dedup-isc-ft-v107-score
float64 0.32
1
| uid
stringlengths 32
32
| text
stringlengths 0
32.5k
| paper_id
stringlengths 1
14
| original_image_filename
stringlengths 5
222
|
|---|---|---|---|---|
0.455394 | fde2a631e14d4c029f2b3f11305fd7e9 | U,I,V time evolution for the untreated case. As stated in Theorem <ref>, ť_V < t̂_I < t_c < t̂_V, where ť_V and t̂_V are the times at which V(t) reaches a local minimum and a local maximum, respectively, t̂_I is the time at which I(t) reaches a local maximum, and t_c is the time at which U(t) reaches . Furthermore, as stated in Remark <ref>, ť_V ≈ 0 and t̂_I ≈ t_c ≈t̂_V. V_undet = 50 (Copies/mL) stands for the undetectable level of the virus load. | 2012.15585 | allpatOL2.jpg |
0.42368 | 322bb701d8b3494cabb278edeec844a8 | ℋ^c(t_tr) is given by the yellow and the blue regions, considering the following system parameters: β=0.5, δ=0.2, p=2 and c=5, with U(t_tr)=3. The critical boundary of ℋ^c(t_tr) represents the critical pairs of η_β and η_p. | 2012.15585 | effectivnness_both_antivirals.jpg |
0.529544 | 15de974d9c7e4640962fe4e19556a8a0 | η_p^c(t_tr) vs t_tr corresponding to the nine patients simulated in Section <ref> | 2012.15585 | etas_c.jpg |
0.393197 | 4e265be668fd476499deedc32ae7665d | Qualitative plot of function f(V) (arbitrary parameters) for different values of ℛ. As it can be seen - independently of the parameter values - if (t_tr)>1 it there exists an interval of values of V, and a corresponding period of time such that f(V)>V. | 2012.15585 | f_V.jpg |
0.530341 | 6a9fd0df19114f878d2762adfdd2575a | Infection-related metrics as function of η_p (t_tr=0.7t^e), for antiviral effectiveness assessment. | 2012.15585 | index_allpatCL_07_te.jpg |
0.516016 | b7a87da3f5dc4ad1a400739b910ca2ac | Infection-related metrics as function of η_p (t_tr=t_DL), for antiviral effectiveness assessment. | 2012.15585 | index_allpatCL_DL.jpg |
0.534723 | 6db3053b91c54dc5b11055cf3f594aee | Infection-related metrics as function of η_p (t_tr=t^e), for antiviral effectiveness assessment. | 2012.15585 | index_allpatCL_te.jpg |
0.504777 | 54f35074348241df88fd562b9d68d346 | Infection-related metrics for antiviral effectiveness assessment as function of η_p and η_β (t_tr=0.7t^e). Patient A | 2012.15585 | index_patACL_07_te.jpg |
0.398407 | 8407e577d5fc44a880f3a8cc21e4ec87 | Time evolution of U(t), I(t), V(t), and (t) corresponding to system (<ref>), with parameters β=0.5, δ=0.2, p=2 and c=5 and initial conditions U_0=3,I_0=0,V_0=0.1 (first row) and U_0=0.5,I_0=0,V_0=0.1 (second row). The firs row corresponds to the case >1+α(0), while the second one corresponds to the case <1+α(0) (with α(0) being approximately zero). In the first case, the minimum of V(t) occurs at ť_V=0.85, the maximum of I occurs at t̂_I=14.86, U(t) crosses 1 at t_c=14.87 and the maximum of V(t) occurs at t̂_V=15.08, which confirms both, the results in Theorem <ref> and Remark <ref>. In the second case, both I(t) and V(t) are strictly decreasing and U(t) remains almost constants, which confirms the results in Theorem <ref> and Remark <ref>. | 2012.15585 | main.jpg |
0.426034 | a4af9550f8a1488ea9ee540b269bcade | Virus time evolution for different treatment times, t_tr=4,6,8,17,20,25 (dpi). Two fixed values of η_p were used, smaller and bigger than η_p^c(t_tr): η_p=0.73 (left) and η_p=0.9 (right), respectively. η_p^c(t_tr)≈ 0.81 for t_tr<t^e. Patient B. | 2012.15585 | patBCL_tr.jpg |
0.409629 | cc7a279c0dd843749fa6e309327fcf8b | Virus time evolution for different treatment times, t_tr=4,6,8,17,20,25 (dpi). Two fixed values of η_p were used, smaller and bigger than η_p^c(t_tr): η_p=0.54 (left) and η_p=0.8 (right), respective. η_p^c(t_tr)≈ 0.6 for t_tr<t^e. Patient E. | 2012.15585 | patECL_tr.jpg |
0.545516 | b39ca3e7780644dd8bc28b17972d9868 | Instrumented cable principle | 2012.15586 | figure1.jpg |
