Datasets:

sswt

/

arxiv_sample_50K

like 0

BornInfeldtype electrodynamics and magnetic black holes ; We investigate a BornInfeldtype model of nonlinear electrodynamics, possessing three parameters, coupled with general relativity. As a particular case BornInfeld electrodynamics is reproduced. There is no singularity of the electric field at the centre of pointlike charged particles and selfenergy of charges is finite in this model. The magnetized black hole is studied and solutions are obtained. We demonstrate that for some parameters of the model the black hole is regular. We find the asymptotic of the metric and mass functions at rrightarrowinfty and rrightarrow 0, and corrections to the ReissnerNordstrom solution. Thermodynamics of black holes is investigated. We calculate the Hawking temperature of black holes and show that black holes are not stable and there are phase transitions in the model under consideration.

A Point Process Model for Generating Biofilms with Realistic Microstructure and Rheology ; Biofilms are communities of bacteria that exhibit a multitude of multiscale biomechanical behaviors. Recent experimental advances have lead to characterizations of these behaviors in terms of measurements of the viscoelastic moduli of biofilms grown in bioreactors and the fracture and fragmentation properties of biofilms. These properties are macroscale features of biofilms; however, a previous work by our group has shown that heterogeneous microscale features are critical in predicting biofilm rheology. In this paper we use tools from statistical physics to develop a generative statistical model of the positions of bacteria in biofilms. We show through simulation that the macroscopic mechanical properties of biofilms depend on the choice of microscale spatial model. Our key finding is that a biologically inspired model of the locations of bacteria in a biofilm is critical to the simulation of biofilms with realistic in silico mechanical properties and statistical characteristics.

Correct Composition of Dephased Behavioural Models ; Scenarios of execution are commonly used to specify partial behaviour and interactions between different objects and components in a system. To avoid overall inconsistency in specifications, various automated methods have emerged in the literature to compose behavioural models. In recent work, we have shown how the theorem prover Isabelle can be combined with the constraint solver Z3 to efficiently detect inconsistencies in two or more behavioural models and, in their absence, generate the composition. Here, we extend our approach further and show how to generate the correct composition as a set of valid traces of dephased models. This work has been inspired by a problem from a medical domain where different care pathways for chronic conditions may be applied to the same patient with different starting points.

Cut, Paste and Learn Surprisingly Easy Synthesis for Instance Detection ; A major impediment in rapidly deploying object detection models for instance detection is the lack of large annotated datasets. For example, finding a large labeled dataset containing instances in a particular kitchen is unlikely. Each new environment with new instances requires expensive data collection and annotation. In this paper, we propose a simple approach to generate large annotated instance datasets with minimal effort. Our key insight is that ensuring only patchlevel realism provides enough training signal for current object detector models. We automatically cut' object instances and paste' them on random backgrounds. A naive way to do this results in pixel artifacts which result in poor performance for trained models. We show how to make detectors ignore these artifacts during training and generate data that gives competitive performance on real data. Our method outperforms existing synthesis approaches and when combined with real images improves relative performance by more than 21 on benchmark datasets. In a crossdomain setting, our synthetic data combined with just 10 real data outperforms models trained on all real data.

Overview of Millimeter Wave Communications for FifthGeneration 5G Wireless Networkswith a focus on Propagation Models ; This paper provides an overview of the features of fifth generation 5G wireless communication systems now being developed for use in the millimeter wave mmWave frequency bands. Early results and key concepts of 5G networks are presented, and the channel modeling efforts of many international groups for both licensed and unlicensed applications are described here. Propagation parameters and channel models for understanding mmWave propagation, such as lineofsight LOS probabilities, largescale path loss, and building penetration loss, as modeled by various standardization bodies, are compared over the 0.5100 GHz range.

WebVision Database Visual Learning and Understanding from Web Data ; In this paper, we present a study on learning visual recognition models from large scale noisy web data. We build a new database called WebVision, which contains more than 2.4 million web images crawled from the Internet by using queries generated from the 1,000 semantic concepts of the benchmark ILSVRC 2012 dataset. Meta information along with those web images e.g., title, description, tags, etc. are also crawled. A validation set and test set containing human annotated images are also provided to facilitate algorithmic development. Based on our new database, we obtain a few interesting observations 1 the noisy web images are sufficient for training a good deep CNN model for visual recognition; 2 the model learnt from our WebVision database exhibits comparable or even better generalization ability than the one trained from the ILSVRC 2012 dataset when being transferred to new datasets and tasks; 3 a domain adaptation issue a.k.a., dataset bias is observed, which means the dataset can be used as the largest benchmark dataset for visual domain adaptation. Our new WebVision database and relevant studies in this work would benefit the advance of learning stateoftheart visual models with minimum supervision based on web data.

Droplet breakup in the liquid drop model with background potential ; We consider a variant of Gamow's liquid drop model, with a general repulsive Riesz kernel and a longrange attractive background potential with weight Z. The addition of the background potential acts as a regularization for the liquid drop model in that it restores the existence of minimizers for arbitrary mass. We consider the regime of small Z and characterize the structure of minimizers in the limit Zto 0 by means of a sharp asymptotic expansion of the energy. In the process of studying this limit we characterize all minimizing sequences for the Gamow model in terms of generalized minimizers.

A Conditional Model of Wind Power Forecast Errors and Its Application in Scenario Generation ; In power system operation, characterizing the stochastic nature of wind power is an important albeit challenging issue. It is well known that distributions of wind power forecast errors often exhibit significant variability with respect to different forecast values. Therefore, appropriate probabilistic models that can provide accurate information for conditional forecast error distributions are of great need. On the basis of Gaussian mixture model, this paper constructs analytical conditional distributions of forecast errors for multiple wind farms with respect to forecast values. The accuracy of the proposed probabilistic models is verified by using historical data. Thereafter, a fast sampling method is proposed to generate scenarios from the conditional distributions which are nonGaussian and interdependent. The efficiency of the proposed sampling method is verified.

