system of @xmath165 waveguides there are @xmath166 possible choices of polarization orientation. this degeneracy is drastically reduced however if one takes into account the coupling term @xmath167. this term favors the configuration where the phase difference @xmath168 is uniform across the waveguides which leaves only two possibilities : either all fields are clockwise polarized or they are all anticlockwise polarized. these states also corresponds to the two maxima of the distribution function @xmath169 (or @xmath170 in stokes parameters) one for each direction of rotation. as previously we assume here that @xmath171 which ensures the convergence of the integrals for positive temperature. the minimal value of the interaction energy (at fixed power) is given by @xmath172 if we assume uniform initial amplitudes @xmath135 and @xmath136 (@xmath114) and neglect the terms that are not proportional to @xmath77 in the limit @xmath78 the average energy per waveguide is given by : @xmath173. \label{def - f}\]] the energy is always bound from below by @xmath174. in the spirit of saddle point approximation we can now plug @xmath175, @xmath176, @xmath177 from eq.([min - thermal]) into linear coupling pre - factor @xmath178 when calculating the partition function. the integration over the phases @xmath179 of the te component is reduced then to calculation of the partition function of the 1d calssical xy model (see @xcite or @xcite for details). in the thermodynamic limit (@xmath180) the contribution of this linear coupling term to the logarithm of the partition function is @xmath181. the remaining gaussian integration over the fluctuations around the minimum of @xmath182 eq.([min - thermal]) is trivial. inserting the resulting expression for the partition function @xmath42 into the system of equations ([h - p]) after some simple algebra we obtain the following transcendental equation for the inverse temperature @xmath40 : @xmath183 with @xmath50 given by eq.([def - f]). for the consistency of our approximation we must demand that the inverse temperature @xmath40 remains finite as @xmath77 goes to infinity (as is the case in the scalar dnlse @xcite). but this is achieved only when @xmath184 so that the l.h.s. vanishes which means that the initial input must be circularly polarised : @xmath145, @xmath185. in other words, in order to obtain a universal (i.e. @xmath77-independent) temperature constant, similar to the scalar case, the initial state must necessarily be the one that minimizes the nonlinear part of the hamiltonian, @xmath182 subject to the intensity constraint @xmath114. the system then becomes effectively scalar and thermalization occurs only in the distribution of the te phase differences @xmath186 while the field in each waveguide remains locked in its original clock- or anti - clockwise circular state of polarization. as in ref.@xcite the gibbs distribution can be approximately factorized so that the intensity of each waveguide @xmath187 has a narrow distribution close to gaussian centered around the conserved mean value of @xmath188. (color online) the statistics of the final thermal state in the limit of large nonlinearity when the initial field in _ each _ waveguide is circular clockwise polarized. (a),(b) the marginal distribution of stokes parameters @xmath141 for anisotropic and isotropic case respectively. (c), (d) the statistics of the intensity and phase difference (of the te component) respectively for different values of the nonlinearity parameter. (e) cross and self field correlation functions for the te and tm components. (f) the evolution of the intensity pdf with distance. the thermalization length @xmath44 is determined as a point where the distribution stabilizes.,width=325] the results of the numerical simulations of universal correlations in the vector case of @xmath189 waveguides are presented in fig. [fig : universal]. as prescribed, we have started from a symmetric initial configurations @xmath145 and uniform, uncorrelated phase distribution for the te components. to ensure constant circular polarization locking the phases of the tm component were obtained by those of the te by adding @xmath190 (corresponding to the clockwise field rotation). in all the figures (unless otherwise specified) the values of the parameters are : @xmath191, @xmath192 and the propagation length is @xmath193 (all in the normalized units). in fig.[fig : universal] (a),(b) we show numerical simulations of the marginal pdf @xmath141 for the anisotropic (@xmath74,@xmath194) and isotropic (@xmath28, @xmath29) cases respectively. one can see that both distributions have narrow maxima at the point @xmath195 corresponding to clockwise circular polarization (as in the initial state). we have found very little qualitative difference between the anisotropic and isotropic cases so figs.[fig : universal](c)-(f) feature only the former. in [fig : universal](c) we can see that the intensity pdf becomes more and more narrowly centred around the average value of 2 and the distribution is close to gaussian with the variance proportional to @xmath196. as for the pdf for the te phase difference, @xmath197, the theoretical prediction @xmath198 is clearly corroborated by the numerics shown in [fig : universal](d) : one can observe a perfect data collapse for different values of @xmath77. similar to universal correlations in the scalar case, one can obtain universal filed correlations in the vector case. in [fig : universal](e) we plot the field correlation functions functions defined as @xmath199 \rangle $] where @xmath200, @xmath201 denote different components of the field, te or tm, and the average is taken both over the waveguide position, @xmath62, and the initial conditions. since in the thermal equilibrium the te and tm components are always @xmath190 out of phase the cross - field correlation @xmath202 is always zero while the self - field correlation functions have identical universal shape : @xmath203 this is illustrated in fig.[fig : universal]e, where all three components are plotted together with the theoretical fit. finally, fig.[fig : universal](f) illustrates how quickly the final thermal distribution is achieved. the initial delta - peak in intensity distribution quickly broadens and settles to a stationary, almost gaussian distribution of the type shown in [fig : universal](c) already at the point @xmath204 (in units of the coupling length) that naturally serves as the thermalization length @xmath44. one can see that thermalization here occurs over just a fraction of the coupling length, i.e. on the scale of @xmath205 mm for a typical algaas waveguide system. of course one can come up with a more rigorous quantitative definitions of @xmath44, by looking e.g. at the saturation of the temperature defined from the simulated data via equipartition theorem @xcite but for the purposes of this paper we can restrict ourselves to the crude estimate above. all other distributions also stabilize at the same length. as for the value of the universal temperature constant, @xmath40, one might think that it can be obtained from eq.([beta - saddle]) by putting the l.h.s. to zero. unfortunately this is not so since the resulting value of @xmath40 differs from the one observed in the numerics (@xmath206) by the factor of almost 2. the reason for such a discrepancy is that the vanishing l.h.s. of eq.([beta - saddle]) generally represents the main terms of approximation (@xmath207). as these terms have now cancelled each other the universal temperature @xmath208 is given by the balance of the next terms in the expansion that are of the order of unity. but it turns out that the neglected terms in the saddle point approximation of the partition function being of the order @xmath209 have nevertheless the contribution to its logarithmic derivative that are in the order of 1, which affects the value of the reciprocal temperature @xmath40. therefore in principle one has to consider the next terms in the saddle point approximation of ([z - vector]). however we have opted for a different approach, namely, we have evaluated the logarithmic derivatives of the integral of @xmath210 numerically and the obtained solution of system ([h - p]) provided us a value of @xmath40 which fitted well the dependencies shown [fig : universal](d)-(e) (and also the intensity pdfs in [fig : universal](c)). from a few separate runs (not shown) it appears also that not only the found universal value @xmath206 is independent of @xmath77 (when @xmath78) but it is apparently independent of the material parameters @xmath143 as well. this fact however requires further thorough verification. in the above section we have shown that if one ensures that the initial field components are locked in the same circular state of polarization for each waveguide, the reciprocal temperature is a universal numerical constant @xmath40, independent on the nonlinearity parameter @xmath77 which implied universal shape of the intensity and phase pdfs as well as the field correlation functions. let us now turn to the more general case when the initial field does not have any preferred state of polarization, i.e. although both te and tm components have constant intensities @xmath135, @xmath136 (@xmath114) their phases are always independent and uniformly distributed. it turns out that although the final thermal state is still given by the constrained minimum of the nonlinear energy corresponding to the circular polarization given by ([min - thermal]), the value of the temperature @xmath40 is no longer a universal constant but depends on the nonlinearity. also there is a drastic increase in the thermalization length, sometimes by up to 3 orders of magnitude. however we shall see below that in this general case the `` universality '' is not lost altogether. rather it is the product @xmath211 and not the temperature itself that is universal so instead of e.g. universal filed correlations as in the scalar case @xcite one has _ universal intensity distribution _ @xmath212 that is not affected by the value of the nonlinearity parameter @xmath77. the width of this distribution (i.e. the variance of the intensity fluctuations) is of the order of @xmath213. it turns out, however, that this new constant @xmath211 is less `` universal '' than the temperature in the above - considered case of initial circular polarization inasmuch as it exhibits strong dependence on parameters @xmath143. we start by noticing that from eq.([beta - saddle]) it follows that if the initial fields are not circularly polarised, i.e. @xmath214 the reciprocal temperature @xmath40 must necessarily be of the order @xmath215 to balance the large terms in the l.h.s. of course ([beta - saddle]) was obtained in the saddle point approximation which strictly speaking no longer applies when the product @xmath211 is in the order of unity. but as we shall presently see the relation @xmath216 also follows from the exact scaling dependence of the partition function valid beyond the saddle point approximation. we will introduce the new notations for the re - scaled temperature @xmath217, and chemical potential @xmath218 that are now both in the order of 1. one can notice that in this regime since @xmath40 scales as @xmath219 one can neglect the coupling energy, @xmath99 (as well as the birefringence term @xmath220) in ([z - vector]) which simplifies the calculations somewhat. in fact this assumption is similar to the one used in section [sec : boundaries] when we determined the infinite temperature boundary @xmath55 with the only difference being that now one can not expand @xmath221 in powers of the argument (since the latter is in the order of unity). regardless of the applicability of the saddle point approximation the numerical simulations demonstrate that a single waveguide probability density function factorises into the product of the pdf of the phase differences in the te polarization, @xmath222 and the pdf for the remaining three jones parameters, @xmath223, with : @xmath224 note that since @xmath225 while @xmath226 the phase pdf @xmath222 is i) non - universal (i.e depends on the nonlinearity) and ii) approaches the uniform distribution as @xmath159 (unlike the scalar case where it remains fixed, c.f. fig.[fig : universal](d)). the intensity distribution is no longer narrow and has a finite width proportional to @xmath227. it is convenient at this stage to change variables from jones to stokes parameters @xmath228 according to the definition ([stokes]). the jacobian of this transformation is equal to @xmath229 and the marginal pdf for the stokes parameters has the form : @xmath230, \end{split}\]] and @xmath231. (color online) the statistics of the final thermal state in the limit of large nonlinearity and no initial phase correlation. (a)-(d) the marginal distribution of stokes parameters @xmath141 for anisotropic (a),(b) and isotropic (d, e) cases respectively. (a) and (c) show numerical simulations while (b) and (d) demonstrate theoretically calculated profiles. (e), (f) the statistics of the intensity for anisotropic (e) and isotropic (f) cases for different values of nonlinearity parameter @xmath77. the phase difference distribution @xmath197 for anisotropic (g) and isotropic (h) cases for different values of the nonlinearity parameter.,width=325] the marginal probability for the intensity as well as the overall normalization, the temperature and the chemical potential can be established by introducing the spherical coordinates on the poincar sphere of radius @xmath129 and integrating out the angular variables. the result reads : @xmath232\times \\ & \,e^{\tilde \beta s_0 ^ 2 (1-\lambda_1)y/4}. \end{split}\]] here @xmath233 is the normalization constant related to the overall partition function of eq.([z - vector]) via @xmath234 (as both linear coupling and the birefringence can be neglected). the full analytical form of the intensity pdf @xmath212 can not be obtained in the general case but one can see that it is i) asymmetric and ii) has a gaussian asymptote @xmath235 $] as @xmath236. again the condition @xmath171 ensures the convergence. the normalized inverse temperature @xmath237 and the chemical potential @xmath238 can then be determined in a standard way from eqs.([h - p]). by closely inspecting the normalization integral @xmath233 one can infer that it has a self - similar form (and so does the partition function) : @xmath239 where the explicit form of the function @xmath240 is given below. plugging this ansatz into ([h - p]) and excluding the terms containing the derivative of @xmath241 we obtain the relation between the (normalized) chemical potential and the temperature : @xmath242 the temperature is then to be obtained from a single transcendental equation : @xmath243 where the derivative is taken with respect to the argument, and @xmath244 is determined via eq.([def - f]). as for the function @xmath245, it is generally only available in quadratures as @xmath246e^{(1-\lambda_1)yz/4}.\]] the solution of the transcendental equation ([eq - beta]), @xmath247 is universal inasmuch as it does not depend on the value of nonlinearity, @xmath77. for given choice of @xmath143 it depends only on the average initial energy ([def - f]) and as long as the latter is not very close to the minimal value @xmath248 (which corresponds to the universal regime considered in the previous section) the solution always exists an is in the order of 1. in our numerical simulations shown in fig.[fig : general] we have chosen a symmetric initial condition with completely randomised phases : @xmath145, @xmath249. from ([def - f]) this corresponds to the value @xmath250. for the anisotropic case (algaas) the numerical solution of ([eq - beta]) yields @xmath251 while for the isotropic one (fused silica) one gets @xmath252. we have found that unlike in the case of universal correlations observed in section [sec : universal] (or the scalar case), the thermalization length now depends on the temperature (which is directly proportional to the nonlinearity). generally, the higher is the temperature, the longer it takes for the system to reach equilibrium. for the highest level on nonlinearity achieved in our simulations (@xmath253) the thermalization length was of the order @xmath254 units of coupling length. this is of course an extreme limit and lower values of nonlinearity (i.e. temperature) produce lower values of the thermalization length (typically @xmath255). in [fig : general](a)-(d) we can see marginal distribution of the 2 stokes parameters @xmath147 for @xmath192. theoretical profiles (b),(d) were obtained by integrating out the @xmath133 variable in the (normalized) distribution ([pdf - stokes]) while the parameters @xmath237 and @xmath238 were determined via ([mu - vector]) and ([eq - beta]). one can observe a good agreement between theory and numerics @xcite. the distribution is bimodal but the peaks (corresponding to circular clockwise and anticlockwise polarisation) are wide enough (the width is in the order of @xmath227). [fig : general](e),(f) compare theoretical prediction given by ([pdf - s0]) with the numerics. one can see that for different values of the nonlinearity @xmath77 the product @xmath256 (and hence the intensity pdf @xmath212) indeed remains the same and instead of universal correlation functions observed in the scalar case or the case of circular initial polarization (fig.[fig : universal](c)) one now has _ universal intensity distribution _ as given by ([pdf - s0]). the opposite case occurs with regards to the angle distribution @xmath197, [fig : general](g), (h). whereas in the the scalar case and in the case of circular polarization (fig.[fig : universal](d)) this pdf remains invariant, now, according to the theoretical prediction ([saddle - pdfs]) the distribution becomes closer and closer to uniform as @xmath77 increases since @xmath248 scales as @xmath77 while @xmath237 and @xmath238 remain constant. in some respect the results of ([fig : general])(e)-(h) mirror those results for the circular polarization [fig : universal](c),(d) and the scalar case (fig.2 of ref.@xcite) but with the angular and intensity distribution trading places. the universal nature of the intensity pdf @xmath212 can also be explained from the energy conservation (a similar argument for the scalar case was put forward in @xcite). in the limit of large @xmath77 most of the energy concentrates in the nonlinear part @xmath160 which is a sum of on - site component given by ([hamiltonian - gamma]). we can introduce the standard deviation for intensities as @xmath257 (where angular brackets denote both averaging over the initial conditions and over waveguides) and rewrite the nonlinear energy in terms of @xmath258. then the following average integral of motion exists : @xmath259=const,\]] where the term @xmath260 $] collectively denotes the contribution from the linear coupling and birefringence. the value of the constant is determined by the initial conditions and since initially all waveguides have the amplitudes, @xmath135, @xmath136, @xmath114, and random uncorrelated phases the initial variance vanishes @xmath261 and one has the following relation : @xmath262.\]] this means that beyond the thermalization distance @xmath44 the first term in the r.h.s. becomes a stationary @xmath77-independent constant of the order of unity and so does the variance of the intensity distribution as clearly seen in [fig : general](e). the situation changes however if one starts from the locked circular polarization state for each waveguide as in section [sec : universal]. then the first term in the r.h.s. of the above changes to @xmath263. but the distribution in this case centres narrowly on the extremal values @xmath264, @xmath265 so that the average vanishes and in this regime the variance scales inversely proportional to @xmath77 as is indeed seen in [fig : universal](c). finally it is instructive to see how the distribution of the polarization state @xmath141 evolves with distance @xmath43, passing from the initial isotropic state through the integrable regime described in section [sec : anticont] and eventually relaxing towards the equilibrium gibbs distribution. the successive snapshots of such evolution are given in fig.[fig : evolution] for the anisotropic case and @xmath192. one can see all successive stages from the uniform distribution on a circle of radius 2 onto the 4-modal distribution that occurs in a transient anticontinuum regime when the energy exchange between the waveguides is still negligible (cf. fig.[fig : anticont](a)) and finally towards the thermal equilibrium state which is achieved when the weak linear inter - waveguide coupling provides uniform mixing between all the degrees of freedom.
Conclusions
in this paper we have considered the structure of a thermalized field dynamics in discrete birefringent waveguide systems. we have defined for the first time an exact boundary in the space of the integrals of motion separating the thermal phase with positive temperature from that corresponding to localized excitations (discrete breathers). we have shown that in the limit of high nonlinearity depending on the choice of the initial conditions the marginal pdf for the stokes parameters @xmath141 relaxes either to a universal, broad, bimodal distribution (with the maxima corresponding to clock- or anti - clockwise circular polarization) or (if initially all waveguides are locked in the same circular polarization) a narrow - peaked one, corresponding to the thermal fluctuation around the initial state. in both cases either the effective `` temperature constant '' @xmath40 is universal (i.e. does not depend on the nonlinearity parameter @xmath77) or generally the product @xmath211 remains fixed as one increases the value of the nonlinearity. also in the limit of strong nonlinearity before reaching the final thermal state the system passes through the state where its dynamic is integrable, corresponding to the anticontinuum limit of the nonlinear polarization dynamics. in this regime the probability distribution of finding a system in a certain polarization state has a complex multimodal structure. as the thermalization sets in the different modes of the distribution gradually disappear leaving only the two maxima corresponding to the circularly polarized states achieving the global minima of the nonlinear coupling energy subject to fixed total power. as a possible continuation of this research one may suggest a more detailed study of the material dependence of the final thermal state (here only two specific examples were considered) as well as a full thermodynamical analysis of the breather region of the phase space as done by rumpf @xcite for the scalar case. the author would like to thank yaron silberberg and omer sidis for illuminating discussions. this work was supported by the marie curie fellowship project `` indigo ''.
The dynamics of stokes parameters in the anticontinuum limit
in this section we recall the exact results for the dynamics of the system of stokes parameters ([stokes - dyn]) (see @xcite). by differentiating the first eq. in ([stokes - dyn]) and using the other two we arrive at the following system : @xmath266 with @xmath267 \\ \beta&=&\frac{1}{2}\,\gamma^2\,(2-\lambda_1)(2 - 3\lambda_1), \quad \omega=2\gamma(1-\lambda_1)\,s_3(0)\end{aligned}\]] subject to initial conditions @xmath268, and @xmath269. the coefficient @xmath40 should not be confused with the reciprocal temperature as defined in other sections. thus in the isotropic case we get an equation for a harmonic oscillator while in the anisotropic case the equation is that of the nonlinear duffing oscillator. the solution in the isotropic case is simple : @xmath270 in the anisotropic case the form of the solution depends on the sign of the coefficient @xmath40 as well as the duffing oscillator energy : @xmath271 here we will only consider the case when @xmath272 so that we have simultaneously @xmath56, @xmath273 and @xmath274. for algaas we have @xmath275 which corresponds exactly to this regime. moreover in this case the potential energy has 2 symmetric maxima with the values @xmath276 and one can prove that for any choice of the initial conditions @xmath277 the difference @xmath278 the effective `` particle '' oscillates in the valley between the two maxima. the solution is expressed via jacobi elliptic function : @xmath279, \\ s_1^*&=&\sqrt{\frac{\alpha}{|\beta|}}\, \frac{\sqrt{2}\kappa}{\sqrt{1+\kappa^2 } }, \quad \kappa=\sqrt{\frac{e_*}{e}}-\sqrt{\frac{e_*}{e}-1 } \leq 1. \nonumber\end{aligned}\]] here @xmath280 is the maximal reachable amplitude of the oscillations, @xmath281 is the elliptic sine function and @xmath282 is the incomplete elliptic integral of the first kind : @xmath283 now using either the two integrals of motion of the main system - i.e. eqs([stokes - integrals]) or integrating the system ([stokes - dyn]) directly one can restore the remaining stokes parameters, @xmath284 and @xmath285. in the anti - continuum limit the averaging over the initial conditions amounts to averaging either ([s1-isotropic]) or ([duffing - solution]) for @xmath133 and using the integrals of motion to obtain the marginal pdf @xmath141. becasue of the jacobian @xmath229 there is a narrow integrable singularity at zero both in ([pdf - stokes]) and in the marginal pdf @xmath141, but the width of this singularity is much smaller than the computational grid and it does not feature prominently in our simulations. | statistical mechanics of two coupled vector fields is studied in the tight - binding model that describes propagation of polarized light in discrete waveguides in the presence of the four - wave mixing.
the energy and power conservation laws enable the formulation of the equilibrium properties of the polarization state in terms of the gibbs measure with positive temperature.
the transition line @xmath0 is established beyond which the discrete vector solitons are created. also in the limit of the large nonlinearity an analytical expression for the distribution of stokes parameters is obtained which is found to be dependent only on the statistical properties of the initial polarization state and not on the strength of nonlinearity.
the evolution of the system to the final equilibrium state is shown to pass through the intermediate stage when the energy exchange between the waveveguides is still negligible.
the distribution of the stokes parameters in this regime has a complex multimodal structure strongly dependent on the nonlinear coupling coefficients and the initial conditions. | 1307.2979 |
Introduction
in this paper all groups are assumed to be finite. the problem of detecting structural properties of a finite group by looking at element orders has been considered by various authors. amiri, jafarian amiri and isaacs in @xcite proved that the sum of element orders of a finite group @xmath0 of order @xmath2 is maximal in the cyclic group of order @xmath2. the problem of minimizing sums of the form @xmath3, where @xmath4 is a positive integer and @xmath5 denotes the order of @xmath6, was considered in @xcite, however there is a mistake in the proof pointed out by isaacs in @xcite. the main point of the argument in @xcite is a pointwise argument, and the strong evidence that it is true suggests to state it as a conjecture. [mainconj] let @xmath0 be a finite group of order @xmath2 and let @xmath7 denote the cyclic group of order @xmath2. there exists a bijection @xmath8 such that @xmath5 divides @xmath9 for all @xmath10. this is proved in @xcite by frieder ladisch in the case in which @xmath0 is solvable. note that the existence of a bijection as in the conjecture is equivalent to the existence of a family @xmath11 of subsets of @xmath0 with the following properties (here @xmath12 denotes euler s totient function) : * the sets @xmath13 are pairwise disjoint and @xmath14. * @xmath15 for all @xmath16, for all @xmath17. * @xmath18 for all @xmath17. indeed, given a bijection @xmath19 as in the conjecture, define @xmath13 to be the preimage via @xmath19 of the set of elements of @xmath7 of order @xmath20, and given a partition as above, define @xmath19 piecewise sending @xmath13 to the set of elements of @xmath7 of order @xmath20. the existence of such a partition is claimed in @xcite with a wrong proof, although this is not the main result of that paper. the main result of @xcite, dealing with the sum @xmath21, is a consequence of our main result (theorem [mainth](1) for @xmath22). although in this paper we do not prove conjecture [mainconj], such conjecture is worth mentioning because it is very much related to our results. let @xmath12 denote euler s totient function, i.e. @xmath23 denotes the number of integers in @xmath24 coprime to @xmath2. in this paper we consider the sum @xmath25 for @xmath26 real numbers and compare it with the case of the cyclic group of size @xmath1. in the case @xmath27, @xmath28 this sum equals the sum of element orders, in the case @xmath29 it equals the sum of the cyclic subgroup sizes. moreover if @xmath22 we get an extension of the case considered in @xcite and the case @xmath30, @xmath31 gives the number of cyclic subgroups. this last case was what motivated us in the beginning, and as a particular case of our main theorem we obtain the following. let @xmath32 denote the number of positive divisors of the integer @xmath2. [motiv] let @xmath0 be a finite group. then @xmath0 has at least @xmath33 cyclic subgroups and @xmath0 has exactly @xmath33 cyclic subgroups if and only if @xmath0 is cyclic. this theorem follows from corollary [cormotiv]. using the same techniques we also prove, in section [sprod], the following : [thprod] let @xmath0 be a finite group of order @xmath2 and let @xmath34. then @xmath35 with equality if and only if @xmath0 is cyclic. we also obtain a very interesting characterization of nilpotency (theorem [mainth](2)) : let @xmath36 be a real number and let @xmath0 be a finite group of order @xmath2. then @xmath37 and equality holds if and only if @xmath0 is nilpotent. let us be more specific about what we actually do in the paper. we prove the following result. [mainth] let @xmath26 be two real numbers, let @xmath0 be a finite group of order divisible by @xmath2 and let @xmath38set @xmath39 and @xmath40. 1. if @xmath41 and @xmath42 then @xmath43 with equality if and only if @xmath0 contains a unique cyclic subgroup of order @xmath4, for every divisor @xmath4 of @xmath2. 2. if @xmath44 then @xmath43 with equality if and only if @xmath0 contains a unique subgroup of order @xmath2 and such subgroup is nilpotent. 3. if @xmath45 and @xmath46 then @xmath47 with equality if and only if @xmath0 is cyclic. 4. if @xmath0 is nilpotent and non - cyclic then the sign of @xmath48 equals the sign of @xmath49. we prove this in section [main]. for the case @xmath50 we use lemma [nov] (a combinatorial tool, which is a key result in this paper) and for the case @xmath51 we adapt the arguments of @xcite. in section [examples], for any positive integer @xmath52, we construct infinitely many finite groups @xmath0 with exactly @xmath53 cyclic subgroups.
The main result
in this section we prove theorem [mainth]. as usual @xmath54 denotes the set of natural numbers (in particular @xmath55). denote by @xmath56 (the mbius function) the map taking @xmath2 to @xmath57 if @xmath2 is divisible by a square different from @xmath58, to @xmath58 if @xmath2 is a product of an even number of distinct primes and to @xmath59 if @xmath2 is a product of an odd number of distinct primes. the following result is well - known. let @xmath60 be two functions such that @xmath61 for all @xmath62. then @xmath63 for all @xmath62. an important example is the following. it is well - known that@xmath64 for all @xmath62. this is because in the cyclic group of order @xmath2 for any divisor @xmath20 of @xmath2 there are exactly @xmath65 elements of order @xmath20. applying the mbius inversion formula we obtain @xmath66. the following is our key combinatorial tool. [nov] let @xmath67 be two functions such that @xmath68 for @xmath69 and @xmath26 two real numbers such that @xmath50 set @xmath70 then we have : 1. write the prime factorizations of @xmath71 and @xmath4 as @xmath72 and @xmath73 where @xmath74 and @xmath75. then @xmath76 in particular @xmath77 always and @xmath78 if and only if one of the following holds : * @xmath79 and @xmath80. * @xmath81 and @xmath82, i.e. @xmath71 and @xmath4 are not coprime. * @xmath83 and @xmath84 for some @xmath85, i.e. @xmath71 is even and @xmath4 is odd. 2. @xmath86. 3. suppose @xmath87 are positive functions @xmath88 with @xmath89 and @xmath90 for all divisors @xmath91 of @xmath2. then @xmath92moreover equality holds if and only if @xmath93 for all divisors @xmath91 of @xmath2 such that @xmath94. 1. observe that since @xmath95 is zero whenever @xmath96 is not square - free, @xmath97, hence we may assume @xmath98. in the following computation the index @xmath96 will be written as @xmath99 where @xmath100. @xmath101 since @xmath102 and @xmath42, @xmath103 is a non - negative number, and it is zero if and only if either @xmath104 for some @xmath105, i.e. @xmath82 and @xmath106, or @xmath107 for some @xmath85, i.e. @xmath30 and either @xmath28 or @xmath108 for some @xmath85, in other words @xmath71 is even and @xmath4 is odd. we have @xmath109 3. since by point (1) @xmath110 for all @xmath111, applying point (2) to @xmath112 and to @xmath113 we obtain @xmath114the statement about equality follows easily. this concludes the proof. for @xmath4 a divisor of @xmath1 set @xmath115that is, the number of cyclic subgroups of @xmath0 of order @xmath4. observe that @xmath0 has exactly @xmath116 elements of order @xmath4. it follows that @xmath0 has exactly @xmath117 elements of order a divisor of @xmath4, that is, elements @xmath6 with the property that @xmath118. the following is a fundamental fact we will use in an essential way. [frob] let @xmath4 be a divisor of @xmath1 and let @xmath119 be the set of elements @xmath10 such that @xmath120. then @xmath4 divides @xmath121. thus we can write @xmath122 where @xmath123 is an integer depending on @xmath4 and @xmath0. let @xmath26 be two real numbers, let @xmath0 be a finite group of order divisible by @xmath2 and let @xmath38set @xmath39 and @xmath40. note that @xmath124. we proceed with the proof of theorem [mainth]. we prove that if @xmath41 and @xmath42 then @xmath43 with equality if and only if @xmath0 contains a unique cyclic subgroup of order @xmath4, for every divisor @xmath4 of @xmath2. observe that @xmath126 apply lemma [nov] to @xmath127, @xmath128, @xmath129, @xmath130. since @xmath41, if @xmath131 then @xmath43 and equality holds if and only if @xmath132 for all divisors @xmath91 of @xmath2, i.e. @xmath0 has a unique cyclic subgroup of order @xmath91 for all divisors @xmath91 of @xmath2. now suppose that @xmath30. then following the above argument we find that @xmath110 with equality if and only if @xmath108 for some @xmath133. in other words, equality holds if and only if whenever @xmath91 is a divisor of @xmath2 such that either @xmath91 is even or @xmath134 is odd, @xmath135. we prove that @xmath135 for all divisors @xmath91 of @xmath2, from which it follows that if @xmath4 is a divisor of @xmath2 then @xmath136, i.e. @xmath0 has a unique cyclic subgroup of order @xmath4. so let @xmath91 be a divisor of @xmath2. if @xmath91 is even then @xmath135, and the same is true if @xmath134 is odd, so now suppose that @xmath91 is odd and @xmath134 is even. in particular @xmath137 divides @xmath2 and is even, hence @xmath138. we prove that @xmath135. if by contradiction @xmath139 then since @xmath140 is a positive integer (by frobenius theorem), @xmath141 hence @xmath142, that is, there are at least @xmath137 elements @xmath10 such that @xmath143. but these elements also verify @xmath144, and @xmath145, i.e. there are exactly @xmath137 elements @xmath6 in @xmath0 verifying @xmath144. this means that every @xmath10 such that @xmath144 verifies @xmath146. now since @xmath134 is even @xmath0 has an element @xmath6 of order @xmath147, and since @xmath91 is odd @xmath148 ; on the other hand @xmath149, a contradiction. conversely, if @xmath150 for all divisor @xmath4 of @xmath2 then if @xmath4 divides @xmath2, @xmath151. we prove that if @xmath152 then @xmath43 with equality if and only if @xmath0 contains a unique subgroup of order @xmath2 and such subgroup is nilpotent. as above, using lemma [nov], we find @xmath43, and equality holds if and only if @xmath132 whenever @xmath91 is a divisor of @xmath2 such that @xmath91 and @xmath134 are coprime. applying this to the case when @xmath91 is a prime power we find that writing @xmath153, @xmath0 has a unique subgroup of order @xmath154 (which must then be normal in @xmath0) for @xmath155, thus the product of such subgroups is the unique subgroup of @xmath0 of order @xmath2, and it is nilpotent. conversely, if @xmath0 has a unique subgroup @xmath156 of order @xmath2 and @xmath156 is nilpotent then every sylow subgroup of @xmath156 is the unique subgroup of @xmath0 of its size. indeed if @xmath157 is a sylow subgroup of @xmath156 and @xmath158 with @xmath159 and @xmath160 is some subgroup of @xmath0 such that @xmath161 then @xmath162 is a subgroup of @xmath0 of order @xmath2 (because @xmath163 being the sylow subgroups of @xmath156 normal in @xmath0), hence @xmath164 and this implies @xmath165. hence if @xmath91 is a divisor of @xmath2 such that @xmath166 then @xmath135. we prove that if @xmath45 and @xmath46 then @xmath47 with equality if and only if @xmath0 is cyclic. the following arguments are the natural generalization of the arguments in @xcite. for @xmath167 a subset of @xmath0 define @xmath168. the following is lemma c in @xcite. [largep] let @xmath169 be the largest prime divisor of the integer @xmath170. then @xmath171. [nphi] let @xmath172 be two real numbers. let @xmath173 be a positive divisor of @xmath2. then @xmath174, and equality holds if and only if one of the following occurs. * @xmath175 and @xmath176. * @xmath177 and each prime divisor of @xmath2 divides @xmath4. since the function @xmath178 is multiplicative we may assume that @xmath179 and @xmath180 with @xmath169 a prime and @xmath181. the inequality @xmath182 becomes @xmath183, i.e. @xmath184 which follows from @xmath185, @xmath172. if equality holds and @xmath175 we find @xmath186. [cosets] let @xmath157 be a cyclic normal sylow @xmath169-subgroup of @xmath0. let @xmath10 and assume that the coset @xmath187 has order @xmath4 as an element of @xmath188. suppose @xmath175. then @xmath189 with equality if and only if @xmath6 centralizes @xmath157. the case @xmath190 is clear, so now assume @xmath191. since @xmath4 divides @xmath5 we can write @xmath192 for some integer @xmath193. then @xmath194 and as @xmath195, we see that @xmath193 is a power of @xmath169. but @xmath4 divides @xmath196, which is not divisible by @xmath169, so @xmath193 and @xmath4 are coprime, and there exists an integer @xmath2 such that @xmath197. now @xmath198 and we write @xmath199 so that @xmath200 because @xmath2 is coprime to @xmath4. also @xmath201 so since @xmath157 is abelian, @xmath202 centralizes @xmath157 if and only if @xmath6 centralizes @xmath157. we can thus replace @xmath6 by @xmath202 and assume that @xmath203. every element of @xmath187 has the form @xmath204 for some element @xmath205, and we argue that @xmath206 with equality if and only if @xmath6 centralizes @xmath207. since @xmath157 is cyclic, @xmath208 is characteristic in @xmath157, and thus @xmath209 and @xmath210 is a subgroup. now @xmath204 is an element of @xmath210, so @xmath211 divides @xmath212, and if equality holds then @xmath210 is cyclic hence @xmath6 centralizes @xmath207, and conversely if @xmath6 centralizes @xmath207 then since @xmath5 and @xmath213 are coprime, @xmath214. since @xmath175 it follows from lemma [nphi] that @xmath215 with equality if and only if @xmath6 centralizes @xmath207. now @xmath216 moreover equality holds if and only if @xmath6 centralizes @xmath207 for every @xmath205, i.e. @xmath6 centralizes @xmath157. let @xmath157 be a cyclic normal sylow @xmath169-subgroup of @xmath0. then @xmath217 with equality if and only if @xmath157 is central in @xmath0. write @xmath218 to denote the order of a coset @xmath187 viewed as an element of @xmath188. applying lemma [cosets] to each coset of @xmath157 in @xmath0 we have @xmath219 by lemma [cosets] equality holds if and only if every element @xmath10 centralizes @xmath157, i.e. @xmath157 is central in @xmath0. we show that if @xmath220 then @xmath0 is cyclic, and we do it by induction on @xmath1. this is a straightforward generalization of the argument used in @xcite (proof of the main theorem). assume @xmath220, i.e. @xmath221, which we can write as @xmath222 where @xmath223. then averaging on @xmath0 and using the fact that @xmath7 has @xmath23 elements of order @xmath2 and lemma [largep], where @xmath169 is the largest prime divisor of @xmath2, we find @xmath224 the strict inequality comes from the fact that we did not count the contribution of the identity element of @xmath7. this implies that there exists at least one element @xmath10 `` not below the average '', i.e. such that @xmath225 let @xmath226. since @xmath46 we have @xmath227 by lemma [nphi], and since @xmath228 we have @xmath229. we deduce that @xmath230hence @xmath169 does not divide @xmath231, in other words @xmath232 contains a sylow @xmath169-subgroup @xmath157 of @xmath0. moreover @xmath233 hence @xmath234 and it follows from the sylow theorem that @xmath235, i.e. @xmath236. corollary [psigp] then implies that @xmath217 with equality if and only if @xmath157 is central in @xmath0. let @xmath160 be the sylow @xmath169-subgroup of @xmath237, so that @xmath238 and @xmath239. we have @xmath240hence @xmath241. by the induction hypothesis we deduce that @xmath188 is cyclic. then @xmath242 and hence @xmath243. thus equality holds in the above chain of inequalities and thus @xmath157 is central in @xmath0. since @xmath157 is central and @xmath188 is cyclic, it follows that @xmath0 is abelian, and as @xmath157 is a sylow subgroup of @xmath0, we know that we can write @xmath244 where @xmath245 is cyclic. thus @xmath0 is a direct product of cyclic groups of coprime orders, and thus @xmath0 is cyclic, as required. we prove that if @xmath0 is nilpotent and non - cyclic then the sign of @xmath48 equals the sign of @xmath49. by lemma [nov] applied to @xmath127, @xmath128, we have @xmath246, hence @xmath247 suppose first that @xmath248 with @xmath169 a prime. we then have @xmath249 where we used that @xmath250. now, for @xmath251 we have @xmath252. therefore @xmath253 if and only if @xmath175, @xmath254 if and only if @xmath41, and @xmath255 if and only if @xmath177. since @xmath0 is non - cyclic, @xmath256 for some @xmath251 hence @xmath257 if @xmath175, @xmath258 if @xmath259 and @xmath260 if @xmath41. now assume that @xmath0 is any nilpotent non - cyclic group, and write @xmath261 as direct product of its sylow subgroups. note that @xmath262 is multiplicative in the sense that if @xmath112 are groups of coprime orders then @xmath263. write @xmath264 and @xmath265. if @xmath266 then, since @xmath0 is non - cyclic, by our above discussion of @xmath169-groups @xmath267 for some @xmath268 hence taking the product we find @xmath269. similarly if @xmath270 then @xmath271 and if @xmath272 then @xmath273
Some observations
consider now the case @xmath274. in this case the result looks as follows. let @xmath32 denote the number of positive divisors of the integer @xmath2. [cormotiv] let @xmath2 be a divisor of @xmath1. then @xmath0 has at least @xmath32 cyclic subgroups of order a divisor of @xmath2. moreover the following are equivalent. 1. for all divisors @xmath4 of @xmath2, @xmath0 has exactly @xmath4 elements @xmath6 such that @xmath120. @xmath0 has exactly @xmath32 cyclic subgroups of order a divisor of @xmath2. 3. the subgroup generated by the cyclic subgroups of @xmath0 of order a divisor of @xmath2 is cyclic and has order @xmath2. we prove that the number of cyclic subgroups of a finite group @xmath0 whose order divides @xmath2 equals @xmath275. let @xmath276 be the distinct cyclic subgroups of @xmath0 of order a divisor of @xmath2. then @xmath277 contains @xmath278 elements of order @xmath279. for @xmath280 write @xmath281 if @xmath282 generate the same cyclic subgroup of @xmath0. then @xmath283 is an equivalence relation in @xmath284 hence @xmath285now the result follows from theorem [mainth] choosing @xmath274. de medts and tarnauceanu in @xcite prove that if @xmath0 is nilpotent then @xmath286, and they conjecture that the converse holds, namely if @xmath286 then @xmath0 is nilpotent. also, they conjecture in @xcite that @xmath287 for all finite groups @xmath0. does the equation @xmath288 detect solvability of @xmath0 for some @xmath289? (this is a version of a question of thompson, cf. @xcite). what structural properties of @xmath0 can be detected by the equation @xmath288 (for fixed @xmath26)? is the set of pairs @xmath289 for which cyclic groups are detected by the equation @xmath288 dense in @xmath290? in the picture above the point @xmath291 does not detect anything, @xmath292 (which corresponds to the number of cyclic subgroups of @xmath0) and @xmath293 (which corresponds to the sum of the orders of the elements of @xmath0) detect cyclicity and whether @xmath294 detects nilpotency is the question of de medts and tarnauceanu. the checked zone detects cyclicity and the thick line detects nilpotency.
Proof of theorem
let @xmath0 be a finite group of order @xmath2. for a divisor @xmath20 of @xmath2 let @xmath295 be the number of elements of @xmath0 of order @xmath20. then @xmath296 hence @xmath297. let @xmath298 be the size of @xmath299 for @xmath300. then @xmath301. we compute @xmath302 where @xmath303. clearly @xmath304 and if @xmath305 then @xmath306. we now compute @xmath307 for @xmath305. if @xmath71 is a power of a prime @xmath169 then @xmath308. now assume this is not the case. let @xmath157 be the set of prime divisors of @xmath71, so that @xmath309. we prove that @xmath310. @xmath311 we used that @xmath312. in conclusion, writing @xmath313 and @xmath314 we have that @xmath315 equals @xmath316 since @xmath317 we obtain that @xmath318 since @xmath319 for all @xmath300 (by frobenius theorem), we obtain @xmath320suppose now that equality holds. then @xmath321 for @xmath322 and @xmath323. in particular if @xmath324 then @xmath325 hence there are @xmath326 elements of @xmath0 of order not dividing @xmath327, in particular the sylow @xmath169-subgroups of @xmath0 are cyclic. in particular if @xmath0 is a @xmath169-group we are done, so now assume that @xmath2 is divisible by at least two distinct primes. let @xmath157 be a sylow @xmath169-subgroup of @xmath0. we prove that @xmath157 is central in @xmath0. to do this it is enough to show that for all prime @xmath193 dividing @xmath2 there is a sylow @xmath193-subgroup @xmath160 of @xmath0 which centralizes @xmath157. if @xmath328 choose @xmath165. now suppose @xmath329. write @xmath330, @xmath331 for the positive integers such that @xmath332, @xmath333 are the largest powers of @xmath334 dividing @xmath2, respectively. it is enough to prove that @xmath0 contains elements of order @xmath335. there are @xmath327 elements of order dividing @xmath327 and @xmath336 elements of order dividing @xmath336, hence there are at least @xmath337 elements of order not dividing @xmath327 and not dividing @xmath336, in other words there are at least @xmath337 elements of order divisible by @xmath335, a suitable power of which has order @xmath335. we only need to make sure that @xmath338. we have @xmath339. since the sylow subgroups of @xmath0 are central and cyclic, @xmath0 is cyclic. the proof is completed.
An example
let @xmath52 be a fixed positive integer. in this section we construct infinitely many finite groups @xmath0 with exactly @xmath340 cyclic subgroups. this is done by applying proposition [subcont] with @xmath341, @xmath342. let @xmath343 be positive integers with @xmath344 (since @xmath345, @xmath346 is even), @xmath347 odd and @xmath348. let @xmath349 where the action of @xmath350 on @xmath351 is given by inversion : @xmath352. [subcont] the number of cyclic subgroups of @xmath0 is @xmath353. for the record, @xmath354 where @xmath355. we will first find the cyclic subgroups generated by elements of the form @xmath356 where @xmath357 divides @xmath347 and then we will look at the elements of the form @xmath358. let @xmath359 and let @xmath357 be a divisor of @xmath347. we compute @xmath360. since @xmath347 is odd, @xmath357 is odd, hence @xmath361 hence @xmath362. this implies that @xmath363 equals @xmath364 if @xmath2 is even and @xmath365 if @xmath2 is odd, implying that @xmath366, which is divisible by @xmath367. moreover clearly @xmath368 is determined by @xmath369 (all the elements of @xmath369 outside @xmath232 are of the form @xmath365) and @xmath370 is determined by @xmath369 because @xmath371. this means that different pairs @xmath372 give rise to distinct cyclic subgroups @xmath369. hence there are @xmath373 such subgroups. hence we have found @xmath373 cyclic subgroups of the form @xmath356 where @xmath357 is a divisor of @xmath347 (so @xmath357 is odd). since @xmath374 commutes with @xmath375 we have @xmath376 hence @xmath377. moreover the elements of @xmath0 of the form @xmath358 (i.e. @xmath368 times an even power of @xmath6) verify @xmath378 hence their order divides @xmath379, on the other hand there are exactly @xmath379 such elements hence they all belong to @xmath380. this proves that @xmath380 is the only cyclic subgroup of @xmath0 of order @xmath379. we have found @xmath381 cyclic subgroups of @xmath0. we have @xmath382 and @xmath383, hence @xmath384 therefore all we are left to show is that the cyclic subgroups we listed are all the cyclic subgroups of @xmath0. we need to show that if @xmath385 is a cyclic subgroup of @xmath0 of order divisible by @xmath367 then @xmath385 is generated by an element of the form @xmath356 with @xmath357 a divisor of @xmath347. say @xmath385 is generated by @xmath386. then @xmath387 is odd, indeed if @xmath387 is even then @xmath368 and @xmath388 commute hence @xmath389, contradicting the fact that @xmath390 is divisible by @xmath367. now, similarly as above, @xmath386 has order @xmath391 where @xmath392 is an odd divisor of @xmath346, that is, a divisor of @xmath347. since @xmath393 there is some integer @xmath394 such that @xmath395 and @xmath394 is coprime to @xmath396. this implies that @xmath397 and @xmath398 equals @xmath356 if @xmath394 is odd, it equals @xmath370 if @xmath394 is even.
Acknowledgements
we are really grateful to andrea lucchini and federico menegazzo for helpful discussions, comments and suggestions. we are also very grateful to the referee for his / her very careful reading of a previous version of the paper. 10 g. frobenius, uber einen fundamentalsatz der gruppentheorie, ii., sitzungsberichte der preussischen akademie weissenstein (1907), 428437. i. m. isaacs and g. r. robinson, on a theorem of frobenius : solutions of @xmath399 in finite groups, amer. monthly 99, no. 4 (1992), 352354. h. salmasian and nsa problems group, minimizing the sum of negative powers of orders of group elements : 10775, amer. monthly 109, no. 3 (2002), t. de medts, m. tarnauceanu, an inequality detecting nilpotency of finite groups (2012). http://arxiv.org/abs/1207.1020 t. de medts, m. tarnauceanu, finite groups determined by an inequality of the orders of their subgroups, bull. simon stevin 15 (2008), 699704. amiri, habib ; jafarian amiri, s. m. ; isaacs, i. m. sums of element orders in finite groups. algebra 37 (2009), no. 9, marius tarnauceanu, http://mathoverflow.net/questions/82547/a-question-on-the-product-of-element-orders-of-a-finite-group (2011). t. de medts, http://mathoverflow.net/questions/104183/order-increasing-bijection-from-arbitrary-groups-to-cyclic-groups (2012). | we prove several results detecting cyclicity or nilpotency of a finite group @xmath0 in terms of inequalities involving the orders of the elements of @xmath0 and the orders of the elements of the cyclic group of order @xmath1.
we prove that, among the groups of the same order, the number of cyclic subgroups is minimal for the cyclic group and the product of the orders of the elements is maximal for the cyclic group. | 1503.00355 |
Introduction
one of the most tantalizing observational discoveries of the past decade has been that the expansion of the universe is speeding up rather than slowing down. an accelerating universe is strongly suggested by observations of type ia high redshift supernovae provided these behave as standard candles. the case for an accelerating universe is further strengthened by the discovery of cosmic microwave background (cmb) anisotropies on degree scales (which indicate @xmath8) combined with a low value for the density in clustered matter @xmath9 deduced from galaxy redshift surveys. all three sets of observations strongly suggest that the universe is permeated by a relatively smooth distribution of ` dark energy'(de) which dominates the density of the universe (@xmath10) and whose energy momentum tensor violates the strong energy condition (@xmath11) so that @xmath12. although a cosmological constant (@xmath13) provides a plausible answer to the conundrum posed by dark energy, it is well known that the unevolving cosmological constant faces serious ` fine tuning'problems since the ratio between @xmath14 and the radiation density, @xmath15, is already a miniscule @xmath16 at the electroweak scale (@xmath17 gev) and even smaller, @xmath18, at the planck scale (@xmath19 gev). this issue is further exacerbated by the ` cosmological constant problem'which arises because the @xmath20-term generated by quantum effects is enormously large @xmath21, where @xmath22 gev is the planck mass @xcite. although the cosmological constant problem remains unresolved, the issue of fine tuning which plagues @xmath20 has led theorists to explore alternative avenues for de model building in which either de or its equation of state are functions of time. (following @xcite we shall refer to the former as quiessence and to the latter as kinessence.) inspired by inflation, the first dark energy models were constructed around a minimally coupled scalar field (quintessence) whose equation of state was a function of time and whose density dropped from a large initial value to the small values which are observed today @xcite. (` tracker'quintessence models had the advantage of allowing the current accelerating epoch to be reached from a large family of initial conditions @xcite.) half a decade after sne - based observations pointed to the possibility that we may be living in an accelerating universe, the theoretical landscape concerning dark energy has evolved considerably (see the reviews *??? * ; *??? * ; *??? * ; *??? * ; *??? in addition to the cosmological constant and quintessence, the current paradigm for de includes the following interesting possibilities : * * dark energy with @xmath23 * @xcite * * the chaplygin gas * whose equation of state drops from @xmath24 at high redshifts to @xmath25 today @xcite * * braneworld models * in which the source for cosmic acceleration rests in the _ gravity sector _ rather than in the matter sector of the theory @xcite * * dark energy models with negative potentials * @xcite * * interacting models of dark matter and dark energy * @xcite * * modified gravity and scalar - tensor theories * @xcite * * dark energy driven by quantum effects * @xcite * * dark energy with a late - time transition in the equation of state * @xcite * * unified models of dark energy and inflation * @xcite etc. faced with the current plethora of dark energy scenarios the concerned cosmologist is faced with two options : \(i) she can test _ every single _ model against observations, \(ii) she can take a more flexible approach and determine the properties of dark energy in a _ model independent manner_. in this paper we proceed along route (ii) and demonstrate that model independent reconstruction brings us face to face with exciting new properties of dark energy. applying the techniques developed in @xcite to a new data set consisting of @xmath26 supernovae from @xcite and an additional 22 supernovae from @xcite we show that the de equation of state which best fits the data evolves from @xmath3 at @xmath27 to @xmath28 today. _ an evolving equation of state of de is favoured by the data over a cosmological constant for a large region in parameter space. _
*model independent reconstruction of dark energy*
supernova observations during the previous decade have been pioneered by two teams : the high - z supernova search team (hzt) @xcite and the supernova cosmology project (scp) @xcite. the enormous efforts made by these two teams have changed the way cosmologists view their universe. a recent analysis @xcite of 172 type ia supernovae by hzt gives the following bounds on the cosmic equation of state (at @xmath29 cl) -1.48 < w < -0.72, [eq : state0] when the 2dfgrs prior @xmath30 is assumed @xcite. a similar bound w < -0.78, [eq : state1] is obtained for a new sample of high - z supernovae by scp @xcite., the dark energy equation of state becomes virtually unbounded from below and has a @xmath31 confidence limit of being @xmath32! @xcite] these results clearly rule out several de contenders including a tangled network of cosmic strings (@xmath33) and domain walls (@xmath34). however a note of caution must be added before we apply ([eq : state0]) or ([eq : state1]) to the wider class of de models discussed in the introduction. impressive as the bounds in ([eq : state0]) & ([eq : state1]) are, they strictly apply only to dark energy having a _ constant equation of state _ since this prior was assumed both in the analysis of the supernova data set as well as in the 2dfgrs study @xcite. aside from the cosmological constant (@xmath13), the topological defect models alluded to earlier and the sine - hyperbolic scalar field potential @xcite no viable de models exist with the property @xmath35. indeed, most models of dark energy (quintessence, chaplygin gas, braneworlds, etc.) can show significant evolution in @xmath0 over sufficiently large look back times. in this paper we shall reconstruct the properties of dark energy _ without assuming any priors _ on the cosmic equation of state. (the dangers of imposing priors on @xmath0 have been highlighted in @xcite and several of our subsequent results will lend support to the conclusions reached in this paper.) cosmological reconstruction is based on the observation that, in a spatially flat universe, the luminosity distance and the hubble parameter are related through the equation @xcite : [eq : h] h(z)=^-1 . thus knowing @xmath36 we can unambiguously determine the hubble parameter as a function of the cosmological redshift. next, the einstein equations h^2 & & _ m + _ de , + q & = & - = _ i (_ i + 3p_i) , [eq : einstein] are used to determine the energy density and pressure of dark energy : [eq : energy] _ de & = & _ critical - _ m = (1 - _ m(x)) , + p_de & = & (q -) , where @xmath37 is the critical density of a frw universe. the equation of state of de @xmath38 follows immediately @xcite [eq : state] w_eff(x) = 2 q(x) - 1 3 (1 - _ m(x)), where @xmath39, @xmath40. in quintessence models and in @xmath20cdm, the equation ([eq : state]) determines the true ` physical'equation of state of dark energy. however the subscript ` eff'in @xmath41 stresses the fact that this quantity should be interpreted as an ` effective'equation of state in de models in which gravity is non - einsteinian or in models in which dark energy and dark matter interact. examples of the former include braneworld models and scalar - tensor theories. it is well known that in a large class of braneworld models the hubble parameter does not adhere to the einsteinian prescription ([eq : einstein]) since it includes explicit interaction terms between dark matter and dark energy @xcite. in this case the equation of state determined using ([eq : state]) can still be used to characterize de, but physical interpretations of @xmath41 need to be treated with caution. and its derivatives. a detailed discussion of these issues can be found in @xcite.] one route towards the meaningful reconstruction of @xmath0 lies in inventing a sufficiently versatile fitting function for either @xmath42 or @xmath43. the parameters of this fitting function are determined by matching to supernova observations and @xmath0 is determined from ([eq : h]) and ([eq : state]). itself @xcite. see @xcite for a summary of different approaches to cosmological reconstruction. non - parametric approaches are discussed in @xcite ; see also @xcite.] our reconstruction exercise will be based upon the following flexible and model independent ansatz for the hubble parameter @xcite h(x) = h_0x^3 + a_0 + a_1x + a_2 x^2^, [eq : taylor] where @xmath44. this ansatz for @xmath43 is exact for the cosmological constant @xmath13 (@xmath45) and for de models with @xmath46 (@xmath47) and @xmath48 (@xmath49). it has also been found to give excellent results for de models in which the equation of state varies with time including quintessence, chaplygin gas, etc. @xcite. the ansatz ([eq : taylor]) is equivalent to the following expansion for de [eq : de1] _ de = (a_0 + a_1x + a_2 x^2 + a_3x^3), where @xmath50 is the present day critical density. the condition @xmath51 allows @xmath52 to _ mimic _ the properties of dark matter at large redshifts (@xmath53 follows from large scale structure constraints). from ([eq : taylor]) and ([eq : de1]) we find @xmath54, the value of @xmath55 in ([eq : taylor]) can be slightly larger than @xmath56 in this case. substituting ([eq : taylor]) into the expression for the luminosity distance we get = _ 1 ^ 1+z. [eq : lumdis] the parameters @xmath57 are determined by fitting ([eq : lumdis]) to supernova observations using a maximum likelihood technique. this ansatz has only three free parameters @xmath58 since @xmath59 for a flat universe. a note of caution : since the ansatz ([eq : de1]) is a truncated taylor expansion in @xmath60 its range of validity is @xmath61, consequently the ansatz - derived @xmath43 and @xmath42 should not be used at higher redshifts. note that the weak energy condition for dark energy @xmath62 has the following form for the ansatz ([eq : taylor]) : [eq : wec] a_0 + a_1 x + a_2 x^2 0, a_1 + 2a_2 x 0 , provided we assume that the @xmath63 term in ([eq : taylor]) is totally due to non - relativistic dark matter and does not include any contribution from dark energy. the demand that the wec ([eq : wec]) be satisfied for all @xmath64 (i.e. in the past as well as in the future) requires @xmath65 to be non - negative. however, the demand that the wec ([eq : wec]) be satisfied in the past (@xmath66) but not necessarily in the future, leads to the somewhat weaker constraint [eq : wec1] a_1 + 2a_2 0, a_2 0 . (models in which @xmath67 for @xmath68 and which violate the wec in the future, have been discussed in @xcite.) = 2.4 in the presence of the term @xmath63 in ([eq : taylor]) has two important consequences : (i) it ensures that the the universe transits to a matter dominated regime at early times (@xmath69), (ii) it allows us to incorporate information (available from other data sets) regarding the current value of the matter density in the universe. this information can be used to perform a maximum likelihood analysis with the introduction of suitable priors on @xmath70. in further analysis we will assume that the @xmath63 term in ([eq : taylor]) does not include any contribution from dark energy. we have also studied simple extensions of the ansatz ([eq : taylor]) by adding new terms @xmath71 and @xmath72. the @xmath71 term allows @xmath0 to become substantially less than @xmath73, thereby providing greater leeway to phantom models. the @xmath72 term allows de to evolve towards equations of state which are more stiff than dust (@xmath74) ; its role is therefore complementary to that of @xmath71. despite the inclusion of these new terms, our best fit to the supernova data presented below does not change significantly(choosing @xmath75 and @xmath76), which points to the robustness of the ansatz ([eq : taylor]) for the given data set. we should add that our reason for choosing an ansatz to fit @xmath43 rather than some other cosmological quantity was motivated by the fact that the hubble parameter is directly related to a fundamental physical quantity the ricci tensor, and is therefore likely to remain meaningful even when other quantities (such as the equation of state) become ` effective '. (this happens for instance, in the case of the braneworld models of dark energy discussed in @xcite.) the rationale for choosing a three parameter ansatz for @xmath43 is the following. the observed luminosity distance determined using type ia supernovae is rather noisy, therefore in order to determine the hubble parameter from @xmath77 and following that the equation of state, one must take two derivatives of a noisy quantity. this difficulty can be tackled in two possible ways : (i) either one smoothes the data over some interval @xmath78 (binning is one possibility), or (ii) we may choose to smooth ` implicitly'by parameterizing @xmath43 through an appropriate fitting function. the number of free parameters @xmath79 in the fit to @xmath43 will be related to the smoothing interval @xmath80 through @xmath81. increasing @xmath82 implies decreasing @xmath80 which results in a rapid growth of errors through @xmath83, and @xmath84 @xcite, therefore in order not to loose too much accuracy in our reconstruction we considered 3 parameter fits for @xmath43 in our paper (these correspond to 2 parameter fits for @xmath0). we now test the usefulness of the ansatz ([eq : taylor]) in reconstructing different dark energy models. the ansatz returns exact values for @xmath20cdm, and @xmath85, @xmath86 quiessence models. in figure [fig : err] we show the accuracy of the ansatz ([eq : taylor]) when applied to several other dark energy models such as tracker quintessence, the chaplygin gas and super - gravity (sugra) models. we plot the deviation of @xmath87 (which is the measured quantity for sne) obtained with the ansatz ([eq : taylor]) from the actual model values. clearly the ansatz performs very well over a significant redshift range for @xmath88 (also see appendix [sec : fits]). in fact, in the redshift range where sne data is available, the ansatz recovers these models of dark energy with less than @xmath89 errors. however it would be appropriate to add a note of caution at this point. although figure [fig : err] clearly demonstrates the usefulness of the ansatz for some de models, its performance vis - a - vis other models of de is by no means guaranteed. by its very construction the ansatz ([eq : taylor]) is expected to have limitations when describing models with a fast phase transition @xcite as well as rapidly oscillating quintessence models @xcite. (the ansatz ([eq : taylor]) can give reasonable results even for these models provided the resulting de behaviour is suitably smoothed.) for this reason, although the bulk of our analysis will be carried out using ([eq : taylor]), we shall supplement it when necessary with other fitting functions, which will provide us with an independent means with which to test the robustness of our reconstruction exercise. * methodology :* for our primary reconstruction, we use a subset of 172 type ia supernovae, obtained by imposing constraints @xmath90 and @xmath91 on the 230 sne sample, as in the primary fit of @xcite. for the ansatz ([eq : lumdis]), we require to fit four parameters : (@xmath92). we may use prior information on @xmath93 (@xmath94, @xcite) and @xmath70 (@xmath95, @xcite). is not model independent since it relies on the @xmath20cdm model to project from redshift space to real space. results coming from the use of this prior should therefore not be taken too literally in the present context. see @xcite for an interesting discussion of related issues.] the measured quantity for this data is @xmath96, therefore the likelihood function is given by & = & n (-), + ^2 & = & _ i=1 ^ 172 () ^2, where @xmath82 is a normalisation constant. therefore, the probability distribution function in the four - space @xmath97 is (h_0,, a_1, a_2) (-) pr(h) pr(h_0). where @xmath98 refers to the priors applied on the parameters of the system. our goal is to reconstruct cosmological parameters such as the equation of state @xmath99, therefore we marginalise over @xmath93 and obtain the probability distribution function in the @xmath100 space : (, a_1, a_2) = (h_0,, a_1, a_2) dh_0. in order to do this, we have to define the bounds of a four - dimensional volume in @xmath101. the bounds of @xmath93 are taken at @xmath102 of the hst prior. for @xmath70, the natural choice is @xmath103. it is not immediately obvious what the bounds should be for @xmath104. we choose a sufficiently large rectangular grid for @xmath105 (roughly corresponding to @xmath106) which includes most known models of dark energy. this bound is merely a matter of convenience and does not affect our results in any way. after marginalisation, we have a grid in @xmath100 space on which @xmath107 is specified at each point. we may now proceed in two ways. firstly, we may choose to fix @xmath70 at a suitable constant value (@xmath88) thereby obtaining a grid in the @xmath108 plane with @xmath109 (the probability if @xmath70 is known to be an exact value) defined at each point. for a particular redshift, we may then calculate @xmath110 at each point of the grid. this would yield results that would hold true if @xmath70 were known exactly. instead of using the exact value of @xmath70, we may use the prior information about it available to us (@xmath111), and calculate @xmath110 at each point of a three - dimensional grid, the probability @xmath112 at each point being known. therefore, at any given redshift @xmath2, @xmath113 can be tagged with a numerical value @xmath114. starting from the best - fit @xmath0 (the value at the peak of the probability distribution), we may move down on either side till @xmath115 of the total area is enclosed under the curve, thus obtaining asymmetric @xmath116 bounds on @xmath0. the @xmath117 bounds can be similarly obtained. = 2.4 in = 2.4 in * results :* we first show preliminary results for which the matter density is fixed at a constant value of @xmath88. a detailed look at the @xmath118 surface in the (@xmath104) plane (figure [fig : chi]) reveals the existence of two minima in @xmath118, a shallower one close to @xmath20cdm (@xmath119), and a deeper minimum at @xmath120. we would like to draw the readers attention to the fact that imposing the prior @xmath121 amounts to disallowing a significant region of parameter space (the unshaded region in figure [fig : chi]). consequently an analysis which assumes @xmath122 loses all information about the region @xmath123 around the deeper minimum! since we have no reason (observational or theoretical) to favour either minimum over the other, we shall always choose the deeper minimum as our best - fit in all the subsequent calculations. = 2.4 in in the figure [fig : dev], we plot the deviation of the squared hubble parameter @xmath124 from @xmath20cdm over redshift for the best - fit. we note that the quantity @xmath124 has a simple linear relationship with the parameters of the fit (eq [eq : taylor]), therefore the errors in this quantity increase with redshift. another quantity of interest is the energy density of dark energy. for this ansatz, @xmath125 (where @xmath126 is the present day critical density). the figure [fig : dens] shows the logarithmic variation of @xmath127 with redshift. in this figure too the errors increase with redshift. an interesting point to note is that initially, dark energy density decreases with redshift, showing the phantom - like nature (@xmath128) of dark energy at lower redshifts of @xmath129, while at higher redshifts, the dark energy density begins to track the matter density. before moving on to the second derivative of the luminosity distance (the equation of state) we may obtain more information from the dark energy density by considering a weighted average of the equation of state : 1+|w= (1+w(z)), [eq : w_avg1] where @xmath130 denotes the total change of the variable between integration limits. this quantity can be elegantly expressed in terms of the difference in energy densities over a range of redshift as 1+|w=. [eq : w_avg2] thus the variation in the dark energy density depicted in figure [fig : dens] is very simply related to the weighted average equation of state! cccc @xmath80&@xmath131&@xmath132&@xmath123 + @xmath133&@xmath134&@xmath135&@xmath136 + + @xmath137&@xmath138&@xmath139&@xmath140 + + @xmath141&@xmath142&@xmath143&@xmath144 + @xmath145 \epsfxsize=2.4 in \epsffile{h_exp_w_mi_tot.epsi } & \epsfxsize=2.4 in \epsffile{h_exp_w_tot.epsi } & \epsfxsize=2.4 in \epsffile{h_exp_w_pl_tot.epsi } \\ \mbox{\bf (a) } & \mbox{\bf (b) } & \mbox{\bf (c) } \end{array}$] . @xmath118 per degree of freedom for best - fit and @xmath20cdm models. @xmath146 is the present value of the equation of state of dark energy in best - fit models. @xmath147 refers to the best fit after imposition of the wec prior @xmath121. [cols="^,^,^,^,^,^,^ ",] = 2.4 in * using priors on age of the universe :* important consistency checks on our best - fit universe may be provided by observations of the age of the universe. unfortunately, estimates of the age of the universe from different methods can produce widely varying results one reason for which is that estimates of the hubble parameter itself can vary significantly. for instance, the hst key project yields @xmath148, while studies of the sunyaev - zeldovich effect in galaxy clusters give a significantly lower value @xmath149 @xcite. estimates of the ages of the oldest globular clusters suggest @xmath150 gyrs, at the @xmath29 confidence level @xcite and this age estimate is consistent with several other measurements including observations of eclipsing spectroscopic binaries @xcite, results from radioactive dating of a metal - poor star @xcite and wmap data @xcite (see also @xcite). the results from the wmap experiment suggest @xmath151 gyrs with a hubble parameter @xmath152, for @xmath20cdm cosmology (which satisfies the wec). adding sdss and sne ia data to wmap, @xcite find an age of @xmath153 gyrs for a slightly closed @xmath20cdm universe with @xmath154. although these results can not be carried over to evolving dark energy models including those implied by our best - fit reconstruction (which violate the wec) they provide an indication of the range within which the age of the universe might vary. keeping in mind these various results, we use two different priors on the hubble parameter : @xmath155 (@xmath132 bound from hst ; @xcite), and @xmath156 (approximate bound from wmap, sdss, sne ia ; @xcite). for each case, we choose three different gaussian priors on the present age of the universe : @xmath157 respectively, and perform the reconstruction for a @xmath88 universe. the results are shown in the figure [fig : age_pr]. we find that, for a hubble parameter of @xmath148, and with an additional prior on the age of the universe @xmath158 gyrs, the best - fit remains nearly the same, showing a rapid evolution of the equation of state from @xmath159 at @xmath4 to @xmath160 at @xmath161, and the errors become narrower. as the age is increased, the best - fit equation of state evolves more slowly, and the @xmath162 also increases (see table [tab : chi_pr]). for the prior @xmath163, we find that the lowest @xmath164 is obtained for the age prior of @xmath165 gyrs, which once again matches our best - fit. it should be noted that the errors are smaller in all cases, even though the @xmath118 may be larger. we must remember that the addition of a new prior which is consistent with the underlying dataset would lead to a natural reduction in errors. however, the addition of a prior inconsistent with the dataset would lead to a shift of the likelihood maximum as well as a reduction in errors, and the results would then fail to reflect the actual information present in the dataset. that this is happening here for the higher values of age can be seen from the fact that although the errors are reduced, the @xmath162 is actually larger. therefore priors from other observations should be added prudently to ensure that they do not lead to incorrect representation of the data. since there is as yet no clear model independent consensus on the age of the universe, the results we obtain in this section should be interpreted with a degree of caution. figure [fig : dec] shows the evolution of the deceleration parameter with redshift. we find that the behaviour of the deceleration parameter for the best - fit universe is quite different from that in @xmath20cdm cosmology. thus, the current value of @xmath166 is significantly lower than @xmath167 for @xmath20cdm (assuming @xmath168). furthermore the rise of @xmath169 with redshift is much steeper in the case of the best - fit model, with the result that the universe begins to accelerate at a comparatively lower redshift @xmath170 (compared with @xmath171 for @xmath20cdm) and the matter dominated regime (@xmath172) is reached by @xmath4. @xmath173 \epsfxsize=2.4 in \epsffile{h_exp_dens_hzt.epsi } & \epsfxsize=2.4 in \epsffile{h_exp_dens_scp.epsi } \\ \mbox{\bf (a) } & \mbox{\bf (b) } \end{array}$] @xmath173 \epsfxsize=2.4 in \epsffile{h_exp_w_hzt.epsi } & \epsfxsize=2.4 in \epsffile{h_exp_w_scp.epsi } \\ \mbox{\bf (a) } & \mbox{\bf (b) } \end{array}$] @xmath173 \epsfxsize=2.4 in \epsffile{h_exp_w_hzt_cons.epsi } & \epsfxsize=2.4 in \epsffile{h_exp_w_scp_cons.epsi } \\ \mbox{\bf (a) } & \mbox{\bf (b) } \end{array}$] based on the above analysis, it is tempting to conclude that the dominant component of the universe today is dark energy with a steeply evolving equation of state which marginally violates the weak energy condition. (of course, the less radical possibility of weakly time dependent dark energy satisfying the weak energy condition remains an alternative, too.) however, before any such dramatic claims are made, we need to check if our results are in any fashion a consequence of inherent bias in the statistical analysis itself, or in the sampling of the data. we therefore perform the following simple exercises to satisfy ourselves of the robustness of our results. * using different subsets of supernova data :* in an attempt to understand how the nature of the reconstructed equation of state is dependent on the distribution of data, we perform the reconstruction exercise on different samples of data. we have confined ourselves to the case where @xmath88 for these exercises. firstly, we may exclude the scp data points from the 172 sne primary fit, leading to a subsample of 130 sne. we call this the hzt sample. figures [fig : data_dens](a) and [fig : data_w](a) show the result of performing the analysis on this subsample without any constraints. the @xmath118 per degree of freedom for the best - fit is @xmath174, which is lower than @xmath175 for this sample. in this case we find that, though the error bars are slightly larger, overall the dark energy density behaves in the same way as before (compare figure [fig : data_dens](a) with figure [fig : dens]), showing phantom like (@xmath128) behaviour at lower redshifts and tracking matter at higher redshifts. the equation of state of dark energy also evolves much in the same way as when the entire sample is used (compare figure [fig : data_w](a) with figure [fig : total](b)), starting at @xmath176 and evolving rapidly to @xmath3. we may also use a sample complementary to this sample, where all the scp data points published till date are considered, along with the low redshift calan - tololo sample. this leads to a sample of 58 sne @xcite, which we call the scp sample. using this sample, we obtain the figures [fig : data_dens](b) and [fig : data_w](b). the best - fit has a chi - squared per degree of freedom : @xmath177, lower than @xmath178 for this sample. we find that here too, the dark energy density initially decreases and then starts tracking matter. the equation of state shows signs of rising steeply at low redshifts, but since the highest redshift in this sample is @xmath179, the behaviour of @xmath180 beyond this redshift can not be predicted, therefore the apparent flattening out of the curve beyond a redshift of one can not be seen in this case. for both these subsets of data, we may repeat the exercise using the prior @xmath181. the results obtained for the equation of state, as seen in figures [fig : data_c](a), (b), are once again commensurate with the results obtained earlier for the full sample (figure [fig : constr](b)). we may therefore conclude from this exercise that subsampling the data does not significantly affect our results, and the steep evolution of the equation of state of dark energy is not a construct of the uneven sampling of the supernovae, but rather, reflects the actual nature of dark energy. * testing our ansatz against fiducial dark energy models :* @xmath145 \epsfxsize=2.4 in \epsffile{h_exp_w_sim_w.epsi } & \epsfxsize=2.4 in \epsffile{h_exp_w_sim_cg.epsi } & \epsfxsize=2.4 in \epsffile{h_exp_w_sim_ev.epsi } \\ \mbox{\bf (a) } & \mbox{\bf (b) } & \mbox{\bf (c) } \end{array}$] @xmath173 \epsfxsize=2.4 in \epsffile{lin_w.epsi } & \epsfxsize=2.4 in \epsffile{lin_w_cons.epsi } \\ \mbox{\bf (a) } & \mbox{\bf (b) } \end{array}$] @xmath173 \epsfxsize=2.4 in \epsffile{bass_w.epsi } & \epsfxsize=2.4 in \epsffile{bass_w_cons.epsi } \\ \mbox{\bf (a) } & \mbox{\bf (b) } \end{array}$] the crucial question of course is whether the reconstructed equation of state of dark energy depends upon the ansatz which is used in the exercise,, whether the behaviour of the equation of state merely reflects a bias in the ansatz itself. in this section we show how the ansatz performs in recovering dark energy models whose equation of state is known, from simulated data. this ansatz was demonstrated to work extremely well when simulations of snap data were used @xcite. however, simulation of snap - like data is an optimistic exercise, since data of this quality is unlikely to be available in the near future. we now demonstrate the accuracy with which the ansatz can recover the fiducial background cosmological model if data is simulated using present - day observational standards. in figures [fig : sim_err] (a), (b), (c), we show how well the ansatz recovers the equation of state for three fiducial models (assuming @xmath88) : \(a) a quiessence dark energy model with a constant equation of state : @xmath182, \(b) a generalised chaplygin gas model with @xmath183 : with @xmath184 and the present - day equation of state @xmath185, which would give rise to an effective equation of state w(z)=-, and \(c) a model with a linearly evolving equation of state : @xmath186, with @xmath187. (for de models with @xmath188 the ansatz is exact therefore we do nt show the results for these cases.) we find that in all three cases, the fiducial model lies within the @xmath189 confidence limits around the best - fit @xmath0. based on this result, we claim that within the @xmath132 error bars, the reconstructed equation of state represents the true properties of dark energy when we use real data. * using other ansatz :* it is also important to check whether the results of our reconstruction can be replicated using other ansatz such as fits to the luminosity distance or the equation of state. many different fits have been suggested in the literature (see for example @xcite, @xcite, @xcite, @xcite). here we choose the fit suggested in @xcite in which the equation of state of dark energy is expanded as w(z)=w_0 + . [eq : lin] the luminosity distance can therefore be expressed as = _ 1 ^ 1+z, + where @xmath190 $]. we find that for this fit, the errors in the equation of state get larger with redshift, however this fit too demonstrates that the equation of state of dark energy increases rapidly with redshift (figure [fig : lin](a)) when no priors are assumed on the equation of state (eos). the @xmath118 per degree of freedom at the best - fit is @xmath191. when the prior @xmath121 is invoked, the best - fit eos remains very close to the @xmath20cdm model (figure [fig : lin](b)). therefore, from this ansatz, we may make the statement that at low redshifts, the equation of state of dark energy shows the same signs of rising steeply with redshift if no priors are assumed on the equation of state, thus supporting our earlier results. the large errors in the equation of state at redshifts of @xmath192 however make it difficult to make any definitive statements about the behaviour of dark energy at high redshifts. a limitation of the fit ([eq : lin]) is that it is unable to describe very rapid variations in the equation of state. an ansatz which accommodates this possibility has been suggested in @xcite w(z)=w_i+, [eq : bass] where @xmath193 is the initial equation of state at high redshifts, @xmath194 is a transition redshift at which the equation of state falls to @xmath195 and @xmath130 describes the rate of change of @xmath0. the resulting luminosity distance has the form : = _ 1 ^ 1+z, + where @xmath196 $]. the results for the analysis using this fit to the equation of state are shown in fig. [fig : bass]. we find that when the reconstruction is done without any priors on the equation of state (figure [fig : bass](a)), the best fit is _ remarkably close _ to the result for ansatz ([eq : taylor]) (figure [fig : total](b)). the @xmath118 per degree of freedom at the minimum is @xmath197 for this fit. the errors in this case are somewhat larger, especially at high redshift. if we constrain @xmath1, then as before, the evolution of the equation of state is much slower (figure [fig : bass](b)). so the reconstruction using this ansatz appears to confirm our earlier results. the above exercises lead us to conclude that our results are neither dependent on the nature of the statistical analysis nor on the manner in which the sne data is sampled. it therefore appears that dark energy with a steeply evolving equation of state provides a compelling alternative to a cosmological constant if data are analysed in a prior - free manner and the weak energy condition is not imposed by hand. as this paper was nearing completion, a new dataset consisting of 23 type ia sne was released by the hzt team @xcite. it is clearly important to check whether or not these new data points corroborate the findings reported in the previous sections. accordingly, we use a subset of 200 type ia sne with @xmath198 from the 230 sne sample of @xcite, and 22 sne with @xmath199 from the new sample to obtain a best - fit for our ansatz with @xmath88. we then plot the magnitude deviation of our best - fit universe from an empty universe with @xmath200 in order to illustrate how well our model fits the data (figure [fig : hubble]). for clarity, we plot the median values of the data points. we obtain medians in redshift bins by requiring that each bin has a width of at least 0.25 in log@xmath2 and contain at least 20 sne. for comparison, we also plot an @xmath20cdm @xmath201 model, as well as ocdm and scdm models. from figure [fig : hubble] we see that our dark energy reconstruction is a much better fit to sne beyond @xmath202 than @xmath20cdm. at low redshifts (@xmath203) the agreement between data and the two models is rather marginal. we now add 22 of the new supernovae (rejecting one with @xmath204) to our existing dataset of 172 supernovae and perform de reconstruction on this new dataset of 194 sne, assuming @xmath88 and no other priors. the resultant figure [fig : w_new] is similar to the figure [fig : total](b), with slightly smaller errors and has a best - fit @xmath205. the above exercises point to the robustness of results reported in previous sections, and indicate that evolving dark energy agrees well with the full data set containing 194 type ia sne.
*the accelerating universe and the _energy conditions_*
the energy conditions : * strong energy condition : @xmath11 (sec), * weak energy condition : @xmath206, @xmath7 (wec) play a vitally important role in our understanding of the accelerating universe, both in the context of inflation and dark energy. we therefore consider it worthwhile to review certain key developments which deepened our understanding of these issues. in an expanding frw universe the sec implies that the universe decelerates while the wec forbids the pressure from becoming too negative. additionally, in the 1960 s and early 1970 s it was noted that energy conditions play a crucial role in the formulation of the singularity theorems in cosmology. indeed, one of the necessary conditions for the existence of an initial / final singularity in big bang cosmology is that matter satisfies both the sec and wec @xcite. by the late 1970 s it became clear that not all forms of matter satisfy the energy conditions. perhaps the best example of a form of matter which satisfied the weak energy condition but violated the strong one is the cosmological constant, introduced into cosmology by einstein in 1917. in addition, the vacuum expectation value of the energy momentum tensor, @xmath207, which describes quantum effects (particle production and vacuum polarization) in an expanding universe could, in certain cases, violate both wec and sec @xcite. (for certain space - times, such as de sitter space, the vacuum energy momentum tensor generates a cosmological constant since @xmath208.) thus by the late 1970 s it was well known that neither of the energy conditions could be held as being sacrosanct. the 1980 s, as we all know, led to great advances in the development of the inflationary paradigm. the inflaton field mimics the behaviour of a cosmological constant over sufficiently small intervals of time and therefore violates the sec. early dark energy models were based on inflaton - type scalars which coupled minimally to gravity (quintessence). quintessence violates the sec but respects the wec. precisely because of the latter property, not _ any _ experimentally obtained @xmath42 is compatible with quintessence, as emphasized in @xcite. (the same observation holds for @xmath43, since the latter can be derived from @xmath42 using equation ([eq : h]).) clearly if observations do indicate that the wec is violated by de then more general (wec - violating) models for de should be seriously considered. one example of wec - violating de is provided by scalar - tensor gravity. scalar - tensor models contain at least two functions of the scalar field (dilaton) describing dark energy. as shown in @xcite, these two functions, namely, the scalar field potential and its coupling to the ricci scalar @xmath209, are sufficiently general to explain _ any _ @xmath43 obtained from observations. the wec can also be effectively violated in de models constructed in braneworld cosmology. it has recently been shown that such models, with @xmath210, are in excellent agreement with supernova data @xcite. since the field equations in these models are derived from a higher dimensional lagrangian the unusually rapid acceleration of the four dimensional universe arises because of the full five dimensional theory and not because of matter which continues to satisfy the energy conditions and whose density remained finite and well behaved at all times @xcite. this behaviour is in contrast to phantom, which assumes a conventional ` perfect fluid'form for the energy - momentum tensor and therefore contains pathological features such as an energy density which diverges in the future and a sound speed which is faster than that of light @xcite. the fact that the observed luminosity distance (derived from supernova observations) is better fit by dark energy violating the wec than either quintessence or a cosmological constant was first noticed by @xcite. caldwell called this ` phantom energy'and showed that larger values of @xmath70 (@xmath211) implied increasingly more negative values for the equation of state (@xmath5) of phantom. caldwell s results have since been confirmed by larger and better quality sne data sets for instance @xcite find that, in the absence of priors being placed on @xmath70, the de equation of state has a @xmath31 confidence limit of being @xmath32! both @xcite and @xcite however work under the assumption that the equation of state of dark energy is unevolving, so that @xmath212 constant. in this paper we have shown that, suspending the wec prior and allowing the dark energy equation of state to evolve brings out dramatically new properties of dark energy. thus the dark energy model which best fits the sne observations has an equation of state which rapidly evolves from @xmath213 at present (@xmath214) to @xmath215 at @xmath4. dark energy therefore appears to have properties which interpolate between those of dark matter (dust) at early times and those of a ` phantom'(@xmath5) at late times.
*conclusions and discussion*
this paper reports the model independent reconstruction of the cosmic equation of state of dark energy in which no priors are imposed on @xmath0. in the literature the imposition of various priors frequently precedes the analysis of observational data sets. such a procedure is well founded and entirely justified when priors are dictated by complementary information such as orthogonal observations coming from different data sets. however, on occasion the use of priors is justified on ` theoretical grounds'and in this case one must be careful so as not to prejudge nature. (compelling theoretical reasons might well reflect our own particular conditioning or set of prejudices!) in the case of the analysis of type ia supernova data, the priors most frequently used have been @xmath212 constant and @xmath216. both confine de to within a narrow class of models. moreover, as shown in @xcite, the imposition of such priors on the cosmic equation of state can, on occasion, lead to gross misrepresentations of reality. in this paper we do not impose any priors on @xmath0 and reconstruct the equation of state of dark energy in a model independent manner. in this case our best fit @xmath0 evolves from @xmath5 at @xmath161, to @xmath3 at @xmath217 (the upper limit is set by observations). this result is robust to changes in the value of @xmath70 and remains in place within the broad interval @xmath218. our reconstruction clearly favours a model of de whose equation of state metamorphoses from @xmath24 in the past to @xmath219 today. an excellent example of a model which has this property is the chaplygin gas @xcite. however, in this model dark energy does not violate the weak energy condition (if it was not already violated initially). our results also lend support to the dark energy models discussed in @xcite in which the de equation of state shows a late - time phase transition. an interesting example of an evolving de model in which @xmath220 at present whereas @xmath221 at earlier times is provided by the braneworld models (called brane1) examined in @xcite which have been shown to agree very well with current supernova observations @xcite. it is also conceivable that the observed rapid growth in the eos might characterise ` unified'models of dark matter (dm) and dark energy (de). we end this paper with a small speculation on this last possibility. since the nature of both dm and de is currently unknown, it may be that a mechanism exists which converts dm (with @xmath24) into de (with @xmath25) in regions with sufficiently high density contrast @xmath222. (this would happen if, for instance, the rate of conversion of dm into de depended upon @xmath223, etc.) since the conversion of dm to de is confined to high peaks of the density field this process will not occur uniformly in the entire universe but will be restricted to regions occupying a small filling fraction (@xmath224) (@xmath225 for regions with @xmath222 ; see for instance @xcite and references therein). this process could commence as early as @xmath226 when the first peaks in a cdm model collapse. since de does not cluster and since @xmath227 grows rapidly as the universe expands, de from high density regions (@xmath225) will spread at the speed of light, percolating through the entire universe (@xmath228) by @xmath4. since the creation of de is tagged to the formation of structure, this model may not encounter the ` coincidence problem'which plagues other scenarios of de including quintessence. (however this model might have problems in producing a sufficiently homogeneous and isotropic distribution of dark energy on the largest scales.) the concrete mathematical framework for a phenomenological model of this kind will be worked out in a companion paper. in summary, evolving de models have been shown to satisfy sne observations just as well (if not better) than the cosmological constant. our best fit equation of state, in the absence of any priors, evolves from @xmath213 at @xmath214 to @xmath215 at @xmath4. indeed, figure [fig : hubble] shows that our best fit eos is better able to account for the relative brightness of supernovae at @xmath229 than @xmath20cdm. however, the evolution in @xmath0 is much weaker if the prior @xmath230 (@xmath231) is imposed. due to the paucity of sne data beyond @xmath232 (till date, there is only a single data point beyond @xmath233, sn1999bf at @xmath234) it is not clear whether @xmath215 is a stable asymptotic value for the reconstructed de equation of state at high redshifts. new supernova data at @xmath235 from ongoing as well as planned surveys (snap) combined with data from other cosmology experiments (cmb, lss, s - z survey s, lensing, etc.) are bound to provide important insights on the nature of dark energy at high redshifts. our results clearly throw open exciting new possibilities for dark energy model building. @xmath236 _ acknowledgments : _ we would like to thank john tonry for several important clarifications and for help in preparing figure [fig : hubble]. we also thank sarah bridle, pier - stefano corasaniti, alessandro melchiorri, yuri shtanov and lesha toporenskii for their useful comments on an earlier version of the paper. one of us (vs) acknowledges useful discussions with salman habib and daniel holz. ua thanks the csir for providing support for this work. as was partially supported by the russian foundation for basic research, grant 02 - 02 - 16817, and by the research program `` astronomy '' of the russian academy of sciences. alam, u. and sahni, v., 2002, astro - ph/0209443. alam, u., sahni, v. and starobinsky, a.a., 2003, jcap * 0304 *, 002, [astro - ph/0302302]. alam, u., sahni, v., saini, t. d., and starobinsky, a.a., 2003,, * 344 *, 1057 [astro - ph/0303009]. alcaniz, j.s., d. and dev, a., 2002,, 067301 [astro - ph/0206448]. amendola, l., 2000,, 043511. barber, a.j.,, 2000,, 444. barris, b. j., 2004,, 571b [astro - ph/0310843]. bartolo, n. and pietroni, m. 2000, 023518. bassett, b.a., kunz, m., silk, j. and ungarelli, c., 2002, mnras, * 336 *, 1217 [astro - ph/0203383]. bertolami, o. and martins, p.j., 2000,, 064007. birrell, n.d. & davies, p.c.w., 1982, _ quantum fields in curved space _, cambridge university press, cambridge. boisseau, b., esposito - farese, g., polarski, d. and starobinsky, a.a., 2000,, 2236 caldwell, r.r., dave, r. and steinhardt, p.j., 1998,, 1582. caldwell, r.r., 2002,, 23 [astro - ph/9908168]. caldwell, r.r., kamionkowski, m. and weinberg, n.n., 2003, phys.rev.lett. * 91 * 071301 [astro - ph/0302506]. carroll, s.m., 2001, living rev.rel. * 4 * 1 [astro - ph/0004075]. carroll, s.m., hoffman, m. and trodden, m., 2003,, 023509 [astro - ph/0301273]. cayrel, r., 2001, nature * 409 *, 691 [astro - ph/0104357]. chaboyer, b. and krauss, l.m., 2002, apj, * 567 *, l45. chiba, t. and nakamura, t., 2000,, 121301(r). chiba, t., okabe, t. and yamaguchi, m, 2000,, 023511. chimento, l.p., jakubi, a.a., pavon, d. and zimdahl, w., 2003, 083513 [astro - ph/0303145]. copeland, e.j., liddle, a.r. and lidsey, j.e., 2001, 023509. corasaniti, p.s. and copeland, e.j., 2003, 063521 [astro - ph/0205544]. corasaniti, p.s., bassett, b.a., ungarelli, c. and copeland, e.j., 2003,, 091303 [astro - ph/0210209]. daly, r.a. and djorgovsky, s.g., 2003,, 9 - 20. [astro - ph/0305197]. damour, t., kogan, i.i. and papazoglou, a., 2002,, 104025 [hep - th/0206044]. deffayet, c., dvali, g. and gabadadze, g., 2002,, 044023 [astro - ph/0105068]. felder, g.n., frolov, a., kofman, l. and linde, a., 2002, phys. rev * 66 *, 023507. frampton, p., 2003 frampton, p. and takahashi, t., 2003,, 135. freedman, w.,, 2001,, 47. gerke, b, & efstathiou, g., 2002, 33, [astro - ph/0201336]. gnedin, n, lahav, o. and rees, m.j., astro - ph/0108034. gratton, r., 2003, a & a, * 408 *, 529, [astro - ph/0307016]. grib, a.a., mamaev, s.g. and mostepanenko, v.m., 1980, _ quantum effects in strong external fields _, moscow, atomizdat (in russian) [english translation : _ vacuum quantum effects in strong fields _, friedmann laboratory publishing, st.petersburg, 1994]. hansen, b., 2002, apj * 574 *, l155. hawking, s.w. and ellis, g.f.r., 1973, _ the large scale structure of space - time _, cambridge university press. hoffman, m., 2003, astro - ph/0307350. huterer, d. and starkman, g., 2003,, 031301, [astro - ph/0207517]. huterer, d. and turner, m.s., 1999,, * 60 *, 081301. johri, v.b., 2003, astro - ph/0311293. kallosh, r., linde, a., prokushkin, s. and shmakova, m., 2002, 123503. kamenshchik, a., moschella, u. and pasquier, v., 2001, 265. knop, r.a., et al., 2003, 102 [astro - ph/0309368] krauss, l.m., 2001, in proceedings, eso - cern - esa symposium on astronomy, cosmology and fundamental physics, march 2002. krauss, l.m., 2001, in international conference on the identification of dark matter, eds. n. spooner and v. kudryavtsev (singapore, world scientific), 1. krauss, l.m. and chaboyer, b., 2001, astro - ph/0111597. kunz, m., corasaniti, p., parkinson, d. and copeland, e.j., 2003, astro - ph/0307346. linder, e.v., 2003, 091301, [astro - ph/0208512]. maeda, k., mizuno, s. and torii, t., 2003, 024033 [gr - qc/0303039]. maor, i., 2002, 123003, [astro - ph/0112526]. marchi, g. de,, 2004,, 971 [astro - ph/0310646]. mcinnes, b., 2002, jhep * 0208 *, 029 [hep - th/0112066]. nakamura, t. and chiba, t., 1999,, * 306 *, 696. nunes, n.j. and lidsey, j.e., 2003, astro - ph/0310882. padmanabhan, t., 2003, phys. rep. * 380 *, 235 [hep - th/0212290]. parker, l. and raval, a., 1999,, 063512, 123502. peebles, p.j.e. and ratra, b., 1998, ap. j. lett. * 325 *, l17. peebles, p.j.e. and ratra, b., 2002, rev.mod.phys. * 75 *, 559 [astro - ph/0207347]. peebles, p.j.e. and vilenkin, a., 1999 063505. percival, w.j.,, 2001,, 1297. perlmutter, s.j.,, 1999,, 565. riess, a.g.,, 1998,, 1009. sahni, v., 2002, class. quantum grav. * 19 *, 3435, [astro - ph/0202076]. sahni, v. and habib, s., 1998,, 1766, [hep - ph/9808204]. sahni, v., saini, t.d., starobinsky, a.a. and alam, u., 2003, 201 [astro - ph/0201498]. sahni, v., sami, m. and souradeep, t., 2002, 023518. sahni, v. and shtanov, yu.v., 2003, jcap * 11 * 014, [astro - ph/0202346]. sahni, v. and starobinsky, a.a., 2000, ijmp d * 9 *, 373 [astro - ph/9904398]. sahni, v. and wang, l., 2000,, 103517 [astro - ph/9910097]. saini, t.d., 2003,, 129. saini, t.d., raychaudhury, s., sahni, v. and starobinsky, a.a., 2000,, * 85 *, 1162. sheth, j.v., sahni, v., shandarin, s.f. and sathyaprakash, b.s., 2003, mnras * 343 *, 22 [astro - ph/0210136]. singh, p., sami, m. and dadhich, n.k., 2003,, 023522 [hep - th/0305110]. spergel, d.n., 2003, suppl. * 148 *, 175 [astro - ph/0210136] starobinsky, a.a., 1998,, 757. tegmark, m., 2002,, 103507. tegmark, m., 2003, astro - ph/0310723. thomson, i.b., 2001,, 3089. tonry, j.l., et al., 2003,, 1, [astro - ph/0305008]. urena - lopez, l.a., matos, t., 2000,, 081302, [astro - ph/0003364]. wang, y and lovelace, g, 2001,, l115. weinberg, s. (1989) rev. phys. * 61 *, 1. weller, j. and albrecht, a., 2002,, 103512 [astro - ph/0106079]. wetterich, c, 1988,, 668. zeldovich, ya.b. (1968) sov. uspekhi * 11 *, 381.
Propagation of errors
we have seen that the error bars on @xmath0 for the analysis using ansatz ([eq : taylor]) are non - monotonic with redshift. low redshift behaviour of the equation of state affects the luminosity distance at all higher redshifts, while high redshift behaviour effects fewer such distances. this leads to an expectation that high-@xmath2 behaviour of the equation of state should be poorly constrained as opposed to the low-@xmath2 behaviour. this seems to contradict the behaviour seen in our figures. to investigate if this could be explained by our specific method of error analysis we describe the fisher matrix error bars below and show that they are almost identical to what we obtain in our method. in an analysis which uses an ansatz with @xmath237 parameters @xmath238, the fisher information matrix is defined to be f_ij, where @xmath239, @xmath240 being the likelihood. for an unbiased estimator, the errors on the parameters will follow the cramr - rao inequality : @xmath241. since the likelihood function is approximately gaussian near the maximum likelihood (ml) point, the covariance matrix for a maximum likelihood estimator is given by (c^-1)_ij. the fisher information matrix is therefore simply the expectation value of the inverse of the covariance matrix at the ml - point. given the covariance matrix, the error on any cosmological quantity @xmath242 is given by : [eq : propagate] _ q^2 = _ i=1^n () ^2 c_ii+2 _ i=1^n _ j = i+1^n () () c_ij. thus the nature of the errors on a quantity will depend essentially on the manner in which it is related to the parameters of the system. we now consider how errors propagate for different cosmological quantities for the polynomial fit to dark energy which we have used for most of the results in this paper : h^2/h_0 ^ 2 = x^3+a_0+a_1 x+a_2 x^2, x=1+z, where @xmath243. if @xmath70 is held constant then the parameters of the system are @xmath244. = 2.4 in = 2.4 in we obtain the covariance matrix in @xmath244 from the ml analysis, and then using equation ([eq : propagate]), calculate the errors on cosmological quantities of interest. for example, the errors on the quantity @xmath245 are given by : _ ^ 2(x) = (x-1)^2 [c_11 + 2 (x+1) c_12+(x+1)^2 c_22]. although the term @xmath246 is negative we find that @xmath247 still _ increases _ with redshift. this is shown in the figure [fig : err_hubb]. the errors shown are approximately similar to those obtained in figure [fig : dev]. the corresponding errors on the equation of state can be calculated using equations ([eq : state2]) and ([eq : propagate]), and has the somewhat more complicated expression : _ w^2(x) =, where @xmath248 and @xmath105 are the mean values of the parameters. although in this case it is difficult to predict the behaviour of error bars, after substituting the numerical values we obtain the error bars that are shown in figure [fig : err_w]. this figure can be compared to the figure [fig : total](b), having almost identical errors. this shows that the nature of our error bars is not an artifact of our specific method of error analysis. however, as shown in figure [fig : lin], a two parameter expansion in @xmath0 shows monotonically deteriorating errors in @xmath0 with the redshift, while the expansion in @xmath249 shows errors that improve with redshift (figure [fig : total](b)). this indicates that the nature of error bars might be affected by which quantity is being approximated. in the limit of infinite terms in the expansion of various quantities all the methods should produce identical result. the practical need for truncating these expansions make these approximations slightly different from each other. more specifically, we require setting of priors f(z) & = & _ i=0^ a_n z^n + a_n & = & 0 ; (n > n_p) where @xmath250 could be @xmath43, @xmath0 or any other physical quantity and @xmath251 is the chosen number of parameters. the non - linear priors in the above equation make different finite expansions inequivalent. since we do not know for certain if the underlying model for the accelerating expansion involves an energy component with negative pressure in a frw setting we are forced to choose one of the alternatives for approximations. we hope that with increasingly high quality data the effect of such truncations will eventually disappear.
Reconstruction of other dark energy models
@xmath173 \epsfxsize=2.4 in \epsffile{h_exp_error_lin.epsi } & \epsfxsize=2.4 in \epsffile{h_exp_error_coras.epsi } \\ \mbox{\bf (a) } & \mbox{\bf (b) } \end{array}$] we have seen in the figure [fig : err] that the ansatz ([eq : taylor]) works well for several physically motivated models of quintessence, chaplygin gas and sugra. in this section we take this exercise further and see how well it can reconstruct some of the other fits to dark energy known in literature. in figures [fig : sim] (a) and (b), we show results for simulations using @xmath88 and two different fits to the equation of state of dark energy : \(b) the non - perturbative @xmath0 suggested in @xcite and @xcite, which has the parameters @xmath256 (the dark energy equation of state today), @xmath257 (the dark energy equation of state at the matter dominated epoch), @xmath258 (the redshift where equation of state changes from @xmath257 to @xmath256), and @xmath130 (the width of transition). for the simulation we again use three sets of values in order of increasing growth rate of @xmath0 : (a) @xmath259, (b) @xmath260, and (c) @xmath261. we find that in both cases, the ansatz recovers the measured quantity to within @xmath89 accuracy in the redshift range important for sne observations. thus we find that even for fits for which the ansatz does not return exact values, it can recover cosmological quantities to a high degree of accuracy. | we reconstruct the equation of state @xmath0 of dark energy (de) using a recently released data set containing 172 type ia supernovae without assuming the prior @xmath1 (in contrast to previous studies).
we find that dark energy evolves rapidly and metamorphoses from dust - like behaviour at high @xmath2 (@xmath3 at @xmath4) to a strongly negative equation of state at present (@xmath5 at @xmath6).
dark energy metamorphosis appears to be a robust phenomenon which manifests for a large variety of sne data samples provided one does not invoke the weak energy prior @xmath7.
invoking this prior considerably weakens the rate of growth of @xmath0.
these results demonstrate that dark energy with an evolving equation of state provides a compelling alternative to a cosmological constant if data are analysed in a prior - free manner and the weak energy condition is not imposed by hand.
cosmology : theory cosmological parameters statistics | astro-ph0311364 |
Introduction
parametric resonance is a fundamental physical phenomenon that is encountered eventually in every area of science. in different disciplines, however, different facets of this rich phenomenon play a major role and are highlighted. parametric instability and multistable regimes in nonlinear dynamics @xcite, noise driven transitions among stable states in statistical physics @xcite, wave mixing and frequency conversion in wave dynamics @xcite are topics of primary interest. in electrical and optical engineering the low - noise properties of parametric amplifiers attract attentions, as well as non - classical statistical properties of the electromagnetic field generated by parametric devices @xcite. in superconducting electronics, the idea of using josephson junctions for quantum limited parametric amplification is under attention and development since the 1980s @xcite. during the last years the field revived by challenges of quantum information technology. the circuit - qed design, initially proposed for qubit manipulation and measurement @xcite, was employed for developing a variety of parametric devices @xcite. the circuit - qed approach is based on a combination of extended linear electromagnetic elements (transmission lines and resonators) with josephson junctions as nonlinear lumped elements. the design is flexible, allowing for diverse methods of parametric pumping, phase preserving and phase sensitive amplification schemes, different numbers or input and output ports, distributed josephson nonlinearities @xcite. the most of developed amplifiers are engineered in such a way that the dominant pump tone is sent through the same port as the signal, and parametric resonance is achieved by mixing them in nonlinear josephson elements. a different method is available for tunable superconducting cavities @xcite. the device consists of a resonator terminated with one (or more) dc - squid(s) that determines the reflection condition at the cavity edge and hence the cavity resonance spectrum. parametric resonance is achieved by rapid modulation of a magnetic flux through the squid with an appropriate frequency. a number of interesting parametric effects have been observed with such a device : phase sensitive amplification @xcite, frequency conversion @xcite, radiation and multistability regimes above the parametric threshold @xcite, quantum entanglement of output photons @xcite, generation of photons out of vacuum noise @xcite - an analog of the dynamical casimir effect @xcite. in this paper we formulate a consistent theory of parametric resonance in a tunable superconducting cavity. we aim at a unified picture of the phenomenon below and above the parametric threshold. to this end we include into consideration the squid nonlinearity, and damping due to connection to a transmission line. the latter provides a stage for studying the parametric amplification. we develop a full nonlinear description of the cavity resonance dynamics and the amplification effect in the classical limit, and study small quantum fluctuations of amplified and radiative fields. for certainty we consider parametric excitation of the main cavity mode @xmath0 by pumping with a frequency @xmath1 close to twice the cavity resonance, @xmath2. the overall picture of nonlinear parametric resonance in the tunable cavity is rather rich and complicated. at very small pump strength the cavity intrinsic dynamics resembles the one of the duffing oscillator @xcite showing a bifurcation of the cavity response and bistability. however, the scattering of an external incidental wave is qualitatively different from the duffing case : the scattering is inelastic, the reflected wave undergoes amplification or deamplification depending on the phase shift between the input tone and the pump (phase sensitive amplification). with increasing pump strength, the amplification effect increases, and at the same time the resonance narrows such that the bifurcation occurs at ever smaller input amplitudes. eventually, while approaching the parametric threshold, the cavity response becomes nonlinear at any small input amplitude. further increase of the pump strength leads to an instability of the cavity zero - amplitude state and the formation of finite - amplitude states accompanied by stationary parametric radiation at the half frequency of the pump. the radiative states are bistable in a certain window of detuning of the pump frequency from the cavity resonance. outside of this interval at red detuning the radiative states coexist with the stable zero - amplitude state (tristability), and the latter one becomes dominant at far red detuning. remarkably all these multistable regimes have been observed in experiment with a high quality tunable cavity @xcite. the multistability regimes are accompanied by random jumps among the stable states induced by thermal or quantum noise. these large amplitude fluctuations have small probability away from the bifurcation points and the parametric threshold, but become significant in the vicinity of these critical points (cf. ref. and references therein). these effects are out of the scope of this paper, here we restrict to small quantum fluctuations around well defined classical states outside of the critical regions, both below and above the parametric threshold. the bifurcation of the duffing oscillator response is employed in josephson bifurcation amplifiers (jba) for dispersive qubit readout @xcite. this method also applies to the parametric regime below the threshold (josephson parametric bifurcation amplifier, jpba). the novel feature here is the possibility to measure _ amplitude _ of the amplified probing tone, which exhibits strong dispersion with respect to the detuning near the threshold, and can be advantageous for high fidelity qubit readout. the parametric radiation above the threshold offers yet another strategy for the qubit readout based on the significant contrast between the strengths of the output radiation above the threshold and the amplified noise below the threshold. the paper is organized as follows. sections [sec : device], [sec : ftcavity] and [sec : losses] are devoted to the development of the theoretical framework for describing parametric resonance in a high quality tunable cavity. in sec. [sec : classical] we consider the nonlinear cavity response to a classical input signal below and above the parametric threshold, and in sec. [sec : application] apply the results for the analysis of parametric amplification and methods of dispersive qubit readout. section [sec : quantum] is devoted to the analysis of quantum fluctuations.
Circuit lagrangian
-cavity is terminated by a dc squid at the right end, and is capacitively coupled to a transmission line at the left end ; the squid is flux biased (phase @xmath3) via inductive coupling to a flux line imposing a driving phase @xmath4 ; @xmath5 and @xmath6 are the phase values at the right and left ends of the cavity, respectively. an incidental signal fed in from the transmission line is reflected, separated from the input, and then analyzed.] the device we study is sketched in fig. [fig : device]. its main part is a tunable superconducting strip line cavity terminated with a squid @xcite. the cavity is weakly coupled to a transmission line that feeds an external microwave signal in and provides means for probing the field inside the cavity. the cavity is a spatially extended system of length @xmath7 with inductance @xmath8 and capacitance @xmath9 per unit length, and the cavity state is characterized by the superconducting phase field @xmath10. we use the lagrangian formalism @xcite to describe the nonstationary dynamics of @xmath10. the lagrangian of the entire device consists of the sum of the lagrangians of the cavity, transmission line, and the coupling, @xmath11 = \mathcal{l}_{\text{cav } } + \mathcal{l}_{tl } + \mathcal{l}_{c } \,.\]] the lagrangian of the cavity in its turn consists of the lagrangian of the bare cavity @xmath12, and the lagrangian of the squid @xmath13 $], @xmath14 + \mathcal{l}_{s}[\phi_d] \nonumber \\ \label{eq : lcav_x_cos_orig0 } & = & { \left(\frac{\hbar}{2 e}\right)^2}\frac{c_0}{2 } \int_0^d d x \left(\dot \phi^2 - v^2 \phi'^2 \right) \\ & + & \left [{ \left(\frac{\hbar}{2 e}\right)^2}\frac{2 c_j}{2 } \dot \phi_d^2 + 2 e_j \cos{f(t) } \cos\phi_d \right] \nonumber \,.\end{aligned}\]] here @xmath15 is the field propagation velocity, @xmath16 is the boundary value of the cavity field at the squid, and @xmath17 is the phase across the squid controlled by external magnetic flux, see fig. [fig : device]. the squid is assumed symmetric for simplicity, with two identical josephson junctions, each having a josephson energy @xmath18 and a capacitance @xmath19. the phase @xmath17 appears in eq. as an external time - dependent parameter that is able to excite parametric resonance. in fact it is a dynamical variable that describes, together with the variable @xmath20, the dynamics of two coupled josephson oscillators of the squid driven by the external electromagnetic field @xmath21. in appendix [sec : squid] we show that in the limit of small @xmath22 the @xmath23-oscillator decouples from the @xmath20-oscillator. moreover, for experimentally relevant circuit parameters, the @xmath23-oscillator follows the drive field adiabatically because the resonance frequency of the @xmath23-oscillator is large compared to a typical resonance frequency of the cavity. a detailed derivation of eq. and the connection of @xmath17 to the external field @xmath21 is provided in appendix [sec : squid]. we assume here that the controlling field @xmath17 is composed of a constant biasing part @xmath24 and a small harmonic oscillation with amplitude @xmath25, @xmath26 it is worth mentioning that the constraint @xmath22 is essential, otherwise the two josephson oscillators become coupled and exhibit complex, even chaotic behavior under external drive @xcite. proceeding to the other components of the device, we suppose the transmission line to have the same characteristic parameters @xmath9 and @xmath8 as the cavity, @xmath27 = { \left(\frac{\hbar}{2 e}\right)^2}\frac{c_0}{2 } \int_{-\infty}^0 d x \left(\dot \phi_{tl}^2 - v^2 \phi_{tl}'^2 \right) \,.\]] the capacitive coupling is described with the lagrangian @xmath28 where @xmath6 and @xmath29 are the field values at the different sides of the coupling capacitor @xmath30.
Parametric dynamics of closed cavity
we first consider the cavity decoupled from the input line, @xmath31. the goal will be to identify the cavity frequency spectrum and investigate the parametric resonance. the lagrangian @xmath32, eq. , explicitly contains two dynamical variables, the phase field @xmath10, and its boundary value @xmath33. variation of the associated action with respect to @xmath10 leads to the wave equation, @xmath34 supplemented by the boundary condition @xmath35 at the open end of the cavity. variation with respect to the boundary value @xmath33 yields the boundary condition, @xmath36 where @xmath37 and @xmath38. under static biasing, @xmath39, the linearized boundary condition of eq. determines the set of cavity eigen modes @xcite, @xmath40 the frequency spectrum @xmath41 is non - equidistant, and can be tuned by varying the bias @xmath24. although the first term at the rhs of eq. can in principle be tuned to zero, at @xmath42, in practice it dominates over the second term, at least for the lowest cavity modes, by virtue of the large parameter @xmath43, where @xmath44 is the josephson plasma frequency. indeed, given typical experimental values, @xmath45 and @xmath46, the plasma frequency is @xmath47, while the cavity fundamental frequency is @xmath48, i.e., by one order of magnitude smaller (for typical cavity parameters@xcite @xmath49, @xmath50, and @xmath51). furthermore, the cavity inductive energy is typically small, @xmath52, compared to the josephson energy @xmath53. taking advantage of this relation, and neglecting the capacitive term in eq. , we get the approximate solutions @xmath54 the solutions of the spectral equation are graphically illustrated in fig. [fig : spectrum](a), while fig. [fig : spectrum](b) shows the cavity spectrum as a function of the parameter @xmath55. , and @xmath56 (solid) and @xmath57 (dashed) ; (b) cavity spectrum @xmath58 vs. @xmath55 according to eq. , the vertical line indicates the value @xmath59 used in (a).] the lagrangian formalism is sufficient for analyzing the classical parametric resonance. to describe the quantum dynamics the hamiltonian approach is more convenient. we derive the cavity hamiltonian by expanding the cavity field over the complete set of cavity eigen modes, @xmath60 where @xmath61 are time - dependent coefficients, and @xmath62 obey eq. . using expansion and noticing that the set of functions @xmath63 is non - orthogonal, we present the lagrangian ([eq : lcav_x_cos_orig0]) after some algebra in the form, @xmath64 - v(q_n, t) \,.\end{aligned}\]] here the `` masses '' of the mode oscillators are given by the expressions, @xmath65 and @xmath66\]] is a nonstationary nonlinear potential that mixes the eigen modes (see appendix [sec : lcav_n] for details of the derivation). it is convenient to absorb the factors @xmath67 and @xmath41 into the rescaled coordinate, @xmath68 and redefine the mode expansion in eq. accordingly. then introducing the conjugated momenta, @xmath69, we arrive at the cavity hamiltonian, @xmath70 for small pumping amplitudes and weak non - linearity, the potential @xmath71 in eq. could be considered perturbatively. however, the perturbative approach does not apply to the case of parametric resonance, when the pumping frequency matches an algebraic sum of the cavity eigen frequencies, @xmath72. in this case the corresponding cavity modes are strongly mixed and undergo complex time evolution. a particular case is the degenerate parametric resonance for @xmath73. in this paper we consider for certainty the degenerate parametric resonance of the fundamental mode, @xmath74. the method outlined below is straightforwardly extended to a non - degenerate parametric resonance. first we perform a canonical transformation corresponding to a transition to the rotating frame with frequency @xmath75. this is conveniently done in terms of a complex variable, @xmath76 for which the transformation reads @xmath77. the equations of motion for the amplitudes @xmath78 read, @xmath79 at this point we take advantage of small values of the pumping amplitude, @xmath25, and the field amplitude, @xmath22, and expand the potential @xmath71 in powers of these small parameters, keeping only the first non - vanishing terms, @xmath80 close to the resonance, @xmath81, the variable @xmath82 depends slowly on time while all the other variables contain rapid time oscillations. after averaging over these oscillations we arrive at the shortened equation of motion for @xmath82 (we skip the mode index @xmath83 below), @xmath84 with the parameters @xmath85 when applying the canonical transformation to coordinate and momentum, @xmath86, and averaging over fast oscillations, the cavity hamiltonian is cast into the form, @xmath87 this hamiltonian corresponds to the metapotential of the parametric lumped element oscillator @xcite, i.e. the degenerate parametric resonance in the cavity is mapped on the one in a lumped element oscillator. the mapping is defined by eqs. and ([eq : def_alphan]), where the effective pump strength @xmath88, and the nonlinearity coefficient @xmath89 are expressed through generic cavity parameters. according to the experimental values discussed in sec. [subsec : cavmodes], parameter @xmath90 in eq. is estimated as @xmath91. for such a small value of @xmath90, the parameters @xmath88 and @xmath89 are approximated, using the spectral equation ([eq : dispersion]), @xmath92 where @xmath93 is the cavity impedance and @xmath94 is the quantum resistance. it follows from these estimates that the effective pump strength @xmath88 is substantially reduced compared to the amplitude of the phase modulation in the squid, and the effective nonlinearity of the cavity oscillator is significantly smaller than the underlying bare nonlinearity of the squid oscillator (@xmath95 for the josephson potential). these remarkable properties result from the fact that the cavity is almost shortcut to the ground at the edge @xmath96 by virtue of large josephson energy in eq. (@xmath97), hence the boundary value of the field amplitude @xmath20 is small. the small values of the effective oscillator parameters are essential for the validity of the resonance approximation. the latter requires the evolution of @xmath98 to take place on a time scale much larger than the period of the cavity fundamental mode, @xmath99, over which the initial hamiltonian is averaged. it is instructive to express the constraints earlier imposed on the phases, @xmath100, in terms of the amplitude @xmath98 and the pump strength @xmath88, @xmath101 or equivalently, @xmath102. in other words, the constraints ([eq : constraint_a]) are more stringent than the ones required for the resonance approximation, @xmath103. on the other hand, these constraints provide sufficient room for the pumping strength to be increased above the parametric threshold beyond the resonance width @xmath104 (see eq. in the next section), @xmath105, for a high quality cavity. in most of our calculations we restrict to the lowest order @xmath106-dependence in eq., however, in some cases it is useful to keep higher order terms. in particular, the second order term @xmath107 will introduce, after averaging over time, a nonlinear shift of the resonator frequency, proportional to @xmath108. this shift is evaluated in eq. in appendix [sec : squid], and in terms of the effective pump strength it reads, @xmath109 this shift could be used in practice for evaluating the actual magnitude of the pump power acting upon the squid, which is usually not known. also, it causes quenching of the parametric instability at large pump strength, as will be shown in sec. [sec : paramresonance].
Cavity coupled to transmission line
the parametric effect in the closed cavity is an idealization. the connection to the external transmission line gives rise to the qualitatively important new features : firstly, the cavity field is allowed to leak out of the cavity, giving rise to the cavity damping, and secondly, an external electromagnetic signal can be fed into the cavity and amplified. our aim in this section will be to include these features into eq. , and derive the relation between the input and output fields, thus preparing the framework for the further investigation of parametric amplification. our derivation closely follows the input - output theory @xcite, (see also illuminative derivations in refs. ). aiming at the analysis of the quantum dynamics of the open cavity, we describe the field in the transmission line in terms of spatial modes, similar to eq. for the cavity, @xmath110 with @xmath111. opening of the cavity invokes also an additional set of modes @xmath112, however, in the weak coupling limit these modes do not contribute in the main approximation and are neglected here. focusing on the effect of cavity damping at weak coupling, we will only keep the cross term in the coupling lagrangian ([eq : lcavtl]), @xmath113 and neglect the quadratic terms, thus neglecting small corrections to the kinetic energies. with this simplification, and retaining only the fundamental mode field in the cavity lagrangian, we write the total lagrangian in the form, @xmath114 the corresponding hamiltonian reads, to first order of the weak coupling (@xmath115), @xmath116 repeating the derivation of the previous section we derive coupled equations of motion for the cavity amplitude @xmath117 and the spectral amplitudes of the transmission line, @xmath118, @xmath119 here we introduced the cavity damping rate, @xmath120 near the parametric resonance the equations of motion for the slow variables, @xmath121, and @xmath122, take the form, after averaging over rapid time oscillations, @xmath123 with @xmath124. we eliminate the transmission line modes from eq. , invoking the solutions of eq. , @xmath125 with initial conditions @xmath126 at time @xmath127, and substituting it into eq. . within the resonance approximation, the factor @xmath128 in the integrand is to be replaced with @xmath129, and the integration over the wave vector @xmath130 to be extended to the entire axis. after making these approximations we arrive at the langevin equation for the cavity amplitude, @xmath131 with the input flux amplitude @xmath132 the amplitude @xmath133 is associated, as shown in appendix [sec : langevin], with the incident (right - going) wave in the transmission line, @xmath134, taken at the boundary @xmath135. the solution of eq. can be equivalently expressed in terms of the amplitude at a future time, @xmath136, @xmath137, which defines the output flux amplitude @xmath138 via a relation similar to eq. with @xmath139 substituting for @xmath140. this output amplitude is associated with the reflected (left - going) wave in the transmission line, @xmath141, taken at @xmath135 (appendix [sec : langevin]). the relation between the output and input amplitudes reads, @xmath142 the parametric pumping couples the cavity field amplitude @xmath98 and its complex conjugate, and it is convenient to rewrite eq. in the matrix form, @xmath143 where @xmath144 the conservative part of the dynamics in eqs., is determined by the effective hamiltonian @xmath145 with @xmath146 from eq. , and @xmath147 being the phase shift between the input amplitude @xmath148 and the pump. besides the damping @xmath149 associated with the opening of the cavity, there might also be internal losses in the cavity, e.g. caused by the cavity resistance. a way to account for these losses is a model with a fictitious transmission line coupled to the cavity, that acts as a scattering channel with a noisy input amplitude @xmath150 and an associated damping rate @xmath151. this would lead to an enhanced damping rate, @xmath152, at the lhs of eq. , and also introduce an additional input term, @xmath153, at the rhs of this equation. the damping effect results in the broadening of the resonance, and if the resonance becomes sufficiently broad, higher cavity modes might also be excited, despite the non - equidistant property of the cavity spectrum. in this case, the isolated mode dynamics of eq. would be replaced by a more complex dynamics of parametrically excited coupled modes. to ensure the validity of the single - mode approximation, the condition @xmath154 must be met, where @xmath155 is the resonance width of the first cavity mode. for @xmath156, and the cavity spectrum given by eq. and parameters of sec. [subsec : cavmodes], the anharmonicity is of the order, @xmath157. this implies that the cavity quality factor @xmath158 should not be less than @xmath159. this corresponds to a small coupling capacitance in eq., @xmath160, assuming that the internal losses are not dominant, @xmath161.
Classical cavity response and radiation
in this section we analyze the cavity response to a noiseless classical input signal. we consider harmonic inputs, which have the form @xmath162, where @xmath163 is the detuning of the input signal from the half frequency of the pump. for the input frequency @xmath164 the cavity response is stationary, and it can be fully analyzed in the nonlinear regime. for detuned inputs, we restrict to small input amplitudes ; at large amplitudes the nonlinear response becomes complex and exhibits a transition to a chaotic regime. we start with the analysis of the intrinsic parametric resonance in the cavity in the absence of input signals, @xmath165. due to the damping, any initial cavity state evolves towards one of the steady states that define the picture of the parametric resonance. these steady states depend crucially on the pump strength @xmath88, and also on the detuning of the pump frequency from the cavity resonance, @xmath166. if @xmath167, only the trivial steady state, @xmath168, exists for all values of the detuning @xmath169. if @xmath170, the trivial state turns unstable within the interval @xmath171, and instead two non - trivial stable steady states, @xmath172, emerge at the threshold @xmath173, and persist for all @xmath174, see fig. [fig : homogeneous](a). these states have identical amplitudes, @xmath175 and are @xmath176-shifted in phase, with @xmath177. in the further red detuned region, @xmath178, the trivial steady state solution, @xmath168, becomes stable again, such that the three stable states coexist there. simultaneously, two new unstable states emerge having the same amplitude, @xmath179. in the limit of @xmath180, @xmath181 (the undamped duffing oscillator), the nontrivial stable and unstable states merge, forming a manifold of marginally stable states with indefinite phase @xmath182 and amplitude @xmath183. the steady states of the damped cavity at @xmath170 originate from the fixed points of the cavity hamiltonian @xmath184, eq. , which are illustrated in the insets of fig. [fig : homogeneous](a) for the mono-, bi-, and tristable regions. the damping @xmath104 introduces the threshold for the emerging nontrivial states, and shifts the positions of the steady states in phase space away from the fixed points. for @xmath185, stable (solid) and unstable (dotted) ; dashed vertical lines separate mono-, bi- and tristable regions ; insets show phase portraits of corresponding regions. (b) boundary of parametric instability without (solid) and with (dotted) account of nonlinear frequency shift, eq., for @xmath186 and @xmath187, cf. eq. ; yellow region corresponds to bistable high - amplitude state, blue region indicates coexistence of stable high - amplitude and zero - amplitude states. (@xmath186).] the pump parameters where new steady states occur are determined by the stability properties of the underlying linear system, characterized by the matrix @xmath188. its determinant, @xmath189, causes divergence at the parametric instability threshold, @xmath190, where the fixed point @xmath168 turns unstable. in a linear system this would lead to exponentially growing solutions in the parameter regime @xmath191 and @xmath192, with a rate @xmath193. in the nonlinear system this global instability is lifted by the bifurcation of the fixed point @xmath168 into the two new stable steady states. the cavity field, as it leaks into the transmission line, generates an outgoing field with the amplitude @xmath194 according to eq. . for the steady state, eq. , the flux radiated into the transmission line amounts to @xmath195 the nonlinear effect of the cavity resonance shift induced by the pump, mentioned in sec. [sec : parametric], eq. , leads to the quenching of the parametric instability at strong pumping as observed in experiment @xcite. by taking into account this shift, the actual pump detuning becomes @xmath196, and the parametric instability condition modifies accordingly, @xmath197 where @xmath198. the modified boundary of parametric instability in the @xmath199-plane is depicted in fig. [fig : homogeneous](b) : the instability region is bounded by the maximum blue detuning, @xmath200, and it is also bounded by a maximum pump strength at given detuning, e.g. @xmath201 at @xmath202. . insets show the basins of attraction of high - amplitude (red and green) and zero - amplitude (blue) states, for @xmath203 (left) and @xmath204 (right) (@xmath205, @xmath186).] in the experiment @xcite, all the described states of the parametrically pumped cavity have been observed : the subthreshold monostable regime at blue detuning, as well as the above - threshold bistable and tristable regimes at red detuning. the visibility of particular stable states in the multistable regime is defined by the probabilities of their occupation, which are determined by the relative areas of the respective basins of attraction, i.e. the phase space regions from which trajectories asymptotically approach the respective state. examples of the attractor basins in the red - detuned region, @xmath206, are shown in the insets of fig. [fig : attr_basins] where the blue basin belongs to the zero - amplitude state, and the red and green attractor basins are those of the high - amplitude states. the relative areas of the latter rapidly decrease and become very small in the far red - detuned region, as shown on the main panel in fig. [fig : attr_basins], implying that these states are much less populated. a similar conclusion is drawn from the calculation of the probability to escape from the high - amplitude states @xcite, which is much larger than the one for the trivial state, @xmath168, at far red detuning. these arguments explain why in the experiment @xcite the boundary of parametric resonance is washed out at red detuning, in contrast to the sharp boundary at blue detuning, which is determined by the threshold for the nontrivial steady states. now we turn to the discussion of the cavity response to a weak signal with zero detuning, @xmath164, and complex amplitude @xmath207. it is instructive to first review the response of the driven duffing oscillator @xcite, which corresponds to the limit @xmath208 in eq. . in this case the detuning @xmath169 refers to the deviation of the input frequency from the cavity resonance. the cavity response is given by the equation, @xmath209 the maximum response is achieved at @xmath210, along the tilted line @xmath211, and amounts to @xmath212, independent of @xmath89. as a consequence of the tilted resonance line, the cavity response can display bistability, with two coexisting stable states, as shown in fig. [fig : asqrsq_delta](a). the bistability emerges above the critical value of the driving amplitude, @xmath213, and at the detunings, @xmath214. the bistability region is confined by the bifurcation lines, @xmath215 \,,\]] forming a wedge in the (@xmath169-@xmath216) plane, as illustrated in fig. [fig : dpdbifurc_ain] with black lines. an ideal duffing cavity fully reflects the input signal, so the amplitude of the output, @xmath217, carries no information about the resonance, @xmath218. such information is only available for a lossy cavity, where @xmath219 on the other hand, the phase @xmath220 of the output signal is sensitive to the position of the resonance. this is the working principle of the josephson bifurcation amplifiers @xcite, where the variation of @xmath220 under sweeping the input power through the bistability region is exploited for the qubit readout. according to eq. vs. pump detuning @xmath169 for different values of the pump strength @xmath88, below threshold, @xmath221 (a, b), and above threshold, @xmath222 (c, d). solid and dashed lines mark stable and instable states, respectively. (@xmath223, @xmath224, @xmath225, @xmath226).] , for different subthreshold values of the pump strength, @xmath227 (@xmath224, @xmath225, @xmath226).] switching on the parametric pumping, @xmath228, qualitatively changes the cavity response. now the amplitude of the cavity field is determined by the equation, @xmath229 \right), \nonumber\\ \label{eq : detd } d & = & \det(\mathcal{a }) = \zeta^2 + \gamma^2 - \epsilon^2 \,.\end{aligned}\]] in the subthreshold regime @xmath167, the cavity response remains qualitatively similar to the duffing oscillator, see figs. [fig : asqrsq_delta](a)-(b). the role of the parametric pumping in this regime is to effectively reduce the damping term, @xmath230. this makes the resonance more narrow and, at the same time, strongly increases the cavity amplitude along the tilted resonance line @xmath210. another important feature is an explicit dependence of the cavity field on the phase shift @xmath147 of the input with respect to the parametric pump. the maximum value of the cavity field is, @xmath231 similar to the duffing limit, this value is independent of the nonlinearity coefficient @xmath89. the maximum response diverges at @xmath232, which can be compared to the resonance catastrophe of a linear parametric oscillator. while in the linear case the divergence occurs at @xmath233, the nonlinearity here shifts the divergence towards an infinite red detuning. as a consequence of the resonance narrowing, the critical bifurcation point moves towards the origin, @xmath234 when @xmath235, as illustrated in fig. [fig : dpdbifurc_ain]. above the threshold, @xmath170, the resonance splits into two branches, as shown in figs. [fig : asqrsq_delta](c)-(d), each branch consisting of two non - degenerate steady states, one pair being stable and the other unstable. these states originate from the degenerate nontrivial states in the absence of an input signal, cf. [fig : homogeneous], the degeneracy being now lifted by the input. the distance between the branches increases with @xmath88. the scattering by the parametrically pumped cavity is always inelastic, in contrast to the duffing cavity, and the output signal in general differs significantly from the input signal, not only in phase but also in the absolute value, @xmath236. using the input - output relation in eq. , and the steady state solution @xmath237 in eq., the output amplitude can be expressed as a function of the input amplitude, @xmath238 with the parameters @xmath239 the relation in eq. maps the points of the unit circle, @xmath240, onto the phase - dependent curve, @xmath241, and determines the phase - dependent gain @xmath242 the @xmath147-dependence of the gain @xmath243 and the quadratures of @xmath244 are illustrated in fig. [fig : rsq_phin]. in the monostable (subthreshold) regime the output amplitude @xmath194 is amplified (@xmath245) or deamplified (@xmath246) depending on the input phase. for @xmath247 the points @xmath241 form a strongly elongated curve in phase space, centered at @xmath248. in the quasilinear regime, where the parameters @xmath249 and @xmath250 in eq. are approximately independent of @xmath147, this curve approaches an ellipse with the half axes @xmath251 giving the maximum / minimum gain factor along those quadratures. for negligible internal losses, @xmath252, the amplified and deamplified quadratures are related according to @xmath253. in the limit @xmath254 the gain factors become equal and reduce to the reflection coefficient of the duffing oscillator, eq. . in the bistable regime above the threshold the corresponding output amplitudes @xmath241 are mapped on two distinct closed curves in phase space, with a @xmath176-phase shift between them, as shown in fig. [fig : rsq_phin]. the offset from the origin is due to the parametric radiation generated by the cavity. -plane, in the subthreshold regime (@xmath255 [blue, red]) and above the threshold (@xmath256 [yellow]) ; insets show the dispersion of the gain with the input phase @xmath147 below threshold (left) and above (right). (@xmath257, @xmath258, @xmath225, @xmath226).] the cavity response has a simple stationary form only when the frequency @xmath259 of the input signal strictly matches the half - frequency of the pump, @xmath75. if the input is time - dependent in the rotating frame, e.g. @xmath260 with @xmath163, the combination of the time - periodic force with the nonlinearity leads to the formation of a region in phase space where the cavity amplitude @xmath98 evolves chaotically, as illustrated in fig. [fig : phasespace] for the bistable regime above the parametric threshold. with increasing input amplitude and detuning a chaotic layer forms around the instable fixed points of the hamiltonian , and its area grows with @xmath261 and @xmath262, see fig. [fig : phasespace]. however, as long as the stable fixed points persist in the presence of the time - dependent drive, the amplitude of the damped cavity evolves into a time - periodic limit cycle around them, and then the time - average of @xmath263 gives only small corrections to the stationary result. driven by detuned input amplitude @xmath260. the trajectories @xmath264 are evaluated in the conservative limit, neglecting the damping term on the lhs of eq. , for (a) @xmath265, (b) @xmath266, and (c) @xmath267. in (b)-(c) trajectories are represented stroboscopically, at times @xmath268, (@xmath269). (@xmath270, @xmath202, @xmath271, @xmath272, @xmath186, @xmath226).] in this section we evaluate the response of an ideal cavity @xmath273 to a detuned signal in the monostable regime, @xmath274. we restrict to a linear response assuming @xmath275. suppose the input signal in eq. consists of two conjugated harmonics, @xmath276 (signal and idler in the terminology of non - degenerate parametric amplification). then the output field, as well as the field in the cavity, will also consist of the combination of the same harmonics. the output amplitudes are related to the input via the equation generalizing eqs. -([eq : def_q]), @xmath277 where @xmath278 the coupling between the conjugated harmonics is a fingerprint of parametric amplification : an input at frequency @xmath279 generates outputs at frequencies @xmath279 and @xmath280, and conversely an output at frequency @xmath279 consists of the contributions of inputs at frequencies @xmath279 and @xmath280. in particular, for @xmath281, eq. yields, @xmath282 amplification of the detuned signal is characterized by two gain factors, direct gain @xmath283, and interconversion gain, @xmath284. these two gains are fundamentally related, @xmath285, which is the consequence of the fundamental property of the matrix elements in eq., @xmath286. for the quantum fields, this property guarantees the unitary relation between the input and output quantum states (see later in sec. [sec : quantum]). for small detuning @xmath287 (single resonance), and large detuning @xmath288 (split resonance), at fixed pump strength @xmath289 ; (b) @xmath290 for fixed detuning @xmath291 and increasing pump strengths, @xmath292 (from bottom to top). (@xmath226).] the amplification of detuned signals possesses another interesting property - the appearance of resonance features, as illustrated in fig. [fig : gain_dw__dpd](a). the resonance structure of the gain is determined by the determinant @xmath293, eq. . it has a single minimum, at @xmath265, within the interval of relatively small detuning, @xmath294, and the gain factor @xmath290 is accordingly single peaked at @xmath164. however, at larger detunings, @xmath295 two resonance peaks emerge, situated symmetrically with respect to @xmath164 at @xmath296 the origin of these resonances can be understood from the behavior of the response function of a conventional damped linear oscillator, @xmath297. at small damping, @xmath298, the resonance is close to the eigen frequency @xmath0, @xmath299. with increasing damping the resonance is pulled towards the zero frequency, and stays at the zero frequency as soon as @xmath300. similarly, the resonances in the response of the linearized parametric oscillator, eq. with @xmath301, are at small @xmath104 close to the oscillator eigen frequencies, @xmath302, as in eq. , but are pulled towards @xmath164 with increasing @xmath104, and eventually merge when @xmath303, eq. .
Amplification and qubit readout
in this section we discuss the application of the parametrically pumped cavity for signal amplification, and for dispersive qubit readout. in what follows we shall neglect internal losses in the cavity and assume @xmath304. the amplification characteristics of the nonlinear parametric cavity depend on many parameters : pump and input strengths and detunings from the cavity resonance, relative phase shift, nonlinearity and damping, which makes the overall picture pretty intricate. the output power @xmath305 as a function of the input power @xmath216 for on - resonance input, @xmath164, is depicted in fig. [fig : drsqdbsq_ain] for various values of pump strengths and pump detunings. the major phenomenon here is the appearance of multistable regimes. the bistable regime establishes already below the threshold, @xmath306, in the red detuning region, @xmath307, as shown on fig. [fig : drsqdbsq_ain](b). above the threshold, the mono-, bi-, and tristable regimes exist at different detunings, as shown on fig. [fig : drsqdbsq_ain](c). moreover, in the latter regime, the output power does not approach zero value at @xmath308 due to the effect of parametric radiation. for the amplification purpose the monostable regime in fig. [fig : drsqdbsq_ain](a) is the most suitable. the output power in this regime depends monotonically on the input power, but exhibits pronounced nonlinearity with increasing pump strength at input power levels @xmath309. the maximum differential gain is achieved at small input power, and for phase shift @xmath310. the gain is controlled by the quantity @xmath249 in eq. , and at large @xmath311, @xmath312 the gain increases while approaching the threshold (cf. fig. [fig : asqrsq_delta](b)), @xmath313, in the quasilinear approximation, and then it is limited by the nonlinearity. let us evaluate this upper bound for the gain at @xmath232 and @xmath202. in this case, @xmath314. extracting the amplitude @xmath237 from eq. , with @xmath315, @xmath316 we get @xmath317 in a similar way we can evaluate the absolute minimum of deamplification. this is achieved at @xmath318, where @xmath319, and @xmath320 leading to the equation for minimum gain, @xmath321 we note that the nonlinear deamplification is more efficient than the amplification : the product of the maximum and minimum nonlinear gains significantly deviates from unity, in contrast to the linear case, @xmath322 with these results we conclude that the maximum amplification (deamplification) efficiency is controlled by the parameter @xmath323, and therefore a relatively small nonlinearity coefficient is required to achieve a large parametric effect. as we will see later, the same conclusion is also valid for the nonclassical properties of the fluctuations. at this point it is appropriate to estimate the output signal - to - noise ratio for parametric amplification, referring to the results of the noise analysis in sec. [sec : noisec]. according to eqs. and the amplified noise increases in the vicinity of the threshold, however, the noise amplification is less efficient than the signal amplification, giving the ratio (for the quasilinear limit), @xmath324 this ratio is large as soon as @xmath325. vs. input power @xmath216, below threshold, @xmath167 (a, b), and above threshold, @xmath326 (c). (a) @xmath327 and @xmath328 (from bottom to top) ; (b) @xmath329 and @xmath330 (from bottom to top) ; the dotted line refers to the duffing limit, @xmath180. (c) @xmath331 and @xmath332 (from top to bottom) for each of the parameters instable branches are indicated by dashed lines. (@xmath272, @xmath186).] amplification of a detuned signal, @xmath333, has qualitatively similar properties in the vicinity of the parametric threshold, @xmath334. here the gain factor @xmath290 has a quasi - lorentzian shape, peaked at @xmath164, as shown in fig. [fig : gain_dw__dpd](b), the maximum gain increases while approaching the parametric threshold, while the bandwidth shrinks to zero. however, the bandwidth can be considerably increased, maintaining rather high gain, by working away from the parametric threshold in the region where the gain peak splits, @xmath335, eq. . here a wide frequency plateau emerges around @xmath265, see fig. [fig : gain_dw__dpd](b), where the gain factor is nearly constant over a frequency interval given by the distance between the resonances, @xmath336. vs. input power @xmath216 for @xmath337 and @xmath338. inset : maximum gain factor g vs. @xmath216 and detuning @xmath169 ; black lines indicate the boundaries of the bistable region (wedge) cf. [fig : dpdbifurc_ain] ; white vertical lines indicate the parameter traces used in the main figure. the average output power @xmath339 is obtained from eq. in the presence of white gaussian noise in the input. (@xmath272, @xmath186).] the bifurcation regime of the cavity nonlinear response in the absence of parametric pumping is employed for dispersive qubit readout using jba @xcite, for a review see ref. and references therein. with this method, the phase shift of a reflected (or transmitted) probing signal is measured while ramping the signal amplitude. the result is sensitive to the detuning of the signal tone from the cavity resonance, which is pulled by the qubit by @xmath340, depending on the qubit state. one may take advantage of the high parametric gain for probing a qubit state by measuring the _ amplitude _ of the output signal instead of the phase shift. the amplified signal exhibits significant dispersion over the cavity - pump detuning thus providing high contrast for the qubit readout. the basis of the method can be understood from fig. [fig : meanrsq_ain] ; here the average output power is plotted against the input power for different detunings below the threshold, @xmath341. the bistability wedge for this pump strength is illustrated in the inset, compare also fig. [fig : dpdbifurc_ain]. the lowest three curves in fig. [fig : meanrsq_ain] correspond to values of the detuning within the monostable regions, either to the right or to the left of the critical bifurcation point, as indicated by white cuts in the inset (in the latter case, @xmath342, the ramped input signal should not cross the bifurcation line). the other two curves correspond to crossing through the bistability wedge or very close to the critical bifurcation point (here the average output power in the presence of classical noise is plotted, which then exhibits a gradual transition from the low- to the high - amplitude branch of the bifurcation curve). the output contrast is extremely sensitive to the detuning : it is up to factor of 10 for detunings differing by a linewidth @xmath104 already at rather small input power, @xmath343. in practice, a cavity frequency pull exerted by the qubit may be of the order @xcite @xmath344, i. e. comparable to the linewidth, @xmath345. the output contrast can be further enhanced by increasing the pump strength towards the threshold. it is also possible to ramp the pump strength rather than input power. the possibility to operate with several parameters gives room for further optimization. vs. detuning @xmath169 and pump strength @xmath88 for @xmath346. the black line separates bistable and tristable regimes, cf. [fig : homogeneous] ; the white line spans between two qubit - state dependent, effective detunings @xmath347 (@xmath348, @xmath349, @xmath350, @xmath186). inset : noise photon number @xmath351, eq., vs. @xmath169, @xmath88 as in main figure.] an alternative strategy for the dispersive qubit readout is provided by parametric _ radiation _ above threshold. this method, illustrated in fig. [fig : rsq_delta_epsilon], is based on the fact that in the absence of an input signal, @xmath352, the output signal is zero in the monostable region below the threshold (at blue detunings), @xmath353, while it is finite above the threshold, @xmath354, where it equals, @xmath355, according to eq. . the maximum contrast is achieved by choosing the pumping strength, @xmath356, and the optimum biasing detuning, @xmath357, as illustrated in fig. [fig : rsq_delta_epsilon]. such a choice guarantees that the blue shifted point, @xmath358, lies in the monostable region close to the threshold, while the red shifted point, @xmath359, lies in the bistable region and not in the tristable region where the trivial cavity state, @xmath360, dominates. then the output radiation power does not depend on @xmath88, @xmath361 this value is to be compared to the noise value in the monostable region below the threshold. the amplified vacuum noise is given by eq. in sec. [sec : quantum] and illustrated in the inset of fig. [fig : rsq_delta_epsilon], @xmath362 since the noise diverges at the threshold, the point @xmath358 is to be chosen not too close to the threshold. it is sufficient to depart from the threshold by @xmath363 to have the noise level, @xmath364. then for @xmath365, the radiation to noise contrast becomes, @xmath366
Quantum fluctuations of cavity field and emitted field
so far we discussed the classical regime of parametric resonance in the tunable cavity. in this section, we extend the formalism to the quantum regime, and investigate the quantum properties of the field inside the cavity, and of the output field. the hamiltonian description of the cavity parametric dynamics is a convenient starting point for the extension to the quantum regime. to this end we revisit eq. of sec. [subsec : hamiltonian] and impose canonical commutation relations, @xmath367 = { \text{i}}\hbar$], on the conjugated variables of the eigen modes of the closed cavity. these commutation relations obviously translate to the commutation relations for the resonant variables, @xmath368 = { \text{i}}\hbar$], because of the canonical nature of the transformations made in sec. [sec : parametric]. the unitary operator, which explicitly defines the corresponding quantum canonical transformation is@xmath369 averaging over rapid oscillations leads to the quantum hamiltonian coinciding with the one in eq. with quantum operators replacing respective classical variables. the quantization of the fundamental mode oscillator implies the quantization of the variable @xmath263 in terms of the conventional commutation relation for the annihilation operator, @xmath370=1 $]. due to the linear coupling of the cavity to the transmission line, eqs. -([eq : coupl_h]), the input - output formalism outlined in sec. [sec : losses] straightforwardly extends to the quantum regime. to this end, the classical amplitudes of the transmission line modes are to be replaced with the bosonic annihilation and creation operators, with @xmath371 = \delta(k - k')$]. from these commutation relations follows the commutation relation for the incoming field operator, @xmath372 = \delta(t - t')$], and similarly for the outgoing field operator @xmath138. the scattering relation, eq. , has the same form in the quantum regime, @xmath373 while the quantum langevin equation for the cavity operator @xmath374 becomes, @xmath375 this quantum langevin equation, together with eq. preserves the commutation relation @xmath376 = 1 $] for the cavity mode, as shown in appendix [sec : commutation]. the conservative part of eq. is a dynamical equation associated with the hamiltonian, @xmath377 the full analytical solution to the nonlinear quantum equation eq. is unknown. in what follows we restrict to the limit of small quantum fluctuations around the classical stationary states. such a restriction is valid far from the bifurcation points and the parametric threshold. some exact results for the critical fluctuations at such points can be found in literature @xcite, also quantum jumps in multistable regimes have been investigated @xcite. to study quantum fluctuations within the framework of a linearized quantum langevin equation, we assume the cavity field operators to be of the form, @xmath378, where @xmath82 is a steady state solution of the classical nonlinear equation, eq. , and @xmath379 describes small quantum fluctuations, @xmath380 similarly, we separate the classical amplitude and quantum fluctuations of the input field in the transmission line, @xmath381, @xmath382. then we expand eq. around @xmath82 up to linear order in the quantum fluctuation @xmath379 to obtain, @xmath383 herein we introduced the effective detuning @xmath384 and the (complex) pump strength @xmath385 by adding the terms proportional to the classical amplitude @xmath82. we note that @xmath82 itself depends on the bare parameters @xmath169 and @xmath88. quantitatively, the parameter regions where this approximation is valid are identified in appendix [sec : validity]. the analysis of eq. goes along the lines of sec. [subsec : detuned], where the response to a classical detuned signal was evaluated. by introducing fourier harmonics of the quantum fluctuations in the transmission line, @xmath386 and similarly in the cavity, the solution of the linear eq. is cast into the form, @xmath387 where @xmath388 cf. it follows from this equation, that modes with frequencies @xmath389 and @xmath390 are coupled pairwise by virtue of the parametric pumping. this property underlines the generation of correlated pairs of photons with frequencies @xmath391, which is analogous to the photon generation under non - degenerate parametric resonance. the denominator in eq. turns to zero at @xmath392, if the relation @xmath393 holds, leading to the divergence of fluctuations at the corresponding parameter values. this happens at the parametric threshold, and at the bifurcation points, and indicates the enhancement of critical fluctuations. the full power spectrum of the field in the cavity consists of the sharp line of the amplified (or generated) classical signal, @xmath394, together with the noise power spectrum, @xmath395, @xmath396 solving eq. and assuming thermal noise in the input field, @xmath397, where @xmath398, we calculate for the noise power spectrum @xmath399 at zero temperature, the noise power spectrum reduces to @xmath400 which can be interpreted as the amplified vacuum noise of the input, manifesting itself as real photons in the cavity. the noise power spectrum in eq. has a resonance structure equivalent to the resonances in the classical response to a detuned signal discussed in sec. [subsec : detuned]. the only difference is that now the effective pump parameters enter eqs. instead of the bare pump parameters, since we allow here for a finite classical amplitude @xmath82. accordingly, a single resonance at @xmath392 is observed under the condition @xmath401 and otherwise two resonances are found at @xmath402 cf. eqs. and. in fig. [fig : nafluc_w] the noise power spectrum @xmath403 is presented as a function of the pump detuning @xmath169 for @xmath404 and @xmath405. in the monostable regime, @xmath406, where @xmath407, the effective pump parameters in eq. are identical to the bare parameters, while in the bistable regime, @xmath174, with @xmath82 given by eq. , they are @xmath408 and @xmath409. the condition identifies the interval @xmath410 around the parametric threshold, where the resonance lies at @xmath411. outside that interval, once the resonance is split, the separation grows with the parameter distance from the threshold, both below and above the threshold. at the parametric threshold itself, @xmath173, the noise power diverges. the total number of photons in the cavity at zero temperature is @xmath412, with the noise power @xmath413 this quantity enters the validity criterium for the linearized langevin equation, eq. , which is analyzed in appendix [sec : validity]. of the cavity field, eq. , vs. pump detuning @xmath169 for @xmath406 the classical cavity amplitude is @xmath414, while @xmath415 for @xmath174 according to eq. . the resonances, eq. , are indicated by the grey lines. (@xmath404, @xmath416, @xmath186).] similar to the in - cavity field, the full power spectrum of the output field consists of the sharp line of the amplified (generated) classical signal, @xmath417, and the noise power spectrum @xmath418, @xmath419 the relation between the input and output field operators is similar to the one for a detuned classical signal in sec. [subsec : detuned], eqs. , @xmath420 with matrix elements now dependent on the effective pump parameters, @xmath421 and @xmath422 is given by eq. . the matrix elements obey the fundamental relation, @xmath423 which provides the correct commutation relation for the output operators, @xmath424 = \delta({\delta_k}- { \delta_k}')$]. equation describes an input - output relation for a linear non - degenerate amplifier @xcite with signal and idler modes having frequencies @xmath389 and @xmath390, respectively, while the input classical tone at @xmath392 plays the role of an additional pump. indeed, the renormalization of the generic pump parameters in eq. is an effect of this additional pump that increases the overall pump strength by @xmath425, and also affects the detuning @xmath169 similar to eq. . we note that eq. is valid both below and above the threshold, and in the latter case it includes the classical parametric radiation acting as an additional pump signal even in the absence of the classical input. with the corresponding renormalization of the quantity @xmath426 that characterizes the amplifier gain, eq., we cast the input - output relation, eq. into the form, @xmath427 where we introduced the standard notation for a non - degenerate parametric amplifier, @xmath428 the mapping in eq. is provided by a unitary two - mode squeezing operator @xcite, @xmath429 \hat b({\delta_k }) s^\dag[\xi],\\ & & s[\xi] = \exp\left(\int_{0}^\infty \!\! d{\delta_k}\left(\xi({\delta_k }) \hat b^\dag({\delta_k }) \hat b^\dag(-{\delta_k }) - \text{h.c. } \right)\right) \,, \nonumber\end{aligned}\]] where @xmath430. this implies that the stationary state of the output field is a pure state provided the input is a pure state. this is true in spite of because the evolution of the total system, including the cavity variable, is formally non - unitary due to the presence of the dissipative term in the langevin equation ([eq : eom_a_linearized]). the noise power spectrum of the output field can be computed from eq. , and for thermal noise input it reads, @xmath431.\end{aligned}\]] at zero temperature this equation reduces to @xmath432 and describes the generation of real photons from the vacuum under parametric resonance. this phenomenon is closely related to the dynamical casimir effect - the creation of real photons from vacuum fluctuations by an accelerated mirror @xcite. here the role of the moving mirror is played by the time - dependent boundary condition, driven by the modulated magnetic flux through the squid. the output noise, being proportional to @xmath395, inherits the resonant behavior of the noise power spectrum in the cavity, as discussed in sec. [sec : noisea] and shown on fig. [fig : nafluc_w]. in the deep subthreshold regime, for very weak pump strength, @xmath433, and in absence of an input signal, @xmath405, eq. takes the form, @xmath434 \left[\gamma^2 + ({ \delta_k}- \delta)^2\right] } \,.\]] in this limit the resonances move towards @xmath435, and the resonant structure of @xmath436, approaches the one computed in @xcite and observed in @xcite. the total photon flux in the output field is @xmath437, with the noise photon flux, @xmath438 at zero temperature. below the parametric threshold, @xmath439, the effective parameters in eq. are identical to the bare ones, and eq. reduces to @xmath440. above the threshold, @xmath441, with @xmath442 given by eq. , eq. becomes @xmath443 the output noise level is illustrated in fig. [fig : nctot_vac_dpd_eps] as a function of @xmath88 and @xmath169 for @xmath416. the right panel demonstrates the effect of back - bending of the threshold line due to the pump - induced frequency shift, eq. [fig : homogeneous](b) in sec. [sec : paramresonance]). the noise is enhanced at the parametric threshold and decreases while moving away from the threshold, there it is estimated as @xmath364 for @xmath444. , eq. , vs. @xmath169 and @xmath88, assuming (a) bare pump detuning, and (b) taking into account the pump - induced frequency shift, eq. . (@xmath405, @xmath186).] since the noise near the parametric threshold becomes strong, it is useful to evaluate the conditions for the output coherent signal dominating over the noise, @xmath445. above the parametric threshold, the signal - to - noise ratios are identical for the output field and the field inside the cavity (for @xmath416), @xmath446 therefore the limitation established by eq. for the field in the cavity applies as well to the output field, @xmath447 below the threshold, the maximum amplified signal is, according to eq., @xmath448 for @xmath202 and @xmath449. comparing this with eq. , we arrive at the constraint on the input signal, @xmath450 this bound is of order @xmath104 for @xmath451, and decreases both at weak pumping and close to the threshold. this is explained, at small @xmath452, by the fact that the amplification of vacuum noise is small, while the classical signal remains finite, and, close to the threshold, by the fact that amplification of the signal is more efficient than the amplification of the noise. the constraint in eq. is qualitatively similar to the one for the field inside the cavity given by eq. . a homodyne detection scheme allows for measurement of the quadratures of the output signal, and characterization of quadrature fluctuations @xcite. with this method, the output field is mixed with a strong classical field of a local oscillator, @xmath453, and the intensity of the mixed signal is measured. this intensity is proportional to the output field quadrature, @xmath454, @xmath455 the phase @xmath456 refers to the phase shift of the local oscillator with respect to the parametric pump ; variation of @xmath456 allows accessing all the quadratures individually. the mean quadrature is determined by the classical output signal @xmath457 separating the classical and quantum components, @xmath458, @xmath459, and using the spectral representation of the noise quadratures, @xmath460, we present the corresponding power spectrum in the form @xmath461, where @xmath462 is the squeezing power spectrum @xcite. note that by virtue of the stationary state of the cavity, @xmath463, hence only symmetric correlations between the sidebands contribute to the integral, i.e. the squeezing power characterizes the two - mode squeezing. we calculate the squeezing power assuming vacuum fluctuations of the input, using eq. for the output field operators. the result reads, @xmath464 \,.\end{aligned}\]] equation @xmath465 corresponds to pure vacuum fluctuations. the noise squeezing power varies with the phase @xmath456, the maximum and minimum values reached at @xmath466 and @xmath467, respectively, with @xmath468 { \text{re}}({\tilde \epsilon }) } { 2{\tilde \zeta}\gamma { \text{re}}({\tilde \epsilon }) + [\gamma^2-{\tilde \zeta}^2+|{\tilde \epsilon}|^2+{\delta_k}^2] { \text{im}}({\tilde \epsilon }) } .\end{aligned}\]] the corresponding extreme values are determined by the quantity @xmath469, and have the form, @xmath470 which is similar to the classical gain of the ideal amplifier, eq., including the relation, @xmath471. however, the maximum squeezing and maximum quadrature gain do not generally correspond to the same value of mixing phase @xmath456. moreover, the amplified classical signal, eq. , contains an additional phase, @xmath220, which is controlled by the input signal phase @xmath147. by varying the latter one may control the signal - to - noise ratio for the quadratures. in fig. [fig : squeeze_w_theta](a - b) the squeezing power @xmath472 is shown for @xmath405, @xmath202, and two different values of @xmath170. the @xmath456-values of maximum and minimum squeezing power are indicated by black lines. , eq. (a - b), and in - cavity noise @xmath473, eq. , (c - d) for @xmath474 (left column) and @xmath475 (right column). the quadrature phases for maximum and minimum squeezing power, eq. are indicated by the black lines. (@xmath202, @xmath416, @xmath186).] it is useful to also quantify the in - cavity squeezing, by calculating the squeezing power for the quadrature operator @xmath476 in analogy to eq. . although the phase @xmath456 in this case is not related to any externally tunable phase, it might be relevant for a quadrature - dependent coupling to a qubit placed in the cavity, or to another transmission line. assuming vacuum fluctuations in the input field we calculate the internal squeezing power using eq., @xmath477 \,.\end{aligned}\]] further evaluation of the minimum uncertainty of the cavity quadrature, @xmath478, results in the value @xmath479, as in the case of linear parametric amplifiers @xcite, i.e. a factor @xmath479 below the vacuum limit. in figs. [fig : squeeze_w_theta](c - d) @xmath480 is illustrated for @xmath416, @xmath202 and @xmath191, in comparison to the external squeezing @xmath481 of figs. [fig : squeeze_w_theta](a - b). as a consequence of the effective detuning @xmath384 in eq., @xmath482 is not symmetric around @xmath392, as is the case for @xmath481. , eq. , vs. @xmath88 for fixed @xmath483 and for @xmath484 (from bottom to top, the corresponding parameter values are marked with crosses on panel (b)). (b) @xmath485 and (c) gain @xmath486 vs. @xmath169 and @xmath147, for fixed @xmath487 (indicated by the vertical line in (a)). (@xmath488, @xmath186).] the two - mode squeezing is a nonclassical property of the amplified noise that originates from the production of noise photons in entangled pairs. further information about the nonclassical properties of the correlated output photons is provided by a two - photon correlation function, and characteristics of two - photon entanglement. we start with evaluating the second - order correlation function @xcite, @xmath489 in the presence of the classical output component, this equation takes the form, @xmath490 explicitly, using eq., we obtain for @xmath491 and input vacuum noise, @xmath492 \bigr] \,. \nonumber\end{aligned}\]] in fig. [fig : g2](a) the normalized correlation function, @xmath493, is presented as a function of the pumping strength @xmath88 for several values of the pump detuning @xmath169. in the duffing limit, @xmath494, all the terms in eq. vanish except of the first one, yielding the coherent state limit, @xmath495. the same is also true for large pumping strength above the threshold, @xmath496. this is explained by the rapid growth of classical radiation power that dominates over the fluctuations, @xmath497 (kinks on the curves at @xmath498). at the intermediate pump strengths both bunching (@xmath499) and antibunching (@xmath500) are possible. for pure output noise in the absence of classical output, @xmath501 (i.e. for @xmath416 below the threshold), only bunching occurs, @xmath502, where the degree of bunching exceeds that of classical chaotic radiation, @xmath503 this can be interpreted as a consequence of the pair production of noise photons. when @xmath504, also antibunching is possible @xcite due to the interplay between the classical and the quantum contribution to the correlation, last line in eq. . it occurs within a relatively narrow window of parameters, @xmath167, @xmath505, @xmath506, for which the phase dependence in the last term in eq. can introduce a sign change. the dependence of @xmath485 as a function of the input phase @xmath147 and the pump detuning @xmath169 is illustrated in fig. [fig : g2](b). pronounced antibunching (blue regions) is observed for @xmath507, and for those values of @xmath147 where the gain approaches unity, @xmath508, compare fig. [fig : g2](c). (solid) and logarithmic negativity @xmath509 (dashed) vs. pump strength @xmath88 for @xmath510 (from top to bottom). the vertical lines mark the values of @xmath88 at which the resonance, eq. , is encountered at the chosen value of @xmath389. (@xmath202, @xmath416, @xmath186).] the degree of entanglement between the two modes with frequencies @xmath389 and @xmath390 can be quantified with the entanglement entropy @xcite, @xmath511 where @xmath512 is the reduced density matrix of one of the involved modes. if these modes are entangled, the entropy takes a positive value, @xmath513. we compute the entanglement entropy for the amplified vacuum noise, using the two - photon wave function of the squeezed state, @xmath514 which is obtained by applying the squeezing operator, eq., to the vacuum input, @xmath515\,|0\rangle$], and using the decomposition equation @xcite. the reduced density matrix has the form, @xmath516 giving the entanglement entropy @xcite, @xmath517 the entropy is nonzero for all @xmath228, and follows closely the squeezing parameter @xmath518, asymptotically approaching the linear dependence, @xmath519, for @xmath520. the entanglement entropy @xmath521 is shown as function of @xmath88 in fig. [fig : logneg] (solid lines), for several values of the detuning @xmath389, and for @xmath405 and @xmath202. for small detuning, @xmath522, the entropy reaches the maximum at the threshold, @xmath232, at which @xmath523 [@xmath524 diverges. with increasing value of the detuning @xmath389 this maximum shifts towards the value of @xmath525, at which @xmath526 exhibits the resonance, eq. . the entropy rapidly decreases above the threshold, analogous to the behaviour of @xmath485, due to the emergence of the classical radiative state, @xmath527, that suppresses @xmath526. a convenient measure of entanglement for gaussian states is provided by the logarithmic negativity @xcite related to the covariance matrix for the two entangled modes. the covariance matrix @xmath528 is defined through a 4-vector composed of the quadratures, @xmath529, @xmath530 then splitting the covariance matrix into @xmath531 submatrices, @xmath532, the logarithmic negativity is defined as @xmath533 where @xmath534 and @xmath535. for entangled states the logarithmic negativity takes positive values. for amplified vacuum noise we obtain a simple result, using eq., @xmath536 i.e. the logarithmic negativity is equal to twice the squeezing parameter @xmath518. the logarithmic negativity @xmath537 is shown in fig. [fig : logneg] with dashed lines. its functional behavior is basically equivalent to that of the entropy @xmath521. our calculation shows that the degree of the two - mode entanglement is significantly enhanced in the presence of the parametric resonance. to evaluate the exact maximum entanglement value one needs to go beyond the quasilinear approximation and include the nonlinear effect. we make a qualitative estimate by taking the function @xmath538 at the threshold, @xmath232, @xmath202, and at @xmath392, and for the cavity field given by eqs. and assuming input power, @xmath539, corresponding to one photon per bandwidth. this yields an estimate, @xmath540 with a numerical constant of order one. this crude estimate seems to agree with more accurate evaluation of the critical fluctuations @xcite. for values @xmath541 achievable in tunable cavities, the entanglement entropy can accordingly reach the values @xmath542. this is significantly larger than the values calculated @xcite for a non - resonant open transmission line with modulated boundary, and also exceeds the values reported for experimental parametric josephson devices @xcite.
Summary
we have developed a consistent theory of parametric resonance in a high quality tunable superconducting cavity. we considered the nonlinear classical dynamics of the cavity both below and above the parametric threshold, and analyzed amplification of external signals, and parametric radiation. we also studied quantum properties of the amplified and radiative fields. the non - equidistance of the cavity frequency spectrum enabled us to formulate the theory of the degenerate parametric resonance in terms of the one encountered in a nonlinear parametric oscillator. we identified the parameters of this effective oscillator as functions of the cavity generic characteristics, and investigated the multistable cavity dynamics in a relevant range of the effective parameters. the operation of the device in the monostable regime as a nonlinear parametric amplifier is characterized with a phase - dependent differential gain, which increases at small input power and reaches the maximum value at the parametric threshold. we found that this maximum value scales with the ratio of the damping coefficient and the nonlinearity coefficient, @xmath323. we also found that the relation between the maximum and minimum gain for an ideal linear amplifier is violated in the nonlinear regime, @xmath543. extremely small values of @xmath89 available in tunable cavities allows for very large gain and strong amplification vs. deamplification contrast. amplification of detuned signals was found to exhibit sideband resonances within a specific region of the cavity parameters. this effect can be used for enhancing the amplification bandwidth while maintaining high gain. the application of the device as a parametric bifurcation amplifier was discussed in regard to dispersive qubit readout. the advantage of the parametric regime compared to the conventional jba is a high sensitivity of the strength of the output signal to the variation of the cavity frequency. this, together with a high amplification gain, provides a potential for improving the fidelity of qubit single shot readout. yet another suggested method for qubit readout is based on a high contrast between the strengths of parametric radiation above the threshold and amplified noise below the threshold. small - amplitude quantum fluctuations around the classical signal were investigated for the in - cavity field and the output field. the limit of small fluctuations is appropriate in a wide range of the device parameters except of small regions of critically enhanced fluctuations close to the bifurcation points and the parametric threshold. the theory is analogous to the one for a quantum linear amplifier. the strength of the amplified noise increases in the vicinity of the threshold in accord with the classical gain. the same is also true for the two - mode squeezing and the entanglement quantified with the entanglement entropy and the logarithmic negativity. at the threshold, the estimated magnitude of the squeezing parameter may reach the values of a few units, exceeding that achievable e.g. in non - resonant josephson mixers. the second order coherence is dominated by strong bunching for small classical inputs, resulting from the production of noise photons in pairs. however, for classical inputs with strength comparable to the vacuum noise, significant antibunching is predicted resulting from the interference of the classical and quantum field components. to conclude, we note that the developed theory straightforwardly extends to the regime of non - degenerate parametric resonance, when the pumping frequency is commensurate with a combination of cavity resonances. similarly, in this case, strongly enhanced amplification gain is to occur near the parametric threshold, as well as strongly enhanced two - mode squeezing and entanglement of the cavity modes selected by the resonance. yet another extension of the theory is readily done for a two - sided cavity parametrically pumped by two squids, attached to both sides of the cavity @xcite. the dynamics of this device is equivalent to the single - sided parametric cavity, provided the squids are operated at the same pump frequency. the parametric resonance is then controlled by an effective pump strength, which depends on the phase shift between the actual pumps. for equal pump amplitudes the parametric effect is maximum for the out - of - phase pumping (`` breathing '' mode), while for the in - phase pumping (`` translational '' mode) the parametric instability is completely suppressed. + _ acknowledgement _ we acknowledge useful discussions with chris wilson, per delsing, gran johansson, konrad lehnert, and tim duty. support from fp-7 ip solid is gratefully acknowledged.
Lagrangian and boundary condition of the flux-tunable cavity
in this appendix we derive the lagrangian of the flux - tunable cavity, eq. , and give arguments for its validity. we start with a description of the squid establishing the connection between the cavity and the pump line. the generalized coordinates of the squid are the superconducting phase @xmath544 at the cavity edge @xmath96, the phase @xmath3 dropping over the inductance @xmath545 of the squid loop, and the phase @xmath546 dropping over the coupling inductance @xmath547 of the pump line, see fig. [fig : device]. the squid is modelled as symmetric, with two identical josephson junctions, each having a josephson energy @xmath18 and a capacitance @xmath19. to simplify notation we assume that the squid is grounded in such a way that its geometric inductance @xmath545 is divided into two equal parts @xmath548, with a phase drop of @xmath23 over each part. thus, the phase difference on one of the josephson junctions is @xmath549 and @xmath550 on the other. the coupling to the flux line is inductive, with a mutual inductance @xmath551. the full squid lagrangian is @xmath552 or, written with the capacitive energy of a josephson junction @xmath37 and the inductive energy of the squid loop @xmath553 @xmath554 separating the @xmath20-dependent terms (first line) from the purely @xmath23-dependent ones (second line), @xmath555 + \mathcal{l}_{s}[f]$], the former can be combined with the bare cavity lagrangian @xmath556 with the inductive energy of the cavity @xmath557. together, these form the lagrangian @xmath32 of the flux - tunable cavity, eq. . for typical cavity and junction dimensions the orders of the three inductive energies in the lagrangian, eqs. -, are distinctly different. the dominant energy, @xmath558, determined by the small geometric inductance of the squid loop (@xmath559), is larger than the josephson energy of the squid, @xmath560, and that dominates over the inductive energy of the cavity, @xmath52 (for @xmath561). furthermore, the josephson plasma frequency @xmath562 is high compared to the fundamental cavity resonance, @xmath48 (compare sec. [subsec : cavmodes]). the equations of motion for @xmath20 and @xmath23, according to the full lagrangian @xmath563 $], @xmath564 describe two coupled nonlinear oscillators. for @xmath565 the equilibrium is @xmath566. in general, the coupled dynamics of nonlinear, driven oscillators features chaotic behaviour. we restrict our analysis to the case @xmath22 and assume that this is fulfilled even in the presence of a resonant excitation by the external field @xmath21. under this condition the equation of motion for @xmath23, eq. , decouples from the other oscillator, @xmath567 and the dynamical equation for @xmath20, eq. , then depends only parametrically on @xmath17, cf.. we suppose the external force of the form @xmath568, @xmath569, and separate the squid phase response @xmath570 into a constant equilibrium shift @xmath24, governed by the equation, @xmath571 and a small harmonic oscillation, @xmath572, driven by @xmath573, @xmath574 assuming @xmath575, we write the stationary solution in the form, @xmath576, @xmath577 where @xmath578 is the frequency of the @xmath23-oscillator, which is much larger than the frequency of the pump, @xmath579. linearized around the equilibrium shift @xmath24, eq. becomes @xmath580 \sin\phi_d \nonumber\\ & + & { e_{l,\text{cav}}}d \phi'_d = 0 \,.\end{aligned}\]] for @xmath581 this boundary condition determines the cavity mode spectrum, eqs. and . further expanding eq. to the second order with respect to @xmath106 leads to the pump induced shift of the cavity frequencies. indeed, averaging over time, we get a correction to the josephson energy, @xmath582, which will modify eq. accordingly. in particular, for the fundamental mode we get from eq. , @xmath583 one could also expand eq. to the second order with respect to @xmath106, which would lead, after the time averaging, to a shift of the static bias @xmath24, and eventually to an additional shift of the cavity frequencies. however, this effect is small, by virtue of the parameter @xmath584, compared to the shift ([eq : pump_frshift]).
Mode representation of cavity lagrangian
in this appendix, we express the lagrangian of the flux - tunable cavity, eq., in the mode representation, eq. , based on the expansion of the cavity field. firstly, making use of eq. , the overlap integrals of the non - orthogonal modes are @xmath585 where we have defined the coefficients @xmath67, eq. . with these, the bulk contribution to the cavity lagrangian becomes @xmath586 \nonumber\\ & + & \frac{1}{2 } \sum_{n, m } \bigl [- { 2c_j \over c_0 } \cos{k_n d } \cos{k_m d } \dot q_n \dot q_m \bigr.\nonumber\\ & & \bigl. + v^2 \cos{k_n d } \left({2c_j \over c_0 } k_m^2 \cos{k_m d } + k_m \sin{k_m d } \right) q_n q_m \bigr] \nonumber \,.\end{aligned}\]] in the remaining boundary contribution of eq. , we firstly separate a time dependent, nonlinear potential term @xmath587 from the harmonic contribution, @xmath588 the mode - representation, eq. , of the harmonic part becomes, using @xmath589, @xmath590 the first term of this cancels directly with a term in the bulk contribution, eq. . further, using the definition of the modes in eq. , we note that @xmath591 leading to further cancellation of terms between the bulk and the boundary contribution. the remaining terms are @xmath592 - v(q_n, t) \,,\end{aligned}\]] with @xmath593. this is the mode representation of the cavity lagrangian in eq. .
Transmission line amplitudes and scattering relation
in this appendix we show the relation of the flux amplitude @xmath594 introduced in eq. to the incoming field, and similarly for the flux amplitude @xmath194 of the outgoing field, as well as their mutual relation given in eq. . the incoming and outgoing fields in the transmission line are defined, respectively, as @xmath595 \\ \label{eq : phiout } \!\!\!\!\!\! \phi_{out}(t) \!&=&\! { e\over \hbar } \sqrt { \!\frac{\hbar}{\pi c_0 } } \!\int_0^{\infty } \!\!\!\! { d k \over \sqrt{\omega_k } } \!\!\left [a_k(t_1) e^{-{\text{i}}\omega_k (t - t_1) } \!+\! \text{h.c. } \right] .\end{aligned}\]] these are based on the solutions of eq. , which is expressed in terms of initial amplitudes @xmath596 at a time @xmath127 in the past, or in terms of final amplitudes @xmath597 at a time @xmath598 in the future, @xmath599 such a definition is justified, as will be shown below, by attributing different propagation directions along the transmission line for the incoming and outgoing field components, which can be separated by circulators and hence have physical meaning. we firstly use the solution to evaluate the field in the transmission line, eq. with @xmath600, @xmath601 } \\ & & + { 4e c_c \over \pi \hbar c_0 d}\sqrt{2 d \omega_0 \over c_0 m_0 } \int_{t_0}^t \!\!d t'p(t') \int_0^\infty \!\!\!d k \cos(\omega_k (t - t')) \cos{kx } \nonumber.\end{aligned}\]] the contribution from the first line can be straightforwardly identified with @xmath602 from eq. , and equals @xmath603. the integral in the second line is evaluated (for @xmath604) @xmath605 \nonumber \\ & = & \frac{d}{d t } \int_0^\infty \frac{d k}{kv } \left [\sin(k v(t - t') - kx) + \sin(k v(t - t') + kx) \right] \nonumber\\ & = & \frac{\pi}{2v } \frac{d}{d t } \left [\text{sgn}(t - t'- x / v) + \text{sgn}(t - t'+ x / v) \right] \nonumber \\ & = & \frac{\pi}{v } \left [\delta(t - t'- x / v) + \delta(t - t'+ x / v) \right] .\end{aligned}\]] the second @xmath169-function gives a contribution at @xmath606, whereas the first, for @xmath604, is not included in the integration limits, @xmath607, and therefore @xmath608 taken together, the field in the transmission line reads @xmath609 alternatively, the transmission line field, eq., can be evaluated from the second solution for the @xmath610, eq. , yielding @xmath611 by subtracting eqs. and at @xmath135 we can establish a relation between the incoming and the outgoing field components, @xmath612 note that the last term can also be expressed by the derivate of the cavity field at @xmath135, @xmath613, using @xmath614 in the weak coupling approximation. we can now evaluate eq. at @xmath615, and insert in eq., such that the transmission line field is expressed as a linear combination of @xmath602 and @xmath616 alone, @xmath617 demonstrating the role of @xmath602 and @xmath616 as incoming and outgoing field components. finally, we want to relate the general input - output relation, eq., with the slow varying amplitudes of the resonant approximation. to that end we separate the fast time oscillation with frequency @xmath75 in eqs. and ([eq : phiout]), @xmath618 where @xmath619 within the resonant approximation, @xmath620, @xmath621, these quantities coincide with the ones defined in sec. [sec : losses], cf. the cavity momentum is expressed in the rotating frame as well, @xmath622 with the slowly time - dependent cavity amplitude @xmath263. by setting these expressions into eq., multiplying with @xmath623 and averaging over fast oscillation, the corresponding input - output relation is obtained in the rotating frame, cf.
Commutation relations
in this appendix we show that the quantum langevin equation, eq. , preserves the commutation relation of the cavity amplitude. to this end we express the solution of eq. in terms of the propagator @xmath625 with the hamiltonian that governs the dynamics of the isolated cavity, @xmath626 (cf. the operator @xmath98 refers to the schrdinger picture and coincides initially with the heisenberg operator @xmath627. at time @xmath628 the solution is @xmath629 where we have used the fact that @xmath184 and @xmath630 commute since @xmath98 is uncorrelated with the operators @xmath596 of the incoming transmission line modes of which @xmath133 is composed. using this solution we are able to evaluate the equal time commutator, @xmath631 = e^{-2\gamma (t - t_0) } \bigl[\, u^{-1}(t, t_0) a(t_0) u(t, t_0)\,,\bigr. \\ & & \hspace*{2 cm } \bigl. u^{-1}(t, t_0) a^\dag(t_0) u(t, t_0) \,\bigr] + { \text{i}}\sqrt{2\gamma } e^{-\gamma (t - t_0)}\nonumber\\ & & \hspace*{0.5cm}\times \int_{t_0}^t \!\!d t'e^{-\gamma t'} \bigl (\left[u^{-1}(t, t_0) a(t_0) u(t, t_0),\, b^\dag(t') \right] \bigr. \nonumber\\ & & \phantom{\hspace*{0.5cm}\times \!\int_{t_0}^t \!\!d t'e^{-\gamma t'} } - \left[b(t'),\, u^{-1}(t, t_0) a^\dag(t_0) u(t, t_0) \right] \bigr) \nonumber\\ & & \hspace*{0.5cm}+ 2\gamma \iint_{t_0}^t d t'd t '' e^{-\gamma(t'+t '') } \left [b(t'),\, b^\dag(t '') \right] \,.\end{aligned}\]] the mixed commutators vanish, again with the argument of initially uncorrelated cavity and transmission line operators, leaving @xmath632 = } \nonumber\\ & & \hspace*{0.5 cm } e^{-2\gamma (t - t_0) } u^{-1}(t, t_0) \left [ a(t_0),\, a^\dag(t_0) \right] u(t, t_0) \nonumber\\ & & \hspace*{0.5 cm } + 2\gamma \iint_{t_0}^t d t'd t '' e^{-\gamma(t'+t '') } \left [b(t'),\, b^\dag(t '') \right] \,.\end{aligned}\]] finally, using @xmath633 = 1 $] and @xmath634 = \delta(t'-t'')$], we arrive at the desired result, @xmath635 = e^{-2\gamma (t - t_0) } - \left(e^{-2\gamma (t - t_0) } - 1 \right) = 1 \,.\]] the invariance of the commutation relation under the langevin evolution, eq. , follows from the correct combination of the damping term, @xmath636, and the fluctuations in the amplitude @xmath594. averaging over fluctuations would violate the exact unitary evolution and break the commutation relation.
