abstract
stringlengths
6
6.09k
id
stringlengths
9
16
time
int64
725k
738k
One of the consequences of persistent technological change is that it force individuals to make decisions under extreme uncertainty. This means that traditional decision-making frameworks cannot be applied. To address this issue we introduce a variant of Case-Based Decision Theory, in which the solution to a problem obtains in terms of the distance to previous problems. We formalize this by defining a space based on an orthogonal basis of features of problems. We show how this framework evolves upon the acquisition of new information, namely features or values of them arising in new problems. We discuss how this can be useful to evaluate decisions based on not yet existing data.
2104.14268
737,909
Structural and magnetic transitions in a double perovskite hosting 5d1 Re ions are discussed on the basis of recently published high-resolution x-ray diffraction patterns [D. Hirai, et al., Phys. Rev. Res. 2, 022063(R) (2020)]. A reported structural transition below room temperature, from cubic to tetragonal symmetry, appears not to be driven by T2g-type quadrupoles, as suggested. A magnetic motif at lower temperature is shown to be composed of two order parameters, associated with propagation vectors k = (0, 0, 1) and k = (0, 0, 0). Findings from our studies, for structural and magnetic properties of Ba2MgReO6, surface in predicted amplitudes for x-ray diffraction at rhenium L2 and L3 absorption edges, and magnetic neutron Bragg diffraction. Specifically, entanglement of anapole and spatial degrees of freedom creates a quadrupole in the neutron scattering amplitude. It would be excluded in an unexpected scenario whereby the rhenium atomic state is a manifold. Also, a chiral signature visible in resonant x-ray diffraction will be one consequence of predicted electronic quadrupole and magnetic dipole orders. A model Re wave function consistent with all current knowledge is a guide to electronic and magnetic multipoles engaged in x-ray and neutron diffraction investigations.
2104.14269
737,909
Robotic applications nowadays are widely adopted to enhance operational automation and performance of real-world Cyber-Physical Systems (CPSs) including Industry 4.0, agriculture, healthcare, and disaster management. These applications are composed of latency-sensitive, data-heavy, and compute-intensive tasks. The robots, however, are constrained in the computational power and storage capacity. The concept of multi-agent cloud robotics enables robot-to-robot cooperation and creates a complementary environment for the robots in executing large-scale applications with the capability to utilize the edge and cloud resources. However, in such a collaborative environment, the optimal resource allocation for robotic tasks is challenging to achieve. Heterogeneous energy consumption rates and application of execution costs associated with the robots and computing instances make it even more complex. In addition, the data transmission delay between local robots, edge nodes, and cloud data centres adversely affects the real-time interactions and impedes service performance guarantee. Taking all these issues into account, this paper comprehensively surveys the state-of-the-art on resource allocation and service provisioning in multi-agent cloud robotics. The paper presents the application domains of multi-agent cloud robotics through explicit comparison with the contemporary computing paradigms and identifies the specific research challenges. A complete taxonomy on resource allocation is presented for the first time, together with the discussion of resource pooling, computation offloading, and task scheduling for efficient service provisioning. Furthermore, we highlight the research gaps from the learned lessons, and present future directions deemed beneficial to further advance this emerging field.
2104.14270
737,909
The differential cross-section in squared momentum transfer of $\rho$, $\rho^0$, $\omega$, $\phi$, $f_{0}(980)$, $f_{1}(1285)$, $f_{0}(1370)$, $f_{1}(1420)$, $f_{0}(1500)$, and $J/\psi$ produced in high energy virtual photon-proton ($\gamma$$^{*} p$), photon-proton ($\gamma p$), and proton-proton ($pp$) collisions measured by the H1, ZEUS, and WA102 Collaborations are analyzed by the Monte Carlo calculations. In the calculations, the Erlang distribution, Tsallis distribution, and Hagedorn function are separately used to describe the transverse momentum spectra of the emitted particles. Our results show that the initial and final-state temperatures increase from lower squared photon virtuality to higher one, and decrease with increasing of center-of-mass energy.
2104.14271
737,909
The first phase of table recognition is to detect the tabular area in a document. Subsequently, the tabular structures are recognized in the second phase in order to extract information from the respective cells. Table detection and structural recognition are pivotal problems in the domain of table understanding. However, table analysis is a perplexing task due to the colossal amount of diversity and asymmetry in tables. Therefore, it is an active area of research in document image analysis. Recent advances in the computing capabilities of graphical processing units have enabled deep neural networks to outperform traditional state-of-the-art machine learning methods. Table understanding has substantially benefited from the recent breakthroughs in deep neural networks. However, there has not been a consolidated description of the deep learning methods for table detection and table structure recognition. This review paper provides a thorough analysis of the modern methodologies that utilize deep neural networks. This work provided a thorough understanding of the current state-of-the-art and related challenges of table understanding in document images. Furthermore, the leading datasets and their intricacies have been elaborated along with the quantitative results. Moreover, a brief overview is given regarding the promising directions that can serve as a guide to further improve table analysis in document images.
2104.14272
737,909
Since the mapping relationship between definitized intra-interventional 2D X-ray and undefined pre-interventional 3D Computed Tomography(CT) is uncertain, auxiliary positioning devices or body markers, such as medical implants, are commonly used to determine this relationship. However, such approaches can not be widely used in clinical due to the complex realities. To determine the mapping relationship, and achieve a initializtion post estimation of human body without auxiliary equipment or markers, a cross-modal matching transformer network is proposed to matching 2D X-ray and 3D CT images directly. The proposed approach first deep learns skeletal features from 2D X-ray and 3D CT images. The features are then converted into 1D X-ray and CT representation vectors, which are combined using a multi-modal transformer. As a result, the well-trained network can directly predict the spatial correspondence between arbitrary 2D X-ray and 3D CT. The experimental results show that when combining our approach with the conventional approach, the achieved accuracy and speed can meet the basic clinical intervention needs, and it provides a new direction for intra-interventional registration.
2104.14273
737,909
A central line of inquiry in condensed matter science has been to understand how the competition between different states of matter give rise to emergent physical properties. Perhaps some of the most studied systems in this respect are the hole-doped LaMnO$_3$ perovskites, with interest in the past three decades being stimulated on account of their colossal magnetoresistance (CMR). However, phase segregation between ferromagnetic (FM) metallic and antiferromagnetic (AFM) insulating states, which itself is believed to be responsible for the colossal change in resistance under applied magnetic field, has until now prevented a full atomistic level understanding of the orbital ordered (OO) state at the optimally doped level. Here, through the detailed crystallographic analysis of the hole-doped phase diagram of a prototype system, we show that the superposition of two distinct lattice modes gives rise to a striped structure of OO Jahn-Teller active Mn$^{3+}$ and charge disordered (CD) Mn$^{3.5+}$ layers in a 1:3 ratio. This superposition leads to an exact cancellation of the Jahn-Teller-like oxygen atom displacements in the CD layers only at the 3/8th doping level, coincident with the maximum CMR response of the manganties. Furthermore, the periodic striping of layers containing Mn$^{3.5+}$, separated by layers of fully ordered Mn$^{3+}$, provides a natural mechanism though which long range OO can melt, a prerequisite for the emergence of the FM conducting state. The competition between insulating and conducting states is seen to be a key feature in understanding the properties in highly correlated electron systems, many of which, such as the CMR and high temperature superconductivity, only emerge at or near specific doping values.
2104.14274
737,909
In recent years, Evolutionary Algorithms (EAs) have frequently been adopted to evolve instances for optimization problems that pose difficulties for one algorithm while being rather easy for a competitor and vice versa. Typically, this is achieved by either minimizing or maximizing the performance difference or ratio which serves as the fitness function. Repeating this process is useful to gain insights into strengths/weaknesses of certain algorithms or to build a set of instances with strong performance differences as a foundation for automatic per-instance algorithm selection or configuration. We contribute to this branch of research by proposing fitness-functions to evolve instances that show large performance differences for more than just two algorithms simultaneously. As a proof-of-principle, we evolve instances of the multi-component Traveling Thief Problem~(TTP) for three incomplete TTP-solvers. Our results point out that our strategies are promising, but unsurprisingly their success strongly relies on the algorithms' performance complementarity.
2104.14275
737,909
In this letter, we investigate the population dynamics in a May-Leonard formulation of the rock-paper-scissors game in which one or two species, which we shall refer to as "weak", have a reduced predation or reproduction probability. We show that in a nonspatial model the stationary solution where all three species coexist is always unstable, while in a spatial stochastic model coexistence is possible for a wide parameter space. We find, that a reduced predation probability results in a significantly higher abundance of "weak" species, in models with either one or two "weak" species, as long as the simulation lattices are sufficiently large for coexistence to prevail. On the other hand, we show that a reduced reproduction probability has a smaller impact on the abundance of "weak" species, generally leading to a slight decrease of its population size -- the increase of the population size of one of the "weak" species being more than compensated by the reduction of the other, in the two species case. We further show that the species abundances in models where both predation and reproduction probabilities are simultaneously reduced may be accurately estimated from the results obtained considering only a reduction of either the predation or the reproduction probability.
2104.14276
737,909
For a gambler with side information, Kelly betting gives the optimal log growth rate of the gambler's fortune, which is closely related to the mutual information between the correct winner and the noisy side information. We show conditions under which optimal Kelly betting can be implemented using single-letter codes. We show that single-letter coding is optimal for a wide variety of systems; for example, all systems with diagonal reward matrices admit optimal single-letter codes. We also show that important classes of systems do not admit optimal single-letter codes for Kelly betting, such as when the side information is passed through a Z channel. Our results are important to situations where the computational complexity of the gambler is constrained, and may lead to new insights into the fitness value of information for biological systems.
2104.14277
737,909
Continuous and multimodal stress detection has been performed recently through wearable devices and machine learning algorithms. However, a well-known and important challenge of working on physiological signals recorded by conventional monitoring devices is missing data due to sensors insufficient contact and interference by other equipment. This challenge becomes more problematic when the user/patient is mentally or physically active or stressed because of more frequent conscious or subconscious movements. In this paper, we propose ReLearn, a robust machine learning framework for stress detection from biomarkers extracted from multimodal physiological signals. ReLearn effectively copes with missing data and outliers both at training and inference phases. ReLearn, composed of machine learning models for feature selection, outlier detection, data imputation, and classification, allows us to classify all samples, including those with missing values at inference. In particular, according to our experiments and stress database, while by discarding all missing data, as a simplistic yet common approach, no prediction can be made for 34% of the data at inference, our approach can achieve accurate predictions, as high as 78%, for missing samples. Also, our experiments show that the proposed framework obtains a cross-validation accuracy of 86.8% even if more than 50% of samples within the features are missing.
2104.14278
737,909
The mapping of lexical meanings to wordforms is a major feature of natural languages. While usage pressures might assign short words to frequent meanings (Zipf's law of abbreviation), the need for a productive and open-ended vocabulary, local constraints on sequences of symbols, and various other factors all shape the lexicons of the world's languages. Despite their importance in shaping lexical structure, the relative contributions of these factors have not been fully quantified. Taking a coding-theoretic view of the lexicon and making use of a novel generative statistical model, we define upper bounds for the compressibility of the lexicon under various constraints. Examining corpora from 7 typologically diverse languages, we use those upper bounds to quantify the lexicon's optimality and to explore the relative costs of major constraints on natural codes. We find that (compositional) morphology and graphotactics can sufficiently account for most of the complexity of natural codes -- as measured by code length.