Interactive Attention Networks for AspectLevel Sentiment Classification ; Aspectlevel sentiment classification aims at identifying the sentiment polarity of specific target in its context. Previous approaches have realized the importance of targets in sentiment classification and developed various methods with the goal of precisely modeling their contexts via generating targetspecific representations. However, these studies always ignore the separate modeling of targets. In this paper, we argue that both targets and contexts deserve special treatment and need to be learned their own representations via interactive learning. Then, we propose the interactive attention networks IAN to interactively learn attentions in the contexts and targets, and generate the representations for targets and contexts separately. With this design, the IAN model can well represent a target and its collocative context, which is helpful to sentiment classification. Experimental results on SemEval 2014 Datasets demonstrate the effectiveness of our model.

Search for supersymmetric partners of third generation quarks in leptonic channels with the ATLAS detector ; Two of the most important parameters in supersymmetry are the masses of the stop and sbottom, the supersymmetric partners of the third generation quarks. A stop mass lighter than 1 TeV is favored theoretically, however, conventional searches based on the simplified models have not produced experimental evidence for a light stop. It is possible that the light stop evades our searches due to a compressed sparticle mass spectrum. Therefore, the searches are extended to cover a broader range of signal scenarios with different mass splittings between the stop, neutralinos, and charginos. The searches are then interpreted in the context of both simplified models and pMSSM models. Searches are also performed for various Rparity violating stop sbottom models. Recent ATLAS results from searches for direct stop sbottom pair production are presented in final states with jets, missing transversemomentum, and leptons. The analyses are based on 36 fb1 of sqrts13 TeV protonproton collision data recorded with ATLAS detector at the LHC in 2015 and 2016.

Exploring Humanlike Attention Supervision in Visual Question Answering ; Attention mechanisms have been widely applied in the Visual Question Answering VQA task, as they help to focus on the areaofinterest of both visual and textual information. To answer the questions correctly, the model needs to selectively target different areas of an image, which suggests that an attentionbased model may benefit from an explicit attention supervision. In this work, we aim to address the problem of adding attention supervision to VQA models. Since there is a lack of human attention data, we first propose a Human Attention Network HAN to generate humanlike attention maps, training on a recently released dataset called Human ATtention Dataset VQAHAT. Then, we apply the pretrained HAN on the VQA v2.0 dataset to automatically produce the humanlike attention maps for all imagequestion pairs. The generated humanlike attention map dataset for the VQA v2.0 dataset is named as HumanLike ATtention HLAT dataset. Finally, we apply humanlike attention supervision to an attentionbased VQA model. The experiments show that adding humanlike supervision yields a more accurate attention together with a better performance, showing a promising future for humanlike attention supervision in VQA.

Mitigating the Impact of Speech Recognition Errors on Chatbot using SequencetoSequence Model ; We apply sequencetosequence model to mitigate the impact of speech recognition errors on open domain endtoend dialog generation. We cast the task as a domain adaptation problem where ASR transcriptions and original text are in two different domains. In this paper, our proposed model includes two individual encoders for each domain data and make their hidden states similar to ensure the decoder predict the same dialog text. The method shows that the sequencetosequence model can learn the ASR transcriptions and original text pair having the same meaning and eliminate the speech recognition errors. Experimental results on Cornell movie dialog dataset demonstrate that the domain adaption system help the spoken dialog system generate more similar responses with the original text answers.

Exploring stellar evolution with gravitationalwave observations ; Recent detections of gravitational waves from merging binary black holes opened new possibilities to study the evolution of massive stars and black hole formation. In particular, stellar evolution models may be constrained on the basis of the differences in the predicted distribution of black hole masses and redshifts. In this work we propose a framework that combines galaxy and stellar evolution models and use it to predict the detection rates of merging binary black holes for various stellar evolution models. We discuss the prospects of constraining the shape of the time delay distribution of merging binaries using just the observed distribution of chirp masses. Finally, we consider a generic model of primordial black hole formation and discuss the possibility of distinguishing it from stellarorigin black holes.

On Categorical Time Series Models With Covariates ; We study the problem of stationarity and ergodicity for autoregressive multinomial logistic time series models which possibly include a latent process and are defined by a GARCHtype recursive equation. We improve considerably upon the existing results related to stationarity and ergodicity conditions of such models. Proofs are based on theory developed for chains with complete connections. This approach is based on a useful coupling technique which is utilized for studying ergodicity of more general finitestate stochastic processes. Such processes generalize finitestate Markov chains by assuming infinite order models of past values. For finite order Markov chains, we also discuss ergodicity properties when some strongly exogenous covariates are considered in the dynamics of the process.

Random Overlapping Communities Approximating Motif Densities of Large Graphs ; A wide variety of complex networks social, biological, information etc. exhibit local clustering with substantial variation in the clustering coefficient the probability of neighbors being connected. Existing models of large graphs capture power law degree distributions Barab'asiAlbert and smallworld properties WattsStrogatz, but only limited clustering behavior. We introduce a generalization of the classical ErdHosR'enyi model of random graphs which provably achieves a wide range of desired clustering coefficient, triangletoedge and fourcycletoedge ratios for any given graph size and edge density. Rather than choosing edges independently at random, in the Random Overlapping Communities model, a graph is generated by choosing a set of random, relatively dense subgraphs communities. We discuss the explanatory power of the model and some of its consequences.