Validity of quantum linearized treatment
having evaluated the magnitude of the quantum fluctuations, we are able to discuss the region of validity of the linearized equation, eq. . two assumptions have been made for the derivation : the amplified signal at frequency @xmath637 has been treated as a classical field, @xmath638, and its magnitude to exceed the amplified external noise, @xmath639. together these conditions are (cf. eq. ), @xmath640 we analyze these conditions separately above and below the threshold, at zero temperature, and at @xmath202 for simplicity. above the threshold, @xmath191, the parametric radiation dominates over the input signal. neglecting the input, @xmath416, we have @xmath641 in accord with eq.. then @xmath642 and @xmath643, and the amplified vacuum noise is, @xmath644 the conditions of eq., @xmath645 are fulfilled everywhere except of the close vicinity of the threshold. near the threshold the external noise dominates, while its role diminishes with growing pump strength. in the limit of very strong pumping, @xmath646, eq. reduces to @xmath647. this result can be understood from a purely hamiltonian argument. the semiclassical limit requires the quantum uncertainty of a state localized in a quantum well, @xmath648, to be much smaller than the total phase - space volume of the well. the latter can be estimated from the separatrix area, @xmath649, cf. second inset of fig. [fig : homogeneous](a). since in the semiclassical limit tunneling between the wells is exponentially suppressed, it is consistent to treat noise as local fluctuations in each well separately. below the threshold, @xmath306, eq. imposes constraints on the input field @xmath650. to be consistent with the linear description of fluctuations, we consider the quasilinear limit of the classical response, @xmath651. with this assumption, the maximum magnitude of the field in the cavity, eq. with @xmath315, reads @xmath652 while the amplified vacuum noise is @xmath653 the number of amplified vacuum photons inside the cavity is small at weak pumping but grows and passes the one - photon level at @xmath654, and becomes dominant while approaching the parametric threshold. using these estimates we extract from eq. the lower bound on the input signal, @xmath655 for very small pump strength, @xmath452, the constraint ([eq : bsq_min1]) reduces to @xmath656, which is qualitatively similar to a high quality duffing cavity, in which the resonant field fed by the input @xmath656 achieves a large (classical) value, @xmath638. the quasilinear approximation in this case, @xmath657 is valid as soon as @xmath658, and it imposes an upper bound on the input, @xmath659. close to the threshold, the amplified signal grows with @xmath88 more rapidly than the noise, and remains dominant at practically all input signals. this regime persists until the nonlinear effect breaks the quasilinear approximation at @xmath660, and the signal amplitude saturates. the corresponding constraint on the input reads, @xmath661 in terms of @xmath88, the upper bound for this regime is given by the condition @xmath662 for experimentally relevant cavity parameters, @xmath663, our estimates for the relative noise strength are therefore valid up to @xmath664. menzel, r. di candia, f. deppe, p. eder, l. zhong, m. ihmig, m. haeberlein, a. baust, e. hoffmann, d. ballester, k. inomata, t. yamamoto, y. nakamura, e. solano, a. marx, and r. gross, phys. lett. * 109 *, 250502 (2012). devoret, in _ quantum entanglement and information processing _, edited by d. esteve, j.m. raimond, and j. dalibard, proceedings of the les houches summer school of theoretical physics, lxiii, 1995 (elsevier, amsterdam, 2004). | we develop a theory of parametric resonance in tunable superconducting cavities.
the nonlinearity introduced by the squid attached to the cavity, and damping due to connection of the cavity to a transmission line are taken into consideration.
we study in detail the nonlinear classical dynamics of the cavity field below and above the parametric threshold for the degenerate parametric resonance, featuring regimes of multistability and parametric radiation.
we investigate the phase - sensitive amplification of external signals on resonance, as well as amplification of detuned signals, and relate the amplifier performance to that of linear parametric amplifiers.
we also discuss applications of the device for dispersive qubit readout. beyond the classical response of the cavity,
we investigate small quantum fluctuations around the amplified classical signals.
we evaluate the noise power spectrum both for the internal field in the cavity and the output field.
other quantum statistical properties of the noise are addressed such as squeezing spectra, second order coherence, and two - mode entanglement. | 1302.3484 |
Introduction
due to the importance of the primes, the mathematicians have been investigating about them since long centuries ago. in 1801, carl gauss, one of the greatest mathematician, submitted that the problem of distinguishing the primes among the non - primes has been one of the outstanding problems of arithmetic @xcite. proving the infinity of prime numbers by euclid is one of the first and most brilliant works of the human being in the numbers theory @xcite. greek people knew prime numbers and were aware of their role as building blocks of other numbers. more, the most natural question asked by human being was this what order prime numbers are following and how one could find prime numbers? until this time, there have been more attempts for finding a formula producing the prime numbers and or a model for appearance of prime numbers among other numbers and although they could be more helpful for developing the numbers theory, however, the complicated structure of prime numbers could not be decoded. during last years, the prime numbers attained an exceptional situation in the field of coding. for example, `` rsa '' system is one of the most applicable system in this field used in industries relying on prime numbers. `` rsa '' system is used in most computerized systems and counted as main protocol for secure internet connections used by states and huge companies and universities in most computerized systems @xcite. on 2004, manindra agrawal and his students in indian institute of technology kanpur could develop an algorithm called aks for detecting prime numbers @xcite. on 2006, 2008, 2009 and recently on 2013, mathematics students in a project called detecting the mersenne prime numbers by computer network gimps succeeded to discover the greatest prime number. all such cases indicate the importance of mersenne theorem or any other approach for finding the largest prime numbers @xcite. generalizing the mersenne theorem, this paper could accelerate finding the largest prime numbers. in addition, there have been provided new equations and algorithm for attaining the largest primes.
Generalization the mersennes theorem
assume that @xmath0 is a natural number greater than 1, @xmath1 related to n and natural numbers @xmath2 and @xmath3 are defined as below : @xmath4 if @xmath1 is a prime number, then @xmath0 is a prime number, too. if @xmath0 is not the prime number so we can write @xmath0 as the multiplication of two natural numbers except @xmath5 meaning : @xmath6 @xmath7 @xmath8 @xmath9 @xmath10 @xmath11 therefore, @xmath1 is not the prime number. so, @xmath0 must be a prime number. this theorem is a generalization for mersenne theorem in which @xmath2 and @xmath3 are arbitrary natural numbers. if in the theorem @xmath12, c is chosen as a multiple to @xmath2 and @xmath13, thus, @xmath1 will not be a prime number. suppose : @xmath14 therefore : @xmath15 @xmath16 @xmath17 @xmath18 @xmath19 @xmath20 the last equality shows that @xmath1 is not a prime number.
A specific state of generalized theorem
suppose @xmath0 is a natural number greater than @xmath5, function @xmath1 related to @xmath0 and natural number @xmath2 are defined as below : @xmath21 if @xmath1 is a prime number, then @xmath0 is a prime number, too. in this theorem @xmath22, based on @xmath0 constant, please consider a sequence @xmath23 we prove that sequence @xmath24 is strictly ascending, i.e. @xmath25 to prove the last inequality, we write : @xmath26 @xmath27 @xmath28 @xmath29 status 1. if @xmath0 is a multiple of @xmath30 : @xmath31 status 2. if @xmath0 is not a multiple of @xmath30 : @xmath32 therefore, inequity is accepted. in this theorem, each number is higher than mersenne number, meaning : @xmath33
Obtaining all primes in given interval
suppose @xmath2 be a natural number and @xmath34 are the primes smaller than or equal @xmath35 and @xmath36, @xmath37 are natural numbers which limitations are intended for them indicated as follows : @xmath38 assume that @xmath39 is a function of @xmath40 which is displayed as bellow : @xmath41 if the @xmath42 and @xmath37 circumstances are followed, @xmath39 can obtain all the primes less than @xmath2. knowing that @xmath39 is odd, because it is non prime, therefore it comprises from two odd numbers except @xmath5, and because @xmath43, @xmath39 has at least a prime factor @xmath44. therefore, @xmath39 is divided at least on one of the prime factors @xmath45. @xmath46 @xmath47 it is clear that above equalities are in discrepancy of the assumption of the theorem. 1. if : @xmath48 2. interval @xmath49 : + it is clear that by putting minimum @xmath37 in the definition @xmath50 minimum @xmath51 followed by minimum @xmath39 is obtained as below : @xmath52 according to recent equation, it is obvious that being as prime number in prime numbers smaller than @xmath53, r may not be divided into prime factors smaller than @xmath54. on the other hand, it is not necessary to see if prime numbers smaller than @xmath53 are divided into @xmath55 to detect it as a prime number. indeed, for obtaining the prime numbers, we only require @xmath56 in @xmath57 to enter the provision of prime factor @xmath58. if @xmath59 is considered as a prime number bigger than @xmath60, we could use @xmath61 instead of @xmath2 in this theorem because prime numbers smaller than @xmath35 include prime numbers smaller than @xmath62. prime numbers smaller than 120 : @xmath63 \ { @xmath64, prime numbers smaller than @xmath35 : @xmath65 } + @xmath66 @xmath67 @xmath68 @xmath69 @xmath70 @xmath71 @xmath72 @xmath73 @xmath74 @xmath75
Relation 1: determination of primes less than a given natural number
suppose @xmath2 be the natural number and @xmath76 are the primes smaller than or equal @xmath35 and also consider that @xmath77 are the primes larger than @xmath35. suppose that @xmath78 and @xmath42 be the members of the natural numbers and also @xmath79 be the members of the account numbers, these variables are selected arbitrarily. function @xmath39 related to values @xmath80, natural numbers @xmath81 and arithmetic numbers @xmath82 are defined as below : @xmath83 @xmath84 if @xmath56 be as the natural number less than @xmath2, then @xmath39 is a prime number. if @xmath39 is not prime number, it has a prime factor @xmath85. on the other side, because @xmath86, @xmath39 has at least one prime factor @xmath87. so, it is arbitrarily supposed that @xmath39 is divisible in @xmath88. @xmath89 @xmath90 because @xmath54 is not denominator of any @xmath91. we have : @xmath92 @xmath93 we reached a contradiction to the assumption. thus, the theorem was verified. we obtain prime numbers smaller than 119. + @xmath94 @xmath95 @xmath96 table 1 [cols="^,^,^,^,^,^",options="header ",] and continuing so suppose that : @xmath97 (the order is considered in the primes) general speaking, the theorem @xmath98 comes true to @xmath99 because @xmath100 includes the same primes. @xmath39 is obviously not divisible to @xmath101 and according to prime of the number @xmath39, we have : @xmath102 to attain prime numbers, we divide the intervals as below : @xmath103 with regard to the relationship easier to be written. in example of the primes less than @xmath104, the rang can be divided into three sections of @xmath105, @xmath106 and @xmath107. then, a distinct relation asserted for each. prime numbers smaller than 48 : @xmath108 @xmath109 @xmath110 @xmath111 and continuing so
An algorithm to obtain the largest prime numbers
by integrating the relations, particularly using the relation 2 and notation 5.4, we can attain an algorithm to obtain the largest prime number. ; one of them is as below : first of all, assign @xmath36 as one @xmath119 so as to obtain the minimum of @xmath0. then in the following equation, give a counter to @xmath0 through the minimum @xmath0 so long as k be a member of the natural numbers @xmath120. meanwhile, @xmath36 should not be even. @xmath121 if @xmath39 is not prime number, it has a prime factor @xmath85. on the other side, because @xmath125, @xmath39 has at least one prime factor @xmath126 therefore, with no interruption in the generality of the subject, we can assume that @xmath39 is divisible on @xmath127 : @xmath128 @xmath129 @xmath130 we reached a contradiction to the assumption. thus, the theorem was verified. 99 p. ribenboim, the little book of bigger primes, 2nd ed. springer science & business media, 2004. w. stein, `` elementary number theory : primes, congruences, and secrets '', book, pp. 1 - 172, 2009. r. crandall and c. b. pomerance, prime numbers : a computational perspective. springer science & business media, 2006. m. agrawal, n. kayal, and n. saxena, `` primes is in p '', ann. 781 - 793, 2004. `` list of known mersenne prime numbers - primenet ''. [online]. available : + http://www.mersenne.org/primes/ [accessed : 08-apr-2015]. | today, prime numbers attained exceptional situation in the area of numbers theory and cryptography. as we know, the trend for accessing to the largest prime numbers due to using mersenne theorem, although resulted in vast development of related numbers, however it has reduced the speed of accessing to prime numbers from one to five years.
this paper could attain to theorems that are more extended than mersenne theorem with accelerating the speed of accessing to prime numbers.
since that time, the reason for frequently using mersenne theorem was that no one could find an efficient formula for accessing to the largest prime numbers.
this paper provided some relations for prime numbers that one could define several formulas for attaining prime numbers in any interval ; therefore, according to flexibility of these relations, it could be found a new branch in the field of accessing to great prime numbers followed by providing an algorithm at the end of this paper for finding the largest prime numbers. | 1503.07688 |
Introduction
the large variety of novel and interesting phenomena of thin - film magnetism results very much from the fact that the magnetic anisotropy, which determines the easy axis of magnetization, can be one or two orders of magnitude larger than in the corresponding bulk systems@xcite. the reorientation transition (rt) of the direction of magnetization in thin ferromagnetic films describes the change of the easy axis by variation of the film thickness or temperature and has been widely studied both experimentally @xcite and theoretically @xcite. an instructive phenomenological picture for the understanding of the rt is obtained by expanding the free energy @xmath0 of the system in powers of @xmath1, where @xmath2 is the angle between the direction of magnetization and the surface normal. neglecting azimuthal anisotropy and exploiting time inversion symmetry yields : @xmath3 the anisotropy coefficients of second (@xmath4) and fourth (@xmath5) order depend on the thickness @xmath6 of the film as well as on the temperature @xmath7. away from the transition point usually @xmath8 holds, and, therefore, the direction of magnetization is determined by the sign of @xmath4 (@xmath9 : out - of - plane magnetization ; @xmath10 : in - plane magnetization). on this basis the concept of anisotropy flow @xcite immediately tells us that the rt is caused by a sign change of @xmath4 while the sign of @xmath5 mainly determines whether the transition is continuous (@xmath11) or step - like (@xmath12). in the case of a continuous transition @xmath5 also gives the width of the transition region. from the microscopic point of view we know that the magnetic anisotropy is exclusively caused by two effects, the dipole interaction between the magnetic moments in the sample and the spin - orbit coupling : @xmath13. while the dipole interaction always favors in - plane magnetization (@xmath14) due to minimization of stray fields, the spin - orbit interaction can lead to both, in - plane and out - of - plane magnetization depending sensitively on the electronic structure of the underlying sample. the spin - orbit anisotropy is caused by the broken symmetry@xcite at the film surface and the substrate - film interface as well as by possible strain@xcite in the volume of the film. it is worth to stress that a strong positive spin - orbit induced anisotropy alone opens up the possibility of an out - of - plane magnetized thin film. the rt must be seen as a competition between spin - orbit and dipole anisotropy. in many thin - film systems both thickness- and temperature - driven rts are observed. although it is clear by inspection of the corresponding phase diagrams @xcite that both types of transitions are closely related to each other, different theoretical concepts are needed to explain their physical origin. the thickness - driven rt is rather well understood in terms of a phenomenological separation of the spin - orbit induced anisotropy constant @xmath15 into a surface term @xmath16 and a volume contribution @xmath17 by the ansatz @xmath18. experimentally, this separation seems to provide a rather consistent picture@xcite despite the fact that in some samples additional structural transitions are present@xcite which clearly restrict its validity. on the theoretical side, basically two different schemes for the calculation of @xmath19 magnetic anisotropy constants have been developed, semi - empirical tight - binding theories@xcite and spin - polarized ab initio total - energy calculations @xcite. in both approaches the spin - orbit coupling is introduced either self - consistently or as a final perturbation. however, these investigations still remain to be a delicate problem because of the very small energy differences involved. neglecting the large variety of different samples, substrates, growth conditions, etc. it is useful for the understanding of the rt to concentrate on two somewhat idealized prototype systems both showing a thickness- as well as a temperature - driven rt. the `` fe - type '' systems @xcite are characterized by a large positive surface anisotropy constant @xmath20 together with a negative volume anisotropy @xmath21 due to dipole interaction. this leads to out - of - plane magnetization for very thin films. for increasing film thickness the magnetization switches to an in - plane direction because the volume contribution becomes dominating@xcite. as a function of increasing temperature a rt from out - of - plane to in - plane magnetization is found for certain thicknesses @xcite. in the `` ni - type '' systems @xcite, the situation is different. here the volume contribution @xmath21 is positive due to fct lattice distortion @xcite, thereby favoring out - of - plane magnetization, while the surface term @xmath20 is negative. for very thin films the surface contribution dominates leading to in - plane magnetization. at a critical thickness, however, the positive volume anisotropy forces the system to be magnetized in out - of - plane direction @xcite, until at a second critical thickness the magnetization switches to an in - plane position again caused by structural relaxation effects. here a so - called anomalous temperature - driven rt from in - plane to out - of - plane magnetization was found recently by farle et al.@xcite. in this article we will focus on the temperature - driven rt which can not be understood by means of the separation into surface and volume contribution alone. here the coefficients @xmath22 and @xmath23 need to be determined for each temperature separately. experimentally, this has been done in great detail for the second - order anisotropy of ni / cu(100)@xcite. the results clearly confirm the existence and position of the rt, but, on the other hand, do not lead to any microscopic understanding of its origin. to obtain more information on the temperature - driven rt theoretical investigations on simplified model systems have proven to be fruitful. despite the fact that in the underlying transition - metal samples the spontaneous magnetization is caused by the itinerant, strongly correlated 3d - electrons, up to now heisenberg - type models have been considered exclusively @xcite. the magnetic anisotropy has been taken into account by incorporating the dipole interaction and an uniaxial single - ion anisotropy to model the spin - orbit - induced anisotropy. using appropriate @xmath19 second - order anisotropy constants as input parameters, both types of rts have been observed within the framework of a self - consistent mean field approximation @xcite as well as by first - order perturbation theory for the free energy @xcite. a continuous rt has been found for @xmath24 layers@xcite taking place over a rather small temperature range. step - like transitions occur as an exception for special parameter constellations only. the rt is attributed to the strong reduction of the surface - layer magnetization relative to the inner layers for increasing temperature leading to a diminishing influence of the surface anisotropy. since the itinerant nature of the magnetic moments is ignored completely in these calculations, it is interesting to compare these results with calculations done within itinerant - electron systems. the present work employs a similar concept but in the framework of the single - band hubbard model@xcite which we believe is a more reasonable starting point for the description of temperature - dependent electronic structure of thin transition - metal films. the paper is organized in the following way : in the next section we define our model hamilton operator. in sec. [sec_f] we will focus on the derivation of the free energy and the second - order anisotropy constants by use of a perturbational approach. the isotropic part of the hamilton operator is treated in sec. [sec_sda]. here we present a generalization of a self - consistent spectral - density approach to the film geometry. in sec. [sec_res] we will show and analyze the results of the numerical evaluations and discuss the possibility of a temperature - driven rt within our model system. we will end with a short conclusion in sec. [sec_con].
Definition of the hamilton operator
the description of the film geometry requires some care. each lattice vector of the film is decomposed into two parts @xmath25 @xmath26 denotes a lattice vector of the underlying two - dimensional bravais lattice with @xmath27 sites. to each lattice point a @xmath6-atom basis @xmath28 (@xmath29) is associated referring to the @xmath6 layers of the film. the same labeling, of course, also applies for all other quantities related to the film geometry. within each layer we assume translational invariance. then a fourier transformation with respect to the two - dimensional bravais lattice can be applied. the considered model hamiltonian consists of three parts : @xmath30 @xmath31 denotes the single - band hubbard model @xmath32 where @xmath33 (@xmath34) stands for the annihilation (creation) operator of an electron with spin @xmath35 at the lattice site @xmath36, @xmath37 is the number operator and @xmath38 is the hopping - matrix element between the lattice sites @xmath36 and @xmath39. the hopping - matrix element between nearest neighbor sites is set to @xmath40. @xmath41 denotes the on - site coulomb matrix element, and @xmath42 is the chemical potential. the second term @xmath43 describes the dipole interaction between the magnetic moments on different lattice sites : @xmath44.\]] here @xmath45 is the distance between @xmath36 and @xmath39, and the unit vector @xmath46 is given by @xmath47. @xmath48 denotes the strength of the dipole interaction, @xmath49 the lattice constant and @xmath50 the number of atoms in the corresponding bulk cubic unit cell. @xmath51 are the spin operators constructed by the pauli spin matrices @xmath52. the expectation value of the spin operator yields the (dimensionless) magnetization vector @xmath53 the magnetization only depends on the layer index @xmath54 because of the assumed translational invariance within the layers. in our approach the uniaxial anisotropy caused by the spin - orbit coupling is taken into account phenomenologically by an effective layer - dependent anisotropy field coupled to the spin operator : @xmath55 the effective field @xmath56 is chosen to be parallel to the film normal @xmath57. to ensure the right symmetry we set : @xmath58 this corresponds to a mean - field treatment of the spin - orbit - induced anisotropy@xcite. the strengths of the anisotropy fields @xmath59 enter as additional parameters and have to be fixed later.
Calculation of the magnetic anisotropy constants
the direction of the magnetization is determined by the minimal free energy @xmath0. the anisotropic contributions @xmath60 and @xmath43 to the hamilton operator ([ham_op]) can be considered as a small perturbation to the isotropic hubbard model (@xmath61). then we can apply a thermodynamic perturbation expansion@xcite of the free energy @xmath0 up to linear order with respect to @xmath62 : @xmath63 here @xmath64 denotes the expectation value taken within the unperturbated hubbard model @xmath31. on the same footing we use a mean - field decoupling for the two - particle expectation values contained in @xmath65. because @xmath0 is calculated to linear order in the anisotropy contributions, only the lowest, i.e. second - order anisotropy constants are considered, and a possible canted phase is neglected. the ratio @xmath66 and thus the width of the transition region have been found to be very small for reasonable strengths of the anisotropy contributions @xcite. within our approach it is, therefore, sufficient to consider the free - energy difference @xmath67 between in - plane and out - of - plane magnetization : @xmath68 @xmath10 and @xmath9 indicate in - plane and out - of - plane magnetization respectively. hence, the reorientation temperature @xmath69 is given by the condition @xmath70. evaluation of @xmath67 yields : @xmath71 the constants @xmath72 contain the effective dipole interaction between the layers and can be calculated separately : @xmath73.\]] @xmath74 is the angle between @xmath46 and the direction of the magnetization. the @xmath72 only depend on the film geometry. for thick films @xmath75 reduces to its continuum value @xmath76 where @xmath77 is the magnetization per atom. to calculate @xmath67 the temperature- and layer - dependent magnetizations @xmath78 of the hubbard film are needed.
Spectral-density approach to the hubbard film
in this section we will focus on the evaluation of the hubbard film @xmath31. ferromagnetism in the hubbard model is surely a strong - coupling phenomenon. the existence of ferromagnetic solutions was recently proven in the limit of infinite dimensions by quantum monte - carlo calculations@xcite. ferromagnetism is favored by a strongly asymmetric bloch density of states (bdos) and by a singularity at the upper band edge as it is found, e. g., for the fcc lattice. the hubbard model constitutes a highly non - trivial many - body problem even for a periodic infinitely extended lattice. even more complications are introduced when the reduction of translational symmetry has to be taken into account additionally. one of the easiest possible approximations to treat a hubbard film is a hartree - fock decoupling, which has been applied previously @xcite. hartree - fock theory, however, is necessarily restricted to the weak - coupling regime and is known to overestimate the possibility of ferromagnetic order drastically. neglecting electron - correlation effects altogether leads to qualitatively wrong results especially for intermediate and strong coulomb interaction @xmath41 @xcite. furthermore, we did not find a realistic temperature - driven rt within this approach. thus we require an approximation scheme which is clearly beyond the hartree - fock solution and takes into account electron correlations more reasonably. on the other hand, it must be simple enough to allow for an extended study of magnetic phase transitions in thin films. for this purpose we apply the spectral - density approach (sda)@xcite which is motivated by the rigorous analysis of harris and lange@xcite in the limit of strong coulomb interaction (@xmath79). the sda can be interpreted as an extension of their @xmath80-perturbation theory@xcite in a natural way to intermediate coupling strengths and finite temperatures and has been discussed in detail for various three - dimensional @xcite as well as infinite - dimensional@xcite lattices. at least qualitatively, it leads to rather convincing results concerning the magnetic properties of the hubbard model. a similar approach has been applied to a multiband hubbard model with surprisingly accurate results for the magnetic key - quantities of the prototype band ferromagnets fe, co, ni@xcite. recently, a generalization of the sda has been proposed to deal with the modifications due to reduced translational symmetry@xcite. in the following we give only a brief derivation of the sda solution and refer the reader to previous papers for a detailed discussion@xcite. the basic quantity to be calculated is the retarded single - electron green function @xmath81. from @xmath82 we obtain all relevant information on the system. its diagonal elements, for example, determine the spin- and layer - dependent quasiparticle density of states (qdos) @xmath83. the equation of motion for the single - electron green function reads : @xmath84 g_{lj\sigma}^{\gamma\beta}(e) = \hbar\delta_{ij}^{\alpha\beta}.\]] here we have introduced the electronic self - energy @xmath85 which incorporates all effects of electron correlations. we adopt the local approximation for the self - energy which has been tested recently for the case of reduced translational symmetry@xcite. if we assume translational invariance within each layer of the film we have @xmath86. after fourier transformation with respect to the two - dimensional bravais lattice the equation of motion ([eq_motion]) is formally solved by matrix inversion. the decisive step is to find a reasonable approximation for the self - energy @xmath87. guided by the exactly solvable atomic limit of vanishing hopping (@xmath88) and by the findings of harris and lange in the strong - coupling limit (@xmath89), a one - pole ansatz for the self - energy @xmath87 can be motivated @xcite. the free parameter of this ansatz are fixed by exploiting the equality between two alternative but exact representations for the moments of the layer - dependent quasiparticle density of states : @xmath90_{+}\bigg\rangle_{t = t'}.\]] here, @xmath91_{+}$] denotes the anticommutator. it can be shown by comparing various approximation schemes @xcite that an inclusion of the first four moments of the qdos (@xmath92) is vital for a proper description of ferromagnetism in the hubbard model. further, the inclusion of the first four moments represents a necessary condition to be consistent with the @xmath80-perturbation theory@xcite. the hartree - fock approximation recovers the first two moments (@xmath93) only, while the so - called hubbard - i solution@xcite reproduces the third moment (@xmath94) as well, but is well known to be hardly able to describe ferromagnetism. taking into account the first four moments to fix the free parameters of our ansatz we end up with the sda solution which is characterized by the following self - energy : @xmath95 the self - energy depends on the spin - dependent occupation numbers @xmath96 as well as on the so - called bandshift @xmath97 which consists of higher correlation functions : @xmath98 a possible spin dependence of @xmath97 opens up the way to ferromagnetic solutions@xcite. ferromagnetic order is indicated by a spin - asymmetry in the occupation numbers @xmath99, and the layer - dependent magnetization is given by @xmath100. the band occupations @xmath101 are given by @xmath102 where @xmath103 is the fermi function. the mean band occupation @xmath104 is defined as @xmath105. although @xmath106 consists of higher correlation functions it can by expressed exactly @xcite via @xmath107 and @xmath108 : @xmath109 \rho_{\alpha\sigma}(e). \label{bas}\end{aligned}\]] equations ([eq_motion]), ([sig_sda]), ([nas]) and ([bas]) build a closed set of equations which can be solved self - consistently. surely a major short - coming of the sda is the fact that quasiparticle damping is neglected completely. recently a modified alloy analogy (maa) has been proposed @xcite which is also based on the exact results of the @xmath80-perturbation theory but is capable of describing quasiparticle damping effects as well. for bulk systems it has been found that the magnetic region in the phase diagram is significantly reduced by inclusion of damping effects. on the other hand, the qualitative behavior of the magnetic solutions is very similar to the sda. an application of the maa to thin film systems is in preparation@xcite.
Results and discussion
the numerical evaluations have been done for an fcc(100) film geometry. in this configuration each lattice site has four nearest neighbors within the same layer and four nearest neighbors in each of the respective adjacent layers. we consider uniform hopping @xmath110 between nearest neighbor sites @xmath36, @xmath39 only. energy and temperature units are chosen such that @xmath111. the on - site hopping integral is set to @xmath112. further, we keep the on - site coulomb interaction fixed at @xmath113 which is three times the band width of the three - dimensional fcc - lattice and clearly refers to the strong - coupling regime. let us consider the isotropic hubbard - film first. there are three model parameters left to vary, the temperature @xmath7, the thickness @xmath6 and the band occupation @xmath114). except for the last part of the discussion we will keep the band occupation fixed at the value @xmath115 and focus exclusively on the temperature and thickness dependence of the magnetic properties. in fig. [fig_1] the layer - dependent density of states of the non - interacting system @xmath116 (@xmath117 `` bloch density of states '' (bdos)) is plotted for a two, three and five layer film with an fcc(100) geometry. the bdos is strongly asymmetric and shows a considerable layer - dependence for @xmath118. considering the moments @xmath119 of the bdos yields that the variance @xmath120 as well as the skewness @xmath121 are reduced at the surface layer compared to the inner layers due to the reduced coordination number at the surface (@xmath122, @xmath123, @xmath124, @xmath125 for @xmath126). in fig. [fig_2] the layer magnetizations @xmath78 as well as the mean magnetization @xmath127 are shown as a function of temperature @xmath7. while symmetry requires the double layer film to be uniformly magnetized, the magnetization shows a strong layer dependence for @xmath118. the magnetization curves of the inner layers (and for the double layer) show the usual brillouin - type behavior. the trend of the surface layer magnetization (@xmath24), however, is rather different. note that @xmath128 depends almost linearly on temperature in the range @xmath129 and for thicknesses @xmath130. compared to the inner layers, the surface magnetization decreases significantly faster as a function of temperature, tending to a reduced curie temperature. however, due to the coupling between surface and inner layers which is induced by the electron - hopping, this effect is delayed and a unique curie - temperature for the whole film is found. the curie temperature @xmath131 increases as a function of the film thickness @xmath6 and saturates already for film - thicknesses around @xmath132 to the corresponding bulk value. a similar behavior was found for a bcc(110) film geometry@xcite. the critical exponent of the magnetization (fig. [fig_3]) is found to be equal to the mean - field value @xmath133 for all thicknesses and all other parameters considered. this clearly reveals the mean - field type of our approximation which is due to the local approximation for the electronic self - energy. the layer - dependent quasiparticle density of states (qdos) is shown in fig. [fig_4] for three temperatures @xmath134. two kinds of splittings are observed in the spectrum. due to the strong coulomb repulsion @xmath41 the spectrum splits into two quasiparticle subbands (`` hubbard splitting '') which are separated by an energy of the order @xmath41. in the lower subband the electron mainly hops over empty sites, whereas in the upper subband it hops over sites which are already occupied by another electron with opposite spin. the latter process requires an interaction energy of the order of @xmath41. the weights of the subbands scale with the probability of the realization of these two situations while the total weight of the qdos of each layer is normalized to 1. therefore, the weights of the lower and upper subbands are roughly given by @xmath135 and @xmath136 respectively. this scaling becomes exact in the strong coupling limit (@xmath89). since the total band occupation (@xmath115) is above half - filling (@xmath137), the chemical potential @xmath42 lies in the upper subband while the lower subband is completely filled. for temperatures below the curie temperature @xmath131 an additional splitting (`` exchange - splitting '') in majority (@xmath138) and minority (@xmath139) spin direction occurs, leading to non - zero magnetization @xmath140. for @xmath19 the majority qdos lies completely below the chemical potential, the system is fully polarized (@xmath141). thus the low - energy subband of the minority spin direction disappears and the minority qdos is exactly given by the bdos of the non - interacting system (fig. [fig_1]) due to vanishing correlation effects in the @xmath139 channel. the reduced surface magnetization at @xmath19 (see fig. [fig_2]) is, therefore, directly caused by the layer - dependent bdos. we like to stress that the spin - splitting does not depend on the size of the coulomb interaction @xmath41 as long as @xmath41 is chosen from the strong coupling limit. contrary to hartree - fock theory the spin splitting saturates as a function of @xmath41 for values of about @xmath142 times the bandwidth of the non - interacting system. the same holds for the curie temperature @xmath131 @xcite. the temperature behavior of the qdos is governed by two correlation effects (fig. [fig_4]). as the temperature is increased, the spin - splitting between @xmath138 and @xmath139 spectrum decreases. this effect is accompanied by a redistribution of spectral weight between the lower and upper subbands along with a change of the widths of the subbands. for @xmath143 one clearly sees that in the minority spectrum weight has been transferred from the upper to lower subband which has reappeared due to non - saturated magnetization at finite temperatures. in the @xmath138 spectrum the opposite behavior is found. for all @xmath144 the spin splitting is significantly larger in the inner layer compared with the surface. at @xmath145 the exchange splitting has disappeared whereas the correlation - induced hubbard splitting is still present. let us consider the question why the surface magnetization shows a tendency towards a reduced curie temperature as seen in fig. [fig_2]. this effect can be understood by the above mentioned moment analysis of the bdos (fig. [fig_1]) : from bulk systems it is known that an asymmetrically shaped bdos is favorable for the stability of ferromagnetism in the hubbard model. in particular, the curie temperatures increase with increasing skewness of the bdos @xcite. the same trend shows up in the present film - system where the skewness of the bdos is higher for the inner layers compared to the surface - layer. note that this argument is somewhat more delicate than in the case of heisenberg films where the reduced surface magnetization is directly caused by the reduced number of interacting sites. in fig. [fig_5] the difference between surface magnetization @xmath128, central - layer magnetization @xmath146, and mean magnetization @xmath147 is analyzed in more detail. for all film thicknesses where this distinction is meaningful (@xmath118), the surface magnetization is reduced with respect to the mean magnetization. this holds not only for very thin films (@xmath132, see also fig. [fig_2]) where some oscillations are present that are caused by the finite film thickness, but also extends to the limit @xmath148 where the two surfaces are well separated and do not interact. the surface and central - layer magnetizations already stabilize for thicknesses around @xmath149. further, fig. [fig_5] clearly shows that the reduction of @xmath128 drastically increases for higher temperatures. for @xmath143 the surface magnetization is reduced to about half the size of the magnetization in the center of the film. the charge transfer due to differing layer occupations @xmath150 is found to be smaller than @xmath151 at @xmath19 and is almost negligible for finite temperatures. we now like to focus on the magnetic anisotropy energy within the model system ([ham_op]). the second - order anisotropy constant @xmath152 is calculated via eq. ([f_anis_end]) which needs as an input the temperature - dependent layer magnetizations of the hubbard film. the dipole constants @xmath72 for an fcc(100) film geometry are found to be : @xmath153, @xmath154, @xmath155, and are set to zero for @xmath156. to simulate both, surface and volume contribution of the spin - orbit induced anisotropy, we choose the effective anisotropy field ([anis_field]) in the surface - layer to be different from its value in the volume of the film : @xmath157 in the perturbational approach only the ratio @xmath158 is important. thus we are left with only two parameters (fixed at @xmath19) to model the rt. in principle these constants could be taken from experiment or from theoretical ground - state calculations. in our opinion, however, this would mean to somewhat over - judge the underlying rather idealized model system. we are mainly interested in the question whether a realistic temperature - driven rt is possible at all in the hubbard film and by what mechanism it is induced. therefore, we choose the parameters @xmath159 and @xmath160 conveniently, guided however by the experimental findings in the fe - type and ni - type scenarios described in the introduction. [fig_6] and fig. [fig_7] show that for appropriate parameters @xmath161 both types of temperature - driven rt can be found within a three - layer film. the same is found for any film thickness @xmath24. in the fe - type situation (fig. [fig_6]) we consider a strong positive surface anisotropy field @xmath162 together with @xmath163. at low temperatures the system is magnetized in out - of - plane direction. as the temperature increases, @xmath164 decreases faster than @xmath165 because of the strong reduction of the surface anisotropy and the magnetization switches to an in - plane position. in principle this kind of rt is possible for all film thicknesses @xmath24. for thicker films @xmath159 has to be rescaled proportional to @xmath6 to compensate the increasing importance of the dipole anisotropy. the ni - type rt from in - plane to out - of - plane magnetization for a three layer system is obtained by a positive volume anisotropy field @xmath166 and a negative surface anisotropy @xmath167. at low temperatures the dipole anisotropy as well as the negative surface anisotropy field lead to an in - plane magnetization. for higher temperatures, however, where the surface anisotropy becomes less important because of the reduced surface magnetization, the positive volume anisotropy field forces the magnetization to switch to an out - of - plane direction. the ratio @xmath168 determines for what thickness @xmath6 this type of temperature - driven rt is possible and scales like @xmath169 for thicker films. note that for both types of rt the values of @xmath159 and @xmath160 are chosen in such a way that the system is close to a thickness - driven rt. in both cases the rt is mediated by the strong decrease of the surface - layer magnetization compared to the inner layers as a function of temperature. finally, we consider the dependence on the band occupation @xmath104. in fig. [fig_8] the ratio @xmath170 of a three - layer film is plotted as a function of the reduced temperature @xmath171 for different band occupations @xmath172. above @xmath173 and below @xmath174 the fcc(100) hubbard film does not have ferromagnetic solutions whereas between @xmath174 and @xmath175 there is a tendency towards first order phase transition as a function of temperature@xcite and no realistic rt is possible. the ratio @xmath170 (fig. [fig_8]) decreases as a function of temperature for all band occupations. for @xmath115 which has been considered in fig. [fig_6] and fig. [fig_7], it changes from @xmath176 at @xmath19 to @xmath177 close to @xmath131. however, for higher band occupations this strong temperature - dependent change in @xmath170 diminishes. from the discussion above it is clear that this is unfavorable for the rt. we can thus conclude from fig. [fig_8] that the possibility of a temperature - driven rt sensitively depends on the band occupation @xmath104. in addition we plotted in fig. [fig_8] the ratio @xmath170 calculated within hartree - fock theory. the result is completely different. at low temperatures we find @xmath178 whereas @xmath179 close to @xmath131. note that the ratio @xmath170 is almost constant for the wide temperature range @xmath180. therefore, a realistic temperature - driven rt is excluded within the hartree - fock approximation. this holds for all parameters that have been considered here.
Conclusion
we have applied a generalization of the spectral - density approach (sda) to thin hubbard films. the sda which reproduces the exact results of the @xmath80-perturbation theory in the strong coupling limit, leads to rather convincing results concerning the magnetic properties. the magnetic behavior of the itinerant - electron film can be microscopically understood by means of the temperature - dependent electronic structure. for an fcc(100) film geometry the layer - dependent magnetizations have been discussed as a function of temperature as well as film thickness. the magnetization in the surface layer is found to be reduced with respect to the inner layers for all thicknesses and temperatures. by analyzing the layer - dependent qdos this reduction can be explained by the fact that in the free bdos both variance and skewness are diminished in the surface - layer compared to the inner layers. the inclusion of the dipole interaction and an effective layer - dependent anisotropy field allows to study the temperature - driven rt. the second - order anisotropy constants have been calculated within a perturbational approach. for appropriate strengths of the surface and volume anisotropy fields both types of rt, from out - of - plane to in - plane (fe - type) and from in - plane to out - of - plane (ni - type) magnetization are found. for the ni - type scenario the inclusion of a positive volume anisotropy is necessary. the rt in our itinerant model system is mediated by a strong reduction of the surface magnetization with respect to the inner layers as a function of temperature. here a close similarity to the model calculations within heisenberg - type systems is apparent, despite the fact that these models completely ignore the itinerant nature of the magnetic moments in the underlying transition - metal samples. contrary to heisenberg films, the band occupation @xmath104 enters as an additional parameter within an itinerant - electron model. we find that the possibility of a rt sensitively depends on the band occupation. the fact that no realistic rt is possible within hartree - fock theory clearly points out the importance of a reasonable treatment of electron correlation effects. | the temperature - driven reorientation transition which, up to now, has been studied by use of heisenberg - type models only, is investigated within an itinerant - electron model.
we consider the hubbard model for a thin fcc(100) film together with the dipole interaction and a layer - dependent anisotropy field.
the isotropic part of the model is treated by use of a generalization of the spectral - density approach to the film geometry.
the magnetic properties of the film are investigated as a function of temperature and film thickness and are analyzed in detail with help of the spin- and layer - dependent quasiparticle density of states. by calculating the temperature dependence of the second - order anisotropy constants
we find that both types of reorientation transitions, from out - of - plane to in - plane (`` fe - type '') and from in - plane to out - of - plane (`` ni - type '') magnetization are possible within our model. in the latter case
the inclusion of a positive volume anisotropy is vital.
the reorientation transition is mediated by a strong reduction of the surface magnetization with respect to the inner layers as a function of temperature and is found to depend significantly on the total band occupation.
-4 mm | cond-mat9804301 |
Introduction
non - relativistic quantum fluids (fermions or bosons) constrained by periodic structures, such as layered or tubular, are found in many real or man - made physical systems. for example, we find electrons in layered structures such as cuprate high temperature superconductors or semiconductor superlattices, or in tubular structures like organo - metalic superconductors. on the experimental side, there are a lot of experiments around bosonic gases in low dimensions, such as : bec in 2d hydrogen atoms @xcite, 2d bosonic clouds of rubidium @xcite, superfluidity in 2d @xmath1he films @xcite, while for in 1d we have the confinement of sodium @xcite, to mention a few. meanwhile, for non - interacting fermions there are only a few experiments, for example, interferometry probes which have led to observe bloch oscillations @xcite. to describe the behavior of fermion and boson gases inside this symmetries, several works have been published. for a review of a boson gas in optical lattices see @xcite, and for fermions @xcite is very complete. most of this theoretical works use parabolic @xcite, sinusoidal @xcite and biparabolic @xcite potentials, with good results only in the low particle energy limit, where the tight - binding approximation is valid. although in most of the articles mentioned above the interactions between particles and the periodic constrictions are taken simultaneously in the system description, the complexity of the many - body problem leads to only an approximate solution. so that the effects of interactions and constrictions in the properties of the system, are mixed and indistinguishable. in this work we are interested in analyzing the effect of the structure on the properties of the quantum gases regardless of the effect of the interactions between the elements of the gas, which we do as precisely as the accuracy of the machines allows us to do. this paper unfolds as follows : in sec. 2 we describe our model which consists of quantum particles gas in an infinitely large box where we introduce layers of null width separated by intervals of periodicity @xmath2. in sec. 3 we obtain the grand potential for a boson and for a fermion gas either inside a multilayer or a multitube structure. from these grand potentials we calculate the chemical potential and specific heat, which are compared with the properties of the infinite ideal gas. in sec. 4 we discuss results, and give our conclusions.
Quantum gases whithin multilayers and multitubes
we consider a system of @xmath3 non - interacting particles, either fermions or bosons, with mass @xmath4 for bosons or @xmath5 for fermions respectively, within layers or tubes of separation @xmath6, @xmath7 = @xmath8 or @xmath9, and width @xmath10, which we model as periodic arrays of delta potentials either in the @xmath11-direction and free in the other two directions for planes, and two perpendicular delta potentials in the @xmath8 and @xmath9 directions and free in the @xmath11 one for tubes. the procedure used here is described in detail in refs. @xcite and @xcite for a boson gas, where we model walls in all the constrained directions using dirac comb potentials. in every case, the schrdinger equation for the particles is separable in @xmath8, @xmath9 and @xmath11 so that the single - particle energy as a function of the momentum @xmath12 is @xmath13. for the directions where the particles move freely we have the customary dispertion relation @xmath14, with @xmath15, @xmath16, and we are assuming periodic boundary conditions in a box of size @xmath17. meanwhile, in the constrained directions, @xmath11 for planes and @xmath18 for tubes, the energies are implicitly obtained through the transcendental equation @xcite @xmath19 with @xmath20, and the dimensionless parameter @xmath21 represents the layer impenetrability in terms of the strength of the delta potential @xmath22. we redefine @xmath23, where @xmath24 is the thermal wave length of an ideal gas inside an infinite box, with @xmath25 the fermi energy and @xmath26 is the density of the gas. the energy solution of eq. ([kpsol]) for has been extensively analized in refs. @xcite and @xcite, where the allowed and forbidden energy - band structure is shown, and the importance of taking the full band spectrum has been demonstrated.
Thermodynamic properties of quantum gases in multilayers and in multitubes
every thermodynamic property may be obtained starting from the grand potential of the system under study, whose generalized form is @xcite @xmath27\bigr\ }, \label{omega}\]] where @xmath28 for bosons, 1 for fermions and 0 for the classical gas, @xmath29 is the kronecker delta function and @xmath30. the ground state contribution @xmath31, which is representative of the bose gas, is not present when we analyze the fermi gas. for a boson gas inside multilayers we go through the algebra described in @xcite, and taking the thermodynamic limit one arrives to @xmath32\bigr) \nonumber \\ & & -\frac{1}{\beta ^{2}}\frac{l^{3}m}{\left (2\pi \right)^{2}\hbar ^{2 } } { \int_{-\infty } ^{\infty } dk_{z}}g_{2 } \bigl\{\exp [-\beta (\varepsilon _ { k_{z}}-\mu)]\bigr\}. \label{omegaboson}\end{aligned}\]] meanwhile, for a fermion gas we get @xmath33\bigr\ }, \label{omegafermion}\]] where @xmath34 and @xmath35 are the bose and fermi - dirac functions @xcite. the spin degeneracy has been taken into account for the development of eq. ([omegafermion]). on the other hand, for a multitube structure we have @xmath36 \nonumber \\ & & -\frac{l^{3}m^{1/2}}{\left (2\pi \right) ^{5/2}\hbar } \frac{1}{\beta ^{3/2 } } \int_{-\infty } ^{\infty } \int_{-\infty } ^{\infty } dk_{x}\ dk_{y}g_{3/2}(e^{-\beta (\varepsilon _ { k_{x}}+\varepsilon _ { k_{y}}-\mu) }) \label{tubosboson}\end{aligned}\]] for a boson gas, and @xmath37\bigr\ } \label{tubosfermion}\]] for a fermion gas. for calculation matters, it is useful to split the infinite integrals into an number @xmath38 of integrals running over the energy bands, taking @xmath38 as large as necessary to acquire convergence. for a gas inside a multilayer structure, the particle number @xmath3 is directly obtained from eqs. ([omegaboson]) and ([omegafermion]). important characteristics can be extracted, such as the critical temperature for a condensating boson gas and the influence of the system parameters on it, @xmath2 and @xmath39, already reported in refs. but for the case of a fermion gas, we focus on the chemical potential since it is closely related to the fermi energy of the system. in this case the number equation is @xmath40\ \bigr\ }, \label{numfer}\]] from which we are able to numerically extract the fermi energy of the system, which corresponds to the chemical potential for @xmath41, over the fermi energy of the ifg @xmath42, namely @xmath43, as a function of the impenetrability parameter @xmath39, whose behavior corresponds to a monotonically increasing curve as @xmath39 increases, being more evident for smaller values of @xmath44. another important feature is the chemical potential of the system over its fermi energy, @xmath45 as a function of the temperature in units of the fermi temperature @xmath46, fig [muvst], for a given value of @xmath47. there is a special interest in fig [muvst], since for certain geometrical configurations the chemical potential shows an anomalous behavior, as will be shown later. also, in this last figure one may notice that for @xmath48 the 3d ifg behavior for the chemical potential is recovered, and that the curve crosses the @xmath8 axis in @xmath49, as has been reported in @xcite. meanwhile, as @xmath50, we have a fermion gas inside a two dimension structure, giving a zero chemical potential at @xmath51, as expected. for planes with @xmath52 and different values of @xmath44.] [muvst] for multitubes with @xmath53 and different values of @xmath39.] we make a similar procedure for the boson an fermi gases inside a multitube structure, the first one being reported in @xcite. but for a fermion gas we start from the equation @xmath54\bigr\ } \label{fermtubos}\]] and extract the chemical potential over the fermi energy of the system, @xmath45, which is probably the feature that attracts greater attention due to its anomalous behavior shown in fig. [muvsttubos], which shows up as an unexpected small hump. another interesting characteristic is that the chemical potential over the ifg fermi energy in every case is lifted as @xmath39 increases due to the presence of the layers, in the same way as the chemical potential of the boson gas started above zero. the _ specific heat _ of a boson gas has been reported in ref. @xcite where we can observe a transition from a 3d system to a 2d one, which becomes evident for certain parameter values and sufficiently high temperatures. at this point is where the advantages of summing over a great amount of allowed energy bands shows its relevance. meanwhile, the specific heat for a fermion gas inside layered arrays is obtained going through the derivatives of eqs. ([omegaboson]) and ([tubosfermion]), leading, after some algebra, to @xmath55\bigr\ } \nonumber \\ & & + 2{\int_{-\infty } ^{\infty } dk_{z } } \ln \bigl\{1+\exp [-\beta (\varepsilon _ { k_{z}}-\mu)] \bigr\ } \{2\varepsilon _ { k_{z}}-\mu + t\frac{d\mu } { dt}\ } \nonumber \\ & & + 2\beta { \int_{-\infty } ^{\infty } dk_{z}}\frac{\varepsilon _ { k_{z } } \bigl\{\varepsilon_{k_{z}}-\mu + t\frac{d\mu } { dt}\bigr\}}{\exp \bigl\{\beta (\varepsilon_{k_{z}}-\mu) \bigr\}+1 } \bigr\ } \label{cvfermiones}\end{aligned}\]] for multiplanes, and @xmath56 for multitubes. for a fermion gas in multilayers.] for a fermion gas in multitubes.] in figs. [cvlayers] and [cvtubos] we show the behavior of the specific heat of layers with a separation among layers of @xmath57 and several impenetrability intensities @xmath39, as a function of the temperature over the system s fermi temperature. it may be observed that, as the barriers disappear (@xmath48), one recovers the ifg specific heat classical value, @xmath58. it is also noticeable that as the value of @xmath59 diminishes, a dimensional crossover signature from 3d to 2d becomes evident, since the the first minimum (going from right to left) in the specific heat deepens towards a value @xmath60 and broadens as @xmath39 increases. in fig. [cvtubos] the dimensional crossover from 3d to 1d is very noticeable as the mentioned minimum drops down to the value @xmath61 which corresponds to one - dimension.
Conclusions
in summary, we have calculated the thermodynamic properties of ideal boson and fermion gases inside periodical structures. in our model the multilayers and multitubes are generated with dirac - comb potentials in either one or two directions, while the particles are free in the remaining directions. just by introducing the planes, the translational symmetry of the particles is broken. this fact reflects in every thermodynamic property of the constrained system. in particular, fermions in multi - tubes progress from a 3d behavior to that 2d and finally to 1d as the wall impenetrability is increased, which is observed in the curves of the specific heat as a function of temperature. there is a critical value of the wall impenetrability for which the system begins to behave in dimensions less than two, which is signaled by the appearance of an anomalous chemical potential. bosons in multitubes show similar dimensional crossover like that expressed by fermions, in addition to the bose - einstein condensation at temperatures below the critical temperature of a ideal bose - gas with the same particle density. + * acknowledgements. * we acknowledge partial support from unam - dgapa - papiit (mxico) # in105011 and in111613 - 3, and conacyt 104917. | we report the thermodynamic properties of bose and fermi ideal gases immersed in periodic structures such as penetrable multilayers or multitubes simulated by one (planes) or two perpendicular (tubes) external dirac comb potentials, while the particles are allowed to move freely in the remaining directions.
although the bosonic chemical potential is a constant for @xmath0, a non decreasing with temperature anomalous behavior of the fermionic chemical potential is confirmed and monitored as the tube bundle goes from 2d to 1d when the wall impenetrability overcomes a critical value. in the specific heat curves
dimensional crossovers are very noticeable at high temperatures for both gases, where the system behavior goes from 3d to 2d and latter to 1d as the wall impenetrability is increased. | 1307.2649 |
Introduction
large scale structures like galaxies and clusters of galaxies are believed to have formed by gravitational amplification of small perturbations @xcite. observations suggest that the initial density perturbations were present at all scales that have been probed by observations. an essential part of the study of formation of galaxies and other large scale structures is thus the evolution of density perturbations for such initial conditions. once the amplitude of perturbations at any scale becomes large, i.e., @xmath0, the perturbation becomes non - linear and the coupling with perturbations at other scales can not be ignored. indeed, understanding the interplay of density perturbations at different scales is essential for developing a full understanding of gravitational collapse in an expanding universe. the basic equations for this have been known for a long time @xcite but apart from some special cases, few solutions are known. a statistical approach to this problem based on pair conservation equation has yielded interesting results @xcite, and these results have motivated detailed studies to obtain fitting functions to express the non - linear correlation function or power spectrum in terms of the linearly evolved correlation function @xcite. it is well known from simulation studies that at the level of second moment, i.e., power spectrum, correlation function, etc., large scales have an important effect on small scales but small scales do not have a significant effect on large scales @xcite. most of these studies used power spectrum as the measure of clustering. results of these simulation studies form the basis for the use of n - body simulations, e.g., from the above results we can safely assume that small scales not resolved in simulations do not effect power spectrum at large scales and can be ignored. substructure can play an important role in the relaxation process. it can induce mixing in phase space @xcite, or change halo profiles by introducing transverse motions @xcite, and, gravitational interactions between small clumps can bring in an effective collisionality even for a collisionless fluid @xcite. thus it is important to understand the role played by substructure in gravitational collapse and relaxation in the context of an expanding background. in particular, we would like to know if this leaves an imprint on the non - linear evolution of correlation function. effect of substructure on collapse and relaxation of larger scales is another manifestation of mode coupling. in this paper, we report results from a study of mode coupling in gravitational collapse. in particular, we study how presence of density perturbations at small scales influences collapse and relaxation of perturbations at larger scales. these effects have been studied in past @xcite but the motivation was slightly different @xcite. we believe it is important to study the issue in greater detail and make the relevance of these effects more quantitative using n - body simulations with a larger number of particles. we also study the reverse process, i.e., how does collapse of perturbations at large scales effect density perturbations at much smaller scales. it is well known that the local geometry of collapse at the time of initial shell crossing is planar in nature @xcite, hence we model density perturbations as a single plane wave in this work. simple nature of the large scale fluctuation allows us to study interaction of well separated scales without resorting to statistical estimators like power spectrum. we are studying the same problem in a more general setting and those results will be reported in a later publication. key features of collapse of a plane wave can be understood using quasi - linear approximations, at least at a qualitative level. initial collapsing phase is well modelled by the zeldovich approximation @xcite, wherein particles fall in towards the centre of the potential well. zeldovich approximation breaks down after orbit crossing as it does not predict any change in the direction of motion for particles, thus in this approximation particles continue to move in the same direction and the size of the collapsed region grows monotonically. in a realistic situation we expect particles to fall back towards the potential well and oscillate about it with a decreasing amplitude, and the collapsed region remains fairly compact. several approximations have been suggested to improve upon the zeldovich approximation @xcite. the adhesion approximation @xcite invokes an effective viscosity : this prevents orbit crossing and conserves momentum to ensure that pancakes remain thin and matter ends up in the correct region. this changes the character of motions in dense regions (no orbit crossing or mixing in the phase space) but predicts locations of these regions correctly. if one assumes that the gravitational potential evolves at the linear rate @xcite, then it can be shown that the collapsed region remains confined. the effective drag due to expanding background slows down particles and they do not have enough energy to climb out of the potential well. thus the process of confining particles to a compact collapsed region results from a combination of expansion of the universe and gravitational interaction of in falling particles. none of the approximations captures all the relevant effects. therefore we must turn to n - body simulations @xcite in order to study collapse and relaxation of perturbations in a complete manner.
Evolution of perturbations
we will consider only gravitational effects here and ignore all other processes. we assume that the system can be described in the newtonian limit. the growth of perturbations is then described by the coupled system of euler s equation and poisson equation in comoving coordinates along with mass conservation, e.g., see @xcite. @xmath1 it is assumed that the density field is generated by a distribution of particles, each of mass @xmath2, position @xmath3. @xmath4 is the present value of hubble constant, @xmath5 is the present density parameter for non - relativistic matter and @xmath6 is the scale factor. in this paper we will consider the einstein - de sitter universe as the background, i.e., @xmath7. these can be reduced to a single non - linear differential equation for density contrast @xcite. @xmath8 where @xmath9 \delta_{\bf k'} \delta_{{\bf k } -{\bf k'} } \nonumber\]] and @xmath10 \;\;\ ; ; \;\;\;\ ; m= \sum\limits_i m_i \nonumber\]] the terms @xmath11 and @xmath12 are the non - linear coupling terms between different modes. @xmath12 couples density contrasts in an indirect manner through velocities of particles (@xmath13). the equation of motion still needs to be solved for a complete solution of this equation, or we can use some ansatz for velocities to make this an independent equation. it can be shown that individual _ virialised _ objects, i.e., objects that satisfy the condition @xmath14 where @xmath15 is the kinetic energy and @xmath16 is the potential energy, do not make any contribution towards growth of perturbations through mode coupling @xcite at much larger scales, i.e., the @xmath17 term is zero. the contribution of mode coupling due to interaction of such objects is not known. approximate approaches to structure formation can be developed by ignoring interaction of well separated scales. the evolution of density perturbations can be modelled as a combination of non - linear collapse at small scales, and the collapsed objects can be displaced using quasi - linear approximations @xcite. these approaches yield an acceptable description of properties of collapsed objects and their distribution for a first estimate. pinocchio @xcite provides sufficient information about halo properties and merger trees for use with semi - analytic models of galaxy formation. the efficacy of these models puts an upper bound on the effects of mode coupling that we are studying here. in this paper we simplify the system by starting with perturbations that have a non - zero amplitude only for two sets of scales. we simulate the collapse of a plane wave by starting with non - zero amplitude of perturbations for the fundamental mode of the simulation box along the @xmath18 axis, the wavenumber of the fundamental mode is denoted by @xmath19. this serves as the large scale perturbation in our study. the amplitude for this mode is chosen so that shell crossing takes place when the scale factor @xmath20. power spectrum for small scale fluctuations is chosen to be non - zero in a range of wave numbers @xmath21 with a constant amplitude across this window, i.e., @xmath22 for @xmath23. a gaussian random realisation of this power spectrum is used for small scale fluctuations. here @xmath24 is the power per logarithmic interval in @xmath25 contributed by small scales (large @xmath25) and @xmath11 is the amplitude of the fundamental mode that gives rise to the plane wave. the ratio of @xmath26 at @xmath27 and for the plane wave is denoted by @xmath28, thus when @xmath29 collapse of perturbations at these scales happens at the same time whereas for @xmath30 perturbations at small scales collapse before the plane wave collapses. we chose the ratio @xmath31 so that there is distinct separation in the scales involved. , we go from a single stream region to a three stream region and so on up to seven stream region near the centre.] collapse of a plane wave of this kind leads to formation of a multi - stream region, we will also use the term pancake to describe this region. figure [phase_space] shows the phase space plot for the plane wave at late stages of collapse for the simulation pm_00l (see table 1 for details of the simulation). velocity of particles is plotted as a function of position, only the @xmath18 component is plotted as there is no displacement or velocity along other directions in this simulation. regions where particles with different velocities can be found are the multi - stream regions. as we approach the centre of pancake located at @xmath32, we go from a single stream region to a three stream region and so on up to seven stream region near the centre. in initial stages, the mass in the pancake increases rapidly as more particles fall in. figure [den_plwave] shows this in terms of over - density which increases sharply from @xmath20 to @xmath33. a significant fraction of the total mass falls into the pancake and the infall velocities for the remaining matter are very small. in this regime the mass of the pancake is almost constant, this can be seen from the panels of figure [den_adh_sub] where the mass enclosed in the pancake region is almost constant from @xmath33 to @xmath34..this table lists parameters of n - body simulations we have used. all the simulations used @xmath35 particles. the first column lists name of the simulation, second column lists the code that was used for running the simulation, third column gives the relative amplitude of small scale power and the plane wave, the fourth column tells us whether the large scale plane wave was present in the simulation or not, and the last column lists the distribution of particles before these are displaced using a realisation of the power spectrum. _ grid _ distribution means that particles started from grid points. _ perturbed grid _ refers to a distribution where particles are randomly displaced from the grid points, this displacement has a maximum amplitude of @xmath36 grid points. such an initial condition is needed to prevent particles from reaching the same position in plane wave collapse as such a situation is pathological for tree codes. the treepm simulations were run with a force softening length equal to the grid length. [cols="<,<,<,<,<",options="header ",] in absence of any substructure the collisionless collapse retains planar symmetry and we have layers of multi - stream regions with the number of streams increasing towards the centre of the pancake. presence of small scale fluctuations can induce transverse motions and these motions are amplified in the pancake. weakly bound substructure can be torn apart due to interaction with rapidly in falling matter. on the other hand, higher average density in the multi stream region can lead to rapid growth of perturbations. it is known that pancakes are unstable to fragmentation due to growth of perturbations @xcite. the velocity field is anisotropic due to infall along one direction, hence the rate at which perturbations grows will also exhibit anisotropies. velocity dispersion along the direction of plane wave collapse is larger than the transverse direction, hence the growth of fluctuations in the transverse plane is expected to be more rapid. if the in falling material contains collapsed substructure, then gravitational interactions between these can induce large transverse velocities. this takes away kinetic energy from the direction of infall, which in turn can lead to more fragmented and thinner multi - stream region. in the following sections we describe the numerical experiments have undertaken in detail, and test the physical ideas and expectations outlined above
Numerical experiments and results
we used a particle - mesh code @xcite and the treepm code @xcite. some simulations were run using the parallel treepm @xcite. treepm simulations used spline softening with softening length equal to the length of a grid cell in order to ensure collisionless evolution. we used force softening assuming a spline kernel @xcite. all the simulations were carried out with @xmath35 particles. table 1 lists parameters of the simulations we have used for this paper. we have used two types of initial distribution of particles. in the _ grid _ distribution particles are located at grid points before being displaced to set up the initial perturbations. _ perturbed grid _ refers to a distribution where particles are randomly displaced from the grid points @xcite, this displacement has a maximum amplitude of @xmath36 grid length. such an initial condition is needed to prevent particles from reaching the same position in plane wave collapse as such a situation is pathological for tree codes. these small displacements do not affect the power spectrum to be realised, pm_@xmath37l and t_@xmath37l were compared to test for any systematic effects. simulations t_@xmath38p and t_@xmath39p were similar to t_@xmath38l and t_@xmath39l except that the small scale fluctuations were restricted to the direction orthogonal to the direction of plane wave. thus the small scale fluctuations had the same form for all @xmath18. these simulations are useful for differentiating between competing explanations for results outlined below. in addition to the n - body simulations listed in table 1, we also carried out one dimensional simulations within the adhesion model @xcite with a finite viscosity following a method similar to the one outlined by weinberg and gunn (1990). and @xmath33 from the pm_00l simulation, dashed lines show the density profile from the t_00l simulation with the same profile.] figure [den_plwave] shows the density profile of the pancake for pm_00l and t_00l simulations at two epochs. these figures demonstrate that the density profiles in these simulations are almost identical, indeed the tiny differences can be attributed to the different initial distribution of particles. we have checked this assertion by running the pm_00l with the perturbed grid initial conditions. the treepm method has a slightly better resolution but it does not induce any new features. this is expected as the force softening length used in the treepm simulations is one grid length, same as the average inter - particle separation and it has been shown than such force softening does not induce two body collisions @xcite. we will mostly use treepm simulations for the remaining part of this study. an important indicator of the role played by substructure is the thickness of the pancake that forms by collapse of the plane wave. if substructure does not play an important role in evolution of large scale perturbations then the thickness of pancake should not change by a significant amount. on the other hand, if substructure does indeed speed up the process of dynamical relaxation then we should see some signature in terms of the thickness of pancake, velocity structure, or both. any such effect will be apparent only at late times as infall of matter into the pancake dominates at early times. dynamical effects of substructure will become important only at late times. figure [pthick] shows a slice from some of the simulations listed in table 1. the plane wave collapses along the vertical axis. configuration at @xmath33 is shown here, the plane wave begins to collapse at @xmath20. different panels in this figure refer to simulation t_00l, t_10l and t_40l. the boundary of the multi stream region is visible clearly in all the slices even though this region is fragmented in the last panel (t_40l). it is clear that the pancake is thinner in simulations with more substructure. a more detailed comparison of simulations with different level of substructure is shown in figure [den_all]. the top panel of this figure shows the averaged over - density as a function of the @xmath18 coordinate, the plane wave collapses along this axis. over - density is averaged over all @xmath40 and @xmath41 for a given interval @xmath42 to obtain the averaged values plotted here. the peak over - density at the centre of the pancake is smaller in simulations with more substructure. the mass enclosed within a given distance of the centre of pancake (defined here as the trough of the potential well of the plane wave) is smaller for more substructure, even though the variation is very small at less than @xmath43 between the extreme cases (see figure [den_adh_sub]). potential wells corresponding to substructure prevent infall into the pancake region. as the amount of substructure is increased, there is visible reduction in the size of the region around the pancake where density is greater than average. the visual impression of figure [pthick] is reinforced by the variation of over - density. the middle panel of figure [den_all] shows the _ rms _ velocities of particles in direction transverse to the plane wave collapse as a function of the @xmath18 coordinate. as in the top panel, averaging is done over all @xmath40 and @xmath41 for a given interval @xmath42. this plot shows that the transverse motions are enhanced in the dense pancake region. the amplitude of transverse motions is larger in simulations with more substructure. size of the region where these motions are significant varies with the amount of substructure, as in case of over - density (top panel). the _ rms _ transverse velocities do not go to zero outside the pancake region, instead these level off to a small residual value. , the direction of collapse for the plane wave. density profile has been averaged over the directions transverse to the collapse of plane wave. the curves are for @xmath33, simulations used are t_00l (solid line), t_05l (dashed line), t_10l (dot - dashed line), t_20l (dotted line) and t_40l (dot - dot - dashed line). the middle panel shows the rms transverse velocities of particles at the same epoch for t_05l (dashed line), t_10l (dot - dashed line), t_20l (dotted line) and t_40l (dot - dot - dashed line). the lower panel shows the rms transverse velocities of collapsed haloes at the same epoch for t_05l (dashed line), t_10l (dot - dashed line), t_20l (dotted line) and t_40l (dot - dot - dashed line).] transverse motions are due to motions of particles in clumps that constitute substructure, due to infall of particles in these clumps, and, transverse motions of clumps as they move towards each other. in order to delineate these effects, we have plotted the _ rms _ velocities for haloes in the last panel of figure [den_all]. these haloes were selected with the friends - of - friends (fof) algorithm using a linking length of @xmath44 grid length. transverse component of the velocity of centre of mass for haloes with more than @xmath45 particles was used for the figure. such a high cutoff for halo members is acceptable because typical haloes have several hundred members, see the following subsection on mass functions. differences between simulations with different amount of substructure are more pronounced than in the middle panel. for simulations with a small amount of substructure, motion of clumps is subdominant and hence the transverse motions are contributed mostly by internal motions and infall. in simulations with more substructure, motions of clumps contribute significantly to the _ rms _ transverse velocity. gravitational attraction of clumps, particularly in close encounters in the pancake region induce the transverse component. collisions are enhanced in the pancake region as the number density of clumps is higher. in order to convince ourselves that transverse motions induced by scattering / collision of clumps is the most likely reason for the reduced thickness of pancakes, we compare simulations t_@xmath38l and t_@xmath39l with t_@xmath38p and t_@xmath39p. in t_@xmath38p and t_@xmath39p simulations, the small scale fluctuations do not have any @xmath18 dependence. in these (t_@xmath38p and t_@xmath39p) simulations there are no clumps but streams of particles that are falling in and this reduces the number of scattering that take place no @xmath18 dependence means that dense streams run into each other head on with grazing collisions happening only rarely. of course, in the simulation the presence of the plane wave leads to breaking of these streams into clumps as the streams are stretched inhomogeneously in the @xmath18 direction. these clumps are aligned parallel to the @xmath18 axis. in the pancake region scattering of these streams occasionally leads to complex patterns. if presence of substructure and its growth in the pancake was the only cause for making the pancake thinner then pancake in these simulations should be thinner as well. figure [pic_lp] shows slices from simulations t_@xmath39l and t_@xmath39p for @xmath33. a slice from the simulation pm_@xmath37l is also plotted here for reference. this visual comparison shows that the pancake is thinner in t_@xmath39l as compared to t_@xmath39p. indeed, the thickness of pancake in t_@xmath39p and pm_@xmath37l is very similar. this reinforces the point that scattering of clumps in the pancake region is the key reason for thinner pancakes. the substructure is helping to confine the pancake to a smaller region. it is interesting to study the collapse of a plane wave in an n - body simulation and compare it with the collapse in the adhesion model @xcite with a finite effective viscosity. we first study the collapse of a plane wave in absence of any substructure, n - body simulations pm_00l for comparison with numerical simulations using the adhesion model with finite effective viscosity. one dimensional adhesion simulations were done using the plane wave with the same amplitude as the n - body simulations. we use the standard method for computing the trajectories of particles in the adhesion model @xcite, a summary of the basic formalism is reproduced here for reference. in adhesion approximation, the equation of motion for a particle is replaced by the burgers equation @xcite. in the one dimensional situation we are considering here, we have : @xmath46 here @xmath47 is the velocity of particles and @xmath48 is the linear growth factor. this equation can be solved by introducing the velocity potential @xmath49, where @xmath50 coincides with the gravitational potential at the initial time. solution has the following form. @xmath51 and, @xmath52 dq .\]] here @xmath53 is the lagrangian position of the particle and @xmath40 is the eulerian position. in this method we integrate the differential equation for particle trajectories. at each time step velocity is calculated by above procedure at grid points and interpolated to particles positions. figure [den_adh] shows the mass enclosed within a distance @xmath54 from the centre of the pancake. the enclosed mass is defined as : @xmath55 here @xmath56 is the density at position @xmath18 and @xmath57 is the centre of the pancake. there is no ambiguity for comparing the results with n - body simulations in case of no substructure as density depends only on @xmath18. while comparing other simulations with the adhesion solution, we will consider density averaged over @xmath40 and @xmath41 directions adhesion model is run only for the one dimensional problem. top panel of figure [den_adh] shows the enclosed mass @xmath58 at @xmath59, middle panel is for @xmath60 and the lower panel is for @xmath61. the solid curve shows the enclosed mass for pm_00l. in the region with a given number of streams, the n - body curve is smooth. jumps in mass enclosed occur at transition from single stream to multi stream region, and at other transitions where the number of streams changes within the multi stream region. all other curves show @xmath58 for adhesion model : dashed curve is for @xmath62, dotted curve is for @xmath63 and the dot - dashed curve is for @xmath64. there is no constant effective viscosity curve that follows the n - body curve closely through the multi stream regions. in regions with a given number of streams, the n - body curve stays around a curve for constant effective viscosity in the adhesion model. a remarkable fact is that the n - body curve for the three stream region at all the epochs follows the adhesion model curve for @xmath65. similar behaviour is seen for the five stream region which follows @xmath66 though the range of scales and epochs over which this can be resolved is somewhat limited. addition of substructure clearly changes the character of the problem and the collapse is no longer one dimensional. however, the scale of the substructure is so small compared to the wavelength of the plane wave that the large scale collapse is still very close to planar. figure [den_adh_sub] shows the mass enclosed within a distance @xmath54 from the centre of the multi stream region for simulations pm_00l, t_10l and t_40l. density is averaged over all @xmath40 and @xmath41 for this plot in the same manner as for figure [den_all]. also plotted in the figure are curves for the adhesion model (@xmath63), where the calculation is done without taking substructure into account. the motivation for such a comparison is to see the effect of substructure on the favoured value of effective viscosity. substructure removes the sharp change in density at the boundaries of @xmath67-stream, @xmath68-stream and @xmath69-stream regions and the curves for t_10l and t_40l are smoother in the pancake region. the finite viscosity curve matches simulations with substructure over a wider range of scales than with pm_00l. there are no other noteworthy differences. mass function of collapsed haloes in these simulations can be used to understand the influence of plane wave collapse on substructure. collapsed structures form in these simulations primarily due to initial density fluctuations at small scales, with some modulation by the collapse of the plane wave. in this section we study the effect of the collapsing plane wave on the mass function of collapsed haloes. these haloes were selected with the friends - of - friends (fof) algorithm using a linking length of @xmath44 grid length. the initial power spectrum has a peak at the scale corresponding to @xmath70 of the simulation box, or @xmath71 grid lengths. thus typical haloes will have a lagrangian radius of about @xmath72 grid lengths and should contain about @xmath73 particles. thus a cutoff of @xmath45 or more particles for haloes is reasonable for this study. in absence of the plane wave, the only perturbations are at small scales. the small scale perturbations are concentrated around a given mass scale and the mass function is also peaked around this mass at early epochs. at late epochs mergers lead to formation of haloes with a larger mass and the range of masses is greater for models with a larger amplitude of fluctuations. figure [slices] shows these features in the distribution of particles. these features can also be seen in figure [massf] where mass fraction @xmath74 for @xmath75, @xmath76 and @xmath77 is plotted in different panels. @xmath74 is the fraction of total mass in collapsed haloes with halo mass above @xmath78. adding the plane wave at a much larger scale than the small scale fluctuations essentially pushes much of the mass into the pancake region, leaving a small fraction of matter in the under dense regions that occupy much of the volume. growth of small scale fluctuations in the under dense regions is inhibited whereas growth of fluctuations in the pancake region is enhanced, this is seen clearly in the slices from simulations shown in figure [slices]. higher background density in the pancake region leads to rapid growth of perturbations, mergers of haloes also lead to formation of massive clumps. these effects become more pronounced at late epochs and result in a shift of mass function towards larger masses, indeed haloes at two distinct mass scales are present. low mass clumps in under dense regions have the mass expected of haloes in regions where small scale fluctuation dominate whereas haloes of a much higher mass are present in the pancake region. figure [massf] shows these two mass scales very clearly. total mass in collapsed haloes does not change significantly with the addition of the plane wave. indeed for simulations t_40l and t_40, mass function is the same at @xmath75 as small scales dominate. at late times (@xmath33), the effect of plane wave makes the mass function of t_05l, t_10l and t_20l similar. not surprisingly, presence of large scale power leads to formation of more massive haloes. however it does not seem to enhance the total mass in collapsed haloes.
Discussion
in this paper we studied the effect of substructure on collapse of a plane wave. the key conclusions of the present study of the role of substructure are : * the pancake formed due to collapse of the plane wave is thinner if the in falling material is formed of collapsed substructure. * we show that collisions between clumps lead to enhancement of velocities transverse to the direction of large scale collapse. * we show that in simulations with substructure where collisions are suppressed, pancakes are not thinner. * thus collision induced enhancement of motions transverse to the collapsing plane wave takes away kinetic energy from the direction of infall and leads to thinner pancakes. * presence of large scale power shifts the mass function towards larger masses. there is, however, no change in the total mass in collapsed haloes. the points outlined above essentially relate to coupling of density fluctuations at well separated scales. each of these points refers to a measurable effect of such a coupling. the nature of large scale fluctuation, a single plane wave, does not allow us to estimate the effect in terms of statistical indicators like the power spectrum. we plan to study these aspects with larger (@xmath79) simulations where the large scale collapse will also be generic. large, high resolution studies are needed as @xmath35 simulations with particle mesh code have not shown any large effect in power spectrum at late times @xcite. from the centre of the multi stream region. the top panel shows the curves for @xmath33. the thick solid curve is for the n - body simulation pm_00l. jumps in the mass enclosed occur at transition from multi stream region with @xmath80 streams to @xmath81 streams, with @xmath82 a non - zero positive integer. all other curves show @xmath58 for adhesion model : dashed curve is for @xmath62, dotted curve is for @xmath63 and the dot - dashed curve is for @xmath64. the middle panel shows the same set curves for @xmath60 and the lower panel is for @xmath34.] from the centre of the multi stream region. the top panel shows the curves for @xmath33. the solid curve is for n - body simulation pm_00l. other simulations are also plotted here t_10l (dashed curve) and t_40l (dot - dashed curve). dotted curve shows the mass enclosed in the one dimensional adhesion model with @xmath63. the lower panel shows the same set of curves for @xmath34 and the middle panel is for @xmath60.] as a function of mass @xmath78 in n - body simulations. the top panel is for @xmath75, the middle panel is for @xmath20 and the lower panel is for @xmath33. curves are shown for t_05 (solid curve), t_05l (thick solid curve), t_10 (dashed curve), t_10l (thick dashed curve), t_20 (dotted curve), t_20l (thick dotted curve), t_40 (dot - dashed curve) and t_40l (thick dot - dashed curve).] another important point to consider is that we have considered two well separated scales for fluctuations and there is no infall once fluctuations at the larger scales collapse. numerical experiments that can shed light on effects of this feature are also required to improve our understanding of issues. we also compared the collapse of a plane wave in an n - body with the collapse in the adhesion model with a finite effective viscosity. we found that : * the adhesion model predicts the variation of density very well with a constant effective viscosity in regions with a given number of streams. * regions with a given number of streams coincide with the adhesion model with the same value of effective viscosity at all epochs.
Acknowledgements
jsb thanks r.nityananda, t.padmanabhan and k.subramanian for useful discussions and suggestions. jsb also thanks varun sahni and uriel frisch for a useful discussion on related issues. numerical experiments for this study were carried out at cluster computing facility in the harish - chandra research institute (http://cluster.mri.ernet.in). this research has made use of nasa s astrophysics data system. bagla j. s. and padmanabhan t. 1994, mnras 266, 227 bagla j. s. and padmanabhan t. 1997a, mnras 286, 1023 bagla j. s. and padmanabhan t. 1997b, pramana 49, 161 bagla j. s. 2002, japa 23, 185 bagla j. s. and ray s. 2003, new astronomy 8, 665 bagla j. s. 2004, to appear in _ current science_. bernardeau f., colombi s., gaztanaga e. and scoccimarro r. 2002, physics reports 367, 1 bertschinger e. 1998, ara&a 36, 599 bond j. r. and myers s. t. 1996a, apjs 103, 1 bond j. r. and myers s. t. 1996b, apjs 103, 41 bond j. r. and myers s. t. 1996c, apjs 103, 63 brainerd t. g., scherrer r. j. and villumsen j. v. 1993, apj 418, 570 couchman h. m. p. and peebles p. j. e. 1998, apj 497, 499 engineer s., kanekar n. and padmanabhan t. 2000, mnras 314, 279 evrard a. e. and crone m. m. 1992, apjl 394, 1 gurbatov s. n., saichev a. i. and shandarin s. f. 1989, mnras 236, 385 hamilton a. j. s., matthews alex, kumar p. and lu edward 1991, apjl 374, 1 hui l. and bertschinger e. 1996, apj 471, 1 jain b., mo h. j. and white s. d. m. 1995, mnras 276l, 25 klypin a. a. and melott a. l. 1992, apj 399, 397 little blane, weinberg david h. and park changbom, 1991 mnras 253, 295 lynden - bell d. 1967, mnras 136, 101 ma chung - pei and boylan - kolchin michael 2004, phys.rev.lett. 93, 021301 ma chung - pei and bertschinger edmund 2003, astro - ph/0311049, apj in press. matarrese s., lucchin f., moscardini l. and saez d. 1992, mnras 259, 437 melott a. l., shandarin s. f., splinter r.j. and suto y. 1997, apjl 479, 79 monaco p., theuns t., taffoni g., governato f., quinn t. and stadel j. 2002, apj 564, 8 monaco p., theuns t. and taffoni g. 2002, mnras 331, 587 nityananda r. and padmanabhan t. 1994, mnras 271, 976 padmanabhan t. 1996, mnras 278l, 29 padmanabhan t. 2002, _ theoretical astrophysics : galaxies and cosmology (vol.3) _, cambridge university press peacock j. a. and dodds s.j. 1996, mnras 280, 19 peacock j. a. 1998, _ cosmological physics _, cambridge university press peebles p. j. e. 1974, a&a 32, 391 peebles p. j. e. 1980, _ large scale structure of the universe _, princeton university press peebles p. j. e. 1985, apj 297, 350 peebles p. j. e. 1990, apj 365, 27 ray s. and bagla j. s. 2004, astro - ph/0405220 sahni v. and coles p. 1995, physics reports 262, 1 shandarin s. f., zeldovich y. b., 1989, rvmp, 61, 185 smith r. e. et al. 2003, mnras 341, 1311 splinter r. j., melott a. l., shandarin s. f. and suto y. 1998, apj 497, 38 springel v., yoshida n. and white s.d.m. 2001, new astronomy 6, 79 subramanian k. 2000, apj 538, 517 taffoni g., monaco p. and theuns t. 2002, mnras 333, 623 valinia a. and shapiro p. r. and martel h. and vishniac e. t. 1997, apj 479, 46 weinberg d. h. and gunn j. e. 1990, mnras 247, 260 weinberg martin d. 2001, mnras 328, 311 zeldovich ya.b. 1970, a&a 5, 84 | we study the interplay of clumping at small scales with the collapse and relaxation of perturbations at much larger scales.
we present results of our analysis when the large scale perturbation is modelled as a plane wave.
we find that in absence of substructure, collapse leads to formation of a pancake with multi - stream regions.
dynamical relaxation of plane wave is faster in presence of substructure.
scattering of substructures and the resulting enhancement of transverse motions of haloes in the multi - stream region lead to a thinner pancake. in turn,
collapse of the plane wave leads to formation of more massive collapsed haloes as compared to the collapse of substructure in absence of the plane wave.
the formation of more massive haloes happens without any increase in the total mass in collapsed haloes.
a comparison with the burgers equation approach in absence of any substructure suggests that the preferred value of effective viscosity depends primarily on the number of streams in a region.
gravitation cosmology : theory dark matter, large scale structure of the universe | astro-ph0408429 |
Introduction
let us consider the following question : a group of people, living in the same country, but in different places, want to move in such a way that the distance among any two of them strictly decreases. unfortunately they can not all fit into one single house. hence, the constraints are : a. anyone can either move from his place to another one s (no matter what the owner of the `` destination house '' does) or stay where he is, b. they can not all move to the same place. is it possible for them to move? this question is a particular case of a more interesting problem involving discrete, possibly infinite, metric spaces (we will come back to the original problem in section [sec : conclusions] at the end of the paper). we say that a couple @xmath0 is a metric space if @xmath1 is a set and @xmath2 is real function, called _ distance _, defined on @xmath3 such that (1) @xmath4 if and only if @xmath5 (for all @xmath6) and (2) @xmath7 for all @xmath8. we do not require the distance to be finite ; if @xmath9 we imagine that @xmath10 and @xmath11 belong to two disjoint components of the space (see below a more detailed discussion on this, in the case of a graph). from (2), with @xmath12, and (1) we have that @xmath13 (for all @xmath14) hence @xmath2 is nonnegative. moreover, again from (2), using @xmath15, and (1), we have that @xmath16 (for all @xmath6), thus, by symmetry, we have @xmath17. for some basic properties of metric spaces see for instance (*??? * chapter 2). an example of a discrete metric space is given by a (nonoriented) graph (see for instance @xcite). roughly speaking, a graph is a (finite or infinite) collection of points, called _ vertices _, along with a set of pairs of vertices, called _ edges_. we say that it is possible to move (in one step) from a vertex @xmath10 to a vertex @xmath11 if and only if @xmath18 is an edge. a _ path _ is a concatenation of edges, and the minimum number of steps required to go from @xmath10 to @xmath11, say @xmath19, is called _ natural distance _ from @xmath10 to @xmath11. to be precise, in order to have an actual distance we have to assume that if we can move in one step from @xmath10 to @xmath11 then we can also go back in one step, that is, if @xmath18 is an edge then @xmath20 is an edge (in this case the graph is called _ nonoriented _). if we can go, in a finite number of steps, from @xmath10 to @xmath11 and back then we say that @xmath10 and @xmath11 _ communicate_. the set of vertices which communicate with @xmath10 is called the _ connected component _ containing @xmath10 ; if there is only one connected component, we say that the graph is _ connected_. of course, if two vertices belong to two disjoint connected components then the distance between them is infinite. in figure [fig:1] there is an example of an infinite graph which is connected if the dashed arrow is an edge and it has two connected components otherwise (the double arrows mean that you can move forth and back between the points, thus the graph is nonoriented). the central concept in this paper is the one of _ contractive map _ (see also @xcite). let @xmath0 and @xmath21 be two metric spaces ; a function @xmath22 is _ contractive _ if and only if for any @xmath6, such that @xmath23, we have @xmath24. the map is called a _ contraction _ if there exists @xmath25 such that @xmath26 for all @xmath6. roughly speaking a map is contractive if it (stricly) reduces the distances between points ; if the ratio of reduction has an upper bound which is (stricly) smaller than 1, then we have a contraction. clearly, any contraction is a contractive map and the two classes coincide on finite metric spaces. we are (mainly) interested in the case @xmath27 and @xmath28. observe that in our original problem, if we denote by @xmath29 the new position of the person who was at @xmath10 before, what we are looking for are nonconstant, contractive maps. any constant map is trivially contractive ; when the converse is true? the main result theorem [main] characterizes all the discrete metric spaces, satisfying a certain property (see below)), which have nonconstant contractive maps. as a simply consequence of theorem [main] we will see that every contractive map on a graph with the natural distance is a constant function if and only if the graph is connected.
Main results and examples
let us consider a discrete metric space @xmath0 (which is a metric space such that for any given point @xmath30 it is possible to find @xmath31 such that @xmath32 for all @xmath33) and let us assume that @xmath34 (the existence of the minimum is implicitly assumed). roughly speaking, according to equation , the distance either equals @xmath35 or it is at least @xmath36 ; this implies that if @xmath37 then @xmath5. we define, once and for all, @xmath38 and we introduce the equivalence relation @xmath39 defined by @xmath40 if and only if there exists @xmath41 such that @xmath42, @xmath43 and @xmath44 for any @xmath45 (if @xmath46). hence, two points @xmath10 and @xmath11 are equivalent if and only if we can reach @xmath11 from @xmath10 by performing a finite number of jumps of length @xmath36. we denote by @xmath47 $], the equivalence class induced by @xmath48, that is, @xmath47:=\{y : y \sim x\}$]. as usual, the _ quotient space _ @xmath49 is the set of the equivalence classes. it is easy to show that, for any @xmath6 such that @xmath23, we have @xmath50 $] if and only if there exists @xmath51 $] satisfying @xmath52. we denote by @xmath53 the metric on @xmath49 defined by @xmath54,[y]):= \inf_{x^\prime \in [x],y^\prime \in [y] } d(x, y)$]. examples of discrete metric spaces satisfying equation are finite metric spaces and (nonoriented) graphs with their natural distance ; in the last case we have @xmath55 and the quotient space @xmath49 is the set the connected components of the graph. we are ready to state and prove the main result of this paper. [main] let @xmath0 be a metric space satisfying equation ; then tfae a. there exists a nonconstant contractive map on @xmath1 ; b. there exists @xmath30 such that @xmath47 \not = x$] ; c. for every @xmath30, we have @xmath47 \not = x$] ; d. @xmath56. moreover there is a one - to - one map from the set of contractive maps @xmath57 into the set @xmath58 of contractive maps from @xmath59 to @xmath1. @xmath60. if @xmath61 is contractive and @xmath62 then @xmath63 which implies, by equation , @xmath64 ; hence @xmath65}$], the function @xmath61 restricted to the class @xmath47 $], is a constant function for any @xmath30. hence, if @xmath61 is nonconstant we have @xmath47 \neq x$]. just remember that @xmath67\}_{x\in x}$] is a partition of @xmath1 ; hence for all @xmath68 we have either @xmath69=[x]$] or @xmath69 \subseteq [x]^c \subsetneq x$]. we have that @xmath71 \not = \emptyset$] ; let us define a function @xmath61 by @xmath72 \\ y_0 & \hbox{if } x\in y, \\ \end{cases}\]] where @xmath73. it is just a matter of easy computation to show that this is a nonconstant contractive map (indeed for any @xmath74 $], @xmath75 we have @xmath76). it is an easy consequence of the fact that for all @xmath78, @xmath79 $] if and only if @xmath69 \neq [x]$] which, in turn, is equivalent to @xmath69 \cap [x]=\emptyset$]. finally, it is easy to show that, given a contractive map @xmath61, the map @xmath80 is well defined by @xmath81):=f(x)$] (since @xmath61 is constant on every class @xmath47 $]). it is straightforward to show that @xmath82 is injective and, clearly, being @xmath61 a contractive map, @xmath83), \phi(f)([y]))=d(f(x),f(y))=d(f(x^\prime),f(y^\prime)) \le d(x^\prime, y^\prime)$] for all @xmath84 and @xmath85. this implies that @xmath83), \phi(f)([y])) \le d_\sim([x],[y])$], thus @xmath80 is a contractive map from @xmath59 to @xmath1. let us observe that if the range of the distance @xmath86 is a finite set (take for instance @xmath1 finite) then any contractive map @xmath61 is actually a contraction and there is @xmath87 such that the @xmath88-th iteration @xmath89 is a constant map for all @xmath90 and @xmath91. indeed, @xmath92, hence if @xmath93 then @xmath94 is constant (say, @xmath95 for all @xmath48). if @xmath96 then @xmath97. clearly @xmath98, that is, @xmath99 is (the unique) fixed point of @xmath61. indeed, it is possible to prove (see (*??? * theorem 9.3)) that for any contraction @xmath61 in a (complete) metric space there exists @xmath100 exists and it is the unique fixed point for @xmath61, that is, @xmath101. a generic contractive map @xmath61 has at most one fixed point, since @xmath102, @xmath103 and @xmath104 implies @xmath105 which is a contradiction. nevertheless, if @xmath86 is not a finite set then the set of fixed points of @xmath61 might be empty as example [ex : empty] shows. in the language of our original problem, @xmath10 is a fixed point if and only if the member of the group living at @xmath10 does not move. moreover if @xmath106 then the three classes of contractive maps @xmath57, contractions @xmath107 and constant functions @xmath108 coincide. [cor : main] let @xmath0 be a metric space such that there are just two couples satisfying equation (namely @xmath109 and @xmath110) ; if @xmath111 then there exists a non constant contractive map. [ex : empty] consider @xmath112 and define the distance as follows @xmath113 where @xmath114 (see figure [fig:2] for a picture of a finite portion of this space along with the distances between `` consecutive '' points). in this case it is easy to prove that the following is a contracting map without fixed points @xmath115 in the following we compare the relations between @xmath57, @xmath107 and @xmath108. clearly @xmath116 ; we provide examples to show that, even if equation holds, all cases are possible. given @xmath117, by @xmath118 we mean the usual characteristic function of @xmath119 (which equals @xmath120 on @xmath119 and @xmath35 elsewhere). the reader is encouraged to verify that, in the following examples, the spaces we define are indeed metric spaces. [ex1] let @xmath121 and @xmath2 be defined by @xmath122, @xmath123, @xmath124 and @xmath35 otherwise ; then @xmath125 is a nonconstant contraction.another example (in the infinite case) is the following. let @xmath112 and define the distance by @xmath126 then @xmath127 is a nonconstant contraction. in the previous examples @xmath128 since the range of the distance is finite. [ex2] let @xmath129 and let @xmath2 be defined as follows @xmath130 in this case it is easy to check that this is a distance and that @xmath131 is a contractive map which is not a contraction while @xmath132 is a (nonconstant) contraction with @xmath133 equal to @xmath134. hence @xmath135. in theorem [main], to prove that every contractive map is a constant map, we require that all the points belong to the same class, that is, we can reach @xmath11 from @xmath10 with a finite number of consecutive steps of length @xmath36 (for any choice of @xmath6, @xmath104). to prove an analogous result for contractions we do not need such a strong property : we simply require that any two different points can be joined by a sequence of consecutive steps whose lengths are arbitrarily close to @xmath36. the following result is used in example [ex3] to construct a space where every contraction is a constant map but there are nonconstant contractive maps. [pro : contrconst] if equation holds and for all @xmath136, @xmath137 there exists @xmath138 such that, @xmath42, @xmath139 and @xmath140 for all @xmath141 then any contraction from @xmath1 into itself is constant. suppose that @xmath142 for all @xmath6 where @xmath143. define @xmath144 ; it is clear that for all @xmath45 we have that @xmath145 which implies @xmath146 and @xmath147. thus @xmath148. [ex3] let @xmath129 and let @xmath2 be defined as follows @xmath149 it is easy to check that this is a distance and that the hypotheses of proposition [pro : contrconst] are satisfied (since @xmath150, we can always take @xmath151 and @xmath152 sufficiently large). hence in this case any contraction is a constant function. nevertheless @xmath131 is a (nonconstant) contractive map. hence @xmath153. let us consider now a nonoriented graph along with its natural distance @xmath2. if @xmath117, @xmath154 is the natural distance restricted to @xmath155 and @xmath156, then condition (ii) of theorem [main] is equivalent to the existence, given any couple of vertices @xmath157, of a finite sequence @xmath158 of vertices in @xmath119 such that @xmath159 for all @xmath160. in particular if @xmath161 then any contractive map on @xmath119 is constant if and only if @xmath119 is a connected subgraph. by taking @xmath162 we have that there are no nonconstant, contractive maps on @xmath1 if and only if @xmath1 is a connected graph.
Conclusions
we come back to our original question : may the group of people move according to the rules (i) and (ii) as stated in section [sec : intro]? we may reasonably suppose that the group is finite and with cardinality strictly greater than @xmath163 (if there are just @xmath163 people then any contractive map is constant and they can not move). hence the metric space is finite and equation is fulfilled. moreover one can assume that to different couples of places correspond different distances (i.e. if @xmath164 and @xmath165 then @xmath166 unless @xmath5 and @xmath167). under these assumptions, according to corollary [cor : main], we know that the group can move. moreover, since the contractive map in this case is a contraction, there exists one (and only one) person who does not move at all (this is the fixed point of the contraction). | we consider discrete metric spaces and we look for nonconstant contractions. we introduce the notion of contractive map and we characterize the spaces with nonconstant contractive maps.
we provide some examples to discussion the possible relations between contractions, contractive maps and constant functions.
finally we apply the main result to the subgraphs of a nonoriented, connected graph.
.6 cm * keywords * : discrete metric space, contractive map, contraction, fixed point, nonoriented graph * ams subject classification * : 47h09, 54e99. | 1011.4163 |
Introduction
the precise determination of the cabibbo - kobayashi - maskawa (ckm) matrix element @xmath8 is a crucial step for @xmath9 physics to pursue phenomena beyond the standard model. in particular, the precision achieved in determining the apex of the unitarity triangle may be limited by @xmath10, even with future high - statistics experiments. the current determination of @xmath10 @xcite is made through inclusive @xcite and exclusive @xcite @xmath9 decays. the heavy quark expansion offers a method to evaluate the hadronic transition amplitude in a systematic way. in particular, at the kinematic end point the exclusive @xmath11 matrix element is normalized in the infinite heavy quark mass limit, and the correction of order @xmath12 vanishes as a consequence of luke s theorem @xcite. it is thus possible to achieve an accuracy on @xmath10 of a few percent. calculations of the @xmath13 (and higher order) deviations from the heavy quark limit have previously been attempted with the non - relativistic quark model and with qcd sum rules. lattice qcd has the potential to calculate exclusive transition matrix elements from first principles. the shapes of the @xmath14 decay form factors have already been calculated successfully with propagating @xcite, static @xcite, and non - relativistic @xcite heavy quarks. on the other hand, a precise determination of the absolute normalization of the form factors has not been achieved. this paper fills that gap for the decay @xmath0. previous lattice calculations were unable to obtain the normalization of the form factors for various reasons. first, the statistical precision of the three point function @xmath15, which is calculated by monte carlo integration, has not been enough. second, perturbative matching between the lattice and the continuum currents has been a large source of uncertainty. since the local vector current defined on the lattice is not a conserved current at finite lattice spacing @xmath16, the matching factor is not normalized even in the limit of degenerate quarks. although one - loop perturbation theory works significantly better with tadpole improvement @xcite, the two - loop contribution remains significant (@xmath17 5 %). last, the systematic error associated with the large heavy quark mass must be understood. previous work with wilson quarks @xcite, for which the discretization error was as large as @xmath18, could not address the @xmath12 dependence in a systematic way when @xmath19. in this paper we present a lattice qcd calculation of the @xmath0 decay form factor. for the heavy quark we use an improved action @xcite for wilson fermions, reinterpreted in a way mindful of heavy - quark symmetry @xcite. discretization errors proportional to powers of @xmath20 do not exist in this approach. instead, discretization errors proportional to powers of @xmath21 remain, although they are intertwined with the @xmath12 expansion. the first extensive application of this approach to heavy - light systems was the calculation @xcite of the heavy - light decay constants, such as @xmath22 and @xmath23. there the lattice spacing dependence was studied from direct calculations at several lattice spacings, and a very small @xmath16 dependence was observed. the third difficulty mentioned above is, thus, no longer a problem. to obtain better precision on the semi - leptonic form factors, we introduce ratios of three - point correlation functions. the bulk of statistical fluctuations from the monte carlo integration cancels between numerator and denominator. furthermore, the ratios are, by construction, identically one in both the degenerate - mass limit and the heavy - quark - symmetry limit. consequently, statistical and all systematic errors, as well as the signal, are proportional to the deviation from one. the first and second difficulties given above are, thus, also essentially cured. the ratio of correlation functions for the calculation of @xmath24 corresponds to the ratio of matrix elements, @xmath25 in which all external states are at rest. the denominator may be considered as a normalization condition of the heavy - to - heavy vector current, since the vector current @xmath26 with degenerate quark masses is conserved in the continuum limit, and its matrix element is, therefore, normalized. as a result the perturbative matching between the lattice and continuum currents gives only a small correction to @xmath27. for the calculation of @xmath3 we define another ratio, corresponding to matrix elements @xmath28 where equality holds when the final - state @xmath29 meson has small spatial momentum. by construction, the ratio produces a value of @xmath30 that vanishes when the @xmath31 quark has the same mass as the @xmath32 quark, as required by current conservation. this method does not work as it stands for the @xmath33 decay form factors. the axial vector current mediates this decay, and it is neither conserved nor normalized. we will deal separately with this case in another paper. this paper is organized as follows. section [sec : form_factors] contains a general discussion of form factors for the exclusive decay @xmath0. sections [sec : hqet_and_1/m_q_expansion] and [sec : lattice_and_hq] discuss heavy quark effective theory and the @xmath12 expansion in the continuum and with the lattice action used here. section [sec : lattice_details] contains details of the numerical calculations. sections [sec : calculation_of_h_+][sec : heavy_quark_mass_dependence_of_h-] present our results. sections [sec : calculation_of_h_+] and [sec : heavy_quark_mass_dependence_of_h+] discuss the form factor @xmath34 and its mass dependence. sections [sec : calculation_of_h_-] and [sec : heavy_quark_mass_dependence_of_h-] do likewise for @xmath30. we compare the results from the fits of the mass dependence to corresponding results from qcd sum rules in sec. [sec : comparison_with_the_qcd_sum_rules]. the values of @xmath24 and @xmath35 at the physical quark masses are combined in sec. [sec : result_for_f] into a result for the form factor @xmath36, which with experimental data determines @xmath10. we give our conclusions in sec. [sec : conclusions].
@xmath0 form factors
the decay amplitude for @xmath0 is parametrized with two form factors @xmath1 and @xmath3 as @xmath37,\]] where @xmath38 and @xmath39 are the velocities of the @xmath9 and @xmath29 mesons, respectively, and @xmath40. the square of the momentum transferred to the leptons is then @xmath41. we denote by the symbol @xmath42 the physical vector current, to distinguish it from currents in heavy quark effective theory (hqet) and in lattice qcd. the differential decay rate reads @xmath43 with @xmath44 at zero recoil (@xmath45, so @xmath2) one expects @xmath36 to be close to one, because of heavy quark symmetry. from ([eq : differential_decay_rate]) a determination of @xmath10 consists of the following three steps : measure @xmath46 in an experiment, extrapolate it to the zero - recoil limit assuming some functional form, and use the theoretical input of @xmath36. in this paper we report on a new calculation of @xmath36 with lattice qcd, which is model independent, at least in principle. the present calculation includes the leading corrections to the heavy - quark limit : radiative corrections to the static limit of @xmath47, the @xmath12 contribution of @xmath48, and the @xmath13 contributions of @xmath49. radiative corrections of order @xmath50 to @xmath35 are not yet available, but these and further corrections, of order @xmath51, @xmath52, etc., could be included in future applications of our numerical technique, once the needed perturbative results become available. an obvious disadvantage in using the @xmath0 decay mode is that the branching fraction is much smaller than the @xmath33 mode. another, but not less important, shortcoming is that the phase - space suppression factor @xmath53 makes the extrapolation of the experimental data to @xmath2 more difficult than for @xmath33, where the corresponding factor is @xmath54. nevertheless, the experimental result of the cleo collaboration @xcite shows that the above method certainly works, even with current statistics. that means that future improvement of the statistics will allow a much better determination of @xmath10, providing an important cross check against other methods.
Hqet and the @xmath12 expansion
many important theoretical results have been obtained for the form factors with hqet. the lagrangian of hqet uses fields of infinitely heavy quarks, so that heavy quark symmetries are manifest. the effects of finite quark mass are included through the @xmath12 expansion and through radiative corrections. for example, at zero recoil the form factor @xmath34 is given by @xmath55,\]] where @xmath56 represents a matching factor relating the vector current in ([eq : definition_of_the_form_factors]) to the current in hqet @xcite. the absence of the @xmath57 term in ([eq:1/m_q - expansion+]) is a result of a symmetry under an interchange of initial and final states in ([eq : definition_of_the_form_factors]), and it is known as a part of luke s theorem @xcite. the same symmetry also restricts the form of the @xmath58 terms. the matching factor, defined so that the identity @xmath59 holds for matrix elements, is an ultraviolet- and infrared - finite function of @xmath60. through one - loop perturbation theory, @xmath61 the two - loop coefficient is also available @xcite. the vector current defined with lattice fermion fields has properties similar to @xmath62. there is a normalization factor @xmath63 defined so that @xmath64 holds for matrix elements. the factor @xmath63 depends strongly on the (lattice) quark masses @xmath65 and @xmath66 @xcite, and its one - loop corrections are large. in the past, such uncertainties in the normalization prevented a calculation of @xmath47 with the sought - after accuracy. one can, however, capture most of the normalization nonperturbatively by writing, with explicit flavor indices, @xmath67 in our ratio ([eq : ratio_1]) the flavor - diagonal factors cancel, so our method avoids the major normalization uncertainties. the remaining radiative correction @xmath68 depends on the ratio of quark masses and the lattice spacing. in the continuum limit, @xmath69 and @xmath70 with @xmath60 fixed, @xmath71 by construction. in the static limit, @xmath72 and @xmath73 with @xmath16 and @xmath60 fixed, @xmath74 because the lattice theory strictly obeys heavy - quark symmetries. in numerical work one is somewhere in between, but the limits imply that @xmath75 is never far from unity. two of us have computed @xmath75 at one loop in perturbation theory @xcite, verifying explicitly that the radiative correction is small. similarly, the ratio ([eq : ratio_2]) is described by the expansion @xmath76 where @xmath77 is a coefficient from matching the currents in ([eq : ratio_2]) to hqet. like @xmath78, it is an ultraviolet- and infrared - finite function of @xmath60, and @xmath79 at leading order. the ratio ([eq : ratio_2]) again captures nonperturbatively most of the renormalization of the lattice currents, apart from a factor @xmath80 to compensate for the difference between the radiative corrections with a fixed lattice cutoff and with no ultraviolet cutoff. in the continuum limit @xmath81, and in the static limit @xmath82. again, explicit calculation verifies that the one loop contribution remains small between the limits. in the rest of this paper, we do not write the matching factors @xmath83 when there is no risk of confusion. in the final result, on the other hand, they are included.
Lattice qcd and heavy quark symmetry
@xcite, it was shown that the usual action for light quarks @xcite can be analyzed in terms of the operators of hqet. therefore, it can be used as the basis of a systematic treatment of heavy quarks on the lattice, even when the quark mass in lattice units, @xmath20, is not especially small. the key is to adjust the couplings in the lattice action so that operators are normalized as they are in hqet. when @xmath84, as is the case for charmed quarks at the smaller lattice spacings in common use, this is essentially automatic, because the higher order terms of the heavy quark expansion come from the dirac term of the lattice action, as in continuum qcd. when @xmath85, as is the case for bottom quarks, one can apply the formalism of hqet to the lattice theory to obtain the normalization conditions, as sketched below. in either case, the kinetic energy is normalized nonperturbatively by tuning the quark mass according to some physical condition. other operators are often normalized perturbatively as an initial approximation but ultimately may be normalized nonperturbatively. in the numerical calculations presented here, we use an action introduced by sheikholeslami and wohlert @xcite, @xmath86 where the index @xmath87 runs over heavy and light flavors. the hopping parameter @xmath88 is related to the bare quark mass, @xmath89 where @xmath90 is the value of @xmath91 needed to make a quark massless. the flavor - independent matrix @xmath92 vanishes except when @xmath93, for some spacetime direction @xmath94. the kinetic energy arises from this term. the gluons field strength @xmath95 is defined on a set of paths shaped liked a four - leaf clover, so @xmath96 is often called the `` clover '' action. with @xmath97 one has the wilson action. for the light quark the clover coupling @xmath98 can be chosen so that there are lattice artifacts of order @xmath99. in our numerical work we take an approximation to the optimal value, leaving an artifact of order @xmath100. for heavy quarks, the clover action ([eq : sw_action]) has the same heavy - quark spin and flavor symmetries as continuum qcd, even at nonzero lattice spacing. consequently, we can use the machinery of hqet to characterize the lattice theory. the same operators as in continuum qcd appear, but the coefficients can differ. through first order in @xmath12 there are three operators in the heavy quark effective hamiltonian, @xmath101 where @xmath102 is a heavy quark field, and the coefficients @xmath103, @xmath104, and @xmath105 depend on the bare mass and the gauge coupling. because the lattice breaks relativistic invariance, the three `` masses '' are not necessarily equal, except as @xmath106. at tree level, the rest mass @xmath107, and the (inverse) kinetic mass @xmath108 the first term can be traced to the dirac term of the lattice action, and the second to the wilson term. the one - loop corrections to @xmath109 and @xmath110 are also available @xcite. the chromomagnetic mass @xmath111 is considered below. in the heavy quark effective theory, the rest mass term @xmath112 commutes with the rest of the hamiltonian and, thus, decouples from the dynamics. as with decay constants @xcite, one can derive the expansions like ([eq:1/m_q - expansion+]) and ([eq:1/m_q - expansion-]) within the lattice theory, and the rest mass disappears from physical amplitudes @xcite. on the other hand, adjusting the bare quark mass so that @xmath113 is the way to normalize the kinetic operator @xmath114 correctly. this normalization can be implemented nonperturbatively by demanding that the energy of a hadron have the correct momentum dependence. in our numerical work we use the @xmath9 and @xmath29 mesons for this purpose. furthermore, one can correctly normalize the chromomagnetic operator @xmath115 by adjusting the clover coupling @xmath98, as a function of the gauge coupling, so that @xmath116. for example, at tree level the desired adjustment is @xmath117. in our numerical work, we choose @xmath98 in a way that sums up tadpole diagrams, which dominate perturbation theory. this amounts to normalizing the chromomagnetic operator perturbatively. in summary, we adjust the bare mass @xmath118 and clover coupling @xmath98 so that the leading effects of the heavy - quark expansion are correctly accounted for @xcite. previous work in the literature chose instead to adjust the bare mass until @xmath119, which introduces an unnecessarily large error, expansion with higher orders.] proportional to @xmath120. under renormalization the heavy quark kinetic energy can mix with the rest mass term in a power divergent way. because the lattice action used here contains both, the rest mass fully absorbs the power divergence. a related problem is the ambiguity owing to renormalons @xcite, which appears in some quantities in hqet or nonrelativistic qcd (nrqcd). it is irrelevant to our work, because we calculate physical quantities, namely the masses of the @xmath9 and @xmath29 mesons and decay amplitude for @xmath0. to complete the correspondence of the lattice theory to hqet we must consider the vector current. at order @xmath12 of hqet @xmath121 where the coefficient @xmath122 depends on the current employed. the heavy - heavy current on the lattice is constructed by defining a rotated field @xcite, @xmath123\psi^f, \label{eq : rotation}\]] where @xmath124 is the quark field in the hopping - parameter form of the action ([eq : sw_action]). then the lattice vector current @xmath125 and @xmath126. both @xmath127 and @xmath128 depend on the gauge coupling, the masses, and (at higher orders) on the dirac matrix in ([eq : vhh]). they are adjusted so that the normalization and momentum dependence of matrix elements matches the continuum, respectively. in particular, at tree level the coefficient in ([eq : vhh]) is @xmath129 and the condition @xmath130 prescribes a condition on @xmath131 @xcite. from the properties of the operators under heavy - quark symmetry, it follows that the @xmath104 and @xmath105 terms in ([eq : h]) could give a contribution to @xmath24, but not to @xmath35 @xcite. on the one hand, these contributions to @xmath24 must be symmetric under interchange of the initial and final states, but, on the other hand, they must vanish when the initial and final quark masses are the same. consequently, there can be no contributions linear in either @xmath104 or @xmath105. our definition of @xmath24 enjoys this property, by construction, because ([eq : ratio_1]) manifestly respects the interchange symmetry. similarly, the @xmath122 terms in ([eq : vhh]) give a contribution only to @xmath35. it must be anti - symmetric under interchange of the initial and final states and must vanish when the initial and final quark masses are the same. our definition of @xmath35, again by construction, ensures that only the combination @xmath132 appears. this feature is taken into account in sec. [sec : heavy_quark_mass_dependence_of_h-]. in ([eq:1/m_q - expansion+]) and ([eq:1/m_q - expansion-]) we seek contributions of order @xmath13. these come from double insertions of the @xmath12 terms in ([eq : h]) and ([eq : vhh]), and from @xmath13 terms implied by the ellipses. remarkably, the latter cancel when @xmath24 and @xmath35 are defined by the double ratios ([eq : ratio_1]) and ([eq : ratio_2]) @xcite. this is easy to understand if one starts with the matrix elements. the @xmath133 and @xmath134 corrections to the action arise from the initial or final state only. to this order, one can factorize them. they drop out of the double ratios, because the numerator and denominator of ([eq : ratio_1]) or ([eq : ratio_2]) contain the same number of @xmath9 and @xmath29 factors. the same applies to @xmath133 and @xmath134 corrections to the current. there may be a nonfactorizable correction to the current with coefficient @xmath135, where the function @xmath136 is unknown, except that in perturbation theory it starts at one loop and that @xmath137. in the long run, one would like to pick up terms of order @xmath138 and higher. because the bottom quark is so heavy, these are dominated by the @xmath139 terms. with the normalization conditions outlined here @xcite, these come automatically from the dirac term, as in continuum qcd. in future work, at smaller lattice spacings, the dirac term will dominate, generating contributions to all orders in @xmath140.
Lattice details
our numerical data are obtained in the quenched approximation on a @xmath141 lattice with the plaquette gluon action at @xmath142. we take a mean - field - improved @xcite value of the clover coupling, which on this lattice is @xmath143. out of 300 configurations generated for our previous work @xcite, we use 200 configurations. we usually define the inverse lattice spacing through the charmonium 1s1p splitting, finding @xmath144 gev. for comparison, with the kaon decay constant @xmath145 gev, and the difference is thought to be part of the error of quenching. because the form factors are dimensionless, the lattice spacing affects them only indirectly, through the adjustment of the quark masses. to investigate the heavy quark mass dependence of the form factors we take @xmath146, 0.089, 0.100, 0.110, 0.119 and 0.125, and consider several combinations for the heavy quarks in the initial and final states. the mass of the spectator light quark is usually taken to be close to that of the strange quark, for which @xmath147. we examine the effect of chiral extrapolation using four @xmath148 values, 0.1405, 0.1410, 0.1415, and 0.1419, for various combinations of the initial and final heavy quark masses @xmath149, 0.110, and 0.119. the critical hopping parameter is @xmath150. for the computation of the matrix element '' instead of `` @xmath151 '' to indicate the @xmath152 meson, and we use `` @xmath9 '' or `` @xmath29 '' for any values of the heavy quark masses.] @xmath153 we calculate the three point correlation function @xmath154 with @xmath155 from ([eq : v_0]) and @xmath156. the light quark propagator is solved with a source at time slice @xmath157, and we place the interpolating field for @xmath9 at @xmath158, where we use the source method. the interpolating fields @xmath9 and @xmath29 are constructed with the 1s state smeared source as in ref. the spatial momentum @xmath159 carried by the final state is taken to be (0,0,0), (1,0,0), (1,1,0), (1,1,1) and (2,0,0) in units of @xmath160, where @xmath161 is the physical size of the box ; in our case, @xmath162. the numerical results presented below are obtained from uncorrelated fits to ratios of these three - point functions. the statistical errors are estimated with the jackknife method. for a subset of the data we have repeated the analysis with correlated fits and the bootstrap method. we find no statistically significant difference. in much of the numerical work presented in this paper, we set the coefficients @xmath131 of the rotation ([eq : rotation]) to zero. from the discussion following ([eq : vhh]) the dependence on @xmath131 enters directly through @xmath122, and indirectly by changing @xmath83. on the scattering matrix elements of the spatial current @xmath163, this should make a small (@xmath164 or so) effect. on the temporal current @xmath165, the effect should be tiny. both expectations are checked at representative choices of the heavy quark masses, and the uncertainty introduced into the spatial current is propagated to the final result.
Calculation of @xmath27
the form factor @xmath166 at zero recoil is obtained directly from the three - point correlation functions ([eq : three_point_correlator]), setting all three momentum to be zero. we define a ratio dependence of the isgur - wise function, which is the infinite mass limit of @xmath1.] @xmath167 in which the exponential dependence on @xmath168 associated with the ground state masses cancels between the numerator and denominator. when the current and two interpolating fields are separated far enough from each other, the contribution of the ground state dominates and @xmath169 suppressing radiative corrections. here we use the definition ([eq : definition_of_the_form_factors]) and the unit normalization of @xmath27 in the equal mass case. thus, we expect @xmath170 to be constant as a function of @xmath168, and its value represents the form factor squared. in fig. [fig : r_btod] we plot the ratio @xmath171 for two representative combinations of mass parameters. we observe a nice plateau extending over about five time slices, and our fit over the interval @xmath172 is shown by the solid line. to see if the plateau is stable under the change of the position of the interpolating field, we repeat the calculation changing the time @xmath173 of the @xmath9-meson interpolating field. the results with @xmath174 and 8 are shown in fig. [fig : r_btod_tb] together with the one with @xmath175. we observe that the plateau is very stable and conclude that the extraction of the ground state is reliable. in the following analysis we use the result with @xmath176, and the numerical data for each @xmath177 are given in table [tab : h+_data]. 0.3em .numerical data for @xmath170, which corresponds to @xmath178, at @xmath147. rows (columns) are labeled by the value of @xmath177 in the initial (final) state. combinations without data have not been calculated in this work. the diagonal elements are 1 by construction. [cols= " <, <, <, <, <, <, < ",] since the systematic errors in @xmath66 and in @xmath65 are correlated, we consider the central and two limiting combinations only. the statistical errors on @xmath24 and @xmath35 are estimated with the jackknife method, so that the resulting precision is better than that obtained by adding in quadrature the errors on coefficients @xmath179. in the physical amplitude @xmath36, which is the linear combination of @xmath24 and @xmath35 given in ([eq : form_factor_relation]), the uncertainty from adjusting the quark masses largely cancels, and the value of @xmath36 is very stable. to obtain the physical result, we must now fold in the radiative correction @xmath75, relating the lattice current to the continuum. two of us recently have calculated this factor to one loop @xcite, and at @xmath180 and @xmath181 they find @xmath182. the lepage - mackenzie scale @xmath183 for the coupling @xmath184 @xcite has also been calculated, and at the same quark masses the result is @xmath185. at @xmath186, @xmath187 and the correction to @xmath188 is @xmath189, taking the error of omitting higher orders to be 20 % of the one - loop correction. a similar one - loop calculation for @xmath190, which modifies @xmath35, is not yet available. we allow, therefore, a systematic uncertainty for this effect. our results for the form factors are @xmath191 where the error estimates are as follows. the first error comes from statistics, after the chiral extrapolation ; the second from adjusting the heavy quark masses ; and the third error from unknown radiative corrections, two loops and higher for @xmath34 and one loop and higher for @xmath30. the chiral extrapolations, which are shown in figs. [fig : h+_chiral] and [fig : h-_chiral], double the statistical errors of table [tab : form_factors], without changing the central values. our main result is the value of the form factor entering the decay rate, at zero recoil. inserting the physical values of the @xmath9 and @xmath29 meson masses and the results ([eq : result_h+]) and ([eq : result_h-]) into ([eq : form_factor_relation]), @xmath192 where errors are from statistics, heavy quark masses, and omitted radiative corrections. the last of these could be reduced substantially by calculating the radiative correction factor @xmath190 to one loop. two sources of uncertainty have yet to be investigated carefully. they are the dependence on the lattice spacing and the effects of the quenched approximation. from our experience with @xmath22 @xcite, we might suppose that these effects are a few percent and @xmath193, respectively. the ratios have been constructed so that all sources of error, including these, vanish for equal heavy quark masses. it is, therefore, our expectation that these percentages apply not to @xmath194 but to @xmath195. that means that these two sources of error should be under good control, just as we have found with the other sources of uncertainty.
Conclusions
in this paper we have shown that precise lattice calculations of the zero - recoil form factors @xmath24 and @xmath35 are possible. the principal technical advance is to consider ratios of matrix elements, in which a large cancellation of statistical and systematic errors takes place. the numerical data are interpreted in a way mindful of heavy quark symmetry @xcite. we find, therefore, that the dependence of the form factors on the heavy quark mass is well described by @xmath12 expansions, and we obtain the coefficients in the expansions. our control over the heavy quark mass dependence allows us to determine the individual form factors @xmath24 and @xmath35, as well as the physical combination @xmath36. the main results ([eq : result_h+])([eq : result_f(1)]) account for most uncertainties, but not the dependence on the lattice spacing or the effect of the quenched approximation. since our method is designed to yield the deviation of @xmath36 from one, we do not expect these qualitatively to spoil the quoted precision. with the proof of principle provided by this work, it should be possible, in the short term, to obtain @xmath36 with control over all sources of uncertainty and an error bar that is small enough to be relevant to the determination of @xmath10.
Acknowledgments
we thank zoltan ligeti and ulrich nierste for comments on the manuscript. high - performance computing was carried out on acpmaps ; we thank past and present members of fermilab s computing division for designing, building, operating, and maintaining this supercomputer, thus making this work possible. fermilab is operated by universities research association inc., under contract with the u.s. department of energy. sh is supported in part by the grants - in - aid of the japanese ministry of education under contract no. axk is supported in part by the doe oji program under contract de - fg02 - 91er40677 and through the alfred p. sloan foundation. 99 for recent reviews, see _ the babar physics book _, edited by p.f. harrison and h.r. quinn (slac - r-504), sec. 8 ; bigi _ et al_. @xcite. p. ball, m. beneke and v.m. braun, phys. rev. * d52 *, 3929 (1995). i. bigi, m. shifman and n. uraltsev, annu. nucl. part. sci. * 47 *, 591 (1997). m. neubert, phys. lett. * b338 *, 84 (1994). m. shifman, n.g. uraltsev and a. vainshtein, phys. rev. * d51 *, 2217 (1995). m.e. luke, phys. lett. * b252 *, 447 (1990). c.w. bernard, y. shen and a. soni, phys. lett. * b317 *, 164 (1993). ukqcd collaboration, k.c. bowler _ et al. _, phys. rev. * d52 *, 5067 (1995). t. bhattacharya and r. gupta, nucl. b (proc. suppl.) * 42 *, 935 (1995) ; nucl. phys. b (proc. suppl.) * 47 *, 481 (1996). j.e. mandula and m.c. ogilvie, nucl. b (proc. suppl.) * 34 *, 480 (1994) ; ` hep - lat/9408006 `. t. draper and c. mcneile, nucl. b (proc. suppl.) * 47 *, 429 (1996). c. bernard _ et al. _, b (proc. suppl.) * 63 *, 374 (1998). j. christensen, t. draper and c. mcneile, nucl. b (proc. suppl.) * 63 *, 377 (1998). s. hashimoto and h. matsufuru, phys. rev. * d54 *, 4578 (1996). lepage and p.b. mackenzie, phys. rev. * d48 *, 2250 (1993). b. sheikholeslami and r. wohlert, nucl. phys. * b259 *, 572 (1985). el - khadra, a.s. kronfeld and p.b. mackenzie, phys. rev. * d55 *, 3933 (1997). el - khadra, a.s. kronfeld, p.b. mackenzie, s.m. ryan and j.n. simone, phys. rev. * d58 * 14506 (1998). jlqcd collaboration, s. aoki _ et al. lett. * 80 *, 5711 (1998). cleo collaboration, m. athanas _ et al. * 79 *, 2208 (1997). m. neubert, nucl. b371 *, 149 (1992). a. czarnecki, phys. lett. * 76 *, 4124 (1996) ; + a. czarnecki and k. melnikov, nucl. b505 *, 65 (1997). a.s. kronfeld and s. hashimoto, nucl. b (proc. suppl.) * 73 *, 387 (1999) ; and (work in progress). b.p.g. mertens, a.s. kronfeld, and a.x. el - khadra, phys. rev. * d58 *, 034505 (1998). kronfeld, nucl. b (proc. suppl.) * 42 * 415 (1995). a.s. kronfeld, fermilab - conf-99/251-t, hep - lat/9909085. see, for example, m. beneke, phys. rep. * 317 *, 1 (1999). l. randall and m.b. wise, phys. b303 *, 135 (1993). boyd and b. grinstein, nucl. * b451 *, 177 (1995). falk and m. neubert, phys. rev. * d47 *, 2965 (1993). t. mannel, phys. rev. * d50 *, 428 (1994). m. neubert and c.t. sachrajda, nucl. phys. * b438 *, 235 (1995). + the borel transforms of wilson coefficients presented there imply a large contribution of order @xmath196 [z. ligeti, (private communication)]. a further implication is that the brodsky - lepage - mackenzie (blm) scale @xcite is _ tiny_. s.j. brodsky, g.p. lepage, and p.b. mackenzie, phys. rev. * d28 *, 228 (1983). m. neubert, phys. rev. * d46 *, 3914 (1992). z. ligeti, y. nir and m. neubert, phys. rev. * d49 *, 1302 (1994). m. neubert, phys. lett. * b306 *, 357 (1993). | a lattice qcd calculation of the @xmath0 decay form factors is presented.
we obtain the value of the form factor @xmath1 at the zero - recoil limit @xmath2 with high precision by considering a ratio of correlation functions in which the bulk of the uncertainties cancels.
the other form factor @xmath3 is calculated, for small recoil momenta, from a similar ratio. in both cases, the heavy quark mass dependence is observed through direct calculations with several combinations of initial and final heavy quark masses.
our results are @xmath4 and @xmath5. for both
the first error is statistical, the second stems from the uncertainty in adjusting the heavy quark masses, and the last from omitted radiative corrections. combining these results
, we obtain a precise determination of the physical combination @xmath6, where the mentioned systematic errors are added in quadrature. the dependence on lattice spacing and the effect of quenching
are not yet included, but with our method they should be a fraction of @xmath7. | hep-ph9906376 |
Introduction
let @xmath1 be two partitions of an integer @xmath2, @xmath3 and let @xmath4 and @xmath5 be two _ coprime _ polynomials of degree @xmath2 having the following factorization patterns : @xmath6 in these expressions we consider the multiplicities @xmath7 and @xmath8, @xmath9, @xmath10 as being given, while the roots @xmath11 and @xmath12 are not fixed, though they must all be distinct. in this paper we study polynomials satisfying ([eq : p - and - q]) and such that _ the degree of their difference @xmath13 attains its minimum_. numerous papers, mainly in number theory, were devoted to the study of such polynomials. [assump] throughout the paper, we always assume that * the greatest common divisor of the numbers @xmath14 is 1 ; * @xmath15. the case of partitions @xmath16, @xmath17 not satisfying the above conditions can easily be reduced to this case (see @xcite). in 1995, zannier @xcite proved that under the above conditions the following statements hold : 1. this bound is always attained, whatever are @xmath16 and @xmath17. [def : dz] a pair of polynomials @xmath19 such that @xmath4 and @xmath5 are of the form and @xmath20 is called _ davenport zannier pair _, or _ dz - pair _ for short. the pair of partitions @xmath21 is called the _ passport _ of the dz - pair. obviously, if @xmath19 is a dz - pair with a passport @xmath21, and if we take @xmath22, @xmath23 where @xmath24, then @xmath25 is also a dz - pair with the same passport. we call such dz - pairs equivalent. [def : over] we say that a dz - pair @xmath19 is _ defined over _ @xmath0 if @xmath26 $]. we say that an equivalence class of dz - pairs is defined over @xmath0 if there exists a representative of this class which is defined over @xmath0. by abuse of language, in what follows, we will use the shorter term `` dz - pair '' to denote also an equivalence class of dz - pairs. in our previous paper @xcite, using the _ theory of dessins denfants _ (see, for example, ch. 2 of @xcite), we established a correspondence between dz - pairs and _ weighted bicolored plane trees_. these are bicolored plane trees whose edges are endowed with positive integral weights. the degree of a vertex is defined as the sum of the weights of the edges incident to this vertex. obviously, the sum of the degrees of black vertices and the sum of the degrees of white vertices are both equal to the total weight of the tree. let @xmath27 and @xmath28 be two partitions of the total weight @xmath2 which represent the degrees of black and white vertices respectively. the pair @xmath21 is called the _ passport _ of the tree in question. [biject] there is a bijection between dz - pairs with a passport @xmath21 on one hand, and weighted bicolored plane trees with the same passport on the other hand. [uni] a weighted bicolored plane tree such that there is no other tree with the same passport is called _ unitree_. general facts of the theory of dessins denfants imply that dz - pairs corresponding to unitrees are defined over @xmath0. basing on our experience, we claim that this class represents a vast majority of dz - pairs defined over @xmath0. the other examples may roughly be subdivided into two categories. the members of the first one are constructed as compositions of dz - pairs corresponding to unitrees. the second category is, in a way, a collection of exceptions. still, the latter category is no less interesting since it involves some subtle combinatorial and group - theoretic invariants of the galois action on dz - pairs and on weighted trees. * the main result of @xcite is the classification of all unitrees. the main result of the present paper is a complete list of the corresponding polynomials. * the final part of @xcite is devoted to the study of galois invariants of weighed trees. in the final part of the present paper we compute the corresponding polynomials. the class of unitrees comprises ten infinite series, denoted from @xmath29 to @xmath30, and ten sporadic trees, denoted from @xmath31 to @xmath32. the pictures of these trees are given below in the text. dz - pairs corresponding to the series from @xmath29 to @xmath30 are presented in sects. [sec : a] to [sec : j] ; those corresponding to the sporadic trees from @xmath31 to @xmath32, in sect. [sec : sporadic]. the galois action is treated in sects. [sec : galois] to [sec : other]. for individual dz - pairs, a computation may turn out to be difficult, sometimes even extremely difficult, but the verification of the result is completely trivial. as to the infinite series, the difficulties grow as a snowball. the `` computational '' part now consists in finding an analytic expression of the polynomials in question, depending on one or several parameters, while the `` verification '' part consists in a _ proof _, which may be rather elaborate. see a more detailed discussion below.
Preliminaries
in 1965, birch, chowla, hall, and schinzel @xcite asked a question which soon became famous:[bchs] _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ let @xmath29 and @xmath33 be two coprime polynomials with complex coefficients ; what is the possible minimum degree of the difference @xmath34? _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ in order for the question to be meaningful we should take @xmath35 and @xmath36 of the same degree and with the same leading coefficient. denote @xmath37, @xmath38, so that @xmath39. let us start with an example. [ex : pair - t] in this example, @xmath40, so that both polynomials @xmath4 and @xmath5 are of degree @xmath41. as to their difference @xmath42, all its coefficients of degrees from 24 down to 6 vanish, so that @xmath43 becomes a polynomial of degree 5. @xmath44 the following two conjectures were proposed in @xcite:[init - problem] 1. for @xmath45, @xmath46, one always has @xmath47. this bound is sharp : that is, it is attained for infinitely many values of @xmath48. the first conjecture was proved the same year by davenport @xcite. the second one turned out to be much more difficult and remained open for 16 years : in 1981 stothers @xcite showed that the bound is in fact attained not only for infinitely many values of @xmath48 but for all of them. a far - reaching generalization of the above result was proved in 1995 by zannier @xcite. let @xmath49 and @xmath50 be two partitions of an integer @xmath2 satisfying the conditions of assumption [assump], and let @xmath4 and @xmath5 be two polynomials of degree @xmath2 having the factorization pattern. then 1. @xmath51. this bound is always attained, whatever are @xmath16 and @xmath17. for the case of cubes and squares considered above we have @xmath52, @xmath53 so that @xmath54 and @xmath55, whence @xmath56 a result equivalent to that of zannier was, in fact, proved, in a very implicit way, by boccara in 1982 @xcite (see also @xcite, page 775). the result of @xcite was purely combinatorial, and relations between combinatorics and polynomials were at the time largely overlooked. recall that a pair of polynomials @xmath19 satisfying and such that the degree of @xmath57 is _ equal _ to the minimum value @xmath58 are called davenport zannier pairs or dz - pairs (definition [def : dz]). the theory of dessins denfants implies that dz - pairs are always defined over the field @xmath59 of algebraic numbers. however, the most interesting case is, without doubt, the one of pairs defined over @xmath0. in 2010, beukers and stewart @xcite undertook a study of dz - pairs of the special type @xmath60, @xmath61, defined over @xmath0. in our paper we study dz - pairs of a general form defined over @xmath0. as we have already said, the framework of our paper is the theory of dessins denfants (see, for example, ch. 2 of @xcite). the main notion of this theory is that of belyi function. for a rational function @xmath62, where @xmath63 is the riemann complex sphere, let us call @xmath64 a _ critical value _ of @xmath65 if the equation @xmath66 has multiple roots. the definition of a belyi function restricted to the planar case is as follows : [def : belyi] a rational function @xmath67 is a _ belyi function _ if @xmath65 has at most three critical values, namely, 0, 1 and @xmath68. [th : riemann] if @xmath62 is a belyi function then : * the preimage @xmath69)$] is a plane map, that is, a connected graph, which is embedded into the sphere in such a way that its edges do not intersect. * the map @xmath70 has a natural bipartite structure : its vertices may be colored in black and white in such a way that each edge would connect vertices of opposite colors. namely, black vertices of @xmath70 are the points @xmath71, and white vertices of @xmath70 are the points @xmath72, the vertex degrees being equal to the multiplicities of the corresponding preimages. * inside each face, there is a unique pole of @xmath65 whose multiplicity is equal to the degree of the face. here the _ degree of a face _ is defined as _ a half _ of the number of surrounding edges. we call this pole the _ center _ of the face in question. in the opposite direction, if @xmath73 is a bicolored plane map then : * there exists a belyi function @xmath65 such that @xmath73 can be realized as a preimage @xmath69)$]. * this function @xmath65 is unique, up to an affine change of the variable @xmath74. * there is a uniquely defined number field @xmath31 corresponding to @xmath73 which is called the _ field of moduli _ of @xmath73. the function @xmath65 can be realized over a number field @xmath75. statements 4 and 5 represent a particular case of riemann s existence theorem. statement 6 follows from the rigidity of the ramified covering @xmath67 and from some general facts of the galois theory. the above theorem, being applied to the dz - pairs, gives the following statement (see more details in @xcite). [prop : dz - belyi] a pair of complex polynomials @xmath19 is a dz - pair with a passport @xmath21 if and only if the rational function @xmath76, where @xmath42, is a belyi function for a bicolored plane map @xmath73 with the following characteristics : * the map @xmath73 has @xmath77 edges, @xmath78 black vertices with the degree distribution @xmath16, and @xmath79 white vertices with the degree distribution @xmath17. the euler formula then implies that the number of faces is @xmath80. * all faces of @xmath73 except the outer one are of degree @xmath81. * the number of the faces of @xmath73 of degree @xmath81 is equal to @xmath82. in other words, the degree distribution of the faces is equal to @xmath83 where @xmath84. furthermore, if @xmath85 is the moduli field of @xmath73, then it is possible to find a corresponding dz - pair such that @xmath86 $]. in other words, in this case the realization field @xmath87 @xmath88see the last statement of theorem [th : riemann]@xmath89 coincides with the field of moduli @xmath31. in particular, an equivalence class of the pair @xmath19 is defined over @xmath0 if and only if the field of moduli of the map @xmath73 is @xmath90. the characteristic which distinguishes the maps corresponding to dz - pairs from other maps is property 2 of the above theorem. we will call the faces other than the outer one _ inner faces_. the maps whose all inner faces are of degree 1 can be easily represented in the form of weighted trees : just merge every sheaf of parallel edges into one edge and indicate the number of edges merged together as the _ weight _ of the corresponding edge of the weighted tree : see fig. [fig : map->tree]. weighted trees are easier to work with than maps. [def : weighted] a _ weighted bicolored plane tree _, or a _ weighted tree _, or just a _ tree _ for short, is a bicolored plane tree whose edges are endowed with positive integral _ weights_. the sum of the weights of the edges of a tree is called the _ total weight _ or the _ degree _ of the tree. the _ degree _ of a vertex is the sum of the weights of the edges incident to this vertex. obviously, the sum of the degrees of black vertices, as well as the sum of the degrees of white vertices, is equal to the total weight @xmath2 of the tree. let the tree have @xmath78 black vertices, of degrees @xmath91, and @xmath79 white vertices, of degrees @xmath92, respectively. then the pair of partitions @xmath21 of the total weight @xmath2 of the tree is called its _ passport_. forgetting the weights and considering only the underlying plane tree, we speak of a _ topological tree_. weighted trees, all of whose edges are of weight 1, will be called _ ordinary trees_. belyi functions for ordinary trees are polynomials (with the only pole at infinity) ; they are usually called _ shabat polynomials_. we call a _ leaf _ a vertex which has only one edge incident to it, whatever is the weight of this edge. by abuse of language, we will also call this edge itself a leaf. the adjective _ plane _ in the above definition means that the cyclic order of branches around each vertex of the tree is fixed, and changing this order will in general produce a different plane tree (though the tree considered as a mere graph, without `` planar '' structure, remains the same). _ all trees considered in this paper will be endowed with the planar structure _ ; therefore, the adjective `` plane '' will often be omitted. the filed of moduli of a unitree is @xmath0, see, e.g., @xcite. therefore, the second part of theorem [prop : dz - belyi] implies the following statement. [uni->q] if a weighted bicolored plane tree is a unitree, then the corresponding equivalence class of dz - pairs is defined over @xmath0. [ex : pair - t.bis] let us consider the tree shown in fig. [fig : t]. it has eight black vertices of degree 3 and twelve white vertices of degree 2, so that its total weight (or degree) is @xmath93. accordingly, @xmath94, and @xmath16 and @xmath17 are the following two partitions of @xmath93 : @xmath95 in the corresponding dz - pair, the polynomial @xmath4 must have eight roots of multiplicity 3, the polynomial @xmath5 must have twelve roots of multiplicity 2. in other words, @xmath96 with @xmath97, and @xmath98 with @xmath99. the difference @xmath42 must be of degree @xmath100. the general results formulated up to now, being applied to this particular tree, imply the following statements : * the mere existence of such a tree implies the existence of polynomials with needed properties. * the fact that there exist polynomials @xmath4 and @xmath5 _ with rational coefficients _ is a consequence of the fact that there exists a unique tree with the passport @xmath101. all this can be affirmed without any computations, just by looking at the picture. as to the polynomials themselves, they are given in example [ex : pair - t]. it turns out that technically it is often much more convenient to work not with the polynomials appearing in dz - pairs but with their _ reciprocals_. [def : reciprocal] for a polynomial @xmath4 of degree @xmath2, its _ reciprocal _ is @xmath102. in many examples, the reciprocals of polynomials forming a dz - pair take the form of initial segments of power series of some special functions. after having observed this phenomenon we learned that it was (re)discovered many times, notably in @xcite, @xcite, @xcite. assume that polynomials @xmath4 and @xmath5 form a dz - pair, so that @xmath103 and denote by @xmath104 the number of edges of the corresponding topological tree. this tree has @xmath105 vertices, therefore it has @xmath106 edges. considering @xmath4 and @xmath5 as power series we may write condition as @xmath107 for the reciprocal polynomials condition is transformed into the following one : @xmath108 where @xmath109 is a polynomial, or, equivalently, to the condition @xmath110 for instance, in the example [ex : pair - t] the polynomials reciprocal to and and to their difference look as follows : @xmath111 the computation of belyi functions has recently become a vast domain of research. a remarkable overview of this activity may be found in @xcite, a paper of 57 pages, with a bibliography of 176 titles. beside a direct approach, involving the solution of a system of polynomial equations, the authors of @xcite also discuss complex analytic methods, modular forms methods, and @xmath78-adic methods. in order to get an idea of the level of difficulty of such a computation let us return once again to example [ex : pair - t]. a naive approach would be to write down polynomials @xmath112 and @xmath113 with indeterminate coefficients @xmath114 and @xmath115, and then equate to zero the coefficients of degrees from 6 to 24 of the difference @xmath34. in this way we get a system of @xmath116 algebraic equations for @xmath117 unknowns. then we may set, for example, @xmath118, @xmath119, and @xmath120. the system thus obtained (19 equations with 19 unknowns) will be of degree @xmath121! obviously, this is not a clever way to proceed. by the way, the solution we are looking for is unique ; all the other solutions of this enormous system are `` parasitic '' ones. for example, the system does not give us any guarantee that the polynomials @xmath29 and @xmath33 obtained as its solution will be coprime. this condition should be added to the system, but this addition will make our situation even worse. notice, however, that, once the result is obtained, its verification is trivial. taking into account the above considerations, we would like to underline one aspect of our work : though we do compute belyi functions for certain individual dessins, the most interesting part of the paper is the computation of belyi functions for _ infinite series _ of dessins which depend on one or several parameters. for infinite series the situation is significantly more complicated than for individual dessins. usually, the first thing to do is to compute quite a few particular cases, sometimes dozens of them (or to use other heuristics whenever possible). then, we need to guess a general pattern of corresponding belyi functions. and, finally, instead of a trivial verification step which was applicable to individual dessins, we should provide a _ proof _, which may turn out to be rather laborious. in the present paper we obviously do not expose the first step of the above procedure. what we do is presenting the final results, that is, the general form of belyi functions in question, and then we give the proofs whenever they are necessary. @xmath122 as it was already said, the unitrees comprise ten infinite series, from @xmath29 to @xmath30, and ten sporadic trees, from @xmath31 to @xmath32. in the subsequent sections we do not strictly follow the `` alphabetic '' order of trees since we prefer to underline the structural properties of belyi functions in question. certain belyi functions are expressed in terms of jacobi polynomials ; there are others which lead to interesting differential relations ; we will also encounter compositions, pad approximants, an application to the hall conjecture, etc.
Stars and binomial series
our first series, called `` series @xmath29 '' in @xcite, is composed of stars - trees, see fig. [fig : a]. all edges except maybe one are of the same weight. this is a three - parametric series. denote the number of leaves of weight @xmath123 by @xmath48 ; then the total weight of the tree is @xmath124. clearly, we may put the only black vertex at @xmath125, put the white vertex of degree @xmath126 at @xmath127, and assume that both @xmath4 and @xmath5 are monic. then @xmath128 and @xmath129 where @xmath29 is a monic polynomial of degree @xmath48 whose roots are the white vertices of degree @xmath123. now, condition takes the form @xmath130 the only thing we need to know is the polynomial @xmath29. the polynomial @xmath131 reciprocal to @xmath29 is the initial segment of the binomial series for @xmath132 up to the degree @xmath48 : @xmath133 [[proof.]] proof. + + + + + + let us pass to reciprocals in : we need to obtain @xmath131 such that @xmath134 let us verify that the polynomial @xmath131 defined in ([a - a]) satisfies the latter equality. we have : @xmath135 where @xmath136 therefore, @xmath137 and @xmath138^s \operatorname*{=}_{x\to 0 } 1 + o(x^{k+1})\]] which concludes the proof. @xmath139 some particular cases of formula were previously found by n.adrianov (unpublished).
Forks and halls conjecture
the two - parametric series of trees shown in fig. [fig : d] was called `` series @xmath140 '' in @xcite. this is the only infinite series of unitrees for which we were able to find the corresponding dz - pairs by computer. let us introduce the following three _ quadratic _ polynomials : [cols= " <, <, < ",] we have @xmath141 and @xmath142, where @xmath143 [[proof.-1]] proof. + + + + + + by, we must prove that @xmath144 clearly, we may assume that the sum of the roots of @xmath29 equals zero. write @xmath145 and calculate, with the help of maple, the first five coefficients of the taylor series in the left - hand side of ([d - maple]). equate now the expressions thus obtained to zero and solve the corresponding system in the unknowns @xmath146. maple returns two solutions : @xmath147 and @xmath148 @xmath149 rejecting the first solution, for which the roots of @xmath29, @xmath33 and @xmath150 coincide, and making an additional normalization by setting the @xmath151, we obtain formulas, and. @xmath139 in 1971, m. hall, jr. @xcite suggested the following two conjectures. 1. there exists a constant @xmath152 such that for all positive integers @xmath153, @xmath154, the inequality @xmath155 holds. the exponent @xmath156 in the above inequality can not be improved. namely, for every @xmath157 there exists a constant @xmath158 such that there are infinitely many pairs of integers @xmath159 satisfying the inequality @xmath160 this first conjecture is neither proved nor disproved. however, a general belief is that in order to be true it should be modified as follows : for each @xmath157 there exists a constant @xmath161 such that for all positive integers @xmath153, @xmath154, the inequality @xmath162 holds. in this form the conjecture is a corollary of the famous @xmath163-conjecture (see, e.g., @xcite, @xcite for further details). as to the second conjecture, in 1982 danilov @xcite proved its stronger version. his result is interesting for us since in his proof he used, in a slightly different normalization, the above polynomials @xmath164, see,,, with the parameters @xmath165. [prop : hall] there exists a constant @xmath150 such that there are infinitely many pairs of integers @xmath159 satisfying the inequality @xmath166 proof. + + + + + + specializing ([a - d]), ([b - d]) and ([c - d]) for @xmath165 and computing the difference @xmath57 we get @xmath167 substituting @xmath168 and dividing both parts by 8 we get @xmath169 let us now consider the factor @xmath170 and try to make it a perfect square ; then will give us a relatively `` small '' difference between a cube and a square. to do that we have to solve the diophantine equation @xmath171 where @xmath172. the last equation is a pell - like equation, that is an equation of the form @xmath173 where @xmath174 is a square - free integer and @xmath175. for @xmath176 this equation is a usual pell equation, and it is well known that any pell equation has infinitely many integer solutions. pell - like equations not necessarily have integer solutions. however, if at least one such solution @xmath177 exists, then we can obtain infinitely many solutions @xmath178 using the following recursion : @xmath179 where @xmath180 is the minimum solution of the equation @xmath181. in our case, @xmath182. equation does have an integer solution @xmath183. returning to , it is easy to verify that for all @xmath184 one has @xmath185 which proves the theorem : there are infinitely many pairs of integers @xmath159 satisfying, with the constant @xmath186. @xmath139 the same polynomials @xmath187 with the parameters @xmath165 were used by dujella @xcite for constructing an infinite series of pairs of polynomials @xmath188 with the following properties : (a) @xmath189, @xmath190 ; (b) @xmath4 and @xmath5 are _ not _ coprime ; (c) @xmath191, so that the minimum degree @xmath192 is not attained, though the discrepancy remains bounded ; (d) in return, @xmath4 and @xmath5 are defined over @xmath0. using other dz - pairs, danilov @xcite and beukers and stewart @xcite obtained results similar to proposition [prop : hall] for the differences between integer powers @xmath193 and @xmath194.
Jacobi polynomials
davenport zannier pairs for the series of trees considered in this section are expressed in terms of jacobi polynomials. the trees in question are constructed as follows. first, we take chain - trees with alternating edge weights @xmath195, see fig. [fig : b]. we must distinguish chains of odd and even length since in one case both ends are of the same color while in the other case they are of different colors. then, we have a right to attach to the end - points an arbitrary number of leaves of the weight @xmath196. in this way we obtain `` odd '' series @xmath197 and `` even '' series @xmath198, see figs. [fig : e1-e3] and [fig : e2-e4]. we call these series `` double brushes ''. note that any of the parameters @xmath199, and also both of them, may be equal to zero. thus, @xmath200 and @xmath201 are particular cases of @xmath202, and @xmath203 and @xmath204 are particular cases of @xmath205. there are two exceptions from the above construction. the first is when the chain part consists of a single edge, so that there is no alternance of weights. we thus obtain the series @xmath150, see fig. [fig : c]. in contrast to the general case, now the weight of leaves may be smaller than the weight of the edge between the leaves. the second exception is when the chain part consists of two edges. in this case it is possible to attach exactly one leaf of weight @xmath196 to one of the ends and exactly two leaves of weight @xmath123 (or @xmath126, to ensure the weight alternance) to the other end. in this way, we get the series of forks @xmath140 already studied in sect. [sec : d]. let us recall some general facts concerning jacobi polynomials ; for more advances and detailed treatment see, for example, @xcite or @xcite. the classical jacobi polynomials @xmath206, @xmath207, are defined for the parameters @xmath208, @xmath209, as orthogonal polynomials with respect to the measure on the segment @xmath210 $], given by the density @xmath211. the restriction @xmath209 is necessary in order to ensure the integrability. the polynomial @xmath206 can also be defined as a unique polynomial solution of the differential equation @xmath212y'+n(n+a+b+1)y { \,=\,}0,\]] satisfying the condition @xmath213 or by the explicit formula @xmath214 notice that equation can be written in the form @xmath215y'+(n+1)(n+a+b)y { \,=\,}0,\]] where @xmath216, implying that the function @xmath217 satisfies. it follows from that @xmath206 are also polynomials in parameters @xmath218 and @xmath219. therefore, their definition can be extended to arbitrary (even complex) values of these parameters. these _ generalized _ jacobi polynomials still satisfy , although they are no longer orthogonal with respect to a measure on the segment @xmath210 $]. similarly, since the function may be represented as a power series in @xmath74 whose coefficients are polynomials in @xmath153, this function satisfies equation for arbitrary @xmath218 and @xmath219. the following key observation will be used in subsequent proofs. if, in the differential operator, we replace @xmath2 with @xmath220, @xmath218 with @xmath221, and @xmath219 with @xmath222, we get exactly the differential operator. therefore, @xmath223 along with satisfies. the last statement, however, should be taken with caution : the subscript @xmath220 must be a non - negative integer since it is the degree of a polynomial. notice that if @xmath218 and @xmath219 do not satisfy the inequalities @xmath209, then the degree in @xmath74 of the polynomial @xmath206 defined by may drop down below @xmath2. indeed, implies that the leading coefficient of @xmath206 is equal to @xmath224 hence, in order to obtain a polynomial of degree @xmath2 we must require that the sum @xmath225 does not take values @xmath226, @xmath227, , @xmath228. in particular, this is always true if @xmath218 and @xmath219 are real and @xmath229 or, equivalently, @xmath230. along with the density @xmath211, which is defined on @xmath210 $], we will use the multivalued complex function @xmath231 (note the change of the sign of the term in the first parenthesis). clearly, this function has three ramification points @xmath232. further, observe that if @xmath233, then any germ of @xmath231 defined near a non - singular point @xmath234 extends to a function @xmath235 which is single - valued in any domain @xmath236 obtained from @xmath237 by removing a simple curve connecting @xmath238 and @xmath81. indeed, in such @xmath236 the function @xmath235 may have a ramification only at infinity. on the other hand, since the analytic continuation of @xmath235 along a loop around infinity is @xmath239, we see that @xmath68 is not a ramification point since @xmath233. in particular, @xmath235 can be expanded into a laurent series at infinity, @xmath240 finally, if @xmath218 and @xmath219 are rational numbers, say @xmath241 then any @xmath235 as above satisfies the condition @xmath242 implying that @xmath235 is defined up to a multiplication by an @xmath104th root of unity, and that for a certain choice of this root the equality @xmath243 holds. by abuse of notation, below we will always use the expression @xmath231 to denote the function @xmath235 which satisfies the equality @xmath243. [lemma] assume that @xmath218 and @xmath219 are rational numbers which satisfy the condition @xmath233. then for any @xmath244 the equality @xmath245 holds. [[proof.-3]] proof. + + + + + + as it was mentioned above, the function satisfies the differential equation, where the function @xmath246 is assumed to be defined on @xmath210 $]. however, since this function is analytic near the origin, we can consider its analytic continuation @xmath247, and the function @xmath248 will satisfy in the domain @xmath236 as above. furthermore, if holds, then @xmath249 implying that the function @xmath250 also satisfies in @xmath236. since the polynomial @xmath206 satisfies the differential equation, we conclude that the functions @xmath251 both satisfy the differential equation @xmath252 where @xmath253\frac{d}{d z}+(n+1)(n+a+b).\]] this implies that the function @xmath254 also satisfies this equation. on the other hand, it is easy to see that if @xmath255 is a function whose laurent expansion at infinity is @xmath256 then @xmath257 where @xmath258 therefore, if @xmath259 satisfies and @xmath260 while @xmath261, we should have either @xmath262 or @xmath263. finally, implies that the leading terms of both @xmath264 and @xmath265 are equal to @xmath266 therefore, the degree of the leading term of their difference @xmath254 is less than @xmath220, hence the only possible case is @xmath263, implying. let @xmath267 be a weighted tree from the series @xmath205 or of its two particular cases @xmath204 or @xmath203, see figs. [fig : e2-e4] and [fig : b]. denote by @xmath268 the number of white vertices of @xmath267 which are not leaves. then the total weight of @xmath267 is equal to @xmath269 and the total number of edges is equal to @xmath270. clearly, @xmath271 for some polynomials @xmath29 and @xmath33 with @xmath272, @xmath273. furthermore, by, we must have : @xmath274 where @xmath275 [pro] the polynomials @xmath4 and @xmath5 may be represented as follows : @xmath276 where @xmath277 is the jacobi polynomial with parameters @xmath278 and @xmath279 [[proof.-4]] proof. + + + + + + since the polynomials @xmath29 and @xmath33 in, are defined in a unique way up to a multiplication by a scalar factor, it is enough to show that @xmath280 where @xmath218 and @xmath219 are given by, and @xmath104, by. represent the left side of as a product of two factors using the formula @xmath281 where @xmath282 it is easy to see that both @xmath283 and @xmath284 are @xmath285 near infinity. let us consider the difference @xmath286 clearly, @xmath287 furthermore, since @xmath288 the inequality @xmath289 holds. therefore, by lemma [lemma], we have : @xmath290 on the other hand, @xmath291 thus, @xmath292 as required. @xmath139 belyi functions for the series @xmath204 and @xmath205 with the parameters @xmath165 were first calculated in the thesis of nicolas magot in 1997 @xcite. a different proof, proposed by don zagier, was given in ch. 2 of @xcite. we used zagier s proof as a model for the above construction. let now @xmath267 be a weighted tree from the series @xmath202 or of its two particular cases @xmath201 and @xmath200, see figs. [fig : e1-e3] and [fig : b]. as above, denote by @xmath268 the number of white vertices of @xmath267 which are not leaves, so that the total weight of @xmath267 is @xmath293, and the total number of edges is @xmath294. now we must find polynomials @xmath4 and @xmath5 such that @xmath295 for some polynomials @xmath29 and @xmath33 with @xmath296 and @xmath297, and @xmath274 where @xmath298 the polynomials @xmath4 and @xmath5 may be represented as follows : @xmath299 where @xmath300 is the jacobi polynomial with the parameters @xmath301 and @xmath302 [[proof.-5]] proof. + + + + + + we must show that @xmath303 where @xmath304 and @xmath104 is defined by. equality is equivalent to the equality @xmath305 where @xmath306 on the other hand, since @xmath307 and @xmath308 it follows from lemma [lemma] that @xmath309 implying in the same way as in proposition [pro] that holds. @xmath139 the series @xmath150 is a particular case of the series @xmath310 of odd length corresponding to the case of @xmath268 equal to zero. in order to adjust the notation (which is slightly different for the series @xmath310 and @xmath150) we must set @xmath311 and change @xmath123 to @xmath126 and @xmath126 to @xmath312 in formulas. thus, @xmath313 where @xmath314 is the jacobi polynomial of degree @xmath315 with parameters @xmath316 while @xmath317 finally, it is clear that the series @xmath200 and @xmath203 (chains of odd and even length) are particular cases of the series @xmath202 and @xmath205, so that the davenport zannier pairs for @xmath200 and @xmath203 are obtained from those for @xmath202 and @xmath205 by setting @xmath318 the above results can be interpreted in terms of pad approximants for the function @xmath211. recall that if @xmath319 is a formal power series, then its _ pad approximant _ of order @xmath320 $] at zero is a rational function @xmath321, where @xmath322 is a polynomial of degree @xmath323 and @xmath324 is a polynomial of degree @xmath325, such that @xmath326 defined in this way, pad approximants do not necessarily exist. however, if an approximant of a given order exists, it is unique. linearizing the problem by requiring that @xmath327 we arrive to the notion of a _ pad form _ @xmath328 of order @xmath320 $]. being defined by linear equations, pad forms always exist (in general, does not imply since @xmath324 may vanish at zero), and the pad form of a given order is defined in a unique way up to a multiplication by a constant. keeping the notation of sect. [sec : e - even] we may now reformulate the condition for @xmath4 and @xmath5 to be a davenport zannier pair for the series @xmath310 of even length as follows (a similar result is also true for the series @xmath310 of odd length). [prop : pade - even] let polynomials @xmath29 and @xmath33 be like in formulas and. then the pair of their reciprocals @xmath329 is the pad form of order @xmath330 $] for the function @xmath211 with parameters @xmath331 [[proof.-6]] proof. + + + + + + since the pairs @xmath19 and @xmath332 are both defined up to a multiplication by a constant, it is enough to show that @xmath333 where @xmath334 by definition of pad forms we have : @xmath335 implying that @xmath336 (here we use formula again though now the factors involved are series by non - negative powers of @xmath74). finally, substituting @xmath337 in place of @xmath74 in and multiplying both sides by @xmath338 we obtain. [prop : pade - odd] let polynomials @xmath29 and @xmath33 be like in formulas and. then the pair of their reciprocals @xmath329 is the pad form of order @xmath339 $] for the function @xmath211 with parameters @xmath340 the proof is similar to the previous one, so we omit it. [rem : pade] from the computational point of view, a great advantage of pad approximants is due to the fact that the equations describing them are linear. this observation remains true even in the case like ours when the polynomials in question are known explicitly. one has to use some astute tricks in order to make maple work with jacobi polynomials whose parameters do not satisfy the condition @xmath209. at the same time, the computation of pad approximants is instantaneous. a vast literature is devoted to the study of pad approximants for some particular functions. this is the case, for example, for the exponential function. to our surprise, we did not find any research concerning pad approximants for the function @xmath211. by the way, our lemma [lemma] can also be reformulated as a result about pad forms for this function.
Series @xmath341 and @xmath342: trees of diameter 4
below we find dz - pairs for the series @xmath341 and @xmath342, see figs. [fig : f] and [fig : g], using their relations to differential equations. for the series @xmath341, which consists of ordinary trees, the corresponding formulas are particular cases of the formulas for _ shabat polynomials _ for trees of diameter four, first calculated by adrianov @xcite. since any tree from the series @xmath341 is ordinary, the degree of @xmath42 is zero, that is @xmath343 for some @xmath344 therefore, in order to describe the corresponding dz - pair it is enough to find @xmath4 and @xmath152. this is equivalent to the finding of the shabat polynomial corresponding to the tree. similarly, for trees from the series @xmath342 the degree of @xmath43 is one, and it is technically easier to provide explicit formulas for @xmath4 and @xmath43 rather than for @xmath4 and @xmath5. we start with the series @xmath342. the polynomial @xmath4 for the series @xmath342 takes the form @xmath345 where @xmath29 is a polynomial of degree @xmath346 whose roots are the black vertices (all of them are of degree @xmath104). notice that the number of these vertices does not coincide with the degree of the central vertex since we have one `` double '' edge. we choose the normalization of @xmath4, @xmath5 and @xmath42 in the following way : * @xmath347 where @xmath29 is monic, @xmath348 ; * the central vertex is placed at @xmath125, so that @xmath349 where @xmath33 is monic, @xmath350 ; the roots of @xmath33 are the white vertices distinct from zero ; * @xmath351 ; this means that the pole inside the only face of degree 1 is placed at @xmath127. thus, we get @xmath352 the polynomial @xmath29 satisfies the differential equation @xmath353 consequently, coefficients @xmath354 of @xmath355 may be found by the following backward recurrence : @xmath356 finally, @xmath357. [[proof.-7]] proof. + + + + + + taking the derivative of the both sides of equality we obtain the equality @xmath358 implying the equality @xmath359 substituting in the last equality the value of @xmath360 from, we obtain @xmath361-ca { \,=\,}x^{k-1}a\left(kb+xb^{\prime}\right)\]] and @xmath362.\]] we now observe that the degree of the left - hand side of the latter equality is @xmath346, while its right - hand side is _ proportional _ to @xmath363. therefore, the expression in the square brackets on the right is some constant @xmath31, and both parts are equal to @xmath364. the constant @xmath31 can be easily found as the leading coefficient of the left - hand side : it is equal to @xmath365. finally, we get the equality @xmath366 which implies. substituting @xmath355 in we obtain. finally, substituting @xmath125 in we obtain @xmath367. @xmath139 let us take @xmath368, so that @xmath369. then the corresponding polynomial looks as follows : @xmath370 where @xmath371 [rem : hyp - geom] polynomial @xmath29 also satisfies the hypergeometric differential equation @xmath372\frac{dy}{dx } - ab\cdot y { \,=\,}0.\end{aligned}\]] indeed, applying the differential operator @xmath373 to both parts of equality we obtain @xmath374^{\prime}+(1-k)\left[ma^{\prime}\cdot(x-1)-a\right]=0,\]] implying @xmath375a^{\prime } - \frac{(1-k)}{m}a { \,=\,}0.\]] therefore, @xmath29 is a solution of the differential equation @xmath376\frac{dy}{dx } + \frac{(1-k)}{m}\,y { \,=\,}0\]] which is a particular case of with @xmath377 for this series we may assume that @xmath378 here @xmath29 is monic and @xmath379 ; namely, @xmath29 is a polynomial whose roots are the black vertices of degree @xmath104. now, @xmath33 is a polynomial whose roots are the white vertices distinct from zero, @xmath380. the polynomials @xmath4 and @xmath5 must satisfy the condition @xmath381 where @xmath382 is a non - zero constant. the polynomial @xmath29 satisfies the differential equation @xmath383\,x^{k-1}.\end{aligned}\]] consequently, coefficients @xmath354 of @xmath355 may be found by the following backward recurrence : @xmath384 finally, the value of @xmath152 in is equal to @xmath385. [[proof.-8]] proof. + + + + + + as above, let us take the derivative of both sides of equation. then we get @xmath386 { \,=\,}x^{k-1}\left(kb+xb^{\prime}\right).\]] we observe that the polynomial @xmath363 is coprime with the factor @xmath387 in the left - hand side, and therefore it must be proportional to the factor @xmath388 which is itself a polynomial of degree @xmath346. therefore, both of them are equal to @xmath364 where the constant @xmath31 can be found as the leading coefficient of @xmath388 ; namely, it is equal to @xmath389. thus, holds. now, substituting @xmath355 in we obtain the recurrence, and substituting @xmath125 in we obtain the value of @xmath152. @xmath139 here, like in the case of the series @xmath342, the polynomial @xmath29 also satisfies the hypergeometric differential equation, and therefore it may be represented through a hypergeometric function. let us take @xmath368, so that @xmath369. then the corresponding polynomial looks as follows : @xmath390 where @xmath391 the above method may be applied to dz - pairs which do not necessary correspond to trees of diameter four or to unitrees. however, in general, it leads to _ differential relations _ between @xmath4 and @xmath5. let us clarify what we mean by considering the problem of the difference between cubes and squares of polynomials, which was at the origin of the whole activity concerning dz - pairs, see @xcite, @xcite. let @xmath392 @xmath393 and @xmath43 be polynomials such that @xmath394 and @xmath395 taking the derivative of both parts of we obtain @xmath396 multiplying now the last equality by @xmath29 and substituting @xmath35 from we obtain the equality @xmath397 implying in its turn the equality @xmath398 since the degree of the right - hand side is @xmath399 while @xmath46, the above equality implies that @xmath400 for some non - zero constant @xmath401. the last expression is a differential equation of the first order with respect to @xmath29 as well as with respect to @xmath33. unfortunately, both @xmath29 and @xmath33 are unknown. thus, it does not give us any immediate information about @xmath29 and @xmath33. still, algebraic equations for coefficients of @xmath29 and @xmath33 obtained from are (mostly) of degree 2 while the equations obtained from are (mostly) of degree 3. differentiating and writing the expression thus obtained as a differential equation with respect to @xmath29 we get : @xmath402 this differential equation is a particular case of the differential equation @xmath403 where @xmath404 is a polynomial of degree at most @xmath405 polynomial solutions of the last equation are called stieltjes polynomials. the polynomials @xmath404 for which has a polynomial solution are called van vleck polynomials. thus, @xmath33 is a van vleck polynomial, and @xmath29 is the corresponding stieltjes polynomial. writing now in the form @xmath406 we obtain that @xmath29 is a van vleck polynomial and @xmath33 is the corresponding stieltjes polynomial. the above observations show that the relations between dz - pairs and differential equations may be deeper than it seems at first glance and deserve further investigation.
Series @xmath407 and @xmath408: decomposable ordinary trees
in this section we consider series @xmath407 (fig. [fig : h]) and @xmath408 (fig. [fig : i]). in both cases the corresponding dz - pairs are obtained with the help of the operation of composition. notice that the trees in question are ordinary (the weights of all edges are equal to 1). as it was mentioned in definition [def : weighted], belyi functions for ordinary trees are called shabat polynomials. the trees of the series @xmath407 are compositions of trees from the series @xmath150 with the parameters @xmath165 and chains of length 2. the expressions of the shabat polynomials for the trees from the series @xmath150 in terms of jacobi polynomials are given in sect. [sec : c]. using the fact that @xmath165 we can also compute them directly. indeed, the trees in question have exactly two vertices of degree greater than 1. putting them into the points @xmath125 and @xmath127 and taking into account that the degree of the corresponding shabat polynomial @xmath409 is @xmath410, we conclude that the derivative of @xmath109 is proportional to @xmath411. therefore, the polynomial @xmath409 itself can be written as @xmath412 then we automatically have @xmath413, while in order to get @xmath414 we must take @xmath415 where @xmath416 is the euler beta function. then, taking the shabat polynomial for the chain with two edges and with two black vertices put to 0 and 1, which is equal to @xmath417 we obtain the following the polynomial @xmath4 for the tree @xmath407 is equal to @xmath418 where @xmath236 is as in and @xmath109 is as in and. the proof is obvious. below are given shabat polynomials @xmath419 for the trees of the series @xmath408. these trees are compositions of trees from the series @xmath150 with @xmath165 and @xmath420, and the stars with three edges. thus, @xmath421, where @xmath109 is a shabat polynomial corresponding to a tree from the series @xmath150, and @xmath236 is a shabat polynomial corresponding to the star with three edges. however, in order to achieve the rationality of the coefficients of @xmath4 we still must find an appropriate normalization of @xmath109. for this purpose, contrary to all traditions, let us put the vertices of degree @xmath48 of the tree from the series @xmath150 into the points @xmath422. then the derivative of the corresponding shabat polynomials @xmath409 must be equal to @xmath423 therefore, @xmath424+b\end{aligned}\]] for some @xmath425. substituting into @xmath409 the critical points @xmath422, we obtain the critical values @xmath426, where @xmath427 setting @xmath428 and choosing @xmath218 in such a way that @xmath429 we obtain a polynomial @xmath430 $] with two critical values @xmath431 taking now @xmath432 (we must take @xmath433 instead of @xmath434 in order to get the colors of the vertices which would correspond to fig. [fig : i]), we obtain the following the polynomial @xmath435 for the tree @xmath408 is equal to @xmath436 where @xmath236 is as in and @xmath109 is as in with @xmath218 and @xmath219 defined by conditions,,. once again, the proof is obvious.
Series @xmath30
this is the last infinite series of unitrees. the degree of this tree, or its total weight, is @xmath437. let us normalize the polynomial @xmath4 so that @xmath438 this means that the black vertex of degree 4 is put at @xmath439, while two black vertices of degree @xmath440 are put at the points @xmath441 for certain @xmath442, @xmath443. all the white vertices are of degree 2 ; therefore, the polynomial @xmath5 has the form @xmath444 for some polynomial @xmath29, @xmath445. further, condition gives us @xmath446 here @xmath440 is the `` overweight '' of the tree (that is, its total weight minus the number of edges of the topological tree). for the reciprocal polynomials this gives (see) @xmath447 here @xmath448 is the number of edges of the topological tree. the reciprocal polynomials @xmath449 and @xmath450 may be represented as follows : @xmath451 where @xmath131 is the initial segment of the series @xmath452 up to the degree @xmath453 : @xmath454 [[proof.-9]] proof. + + + + + + let @xmath455 where @xmath456 computing @xmath457 we get @xmath458.\end{aligned}\]] thus, for any value of the parameter @xmath218 we have @xmath459 and therefore, in order to obtain, we only have to show that for @xmath460 the constant term of @xmath461 is equal to zero, or, equivalently, the coefficient in front of @xmath462 in the series @xmath463 vanishes. let us write the second factor of the latter expression explicitly : @xmath464 notice that this series involves only even powers. multiplying it by @xmath465 we see that the coefficient in front of @xmath462 in @xmath452 is the sum of the coefficients in front of @xmath462 and @xmath466 in. therefore, we must ensure that @xmath467 collecting similar terms we get @xmath468 which gives @xmath460. @xmath139 let us take @xmath469. then we have : @xmath470 further, @xmath471 notice that the term with @xmath472 is missing. finally, @xmath473
Sporadic trees
as it was explained previously, in sect. [sec : computation], the verification of the results given below is trivial. therefore, we present nothing else but the polynomials themselves. @xmath474 @xmath475 @xmath477 @xmath479 this tree is symmetric, with the symmetry of order 3. therefore, @xmath4, @xmath5, @xmath43 are polynomials in @xmath480. @xmath482 this triple was found in beukers and stewart @xcite (only the polynomial @xmath4 is given in their paper, but it uniquely determines two other polynomials). @xmath483 once again, the answer is taken from @xcite, with a slight renormalization. @xmath484 this tree is the only sporadic tree from the adrianov s list of ordinary unitrees. correspondingly, @xmath4 is a shabat polynomial : the polynomial @xmath43 is a constant. note that the positions of certain black vertices are rational : @xmath485 @xmath486 the tree @xmath43 is the `` square '' of the tree @xmath87 : it is symmetric, with the symmetry of order 2, and one of its `` halves '' is equal to @xmath87. therefore, we may take the polynomials for the tree @xmath87 and insert @xmath487 instead of @xmath74. @xmath488 @xmath489 notice that the second factor in @xmath4, the one which is `` cubed '', does not contain the term with @xmath74 : this is not a misprint. the picture of this tree is given in example [ex : pair - t.bis], and the corresponding polynomials are given in example [ex : pair - t].
Trees defined over @xmath0 by virtue of galois + theory
recall that the passport of a (bicolored weighted plane) tree is a pair of partitions @xmath1, where @xmath2 is the degree (or the total weight) of the tree, @xmath16 represents the set of degrees of its black vertices, and @xmath17 represents the set of degrees of its white vertices. [def : comb - orbit] the set of weighted trees with the same passport is called _ combinatorial orbit_. unitrees represent, in fact, combinatorial orbits consisting of a unique tree. usually, a dz - pair corresponding to a tree is defined over a number field whose degree is equal to the size of the combinatorial orbit to which this tree belongs. this is why unitrees are always defined over @xmath0. there exist, however, other galois invariants which may split a combinatorial orbit into several distinct galois orbits. in this way we may obtain certain trees which are not unitrees but which are still defined over @xmath0. in @xcite we gave several such examples. here we present the corresponding dz - pairs. a bicolored map may be characterized by a pair of permutations acting on the set of its edges : one permutation represents the cyclic order (in the positive direction) of the edges around black vertices, the other one, the cyclic order around white vertices. for example, the map shown in fig. [fig : pgl27], is represented by the pair of permutations @xmath491 it turns out that the permutation group @xmath492 is equal to @xmath490. this group, which is called _ monodromy group _, is a galois invariant. since this tree is the only one in its combinatorial orbit whose monodromy group is @xmath490, it is defined over @xmath0. @xmath493 the combinatorial orbit to which this tree belongs, that is, the set of trees with the passport @xmath494, contains six trees. the five remaining trees constitute a single galois orbit ; the corresponding dz - pairs (or, we may say, the trees themselves) are defined over the splitting field of the polynomial @xmath495 the combinatorial orbit corresponding to the passport @xmath496, consists of five trees. one of them, shown in fig. [fig : pgl27-bis], has the monodromy group @xmath490. therefore, it is defined over @xmath0. its dz - pair is given below. @xmath497 one of the trees in this combinatorial orbit is symmetric (see fig. [fig : sym]) and is therefore also defined over @xmath0. the corresponding polynomials are @xmath498 the three remaining trees constitute a single galois orbit and are defined over the splitting field of the polynomial @xmath499 the _ duality _ for the bicolored maps is defined as follows : * a map and its dual share their white vertices ; * black vertices of each map correspond to the faces of the dual map ; * edges of the dual map connect the centers of the faces of the initial map to the white vertices which lie on the border of these faces. see details and examples in @xcite. a map is _ self - dual _ if it is isomorphic to its dual. self - duality is a galois invariant. the maps corresponding to weighted trees may well be self - dual. let @xmath500 be two positive integers, and @xmath501. we consider the trees with the black partition @xmath502 and the white partition @xmath503. the partition representing the face degrees is @xmath504. we notice that @xmath505 ; therefore, the corresponding combinatorial orbit _ may _ contain self - dual trees. it is easy to verify that this combinatorial orbit consists of @xmath506 trees, and that only one of them is self - dual, namely, the tree shown in fig. [fig : self - dual]. therefore, this tree is defined over @xmath0. put the white vertices at the points @xmath439 and @xmath127 so that @xmath507 (notice that both powers are odd). observe now that this polynomial is `` antipalindromic '' : if we write it as @xmath508 then @xmath509, @xmath510, this fact trivially follows from the equality @xmath511. because of this, the coefficient in front of the `` middle '' degree @xmath512 is zero. therefore, if we take the higher degrees from @xmath513 to @xmath105, what will remain is a polynomial of degree @xmath514. in other words, @xmath515 where @xmath516. setting now @xmath517 we see that @xmath4, @xmath5 is a dz - pair with required properties. notice that the polynomial @xmath518 is reciprocal to @xmath519. geometrically, this means that if @xmath520 are the positions of the black vertices of degree 1 (here @xmath521), then the centers of the faces of degree 1 are @xmath522. together with the fact that the position of the black vertex of degree @xmath105 is @xmath125 while the center of the face of degree @xmath105 is @xmath68, this shows that the map in question is indeed self - dual. let us take, for example, @xmath523, @xmath524. then @xmath525 where @xmath526 and @xmath527. the combinatorial orbit corresponding to the passport @xmath528, @xmath529 is shown in fig. [fig : zannier-5]. it consists of four trees (recall that the invisible white vertices are middle points of the edges), and is divided into three galois orbits. the tree @xmath218 is the only one which is symmetric with the symmetry of order 3. therefore, it is defined over @xmath0. the corresponding polynomials were computed by b.birch in 1965 @xcite. they look as follows (notice that they are polynomials in @xmath480) : @xmath530 the trees @xmath219 and @xmath152 are symmetric with the symmetry of order 2 with respect to an (invisible) white vertex. they are also mirror symmetric to each other ; therefore, the complex conjugation sends one of the trees to the other. thus, we may conclude that this couple of trees constitutes a separate galois orbit, and this orbit is defined over an imaginary quadratic field. the corresponding polynomials were computed in 2005 by shioda @xcite and, indeed, they are defined over the field @xmath531. we do not present these polynomials here. the tree @xmath532 does not have any particular combinatorial properties. (it is known that the mirror symmetry of a dessin is not a galois invariant.) but _ it remains alone _, that is, it constitutes a galois orbit containing a single element. therefore, it is defined over @xmath0. the corresponding polynomials were computed in 2000 by n.elkies @xcite. they look as follows : @xmath533 by the way, a naive approach mentioned in sect. [sec : computation], namely, taking polynomials @xmath29 and @xmath33 of degrees 10 and 15 respectively with indeterminate coefficients and equating to zero the coefficients of degrees from 7 to 30 of @xmath534, would, this time, lead us to a system of polynomial equations of degree @xmath535. it took 40 years (from 1965 to 2005) to compute all the four dz - pairs of this example, but the fact that there are exactly four non - equivalent solutions and that two of them are defined over @xmath0 while the other two are defined over an imaginary quadratic field, can be immediately seen from the picture without any computation.
Some sporadic examples of beukers and + stewart @xcite
all the polynomials in this section which correspond to the _ asymmetric _ trees are taken from the above - cited article @xcite. the normalization sometimes is changed. the goal of this section is to show the _ combinatorial reasons _ of appearance of these sporadic examples. the passport shows that we are treating here the problem of the minimum degree of the difference @xmath537 where @xmath538, @xmath539. the combinatorial orbit consists of two trees, see fig. [fig : split-7 - 3]. one of them is symmetric, the other one is not ; therefore, both are defined over @xmath0. the triple corresponding to the asymmetric tree is as follows : @xmath540 the triple corresponding to the symmetric tree may be computed as follows : 1. compute the polynomials corresponding to a branch of this three - branch tree, that is, to a tree of the series @xmath29 (see sect. [sec : a]) with the parameters @xmath541, @xmath542, @xmath543. 2. make the change of variables @xmath544 in order to put the white vertex of degree 1 to the point @xmath125 ; thus, the polynomial @xmath435, instead of being @xmath545, becomes @xmath546 ; it is convenient to change its sign and to get @xmath547. insert @xmath480 instead of @xmath74. by pure convenience we add to the above operations one more : instead of taking @xmath548 we take @xmath549. this permits us to avoid fractional coefficients. the resulting triple is @xmath550 the passport corresponds to the problem of the minimum degree of the difference @xmath552 where @xmath538, @xmath553. the combinatorial orbit consists of two trees, see fig. [fig : split-8 - 3]. one of them is symmetric, the other one is not ; therefore, both are defined over @xmath0. the triple corresponding to the asymmetric tree looks as follows : @xmath554 the triple corresponding to the symmetric tree may be computed as follows : 1. compute the polynomials corresponding to the series @xmath205 (see sect. [sec : e - even]) with @xmath555, @xmath556, @xmath557, @xmath558. 2. make the change of variables @xmath559 in order to move the (left) black vertex of degree 4 from @xmath238 to 0. insert @xmath487 instead of @xmath74. we omit the resulting polynomials. this time we deal with the problem @xmath561, @xmath538, @xmath562. the combinatorial orbit corresponding to this passport contains three trees, see fig. [fig : split-10 - 3]. these trees have three different symmetry types, hence all of them are defined over @xmath0. the polynomials for the asymmetric tree look as follows : @xmath563 the polynomials for the tree with the symmetry of order 2 is computed in the same way as in sect. [sec : split2]. the parameters of the tree of the type @xmath205 are @xmath564, @xmath542, @xmath557, @xmath565 ; then we must replace @xmath74 with @xmath566, and insert @xmath487 instead of @xmath74. the polynomials for the tree with the symmetry of order 3 is computed in the same way as in sect. [sec : split1]. the parameters of the tree of the type @xmath29 are @xmath541, @xmath542, @xmath469 ; then we must replace @xmath74 with @xmath567, and insert @xmath480 instead of @xmath74. we finish this section with an example which shows that the combinatorial methods, while being very powerful, are, however, not all - powerful. there are several trees with the passport @xmath568, and one of them, shown in fig. [fig : why], is defined over @xmath0 without any apparent reason. all known combinatorial and group - theoretic galois invariants fail to explain this phenomenon. all we can say is that the corresponding system has rational solutions `` by chance ''. the polynomials @xmath4 and @xmath5 for the tree of fig. [fig : why] are as follows : @xmath569 the polynomial @xmath43 here is of degree 32, and it is too cumbersome, so we do not write it explicitly.