0.488742 | 4d15c6a8389343acad28f89ec32a40d8 | Medium CDPR congestion | 2012.15586 | figure1e11.jpg |
0.424216 | c164f0fe8621425b93a8ad0ea2fc0a66 | 2^nd example Δρ(t) | 2012.15586 | figure1e21.jpg |
0.443938 | aa77fc37a0014547a35ba76e8d9cf9d7 | 3^rd example Δρ(t) | 2012.15586 | figure1e31.jpg |
0.431651 | 4f79273135a549669df50f8cd15922ae | 3^rd example Δρ(t) (after changement) | 2012.15586 | figure1e32.jpg |
0.491209 | 8d644dbb87c94b458908a15b0c7c32ca | Instrumented cable | 2012.15586 | figure2.jpg |
0.406111 | d4a87b280fe84fc5a6b07d7a5df0db41 | 1^st example Δρ(t) | 2012.15586 | figure2e11.jpg |
0.511604 | 6a7e311cf413470e98bf6a4c4283e5a3 | Configuration during passage of mark M_i by sensor S_j | 2012.15586 | figure3.jpg |
0.421661 | 5f332cd580924b57bada86f219ddcf1e | Placement of marks on cable | 2012.15586 | figure41.jpg |
0.3999 | 81a41dd54717466dafb912d47340cb50 | Placement of sensors on support | 2012.15586 | figure42.jpg |
0.513223 | 470f1a4b8ea34182b458b6412d82ed6e | Autocalibration flowchart | 2012.15586 | figure5.jpg |
0.386737 | ecc44cea2327483abff49a63ee1104e6 | Closed-loop autocalibration approach | 2012.15586 | figure6.jpg |
0.442722 | 02e32bdf62c848d39e19021386b05b42 | Calibration stroke | 2012.15586 | figure7.jpg |
0.448525 | 18aa69b686b84b52badbc2ac1879e59e | Autocalibration example Δρ(t) | 2012.15586 | figureae1.jpg |
0.463713 | 21118c9382af46ca9703101f15142ecb | Autocalibration scenario (1) | 2012.15586 | figureae2.jpg |
0.476879 | 3911744dd6154c0f93bde631f9f7a08b | Autocalibration scenario (2) | 2012.15586 | figureae3.jpg |
0.486431 | f0d95a0a4f484b829f01222337ebbbfb | Autocalibration scenario (3) | 2012.15586 | figureae4.jpg |
0.429988 | ae92989629ba452ebf654ff32e954998 | Charge photon coupling extraction for ICT 1. The experimental data (blue) is fitted (orange) with Eq. <ref> and Eq. <ref> to extract the relevant parameters. | 2012.15588 | Charge_photon_coupling_TP1.jpg |
0.40996 | ec2f676226d741ee977ff1459038ea18 | Charge photon coupling extraction for ICT 2. The experimental data (blue) is fitted (orange) with Eq. <ref> and Eq. <ref> to extract the relevant parameters. | 2012.15588 | Charge_photon_coupling_TP2.jpg |
0.415273 | 3112f4e5c5a34c62bdb01ee2b67e7485 | Double quantum dot device & working points. (a) Simplified 3D schematic of a split-gate, silicon-on-insulator field-effect transistor. A LC resonator wired to gate 1 is used for reflectometry readout. Static voltages and are applied to gates 1 and 2. Using bias tees, they are combined with a ∼500 reflectometry tone () and a 1-20 microwave spectroscopy tone (), respectively. The inset shows a false color scanning electron micrograph of the device (scale bar: 100). (b), (d) Phase response of the LC resonator as a function of and showing inter-dot charge transitions ICT 2 and ICT 3 for even and odd parity, respectively. The insets show the equivalent one- and two-electron charge configurations just above and below ICT 3 and ICT 2, respectively. The first (second) number represents the equivalent hole occupation in the dot under gate 1 (gate 2). (c), (e) Phase response as a function of and B_z at fixed (see dashed line in (b) and (d)) revealing the ground-state evolution in magnetic field. | 2012.15588 | Fig1.jpg |
0.465541 | d27c81a86f31478cb37787e449d873b2 | Photon assisted spectroscopy at zero magnetic field. (a) Phase response of the resonator as a function of gate voltage, , and microwave frequency, , at zero magnetic field. The output power of the microwave generator is adjusted for each in order to deliver a constant power at the device D. In addition to the central interdot transition signal vanishing at , two side branches mark photon-assisted charge transitions between the quantum dots. (b) Theoretical simulation of the driven DQD phase response. The central dip at ε = 0 vanishes when the excitation energy matches Δ. (c) Multi-photon processes arising at higher driving power. Additional side branches appear at one half and one third of the side-branch frequency in (a) indicating two-photon and three-photon process, respectively. (d) Energy diagram of a DQD near the “” ↔ “” transition at zero magnetic field. The double arrows mark the processes giving rise to the branches observed in (a). | 2012.15588 | Fig2.jpg |