Localized Mode and Nonergodicity of a Harmonic Oscillator Chain ; We present a simple and microscopic physical model that breaks the ergodicity. Our model consists of coupled classical harmonic oscillators, and the motion of the tagged particle obeys the generalized Langevin equation satisfying the second fluctuation dissipation theorem. It is found that although the nonergodicity strength, which is expected to detect the ergodicity breaking, for this model vanishes, the velocity auto correlation function of the tagged particle asymptotically oscillates. We analyze the model by using the molecular dynamics and the exact diagonalization as well as the rigorous mapping to the generalized Langevin equation. Our analysis reveals that the asymptotic oscillation is caused by a localized mode with an isolated frequency from the continuous phonon spectrum.

Cosmological constraints on gravity models ; In this paper we place observational constraints on the wellknown gammagravity fR model using the latest cosmological data, namely we use the latest growth rate, Cosmic Microwave Background, Baryon Acoustic Oscillations, Supernovae type Ia and Hubble parameter data. Performing a joint likelihood analysis we find that the gammagravity model is in very good agreement with observations. Utilizing the AIC statistical test we statistically compare the current fR model with LambdaCDM cosmology and find that they are statistically equivalent. Therefore, gammagravity can be seen as a useful scenario toward testing deviations from General Relativity. Finally, we note that we find somewhat higher values for the fR bestfit values compared to those mentioned in the past in the literature and we argue that this could potential alleviate the halomass function problem.

Impact of humanhuman contagions in the spread of vectorborne diseases ; This article is aimed at proposing a generalization of the RossMacdonald model for the transmission of Vectorborne diseases in which humantohuman contagions are also considered. We first present this generalized model by formulating a mean field theory, checking its validity by comparing to numerical simulations. To make the premises of our model more realistic, we adapt the mean field equations to the case in which human contacts are described by a complex network. In this case we are also able to derive an analytical expression for the epidemic threshold. In both the meanfield and networkbased models, we estimate the value of the epidemic threshold which corresponds to the boundary between the diseasefree and epidemic regimes. The expression of this threshold allows us to discuss the impact that humantohuman contagions have on the spread of vectorborne diseases.

Random super matrices with an external source ; In the past we have considered Gaussian random matrix ensembles in the presence of an external matrix source. The reason was that it allowed, through an appropriate tuning of the eigenvalues of the source, to obtain results on nontrivial dual models, such as Kontsevich's Airy matrix models and generalizations. The techniques relied on explicit computations of the kpoint functions for arbitrary N the size of the matrices and on an Nk duality. Numerous results on the intersection numbers of the moduli space of curves were obtained by this technique. In order to generalize these results to include surfaces with boundaries, we have extended these techniques to supermatrices. Again we have obtained quite remarkable explicit expressions for the kpoint functions, as well as a duality. Although supermatrix models a priori lead to the same matrix models of 2dgravity, the external source extensions considered in this article lead to new geometric results.

Multiwavelength Observations of Relativistic Jets from General Relativistic Magnetohydrodynamic Simulations ; This work summarizes a program intended to unify three burgeoning branches of the highenergy astrophysics of relativistic jets general relativistic magnetohydrodynamic GRMHD simulations of everincreasing dynamical range, the microphysical theory of particle acceleration under relativistic conditions, and multiwavelength observations resolving everdecreasing spatiotemporal scales. The process, which involves converting simulation output into time series of images and polarization maps that can be directly compared to observations, is performed by 1 selfconsistently prescribing models for emission, absorption, and particle acceleration and 2 performing timedependent polarized radiative transfer. M87 serves as an exemplary prototype for this investigation due to its prominent and wellstudied jet and the imminent prospect of learning much more from Event Horizon Telescope EHT observations this year. Synthetic observations can be directly compared with real observations for observational signatures such as jet instabilities, collimation, relativistic beaming, and polarization. The simplest models described adopt the standard equipartition hypothesis; other models calculate emission by relating it to current density or shear. These models are intended for application to the radio jet instead of the higher frequency emission, the disk and the wind, which will be subjects of future investigations.

RETURNN as a Generic Flexible Neural Toolkit with Application to Translation and Speech Recognition ; We compare the fast training and decoding speed of RETURNN of attention models for translation, due to fast CUDA LSTM kernels, and a fast pure TensorFlow beam search decoder. We show that a layerwise pretraining scheme for recurrent attention models gives over 1 BLEU improvement absolute and it allows to train deeper recurrent encoder networks. Promising preliminary results on max. expected BLEU training are presented. We are able to train stateoftheart models for translation and endtoend models for speech recognition and show results on WMT 2017 and Switchboard. The flexibility of RETURNN allows a fast research feedback loop to experiment with alternative architectures, and its generality allows to use it on a wide range of applications.

LiebSchultzMattis type theorem with higherform symmetry and the quantum dimer models ; The LiebSchultzMattis theorem dictates that a trivial symmetric insulator in lattice models is prohibited if lattice translation symmetry and U1 charge conservation are both preserved. In this paper, we generalize the LiebSchultzMattis theorem to systems with higherform symmetries, which act on extended objects of dimension n 0. The prototypical lattice system with higherform symmetry is the pure abelian lattice gauge theory whose action consists only of the field strength. We first construct the higherform generalization of the LiebSchultzMattis theorem with a proof. We then apply it to the U1 lattice gauge theory description of the quantum dimer model on bipartite lattices. Finally, using the continuum field theory description in the vicinity of the RokhsarKivelson point of the quantum dimer model, we diagnose and compute the mixed 't Hooft anomaly corresponding to the higherform LiebSchultzMattis theorem.