2104.14279
737,909
In cataract surgery, the operation is performed with the help of a microscope. Since the microscope enables watching real-time surgery by up to two people only, a major part of surgical training is conducted using the recorded videos. To optimize the training procedure with the video content, the surgeons require an automatic relevance detection approach. In addition to relevance-based retrieval, these results can be further used for skill assessment and irregularity detection in cataract surgery videos. In this paper, a three-module framework is proposed to detect and classify the relevant phase segments in cataract videos. Taking advantage of an idle frame recognition network, the video is divided into idle and action segments. To boost the performance in relevance detection, the cornea where the relevant surgical actions are conducted is detected in all frames using Mask R-CNN. The spatiotemporally localized segments containing higher-resolution information about the pupil texture and actions, and complementary temporal information from the same phase are fed into the relevance detection module. This module consists of four parallel recurrent CNNs being responsible to detect four relevant phases that have been defined with medical experts. The results will then be integrated to classify the action phases as irrelevant or one of four relevant phases. Experimental results reveal that the proposed approach outperforms static CNNs and different configurations of feature-based and end-to-end recurrent networks.
2104.14280
737,909
Ubiquitous internet access is reshaping the way we live, but it is accompanied by unprecedented challenges to prevent chronic diseases planted in long exposure to unhealthy lifestyles. This paper proposes leveraging online shopping behaviors as a proxy for personal lifestyle choices to freshen chronic disease prevention literacy targeted for times when e-commerce user experience has been assimilated into most people's daily life. Here, retrospective longitudinal query logs and purchase records from millions of online shoppers were accessed, constructing a broad spectrum of lifestyle features covering assorted product categories and buyer personas. Using the lifestyle-related information preceding their first purchases of prescription drugs, we could determine associations between online shoppers' past lifestyle choices and if they suffered from a particular chronic disease. Novel lifestyle risk factors were discovered in two exemplars -- depression and diabetes, most of which showed cognitive congruence with existing healthcare knowledge. Further, such empirical findings could be adopted to locate online shoppers at high risk of chronic diseases with fair accuracy (e.g., [area under the receiver operating characteristic curve] AUC=0.68 for depression and AUC=0.70 for diabetes), closely matching the performance of screening surveys benchmarked against medical diagnosis. Unobtrusive chronic disease surveillance via e-commerce sites may soon meet consenting individuals in the digital space they already inhabit.
2104.14281
737,909
Clinicians conduct routine diagnosis by scrutinizing signs and symptoms of patients in treating epidemics. This skill evolves through trial-and-error and improves with time. The success of the therapeutic regimen relies largely on the accuracy of interpretation of such sign-symptoms, based on which the clinician ranks the potent causes of the epidemic and analyzes their interdependence to devise sustainable containment strategies. This study proposed an alternative medical front, a VIRtual DOCtor (VIRDOC), that can self-consistently rank key contributors of an epidemic and also correctly identify the infection stage, using the language of statistical modelling and Machine Learning. VIRDOC analyzes medical data and then translates these into a vector comprising Multiple Linear Regression (MLR) coefficients to probabilistically predict scores that compare with clinical experience-based assessment. The VIRDOC algorithm, risk managed through ANOVA, has been tested on dengue epidemic data (N=100 with 11 weighted sign-symptoms). Results highly encouraging with ca 75% accurate fatality prediction, compared to 71.4% from traditional diagnosis. The algorithm can be generically extended to analyze other epidemic forms.
2104.14282
737,909
We present a new uncertainty principle for risk-aware statistical estimation, effectively quantifying the inherent trade-off between mean squared error ($\mse$) and risk, the latter measured by the associated average predictive squared error variance ($\sev$), for every admissible estimator of choice. Our uncertainty principle has a familiar form and resembles fundamental and classical results arising in several other areas, such as the Heisenberg principle in statistical and quantum mechanics, and the Gabor limit (time-scale trade-offs) in harmonic analysis. In particular, we prove that, provided a joint generative model of states and observables, the product between $\mse$ and $\sev$ is bounded from below by a computable model-dependent constant, which is explicitly related to the Pareto frontier of a recently studied $\sev$-constrained minimum $\mse$ (MMSE) estimation problem. Further, we show that the aforementioned constant is inherently connected to an intuitive new and rigorously topologically grounded statistical measure of distribution skewness in multiple dimensions, consistent with Pearson's moment coefficient of skewness for variables on the line. Our results are also illustrated via numerical simulations.
2104.14283
737,909
We continue studying $6D, {\cal N}=(1,1)$ supersymmetric Yang-Mills (SYM) theory in the ${\cal N}=(1,0)$ harmonic superspace formulation. Using the superfield background field method we explore the two-loop divergencies of the effective action in the gauge multiplet sector. It is explicitly demonstrated that among four two-loop background-field dependent supergraphs contributing to the effective action, only one diverges off shell. It is also shown that the divergences are proportional to the superfield classical equations of motion and hence vanish on shell.
2104.14284
737,909
Path tracking system plays a key technology in autonomous driving. The system should be driven accurately along the lane and be careful not to cause any inconvenience to passengers. To address such tasks, this paper proposes hybrid tracker based optimal path tracking system. By applying a deep learning based lane detection algorithm and a designated fast lane fitting algorithm, this paper developed a lane processing algorithm that shows a match rate with actual lanes with minimal computational cost. In addition, three modified path tracking algorithms were designed using the GPS based path or the vision based path. In the driving system, a match rate for the correct ideal path does not necessarily represent driving stability. This paper proposes hybrid tracker based optimal path tracking system by applying the concept of an observer that selects the optimal tracker appropriately in complex road environments. The driving stability has been studied in complex road environments such as straight road with multiple 3-way junctions, roundabouts, intersections, and tunnels. Consequently, the proposed system experimentally showed the high performance with consistent driving comfort by maintaining the vehicle within the lanes accurately even in the presence of high complexity of road conditions. Code will be available in https://github.com/DGIST-ARTIV.
2104.14285
737,909
Advancing models for accurate estimation of food production is essential for policymaking and managing national plans of action for food security. This research proposes two machine learning models for the prediction of food production. The adaptive network-based fuzzy inference system (ANFIS) and multilayer perceptron (MLP) methods are used to advance the prediction models. In the present study, two variables of livestock production and agricultural production were considered as the source of food production. Three variables were used to evaluate livestock production, namely livestock yield, live animals, and animal slaughtered, and two variables were used to assess agricultural production, namely agricultural production yields and losses. Iran was selected as the case study of the current study. Therefore, time-series data related to livestock and agricultural productions in Iran from 1961 to 2017 have been collected from the FAOSTAT database. First, 70% of this data was used to train ANFIS and MLP, and the remaining 30% of the data was used to test the models. The results disclosed that the ANFIS model with Generalized bell-shaped (Gbell) built-in membership functions has the lowest error level in predicting food production. The findings of this study provide a suitable tool for policymakers who can use this model and predict the future of food production to provide a proper plan for the future of food security and food supply for the next generations.
2104.14286
737,909
A comprehensive analysis of energy requirements and emissions associated with electric vehicles, ranging from mining and making the rare-earth magnets required in electric motor to assembling the Li-ion battery, including charging and regular running of the electric vehicles has been performed. A simple, analytical procedure is used to determine the embodied energy and emissions. The objective is to assess the potential of electric cars to reduce green house gases emission to limit global warming to < 1.5 degrees C by the Year 2050 as per IPCC recommendations and also to compare them with conventional fuel driven cars. The combined embodied energy for Nd- and Dy-metals production which are required in electric motors and battery assembly for 150 million cars, projected to be on the road in the year 2050 is ~ 1500 TWh and the CO2 emissions is found to be > 600 MT. The emissions includes carbon intensity of electrical energy required to run these electric vehicles. The projected emissions due to fossil fuels, gasoline production as well as burning it in combustion engines however is only 412 MT, far less than that due to electric vehicles. The main contributor to emissions from electric vehicles is the battery assembling process which releases ~ 379 MT of CO2-e gases. The emissions from both electric vehicles as well as combustion engine vehicles scale linearly with the number of vehicles, indicating that a breakeven is not possible with the currently available manufacturing technologies. These results clearly show that significant technological developments have to take place in electric vehicles so that they become environmentally better placed compared to combustion engine based cars.
2104.14287
737,909
In a system with inversion symmetry broken, a second-order nonlinear Hall effect can survive even in the presence of time-reversal symmetry. In this work, we show that a giant nonlinear Hall effect can exist in twisted bilayer WTe2 system. The Berry curvature dipole of twisted bilayer WTe2 ({\theta} = 29.4{\deg}) can reach up to ~1400 {\AA}, which is much larger than that in previously reported nonlinear Hall systems. In twisted bilayer WTe2 system, there exists abundant band anticrossings and band inversions around the Fermi level, which brings a complicated distribution of Berry curvature, and leading to the nonlinear Hall signals exhibit dramatically oscillating behavior in this system. Its large amplitude and high tunability indicate that the twisted bilayer WTe2 can be an excellent platform for studying the nonlinear Hall effect.
2104.14288
737,909
Neural ODE Processes approach the problem of meta-learning for dynamics using a latent variable model, which permits a flexible aggregation of contextual information. This flexibility is inherited from the Neural Process framework and allows the model to aggregate sets of context observations of arbitrary size into a fixed-length representation. In the physical sciences, we often have access to structured knowledge in addition to raw observations of a system, such as the value of a conserved quantity or a description of an understood component. Taking advantage of the aggregation flexibility, we extend the Neural ODE Process model to use additional information within the Learning Using Privileged Information setting, and we validate our extension with experiments showing improved accuracy and calibration on simulated dynamics tasks.
2104.14290
737,909
Control of molecular orientation is emerging as crucial for the characterization of the stereodynamics of kinetics processes beyond structural stereochemistry. The special role played in chiral discrimination phenomena has been particularly emphasized by the authors after their extensive probes of experimental control of molecular alignment and orientation. In this work, the role of the orientation has been demonstrated for the first time in first-principles molecular dynamics simulations: stationary points characterized on potential energy surfaces have been calculated for the study of chemical reactions occurring between the bisulfide anion HS- and oriented prototypical chiral molecules CHFXY (where X = CH3 or CN and Y = Cl or I). The important reaction channels are those corresponding to bimolecular nucleophilic substitution (SN2) and to bimolecular elimination (E2): their relative role has been assessed and alternative pathways due to the mirror forms of the oriented chiral molecule are revealed by the different reactivity of the two enantiomers of CHFCNI in SN2 reaction.