0.463238 | b2c9d046305a464d95e7948b5c3d20e6 | Photon assisted spectroscopy at finite magnetic field for an even-parity inter-dot charge transition. (a) Phase response of the LC resonator around the spin-orbit anticrossing of with as a function of and at B_z=600. The dashed horizontal lines delimit regions in which the spectroscopy tone power is held constant at room temperature. The dash-dotted vertical line marks position of the ICT at zero magnetic field, corresponding to ε = 0. The strong vertical structure is associated with the spin-orbit anticrossing at ε =. It vanishes at = /h =. The three side branches around the central signal correspond to photon induced charge transitions between the quantum dots, as indicated in (c). (b) Corresponding simulated phase response of the driven DQD. (c) Energy diagram of DQD of around a “” ↔ “” transition at finite magnetic field and with g_1 ≠ g_2. The photon-induced charge transitions responsible for the side branches in (a) and (b) are indicated by arrows and corresponding symbols. | 2012.15588 | Fig3.jpg |
0.424506 | 2f1ef80c57074481beb88e2ded89deb7 | Photon assisted spectroscopy at finite magnetic field for an odd-parity inter-dot charge transition. (a) Phase response of the LC resonator around zero detuning as a function of and at B_z = 1.3. The dispersive response due to the charge qubit is visible as the vertical feature at ≈-737.3. At resonance, 2h = Ω, the resonator undergoes as well a phase shift (white dashed line). Around 12 spin-orbit anticrossings between states with opposite spin localized in different dots are detectable in the driven response. (b) Energy diagram of a single hole in a DQD (“” ↔ “” transition) at finite magnetic field with g_1 ≃ g_2. At zero detuning, the down spin states of the left and right dot undergo an anticrossing due to tunneling t. A small anticrossing due to the spin-orbit interaction appears between the down spin states of one dot and the up spin states of the other dot. Microwave induced transitions that give rise to the two branches are highlighted with arrows. | 2012.15588 | Fig4.jpg |
0.455891 | 8ddeaa4aa7354281a96e189fcda0ecac | Spin orbit interaction extraction for ICT 1. Phase response of the LC resonator as a function of and is shown on the left and a line-cut (black dashed line) is shown on the right. The line-cut is fitted with Eq. <ref> to extract the framed parameters. | 2012.15588 | TP1_deltaSO_fit_-50_dBm.jpg |
0.430152 | 75e1a4495bf14f85b7edd4f34208a66c | Phase response of the driven double quantum dot as given by Eqs. (<ref>) and (<ref>) for the one and two photon processes at zero magnetic field. The parameters are Q=24, ħ=2, ħ g_c=0.15. The double quantum dot is modeled by the charge tunneling t=11.8, tunneling with spin flip t_so=8.3, and g-factors g_1=4, g_2=1.91. Detuning energy ε is counted from the point of charge degeneracy at zero magnetic field. The amplitude of the resonant drive is A_ exc=15. The phonon equilibrium temperature is significantly smaller than all the energy separations between the levels, which corresponds to the limit Γ_↑≪Γ_↓. With our environment model we set for simplicity S_AA(E)/ħ=0.5 and S_AA(0)/ħ=S_AA()/ħ=0.5. To account for quasistatic noise due to the electrical environment we furthermore convolute the computed signal with a Gaussian function of width σ=5. | 2012.15588 | TP2SpectroHighPowTh.jpg |