Yet more examples
we have already seen two examples (see sect. [sec : split1] and [sec : split2]) of combinatorial orbits of size 2 which, instead of being defined over a quadratic field, split in two orbits defined over @xmath0 because the trees in question have different orders of symmetry. here we present an infinite series of such examples. the trees in question have the passport @xmath570 for @xmath571, see fig. [fig : sym - and - not]. belyi function for the symmetric tree looks as follows : @xmath572 belyi function for the asymmetric tree looks as follows : @xmath573 in both cases, the white vertex of degree 4 lies at @xmath125. the expressions for belyi functions give us the polynomials @xmath4 and @xmath43. in order to prove the correctness of the above expressions we need to verify two things : for both @xmath574 and @xmath575, we have (a) @xmath576 ; (b) first three derivatives of @xmath577 at @xmath125 vanish. let us return to the problem of the minimum degree of the difference @xmath534, the question from which this whole line of research started (see @xcite. we have seen that when @xmath45, @xmath46, we have @xmath578. for @xmath579, the computation becomes exceedingly difficult, and there is practically no hope to find solutions defined over @xmath0. however, if we are not so demanding and accept a solution with the degree of @xmath534 slightly greater than @xmath192, then sometimes we can find a needed solution. let us take a polynomial @xmath29 with one double root, so that @xmath35 would have one root of multiplicity 6 and all the other roots of multiplicity 3. the corresponding tree would have one vertex less and therefore one face more. the tree in fig. [fig : deg9-instead-8] corresponds to @xmath580. it is the `` cube '' of the tree @xmath109, see sect. [sec : tree - s]. therefore, all we have to do is to insert @xmath480 instead of @xmath74 in the formulas of section [sec : tree - s]. @xmath581 when all the roots of @xmath29 and @xmath33 are distinct, the polynomial @xmath43 has @xmath192 distinct roots. let us accept @xmath43 with a multiple root (thus, its degree will be greater that @xmath192). the tree in fig. [fig : deg - relaxed] gives such and example. it corresponds to @xmath368, and @xmath582. the polynomials for this tree look as follows : @xmath583 fedor pakovich was partially supported by isf grants no. 639/09 and 779/13. he is also grateful to the max plank institute for mathematics for the hospitality and support. alexander zvonkin was partially supported by the research grant graal anr-14-ce25 - 0014. 99 handbook of mathematical functions : with formulas, graphs, and mathematical tables. dover, 1972. on generalized chebyshev polynomials corresponding to planar trees of diameter 4, * 13 * (2007), no. 6, 1933 (in russian) ; translation in _ j. math (n. y.) vol. * 158 * (2009), no. 1, 1121 arithmetic theory of graphs on surfaces, ph.d. thesis, moscow state university, 1997, 116 pp. (in russian). neighboring powers, _ journal of number theory _, 2010, vol. * 130 *, 660679. on the difference @xmath584, _ det kongelige norske videnskabers selskabs forhandlinger _ (trondheim), 1965, vol. * 38 *, 6569. cycles comme produit de deux permutations de classes donnes, _ discrete math. _, 1982, vol. * 58 *, 129142. the diophantine equation @xmath585 and hall s conjecture. _ matematicheskie zametki _, 1982, vol. * 32 *, no. 3, 273275. diophantine equations @xmath586. _ matematicheskie zametki _, 1989, vol. * 46 *, no. 6, 3845. on @xmath587, _ det kongelige norske videnskabers selskabs forhandlinger _ (trondheim), 1965, vol. * 38 *, 8687. on hall s conjecture. _ acta arith. _, 2011, vol. * 147 *, no. 4, 397 - 402. realizability of branched coverings of surfaces, _ trans. soc. _, 1984, vol. * 282 *, no. 2, 773790. rational points near curves and small non - zero @xmath588 via lattice reduction. in : wieb bosma, ed., algorithmic number theory, lecture notes in computer science, vol. * 1838 *, springer - verlag, 2000, 3363. the diophantine equation @xmath585. in : computers in number theory, academic press, london new york, 1971, 173198. graphs on surfaces and their applications. springer - verlag, 2004. old and new conjectured diophantine inequalities. _, new ser., 1990, vol. * 23 *, no. 1, 3775. cartes planaires et fonctions de belyi : aspects algorithmiques et exprimenataux. thesis, universit bordeaux i, 1997, 144 pp. minimum degree of the difference of two polynomials over @xmath0, and weighted plane trees. _ selecta mathematica, new ser. _, 2014, vol. * 20 *, no. 4, 10031065. see also arxiv:1306.4141v1. elliptic surfaces and davenport stothers triples, _ comment. univ. st. pauli _, 2005, vol. * 54 *, no. 1, 4968. on computing belyi maps, arxiv:1311.2529v3 (may 2014) polynomial identities and hauptmoduln, _ quart. j. math. oxford _, ser. 2, 1981, vol. * 32 *, no. 127, 349370. orthogonal polynomials. colloquium publications, vol. * 23 *, 1939 (reedited in 1992). on davenport s bound for the degree of @xmath589 and riemann s existence theorem, _ acta arithmetica _, 1995, vol. * 71 *, no. 2, 107137. labri, umr 5800, universit de bordeaux, 33400 talence, france ; e - mail : zvonkin@labri.fr, and the chebyshev mathematical laboratory at the saint - petersburg state university, 29b, 14th line, vasilyevsky island, saint - petersburg 199178, russia. | in this paper we study pairs of polynomials with a given factorization pattern and such that the degree of their difference attains its minimum.
we call such pairs of polynomials _ davenport zannier pairs _, or dz - pairs for short.
the paper is devoted to the study of dz - pairs _ with rational coefficients_. in our earlier paper @xcite, in the framework of the _ theory of dessins denfants _, we established a correspondence between dz - pairs and _ weighted bicolored plane trees_. these are bicolored plane trees whose edges are endowed with positive integral weights. when such a tree is uniquely determined by the set of black and white degrees of its vertices, it is called _ unitree _, and the corresponding dz - pair is defined over @xmath0. in @xcite, we classified all unitrees. in this paper, we compute all the corresponding polynomials. in this way,
the present paper is a sequel of @xcite. in the final part of the paper we present some additional material concerning the galois theory of dz - pairs and weighted trees. | 1509.07973 |
Introduction
the discontinuous galerkin method @xcite nowadays is a well - established method for solving partial differential equations, especially for time - dependent problems. it has been thoroughly investigated by cockburn and shu as well as hesthaven and warburton, who summarized many of their findings in @xcite and @xcite, respectively. concerning maxwell s equations in time - domain, the dgm has been studied in particular in @xcite. the former two apply tetrahedral meshes, which provide flexibility for the generation of meshes also for complicated structures. the latter two make use of hexahedral meshes, which allow for a computationally more efficient implementation @xcite. in @xcite the authors state that the method can easily deal with meshes with hanging nodes since no inter - element continuity is required, which renders it particularly well suited for @xmath2-adaptivity. indeed, many works are concerned with @xmath0-, @xmath1- or @xmath2-adaptivity within the dg framework. the first published work of this kind is presumably @xcite, where the authors consider linear scalar hyperbolic conservation laws in two space dimensions. for a selection of other publications see @xcite and references therein. the latter three are concerned with the adaptive solution of maxwell s equations in the time - harmonic case. in this article, we are concerned with solving the maxwell equations for electromagnetic fields with arbitrary time dependence in a three - dimensional domain @xmath3. they read [eq : maxwell] @xmath4 with the spatial variable @xmath5 and the temporal variable @xmath6 subject to boundary conditions specified at the domain boundary @xmath7 and initial conditions specified at time @xmath8. the vectors of the electric field and flux density are denoted by @xmath9 and @xmath10 and the vectors of the magnetic field and flux density by @xmath11 and @xmath12. the electric current density is denoted by @xmath13. however, we assume the domain to be source free and free of conductive currents (@xmath14). furthermore, we assume heterogeneous, linear, isotropic, non - dispersive and time - independent materials in the constitutive relations @xmath15 the material parameters @xmath16 and @xmath17 are the magnetic permeability and dielectric permittivity. at the domain boundary, we apply either electric (@xmath18) or radiation boundary conditions (@xmath19), where @xmath20 denotes the local speed of light @xmath21. we also introduce the electromagnetic energy @xmath22 contained in a volume @xmath23 obtained by integrating the energy density @xmath24 as @xmath25 this paper focuses on a general formulation of the dgm on non - regular hexahedral meshes as well as the projection of solutions during mesh adaptation. the issues of optimality of the projections and stability of the adaptive algorithm are addressed. special emphasis is put on discussing the computational efficiency. to the best of our knowledge, this is the first publication dealing with dynamical @xmath2-meshes for the maxwell time - domain problem employing the dg method in three - dimensional space. as they are key aspects of adaptive and specifically @xmath2-adaptive methods, we will also address the issues of local error and smoothness estimation. this includes comments on the computational efficiency of the estimates. as estimators are not at the core of this article the discussion is, however, rather short.
Discontinuous galerkin discretization on non-regular hexahedral grids
we perform a tesselation of the domain of interest @xmath26 into @xmath27 hexahedra @xmath28 such that the tesselation @xmath29 is a polyhedral approximation of @xmath26. the tesselation is not required to be regular, however, it is assumed to be derivable from a regular root tesselation @xmath30 by means of element bisections. the number of element bisections along each cartesian coordinate, which is required to an obtain element @xmath31 of @xmath32 is referred to as the refinement levels @xmath33. as we allow for anisotropic bisecting the refinement levels of one element may differ. in case of isotropic refinement we simply use @xmath34. the intersection of two neighboring elements @xmath35 is called their interface, which we denote as @xmath36. as we consider non - regular grids, every face @xmath37 of a hexahedral element may be partitioned into several interfaces depending on the number of neighbors @xmath38 such that @xmath39. the face orientation is described by the outward pointing unitary normal @xmath40. the union of all faces is denoted as @xmath41, and the internal faces @xmath42 are denoted as @xmath43. finally, the volume, area and length measures of elements, interfaces, faces and edges are referred to as @xmath44, @xmath45, @xmath46 and @xmath47, where @xmath48 denotes any of the cartesian coordinates. every element of the tesselation @xmath32 is related to a master element @xmath49 ^ 3 $] through the mapping @xmath50 @xmath51 where @xmath52 denotes the element center. multiplying maxwell s equations ([eq : maxwell]) by a test function @xmath53, integrating over @xmath54 and performing integration by parts yields [eq : weakmaxwell] @xmath55 where the explicit dependencies of @xmath56 and @xmath6 have been omitted. equations ([eq : weakmaxwell]) constitute the generic weak formulation of the time - dependent maxwell s equations. in the following, we will replace the exact field solutions @xmath9 and @xmath11 by approximations using the discontinuous galerkin framework. the space and time continuous electromagnetic field quantities are approximated on @xmath32 as @xmath57 where @xmath58. the element - local approximation @xmath59 reads @xmath60 with the polynomial basis functions @xmath61 and the time - dependent vector of coefficients @xmath62^\text{t }, \]] representing the numerical degrees of freedom. the basis functions are defined with element - wise compact support, which is an essential property of dg methods @xmath63 we define the basis functions on the master element @xmath64 and obtain the element - specific basis through the mapping @xmath50 as @xmath65 we employ cartesian grids and tensor product basis functions of the form @xmath66 where @xmath1 is a multi - index obtained from all @xmath67. we denote by @xmath68 the local maximum approximation orders of element @xmath28. the finite element space (fes) @xmath69 spanned by the basis functions is given by the tensor product of the respective one - dimensional spaces @xmath70 the approximation may, thus, make use of different orders @xmath71 in each of the coordinate directions, where we drop the subscript if they are equal. the basis functions are legendre polynomials scaled such that @xcite @xmath72 in the following the dependence of the spatial and temporal variable is not written down explicitly. if now we were to substitute the exact electromagnetic field solution @xmath73 for its approximation the surface integral term of ([eq : weakmaxwell]) can not be evaluated straightforwardly at the internal faces @xmath43. this is due to the ambiguity of the dg approximation at any interface as a result of ([eq : approx]) and ([eq : basisfcts]). weak continuity at internal faces is obtained locally by introducing numerical interface fluxes as @xmath74 where @xmath75 is a unique interface value computed solely from @xmath76 and @xmath77, where @xmath78 is a neighboring element. common choices include centered and upwind fluxes. the centered interface value is given as @xmath79 computing the upwind value is more involved. it is obtained as the exact solution of maxwell s equations for piece - wise constant initial data after an infinitesimal time span, which is referred to as the riemannian problem @xcite. for the @xmath80-component of the electric and magnetic field at an interface with normal @xmath81 they read [eq : upw] @xmath82 with the intrinsic impedance and admittance @xmath83 other components are obtained by cycling the component indices and signs. note that centered fluxes preserve the hamiltonian structure of maxwell s equations while this property does not carry over to the semi - discrete equations when applying upwind fluxes due to the mixing of electric and magnetic quantities in ([eq : upw]). consequently, an energy conservation property @xcite can be obtained with the centered flux formulation only, determining the kind of time integration schemes to be used as well @xcite. our implementation includes both flux types. having resolved the ambiguity at interfaces, we insert the approximations ([eq : approx]) into the weak formulation ([eq : weakmaxwell]) and follow the galerkin procedure yielding the semi - discrete dg formulation [eq : dgmaxwell] @xmath84 @xmath85, @xmath86. the volume integrals are referred to as the mass and stiffness terms, the surface integrals represent face fluxes. note that no assumptions on the grid regularity have been made in the derivation. due to the strictly element - local support of the basis and test functions, the dgm is highly suited for the application on non - regular grids. the actual difference of the refinement levels @xmath87 and @xmath88 of neighboring elements, i.e., the level of hanging nodes, plays a minor role as shown in the following. inspecting equations ([eq : dgmaxwell]) it is seen that the mass and stiffness terms are not affected by the grid regularity as they are strictly local to the element @xmath28. the flux term, however, involves neighboring elements as well. decomposing the surface integral into the six contributing face integrals @xmath89 and considering centered fluxes for brevity each of these can be expressed as @xmath90.\]] accounting for the kind of non - regular grids described above, i.e. grids obtained from a regular root tesselation, requires no more than summing up the contributions of all neighboring elements to the total flux. this is independent of the hanging node levels as well as the actual number of neighboring elements. inserting the approximation ([eq : elocal]) into ([eq : faceinterfaces]) yields @xmath91.\]] again, the first integral term does not depend on the grid regularity. assuming @xmath92 to be aligned with the @xmath93-coordinate and to point towards positive direction it amounts to @xmath94 due to the basis function scaling ([eq : orthogonalbasis]). the second integral term can be expressed as @xmath95 in this case the orthogonality property of the basis functions is lost due to non - identical supports of @xmath96 and @xmath97. we gather the terms ([eq : faceinterfaceregular]) and ([eq : faceinterfaceirregular]) in the interior and exterior flux matrices @xmath98 and @xmath99. following to the usual notation the sign indicates the evaluation from the interior and exterior side of the interface. any non - regularity of the grid is now concealed within @xmath100, which reduces to the standard form on regular grids. for high level hanging nodes the number of integrals to compute quickly becomes large, imposing a heavy computational burden if integration is performed at run time. however, as the integrals @xmath101 in ([eq : faceinterfaceirregular]) do not include the actual approximation but basis functions only, they can be precomputed analytically (making use of the master basis functions) and stored in tabulated form in the code. this has to be done for all combinations of @xmath102 and @xmath103 as well as for each possible edge overlap according to the respective difference in the refinement levels @xmath104 (cf. [fig : nonregularinterface]). the number of possible overlaps grows as @xmath105. we tabulated the integrals up to @xmath106 and for basis functions up to order six, yielding 247 matrices @xmath107 of size @xmath108. in the isotropic refinement case @xmath106 corresponds to one element interfacing with @xmath109 neighbors. in the case of even larger differences in the refinement levels of neighboring elements, which are unlikely to occur a numerical integration is invoked at run time. if the neighboring element has a smaller instead of higher refinement level the respective transposed matrix @xmath110 is applied. for upwind fluxes, the interior and exterior flux matrices do not change, however, they are applied to both, the electric and the magnetic field due to ([eq : upw]). -axis for a better visualization. the left hand root element has been refined several times, the right hand element is at root level. the interface i connects an element of refinement levels @xmath111 with the root level element. the tick marks indicate possible locations for the imprint of elements of these refinement levels. the actual imprint on the root element face fills the first and sixth slab along the @xmath80- and @xmath112-axis, respectively. the interface ii fills the respective second slab along the @xmath80-axis.,scaledwidth=60.0%] in order to further enhance computational performance, all combinations of @xmath113 and the integrals @xmath114 arising form the stiffness terms of ([eq : dgmaxwell]) are evaluated and tabulated as well. precomputing the interface integrals maintains the high computational efficiency of the dg methods also for non - regular grids. using matrix notation, the semi - discrete dg maxwell equations ([eq : dgmaxwell]) read @xmath115 where * s * and * z * denotes the stiffness and impedance matrix. the matrix operator on the right hand side of ([eq : maxwellsemimatrix]) represents a weak dg curl operator. choosing @xmath116 as either zero or one yields centered or upwind fluxes, respectively. by applying centered fluxes the hamiltonian structure of maxwell s equations in continuum is preserved, whereas upwind fluxes lead to a mixed form. symplectic explicit time integration can be applied in the former case but not in the latter one @xcite. for examples of symplectic time integration for maxwell s equations in the dg framework see, e.g., @xcite. in @xcite upwind fluxes and runge - kutta schemes are applied for the time integration, where the latter one is concerned with maxwell s equations.
Local refinement techniques
the adaptation techniques presented in the following are based on projections between the finite element spaces introduced in ([eq : dgtensorspace]). the projection operators have been introduced in @xcite, however, they are included for completeness. also, we address the issue of stability in depth and amended this section with examples. the approximation @xmath117 to a given function @xmath118 in the fes @xmath69 is obtained by performing an orthogonal projection. the projection is carried out in an element - wise manner, by means of the projection operator @xmath119 given by @xmath120 where @xmath121 denotes the inner product @xmath122 on the element @xmath28 with the associated 2-norm @xmath123. when applied successively to all elements and all components of given initial conditions of the electric field, @xmath124, and the magnetic field, @xmath125, the respective dg approximations @xmath126 and @xmath127 are obtained. these approximations are optimal in the sense that the projection errors @xmath128 are orthogonal to the space of basis functions @xmath69 @xmath129,\, \varphi_i^p = \hat{\varphi}^p \circ g_i^{-1},\, \varphi^p \in \mathcal{v}^p\]] as stated above @xmath0-refinement is achieved by means of element bisections along the coordinate directions, where we allow for anisotropic refinements. the refined elements are referred to as the left and right hand side element @xmath130 and @xmath131 with basis functions denoted as @xmath132 and @xmath133 spanning the spaces @xmath134 and @xmath135 in a full analogy to @xmath69 defined in ([eq : dgtensorspace]). the approximation orders @xmath136 and @xmath137 in each child element do not have to be identical, neither are they required to be equal to the respective order @xmath71 of the parent element. the direct sum of the spaces @xmath138 and @xmath139 is denoted by @xmath140 @xmath141 in the following, the projection can be applied in order to project an approximation given in an element @xmath28 to the fes associated with an @xmath0-refined or @xmath0-reduced element. for @xmath0-refinement this yields @xmath142 due to the tensor product character of the basis, this can be expressed as @xmath143 for the left and right child, respectively. if refinement is carried out along one coordinate only, e.g. @xmath80, this further simplifies to @xmath144 where @xmath145 denotes the kronecker delta. note that above we loop over @xmath146, whereas in ([eq : projsolutionrefine2]) the loop parameter is @xmath1. as, moreover, @xmath147 vanishes for any @xmath148, we can limit the above loop to the range @xmath149 $], which reduces the number of addends to the minimum possible. for the merging of elements, the approximation within the parent element, @xmath28, is considered to be given piece - wise within its child elements. the projection reads @xmath150 where the simplifications ([eq : projsolutionrefine2]) and ([eq : projsolutionrefine3]) apply. for the case of @xmath1-enrichments, the local fes are amended with the @xmath151 order basis functions @xmath152 where any (non - zero) number of the local maximum approximation orders @xmath71 may be increased. also, an enrichment by more than one higher order basis function is possible. formally, we perform the orthogonal projection ([eq : projectionoperator]), however, due to the orthogonality property of the basis functions the coefficients @xmath153 remain unaltered under a projection from @xmath154 to @xmath155. practically, we simply extend the local vectors of coefficients @xmath156 with the new coefficients @xmath157, which are initialized to zero. conversely, for the case of a @xmath1-reduction, the local fes is reduced by discarding the @xmath71-order basis functions @xmath158 again, by virtue of the orthogonality, we find that the coefficients @xmath159 are deleted from the local vectors of coefficients while the coefficients @xmath160 remain unaltered. we denote by @xmath161 the projection of the global approximation @xmath162 from the current discretization to another one obtained by local @xmath0- and @xmath1-adaptations. an approximation @xmath163 with coefficients according to is optimal in the sense of ([eq : dgapproxerror]). the approximations within refined and merged elements with coefficients obtained through the orthogonal projections and are, hence, optimal in the same sense. if @xmath164 and @xmath165 holds true for all @xmath48, the fes @xmath166 is a subspace of @xmath140 (cf. ([eq : legendrespaceunion])) and every function of @xmath166 is representable in @xmath140 but not vice versa. in this case, a given approximation is exactly represented within an element under @xmath0-refinement but not under @xmath0-reduction. see fig. [fig : adaplegendreremarks] for an example. [cc](a) [cc](b) [cc]@xmath167 [cc]@xmath168 [cc]@xmath169 [cc]@xmath170 [cc]@xmath171 [lb]@xmath80 [cb]@xmath172 and @xmath173, the approximations of the parent and child elements agree point - wise. the projection to a merged element shown in (b) can, in general, not be exact due to the discontinuity., title="fig:",scaledwidth=95.0%] since the projections ([eq : projsolutionrefine]) for performing @xmath0-refinement are independent of the actual approximation, we also tabulated the projection operators @xmath174 and @xmath175 (expressed in master basis functions) yielding the matrix operators @xmath176 and @xmath177, where the superscript denotes that the refinement level @xmath178 is increased. accordingly, we make use of the matrix operators @xmath179 and @xmath180 for evaluating the projections of ([eq : projsolutionreduce]) in the case of element merging. the matrix operators are related as @xmath181 this allows for the computation of the approximations within adapted elements by means of efficient matrix - vector multiplications. as all projection matrices are triangular the evaluation can be carried out as an in - place operation requiring no allocation of temporary memory. the global approximation associated with an adapted grid is computed as @xmath182. it can be considered as initial conditions applied on the new discretization obtained by performing the refinement operations. assuming stability of the time stepping scheme (cf. @xcite), it is sufficient to show that the application of the projection operators at some time @xmath183 does not increase the electromagnetic energy associated with the approximate dg solution, i.e., @xmath184 in this case it follows @xmath185 and, thus, stability of the adaptive scheme. following ([eq : energydensity]) the energy associated with element @xmath28 is given as @xmath186 as a consequence of ([eq : projsolutionrefine2]), it is sufficient to show that the energy ([eq : dgenergy]) is non - increasing during any adaptation involving one coordinate only. [[sec : h - refinement-1]] @xmath0-refinement + + + + + + + + + + + + + + + + + + + + + + + + + + for the following discussion of stability it is assumed that refinement is carried out along the @xmath80-coordinate. also we assume the maximum approximation orders @xmath187 and @xmath188 to be identical. it is clarified later, that this does not pose a restriction to the general validity of the results. in the case of @xmath0-refinement, the operators @xmath189 and @xmath190 project from the space @xmath166 to the larger space @xmath140. following the argument of paragraph [sec : optimality] on optimality, any function defined in the space @xmath166 is exactly represented in @xmath140. the conservation of the discrete energy is a direct consequence as the approximation in the parent and child elements are point - wise identical. we find the following relation for the 2-norms of the respective local vectors of coefficients @xmath191 the exemplary parent element approximation plotted in fig. [fig : adaplegendreremarks]a has a maximum order of @xmath192 with all coefficients equal to one. the coefficients of the child element approximations and the square values of their 2-norms are given in tab. [tab : adapdgcoeffsr]. if the vector @xmath193 is considered to be either the vector of coefficients of the electric field @xmath194 or the magnetic field @xmath195 the result agrees with ([eq : dgenergy1drefine])..parent and child element coefficients of the function plotted in fig. [fig : adaplegendreremarks]a [cols="^,^,^,^,^,^,^,^,>",options="header ",] we chose a maximum @xmath0-refinement level of two and the local element order to vary in between zero and four. the local energy density, introduced in ([eq : energydensity]), serves as the criterion for controlling the adaptation procedure. denoting by @xmath196 the average energy density of element @xmath31 and by @xmath197 the normalized energy density with @xmath198, we assigned the local refinement levels according to @xmath199 and polynomial orders as @xmath200 with @xmath201. the initial discretization consisted of @xmath202 elements. during the simulation the number of elements varies and grew strongly after scattering from the reflector took place, when it reached close to 800,000 elements corresponding to slightly more than 55 million dof. for comparison, we note that employing the finest mesh resolution globally as well as fourth order approximations uniformly would lead to approximately 7.5 billion (@xmath203) dof. this corresponds to a factor of approximately 130 in terms of memory savings. we emphasize that the simulations were carried out on a single machine. the implementation takes full advantage of multi - core capabilities through openmp parallelization. the numerous run - time memory allocations and deallocations are handled through a specialized memory management library based on memory blocking, which we implemented for supporting the main code @xcite. figure [fig : eygrids] depicts cut - views of the @xmath112-component of the electric field and the respective @xmath2-mesh at three instances in time. note that the scaling of the electric field differs for every time instance, which is necessary to allow for a visual inspection. the enlargement shows details of the computational grid. all elements are of hexahedral kind, however, we make use of the common tensor product visualization technique (cf. @xcite) using embedded tetrahedra for displaying the three tensor product orders (out of which only @xmath204 and @xmath205 are visible in the depicted @xmath206-plane). as we employed isotropic @xmath0- as well as @xmath1-refinement all tetrahedra associated with one element share the same color. figure [fig : signals] shows plots of the outgoing and reflected waveform recorded along the waveguide center. the blue dashed line was obtained with the commercial cst microwave studio software @xcite on a very fine mesh and serves as a cross comparison result. -component of the electric field (top panel) and the computational grid (bottom panel) at three instances in time. the enlargement shows details of the grid. we employ hexahedral elements for the computation but make use of embedded tetrahedra for displaying the tensor product orders in the grid view. as isotropic @xmath1-refinement was employed in this examples all tetrahedra associated with one element share a common color. note that different scalings are used for the time instances in the top panel.,scaledwidth=100.0%]
Conclusions
we presented a discontinuous galerkin formulation for non - regular hexahedral meshes and showed that hanging nodes of high level can easily be included into the framework. in fact, any non - regularity of the grid can be included in a single term reflecting the contribution of neighboring elements to the local interface flux. we demonstrated that the method can be implemented such that it maintains its computational efficiency also on non - regular and locally refined meshes as long as the mesh is derived from a regular root tesselation by means of element bisections. this is achieved by extensive tabulations of flux and projection matrices, which are obtained through (analytical) precomputations of integral terms. we also presented local refinement techniques for @xmath0- and @xmath1-refinements, which are based on projections between finite element spaces. these projections were shown to guarantee minimal projection errors in the @xmath207-sense and to lead to an overall stable time - domain scheme. local error and smoothness estimates have been addressed, both of them relate to the size of the interface jumps of the dg solution. we considered the simulation of a smooth and a non - smooth waveform in a one - dimensional domain for validating the error and smoothness estimates. as an application example in three - dimensional space the backscattering of a broadband waveform from a radar reflector was considered. in this example the total wave propagation distance corresponds to approximately sixty wavelengths involving thousand of local mesh adaptations. as the implementation of the derived error and smoothness estimates for three - dimensional problems is subject of ongoing work, we chose to drive the grid adaptation using the energy density as refinement indicator. crosschecking with a result obtained using a commercial software package showed good agreement. l. fezoui, s. lanteri, s. lohrengel, s. piperno, convergence and stability of a discontinuous galerkin time - domain method for the 3d heterogeneous maxwell equations on unstructured meshes, esaim - math model num 39 (6) (2005) 11491176. d. wirasaet, s. tanaka, e. j. kubatko, j. j. westerink, c. dawson, a performance comparison of nodal discontinuous galerkin methods on triangles and quadrilaterals, int j numer meth fluids 64 (10 - 12) (2010) 13361362. l. krivodonova, j. xin, j. remacle, n. chevaugeon, j. flaherty, shock detection and limiting with discontinuous galerkin methods for hyperbolic conservation laws, appl numer math 48 (3 - 4) (2004) 323338. smove, a program for the adaptive simulation of electromagnetic fields and arbitrarily shaped charged particle bunches using moving meshes, technical documentation : _ - school - ce.de / files2/schnepp / smove/_. | a framework for performing dynamic mesh adaptation with the discontinuous galerkin method (dgm) is presented.
adaptations include modifications of the local mesh step size (@xmath0-adaptation) and the local degree of the approximating polynomials (@xmath1-adaptation) as well as their combination.
the computation of the approximation within locally adapted elements is based on projections between finite element spaces (fes), which are shown to preserve an upper limit of the electromagnetic energy.
the formulation supports high level hanging nodes and applies precomputation of surface integrals for increasing computational efficiency.
error and smoothness estimates based on interface jumps are presented and applied to the fully @xmath2-adaptive simulation of two examples in one - dimensional space.
a full wave simulation of electromagnetic scattering form a radar reflector demonstrates the applicability to large scale problems in three - dimensional space.
discontinuous galerkin method, dynamic mesh adaptation, @xmath2-adaptation, maxwell time - domain problem, large scale simulations 65m60, 78a25 | 1109.0856 |
Introduction
incidence of energetic ion - beam causes the displacement and even removal of atoms near the surface. these whole processes are conventionally referred to as ion - beam - sputtering (ibs). simultaneously, the healing kinetics proceeds via mass transport to minimize the surface free energy of the modified surface. those two competing processes often produce patterns of nano - dots / holes or ripples depending on the incidence angle of the ion - beam. ibs is one of the most versatile tools to fabricate nano patterns in various sizes and shapes by controlling physical variables, and applicable to a wide range of materials from metals, insulators to organic materials. ibs has, thus, drawn attention as a representative method for triggering physical self - assembly.@xcite continuum models have mostly elucidated the pattern formation by ibs. in their seminal work, bradley and harper (bh)@xcite proposed a model of the pattern evolution ; they took the erosion into account according to sigmund picture@xcite in the linear approximation, while the diffusion of adatoms according to mullin s model.@xcite according to sigmund, the sputter yield at a position on the surface is proportional to the energy deposited at that position by an incident ion. the energy depends on gaussian ellipsoidal function of the distance from a terminal position of an incident ion to the position at the surface.@xcite according to mullin s picture, the adatom current is proportional to the gradient of the surface free energy which is in proportion to the curvature at the position of interest.@xcite bh model and its non - linear extensions@xcite that incorporate surface - confined mass flow, redeposition and surface damping well reproduce, although qualitatively yet, many features of the pattern formation by ibs. those models, though, tacitly assume amorphous surface. refined models have also been proposed taking the crystallinity@xcite of the substrate and anisotropy@xcite of the surface and sputter geometry combined into account. polycrystals consist of the grains whose mean size can be comparable with characteristic size of the structures formed by ibs.@xcite polycrystals offer unique environment for the pattern formation by ibs contrasted with the homogeneous substrates such as the amorphous and the single crystals. a challenging question is, then, whether and in the case how the grained structure of the polycrystalline substrate affects the pattern formation and its temporal evolution. recently, toma _ et al._@xcite have studied the ripple evolution of polycrystalline au films by ibs. with continued sputtering, the initially rough surface leveled off, and the pattern evolved as if it did on single crystalline au surface in regards to the surface width and ripple wavelength. this observation leads them to the conclusion that the polycrystallinity did not influence the pattern evolution. on the other hand, kere _ et al._@xcite reported that the variation of the sputter yield depending on the orientation of the grains induced the surface instability during ibs, and well reproduced the morphological evolution of polycrystalline ni films without invoking the instability owing to the curvature dependent erosion@xcite. the two conclusions on the effects of the polycrystallinity contradict to each other. highly oriented pyrolytic graphite (hopg) is polycrystalline and composed of the large grains, compared with the metallic films such as au@xcite and ni@xcite. hopg would, thus, offer the opportunity to study the effects of the grained structure on the pattern formation in the limit opposite to that of the metallic films. several groups have previously worked on patterning hopg by ibs.@xcite their studies are, however, little motivated by the interests in the effects of polycrystallinity on pattern formation. in this work, we study the effects of the grained structure on the pattern formation, employing two different kinds of graphites, hopg and natural graphite (ng). the mean grain size of ng is distinctly larger than that of hopg. this controlled experiment clearly tells that the grain boundaries play a critical role in determining the mean uninterrupted length of the ripples along their ridges or the coherence length @xmath0 and the surface width @xmath1, while they little influence the ripple wavelength. this highly anisotropic effects on the ripple evolution are attributed to the intricate roles of the grain boundaries in the temporal evolution of the primordial islands to the ripples during the pattern formation.
Experiments
the ion - beam - sputtering of both hopg (zya grade, spi) and ng (donated from union carbide) samples were performed in a high vacuum chamber whose base pressure was 5@xmath2 10@xmath3 torr. the ion - irradiated surface is characterized _ ex situ _ by atomic force microscopes (afm) in both the contact (afm, psi, autoprobe cp) and the noncontact modes (xei-100, park systems). sputtering is performed by irradiation of ar@xmath4 ion - beam with its beam diameter @xmath5 mm, incident ion energy @xmath6, 2 kev, at a polar angle of incidence @xmath7, 78@xmath8 from the global surface normal. the partial pressure of ar@xmath4, @xmath9 and the ion flux @xmath10 are 1.2 @xmath2 10@xmath11 torr and 0.3 ions nm@xmath12 s@xmath13, respectively. the sample temperature is kept around room temperature by limiting each sputter period to 1 minute with interval for 10 minutes, unless mentioned otherwise. the raman spectra of the samples were obtained by using a micro - raman spectroscopy system.@xcite the 514.5 nm (2.41 ev) line of an ar ion laser was used as the excitation source and the laser power was kept below 1mw to avoid unintentional heating. the laser beam was focused - spot size @xmath14m - onto the graphite sample by a x 50 microscope objective lens (0.8 n.a.), and the scattered light was collected and collimated by the same objective. the collected raman scattered light was dispersed by a jobin - yvon triax 550 spectrometer and detected by a liquid - nitrogen - cooled charge - coupled - device detector. the spectral resolution was about 1 cm@xmath13. the grazing incidence x - ray diffraction (gid) of the samples was performed with 20 kev photons (@xmath16 = 0.62 @xmath17) in the 5a beam line of pohang light source in korea. the incident angle of the x - ray was kept to be 0.1@xmath8 from the global sample plane to reduce both the beam penetration depth and bulk diffuse scattering.
Results
in fig. 1(a), the raman spectrum of hopg before ibs shows three bands, labeled as g, g * and 2d, typical of @xmath18 hopg.@xcite after an extended sputtering with an ion fluence @xmath19, flux times total sputter time, @xmath19 = 5625 ions nm@xmath12, its surface is patterned by ripples as shown in the inset of fig. 1. two new bands d and d develop as shown in fig. 1(b), while both g and 2d bands notably weaken as previously reported.@xcite both d and d bands originate from defective carbon atoms.@xcite the observed spectral changes tell that sputtering produces defective carbons at the cost of the @xmath18 carbons. the sample, however, largely remains crystalline ; the spectral shape of the _ amorphous _ carbon shows broad adjoined peaks of d and g bands,@xcite while the present spectrum in fig. 1 shows a well - defined d band with still intense and sharp g band. since the raman spectrum could also sample the @xmath18 layers beneath the damaged surface layers, we exfoliated the surface layers of the hopg after transferring it to silicon oxide substrate. 1(c) is a raman spectrum of the exfoliated surface layers that should be thinner than 10 layers, because the intensity of the 2d band is now comparable with that of the g band.@xcite in fig. 1(c), we still find a sharp g band with negligible d band, confirming the crystallinity of the sputtered surface. et al._@xcite also observed the development of both d and d bands upon ibs by ar@xmath4. their intensities monotonically increase with the increase of @xmath6 (@xmath20 10 kev) for the same @xmath19, indicating that the ion - beam creates defects in the deeper region and/or more effectively for the larger @xmath6. for the present sputtering condition, @xmath6 is 2 kev, small to critically damage the surface region, and leaves the sample in largely crystalline state. 2(a) shows omega(@xmath21)-rocking profiles of (0,0,6) peak for both @xmath22 hopg and ng. the profile from the hopg reveals a broad peak, indicating wide angular distribution of the grains. in contrast, ng shows sharp peaks that originate from large crystalline grains, indicating single crystal - like character of ng. (see also afm images in figs. 3(a) and (b).) c|c|c|c|p22mm|p22 mm & & & & & & & & lateral & vertical hopg & 2.4600 & 6.7119 & @xmath23 & @xmath24 & @xmath25 ng & 2.4605 & 6.7082 & @xmath26 & n / a & @xmath27 table 1 summarizes the lateral and vertical correlation lengths of both the @xmath22 hopg and ng. they are obtained by williamson - hall (wh) plots@xcite(plots not shown) of both peak width and peak rocking width versus the position of (0,0,l) peaks. the lateral correlation length of hopg is @xmath28 350 nm, comparable with the grain size observed in fig. 3(a), a typical afm image of a @xmath18 hopg. the mean grain size of hopg is still quite large compared with those of metallic films by an order of magnitude.@xcite the lateral correlation length of ng can not be determined by wh plots, because the peak width becomes smaller than the resolution limit of the detector as l of (0,0,l) peaks becomes small. this result indicates that the lateral correlation length of ng should be much larger than that of hopg, as also suggested by the afm images of both hopg and ng respectively in figs. 3(a) and (b). each grain of ng should behave as single crystalline graphite. 2 (b) and (c) show reciprocal space maps (rsms) around (101) peak of a @xmath22 hopg with @xmath19 = 5625 ions nm@xmath12, for q@xmath29 (reciprocal vector) (b) parallel to the ridge of the ripple and (c) perpendicular to the ridge. the blue dashed lines in both maps are equi - q plots, originating from the angular (rotational) spread of the grain orientations analogous to the powder diffraction pattern. note that a feature indicated by the white line is observed in the map perpendicular to the ripple, fig. 2(c), but not for the map along the ripple. this is attributed to the formation of high index facets along the side wall of the ripples with the facet normal, @xmath30 from the global surface normal. such a facet formation points to the crystalline surface layers, consistently with the conclusion from the raman spectroscopy in fig.1. figures 3(a) and (b) show ripple patterns on (a) hopg(0001) and (b) ng(0001), formed after extended ibs (@xmath31 ions nm@xmath12) under the same sputter condition. the mean uninterrupted ripple lengths or the coherence lengths @xmath0s show notable difference between hopg and ng, 0.41 @xmath32 0.06 @xmath33 m for hopg in fig. 3(a), much shorter than that for ng, @xmath34 @xmath33 m (not properly determined by the limited image size) in fig. the error limit is set from the standard deviation of the measured @xmath0s that are obtained by manually measuring the uninterrupted lengths of the ripples along their ridges from the images. in addition, the surface width @xmath1 of the patterned hopg in fig. 3(a) is @xmath3513.46 nm, notably larger than that of ng in fig. 3(b), @xmath35 9.6 nm for the same sputter condition. that difference in @xmath1 is much larger than the difference between the initial surface widths of hopg (@xmath36 nm) and ng (@xmath37 nm), and should have resulted from ibs. since the conspicuous difference between the hopg and ng lies in their mean grain sizes, the grained structure of the graphite should have affected both @xmath0 and @xmath1. in contrast, @xmath16s are almost identical, @xmath38 nm and @xmath39 nm, respectively for hopg and ng. (taking line profiles from each image, we obtain the mean ripple wavelength @xmath16.) the error limit is set from the standard deviation of the ripple wavelengths taken from the line profiles. @xmath16s are smaller than the grain size, @xmath40 350 nm for hopg. still 2 out of the 7 ripples of hopg meet the grain boundary for hopg, while the portion of the ripples neighboring the grain boundaries should be much smaller for ng due to much larger grain size than that for hopg. the present observation that @xmath16 is independent of the grain size indicates that the grain boundaries little affect coarsening of the ripple or that the grain boundaries parallel to the ridge direction do not hinder the mass transport across them. similar conclusion is also drawn for polycrystalline metal films.@xcite a common feature of the ripples on both hopg and ng is that the pronounced protuberances or the buds terminate ripples on the side facing the ion - beam. (also see the inset of fig. 1 and the following figures.) the other carbon allotropes such as diamond and tetrahedral amorphous carbon do not show such a bud structures upon ibs.@xcite polymer(pdms) show a labyrinthine pattern under the same sputter condition. 4(a)) layered structure may be a requirement for the formation of such protuberances. muscovite mica, however, shows ripple pattern with no such protuberances under the same sputtering condition. 4(b)) mica is another layered material, but the interlayer bonding is stronger and constituent atoms are heaver than for graphite. mica should then be less modified than graphite by the incidence of the same ion - beam. since the pronounced protuberance needs high sputter yield producing large mass transport, we tentatively conclude that the bud termination of the ripples requires layered structure with weak interlayer bonding and/or light constituents of each layer.
Discussion
the growth kinetics during the initial pattern formation on hopg offers insights on how the grain boundaries influence the growth of the ripples in such a highly anisotropic fashion as observed in fig. 5(a) shows an image at an incipient stage of pattern formation on hopg with @xmath41 ions nm@xmath12. the surface is covered by the oval - shaped islands elongated along the ion - beam direction. some of the islands have already developed tails or the incipient ripples along the ion - beam direction. the islands look distributed randomly, indicating little preference in the island formation between the grain boundaries and the terraces, which is also the case for ng. (figure not shown) as sputtering proceeds with @xmath42 ions nm@xmath12, we observe that the islands have followed two different paths in their temporal evolution. 5(b)) one kind of islands have grown to pronounced protuberances in the shape of buds as observed in fig. note that the buds always form in the side of the ripple facing the ion - beam. their tails have also grown to form segmented ripples that are terminated by the adjacent buds along their ridges. each bud, thus, works as the birth place of the segmented ripple and simultaneously as the terminus of the adjacent ripple growing along the ion - beam direction. ripple growth via elongation of the tails has also been observed for the polycrystalline ni films.@xcite the other kind of islands become affiliated with the neighboring ripples, and barely idetifiable as nodes along their ridges.(fig. 5(b)) only the islands in the form of the buds remain distinguished, and the density of the islands looks significantly diminished as compared with that in fig. note that the growth of the ripple occurs via the linkage - which can occur multiply - of the ripples along the ion - beam direction, for which the nodes work as linker. due to the linkage of the neighboring ripples that do not meet along a line, the elongated ripples slightly meander throughout their length as shown in fig. 5(b) and also in fig. 3(b). with the continued sputtering with @xmath43 ions nm@xmath12, each ripple in fig. 5(c) has grown longer, wider and shows the less modulation in both the height and width than the ripples in fig. the buds have also become further enlarged. contrastingly, the nodes are not identified any more along the ridge, indicating the the mass redistribution around each node is very efficient. with the further sputtering with @xmath44 ions nm@xmath12, ripples are so coarse that adjacent ripples are almost in touch. as a result, the side - by - side coalescence of adjacent ripples is frequently observed as seen in the circled area in fig. such coalescence of the ripples improves the order of the ripple pattern, because ibs drives the coalescence of the adjacent ripples, leading to the alignment of the merged ripples along the ion - beam direction. et al._@xcite also observe in their kinetic monte carlo simulation that the coalescence improves the order of the ripples by eliminating defects in the pattern. note that the distribution of the buds is not so random in fig. 5(d) as that of the incipient islands in fig. the buds often locally group or separated by region depleted of the buds, conspicuously in fig. 5(c) and also in fig. such grouping of the islands are more and more frequently observed with the larger @xmath19. the distance between the two neighboring groups in figs. 5(c) and (d) is around 400 nm that is similar to the coherence length @xmath0 of the ripple pattern and also the mean grain size. this suggests the close correlation between the grouping of the islands and the grained structure. from the temporal evolution of the patterns in fig. 5, the growth kinetics of the two kinds of primordial islands can be summarized as follows ; the @xmath45 kind of islands or 1) the buds form only on the side of each ripple facing the ion - beam. 2) the bud grows up as sputter proceeds. 3) the buds tend to locally group. 4) the mean distance between the neighboring buds along the ion - beam direction is similar to the mean grain size. the @xmath46 kind of islands or 1) the nodes reside in the middle of the ripples or each of the nodes stay inbetween two neighboring buds defining a ripple. 2) the nodes swiftly die out as sputter proceeds, indicating efficient diffusion processes of adspecies around them. upon those observation, we propose the following picture for the ripple growth on hopg(0001) ; the buds form at the grain boundaries, while the nodes on the terraces of the grains. in the early stage of sputtering, the grain boundaries are not influential on the formation of the primordial islands because the roughness of the @xmath18 surface is of the atomic scale as shown in figs. 3(a) and (b). with the continued sputtering, however, the initially embedded side walls of the grains become exposed. the side wall - formed of the edges of the graphenes - is more easily modified by the incident ion - beam than the surface of the graphite surface. in addition, the ion - beam shed the enhanced ion flux on the side walls, especially those facing the incident ion - beam due to the reduced angle from the local surface normal, and there. with all the effects in synergy, the growth of the islands becomes escalated near the grain boundaries facing the incident ion - beam, and the buds selectively decorate them. on the other hand, the diffusion of adspecies is highly efficient on the terrace due to strong intraplanar @xmath47 bond of the graphite surface.@xcite then, the islands on the terrace becomes linked to the adjacent ripple to form the nodes as seen in fig. 5(b) that becomes swiftly transformed to be a part of the respective, now elongated ripple as in figs. 5(c) and 5(d). in the long run, each ripple extends via the linkage of the ripples from a bud on a grain boundary facing the ion - beam to the other bud on the adjacent grain boundary. @xmath0 of the ripple pattern should, thus, be congruent to the mean grain size, as actually observed in the experiments. the present picture can also elucidate the experimental observation that the surface width @xmath1 of the patterned hopg is always larger than that of ng for the same sputter condition. in fig. 3, for example, @xmath48 of the patterned hopg is @xmath49 nm, larger than that of ng, @xmath50 nm. the mean height of the nano buds from the adjacent ridge of the ripple is 9.9 @xmath32 0.7 nm for both hopg and ng, and is a major source of the surface width. since the density of the nano buds is higher for hopg due to the smaller grain size than that for ng, @xmath1 is, thus, larger for hopg than for ng. this observation assures the significant role of the grain boundaries in the morphological evolution of the ripple pattern. experimental observation of the similar @xmath16s for both hopg and ng in fig. 2 indicates that the grain boundaries do not affect the coarsening of the ripple. under the present experimental condition, the mass transport across the grain boundary, transversally to the ion - beam is not driven by thermal diffusion that should be seriously hampered by the grain boundary. instead, athermal processes such as sputter - induced 1) solid flow@xcite and/or 2) ballistic diffusion@xcite must play the major roles. such athermal diffusion could be substantial for graphite, because 1) the small mass of constituent of hopg or carbon compared with that of the projectile or argon leads to large momentum transfer from the incident ar ion. habenicht@xcite observed linear dependence of @xmath16 on incident ion energy, consistent with a prediction for the ion - induced solid flow for hopg@xcite recently, amorphous carbon sputtered by xe@xmath51 of 0.2 to 10 kev showed significant mass redistribution by the incident ion, which was also found the major source of the surface instability.@xcite those results support the argument of the significant contribution from the sputter - induced mass transport to the pattern formation on hopg(0001). recently, kere _ et al._@xcite proposed a model for the pattern formation of polycrystalline (ni) films. the variation of the sputter yield depending on the orientations of the grains forming the film can well reproduce the observed pattern evolution on ni films formed under various sputter conditions without invoking the well - known instability due to curvature dependent erosion of surface by ibs.@xcite the @xmath18 hopg is largely formed of the grains differing only in relative azimuthal angle, but sharing (0001) basal plane with minute mosaic estimated from the omega - rocking width in table i. the sputter yields among grains should not significantly differ. with continued sputtering, however, the side walls at the grain boundaries in the direction facing the ion - beam becomes exposed, and have distinctly larger sputter yield than (0001) basal plane. this - dynamically driven - inhomogeneous surface with respect to the sputter yield is similar to the situation for the polycrystalline ni film, and actually leads to an instability or to the formation of the buds near the grain boundaries. this is an aspect in common with the model of kere _ et al._@xcite. away from the grain boundary or on terrace, however, the ripple formation should follow the same mechanism for both hopg and ng as expected for single crystalline surface. in the previous work with the polycrystalline au films@xcite, the large height fluctuation of au grains leads to shadowing instability. the present hopg(0001) is distinct from the metallic film, because the lateral grain size is much larger than the characteristic wavelength and the initial surface width is very small, less than 1 nm. due to the small surface height fluctuation, the shadowed area is negligible. for the ni film whose initial surface width is also less than 1 nm, kere _ et al._@xcite also observe the rapid reduction of the shaded area as sputter proceeds, and the negligible effects of shadowing. accordingly, the temporal evolution of hopg do not show any sign of the initial smoothening observed for the au film, but show monotonic increases of both @xmath1.@xcite instead, the grain boundaries play significant role in the determination of both @xmath0 and @xmath1 of the ripple pattern in contrast to the case of the au films.
Summary and conclusions
we investigated the effects of the grained structure on the pattern formation by ion - beam - sputtering (ibs), employing two different graphites : highly oriented pyrolytic graphite (hopg) and natural graphite (ng) whose mean grain size is distinctly larger than that of hopg. each ripple runs from a pronounced protuberance or a bud at one grain boundary to that at the adjacent boundary. the buds originate from the side walls of the grains - formed of the edges of the graphenes, thus having high sputter yield- that become exposed at the grain boundaries facing the ion - beam by continued sputtering. since hopg is composed of the smaller grains than ng, so is the mean uninterrupted ripple length of hopg. due to the higher bud density on hopg, the surface width of the patterned hopg is larger than that of ng, well elucidating the experimental observations. on the other hand, the wavelengths on both hopg and ng are similar, indicating that the ripple coarsening across the grain boundary proceeds via athermal processes such as sputter - induced viscous flow or ballistic diffusion.@xcite in short, the grain boundary of the graphite significantly affects the morphological evolution of the ripple pattern on graphite, but in highly anisotropic fashion. m. stepanova and s. k. dew, j. phys. : condens. matter * 21 *, 224014 (2009). m. heggie, b. r. eggen, c. p. ewels, p. leary, s. ali, g. jungnickel, r. jones, and p. r. briddon, electrochem. 98 *, 60 (1998). i. v. lebedeva, a. a. knizhnik, a. m. popov, o. v. ershova, y. e. lozovik, and b. v. potapkin, j. chem. phys. * 134 *, 104505 (2011). | employing graphites having distinctly different mean grain sizes, we study the effects of polycrystallinity on the pattern formation by ion - beam - sputtering. the grains influence the growth of the ripples in highly anisotropic fashion ; both the mean uninterrupted ripple length along its ridge and the surface width depend on the mean size of the grains, which is attributed to the large sputter yield at the grain boundary compared with that on terrace.
in contrast, the ripple wavelength does not depend on the mean size of the grains.
coarsening of the ripples - accompanying the mass transport across the grain boundaries - should not be driven by thermal diffusion, rather by ion - induced processes. | 1402.3379 |
Introduction
the x - ray source is a 7.7 s pulsar in a highly compact binary system of orbital period 2485 s. it is unusual in that it is one of the few low mass x - ray binary systems to contain an x - ray pulsar. while the x - ray emission is strongly modulated by the pulsar, there is no evidence for doppler shifts induced by the orbital motion of the source, despite extensive searches. the implies that the projected orbital radius of the neutron star is small, i.e., @xmath2sin @xmath3 @xmath4 13 m - lt - s (levine et al. optical pulsations were first detected by ilovaisky et al. (1978) and interpreted as x - ray re - processing near to, or along, the line of sight to the x - ray source. middleditch et al. (1981) found a single low frequency side - lobe which they interpret as arising from the optical re - processing of the primary x - rays on the companion star. assuming the pulsar spins in the same sense as the orbital motion, these photons will be shifted to a lower frequency by the rotation frequency of the binary orbit. from the observed frequency shift of 0.4 mhz an orbital period of 2485 s and a projected semimajor axis of 0.4 lt - s is inferred. the current picture of is of a highly compact system comprising a neutron star of mass @xmath11 m@xmath5, with a 0.08@xmath6 main sequence or 0.02@xmath6 white dwarf companion (verbunt et al. 1990). for the first decade after its discovery was rapidly spinning - up at a rate of @xmath7@xmath8 yr@xmath9. however, long term monitoring by the burst and transient source experiment (batse) on - board the compton gamma - ray observatory beginning in 1991 april found that @xmath10, and hence the accretion torque, had changed sign, resulting in a spin - down at nearly the same rate (wilson et al. it is estimated that the reversal occurred in mid-1990. observations of during the earlier spin - up phase found that the phase averaged spectrum could be modeled by a blackbody of temperature, kt, @xmath10.6 kev together with a power - law of photon index, @xmath11, of @xmath11 (e.g., pravdo et al. 1979 ; kii et al. 1986). in the 210 kev energy range the pulse profile consisted of a narrow pulse with a `` notch '', while at higher and lower energies this evolved into a roughly sinusoidal shape (levine et al. 1988 ; mavromatakis 1994). this strong energy dependence may result from anisotropic radiative transfer in a strongly magnetized plasma (kii et al. 1986). in addition to periodic pulsations, also exhibits quasi periodic behavior. both the x - ray and optical intensities show correlated flaring on timescales of @xmath11000 s (joss et al. 1978). the origin of this behavior is unknown. a 40 mhz quasi periodic oscillation (qpo) has been detected in x - rays (shinoda et al. 1990) and more recently in the optical band (chakrabarty et al. 1997). finally, the recent observation of an emission line complex near 1.0 kev by angelini et al. (1995) is particularly interesting. this emission is interpreted as arising primarily from ne k rather than from fe l, based on the measured line energies and intensities. this result suggests that the companion star has evolved past its hydrogen burning stage. ne is a by - product of he burning and therefore its overabundance suggests that the star is burning, or has burnt, he.
Observations
the low - energy concentrator spectrometer (lecs) is one of 5 scientific instruments on - board the _ bepposax _ satellite (boella et al. 1997). the lecs is described in parmar et al. (1997) and consists of a nest of conical approximation to wolter i x - ray mirrors which produce a focused beam of x - rays on an imaging gas scintillation proportional counter. the detector employs a driftless design which, in conjunction with an extremely thin entrance window, ensures an extended low energy response. the nominal energy range of the instrument is 0.110 kev and the full width at half maximum (fwhm) energy resolution is 19% at 1 kev. the field of view (fov) is 37@xmath12 diameter and the spatial resolution is 5.1@xmath12 fwhm at 1 kev. _ bepposax _ was launched into a 600 km equatorial orbit on 1996 april 30. was observed twice by the lecs during the science verification phase on 1996 august 6 from 12:00 ut to 21:50 ut and on august 9 02:25 ut to august 11 00:00 ut. the total source exposure is 40 ks and the mean lecs count rate is @xmath13 s@xmath9. because of the large spatial extent of the point spread function, background data were acquired separately by viewing blank regions of sky. background subtraction is not critical for such a bright source.
Spectral analysis
source events were obtained from both observations using the standard extraction radius of 8@xmath12, centered on the source centroid. the extracted data were rebinned to have bin widths of 0.25 @xmath14 the fwhm energy resolution. this is to help ensure an unbiased fit across the energy range, while preserving sensitivity to line features. since the lecs response is dependent on both position within the fov and extraction radius, the appropriate matrix was created using saxledas version 1.4.0 (lammers 1997). the spectrum was first fit with an absorbed blackbody plus power - law model yielding a @xmath15 of 115 for 84 degrees of freedom (dof). inspection of the residuals (see fig. [4uspec]) shows an excess centered around 1 kev, similar to that seen by angelini et al. (1995) with the asca solid - state imaging spectrometer (sis). the addition of a gaussian line feature to the model results in a @xmath15 of 92 for 81 dof. an f - test shows that the improvement is significant at the @xmath16 99.99% level. the best - fit line energy and flux are 1.05 @xmath17 kev and (4.58 @xmath18) @xmath19 photons @xmath20 s@xmath9, respectively. (all uncertainties are quoted at 68% confidence.) the equivalent width (ew) is 47.6 @xmath0 13.5 ev. the spectrum is shown in fig. [4uspec] with the line intensity set to zero. the best - fit values of kt and @xmath11 are 0.33 @xmath0 0.02 kev and 0.61 @xmath0 0.02, respectively. the derived absorption column is (6.9 @xmath0 2.0)@xmath21 atoms @xmath20, consistent with the interstellar value in the direction of (daltabuit & meyer 1972 ; dickey & lockman 1990). the best - fit line energy agrees well with that expected from a blend of 80% ne ly-@xmath11 (1.021 kev) and 20% ne he-@xmath22 (1.084 kev) emission, consistent with the asca result of angelini et al. the values of kt and @xmath11 are also similar to those measured using asca, but both a factor of 2 lower than earlier heao-1 (pravdo et al. 1979) and @xmath23 (kii et al. 1986) values. the blackbody radius, r@xmath24, of 1.0 d@xmath25 km, where d@xmath26 is the distance in kpc, is also consistent with asca. the 0.510 kev luminosity is 2.0 @xmath14 10@xmath27 ergs s@xmath9 d@xmath28 which is @xmath140% lower than measured by asca three years earlier, and a factor of 6 lower than that derived using the _ einstein _ solid state spectrometer in 1979 march (angelini et al. 1994). in the case of asca, angelini et al. (1995) attribute the spectral and luminosity differences to the torque reversal, since the heao-1, @xmath23, and _ einstein _ measurements were carried out prior to mid-1990. the asca and _ bepposax _ measurements were performed during the present spin - down phase. ll parameter & value + + n@xmath29 & 1.10 @xmath0 0.20 + @xmath11 & 0.64 @xmath0 0.02 + kt & 0.29 @xmath0 0.01 + r@xmath24 & 1.4 @xmath0 0.8 + @xmath15/@xmath30 & 115/84 + + n@xmath29 & 0.69 @xmath0 0.20 + @xmath11 & 0.61 @xmath0 0.02 + kt & 0.33 @xmath0 0.02 + r@xmath24 & 1.0 @xmath0 0.5 + e@xmath31 & 1.05 @xmath32 + @xmath33 & 0.04 @xmath0 0.04 + ew & 48 @xmath0 14 + flux@xmath31 & 4.6 @xmath18 + @xmath15/@xmath30 & 92/81 + + n@xmath29 & 0.81 @xmath0 0.26 + @xmath11 & 0.62 @xmath0 0.02 + kt & 0.33 @xmath0 0.02 + r@xmath24 & 1.0 @xmath0 0.6 + o he-@xmath11 0.568 kev flux & 5.3 @xmath34 + ew & 34 @xmath0 34 + ne ly-@xmath11 1.021 kev flux & 3.2 @xmath0 1.5 + ew & 28 @xmath0 12 + ne he-@xmath22 1.084 kev flux & 1.5 @xmath0 1.3 + ew & 13 @xmath0 10 + fe k-@xmath11 6.400 kev flux & 0.7 @xmath0 0.5 + ew & 38.6 @xmath0 22.9 + @xmath15/@xmath30 & 88/80 + in order to investigate whether the single narrow feature seen by the lecs is consistent with the emission line complex observed by the asca sis, the feature at 1.05 kev was replaced by the blend of lines used in the asca fit (see table 1 of angelini et al. each line was added separately with its energy and width fixed at the measured asca values. the only @xmath35 for inclusion in the final fit was that @xmath15 must reduce. only three lines satisfied this criteria ; the ne ly-@xmath11 line at 1.021 kev, the ne he-@xmath22 line at 1.084 kev, and the o he-@xmath11 line at 0.568 kev. surprisingly, the fit did not require an ne he-@xmath11 line and we derive an upper flux limit at the 90% confidence level of 2.2 @xmath14 10@xmath36 photons @xmath20 s@xmath9. phase - resolved asca spectra reveal the presence of a weak iron k feature over a narrow range of pulse phases (angelini et al. including such a line in the current model gives a @xmath15 of 88 for 80 dof. best - fit spectral parameters are listed in table 1. the derived o he-@xmath11 and ne he-@xmath22 line intensities are consistent with the asca values reported in angelini et al. (1995). however, the ne ly-@xmath11 line intensity is a factor of 3 lower and the 90% confidence upper flux limit to any ne he-@xmath11 emission is lower by a factor of @xmath1 6. [4u10per] we next investigated an alternate spectral model for in which the power - law plus blackbody is supplemented by emission from an optically - thin collisionally ionized plasma (specifically the vmekal model in xspec v.9.01). in principle, this would allow us to estimate the elemental abundances necessary to produce the excess emission around 1 kev, in a similar manner to angelini et al. the abundances of ne and fe were allowed to vary while the abundances of the other elements were fixed at the photospheric values of anders & grevesse (1989). both high ne and fe over - abundances gave acceptable fits to the data with @xmath15 s comparable to the `` ne complex '' fit. however, the lecs spectrum is of insufficient quality to determine meaningful limits to these abundances.
Temporal analysis
the barycentric pulse period during the lecs observations of @xmath37 s is in good agreement with the predicted value of 7.667943 @xmath0 0.000006 s derived from batse data (chakrabarty 1997). the pulse profiles in the energy ranges 0.51, 1.03.0, and 3.010 kev, are shown in fig. [4u10per]. the overall shape and energy dependence are similar to those seen by asca. the pulse profile in the lowest energy band in fig. [4u10per] may be consistent with that seen during spin - up (e.g., pravdo et al. 1979), but with a reduced amplitude. [4ups] figure [4ups] shows a white noise subtracted, power density spectrum of the lecs data from which quasi periodic oscillations (qpos) are apparent. the center frequency is 0.049 @xmath0 0.002 hz and the fwhm is 0.015 @xmath38 hz. the qpo strength is 18 @xmath39% _ rms _ of the mean count rate. within uncertainties, the qpo amplitude is the same in the 0.52.0 kev and 2.010.0 kev energy ranges. the qpo centroid frequency, width, and amplitude are consistent with the asca measurements (angelini et al. the amplitude and width are also consistent with the @xmath40 values (shinoda et al. 1990), but the centroid frequency is not. this change may be related to the torque reversal.
Discussion
the lecs data confirm the recent asca detection of excess emission near 1 kev from. in addition, the best - fit spectral parameters confirm that the spectral shape remains changed following the mid-1990 torque reversal. both kt and @xmath11 decreased by a factor of @xmath12, while the x - ray luminosity decreased by a factor of 6. as before the reversal, the new parameters appear stable with time. chakrabarty et al. (1997) report that the intensity of the source is steadily decreasing with time. the lecs results support this, since the 0.510 kev source intensity is 0.6 that measured by asca. extrapolating from previous measurements (see fig. 8 of chakrabarty et al. 1997), the expected decrease is a factor of @xmath10.7. the shape of the pulse profile is also different from that measured before the reversal. prior to the reversal, the profile was strongly energy and phase dependent (e.g., see levine et al. 1988), while the lecs pulse profile has the same shape, but a variable amplitude, over the 1.010.0 kev energy range (see fig. we thank chris butler, luigi piro, and the staff of the _ bepposax _ science data center in rome. lorella angelini is thanked for discussions and deepto chakrabarty for providing the batse pulse period. t. oosterbroek acknowledges an esa research fellowship. the _ bepposax _ satellite is a joint italian and dutch programme. | we report on observations of the x - ray pulsar by the lecs instrument on - board _
bepposax_. we confirm the recent asca discovery of excess emission near 1 kev (angelini et al.
1995).
the pulse period of 7.66794 @xmath0 0.00004 s indicates that the source continues to spin - down.
the phase averaged spectrum is well fit by an absorbed power - law of photon index 0.61 @xmath0 0.02 and a blackbody of temperature 0.33 @xmath0 0.02 kev, together with an emission feature at 1.05 @xmath0 0.02 kev.
this spectral shape is similar to that observed by asca during the spin - down phase, but significantly different from measurements during spin - up.
this suggests that the change in spectrum observed by asca may be a stable feature during spin - down intervals.
the source intensity is a factor @xmath12 lower than observed by asca three years earlier, confirming that continues to become fainter with time. | astro-ph9706021 |
Introduction
cadmium arsenide, known for decades as an inverted - gap semiconductor, has recently been shown to be a three - dimensional dirac semimetal. these materials, with a massless dirac dispersion throughout the bulk, are the 3d analogs of graphene, and cd@xmath0as@xmath1 is foremost among them : stable, high - mobility, and nearly stoichiometric. it displays giant magnetoresistance, hosts topologically nontrivial fermi - arc states on its surface, and is predicted to serve as a starting point from which to realize a weyl semimetal, quantum spin hall insulator, or axion insulator. ultrafast spectroscopy, which monitors changes in a sample s optical properties after excitation by a short laser pulse, has in many materials provided a time - resolved probe of basic carrier relaxation processes such as electron - electron and electron - phonon scattering and carrier diffusion. calculations for dirac and weyl semimetals predict that photoexcited electrons will, anomalously, cool linearly with time once their elergy drops below that of the lowest optical phonon. nothing, however, is known of cadmium arsenide s ultrafast properties. here we use the transient - grating method, which measures both the magnitude and phase of the complex change of reflectance. our measurements reveal two processes, distinct in lifetime and in phase, by which the sample s reflectance recovers after photoexcitation. analysis of the signal s phase allows us to identify changes in both the real and the imaginary parts of the index of refraction, @xmath2. the fastest response, with a lifetime of 500 fs, is a reduction in the absorptive part, @xmath3, which we attribute to photoexcited electrons filling of states near the excitation energy. the longer - lived response is an increase in @xmath4 and arises from the filling of states at much lower energy. these observations reveal a two - stage cooling process, which we suggest may proceed first through optical phonons, then through acoustic.
Methods
we measured two samples of cd@xmath0as@xmath1. sample 1 had well - defined crystal facets and measured a few millimeters in each dimension. it was grown by evaporation of material previously synthesized in argon flow and was annealed at room - temperature for several decades. such annealing is known to increase electron mobility and to decrease electron concentration. indeed, hall measurements on a sample of the same vintage give electron density @xmath5 @xmath6 (roughly independent of temperature), metallic resistivity, and mobility @xmath7 @xmath8 at 12 k. x - ray powder diffraction gives lattice parameters in agreement with previous reports. sample 2 was grown in an argon - purged chamber by cvd in the form of a platelet ; the surface was microscopically flat and uniform. the ratio of the main cd and as peaks seen in energy - dispersive x - ray spectroscopy corresponds to cd@xmath0as@xmath1, indicating proper stoichiometry. though its transport was not unambiguously metallic, in our experiment samples 1 and 2 behaved identically. this is consistent with the interpretation given below, that our ultrafast signal arises from the dynamics of high - energy electrons. we use the transient - grating method to measure the change, @xmath9, in reflectance after photoexcitation. a pair of pump pulses interfere at the sample, exciting electrons and holes in a sinusoidal pattern. the sinusoidal variation in @xmath10 caused by this excitation is the `` grating. '' time - delayed probe pulses reflect and diffract off of the grating. the experimental geometry is shown in fig. we use a diffractive - optic beamsplitter to generate the pair of pump pulses. as these pulses converge on the sample, they make angles @xmath11 with the surface normal, creating a grating of wavevector @xmath12. (here @xmath13 is the light s wavelength.) two probe pulses are incident on the sample at the same angles, @xmath11. the difference in their wavevectors equals @xmath14, so when each probe diffracts off of the grating, it is scattered to be collinear with the other probe. orders. a concave mirror (ccm) focuses the two probes onto the sample (s), at an angle @xmath15 from the normal. diffracted beams (dashed) scatter through @xmath16, so that each diffracted probe is collinear with the opposite reflected probe. pump beams (not shown) follow the same paths. however, pump beam paths are tipped slightly out of the page, and probe beams slightly into the page. thus the pumps are not collinear with the probes, nor are the reflected beams collinear with the incident ones., width=2,height=0] this geometry allows for simple heterodyne detection of the diffracted probe : rather than provide a separate `` local oscillator '' beam, the reflected beam from one probe acts as a local oscillator for the diffracted beam from the other probe. if an incident probe has electric field @xmath17, then the reflected and diffracted probe fields are, respectively, @xmath18 here @xmath19 is the complex reflectance, @xmath20 is the order of diffraction, and @xmath21 is a geometric phase due to the grating s spatial location. @xmath21 can not be measured, but it can be changed controllably. heterodyne detection of @xmath22 improves signal, and we suppress noise by modulation of @xmath21 and lock - in detection. the transient - grating signal is proportional to @xmath23 each measurement is repeated with the grating shifted by a quarter wavelength, giving the real and imaginary parts of @xmath24. in the absence of measurable diffusion, as seen here, @xmath25. the laser pulses have wavelength near 810 nm, duration 120 fs, repetition rate 80 mhz, and are focused to a spot of diameter 114 @xmath26 m. the pump pulses have fluence @xmath27 at the sample of @xmath28 @xmath26j/@xmath29 ; the probe pulses are a factor of 10 weaker. at 810 nm cd@xmath0as@xmath1 has index of refraction @xmath30, giving @xmath31. the absorption length is of order 45 nm and the reflectivity is 35%, so at our highest fluence each pair of pump pulses photoexcites electrons and holes at a mean density of @xmath32 @xmath6. measurements were taken at temperatures @xmath33 k and 80 k, and one at 115 k.
Results
@xmath33 k, @xmath34 @xmath26m@xmath35, @xmath36 @xmath26j/@xmath29. * (a), (b) * are @xmath37, @xmath38 diffracted orders, respectively.,width=3,height=1] examples of the data obtained appear in fig. [transients]. all of our data fit well to the form : @xmath39 the data s three most salient features are each evident. first, the signal returns to equilibrium through two distinct decay processes, the first with @xmath40 fs and the second with @xmath41 ps. second, the two decay processes differ distinctly in complex phase. finally, as shown in fig. [taus], the decays are insensitive to both @xmath14 and @xmath27. of these observations, the complex phase will play the key role in our identification, below, of the causes of the two decay processes. and @xmath42 are roughly constant _ vs_. pump fluence. * (c) * and * (d) * : @xmath43 and @xmath44 are roughly constant _ vs_. @xmath45. @xmath42 is consistent with diffusion coefficients from @xmath46 (horizontal line) to @xmath47 @xmath29/s (sloped line).,width=3,height=1] in fact, the transient reflectance is even less sensitive to experimental conditions than fig. [taus] indicates. we varied the conditions sample, @xmath48, @xmath27, @xmath20, and @xmath14to measure 32 distinct @xmath9 curves ; we saw little variation in any of the fitting parameters of eq. [twoexp]. the relative size of the two decay processes is constant, @xmath49. the constant term increases from @xmath50 at 80 k to @xmath51 at 295 k, but always remains small. we attribute the @xmath52 term to lattice heating, for which we present qualitative evidence in the supplemental material. transient - grating experiments are often used to measure the diffusivity @xmath53 of photoexcited species. in the presence of diffusion, the diffracted signal @xmath24 decays faster than @xmath9 because carriers diffuse from the grating s peaks to its troughs. this effect is stronger at higher @xmath14, because the peak - to - trough distance is shorter. however, fig. [taus] (d) shows that @xmath42 is independent of @xmath14, consistent with @xmath46. we caution against assigning too much weight to this negative result. the sloped line in fig. [taus] (d) shows that our data exclude only @xmath54 @xmath29/s a distinctly high upper bound. so the carriers likely do diffuse, but relax so quickly that they do not diffuse through an appreciable fraction of the grating s wavelength. the situation for @xmath55 is similar : fig. [taus] (c). our typical measurement, of @xmath37, is not sensitive to the multiplication of eq. [twoexp] by an overall phase. however, by additionally measuring @xmath56, it is possible to determine the absolute phase of @xmath57. we have done several such measurements on each sample ; one appears in fig. [transients] (b). we can then calculate @xmath58 and similarly for the signal s @xmath59 and @xmath52 components. though the half - angle in eq. [halfangle] can take two values differing by @xmath60, this ambiguity is easily resolved. the photoinduced change in reflectivity is @xmath61 ; we measure @xmath62 and choose the angles @xmath63 to reproduce its sign, shown in figs. [phasefig] (a) and [phasefig] (b). , measured. * (b) * : @xmath62, calculated from our mean fit parameters. the sign of each component is chosen to match the shape of the measured curve. * (c) * : transient change, @xmath64, in index of refraction calculated from our mean fit parameters. imaginary part, dashed, accounts for most of the fast decay. real part, solid, accounts for most of the slow decay and the constant term.,width=2,height=1] we now use these angles to determine the photoinduced change in @xmath10. the reflectance changes after photoexcitation by @xmath65\delta n(t)$]. for cadmium arsenide, the bracketed factor has argument @xmath66, so @xmath67. we obtain, finally, @xmath68, @xmath69, and @xmath70. this result is surprisingly simple. the signal s faster component results from a negative @xmath71a reduction in absorption and the slower from a positive @xmath72a decrease in the light s phase velocity. the calculated @xmath73 appears in fig. [phasefig] (c). for cd@xmath0as@xmath1, both the real and imaginary parts of @xmath64 appear in @xmath74, and they may be distinguished by the time - scales of their decays.
Discussion
the key questions in interpreting these two decay processes are what has been excited, and by what means it relaxes. our excitation energy @xmath75 ev is well beyond the region of cadmium arsenide s dirac - like dispersion, and, though optical transitions near 1.5 ev are believed to occur at the @xmath76 point, transitions are allowed between electrons and holes of several different bands. cadmium arsenide s large unit cell hosts over 200 phonon branches ; infrared and raman measurements detect a few dozen, with energies from 3.2 mev to 49 mev. (the deficit of detected branches is attributed to a weak polarizability.) considering the abundance of excited states and relaxation pathways available, we can not hope to identify precise processes of excitation or relaxation. nonetheless, the optical signal s phase constrains our interpretation significantly. photoexcitation changes a sample s reflectance by changing its frequency - dependent absorption coefficient. leaving aside the possibility of changes to the band structure, it does so either by occupying excited states or by changing the free carriers absorption. our experiment s probe photons have the same energy @xmath77 as those of the pump. therefore excited electrons fill phase - space effectively, reducing absorption at @xmath77, and causing the negative @xmath71 observed in our fast decay process. this picture remains valid even as electrons scatter away from their initial excited energy @xmath78. carrier - carrier scattering gradually creates a thermal distribution of electrons at elevated temperature. if this process is fast compared to the carriers energy loss, their mean energy remains nearly @xmath78, and they occupy states both below and above @xmath78. such a distribution results in @xmath79, just as does the conceptually simpler case of phase - space filling exactly at @xmath78. our signal s slower component has @xmath80, which, according to the kramers - kronig relation, must result either from increased absorption at @xmath81 or from decreased absorption at @xmath82. we can eliminate the former as the cause of our signal. if absorption increases at all, it should do so at low frequency due to enhanced free - carrier (intraband) conductivity ; this would cause a negative @xmath72 that we do not observe. on the other hand, there is a straightforward mechanism for decreased absorption at @xmath82 : as electrons and holes lose their excess energy, they fill phase space at progressively lower energies. kramers - kronig analysis using a simplified density of states suggests that, by the time @xmath83 becomes mostly real, the carriers mean energy should be @xmath84 or less ; our data show that cooling of this magnitude occurs within 500 fs. we attribute this cooling to phonons rather than to carrier - carrier scattering, since there are too few cool, background electrons compared to the hot, photoexcited ones (an order of magnitude fewer for sample 1 and at our highest fluence). the subsequent dynamics of @xmath72 indicates that once carriers reach low energy, their relaxation slows to give @xmath85 ps. possibly cooling slows when the carriers excess energy falls below that of the lowest optical phonon, as occurs in graphene and as recently preicted for weyl and 3d dirac semimetals. however, for cd@xmath0as@xmath1 this energy is just 15 mev. other possible relaxation processes include electron - impurity scattering or electron - electron scattering with plasmon emission. however, we suggest that after the initial 500-fs cooling the carriers and optical phonons have equilibrated ; further cooling requires the slower emission of acoustic phonons. this picture fits the measured time - scale : electron - lattice cooling in bismuth, a semimetal, occurs in 5 ps. we may gain insight into the two decay processes we observe in cadmium arsenide by considering another dirac semimetal, graphene. photoexcitation of graphene initially produces electrons and holes with separate chemical potentials. within the pulse duration, these carriers partially equilibrate with optical phonons ; they then quickly occupy the dirac cone and enhance the intraband conductivity, and recombine in less than a picosecond. the chemical potential reverts to its original level, but because carriers are still hot they continue to occupy high - energy states, filling phase - space and reducing optical absorption. these hot carriers finally relax _ via _ optical, then acoustic, phonons. our measurements indicate that some of the same processes occur in cadmium arsenide, but possibly not all. we do not know whether carriers relax into the dirac cone, but the weakness of cadmium arsenide s photoluminescence suggests that many do. we also can not conclude that, as in graphene, photoexcitation produces electrons and holes with separate chemical potentials ; time - resolved photoemission and thz could more directly detect changes in carrier population and conductivity. in conclusion, we have shown that after photoexcitation cadmium arsenide relaxes in two distinct stages, irrespective of sample, fluence, and temperature. first, carriers fill phase - space at the pump energy, but relax within 500 fs to lower energy. these low - energy carriers relax further with a time - scale of 3.1 ps ; the lattice finally reaches high temperature. this result may guide further ultrafast measurements on cd@xmath0as@xmath1 and other dirac and weyl semimetals.
Acknowledement
this work was supported by the national science foundation grant no. dmr-1105553.
Sample properties
here we describe further our two samples of cd@xmath0as@xmath1. fig. [xray] shows the x - ray powder diffraction pattern from samples of the same vintage as sample 1. the data were fit using rietveld refinement, giving lattice parameters @xmath86 and @xmath87 with space group symmetry @xmath88, consistent with other recent experiments. no peak corresponding to an impurity phase was detected. [samplediagnostics] (a) shows the resistivity of a sample of the same vintage. the resistivity is metallic, and at low temperature is nearly as small as that measured in samples exhibiting confirmed dirac - semimetal behavior. for a sample of type 1. for a sample of type 2, we show * (b) * the zero - bias resistance @xmath89 and * (c) * the differential resistance @xmath90 _ vs_. voltage @xmath91, averaged over the temperatures measured.,width=3,height=1] for a sample of the same vintage as sample 2, the current - voltage characteristics were measured by placing it across a gap formed by two indium electrodes. the dc photoconductivity at 300 k was negative, consistent with the optical heating of free carriers. the differential resistance, fig. [samplediagnostics] (b) and (c), is not unambiguously metallic : it depends non - monotonically on temperature and is bias - dependent. nonetheless, in our experiment samples 1 and 2 behaved identically.
Description of the transient grating
though transient - grating spectroscopy has a long history, the method has advanced considerably in recent decades. here we provide additional details of the method and analysis used in this work. using the notation of eq. [notation], the reflected and diffracted fields are, respectively, @xmath92 it is instructive to compare heterodyne detection to the traditional, homodyne - detected transient - grating signal, in which one simply measures the diffracted beam, @xmath93. the homodyne signal is second - order in @xmath24. unfortunately, the photoinduced changes in a sample s optical response@xmath24 and @xmath9are typically quite small. one advantage of heterodyne detection is a large increase in signal, for it has several terms of first or zeroth order : latexmath:[\[\begin{aligned } indeed, the second - order terms are negligible, giving latexmath:[\[\begin{aligned } the term of interest is the one proportional to @xmath24 ; it is also the only one depending on @xmath21. to isolate this term, we use the coverslip discussed below to modulate @xmath21 at 95 hz, and filter our signal through a lock - in amplifier. this procedure acts as a derivative @xmath95, giving a signal proportional to @xmath96 equivalent to eq. [eq2]. key to the heterodyne detection of the transient grating is the ability to control and modulate @xmath21, by controlling the grating s spatial position. we do this by passing one of the incident pump beams obliquely through a thin, glass coverslip. fine adjustments of the coverslip s angle change the beam s path - length, adding or subtracting phase relative to the other pump beam. the coverslip is mounted on both a torsional oscillator and a stepping rotation stage. the former allows us to modulate the coverslip s angle rapidly and sinusoidally, for lock - in detection ; the latter allows us to change the angle in calibrated increments. below, when we discuss measurement `` at '' a particular coverslip position, we mean the coverslip s _ central _ position, about which it oscillates at 95 hz. to maintain the spatial and temporal overlap of the beams that converge on the sample, we introduce three similar coverslips into the paths of both probe beams and of the other pump. their orientations are fixed, but are similar to that of the modulated coverslip. to obtain data such as that shown in fig. [transients] (a), we set the coverslip to a position corresponding to an arbitrary, unknown @xmath21, and measure using the @xmath37 diffracted probe. we then shift to @xmath97 and measure again. we obtain @xmath98 and @xmath99 and define these, respectively, as the real and imaginary parts of our signal : re@xmath100 and im@xmath100. fitting to the form : @xmath101 defines the set of angles @xmath102, @xmath103, @xmath104. the superscript indicates that @xmath37. hereafter we consider just one of the signal s @xmath105, @xmath59, and @xmath52 components ; the same equations apply to each. comparison of eqs. [real], [imaginary], and [twoexp] shows that @xmath106 because re@xmath107 and im@xmath108. up to this point, @xmath109 is arbitrary, because @xmath21 is arbitrary. we next describe how measurement of the @xmath56 diffracted order allows us to eliminate @xmath21 and to determine @xmath63. our measurements of the @xmath56 diffracted order are done at the same coverslip positions as for the @xmath110 order. these correspond to grating phases of @xmath21 and @xmath111. our transient - grating signals are, respectively, @xmath112 and @xmath113 from which @xmath114 we can then calculate @xmath115 equivalent to eq. [halfangle]. above, we state that @xmath117 because the diffusion is negligible. here we clarify the reasoning, which may otherwise appear circular. @xmath9 must equal the @xmath116 limit of @xmath24. we observe that, within experimental error, @xmath24 is the same at all @xmath14, including at some rather low @xmath14. therefore our measured @xmath24 does in fact equal @xmath9. in other words, we never assume that the diffusion is negligible ; we observe it. nm and @xmath47 @xmath29/s. if @xmath53 is lower, the density will be even more nearly constant.),width=2,height=1] because carriers are photoexcited within an absorption length @xmath118 of just 45 nm, they will diffuse away from the sample s surface. as their density at the surface drops, @xmath64 will decrease. surprisingly, however, this effect has little influence on the time constants @xmath55 and @xmath42 measured in this experiment. the carriers initial exponential distribution quickly evolves to be nearly gaussian, and diffusion broadens a gaussian s width only as @xmath119. (see, for instance, sheu _ et al. _, particularly eq. 3 and fig. 5a.) fig. [inward] makes this argument more quantitative. we used the diffusion equation with no relaxation term_i.e. _ for conserved particle number to model carriers diffusion away from the sample s surface, and we plot the density at the surface as a function of time. the largest drop in density occurs at very early times, before any of the data shown in fig. [transients]. the drop at later times is not nearly enough to account for the experimentally observed decays. for this reason we conclude that our relaxation rates are little influenced by inward diffusion.
Optical heating
above, we measure the photoinduced change of reflectance, @xmath9, and argue that it is related to the cooling of optically heated electrons. here we describe qualitative evidence for the electrons high temperature. when the laser was incident on the sample, we saw that the illuminated spot glowed with broadband visible light ; it looked like incandescence, of a reddish hue. some locations on the sample surface glowed more than others ; however, we excluded surface contamination as a cause of the light emission by visual examination of the sample and by cleaning with acetone and methanol. et al_. measured a similar effect in graphene, caused by thermal emission by electrons heated to several thousand kelvin. these electrons were partially equilibrated with the optical phonons. after equilibration with all phonon modes, the lattice temperature was estimated to be around 700 k. the more complex band structure of cd@xmath0as@xmath1 precludes the quantitative analysis of lui _ et al_., but we expect that emission from our sample is caused by similarly heated electrons. above, we attribute our signal s small, constant component @xmath52 to lattice heating. it is unremarkable that energy deposited by the laser should eventually find its way to the lattice. however, given the modest optical power used in our experiment tens of milliwats for the pump beams this heating maifests itself in a surprising way that may serve to caution future experimenters. we observed that the direction of the probe beam s specular reflection from the sample s surface could vary by about @xmath120, depending on whether the more powerful pump beam was incident on the sample or blocked. this change was reproducible over dozens of cycles, and occurred with a time constant of several seconds. reflection remained specular, but the orientation of the sample s surface evidently shifted. after many cycles, sample 1 s surface showed small cracks. we explain this observation as follows. the thermal conductivity of cadmium arsenide is low, of order 1 w / k - m, leading to large temperature gradients. the material suffers several structural phase transitions at elevated temperature, the lowest at 503 k ; combined with temperature gradients, these could create strains that move the sample s surface slightly. note that a transient - grating signal can not be measured when the reflected beam is shifting. we were able to obtain data because samples neither shifted nor glowed when exposed to atmosphere, perhaps due to convective cooling. for low - temperature measurement under vacuum, sample 2 glowed and shifted only rarely, evidently depending on which part of the sample was illuminated. the sample was never measured while incandescing ; nonetheless, it is likely that even when the sample was cooled to 80 k, the measured spot was much hotter. both a second platelet - like sample and sample 1 glowed and shifted more consistently and could not be measured under vacuum. these observations suggest the value of thin - film samples or of laser systems with lower average power. 10 [2]#2 [1]#1 [1]#1,,,, and, ``, '' (12), ().,,,,,,,,,,,,, ``, '' ().,,,,,,,,,,,,,,,,,, ``, '' (7), ().,,,,, and, ``, '' (2), ().,,,,, and, ().,,,,, and,, ().,,,,,,,,,,,,,,,,,,,,,,, ``, '',, and,,, ().,,, and, ``, '' (20), () r. lundgren and g. a. fiete, `` electronic cooling in weyl and dirac semimetals. '' arxiv:1502.07700 (2015)., ``, '' (2 - 3), (). and, ``, '', (). see appendices for the samples transport properties, additional description of the transient grating, and a discussion of optical heating of the samples.,, and, `` ultrafast heterodyne - detected transient- grating spectroscopy using diffractive optics '' (6) ().,, and, ``, '' (16) (). and, ``, '' (18), ().,, and, ``, '' (3), (). errors and error - bars are standard deviation of the mean. the lack of measurable diffusion is perhaps to be expected. photoexcited electrons and holes must move together to preserve local charge neutrality, resulting in ambipolar diffusion. in an @xmath10-type sample, the ambipolar diffusivity nearly equals the holes diffusivity, which is likely much less than the electrons. and, ``, '', ().,, and, ``, '', 2, ().,,, and, ``, '' (2), ().,,, and, ``, '' (4), (). note that electronic energy of @xmath78 corresponds to optical transitions at @xmath77. and, ``, '' (20), ().,,,,, and, ``, '' (11), ().,,,, and, `` free - carrier relaxation and lattice heating in photoexcited bismuth '', ().,,,,,,, and, ``, '' (1), ().,,, and, ``, '' (12), ().,,,,,,,,, and, ``, '' (12), ().,,,,,,, and, ``, '' (15), ().,, and, ``, '', ().,,,,, and, `` the crystal and electronic structures of cd@xmath0as@xmath1, the three - dimensional analogue of graphene '', ().,,,,,, and, `` quantum transport evidence for a three - dimensional dirac semimetal phase in cd@xmath0as@xmath1 '', ().,, and, (,). and, `` transient grating optical heterodyne detected impulsive stimulated raman - scattering in simple liquids '' (9), ().,, and, `` optical heterodyne - detection of impulsive stimulated raman - scattering in liquids '' (20) ().,, and, `` how to make femtosecond pulses overlap '' (17) ().,,,,, and, `` observation of spin coulomb drag in a two - dimensional electron gas '' (7063), (). the characteristic time - scale of the diffusive process is @xmath121. in this experiment @xmath122 ps. indeed, this is the _ only _ time - scale in the problem, so that the @xmath123-axis in fig. [inward] may we re - scaled by the factor @xmath124 to adapt it to other experiments. and, ``, '' (11), (). and, ``, '' (1), (). | we report ultrafast transient - grating measurements of crystals of the three - dimensional dirac semimetal cadmium arsenide, cd@xmath0as@xmath1, at both room temperature and 80 k. after photoexcitation with 1.5-ev photons, charge - carriers relax by two processes, one of duration 500 fs and the other of duration 3.1 ps. by measuring the complex phase of the change in reflectance
, we determine that the faster signal corresponds to a decrease in absorption, and the slower signal to a decrease in the light s phase velocity, at the probe energy.
we attribute these signals to electrons filling of phase space, first near the photon energy and later at lower energy.
we attribute their decay to cooling by rapid emission of optical phonons, then slower emission of acoustic phonons.
we also present evidence that both the electrons and the lattice are strongly heated.
+ + the following article appeared in _ applied physics letters _ and may be found at + http://scitation.aip.org/content/aip/journal/apl/106/23/10.1063/1.4922528 + (this version of the article differs slightly from the published one.) + + _ copyright 2015 american institute of physics.
this article may be downloaded for personal use only.
any other use requires prior permission of the author and the american institute of physics. _ | 1502.07007 |
Introduction
critical properties of structurally disordered magnets remain a problem of great interest in condensed matter physics, as far as real magnetic crystals are usually non - ideal. commonly, in the theoretical studies, as well as in the mc simulations, one considers point - like uncorrelated quenched non - magnetic impurities @xcite. however, in real magnets one encounters non - idealities of structure, which can not be modeled by simple point - like uncorrelated defects. indeed, magnetic crystals often contain defects of a more complex structure : linear dislocations, planar grain boundaries, three - dimensional cavities or regions of different phases, embedded in the matrix of the original crystal, as well as various complexes (clusters) of point - like non - magnetic impurities @xcite. therefore, a challenge is to offer a consistent description of the critical phenomena influenced by the presence of such complicated defects. different models of structural disorder have arisen as an attempt to describe such defects. in this paper we concentrate on the so - called long - range - correlated disorder when the point - like defects are correlated and the resulting critical behaviour depends on the type of this correlation. several models have been proposed for description of such a dependence @xcite, a subject of extensive analytical and numerical @xcite treatment. a common outcome of the above studies is that although the concentration of non - magnetic impurities is taken to be far from the percolation threshold, in the region of weak dilution, the impurities make a crucial influence on an onset of ordered ferromagnetic phase. given that the pure (undiluted) magnet possesses a second - order phase transition at certain critical temperature @xmath0, an influence of the weak dilution may range from the decrease of @xmath0 to the changes in the universality class and even to the smearing off this transition @xcite. moreover, the critical exponents governing power low scaling in the vicinity of @xmath0 may depend on the parameters of impurity - impurity correlation. to give an example, the harris criterion, which holds for the energy - coupled uncorrelated disorder @xcite is modified when the disorder is long - range correlated @xcite. in particular, when the impurity - impurity pair correlation function @xmath1 decays at large distances @xmath2 according to a power law : @xmath3 the asymptotic critical exponents governing magnetic phase transition (and hence the universality class of the transition) do change if @xcite @xmath4 where @xmath5 is the correlation length critical exponent of the undiluted magnet. the above condition ([2]) holds for @xmath6, @xmath7 being the space (lattice) dimension. for @xmath8 the usual harris criterion @xcite is recovered and condition ([2]) is substituted by @xmath9. the fact, that the power of the correlation decay might be a relevant parameter at @xmath6 can be easily understood observing an asymptotics of the fourier transform @xmath10 of @xmath1 at small wave vector numbers @xmath11. from ([1]) one arrives at @xmath12, which for @xmath6 leads to a singular behaviour at @xmath13. as far as the small @xmath11 region defines the criticality, the systems with @xmath6 are good candidates to manifest changes in the critical behaviour with respect to their undiluted counterparts. on contrary, impurity - impurity correlations at @xmath8 do not produce additional singularities with respect to the uncorrelated point - like impurities, therefore they are referred to as the short - range correlated. in turn, the disorder characterized by eq. ([1]) with @xmath6 is called the long - range correlated. there are different ways to model systems with the long - range - correlated disorder governed by eq. ([1]). the most direct interpretation relies on the observation that the integer @xmath14 in eq. ([1]) corresponds to the large @xmath2 behaviour of the pair correlation function for the impurities in the form of points (@xmath15), lines (@xmath16), and planes (@xmath17) @xcite. since the last two objects extend in space, the impurities with @xmath6 sometimes are called the extended ones. note that the isotropic form of the pair correlation function ([1]) demands random orientation of such spatially extended objects. non - integer @xmath14 sometimes are treated in terms of a fractal dimension of impurities, see e.g. @xcite. besides energy - coupled disorder, the power - low correlation decay ([1]) is relevant for the thermodynamic phase transition in random field systems @xcite, percolation @xcite, scaling of polymer macromolecules at presence of porous medium @xcite, zero - temperature quantum phase transitions @xcite. our paper was stimulated by the observations of obvious discrepancies in the state - of - the - art analysis of criticality in three - dimensional ising magnets with the long - range - correlated disorder governed by eq. ([1]). indeed, since for the pure @xmath18 ising model @xmath19 @xcite, the long - range correlated disorder should change its universality class according to eq. ([2]). whereas both theoretical and numerical studies agree on the validity of extended harris criterion ([2]) and bring about the new universality class @xcite, the numerical values of the critical exponents being evaluated differ essentially. we list the values of the exponents found so far by different approaches in table [tab1] and refer the reader to the section [ii] for a more detailed discussion of this issue. here, we would like to point out that presently the results of each of existing analytical approaches (refs. @xcite and @xcite) is confirmed by only one numerical simulation (refs. @xcite and @xcite, respectively). to resolve such a bias, we perform mc simulations of a @xmath18 ising model with extended impurities and evaluate critical exponents governing ferromagnetic phase transition. as it will become evident from the further account, our estimates for the exponents differ from the results of two numerical simulations performed so far @xcite and are in favour of a non - trivial dependency of the critical exponents on the peculiarities of long - range correlations decay. lllll reference & @xmath20 & @xmath21 & @xmath22 & @xmath23 + + weinrib, halperin, @xcite & 1 & 2 & 1/2 & 0 + prudnikov _ et al. _, @xcite & 0.7151 & _ 1.4449 _ & _ 0.3502 _ & -0.0205 + ballesteros, parisi, @xcite@xmath24 & 1.012(10) & _ 1.980(16) _ & _ 0.528(7) _ & 0.043(4) + @xcite@xmath25 & 1.005(14) & _ 1.967(23)_&_0.524(9) _ & 0.043(4) + prudnikov _ et al. _, @xcite@xmath24 & 0.719(22) & _ 1.407(24) _ & 0.375(45) & _ 0.043(93) _ + @xcite@xmath25 & 0.710(10) & 1.441(15)&0.362(20) & _ -0.030(7) _ + the outline of the paper is the following. in the next section we make a brief overview of the results of previous analysis of the 3d ising model with long - range - correlated impurities paying special attention to the former mc simulations. details of our mc simulations are explained in sections [iii] and [iv]. there, we formulate the model and define the observables we are interested in. we analyze statistics of typical and rare events taking the magnetic susceptibility as an example in section [iv]. the numerical values of the exponents are evaluated there by the finite - size scaling technique. section [v] concludes our study.
Overview of previous analytical and mc results
currently, there exist two different analytical results for the values of critical exponents of the 3d ising model with long - range - correlated impurities. the first one is due to weinrib and halperin, who formulated the model of a @xmath26-vector magnet with quenched impurities correlated via the power law ([1]) @xcite. they used the renormalzation group technique carrying out a double expansion in @xmath27, @xmath28, considering @xmath29 and @xmath30 to be of the same order of magnitude, and estimating values of the critical exponents to the first order in this expansion. further, they conjectured the obtained first order result for the correlation length critical exponent @xmath31 to be an exact one and to hold for any value of spin component number @xmath26 provided that @xmath32. by complementing eq. ([3]) with the first order value of the pair correlation function critical exponent, @xmath33 @xcite, the other critical exponents can be obtained via familiar scaling relations. for @xmath34 the exponents are listed in table [tab1]. the second theoretical estimate was obtained by prudnikov _ @xcite in the field theoretical renormalization group technique by performing renormalization for non - zero mass at fixed space dimension @xmath18. their two - loop calculations refined by the resummation of the series obtained bring about a non - trivial dependence of the critical exponents both on @xmath26 and on @xmath14. we list their values of the exponents for @xmath35 and @xmath34 in the second row of table [tab1]. in particular, one observes that the correlation length exponent @xmath20 obtained in ref. @xcite differs from the value predicted by ([3]) by the order of 25 %. as for the reason of such discrepancy the authors of ref. @xcite point to a higher order of the perturbation theory they considered together with the methods of series summation as well as to taking into account the graphs which are discarded when the @xmath36-expansion is being used. let us note however, the qualitative agreement between the above analytical results : both renormalization group treatments, refs. @xcite and @xcite, predict that the new (long - range - correlated) fixed point is stable and reachable for the condition considered and hence the 3d ising model with long - range - correlated impurities belongs to a new universality class. these are only the numerical values of the exponents which call for additional verification. in such cases one often appeals to the numerical simulations. indeed, two simulations were performed, and, strangely enough, the results again split into two groups : whereas the simulation of ballesteros and parisi @xcite is strongly in favour of the theoretical results of weinrib and halperin, the simulation of prudnikov _ et al. _ @xcite almost exactly reproduces former theoretical results of this group @xcite as one can see from table 1. there, we give the results for the exponents obtained in simulations complementing them by those that follow if one uses familiar scaling relations (the last are shown in italic). since our own study will rely on the simulation technique as well, we describe an analysis performed in refs. @xcite in more details. in ref. @xcite the simulations were done for two different types of the long - range - correlated disorder referred by the authors as the gaussian and the non - gaussian one. in the first case, the point - like defects are scattered directly on the sites of a 3d simple cubic lattice according to the desired decay of the pair correlation function ([1]), @xmath34. in the second case, the defects form the lines directed along the randomly chosen axes and, as was mentioned before, their impurity - impurity pair correlation function should also decay at large @xmath2 as @xmath37). the mc algorithm used in ref. @xcite is a combination of a single - cluster wolff method and a swendsen - wang algorithm. after a fixed number of a single - cluster updates one swendsen - wang sweep is performed. the procedure was called a mc step (mcs). the number of single cluster flips was chosen such that the autocorrelation time @xmath38 was typically 1 mcs. for thermalization of the system @xmath39 mcss were performed and then various observables were measured. the simulation was done on the lattice of sizes @xmath40 and @xmath41. then @xmath39 mcss have been used for measurements on @xmath42=20000 different samples for @xmath43 and on @xmath42=10000 samples for @xmath44. the procedure seems to be quite safe. values of the critical exponents @xmath20 and @xmath23 obtained in ref. @xcite in the case of gaussian disorder (point - like long - range - correlated defects) are given in table [tab1] for the impurities concentrations @xmath45, @xmath46. it was stated in the paper, that for the non - gaussian disorder (lines of defects) the estimates of the exponents are comparable with analytic calculations of weinrib and halperin @xcite. in ref. @xcite the mc simulation of the critical behavior of the three - dimensional ising model with long - range - correlated disorder at criticality was performed by means of the short - time dynamics and the single - cluster wolff method. randomly distributed defects had a form of lines and resembled the `` non - gaussian disorder '' of ref. however, in contrast to ref. @xcite, a condition of lines mutual avoidance was implemented. according to ref. @xcite, a situation when crossing of the lines of defects is allowed is not described by the weinrib - halperin model. in the short - time critical dynamics method, the concentration of spins was chosen @xmath45 and the cubic lattices of sizes @xmath47 were considered. as a mc method was used a metropolis algorithm. resulting numbers for the exponents are given in the 5th row of table [tab1]. additional simulations in the equilibrium state were performed in ref. @xcite to verify the reliability of results obtained by means of the short - time critical dynamics. the single - cluster wolff algorithm was used for simulation and a finite size scaling for evaluation of the critical exponents. one mcs was defined as 5 cluster flips, @xmath48 mcss were discarded for equilibration and @xmath49 for measurement. disorder averaging was typically performed over @xmath48 samples. again, the procedure seems to be safe. the values of critical exponents obtained in the simulations are quoted in the 6th row of table [tab1]. with the above information at hand we started our numerical analysis as explained in the forthcoming sections.
Model and observables
we consider a @xmath18 ising model with non - magnetic sites arranged in a form of randomly oriented lines (see fig. [fig_1]). the hamiltonian reads : @xmath50 where @xmath51 means the nearest neighbour summation over the sites of a s.c. lattice of linear size @xmath52, @xmath53 is the interaction constant, ising spins @xmath54, @xmath55 is the occupation number for the @xmath56-th site and periodic boundary conditions are employed. non - magnetic sites (@xmath57) are quenched in a fixed configuration and form lines, as shown in fig. [fig_1]. concentration @xmath58 of the magnetic sites is taken to be far above the percolation threshold. [fig_1] shows one possible configuration for non - magnetic impurity lines. during the simulations, one generates a large number of such disorder realizations this is done in our study using the following procedure. the lines of impurities are generated one by one, each along randomly chosen axis and the final disorder realization is accepted with the probability @xmath59 using rejection method. here @xmath60 is a required value for the concentration of magnetic sites and @xmath61 is the dispersion of the resulted gaussian distribution in @xmath58 (see fig. [hist]). the average value for the impurity concentration is equal to @xmath62. in the simulations presented here we paid special attention to the width of this distribution and consider both broad and narrow cases, see sections [ivb] and [ivc]). the observables saved during each step of the mc simulations for a given disorder realization will be the instantaneous values for the internal energy @xmath63 and magnetisation @xmath64 per spin, defined as @xmath65 @xmath66 where number of magnetic sites is @xmath67, total number of sites is @xmath68. using the histogram reweighting technique @xcite we compute the following expectation values at the critical temperature region for a given disorder realization : @xmath69 as functions of an inverse temperature @xmath70. in ([99]) the angular brackets @xmath71 stand for the thermodynamical averaging in one sample of disorder. for a given disorder configuration, one can evaluate the temperature behaviour of the magnetic susceptibility @xmath72 and magnetic cumulants @xmath73, @xmath74 : @xmath75 in order to refer to the physical quantities, the observables are to be averaged over different disorder configurations, denoted hereafter by an overline : @xmath76. two ways of averaging can be found in the literature, which we will consider for the case of finding the maximum value for the susceptibility. at each disorder realization an individual curve for the susceptibility @xmath72 as a function of the temperature @xmath77 is obtained (using histogram reweighting technique). in the first method of averaging, depicted in fig. [fig_11],a the maximal values of each individual curves are averaged over all disorder realizations. this type of averaging, referred hereafter as _ averaging a _, will be denoted as @xmath78. \a) _ averaging a _ b) _ averaging b _ an alternative method is shown in fig. [fig_11],b : at first one performs a configurational averaging of the susceptibility, i.e. the single averaged curve is evaluated for @xmath79. then the maximal value for this curve, @xmath80, is found. hereafter, this method will be referred as _ averaging b_. both _ averaging a _ and _ averaging b _ will be exploited in our analysis. to evaluate the critical exponents, a number of characteristics will be used with known finite size scaling : * _ averaging a _ : @xmath81 here, @xmath78, @xmath82, and @xmath83 are the averaged over disorder configurations maximal values of the magnetic susceptibility and magnetic cumulants temperature derivatives. the value @xmath84 is calculated for the temperature that corresponds to the magnetic susceptibility maximum. * _ averaging b _ : + @xmath85 + @xmath86 + where @xmath87 note, that quantities ([111a]) are ill - defined if _ averaging a _ is considered. moreover, as one can see from the definitions ([22a]), ([22b]) there are different ways to define magnetic cumulants for the disordered system (see refs. @xcite for more discussion). we performed our numerical simulations using the swendsen - wang cluster algorithm. the main reason is, that with the diluted system one often performs simulations at the temperature which is far from the native phase transition point of each particular disorder realization. as the result, the single clusters can be of rather small size and the wolff one algorithm, used by us before @xcite is found to be less effective. the following set of lattice sizes @xmath88 (@xmath89 in some cases) is used and performing the finite - size - scaling analysis is performed. the magnetic site concentration is chosen to be @xmath45 which is far from the percolation threshold and this also allows comparison with the previous simulations @xcite. another reason for such a choice is that for the 3d ising model with uncorrelated impurities the correction to scaling terms were found to be minimal at @xmath45 @xcite, (obviously, this does not mean that @xmath45 minimizes correction - to - scaling terms at any level of impurities correlation). the number of samples ranges from @xmath90 to @xmath48 for all lattice sizes. the simulations are performed at the finite - size critical temperature @xmath91 which is found from the maximum of magnetic susceptibility @xmath72 at each lattice size @xmath52 and the reweighting technique for the neighbouring temperatures is used. note, that in refs. @xcite all simulation were performed at the critical temperature of an infinite system. similarly to our other previous papers @xcite we start simulation from estimating the critical temperature of a finite - size system as a function of the lattice size @xmath52, @xmath91. for the smallest lattice size @xmath92, the preliminary simulation is performed at @xmath93 first (@xmath94 is the critical temperature for the pure system). the @xmath95 is located then from the maximum of the susceptibility. at the same time, we estimate various autocorrelation times to control the error due to time correlations. for the next larger lattice size, e.g. @xmath96, preliminary simulation is performed at @xmath95 and then @xmath97 is located again from the maximum of the susceptibility at @xmath96. the process is repeated for all lattice sizes. in this way we estimate @xmath91 and the energy autocorrelation time @xmath98. the results are given in table [tab2]. one can see that the finite size system critical temperature obtained via _ averaging a _ differs from those obtained via and _ averaging b_. the difference is not too large, the next simulations were performed at the critical temperature obtained by _ averaging a_. in evaluation of the critical indices we followed the standard finite - size - scaling scheme as described e. g. in ref. the estimates for the correlation length, magnetic susceptibility, and spontaneous magnetization critical exponents @xmath20, @xmath21 and @xmath22 can be obtained from the fss behaviour of the following quantities : * _ averaging a _ : @xmath99 in eqs. ([111b]), ([111c]), the correction - to - scaling terms have been omitted. taking them into account the fss behaviour of the quantity @xmath100 attains the form : @xmath101 where @xmath102 is the leading exponent, @xmath103 is the correction - to - scaling exponent, and @xmath104 is a correction - to - scaling amplitude. hereafter, we perform the finite - size - scaling analysis in several ways. first, assuming the correction - to - scaling terms to be small we will fit the data for the observables to the leading power laws ([111b]), ([111c]) only. besides that, we will also perform the fits with correction - to - scaling terms ([111d]) as explained below.
Statistics of the observables. preliminary estimates of the critical exponents
in this section, applying the above described formalism, we give preliminary estimates of the critical exponents. as we will see, the estimates for the same exponent obtained on the base of fss of different quantities differ from each other. moreover, similar differences arise when the results obtained via _ averaging a _ and _ averaging b _ are compared. we will discuss possible reasons for such behaviour and will analyze them further in the forthcoming section. the starting point of the analysis performed in this section is connected with the fact that at the same concentration of impurities, different samples vary configurationally (geometrically) as well as in their thermodynamical characteristics. in systems with uncorrelated dilution these variations are also present (see e.g. ref. @xcite) although, as we will see below, these are less pronounced than in the case of the long - range - correlated disorder. moreover, although the mean concentration of magnetic sites is well - defined for an ensemble samples (and chosen to be @xmath45), the concentration of magnetic sites in each sample vary from the mean one. such deviations are caused by possible intersections of impurity lines in different samples, besides not for all lattice sizes one can reach given concentration of impurities by an integer number of lines. therefore, to start an analysis one has to make a choice of a set of sample specifying allowed values of the concentration dispersion. we start an analysis by dealing with the set of samples that are characterized by dispersion of concentrations @xmath105. the sample - to - sample variations increase with the increase of the lattice size. as an example, we show in fig. [fig2] the values of the susceptibility for a given disorder realization calculated via _ averaging a _ (@xmath78, fig. 3a-3c) and _ averaging b _ (@xmath106, fig. 3d-3f) for three typical lattice sizes @xmath107. each point in these plots represents the data obtained for a given sample (i.e. given disorder realization). solid lines are made up of the average values, where at each @xmath108 the averaging is done on the interval @xmath109. one can see that the averages saturate at @xmath108 larger than two hundred samples. from this in particular one can conclude that considered here number of samples @xmath110 is quite sufficient for further analysis. one observes that for some samples @xmath78 and @xmath106 differ significantly from the average values. when this is the case, we will call such an event a rare one. the values that are close to the maximal one will be referred to as typical events. we would like to stress that for _ averaging b _ for the lattice size @xmath111 most of rare events produce the values for the susceptibility below the average value, while for @xmath112 the situation is reversed, see figure [fig2]. at @xmath113 the distribution of rare events is aproximately symmetric. the situation is more homogenous for the _ averaging a_. the averaging over too small number of disorder realizations leads to typical (i.e. most probable) values instead of averaged ones. indeed, as can be seen in fig. [fig3] for @xmath112, the probability distributions of @xmath78 and @xmath106 possess long tails of rare events with larger values of susceptibility. the samples which correspond to the rare events give a large contribution to the average and shift it far from the most probable values. the value of @xmath106 calculated with the complete probability distribution in shown in fig. [fig3] by a dashed line (blue on - line). another characteristic value shown in fig. [fig3] by lines (solid) are the median values @xmath114, defined as the values of @xmath72 where the integrated probability distributions are equal to 0.5. a distance between the average and median values may serve to quantify the asymmetricity of the probability distribution. figures [fig5a] and [fig6a] show contributions of typical and rare events to the temperature behaviour of the observables under discussion. as an example of typical events, we plot in fig. [fig5a] the magnetic susceptibility @xmath72, binder s cummulant @xmath74, and magnetisation @xmath115 selecting the samples where susceptibilities have the same values as @xmath116 at @xmath117 for the largest lattice size @xmath112. few examples of rare events corresponding to large values of @xmath72 are shown in fig. [fig6a]. the bold lines in the figures show averages over all disorder realizations (all samples). the thin lines (color on - line) correspond to different samples, as shown in the legends to the plots [fig5a]a, [fig6a]a. it follows from the above analysis that the rare events are obviously present in the random variables distributions and give a contribution to the average values of physical quantities. law to choose the samples during the mc experiment (as it was done at the initial phase of our analysis, see ref. @xcite) modifies the distribution. such modifying leads to different results for the exponents. we thank the referee for the comments on this point.] before passing to evaluation of the critical exponents let us recall that as it was noted at the beginning of section [v] we perform the simulations for the samples with varying concentration of magnetic sites, keeping the mean concentration equal to @xmath45 and limiting dispersion of concentrations by @xmath105 ([probp]). following refs. @xcite it is natural to call such situation the grand canonical disorder. afterwards, in section [ivc] we will consider the case of canonical disorder, when we will choose the samples much more close in the concentration of magnetic sites (taking @xmath118). in the finite - size - scaling technique @xcite, the critical exponents are calculated by fitting the data set for the observables evaluated at various lattice sizes to the power laws ([111b]), ([111d]) as was already mentioned in section [iii]. this procedure will be used for the evaluation of the critical exponents @xmath20, @xmath22, and @xmath21 based on the observables measured during the mc simulations. corresponding finite - size - scaling plots are shown in figs. [fig5], [fig7], the numbers are given in table [tab3]. in fig. [fig5], we give the log - log plots of the maximum values of the configurationally averaged derivatives of binder cumulants @xmath119 and @xmath120. the scaling of these quantities is governed by the correlation length critical exponent @xmath20, see eqs. ([111b]). [fig7] shows the size dependence of the configurationally averaged magnetic susceptibility @xmath78 (fig. [fig7]a) and magnetisation @xmath84 (fig. [fig7]b). these values are expected to manifest a power - law scaling with exponents @xmath121 and @xmath122, see eq. ([111b]). | we analyze a controversial topic about the universality class of the three - dimensional ising model with long - range - correlated disorder. whereas both theoretical and numerical studies agree on the validity of extended harris criterion (a. weinrib, b.i.
halperin, phys.
rev.
b 27 (1983) 413) and indicate the existence of a new universality class, the numerical values of the critical exponents found so far differ essentially. to resolve this discrepancy
we perform extensive monte carlo simulations of a 3d ising model with non - magnetic impurities being arranged in a form of lines along randomly chosen axes of a lattice.
the swendsen - wang algorithm is used alongside with a histogram reweighting technique and the finite - size scaling analysis to evaluate the values of critical exponents governing the magnetic phase transition.
our estimates for these exponents differ from both previous numerical simulations and are in favour of a non - trivial dependency of the critical exponents on the peculiarities of long - range correlations decay. , , , random ising model, long - range - correlated disorder, monte carlo, critical exponents 05.10.ln, 64.60.fr, 75.10.hk | cond-mat0611568 |
Introduction
the development of coherent electromagnetic sources of higher and higher intensity with increasing photon energy up to the hard x - ray range motivates the theoretical study of the change of the processes of strongly bound systems, such as nuclear processes, by these intense fields ledingham. in this paper, the change of the nuclear photoeffect due to the presence of an intense coherent electromagnetic field is studied. this process is analogous to the laser - assisted x - ray photo effect (x - ray absorption), a process which was already discussed @xcite in the late 80 s taking into account gauge invariance @xcite, @xcite. the laser - assisted nuclear photoeffect (lanp) and the laser - assisted x - ray photo effect (x - ray absorption) are laser - assisted bound - free transitions. the difference between them lies in the charged particle (proton or electron, respectively) which takes part in these processes. although the lanp was recently investigated far from the threshold and neglecting the effect of the coulomb field of the remainder nucleus @xcite, in the case of the laser - assisted x - ray absorption processes it was found that the most interesting changes due to the presence of the laser field appear near the threshold @xcite, @xcite. thus, applying the results of @xcite, the lanp is reexamined in a gauge invariant manner and near the threshold, where the hindering effect of the coulomb field of the remainder nucleus is very large so that it must be taken into account. the effect of the coulomb field of the remainder nucleus on the transition rate is approximately taken into account. the laser - modified differential cross section is compared to the laser - free differential cross section, and it is shown that their ratio does not depend on nuclear parameters in the two types of initial nuclear states investigated and on the state of polarization of the @xmath0 radiation, but it has only a laser parameter dependence. the process investigated can be symbolically written as @xmath1where @xmath2 denotes the target nucleus of mass number @xmath3 and of charge number @xmath4. the target nucleus absorbs a @xmath0 photon symbolized by @xmath5, and @xmath6 laser photons take part in the process which is symbolized by @xmath7. @xmath8 and @xmath9 correspond to @xmath10 laser photon emission and absorption, respectively. as a result, a free proton @xmath11 is emitted and the remainder nucleus is @xmath12. the calculation is made in the radiation @xmath13 gauge, and in the long wavelength approximation (lwa) of the electromagnetic fields, the recoil of the remainder nucleus and the initial momentum carried by the laser and @xmath0 fields are neglected. in the case of a circularly polarized monochromatic wave for the vector potential of a laser field, @xmath14 $] is used. @xmath15 is the angular frequency of the laser. the amplitude of the corresponding electric field @xmath16. the frame of reference is spanned by the unit vectors @xmath17, @xmath18 and @xmath19. the vector potential describing the gamma radiation is @xmath20, with @xmath21 the energy and @xmath22 the unit vector of the state of polarization of the @xmath0 photon, and @xmath23 the volume of normalization.
Gauge invariant s-matrix element of laser-modified proton emission in nuclear photoeffect
it is shown in @xcite that the electromagnetic transition amplitudes of a particle (proton) of rest mass @xmath24 and of charge @xmath25 in the presence of a laser field are determined by the matrix elements of the operator @xmath26 with the eigenstates of the instantaneous energy operator @xmath27 in both (@xmath28 and @xmath29) gauges. (@xmath25 is the elementary charge and the superscript @xmath30 refers to the gauge.) accordingly, the gauge - independent s - matrix element can be written as@xmath31where @xmath32 and @xmath33 are the initial and final states of the proton in the same gauge and @xmath34 is the reduced planck constant. our calculation is carried out in the radiation @xmath35 gauge because of the choice of the final state of the proton (see below). the initial state of the proton has the form @xmath36where @xmath37 is a stationary nuclear state of separation energy @xmath38 of the proton. the @xmath39 factor, where@xmath40, appears because of gauge transformation since @xmath41 is the eigenfunction of the instantaneous energy operator, @xmath42 in the @xmath28 gauge. @xmath43 is the nuclear potential and @xmath44 is the coulomb potential felt by the proton initially, and the superscript @xmath45 refers to the @xmath28 gauge. the modification of the initial state due to the laser field is neglected since the direct effect of the intense laser field on the nucleus has been found to be negligible @xcite at the laser parameters discussed. it is also supposed that the initial nucleus does not have an excited state which is resonant or nearly resonant with the applied @xmath0 radiation. if similarly to @xcite the modification of the final state due to the strong interaction is neglected, then in the final state and in the @xmath29 gauge the instantaneous energy operator @xmath46 reads @xmath47where the superscript @xmath48 refers to the radiation @xmath35 gauge and @xmath49 is the coulomb potential of the remainder nucleus. an approximated solution of @xmath50, i.e. an approximated time dependent state of a particle in the laser plus coulomb fields, is the coulomb - volkov solution of a proton of wave number vector @xmath51 @xcite, @xcite : @xmath52 here @xmath53 is the coulomb function, i.e. the wave function of a free proton in a repulsive coulomb field of charge number @xmath54, @xmath23 denotes the volume of normalization, @xmath55 is the relative coordinate of the two particles. @xmath56), \label{hyperg}\]]where @xmath57is the sommerfeld parameter, with @xmath58 the fine structure constant, and it is supposed that @xmath24 is much less than the rest mass of the remainder nucleus. @xmath59 is the confluent hypergeometric function and @xmath60 is the gamma function @xcite. the function @xmath61 \label{ft}\]]where @xmath62here the polar angles of the wave number vector @xmath63 of the outgoing proton are @xmath64 and @xmath65, i.e. they are the polar angles of the direction in which the proton is ejected. in the low - energy range (@xmath66, where @xmath48 is the radius of a nucleon) and for @xmath67 the long wavelength approximation yields @xmath68 -1 } }, \label{cb2}\]]which is the square root of the so - called coulomb factor. (the coulomb factor @xmath69 describes well e.g. the coulomb correction to the spectrum shape of beta decay @xcite.) for the final state of a proton of wave number vector @xmath63, the lwa of the nonrelativistic coulomb - volkov solution @xmath70 is used. @xmath71with @xmath72, that is the energy of the outgoing proton in the intense field, where @xmath73 is the ponderomotive energy. substituting @xmath74, @xmath75 into @xmath76 and using @xmath77 one can obtain the following form of the gauge - independent s - matrix element as @xmath78f^{\ast } (t)\frac{\partial } { \partial t}g\left [\overrightarrow{q}\left (t\right) \right] dt, \label{sfi}\]]where @xmath79is the fourier transform of the initial stationary nuclear state @xmath37 of the proton and @xmath80(equation @xmath81 can be obtained directly with the aid of eq.(27) of @xcite.) using the @xmath82 identity@xmath83with @xmath77, i.e. @xmath84. the @xmath85 term of the last factor of @xmath86 can be neglected if the pure intense field induced proton stripping process is negligible since this term describes the process without the gamma photon. furthermore, the ratio of the amplitudes of @xmath87 and @xmath88 equals @xmath89. therefore the @xmath90 approximation is justified to use, where @xmath91. the relative strength of the @xmath92 and @xmath93 terms is characterized by the parameter @xmath94. in the laser free case @xmath95, where @xmath96 is the difference of the photon energy and the proton separation energy. numerical estimation shows that @xmath97 near @xmath98 @xmath99 used here and in the case of laser photon energy and intensity values discussed. therefore the @xmath100 term is the leading one in the last factor of @xmath86. as to the radiation field dependence of @xmath101, the effect of @xmath102 is negligible in @xmath103 and thus @xmath104. it was shown above that the amplitude of oscillation of @xmath105 due to the intense field can be neglected. therefore @xmath106 can be used in @xmath101, and @xmath107 results @xmath108 using the jacobi - anger formula in the fourier series expansion of @xmath109 @xcite the s - matrix element can be written as@xmath110i}{v}\chi _ { c}(q)\times \label{sfi2 } \\ & & \times \left (\frac{\partial } { \partial q}g\right) _ { q = q}\frac{e\sqrt{2\pi \hbar \omega _ { \gamma } } } { \hbar } \xi j_{n}(\alpha) e^{-in\eta _ { 0 } }, \notag\end{aligned}\]]where @xmath111, @xmath112 is a bessel function of the first kind, and @xmath113 the terms, that are small if the @xmath114 condition is fulfilled (@xmath115 and @xmath116 see below), were neglected in the calculation.
Gauge invariant differential cross section of laser-modified proton emission in nuclear photoeffect
in this paper two cases of initial nuclear energy @xmath117 of the initial state having a different type of space - dependent part @xmath118 are considered in order to show the general nature of the effect of the laser on the process. the one case is the @xmath119 one - proton halo isotope of separation energy @xmath120 @xmath121 firestone and of initial state @xmath122, with @xmath123 and @xmath24 the rest mass of the proton @xmath124. although the proton rest mass is more than 12 % of the total rest mass of @xmath119 and the approximation, the fact that @xmath24 is much less than the rest mass of the remainder nucleus, is not very good, but following @xcite we investigate @xmath119. in the other case the initial state is the @xmath125 shell model state @xcite of the form @xmath126, with @xmath127 ^{1/2}/\gamma (\frac{3}{2})$] where @xmath128 is the quantum number of the nuclear shell model and @xmath129 denotes the gamma function. @xmath130 is the confluent hypergeometric function, @xmath131, @xmath132, @xmath24 is the nucleon rest mass and @xmath133 is the shell model angular frequency (@xmath134 @xmath121, @xmath135 is the nucleon number @xcite, and @xmath136 @xmath137). the differential cross section of lanp has the form@xmath138where @xmath139 is the differential solid angle around the direction of the outgoing proton. @xmath140 is the smallest integer fulfilling the @xmath141 condition and @xmath142 is the same in both cases of the initial state. (the cases @xmath8 and @xmath9 correspond to @xmath10 laser photon emission and absorption, respectively.) the partial differential cross section @xmath143with@xmath144 _ { q = q_{n}}^{2}.\]]here @xmath145 } /\hbar $], @xmath146, @xmath147 is the reduced compton wavelength of the proton and @xmath148.@xmath149@xmath150 is a bessel function of the first kind, with @xmath151. in the variable @xmath152 the coulomb factor @xmath153 reads as@xmath154 -1\right] }, \label{fjk}\]]where @xmath155 with @xmath156 the charge number of the remainder nucleus. the coulomb factor causes a strong hindering of the effect in both the laser assisted and laser free cases. near the threshold @xmath157 the @xmath158 _ { q_{n}}^{2}\varpropto \varepsilon _ { n}^{2}$] in the case of the two types of initial state discussed and @xmath159here @xmath160 is constant, it depends on the form of the initial state and@xmath161 in the laser free case @xmath162 using @xmath163, @xmath164 and @xmath165 at @xmath166, the differential cross section near above the threshold@xmath167with @xmath168 ^{-1}$].
Numerical results
the ratio @xmath48 of the laser - assisted and the laser - free differential cross sections, @xmath169where @xmath170 [@xmath171 divided by @xmath172, equals the ratio of the rates of the corresponding processes in an elementary solid angle in a given direction of the outgoing proton. the rate of change in one channel, @xmath173with @xmath174 in the variable @xmath152. so @xmath175 and @xmath48 describe the change caused by the intense coherent field independently of the state of @xmath0 polarization and the initial states applied. in our numerical calculation, the laser photon energy @xmath176 @xmath177. first the case in which the outgoing proton moves in the plane of polarization of the laser beam (@xmath178 @xmath179) is investigated in the case of @xmath180. figure 1 shows the laser photon number dependence of @xmath175 at @xmath98 @xmath99 with laser intensity @xmath181 @xmath182. the intensity dependence of @xmath48 has been investigated with @xmath98 @xmath99. @xmath48 increases linearly from @xmath183 at @xmath184 @xmath182 up to @xmath185 at @xmath181 @xmath182. figure 2 depicts the @xmath142 dependence of @xmath48 with @xmath181 @xmath182. figure 3 shows the @xmath64 dependence of @xmath175 in the case @xmath181 @xmath182 at (a) @xmath186 (b) @xmath187 (c) @xmath188 and (d) @xmath189. finally fig. 4 depicts the @xmath64 dependence of @xmath175 at @xmath188 with different laser intensities : (a) @xmath190 @xmath182 (b) @xmath191 @xmath182 (c) @xmath181 @xmath182. the numerical calculation in all the cases discussed above has been repeated at @xmath192, and negligible change has been found.
Discussion and summary
to compare our results and the results of @xcite we investigate our formulas in the @xmath193 limit. in this case @xmath194, @xmath195, @xmath196and @xmath197thus in this limit the @xmath6 dependence disappears from the argument of the bessel function and @xmath198 the cross sections obtained in @xcite are symmetric in @xmath6 around @xmath199 (see the figures of @xcite) and the total cross section is asserted to be unaffected by the laser radiation. this corresponds to @xmath200. in contrast, our result is significantly asymmetric in @xmath6 (see fig. 1.), which is a consequence of the @xmath152 dependence of @xmath175 [see @xmath201. the change (increase) of the kinetic energy of the proton is manifested in the increase of @xmath152 from @xmath202 up to @xmath203. the increase of @xmath152 with increasing @xmath6 causes an asymmetry in the @xmath6 dependence of fig. the sum of the changes (increments) in the different channels results in a moderate increment of @xmath48 (@xmath204) as can be seen in fig. 2. summarizing, one can say that near the threshold, @xmath48 which measures the change of the rate of the lanp, has minor @xmath142 and intensity dependence, and negligible @xmath54 dependence. furthermore, @xmath175, which is the rate of change in one channel (at a definite laser photon number), has a significant laser photon number and @xmath64 dependences. @xmath48 and @xmath175 are the same in the cases of the two different initial states considered. since the numerical results obtained seem to be independent of the initial nuclear states chosen, it can be expected that @xmath48 and @xmath175 have a minor dependence from the form of the initial state in general. regarding the experimental situation at such a high intensity, it is hard to distinguish photo - protons from background protons that arise as a result of interaction with the hot electron plasma created by the intense laser field. we also have to mention that in an experiment the gamma ray pulse from an accelerator must be synchronized with an intense (e.g. attosecond) pulse of a laser system. moreover, in the case of @xmath119, which has a short half - life of about @xmath205 @xmath206, the @xmath119 nuclei must be created in situ in the laser beam by a nuclear reaction. fortunately, most of the heavier nuclei have protons of @xmath125 shell model state (the other case investigated) in their stable ground state. the wispy target determined by the focal spot of the focused intense laser beam, the low repetition rate of the laser system, and the angular resolution of the proton detector together make it very challenging to carry out a successful near - threshold, laser - modified proton emission experiment that could produce significant counting statistics. | the change of the probability of proton emission in the nuclear photoeffect due to an intense coherent (laser) field is discussed near the threshold, where the hindering effect of the coulomb field of the remainder nucleus is essential.
the ratio of the laser - assisted and laser - free differential cross section is deduced and found to be independent of the polarization state of the @xmath0 field and the two types of initial nuclear state considered.
the numerical values of this ratio are given at some characteristic parameters of the intense field. | 1303.1268 |
Introduction
chiral perturbation theory is a low energy effective field theory of the strong interaction. the work @xcite presents analytic expressions for the two - loop contribution to the pion mass and decay constant in su(3) chiral perturbation theory with suitable expansions in powers of @xmath3. in an upcoming work @xcite, we will present analogous expressions for the pion decay constant. work is also underway to find similar simple analytic representations for the kaon and eta mass and decay constants to two loops. due to the goldstone nature of the particles involved, scalar, tensor and derivatives of sunset diagrams appear in these calculations, with various mass configurations and with up to three distinct masses. much work has been done on sunset diagrams (an incomplete list is given in references @xcite-@xcite), and a variety of analytic results exist in the literature for the one - and two - mass scale configurations @xcite. papers directly relevant to this work are the following. in @xcite, analytic results have been given for the master integrals at the pseudothreshold @xmath4 and threshold @xmath5, the former of which may be used to obtain the single, and many of the double, mass scale analytic expressions. gasser and saino @xcite use integral representations to give results in closed form for several basic two - loop integrals appearing in chpt, including the sunset, with one mass - scale. for unequal masses, fully analytic results are given by @xcite gives in terms of newly defined elliptic generalizations of the clausen and glaisher functions, but the application of methods or approximation schemes that give the three mass scale sunsets as expansions in powers of the mass ratio allow for a more transparent interpretation of the results being considered. in @xcite, just such an expansion is given for the most general sunset integral in terms of lauricella functions. however, none of the series presented in @xcite converge for the physical values of the meson masses. the interest in analytic or semi - analytic expressions arises from the desire to make as direct a contact as possible with results in lattice field theories. recent advances in lattice qcd now allow for quark masses in these theories to be varied independently, allowing for realistic quark masses. the availability of analytic results for pseudo - scalar masses and decay constants, for example, would allow for easy and computationally efficient comparison with lattice results. aside from the derivation of analytic expressions for the pseudo - scalar meson masses and decay constants to two - loops, the application of sunset diagrams to chiral perturbation theory is also of general interest. in this context, sunset diagrams have been studied quite early (@xcite), where not only the single mass scale sunset (which appears in su(2) chiral perturbation theory) is considered, but also the cases with more than one mass scale which are common in the su(3) theory. in su(3) chiral perturbation theory, the sunset is the simplest diagram that appears at two loops, and a careful study of it paves the way for the study of the other diagrams that appear at this order (i.e. vertices, boxes and acnodes). the work @xcite gives a terse but comprehensive summary of results. another possible use of the sunsets is to expand them out using methods such as expansion in regions @xcite, and then use this to reduce the su(3) low energy constants to the su(2) ones. the process of relating the su(3) to su(2) low - energy - constants has been done using an alternative method in @xcite but it has not yet been done for the full set of low - energy - constants at next - to - next - to - leading order. it must be noted in the context of @xcite that the sunset technology is also important when considering vertices, as many of the latter get related to the sunsets when using, for example, the method of expansion by regions. in this paper, we use the mellin - barnes method to derive results for all the single and double mass scale integrals. it has been shown in @xcite that the mellin - barnes method is an efficient one for obtaining expansions in ratios of two mass scales should they appear in feynman diagrams in general. this work therefore serves as an independent verification of the existing results in the literature. the mellin - barnes method is also an appropriate tool for chiral perturbation theory applications as it ab initio allows us to express the integrals as expansions in mass ratios. a further reason for mellin - barnes as our tool of choice is the availability of powerful public computer packages in this approach. the availability of such codes has made such a study of sunsets (and two - loop diagrams in general) in chiral perturbation theory much more accessible. the mathematica based package ` tarcer ` @xcite applies the results of tarasov s work @xcite to recursively reduce all sunset diagrams to the master integrals. several packages @xcite have automatized many aspects of the application of mellin - barnes methods to feynman integrals. the sunsets appearing in chiral perturbation theory have been implemented numerically in the package ` chiron ` @xcite using the methods of @xcite. one of the goals of the present work is to improve on this implementation. in addition, there are two other packages ` bokasum ` @xcite and ` tsil ` @xcite that can be used to numerically calculate sunset integrals. we present along with this paper several mathematica notebooks (lodged as ancillary files along with the arxiv submission) which contain the details of our calculations, as well as a demonstration of how to apply the above packages to the calculation of sunset integrals. the notebooks are thoroughly annotated, and can be used in a stand - alone capacity, or in conjunction with this note. these may also serve as a pedagogical introductions to the analytic evaluation of sunsets diagram. the primary goal of this paper is to show the use of the packages of @xcite ; the results as presented here have been checked in a number of other ways as well. the relations from @xcite have been implemented independently using ` form ` @xcite. the expansions around @xmath6 were also derived using the methods of @xcite and numerical results have been compared with the results from analytical expressions of @xcite. this paper is organized as follows. in section 2, we give the five different sunset configurations that will be explicitly considered in this work, and show from where they arise. in section 3 we give an overview of the sunset integrals, their divergences, and their renormalization in chiral perturbation theory. in section 4, we briefly discuss the mellin - barnes method of evaluating feynman integrals. in section 5, we demonstrate the use of the package ` tarcer ` @xcite to reduce the tensor and derivatives of the sunsets to master integrals. in section 6, we explain the use of the packages @xcite to derive the results for the one - mass scale master integral. we also explain how the ` tarcer ` package @xcite alone can be used to derive this result. in section 7, we describe briefly the two different categories of two - mass scale sunset diagrams and their evaluation, and present a complete set of results in appendix a. in section 8, we explain how three mass scale sunsets can be handled either by means of an expansion in the external momentum, or by a more sophisticated application of the mellin - barnes method to. in section 9, we present a one - dimensional integral representation of an important configuration that arises in the su(2) chiral perturbation theory, and in section 10 with a discussion of some numerical issues of the new results presented herein. we conclude in section 11 with a discussion of the import and limitations of this work, and possible future work in this field. in appendix b, we give a brief description of all the public codes used in this work, and in appendix c, we present a dictionary that allows for an easy translation between the definition used in this work for the sunset and other integrals, and those used in the various programs and papers. in appendix d, we list the ancillary files provided with this paper.
The meson masses and decay constants to two loops
expressions for the pseudoscalar meson masses and decay constants in two loop chiral perturbation theory are given in appendix a of @xcite. as a concrete example, the pion mass is given by : @xmath7 where @xmath8 is the bare mass, @xmath9 is the one - loop contribution, @xmath10 is the two - loop model - dependent counterterm contribution, and @xmath11 is the chiral loop contribution. it is in this last term that the sunset integrals appear : @xmath12 the @xmath13 in the above expression refer to the scalar sunset integral @xmath14 as defined in eq.([sunsetdef]) of section 3, where the first three arguments pertain to the masses entering the propagators, and the last is the square of the energy entering the loop. the @xmath15 and @xmath16 are the scalar integrals that make up the passarino - veltman decomposition of vector and tensor sunsets, and are defined precisely in eq.([h1]) and eq.([h21]) respectively. in the case of the meson decay constants, in addition to the variety of sunset integrals appearing above, also appear derivatives of the sunsets (i.e.@xmath17, @xmath18 and @xmath19). the work of finding an analytic expression for the pion mass (as well as the other pseudoscalar meson masses and decay constants) reduces to analytically evaluating these sunset integrals. in the subsequent sections of this paper, we explain how to analytically evaluate each of the different types of integrals appearing in expressions such as eq.([pionmass]) above. in particular, we show in detail how to evaluate the following integrals as representative of the different types of integrals and the different types of mass configurations that may appear in expressions for the pseudoscalar masses and decay constants : ''' '' * integral * & * characteristic * + ''' '' @xmath20 & one mass scale + ''' '' @xmath21 & two mass scales + ''' '' @xmath22 & three mass scales with smallest parameter as external momentum + ''' '' @xmath23 & three mass scales with an internal mass as smallest parameter + ''' '' @xmath24 & tensor sunset derivative + the evaluation of all these integrals requires writing them in terms of master integrals, and then analytically evaluating the master integrals. this is explained in greater detail in the next section. the analytic evaluation of the master integrals can be done using a variety of methods, and many of these have previously been used to derive the plethora of results that exist in the literature. in this paper, we use the mellin - barnes approach, which appears to be the most efficient method by which to evaluate the three mass scale integrals, such as @xmath23 that appears in the expressions for eta mass and decay constant. the integrals given in the table above are all amenable to a mellin - barnes treatment. however, for @xmath22, we instead take an expansion in the external momentum @xmath25, as it provides a result that is as accurate as a mellin - barnes expansion (to the same order) but that is much easier to calculate. a similar expansion can not be done for @xmath26 in either the external momentum @xmath27 due to poor convergence, or in @xmath3 as it gives rise to an infrared divergence.
Sunset integrals
the sunset integral, shown in fig.([sunset]), is defined as : @xmath28^{\alpha } [r^2-m_2 ^ 2]^{\beta } [(q+r - p)^2-m_3 ^ 2]^{\gamma } } \label{sunsetdef}\end{aligned}\]] vector and tensor sunset integrals have four - momenta, such as @xmath29 or @xmath30, sitting in the numerator. two tensor integrals that appear in the calculation of meson masses and decay constants in chiral perturbation theory are : @xmath31^{\alpha } [r^2-m_2 ^ 2]^{\beta } [(q+r - p)^2-m_3 ^ 2]^{\gamma } } \nonumber \\ \nonumber \\ h_{\mu \nu}^d \{m_1,m_2,m_3 ; p^2\ } = \frac{1}{i^2 } \int \frac{d^dq}{(2\pi)^d } \frac{d^dr}{(2\pi)^d } \frac{q_{\mu } q_{\nu}}{[q^2-m_1 ^ 2]^{\alpha } [r^2-m_2 ^ 2]^{\beta } [(q+r - p)^2-m_3 ^ 2]^{\gamma } } \nonumber\end{aligned}\]] these may be decomposed into linear combinations of scalar integrals via the passarino - veltman decomposition as : @xmath32 to obtain the scalar integral @xmath33, we take the scalar product of @xmath34 with @xmath35 : @xmath36^{\alpha } [r^2-m_2 ^ 2]^{\beta } [(q+r - p)^2-m_3 ^ 2]^{\gamma } } \equiv \frac{1}{p^2 } \langle \langle q.p \rangle \rangle \label{h1}\end{aligned}\]] where we have defined @xmath37 as the scalar sunset diagram with unit powers of the propagators, and with @xmath38 in the numerator. similarly, @xmath16 may be expressed as : @xmath39 in @xcite tarasov has shown by using the method of integration by parts that all sunset diagrams, including those of higher than @xmath40 dimensions, may be rewritten as linear combinations of a set of four master integrals and bilinears of one - loop tadpole integrals. these basic integrals are @xmath41 and the one - loop tadpole integral : @xmath42 application of tarasov s relations becomes crucial when evaluating another class of integrals that show up in chiral perturbation theory calculations, namely the derivatives of scalar and tensor sunsets (e.g. @xmath43). these may be evaluated by means of the following well - known formula relating derivatives and integrals in different dimensions @xcite : @xmath44 the mathematica package ` tarcer ` @xcite automatizes the reduction of any sunset integral to the master integrals. many results exist in the literature regarding these master integrals. one result that we use frequently in the subsequent sections is that of the two - mass scale master integral with zero external momentum @xmath45. this is given in @xcite as : @xmath46 - \frac{x}{2 } \text { ln}^2 \left [x \right] + \left (2 + x \right) \left [\frac{\pi^2}{12 } + \frac{3}{2 } \right] \bigg\ } \nonumber \\ & - (\mu^2)^{-2\epsilon } \bigg\ { m^2 \log \left (\frac{m^2}{\mu^2 } \right) \left [1 - \log \left (\frac{m^2}{\mu^2 } \right) \right] + 2 m^2 \log \left (\frac{m^2}{\mu^2 } \right) \left [1 - \log \left (\frac{m^2}{\mu^2 } \right) \right] \bigg\ } \nonumber \\ & + \frac{m^2}{2 } \bigg\ { \bigg [2 + x \bigg] \frac{1}{\epsilon^2 } + \bigg [x \left (1 - 2 \log \left (\frac{m^2}{\mu^2 } \right) \right) + 2 \left (1 - 2 \log \left (\frac{m^2}{\mu^2 } \right) \right) \bigg] \frac{1}{\epsilon } \bigg\ } + \mathcal{o}(\epsilon) \label{k111}\end{aligned}\]] where @xmath47, \qquad \sigma = \sqrt{1-\frac{4}{x } } \end{aligned}\]] eq.([k111]) above is given in the modified version of the @xmath48 scheme normally used in chiral perturbation theory (@xmath49). the change from the minimal subtraction (ms) scheme to @xmath49 involves making the replacement : @xmath50 analytical expressions for the divergent parts of the sunset master integrals have been derived in @xcite, amongst other places. the following are the divergent parts of the master integrals in the @xmath49 scheme : @xmath51 \frac{1}{\epsilon^2 } \nonumber \\ & + \left [m_1 ^ 2+m_3 ^ 2+m_3 ^ 2 - \frac{s}{2 } - 2m_1 ^ 2 \log \left(\frac{m_1 ^ 2}{\mu ^2}\right) - 2m_2 ^ 2 \log \left(\frac{m_2 ^ 2}{\mu ^2}\right)- 2m_3 ^ 2 \log \left(\frac{m_3 ^ 2}{\mu ^2}\right) \right] \frac{1}{\epsilon } \bigg\ } \nonumber \\ h_{\{2,1,1\}}^{div } & \{m_1,m_2,m_3;s\ } = \frac{1}{512 \pi^4 } \bigg\ { \frac{1}{\epsilon^2 } - \left [1 + 2 \log \left(\frac{m_1 ^ 2}{\mu ^2}\right) \right] \frac{1}{\epsilon } \bigg\ } \nonumber \\ h_{\{1,2,1\}}^{div } & \{m_1,m_2,m_3;s\ } = \frac{1}{512 \pi^4 } \bigg\ { \frac{1}{\epsilon^2 } - \left [1 + 2 \log \left(\frac{m_2 ^ 2}{\mu ^2}\right) \right] \frac{1}{\epsilon } \bigg\ } \nonumber \\ h_{\{1,1,2\}}^{div } & \{m_1,m_2,m_3;s\ } = \frac{1}{512 \pi^4 } \bigg\ { \frac{1}{\epsilon^2 } - \left [1 + 2 \log \left(\frac{m_3 ^ 2}{\mu ^2}\right) \right] \frac{1}{\epsilon } \bigg\ } \label{divparts}\end{aligned}\]] in the remainder of this paper, unless explicitly stated, @xmath52 will be used to denote the finite part of the sunset integral in the @xmath49 scheme. eq.([chiralmsbar]) may be reverse engineered and used in combination with eq.([divparts]) to find the full results in any other subtraction scheme.
The mellin-barnes method
we give a brief overview of the basic mellin - barnes approach to feynman integrals here. for a more comprehensive overview see @xcite. the mellin transform is defined as follows : @xmath53(s) = \int\limits_0^\infty f(t)t^{s-1}dt, \hspace{0.2 in } s \in \mathcal{c}\end{aligned}\]] its inverse is given by : @xmath54(x) = \frac{1}{2\pi i } \int\limits_{c - i\infty}^{c+i\infty } x^{-s}g(s)ds\end{aligned}\]] the following formula derived from the inverse mellin transform is used in high energy physics to write massive propagators as combinations of massless propagators : @xmath55 the expression obtained after application of this formula and evaluation of the momentum integral is known as the mellin - barnes representation of a feynman integral. in some cases, it may be possible to simplify the mellin - barnes representation of an integral by the application of the following two barnes lemmas @xcite : @xmath56 where @xmath57 the evaluation of the mellin - barnes integrals may then be performed either numerically, or analytically by the addition of residues. in case of multiple mellin - barnes parameters, results from the theory of several complex variables may have to be used for analytic evaluation @xcite.
Derivative and tensor sunsets: @xmath58
in this section, we demonstrate how to handle both the tensor sunset integrals, as well as the derivatives of the sunsets, by reducing them to master integrals. in particular, we show how to evaluate the integral @xmath59, by making extensive use of the package ` tarcer ` @xcite. the computer implementation of what follows is given in the ancillary file ` reductiontomi.nb `. the first step is to decompose @xmath60 into master integrals. from eq.([h21]), we have : @xmath61 differentiating with respect to @xmath62 gives : @xmath63 the next step involves evaluating the scalar sunset integrals with @xmath64 and @xmath65 in the numerator. the following command allows us to express the first of these integrals in terms of the master integrals. + ` tarcerrecurse[tfi[d, s, { 0, 0, 2, 0, 0 }, { { 1, mpi }, { 0, 0},{0, 0},{1, mk},{1, mk } }]] ` + the output, @xmath66, is a function of the dimensional parameter @xmath40, the external momentum @xmath62, the masses @xmath67 and @xmath68, the integrals @xmath69, @xmath70, @xmath71, @xmath72 and @xmath73. this expression is then differentiated with respect to @xmath62, the resulting expression, @xmath74, also being a function of the same parameters and integrals as @xmath66, but in addition also being a function of the differentiated master integrals @xmath75, @xmath76, @xmath77. each of these differentiated master integrals can be expressed as a sunset integral in a higher (@xmath78) dimension by use of eq.([dh]), and each of these higher dimensional sunsets can in turn be expressed in terms of the @xmath40 dimensional master integrals by further use of ` tarcer `. for example, the integral @xmath79 is equal to @xmath80. by use of the command : + ` tarcerrecurse[tfi[d+2, s, { { 3, mpi }, { 0, 0},{0, 0},{2, mk},{2, mk } }]] ` + we get an expression for @xmath81 in terms of @xmath40 dimensional master integrals. we repeat this process for each of the differentiated master integrals that appear, and substitute them (and @xmath25) into the expression for @xmath74. we can similarly obtain an expression for @xmath82 and @xmath83, and substituting all these expressions into eq.([h21prime]) with @xmath25 gives us our desired expression for @xmath59. the expressions we obtain for @xmath84 and @xmath19, given in the notebook ` reductiontomi.nb `, have been positively checked against expressions obtained from a direct differentiation of eq.(2.13) and eq.(2.14) respectively of @xcite.
Single mass scale sunset: @xmath85
all one mass scale case sunset integrals can be reduced to a single master integral, namely @xmath86 where @xmath87 is the mass in question. below, we show how to evaluate the one mass scale sunset integral @xmath88, and therefore give a pedagogical demonstration of the use of the mellin - barnes approach to evaluating feynman integrals. we also demonstrate the use of the public packages @xcite and @xcite. the accompanying mathematica notebook ` onemassmb.nb ` has a detailed computer implementation of what follows. we begin by applying eq.([mbformula]) to the definition of the sunset integral eq.([sunsetdef]), and then evaluating the integrals over the loop momenta, to get the following mellin - barnes representation : @xmath89 to make contact with results in the literature, we extract a factor of @xmath90. the above is also obtained automatically by use of the public code @xcite. the next step is to resolve (i.e separate) the singularities in @xmath91 and the finite part by shifting the contour across the points @xmath92 and @xmath93. this can be done in an automatic manner by use of the package @xcite. the result is an expression consisting of two terms : @xmath94 the first term contains the divergences, and the second piece is a finite one - fold contour integral which is to be evaluated by adding up residues. since the singularities in @xmath91 have been extracted, we can set @xmath91 to 0 in the second term. expressing the divergent piece as a laurent series around @xmath95, we get : @xmath96 the convergent piece is calculated by summing up the residues at the points @xmath97. the residues at non - zero integers @xmath98 for @xmath99 are given by : @xmath100 summing this up from @xmath101 to @xmath102 gives : @xmath103 the residue at @xmath104 = 0 is : @xmath105 combining the convergent and divergent pieces, we get the full result, expressed as a laurent series in @xmath91 : @xmath106 by pulling out a factor of @xmath107 and setting @xmath87 to 1, this can be expressed more succinctly as : @xmath108 this reproduces the result derived in eq.(13) of @xcite. the result given in eq.([gs1mass]) above is prior to the application of any subtraction procedure. the @xmath49 scheme may be applied by multiplying both terms of eq.([intmbdef]) by the additional factor : @xmath109^{2\epsilon}\end{aligned}\]] the inclusion of these two factors gives the following result for the @xmath49 subtracted single mass scale sunset integral : @xmath110 the ` tarcer ` package @xcite has the added functionality of performing a laurent series expansion in the small parameter @xmath111 for the master integrals. the command for such an expansion is : ` tarcerexpand [expression, d \rightarrow 4- 2\epsilon] ` for one mass - scale sunsets, using this feature, ` tarcer ` can be used directly to derive expressions for the integrals @xmath14, @xmath33, @xmath112, @xmath113, @xmath18, @xmath114, i.e. for all the sunset results that appear in @xcite. this has been demonstrated in the notebook ` onemasstarcer.nb `, in which is derived a very comprehensive set of relations with detailed annotations, and completely verifies all the sunset relations in @xcite. note that the ` tarcerexpand ` command has been found to work for all the cases of interest, since this is a pure single mass scale example. we find that for other more complicated mass configurations, including the case when we have a single mass scale with @xmath6, this command is unable to reproduce the laurent expansion of the integral. however, that ` tarcer ` can reproduce all the results for the sunsets in @xcite so efficiently indicates the power and utility of this package.
Two mass scale sunsets
there are eight possible independent mass configurations of the sunset master integrals with two masses. three of these fall into the pseudothreshold configurations, in which @xmath116. in the two - loop calculation of the pseudoscalar meson masses and decay constants, these are the only two - mass configurations that arise. results for the pseudothresholds, calculated directly using an integral representation of the sunsets, are given in @xcite. we rederived the three pseudothreshold results @xmath117, @xmath118 and @xmath119 using mellin - barnes representations, and expressions for these are given below : @xmath120\end{aligned}\]] @xmath121\end{aligned}\]] @xmath122\end{aligned}\]] where @xmath123. these results are valid for all real values of @xmath124. the other two mass pseudothreshold expressions may be obtained from the above by a simple re - ordering of the masses and indices. in the notebook ` twomasspt.nb `, we demonstrate the above calculations by means of the example @xmath125. the evaluation of non - pseudothreshold two mass sunset configurations results in three complications that do not arise in the pseudothreshold case. firstly, their mellin - barnes representation is a linear combination of complex - plane integrals of which at least one is two - fold, and which therefore requires a more sophisticated approach in its evaluation. these two - fold mellin - barnes integrals result in nested infinite sums, many of which can not be expressed as common analytic functions. therefore, completely analytic expressions for these integrals can not be obtained easily, and we are forced instead to take as many terms of these sums as yields the degree of accuracy we desire. secondly, the specific form of these infinite series depends on the numerical values of the two masses @xmath87 and @xmath126, or more specifically their ratio @xmath127. thirdly, there exists a range of values of @xmath128 for which it is not possible to use the mellin - barnes method (given the current state of the art) to evaluate these integrals. for these values of @xmath128, recourse must be had to other techniques, such as expansion in the external momentum @xmath62. the non - pseudothreshold mass configurations do not appear in the calculation of the pseudoscalar meson masses and decay constants to two - loops in chiral perturbation theory, but they may appear elsewhere. thus for completeness we provide results for these as well in appendix a. the notebook ` twomassresults.nb ` contains all the pseudothreshold and non - pseudothreshold two mass scale sunset integrals.
Three mass scale sunsets
three mass scale sunset integrals result in two - fold mellin barnes representations, which can be evaluated using the method of @xcite. however, for purposes of evaluating the pion mass and decay constant, we take an expansion in the external momentum @xmath62 : @xmath130 for the pion mass and decay constant the external momentum is always @xmath131, which is much smaller than the @xmath132 and @xmath133 that can appear in the propagators. therefore, the above series converges fairly fast, and only a few of higher order terms are required. for integrals with @xmath134 or @xmath27, the mellin - barnes approach may be more suitable. the derivatives of the integrals above can be evaluated using a combination of eq.(6) and ` tarcer ` @xcite. it turns out that derivatives to all orders of the sunset integral with @xmath6 can be expressed in terms of the single master integral @xmath135 given in eq.([k111]). [[two - fold - mellin - barnes - representations - hd_111-left - m_k2-m_k2-m_pi2-m_eta2-right]] two - fold mellin - barnes representations : @xmath136 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ for the three mass scale sunset integrals in which the external momentum is not the smallest parameter, such as those that appear in the kaon and eta masses and decay constants, the expansion in @xmath62 does not converge well. an expansion in one of the propagator masses must also be precluded as they lead to infrared divergences. the simplest method by which to obtain analytic expressions for these integrals to the order desired is by evaluating their two - fold mellin - barnes representation, a detailed explanation of which is given in @xcite. in this section, we list the main intermediate results in the evaluation of @xmath137 to exemplify the method in brief. the first step is to find the mellin - barnes representation of the integral @xmath137 and to resolve its singularity structure. this can be done semi - automatically by a combined use of the packages ` ambre.m ` and ` mb.m `. the result is a linear combination of four parts. the first consists of the divergent parts and the finite part containing the @xmath138-scale dependent logarithms. the second and third parts are one - fold mellin - barnes integrals, the evaluation of which can be performed by simply adding up residues up to the desired order in powers of the mass ratio. the fourth part is proportional to the two - fold mellin - barnes representation : @xmath139 where @xmath140, @xmath141, @xmath142, @xmath143. the singularity structure above gives rise to four distinct cones, i.e. the above integral will converge to four distinct expressions depending on the particular value of the mass ratios @xmath144 and @xmath145. these regions are given in table [3massconestable] and plotted in fig.([3massconesfigure]). g. amoros, j. bijnens and p. talavera, nucl. b * 568 * (2000) 319 [hep - ph/9907264]. f. a. berends, a. i. davydychev and n. i. ussyukina, phys. b * 426 * (1998) 95 [hep - ph/9712209]. j. gasser and m. e. sainio, eur. j. c * 6 * (1999) 297 [hep - ph/9803251]. j. van der bij and m. j. g. veltman, nucl. b * 231 * (1984) 205. doi:10.1016/0550 - 3213(84)90284 - 0 f. hoogeveen, nucl. b * 259 * (1985) 19. doi:10.1016/0550 - 3213(85)90295 - 0 a. i. davydychev and j. b. tausk, nucl. b * 397 * (1993) 123. g. weiglein, r. scharf and m. bohm, nucl. b * 416 * (1994) 606 doi:10.1016/0550 - 3213(94)90325 - 5 [hep - ph/9310358]. a. i. davydychev and v. a. smirnov, nucl. b * 554 * (1999) 391 doi:10.1016/s0550 - 3213(99)00269 - 2 [hep - ph/9903328]. s. laporta and e. remiddi, nucl. b * 704 * (2005) 349 doi:10.1016/j.nuclphysb.2004.10.044 [hep - ph/0406160]. o. v. tarasov, phys. b * 638 * (2006) 195 doi:10.1016/j.physletb.2006.05.033 [hep - ph/0603227]. m. caffo, h. czyz, s. laporta and e. remiddi, nuovo cim. a * 111 * (1998) 365 [hep - th/9805118]. m. caffo, h. czyz, s. laporta and e. remiddi, acta phys. b * 29 * (1998) 2627 [hep - th/9807119]. m. caffo, h. czyz and e. remiddi, nucl. b * 581 * (2000) 274 doi:10.1016/s0550 - 3213(00)00274 - 1 [hep - ph/9912501]. m. argeri, p. mastrolia and e. remiddi, nucl. b * 631 * (2002) 388 doi:10.1016/s0550 - 3213(02)00176 - 1 [hep - ph/0202123]. m. caffo, h. czyz, m. gunia and e. remiddi, comput. commun. * 180 * (2009) 427 doi:10.1016/j.cpc.2008.10.011 [arxiv:0807.1959 [hep - ph]]. f. jegerlehner and m. y. kalmykov, nucl. b * 676 * (2004) 365 doi:10.1016/j.nuclphysb.2003.10.012 [hep - ph/0308216]. l. adams, c. bogner and s. weinzierl, j. math. phys. * 56 * (2015) no.7, 072303 doi:10.1063/1.4926985 [arxiv:1504.03255 [hep - ph]]. p. post and j. b. tausk, mod. a * 11 * (1996) 2115 [hep - ph/9604270]. f. a. berends, m. buza, m. bohm and r. scharf, z. phys. c * 63 * (1994) 227. doi:10.1007/bf01411014 b. a. kniehl, a. v. kotikov, a. onishchenko and o. veretin, nucl. b * 738 * (2006) 306 [hep - ph/0510235]. m. y. kalmykov and b. a. kniehl, phys. b * 714 * (2012) 103 doi:10.1016/j.physletb.2012.06.045 [arxiv:1205.1697 [hep - th]]. s. groote, j. g. korner and a. a. pivovarov, eur. j. c * 72 * (2012) 2085 doi:10.1140/epjc / s10052 - 012 - 2085-z [arxiv:1204.0694 [hep - ph]]. h. czyz, a. grzelinska and r. zabawa, phys. b * 538 * (2002) 52 [hep - ph/0204039]. r. kaiser and j. schweizer, jhep * 0606 * (2006) 009 [hep - ph/0603153] j. gasser, c. haefeli, m. a. ivanov and m. schmid, phys. * 41 * (2010) 939. s. friot, d. greynat and e. de rafael, phys. b * 628 * (2005) 73 [hep - ph/0505038]. j. bijnens, eur. j. c * 75 * (2015) 1, 27 [arxiv:1412.0887 [hep - ph]]. r. mertig and r. scharf, comput. commun. * 111 * (1998) 265 [hep - ph/9801383]. o. v. tarasov, nucl. b * 502 * (1997) 455 [hep - ph/9703319]. j. gluza, k. kajda and t. riemann, comput. commun. * 177 * (2007) 879 [arxiv:0704.2423 [hep - ph]]. a. v. smirnov and v. a. smirnov, eur. j. c * 62 * (2009) 445 [arxiv:0901.0386 [hep - ph]]. m. czakon, comput. commun. * 175 * (2006) 559 [hep - ph/0511200]. d. kosower, http://www.hepforge.org/downloads/mbtools/barnesroutines-1.0.tar.gz v. a. smirnov, springer tracts mod. * 250 *, 1 (2012). s. friot and d. greynat, j. math. * 53 * (2012) 023508 [arxiv:1107.0328 [math - ph]]. b. jantzen, j. math. * 54 * (2013) 012304 [arxiv:1211.2637 [math - ph]]. k. a. olive _ et al. _ [particle data group collaboration], chin. c * 38 * (2014) 090001. doi:10.1088/1674 - 1137/38/9/090001 j. a. m. vermaseren, math - ph/0010025. | we demonstrate the use of several code implementations of the mellin - barnes method available in the public domain to derive analytic expressions for the sunset diagrams that arise in the two - loop contribution to the pion mass and decay constant in three - flavoured chiral perturbation theory.
we also provide results for all possible two - mass configurations of the sunset integral, and derive a new one - dimensional integral representation for the one mass sunset integral with arbitrary external momentum.
thoroughly annotated mathematica notebooks are provided as ancillary files, which may serve as pedagogical supplements to the methods described in this paper.
lu tp 16 - 43 + august 2016 * an analytic approach to sunset diagrams in chiral perturbation theory : theory and practice * + @xmath0 centre for high energy physics, indian institute of science, + bangalore-560012, karnataka, india + @xmath1department of astronomy and theoretical physics, lund university, + slvegatan 14a, se 223 - 62 lund, sweden + @xmath2department of physics and astronomy, university of delaware, + newark, de 19716, usa + | 1608.02386 |
Introduction
wolf - rayet (wr) stars represent the final phase in the evolution of very massive stars prior to core - collapse, in which the h - rich envelope has been stripped away via either stellar winds or close binary evolution, revealing products of h - burning (wn sequence) or he - burning (wc sequence) at their surfaces, i.e. he, n or c, o (crowther 2007). wr stellar winds are significantly denser than o stars, as illustrated in fig. [wrross], so their visual spectra are dominated by broad emission lines, notably heii @xmath04686 (wn stars) and ciii @xmath04647 - 51, ciii @xmath05696, civ @xmath05801 - 12 (wc stars). the spectroscopic signature of wr stars may be seen individually in local group galaxies (e.g. massey & johnson 1998), within knots in local star forming galaxies (e.g. hadfield & crowther 2006) and in the average rest frame uv spectrum of lyman break galaxies (shapley et al. 2003). in the case of a single massive star, the strength of stellar winds during the main sequence and blue supergiant phase scales with the metallicity (vink et al. consequently, one expects a higher threshold for the formation of wr stars at lower metallicity, and indeed the smc shows a decreased number of wr to o stars than in the solar neighbourhood. alternatively, the h - rich envelope may be removed during the roche lobe overflow phase of close binary evolution, a process which is not expected to depend upon metallicity. wr stars represent the prime candidates for type ib / c core - collapse supernovae and long, soft gamma ray bursts (grbs). this is due to their immediate progenitors being associated with young massive stellar populations, compact in nature and deficient in either hydrogen (type ib) or both hydrogen and helium (type ic). for the case of grbs, a number of which have been associated with type ic hypernovae (galama et al. 1998 ; hjorth et al. 2003), a rapidly rotating core is a requirement for the collapsar scenario in which the newly formed black hole accretes via an accretion disk (macfadyen & woosley 1999). indeed, wr populations have been observed within local grb host galaxies (hammer et al. 2006). at solar metallicity, single star models predict that the core is spun down either during the red supergiant (via a magnetic dynamo) or wolf - rayet (via mass - loss) phases. the tendency of grbs to originate from metal - poor environments (e.g. stanek et al. 2006) suggests that stellar winds from single stars play a role in their origin since roche lobe overflow in a close binary evolution would not be expected to show a strong metallicity dependence. in this article, evidence in favour of a metallicity dependence for wr stars is presented, of application to the observed wr subtype distribution in local group galaxies, plus properties of wr stars at low metallicity including their role as grb progenitors.
Wr subtype distribution
historically, the wind properties of wr stars have been assumed to be metallicity independent (langer 1989), yet there is a well known observational trend to earlier, higher ionization, wn and wc subtypes at low metallicity as illustrated in fig. [wrpop], whose origin is yet to be established. mass - loss rates for wn stars in the milky way and lmc show a very large scatter. the presence of hydrogen in some wn stars further complicates the picture since wr winds are denser if h is absent (nugis & lamers 2000). this is illustrated in fig. [wn_mdot], which reveals that the wind strengths of (h - rich) wn winds in the smc are lower than corresponding h - rich stars in the lmc and milky way (crowther 2006). [wc_mdot] shows that the situation is rather clearer for wc stars, for which lmc stars reveal @xmath10.2 dex lower mass - loss rates than milky way counterparts (crowther et al. 2002) the observed trend to earlier subtypes in the lmc (fig. [wrpop]) was believed to originate from a difference in carbon abundances relative to galactic wc stars (smith & maeder 1991), yet quantitative analysis reveals similar carbon abundances (koesterke & hamann 1995 ; crowther et al. 2002). theoretically, nugis & lamers (2002) argued that the iron opacity peak was the origin of the wind driving in wr stars, which grfener & hamann (2005) supported via an hydrodynamic model for an early - type wc star in which lines of fe ix - xvii deep in the atmosphere provided the necessary radiative driving. vink & de koter (2005) applied a monte carlo approach to investigate the metallicity dependence for cool wn and wc stars revealing @xmath2 where @xmath3=0.86 for wn stars and @xmath3=0.66 for wc stars for 0.1 @xmath4. the weaker wc dependence originates from a decreasing fe content and constant c and o content at low metallicity. empirical results for the solar neighbourhood, lmc and smc presented in figs. [wn_mdot][wc_mdot] are broadly consistent with theoretical predictions, although detailed studies of individual wr stars within galaxies broader range in metallicity would provide stronger constraints. theoretical wind models also predict smaller wind velocities at lower metallicity, as is observed for wo stars, which are presented in fig. [wo] (crowther & hadfield 2006). the impact of a metallicity dependence for wr winds upon spectral types is as follows. at high metallicity, recombination from high to low ions (early to late subtypes) is very effective in very dense winds, whilst the opposite is true for low metallicity, low density winds. the situation is illustrated in the upper panel of fig. [wc_wo], where we present synthetic wc spectra obtained from identical models except that their wind densities differ by a factor of 10, and the weak wind model is assumed to be extremely fe - poor (adapted from crowther & hadfield 2006). the high wind density case has a wc4 spectral type whilst the low wind density case has an earlier wo subtype. crowther et al. (2002) noted that a further increase in wind density by a factor of 2 predicts a wc7 subtype. stellar temperatures further complicates this picture, such that the spectral type of a wr star results from a subtle combination of ionization and wind density, in contrast with normal stars.
Wr populations at low metallicity
the effect of reduced wr wind densities at low metallicity on wr populations is as follows. wr optical recombination lines will (i) decrease in equivalent width, since their strength scales with the square of the density, and (ii) decrease in line flux, since the lower wind strength will reduce the line blanketing, resulting in an increased extreme uv continuum strength at the expense of the uv and optical. the equivalent widths of optical emission lines in smc wn stars are well known to be lower than milky way and lmc counterparts (conti et al. 1989). to date, the standard approach for the determination of unresolved wr populations in external galaxies has been to assume metallicity independent wr line fluxes obtained for milky way and lmc stars (schaerer & vacca 1998) regardless of whether the host galaxy is metal - rich (mrk 309, schaerer et al. 2000) or metal - poor (i zw18, izotov et al. 1997). ideally, one would wish to use wr template stars appropriate to the metallicity of the galaxy under consideration. unfortunately, this is only feasible for the lmc, smc and solar neighbourhood, since it is challenging to isolate individual wr stars from ground based observations in more distant galaxies, which span a larger spread in metallicity. line luminosities for optical emission lines in lmc and smc wr stars are compared in table [flux], illustrating significantly lower (factor of 56) luminosities for the lower metallicity of the smc..mean heii @xmath04686 line luminosities for magellanic cloud wn stars including known binaries (crowther & hadfield 2006) [cols="<,^,>,^, > ",] reduced wr line fluxes are also predicted for wr atmospheric models at low metallicity if one follows the metallicity dependence from vink & de koter (2005), such that wr populations inferred from schaerer & vacca (1998) at low metallicity may underestimate actual populations by an order of magnitude. this is potentially problematic for single star evolutionary models at very low metallicities (@xmath5) since the wr populations inferred for i zw18 and sbs0335 - 052e using milky way line fluxes compare well with evolutionary models (e.g. izotov et al. 1997 ; papaderos et al. if wr populations are in fact a factor of @xmath110 larger, similar to that of the smc, close binary evolution would represent the most likely origin for such large wr populations. cc + & + for magellanic cloud metallicity starburst galaxies, one may employ appropriate template spectra (e.g. crowther & hadfield 2006) to reproduce wr features, as shown in the upper panel of fig. [ngc3125] for the starburst galaxy ngc 3125. indeed, consistent fits to the blue and yellow wr bumps may be achieved for the a1 cluster within ngc 3125 using lmc template wr stars, as shown in the lower panels of fig. [ngc3125] (hadfield & crowther 2006).
Ionizing fluxes and grb progenitors
schmutz et al. (1992) demonstrated that the ionizing fluxes from wr stars soften as wind density increases. consequently, a metallicity dependence for wr wind strengths implies that wr ionizing flux distributions soften at increased metallicity, as demonstrated by smith et al. indeed, relatively soft ionizing fluxes are observed in the super - solar metallicity wr starburst galaxy ngc 3049 (gonzalez delgado et al. 2002). at low metallicities, one anticipates a combination of weak uv and optical spectral lines from wr stars (i.e. weak stellar heii @xmath04686) but very strong h and he lyman continua (i.e. strong nebular heii @xmath04686), as is indicated in the lower panel of fig. [wc_wo]. indeed, low metallicity star forming galaxies display strong nebular heii @xmath04686, although shocks from supernovae remnants may also contribute to nebular emission. the typically environment of nearby (@xmath6) long duration grbs is unusually metal - poor, as emphasized by stanek et al. (2006) with respect to star forming galaxies from the sloan digital sky survey. reduced wr mass - loss rates at low metallicity will lead to reduced densities in the immediate environment of grbs with respect to typical wr stars, as is observed (chevalier et al. in addition, massive single stars undergoing homogeneous evolution in which wr mass - loss rates are low may maintain their rapidly spinning cores through to core - collapse (yoon & langer 2005 ; langer & norman 2006).
Summary
observational and theoretical evidence supports reduced wind densities and velocities for low metallicity wr stars, which addresses the relative wr subtype distribution in the milky way and magellanic clouds, plus the reduced wr line strengths in the smc with regard to the galaxy and lmc. the primary impact at low metallicity is as follows ; (a) an increased wr population due to lower line fluxes from individual stars, of particular relevance to i zw18 and sbs0335 - 052e ; (b) harder ionizing fluxes from wr stars, potentially responsible for the strong nebular heii @xmath04686 seen in low metallicity hii galaxies ; (c) responsible for the reduced density of grb environments with respect to solar metallicity wr counterparts. finally, a metallicity dependence for wr winds may help to reconcile the relative number of wn to wc stars observed in surveys (e.g. massey & johnson 1998) with evolutionary predictions. evolutionary models for which rotational mixing is included yet metallicity dependent wr winds are not (meynet & maeder 2003) fail to predict the high n(wc)/n(wn) ratio observed at high metallicities (hadfield et al. 2005), whilst models which account for the vink & de koter (2005) wr wind dependence compare much more favourably with observations (eldridge & vink 2006), in spite of the neglect of rotational mixing. chevalier, r.a. li, z.- y. & fransson, c. 2004 the diversity of gamma - ray burst afterglows and the surroundings of massive stars _ apj _ * 606 *, 369 conti, p.s. garmany, c.d. & massey p., 1989 spectroscopic studies of wolf - rayet stars. v - optical spectrophotometry of the emission lines in small magellanic cloud stars _ apj _ * 341 *, 113 crowther, p.a. 2006 observed metallicity dependence of winds from wr stars. in _ stellar evolution at low metallicity : mass - loss, explosions, cosmology _ (eds : h. lamers, n. langer & t. nugis), asp conf series in press crowther, p.a. 2007 physical properties of wolf - rayet stars _ _ in preparation crowther, p.a. & hadfield, l.j. 2006 reduced wolf - rayet line luminosities at low metallicity _ a&a _ * 449 *, 711 crowther, p.a. dessart, l. hillier, d.j. abbott j.b. & fullerton a.w. 2002 stellar and wind properties of lmc wc4 stars. a metallicity dependence for wolf - rayet mass - loss rates _ a&a _ * 392 *, 653 eldridge, j.j. & 2006 implications of the metallicity dependence of wolf - rayet winds _ a&a _ * 452 *, 295 galama t.j. 1998 an unusual supernova in the error box of the gamma - ray burst of 25 april 1998 _ nat _ * 395 *, 670 gonzalez delgado, r.m. leitherer, c. stasinska, g. & heckman, t.m. 2002 the massive stellar content in the starburst ngc 3049 : a test for hot - star models _ apj _ * 580 *, 824 grfener, g. & hamann, w.- r. 2005 hydrodynamic model atmospheres for wr stars. self - consistent modeling of a wc star wind _ a&a _ * 432 *, 633 hadfield, l.j. & crowther, p.a. 2006 how extreme are the wolf - rayet clusters in ngc3125? _ mnras _ * 368 *, 1822 hammer, f., flores, h., schaerer, d., dessauges - zavadsky, m., le floch, e. & puech, m. 2006 detection of wolf - rayet stars in host galaxies of gamma - ray bursts (grbs) : are grbs produced by runaway massive stars ejected from high stellar density regions? _ a&a _, in press (astro - ph/0604461) hjorth, j., sollerman, j., moller, p. et al. 2003 a very energetic supernova associated with the -ray burst of 29 march 2003 _ nat _ * 423 *, 847 izotov, y.i. foltz, c.b. green, r.f. guseva, n.g. & thuan, t.x. 1997 i zw 18 : a new wolf - rayet galaxy _ apj _ * 487 *, l37 koesterke, l. & hamann, w.- r. 1995 spectral analyses of 25 galactic wolf - rayet stars of the carbon sequence _ a&a _ * 299 *, 503 langer, n., 1989 mass - dependent mass loss rates of wolf - rayet stars _ a&a _ * 220 *, 135 langer, n. & norman, c. 2006 on the collapsar model of long gamma - ray bursts : constraints from cosmic metallicity evolution _ apj _ * 638 *, l63 macfadyen, a.i. & woosley s.e. 1999 collapsars : gamma - ray bursts and explosions in `` failed supernovae '' _ apj _ * 524 *, 262 massey, p. & johnson, o., 1998 evolved massive stars in the local group. ii. a new survey for wolf - rayet stars in m33 and its implications for massive star evolution : evidence of the `` conti scenario '' in action _ apj _ * 505 *, 793 meynet, g. & maeder a. 2005 stellar evolution with rotation. xi. wolf - rayet star populations at different metallicities _ a&a _ * 429 * 581 nugis, t. & lamers, h.j.g.l.m., 2000 mass - loss rates of wolf - rayet stars as a function of stellar parameters _ a&a _ * 360 *, 227 nugis, t. & lamers, h.j.g.l.m. 2002 the mass - loss rates of wolf - rayet stars explained by optically thick radiation driven wind models _ a&a _ * 389 *, 162 papaderos, p., izotov, y.i., guseva, n.g., thuan t.x. & fricke k.j. 2006 oxygen abundance variations in the system of the two blue compact dwarf galaxies sbs 0335 - 052e and sbs 0335 - 052w _ a&a _ in press (astro - ph/0604270) schaerer, d. & vacca, w.d. 1998 new models for wolf - rayet and o star populations in young starbursts _ apj _ * 497 *, 618 schaerer, d., guseva, n.g., izotov, y.i. & thuan, t.x. 2000 massive star populations and the imf in metal - rich starbursts _ _ * 362 *, 53 schmutz, w. leitherer, c. & gruenwald, r. 1992 theoretical continuum energy distributions for wolf - rayet stars _ pasp _ * 104 *, 1164 shapley, a.e., steidel, c.s., pettini, m. & adelberger, k.l. 2003 rest - frame ultraviolet spectra of z 3 lyman break galaxies _ apj _ * 588 *, 65 smith, l.f. & maeder, a. 1991 comparison of predicted and observed properties of wc stars - explanation of the subtype gradient in galaxies _ a&a _ * 241 *, 77 smith, l.j. norris, r.p.f. & crowther, p.a. 2002 realistic ionizing fluxes for young stellar populations from 0.05 to 2@xmath7 _ mnras _ * 337 *, 1309 stanek, k.z., gnedin, o.y. beacom, j.f. et al. 2006 protecting life in the milky way : metals keep the grbs away _ apj _ in press (astro - ph/0604113) vink, j.s. & de koter a. 2005 on the metallicity dependence of wolf - rayet winds _ a&a _ * 442 *, 587 vink, j.s., de koter, a. & lamers, h.j.g.l.m. 2001 mass - loss predictions for o and b stars as a function of metallicity _ a&a _, * 369 *, 574 yoon, s - c & langer, n. 2005 evolution of rapidly rotating metal - poor massive stars towards gamma - ray bursts _ a&a _ * 443 *, 643 | observational and theoretical evidence in support of metallicity dependent winds for wolf - rayet stars is considered.
well known differences in wolf - rayet subtype distributions in the milky way, lmc and smc may be attributed to the sensitivity of subtypes to wind density.
implications for wolf - rayet stars at low metallicity include a hardening of ionizing flux distributions, an increased wr population due to reduced optical line fluxes, plus support for the role of single wr stars as gamma ray burst progenitors. | astro-ph0606580 |
Introduction
the spectroscopic signature of the presence of @xmath1li in the atmospheres of metal - poor halo stars is a subtle extra depression in the red wing of the @xmath2li doublet, which can only be detected in spectra of the highest quality. based on high - resolution, high signal - to - noise vlt / uves spectra of 24 bright metal - poor stars, (*??? * asplund (2006)) report the detection of @xmath1li in nine of these objects. the average @xmath1li/@xmath2li isotopic ratio in the nine stars in which @xmath1li has been detected is about 4% and is very similar in each of these stars, defining a @xmath1li plateau at approximately @xmath5li@xmath6 (on the scale @xmath7h@xmath8). a convincing theoretical explanation of this new @xmath1li plateau turned out to be problematic : the high abundances of @xmath1li at the lowest metallicities can not be explained by current models of galactic cosmic - ray production, even if the depletion of @xmath1li during the pre - main - sequence phase is ignored (see reviews by e.g. (*??? * christlieb 2008), (*??? * cayrel 2008), prantzos 2010 [this volume] and references therein). a possible solution of the so - called ` second lithium problem'was proposed by (*??? * cayrel (2007)), who point out that the intrinsic line asymmetry caused by convection in the photospheres of metal - poor turn - off stars is almost indistinguishable from the asymmetry produced by a weak @xmath1li blend on a presumed symmetric @xmath2li profile. as a consequence, the derived @xmath1li abundance should be significantly reduced when the intrinsic line asymmetry in properly taken into account. using 3d nlte line formation calculations based on 3d hydrodynamical model atmospheres computed with the co@xmath0bold code ((*??? * freytag 2002), (*??? * wedemeyer 2004), see also http://www.astro.uu.se/@xmath9bf/co5bold_main.html), we quantify the theoretical effect of the convection - induced line asymmetry on the resulting @xmath1li abundance as a function of effective temperature, gravity, and metallicity, for a parameter range that covers the stars of the (*??? * asplund (2006)) sample. a careful reanalysis of individual objects is under way, in which we consider two alternative approaches for fixing the residual line broadening, @xmath10, the combined effect of macroturbulence (1d only) and instrumental broadening, for given microturbulence (1d only) and rotational velocity : (i) treating @xmath10 as a free parameter when fitting the li feature, (ii) deriving @xmath10 from additional unblended spectral lines with similar properties as lii@xmath4. we show that method (ii) is potentially dangerous, because the inferred broadening parameter shows considerable line - to - line variations, and the resulting @xmath1li abundance depends rather sensitively on the adopted value of @xmath10. +
3d hydrodynamical simulations and spectrum synthesis
the hydrodynamical atmospheres used in the present study are part of the cifist 3d model atmosphere grid ((*??? * ludwig 2009)). they have been obtained from realistic numerical simulations with the co@xmath0bold code which solves the time - dependent equations of compressible hydrodynamics in a constant gravity field together with the equations of non - local, frequency - dependent radiative transfer in a cartesian box representative of a volume located at the stellar surface. the computational domain is periodic in @xmath11 and @xmath12 direction, has open top and bottom boundaries, and is resolved by typically 140@xmath13140@xmath13150 grid cells. the vertical optical depth of the box varies from @xmath14 (top) to @xmath15 (bottom), and the radiative transfer is solved in 6 or 12 opacity bins. further information about the models used in the present study is compiled in table[tab1]. each of the models is represented by a number of snapshots, indicated in column (6), chosen from the full time sequence of the corresponding simulation. these representative snapshots are processed by the non - lte code nlte3d that solves the statistical equilibrium equations for a 17 level lithium atom with 34 line transitions, fully taking into account the 3d thermal structure of the respective model atmosphere. the photo - ionizing radiation field is computed at @xmath16 frequency points between @xmath17 and 32407 , using the opacity distribution functions of @xcite to allow for metallicity - dependent line - blanketing, including the hi h@xmath18 and hi hi quasi - molecular absorption near @xmath19 and @xmath20 , respectively. collisional ionization by neutral hydrogen via the charge transfer reaction h(@xmath21) + li(@xmath22) @xmath23 li@xmath18(@xmath24) + h@xmath25 is treated according to @xcite. more details are given in @xcite. finally, 3d nlte synthetic line profiles of the lii @xmath26 doublet are computed with the line formation code linfor3d (http://www.aip.de/@xmath9mst/linfor3d_main.html), using the departure coefficients @xmath27=@xmath28 provided by nlte3d for each level @xmath29 of the lithium model atom as a function of geometrical position within the 3d model atmospheres. as demonstrated in fig.[fig1], 3d nlte effects are very important for the metal - poor dwarfs considered here : they strongly reduce the height range of line formation such that the 3d nlte equivalent width is smaller by roughly a factor 2 compared to 3d lte. ironically, the line strength predicted by standard 1d mixing - length models in lte are close to the results obtained from elaborate 3d nlte calculations. we note that the half - width of the 3d nlte line profile, fwhm(nlte)=8.5 km / s, is larger by about 10% : fwhm(lte)=@xmath30 and @xmath31 km / s, respectively, before and after reducing the li abundance such that 3d lte and 3d nlte equivalent widths agree. this is because 3d lte profile senses the higher photosphere where both thermal and hydrodynamical velocities are lower. however, the nlte line profile is significantly less asymmetric than the lte profile, even if the latter is broadened to the same half - width (fig.[fig1], bottom panel). + .list of models used in the present study. columns (2)-(6) give effective temperature, surface gravity, metallicity, number of opacity bins used in the radiation hydrodynamics simulation, and number of snapshots selected for spectrum synthesis. the equivalent width of the synthetic 3d non - lte @xmath2li doublet at @xmath26 , assuming a(li)=2.2 and no @xmath1li, is given in column (7). columns (8) and (9) show @xmath32(li) and @xmath33(li), the @xmath1li/@xmath2li isotopic ratio inferred from fitting this 3d non - lte line profile with two different kinds of 1d profiles, in each case assuming a rotational broadening of @xmath34 = 0.0 / 2.0 km / s, respectively (see text for details). [cols="^,^,^,^,^,^,^,^,^ ",]
Conclusions
the @xmath1li/@xmath2li isotopic ratio derived from fitting of the lii doublet with 3d nlte synthetic line profiles is shown to be about 1% to 2% lower than what is obtained with 1d lte profiles. based on this result, we conclude that only @xmath35 out of the @xmath36 stars of the @xcite sample would remain significant @xmath1li detections when subjected to a 3d non - lte analysis, suggesting that the presence of @xmath1li in the atmospheres of galactic halo stars is rather the exception than the rule, and hence does not necessarily constitute a _ cosmological _ @xmath1li problem. if we adopt the approach by @xcite, relying on additional spectral lines to fix the residual line broadening, the difference between 3d nlte and 1d lte results increases even more, as far as we can judge from our case study hd74000. until the 3d nlte effects are fully understood for all involved lines, we consider this method as potentially dangerous. | based on 3d hydrodynamical model atmospheres computed with the co@xmath0bold code and 3d non - lte (nlte) line formation calculations, we study the effect of the convection - induced line asymmetry on the derived @xmath1li abundance for a range in effective temperature, gravity, and metallicity covering the stars of the (*???