2104.14292
737,909
An energetic muon beam is an attractive key to unlock new physics beyond the Standard Model: the lepton flavor violation or the anomalous magnetic moment, and also is a competitive candidate for the expected neutrino factory. Lots of the muon scientific applications are limited by low flux cosmic-ray muons, low energy muon sources or extremely expensive muon accelerators. An prompt acceleration of the low-energy muon beam is found in the beam-driven plasma wakefield up to $\mathrm{TV/m}$. The muon beam is accelerated from $275\mathrm{MeV}$ to more than $10\mathrm{GeV}$ within $22.5\mathrm{ps}$. Choosing the injection time of the muon beam in a proper range, the longitudinal spatial distribution and the energy distribution of the accelerated muon beam are compressed. The efficiency of the energy transfer from the driven electron beam to the muon beam can reach $20\%$. The prompt acceleration scheme is a promising avenue to bring the expected neutrino factory and the muon collider into reality and to catch new physics beyond the Standard Model.
2104.14293
737,909
In this paper, we question if self-supervised learning provides new properties to Vision Transformer (ViT) that stand out compared to convolutional networks (convnets). Beyond the fact that adapting self-supervised methods to this architecture works particularly well, we make the following observations: first, self-supervised ViT features contain explicit information about the semantic segmentation of an image, which does not emerge as clearly with supervised ViTs, nor with convnets. Second, these features are also excellent k-NN classifiers, reaching 78.3% top-1 on ImageNet with a small ViT. Our study also underlines the importance of momentum encoder, multi-crop training, and the use of small patches with ViTs. We implement our findings into a simple self-supervised method, called DINO, which we interpret as a form of self-distillation with no labels. We show the synergy between DINO and ViTs by achieving 80.1% top-1 on ImageNet in linear evaluation with ViT-Base.
2104.14294
737,909
The purpose of this paper is to prove that if on a commutative hypergroup an exponential monomial has the property that the linear subspace of all sine functions in its variety is one dimensional, then this exponential monomial is a linear combination of generalized moment functions.
2104.14295
737,909
BACKGROUND: Software engineering is a human activity. People naturally make sense of their activities and experience through storytelling. But storytelling does not appear to have been properly studied by software engineering research. AIM: We explore the question: what contribution can storytelling make to human--centric software engineering research? METHOD: We define concepts, identify types of story and their purposes, outcomes and effects, briefly review prior literature, identify several contributions and propose next steps. RESULTS: Storytelling can, amongst other contributions, contribute to data collection, data analyses, ways of knowing, research outputs, interventions in practice, and advocacy, and can integrate with evidence and arguments. Like all methods, storytelling brings risks. These risks can be managed. CONCLUSION: Storytelling provides a potential counter--balance to abstraction, and an approach to retain and honour human meaning in software engineering.
2104.14296
737,909
Training Automatic Speech Recognition (ASR) models under federated learning (FL) settings has recently attracted considerable attention. However, the FL scenarios often presented in the literature are artificial and fail to capture the complexity of real FL systems. In this paper, we construct a challenging and realistic ASR federated experimental setup consisting of clients with heterogeneous data distributions using the French Common Voice dataset, a large heterogeneous dataset containing over 10k speakers. We present the first empirical study on attention-based sequence-to-sequence E2E ASR model with three aggregation weighting strategies -- standard FedAvg, loss-based aggregation and a novel word error rate (WER)-based aggregation, are conducted in two realistic FL scenarios: cross-silo with 10-clients and cross-device with 2k-clients. In particular, the WER-based weighting method is proposed to better adapt FL to the context of ASR by integrating the error rate metric with the aggregation process. Our analysis on E2E ASR from heterogeneous and realistic federated acoustic models provides the foundations for future research and development of realistic FL-based ASR applications.
2104.14297
737,909
In this paper we develop and significantly extend the thermal phase change model, introduced in [12], describing the process of paraffinic wax layer formation on the interior wall of a circular pipe transporting heated oil, when subject to external cooling. In particular we allow for the natural dependence of the solidifying paraffinic wax conductivity on local temperature. We are able to develop a complete theory, and provide efficient numerical computations, for this extended model. Comparison with recent experimental observations is made, and this, together with recent reviews of the physical mechanisms associated with wax layer formation, provide significant support for the thermal model considered here.
2104.14298
737,909
The four-bar linkage is a basic arrangement of mechanical engineering and represents the simplest movable system formed by a closed sequence of bar-shaped bodies. Although the mechanism can have in general a spatial arrangement, we focus here on the prototypical planar case, starting however from a spatial viewpoint. The classification of the mechanism relies on the angular range spanned by the rotational motion of the bars allowed by the ratios among their lengths and is established by conditions for the existence of either one or more bars allowed to move as cranks, namely to be permitted to rotate the full 360 degrees range (Grashof cases), or as rockers with limited angular ranges (non-Grashof cases). In this paper, we provide a view on the connections between the "classic" four-bar problem and the theory of 6j symbols of quantum mechanical angular momentum theory, occurring in a variety of contexts in pure and applied quantum mechanics. The general case and a series of symmetric configurations are illustrated, by representing the range of existence of the related quadrilaterals on a square "screen" (namely as a function of their diagonals) and by discussing their behavior according both to the Grashof conditions and to the Regge symmetries, concertedly considering the classification of the two mechanisms and that of the corresponding objects of the quantum mechanical theory of angular momentum. An interesting topological difference is demonstrated between mechanisms belonging to the two Regge symmetric configurations: the movements in the Grashof cases span chirality preserving configurations with a 2 pi-cycle of a rotating bar, while by contrast the non-Grashof cases span both enantiomeric configurations with a 4 pi-cycle.
2104.14299
737,909
Path planning is an important topic in robotics. Recently, value iteration based deep learning models have achieved good performance such as Value Iteration Network(VIN). However, previous methods suffer from slow convergence and low accuracy on large maps, hence restricted in path planning for agents with complex kinematics such as legged robots. Therefore, we propose a new value iteration based path planning method called Capability Iteration Network(CIN). CIN utilizes sparse reward maps and encodes the capability of the agent with state-action transition probability, rather than a convolution kernel in previous models. Furthermore, two training methods including end-to-end training and training capability module alone are proposed, both of which speed up convergence greatly. Several path planning experiments in various scenarios, including on 2D, 3D grid world and real robots with different map sizes are conducted. The results demonstrate that CIN has higher accuracy, faster convergence, and lower sensitivity to random seed compared to previous VI-based models, hence more applicable for real robot path planning.
2104.14300
737,909
Analyzing the financial benefit of marketing is still a critical topic for both practitioners and researchers. Companies consider marketing costs as a type of investment and expect this investment to be returned to the company in the form of profit. On the other hand, companies adopt different innovative strategies to increase their value. Therefore, this study aims to test the impact of marketing investment on firm value and systematic risk. To do so, data related to four Arabic emerging markets during the period 2010-2019 are considered, and firm share price and beta share are considered to measure firm value and systematic risk, respectively. Since a firm's ownership concentration is a determinant factor in firm value and systematic risk, this variable is considered a moderated variable in the relationship between marketing investment and firm value and systematic risk. The findings of the study, using panel data regression, indicate that increasing investment in marketing has a positive effect on the firm value valuation model. It is also found that the ownership concentration variable has a reinforcing role in the relationship between marketing investment and firm value. It is also disclosed that it moderates the systematic risk aligned with the monitoring impact of controlling shareholders. This study provides a logical combination of governance-marketing dimensions to interpret performance indicators in the capital market.
2104.14301
737,909
Observation of quantum phenomena in cryogenic, optically cooled mechanical resonators has been recently achieved by a few experiments based on cavity optomechanics. A well-established experimental platform is based on a thin film stoichiometric ($ Si_3 N_4 $) nanomembrane embedded in a Fabry-Perot cavity, where the coupling with the light field is provided by the radiation pressure of the light impinging on the membrane surface. Two crucial parameters have to be optimized to ensure that these systems work at the quantum level: the cooperativity $ C$ describing the optomechanical coupling and the product $ Q \times \nu$ (quality factor - resonance frequency) related to the decoherence rate. A significant increase of the latter can be obtained with high aspect-ratio membrane resonators where uniform stress dilutes the mechanical dissipation. Furthermore, ultra-high $Q \times \nu$ can be reached by drastically reducing the edge dissipation via clamp-tapering and/or by soft-clamping, virtually a clamp-free resonator configuration. In this work, we investigate, theoretically and experimentally, the edge loss mechanisms comparing two state-of-the-art resonators built by standard micro/nanofabrication techniques. The corresponding results would provide meaningful guidelines for designing new ultra-coherent resonating devices.
2104.14302
737,909
For the measurement of the dynamics of fusion-born alpha particles $E_\alpha \leq 3.5$ MeV in ITER using collective Thomson scattering (CTS), safe transmission of a gyrotron beam at mm-wavelength (1 MW, 60 GHz) passing the electron cyclotron resonance (ECR) in the in-vessel tokamak `port plug' vacuum is a prerequisite. Depending on neutral gas pressure and composition, ECR-assisted gas breakdown may occur at the location of the resonance, which must be mitigated for diagnostic performance and safety reasons. The concept of a split electrically biased waveguide (SBWG) has been previously demonstrated in [C.P. Moeller, U.S. Patent 4,687,616 (1987)]. The waveguide is longitudinally split and a kV bias voltage applied between the two halves. Electrons are rapidly removed from the central region of high radio frequency electric field strength, mitigating breakdown. As a full scale experimental investigation of gas and electromagnetic field conditions inside the ITER equatorial port plugs is currently unattainable, a corresponding Monte Carlo simulation study is presented. Validity of the Monte Carlo electron model is demonstrated with a prediction of ECR breakdown and the mitigation pressure limits for the above quoted reference case with $^1$H$_2$ (and pollutant high $Z$ elements). For the proposed ITER CTS design with a 88.9 mm inner diameter SBWG, ECR breakdown is predicted to occur down to a pure $^1$H$_2$ pressure of 0.3 Pa, while mitigation is shown to be effective at least up to 10 Pa using a bias voltage of 1 kV. The analysis is complemented by results for relevant electric/magnetic field arrangements and limitations of the SBWG mitigation concept are addressed.
2104.14303
737,909
Coded modulation with probabilistic amplitude shaping (PAS) is considered for intensity modulation/direct detection channels with a transmitter peak-power constraint. PAS is used to map bits to a uniform PAM-6 distribution and outperforms PAM-8 for rates up to around 2.3 bits per channel use. PAM-6 with PAS also outperforms a cross-shaped QAM-32 constellation by up to 1 dB and 0.65 dB after bit-metric soft- and hard decoding, respectively. An alternative PAM-6 scheme based on a framed cross-shaped QAM-32 constellation is proposed that shows similar gains.
2104.14304
737,909
Latent heat thermal energy storage (LHTES) has been recommended as an effective technology to the thermal management system of space exploration for its excellent ability of storing thermal energy. However, it is well known that the low thermal conductivity of phase change material (PCM) seriously weakens the heat charging and discharging rates of LHTES system. In present study, the electrohydrodynamic (EHD), which is a popular active heat transfer enhancement technology, is introduced to enhance the PCM melting in a shell-tube LHTES system under microgravity. In our numerical simulations, we mainly focus on the combined effects of the electric Rayleigh number $T$ and the eccentricity $\Gamma$ on the melting performance under microgravity. Based on the numerical results, it is found that in contrast to the case without the electric field, the presence of the electric field causes the heat transfer efficiency and melting behavior of LHTES system to be enhanced significantly. In addition, our results indicates that the EHD technique always shows a good performance in accelerating the melting process even under microgravity, and specially, the maximum time saving in some cases is more than $90\%$. Furthermore, we note that although the concentric annulus is always the optimal configuration under no-gravity condition, the optimal eccentric position of the internal tube strongly depends on the electric Rayleigh number if the gravity effects are taken into account.