0.4596 | 96338546afd141058d3bb73b6fe3417c | Alpha factor extraction for ICT 2. Phase response of the LC resonator as a function of and (with power calibration) is shown on the left. The scatter plot on the right retraces the positions of the central line as ell as the right dip-peak structure from the left graph. The orange curve is a fit using Eq. <ref> which yields the framed results. | 2012.15588 | TP2_alpha_fit.jpg |
0.431427 | 1321e19f83364017ab32bdaed742b564 | Tunnel coupling extraction for ICT 2. Phase response of the LC resonator as a function of and is shown on the left and a line-cut (black dashed line) is shown on the right. The line-cut is fitted with Eq. <ref> to extract the framed parameters. | 2012.15588 | TP2_delta_fit_-50_dBm.jpg |
0.422942 | bfa5c08c2aa14e0dbcd55c17f236a8e5 | g-factors extraction. Simulated response of the driven DQD system at finite magnetic field in the case of an even charge parity (top panel) and the corresponding schematic of the energy diagram (bottom panel) highlighting the transition energies E_f1 and E_f2 necessary for g-factors extraction. | 2012.15588 | gFactorsExtraction.jpg |
0.411401 | 5085d90284d54b318e81ea4048582d2e | The one-parameter model of LULESH evaluating the effect of the number of MPI ranks r per node. Computational kernels experience slowdown because of hardware contention when many processes occupy the same socket. | 2012.15592 | contention-plot-small.jpg |
0.46845 | 185613c7329b4e208139049444ee3a20 | The Score-P instrumentation overhead of LULESH on the Skylake cluster. Plot scales are tuned to enhance visibility. | 2012.15592 | overhead-lulesh-small.jpg |
0.471566 | b52d3d425d9a40e581c7a820610e4ee8 | The Score-P instrumentation overhead of MILC on the Skylake cluster. Plot scales are tuned to enhance visibility. | 2012.15592 | overhead_milc.jpg |
0.436684 | d2a3559f31524664b7180436888db7ee | The basic concept of . A tainted run of the program provides information that improves the empirical performance modeling process along the dimensions cost, quality, and validity of the resulting models. | 2012.15592 | overview-new.jpg |
0.422569 | eb1defab91e5436ba30f3a5f875529cb | The processing pipeline of . All four major steps of empirical modeling are improved with the program information provided by taint analysis. | 2012.15592 | workflow-new.jpg |
0.40761 | fdba757c5f6d42f2b73b2dc9dc8de48c | The red points are the graph of T_2n+1^-1y_n|_n=100 and the green line is y=(3τ+1)/2x. | 2012.15594 | equi_n_100.jpg |
0.379488 | f7a64581a5004b7c90e57d66df04a7d6 | The red points are the graph of (θ_5,n)_n∈[-50,50] and the green line is y=(3τ+1)/2x. | 2012.15594 | minimal_n_5.jpg |
0.505779 | 3653524055b94357a1de63c18f27f704 | The red points are the graph with n=100 and the green line is y=sign(x)x^2. | 2012.15594 | without_rot.jpg |
0.401147 | 6dabbebfe7f64ca082f4b0bb0b582114 | extends with range queries, is efficient and performs well (RocksDB). | 2012.15596 | Headpicture.jpg |
0.438128 | ca504faf340746baa4c843c80d54660b | (a) Mixed trace-sizes increase the number of PMHF, and (b) improve the error-correction. | 2012.15596 | PMHF-ErrCorrection.jpg |
0.385662 | 8a1d1f124be54540a6f3877377cf11c5 | space management: (a) basic with uniform-sized traces are stored overlapped in a bit-array; and (b) mixed trace-sizes and segment-based space allocation with different error-control policies: exact storage or replicating hashing. | 2012.15596 | Space-Allocation.jpg |
0.444538 | 62197f68ff6b4df2900d422575638854 | While (a) hashing in improves random distribution, (b) piecewise-monotone hashing in preserves partial order (within a trace) assisting range-querying, and is (c) hierarchical in DTT. | 2012.15596 | Std-Monotone-HF.jpg |