* asplund (2006)) sample.
when the asymmetry effect is taken into account for this sample of stars, the resulting @xmath1li/@xmath2li ratios are reduced by about 1.5% on average with respect to the isotopic ratios determined by (*???
* asplund (2006)).
this purely theoretical correction diminishes the number of significant @xmath1li detections from 9 to 4 (2@xmath3 criterion), or from 5 to 2 (3@xmath3 criterion). in view of this result
the existence of a @xmath1li plateau appears questionable.
a careful reanalysis of individual objects by fitting the observed lithium @xmath4 doublet both with 3d nlte and 1d lte synthetic line profiles confirms that the inferred @xmath1li abundance is systematically lower when using 3d nlte instead of 1d lte line fitting.
nevertheless, halo stars with unquestionable @xmath1li detection do exist even if analyzed in 3d - nlte, the most prominent example being hd84937. | 1001.3274 |
Introduction
we consider @xmath0 diffeomorphisms on riemannian surfaces. our goal is to study the hyperbolic properties of a class of maps exhibiting a homoclinic tangency associated to a fixed saddle point @xmath1, as in figure [map]. homoclinic tangencies inside the limit set, width=226] we assume without loss of generality that we are working on @xmath2 and in the standard euclidean norm. we recall that a compact invariant set @xmath3 is * uniformly hyperbolic * if there exist constants @xmath4 and a continuous, @xmath5-invariant, decomposition @xmath6 of the tangent bundle over @xmath7 such that for all @xmath8 and all @xmath9 we have @xmath10 by standard hyperbolic theory, every point @xmath11 in @xmath7 has stable and unstable manifolds @xmath12 tangent to the subspaces @xmath13 and @xmath14 respectively, and thus in particular _ transversal _ to each other. the presence of the tangency therefore implies that the dynamics on @xmath15 can not be uniformly hyperbolic. we emphasize at this point that, in the case we are considering, the homoclinic tangency is accumulated by transverse homoclinic orbits which in turn are accumulated by periodic points. thus it constitutes an intrinsic obstruction to uniform hyperbolicity which can not be resolved by simply ignoring the orbit of tangency. most of the classical theory of homoclinic bifurcations for diffeomorphisms (see @xcite and references therein) considers the unfolding of homoclinic tangencies _ external _ to the set @xmath7 to which they are associated, thus causing no real issues with the hyperbolicity at the bifurcation parameter. the main goal of such a theory has often been to study the hyperbolicity and the occurrence of tangencies in a neighbourhood of the orbit of tangency _ after _ the bifurcation. the presence of an internal tangency gives rise to a more subtle situation and it has only recently been shown that this can actually occur as a _ first bifurcation _ part of the motivation of the present paper is to study the global dynamics and hyperbolicity _ at this bifurcation parameter_. another part of the motivation for this result is to give an example of a compact invariant set which is as `` uniformly '' hyperbolic as possible in the ergodic theory sense, but still not uniformly hyperbolic. to formulate this result precisely, let @xmath16 denote the set of all @xmath17-invariant probability measures @xmath18 on @xmath7. by the classical multiplicative ergodic theorem of oseledet s there is a well defined set @xmath19 of _ lyapunov exponents _ associated to the measure @xmath18, we give the precise definitions below. we say that the measure @xmath20 is _ hyperbolic _ if all the lyapunov exponents are non - zero. the existence of an invariant measure with non - zero lyapunov exponents indicates a minimum degree of hyperbolicity in the system. a stronger requirement is that all invariant measures @xmath18 are hyperbolic and of course an even stronger requirement is that they are all `` uniformly '' hyperbolic in the sense that all lyapunov exponents are uniformly bounded away from 0. this condition is clearly satisfied for uniformly hyperbolic systems but, as we show in this paper, it is strictly weaker. the class of examples we are interested in were first introduced in @xcite and constitute perhaps the simplest situation in which an internal tangency can occur as a first bifurcation. in section [maps] we give the precise definition of this class. for this class we shall then prove the following all lyapunov exponents of all measures in @xmath21 are uniformly bounded away from zero. as an immediate corollary we have the following statement which is in itself non - trivial and already remarkable. @xmath22 is uniformly hyperbolic on periodic points. we recall that uniform hyperbolicity on periodic points means that there exists constants @xmath23 such that for each periodic point @xmath24 of period @xmath25, the derivative @xmath26 has two distinct real eigenvalues @xmath27 with @xmath28 notice that the bounds for the eigenvalues are exponential in @xmath29. as far as we know this is the first known example of this kind, although it is possible, and indeed even likely, that such a property should hold for more complex examples such as benedicks - carleson and mora - viana parameters in hnon - like families @xcite and horseshoes at the boundary of the uniform hyperbolicity domain of the hnon family. the weaker result on the uniform hyperbolicity of periodic points has been proved recently for both cases in @xcite and @xcite respectively.. other known examples of non - uniformly hyperbolic diffeomorphisms include cases in which the lack of uniformity comes from the presence of `` neutral '' fixed or periodic points. in these cases, the dirac-@xmath30 measures on such periodic orbits are invariant and have a zero lyapunov exponent. it is interesting to view our result in the light of some recent work which appears to go in the opposite direction : if a compact invariant set @xmath31 admits an invariant splitting @xmath32 such that, for a total measure set of points @xmath33, the lyapunov exponents are positive in @xmath34 and negative in @xmath35, then @xmath31 is uniformly hyperbolic, @xcite. here, the lyapunov exponents are not even required to be uniformly bounded away from zero. thus the existence of at least one orbit as which the splitting degenerates is a necessary condition for a situation such as the one we are considering, in which the lypaunov exponents are all non - zero but @xmath7 is strictly not uniformly hyperbolic. the concept of uniform hyperbolicity of the periodic points and of measures plays an important role in the general theory of one - dimensional dynamics. in some situations, such as for certainclasses of smooth non - uniformly expanding unimodal maps these notions have been been shown to be equivalent to each other and to various other properties usually associated to uniform hyperbolicity such as exponential decay of correlations @xcite. higher dimensional cases are generally much harder due to the complexity of the geometrical and dynamical structure of the examples and progress is only starting to be made. after defining the class of systems of interest in section [maps], in section [splitting] we construct a field of cones which are invariant under the derivative map. in section [periodicpoints] we show that some * non - uniform hyperbolicity * is satisfied off the orbit of tangency : the expanding and contracting directions of points which are very close to the orbit of tangency are almost aligned and therefore the contraction and/or expansion may take arbitrarily long time to become effective. we formalize this by saying that there exists a constant @xmath36, _ depending on the point _ @xmath11 (compare with), such that for all @xmath11 not in the orbit of tangency we have @xmath37 with @xmath38 arbitrarily small near the points of the orbit of tangency, i.e. @xmath39 one way to show that the non - uniformity is not too extreme is to show that there is a _ uniformly hyperbolic core _ in the following sense : there exists a region @xmath40 containing one point of the orbit of tangency, and there exists constants @xmath23 such that, for each point @xmath41 that returns to @xmath40 at the @xmath42-th iterate, we have @xmath43 in fact, we show that this uniform hyperbolicity occurs well before the return of the orbit of @xmath44 to @xmath40. those exponential estimates are valid from the moment that the orbit of @xmath44 leaves a certain neighborhood of the fixed point @xmath1, and they keep working as long as the orbit wanders around @xmath45. finally, in section [lyapunov] we show that this implies the statement of our main theorem.
Horseshoes with internal tangencies
we start with a geometrical definition of the class of maps under consideration. let @xmath46 be a @xmath47 diffeomorphism and let @xmath48\times [0,1] $] denote the unit square in @xmath49. we suppose that there exist 5 `` horizontal '' regions in @xmath50 which are mapped as in figure [the map] : @xmath51 for @xmath52. the map @xmath53,width=340] we suppose that regions @xmath54 are mapped affinely to their images, with derivative @xmath55 in @xmath56 and @xmath57 respectively, for two constants @xmath58 satisfying @xmath59 and @xmath60. more explicitly we suppose that @xmath61 in in @xmath62 (@xmath63), @xmath64 in @xmath57 (@xmath65) and that @xmath66 is a vertical strip parallel to @xmath67. in particular there is a hyperbolic fixed point at the origin @xmath68 whose stable and unstable manifolds contain the lower and left hand side of the square @xmath50 respectively. we emphasize that the explicit nature of the map in these regions is for simplicity only, and to allow us to concentrate on the strategy for dealing with the tangency. this could be weakened significantly, for example by assuming only some uniformly hyperbolic structure in regions @xmath69. region @xmath70 is mapped outside the square @xmath50 and thus we do not need to make any particular assumption on its form. region 4 is mapped to a `` fold '' @xmath71 which contains a point of the orbit of tangency. here we make some `` non - degeneracy assumptions '' essentially stating that vertical lines in @xmath72 are mapped to non - degenerate parabolas. more precisely, we suppose that there is a region @xmath73 bounded by two disjoint curves so that points in @xmath74, are mapped inside region @xmath75 with second coordinate greater than @xmath76. then, for each @xmath77 in @xmath78 $] we have that @xmath79 is contained in the graph of the map @xmath80 for some fixed @xmath81 and some @xmath82. thus all vertical lines in @xmath83 are mapped to parabolas with constant curvature @xmath84. again, this could be weakened significantly although we do assume for less trivial reasons that the curvature @xmath84 is sufficiently large in relation to the constants @xmath85 and @xmath86. notice that the point @xmath87 is a point of tangency of the stable and unstable manifolds of the fixed point at the origin. to control the global properties of the family of parabolas we also assume that @xmath88 and that, for every @xmath89, @xmath90 a completely explicit example of a map @xmath17 satisfying all these conditions is given in @xcite. we now define the limit set @xmath91 it follows from the construction that @xmath3 is non empty and contains (at least) one orbit of tangency between stable and unstable manifolds. our assumptions make it easy to control the dynamics in regions @xmath69 and therefore our main objective is to control the dynamics of points returning to the fold @xmath92 and to a neighbourhood of the tangency. thus we concentrate on the _ first return map _ from @xmath93 to itself and show that this map satisfies some strong hyperbolicity estimates from which we can deduce our main estimate on the hyperbolicity of periodic points.
The conefield
consider the foliation @xmath94, of @xmath95, whose leaves are images of vertical lines by the map @xmath96 (in @xmath97, the leaves are parabolas). to each point @xmath98 we associate the tangent direction, @xmath99, to the leaf of @xmath94 which passes through @xmath98, at this point (@xmath99 is parallel to @xmath100. let @xmath101 be the angle between @xmath99 and the horizontal. there is only one point for which @xmath102 and this is the point of tangency @xmath103. at all other points we have with @xmath104. then, we divide @xmath7 into 3 regions, according to the value of @xmath105 : @xmath106 notice that @xmath107 for each point @xmath108 we define a _ cone _ @xmath109 of vectors in the tangent space @xmath110 as follows : let @xmath111, and @xmath112 for any integer @xmath113. all cones will be centered on the vertical and, for convenience, we define a _ standard cone _ @xmath114 for @xmath115 in @xmath116 we simply let @xmath117. for @xmath118, we define cones which extend to each side of the vertical by three times the (cotangent of the) angle of the line @xmath119 defined above. more precisely, recall that @xmath120 is tangent to the graph of @xmath121, see, and therefore @xmath122. then we let @xmath123 before defining the cones in the remaining points in @xmath124 we show that the conefield defined above in @xmath125 is invariant under the first return map to @xmath125. as a first step towards proving this invariance we estimate the first return time. [return] consider @xmath126, and @xmath127 the smallest positive integer such that @xmath128. then @xmath129 note that, since @xmath130, we have @xmath131, provided that @xmath132. in fact, the images @xmath133 of the point @xmath115 will stay in region @xmath134 while @xmath113 is such that @xmath135. since @xmath136, there must be at least one more linear iterate of @xmath133 before its orbit visits the pre - image of @xmath137. so, we have @xmath138. analogously, using the inverse map, we have @xmath139. we shall show below that these estimates mean that cones are sufficiently contracted before returning, guaranteeing that they are mapped back strictly into existing cones even if they started out very wide. in particular we shall use the following simple [returncor] consider @xmath126, and @xmath127 the smaller positive integer such that @xmath128. then for this @xmath127 we have @xmath140 and @xmath141 the first inequality follows from which gives @xmath142 the second follows also by by a similar straightforward calculation. we are now ready to prove the invariance of the conefield for the first return map. [inclusion] there exists @xmath143 such that, if @xmath144 in the definition of @xmath96, @xmath145 and @xmath146 is the first positive iterate of @xmath115 in @xmath147, then @xmath148 since @xmath149 is centered in the vertical line, we have, after applying @xmath150 to the vectors of @xmath151, that @xmath152 is a cone centered in the vertical line at the point @xmath153 whose width is @xmath154 times the width of @xmath151. after applying @xmath155 to this cone, since @xmath136, we obtain a cone centered in @xmath156 with width smaller than @xmath157 times the original width. there are many cases to be considered depending on the location of @xmath115 and @xmath146 in @xmath147. here we present the case where @xmath115 and @xmath146 are both in @xmath158. the other cases are made following the same steps. let @xmath115 and @xmath146 be in @xmath158. then @xmath159 and @xmath160 since @xmath161, it satisfies @xmath162 this implies that a cone centered in @xmath163 with width @xmath164 will be properly contained in @xmath165. then, in order to obtain @xmath166 we need @xmath167 or, equivalently, @xmath168 we now distinguish two cases. in the case where @xmath169 we have @xmath170 and therefore it is enough to show that @xmath171. by this follows as long as we have @xmath172 for all @xmath173 (remember that we are working with points of @xmath78\times[0,1]$]). since @xmath174 is fixed and negative, the condition holds if @xmath175 is big enough. in the case where @xmath176 we argue in the same way and, using, reduce the problem to showing that @xmath177 for any @xmath178. again this follows as long as @xmath84 is sufficiently large. the other cases follow analogously. we can now extend the conefield to the set @xmath179, in a natural way by considering the images of of all cones in @xmath125 and taking slightly wider cones at each point. in this way we obtain a conefield defined at every point outside the orbit of tangency such that @xmath180 for every point @xmath181. notice that these cones can be arbitrarily wide close to the orbit of tangency. for points @xmath181 which do not enter @xmath125 in either forward or backward time, we simply let @xmath109 be the standard cone @xmath182, see. for points @xmath181 which intersect @xmath183 only in forward time, we define @xmath184 where @xmath185 is the first time for which @xmath186 and @xmath187 is that standard cone. the stable cone field is defined assigning to each @xmath188, the closure of the complement of @xmath189, and satisfies the inclusion condition for the inverse map.
Uniform hyperbolicity for the escape time
we fix a neighbourhood @xmath40 of the tangency point @xmath190 of radius @xmath191. let @xmath192. we start with a simple estimate of the number of iterations @xmath127 it takes before @xmath146 falls outside the set @xmath78\times[0,1/3]$] (@xmath127 is the escape time of @xmath115). [pontos] we have @xmath193. note that @xmath40 is contained in the domain where @xmath194 is linear, and, as long as the second coordinate of the image of a point of @xmath31 is less than @xmath195, it is contained in @xmath134, the second coordinate of @xmath115 being multiplied by @xmath196 at each iteration. for @xmath197 as above, let @xmath198 be a vector contained in @xmath151 and @xmath199. [vetores] we have @xmath200 as we saw in the last section, since @xmath201 is linear in @xmath40, we have @xmath202 [expansao]with the notation above, if @xmath175 is sufficiently big, then @xmath203 since @xmath204, we have that @xmath205 using lemmas [vetores] and [pontos], we obtain @xmath206 by lemma [pontos] we have that @xmath207 and therefore @xmath208 since @xmath209, we can write @xmath210 the result follows if @xmath175 is sufficiently large (greater than @xmath211 in this case).
Lyapunov exponents
let @xmath212 be the connected component of @xmath213 containing @xmath214, and @xmath215. in this section we estimate the growth of the unstable vectors of points in @xmath216 outside @xmath217, under the action of @xmath155. we also compute bounds for the lyapunov exponents of @xmath96, and prove the main result. first we claim that, if @xmath219, we have @xmath222. indeed, by construction, if @xmath223 and at distance at least @xmath191 from the tangency point @xmath190, then we have that @xmath224, and the width of @xmath151 is less than @xmath225. notice that vectors in @xmath226 grow by at least a constant factor @xmath218 at each iteration where @xmath218 is the rate of growth of the vector @xmath227 by the linear map @xmath228). if @xmath115 is a point outside the set @xmath217, then we have already @xmath229, and the estimate applies to the unstable vectors of @xmath115. let @xmath230. until now we proved that, if the orbit of @xmath115 visits the set @xmath40, then when it leaves the set @xmath217, it has accumulated exponential growth to the unstable vectors by a factor @xmath231 as we have seen in lemma [expansao]. as shown above, iterates outside @xmath217 contribute as well with the same factor. similar calculations hold for the stable vectors, and we define @xmath232 through the analogous estimates for @xmath233. we remark that in particular these estimates improve significantly on the growth estimates of @xcite which already imply the existence of a non - uniformly hyperbolic splitting as defined in the introduction. [sequencia] if @xmath234 does not belong to the orbit of homoclinic tangency, @xmath235, and @xmath236, then there exist sequences @xmath237 and @xmath238, and a positive number @xmath239 such that @xmath240 and @xmath241. let @xmath234 be a point in @xmath31 outside the orbit of homoclinic tangency. for @xmath244, let @xmath245 and @xmath246. let @xmath247. if @xmath248 is non - empty and infinite, let @xmath249 and consider @xmath250 as the smaller natural number bigger than @xmath251 such that @xmath252. in this case, we have @xmath253, for any @xmath254. as a consequence of the estimates in lemmas ([expansao]) and ([first]), we have that if @xmath248 is empty, then, there are three cases to consider. if the orbit of @xmath234 is not in @xmath217, then we have exponential growth beginning at the first iterate. if @xmath263 and leaves it eventually, we wait until it does so to have exponential growth, and argue as before. if @xmath264 and never leaves it, then @xmath265 for all @xmath244, meaning that @xmath234 is in the local stable manifold of @xmath266. in this case, any non - horizontal vector based in @xmath234 will eventually grow exponentially by a factor @xmath267. let @xmath269 be an invariant measure, and @xmath275 the subset of @xmath276 for which the lyapunov exponents exist. then, by the oseledet s theorem @xmath277. by lemma ([sequencia]) all lyapunov exponents are outside the interval @xmath278, for all @xmath279. | we study the hyperbolicity of a class of horseshoes exhibiting an _ internal _ tangency, i.e. a point of homoclinic tangency accumulated by periodic points.
in particular these systems are strictly _ not _ uniformly hyperbolic.
however we show that all the lyapunov exponents of all invariant measures are uniformly bounded away from 0.
this is the first known example of this kind. | math0306041 |
Introduction
quantum walks @xcite have been proposed as potentially useful components of quantum algorithms @xcite. in recent years these systems have been studied in detail and some progress has been made in developing new quantum algorithms using either continuous @xcite or discrete @xcite versions of quantum walks. the key to the potential success of quantum walks seems to rely on the ability of the quantum walker to efficiently spread over a graph (a network of sites) in a way that is much faster than any algorithm based on classical coin tosses. quantum interference plays an important role in quantum walks being the crucial ingredient enabling a faster than classical spread. for this reason, some effort was made in recent years in trying to understand the implications of the process of decoherence for quantum walks @xcite. decoherence, an essential ingredient to understand the quantum classical transition @xcite, could turn the quantum walk into an algorithm as inefficient as its classical counterpart. the models studied in this context can be divided in two classes depending on how the coupling with an external environment is introduced. in fact, a quantum walk consists of a quantum particle that can occupy a discrete set of points on a lattice. in the discrete version, the walker carries a quantum coin, which in the simplest case can be taken as a spin-@xmath0 degree of freedom. the algorithm proceeds so that the walker moves in one of two possible directions depending on the state of the spin (for more complex regular arrays, a higher spin is required). so, in this context it is natural to consider some decoherence models where the spin is coupled to the environment and others where the position of the walker is directly coupled to external degrees of freedom. the specific system in which the algorithm is implemented in practice will dictate which of these two scenarios is more relevant. several experimental proposals to implement discrete quantum walks in systems such as ion traps @xcite, cavity qed @xcite, and optical lattices @xcite have been analyzed (see also ref. @xcite for a recent nmr implementation of a continuous quantum walk). the main effect of decoherence on quantum walks is rather intuitive : as the interaction with the environment washes out quantum interference effects, it restores some aspects of the classical behavior. for example, it has been shown that the spread of the decohered walker becomes diffusion dominated proceeding slower than in the pure quantum case. this result was obtained both for models with decoherence in the coin and in the position of the walker @xcite. however, it is known that classical correspondence in these systems has some surprising features. for example, for models with some decoherence in the quantum coin the asymptotic dispersion of the walker grows diffusively but with a rate that does not coincide with the classical one @xcite. also, a small amount of decoherence seems to be useful to achieve a quantum walk with a significant speedup @xcite. in this work we will revisit the quantum walk on a cycle (and on a line) considering models where the quantum coin interacts with an environment. the aim of our work is twofold. first we will use phase - space distributions (i.e., discrete wigner functions) to represent the quantum state of the walker. the use of such distributions in the context of quantum computation has been proposed in ref. @xcite, where some general features about the behavior of quantum algorithms in phase space were noticed. a phase - space representation is natural in the case of quantum walks, where both position and momentum play a natural role. our second goal is to study the true nature of the transition from quantum to classical in this kind of model. we will show that models where the environment is coupled to the coin are not able to induce a complete transition to classicality. this is a consequence of the fact that the preferred observable selected by the environment is the momentum of the walker. this observable, which is the generator of discrete translations in position, plays the role of the `` pointer observable '' of the system @xcite. therefore, as we will see, the interaction with the environment being very efficient in suppressing interference between pointer states preserves the quantum interference between superpositions of eigenstates of the conjugate observable to momentum (i.e., position). again, the use of phase - space representation of quantum states will be helpful in developing an intuitive picture of the effect of decoherence in this context. the paper is organized as follows : in sec. ii we review some basic aspects of the quantum walk on the cycle. we also introduce there the phase - space representation of quantum states for the quantum walk and discuss some of the main properties of the discrete wigner functions for this system. in sec. iii we introduce a simple decoherence model and show the main consequences on the quantum walk algorithm. in sec. iv we present a summary and our conclusions.
Quantum walks and their phase-space representation
the quantum walks on an infinite line or in a cycle with @xmath1 sites are simple enough systems to be exactly solvable. for the infinite line the exact solution was presented in ref. the case of the cycle was first solved in ref. however, the exact expressions are involved enough to require numerical evaluation to study their main features. here we will review the main properties of this system presenting them in a way which prepares the ground to use phase - space representation for quantum states (we will focus on the case of a cycle, the results for the line can be recovered from ours with @xmath2). for a quantum walk in a cycle of @xmath1 sites, the hilbert space is @xmath3, where @xmath4 is the space of states of the walker (an @xmath1-dimensional hilbert space) and @xmath5 is the two dimensional hilbert space of the quantum coin (a spin @xmath0). the algorithm is defined by a unitary evolution operator which is the iteration of the following map : @xmath6 here @xmath7 is the hadamard operator acting on the hilbert space of the quantum coin [@xmath8, @xmath9 being the usual @xmath10 pauli matrices]. the operator @xmath11 is the cyclic translation operator for the walker, which in the position basis is defined as @xmath12 (mod @xmath1). it is worth stressing that the operator @xmath13 is nothing but a spin controlled translation acting as @xmath14. so, the map @xmath15 consists of a spin controlled translation preceded by a coin toss, which is implemented by the hadamard operation (the use of the hadamard operator in this context is not essential and can be replaced by almost any unitary operator on the coin @xcite). the notion of phase - space is natural in the context of this kind of quantum walk. in fact, the position eigenstates @xmath16 form a basis of the walkers hilbert space @xmath4. the conjugate basis is the so called momentum basis @xmath17. position and momentum bases are related by the discrete fourier transform, i.e., @xmath18 the cyclic translation operator @xmath11 that plays a central role in the quantum walk is diagonal in momentum basis since @xmath19. this simply indicates that momentum is nothing but the generator of finite translations. as a consequence of this, the total unitary operator defining the quantum walk algorithm is also diagonal in such basis. this fact, which was noticed before by several authors, enables a simple exact solution of the quantum walk dynamics. indeed, we can write the initial state of the system using the momentum basis of the walker as (below @xmath20 denotes the total density matrix of the system formed by the walker and the coin) @xmath21 where @xmath22 is the initial state of the quantum coin (we assume that the initial state of the combined system is a product, but this assumption can be relaxed). after @xmath23 iterations of the quantum map the reduced density matrix of the walker (denoted as @xmath24) is @xmath25,\end{aligned}\]] where the operator @xmath26 is defined as @xmath27 all the temporal dependence is contained in the function @xmath28 which can be exactly computed in a straightforward way : one should first expand the initial state @xmath22 in a (nonorthogonal) basis of operators of the form @xmath29 (@xmath30), where @xmath31 are the eigenstates of the operator @xmath26 (i.e., @xmath32). the explicit expressions for the eigenstates @xmath31 and the eigenvalues @xmath33 will not be given here since they can be found in the literature (see, for example, @xcite). after doing this the evolution of the quantum state is fully determined by the equation @xmath34 below we will describe the properties of this solution using a phase - space representation for the quantum state of the system. wigner functions @xcite are a powerful tool to represent the state and the evolution of a quantum system. for systems with a finite - dimensional hilbert space, the discrete version of wigner functions was introduced using different methods (see ref. @xcite). we will follow the approach and notation used in ref. @xcite, where these phase - space distributions were applied to study properties of quantum algorithms. for completeness, we will give here the necessary definitions and outline some of the most remarkable properties of the discrete wigner functions. for a system with an @xmath1-dimensional hilbert space the discrete wigner function can be defined as the components of the density matrix in a basis of operators defined as @xmath35 these are the so called phase - space point operators. they are defined in terms of the cyclic shift @xmath11 (which in the position basis acts as @xmath36), the reflection operator @xmath37 (which in the position basis acts as @xmath38), and the momentum shift @xmath39 (which generates cyclic displacements in the momentum basis, i.e., @xmath40). phase - space operators are unitary, hermitian and form a complete orthogonal basis of the space of operators (they are orthogonal in the hilbert schmidt inner product since they satisfy that @xmath41=n\delta_{q, q'}\delta_{p, p'}$]). expanding the quantum state in the @xmath42 basis as @xmath43 the coefficients @xmath44 are the discrete wigner functions of the quantum state, which are obtained as @xmath45. \label{wigner}\]] this function has three remarkable properties that almost give it the status of a probability distribution. the first two properties are evident : wigner functions are real numbers (a consequence of the hermiticity of the phase - space operators) and they provide a complete description of the quantum state (a consequence of the completeness of the basis of such operators). the third property is less obvious : marginal probability distributions can be obtained by adding values of the wigner function along lines in phase - space. for this to be possible, it turns out that the phase - space has to be defined as a grid of @xmath46 points where @xmath44 is given at each point precisely by eq. ([wigner]). thus, adding the values of the wigner function over all points satisfying the condition @xmath47 one gets the probability to detect an eigenstate of the operator @xmath48 with eigenvalue @xmath49 (the sum is equal to zero if such eigenvalue does not exist). in particular, adding the wigner function along vertical lines @xmath50 one obtains the probability to detect eigenstates of the operator @xmath51, with eigenvalues given by @xmath49. these numbers are equal to zero if @xmath52 is odd and they are equal to the probability for measuring the position eigenstate @xmath53 when @xmath52 is even. complementary, adding values of the wigner function along horizontal lines enables us to compute the probability to detect a momentum eigenstate. a final remark about properties of the discrete wigner function is in order. figure [wigner - basicas] shows the wigner function of a position eigenstate @xmath54 and of a superposition of two position eigenstates, such as @xmath55. as we see, in the first case the function is positive on a vertical line located at @xmath56 and is oscillatory on a vertical line located at @xmath57. the interpretation of these oscillations is clear. it is well known that wigner functions display oscillatory regions whenever there is interference between two pieces of a wave packet. in this case, the cyclic boundary conditions we are imposing (that originate from the fact that @xmath11 and @xmath39 are cyclic shift operators) generate a mirror image for every phase - space point. thus, the oscillating strip can be interpreted as the interference between the positive strip and its mirror image. for the case of a quantum state which is a superposition of two position eigenstates, we observe two positive vertical lines with the usual interference fringes in between them. all these vertical lines have their corresponding oscillatory counterparts originated from the boundary conditions which are located at a distance @xmath1. in what follows we will show wigner functions for typical states of a quantum walker. for a localized state (up) and for a delocalized superposition state (down). the dimension of the hilbert space is @xmath58. the horizontal (vertical) axis corresponds to position (momentum). color code is such that red (blue) regions correspond to positive (negative) values while white corresponds to zero (@xmath59, @xmath60, @xmath61).,title="fig:",scaledwidth=28.0%] + for a localized state (up) and for a delocalized superposition state (down). the dimension of the hilbert space is @xmath58. the horizontal (vertical) axis corresponds to position (momentum). color code is such that red (blue) regions correspond to positive (negative) values while white corresponds to zero (@xmath59, @xmath60, @xmath61).,title="fig:",scaledwidth=28.0%] ''' '' for an initial state where the walker starts at a given position and the spin is initially unbiased [@xmath62, the behavior of the quantum walker starts to deviate from its classical counterpart at early times (in this paper we will only consider unbiased initial states for the quantum coin). the phase - space representation of the state is shown in fig. [wigner - local - nodeco] and makes evident that a peculiar pattern of quantum interference fringes develops between the different pieces of the wave packet. the consequence of these interference effect is evident also when one looks at the probability distribution for different positions. this distribution is shown in fig. [probab - nodeco] and has been previously studied in the literature (see @xcite). like its classical counterpart, at a given time @xmath23 the state initially located at @xmath63 has support only on states @xmath64 satisfying that @xmath65 adds up to an even number. however, in general the quantum distribution differs from the classical one, exhibiting peaks located at @xmath66 and a plateau of height @xmath67 around @xmath63. after some time the wigner function of the quantum walker develops a shape that resembles a thread, as it is clear in the pictures. for this reason we will call this a thread state. + iterations for an initially localized walker with unbiased spin. @xmath68, @xmath69. we only plot the function for sites such that @xmath65 adds to an even number (solid), and also include the classical distribution (dotted).,scaledwidth=30.0%] ''' '' + a) @xmath70 + b) @xmath71 + c) @xmath72 ''' '' it is also interesting to analyze the evolution of the quantum walk for delocalized initial states. in particular, we will consider an initial state that is a coherent superposition of two position eigenstates (whose wigner function was already displayed in fig. [wigner - basicas]). we find a wigner function that develops into a sum of two threads with a region in between where interference fringes are evident. this is displayed in fig. [wigner - delocal - nodeco]. some properties of the quantum walk for this kind of delocalized initial states were analyzed in ref. @xcite where it was noticed that the asymptotic probability distribution can be rather different from the one obtained from a localized initial state. below, we will show that the process of decoherence affects localized and delocalized initial states in a rather different way. + a) @xmath70 + b) @xmath73 + c) @xmath74 ''' ''
Decoherence and the transition from quantum to classical
we will consider a quantum walk where the quantum coin couples to an environment. to describe such coupling we will use a model which was introduced and studied in detail in ref. @xcite. in that paper it was shown that one can mimic the coupling to an external environment by a sequence of random rotations applied to the quantum coin, which have the effect of scrambling the spin polarization. more precisely, these kicks will be generated by the evolution operator @xmath75 where the angles @xmath76 take random values and @xmath77 is a fixed vector specifying the rotation axis. the virtue of this model is not only its simplicity but also the fact that can be experimentally implemented in a controllable manner using, for example, nmr techniques. after the application of one step of the quantum walk algorithm and one kick the evolution of the total system is @xmath78 to obtain a closed expression for the reduced density matrix of the walker for an ensemble of realizations of the random variables @xmath76 we follow the method proposed in ref. @xcite (see ref. @xcite for a similar approach) : assuming that these angles are randomly distributed in an interval (@xmath79,@xmath80), this density matrix is @xmath81\ \label{rodet}\end{aligned}\]] expanding the initial state in the momentum basis as before enables us to simplify this expression. in fact, after doing this one can integrate over the random variables to find @xmath82. \label{fkkp}\end{aligned}\]] here @xmath83 is a superoperator acting on the spin state (depending on direction @xmath77 of the kicks) as @xmath84 where @xmath85 is a parameter related to the strength of the noise (notice that @xmath86 corresponds to unitary evolution, i.e., to @xmath87). one can find a simple matrix representation for the superoperator @xmath83 for different choices of the rotation axis by writing @xmath22 in the basis formed by the identity and the pauli operators. in the appendix we show the explicit form of this matrix representation, which is helpful in finding exact and numerical solutions to the problem. in what follows we will show results for the case @xmath88 (the other cases are qualitatively similar). to find the state of the walker at arbitrary times we simply need to find eigenstates and eigenvalues of the superoperator @xmath83. this can always be done numerically and also analytically in the interesting case of @xmath89, which can be denoted as `` total decoherence ''. in such case, the exact solution turns out to be @xmath90 several features of the decoherence effect are evident in the above formula. the environment produces a tendency towards diagonalization of the density matrix of the walker in the momentum basis (matrix elements with @xmath91 are maximally suppressed). the decay of nondiagonal elements is exponential in time, as already discussed in ref. it is also clear that momentum eigenstates are not affected by the interaction since they are eigenstates of the full evolution [in fact, from eqs. ([fkkp]) and ([superop]) follows @xmath92. in this sense they are perfect pointer states for this model. in what follows we will present results concerning the evolution of several initial quantum states. + a) @xmath93 + b) @xmath94 + c) @xmath95 ''' '' + a) @xmath96 + b) @xmath97 + c) @xmath98 ''' '' the effect of decoherence on the evolution of states which are initially localized in position has been analyzed elsewhere @xcite. as shown in fig. [wigner - local - deco], the wigner function of the evolved quantum state gradually loses its oscillatory nature. thus, instead of a thread state the interaction with the environment gradually produces a mixed state with a binomial distribution in the position direction (which has an approximately gaussian shape for large @xmath23) but remains constant along the momentum direction. it is worth noticing that for any value of @xmath99 the resulting state has support only on position eigenstates satisfying that the sum @xmath100 is equal to an even number, as it was already pointed out for both the classical distribution (@xmath89) and the purely quantum one (@xmath86) in the preceding section. it is interesting to notice that the process of decoherence has a rather simple interpretation when represented in phase - space : decoherence in phase - space is roughly equivalent to diffusion in the position direction. this is not unexpected : in fact, in ordinary quantum brownian motion models a coupling to the environment through position (momentum) gives rise to a momentum (position) diffusion term in the evolution equation for the wigner function. the situation here is quite similar, since the walker effectively couples with the environment through its momentum. this is indeed the case because the environment interacts with the quantum coin which is itself coupled with the walker through the displacement operator which is diagonal in the momentum basis. therefore, the decoherence effect on the wigner function is expected to correspond to diffusion along the position direction. ''' '' as noticed before, if one considers initial states where the walker is localized in a well defined position, one can see that the probability distribution for the different positions of the walker gradually tends to the classical one by increasing the coupling strength @xmath101 from @xmath87 (no kicks) to @xmath102. this is shown in fig. [probab - deco]. + from the above analysis we could be tempted to conclude that the interaction between the quantum coin and the environment induces the classicalization of the walker. however, this is not the case. the process of decoherence induced in this way is not complete. this is most clearly seen by analyzing how it is that the interaction with the environment affects initial states of the quantum walker which are not initially localized. in fig. [wigner - delocal - deco] we show the wigner function of an initially delocalized state (shown in fig. [wigner - basicas]) under full decoherence. we can clearly see that decoherence does not erase all quantum interference effects. in fact, as mentioned above, the interaction with the environment induces diffusion along the position direction. therefore, interference fringes which are aligned along the position direction are immune to decoherence. thus, the final state one obtains from a superposition of two position eigenstates is not the mixture of two binomial states but a coherent superposition of them. this peculiar behavior is easily understood by noticing that this is a simple consequence of the fact that momentum eigenstates are pointer states : decoherence is effective in destroying superpositions of pointer states but highly inefficient in destroying superpositions of eigenstates of the conjugate observable (position). by analyzing the entropy of the reduced density matrix of the walker one can get a more quantitative measure of the degree of decoherence achieved as a consequence of the interaction with the environment. for convenience we will not examine the von neumann entropy @xmath103 but concentrate on the linear entropy defined as @xmath104)$], which is easier to calculate. this entropy provides a lower bound to @xmath103 @xcite. it is possible to show that almost no entropy is produced by the decay of the coherence present in the initially delocalized superposition state. in fact, this can be seen by comparing the entropy produced from the initially delocalized superposition and the one originated from an initial state in which the walker is prepared in an equally weighted mixture of two positions. these entropies can be seen in fig. [entropy]. the initial entropy of the mixture is 1 bit [@xmath105. it is quite clear from the curves shown in such figure that the entropy arising from the initial mixture remains to be 1 bit higher than the one originated from the initial coherent superposition. thus, the quantum coherence present in the initial state is robust under the interaction with the environment and does not decay at all. [entropy] shows another interesting feature : one would naively expect a monotonic dependence of the entropy with the coupling to the environment (which is parametrized by @xmath101). however this is not the case since the curves in fig. [entropy] intersect. this peculiar effect is made more evident in fig. [entropy - rate] where we study the entropy at a fixed time as a function of the coupling strength. in this figure a clear indication of a nontrivial behavior is seen : for early times the entropy grows slowly with @xmath101 and exhibits a flat plateau for large values of @xmath101. however, as time progresses a peak develops : the largest value of entropy at a given time is not achieved by the largest coupling. to the contrary, the largest entropy is attained by an intermediate coupling @xmath106, whose value decreases with time. the fact that for a given time the maximal entropy is not achieved by the maximal coupling to the environment is counterintuitive. as entropy is a measure of the spread of a distribution, this strange behavior can be rephrased as a manifestation of the counterintuitive fact that the decohered state (which is approximately diagonal in position basis) has a probability distribution that is more spread for @xmath107 than for @xmath108. a possible explanation for this peculiar behavior is the following : for high values of the coupling to the environment the state rapidly becomes classical and the spread in position grows diffusively, as in the classical random walk. when the coupling to the environment is not strong, our result seems to indicate that the state of the walker remains `` quantum '' for a longer time during which it spreads at a rate faster than classical. when this quantum state finally decoheres it may end up having a larger entropy than the one attained for high coupling simply because it is spread over a wider range of positions. we speculate that there could be a relation between this peculiar feature and the properties that make some degree of decoherence useful for quantum walks as discussed by kendon and tregenna in @xcite. the value of the @xmath106 introduced above depends on both @xmath1 and @xmath23 and could be related to the position of the minima reported in refs. @xcite. for example, in a cycle regime (@xmath109) @xmath106 diminishes with increasing @xmath1 as it is also the case for the position of the minima of the so called _ quantum mixing time _ @xcite. (solid), @xmath110 (dotted), @xmath111 (dash dotted), @xmath112 (dashed). the top (down) plot corresponds to an initial state which is an equally weighted mixture (superposition) of two position eigenstates. @xmath113.,title="fig:",scaledwidth=32.0%] + (solid), @xmath110 (dotted), @xmath111 (dash dotted), @xmath112 (dashed). the top (down) plot corresponds to an initial state which is an equally weighted mixture (superposition) of two position eigenstates. @xmath113.,title="fig:",scaledwidth=32.0%] ''' '' for various values of time : @xmath114 (solid), @xmath115 (dashed), @xmath116 (dotted), @xmath117 (@xmath118), @xmath119 (@xmath120), @xmath121 (dash dotted). the initial state of the walker is well localized and the initial state of the quantum coin is unbiased. @xmath122 and all the curves are below the saturation regime [@xmath123. it is evident that, after some time, the maximum value of the entropy is not achieved by the maximum value of the coupling strength.,scaledwidth=32.0%] ''' ''
Conclusions
the use of phase - space representation enables us to develop some intuition about the nature of the decoherence process in the kind of quantum walk analyzed in this paper. by coupling the quantum coin to an environment we obtain a decoherence model which is roughly equivalent to position diffusion. as we mentioned above, this is a natural result whose origin can be traced back to the way in which the system effectively couples to the environment (via the momentum operator). the relation between decoherence and position diffusion can also be established by analyzing in more detail the structure of the superoperator @xmath83 [given in eq. ([superop])]. let us consider the form of the superoperator after @xmath23 iterations. if we use eq. ([superop]) we can easily see that, as each iteration doubles the number of terms, we will have an expression with @xmath124 terms each one of which has pauli operators applied at different times. to obtain the function @xmath125 one should compute the trace over the quantum coin. in each of the @xmath124 terms we can move the pauli operator @xmath126 towards the outside of the expression and cancel them due to the cyclic property of the trace. for the case @xmath88 it is easy to show that the only remaining effect of the pauli operators (that in this case anti commute with the hadamard operator) is to reverse the direction of the rotation in @xmath26 defined in eq. ([mofk]). the final expression can be shown to be @xmath127 where @xmath128 and @xmath129 (we use the convention @xmath130). therefore, the superoperator is the sum of @xmath124 terms each one of which contains a contribution that is identical to that of a quantum walk where the direction of the walker is chosen at random after the first step. each of the @xmath124 terms is labeled by a @xmath23bit string @xmath131 and corresponds to a quantum walk where the direction of the @xmath132th step (@xmath133) is reversed if and only if @xmath134 is odd. in the limit of total decoherence each of these terms has equal weight. therefore the final state is simply the average over an ensemble where each member corresponds to each of the @xmath135 possible choices of two directions (forward or backward) for the @xmath136 steps (notice that the direction of the first step is not affected by the decoherence model we chose). for this type of decoherence it is clear that the quantum walk becomes a random walk. the relation between decoherence and position diffusion is quite evident in this way. it is worth pointing out that similar models of decoherence were considered in ref. @xcite in a different context. other decoherence models have been analyzed for quantum walks @xcite, where the effective coupling to the environment is through the position observable. in such case, we expect decoherence to correspond to diffusion along the momentum direction. combining the two types of decoherence (i.e., considering coupling to the environment via the quantum coin and the position of the quantum walker) the initial state corresponding to a superposition of two positions would finally decay into a mixture of two binomial states (see ref. @xcite for similar results obtained when studying decoherence models with a natural phase - space representation in a finite quantum system evolving under various quantum maps). the above conclusions are generic for any model in which decoherence is due to the coupling of the quantum coin to an environment. an interesting class of models, based on the use of quantum multi baker maps @xcite, has been studied. in such models one replaces the quantum coin with a quantum system with a higher - dimensional hilbert space. the total space of states is then @xmath137. here @xmath138 is the dimensionality of the system which plays the role of the quantum coin and is considered to be an even number (@xmath139, so that we can always consider @xmath140). the dynamics for a quantum multi baker map is defined in terms of the unitary operator [that replaces eq. ([onestep])] : @xmath141 where @xmath142 is the unitary operator defining the so called `` quantum baker map '' (see refs. @xcite) and @xmath143 is a pauli operator acting on the hilbert space @xmath5 (the most significant qubit of the internal space @xmath140). the properties of the operator @xmath142 have been widely studied in the literature @xcite : the map faithfully represents a classically chaotic system (in the large @xmath138 limit). from the point of view of the quantum walker the situation is quite similar to the one we studied in this paper. one can describe a quantum multi baker map as an ordinary quantum walk where the quantum coin (whose hilbert space is @xmath5) interacts with an environment (whose hilbert space is @xmath144). the interaction is modeled by the quantum baker map acting on the total internal hilbert space @xmath145, which also replaces the usual hadamard step in eq. ([onestep]). as the quantum baker map is chaotic, the state of the quantum coin will be roughly randomized after each iteration. thus, the effect should be similar to the one we described here (where the quantum coin is subject to a noisy evolution). however, after a large number of iterations (of the order of @xmath138) all the possible orthogonal directions available in the internal space of the quantum coin would have been explored. one should therefore expect that this model will stop being effective in producing decoherence after such time. recent studies of quantum multi baker systems agree with these expectations (see ref. @xcite where a transition from diffusive to ballistic behavior after a time of the order of @xmath138 has been analyzed). in any case, based on the results of our work we believe that in quantum multi baker systems the relative stability of initially delocalized superpositions will also be observable. we thank marcos saraceno and augusto roncaglia for useful discussions and assistance during several stages of this work. was partially supported by funds from fundacin antorchas, anpcyt and ubacyt.
Density matrix under decoherence
a matrix representation for the superoperator @xmath83 can be obtained for @xmath146. we will write this superoperator in the basis formed by the identity and the pauli matrices (we use the standard ordering of the basis as @xmath147). in such basis the matrix of @xmath83 is @xmath148 where [@xmath149. notice they all converge to the same matrix when @xmath86 (no decoherence). + to compute eq. ([fkkplanteodeco]) we need to diagonalize @xmath83. although the matrix representation of the superoperator is rather sparse, the eigenvalues and eigenvectors are quite cumbersome for an arbitrary value of @xmath99 (including @xmath86, no decoherence), so it was more convenient to use numerical techniques. however, for the special case of complete decoherence (@xmath89) it is possible to obtain a simple formula and the final result for @xmath125. the result for @xmath154 is given in eq. ([fkkyz]). n. shenvi, j. kempe, k. birgitta whaley, phys. a * 67 *, 052307 (2003). v. kendon and b. tregenna, phys. a * 67 *, 042315 (2003). brun, h.a. carteret, a. ambainis, phys. a * 67 *, 032304 (2003). v. kendon and b. tregenna, e - print quant - ph/0301182. for a review see j.p. paz and w.h. zurek, in _ coherent matter waves, les houches session lxxii _, edited by r. kaiser, c. westbrook and f. david (springer verlag, berlin, 2001), pp. 533614, e - print quant - ph/0010011. w. dr, r. raussendorf, v.m. kendon, h.j. briegel, phys. a * 66 *, 052319 (2002). j. du, h. li, x. xu, m. shi, j. wu, x. zhou, r. han, phys. rev. a * 67 *, 042316 (2003). c. miquel, j.p. paz, m. saraceno, phys. a * 65 *, 062309 (2002). zurek, phys. d, * 24 *, 15161525 (1981). a. nayak and a. vishwanath, e - print quant - ph/0010117 ; see also a. ambainis, e. bach, a. nayak, a. vishwanath, j. watrous, proc. 33rd annual acm symposium on theory of computing (stoc 2001), 3749. m. scully, m. hillery and e. wigner, phys. rep. * 106 *, 121 (1984). g. teklemariam, e.m. fortunato, c.c. lpez, j. emerson, j.p. paz, t.f. havel, d.g. cory, phys. a * 67 *, 062316 (2003). the linear entropy @xmath155)$] shares two important properties of the von neumann entropy @xmath156 $] : (@xmath157) it is non - negative and vanishes only if @xmath20 is pure ; (@xmath158) its maximum value is @xmath159 and it is reached iff @xmath160. in addition to this, @xmath161 provides a lower bound to @xmath103 since @xmath162. the use of this entropy can be found, for example, in @xcite and also in i. garca - mata, m. saraceno, m.e. spina, phys. lett. * 91 *, 064101 (2003) ; d. monteoliva and j.p. paz, phys. * 85 *, 3373 (2000). | we analyze the quantum walk on a cycle using discrete wigner functions as a way to represent the states and the evolution of the walker.
the method provides some insight on the nature of the interference effects that make quantum and classical walks different.
we also study the behavior of the system when the quantum coin carried by the walker interacts with an environment.
we show that for this system quantum coherence is robust for initially delocalized states of the walker.
the use of phase - space representation enables us to develop an intuitive description of the nature of the decoherence process in this system. | quant-ph0308104 |
Introduction
carbon chains play a central role in the chemistry and spectroscopy of interstellar space. the detection of cyanopolyacetylenes in dense interstellar clouds @xcite led to the suggestion by @xcite that carbon chain species be considered as candidates for the diffuse interstellar bands (dibs), which are found in the 4000 - 8500 spectral region of stars reddened by interstellar dust @xcite. since then, many other molecules with a carbon chain backbone have been identified at radio frequencies in dense clouds @xcite. meanwhile, advances in laboratory measurements have provided an understanding of the types and sizes of carbon chains which have strong electronic transitions in the dib range @xcite. it is thus somewhat surprising that as yet among the bare carbon species only diatomic c@xmath11 has been identified in interstellar clouds where dib are detected. the 4052 electronic band system of c@xmath0 was first detected in comets @xcite and then in circumstellar shells by infrared spectroscopy @xcite. most recently c@xmath0 was identified in a dense cloud using sub - mm measurements of its low frequency bending mode and n@xmath12(c@xmath0)@xmath13 @xmath7 was estimated @xcite. @xcite established an upper limit of 5@xmath1410@xmath15 @xmath7 for the column density of c@xmath0 in the direction of @xmath2 oph, some two orders of magnitude lower than that set by @xcite. @xcite made a tentative detection of c@xmath0 towards an eighth magnitude star in the same part of the sky, hd 147889, at a column density of 4@xmath1410@xmath16 @xmath7. unfortunately, the star turned out to be a (previously unknown) double - lined spectroscopic binary which limited their sensitivity. this letter presents the detection of c@xmath0 towards three stars and infers the column densities in the diffuse clouds. although c@xmath17, c@xmath18 and c@xmath19 were not detected, upper limits for their column densities are estimated. diatomic species, such as ch, cn, c@xmath11, and ch@xmath20, have been detected towards two of the stars chosen and their column densities are considered standards with which to compare models for the physical and chemical processes in diffuse regions @xcite. the four bare carbon chains, c@xmath0, c@xmath17, c@xmath18 and c@xmath19, were selected for the present search because gas phase electronic transitions for these species have been identified in the laboratory in the 4000 - 5500 region and their oscillator strengths are known (see table [results]).
The observations
observations of the reddened stars @xmath2 oph (hd 149757), @xmath2 per (hd 24398) and 20 aql (hd 179406) were made with the gecko echellette spectrograph on 2000 july 16 and 19, fiber fed from the cassegrain focus of the canada - france - hawaii 3.6-m telescope (cfht) @xcite. all three stars have a visual extinction, a@xmath21, near 1 and were chosen because they are bright with sharp interstellar k i lines indicating either single clouds or little doppler distortion (in the case of @xmath2 oph, @xcite resolved the c@xmath11 at 8756 into two close velocity components separated by 1.1 km s@xmath22). the detector was a rear illuminated eev1 ccd (13.5 @xmath23m@xmath24 pixels) and the spectral regions were centered at 4047 in the 14th order, and at 5060 and 5400 in the 11th and 10th orders, respectively. the ultraviolet gecko prism was used to isolate the 14th order, the blue grism for the 11th order, while the stock cfht filter # 1515 was used for the 10th order observations. individual spectra had exposure times ranging from 5 to 20 minutes and were obtained with continuous fiber agitation to overcome modal noise. the resulting combined spectra for the individual stars at each wavelength had unusually high signal - to - noise ratios (s / n@xmath5800 - 4000) for ccd observations. the th / ar comparison arc spectra, taken before and after each spectrograph wavelength reconfiguration, had a typical fwhm of 2.8 pixels, which corresponds to resolutions of @xmath25 = 121000, 113000 and 101000 at 4047, 5060 and 5400 , respectively. processing of the spectra was conventional. groups of biases were taken several times throughout each night and at each grating setting a series of flat - field spectra of a quartz - iodide lamp were recorded. the biases and appropriate flats were averaged and used to remove the zero - level offset and pixel - to - pixel sensitivity variations of the ccd. one - dimensional spectra were extracted using standard iraf routines. spectra of vega and other unreddened stars were used to search for contaminating telluric water vapor lines and stellar photospheric features. heliocentric corrections were applied to each spectrum. the observations are summarised in table [observations] which lists exposure times and s / n per pixel for each spectral region. the final column gives the radial velocities measured from the interstellar k i 4044.1 and 4047.2 lines. these velocities have been applied to each spectrum to put the interstellar features on a laboratory scale.
The results
the @xmath3 origin band of c@xmath0 is quite clearly detected towards all three stars. figure [figure1] compares the observed spectra with a simulated c@xmath0 absorption spectrum based on the spectrograph resolution and assuming a boltzmann distribution in the ground state rotational levels with a temperature of 80 k. the continuum noise level in the observations is @xmath50.1 %. low order polynomials have been applied to the stellar data to give a level continuum (base line) and, in the case of @xmath2 per, a weak, broad stellar feature at 4053.2 has been removed. residual broad features in the final spectra are only a few tenths of a percent deep, much less than in the original, and they in no way mask the sharp c@xmath0 lines. in the simulation the rotational line intensities were calculated using the hnl - london factors, while the line positions were taken from the laboratory measurements (this avoids the problem of a perturbation affecting the low @xmath4 ground state levels which is not accounted for by the fitted spectroscopic constants). the individual rotational p, q and r lines are clearly resolved in the spectra of all three stars. table [lines] lists the observed positions and equivalent widths of each rotational line assigned in the spectra of the three stars. the table also gives the corresponding positions measured for these transitions in the laboratory (gausset et al. the positions of some 30 lines in the spectrum of @xmath2 oph agree with the laboratory data to within 0.1 cm@xmath22 providing an unambiguous identification of c@xmath0 in these diffuse clouds. figures [figurec2p], [figurec2 m], and [figurec3 m] show the results of equivalent searches for @xmath27 origin band of c@xmath17 at 5078.1 (@xmath2 oph only), the @xmath28 origin band of c@xmath18 at 5415.9 and the @xmath29 origin band of c@xmath19 at 4040.4, together with simulated spectra for these transitions at 80 k based upon the published spectroscopic constants (@xcite, @xcite, @xcite). in the cases of c@xmath30 and c@xmath31 the linewidth was assumed to be determined by the spectrograph resolution. for c@xmath32 the excited state has been identified as a short - lived feshbach resonance and the measured natural linewidth of @xmath51 cm@xmath22 is employed in the simulation. weak telluric lines were been removed from the c@xmath17 and c@xmath18 observations using standard procedures. unlike the 4050 region for c@xmath0, each of these spectral regions is contaminated by weak stellar features. nonetheless, for the c@xmath17 and c@xmath18 ions, there are no sharp features corresponding to the rotational lines in the simulations. on the other hand, for c@xmath32 there are features in the spectrum of @xmath2 per and (much less convincing) in the magnified plot for @xmath2 oph which appear to coincide in position and shape with the simulated band heads. it is unlikely that these are due to c@xmath32 because the stellar lines in @xmath2 per have exactly the same shape as the coincident features. for @xmath2 oph the whole c@xmath32 spectrum sits within a weak stellar feature (the lines are broadened by rapid rotation @xmath5400 km s@xmath22). the photospheric lines in @xmath2 oph show nonradial pulsation ` ripples'which will be washed out to some extent by the long exposure time employed. the spectrum of 20 aql, which normally has the strongest interstellar lines of the three, is free of stellar features but has no features coincident with the c@xmath32 simulation. it is concluded that, while interstellar c@xmath32 might be absent for 20 aql and present for the other two stars, it is more likely that in the latter cases the features are instead stellar. table [results] gives the measured equivalent widths for the most intense c@xmath0 line (q(8)) for each star together with an 1@xmath33 error estimate. for c@xmath17, c@xmath18 and c@xmath19, 3@xmath33 detection limits are given. the 1@xmath33 level errors and detection limits are derived from : where the 1@xmath33 limiting equivalent width, @xmath35, and the fwhm of the feature, @xmath36, are both measured in , the spectrograph dispersion, @xmath37, in pixel@xmath22, and s / n is the signal to noise per pixel. from the simulations, @xmath36 = 0.045, 1.0, 0.13 and 0.045 for c@xmath0, c@xmath19, c@xmath17 and c@xmath18, respectively. in the case of c@xmath0, equivalent widths, @xmath35, were determined for each rotational line (varying between 0.3 - 2.7@xmath38) and, in combination with the transition oscillator strength, f@xmath39, and hnl - london factors, the column densities, @xmath40(c@xmath0), of each rotational level (@xmath4) in the ground electronic state were calculated @xcite. in cases where several rotational lines originating from the same level were assigned (e.g. p(8), q(8), r(8)) the mean of the determined column densities was taken. figure [figure5] shows a boltzmann plot of ln(@xmath41) vs. the rotational energy @xcite where the slope is inversely proportional to the rotational temperature. among the lowest rotational levels (@xmath4214) the populations are reasonably approximated by a distribution at 50 - 70 k, whereas the higher rotational levels correspond to a temperature of 200 - 300 k. the simulation in figure 1 uses 80 k as this represents an average temperature for the entire rotational population and allows both the high and low @xmath4 lines to be identified. the high temperature component of the distribution is apparent in the astronomical spectra where the r band head and the higher q lines are more intense than in the simulation (figure 1). @xcite also found a bimodal population distribution for c@xmath11 in diffuse clouds, with similar characteristic temperatures for the low and high @xmath4 values. the lower temperature is interpreted as the kinetic energy of the cloud and for both c@xmath11 and c@xmath0 the values obtained are comparable to those used in models of diffuse clouds @xcite. the higher temperature component is attributed to repopulation of the levels in the ground electronic state by radiative pumping from excited states. in the case of c@xmath0 it is expected that both the @xmath43 state and the higher lying @xmath44 state will contribute to the radiative pumping. the sensitivity of these measurements is such that @xmath40(c@xmath0) in the range 0.2 - 2@xmath45 @xmath7 is determined for rotational levels up to @xmath4=30. the @xmath40 values were summed to give the estimated lower limits in the range 1 - 2@xmath6 @xmath7 for the total column density, @xmath46(c@xmath0), in table [results]. a previous search for c@xmath0 in the direction of @xmath2 oph did not identify the molecule @xcite. it is unclear why this was the case as, in the light of the present observations, the signal - to - noise quoted for these measurements was adequate and the upper limit given was some thirty times lower than the column density reported here. the present measurements for the column densities of c@xmath0 are of the same order of magnitude as the tentative estimate for a translucent cloud @xcite and can be compared to those of other polyatomic molecules observed in diffuse interstellar clouds. column densities (also towards @xmath2 oph) in the 10@xmath47 @xmath7 range have been inferred for hco@xmath20, c@xmath11h and c@xmath0h@xmath11 from observations in the mm region by @xcite, while h@xmath48 has been identified in diffuse regions and n@xmath12(h@xmath48) estimated @xmath49 @xmath7 by @xcite. column densities of c@xmath11 towards @xmath2 oph and @xmath2 per have been determined in the 2 - 3@xmath50 @xmath7 range by @xcite. a current model of the diffuse clouds by @xcite predicts n@xmath51(c@xmath11)/n@xmath12(c@xmath0) @xmath5 20 (on a 10@xmath52 year time scale), implying n@xmath12(c@xmath0)@xmath53 @xmath7, in agreement with the values deduced from the table [results]. the main production route to c@xmath0 is presumed to be the dissociative recombination process : c@xmath0h@xmath20 + @xmath54 @xmath55 c@xmath0 + h, where c@xmath0h@xmath20 is produced from smaller species by c@xmath20 ion insertion. under conditions where ultra - violet radiation penetrates, photodissociation of c@xmath0 takes place at a threshold of 1653. as the strong @xmath44@xmath56@xmath57 electronic transition of c@xmath0 is predicted to occur around 1700 @xcite, the dissociation process, c@xmath0 @xmath55 c@xmath11 + c, may be an important destruction pathway in diffuse clouds. although only upper limits for the column densities of c@xmath19, c@xmath17 and c@xmath18 could be presently established, these species are of interest as small ionic carbon fragments play a crucial role in the ion molecule schemes for diffuse cloud chemistry @xcite. the c@xmath17 ion is the only bare carbon cation for which the gas phase electronic spectrum is known @xcite. in diffuse clouds it is supposed to be the product of the fundamental step : c@xmath20 + ch @xmath55 c@xmath17 + h. its main destruction mechanism is hydrogenation : c@xmath17 + h@xmath11 @xmath55 c@xmath11h@xmath20 + h, which dominates over recombination with electrons in diffuse regions. the diffuse cloud model @xcite predicts a c@xmath17 abundance a factor of 10@xmath58 lower than c@xmath11, implying a column density @xmath59 @xmath7, in accord with the upper limit in table [results]. published models for diffuse regions do not include the smallest pure carbon anions, c@xmath18 and c@xmath19 in their reaction libraries. unlike c@xmath17, c@xmath18 does not react with h@xmath11 so its main destruction mechanism is expected to be photodetachment. the similar rotational line widths and oscillator strengths of the c@xmath18 and c@xmath17 transitions lead to similar upper limits for their total column densities. the width of the unresolved bands for c@xmath19 and the presence of weak stellar features in this spectral region means that a higher column density of this ion could have escaped detection. the detection of c@xmath0 provides a powerful incentive for the laboratory study of the electronic transitions of longer carbon chains in the gas phase with the aim of comparison with dib data. the question as to what types and sizes of carbon chains will have strong transitions in the 4000 - 9000 range has already been answered : for example, @xmath44@xmath56@xmath57 transitions of the odd - number bare chains, c@xmath60, @xmath61=8 - 30 @xcite. the existence of linear carbon chains up to c@xmath62 has been confirmed by the observation of their electronic spectra in neon matrices @xcite. as the oscillator strength scales almost linearly with the length of the molecule, one can expect f@xmath6310 - 20 for these carbon chains. with such an oscillator strength, a species with a column density @xmath64 @xmath7 would be enough to give rise to a strong dib, with an equivalent width of 1. in view of the column density @xmath510@xmath16 @xmath7 for c@xmath0 determined in this work for three diffuse clouds, this appears to be a reasonable expectation. the support of the swiss national science foundation (project no. 20 - 055285.98), the canadian natural sciences and engineering research council and the national research council of canada is gratefully acknowledged. the authors thank the staff of the cfht for their care in setting up the fiber feed and agitator, thereby making such high signal - to - noise spectra possible. clccccccccl @xmath2 per & b1 ib & 2.85 & 0.28 & 4800 & 1200 & & & 2700 & 1900 & + 13.91 @xmath650.24 + @xmath2 oph & o9.5 v & 2.56 & 0.30 & 5400 & 2400 & 5400 & 4000 & 3000 & 2200 & @xmath5614.53 @xmath650.18 + 20 aql & b3 v & 5.36 & 0.27 & 10800 & 800 & & & 8400 & 900 & @xmath5612.53 @xmath650.08 + cc|lc|lc|lc 4049.784 & r(22) & & & 4049.770@xmath66&1.016 & 4049.795@xmath66&1.708 + 49.770 & r(24) & & & 49.770@xmath66&1.016 & 49.795@xmath66&1.708 + 49.810 & r(20) & 4049.782@xmath66 & 1.658 & 49.807@xmath66&1.162 & & + 49.784 & r(26) & 49.782@xmath66 & 1.658 & 49.807@xmath66&1.162 & & + 49.861 & r(18) & 49.865 & 0.309 & 49.877 & 0.511 & 49.865 & 0.821 + 49.963 & r(16) & 49.959 & 0.726 & 49.961 & 0.773 & & + 50.081 & r(14) & 50.079 & 0.792 & 50.091 & 0.759 & & + 50.206 & r(12) & 50.198 & 0.920 & 50.198 & 0.680 & 50.211 & 1.165 + 50.337 & r(10) & 50.329 & 1.034 & 50.342 & 0.679 & 50.339 & 1.450 + 50.495 & r(8) & 50.483 & 1.525 & 50.497 & 0.383 & 50.489 & 2.204 + 50.670 & r(6) & 50.669 & 1.562 & 50.662 & 0.943 & 50.667 & 2.330 + 50.865 & r(4) & 50.863 & 1.018 & 50.864 & 1.068 & 50.853 & 2.225 + 51.069 & r(2) & 51.073 & 0.896 & & & 51.074 & 1.657 + 51.309 & r(0) & 51.267@xmath67 & 0.371 & & & 51.386@xmath67&0.730 + 51.461 & q(2) & 51.457 & 1.045 & 51.457 & 0.923 & 51.455 & 1.180 + 51.521 & q(4) & 51.515 & 2.187 & 51.518 & 0.752 & 51.506 & 1.876 + 51.590 & q(6) & 51.586 & 2.719 & 51.588 & 2.183 & 51.583 & 1.965 + 51.682 & q(8) & 51.679 & 2.336 & 51.680 & 2.016 & 51.669 & 2.229 + 51.793 & q(10) & 51.788 & 2.138 & 51.795 & 2.294 & 51.787 & 2.508 + 51.929 & q(12) & 51.930 & 1.060 & 51.922 & 2.020 & 51.929 & 2.186 + 52.062 & p(4) & 52.074@xmath66&1.266 & 52.085@xmath66&1.043 & 52.065@xmath66 & 3.130 + 52.089 & q(14) & 52.074@xmath66&1.266 & 52.085@xmath66&1.043 & 52.065@xmath66 & 3.130 + 52.271 & q(16) & 52.262 & 1.217 & & & 52.271 & 1.821 + 52.424 & p(6) & & & & & 52.456 & 2.499 + 52.473 & q(18) & 52.466 & 1.233 & & & & + 52.698 & q(20) & 52.701 & 0.923 & & & & + 52.792 & p(8) & 52.784 & 0.605 & & & 52.772 & 1.592 + 52.900 & q(22) & 52.929 & 0.985 & & & 52.939 & 1.122 + 53.180 & p(10) & 53.197@xmath66&1.883 & & & + 53.207 & q(24) & 53.197@xmath66&1.883 & & & + 53.590 & p(12) & 53.593 & 0.678 & & & 53.588 & 1.237 + 53.795 & q(28) & 53.786 & 0.469 & & & 53.781 & 1.098 + 54.112 & q(30) & 54.113 & 0.523 & & & & + 54.459 & p(16) & 54.445 & 0.870 & & & & + 54.908 & p(18) & 54.904 & 1.122 & & & & + c@xmath0&@xmath3&4051.5&0.016&@xmath2 oph&2.3@xmath650.1&20&1.6 + & & & & 20 aql&2.2@xmath650.3&19&2.0 + & & & & @xmath2 per & 2.0@xmath650.2&17&1.0 + c@xmath19&@xmath29&4040.4&0.04&@xmath2 oph&6.0&&@xmath680.3 + & & & & 20 aql&20&&@xmath681.2 + & & & & @xmath2 per&12&&@xmath680.7 + c@xmath17&@xmath27&5066.9&0.025&@xmath2 oph&0.35&@xmath681.1&@xmath680.04 + c@xmath18&@xmath28&5408.6&0.044&@xmath2 oph & 0.30&@xmath680.5&@xmath680.02 + & & & & 20 aql&0.55&@xmath680.9&@xmath680.03 + & & & & @xmath2 per&0.35&@xmath680.6&@xmath680.02 + | the smallest polyatomic carbon chain, c@xmath0, has been identified in interstellar clouds (a@xmath11 mag) towards @xmath2 ophiuchi, 20 aquilae, and @xmath2 persei by detection of the origin band in its @xmath3 electronic transition, near 4052.
individual rotational lines were resolved up to @xmath4=30 enabling the rotational level column densities and temperature distributions to be determined.
the inferred limits for the total column densities (@xmath51 to 2@xmath6 @xmath7) offer a strong incentive to laboratory and astrophysical searches for the longer carbon chains.
concurrent searches for c@xmath8, c@xmath9 and c@xmath10 were negative but provide sensitive estimates for their maximum column densities. | astro-ph0102449 |
Introduction
radiation pressure on spectral lines (line force) driving a wind from an accretion disk is the most promising hydrodynamical (hd) scenario for agn outflows. within this framework, a wind is launched from the disk by the local disk radiation at radii where the disk radiation is mostly in the ultraviolet (uv ; e.g., shlosman, vitello & shaviv 1985 ; murray et al. 1995, mcgv hereafter). such a wind is continuous and has mass loss rate and velocity which are are capable of explaining the blueshifted absorption lines observed from many agn, if the ionization state is suitable. such winds have the desirable feature that they do not rely on unobservable forces or fields for their motive power. however, detailed tests of this idea via modelling is challenging. the wind dynamics are coupled to the ionization and opacity properties of the gas, and the location and nature of the radiation sources is not well understood. previous models by us (proga stone and kallman 2000, hereafter psk) relied on uv radiation from the black hole in order to make the flow steady and to impart a strong radial component. in this paper we report new calculations which relax this requirement, and which demonstrate quantitative consistency between disk winds and observations. uv driven disk winds in agn are motivated by analogy with winds from hot stars, which have been explored in great detail (eg. lamers and cassinelli 1999 and references therein). but they differ in important ways, including the role of rotation near a keplerian disk, the non - uniform disk temperature distribution, and the influence of the strong x - ray flux from the inner disk and black hole. rotation acts to make the vertical component of gravity increase with height near the disk plane, and can also affect the wind trajectory when height is comparable to the radius. close to the disk plane the wind is driven by photons emitted locally, but at greater heights the radiation spectrum and the driving momentum depend on position ; the increase in disk temperature at small radius can result in a net outward component to the momentum. the wind density and mass loss rate can be estimated from the observed agn uv luminosity. when compared with the x - ray flux from the agn, the density is low enough that the gas is predicted to be highly ionized. if so, the opacity needed for efficient driving and for line formation will be very small and the wind will fail. on the other hand, estimates for the column density in the radial direction close to the disk surface are high enough that a portion of the wind can be shielded from this ionization. viable models for disk winds must account self - consistently for the wind driving, ionization, and self shielding. mcgv pointed this out and postulated the existence of ` hitchhiking'gas which is not driven by uv but which provides the shielding. their calculation was a one - dimensional time independent quasi - radial flow. proga, stone & kallman (2000, psk hereafter) were able to to relax some of the mcgv assumptions, and explored consequences of a radiation driven disk wind model by performing 2.5 dimensional time - dependent hd simulations. the most challenging component of fluid dynamics calculations is treatment of the radiative transfer. in the case of disk wind calculations this limitation forces an inexact treatment of the radiative transfer, by assuming parameterized values for the x - ray and uv continuum opacities. psk found that a disk accreting onto a @xmath0 black hole at the rate of 1.1 @xmath1 (@xmath2) can launch a wind at @xmath3 cm from the central engine. the x - rays from the central object are significantly attenuated by the disk atmosphere so they can not prevent the uv radiation from pushing matter away from the disk. however, the x - rays can overionize the gas in the flow high above the disk and decrease the wind velocity. for a reasonable x - ray opacity, e.g., @xmath4, the disk wind can be accelerated by the central uv radiation to velocities of up to 15000 @xmath5 at a distance of @xmath6 cm from the central engine. the covering factor of the disk wind is @xmath7. the wind is unsteady and consists of an opaque, slow vertical flow near the disk that is bounded on the polar side by a high - velocity stream. a typical column density radially through the fast stream is a few @xmath8 so that the stream is optically thin to the uv radiation but optically thick to the x - rays. a key issue in modeling radiation pressure driven winds from disks is the relative importance of the driving by photons emanating from near the black hole, i.e. at radii smaller than the computational grid, compared with photons emitted by the disk at radii within the computational grid. models for cv disk winds, which resemble agn disk winds except for an absence of strong x - rays, show that strong steady flows require a central source of uv. motivated by this, psk as well as mcgv considered the situation where the central radiation in the uv accelerates the gas in the radial direction. a consequence of this assumption is that an efficiently driven wind must attenuate the x - rays, in order to avoid over - ionization, but must transmit the uv, in order to be driven. this places an intrinsic limit on the radial column density, and hence on the mass loss rate, and can also influence the geometry of the flow. our goal in the present paper is to explore what happens when the assumption of a strong central flux is relaxed in the psk models, and to examine other consequences of the line - driven disk wind model. in particular, we check how robust is the disk wind solution and whether the solution predicts synthetic line profiles capable of reproducing line profiles observed in agn. in 2, we summarize the key elements of our calculations. we present the results from disk wind simulations and synthetic line profile calculations in 2.1 and 2.2, respectively. the paper ends in 4, with our conclusions and discussion.
Method
in this paper we extend work by psk by relaxing some of their assumptions and simplifications. our 2.5-dimensional hd numerical method is in most respects as described by psk. here we only describe the key elements of the method and list the changes we made. we refer a reader to psk for details. as in psk, we apply line - driven stellar wind models (castor, abbott & klein 1975, hereafter cak) to winds driven from accretion disks (see also proga, stone & drew 1998 ; 1999, psd98 and psd99 hereafter). we specify the radiation field of the disk by assuming that the temperature follows the radial profile of the optically thick accretion disk (shakura & sunyaev 1973). we account for some of the effects of photoionization. in particular, we calculate the gas temperature assuming that the gas is optically thin to its own cooling radiation. we also take into account some of the effects of photoionization on the line force. in particular, we compute the parameters of the line force using a value of the photoionization parameter, @xmath9 and the analytical formulae due to stevens & kallman (1991). this procedure is computationally efficient and gives approximate estimates for the number and opacity distribution of spectral lines for a given @xmath9 without detail information about the ionization state (see stevens & kallman 1991). additionally, we take into account the attenuation of the x - ray radiation by computing the x - ray optical depth in the radial direction. we modify psk s method as follows : (i) we decrease the inner and outer radius of the computational domain (see 3 for details) ; (ii) to compute the radiation force due to lines, we use the intensity of the radiation integrated over the uv band only (i.e., between @xmath10 and @xmath11, see also proga 2003). (in psk we took the uv flux to be a constant fraction of the flux from the black hole) ; (iii) we exclude entirely the central object radiation force ; (iv) we compute the x - ray optical depth in the radial direction by considering only electron scattering (in psk we took the x - ray opacity to be @xmath12 @xmath13) ; (v) we do not explicitly treat continuum opacity for photons emitted from the disk, although the self - shielding of the lines is taken into account by our sobolev treatment of the line force (in psk we allowed the disk radiation to be attenuated in the radial direction with an opacity @xmath14 g@xmath15 @xmath13). the changes in our calculations extend the range of validity of calculations of psk. the decrease of the inner radius of the computation domain allows us to capture as much as possible of the uv emitting disk within our grid. the second change makes our calculations as self - consistent as possible without making them computationally prohibitive, i.e., there is no wind launching from very large radii where the disk is cold and radiates few uv photons, and from very small radii where the disk effective temperature or gas temperature, or both, are too high to permit enough spectral lines. the third change was motivated by the results from psk : we want to test the importance of attenuation of the central uv flux as a limit to the acceleration of gas to high velocities, and also the extent to which the central flux is needed to make a strong and steady flow. by neglecting the radiation force from the central object we are exploring relatively unfavorable conditions for wind acceleration. the only force that can accelerate the wind is the radiation force due to the disk. the fourth change was also motivated by our wish to explore relatively unfavorable conditions for attenuation of x - rays by disk winds. using this treatment, our method gives a lower limit for the optical depth to ionizing photons. the fifth change to the calculations means that we allow all disk photons emitted toward a given point in the wind to reach this point. this may correspond to an overestimate of the disk radiation force. however, exact transfer of uv continuum is computationally prohibitive and is treated in the sobolev approximation even in one dimensional hot star wind models. we briefly discuss some of the consequences of our changes in .4.
Results
we present here results for the same model parameters as in psk with only two exceptions (see below). we assume the mass of the non - rotating black hole, @xmath16. to determine the radiation field from the disk, we assume the mass accretion rate @xmath17 m@xmath18 yr@xmath15. these system parameters yield the disk luminosity, @xmath19 of 50% of the eddington luminosity and the disk inner radius, @xmath20 cm, where @xmath21 is the schwarzschild radius of a black hole. the radiation field from the central engine is specified by its luminosity @xmath22 with @xmath23 set to 1, its fraction in the uv band (@xmath24) and its fraction in the x - rays (@xmath25). in psk, @xmath26 and these are the only changes to the model parameters we made. the spectral energy distribution of the ionizing radiation is not well known, our choice of values for @xmath27 and @xmath28 is guided by the results from zheng et al. (1997) and laor et al. (1997) [e.g., see figure 6 in the latter, for a comparison of the spectral energy distributions found for various samples of qsos.] as mentioned in 2, we include the central radiation only as a source of ionizing photons and exclude its contribution to the radiation force. there are two definitions of the photoionization parameter used in literature : @xmath9 and @xmath29 (e.g., krolik 1999). the former is based on the ionizing flux while the latter on the number density of the ionizing photons. for the adopted spectral energy distribution, the conversion between the two is as follows : @xmath30. our computational domain is defined to occupy the radial range @xmath31, and the angular range @xmath32. the @xmath33 domain is discretized into zones. our numerical resolution consists of 100 and 140 zones in the @xmath34 and @xmath35 directions, respectively. we use fixed zone size ratios, @xmath36 and @xmath37. figure 1 shows the instantaneous density, temperature and photoionization parameter distributions and the poloidal velocity field of the model. the wind speed at the outer boundary is 2000 to 12000 km @xmath38. this corresponds to a dynamical time of @xmath390.2 yrs for the material at @xmath40. figure 1 shows results at the end of the simulation after @xmath41 years. although the flow is still weakly time - dependent after this time has elapsed, the gross properties of the flow (e.g., the mass loss rate and the radial velocity at the outer boundary), settle down to steady time - averages. as in the flow found by psk, the wind has 3 components : (i) a hot, low density flow in the polar region (ii) a dense, warm and fast equatorial outflow from the disk, (iii) a transitional zone in which the disk outflow is hot and struggles to escape the system. the main difference with the results of psk is that here the transitional zone is much more prominent, and it occupies a large fraction of the computational domain. in the polar region, the density is very small and close to the lower limit that we set on the grid, i.e., @xmath42 g @xmath43. the line force is negligible because the matter is highly ionized as indicated by a very large photoionization parameter (@xmath44). the gas temperature is close to the compton temperature of the x - ray radiation. the matter in the polar region is pulled by the gravity from the outer boundary, which is an artifact of the boundary conditions (e.g., we do not model a jet which is likely to propagate through the polar region). however, this region is sometimes filled with gas which is launched from the disk with a large vertical velocity in an episodic manner. the outflowing wind itself has distinct two components : hot and warm. the two components are launched from the disk by the line force. the gas density at the disk atmosphere and wind base is @xmath45, so the photoionization parameter is low (log(@xmath9)@xmath46 - 5) despite the strong central radiation. however as the flow from the inner part of the disk is accelerated by the line force its density decreases. for the disk wind launched at small radii (@xmath47), this decrease of the density causes an increase in the gas temperature and the photoionization parameter. as this process proceeds the gas becomes fully ionized and loses all driving lines. the wind speed when this happens is not generally great enough to allow the gas to escape, and it tends to fall back toward the disk. this ` failed wind'has an effect on the remainder of the flow, both due to its shielding effect on the central x - rays and due to its pressure as it falls toward the disk. the disk wind launched at large radii does not become overionized downstream despite a density decrease. the photoionization parameter in this wind is kept low because of the large column density toward the source of the ionizing photons. consequently, the outer disk wind is not only launched but also accelerated by the disk line force. we find that the outer wind becomes radial relatively close to the disk. this may seem surprising because we set to zero the radiation force in the radial direction due to the central object. there are three effects responsible for the flow to be equatorial : (i) overionization of the vertical part of the wind by the central object radiation ; (ii) the ram pressure of the failed wind launched at smaller radii that falls back after it is overionized ; and (iii) the line force due to disk photons emitted interior to a given point in the wind. the two first effects dominate at smaller radii and close to the interface between the hot and warm flow whereas the third effect dominates above the disk at large radii where the fore - shortening of the disk radiation weakens. the flows we find are time dependent and have density and velocity fluctuations of @xmath48 even after the flow has evolved for many dynamical times. in particular, there are occasionally regions along the equator with the gas densities lower than those for hydrostatic equilibrium. from those low density regions, the radiation force can launch a wind with an acceleration length scale shorter than for a cak - like wind [the line force is a strong function of the gas density (cak)]. in fact, we observe that the line acceleration can be so efficient that some gas can reach velocities comparable to the escape velocity before it is eventually overionized. this flow behavior manifests itself as erratic high velocity ejections of gas from the inner disk. figure 2 presents the run of the density, radial velocity, mass flux density, accumulated mass loss rate, photoionization parameter and column density as a function of the polar angle, @xmath35, at the outer boundary, @xmath49 cm from figure 1. the accumulated mass loss rate, @xmath50 and the column density, @xmath51 are computed as in psk (see eq. 13 in psd 98 and eq. 24 in psk for the definitions of @xmath50 and @xmath51, respectively). the gas density is very low, i.e., @xmath52, for @xmath35 between @xmath53 and @xmath54. then the gas density increases with @xmath35 between @xmath54 and @xmath55 to the level of a few @xmath56. for @xmath57, the density decreases to the level of @xmath52 at @xmath58. then the density increases again with @xmath35 and reaches the maximum of a few @xmath56 at @xmath59. this is followed by a decrease of the density with the minimum of a few @xmath60 at @xmath61. for, @xmath62, the density sharply increases, as might be expected of a density profile determined by hydrostatic equilibrium. the radial velocity is @xmath63 km @xmath38 for @xmath64 and has a broad flat maximum at the level of @xmath65 km @xmath38 for @xmath35 between @xmath54 and @xmath66. for @xmath35 between @xmath66 and @xmath67, the radial velocity is negative and then gradually increases to @xmath68 km @xmath38 at @xmath69. this is followed by a drop of @xmath70 to a very small value near the equator. we note that @xmath70 stays nearly constant at the level of @xmath71 km @xmath38 for @xmath72. the accumulated mass loss rate is negligible for @xmath73 because of the very low gas density and velocity. for @xmath74, @xmath75 increases to @xmath76 at @xmath77. for @xmath35 between @xmath55 and @xmath78, the accumulated mass loss rate stays nearly constant. for @xmath79, the accumulated mass loss rate increases to @xmath80 at @xmath81. the column density in the wind increases gradually with @xmath35. in particular, it increases from @xmath82 at @xmath83 to @xmath84 at @xmath85. for @xmath86, @xmath51 continues to increase and indicates that the wind is optically thick to electron scattering at @xmath87 for the central object radiation. the column densities greater than @xmath88 are effectively infinite, and represent complete obscuration of the central object. winds with such high @xmath51 can still be radiation driven owing to the difference between the column the disk radiation sees and the column the central source radiation sees. we expect that the disk wind obscuration can change the ratio between the number of all qsos and bal qsos, i.e., qsos, viewed by an observer almost edge - on, would not be detected or identified as qso. the photoionization parameter is very high, @xmath44, for @xmath40 because of the very low density in the polar region. however, over next @xmath89, @xmath9 drops by many order of magnitude owing to the increase of the column density. the @xmath35 profiles of the flow properties show that the overoinized outflow can contribute to the total mass loss rate at the level of a few per cent. our new simulations of agn disk wind confirm the main result from psk : the disk atmosphere can shield itself so that the local disk radiation can launch gas off the disk photosphere. here we find that this results holds even when the condition for shielding are unfavorable : the ionizing photons are only scattered on electrons but not absorbed by the shielding gas. our new simulations also provide some new insights to the acceleration. in particular, we find the local disk force suffices to accelerate the disk wind to high velocities in the radial direction provided the wind does not change the geometry of the disk radiation by continuum scattering and absorption processes. the main difference between our disk wind solution and the solution presented in psk is the properties and behavior of the hot component of the disk outflow which struggles to escape the system. in psk, this component occupied a relatively narrow @xmath35 range (@xmath90) above the disk wind, whereas here @xmath91. the key reason for this difference is the strength of the radial line force relative to the latitudinal line force acting on the inner disk wind. in psk, the radial force is strong and accelerates the inner wind on a relatively long length scale (of order of the wind launch radius). a small fraction of the inner flow is overionized and fails to develop into a strong outflow. this overionized gas falls on the freshly launched disk wind and introduces weak perturbations to the disk wind so that the wind is not significantly slowed down nor disrupted. here, on the other hand, the radial force is relatively weak and the flow from the innermost disk tends to be vertical. the flow geometry is important for how long the flow can stay shielded : for the vertical geometry, the flow height can be larger than the height of the shielding gas whereas for the radial geometry, the flow height can stay smaller than the height of the shielding gas even at very large distances from the launch point. therefore vertical flow is more susceptible to overionization than radial flow ; a much larger fraction of the inner flow is overionized in the vertical flow than in the radial flow. in our simulations the overionized flow is relatively dense and as it falls on the disk it disrupts the freshly launched the disk wind. we note that the mass loss rate in our simulations is smaller than in psk s simulation by a factor of @xmath39 2. this is an unexpected result because here we allow launching a wind from radii smaller than those in psk and a simple scaling law for line - driven disk wind models suggests that mass loss rate should increase with decreasing radius (e.g., psd98 ; proga & kallman 2002). additionally, one can argue that because of this scaling, overionization should become less of a problem with decreasing radius (proga & kallman 2002). what is then responsible for this low mass loss rate? our analysis of the flow properties and time evolution shows that at the early phase of the evolution, the mass flux density is consistent with the expectations based on the above arguments. however, as the inner disk wind becomes overionized and starts falling back toward the equator the mass loss rate decreases because the average gas density at the wind base increases which reduces the driving line force. additionally, the infalling gas is dense and can significantly slow the outflow which it encounters on its way toward the disk. there are then a few factors which reduce the mass loss rate but the primary factor is the flow overionization. our time - dependent disk wind solutions differ from those found by psd98, in which it was shown that cv disk winds are unsteady in time when the wind is driven solely by disk radiation. in the latter, the flow could also fall back on the disk, but it was not strong enough to cause a significant decrease in the mass loss rate. furthermore, in the psd98 simulations there was no radiation capable of fully ionizing the wind so that regions of relatively low density, above the falling gas, could be accelerated to high velocities and contribute to the total mass loss rate. simulations presented in psk and here can serve as a proof - of - concept for the radiation driven disk wind model of outflows in agn. some consequences of the radiation driven disk wind model have been explored in the context of cv and low mass x - ray binaries (lmxb ; e.g., see psd98 and proga & kallman 2002, respectively). generally, the insights from those models allow us to explain systems with strong evidence for uv absorbing disk wind such as cvs and systems without uv absorbing disk wind such as lmxb. in the case of cvs, the model also predicts uv line profiles which are capable of reproducing observations (proga 2003). in this section, we present synthetic line profiles predict by our model and briefly discuss their relevance to agn observations. we present line profiles for the c iv @xmath92 transition. we compute the profiles by integrating flux from the entire disk at a given orientation (i.e., as in the dynamical models, the continuum source has a finite size). the line profile calculation performed here are exactly as in proga et al. (2002) with the exception that we set the c iv abundance to one in the regions with the gas temperature from 8000 k to 420000 k and to zero elsewhere. for the adopted model parameters, the upper limit for the gas density corresponds to the disk effective temperature at @xmath93. we refer a reader to proga et al. (2002) for details on line - profile calculations. the c iv abundance could be in principle, computed self - consistently for given wind structure and radiation field. however, we save this for a future paper and focus on gaining some insights from the multidimensional models of disk winds and predicted profiles. to illustrate the effect of viewing angle on the line profiles, figure 3 shows line profiles for five inclination angles : @xmath94, and @xmath95. we limit our line profile calculation to the absorption contribution only. modeling of the emission component depends on the influence of thermal line emission, and we will discuss this in a future paper. figure 3 clearly shows that the line profiles are very sensitive to the inclination angle. in particular, for @xmath96 the line absorption is broad and nearly symmetric whereas for @xmath97 just @xmath54 higher a strong blueshifted absorption dominates the line profile (compare figs 3c and 3d). we note that for @xmath98 the profiles is more symmetric than for @xmath96. in the latter case, the blueshifted absorption is somewhat stronger than the redshifted absorption. overall our line profiles are representative of generic line profiles for a bipolar wind from a rotating disk (e.g. proga et al. 2002 and references therein). for example, for a relatively high inclination angles (@xmath99) the profiles are affected by the absorption in the rotating base of the wind, so blueshifted as well as redshifted absorption is present (see proga et al. 2002, and references therein). the presence of the expanding wind manifests itself in the profiles by a blueshifted absorption for inclination angles where the continuum source is viewed through the expanding wind. in our wind model, the disk wind with a low @xmath9 (i.e., high c iv abundance) has a half - opening angle of @xmath100. therefore only for @xmath101, our line profiles show blueshifted absorption stronger than redshifted absorption. in general, the line profiles predicted by our models are similar to those predicted by the model for cv winds (proga 2003). however, there are two important differences : (i) for low @xmath97, cv wind models predict a blueshifted absorption whereas here little absorption is present at those angles and (ii) for all @xmath97, cv wind models predict that the maximum of the blueshifted absorption is line - centered whereas here the maximum can be blueshifted by as much as @xmath102 km @xmath38. the reason for these differences is the fact that in cv wind models, the base of the wind covers the whole continuum source whereas here the continuum source extends inside the computational domain and therefore the continuum is only partially covered by the flow. our line profiles are consistent with many of the properties of profiles observed in broad absorption line qsos (e.g., korista et al. 1993 ; hall et al 2002). in particular, the synthetic line profiles can be as strong and broad (up to 30000 km @xmath38) as those observed. additionally, for certain inclination angles the maximum of the blueshifted absorption is far from the line center - this is consistent with so - called detached troughs observed in some qsos. some of the profiles shown in fig. 3 display prominent redshifted absorption. in observed lines, this component is likely to be affected by the emission contribution to the line (e.g., proga 2003). the line emission can fill in the redshifted absorption and also some of the highly blueshifted absorption produced by the wind near the equator where the rotational velocity exceeds the expansion velocity. for example, proga (2003) showed that for high inclinations (@xmath103), the profile has double - peaked emission (figs. 1e and 1j in proga 2003).
Conclusions and discussion
in this paper we have presented numerical simulations of the radiation - driven disk wind scenario for agn outflows. these models provide insights into the wind s capability to shield itself from ionizing radiation. we performed axisymmetric time - dependent hydrodynamic calculations similar to those of proga, stone and kallman (2000). we studied the robustness of the radiation launching and acceleration of the wind for relatively unfavorable conditions. in particular, we have taken into account the central engine radiation as a source of ionizing photons but neglected its contribution to the radiation force. additionally, we have accounted for the attenuation of the x - ray radiation by computing the x - ray optical depth in the radial direction assuming that only electron scattering contributes to the opacity. our new simulations confirm the main result from psk : the disk atmosphere can shield itself from external x - rays so that the local disk radiation can launch gas off the disk photosphere. we have also found that the local disk force suffices to accelerate the disk wind to high velocities in the radial direction. this is true provided the wind does not change significantly the geometry of the disk radiation by continuum scattering and absorption processes, and we discuss the plausibility of this requirement. overall our models are consistent with observations of broad absorption line (bal) qsos. additionally, synthetic profiles of a typical resonance ultraviolet (uv) line, predicted by our models, promise to reproduce observed bals and it is likely that the model can account for other agn outflows observed in uv as well as in x - rays. strong ionizing radiation has long been recognized as a serious challenge to theories of agn outflows absorbing the uv radiation. radiation - driven disk wind models are particularly affected by this problem because the overionization of the wind can prevent the wind from being driven. but in other models, including a magnetic one, overionization also needs to be addressed owing to the requirement for sufficient un - ionized gas to account for the observed lines. here we find that an indirect effect of overionization can reduce but not prevent the development of a wind, for our choice of model parameters : the so - called failed inner disk wind. if the density of the failed wind is relatively high then the ram pressure of such a wind can seriously slow down the disk wind which otherwise could be driven by radiation. we emphasize that this damaging effect of the failed wind can be important to explain agn winds within the radiation driven disk wind scenario rather than being used as an argument against the scenario. in particular, in this scenario the accretion disk, with parameters as adopted here (see 3), should produce a powerful wind from radii as small as @xmath104 because the radiation at those radii is mostly in the uv. however, such a wind would be too fast to be consistent with agn observations. one of plausible solutions of this problem is to allow for the energy dissipation at small radii to occur not only inside but also outside the disk (i.e., in the disk corona). consequently, the uv radiation from the innermost disk would be reduced while ionizing radiation would be increased. the two factors would reduce launching of the wind. a complimentary solution is that the failed wind can disrupt development of a wind up to a radius of a few @xmath105. we have performed many simulations to study the sensitivity of disk wind solutions to the black hole mass and disk luminosity. these results will appear in proga et al.(2004) (in preparation). here we only mention that the wind solution is very sensitive to the luminosity compared to the eddington limit. in particular, we find that for the model parameters as in .3 but with the luminosity reduced by a factor of 5 (from 50% to 10% of the eddington limit) there is no disk wind. the primary reason for this luminosity sensitivity is the fact that the mass flux density of the wind decreases strongly with decreasing disk luminosity and the wind is more subject to overionization (e.g. proga & kallman 2002). our simulations show also that for a fixed eddington fraction (e.g., 50%) it is easier to produce a wind for @xmath106 than for @xmath107. this result is a consequence of the decrease of the uv contribution to the disk total radiation with decreasing mass of the bh in the shakura sunyaev disk model for a fixed eddington fraction. we have also performed many simulations to study the sensitivity of our main results to the model parameters. the most important parameter of the model is the x - ray opacity. this quantity could be computed from first principles for given spectral energy distribution, strength and geometry of the radiation, and the structure and chemical composition of the flow. however, such an approach to @xmath108 s calculations is not computationally feasible and therefore we have explored asymptotic approximations, e.g., assuming that @xmath109 (corresponding to the thomson cross section). as we discussed above, such a simplistic approach yields rather conservative results for the wind strength. in our sensitivity studies, we have computed wind models using a more realistic @xmath108. in particular, we computed @xmath110 based on the analytic fit to the detailed photoionization calculations using the photoionization code xstar. we find that @xmath110 can be expressed a function of the photoionization parameter, @xmath9 in the following way : @xmath111, in units of @xmath112. we found that as expected a wind with this higher @xmath110 is stronger (e.g., in terms of @xmath113 and the opening angle) the model solution described here. the radiative transfer of continuum photons inside the disk wind is as critical as the wind photoionization structure in modeling disk winds. to accelerate the wind to velocities comparable to the escape velocity we should not rely on the radiation which originates interior to the wind zone because the wind will shield itself from this radiation in the same way as it shields itself from the external ionizing radiation. here we explore a configuration where the ionizing radiation comes from a point source located at the center while the driving radiation comes from the extended source, (i.e., the uv emitting disk). calculations of the radiative transfer of the former is relatively simple as it can be approximated by 1d solutions for the optical depth. however, the radiative transfer of the latter is intrinsically 3d and computationally expensive. we argue that the uv continuum photons emitted from the disk into the overlying disk wind can propagate through the disk without large changes of the net radiation flux, hence our assumption that the wind is optically thin to the disk photons. on the other hand, x - rays which are emitted outside the disk wind suffer many scatterings and absorptions once they reach the wind zone. consequently, the x - rays do not penetrate deep inside the wind zone. the x - ray transfer in the case we consider is analogous to the problem of disk irradiation by an external source whereas the uv transfer is analogous to the problem of radiation transfer in an extended atmosphere. our simulations show how the line force can produce a fast, highly ionized wind : if the density near the wind base is relatively low and the base is shielded over long enough length scale above the disk then the local disk radiation can accelerate gas to relatively high velocities (even exceeding the escape velocity) before the gas becomes overionizated. at large radii, such a wind would be transparent to the uv radiation but may be able to absorb x - rays. we plan to present models for the x - ray properties of disk winds in a future paper. we thank n. arav, m.c., begelman, j.r. gabel, p.b. hall, g.t. richards, and j.m. stone for useful discussions. we also thank an anonymous referee for comments that helped us clarify our presentation. we acknowledge support provided by nasa through grant hst - ar-09947 from the space telescope science institute, which is operated by the association of universities for research in astronomy, inc., under nasa contract nas5 - 26555. dp also acknowledges support from nasa ltsa grants nag5 - 11736 and nag5 - 12867. we acknowledge support from the w. m. keck foundation, which purchased the jila 74-processor keck cluster. | we explore consequences of a radiation driven disk wind model for mass outflows from active galactic nuclei (agn).
we performed axisymmetric time - dependent hydrodynamic calculations using the same computational technique as proga, stone and kallman (2000).
we test the robustness of radiation launching and acceleration of the wind for relatively unfavorable conditions.
in particular, we take into account the central engine radiation as a source of ionizing photons but neglect its contribution to the radiation force.
additionally, we account for the attenuation of the x - ray radiation by computing the x - ray optical depth in the radial direction assuming that only electron scattering contributes to the opacity.
our new simulations confirm the main result from our previous work : the disk atmosphere can shield itself from external x - rays so that the local disk radiation can launch gas off the disk photosphere.
we also find that the local disk force suffices to accelerate the disk wind to high velocities in the radial direction.
this is true provided the wind does not change significantly the geometry of the disk radiation by continuum scattering and absorption processes ; we discuss plausibility of this requirement.
synthetic profiles of a typical resonance ultraviolet line predicted by our models are consistent with observations of broad absorption line (bal) qsos. | astro-ph0408293 |
Introduction
the existence of feeble magnetic fields of several microgauss in our galaxies @xcite, as well as of gigagauss in intense laser - plasma interaction experiments @xcite and of billions of gauss in compact astrophysical objects @xcite (e.g. super dense white dwarfs, neutron stars / magnetars, degenerate stars, supernovae) is well known. the generation mechanisms for seed magnetic fields in cosmic / astrophysical environments are still debated, while the spontaneous generation of magnetic fields in laser - produced plasmas is attributed to the biermann battery @xcite (also referred to as the baroclinic vector containing non - parallel electron density and electron temperature gradients) and to the return electron current from the solid target. computer simulations of laser - fusion plasmas have shown evidence of localized anisotropic electron heating by resonant absorption, which in turn can drive a weibel - like instability resulting in megagauss magnetic fields @xcite. there have also been observations of the weibel instability in high intensity laser - solid interaction experiments @xcite. furthermore, a purely growing weibel instability @xcite, arising from the electron temperature anisotropy (a bi - maxwellian electron distribution function) is also capable of generating magnetic fields and associated shocks @xcite. however, plasmas in the next generation intense laser - solid density plasma experiments @xcite would be very dense. here the equilibrium electron distribution function may assume the form of a deformed fermi - dirac distribution due to the electron heating by intense laser beams. it then turn out that in such dense fermi plasmas, quantum mechanical effects (e.g. the electron tunneling and wave - packet spreading) would play a significant role @xcite. the importance of quantum mechanical effects at nanometer scales has been recognized in the context of quantum diodes @xcite and ultra - small semiconductor devices @xcite. also, recently there have been several developments on fermionic quantum plasmas, involving the addition of a dynamical spin force @xcite, turbulence or coherent structures in degenerate fermi systems @xcite, as well as the coupling between nonlinear langmuir waves and electron holes in quantum plasmas @xcite. the quantum weibel or filamentational instability for non - degenerate systems has been treated in @xcite. in this work, we present an investigation of linear and nonlinear aspects of a novel instability that is driven by equilibrium fermi - dirac electron temperature anisotropic distribution function in a nonrelativistic dense fermi plasma. specifically, we show that the free energy stored in electron temperature anisotropy is coupled to purely growing electromagnetic modes. first, we take the wigner - maxwell system @xcite with an anisotropic fermi - dirac distribution for the analysis of the linearly growing electromagnetic perturbations as a function of the physical parameters. second, we use a fully kinetic simulation to assess the saturation level of the magnetic fields as a function of the growth rate. the treatment is restricted to transverse waves, since the latter are associated with the largest weibel instability growth rates. the nonlinear saturation of the weibel instability for classical, non - degenerate plasmas has been considered elsewhere @xcite.
Basic equations
it is well known @xcite that a dense fermi plasma with isotropic equilibrium distributions does not admit any purely growing linear modes. this can be verified, for instance, from the expression for the imaginary part of the transverse dielectric function, as derived by lindhard @xcite, for a fully degenerate non - relativistic fermi plasma. it can be proven (see eq. (30) of @xcite) that the only exception would be for extremely small wavelengths, so that @xmath0, where @xmath1 is the wave number and @xmath2 the characteristic fermi wave number of the system. however, in this situation the wave would be super - luminal. on the other hand, in a classical vlasov - maxwell plasma containing anisotropic electron distribution function, we have a purely growing weibel instability @xcite, via which dc magnetic fields are created. the electron temperature anisotropy arises due to the heating of the plasma by laser beams @xcite, where there is a signature of the weibel instability as well. in the next generation intense laser - solid density plasma experiments, it is likely that the electrons would be degenerate and that electron temperature anisotropy may develop due to an anisotropic electron heating by intense laser beams via resonant absorption, similar to the classical laser plasma case @xcite. in a dense laser created plasma, quantum effects must play an important role in the context of the weibel instability. in order to keep the closest analogy with the distribution function in phase space for the classical plasma, we shall use the wigner - maxwell formalism for a dense quantum plasma @xcite. here the distribution of the electrons is described by the wigner pseudo - distribution function @xcite, which is related to the fermi - dirac distribution widely used in the random phase approximation @xcite. proceeding with the time evolution equation for the wigner function (or quantum vlasov equation @xcite), we shall derive a modified dispersion relation accounting for a wave - particle duality and an anisotropic wigner distribution function that is appropriate for the fermi plasma. the results are consistent with those of the random phase approximation, in that they reproduce the well - known transverse density linear response function for a fully degenerate fermi plasma @xcite. consider linear transverse waves in a dense quantum plasma composed of the electrons and immobile ions, with @xmath3, where @xmath4 is the wave vector and @xmath5 is the wave electric field. following the standard procedure, one then obtains the general dispersion relation @xcite for the transverse waves of the wigner - maxwell system @xmath6 where @xmath7 is the frequency, @xmath8 is the speed of light in vacuum, @xmath9 is the planck constant divided by @xmath10, @xmath11 the rest electron mass, @xmath12 the unperturbed plasma number density, @xmath13 the electron plasma frequency, @xmath14 is the velocity vector, and @xmath15 is the equilibrium wigner function associated to fermi systems. for spin @xmath16 particles, the equilibrium pseudo distribution function is in the form of a fermi - dirac function. here we allow for velocity anisotropy and express @xmath17 + 1 } \,,\]] where @xmath18 is the chemical potential, @xmath19 the boltzmann constant, and the normalization constant is @xmath20 here @xmath21 is a polylogarithm function @xcite. also, @xmath22 $], where @xmath23 and @xmath24 are related to velocity dispersion in the direction perpendicular and parallel to @xmath25 axis, respectively. in the special case when @xmath26, the usual fermi - dirac equilibrium is recovered. the chemical potential is obtained by solving the normalization condition ([e3]), yielding, in particular, @xmath27 in the limit of zero temperature, where @xmath28 is the fermi energy. also, the fermi - dirac distribution @xmath29, where @xmath4 is the appropriated wave vector in momentum space, is related to the equilibrium wigner function ([e2]) by @xmath30, with the factor @xmath31 coming from spin @xcite. however, these previous works refer to the cases where there is no temperature anisotropy. notice that it has been suggested @xcite that in laser plasmas the weibel instability is responsible for further increase of @xmath24 with time. inserting ([e2]) into ([e1]) and integrating over the perpendicular velocity components, we obtain @xmath32 where @xmath33\bigr\ } \nonumber \\ & - & { \rm li}_{2}\bigl\{-\exp\bigl[-\bigl(\nu - \frac{h}{2}\bigl)^2 + \beta\mu\bigl]\bigr\}\biggr) \,. \nonumber \end{aligned}\]] in ([e5]), @xmath34 is the dilogarithm function @xcite, @xmath35 is a characteristic parameter representing the quantum diffraction effect, @xmath36, and @xmath37, with @xmath38. in the simultaneous limit of a small quantum diffraction effect (@xmath39) and a dilute system (@xmath40), it can be shown that @xmath41, where @xmath42 is the standard plasma dispersion function @xcite. it is important to notice that either ([e1]) or ([e4]) reproduces the transverse dielectric function calculated from the random phase approximation for a fully degenerate quantum plasma @xcite, in the case of an isotropic system. the simple way to verify this equivalence is to put @xmath43 in ([e1]) and then take the limit of zero temperature, so that @xmath44 for @xmath45, and @xmath46 otherwise, where @xmath47 is the fermi velocity. however, to the best of our knowledge, there is no corresponding calculation for an anisotropic fermi equilibrium, as necessary in laser - solid interaction experiments with an anisotropic electron heating due to resonant absorption. also notice that in this letter we are mainly interested in the real part of the transverse response function, since we are looking for purely growing instabilities (@xmath48), so that the contribution from the poles at ([e4]) is not relevant.
Numerical results
(@xmath49) and @xmath50, relevant for the next generation inertially compressed material in intense laser - solid density plasma interaction experiments. the temperature anisotropies are @xmath51 (dashed line), @xmath52 (solid line) and @xmath53 (dotted line), yielding, respectively, @xmath54, @xmath55 and @xmath56., width=321] (@xmath49). here the temperature anisotropy is @xmath52. we used @xmath57 (dashed line), @xmath50 (solid line) and @xmath58 (dotted line), yielding @xmath59, @xmath60 and @xmath61, respectively.,width=321] we next solve our new dispersion relation ([e4]) for a set of parameters that are representative of the next generation laser - solid density plasma interaction experiments. the normalization condition ([e3]) can also be written as @xmath62=(4/3\sqrt{\pi})(\beta{\cal e}_f)^{3/2}$], which is formally the same relation holding for isotropic fermi - dirac equilibria @xcite. for a given value on the product @xmath63 and the density, this relation yields the value @xmath64, from which the temperatures @xmath65 and @xmath66 can be calculated, if we know @xmath67. consider only purely growing modes. from the definition ([e5]), one can show that @xmath68 when @xmath69 for a finite wavenumber @xmath1. from ([e4]) we then obtain the maximum wavenumber for instability as @xmath70. when @xmath71, the range of unstable wavenumbers shrinks to zero. in figs. 1 and 2, we have used the electron number density @xmath72, which can be obtained in laser - driven compression schemes. the growth rate for different values on @xmath67 is displayed in fig. we see that the maximum unstable wavenumber is @xmath70, as predicted, and that the maximum growth rate occurs at @xmath73. figure 1 also reveals that the maximum growth rate of the instability is almost linearly proportional to @xmath74. in fig. 2, we have varied the product @xmath63, which is a measure of the degeneracy of the quantum plasma. we see that for @xmath63 larger than @xmath75, the instability reaches a limiting value, which is independent of the temperature, while thermal effects start to play an important role for @xmath63 of the order unity. (top panel) and @xmath76 (bottom panel) as a function of space and time, for @xmath50 and @xmath77. the magnetic field has been normalized by @xmath78. we see a nonlinear saturation of the magnetic field components at an amplitude of @xmath79.,width=321] , over the simulation box (top panel), and the logarithm of the magnetic field maximum (bottom panel) as a function of time, for @xmath52 and @xmath50. the magnetic field has been normalized by @xmath78. from the logarithmic slope of the magnetic field in the linear regime we find @xmath80., width=321] from several numerical solutions of the linear dispersion relation, we have been able to deduce an approximate scaling law for the instability as @xmath81, where the constant is approximately @xmath82. using that @xmath83, we have @xmath84 for the maximum growth rate of the weibel instability in a degenerate fermi plasma. this scaling law, where the growth rate depends on the fermi energy and the temperature anisotropy, should be compared to that of a classical plasma @xcite, where the growth rate depends on the thermal energy and the temperature anisotropy. for a maxwellian plasma, it has been found @xcite that the weibel instability saturates nonlinearly once the magnetic bounce frequency @xmath85 has increased to a value comparable to the linear growth rate. in order to assess the nonlinear behavior of the weibel instability for a degenerate plasma, we have carried out a kinetic simulation of the wigner - maxwell system. we have assumed that the quantum diffraction effect is small, so that the simulation of the wigner equation can be approximated by simulations of the vlasov equation by means of an electromagnetic vlasov code @xcite. as an initial condition for the simulation, we used the distribution function (2). in order to give a seed for any instability, the plasma density was perturbed with low - frequency fluctuations (random numbers). the results are displayed in figs. 3 and 4, for the parameters @xmath50 and @xmath52, corresponding to the solid lines in figs. 1 and 2. figure 3 shows the magnetic field components as a function of space and time. we see that the magnetic field initially grows, and saturates to steady state magnetic field fluctuations with an amplitude of @xmath86. the maximum amplitude of the magnetic field over the simulation box as a function of time is shown in fig. 4, where we see that the magnetic field saturates at @xmath87, while the linear growth rate of the most unstable mode is @xmath88. similar to the classical maxwellian plasma case @xcite, we can thus estimate the magnetic field (in tesla) as @xmath89 for a degenerate fermi plasma. for our parameters parameters relevant for intense laser - solid interaction experiments, we will thus have magnetic fields of the order @xmath90 (one gigagauss).
Conclusion
in conclusion, we have demonstrated the existence of the weibel instability for a wigner - maxwell dense quantum plasma, taking into account an anisotropic fermi - dirac equilibrium distribution function and the quantum diffraction effect. numerically solving the dispersion relation for transverse waves, we found the dependence of the growth rate on the fermi energy and the temperature anisotropy. the nonlinear saturation level of the magnetic field was found by means of kinetic simulations, which show a linear dependence between the growth rate and the saturated magnetic field. the present results may account for intense magnetic fields in dense quantum plasmas, such as those in the next generation of intense laser - solid density plasma interaction experiments. 99 l. w. wildrow, rev. phys. * 74 *, 775 (2002) ; p. p. kronberg, phys. today * 55 *, 40 (2002). m. tatarakis _ et al. _, nature (london) * 415 *, 280 (2002) ; u. wagner _ et al. _, phys. e * 70 *, 026401 (2004). et al. _, j. * 395 *, 622 (1992) ; p. romatschke and r. venugopalan, phys. lett. * 96 *, 062302 (2006). p. biermann, z. naturforsch. a * 5 *, 65 (1950). k. estabrook, phys. lett. * 41 *, 1808 (1978). m. s. wei _ et al. _, e * 70 *, 056412 (2004). e. s. weibel, phys. * 2 *, 83 (1959). m. tzoufras _ et al. * 96 *, 105002 (2006). v. m. malkin, n. j. fisch and j. s. wurtele, phys. rev. e * 75 *, 026404 (2007). g. manfredi, fields inst. commun. * 46 *, 263 (2005). c. l. gardner and c. ringhofer, phys. e * 53 *, 157 (1996). l. k. ang and p. k. zhang, phys. * 98 *, 164802 (2007). g. brodin and m. marklund, new j. phys. * 9 *, 277 (2007) m. marklund, b. eliasson and p. k. shukla, phys e * 76 *, 067401 (2007). g. brodin and m. marklund, phys. e * 76 *, 055403 (2006). m. marklund and g. brodin, phys. 96 *, 025001 (2007). p. k. shukla and b. eliasson, phys. lett. * 96 *, 245001 (2006). d. shaikh and p. k. shukla, phys. lett. * 99 *, 125002 (2007). d. jovanovic and r. fedele, phys. a * 364 *, 304 (2007). a. bret, phys. plasmas * 14 *, 084503 (2007). f. haas, phys. plasmas * 15 *, 022104 (2008). f. haas, phys. of plasmas * 12 *, 062117 (2005). f. califano, f. pegoraro, s. v. bulanov and a. mangeney, phys. e * 57 *, 7048 (1998). d. pines and p. nozires, _ the theory of quantum liquids _ (w. a. benjamin, new york, 1966). j. lindhard, k. dan. selsk. mat. * 28 *, 1 (1954). e. cockayne and z. h. levine, phys. rev. b * 74 *, 235107 (2006). l. klimontovich and v. p. silin, in _ plasma physics _, edited by j. drummond (mcgraw - hill, new york, 1961), and many references therein for the wigner - maxwell method for quantum plasmas. e. wigner, phys. rev. * 40 *, 749 (1932). l. klimontovich and v. p. silin, zh. teor. fiz. * 23 *, 151 (1952). m. abramowitz and i. a. stegun (eds.), _ handbook of mathematical functions with formulas, graphs and mathematical tables _ (dover, new york, 1972). l. lewin, _ polylogarithms and associated functions _ (north - holland, new york, 1981). o. ross, phys. rev. * 119 *, 1174 (1960). n. r. arista and w. brandt, phys. a * 29 *, 1471 (1984). w. p. leemans _ et al. a * 46 *, 1091 (1992). b. d. fried and s. d. conte, _ the plasma dispersion function _ (academic press, london, 1961). b. h. brandsen and c. j. jochain, _ introduction to quantum mechanics _ (john wiley & sons inc., new york, 1989). n. a. krall and a. w. trivelpiece, _ principles of plasma physics _ (mcgraw - hill, new york, 1973). r. c. davidson _ et al. _, fluids * 15 *, 317 (1972). b. eliasson, j. comput. phys. * 225 *, 1508 (2007). | we present an investigation for the generation of intense magnetic fields in dense plasmas with an anisotropic electron fermi - dirac distribution. for this purpose, we use a new linear dispersion relation for transverse waves in the wigner - maxwell dense quantum plasma system.
numerical analysis of the dispersion relation reveals the scaling of the growth rate as a function of the fermi energy and the temperature anisotropy.
the nonlinear saturation level of the magnetic fields is found through fully kinetic simulations, which indicates that the final amplitudes of the magnetic fields are proportional to the linear growth rate of the instability.
the present results are important for understanding the origin of intense magnetic fields in dense fermionic plasmas, such as those in the next generation intense laser - solid density plasma experiments. | 0803.1036 |
The gluon propagator at zero temperature
in lattice qcd, the finite lattice spacing and finite lattice volume effects on the gluon propagator can be investigated with the help of lattice simulations at several lattice spacings and physical volumes. here we report on such a calculation. for details on the lattice setup see @xcite. in figure [fig : gluevol], we show the renormalized gluon propagator at @xmath0 gev for all lattice simulations. note that we compare our data with the large volume simulations performed by the berlin - moscow - adelaide collaboration @xcite see @xcite for details. in each plot we show data for a given value of @xmath1, i.e. data in the same plot has the same lattice spacing. the plots show that, for a given lattice spacing, the infrared gluon propagator decreases as the lattice volume increases. for larger momenta, the lattice data is less dependent on the lattice volume ; indeed, for momenta above @xmath2900 mev the lattice data define a unique curve. we can also investigate finite volume effects by comparing the renormalized gluon propagator computed using the same physical volume but different @xmath1 values. we are able to consider 4 different sets with similar physical volumes see figure [fig : gluespac]. although the physical volumes considered do not match perfectly, one can see in figure [fig : gluespac] that for momenta above @xmath2 900 mev the lattice data define a unique curve. this means that the renormalization procedure has been able to remove all dependence on the ultraviolet cut - off @xmath3 for the mid and high momentum regions. however, a comparison between figures [fig : gluevol] and [fig : gluespac] shows that, in the infrared region, the corrections due to the finite lattice spacing seem to be larger than the corrections associated with the finite lattice volume. in particular, figure [fig : gluespac] shows that the simulations performed with @xmath4, i.e., with a coarse lattice spacing, underestimate the gluon propagator in the infrared region. in this sense, the large volume simulations performed by the berlin - moscow - adelaide collaboration provide a lower bound for the continuum infrared propagator.
The gluon propagator at finite temperature
we also aim to study how temperature changes the gluon propagator. at finite temperature, the gluon propagator is described by two tensor structures, @xmath5 where the transverse and longitudinal projectors are defined by @xmath6 the transverse @xmath7 and longitudinal @xmath8 propagators are given by @xmath9 @xmath10 on the lattice, finite temperature is introduced by reducing the temporal extent of the lattice, i.e. we work with lattices @xmath11, with @xmath12. the temperature is defined by @xmath13. in table [tempsetup] we show the lattice setup of our simulation. simulations in this section have been performed with the help of chroma library @xcite. for the determination of the lattice spacing we fit the string tension data in @xcite in order to have a function @xmath14. note also that we have been careful in the choice of the parameters, in particular we have only two different spatial physical volumes : @xmath15 and @xmath16. this allows for a better control of finite size effects..lattice setup used for the computation of the gluon propagator at finite temperature. [cols="^,^,^,^,^,^",options="header ",] [tempsetup] figures [fig : transtemp] and [fig : longtemp] show the results obtained up to date. we see that the transverse propagator, in the infrared region, decreases with the temperature. moreover, this component shows finite volume effects ; in particular, the large volume data exhibits a turnover in the infrared, not seen at the small volume data. the longitudinal component increases for temperatures below @xmath17. then the data exhibits a discontinuity around @xmath18, and the propagator decreases for @xmath19. the behaviour of the gluon propagator as a function of the temperature can also be seen in the 3d plots shown in figure [fig:3dtemp]. as shown above, data for different physical (spatial) volumes exhibits finite volume effects. this can be seen in more detail in figure [fig : finvoltemp], where we show the propagators for two volumes at t=324 mev. moreover, we are also able to check for finite lattice spacing effects at t=305 mev, where we worked out two different simulations with similar physical volumes and temperatures, but different lattice spacings. for this case, it seems that finite lattice spacing effects are under control, with the exception of the zero momentum for the transverse component see figure [fig : lattspactemp]. our results show that a better understanding of lattice effects is needed before our ultimate goal, which is the modelling of the propagators as a function of momentum and temperature.
Acknowledgments
paulo silva is supported by fct under contract sfrh / bpd/40998/2007. work supported by projects cern / fp/123612/2011, cern / fp/123620/2011 and ptdc / fis/100968/2008, projects developed under initiative qren financed by ue / feder through programme compete. | we study the landau gauge gluon propagator at zero and finite temperature using lattice simulations.
particular attention is given to the finite size effects and to the infrared behaviour. | 1207.7324 |
Introduction
in industrial inspection, there is an ever - growing demand for highly accurate, non - destructive measurements of three - dimensional object geometries. a variety of optical sensors have been developed to meet these demands @xcite. these sensors satisfy the requirements at least partially. numerous applications, however, still wait for a capable metrology. the limitations of those sensors emerge from physics and technology the physical limits are determined by the wave equation and by coherent noise, while the technological limits are mainly due to the space - time - bandwidth product of electronic cameras. closer consideration reveals that the technological limits are basically of information - theoretical nature. the majority of the available optical 3d sensors need large amounts of raw data in order to obtain the shape. a lot of redundant information is acquired and the expensive channel capacity of the sensors is used inefficiently @xcite. a major source of redundancy is the shape of the object itself : if the object surface @xmath0 is almost planar, there is similar height information at each pixel. in terms of information theory the surface points of such objects are `` correlated '' ; their power spectral density @xmath1 decreases rapidly. in order to remove redundancy, one can apply spatial differentiation to whiten the power spectral density (see fig. [spectra]). fortunately, there are optical systems that perform such spatial differentiation. indeed, sensors that acquire just the local slope instead of absolute height values are much more efficient in terms of exploiting the available channel capacity. further, reconstructing the object height from slope data reduces the high - frequency noise since integration acts as a low - pass filter. there are several sensor principles that acquire the local slope : for _ rough surfaces _, it is mainly the principle of shape from shading @xcite. for _ specular surfaces _, there are differentiating sensor principles like the differential interference contrast microscopy or deflectometry @xcite. deflectometric scanning methods allow an extremely precise characterization of optical surfaces by measuring slope variations as small as @xmath2arcsec @xcite. full - field deflectometric sensors acquire the two - dimensional local gradient of a (specular) surface. using `` phase - measuring deflectometry '' (pmd) @xcite, for example, one can measure the local gradient of an object at one million sample points within a few seconds. the repeatability of the sensor described in @xcite is below @xmath3arcsec with an absolute error less than @xmath4arcsec, on a measurement field of @xmath5 mm @xmath6 @xmath5 mm and a sampling distance of @xmath7 mm. in several cases it is sufficient to know the local gradient or the local curvature ; however, most applications demand the height information as well. as an example we consider eyeglass lenses. in order to calculate the local surface power of an eyeglass lens by numerical differentiation, we only need the surface slope and the lateral sampling distance. but for quality assurance in an industrial setup, it is necessary to adjust the production machines according to the measured shape deviation. this requires height information of the surface. another application is the measurement of precision optics. for the optimization of these systems sensors are used to measure the local gradient of wavefronts @xcite. to obtain the shape of these wavefronts, a numerical shape reconstruction method is needed. in the previous section we stated that measuring the gradient instead of the object height is more efficient from an information - theoretical point of view, since redundant information is largely reduced. using numerical integration techniques, the shape of the object can be reconstructed locally with high accuracy. for example, a full - field deflectometric sensor allows the detection of local height variations as small as _ a few nanometers_. however, if we want to reconstruct the global shape of the object, low - frequency information is essential. acquiring _ solely the slope _ of the object reduces the low - frequency information substantially (see fig. [spectra]). in other words, we have a lot of _ local _ information while lacking _ global _ information, because we reduced the latter by optical differentiation. as a consequence, small measuring errors in the low - frequency range will have a strong effect on the overall reconstructed surface shape. this makes the reconstruction of the global shape a difficult task. furthermore, one - dimensional integration techniques can not be easily extended to the two - dimensional case. in this case, one has to choose a path of integration. unfortunately, noisy data leads to different integration results depending on the path@xcite. therefore, requiring the integration to be path independent becomes an important condition (`` integrability condition '') for developing an optimal reconstruction algorithm (see sections [problem] and [optimal]).
Problem formulation
we consider an _ object surface _ to be a twice continuously differentiable function @xmath8 on some compact, simply connected region @xmath9. the integrability condition implies that the gradient field @xmath10 is _ curl free _, i.e. every path integral between two points yields the same value. this is equivalent to the requirement that there exists a potential @xmath11 to the gradient field @xmath12 which is unique up to a constant. most object surfaces measurable by deflectometric sensors fulfill these requirements, or at least they can be decomposed into simple surface patches showing these properties. measuring the gradient @xmath12 at each sample point @xmath13 yields a discrete vector field @xmath14. these measured gradient values usually are contaminated by noise the vector field is not necessarily curl free. hence, there might not exist a potential @xmath11 such that @xmath15 for all @xmath16. in that case, we seek a least - squares approximation, i.e. a surface representation @xmath11 such that the following error functional is minimized @xcite : @xmath17 ^ 2 + \left[z_y({\bf x}_i)-q({\bf x}_i)\right]^2.\]]
Related work
in the case of one - dimensional data, integration is a rather straightforward procedure which has been investigated quite extensively @xcite. in case of two - dimensional data, there exist mainly two different approaches to solve the stated problem @xcite. on the one hand, there are _ local methods _ : they integrate along predetermined paths. the advantage of these methods is that they are simple and fast, and that they reconstruct small local height variations quite well. however, they propagate both the measurement error and the discretization error along the path. therefore, they introduce a global shape deviation. this effect is even worse if the given gradient field is not guaranteed to be curl free : in this case, the error also depends on the chosen path. on the other hand, there are _ global methods _ : they try to minimize @xmath18 by solving its corresponding euler - lagrange equation@xcite @xmath19 where @xmath20 and @xmath21 denote the numerical @xmath22- and @xmath23-derivative of the measured data @xmath24 and @xmath25, respectively. the advantage of global methods is that there is no propagation of the error ; instead, the error gets evenly distributed over all sample points. unfortunately, the implementation has certain difficulties. methods based on finite differences are usually inefficient in their convergence when applied to strongly curved surfaces @xcite. therefore, they are mainly used for nearly planar objects. another approach to solve eq. ([poisson]) is based on fourier transformation @xcite. integration in the fourier domain has the advantage of being optimal with respect to information - theoretical considerations @xcite. however, fourier methods assume a periodic extension on the boundary which can not be easily enforced with irregularly shaped boundaries in a two - dimensional domain. in general, it is crucial to note that the reconstruction method depends on the slope - measuring sensor and the properties of the acquired data. for example, slope data acquired by shape from shading is rather noisy, exhibits curl, and is usually located on a full grid of millions of points. here, a fast subspace approximation method like the one proposed by frankot and chellappa @xcite is appropriate. on the other hand, wavefront reconstruction deals with much smaller data sets, and the surface is known to be rather smooth and flat. in this case, a direct finite - difference solver can be applied @xcite. deflectometric sensors deliver a third type of data : it consists of very large data sets with rather small noise and curl, but the data may not be complete, depending on the local reflectance of the measured surface. furthermore, the measuring field may have an unknown, irregularly shaped boundary. these properties render most of the aforementioned methods unusable for deflectometric data. in the following sections, we will describe a surface reconstruction method which is especially able to deal with slope data acquired by sensors such as phase - measuring deflectometry.
Shape reconstruction
the desired surface reconstruction method should have the properties of both _ local and global _ integration methods : it needs to _ preserve local details _ without propagating the error along a certain path. it also needs to minimize the error functional of eq. ([approx_functional]), hence yielding a _ globally optimal solution _ in a least - squares sense. further, the method should be able to deal with _ irregularly shaped boundaries _, _ missing data points _, and it has to be able to reconstruct surfaces of a large variety of objects with steep slopes and _ high curvature values_. it should also be able to _ handle large data sets _ which may consist of some million sample points. we now show how to meet these challenges using an analytic interpolation approach. a low noise level allows _ interpolation _ of the slope values instead of approximation. interpolation is a special case which has the great advantage that we can ensure that small height variations are preserved. in this paper we will only focus on the interpolation approach as analytic reconstruction method. for other measurement principles like shape from shading, an approximation approach might be more appropriate. the basic idea of the integration method is as follows : we seek an analytic interpolation function such that its gradient interpolates the measured gradient data. once this interpolation is determined, it uniquely defines the surface reconstruction up to an integration constant. to obtain the analytic interpolant, we choose a generalized hermite interpolation approach employing _ radial basis functions _ (rbfs) @xcite. this method has the advantage that it can be applied to _ scattered data_. it allows us to integrate data sets with holes, irregular sampling grids, or irregularly shaped boundaries. furthermore, this method allows for an optimal surface recovery in the sense of eq. ([approx_functional]) (see section [optimal] below). in more detail : assuming that the object surface fulfills the requirements described in section [problem], the data is given as pairs @xmath26, where @xmath27 and @xmath28 are the measured slopes of the object at @xmath29 in @xmath22- and @xmath23-direction, respectively, for @xmath30. we define the interpolant to be @xmath31 where @xmath32 and @xmath33, for @xmath34, are coefficients and @xmath35 is a radial basis function. hereby, @xmath36 and @xmath37 denote the analytic derivative of @xmath38 with respect to @xmath22 and @xmath23, respectively. this interpolant is specifically tailored for gradient data @xcite. to obtain the coefficients in eq. ([interpolant]) we match the analytic derivatives of the interpolant with the measured derivatives : @xmath39 this leads to solving the following system of linear equations @xcite : @xmath40 \\[-0.7em] $ \phi_{x y}(\mathbf{x}_i-\mathbf{x}_j)$ & $ \phi_{y y}(\mathbf{x}_i-\mathbf{x}_j)$\\ \end{tabular } } \right)}_{a \, \in \, m^{2n\times2n } } \underbrace{\left ( \mbox{\begin{tabular}{c } $ \alpha_i$\\ \\[-0.7em] \\[-0.7em] $ \beta_i$ \end{tabular } } \right)}_{\alpha \, \in \, m^{2n\times1 } } = \underbrace{\left ( \mbox{\begin{tabular}{c } $ p(\mathbf{x}_j)$\\ \\[-0.7em] \\[-0.7em] $ q(\mathbf{x}_j)$ \end{tabular } } \right)}_{d \, \in \, m^{2n\times1}}.\]] + using the resulting coefficients @xmath41 we then can apply the interpolant in eq. ([interpolant]) to reconstruct the object surface. for higher noise levels an _ approximation _ approach is recommended. in this case, we simply reduce the number of basis functions so that they do not match the number of data points any more. the system @xmath42 in eq. ([linear_system]) then becomes overdetermined and can be solved in a least - squares sense. the interpolation approach employing radial basis functions has the advantage that it yields a _ unique _ solution to the surface recovery problem : within this setup, the interpolation matrix in eq. ([linear_system]) is always symmetric and positive definite. further, the solution satisfies a _ minimization principle _ in the sense that the resulting analytic surface function has minimal energy @xcite. we choose @xmath38 to be a wendland s function @xcite, @xmath43, with @xmath44 this has two reasons : first, wendland s functions allow to choose their continuity according to the smoothness of the given data. the above wendland s function leads to an interpolant which is three - times continuously differentiable, hence guaranteeing the integrability condition. second, the compact support of the function allows to adjust the support size in such a way that the solution of eq. ([linear_system]) is stable in the presence of noise. it turns out that the support size has to be chosen rather large in order to guarantee a good surface reconstruction @xcite. the amount of data commonly acquired with a pmd sensor in a single measurement is rather large : it consists of about one million sample points. this amount of data, which results from a measurement with high lateral resolution, would require the inversion of a matrix with @xmath45 entries (eq. ([linear_system])). since we choose a large support size for our basis functions to obtain good numerical stability the corresponding matrix is not sparse. it is obvious that this large amount of data can not be handled directly by inexpensive computing equipment in reasonable time. to cope with such large data sets we developed a method which first splits the data field into a set of overlapping rectangular patches. we interpolate the data on each patch separately. if the given data were height information only, this approach would yield the complete surface reconstruction. for slope data, we interpolate the data and obtain a surface reconstruction _ up to a constant of integration _ (see fig. [mirror_fit](a)) on each patch. in order to determine the missing information we apply the following fitting scheme : let us denote two adjacent patches as @xmath46 and @xmath47 and the resulting interpolants as @xmath48 and @xmath49, respectively. since the constant of integration is still unknown the two interpolants might be on different height levels. generally, we seek a correcting function @xmath50 by minimizing @xmath51 this fitting scheme is then propagated from the center toward the borders of the data field to obtain the reconstructed surface on the entire field (see fig. [mirror_fit](b)). in the simplest case, the functions @xmath52 are chosen to be constant on each patch, representing the missing constant of integration. if the systematic error of the measured data is small, the constant fit method is appropriate since it basically yields no error propagation. for very noisy data sets it might be better to use a planar fit, i.e. @xmath53, to avoid discontinuities at the patch boundaries. this modification, however, introduces a propagation of the error along the patches. the correction angle required on each patch to minimize eq. ([lsqfit]) depends on the noise of the data. numerical experiments have shown that in most cases the correction angle is at least ten times smaller than the noise level of the measured data. using this information we can estimate the global height error which, by error propagation, may sum up toward the borders of the measuring field @xcite : @xmath54 where @xmath55 is the standard deviation of the correction angles, @xmath56 is the patch size, and @xmath57 is the number of patches. suppose we want to integrate over a field of @xmath5 mm (which corresponds to a typical eyeglass diameter), assuming a realistic noise level of @xmath58arcsec and a patch size (not including its overlaps) of @xmath59 mm. with our setup, this results in @xmath60 patches, with a maximal tipping of @xmath61arcsec per patch. according to eq. ([globaltilt]), the resulting global error caused by propagation of the correction angles is _ only @xmath62nm_. we choose the size of the patches as big as possible, provided that a single patch can still be handled efficiently. for the patch size in the example, @xmath63 points (including @xmath64 overlap) correspond to a @xmath65 interpolation matrix that can be inverted quickly using standard numerical methods like cholesky decomposition. a final remark concerning the runtime complexity of the method described above : the complexity can be further reduced in case the sampling grid is regular. since the patches all have the same size and the matrix entries in eq. ([linear_system]) only depend on the distances between sample points, the matrix can be inverted once for all patches and then applied to varying data on different patches, as long as the particular data subset is complete. note that eq. ([interpolant]) can be written as @xmath66, where @xmath67 is the evaluation matrix. then, by applying eq. ([linear_system]) we obtain @xmath68, where the matrix @xmath69 needs to be calculated only once for all complete patches. if samples are missing, however, the interpolation yields different coefficients @xmath70 and hence forces to recompute @xmath69 for this particular patch. using these techniques, the reconstruction of @xmath71 surface values from their gradients takes about @xmath72 minutes on a current personal computer.
Results
first, we investigated the stability of our method with respect to noise. we simulated realistic gradient data of a sphere (with @xmath5 mm radius, @xmath5 mm @xmath6 @xmath5 mm field with sampling distance @xmath73 mm, see figure [noise_sphere](a)) and added uniformly distributed noise of different levels, ranging from @xmath74 to @xmath75arcsec. we reconstructed the surface of the sphere using the interpolation method described in section [method]. hereby, we aligned the patches by only adding a constant to each patch. the reconstruction was performed for @xmath76 statistically independent slope data sets for each noise level. depicted in figure [noise_sphere](b) is a cross - section of the absolute error of the surface reconstruction from the ideal sphere, for a realistic noise level of @xmath58arcsec. the absolute error is less than @xmath77 nm _ on the entire measurement field_. the local height error corresponding to this noise level is about @xmath78 nm. this demonstrates that the dynamic range of the global absolute error with respect to the height variance (@xmath79 mm) of the considered sphere is about @xmath80. the graph in figure [noise_sphere](c) depicts the mean value and the standard deviation (black error bars) of the absolute error of the reconstruction, for @xmath76 different data sets and for different noise levels. it demonstrates that for increasing noise level the absolute error grows only linearly (linear fit depicted in gray), and even for a noise level being the _ fifty - fold of the typical sensor noise _ the global absolute error remains in the _ sub - micrometer regime_. this result implies that the reconstruction error is smaller than most technical applications require. another common task in quality assurance is the detection and quantification of surface defects like scratches or grooves. we therefore tested our method for its ability to reconstruct such local defects that may be in a range of only a few nanometers. for this purpose, we considered data from a pmd sensor for small, specular objects. the sensor has a resolvable distance of @xmath81 m laterally and a local angular uncertainty of about @xmath76arcsec @xcite. in order to quantify the deviation of the perfect shape, we again simulated a sphere (this time with @xmath76 mm radius and @xmath82 mm @xmath6 @xmath82 mm data field size). we added parallel, straight grooves of varying depths from @xmath83 to @xmath4 nm and of @xmath84 m width and reconstructed the surface from the modified gradients. the perfect sphere was then subtracted from the reconstructed surface. the resulting reconstructed grooves are depicted in figure [groove_sphere](a). the grooves ranging from @xmath4 down to @xmath72 nm depth are clearly distinguishable from the plane. figure [groove_sphere](b) shows that all reconstructed depths agree fairly well with the actual depths. note that each groove is determined by only @xmath72 inner sample points. the simulation results demonstrate that our method is almost free of error propagation while preserving small, local details of only some nanometers height. so far, we used only simulated data to test the reconstruction. now, we want to demonstrate the application of our method to a real measurement. the measurement was performed with a phase - measuring deflectometry sensor for very small, specular objects. it can laterally resolve object points with a distance of @xmath85 nm, while having a local angular uncertainty of about @xmath86arcsec. the object under test is a part of a wafer with about @xmath87 nm height range. the size of the measurement field was @xmath4 m @xmath6 @xmath5 m. depicted in figure [wafer] is the reconstructed object surface from roughly three million data values. both the global shape and local details could be reconstructed with high precision.
Conclusion
we motivated why the employment of optical slope - measuring sensors can be advantageous. we gave a brief overview of existing sensor principles. the question that arose next was how to reconstruct the surface from its slope data. we presented a method based on radial basis functions which enables us to reconstruct the object surface from noisy gradient data. the method can handle large data sets consisting of some million entries. furthermore, the data does not need to be acquired on a regular grid it can be arbitrarily scattered and it can contain data holes. we demonstrated that, while accurately reconstructing the object s global shape, which may have a _ height range of some millimeters _, the method preserves _ local height variations on a nanometer scale_. a remaining challenge is to improve the runtime complexity of the algorithm in order to be able to employ it for inline quality assurance in a production process.
Acknowledgments
the authors thank the deutsche forschungsgemeinschaft for supporting this work in the framework of sfb 603. i. weingrtner and m. schulz, `` novel scanning technique for ultra - precise measurement of slope and topography of flats, aspheres, and complex surfaces, '' in _ optical fabrication and testing _, (1999), no. 3739 in proc. spie, pp. 27482. m. c. knauer, j. kaminski, and g. husler, `` phase measuring deflectometry : a new approach to measure specular free - form surfaces, '' in _ optical metrology in production engineering _,, w. osten and m. takeda, eds. (2004), no. 5457 in proc. spie, pp. 366376. j. kaminski, s. lowitzsch, m. c. knauer, and g. husler, `` full - field shape measurement of specular surfaces, '' in _ fringe 2005, the 5th international workshop on automatic processing of fringe patterns _,, w. osten, ed. (springer, 2005), pp. 372379. s. lowitzsch, j. kaminski, m. c. knauer, and g. husler, `` vision and modeling of specular surfaces, '' in _ vision, modeling, and visualization 2005 _,, g. greiner, j. hornegger, h. niemann, and s. m., eds. (akademische verlagsgesellschaft aka gmbh, 2005), pp. 479486. j. pfund, n. lindlein, j. schwider, r. burow, t. blumel, and k. e. elssner, `` absolute sphericity measurement : a comparative study of the use of interferometry and a shack - hartmann sensor, '' opt. * 23 *, 742744 (1998). a. agrawal, r. chellappa, and r. raskar, `` an algebraic approach to surface reconstruction from gradient fields, '' in `` ieee international conference on computer vision (iccv), '', vol. 1 (2005), vol. 1, pp. 174181. c. elster and i. weingrtner, `` high - accuracy reconstruction of a function @xmath88 when only @xmath89 or @xmath90 is known at discrete measurement points, '' in `` x - ray mirrors, crystals, and multilayers ii, '' (2002), no. 4782 in proc. spie, pp. 1220. s. ettl, j. kaminski, and g. husler, `` generalized hermite interpolation with radial basis functions considering only gradient data, '' in _ curve and surface fitting : avignon 2006 _,, a. cohen, j.- l. merrien, and l. l. schumaker, eds. (nashboro press, brentwood, 2007), pp. 141149. of a typical smooth object and (b) its derivative @xmath91, together with (c) the power spectral density of the surface @xmath1 and (d) of its slope @xmath92 (all in arbitrary units). the power spectral density shows that differentiation reduces redundancy, contained in the low frequencies., width=627] nm down to @xmath83 nm. after the reconstruction, the sphere was subtracted to make the grooves visible. the actual, reconstructed grooves are depicted in (a) full - field and in (b) cross - section.,width=529] | we present a novel method for reconstructing the shape of an object from measured gradient data.
a certain class of optical sensors does not measure the shape of an object, but its local slope.
these sensors display several advantages, including high information efficiency, sensitivity, and robustness. for many applications, however, it is necessary to acquire the shape, which must be calculated from the slopes by numerical integration.
existing integration techniques show drawbacks that render them unusable in many cases.
our method is based on approximation employing radial basis functions.
it can be applied to irregularly sampled, noisy, and incomplete data, and it reconstructs surfaces both locally and globally with high accuracy. | 0710.4278 |
Introduction
the study of tachyons and their condensation processes has been a longstanding challenge in string theory. during recent years, the interest in branes has added significantly to the relevance of this issue since open string tachyons occur in various configurations of stable branes, the most important being the brane - antibrane system. fortunately, unstable brane configurations turned out to be much more tractable than bulk backgrounds with instabilities such as e.g. the 26-dimensional bosonic string. in fact, various different approaches have been employed over the last five years, including string field theory (see e.g. @xcite and @xcite for a review), effective field theory models (see @xcite for some early work), and world - sheet conformal field theory (see e.g. @xcite and references therein). in @xcite, sen initiated the study of _ exact time - dependent solutions_. one goal of his work was to find a world - sheet description of so - called s - branes @xcite, i.e. of branes that are localized in time. in this context he proposed to add a @xmath1-shaped boundary interaction to the usual world - sheet model of d - branes in a flat space - time, @xmath2 \ = \ \frac{1}{4\pi } \int_\sigma d^2 z \, \eta_{\mu\nu } \partial x^\mu \bar \partial x^\nu + \int_{\partial \sigma } d u \, \lambda \cosh x^0(u) \ \.\]] here, @xmath3 denotes the time - coordinate and we have chosen units in which @xmath4. with such an interaction term the open string tachyon becomes large at early and late times and hence, according to an earlier proposal of sen @xcite, is believed to dissolve the brane for @xmath5. unfortunately, the world - sheet model ([senact]) appears to be ill - defined without an additional prescription to select the proper vacuum. to resolve this issue, sen argued that the appropriate choice for super - string computations would be determined by wick rotation from the corresponding euclidean theory. this suggestion relates the study of open string tachyon condensation to the so - called boundary sine - gordon model @xmath6 \ = \ \frac{1}{4\pi } \int_\sigma d^2 z\, \partial x \bar \partial x + \int_{\partial \sigma } du \, \lambda \cos x(u)\]] in which @xmath7 is a field with space - like rather than time - like signature. all spatial coordinates @xmath8 have been suppressed here since their contribution to the model is not affected by the interaction. the boundary sine - gordon theory has been studied rather intensively, see e.g. @xcite. let us briefly review some of the most important results. to begin with, we point out that the boundary interaction is exactly marginal so that the theory is conformal for all values of @xmath9. properties of the model, however, depend crucially on the strength @xmath9 of the boundary potential. in fact, it is well known that variations of @xmath9 allow to interpolate smoothly between neumann and dirichlet boundary conditions. the former appear for all integer values of @xmath9 (in particular for @xmath10, of course) while the latter are reached when @xmath11. at these points, the theory describes a one - dimensional infinite lattice of d0 branes. from the geometric pictures we can infer that the spectra of boundary excitations must also depend drastically on the parameter @xmath9. in fact, at the points @xmath12 with neumann boundary conditions, the open string spectrum is continuous. if we now start tuning @xmath9 away from these special values, the spectrum develops band gaps which become wider and wider until we reach the dirichlet points @xmath13 at which the spectrum is discrete. the first computation of the spectrum for generic values of @xmath9 can be found in @xcite (see also @xcite for a much more elegant derivation). despite of these significant insights into the structure of the boundary sine - gordon model, there are several important quantities that remain unknown. this applies in particular to the boundary 2- and 3-point functions and the bulk - boundary operator product expansions. in the string theoretic context, these missing data determine the time - dependence of open string couplings on a decaying brane and the back - reaction to the bulk geometry. our desire to understand such important quantities is one of the main motivations to study the following, closely related world - sheet theory @xcite, @xmath14 \ = \ \frac{1}{4\pi } \int_\sigma d^2 z \, \eta_{\mu\nu } \partial x^\mu \bar \partial x^\nu + \int_{\partial \sigma } d u \, \mu_b \exp x^0(u) \ \.\]] this model has been named time - like boundary liouville theory. since the tachyon vanishes in the far past, the model seems to describe a half - brane, i.e. a brane that dissolves as time progresses. after wick rotation @xmath15, the theory of the time coordinate becomes. this might be relevant for comparisons with previous results in models with exponential interactions.] @xmath16 \ = \ \left (\frac{1}{4\pi } \int_\sigma d^2 z\, \partial x \bar \partial x + \mu \exp 2b x(z,{\bar{z } }) + \int_{\partial \sigma } du \, \mu_b \exp b x(u)\right)^{\mu=0}_{b = i } \ \.\]] we have written the interaction term for general parameters @xmath17 and also added a similar interaction in the bulk, mainly to emphasize the relation with boundary liouville theory. what makes this relation so valuable for us is the fact that boundary liouville theory has been solved over the last years @xcite. needless to stress that the solution includes explicit formulas for the bulk - boundary structure constants @xcite and the boundary 3-point couplings @xcite. there is one crucial difference between liouville theory and the model we are interested in : whereas the usual liouville model is defined for real @xmath17, our application requires to set @xmath18. a continuation of results in liouville theory from @xmath19 to @xmath0 (i.e. @xmath18) might appear to be a rather daring project, even more so as a naive inspection of the world - sheet action would suggest the @xmath0 model could not possibly be unitary. nevertheless, we shall show below that the theory is entirely well - defined and unitary. for the pure bulk theory, a similar result was established in @xcite. it was shown there that the bulk 3-point couplings of liouville theory possess a @xmath18 limit which is well - defined for real momenta of the participating closed strings. the @xmath18 theory, however, is no longer analytic in the momenta. furthermore, the limit turned out to agree with the @xmath0 limit of unitary minimal models which was constructed by runkel and watts in @xcite. our analysis here extends these findings to the boundary theory. we shall see that the latter is considerably richer than the bulk model. some qualitative properties of the @xmath0 boundary model can be understood using no more than a few general observations. to begin with, let us reconsider the simpler bulk model. recall from ordinary liouville theory that it has a trivial dependence on the coupling constant @xmath20. since any changes in the coupling can be absorbed in a shift of the zero mode, one can not vary the strength of the interaction. this feature of liouville theory persists when the parameter @xmath17 moves away from the real axis into the complex plane. as we reach the point @xmath18, our model seems to change quite drastically : at this point, the ` liouville wall'disappears and the potential becomes periodic. standard intuition therefore suggests that the spectrum of closed string modes develops gaps at @xmath18. but since the strength of the interaction can not be tuned in the bulk theory, the band gaps must be point - like. though our argument here was based on properties of the classical action which we can not fully trust, the point - like band - gaps are indeed a characteristic property of the @xmath0 bulk theory @xcite. these band - gaps also explain why the couplings of liouville theory cease to be analytic when we reach @xmath0. for the boundary theory, we can go through a similar argument, but the consequences are more pronounced. in the presence of a boundary, liouville theory contains a second coupling constant @xmath21 which controls the strength of an exponential interaction on the boundary of the world - sheet and is a real parameter of the model. in fact, the freedom of shifting the zero mode can only be used for one of the couplings @xmath20 or @xmath21. once more, the boundary potential becomes periodic at @xmath18 and hence the open string spectrum should develop gaps, as in the case of the bulk model. but this time, the width of these gaps can be tuned by changes of the parameter @xmath21. hence, we expect the boundary liouville theory to possess gaps of finite width at @xmath0. our exact analysis will confirm this outcome of the discussion. after these more qualitative remarks on the model we are about to construct, we shall summarize our main results in more detail. in section 3 we shall provide the boundary states for a family of boundary conditions of the @xmath0 liouville theory that is parametrized by one parameter @xmath22. formulas for these boundary states are trivially obtained from the corresponding expressions in @xmath23 liouville theory and, as we shall also explain, through particular limits of boundary theories in minimal models. we shall then use both approaches to establish the existence of finite band gaps in the boundary spectrum of the @xmath0 boundary theories. more precisely, we show that boundary fields @xmath24 in the boundary theory with parameter @xmath25 carry a momentum whose allowed values are taken from the following set @xmath26 when @xmath25 is a half - integer, the set @xmath27 fills the entire real line. as we vary @xmath25 away from such half - integer values, the allowed momenta are restricted to intervals of decreasing width until we reach integer values of @xmath25 where the spectrum of boundary fields becomes discrete. in this sense the parameter @xmath25 interpolates between neumann - type and dirichlet - type boundary conditions. let us note that the boundary conditions with @xmath28 are related to those constructed by runkel and watts in @xcite. all other boundary theories, however, are new. it is worthwhile pointing out how closely the @xmath0 limit of liouville theory mimics the behavior of the boundary sine - gordon model. in both theories we can interpolate smoothly between continuous and discrete boundary spectra. even the geometric interpretation of the dirichlet points is very similar : in the context of the @xmath0 liouville model they describe a semi - infinite array of point - like branes from which individual branes can be removed through shifts of the boundary parameter @xmath25. let us point out that finite arrays of such point - like branes are described by the @xmath0 limit of zz branes. that parametrize the zz branes correspond to the transverse position and the length of the finite array.] hence, the picture we suggest for the dirichlet points of the @xmath0 fzzt branes provides a nice geometric interpretation for the well - known relation of fzzt and zz branes in liouville theory @xcite. there are two important quantities that we shall construct for all these new boundary theories. these are the boundary 2-point function and the bulk - boundary 2-point function. for the former we obtain @xmath29 \\[3 mm] \mbox { where } & & \r^{s_1s_2}_{c=1}(p) \ = \ \r^{is_1\ is_2}_{c=25 } (1-p)^{-1}\ \. \end{aligned}\]] detailed formulas for @xmath30 (see eq. ([2ptfctres1])) and for the so - called reflection amplitude @xmath31 are spelled out in section 4. here, we have expressed the result in terms of the reflection amplitude for the theory at @xmath32. it is quite remarkable that the reflection amplitudes of the two models are related in such a simple way. let us point out, though, that the boundary reflection amplitude for the @xmath0 theory is only defined within a subset @xmath33 of momenta @xmath34. it is more difficult to write down the bulk - boundary coupling @xmath35 of the @xmath0 model, i.e. the coefficient in the 2-point function of a bulk field @xmath36 with one of the allowed boundary fields, @xmath37 we shall prove below that this coupling @xmath35 is given by @xmath38 & & \hspace*{-3.5cm}\nonumber \times \,\bigg ( \frac{e^{-2\pi ip_{\alpha } (s+1)}}{1-e^{-4\pi ip_{\alpha } } } \frac{g(y_0)^2 e^{-\frac{ih (y_{0})}{2\pi}}}{y_0^{2 } \, h''(y_0)}\, \exp \int_0^\infty \frac{dt}{t } \ \frac{\left(e^{-p_\b t } - e^{2p_\a t}\right) \sinh^2\frac{p_{\b}t}{2}}{\sinh^2\frac{t}{2}}-\ (p_{\alpha}\rightarrow -p_{\alpha }) \bigg) \end{aligned}\]] where @xmath39 and the functions @xmath40 and @xmath41 are defined through @xmath42 h (y)&= & { \text{li}_{2}}(a_{0})-{\text{li}_{2}}(a_{0}y)+{\text{li}_{2}}(b_{0})-{\text{li}_{2}}(b_{0}y)\nonumber \\ & & -{\text{li}_{2}}(c_{0})+{\text{li}_{2}}(c_{0}y)-{\text{li}_{2}}(1)+{\text{li}_{2}}(y)+ \log y \log z_{0 } \ \.\end{aligned}\]] here, @xmath43 denotes the dilogarithm and we have abbreviated @xmath44 and the parameter @xmath45 is one of the two solutions of the following quadratic equation @xmath46 for more details and explicit formulas, see section 5. we also show that the coefficients in the bulk - boundary operator product expansion vanish whenever the open string momentum @xmath47 lies within the band gaps, i.e. when @xmath48. this provides another non - trivial consistency check for the couplings we propose. let us add that correlators with boundary insertions in a similar model where also discussed recently in @xcite. the techniques used there, however, only allowed to determine such correlations functions for a discrete set of boundary momenta @xmath49.
The @xmath0 limit of the bulk theory
in this section we shall mainly review some results on the @xmath0 limit of bulk liouville theory @xcite (see also @xcite). our derivation of the 3-point couplings, however, is different from the one given in @xcite. the new construction is simpler and uses some of the same techniques that we shall also employ to analyze the boundary model later on. to begin with, let us recall a few standard facts about the usual @xmath50 liouville theory. as in any bulk conformal field theory, the exact solution of the liouville model is entirely determined by the structure constants of the 3-point functions for the (normalizable) primary fields @xmath51 where @xmath52. these fields are primaries with conformal weight @xmath53 under the action of the two virasoro algebras whose central charge is @xmath54. for real values of @xmath17, the couplings of three such fields are given by the following expression @xmath55 where @xmath56 and @xmath57 are linear combinations of @xmath58, @xmath59 @xmath60 is a quotient of ordinary @xmath61-functions and the function @xmath62 is defined in terms of barnes double gamma function @xmath63 (see appendix a.2) by @xmath64 the solution ([bulktpf]) was first proposed several years ago by h. dorn and h.j. otto @xcite and by a. and al. zamolodchikov @xcite. crossing symmetry of the conjectured 3-point function was then checked analytically in two steps by ponsot and teschner @xcite and by teschner @xcite. it is well known that barnes double gamma function is analytic for @xmath65. in the limit @xmath18, however, the function becomes singular. nevertheless, the combinations of double gamma functions that appear in the various couplings of liouville theory turn out to be well defined, though they are no longer analytic. most of our analysis of the @xmath18 limit is based on the following relation between double gamma functions with parameter @xmath17 and @xmath66, @xmath67 which holds for @xmath68. here we have introduced @xmath69 and the so - called q - pochhammer symbols @xmath70 (see appendix a.1). relation ([dgdgrel]) follows from a rotation of the contour in the integral representation ([barnesg]) of the double gamma function. the factor involving the q - pochhammer symbols assembles contributions from the poles in the integrand of the double gamma function (see appendix a.2 for details). using formula ([pochasymp]) for the behavior of our q - pochhammer symbols at @xmath18 we conclude @xmath71 - 1/2) } \right)^{-(p+1-n)/2 } \, \times \\[2 mm] & & \hspace*{2 cm } \times \, e^{\frac{1}{\e } \left (li_2(e^{-2\pi i p }) - li_2(1)\right) } \gamma_2^{-1}(1-p|1) \, \left(1 + o(\e^{0})\right) \label{gasym } \nonumber\end{aligned}\]] with @xmath72 and @xmath73 $] denoting the integer part of @xmath34. note that the double gamma function diverges as we send @xmath74 (or @xmath75) with a divergent factor involving the dilogarithm @xmath76. in the expression for closed string couplings, barnes double gamma functions appears only through @xmath62, and using a standard dilogarithm identity (see appendix a.3) one may show that @xmath77 - 1/2) } \, \upsilon (1-p|1)^{-1 } \ \.\]] here, @xmath78 is a continuous periodic function which is quadratic on each interval @xmath79 $] for @xmath80, @xmath81 - 1/2)^{2 } \ \.\]] observe that the divergent factor in @xmath62 is much simpler than for the double gamma functions it is built from. in the function @xmath62, the divergence comes from a product of q - pochhammer symbols. the latter is closely related to jacobi s function @xmath82 and its divergence may be controlled more directly using modular properties of @xmath83-functions (see @xcite). using once more the formula ([dgdgrel]) and some simple facts about barnes double gamma function is is not difficult to show that @xmath84 with the help of the two asymptotic expressions and , we find @xmath85 - 1/2) } \prod _ { j=1}^{3 } \frac{e^{\pi i p_{j}}e^{-\pi i (2p_{j}-1) (p_{j}-[2p_{j}]-1/2)}}{e^{-\pi i 2\tilde{p}_{j } (\tilde{p}_{j}-[2\tilde{p}_{j}]-1/2)}}\\ & \ \times \ \frac{\upsilon (1 - 2\tilde{p}|1)}{\upsilon (1|1) } \prod _ { j=1}^{3}\frac{\upsilon(1 - 2\tilde{p}_{j}|1) } { \upsilon (1 - 2p_{j}|1)}\ (1+o (\epsilon^{0 })) \label{bulktpfasymp}\end{aligned}\]] where @xmath86 it can be shown that @xmath87, so that the exponential is never diverging, but it can suppress the whole three - point coupling. we write @xmath88+f_{j}$] and find that @xmath89 is zero if either @xmath90+[2p_{2}]+[2p_{3}]\ \ \text{even, and } \ |f_{1}-f_{2}|\leq f_{3}\leq \min \{f_{1}+f_{2},2-f_{1}-f_{2 } \}\ \, \]] or @xmath91+[2p_{2}]+[2p_{3}]\ \ \text{odd, and } \ |f_{1}-f_{2}|\leq 1-f_{3}\leq \min \ { f_{1}+f_{2},2-f_{1}-f_{2}\ } \ \, \]] otherwise @xmath89 is strictly greater than zero and the three - point coupling vanishes for @xmath92. in more formal terms, the function @xmath93 assumes the value @xmath94 when one of the above conditions is fulfilled and it vanishes otherwise. with the new function @xmath95 being introduced, we can use eq. to recast the 3-point couplings in the form @xmath96 } \, p (2p_{1},2p_{2},2p_{3 }) \times \nonumber\\[2 mm] & & \hspace*{3 cm } \times \ \frac{\upsilon (1 - 2\tilde{p}|1)}{\upsilon (1|1) } \prod _ { j=1}^{3}\frac{\upsilon (1 - 2\tilde{p}_{j}|1)}{\upsilon (1 - 2p_{j}|1) } \ \. \label{bulk3ptfunc}\end{aligned}\]] in our conventions, @xmath97, so up to a factor @xmath98 we reproduce precisely the result of @xcite (see eqs.(4.8) and (5.3) of that article). to obtain a finite limit, the bulk fields should be renormalized by a factor @xmath99 and correlators on the sphere receive an additional factor @xmath100 due to a rescaling of the vacuum (see section 4.2 in @xcite for a related discussion). this prescription produces a finite 3-point function and it does not affect the 2-point function. at first, the form of the couplings, and in particular the non - analytic factor @xmath95, may seem a bit surprising. it is therefore reassuring that exactly the same couplings emerged several years ago through a @xmath0 limit of unitary minimal models @xcite. runkel and watts investigated the consistency of these couplings and demonstrated that crossing symmetry holds when all the half - integer momenta @xmath34 are removed from the spectrum. as we have argued in the introduction, the appearance of such point - like band - gaps is rather natural from the point of view of liouville theory.
The @xmath0 limit of the boundary theory
now we are ready to analyze the boundary model. we shall begin with the so - called fzzt boundary conditions of liouville theory. they describe extended branes with an exponential potential for open strings (see eq. ([blact])). taking the @xmath0 limit of their 1-point function is straightforward and results in an expression with analytic dependence on the closed string momentum. these @xmath0 boundary states are parametrized through one real parameter @xmath25 and for non - integer values of @xmath25 they can be reproduced by taking an appropriate limit of boundary conditions in unitary minimal models. the latter construction will provide a first derivation of the spectrum of boundary vertex operators, including the precise position and width of the band - gaps. when @xmath28, the bands become point - like. at these special values of the parameter @xmath25, our two constructions through liouville theory and minimal models do not produce the same boundary states, though they still give closely related results. from minimal models we obtain the discrete set of boundary conditions that is already contained in the work of runkel and watts. these agree with the @xmath0 limit of zz branes in liouville theory and they possess a simple geometric interpretation as finite arrays of point - like branes. the relation between zz and fzzt branes in liouville theory then implies that we can interpret fzzt branes with @xmath28 as half - infinite arrays of point - like branes in the @xmath0 model. it is well known that for real @xmath17, liouville theory admits a 1-parameter family of boundary conditions which correspond to branes stretching out along the real line. the boundary state of this theory was found in @xcite, @xmath101 & & \mbox{where } \ \ \ \mu_b^2 \sin \pi b^2 \ = \ \mu \cosh^2 \pi s b \nonumber \end{aligned}\]] and @xmath102, as usual. note that the coupling on the right hand side is analytic in @xmath17 and hence there is no problem to extend the fzzt brane boundary state into the @xmath0 theory, @xmath103 this boundary state is related to its lorentzian analogue @xcite by analytic continuation. it is, however, not entirely obvious that eq. ([1ptfceq1]) really provides the one - point function of a consistent boundary conformal field theory. we shall argue below that this is the case, at least as long as the parameter @xmath25 remains real. let us now see how such a boundary state of the @xmath0 model can arise from a limit of boundary minimal models. to begin with, we need to set up a few notations. we shall label minimal models by some integer @xmath104 so that @xmath105. for representations of the corresponding virasoro algebra we use the kac labels @xmath106 with @xmath107 and @xmath108 along with the usual identification @xmath109. recall that the boundary conditions of the theory are in one to one correspondence with the representations, i.e. that they are also labeled by kac labels @xmath110. the associated one - point functions are given by @xmath111 where @xmath112 and @xmath113 is the conformal dimension of the primary field with kac labels @xmath106, @xmath114 here, we have introduced the new quantity @xmath115 that can be regarded as a discrete analogue of the continuous momentum @xmath34 when @xmath116. in taking the limit @xmath117, the bulk fields @xmath118 of minimal models approach the fields @xmath119 in the continuum theory. the momentum variable @xmath34 is related to the kac - labels by @xmath120 in other words, the integer part of the quantity @xmath121 is determined by the distance of the kac label @xmath106 from the diagonal @xmath122, whereas the fractional part @xmath123 $] corresponds to the (rescaled) position along the diagonal. for the a - series of minimal models, the bulk spectrum is obtained by filling every point of the kac table above the diagonal. when we send @xmath124 to infinity, approximately the same number of fields appears at any given distance from the diagonal of the kac table and these are distributed almost homogeneously along the diagonal @xmath125. hence, the spectrum of @xmath34 in the @xmath0 limit is homogeneous. only integer values of @xmath121 are not part of this spectrum since the kac labels satisfy @xmath126. this leads to the point - like band - gaps in the spectrum of bulk fields that we also saw emerging from liouville theory. to obtain a one - parameter family of branes with 1-point function given by eq ([1ptfceq1]) we propose the following ansatz, @xmath127 + [s] + 1 \,, \, [f_s m]+ 1 \,) \ \.\]] here, @xmath22 is the parameter of the resulting limiting boundary condition. @xmath128 $] and @xmath129 denote the integer and fractional part of @xmath25, respectively. the prescription ([prescript]) instructs us to pick a boundary condition that is represented by a point at distance @xmath128 $] from the diagonal of the kac table and to scale its projection to the diagonal with @xmath124. our first aim now is to show that this recipe reproduces the 1-point function ([1ptfceq1]) in the @xmath0 liouville theory. the main idea is to rewrite the product of @xmath130-functions in the numerator of eq. ([1ptfmm]) as follows @xmath131+[s]+1) (r / t - r') \, \sin \pi ([f_s m] + 1)(r - r't) \nonumber \\[2 mm] & & \hspace*{3 cm } \stackrel{m\rightarrow \infty}{\sim } \ \cos 2 \pi p s - \cos 4 \pi p ([f_s m]+1+[s]/2) + \dots \nonumber \end{aligned}\]] with @xmath132 being the momentum in the @xmath0 theory. provided @xmath133, the second term in the previous formula oscillates rapidly when we send @xmath124 to infinity and hence we can drop it in the limit (recall that all correlation functions should be understood as distributions in momentum space). together with the other factors in the 1-point function ([1ptfmm]) of minimal models, a short analysis similar to the one carried out in @xcite gives @xmath134 the result agrees with the 1-point function ([1ptfceq1]) of the @xmath0 liouville model up to some prefactors which may be absorbed by an appropriate renormalization of the bulk fields. if, on the other hand, the label @xmath25 is an integer, then our limit of boundary conditions in minimal models becomes @xmath135 the last identification with the @xmath0 limit of boundary states in liouville theory holds up to some factors that are due to the slightly different normalizations of bulk fields in minimal models and liouville theory (see @xcite). our result ([mml]) is very reminiscent of a similar relation between localized and extended branes in liouville theory @xcite and we shall comment on the precise connection and its geometric interpretation in the @xmath0 model below. now that we have identified the construction of the boundary state from minimal models, it is instructive to study the associated spectrum of boundary fields. according to the usual rules @xcite, the boundary spectrum of the brane @xmath110 contains fields with kac labels @xmath136 and @xmath137. using the identification @xmath138 we can assume that @xmath139. hence, the kac labels of boundary fields fill every second lattice point within two triangles in the kac table (see figure 1). in the first triangle with corner at @xmath140, the difference @xmath141 of kac labels is even. for kac labels in the second triangle, on the other hand, @xmath141 is an odd integer since the reflection @xmath142 shifts the difference of kac labels by one unit. recall that the quantity @xmath141 measures the distance of a point in the kac table from the diagonal and thereby determines the integer part of @xmath121 in the @xmath0 theory. we conclude that the two triangles contribute boundary fields with @xmath143 $] being even or odd integer, respectively. furthermore, points from the first triangle are uniformly distributed in the direction along the diagonal up to a maximal value @xmath144. a similar observation holds true for the second triangle. after we have sent @xmath124 to infinity, this implies that the spectrum of @xmath34 is given by @xmath145 hence, the momentum @xmath34 has bands of width @xmath146 centered around integer momentum @xmath34. the gaps between these bands widen while we vary @xmath25 between a half - integer and an integer value. once we reach a point @xmath147, the boundary spectrum becomes discrete. branes in the @xmath0 limit of minimal models with a discrete open string spectrum have also been constructed by runkel and watts (see also @xcite). actually, the branes that we obtained through the limit of minimal models when @xmath147 are identical to the limit of @xmath148 branes in minimal models and hence to the discrete boundary conditions that were found by runkel and watts. we wish to point out that these branes may also be obtained from the so - called zz branes of liouville theory. in general, the zz branes are labeled by a pair @xmath149 of positive integers and they possess the following 1-point functions, @xmath150 if we send the parameter @xmath17 to @xmath18, these 1-point functions assume the following form @xmath151 we can rewrite the coupling on the right hand side with the help of the trigonometric identity @xmath152 to read off the following geometric interpretation of the zz brane with label @xmath149 in the @xmath0 theory : it corresponds to an array of @xmath153 point - like branes that are distributed equidistantly such that the rightmost brane is located at @xmath154. this interpretation agrees with the geometry of branes that emerges from the coset construction of minimal models (see @xcite). with this in mind, we shall also be able to interpret our fzzt branes at the dirichlet points @xmath28. let us recall that the @xmath0 limit of fzzt branes with integer label @xmath25 differs from the limit of minimal model branes. the precise relation between their two boundary states is given in eq. ([mml]). on its right hand side we consider the difference of one - point couplings for two fzzt branes whose labels differ by two units. this is claimed to be the same as the one - point coupling to the discrete brane with label @xmath148. therefore, the brane configuration that is associated to the fzzt brane with integer label @xmath25 contains an additional point - like object with transverse position parameter @xmath25 along the real line. the latter gets removed as we shift from @xmath25 to @xmath155. our observation suggests to think of the @xmath0 fzzt branes with integer @xmath25 as being built up from discrete objects, or more specifically, as some half - infinite array of point - like objects. they occupy points with even or odd integer position @xmath156 in target space, depending on whether @xmath25 is odd or even. changes of @xmath25 by one unit should therefore correspond to a simple translation of the entire array in the target space. at least qualitatively, such an interpretation in terms of point - like branes appears consistent with our previous study of brane spectra on fzzt branes. in fact, we noticed before that their bands shrink to points when @xmath25 reaches integer values. let us also stress that our geometric picture of the branes with integer @xmath25 is very similar to the interpretation of the dirichlet points in the boundary sine - gordon model (see introduction). more support for the proposed geometric pictures comes from considering spectra involving the zz branes in the @xmath0 theory. the annulus amplitude for two zz branes with labels @xmath149 and @xmath157 was found in @xcite to be of the form @xmath158 \mbox{where } \ \ \chi_{m, n}(q) & = & \eta^{-1}(q) \, \left (q^{-(m / b+nb)^2/4}-q^{-(m / b - nb)^2/4}\right) \\[2 mm] & \stackrel{b = i}{= } & \eta^{-1}(q) \, \left ( q^{(m - n)^2/4}-q^{(m+n)^2/4}\right)\ \. \end{aligned}\]] note that in the @xmath0 theory, excitations of the zz branes have non - negative conformal dimensions. since @xmath159 are characters of irreducible virasoro representations at @xmath0, the expression for the annulus amplitude coincides with what is obtained from the limit of cardy type branes in unitary minimal models. a particularly simple case arises from setting @xmath160 and @xmath161, @xmath162 according to our previous discussion, this amplitude encodes the spectrum of open strings between a single point - like brane and an array of length @xmath163. for the annulus amplitude between a zz brane @xmath149 and an fzzt brane at parameter @xmath25 one has @xmath164 \mbox{where } \ \ \chi_p(q) & = & \eta^{-1}(q) \, q^{p^2}\ \. \nonumber \end{aligned}\]] here, @xmath165 denotes a summation in steps of two. let us point out that once more, the open string excitations between zz and fzzt branes have real, non - negative conformal dimensions when @xmath18. this is in contrast to the situation for @xmath23 where the spectrum contains fields with complex dimensions if @xmath166 or @xmath167. the spectrum between the @xmath140 branes and an fzzt brane of parameter @xmath25 contains a single primary field with conformal dimension @xmath168, in perfect agreement with the construction from unitary minimal models. in fact, the construction of the @xmath0 brane we have proposed along with cardy s rule implies that there appears a single primary with kac label given by eq. ([prescript]) in the corresponding spectrum of the @xmath169 minimal model. when we send @xmath124 to infinity, this field approaches a @xmath0 primary field with momentum @xmath170 + [s] - [f_s m] + \frac{1}{2(m+1) } \, (2[f_s m] + [s]+2) + \mathcal{o}(1/m^2) \ \stackrel{m \rightarrow \infty}{\longrightarrow } \ [s] + f_s \ = \ s \ \.\]] hence, there is perfect agreement between the annulus amplitudes in the @xmath0 limit of unitary minimal models and of liouville theory. let us finally return to the analysis of our fzzt branes at the dirichlet points. since we already understand the effect of shifts in @xmath25, we can restrict the following discussion to the case @xmath171. we would like to probe this fzzt brane with the zz brane @xmath140. the amplitude ([zzfzzt]) then becomes @xmath172 in the second equality, we inserted formula ([ann1l]). this result nicely confirms our interpretation of the fzzt brane with label @xmath171 as an infinite array of point - like branes.