2104.14305
737,909
We construct recently introduced palatial NC twistors by considering the pair of conjugated (Born-dual) twist-deformed $D=4$ quantum inhomegeneous conformal Hopf algebras $\mathcal{U}_{\theta }(su(2,2)\ltimes T^{4}$) and $\mathcal{U}_{\bar{\theta}}(su(2,2)\ltimes\bar{T}^{4}$), where $T^{4}$ describe complex twistor coordinatesand $\bar{T}^{4}$ the conjugated dual twistor momenta. The palatial twistors are suitably chosen as the quantum-covariant modules (NC representations) of the introduced Born-dual Hopf algebras. Subsequently we introduce the quantum deformations of $D=4$ Heisenberg-conformal algebra (HCA) $su(2,2)\ltimes H^{4,4}_\hslash$ ($H^{4,4}_\hslash=\bar{T}^4 \ltimes_\hslash T_4$ is the Heisenberg algebra of twistorial oscillators) providing in twistorial framework the basic covariant quantum elementary system. The class of algebras describing deformation of HCA with dimensionfull deformation parameter, linked with Planck length $\lambda_p$ will be called the twistorial DSR (TDSR) algebra, following the terminology of DSR algebra in space-time framework. We shall describe the examples of TDSR algebra linked with Palatial twistors which are introduced by the Drinfeld twist and by the quantization map in $H_\hslash^{4,4}$. We introduce as well generalized quantum twistorial phase space by considering the Heisenberg double of Hopf algebra $\mathcal{U}_\theta(su(2,2)\ltimes T^4).$
2104.14306
737,909
We report the detection of a short-lived narrow quasi-periodic oscillation (QPO) at ~88 mHz in an Insight-HXMT observation during the soft state of the persistent black hole high mass X-ray binary Cygnus X-1. This QPO is significantly detected in all the three instruments of Insight-HXMT, so in the broad energy range 1-250 keV. The fractional RMS of the QPO does not show significant variations above 3 keV (~5%) while decreases at lower energy (~2%). We show that this QPO is different from the type-A, -B, and -C QPOs usually observed in black hole X-ray binaries. We compare QPOs at similar frequencies that have been previously detected in another persistent high mass X-ray binaries in the soft state, we speculate that such QPOs might relate to some local inhomogeneity rarely formed in the accretion flow of wind-fed accretion systems.
2104.14307
737,909
Aberrations limit scanning fluorescence microscopy when imaging in scattering materials such as biological tissue. Model-based approaches for adaptive optics take advantage of a computational model of the optical setup. Such models can be combined with the optimization techniques of machine learning frameworks to find aberration corrections, as was demonstrated for focusing a laser beam through aberrations onto a camera [arXiv:2007.13400]. Here, we extend this approach to two-photon scanning microscopy. The developed sensorless technique finds corrections for aberrations in scattering samples and will be useful for a range of imaging application, for example in brain tissue.
2104.14308
737,909
Improving the clock stability is of fundamental importance for the development of quantum-enhanced metrology. One of the main limitations arises from the randomly-fluctuating local oscillator (LO) frequency, which introduces "phase slips" for long interrogation times and hence failure of the frequency-feedback loop. Here we propose a strategy to improve the stability of atomic clocks by interrogating two out-of-phase state sharing the same LO. While standard Ramsey interrogation can only determine phases unambiguously in the interval $[-\pi/2,\pi/2]$, the joint interrogation allows for an extension to $[-\pi,\pi]$, resulting in a relaxed restriction of the Ramsey time and improvement of absolute clock stability. Theoretical predictions are supported by ab-initio numerical simulation for white and correlated LO noise. While our basic protocol uses uncorrelated atoms, we have further extended it to include spin-squeezing and further improving the scaling of clock stability with the number of atoms. Our protocol can be readily tested in current state-of-the-art experiments.
2104.14309
737,909
We present a Dicke state preparation scheme which uses global control of $N$ spin qubits: our scheme is based on the standard phase estimation algorithm, which estimates the eigenvalue of a unitary operator. The scheme prepares a Dicke state non-deterministically by collectively coupling the spins to an ancilla qubit via a $ZZ$-interaction, using $\ceil*{\log_2 N} + 1$ ancilla qubit measurements. The preparation of such Dicke states can be useful if the spins in the ensemble are used for magnetic sensing: we discuss a possible realization using an ensemble of electronic spins located at diamond Nitrogen-Vacancy (NV) centers coupled to a single superconducting flux qubit. We also analyze the effect of noise and limitations in our scheme.
2104.14310
737,909
We revisit large spectroscopic data sets for field stars from the literature to derive the upper Li envelope in the high metallicity regime in our Galaxy. We take advantage of Gaia EDR3 data and state-of-the-art stellar models to precisely determine the position of the sample dwarf stars in the Hertzsprung-Russell diagram. The highest Li abundances are found in field metal-rich warm dwarfs from the GALAH survey, located on the hot side of the Li-dip. Their mean Li value agrees with what was recently derived for warm dwarfs in metal-rich clusters, pointing towards a continuous increase of Li up to super-solar metallicity. However, if only cool dwarfs are considered in GALAH, as done in the other literature surveys, it is found that the upper Li envelope decreases at super-solar metallicities, blurring the actual Li evolution picture. We confirm the suggestion that field and open cluster surveys that found opposite Li behaviour in the high metallicity regime do not sample the same types of stars: The first ones, with the exception of GALAH, miss warm dwarfs that can potentially preserve their original Li content. Although we can discard the bending of the Li upper envelope at high metallicity derived from the analysis of cool star samples, we still need to evaluate the effects of atomic diffusion on warm, metal-rich early-F and late-A type dwarfs before deriving the actual Li abundance at high metallicity.
2104.14311
737,909
We consider the Navier-Stokes system in three dimensions perturbed by a transport noise which is sufficiently smooth in space and rough in time. The existence of a weak solution was proved recently, however, as in the deterministic setting the question of uniqueness remains a major open problem. An important feature of systems with uniqueness is the semigroup property satisfied by their solutions. Without uniqueness, this property cannot hold generally. We select a system of solutions satisfying the semigroup property with appropriately shifted rough path. In addition, the selected solutions respect the well accepted admissibility criterium for physical solutions, namely, maximization of the energy dissipation. Finally, under suitable assumptions on the driving rough path, we show that the Navier-Stokes system generates a measurable random dynamical system. To the best of our knowledge, this is the first construction of a measurable single-valued random dynamical system in the state space for an SPDE without uniqueness.
2104.14312
737,909
We develop a reliable, fully automatic method for the detection of coronal holes, that provides consistent full-disk segmentation maps over the full solar cycle and can perform in real-time. We use a convolutional neural network to identify the boundaries of coronal holes from the seven EUV channels of the Atmospheric Imaging Assembly (AIA) as well as from line-of-sight magnetograms from the Helioseismic and Magnetic Imager (HMI) onboard the Solar Dynamics Observatory (SDO). For our primary model (Coronal Hole RecOgnition Neural Network Over multi-Spectral-data; CHRONNOS) we use a progressively growing network approach that allows for efficient training, provides detailed segmentation maps and takes relations across the full solar-disk into account. We provide a thorough evaluation for performance, reliability and consistency by comparing the model results to an independent manually curated test set. Our model shows good agreement to the manual labels with an intersection-over-union (IoU) of 0.63. From the total of 261 coronal holes with an area $>1.5\cdot10^{10}$ km$^2$ identified during the time range 11/2010 - 12/2016, 98.1% were correctly detected by our model. The evaluation over almost the full solar cycle no. 24 shows that our model provides reliable coronal hole detections, independent of the level of solar activity. From the direct comparison over short time scales of days to weeks, we find that our model exceeds human performance in terms of consistency and reliability. In addition, we train our model to identify coronal holes from each channel separately and show that the neural network provides the best performance with the combined channel information, but that coronal hole segmentation maps can be also obtained solely from line-of-sight magnetograms.
2104.14313
737,909
The new generation of pre-trained NLP models push the SOTA to the new limits, but at the cost of computational resources, to the point that their use in real production environments is often prohibitively expensive. We tackle this problem by evaluating not only the standard quality metrics on downstream tasks but also the memory footprint and inference time. We present MOROCCO, a framework to compare language models compatible with \texttt{jiant} environment which supports over 50 NLU tasks, including SuperGLUE benchmark and multiple probing suites. We demonstrate its applicability for two GLUE-like suites in different languages.
2104.14314
737,909
In many multiagent environments, a designer has some, but limited control over the game being played. In this paper, we formalize this by considering incompletely specified games, in which some entries of the payoff matrices can be chosen from a specified set. We show that it is NP-hard for the designer to make these choices optimally, even in zero-sum games. In fact, it is already intractable to decide whether a given action is (potentially or necessarily) played in equilibrium. We also consider incompletely specified symmetric games in which all completions are required to be symmetric. Here, hardness holds even in weak tournament games (symmetric zero-sum games whose entries are all -1, 0, or 1) and in tournament games (symmetric zero-sum games whose non-diagonal entries are all -1 or 1). The latter result settles the complexity of the possible and necessary winner problems for a social-choice-theoretic solution concept known as the bipartisan set. We finally give a mixed-integer linear programming formulation for weak tournament games and evaluate it experimentally.
2104.14317
737,909
The decoupling of heavy fields as required by the Appelquisst-Carazzone theorem plays a fundamental role in the construction of any effective field theory. However, it is not a trivial task to implement a renormalization prescription that produces the expected decoupling of massive fields, and it is even more difficult in curved spacetime. Focused on this idea, we consider the renormalization of the one-loop effective action for the Yukawa interaction with a background scalar field in curved space. We compute the beta functions within a generalized DeWitt-Schwinger subtraction procedure and discuss the decoupling in the running of the coupling constants. For the case of a quantized scalar field, all the beta function exhibit decoupling, including also the gravitational ones. For a quantized Dirac field, decoupling appears almost for all the beta functions. We obtain the anomalous result that the mass of the background scalar field does not decouple.
2104.14318
737,909
Every x-adjustment in the so-called xVA financial risk management framework relies on the computation of exposures. Considering thousands of Monte Carlo paths and tens of simulation steps, a financial portfolio needs to be evaluated numerous times during the lifetime of the underlying assets. This is the bottleneck of every simulation of xVA. In this article, we explore numerical techniques for improving the simulation of exposures. We aim to decimate the number of portfolio evaluations, particularly for large portfolios involving multiple, correlated risk factors. The usage of the Stochastic Collocation (SC) method, together with Smolyaks sparse grid extension, allows for a significant reduction in the number of portfolio evaluations, even when dealing with many risk factors. The proposed model can be easily applied to any portfolio and size. We report that for a realistic portfolio comprising linear derivatives, the expected reduction in the portfolio evaluations may exceed 6000 times, depending on the dimensionality and the required accuracy. We give illustrative examples and examine the method with realistic multi-currency portfolios.