0.440989 | 89fe5dca97854ccaaf6f81a0aa614d76 | Organization of a (a) trace-tree and path-encoding; (b) example of the trace-tree representation of 129 and 137 in (c) an excerpt of the DTT. | 2012.15596 | TraceTree.jpg |
0.482348 | 7826e595bf0e49bd8ee445f2ba4d6292 | relies (a) on a dyadic interval scheme; which is represented as a (b_1) Dyadic Trace-Tree where each data point (b_2) represents an interval and is encoded with a Trace-Tree scheme and (c) piecewise- monotone hashing functions. Upon insertion they place it in a bit array, which is used in (d) point- and (e) range-lookups. | 2012.15596 | bloomRF_example.jpg |
0.466503 | e58c38b11eb941deb2b3292372d82775 | has good performance for a variety of ranges and workload distributions in RocksDB. | 2012.15596 | exp1.jpg |
0.406103 | 5a208297a12b46cf880139959b2b8bf4 | is efficient, with better performance for different space budgets and query ranges in RocksDB. | 2012.15596 | exp2.jpg |
0.498002 | b3d576cce38f4dff8a5e14776afefdd8 | Impact of insertions on probe performance. | 2012.15596 | exp3.jpg |
0.437909 | c182bf6223d548f4a9f3835ee1505ee2 | with (a) strings and (b) floats. | 2012.15596 | exp4.jpg |
0.395963 | b5f0ab9f62174ae8993bbee218ff7243 | Dual-Attribute outperforms two separate s on individual attributes. | 2012.15596 | exp5.jpg |
0.448246 | 2e62b9636c0e43c6aca83c26252cdb01 | Rogue wave solutions with 𝒢𝒜[1] for (a) 3-NLS equation, (c) 5-NLS equation and (e) 7-NLS equation. Rogue wave solutions with 𝒜_n+1/2[1] for (b) 3-NLS equation, (d) 5-NLS equation and (f) 7-NLS equation. | 2012.15603 | GAA.jpg |
0.416397 | fa97744e262245e9b761166dc2f98933 | The second-order rogue wave solution with 𝒢𝒜[2] for (a) 3-NLS equation and (c) 5-NLS equation. The second-order rogue wave solution with 𝒜_n+1/2[2] for (b) 3-NLS equation and (d) 5-NLS equation. | 2012.15603 | GAAhigh.jpg |
0.439681 | 3a02552ad0b648b3b12dbdd3c1e7cd92 | Profiles of fundamental RW solutions of the 6-NLS equation with ζ=1, α_0=-1/2 in Eq. (<ref>). | 2012.15603 | RW1.jpg |
0.460811 | d9a836eede864948ad34f704bcffdea2 | Rogue wave solutions q_1^[1]_2/a_1 (a, c, e) and q_2^[1]_2/a_2 (b, d, f) of 3-NLS equation in Eqs. (<ref>)-(<ref>) with parameters: (a, b) ζ=1, α_0=-100, α_1=-i, α_2=-2; (c, d) ζ=1, α_0=100, α_1=0, α_2=-2; (e, f) ζ=1, α_0=α_1=0, α_2=-2. | 2012.15603 | RW2.jpg |
0.455368 | 05d49a26b7f34dae850bf32d4581226e | Rogue wave solution q_1^[1]_3/a_1 (a, c, e) and q_2^[1]_3/a_2 (b, d, f) of 3-NLS equation in Eqs. (<ref>), (<ref>) and (<ref>) with parameters: (a, b) ζ=1, α_0=225/2, α_1=-225i, α_2=3, α_3=-6i; (c, d) ζ=1, α_0=-225/2, α_1=225i, α_2=-3, α_3=-6i; (e, f) ζ=1, α_0=-128, α_1=α_2=0, α_3=-6i. | 2012.15603 | RW3.jpg |
0.516581 | ee71d2bb3681441188ca99feaac27087 | Rogue wave solutions q_1^[1]_4/a_1 (a, c, e) and q_2^[1]_4/a_2 (b, d, f) of 4-NLS equation in Eqs. (<ref>), (<ref>) and (<ref>) with parameters: (a, b) ζ=1, α_0=5^49, α_1=250i, α_2=500, α_3=12i, α_4=24; (c, d) ζ=1, α_0=5^49, α_1=0, α_2=-500, α_3=-24i, α_4=24; (e, f) ζ=1, α_0=10^4, α_1=-300, α_2=0, α_3=-6i, α_4=24. | 2012.15603 | RW4.jpg |
0.478842 | ae4a116fe1724e6eb54243ef5bd7711c | Rogue wave solutions q_1^[1]_4/a_1 (a, c, e) and q_2^[1]_4/a_2 (b, d, f) of the 4-NLS equation given in Eqs. (<ref>), (<ref>) and (<ref>) with parameters: (a, b) ζ=1, α_0=10^4, α_1=-300i, α_2=0, α_3=-6i, α_4=24; (c, d) ζ=1, α_0=15^4, α_1=0, α_2=900, α_3=0, α_4=24; (e, f) ζ=1, α_0=α_1=α_2=0, α_3=-19i, α_4=8. | 2012.15603 | RW4a.jpg |