The boundary 2-point function
in this section we shall discuss how the boundary 2-point function of fzzt branes in liouville theory can be continued to @xmath0. this will allow us to derive the exact boundary spectrum, in particular the position and width of the expected band gaps, from liouville theory. our results will confirm nicely the outcome of our previous discussion in the context of unitary minimal models. let us recall that the boundary 2-point functions for the fzzt branes in @xmath50 liouville theory are given by @xmath173 \ \.\]] the coefficient in front of the first term has been set to @xmath174 by an appropriate normalization of the boundary fields. once this freedom of normalization has been fixed, the coefficient of the second term is entirely determined by the physics. it takes the form @xmath175 where @xmath176 and @xmath177. the special function @xmath178 is constructed as a ratio of two barnes double gamma functions (see appendix a.2 for details). @xmath31 is known as the _ reflection amplitude _ since it describes the phase shift of wave functions that occurs when open strings spanning between the fzzt branes with parameters @xmath179 and @xmath180 are reflected by the liouville potential. our aim now is to continue these expressions to the @xmath0 model (see also @xcite for a first attempt in this direction). using once more the general formula ([gasym]) for the behavior of barnes double gamma function at @xmath181, it is not hard to see that the boundary reflection amplitude @xmath31 of liouville theory vanishes outside a discrete set of open string momenta. this behavior is unacceptable since it does not allow for a sensible physical interpretation. actually, it does not originate from the physics of the model but rather from an inappropriate choice of the limiting procedure. to see this we recall that the boundary spectrum of the theory is expected to develop gaps of finite width. for such gaps to emerge from liouville theory, it is necessary that the corresponding states in the liouville model become non - normalizable. but our normalization of boundary fields was chosen such that all states in the boundary theory have unit norm. hence, the normalization we have chosen above is clearly not appropriate for the limit we are about to take. there exists a distinguished normalization of the boundary fields in which the reflection amplitude of the boundary theory becomes trivial. since the usual ` liouville wall'ceases to exist at the point @xmath18, the trivialization of the boundary reflection amplitude seems quite well adapted to our physical setup. explicitly, the new normalization of the boundary fields is given by @xcite, @xmath182 with @xmath183 it is straightforward to rewrite the 2-point function in terms of the fields @xmath184. the result is @xmath185 & & \hspace*{-4 cm } \times\, \frac{\gamma(1 + 2ibp_1) \gamma(2ib^{-1 } p_1) } { \gamma(1-ibs_1)\ga(-ib^{-1}s_1)\ga(1-ibs_2)\ga(-ib^{-1}s_2) } \ \left[\, \delta(q_{b}- \b_1 - \b_2) + \delta(\b_1-\b_2)\, \right] \ \. \nonumber \end{aligned}\]] here we use @xmath186 and the term @xmath187 in the denominator denotes the bulk 3-point couplings of liouville theory evaluated at the points @xmath188 and @xmath189. in writing this expression we have omitted a factor that is constant in the momentum @xmath190 and invariant under the exchange of @xmath179 and @xmath180 since this can easily be absorbed in a redefinition of the boundary fields. when rewritten in the new normalization, the only nontrivial term in the expression ([new2ptf]) is the bulk 3-point function. hence, we can use the results of @xcite (see also section 2) to continue the boundary 2-point function to @xmath0. note, however, that the bulk 3-point coupling needs to be inverted. this is possible whenever it is non - zero. since the coupling @xmath191 vanishes whenever the factor @xmath192 does, we conclude that boundary fields @xmath193 exist for @xmath194 where @xmath195 fields which are labeled by momenta outside this range @xmath33 do not correspond to normalizable states of the model. on a single brane with label @xmath196 we obtain in particular @xmath197 here, @xmath198 $] denotes the fractional part of the boundary parameter as before. this is exactly the same spectrum we found in our discussion of boundary conditions in the @xmath0 limit of unitary minimal models! our result for the 2-point functions of boundary fields in the spectrum of the model is given by @xmath199 \nonumber \\[2 mm] \mbox { where } \ \ e^{q(\a_1,\a_2,\a_3) } & : = & \frac{{\upsilon}(1 + 2i\ta|1)}{{\upsilon}(1|1) } \ \prod_{j=1}^3 \, \frac{{\upsilon}(1 + 2i \ta_j)}{{\upsilon}(1 + 2i\a_j)}\ \. \nonumber \end{aligned}\]] in a final step, we change the normalization of the fields @xmath200 in the @xmath0 theory again so that they possess unit norm, @xmath201 with @xmath202 in an abuse of notation, we have denoted the normalized boundary fields of the @xmath0 model by the same letter @xmath203 as in @xmath50 liouville theory. nevertheless it important to keep in mind that they are not justs limits of the corresponding fields in ordinary liouville theory. after this final change of normalization, the boundary 2-point functions take the form so that the limit of the properly renormalized 2-point function is finite.] @xmath204 \\[3 mm] & & \hspace*{-.5 cm } \mbox { where } \ \ \r^{s_1s_2}_{c=1}(p) \ = \ \r^{is_1\ is_2}_{c=25 } (1-p)^{-1 } \ \ \mbox{and } \ \ { \cal{n}}^{s_1s_2}(p) \ = \ -2\epsilon p_1 (-1)^{[2p_1]+[s_1]+[s_2]}\nonumber \end{aligned}\]] for all @xmath205. note that we have been able to express the `` boundary reflection amplitude '' of the @xmath0 theory through the corresponding quantity of the @xmath206 model. we shall comment on this interesting outcome of our computation in the section 6.
The bulk-boundary coupling constants
our final aim is to compute the bulk - boundary couplings of the @xmath0 boundary liouville models. for liouville theory with real @xmath17, expressions for these couplings were provided by hosomichi in @xcite. we shall depart from these formulas to derive the corresponding couplings in the @xmath0 model. the analysis turns out to be rather intricate. nevertheless, the final answer is in some sense simpler than for @xmath50. let us begin by reviewing hosomichi s formula for the bulk - boundary correlation function latexmath:[\[\langle \phi_{\alpha } (z,\bar{z})\tilde \psi^{ss}_{\beta } (u)\rangle \ = \ |z-\bar{z}|^{h_{\beta}-2h_{\alpha } } labeled by a parameter @xmath25. in our normalization ([normalisation]) for the boundary fields @xmath208 the couplings @xmath209 read @xmath210 \times \ \frac{\gamma_{2 } (q_{b}-\beta|b)\gamma_{2 } (2q_{b}-2\alpha -\beta|b)\gamma_{2 } (2\alpha -\beta|b)\gamma_{2 } (q_{b}+is|b)\gamma_{2 } (q_{b}-is|b)}{\gamma_{2 } (q_{b}-\beta -is|b)\gamma_{2 } (q_{b}-\beta + is|b)\gamma_{2 } (\beta|b)\gamma_{2 } (2\alpha|b)\gamma_{2 } (q_{b}-2\alpha|b) } \label{bbcorrf}\end{gathered}\]] where @xmath211 is given by the integral @xmath212 while sending @xmath17 to @xmath213, an infinite number of poles and zeroes of the integrand approach the integration path. it turns out to be more convenient to evaluate the integral by cauchy s theorem. for @xmath214 we can close the integration contour and rewrite the sum over residues as a combination of basic hypergeometric series (see appendix a.1), @xmath215 & \quad \times \ { } _ { 2}\phi_{1 } \big (e^{-2\pi ib^{-1 } (\beta + 2\alpha -q_{b})},e^{-2\pi ib^{-1}\beta},e^{-4\pi i b^{-1}\alpha};e^{-2\pi ib^{-2}};e^{2\pi ib^{-1 } (\beta -q_{b}+is)}\big) \nonumber \\[2 mm] & \quad \times \ e^{-\pi s (\beta + 2\alpha -q_{b})}s (\beta|b)s (\beta + 2\alpha -q_{b}|b)s (q_{b}-2\alpha|b) + \ (\alpha\rightarrow q_{b}-\alpha)\ \.\label{integral}\end{aligned}\]] our previous results allow us to estimate the behavior of the double gamma functions and its close relative @xmath178 in the vicinity of @xmath18. but the function @xmath216 has not appeared in our discussion before. hence, in order to continue our analysis, we need to study the basic hypergeometric series in the limit when the parameter @xmath217 approaches @xmath218. we shall begin with a separate treatment of the two factors @xmath219 before we combine the results to address the full coupling @xmath209. in dealing with one of the basic hypergeometric series, the basic idea is to approximate the series (we denote the summation parameter by @xmath220) through an integral and to evaluate the latter using a saddle - point analysis. we will not give a rigorous derivation but shall content ourselves with a rough sketch of the argument. this will help to understand the main features of our final formulas. let us introduce the variable @xmath221. using the asymptotic behavior of the q - pochhammer symbols (see appendix b) we obtain @xmath222 where @xmath223 & & -{\text{li}_{2}}(c)+{\text{li}_{2}}(cy)-{\text{li}_{2}}(1)+{\text{li}_{2}}(y)+\log y \, \log z \ \, \end{aligned}\]] and @xmath224 the asymptotics of the integral in eq. ([qhypasym]) can be determined by the method of steepest descent. the leading contribution comes from saddle - points @xmath225, i.e. from points satisfying the condition @xmath226. an elementary computation shows that such saddle - points are obtained as a solution of the quadratic equation @xmath227 once we have understood which saddle - points @xmath225 contribute to the integral, we end up with an expansion of the form @xmath228 the most difficult part, however, is to find the right saddle - points. to begin with, the saddle - point equation ([saddlepeq]) has two different solutions, at least for generic choice of the momenta. but this is not the full story since the function @xmath41 is a complex function with branch - cuts. hence, the description of the saddle point @xmath229 is not complete before we have specified on which branch the relevant saddle - point is located. we were not able to solve the problem in full generality, but for the combination of parameters in the problem at hand, we could find the saddle - points through a comparison with numerical studies (mathematica). let us state the result for the first q - hypergeometric function appearing in eq. ([integral]). here, the parameter @xmath217 is @xmath230 and thus the prefactor @xmath231 of @xmath41 in the exponent becomes @xmath232, where @xmath233 as introduced in section 2. note that the arguments @xmath234 and @xmath235 depend on @xmath17 and thus on @xmath99 and hence they need to be expanded around @xmath236. we shall use the symbols @xmath237 and @xmath238 to denote the leading terms, i.e. @xmath239 where we have set @xmath240 and @xmath241. since the function @xmath41 in the exponential comes with an extra factor @xmath242, the sub - leading term in the @xmath99-expansion of @xmath41 contributes an extra factor which we shall combine with @xmath243 into a new function @xmath40. before we spell out our results on the saddle points, we also have to specify the branch we use for @xmath244. we found it most convenient to choose @xmath245-s+[s])$]. in the end we obtain, @xmath246 with @xmath247 the two solutions of the saddle - point equation are denoted by @xmath248, and they are given explicitly by @xmath249 & \ \ = \ \left\{\begin{array}{l } e^{\pi i (2p_{\alpha } + p_{\beta }) } \frac{\cos (2\pi p_{\alpha})\sin (\pi s)}{\sin (\pi (p_{\beta}+s))}\bigg (1\mp i\tan (2\pi p_{\alpha})\sqrt{1-\frac{\sin (\pi p_{\beta})^{2}}{\sin (\pi s)^{2}\sin (2\pi p_{\alpha})^{2 } } } \bigg)\\[1 mm] \quad \quad \text{for}\ |\sin (\pi p_{\beta})|<|\sin (\pi s)\sin (2\pi p_{\alpha})|\\[3 mm] e^{\pi i (2p_{\alpha } + p_{\beta }) } \frac{\cos (2\pi p_{\alpha})\sin (\pi s)}{\sin (\pi (p_{\beta}+s))}\bigg(1\mp \sqrt{\frac{\sin^{2 } (\pi p_{\beta})-\sin^{2 } (2\pi p_{\alpha})\sin^{2 } (\pi s)}{\sin^{2}(\pi s)\cos^{2 } (2\pi p_{\alpha }) } } \bigg)\\[1 mm] \quad \quad \text{for}\ |\sin (\pi s)|\geq |\sin (\pi p_{\beta})|\geq |\sin (\pi s)\sin (2\pi p_{\alpha})| \\[3 mm] e^{\pi i (2p_{\alpha } + p_{\beta }) } \frac{\cos (2\pi p_{\alpha})\sin (\pi s)}{\sin (\pi (p_{\beta}+s))}\bigg(1\mp \frac{\sin (\pi (s+2p_{\alpha}))}{\sin (\pi s)\cos (2\pi p_{\alpha }) } \sqrt{\frac{\sin^{2 } (\pi p_{\beta})-\sin^{2 } (2\pi p_{\alpha})\sin^{2 } (\pi s)}{\sin ^{2 } (\pi (s+2p_{a })) } } \bigg)\\[1 mm] \quad \quad \text{for}\ |\sin (\pi p_{\beta})|\geq |\sin (\pi s)| \end{array } \right.\end{aligned}\]] the last of these three different cases concerns the band gaps, i.e. momenta @xmath250. the first two cases, on the other hand, apply to the bands so that all bands appear to be split into an inner (first case) and an outer (second case) region, depending on the value of the bulk momentum @xmath251. analyzing the limit in the outer part of the bands turns out to be the most difficult task because the two saddle - points @xmath252 are of the same order, i.e. @xmath253, and there are at least some subsets in momentum space where they both contribute. within the inner region and the gap, the saddle - point @xmath254 dominates so that only the summand @xmath255 remains in our asymptotic expansion. the exact description of our findings for the intermediate region is rather cumbersome. fortunately, the distinction between inner and outer parts of the band disappears once we combine our two basic hypergeometric series into the bulk - boundary coupling. since our main goal is to provide a formula for the @xmath0 limit of @xmath209, we shall not attempt to present our findings for @xmath256 in the outer region. instead, let us recall from above that we also need to specify the branch on which the saddle - points sit. the prescription is as follows : if @xmath257, then we choose the branch by moving it from the fundamental branch first to @xmath258, infinitesimally above / below the real axis depending on whether the fractional part @xmath259 $] is greater or smaller than @xmath260. then we move it on the shortest arc to @xmath225. having completed the analysis for the first hypergeometric function we now turn to the second for which an analogous investigation gives the following results, @xmath261 and @xmath262 note that in this case @xmath263 is directly given by @xmath264. the relevant saddle - points are @xmath265 our discussion of the three different cases and the choice of branches carries over from the previous discussion without any significant changes. before we return to the study of @xmath209 it is necessary to make one more important observation which relates the two saddle - points @xmath266 and @xmath267. it is not difficult to see that @xmath268 solves the saddle - point equation ([saddlepeq]) of @xmath266. indeed we find for @xmath269 that @xmath270. this fact along with some simple formulas in appendix a.3 allows to re - express the final formulas entirely in terms of @xmath271. now we have all ingredients at our disposal to analyze the asymptotic behavior of the bulk - boundary correlator. we must combine the divergences coming from the q - hypergeometric functions with the divergences of the prefactors in eq. ([bbcorrf]). since we know already that boundary fields with @xmath272 do not exist, we can restrict our evaluation of @xmath209 to the inner and outer band regions (see discussion above). in the inner region, only one saddle - point contributes and the divergent terms may be determined easily from the formulas we have provided. it then turns out that they cancel each other thanks to a rather non - trivial dilogarithm identity which is attributed to ray @xcite. for convenience we state ray s identity in appendix a.3 (eq.([5term])). hence, we are left with a finite result. after a tedious but elementary computation we obtain @xmath273 where @xmath274. in the derivation we used the fact that @xmath275 and @xmath276 are the two solutions of the same quadratic saddle - point equation along with a few elementary identities which we list in appendix a.3. within the outer regions of the bands, the q - hypergeometric functions can have contributions from both saddle - points. it is possible to show that out of the four possible terms in our product of two q - hypergeometric functions, only two appear at any point in momentum space. moreover, after the divergences of the prefactors have been taken into account, only one of the two terms is free of divergences. the other keeps a rapidly oscillating factor. the cancellation is again due to ray s identity and the finite contributions come from the combination @xmath277. in the end, the result for the outer region of the band is therefore given by the same formula as for the inner region. it now remains to change the normalization of our boundary fields. we pass from the fields @xmath208 to @xmath203 using the prescription ([newnormalisation]), as before. our result then reads @xmath278 & \quad \times \ \big(\frac{e^{-2\pi i p_{\alpha } (s+1)}}{1-e^{-4\pi ip_{\alpha } } } \frac{g (y_{+})^{2}e^{-\frac{ih (y_{+})}{2\pi}}}{y_{+}^{2}h '' (y_{+})}\frac{\gamma_2 (1|1)\gamma_2 (1 + 2p_{\beta}|1)\gamma_2 (1 - 2p_{\alpha}|1)^{2}}{\gamma_2 (1 - 2p_{\alpha}-p_{\beta}|1)\gamma_2 (1 - 2p_{\alpha}+p_{\beta}|1)\gamma_2 (1+p_{\beta } |1)^{2 } } \\ & \quad \quad \quad \ -\ (p_{\alpha}\to -p_{\alpha})\big)\ \.\end{aligned}\]] this is the formula we anticipated in the introduction. for a direct comparison one has to employ the integral representation ([barnesg]) of barnes double gamma function. before we conclude this section, we would like to comment briefly on the band - gaps. actually, we can use the analysis of the bulk - boundary couplings to confirm our previous statement that boundary fields @xmath279 with @xmath280 decouple from the theory. to this end, we shall focus on the coefficients @xmath281 of the bulk boundary operator product expansion rather than the couplings @xmath35. the former are related to latter by multiplication with the boundary 2-point function. our claim is that the coefficients @xmath282 vanish within the band - gaps. as we have seen before, the 2-point function contributes a factor that diverges when @xmath272. with the help of our asymptotic expansion formulas in this section it is possible to show that the couplings @xmath35 always diverge slower hence, boundary fields whose momenta lie in the gaps can not be excited when a bulk field approaches the boundary.
Conclusion and open problems
in this work we have constructed boundary conditions of the euclidean @xmath0 liouville model. in particular we argued that there exists a family of boundary conditions that is parametrized by one real parameter @xmath22. we provided explicit expressions for their boundary states (see eq. ([1ptfceq1])), the boundary 2-point function ([2ptfctres1],[reflb]), and the bulk - boundary coupling ([bcoupl]). the only quantity that we are missing for a complete solution of the model is the boundary 3-point coupling. in the case of liouville theory, the corresponding formulas have been found by ponsot and teschner in @xcite. we believe that the analysis of their @xmath0 limit can proceed along the lines of the studies we have presented in section 5, but obviously this program remains to be carried out in detail. a crucial ingredient in our study was the formula ([gasym]) for the asymptotics of barnes double gamma function. in section 2, the latter enabled us to provide a new derivation of the bulk couplings in the @xmath0 liouville model. our approach here is simpler and more general than the one that was developed in @xcite. with this technical progress it is now also possible to calculate the bulk couplings of non - rational conformal field theories with @xmath116. in contrast to the @xmath0 case, the models with @xmath116 are certainly non - unitary. we will comment on the results and their relation with minimal models elsewhere. possible applications of such developments include the 2-dimensional cigar background with @xmath283 @xcite and similar limits of the associated boundary theories @xcite. all this research, however, was mainly motivated by the desire to construct an exact conformal field theory model for the homogeneous condensation of open string tachyons, such as the tachyon on an unstable d0 brane in type iib theory. as we explained in the introduction, the underlying world - sheet theory is a lorentzian version of the model we have considered in this work. but since the couplings of our theory are not analytic in the momenta (with the boundary state being the only exception), this lorentzian theory can not be obtained by a simple wick rotation from the solution we described here. instead, it was suggested in @xcite to perform the wick rotation before sending @xmath17 to its limiting value @xmath18. such a prescription makes sense because the couplings of liouville theory are analytic in the momenta as long as @xmath17 is not purely imaginary. it is certainly far from obvious that the wick rotated couplings again possess a well - defined @xmath18 limit. preliminary studies of this issue show that our analysis of the boundary 2-point function extends to the lorentzian case. in the time - like model, the spectrum of boundary fields is continuous (there are no gaps) and their 2-point function is still given by eq. ([2ptfctres1]) with a reflection amplitude @xmath284 this result does not agree with the proposal in @xcite. note, however, that in the lorentzian case the correlation functions can depend on the details of how @xmath17 approaches @xmath18. such a behavior might be related to a choice of vacuum. let us also anticipate that consistency of our expression for the boundary 2-point coupling with the half - brane boundary states @xcite can be checked by means of the modular bootstrap. a related observation, though with a somewhat problematic domain of open string momenta, has also been made in @xcite. concerning the bulk - boundary coupling, our investigations are not complete yet. but before having worked out the expression in the lorentzian model, it is worthwhile comparing our formula ([bcoupl]) for the bulk - boundary coupling with a corresponding expression in the time - like theory that was suggested recently in @xcite (eq. [4.14] of that paper). in fact, the exponential in the second line of our formula is identical to a corresponding factor in the work of balasubramanian et. it will be interesting to determine the other factors through our approach. we shall return to this issue in a forthcoming publication. * acknowledgements : * we would like to thank v. balasubramanian, p. etingof, j. frhlich, m.r. gaberdiel, k. graham, g. moore, b. ponsot, i. runkel, j. teschner and g. watts for very helpful discussions and some crucial remarks. part of this research has been carried out during a stay of vs with the string theory group of rutgers university and of both authors at the erwin schroedinger institute for mathematical physics in vienna. we are grateful for their warm hospitality during these stays.
Some background on special functions
in this first appendix we collect a few results on the special functions which appear in the analysis of the @xmath0 boundary liouville model. we shall start with some q - deformed special functions and then introduce barnes double gamma functions and certain closely related special functions. dilogarithms and some of their properties are finally reviewed in the third subsection. one of the most basic objects in the theory of q - deformed special functions is the finite q - pochhammer symbol (see e.g. @xcite) @xmath285 its limit for @xmath286 exists for @xmath287 and is denoted by @xmath288 the tilde is used to avoid confusion with our convention for @xmath289 in the rest of the text. with the help of the the finite q - pochhammer symbol we can now introduce the basic hypergeometric series @xmath290. this q - deformation of the hypergeometric function @xmath291 is defined as @xmath292 the interested reader can find many basic properties of these functions in the literature (see e.g. @xcite). all the properties we need in the main text are stated and derived there. barnes double @xmath61-function @xmath293 is defined for @xmath294 and complex @xmath17 with @xmath295 (see @xcite), and can be represented by an integral (for @xmath296), @xmath297 whenever @xmath298. here we have also introduced the symbol @xmath299 throughout the main text, we often use the following special combinations of double gamma functions, @xmath300 s (x|b) & = & \gamma_{2 } (x|b) \, \gamma_{2 } (q_{b}-x|b)^{-1}\ \.\end{aligned}\]] for the latter, we would also like to spell out an integral representation that easily follows from the formula ([barnesg]) above, @xmath301 this representation is valid for @xmath302. let us also note that @xmath178 is unitary in the sense that @xmath303 the function @xmath178 is also closely related to ruijsenaars hyperbolic gamma function @xmath304 (see @xcite), @xmath305 many further properties of double gamma functions, in particular on the position of their poles and shift properties, can be found in the literature (see e.g. @xcite). here we would like to prove one relation that involves the q - pochhammer symbols we introduced in the previous subsection. to this end, we depart from the integral representation ([barnesg]). let us denote the integrand in formula ([barnesg]) by @xmath306, s.t. the function @xmath243 has the property @xmath308 we evaluate the integral by closing the contour in the first quadrant, @xmath309 & = & -\log \gamma_{2 } (-ib+ix|-ib) - \int_{0}^{\infty } \frac{dt}{2 t } (\tfrac{q_{b}}{2}-x)^{2 } (e^{-it}-e^{-t})-2\pi i \sum_{\text{poles}\ t_{n}}\!\!\text{res}_{t = t_{n } } f (t, x, b) \nonumber\end{aligned}\]] where we have used the formula ([frot]) in passing to the second line. the integral in the second term is @xmath310 the poles appearing in the sum over residues are at @xmath311 with @xmath312 hence we find @xmath313 replacing @xmath314 through @xmath315 and changing the order of summation, we obtain @xmath316 with @xmath317. we can now re - express the infinite product with the help of q - pochhammer symbols (see appendix a.1). inserting this result and eq. ([hilfsintegral]) into eq. ([tempresult]) we finally arrive at @xmath318 this formula is the starting point for our evaluation of the asymptotic behavior of @xmath319 near @xmath320. we shall return to this issue in appendix b. there is one more special function that plays an important role in our analysis : euler s dilogarithm (l. euler 1768). it is defined by @xmath321 for @xmath322. at @xmath323, we find @xmath324. the dilogarithm has the integral representation @xmath325 it can be analytically continued with a branch cut along the real axis from 1 to @xmath326. at @xmath323, the dilogarithm is still continuous, but not differentiable. properties of dilogarithms are used frequently to derive the formulas that appear in the main text. here we list the most relevant properties @xmath327 { \text{li}_{2}}(z)+{\text{li}_{2}}(1-z) & = & { \text{li}_{2}}(1)-\log (z)\, \log (1-z)\\[2 mm] { \text{li}_{2}}(-z)+{\text{li}_{2}}(-1/z) & = & 2{\text{li}_{2}}(-1)-\frac{1}{2}\log ^{2 } (z)\\[2 mm] \label{liclosetoone } { \text{li}_{2}}(e^{-\epsilon }) & = & { \text{li}_{2}}(1)- (1-\log \epsilon)\epsilon + o (\epsilon) \ \. \end{aligned}\]] many more properties of the dilogarithm can be found in the literature (see e.g. @xcite). of particular relevance for our analysis of the bulk - boundary 2-point function is the following equality @xmath328 & & \hspace*{-3 cm } = \ { \text{li}_{2}}(z)-{\text{li}_{2}}(z\tfrac{\eta_{1}\eta_{2}}{\zeta_{1 } \zeta_{2}})+\log z \log (x_{1}x_{2}\zeta_{1}\zeta_{2 }) \ \ \end{aligned}\]] here, @xmath329 are the two solutions of the quadratic equation @xmath330 equation ([5term]) is derived from the usual 5-term relation of the dilogarithm (see e.g. @xcite) with the help of the following list of relations that hold for any two solutions @xmath331 of the equation ([quad]), @xmath332 \left(1-z\, \frac{\eta_1\eta_2}{\zeta_1\zeta_2}\right) (1-\zeta_i x_1) (1-\zeta_i x_2) & = & \frac{(\zeta_i - \eta_1)(\zeta_i-\eta_2)}{\zeta_i } \label{eq2 } \\[2 mm] x_1 x_2 & = & \frac{1-z}{\zeta_1\zeta_2 - z \eta_1 \eta_2 } \label{eq3 } \ \. \end{aligned}\]] we leave the details to the reader. let us note that the properties ([eq1]) to ([eq3]) are also used frequently to simplify the final formula for the bulk - boundary 2-point function.
Asymptotic behavior of q-pochhammer symbols
in order to evaluate the behavior of the double gamma function near @xmath18 one can start from our equation ([dgqpoch]). since the double gamma function that appears on the right hand side of this equation is analytic at @xmath18, the main issue is to understand the asymptotic behavior of the q - pochhammer symbol. this is what we are concerned with here. to begin with, we shall parametrize @xmath217 by @xmath333. in general, the limit can depend on the way we send @xmath334 to zero. we start by taking the logarithm of the definition , @xmath335)\frac{d}{dx}\big(\log (1-a\,\tilde{q}^{z/2+x }) \big)\big|_{x = t}\, dt \ \.\end{aligned}\]] here, we used the euler - maclaurin sum formula to re - express the sum as an integral, @xmath336 is the first bernoulli polynomial, @xmath337 note that the branch of the logarithm @xmath338 is chosen such that it approaches @xmath339 when we send @xmath156 along the real axis toward @xmath326. the first integral is given by the dilogarithm (see appendix [dilog] for the definition and properties), so we obtain @xmath340 where @xmath341 denotes the integral containing the bernoulli polynomial, @xmath342) \frac{(-a)\log (\tilde{q})\, \tilde{q}^{z/2+t}}{1-a\, \tilde{q}^{z/2+t}}\, dt \ \.\]] as long as @xmath343, @xmath344 is suppressed by @xmath345. on the other hand, if @xmath346, then the integral gives a non - trivial contribution. let us explain that in more detail. we rewrite the integral as @xmath347) \frac{1}{a^{-1}\, e^{\epsilon z/2}\, e^{t}-1}dt \ \.\]] when @xmath99 goes to zero, the strongly oscillating sign of the bernoulli polynomial suppresses the integral, so that generically @xmath344 vanishes in that limit. the leading contribution comes from the pole of the integrand, and if it approaches the path of integration when sending @xmath99 to zero, we can get a finite contribution. for @xmath343, the pole is always outside of the integration region, but for @xmath348, the pole might come close to the real axis. we can take this effect into account by extracting the pole part from the integrand, @xmath349 the regular part will not contribute to the integral in the limit @xmath92, so we can rewrite @xmath344 as @xmath350)\frac{1}{t-\frac{\log a}{\epsilon}+\frac{z}{2 } } + o (\epsilon^{0})\ \.\]] the remaining integral is given by the difference of @xmath351 and the first terms of its stirling series, @xmath352 as long as @xmath353 and @xmath354, the limit of @xmath355 vanishes because the @xmath61-function approaches its sterling approximation. if @xmath346, we find @xmath356 \ \.\]] this formula together with describe the asymptotics of q - pochhammer symbols in the cases we shall need for our applications. in the following we want to apply the gained insight in a number of special cases. first, let us write down the full asymptotics for @xmath346, @xmath357 where we used from appendix [dilog] to expand the dilogarithm close to 1. we observe that this expression has zeroes for @xmath358 which is consistent with the general definition . for integer @xmath235 these asymptotics can alternatively be derived using modular properties of dedekind s @xmath359-function. we are often interested in q - pochhammer - symbols of the form @xmath360 with some function @xmath361. in most cases, it is enough to expand @xmath243 to first order around @xmath362, and we find @xmath363 this relation breaks down if @xmath364 and @xmath365 for some @xmath366. let us look at an example. set @xmath367 and @xmath368 for @xmath369. then we obtain @xmath370 by carefully choosing the correct branch of the logarithm we can rewrite the result as @xmath371 - 1/2) } \big)^{(1-z - p)/2 } \ \big (1+o (\epsilon^{0 }) \big) \ \, \]] where @xmath73 $] denotes the largest integer smaller or equal @xmath34. let us finally look at the behavior of @xmath372 when @xmath373 is close to 1. obviously @xmath374, and the derivative is easily obtained as @xmath375 similar formulas for the asymptotics of q - pochhammer symbols have been derived before (see in particular @xcite). a. sen and b. zwiebach, _ tachyon condensation in string field theory _, jhep * 0003 * (2000) 002 [hep - th/9912249]. w. taylor and b. zwiebach, _ d - branes, tachyons, and string field theory _, [hep - th/0311017]. m. r. garousi, _ tachyon couplings on non - bps d - branes and dirac - born - infeld action _ b * 584 * (2000) 284 [hep - th/0003122]. a. sen, _ tachyon matter _, jhep * 0207 * (2002) 065 [hep - th/0203265]. j. a. harvey, d. kutasov and e. j. martinec, _ on the relevance of tachyons _, [hep - th/0003101]. v. schomerus, _ lectures on branes in curved backgrounds _, class. * 19 * (2002) 5781 [hep - th/0209241]. a. sen, _ rolling tachyon _, jhep * 04 * (2002) 048 [hep - th/0203211]. m. gutperle and a. strominger, _ spacelike branes _, jhep * 0204 * (2002) 018 [hep - th/0202210]. a. sen, _ universality of the tachyon potential _, jhep * 9912 * (1999) 027 [hep - th/ 9911116]. j. callan, curtis g., i. r. klebanov, a. w. w. ludwig and j. m. maldacena, _ exact solution of a boundary conformal field theory _, nucl. (1994) 417448 [hep - th/9402113]. j. polchinski and l. thorlacius, _ free fermion representation of a boundary conformal field theory _ * d50 * (1994) 622626 [hep - th/9404008]. p. fendley, h. saleur and n. p. warner, _ exact solution of a massless scalar field with a relevant boundary interaction _, nucl. * b430 * (1994) 577596 [hep - th/9406125]. a. recknagel and v. schomerus, _ boundary deformation theory and moduli spaces of d - branes _, nucl. phys. * b545 * (1999) 233282 [hep - th/9811237]. m. r. gaberdiel and a. recknagel, _ conformal boundary states for free bosons and fermions _, jhep * 11 * (2001) 016 [hep - th/0108238]. m. gutperle, a. strominger, _ timelike boundary liouville theory _, [hep - th/0301038]. i. affleck, w. hofstetter, d. r. nelson and u. schollwock, _ non - hermitian luttinger liquids and flux line pinning in planar superconductors _, [cond - mat/0408478]. h. dorn and h. j. otto, _ two and three point functions in liouville theory _, nucl. * b429 * (1994) 375388 [hep - th/9403141]. a. b. zamolodchikov and a. b. zamolodchikov, _ structure constants and conformal bootstrap in liouville field theory _, nucl. * b477 * (1996) 577605 [hep - th/9506136]. b. ponsot and j. teschner, _ liouville bootstrap via harmonic analysis on a noncompact quantum group _, [hep - th/9911110]. j. teschner, _ liouville theory revisited _, class. * 18 * (2001) r153r222 [hep - th/0104158]. j. teschner, _ a lecture on the liouville vertex operators _, [hep - th/0303150]. v. fateev, a. b. zamolodchikov and a. b. zamolodchikov, _ boundary liouville field theory. i : boundary state and boundary two - point function _, [hep - th/0001012]. j. teschner, _ remarks on liouville theory with boundary _, [hep - th/0009138]. k. hosomichi, _ bulk - boundary propagator in liouville theory on a disc _, jhep * 11 * (2001) 044 [hep - th/0108093]. b. ponsot and j. teschner, _ boundary liouville field theory : boundary three point function _, nucl. phys. * b622 * (2002) 309327 [hep - th/0110244]. a. b. zamolodchikov and a. b. zamolodchikov, _ liouville field theory on a pseudosphere _, [hep - th/0101152]. v. schomerus, _ rolling tachyons from liouville theory _, jhep * 0311 * (2003) 043 [hep - th/0306026]. i. runkel and g. m. t. watts, _ a non - rational cft with c = 1 as a limit of minimal models _, jhep * 09 * (2001) 006 [hep - th/0107118]. e. j. martinec, _ the annular report on non - critical string theory _, [hep - th/0305148]. j. teschner, _ on boundary perturbations in liouville theory and brane dynamics in noncritical string theories _, jhep * 0404 * (2004) 023 [hep - th/0308140]. k. r. kristjansson and l. thorlacius, _ c = 1 boundary conformal field theory revisited _, class. * 21 *, s1359 (2004) [hep - th/0401003]. a. strominger and t. takayanagi, _ correlators in timelike bulk liouville theory _, [hep - th/0303221]. f. larsen, a. naqvi and s. terashima, _ rolling tachyons and decaying branes _, jhep * 02 * (2003) 039 [hep - th/0212248]. j. l. cardy, _ boundary conditions, fusion rules and the verlinde formula _, nucl. b * 324 * (1989) 581. k. graham, i. runkel and g. m. t. watts, _ minimal model boundary flows and c = 1 cft _, nucl. b * 608 * (2001) 527 [hep - th/0101187]. s. fredenhagen and v. schomerus, _ d - branes in coset models _, jhep * 0202 * (2002) 005 [hep - th/0111189]. g. ray, _ multivariable polylogarithm identities _, in _ structural properties of polylogarithm _, l. lewin (ed), math. surveys and monographs 37 (1991) 123. y. hikida and t. takayanagi, _ on solvable time - dependent model and rolling closed string tachyon _, [hep - th/0408124]. s. ribault and v. schomerus, _ branes in the 2-d black hole _, jhep * 0402 *, 019 (2004) [hep - th/0310024]. s. fredenhagen and v. schomerus, _ on minisuperspace models of s - branes _, jhep * 0312 *, 003 (2003) [hep - th/0308205]. j. l. karczmarek, h. liu, j. maldacena and a. strominger, _ uv finite brane decay _, jhep * 0311 *, 042 (2003) [hep - th/0306132]. v. balasubramanian, e. keski - vakkuri, p. kraus and a. naqvi, _ string scattering from decaying branes _, [hep - th/0404039]. g. e. andrews, r. askey and r. roy, _ special functions_. cambridge university press, cambridge, 1999. m. jimbo and t. miwa, _ qkz equation with @xmath376 and correlation functions of the xxz model in the gapless regime _, j. phys. * a29 * (1996) 29232958 [hep - th/9601135]. a. n. kirillov, _ dilogarithm identities _, prog. * 118 * (1995) 61 [hep - th/9408113]. r. j. mcintosh, _ some asymptotic formulae for q - shifted factorials _, the ramanujan journal * 3 * (1999) 205. | the @xmath0 liouville theory has received some attention recently as the euclidean version of an exact rolling tachyon background. in an earlier paper
it was shown that the bulk theory can be identified with the interacting @xmath0 limit of unitary minimal models. here
we extend the analysis of the @xmath0-limit to the boundary problem.
most importantly, we show that the fzzt branes of liouville theory give rise to a new 1-parameter family of boundary theories at @xmath0.
these models share many features with the boundary sine - gordon theory, in particular they possess an open string spectrum with band - gaps of finite width.
we propose explicit formulas for the boundary 2-point function and for the bulk - boundary operator product expansion in the @xmath0 boundary liouville model. as a by - product of our analysis
we also provide a nice geometric interpretation for zz branes and their relation with fzzt branes in the @xmath0 theory. | hep-th0409256 |
Introduction
one of the first striking applications of gromov s theory of pseudoholomorphic curves @xcite was that a closed exact lagrangian immersion @xmath4 inside a liouville manifold must have a double - point, given the assumption that it is hamiltonian displaceable. gromov s result has the following contact - geometric reformulation, which will turn out to be useful. consider the so - called _ contactisation _ @xmath5 of the liouville manifold @xmath6, which is a contact manifold with the choice of a contact form. recall that a (generic) exact lagrangian immersion @xmath4 lifts to a legendrian (embedding) @xmath7. one says that @xmath1 is _ horizontally displaceable _ given that @xmath8 is hamiltonian displaceable. the above result thus translates into the fact that a horizontally displaceable legendrian submanifold @xmath1 must have a _ reeb chord _ for the above standard contact form i.e. a non - trivial integral curve of @xmath9 having endpoints on @xmath1. a similar result holds for legendrian submanifolds of boundaries of subcritical weinstein manifolds, as proven in @xcite by mohnke. in the spirit of arnold @xcite, the following conjectural refinement of the above result was later made : the number of reeb chords on a chord - generic legendrian submanifold @xmath7 whose lagrangian projection is hamiltonian displaceable is at least @xmath10. however, as was shown by sauvaget in @xcite by the explicit counter - examples inside the standard contact vector space @xmath11, @xmath12, the above inequality is not true without additional assumptions on the the legendrian submanifold ; also, see the more recent examples constructed in @xcite by ekholm - eliashberg - murphy - smith. the latter result is based upon the h - principle proven in @xcite by eliashberg - murphy for lagrangian cobordisms having loose negative ends in the sense of murphy @xcite. on the positive side, the above arnold - type bound has been proven using the legendrian contact homology of the legendrian submanifold, under the additional assumption that the legendrian contact homology algebra is sufficiently well - behaved. legendrian contact homology is a legendrian isotopy invariant independently constructed by chekanov @xcite and eliashberg - givental - hofer @xcite, and later developed by ekholm - etnyre - sullivan @xcite. this invariant is defined by encoding pseudoholomorphic disc counts in the legendrian contact homology differential graded algebra (dga for short) which usually is called the _ chekanov - eliashberg algebra _ of the legendrian submanifold. in the case when the chekanov - eliashberg algebra of a legendrian admits an augmentation (this should be seen as a form of non - obstructedness for its floer theory), the above arnold - type bound was proven by ekholm - etnyre - sullivan in @xcite and by ekholm - etnyre - sabloff in @xcite. in @xcite, the authors generalised this proof to the case when the chekanov - eliashberg algebra admits a finite - dimensional matrix representation, in which case the same lower bound also is satisfied. the above arnold - type bound is also related to the one regarding the number of hamiltonian chords between the zero - section in @xmath13 (or, more generally, any exact closed lagrangian submanifold of a liouville manifold) and its image under a generic hamiltonian diffeomorphism. namely, such hamiltonian chords correspond to reeb chords on a legendrian lift of the union of the lagrangian submanifold and its image under the hamiltonian diffeomorphism. in fact, as shown by laudenbach - sikorav in @xcite, the number of such chords is bounded from below by the stable morse number of the zero - section (and hence, in particular, it is bounded from below by half of the betti numbers of the disjoint union of _ two _ copies of the zero - section). arnold originally asked whether this bound can be improved, and if in fact the _ morse number _ of the zero - section is a lower bound. however, this question seems to be out of reach of current technology. on the other hand, we note that the stable morse number is equal to the morse number in a number of cases ; see @xcite as well as section [sec : gendefs] below for more details. finally, we mention the remarkable result by ekholm - smith in @xcite, which shows that the smooth structure itself can predict the existence of more double points than the original bound given in terms of the homology. namely, a @xmath14-dimensional manifold @xmath15 for @xmath16 that admits a legendrian embedding having precisely one transverse reeb chord in the standard contact space must be _ diffeomorphic _ to the standard sphere unless @xmath17. also see @xcite for similar results in other dimensions. in this paper, we will explore a priori lower bounds for the number of reeb chords on a legendrian submanifold @xmath7, given that it admits an exact lagrangian filling @xmath18 inside the symplectisation. recall that the condition of admitting an exact lagrangian filling is invariant under legendrian isotopy ; see e.g. @xcite. the bound will be given in terms of the simple homotopy type of @xmath19. first, we recall that such a legendrian submanifold automatically has a well - behaved chekanov - eliashberg algebra ; namely, an exact lagrangian filling induces an augmentation by @xcite. in the case when the projection of @xmath1 to @xmath20 is displaceable, the aforementioned result can thus be applied, giving the above arnold - type bound. however, in this case, there are even stronger bounds that can be obtained from the topology of the exact lagrangian filling @xmath19 (and without the assumption of horizontal displaceability). see section [sec : wrapped] below for previous such results as well as an outline of the proof, which is based upon seidel s isomorphism in wrapped floer homology. this is also the starting point of the argument that we will use in order to prove our results here. in the following we assume that a legendrian submanifold @xmath21 is chord - generic and has an exact lagrangian filling @xmath22. here @xmath23 denotes the coordinate on the first @xmath24-factor. in particular, the set of reeb chords @xmath25 of @xmath1 is finite. further, the set of reeb chords @xmath26 in degree @xmath27 will be denoted by @xmath28, where the grading is induced by the conley - zehnder index modulo the maslov number @xmath29 of @xmath19 as defined in @xcite. observe that @xmath30 in particular implies that the first chern class of @xmath6 vanishes on @xmath31. for a group @xmath32 being the epimorphic image of @xmath33, consider the morse homology complex @xmath34),\partial_f)$] of @xmath19 with coefficients in the group ring @xmath35 $] twisted by the fundamental group, where @xmath36 is a unital commutative ring and @xmath37 is a morse function satisfying @xmath38 outside of a compact set. (the generators of this complex are graded by the morse index, and the differential counts negative gradient flow lines.) [mainthmsmplehomeqnts] let @xmath22 be an exact lagrangian filling of an @xmath39-dimensional closed legendrian submanifold @xmath40 with fundamental group @xmath41 and maslov number @xmath29. * in the case when the filling is spin and when @xmath30, the morse homology complex @xmath42),\partial_f)$] is simple homotopy equivalent to a @xmath43$]-equivariant complex @xmath44\langle \mathcal{q}_{n-\bullet}(\lambda) \rangle,\partial)$] ; * in the general case, it follows that the complex @xmath45),\partial_f)$] is homotopy equivalent in the category of @xmath32-equivariant complexes to a complex @xmath46\langle \mathcal{q}_{n-\bullet}(\lambda) \rangle,\partial)$] with grading in @xmath47. here we can always take @xmath48, while we are free to choose an arbitrary unital commutative ring in the case when @xmath19 is spin. we prove theorem [mainthmsmplehomeqnts] in section [maintheoremanditsconsequences]. now let @xmath49 denote the stable morse number of a manifold @xmath50 with possibly non - empty boundary, see definition [defn : stablemorse]. using theorem [mainthmsmplehomeqnts] and the adaptation of (*??? * theorem 2.2) to the case of manifolds with boundary (see proposition [prop : morse]), the following result is immediate : [maininequalitystablemorsenumberofafilling] suppose that @xmath51 is a chord - generic closed legendrian submanifold admitting an exact lagrangian filling @xmath19 which is spin and has vanishing maslov number. it follows that the bound latexmath:[\]] where @xmath604 is a subgroup of @xmath605 $] generated by @xmath606-[\phi(h)]$]. the proof can, for example, be deduced from (*??? * proposition 29). then we describe a series of finite solvable groups @xmath607 with the property that @xmath608, @xmath609)=1 $]. let @xmath610 be a finite field with @xmath611 elements, where @xmath612 is a prime number. we define a group @xmath613, where @xmath614 acts on the additive group @xmath615 by @xmath616, @xmath617, @xmath618. it is easy to see that @xmath607 is a solvable group. this follows from the existence of the following subnormal series @xmath619 where @xmath620 and @xmath621. then we observe that from formula [absemiddirprreltohsh] it follows that @xmath622\simeq { \mathbb{f}}^{\ast}_q$] is a non - trivial cyclic group. therefore, @xmath609)=1 $]. finally we prove that @xmath608. since @xmath623 and @xmath624, we get that @xmath625. we first show that @xmath626. assume that @xmath607 has a generating set @xmath627 with @xmath628, @xmath629 and @xmath630. then, note that every element in the group generated by @xmath631 has a form @xmath632, where @xmath633, @xmath634, and @xmath635. this leads to the contradiction with the fact that @xmath623. then we take a set of generators @xmath631 and @xmath636 with the property that @xmath330 is a generator of @xmath637 (@xmath637 is a cyclic group). such an element definitely exists since if all the elements of @xmath631 are of the form @xmath638, where @xmath496 is not a generator of @xmath637, then @xmath631 is not a generating set of @xmath607. again, @xmath637 is a cyclic group of order @xmath639, and the order of @xmath330 that we denote by @xmath640 is coprime to @xmath641. this implies that @xmath642 is coprime to @xmath641, and hence none of the primes which divide @xmath643 will divide @xmath644=|{\mathbb{f}}^m_q|$]. let @xmath282 be the set of primes which divide @xmath640. then @xmath645 and @xmath646 are two hall @xmath282-subgroups, and hence by theorem [conjpihallsubgr] they are conjugate by some element @xmath647. this implies that @xmath631, after conjugation, contains an element @xmath648, where @xmath649 is a generator of @xmath637. we also would like to mention that it is possible to find @xmath173 explicitly without relying on the theory of hall @xmath282-subgroups. then, already knowing that @xmath626, we can apply the previous argument and see that, in fact, @xmath650. together with the fact that @xmath625 we get that @xmath608. using proposition [constrrealizoffundgroup], we construct an exact lagrangian filling @xmath651 of a legendrian @xmath652 inside the standard contact vector space. then theorem [intrineqabssi] tells us that @xmath653 is satisfied for all representatives. on the other hand, the bound given by seidel s isomorphism is @xmath654 finally, note that the difference between the previous two bounds gets arbitrarily large as @xmath655.
Acknowledgements
we would like to thank franois charette and jarek kdra for very helpful conversations and interest in our work. in addition, we are grateful to the referee of an earlier version of this paper for many valuable comments and suggestions. b. chantraine, g. dimitroglou rizell, p. ghiggini, and r. golovko, _ floer homology and lagrangian concordance _, proceedings of 21st gkova geometry - topology conference 2014, 76113, gkova geometry / topology conference (ggt), gkova, 2015. t. ekholm, _ rational sft, linearized legendrian contact homology, and lagrangian floer cohomology _, perspectives in analysis, geometry, and topology. on the occasion of the 60th birthday of oleg viro, volume 296, pages 109145. springer, 2012. k. fukaya, p. seidel and i. smith, _ the symplectic geometry of cotangent bundles from a categorical viewpoint _, homological mirror symmetry, lecture notes in phys. 757, springer, berlin, 2009, 126. | assume that we are given a closed chord - generic legendrian submanifold @xmath0 of the contactisation of a liouville manifold, where @xmath1 moreover admits an exact lagrangian filling @xmath2 inside the symplectisation. under the further assumptions that this filling is spin and has vanishing maslov class, we prove that the number of reeb chords on @xmath1 is bounded from below by the stable morse number of @xmath3. given a general exact lagrangian filling @xmath3
, we show that the number of reeb chords is bounded from below by a quantity depending on the homotopy type of @xmath3, following ono - pajitnov s implementation in floer homology of invariants due to sharko.
this improves previously known bounds in terms of the betti numbers of either @xmath1 or @xmath3. | 1510.08838 |
Introduction
the top quark is the heaviest standard model (sm) particle which has been discovered so far, and might be the first place in which new physics effects could appear. while the top quark has been discovered and studied in some details at the tevatron, many of its properties are being studied at the lhc with high precision. since new physics can show up in the couplings of the top quark with other sm particles, in particular gauge bosons, the precise measurement of the top couplings among the top quark and gauge bosons is important. in particular, the top quark is copiously produced at the lhc, the lhc experiments are the places to probe these couplings@xcite. at the lhc, top quarks are produced primarily via two independent mechanisms. the dominant production mechanism is the pair production processes @xmath1, @xmath2 and second is single top production via electroweak interactions involving the @xmath3 vertex. single top quarks at the lhc are produced in three different modes : the s - channel (the involved @xmath4 is time - like), the t - channel mode (the involved @xmath4 is space - like), and the @xmath0 production (the @xmath4 boson is real). despite single top has a smaller cross section than top pair production, it can play an important role in top quark physics at lhc because this channel has potential to allow a direct measurement of @xmath5 ckm matrix elements as well as its sensitivity to various new physics models. however sufficient integrated luminosity and improved method of analysis can help us achieve detection of single top events at the lhc. the first observable in single top study is the total cross section and measurement of any possible deviation from predicted value by the sm. therefore, it is worthwhile to investigate the effects of physics beyond sm on single top quark production. as it is well - known sm has been successfully predicted experimental measurements with a great precision. nevertheless, it is commonly accepted that it is a valid effective lagrangian which is applicable at low energies. in many beyond sm theories which have been studied to date, reduction to the sm at low energies proceeds via decoupling of heavy particle with masses of order @xmath6. there have been many attempts to study the sensitivity of the lhc observables to various effective operators @xcite. our goal in this paper is to study the effect of anomalous couplings of the top quark with gluon via @xmath0-channel of single top at the lhc. we will assume that new physics effects in @xmath0 single top production are induced by consideration of an effective lagrangian. here, we confine our studies to interaction of mass dimension @xmath7 after spontaneous symmetry breaking. the total statistical and systematic uncertainties in the measurement of cross section for this process at the lhc is about @xmath8 with an integrated luminosity of @xmath9@xcite. the top quark is the heaviest quark, therefore effect of new physics on its coupling are expected to be larger than for any other fermions and deviation with respect to the sm predictions might be detectable. the rest of this paper is organized as follows : in the next section, including the effective lagrangian for @xmath10 coupling, we calculate the analytical expression for the single @xmath0 top cross section production at the lhc. in section 3, we present the dependency of observables which we study at the lhc. then, we find the allowed regions in parameters space of our effective lagrangian and compare our results with the results obtained from observables in production of @xmath11 at lhc and edm of top quark. the conclusions are given in section 4.
Framework and analytical calculations
in this section, we introduce a model independent effective lagrangian for the vertex of @xmath10 and look for any possible deviation from sm prediction in production of @xmath12-channel at the lhc. in this approach, we assume that sm modified by an addition 5-dimensional lagrangian which include interaction of top pair and gluon and consider the effect of this lagrangian in production of @xmath0 single top production. coefficients of this lagrangian parameterize the low energy effects of the underlying high scale physics. as it is well - known, the gauge invariant effective lagrangian for the interactions between the top quark and gluons which include the cedm and cmdm form factors is given by : @xmath13 t g^{a}_{\nu}\end{aligned}\]] where @xmath14 are the @xmath15 color matrices and @xmath16 and @xmath17 are, respectively, the cmdm and cedm forms factors of the top quark. notice that a sizable non - zero cedm would be the signal of a new type of cp - violating interaction beyond ckm phase and can contribute to electric dipole moment (edm) of neutron. for this reason, experimental upper bound on edm of neutron constraint this coupling @xcite,@xcite. in the following, we use this constraint on parameters space of above lagrangian and compare these bounds with our result which arise from @xmath12 single top production at lhc. assuming @xmath18, where @xmath6 is the scale of new physics, the form factors can be approximated by @xmath19 where @xmath20 and @xmath21 are independent of @xmath22. the cmdm of the top quark is then given by @xmath23,while cedm is @xmath24. = 1.6 in as it is shown in fig. [twfig], @xmath10 effective lagrangian can contribute to the right diagram in @xmath0 of single top production at the lhc. we calculate the amplitude of @xmath25 process including cedm and cmdm effect. this amplitude is given by : @xmath26 \\ \nonumber\end{aligned}\]] where @xmath27 and @xmath28 are respectively weak and strong coupling constants and @xmath29, @xmath30 are the masses of top and w gauge boson and @xmath5 is ckm matrix element. the explicit forms for @xmath31 (@xmath32) in terms of mandelstam variables, @xmath21 and @xmath33 are given in eqs. ([f1]-[f6]) of the appendix.
Sensitivity analysis for anomalous @xmath10 couplings
in this section, we study the total cross section of @xmath0 single top production at the lhc and study the effect of cedm and cmdm couplings on it. recently, @xmath34 collaboration reported the measured value of cross section of @xmath0 single top production at the center - of - mass energy @xmath35 with an integrated luminosity of @xmath36@xcite : @xmath37.\label{exp}\end{aligned}\]] this measurement is in agreement with the sm expectation @xmath38 @xcite. the hadronic cross section for production of @xmath0 can be obtained by integrating over the parton level cross section convoluted with the parton distribution functions : @xmath39 where @xmath40 are the parton structure functions of proton. the parameters @xmath41 and @xmath42 are the parton momentum fractions and @xmath43 is the factorization scale. in this paper, we obtain the direct constraints on dipole operators including top quark in the above effective lagrangian approach. we consider the total cross section of @xmath0 single top production at the lhc and study the effect of cmdm and cedm coupling on it. for this study, we consider the relative change in cross section which is defined as : @xmath44 where @xmath45 is total cross section in the presence of cmdm and cedm couplings. the relative change in cross section of the single top production at the lhc are shown in fig. [deltacrossk] and [deltacrosskt]. in these figures, we consider that only cmdm or cedm coupling exists and display the relative change in cross section of @xmath46 versus @xmath21 and @xmath20. in fig. [deltacrossk]-a, we have set @xmath47. in fig. [deltacrossk]-b, as explained in the caption, we have set @xmath48 which might be measured in ongoing run of the lhc. in the center of mass of energy @xmath49, leading order sm cross section of @xmath50 have been obtained @xmath51 @xcite. to calculate @xmath46, we have used the cteq6.6 m @xcite, mstw2008 @xcite and alekhin2 @xcite as the parton structure functions (pdf). the green curve (dashed), the pink curve (line) and blue curve (dotted) are corresponding to cteq6.6 m, mstw2008 and alekhin2 structure functions, respectively. an interesting observation from fig. [deltacrossk] and [deltacrosskt] is that the correction to @xmath0-channel cross section due to cmdm and cedm is sensitive to the choice of parton distribution function, in particular at large values of @xmath21 and @xmath20. as it is seen in these figures, different structure functions change the value of @xmath46 more than @xmath52 for large values of @xmath21 and @xmath20. considering the cteq pdf, the presence of @xmath21 or @xmath20 can change total cross section more than @xmath53. at small value in the range of @xmath54 $], @xmath55/@xmath56 is almost robust against the choice of pdf. /@xmath56 versus @xmath21. in this figure, we have set @xmath47. the green curve (dashed), the pink curve (line) and blue curve (dotted) respectively correspond to cteq6.6 m @xcite, mstw2008 @xcite and alekhin2 @xcite structure functions. b) similar to fig. a except that @xmath48. the horizontal small dotted (yellow dark), violet and dot - dashed (green) lines respectively correspond to @xmath57, @xmath58 and @xmath59 uncertainties in the measurement of @xmath50.,title="fig:"]/@xmath56 versus @xmath21. in this figure, we have set @xmath47. the green curve (dashed), the pink curve (line) and blue curve (dotted) respectively correspond to cteq6.6 m @xcite, mstw2008 @xcite and alekhin2 @xcite structure functions. b) similar to fig. a except that @xmath48. the horizontal small dotted (yellow dark), violet and dot - dashed (green) lines respectively correspond to @xmath57, @xmath58 and @xmath59 uncertainties in the measurement of @xmath50.,title="fig : "] (a)(b) /@xmath56 versus @xmath20 for the center - of - mass energy of @xmath60 (left) and @xmath61 (right). input parameters are similar to fig. [deltacrossk].,title="fig:"]/@xmath56 versus @xmath20 for the center - of - mass energy of @xmath60 (left) and @xmath61 (right). input parameters are similar to fig. [deltacrossk].,title="fig : "] (a)(b) fig. [deltacrossk] demonstrates that the effect of the presence of cmdm can change the total cross section more than @xmath62 for @xmath63. the horizontal small dotted (yellow dark), violet and dot - dashed (green) lines respectively correspond to @xmath57, @xmath58 and @xmath59 uncertainty in the measurement of @xmath50. as it was mentioned, the total statistical and systematic uncertainties in the measurement of cross section for @xmath0 single top production at the lhc is about @xmath8 with an integrated luminosity of @xmath9 at center - of - mass energy @xmath64. with @xmath52 uncertainty in future measurement of the total cross section, from fig. [deltacrossk] ([deltacrosskt]), we can put upper bound on @xmath21 (@xmath20) down to @xmath65 (@xmath66). if in forthcoming run of lhc (in @xmath61), we measure the total cross section even with @xmath67 uncertainty, we can constrain @xmath21 (@xmath20) down to @xmath68 (@xmath69). as it is seen in fig. [deltacrosskt], in spite of cmdm (@xmath21), the cross section is symmetric with respect to @xmath20 because cedm coupling enters in the cross section in even powers when @xmath70. therefore, in this situation, the cross section is not a cp violating observable. another observation from fig. [deltacrossk] is that the effect of presence of cmdm (@xmath21), in spite of cedm (@xmath20), can be destructive. if the total uncertainties in the measurement of cross section is less than @xmath67, we can distinguish between cedm and cmdm effects. in figs. [scater], red area depicts ranges of parameters space in cmdm (@xmath21) and cedm (@xmath20) couplings plane for which prediction of effective lagrangian (equation [lag]) on @xmath0 single top production at lhc are consistent with experimental measurements. when performing such study, one should take into account constraints from other studies. there exist many direct and indirect constraints on dipole operator. presently, the most sensitive observable obtained from mercury and neutron edm. in @xcite, it is shown that the neutron edm constrains the top cedm to be @xmath71. moreover, the constraints from @xmath72 and the top electric dipole moment, provide weaker bound on @xmath20. furthermore, the cedm and cmdm couplings of the top quark directly affect on top pair production at hadron colliders @xcite. we have borrowed the results of lhc constraints which come from @xmath73 total cross section on cedm and cmdm of top from reference @xcite. these results have been shown in fig. [scater]. in this figure green line depicts neutron edm constraint on @xmath20. hatched cyan shaded area depicts the allowed region which is consistent with @xmath73 total cross section at lhc. yellow area shows allowed region which is consistent with spectrum measurement of @xmath74 at lhc. it is remarkable that the allowed regions of top pair production cross section and spectrum measurement of @xmath74 at the lhc overlap with allowed region of @xmath0 single top production at lhc. as it can be seen in this figure, @xmath0 single top constraints on @xmath21 and @xmath20 are stronger than direct constraints which come from @xmath73 total cross section and spectrum measurement of @xmath74 at lhc.
Concluding remarks
in this paper, based on the effective lagrangian approach, we modify the sm by the additional 5-dimensional operators which include the interaction of top quark with gluon and consider the effect of this lagrangian in production of @xmath0 single top. coefficients of this lagrangian are related to cedm and cmdm form factors. we consider the total cross section of @xmath0 single top production at the lhc and study the effect of cedm and cmdm couplings on it. we have found the allowed regions in parameters space of cedm and cmdm in such a way that the experimental measurement of @xmath46 is satisfied. we also investigate the effect of different pdfs on @xmath0 single top production at lhc as a function of @xmath21 and @xmath20. we have shown that deviation of the @xmath0-channel single top cross section from the sm value is significant. we consider constraints on cedm and cmdm couplings which come from @xmath73 total cross section and spectrum measurement of @xmath74 at the lhc and compare them with our results. we have shown that constraints on @xmath21 and @xmath20 which arise from @xmath0 single top production at lhc are comparable with the ones coming from @xmath73 production at the lhc. it is notable that with the current @xmath0 cross section precision measurement, tight bounds are obtained and therefore with more precision measurements in future even more stringent limits than @xmath73 cross section could be achieved.
Acknowledgement
we would like to thank h. kanpour for helping us in technical issues.
Appendix
here, we list the formulas of @xmath31 which have been applied in calculation of single top production cross section. notice that mass dimension of @xmath31 are not equal. @xmath75 @xmath76 @xmath77 @xmath78 @xmath79 @xmath80\end{aligned}\]] where @xmath81 and @xmath82 are mandelstam variables, @xmath29, @xmath30 are the masses of top and w gauge boson. the parameters @xmath20 and @xmath21 are, respectively cedm and cmdm couplings. f. -p. schilling, int. j. mod. a * 27 * (2012) 1230016 [arxiv:1206.4484 [hep - ex]]. w. buchmuller and d. wyler, nucl. b * 268 * (1986) 621 ; c. arzt, m. b. einhorn and j. wudka, nucl. b * 433 * (1995) 41 [hep - ph/9405214] ; g. j. gounaris, m. kuroda and f. m. renard, phys. d * 54 * (1996) 6861 [hep - ph/9606435] ; a. cordero - cid, m. a. perez, g. tavares - velasco and j. j. toscano, phys. d * 70 * (2004) 074003 [hep - ph/0407127] ; b. grzadkowski, z. hioki, k. ohkuma and j. wudka, nucl. b * 689 * (2004) 108 [hep - ph/0310159] ; b. grzadkowski, m. iskrzynski, m. misiak and j. rosiek, jhep * 1010 * (2010) 085 [arxiv:1008.4884 [hep - ph]] ; r. torre, arxiv:1005.4801 [hep - ph]. s. chatrchyan _ et al. _ [cms collaboration], phys. [arxiv:1209.3489 [hep - ex]]. e. o. iltan, phys. d * 65 * (2002) 073013 [hep - ph/0111038]. j. f. kamenik, m. papucci and a. weiler, phys. d * 85 * (2012) 071501 [arxiv:1107.3143 [hep - ph]]. n. kidonakis, phys. d * 82 * (2010) 054018 [arxiv:1005.4451 [hep - ph]]. d. atwood, a. kagan and t. g. rizzo, phys. d * 52 * (1995) 6264 [hep - ph/9407408] ; p. haberl, o. nachtmann and a. wilch, phys. d * 53 * (1996) 4875 [hep - ph/9505409] ; c. degrande, j. -m. gerard, c. grojean, f. maltoni and g. servant, jhep * 1103 * (2011) 125 [arxiv:1010.6304 [hep - ph]] z. hioki and k. ohkuma, eur. phys. j. c * 65 * (2010) 127 [arxiv:0910.3049 [hep - ph]] ; z. hioki and k. ohkuma, eur. j. c * 71 * (2011) 1535 [arxiv:1011.2655 [hep - ph]] ; s. s. biswal, s. d. rindani and p. sharma, top quark using its polarization in pair production at hadron colliders, arxiv:1211.4075 [hep - ph]. j. pumplin, d. r. stump, j. huston, h. l. lai, p. m. nadolsky and w. k. tung, jhep * 0207 * (2002) 012 [hep - ph/0201195]. a. d. martin, w. j. stirling, r. s. thorne and g. watt, eur. j. c * 63 * (2009) 189 [arxiv:0901.0002 [hep - ph]]. s. alekhin, j. blumlein, s. klein and s. moch, arxiv:0908.3128 [hep - ph]. | we study the effects of chromoelectric and chromomagnetic dipole moments (cedm and cmdm) on the production cross section of single top @xmath0-channel at the lhc based on the effective lagrangian approach.
we show that the impact of cedm and cmdm could be large. using the experimental measurement of the @xmath0-channel cross section,
constraints on cedm and cmdm are extracted.
these constraints are comparable with the ones obtained from the top pair analysis.
0.65 cm 0.5 cm * seyed yaser ayazi, hoda hesari and mojtaba mohammadi najafabadi*0.5 cm school of particles and accelerators, institute for research in fundamental sciences (ipm), p.o.
box 19395 - 5531, tehran, iran | 1307.1846 |
Introduction
the recent enormous improvement of our knowledge of the neutrino oscillation parameters suggests a detailed investigation of the current constraints on the neutrino mass matrix. most of these constraints depend on the assumed nature of neutrinos (dirac or majorana). the structure of this paper is as follows. after a brief general discussion of the light - neutrino mass matrix in section [section - neutrino_mass_matrix], we will investigate the implications of the currently available data on the majorana neutrino mass matrix in section [section - majorana]. in section [section - dirac] we will discuss constraints on the neutrino mass matrix in the dirac case. finally we will conclude in section [section - conclusions].
The light-neutrino mass matrix
in this paper we assume that there are exactly three light neutrino mass eigenstates with masses smaller than @xmath0, _ i.e. _ we assume that there are no light sterile neutrinos. by the term `` neutrino mass matrix '' we thus always mean the @xmath1 mass matrix of the three light neutrinos. if neutrinos are _ majorana particles _, we assume that there is a (possibly effective) mass term @xmath2 where @xmath3 is a complex symmetric @xmath1-matrix. such a mass term directly arises from the type - ii seesaw mechanism and can be effectively generated via the seesaw mechanisms of type i and iii. if neutrinos are dirac particles, _ i.e. _ if the total lepton number is conserved, we assume the existence of three right - handed neutrino fields @xmath4 leading to the mass term @xmath5 where @xmath6 is an arbitrary complex @xmath1-matrix. before we can discuss any constraints on the neutrino mass matrix, we have to specify a basis in flavour space. in models involving flavour symmetries, the chosen matrix representations of the flavour symmetry group specify the basis. since we will at this point not assume any flavour symmetries in the lepton sector, we are free to choose a basis. for simplicity we will always choose a basis in which the charged - lepton mass matrix is given by @xmath7
Constraints on the majorana neutrino mass matrix
in the basis specified by equation ([mldiag]) the majorana neutrino mass matrix has the form @xmath8 where @xmath9 is the lepton mixing matrix and the @xmath10 (@xmath11) are the masses of the three light neutrinos. as any unitary @xmath1-matrix, @xmath9 can be parameterized by six phases and three mixing angles. we will use the parameterization @xmath12 with @xmath13 the phases @xmath14 and @xmath15 are unphysical since they may be eliminated by a suitable redefinition of the charged - lepton fields. on the contrary, @xmath16 and @xmath17 are physical in the case of majorana neutrinos and are therefore referred to as the majorana phases. @xmath18 denotes the well - known unitary matrix @xmath19 where @xmath20 and @xmath21 are the sines and cosines of the three mixing angles, respectively. the phase @xmath22 is responsible for a possible cp violation in neutrino oscillations (also in the dirac case) and is therefore frequently referred to as the dirac cp phase. the fact that the neutrino masses are the _ singular values _ of @xmath3 allows to derive a generic upper bound on the absolute values @xmath23. from linear algebra it is known that the absolute value of an element of a matrix is smaller or equal its largest singular value. for the neutrino mass matrix this implies @xcite @xmath24 since this bound is valid for _ any _ matrix, it holds also for dirac neutrinos. the strongest bounds on the absolute neutrino mass scale come from cosmology, where the sum of the masses of the light neutrinos is usually constrained to be at most of the order @xmath25see _ e.g. _ the list of upper bounds in @xcite. from this we deduce the approximate upper bound @xmath26 leading to @xmath27 in @xcite also an analytical lower bound on the @xmath23 is provided. defining @xmath28 one can show that @xmath29 note that this lower bound is independent of the majorana phases @xmath16 and @xmath17. unlike the generic upper bound discussed before, the lower bound ([lowerbound]) is valid only for majorana neutrinos. numerically evaluating this lower bound using the results of the global fits of oscillation data of @xcite only for two matrix elements leads to non - trivial lower bounds. the lower bounds in units of ev for these matrix elements are given by @xcite : [cols="<,<,<,^,^,^ ",] all of these five correlations may be subsumed as `` if one matrix element is small, the other one must be large. '' an example for such a correlation plot can be found in figure [fig : m11-m33normal]. in the case of an inverted neutrino mass spectrum, there are no correlations manifest at the @xmath30-level. it is important to note that while at the @xmath30-level the correlation plots based on the global fits of @xcite and @xcite agree, this is not true at the @xmath31-level for further details see @xcite.
Constraints on the dirac neutrino mass matrix
in analogy to the majorana case, we will study the @xmath1 dirac neutrino mass matrix @xmath6 in the basis where the charged - lepton mass matrix is diagonal see equation ([mldiag]). in this basis @xmath6 takes the form @xmath32 where @xmath33 is a unitary @xmath1-matrix. @xmath33 can be eliminated by considering the matrix @xmath34 since all observables accessible by current experimental scrutiny are contained in @xmath35, all matrices @xmath6 leading to the same @xmath35 are _ indistinguishable _ from the experimental point of view. therefore, the nine parameters of @xmath36 have to be treated as _ free _ parameters. consequently, in stark contrast to the majorana case, in the dirac case the neutrino mass matrix has at least nine free parameters (even if the mixing matrix and the neutrino masses are known). this freedom of choosing @xmath33 has important consequences for the analysis of @xmath6. obviously it is much harder to put constraints on the elements of @xmath6 than in the majorana case. the freedom of choosing @xmath33 even allows to set several elements of @xmath6 to zero without changing the physical predictions. this directly follows from the fact that every matrix can be decomposed into a product of a unitary matrix and an upper triangular matrix. thus there is a choice of @xmath33 such that @xmath37 similarly, by multiplication of @xmath6 by one of the six @xmath1 permutation matrices generated by @xmath38 from the left, one can arbitrarily permute the rows of @xmath6 without changing any physical predictions @xcite. however, it is most important to note that the freedom of choosing @xmath33 holds only as long as we do not impose a symmetry in the lepton sector. namely, a flavour symmetry which acts non - trivially in the neutrino sector imposes constraints on the form of the neutrino mass matrix @xmath6, _ i.e. _ not only on the neutrino masses and @xmath9 but also on @xmath33. consequently, a choice of @xmath33, _ e.g. _ such that @xmath6 is upper triangular, will in general be incompatible with the flavour symmetry. nevertheless, if we want to set bounds on the elements of @xmath6 (without introducing flavour symmetries), we indeed have the freedom of arbitrarily choosing the matrix @xmath33. therefore, examining bounds and correlations of the elements of @xmath6 is much less elucidating than in the majorana case. that said, studies of @xmath6 such as the question for the allowed cases of texture zeros in @xmath6 are still of great interest. in the following we will shortly comment on the allowed cases of texture zeros in @xmath6 under the assumption that @xmath39 is diagonal. a detailed analysis has been done by hagedorn and rodejohann in @xcite, which provides a classification of all possible texture zeros in this framework. we repeated the analysis of @xcite of the allowed cases of five, four and three texture zeros are investigated, the result being that all these cases are allowed and do not show any relations among the observables.] in @xmath6 based on the global fit results of @xcite. our numerical results are in perfect agreement with the analysis of @xcite. however, there are some previously allowed cases of texture zeros which can be excluded due to the new data, namely precisely those which lead to a vanishing or too small value (@xmath40) of @xmath41, @xmath42, @xmath43, @xmath44, @xmath45, @xmath46@xmath47, @xmath48@xmath49, @xmath50 (inverted spectrum) and @xmath51 (inverted spectrum) in the notation of @xcite. consequently, all cases of five texture zeros in @xmath6 are now excluded, and among the cases of four texture zeros only @xmath50 (normal spectrum), @xmath51 (normal spectrum) as well as @xmath52@xmath53 remain valid.
Conclusions
in the case of majorana neutrinos, the absolute values of the elements of the light - neutrino mass matrix @xmath3 can be described by nine parameters, of which seven are constrained by experiments / observations. the by now very precise knowledge of the oscillation parameters therefore allows detailed studies of the elements of @xmath3, including their allowed ranges and their correlations. the situation is quite different in the case of dirac neutrinos, where the neutrino mass matrix @xmath6 is by far not uniquely determined, even if the neutrino masses and the mixing matrix are known. therefore putting bounds on the elements of @xmath6 is much harder than in the majorana case. nevertheless studies of @xmath6 are possible, for example the analysis of texture zeros in @xmath6. we reinvestigated the allowed texture zeros of @xmath6 in the basis where the charged - lepton mass matrix is diagonal. our results agree with the original analysis @xcite, the only difference being that by now we know that @xmath54, which excludes some previously viable types of texture zeros. * acknowledgments : * the author wants to thank the organizers for the invitation to the delightful and interesting conference. this work is supported by the austrian science fund (fwf), project no. p 24161-n16. | we analyse the mass matrix of the three light neutrinos in the basis where the charged - lepton mass matrix is diagonal and discuss constraints on its elements for the majorana and the dirac case. | 1310.4455 |
Introduction
multiple - input, multiple - output(mimo) wireless transmission systems have been intensively studied during the last decade. the alamouti code @xcite for two transmit antennas is a novel scheme for mimo transmission, which, due to its orthogonality properties, allows a low complexity maximum - likelihood (ml) decoder. this scheme led to the generalization of stbcs from orthogonal designs @xcite. such codes allow the transmitted symbols to be decoupled from one another and single - symbol ml decoding is achieved over _ quasi static _ rayleigh fading channels. even though these codes achieve the maximum diversity gain for a given number of transmit and receive antennas and for any arbitrary complex constellations, unfortunately, these codes are not @xmath2, where, by a @xmath2 code, we mean a code that transmits at a rate of @xmath3 complex symbols per channel use for an @xmath4 transmit antenna, @xmath5 receive antenna system. the golden code @xcite is a full - rate, full - diversity code and has a decoding complexity of the order of @xmath6 for arbitrary constellations of size @xmath7 the codes in @xcite and the trace - orthogonal cyclotomic code in @xcite also match the golden code. with reduction in the decoding complexity being the prime objective, two new full - rate, full - diversity codes have recently been discovered : the first code was independently discovered by hottinen, tirkkonen and wichman @xcite and by paredes, gershman and alkhansari @xcite, which we call the htw - pga code and the second, which we call the sezginer - sari code, was reported in @xcite by sezginer and sari. both these codes enable simplified decoding, achieving a complexity of the order of @xmath8. the first code is also shown to have the non - vanishing determinant property @xcite. however, these two codes have lesser coding gain compared to the golden code. a detailed discussion of these codes has been made in @xcite, wherein a comparison of the codeword error rate (cer) performance reveals that the golden code has the best performance. in this paper, we propose a new full - rate, full - diversity stbc for @xmath9 mimo transmission, which has low decoding complexity. the contributions of this paper may be summarized (see table [table1] also) as follows : * the proposed code has the same coding gain as that of the golden code (and hence of that in @xcite and the trace - orthonormal cyclotomic code) for any qam constellation (by a qam constellation we mean any finite subset of the integer lattice) and larger coding gain than those of the htw - pga code and the sezginer - sari code. * compared with the golden code and the codes in @xcite and @xcite, the proposed code has lesser decoding complexity for all complex constellations except for square qam constellations in which case the complexity is the same. compared to the htw - pga code and the sezginer - sari codes, the proposed code has the same decoding complexity for all non - rectangular qam [fig [fig2]] constellations. * the proposed code has the non - vanishing determinant property for qam constellations and hence is diversity - multiplexing gain (dmg) tradeoff optimal. the remaining content of the paper is organized as follows : in section [sec2], the system model and the code design criteria are reviewed along with some basic definitions. the proposed stbc is described in section [sec3] and its non - vanishing determinant property is shown in section [sec4]. in section [sec5] the ml decoding complexity of the proposed code is discussed and the scheme to decode it using sphere decoding is discussed in section [sec6]. in section [sec7], simulation results are presented to show the performance of the proposed code as well as to compare with few other known codes. concluding remarks constitute section [sec8]. _ notations : _ for a complex matrix @xmath10 the matrices @xmath11, @xmath12 and @xmath13 $] denote the transpose, hermitian and determinant of @xmath10 respectively. for a complex number @xmath14 @xmath15 and @xmath16 denote the real and imaginary part of @xmath14 respectively. also, @xmath17 represents @xmath18 and the set of all integers, all real and complex numbers are denoted by @xmath19 @xmath20 and @xmath21 respectively. the frobenius norm and the trace are denoted by @xmath22 and @xmath23 $] respectively. the columnwise stacking operation on @xmath24 is denoted by @xmath25 the kronecker product is denoted by @xmath26 and @xmath27 denotes the @xmath28 identity matrix. given a complex vector @xmath29^t,$] @xmath30 is defined as @xmath31^t\]] and for a complex number @xmath32, the @xmath33 operator is defined by @xmath34.\]] the @xmath33 operator can be extended to a complex @xmath35 matrix by applying it to all the entries of it.
Code design criteria
a finite set of complex matrices is a stbc. a @xmath35 linear stbc is obtained starting from an @xmath35 matrix consisting of arbitrary linear combinations of @xmath36 complex variables and their conjugates, and letting the variables take values from complex constellations. the rate of such a code is @xmath37 complex symbols per channel use. we consider rayleigh quasi - static flat fading mimo channel with full channel state information (csi) at the receiver but not at the transmitter. for @xmath9 mimo transmission, we have @xmath38 where @xmath39 is the codeword matrix, transmitted over 2 channel uses, @xmath40 is a complex white gaussian noise matrix with i.i.d entries, i.e., @xmath41 and @xmath42 is the channel matrix with the entries assumed to be i.i.d circularly symmetric gaussian random variables @xmath43. @xmath44 is the received matrix. [def1]@xmath45 if there are @xmath36 independent information symbols in the codeword which are transmitted over @xmath46 channel uses, then, for an @xmath47 mimo system, the code rate is defined as @xmath48 symbols per channel use. if @xmath49, where @xmath50, then the stbc is said to have @xmath51 @xmath52. considering ml decoding, the decoding metric that is to be minimized over all possible values of codewords @xmath53 is given by @xmath54 [def2]@xmath55 the ml decoding complexity is given by the minimum number of symbols that need to be jointly decoded in minimizing the decoding metric. this can never be greater than @xmath36, in which case, the decoding complexity is said to be of the order of @xmath56. if the decoding complexity is lesser than @xmath56, the code is said to admit simplified decoding. [def3]@xmath57 for any stbc @xmath53 that encodes @xmath36 information symbols, the @xmath58 matrix @xmath59 is defined by the following equation @xmath60 where @xmath61^t$] is the information symbol vector the code design criteria @xcite are : (i) @xmath62 @xmath63 to achieve maximum diversity, the codeword difference matrix @xmath64 must be full rank for all possible pairs of codewords and the diversity gain is given by @xmath65 (ii) @xmath66 @xmath63 for a full ranked stbc, the minimum determinant @xmath67, defined as @xmath68\]] should be maximized. the coding gain is given by @xmath69, with @xmath4 being the number of transmit antennas. for the @xmath9 mimo system, the target is to design a code that is full - rate, i.e transmits 2 complex symbols per channel use, has full - diversity, maximum coding gain and allows low ml decoding complexity.
the proposed stbc
in this section, we present our stbc for @xmath1 mimo system. the design is based on the class of codes called co - ordinate interleaved orthogonal designs (ciods), which was studied in @xcite in connection with the general class of single - symbol decodable codes and, specifically for 2 transmit antennas, is as follows. the ciod for @xmath0 transmit antennas @xcite is + @xmath70\]] where @xmath71 are the information symbols and @xmath72 and @xmath73 are the in - phase (real) and quadrature - phase (imaginary) components of @xmath74 respectively. notice that in order to make the above stbc full rank, the signal constellation @xmath75 from which the symbols @xmath76 are chosen should be such that the real part (imaginary part, resp.) of any signal point in @xmath75 is not equal to the real part (imaginary part, resp.) of any other signal point in @xmath75 @xcite. so if qam constellations are chosen, they have to be rotated. the optimum angle of rotation has been found in @xcite to be @xmath77 degrees and this maximizes the diversity and coding gain. we denote this angle by @xmath78 the proposed @xmath79 stbc @xmath80 is given by @xmath81 where * the four symbols @xmath82 and @xmath83, where @xmath75 is a @xmath84 degrees rotated version of a regular qam signal set, denoted by @xmath85 which is a finite subset of the integer lattice, and @xmath86 to be precise, @xmath87 * @xmath88 is a permutation matrix designed to make the stbc full rate and is given by @xmath89.$] * the choice of @xmath90 in the above expression should be such that the diversity and coding gain are maximized. a computer search was done for @xmath90 in the range @xmath91 $]. the optimum value of @xmath90 was found out to be @xmath92. explicitly, our code matrix is @xmath93 @xmath94 \\\]] the minimum determinant for our code when the symbols are chosen from qam constellations is @xmath95, the same as that of the golden code, which will be proved in the next section. the generator matrix for our stbc, corresponding to the symbols @xmath76, is as follows : @xmath96\]] it is easy to see that this generator matrix is orthonormal. in @xcite, it was shown that a necessary and sufficient condition for an stbc to be _ information lossless _ is that its generator matrix should be unitary. hence, our stbc has the _ information losslessness _ property.
Nvd property and the dmg optimality
in this section it is shown that the proposed code has the non - vanishing determinant (nvd) property @xcite, which in conjunction with full - rateness means that our code is dmg tradeoff optimal @xcite. the determinant of the codeword matrix @xmath53 can be written as @xmath97.\]] using @xmath98 and @xmath99 in the equation above, we get, @xmath100\\ & = & \big((s_1+s_2)+(s_1-s_2)^*\big)\big((s_1+s_2)-(s_1-s_2)^*\big) { } \nonumber\\ & & { } -j[\big((s_3+s_4)+(s_3-s_4)^*\big)\big((s_3+s_4)-(s_3-s_4)^*\big)].\end{aligned}\]] since @xmath101, with @xmath102, @xmath103, a subset of @xmath104 $], defining @xmath105, @xmath106, @xmath107 and @xmath108, with @xmath109 and @xmath110 $], we get @xmath111\\ & = & e^{j2\theta_g}a^2-e^{-j2\theta_g}b^2 - j[e^{j2\theta_g}c^2-e^{-j2\theta_g}d^2].\end{aligned}\]] since @xmath112, we get @xmath113 for the determinant of @xmath53 to be 0, we must have @xmath114 the above can be written as @xmath115 where @xmath116 and clearly @xmath117 $]. it has been shown in @xcite that holds only when @xmath118, i.e., only when @xmath119. this means that the determinant of the codeword difference matrix is 0 only when the codeword difference matrix is itself 0. so, for any distinct pair of codewords, the codeword difference matrix is always full rank for any constellation which is a subset of @xmath104 $]. also, the minimum value of the modulus of r.h.s of can be seen to be @xmath120. so, @xmath121. in particular, when the constellation chosen is the standard qam constellation, the difference between any two signal points is a multiple of 2. hence, for such constellations, @xmath122, where @xmath53 and @xmath123 are distinct codewords. the minimum determinant is consequently 16/5. this means that the proposed codes has the non - vanishing determinant (nvd) property @xcite. in @xcite, it was shown that full - rate codes which satisfy the non - vanishing determinant property achieve the optimal dmg tradeoff. so, our proposed stbc is dmg tradeoff optimal.
Simplified decoding using sphere decoder
in this section, it is shows that sphere decoding can be used to achieve the decoding complexity of @xmath8. it can be shown that can be written as @xmath147 where @xmath148 is given by @xmath149 with @xmath150 being the generator matrix as in and @xmath151^t\]] with @xmath152 drawn from @xmath75, which is a rotation of the regular qam constellation @xmath85. let @xmath153^t$] then, @xmath154 where @xmath155 is @xmath156 $] with @xmath157 being a rotation matrix and is defined as follows @xmath158.\]] so, can be written as @xmath159 where @xmath160. using this equivalent model, the ml decoding metric can be written as @xmath161 on obtaining the qr decomposition of @xmath162, we get @xmath163= @xmath164, where @xmath165 is an orthonormal matrix and @xmath166 is an upper triangular matrix. the ml decoding metric now can be written as @xmath167 if @xmath168 $], where @xmath169 are column vectors, then @xmath170 and @xmath171 have the general form obtained by @xmath172 process as shown below @xmath173\]] where @xmath174 are column vectors, and @xmath175\]] where @xmath176, @xmath177 @xmath178 @xmath179 it can be shown by direct computation that @xmath171 has the following structure @xmath180\]] where @xmath181 stands for a possibly non - zero entry. the structure of the matrix @xmath182 allows us to perform a 4 dimensional real sphere decoding (sd) @xcite to find the partial vector @xmath183^t$] and hence obtain the symbols @xmath124 and @xmath125. having found these, @xmath126 and @xmath127 can be decoded independently. observe that the real and imaginary parts of symbol @xmath126 are entangled with one another because of constellation rotation but are independent of the real and imaginary parts of @xmath127 when @xmath124 and @xmath125 are conditionally given. having found the partial vector @xmath184^t$], we proceed to find the rest of the symbols as follows. we do two parallel 2 dimensional real search to decode the symbols @xmath126 and @xmath127. so, overall, the worst case decoding complexity of the proposed stbc is 2@xmath8. this is due to the fact that 1. a 4 dimensional real sd requires @xmath185 metric computations in the worst possible case. two parallel 2 dimensional real sd require @xmath186 metric computations in the worst case. this decoding complexity is the same as that achieved by the htw - pga code and the sezginer - sari code. though it has not been mentioned anywhere to the best of our knowledge, the ml decoding complexity of the golden code, dayal - varanasi code and the trace - orthogonal cyclotomic code is also @xmath146 for square qam constellations. this follows from the structure of the @xmath171 matrices for these codes which are counterparts of the one in. the @xmath171 matrices of these codes are similar in structure and as shown below : @xmath187\]] table [table1] presents the comparison of the known full - rate, full - diversity @xmath1 codes in terms of their ml decoding complexity and the coding gain.
Simulation results
fig [4qam] shows the codeword error performance plots for the golden code, the proposed stbc and the htw - pga code for the 4-qam constellation. the performance of the proposed code is the same as that of the golden code. the htw - pga code performs slightly worse due to its lower coding gain. fig [16qam], which is a plot of the cer performance for 16-qam, also highlights these aspects. table [table1] gives a comparison between the well known full - rate, full - diversity codes for @xmath9 mimo. [cols="^,^,^,^,^ ",]
Concluding remarks
in this paper, we have presented a full - rate stbc for @xmath9 mimo systems which matches the best known codes for such systems in terms of error performance, while at the same time, enjoys simplified - decoding complexity that the codes presented in @xcite and @xcite do. recently, a rate-1 stbc, based on scaled repetition and rotation of the alamouti code, was proposed @xcite. this code was shown to have a hard - decision performance which was only slightly worse than that of the golden code for a spectral efficiency of @xmath188, but the complexity was significantly lower.
Acknowledgement
this work was partly supported by the drdo - iisc program on advanced research in mathematical engineering. j. c. belfiore, g. rekaya and e. viterbo, `` the golden code : a @xmath9 full rate space - time code with non - vanishing determinants, '' _ ieee trans. inf. theory _, 51, no. 4, pp. 1432 - 1436, april 2005. j. paredes, a.b. gershman and m. gharavi - alkhansari, a @xmath9 space - time code with non - vanishing determinants and fast maximum likelihood decoding, " in proc _ ieee international conference on acoustics, speech and signal processing(icassp 2007), _ vol. 2, pp.877 - 880, april 2007. s. sezginer and h. sari, `` a full rate full - diversity @xmath9 space - time code for mobile wimax systems, '' in proc. _ ieee international conference on signal processing and communications _, dubai, july 2007. p. elia, k. r. kumar, s. a. pawar, p. v. kumar and h. lu, explicit construction of space - time block codes : achieving the diversity - multiplexing gain tradeoff , _ ieee trans. inf. theory _, 52, pp. 3869 - 3884, sept. 2006. v.tarokh, n.seshadri and a.r calderbank,"space time codes for high date rate wireless communication : performance criterion and code construction , _ ieee trans. inf. theory _, 744 - 765, 1998. h. yao and g. w. wornell, `` achieving the full mimo diversity - multiplexing frontier with rotation - based space - time codes, '' in _ proc. allerton conf. on comm. control and comput., _ monticello, il, oct. 2003. | this paper presents a low - ml - decoding - complexity, full - rate, full - diversity space - time block code (stbc) for a @xmath0 transmit antenna, @xmath0 receive antenna multiple - input multiple - output (mimo) system, with coding gain equal to that of the best and well known golden code for any qam constellation.
recently, two codes have been proposed (by paredes, gershman and alkhansari and by sezginer and sari), which enjoy a lower decoding complexity relative to the golden code, but have lesser coding gain.
the @xmath1 stbc presented in this paper has lesser decoding complexity for non - square qam constellations, compared with that of the golden code, while having the same decoding complexity for square qam constellations.
compared with the paredes - gershman - alkhansari and sezginer - sari codes, the proposed code has the same decoding complexity for non - rectangular qam constellations.
simulation results, which compare the codeword error rate (cer) performance, are presented. | 0806.1819 |
Introduction
the theme of @xmath0 (parity time) symmetric systems was initiated in the works of bender and collaborators @xcite as an alternative to the standard quantum theory, where the hamiltonian is postulated to be hermitian. the principal conclusion of these works was that @xmath0-invariant hamiltonians, which are not necessarily hermitian, may still give rise to completely real spectra, thus being appropriate for the description of physical settings. in terms of the schrdinger - type hamiltonians, which include the usual kinetic - energy operator and the potential term, @xmath3, the @xmath0-invariance admits complex potentials, subject to constraint that @xmath4. recent developments in optics have resulted in an experimental realization of the originally theoretical concept of the @xmath0-symmetric hamiltonians, chiefly due to the work by christodoulides and co - workers christo1 (see also @xcite). it has been demonstrated that the controllable imposition of symmetrically set and globally balanced gain and loss may render optical waveguiding arrays a fertile territory for the construction of @xmath0-symmetric complex potentials. the first two such realizations made use of couplers composed of two waveguides with and without loss @xcite (so - called passive @xmath5couplers), or, in more standard " form, a pair of coupled waveguides, one carrying gain and the other one loss @xcite. in fact, more general models of linearly coupled active (gain - carrying) and passive (lossy) intrinsically nonlinear waveguides, without imposing the condition of the gain - loss balance, were considered earlier, and stable solitons were found in them @xcite, including exact solutions @xcite (see also a brief review in ref. recently, an electronic analog of such settings has also been implemented @xcite. configurations with a hidden @xmath6 symmetry have been identified also in fine - tuned parameter regions of microwave billiards @xcite. effects of the nonlinearity in a gross - pitaevski equation on the @xmath7 properties of a bose - einstein condensate have been analyzed in @xcite. the possibility to engineer @xmath0-symmetric _ oligomers _ (coupled complexes of a few loss - and gain - carrying elements) @xcite, which may include nonlinearity, was an incentive to a broad array of additional studies on both the few - site systems and entire @xmath0-symmetric lattices @xcite. more recently, nonlinear @xmath0-symmetric systems, incorporating @xmath0-balanced nonlinear terms, have drawn considerable interest too @xcite-@xcite. most of the @xmath0-invariant systems considered thus far have been one - dimensional (1d) in their nature, although the stability of solitons in 2d periodic @xmath0-symmetric potentials has also been recently investigated @xcite. actually, 2d arrays of optical waveguides can be readily built @xcite (the same is true about other quasi - discrete systems, including electrical ones), hence, a natural question is whether @xmath0-symmetric oligomers (and ultimately lattices built of such building blocks) can be created in a 2d form. this work aims to make a basic step in this direction, by introducing fundamental 2d plaquettes consisting, typically, of four sites (in one case, it will be a five - site cross). these configurations, illustrated by fig. [modes], are inspired by earlier works on 2d hamiltonian lattices described by discrete nonlinear schrdinger equations @xcite, where diverse classes of modes, including discrete solitary vortices malomed, pelin, have been predicted and experimentally observed moti, yuri. the plaquettes proposed herein should be straightforwardly accessible with current experimental techniques in nonlinear optics, as a straightforward generalization of the coupler - based setting reported in ref. we start from the well - established hamiltonian form of such plaquettes in the conservative form, gradually turning on the gain - loss parameter (@xmath2), as the strength of the @xmath0-invariant terms, to examine stationary states supported by the plaquettes, studying their stability against small perturbations and verifying the results through direct simulations. actually, in this work we focus on those (quite diverse, although, obviously, not most generic) modes that can be found in an analytical form, while their stability is studied by means of numerical methods. the analytical calculations and the manifestations of interesting features, such as a potential persistence past the critical point of the linear @xmath0 symmetry, are enabled by the enhanced symmetry of the modes that we consider below. it is conceivable that additional asymmetric modes may exist too within these 2d configurations. our principal motivation for studying the above systems stems from the fact that realizations of @xmath0-symmetry e.g. within the realm of nonlinear optics will be inherently endowed with nonlinearity. hence, it is only natural to inquire about the interplay of the above type of linear systems with the presence of nonlinear effects. in addition to this physical argument, there exists an intriguing mathematical one which concerns the existence, stability and dynamical fate of the nonlinear states in the presence of @xmath0-symmetric perturbations. in particular, previous works @xcite point to the direction that neither the existence, nor the stability of @xmath0-symmetric nonlinear states mirrors that of their linear counterparts (or respects the phase transition of the latter generically). the presentation of our results is structured as follows. section [symmetry] contains a part of the analytical results, including a detailed analysis of the @xmath8symmetry properties of the nonlinear schrdinger type model, as well as the spectral properties of the linear hamiltonian subsystems. section [numerics] is devoted to the existence, stability and dynamics of stationary modes in the nonlinear systems. beside analytical results, it contains a detailed presentation of the numerical findings. in section [conclu] we summarize conclusions and discuss directions for future studies.
The setup and symmetry properties[symmetry]
the dynamics of the 2d plaquettes that we are going to consider is described by a multicomponent nonlinear schrdinger equation (nlse) @xmath9built over a transposition - symmetric linear @xmath10 hermitian matrix hamiltonian @xmath11 and an additional nonlinear @xmath10 matrix operator, @xmath12. to understand the symmetry properties of this nlse, we first analyze the associated linear problem @xmath13and check then whether the symmetry is preserved by the nonlinear term, @xmath14. the analysis can be built, in a part, on techniques developed for other nonlinear dynamical systems with symmetry preservation nonlin-1,nonlin-2,nonlin-3,nonlin-4,nonlin-5,nonlin-6,nonlin-7,nonlin-8. for the present setups, the time reversal operation @xmath15 can be defined as the combined action of a scalar - type complex conjugation @xmath16, @xmath17, and the sign change of time, @xmath18, in full accordance with wigner s original work which introduced these concepts @xcite. for the linear schrdinger equation ([a2]) and its solutions @xmath19this implies @xmath20where the overbar denotes complex conjugation. from the actual form of the gain - loss arrangements in the 2d plaquettes we can conjecture the existence of certain plaquette - dependent parity operators @xmath21, with @xmath22, which will render the hamiltonians @xmath8symmetric, @xmath23=0 $]. to find an explicit representation of these parity operators @xmath21, we use the following ansatz, @xmath24=0 \label{a2 - 3}\]]together with the pseudo - hermiticity condition @xmath25the latter follows trivially from eq. ([a2 - 3]), @xmath23=0 $], and the transposition symmetry, @xmath11. these parity operators @xmath21 will be used to check whether the corresponding nonlinear terms @xmath26 satisfy the same @xmath27symmetry. in contrast to linear setups with the @xmath28symmetry being either exact (@xmath29=0 $], @xmath30) or spontaneously broken (@xmath29=0 $], @xmath31), the nonlinear setups considered in the present paper allow for sectors of exact @xmath28symmetry (@xmath32=0 $], @xmath30) and of broken @xmath28symmetry (@xmath31@xmath33@xmath34\neq 0 $]), as it is common for nonlinear @xmath6-symmetric systems @xcite-@xcite. we start from the 2d plaquette of 0 + 0- type depicted as configuration (a) in fig. [modes]. this plaquette has only two (diagonally opposite) nodes carrying the gain and loss, while the other two nodes bear no such effects. the corresponding dynamical equations for the amplitudes at the four sites of this oligomer are @xmath35where @xmath36 is the above - mentioned gain - loss coefficient, and @xmath37 is a real coupling constant. the nonlinearity coefficients are scaled to be @xmath38 (we use time @xmath39 as the evolution variable, although in the mathematically equivalent propagation equations for optical waveguides @xmath39 has to be identified with the propagation distance, @xmath40). denoting @xmath41, the matrices @xmath42 and @xmath26 in ([a1]) take the form of @xmath43to find the parity matrix @xmath21 which renders the linear hamiltonian @xmath8symmetric, @xmath23=0 $], we use the pseudo - hermiticity condition ([a2 - 4]) and notice that @xmath44obviously, the following relations should hold : @xmath45=0 \nonumber \\ \mathcal{p}h_{l,1}\mathcal{p}=-h_{l,1 } & \qquad \longrightarrow \qquad & \left\ { \mathcal{p},h_{l,1}\right\ } = 0. \label{a6}\end{aligned}\]]the first of these conditions together with @xmath22, @xmath46 reduces the possible form of the parity transformation to one of the three types, @xmath47where @xmath48 are the usual pauli matrices. taking into account that @xmath49, the anti - commutativity condition in eqs. ([a6]) singles out the only possible parity matrix : @xmath50for the linear transformation, i.e., the matrix interchanging @xmath51 and @xmath52, as well as @xmath53 and @xmath54. one then immediately checks that @xmath55hence, in contrast to the linear component @xmath42, the nonlinear terms @xmath26 corresponding to eqs. ([zpzm1]) are not @xmath8symmetric, in the usual matrix sense. rather, the symmetry properties of the nonlinear terms have to be considered in the context of the nonlinear schrdinger equation itself. acting with @xmath56 on eq. ([a1]) we observe that @xmath57, \notag \\ i\partial _ { t}(\mathcal{p}\mathbf{t}\mathbf{u }) & = & h_{l}\mathcal{p}\mathbf{t}\mathbf{u}+h_{nl}(\mathcal{p}\mathbf{t}\mathbf{u})(\mathcal{p}\mathbf{t}\mathbf{u})\label{a13}.\end{aligned}\]] hence, the full nlse system ([zpzm1]) remains invariant if we define the @xmath58 transformation of the vectorial wave function obeying this system as follows : @xmath59 this is in full analogy to the condition of exact @xmath27symmetry for the corresponding linear schrdinger equation, nonlinearity matrices @xmath60 of more general type than that in ([nl-1]) and ([nl-2]) may be envisioned which may produce @xmath61symmetric solutions @xmath62 with less simple time dependence. a detailed analysis of such systems will be presented elsewhere.]. but the condition of _ spontaneously broken _ @xmath28symmetry (@xmath29=0 $], @xmath31) is replaced by the condition of _ completely broken _ @xmath28symmetry (@xmath31@xmath63@xmath64\neq 0 $]). in contrast to the present 2d plaquettes, which are mainly motivated by feasible experimental realizations, one can envision more sophisticated setups with @xmath65. this will lead to a new type of partial (or intermediate) @xmath28symmetry (to be considered elsewhere), which for solutions @xmath66 with broken @xmath28symmetry will keep the nonlinear term @xmath26 explicitly @xmath8symmetric (@xmath67pseudo - hermitian) in the matrix sense, but not @xmath28symmetric (under inclusion of the explicit time reversal @xmath18) in the sense of the nlse system. for configurations (b) and (c) in fig. [modes], @xmath68 and @xmath26 are still given by eqs. ([a3]) and ([a5]), but with @xmath69respectively. hence, relations ([a6]) are valid for both configurations (b) and (c) as well. using ([a7]) in @xmath70, we find a richer variety of parity operators @xmath21 than for configuration (a). configuration (b) allows for @xmath71whereas configuration (c) may be associated with @xmath72for configuration (d), we have @xmath73and simple computer algebra gives again two possible parity operators : @xmath74 in strong structural analogy to configuration (b). one verifies that @xmath75 holds also for configurations (b), (c) and (d), hence all 2d plaquettes considered in the present paper are not @xmath28symmetric in the usual matrix sense. next, we turn to the eigenvalue problems of the linear setups associated with plaquettes (a) - (d), i.e., to solutions of the equation @xmath76from the corresponding characteristic polynomials, @xmath77, @xmath78 = 0, \notag \\ (\mathrm{b}):\qquad \qquad & (e^{2}+\gamma ^{2})\left [e^{2}-(4k^{2}-\gamma ^{2})\right] = 0, \notag \\ (\mathrm{c}):\qquad \qquad & e^{4}-2(2k^{2}-\gamma ^{2})e^{2}+\gamma ^{4}=0, \notag \\ (\mathrm{d}):\qquad \qquad & e(e^{2}+\gamma ^{2})\left [e^{2}-(4k^{2}-\gamma ^{2})\right] = 0,\end{aligned}\]]we find @xmath79obviously, the matrix hamiltonians @xmath42 for plaquettes (a) and (d) are not of full rank. for plaquette (a) we find @xmath80, and @xmath42 has a two - dimensional kernel space, @xmath81. for plaquette (d) we find @xmath82 and @xmath83, where @xmath84. moreover, we see that the spectrum for plaquette (d), up to the additional eigenvalue @xmath85, coincides with that for (b). the different eigenvalues of the 4-node plaquettes displayed in eqs. ([a15 - 2]) show that these plaquettes are also physically not equivalent. equivalence classes of nonlinear 4-node plaquettes with isospectral linear hamiltonians @xmath86 but different pairwise couplings have been considered, e.g., in @xcite. for plaquettes (a), (b) and (d) an exceptional point (ep) occurs at @xmath87, being associated with a branching of the eigenvalue pair @xmath88 @xmath89 in the case of plaquette (a), all four eigenvalues are involved in the branching at @xmath90, where @xmath91. via jordan decomposition (e.g., with the help of the corresponding linear algebra tool of mathematica) we find that @xmath92i.e., a spectral degeneration of the type @xmath93 in arnold s notation @xcite, or, in other words, a third - order ep with a single decoupled fourth mode. hence, plaquettes of type (a) may serve as an easily implementable testground for the experimental investigation of third - order eps (see e.g. @xcite). for plaquettes (b) and (d) we have second - order eps at @xmath87, similar as for plaquette (c) where a pair of second - order eps occurs at @xmath94 with @xmath95, @xmath96. from the eigenvalues in ([a15 - 2]) we read off the @xmath27symmetry content of the four types of plaquettes. the sector of exact @xmath28symmetry (i.e., the sector with all eigenvalues purely real, @xmath97) corresponds to @xmath98i.e., for plaquettes (b) and (d) the @xmath28symmetry is spontaneously broken as soon as the gain - loss coupling is switched on, namely for @xmath99.
Existence, stability and dynamics of nonlinear statesnumerics
in this section, we seek stationary solutions of the type @xmath100constructed over constant vectors @xmath101. according to ([a13]) and ([a13 - 2]) such solutions will be @xmath61symmetric provided it holds @xmath102 for some @xmath103. we will test these symmetry properties for the solutions to be obtained. we note that restricting the explicit analysis to stationary solutions of the type ([e]) we by construction exclude from this analysis @xmath61violating solutions with @xmath104 which are necessarily non - stationary. a useful technical tool to facilitate the explicit derivation of stationary solutions @xmath105 are conservation equations of the type @xmath106constructed from eq. ([a1]) and its adjoint, where @xmath107 denotes an arbitrary constant matrix. the most simplest of them can be found via eqs. ([a2 - 4]), ([a3]) and eq. ([a5]) to be @xmath108 \mathbf{u}. \label{a18 - 2}\end{aligned}\]]for stationary equations @xmath109 the time - dependent phase factors @xmath110 cancel so that the left - hand - sides of these relations vanish, yielding simple algebraic constraints on the right - hand - sides. from eq. ([a18 - 2]) we see that for stationary solutions the @xmath6 inner product quantum mechanics (ptqm) the @xmath7 inner product was introduced first by znojil in @xcite in 2001. immediately afterwards, it was interpreted by japaridze as indefinite inner product @xcite in a krein space and generalized by mostafazadeh to the @xmath111metric in the context of pseudo - hermitian hamiltonians @xcite. finally, it was used by bender, brody and jones in 2002 to construct the positive definite @xmath112 inner product @xcite. for oligomer settings (of plaquettes or other few site configurations), it can be employed, e.g., to derive a simple algebraic constraint or as a criterion of the numerical accuracy of the evolutionary dynamics (especially since the solutions rapidly acquire very large amplitudes when unstable, as will be seen below). it also turned out useful in @xcite.] will remain conserved (@xmath113const) regardless of the violated @xmath114pseudo - hermiticity, @xmath115, characteristic for of our specific nonlinear plaquette couplings (see eq. ([nl-2])). subsequently, we first derive classes of stationary solutions @xmath105 explicitly. then, we analyze the stability of small perturbations over these stationary solutions by the linearization, via ansatz @xmath116 + o(\delta ^{2}),\qquad |\delta |\ll 1,\]]where @xmath117 is the small amplitude of the perturbation. exponents @xmath118 can be defined as wick - rotated eigenvalues from the corresponding @xmath119 perturbation matrix @xmath120 (see, e.g., @xcite for more details) : @xmath121 where @xmath122\mathbf{u}\]]characterizes the stationary problem, and the elements of the matrix @xmath120 are evaluated at @xmath123. linear stability is ensured for @xmath124, whereas @xmath125 corresponds to growing and decaying modes, i.e., exponential instabilities. substituting ansatz ([e]) for the stationary solutions in eqs. ([a1]), ([a18 - 1]) and ([a18 - 2]) we obtain the following algebraic equations : @xmath126@xmath127and @xmath128 \mathbf{u, } \notag \label{a31 } \\ 0 & = & \left (|a|^{2}-|c|^{2}\right) (\bar{a}c-\bar{c}a)+\left ( equations can be analyzed via the madelung substitution (i.e., via amplitude - phase decomposition), @xmath129without loss of generality, we may fix @xmath130. for arbitrary phase factors in ([aa]), eqs. ([a30]) and ([a31]) are satisfied by @xmath131 and @xmath132. using this condition in eq. ([zpzm2]) and dividing each equation ([zpzm2]) by the phase factor on its left - hand side, one obtains the imaginary parts of the resulting equations : @xmath133 = 2kb\sin \left (\frac{\phi _ { b}+\phi _ { d}}{2}-\phi _ { a}\right) \cos \left (\frac{\phi _ { b}-\phi _ { d}}{2}\right), \notag \\ \gamma b & = & ka\left [\sin (\phi _ { a}-\phi _ { b})+\sin (\phi _ { c}-\phi _ { b})\right] = 2ka\sin \left (\frac{\phi _ { a}+\phi _ { c}}{2}-\phi _ { b}\right) \cos \left (\frac{\phi _ { a}-\phi _ { c}}{2}\right), \notag \\ 0 & = & kb\left [\sin (\phi _ { b}-\phi _ { c})+\sin (\phi _ { d}-\phi _ { c})\right] = 2kb\sin \left (\frac{\phi _ { b}+\phi _ { d}}{2}-\phi _ { c}\right) \cos \left (\frac{\phi _ { b}-\phi _ { d}}{2}\right), \notag \\ -\gamma b & = & ka\left [\sin (\phi _ { a}-\phi _ { d})+\sin (\phi _ { c}-\phi _ { d})\right] = 2ka\sin \left (\frac{\phi _ { a}+\phi _ { c}}{2}-\phi _ { d}\right) \cos \left (\frac{\phi _ { a}-\phi _ { c}}{2}\right). \notag \\ & & \label{zpzm_add}\end{aligned}\]]for @xmath130 the first of these equations implies @xmath134, hence either @xmath135 (case 1) or @xmath136 (case 2). in case 1, we conclude from the third equation that either @xmath137 and @xmath138 (case 1a), or @xmath139 and @xmath140 is arbitrary (case 1b). in case 2 the third equation is satisfied automatically. in all the three cases, the second and the fourth equation are compatible. they give @xmath141returning to the phase - factor divided equations ([zpzm2]) and considering their real parts, we find @xmath142 + a^{3 }, \notag \label{a33 } \\ eb & = & ka\left [\cos (\phi _ { a}-\phi _ { b})+\cos (\phi _ { c}-\phi _ { b})\right] + b^{3 }, \notag \\ ea & = & kb\left [\cos (\phi _ { b}-\phi _ { c})+\cos (\phi _ { d}-\phi _ { c})\right] + a^{3 }, \notag \\ eb & = & ka\left [\cos (\phi _ { a}-\phi _ { d})+\cos (\phi _ { c}-\phi _ { d})\right] + b^{3}.\end{aligned}\]]the pairwise compatibility of the first and third, as well as of the second and fourth equations requires @xmath143for case 1a, these conditions are trivially satisfied, whereas for the remaining cases they lead to further restrictions : @xmath144 in this way the phase angles are fixed for all the three cases and we can turn to the amplitudes. the corresponding equation sets reduce to @xmath145 in the latter two cases (1b and 2) the amplitudes and phases completely decouple and we have @xmath146case 1a allows for a richer behavior. equating the terms @xmath147 in the upper two equations ([a36]) leads to the constraint @xmath148which can be resolved by @xmath149 (case 1aa) as well as by @xmath150 (case 1ab). the analysis of these two cases can be completed with the help of the relation @xmath151 from eq. ([a32]). as result we obtain the following set of stationary solutions : @xmath152 from eq. ([a31]), it can also be seen that either @xmath131 or if @xmath153, then @xmath154 must be true. here, we use the information available so far to check the @xmath28symmetry content of the solutions ([a39 - 1aa]) - ([a39 - 2]) explicitly. for stationary solutions @xmath155, @xmath156 the @xmath28symmetry condition ([a13 - 2]) implies @xmath157taking into account that @xmath16 acts as complex conjugation, we see from the explicit structure of @xmath158 in eq. (a8) that a stationary solution is @xmath28symmetric if, with @xmath130, it has @xmath138 and @xmath159 (up to a common phase shift). additionally, the amplitudes have to coincide pairwise : @xmath131, @xmath132. for eqs. ([a39 - 1aa]) - ([a39 - 2]) this means that all case-1 stationary solutions with @xmath138 are @xmath28symmetric in their present form. the case-2 mode becomes explicitly @xmath8symmetric after a global @xmath160 multiplication by a phase factor : @xmath161 ^{t }, \notag \\ \mathcal{p}\mathcal{t}\mathbf{v}_{0 } & = & \mathbf{v}_{0},\end{aligned}\]]where @xmath162 has to be chosen in eq. ([a39 - 2]). we note that this procedure is effectively equivalent to a redefinition of the original phase constraint : @xmath163 at the very beginning of the calculations in eq. ([aa]). the linear stability analysis was performed numerically. subsequently we present corresponding graphical results. the plaquettes (b) - (d) can be analyzed in a similar way. for brevity s sake, in fig. [figzpzm1] we present only the basic numerical results, by means of the following symbols : [figzpzm1] * case 1aa with @xmath164 blue circles ; * case 1aa with @xmath165 red crosses ; * case 1ab green stars ; * case 2 black squares ; * case 1b is not depicted explicitly because it corresponds to point configurations without gain - loss (@xmath166) and to exceptional point configurations @xmath90. figure [figzpzm1] presents the mode branches (their amplitudes, phases, and also their stability) over the gain - loss parameter @xmath2, starting from the conservative system at @xmath166. the same symbols are used in fig. [figzpzm2], which displays typical examples of the spectral plane @xmath167 for stability eigenvalues @xmath168 of the linearization ; recall that the modes are unstable if they give rise to @xmath169. explicitly we observe the following behavior. * case 1aa with @xmath164 blue circlesaccording to fig. [figzpzm2], the present solution is stable. notice that, although featuring a phase profile, it can not be characterized as a vortex state (the same is true for some other configurations carrying phase structure). interestingly, the relevant configuration is generically stable bearing two imaginary pairs of eigenvalues. * case 1aa with @xmath165 red crosses.obviously, this kind of solutions as well as the previous one exist up to the exceptional point @xmath90 of the @xmath0-symmetry breaking in the linear system, where the two branches collide and disappear (leave the stationary regime and become nonstationary). as seen in fig. figzpzm2, the present branch has two eigenvalue pairs which are purely imaginary for small @xmath2, but become real (rendering the configuration unstable) at @xmath170 and then @xmath171, respectively. ultimately, these pairs of unstable eigenvalues collide at the origin of the spectral plane with those of the previous branch (blue circles). * case 1ab green stars.this stationary solution has a number of interesting features. firstly, it is the only one among the considered branches which has two unequal amplitudes. secondly, it exists past the critical point @xmath90 of the linear system, due to the effect of the nonlinearity (the extension of the existence region for nonlinear modes was earlier found in 1d couplers @xcite and oligomers @xcite). furthermore, this branch has three non - zero pairs of stability eigenvalues, two of which form a quartet for small values of the gain - loss parameter, while the third is imaginary (i.e., the configuration is unstable due to the real parts of the eigenvalues within the quartet). at @xmath172, the eigenvalues of the complex quartet collapse into two imaginary pairs, rendering the configuration stable, in a narrow parametric interval. at @xmath173, the former imaginary pair becomes real, destabilizing the state again, while subsequent bifurcations of imaginary pairs into real ones occur at @xmath174 and @xmath175 (at the latter point, all three non - zero pairs are real). shortly thereafter, two of these pairs collide at @xmath176 and rearrange into a complex quartet, which exists along with the real pair past that point. * case 2 black squares. in contrast to all other branches, this one is _ always _ unstable. one of the two nonzero eigenvalue pairs is always real (while the other is always imaginary), as seen in fig. this branch also terminates at the exceptional point @xmath90, as relation @xmath177 can not hold at @xmath178. this branch collides with the two previous ones via a very degenerate bifurcation (that could be dubbed a double saddle - center " bifurcation), which involves 3 branches instead of two as in the case of the generic saddle - center bifurcation, and two distinct eigenvalue pairs colliding at the origin of the spectral plane. by means of direct simulations, we have also examined the dynamics of the modes belonging to different branches in fig. [stabzpzm]. the stable blue - circle branch demonstrates only oscillations under perturbations. this implies that, despite the presence of the gain - loss profile, none of the perturbation eigenmodes grows in this case. nevertheless, the three other branches ultimately manifest their dynamical instability, which is observed through the growth of the amplitude at the gain - carrying site [b, in fig. [modes](a)] at the expense of the lossy site (d). that is, the amplitude of the solution at the site with the gain grows, while the amplitude of the solution at the dissipation site loses all of its initial power. depending on the particular solution, passive sites (the ones without gain or loss, such as a and c) may be effectively driven by the gain (as in the case of the black - square - branch, where the site a is eventually amplified due to the growth of the amplitude at site b) or by the loss (red - cross and green - star branches, where, eventually, the amplitudes at both a and c sites lose all of their optical power). we now turn to the generalized (not exactly @xmath56-symmetric) configurationsymmetry, spontaneously broken @xmath61symmetry and completely broken @xmath61symmetry see the discussion of eqs. ([a13]) and ([a13 - 2]).] featuring the alternation of the gain and loss along the plaquette in panel (b) of fig. [modes]. indeed, the absence of @xmath0-symmetry in this case is mirrored in the existence of imaginary eigenvalues in the linear problem of eqs. ([a15 - 2]), as soon as @xmath99. the corresponding nonlinear solutions (with @xmath104) are not covered by the stationary solution ansatz ([e]). stationary solutions (with @xmath179) solely belong to dynamical regimes below the concrete @xmath61thresholds. apart from the two @xmath61symmetry violating solutions, there should exist at least two stationary solutions which we construct in analogy to [cf. ([zpzm2])] from @xmath180 substituting the madelung representation ([aa]) and setting @xmath181 (for illustration purposes, we focus here only on this simplest case), we obtain @xmath182further, fixing @xmath183, eqs. ([phi1]) and ([phi2]) yield @xmath184obviously, the solution terminates at point @xmath90. similar to what was done above, the continuation of this branch and typical examples of its linear stability are shown in figs. [figpmpm2] and [figpmpm3], respectively. from here it is seen that the blue - circle branch, which has a complex quartet of eigenvalues, is always unstable. in fact, the gain - loss alternating configuration is generally found to be more prone to the instability. the red - cross branch is also unstable via a similar complex quartet of eigenvalues. this quartet, however, breaks into two real pairs for @xmath185, and, eventually, the additional imaginary eigenvalue pair becomes real too at @xmath186, making the solution highly unstable with three real eigenvalue pairs. the manifestation of the instability is shown in fig. [stabpmpm], typically amounting to the growth of the amplitudes at one or more gain - carrying sites. [figpmpm3]
The plaquette of the ++-
we now turn to the plaquette in fig. [modes](c), which involves parallel rows of gain and loss. in this case, the stationary equations are @xmath187 in this case too, we focus on symmetric states of the form of @xmath181 [see eq. ([aa])], which gives rise to two solutions displayed in fig. figppmm, represented by the following analytical solutions : @xmath188 @xmath189the analysis demonstrates that the branch with the upper sign in eq. (+ -1) is always unstable (through two real pairs of eigenvalues), as shown by blue circles in fig. [figppmm]. on the other hand, the branch denoted by the red crosses, which corresponds to the lower sign in eq. ([+ -1]) is stable up to @xmath190, and then it gets unstable through a real eigenvalue pair. the black - squares branch with the upper sign in eq. (+ -3) is always stable, while the green - star branch with the lower sign in eq. ([+ -3]) is always unstable. at the linear-@xmath0-symmetry breaking point @xmath191, we observe a strong degeneracy, since all the three pairs of eigenvalues for two of the branches (in the case of the blue circles, two real and one imaginary, and in the case of red crosses one real and two imaginary) collapse at the origin of the spectral plane. on the other hand, the black - squares branch is always stable with three imaginary eigenvalue pairs, while the green - star branch has two imaginary and one real pair of eigenvalues. between the latter two, there is again a collision of a pair at the origin at the critical condition, @xmath191. direct simulations, presented for @xmath192 in fig. [stabppmm], demonstrate the stability of the lower - sign black - squares branch, while the instability of the waveform associated with the blue circles and the green stars leads to the growth and decay of the amplitudes at the sites carrying, respectively, the gain and loss. notice that at the parameter values considered here, the red - cross branch is also dynamically stable as shown in the top right panel of fig. [stabppmm]. [figppmm] lastly, motivated by the existence of known cross "- shaped discrete - vortex modes in 2d conservative lattices, in addition to the fundamental discrete solitons @xcite, we have also examined the five - site configuration, in which the central site does not carry any gain or loss, while the other four feature a @xmath0-balanced distribution of the gain and loss, as shown in panel (d) of fig. [modes]. seeking for stationary states with propagation constant, @xmath193 [instead of @xmath194 in eq. ([e]), as in this case we reserve label @xmath194 for one of the sites of the 5-site plaquette in fig. [modes](d)], we get : @xmath195similarly as before, we use the madelung decomposition @xmath196, cf. ([aa]), and focus on the simplest symmetric solutions with @xmath197. without the loss of generality, we set @xmath138, reducing the equations to @xmath198 we report here numerical results for parameters @xmath199 (smaller @xmath193 yields similar results but with fewer solution branches). we have identified five different solutions in this case, see figs. [figpmzpm1] and figpmzpm2 for the representation of the continuation of the different branches, and for typical examples of their stability (the latter is shown for @xmath200, @xmath201 and @xmath202). there are two branches (green stars and black squares) that only exist at @xmath203 colliding and terminating at that point. one of them has three real eigenvalue pairs and one imaginary pair, while the other branch has two real and two imaginary pairs. two real pairs and one imaginary pair of green stars collide with two real pairs and one imaginary pair of black squares, respectively, while the final pairs of the two branches (one imaginary for the green stars and one real for the black squares) collide at the origin of the spectral plane. these collisions take place at @xmath204, accounting for the saddle - center bifurcation at the point where those two branches terminate. on the other hand, there exist two more branches (red crosses and magenta diamonds in fig. [figpmzpm1]), which collide at @xmath205. one of these branches (the less unstable one, represented by magenta diamonds) bears only an instability induced by an eigenvalue quartet, while the highly unstable branch depicted by the red crosses has four real pairs (two of which collide on the real axis and become complex at @xmath206). last but not least, the blue circles branch does not terminate at @xmath207, but continues to larger values of the gain - loss parameter, @xmath208. it is also unstable (as the one represented by the magenta diamonds) due to a complex quartet of eigenvalues. the dynamics of the solutions belonging to these branches is shown in fig. [stabpmzpm]. for the branches depicted by black squares and green stars (recall that they disappear through the collision and the first saddle - center bifurcation at @xmath204), the perturbed evolution is fairly simple : the amplitudes grow at the gain - carrying sites and decay at the lossy ones, while the central passive site (c) stays almost at zero amplitude. for the other branches, the amplitudes also grow at the two gain - carrying sites and decay at the lossy elements, while the passive site may be drawn to either the growth or decay. [figpmzpm1] [figpmzpm2]
Conclusions and future challenges[conclu]
in the present work, we have proposed generalizations of the one - dimensional @xmath0-symmetric nonlinear oligomers into two - dimensional plaquettes, which may be subsequently used as fundamental building blocks for the construction of @xmath0-symmetric two - dimensional lattices. in this context, we have introduced four basic types of plaquettes, three of which in the form of four - site squares. the final one was in the form of the five - site cross, motivated by earlier works on cross - shaped (alias rhombic or site - centered) vortex solitons in the discrete nonlinear schrdinger equation. our analysis was restricted to modes which could be found in the analytical form, while their stability against small perturbations was analyzed by means of numerical methods. even within the framework of this restriction, many effects have been found, starting from the existence of solution branches that terminate at the critical points of the respective linear @xmath0-symmetric systems e.g., in the settings corresponding to plaquettes (a) and (c) in fig. [modes]. the bifurcation responsible for the termination of the pair of branches may take a complex degenerate form [such as the one in the case of setting (a)]. other branches were found too, that continue to exist, due to the nonlinearity, past the critical points of the underlying linear systems. in addition, we have identified cases [like the gain - loss alternating pattern (b) or the cross plaquette of type (d)] when the @xmath0 symmetry is broken immediately after the introduction of the gain - loss pattern. the spectral stability of the different configurations was examined. most frequently, the stationary modes are unstable, although stable branches were found too [e.g., in settings (a) and (c)]. we have also studied the perturbed dynamics of the modes. the evolution of unstable ones typically leads to the growth of the amplitudes at the gain - carrying sites and decay at the lossy ones. it was interesting to observe that the passive sites, without gain or loss, might be tipped towards growth or decay, depending on the particular solution (and possibly on specific initial conditions). the next relevant step of the analysis may be to search for more sophisticated stationary modes (that plausibly can not be found in an analytical form), produced by the _ symmetry breaking _ of the simplest modes considered in this work, cf. ref. the difference of such modes from the @xmath0-symmetric ones considered in the present work is the fact that modes with the unbroken symmetry form a continuous family of solutions, with energy @xmath194 depending on the solution s amplitude, see eq. ([e]). this feature, which is generic to conservative nonlinear systems, is shared by @xmath0-symmetric ones, due to the automatic " balance between the separated gain and loss. on the other hand, the breaking of the symmetry gives rise to the typical behavior of systems with competing, but not explicitly balanced, gain and loss, which generate a single or several _ attractors _, i.e., _ isolated _ solutions with a single or several values of the energy, rather than a continuous family. a paradigmatic example of the difference between continuous families of solutions in conservative models and isolated attractors in their (weakly) dissipative counterparts is the transition from the continuous family of solitons in the usual nlse to a pair of isolated soliton solutions, one of which is an attractor (and the other is an unstable solution playing the role of the separatrix between attraction basins, the stable soliton and the stable zero solution) in the complex ginzburg - landau equation, produced by the addition of the cubic - quintic combination of small dissipation and gain terms to the nlse @xcite. as concerns the systems considered in the present work, in the context of the breaking of the @xmath0 symmetry it may also be relevant to introduce a more general nonlinearity, which includes @xmath0-balanced cubic gain and loss terms, in addition to their linear counterparts (cf. @xcite and @xcite). nevertheless, it should also be noted that the issue of potential existence of isolated solutions versus branches of solutions in @xmath0-symmetric systems is already starting to be addressed in the relevant literature (including in plaquette - type configurations), as in the very recent work of @xcite. moreover, the present work may pave the way to further considerations of two - dimensional @xmath0-symmetric lattice systems, and even three - dimensional ones. in this context, the natural generalization is to construct periodic two - dimensional lattices of the building blocks presented here, and to identify counterparts of the modes reported here in the infinite lattices, along with new modes which may exist in that case. on the other hand, in the three - dimensional realm, the first step that needs to be completed would consist of the examination of a @xmath0-symmetric cube composed of eight sites, and the nonlinear modes that it can support. this, in turn, may be a preamble towards constructing full three - dimensional @xmath0-symmetric lattices. these topics are under present consideration and will be reported elsewhere.
Acknowledgments
ug thanks holger cartarius and eva - maria graefe for useful discussions. pgk gratefully acknowledges support from the national science foundation under grant dms-0806762 and cmmi-1000337, as well as from the alexander von humboldt foundation and the alexander s. onassis public benefit foundation. pgk and bam also acknowledge support from the binational science foundation under grant 2010239.
References
99 c. m. bender and s. boettcher, phys. rev. lett. * 80 *, 5243 (1998) ; c. m. bender, s. boettcher and p. n. meisinger, j. math. phys. * 40 *, 2201 (1999). z. h. musslimani, k. g. makris, r. el - ganainy and d. n. christodoulides, phys. lett. * 100 *, 030402 (2008) ; k. g. makris, r. el - ganainy, d. n. christodoulides and z. h. musslimani, phys. a * 81 *, 063807 (2010). b. a. malomed and h. g. winful, phys. rev. e * 53 *, 5365 (1996) ; h. sakaguchi and b. a. malomed, physica d * 147 *, 273 (2000) ; w. j. firth and p. v. paulau, eur. j. d * 59 *, 13 (2010) ; p. v. paulau, d. gomila, p. colet, n. a. loiko, n. n. rosanov, t. ackemann, and w. j. firth, opt. express * 18 *, 8859 (2010) ; a. marini, d. v. skryabin, and b. a. malomed, _ ibid_. * 19 *, 6616 (2011) ; p. v. paulau, d. gomila, p. colet, b. a. malomed, and w. j. firth, phys. rev. e * 84 *, 036213 (2011). m. znojil, rendic. circ. mat. palermo, ser. ii, suppl. * 72 * (2004), 211 - 218, math - ph/0104012. g. s. japaridze, j. phys. a : math. theor. * 35 *, 1709 - 1718 (2002), quant - ph/0104077. a. mostafazadeh, j. math. phys. * 43 *, 205 - 214 (2002), math - ph/0107001. c. m. bender, d. c. brody, and h. f. jones, phys. lett. * 89 *, 270401 (2002), quant - ph/0208076. | we introduce four basic two - dimensional (2d) plaquette configurations with onsite cubic nonlinearities, which may be used as building blocks for 2d @xmath0-symmetric lattices. for each configuration
, we develop a dynamical model and examine its @xmath1symmetry.
the corresponding nonlinear modes are analyzed starting from the hamiltonian limit, with zero value of the gain - loss coefficient, @xmath2.
once the relevant waveforms have been identified (chiefly, in an analytical form), their stability is examined by means of linearization in the vicinity of stationary points.
this reveals diverse and, occasionally, fairly complex bifurcations.
the evolution of unstable modes is explored by means of direct simulations. in particular,
stable localized modes are found in these systems, although the majority of identified solutions is unstable. | 1204.5530 |
Introduction
the comparison between chemical abundances of deuterium, helium, and lithium predicted by bbn models with current empirical estimates is one of the most viable method to constrain the physical mechanisms and the cosmology which governed the nucleosynthesis of primordial abundances (olive, steigman, & walker 2000). as far as the primordial he content is concerned, current empirical estimates are mainly based on measurements of nebular emission lines in low - metallicity, extragalactic hii regions (izotov, thuan, & lipovetsky 1997 ; olive, steigman, & skillman 1997). recent he determinations present small observational errors (@xmath7%), but large uncertainties between independent measurements : @xmath8 by olive & steigman (1995) against @xmath9 by izotov & thuan (1998). this evidence suggests that current he abundances are still dominated by systematic errors. in fact, viegas, gruenwald, & steigman (2000) and gruenwald, steigman, & viegas (2001) in two detailed investigations on the ionization correction for unseen neutral and doubly - ionized he in hii regions, found that he estimates should be reduced by 0.006 (@xmath10), a quantity which is a factor of 2 - 3 larger than typical statistical errors quoted in the literature. moreover and even more importantly, pistinner et al. (1999) on the basis of a new grid of stellar atmosphere models for ob stars found that the inclusion of both nlte and metal - line blanketing effects causes an increase of the order of 40% in the ratio of he to h ionizing photons. this evidence together with uncertainties due to the occurrence of stellar winds, shocks, temperature fluctuations (izotov, thuan, & lipovetsky 1997 ; pistinner et al. 1999 ; peimbert, peimbert, & luridiana 2001 ; sauer, & jedamzik 2001, and references therein) and of peculiar nebular dynamics certainly affects the he abundance estimates based on giant extragalactic hii regions. in addition it is worth mentioning that the hii regions used for determining the cosmological helium abundance could have been somewhat polluted by the stellar yields of the pristine type ii supernovae, and in turn the empirical he abundances in these stellar systems should be corrected for self - pollution by massive stars. a plain evidence of this occurrence has been recently provided by aloisi, tosi, & greggio (1999), and stlin (2000). on the basis of deep hst optical and nicmos data they have resolved the stellar content of i zw 18 and found evidence that this blue compact galaxy hosts a relatively old population of asymptotic giant branch stars (@xmath11 0.1 - 5 gyr). on the other hand, the comparison between star counts of horizontal branch (hb, central he burning phase) and red giant (rg, h shell burning phase) stars in galactic globular clusters (ggcs) with the lifetimes predicted by evolutionary models, the so - called r parameter (iben 1968), supplies upper limits to primordial he mass fraction of the order of 0.20 (sandquist 2000 ; zoccali et al. 2000). however, such estimates should be cautiously treated (bono et al. 1995 ; cassisi et al. 1998), since they are hampered by current uncertainties on the nuclear cross - section of the @xmath12 reaction (buchmann 1996). note that spectroscopic measurements of he abundances in low - mass population ii stars are useless for constraining the primordial he content, because the he lines are either too faint (low - temperature stars) or affected by gravitational settling such as high temperature hb stars (giannone & rossi 1981 ; moheler et al.. however, empirical and statistical errors affecting abundance determinations of primordial deuterium, @xmath13he, and lithium could be significantly larger than for he (sasselov & goldwirth 1995 ; olive et al. moreover, the primordial he content plays a paramount role in constraining both stellar ages and cosmic distances, since the mass - luminosity (m / l) relation of low and intermediate - mass stars during h and he burning phases depends on @xmath0 (bono et al. 2000). at the same time, at fixed he to metal enrichment ratio the he abundance adopted to model evolutionary and pulsational properties of metal - rich stellar structures does depend on @xmath0 as well (bono et al. 1997 ; zoccali et al. 2000). the physical baryon density of the universe is one of the observables that can be determined with high accuracy using measurements of cmb anisotropies at intermediate and small angular scales (see e.g., hu et al. 2000, and references therein). it goes without saying that this observable plays a key role not only to assess the plausibility of the physical assumptions adopted in bbn models (tegmark & zaldarriaga 2000) but also for constraining the intrinsic accuracy of current primordial abundance estimates. according to the joint analysis of both boomerang and maxima-1 data, it has been estimated at 68% confidence level a baryon density @xmath14 (jaffe et al. 2001). on the basis of this observable esposito et al. (2001) found that the new cmb measurements are inconsistent at more that 3@xmath4 with both standard and degenerate bbn models. on the other hand, the latest analysis of the boomerang results, which has improved the removal of systematics from the data (netterfield et al. 2001), found @xmath15 (de bernardis et al. 2001), in very good agreement with the bbn value. the same conclusion has been derived from the analysis of the ground - based cmb observations performed by the dasi interferometer (halverson et al. 2001), which also found @xmath15 (pryke et al. this notwithstanding, the new analysis of the maxima data (lee et al. 2001), which extended the high @xmath5 coverage of the power spectrum measurement, still points towards somewhat higher values of the physical baryon density : @xmath16 (stompor et al. 2001). finally, we mention the measurements of the cmb power spectrum at @xmath171000 by the cosmic background imager. from these observations, and by assuming a flat cosmological model, the likelihood for the physical baryon density is found to peak at @xmath18=0.009 (padin et al. current physical baryon densities based on cmb measurements and bbn models would imply that @xmath0 might range from roughly 0.24 to approximately 0.26. the main aim of this investigation is to constrain @xmath0 on the basis of recent cmb measurements and two stellar observables that depend on @xmath0, namely the trgb luminosity and the zahb luminosity. in 2 we discuss in detail the adopted theoretical framework as well as the comparison between predicted and empirical observables. the effect of a change in @xmath0 abundance on the _ uv - upturn _ as well as on cmb anisotropies are presented in 3 and 3.1 respectively. our conclusions and final remarks are briefly mentioned in 4.
Stellar constraints on @xmath0
on the basis of the new cmb measurements and bbn models esposito et al. (2000) estimated at 68% confidence level a @xmath0 abundance ranging from 0.249 to 0.254. however, recent investigations (tegmark & zaldarriaga 2000 ; padin et al. 2001 ; stompor et al. 2001 ; tegmark, zaldarriaga, & hamilton 2001) suggest on the basis of cmb measurements and of the simplest flat inflation model that the baryon density should range from 0.009 to 0.045. the upper limit taken at face value would imply a larger primordial he content. as a generous but still plausible primordial he content we adopted @xmath3. to assess the impact that such a determination has on stellar structures we selected two observables, namely the luminosity of zahb stars and the luminosity of the tip of trgb stars. the previous observables refer to stars belonging to ggcs. the reasons why we selected these observables are the following : 1) the stellar population in ggcs are among the oldest stars in the galaxy, and therefore they are the best laboratory to investigate the primordial he content. the comparison between theory and observations in ggcs is more predictable when compared with field, halo stars, since it relies on stars that are coeval, located at the same distance, and chemically homogeneous, since the latter was somewhat contaminated by the debris of the first stellar generation.]. 2) the zahb in ggcs marks the phase in which the stars are mainly supported by @xmath19 reaction in the stellar center and it is a well - defined observational feature. the zahb luminosity depends on the he core mass and an increase of 15% in y causes an increase in the luminosity of approximately 0.16 mag (sweigart & gross 1976 ; raffelt 1990). 3) according to current evolutionary prescriptions the trgb phase marks the onset of central he burning (he - core flash) in low - mass stars and its luminosity strongly depends on the he - core mass, and in turn on the initial he content. in fact, sweigart & gross (1978) found that an increase of 15% in y causes a decrease in the trgb luminosity of 0.1 mag. finally, we mention that the previous observables are virtually unaffected by cluster age (@xmath20 gyr ; vandenberg, stetson, & bolte 1996), since for stellar ages larger than @xmath116 gyr both the zahb and the trgb luminosities do not depend on age (castellani, deglinnocenti, & luridiana 1993 ; lee, freedman, & madore 1993 ; cassisi & salaris 1997 ; salaris & cassisi 1998). 1 shows the comparison between predicted and observed zahb luminosity at the rr lyrae effective temperature (@xmath21) as a function of global metallicity abundances (salaris et al. 1993 ; vandenberg et al. 2000).]. the approach adopted to estimate the empirical bolometric magnitudes as well as their errors have been discussed by de santis & cassisi (1999). the solid and the dashed line show the zahb luminosities predicted by cassisi & salaris (1997) and by vandenberg et al. the comparison between theory and observations suggests that hb models constructed by the previous authors are in good agreement with empirical data. on the other hand, the zahb luminosities predicted by hb models constructed by assuming @xmath3 (dotted line) are @xmath22 brighter than the observed ones. note that current predictions on zahb luminosities are still controversial, since there is a mounting evidence that recent hb models based on new input physics (equation of state, neutrino energy loss rates, conductive opacities) are systematically brighter than observed and predicted by canonical hb models (cassisi et al. 1999 ; bono, castellani, & marconi 2000). however, such a conundrum does not affect our conclusion, because the systematic shift in the luminosity showed by he - enhanced hb models is a differential effect. 2 shows the comparison between theory and observations for the trgb bolometric magnitudes in a sample of 12 ggcs as a function of global metallicity. apparent bolometric magnitudes were estimated by frogel, persson, & cohen (1983) and by ferraro et al. (2000). the distance moduli adopted to derive the absolute magnitudes were estimated by comparing predicted zahb luminosities at fixed chemical composition, derived by adopting the same theoretical framework, with the observed distribution of hb stars in the color - magnitude diagram of each individual cluster. theoretical predictions (solid line) refer to models constructed by adopting a canonical initial he content of 0.23 and global metallicities ranging from [m / h]=-2.4 to -0.4. the top panel shows that the predicted luminosities are systematically brighter than the observed ones. this mismatch between theory and observations is expected and caused by the fact that the empirical estimates are hampered by sample size and also by the decrease in the lifetimes of the stellar structures approaching the trgb. by taking into account the typical sample sizes of cluster rgb stars, salaris & cassisi (1997) found that the 30% of clusters should lay within 0.1 mag (dotted line) the predicted trgb luminosity and the 70% within 0.3 mag (dashed line) the predicted ones. data plotted in the top panel support, within current uncertainties, this prediction. as a matter of fact, more than 70% of the empirical trgb bolometric magnitudes lay within the expected range (dashed and dotted lines) and only a few clusters attain magnitudes close to predicted @xmath23 values. the bottom panel of fig. 2 shows the same empirical data plotted in the top panel. theoretical predictions on trgb bolometric magnitudes are based on evolutionary models constructed by adopting an initial he content of 0.26. at the same time, the absolute bolometric magnitudes were estimated by adopting the distance moduli obtained by comparing observed hb stars with the zahb luminosities predicted by hb models constructed by adopting @xmath3, i.e. the same he content adopted in he enhanced models of fig. a glance at the data plotted in this panel shows that more than 70% of the clusters in our sample attain bolometric magnitudes similar or even brighter than predicted by he enhanced models. this finding is at odds with the straightforward statistical arguments mentioned above, and indeed only two measurements lay within 0.3 mag from the predicted trgb bolometric magnitudes.
The dependence of the _uv-upturn_ on @xmath0
stellar observables discussed in the previous section can be extended over local group (lg) galaxies or slightly beyond, i.e. on scales of the order of a few mpc. moreover, current he estimates are based on spectroscopic measurements of hii regions in extremely metal - poor blue compact galaxies. these systems are located outside the lg and their typical distances are, within the uncertainties, of the order of 200 - 300 mpc. this means that previous stellar observables and hii regions can hardly be adopted on gpc scale even by using the largest telescopes. however, the he abundance on these scales can be probed on the basis of the _ uv - upturn_. this phenomenon shows up as a sharp rise in the spectra of elliptical and s0 galaxies for wavelengths smaller than 2500 @xmath24. according to both theoretical (greggio & renzini 1990 ; castellani & tornamb 1991 ; castellani et al. 1994) and empirical (burstein et al. 1988 ; brown et al. 2000) evidence the _ uv - upturn _ is driven by the progeny of extreme and hot hb stars, namely agb - manqu and post - early - agb (castellani & tornamb 1991). this observable presents a strong dependence on age, since he abundance affects the mass of main sequence turn off stars, the hb morphology, and in turn the uv emission of old stellar populations. as a consequence, the _ uv - upturn _ is a powerful age indicator in elliptical galaxies (tantalo et al. 1996 ; greggio & renzini 1999). however, as already mentioned in the previous section the evolutionary properties of hb stars do depend on the primordial he content. in particular, an increase in y causes, at fixed age, an increase in the zahb luminosity. current hb evolutionary models for metal - poor and metal - rich structures (@xmath25) constructed (zoccali et al. 2000) by adopting different he contents (@xmath26) suggest that the hb luminosity at @xmath21 scales with the initial he content according to the following relation : @xmath27. a quite similar value was also suggested by raffelt (1990) on the basis of hb models computed by sweigart & gross (1976, 1978). thus suggesting that evolutionary predictions on this ratio are quite robust. this means that in metal - rich populations, which is typical of e and s0 galaxies, an increase in y of a factor of two (0.23 vs 0.46) causes an increase in @xmath28 of the order of 0.40 dex. this result is a rough estimate of the impact of y on the hb luminosity, it can not be easily extrapolated to the _ uv - upturn_. in fact, the crucial parameters for this phenomenon are the changes in hb and post - hb evolutionary lifetimes as well as the uv flux and not the bolometric one. this means that a detailed quantitative estimate does require synthetic models which simultaneously account for both h and he burning phases, and in turn for the change in the spectral energy distribution (sed) of the entire field population (brown et al. this notwithstanding, we are interested in estimating the dependence of the _ uv - upturn _ on he content at intermediate redshifts, and therefore we constructed four metal - poor (z=0.0001) extreme hb structures at different he contents, namely 0.15, 0.23, 0.35, and 0.5 (see fig. the he - core masses of these structures were estimated by constructing h - burning evolutionary tracks that reach the rgb tip with an age of @xmath11 10 gyr. we find that the he - core masses for the previous compositions are 0.523, 0.511, 0.481, and 0.433 @xmath29 respectively. to estimate the impact of he content on the uv emission of agb - manqu stars we selected a typical effective temperature for these structures, namely @xmath30. one finds that the total mass of hb structures at the selected compositions range from @xmath31 0.53 (@xmath32) to 0.45 (@xmath33). the evolution was followed from central he - burning till the beginning of the white dwarf cooling sequence (see fig. 3), and then we estimated the total spectral energy distribution (sed) for the four structures according to the individual evolutionary lifetimes. it turns out that the uv emission of extreme horizontal branch (ehb) structures is relatively sensitive to the he content, and indeed an increase from @xmath34 to @xmath35 and @xmath33 causes a decrease in the uv emission by roughly 10% and 22% respectively. at the same time, a decrease from @xmath34 to @xmath32 causes an increase by @xmath119%. this effect is due to the fact that among ehb structures an increase in y causes, at fixed age and effective temperature, a decrease in the he - core mass and in turn in the zahb luminosity. this means that he - rich ehb structures spend a substantial portion of their central he - burning lifetime at lower luminosities. therefore their total uv emission decreases when compared with he - poor structures. moreover, we also find that an increase in the he content causes in metal - poor structures a decrease in the range of stellar masses evolving at high temperatures during he - burning. in fact, we find that the largest mass that evolve as agb - manqu slightly decreases from 0.54 at @xmath32 to 0.52 at @xmath35. this finding further strengthens the evidence that in metal - poor structure an increase in the he abundance causes a decrease in the uv emission. note that such a trend is at odds with the behavior in metal - rich structure, and indeed dorman et al. (1993) found that an increase in y causes an increase in the largest mass evolving as agb - manqu, and in turn in the uv emission. the difference seems to be due to the fact that the evolution of metal - poor ehb structures is mainly governed by central he burning, and therefore by the he - core mass at the tip of the rgb. on the contrary the h - burning shell is more efficient in metal - rich ehb structures. this means that the he - core mass in these structures undergoes a mild increase during the central he burning phase, and therefore the range of stellar masses evolving as agb - manqu increases as well. once again we note that current arguments are preliminary but plausible speculations of the impact of he - content on the _ uv - upturn_. however, a firm quantitative evaluation requires the calculation of synthetic population models that account for both agb - manqu and post - early - agb structures. this seems a promising result, since a substantial increase / decrease in the he content should cause, at fixed look - back time and similar star formation histories, a decrease / increase in the scatter of the empirical average restframe 1550-v colors, i.e. the fingerprint of the _ uv - upturn_. moreover and even more importantly, present - day instrumentation allowed the detection and measurement of the _ uv - upturn _ in intermediate redshift (@xmath36) e galaxies (brown et al. 2000b). by assuming a hubble constant @xmath37 km s@xmath38 mpc@xmath38 and an eistein - de sitter cosmological model one finds that the comoving distance @xmath39 and the look - back time @xmath40 of this e galaxy are @xmath41 gpc and @xmath42 gyr respectively. on the other hand, if we assume a high lambda cosmological model (@xmath43 @xmath44) one finds @xmath45 gpc and @xmath46 gyr. this is the reason why we constructed ehb structures by adopting he - core masses at evolutionary ages of approximately 10 gyr. therefore this evidence and our finding seem to support the use of the _ uv - upturn _ to trace the he content up to gpc scales. as it has been emphasized countless times in the literature (see e.g., jungman et al. 1996 ; hu & white 1996 ; kamionkowski & kosowsky 1999), observations of cmb anisotropies provide a powerful way of setting tight constraints on the value of most cosmological parameters. in particular, the angular power spectrum of cmb temperature fluctuations depends both on the primordial fluctuations which seeded structure formation in the universe and on the physical processes occurring before the recombination in the baryon - photon plasma. within the framework of inflationary adiabatic models, these processes leave a characteristic imprint in the angular power spectrum in the form of a series of harmonic peaks (usually named `` acoustic peaks '' in the literature) whose height and position is sensitively dependent on the parameters of the cosmological model. while the parameter which is most robustly determined from measurements of the cmb power spectrum is undoubtedly the total energy density of the universe, many other parameters are measurable with striking precision. among them, the physical density of baryonic matter in the universe, whose value affects the height ratio of odd and even peaks in the spectrum : in particular, a high baryon density enhances the height of the first peak with respect to the second, and vice - versa. this effect is quite relevant even for small variations of @xmath18, as shown in fig. the dependence of the cmb anisotropy on the primordial he mass fraction is, on the contrary, quite weak. as shown in figure 5, the effect on the first peak is just about @xmath47, even for an unrealistically large range of values @xmath48. one has to go beyond the second peak in order to obtain effects larger than @xmath49. we stress the fact that the effect of both the baryon density and the primordial he abundance on the cmb pattern is only relevant at small angular scales (corresponding to @xmath6 in the power spectrum). these scales are only weakly affected by the uncertainty deriving from the limited coverage of the observed region, and from the so called `` cosmic variance '' resulting from the fact that we can only observe one statistical realization (our sky) drawn from the underlying cosmological model. on the other hand, the high @xmath5 end of the spectrum is unfortunately the one which is currently most at risk of being affected by unknown systematics such as pointing inaccuracies, poorly known beam pattern, residual instrumental noise, etc. until recent times, no high resolution observation of the cmb anisotropy pattern was available. consequently, the structure of peaks in the power spectrum could not be resolved with the accuracy needed to obtain a precise measurement of the baryon density. the situation has dramatically changed after recent observations. the maxima and boomerang experiments (hanany et al. 2000 ; de bernardis et al. 2000) produced the first high - resolution maps of the cmb, and measured the cmb angular power spectrum on a wide range of @xmath5 : @xmath50, corresponding to angular scales @xmath51. this has provided tight constraints on the main parameters of the inflationary adiabatic model (balbi et al. 2000 ; lange et al. 2000 ; jaffe et al. in particular, the constraints on the physical baryon density from the joint analysis of the maxima and boomerang data, @xmath14, was found to be higher than the one derived from primordial nucleosynthesis considerations @xmath52 (see e.g., burles, nollett & turner 2001a ; burles et al. 1999) although the bbn value fall within the cmb 95% confidence interval. this stirred some discussion about the existence of a conflict between cmb and bbn and possible explanation for it (for example, burles, nollett & turner 2001b ; kurki - suonio & sihvola 2001 ; esposito et al. 2001 ; di bari & foot 2001 ; lesgourgues & peloso 2000). it is remarkable, however, that these first limits from the cmb agree, within 2@xmath4, with those from the bbn, which are derived using a different set of measurements.
Summary and final remarks
recent measurements of the cmb anisotropy provided the unique opportunity to evaluate several fundamental cosmological parameters and to supply for all of them a preliminary but plausible estimate of their error budget. the impact of these new measurements on cosmological models uncorked a flourishing literature. however, it is not easy to assess on a quantitative basis to what extent current differences in the physical baryon density derived from cmb observations are caused by deceptive systematic errors. as a matter of fact, cmb measurements of @xmath53 range from 0.009 (padin et al. 2001) to @xmath54 (95% confidence level, stompor et al. 2001). according to bbn models the new measurements imply that @xmath0 might range from approximately 0.24 to roughly 0.26. by adopting the upper limit on @xmath0 we investigated the impact of the change on two stellar observables, namely the zahb luminosity and the luminosity of the tip of the rgb. the main outcome of our analysis is that an increase in the primordial he content from the canonical @xmath34 to @xmath3 does not seem to be supported by the comparison between current theoretical predictions and empirical data. we found that the _ uv - upturn _ can be adopted to estimate the primordial he content. in fact, numerical experiments suggest that an increase of @xmath0 from 0.23 to 0.50 causes a decrease in the uv emission at least of the order of 20%. this is a preliminary rough estimate based on the assumption that agb - manqu structures are the main sources of the _ uv - upturn_. an interesting feature of this observable is that current instruments can allow us to measure the _ uv - upturn _ up to distances of the order of gpcs. note that to supply quantitative estimates of @xmath0 on the basis of the comparison between synthetic and observed _ uv - upturns _ it is necessary to account for the sed typical of complex stellar populations as a function of redshift (tantalo et al. 1996 ; yi et al. however, theoretical predictions should be cautiously treated, since uv flux when moving from low to high metal contents strongly depends on the efficiency of the mass loss as well as on the he to metal enrichment ratio (greggio & renzini 1999). the scenario has been further complicated by recent spectroscopic measurements of hot hb stars (@xmath55 k) in metal - poor ggcs (m15, m13). in fact, behr et al. (2000) and behr, cohen, & mccarthy (2000) found that in these stars the iron abundance is enhanced by 1 - 2 order of magnitudes, whereas the he content is depleted by at least one order of magnitude respect to solar abundance. unfortunately, we still lack quantitative estimates of the impact that such a peculiarities have on the uv emission. theoretical and empirical arguments support the evidence that the density of baryons in the universe is homogeneous (copi, olive, & schramm 1995). the same outcome applies to large scale chemical inhomogeneities (copi, olive, & schramm 1996). however, it has been recently suggested by dolgov & pagel (1999, hereinafter dp) a new cosmological model that predicts a substantial spatial variation in the primordial chemical composition and a small baryon density variation. this investigation was triggered by a difference of one order of magnitude in the deuterium abundance of damped @xmath56 systems along the line of sight of high - redshift (@xmath57) qsos (dodorico et al. 2001 ; steigman et al. the scenario developed by dp relies on a model of leptogenesis (dolgov 1992) in which takes place a large lepton asymmetry and this asymmetry undergoes strong changes on spatial scales ranging from mega to giga pcs. the key feature of this model is to predict a large and varying lepton asymmetry and a small baryon asymmetry. within this theoretical framework the he mass fraction in deuterium - rich regions should range from 35% to 60%, while the @xmath58 one should increase up to @xmath59, while the variation of the photon temperature should be @xmath60. obviously, the hypothesis that current changes in baryon density are due to real spatial variations is premature as any further speculative issue. future full - sky cmb observations from space missions such as nasa s map (wright 1999) and esa s planck (mandolesi et al. 1998) will play a crucial role to properly address the problem of the spatial variation, since they will supply a larger sensitivity up to very high @xmath5 (@xmath61) and an improved control on systematics. we warmly thank claudia maraston for kindly providing us the spectral energy distribution of current hb models. we also ackowledge an anonymous referee for his / her useful suggestions that improved the readability of the paper. this work was supported by murst / cofin2000 under the project : `` stellar observables of cosmological relevance '' (g. b., r. b., & s. c.). aloisi, a., tosi, m., & greggio, l. 1999, aj, 118, 302 balbi, a., et al., 2000, apjl, 545, l1 behr, b. b., cohen, j. g., mccarthy, j. k. 2000, apj, 531, l37 behr, b. b., djorgovski, s. g., cohen, j. g., mccarthy, j. k., cot, p., piotto, g., zoccali, m. 2000, apj, 528, 849 bono, g., caputo, f., cassisi, s., incerpi, r., & marconi, m. 1997, apj, 483, 811 bono, g., caputo, f., cassisi, s., marconi, m., piersanti, l., & tornamb, a. 2000, apj, 543, 955 bono, g., castellani, v., deglinnocenti, s., & pulone, l. 1995, a&a, 297, 115 bono, g., castellani, v., & marconi, m. 2000, apj, 532, l129 brown, t. m., bowers, c. w., kimble, r. a., sweigart, a. v., & ferguson, h. c. 2000, apj, 532, 308 brown, t. m., ferguson, h. c., stanford, s. a., & deharveng, j 1998, apj, 504, 113 burles, s., nollett, k.m., & turner, m. 2001a, apjl, 552, l1 burles, s., nollett, k.m., & turner, m. 2001b, phys. d, 63, 063512 burles, s., et al., 1999, phys. lett., 82, 4176 burstein, d., bertola, f., buson, l. m., faber, s. m., & lauer, t. r. 1988, apj, 328, 440 cassisi, s., castellani, v., deglinnocenti, s., & weiss, a. 1998, a&as, 129, 267 cassisi, s., castellani, v., deglinnocenti, s., salaris, m., & weiss, a. 1999, a&as, 134, 103 cassisi, s., & salaris, m. 1997, mnras, 285, 593 castellani, m., castellani, v., pulone, l., & tornamb, a. 1994, a&a, 282, 711 castellani, v., deglinnocenti, s., & luridiana, v. 1993, a&a, 272, 442 castellani, m., & tornamb, a. 1991, apj, 381, 393 copi, c. j., olive, k. a., & schramm, d. n. 1995, apj, 451, 51 copi, c. j., olive, k. a., & schramm, d. n. 1996, astro - ph/9606156 de bernardis, p., et al. 2000, nature, 404, 955 de bernardis, p., et al. 2001, apj, submitted, astro - ph/0105296 de santis, r., & cassisi, s. 1999, mnras, 308, 97 di bari, p. & foot, r. 2001, phys. d, 63, 043008 dodorico, s., dessauges - zavadsky, m., & molaro, p. 2001, a&a, 368, l21 dorman, b., rood, r. t., & oconnell, r. w. 1993, apj, 419, 596 dolgov, a. d. 1992, phys., 222, 6 dolgov, a. d., & pagel, b. e. j. 1999, newa, 4, 223 (dp) esposito, s., et al., 2001, d, 63, 043004 ferraro, f. r., montegriffo, p., origlia, l., & fusi pecci, f. 2000, aj, 119, 1282 frogel, j. a., persson, s. e., & cohen, j. g. 1983, apjs, 53, 713 giannone, p., & rossi, l. 1981, a&a, 102, 386 greggio, l., & renzini, a. 1990, apj, 364, 35 greggio, l., & renzini, a. 1999, mem. astr. it., 70, 691 gruenwald, r., steigman, g. & viegas, s. m. 2000, apj, accepted, astro - ph/0109071 halverson, n.w. 2001, apj, submitted, astro - ph/0104489 hanany, s., et al., 2000, apjl, 545, l5 hu, w. & white, m. 1996, apj, 471, 30 iben, i. jr. 1968, nature, 220, 143 izotov, y. i., & thuan, t. x. 1998, apj, 497, 227 izotov, y. i., thuan, t. x., & lipovetsky, v. a. 1997, apjs, 108, 1 jaffe, a., et al. 2001, phys. 86, 3475 jungman, g., et al. 1996, phys.rev. d54, 1332 kamionkowski, m. & kosowsky, a. 1999, ann.rev.nucl.part.sci. 49, 77 - 123 kurki - suonio, h. & sihvola, e., 2001, phys. d, 63, 083508 lange, a. et al., 2001, d, 63, 042001 lesgourgues, j. & peloso, m. 2000, phys. d, 62, 081301 lee, m. g., freedman, w. l., & madore, b. f. 1993, apj, 417, 533 lee, a.t., et al. 2001, apj, 561, l1 mandolesi, n., et al. 1998, planck low frequency instrument, a proposal submitted to esa moehler, s., sweigart, a. v., landsman, w. b., heber, u., & catelan, m. 1999,, 346, l1 netterfield, b., et al. 2001, apj, submitted, astro - ph/0104460 olive, k. a., & steigman, g. 1995, apjs, 97, 49 olive, k. a., steigman, g., & skillman, e. d. 1997, apj, 483, 788 olive, k. a., steigman, g., & walker, t. p. 2000, phys., 333, 389 stlin, g. 2000, apj, 535, l99 padin, s., et al. 2001, apj, 549, l1 peimbert, m., peimbert, a., & luridiana, v. 2001, apj, accepted, astro - ph/0107189 pistinner, s. l., hauschildt, p. h., eichler, d., & baron, e. a. 1999, mnras, 302, 684 pryke, c., et al., 2001, apj, submitted, astro - ph/0104490 puget, j. l., et al. 1998, planck high frequency instrument, a proposal submitted to esa raffelt, g. g. 1990, apj, 365, 559 salaris, m., & cassisi, s. 1997, mnras, 289, 406 salaris, m., & cassisi, s. 1998, mnras, 298, 166 salaris, m., chieffi, a., & straniero, o. 1993, apj, 414, 580 sandquist, e. l. 2000, mnras, 313, 571 sasselov, d., & goldwirth, d. 1995, apj, 444, l5 sauer, d., & jedamzik, k. 2001, a&a, submitted, astro - ph/0104392 stompor, r. et al. 2001, apj, 561, l7 sweigart, a. v., & gross, p. g. 1976, apjs, 32, 367 sweigart, a. v., & gross, p. g. 1978, apjs, 36, 405 tantalo, r., chiosi, c., bressan, a., & fagotto, f. 1996, a&a, 311, 361 tegmark, m., & zaldarriaga, m. 2000, apj, 544, 30 tegmark, m., zaldarriaga, m., & hamilton a. j. s. 2001, phys. d., 63, 043007 vandenberg, d. a., stetson, p. b., & bolte, m. 1996, ara&a, 34, 461 vandenberg, d. a., swenson, f. j., rogers, f. j.,, iglesias, c. a., & alexander, d. r. 2000, apj, 532, 430 viegas, s. m., gruenwald, r., & steigman, g. 2000, apj, 531, 813 wright, e.l., 1999, new astr.rev., 43, 257 yi, s., lee, y.- w., woo, j.- h., park, j.- h., demarque, p., oemler, a. jr. 1999, apj, 513, 128 zoccali, m., cassisi, s., piotto g., bono g., & salaris m. 1999, apj, 518, l49 zoccali, m., cassisi, s., bono, g., piotto, g., rich, r. m., & djorgovski, s. g. 2000, apj, 538, 289 | we present the results of a joint investigation aimed at constraining the primordial he content (@xmath0) on the basis of both the cosmic microwave background (cmb) anisotropy and two stellar observables, namely the tip of the red giant branch (trgb) and the luminosity of the zero age horizontal branch (zahb).
current baryon density estimates based on cmb measurements cover a wide range values @xmath1, that according to big bang nucleosynthesis (bbn) models would imply @xmath2.
we constructed several sets of evolutionary tracks and hb models by adopting @xmath3 and several metal contents.
the comparison between theory and observations suggests that zahb magnitudes based on he - enhanced models are 1.5@xmath4 brighter than the empirical ones.
the same outcome applies for the trgb bolometric magnitudes.
this finding somewhat supports a @xmath0 abundance close to the canonical 0.23 - 0.24 value.
more quantitative constraints on this parameter are hampered by the fact that the cmb pattern shows a sizable dependence on both @xmath0 and the baryon density only at small angular scales, i.e. at high @xmath5 in the power spectrum (@xmath6).
however, this region of the power spectrum could be still affected by deceptive systematic uncertainties. finally, we suggest to use the _ uv - upturn _ to estimate the he content on gpc scales.
in fact, we find that a strong increase in @xmath0 causes in metal - poor, hot hb structures a decrease in the uv emission. | astro-ph0112064 |
Acknowledgments
the author would like to thank c. bunster, c. martnez, r. troncoso, s. willison, and j. zanelli for useful comments. the author thanks kjell tangen for the crucial information of the paper by fonarev. this work was supported by the grant nos. 1071125 from fondecyt (chile) and the grant - in - aid for scientific research fund of the ministry of education, culture, sports, science and technology, japan (young scientists (b) 18740162). 99 h. stephani, d. kramer, m. maccullum, c. hoenselaers and e. herlt, _ exact solutions of einsteins field equations : second edition _, (cambridge university press, cambridge, 2003). a. krasiski, _ inhomogeneous cosmological models _ (cambridge : cambridge university press, 1997). peebles, _ principles of physical cosmology _ (princeton univ. press, 1976). linde, _ particle physics and inflationary cosmology _ (harwood academic publishers, chur, switzerland 1990). a. liddle, _ an introduction to modern cosmology _ (wiley, 2003). s.w. hawking and g.f.r. ellis, _ `` the large scale structure of space - time '' _ (cambridge university press, cambridge, england, 1973). `` general relativity '' _ (university of chicago press, chicago, united states of america, 1983). roberts, gen. rel. grav. * 21 *, 907 (1989). v. husain, e.a. martinez and d. nez, d * 50 *, 3783 (1994). choptuik, phys. * 70 *, 9 (1993). c. gundlach, liv. * 2 *, 4 (1999). d. christodoulou, commun. phys. * 109 *, 613 (1987) ; commun. pure appl. math. * 44 *, 339 (1991) ; commun. pure appl. math. * 46 *, 1131 (1993) ; ann. of math. * 140 *, 607 (1994) ; ann. of math. * 149 *, 187 (1999). townsend, jhep * 11 *, 042 (2001). a.m. green and j.e. lidsey, phys. d * 61 *, 067301 (2000). r. emparan and j. garriga, jhep * 05 *, 028 (2003). fonarev, class. quantum grav. * 12 *, 1739 (1995). y. kitada and k. maeda, phys. d * 45 *, 1416 (1992) ; y. kitada and k. maeda, class. quantum grav. * 10 *, 703 (1993) ; a.r. liddle, a. mazumdar and f.e. schunck, phys. d * 58 *, 061301 (1998) ; k.a. malik and d. wands, phys. d * 59 *, 123501 (1999) ; a.a. coley and r.j. van den hoogen, phys. d * 62 *, 023517 (2000) ; z - k. guo, y - s. piao, r - g. cai and y - z. zhang, phys. b * 576 *, 12 (2003). hayward, phys. d*49 *, 6467 (1994). m. heusler, j. math. * 33 *, 3497 (1992). j. d. bekenstein, phys. rev. d * 51 *, 6608 (1995). d. sudarsky, class. * 12 *, 579 (1995). t. torii, k. maeda and m. narita, phys. rev. d * 59 *, 064027 (1999). t. torii, k. maeda and m. narita, phys. d * 64 *, 044007 (2001). | we discuss the physical interpretation of a dynamical and inhomogeneous spherically symmetric solution obtained by fonarev for a scalar field with an exponential potential. there is a single parameter @xmath0 in the solution which can be set to @xmath1 if it is non - zero, in addition to the steepness parameter @xmath2 in the potential.
the spacetime is conformally static and asymptotically flat friedmann - robertson - walker spacetime.
the solution reduces to the friedmann - robertson - walker solution for @xmath3.
there are two curvature singularities, of which one is a timelike central singularity and the other is a big - bang or big - crunch type singularity.
depending on the parameters, the spacetime can possess a future outer trapping horizon in the collapsing case.
then the solution represents a dynamical black hole in the sense of hayward although there is a locally naked singularity at the center and no black - hole event horizon.
this demonstrates a weak point of the local definition of a black hole in terms of a trapping horizon.
einstein equations are so complicated simultaneous nonlinear partial differential equations that it is hopeless to obtain general solutions.
thus, spacetime symmetries are often assumed to make the system more tractable.
although such spacetimes with high symmetries are idealized ones, they have occupied an important position in the history of gravitation research as touchstones to know the essential physics @xcite. for example, the friedmann - robertson - walker (frw) cosmological model plays a central role in modern cosmology as the zeroth - order approximation of the present universe, on which the behavior of the density perturbations has been particularly investigated to clarify the origin of the large - scale structure of the universe or to determine the cosmological parameters from the observations of the cosmic microwave background @xcite. from the analyses of the stationary and asymptotically flat black - hole spacetime such as the schwarzschild or kerr solution,
tremendous results have been derived, among which the most remarkable one is the black - hole thermodynamics @xcite. on the other hand, there are many open problems on the dynamical aspects of einstein equations such as dynamical black holes or their formations. in homogeneous or stationary spacetimes,
einstein equations reduce to a set of ordinary differential equations, which is comparatively easy to handle, however, the formation or the growth of a black hole is essentially a dynamical and inhomogeneous process, where we have to struggle with a set of partial differential equations.
for this reason, numerical methods have often been used to study such systems.
nevertheless, the few dynamical and inhomogeneous exact solutions have been found.
such precious solutions should be intensively investigated to complement the numerical works.
scalar fields are fundamental fields which naturally exist in a variety of theories. in spherically symmetric spacetimes,
the system with the simplest massless scalar field has been fully investigated.
there are two important dynamical and inhomogeneous exact solutions in this case.
the first one was obtained by roberts @xcite, and subsequently the other one was obtained by husain, martinez and nez @xcite.
the dynamical aspects of the system with a massless scalar field have been investigated in the numerical studies of the gravitational collapse, especially in the context of critical phenomena pioneered by choptuik @xcite (see @xcite for the review).
the analytic proof of the cosmic censorship hypothesis by christodoulou is a significant milestone @xcite.
although these two exact solutions are not directly related to these results, the potential importance of them would go without saying.
the case with potentials is a natural generalization of the massless case. in various theories,
scalar fields have their specific potentials. among them, an exponential potential arises naturally in supergravity @xcite or theories obtained through dimensional reduction to effective four - dimensional theories @xcite.
indeed, the existence of exact solutions in such systems must be useful as a touchstone for the future research.
the generalized husain - martinez - nez (hmn) solution in the presence of an exponential potential was obtained by fonarev @xcite.
however, the properties of the solution have not been studied in details and the physical interpretation of the solution is still not clear. in this letter, we discuss the physical interpretation of the fonarev solution.
we investigate the properties of the trapping horizon and show the global structure of the solution.
we adopt the units such that @xmath4.
we begin with the action which describes the system with a scalar field with an exponential potential : @xmath5, \label{action}\]] where @xmath6 is the potential of a scalar field @xmath7 given by @xmath8 where @xmath9 and @xmath2 are real constants.
a scalar field with an exponential potential has been particularly studied in spatially homogeneous cosmology @xcite.
if @xmath10, then @xmath7 is massless and @xmath2 is meaningless.
we assume that @xmath9 is non - negative.
then the energy - momentum tensor for a scalar field is given by @xmath11 the einstein equation is @xmath12 while equations of motion for @xmath7 is given by @xmath13 the spherically symmetric solution obtained by fonarev @xcite is @xmath14, \label{sol } \\
\phi & = & \frac{1}{4\sqrt{\pi}}\ln\biggl[d\biggl(1-\frac{2w}{r}\biggl)^{\sqrt{2/(\lambda^2 + 2)}}a^{\sqrt{2}\lambda}\biggl],\label{sol2}\end{aligned}\]] where @xmath15^{-\sqrt{2}/\lambda}.\end{aligned}\]] here @xmath0 and @xmath16 are constants.
when we set @xmath3, the solution reduces to the frw solution.
hereafter we only consider the case with @xmath17.
the potential form is then given by @xmath18 so the steepness parameter @xmath2 must satisfy the relation @xmath19 and @xmath20 for non - negative potential, which corresponds to @xmath21 with @xmath22.
this solution is asymptotically frw solution for @xmath23. for @xmath24 (corresponding to @xmath25),
the asymptotic frw solution is accelerated, while for @xmath26 (corresponding to @xmath27), it is decelerated.
when we set @xmath28, this solution reduces to the hmn solution for a massless scalar field, in which @xmath29 is an arbitrary constant and meaningless @xcite. because this solution is invariant for @xmath30 with @xmath31
, we consider only the region of @xmath32. for @xmath33
, there are central curvature singularities at @xmath34 and @xmath35, which are both timelike.
thus, the domain of definition for @xmath36 is @xmath37 for @xmath38, while it is @xmath39 for @xmath40.
then, we can set @xmath41 for positive (negative) @xmath0 without loss of generality by the coordinate transformations @xmath42 and @xmath43 $] and the redefinition of the constant @xmath16.
in addition to them, @xmath44 is a null curvature singularity for @xmath24, in which case @xmath45 corresponds to past (future) infinity, while @xmath46 is a spacelike curvature singularity for @xmath26, in which case @xmath47 corresponds to future (past) infinity.
these are the big - bang or big - crunch type singularities.
the domain of definition for @xmath48 is thus @xmath49 and @xmath50.
the physical (areal) radius @xmath51 is given by @xmath52.
@xmath51 is a monotonically increasing function of @xmath36 for @xmath37 and @xmath23 corresponds to the spacelike infinity. for @xmath24,
@xmath51 is a monotonically increasing (decreasing) function of @xmath48 for @xmath53, i.e., the spacetime is expanding (collapsing).
for @xmath26, on the other hand, it is a monotonically increasing (decreasing) function of @xmath48 for @xmath54. here
let us consider whether the scalar field has a non - trivial configuration or not
. we should be careful that there is no natural time - slicing in general spherically symmetric spacetimes, so the derivative of the scalar field must have a spacelike portion to have the non - trivial configuration in a correct sense. for the fonarev solution,
we have @xmath55 and there is indeed a region with @xmath56. the misner - sharp mass @xmath57 is given by @xmath58 where we have @xmath59 it is noted that @xmath60 and @xmath61 for @xmath21 with @xmath22.
a trapping horizon is obtained by @xmath62, or equivalently @xmath63, which is given by @xmath64 it is noted that outgoing (ingoing) null geodesics are trapped on the trapped regions in the collapsing (expanding) regions.
the normal vector @xmath65 to the surface @xmath66 is given by @xmath67.
we obtain @xmath68 ^ 2g(r)}{a^2(2\alpha^2 - 1)^2[r-(1+\alpha)w]^4},\\ g(r)&:=&(3\alpha^2 - 2)r^2 \nonumber \\ & & -2(1+\alpha)(4\alpha^2-\alpha-2)wr \\ & & + (4\alpha^2 - 3)(1+\alpha)^2w^2\end{aligned}\]] for both signs in eq.
([t - horizon]).
the function @xmath69 is negative for @xmath70, which implies that the trapping horizon is timelike.
the case of @xmath71 is rather complicated.
we obtain @xmath72 which is non - positive for @xmath73.
the solution of @xmath74 exists only for @xmath75. for @xmath76
, we have @xmath77.
thus, for @xmath78, the trapping horizon is timelike for @xmath79, while for @xmath80, the trapping horizon is timelike (spacelike) for @xmath81, where @xmath82 is defined by @xmath83.
next let us consider the case with @xmath84.
the solutions of @xmath74 are then given as @xmath85 defined by @xmath86}{2(3\alpha^2 - 2)}.\end{aligned}\]] we can show that @xmath87 holds for @xmath88, @xmath89, and @xmath90 with equality holding for @xmath91, while @xmath92 holds for @xmath93.
we can also show that @xmath94 holds for @xmath95, while @xmath96 holds for @xmath88, @xmath97, and @xmath90.
thus, it is concluded that the trapping horizon is timelike for @xmath93 and spacelike for @xmath91, while in other cases it has both timelike and spacelike portions. for @xmath98,
the trapping horizon is timelike for @xmath99 and spacelike for @xmath100. for @xmath95,
the trapping horizon is timelike for @xmath101 and @xmath102, while it is spacelike for @xmath103.
we have now found that the properties of the trapping horizon are different from that in the hmn solution corresponding to @xmath91, which is spacelike.
a local definition of a black hole in terms of a trapping horizon was given by hayward @xcite.
then, a future outer trapping horizon is a black - hole horizon, which corresponds to a spacelike trapping horizon in the collapsing region. on the other hand,
a spacelike trapping horizon in the expanding region is a future inner trapping horizon corresponding to a white - hole or cosmological horizon.
thus, there exists a black - hole trapping horizon for @xmath88 (corresponding to @xmath104) and @xmath105 (corresponding to @xmath106). in summary,
the solution represents a dynamical black hole in the sense of hayward with the non - trivial configuration of a scalar field, i.e., a scalar hair, in the collapsing case with @xmath107, where the asymptotic frw solution is decelerating.
the penrose diagrams of the fonarev solution are given in fig.
[penrose].
the penrose diagrams for the fonarev solution in the collapsing case with @xmath33 and (a) @xmath108 and (b) @xmath26.
a zigzag line corresponds to a curvature singularity.
the central timelike singularity is at @xmath109 and @xmath34 for @xmath110 and @xmath111, respectively. for @xmath3,
it is replaced by a regular center.
the diagrams in the expanding cases are obtained by setting the figures upside - down.] in the static and asymptotically flat case, the scalar no - hair theorem is available, in which the scalar field must have a trivial configuration for a black - hole solution for the arbitrary positive semidefinite potential @xcite.
this scalar no - hair theorem was extended into the asymptotically de sitter case in the presence of the convex potential @xcite. on the other hand,
a counter example was numerically constructed in the asymptotically anti - de sitter case, which violates the null energy condition @xcite.
if the scalar no - hair theorem also holds in general dynamical and inhomogeneous spherically symmetric spacetimes with certain energy conditions, a scalar field must have a trivial configuration for black - hole solutions.
the contraposition of this statement is that if a scalar field has a non - trivial configuration, the solution is not a black - hole solution.
then, does the fonarev solution suggest that the scalar no - hair theorem can be extended into general spherically symmetric spacetimes?
the answer is `` no '' in the sense of hayward.
however, here we face with the subtlety of the definition of a black hole. although the solution represents a dynamical black hole in the sense of hayward, the singularity inside that trapping horizon is the big - crunch type spacelike singularity, and the central singularity is timelike and not covered by trapped surfaces. furthermore, there is no black - hole event horizon in this spacetime since there is no future null infinity. from these viewpoints,
it might be proper to say that the fonarev solution as well as the hmn solution represents not a black hole but a naked singularity.
then, the answer to the above question turns to be `` yes ''.
these solutions may demonstrate a weak point of the local definition of a black hole in terms of a trapping horizon. in order to rule out such solutions
, we should require additional conditions in the definition by a trapping horizon or another local definition of a black hole.
further investigations are needed to have much insight into this problem. | 0704.2731 |
Introduction
the @xmath0 ncsm / rgm was presented in @xcite as a promising technique that is able to treat both structure and reactions in light nuclear systems. this approach combines a microscopic cluster technique with the use of realistic interactions and a consistent @xmath0 description of the nucleon clusters. the method has been introduced in detail for two - body cluster bases and has been shown to work efficiently in different systems @xcite. however, there are many interesting systems that have a three - body cluster structure and therefore can not be successfully studied with a two - body cluster approach. the extension of the ncsm / rgm approach to properly describe three - body cluster states is essential for the study of nuclear systems that present such configuration. this type of systems appear, @xmath3, in structure problems of two - nucleon halo nuclei such as @xmath1he and @xmath4li, resonant systems like @xmath5h or transfer reactions with three fragments in their final states like @xmath6h(@xmath6h,2n)@xmath2he or @xmath6he(@xmath6he,2p)@xmath2he. recently, we introduced three - body cluster configurations into the method and presented the first results for the @xmath1he ground state @xcite. here we present these results as well as first results for the continuum states of @xmath1he within a @xmath2he+n+n basis.