2104.14319
737,909
Recently, researchers have utilized neural networks to accurately solve partial differential equations (PDEs), enabling the mesh-free method for scientific computation. Unfortunately, the network performance drops when encountering a high nonlinearity domain. To improve the generalizability, we introduce the novel approach of employing multi-task learning techniques, the uncertainty-weighting loss and the gradients surgery, in the context of learning PDE solutions. The multi-task scheme exploits the benefits of learning shared representations, controlled by cross-stitch modules, between multiple related PDEs, which are obtainable by varying the PDE parameterization coefficients, to generalize better on the original PDE. Encouraging the network pay closer attention to the high nonlinearity domain regions that are more challenging to learn, we also propose adversarial training for generating supplementary high-loss samples, similarly distributed to the original training distribution. In the experiments, our proposed methods are found to be effective and reduce the error on the unseen data points as compared to the previous approaches in various PDE examples, including high-dimensional stochastic PDEs.
2104.14320
737,909
The nanoscale mode volumes of surface plasmon polaritons have enabled plasmonic lasers and condensates with ultrafast operation. Most plasmonic lasers are based on noble metals, rendering the optical mode structure inert to external fields. Here, we demonstrate active magnetic-field control over lasing in a periodic array of Co/Pt multilayer nanodots immersed in an IR-140 dye solution. We exploit magnetic circular dichroism (MCD) at the excitation wavelength to modify the optical absorption of the nanodots as a function of their magnetization. Under circularly polarized excitation, angle-resolved photoluminescence measurements reveal a transition between lasing action and non-lasing emission as the nanodot magnetization is reversed. Our results introduce magnetization as a means of externally controlling plasmonic nanolasers, complementary to the modulation by excitation, gain medium, or substrate. Further, the results show how the effects of magnetization on light that are inherently weak become prominent at the lasing regime, inspiring studies of topological photonics.
2104.14321
737,909
In this paper we continue the discussion about relations between exponential polynomials and generalized moment generating functions on a commutative hypergroup. We are interested in the following problem: is it true that every finite dimensional variety is spanned by moment functions? Let $m$ be an exponential on $X$. In our former paper we have proved that if the linear space of all $m$-sine functions in the variety of an $m$-exponential monomial is (at most) one dimensional, then this variety is spanned by moment functions generated by $m$. In this paper we show that this may happen also in cases where the $m$-sine functions span a more than one dimensional subspace in the variety. We recall the notion of a polynomial hypergroup in $d$ variables, describe exponentials on it and give the characterization of the so called $m$-sine functions. Next we show that the Fourier algebra of a polynomial hypergroup in $d$ variables is the polynomial ring in $d$ variables. Finally, using Ehrenpreis--Palamodov Theorem we show that every exponential polynomial on the polynomial hypergroup in $d$ variables is a linear combination of moment functions contained in its variety.
2104.14322
737,909
JSON is a popular file and data format that is precisely specified by the IETF in RFC 8259. Yet, this specification implicitly and explicitly leaves room for many design choices when it comes to parsing and generating JSON. This yields the opportunity of diverse behavior among independent implementations of JSON libraries. A thorough analysis of this diversity can be used by developers to choose one implementation or to design a resilient multi-version architecture. We present the first systematic analysis and comparison of the input / output behavior of 20 JSON libraries, in Java. We analyze the diversity of architectural choices among libraries, and we execute each library with well-formed and ill-formed JSON files to assess their behavior. We first find that the data structure selected to represent JSON objects and the encoding of numbers are the main design differences, which influence the behavior of the libraries. Second, we observe that the libraries behave in a similar way with regular, well-formed JSON files. However, there is a remarkable behavioral diversity with ill-formed files, or corner cases such as large numbers or duplicate data.
2104.14323
737,909
Motivated by the recent discovery of superconductivity in infinite-layer nickelate thin films, we report on a synthesis and magnetization study on bulk samples of the parent compounds ${R}$NiO$_{2}$ (${R}$=La, Pr, Nd). The frequency-dependent peaks of the AC magnetic susceptibility, along with remarkable memory effects, characterize spin-glass states. Furthermore, various phenomenological parameters via different spin glass models show strong similarity within these three compounds as well as with other rare-earth metal nickelates. The universal spin-glass behaviour distinguishes the nickelates from the parent compound CaCuO$_{2}$ of cuprate superconductors, which has the same crystal structure and $d^9$ electronic configuration but undergoes a long-range antiferromagnetic order. Our investigations may indicate a distinctly different nature of magnetism and superconductivity in the bulk nickelates than in the cuprates.
2104.14324
737,909
We present the far ultraviolet (FUV) imaging of the nearest Jellyfish or Fireball galaxy IC3418/VCC 1217, in the Virgo cluster of galaxies, using Ultraviolet Imaging Telescope (UVIT) onboard the ASTROSAT satellite. The young star formation observed here in the 17 kpc long turbulent wake of IC3418, due to ram pressure stripping of cold gas surrounded by hot intra-cluster medium, is a unique laboratory that is unavailable in the Milkyway. We have tried to resolve star forming clumps, seen compact to GALEX UV images, using better resolution available with the UVIT and incorporated UV-optical images from Hubble Space Telescope archive. For the first time, we resolve the compact star forming clumps (fireballs) into sub-clumps and subsequently into a possibly dozen isolated stars. We speculate that many of them could be blue supergiant stars which are cousins of SDSS J122952.66+112227.8, the farthest star (~17 Mpc) we had found earlier surrounding one of these compact clumps. We found evidence of star formation rate (4 - 7.4 x 10^-4 M_sun per yr ) in these fireballs, estimated from UVIT flux densities, to be increasing with the distance from the parent galaxy. We propose a new dynamical model in which the stripped gas may be developing vortex street where the vortices grow to compact star forming clumps due to self-gravity. Gravity winning over turbulent force with time or length along the trail can explain the puzzling trend of higher star formation rate and bluer/younger stars observed in fireballs farther away from the parent galaxy.
2104.14325
737,909
The polarization properties of the elastic electron scattering on H-like ions are investigated within the framework of the relativistic QED theory. The polarization properties are determined by a combination of relativistic effects and spin exchange between the incident and bound electrons. The scattering of a polarized electron on an initially unpolarized ion is fully described by five parameters. We study these parameters for non-resonant scattering, as well as in the vicinity of LL resonances, where scattering occurs through the formation and subsequent decay of intermediate autoionizing states. The study was carried out for ions from $\txt{B}^{4+}$ to $\txt{Xe}^{53+}$. Special attention was paid to the study of asymmetry in electron scattering.
2104.14326
737,909
Cascade prediction estimates the size or the state of a cascade from either microscope or macroscope. It is of paramount importance for understanding the information diffusion process such as the spread of rumors and the propagation of new technologies in social networks. Recently, instead of extracting hand-crafted features or embedding cascade sequences into feature vectors for cascade prediction, graph neural networks (GNNs) are introduced to utilize the network structure which governs the cascade effect. However, these models do not take into account social factors such as personality traits which drive human's participation in the information diffusion process. In this work, we propose a novel multitask framework for enhancing cascade prediction with a personality recognition task. Specially, we design a general plug-and-play GNN gate, named PersonalityGate, to couple into existing GNN-based cascade prediction models to enhance their effectiveness and extract individuals' personality traits jointly. Experimental results on two real-world datasets demonstrate the effectiveness of our proposed framework in enhancing GNN-based cascade prediction models and in predicting individuals' personality traits as well.
2104.14327
737,909
In this paper, we establish the entropy-entropy production estimate for the ES-BGK model, a generalized version of the BGK model of the Boltzmann equation introduced for better approximation in the Navier-Stokes limit. Our result improves the previous entropy production estimate [39] in that (1) the full range of Prandtl parameters $-1/2\leq\nu <1$ including the critical case $\nu=-1/2$ is covered, and (2) a sharper entropy production bound is obtained. An explicit characterization of the coefficient of the entropy-entropy production estimate is also presented.
2104.14328
737,909
A new method is proposed for the solution of the data-driven optimal transport barycenter problem and of the more general distributional barycenter problem that the article introduces. The method improves on previous approaches based on adversarial games, by slaving the discriminator to the generator, minimizing the need for parameterizations and by allowing the adoption of general cost functions. It is applied to numerical examples, which include analyzing the MNIST data set with a new cost function that penalizes non-isometric maps.
2104.14329
737,909
Models of stellar structure and evolution can be constrained using accurate measurements of the parameters of eclipsing binary members of open clusters. Multiple binary stars provide the means to tighten the constraints and, in turn, to improve the precision and accuracy of the age estimate of the host cluster. In the previous two papers of this series, we have demonstrated the use of measurements of multiple eclipsing binaries in the old open cluster NGC6791 to set tighter constraints on the properties of stellar models than was previously possible, thereby improving both the accuracy and precision of the cluster age. We identify and measure the properties of a non-eclipsing cluster member, V56, in NGC\,6791 and demonstrate how this provides additional model constraints that support and strengthen our previous findings. We analyse multi-epoch spectra of V56 from FLAMES in conjunction with the existing photometry and measurements of eclipsing binaries in NGC6971. The parameters of the V56 components are found to be $M_{\rm p}=1.103\pm 0.008 M_{\odot}$ and $M_{\rm s}=0.974\pm 0.007 M_{\odot}$, $R_{\rm p}=1.764\pm0.099 R_{\odot}$ and $R_{\rm s}=1.045\pm0.057 R_{\odot}$, $T_{\rm eff,p}=5447\pm125$ K and $T_{\rm eff,s}=5552\pm125$ K, and surface [Fe/H]=$+0.29\pm0.06$ assuming that they have the same abundance. The derived properties strengthen our previous best estimate of the cluster age of $8.3\pm0.3$ Gyr and the mass of stars on the lower red giant branch (RGB), which is $M_{\rm RGB} = 1.15\pm0.02M_{\odot}$ for NGC6791. These numbers therefore continue to serve as verification points for other methods of age and mass measures, such as asteroseismology.
2104.14330
737,909
In a previous paper, we computed the energy density and the non-linear energy cascade rate for transverse kink waves using Elsasser variables. In this paper, we focus on the standing kink waves, which are impulsively excited in coronal loops by external perturbations. We present an analytical calculation to compute the damping time due to the non-linear development of the Kelvin-Helmholtz instability. The main result is that the damping time is inversely proportional to the oscillation amplitude. We compare the damping times from our formula with the results of numerical simulations and observations. In both cases we find a reasonably good match. The comparison with the simulations show that the non-linear damping dominates in the high amplitude regime, while the low amplitude regime shows damping by resonant absorption. In the comparison with the observations, we find a power law inversely proportional to the amplitude $\eta^{-1}$ as an outer envelope for our Monte Carlo data points.