0.535737 | e8a1f95388af4595ae9ca46f22f9474d | Degree distribution of the emergent network of imprinted WS networks after the subtraction of ten photons, using the same data as Fig. <ref>. Complete histograms (top) were each obtained by combining all the nodes of 74 random realizations of a 100-node network (see also Fig. <ref>). Distance-resolved histograms (bottom) are obtained by grouping nodes based on their network distance (in the imprinted network) to the node of photon subtraction. Network distance is indicated by color code and labeled by a number (zero being the node of subtraction). Photon subtraction shifts the degree distribution to higher values for the nodes of photon subtraction (distance 0) and those at distance 2. At distance 3 and beyond, the effects are negligible. At distance 1 the distribution is affected in a non-trivial way depending on p_WS. The value of p_WS also influences the relative importance of distance-induced features, e.g., for p_WS=0.05 we find a larger fraction of nodes at distance 3 or beyond. | 2012.15608 | Distance-Degree-WS.jpg |
0.431348 | c304cb6cc80549e88d8976f0aa75f7ad | Nearest-neighbor (1) degree distribution of the emergent network after the subtraction of ten photons, using the same data as Fig. <ref>. Top row: complete nearest-neighbor histograms. Bottom row: histograms obtained by grouping nearest-neighbor nodes based on the number of other nearest-neighbor nodes they are connected to in the nearest-neighbour sub-network. The connectivity in the nearest-neighbor sub-network is highlighted by the nodes represented next to the histogram. Nearest neighbors of the node of photon subtraction are more strongly affected by the non-Gaussian operation when they are connected to other nearest neighbors. Nodes that are not connected to any other nearest neighbors (darkest purple) are shifted to lower degrees as compared to the Gaussian distributions in Fig. <ref>. | 2012.15608 | Distance-One-Degree-WS.jpg |
0.442421 | 5950f6a5b17246bd8adbfae575da248c | Imprinted networks 𝒜 (left column) give rise to emergent networks 𝔸 (middle column), which can then further undergo photon subtraction (right column). Photon subtraction at indicated red node; node sizes show the degree. Imprinted networks include Barabási-Albert with m=1 (top row) and Watts-Strogatz generated from a regular one-dimensional network in which every node is connected to its k=5 nearest neighbors with a rewiring probability p_WS=0.05 (bottom row). | 2012.15608 | Example-networks.jpg |
0.431268 | f3c5516cf5ae44c09d3322322532486f | BA emergent network statistics in Gaussian case. Histogram of degree and clustering for the imprinted BA network with m=1 and m=2 (top row) and associated emergent network of photon number correlations (bottom row) in the Gaussian case, i.e., when no photon is subtracted. | 2012.15608 | Gaussian-BA.jpg |
0.412023 | 6c2a9d939653475a877172fed54d32ca | WS emergent network statistics in Gaussian case. Histogram for the degree and clustering for the imprinted networks (top row) and for the emergent network of photon number correlations in the Gaussian case (when no photon is subtracted) (bottom row). Here results on the WS network model with p_WS=0.05, 0.2 and 0.6 are reported. | 2012.15608 | Gaussian-WS.jpg |
0.480895 | a340751681ac4a66aa19665e686dfd1a | Distance-resolved moments for the degree statistics in the emergent photon-number correlation network, resulting from imprinted BA networks with parameters m=1 and m=2. Gaussian states (cyan) and ten-photon subtracted states (red). The same observations hold as for the WS networks in Fig. <ref>, except for the distance 1 nodes in imprinted structures with m=1. The latter is explained in detail in Sec. <ref>. | 2012.15608 | Moment-Analysis-BA_alt.jpg |