Formalism
the extension of the ncsm / rgm approach to properly describe three - cluster configurations requires to expand the many - body wave function over a basis @xmath7 of three - body cluster channel states built from the ncsm wave function of each of the three clusters, @xmath8 @xmath9^{(j^{\pi}t) } \times \frac{\delta(x-\eta_{a_2-a_3})}{x\eta_{a_2-a_3 } } \frac{\delta(y-\eta_{a - a_{23}})}{y\eta_{a - a_{23}}}\,, \label{eq:3bchannel } \end{aligned}\]] where @xmath10 is the relative vector proportional to the displacement between the center of mass (c.m.) of the first cluster and that of the residual two fragments, and @xmath11 is the relative coordinate proportional to the distance between the centers of mass of cluster 2 and 3. in eq. ([eq1]), @xmath12 are the relative motion wave functions and represent the unknowns of the problem and @xmath13 is the intercluster antisymmetrizer. projecting the microscopic @xmath14-body schrdinger equation onto the basis states @xmath15, the many - body problem can be mapped onto the system of coupled - channel integral - differential equations @xmath16 g_{\nu}^{j^\pi t}(x, y) = 0,\label{eq:3beq1 } \end{aligned}\]] where @xmath17 is the total energy of the system in the c.m. frame and @xmath18 are integration kernels given respectively by the hamiltonian and overlap (or norm) matrix elements over the antisymmetrized basis states. finally, @xmath19 is the intrinsic @xmath14-body hamiltonian. in order to solve the schrdinger equations ([eq:3beq1]) we orthogonalize them and transform to the hyperspherical harmonics (hh) basis to obtain a set of non - local integral - differential equations in the hyper - radial coordinate, @xmath20 which is finally solved using the microscopic r - matrix method on a lagrange mesh. the details of the procedure can be found in @xcite. at present, we have completed the development of the formalism for the treatment of three - cluster systems formed by two separate nucleons in relative motion with respect to a nucleus of mass number a@xmath21.
Application to @xmath2he+n+n.
it is well known that @xmath1he is the lightest borromean nucleus @xcite, formed by an @xmath2he core and two halo neutrons. it is, therefore, an ideal first candidate to be studied within this approach. in the present calculations, we describe the @xmath2he core only by its g.s. wave function, ignoring its excited states. this is the only limitation in the model space used. we used similarity - renormalization - group (srg) @xcite evolved potentials obtained from the chiral n@xmath6lo nn interaction @xcite with @xmath22 = 1.5 @xmath23. the set of equations ([rgmrho]) are solved for different channels using both bound and continuum asymptotic conditions. we find only one bound state, which appears in the @xmath24 channel and corresponds to the @xmath1he ground state. [[ground - state]] ground state + + + + + + + + + + + + [tab : a] lccc approach & & e@xmath25(@xmath2he) & e@xmath25(@xmath1he) + ncsm / rgm & (@xmath26=12) & @xmath27 mev & @xmath28 mev + ncsm & (@xmath26=12) & @xmath29 mev & @xmath30 mev + ncsm & (extrapolated) & @xmath31 mev & @xmath32 mev + the results for the g.s. energy of @xmath1he within a @xmath2he(g.s.)+n+n cluster basis and @xmath26 = 12, @xmath33 = 14 mev harmonic oscillator model space are compared to ncsm calculations in table [tab : a]. at @xmath34 12 the binding energy calculations are close to convergence in both ncsm / rgm and ncsm approaches. the observed difference of approximately 1 mev is due to the excitations of the @xmath2he core, included only in the ncsm at present. therefore, it gives a measure of the polarization effects of the core. the inclusion of the excitations of the core will be achieved in a future work through the use of the no - core shell model with continuum approach (ncsmc) @xcite, which couples the present three - cluster wave functions with ncsm eigenstates of the six - body system. contrary to the ncsm, in the ncsm / rgm the @xmath2he(g.s.)+n+n wave functions present the appropriate asymptotic behavior. the main components of the radial part of the @xmath1he g.s. wave function @xmath35 can be seen in fig. ([fig:1]) for different sizes of the model space demostrating large extension of the system. in the left part of the figure, the probability distribution of the main component of the wave function is shown, featuring two characteristic peaks which correspond to the di - neutron and cigar configurations. a thorough study of the converge of the results with respect to different parameters of the calculation was presented in @xcite, showing good convergence and stability. he+@xmath36+@xmath36 relative motion wave function for the @xmath37 ground state, @xmath38 and @xmath39 are the distances between the two neutrons and between the @xmath40 particle and center of mass of the two neutrons, respectively. in the right, the three main components of the radial part of the @xmath1he g.s. wave functions @xmath35 for @xmath26=6,8,10, and 12., title="fig:",height=226] he+@xmath36+@xmath36 relative motion wave function for the @xmath37 ground state, @xmath38 and @xmath39 are the distances between the two neutrons and between the @xmath40 particle and center of mass of the two neutrons, respectively. in the right, the three main components of the radial part of the @xmath1he g.s. wave functions @xmath35 for @xmath26=6,8,10, and 12., title="fig:",height=207] [[continuum - states]] continuum states + + + + + + + + + + + + + + + + the use of three - cluster dynamics is essential for describing @xmath1he states in the continuum. therefore, this formalism is ideal for such study. using continuum asymptotic conditions, we solved the set of equations ([rgmrho]) in order to obtain the low - energy phase shifts for the @xmath41 and @xmath42 channels in the continuum. in our preliminary results, we obtain the experimentally well - known @xmath43 resonance as well as a second low - lying @xmath44 resonance recently measured at ganil @xcite. a resonance is also found in the @xmath45 channel while no low - lying resonances are present in the @xmath46 or @xmath47 channels. in fig. [fig:2] some of the preliminary phase shifts for different channels are shown. results for bigger model spaces and a study of their stability respect to the parameters in the formalism are presently being calculated and will be presented elsewhere. he for different @xmath48 channels.,height=226]
Conclusions
in this work, we present an extension of the ncsm / rgm which includes three - body dynamics in the formalism. this new feature permits us to study a new range of systems that present three - body configurations. in particular, we presented results for both bound and continuum states of @xmath1he studied within a basis of @xmath2he+n+n. the obtained wave functions feature an appropriate asymptotic behavior, contrary to bound - state @xmath0 methods such as the ncsm. computing support for this work came from the llnl institutional computing grand challenge program and from an incite award on the titan supercomputer of the oak ridge leadership computing facility (olcf) at ornl. prepared in part by llnl under contract de - ac52 - 07na27344. support from the u.s. doe / sc / np (work proposal no. scw1158) and nserc grant no. 401945 - 2011 is acknowledged. triumf receives funding via a contribution through the canadian national research council. 13 quaglioni s, navrtil p (2008) ab initio many - body calculations of n-3h, n-(4)he, p-(3,4)he, and n-(10)be scattering. phys rev lett 101:092501 quaglioni s, navrtil p (2009) ab initio many - body calculations of nucleon - nucleus scattering. phys rev c 79:044606 navrtil p, quaglioni s (2011) ab initio many - body calculations of deuteron - he-4 scattering and li-6 states. phys rev c 83:044609 navrtil p, quaglioni s (2012) ab initio many - body calculations of the h-3(d, n)he-4 and he-3(d, p)he-4 fusion reactions. phys rev lett 108:042503 quaglioni s, romero - redondo c, navrtil p (2013) three - cluster dynamics within an ab initio framework. phys rev c 88:034320 tanihata i (1996) neutron halo nuclei. j phys g 22:157 - 198 tanihata i, hamagaki h, hashimoto o, shida y, yoshikawa n, sugimoto k, yamakawa o, kobayashi t, takahashi n (1985) measurements of interaction cross sections and nuclear radii in the light p - shell region. phys rev lett 55:2676 - 2679 bogner sk, furnstahl rj, perry rj (2007) similarity renormalization group for nucleon - nucleon interac- tions. phys rev c 75:061001 roth r, reinhardt s, hergert h (2008) unitary correlation operator method and similarity renormalization group : connections and differences. phys rev c 77:064003 entem dr, machleidt r (2003) accurate charge - dependent nucleon - nucleon potential at fourth order of chiral perturbation theory. phys rev c 68:041001 baroni s, navrtil p, quaglioni s (2013) ab initio description of the exotic unbound @xmath49he nucleus. phys rev lett 110:022505 baroni s, navrtil p, quaglioni s (2013) unified ab initio approach to bound and unbound states : no - core shell model with continuum and its application to @xmath49he. phys rev c 87:034326 mougeot x, lapoux v, mittig w, alamanos n, auger f, et al (2012) new excited states in the halo nucleus he-6. phys lett b 718:441 - 446 | we introduce an extension of the @xmath0 no - core shell model / resonating group method (ncsm / rgm) in order to describe three - body cluster states.
we present results for the @xmath1he ground state within a @xmath2he+n+n cluster basis as well as first results for the phase shifts of different channels of the @xmath2he+n+n system which provide information about low - lying resonances of this nucleus.
example.eps gsave newpath 20 20 moveto 20 220 lineto 220 220 lineto 220 20 lineto closepath 2 setlinewidth gsave.4 setgray fill grestore stroke grestore | 1311.4595 |
Introduction
the cataclysmic variable hu aqr currently consists of a 0.80 white dwarf that accretes from a 0.18 main - sequence companion star. the transfer of mass in the tight @xmath12 orbit is mediate by the emission of gravitational waves and the strong magnetic field of the accreting star. since its discovery, irregularities of the observed - calculated variations have led to a range of explanations, including the presence of circum - binary planets. detailed timing analysis has eventually led to the conclusion that the cv is orbited by two planets @xcite, a 5.7 planet in a @xmath13 orbit with an eccentricity of @xmath14 and a somewhat more massive (7.6) planet in a wider @xmath15 and eccentric @xmath16 orbit @xcite. although, the two - planet configuration turned out to be dynamically unstable on a 100010,000 year time scale (*??? * see also [sect : stability]), a small fraction of the numerical simulations exhibit long term dynamical stability (for model b2 in *??? * see tab.[tab : huaqr] for the parameters). it is peculiar to find a planet orbiting a binary, in particular around a cv. while planets may be a natural consequence of the formation of binaries @xcite, planetary systems orbiting cvs could also be quite common. in particular because of recently timing residual in nn serpentis, dp leonis and qs virgo @xcite were also interpreted is being caused by circum - cv planets. although the verdict on the planets around hu aqr (and the other cvs) remains debated (tom marsh private communication, and *??? *), we here demonstrate how a planet in orbit around a cv, and in particular two planets, can constrain the cv evolution and be used to reconstruct the history of the inner binary. we will use the planets to perform a precision reconstruction of the binary history, and for the remaining paper we assume the planets to be real. because of their catastrophic evolutionary history, cvs seem to be the last place to find planets. the original binary lost probably more than half its mass in the common - envelope phase, which causes the reduction of the binary separation by more than an order of magnitude. it is hard to imagine how a planet (let alone two) can survive such turbulent past, but it could be a rather natural consequence of the evolution of cvs, and its survival offers unique diagnostics to constrain the origin and the evolution of the system.
The evolution of a cv with planets
after the birth of the binary, the primary star evolved until it overflowed it roche lobe, which initiated a common - envelope phase. the hydrogen envelope of the primary was ejected quite suddenly in this episode @xcite, and the white dwarf still bears the imprint of its progenitor : the mass and composition of the white dwarf limits the mass and evolutionary phase of its progenitor star at the moment of roche - lobe overflow (rlof). for an isolated binary the degeneracy between the donor mass at the moment of rlof (@xmath17), its radius @xmath18 and the mass of its core @xmath19 can not be broken. the presence of the inner planet in orbit around hu aqr @xcite allows us to break this degeneracy and derive the rate of mass loss in the common - envelope phase. the outer planet allows us to validate this calculation and in addition to determine the conditions under which the cv was born. the requirement that the initial binary must have been dynamically stable further constrains the masses of the two stars and their orbital separation. during the cv phase little mass is lost from the binary system @xmath20constant (but see *??? *), and the current total binary mass (@xmath21) was not affected by the past (and current) cv evolution @xcite. the observed white dwarf mass then provides an upper limit to the mass of the core of the primary star at the moment of roche - lobe contact, and therefore also provides a minimum to the companion mass via @xmath22. with the mass of the companion not being affected by the common envelope phase, we constrain the orbital parameters at the moment of rlof by calculating stellar evolution tracks to measure the core mass @xmath19 and the corresponding radius @xmath23 for stars with zero - age main - sequence mass @xmath24. in fig.[fig : amcoreformzams3msun] we present the evolution of the radius of a 3 star as a function of @xmath19, which is a measure of time we adopted the henyey stellar evolution code mesa @xcite to calculate evolutionary track of stars from @xmath25 to 8 using amuse @xcite to run mesa and determine the mass of the stellar core. the latter is measured by searching for the mass - shell in the stellar evolution code for which the relative hydrogen fraction @xmath26. at the moment of rlof the core mass is @xmath19 and the stellar radius @xmath27. via the relation for the roche radius @xcite, we can now calculate the orbital separation at the moment of rlof @xmath28 as a function of @xmath17. this separation is slightly larger than the initial (zero - age) binary separation @xmath29 due to the mass lost by the primary star since its birth @xmath30. the long (main - sequence) time scale in which this mass is lost guarantees an adiabatic response to the orbital separation, i.e. @xmath31 constant. for each @xmath24 we now have a range of possible solutions for @xmath28 as a function of @xmath19 and @xmath32. this reflects the assumption that the total mass (@xmath33) in the observed binary with mass @xmath34 is conserved throughout the evolution of the cv. in fig.[fig : amcoreformzams3msun] we present the corresponding stellar radius @xmath18 and @xmath35 as a function of @xmath19 for @xmath36. this curve for @xmath28 is interrupted when rlof would already have been initiated earlier for that particular orbital separation. we calculate this curve by first measuring the size of the donor for core mass @xmath19, and assuming that the primary fills its roche - lobe we calculate the orbital separation at which this happens. during the common envelope phase the primary s mantle is blown away beyond the orbit of the planets. the latter responds to this by migrating from the orbits in which they were born (semi - major axis @xmath37 and eccentricity @xmath38, the subscript @xmath39 indicates the inner planet, we adopt a @xmath40 to indicate the outer planet) to the currently observed orbits. using first order analysis we recognize two regimes of mass loss : fast and slow. in the latter case the orbit expands adiabatically without affecting the eccentricity : the minimum possible expansion of the planet s orbit is achieved when the common envelope is lost adiabatically. fast mass loss leads to an increase in the eccentricity as well and may even cause the planet to escape @xcite. a planet born at the shortest possible orbital separation to be dynamically stable will have @xmath41 @xcite, which is slightly smaller than the distance at which circum binary planets tend to form @xcite. in fig.[fig : amcoreformzams3msun] we present a minimum to the semi - major axis for a planet that was born at @xmath42 and migrated by the adiabatic loss of the hydrogen envelope from the primary star in the common - envelope phase. the planet can have migrated to a wider orbit, but not to an orbit smaller than the solid black curve (indicated with @xmath43) in fig.[fig : amcoreformzams3msun]. for the 3 star, presented in fig.[fig : amcoreformzams3msun], rlof can successfully result in the migration of the planet to the observed separation in hu aqr for @xmath44, which occurs for @xmath45. a core mass @xmath46 would, for a 3 primary star, result in an orbital separation that exceeds that of the inner planet in hu aqr ; in this case the core mass of the primary star must have been smaller than 0.521. another constraint on the initial binary orbit is provided by the requirement that the mass transfer in the post common - envelope binary should be stable when the companion starts to overfill its roche lobe. to guarantee stable mass transfer we require that @xmath47. the thick part of the red curve in fig.[fig : amcoreformzams3msun] indicates the valid range for the initial orbital separation and core - mass for which the observed planet can be explained ; the thin parts indicate where these criteria fail. we repeat the calculation presented in fig.[fig : amcoreformzams3msun] for a range of masses from @xmath48 to 8 with steps of 0.02, the results are presented as the shaded region in fig.[fig : am0_distribution_hu].
Effect of the common envelope phase on the planetary system
the response of the orbit of the planet to the mass loss depends on the total amount of mass lost in the common envelope and the rate at which it is lost. numerical common - envelope studies indicate that for an in - spiraling binary @xmath49 @xcite. at this rate the entire envelope @xmath505.8 is expelled well within one orbital period of the inner planet, which leads to an impulsive response and the possible loss (for @xmath51) of the planet. the fact that the hu aqr is orbited by a planet indicates that at the distance of the planet @xmath52. the eccentricity of the inner planet in hu aqr (see tab.[tab : huaqr]) can be used to further constrain the rate at which the common - envelope was lost from the planetary orbit. the higher eccentricity of the outer planet indicates a more impulsive response, which is a natural consequence of its wider orbits with the same @xmath53. this regime between adiabatic and impulsive mass loss is hard to study analytically @xcite. [cols="<,<,<,<,<,<,<,<",options="header ",] we calculate the effect of the mass loss on the orbital parameters by numerically integrating the planet orbit. the calculations are started by selecting initial conditions for the zero - age binary hu aqr @xmath24, @xmath29 and consequently @xmath19 from the available parameter space (shaded area) in fig.[fig : am0_distribution_hu], and integrate the equations of motion of the inner planet with time. planets ware assumed to be born in a circular orbit (@xmath54) in the binary plane with semi - major axis @xmath37. the equations of motions are integrated using the high - order symplectic integrator huayno @xcite via the amuse framework. during the integration we adopt a constant mass - loss rate @xmath53 applied at every 1/100th of an orbit, and we continued the calculation until the entire envelope is lost (see [sect : ce] and fig.[fig : am0_distribution_hu]), at which time we measure the final semi - major axis and eccentricity of the planetary orbit. during the integration we allow the energy error to increase up to at most @xmath55. by repeating this calculation while varying @xmath37 and @xmath56 we iterate (by bisection) until the result is within 1% of the observed @xmath57 and @xmath58 of the inner planet observed in hu aqr. the converged results of these simulations are presented in fig.[fig : am0_distribution_hu] (circles), and these represent the range of consistent values for the inner planet s orbital separation @xmath59752 as a function of @xmath608 and consistently reproduce the observed inner planet when adopting @xmath610.267/yr. the highest value for @xmath53 is reached for @xmath62 at an initial orbital separation of @xmath63. the orbital solution for the inner planet is insensitive to the semi - major axis of the zero - age binary @xmath29 (for a fixed @xmath24), and each of these solutions were tested for dynamical stability, which turned out to be the case irrespective of the initial binary semi - major axis (as discussed in [sect : stability]). we now adopt the in [sect : innerplanet] measured value of @xmath53 to integrate the orbit of the outer planet. the effect of the mass outflow on the planet is proportional to the square of the density in the wind at the location of the planet. we correct for this effect by reducing the mass loss rate in the common envelope that affects the outer planet by a factor @xmath64. we use the same integrator and assumptions about the initial orbits as in [sect : innerplanet], but we adopt the value of @xmath53 from our reconstruction of the inner planet (see [sect : innerplanet]). to reconstruct the initial orbital separation of the outer planet @xmath65, we vary this value (by bisection) until the final semi - major axis is within 1% of the observed orbit (see tab.[tab : huaqr]). the results are presented in fig.[fig : am0_distribution_hu] (triangles). the post common - envelope eccentricity of the outer planet then turn out to be @xmath66.
Stability of the initial system
after having reconstructed the initial conditions of the binary system with its two planets we test its dynamical stability by integrating the entire system numerically for 1myr using the huayno integrator @xcite. to test the stability we check the semi - major axis and eccentricity of both planets every 100years. if any of these parameters change by a factor of two compared to the initial values or if the orbits cross we declare the system unstable, otherwise they are considered stable. the calculations are repeated with the 4th order hermite predictor - corrector integrator ph4 @xcite within amuse to verify that the results are robust, which turned out to be the case. we then repeated this calculation ten times with random inital orbital phases and again with a 1% gaussian variation in the initial planetary semi - major axes. in fig.[fig : am0_distribution_hu] we present the resulting stable systems by coloring them red (circled) and blue (triangles), the unstable systems are represented by open symbols. from the wide range of possible systems that can produce hu aqr only a small range around @xmath67 turns out to be dynamically stable. the eccentricity of the outer orbit of the stable systems (which ware stable for initial conditions within 1%) @xmath68, which is somewhat smaller than the observed value for hu aqr (*??? * @xmath69). these values are obtained with @xmath70. the small uncertainty in the derived value of @xmath53 is a direct consequence of its sensitivity to @xmath38 and the small error on @xmath24 from the requirement that the initial system is dynamically stable.
Discussion and conclusions
we have adopted the suggestive results from the timing analysis of hu aqr, that the cv is orbited by two planets, to reconstruct the evolution of this complex system. a word of caution is well placed in that these observations are not confirmed, and currently under debate (tom marsh private communication, and comments by the referee). however, the predictive power that such an observation would entail is interesting. the possibility to reconstruct the initial conditions of a cv by measuring the orbital parameters of two circum binary planets is a general result that can be applied to other binaries. for cvs in particular it enables us to constrain the value of fundamental parameters in the common - envelope evolution. this in itself makes it interesting to perform this theoretical exercise, irrespective of the uncertainty in the observations. on the other hand, the consistency between the observations and the theoretical analysis give some trust to the correctness of these observations. the presence of one planet in an eccentric orbit around a cv allow us to calculate the rate at which the common - envelope was lost from the inner binary. a single planet provides insufficient information to derive the initial mass of the primary star, but allows us to derive the initial binary separation and planetary orbital separation to within about factor of 5, and the initial rate of mass loss from the common envelope to about a factor 2. a second planet can be used to further constrain these parameters to a few per cent accuracy and allows us to make a precision reconstruction of the evolution of the cv. we have used the observed two planets in orbit around the cv hu aqr to reconstruct its evolution, to derived its initial conditions (primary mass, secondary mass, orbital separation, and the orbital separations of both planets) and to measure the rate of mass lost in the common - envelope parameters @xmath53. by comparing the binary parameters at birth with those after the common - envelope phase we subsequently calculate the two parameters @xmath71 and @xmath72. the measured rate of mass loss for hu aqr of @xmath73 from the inner planetary orbit, which from the binary system itself would entail a mass - loss rate of @xmath74, when we adopt the initial binary to have a semi - major axis of @xmath75, which is bracketed by our derived range of @xmath76160. this is consistent with a mass - loss rate of @xmath77 from numerical common - envelope studies @xcite. by adopting that the binary survives its common envelope at a separation between @xmath78 (at which separation the secondary star will just fill it s roche - lobe to the white dwarf) and @xmath79 (for gravitational wave radiation to drive the binary into roche - lobe overflow within 10gyr), we derive the value of @xmath802.0 (for @xmath75 we arrive at @xmath81). this value is a bit small compared to numerous earlier studies, which tend to suggest @xmath82. the alternative @xmath72-formalism for common - envelope ejection gives a value of @xmath831.80 (for @xmath75 we arrive at @xmath84), which is consistent with the determination of @xmath72 in 30 other cvs @xcite. the inner planet in hu aqr formed at @xmath85@xmath86, with a best value of @xmath87, which is consistent with the planets found to orbit other binaries, like kepler 16 @xcite and for kepler 34 and 35 @xcite, although these systems have lower primary mass and secondary mass stars. it seems unlikely that more planets were formed inside the orbit of the inner most planet, even though currently there is sufficient parameter space for many more stable planets ; in the zero - age binary there has not been much room for forming additional planets further in. it is however possible that additional planets formed further out and those, we predict, will have even higher eccentricity than those already found. * acknowledgements * it is a pleasure to thank edward p.j. van den heuvel, tom marsh, inti pelupessy, nathan de vries, arjen van elteren and the anonymous referee for comments on the manuscript and discussions. this work was supported by the netherlands research council nwo (grants # 612.071.305 [lgm], # 639.073.803 [vici] and # 614.061.608 [amuse]) and by the netherlands research school for astronomy (nova). | cataclysmic variables (cvs) are binaries in which a compact white dwarf accretes material from a low - mass companion star.
the discovery of two planets in orbit around the cv hu aquarii opens unusual opportunities for understanding the formation and evolution of this system.
in particular the orbital parameters of the planets constrains the past and enables us to reconstruct the evolution of the system through the common - envelope phase. during this dramatic event the entire hydrogen envelope of the primary star
is ejected, passing the two planets on the way.
the observed eccentricities and orbital separations of the planets in hu aqr enable us to limit the common - envelope parameter @xmath0 or @xmath1 and measure the rate at which the common envelope is ejected, which turns out to be copious.
the mass in the common envelope is ejected from the binary system at a rate of @xmath2.
the reconstruction of the initial conditions for hu aqr indicates that the primary star had a mass of @xmath3 and a @xmath4 companion in a @xmath5160 (best value @xmath6) binary.
the two planets were born with an orbital separation of @xmath7 and @xmath8 respectively. after the common envelope, the primary star turns into a @xmath9 helium white dwarf, which subsequently accreted @xmath10 from its roche - lobe filling companion star, grinding it down to its current observed mass of @xmath11.
methods : numerical planets and satellites : dynamical evolution and stability planet star interactions planets and satellites :
formation stars : individual : hu aquarius stars : binaries : evolution | 1210.5540 |
Introduction
in recent years loop quantum cosmology (lqc) has inspired realisation of the cosmological scenario in which the initial singularity is replaced by the bounce. in this picture, the universe is initially in the contracting phase, reaches the minimal, nonzero volume and, thanks to quantum repulsion, evolves toward the expanding phase. such a scenario has been extensively studied with use of the numerical methods @xcite. however, as it was shown for example in @xcite exact solutions for bouncing universe with dust and cosmological constant can be found. the aim of the present paper is to show that analytical solutions can also be obtained for the bouncing models arising from lqc. the main advantage of such exact solutions is that they allow for investigations in whole ranges of the parameter domains. in this paper we consider the flat frw model with a free scalar field and with the cosmological constant. quantum effects are introduced in terms of correction to the classical theory. generally one considers two types of of quantum correction : correction from inverse volume and holonomy corrections. the leading effect of the volume corrections is the appearance of the super - inflationary phase. the effect of holonomy corrections, on the other hand, is the appearance of a bounce instead of singularity. the aim of this paper is to investigate analytically these effects in a flat frw model. that is to say, we neglect corrections from inverse volume, these effects however, has been extensively studied elsewhere. moreover, these two types of corrections are not equally important in the same regimes. the inverse volume corrections are mainly important for small values of the scale factor, whereas holonomy corrections are mainly important for large values of the hubble parameter. in other words, when the minimal scale factor (during the bounce) is large enough, the effects of inverse volume corrections can be neglected. the flat frw model in the loop quantum cosmology has been first investigated in the pioneer works of bojowald @xcite and later improved in the works of ashtekar, pawowski and singh @xcite. bojowald s original description of the quantum universe in currently explored in the number of works and regarded as a parallel line of research @xcite. in the present paper, we restrict ourselves to the flat frw models arising in the framework proposed by ashtekar and co - workers. beside the flat models this approach has also been applied to the frw @xmath2 models in @xcite and bianchi i in @xcite. in these models the unambiguity in the choice of the elementary area for the holonomy corrections appear. in the present paper we consider two kind of approaches to this problem : the so called @xmath3scheme and @xmath4scheme (for a more detailed description see appendix [appendix1]). we find analytical solutions for the considered models in these two schemes. the hamiltonian of the considered model is given by @xmath5 ^ 2 + \frac{1}{2 } \frac { p_{\phi}^2 } { { |p|}^{3/2 } } + { |p|}^{3/2}\frac{\lambda}{8\pi g}. \label{model}\]] in appendix [appendix1] we show the derivation of this hamiltonian in the loop quantum gravity setting. the canonical variables for the gravitational field are @xmath6 and for the scalar field @xmath7. the canonical variables for the gravitational field can be expressed in terms of the standard frw variables @xmath8. where the factor @xmath9 is called barbero - immirzi parameter and is a constant of the theory, and @xmath10 is the volume of the fiducial cell. the volume @xmath10 is just a scaling factor and can be chosen arbitrarily in the domain @xmath11. since @xmath12 is the more natural variable than @xmath13 here, we present mostly @xmath14 in the figures. @xmath13 is always the positive square root of @xmath12 so the shape of the graphs would be essentially the same when drawn with @xmath13. the equations of motions can now be derived with the use of the hamilton equation @xmath15 where the poisson bracket is defined as follows @xmath16 \nonumber \\ & + & \left[\frac{\partial f}{\partial \phi}\frac{\partial g}{\partial p_{\phi } } - \frac{\partial f}{\partial p_{\phi}}\frac{\partial g}{\partial \phi } \right]. \end{aligned}\]] from this definition we can immediately retrieve the elementary brackets @xmath17 with use of the hamiltonian ([model]) and equation ([hameq]) we obtain equations of motion for the canonical variables @xmath18 ^ 2 \right\ } \nonumber \\ & -&\text{sgn}(p)\frac{\kappa \gamma}{4 } \frac{p_{\phi}^2}{{|p|}^{5/2 } } + \text{sgn}(p) \frac{\lambda \gamma } { 2}\sqrt{|p| }, \nonumber \\ \dot{\phi } & = & { |p|}^{-3/2 } p_{\phi }, \nonumber \\ \dot{p_{\phi } } & = & 0, \label{equations}\end{aligned}\]] where @xmath19. the hamiltonian constraint @xmath20 implies @xmath21 ^ 2 = \frac{\kappa } { 3 } \frac{1}{2 } \frac{p_{\phi}^2}{{|p|}^3 } + \frac{\lambda}{3}. \label{constraint}\]] the variable @xmath22 corresponds to the dimensionless length of the edge of the elementary loop and can be written in the general form @xmath23 where @xmath24 and @xmath25 is a constant @xmath26 (this comes from the fact that @xmath27 is positively defined). the choice of @xmath28 and @xmath25 depends on the particular scheme in the holonomy corrections. in particular, boundary values correspond to the cases when @xmath22 is the physical distance (@xmath29, @xmath3scheme) and coordinate distance (@xmath30, @xmath4scheme). however, the @xmath30 case does not lead to the correct classical limit. when @xmath31, the classical limit can not be recovered either. only for negative values of @xmath28 is the classical limit @xmath32 correctly recovered @xmath33 strict motivation of the domain of the parameter @xmath28 comes form the investigation of the lattice states @xcite. the number of the lattice blocks is expressed as @xmath34 where @xmath35 is the average length of the lattice edge. this value is connected to the earlier introduced length @xmath22, namely @xmath36. during evolution the increase of the total volume is due to the increase of the spin labels on the graph edges or due to the increase of the number of vortices. in this former case the number of the lattice blocks is constant during evolution, @xmath37. otherwise, when the spin labels do not change, the number of vortices scales with the volume, @xmath38. the physical evolution correspond to something in the middle, it means the power index lies in the range @xmath39 $]. applying the definition of @xmath40 we see that the considered boundary values translate to the domain of @xmath28 introduced earlier, @xmath41 $]. more detailed investigation of the considered ambiguities can be found in the papers @xcite and in the appendix [appendix1]. combining equations ([constraint]), ([correction]) and the first one from the set of equations ([equations]) we obtain @xmath42 where new parameters are defined as follow @xmath43 equation ([mainequation]) is, in fact, a modified friedmann equation @xmath44 where the effective constants are expressed as follow @xmath45, \\ \lambda_{\text{eff } } & = & \lambda \left[1 - 2\frac{\rho}{\rho_{\text{c}}}-\frac{\lambda}{\kappa \rho_{\text{c } } } \right],\end{aligned}\]] and @xmath46 we will study the solutions of the equation ([mainequation]) for both @xmath3scheme and @xmath4scheme. the organisation of the text is the following. in section [sec : nolambda] we consider models with @xmath47. we find solutions of the equations ([mainequation]) for both @xmath3scheme and @xmath4scheme. next, in section [sec : lambda] we add to our considerations a non - vanishing cosmological constant @xmath1. we carry out an analysis similar to the case without lambda. we find analytical solutions of the equation([mainequation]) for @xmath3scheme. then, we study the behaviour of this case in @xmath4scheme. in section [sec : summary] we summarise the results.
Models with @xmath47
in this section we begin our considerations with the model without @xmath1. equation ([mainequation]) is then simplified to the form @xmath48 we solve this equation for both @xmath3scheme and @xmath4scheme in the present section. in the @xmath3scheme, as is explained in the appendix [appendix1], the @xmath22 is expressed as @xmath50 where @xmath51 is the area gap. so @xmath52 and @xmath29. to solve ([equation1]) in the considered scheme we introduce a new dependent variables @xmath53 in the form @xmath54 with use of the variable @xmath53 the equation ([equation1]) takes the form @xmath55 and has a solution in the form of a second order polynomial @xmath56 where @xmath57 is a constant of integration. we can choose now @xmath58, so that the minimum of @xmath59 occurs for @xmath60. going back to the canonical variable @xmath12 we obtain a bouncing solution @xmath61^{1/3}. \label{sol1}\]] the main property of this solutions is that @xmath14 never reaches zero value for non vanishing @xmath62. the minimal value of @xmath14 is given by @xmath63^{1/3}. \label{pmin}\]] we show the solution ([sol1]) in the fig. [solution1]. in the @xmath3scheme. the canonical variable @xmath12 is expressed in the @xmath64 $] units and time @xmath65 in the @xmath66 $] units.,width=226] the dynamical behaviour in this model is simple. for negative times we have a contracting pre - big bang universe. for @xmath67 we have a big bang evolution from minimal @xmath68 ([pmin]). it is, however, convenient to call this kind of stage big bounce rather than big bang because of initial singularity avoidance. in the @xmath4scheme the @xmath22 is expressed as @xmath70 so @xmath71 and @xmath30. to solve ([equation1]) in this scheme, we change the time @xmath72 and introduce a new variable @xmath73 as follows @xmath74 applying this new parametrisation to the equation ([equation1]) leads to the equation in the form @xmath75 which has a solution @xmath76 we can now choose @xmath77 so that the minimal value of @xmath78 occurs for @xmath79. now we can go back to the initial parameters @xmath12 and @xmath65, then @xmath80^{1/4 } \label{time1}\end{aligned}\]] introducing the variable @xmath81 we can rewrite integral ([time1]) to the simplest form @xmath82^{1/4 }, \end{aligned}\]] and the solution of such integral is given by @xmath83^{1/4 } dy & = & \frac{2}{3 } x\left(1+x^2 \right)^{1/4 } \nonumber \\ & + & \frac{1}{3 } x { _ 2f_1}\left [\frac{1}{2},\frac{3}{4 }, \frac{3}{2 } ; -x^2 \right], \end{aligned}\]] where @xmath84 is the hypergeometric function, defined as @xmath85 = \sum_{k=0}^{\infty } \frac{(a_1)_k... (a_p)_k}{(b_1)_k... (b_q)_k}\frac{x^k}{k!},\]] where @xmath86 is the pochhammer symbol defined as follow @xmath87 this solution is very similar to the one in @xmath3scheme. however, the time parametrisation is expressed in a more complex way. we show this solution in the fig. [solution2]. in the @xmath4scheme. the canonical variable @xmath12 is expressed in @xmath64 $] units and time @xmath65 in the @xmath66 $] units.,width=226] in this case minimal value of @xmath14 is expressed as @xmath88
Models with @xmath89
in this section, we investigate the general model with non vanishing cosmological constant. it will be useful to write equation ([mainequation]) in the form @xmath90\times \nonumber \\ & \times & \left [1- \frac{\omega_{\text{iii}}}{\omega_{\text{i}}}{|p|}^{2n-2 } - \frac{\omega_{\text{v}}}{\omega_{\text{ii}}}{|p|}^{2n+1 } \right]. \label{equation22}\end{aligned}\]] we see that when we perform the multiplication in this equation and define @xmath91 we recover equation ([mainequation]). in this and the next section we use equivalently @xmath92 and @xmath93 to simplify notation. in this case equation ([equation22]) can be rewritten to the form @xmath94 where the parameters are defined as @xmath95, \\ \gamma_{\text{ii } } & = & \omega_{\text{i } } - \omega_{\text{iv } } = \frac{2}{3}\kappa p^2_{\phi}\left[1 - 2\frac{\lambda}{3}\gamma^2 \xi^2 \right], \\ \gamma_{\text{iii } } & = & \omega_{\text{iii}}= \frac{1}{9 } \kappa^2 \gamma^2 \xi^2 p^4_{\phi }. \end{aligned}\]] to solve equation ([equation2]) we re - parametrise the time variable @xmath96 and introduce a new variable @xmath12 as follows @xmath97 this change of variables leads to the equation in the form @xmath98 where @xmath99 }, \\ y_2 & = & -\frac{1}{2 } \frac{\kappa p^2_{\phi}}{\lambda}.\end{aligned}\]] there are three general types of solutions corresponding to the values of the parameters @xmath100. we summarise these possibilities in the table below. @xmath101 it is important to notice that the product @xmath102 is negative in all cases. this property will be useful to solve equation ([equation3]). in the fig. [roots1] we show values of the roots @xmath103 and @xmath104 as functions of @xmath1. and @xmath104 as functions of @xmath1. region 1 corresponds to oscillatory solution, region 2 to bouncing solution. there is no physical solutions in region 3. the parameter @xmath1 is expressed in units of @xmath105$].,width=226] thus, there are two values of cosmological constants where the signs of the roots change, namely @xmath47 and @xmath106 where we have used the value of the @xmath107 calculated in the work @xcite. more recent investigation of the black hole entropy indicate however that the value of the barbero - immirzi parameter is @xmath108 @xcite. in particular, meissner has calculated @xmath109 @xcite. however, this freedom in the choice of @xmath9 does not change the qualitative results. only the region of the parameter space where the particular kind of motion is allowed can be shifted. we now perform a change of variables in equation ([equation3]) in the form @xmath110 and @xmath111 we also introduce the parameters @xmath112 and then equation ([equation3]) takes the form @xmath113 the equation ([equation4]) is the equation of shifted harmonic oscillator and its solution is @xmath114.\]] where we set the integration constant to zero. it is now possible to go back to the initial parameters @xmath12 and @xmath65 which are expressed as follows @xmath115 { \frac{6\alpha}{\lambda_{0,\bar{\mu } } } } } { \sqrt[3] { 1 - 2\frac{\lambda}{\lambda_{0,\bar{\mu } } } + \cos \left(3\sqrt{|\upsilon|}u\right) } }, \label{sol11 } \\ t(u) & = & \int_0^u du^ {'} p^3(u^ {'}), \label{sol22}\end{aligned}\]] the integral ([sol22]) can be easily solved but solutions depend on the parameters in ([sol11]). we define now an integral @xmath116 so the expression ([sol22]) can now be written as @xmath117 solutions of integral ([mainintegral1]) depend on the value of parameter @xmath13 and read @xmath118, \nonumber \\ i(x, a= 1) & = & \tan \left(\frac{x}{2 } \right), \nonumber \\ i(x,0<|a|<1) & = & -\frac{2}{\sqrt{1-a^2 } } \text{arctanh } \left [\frac{(a-1)\tan\left (\frac{x}{2 } \right)}{\sqrt{1-a^2 } } \right], \nonumber \\ i(x, a=0) & = & \ln \left (\frac{1}{\cos x } - \tan x \right). \nonumber\end{aligned}\]] in fig. [solution4] we show the solution for @xmath119 or equivalently with parameter @xmath120. . oscillating curve (green) is @xmath121 and the increasing curve (red) is @xmath122. the canonical variable @xmath12 is expressed in the @xmath64 $] units and time @xmath65 in the @xmath66 $] units.,width=226] with @xmath119. the canonical variable @xmath12 is expressed in units of @xmath64 $] and time @xmath65 in @xmath66$].,width=226] in fig. [solution5] we show the solution for @xmath123 which corresponds to @xmath124. . top curve (green) shows @xmath121 and bottom curve (red) is @xmath122. the canonical variable @xmath12 is expressed in units of @xmath64 $] and time @xmath65 in @xmath66$].,width=226] with @xmath123. the canonical variable @xmath12 is expressed in units of @xmath64 $] and time @xmath65 in @xmath66$].,width=226] in this section we study the last case in which @xmath30 and @xmath125. equation ([equation22]) can then be written in the form @xmath126 where @xmath127 and the polynomial s coefficients are expressed as @xmath128 the discriminant of polynomial ([polynomial]) is @xmath129 inserting the values of parameters @xmath130 listed above we obtain @xmath131.\end{aligned}\]] for @xmath132, or @xmath133 polynomial ([polynomial]) has three real roots. when relation ([rel]) is fulfilled and @xmath134, oscillatory solutions occur. for @xmath119 equation ([equation6]) has bouncing type solutions. this can be seen when we redefine equation ([equation6]) to the point particle form. this approach is useful for qualitative analysis and will be fully used in the next section. however, here we will use it to distinguish between different types of solutions. introducing a new time variable @xmath135 we can rewrite equation ([equation6]) to the form @xmath136 where the potential function @xmath137 equation ([hamconstr]) has the form of hamiltonian constraint for a point particle in a potential well. we see that for @xmath119 potential ([pot]) has only one extremum (for @xmath138) and only a bouncing solution is possible. for @xmath119 potential ([pot]) has a minimum for @xmath139. in this case physical solutions correspond to the condition @xmath140 and the energy of the imagined particle in the potential well is greater than the minimum of well. the particle then oscillates between the boundaries of the potential. the condition @xmath140 is equivalent to relation ([rel]) calculated from the discriminant. upon introducing the parameters @xmath141 and the variable @xmath142 @xmath143 then equation ([equation6]) takes on the form of the weierstrass equation @xmath144 detailed analysis and plots of solutions of this equation for different values of parameters can be found in the appendix to the article @xcite. the solution of equation ([weierstrass]) is the weierstrass @xmath145-function @xmath146 where @xmath147 so the parametric solution for the parameter @xmath12 is @xmath148\]] the time variable can be expressed as the integral @xmath149 in the fig. [solution9] and [solution9a] we show an exemplary parametric bounce solution with @xmath119 and a possible oscillatory solution with @xmath134. with @xmath119. the canonical variable @xmath12 is expressed in units of @xmath64 $], @xmath53 is dimensionless.,width=226] with @xmath134 and relation ([rel]) satisfied. the canonical variable @xmath12 is expressed in units of @xmath64 $] and @xmath53 is dimensionless.,width=302] in general the solution is expressible as an explicit function of time, by means of the so called abelian functions. however, we chose not to employ them here as the appropriate formulae are not as clear, and the commonly used numerical packages do not allow for their direct plotting. the fact of existence of such solutions allows us to assume that the above integral is well defined, and so is the solution itself.
Qualitative methods of differential equations in study evolutional paths
the main advantage of using qualitative methods of differential equations (dynamical systems methods) is the investigation of all solutions for all admissible initial conditions. we demonstrate that dynamics of the model can be reduced to the form of two dimensional autonomous dynamical system. in our case the phase space is @xmath150 @xmath151. first we can find the solutions corresponding to vanishing of the right hand sides of the system which are called critical points. the information about their stability and character is contained in the linearisation matrix around a given critical point. in the considered case, the dynamical system is of the newtonian type. for such a system, the characteristic equation which determines the eigenvalues of the linearisation matrix at the critical point is of the form @xmath152 where @xmath153 is a potential function and @xmath154, @xmath13 being the scale factor. as it is well known for dynamical systems of the newtonian type, only two types of critical points are admissible. if the diagram of the potential function is upper convex then eigenvalues are real and of opposite signs, and the corresponding critical point is of the saddle type. in the opposite case, if @xmath155 then the eigenvalues are purely imaginary and conjugate. the corresponding critical point is of the centre type. equation ([equation22]) can be written in the form @xmath156 where @xmath157 or equivalently as @xmath158 where we have made the following time reparametrisation @xmath159 @xmath160 now we are able to write the hamiltonian constraint in the form analogous to the particle of the unit mass moving in the one dimensional potential well @xmath161 where the potential function @xmath162 as we can see, the constant @xmath163 plays the role of the total energy of the fictitious particle. the domain admissible for motion in the configuration space is determined by the condition @xmath164. a dynamical system of the hamiltonian type has the following form @xmath165 where the prime denotes differentiation with respect to a new re - parametrised time variable which is a monotonous function of the original, cosmological time. the structure of the phase plane is organised by the number and location of critical points. in our case, critical points in the finite domain are located only on the line @xmath166 and the second coordinate is determined form the equation @xmath167 which is @xmath168 the number of critical points depends on the value of @xmath1 and @xmath28. we can distinguish two cases : * for @xmath169 : * for @xmath29 : the full analysis of the behaviour of trajectories requires investigation also at the infinity. to this aim we introduce radial coordinates on the phase plane for compactification of the plane by adjoining the circle at infinity @xmath170, @xmath171. the phase portraits for both cases are shown at figs. [fig:1] and [fig:2]. for the general case of @xmath134 we can write the parametric equation of the boundary @xmath172 of the physically admissible region in the phase space, namely @xmath173 where the dot denotes differentiation with respect to cosmological time @xmath65. this equation greatly simplifies for the special case @xmath29, and we receive the value of the hubble parameter at the boundary @xmath174 a) b) a) b) in this special case @xmath172 and @xmath30 we can integrate eq.(1) for @xmath134 and choosing the integration constant equal to zero @xmath175 which gives the maximal value o parameter @xmath12 @xmath176 in the same case with @xmath119 and choosing integration constant equal to zero we have @xmath177 for @xmath172 and @xmath178 in the case @xmath29 we obtain the de sitter solution @xmath179 these solutions represent the lines on the boundaries of the physically admissible domains.
Summary
we have studied dynamics and analytical solutions of the flat friedmann - robertson - walker cosmological model with a free scalar field and the cosmological constant, modified by the holonomy corrections of loop quantum gravity. we performed calculations in two setups called @xmath3scheme and @xmath4scheme, explained in the appendix a. we have explored whole @xmath180 range and whole allowed @xmath181 range. in the case of @xmath3scheme resulting solutions are of oscillating type for @xmath182 and bouncing type for @xmath183. for @xmath184 there are no physical solutions. in the case of @xmath4scheme for @xmath182 bouncing solutions occur. when both @xmath185 and relation ([rel]) are fulfilled, oscillatory behaviour occurs. otherwise, bouncing solutions appear. in all considered cases with @xmath186 the initial singularity is avoided. we have investigated the evolutional paths of the model, from the point of view of qualitative methods of dynamical systems of differential equations. we found that in the special case of @xmath29 the boundary trajectory @xmath187 approaches the de sitter state ; and demonstrate that in the case of positive cosmological constant there are two types of dynamical behaviours in the finite domain. for the case of @xmath30 there appear oscillating solutions without the initial and final singularities, and that they change into bouncing for @xmath188. the results of this paper can give helpful background dynamics to study variety of physical phenomenas during the bounce epoch in loop quantum cosmology. for example, the interesting question of the fluctuations like gravitational waves@xcite or scalar perturbations @xcite during this period. we tried to show that numerical calculations can be `` shifted '' one step further, since the basic model is explicitly solvable, and can be treated as starting ground for more complex problems, like the above, which can not be solved analytically. authors are grateful to orest hrycyna for discussion. this work was supported in part by the marie curie actions transfer of knowledge project cocos (contract mtkd - ct-2004 - 517186).
Flat frw model and holonomy corrections in loop quantum gravity
in this appendix, we derive the form of the hamiltonian ([model]) considered in the paper. the frw @xmath0 spacetime metric can be written as @xmath189 where @xmath190 is the lapse function and the spatial part of the metric is expressed as @xmath191 in this expression @xmath192 is fiducial metric and @xmath193 are co - triads dual to the triads @xmath194, @xmath195 where @xmath196 and @xmath197. from these triads we construct the ashtekar variables @xmath198 where @xmath199 note that the gaussian constraint implies that @xmath200 leads to the same physical results. the factor @xmath9 is called barbero - immirzi parameter. in the definition ([a]) the spin connection is defined as @xmath201}+\frac{1}{2}e^c_k e^l_a \partial_{[c}e^l_{b] }), \]] and the extrinsic curvature is defined as @xmath202,\]] which corresponds to @xmath203. the scalar constraint, in ashtekar variables, has the form @xmath204 } \right], \label{scalar}\end{aligned}\]] where field strength is expressed as @xmath205 with use of ([a]),([e]) and ([str]) the hamiltonian ([scalar]) assumes the form @xmath206 where we have assumed a gauge of @xmath207. the constant @xmath208 is the volume of the fiducial cell. this volume can be chosen arbitrarily. it is convenient to absorb the factor @xmath208 by redefinition @xmath209 the holonomy along a curve @xmath210 is defined as follows @xmath211 where @xmath212 and @xmath213 are the pauli matrices. from this definition we can calculate holonomy in the direction @xmath214 and the length @xmath215 @xmath216 where we used definition of the ashtekar variable @xmath217 ([a]). from such a particular holonomies we can construct holonomy along the closed curve @xmath218. this curve is schematically presented on the diagram below. (10,10) (3,3)(1,0)2 (5,3)(1,0)2 (7,3)(0,1)2 (7,5)(0,1)2 (7,7)(-1,0)2 (5,7)(-1,0)2 (3,7)(0,-1)2 (3,5)(0,-1)2 (4.5,2)(1,0)1 (5.5,8)(-1,0)1 (8,4.5)(0,1)1 (2,5.5)(0,-1)1 (4.5,1.2)@xmath214 (8.3,5)@xmath219 (4.5,8.5)@xmath220 (0.2,5)@xmath221 (4.5,3.4)@xmath222 (5.8,4.6)@xmath223 (4.5,6)@xmath224 (3.3,4.6)@xmath225 this holonomy can be written as @xmath226 + \mathcal{o}(\mu^3) \right] \nonumber \\ & = & \mathbb{i } + \mu^2 v_0^{2/3 } f^k_{ab } \tau_k { ^oe^a_i } { ^oe^b_j } + \mathcal{o}(\mu^3), \label{deriv}\end{aligned}\]] where we have used baker - campbell - hausdorff formula and the fact that for flat frw, field strength ([str]) simplifies to the form @xmath227. now, equation ([deriv]) can be simply rewritten to the form @xmath228}{\mu^2 v_0^{2/3 } } { ^o\omega^i_a}{^o\omega^j_b}. \label{lim}\]] the trace in this equation can be calculated with use of definition ([hol2]) @xmath229 = - \frac{\epsilon_{kij}}{2 } \sin^2\left(\mu c \right). \label{tr}\]] in loop quantum gravity the limit @xmath230 in the formula ([lim]) does not exist because of existence of the area gap. the area gap corresponds to the minimal quantum of area @xmath231, which arises as the first non - zero eigenvalue of the are operator @xcite. so instead of the limit in equation ([lim]) we should stop shrinking the loop at the appropriate minimal area @xmath232. this area corresponds to the area intersected by the loop. for the holonomy in the direction @xmath214 the area is @xmath233, as explained in the diagram below (where @xmath234). the limit @xmath239 corresponds to @xmath240. now, we must connect the area @xmath233 with the length @xmath27. we can choose that area @xmath233 to correspond to physical area @xmath241 or to the area @xmath242. so in the case @xmath243 we have the limit @xmath244 this approach we call @xmath3scheme. in the case @xmath245 where @xmath246 correspond to the eigenvalue @xmath247 and taking the limit @xmath248 which we call @xmath4scheme. in the quantum version, we can combine equations ([lim]) and ([tr]), and write @xmath249 whereas in the classical case we have @xmath250 so, from these two equations we see that quantum effects can be introduced by a replacement @xmath251 in the classical expressions. the hamiltonian ([hami]) with holonomy correction takes the form @xmath5 ^ 2.\]] p. singh, k. vandersloot and g. v. vereshchagin, phys. d * 74 * (2006) 043510 [arxiv : gr - qc/0606032]. a. ashtekar, t. pawlowski and p. singh, phys. d * 73 * (2006) 124038 [arxiv : gr - qc/0604013]. t. stachowiak and m. szydlowski, phys. b * 646 * (2007) 209 [arxiv : gr - qc/0610121]. a. ashtekar, t. pawlowski and p. singh, phys. rev. lett. * 96 * (2006) 141301 [arxiv : gr - qc/0602086]. m. bojowald, class. * 19 * (2002) 2717 [arxiv : gr - qc/0202077]. m. bojowald, class. * 20 * (2003) 2595 [arxiv : gr - qc/0303073]. a. ashtekar, t. pawlowski and p. singh, phys. d * 74 * (2006) 084003 [arxiv : gr - qc/0607039]. a. ashtekar, t. pawlowski, p. singh and k. vandersloot, phys. d * 75 * (2007) 024035 [arxiv : gr - qc/0612104]. m. bojowald, phys. d * 74 * (2007) 081301 [arxiv : gr - qc/0608100]. m. bojowald, arxiv:0801.4001 [gr - qc]. k. vandersloot, phys. d * 75 * (2007) 023523 [arxiv : gr - qc/0612070]. l. szulc, w. kaminski and j. lewandowski, class. * 24 * (2007) 2621 [arxiv : gr - qc/0612101]. l. szulc, arxiv:0707.1816 [gr - qc]. d. w. chiou, arxiv:0710.0416 [gr - qc]. d. w. chiou, arxiv : gr - qc/0703010. m. bojowald, gen. grav. * 40 *, 639 (2008) [arxiv:0705.4398 [gr - qc]]. m. p. dabrowski and t. stachowiak, annals phys. * 321 * (2006) 771 [arxiv : hep - th/0411199]. a. ashtekar, j. baez, a. corichi and k. krasnov, phys.. lett. * 80 * (1998) 904 [arxiv : gr - qc/9710007]. m. domagala and j. lewandowski, class. * 21 * (2004) 5233 [arxiv : gr - qc/0407051]. k. a. meissner, class. * 21 * (2004) 5245 [arxiv : gr - qc/0407052]. m. bojowald and g. m. hossain, arxiv:0709.2365 [gr - qc]. j. mielczarek and m. szydlowski, phys. b * 657 * (2007) 20 [arxiv:0705.4449 [gr - qc]]. j. mielczarek and m. szydlowski, arxiv:0710.2742 [gr - qc]. m. bojowald, h. h. hernandez, m. kagan, p. singh and a. skirzewski, phys. d * 74 * (2006) 123512 [arxiv : gr - qc/0609057]. m. bojowald, gen. * 38 * (2006) 1771 [arxiv : gr - qc/0609034]. a. ashtekar and j. lewandowski, class. * 14 * (1997) a55 [arxiv : gr - qc/9602046]. | in this paper we study the flat (@xmath0) cosmological frw model with holonomy corrections of loop quantum gravity. the considered universe contains a massless scalar field and the cosmological constant @xmath1.
we find analytical solutions for this model in different configurations and investigate its dynamical behaviour in the whole phase space.
we show the explicit influence of @xmath1 on the qualitative and quantitative character of solutions. even in the case of positive @xmath1
the oscillating solutions without the initial and final singularity appear as a generic case for some quantisation schemes. | 0801.0502 |
Introduction
in the standard model @xmath3 and @xmath4 are not mass eigenstates. instead we have (the small cp - violating effects are neglected) @xmath5 so the time evolution of the @xmath6 states looks like @xmath7 where @xmath8 is the mass eigenvalue and @xmath9 - the corresponding width. it follows from (1) and (2) that the probability for @xmath3 meson not to change its flavour after a time @xmath10 from the creation is @xmath11 and the probability to convert into the @xmath4 meson @xmath12where @xmath13 is the average width and @xmath14. so @xmath15 mass difference between the @xmath16 mass eigenstates defines the oscillation frequency. standard model predicts @xcite that @xmath17, @xmath18 being the cabibbo - kobayashi - maskawa matrix element. therefore the mixing in the @xmath19 meson system proceeds much more faster than in the @xmath20 system. the total probability @xmath21 that a @xmath3 will oscillate into @xmath4 is @xmath22 in the first @xmath23-mixing experiments @xcite just this time integrated mixing probability was measured. the result @xcite @xmath24 shows that in the @xmath1 system @xmath25 is expected. in fact the allowed range of @xmath26 is estimated to be between @xmath27 and @xmath28 in the standard model @xcite. such a big value of @xmath26 makes impossible time integrated measurements in the @xmath1 system, because @xmath21 in (5) saturates at @xmath29 for large values of x. although it was thought that unlike the kaon system for the @xmath16 mesons the decay width difference can be neglected @xcite, nowadays people is more inclined to believe the theoretical prediction @xcite that the @xmath30 transition, with final states common to both @xmath1 and @xmath31, can generate about 20% difference in lifetimes of the short lived and long lived @xmath1-mesons @xcite. but we can see from the (3 @xmath32 5) formulas that the effect of nonzero @xmath33 is always @xmath34 and so of the order of several percents, because @xmath35 is expected. in the following we will neglect this effect and will take @xmath36, though in some formulas @xmath33 is kept for reference reason. the development of high precision vertex detectors made it possible to measure @xcite in the @xmath23 system the time dependent asymmetry @xmath37 the same techniques can also be applied to the @xmath38 system. recently the atlas detector sensitivity to the @xmath26 parameter was studied @xcite using @xmath39 decay chain for @xmath1 meson reconstruction. it was shown that @xmath26 up to 40 should be within a reach @xcite. the signal statistics could be increased by using other decay channels, like @xmath40. the purpose of this note is to study the usefulness of the decay chain @xmath41 for @xmath1 meson reconstruction in the atlas @xmath1-mixing experiments.
Event simulation
about 20 000 following b - decays were generated using the pythia monte carlo program @xcite @xmath42 = @xmath43 = @xmath44 + @xmath45 = @xmath46 + @xmath47 + @xmath45 = @xmath48 + @xmath49 the impact parameter was smeared using the following parameterized description of the impact parameter resolution @xmath50 where resolutions are in @xmath51 and @xmath52 is the angle with respect to the beam line. it was shown in @xcite that this parameterized resolution reasonably reproduces the results obtained by using the full simulation and reconstruction programs. for the transverse momentum resolution an usual expression @xcite @xmath53 was assumed. track reconstruction efficiencies for various particles were taken from @xcite. because now we have 6 particles in the final state instead of 4 for the @xmath2 decay channel, we expect some loss in statistics due to track reconstruction inefficiencies, but the effect is not significant because the investigation in @xcite indicates a high reconstruction efficiency of 0.95.
Event reconstruction
the topology of a considered @xmath19 decay chain is shown schematically in a figure : (150,110) (40,40) (40,40)(-3,-2)30 (12,17)*@xmath54 * (40,40)(2,1)40 (62,45)*@xmath55 * (80,60)(3,-1)15 (83,52)*@xmath43 * (95,55)(3,1)33 (130,65)*@xmath56 * (95,55)(3,0)35 (133,55)*@xmath57 * (95,55)(3,-2)35 (130,35)*@xmath58 * (80,60)(1,3)10 (85,90)*@xmath59 * (80,60)(2,3)20 (102,90)*@xmath58 * (80,60)(3,2)30 (112,80)*@xmath59 * the @xmath1 decay vertex reconstruction was done in the following three steps. first of all the @xmath60 was reconstructed by finding three charged particles presumably originated from the @xmath60 decay and fitting their tracks. for this goal all combinations of the properly charged particles were examined in the generated events, assuming that two of them are kaons and one is pion. the resulting invariant mass distribution is shown in fig.1a for signal events. the expected @xmath60 peak is clearly seen along with moderate enough combinatorial background. cuts on @xmath61 , @xmath62 and @xmath63 were selected in order to optimize signal to background ratio. to select one more cut on @xmath64, the information about the invariant mass resolution is desirable. 2a shows the reconstructed @xmath60 meson from its true decay products. the finite invariant mass resolution is due to applied track smearing and equals approximately to @xmath65. after @xmath60 meson reconstruction, @xmath66 meson was searched in three particle combinations from the remaining charged particles, each particle in the combination being assumed to be a pion. 1b shows a resulting invariant mass distribution for signal events. because of huge width of @xmath66, signal to background separation is not so obvious in this case. if @xmath66 is reconstructed from its true decay products as in fig. 2b, its width is correctly reproduced. to draw out @xmath66 from the background, further cuts were applied on @xmath67 , @xmath68 , @xmath69 and @xmath70. at last @xmath19 decay vertex was fitted, using reconstructed @xmath60 and @xmath66. almost the same resolution in the @xmath1-decay proper time was reached @xmath71, as in @xcite. the corresponding resolution in the b - meson decay length in the transverse plane is @xmath72. the relevant distributions are shown in fig.3.
Signal and background
branching ratios and signal statistics for the @xmath73 channel are summarized in table 1. note that we use an updated value for br(@xmath74) from @xcite. @xmath19 branching ratios are still unknown experimentally. neglecting su(3) unitary symmetry breaking effects, we have taken br(@xmath73)@xmath75 br(@xmath76). + branching ratios and signal statistics for @xmath77. + [cols="<,^,<",options="header ",] as we see, about 2065 reconstructed @xmath19 are expected after one year run at @xmath78 luminosity. the corresponding number of events within one standard deviation (@xmath79) from the @xmath19 mass equals 1407. this last number should be compared to 2650 signal events, as reported in @xcite, when @xmath80 decay channel is used. events which pass the first level muon trigger (latexmath:[$p_t > 6~ gev / c,~ background can come from other @xmath16 decays of the same or higher charged multiplicity, and from random combinations with some (or all) particles originating not from a @xmath16 decay (combinatorial background). the following channels were considered and no significant contributions were found to the background : * @xmath82. these events do nt pass the analysis cuts, because the @xmath83 mass is shifted from the @xmath60 mass by about 100 mev, and so does the @xmath20 mass compared to the @xmath19 mass. * @xmath84 followed by @xmath85. taking @xmath86 from @xcite, we see that the expected number of @xmath87 events, originated from this source, is only five times less than the expected number of truly signal events. but the decay topology for this decay chain is drastically different (1 + 5, not 3 + 3) and therefore it is unexpected that significant amount of the b - decays will be simulated in this way. + note that even for @xmath88 decay channel the similar background is negligible @xcite, although @xmath89 is about 44 times bigger than @xmath90. * @xmath91. about 10 000 such events were generated by pythia and then analyzed. using br(@xmath92 from @xcite and assuming that @xmath93 decay goes through @xmath94 oscillations : @xmath95, and therefore @xmath96, we have got fig.4. it is seen from this figure that because of @xmath97 mass shift, the contribution of this channel to the background proves to be negligible. + note that fig.4 refers to the total number of the @xmath91 events. in fact the distribution of these events with regard to the decay proper time is oscillatory, @xmath98 (not @xmath26) defining the oscillation frequency. so in general this will result in oscillatory dilution factor. the conclusion that this dilution factor is irrelevent relies on the fact that no candidate event was found with invariant mass within one standard deviation from the @xmath1 mass for @xmath99 integrated luminosity. a huge monte - carlo statistics is needed for combinatorial background studies. no candidate event with @xmath100 was found within @xmath101 inclusive @xmath102 events. this indicates that signal / background ratio is expected to be not worse than 1:1.
Dilution factors
the observation of the @xmath103 oscillations is complicated by some dilution factors. first of all the decay proper time is measured with some accuracy @xmath104. from previous discussions we know that in our case @xmath105 is expected. due to this finite time resolution, the observed oscillations are convolutions of the expressions (3) and (4) given above with a gaussian distribution. for example @xmath106 } \frac{ds}{\sqrt{2\pi}\sigma } \sim \nonumber \\ & & \frac{1}{2 } e^{-\frac{\gamma t}{\hbar}}\left (\cosh{\frac{\delta \gamma } { 2\hbar}}(t-\sigma \frac{\sigma}{\tau})-d_{time}\cos { \frac{\delta m}{\hbar}(t-\sigma \frac{\sigma}{\tau }) } \right) \ ; \ ;, \label{eq9 } \end{aligned}\]] where @xmath107 }, ~\tau = \frac{\hbar}{\gamma}$]. so the main effect of this smearing is the reduction of the oscillation amplitude by @xmath108. this is quite important in the @xmath1 system where @xmath109. there is also a time shift @xmath110 in (9). this time shift does not really effect the observability of the oscillations and we will neglect it. in fact (9) is valid only for not too short decay times @xmath111, because in (3) and (4) distributions @xmath112 is assumed. another reduction in the oscillation amplitude is caused by the particle/ antiparticle mistagging at t=0. in our case particle / antiparticle nature of the @xmath16 meson is tagged by the lepton charge in the semileptonic decay of the associated beauty hadron. mistagging is mainly due to * @xmath113 oscillations : accompanying b - quark can be hadronized as a neutral @xmath16 meson and oscillate into @xmath114 before semileptonic decay. * @xmath115 cascade process, then the lepton is misidentified as having come directly from the @xmath16-meson and associated to the @xmath116 decay. * leptons coming from other decaying particles (k,@xmath117,...). * detector error in the lepton charge identification. let @xmath118 be the mistagging probability. if we have tagged @xmath119 @xmath3 mesons, among them only @xmath120 are indeed @xmath3-s and @xmath121 are @xmath4-s misidentified as @xmath3-s. so at the proper time @xmath10 we would observe (@xmath122 due to cpt invariance) @xmath123= \frac{n}{2}e^{-\frac{\gamma t}{\hbar } } \left [\cosh{\frac{\delta \gamma t}{2\hbar}}-(1 - 2\eta) \cos { \frac{\delta m \, t}{\hbar } } \right] \nonumber \end{aligned}\]] decays associated to the @xmath4 meson and therefore @xmath124 \ ; \ ; \ ;.\]] so the dilution factor due to mistagging is @xmath125. in our studies we have taken @xmath126, as in @xcite. finally the dilution can emerge from background. suppose that apart from @xmath127 events with @xmath128 oscillations we also have @xmath129 additional background events. half of them will simulate @xmath114 meson and half of them b meson (assuming asymmetry free background). so the observed number of would be @xmath128 oscillations will be @xmath130 and the oscillation amplitude will be reduced by an amount @xmath131 neglecting the proper time dependence of this dilution factor (that is supposing that the background is mainly due to @xmath16-hadron decays and therefore has approximately the same proper time exponential decay as the signal @xcite),we have taken @xmath132 which corresponds to the 2:1 signal / background ratio.
Prospects for @xmath26 measurements
for @xmath133 integrated luminosity the number of reconstructed @xmath19-s would reach @xmath134 from the analyzed channel alone. another @xmath135-s are expected from the @xmath136 channel @xcite. for events in which @xmath19 meson does not oscillate before its decay, the @xmath137 meson and the tagging muon have equal sign charges. if the @xmath19 meson oscillates, opposite charge combination emerges. the corresponding decay time distributions are @xmath138 d is the product of all dilution factors and @xmath119 is the total number of reconstructed @xmath19-s. the unification of samples from @xmath73 and @xmath80 decay channels allows to increase @xmath26 measurement precision. fig.7 and fig.8 show the corresponding @xmath139 asymmetry plots for @xmath140 and 35.
Conclusions
it seems to us that @xmath141 decay channel is almost as good for the @xmath1-mixing exploration as previously studied @xmath142 and enables us to increase signal statistics about 1.5 times. further gain in signal statistics can be reached @xcite by using @xmath143 decay mode and considering other decay channels of @xmath60. these possibilities are under study. we refrain from giving any particular value of @xmath26 as an attainable upper limit. too many uncertainties are left before a real experiment will start. note, for example, that about two times bigger branching ratios for both @xmath144 and @xmath145 decay channels are predicted in @xcite. @xmath146 as a @xmath147 production cross section can also have significant variation in real life @xcite. so although the results of this investigation strengthen confidence in reaching @xmath26 as high as 40 @xcite, it should be realized that some theoretical predictions about @xmath1-physics and collider operation were involved and according to t.d.lees first law of physicist @xcite `` without experimentalist, theorist tend to drift ''. however maybe it is worthwhile to recall his second law also `` without theorist, experimentalists tend to falter ''.
Acknowledgements
many suggestions of p.eerola strongly influenced this investigation and lead to considerable improvement of the paper. communications with s.gadomski and n.ellis are also appreciated. authors thank n.v. makhaldiani for drawing their attention to t.d.lees paper. 99 a. ali, d. london, j. phys. * g19 * (1993), 1069. see for example : ua1 coll. albajar et al., phys. lett. * b186 * (1987), 247 ; (1991), 171. cleo coll., j. bartelt et al., phys. * 71 * (1993), 1680. argus coll., h. albrecht et al., z. phys. * c55 * (1992), 357. aleph coll., d. buskulic et al., phys. lett. * 284 * (1992), 177. opal coll. acton et al., phys. lett. * b276 * (1992), 379., b. adeva et al., phys. lett. * b288 * (1992), 395. delphi coll., p. abreu et al., phys. lett. * b332 * (1994), 488. moser, b - mixing. talk given at the @xmath148 international symposium on heavy flavour physics, montreal, canada, 1993. cern - ppe/93 - 164. a. ali, d. london, cp violation and flavour mixing in the standard model, desy-95 - 148 (hep - ph/9508272). a. ali, d. london, implications of the top quark mass measurement for the ckm parameters, @xmath26 and cp asymmetries. cern - th.7398/94 (hep - ph/9408332). buras, w. slominsski, h. steger, nucl. b245 * (1984), 369. franzini, phys. * 173 * (1989), 1. voloshin et al., * 46 * (1987), 181. a. datta, e.a. paschos, u. trke, phys. b196 (1987), 382. bigi, lifetimes of heavy - flavour hadrons - whence and whither? und - hep-95-big06 (hep - ph/9507364). i. dunietz, @xmath38 mixing, cp violation and extraction of ckm phases from untagged @xmath1 data samples. fermilab - pub-94/361-t (hep - ph/9501287). aleph coll., d. decamp et al., phys. b313 * (1993), 498. aleph coll., d. buskulic et al., cern - ppe/93 - 99. opal coll., r. akers et al., cern - ppe/94 - 43. p. eerola, s. gadomski, b. murray, @xmath19-mixing measurement in atlas, atlas internal note phys - no-039, 1994. atlas technical proposal, cern / lhcc/94 - 43, 1994. t.sjstrand, pythia 5.7 and jetset 7.4 : physics and manual, cern - th.7112/93, 1993. review of particle properties, phys. rev. * d50 *, 1994. s. rudaz, m.b. voloshin, phys. lett. * b252 * (1990), 443. p. eerola et al., asymmetries in @xmath16 decays and their experimental control, atlas internal note phys - no-054, 1994. p. camarri, a. nisati, time - dependent analysis of cp - asymmetries in the @xmath149 system, atlas internal note phys - no-065, 1995. p. blasii, p. colangelo, g. nandulli, phys. b283(1992), 434. p. eerola, measurement of cp - violation in b - decays with the atlas experiment, atlas internal note, phys - no-009, 1992. t. d. lee, the evolution of weak interactions, talk given at the symposium dedicated to jack steinberger, geneva, 1986, cern 86 - 07. | the usefulness of the @xmath0 decay chain is investigated for the @xmath1-reconstruction in the future atlas @xmath1-mixing experiment.
it is shown that this decay channel is almost as suitable for this purpose as previously studied @xmath2.
-10 mm | hep-ph9510416 |
Introduction
a diffusive electron - electron interaction correction (eec) to the conductivity was predicted theoretically @xcite about 30 years ago. in 2d system it is proportional to @xmath5 (@xmath6 is the momentum relaxation time) and grows in amplitude as temperature decreases. a way to experimentally single - out eec among other numerous effects is based on its property not to affect hall component of magnetoconductivity tensor @xmath7 in perpendicular magnetic field @xcite. eec therefore gives birth to temperature - dependent and parabolic with field contribution to the diagonal magnetoresistance @xmath8 and correction to the hall coefficient @xmath9, both being proportional to @xmath5. the predicted features were observed in numerous experiments, mainly with n - type gaas - based 2d systems @xcite. however, the quantitative level of agreement between theory and experiment was achived only in the 2000s by minkov group @xcite from simultaneous analysis of both hall and diagonal components of resistivity tensor. the suggested method was later approbated by others @xcite. we note, that zeeman splitting effects were negligible in most of the studied systems. zeeman splitting was predicted to decrease the eec value @xcite, the physical interpretation of this effect introduced later in ref. @xcite consists in decreasing the effective number of triplet channels with field. for the diffusive regime @xmath10, the eec is predicted to be quadratic - in - field in the low field limit, and proportional to logarithm of field in the high field limit. experimentally, however, the effect of zeeman splitting on eec in the diffusive regime was only briefly considered in refs. @xcite. for the most ubiquitous 2d system known, 2deg in si - mosfet, which fits well all theory requirements, no convincing measurements of the eec have been done so far. at the same time this system demonstrates positive magnetoresistance in parallel field, the behavior expected for eec. in the 1980-s there were attempts to reveal eec in si from temperature and magnetic field dependences @xcite of resistivity ; these attempts were based on not yet developed theoretical concepts and did not lead to a self - consistent picture of magnetotransport. interest to zeeman splitting effects was resumed in 1997 with observation of a huge rise in resistivity of 2deg in clean si - mosfets in parallel magnetic field @xcite, close to metal - to - insulator transition. the interest further increased with interpretation of this magnetoresistance as a signature of magnetic quantum phase transition @xcite. in the 2000s several attempts to treat the parallel field magnetoresistance (mr) in terms of renormalization - group approach were taken both theoretically @xcite and experimentally @xcite. this approach is in fact self - consistent generalization of the eec for arbitrary interaction strength and conduction. independently, another theoretical approach was developed in refs. @xcite, and successfully applied @xcite, which accounts for resistivity increase with field simply by renormalization of the density of states and single impurity scattering time. the latter effect is essentially different from logarithmic eec which emerges from multiple electron - impurity scattering. the experimental situation, however, is more complicated : studies @xcite showed a strong effect of disorder on the parallel field magnetoresistance, that was discussed in terms of the band tail effects in refs. moreover, detailed studies of the mr on different material systems @xcite did demonstrate quantitative disagreement between the fitted - to - eec theory temperature- and magnetic field dependences of the conductivity. to summarize the present state of the field, there seems to be a common agreement on the zeeman nature of parallel field mr in 2d carrier systems. however, two conceptually different underlining mechanisms of mr were put forward : (i) eec (multiple - scattering effect) and (ii) screening change in magnetic field (single - scattering effect). which of them is responsible for the experimentally observed strong mr in parallel field? the answer is especially crucial in the vicinity of the metal - to - insulator transition, where the mr is dramatically strong. unfortunately, both theories become inapplicable in this regime of small conductances @xmath11. to address this issue, we have chosen to approach the problem from the large conductance regime, where both theories have solid ground, though the mr is low. in our paper we contest possible origin of the parallel field magnetoresistance of weakly interacting 2d electron gas. in order to study eec, we take detailed measurements of the magnetoresistance tensor in tilted field, and analyze the data using the procedure developed in refs. we stress that our approach does not rely on any particular microscopic theory, rather, it is _ ab - initio _ phenomenological and uses only general property of the eec in the diffusive regime to affect @xmath12 solely. for the experiments we have chosen the simplest model system, the 2d electron gas in si in diffusive regime @xmath13, @xmath14. to vary the strength of the zeeman splitting, and thus, the eec magnitude, we tilted magnetic field with respect to the 2d plane. this procedure allowed us to extract eec on top of other magnetoresistivity effects and to establish two principally different regions : (i) high - field region, where eec depends on total field and quantitatively agrees with the theoretically predicted @xmath15 asymptotics, and (ii) low - field region, where eec unexpectedly depends on perpendicular field component, grows with field and does not match existing theories. our observations suggest a new insight on the origin of the parallel field mr : (i) the high - field small and @xmath16independent eec @xmath17, revealed in the current study, can not be responsible for large and @xmath18-dependent parallel field mr, and (ii) application of even low @xmath19 perpendicular field component strongly suppresses mr. the latter fact proves not purely spin nature of the parallel field mr and points at the incompleteness of the existing theory of mr. the paper is organized as follows : after a brief description of experimental details in section ii, we give theoretical background in section iii, and describe the results in section iv, first on the eec, then on the experimental proof of the zeeman origin of the magnetic field effect on eec, and further we compare the known eec with the parallel field magnetoresistance to show that the eec is not the main origin of the parallel field magnetoresistance. in section v we discuss the obtained result and suggest possible directions of further development.
Experimental details
the ac - measurements (13 to 73hz) of the resistivity were performed at temperatures 0.3 - 25k in magnetic fields up to 15 t with two (100) si - mos samples with 200 nm oxide thickness : si-40, (peak mobility @xmath20m@xmath21/vs at @xmath22 k), si-24 (0.22m@xmath23/vs). the samples were lithographically defined as rectangular hall bars of the size 0.8@xmath245 mm@xmath21. to obtain the data for different orientations of magnetic field relative to the 2d plane, the sample platform was rotated _ in situ _ at low temperature using a step motor. because of the smallness of the studied effects and unavoidable misalignment of potential contacts there always was some asymmetry (within less than a few percents) of both @xmath25 and @xmath26 with respect to field reversal at a constant @xmath27 (@xmath28 na) current direction. to compensate this asymmetry the field was swept from positive to negative values and the data were symmetrized. the alignment of the sample parallel to magnetic field was done using the resistivity peak due to weak localization. the carrier density @xmath29 was varied by the gate voltage in the range @xmath30 @xmath31. the field was tilted in the @xmath32-plane, always perpendicular to the current direction, which is however not crucial for si because of weak spin - orbit coupling @xcite. we have selected samples with moderate carrier mobility in order to ensure studies deep in the diffusive regime @xmath33, where eec theory should be applicable, and at the same time to achieve @xmath34 in available magnetic field. other features which make si - mos system preferable for this study are (i) short - range and uncorrelated scatterers which make motion diffusive, (ii) large conductance @xmath35, which ensure quantum correction theory applicable, (iii) very thin potential well @xmath36 nm which excludes orbital effects in parallel magnetic fields up to 20 t, and (iv) filling of the lowest size quantization subband solely for @xmath37@xmath31 with effective mass @xmath38 @xcite.
Theoretical background
the idea of the present study is based on a property of the e - e correction to affect only diagonal component of the conductivity tensor @xcite : @xmath39 correspondingly, the procedure of the eec extraction @xcite for arbitrary (@xmath40) value includes the following steps : (i) reversing the measured resistivity tensor one calculates the conductivity tensor. (ii) from @xmath7 one finds the mobility @xmath41 value, using experimentally determined density @xmath29 @xcite. (iii) subtracting @xmath42 from @xmath12 one finds the correction @xmath43. the mobility @xmath44 value is a by - product of this algorithm. if the conductivity is large, @xmath45, the above algorithm leads to the following expressions for the resistivity components : @xmath46 \label{aamrtensor}\]] @xmath47 as we checked, these formulae are valid for all our data with @xmath48 kohm. in practice, both @xmath49 and mobility @xmath41 in eqs. ([aamctensor],[aamrtensor],[simpleds]) might depend on @xmath50 and @xmath18 thus incorporating other possible magnetoresistance effects. nevertheless, as follows from eq.([simpleds]), no matter how the mobility depends on field, the correction to the hall coefficient arises from eec solely. depending on the geometry of experiment the following cases are possible : (i) the field is directed normally to the sample plane, @xmath51, during the sweep (this case refers to most of the previous studies). (ii) the field is inclined by angle @xmath52 relative to the 2d plane, in order to enhance zeeman effects which depend on @xmath53. (iii) the field magnitude @xmath54 remains constant while sample is rotated relative to the field direction ; in the latter case one expects the zeeman effects to remain unchanged, thus leading to the angle - independent @xmath49. we note here that in the case of geometry (ii) data processing requires knowledge of the ratio @xmath55 solely, rather than the tilt angle. according to our definition of density @xcite, the @xmath55 ratio is obtained from the measured linear - in - total - field high temperature limit of the hall resistance @xmath56. we find this limit by linear extrapolation of @xmath57 to the highest temperatures (typically 15k), and by ignoring low - field effects (see below). making use of geometries (ii) and (iii) is the key feature of the present study. we note here that the @xmath3-factor in si is large (@xmath58) and isotropic, which insures the zeeman effects to be tilt independent. in systems with anisotropic @xmath3-factor, like holes in gaas @xcite one should take into account different components of the @xmath3-factor tensor, which complicates the problem. according to the theory of interaction corrections @xcite in its modern form @xcite, the eec value at zero magnetic field in the diffusive regime is @xcite. @xmath59 where @xmath60 is a fermi - liquid constant, @xmath61 is the number of triplet channels of interaction, @xmath62- valley degeneracy (in the original formula @xcite @xmath63). in (001) si - mosfets, electron system is 2-fold valley degenerate, and the degeneracy, if perfect, should increase @xmath64 to 15 @xcite. in fact, however, this degeneracy is never perfect, because of the two sample - dependent parameters, a finite valley splitting @xmath65 @xcite, and intervalley scattering time @xmath66 k@xmath67 for the sample si-40 @xcite. both effects decrease the number of triplet terms, which was described theoretically @xcite and studied experimentally @xcite. the problem with mosfets is that @xmath65 (which is typically less than @xmath68) can hardly be measured directly in zero and low magnetic fields, because valley splitting of shubnikov - de haas oscillations does not exceed level broadening and hence can not be resolved @xcite. one could treat @xmath65 as an additional free parameter, which vary in the range from 0 to @xmath68, strongly affecting predictions of the e - e interaction correction theory @xcite. within the present study, uncertain @xmath65 value affects only the effective number of valleys @xmath62 that can vary from @xmath69 to @xmath70. we therefore can use @xmath62 as adjustable parameter, which quantifies effective degree of valley multiplicity. in 2001, the eec was recalculated by zala et al @xcite and a new `` ballistic '' contribution was introduced, that gave explanation to @xmath71 dependence observed in different 2d systems in the regime @xmath72 @xcite. it was shown experimentally @xcite that the ballistic and diffusive corrections differ fundamentally : diffusive eec does not affect hall component of conductivity tensor (see eq. [aamctensor]), whereas ballistic contribution is basically renormalization of the single impurity scattering time or mobility. in the present study we do not consider ballistic contribution, rather, we extract from the experimental data and analyse the diffusive part of the eec solely. theoretical prediction for the zeeman splitting dependence of the eec was first given in @xcite : @xmath73 where @xmath74, and the two asymptotics for @xmath75 function are : @xmath76 @xmath77 within the same theoretical formalism, the zeeman splitting effect on ballistic and diffusive corrections to magnetoresistance in parallel field was recalculated in ref. @xcite. although the theory was successfully used to fit some data on parallel field magnetoresistance @xcite, the procedure of the comparison with theory is ill - defined and requires careful separation of the diffusive and ballistic eec contributions in the crossover regime. indeed, theoretical eec is a sum of two contributions with absolutely different structure : (i) a ballistic one which comes from the renormalization of single impurity scattering and (ii) diffusive eec for which @xmath78. since our method catches only diffusive part, we briefly discuss below modern theoretical expressions in the diffusive regime @xmath13 solely @xmath79 this low field limit is close to @xcite for small @xmath80. in the high field limit, the diffusive contribution from ref. @xcite is given by : @xmath81 noteworthy, functional dependence on @xmath60 is the same for eqs. ([znamaineq]) and ([znabpar1]). this fact has a transparent physical meaning : application of high field @xmath82 suppresses temperature dependence of only @xmath83 triplets with @xmath84. this suppresion comes from expansion of the @xmath85. correspondingly, if we define @xmath86 we may rewrite the theoretical expectation for the low temperature (@xmath87) high field (@xmath88) asymptotics : @xmath89 \label{lowtbasympt}\]] here @xmath90 is a @xmath16 and @xmath91 independent term. this expression allows one to compare experimental data on low - temperature high - field asymptotics of the eec with the microscopic theory predictions. it s meaning is as follows : in the high field limit @xmath82, the magnetic field dependence is not affected by temperature. in appendix a we show that from the practical point of view this limit is achieved already for @xmath92. to conclude this section, studying magnetoresistance in purely parallel field is insufficient to disentangle the ballistic and eec contributions. tilting the field enables one to overcome this drawback. as shown in the next section, application of high tilted fields allows us to achieve in experiment the asymptotical @xmath93 behavior, for which there is a firm theory prediction.
Results
this section is organized as follows : in the first subsection we discuss phenomenology, low - field regime and experimental proof of the eec dependence on modulus of magnetic field rather than on its direction. in the second subsection we consider high - field asymptotics of the eec, and compare them with the fermi - liquid theory expectations. in the third subsection we discuss magnetoresistance in parallel and tilted fields. we explore mr in the range of fields @xmath94, i.e. in the domain between weak localization, @xmath95), and shubnikov - de haas oscillations (here @xmath96 is the transport mean free path). figure [perpfield]a shows magnetoresistance @xmath0 and hall resistance @xmath1 versus @xmath97 in field up to @xmath98 t. the data were collected at various temperatures (0.6 - 4.2k) in a standard geometry (magnetic field perpendicular to the sample plane) for sample si-40 at @xmath99@xmath31. the set of curves for si [fig. [perpfield](a)] does not look similar to numerous data for n - gaas 2d systems @xcite. indeed, there are two important features : (i)unlike gaas, for si in high fields (@xmath100 t in fig. 1(a)) the mr gets weaker and ultimately even changes sign to positive as temperature decreases ; (ii) the shubnikov - de haas oscillations start to appear already in fields @xmath101, due to short - range disorder. because of the above features, the eec is not seen straightforwardly from the data. therefore, in order to extract it we use the procedure described in the previous section. at the first step we invert the resistivity tensor and obtain the conductivity one (solid lines in figs. [perpfield]c, [perpfield]e). hall component of the conductivity tensor allows one to calculate mobility @xmath41 using eq. ([aamctensor]) (shown in fig. [perpfield]d). surprisingly, the mobility appears to be field - dependent (unlike that in experiments with gaas @xcite) : in strong fields (@xmath102 t), the higher is the temperature the stronger is the field dependence. this effect is beyond the scope of the present study, we note only that the sign of the @xmath103-dependence is in - line with expectations from memory effects @xcite though such effects do not produce temperature dependence. field dependent mobility was suggested to arise also from ballistic correction @xmath104 @xcite. even though such correction looks qualitatively similar to our data, it is hardly relevant, because originates from multiple cyclotron returns and should be valid for @xmath105, which is not the case. in weak fields, the hall mobility increases as @xmath50 decreases due to increase of the hall slope. having the mobility known, we calculate the drude part of @xmath12, @xmath106 [shown by dotted lines in fig. [perpfield](e)]. correction to conductivity @xmath49 is calculated as a difference between @xmath12 and its drude expectation (see inset to fig. [perpfield]e for the graphical definition). the resultant @xmath107 is shown by solid lines in fig. [perpfield]f. as it is clear from eq. ([simpleds]), the correction resembles behavior of the hall coefficient @xmath9 (shown in fig. [perpfield]f) : the larger the hall coefficient, the less the correction. all @xmath108 dependences (collected at different temperatures, densities, tilt angles and for different samples) manifest similar behavior : in low fields (region a in fig. [perpfield]g) @xmath49 grows as @xmath50 increases, then reaches a maximum and decreases in the high field region b. in the low field region a, the feature in @xmath49 originates from non - linearity of the hall resistance with field (as seen from fig. [perpfield]f). similar low - field feature was observed in numerous previous studies with various 2d systems @xcite ; it is still poorly understood. in appendix b we summarize our observations on the low - field hall nonlinearity and argue that this low - field feature does not follow contemporary theories. empirically, the boundary between the regions a and b (a point where @xmath49 is maximal) roughly follows the equation @xmath109. the correction decreasing with increasing field in the region b fig. [perpfield]g is qualitatively in line with zeeman splitting effect @xcite. the latter should be direction independent, as discussed above. to check this property we set the tilt angle to 45@xmath110, measured both @xmath0 and @xmath1 for the same temperatures and recalculated the eec. the resultant eec versus _ total _ magnetic field is shown in fig. [perpfield]g by dashed lines. in the high - field region b, the monotonic parts of the eec data (ignoring shubnikov - de haas oscillations) in perpendicular and tilted field are quantitatively similar to each other, even though the perpendicular component of the field differs by a factor of @xmath111. this observation proves the zeeman nature of the eec in the high - field region. another instructive way to check whether @xmath49 depends on the _ total _ field is to rotate the sample in constant field. during these measurements, the tilt angle @xmath52 was slowly swept at a constant rate using a step motor. we used positions of the specific sharp mirror symmetric features in the @xmath112 dependence at @xmath113, @xmath114, @xmath115 to calibrate angle and calculate perpendicular component of the field. figure [spining] shows @xmath116 and @xmath117 for the same electron density and different temperatures. the inset to fig. [spining] shows the corresponding eec. it is easy to see that gradual changes of @xmath49 with @xmath118, which were seen in fig. [perpfield]g almost disappeared in the inset to fig. 2 ; this demonstrates that eec remains unchanged in constant total field. we stress the importance of this `` rotating field '' experiment ; indeed, the coincidence of @xmath107 curves in fig. 1f for different tilt angles is not a complete proof for direction - independence of the eec, because it does not exclude corrections to conductivity @xmath119 (see e.g. ref.@xcite). in the latter case one would see just a constant shift @xmath120 in tilted field sweep experiment (fig. 1f) with no visible change in functional form of the @xmath121 dependence. in contrast, in the rotating field experiment such corrections would reveal themselves as inclined @xmath122-lines, which is not the case in the insert to fig. the discussed above field direction independence of the eec was tested with two samples in the density range @xmath123 @xmath31 which corresponds to @xmath124 @xmath125, and in the field range @xmath126. the data in the high field region b (@xmath127) should follow the asymptotics of eq.([lowtbasympt]). the field range for observing the logarithmic - in - field behavior, however, is limited on the low - field side by low - field hall feature (at @xmath128 t) and, on the high field side, by the onset of shubnikov - de haas oscillations (at @xmath129 t). for this reason, the field range for fitting the data with @xmath4-dependence is less than one decade. we performed detailed measurements of magnetoresistivity tensor at different tilt angles in magnetic fields up to 15 t. the high - field asymptotes were fitted with a function @xmath130 where @xmath131 and @xmath132 are two positive adjustable parameters, common for all curves. according to eq.([lowtbasympt]) : @xmath133 and @xmath134. example of the data and corresponding eec is shown in fig. [15tesla]. one can see that the @xmath107 dependencies for different temperatures follow almost parallel to each other being only shifted vertically ; exactly such behavior should be expected for the eec correction according to eq.([10]). the temperature prefactor @xmath131 can be easily found from the corresponding @xmath135- dependence (see the inset to fig. [15tesla]b) ; for the given dataset @xmath136. the data in the inset also demonstrates that the obtained @xmath131 value is field - independent for @xmath137 t, which confirms that the analyzed data follow the high - field asymptotic behavior. as for the field - dependence prefactor @xmath132, since the accessible field interval is less than a decade and we do nt know how far the low - field correction (caused by the hall anomaly) may extend, we can only roughly estimate @xmath138. its lower bound is found from the low - field, @xmath139 t, data and the upper bound - from the high field (@xmath140 t) data. the next logical step would be to analyze the density dependences of the two prefactors, @xmath131 and @xmath132, and to check their consistency with each other and with a microscopic theory. figure [vg21] shows magnetoresistance and the corresponding eec for elevated density @xmath141 @xmath31 and for the same tilt angle. since conductivity increases here by a factor of 2.5, the effect of eec becomes less pronounced on top of drude conductivity, nevertheless both temperature and magnetic field dependences of the eec remain approximately the same : @xmath142 ; @xmath143. the resulting @xmath131 and @xmath132-values are summarized in table [f0summarytable]. as we argued above, these parameters do not demonstrate a pronounced density dependence. this fact is reasonable because in the studied range of densities (i) interaction parameter @xmath144 @xcite is rather small @xmath145, and (ii) conductivity is large compared to the quantum unit to ensure smallness of the renormalization effects@xcite. in table [f0summarytable] we present @xmath60 calculated from @xmath131 and @xmath132- values using eqs. ([eq : f0s]),([lowtbasympt]) for the effective valley multiplicity @xmath70 and 2. for comparison, we also show the @xmath60-values experimentally determined from shubnikov - de haas oscillations @xcite and from @xmath146 ballistic dependences @xcite. firstly, @xmath147 extracted from @xmath131 decreases with density. secondly, @xmath147 values extracted from @xmath131 for @xmath69 is rather close to the earlier measured values. if we adopt the effective valley multiplicity @xmath62 to be somewhat smaller than 2, the agreement will become even better. when making such comparison with earlier data one should keep in mind that previous results were obtained in high - temperature ballistic regime, where both valleys contribute equally and @xmath69 exactly. one should also take into account that the @xmath147 values for the diffusive (this work) and ballistic (previous data @xcite) regimes should not coincide, for the reasons discussed in ref. we therefore conclude the temperature dependence of the diffusive eec (i.e., @xmath131-values) to agree quantitatively with earlier data and to agree qualitatively with theory. possible reasons for large uncertainty in @xmath132-values and their poor consistency with @xmath131-values may be (i) too narrow range of fields accessible for identifying @xmath4 dependence (see above), (ii) the effective number of valleys (@xmath148) may be different in the field and temperature dependences. for the lowest studied density @xmath149 @xmath31, the low - field feature broadens and obscures observation of the decreasing logarithmic @xmath121 dependence ; this prevented us from measuring the respective @xmath132 values. to summarize this section, we presented above three firm experimental observations on the high - field and low - temperature behavior of the eec, as follows : (i) the observed @xmath43 dependences do not demonstrate any anisotropy with respect to the field direction, (ii) @xmath49 is linear both in @xmath15 and @xmath150 in line with the expected asymptotics (eq. [lowtbasympt]), and (iii) the observed field and temperature dependences have an anticipated magnitude. in total, the above listed facts prove that the field and temperature dependences of conductivity indeed represent the eec. the prefactor in the temperature dependence of @xmath49 reproduces @xmath147 values that are decreasing with density and are reasonably consistent with earlier data. the prefactor in the @xmath15-dependence does not contradict the microscopic theory and earlier data though it was determined with rather large uncertainty..summary of interaction parameters found from the eec measurements.[f0summarytable] [cols="^,^,^,^,^,^",options="header ",] a nonmonotonic magnetoresistance in perpendicular field and at low temperatures in si and p - sige has been observed since 1982 @xcite : as field increases, the drop in @xmath0 due to weak localization is followed by the resistivity increase (see, e.g. fig. [perpfield]a). surprisingly, the effect has not been discussed and understood yet. as can be seen from our data (fig. [15tesla]a), field tilting makes this effect even more pronounced, thus pointing to the zeeman nature of the nonmonotonic magnetoresistance. qualitatively, this seems to be transparent : the larger is the zeeman field, the stronger is the positive component of magnetoresistance. in the strictly parallel field orientation there is no weak localization suppression and the magnetoresistance is purely positive. following castellani et al. @xcite, the magnetoresistance in the diffusive regime for 2d systems was often attributed to eec @xcite and used to evaluate interaction parameter @xcite. however, whether the observed parallel field magnetoresistance may be entirely attributed to eec, has not been proven so far. having developed a technique to measure zeeman field dependence of the eec directly, with no adjustable parameters, we can compare now eec with magnetoresistance. figure [parallel]a shows a set of magnetoresistance curves collected at different temperatures for @xmath99 @xmath31 in the parallel field orientation. high field asymptotics of eec for the same sample and same density @xmath151 is also shown in fig.[parallel]a by dashed line (the prefactor 0.15 for this particular density is found in the previous section). the observed magnetoresistance appears to be (i) an order of magnitude larger than eec, and (ii) has much stronger temperature dependence ; therefore, it is mainly of a different origin. this is one of the key results of the present study, it brings into a question the validity of usage parallel field magnetoresistance for finding the interaction constant. we have shown above that the multiple scattering (eec) approach leads to much smaller magnetoresistance than that observed ; we test below whether or not single impurity scattering processes play dominant role in magnetoresistance. we show in fig. [parallel]a (dotted lines) the magnetocondutivity calculated according to eq. (4) from ref. @xcite (two - valley version of theory @xcite) for a system with @xmath152, @xmath153, and @xmath154. the calculated theory curve seems to demonstrate a reasonable agreement with the same experimental data. this agreement (in contrast to above sharp disagreement - see the dashed curve in fig. [parallel]a) has a simple physical explanation : the calculated curve comprises mainly (more than 65 %) single impurity scattering renormalization, and only 35% of the diffusive eec. we believe that by using two different @xmath60 values for ballistic and diffusive contributions @xcite and by introducing finite valley splitting and intervalley scattering rate @xcite (i.e. about 4 adjustable parameters) one can achieve a perfect agreement with experiment. this exercise reproduces an apparent successful comparison of the parallel field magnetoresistance data in si - mosfets@xcite with theory@xcite. is this the end of the story and complete understanding of the magnetoresistance in the 2d system? our answer is no ; in order to argue this point we apply perpendicular field. theoretically @xcite the zeeman field - induced magnetoresistance should remain the same in tilted field. figure [parallel]b shows magnetoresistance at fixed @xmath155 k and @xmath99 @xmath31 for various tilt angles. in perpendicular field (0@xmath110 curve), as field increases from zero, weak localization becomes suppressed, leading to the negative magnetoresistance ; further we observe almost flat magnetoresistance until the onset of shubnikov - de haas oscillations. such classically flat magnetoresistance is a signature of a field - independent mobility. in the opposite limit of parallel field (90@xmath110 curve) we observe a huge, 25% magnetoresistance in field of 6 t. this fact itself is remarkable because we did not expect much difference between the parallel and perpendicular configurations in fields @xmath156 within the above simple considerations. although we do not believe in orbital origin of the parallel field magnetoresistance, let us assume for a moment that only parallel component of the field affects mobility. if this is so, then by tilting the field from purely in - plane direction (@xmath157) to slightly out of plane (75@xmath110), when parallel component of the field decreases only by 4%, we would expect roughly the same positive magnetoresistance as that in the purely parallel field. surprisingly, after suppression of weak localization, the observed positive upturn in magnetoresistance is quite shallow and is no more than @xmath1586%! for intermediate tilt angle (45@xmath110), the magnetoresistance (fig. 5b) only slightly deviates from the perpendicular field data, which further confirms suppression of the parallel field mr by perpendicular component. to summarize our observations, the positive magnetoresistance that is weak in perpendicular and tilted fields increases dramatically when the field turns parallel to the 2d plane. such behavior is counterintuitive and qualitatively different from that in gaas - based structures.
Discussion
in the present study, by using magnetoresitance tensor measurements technique in tilted magnetic field, we extracted eec and explored zeeman effects in conductivity. we demonstrated fruitfulness of this approach and revealed the anticipated @xcite high - field logarithmic asymptotics of the eec. our measurements also reveal several puzzling features of the magnetotransport in si - mosfets. firstly, the low - field drop of the eec (i.e., the increase in hall resistance), detected in our experiments, lacks an explanation. in order to clarify this issue one requires more precise very - low - field hall measurements with variety of samples ranging in mobility and conductance values, as well as theoretical framework to treat the problem. another important observation is that the parallel field magnetoresistance is not produced by eec solely. this observation poses a question of the applicability of the eec theory to determination of the interaction constant from the parallel field magnetoresistance @xcite. indeed, our data prove that there is another unexplored mechanismthat also contributes to parallel field magnetoresistance ; it s necessary to disentangle different mr mechanisms, before extracting the fermi - liquid constant. finally, our key result is that the parallel field magnetoresistance is suppressed by rather insignificant nonquantizing perpendicular field component. this result directly contradicts to the predictions of the electron - electron interaction theory @xcite. were the parallel field magnetoresistance caused by eec, such behavior could be straightforvardly explained by complete suppression of magnetoresistance already at @xmath159, according to eq. ([aamrtensor]). however, we have shown above, that eec is too small and alone can not be a reason for the parallel field magnetoresistance. according to eq. ([aamrtensor]) it means that the parallel field mr comes from mobility renormalization. there are several potential mechanisms of such renormalization : (i) mobility may drop with parallel field due to finite width effects @xcite. this mechanism, however, is not supported by our data (compare the 90@xmath110- and 75@xmath110-curves in fig. [parallel] that show a factor of 3 diminished magnetoresistance in almost the same parallel field), and is not very probable because of thin potential well in si - mosfets (@xmath160 nm) compared to magnetic length @xmath161[nm]@xmath162[t])@xmath163 ; (ii) mobility may depend on zeeman splitting in agreement with the predictions of the screening theory @xcite. the latter mechanism should produce direction - independent positive magnetoresistance, exactly the behavior observed in heavily - doped multisubband thin si quantum wells@xcite. this however does not reconcile with our data. (iii) surface roughness could also affect mr through the weak localization suppression @xcite, however, this mechanism is hardly relevant because it produces a negative mr in contrast to our observations. experimentally, we observe a picture opposite to the common sense arguments : parallel field magnetoresistance is strongly suppressed when sample is rotated in fixed total field. this observation is not a property of particular studied samples, rather, there are numerous examples in literature where strong positive mr in parallel field coexists with almost shallow or even negative mr in perpendicular or tilted field : for p - gaas @xcite, strained si @xcite, si / sige quantum wells @xcite. dramatic delocalizing effect of the perpendicular field was also reported for high mobility si - mosfet samples in the low - density / high resistivity (@xmath164) regime in the vicinity of the metal - insulator transition @xcite. such behavior is not explained at all, and the question why parallel field mr is suppressed by the perpendicular field component remains open. we believe that the parallel field magnetoresistance suppression in perpendicular field is somehow related to the low - field hall feature (region a in fig. [perpfield]g). the empiric crossover between the low- and high - field regimes @xmath165 points to the localization - related nature of the low - field hall feature. we, therefore, suggest the following model : a small part of electrons is localizedin low binding - energy states and coexist with mobile electrons. we stress that these states are different from conventional tail of localized states located below the mobility edge @xcite. the localized electrons do not contribute to hall effect at @xmath166 because they can be treated as zero mobility carriers. let us assume also that the localized states promote strong parallel field mr. this assumption does not contradict to the empirical observations of the disorder effect in mr @xcite. application of perpendicular field changes the symmetry class of the system, favors delocalization of these electrons and hence decreases the response of a 2d system to parallel magnetic field. such explanation initiates questions about the source of 10 - 30% electrons that are out of the game at @xmath166 and about their influence on the parallel field mr. the tilted field approach suggested in this paper might be applied for similar study of eec with high - mobility si - mosfets where strong metallic conductivity (@xmath167) and metal - insulator transition are considered to be driven by the eec @xcite. this task is however rather challenging experimentally because it requires a combination of millikelvin temperatures (to insure the deep diffusive limit @xmath10 for an order of magnitude higher mobility samples), field sweep and sample rotation.
Conclusion
in this paper we applied phenomenological technique of resistivity tensor analysis to check long - standing prediction by lee and ramakrishnan @xcite about zeeman splitting dependence of the electron - electron interaction correction (eec) to conductivity in the diffusive regime @xmath168. our measurements reveal distinctly different behaviors of the magnetotransport in the two domains of perpendicular field, the low @xmath118-field (lf) and the high @xmath118-field (hf) one. in the lf domain, the 2d system demonstrates a strong and @xmath18-dependent magnetoresistance versus @xmath169 field, whereas the hall angle exceeds the drude value by up to 30%, depends on @xmath118, and grows as @xmath18 decreases. this hall anomaly obscures determination of the quantum interaction correction from the magnetotransport data in the lf domain. in the hf domain, the hall anomaly is washed - out by the @xmath118 field that enables extracting the interaction quantum correction from experimental data. the eec was found to be field direction independent, and linear in both @xmath150 and @xmath15, as expected. remarkably, the magnitude of the experimentally determined eec appeared to be more than a factor of 10 smaller than the parallel field magnetoconductance. thus, the parallel field mr observed in si - mosfets is not explained by the eec solely. even more surprisingly, the observed strong parallel field mr quickly diminishes when the perpendicular field component is applied on top of the parallel field. this fact is the direct evidence for non - purely zeeman origin of the parallel field magnetoresistance. in total, our findings point at the incompleteness of the existing theory of magnetotransport in interacting and disordered 2d systems : too strong parallel field magnetoresistance, its suppression by perpendicular magnetic field, and the low - field hall anomaly require an explanation. we believe that these three phenomena are interrelated and originate from a destructive action of the perpendicular field on the localized states.
Acknowledgements
we thank g.m. minkov, a.v. germanenko, a.a. sherstobitov, i.s. burmistrov, and a.m. finkelstein for discussions, and s.i. dorozhkin for stimulating criticism of the preliminary results. the work was supported by rfbr (grants 12 - 02 - 00579), by russian ministry of education and science (grant no 8375), and using research equipment of the shared facilities center at lpi.
Appendix a: test of the applicability of the high-field asymptotics
in the original papers @xcite, the interaction correction field dependence, @xmath170, is expressed via an integral with known low and high - field asymptotics. this integral is inconvenient in handling, therefore, we used its analytical asymptotics eq. ([lowtbasympt]) instead. in fig. [asymptotics] we compare the exact result @xmath170 for several @xmath171 values [eq. (15) in ref. @xcite, note that the dimensionless field scale differs by a factor @xmath172 and high - field linear - in-@xmath15 asymptotics, eq. ([lowtbasympt]). to exclude interaction constant dependence in the high - field limit all results were divided by @xmath173. figure [asymptotics] shows that for @xmath92 the exact result for different @xmath171 become indistinguishable from the logarithmic high - field asymptotics. at the same time, in the low field regime the curves substantially deviate from each other : the closer @xmath171 is to -1, the stronger is the magnetoconductance. this fact has simple physical explanation : in the high - field limit, @xmath83 triplets (out of all @xmath174 triplet terms) become suppressed, thus excluding any dependence on @xmath171. in the low field regime, the closer @xmath175 is to -1 (i.e. to the stoner instability), the larger is the effective @xmath3-factor, and the stronger is the response of conductivity to magnetic field. alternative approach used in ref. @xcite is based on the approximation of the crossover function (eq.(4) from ref. @xcite). for comparison, we present this approximation also in fig. [asymptotics]. from the practical point of view, the condition @xmath92 means that the total magnetic field @xmath50[t] should exceed @xmath176[k]. this inequality is valid for our low - temperature high - field data shown in figs. [15tesla] and [vg21].
Appendix b: low-field nonlinearity of the hall resistance
nonlinearity of the hall resistance in the low field domain affects both @xmath103 and @xmath121 dependences due to the calculation procedure, based on eq. [aamctensor]. most of scattering mechanisms and quantum corrections, including low - field weak localization and maki - thompson corrections, renormalize mobility @xmath41 and do not renormalize hall effect @xcite. eec does affect hall resistivity but has no peculiarities close to zero field. the low field hall feature is thus unexpected in theory, though its experimental observation is not surprising. the lack of agreement between low - field hall resistance and quantum correction theory was first pointed by ovadyahu @xcite and reported several times since then @xcite. although there are some theoretically suggested mechanisms @xcite, this phenomenon is still poorly understood. let us briefly summarize our experimental observations of the low - field nonlinearity in the hall resistance in si - mosfets : (i) the effect weakens as temperature increases (see e.g. fig. [15tesla]) ; (ii) the amplitude of the effect may achieve @xmath177 at low temperatures, i.e. variations in the hall angle are as large as 30% ; (iii) the @xmath107 dependence is determined by perpendicular field component. to prove this we compare in fig. 7 @xmath107 for perpendicular field orientation, @xmath122 and @xmath107 for @xmath178 ; (iv) the field range, where the hall resistance is nonlinear, broadens as density (and hence conductivity) decreases. (v) the effect is observed in low - mobility si - mosfets (@xmath179 m@xmath21/vs) in the temperature range @xmath180k (@xmath181). for high mobility samples (@xmath182 m@xmath21/vs) we did nt observe any nonlinearity in the hall resistance down to 0.05k (@xmath183). it might be that the low - field hall nonlinearity is somehow related to weak localization, though weak localization itself does not produce correction to hall coefficient. indeed, let us estimate transport field @xmath184, (@xmath185), that is the typical value of perpendicular field component where weak localization is suppressed. by substituting the transport mean free path @xmath96 from the drude formula @xmath186 for a two - valley system, we get : @xmath187, where @xmath188 is the dimensionless drude resistivity, and @xmath189 for simplicity was taken equal to @xmath190 ; this simplification is justified for @xmath191 kohm. for practical use @xmath192= 0.062\times n[10^{12 } { \rm cm}^{-2}]\times (\rho_d[{\rm kohm}])^2 $]. the coincidence of the estimated @xmath193 t value for the most intensively discussed density @xmath99@xmath31 with characteristic field of the hall anomaly suppression at the lowest temperature (see figs. 1, 3, 4) points to the relationship between weak localization and the anomalous low - field hall slope. observation (iv) is in line with this scenario, because growth of the resistivity pushes @xmath194 to higher fields. a mechanism of the weak - localization - related low - field nonlinearity in the hall resistance was suggested in ref. @xcite, where the effect was interpreted as the second - order correction to conductivity ; i.e. a crossed term of the eec (which controls the amplitude of the effect) and weak localization (which controls the magnetic field dependence). in our case, however, the sign of the effect is opposite to that of ref. @xcite, and the amplitude of the effect is much stronger. similar sign of the effect (i.e. hall coefficient decreasing with field down to its nominal value @xmath195) follows from memory effects @xcite. however, the memory effects do not produce temperature dependent correction to hall effect, as we observed. interaction in the cooper (particle - particle) channel @xcite produces the so - called dos correction that is suppressed in magnetic fields of the order of @xmath196, similar to what is seen in experiment. the observed effect however is too big to be explained by quantum correction : indeed, were the hall nonlinearity, @xmath197, caused by dos correction, one would observe an enormous temperature dependence of the conductivity (of the insulting sign) at zero field @xmath198, much stronger than both, eec and weak localization. the absence of such temperature dependence in our data points to the irrelevance of the dos correction to the observed weak field hall anomaly. another model that might explain field dependent hall effect is a classical multi - component system. however, this model can hardly be applicable to our data because the hall coefficient nonlinearity should be accompanied with a large positive magnetoresistance, and should occur in fields @xmath199, i.e. much higher than we observe. 99 b. l. altshuler, a. g. aronov, p. a. lee, phys.. lett. * 44 *, 1288 (1980). b. l. altshuler, d. khmelnitzkii, a. i. larkin, p. a. lee, phys. b * 22 *, 5142 (1980). paalanen, d.c. tsui, j.c.m. hwang, phys. * 51 *, 2226 (1983). r. taboryski, e. veje, p. e. lindelof, phys. b * 41 *, 3287 - 3290 (1990). k. k. choi, d. c. tsui, s. c. palmateer, phys. b * 33 *, 8216 - 8227 (1986). w. poirier, d. mailly, m. sanquer, 3710 (1998). minkov, o.e. rut, a.v. germanenko, a.a. sherstobitov, v.i. shashkin, o.i. khrykin, v.m. daniltsev, phys. b * 64 *, 235327 (2001). l. li, y.y. proskuryakov, a.k. savchenko, e.h. linfield, d.a. ritchie, phys. 90 *, 076802 (2003). v. t. renard, i. v. gornyi, o. a. tkachenko, v. a. tkachenko, z. d. kvon, e. b. olshanetsky, a. i. toropov, j. c. portal, phys. b * 72 *, 075313 (2005). goh, m.y. simmons, a.r. hamilton, phys. b * 77 *, 235410 (2008). p.a. lee, t.v. ramakrishnan, phys. rev. b*26 *, 4009 (1982). g. zala, b.n. narozhny, i.l. aleiner, phys. b * 65 *, 020201(r) (2001). g.m. minkov, a.a. sherstobitov, a.v. germanenko, o.e. rut, v.a. larionova, b.n. zvonkov, phys. rev. b * 72 *, 165325 (2005). p.t. coleridge, a.s. sachrajda, p. zawadzki, prb,*65 *, 125328, (2002). minkov, a.v. germanenko, o.e. rut, a.a. sherstobitov, phys. b * 85 *, 125303 (2012). d.j. bishop, r.c. dynes, d.c. tsui, phys. b * 26 *, 773 (1982). davies m. pepper, journar physics c : solid state physics * 16 *, l679, (1983). burdis, c.c. dean, phys.rev. b * 38*(5), 3269 (1988). d. simonian, s.v. kravchenko, m.p. sarachik, v.m. pudalov, phys. * 79 *, 2304 (1997). pudalov, g. brunthaler, a. prinz, g. bauer, jetp lett. * 67 *, 887 (1997). e. abrahams, s.v. kravchenko, m.p. sarachik, rev. phys.*73 *, 251 (2001) s. a. vitkalov, h. zheng, k. m. mertes, m. p. sarachik, t. m. klapwijk phys. 87 *, 086401, (2001). a. a. shashkin, s. v. kravchenko, v. t. dolgopolov, t. m. klapwijk phys. lett. * 87 *, 086801 (2001). burmistrov, n.m. chtchelkatchev, phys. b * 77 *, 195319, (2008). s. anissimova, s.v. kravchenko, a. punnoose, a.m. finkelstein, t.m. klapwijk, nature phys. * 3 *, 707 (2007). d. a. knyazev, o. e. omelyanovskii, v. m. pudalov, i. s. burmistrov, pisma v zhetf * 84*(12), 780 - 784 (2006) ; [jetp lett. * 84*(12), 662 (2006)]. v. t. dolgopolov, a. gold, jetp letters, * 71*,27 - 30, (2000). s. das sarma, e.h. hwang, phys. b * 72 *, 205303, (2005). s. das sarma, e.h. hwang, phys. b * 72 *, 035311, (2005). j. m. broto, m. goiran, h. rakoto, a. gold, v. t. dolgopolov, phys. b * 67 *, 161304 (2003). v. m. pudalov, g. brunthaler, a. prinz, g. bauer arxiv : cond - mat/0103087 (2001). v. m. pudalov, g. brunthaler, a. prinz, g. bauer phys. * 88 *, 076401 (2002). a. v. gold, v. t. dolgopolov, jetpl (2002). s. a. vitkalov, m. p. sarachik, t. m. klapwijk, phys. b * 65 *, 201106 (2002) n. n. klimov, d. a. knyazev, o. e. omelyanovskii, v. m. pudalov, h. kojima, m. e. gershenson, phys. rev. b * 78 *, 195308, (2008). v. m. pudalov, m. e. gershenson, h. kojima, g. brunthaler, a. prinz, g. bauer, phys. lett. * 91 *, 126403 (2003). s. a. vitkalov, k. james, b. n. narozhny, m. p. sarachik, t. m. klapwijk, phys. b * 67 *, 113310 (2003). t. ando, a. b. fowler, f. stern ; rev. 54 *, 437 (1982). a. houghton, j.r. senna, s.c. ying, phys. rev. b * 25 *, 2196 (1982). the density value @xmath29 was determined preliminary with @xmath200 uncertainty from shubnikov - de - haas oscillations, and adjusted more precisely to make @xmath201 in the high - temperature limit. such procedure was substantiated in ref.@xcite. a.m. finkelstein, zh. exp. phys. * 84 *, 168 - 189 (1983). a. punnoose, phys. b * 81 *, 035306 (2010) g. zala, b.n. narozhny, i.l. aleiner, phys. b * 64 *, 214204 (2001). original formula from ref.@xcite contains @xmath202 instead of @xmath203. we omit the term @xmath204 because it is just a constant shift, which enters to mobility only and does not affect conductivity tensor in eq. ([aamctensor]) as eec does (for argumentation, see ref. a. punnoose, a.m. finkelstein, phys. lett. * 88 *, 016802 (2001). kuntsevich, n.n. klimov, s.a. tarasenko, n.s. averkiev, v.m. pudalov, h. kojima, m.e. gershenson, phys. b * 75 *, 195330 (2007). a. punnoose, a.m. finkelstein, a. mokashi, s. v. kravchenko, phys. b * 82 *, 201308 (2010). pudalov, a. punnoose, g. brunthaler, a. prinz, g. bauer, arxiv : cond - mat/0104347 (2001). proskuryakov, a.k. savchenko, s.s. safonov, m. pepper, m.y. simmons, d.a. ritchie, phys. lett. * 89 *, 076406 (2002). g. m. minkov, a. v. germanenko, o. e. rut, a. a. sherstobitov, v. a. larionova, a. k. bakarov, and b. n. zvonkov phys. b * 74 *, 045314, (2006). a. dmitriev, m. dyakonov, r. jullien, phys. rev. lett. * 89 *, 266804, (2002). gornyi, a.d. mirlin, phys. b * 69 *, 045313, (2004). y. zhang, p. dai, m.p. sarachik, phys. b * 45 *, 6301 (1992). newson, m. pepper, e.y. hall, g. hill, j. phys. c * 20 *, 4369 (1987). g.m. minkov, a.v. germanenko, o.e. rut, a.a. sherstobitov, b.n. zvonkov, phys. b 82, 035306 (2010). altshuler, a.g. aronov, a.i. larkin, d.e. khmelnitskii, jetp, * 54 *, 411, (1981). minkov, o.e. rut, a.v. germanenko, a.a. sherstobitov, v.i. shashkin, o.i. khrykin, b.n. zvonkov, phys. b. * 67 *, 205306, (2003). pudalov, m.e. gershenson, h. kojima, n. butch, e.m. dizhur, g. brunthaler, a. prinz, g. bauer, phys.. lett. * 88 *, 196404 (2002). o. prus, m. reznikov, u. sivan, v.m. pudalov, phys. 88 *, 016801 (2001). r. raimondi, c. castellani, c. dicastro, phys. b * 42 *, 4724, (1990). k. michaeli, k. s. tikhonov, a. m. finkelstein, phys. b * 86 *, 014515 (2012). a. p. dmitriev, v. yu. kachorovskii, phys. b * 77 *, 193308 (2008). p. m. mensz, r. g. wheeler phys. b * 35 *, 2844, (1987). a. s. troup, j. wunderlich, d.a. williams, journal of applied physics, * 101 *, 033701 (2007). m. kumar, g. moria, f. capotondia, g. biasiola, l. sorba, sol. commun, * 135 *, 5761, (2005). k. lai, w. pan, d.c. tsui, s.a. lyon, m. muhlberger, f. schaffler, phys. b * 72 *, 081313(r), (2005). t. okamoto, m. ooya, k. hosoya, s. kawaji, phys. rev. b * 69 *, 041202(r) (2004) s.v. kravchenko, d. simonian, m. p. sarachik, a. d. kent, v. m. pudalov, phys. b * 58 *, 3553 (1998). a. punnoose, a. m. finkelstein, science * 310 *, 289 (2005). b. l. altshuler, a. g. aronov, in _ electron - electron interactions in disordered systems _, edited by a. l. efros and m. pollak (elsevier, amsterdam, 1985) e. tousson, z. ovadyahu, phys. b * 38 *, 12290 (1988). | we study diffusive electron - electron interaction correction to conductivity by analyzing simultaneously @xmath0 and @xmath1 for disordered 2d electron systems in si in tilted magnetic field. tilting the field
is shown to be a straightforward tool to disentangle spin and orbital effects.
in particular, by changing the tilt angle we prove experimentally that in the field range @xmath2 the correction depends on modulus of magnetic field rather than on its direction, which is expected for a system with isotropic @xmath3-factor. in the high - field limit
the correction behaves as @xmath4, as expected theoretically (lee, ramakrishnan, phys.
rev. b*26 *, 4009 (1982)).
our data prove that the diffusive electron - electron interaction correction to conductivity is not solely responsible for the huge and temperature dependent magnetoresistance in parallel field, typically observed in si - mosfets. | 1305.3736 |
Monte carlo studies of the fnc
since the spring of 1995, the fnc has been operating in the zeus experiment at hera where it is used in the study of leading neutrons produced at small angles with energies @xmath92 100 gev. because the calorimeter has been calibrated and tested only in beams of energy up to 120 gev, and because the top was not present for the beam tests, we must rely on a monte carlo simulation to predict the response of the calorimeter to high energy particles. we have modeled the fnc using the geant 3.13@xcite program, upon which the simulation of the full zeus detector is based. in this section we present some results from the simulation which can be compared to our test beam data. for 120 gev electrons and pions incident on the center of towers 5 and 6 the geant simulation predicts an electron to hadron response ratio of 0.98, in agreement with the measured value of 0.96. the simulated response to pions incident on the center of each tower is shown in fig. [mc_eres]. the energy loss due to leakage when the beam is incident near the edge of the calorimeter is also in agreement with the data as is the degradation of the energy resolution (compare with the data shown in fig. [ptower]). the monte carlo gives an energy resolution due to shower fluctuations alone, that is, neglecting fluctuations due to photostatistics, of 0.66/@xmath35 for hadrons incident at the center of towers 5 and 6. we have also used the monte carlo to predict, at higher energies, the overall energy response of the calorimeter and its expected energy resolution due to shower fluctuations. the monte carlo predicts that for neutrons centered on the calorimeter without the top, the energy response is linear up 800 gev. fitting the energy resolution as a function of incident energy, we find @xmath93 where @xmath94 is in gev. if fluctuations due to photostatistics are added, the sampling constant 0.54 will increase to 0.58. the top section is present at hera, but the neutrons are predominantly incident on towers 7 and 8. just as for data, 120 gev pions were studied with a grid over the face of the calorimeter. [mc_ys] shows the predicted dependence of @xmath15 on @xmath12, the vertical impact position, for three values of horizontal impact position, @xmath95, 10.1, and 20.1 cm. @xmath15 is calculated with logarithmic weights and a cutoff parameter of f=10%. the simulation shows behavior and biases similar to the data (fig. [ys], lower inset) ; in particular, the value of @xmath15 is biased towards the nearest tower center. the @xmath12 residual distribution is gaussian with mean 0 and width 7.3 cm/@xmath35 (compare to the data width of 8.0 cm/@xmath35). the @xmath11 residual distribution is gaussian with mean 0 and width 10.3 cm/@xmath35. this width is due to transverse shower fluctuations. if photostatistics fluctuations are also included, the width constant increases to 20.7 cm. this should be compared with the data value of 22.3 cm. the simulated energy response for pions as a function of position, shown in fig. [mc_eofy], is in agreement with the data shown in fig. [eofy]. fig [mc_ywidth] shows simulations of the vertical shower widths for 70 and 120 gev electrons and pions. the monte carlo results are narrower than the data shown in fig. [ehsep].
Summary
we have designed and constructed a lead scintillator sandwich calorimeter for the zeus experiment at hera. the calorimeter is divided into 5 cm vertical towers read out on two sides with wavelength shifting light guides coupled to photomultiplier tubes. the calorimeter was tested in the h6 beam line at cern with 120 gev electrons, muons, pions and protons. electrons can be cleanly separated from hadrons using the energy weighted vertical width of a shower. at 120 gev the calorimeter is slightly over compensating, with an electron to hadron response ratio of 0.96 and an energy resolution 6% at 120 gev. the horizontal position resolution, measured by charge sharing between the two sides, is 20 cm/@xmath35, assuming a @xmath96 dependence ; the vertical position resolution, measured by energy sharing between the towers, is 10 cm/@xmath35. by using energetic neutrons produced by proton interactions in a lucite target, the overall energy scale was determined to 1.5%. since the spring of 1995, the calorimeter has operated successfully in the hera tunnel 105.6 m downstream of the zeus detector on the zero degree line.
Acknowledgments
we thank t. tymieniecka for modeling the fnc during the design stage, and using fluka to study its energy resolution and @xmath47 response. e. borsato, f. czempik, c. fanin, r. fernholz, t. kiang, k. loeffler, h. stehfest, v. sturm, and k. westphal provided us with much help constructing the calorimeter, shipping it to cern, and installing it in the hera tunnel. m. roseman helped with the pmt tests and at cern. we thank m. brki for his assistance in setting up the computer readout system for the cern tests. b. racky kindly arranged for us the transport of our @xmath97co source to and from cern. we also thank f. dittus for making available the adjustable table on which we mounted the calorimeter, and k. elsener for his invaluable assistance with the h6 beam. we are also grateful to cern for making the beam test possible. j. prentice helped with cosmic ray tests during the initial stages. the zeus collaboration has been continually enthusiastic and supportive, in particular, r. klanner, g. wolf and u. koetz. we also thank the hera machine group who helped install the fnc, and who provided the beam line modifications which greatly enhance the performance for physics of the fnc. finally, we especially thank the desy directorate for the continual interest they have shown in the project, and for the financial support they provided. d. bintinger, in _ proceedings of the workshop on calorimetry for the supercollider _, tuscaloosa, al, march 13 - 17, 1989, r. donaldson and m. g. d. gilchriese, ed., (world scientific, teaneck, nj, 1989), p. 91. | a lead scintillator sandwich sampling calorimeter has been installed in the hera tunnel 105.6 m from the central zeus detector in the proton beam direction.
it is designed to measure the energy and scattering angle of neutrons produced in charge exchange @xmath0 collisions. before installation
the calorimeter was tested and calibrated in the h6 beam at cern where 120 gev electrons, muons, pions and protons were made incident on the calorimeter.
in addition, the spectrum of fast neutrons from charge exchange proton - lucite collisions was measured.
the design and construction of the calorimeter is described, and the results of the cern test reported.
special attention is paid to the measurement of shower position, shower width, and the separation of electromagnetic showers from hadronic showers.
the overall energy scale as determined from the energy spectrum of charge exchange neutrons is compared to that obtained from direct beam hadrons. hep - ex/9701015 + desy 97 - 006 + design and test of a + forward neutron calorimeter + for the zeus experiment + the zeus fnc group + s. bhadra@xmath1, i. bohnet@xmath2, m. cardy@xmath1, u. dosselli@xmath3, c.- p.
fagerstroem@xmath1, w. frisken@xmath1, k. furutani@xmath1, d. hanna@xmath4, u. holm@xmath2, k. f. johnson@xmath5, m. khakzad@xmath1, g. levman@xmath6, j. n. lim@xmath4, b. loehr@xmath7, j. f. martin@xmath6, c. muhl@xmath7, t. neumann@xmath2, m. rohde@xmath7, w. b. schmidke@xmath1, d. g. stairs@xmath4, h. tiecke@xmath8, c. voci@xmath3 + [cols= " >, < ",] the photomultipliers were read out with lecroy 4300 charge integrating adcs.
the adcs were gated so that 160 ns of the pulse from the calorimeter was integrated. during data taking the adc pedestals
were regularly measured, and the gains were monitored with an internal test function in the units.
the gains were also checked daily using an external charge injector. over the course of the beam test, which lasted one week,
individual adc gains varied less than 1 part per mil, and all gains were uniform to within 5 parts per mil. in the following we use a right - handed coordinate system with @xmath9 vertically up and @xmath10 into the face of the calorimeter.
the origin (0,0) is taken to be between the two central towers halfway between the sides (the center of the face of the calorimeter). @xmath11 and @xmath12 are measurements of @xmath13 and @xmath9 based on the table position ; @xmath14 and @xmath15, measurements based on energy deposits in the fnc.
the vertical position of a shower is determined using the centroid method by taking an energy weighted average of the the tower positions : @xmath16 where the sum is over towers in the front part of the fnc, and the weights @xmath17 are functions of the energy deposits, @xmath18, in each tower.
the weights are chosen to minimize bias and to maximize resolution in the position measurement.
the simplest method of estimating @xmath9 corresponds to choosing weights which are linear in energy, @xmath19.
awes et al.@xcite have suggested that the weights be chosen to have instead a logarithmic dependence on the energy deposits : @xmath20, \label{eqn_w}\]] where @xmath21, and @xmath22 is a fractional cutoff.
when the energy deposited in a tower satisfies @xmath23, that tower is not used in the determination of the vertical position.
the logarithmic dependence of the @xmath17 more closely matches the true transverse profile of energy deposits in a hadronic shower than the linear dependence.
this estimator is simple, but it has an undetermined free parameter @xmath24 which may be a function of the deposited energy.
in addition the optimal value of the cutoff parameter may depend on particle type, electron or hadron.
the vertical segmentation also allows us to estimate the size of the shower using the rms vertical shower width : @xmath25 since the width of electromagnetic showers is much less than the width of hadronic showers, this estimator enables us to distinguish incident electrons from incident hadrons
. the optimal weights for the width determination are not necessarily the same as those for the position measurement.
since we use the width @xmath26 to separate electromagnetic from hadronic showers, it is most convenient to have an estimator that does not depend on knowledge of the nature of the incident particle. otherwise an iterative procedure must be employed.
as we will discuss in more detail below, linear weights for the shower size measurement give a hadron - electron separation which is close to that obtained with logarithmic weights whose cutoff parameter has been optimized.
the horizontal position of a shower can be estimated from the division of scintillation light between the left and right sides of the calorimeter : @xmath27 where @xmath28 is the effective attenuation length of the scintillator strips, and @xmath29 is the charge from the left(right) pmt.
if the light attenuation behaves as @xmath30 where @xmath31, @xmath32 are constants, and @xmath13 is the displacement of the incident particle from the center of the tower, then @xmath14 is independent of @xmath32 and @xmath33.
the calorimeter was calibrated with 120 gev electrons incident on the center of each tower.
the high voltage for the tubes in the front was adjusted so that their mean response to 120 gev electrons as measured in the lecroy adcs was 100 pc (400 counts). after lifting and rotating the calorimeter by 180 degrees to make electrons incident on the rear towers, the high voltage of each rear tube
was set so that the response of the rear section to a muon passing through the calorimeter would approximately equal the response of the front section. since the ratio of the number of front layers to the number of rear layers is 95/39, the gain is considerably higher in the rear ; however, the energy deposits in the rear for electrons and hadrons are small, so there is no saturation problem for the very high energy particles at hera.
figure [energy]a shows the distribution of deposited energy, after calibration, for 120 gev electrons centered on towers 3 through 7, inclusive.
pion contamination has been removed be requiring that the vertical shower width be less than 3 cm (section [separation]).
the energy resolution for electrons centered on a tower averaged 33%/@xmath34 independent of tower.
electrons centered in the edge towers had an energy response reduced by 3% due to transverse energy leakage.
the response of the calorimeter to electrons is dependent on the position of the electron with respect to the gap between the scintillator strips.
figure [crack_width]a shows the energy deposited in tower 6 when the electron beam is scanned from tower 6 to tower 7.
the total signal for electrons incident between two towers is 90% of the signal of an electron incident on the center of a tower.
moreover, the energy resolution is degraded to 60%/@xmath35.
these effects are due to the narrowness of electromagnetic showers compared to the width of the region between strips.
the small width of the shower compared with the strip width, combined with the reduction in response at the tower boundaries, is responsible for the large bias towards the tower center in the measured electron vertical position shown in fig.
[crack_pos].
the bias grows with increasing cutoff parameter since for sufficiently large @xmath24 only the struck tower contributes to the sums in equations [eqn_y] and [eqn_yw].
the rms deviation of the distribution also tends to increase.
the variation with vertical position of the shower width (linear weights) of electrons is shown in fig.
[crack_width]b, and is small compared to the average width of hadronic showers (see section [separation]). electrons were also used to determine the effective attenuation length for light in the scintillator strips.
one half of each tower was scanned horizontally.
for each position the response of the right and the left photomultiplier tubes was plotted as a function of the offset @xmath13 of the beam from the center of the calorimeter, positive @xmath13 representing an offset away from the pmt.
figure [attenl] shows the resulting attenuation curve for a typical tower. the response is well represented by equation [eqn_atten]
the tower - to - tower average value of @xmath31 was found to be 1.08@xmath36 @xmath37 with a 3% rms spread, corresponding to an average effective attenuation length of 92 cm ; @xmath32 averaged 9.60@xmath38 @xmath39 with a 20% rms spread. to compare the efficiencies of the wavelength shifting material in the front and rear section we measured the number of photoelectrons collected by the photomultiplier tubes, @xmath40, which is related to the width, @xmath41, of the difference distribution by @xmath42 where @xmath43 is the mean of the sum. by rotating the calorimeter 180 degrees to make electrons incident both front and rear, we found that although in the rear we averaged 92 pe / gev, in the front we obtained only 26 pe / gev.
this difference is due to different wls lengths, wls material, and optical geometry between front and rear, and it worsens the @xmath13 position resolution by a factor of @xmath44 compared to the estimate in section [description].
photostatistics makes only a small contribution to the energy resolution since, for 120 gev electrons and 26 pe / gev, the fluctuations are only about 2%.
electron showers penetrate only about an interaction length into the calorimeter.
to refine the calibration constants determined by electrons, 120 gev pions were made incident on the center of each tower. using these data, correction factors were determined which equalized the response of each pmt to hadrons.
the correction factors for hadrons were forced to average 1.0, and had an rms spread of 5%.
the distribution of deposited energy for 120 gev hadrons (pions and protons) incident on the center of the calorimeter, between towers 5 and 6, is shown in fig. [energy]b. for pions incident in the center of each tower, the average response of the calorimeter
is shown in fig.
[ptower].
the curves are quadratic in @xmath45.
leakage loss is important whenever the beam is outside the four central towers.
when the beam is incident in the center of the calorimeter, between towers 5 and 6, about 2.5% of the energy is lost compared to the case when the beam is in the center of tower 5 or 6 because of the gap between the scintillator strips (see inset of fig.
[eofy]).
parameterized as @xmath46, the energy resolution for hadrons incident at the center of towers 5 and 6 is 0.70/@xmath35 (fig.
[ptower]).
this improves to 0.62/@xmath35 for hadrons incident at the center of the calorimeter, between towers 5 and 6.
these results can be compared to the results for electrons discussed above.
the electron to hadron ratio @xmath47 is 0.96 when the beams are incident on the center of a tower, but it falls to about 0.9 when both are incident on the center of the calorimeter in the gap between towers 5 and 6. for electrons centered on a tower, but hadrons incident on the center of the calorimeter @xmath48 (see fig. [energy]).
if the electron calibration constants are used instead of the hadron calibration constants, @xmath47 changes by 0.01. as expected from the ratio of lead to scintillator thickness, the calorimeter is over compensating.
this effect is magnified for electrons by the gap between the scintillator strips ; however, since compared to the electrons the electromagnetic component of a hadronic shower is more spread out transversely, we expect the ` effective'@xmath47 to be higher. in order to study energy leakage and position resolution for hadrons, the face of the calorimeter
was scanned with 120 gev pion over an @xmath49 grid. using this grid the effective attenuation length of the scintillator
was determined with hadronic showers. the resulting average attenuation length, 91.9 cm, was in good agreement with the electron scan results.
figure [ys] shows the true vertical position of the calorimeter as a function of its position determined using the centroid algorithm with linear weights (@xmath19).
the effect of the finite strip size is clearly evident in the oscillation superimposed on an average behavior.
in addition, towards the edge of the calorimeter the measured value falls below the true value because of leakage.
the average behavior can be parameterized conveniently as @xmath50.
the residuals, shown in the upper inset, follow an approximate sine curve with period 5 cm, the tower height. the overall vertical resolution is determined by plotting the @xmath9 residuals for the centroid algorithm with the polynomial correction for all points of the grid.
the residual distribution is well fit by a sum of two gaussian (fig.
[yres]a).
the largest component has a weight of 0.91 and an rms width of 0.8 cm for 120 gev pions.
because this width is larger than the size of the oscillation, the @xmath9 position can not be corrected for the oscillation on an event by event basis.
in contrast to linear weights, logarithmic weights depend on an arbitrary cutoff parameter @xmath24 which may be chosen to optimize the position resolution, minimizing bias.
residuals for logarithmic weights with @xmath51 are shown in the lower inset in fig.
[ys] and in fig.
[yres]b.
the bias due to leakage has been considerably reduced (@xmath52), and the resolution improved, compared with the use of linear weights.
a measure of the effective overall resolution is @xmath53 @xmath54 where @xmath55 is the variance of gaussian @xmath56, and @xmath57 is its fractional contribution to the overall distribution.
the resulting @xmath53 is plotted in fig.
[fval] as a function of @xmath24.
the best vertical position resolution for 120 gev pions is 8 cm/@xmath35 for @xmath58, where the curve has a minimum. like the @xmath9 residuals, the distribution of @xmath13 residuals is also well represented by a sum of two gaussians : 97% with @xmath59 cm/@xmath35, and 3% with @xmath60 cm/@xmath35, giving @xmath61 cm/@xmath35.
the resolution in @xmath13 is much poorer than in @xmath9 ; nevertheless, it is close to the value expected since for 1 gev incident energy @xmath62 the energy response varies over the grid as a result of two competing effects : a) energy leakage, and b) light enhancement. as the beam is moved towards the edge of the calorimeter, energy is lost due to leakage, and the measured signal tends to decrease. on the other hand, as the beam moves horizontally away from the center the light collected by the pmts increases because of the characteristic light attenuation of the scintillator strips : @xmath63 the net result is that the signal increases for horizontal displacements at fixed @xmath9.
the increase is approximately linear in @xmath13 with @xmath64.\]] figure [eofy] shows the energy response as a function of y for all grid values with the @xmath13 dependence removed.
energy loss due to the gap between the scintillator strips is clearly visible as is also loss due to leakage.
the leakage component has a quadratic dependence in @xmath45 ; the loss due to the gaps between strips behaves as a superposed oscillation.
a calorimetric measurement with the capability of distinguishing between electromagnetic and hadronic showers is important for physics studies.
because electromagnetic showers are much narrower than hadronic showers, the measurement of the shower width allows a separation of particles by type.
the use of logarithmic weights for the determination of the vertical shower position reduces the bias due to the finite strip size and lateral leakage, and improves the resolution ; however, the optimal choice of the cutoff parameter depends on incident particle type. at 120 gev
, @xmath24 should be chosen to be about 10% for hadrons, but should be close to 0 (no cutoff) for electrons.
the choice of cutoff for the width measurement can be chosen to optimize the separation of electromagnetic and hadronic showers. to estimate the optimal value of @xmath24 we define @xmath65 as a measure of the electron - hadron separation, where @xmath66 and @xmath67 are the means and standard deviations, respectively, of the width distributions for hadrons (@xmath68) and electrons (@xmath69).
all are implicitly functions of the cutoff @xmath24, so the value of @xmath24 can be chosen to maximize @xmath70.
the dependence of @xmath70 on @xmath24 for 120 gev electrons and hadrons is shown in fig.
[ehsep].
@xmath70 is insensitive to @xmath24 at low values of @xmath24, but falls approximately linearly for @xmath71%.
linear weights give the result shown by the dashed line, @xmath72, close to best values obtained with logarithmic weights, and have the advantage of being independent of @xmath24, which may depend on the incident energy. as a result
, we use linear weights for measurement of shower widths instead of the logarithmic weights chosen for the shower centroid measurement. the vertical width distribution
is plotted in fig.
[ehsep]b and c for 70 and 120 gev electrons and pions incident on towers 2 through 9.
although the distributions do not change much with energy, there is a small reduction in spread and decrease in mean.
figures [ehsep]c and [ywidthy]a show that mean and standard deviation of the shower width distribution for hadrons does not change significantly with the incident vertical position. a @xmath26 cut chosen in the range 2 - 4 cm will clearly separate electrons from hadrons.
only 2.3% of hadrons incident at the center of the calorimeter will be lost with a @xmath26 cut of 3.5 cm ; the loss rises to 4.0% for hadrons incident at the center of a strip (fig.
[ehsep]).
the same analysis for electrons is complicated by pion contamination in the electron sample, which is shown as a broken line in fig.
[ehsep]. to subtract the pion contamination, we normalize the width distribution for pions to the raw (unsubtracted) distribution for electrons in the region @xmath73 cm.
the result is shown in fig.
[ywidthy]b for electrons incident in the center of a tower, and electrons scanned vertically from tower 6 to 7.
the inset shows the tail of the distribution for centered electrons.
about 1.5% of the electron showers have a width greater than 3.5.
although the signal from penetrating muons is low because of their small energy deposits, they can be useful for calibrating and monitoring the fnc in hera using either cosmic rays or beam halo muons.
to study the response of the calorimeter to high energy muons, a special trigger was implemented for beam particles which penetrated the calorimeter.
figure [muon] shows a typical energy distribution for a muon in the bottom front tower. the peak position @xmath74 is determined by fitting the distribution to a convenient functional form : @xmath75 and @xmath76, @xmath31, @xmath74, and @xmath77 are parameters.
when @xmath78, @xmath79 is the moyal function@xcite, an analytic approximation for the landau distribution.
the tower - to - tower spread of the peak position is 4%.
the absolute energy scale is best determined by making a beam of known energy incident on the calorimeter.
if deuterons were to be accelerated at hera, nuclear stripping reactions would make neutrons of known energy incident on the calorimeter. at present only protons
are accelerated, and the absolute energy scale must be determined using beam gas interactions which produce leading neutrons by charge exchange. if deuterons were accelerated at hera in order to test the feasibility of using charge exchange reactions to determine the absolute energy scale, a short experiment was performed.
figure [cern_tb] shows the experimental arrangement.
a lucite target was placed in the beam 20 m in front of the calorimeter. immediately after the target
a sweeping magnet was tuned to deflect beam particles downward. with the magnet tuned to full field beam particles (protons and pions)
were deflected downwards by 11 cm at the calorimeter.
it was necessary to devise a trigger that ensured that only events arising from the impact of a neutral particle in the central part of the calorimeter were accepted.
crossed finger counters in front of the target defined the beam position (see table [countertab] and fig.
[cern_tb]).
the trigger required a coincidence of these two counters and energy deposited in the center of the calorimeter for the trigger, the deposited energy was measured by summing and discriminating the signals from the last dynode of the pmts of the four central towers. in order to veto
charged particles, three scintillation counters were placed directly in front of the calorimeter. in addition, only showers near the center of the calorimeter were considered in order to reduce energy leakage. in summary
the data were required to satisfy : @xmath80 where @xmath81 the shower position @xmath14 was determined by light division ; @xmath15, by the centroid method with logarithmic weights. the observed energy distribution is shown in fig. [neutron].
in the same figure, the hatched histogram shows the response of the calorimeter to 120 gev hadrons.
the mean value of the latter distribution determines the energy scale for the figure.
the one pion exchange (ope) model of the reaction @xmath82 predicts that the cross section, integrated over scattering angles, is given by @xmath83 where @xmath84 is the flux of virtual pions associated with the incoming proton ; @xmath85 is the total pion - target cross section for a virtual pion beam energy of @xmath86.
any dependence of @xmath87 on the virtuality of the exchanged pion has been ignored in equation [eqn_ope], so @xmath87 can be estimated using a parameterization (@xmath88 gev), or an interpolation table (@xmath89 gev)@xcite.
several forms have been postulated for the flux factor @xmath84 in theoretical and phenomenological studies of ope.
they fall into two classes : regge@xcite and light cone@xcite.
accordingly, to compare data and theory, we extended the herwig@xcite monte carlo program to generate events according to ope folded with the measured experimental resolutions.
the predicted energy distribution, for the light cone flux factor@xcite, adjusted for resolution and acceptance is shown as the open histogram in fig.
[neutron].
the ope prediction is normalized to the data by the number of events. as can be seen, agreement is good for energies larger than 100 gev.
the energy scale was varied and then compared with the ope prediction for @xmath90 gev by computing the @xmath91 of the difference distribution.
the variation of @xmath91 with the scale factor is shown in the inset to fig.
[neutron].
the minimum occurs for a scale factor of 0.985.
if the regge@xcite form for the flux factor is used instead, the same procedure yields a minimum at 1.003, in complete agreement with the incident proton beam ; in this case, however, the minimum @xmath91 is worse (17 rather than 10).
we conclude that using charge exchange neutrons the absolute energy scale can be determined to better than 1.5% with the chief source of error being the theoretical uncertainty on the form of the virtual pion flux factor. | hep-ex9701015 |
Introduction
observations of starburst galaxies can provide vital insights into the processes and spectral characteristics of massive star formation regions. in such regions the physical conditions are similar to those that existed at the time of collapse and formation of galaxies in the early universe, and they can also provide an understanding of early galaxy evolution. the _ infrared astronomical satellite _ (iras) made the key discovery of large numbers of infrared luminous galaxies, similar to those found by @xcite. many of these are dominated by intense star formation @xcite in which the luminosity of the young hot stars heats the surrounding dust, producing large amounts of infrared radiation. the theoretical tools required to interpret the spectra of such galaxies are now available. for example, detailed stellar population synthesis models have been developed for both instantaneous and continuous starbursts and using these models, one is able to derive parameters such as the starburst age and metallicity from the continuous spectrum. in such models, the stellar initial mass function (imf), star formation rate (sfr) and stellar atmosphere formulations are all adjustable initial parameters. the emission line spectrum of the starburst provides constraints on the physical parameters for the ionized gas and the interstellar medium in general. in particular, the gas density, temperature and pressure can be derived directly from such observations, and the total rates of star formation can be estimated from the luminosity in the balmer lines of hydrogen for the objects without large quantities of dust at least (_ eg _ @xcite). using the ionizing uv radiation fields produced by stellar population synthesis models in conjunction with detailed self - consistent photoionization models such as mappings iii @xcite or cloudy @xcite we can now generate models for any region or starburst. in such models it is vital to include a self - consistent treatment of dust physics and the depletion of various elements out of the gas phase. since the nebular emission line spectrum is very sensitive to the hardness of the ionizing euv radiation, optical line ratio diagnostic diagrams provide an important constraint on the shape of the euv spectrum and these may also be used to estimate the mean ionization parameter and metallicity of the galaxies. such optical diagnostic diagrams were first proposed by @xcite to classify galaxies into starburst or agn type, since agn have a much harder ionizing spectrum than hot stars. the classification scheme was revised by @xcite and @xcite, hereafter vo87. these revised diagnostics are used here. for both schemes, the line diagnostic tools are based on emission - line intensity ratios which turn out to be particularly sensitive to the hardness of the euv radiation field. in an earlier paper @xcite, we theoretically recalibrated the extragalactic region sequence using these line diagnostic diagrams and others, in order to separate and quantify the effects of abundance, ionization parameter and continuous vs. instantanteous burst models. the theoretical region models were generated by the mappings iii code which uses as input the euv fields predicted by the stellar population synthesis models pegase v2.0 @xcite and starburst99 @xcite. dust photoelectric heating and the gas - phase depletion of the heavy elements were treated in a self - consistent manner. this work found that the high surface brightness isolated extragalactic regions are in general excited by young clusters of ob stars, and that, in this case, the ionizing euv spectra and region emission line spectra predicted by the pegase and starburst99 codes are essentially identical. for starburst galaxies, in which the starburst has a luminosity comparable to the luminosity of the host galaxy, the situation is rather different. in these objects, intense star formation is likely to continue over at least a galactic dynamical timescale, and therefore the assumption of a continuous rather than an instantaneous burst of star formation would be more accurate. as a consequence, the assumptions which go into the theoretical stellar mass loss formulations and evolutionary tracks are likely to play a much more important role in the modeling. furthermore, for starbursts continued for more than a few myr, the wolf - rayet (w - r) stars can play an important part in determining both the intensity and shape of the euv spectrum. for the w - r stars, the uncertain assumptions made about the stellar lifetimes, wind mass - loss rates, the velocity law in the stellar wind, and the atmospheric opacities play a critical role in determining the spectral shape and intensity of the emergent euv flux predicted by theory. in this paper, we present new grids of theoretical models (based on the assumption of continuous star formation) which again combine region models generated by the mappings iii code with input euv fields given by the stellar population spectral synthesis models pegase 2 and starburst99. these models are used in conjuction with our large observational data set described in @xcite and @xcite to place new _ observational _ constraints on the shape of the euv ionizing radiation field. since the two stellar population spectral synthesis codes provide a wide choice of stellar mass loss formulations, evolutionary tracks and stellar atmospheric transfer models, they provide strikingly different predictions about the shape and intensity of the euv field as a function of stellar age. in this paper, we use these to separate and quantify the effects of the stellar atmospheric models and the evolutionary tracks used on the optical diagnostic diagrams. in particular, we will show that the models which give the hardest euv spectrum below the ionization limit, but which have relatively few photons above this limit, provide the best empirical fit to the distribution of starburst galaxies on the optical line ratio diagnostic diagrams. these new observational constraints should prove very helpful to theoreticians modeling the late stages of stellar evolution in massive stars. this paper is structured as follows. our observational comparison sample is described in section 2. the stellar population synthesis models used to calculate the euv ionizing radiation field are presented in section 3. our theoretical starburst models were produced using the photoionization and shock code mappings iii and are described in section 4. wolf - rayet emission in our sample of starburst galaxies is discussed in section 5 and the effect of continuum metal opacities on the euv ionizing continuum is discussed in section 6. a new theoretical classification scheme for starbursts and agn is presented in section 7 and our main conclusions are summarised in section 8.
The observational comparison sample
we have selected a large sample of 285 warm iras galaxies covering a wide range of infrared luminosities. the warm selection criterion ensures that our sample contains galaxies with either concentrated star formation @xcite, agn @xcite, or in many cases, a combination of the two. our sample has been selected from the catalogue by @xcite and consists of all objects south of declination @xmath1 with the additional following criteria ; \1. flux at 60 @xmath2 with moderate or high quality detections at 25, 60 and 100 @xmath3 \2. redshift @xmath4 8000 km s@xmath5 for @xmath6 and @xmath7 30000 km s@xmath5 for @xmath8 \3. galactic latitude @xmath9 15@xmath10, and declination @xmath11 + 10@xmath12 \4. warm fir colours ; @xmath13 and @xmath14 where @xmath15,@xmath16 and @xmath17 are the iras fluxes at 25, 60 and 100 @xmath3 respectively. these selection criteria ensure that that the galaxies in the sample are well - resolved, and that the sample has a large dynamical range in luminosity, so that luminosity dependent effects can, in principle, be investigated. high resolution (50 km s@xmath5 at @xmath18) optical spectra with useable s / n (@xmath19) ratios were obtained for 225 of the galaxies in our sample. spectra were obtained in the red and blue wavelength ranges using the double beam spectrograph on the mount stromlo and siding springs 2.3 m telescope. these selection criteria and the full details of our observations are discussed in detail in @xcite and @xcite. all 225 objects were classified into agn, starburst and liner types using new theoretical classification lines on the @xcite (hereafter vo87) diagnostic diagrams. we found 157 starburst galaxies using the vo87 diagrams and semi - empirical classification scheme which provide the primary observational comparison for the starburst spectral modeling presented here. we believe there is little contamination by obscured agn in this starburst sample for two reasons. firstly, in @xcite, we have shown that the optical diagnostic diagrams are extremely sensitive to the presence of an agn. an agn which contributes only 20% to the optical emission increases the line ratios sufficiently that it would be classified as an agn. secondly, we found very few starburst galaxies with warm colors (@xmath20). since warm colors usually mean that the galaxy is energetically dominated by an agn we can conclude that the fraction of obscured agn in our starburst galaxies must be low. these results are consistent with the infrared studies of @xcite and @xcite, who showed that ultraluminous infrared galaxies classified as starbursts also show a lack of an energetically important agn.
Stellar population synthesis models
the models we have utilized are described in detail in @xcite, and we describe them again briefly here. we have used both the pegase 2 and starburst99 codes to model the starbursts in our sample. the pegase 2 code uses the @xcite grid of atmospheres covering the entire hertzsprung - russell diagram (hrd) plus @xcite planetary nebula nuclei (pnn) atmospheres for stars with high effective temperatives (t @xmath21 50000 k) (hereafter known as the clegg & middlemass atmospheres). the @xcite grid (hereafter called the lejeune grid) is derived from three sets of atmosphere calculations, the bulk being the @xcite models with smaller specialized cool star models by @xcite and @xcite. lejeune, cuisinier & buser incorporated observational flux corrections into these models for a range of stellar temperatures, but does not alter the @xmath22k models of kurucz. the starburst 99 code also uses the plane - parallel atmospheric lejeune grid. for stars with strong winds it offers the choice of the lejeune grid or the @xcite extended model atmospheres (hereafter known as the schmutz atmospheres). the prescription for the switch between extended and plane - parallel atmospheres is the same as in @xcite. we ran models for both the `` standard '' mass loss rates and the `` enhanced '' mass loss rates described in @xcite. we found that for the line diagnostic ratios used here, the mass loss prescriptions agree to within 0.03 dex. both codes follow the theoretical stellar tracks from the zero age main sequence (zams) to their final stages. these stages include the asymptotic and post - asymptotic giant branch phases for intermediate - mass stars. the padova tracks @xcite are used in pegase 2 and the geneva tracks (@xcite) are used in starburst99. the padova tracks overshoot for masses @xmath23 and use a higher ratio of the overshooting distance to the pressure scale height and down to lower masses than the geneva tracks, which include overshooting only above 1.5 @xmath24. both tracks use the opal opacities ; @xcite for the padova tracks and @xcite for the geneva tracks. both sets of tracks assume similar mixing lengths. helium contents of 0.28 and 0.30 are used for the padova and geneva tracks respectively. the padova tracks have a higher resolution in mass and time. clear differences between these two sets of tracks (at solar metallicity) can be seen by comparing figure (5) in @xcite and figure (7) in @xcite. assuming a standard initial mass function, the choices offered by the two spectral synthesis modeling codes allows sufficient flexibility to separately investigate, and to quantify, the effect that either the stellar atmospheres or the stellar evolutionary tracks have upon the theoretical starburst model line intensity ratios as a function of age. for example we can run the starburst99 code with either the lejeune atmospheres or the lejeune plus schmutz atmospheres to investigate the effect of extended atmospheres, or we could compare the pegase 2 and starburst99 codes to quantify the effect of the differences between the padova and the geneva tracks and the effect of the different stellar atmosphere models incorporated into these codes. we are confident that this comparison is valid, since in our earlier paper @xcite, we compared the ionizing euv spectra and region emission spectra predicted by the pegase and starburst99 codes for zero age clusters, and found these to be essentially identical. in the modeling of the starburst emission spectra, we distinguish between a zero - age _ instantaneous _ star formation case, and _ continuous _ starburst models in which a balance between star birth and star death is set up for all stellar masses which contribute significantly to the euv spectrum. the hydrogen - burning lifetime of massive stars is approximately @xmath25 myr, so in practice this condition is satisfied for any starburst which lasts longer than about 6 myr. this gives a dynamical balance between star births and star deaths for all masses greater than about @xmath26, and is also time enough for the wolf rayet stars to produce their full contribution to the euv spectrum. figure (1) compares predicted solar metallicity euv spectra produced by the pegase code which uses the lejeune stellar atmospheres plus clegg & middlemass atmospheres for stars with high effective temperatives (t @xmath21 50000k), and the starburst99 codes which use either the lejeune stellar atmospheres grid or the lejeune atmospheres plus schmutz extended model atmospheres. the pegase models used range from ages of @xmath27 myr, and the starburst99 models cover ages of @xmath28 myr. no further evolution in the shape of the euv spectrum is seen after 6 and 8 myr for the pegase and starburst99 models respectively. after a few myr of evolution, quite marked differences in the euv spectrum develop. the comparison of figure (1b) and figure (1c) shows that the schmutz extended atmospheres produce far more ionizing radiation at frequencies above the ionization limit than do the lejeune atmospheres, but the diffences are much less marked at lower energies. the differences between figures (1a - c) are due to a combination of the different evolutionary tracks, and the stellar atmosphere models used. figures (2) and (3) show the euv spectra from pegase 2 and starburst99 for metallicities 0.2 and 2 @xmath29 solar. we can observe clear differences in the euv spectra with increasing metallicity. in particular, the euv spectrum becomes harder for lower age models. this is expected if high mass stars are responsible for the euv field in this regime. at higher metallicities, the high mass stars make a larger contribution to the euv radiation field at younger burst ages. the most likely cause of the difference between the euv fields in figures (1a - c) is to be found in the different stellar atmospheric models used for the high mass stars, especially for the wolf - rayet stars. since it was specifically constructed to model starbursts, the starburst99 code uses a more theoretically sophisticated approach to modeling the euv spectrum. in the pegase 2 models, stars with effective temperatures greater than 50000 k (which includes wolf - rayet stars) are modeled by the clegg & middlemass planetary nebulae nuclei (pnn) atmospheres. these stars have much higher surface gravity than wolf - rayet stars, and so we would expect the atmospheric blanketing to be quite different. in starburst99 code, stars with strong stellar winds (which includes wolf - rayet stars) are modeled by the schmutz wolf - rayet atmospheres. these include h and he opacities, but do not include heavy element opacities. the euv spectrum emergent from a w - r model atmosphere depends critically on the fraction of ionizing photons which have been used up to maintain the ionization of the w - r wind region. this is determined by the size of the emission measure of the atmosphere ; @xmath30 this parameter is proportional to the the product @xmath31, where @xmath32 is the mass - loss rate, @xmath33 the terminal velocity of the wind, and @xmath34 is the photospheric radius of the star. this product is the @xcite density parameter. models with the same density parameter display very similar emission line equivalent widths, but the total scaling in luminosity depends on @xmath35 the density parameter can also be expressed in terms of a `` transformed radius '', @xmath36 : @xmath37 ^{2/3}\]] where @xmath38 is a (normalizing) reference velocity, and @xmath39 is a (normalizing) reference mass - loss rate. again, models with similar values of the transformed radius give similar spectra. stars which use a greater fraction of their euv photons in maintaining the photoionization of their extended atmospheres would show a lower - intensity, harder euv spectrum below the ionization edge, and would be expected to exhibit more atmospheric blanketing by heavy elements.
Starburst modeling
to model the starburst spectrum as a function of age of the exciting stars, metallicity and ionization parameter, we input the ionizing spectrum from the both continuous and instantaneous pegase and starburst99 models into the mappings iii code. the photoionization modeling carried out with mappings iii for this analysis is described in @xcite and is described briefly here. we computed plane parallel, isobaric models with electron density of 350 @xmath40, which is the average electron density of the individual (frequently unresolved) regions within the 1 kpc slit aperture extracted for the starbursts in our sample. the electron density was found using the flux in the [] @xmath41 and [] @xmath42 forbidden lines from our spectra in conjunction with a 5 level model atom using mappings iii. electron densities for the galaxies in our sample can be found in @xcite. the ionization parameter @xmath43 (cm s@xmath5) is defined on the inner boundary of the nebula, that is, the boundary nearest the exciting star. this dimensional ionization parameter can be readily transformed to the more commonly used dimensionless ionization parameter through the identity @xmath44 dust physics is treated explicitly through the absorption of the radiation field on grains, grain charging and photoelectric heating by the grains. we do not yet calculate the re - emission spectrum of dust in the ir but this is currently being implemented. the dust model consists of silicate grains (100 @xmath45 1000) and small amorphous organic grains (10 @xmath45 100), with a size distribution following a @xcite power - law and spherical geometry. the range of sizes is chosen so as to give depletion factors relative to solar similar to those observed by uv interstellar absorption measurements of stars seen through warm diffuse clouds in the local interstellar medium the mappings iii dust model also provides the observed absorption per hydrogen atom for solar metallicity @xcite : @xmath46. photoelectric yields are found using a more conservative yield curve of the same form as @xcite. photoelectric grain currents are found using @xcite using the dust absorption data from @xcite. collisions with electrons and protons are considered assuming the standard `` sticking '' coefficients, following @xcite and @xcite. more details of the dust physics in mappings iii can be found in @xcite. the undepleted solar abundances are assumed to be those of @xcite ; these abundances and the depletion factors adopted for each element in the starburst modeling are shown in table 1. for non - solar metallicities we assume that both the dust model and the depletion factors are unchanged, since we have no way of estimating what they may be otherwise. all elements except nitrogen and helium are taken to be primary nucleosynthesis elements. it is known that this assumption may be incorrect in systems where the time history of star formation in the galaxy is different, or where galactic winds are important. for example, the o / fe ratio is different from its solar value in both the lmc and the smc @xcite. again, we are forced use the simplest assumptions possible in the absence of a more detailed understanding of the chemical evolution of starburst galaxies. for helium, we assume a primary nucleosynthesis component in addition to the primordial value derived from @xcite. this primary component is matched empirically to provide the observed abundances at smc, lmc and solar abundances ; @xcite, @xcite. by number, the he / h ratio is : @xmath47 nitrogen is assumed to be a secondary nucleosynthesis element above metallicities of @xmath48 solar, but as a primary nucleosynthesis element at lower metallicities. this is an empirical fit to the observed behaviour of the n / o ratio in regions (van zee, haynes & salzer, 1997). by number : @xmath49 the abundances and depletion factors for each element used can be found in table [table1]. the ionization parameter was varied from @xmath50 to @xmath51, and the metallicities varied from 0.01 to 3 solar for pegase and 0.05 to 2 solar for starburst99. the metallicity values used are restricted by the stellar tracks used by the population synthesis models. figure (4) shows that, due to the absence of wolf - rayet stars, the use of similar zero - age main sequences, and the use of identical model atmospheres for massive stars, the shape of the euv spectrum for both the pegase 2 and starburst99 models is almost identical for instantaneous burst models. as a consequence, they produce almost identical optical line ratios in the photoionization models. the results from these models are compared with the observational data set on the vo87 line ratio diagnostic diagrams in figures (4), (5) and (6). note that the major problem seen with these models is that many starburst galaxies are found in a region lying above and to the right of the `` fold '' in the ionization parameter : metallicity surface. this presents a problem, since these points lie in a `` forbidden zone '' of line ratio space which can not be reached by any combination of metallicity or ionization parameter. the only way to make models which fall into this region of the the diagnostic diagrams is either to mix in another type of excitation _ i.e. _ shocks or a power - law ionizing radiation field or simply to use a harder euv ionizing spectrum, particularly in the 1 - 4 rydberg region. in any event, we should not be too surprised that the instantaneous models do not provide a very good fit to the observed starbursts, since many of these objects are seen in merging pairs of galaxies, and in this case we would theoretically expect star formation to continue over a galactic dynamical timescale. thus massive clusters associated with individual (usually unresolved) regions should have a wide variety of ages. direct evidence for continuous star formation, or older regions, is seen in many of our starburst spectra. these show either a low equivalent width in h@xmath52 (which often indicates dilution by an older stellar population) or clear evidence of h@xmath53 absorption in the stellar continuum. both of these indicate that star formation has continued over at least several myr @xcite. in @xcite, we showed that the line ratio usually used for measuring the ionization parameter ; [] @xmath545007 / [] @xmath553726,9, is indeed a good diagnostic and that the [] @xmath54 6584 / [] @xmath553726,9 ratio gives the best diagnostic of abundance, as it is monotonic between 0.1 and over 3.0 times solar metallicity. the wavelength range of our spectra was not sufficient for us to observe [], however we present the grids of the instantaneous models for the [] @xmath556548,84 / [] @xmath563726,9 _ vs. _ [] @xmath545007 /h@xmath52 diagram and the [] @xmath556548,84/ [] @xmath573726,9 _ vs. _ [] @xmath545007/ [] @xmath553726,9 in figures (7) and (8) for the use of the astronomical community. as we have seen, when star formation continues over several myr, the ionizing spectrum evolves until a dynamic balance between stellar births and stellar deaths has been set up for all intial stellar masses which are important in producing euv photons. as figure (1) shows, this occurs after 6 and 8 myr for the pegase and starburst99 models respectively, and so these cluster ages were assumed for the continuous star formation models presented in this section. the results for the continuous models shown on the vo87 line ratio diagnostic diagrams are presented in figures (9 - 11) for the pegase 2 models with (a) the lejeune plus clegg & middlemass pnn model atmospheres, (b) the starburst99 models with the lejeune plus schmutz model atmospheres and (c) the starburst99 models with the lejeune atmospheres respectively. let us first consider the two sets of starburst99 models. the differences seen here simply reflect the differences in the stellar atmospheric models. the lejeune plus schmutz model atmospheres show little change in spectral slope between 1 and 3 ryd as a function of cluster age. it is therefore not surprising that these continuous star formation models give very similar results to the zero age instantaneous models of figures (4 - 8). in these continuous star formation models the w - r stars provide a radiation field of appreciable strength above the ionization limit (eg. 4 - 8 ryd in figures 1 - 3). in this case, we might expect to detect lines in the optical spectrum. the detection of a nebular @xmath58 line would provide an important observational diagnostic in support of the schmutz extended atmospheric modeling and is discussed in the following section. these models with schmutz extended atmospheres have exactly the same difficulty in explaining the position of the observed points as do the instantaneous models, in that the model grid fails to overlap about half of the observed points, indicating the need for a harder ionizing spectrum than these tracks and atmospheres provide. the second of the starburst99 grids, for continuous starbursts using the geneva tracks along with the lejeune atmospheres provides an even softer radiation field in the @xmath59 ryd energy range. in this case, the theoretical grid falls below and to the left of the majority of the observed points on all three of the vo87 plots. this combination of tracks and atmospheres is therefore excluded with a good degree of certainty. finally, consider the pegase 2 models, which use the padova tracks with the lejeune plus clegg & middlemass pnn extended atmospheres. these models are characterised by the hardest radiation field in the 1 - 4 ryd region, and are the only ones that show the spectrum becoming harder as the w - r stars turn on. this is a direct consequence of the clegg & middlemass pnn atmosphere models used for stars with effective temperatures greater than 50000 k. these are the only set of models which encompass nearly all of the observed starbursts on all three of the vo87 line diagnostic plots. furthermore, individual objects fall into similar regions of the theoretical grid in all three diagnostic plots, indicating that there is some degree of consistency achieved here. in essence, the ionization parameter appears to be limited to the range @xmath60 and the metallicity range seems to cover most of the range ; from about @xmath61 solar up to nearly 3 times solar. however, the low - metallicity objects appear to be rather rare in our sample, and most of the starbursts are consistent with a metallicity 1 - 3 times solar. we present the grids of the continuous models for the [] @xmath556548,84 / [] @xmath563726,9 _ vs. _ [] @xmath545007 /h@xmath52 diagram and the [] @xmath556548,84/ [] @xmath573726,9 _ vs. _ [] @xmath545007/ [] @xmath553726,9 in figures (12) and (12) for the use of the astronomical community. the starburst models presented here are based on pure photoionization models, and therefore do not incorporate the effect of mechanical luminosity from supernovae shocks. the primary effect of the release of mechanical energy through shocks is to move the theoretical grids upwards, and lightly to the right on the vo87 line ratio diagnostic plots. the size of this effect depends on the relative importance of the shock and photoionization luminosity. the input mechanical energy luminosity @xmath62 produced by supernova events and winds is converted into optical line emission through radiative shocks. the importance of this line emission relative to that produced by photoionization depends on fraction of this energy, @xmath63, which has been converted to h@xmath53 flux compared to the h@xmath53 flux produced by recombinations in the photoionized plasma : @xmath64 where @xmath65 is the effective recombination coefficient of hydrogen, and @xmath66 is the number of ionizing photons being produced by the hot stars in the cluster. in the case of a supernova remnant, the shock becomes radiative, and the sedov - taylor (adiabatic) expansion phase is terminated when its cooling timescale, @xmath67, becomes shorter than the dynamical expansion timescale of the shell, @xmath68. from standard sedov - taylor theory, @xmath69 where @xmath70 is the radius of the supernova remnant, @xmath71 is the shock wave velocity, and @xmath72 is the time since the supernova explosion. the radiative shock - wave theory of dopita & sutherland (1996) shows that the cooling timescale, in years, can be expressed as : @xmath73 where @xmath74 is the shock velocity in units of 100 km s@xmath75 @xmath76 is the metallicity of the plasma relative to solar, and @xmath77 is the pre - shock density. we therefore find that, with @xmath78 @xmath79, a supernova remnant typically become radiative at a radius of 1 pc, when the shock velocity is 600 km s@xmath80 at this point, the supernova remnant is @xmath81600 years old, and its expansion timescale is @xmath811500 yr. we therefore used a shock model with a velocity of 600 km s@xmath5 in a spherical nebula of solar metallicity and @xmath78 @xmath79 to calculate the possible contribution by supernova remnants to our photoionization models. we assumed a total mechanical luminosity in snr of @xmath82 @xcite. we calculated a star formation rate of @xmath83 from the average @xmath84for the starbursts in our sample, following the prescription given in @xcite (note that @xmath84 is defined as @xmath85 in @xcite). assuming the ir luminosity is distributed evenly throughout the galaxy and a minimum size of 7 kpc, the sfr reduces to @xmath86 in the observed aperture. we note that this is still higher than that found using the h@xmath87 luminosity of our template starburst (within the 1 kpc aperture) of @xmath88. this is to be expected as dust absorption will reduce the h@xmath87 derived sfr compared with that derived using the fir luminosity. we computed a 600 km / s shock model and a spherical ionized precursor with a shocked spherical radius of 1 pc. this model had a mechanical luminosity of @xmath89 erg / s. as we expect the total mechanical luminosity to be @xmath90 erg / s, we expect to obtain on average 11.2 snr within our 1 kpc aperture at any one time. the luminosity produced in the shock and precursor are given in table (2). from this snr model, the contribution to h@xmath52 emission in the starbursts due to snrs would be about 16 - 20%. the contribution to the h@xmath52 flux from starbursts is determined more or less directly by the star formation history and the imf. the contribution to h@xmath52 from snrs is determined by the velocity and the total area covered by the snr shocks, which in turn depend on the number of snrs (determined by the sfr and imf) and weakly on the density. at low densities, snrs are larger and have lower velocities than at higher densities, and have similar overall luminosities in h@xmath52. the [] emission from the starbursts, on the other hand, is not a function of the sfr, rather it is a measure of the ionization parameter @xmath43 of the radiation field within the starburst which is a function of the density and mass distribution (ie filaments vs uniform) of the gas. thus the [] in the starburst models varies from bright to negligible levels compared to h@xmath52. the [] mission from the snrs is constrained by the chosen model geometry and is therefore determined by the sfr and the average density, not by the global ionization parameter. the ionization parameter for each snr is determined internally by the snr model without reference to the global starburst value. the observed total @xmath91@xmath92/{\rm h}\beta)$] ratio is the sum of the starburst and snr contributions ; @xmath93/{\rm h}\beta) = \log([{\rm oiii}]_{\rm starb } + [{ \rm oiii}]_{\rm snr }) - \log({\rm h\beta_{starb}+h\beta_{snr}}).\]] thus, in the limit of @xmath94_{\rm starb}=0 $] due to low global @xmath43 in the starbursts, the observed @xmath91@xmath92/{\rm h}\beta)$] ratio will reach a lower limit of @xmath93_{\rm snr }) - \log({\rm h\beta_{starb}+h\beta_{snr } }) \ge 0.0\]] with a density of 350 @xmath79 and the sfr used here (@xmath86 in the observed aperture of 1 square kpc). this lower limit is not observed, and the actual lower limit is @xmath95/h}\beta) \approx -1.0 $], a factor of 10 less than the snr model limit. clearly the strong @xmath94_{\rm snr}$] contribution modeled here is not compatible with the observations. this incompatibility may be due to either the number of snr we derived, or that the average density in the snr environment is less than 350 @xmath79. we conclude that the snr contribution to the @xmath91@xmath92/{\rm h}\beta)$] ratio is @xmath96% and is probably a factor of 10 less (@xmath97%), and can be neglected for the starburst models derived here. we are currently investigating this snr contribution to starbursts further by considering a range of densities and lower shock velocities to determine the level at which the snrs model is compatible with the observations. we expect to find that velocities in the 200 - 300 km / s range will be more compatible with observations, resulting in a small @xmath94_{\rm snr}$] contribution, negligible in all but the starbursts with the lowest ionization parameter.
Wolf-rayet emission in starburst galaxies
wolf - rayet features were first found in the dwarf emission galaxy he2 - 10 @xcite. as more such features were discovered in galaxies, @xcite defined wolf - rayet (w - r) galaxies as those galaxies which contain broad stellar emission lines in their spectra and therefore contain large numbers of w - r stars. the large numbers of w - r stars are thought to be a result of present or very recent star formation. @xcite searched for w - r emission in a sample of blue emission - line galaxies. they found a positive detection in one galaxy and suspected in 14 others and suggested that w - r stars are preferentially detected in low redshift galaxies. the first catalogue of w - r galaxies was presented by @xcite, showing that w - r galaxies can be easily distinguished by their broad @xmath544686 [] emission feature, or in some cases a broad line at 4640 due to. high resolution long - slit observations of a sample of wolf - rayet galaxies were carried out recently by @xcite. nearly all the galaxies in their sample show broad w - r emission consisting of an unresolved blend of @xmath544640, @xmath544650, [] @xmath544658, and @xmath544686 emission lines. they also found weaker w - r emission lines @xmath544512 and @xmath544565 in some galaxies. the signal - to - noise ratios of our spectra are not sufficient in individual galaxies to detect these signatures of w - r emission. we have therefore constructed a template `` average '' warm infrared starburst spectrum from the 56 starburst galaxies in our sample which have snrs at h@xmath53 of 60 or greater and for which the zero - redshift blue wavelength cut - off is lower than @xmath984620. we note that this average spectrum is not representative of our sample of warm infrared starbursts because selecting for high snr galaxies may bias the average spectrum towards young and hence more luminous starbursts. however, it is useful simply to assist in the search for w - r features. the average spectrum is shown in figure (14) and in detail in figure (15). the positions of the expected w - r features are marked on figure (15). the mean age of the average warm infrared starburst can be estimated from the h@xmath53 absorption equivalent width, bearing in mind that dust absorption may also contribute to the absorption line profile and has not been taken into account. the h@xmath53 absorption equivalent width was found by simultaneously fitting the h@xmath53 absorption and emission lines with gaussians in the iraf task _ ngaussfits_. we find that the h@xmath53 absorption equivalent width is @xmath99 which corresponds to an upper limit age of @xmath100 myr for a continous star formation model at solar metallicity @xcite. however, this is probably an underestimate, since the h@xmath53 absorption equivalent width will be reduced by any underlying old star population. nonetheless, there is likely to be likely a selection affect towards more youthful starbursts, in that the bright starbursts tend to be both younger and intrinsically more luminous. in the `` average '' spectrum of figure (15), there appears to be marginal detections of [] @xmath984658, and @xmath984686 at the 2@xmath101 level. these lines do not appear broad, however this may simply be an effect of the low snr for these lines. the @xmath984686 emission in the majority of our galaxies provides an important constraint on the schmutz extended atmospheric modeling. the @xmath102(@xmath103)ratio for our template starburst galaxy is @xmath104. both the mappings iii models with pegase and starburst99 (lejeune atmospheres) stellar ionizing continua produce @xmath102(@xmath103) around @xmath105, while the mappings iii models with starburst99 (lejeune + schmutz atmospheres) stellar ionizing continuum produces @xmath102(@xmath103)around @xmath106, consistent with that observed with our template starburst galaxy. we note again that our template starburst galaxy has been composed of the starbursts in our sample with the highest snrs, and so may be biased towards the most luminous and therefore the youngest starbursts in our sample. this @xmath103ratio for our template starburst may therefore be biased towards those with more w - r stars than the average starburst galaxy in our sample. clearly we require a hard euv field in the 1 - 4 ryd regime to model our starburst galaxies on the optical diagnostic diagrams such as that provided with the pegase 2 model. however, as we would expect the schmutz extended atmospheric modeling to be more appropriate to the modeling of our starburst galaxies, we believe that the inclusion of continuum metal opacity (or continuum metal blanketing) in the stellar population synthesis models using the schmutz extended atmospheres is a possible solution.
Continuum metal opacity
for many years, optical depth estimates in the ism were found using the hydrogen 21 cm line and the assumption of a uniform gas density. as a result of the low optical depths obtained, it was thought that the ism should be opaque to radiation in the euv @xcite. the discovery that the ism is inhomogeneous overturned this conclusion and @xcite established an estimate of the effective absorption cross - section of the ism at euv wavelengths using determinations of cross - sections and abundances. @xcite showed that some euv radiation should be able to be observed over considerable distances. more recent determinations of cross - sections and abundances have been used by @xcite to provide a new estimate of the effective absorption cross - section of the ism at euv wavelengths. the relative difficulty in obtaining euv spectra to compare with theoretical estimates of opacity in the euv continuum described above has decreased significantly over the last decade with the aid of euv telescopes such as euve (extreme ultraviolet explorer), rosat wide field camera, alexis (array of low energy x - ray imaging sensors), fuse (far ultraviolet spectroscopic explorer), and the extreme - ultraviolet imaging telescope (eit). these telescopes have allowed observations of many stellar objects which have greatly advanced the modeling of stellar atmospheres (*??? * eg.) and have allowed the euv study of some seyfert galaxies (eg. however the euv spectrum in starburst galaxies continues to remain unseen as a result of the weakness of the euv continuum in starburst galaxies due to absorption. it is therefore necessary to rely on theoretical predictions of the euv spectrum from stellar population synthesis codes. although the euv spectrum must be estimated using theoretical modeling, the reprocessing of the euv spectrum into optical emission lines allows limits to be placed on the shape of the euv spectrum in starburst galaxies. as concluded in previous sections, we require a harder euv field between the 1 - 4 ryd regime than is provided by the stellar population synthesis models with schmutz extended atmospheres. currently, only continuous opacities due to helium are included in the models, as continuous metal opacities are relatively unimportant for many studies. we believe that the inclusion of continuum metal opacities in these models would provide a suitable shape in the euv, enabling the models to reproduce the starburst sequence on the standard optical diagnostic diagrams. continuum metal opacities, as with hydrogen and helium opacities, are a result of bound - free transitions, ie photoionization of the metals. continuum metal opacity would allow some fraction of the radiation with energies greater than 4 ryd to be absorbed and re - emitted less than 4 ryd. the fraction of radiation absorbed by metals depends on their individual absorption cross - sections and abundances. the resulting euv continuum would have a softer continuum above the ionization limit, as a result of carbon opacities. the spectrum would also have a harder but fainter euv field between the 1 - 4 ryd regime, as required by the position of our starburst galaxies on the optical diagnostic diagrams. note that if this were the case, we might expect that the models with schmutz extended atmospheres would produce too much flux in the @xmath107 line compared with our observations, contrary to our findings. we conclude that while continuum metal blanketing may be a possible solution to the discrepancy seen between our observations and the models using schmutz extended atmospheres, it may not be the only solution.