2104.14331
737,909
Network dismantling aims to degrade the connectivity of a network by removing an optimal set of nodes and has been widely adopted in many real-world applications such as epidemic control and rumor containment. However, conventional methods usually focus on simple network modeling with only pairwise interactions, while group-wise interactions modeled by hypernetwork are ubiquitous and critical. In this work, we formulate the hypernetwork dismantling problem as a node sequence decision problem and propose a deep reinforcement learning (DRL)-based hypernetwork dismantling framework. Besides, we design a novel inductive hypernetwork embedding method to ensure the transferability to various real-world hypernetworks. Generally, our framework builds an agent. It first generates small-scale synthetic hypernetworks and embeds the nodes and hypernetworks into a low dimensional vector space to represent the action and state space in DRL, respectively. Then trial-and-error dismantling tasks are conducted by the agent on these synthetic hypernetworks, and the dismantling strategy is continuously optimized. Finally, the well-optimized strategy is applied to real-world hypernetwork dismantling tasks. Experimental results on five real-world hypernetworks demonstrate the effectiveness of our proposed framework.
2104.14332
737,909
We present MoonLight, a tool for monitoring temporal and spatio-temporal properties of mobile and spatially distributed cyber-physical systems (CPS). In the proposed framework, space is represented as a weighted graph, describing the topological configurations in which the single CPS entities (nodes of the graph) are arranged. Both nodes and edges have attributes modelling physical and logical quantities that can change in time. MoonLight is implemented in Java and supports the monitoring of Spatio-Temporal Reach and Escape Logic (STREL). MoonLight can be used as a standalone command line tool, as a Java API, or via Matlab interface. We provide here some examples using the Matlab interface and we evaluate the tool performance also by comparing with other tools specialized in monitoring only temporal properties.
2104.14333
737,909
We extend the construction of equilibria for linear-quadratic and mean-variance portfolio problems available in the literature to a large class of mean-field time-inconsistent stochastic control problems in continuous time. Our approach relies on a time discretization of the control problem via n-person games, which are characterized via the maximum principle using Backward Stochastic Differential Equations (BSDEs). The existence of equilibria is proved by applying weak convergence arguments to the solutions of n-person games. A numerical implementation is provided by approximating n-person games using finite Markov chains.
2104.14334
737,909
The dramatic increase in sensitivity, spectral coverage and resolution of radio astronomical facilities in recent years has opened new possibilities for observation of chemical differentiation and isotopic fractionation in protostellar sources to shed light on their spatial and temporal evolution. In warm interstellar environments, methanol is an abundant species, hence spectral data for its isotopic forms are of special interest. In the present work, the millimeter-wave spectrum of the $^{13}$CH$_3$OD isotopologue has been investigated over the region from 150$-$510 GHz to provide a set of transition frequencies for potential astronomical application. The focus is on two types of prominent $^{13}$CH$_3$OD spectral groupings, namely the $a$-type $^qR$-branch multiplets and the $b$-type $Q$-branches. Line positions are reported for the $^qR(J)$ clusters for $J = 3$ to 10 for the $v_{\rm t} = 0$ and 1 torsional states, and for a number of $v_{\rm t} = 0$ and 1 $^rQ(J)$ or $^pQ(J)$ line series up to $J = 25$. The frequencies have been fitted to a multi-parameter torsion-rotation Hamiltonian, and upper level excitation energies have been calculated from the resulting molecular constants.
2104.14340
737,909
Impurities hosted in semiconducting solid matrices represent an extensively studied platform for quantum computing applications. In this scenario, the so-called flip-flop qubit emerges as a convenient choice for scalable implementations in silicon. Flip-flop qubits are realized implanting phosphorous donor in isotopically purified silicon, and encoding the logical states in the donor nuclear spin and in its bound electron. Electrically modulating the hyperfine interaction by applying a vertical electric field causes an Electron Dipole Spin Resonance (EDSR) transition between the states with antiparallel spins $\{|\downarrow\Uparrow\rangle,|\uparrow\Downarrow\rangle\}$, that are chosen as the logical states. When two qubits are considered, the dipole-dipole interaction is exploited allowing long-range coupling between them. A universal set of quantum gates for flip-flop qubits is here proposed and the effect of a realistic 1/f noise on the gate fidelity is investigated for the single qubit $R_z(-\frac{\pi}{2})$ and Hadamard gate and for the two-qubit $\sqrt{iSWAP}$ gate.
2104.14341
737,909
We study the problem of fairly allocating indivisible items to agents with different entitlements, which captures, for example, the distribution of ministries among political parties in a coalition government. Our focus is on picking sequences derived from common apportionment methods, including five traditional divisor methods and the quota method. We paint a complete picture of these methods in relation to known envy-freeness and proportionality relaxations for indivisible items as well as monotonicity properties with respect to the resource, population, and weights. In addition, we provide characterizations of picking sequences satisfying each of the fairness notions, and show that the well-studied maximum Nash welfare solution fails resource- and population-monotonicity even in the unweighted setting. Our results serve as an argument in favor of using picking sequences in weighted fair division problems.
2104.14347
737,909
We investigate the invariance of the Gibbs measure for the fractional Schrodinger equation of exponential type (expNLS) $i\partial_t u + (-\Delta)^{\frac{\alpha}2} u = 2\gamma\beta e^{\beta|u|^2}u$ on $d$-dimensional compact Riemannian manifolds $\mathcal{M}$, for a dispersion parameter $\alpha>d$, some coupling constant $\beta>0$, and $\gamma\neq 0$. (i) We first study the construction of the Gibbs measure for (expNLS). We prove that in the defocusing case $\gamma>0$, the measure is well-defined in the whole regime $\alpha>d$ and $\beta>0$ (Theorem 1.1 (i)), while in the focusing case $\gamma<0$ its partition function is always infinite for any $\alpha>d$ and $\beta>0$, even with a mass cut-off of arbitrary small size (Theorem 1.1 (ii)). (ii) We then study the dynamics (expNLS) with random initial data of low regularity. We first use a compactness argument to prove weak invariance of the Gibbs measure in the whole regime $\alpha>d$ and $0<\beta < \beta^\star_\alpha$ for some natural parameter $0<\beta^\star_\alpha\sim (\alpha-d)$ (Theorem 1.3 (i)). In the large dispersion regime $\alpha>2d$, we can improve this result by constructing a local deterministic flow for (expNLS) for any $\beta>0$. Using the Gibbs measure, we prove that solutions are almost surely global for $0<\beta \ll\beta^\star_\alpha$, and that the Gibbs measure is invariant (Theorem 1.3 (ii)). (iii) Finally, in the particular case $d=1$ and $\mathcal{M}=\mathbb{T}$, we are able to exploit some probabilistic multilinear smoothing effects to build a probabilistic flow for (expNLS) for $1+\frac{\sqrt{2}}2<\alpha \leq 2$, locally for arbitrary $\beta>0$ and globally for $0<\beta \ll \beta^\star_\alpha$ (Theorem 1.5).
2104.14348
737,909
Recognition of hand gestures is one of the most fundamental tasks in human-robot interaction. Sparse representation based methods have been widely used due to their efficiency and low requirements on the training data. Recently, nonconvex regularization techniques including the $\ell_{1-2}$ regularization have been proposed in the image processing community to promote sparsity while achieving efficient performance. In this paper, we propose a vision-based human arm gesture recognition model based on the $\ell_{1-2}$ regularization, which is solved by the alternating direction method of multipliers (ADMM). Numerical experiments on realistic data sets have shown the effectiveness of this method in identifying arm gestures.
2104.14349
737,909
Recent years have seen tremendous progress in the theoretical understanding of quantum systemsdriven dissipatively by coupling them to different baths at their edges. This was possible because of the concurrent advances in the models used to represent these systems, the methods employed, and the analysis of the emerging phenomenology. Here we aim to give a comprehensive review of these three integrated research directions. We first provide an overarching view of the models of boundary driven open quantum systems, both in the weak and strong coupling regimes. This is followed by a review of state-of-the-art analytical and numerical methods, both exact, perturbative and approximate. Finally,we discuss the transport properties of some paradigmatic one-dimensional chains, with an emphasis on disordered and quasiperiodic systems, the emergence of rectification and negative differential conductance, and the role of phase transitions.
2104.14350
737,909
A remarkable consequence of the Hohenberg-Kohn theorem of density functional theory is the existence of an injective map between the electronic density and any observable of the many electron problem in an external potential. In this work, we study the problem of predicting a particular observable, the band gap of semiconductors and band insulators, from the knowledge of the local electronic density. Using state-of-the-art machine learning techniques, we predict the experimental band gaps from computationally inexpensive density functional theory calculations. We propose a modified Behler-Parrinello (BP) architecture that greatly improves the model capacity while maintaining the symmetry properties of the BP architecture. Using this scheme, we obtain band gaps at a level of accuracy comparable to those obtained with state of the art and computationally intensive hybrid functionals, thus significantly reducing the computational cost of the task.
2104.14351
737,909
In this paper, we investigate the problem of prescribing Webster scalar curvatures on compact pseudo-Hermitian manifolds. In terms of the method of upper and lower solutions and the perturbation theory of self-adjoint operators, we can describe some sets of Webster scalar curvature functions which can be realized through pointwise CR conformal deformations and CR conformally equivalent deformations respectively from a given pseudo-Hermitian structure.
2104.14358
737,909
Operational urban transport models require to gather heterogeneous sets of data and often integrate different sub-models. Their systematic validation and reproducible application therefore remains problematic. We propose in this contribution to build transport models from the bottom-up using scientific workflow systems with open-source components and data. These open models are aimed in particular at estimating congestion of public transport in all UK urban areas. This allows us building health indicators related to public transport density in the context of the COVID-19 crisis, and testing related policies.
2104.14359
737,909
Video salient object detection (VSOD) is an important task in many vision applications. Reliable VSOD requires to simultaneously exploit the information from both the spatial domain and the temporal domain. Most of the existing algorithms merely utilize simple fusion strategies, such as addition and concatenation, to merge the information from different domains. Despite their simplicity, such fusion strategies may introduce feature redundancy, and also fail to fully exploit the relationship between multi-level features extracted from both spatial and temporal domains. In this paper, we suggest an adaptive local-global refinement framework for VSOD. Different from previous approaches, we propose a local refinement architecture and a global one to refine the simply fused features with different scopes, which can fully explore the local dependence and the global dependence of multi-level features. In addition, to emphasize the effective information and suppress the useless one, an adaptive weighting mechanism is designed based on graph convolutional neural network (GCN). We show that our weighting methodology can further exploit the feature correlations, thus driving the network to learn more discriminative feature representation. Extensive experimental results on public video datasets demonstrate the superiority of our method over the existing ones.
2104.14360
737,909
This paper provides maximal function characterizations of anisotropic Triebel-Lizorkin spaces associated to general expansive matrices for the full range of parameters $p \in (0,\infty)$, $q \in (0,\infty]$ and $\alpha \in \mathbb{R}$. The equivalent norm is defined in terms of the decay of wavelet coefficients, quantified by a Peetre-type space over a one-parameter dilation group. For the Banach space regime $p,q \geq 1$, we use this characterization to prove the existence of frames and Riesz sequences of dual molecules for the Triebel-Lizorkin spaces; the atoms are obtained by translations and anisotropic dilations of a single function, where neither the translation nor dilation parameters are required to belong to a discrete subgroup. Explicit criteria for molecules are given in terms of smoothness, decay and moment conditions.