A SemiAnalytical Line Transfer SALT Model II the effects of a BiConical geometry ; We generalize the semianalytical line transfer SALT model recently introduced by Scarlata Panagia 2015 for modeling galactic outflows, to account for biconical geometries of various opening angles and orientations with respect to the lineofsight to the observer, as well as generalized velocity fields. We model the absorption and emission component of the line profile resulting from resonant absorption in the biconical outflow. We show how the outflow geometry impacts the resulting line profile. We use simulated spectra with different geometries and velocity fields to study how well the outflow parameters can be recovered. We find that geometrical parameters including the opening angle and the orientation are always well recovered. The density and velocity field parameters are reliably recovered when both an absorption and an emission component are visible in the spectra. This condition implies that the velocity and density fields for narrow cones oriented perpendicular to the line of sight will remain unconstrained.

Approximating the Void Learning Stochastic Channel Models from Observation with Variational Generative Adversarial Networks ; Channel modeling is a critical topic when considering designing, learning, or evaluating the performance of any communications system. Most prior work in designing or learning new modulation schemes has focused on using highly simplified analytic channel models such as additive white Gaussian noise AWGN, Rayleigh fading channels or similar. Recently, we proposed the usage of a generative adversarial networks GANs to jointly approximate a wireless channel response model e.g. from real black box measurements and optimize for an efficient modulation scheme over it using machine learning. This approach worked to some degree, but was unable to produce accurate probability distribution functions PDFs representing the stochastic channel response. In this paper, we focus specifically on the problem of accurately learning a channel PDF using a variational GAN, introducing an architecture and loss function which can accurately capture stochastic behavior. We illustrate where our prior method failed and share results capturing the performance of such as system over a range of realistic channel distributions.

FLRW cosmological models with quark and strange quark matters in fR,T gravity ; In this paper, we have studied the magnetized quark matter QM and strange quark matter SQM distributions in the presence of fR,T gravity in the background of FriedmannLemaitreRobertsonWalker FLRW metric. To get exact solutions of modified field equations we have used fR,T R 2 fT model given by Harko et al. with two different parametrization of geometrical parameters textiti.e. the parametrization of the deceleration parameter q , and the scale factor a in hybrid expansion form. Also, we have obtained Einstein Static Universe ESU solutions for QM and SQM distributions in fR,T gravity and General Relativity GR. All models in fR,T gravity and GR for FRW and ESU Universes with QM also SQM distributions, we get zero magnetic field. These results agree with the solutions of Aktacs and Aygun in fR,T gravity. However, we have also discussed the physical consequences of our obtained models.

TwoStage Residual Inclusion under the Additive Hazards Model An Instrumental Variable Approach with Application to SEERMedicare Linked Data ; Instrumental variable is an essential tool for addressing unmeasured confounding in observational studies. Two stage predictor substitution 2SPS estimator and two stage residual inclusion2SRI are two commonly used approaches in applying instrumental variables. Recently 2SPS was studied under the additive hazards model in the presence of competing risks of timetoevents data, where linearity was assumed for the relationship between the treatment and the instrument variable. This assumption may not be the most appropriate when we have binary treatments. In this paper, we consider the 2SRI estimator under the additive hazards model for general survival data and in the presence of competing risks, which allows generalized linear models for the relation between the treatment and the instrumental variable. We derive the asymptotic properties including a closedform asymptotic variance estimate for the 2SRI estimator. We carry out numerical studies in finite samples, and apply our methodology to the linked Surveillance, Epidemiology and End Results SEER Medicare database comparing radical prostatectomy versus conservative treatment in earlystage prostate cancer patients.

General multilevel Monte Carlo methods for pricing discretely monitored Asian options ; We describe general multilevel Monte Carlo methods that estimate the price of an Asian option monitored at m fixed dates. Our approach yields unbiased estimators with standard deviation Oepsilon in Om 1epsilon2 expected time for a variety of processes including the BlackScholes model, Merton's jumpdiffusion model, the SquareRoot diffusion model, Kou's double exponential jumpdiffusion model, the variance gamma and NIG exponential Levy processes and, via the Milstein scheme, processes driven by scalar stochastic differential equations. Using the Euler scheme, our approach estimates the Asian option price with root mean square error Oepsilon in Omlnepsilonepsilon2 expected time for processes driven by multidimensional stochastic differential equations. Numerical experiments confirm that our approach outperforms the conventional Monte Carlo method by a factor of order m.

Adversarial Constraint Learning for Structured Prediction ; Constraintbased learning reduces the burden of collecting labels by having users specify general properties of structured outputs, such as constraints imposed by physical laws. We propose a novel framework for simultaneously learning these constraints and using them for supervision, bypassing the difficulty of using domain expertise to manually specify constraints. Learning requires a blackbox simulator of structured outputs, which generates valid labels, but need not model their corresponding inputs or the inputlabel relationship. At training time, we constrain the model to produce outputs that cannot be distinguished from simulated labels by adversarial training. Providing our framework with a small number of labeled inputs gives rise to a new semisupervised structured prediction model; we evaluate this model on multiple tasks tracking, pose estimation and time series prediction and find that it achieves high accuracy with only a small number of labeled inputs. In some cases, no labels are required at all.

A Stochastic Decoder for Neural Machine Translation ; The process of translation is ambiguous, in that there are typically many valid trans lations for a given sentence. This gives rise to significant variation in parallel cor pora, however, most current models of machine translation do not account for this variation, instead treating the prob lem as a deterministic process. To this end, we present a deep generative model of machine translation which incorporates a chain of latent variables, in order to ac count for local lexical and syntactic varia tion in parallel corpora. We provide an in depth analysis of the pitfalls encountered in variational inference for training deep generative models. Experiments on sev eral different language pairs demonstrate that the model consistently improves over strong baselines.