Extreme starburst classification line
in order to place a theoretical upper limit for starburst models on the optical diagnostic diagrams, we used the pegase 2.0 grids, since these provide the hardest euv spectrum, and therefore give a theoretical grid which is placed both highest and furthest to the right on the vo87 line ratio diagnostic diagrams. we have shown that, with a realistic range of metallicities (@xmath108) and ionization parameter @xmath43 (cm / s) in the range @xmath109 (or @xmath110), continuous starburst models produced by any modeling procedure always fall below and to the left of an empirical limit on the [] /@xmath111 _ vs. _ [] /@xmath112, [] /@xmath111 _ vs. _ [] /@xmath112 and [] /@xmath111 _ vs. _ [] /@xmath112 diagrams. this is due to the two - parameter grid of the ionization parameter and metallicity folding back upon itself. therefore, no combination of these parameters can generate a theoretical point above this fold. the lines dividing the theoretical starburst region from objects of other types of excitation are shown in figure (16). these can be parametrized by the following simple fitting formulae, which have the shape of rectangular hyperbolae ; @xmath113\,\lambda 5007}}{{\rm h}\beta}\right) = \frac{0.61 } { \log ({ \rm [nii]}/{\rm h}\alpha) -0.47}+1.19\]] @xmath114\,\lambda 5007}{{\rm h}\beta}\right) = \frac{0.72 } { \log ({ \rm [sii]\,\lambda \lambda 6717,31}/{\rm h}\alpha) -0.32 } + 1.30\]] @xmath114\,\lambda 5007}{{\rm h}\beta}\right) = \frac{0.73 } { \log ({ \rm [oi]\,\lambda 6300}/{\rm h}\alpha) + 0.59}+1.33\]] the shape and position of this maximum starburst line has not been previously established from theoretical models, since the shape of the ionizing spectrum of the cluster has not been known to sufficient accuracy. vo87 have attempted to establish both the position and the shape of this boundary in a semi - empirical way, using both observational data from the literature and a combination of models available to them at that time. the theoretical boundaries for starbursts defined by equations (1)-(3) provide us for the first time with a theoretical (as opposed to a semi - empirical boundary) for the region occupied by starbursts in these diagnostic plots. in view of the potential for errors in the modeling which may flow from errors in the assumptions made in the chemical abundances, chemical depletion factors, the slope of the initial mass function, and the evolutionary tracks and the stellar atmosphere models used, we have indicated a `` best guess '' estimate of these errors as dashed lines in figure (16). these theoretical upper limits for starburst galaxies have been used in @xcite to classify the galaxies in our sample along with an extreme mixing line produced using our shock modeling to classify galaxies into starburst, liner and agn types. the strength of our theoretical starburst classification line can be seen by observing the number of galaxies which have ` ambiguous'classifications. these galaxies are those which fall within the starburst region of one or two of the diagnostic diagram(s) and the agn region of the remaining diagram(s). we found that only 6% of our sample have ambiguous classifications using our theoretical extreme starburst line, compared with 16% ambiguous classifications using the standard vo87 classification lines. these results indicate that our theoretical starburst line is a reliable tool for optically classifying galaxies into starburst and agn types, and is more consistent from diagram to diagram than the conventional vo87 method.
Conclusions
we have presented a comparison between the stellar population synthesis models pegase 2 and starburst99 using a large sample of 157 warm infrared starburst galaxies. the main differences between the two synthesis models are the stellar tracks and stellar atmosphere prescriptions. pegase and starburst99 were used to generate the spectral energy distribution (sed) of the young star clusters. mappings iii was used to compute photoionization models which include a self - consistent treatment of dust physics and chemical depletion. the standard optical diagnostic diagrams are indicators of the hardness of the euv radiation field in the starburst galaxies. these diagnostic diagrams are most sensitive to the spectral index of the ionizing radiation field in the 1 - 4 rydberg region. we find that warm infrared starburst galaxies contain a relatively hard euv field in this region. the pegase ionizing stellar continuum is harder in the 1 - 4 ryd range than that of starburst99, most likely due to the different stellar atmospheres used for wolf - rayet stars. we have constructed an average spectrum of the high snr warm infrared starbursts in our sample in order to look for wolf - rayet signatures. we find detections of @xmath544658, and @xmath544686 at the 2@xmath115 level, indicating w - r activity, and constraining the schmutz extended atmospheric modeling. we require a hard euv field in the 1 - 4 ryd regime to model our starburst galaxies on the optical diagnostic diagrams such as that provided with the pegase 2 model. however, as we would expect the schmutz extended atmospheric modeling to be more appropriate to the modeling of our starburst galaxies, we believe that one solution would be to include continuum metal blanketing in the stellar population synthesis models using the schmutz extended atmospheres. continuum metal blanketing would allow much of the radiation with energies greater than 4 ryd to be absorbed and re - emitted at energies less than 4 ryd. the resulting euv continuum would have a softer continuum above the ionization limit and a harder euv field between the 1 - 4 ryd regime, as required by the position of our starburst galaxies on the optical diagnostic diagrams. snrs could also contribute to the hardness of the euv field, although our models and observations suggest that this is likely to be @xmath96%, insufficient to cause the discrepancy between our starburst galaxies and the models using schmutz extended atmospheres. we use the starburst grids produced with the pegase euv ionizing radiation field and our mappings iii models to parametrize an extreme starburst line which is useful in classifying galaxies into starburst or agn types. in a previous paper @xcite, we showed that this theoretical classification scheme produces reliable classifications with less ambiguity than the classical @xcite empirical method.
Acknowledgements
we thank claus leitherer and brigitte rocca - volmerange for helpful discussions and for the use of starburst99 and pegase 2. we thank the referee for his constructive comments which helped this to be a better paper. we also thank the staff at siding springs observatories for their assistance during our spectroscopy observations. l. kewley gratefully acknowledges support from the australian academy of science young researcher scheme and the french service culturel & scientifique. lrr h & 0.00 & 0.00 + he & -1.01 & 0.00 + c & -3.44 & -0.30 + n & -3.95 & -0.22 + o & -3.07 & -0.22 + ne & -3.91 & 0.00 + mg & -4.42 & -0.70 + si & -4.45 & -1.00 + s & -4.79 & 0.00 + ar & -5.44 & 0.00 + ca & -5.64 & -2.52 + fe & -4.33 & -2.00 + | we have modeled a large sample of infrared starburst galaxies using both the pegase v2.0 and starburst99 codes to generate the spectral energy distribution (sed) of the young star clusters.
pegase utilizes the padova group tracks while starburst99 uses the geneva group tracks, allowing comparison between the two.
we used our mappings iii code to compute photoionization models which include a self - consistent treatment of dust physics and chemical depletion.
we use the standard optical diagnostic diagrams as indicators of the hardness of the euv radiation field in these galaxies.
these diagnostic diagrams are most sensitive to the spectral index of the ionizing radiation field in the 1 - 4 rydberg region.
we find that warm infrared starburst galaxies contain a relatively hard euv field in this region.
the pegase ionizing stellar continuum is harder in the 1 - 4 rydberg range than that of starburst99.
as the spectrum in this regime is dominated by emission from wolf - rayet (w - r) stars, this difference is most likely due to the differences in stellar atmosphere models used for the w - r stars.
the pegase models use the @xcite planetary nebula nuclei (pnn) atmosphere models for the wolf - rayet stars whereas the starburst99 models use the @xcite wolf - rayet atmosphere models.
we believe that the @xcite atmospheres are more applicable to the starburst galaxies in our sample, however they do not produce the hard euv field in the 1 - 4 rydberg region required by our observations.
the inclusion of continuum metal blanketing in the models may be one solution.
supernova remnant (snr) shock modeling shows that the contribution by mechanical energy from snrs to the photoionization models is @xmath0%.
the models presented here are used to derive a new theoretical classification scheme for starbursts and agn galaxies based on the optical diagnostic diagrams. | astro-ph0106324 |
Introduction
the hubble space telescope and advances in ground based observing have greatly increased our knowledge of the galaxy population in the distant universe. however, the nature of these galaxies and their evolutionary connections to local galaxies remain poorly understood. luminous, compact, star forming galaxies appear to represent a prominent phase in the early history of galaxy formation @xcite. in particular : * the number density of luminous, compact star forming galaxies rises significantly out to z @xmath1 1 @xcite. * the lyman break galaxies at z @xmath12 2 seen in the hubble deep field are characterized by very compact cores and a high surface brightness @xcite. * sub - millimeter imaging has revealed distant galaxies (z @xmath1 2@xmath134), half of them compact objects, which may be responsible for as much as half of the total star formation rate in the early universe @xcite. however, little is definitively known of their physical properties, or how they are related to subsets of the local galaxy population. a classification for known examples of intermediate redshift (0.4 @xmath0 z @xmath0 0.7) luminous, blue, compact galaxies, such as blue nucleated galaxies, compact narrow emission line galaxies, and small blue galaxies, has been developed by @xcite in order to be able to choose samples over a wide redshift range. they have found that the bulk of these galaxies, collectively termed luminous compact blue galaxies (lcbgs), can be distinguished quantitatively from local normal galaxies by their blue color, small size, high luminosity, and high surface brightness. (see 2.1 for more detail.) from studies at intermediate redshifts, it has been found that lcbgs are a heterogeneous class of vigorously starbursting, high metallicity galaxies with an underlying older stellar population @xcite. while common at intermediate redshifts, they are rare locally @xcite and little is known about the class as a whole, nor their evolutionary connections to other galaxies. lcbgs undergo dramatic evolution : at z @xmath1 1, they are numerous and have a total star formation rate density equal to that of grand - design spirals at that time. however, by z @xmath1 0, the number density and star formation rate density of lcbgs has decreased by at least a factor of ten @xcite. since the lcbg population is morphologically and spectroscopically diverse, these galaxies are unlikely to evolve into one homogeneous galaxy class. @xcite and @xcite suggest that a subset of lcbgs at intermediate redshifts may be the progenitors of local low - mass dwarf elliptical galaxies such as ngc 205. alternatively, @xcite and @xcite suggest that others may be disk galaxies in the process of building a bulge to become local l@xmath2 spiral galaxies. clearly, to determine the most likely evolutionary scenarios for intermediate redshift lcbgs, it is necessary to know their masses and the timescale of their starburst activity. are they comparable to today s massive or low - mass galaxies? are they small starbursting galaxies which will soon exhaust their gas and eventually fade? or are they larger galaxies with only moderate amounts of star formation? only kinematic line widths that truly reflect the masses of these galaxies, as well as measures of their gas content and star formation rates, can answer these questions. using ionized gas emission line widths, @xcite, @xcite, and @xcite, have found that lcbgs have mass - to - light ratios approximately ten times smaller than typical local l@xmath2 galaxies. however, since ionized gas emission lines may originate primarily from the central regions of galaxies, their line widths may underestimate the gravitational potential @xcite. h emission lines provide a better estimate of the total galaxy mass as they measure the gravitational potential out to larger galactic radii. observations of both h and co (the best tracer of cold h@xmath14), combined with star formation rates, are necessary to estimate the starburst timescales. with current radio instrumentation, h and co can only easily be measured in very nearby lcbgs, at distances @xmath0 150 mpc for h , and @xmath0 70 mpc for co. therefore, to understand the nature and evolutionary possibilities of higher redshift lcbgs, we have undertaken a survey in h 21 cm emission and multiple rotational transitions of co of a sample of 20 local lcbgs, drawn from the sloan digital sky survey @xcite. this work, paper i, reports the optical photometric properties of our sample and the results of the h 21 cm portion of the survey, including dynamical masses and comparisons with local galaxy types. paper ii @xcite will report the results of a survey of the molecular gas conditions. knowledge of the dynamical masses, combined with gas masses and star formation rates, constrains the evolutionary possibilities of these galaxies. nearby blue compact galaxies (bcgs) have been studied extensively at radio and optical wavelengths since @xcite originated the term `` compact galaxy '' and @xcite distinguished between `` red '' and `` blue '' compact galaxies. the term bcg typically refers to galaxies with a compact nature, a high mean surface brightness, and emission lines superposed on a blue continuum. however, many different selection criteria have been used, leading to various definitions of bcgs and samples with a range of properties. for example, the term `` dwarf '' has been used to mean bcgs fainter than @xmath1317 (e.g. thuan & martin 1981 ; kong & cheng 2002) or @xmath1318 blue magnitudes (e.g. taylor et al. 1994), or an optical diameter less than 10 kpc @xcite. the term `` blue '' has been used to mean blue on the palomar sky survey plate (e.g. gordon & gottesman 1981), or to have emission lines superposed on a blue background (e.g. thuan & martin 1981). the term `` compact '' has been used to mean smaller than 1 kpc in optical diameter (e.g. thuan & martin 1981), or qualitatively compact (e.g. doublier et al. some, for example @xcite, began using the term blue compact dwarf to refer to the common low luminosity, low metallicity bcgs. as high luminosity (@xmath1l@xmath2), nearby bcgs are rare, many began to use the term blue compact dwarf for all nearby bcgs, regardless of luminosity. there are only a few of the rare local lcbgs (by the @xcite definition) in previous bcg surveys. note that @xcite have recently been studying local luminous bcgs. however, their selection criteria are not as stringent as that of @xcite, and they have been focusing on very low metallicity (less than 15% solar) galaxies. intermediate redshift lcbgs with rest - frame properties matching the class definition of @xcite have metallicities at least 40% solar @xcite. four of our local lcbgs have metallicities available in the literature, which range from 40 @xmath13 70 % solar @xcite. by using the selection criteria of @xcite, we ensure that our study is of those local lcbgs defined to be analogs to the widely studied higher redshift lcbgs. the observations, including sample selection and data reduction, are described in @xmath152. the optical photometric properties, h 21 cm spectra, measurements, and derived properties are presented in @xmath153, and analyzed in @xmath154. we compare the derived physical properties to local normal galaxies and higher redshift lcbgs in @xmath155, and conclude in @xmath156. we assume h@xmath16 = 70 km s@xmath17 mpc@xmath17 throughout. when we compare our results to those of other authors, we scale their results to this value.
The observations
when @xcite compared intermediate redshift lcbgs with local normal galaxies, they found that lcbgs can be isolated quantitatively on the basis of color, surface brightness, image concentration, and asymmetry. color and surface brightness were found to give the best leverage for separating lcbgs from normal galaxies. specifically, lcbgs can be defined by a region limited to b@xmath13v @xmath18 0.6, and a b - band surface brightness within the half - light radius, sbe, brighter than 21 b - mag arc sec@xmath19. this simple definition differs only slightly from the formal definition in @xcite which uses a color - dependent sbe. @xcite also applied a luminosity cut - off of 25% l@xmath20m@xmath21 @xmath18 @xmath1318.5), to distinguish lcbgs from blue compact galaxies. lcbgs are not so extreme that they are completely separated from the continuum of normal galaxies. the sharp borders used to classify them are artificial, but serve the purpose of defining similar objects over a range of redshifts. lcbgs at intermediate redshifts (z @xmath1 0.6) can be studied in deep spectroscopic surveys (i @xmath1 24), while the brightest lcbgs at high redshifts (z @xmath1 3) require very deep spectroscopic surveys (i @xmath1 26). therefore, lcbgs selected in this manner are observable over a wide range of redshifts. using these color, surface brightness, and luminosity criteria, we selected our sample of local lcbgs from the sloan digital sky survey (sdss). begun in 2000, this survey will ultimately image one quarter of the sky using a large format ccd camera on a 2.5 m telescope at apache point observatory in new mexico. images are taken in five broad bands (u, g, r, i, z) which range from 3540 (u) to 9130 (z). the survey has a limiting magnitude of 22.2 in g (4770 ) and r (6230 ), our two bands of interest. after searching through approximately one million galaxies (@xmath11500 degree@xmath22 on the sky), we identified only 16 nearby (d @xmath0 70 mpc) lcbgs. this distance cut - off was chosen to ensure our galaxies could be detected quickly in both h and co. we added four markarian galaxies from the literature which fulfilled our selection criteria and were not yet surveyed by sdss, for a total of 20 local lcbgs. the color and surface brightness characteristics of our local sample, as well as higher redshift (0.4 @xmath0 z @xmath0 1) samples of lcbgs, are compared to other nearby galaxies in figure 1. we note that while our sample is not complete, it is representative of local lcbgs (see castander et al. 2004). as at higher redshifts @xcite, our local lcbg sample is a morphologically heterogeneous mixture of galaxies, including sb to sc spirals, s0 galaxies, polar ring galaxies, peculiar galaxies, and h liners and starbursts, as classified by hyperleda and the nasa@xmath23ipac extragalactic database (ned). approximately one quarter of these galaxies are not classified in either database. roughly one third are members of multiple, sometimes interacting or merging, systems. sdss and digitized sky survey images of our sample of 20 galaxies are shown in figures 2 and 3. the properties of our sample of local lcbgs are described below, and listed in table 1. _ full sdss galaxy designation of the form sdss jhhmmss.ss@xmath24ddmmss.s, in the j2000 system. in the remainder of the paper, individual galaxies are referred to by an abbreviated sdss name of the form sdssjhhmm@xmath24ddmm. the four non - sdss galaxies are referred to by their markarian (mrk) names. _ alternate name. _ mrk and/or ngc designation, if any. _ d@xmath25. _ hubble distance, in mpc, calculated from the optical redshift (from sdss, except for the non - sdss galaxies for which we use ned redshifts.) _ m@xmath21. _ absolute blue magnitude calculated from the apparent blue magnitude, m@xmath21, using d@xmath25. for the sdss galaxies, m@xmath21 = g + 0.30(g@xmath13r) + 0.18, b@xmath13v, and r@xmath26(b) are based on synthetic spectra which fit the observed spectral energy distribution of lcbgs.]. the r and g magnitudes were calculated from the sdss `` petrosian '' and `` model '' magnitudes : r = r(petro) and g = r(petro) @xmath13 [r(model) @xmath13 g(model)] @xcite. the sdss galactic reddening corrections were applied, but no k corrections were applied as our galaxies are nearby, and many are of unknown spectral type. the k correction is at most 0.015 magnitudes in r and 0.038 magnitudes in g @xcite. no correction was applied for extinction due to inclination, also because of the uncertainty in spectral types. for the non - sdss galaxies, the `` total apparent blue magnitudes '' were used from hyperleda, and corrected for galactic reddening using extinction values from ned. _ b@xmath13v. _ the color transformation is b@xmath13v @xmath27 0.900(g@xmath13r) + 0.145@xmath28, where the magnitudes were calculated as above, for the sdss galaxies. the `` total b@xmath13v colors '' from hyperleda were used for the non - sdss galaxies. they were corrected for reddening using extinction values from ned. _ average surface brightness in the b band within the half light or effective radius in b - mag arc sec@xmath19. for the sdss galaxies, sbe(b) = m@xmath21 + 2.5 @xmath29 [2@xmath30r@xmath31(b)], where r@xmath26(b), the effective or half - light radius in the b band, was calculated from the `` petrosian '' effective radii in the g and r bands [r@xmath26(b) = 1.30 @xmath32(g) @xmath13 0.300 @xmath32(r)]@xmath28. we corrected sbe for cosmological dimming by subtracting 7.5 @xmath29(1 + z), where the redshift, z, was calculated using the h velocity (@xmath153.1). for the non - sdss galaxies, we used sbe from hyperleda (termed the `` mean effective surface brightness ''), and then corrected it for cosmological dimming in the same way. _ ned type. _ morphological type as given by ned, which does not use a homogeneous system. _ hyperleda type. _ morphological type as given by hyperleda, which uses the @xcite system. _ other sources in beam? _ indicates whether or not each galaxy has other sources at similar velocities within the 9@xmath33.2 beam of the green bank telescope (used for the h observations), as given by ned. we compared hyperleda and sdss magnitudes for our sdss selected galaxies to ensure that there are no systematic offsets between the two. fourteen galaxies had magnitudes measured in both systems. the median difference between sdss and hyperleda magnitudes (as calculated in @xmath152.1.2) is 0.1 magnitudes. since the standard deviation of these differences is @xmath34 0.8 magnitudes, we estimate the uncertainty in the median magnitude difference to be 0.2 magnitudes. therefore, there is no significant difference between the hyperleda and sdss magnitudes. unfortunately, while fourteen galaxies have magnitudes measured in both systems, only two sdss selected galaxies have b@xmath13v and sbe entries in hyperleda ; for these two galaxies there are no significant differences between the hyperleda and sdss values. h 21 cm observations of 19 of the 20 nearby lcbgs in our sample were made with the 100 meter green bank telescope (gbt) at the national radio astronomy observatory in green bank, west virginia between 2002 november 30 and december 6. the h spectrum of sdssj0934@xmath240014 was acquired by @xcite during gbt commissioning on 2001 november 29. the main beam half power width is 9@xmath33.2 at 21 cm @xcite. both linear and circular polarizations were observed using the l - band receiver (1.15 @xmath13 1.73 ghz). position - switching was used with an offset of @xmath1318@xmath33 in right ascension. the spectral processor was used with a bandwidth of 20 mhz (@xmath14000 km s@xmath17) for most galaxies, although a few were observed with a bandwidth of 5 mhz (@xmath11000 km s@xmath17). the sample time was 60 s, with each galaxy observed for between 6 and 30 minutes, for a peak signal - to - noise of at least five. the only exceptions were sdssj0218@xmath130757 and sdssj0222@xmath130830, marginal detections, which were observed for 52 minutes each. the initial data calibration and reduction was performed using the aips++ single dish analysis environment `` dish. '' the position - switched data were calibrated in the standard way taking the difference of the on and off scans divided by the off scans. the dish package automatically calibrates the data into temperature units using tabulated values for the position of the telescope. the individual scans were then combined and a first order baseline was fit to the line - free regions and removed. our wide bandwidths ensured a sufficient line - free region. the individual polarizations were then combined, except when one polarization was significantly noisier than the other, or was affected by radio interference. finally, each spectrum was smoothed to @xmath112 km s@xmath17 channels using boxcar smoothing. the rest of the reduction and analysis was done using our own procedures written in interactive data language (idl). to convert our data from temperature to flux density units, we observed the radio galaxy 3c295. it makes an ideal calibration source as it has not varied by more than @xmath11@xmath35 since 1976 for 2.8 cm @xmath36 @xmath37 @xmath36 21 cm @xcite. comparing our observations of 3c295 with those of @xcite, we found the gain of the gbt to be 1.9 k jy@xmath17 and applied this calibration to our data. finally, obvious noise spikes were clipped out of the data. data of sdssj0934@xmath240014, the galaxy observed during telescope commissioning, were reduced and calibrated by @xcite. the spectrum was then smoothed to @xmath112 km s@xmath17 channels. the central 800 km s@xmath17 of the final h 21 cm spectra of the 20 local lcbgs are shown in figure 4. each galaxy s velocity calculated from optical redshifts is indicated with a triangle. dashed lines indicate the 20% crossings used to measure the line widths (@xmath153.1).
Results
all 20 galaxies were detected in the 21 cm line of h, although two, sdssj0218@xmath130757 and sdssj0222@xmath130830, were only detected at the 3 and 4 @xmath38 (respectively) level. h measurements and derived optical and h quantities are listed in tables 2, 3 and 4 and described in the following sections. since the gbt beam is 9@xmath33.2 at 21 cm, the emission from the target galaxies and any other galaxies within the beam and at similar velocities are blended into one spectrum. (see table 1 for those galaxies with other sources at similar velocities within the gbt beam.) we find fairly broad h profiles (126 km s@xmath17 @xmath18 w@xmath39 @xmath18 362 km s@xmath17) and a variety of profile shapes, including double - horned, flat - topped and gaussian. over a third of our profiles appear asymmetric and only half of these are from galaxies known to have other sources within the gbt beam. these asymmetries may indicate asymmetries in the gas density distribution and/or disk kinematics, or unidentified galaxies within the beam @xcite. estimates of the dynamical masses of our local sample of lcbgs constrain their evolutionary possibilities. we estimate the dynamical mass, m@xmath40, within a radius, r, as @xmath41 the constant c@xmath14 is a geometry - dependent factor which depends on the galactic light profile. for example, the king models of @xcite give 0.9 @xmath0 c@xmath14 @xmath0 1.5, depending on the ratio of tidal to core radius. as we do not have information on the light profiles of our galaxy sample, we simply choose c@xmath14 = 1. the line width at 20@xmath35, corrected for the effects of inclination, w@xmath42, is used as a measure of twice the maximum rotational velocity, v@xmath43 (e.g. *??? w@xmath39 was measured at the points equal to 20% of the peak flux. the first crossing at 20@xmath35 on each side of the emission line was used, once the spectrum became distinguishable from the noise. these crossings are indicated by dashed lines in figure 4. we find w@xmath39 ranging from 126 @xmath13 362 km s@xmath17. when @xcite studied lcbgs at intermediate redshifts, they found a similar range of line widths, using optical emission lines. the recessional velocities in the barycentric system, v@xmath44, were calculated as the midpoint between the 20@xmath35 crossings. the uncertainties listed in table 2 for w@xmath39, v@xmath44, and the galaxy distance (d) are from uncertainties in measuring the exact w@xmath39 crossings. two galaxies have multiple crossings at the 20% flux level, once the spectrum is distinguishable from the noise. one of these, sdssj0834@xmath240139, has a wing on the h spectrum which extends over 200 km s@xmath17. neither sdssj0834@xmath240139 nor its nearby companion have optical velocities coincident with the peak of the h spectrum ; instead, both have velocities in the wing of the spectrum (see figure 4). for this reason, we used w@xmath39 measured from the last crossing, which is at the edge of the wing. the other galaxy with multiple crossings at 20% is sdssj0904@xmath245136, which has a low level extension to one side of the spectrum. we used the first crossing at 20%. the uncertainties in these two w@xmath39 measurements are reflected in the large errors associated with w@xmath39 and derived quantities. we have initiated an observing program at the very large array (vla) to map these local lcbgs in h to disentangle target galaxy emission from any other galaxies in the field, improving the mass estimates. to correct the line width for the effects of inclination (_ i _), we divided w@xmath39 by @xmath45(_i _), to give w@xmath42. each sdss galaxy s inclination was approximated from the sdss data by : @xmath46 the sdss isophotal major and minor axes in the g - band (4770 ) were used, except for sdssj0943@xmath130215 where no g band data were available. in that case, r - band (6230 ) data were used. the sdss isophotal axes are derived from the ellipticity of the 25 magnitude arc sec@xmath19 isophote in each band. inclinations derived from sdss isophotal axes in r, i (7630 ), and z (9130 ) bands agree with inclinations derived from g - band data to within 8@xmath47. the inclinations from hyperleda were used for the markarian galaxies. thirteen of our sdss galaxies also have inclinations available in hyperleda, which are calculated from the apparent flattening and morphological type of the galaxies. we estimate the dispersion of the differences between sdss and hyperleda inclinations is 12@xmath47. a further correction can be made to w@xmath42 to account for random motions, giving w@xmath48. as outlined in @xcite, @xmath49 - 2w_t^2\exp-(w_{20}/w_c)^2\]] and @xmath50 w@xmath51 is the random motion component of the line width ; w@xmath51 = 38 km s@xmath17. w@xmath52 characterizes the transition region between linear and quadrature summation of rotational and dispersive terms ; w@xmath52 = 120 km s@xmath17. the formula degenerates to linear summation for giant galaxies, and quadrature summation for dwarf galaxies @xcite. this correction for random motions decreases the line width of the local lcbg sample by 26 to 38 km s@xmath17, depending on the rotational velocity. it is crucial, when comparing dynamical masses from different studies, that the rotational velocities be calculated in the same way. in general, we calculate the rotational velocities from w@xmath53. however, when we wish to compare our sample with others (e.g. *??? *) who have not applied this correction for random motions to their line widths, we also do not apply this correction. we indicate dynamical masses calculated without correcting w@xmath42 for random motions as m@xmath54. unless indicated as such, all dynamical masses are calculated using the line width corrected for random motions, w@xmath53. note that in all cases, the line widths have been corrected for inclination. it is also crucial, when comparing dynamical masses from different studies, that they be measured within the same radius. typically, the h radius (r@xmath3) is measured at 1 m@xmath7 pc@xmath19 and is used to estimate the total enclosed mass of a galaxy. however, it is only possible to measure r@xmath3 with interferometers in nearby galaxies. by practical necessity then, some other radius must be adopted for our sample and the galaxy s mass is assumed to be spherically distributed in that radius. two different radii are commonly used : r@xmath26, the effective or half - light radius, and r@xmath10, the isophotal radius at the limiting surface brightness of 25 b - magnitudes arc sec@xmath19. we have measurements of both r@xmath26 (in r and b band from sdss) and r@xmath10 (in b - band from hyperleda) for most galaxies. @xcite found that r@xmath10 = 2.5 r@xmath26 for nearby emission line galaxies measured in r - band. when we compare our values of r@xmath26 to r@xmath10, we find medians of @xmath55 and @xmath56 for the galaxies with no r@xmath10 (sdssj0218@xmath130757, sdssj0222@xmath130830 and sdssj1118@xmath246316) or r@xmath26 (non - sdss galaxies) measurements available, we estimate these quantities from the above relations. we find our local sample of lcbgs have dynamical masses, measured within r@xmath10, ranging from 3@xmath410@xmath6 to 1@xmath410@xmath11 m@xmath7, with a median of 3@xmath410@xmath57 m@xmath7. the dynamical masses measured within the effective radius (measured in both the r and b - band) range from 8@xmath410@xmath5 to 3@xmath410@xmath57 m@xmath7, with a median of 8@xmath410@xmath58 m@xmath7. figure 5 compares the dynamical masses measured for local lcbgs to nearby galaxies of hubble type s0a through i m, where `` m '' indicates magellanic, or low luminosity @xcite. note that @xcite did not correct line widths for random motions as we did. therefore, in figure 5 we compare m@xmath54(r @xmath59r@xmath10) which have not been corrected for random motions. these dynamical masses are higher than those we list in table 3, which have been corrected for random motions. as seen in figure 5, some local lcbgs have dynamical masses as large as l@xmath2 galaxies (@xmath110@xmath11 m@xmath7, roberts & haynes 1994). however, at least 75@xmath35 have dynamical masses approximately an order of magnitude smaller than typical local l@xmath2 galaxies, consistent with observations of lcbgs at higher redshifts (as discussed in @xmath15 1.1). however, our sample includes galaxies down to 0.25 l@xmath2. galaxies of this luminosity typically have dynamical masses (within r@xmath10) @xmath12@xmath410@xmath57 m@xmath7 @xcite. therefore, it is more appropriate to compare mass - to - light ratios ; see @xmath154.1. we estimate the random errors associated with the dynamical masses to be approximately 50%. this estimate is made from uncertainties in the inclination (@xmath34 12@xmath47, from the earlier comparison of hyperleda and sdss inclinations) ; uncertainties in radii measurements (as given by hyperleda and sdss) ; and uncertainties in measuring w@xmath39 (as discussed above). however, the random errors are overwhelmed by the systematic uncertainties when estimating dynamical masses. we have chosen a structural constant of c@xmath14 = 1, but this could be as high as 1.5, as discussed earlier. even more significant is that 45% of our galaxy sample have other galaxies at similar velocities in the beam of the gbt. as the h emission from all sources in the beam and within the bandwidth is blended into one spectrum, we may be overestimating the masses of these galaxies. finally, a fifth of our galaxies have low inclinations (@xmath18 40@xmath47) which lead to large corrections and therefore increasing uncertainties in the rotational velocities and overestimations of the dynamical masses. the dynamical masses we report may be viewed as upper limits we are certainly overestimating the masses in many cases, but not underestimating them. we are pursuing a follow - up program with both the arecibo radio telescope and the vla to address these issues. at 21 cm the arecibo radio telescope has a beam size of 3@xmath33.1 @xmath4 3@xmath33.5, a third the size of the gbt beam. this decreases the number of target galaxies observed with other galaxies in the beam, decreasing the number of galaxies with overestimates of dynamical masses. we have already observed over 40 lcbgs at arecibo, but results are not yet available. the vla in `` b '' configuration provides the ability to map our galaxies in 21 cm emission to a resolution of 5@xmath33@xmath33. we have begun a survey of those lcbgs with companions. vla maps allow us to disentangle the emission from each galaxy and more accurately estimate the dynamical mass of the target galaxy. along with dynamical mass, a knowledge of the amount of fuel, i.e. atomic and molecular gas, available for the starburst activity, is critical for narrowing down the evolutionary possibilities for lcbgs. when combined with star formation rates, this gives an estimate of the maximum length of the starburst at the current rate of star formation. we defer a full discussion of this until paper ii @xcite where we present our measurements of the co content of these galaxies, but present the h results here. the h masses are given by : @xmath60 @xcite, where the distance (measured from h ) is d = v@xmath44 h@xmath61. the total h flux, @xmath62, was calculated by numerically integrating under the spectrum where it is distinguishable from the noise. for the two marginal detections, sdssj0218@xmath130757 and sdssj0222@xmath130830, h masses were calculated using the optical (sdss) recessional velocities. we find local lcbgs have h masses ranging from 5@xmath410@xmath5 to 8@xmath410@xmath6 m@xmath7, with a median of 5@xmath410@xmath6 m@xmath7. these span the range of h masses in nearby galaxies across the hubble sequence ; the median is that of nearby late - type spiral galaxies @xcite. the uncertainties listed for m@xmath3 (table 3) include uncertainties in the distance and integrated flux density, as listed in table 2. however, as in the discussion of dynamical mass uncertainties in @xmath153.1, the h masses for nearly half of our galaxies are most likely overestimates due to the presence of other galaxies within the gbt beam.
Analysis
dynamical mass - to - light ratios indicate whether our local sample of lcbgs are under - massive for their luminosities. we calculated the total blue luminosities, l@xmath21, of our galaxies from m@xmath21, assuming m@xmath63 = 5.48 @xcite. we find the median mass - to - light ratio, m@xmath40(r @xmath18 r@xmath10) l@xmath8 = 3 m@xmath7 l@xmath64, with a minimum of 0.6 and a maximum of 9 m@xmath7 l@xmath64. @xcite find that m@xmath40(r @xmath18 r@xmath10) l@xmath8 ranges from 3.6 @xmath13 10 m@xmath7 l@xmath64 in local galaxies across the hubble sequence. the median m@xmath40(r @xmath18 r@xmath10) l@xmath8 is fairly constant ranging from 5 to 7 m@xmath7 l@xmath64. however, as discussed in @xmath153.1, to compare our mass - to - light ratios to those of local, normal hubble - types in @xcite, we must use dynamical masses which have not been corrected for random motions. figure 6 compares such mass - to - light ratios for our local sample of lcbgs and nearby hubble types. while some local lcbgs have mass - to - light ratios equal to or greater than the median for local hubble types, approximately half the local lcbgs have smaller mass - to - light ratios. many of the galaxies with mass - to - light ratios typical of local hubble types have other galaxies within the beam, possibly leading to an overestimation of the dynamical mass - to - light ratio of the target galaxy. therefore, in general, lcbgs tend to have mass - to - light ratios smaller than local normal galaxies. that is, most are small galaxies undergoing large amounts of star formation and not simply large galaxies with moderate amounts of star formation. we calculated the gas mass fraction, m@xmath3 m@xmath65 (r @xmath18 r@xmath3) for the local sample of lcbgs as a measure of their atomic gas richness. we do not have interferometric observations of these galaxies, so we estimated the hydrogen radius, r@xmath3, from optical radii. @xcite found for nearby spiral galaxies that r@xmath3 = 2 r@xmath10, where r@xmath3 is measured at the 1 m@xmath7 pc@xmath19 level. however, @xcite found r@xmath3 ranging from 3 @xmath13 5 r@xmath10 for a small sample of local h galaxies which are similar to lcbgs, but much less luminous (m@xmath21 @xmath66 @xmath1316). @xcite studied a large sample of nearby galaxies and found that galaxies with smaller optical radii have larger h extensions. @xcite have measured r@xmath3 in one of our lcbgs, mrk 314, using the vla. they find r@xmath3 = 4 r@xmath10. therefore, although we estimated r@xmath3 as 2 @xmath4 r@xmath10 following @xcite, we note it may be an underestimate, which would cause us to overestimate the gas mass fraction. we are undertaking interferometric observations of some of our local sample of lcbgs to directly measure r@xmath3. this will be addressed in a future paper. in local lcbgs, we find the gas mass fraction, m@xmath3 m@xmath65, ranges from 0.03 to 0.2, with a median of 0.08. that is, it ranges from normal hubble type spirals (@xmath18 0.1 @xcite) to very gas rich galaxies such as nearby h and irregular galaxies studied by @xcite with gas fractions ranging from 0.2 @xmath13 0.4. this is similar to the range of 0.01 @xmath13 0.5 found by @xcite for local blue compact galaxies, most of them less luminous than lcbgs. the fraction of hydrogen mass to blue luminosity provides a distance independent alternative measure of the gas richness of galaxies. we find our local lcbgs have m@xmath3 l@xmath8 ranging from 0.09 to 2 m@xmath7 l@xmath67, with a median of 0.4 m@xmath7 l@xmath67. this spans the range of nearby galaxies from s0a through i m, the median corresponding to late - type spirals @xcite. @xcite studied a range of blue compact galaxies, ranging from faint blue compact dwarfs to lcbgs, and found the same median value, 0.4 m@xmath7 l@xmath67. the tully - fisher relationship, a relation between line width and luminosity, is a fundamental scaling relation for non - interacting spiral galaxies. an absolute blue magnitude versus h line width version of the tully - fisher relationship is shown in figure 7. we have compared our local lcbgs with the tully - fisher relation found for @xmath14500 normal galaxies within @xmath140 mpc @xcite. as seen in figure 7, the location of local lcbgs is consistent with the @xcite relation, although with a large scatter. @xcite have a 1 @xmath38 scatter of 0.3 magnitudes in m@xmath21, while the local lcbgs have a 1 @xmath38 scatter of 0.9 magnitudes in m@xmath21. approximately half the galaxies (without other sources in the beam) lie to the left of the tully - fisher relationship, indicating they have lower masses than expected from their luminosities, although this result is not statistically significant. this is consistent with the distribution of lcbg and hubble type galaxy mass - to - light ratios in figure 6. seven of the galaxies lying to the right of the tully - fisher relationship have other galaxies within the gbt beam, possibly leading to overestimations of the line widths. note that in our comparison of magnitudes, masses, and mass - to - light ratios, as in all other comparisons in this paper, we have adjusted all values to a hubble constant of h@xmath16 = 70 km s@xmath17 mpc@xmath17
Discussion: comparison with local galaxy types
we have observed 20 local lcbgs chosen with the same selection criteria as those lcbgs common at higher redshifts. lcbgs can not remain lcbgs for a long period of time : the number density of lcbgs decreases by at least a factor of ten from z @xmath12 0.5 to today. that is, while lcbgs were common in the past, there are very few today. out of approximately a million nearby galaxies observed by the sdss, only about a hundred are lcbgs. (see castander et al. 2004 for a discussion of the local space density of lcbgs.) therefore, lcbgs must evolve into some other galaxy type. from studies of intermediate redshift lcbgs, @xcite and @xcite proposed that some may be progenitors of local dwarf elliptical galaxies. alternatively, @xcite and @xcite proposed that some may be disk galaxies in the process of forming a bulge to become present - day l@xmath2 spiral galaxies. two pieces of information are crucial to determining if these possibilities are likely. it is necessary to know the dynamical masses and the duration of the starbursts. we have measured the dynamical masses of our local sample of lcbgs using h . the length of the starburst will not be estimated until we discuss our molecular gas survey in paper ii @xcite. however, we can use our measurements of line width and radius to identify local galaxies comparable to lcbgs, independent of the evolutionary stage of their stellar populations. in figure 8 we plot the effective or half - light radius, r@xmath26, versus the velocity dispersion, @xmath38, for our sample of local lcbgs, intermediate redshift (0.4 @xmath0 z @xmath0 1) lcbgs @xcite, and local samples of elliptical, spiral, magellanic spiral, irregular, and dwarf elliptical galaxies (guzmn et al. 1996 ; hyperleda). we measured the @xmath38 of our sample of galaxies from the h spectra by measuring the moments of the spectra. the measurements were made in the same way as we described in @xmath153.2 for measuring @xmath62, the zeroth moment. we find similar results if we simply scale w@xmath39 to @xmath38 by assuming the spectra are gaussian, that is, dividing w@xmath39 by 3.6. the values of @xmath38 for the other galaxy samples are from optical line width measurements, except the measurements for magellanic spirals are from 21 cm h line widths. optical line widths may be smaller than h line widths. for example, @xcite studied a sample of nearby blue compact galaxies and found the ionized gas emission line widths to be systematically smaller than the neutral hydrogen emission line widths. on average, the ratio of w@xmath39 measured from h to w@xmath39 measured from h was 0.66 for their sample of 11 blue compact galaxies. however, the galaxies in @xcite s sample tend to be smaller both in effective radius and h line width than our sample of local lcbgs, suggesting that the difference between the optical and h line widths may be smaller as well @xcite. the r@xmath26 versus @xmath38 plot allows us to compare the dynamical mass properties of our sample of lcbgs with the other galaxy types. these properties are expected to remain constant despite luminosity evolution. we find that while some local lcbgs have @xmath38 consistent with local spiral galaxies, they tend to be too small in r@xmath26. however, local lcbgs are consistent with the smaller spiral galaxies, magellanic spirals. they are inconsistent with elliptical galaxies, but are consistent with the most massive dwarf elliptical and irregular galaxies. there is much confusion by what various authors mean when discussing dwarf elliptical galaxies (sometimes called spheroidal galaxies). we are referring to galaxies such as ngc 205 and _ not _ the less massive and less luminous galaxies like draco and carina. finally, when we compare our local sample of lcbgs with those lcbgs observed at intermediate redshifts, we find that they occupy the same region of r@xmath26 @xmath13 @xmath38 space, suggesting we are indeed examining a similar mass range in both samples of galaxies. we also compared local lcbgs to other galaxy types in w@xmath39 sin@xmath17(_i _) versus r@xmath10 space, where w@xmath39 was measured from h emission for all samples. this allowed us to investigate if the wavelength used to measure the line width, or correcting the line width for inclination, was having an effect on our interpretation of which galaxy types most resemble lcbgs. we could only compare local lcbgs, spirals, magellanic spirals, and irregular galaxies, as h observations of intermediate redshift lcbgs and dwarf ellipticals are rare or non - existent. our findings were entirely consistent with the interpretation from figure 8. local lcbgs are therefore consistent with the dynamical mass properties of the most massive dwarf ellipticals and irregulars, and lower mass or magellanic spirals. these classes vary in color and magnitude. knowledge of the amount of molecular gas, time scale of starburst and fading, along with the ability of the galaxy to retain its interstellar medium, will allow us to discriminate between these remaining possibilities. we will begin this work in paper ii @xcite. given the diverse nature of lcbgs, it is likely that multiple scenarios apply, each to a different subset of lcbgs.
Conclusions
we have performed a single dish 21 cm h survey of 20 local lcbgs chosen to be local analogs to the numerous lcbgs studied at intermediate redshifts (0.4 @xmath0 z @xmath0 0.7). our findings have verified results from intermediate redshift lcbg studies. we have found that local lcbgs are a morphologically heterogeneous mixture of galaxies. they are typically gas - rich, with median values of m@xmath3 = 5@xmath410@xmath6 m@xmath7 and m@xmath3 l@xmath8 = 0.4 m@xmath7 l@xmath9. approximately half have mass - to - light ratios approximately ten times smaller than local galaxies of all hubble types at similar luminosities, confirming that these are indeed small galaxies undergoing vigorous bursts of star formation. this proportion is likely an underestimate, as nearly half our sample of galaxies may have dynamical mass overestimates. by comparing line widths and radii with local galaxy populations, we find that local lcbgs are consistent with magellanic spirals, and the more massive irregulars and dwarf ellipticals. measurements of the length of starburst, amount of fading, and ability of these galaxies to retain their interstellar media will help to constrain the evolutionary possibilities of this galaxy class. we begin to address these issues in paper ii @xcite, where we present the results of a molecular gas survey of these same local lcbgs. _ acknowledgments _ we thank the referee for helpful comments which improved the quality of this paper. we thank rick fisher for providing the h spectrum of sdssj0934 + 0014. we also thank the operators and staff at the gbt for their help with the observing and reduction, and their hospitality. support for this work was provided by the nsf through award gssp02 - 0001 from the nrao. d. j. p. acknowledges generous support from an nsf mps distinguished international postdoctoral research fellowship, nsf grant ast0104439. r. g. acknowledges funding from nasa grant ltsa nag5 - 11635. funding for the creation and distribution of the sdss archive has been provided by the alfred p. sloan foundation, the participating institutions, the national aeronautics and space administration, the national science foundation, the u.s. department of energy, the japanese monbukagakusho, and the max planck society. the sdss web site is http://www.sdss.org/. we have made extensive use of hyperleda (http://www-obs.univ-lyon1.fr/hypercat/) and the nasa / ipac extragalactic database (ned) which is operated by the jet propulsion laboratory, california institute of technology, under contract with the national aeronautics and space administration (http://nedwww.ipac.caltech.edu/). the digitized sky surveys were produced at the space telescope science institute under u.s. government grant nag w-2166. the images of these surveys are based on photographic data obtained using the oschin schmidt telescope on palomar mountain and the uk schmidt telescope. llcccclcc mrk 297 & ngc 6052, ngc 6064 & 67 & @xmath1321.0 & 0.4 & 20.6 & & sc & y + mrk 314 & ngc 7468 & 30 & @xmath1318.5 & 0.4 & 20.2 & e3, pec (polar ring?) & e & n + mrk 325 & ngc 7673, mrk 325 & 49 & @xmath1320.0 & 0.4 & 20.0 & sac?, pec, h ii starburst & sc & y + mrk 538 & ngc 7714 & 40 & @xmath1320.1 & 0.4 & 20.2 & sb(s)b, pec, h ii liner & sbb & y + sdss j011932.95 + 145219.0 & ngc 469 & 59 & @xmath1318.9 & 0.4 & 20.3 & & & y + sdss j021808.75@xmath13075718.0 & & 69 & @xmath1318.8 & 0.5 & 20.2 & & & n + sdss j022211.96@xmath13083036.2 & & 67 & @xmath1318.6 & 0.4 & 20.1 & & & n + sdss j072849.75 + 353255.2 & & 56 & @xmath1318.9 & 0.4 & 20.3 & s? & sbc & n + sdss j083431.70 + 013957.9 & & 59 & @xmath1319.1 & 0.6 & 20.6 & sb(s)b & sbb & y + sdss j090433.53 + 513651.1 & mrk 101 & 68 & @xmath1319.7 & 0.6 & 20.3 & s & sc & n + sdss j091139.74 + 463823.0 & mrk 102 & 61 & @xmath1319.3 & 0.5 & 19.1 & s? & & n + sdss j093410.52 + 001430.2 & mrk 1233 & 70 & @xmath1319.8 & 0.3 & 19.9 & sb & sbc & y + sdss j093635.36 + 010659.8 & & 71 & @xmath1319.1 & 0.6 & 21.0 & & & y + sdss j094302.60@xmath13021508.9 & & 68 & @xmath1319.3 & 0.5 & 20.4 & s0 & s0 & n + sdss j111836.35 + 631650.4 & mrk 165 & 46 & @xmath1318.6 & 0.4 & 19.4 & compact starburst & & n + sdss j123440.89 + 031925.1 & ngc 4538 & 67 & @xmath1319.2 & 0.6 & 21.0 & s pec & sbc & n + sdss j131949.93 + 520341.1 & & 67 & @xmath1318.7 & 0.2 & 20.1 & & & y + sdss j140203.52 + 095545.6 & ngc 5414, mrk 800 & 61 & @xmath1319.7 & 0.5 & 19.7 & pec & & y + sdss j150748.33 + 551108.6 & & 48 & @xmath1318.9 & 0.4 & 20.8 & s & sbc & n + sdss j231736.39 + 140004.3 & ngc 7580, mrk 318 & 63 & @xmath1319.3 & 0.6 & 20.4 & s? & sbc & n + lllll mrk 297 & 4739 @xmath34 17 & 68 @xmath34 0.2 & 362 @xmath34 17 & 6.5 @xmath34 0.08 + mrk 314 & 2081 @xmath34 17 & 30 @xmath34 0.2 & 188 @xmath34 17 & 12 @xmath34 0.1 + mrk 325 & 3427 @xmath34 17 & 49 @xmath34 0.2 & 202 @xmath34 17 & 11 @xmath34 0.2 + mrk 538 & 2798 @xmath34 17 & 40 @xmath34 0.2 & 240 @xmath34 17 & 20 @xmath34 0.2 + sdssj0119 + 1452 & 4098 @xmath34 17 & 59 @xmath34 0.2 & 266 @xmath34 17 & 2.4 @xmath34 0.1 + sdssj0218@xmath130757 & & & & 0.41 @xmath34 0.07 + sdssj0222@xmath130830 & & & & 0.61 @xmath34 0.09 + sdssj0728 + 3532 & 3953 @xmath34 17 & 56 @xmath34 0.2 & 216 @xmath34 17 & 7.9 @xmath34 0.1 + sdssj0834 + 0139 & 4215 @xmath34 160 & 60 @xmath34 2 & 329 @xmath34 160 & 6.9 @xmath34 0.2 + sdssj0904 + 5136 & 4782 @xmath34 103 & 68 @xmath34 2 & 204 @xmath34 103 & 4.5 @xmath34 0.2 + sdssj0911 + 4636 & 4281 @xmath34 17 & 61 @xmath34 0.2 & 151 @xmath34 17 & 2.0 @xmath34 0.1 + sdssj0934 + 0014 & 4860 @xmath34 17 & 69 @xmath34 0.2 & 332 @xmath34 17 & 4.8 @xmath34 0.2 + sdssj0936 + 0106 & 4920 @xmath34 17 & 70 @xmath34 0.2 & 255 @xmath34 17 & 3.3 @xmath34 0.09 + sdssj0943@xmath130215 & 4823 @xmath34 17 & 69 @xmath34 0.2 & 230 @xmath34 17 & 2.6 @xmath34 0.06 + sdssj1118 + 6316 & 3218 @xmath34 17 & 46 @xmath34 0.2 & 152 @xmath34 17 & 2.2 @xmath34 0.07 + sdssj1234 + 0319 & 4685 @xmath34 17 & 67 @xmath34 0.2 & 269 @xmath34 17 & 4.4 @xmath34 0.1 + sdssj1319 + 5203 & 4619 @xmath34 17 & 66 @xmath34 0.2 & 202 @xmath34 17 & 7.7 @xmath34 0.09 + sdssj1402 + 0955 & 4267 @xmath34 17 & 61 @xmath34 0.2 & 343 @xmath34 17 & 6.8 @xmath34 0.2 + sdssj1507 + 5511 & 3373 @xmath34 17 & 48 @xmath34 0.2 & 126 @xmath34 17 & 3.7 @xmath34 0.1 + sdssj2317 + 1400 & 4413 @xmath34 17 & 63 @xmath34 0.2 & 254 @xmath34 17 & 6.2 @xmath34 0.1 + llccccccc mrk 297 & 7.0 @xmath34 0.1 & 42 & 486 & 243 & 8.0 & 11 & 2.0 & 2.7 + mrk 314 & 2.5 @xmath34 0.05 & 65 & 170 & 85 & 3.8 & 0.63 & 0.94 & 0.16 + mrk 325 & 6.3 @xmath34 0.1 & 43 & 242 & 121 & 9.6 & 3.3 & 2.4 & 0.83 + mrk 538 & 7.6 @xmath34 0.1 & 50 & 264 & 132 & 11 & 4.5 & 2.8 & 1.1 + sdssj0119 + 1452 & 2.0 @xmath34 0.1 & 84 & 230 & 115 & 5.6 & 1.7 & 1.6 & 0.48 + sdssj0218@xmath130757 & 0.47 @xmath34 0.08 & 52 & & & 4.9 & & 1.3 & + sdssj0222@xmath130830 & 0.65 @xmath34 0.1 & 48 & & & 4.2 & & 1.2 & + sdssj0728 + 3532 & 6.0 @xmath34 0.1 & 34 & 321 & 160 & 5.4 & 3.3 & 1.2 & 0.74 + sdssj0834 + 0139 & 5.9 @xmath34 0.5 & 83 & 293 & 147 & 6.5 & 3.3 & 1.6 & 0.82 + sdssj0904 + 5136 & 4.9 @xmath34 0.3 & 34 & 301 & 150 & 7.4 & 3.9 & 2.0 & 1.0 + sdssj0911 + 4636 & 1.7 @xmath34 0.1 & 33 & 221 & 111 & 5.6 & 1.6 & 1.2 & 0.35 + sdssj0934 + 0014 & 5.4 @xmath34 0.2 & 51 & 378 & 189 & 6.8 & 5.7 & 1.9 & 1.5 + sdssj0936 + 0106 & 3.8 @xmath34 0.1 & 48 & 293 & 146 & 7.1 & 3.5 & 2.0 & 1.0 + sdssj0943@xmath130215 & 2.9 @xmath34 0.07 & 84 & 194 & 97 & 4.9 & 1.1 & 1.8 & 0.39 + sdssj1118 + 6316 & 1.1 @xmath34 0.04 & 82 & 123 & 61 & 3.1 & 0.27 & 0.91 & 0.08 + sdssj1234 + 0319 & 4.7 @xmath34 0.1 & 52 & 293 & 147 & 6.7 & 3.4 & 2.1 & 1.1 + sdssj1319 + 5203 & 7.9 @xmath34 0.1 & 44 & 239 & 120 & 5.3 & 1.8 & 1.2 & 0.41 + sdssj1402 + 0955 & 6.0 @xmath34 0.2 & 55 & 372 & 186 & 8.7 & 7.0 & 1.6 & 1.3 + sdssj1507 + 5511 & 2.0 @xmath34 0.07 & 44 & 144 & 72 & 7.9 & 0.95 & 1.8 & 0.21 + sdssj2317 + 1400 & 5.8 @xmath34 0.1 & 31 & 420 & 210 & 7.3 & 7.5 & 1.7 & 1.8 + lccccc mrk 297 & 0.03 & 39 & 0.2 & 3 & 0.7 + mrk 314 & 0.2 & 3.9 & 0.7 & 2 & 0.4 + mrk 325 & 0.1 & 16 & 0.4 & 2 & 0.5 + mrk 538 & 0.08 & 17 & 0.5 & 3 & 0.7 + sdssj0119 + 1452 & 0.06 & 5.6 & 0.4 & 3 & 0.9 + sdssj0218 - 0757 & & 5.2 & 0.09 & & + sdssj0222 - 0830 & & 4.3 & 0.2 & & + sdssj0728 + 3532 & 0.09 & 5.6 & 1 & 6 & 1 + sdssj0834 + 0139 & 0.1 & 6.8 & 0.9 & 5 & 1 + sdssj0904 + 5136 & 0.07 & 12 & 0.4 & 3 & 0.9 + sdssj0911 + 4636 & 0.05 & 8.2 & 0.2 & 2 & 0.4 + sdssj0934 + 0014 & 0.05 & 13 & 0.4 & 4 & 1 + sdssj0936 + 0106 & 0.06 & 6.8 & 0.6 & 5 & 2 + sdssj0943 - 0215 & 0.1 & 8.2 & 0.4 & 1 & 0.5 + sdssj1118 + 6316 & 0.2 & 4.3 & 0.3 & 0.6 & 0.2 + sdssj1234 + 0319 & 0.07 & 7.4 & 0.6 & 5 & 1 + sdssj1319 + 5203 & 0.2 & 4.7 & 2 & 4 & 0.9 + sdssj1402 + 0955 & 0.04 & 12 & 0.5 & 6 & 1 + sdssj1507 + 5511 & 0.1 & 5.6 & 0.4 & 2 & 0.4 + sdssj2317 + 1400 & 0.04 & 8.2 & 0.7 & 9 & 2 + . the spectra have been smoothed to a resolution of @xmath112 km s@xmath17 and only the central 800 km s@xmath17 are shown. the dashed lines indicate the 20% crossings used to measure the line width at 20%. the triangles indicate the recessional velocities calculated from sdss redshifts for the sdss galaxies ; velocities from ned redshifts are shown for the non - sdss galaxies. those galaxies with other sources at similar velocities within the gbt beam are indicated by a star in the upper right corner. [fig4]] (r @xmath59r@xmath10), for local lcbgs, as measured from h observations, is shown as a gray histogram. for comparison, the range of dynamical masses for local hubble type galaxies are indicated with black boxes. the `` m '' in `` sm '' and `` i m '' indicates magellanic or low - luminosity spirals and irregulars. note that in order to compare lcbg dynamical masses with @xcite s results for local hubble types, the dynamical masses plotted here do not include a line width correction for random motions. [fig5]] ) to l@xmath21 ratios for local lcbgs is shown as a gray histogram. for comparison, the range of mass - to - light ratios for local hubble type galaxies (s0a to i m) @xcite are shown. as in figure 5, for accurate comparisons, these mass - to - light ratios do not include a line width correction for random motions. many lcbgs tend to have mass - to - light ratios smaller than local normal galaxies, consistent with findings at intermediate redshifts. note that we may have overestimated the dynamical masses for most of the lcbgs with higher mass - to - light ratios. [fig6]] ) versus absolute blue magnitude (m@xmath21). local lcbgs are indicated by filled circles. those galaxies with other sources at similar velocities within the gbt beam are circled ; their line widths may be overestimated. the solid line indicates the tully - fisher relationship from @xcite ; the dotted lines indicate their 1 @xmath38 scatter of 0.3 m@xmath21. the location of our local lcbgs is consistent with the tully - fisher relationship, but with a higher 1 @xmath38 scatter of 0.9 m@xmath21. [fig7]] ) versus the the velocity dispersion (@xmath38) is plotted for our local lcbgs (filled circles) and intermediate redshift (0.4 @xmath0 z @xmath0 1) lcbgs (open circles) @xcite. the regions occupied by the bulk of other local galaxy types@xmath13ellipticals (e), spirals (s), magellanic spirals (sm), dwarf ellipticals (de), and irregulars (irr)@xmath13are indicated (guzmn et al. 1996, hyperleda). we have also indicated the approximate locations for some representative galaxies : draco, ngc 205 and m 31 (hyperleda). local lcbgs are consistent with higher mass irregulars and dwarf ellipticals, and lower mass or magellanic spirals, as well as intermediate redshift lcbgs. | we present single - dish h spectra obtained with the green bank telescope, along with optical photometric properties from the sloan digital sky survey, of 20 nearby (d @xmath0 70 mpc) luminous compact blue galaxies (lcbgs). these @xmath1l@xmath2, blue, high surface brightness, starbursting galaxies were selected with the same criteria used to define lcbgs at higher redshifts.
we find these galaxies are gas - rich, with m@xmath3 ranging from 5@xmath410@xmath5 to 8@xmath410@xmath6 m@xmath7, and m@xmath3 l@xmath8 ranging from 0.2 to 2 m@xmath7 l@xmath9, consistent with a variety of morphological types of galaxies.
we find the dynamical masses (measured within r@xmath10) span a wide range, from 3@xmath410@xmath6 to 1@xmath410@xmath11 m@xmath7.
however, at least half have dynamical mass - to - light ratios smaller than nearby galaxies of all hubble types, as found for lcbgs at intermediate redshifts. by comparing line widths and
effective radii with local galaxy populations, we find that lcbgs are consistent with the dynamical mass properties of magellanic (low luminosity) spirals, and the more massive irregulars and dwarf ellipticals, such as ngc 205. | astro-ph0407438 |
Introduction
it has become clear in the past several years that copper oxide materials are among the most complex systems studied in condensed matter physics, and show many unusual normal - state properties. the complications arise mainly from (1) strong anisotropy in the properties parallel and perpendicular to the cuo@xmath0 planes which are the key structural element in the whole copper oxide superconducting materials, and (2) extreme sensitivity of the properties to the compositions (stoichiometry) which control the carrier density in the cuo@xmath0 plane @xcite, while the unusual normal - state feature is then closely related to the fact that these copper oxide materials are doped mott insulators, obtained by chemically adding charge carriers to a strongly correlated antiferromagnetic (af) insulating state, therefore the physical properties of these systems mainly depend on the extent of dopings, and the regimes have been classified into the underdoped, optimally doped, and overdoped, respectively @xcite. the normal - state properties of copper oxide materials in the underdoped and optimally doped regimes exhibit a number of anomalous properties in the sense that they do not fit in the conventional fermi - liquid theory @xcite, and the mechanism for the superconductivity in copper oxide materials has been widely recognized to be closely associated with the anisotropic normal - state properties @xcite. among the striking features of the normal - state properties in the underdoped and optimally doped regimes, the physical quantity which most evidently displays the anisotropic property in copper oxide materials is the charge dynamics @xcite, which is manifested by the optical conductivity and resistivity. it has been show from the experiments that the in - plane charge dynamics is rather universal within the whole copper oxide materials @xcite. the in - plane optical conductivity for the same doping is nearly materials independent both in the magnitude and energy dependence, and shows the non - drude behavior at low energies and anomalous midinfrared band in the charge - transfer gap, while the in - plane resistivity @xmath1 exhibits a linear behavior in the temperature in the optimally doped regime and a nearly temperature linear dependence with deviations at low temperatures in the underdoped regime @xcite. by contrast, the magnitude of the c - axis charge dynamics in the underdoped and optimally doped regimes is strongly materials dependent, _ i.e. _, it is dependent on the species of the building blocks in between the cuo@xmath0 planes @xcite. in the underdoped and optimally doped regimes, the experimental results @xcite show that the ratio @xmath2 ranges from @xmath3 to @xmath4, this large magnitude of the resistivity anisotropy reflects that the c - axis mean free path is shorter than the interlayer distance, and the carriers are tightly confined to the cuo@xmath0 planes, and also is the evidence of the incoherent charge dynamics in the c - axis direction. for the copper oxide materials without the cu - o chains in between the cuo@xmath0 planes @xcite, such as la@xmath5sr@xmath6cuo@xmath7 systems, the transferred weight in the c - axis conductivity forms a band peaked at high energy @xmath8, and the low - energy spectral weight is quite small and spread over a wide energy range instead of forming a peak at low energies, in this case the behavior of the c - axis temperature dependent resistivity @xmath9 is characterized by a crossover from the high temperature metallic - like to the low temperature semiconducting - like @xcite. however, for these copper oxide materials with the cu - o chains in between the cuo@xmath0 planes @xcite, such as yba@xmath0cu@xmath10o@xmath11 systems, the c - axis conductivity exhibits the non - drude behavior at low energies and weak midinfrared band, moreover, this weak midinfrared band rapidly decrease with reducing dopings or increasing temperatures, while the c - axis resistivity @xmath9 is linear in temperatures in the optimally doped regime, and shows a crossover from the high temperature metallic - like behavior to the low temperature semiconducting - like behavior in the underdoped regime @xcite. therefore there are some subtle differences between the chain and no - chain copper oxide materials. the c - axis charge dynamics of copper oxide materials has been addressed from several theoretical viewpoints @xcite. based on the concept of dynamical dephasing, leggett @xcite thus proposed that the c - axis conduction has to do with scatterings from in - plane thermal fluctuations, and depends on the ratio of the interlayer hopping rate of cuo@xmath0 sheets to the thermal energy. while the theory of tunneling c - axis conductivity in the incoherent regime has been given by many researchers @xcite. based on a highly anisotropic fermi - liquid, some effect from the interlayer static disorder or dynamical one has been discussed @xcite. the similar incoherent conductivity in the coupled fermion chains has been in more detail studied by many authors within the framework of the non - fermi - liquid theory @xcite. moreover, the most reliable result for the c - axis charge dynamics from the model relevant to copper oxide materials has been obtained by the numerical simulation @xcite. it has been argued that the in - plane resistivity deviates from the temperature linear behavior and temperature coefficient of the c - axis resistivity change sign, showing semiconducting - like behavior at low temperatures are associated with the effect of the pseudogap @xcite. to shed light on this issue, we, in this paper, apply the fermion - spin approach @xcite to study the c - axis charge dynamics by considering the interlayer coupling. the paper is organized as follows. the theoretical framework is presented in sec. ii. in the case of the incoherent interlayer hopping, the c - axis current - current correlation function (then the c - axis optical conductivity) is calculated in terms of the in - plane single - particle spectral function by using standard formalisms for the tunneling in metal - insulator - metal junctions @xcite. within this theoretical framework, we discuss the c - axis charge dynamics of the chain copper oxide materials in sec. it is shown that the c - axis charge dynamics of the chain copper oxide materials is mainly governed by the scattering from in - plane charged holons due to spinon fluctuations, and the behavior of the c - axis resistivity is the metallic - like in the optimally doped regime and the semiconducting - like in the underdoped regime at low temperatures. in sec. iv, the c - axis charge dynamics of the no - chain copper oxide materials is discussed. our result shows that the scattering from the in - plane fluctuation incorporating with the interlayer disorder dominates the c - axis charge dynamics for the no - chain copper oxide materials. in this case, the c - axis resistivity exhibits the semiconducting - like behavior in the underdoped and optimally doped regimes at low temperatures. v is devoted to a summary and discussions. our results also show that the crossover to the semiconducting - like range in @xmath9 is obviously linked with the crossover from the temperature linear to the nonlinear range in @xmath1, and the common origin for these crossovers is due to the existence of the holon pseudogap at low temperatures and lower doping levels.
Theoretical framework
among the microscopic models the most simplest for the discussion of doped mott insulators is the @xmath12-@xmath13 model @xcite, which is originally introduced as an effective hamiltonian of the hubbard model in the strong coupling regime, where the electron become strongly correlated to avoid the double occupancy. the interest in the @xmath12-@xmath13 model is stimulated by many researchers suggestions that it may contain the essential physics of copper oxide materials @xcite. on the other hand, there is a lot of evidence from the experiments and numerical simulations in favour of the @xmath12-@xmath13 model as the basic underlying microscopic model @xcite. within each cuo@xmath0 plane, the physics property may be described by the two - dimensional (2d) @xmath12-@xmath13 model, @xmath14 supplemented by the on - site local constraint @xmath15 to avoid the double occupancy, where @xmath16, @xmath17 is the lattice constant of the square planar lattice, which is set as the unit hereafter, @xmath18 refers to planar sites within the l - th cuo@xmath0 plane, @xmath19 (@xmath20) are the electron creation (annihilation) operators, @xmath21 are the spin operators with @xmath22 as the pauli matrices, and @xmath23 is the chemical potential. then the hopping between cuo@xmath0 planes is considered as @xcite @xmath24 where @xmath25, @xmath26 is the interlayer distance, and has been determined from the experiments @xcite as @xmath27. as mentioned above, the experimental results show that the c - axis charge dynamics in the underdoped and optimally doped regimes is incoherent, therefore the c - axis momentum can not be defined @xcite. moreover, the absence of the coherent c - axis charge dynamics is a consequence of the weak interlayer hopping matrix element @xmath28, but also of a strong intralayer scattering, _ i.e. _, @xmath29, and therefore the common cuo@xmath0 planes in copper oxide materials clearly dominate the most normal - state properties. in this case, the most relevant for the study of the c - axis charge dynamics is the results on the in - plane conductivity @xmath30 and related single - particle spectral function @xmath31. since the strong electron correlation in the @xmath12-@xmath13 model manifests itself by the electron single occupancy on - site local constraint, then the crucial requirement is to impose this electron on - site local constraint for a proper understanding of the physics of copper oxide materials. to incorporate this local constraint, the fermion - spin theory based on the charge - spin separation has been proposed @xcite. according to the fermion - spin theory, the constrained electron operators in the @xmath12-@xmath13 model is decomposed as @xcite, @xmath32 with the spinless fermion operator @xmath33 keeps track of the charge (holon), while the pseudospin operator @xmath34 keeps track of the spin (spinon). the main advantage of this approach is that the electron on - site local constraint can be treated exactly in analytical calculations. in this case, the low - energy behavior of the @xmath12-@xmath13 model (2) in the fermion - spin representation can be written as @xcite, @xmath35 where @xmath36 $], the in - plane holon particle - hole parameter @xmath37, and @xmath38 and @xmath39 are the pseudospin raising and lowering operators, respectively. as a consequence, the kinetic part in the @xmath12-@xmath13 model has been expressed as the holon - spinon interaction in the fermion - spin representation, which dominates the charge and spin dynamics in copper oxide materials in the underdoped and optimally doped regimes. the spinon and holon may be separated at the mean - field level, but they are strongly coupled beyond mean - field approximation (mfa) due to fluctuations. the mean - field theory within the fermion - spin formalism in the underdoped and optimally doped regimes without af long - range - order (aflro) has been developed @xcite, and the in - plane mean - field spinon and holon green s functions @xmath40 and @xmath41 have been evaluated @xcite as, @xmath42 @xmath43 respectively, where @xmath44 $], @xmath45, @xmath46, @xmath47, @xmath48 is the number of the nearest neighbor sites at the plane, while the in - plane mean - field spinon spectrum @xmath49-[\alpha c_{z}+{1\over 4z}(1-\alpha)]\right) (\epsilon\gamma_{k}-1) \nonumber \\ + \lambda^{2}\left (\alpha\epsilon[{1\over 2}\chi\gamma_{k}+ { 1\over z}\chi_{z}]-{1\over 2}\epsilon[\alpha c+{1\over 2z } (1-\alpha)]\right) (\gamma_{k}-\epsilon),\end{aligned}\]] and the in - plane mean - field holon spectrum @xmath50, with the in - plane spinon correlation functions @xmath51, @xmath52, @xmath53, and @xmath54. in order not to violate the sum rule of the correlation function @xmath55 in the case without aflro, the important decoupling parameter @xmath56 has been introduced in the mean - field calculation @xcite, which can be regarded as the vertex correction. the mean - field order parameter @xmath57, @xmath58, @xmath59, @xmath60, @xmath61 and chemical potential @xmath23 have been determined @xcite by the self - consistent equations. within the 2d @xmath12-@xmath13 model, the in - plane charge dynamics in copper oxide materials has been discussed @xcite by considering fluctuations around this mean - field solution, and the result exhibits a behavior similar to that seen in the experiments @xcite and numerical simulations @xcite. in the framework of the charge - spin separation, an electron is represented by the product of a holon and a spinon, then the external field can only be coupled to one of them. according to the ioffe - larkin combination rule @xcite, the physical c - axis conductivity @xmath62 is given by, @xmath63 where @xmath64 and @xmath65 are the contributions to the c - axis conductivity from holons and spinons, respectively, and can be expressed @xcite as, @xmath66 with @xmath67 and @xmath68 are the holon and spinon c - axis current - current correlation function, respectively, which are defined as, @xmath69 within the hamiltonian (4), the c - axis current densities of spinons and holons are obtained by the time derivation of the polarization operator using heisenberg s equation of motion as, @xmath70 and @xmath71, respectively, with @xmath72 is the effective interlayer holon hopping matrix element, and the order parameters @xmath73 and @xmath74 are defined @xcite as @xmath75, and @xmath76, respectively. as in the previous discussions @xcite, a formal calculation for the spinon part shows that there is no the direct contribution to the current - current correlation from spinons, but the strongly correlation between holons and spinons is considered through the spinon s order parameters entering in the holon part of the contribution to the current - current correlation, therefore the charge dynamics in copper oxide materials is mainly caused by charged holons within the cuo@xmath0 planes, which are strongly renormalized because of the strong interaction with fluctuations of surrounding spinon excitations. in the case of the incoherent charge dynamics in the c - axis direction, _ i.e. _, the independent electron propagation in each layer, the c - axis holon current - current correlation function is then proportional to the tunneling rate between just two adjacent planes, and can be calculated in terms of the in - plane holon green s function @xmath77 by using standard formalisms for the tunneling in metal - insulator - metal junctions @xcite as, @xmath78 where @xmath79 is the matsubara frequency. therefore the c - axis current - current correlation function is essentially determined by the property of the in - plane full holon propagator @xmath80, which can be expressed as, @xmath81 where the holon self - energy has been obtained by considering the second - order correction due to the spinon pair bubble as @xcite, @xmath82-n_{b}(\omega_{p+p'}) n_{b}(-\omega_{p'})\over i\omega_{n}+\omega_{p+p'}-\omega_{p'}- \xi_{p+k } } \right. \nonumber \\ + { n_{f}(\xi_{p+k})[n_{b}(\omega_{p+p'})-n_{b}(-\omega_{p'})]+ n_{b}(\omega_{p'})n_{b}(\omega_{p+p'})\over i\omega_{n}+\omega_{p'} + \omega_{p+p'}-\xi_{p+k } } \nonumber \\ \left. -{n_{f}(\xi_{p+k)}[n_{b}(\omega_{p+p'}) -n_{b}(-\omega_{p'})]-n_{b}(-\omega_{p'})n_{b}(-\omega_{p+p'}) \over i\omega_{n}-\omega_{p+p'}-\omega_{p'}-\xi_{p+k}}\right), \end{aligned}\]] with @xmath83 and @xmath84 are the fermi and bose distribution functions, respectively. for the convenience in the following discussions, the above full holon in - plane green s function @xmath80 also can be expressed as frequency integrals in terms of the spectral representation as, @xmath85 with the in - plane holon spectral function @xmath86. then the c - axis optical conductivity in the present theoretical framework is expressed @xcite as @xmath87.
Charge dynamics for the chain copper oxide materials
we firstly consider the chain copper oxide materials. from the experiments testing the c - axis charge dynamics @xcite, it has been shown that the presence of the rather conductive cu - o chains in the underdoped and optimally doped regimes can reduce the blocking effect, and therefore the c - axis charge dynamics in this system is effected by the same electron interaction as that in the in - plane. in this case, we substitute eq. (14) into eq. (11), and evaluate the frequency summations, then the c - axis optical conductivity for the chain copper oxide materials can be obtained as @xcite, @xmath88 where the in - plane momentum is conserved. this c - axis conductivity @xmath89 has been calculated numerically, and the result at the doping @xmath90 (solid line), @xmath91 (dashed line), and @xmath92 (dot - dashed line) for the parameters @xmath93, @xmath94, and @xmath95 at the temperature @xmath96=0 is shown in fig. 1, where the charge @xmath97 has been set as the unit. from fig. 1, we find that there is two bands in @xmath62 separated at @xmath98, the higher - energy band, corresponding to the `` midinfrared band '' in the in - plane optical conductivity @xmath30 @xcite, shows a broad peak at @xmath99, in particular, the weight of this band is strongly doping dependent, and decreasing rapidly with dopings, but the peak position does not appreciably shift to higher energies. on the other hand, the transferred weight of the lower - energy band forms a sharp peak at @xmath100, which can be described formally by the non - drude formula, and our analysis indicates that this peak decay is @xmath101 at low energies as in the case of @xmath30 @xcite. in comparison with @xmath30 @xcite, the present result also shows that the values of @xmath89 are by @xmath102 orders of magnitude smaller than those of @xmath30 in the corresponding energy range. the finite temperature behavior of @xmath89 also has been discussed, and the result shows that @xmath89 is temperature dependent, the higher - energy band is severely suppressed with increasing temperatures, and vanishes at higher temperatures. these results are qualitatively consistent with the experimental results @xcite of the chain copper oxide materials and numerical simulations @xcite. with the help of the c - axis conductivity, the c - axis resistivity can be obtained as @xmath103, and the numerical result at the doping @xmath90 and @xmath92 for the parameters @xmath93, @xmath104, and @xmath95 is shown in fig. 2(a) and fig. 2(b), respectively. in the underdoped regime, the behavior of the temperature dependence of @xmath9 shows a crossover from the high temperature metallic - like (@xmath105) to the low temperature semiconducting - like (@xmath106), but the metallic - like temperature dependence dominates over a wide temperature range. therefore in this case, there is a general trend that the chain copper oxide materials show nonmetallic @xmath9 in the underdoped regime at low temperatures. while in the optimally doped regime, @xmath9 is linear in temperatures, and shows the metallic - like behavior for all temperatures. these results are also qualitatively consistent with the experimental results of the chain copper oxide materials @xcite and numerical simulation @xcite.
Charge dynamics for the no-chain copper oxide materials
now we turn to discuss the c - axis charge dynamics of the no - chain copper oxide materials. it has been indicated from the experiments @xcite that for the no - chain copper oxide materials the doped holes may introduce a disorder in between the cuo@xmath0 planes, contrary to the case of the chain copper oxide materials @xcite, where the increasing doping reduces the disorder in between the cuo@xmath0 planes due to the effect of the cu - o chains. therefore for the no - chain copper oxide materials, the disorder introduced by doped holes residing between the cuo@xmath0 planes modifies the interlayer hopping elements as the random matrix elements. in this case, only the in - plane holon density of states (dos) @xmath107 enters the holon current - current correlation function as in disordered systems @xcite, and after a similar discussion as in sec. ii, we find that the corresponding momentum - nonconserving expression of the c - axis conductivity @xmath108 for the no - chain copper oxide materials is obtained by the replacement of the in - plane holon spectral function @xmath109 in eq. (15) with the in - plane holon dos @xmath110 as @xcite, @xmath111 where the @xmath112 is some average of the random interlayer hopping matrix elements @xmath113. we have performed a numerical calculation for this c - axis conductivity @xmath114, and the result at the doping @xmath115 (solid line), @xmath116 (dashed line), and @xmath92 (dash - dotted line) for the parameters @xmath93, @xmath117 with the temperature @xmath118 is plotted in fig. this result shows that the c - axis conductivity @xmath108 contains two bands, the higher - energy band, corresponding to the midinfrared band in the in - plane conductivity @xmath30 @xcite, shows a broad peak at @xmath119. the weight of this band is increased with dopings, but the peak position does not appreciably shift to lower energies. as a consequence of this pinning of the transferred spectral weight, the weight of the lower - energy band, corresponding to the non - drude peak in @xmath30 @xcite, is quite small and does not form a well - defined peak at low energies in the underdoped and optimally doped regimes. in this case, the conductivity @xmath108 at low energies can not be described by the over - damped drude - like formula even an @xmath120-dependence of the in - plane holon scattering rate has been taken into consideration. in comparison with the momentum - conserving @xmath89, our result also shows that at low energies the suppression of the momentum - nonconserving @xmath114 is due to the disordered effect introduced by doped holes residing between the cuo@xmath0 planes. these results are qualitatively consistent with the experimental results of the no - chain copper oxide materials @xcite. for the further understanding the transport property of the no - chain copper oxide materials, we have also performed the numerical calculation for the c - axis resistivity @xmath121, and the results at the doping @xmath115 and @xmath92 for the parameters @xmath93, @xmath117, and @xmath122 are shown in fig. 4(a) and fig. 4(b), respectively. in accordance with the c - axis conductivity @xmath108, the behavior of the c - axis resistivity @xmath123 in the underdoped and optimally doped regimes is the semiconducting - like at low temperatures, and metallic - like at higher temperatures. in comparison with the in - plane resistivity @xmath1 @xcite, it is shown that the values of the c - axis resistivity @xmath123 for the no - chain copper oxide materials are by @xmath124 orders of magnitude larger than these of the in - plane resistivity @xmath1 in the corresponding energy range, which are also qualitatively consistent with the experimental results of the no - chain copper oxide materials @xcite.
Summary and discussions
in the above discussions, the central concerns of the c - axis charge dynamics in copper oxide materials are the two dimensionality of the electron state and incoherent hopping between the cuo@xmath0 planes, and therefore the c - axis charge dynamics in the present fermion - spin picture based on the charge - spin separation is mainly determined by the in - plane charged holon fluctuation for the chain copper oxide materials and the in - plane charged holon fluctuation incorporating with the interlayer disorder for the no - chain copper oxide materials. in comparison with the in - plane resistivity @xmath1 @xcite, it is shown that the crossover to the semiconducting - like range in @xmath9 is obviously linked with the crossover from the temperature linear to the nonlinear range in @xmath1, _ i.e. _, they should have a common origin. in the fermion - spin theory @xcite, the charge and spin degrees of freedom of the physical electron are separated as the holon and spinon, respectively. although both holons and spinons contributed to the charge and spin dynamics, it has been shown that the scattering of spinons dominates the spin dynamics @xcite, while the results of the in - plane charge dynamics @xcite and present c - axis charge dynamics show that scattering of holons dominates the charge dynamics, the two rates observed in the experiments @xcite are attributed to the scattering of two distinct excitations, spinons and holons. it has been shown that an remarkable point of the pseudogap is that it appears in both of spinon and holon excitations @xcite. the present study indicates that the observed crossovers of @xmath125 and @xmath126 for copper oxide materials seem to be connected with the pseudogap in the in - plane charge holon excitations, which can be understood from the physical property of the in - plane holon dos. the numerical result of the in - plane holon dos @xmath110 at the doping @xmath92, @xmath90, and @xmath127 for the parameter @xmath93 at the temperature @xmath96=0 is shown in fig. 5(a), fig. 5(b), and fig. 5(c), respectively. for comparison, the corresponding mean - field result (dashed line) is also shown in fig. 5. while the in - plane holon density of states @xmath110 in the underdoping @xmath92 as a function of energy for the temperature (a) @xmath118, (b) @xmath128, and (c) @xmath129 is plotted in fig. 6. from fig. 5 and fig. 6, we therefore find that the in - plane holon dos in mfa consists of the central part only, which comes from the noninteracting particles as pointed in ref. @xcite. after including fluctuations the central part is renormalized and two side bands @xcite and a v - shape holon pseudogap near the chemical potential @xmath23 in the underdoped regime appear. but these two side bands are almost doping and temperature independent, while the v - shape holon pseudogap is doping and temperature dependent, and grows monotonously as the doping @xmath130 decreases, and disappear in the overdoped regime. moreover, this holon pseudogap also decreases with increasing temperatures, and vanishes at higher temperatures. since the full holon green s function (then the holon spectral function and dos) is obtained by considering the second - order correction due to the spinon pair bubble, then the holon pseudogap is closely related to the spinon fluctuation. for small dopings and lower temperatures, the holon kinetic energy is much smaller than the magnetic energy, _ i.e. _, @xmath131, in this case the magnetic fluctuation is strong enough to lead to the holon pseudogap. this holon pseudogap would reduce the in - plane holon scattering and thus is responsible for the metallic to semiconducting crossover in the c - axis resistivity @xmath126 and the deviation from the temperature linear behavior in the in - plane resistivity @xmath125 @xcite. this holon pseudogap will also lead to form the normal - state gap in the system, and the similar result has been obtained from the doped _ kagom _ and triangular antiferromagnets @xcite, where the strong quantum fluctuation of spinons due to the geometric frustration leads to the normal - state gap. with increasing temperatures or dopings, the holon kinetic energy is increased, while the spinon magnetic energy is decreased. in the region where the holon pseudogap closes, at high temperatures or at higher doping levels, the charged holon scattering would give rise to the temperature linear in - plane resistivity as well as the metallic temperature dependence of the c - axis resistivity. our results also show that @xmath1 is only slightly affected by this holon pseudogap @xcite, while @xmath9 is more sensitive to the underlying mechanism. in summary, we have studied the c - axis charge dynamics of copper oxide materials in the underdoped and optimally doped regimes within the @xmath12-@xmath13 model by considering the incoherent interlayer hopping. our result shows the c - axis charge dynamics for the chain copper oxide materials is mainly governed by the scattering from the in - plane fluctuation, and the c - axis charge dynamics for the no - chain copper oxide materials is dominated by the scattering from the in - plane fluctuation incorporating with the interlayer disorder, which would be suppressed when the holon pseudogap opens at low temperatures and lower doping levels, leading to the crossovers to the semiconducting - like range in the c - axis resistivity @xmath9 and the temperature linear to the nonlinear range in the in - plane resistivity @xmath1. because copper oxide materials are very complex systems, it is also possible that the actual c - axis conductivity may be a linear combination of the momentum - conserving @xmath89 and momentum - nonconserving @xmath108, but we believe that @xmath89 should be the major part of the c - axis conductivity in the chain copper oxide materials, while @xmath132 should be the major part of the c - axis conductivity in the no - chain copper oxide materials. the authors would like to thank professor ru - shan han, professor h. q. lin, and professor t. xiang for helpful discussions. this work was supported by the national natural science foundation under grant no. 19774014 and the state education department of china through the foundation of doctoral training. s. l. cooper _ et al. b*47 *, 8233 (1993) ; s. l. cooper, _ et al. _, lett. * 70 *, 1533 (1993) ; j. schtzmann _ et al. * 73 *, 174 (1994) ; a. v. puchkov _ et al. _, 77 *, 1853 (1996) ; c. c. homes _ et al. * 71 *, 1645 (1993) ; | the c - axis charge dynamics of copper oxide materials in the underdoped and optimally doped regimes has been studied by considering the incoherent interlayer hopping.
it is shown that the c - axis charge dynamics for the chain copper oxide materials is mainly governed by the scattering from the in - plane fluctuation, and the c - axis charge dynamics for the no - chain copper oxide materials is dominated by the scattering from the in - plane fluctuation incorporating with the interlayer disorder, which would be suppressed when the holon pseudogap opens at low temperatures and lower doping levels, leading to the crossovers to the semiconducting - like range in the c - axis resistivity and the temperature linear to the nonlinear range in the in - plane resistivity. | cond-mat0001260 |
Introduction
the study of cellular processes at the single - molecule level is a flourishing field of research in biology. individual molecules labeled with sub - micron markers can now be tracked in a cellular environment, and quantitative information about their dynamics can be obtained by reconstructing their trajectory. one of the most used techniques for this purpose is single - molecule fluorescence microscopy (smfm), which relies on a labeling with nanometer - sized fluorescent markers such as organic dyes or quantum dots. but standard smfm provides no information on the axial position of the marker, limiting this technique to 2d tracking. recent improvements of smfm such as astigmatism optic@xcite, @xmath1pi microscopy @xcite, double - helix psf @xcite, or multi - planes detection @xcite have made possible 3d tracking. since the depth of field of these techniques is limited to a few microns, 3d tracking of molecules that explore larger distances in the thickness of a sample requires to continuously adjust the position of the focal plane of the microscope objective, which strongly limits the time resolution. digital holographic microscopy (dhm) @xcite circumvents this drawback. in dhm, a ccd camera records the interference pattern between the light scattered by the sample and a reference wave, and a single shot is sufficient to determine the 3d positions of scatterers embedded in a non - diffusing environment, over a depth of typically a hundred of microns. as the scattering cross section of a particle scales as the sixth power of its radius @xcite, how easily and accurately a particle can be detected strongly depends on its size. several publications demonstrate the tracking of micron - sized colloids by using in - line holography @xcite, with a localization accuracy in the nanometer range through the use of high numerical aperture (na) microscope objectives. for example, with a @xmath2 na 1.4 oil immersion objective, cheong et al. @xcite reported lateral and axial localization accuracies of 4 and 20 nm respectively. this result was obtained with polystyrene spheres of diameter @xmath3 m, whose scattering cross section is quite large. the tracking of @xmath4 nm particles, whose scattering cross section is extremely low, is much more difficult, and, as far as we know, has not been demonstrated using in - line holography. yet the detection of such small particles is possible using dhm in a off - axis geometry @xcite with a noise level as low as possible @xcite and using good light scatterers, such as gold nanobeads @xcite. by this way, @xmath5 to @xmath6 nm gold particles embedded in an agarose gel have been detected and localized @xcite. since gold particles are not toxic for cells, they can be used as markers in biology @xcite, and d = 40 nm gold nanobeads fixed on the membrane receptors of a living cell have been localized @xcite. more recently, 3d tracking of batio3 particles with second harmonic generation dhm was also demonstrated @xcite. here, the main advantage of dhm, with respect to in - line holography, is the possibility to independently adjust the intensity of the illumination and reference beams in order to get the best detection sensitivity, by adjusting the reference beam intensity @xcite, and the largest signal, by adjusting the illumination beam intensity. combining dhm with dark field illumination allows then to detect nanometer - sized particles, as the sample can be illuminated with an intensity as large as possible, while avoiding a saturation of the camera @xcite. for example, atlan et al. and warnasooriya et al. uses total internal reflection (tir) to detect and localize @xmath5 nm and @xmath7 nm particles @xcite. but the tir configuration used in these experiments yield a standing wave which does not allow to track moving particles : when a moving particle crosses a node, the illumination (and thus the signal) goes down to zero, and the particle is lost. dubois et al. uses another dark field illumination configuration that focuses the illumination on a mask @xcite. since the illumination is parallel to the optical axis, no standing wave can appear, but since the illumination passes through the microscope objective, one expects parasitic reflections of the illumination beam. in this paper, we present a digital holographic microscopy technique which makes possible to track @xmath8 nm gold particles 3d diffusing in water. the illumination is parallel to the optical axis to prevent the formation of standing waves, and the holographic signal is collected by a na=0.5 dark field reflecting microscope objective. with such objective, the illumination beam is masked before the microscope objective and parasitic reflections are avoided. this yields high dynamic dark field illumination, which makes possible to detect, localize and track @xmath8 nm particles. first we describe the setup, which combines dark - field microscopy and off - axis holography. then we present the algorithm of reconstruction, our procedure to localize the beads, and describe how we can reach a real - time localization by performing calculations on a graphic card. finally we show that our setup allows us to track gold nanoparticles in motion with a lateral (@xmath9) resolution of @xmath10 nm and an axial (@xmath11) resolution of @xmath12. since na=0.5, the resolution (especially in @xmath11) is lower than with na=1.4 in - line holography @xcite. we also show that the depth of field of our holographic microscope is made two orders of magnitude larger than in optical microscopy.
Digital holography setup
our dhm experimental setup, depicted in fig. [fig : setup], is designed to investigate the brownian motion of @xmath13 diameter gold particles diffusing in a @xmath14 (length @xmath15 width @xmath15 height) cell chamber (ibidi @xmath16) filled with water. the concentration of nanoparticles is adjusted to @xmath17 particles / mm@xmath18 to have a few particles per field of view. the light source is a @xmath19 diode pump solid state laser (crystal laser) with a short coherence length (@xmath20) to avoid parasitic interferences raising from reflexions between the optical elements. the laser beam is split into two beams by a polarizing beam splitter (pbs), a half - wave plate before the pbs setting the ratio of energy between the emerging beams. the reference beam passes through a dove prism fixed on a micrometer translation stage to adjust the length of the optical path. this beam is spatially filtered through a @xmath21 m diameter pinhole and then expanded as to uniformly cover the ccd chip of the camera (512 @xmath15 512 pixels, andor luca r). the illumination beam is focalized on the sample with a plano - convex lens of focal length @xmath22 cm (waist diameter @xmath23, laser intensity @xmath24 w/@xmath25). the light scattered by the beads is collected in transmission with a dark - field reflecting objective (edmund optics, reflx series) of @xmath26 and @xmath27 magnification. a small mask on the input of the objective limits the collection of light between @xmath28 and @xmath26, so the illumination beam is totally blocked after passing through the sample. this dark - field configuration prevents the saturation of the ccd chip. a non - polarizing @xmath29 beam splitter behind the objective combines the scattered light with the reference beam. the ccd camera records the interference pattern on @xmath30 frames with a @xmath31 rate. the last beam - splitter is tilted by few degrees to be in off - axis configuration.
Numerical reconstruction
in digital holography, the ccd sensor records an intensity @xmath32 which is the interference of the reference beam with the light scattered by the nano - objects. @xmath32 is thus a sum of 4 terms : @xmath33 where @xmath34 is the intensity of the reference beam, @xmath35 the intensity of the scattered light and @xmath36 are the electric fields for the reference beam and the scattered light respectively. as the scattering cross section of a @xmath0 diameter gold particle calculated with the rayleigh - mie scattering model is about @xmath37 at @xmath19 @xcite, the integration over the collection solid angle of the objective (@xmath38) gives @xmath39 for a single particle. thus @xmath35 can be neglected compared to the other terms of eq.([eq : intensity]). the field @xmath40, the amplitude of which is about @xmath41, contains the phase of the scattered light necessary to a 3d localization of the particle. in the fourier space, the tilted beam splitter adds a spatial frequency on the two conjugates terms of interference, and thus the different terms of @xmath32 are spatially separated, i.e. @xmath34 remains centered on the zero frequency of the fourier plane while the two cross terms of interference are centered on the spatial frequency induced by the off - axis geometry. since 3d reconstruction is a time consuming task even using recent multi - cores processors, we developed parallel calculations on a graphic card (nvidia geforce gt470, 448 cores) @xcite. our ` c++ ` based algorithm uses the nvidia cuda library to decompose the calculations on the gpus of the card. among the existing methods to reconstruct holograms, the most common is the convolution method described by schnars et al. @xcite. the main drawback of this method is that the pixel size of the reconstructed image depends on the distance of reconstruction, so that the image of a thick sample is distorted. therefore this method is not convenient for 3d tracking, as the lateral scale depends on the depth of reconstruction considered. here we chose to reconstruct the holograms with the angular spectrum method @xcite, which, by compensating the sphericity of the signal wave, allows to reconstruct the hologram without distortion. our algorithm can be decomposed in 6 steps : i. _ subtraction of the background : _ : : in order to increase the signal - to - noise ratio, we subtract from the last recorded frame the average of the ten previous frames. phase shifting holography @xcite is also an effective technique for reducing noise, but the minimal delay @xmath42 ms between two frames, which is driven by our camera, is too long to use this technique for nanoparticle tracking. this step of calculation is performed only for particles in motion (i.e. we skipped this step for the results presented in [fixed]). numerical correction of the signal wave sphericity : _ : : the hologram @xmath32 is multiplied by a complex phase matrix @xmath43 to compensate the sphericity induced by the microscope objective on the signal wave : @xmath44 where @xmath45 is the local radius of curvature of the wave on the ccd plane. if the reference wave is a plane wave, this distance @xmath45 is also the distance between the ccd chip and the back focal plane of the objective. _ first fft : _ : : the direct fourier transform of the corrected hologram is calculated using the cuda cufft library : @xmath46.\]] fig. [fig : recons](a) shows the intensity @xmath47 in the _ k - space _, in logarithmic scale in the case where the background is not removed (step (i) skipped). in the middle of fig. [fig : recons](a), the zero - order appears as a square because of the multiplication by the matrix @xmath43. the term related to @xmath40 is in the down - right corner, centered on the spatial frequency induced by the off - axis geometry. the term related to @xmath48 is centered on the conjugate frequency (top - left corner). at this step, the calculation is equivalent of the reconstruction in one fft (convolution method) of the hologram at the distance @xmath45 described in the previous step. since the back focal plane of our microscope objective coincides with the output pupil plane, which is common for high magnification objective, we see on the down - right corner a sharp reconstruction of the output pupil plane. if we change in step (i) the parameter @xmath45 to @xmath49, the image of the output pupil would be sharp in the top - left corner, while the term related to @xmath40 would be blurred. [fig : recons](b) shows @xmath47 in the _ k - space _ when the background is removed (step (i) performed). the zero - order term in the middle of the _ k - space _ is largely removed compare to fig. [fig : recons](a), which reduces the recovery between the zero - order and @xmath40. _ spatial filtering and centering : _ : : to remove the zero - order term and replace the term related to @xmath40 in the middle of the fourier plane, a round numerical filter which matches with the output pupil of the objective is applied. since the shape of the pupil is sharp in the k - space, we can isolate precisely the pixels containing the signal, minimizing the loss of information. to more precisely calibrate the radius and the center of the filter, we used a diffusive paper as a sample before performing experiments. since the paper scatters light uniformly, all the spatial frequencies that the microscope objective can collect are recorded. [fig : recons](d) shows the intensity @xmath47 in logarithmic scale when the sample is replaced by a diffusing paper. we clearly see the shape of the output pupil of the objective and the dark - field mask in the center. we set the filter mask to match with the shape of the output pupil (white dotted circle in fig. [fig : recons](d)). the filtered part is then translated into the middle of a @xmath50 calculation grid in order to compensate the off - axis shift. v. _ propagation : _ : : the matrix obtained is multiplied by a propagation matrix @xmath51, which is the exact form of the matrix propagation as given by kim et al. @xcite : @xmath52 where @xmath53 to propagate the hologram by a distance @xmath11 in the axial direction. @xmath54 is the magnified pixel size. these equations are suited for holograms of @xmath50 pixels. _ second fft : _ : : finally the inverse fft is calculated. for each hologram, the steps (v) and (vi) are repeated in order to get a stack of the scattered field at different depths, with a propagation step @xmath55 : @xmath56,\]] where @xmath57 is an integer. in logarithmic scale. the zero - order term appears as a red square distorted by the multiplication with the phase matrix @xmath43. the term of interest is located in the down - right corner. b : @xmath58 when the average of the ten previous holograms is subtracted before calculating the fft. the zero - order term is largely removed, so the recovery between this term and the region of interest in the down - right corner is reduced. c : two - dimensional reconstruction at a fixed depth of the sample. a gold nanobead (1) is localized in this plan and the shape of the intensity of the field scattered by other beads at other depths (2 and 3) is visible. d : intensity @xmath58 in logarithmic scale when the sample is replaced by a diffusive paper. the area of the output pupil of the objective is sharply defined. the white - dotted circle shows the mask of the numerical filter used for the reconstruction.,height=10] an example of reconstruction is shown in fig. [fig : recons](c). we can see a nanoparticle (1) localized in the considered plane and the intensity of the field scattered by two particles at different depths (2 and 3). the reconstruction of a @xmath59 thick volume requires to calculate one fft to reconstruct the hologram in the output pupil plane, then to calculate 500 inverse ffts for the slices of the stack. these 501 ffts typically require one minute when the calculation is performed on the cpu, even with a recent multi - cores computer. we reduce this time by a factor of 30 when the calculation is parallelized on the graphic card. once the three dimensional map of the scattered field is calculated, the beads are localized by pointing the local maxima of the field s intensity. [fig : psf] shows the field s intensity in the @xmath60,@xmath61 and @xmath62 directions. the full width at half maximum (fwhm) is about @xmath63 in lateral direction, and @xmath64 in the axial direction as expected for the point spread function associated with our microscope objective@xcite. in order to reach a sub - pixel size resolution, the intensity is fitted by a gaussian curve in @xmath60 and @xmath61 using the pixel for which the intensity is maximum and the two adjacent pixels. as the intensity profile in @xmath62 is not gaussian and 10 times larger than in @xmath60 and @xmath61, we fit the maximum of the peak with a parabola using the pixel @xmath65 for which the intensity is maximum and the two adjacent pixels @xmath66 and @xmath67 (fig. [fig : psf](d)). we chose this simple localization method because programming a more elaborate fit, as t - matrix theory based computation @xcite, with cuda (programming language of the graphic card) is more complicated and would considerably slow down the process. our method shows good performance (see fig. [fig : reso] and fig. [fig : error]), but a better resolution may be achieved using t - matrix theory.
Results
to evaluate the lateral and axial localization accuracy of our setup, we localize a single @xmath0 diameter gold particle embedded in a 1% agarose gel. the localization accuracy is evaluated by calculating the standard deviation of 200 positions of the bead obtained from successive frames, with an exposure time for each frame @xmath68. for a particle in the focal plane of the objective, we found a lateral localization accuracy of @xmath69 and an axial localization accuracy of @xmath70 (fig. [fig : reso](a) and [fig : reso](b)). then the distance between the particle and the focal plane was increased by steps of @xmath71. for each step we recorded 200 holograms and determined the mean position of the bead as well as the lateral and axial accuracy. [fig : error] compares the mean axial position of the particle with the mechanical displacement along @xmath62 of the sample, which are in excellent agreement. the localization accuracy in @xmath60,@xmath61 and @xmath62 as a function of the axial position of the bead is shown in fig. [fig : reso](a) and [fig : reso](b). while the lateral accuracy is constant (@xmath69) for @xmath72 nm, the axial accuracy slightly depends on @xmath62. it is about @xmath70 around @xmath73, then decreases to @xmath74 for @xmath72 nm. this accuracy strongly increases for @xmath75, and for @xmath76, the localization of the particle is not possible because the scattered signal level reaches the noise level. the local maximum at @xmath73 observed on the axial accuracy curve (fig. [fig : reso](b)) shows that the localization is not optimal when the gold particle is the focal plane of the objective. in this case the particle is imaged on a small area of the ccd chip, so that the interference pattern spreads on a small number of pixels, which degrades the quality of the reconstruction @xcite. in the case of particle tracking, the position of the focal plane in the sample has to be fixed. to minimize the spherical aberrations due to the presence of the coverslip, the focal plane should coincide with the sample - coverslip interface. in this configuration, a moving particle can not cross the focal plane, so that the localization accuracy remains optimal and constant over a depth of @xmath77, that corresponds to the part between abscissa 0 and abscissa 250 in fig. [fig : reso](a) and [fig : reso](b). yet tracking particles remains possible when the focal plane is set above the sample - coverslip interface as demonstrated below, as the cost of a slightly worse axial localization accuracy. we now consider gold particles in brownian motion. according to the stokes - einstein equation, the diffusion coefficient @xmath78 of the particles is given by : @xmath79 where @xmath80 is the boltzmann constant, @xmath81 the room temperature (@xmath82), @xmath83 the viscosity of water (1.0mpa.s at @xmath82) and @xmath84 nm the radius of the nanobead (size dispersity given by the provider bbinternational). the exposure time is @xmath68 and the time between two frames @xmath85. the mean distance traveled along one direction by a brownian particle during @xmath86 is @xmath87, which is smaller than the lateral size of the magnified pixels (@xmath88). consequently, the signal from a particle is not blurred over several pixels during @xmath86. the mean distance covered along a given axis during @xmath89 is @xmath90, which corresponds to approximately @xmath91 pixels. as shown in fig. [fig : motion], our method allows us to simultaneously track several particles. a volume of @xmath92 (@xmath93), i.e. @xmath94 pixels, can be reconstructed from a single hologram, and the localization method described in the section _ [method] _ can be performed to localize several beads with a sub - pixel accuracy. by repeating the algorithm for successive frames, we could for instance reconstruct the trajectories of 3 gold particles diffusing in water (fig. [fig : motion]). we were able to track particles during up to 10 s (@xmath95 frames). since the time needed to reconstruct a volume of @xmath94 pixels is about 0.5 s (i.e. much larger than @xmath96 ms), reconstruction is necessarily a post - processing process in this case. in contrast, when a single particle is tracked, 3d localization in a given frame requires to reconstruct only a few slices around the position of the particle in the previous frame. for a @xmath0 particle, the mean distance traveled along @xmath62 during @xmath89 is @xmath97, thus only 24 reconstructions (12 reconstructions above / below the previous position of the particle) are sufficient to find the new position of the particle. the calculation of these 24 slices from the hologram requires @xmath98, which is smaller than @xmath96 ms. real - time tracking is thus possible for a single particle under the condition that reconstruction is performed fast enough, for instance by using a graphic card unit as described above. to evidence the brownian motion, we calculated the 3d mean square displacement (msd) of a bead from the red trajectory plotted in fig. [fig : motion] : @xmath99 where @xmath100 is the total number of positions, as well as the 1d msd along the directions @xmath60, @xmath61, and @xmath62. for a brownian motion with a diffusion constant @xmath78, the msd curve depends linearly on time, and the slope of the curve is @xmath101, where @xmath102 for the 3d msd (fig. [fig : msd](a)) and @xmath103 for 1d msd (fig. [fig : msd](b), (c) and (d)). as expected for a brownian motion, experimental 3d and 1d msd depend linearly on time, and a fit of the first six points of the curves gives @xmath104 along @xmath60, @xmath105 along @xmath61, @xmath106 along z and @xmath107 for the 3d motion. these values are in agreement with the theoretical value @xmath108 predicted by eq. ([stokes]).
Conclusion
in this paper, we show that digital holographic microscopy can be used to track @xmath109 gold particles diffusing in water. as the intensity of the light scattered by such nanoparticles is height orders of magnitude smaller than that of the excitation light, we combined holography to standing wave free dark - field microscopy to completely block the illumination beam, thereby preventing a saturation of the ccd chip of the camera. a single hologram, recorded with an exposure time of only 1 ms, is sufficient to localize several particles in a 250 @xmath110 m thick sample, with a lateral (@xmath9) localization accuracy of @xmath69 and an axial (@xmath11) localization accuracy of @xmath74. as our dark field microscope involves a na=0.5 reflecting microscope objective, the resolution, especially in @xmath11, obtained here with nanoparticles (@xmath8 nm), is lower than the resolution reached in in - line holography using a na=1.4 objective and micron - sized objects (@xmath3 m) @xcite. this is the cost to pay for the detection of nanometer - sized particles. we were able to reconstruct particle trajectories, evidence the brownian nature of the motion and determine the related diffusion coefficient. the accuracy achieved by our setup is comparable with that reached with super - resolution microscopy : a localization accuracy of @xmath111 nm in x and y, and @xmath112 nm in z has been reported for fluorescence dye or quantum dots imaged with storm microscopy using optical astigmatism @xcite, and an accuracy of @xmath113 nm over a depth of @xmath114 m has been reached using a double helix psf @xcite. severals results of 3d tracking of quantum dots using palm microscopy report an axial accuracy between @xmath113 and @xmath115 nm, over a depth of typically @xmath116 m @xcite. palm and storm microscopy reach higher 3d localization accuracies than our dhm setup, but over a depth two order of magnitude lower. this ability to track nanoparticles up to @xmath117 m from the focal plane with a constant localization accuracy is the strength of dhm compare to super - resolution techniques. our results pave the way for the use of gold nanobeads as markers in more complex media such as cellular environment.
Acknowledgments
this work was supported by funds from the french national research agency (anr simi 10 and anr 3d brom), centre national de la recherche scientifique (cnrs) and cole normale suprieure (ens). the authors thank mathieu coppey, fred etoc and jasmina dikic for their suggestions and careful reading. | we present a new technique that combines off - axis digital holography and dark field microscopy to track @xmath0 gold particles diffusing in water.
we show that a single hologram is sufficient to localize several particles in a thick sample with a localization accuracy independent of the particle position. from our measurements
we reconstruct the trajectories of the particles and derive their 3d diffusion coefficient.
our results pave the way for quantitative studies of the motion of single nanoparticle in complex media.
99 b. huang, w. wang, m. bates, and x. zhuang, `` three - dimensional super - resolution imaging by stochastic optical reconstruction microscopy, '' science * 319, * 810 (2008).
j. bewersdorf, b.t.
bennett, and k.l.
knight, `` h2ax chromatin structures and their response to dna damage revealed by 4pi microscopy, '' proc.
nat.
acad.
sci.
usa * 103, * 18137 (2006).
s.r.
pavani, m.a.
thompson, j.s.
biteen, s.j.
lord, n. liu, r.j.
twieg, r. piestun, and w.e.
moerner, `` three - dimensional, single - molecule fluorescence imaging beyond the diffraction limit by using a double - helix point spread function, '' proc.
nat.
acad.
sci.
usa * 106, * 2995 (2009).
m.f juette, t.j.
gould, m.d.
lessard, m.j.
mlodzianoski, b.s.
nagpure, b.t.
bennett, s.t.
hess, and j. bewersdorf, `` three - dimensional sub100 nm resolution fluorescence microscopy of thick samples, '' nature methods * 5, * 527529 (2008).
s. ram, p. prabhat, j. chao, e.s.
ward, and r.j.
ober, `` high accuracy 3d quantum dot tracking with multifocal plane microscopy for the study of fast intracellular dynamics in live cells, '' biophysical journal * 95, * 6025 (2008).
u. schnars and w. jptner, `` direct recording of holograms by a ccd target and numerical reconstruction, '' 179181 (1994).
e. leith and j. upatniek, `` wavefront reconstruction photography, '' physics today * 18, * 26 (1965).
h.c.
van de hulst, `` light scattering by small particles, '' (dovers publications inc, 1957).
f.c.
cheong, b. sun, r. dreyfus, j. amato - grill, k. xiao, l. dixon, and d.g.
grier, `` flow visualization and flow cytometry with holographic video microscopy, '' 1307113079 (2009).
f.c.
cheong, s. duarte, s.h.
lee, and d.g.
grier, `` holographic microrheology of polysaccharides from streptococcus mutans biofilms '' rheologica acta * 48, * 109115 (2009).
f.c.
cheong, b.j.
krishnatreya, and d.g.
grier, `` strategies for three - dimensional particle tracking with holographic video microscopy, '' 1356313573 (2010).
w. xu, m.h.
jericho, h.j.
kreuzer, and i.a.
meinertzhagen, `` tracking particles in four dimensions with in - line holographic microscopy, '' 164166 (2003).
j. sheng, e. malkiel, and j. katz, `` digital holographic microscope for measuring three - dimensional particle distributions and motions, '' 38933901 (2006).
s.h lee, y. roichman, g.r.
yi, s.h.
kim, s.m.
yang, a. van blaaderen, p. van oostrum, and d.g.
grier, `` characterizing and tracking single colloidal particles with video holographic microscopy, '' 1827518282 (2007).
m. speidel, l. friedrich, and a.rohrbach, `` interferometric 3d tracking of several particles in a scanning laser focus, '' 10031015 (2009).
e. leith and j. upatnieks, `` microscopy by wavefront reconstruction, '' 569-570 (1965).
e. cuche, p. marquet, and c. depeursinge, `` spatial filtering for zero - order and twin - image elimination in digital off - axis holography, '' 40704075 (2000).
m. gross and m. atlan, digital holography with ultimate sensitivity 909911 (2007) p.k.
jain, k.s.
lee, i.h.
el - sayed, and m.a.
el - sayed, `` calculated absorption and scattering properties of gold nanoparticles of different size, shape, and composition : applications in biological imaging and biomedicine, '' j. phys.
chem.
b * 110, * 7238-7248 (2006).
m. atlan, m. gross, p. desbiolles, .
absil, g. tessier, and m. coppey - moisan, `` heterodyne holographic microscopy of gold particles, '' 500502 (2008).
l. cognet, c. tardin, d. boyer, d. choquet, p. tamarat, and b. lounis, `` single metallic nanoparticle imaging for protein detection in cells, '' proc.
nat.
acad.
sci.
usa * 100, * 11350 (2003).
n. warnasooriya, f. joud, f. bun, g. tessier, m. coppey - moisan, p. desbiolles, m. atlan, m. abboud, and m. gross, `` imaging gold nanoparticles in living cell environments using heterodyne digital holographic microscopy, '' 32643273 (2010).
e. shaffer, p. marquet, and c. depeursinge, `` real time, nanometric 3d - tracking of nanoparticles made possible by second harmonic generation digital holographic microscopy, '' 1739217403 (2010).
f. verpillat, f. joud, m. atlan, and m. gross, digital holography at shot noise level journal of display technology * 6, * 455464 (2010) f. dubois and p. grosfils, dark - field digital holographic microscopy to investigate objects that are nanosized or smaller than the optical resolution 26052607 (2008) b. samson, f. verpillat, m. gross, and m. atlan, `` video - rate laser doppler vibrometry by heterodyne holography, '' 14491451 (2011).
t. shimobaba, y. sato, j. miura, m. takenouchi, and t. ito, `` real - time digital holographic microscopy using the graphic processing unit, '' 1177611780 (2008).
l. ahrenberg, a.j.
page, b.m.
hennelly, j.b.
mcdonald, and t.j.
naughton, `` using commodity graphics hardware for realtime digital hologram view - reconstruction, '' j. display technol.
* 5, * 111 (2009).
h. kang, f. yara, and l. onural, `` graphics processing unit accelerated computation of digital holograms, '' 137143 (2009).
f. le clerc, l. collot, and m. gross, `` numerical heterodyne holography with two - dimensional photodetector arrays, '' 716718 (2000).
l. yu and m.k.
kim, `` wavelength - scanning digital interference holography for tomographic three - dimensional imaging by use of the angular spectrum method, '' 20922094 (2005).
i. yamaguchi and t. zhang, `` phase - shifting digital holography, '' 1261270 (1997).
m.k.
kim, l. yu, and c.j.
mann, `` interference techniques in digital holography, '' journal of optics a : pure and applied optics * 8, * s518s523 (2006).
m. j. nasse and j. c. woehl, `` realistic modeling of the illumination point spread function in confocal scanning optical microscopy, '' a * 27, * 295302 (2010).
psf lab, http://onemolecule.chem.uwm.edu/software. c. fournier, l. denis, and t. fournel, `` on the single point resolution of on - axis digital holography, '' a * 27, * 18561862 (2010). | 1112.4737 |
Historical notes
soon after the fermi - dirac form (in 1926) of statistical mechanics was proposed for particles which obey pauli s exclusion principle (in 1925), fowler (1926) realized that the electron degeneracy pressure can balance for those stars, called as white dwarfs, discovered by astronomers in 1914. by a numerical calculation (1931) for a polytropic gas of extremely relativistic electrons, chandrasekhar found a unique mass, which was interpreted as a mass limit of white dwarfs. landau (1932) presented an elementary explanation of the chandrasekhar limit by considering the lowest total energy of stars, and recognized that increasing density favors energetically the formation of neutrons, discovered only several months before by chadwick, through the action @xmath0. a very massive object with much high density may have almost neutrons in the chemical equilibrium, which was then called as _ neutron stars _ (nss). detailed calculations of ns structures showed (e.g., oppenheimer & volkoff 1939) that an ns can have a mass of @xmath1, but is only @xmath2 km in radius, which makes it hard to be observed by astronomers. however, on one hand, a few authors do investigate possible astrophysical implications of nss. for example, baade & zwicky (1934) combined the researches of supernovae, cosmic rays, and nss, and suggested that nss may form after supernovae ; pacini (1967) even proposed that the stored energy in rotational form of an ns could be pumped into the supernova remnant by emitting electromagnetic waves. on the other hand, ns models were developed with improved treatments of equation of states, involving not only \{@xmath3 }, but also mesons and hyperons. the cooling behavior of nss was also initiated in 1960s due to the discovery of x - ray sources which were at first though mistakenly to be the thermal emission from nss. the discovery of _ radio pulsars _ by hewish & bell (and their coauthors 1968) is a breakthrough in the study, and this kind of stars were soon identified as spinning nss by gold (1968). since then more and more discoveries in other wave bands broaden greatly our knowledge about these pulsar - like compact stars (plcss), including x - ray pulsars, x - ray bursts, anomalous x - ray pulsars, soft @xmath4-ray repeaters, and rosat - discovered `` isolated neutron stars ''. it is still a current concept among astrophysicists that such stars are really nss. ns studies are therefore in two major directions : 1, the emission mechanisms for the stars, both rotation - powered and accretion - powered ; 2, the ns interior physics. however, neutrons and protons are in fact _ not _ structureless points although they were thought to be elementary particles in 1930s ; they (and other hadrons) are composed of _ quarks _ proposed by gell - mann and zweig, respectively, in 1964. the quark model for hadrons developed effectively in 1960s, and ivanenko & kurdgelaidze (1969) began to suggest a quark core in massive compact stars. itoh (1970) speculated about the exist of 3-flavor _ full _ quark stars (since only @xmath5, @xmath6 and @xmath7 quarks were recognized at that time), and even calculated the hydrostatic equilibrium of such quark stars which are now called as _ strange stars _ (sss). is it possible that strange stars really exist in nature? the possibility increases greatly if the bodmer - witten s conjecture is correct : bodmer (1971) initiated the discussion of quark matter with lower energy per baryon than normal nucleus, whereas witten (1984) considered an assumption of stable 3-flavor quark matter in details and discussion extensively three aspects related (1, dark baryon and primordial gravitational wave due to the cosmic separation in the qcd epoch ; 2, strange quark stars ; 3, cosmic rays). farhi & jaffe s (1984) calculation in the mit bag model showed that the energy per baryon of strange matter is lower than that of nucleus for qcd parameters within rather wide range although we can hardly prove weather the bodmer - witten s conjecture is correct or not from first principles. haensel, zdunik & schaeffer (1986) and alcock, farhi & olinto (1986) then modelled sss, and found that sss can also have typical mass (of @xmath8) and radius (of @xmath2 km), which mean that _ the pulsar - like compact stars believed previously to be nss might actually be sss_. yet the most important and essential thing in the study is : how to distinguish sss from nss observationally? more and more ss candidates appeared recently in literatures (e.g., bombaci 2002, xu 2002). it is generally suggested that sss as radio pulsars, the most popular ones of plcss, should have crusts (with mass @xmath9) being similar to the outer crusts of nss (witten 1984, alcock et al. but this view was criticized by xu & qiao (1998), who addressed that _ bare _ strange stars (bsss, i.e., sss without crusts) being chosen as the interior of radio pulsars have three advantages : 1, the spectral features ; 2, the bounding energy ; and 3, the core collapse process during supernova. it is thus a new window to distinguish bsss from nss via their magnetosphere and surface radiation according to the striking differences between the exotic quark surfaces of bsss and the normal matter surfaces of nss. with regard to the possible methods of finding strange stars in literatures, hard evidence to identify a strange star may be found by studying only the surface conditions since the other avenues are subject to many complex nuclear and/or particle physics processes that are poorly known. thanks to those advanced x - ray missions, it may be a very time for us to identify real strange stars in the universe. it is worth mentioning that, though some authors may name a general term `` _ neutron star _ '', regardless of that the stars are `` neutron '' or `` strange '', it is actually not suitable to call a real ss to be an ns since no _ neutron _ in an ss.
The standard model of particle physics and the quark matter
one of the most great achievements in the last century is the construction of the standard model in particle physics (e.g., cottingham & greenwood 1998), which asserts that the material in the universe is made up of elementary fermions (divided into quarks and leptons) interacting though gauge bosons : photon (electromagnetic), w@xmath10 and z@xmath11 (weak), 8 types of gluons (strong), and graviton (gravitational). there are totally 62 types of `` building blocks '' in the model. besides the 13 types of gauge bosons, there are three generations of fermions (1st : \{@xmath12, e ; u, d }, 2nd : \{@xmath13, @xmath14 ; c, s }, and 3rd : \{@xmath15, @xmath16 ; t, b}. note that each types of quarks has three colors) and their antiparticles. the final one, which is still not discovered, is the higgs particle that is responsible to the origin of mass. it is a first principle in the yang - mills theory that an interaction satisfies a corresponding local gauge symmetry, sometimes being broken spontaneously due to vacuum phase transition. the gauge theory for electromagnetic and weak interactions is very successful, with a high precision in calculation ; whereas one can treat gravity using einstein s general relativity theory if the length scale is not as small as the plank scale (@xmath17 cm) although a gauge theory of gravity is still not successful. as for the gauge theory for strong interaction, the quantum chromodynamics (qcd), however, is still developing, into which many particle physicists are trying to make efforts. nonetheless, qcd has two general properties. for strong interaction in small scale (@xmath18 fm), i.e., in the high energy limit, the interacting particles can be treated as being _ asymptotically free _ ; a perturbation theory of qcd (pqcd) is possible in this case. whereas in larger scale (@xmath19 fm), i.e., in the low energy limit, the interaction is very strong, which results in _ color confinement_. the pqcd is not applicable in this scale (many non - perturbative effects appear then), and a strong interaction system can be treated as a system of hadrons in which quarks and gluons are confined. in this limit, we still have effective means to study color interaction : 1, the lattice formulation (lqcd), with the discretization of space - time and on the base of qcd, provide a non - perturbative framework to compute numerically relations between parameters in the standard model and experiments by first principles ; 2, phenomenological models, which rely on experimental date available at low energy density, are advanced for superdense hadronic and/or quark matter. these two properties result in two distinct phases (rho 2001) of hadronic matter, depicted in the qcd phase diagram in terms of temperature @xmath20 vs. baryon chemical potential @xmath21 (or baryon number density). hadron gas phase locates at the low energy - density limit where both @xmath20 and @xmath21 are relatively low, while a new phase called _ quark gluon plasma _ (qgp) or _ quark matter _ appears in the other limit when @xmath20 _ or _ @xmath21 is high although this new state of matter is still not found with certain yet. it is therefore expected that there is a kind of phase transition from hadron gas to qgp (or reverse) at critical values of @xmath20 and @xmath21. actually a deconfinement transition is observed in numerical simulations of lqcd for zero chemical potential @xmath22, when @xmath23 mev. can we find real quark matter? certainly we may improve very much the knowledge about the strong interaction by studying the matter s various properties if a qgp state is identified in hand. one way is to create high energy - density fireball in laboratory through the collisions of relativistic heavy ions in accelerators. quark matter is expected if the center of the fireball reaches a temperature of @xmath24, but the qgp is hadronized soon and there are final states of hadrons detected. it is thus a challenge to find clear signatures of qgp without ambiguousness in these experiments. another way is to search celestial bodies which contain quark matter via cosmological or stellar processes. strange stars ( 3) are very probably such kind of objects, which are a kind of bulk qgp with mass @xmath25, composed of nearly equal number of up, down, and strange quarks (called as strange quark matter, or strange matter). some possible observational signatures of sss are also addressed in literatures. the first way to detect this new form of matter is in terrestrial laboratory physics (lab - physics), whereas the second is in astrophysics. they compensate each other. up to date, the research both in lab - physics and in astrophysics faces a general difficulty in finding quark matter : to search a _ clear _ signature for its existence. in fact, the investigations in lab - physics and astrophysics are in two _ different _ regions in the qcd phase diagram. available monte carlo simulations of lqcd are only applicable for cases with @xmath22, and thus give valuable guidance for lab - physics. however, it is a very different story with sss, where the density effects dominate (@xmath26). it turns out, for technical reasons, to be extremely difficult to study strange matter by lqcd, and we have to rely on phenomenological models to speculate on the properties of sss by extrapolating our knowledge at nuclear matter density. for an ss, with high @xmath21 but low @xmath20, one can not naively think that it is a simple qgp ; in fact many interesting phenomena, e.g., color superconductivity (alford et al. 2001) of strange matter, are discussed. in this meaning, sss provide only examples to study such hadronic system at high density, and one thus has to learn `` experimental date '' from astrophysics.
Strange stars
that strange stars (rather than neutron stars) are residual after core - collapse type supernova explosion depends upon 1, quark deconfinement occurs, 2, strange matter in bulk is absolutely stable (bodmer - witten s conjecture). unfortunately no certain answer to these issues is obtained theoretically ( 2). nevertheless, we still can obtain general concepts about those two requirement for forming sss by following simple arguments. in the view of bag model (called as the `` bag constant '').], the vacuum in and out hadrons is different ; in - hadron is of `` pqcd vacuum '' where the strong interaction is weak and pqcd is applicable, but out - hadron is of `` qcd vacuum '' where the nonlinear strong coupling between quarks or gluons is in control. as baryon number density @xmath27 gets higher and higher, the pqcd vacuum in nucleons becomes more and more dominated ; if nucleon keeps a radius @xmath28 fm, the qcd vacuum disappears when @xmath29 (@xmath30 is the density of ordinary nuclear matter). depending on rotation frequency and stellar mass, such density are easily surpassed in the cores of neutron stars (weber 1999). therefore the first requirement quark deconfinement may be satisfied. in case of @xmath31, @xmath5 and @xmath6 quarks deconfined from nucleons could have a fermi energy quark and 2 @xmath6 quarks, and the fermi energy @xmath32 (@xmath33 the number density of quarks). if @xmath34, @xmath35 mev for @xmath5 quark and @xmath36 mev for @xmath6 quark.] @xmath37 mev, which is much larger than the mass of @xmath7 quark @xmath38 mev ; the system should be energetically favorable by opening of a third flavor degree of freedom (@xmath7 quark) since high - energy @xmath5 and @xmath6 quarks are expected to decay into @xmath7 quarks via weak interactions. the second requirement may thus also be reasonable. it is worth addressing that, if only the first requirement is satisfied while strange matter is not absolutely stable but only metastable, a mixed phase where bulk quark and nuclear matter could coexist over macroscopical distances may appear in ns cores (such nss are called as `` mixed stars '', see, e.g., heiselberg & hjorth - jensen 2000, for a review). these two requirements may lead to the existence of strange matter in nature. in a simplified version of the bag model, assuming quarks are massless and noninteracting, we then have quark pressure @xmath39 (@xmath40 is the quark energy density) ; the total energy density is @xmath41 but the total pressure is @xmath42. one therefore have the equation of state for strange matter, @xmath43 (actually strange matter may have baryons from several hundreds, called as strangelets, to about that of our sun, called as strange stars.) one can obtain the mass @xmath44 and radius @xmath45 of an ss by integrating numerically the tov equation, with a strange matter equation of state [e.g., eq.(1)], assuming a certain central density (or pressure). the mass - radius relations of sss calculated with various strange matter equations of states are very different from that of nss, which may provide a possible way to differentiate sss from nss ( 3.2). but an ss usually can have a mass of @xmath46, and correspondent radius @xmath2 km. it is worth noting that the apparent temperature @xmath47 and radius @xmath48 (also called as radiation radius, defined by @xmath49, where @xmath50 is the apparent luminosity at infinity and @xmath51 is the stefan - boltzmann constant), i.e., the values observed at infinity, is not that observed at the stellar surface (@xmath45 is also the schwartzschild radius coordinate at the surface, defined by @xmath52). for a compact star with mass @xmath44 and radius @xmath45, the relations are (haensel 2001) @xmath53, @xmath54, where @xmath55 is the schwartzschild radius. as the strange quarks are more massive (@xmath38 mev) than the up (@xmath56 mev) and down (@xmath57 mev) quarks, some electrons are required to keep the chemical equilibrium of an ss. this brings some interesting properties near the bare quark surface. since quarks can be bound through strong interaction, whereas electrons by much weaker interaction (only the electromagnetism), the electrons can thus spread out the quark surface and be distributed in such a way that a strong outward static electric field is formed. adopting a simple thomas - fermi model, one can deduce analytical expressions for the electron number density @xmath58 and the electric field @xmath59 above the surface (xu & qiao 1999), @xmath60 where @xmath61 is a measured height above the quark surface, @xmath62 cm). very strong electric field, @xmath63 v / cm from eq.(2), should be near the quark surface, which makes it possible to support a normal - matter crust with mass @xmath64 (alcock et al. sss covered by such crusts are called as crusted strange stars (csss), while sss without crusts as bare strange stars (bsss). can radio pulsars, the largest population of plcss, be bsss? the answer was no in alcock et al s (1986) paper : `` pulsar emission mechanisms which depend on the stellar surface as a source of plasma will not work if there is a bare quark surface '' and `` the universe is a dirty environment and a bare strange star may readily accrete some ambient material ''. their first point is certainly incomplete, because @xmath65 pairs produced rapidly in strong electro - magnetic fields [as expressed in eq.(2), or induced by the unipolar effect] should create a corotating magnetosphere although no charged particles can be pulled out into the magnetosphere (xu & qiao 1998). a nascent protostrange star should be bare because of strong mass ejection and high temperature (usov 1998) after the supernova detonation flame ; an ss can hardly accrete due to rapid rotating and strong magnetic field (xu et al. 2001) ; even accretion is possible, a crust still can not form as long as the accretion rate is not much larger than the eddinton one (xu 2002). their second point is thus also not reasonable. therefore radio pulsars could be bsss. furthermore, it is found that bsss may be better for explaining the observations of radio pulsars as well as other plcss ( 3.3). although the ss idea is not new, ss identification becomes a hot topic only in recent years because of advanced techniques in space. since sss may have similar masses and radii, which are conventional quantities observable in astronomy, to that of nss, it is very difficult to find observational signals of quark matter in the plcss. it was argued that the ss and ns cooling behaviors could be distinguishable, since sss may cool much faster than nss (e.g. pizzochero 1991). however, recent more complete analyses on this issue indicate that this may be impossible except for the first @xmath6630 years after their births (schaab et al. nevertheless, the author thinks there may still be three effective ways to do. the minimum rotation periods of sss are smaller than that of nss. rotating stars composted of ideal fluid are subject to rotation - mode instability, which leads to the loss of rotation energy by gravitational radiation and results in substantial spindown. however the matter of a real star is not ideal but have viscosity ; the calculated bulk viscosity, based on the work of wang & lu (1984), of strange matter is much higher than that of neutron matter although their shear viscosities are similar ; therefore sss could have smaller periods at which their higher viscosity can prevent them from developing the instability (madsen 1998, 2000). the 2.14 ms optical source in sn 1987a (middleditch et al 2000) should be an ss if being confirmed in further observations. the approximate mass - radius (@xmath67) relations of sss (@xmath68) are in surprising contrast to that of nss (@xmath69), and sss can have much small radii. comparisons of observation - determined relations in x - ray binaries with modelled ones may thus tell if an object is an ns or an ss (li et al. also, a plcs with radius @xmath70 km could be an ss (drake et al. 2002). there are striking differences between the surfaces of bsss and that of nss. the very properties of the quark surface, e.g., strong bounding of particles, abrupt density change from @xmath71g/@xmath72 to @xmath73 [eq.(1)] in @xmath19 fm, and strong electric fields, may eventually help us to identify of a bss ( 3.3). three parts of possible evidence for bsss are discussed below. although pulsar emission mechanism is not well understood, the rs - type (ruderman & sutherland 1975) sparking model is still the popular one to connect magnetospheric dynamics with general observations, with an `` user friendly '' nature. maybe the strongest support to the rs - type vacuum gap model is the drifting subpulses observed from some pulsars. however rs model faces at least two difficulties for nss : the binding energy problem and the antipulsar issue, which can be solved completely if radio pulsars are bsss (xu et al. 1999). the soft @xmath4-ray burst of sgr 0526 - 66, with peak luminosity @xmath74, needs ultra - strong field (@xmath63 g) to constrain the fireball. an alternative bounding is through the quark surface ; and it may be natural to explain the bursting energy and the light curves in a framework that a comet - like object falls to a bss (zhang et al. 2000, usov 2001). bsss are expected to have featureless spectra (of both surface thermal and magnetospheric non - thermal components) since no ion is above the quark surface or in the magnetosphere, except for electron cyclotron lines due to the landau levels appeared in strong fields (xu & qiao 1998, xu 2002). recent observations known hitherto of several plcss actually show featureless spectra except for two sources 1e1207 and sgr1806, the lines of which may originate from landau level transitions in suitable field strength for the space facilities (astro - ph/0207079).
Conclusions
the theoretical bases of sss are, to some extent, solid in physics and the formation of strange matter stars is possible in astrophysics ; sss could thus exist. although each of the observed phenomena from plcss may be interpreted under the ns regime with unusual or artificial physical properties, it could be a natural way to understand the observations by updating nss with sss. | a pedagogical overview of strange quark matter and strange stars is presented. after a historical notation of the research and an introduction to quark matter,
a major part is devoted to the physics and astrophysics of strange stars, with attention being paid to the possible ways by which neutron stars and strange stars can be distinguished in astrophysics.
recent possible evidence for _ bare _ strange stars is also discussed.
# 1_#1 _ # 1_#1 _ = # 1 1.25 in.125 in.25 in | astro-ph0211348 |
Introduction
intensity modulated radiation therapy (imrt) is usually used for head and neck cancer patients because it delivers highly conformal radiation doses to the target with reduction of toxicity to normal organs, as compared with conventional radiation therapy techniques @xcite. volumetric modulated arc therapy (vmat) is a novel imrt technique. vmat has less mu, less treatment time, high quality planning and more efficiency than static gantry angle imrt @xcite. during vmat the linear accelerator (linac) control system changes the dose rate and the multi leaf collimator (mlc) positions while gantry is rotating around the patient. collimator angle is usually rotated in the plans of vmat to reduce radiation leakage between mlc leaves. at a zero angle, the leakage between mlc leaves accumulates during the gantry rotation and the summed leakage results in unwanted dose distributions, which can not be controlled by optimization. at different collimator angles, the unwanted doses can be controlled by dose constraints in the optimization procedure so that we can reduce the unwanted doses. the optimal collimator angle for vmat plan is thus required to be determined. there are several factors for consideration in the choice of the collimator angle of the vmat plan. among them we concentrated on the accuracy of the vmat delivery. we studied the effect of the collimator angle on the results of dosimetric verifications of the vmat plan for nasopharyngeal cancer (npc).
Materials and methods
ten patients with late - stage nasopharyngeal cancer were treated with concurrent chemo radiation therapy (ccrt). eight patients had stage iii disease and 2 patients had stage iv disease according to american joint committee on cancer staging system 7. nine patients were male and 1 patient was female. one radiation oncologist delineated radiation targets and organs at risk (oars). the clinical target volume (ctv) included the primary nasopharyngeal tumor, neck nodal region and subclinical disease. considering the setup uncertainty, margins ranging from 3 - 10 mm were added to each ctv to create a planning target volume (ptv). reduced - field techniques were used for delivery of the 66 - 70 gy total dose. the treatment plan course for each patient consisted of several sub - plans. in this study, we selected the first plan with prescribed doses of 50 - 60 gy in 25 - 30 fractions to study the effect of the collimator angles on dosimetric verifications of the vmat. the radiation treatment planning system eclipse v10.0.42 (varian medical systems, usa) was used to generate vmat plans. the vmat (rapidarc : varian) plans were generated for clinac ix linear accelerator using 6 mv photons. the clinac ix is equipped with a millennium 120 mlc that has spatial resolution of 5 mm at the isocenter for the central 20 cm region and of 10 mm in the outer 2@xmath110 cm region. the maximum mlc leaf speed is 2.5 cm / s and leaf transmission is 1.8%. dosimetric leaf gap of the mlc was measured using the procedure recommended by varian medical systems. the value of the dosimetric leaf gap was 1.427 mm for 6 mv photons. for volume dose calculation, grid size of 2.5 mm, inhomogeneiy correction, the anisotropic analytical algorithm (aaa) v10.0.28 and the progressive resolution optimizer (pro) v10.0.28 were used in all plans. vmat plans for npc patients were composed of 2 coplanar full arcs in 181 - 179 degree clockwise and 179 - 181 degree counterclockwise directions. the 2 full - arc delivery was expected to achieve better target coverage and conformity than the single arc @xcite. we generated 10 vmat plans (plan set a) with different collimator angles for each patient. ten collimator angles for the first arc were 0, 5, 10, 15, 20, 25, 30, 35, 40 and 45 degrees. for the second arc, the collimator angle was selected explementary to the collimator angle of the first arc in the same plan, i.e., the 2 collimator angles added up to 360 degree. the average field size of vmat plans was 22 @xmath1 22 @xmath2. we used the same dose constraints for all the 10 vmat plans and optimization was conducted for each plan. the maximum dose rate was 600 mu / min. the target coverage was aimed to achieve a 100% volume covered by 95% of prescribed dose. optimization of each plan resulted in different fluences and different mlc motions for each plan. therefore we had 2 variables, i.e., the collimator angle and mlc motions. to simplify the analysis we generated another set of 10 plans (plan set b) with the same mlc motions and different collimator angles for each patient. the mlc motions were those of the plan with 30 degree collimator angle. the plans in this set had different dose distributions and usually can not be used for treatment purposes excepting the plan with a 30 degree collimator angle. we performed patient specific quality assurances (qa) of 2 sets of 10 vmat plans for each patient. the measurements were made by the 2-dimensional ion chamber array matrixx (iba dosimetry, germany) @xcite. the matrixx has 1020 pixel ion chambers arranged in a 32@xmath132 matrix covering 24.4@xmath124.4 @xmath2. each ion chamber has the following dimensions : 4.5 mm in diameter, 5 mm in height and a sensitive volume of 0.08 @xmath3. the distance between chambers is 7.619 mm. the matrixx has an intrinsic buildup and backscatter thicknesses of 0.3 mm and 3.5 mm, respectively. the matrixx was placed between solid water phantoms multicube (iba dosimetry, germany) (figure [fig1]) so that thickness of total buildup and backscatter was 5 cm (figure [fig2]). the source to surface distance was 95 cm with the measurement plane of the matrixx at the isocenter of the linac. measurement was done for each arc in the plan ; therefore, we conducted 40 measurements for each patient and the total number of measurements was 400. the angular dependence of the matrixx was corrected after the measurements using the gantry angle sensor @xcite (iba dosimetry, germany). the comparison between the calculations and the measurements were made by @xmath0-index (2%/2 mm, 3%/3 mm) analysis @xcite using omnipro imrt v1.7b (iba dosimetry, germany). the @xmath0-index was calculated only for the regions that have dose values above 10% @xcite in the measured area.
Results and discussion
average @xmath0-index passing rates of patient specific qas were given in table [table1]. the results were averaged over the 2 arcs and 10 patients. because the 2 arcs in each vmat plan rotated almost 360 degrees and the measurement set - up is mirror symmetric about the measurement plane (@xmath4 plane in figure [fig2]) of the matrixx detector and a vertical plane passing through the isocenter (@xmath5 plane in figure [fig2]) the arc with collimator angle @xmath6 is symmetric to the arc with collimator angle @xmath7. therefore we regarded the collimator angle of the second arc, which was equal to 360 minus the collimator angle of the first arc, as the collimator angle of the first arc in the analysis..@xmath0-index passing rates of the patient specific qas as a function of the collimator angle [cols="^,^,^,^,^ ",] [table1] maximum difference between @xmath0-index (2%/2 mm) passing rates of plans in plan set a for each patient ranged from 2.83% to 14.32% and the average value was 8.44@xmath84.24%. using the 3%/3 mm criteria the maximum difference ranged from 1.46% to 5.60% and the average value was 3.67@xmath81.29%. maximum difference between @xmath0-index (2%/2 mm) passing rates of plans in plan set b for each patient ranged from 3.71% to 10.44% and the average value was 7.97@xmath82.17%. using the 3%/3 mm criteria the maximum difference ranged from 1.46% to 7.23% and the average value was 4.69@xmath82.51%. 2-dimensional dose distributions calculated by the eclipse treatment planning system, dose distributions measured by the matrixx detector and @xmath0-index (3%/3 mm) distributions of 1 patient plans in the plan set a for collimator angle 5 and 35 degree were shown in figure [fig3 - 1] and [fig3 - 2], respectively. the passing rate for the 35 degree collimator angle was less than the passing rate for the 5 degree collimator angle. -index (3%/3 mm) distributions. in the @xmath0-index distributions red color indicates the region where the 3%/3 mm criteria failed., width=566] ) plan in the plan set a for collimator angle 35. the first figure is the 2-dimensional dose distribution calculated by the eclipse treatment planning system. the second one is the dose distribution measured by the matrixx detector. the last one is the @xmath0-index (3%/3 mm) distributions. in the @xmath0-index distributions red color indicates the region where the 3%/3 mm criteria failed., width=566] the increase in collimator angle resulted in decreased @xmath0-index passing rates, as shown in figure [fig4]. in the figure, passing rates were normalized to the value of 0 degree. black and white squares indicated @xmath0-index (2%/2 mm) and (3%/3 mm) passing rates, respectively, averaged over plan set a. black and white triangles indicated @xmath0-index (2%/2 mm) and (3%/3 mm) passing rates, respectively, averaged over plan set b. -index passing rates (2%/ 2 mm) of patient specific delivery qas as a function of the collimator angle., width=377] there were statistically significant negative correlations between the collimator angle and the @xmath0-index passing rates. pearson correlation coefficients for pair - wise ratings of the @xmath0-index (2%/2 mm) and (3%/3 mm) passing rates of plans in the plan set a and b were -0.524 and -0.412, respectively with p - values @xmath9 0.001. for accuracy of vmat a smaller collimator angle is better, and for mlc leakage a larger collimator angle is better, we were thus required to make a compromise. based on this study, in our hospital the collimator angles of the vmat plans for head and neck patients range between 15 - 25 degrees because the average @xmath0-index passing rates were above or near to 90% for the 2%/2 mm criteria and 97% for the 3%/3 mm criteria, as shown in the results of the passing rates for the plan set a (table [table1]). in other hospitals these results can be somewhat different because they have different vmat delivery systems and diffterent vmat planning systems. we think that they can find optimal collimator angles by conducting the similar measurements described in this article. although not included in this article, we performed the patient specific qas for other treatment sites with smaller field sizes that are @xmath9 13 @xmath1 13 @xmath2. maximum difference of the passing rates for vmat plans with various collimator angles was @xmath9 1.5%. collimator angle does not affect the accuracy of the vmat delivery with small field sizes. the accuracy of radiation delivery by the linac depends on geometrical accuracies such as gantry isocentricity, collimator isocentricity and mlc position. it was reported that leaf limiting velocity, mlc position and mechanical isocenter varied at different collimator and gantry angles @xcite. this may explain the @xmath0-index passing rates dependence on the collimator angle. further study is needed to investigate the origin of the collimator angle dependence of the accuracy of vmat delivery. the quality of the plan itself is another factor for consideration in the choice of the collimator angle of the vmat plan. optimized dose distributions with the same dose constraints can vary according to the collimator angle of the vmat plan. further study is needed to evaluate the quality of vmat plans with different collimator angles.
Conclusions
we found that the results of the patient specific qas for vmat plans using the 2-dimensional ion chamber array matrixx are dependent on the collimator angle of the vmat plans. the @xmath0-index (2%/2 mm) and (3%/3 mm) passing rates were negatively correlated with the collimator angle. we showed that collimator angles of the vmat plans for head and neck cancer patients range between 15 - 25 degrees resulting in the average @xmath0-index passing rates above or near to 90% for the 2%/2 mm criteria and 97% for the 3%/3 mm criteria. f. k. lee et al., med. 39, 44 (2014). m w. k kan et al., j. appl. 13, 6 (2012). s. a. syam kumar et al., rep. radiother. 8, 87 (2013). a. holt et al., radiat. 8, 26 (2013). j. alvarez - moret et al., raiat. oncol. 5, 110 (2010). t. lee et al., j. appl. 12, 4 (2011). x. jin et al., med. 38, 418 (2013). j. herzen, et al. 52, 1197 (2007). l. d. wolfsberger et al., j. appl. 11, 1 (2010). m. rao et al., med. 37, 1350 (2010) s. korreman et al., acta oncologica. 45, 185 (2009) m. stasi et al. 39, 7626 (2012). g. a. ezzell et al., med. 36, 5359 (2009). c. c. ling et al. j. radiat. 72(2), 575 (2008). m. okumura et al., phys. 55, 3101 (2010). | collimator angle is usually rotated when planning volumetric modulated arc therapy (vmat) due to the leakage of radiation between multi - leaf collimator (mlc) leaves.
we studied the effect of the collimator angles on the results of dosimetric verification of the vmat plans for head and neck patients.
we studied vmat plans for 10 head and neck patients.
we made 2 sets of vmat plans for each patient.
each set was composed of 10 plans with collimator angles of 0, 5, 10, 15, 20, 25, 30, 35, 40, 45 degrees.
plans in the first set were optimized individually and plans in the second set shared the 30 degree collimator angle optimization.
two sets of plans were verified using the 2-dimensional ion chamber array matrixx (iba dosimetry, germany).
the comparison between the calculation and measurements were made by the @xmath0-index analysis.
the @xmath0-index (2%/2 mm) and (3%/3 mm) passing rates had negative correlations with the collimator angle.
maximum difference between @xmath0-index (3%/3 mm) passing rates of different collimator angles for each patient ranged from 1.46% to 5.60% with an average of 3.67%.
there were significant differences (maximum 5.6%) in the passing rates of different collimator angles.
the results suggested that the accuracy of the delivered dose depends on the collimator angle.
these findings are informative when choosing a collimator angle in vmat plans. | 1503.03569 |
Introduction
@xcite found that the galactic cepheids follow a spectral type that is independent of their pulsational periods at maximum light and gets later as the periods increase at minimum light. * hereafter skm) used radiative hydrodynamical models to explain these observational phenomena as being due to the location of the hydrogen ionization front (hif) relative to the photosphere. their results agreed very well with code s observation. skm further used the stefan - boltzmann law applied at the maximum and minimum light, together with the fact that radial variation is small in the optical @xcite, to derive : @xmath3 where @xmath4 are the effective temperature at the maximum / minimum light, respectively. if @xmath5 is independent of the pulsation period @xmath6 (in days), then equation (1) predicts there is a relation between the @xmath7-band amplitude and the temperature (or the colour) at minimum light, and vice versa. in other words, if the period - colour (pc) relation at maximum (or minimum) light is flat, then there is an amplitude - colour (ac) relation at minimum (or maximum) light. equation (1) has shown to be valid theoretically and observationally for the classical cepheids and rr lyrae variables @xcite. for the rr lyrae variables, @xcite and @xcite used linear and non - linear hydrodynamic models of rrab stars in the galaxy to explain why rrab stars follow a flat pc relation at _ minimum _ light. later, @xcite used macho rrab stars in the lmc to prove that lmc rrab stars follow a relation such that higher amplitude stars are driven to cooler temperatures at maximum light. similar studies were also carried out for cepheid variables, as in skm, @xcite, (*??? * hereafter paper i) and (*??? * hereafter paper ii). in contrast to the rr lyrae variables, cepheids show a flat pc relation at the _ maximum _ light, and there is a ac relation at the minimum light. therefore, the pc relation and the ac relation are intimately connected. all these studies are in accord with the predictions of equation (1). in paper i, the galactic, large magellanic cloud (lmc) and small magellanic cloud (smc) cepheids were analyzed in terms of the pc and ac relations at the phase of maximum, mean and minimum light. one of the motivations for this paper originates from recent studies on the non - linear lmc pc relation (as well as the period - luminosity, pl, relation. see paper i ; *??? * ; *??? * ; *??? *) : the optical data are more consistent with two lines of differing slopes which are continuous or almost continuous at a period close to 10 days. paper i also applied the the @xmath2-test @xcite to the pc and ac relations at maximum, mean and minimum @xmath7-band light for the galactic, lmc and smc cepheids. the @xmath2-test results implied that the lmc pc relations are broken or non - linear, in the sense described above, across a period of 10 days, at mean and minimum light, but only marginally so at maximum light. the results for the galactic and smc cepheids are similar, in a sense that at mean and minimum light the pc relations do not show any non - linearity and the pc(max) relation exhibited marginal evidence of non - linearity. for the ac relation, cepheids in all three galaxies supported the existence of two ac relations at maximum, mean and minimum light. in addition, the cepheids in these three galaxies also exhibited evidence of the pc - ac connection, as implied by equation (1), which give further evidence of the hif - photosphere interactions as outlined in skm. to further investigate the connection between equation (1) and the hif - photosphere interaction, and also to explain code s observations with modern stellar pulsation codes, galactic cepheid models were constructed in paper ii. in contrast to skm s purely radiative models, the stellar pulsation codes used in paper ii included the treatment of turbulent convection as outlined in @xcite. one of the results from paper ii was that the general forms of the theoretical pc and ac relation matched the observed relations well. the properties of the pc and ac relations for the galactic cepheids with @xmath8 can be explained with the hif - photosphere interaction. this interaction, to a large extent, is independent of the pulsation codes used, the adopted ml relations, and the detailed input physics. the aim of this paper is to extend the investigation of the connections between pc - ac relations and the hif - photosphere interactions in theoretical pulsation models of lmc cepheids, in addition to the galactic models presented in paper ii. in section 2, we describe the basic physics of the hif - photosphere interaction. the updated observational data, after applying various selection criteria, that used in this paper are described in section 3. in section 4, the new empirical pc and ac relations based on the data used are presented. in section 5, we outline our methods and model calculations, and the results are presented in section 6. examples of the hif - photosphere interaction in astrophysical applications are given in section 7. our conclusions & discussion are presented in section 8. throughout the paper, short and long period cepheid are referred to cepheids with period less and greater than 10 days, respectively.
The physics of hif-photosphere interactions
the partial hydrogen ionization zone (or the hif) moves in and out in the mass distribution as the star pulsates. it is possible that the hif will interact with the photosphere, defined at optical depth (@xmath9) of 2/3, at certain phases of pulsation. for example, skm suggested that this happened at maximum light for the galactic cepheids, as the hif is so far out in the mass distribution that the photosphere occurs right at the base of the hif. the sharp rise of the opacity wall (where the mean free path goes to zero) due to the existence of hif prevents the photosphere moving further into the mass distribution and hence erases any `` memory '' of global stellar conditions, including the underlying pc relation. this lead to a flat relation between period & temperature, period & colour and period & spectral type at maximum light, as seen in skm and paper ii. at other phases, since the hif does not interact with the photosphere, the temperature of the star (or the colour) follows the underlying global pc relation. the hif - photosphere interaction also relies on the properties of the saha ionization equation and the structural properties of the outer envelopes of cepheids. it is well known that the partition functions in the saha ionization equation are formally divergent unless some atomic physics is used to truncate them. in the pulsation codes we used, we approximate the partition functions of various atoms by their ground state statistical weights. the properties of the saha ionization equation in cepheid envelopes are such that hydrogen starts to ionize at a temperature that is almost independent of density, for a certain range of low densities. outside of this range of density, the density dependence increases. thus, when the photosphere is very close to, or engaged with the hif and the density of these regions is reasonably low, the temperature of the photosphere is less dependent on the surrounding density and hence the global stellar parameters. at higher densities, the temperature at which hydrogen ionizes becomes more sensitive to density and hence more sensitive to global stellar parameters. if the photosphere is far from the hif, or disengaged, then the location of the photosphere and hence the temperature of the photosphere, is again strongly dependent on density and hence on global stellar parameters. that is why the photosphere needs to be close to, or engaged with the hif for this effect to take place. moreover, this dependence on density is not sharp so that for `` low '' and `` high '' densities the density dependence of the photospheric temperature is weak and strong respectively. an examination of figure 15.1 in @xcite demonstrates that this is plausible. thus as the star pulsates, the photospheric temperature has a density dependence that can be strong or weak depending on phase. an example where the density dependence is weak are the galactic long period cepheids at maximum light (skm, paper ii) : these cepheids display a flat pc relation at maximum light. these properties of the hif - photosphere interaction can, in turn, affect the temperature of the photosphere and hence the colour of the cepheid. here we investigate the idea that lmc cepheids with periods below 10 days are such that the hif and photosphere are engaged through most of the pulsation cycle. at periods greater than 10 days, the photosphere only engages with the hif at maximum light. the transition is sharp because the photosphere is either at the base of the hif or it is not. the transition occurs because as the period increases, the @xmath10 ratio increases and this implies the hif is located further inside in the mass distribution, changing the phase at which it can interact with the photosphere @xcite. the structure of galactic cepheids is such that this interaction only occurs at maximum light, even for cepheids with periods shorter than 10 days.
Updated lmc cepheid data
in paper i, we constructed the light curves of fundamental mode cepheids in the lmc by using the extensive photometric dataset in the ogle (optical gravitational lensing experiment) database. however, the dataset used in paper i was downloaded in 2002, prior to the updated version of the dataset that was available after april 24, 2003 (ogle website, udalski 2004 [private communication]). the updated version includes additional @xmath7- and @xmath11-band data for most of the cepheids. in addition, the periods have been refined by the ogle team using the complete set of photometric data. due to these reasons, we decided to repeat the light curve construction @xcite with the updated data and periods. since the cepheids in the ogle database are truncated at @xmath12, due to the saturation of the ccd detector for the longer period (hence brighter) cepheids @xcite, we include some additional lmc cepheid data from @xcite, @xcite and @xcite to extend the period coverage to @xmath13 in our sample. the requirements that govern our choice of the published photometric data are : (a) latest observations that use the modern day ccd cameras ; (b) high quality data with large number of data points per light curve, which provide uniform phase coverage and small scatter of the light curve ; and (c) as homogeneous as possible (i.e., from a minimal number of sources) to avoid any additional systematic errors. these requirements are essential to construct accurate light curves to allow the estimation of colours and magnitudes at maximum, mean and minimum light for our pc and ac study. hence we did not include some of the older photometric data in this study. the photometric data of all cepheids, comprising 771 from ogle database, 14 from @xcite+@xcite and 39 from @xcite, were mainly fit with @xmath14 to @xmath15 fourier expansions (@xmath16 is the order of fourier expansion) using the simulated annealing method described in @xcite to the @xmath7- and @xmath11-band photometric data. this is in contrast to paper i that only applied @xmath14 fourier fits. however, for some of the ogle long period cepheids (@xmath17 days), it was found out that the quality of the fitted light curves could be improved by using a higher order fourier expansion, hence we extended the fit to @xmath18 for these long period cepheids. all the fitted light curves were visually inspected and the best - fit light curves from the different orders of the fourier expansions were selected. to the best of our knowledge, this analysis also represents a major improvement in the fourier analysis of the ogle data. the extinction is corrected with the standard procedure, i.e. @xmath19 with @xmath20 and @xmath21 @xcite. the values of @xmath22 for each ogle cepheids are taken from the ogle database @xcite, while for the cepheids in @xcite+@xcite and in @xcite, the values of @xmath22 are adopted from @xcite and/or @xcite. to guard against some `` bad '' cepheids or other contamination in our sample, and select only the good cepheids in both bands, we removed some cepheids in the sample according to the following criteria (see also *??? *) : = 7.5 cm 1. cepheids without @xmath7- and/or @xmath11-band photometry, or the number of data per light curve (in either bands or both) is too low to fit a @xmath14 fourier expansion. cepheids with poorly fitted or unacceptable @xmath7- and/or @xmath11-band light curves in the sample, such as those with a large scatter of data points or with bad - phase coverage (large gaps between the phased data points). most of the magnitudes, as well as the colours, at the maximum and/or minimum light from these fitted light curves are very uncertain. cepheids with possible duplicity in the ogle sample. some of the possible duplicated cepheids were removed in the ogle database by consulting table 4 of @xcite. 4. cepheids with unusual colour. we first plot out (as in figure [figcut][a]) the extinction corrected pc relation at mean light. the plot shows that there are number of outliers in the period - colour plane, mostly with @xmath23. the presence of these outliers is probably due to : (a) their extinction is either over- or under - estimated ; (b) they have blue or red companions that can not be resolved due to the problems of blending ; or (c) other unknown physical reasons. a detailed investigation of these outliers is beyond the scope of this paper, but it is clear that they should be removed from the sample. these outliers are removed with the adopted colour - cut of @xmath24, a compromise between maximizing the number of cepheids in the sample and excluding the cepheids with unusual colour. cepheids with unusually low (or high) amplitude. some cepheids with unusually low @xmath7- and @xmath11-band amplitudes were found in the sample. their amplitudes are typically @xmath25 times smaller as compared to the amplitudes of other cepheids at given period. some examples of the light curves for these low amplitude cepheids are given in @xcite. in addition, most of the light curves for these low amplitude cepheids can be fitted with @xmath14 fourier expansion, while other cepheids with `` normal '' amplitude may require higher order fits. @xcite has briefly discussed some possible physical reasons for these cepheids to have such low amplitudes, e.g. they are just entering or leaving the fundamental mode instability strip @xcite or they have different chemical composition (see, e.g., *??? the detailed investigation of these low amplitudes cepheids is beyond the scope of this paper. here, we apply a conservative amplitude cut of @xmath26 mag. in the @xmath7-band to remove the low amplitude cepheids. besides that, we also remove ogle-286532 (with unusually low amplitude) and hv-2883 (with unusually high amplitude) as they are clear outliers in the @xmath27-amplitude plot (not shown, but see *??? note that @xcite applied a cut of @xmath28 mag. to remove the low amplitude cepheids in ngc 6822. other examples of removing the low - amplitude cepheids can also be found in @xcite. 6. cepheids with @xmath29 and @xmath30. in order to guard against possible contamination from the first overtone cepheids @xcite and to be consistent with the previous studies @xcite, we removed cepheids with @xmath29 (see further justification in @xcite). regarding the removal of cepheids with @xmath30, a preliminary analysis of the pc relation reveals that few of the longest cepheids should be removed from the sample, because they are clear outliers in the pc plot at maximum light (see upper panel of figure [figpcmax]). without these longest period cepheids, the pc(max) relation for the long period cepheids is flat, which is consistent with the results found in paper i. the hypothesis of the hif - photosphere interaction also suggests the flatness of the pc relation at maximum light for long period cepheids. however, as the period gets longer (with @xmath31>1.5 $]), the photosphere disengages from the hif @xcite. these longest period cepheids have biased the slope of the pc(max) relation by making the slope becomes steeper. these selection criteria are guided mainly by the philosophy that it is better to lose some `` bad '' but real cepheids rather than including those spurious and doubtful cepheids in the sample @xcite, or those with bad fitted light curves that will give inaccurate measurements of the maximum and minimum light. hence, the final sample consists of 641 lmc cepheids that will be considered further. the locations of the outliers from various selection criteria are shown in figure [figcut] for the pc(mean) relation, @xmath7-band pl relation, @xmath32-@xmath27 relation, where @xmath33 are the fourier amplitudes. see @xcite and @xcite for details.] and the colour - magnitude diagram (cmd). note that some of the outliers are located within the `` good '' cepheids. however they can be eliminated due to various physical reasons as given above, especially those with poorly fit light curves that will give inaccurate measurements at maximum, mean and minimum light. a simple sigma - clipping algorithm (e.g., *??? *) will not be able to remove these outliers @xcite.
The new empirical pc and ac relations
to construct the empirical pc & ac relations, we used the following quantities from the fourier fits to the cepheid data as obtained from previous section : * @xmath7-band amplitude : the difference of the numerical maximum and minimum from the fourier expansion, @xmath34. * @xmath35 : defined as @xmath36, where @xmath37 is the @xmath11-band magnitude at the same phase as @xmath38. * @xmath39 : defined as @xmath40, where @xmath41 is the mean value from the fourier expansion (see *??? this is very similar to the conventional definition of the mean colour, @xmath42, where @xmath43 denotes the intensity mean. * @xmath44 : defined as @xmath45, where @xmath46 is the @xmath11-band magnitude at the same phase as @xmath47. @xmath47 is the @xmath7-band magnitude closest to @xmath48, the mean value from fourier expansion. * @xmath49 : defined as @xmath50, where @xmath51 is the @xmath11-band magnitude at the same phase as @xmath52. these quantities have been corrected for extinction as mentioned in previous section. the empirical lmc pc and ac relations at maximum, mean and minimum light for all, long and short period cepheids are summarized in table [c9tabpc] & [c9tabac], and the corresponding plots are presented in figure [c9figpc] & [c9figac], respectively. lccc phase & @xmath53 & @xmath54 & @xmath55 + + maximum & @xmath56 & @xmath57 & 0.099 + mean & @xmath58 & @xmath59 & 0.075 + phmean & @xmath60 & @xmath61 & 0.081 + minimum & @xmath62 & @xmath63 & 0.075 + + maximum & @xmath64 & @xmath65 & 0.098 + mean & @xmath66 & @xmath67 & 0.075 + phmean & @xmath68 & @xmath69 & 0.092 + minimum & @xmath70 & @xmath71 & 0.081 + + maximum & @xmath72 & @xmath73 & 0.097 + mean & @xmath74 & @xmath75 & 0.074 + phmean & @xmath76 & @xmath77 & 0.078 + minimum & @xmath78 & @xmath79 & 0.073 + lccc phase & @xmath53 & @xmath54 & @xmath55 + + maximum & @xmath80 & @xmath81 & 0.093 + mean & @xmath82 & @xmath83 & 0.092 + phmean & @xmath84 & @xmath85 & 0.099 + minimum & @xmath86 & @xmath87 & 0.098 + + maximum & @xmath88 & @xmath89 & 0.086 + mean & @xmath90 & @xmath91 & 0.085 + phmean & @xmath92 & @xmath93 & 0.103 + minimum & @xmath94 & @xmath95 & 0.082 + + maximum & @xmath96 & @xmath97 & 0.076 + mean & @xmath98 & @xmath99 & 0.074 + phmean & @xmath100 & @xmath101 & 0.078 + minimum & @xmath102 & @xmath103 & 0.079 + to test the non - linearity of the pc and ac relations, or the `` break '' at a period of 10 days, we apply the @xmath2-test as given in paper i and in @xcite. the null hypothesis in the @xmath2-test is single line regression is sufficient, while the alternate hypothesis is that two lines regressions with a discontinuity (a break) at 10 days is necessary to fit the data. the probability @xmath104, under the null hypothesis, can be obtained with the corresponding @xmath2-values and the degrees of freedom. in general, the large value of @xmath2 (equivalent to the small value of @xmath105 $]) indicates that the null hypothesis can be rejected. for our sample, @xmath106 when @xmath107 (the 95% confident level), therefore the null hypothesis can be rejected if the @xmath2-value is greater than @xmath108 with more than 95% confident level and the data is more consistent with the two - line regression. a glance of table [c9tabpc] and figure [c9figpc] suggests that the lmc pc relations are broken at maximum, mean and minimum light. these are confirmed with the @xmath2-test results with @xmath109. similarly, the @xmath2-test results for the ac relation are : @xmath110. hence, the lmc pc and ac relations are non - linear (hence broken) at maximum, means and minimum light. note that the flatness of the long period pc(max) relation as given in table [c9tabpc] (@xmath64) is in good agreement with the slope found in paper i (@xmath111). recall that equation (1) predicts that if the pc relation is flat at maximum light, then there is a correlation between the amplitude and the colour at minimum light. this is seen in table [c9tabac] (and in figure [c9figac]) for the long period ac(min) relation, with a slope of @xmath112.
Methods and lmc models
the stellar pulsation codes we used are both linear @xcite and non - linear @xcite. these codes, which include a 1-d turbulent convection recipe @xcite, are the same as in paper ii. briefly speaking, the codes take the mass (@xmath113), luminosity (@xmath114), effective temperature (@xmath115) and chemical composition (@xmath116) as input parameters. the chemical composition is set to be @xmath117 to represent the lmc hydrogen and metallicity abundance (by mass). the mass and luminosity are obtained from the ml relations calculated from evolutionary models. the @xmath115 are chosen to ensure the models oscillate in the fundamental mode and located inside the cepheid instability strip. the pulsation periods for the models are obtained from a linear non - adiabatic analysis @xcite. all other parameters used in the pulsation codes had the same values for the lmc and galactic models (paper ii). this included the @xmath118 parameters that are part of the turbulent convection recipe, though see section 8. of course, one variable parameter was the metallicity. the only other difference between this study and paper ii, besides the metallicity, is the value set for the artificial viscosity parameter, @xmath119. in this study, we set @xmath120 for the lmc models to improve the shape of the theoretical light curves, in contrast to the value of @xmath121 used for the galactic models. in paper ii, the ml relations are adopted from @xcite and @xcite. in order to be consistent with previous work, the ml relations used in this paper will also be adopted from these two sources. however, @xcite only provided two ml relations, one for @xmath122 which are used in paper ii, and another one for @xmath123. hence we have to adopt the second ml relation for the lmc models. even though the lmc metallicity is higher than @xmath123, the lmc is still considered as a low metallicity system in the literature. hence the @xcite ml relation can be approximately applied for the lmc models. an anonymous referee pointed out that an interpolation of the @xcite ml relations between @xmath122 and @xmath123 should also be used. we have included the interpolated ml relation in our model calculations. in the context of the hif - photosphere interaction, it is the ml relation which dictates at what period and at what phases this will occur. stellar evolutionary theory changes the ml relation as a function of metallicity. hence the coefficients of the ml relation are important in determining the nature of the hif - photosphere interaction (paper ii). in short, the ml relations used are : ccccccc @xmath113 & @xmath124 & @xmath125 & @xmath126 & @xmath127 & @xmath128 & @xmath129 + + 11.0 & 4.375 & 5050 & 46.4155 & 0.124 & 28.98 & -0.118 + 10.0 & 4.236 & 5100 & 35.6727 & 0.091 & 22.92 & -0.093 + 9.50 & 4.161 & 5250 & 28.2406 & 0.094 & 18.68 & -0.046 + 9.10 & 4.099 & 5260 & 25.3960 & 0.082 & 16.92 & -0.042 + 8.75 & 4.042 & 5310 & 22.3804 & 0.076 & 15.07 & -0.027 + 8.40 & 3.982 & 5380 & 19.3886 & 0.071 & 13.20 & -0.008 + 7.95 & 3.902 & 5330 & 17.7750 & 0.055 & 12.09 & -0.027 + 7.00 & 3.717 & 5410 & 12.6085 & 0.035 & 8.722 & -0.018 + 6.55 & 3.620 & 5490 & 10.2940 & 0.031 & 7.183 & -0.006 + 6.40 & 3.587 & 5485 & 9.81474 & 0.027 & 6.853 & -0.010 + 6.00 & 3.493 & 5510 & 8.37226 & 0.020 & 5.866 & -0.014 + 5.90 & 3.468 & 5500 & 8.12498 & 0.017 & 5.691 & -0.018 + 5.80 & 3.443 & 5525 & 7.69466 & 0.017 & 5.400 & -0.015 + 5.70 & 3.418 & 5560 & 7.23505 & 0.017 & 5.090 & -0.009 + 5.30 & 3.312 & 5600 & 6.01283 & 0.012 & 4.244 & -0.009 + + 7.20 & 4.272 & 5380 & 40.2561 & 0.275 & 24.51 & -0.162 + 6.80 & 4.192 & 5380 & 35.4374 & 0.264 & 21.91 & -0.122 + 6.20 & 4.063 & 5410 & 28.2629 & 0.225 & 17.94 & -0.076 + 5.95 & 4.005 & 5420 & 25.6378 & 0.211 & 16.43 & -0.060 + 5.40 & 3.869 & 5510 & 19.4314 & 0.170 & 12.82 & -0.015 + 5.15 & 3.803 & 5510 & 17.5523 & 0.160 & 11.66 & -0.007 + 4.65 & 3.660 & 5490 & 14.3143 & 0.131 & 9.611 & -0.010 + 4.20 & 3.518 & 5510 & 11.3659 & 0.101 & 7.729 & -0.011 + 4.00 & 3.450 & 5545 & 10.0011 & 0.089 & 6.854 & -0.005 + 3.95 & 3.432 & 5540 & 9.77393 & 0.085 & 6.701 & -0.008 + 3.80 & 3.378 & 5550 & 8.94637 & 0.075 & 6.157 & -0.009 + 3.70 & 3.341 & 5575 & 8.31297 & 0.070 & 5.745 & -0.005 + 3.65 & 3.322 & 5570 & 8.10751 & 0.066 & 5.605 & -0.008 + 3.60 & 3.302 & 5530 & 8.09994 & 0.058 & 5.583 & -0.022 + 3.60 & 3.302 & 5600 & 7.71463 & 0.065 & 5.352 & -0.001 + + 6.80 & 4.092 & 5280 & 30.9661 & 0.207 & 19.51 & -0.091 + 5.20 & 3.701 & 5340 & 15.9744 & 0.094 & 10.63 & -0.050 + 4.40 & 3.457 & 5460 & 10.0562 & 0.056 & 6.898 & -0.028 + 4.20 & 3.389 & 5550 & 8.51317 & 0.054 & 5.906 & -0.007 + 3.80 & 3.243 & 5630 & 6.46303 & 0.039 & 4.532 & -0.002 + 1. ml relation given in @xcite : @xmath130 2. ml relation given in @xcite : @xmath131 3. ml relation interpolated between two @xcite relations at @xmath132 and @xmath123 to yield a relation at @xmath133 : @xmath134 the units for both @xmath113 and @xmath114 are in solar units. note that these ml relations cover reasonably broad @xmath10 ratios given in the literature. the input parameters for the lmc models with these ml relations and the periods calculated from linear non - adiabatic analysis are given in table [tabinput]. after the full amplitude models are constructed from the pulsation codes, the temperature and the opacity profile can be plotted in terms of the internal mass distribution (@xmath135 $], where @xmath136 is mass within radius @xmath137 and @xmath113 is the total mass) at a given phase of pulsation. as in paper ii, the locations of the hif (sharp rise in the temperature profile) and photosphere (at optical depth @xmath138) can be identified in the temperature profile. to quantify the hif - photosphere interaction (if the photosphere is next to the base of the hif or not, see also paper ii), we calculate the `` distance '', @xmath139, in @xmath140 between the hif and the photosphere from the temperature profile. the definition of @xmath139 can be found in paper ii. a small @xmath139 means there is a hif - photosphere interaction, and vice versa. the theoretical quantities from the models can be compared to the observed quantities using the following prescriptions : ccccc @xmath6 & @xmath141 & @xmath5 & @xmath142 & @xmath143 + + 46.4155 & 27481.28 & 5445.00 & 18329.74 & 4826.00 + 35.6727 & 19742.83 & 5434.01 & 14321.20 & 4962.65 + 28.2406 & 16894.15 & 5502.35 & 12093.99 & 4978.72 + 25.3960 & 14460.62 & 5562.28 & 10570.70 & 5010.85 + 22.3804 & 12772.75 & 5615.34 & 9283.257 & 5069.43 + 19.3886 & 11263.84 & 5687.98 & 8143.460 & 5145.20 + 17.7750 & 9034.392 & 5561.39 & 7044.743 & 5147.04 + 12.6085 & 5637.467 & 5528.15 & 4800.840 & 5278.11 + 10.2940 & 4371.170 & 5537.42 & 3839.001 & 5377.80 + 9.81474 & 4004.569 & 5518.39 & 3580.396 & 5383.61 + 8.37226 & 3217.688 & 5596.34 & 2929.354 & 5437.17 + 8.12498 & 3032.639 & 5581.53 & 2793.441 & 5438.96 + 7.69466 & 2864.745 & 5602.37 & 2640.439 & 5467.54 + 7.23505 & 2705.430 & 5639.03 & 2490.271 & 5504.98 + 6.01283 & 2105.589 & 5669.57 & 1986.586 & 5567.91 + + 40.2561 & 23500.98 & 5821.99 & 11853.38 & 4947.11 + 35.4374 & 19604.60 & 5830.18 & 9898.124 & 4962.26 + 28.2629 & 14661.29 & 5875.98 & 7659.896 & 5057.96 + 25.6378 & 12836.04 & 5884.01 & 6916.299 & 5108.02 + 19.4314 & 9538.226 & 5985.21 & 5555.179 & 5303.60 + 17.5523 & 8035.700 & 5921.07 & 4947.967 & 5142.58 + 14.3143 & 5490.345 & 5783.01 & 3639.964 & 5164.11 + 11.3659 & 3850.971 & 5814.49 & 2722.473 & 5242.47 + 10.0011 & 3286.030 & 5817.00 & 2377.950 & 5300.52 + 9.77393 & 3136.139 & 5801.77 & 2301.751 & 5296.73 + 8.94637 & 2738.003 & 5798.30 & 2071.106 & 5331.19 + 8.31297 & 2499.416 & 5794.49 & 1917.696 & 5365.34 + 8.10751 & 2374.713 & 5780.35 & 1850.048 & 5369.40 + 8.09994 & 2235.292 & 5723.18 & 1796.586 & 5355.41 + 7.71463 & 2271.393 & 5805.49 & 1772.270 & 5408.34 + + 30.9661 & 14958.03 & 5611.77 & 8486.049 & 4963.81 + 15.9744 & 5697.042 & 5665.97 & 4209.487 & 5079.81 + 10.0562 & 3197.451 & 5670.64 & 2560.871 & 5283.85 + 8.51317 & 2723.566 & 5716.48 & 2201.628 & 5382.21 + 6.46303 & 1852.537 & 5684.63 & 1593.714 & 5507.38 + = 7.5 cm 1. as in paper ii, we use the basel atmosphere database @xcite to construct a fit giving temperature and effective gravity as a function of @xmath144 colour. the effective gravity is obtained at the appropriate phase from the models (see paper ii). these prescriptions are used to convert the temperatures to the @xmath144 colours. the bolometric corrections (@xmath145) are obtained in a similar manner. the anonymous referee has suggested that @xmath144 may not be a good way to convert between temperature and colour unless both of the micro - turbulence and surface gravity are included. as indicated above this is the case, and in any case our results and those of paper ii for galactic models, show good agreement between theory and observations. a number of previous authors have used this method and some authors commented that this colour can be used as an indicator of temperature (e.g. *??? * ; *??? the empirical relations we studied in this series were also mainly in the @xmath144 colour. 2. in addition to the basel atmosphere, we also use the atmosphere fit from @xcite, referring this as the sbt atmosphere in our paper. the sbt atmosphere does include both of the effective gravity and the micro - turbulence in their table 6 for the temperature and colour conversion. these conversions are tabulated for two micro - turbulence velocities of @xmath146 and @xmath147, as well as for various metallicities. to apply these conversions to our lmc models, we first interpolated the conversions between @xmath148=0.0 $] and @xmath148=-0.5 $] to @xmath148=-0.3 $], which is appropriate for the lmc metallicity. the @xmath144 colours at the maximum, mean and minimum light are then obtained from the given effective temperature and the effective gravity for both of the micro - turbulence velocities. we use the prescriptions given in @xcite to convert the observed colours to the temperatures appropriate for the lmc data as follows : @xmath149 + note that these functions are also obtained from the basel atmosphere database. we can compare the colours obtained from the basel and sbt atmosphere for our models constructed in this paper. the results are presented in figure [figatmos]. from this figure it can be seen that the colours obtained from both of the atmosphere fits agree within @xmath150mag. the difference is even smaller if the micro - turbulence velocity of @xmath147 is used. this indicates that the @xmath144 colours can be used to indicate the temperature. since the results of our models are qualitatively compared to the observations (see next section) and not used to quantitatively derive any theoretical pc and/or ac relations, an accuracy of @xmath151mag., independent of period, from the atmosphere fit is acceptable and would not cause problems for our results. note that the sbt atmosphere are only defined for @xmath152 and @xmath153, few of our models either the @xmath115 or @xmath154 or both are beyond these ranges at certain phases, hence no colours can be obtained from the sbt atmosphere (for example some points are missing at minimum light for few of the long period models, as shown in figure [figatmos]). due to these reasons, we continue adopt the basel atmosphere fits to convert the temperatures and @xmath144 colours, after taking account of the effective gravity in the fits, as a function of phase.