2104.14361
737,909
In recent years, data and computing resources are typically distributed in the devices of end users, various regions or organizations. Because of laws or regulations, the distributed data and computing resources cannot be directly shared among different regions or organizations for machine learning tasks. Federated learning emerges as an efficient approach to exploit distributed data and computing resources, so as to collaboratively train machine learning models, while obeying the laws and regulations and ensuring data security and data privacy. In this paper, we provide a comprehensive survey of existing works for federated learning. We propose a functional architecture of federated learning systems and a taxonomy of related techniques. Furthermore, we present the distributed training, data communication, and security of FL systems. Finally, we analyze their limitations and propose future research directions.
2104.14362
737,909
In collaborative robotic cells, a human operator and a robot share the workspace in order to execute a common job, consisting of a set of tasks. A proper allocation and scheduling of the tasks for the human and for the robot is crucial for achieving an efficient human-robot collaboration. In order to deal with the dynamic and unpredictable behavior of the human and for allowing the human and the robot to negotiate about the tasks to be executed, a two layers architecture for solving the task allocation and scheduling problem is proposed. The first layer optimally solves the task allocation problem considering nominal execution times. The second layer, which is reactive, adapts online the sequence of tasks to be executed by the robot considering deviations from the nominal behaviors and requests coming from the human and from robot. The proposed architecture is experimentally validated on a collaborative assembly job.
2104.14363
737,909
How does your brain decide what you will do next? Over the past few decades compelling evidence has emerged that the basal ganglia, a collection of nuclei in the fore- and mid-brain of all vertebrates, are vital to action selection. Gurney, Prescott, and Redgrave published an influential computational account of this idea in Biological Cybernetics in 2001. Here we take a look back at this pair of papers, outlining the "GPR" model contained therein, the context of that model's development, and the influence it has had over the past twenty years. Tracing its lineage into models and theories still emerging now, we are encouraged that the GPR model is that rare thing, a computational model of a brain circuit whose advances were directly built on by others.
2104.14364
737,909
This paper proposes a multivariable extremum seeking scheme using Fast Fourier Transform (FFT) for a network of subsystems working towards optimizing the sum of their local objectives, where the overall objective is the only available measurement. Here, the different inputs are perturbed with different dither frequencies, and the power spectrum of the overall output signal obtained using FFT is used to estimate the steady-state cost gradient w.r.t. each input. The inputs for the subsystems are then updated using integral control in order to drive the respective gradients to zero. This paper provides analytical rules for designing the FFT-based gradient estimation algorithm and analyzes the stability properties of the resulting extremum seeking scheme for the static map setting. The effectiveness of the proposed FFT-based multivariable extremum seeking scheme is demonstrated using two examples, namely, wind farm power optimization problem, and a heat exchanger network for industrial waste-to-heat recovery.
2104.14365
737,909
In this paper, we study the Erd\H{o}s-Falconer distance problem in five dimensions for sets of Cartesian product structure. More precisely, we show that for $A\subset \mathbb{F}_p$ with $|A|\gg p^{\frac{13}{22}}$, then $\Delta(A^5)=\mathbb{F}_p$. When $|A-A|\sim |A|$, we obtain stronger statements as follows: If $|A|\gg p^{\frac{13}{22}}$, then $(A-A)^2+A^2+A^2+A^2+A^2=\mathbb{F}_p.$ If $|A|\gg p^{\frac{4}{7}}$, then $(A-A)^2+(A-A)^2+A^2+A^2+A^2+A^2=\mathbb{F}_p.$
2104.14366
737,909
White dwarfs, the most abundant stellar remnants, provide a promising means of probing dark matter interactions, complimentary to terrestrial searches. The scattering of dark matter from stellar constituents leads to gravitational capture, with important observational consequences. In particular, white dwarf heating occurs due to the energy transfer in the dark matter capture and thermalisation processes, and the subsequent annihilation of captured dark matter. We consider the capture of dark matter by scattering on either the ion or the degenerate electron component of white dwarfs. For ions, we account for the stellar structure, the star opacity, realistic nuclear form factors that go beyond the simple Helm approach, and finite temperature effects pertinent to sub-GeV dark matter. Electrons are treated as relativistic, degenerate targets, with Pauli blocking, finite temperature and multiple scattering effects all taken into account. We also estimate the dark matter evaporation rate. The dark matter-nucleon/electron scattering cross sections can be constrained by comparing the heating rate due to dark matter capture with observations of cold white dwarfs in dark matter-rich environments. We apply this technique to observations of old white dwarfs in the globular cluster Messier 4, which we assume to be located in a DM subhalo. For dark matter-nucleon scattering, we find that white dwarfs can probe the sub-GeV mass range inaccessible to direct detection searches, with the low mass reach limited only by evaporation, and can be competitive with direct detection in the $1-10^4$ GeV range. White dwarf limits on dark matter-electron scattering are found to outperform current electron recoil experiments over the full mass range considered, and extend well beyond the $\sim 10$ GeV mass regime where the sensitivity of electron recoil experiments is reduced.
2104.14367
737,909
Centrality measures identify the most important nodes in a complex network. In recent years, multilayer networks have emerged as a flexible tool to create increasingly realistic models of complex systems. In this paper, we generalize matrix function-based centrality and communicability measures to the case of layer-coupled multiplex networks. We use the supra-adjacency matrix as the network representation, which has already been used to generalize eigenvector centrality to temporal and multiplex networks. With this representation, the definition of single-layer matrix function-based centrality measures in terms of walks on the networks carries over naturally to the multilayer case. Several aggregation techniques allow the ranking of nodes, layers, as well as node-layer pairs in terms of their importance in the network. We present efficient and scalable numerical methods based on Krylov subspace techniques and Gauss quadrature rules, which provide a high accuracy in only a few iterations and which scale linearly in the network size under the assumption of sparsity in the supra-adjacency matrix. Finally, we present extensive numerical studies for both directed and undirected as well as weighted and unweighted multiplex networks. While we focus on social and transportation applications the networks' size ranges between $89$ and $2.28 \cdot 10^6$ nodes and between $3$ and $37$ layers.
2104.14368
737,909
Tuning two dimensional nanomaterial's structural and electronic properties has facilitated the new research paradigm in electronic device applications. In this work, the first principles density functional theory based methods are used to investigate the structural, electronic, and transport properties of an orthorhombic diboron dinitride based polymorph. Interestingly, it depicts a low band gap semiconducting nature with a robust anisotropic behaviour compared to the hexagonal boron nitride, which is an insulator and isotropic. We can also tune the structural and electronic properties of the semiconducting B2N2 based structure through an external inplane mechanical strain. Further, by employing the Landauer Buttiker approach, the electronic transmission function, and electric current calculations reveal that the diboron dinitride based polymorph shows a robust direction dependent anisotropy of the quantum transport properties. We have demonstrated the direction dependence of the electric current in two perpendicular directions, where we have observed an electric current ratio of around 61.75 at 0.8 V. All these findings, such as directional dependence anisotropy in transmission function, current voltage characteristics, and bandgap tunning, suggest that the applicability of such B2N2 based monolayer can be promising for futuristic electronic device applications.
2104.14369
737,909
In this paper, we propose a novel subspace learning framework for one-class classification. The proposed framework presents the problem in the form of graph embedding. It includes the previously proposed subspace one-class techniques as its special cases and provides further insight on what these techniques actually optimize. The framework allows to incorporate other meaningful optimization goals via the graph preserving criterion and reveals spectral and spectral regression-based solutions as alternatives to the previously used gradient-based technique. We combine the subspace learning framework iteratively with Support Vector Data Description applied in the subspace to formulate Graph-Embedded Subspace Support Vector Data Description. We experimentally analyzed the performance of newly proposed different variants. We demonstrate improved performance against the baselines and the recently proposed subspace learning methods for one-class classification.
2104.14370
737,909
In this paper, we introduce structured sparsity estimators in Generalized Linear Models. Structured sparsity estimators in the least squares loss are introduced by Stucky and van de Geer (2018) recently for fixed design and normal errors. We extend their results to debiased structured sparsity estimators with Generalized Linear Model based loss. Structured sparsity estimation means penalized loss functions with a possible sparsity structure used in the chosen norm. These include weighted group lasso, lasso and norms generated from convex cones. The significant difficulty is that it is not clear how to prove two oracle inequalities. The first one is for the initial penalized Generalized Linear Model estimator. Since it is not clear how a particular feasible-weighted nodewise regression may fit in an oracle inequality for penalized Generalized Linear Model, we need a second oracle inequality to get oracle bounds for the approximate inverse for the sample estimate of second-order partial derivative of Generalized Linear Model. Our contributions are fivefold: 1. We generalize the existing oracle inequality results in penalized Generalized Linear Models by proving the underlying conditions rather than assuming them. One of the key issues is the proof of a sample one-point margin condition and its use in an oracle inequality. 2. Our results cover even non sub-Gaussian errors and regressors. 3. We provide a feasible weighted nodewise regression proof which generalizes the results in the literature from a simple l_1 norm usage to norms generated from convex cones. 4. We realize that norms used in feasible nodewise regression proofs should be weaker or equal to the norms in penalized Generalized Linear Model loss. 5. We can debias the first step estimator via getting an approximate inverse of the singular-sample second order partial derivative of Generalized Linear Model loss.
2104.14371
737,909
The underspecification of most machine learning pipelines means that we cannot rely solely on validation performance to assess the robustness of deep learning systems to naturally occurring distribution shifts. Instead, making sure that a neural network can generalize across a large number of different situations requires to understand the specific way in which it solves a task. In this work, we propose to study this problem from a geometric perspective with the aim to understand two key characteristics of neural network solutions in underspecified settings: how is the geometry of the learned function related to the data representation? And, are deep networks always biased towards simpler solutions, as conjectured in recent literature? We show that the way neural networks handle the underspecification of these problems is highly dependent on the data representation, affecting both the geometry and the complexity of the learned predictors. Our results highlight that understanding the architectural inductive bias in deep learning is fundamental to address the fairness, robustness, and generalization of these systems.
2104.14372
737,909
Heat engines are fundamental physical objects to develop nonequilibrium thermodynamics. The thermodynamic performance of the heat engine is determined by the choice of cycle and time-dependence of parameters. Here, we propose a systematic numerical method to find a heat engine cycle to optimize some target functions. We apply the method to heat engines with slowly varying parameters and show that the method works well. Our numerical method is based on the genetic algorithm which is widely applied to various optimization problems.
2104.14373
737,909
Benefitting from insensitivity to light and high penetration of foggy environments, infrared cameras are widely used for sensing in nighttime traffic scenes. However, the low contrast and lack of chromaticity of thermal infrared (TIR) images hinder the human interpretation and portability of high-level computer vision algorithms. Colorization to translate a nighttime TIR image into a daytime color (NTIR2DC) image may be a promising way to facilitate nighttime scene perception. Despite recent impressive advances in image translation, semantic encoding entanglement and geometric distortion in the NTIR2DC task remain under-addressed. Hence, we propose a toP-down attEntion And gRadient aLignment based GAN, referred to as PearlGAN. A top-down guided attention module and an elaborate attentional loss are first designed to reduce the semantic encoding ambiguity during translation. Then, a structured gradient alignment loss is introduced to encourage edge consistency between the translated and input images. In addition, pixel-level annotation is carried out on a subset of FLIR and KAIST datasets to evaluate the semantic preservation performance of multiple translation methods. Furthermore, a new metric is devised to evaluate the geometric consistency in the translation process. Extensive experiments demonstrate the superiority of the proposed PearlGAN over other image translation methods for the NTIR2DC task. The source code and labeled segmentation masks will be available at \url{https://github.com/FuyaLuo/PearlGAN/}.