Hidden symmetries in mixmastertype universe ; A model of multidimensional mixmastertype vacuum universe is considered. It belongs to a class of pseudoEuclidean chains characterized by root vectors. An algebraic approach of our investigation is founded on construction of Cartan matrix of the spacelike root vectors in Wheeler DeWitt space. Kac Moody algebras can be classified according to their Cartan matrix. By this way a hidden symmetry of the model considered is revealed. It is known, that gravitational models which demonstrate chaotic behavior are associated with hyperbolic Kac Moody algebras. The algebra considered in our paper is not hyperbolic. The square of Weyl vector is negative. The mixmastertype universe is associated with a simplylaced Lorentzian Kac Moody algebra. Since the volume of the configuration space is infinite, the model is not chaotic.

What do RNN Language Models Learn about FillerGap Dependencies ; RNN language models have achieved stateoftheart perplexity results and have proven useful in a suite of NLP tasks, but it is as yet unclear what syntactic generalizations they learn. Here we investigate whether stateoftheart RNN language models represent longdistance fillergap dependencies and constraints on them. Examining RNN behavior on experimentally controlled sentences designed to expose fillergap dependencies, we show that RNNs can represent the relationship in multiple syntactic positions and over large spans of text. Furthermore, we show that RNNs learn a subset of the known restrictions on fillergap dependencies, known as island constraints RNNs show evidence for whislands, adjunct islands, and complex NP islands. These studies demonstrates that stateoftheart RNN models are able to learn and generalize about empty syntactic positions.

A Theory of Enzyme Chemotaxis Comparison Between Experiment and Model ; Enzymes show two distinct transport behaviors in the presence of their substrates in solution. First, their diffusivity enhances with increasing substrate concentration. In addition, enzymes perform directional motion toward regions with high substrate concentration, termed chemotaxis. While a variety of enzymes has been shown to undergo chemotaxis, there remains a lack of quantitative understanding of the phenomenon. Here, we provide a general expression for the active movement of an enzyme in a concentration gradient of its substrate. The proposed model takes into account both the substratebinding and catalytic turnover step, as well as the enhanced diffusion effect. We have experimentally measured the chemotaxis of a fast and a slow enzyme urease under catalytic conditions, and hexokinase for both full catalysis and for simple noncatalytic substrate binding. There is good agreement between the proposed model and the experiments. The model is general, has no adjustable parameters, and only requires three experimentally defined constants to quantify chemotaxis enzymesubstrate binding affinity Kd, MichaelisMenten constant KM and level of diffusion enhancement in the associated substrate alpha.

Information generating, sharing and manipulating SourceReservoirSink model of selforganizing dissipative structures ; Informationtheoretic description of the signal transmitter, the channel and receiver is extended to the network of selforganizing dissipative structures consisting of a source, a reservoir and a sink. The information generation by the source is subjected to controlled manipulation by the reservoir before being transmitted to the sink. The reservoir can have memory and variable capacity for information storage. The role of the reservoir in building the manipulative capacity for information storage and selective sharing is illustrated by the characteristic of asymmetric exchange between the reservoir and the sink. A boxmodel is used to develop the model to represent material, process and information sharing among the source, the reservoir and the sink. The model is applied to selforganizing carbon cages with the enddirected evolution of the buckyball.

Jump to better conclusions SCAN both left and right ; Lake and Baroni 2018 recently introduced the SCAN data set, which consists of simple commands paired with action sequences and is intended to test the strong generalization abilities of recurrent sequencetosequence models. Their initial experiments suggested that such models may fail because they lack the ability to extract systematic rules. Here, we take a closer look at SCAN and show that it does not always capture the kind of generalization that it was designed for. To mitigate this we propose a complementary dataset, which requires mapping actions back to the original commands, called NACS. We show that models that do well on SCAN do not necessarily do well on NACS, and that NACS exhibits properties more closely aligned with realistic usecases for sequencetosequence models.

From Unified Field Theory to the Standard Model and Beyond ; One hundred years ago this year attempts began to generalise general relativity with the ambition of incorporating electromagnetism alongside gravitation in a unified field theory. These developments led to gauge theories and models with extra spatial dimensions that have greatly influenced the modernday pursuit of a unification scheme incorporating the Standard Model of particle physics, again ideally together with gravity. In this paper we motivate a further natural generalisation from extra spatial dimensions at an elementary level which is found to much more directly accommodate distinctive features of the Standard Model. We also investigate the potential to uncover new physical phenomena, making a case in the neutrino sector for one lefthanded neutrino state to be massless, and emphasise the opportunity for a close collaboration between theory and experiment. The new theory possesses a very simple interpretation regarding the underlying source of these empirical structures.

Marginal Structural Models for Timevarying Endogenous Treatments A TimeVarying Instrumental Variable Approach ; Robins 1998 introduced marginal structural models MSMs, a general class of counterfactual models for the joint effects of timevarying treatment regimes in complex longitudinal studies subject to timevarying confounding. He established identification of MSM parameters under a sequential randomization assumption SRA, which essentially rules out unmeasured confounding of treatment assignment over time. In this technical report, we consider sufficient conditions for identification of MSM parameters with the aid of a timevarying instrumental variable, when sequential randomization fails to hold due to unmeasured confounding. Our identification conditions essentially require that no unobserved confounder predicts compliance type for the timevarying treatment, the longitudinal generalization of the identifying condition of Wang and Tchetgen Tchetgen 2018. Under this assumption, We derive a large class of semiparametric estimators that extends standard inverseprobability weighting IPW, the most popular approach for estimating MSMs under SRA, by incorporating the timevarying IV through a modified set of weights. The set of influence functions for MSM parameters is derived under a semiparametric model with sole restriction on observed data distribution given by the MSM, and is shown to provide a rich class of multiply robust estimators, including a local semiparametric efficient estimator.