Results from the models
cccccccc @xmath6 & @xmath155 & @xmath156 & @xmath157 & @xmath158 & @xmath159 & @xmath160 (asc) & @xmath160 (des) + + 46.4155 & 24249.2 & 24423.695 & 5330.06 & 24282.129 & 4882.55 & 5319.42 & 4880.66 + 35.6727 & 17207.7 & 17078.471 & 5293.90 & 17188.051 & 4922.99 & 5305.14 & 4924.47 + 28.2406 & 14484.7 & 14466.869 & 5457.79 & 14529.348 & 5073.65 & 5459.55 & 5069.64 + 25.3960 & 12540.8 & 12461.087 & 5443.66 & 12544.735 & 5096.58 & 5452.77 & 5096.18 + 22.3804 & 10997.1 & 10996.587 & 5493.44 & 11039.123 & 5157.95 & 5493.51 & 5152.87 + 19.3886 & 9587.02 & 9560.1922 & 5545.87 & 9618.0583 & 5232.85 & 5549.58 & 5228.51 + 17.7750 & 7983.35 & 7988.5457 & 5473.94 & 7994.9159 & 5213.83 & 5473.06 & 5211.77 + 12.6085 & 5208.60 & 5208.1564 & 5493.34 & 5202.6440 & 5349.00 & 5493.36 & 5350.84 + 10.2940 & 4170.31 & 4162.1135 & 5560.81 & 4174.0993 & 5450.34 & 5564.43 & 5449.06 + 9.81474 & 3860.04 & 3861.4313 & 5556.01 & 3852.8715 & 5446.51 & 5555.34 & 5449.11 + 8.37226 & 3109.66 & 3106.7337 & 5570.85 & 3110.1806 & 5481.59 & 5572.55 & 5481.37 + 8.12498 & 2939.41 & 2935.0599 & 5553.67 & 2939.6504 & 5476.34 & 5556.38 & 5476.23 + 7.69466 & 2775.73 & 2775.1245 & 5581.66 & 2777.4057 & 5501.67 & 5582.05 & 5500.77 + 7.23505 & 2618.53 & 2620.1965 & 5620.29 & 2618.8674 & 5534.39 & 5619.15 & 5534.20 + 6.01283 & 2051.97 & 2050.8394 & 5646.29 & 2051.3277 & 5583.97 & 5647.27 & 5584.50 + + 40.2561 & 18754.3 & 18580.019 & 5704.60 & 18824.542 & 5144.86 & 5718.84 & 5141.42 + 35.4374 & 15896.6 & 15885.248 & 5749.07 & 15861.179 & 5147.12 & 5750.16 & 5158.63 + 28.2627 & 11157.8 & 11188.594 & 5692.54 & 11143.155 & 5111.26 & 5688.21 & 5112.18 + 25.6378 & 10388.8 & 10366.128 & 5782.33 & 10358.696 & 5184.81 & 5785.59 & 5188.79 + 19.4314 & 7743.89 & 7741.3087 & 5871.14 & 7750.7288 & 5332.20 & 5871.60 & 5330.76 + 17.5523 & 6538.19 & 6537.3819 & 5833.02 & 6532.0823 & 5310.84 & 5833.20 & 5312.28 + 14.3143 & 4564.83 & 4621.4163 & 5750.64 & 4565.7876 & 5280.59 & 5732.80 & 5280.34 + 11.3659 & 3289.17 & 3288.4584 & 5711.02 & 3281.3799 & 5328.36 & 5711.31 & 5331.32 + 10.0011 & 2811.35 & 2823.6619 & 5726.94 & 2800.7972 & 5376.66 & 5721.20 & 5381.50 + 9.77393 & 2702.54 & 2712.0209 & 5715.81 & 2692.4263 & 5378.63 & 5711.29 & 5383.51 + 8.94637 & 2385.34 & 2395.4576 & 5704.28 & 2388.3903 & 5412.32 & 5699.14 & 5410.62 + 8.31297 & 2189.29 & 2193.3351 & 5708.03 & 2188.7842 & 5448.47 & 5705.90 & 5448.80 + 8.10751 & 2094.38 & 2103.6034 & 5698.27 & 2093.0120 & 5449.90 & 5693.36 & 5450.83 + 8.09994 & 2004.90 & 1998.5077 & 5648.26 & 1999.8838 & 5419.02 & 5651.76 & 5422.70 + 7.71463 & 2004.07 & 2009.2157 & 5713.42 & 2004.9007 & 5490.38 & 5710.72 & 5489.80 + + 30.9661 & 12359.6 & 12202.760 & 5566.66 & 12354.904 & 5029.65 & 5585.98 & 5030.07 + 15.9744 & 5011.60 & 5031.4017 & 5550.08 & 5024.6207 & 5173.09 & 5544.24 & 5169.84 + 10.0562 & 2860.25 & 2866.2012 & 5593.28 & 2859.5002 & 5346.96 & 5590.81 & 5347.35 + 8.51317 & 2447.87 & 2441.2040 & 5643.58 & 2453.1647 & 5463.35 & 5646.11 & 5460.12 + 6.46303 & 1749.72 & 1749.6046 & 5719.11 & 1744.6691 & 5577.03 & 5719.24 & 5581.12 + the effective temperatures for the full amplitude models in table [tabinput] at the corresponding maximum and minimum light (or luminosity) are given in table [c9tabmaxmin]. for the effective temperatures at mean light, the temperatures for the mean light at ascending and descending branch of the light (or luminosity) curve are not the same (e.g., in paper ii), hence table [c9tabmean] gives the effective temperature at these phases for our lmc models. the layout of table [c9tabmean] is the same as table 3 from paper ii. following paper ii, the locations of the photosphere can be identified in the temperature and opacity profiles. these are displayed in figure [c9bono4]-[c9chiosi13] with a @xmath161, a @xmath162 and a @xmath23 model, respectively. the left and right panels of figure [c9bono4]-[c9chiosi13] are the temperature and opacity profiles respectively. the photospheres are marked as filled circles in these figures. finally, the plots of the @xmath139, the `` distance '' between the photosphere and the hif from the temperature profiles, as a function of pulsating period for the lmc models are presented in figure [c9deltalmc] with the three ml relations used. in paper ii, it is found that the distribution of @xmath139 as a function of period is almost independent of the adopted ml relation. this is also seen in the lmc models as depicted in figure [c9deltalmc]. figure [c9bono4]-[c9chiosi13] and figure [c9deltalmc] bear witness to the fact that at maximum light, the photosphere lies at the base of the hif for all of the models. although there is a slight deviation for some longer period models, the location of the photosphere is close to the hif within the error bars (which are defined as the coarseness of the grid points around the location of the hif). as in paper ii for the galactic models, the closeness of the photosphere to the base of the hif, for reasonably low densities, results in a flat or almost flat pc relation for the long period lmc cepheids. in the case of minimum light, even though figure [c9deltalmc] implies that @xmath163 is nearly constant across the period range and the photosphere is near the base of the hif, as in the case of maximum light, @xmath163 does follow a shallow correlation with period after 10 days. judging from the error bars of @xmath163 and from figure [c9bono4]-[c9chiosi13], there is tentative evidence that the photosphere is disengaged from the hif for @xmath161 at minimum light. hence the temperatures or the colours at minimum light are more dependent on period for @xmath164-@xmath27 relation may not be correlated with the slopes of the pc relation] and the global properties. theoretical quantities that can be computed from the models and compared with data include the pulsation periods, the @xmath7-band amplitudes and the fourier parameters, the temperatures and colours at the maximum, mean and minimum light. these are the pc plots, the ac plots, the period - temperature plots and the fourier parameters plots portrayed in figures [c9modelpc]-[c9lmcfourier]. the temperatures in table [c9tabmaxmin] & [c9tabmean], after conversion to the @xmath144 colours as mentioned in previous section, are superimposed along with the observed lmc pc relations as plotted in figure [c9modelpc]. similarly, figure [c9modelpt] graphs the same quantities but on the @xmath165-@xmath27 plane with the observed @xmath144 colours converted to temperatures using the prescriptions given in section 5. the theoretical bolometric light curves are converted to the @xmath7-band light curves with the bolometric corrections obtained from the basel database mentioned previously. from the theoretical @xmath7-band light curves, the amplitudes can be estimated and these are displayed in figure [c9modelac] along with the colours from models to compare with the empirical ac relations. the fourier parameters of the theoretical @xmath7-band light curves can also be obtained with (@xmath166) fourier expansion. these fourier parameters are compared with the observational data in figure [c9lmcfourier]. several features are noticed from figure [c9modelpc]-[c9lmcfourier] : 1. the general trends of the models qualitatively match the observational data. there are greater discrepancies between the data and short period models, particularly in matching the observed light curve amplitudes. 2. the models with the ml relation from @xcite, with lower @xmath10 ratio, do better in matching the observations. these models also tend to lie near the envelopes of the pc, ac, @xmath165-@xmath27 and @xmath167-@xmath27 relations defined by the observational data. 3. the slopes of the period - colour (or period - temperature) relations at maximum and minimum light from the models roughly match the observational data, i.e., the theoretical pc(max) relation is approximately flat and there is a relation at minimum light. 4. the temperatures from the models with the @xcite ml relation is cooler (hence redder) than the models with the @xcite ml relation and the observed data at maximum light. in contrast, the temperatures (or the colours) at minimum light from the models with these two ml relations are consistent with each other and are located near the blue edge of the observed data. the means at the descending branches are in better agreement with the observed data than the means at the ascending branches. this is because the observed means, @xmath168, are obtained mostly from the descending branches. though previous researchers have noted that temperatures on the ascending and descending branches are not the same at mean light (as cepheids exhibit loops in cmd), what is new here is the way the nature of the hif changes during the pulsation. the behaviors of the models from the interpolated @xcite ml are closer to the models from @xcite ml relation because their slopes are very similar. the amplitudes of the theoretical light curves (in both of the bolometric and @xmath7-band light curves) are smaller than the observations at given period, especially for the models with the @xcite ml relation. these can be seen from the ac relations as given in figure [c9modelac] and the left panels of figure [c9lmcfourier]. overall, some agreements and disagreements are found between the theoretical quantities and the observational data. it is also found out that there are some problems associated with the pulsation codes when the lmc models are constructed : these include the smaller amplitude of the model light curves and the cooler temperatures at the maximum light (especially with @xcite ml relation). note that from equation (1), cooler temperatures at maximum light imply that the amplitudes will be lower at given period. varying other parameters in the pulsation codes, including the @xmath118 parameters, does not improve the situation, though perhaps a more detailed and systematic study of the dependence of lmc cepheid pulsation models on the @xmath169 parameters could resolve this situation. however, we believe that the qualitative nature of the photosphere - hif interactions as given in figure [c9deltalmc] will still hold even in models which fare better in mimicking observed amplitudes. this is in part because figure [c9deltalmc] suggest that the behaviors of @xmath139 as a function of period are nearly, though not completely, independent of amplitudes, as the models with @xcite ml relation have higher amplitudes (although still smaller than the observations) than the models with the @xcite ml relation. however, better codes that fix these problems or the 3-d convection codes are needed in the future studies. the temperature profiles from the galactic models given in paper ii and the lmc models are compared in figure [c9pt] at maximum and minimum light. the upper panels of figure [c9pt] suggest that at maximum light, the photosphere is not far from the base of the hif in both of the galactic and the lmc models. in contrast, the photosphere is further away from the hif in the galactic models than the lmc models at minimum light. the hif is located further out in the mass distribution for the galactic models. the plots of the @xmath139-@xmath27 relation from the galactic and lmc models at maximum and minimum light are also compared in figure [c9delta]. it can be seen from the figure that at maximum light, the behavior of both galactic and lmc models is similar, where the photosphere is near the base of the hif. at minimum light, the long period models show that the photosphere is disengaged from the hif, while the behavior of the short period models is different between the galactic and lmc models. the photosphere of the short period lmc models seems to be located closer to the hif at minimum light, but it is not the case for the short period galactic models. this could lead to shallower slopes of the pc(min) relation seen in the lmc cepheids as compared to the galactic counterparts. in terms of the hif - photosphere interaction, there is some tentative evidence from the models that the lmc long period cepheids behave like the galactic cepheids, while the short period lmc cepheids behave like the rr lyrae stars at minimum light. figure [c9density] graphs the density (defined as @xmath170, where @xmath7 is the specific volume) at the photosphere as a function of the period of the model at minimum, maximum and ascending and descending mean light. galactic models generally tend to have the lowest density and, in particular, have significantly lower densities at minimum light than the lmc models. we note that the galactic models always have a photospheric density lower than about @xmath171 whereas the photospheric density for the lmc models only falls below this figure after a period of 10 days. at maximum light, all long period models have a low photospheric density. what we get from this figure is that it provides some evidence that there is a difference in photospheric density between the lmc and galactic models. moreover, this difference appears to be consistent with what is required by our theoretical scenario : short period lmc models have a higher photospheric density than their galactic counterparts. however, for a discussion of some caveats, see section 8.
Applications
we now discuss two important applications of the photosphere - hif interaction : reddening corrections and the explanation of the observed non - linear lmc pl (and pc) relations. @xcite original interest in the spectral properties of cepheids at maximum light was to estimate reddening. skm used this to correct a number of reddening for galactic cepheids. @xcite used equation (1) and the theoretical explanation provided in skm to derive a relation linking the colour excess to the colour at maximum light, the @xmath7-band amplitude and the period. such a relation is predicted from equation (1). @xcite estimates the error with this method to be comparable to other multi - colour methods. a more interesting application of the hif - photosphere interaction is to explain the recent detected non - linear lmc pl relation as presented in @xcite, paper i, @xcite and @xcite. paper i used the @xmath2-test to provide strong statistical evidence that the optical cepheid pl relation at mean light in the lmc is non - linear around a period close to 10 days. @xcite used the macho and 2mass datasets together with additional long period cepheids from the literature to further support the existence of non - linear lmc pl relation in the optical and near infra - red wave - bands. in contrast, current data indicate that the galactic pl relation is linear at mean light @xcite. non - linearity of the lmc pl relations can be tested using the @xmath2-test with the data given in section 3. the empirical results of the fitted lmc pl relations at maximum, mean and minimum light using the updated data are presented in table [c9tabpl]. the plots of the pl relations at maximum / minimum light and at mean light are shown in figure [c9plmaxmin] & [c9plmean], respectively. the @xmath2-test results for these pl relations are : @xmath172, and @xmath173. the large @xmath2-values for both @xmath7- and @xmath11-band pl relations at mean and minimum light strongly indicate that the pl relations at these two phases are not linear, and the data is better described with the broken (i.e, two regressions) pl relation. however, the small @xmath2-values at maximum light, with corresponding @xmath174-values of @xmath175 and @xmath176 for the @xmath7- and @xmath11-band pl(max) relations respectively, show that the null hypothesis of the @xmath2-test can not be rejected (a value of @xmath177 and/or @xmath178 is required for doing this). hence there is no observed break seen in the pl(max) relation and the data is consistent with single line regression. note that the same slopes of the pl(max) relations for long period cepheids in both bands are consistent of the finding that the pc(max) relation is flat for these cepheids. lcccccc phase & @xmath179 & @xmath180 & @xmath181 & @xmath182 & @xmath183 & @xmath184 + + maximum & @xmath185 & @xmath186 & 0.260 & @xmath187 & @xmath188 & 0.170 + mean & @xmath189 & @xmath190 & 0.208 & @xmath191 & @xmath192 & 0.141 + minimum & @xmath193 & @xmath194 & 0.204 & @xmath195 & @xmath196 & 0.140 + + maximum & @xmath197 & @xmath198 & 0.257 & @xmath199 & @xmath200 & 0.174 + mean & @xmath201 & @xmath202 & 0.228 & @xmath203 & @xmath204 & 0.158 + minimum & @xmath205 & @xmath206 & 0.243 & @xmath207 & @xmath208 & 0.174 + + maximum & @xmath209 & @xmath210 & 0.260 & @xmath211 & @xmath212 & 0.169 + mean & @xmath213 & @xmath214 & 0.203 & @xmath215 & @xmath216 & 0.138 + minimum & @xmath217 & @xmath218 & 0.194 & @xmath219 & @xmath220 & 0.131 + our tentative theoretical explanation for the non - linear nature of the lmc pl relations across a period of 10 days replies on the hif - photosphere interaction. @xcite and @xcite have established the connection between the pc and pl relations : both these relations arise from the more general plc relation. these relations refer to quantities evaluated at mean light. the existence of such a connection relies on the period - mean density theorem, the instability strip and the stefan - boltzmann law. if we assume the stefan - boltzmann law can be applied at every phase, then it is straightforward to show that a plc relation (though possibly with different coefficients) exists at every phase point. thus the standard plc relation and indeed the pc and pl relation expresses at mean light are just the averages of the same relations at different phases points. consequently one way to understand the behavior of plc / pl / pc relations at mean light is to understand their behavior at different phase points. what we try to do in this paper is point out some evidence from our models that shows how the changing behavior of the pc relations at different phases can, in principle, arise from a consideration of the photosphere - hif interaction at these phases. since the mean light pc and pl relation are the average of those at all phases, these properties can affect the pc and, as a consequence, the pl relation (via the plc relation). in fact, the new data with superb phase resolution from such micro - lensing projects such as ogle and macho demands a multiphase analysis. this approach can potentially lead to a deeper understanding of the pulsation and evolution of cepheid variables. for example, @xcite looked at pc relations in the galaxy and lmc as a function of phase. they found that short and long period lmc cepheids have a shallower and steeper slope at most pulsation phases than galactic cepheids respectively.
Conclusion and discussion
in this paper, we have confronted updated pc and ac relations at maximum, mean and minimum light for lmc cepheids observed by the ogle team, and additional cepheids from the literature, with theoretical, full amplitude pulsation models of lmc cepheids. the observed pc and ac relations provide compelling evidence of a non - linearity or break at a period of 10 days. we also constructed theoretical cepheid pulsation models appropriate for the lmc using the florida pulsation codes @xcite to study the hif - photosphere interaction. the empirical results presented in this paper, as well as in other papers such as @xcite and @xcite, provide strong empirical evidence that the pc and pl relations for the lmc cepheids are non - linear, in the sense described in previous sections. issues such as extinction and a lack of long period cepheids that may cause the non - linear lmc pl and pc relations have been addressed and argued against in paper i, @xcite and @xcite, and will not be repeated here. other arguments against the non - linear lmc pl relation include the results presented in @xcite, as the authors found no evidence for a non - linear pl relation in the lmc at @xmath221-bands. however, @xcite treated the data of @xcite extensively and found, in a statistically rigorous way, that the reason why @xcite found linear @xmath221 pl relations, is due to the small number of short period cepheids (@xmath222) in their sample. @xcite also reduce the number of ogle / macho lmc cepheids and show how the @xmath2-test can produce a non - significant result when the number of short / long period cepheids become small. instead, using the 2mass data that are cross - correlated with macho cepheids, @xcite have found that the lmc @xmath223-band pl relations are non - linear - band than in @xmath224-band, as shown in @xcite.] and the @xmath225-band pl relation starts to become linear. @xcite also discussed why this is the case. another argument against the non - linear pl relation is that the pl relation should be universal, as found in @xcite. we argue that their results are based on a handful of cepheids (@xmath226) and on short periods cepheids in a cluster whose membership to the lmc is in question. their shallower galactic pl relation based on the revised infra - red surface brightness method also contradicts the steeper galactic pl relation based on independent methods from open cluster main - sequence fitting @xcite. it is worthwhile to point out that our sample selection does not affect the detection on non - linear lmc pl relation at mean light. since the mean magnitudes of a cepheid light curve is less affected by our constrains on selecting the cepheids with good light curves, we can use the published (reddening corrected) mean magnitudes to test the non - linear lmc pl relation. the anonymous referee kindly provided a large sample of lmc cepheids that combined the published mean @xmath7-band magnitudes from the ogle, @xcite and @xcite datasets. there are a total of 115 long period cepheids in this sample and the @xmath2-test still return a significant detection of the non - linear lmc pl relation. the ogle+@xcite combined data also give very similar results. similar tests have also been done in @xcite by using the macho data alone and the macho+@xcite combined data. the non - linear lmc pl relation is still present from the @xmath2-test results on these two datasets. therefore we believe our sample selection does not affect the detection of the non - linear lmc pl relation. the detection of non - linear lmc pl relation from totally independent ogle and macho data, using totally independent reddening estimates, suggested that this non - linearity is real and our paper is the first attempt to theoretically explain this non - linearity in terms of the hif - photosphere interaction. due to small number of lmc models, it is impossible to derive the theoretical pc and ac relations with a small error on the slope and compare directly to the empirical relations. however, these lmc models can be qualitatively compared to the observations by converting some physical quantities to the observable quantities and vice versa, such as the temperature - colour conversion. hence we compared our model light curves to the observations in terms of theoretical pc and ac relations at the phases of maximum, mean and minimum light and also in terms of the fourier parameters from theoretical light curves with observations. the theoretical quantities from the models generally agree with the observations, but it was found out that these models tend to have smaller amplitudes and (hence) the temperature is cooler at maximum light than the real cepheids. though our models have some drawbacks in this comparison, our main interest is in comparing the interaction of the photosphere and hif as a function of phase with similar results presented in paper ii for galactic cepheid pulsation models. the aim is _ not _ to compare our models rigorously with observations but rather to study models which match observations reasonably well in the context of the theoretical framework described in previous sections and in paper i & ii. nevertheless we argued that the qualitative nature of the photosphere - hif interaction is not seriously affected by these problems. our postulate is that at certain phases, this interaction can affect the pc relation due to the properties of the saha ionization equation : specifically for reasonably low densities in cepheid envelopes, hydrogen ionizes at a temperature that is almost independent of period. consequently, when the photosphere is located at the base of the hif, the photospheric temperature and hence the colour is almost independent of period. however, when this engagement occurs, but the density is greater, then the temperature at which hydrogen ionizes again becomes sensitive to global surroundings and hence on period. when the photosphere is not engaged with the hif in this way, its temperature is again dependent on period and global stellar parameters. for galactic cepheids, this hif - photosphere interaction occurs mainly at maximum light for cepheids with @xmath227 (paper ii). at minimum light, there is a strong correlation between the hif - photosphere distance and period leading to a definite ac relation at minimum light for galactic cepheids (skm, paper i & ii). in this paper, we have found tentative evidence that, for short period lmc models which match observations in the period - color plane, the hif - photosphere interaction occurs at most phases but at densities which are too high to produce a flat pc relation. why would these short period lmc cepheids be different in this regard to short period galactic cepheids? one possibility could be that this is partly because these lmc cepheids are hotter than their galactic counterparts @xcite. the hif - photosphere are disengaged for most of the pulsation cycle for long period lmc cepheids. this happens because as the period increases, so does the @xmath10 ratio which pushes the hif further inside the mass distribution. when the hif - photosphere are disengaged in this way, the photospheric temperature is more dependent on density and hence on period. the change is sudden because the hif - photosphere are either engaged or they are not. this can lead to a sudden change in the pc relation at 10 days as shown by the observations @xcite. however, at maximum light the hif - photosphere are engaged at low densities for long period lmc cepheids leading to the observed flat pc relation for these stars. taken together with equation (1), this theoretical scenario is consistent with the observed pc - ac behavior described in paper i and in this study. the anonymous referee has noted that these suggestions about photospheric density can be tested by spectroscopic means. we now enumerate some caveats to our argument that could be addressed in future papers. 1. since the smc pc relation at mean light is linear (e.g., paper i), how do smc (i.e., metal - poor) models fit into the theoretical scenario outlined in this paper and paper ii, if at all? this is a difficult question and its full answer is beyond the scope of this paper however, as the metallicity decreases, we do note that the smc has a different ml relation to the lmc and galaxy and so does the temperatures associated with the instability strip. these will change the relative location of the hif and photosphere @xcite and possibly alter the phase at which they interact. further the amplitudes for smc cepheids are smaller due to the lower metallicity @xcite. this will also affect the hif - photosphere interaction. one difference which can be consistent with this is the fact that the pc relation at maximum light in the smc is not flat (see paper i) but it is the case for the galaxy and lmc pc relations. this indicates that at maximum light, there is less interaction between the hif and photosphere at low densities. this leads to an observed linear pc relation at mean light for the smc cepheids. these will be investigated further in a future paper in this series. 2. could the well - known hertzsprung progression play any part in causing the observed changes in the galactic and lmc pc relations? it may also be that higher order overtones becoming unstable or stable, though with the fundamental mode still being dominant, may also have an impact on the pc relation in some as yet unknown way (paper ii). 4. the behavior of short period lmc cepheids still needs to be understood, for example, what causes the difference between the bottom left panels of figures [c9deltalmc] and [c9delta]? that is, why is it that for short period galactic / lmc cepheids, the hif - photosphere are disengaged / engaged? our experience suggests that constructing short period full amplitude fundamental mode cepheids requires more care than the long period case because the first overtone has a non - negligible growth rate. because of this we feel a thorough study of these short period cepheids merits a separate paper. 5. would more advanced pulsation codes which, for example, can match the observed amplitudes and which contain a more accurate model of time dependent turbulent convection, yield similar results, especially for figure [c9delta]? could such codes fare better in modeling short period lmc cepheids?
Acknowledgments
smk acknowledges support from hst - ar-10673.04-a. we thank an anonymous referee for several useful suggestions and providing the data for our testing. we would also like to thank e. antonello, r. buchler & j. kwan for useful discussions, and r. bell & m. marengo for the discussion regarding the atmosphere fits. | period - colour (pc) and amplitude - colour (ac) relations are studied for the large magellanic cloud (lmc) cepheids under the theoretical framework of the hydrogen ionization front (hif) - photosphere interaction.
lmc models are constructed with pulsation codes that include turbulent convection, and the properties of these models are studied at maximum, mean and minimum light. as with galactic models, at maximum light the photosphere is located next to the hif for the lmc models. however very different behavior is found at minimum light. the long period (@xmath0days)
lmc models imply that the photosphere is disengaged from the hif at minimum light, similar to the galactic models, but there are some indications that the photosphere is located near the hif for the short period (@xmath1 days) lmc models.
we also use the updated lmc data to derive empirical pc and ac relations at these phases.
our numerical models are broadly consistent with our theory and the observed data, though we discuss some caveats in the paper.
we apply the idea of the hif - photosphere interaction to explain recent suggestions that the lmc period - luminosity (pl) and pc relations are non - linear with a break at a period close to 10 days.
our empirical lmc pc and pl relations are also found to be non - linear with the @xmath2-test.
our explanation relies on the properties of the saha ionization equation, the hif - photosphere interaction and the way this interaction changes with the phase of pulsation and metallicity to produce the observed changes in the lmc pc and pl relations.
cepheids stars : fundamental parameters | astro-ph0601446 |
Introduction
of the 77 extrasolar planets currently listed by the iau working group on extrasolar planets] (including planet candidates published in a refereed journals with @xmath0@xmath2 10 m@xmath1), only three systems have been found to harbor planets in circular orbits (e @xmath2 0.1) orbits beyond 0.5 au 47 uma (fischer et al. 2002 ; butler & marcy 1996), hd 27442 (butler et al. 2001), and hd 4208 (vogt et al. 2002). with 13 `` 51 peg type '' planets (p @xmath2 5 d), and @xmath360 eccentric planets (e @xmath4 0.1), the long period circular orbits are the rarest of the three types of planetary systems to emerge over the last 8 years. with one exception, all the iau working group list planets orbit within 4 au of their host stars. as all these planets have been discovered via the precision doppler technique, there is a strong selection bias toward discovering systems with small semimajor axes. unsurprisingly, the only extrasolar planet so far found to orbit beyond 4 au was detected by the precision doppler survey that has been gathering data the longest (marcy et al. 2002). perhaps the most critical question facing the field of extrasolar planetary science is `` are solar system analogs (ie. systems with giants planets in circular orbits beyond 4 au and small rocky planets orbiting in the inner few au) ubiquitous, or rare? '' existing precision doppler surveys will become sensitive to giant planets orbiting beyond 4 au by the end of this decade, though only those programs with long term precision of 3 or better will be able to determine if the orbits of such planets are eccentric or circular (butler et al. 2001, figure 11). we report here a new extrasolar planet in an approximately circular orbit beyond 3 au, discovered with the 3.9 m anglo australian telescope (aat). the anglo - australian planet search program is described in section 2. the characteristics of the host star and the precision doppler measurements are presented in section 3. a discussion follows.
The angloaustralian planet search
the anglo - australian planet search began in 1998 january, and is currently surveying 250 stars. fourteen planet candidates with @xmath0ranging from 0.2 to 10 m@xmath1 have first been published with aat data (tinney et al. 2001 ; butler et al. 2001 ; tinney et al. 2002a ; jones et al. 2002a ; butler et al. 2002 ; jones et al. 2002b ; tinney et al. 2003a ; jones et al. 2003), and an additional four planet candidates have been confirmed with aat data (butler et al. 2001). precision doppler measurements are made with the university college london echelle spectrograph (ucles) (diego et al. 1990). an iodine absorption cell (marcy & butler 1992) provides wavelength calibration from 5000 to 6000 . the spectrograph psf and wavelength calibration are derived from the embedded iodine lines (valenti et al. 1995 ; butler et al. this system has demonstrated long term precision of 3 (butler et al. 2001), similar to (if not exceeding) the iodine systems on the lick 3-m (butler et al. 1996 ; 1997) and the keck 10-m (vogt et al.
Hd 70642
hd 70642 (hip 40952, sao 199126) is a nearby g5 dwarf, at a distance of 28.8 pc (perryman et al. 1997), a @xmath5 magnitude of 7.17, and an absolute magnitude of @xmath6 = 4.87. the star is photometrically stable within hipparcos measurement error (0.01 magnitudes). the star is chromospherically inactive, with log@xmath7(hk) @xmath8 @xmath94.90 @xmath100.06, determined from aat / ucles spectra of the ca ii h&k lines (tinney et al. 2003b ; tinney et al. figure 1 shows the h line compared to the sun. the chromospherically inferred age of hd 70642 is @xmath34 gyr. spectral synthesis (lte) of our aat / ucles spectrum of hd 70642 yields t@xmath11 @xmath85670 @xmath1020 k and @xmath12@xmath8 2.4 @xmath101 consistent with its status as a middle aged g5 dwarf. like most planet bearing stars, hd 70642 is metal rich relative to the sun. we estimate [fe / h] @xmath8 @xmath130.16 @xmath100.02 from spectral synthesis, in excellent agreement with the photometric determination of eggen (1998). while ni tracks fe for most g & k dwarfs, the [ni / h] @xmath8 @xmath130.22 @xmath100.03 appears slightly high for hd 70642. the mass of hd 70642 estimated from @xmath14@xmath5, m@xmath15, and [fe / h] is 1.0 @xmath100.05 m@xmath16 . a total of 21 precision doppler measurements of hd 70642 spanning more than 5 years are listed in table 1 and shown in figure 2. the solid line in figure 2 is the best fit keplerian. the keplerian parameters are listed in table 2. the reduced @xmath17 of the keplerian fit is 1.4. figure 3 is a plot of orbital eccentricity vs. semimajor axis for the planet orbiting hd70642, for extrasolar planets listed by the iau working group on extrasolar planets, and solar system planets out to jupiter. hd 70642b joins 47 uma c (fischer et al. 2002) as the only planets yet found in an approximately circular (e @xmath18 0.1) orbit beyond 3 au.
Discussion
prior to the discovery of extrasolar planets, planetary systems were predicted to be architecturally similar to the solar system (lissauer 1995 ; boss 1995), with giant planets orbiting beyond 4 au in circular orbits, and terrestrial mass planets inhabiting the inner few au. the landscape revealed by the first @xmath380 extrasolar planets is quite different. extrasolar planetary systems have proven to be much more diverse than imagined, as predicted by lissauer (1995), `` the variety of planets and planetary systems in our galaxy must be immense and even more difficult to imagine and predict than was the diversity of the outer planet satellites prior to the voyager mission. '' the discovery here of a jupiter mass planet in a circular orbit highlights the existence, but also the rarity, of giant planets that seem similar to the original theoretical predictions. review of all the known extrasolar planets, both those described in refereed, published journals (@xmath19) and those in the larger list of claimed extrasolar planets (@xmath20) shows that @xmath37% of extrasolar planets orbiting beyond 0.5 au reside in circular orbits (@xmath210.1). further detections of planets beyond 1 au are needed to determine if circular orbits are more common for planets that orbit farther from the host star. over the next decade precision doppler programs will continue to be the primary means of detecting planets orbiting stars within 50 parsecs. by the end of this decade, doppler programs carried out at precisions of 3 or better by our group, and by others (e.g., mayor & santos 2002), will be sensitive to jupiter and saturn mass planets orbiting beyond 4 au. the central looming question is whether these planets will commonly be found in circular orbits, or if the architecture of the solar system is rare. of the greatest anthropocentric interest are planets in intrinsically circular orbits, as opposed to the short period planets in tidally circularized orbits. nasa and esa have made plans for new telescopes to detect terrestrial mass planets. transit missions such as corot, kepler and eddington may be sensitive to terrestrial mass planets orbiting near 1 au, providing valuable information on the incidence of such planets. as terrestrial planets make photometric signatures of 1 part in 10,000, these missions may be subject to interpretive difficulties that already challenge current ground based transit searches for jupiter sized planets. transit missions can not determine orbital eccentricity, and thus can not determine if planets are solar system analogs. these space based transit missions are targeting stars at several hundred parsecs, making follow up by other techniques difficult. ground based interferometric astrometry programs at keck and vlt are projected to begin taking data by the end of this decade. these programs are complementary to existing precision doppler velocity programs in that they are most sensitive to planets in distant orbits. like doppler velocities, astrometry needs to observe one or more complete orbits to make a secure detection and solve for orbital parameters. it is thus likely that the first significant crop of ground based astrometry planets will emerge after 2015. the nasa space interferometry mission (sim) is scheduled to launch in 2009. a key objective of the sim mission is the detection of planets as small as 3 earth masses in 1 au orbits around the nearest stars. sim offers the promise of determining actual masses of terrestrial planets, thereby securing their status unambiguously. simulations based on the sim measurement specifications, along with the proposed target stars, the 5 year mission lifetime, and a planet mass function extrapolated to the earth mass regime, yield predictions that as many as @xmath35 terrestrial planets could be found (ford & tremaine 2003). giant planets orbiting 25 au from the host stars are also detectable with sim, though the orbital parameters will not be not well determined in a 5 year mission. a 10 year sim mission yields significantly better orbital determination for jupiter analogs. overall the detection capabilities of sim are similar to existing precision doppler programs with a precision of 3 (ford & tremaine 2003), thereby providing confirmation of known planets and unambiguous masses. direct imaging missions such as the nasa terrestrial planet finder (tpf) and the esa darwin mission have the primary goal of detecting earth like planets and obtaining low resolution spectra that might reveal biomarkers. such missions will not return dynamical information and hence will not directly reveal the masses of detected planets. current plans call for the launching of such missions around 2015, perhaps optimistically. we expect that continued doppler measurements, as well as future astrometric missions, will contribute significantly to the interpretation of the unresolved companions detected by tpf / darwin. we acknowledge support by nsf grant ast-9988087, nasa grant nag5 - 12182, and travel support from the carnegie institution of washington (to rpb), nasa grant nag5 - 8299 and nsf grant ast95 - 20443 (to gwm), and by sun microsystems. we also wish to acknowledge the support of the nasa astrobiology institute. we thank the australian (atac) and uk (patt) telescope assignment committees for allocations of aat time. we are grateful for the extraordinary support we have received from the aat technical staff e. penny, r. paterson, d. stafford, f. freeman, s. lee, j. pogson, and g. schaffer. we would especially like to express our gratitude to the aao director, brian boyle. the aat planet search program was created and thrived under brian s tenure as director in large measure because of his critical support and encouragement. the australia telescope national facility (atnf) is fortunate to have brian as their new director. butler, r. p. & marcy, g. w. 1997, `` the near term future of extrasolar planet searches '', brown dwarfs and extrasolar planets, held on tenerife, 17 - 21 march 1997, ed. r. rebolo, e.l. martin, & m.r. zapatero osorio, asp conference series, vol. 134, 162 rrr [vel70642] 830.1082 & -25.8 & 4.2 + 1157.2263 & -36.4 & 4.4 + 1213.1051 & -42.6 & 4.3 + 1236.0850 & -37.5 & 5.2 + 1630.0095 & -15.9 & 3.6 + 1717.8810 & -9.4 & 3.9 + 1920.1348 & 15.0 & 4.8 + 1983.9687 & 13.7 & 5.8 + 2009.0210 & 13.2 & 4.6 + 2060.8744 & 27.7 & 3.4 + 2420.9072 & 12.7 & 3.1 + 2424.8981 & 11.4 & 3.1 + 2455.8416 & 20.0 & 2.8 + 2592.2229 & 12.7 & 2.9 + 2595.2255 & 15.2 & 3.4 + 2710.0700 & 0.8 & 3.0 + 2744.9571 & 0.3 & 3.1 + 2747.9155 & -4.2 & 2.7 + 2749.9755 & -7.4 & 2.2 + 2751.9384 & -5.6 & 2.4 + 2785.9082 & -3.4 & 2.4 + lcc [orbit] orbital period @xmath22 (days) & 2231 & 400 + velocity semiamplitude @xmath23 (ms@xmath24) & 32 & 5 + eccentricity @xmath25 & 0.10 & 0.06 + periastron date (julian date) & 2451749 & 300 + @xmath26 (degrees) & 277 & 75 + m@xmath27 (m@xmath1) & 2.0 + semimajor axis (au) & 3.3 + n@xmath28 & 21 + rms (ms@xmath24) & 4.0 + | precision doppler velocity measurements from the 3.9m anglo - australian telescope reveal a planet with a 6 year period orbiting the g5 dwarf hd 70642.
the a = 3.3 au orbit has a low eccentricity (e = 0.1), and the minimum (@xmath0) mass of the planet is 2.0 m@xmath1.
the host star is metal rich relative to the sun, similar to most stars with known planets.
the distant and approximately circular orbit of this planet makes it a member of a rare group to emerge from precision doppler surveys. | astro-ph0307066 |
Introduction
peculiar velocity surveys covering a fair fraction of the sky are now reaching to 6000 and beyond (@xcite, @xcite, @xcite, @xcite, @xcite, @xcite) and are being interpreted as evidence for substantial flows on these scales (@xcite, @xcite, @xcite, @xcite, @xcite, @xcite). however, the amplitude, direction, and scale of these flows remain very much in contention, with resulting uncertainties in the theoretical interpretation and implications of these measurements (@xcite, @xcite). indeed, recently published conflicting results suggest that the motion of the lg is either due, or is not due, to material within 6000, and that _ iras _ galaxies either trace, or do not trace, the dark matter which gives rise to the observed peculiar velocities. the most recent potent reconstruction of the markiii velocities (@xcite) shows that the bulk velocity can be decomposed into two components arising from the mass fluctuation field within the sphere of radius @xmath3 about the lg and a component dominated by the mass distribution outside that volume. for convenience, we refer to this boundary at @xmath3 as the `` supergalactic shell '' since it includes the main local attractors in the supergalactic plane, the great attractor and perseus - pisces. this new analysis shows dominant infall patterns by the ga and pp but very little bulk flow within the supergalactic shell. the tidal component inside this volume is dominated by a flow of amplitude @xmath4 in the supergalactic direction @xmath5, which is likely generated by the external mass distribution on very large scales (see also @xcite, @xcite). this interpretation is also supported by an increasingly large number of tf / fp investigations (based on the distribution and motion of abell clusters) which report the detection of streaming motions of amplitudes greater than 700 beyond @xmath6 and away from the cmb dipole (@xcite, @xcite, @xcite, @xcite). other investigations using nearly homogeneous samples of galaxies within and outside the supergalactic shell find motion consistent with the amplitude and direction of the cmb dipole @xcite. this suggests that the reflex motion of the local group could be explained by material contained within the supergalactic shell. this confusion stems, in large part, in our inability to perfectly match the many heterogeneous samples for flow studies into one self - consistent homogeneous catalogue. much of the problem lies in the fact that, with the exception of a few surveys beyond @xmath7 (@xcite, @xcite, @xcite), none of the surveys within the supergalactic sphere sample the _ entire _ sky uniformly.
The markiii catalog of galaxy peculiar velocities
in an attempt to overcome this problem, two of us (jw & sc @xmath8 collaborators) have recently combined the major distance - redshift surveys from both hemispheres (published before 1994) into a catalog of 3100 galaxies (@xcite), but showed that full homogenization at the @xmath9% level, the minimum required for a @xmath10 bulk flow detection at 6000, can not be achieved. due to subjective reduction techniques and varying selection criteria, fundamental uncertainties remain when trying to match greatly disparate tf datasets (@xcite). furthermore, a revised calibration of the markiii tf zero - points based on maximal agreement with the peculiar velocities predicted by the iras 1.2jy redshift survey suggests a possible source of systematic error for the data sets which cover the pp cone (@xcite). this uncertainty has not seriously affected mass density reconstructions within the supergalactic shell (@xcite) but it could lead to spurious estimates of the bulk flows on larger scales. a newer calibration of the courteau / faber catalogue of northern spirals, not included in markiii, has been published (@xcite, @xcite) but a revision of the markiii catalogue is in progress (@xcite).
A new all-sky survey: shellflow
the need to tie all existing data bases for cosmic flow studies in an unambiguous fashion is clear. to that effect, we initiated a new survey in 1996 using noao facilities to measure tf distances for a complete, full - sky sample of sb@xmath0sc galaxies in the supergalactic shell for which we will obtain _ precise _ and _ uniform _ photometric and spectroscopic data. this will be the first well - defined full - sky survey to sample this scale, free of uncertainties from matching heterogeneous data sets. the sfi survey of giovanelli @xcite resembles ours in its scope and sky coverage, but it relies on a separate dataset (@xcite) for coverage of the southern sky and thus can not attain full - sky homogeneity. our survey, on the other hand, is designed from the outset to be homogeneous to the minimum level required for unambiguous bulk flow detection at the supergalactic shell. because of the overlap with existing surveys at comparable depth (markiii + sfi), this new compilation will be of fundamental importance in tying the majority of existing data sets together in a uniform way, which will greatly increase their usefulness for global analyses of mass fluctuations in the universe. our sample is selected from the optical redshift survey (@xcite), consisting of galaxies over the whole sky with m@xmath11 and @xmath12 from the ugc, eso, and esgc (@xcite). it includes all non - interacting sb and sc galaxies with redshifts between 4500 and 7000 from the local group and inclinations between @xmath13 and @xmath14, in regions where burstein - heiles extinction is less than 03. this yields an all - sky catalog of 297 galaxies. following the approach of @xcite, we use the sample itself to calibrate the distance indicator relation ; this mitigates the need to tie the sample to external tf calibrators such as clusters (although it precludes measurement of a monopole term in the velocity field). given a tf fractional distance error of 20%, the statistical uncertainty on a bulk flow from @xmath15 galaxies at common distance @xmath16 is @xmath17. as the measured (and much contested) bulk motions on these scales are of the order of 300, a detection of high statistical significance is well within reach.
Results and analysis
data taking and reduction techniques follow the basic guidelines of previous optical tf surveys (@xcite, @xcite, @xcite, @xcite). our survey is now complete, which is essential to achieve our statistical requirements and ensure a rigorous analysis. the spectroscopy relies on measurement of h@xmath18 rotation velocities at 2.2 disk scale lengths for the tightest tf calibration and best match to analogous 21 cm line widths (@xcite, @xcite). the photometry is based on the kron - cousins @xmath19 and @xmath20 systems which will allow direct matching with two largest tf field samples to date (@xcite,@xcite). one of the key features of this study is not only its all - sky sample selection but the independent duplication of all data reductions (by at least 2, if not 3, of us). these reductions and a first flow analysis based on the shellflow sample alone should be published soon (@xcite). we also plan a more extensive analysis using the recalibrated markiii combined with other new catalogs not included in the original markiii. 99 corwin, h. g, & skiff, b. a. 1994, extension to the southern galaxies catalogue, in preparation clutton - brock, m., & peebles, p.j.e. 1981, aj, 86, 1115 courteau, s. 1992, phd. thesis, uc santa cruz courteau, s., faber, s.m., dressler, a., & willick, j.a. 1993, apj, 412, l51 courteau, s. 1996, apjs, 103, 363 courteau, s. 1997, aj, 114, 2402 courteau, s. (+ shellflow team) 1999 (in preparation) da costa, l. n., freudling, w., wegner, g., giovanelli, r., haynes, m.p., & salzer, j.j. 1996, apj, 468, l5 dekel, a., eldar, a., kolatt, t., yahil, a., willick, j. a., faber, s. m., courteau, s., & burstein, d. 1999, apj (submitted) giovanelli, r., haynes, m.p., freudling, w., da costa, l. n., salzer, j.j., & wegner, g. 1998, apjl, in print, astro - ph/9807274 han, m.- s., & mould, j. r. 1992, apj, 396, 453 hudson, m. (+ smac team) 1998 (in preparation) lauer, t. r., & postman, m. 1994, apj, 425, 418 mathewson, d. s., & ford, v. l. 1994, apj, 434, l39 mathewson, d. s., ford, v. l, & buchhorn, m. 1992, apjs, 81, 413 [m92] postman, m. 1995, in _ dark matter _, aip conf. series 336, 371 postman, m., & lauer, t. r. 1995, apj, 440, 28 riess, a., press, w., & kirshner, r. p. 1995, apj, 445, l91 santiago, b. x., strauss, m. a., lahav, o., davis, m., & huchra, j. p. 1995, apj, 446, 457 scaramella, r., 1989, nature, 338, 562 schlegel, d. 1996, phd. thesis, uc berkeley strauss, m.a. 1996, in _ critical dialogues in cosmology _, ed. neil turok (singapore : world scientific) strauss, m.a., & willick, j.a. 1995, physics reports, 261, 271 willick, j. a., courteau, s., faber, s. m., burstein, d., dekel, a., & strauss, m. a. 1997, apjs, 109, 333 willick, j. a. & strauss, m. s. 1998, apj, in press (astro - ph/9801307) willick, j.a. 1998, apj (submitted) willick, j.a. 1999 (in preparation) | we present a new optical tully - fisher (tf) investigation for a complete, full - sky sample of 297 sb@xmath0sc spirals with redshifts between 4500 and 7000.
the survey was specifically designed to provide _ uniform, well - calibrated _ data over both hemispheres. all previous tf surveys within the supergalactic shell (@xmath1) have relied on matching separate data sets in the northern and southern hemispheres and thus can not attain full - sky homogeneity.
analyses of the cosmological dipole and peculiar velocities based on these studies have produced contradictory claims for the amplitude of the bulk flow and whether it is generated by internal or external mass fluctuations.
with shellflow, and further zero - point calibration of existing tf data sets, we expect a high - accuracy detection of the bulk flow amplitude and an unambiguous characterization of the tidal field at 6000.
# 1#1@xmath2 | astro-ph9809224 |
Introduction
numerous precision tests of the standard model (sm) and searches for its possible violation have been performed in the last few decades, serving as an invaluable tool to test the theory at the quantum level. they have also provided stringent constraints on many `` new physics '' (np) scenarios. a typical example is given by the measurements of the anomalous magnetic moment of the electron and the muon, where recent experiments reached the fabulous relative precision of 0.7 ppb @xcite and 0.5 ppm,@xcite respectively. these experiments measure the so - called gyromagnetic factor @xmath1, defined by the relation between the particle s spin @xmath5 and its magnetic moment @xmath6, = g, where @xmath7 and @xmath8 are the charge and mass of the particle. in the dirac theory of a charged point - like spin-@xmath9 particle, @xmath10. quantum electrodynamics (qed) predicts deviations from dirac s value, as the charged particle can emit and reabsorb virtual photons. these qed effects slightly increase the @xmath1 value. it is conventional to express the difference of @xmath1 from 2 in terms of the value of the so - called anomalous magnetic moment, a dimensionless quantity defined as @xmath11. the anomalous magnetic moment of the electron, @xmath12, is rather insensitive to strong and weak interactions, hence providing a stringent test of qed and leading to the most precise determination of the fine - structure constant @xmath13 to date.@xcite on the other hand, the @xmath1@xmath2@xmath3 of the muon, @xmath14, allows to test the entire sm, as each of its sectors contributes in a significant way to the total prediction. compared with @xmath12, @xmath14 is also much better suited to unveil or constrain np effects. indeed, for a lepton @xmath15, their contribution to @xmath16 is generally expected to be proportional to @xmath17, where @xmath18 is the mass of the lepton and @xmath19 is the scale of np, thus leading to an @xmath20 relative enhancement of the sensitivity of the muon versus the electron anomalous magnetic moment. this more than compensates the much higher accuracy with which the @xmath1 factor of the latter is known. the anomalous magnetic moment of the @xmath0 lepton, @xmath21, would suit even better ; however, its direct experimental measurement is prevented by the relatively short lifetime of this lepton, at least at present. the existing limits are based on the precise measurements of the total and differential cross sections of the reactions @xmath22 and @xmath23 at lep energies. the most stringent limit, @xmath24 at 95% confidence level, was set by the delphi collaboration,@xcite and is still more than an order of magnitude worse than that required to determine @xmath21. in the 1990s it became clear that the accuracy of the theoretical prediction of the muon @xmath1@xmath2@xmath3, challenged by the e821 experiment underway at brookhaven,@xcite was going to be restricted by our knowledge of its hadronic contribution. this problem has been solved by the impressive experiments at low - energy @xmath25 colliders, where the total hadronic cross section (as well as exclusive ones) were measured with high precision, allowing a significant improvement of the uncertainty of the leading - order hadronic contribution.@xcite as a result, the accuracy of the sm prediction for @xmath14 now matches that of its measurement. in parallel to these efforts, very many improvements of all other sectors of the sm prediction were carried on by a large number of theorists (see refs. for reviews). all these experimental and theoretical developments allow to significantly improve the theoretical prediction for the anomalous magnetic moment of @xmath0 lepton as well. in this article we review and update the sm prediction of @xmath21, analyzing in detail the three contributions into which it is usually split : qed, electroweak (ew) and hadronic. updated qed and ew contributions are presented in secs. [sec : qed] and [sec : ew] ; new values of the leading - order hadronic term, based on the recent low energy @xmath4 data from babar, cmd-2, kloe and snd, and of the hadronic light - by - light contribution are presented in sec. [sec : had]. the total sm prediction is confronted to the available experimental bounds on the @xmath0 lepton @xmath1@xmath2@xmath3 in sec. [sec : sm], and prospects for its future measurements are briefly discussed in sec. [sec : conc], where conclusions are drawn.
Qed contribution to @xmath26
the qed part of the anomalous magnetic moment of the @xmath0 lepton arises from the subset of sm diagrams containing only leptons and photons. this dimensionless quantity can be cast in the general form:@xcite a_^qed = a_1 + a_2 () + a_2 () + a_3 (,), [eq : atauqedgeneral] where @xmath27, @xmath28 and @xmath29 are the electron, muon and @xmath0 lepton masses, respectively. the term @xmath30, arising from diagrams containing only photons and @xmath0 leptons, is mass and flavor independent. in contrast, the terms @xmath31 and @xmath32 are functions of the indicated mass ratios, and are generated by graphs containing also electrons and/or muons. the functions @xmath33 (@xmath34) can be expanded as power series in @xmath35 and computed order - by - order a_i = a_i^(2) () + a_i^(4) () ^2 + a_i^(6) () ^3 + a_i^(8) () ^4 +. only one diagram is involved in the evaluation of the lowest - order (first - order in @xmath13, second - order in the electric charge) contribution it provides the famous result by schwinger @xmath36.@xcite the mass - dependent coefficients @xmath31 and @xmath32, discussed below, are of higher order. they were derived using the latest codata@xcite recommended mass ratios : m_/m_e & = & 3477.48 (57) [eq : rte] + m_/m _ & = & 16.8183 (27). [eq : rtm] the value for @xmath29 adopted by codata in ref., @xmath37 mev, is based on the pdg 2002 result.@xcite it remained unchanged until very recently (see refs. ), when preliminary results of two new measurements (from the belle@xcite and kedr@xcite detectors) were reported. the central values of the new mass values are slightly lower than the current world average value, but agree with it within the uncertainties, which are approaching that of the world average value (used in this work). seven diagrams contribute to the fourth - order coefficient @xmath38, one to @xmath39 and one to @xmath40. they are depicted in fig. [fig : qed2]. as there are no two - loop diagrams contributing to @xmath41 that contain both virtual electrons and muons, @xmath42. the mass - independent coefficient has been known for almost fifty years:@xcite a_1^(4) & = & + + (3) - 2 + & = & -0.328 478 965 579 193 78 , [eq : a14] where @xmath43 is the riemann zeta function of argument @xmath44. for @xmath45, @xmath46 or @xmath0, the coefficient of the two - loop mass - dependent contribution to @xmath47, @xmath48, with @xmath49, is generated by the diagram with a vacuum polarization subgraph containing the virtual lepton @xmath50. this coefficient was first computed in the late 1950s for the muon @xmath1@xmath2@xmath3 with @xmath51, neglecting terms of @xmath52.@xcite the exact expression for @xmath53 was reported by elend in 1966.@xcite however, its numerical evaluation was considered tricky because of large cancellations and difficulties in the estimate of the accuracy of the results, so that common practice was to use series expansions instead.@xcite taking advantage of the properties of the dilogarithm @xmath54,@xcite the exact result was cast in ref. in a very simple and compact analytic form, valid, contrary to the one in ref. , also for @xmath55 (the case relevant to @xmath56 and part of @xmath57) : a_2^(4)(1/x) & = & - - + x^2 (4 + 3x) + (1 - 5 x^2) + & & + + & & + x^4. [eq : ea24] for @xmath58, gives @xmath59 ; of course, this contribution is already part of @xmath38 in. numerical evaluation of with the mass ratios given in eqs. ([eq : rte])([eq : rtm]) yields the two - loop mass - dependent qed contributions to the anomalous magnetic moment of the @xmath0 lepton,@xcite a_2^(4)(m_/m_e) & = & 2.024 284 (55), [eq : ta24e] + a_2^(4)(m_/m _) & = & 0.361 652 (38). [eq : ta24 m] these two values are very similar to those computed via a dispersive integral in ref. (which, however, contain no estimates of the uncertainties). equations ([eq : ta24e]) and ([eq : ta24 m]) are also in agreement (but more accurate) with those of ref. . adding up eqs. ([eq : a14]), ([eq : ta24e]) and ([eq : ta24 m]) one gets:@xcite c_^(4) = 2.057 457 (93) [eq : tc2] (note that the uncertainties in @xmath60 and @xmath61 are correlated). the resulting error @xmath62 leads to a @xmath63 uncertainty in @xmath41. more than one hundred diagrams are involved in the evaluation of the three - loop (sixth - order) qed contribution. their analytic computation required approximately three decades, ending in the late 1990s. the coefficient @xmath64 arises from 72 diagrams. its exact expression, mainly due to remiddi and his collaborators, reads:@xcite a_1^(6) & = & ^2 (3) - (5) - ^4 + + + & & + (3) - ^2 2 + ^2 + + & & + + & = & 1.181 241 456 587 . [eq : a16] this value is in very good agreement with previous results obtained with numerical methods.@xcite the calculation of the exact expression for the coefficient @xmath65 for arbitrary values of the mass ratio @xmath66 was completed in 1993 by laporta and remiddi @xcite (earlier works include refs. ). let us focus on @xmath67 (@xmath68, @xmath69). this coefficient can be further split into two parts : the first one, @xmath70, receives contributions from 36 diagrams containing either electron or muon vacuum polarization loops,@xcite whereas the second one, @xmath71, is due to 12 light - by - light scattering diagrams with either electron or muon loops.@xcite the exact expressions for these coefficients are rather complicated, containing hundreds of polylogarithmic functions up to fifth degree (for the light - by - light diagrams) and complex arguments (for the vacuum polarization ones) they also involve harmonic polylogarithms.@xcite series expansions were provided in ref. for the cases of physical relevance. using the recommended mass ratios given in eqs. ([eq : rte]) and ([eq : rtm]), the following values were recently computed from the full analytic expressions:@xcite a_2^(6)(m_/m_e,) & = & 7.256 99 (41) [eq : ta26evac] + a_2^(6)(m_/m_e,) & = & 39.1351 (11) [eq : ta26elbl] + a_2^(6)(m_/m_,)&= & -0.023 554 (51) [eq : ta26mvac] + a_2^(6)(m_/m_,)&= & 7.033 76 (71). [eq : ta26mlbl] almost identical values were obtained employing the approximate series expansions of ref. : 7.25699(41), 39.1351(11), @xmath20.023564(51), 7.03375(71).@xcite the previous estimates of ref. were different : 10.0002, 39.5217, 2.9340, and 4.4412 (no error estimates were provided), respectively ; they are superseded by the results in eqs. ([eq : ta26evac])([eq : ta26mlbl]), derived from the exact analytic expressions. the estimates of ref. compare slightly better : 7.2670, 39.6, @xmath72, 4.47 (no errors provided). in the specific case of @xmath73, the values of refs. and differ from because their derivations did not include terms of @xmath74, which turn out to be unexpectedly large. the sums of eqs. ([eq : ta26evac])([eq : ta26elbl]) and ([eq : ta26mvac])([eq : ta26mlbl]) are a_2^(6)(m_/m_e) & = & 46.3921 (15), [eq : ta26e] + a_2^(6)(m_/m_)&= & 7.010 21 (76). [eq : ta26 m] the contribution of the three - loop diagrams with both electron- and muon - loop insertions in the photon propagator was calculated numerically from the integral expressions of ref. , obtaining:@xcite a_3^(6)(m_/m_e, m_/m _) = 3.347 97 (41). [eq : ta36] this value disagrees with the results of refs. (1.679) and (2.75316). combining the three - loop results of eqs. ([eq : a16]), ([eq : ta26e]), ([eq : ta26 m]) and ([eq : ta36]) one finds the sixth - order qed coefficient,@xcite c_^(6) = 57.9315 (27). [eq : tc3] the error @xmath75 induces a @xmath76 uncertainty in @xmath41. the order of magnitude of the three - loop contribution to @xmath41, dominated by the mass - dependent terms, is comparable to that of ew and hadronic effects (see later). contrary to the case of the electron and muon @xmath1@xmath2@xmath3, qed contributions of order higher than three are not known.@xcite (an exception is the mass- and flavor - independent term @xmath77,@xcite which is however expected to be a very small part of the complete four - loop contribution.) adding up all the above contributions and using the new value of @xmath13 derived in refs. and, @xmath78, one obtains the total qed contribution to the @xmath1@xmath2@xmath3 of the @xmath0 lepton,@xcite a_^qed = 117 324 (2) 10 ^ -8. [eq : tqed] the error @xmath79 is the uncertainty @xmath80 assigned to @xmath41 for uncalculated four - loop contributions. as we mentioned earlier, the errors due to the uncertainties of the @xmath81 and @xmath82 terms are negligible. the error induced by the uncertainty of @xmath13 is only @xmath83 (and thus totally negligible).
Electroweak contribution to @xmath26
with respect to schwinger s contribution, the ew correction to the anomalous magnetic moment of the @xmath0 lepton is suppressed by the ratio @xmath84, where @xmath85 is the mass of the @xmath86 boson. numerically, this contribution is of the same order of magnitude as the three - loop qed one. the analytic expression for the one - loop ew contribution to @xmath26, due to the diagrams in fig. [fig : ew1], is:@xcite ^ew () =, [eq : ewoneloop] where @xmath87 is the fermi coupling constant,@xcite @xmath88, @xmath85 and @xmath89 are the masses of the @xmath90, @xmath86 and higgs bosons, and @xmath91 is the weak mixing angle. closed analytic expressions for @xmath92 taking exactly into account the @xmath93 dependence (@xmath94 higgs, or other hypothetical bosons) can be found in refs. . following ref. , we employ for @xmath95 the on - shell definition,@xcite @xmath96, where @xmath97,@xcite and @xmath85 is the theoretical sm prediction of the @xmath86 mass. the latter can be easily derived from the simple analytic formulae of ref. (see also refs. ), =, [eq : fops] (on - shell scheme ii with @xmath98,@xcite @xmath99,@xcite and @xmath100 gev,@xcite) leading to @xmath101 for @xmath102. this result should be compared with the direct experimental value @xmath103,@xcite which corresponds to a very small @xmath89. in any case, these shifts in the @xmath85 prediction induced by the variation of @xmath89 from 114.4 gev, the current lower bound at 95% confidence level,@xcite up to a few hundred gev, only change @xmath92 by amounts of @xmath104. from, including the tiny @xmath105 corrections of refs., for @xmath102 we get ^ew () = 55.1(1) 10 ^ -8. [eq : ewoneloopn] the uncertainty encompasses the shifts induced by variations of @xmath89 from 114 gev up to a few hundred gev, and the tiny uncertainty due to the error in @xmath29. the estimate of the ew contribution in ref., @xmath106, obtained from the one - loop formula (without the small corrections of order @xmath107), is similar to our value in. however, its uncertainty (@xmath108) is too small, and it does nt contain the two - loop contribution which, as we ll discuss in the next section, is not negligible. the two - loop ew contributions to @xmath16 (@xmath45, @xmath46 or @xmath0) were computed in 1995 by czarnecki, krause and marciano.@xcite this remarkable calculation leads to a significant reduction of the one - loop prediction. navely one would expect the two - loop ew contribution @xmath109 to be of order @xmath110, but this turns out not to be so. as first noticed in the early 1990s,@xcite @xmath111 is actually quite substantial because of the appearance of terms enhanced by a factor of @xmath112, where @xmath113 is a fermion mass scale much smaller than @xmath85. the two - loop contribution to @xmath114 involves 1678 diagrams in the linear t hooft - feynman gauge@xcite (as a check, the authors of refs. and employed both this gauge and a nonlinear one in which the vertex of the photon, the @xmath86 and the unphysical charged scalar vanishes). it can be divided into fermionic and bosonic parts ; the former, @xmath115, includes all two - loop ew corrections containing closed fermion loops, whereas all other contributions are grouped into the latter, @xmath116. the expressions of ref. for the bosonic part were obtained in the approximation @xmath117, computing the first two terms in the expansion in @xmath118, and expanding in @xmath95, keeping the first four terms in this expansion (this number of powers is sufficient to obtain an exact coefficient of the large logarithms @xmath119). recent analyses of the ew bosonic corrections of the @xmath1@xmath2@xmath3 of the muon@xcite relaxed these approximations, providing analytic results valid also for a light higgs. considering the present @xmath120 gev lower bound,@xcite we can safely employ the results of ref. , obtaining, for @xmath102 @xmath121. the neglected terms are of @xmath122. the fermionic part of @xmath123 contains the contribution of diagrams with light quarks ; they involve long - distance qcd for which perturbation theory can not be employed. in particular, these hadronic uncertainties arise from two types of two - loop diagrams : the hadronic photon@xmath90 mixing, and quark triangle loops with the external photon, a virtual photon and a @xmath90 attached to them (see fig. [fig : ew2]). the hadronic uncertainties mainly arise from the latter ones. two approaches were suggested for their study : in ref. the nonperturbative effects where modeled introducing effective quark masses as a simple way to account for strong interactions. in view of the high experimental precision of the @xmath1@xmath2@xmath3 of the muon, a more realistic treatment of the relevant hadronic dynamics was introduced in ref. within a low - energy effective field theory approach, later on developed in the detailed analyses of refs. . however, from a numerical point of view, the discrepancy between the results provided by these two different approaches turns out to be irrelevant for the present interpretation of the experimental result of the muon @xmath1@xmath2@xmath3, in spite of its precision. the use of effective quark masses for the study of the @xmath1@xmath2@xmath3 of the @xmath0 whose experimental precision is a far cry from that of the muon! thus appears to be sufficient at present. the tiny hadronic @xmath124@xmath90 mixing terms can be evaluated either in the free quark approximation or via a dispersion relation using data from @xmath25 annihilation into hadrons ; the difference was shown to be numerically insignificant.@xcite references and contain simple approximate expressions for the contributions of the diagrams with fermion triangle loops shown in fig. [fig : ew2] (right). in general, for a lepton @xmath45, @xmath46 or @xmath0, neglecting small mass - ratios, they are a_l^ew () _ f, [eq : ew2lfd] where @xmath125 is the third component of the weak isospin of the fermion @xmath126 in the loop, @xmath127 is its charge, @xmath128 its number of colors (3 for quarks, 1 for leptons), and c(f) \ { ll (^2/m_l^2) + 5/6 & + (^2/m_l^2) - 8 ^2/27 + 11/18 & + (^2/m_f^2) -2 & + (m_top^2/^2) + + + (5/18)(^2/m_top^2) -4/3 & . [eq : ew2lfdlogs] the contribution of the top - quark triangle loop diagram of fig. [fig : ew2] (right) with the @xmath90 boson replaced by the neutral goldstone boson (@xmath129) has also been included in this expression.@xcite it is clear from eqs. ([eq : ew2lfd])([eq : ew2lfdlogs]) that the logarithms @xmath130 cancel in sums over all fermions of a given generation, as long as @xmath131, due to the no - anomaly condition @xmath132 valid within every generation.@xcite this does not occur for the third generation due to the large mass of the top quark. note that this short - distance cancellation does not get modified by strong interaction effects on the quark triangle diagrams.@xcite contrary to the case of the muon @xmath1@xmath2@xmath3, where all fermion masses, with the exception of @xmath27, enter in @xmath133, the approximate expressions in show that this is not the case for the @xmath1@xmath2@xmath3 of the @xmath0 lepton. indeed, due to the high infrared cut - off set by @xmath29, for @xmath134 does not depend on any fermion mass lighter than @xmath29 ; apart from @xmath29, it only depends on @xmath135 and @xmath136, the masses of the top and bottom quarks (assuming @xmath137). the charm contribution requires some care, as the crude approximation provided by for a charm lighter than the @xmath0 lepton is valid only if @xmath138. clearly, this is not a good approximation, and the spurious shift induced by when @xmath139 is varied across the @xmath29 threshold is of @xmath140. one possibility is to use with @xmath139 equal to @xmath29.@xcite better still, we numerically integrated the exact expressions for @xmath141 provided in ref. for arbitrary values of @xmath113, obtaining a smooth dependence on the value of @xmath139. for completeness we repeated this detailed analysis for all light fermions. as expected, the result depends very mildly on the values chosen for their masses. employing the values @xmath142 gev, @xmath143 gev, @xmath144 gev and @xmath145 gev, and adding to the contribution of the remaining fermionic two - loop diagrams studied in ref., for @xmath146 we obtain @xmath147. in this evaluation we also included the tiny @xmath122 contribution of the @xmath124@xmath90 mixing diagrams, suppressed by (@xmath148 for quarks and (@xmath149 for leptons, via the explicit formulae of ref. . the sum of the fermionic and bosonic two - loop ew contributions described above gives @xmath150, a 14% reduction of the one - loop result. the leading - logarithm three - loop ew contributions to the muon @xmath1@xmath2@xmath3 were determined to be extremely small via renormalization - group analyses.@xcite we assigned to our @xmath0 lepton @xmath1@xmath2@xmath3 ew result an additional uncertainty of @xmath151 \!\sim\! o(10^{-9})$] to account for these neglected three - loop effects. adding @xmath152 to the one - loop value of we get our total ew correction (for @xmath146) ^ew = 47.4 (5) 10 ^ -8. [eq : tew] the uncertainty allows @xmath89 to range from 114 gev up to @xmath153 gev, and reflects the estimated errors induced by hadronic loop effects (@xmath154 and @xmath155 can vary between 70 mev and 400 mev), neglected two - loop bosonic terms, and the missing three - loop contribution. it also includes the small errors due to the uncertainties in @xmath135 and @xmath29. the value in is in agreement with the prediction @xmath156,@xcite with a reduced uncertainty. as we mentioned in sec. [subsec : ew1], the ew estimate of ref., @xmath106, mainly differs from in that it does nt include the two - loop corrections.
The hadronic contribution
in this section we will analyze @xmath157, the contribution to the @xmath0 anomalous magnetic moment arising from qed diagrams involving hadrons. hadronic effects in (two - loop) ew contributions are already included in @xmath114 (see the previous section). similarly to the case of the muon @xmath1@xmath2@xmath3, the leading - order hadronic contribution to the @xmath0 lepton anomalous magnetic moment is given by the dispersion integral:@xcite @xmath158 where the kernel @xmath159 is a bounded function of energy monotonously increasing to unity at @xmath160, and @xmath161 is the total hadronic cross section of the @xmath25 annihilation in the born approximation. in fig. [fig : rat] we plot the ratio of the kernels in the @xmath0 lepton and muon case. clearly, although the role of the low energies is still very important, the different structure of @xmath162 compared to @xmath163, induced by the higher mass of the @xmath0, results in a relatively higher role of the larger energies. the history of these calculations is not as rich as that of the muon. the first calculation performed in 1978 in ref. was based on experimental data available at that time below 7.4 gev, whereas at higher energies the asymptotic qcd prediction was used. ten years later, a rough estimate was made in ref. based on low energy @xmath25 data. in ref. the contribution of the @xmath164 meson was estimated by integrating the approximation obtained using the breit - wigner curve, while other contributions used the data. the accuracy of the calculation was considerably improved in refs. where, below 40 gev, only data were used. in ref., data were only used below 3 gev (together with the experimental parameters of the @xmath165 and @xmath166 family states). in our opinion this can significantly underestimate the resulting uncertainty. in addition, in the same reference, data from @xmath0 lepton decays were extensively used ; as it is known today, this leads to higher spectral functions than in @xmath25 case,@xcite and can therefore overestimate the result. the results of these calculations are summarized in table [tab : atau]. [tab : atau] for completeness, in the second part of table [tab : atau] we also show purely theoretical estimates. the analysis based on qcd sum rules performed in ref. gives results which strongly depend on the choice of quark and gluon condensates. qcd sum rules are also used in ref. . in ref. the authors use a nonlocal constituent quark model for the description of the photon vacuum polarization function @xmath167 at space - like momenta and obtain @xmath168, close to the estimates based on the experimental data. they also show that a simpler model with constituent quark masses independent of momentum is strongly dependent on the values chosen for the quark masses. for example, with @xmath169 mev and @xmath170 550 mev their result is @xmath171, i.e., significantly smaller than the previous estimate. they could reproduce the value @xmath172 using @xmath173 mev. in a recent analysis using the instanton liquid model the author obtains @xmath174.@xcite all these estimates somewhat undervalue the hadronic contribution and have rather large uncertainties. we updated the calculation of the leading - order contribution using the whole bulk of experimental data below 12 gev, which include old data compiled in refs. , as well as the recent datasets from the cmd-2@xcite and snd@xcite experiments in novosibirsk, and from the radiative return studies at kloe in frascati@xcite and babar at slac.@xcite the improvement is particularly visible in the channel @xmath175, where four new independent measurements exist in the most important @xmath164 meson region : cmd-2,@xcite snd,@xcite and kloe@xcite (see fig. [fig : pi]). our result is a_^hlo = 337.5 (3.7) 10 ^ -8 [eq : thlo] (we recently presented a preliminary estimate of this value in ref. ). the breakdown of the contributions of different energy regions as well as their relative fractions in the total leading - order contribution are given in table [tab : at]. the contribution of the @xmath164 meson energy range is still important, but its relative weight is smaller than in the case of the muon anomaly, 51.3% compared to about 72%.@xcite the contributions of the narrow resonances (@xmath165 and @xmath166 families) are included in the corresponding energy regions. it is worth noting that uncertainties of the contributions from the hadronic continuum are larger than that of the very precise @xmath176 one. the overall uncertainty is 2.5 times smaller than that of the previous data - based prediction.@xcite [tab : at] the hadronic higher - order @xmath177 contribution @xmath178 can be divided into two parts : @xmath179 the first one is the @xmath82 contribution of diagrams containing hadronic self - energy insertions in the photon propagators. it was determined by krause in 1996:@xcite a_^hho()= 7.6 (2) 10 ^ -8. [eq : thhovac] note that navely rescaling the muon result by the factor @xmath180 (as it was done in ref. ) leads to the totally incorrect estimate @xmath181 (the @xmath182 value is from ref. ) ; even the sign is wrong! the second term, also of @xmath82, is the hadronic light - by - light contribution. similarly to the case of the muon @xmath1@xmath2@xmath3, this term can not be directly determined via a dispersion relation approach using data (unlike the leading - order hadronic contribution), and its evaluation therefore relies on specific models of low - energy hadronic interactions with electromagnetic currents. actually, very few estimates of @xmath183 exist in the literature,@xcite and all of them were obtained simply rescaling the muon results @xmath184 by a factor @xmath180. following this very nave procedure, the @xmath185 estimate varies between @xmath186\times (m_{\tau}^2/m_{\mu}^2)= 23(11)\times 10^{-8 } $], and @xmath187\times (m_{\tau}^2/m_{\mu}^2)= 38(7)\times 10^{-8 } $], according to the values chosen for @xmath188 from refs. and, respectively. these very nave estimates fall short of what is needed. consider the function @xmath71, the three - loop qed contribution to the @xmath1@xmath2@xmath3 of a lepton of mass @xmath18 due to light - by - light diagrams involving loops of a fermion of mass @xmath189 (see sec. [subsec : qed3]). the exact expression of this function, computed in ref. for arbitrary values of the mass ratio @xmath66, is rather complicated, but series expansions were provided in the same article for the cases of physical relevance. in particular, if @xmath190, then @xmath191. this implies that, for example, the (negligible) part of @xmath192 due to diagrams with a top - quark loop can be reasonably estimated simply rescaling the corresponding part of @xmath188 by a factor @xmath180. on the other hand, to compute the dominant contributions to @xmath192, i.e.those induced by the light quarks, we need the opposite case : @xmath193. in this limit, @xmath71 does not scale as @xmath194, and a nave rescaling of @xmath184 by @xmath180 to derive @xmath195 leads to an incorrect estimate. we therefore decided to perform a parton - level estimate of @xmath192 based on the exact expression for @xmath71 using the quark masses recently proposed in ref. for the determination of @xmath184 : @xmath196 mev, @xmath197 mev, @xmath198 gev and @xmath199 gev (note that with these values the authors of ref. obtain @xmath200, in perfect agreement with the value in ref. see also ref. for a similar earlier determination). we obtain a_^hho()= 5 (3) 10 ^ -8. [eq : thholbl] this value is much lower than those obtained by simple rescaling of @xmath188 by @xmath180. the up - quark provides the dominating contribution ; the uncertainty @xmath201 allows @xmath154 to range from 70 mev up to 400 mev. further independent studies (following the approach of ref. , for example) would provide an important check of this result. the total hadronic contribution to the anomalous magnetic moment of the @xmath0 lepton can be immediately derived adding the values in eqs. ([eq : thlo]), ([eq : thhovac]) and ([eq : thholbl]), a_^had = a_^hlo+ a_^hho()+ a_^hho () = 350.1 (4.8) 10 ^ -8. [eq : thad] errors were added in quadrature.
The standard model prediction for @xmath21
we can now add up all the contributions discussed in the previous sections to derive the sm prediction for @xmath21 : a_^sm = a_^qed + a_^ew + a_^hlo + a_^hho () + a_^hho (), [eq : sm] where @xmath202 (the sum of the hadronic contributions is given in). adding errors in quadrature, our final result is a_^sm = 117 721 (5) 10 ^ -8. [eq : nsm] the present pdg limit on the anomalous magnetic moment of the @xmath0 lepton was derived in 2004 by the delphi collaboration from @xmath22 total cross section measurements at @xmath203 between 183 and 208 gev at lep2:@xcite -0.052 < a _ < 0.013 [eq : exp_delphi1] at 95% confidence level. the authors of ref. also quote their result in the form of central value and error : a _ = -0.018 (17). [eq : exp_delphi2] comparing this result with (their difference is roughly one standard deviation), it is clear that the sensitivity of the best existing measurements is still more than an order of magnitude worse than needed. a reanalysis of various measurements of the cross section of the process @xmath204, the transverse @xmath0 polarization and asymmetry at lep and sld, as well as of the decay width @xmath205 at lep and tevatron, allowed to set a stronger model - independent limit:@xcite -0.007 < a _ 0.005. other limits on @xmath21 can be found in refs. .
Conclusions
in this article we reviewed and updated the sm prediction of the @xmath0 lepton @xmath1@xmath2@xmath3. updated qed and electroweak contributions were presented, together with new values of the leading - order hadronic term, based on the recent low energy @xmath4 data from babar, cmd-2, kloe and snd, and of the hadronic light - by - light contribution. these results were confronted in sec. [sec : sm] to the available experimental bounds on the @xmath0 lepton anomaly. as we already mentioned in the introduction, quite generally, np associated with a scale @xmath19 is expected to modify the sm prediction of the anomalous magnetic moment of a lepton @xmath15 of mass @xmath18 by a contribution @xmath206. therefore, given the large factor @xmath207, the @xmath1@xmath2@xmath3 of the @xmath0 lepton is much more sensitive than the muon one to ew and np loop effects that give contributions @xmath208, making its measurement an excellent opportunity to unveil (or just constrain) np effects. another interesting feature can be observed comparing the magnitude of the ew and hadronic contributions to the muon and @xmath0 lepton @xmath1@xmath2@xmath3. the ew contribution to the @xmath1@xmath2@xmath3 of the @xmath0 is only a factor of seven smaller than the hadronic one, compared to a factor of 45 for the @xmath1@xmath2@xmath3 of the muon. also, while the ew contribution to @xmath209 is only a factor of three larger than the present uncertainty of the hadronic contribution, this factor raises to 10 for the @xmath0 lepton. if a np contribution were of the same order of magnitude as the ew one, from a purely theoretical point of view, the @xmath1@xmath2@xmath3 of the @xmath0 would provide a much cleaner test of the presence (or absence) of such np effects than the muon one. indeed, if this were the case, such a np contribution to the @xmath0 lepton @xmath1@xmath2@xmath3 would be much larger than the hadronic uncertainty, which is currently the limiting factor of the sm prediction. unfortunately, the very short lifetime of the @xmath0 lepton makes it very difficult to determine its anomalous magnetic moment by measuring its spin precession in the magnetic field, like in the muon @xmath1@xmath2@xmath3 experiment.@xcite instead, experiments focused on high - precision measurements of the @xmath0 lepton pair production in various high - energy processes, comparing the measured cross sections with the qed predictions.@xcite as we can see from, the sensitivity of the best existing measurements is still more than an order of magnitude worse than that required to determine @xmath21. nonetheless, the possibility to improve such a measurement is certainly not excluded. for example, it was suggested to determine the @xmath0 lepton @xmath1 factor taking advantage of the radiation amplitude zero which occurs at the high - energy end of the lepton distribution in radiative @xmath0 decays.@xcite this method requires a very good energy resolution and could perhaps be employed at a @xmath0-charm or @xmath210 factory also benefiting from the possibility to collect very high statistics. it is not clear whether the huge data samples at @xmath210 factories will result in a corresponding gain for the limits on @xmath21. indeed, lep measurements were rather limited by systematic uncertainties, which were of the order of 2 - 3% for the discussed processes and, until now, experiments at @xmath210 factories have not yet reached such a level of accuracy in the absolute measurements of the total cross sections. however, a search for the @xmath0 lepton electric dipole moment at belle@xcite showed that with the appropriate choice of observables, using full information about events, the improvement in sensitivity can be proportional to the square root of luminosity, i.e., determined mainly by statistics. one can hope that this is also the case with the determination of @xmath21. a similar method to study @xmath21 using radiative @xmath86 decays and potentially very high data samples at lhc was suggested in ref. . yet another method would use the channeling in a bent crystal similarly to the suggestion for the measurement of magnetic moments of short - living baryons.@xcite this method has been successfully tested by the e761 collaboration at fermilab, which measured the magnetic moment of the @xmath211 hyperon.@xcite in the case of the @xmath0 lepton, it was suggested to use the decay @xmath212, which would produce polarized @xmath0 leptons.@xcite in 1991, when this suggestion was published, the idea seemed completely unlikely. however, in the era of @xmath210 factories, when the decay @xmath212 is already observed by the belle collaboration,@xcite and the possibility of a super-@xmath210 factory is actively discussed, this is no longer a dream. even more promising could be the realization of this idea in a dedicated experiment at a hadron collider with its huge number of @xmath210 mesons produced and a more suitable geometry. we believe that a detailed feasibility study of such an experiment, as well as further attempts to improve the accuracy of the theoretical prediction for @xmath21, are quite timely.
Acknowledgments
we would like to thank m. giacomini and f.v. ignatov for many valuable comments and collaborations on topics presented in this manuscript. we are greatly indebted to a. vainshtein for communications concerning the hadronic light - by - light contribution to @xmath21 and to f. jegerlehner for many fruitful discussions. we are also grateful to k. inami for an interesting discussion on the feasibility of @xmath21 measurements at @xmath210 factories. thanks the dipartimento di fisica, universit di padova and infn, sezione di padova, where part of this work was done, for its hospitality. the work of s.e. was supported in part by the grants of rfbr 06 - 02 - 04018 and 06 - 02 - 16156 as well as by the grant of dfg gz : 436 rus 113/769/0 - 2. is grateful to the instituto de fsica da universidade federal da bahia, brasil, for the hospitality during a visit when this manuscript was finalized. the work of m.p. was supported in part by the european community s marie curie research training networks under contracts mrtn - ct-2004 - 503369 and mrtn - ct-2006 - 035505. all diagrams were drawn with jaxodraw.@xcite g.w. bennett _ et al. _, phys. d * 73 * (2006) 072003 ; phys. * 92 * (2004) 161802 ; phys. * 89 * (2002) 101804 ; _ ibid. _ * 89 * (2002) 129903 ; h.n. brown _ et al. _, phys. rev. lett. * 86 * (2001) 2227. g. gabrielse, d. hanneke, t. kinoshita, m. nio, and b. odom, phys. rev. lett. * 97 * (2006) 030802. k. melnikov and a. vainshtein, _ theory of the muon anomalous magnetic moment _, springer, 2006 ; m. passera, nucl. * 155 * (2006) 365 ; m. davier and w.j. marciano, annu. nucl. part. * 54 * (2004) 115 ; m. knecht, arxiv : hep - ph/0307239 ; a. nyffeler, acta phys. b * 34 * (2003) 5197 ; k. melnikov, int. j. mod. phys. a * 16 * (2001) 4591 ; hughes and t. kinoshita, rev. * 71 * (1999) s133 ; a. czarnecki and w.j. marciano, nucl. * 76 * (1999) 245. m. passera, j. phys. g * 31 * (2005) r75. t. kinoshita and w.j. marciano, in _ quantum electrodynamics _, edited by t. kinoshita (world scientific, singapore, 1990), pp. 419478. k. hagiwara _ et al. _ [particle data group], phys. rev. d * 66 * (2002) 010001. s. eidelman _ et al. _ [particle data group], phys. b * 592 * (2004) 1. w.- yao _ et al. _ [particle data group], j. phys. g * 33 * (2006) 1. elend, phys. lett. * 20 * (1966) 682 ; * 21 * (1966) 720(e). samuel and g. li, phys. d * 44 * (1991) 3935 ; * 46 * (1992) 4782(e) ; * 48 * (1993) 1879(e). g. li, r. mendel, and m.a. samuel, phys. d * 47 * (1993) 1723. a. czarnecki amd m. skrzypek, phys. b * 449 * (1999) 354. s. narison, phys. b * 513 * (2001) 53 ; * 526 * (2002) 414(e). mignaco and e. remiddi, nuovo cim. a * 60 * (1969) 519 ; r. barbieri and e. remiddi, phys. b * 49 * (1974) 468 ; r. barbieri, m. caffo and e. remiddi, _ ibid. _ * 57 * (1975) 460 ; m.j. levine, e. remiddi and r. roskies, phys. d * 20 * (1979) 2068 ; s. laporta and e. remiddi, phys. b * 265 * (1991) 182 ; s. laporta, phys. d * 47 * (1993) 4793 ; phys. b * 343 * (1995) 421 ; s. laporta and e. remiddi, phys. b * 356 * (1995) 390. s. laporta and e. remiddi, phys. b * 379 * (1996) 283. t. kinoshita, phys. * 75 * (1995) 4728 ; _ quantum electrodynamics _, edited by t. kinoshita (world scientific, singapore, 1990), pp. 218321. s. laporta, nuovo cim. a * 106 * (1993) 675. s. laporta and e. remiddi, phys. b * 301 * (1993) 440. t. kinoshita, nuovo cim. b * 51 * (1967) 140 ; b.e. lautrup and e. de rafael, phys. * 174 * (1968) 1835 ; r. barbieri and e. remiddi, nucl. b * 90 * (1975) 233 ; b.e. lautrup and m.a. samuel, phys. b * 72 * (1977) 114. e. remiddi and j.a.m. vermaseren, int. j. mod. phys. a * 15 * (2000) 725 ; t. gehrmann and e. remiddi, comput. commun. * 141 * (2001) 296 ; * 144 * (2002) 200 ; d. matre, comput. * 174 * (2006) 222. samuel, g. li and r. mendel, phys. * 67 * (1991) 668 ; * 69 * (1992) 995(e). t. kinoshita and m. nio, phys. d * 73 * (2006) 013003. t. kinoshita and m. nio, phys. d * 70 * (2004) 113001 ; _ ibid. _ d * 73 * (2006) 053007. r. jackiw and s. weinberg, phys. d * 5 * (1972) 2396 ; i. bars and m. yoshimura, phys. d * 6 * (1972) 374 ; g. altarelli, n. cabibbo and l. maiani, phys. b * 40 * (1972) 415 ; w.a. bardeen, r. gastmans and b. lautrup, nucl. b * 46 * (1972) 319 ; k. fujikawa, b.w. lee and a.i. sanda, phys. rev. d * 6 * (1972) 2923. ternov, v.n. rodionov and a.i. studenikin, yad. * 37 * (1983) 1270 (sov. j. nucl. phys. * 37 * (1983) 755) ; annalen phys. * 46 * (1989) 303 ; a.i. studenikin, zh. eksp. * 97 * (1990) 1407 (sov. jetp * 70 * (1990) 795) ; fiz. chast. atom. yadra * 21 * (1990) 605 (sov. j. part. * 21 * (1990) 259) ; arxiv : hep - ph/9808219 ; yad. * 62 * (1999) 2248 (phys. atom. * 62 * (1999) 2071). a. czarnecki, w.j. marciano and a. vainshtein, phys. d * 67 * (2003) 073006. a. sirlin, phys. d * 22 * (1980) 971. a. ferroglia, g. ossola, m. passera and a. sirlin, phys. d * 65 * (2002) 113002. g. degrassi, p. gambino, m. passera, a. sirlin, phys. b * 418 * (1998) 209 ; g. degrassi and p. gambino, nucl. b * 567 * (2000) 3 ; m. awramik, m. czakon, a. freitas and g. weiglein, phys. d * 69 * (2004) 053006 ; phys. * 93 * (2004) 201805. h. burkhardt and b. pietrzyk, phys. d * 72 * (2005) 057501. e. brubaker _ [tevatron electroweak working group], arxiv : hep - ex/0608032. r. barate _ et al. _, b * 565 * (2003) 61. a. czarnecki, b. krause and w.j. marciano, phys. d * 52 * (1995) 2619. a. czarnecki, b. krause and w.j. marciano, phys. * 76 * (1996) 3267. kukhto, e.a. kuraev, z.k. silagadze, a. schiller, nucl. b * 371 * (1992) 567. t. kaneko and n. nakazawa, arxiv : hep - ph/9505278. s. heinemeyer, d. stckinger and g. weiglein, nucl. b * 699 * (2004) 103 ; + t. gribouk and a. czarnecki, phys. d * 72 * (2005) 53016. s. peris, m. perrottet and e. de rafael, phys. b * 355 * (1995) 523. m. knecht, s. peris, m. perrottet and e. de rafael, jhep * 0211 * (2002) 003. a. czarnecki, w.j. marciano and a. vainshtein, acta phys. b * 34 * (2003) 5669. a. vainshtein, phys. b * 569 * (2003) 187. a. czarnecki and b. krause, nucl. instrum. a * 389 * (1997) 314. kuraev, t.v. kukhto and a. schiller, yad. * 51 * (1990) 1631 (sov. j. nucl. * 51 * (1990) 1031). g. degrassi and g.f. giudice, phys. d * 58 * (1998) 053007. c. bouchiat and l. michel, j. phys. radium * 22 * (1961) 121 ; l. durand, phys. * 128 * (1962) 441 ; erratum - ibid. * 129 * (1963) 2835 ; s.j. brodsky and e. de rafael, phys. * 168 * (1968) 1620 ; m. gourdin and e. de rafael, nucl. b * 10 * (1969) 667. s. narison, j. phys. g * 4 * (1978) 1849. akhmetshin _ et al. _, phys. b * 551 * (2003) 27 ; r.r. akhmetshin _ et al. _, phys. b * 562 * (2003) 173 ; r.r. akhmetshin _ et al. _, phys. b * 580 * (2004) 119 ; r.r. akhmetshin _ et al. _, phys. b * 595 * (2004) 101 ; r.r. akhmetshin _ et al. _, phys. b * 605 * (2005) 26 ; v.m. aulchenko _ et al. _, jetp lett. * 82 * (2005) 743 (zhetf pisma * 82 * (2005) 841) ; r.r. akhmetshin _ et al. _, phys. b * 642 * (2006) 203 ; v.m. aulchenko _ et al. _, jetp lett. * 84 * (2006) 413 (zhetf pisma * 84 * (2006) 491). b. aubert _ et al. _, phys. d * 69 * (2004) 011103 ; b. aubert _ et al. _, phys. d * 70 * (2004) 072004 ; b. aubert _ et al. _, phys. d * 71 * (2005) 052001 ; b. aubert _ et al. _, phys. d * 73 * (2006) 052003. b. krause, phys. b * 390 * (1997) 392. m. knecht and a. nyffeler, phys. d * 65 * (2002) 73034 ; m. knecht, a. nyffeler, m. perrottet and e. de rafael, phys. * 88 * (2002) 71802 ; a. nyffeler, acta phys. b * 34 * (2003) 5197 ; m. hayakawa and t. kinoshita, phys. d * 57 * (1998) 465 ; d * 66 * (2002) 019902(e) ; j. bijnens, e. pallante and j. prades, nucl. b * 474 * (1996) 379 ; _ ibid. _ b * 626 * (2002) 410. k. melnikov and a. vainshtein, phys. d * 70 * (2004) 113006. j. erler and g.t. sanchez, phys. * 97 * (2006) 161801. pivovarov, phys. at. * 66 * (2003) 902 (yad * 66 * (2003) 934). gonzalez - sprinberg, a. santamaria and j. vidal, nucl. b * 582 * (2000) 3. p. achard _ et al. _, phys. b * 585 * (2004) 53 ; m. acciarri _ et al. _, phys. b * 434 * (1998) 169 ; k. ackerstaff _ et al. _, phys. b * 431 * (1998) 188 ; r. escribano and e. masso, phys. b * 395 * (1997) 369 ; _ ibid. _ b * 301 * (1993) 419 ; j.a. grifols and a. mendez, phys. b * 255 * (1991) 611 ; _ ibid. _ b * 259 * (1991) 512(e) ; f. del aguila, m. sher, phys. b * 252 *, 116 (1990) ; g. domokos, s. kovesi - domokos, c. vaz and d. wurmser, phys. rev. d * 32 * (1985) 247 ; d.j. silverman and g.l. shaw, phys. d * 27 * (1983) 1196. | this article reviews and updates the standard model prediction of the @xmath0 lepton @xmath1@xmath2@xmath3. updated qed and electroweak contributions
are presented, together with new values of the leading - order hadronic term, based on the recent low energy @xmath4 data from babar, cmd-2, kloe and snd, and of the hadronic light - by - light contribution.
the total prediction is confronted to the available experimental bounds on the @xmath0 lepton anomaly, and prospects for its future measurements are briefly discussed. | hep-ph0701260 |
Results
we begin with the definitions of the essential concepts of our model. * random geometric graph * consists of @xmath10 agents randomly distributed in a unit square @xmath11. each agent has an interaction range defined by @xmath12, where @xmath5 is the local interaction radius. two agents are connected if they fall in each others interaction range. the choice of network topology, denoted as @xmath13, impacts the boundary conditions. some studies, like @xcite, choose the natural topology of the unit square which leads to the free boundary condition. in this paper, we assume that @xmath13 is a torus, imposing the periodic boundary condition. consequently, the opinion dynamics is free of boundary effects until the correlation length of the opinions grows comparable to the length scale of @xmath13. * microstate * of a network is given by a spin vector @xmath14 where @xmath15 represents the opinion of the @xmath16 individual. in the ng, the spin value is assigned as follows : @xmath17 the evolution of microstate is given by spin updating rules : at each time step, two neighboring agents, a speaker @xmath18 and a listener @xmath19 are randomly selected, only the listener s state is changed (lo - ng). the word sent by the speaker @xmath18 is represented by @xmath20, @xmath21 if the word is a and @xmath22 if the word is b. @xmath20 is a random variable depending on @xmath15. the updating rule of the ng can be written as : @xmath23 * macrostate * is given by @xmath24, @xmath25 and @xmath26, the concentrations of agents at the location @xmath27 with opinion a, b and ab, respectively, that satisfy the normalization condition @xmath28. we define @xmath29 as the local order parameter (analogous to magnetization "), and @xmath30 as the local mean field @xmath31 finally, @xmath32 denotes the probability for an agent to receive a word a if it is located at @xmath27. through the geographic coarsening approach discussed in more detail in * methods *, we obtain the mean - field equation describing the evolution of macrostate @xmath33 while the macrostate itself is defined as @xmath34 there are two characteristic length scales in this system, one is the system size (which is set to @xmath35), the other is the local interaction radius @xmath5. so regarding the correlation length or the typical scale of opinion domains @xmath36, the dynamics can be divided into two stages : (1) @xmath36 is smaller or comparable to @xmath5 ; (2) @xmath37. in the second stage, the consensus is achieved when @xmath36 grows up to 1. figure [meanfield] present snapshots of solution of the mean - field equation. they illustrate the formation of opinion domains and the coarsening of the spatial structure. ). snapshots are taken at @xmath38, the scale of opinion domains are much bigger than @xmath39. black stands for opinion a, white stands for opinion b and gray stands for the coexistence of two types of opinions. the consensus is achieved at @xmath40.,scaledwidth=95.0%] to study the spatial coarsening, we consider the pair correlation function @xmath41 defined by the conditional expectation of spin correlation. @xmath42.\]] figure [meanfield] implies there exists a single characteristic length scale @xmath43 so that the pair correlation function has a scaling form @xmath44, where the scaling function @xmath45 does not depend on time explicitly. for coarsening in most systems with non - conserved order parameter such as the opinion dynamics on a d - dimensional lattice, the characteristic length scale is @xmath46@xcite. according to the numerical results in fig. [scaling], the length scale for opinion dynamics on rgg at the early stage (t=30,50) is also @xmath8, but at the late stage (t=100,200,400), the length scale @xmath47 fits more precisely simulation results than the previous one. for the pair correlation function at times @xmath48. overlapped curves indicate correct scaling of @xmath49. simulations are done for the case @xmath50, @xmath39, @xmath51. @xmath49 is normalized by the length scale (a) @xmath8 and (b) @xmath52.,scaledwidth=95.0%] here, we find all the possible stationary solutions of the mean - field equation eq. ([eqn_macro]). taking @xmath53, we obtain @xmath54 the eigenvalues of the linear dynamical system @xmath55 are both negative, so @xmath56 is stable. applying the definition of @xmath57 and @xmath30, we have @xmath58 @xmath59 once we solve the above integral equation, we can retrieve the stationary macrostate @xmath60 by eq. ([local_equil]). taking @xmath30 as a constant, we find three solutions @xmath61 or @xmath62. @xmath61 are both asymptotically stable, while @xmath63 is unstable. another class of solution is obtained by taking @xmath64 (or similarly @xmath65). the solution consists of an even number of stripe - like opinion domains demarcated by two types of straight intermediate layers parallel to one side of the unit square @xmath13 as shown in fig. [stationary_solution](b). with the boundary condition @xmath66 or vice versa, we solve the two types of intermediate layers @xmath67 as shown in fig. [stationary_solution] (a). the intermediate layers are of the scale @xmath5 and can be placed at arbitrary @xmath68. this class of solution is neutrally stable. finally, there is another class of solution shown in fig. [stationary_solution](c) with intermediate layers both in @xmath69 and @xmath70 directions and opinion domains assigned as a checker board. this type of solution is unstable at the intersections of two types of intermediate layers. the latter two classes of solutions can be easily generalized to the cases when the intermediate layers are not parallel to x or y axis. later we will show that in stationary solutions all the curvature of the opinion domain boundary has to be 0, so the solutions mentioned above are the only possible stationary solutions. . @xmath71 is the location of the intermediate layer. the slope of the intermediate layer at @xmath72 is about @xmath73. (b) stripe - like stationary solution, neutrally stable. (c) checker - board - like stationary solution, unstable.] in conclusion, considering the stability, the final state of the macrostate dynamics can be : (1) all a or all b consensus states which are both asymptotically stable, (2) stripe - like solution. the probability for the dynamics stuck in the stripe - like state before achieving full consensus is roughly @xmath74 in analogy to similar cases in continuum percolation and spin dynamics @xcite. one important observation regarding the macrostate dynamics is that the change of local mean field is usually much slower than the convergence of local macrostate @xmath75 to its local equilibrium @xmath56. let @xmath76, @xmath77, @xmath78. the following equation shows the exponential rate of convergence of @xmath75 : @xmath79 the largest eigenvalue is @xmath80. so the typical time scale @xmath81 of the convergence of the local macrostate is @xmath82 which is independent of time and system size. the typical time scale @xmath83 of the change of the local mean field is inversely proportional to the propagation speed @xmath84 of opinion domain boundaries, and as we will show later, is of the order @xmath85 where @xmath86 is the curvature of the opinion domain boundary. therefore, @xmath87 for both long time (@xmath86 grows to infinity along with the time @xmath6) and big systems (in the sense that @xmath88). [adiabatic] shows the significant two - time - scale separation observed in numerical results. the equilibrium value of the local order parameter, @xmath89, can be predicted by the local mean field, @xmath90. in fig. [adiabatic], we present the empirical local order parameter @xmath91 for different local mean field values @xmath92 and show that it is very close to its local equilibrium @xmath89. from numerical simulation and the equilibrium @xmath89 for different local mean fields @xmath92 s. the solid line stands for @xmath89. the error bars present the means and standard deviations of @xmath91 that is the empirical local order parameter for the given @xmath92 in numerical simulations.,title="fig:",scaledwidth=75.0%] + since @xmath93, we have @xmath94 \label{eq_s}.\]] this ode is quite relevant to reaction - diffusion systems. on the right hand side, the coefficient @xmath95 is easy to get rid of by scaling the time @xmath96. after the scaling, the first term is diffusive since @xmath97 is the continuum approximation of the laplace operator on rgg network acting on @xmath91. the second term @xmath98 is the local reaction term. though classified rigorously, it is non - local as defined in reaction - diffusion system, it represents a reaction in local neighborhood @xmath99. the adiabaticity of the dynamics implies that the diffusion is much slower than the local reaction. we can obtain an approximated ode for slow time scale dynamics in a closed form by estimating @xmath100 by its local equilibrium @xmath101. @xmath102 the qualitative behavior of the reaction - diffusion system is determined by the linear stability of the reaction term @xcite. in this sense, ([slow]) provides clear differentiation between dynamics in our model and in the voter model, the majority game and the glauber ordering. taking a similar approach, we find that the voter model on rgg is purely diffusive, i.e. the reaction term is 0. for the glauber ordering, the reaction term is @xmath103 in which @xmath104 is the inverse of temperature and @xmath105 is the interaction intensity. [reaction_term] shows the reaction term @xmath106 for the voter, ng, and glauber ordering (go) at different temperatures. the majority game, ng, and glauber ordering at zero temperature have reaction terms with the same equilibria and stability (@xmath107 stable, @xmath62 unstable). thus, the mean - field solutions of these models behave similarly. however, at the level of the discrete model, the ng on rgg will always go to a microstate corresponding to some stationary mean - field solution, while glauber ordering at zero temperature on rgg may get stuck in one of many local minima of its hamiltonian. for voter, ng and glauber ordering (go) at different temperatures.,title="fig:",scaledwidth=75.0%] + the evolution of the opinion domains is governed by a very simple rule. the boundary of opinion domains propagate at the speed @xmath84 that is proportional to its curvature @xmath108, i.e. @xmath109. here, @xmath110 is a constant defined by the average degree @xmath9. in * methods *, we provide a heuristic argument and using the perturbation method prove this relation for the mean - field equation, i.e. the case when @xmath111. this behavior is common for many reaction - diffusion systems and it is qualitatively the same as the behavior of glauber ordering at zero temperature @xcite. following the rule of boundary evolution minimizes the length of the domain boundary. a direct consequence of this fact is that if any stationary solution exists, its boundaries must be all straight (geodesic), confirming our conclusion about the stationary solutions found in the previous paragraph. since global topology is irrelevant to our derivation, this relation applies also to other two - dimensional manifolds. the manifold considered here is the torus embedded in 2d euclidean space. however, for the standard torus embedded in 3d euclidean space, the topology is the same but metrics are not, hence the geodesics are different. therefore, there are quite different and more complicated stationary solutions there. another example is the sphere in 3d space. on the sphere, the only inhomogeneous stationary solution consists of two hemispherical opinion domains, because the great circle is the only closed geodesic on a sphere. the numerical result presented in fig. [propagation_speed] confirms this relation. in fig. [propagation_speed], we gather @xmath112 data points from numerical solutions of the macrostate equation, using different initial conditions, taking snapshots at different times, tracking different points on the boundary and calculating the local curvature radius @xmath86 and boundary propagation speed @xmath84. these data points in the double - log plot are aligned well with the straight line with slope @xmath113. the curve formed by data points is jiggling with some period because we implemented the numerical method on a square lattice, so the numerical propagation speed is slightly anisotropic. of the boundary of opinion domains vs. curvature radius @xmath86. data points are gathered from 100 runs of macrostate equation with different initial conditions. in each run, propagation speed and curvature radius are calculated for 10000 points on the boundary.,title="fig:",scaledwidth=80.0%] + another way to confirm this rule is to consider a round opinion domain with initial radius @xmath114. given @xmath115, the size of this opinion domain decreases as @xmath116 when @xmath117. this relation is shown in fig. [shrink]. in simulations, the size of opinion domain @xmath118 is evaluated by @xmath119 where @xmath120,@xmath121 are the total numbers of a, ab nodes, respectively. we also observe from this plot that @xmath110 increases with average degree @xmath122 and converges to its upper bound when @xmath123. as a function of time with @xmath124. the initial radius of the opinion domain is @xmath125. the straight line is for the numerical solution of mean - field equation. the dotted line and dash line are for simulation of discrete model with @xmath126 and @xmath127, as well as with @xmath128 and @xmath129, respectively.,title="fig:",scaledwidth=80.0%] + we now consider influencing the consensus by committed agents. in sociological interpretation, a committed agent is one who keeps its opinion unchanged forever regardless of its interactions with other agents. the effect of committed agents in the ng has been studied in @xcite. a critical fraction of committed agent @xmath130 is found for ng - lo on complete graph which is also relevant here. in our setting, a fraction @xmath131 of agents are committed to opinion a, and all other agents are uncommitted and hold opinion b. the macrostate with committed agents is still defined by eq. ([eqn_macro]), but the definition of local mean field @xmath92 is replaced by @xmath132 generally, @xmath131 can vary on the x - y plane, but we only consider the case that @xmath131 is a constant, i.e. the committed agents are uniformly distributed. then we reanalyze the stationary solution. firstly there is a critical committed fraction @xmath133 which is exactly the same as that on complete graph. when @xmath134, the only stationary solution is @xmath135 and it is stable. when @xmath136, there are three solutions, of which two @xmath135, @xmath137 are stable, and the third @xmath138 is unstable. besides there is a class of stationary solutions when the committed fraction is below the critical. they are analogues of the stripe - like solutions in the non - committed agent case. the evolution of the boundary can be interpreted as a mean curvature flow. in such view, the fraction of agents committed to a opinion exert a constant pressure on the boundary surface from the side of a opinion domain. similarly, agents holding opinion b exert a pressure from the side of b opinion domain. so the stationary solution will contain opinion domains in the form of disks with critical radius @xmath139 for which the pressure arising from the curvature offsets the pressure from the committed fraction ; thus we have @xmath140. this type of stationary solutions are unstable, the round disk of the opinion domain will grow when @xmath141 and will shrink when @xmath142. in the first case, when the typical length scale of opinion domains grows beyond @xmath143, the system will achieve consensus very quickly. on the basis of the above stability analysis, we then analyze the dependence of the consensus time on the system size @xmath10, the committed fraction @xmath131, and the average degree @xmath122, and show our conclusions are consistent with the numerical results in fig. [committed2] which for a given fraction @xmath110 (@xmath144 in the figure) depicts the time for @xmath110-consensus in which at least fraction @xmath110 of agents hold the same opinion. the time to achieve @xmath145consensus is independent of @xmath10 both according to the mean field prediction and numerical plots. when @xmath134 the dynamics will converge to its unique local equilibrium @xmath146 at all locations simultaneously. the consensus in this case is close to that on the complete network, especially when @xmath122 tends to infinity. in the opposite case, when @xmath136, the process to consensus is twofold - before and after the a opinion domain achieves the critical size @xmath143. after this criticality, the process is just the extension of the opinion domain driven by the mean field eq. ([eqn_macro]). this stage is relatively fast and the consensus time is dominated by the duration of the other stage, the one before the criticality, in which the dynamic behavior is a joint effect of the mean field and the random fluctuation we neglect in mean field analysis. assuming the dynamics was purely driven by the random fluctuation, the typical length scale of opinion domains would have the scaling @xmath147 at the early stage, hence the time scale of this stage would be @xmath148, i.e. @xmath149. if the dynamics was purely driven by the mean field, the a opinion domain would never achieve the critical size. the actual dynamics behavior is in between the two extreme cases. when @xmath122 decreases, the fluctuation level relative to the mean field becomes higher, leading to faster consensus. in fig. [committed2], linear regression for the data points @xmath150 gives @xmath151 in which @xmath152 for @xmath153 respectively, where @xmath154 is the value corresponding to the purely random extreme case. two observations here may have meaningful sociological interpretation : \(1) when @xmath123, for both @xmath136 and @xmath134, the dynamics behavior converges to that on the complete networks, though the rgg itself may be far from the complete network (with @xmath5 kept constant, the diameter of the rgg network is much larger than 1). \(2) when @xmath136, committed agents are more powerful in changing the prevailing social opinion on rggs with low average degree. it is similar to the result of the previous study@xcite on the social dynamics on sparse random networks but the `` more powerful '' is in a different sense meaning not the smaller tipping fraction (with longer consensus time), but the shorter time to consensus. for @xmath155-consensus for ng on rgg with different fractions (@xmath131) of committed agents with direct simulation on networks with average degrees @xmath156, and network sizes @xmath157. the solid black curve shows the mean field prediction of ng on the complete network(cn) as the limit case when @xmath123. the slope of the red data points near @xmath133 is -2.19.,title="fig:",scaledwidth=80.0%] +
Discussion
on rggs, the average degree of an agent @xmath158 is an important structural parameter which also strongly impacts the local dynamic behavior. there are two critical values of @xmath122 : one is for the emergence of the giant component, @xmath159 @xcite ; the other one, @xmath160, only applicable for finite - size networks, is for the emergence of the giant component with all vertices belonging to it. in this paper, we only considered the case when @xmath122 is above the critical value @xmath161 so that the network is connected @xcite. we mainly focused on analyzing the mean - field equation of the ng dynamics on rgg. we predicted a number of behaviors, including the existence of metastable states, the two - time - scale separation, and the dependence of the boundary propagation speed on the boundary curvature. we demonstrated in detail that the evolution of the spatial domains for the two - word lo - ng is governed by coarsening dynamics, similar to the broader family of generalized voter - like models with intermediate states @xcite. however, there are still some behaviors that can not be explained by the mean - field equation, such as : (_ i _) in the large @xmath6 limit, the scaling of correlation length is not @xmath8 as on the 2-@xmath162 regular lattice, but @xmath163 ; (_ ii _) the propagation speed increases along with the average degree @xmath122 and its upper - bound is the mean field prediction, i.e., the @xmath164 case. so the major limitation of the mean - field equation derived from the geographic coarsening approach is that it neglects the fluctuation among replicas (see * methods *) and loses the information about @xmath122. the dependence of the dynamics on @xmath122 is left for further study.
Methods
first, we provide the equation for the evolution of microstate @xmath165. denote the probabilities for @xmath15 taking values @xmath35, @xmath62, @xmath113 as @xmath166, @xmath167 and @xmath168, respectively. the master equation for spin @xmath15 is given by @xmath169 where @xmath170 is the probability for the @xmath16 agent receiving a signal @xmath171, while @xmath172 is the local mean field defined as the average of the neighboring spins, @xmath173 where @xmath174 is the coordinate of the @xmath16 agent and @xmath175 is the degree of the @xmath16 agent. master equations for all spins together describe the evolution of microstate. the motivation for geographic coarsening comes from the fact that rgg is embedded in a geographic space, so we may skip the level of agents and relate the opinion states directly to the geographic coordinates. instead of taking into account the opinion of every agent, in geographic coarsening, we consider the concentration of agents with different opinions at a specific location. @xmath176, @xmath57, @xmath177 are continuously differentiable w.r.t. @xmath69, @xmath70 and @xmath6. we derive the equation of @xmath176 from eq. ([eqn_micro]) by taking limits, @xmath178 the limit is done as follows @xmath179 \;.\]] the coarsening based on the above limit is valid under either of the following two assumptions. the first is when the rgg is very dense (@xmath180) and @xmath181 keeps constant. the second is when we consider @xmath182 replicas of rgg with the same set of parameters, and the summation above is over all replicas. in addition, @xmath183 keeps constant. our derivation is actually based on the second assumption. however under the first assumption, the fluctuation is vanishing, and the dynamics behavior of a single run converges to the mean field result. we show here the qualitative property of the boundary evolution by a heuristic argument. we consider a solution with the form @xmath184 where @xmath185 and with boundary conditions @xmath186 and @xmath187. @xmath188 has an intermediate layer at @xmath86 as shown in fig. [stationary_solution](a), so near @xmath86, @xmath189. when @xmath190, using moving coordinate @xmath191 where @xmath192 is unit wave vector, @xmath193 is spatial coordinate and @xmath84 is the wave speed, eq. ([eq_s]) becomes @xmath194 here, @xmath92 can be approximated by @xmath195 then we make perturbation on eq. ([traveling_wave]) @xmath196, @xmath197, @xmath198 and so on, requiring @xmath199, and obtain the equation for @xmath200 @xmath201 at @xmath202, @xmath203, @xmath204, @xmath205 and @xmath206 is @xmath207, the above equation becomes @xmath208, so @xmath209
Acknowledgment
this work was supported in part by the army research laboratory under cooperative agreement number w911nf-09 - 2 - 0053, by the army research office grant nos w911nf-09 - 1 - 0254 and w911nf-12 - 1 - 0546, and by the office of naval research grant no. n00014 - 09 - 1 - 0607. the views and conclusions contained in this document are those of the authors and should not be interpreted as representing the official policies either expressed or implied of the army research laboratory or the u.s. government.
Author contributions
z.w., c.l., g.k. and b.k.s. designed the research ; z.w implemented and performed numerical experiments and simulations ; z.w., c.l., g.k. and b.k.s. analyzed data and discussed results ; z.w., g.k. and b.k.s. wrote and reviewed the manuscript.
Competing financial interests
the authors declare no competing financial interests. lu, q., korniss, g. & szymanski, b.k. naming games in spatially - embedded random networks. in _ proceedings of the 2006 american association for artificial intelligence fall symposium series, interaction and emergent phenomena in societies of agents _ (aaai press, menlo park, ca 2006), pp. 148155 ; arxiv : cs/0604075v3. | we investigate the two - word naming game on two - dimensional random geometric graphs. studying this model advances our understanding of the spatial distribution and propagation of opinions in social dynamics.
a main feature of this model is the spontaneous emergence of spatial structures called opinion domains which are geographic regions with clear boundaries within which all individuals share the same opinion.
we provide the mean - field equation for the underlying dynamics and discuss several properties of the equation such as the stationary solutions and two - time - scale separation. for the evolution of the opinion
domains we find that the opinion domain boundary propagates at a speed proportional to its curvature.
finally we investigate the impact of committed agents on opinion domains and find the scaling of consensus time.
@xmath0 department of mathematical sciences, rensselaer polytechnic institute, 110 8@xmath1 street, troy, ny, 12180 - 3590 usa + @xmath2 department of physics, applied physics and astronomy, rensselaer polytechnic institute, 110 8@xmath1 street, troy, ny, 12180 - 3590 usa + @xmath3 department of computer science, rensselaer polytechnic institute, 110 8@xmath1 street, troy, ny, 12180 - 3590 usa + @xmath4 e - mail : zhangw14@rpi.edu relevant features of social and opinion dynamics @xcite can be investigated by prototypical agent - based models such as the voter model @xcite, the naming game @xcite, or the majority model @xcite.
these models typically include a large number of individuals, each of which is assigned a state defined by the social opinions that it accepts and updates its state by interacting with its neighbors.
opinion dynamics driven by local communication on geographically embedded networks is of great interest to understanding the spatial distribution and propagation of opinions. in this paper
we investigate the naming game (ng) on random geometric graphs as a minimum model of this type.
we focus on the ng but will also compare it with other models of opinion dynamics.
the ng @xcite was originally introduced in the context of linguistics and spontaneous emergence of shared vocabulary among artificial agents @xcite to demonstrate how autonomous agents can achieve global agreement through pair - wise communications without a central coordinator. here
, we employ a special version of the ng, called the two - word @xcite listener - only naming game (lo - ng) @xcite. in this version of the ng
, each agent can either adopt one of the two different opinions a, b, or take the neutral stand represented by their union, ab. in each communication, a pair of neighboring agents are randomly chosen, the first one as the speaker and the second one as the listener.
the speaker holding a or b opinion will transmit its own opinion and the neutral speaker will transmit either a or b opinion with equal probability. the listener holding a or b opinion will become neutral when it hears an opinion different from its own and the neutral listener will adopt whatever it hears.
detailed instances are shown in supplementary table.
consensus formation in the original ng (and its variations) on various regular and complex networks have been studied @xcite. in particular, the spatial and temporal scaling properties
have been analyzed by direct simulations and scaling arguments in spatially - embedded regular and random (rgg) graphs @xcite.
these results indicated @xcite that the consensus formation in these systems is analogous to coarsening @xcite. in this paper, we further elucidate on the emerging coarsening dynamics in the two - word lo - ng on rgg by developing mean - field (or coarse - grained) equations for the evolution of opinions.
our method of relating microscopic dynamics to macroscopic behavior shares similar features with those leading to effective fokker - planck and langevin equations in a large class of opinion dynamic models (including generalized voter models with intermediate states) @xcite. a random geometric graph (rgg), also referred to as a spatial poisson or boolean graph, models spatial effects explicitly and therefore is of both technological and intellectual importance @xcite. in this model, each node is randomly assigned geographic coordinates and two nodes are connected if the distance between them is within the interaction radius @xmath5.
another type of network with geographic information is the regular lattice.
fundamental models for opinion dynamics on regular lattice has been intensively studied @xcite. in many aspects,
opinion dynamics behaves similarly on rggs and regular lattices with the same dimensionality, but in our study, we also observe several differences.
for example, the length scale of spatial coarsening for large @xmath6 is @xmath7 on rgg while it is @xmath8 on regular lattice.
more generally, concerning the spatial propagation of opinions in social systems or agreement dynamics in networks of artificial agents, random geometric graphs are more realistic for a number of reasons : (_ i _) rgg is isotropic (on average) while regular lattice is not ; (_ ii _) the average degree @xmath9 for an rgg can be set to an arbitrary positive number, instead of a small fixed number for the lattice ; (_ iii _) rggs closely capture the topology of random networks of short - range - connected spatially - embedded artificial agents, such as sensor networks.
an important feature of the ng which makes it different from other models of opinion dynamics, e.g., voter model, is the spontaneous emergence of clusters sharing the same opinion.
generally these opinion clusters are communities closely connected within the network.
this feature of the ng can be used to detect communities of social networks @xcite. for the ng on random geometric graph concerned in this paper, the clusters form a spatial structure to which we refer to as _ opinion domains _ and which are geographic regions in which all nodes share the same opinion.
a number of relevant properties of the ng on random geometric graph have been studied by direct individual - based simulations and discussed in @xcite, such as the scaling behavior of the consensus time and the distribution of opinion domain size.
in contrast to these previous works, here we develop a coarse - grained approach and focus on the spatial structure of the two - word lo - ng, such as the correlation length, shape, and propagation of the opinion domains. in this paper
, we provide the mean - field (or coarse - grained) equation for the ng dynamics on rgg. by analyzing the mean - field equation
, we list all possible stationary solutions and find that the ng may get stuck in stripe - like metastable states rather than achieve total consensus.
we find significant two - time - scale separation of the dynamics and retrieve the slow process governed by reaction - diffusion system. using this framework, we identify similarities and differences between ng and other relevant models of opinion dynamics, such as voter model, majority game and glauber ordering.
next, we present the governing rule of the opinion domain evolution, that in the late stage of dynamics, the propagation speed of the opinion domain boundary is proportional to its curvature.
thus, an opinion domain can be considered as a mean curvature flow making many results of the previous works applicable here @xcite.
finally we investigate the impact of committed agents. the critical fraction of committed agents found in the case of the ng on complete graph is also present here.
we discuss the dependence of the consensus time on the system size, the committed fraction and the average degree. | 1401.0115 |
Introduction
understanding the growth history of supermassive black holes (smbhs) is one of the fundamental issues in studies of galaxy formation and evolution. the intimate connection between smbhs and host galaxies is evidenced through empirical correlations between the masses of smbhs (m@xmath8) and the overall properties of the host galaxy spheroids (e.g., magorrian et al. 1998 ; ferraresse & merritt 2000 ; gebhardt et al. the cosmic evolution of these scaling relationships has been investigated in the literature, where a tentative evolution has been reported utilizing observational approaches (e.g., peng et al. 2006 ; woo et al. 2006, 2008 ; treu et al. 2007 ; merloni et al. 2010 ; bennert et al. 2010, 2011 ; hiner et al. 2012 ; canalizo et al. 2012). in order to provide better empirical constraints on the cosmic growth of smbhs and its connection to galaxy evolution, reliable m@xmath8 estimation at low and high redshifts is of paramount importance. the m@xmath8 can be determined for type 1 agn with the reverberation mapping (rm, peterson 1993) method or the single - epoch (se, wandel et al. 1999) method under the virial assumption : @xmath9, where @xmath10 is the gravitational constant. the size of the broad - line region (blr), @xmath11, can be directly measured from rm analysis (e.g., peterson et al 2004 ; bentz et al. 2009 ; denney et al. 2010 ; barth et al. 2011b ; grier et al. 2012) or indirectly estimated from the monochromatic agn luminosity measured from se spectra based on the empirical blr size - luminosity relation (kaspi et al. 2000, 2005 ; bentz et al. 2006, 2009, 2013). the line - of - sight velocity dispersion, @xmath12, of blr gas can be measured either from the broad emission line width in the rms spectrum (e.g., peterson et al. 2004) obtained from multi - epoch rm data or in the se spectra (e.g., park et al. 2012b), while the virial factor, @xmath13, is the dimensionless scale factor of order unity that depends on the geometry and kinematics of the blr. currently, an ensemble average, @xmath14, is determined empirically under the assumption that local active and inactive galaxies have the same @xmath15 relationship (e.g., onken et al. 2004 ; woo et al. 2010 ; graham et al. 2011 ; park et al. 2012a ; woo et al. 2013) and recalibrated to correct for the systematic difference of line widths in between the se and rms spectra (e.g., collin et al. 2006 ; park et al. 2012b). the rm method has been applied to a limited sample (@xmath16) to date, due to the practical difficulty of the extensive photometric and spectroscopic monitoring observations and the intrinsic difficulty of tracing the weak variability signal across very long time - lags for high - z, high - luminosity qsos. in contrast, the se method can be applied to any agn if a single spectrum is available, although this method is subject to various random and systematic uncertainties (see, e.g., vestergaard & peterson 2006, collin et al. 2006 ; mcgill et al. 2008 ; shen et al. 2008 ; denney et al. 2009, 2012 ; richards et al. 2011 ; park et al. 2012b). in the local universe, the se mass estimators based on the h@xmath4 line are well calibrated against the direct h@xmath4 rm results (e.g., mclure & jarvis 2002 ; vestergaard 2002 ; vestergaard & peterson 2006 ; collin et al. 2006 ; park et al. 2012b). for agns at higher redshift (@xmath17), rest - frame uv lines, i.e., or, are frequently used for m@xmath8 estimation since they are visible in the optical wavelength range. unfortunately the kinds of accurate calibration applied to h@xmath4-based se bh masses are difficult to achieve for the mass estimators based on the and lines, since the corresponding direct rm results are very few (see peterson et al. 2005 ; metzroth et al. 2006 ; kaspi et al. 2007). instead, se m@xmath8 based on these lines can be calibrated indirectly against either the most reliable h@xmath4 rm based masses (e.g., vestergaard & peterson 2006 ; wang et al. 2009 ; rafiee & hall 2011a) or the best calibrated h@xmath4 se masses (mcgill et al. 2008 ; shen & liu 2012, sl12 hereafter) under the assumption that the inferred m@xmath8 is the same whichever line is used for the estimation. while several studies demonstrated the consistency between based and h@xmath4 based masses (e.g., mclure & dunlop 2004 ; salviander et al. 2007 ; mcgill et al. 2008 ; shen et al. 2008 ; wang et al. 2009 ; rafiee & hall 2011a ; sl12), the reliability of utilizing the line is still controversial, since can be severely affected by non - virial motions, i.e., outflows and winds, and strong absorption (e.g, leighly & moore 2004 ; shen et al. 2008 ; richards et al. 2011 ; denney 2012). other related concerns for the line include the baldwin effect, the strong blueshift or asymmetry of the line profile, broad absorption features, and the possible presence of a narrow line component (see denney 2012 for discussions and interpretations of the issues). several studies have reported a poor correlation between and h@xmath4 line widths and a large scatter between and h@xmath4 based masses (e.g., baskin & laor 2005 ; netzer et al. 2007 ; sulentic et al. 2007 ; sl12 ; ho et al. 2012 ; trakhtenbrot & netzer 2012). on the other hand, other studies have shown a consistency between them and/or suggested additional calibrations for bringing and h@xmath4 based masses further into agreement. (e.g., vestergaard & peterson 2006 ; kelly & bechtold 2007 ; dietrich et al. 2009 ; greene et al. 2010 ; assef et al. 2011 ; denney 2012). given the practical importance of the line, which can be observed with optical spectrographs over a wide range of redshifts (@xmath18), in studying high - z agns, it is important and useful to calibrate the based m@xmath8 estimators. vestergaard & peterson 2006 (vp06 hereafter) have previously calibrated mass estimators against h@xmath4 rm masses, providing widely used m@xmath8 recipes. since then, however, the h@xmath4 rm sample has been expanded and many of rm masses have been updated based on various recent rm campaigns (e.g., bentz et al. 2009 ; denney et al. 2010 ; barth 2011a, b ; grier et al. 2012). at the same time, new uv data became available for the rm sample, substantially increasing the quality and quantity of available uv spectra for the rm sample. in this paper we present the new calibration of the based m@xmath8 estimators utilizing the highest quality uv spectra and the most updated rm sample. in section 2 we describe the sample of h@xmath4 reverberation mapped agn having available uv spectra. section 3 describes our detailed spectral analysis of the emission line complex to obtain the relevant luminosity and line width measurements necessary for estimating se m@xmath8. we provide the updated se m@xmath8 calibration in section 4 and conclude with a discussion and summary in section 5. we adopt the following cosmological parameters to calculate distances in this work : @xmath19 km s@xmath20 mpc@xmath20, @xmath21, and @xmath22.
Sample and data
for our analysis, we start with the reverberation mapped agn sample, which is considered as a calibration base with reliable mass estimates. to date, there are @xmath23 agns, for which h@xmath4 reverberation based masses are available (peterson et al. 2004 ; denney et al. 2006 ; bentz et al. 2009 ; denney et al. 2010 ; barth et al. 2011a, b ; grier et al. 2012). of those @xmath23 objects, we selected @xmath24 agns whose archival uv spectra are available from _ international ultraviolet explorer _ (_ iue _) or _ hubble space telescope _ (_ hst _) data archives. first, we collected all available uv spectra covering the spectral region from the public archives. if there were multiple spectra for a given individual object, either multiple epochs taken with the same instrument or from multiple instruments, we combined the spectra for each instrument by using a standard weighted average method to get the better signal - to - noise (s / n) ratio. at the same time, we tried to keep contemporaneity as far as possible. then we selected the best quality spectra for each object based on visual inspection and by setting a limiting s / n ratio of @xmath25 per pixel, which was measured in an emission - line free region of the continuum near 1450 or 1700 (see denney et al. 2013 for the s / n related issues). among these @xmath24 objects, we excluded four agns (i.e., ngc 3227, ngc 4151, pg 1411 + 442, pg 1700 + 518) because they are severely contaminated with absorption features. other 9 agns (i.e., mrk 79, mrk 110, mrk 142, mrk 1501, ngc 4253, ngc 4748, ngc 6814, pg 0844 + 349, pg 1617 + 175) were also excluded due to the low quality and unreliability of uv spectra. thus, our sample contains @xmath26 agns. table [tab : optdata] lists the agns in the sample and their properties. note that we adopt the updated virial factor, @xmath27 (park et al. 2012a ; woo et al. 2013). compared to the previous sample of vp06, one object, mrk 290, is newly included and seven objects (i.e., 3c 120, mrk 335, ngc 3516, ngc 4051, ngc 4593, ngc 5548, pg 2130 + 099) have updated reverberation m@xmath8 (denney et al. 2006, bentz et al. 2009 ; denney et al. 2010 ; grier et al. one object, pg 0804 + 761, that was excluded by vp06, is included since it has a new high - quality uv spectrum from the _ cosmic origins spectrograph _ (cos) aboard _ contrary to vp06, ngc 4151 is omitted in this work due to the strong absorption features near the line center (see section 3). in summary, @xmath28 agns have recent high - quality uv spectra from _ hst _ cos , @xmath29 km s@xmath30) by smoothing and re - binning by 7-pixels.] compared to vp06. for the remaining objects, uv spectra were obtained from the _ space telescope imaging spectrograph _ (stis) aboard _ hst _ for one object, from the _ faint object spectrograph _ (fos) aboard _ hst _ for eight objects, and from _ short - wavelength prime _ (swp) camera aboard _ iue _ for four objects as listed in table [tab : uvdata]. we corrected the galactic extinction using the values of @xmath31 from the recalibration of schlafly & finkbeiner (2011) listed in the nasa / ipac extragalactic database (ned) and the reddening curve of fitzpatrick (1999).
Spectral measurements
in order to calibrate the m@xmath8 estimator, we measured the line width of and the continuum luminosity at 1350 , following the multi - component fitting procedure developed by park et al. (2012b) with a modification for the region. we first fitted a single power - law continuum using the typical emission - line - free windows in both sides of (i.e., @xmath32 or @xmath33 and @xmath34 ), which were slightly adjusted for each spectrum to avoid the contaminating absorption and emission features. we did not subtract the iron emission, since it is generally too weak to constrain at least in our data sets, although we indeed tested the pseudocontinuum model by including the uv template from vestergaard & wilkes (2001). after subtracting the best - fit power - law continuum, we simultaneously fitted the complex region with the multi - component model consisting of a gaussian function for the] @xmath01486, a gaussian function for the @xmath01531whenever clearly seen, a gaussian function + a sixth - order gauss - hermite series for the @xmath01549, two gaussian functions for the @xmath01640, and a gaussian function for the] @xmath01663. note that we fitted the 1600 feature, which is contaminating the red wing of, with a broad component (cf. appendix a. in fine et al. 2010 ; marziani et al. 2010). in the fitting process, the centers of,],], and emission line components were fixed to be laboratory wavelengths. we suppressed some components in,],], and lines based on empirical tests with and without such components. narrow absorption features were excluded automatically in the calculation of @xmath35 statistics by masking out the 3 sigma outliers below the smoothed spectrum (cf. shen et al. 2011). strong broad absorption features around the line center were also masked out manually by setting exclusion windows from visual inspection. although it is still controversial whether or not to remove a narrow emission - line component from before measuring the width, we use the full line profile of, i.e., without subtracting a narrow emission - line component, in order to be consistent with other studies (vp06, shen et al. 2011, assef et al. 2011 ; ho et al. we measured the continuum luminosities at @xmath36 and @xmath37 from the power - law model and measured the line widths (fwhm and @xmath38) from the best - fit model (i.e., a gaussian function + a sixth - order gauss - hermite series) as shown in figure [fig : all_spec_hst]. the measured line widths were corrected for the instrumental resolution following the standard practice by subtracting the instrumental resolution from the measured velocity in quadrature. in figure 2, we explicitly show spectra and best - fit models for the objects showing absorption features at the center of. note that the fitting results are uncertain for these objects, in particular ngc 4051 (see section 4.1). to assess measurement uncertainties of the line width and continuum luminosity, we applied the monte carlo flux randomization method used by park et al. (2012b ; see also shen et al. 2011). using the @xmath39 realizations of resampled spectra made by randomly scattering flux values based on the flux errors, we fitted and measured the line width and continuum flux, and adopted the standard deviation of the distribution as the measurement uncertainties for individual objects as listed in table [tab : uvdata]. and @xmath37 , respectively. the bottom panel compares the fwhm and line dispersion (@xmath38), where both were measured from the full line profile. the ratio between fwhm and @xmath38 is close to one (dashed line), indicating the line profile is more peaky than gaussian (dotted line). objects with new uv spectra from the _ hst _ cos is denoted with red filled circles. average offset and 1@xmath38 scatter are given in the lower right corner in each panel., title="fig:",scaledwidth=45.0%] + and @xmath37 , respectively. the bottom panel compares the fwhm and line dispersion (@xmath38), where both were measured from the full line profile. the ratio between fwhm and @xmath38 is close to one (dashed line), indicating the line profile is more peaky than gaussian (dotted line). objects with new uv spectra from the _ hst _ cos is denoted with red filled circles. average offset and 1@xmath38 scatter are given in the lower right corner in each panel., title="fig:",scaledwidth=45.0%] figure [fig : uv_measurements] presents the continuum luminosities measured at @xmath36 and @xmath37 , respectively, which are commonly adopted for the m@xmath8 estimator. since they are almost identical, we choose to use @xmath3 for the mass estimator. the comparison between fwhm and @xmath38 of is plotted in the bottom panel of figure [fig : uv_measurements]. it shows on average, a one - to - one relation between fwhm and @xmath38, indicating that the profile is more peaky than a gaussian profile, although there is large scatter. scatter are given in the lower right corner in each panel., title="fig:",scaledwidth=43.0%] + scatter are given in the lower right corner in each panel., title="fig:",scaledwidth=43.0%] + scatter are given in the lower right corner in each panel., title="fig:",scaledwidth=43.0%] we compare our measurements with those in vp06 in figure [fig : comapre_uv_measure_vp06], using the common sample (23 out of 27 objects given in their table 2, except for mrk 79, mrk 110, ngc 4151, and pg 1617 + 175). since there are multiple measurements in vp06, we here show weighted average values of vp06 measurements for the purpose of comparison. for the comparison of @xmath3, there is 0.24 dex scatter, which may stem from a combination of the differences, e.g., adopted spectra and the galactic extinction correction, between our study and vp06. we used the combined single spectra with the best quality while vp06 used all available se spectra for each object. especially for the objects with the new _ hst _ cos spectra (red filled circles) observed in different epochs, there could be an intrinsic difference due to the variability. in the case of the galactic extinction correction, we utilized the recent values of @xmath31 listed in the ned taken from the schlafly & finkbeiner (2011) recalibration, while vp06 used the original values from schlegel et al. (1998). when comparing our @xmath40 measurements to those of vp06, a slight positive systematic trend seems to be present (middle panel of figure [fig : comapre_uv_measure_vp06]). the most likely origin of this trend is the difference in the adopted line width measurement methods between vp06 and this work. based on the investigation by fine et al. (2010), we modeled the complex region with multiple components and measured line dispersion from the de - blended line model profile, whereas vp06 measured line dispersion from the data without functional fits by limiting the line profile range to @xmath41 km s@xmath30 of the line center, regardless of the intrinsic line width of each civ profile. thus, the line dispersion measured by vp06 will be biased if line wings are extended much further than the fixed line limit (i.e., underestimation) or the wings are smaller than the fixed line limit (i.e., overestimation by including other features). we avoid these biases by de - blending the line from other lines using the multi - component fitting analysis and measuring the line widths from the best - fit models. since the line dispersion is more sensitive to the line wings than the line core, the decomposition and thus recovery of the line wing profile from contaminating lines is essential. the bottom panel of figure [fig : comapre_uv_measure_vp06] compares fwhm@xmath42, indicating on average consistency between vp06 and this work, except for a few outliers. this is because fwhm is less sensitive to the line wings than @xmath43, hence the difference in the measuring method does not generate significant difference in measurements. note that although fwhm is sensitive to the narrow - line component, both vp06 and our study used the full line profile without decomposing the broad and narrow components. instead, another source of discrepancy comes from the fact that vp06 measured the line width directly from the data while in this study the best - fit models were used for line width measurements. thus, there may be object - specific differences depending on how the absorptions above the `` half - maximum '' flux level were dealt with by vp06, and how well the functional fits represent the peak of the profile in our study.
Updating the calibration of the se @xmath44 estimator
-based se bh masses (top) and fwhm - based se bh masses (bottom) to the h@xmath4 rm - based bh masses. the new sample from the recent rm results is marked with a blue open square. the regressed parameters (@xmath45) with the uncertainty estimates are given in the upper left part in each panel., title="fig:",scaledwidth=45.0%] + -based se bh masses (top) and fwhm - based se bh masses (bottom) to the h@xmath4 rm - based bh masses. the new sample from the recent rm results is marked with a blue open square. the regressed parameters (@xmath45) with the uncertainty estimates are given in the upper left part in each panel., title="fig:",scaledwidth=45.0%] by adopting h@xmath4 rm - based masses as true m@xmath8 (see table 1), we calibrate the mass estimators by fitting @xmath46 ~=~ \alpha & + & \beta ~ \log\left(\frac{l_{1350}}{10^{44}~\rm erg~s^{-1}}\right) \nonumber\\ & ~+~ & \gamma ~ \log \left[\frac{\vardelta v(\textrm{{\ion{c}{4}}})}{1000~ \rm km~s^{-1}}\right]~,\end{aligned}\]] where @xmath3 is the monochromatic continuum luminosity at 1350 and @xmath47 is the line width of, either fwhm@xmath1 or @xmath2. we regress equation [eq : calibration] to determine the free parameters (@xmath45) using the ` fitexy ` estimator implemented in park et al. (2012a). note that this approach is different from that used by vp06, who adopted the luminosity slope from the size - luminosity relation and fixed the velocity slope to @xmath48. instead, this method is consistent with the recent approaches described by wang et al. (2009), rafiee & hall (2011b), and shen & liu (2012). because a non - linear dependence is often observed between the line widths of h@xmath4 and line (especially based on the fwhm ; see denney 2012 for a likely physical interpretation), leaving @xmath49, @xmath4, and @xmath50 as free parameters arguably results in a better statistical regression by accommodating the possible covariance between luminosity and line width. in figure [fig : recali3], we present the final best - fit calibration results for -based mass estimators by directly comparing the se masses with the h@xmath4 rm - based masses, using equation [eq : calibration]. the regression results with various conditions and the previous calibrations from the literature are listed in table [tab : calibration]. we adopt the regression results without ngc 4051, which is subject to the largest measurement errors among our sample since modeling the line of this object is highly uncertain due to the strong absorption at the center (see fig. [fig : obj_strabs]). in addition, the variability of ngc 4051 is expected to have a large amplitude since it is the lowest - luminosity object in our sample. thus, ngc 4051 can add large scatter to the regression and potentially skew the slope because there is only a single object at the low - mass regime. thus, excluding this object possibly will lead to less biased results in terms of sample selection and measurement uncertainties. we will present the results without ngc 4051 hereafter unless explicitly stated. the slope of the velocity term, when it is treated as a free parameter, is closer to the virial assumption (i.e., 2) for the @xmath43-based estimator, i.e., @xmath51 (@xmath52 if ngc 4051 is included) than for the fwhm@xmath42-based estimator, i.e., @xmath53 (@xmath54 if ngc 4051 is included). this reinforces the use of @xmath38 for characterizing the line width of, as suggested by denney (2012, 2013 ; see also peterson et al. 2004 and park et al. 2012b for the case of h@xmath4). the slope of the luminosity term (i.e., @xmath55 for @xmath38 ; @xmath56 for fwhm) is almost consistent to that of photoionization expectation (i.e., 0.5 ; bentz et al. 2006) within the uncertainty. this may indicates that asynchronism between h@xmath4 and measurements does not introduce a significant overall difference for our high - luminosity, high - quality calibration sample. in this calibration, we treat @xmath4 and @xmath50 as free parameters in addition to @xmath49. letting @xmath4 be a free parameter is required to reduce luminosity dependent systematics since we are dealing with non - contemporaneous h@xmath4 and measurements, which is expected to be not necessarily linear. in addition, it is currently questionable to directly adopt the size - luminosity relation (e.g., kaspi et al. 2007) for the estimator since it is based on such a small sample. relaxing the constraint of @xmath57 for fwhm@xmath42 can be corroborated by the investigation by denney (2012), which shows that there are severe biases in measuring fwhm@xmath42 due to the non - variable component and dependence on the line shape. these systematic uncertainties may be properly calibrated out by taking @xmath50 as a free parameter. in the case of @xmath43, however, a similar systematic does not seem to be present for @xmath43 (see denney 2012). even if we allow @xmath50 to be free, the regression slope for @xmath43 is consistent to the virial expectation (i.e., 2) within @xmath58 uncertainty, thus we opt to fixing @xmath50 to a value of 2, avoiding systematic uncertainties due to small number statistics or sample - specific systematics. thus, here we provide the best estimator for the -based m@xmath8 (see also, fig. [fig : recali3]) as @xmath59 & ~=~ & (6.71\pm0.07) \nonumber\\ & ~+~ & (0.50\pm0.07) ~ \log\left(\frac{l_{1350}}{10^{44}~\rm erg~s^{-1}}\right) \nonumber\\ & ~+~ & 2 ~ \log \left[\frac{\sigma (\textrm{{\ion{c}{4}}})}{1000~ \rm km~s^{-1}}\right]\end{aligned}\]] with the statistical scatter against rm masses of @xmath60 dex and @xmath61 & ~=~ & (7.48\pm0.24) \nonumber\\ & ~+~ & (0.52\pm0.09) ~ \log\left(\frac{l_{1350}}{10^{44}~\rm erg~s^{-1}}\right) \nonumber\\ & ~+~ & (0.56\pm0.48) ~ \log \left[\frac{\rm fwhm (\textrm{{\ion{c}{4}}})}{1000~ \rm km~s^{-1}}\right]\end{aligned}\]] with the statistical scatter against rm masses of @xmath62 dex. apart from the interpretation of values of zero point and slopes, it is worth noting that these estimators are the best calibrated ones to reproduce h@xmath4 rm masses as closely as possible for the current sample and data sets. -based (top) and fwhm - based (middle) m@xmath8 respectively calculated using the estimators in this study and the estimators of vp06. in the bottom panel, m@xmath8 calculated with the fwhm - based estimator from sl12 is compared to our mass estimates. average offset and 1@xmath38 scatter are given in the lower right corner in each panel., title="fig:",scaledwidth=43.0%] + -based (top) and fwhm - based (middle) m@xmath8 respectively calculated using the estimators in this study and the estimators of vp06. in the bottom panel, m@xmath8 calculated with the fwhm - based estimator from sl12 is compared to our mass estimates. average offset and 1@xmath38 scatter are given in the lower right corner in each panel., title="fig:",scaledwidth=43.0%] + -based (top) and fwhm - based (middle) m@xmath8 respectively calculated using the estimators in this study and the estimators of vp06. in the bottom panel, m@xmath8 calculated with the fwhm - based estimator from sl12 is compared to our mass estimates. average offset and 1@xmath38 scatter are given in the lower right corner in each panel., title="fig:",scaledwidth=43.0%] in figure [fig : comapre_vp06_sl12], we present the systematic difference of m@xmath8 estimates based on our new estimators (equations [eq : final_cal_sigma] and [eq : final_cal_fwhm]) compared to the previous estimators from vp06 and sl12, respectively, using line width and l@xmath63 measurements. the @xmath43-based m@xmath8 estimates show almost consistent results to vp06 with a slight offset of @xmath64, which is expected from a difference in the adopted values of the virial factor (i.e., @xmath27 here versus @xmath65 in vp06). in contrast, the comparison of fwhm@xmath42-based masses, respectively estimated with our recipe and with that of vp06, shows large scatter and a systematic trend that the vp06 recipe underestimates m@xmath8 in the low - mass regime and overestimates m@xmath8 in the high - mass regime, compared to our recipe. the bottom panel of figure [fig : comapre_vp06_sl12] shows that the sl12 recipe systematically overestimates m@xmath8 over the whole dynamic range of the sample (i.e., @xmath66 @xmath67). this is understandable because sl12 used the fwhm@xmath68-based m@xmath8 in vp06 as a fiducial one and recalibrate the estimator using their high - mass (@xmath69 @xmath67) qsos. thus, the calibration performed by sl12 in their limited dynamical range inherits the overestimation behavior of the vp06 recipe with respect to our recipe, and propagates it into the low mass regime with larger effect. in addition, sl12 subtracted a narrow component before measuring fwhm of, leading to an overestimated fwhm@xmath42, compared to vp06 and our methods. we note that a large dynamic range is necessary for better calibration and investigation of the biases, as pointed out by sl12. in order to explicitly compare the calibration methods used in here and vp06, we regress equation [eq : calibration] by fixing @xmath4 and/or @xmath50 with the adopted values in vp06 as listed in table [tab : calibration]. for the @xmath38-based mass estimator, we obtain almost same calibration result (@xmath70) to that of vp06 (@xmath71) using the sample including ngc 4051. when ngc 4051 is exclude, the zero points reduces slightly (@xmath72) and intrinsic scatter becomes smaller. it is interesting to see the consistency of the @xmath38-based calibration between our study and vp06, despite the systematic bias in @xmath40 measurements of vp06 as shown in section [compare_vp06]. we interpret this as follows. although there is a bias in the vp06 measurement method for @xmath38, due to their choice of line limits (i.e., @xmath41 km s@xmath30), their @xmath38-based m@xmath8 measurements serendipitously scatter evenly below and above the central point of the mass scale of the sample, consequently resulting in a similar zero point regardless of the bias in @xmath38 measurements. in the end the calibrations are very similar, however, the intrinsic scatter of our calibration is smaller than that of vp06, which demonstrates a general increase in accuracy of our @xmath38-based masses over those of vp06, advocating for our @xmath38 measurement prescription. to demonstrate the effect of our new estimators on m@xmath8 studies, we present in figure [fig : sdss_dr7_quasar] the m@xmath8 distribution of the sdss qso sample as a function of redshift, based on various mass estimators. these masses are calculated using the fwhm@xmath42 measurements from shen et al. (2011), who provides only fwhm measurements using sdss dr7 spectra. note that fwhm@xmath42-based m@xmath8 determined with our new estimator is on average smaller by @xmath73 dex than that calculated with the previous estimator by vp06, since the vp06 recipe tends to overestimate m@xmath8 in the high - mass regime as explained in section 4.2. in contrast, there is a smooth transition between -based masses estimated from the recipe of wang et al. (2009) and -based masses from our new calibration since both estimators are based on the same calibration scheme (see section [sec : calibration]). kelly & shen (2013) derived a predicted maximum m@xmath8 of broad - line qsos as a function of redshift (their figure 7), showing a slight trend that the maximum m@xmath8 was larger at higher redshift. however, this subtle trend may simply be a result of systematic overestimation of masses at high redshift because their mass determination was based on the vp06 recipe. adopting our new mass estimator may eliminate such a trend.
Discussion and conclusions
we investigated the calibration of m@xmath8 estimators based on the updated sample of @xmath26 agns, for which both h@xmath4 reverberation masses and uv archival spectra were available. the sample of agns with rm masses as well as the uv spectra including have been expanded and updated since the calibrations performed by vp06 ; it is therefore useful to revisit the calibration of the -based mass estimators to provide the most consistent virial m@xmath8 estimates using the emission line. major differences of the calibration method between vp06 and the current study is twofold. first, we derived line widths (i.e., line dispersion and fwhm) from the spectral fits by performing multi - component fitting on the complex region to accurately de - blend from other contaminating lines while vp06 measured the line width of civ directly from the spectra. when `` applying '' a se scaling relationship to calculate m@xmath8, it is important to use the same fitting and line width measurement prescriptions that were used in `` calibrating '' the scaling relationship because significant systematic differences can arise in m@xmath8estimates if different analysis and measurement techniques are utilized (e.g., assef et al. 2011, denney 2012, sl12, park et al. second, we treated the slope parameters (i.e., @xmath4 and @xmath50) in the virial m@xmath8 equation (i.e., eq. [eq : calibration]) as free parameters as in wang et al. (2009), which is particularly important for fwhm. we provided the best - fit calibrations for both @xmath43- and fwhm@xmath42-based mass estimators. while we presented a consistent estimator for the @xmath38-based masses to that of vp06, we obtained significantly different m@xmath8 estimator for the fwhm - based masses, presumably due to relaxing the constraint of the virial expectation (i.e., @xmath74) to mitigate the fwhm - dependent biases. we generally recommend to use the @xmath38-based mass estimator if the @xmath38 measurement is available, as it shows the better consistency with the virial relation, and @xmath38-based masses show a smaller scatter than the fwhm - based masses when compared to h@xmath4 rm - based masses. using @xmath43-based estimator is also preferred by denney (2012), who showed that the @xmath43 measured from mean spectra is the better tracer of the broad - line velocity field than the fwhm@xmath42 since fwhm of is much more affected by the non - variable core component. compared to the previously calibrated fwhm@xmath42 estimator by vp06, our new estimator shows a systematic trend as a function of mass. the vp06 recipe overestimates m@xmath8 in the high - mass regime (i.e., @xmath75) while it underestimate m@xmath8 in the low - mass regime (i.e., @xmath76), compared to masses based on our new estimator. this systematic discrepancy is due to a combination of effects, including difference in the rm sample and updated rm masses, newly available uv spectra, emission - line fitting method, and calibration method. for the sdss quasar sample (shen et al. 2011), we find that m@xmath8 estimates based on our new estimator are systematically smaller by @xmath73 dex than those based on the previous recipe of vp06. one of the main differences in calibrating the fwhm@xmath42-based mass estimator is that we fit the exponent of velocity (@xmath77) as in eq. 3, instead of adopting @xmath78 as in vp06. this provides effectively the same effect as adopting a varying virial factor. if a constant virial factor is used for mass determination, then fwhm - based masses will show systematic difference compared to @xmath38-based masses, since the fwhm/@xmath38 ratio has a broad distribution, while the fiducial rm masses are derived with @xmath38 measurements from rms spectra. to resolve this issue, collin et al. (2006), for example, introduced varying virial factors for the fwhm@xmath68-based mass estimator depending on the range of the line widths. in our case, relaxing the virial (fwhm@xmath79) requirement in calibrating fwhm@xmath42-based masses against @xmath80-based rm masses provides virtually the same effect as adopting a varying virial factor, thus resulting in better consistency with @xmath38-based masses. it also mitigates the bias caused by the contamination of the non - variable emission component, where fwhm@xmath42-based masses in objects with peaky (boxy) profiles are under- (over-) estimated with previous fwhm@xmath42-based mass estimators. the calibration of mass estimators provided in this study still suffers from a sample bias as in the case of vp06. the incompleteness or lack of low - mass objects (i.e., @xmath81) in the current sample will be resolved when new _ hst _ stis observations become available for six low - mass reverberation - mapped agns (go-12922, pi : woo). however, the extrapolation of this calibration to high - luminosity, higher - redshift agns more similar to the sdss sample can only be realized with the extension of the rm sample to this regime an endeavor that we strongly advocate. apart from the calibration analysis performed in this and previous studies, several schemes to correct for the -based masses have been suggested in the literature to reduce the large scatter between the -based masses and the h@xmath4-based masses. for example, assef et al. (2011) suggested a prescription to reduce mass residuals using the ratio of the rest - frame uv to optical continuum luminosities based on a sample of 12 lensed quasars. shen & liu (2012) reported a poor correlation with large scatter between fwhm@xmath68 and fwhm@xmath1 using a sample of 60 high - luminosity qsos, showing that some part of the scatter correlated with the blueshift of with respect to h@xmath4. they suggested a correction for the fwhm of and] lines as a function of the blueshift. recently, denney (2012) showed that the line profile consists of both non - variable and variable components based on the sample of seven agns with reverberation data, and concluded that this non - variable component is a main source of the large scatter of the se m@xmath8. they provided an empirical correction for the fwhm - based mass depending on the line shape as parameterized as the ratio of fwhm to the line dispersion. since the line region is more likely to be affected by non - virial motions such as outflows and winds than the lower ionization line region, such as h@xmath4 (e.g., shen et al. 2008 ; richards et al. 2011), aforementioned corrections are also important and worth investigating further with a larger sample with enlarged dynamic range. in general, correcting for possible systematic biases and providing accurate m@xmath8 estimates is crucial for studies of the cosmic evolution of bh population, particularly at high - redshifts (e.g., fine et al. 2006 ; shen et al. 2008, 2011, shen & kelly 2012 ; kelly & shen 2013). thus, it is important to ensure a reliable calibration at the high - mass end (@xmath82 @xmath67) since the mass estimators are most applicable to high - mass agns at high - redshift utilizing optical spectroscopic data from large agn surveys. note that the current rm sample used for calibrating mass estimators still suffers from the lack of high - luminosity agns, suggesting that the rm sample may not best represent the high - luminosity qsos at high - redshifts, i.e., sdss qsos. thus, obtaining direct reverberation mapping results for high - mass qsos will be even more useful (see kaspi et al. 2007 for tentative results), despite the practical observational challenges. such measurements will be used to better determine a reliable size - luminosity relation and to directly investigate non - varying component of. we thank the anonymous referee for constructive suggestions and charles danforth for helpful comments for the _ hst _ cos data co - addition. this work was supported by the national research foundation of korea (nrf) grant funded by the korea government (mest) (no. 2012 - 006087). has received funding from the people programme (marie curie actions) of the european union s seventh framework programme fp7/2007 - 2013/ under rea grant agreement no. 300553. lcccccc 3c120 & @xmath83 & @xmath84 & @xmath85 & @xmath86 & @xmath87 & 6 + 3c390.3 & @xmath88 & @xmath89 & @xmath90 & @xmath91 & @xmath92 & 1 + ark120 & @xmath93 & @xmath94 & @xmath95 & @xmath96 & @xmath97 & 1 + fairall9 & @xmath98 & @xmath99 & @xmath100 & @xmath101 & @xmath102 & 1 + mrk279 & @xmath103 & @xmath104 & @xmath105 & @xmath106 & @xmath107 & 1 + mrk290 & @xmath108 & @xmath109 & @xmath110 & @xmath111 & @xmath112 & 4 + mrk335 & @xmath113 & @xmath114 & @xmath115 & @xmath116 & @xmath117 & 6 + mrk509 & @xmath118 & @xmath119 & @xmath120 & @xmath121 & @xmath122 & 1 + mrk590 & @xmath123 & @xmath124 & @xmath125 & @xmath126 & @xmath127 & 1 + mrk817 & @xmath128 & @xmath129 & @xmath130 & @xmath131 & @xmath132 & 1 + ngc3516 & @xmath133 & @xmath134 & @xmath135 & @xmath136 & @xmath137 & 4 + ngc3783 & @xmath138 & @xmath139 & @xmath140 & @xmath141 & @xmath142 & 1 + ngc4051 & @xmath143 & @xmath144 & @xmath145 & @xmath146 & @xmath147 & 4 + ngc4593 & @xmath148 & @xmath149 & @xmath150 & @xmath151 & @xmath152 & 2 + ngc5548 & @xmath153 & @xmath154 & @xmath155 & @xmath156 & @xmath157 & 3, 5 + ngc7469 & @xmath158 & @xmath159 & @xmath160 & @xmath161 & @xmath162 & 1 + pg0026 + 129 & @xmath163 & @xmath164 & @xmath165 & @xmath166 & @xmath167 & 1 + pg0052 + 251 & @xmath168 & @xmath169 & @xmath170 & @xmath171 & @xmath172 & 1 + pg0804 + 761 & @xmath173 & @xmath174 & @xmath175 & @xmath176 & @xmath177 & 1 + pg0953 + 414 & @xmath178 & @xmath179 & @xmath180 & @xmath181 & @xmath182 & 1 + pg1226 + 023 & @xmath183 & @xmath184 & @xmath185 & @xmath186 & @xmath187 & 1 + pg1229 + 204 & @xmath188 & @xmath189 & @xmath190 & @xmath191 & @xmath192 & 1 + pg1307 + 085 & @xmath168 & @xmath193 & @xmath194 & @xmath195 & @xmath196 & 1 + pg1426 + 015 & @xmath197 & @xmath198 & @xmath199 & @xmath200 & @xmath201 & 1 + pg1613 + 658 & @xmath202 & @xmath203 & @xmath204 & @xmath205 & @xmath206 & 1 + pg2130 + 099 & @xmath207 & @xmath208 & @xmath209 & @xmath210 & @xmath211 & 6 [tab : optdata] lclcccccccc 3c120 & _ iue_/swp & 1994 - 02 - 19,27;1994 - 03 - 11 & @xmath212 & 0.263 & @xmath213 & @xmath214 & @xmath215 & @xmath216 & @xmath217 & + 3c390.3 & _ hst_/fos & 1996 - 03 - 31 & @xmath218 & 0.063 & @xmath219 & @xmath220 & @xmath221 & @xmath222 & @xmath223 & + ark120 & _ hst_/fos & 1995 - 07 - 29 & @xmath224 & 0.114 & @xmath225 & @xmath226 & @xmath227 & @xmath228 & @xmath229 & + fairall9 & _ hst_/fos & 1993 - 01 - 22 & @xmath230 & 0.023 & @xmath231 & @xmath232 & @xmath233 & @xmath234 & @xmath235 & + mrk279 & _ hst_/cos & 2011 - 06 - 27 & @xmath236 & 0.014 & @xmath237 & @xmath238 & @xmath239 & @xmath240 & @xmath241 & + mrk290 & _ hst_/cos & 2009 - 10 - 28 & @xmath230 & 0.014 & @xmath242 & @xmath243 & @xmath244 & @xmath245 & @xmath246 & + mrk335 & _ hst_/cos & 2009 - 10 - 31;2010 - 02 - 08 & @xmath247 & 0.032 & @xmath248 & @xmath249 & @xmath250 & @xmath251 & @xmath252 & + mrk509 & _ hst_/cos & 2009 - 12 - 10,11 & @xmath253 & 0.051 & @xmath254 & @xmath255 & @xmath256 & @xmath257 & @xmath258 & + mrk590 & _ iue_/swp & 1991 - 01 - 14 & @xmath224 & 0.033 & @xmath259 & @xmath260 & @xmath261 & @xmath262 & @xmath263 & + mrk817 & _ hst_/cos & 2009 - 08 - 04;2009 - 12 - 28 & @xmath264 & 0.006 & @xmath265 & @xmath266 & @xmath267 & @xmath268 & @xmath269 & + ngc3516 & _ hst_/cos & 2010 - 10 - 04;2011 - 01 - 22 & @xmath270 & 0.038 & @xmath271 & @xmath272 & @xmath273 & @xmath274 & @xmath275 & abs + ngc3783 & _ hst_/cos & 2011 - 05 - 26 & @xmath247 & 0.105 & @xmath276 & @xmath277 & @xmath278 & @xmath279 & @xmath280 & + ngc4051 & _ hst_/cos & 2009 - 12 - 11 & @xmath281 & 0.011 & @xmath282 & @xmath283 & @xmath284 & @xmath285 & @xmath286 & abs + ngc4593 & _ hst_/stis & 2002 - 06 - 23,24 & @xmath287 & 0.022 & @xmath288 & @xmath289 & @xmath290 & @xmath291 & @xmath292 & + ngc5548 & _ hst_/cos & 2011 - 06 - 16,17 & @xmath293 & 0.018 & @xmath294 & @xmath295 & @xmath296 & @xmath297 & @xmath298 & abs + ngc7469 & _ hst_/cos & 2010 - 10 - 16 & @xmath299 & 0.061 & @xmath300 & @xmath301 & @xmath302 & @xmath303 & @xmath304 & + pg0026 + 129 & _ hst_/fos & 1994 - 11 - 27 & @xmath305 & 0.063 & @xmath306 & @xmath307 & @xmath308 & @xmath309 & @xmath310 & + pg0052 + 251 & _ hst_/fos & 1993 - 07 - 22 & @xmath311 & 0.042 & @xmath312 & @xmath313 & @xmath314 & @xmath315 & @xmath316 & + pg0804 + 761 & _ hst_/cos & 2010 - 06 - 12 & @xmath317 & 0.031 & @xmath318 & @xmath319 & @xmath320 & @xmath321 & @xmath316 & + pg0953 + 414 & _ hst_/fos & 1991 - 06 - 18 & @xmath218 & 0.012 & @xmath322 & @xmath323 & @xmath324 & @xmath325 & @xmath325 & + pg1226 + 023 & _ hst_/fos & 1991 - 01 - 14,15 & @xmath326 & 0.018 & @xmath327 & @xmath328 & @xmath329 & @xmath330 & @xmath331 & + pg1229 + 204 & _ iue_/swp & 1982 - 05 - 01,02 & @xmath332 & 0.024 & @xmath333 & @xmath334 & @xmath335 & @xmath336 & @xmath337 & + pg1307 + 085 & _ hst_/fos & 1993 - 07 - 21 & @xmath338 & 0.030 & @xmath339 & @xmath340 & @xmath341 & @xmath342 & @xmath343 & + pg1426 + 015 & _ iue_/swp & 1985 - 03 - 01,02 & @xmath344 & 0.028 & @xmath345 & @xmath346 & @xmath347 & @xmath348 & @xmath349 & + pg1613 + 658 & _ hst_/cos & 2010 - 04 - 08,09,10 & @xmath350 & 0.023 & @xmath351 & @xmath352 & @xmath353 & @xmath354 & @xmath355 & + pg2130 + 099 & _ hst_/cos & 2010 - 10 - 28 & @xmath356 & 0.039 & @xmath357 & @xmath358 & @xmath359 & @xmath360 & @xmath361 & [tab : uvdata] lccccccc + + @xmath362 & @xmath363 & @xmath364 & @xmath48 & @xmath365 & & & vp06 + @xmath362 & @xmath366 & @xmath364 & @xmath48 & @xmath367 & & & vp06 + fwhm & @xmath368 & @xmath364 & @xmath48 & @xmath369 & & & vp06 + fwhm & @xmath370 & @xmath364 & @xmath48 & @xmath371 & & & vp06 + fwhm & @xmath372 & @xmath373 & @xmath374 & & @xmath375 & @xmath376 & sl12 + + + @xmath362 & @xmath377 & @xmath378 & @xmath379 & @xmath380 & @xmath381 & @xmath382 & + fwhm & @xmath383 & @xmath384 & @xmath385 & @xmath386 & @xmath381 & @xmath387 & + + + @xmath362 & @xmath388 & @xmath389 & @xmath48 & @xmath390 & @xmath381 & @xmath391 & + fwhm & @xmath392 & @xmath393 & @xmath48 & @xmath394 & @xmath381 & @xmath395 & + + + @xmath362 & @xmath396 & @xmath397 & @xmath398 & @xmath399 & @xmath381 & @xmath400 & + fwhm & @xmath401 & @xmath397 & @xmath402 & @xmath403 & @xmath381 & @xmath387 & + + + @xmath362 & @xmath404 & @xmath397 & @xmath48 & @xmath380 & @xmath381 & @xmath405 & + fwhm & @xmath406 & @xmath397 & @xmath48 & @xmath394 & @xmath381 & @xmath407 & + + + @xmath362 & @xmath408 & @xmath364 & @xmath48 & @xmath409 & @xmath381 & @xmath410 & + fwhm & @xmath411 & @xmath364 & @xmath48 & @xmath412 & @xmath381 & @xmath413 & + + + @xmath362 & @xmath414 & @xmath384 & @xmath415 & @xmath399 & @xmath381 & @xmath416 & + fwhm & @xmath417 & @xmath418 & @xmath419 & @xmath420 & @xmath381 & @xmath421 & best - fit + + + @xmath362 & @xmath422 & @xmath423 & @xmath424 & @xmath425 & @xmath381 & @xmath426 & best - fit + fwhm & @xmath427 & @xmath428 & @xmath48 & @xmath394 & @xmath381 & @xmath429 & + + + @xmath362 & @xmath430 & @xmath397 & @xmath431 & @xmath425 & @xmath381 & @xmath416 & + fwhm & @xmath432 & @xmath397 & @xmath433 & @xmath434 & @xmath381 & @xmath421 & + + + @xmath362 & @xmath435 & @xmath397 & @xmath48 & @xmath425 & @xmath381 & @xmath426 & + fwhm & @xmath436 & @xmath397 & @xmath48 & @xmath437 & @xmath381 & @xmath438 & + + + @xmath362 & @xmath439 & @xmath364 & @xmath48 & @xmath425 & @xmath381 & @xmath440 & + fwhm & @xmath441 & @xmath364 & @xmath48 & @xmath394 & @xmath381 & @xmath442 & [tab : calibration] | we present the single - epoch black hole mass estimators based on the @xmath01549 broad emission line, using the updated sample of the reverberation - mapped agns and high - quality uv spectra. by performing multi - component spectral fitting analysis
, we measure the line widths (fwhm@xmath1 and line dispersion, @xmath2) and the continuum luminosity at 1350 (@xmath3) to calibrate the -based mass estimators. by comparing with the h@xmath4 reverberation - based masses,
we provide new mass estimators with the best - fit relationships, i.e., @xmath5 and @xmath6.
the new -based mass estimators show significant mass - dependent systematic difference compared to the estimators commonly used in the literature. using the published sloan digital sky survey qso catalog
, we show that the black hole mass of high - redshift qsos decreases on average by @xmath7 dex if our recipe is adopted. | 1304.7281 |
[sec:intro]introduction
the experimental evidence of the theoretically predicted skyrmions in non - centrosymmetric compounds with dzyloshinskii - moriya interaction has intrigued many scientists over the last years.@xcite recently, the preparation of thin films of b20 type mnsi on silicon substrates @xcite has offered promising prospects with regard to possible applications in future spintronic devices. on the one hand mnsi films offer a variety of interesting magnetic phases and on the other hand they are easy to integrate into devices due to the use of silicon as substrate material being well established in technology. the benefit of thin films compared to bulk material is the existence of the skyrmion phase in an extended region of the magnetic phase diagram due to the uniaxial anisotropy.@xcite this pioneers new opportunities for data storage devices. the drawback using mnsi films is the low magnetic ordering temperature, which is considerably below liquid nitrogen temperature. therefore, it is the aim to find compounds with similar spin order at higher temperatures. a suitable candidate is the b20 compound mnge (bulk lattice constant of 4.795) with a magnetic ordering temperature @xmath0 of 170k.@xcite the magnetic ground state of mnge is a helical spin structure with a helix length between 3 nm at lowest temperatures and 6 nm near @xmath0.@xcite the helix axis is due to magnetic anisotropy pinned along @xmath1001@xmath2,@xcite but rotates into field direction in an applied field. recently, a large topological hall effect exceeding by 40 times that of mnsi was observed in mnge which was attributed to a skyrmion phase in analogy to mnsi.@xcite. further evidence for the existence of skyrmions was given by small angle neutron scattering experiments.@xcite unfortunately, the synthesis of mnge is considerably laborious, since it forms only under high pressure and temperatures between 600 and 1000@xmath3c.@xcite however, molecular beam epitaxy (mbe) allows for thin film growth under strong non - equilibrium conditions. nevertheless, there has been no successfull attempt to grow mnge on ge, since mn and ge tends to form mn@xmath4ge@xmath5.@xcite the use of si(111) as substrate offers the opportunity to prepare a seedlayer of mnsi, which realizes the b20 crystal structure for mnge growth. the lattice constant of mnge within the (111) plane matches that of si with a misfit of only 2@xmath6, thus, compressively strained mnge films may be grown on si(111) substrates. in this paper we show a preparation method for mnge thin films on si substrates with the aid of a mnsi seedlayer. the structure and morphology of the films have been investigated by reflection high - energy electron diffraction (rheed), atomic force microscopy (afm) and x - ray diffraction (xrd). to determine the physical properties of the samples magnetization and magnetoresistance measurements have been performed.
Film growth and structural characterization
for the growth of mnge thin films p - doped si(111) substrates were used, which possess a resistivity between 1 and 10 @xmath7 cm at room temperature. prior to film deposition the substrates were heated to 1100@xmath3c under uhv conditions in order to remove the oxide layer and to achieve a clean and flat surface with 7@xmath87-reconstruction, which was verified by in - situ rheed investigations. the depostion of mn and ge directly on the si(111) surface does not produce b20 mnge films but results in a mn@xmath4ge@xmath5 layer. in order to establish the b20 crystal structure a 5mn layer was deposited onto the si surface and heated to 300@xmath3c subsequently. this procedure provides for the formation of a thin mnsi seedlayer with a thickness of 10. in a second step, mnge is codeposited by simultanoeus evaporation of mn and ge from an effusion cell and an electron beam evaporator, respectively. during film growth with a rate of 0.15 / s the substrate is held at a temperature of 250@xmath3c. $] crystal direction and b) line scans across the rheed streaks for the mnge film in comparison with the si substrate. the scans were taken parallel to the shadow edge.,scaledwidth=35.0%] the mnge films have been investigated by in - situ rheed in order to determine their structure and morphology. the rheed pattern of a 135 mnge film observed along the @xmath9 $] direction of the si substrate indicates two - dimensional film growth [fig. [fig : rheed](a)]. the arrangement of the streaks is very similar to the pattern of mnsi thin films,@xcite and suggests that mnge sustains the b20 crystal structure provided by mnsi seedlayer. the uniformity of the intensity of the detected streaks implies a flat surface of a size comparable to the area contributing to the rheed pattern of around 100 nm in diameter. line scans across rheed patterns of a 135 mnge film [fig. [fig : rheed](b)] compared to the si substrate reveal a nearly pseudomorphic growth of the mnge layer. however, a small deviance of the mnge streaks from the corresponding si reflections indicates that the mnge lattice has at least partly relaxed from the compressive strain imposed by the substrate. + afm images of films with thicknesses of 45, 90 and 135 give evidence that island growth of vollmer - weber type is the predominant growth mode [fig. [fig : afm]]. the thinnest film of 45 thickness [fig. [fig : afm](a)] consists of islands with a typical diameter of 100 nm separated by valleys of similar size. with increasing film thickness the islands are enlarged and gradually fill the space between them. for the 135 film only very thin valleys of a few nm can be observed [fig. [fig : afm](c)], and the morphology has transformed into elongated islands with a length of up to 2@xmath10 m and a width of around 200 nm. $] and @xmath11 $] crystal directions. inset : intensity plot along the @xmath12 $] direction.,scaledwidth=40.0%] x - ray diffraction measurements were performed using synchrotron radiation with @xmath13 at the swiss - norwegian beamline bm1a of the esrf (grenoble, france) with the pilatus@snbl diffractometer. the investigation of the 135 film confirms the b20 crystal structure of the mnge. in fig. [fig : xray] the (111) and (333) peaks of the si substrate and the mnge thin film are clearly resolved as single crystal peaks. the inset shows an integrated diffraction pattern along the [111] direction. from the position of the mnge(111) peak the lattice parameter of the mnge film of (4.750 @xmath14 0.004) is obtained, which is 1% smaller than the value for bulk mnge due to compressive strain.
Magnetic measurements
the magnetic characterization of the mnge films was carried out using a quantum design mpms-5s squid magnetometer. for films of different thickness the temperature dependence of the magnetic susceptibility has been measured in the range from 5k to 300k in an applied magnetic field of 10mt [fig. [fig : susceptibility]]. below 40k the susceptibility slightly increases due to the mnsi seedlayer, that orders magnetically in this temperature range. the measurements were normalized with respect to the saturation magnetization of the mnsi seedlayer, because this layer is the same for all three films. the susceptibility of mnge films exhibits an ordering temperature of @xmath15k indicated by a broad peak. regarding mnge bulk material, the susceptibility shows a qualitatively similar behavior with a lower @xmath16k.@xcite an enhancement of the ordering temperature has also been observed for mnsi thin films.@xcite in contrast to mnsi thin films, no thickness dependence of @xmath0 can be detected for films between 45 and 135. possibly, the spin - spin correlation length is shorter than the value for mnsi films (7 monolayers) @xcite, i.e. the thickness dependence may occur for mnge when the films are thinner than investigated in this work. both materials belong to the b20 compounds, which possess a helical spin structure, since their magnetic properties are governed by the interplay of ferromagnetic exchange and dzyaloshinskii - moriya interactions. nevertheless, the susceptibility of mnge shows a behavior which is typical for antiferromagnetic order, whereas for mnsi an increase to a constant value of magnetization towards low temperature is observed. the helix length in mnsi is very long (18 nm) and, thus, the local magnetic structure is related to ferromagnetism. a small field easily deforms the soft helix and induces a net magnetization. in the case of mnge the helix is more rigid. therefore, at low temperature no net magnetic moment is induced by a field as small as 10mt. the helix wavelength is connected to the dzyaloshinskii constant @xmath17 via @xmath18, where @xmath19 is the magnetic stiffness.@xcite since the helix in mnge is extremely short (@xmath20nm)@xcite the dzyaloshinskii constant is expected to be large, and the magnetic structure is very close to an antiferromagnet. field dependent magnetization measurements at @xmath21k were carried out on the same three samples as in fig. [fig : susceptibility]. for all samples the magnetization increases in fields up to 1 t [fig. [fig : magnetization]]. the inset of fig. [fig : magnetization] shows a magnetization measurement on the 135 film in fields up to 5 t, which reveals that saturation is reached around 1 t. this is in agreement with measurements performed on bulk mnge at temperatures close to @xmath0.@xcite since the helix length is much shorter than the size of the mnge islands, the magnetization behavior is not expected to be different from the bulk. the magnetic moment per mn atom was calculated assuming that the complete amount of mn deposited during growth has reacted to mnge. however, since the magnetic moment is only half of the bulk value, some part of the deposited mn did not form mnge. furthermore, we observe an apparently larger magnetic moment for thicker films. this can be explained by the fact that especially in the beginning of mnge growth not every mn atom is incorporated into the mnge crystal. evidence for this is also given by magnetization measurements at 5k, where mainly the ordering of the mnsi seedlayer is observed, since mnge only gives a linear contribution in the considered field range. we observe a moment of the mnsi layer that is about twice as large as expected for 10 mnsi. thereby, we conclude that some part of the deposited mn has reacted with si from the substrate to form mnsi. +
Magnetoresistivity
resistivity measurements were performed on the 135 film using the van - der - pauw method. the sample was found to be metallic, and the residual resistivity at 3k was determined as 83 @xmath22 cm. the field dependence of the resistivity was measured in magnetic fields up to 5 t for several temperatures. in fig. [fig : mr] three curves are depicted, which represent the magnetoresistivity defined by @xmath23. the data were obtained at temperatures between 60k and 100k, where the sample is in a magnetically ordered state. the mr effect is negative for all temperatures and fields and exhibits no remarkable features in the investigated field range. for comparison the equivalent data for a 19 nm film mnsi is shown in the inset of fig. [fig : mr]. in the case of mnsi the critical magnetic field, where the spins align ferromagnetically, occurs around 1 t. at this field a clear kink accompanied by a change in curvature is observed. furthermore the size of the mr effect is considerably larger for mnsi. regarding mnge the absence of magnetic phase transitions in moderate magnetic fields and the smallness of the magnetoresistivity evidences that the helical structure is more rigid than in mnsi. as discussed in the previous paragraph this can be ascribed to a stronger dzyaloshinskii - moriya interaction.
Conclusion
in this work we have proved that we succeeded in growing crystalline mnge as a thin film on a si(111) substrate. the film adopts the b20 structure from a thin seedlayer of mnsi prepared prior to mnge growth. morphological investigations using rheed and afm give evidence that the mnge thin films consist of islands with a flat surface, which enlarge during growth. the b20 structure was confirmed by xrd and the lattice parameter was determined to be 1% smaller than in bulk mnge due to compressive strain imposed by the si substrate. + although the magnetic properties of mnge thin films are found to be qualitatively similar to bulk, the ordering temperature is enhanced to 200k. in magnetoresistivity measurements no critical fields were observed up to 5 t. compared to mnsi, the helix in mnge is shorter and more rigid than in mnsi. therefore, the magnetic structure is related to antiferromagnetism rather than to ferromagnetism. we would like to thank dmitry chernyshov for his support with the x - ray measurements at the european synchrotron radiation facility. the afm measurements were performed at the institute of semiconductor technology in braunschweig. we thank alexander wagner for his help with the equipment. 99 s. mhlbauer, b. binz, f. jonietz, c. pfleiderer, a. rosch, a. neubauer, r. georgii, and p. bni, science * 323 *, 915 (2009). a. neubauer, c. pfleiderer, b. binz, a. rosch, r. ritz, p. g. niklowitz, and p. bni, phys. * 102 *, 186602 (2009). u. k. rler, a. n. bogdanov, and c. pfleiderer, nature * 442 *, 797 (2006). j. engelke, t. reimann, l. hoffmann, s. gass, d. menzel, and s. sllow, j. phys.. jpn. * 81 *, 124709(2012). e. a. karhu, u. k. rler, a. n. bogdanov, s. kahwaji, b. j. kirby, h. fritzsche, m. d. robertson, c. f. majkrzak, and t. l. monchesky, phys. b * 85 *, 094429 (2012). m. n. wilson, e. a. karhu, a. s. quingley, u. k. rler, a. b. butenko, a. n. bogdanov, m. d. robertson, and t. l. monchesky, phys. b * 86 *, 144420 (2012). a. b. butenko, a. a. leonov, u. k. rler, and a. n. bogdanov, phys. rev. b * 82 *, 052403 (2010). y. li, n. kanazawa, x. z. yu, a. tsukazaki, m. kawasaki, m. ichikawa, x. f. jin, f. kagawa, and y. tokura, phys. * 110 *, 117202 (2013). h. takizawa, t. sato, t. endo, and m. shimada, j. solid state chem. * 73 *, 40 (1988). n. kanazawa, y. onose, t. arima, d. okuyama, k. ohoyama, s. wakimoto, k. kakurai, s. ishiwata, and y. tokura, phys. lett. * 106 *, 156603 (2011). o. l. makarova, a. v. tsvyashchenko, g. andre, f. porcher, l. n. fomicheva, n. rey, and i. mirebeau, phys. rev. b. * 85 *, 205205 (2012). n. kanazawa, j.- h. kim, d. s. inosov, j. s. white, n. egetenmeyer, j. l. gavilano, s. ishiwata, y. onose, t. arima, b. keimer, and y. tokura, phys. b * 86 *, 134425 (2012). s. olive - mendez, a. spiesser, l. a. michez, v. le thanh, a. glachant, j. derrien, t. devillers, a. barski, and m. jamet, thin solid films * 517 *, 191 (2008). r. gunnella, l. morresi, n. pinto, r. murri, l. ottaviano, m. passacantando, f. dorazio, and f. lucari, surf. sci. * 577 *, 22 (2005). e. karhu, s. kahwaji, and t. l. monchesky, phys. rev. b * 82 *, 184417 (2010). | mnge has been grown as a thin film on si(111) substrates by molecular beam epitaxy.
a 10 layer of mnsi was used as seedlayer in order to establish the b20 crystal structure.
films of a thickness between 45 and 135 have been prepared and structually characterized by rheed, afm and xrd.
these techniques give evidence that mnge forms in the cubic b20 crystal structure as islands exhibiting a very smooth surface.
the islands become larger with increasing film thickness.
a magnetic characterization reveals that the ordering temperature of mnge thin films is enhanced compared to bulk material.
the properties of the helical magnetic structure obtained from magnetization and magnetoresistivity measurements are compared with films of the related compound mnsi. the much larger dzyaloshinskii - moriya interaction in mnge results in a higher rigidness of the spin helix. | 1308.4514 |
Subsets and Splits