0.473167 | 155847e1e10d46d088e01b830e0d5fc7 | Distance-resolved moments for the degree statistics in the emergent photon-number correlation network, resulting from imprinted WS networks with rewiring probabilities p_WS=0.05, p_WS=0.2, and p_WS=0.6. Gaussian states (cyan) and ten-photon subtracted states (red). Photon subtraction mean and variance are affected in the same way for all networks, showing that photon subtraction has the global tendency of increasing the degree and widening the distribution of nodes up to distance 2. The effect on higher moments depends on the value of p_WS, showing that photon subtraction also affects the fine structure of the degree distribution in a more subtle way that depends on the network topology. At distance 3 and beyond we see no effect. | 2012.15608 | Moment-Analysis-WS_alt.jpg |
0.438138 | b3f67b7105d24cfebb4da43f1c7d4b0b | Analysis and structures for an imprinted WS network. Generated from a regular one-dimensional network in which every node is connected to its k=5 nearest neighbors with a rewiring probability p_WS=0.05. Top row: imprinted complex network (left, node with highest degree is highlighted in red); and emergent networks before subtraction (Gaussian, middle) and after subtraction (Ten-photon subtracted, right) with their corresponding degree distributions. Middle row: degree distribution of the photon-subtracted case (zoom on right) is broken up, in the central panel, according to the distance (0, 1, 2, or ⩾ 3) of nodes from the subtraction node. The color code used for different distances is adopted for the nodes in the imprinted network (left). Lower row: Structure of the next-neighbor nodes (distance 1) is highlighted (left); zoom of the degree distribution at distance one (middle); statistics is broken up according to the connectivity of the different nodes (right). | 2012.15608 | Overview_WS.jpg |
0.409726 | c669012d39714a35b6b58c161f9225f9 | Histograms (A) and moments (B) for the clustering distributions for the same networks as Fig. <ref>. Colors indicate degree data prior to (cyan, dots) and after (red, squares) the subtraction of ten photons in the node with the highest connectivity. Photon subtraction shifts and widens the distribution, as shown by the histrograms and quantified by the mean and variance. The higher moments and histrograms indicate that the finer structure of these distributions depend strongly on the value of p_WS. | 2012.15608 | Ten-photon-Clustering-WS.jpg |
0.465952 | 6f61057e2a6c4c5dbd6254fe57d08094 | Histograms (A) and moments (B) for the degree distributions of the emergent correlation network obtained from a WS imprinted structure with rewiring probabilities p_WS=0.05,0.2 and 0.6. Colors indicate degree data prior to (cyan, dots) and after (red, squares) the subtraction of ten photons in the node with the highest connectivity. Data were each obtained by combining 74 random realizations of a 100-node network. The WS imprinted network is obtained by starting from a one-dimensional regular network where each node is connected to its k=5 nearest neighbors. Both moments and histograms show how photon subtraction changes the bulk of the distribution by increasing the mean degree and the width of the distribution (i.e., variance). The higher moments and histograms also show that the finer structure in the tails of the degree distribution depends strongly on the value of p_WS. | 2012.15608 | Ten-photon-Degree-WS.jpg |
0.449113 | 41ea130585d64f289abf399530cc348f | Network representation of a two dimensional quantum cluster state, here pictured as a network where the nodes represent the bosonic modes and the links represent C_Z operations between pairs of nodes. | 2012.15608 | clusterPP2.jpg |
0.413757 | 3b773130a6194f8596e9c880b287afa4 | Histogram of the degree distribution for 100 networks of the three types shown in Fig. <ref>, BA (purple), WS (green), ER (orange). On the left the linear scale is used while on the right the scale is double-logarithmic to emphasize the appearance of the power-law tail for the BA networks. The average degree is ⟨ D⟩= 4.44 (BA), 4.0 (WS) and 4.09 (ER). | 2012.15608 | histofin.jpg |