Cohen Forcing and Inner Models ; Given an inner model W subset V and a regular cardinal kappa, we consider two alternatives for adding a subset to kappa by forcing the Cohen poset Addkappa,1, and the Cohen poset of the inner model Addkappa,1W. The forcing from W will be at least as strong as the forcing from V in the sense that forcing with the former adds a generic for the latter if and only if the two posets have the same cardinality. On the other hand, a sufficient condition is established for the poset from V to fail to be as strong as that from W. The results are generalized to Addkappa,lambda, and to iterations of Cohen forcing where the poset at each stage comes from an arbitrary intermediate inner model.

Scalable visualisation methods for modern Generalized Additive Models ; In the last two decades the growth of computational resources has made it possible to handle Generalized Additive Models GAMs that formerly were too costly for serious applications. However, the growth in model complexity has not been matched by improved visualisations for model development and results presentation. Motivated by an industrial application in electricity load forecasting, we identify the areas where the lack of modern visualisation tools for GAMs is particularly severe, and we address the shortcomings of existing methods by proposing a set of visual tools that a are fast enough for interactive use, b exploit the additive structure of GAMs, c scale to large data sets and d can be used in conjunction with a wide range of response distributions. All the new visual methods proposed in this work are implemented by the mgcViz R package, which can be found on the Comprehensive R Archive Network.

TFReplicator Distributed Machine Learning for Researchers ; We describe TFReplicator, a framework for distributed machine learning designed for DeepMind researchers and implemented as an abstraction over TensorFlow. TFReplicator simplifies writing dataparallel and modelparallel research code. The same models can be effortlessly deployed to different cluster architectures i.e. one or many machines containing CPUs, GPUs or TPU accelerators using synchronous or asynchronous training regimes. To demonstrate the generality and scalability of TFReplicator, we implement and benchmark three very different models 1 A ResNet50 for ImageNet classification, 2 a SNGAN for classconditional ImageNet image generation, and 3 a D4PG reinforcement learning agent for continuous control. Our results show strong scalability performance without demanding any distributed systems expertise of the user. The TFReplicator programming model will be opensourced as part of TensorFlow 2.0 see httpsgithub.comtensorflowcommunitypull25.

Fast Approximation and Estimation Bounds of Kernel Quadrature for Infinitely Wide Models ; An infinitely wide model is a weighted integration int varphix,v d muv of feature maps. This model excels at handling an infinite number of features, and thus it has been adopted to the theoretical study of deep learning. Kernel quadrature is a kernelbased numerical integration scheme developed for fast approximation of expectations int fx d px. In this study, regarding the weight mu as a signed or complexvectorvalued distribution of parameters, we develop the general kernel quadrature GKQ for parameter distributions. The proposed method can achieve a fast approximation rate Oep with parameter number p, which is faster than the traditional Barron's rate, and a fast estimation rate widetildeO1n with sample size n. As a result, we have obtained a new normbased complexity measure for infinitely wide models. Since the GKQ implicitly conducts the empirical risk minimization, we can understand that the complexity measure also reflects the generalization performance in the gradient learning setup.

Parametrizations of Dark Energy Models in the Background of General Noncanonical Scalar Field in Ddimensional Fractal Universe ; We explore noncanonical scalar field model in the background of nonflat Ddimensional fractal Universe with cosmological constant Lambda on the condition that the matter and scalar field are separately conserved. The potential V, scalar field phi, function f, densities, Hubble parameter and deceleration parameter can be expressed in terms of the redshift z and these depend on the equation of state parameter wphi. We also investigate four kinds of well known parametrization models and graphically we have analyzed the natures of potential, scalar field, function f, densities, the Hubble parameter and deceleration parameter. As a result, the best fitted values of the unknown parameters w0,w1 of the parametrizations models due to the joint data analysis SNIaBAOCMBHubble are found. Furthermore, the minimum values of chi2 function are obtained. Also we have plotted the graphs for different confidence levels 66, 90 and 99 contours for w0,w1 by fixing the other parameters.

Squared English Word A Method of Generating Glyph to Use Super Characters for Sentiment Analysis ; The Super Characters method addresses sentiment analysis problems by first converting the input text into images and then applying 2DCNN models to classify the sentiment. It achieves state of the art performance on many benchmark datasets. However, it is not as straightforward to apply in Latin languages as in Asian languages. Because the 2DCNN model is designed to recognize twodimensional images, it is better if the inputs are in the form of glyphs. In this paper, we propose SEW Squared English Word method generating a squared glyph for each English word by drawing Super Characters images of each English word at the alphabet level, combining the squared glyph together into a whole Super Characters image at the sentence level, and then applying the CNN model to classify the sentiment within the sentence. We applied the SEW method to Wikipedia dataset and obtained a 2.1 accuracy gain compared to the original Super Characters method. For multimodal data with both structured tabular data and unstructured natural language text, the modified SEW method integrates the data into a single image and classifies sentiment with one unified CNN model.

Addressing Overfitting on Pointcloud Classification using Atrous XCRF ; Advances in techniques for automated classification of pointcloud data introduce great opportunities for many new and existing applications. However, with a limited number of labeled points, automated classification by a machine learning model is prone to overfitting and poor generalization. The present paper addresses this problem by inducing controlled noise on a trained model generated by invoking conditional random field similarity penalties using nearby features. The method is called Atrous XCRF and works by forcing a trained model to respect the similarity penalties provided by unlabeled data. In a benchmark study carried out using the ISPRS 3D labeling dataset, our technique achieves 84.97 in term of overall accuracy, and 71.05 in term of F1 score. The result is on par with the current best model for the benchmark dataset and has the highest value in term of F1 score.