2104.14374
737,909
One of the most common problems of weakly supervised object localization is that of inaccurate object coverage. In the context of state-of-the-art methods based on Class Activation Mapping, this is caused either by localization maps which focus, exclusively, on the most discriminative region of the objects of interest or by activations occurring in background regions. To address these two problems, we propose two representation regularization mechanisms: Full Region Regularizationwhich tries to maximize the coverage of the localization map inside the object region, and Common Region Regularization which minimizes the activations occurring in background regions. We evaluate the two regularizations on the ImageNet, CUB-200-2011 and OpenImages-segmentation datasets, and show that the proposed regularizations tackle both problems, outperforming the state-of-the-art by a significant margin.
2104.14375
737,909
Thermal dissociation and recombination of molecular hydrogen, H_2, in the atmospheres of ultra-hot Jupiters (UHJs) has been shown to play an important role in global heat redistribution. This, in turn, significantly impacts their planetary emission, yet only limited investigations on the atmospheric effects have so far been conducted. Here we investigate the heat redistribution caused by this dissociation/recombination reaction, alongside feedback mechanisms between the atmospheric chemistry and radiative transfer, for a planetary and stellar configuration typical of UHJs. To do this, we have developed a time-dependent pseudo-2D model, including a treatment of time-independent equilibrium chemical effects. As a result of the reaction heat redistribution, we find temperature changes of up to $\sim$400 K in the atmosphere. When TiO and VO are additionally considered as opacity sources, these changes in temperature increase to over $\sim$800 K in some areas. This heat redistribution is found to significantly shift the region of peak atmospheric temperature, or hotspot, towards the evening terminator in both cases. The impact of varying the longitudinal wind speed on the reaction heat distribution is also investigated. When excluding TiO/VO, increased wind speeds are shown to increase the impact of the reaction heat redistribution up to a threshold wind speed. When including TiO/VO there is no apparent wind speed threshold, due to thermal stabilisation by these species. We also construct pseudo-2D phase curves from our model, and highlight both significant spectral flux damping and increased phase offset caused by the reaction heat redistribution.
2104.14376
737,909
We generalize solid-state tight-binding techniques for the spectral analysis of large superconducting circuits. We find that tight-binding states can be better suited for approximating the low-energy excitations than charge-basis states, as illustrated for the interesting example of the current-mirror circuit. The use of tight binding can dramatically lower the Hilbert space dimension required for convergence to the true spectrum, and allows for the accurate simulation of larger circuits that are out of reach of charge basis diagonalization.
2104.14377
737,909
In this study, the stability dependence of turbulent Prandtl number ($Pr_t$) is quantified via a simple analytical approach. Based on the conventional budget equations, a hybrid length scale formulation is first proposed and its functional relationships to well-known length scales are established. Next, the ratios of these length scales are utilized to derive an explicit relationship between $Pr_t$ and gradient Richardson number. The results predicted by the proposed formulation are compared against other competing formulations as well as published datasets.
2104.14378
737,909
We propose a new approach to train a variational information bottleneck (VIB) that improves its robustness to adversarial perturbations. Unlike the traditional methods where the hard labels are usually used for the classification task, we refine the categorical class information in the training phase with soft labels which are obtained from a pre-trained reference neural network and can reflect the likelihood of the original class labels. We also relax the Gaussian posterior assumption in the VIB implementation by using the mutual information neural estimation. Extensive experiments have been performed with the MNIST and CIFAR-10 datasets, and the results show that our proposed approach significantly outperforms the benchmarked models.
2104.14379
737,909
We propose and implement a Privacy-preserving Federated Learning (PPFL) framework for mobile systems to limit privacy leakages in federated learning. Leveraging the widespread presence of Trusted Execution Environments (TEEs) in high-end and mobile devices, we utilize TEEs on clients for local training, and on servers for secure aggregation, so that model/gradient updates are hidden from adversaries. Challenged by the limited memory size of current TEEs, we leverage greedy layer-wise training to train each model's layer inside the trusted area until its convergence. The performance evaluation of our implementation shows that PPFL can significantly improve privacy while incurring small system overheads at the client-side. In particular, PPFL can successfully defend the trained model against data reconstruction, property inference, and membership inference attacks. Furthermore, it can achieve comparable model utility with fewer communication rounds (0.54x) and a similar amount of network traffic (1.002x) compared to the standard federated learning of a complete model. This is achieved while only introducing up to ~15% CPU time, ~18% memory usage, and ~21% energy consumption overhead in PPFL's client-side.
2104.14380
737,909
We study the probability that an $(n - m)$-dimensional linear subspace in $\mathbb{P}^n$ or a collection of points spanning such a linear subspace is contained in an $m$-dimensional variety $Y \subset \mathbb{P}^n$. This involves a strategy used by Galkin--Shinder to connect properties of a cubic hypersurface to its Fano variety of lines via cut and paste relations in the Grothendieck ring of varieties. Generalizing this idea to varieties of higher codimension and degree, we can measure growth rates of weighted probabilities of $k$-planes contained in a sequence of varieties with varying initial parameters over a finite field. In the course of doing this, we move an identity motivated by rationality problems involving cubic hypersurfaces to a motivic statistics setting associated with cohomological stability.
2104.14381
737,909
We investigate the rotating quark matter in the three-flavor Nambu and Jona-Lasinio (NJL) model. The chiral condensation, spin polarization and number susceptibility of strange quark are carefully studied at finite temperature without or with finite chemical potential in this model. We find that the rotation suppresses the chiral condensation and enhances the first-order quark spin polarization, however for the second-order quark spin polarization and quark number susceptibility the effect is very interesting, in the case of zero chemical potential which have a jump structure when the first-order phase transitions take place. When extending to the situation with finite chemical potential, we find the angular velocity also plays a crucial role, at small or large enough angular velocity the chemical potential enhances the susceptibility, however in the middle region of angular velocity the effect of the chemical potential is suppressed by the angular velocity and susceptibility can be changed considerably, which can be also observed that the quark number susceptibility has two maximum value. Furthermore, it is found that at sufficiently large angular velocity the contributions played by light quark and strange quark to these phenomena are almost equal. We expect these studies to be used to understand the chiral symmetry breaking and restoration as well as probe the QCD phase transition.
2104.14382
737,909
Real-world data is usually segmented by attributes and distributed across different parties. Federated learning empowers collaborative training without exposing local data or models. As we demonstrate through designed attacks, even with a small proportion of corrupted data, an adversary can accurately infer the input attributes. We introduce an adversarial learning based procedure which tunes a local model to release privacy-preserving intermediate representations. To alleviate the accuracy decline, we propose a defense method based on the forward-backward splitting algorithm, which respectively deals with the accuracy loss and privacy loss in the forward and backward gradient descent steps, achieving the two objectives simultaneously. Extensive experiments on a variety of datasets have shown that our defense significantly mitigates privacy leakage with negligible impact on the federated learning task.
2104.14383
737,909
Motivated by the quantum speedup for dynamic programming on the Boolean hypercube by Ambainis et al. (2019), we investigate which graphs admit a similar quantum advantage. In this paper, we examine a generalization of the Boolean hypercube graph, the $n$-dimensional lattice graph $Q(D,n)$ with vertices in $\{0,1,\ldots,D\}^n$. We study the complexity of the following problem: given a subgraph $G$ of $Q(D,n)$ via query access to the edges, determine whether there is a path from $0^n$ to $D^n$. While the classical query complexity is $\widetilde{\Theta}((D+1)^n)$, we show a quantum algorithm with complexity $\widetilde O(T_D^n)$, where $T_D < D+1$. The first few values of $T_D$ are $T_1 \approx 1.817$, $T_2 \approx 2.660$, $T_3 \approx 3.529$, $T_4 \approx 4.421$, $T_5 \approx 5.332$ (the $D=1$ case corresponds to the hypercube and replicates the result of Ambainis et al.). We then show an implementation of this algorithm with time complexity $\text{poly}(n)^{\log n} T_D^n$, and apply it to the Set Multicover problem. In this problem, $m$ subsets of $[n]$ are given, and the task is to find the smallest number of these subsets that cover each element of $[n]$ at least $D$ times. While the time complexity of the best known classical algorithm is $O(m(D+1)^n)$, the time complexity of our quantum algorithm is $\text{poly}(m,n)^{\log n} T_D^n$.
2104.14384
737,909
Few-shot classification aims to recognize unseen classes with few labeled samples from each class. Many meta-learning models for few-shot classification elaborately design various task-shared inductive bias (meta-knowledge) to solve such tasks, and achieve impressive performance. However, when there exists the domain shift between the training tasks and the test tasks, the obtained inductive bias fails to generalize across domains, which degrades the performance of the meta-learning models. In this work, we aim to improve the robustness of the inductive bias through task augmentation. Concretely, we consider the worst-case problem around the source task distribution, and propose the adversarial task augmentation method which can generate the inductive bias-adaptive 'challenging' tasks. Our method can be used as a simple plug-and-play module for various meta-learning models, and improve their cross-domain generalization capability. We conduct extensive experiments under the cross-domain setting, using nine few-shot classification datasets: mini-ImageNet, CUB, Cars, Places, Plantae, CropDiseases, EuroSAT, ISIC and ChestX. Experimental results show that our method can effectively improve the few-shot classification performance of the meta-learning models under domain shift, and outperforms the existing works.
2104.14385
737,909
Visual domain randomization in simulated environments is a widely used method to transfer policies trained in simulation to real robots. However, domain randomization and augmentation hamper the training of a policy. As reinforcement learning struggles with a noisy training signal, this additional nuisance can drastically impede training. For difficult tasks it can even result in complete failure to learn. To overcome this problem we propose to pre-train a perception encoder that already provides an embedding invariant to the randomization. We demonstrate that this yields consistently improved results on a randomized version of DeepMind control suite tasks and a stacking environment on arbitrary backgrounds with zero-shot transfer to a physical robot.
2104.14386
737,909
The $\phi^4$ double-well theory admits a kink solution, whose rich phenomenology is strongly affected by the existence of a single bound excitation called the shape mode. We find that the leading quantum correction to the energy needed to excite the shape mode is $-0.115567\lambda/m$ in terms of the coupling $\lambda/4$ and the meson mass $m$ evaluated at the minimum of the potential. On the other hand, the correction to the continuum threshold is $-0.433\lambda/m$. A naive extrapolation to finite coupling then suggests that the shape mode melts into the continuum at the modest coupling of $\lambda/4\sim 0.106 m^2$, where the $\mathbb{Z}_2$ symmetry is still broken.
2104.14387
737,909