0.527843 | bfc454656dd54c47a0e6a59eec3e2a7c | Left column, first row: sketch of nonlinear multi-mode χ^(2) optical processes that demonstrate the deterministic implementation of large CV networks and mode-dependent multi-photon subtraction. Second row: circuit representation of the two processes. Third row: graphical representation of the so-called cluster states (imprinted networks) where the nodes are the different modes of the field and edges correspond to C_Z gates between modes. Last and fourth row: emergent networks where edges are photon-number correlations between modes. Right column: same as left but considering imprinted cluster with shapes typical of complex network models. | 2012.15608 | sketch2.jpg |
0.440383 | 9b1e9ec9efac44f8ba11e33d6cdde141 | ar | 2012.15613 | ar_sentence_length.jpg |
0.425503 | 235b541f3a484e139465808f5180a713 | Proportion of continued words | 2012.15613 | continuation.jpg |
0.397326 | ad66bbba4f434ca7a53746aba0430bba | We consider all languages and models. | 2012.15613 | corr_all_models.jpg |
0.41893 | cf1a199cb0764f4581016dacc80d3d2e | For the proportion of continued words and the fertility, we consider fully fine-tuned mBERT, the MonoModel-* models, and the mBERTModel-* models. For the pretraining corpus size, we consider the original monolingual models and the MonoModel-MonoTok models. | 2012.15613 | corr_id.jpg |
0.404898 | 4c21279a38a24dfe83792ceb64b657b4 | Spearman's ρ correlation of a relative decrease in the proportion of continued words (Cont. Proportion), a relative decrease in fertility, and a relative increase in pretraining corpus size with a relative increase in downstream performance over fully fine-tuned mBERT. For the proportion of continued words and the fertility, we consider fully fine-tuned mBERT, the MonoModel-* models, and the mBERTModel-* models. For the pretraining corpus size, we consider the original monolingual models and the MonoModel-MonoTok models. We exclude the ID models (see Appendix <ref> for the clarification). | 2012.15613 | corr_no_id.jpg |
0.467121 | 99df52942a7c447eb623d259aa8d7520 | en | 2012.15613 | en_sentence_length.jpg |
0.434294 | 04ce7cf89285408481daf750040e512f | Subword fertility | 2012.15613 | fertility.jpg |
0.589223 | def6b83e8a454a238e989d9d0465e279 | fi | 2012.15613 | fi_sentence_length.jpg |
0.482115 | aacdf1b8608e40eb8021a74bf3e59d43 | id | 2012.15613 | id_sentence_length.jpg |
0.539673 | 07a70f3929da4c3eb601ebdd8efd5521 | ja | 2012.15613 | ja_sentence_length.jpg |
0.572598 | 46ae2bd2dd714cffb15a209958bc6ec3 | ko | 2012.15613 | ko_sentence_length.jpg |
0.530274 | 257228b07d91421e8e576ca968cf8b0f | ru | 2012.15613 | ru_sentence_length.jpg |
0.404128 | cfdd4b1bfb46498a9a1a5dcebd90e9e5 | Pretraining corpus size | 2012.15613 | size_comparison.jpg |
0.533787 | 46d5af7d06994e93a1f2365785595a3d | tr | 2012.15613 | tr_sentence_length.jpg |
0.388887 | fb9fd715c6c84d93872c14bdf1cf35eb | Proportion of unknown tokens in respective monolingual corpora tokenized by monolingual models vs. mBERT. | 2012.15613 | unknowns.jpg |
0.553588 | 823d730d3e5f4645896bd945d52bd8eb | zh | 2012.15613 | zh_sentence_length.jpg |
0.483459 | eba61e3b435e41c1b62954a53c5bd656 | Frequency-domain power allocation. | 2012.15615 | 4_wav.jpg |
0.467583 | b4d3e474dd2e48efb3069461bae45897 | Time-domain waveform. | 2012.15615 | 5_wav.jpg |
0.384025 | 174f27464608494b86245b4f6e985979 | Power region of harvested DC power with K=2, U=16, and N=8. | 2012.15615 | boundary8.jpg |
0.448498 | 0ba6ee06ad784585a7c579190d1922f5 | Frequency response of the multipath channel. | 2012.15615 | channels6.jpg |
0.465938 | 955c217790f04fdb9a1040a2be06aded | Rectenna circuit model. | 2012.15615 | circuit_1.jpg |
0.433935 | 9b20ba0a73c441be97c59f370913cb12 | The behaviour of the nonlinear rectenna model given in (<ref>). | 2012.15615 | diode_model_46.jpg |