Bayesian Nonparametric Adaptive Spectral Density Estimation for Financial Time Series ; Discrimination between nonstationarity and longrange dependency is a difficult and longstanding issue in modelling financial time series. This paper uses an adaptive spectral technique which jointly models the nonstationarity and dependency of financial time series in a nonparametric fashion assuming that the time series consists of a finite, but unknown number, of locally stationary processes, the locations of which are also unknown. The model allows a nonparametric estimate of the dependency structure by modelling the autocovariance function in the spectral domain. All our estimates are made within a Bayesian framework where we use aReversible Jump Markov Chain Monte Carlo algorithm for inference. We study the frequentist properties of our estimates via a simulation study, and present a novel way of generating time series data from a nonparametric spectrum. Results indicate that our techniques perform well across a range of data generating processes. We apply our method to a number of real examples and our results indicate that several financial time series exhibit both longrange dependency and nonstationarity.

Renormalization Effects on Electric Dipole Moments in Electroweakly Interacting Massive Particle Models ; We study the renormalization effects on electric dipole moments in the models with new electroweakly interacting massive fermions. The electric dipole moments are generated by the effective operators which arise from integrating out heavy particles at some scale in the models. We give the renormalization group equation for the Wilson coefficients of the effective operators from the scale where the operators are generated to the electroweak scale. Our numerical studies focus on the electric dipole moments in the minisplit supersymmetric scenario and the electroweakly interacting massive particle dark matter scenario. It turns out that the renormalization effects can give an enhancement factor being of the order of O10 in the minisplit scenario and being more than two in the minimal dark matter model.

Kinetic modeling of alcohol consumption ; In most countries, alcohol consumption distributions have been shown to possess universal features. Their unimodal rightskewed shape is usually modeled in terms of the Lognormal distribution, which is easy to fit, test, and modify. However, empirical distributions often deviate considerably from the Lognormal model, and both Gamma and Weibull distributions appear to better describe the survey data. In this paper we explain the appearance of these distributions by means of classical methods of kinetic theory of multiagent systems. The microscopic variation of alcohol consumption of agents around a universal emphsocial accepted value of consumption, is built up introducing as main criterion for consumption a suitable value function in the spirit of the prospect theory of Kahneman and Twersky. The mathematical properties of the value function then determine the unique macroscopic equilibrium which results to be a generalized Gamma distribution. The modeling of the microscopic kinetic interaction allows to clarify the meaning of the various parameters characterizing the generalized Gamma equilibrium.

Nonlinear generalization of the monotone single index model ; Single index model is a powerful yet simple model, widely used in statistics, machine learning, and other scientific fields. It models the regression function as ga,x, where a is an unknown index vector and x are the features. This paper deals with a nonlinear generalization of this framework to allow for a regressor that uses multiple index vectors, adapting to local changes in the responses. To do so we exploit the conditional distribution over functiondriven partitions, and use linear regression to locally estimate index vectors. We then regress by applying a kNN type estimator that uses a localized proxy of the geodesic metric. We present theoretical guarantees for estimation of local index vectors and outofsample prediction, and demonstrate the performance of our method with experiments on synthetic and realworld data sets, comparing it with stateoftheart methods.

Cooperative Learning of Disjoint Syntax and Semantics ; There has been considerable attention devoted to models that learn to jointly infer an expression's syntactic structure and its semantics. Yet, citetNangiaB18 has recently shown that the current best systems fail to learn the correct parsing strategy on mathematical expressions generated from a simple contextfree grammar. In this work, we present a recursive model inspired by newciteChoiYL18 that reaches near perfect accuracy on this task. Our model is composed of two separated modules for syntax and semantics. They are cooperatively trained with standard continuous and discrete optimization schemes. Our model does not require any linguistic structure for supervision and its recursive nature allows for outofdomain generalization with little loss in performance. Additionally, our approach performs competitively on several natural language tasks, such as Natural Language Inference or Sentiment Analysis.

General framework for testing PoissonVoronoi assumption for real microstructures ; Modeling microstructures is an interesting problem not just in Materials Science but also in Mathematics and Statistics. The most basic model for steel microstructure is the PoissonVoronoi diagram. It has mathematically attractive properties and it has been used in the approximation of single phase steel microstructures. The aim of this paper is to develop methods that can be used to test whether a real steel microstructure can be approximated by such a model. Therefore, a general framework for testing the PoissonVoronoi assumption based on images of 2D sections of real metals is set out. Following two different approaches, according to the use or not of periodic boundary conditions, three different model tests are proposed. The first two are based on the coefficient of variation and the cumulative distribution function of the cells area. The third exploits tools from to Topological Data Analysis, such as persistence landscapes.

Glidar3DJ A ViewInvariant gait identification via flash lidar data correction ; Gait recognition is a leading remotebased identification method, suitable for realworld surveillance and medical applications. Modelbased gait recognition methods have been particularly recognized due to their scale and viewinvariant properties. We present the first modelbased gait recognition methodology, mathcalGlidar3DJ using a skeleton model extracted from sequences generated by a single flash lidar camera. Existing successful modelbased approaches take advantage of high quality skeleton data collected by Kinect and Mocap, for example, are not practicable for applications outside the laboratory. The low resolution and noisy imaging process of lidar negatively affects the performance of stateoftheart skeletonbased systems, generating a significant number of outlier skeletons. We propose a rulebased filtering mechanism that adopts robust statistics to correct for skeleton joint measurements. Quantitative measurements validate the efficacy of the proposed method in improving gait recognition.

Light Mediators in Anomaly Free U1X Models I Theoretical Framework ; We examine theoretical features of U1X extensions of the Standard Model whose quantum anomalies are canceled per generation. Similarly to other versions, the theory consists of a TwoHiggsDoublet Model plus a scalar singlet embedded into the SM otimes U1X gauge group, and introduces small modifications to the Zboson interactions. These changes can be minimized by exclusively charging righthanded fermions under the new Abelian symmetry, and are compensated by the neutral Xboson exchange. Nonuniversality of fermion couplings can also be achieved by requiring one single Xcharged family. In general, X gauge bosons can be separated into A' dark photons and Z' subsets, distinguished by the presence of axialvector currents. A' physics is commonly simpler to constrain and therefore favored by experimental tests. Finally, the model can be UV completed both by stable chi fermions or by righthanded neutrinos. The prior case may provide cold WIMPs in the theory.

Multilingual Factor Analysis ; In this work we approach the task of learning multilingual word representations in an offline manner by fitting a generative latent variable model to a multilingual dictionary. We model equivalent words in different languages as different views of the same word generated by a common latent variable representing their latent lexical meaning. We explore the task of alignment by querying the fitted model for multilingual embeddings achieving competitive results across a variety of tasks. The proposed model is robust to noise in the embedding space making it a suitable method for distributed representations learned from noisy corpora.

Domain adaptation for partofspeech tagging of noisy usergenerated text ; The performance of a Partofspeech POS tagger is highly dependent on the domain ofthe processed text, and for many domains there is no or only very little training data available. This work addresses the problem of POS tagging noisy usergenerated text using a neural network. We propose an architecture that trains an outofdomain model on a large newswire corpus, and transfers those weights by using them as a prior for a model trained on the target domain a dataset of German Tweets for which there is very little annotations available. The neural network has two standard bidirectional LSTMs at its core. However, we find it crucial to also encode a set of taskspecific features, and to obtain reliable sourcedomain and targetdomain word representations. Experiments with different regularization techniques such as early stopping, dropout and finetuning the domain adaptation prior weights are conducted. Our best model uses external weights from the outofdomain model, as well as feature embeddings, pretrained word and subword embeddings and achieves a tagging accuracy of slightly over 90, improving on the previous state of the art for this task.

Bouncing cosmology in an Extended Theory of Gravity ; We have investigated some bouncing models in the framework of an extended gravity theory where the usual Ricci scalar in the gravitational action is replaced by a sum of the Ricci scalar and a term proportional to the trace of the energy momentum tensor. The dynamical parameters of the model are derived in most general manner. We considered two bouncing scenarios through an exponential and a power law scale factor. The non singular bouncing models also favour a late time cosmic speed up phenomenon. The dynamical behaviour of the equation of state parameter is studied for the models. It is observed that, near the bounce, the dynamics is substantially affected by the coupling parameter of the modified gravity theory as compared to the parameters of the bouncing scale factors.

ScoreDriven Exponential Random Graphs A New Class of TimeVarying Parameter Models for Dynamical Networks ; Motivated by the increasing abundance of data describing realworld networks that exhibit dynamical features, we propose an extension of the Exponential RandomGraph Models ERGMs that accommodates the time variation of its parameters. Inspired by the fast growing literature on Dynamic Conditional Scoredriven models each parameter evolves according to an updating rule driven by the score of the ERGM distribution. We demonstrate the flexibility of the scoredriven ERGMs SDERGMs, both as data generating processes and as filters, and we show the advantages of the dynamic version with respect to the static one. We discuss two applications to timevarying networks from financial and political systems. First, we consider the prediction of future links in the Italian interbank credit network. Second, we show that the SDERGM allows to discriminate between static or timevarying parameters when used to model the dynamics of the US congress covoting network.

Physicsinformed Autoencoders for Lyapunovstable Fluid Flow Prediction ; In addition to providing highprofile successes in computer vision and natural language processing, neural networks also provide an emerging set of techniques for scientific problems. Such datadriven models, however, typically ignore physical insights from the scientific system under consideration. Among other things, a physicsinformed model formulation should encode some degree of stability or robustness or wellconditioning in that a small change of the input will not lead to drastic changes in the output, characteristic of the underlying scientific problem. We investigate whether it is possible to include physicsinformed prior knowledge for improving the model quality e.g., generalization performance, sensitivity to parameter tuning, or robustness in the presence of noisy data. To that extent, we focus on the stability of an equilibrium, one of the most basic properties a dynamic system can have, via the lens of Lyapunov analysis. For the prototypical problem of fluid flow prediction, we show that models preserving Lyapunov stability improve the generalization error and reduce the prediction uncertainty.

ODE2VAE Deep generative second order ODEs with Bayesian neural networks ; We present Ordinary Differential Equation Variational AutoEncoder ODE2VAE, a latent second order ODE model for highdimensional sequential data. Leveraging the advances in deep generative models, ODE2VAE can simultaneously learn the embedding of high dimensional trajectories and infer arbitrarily complex continuoustime latent dynamics. Our model explicitly decomposes the latent space into momentum and position components and solves a second order ODE system, which is in contrast to recurrent neural network RNN based time series models and recently proposed blackbox ODE techniques. In order to account for uncertainty, we propose probabilistic latent ODE dynamics parameterized by deep Bayesian neural networks. We demonstrate our approach on motion capture, image rotation and bouncing balls datasets. We achieve stateoftheart performance in long term motion prediction and imputation tasks.

The Shape of Data Intrinsic Distance for Data Distributions ; The ability to represent and compare machine learning models is crucial in order to quantify subtle model changes, evaluate generative models, and gather insights on neural network architectures. Existing techniques for comparing data distributions focus on global data properties such as mean and covariance; in that sense, they are extrinsic and uniscale. We develop a firstofitskind intrinsic and multiscale method for characterizing and comparing data manifolds, using a lowerbound of the spectral variant of the GromovWasserstein intermanifold distance, which compares all data moments. In a thorough experimental study, we demonstrate that our method effectively discerns the structure of data manifolds even on unaligned data of different dimensionalities; moreover, we showcase its efficacy in evaluating the quality of generative models.

Online MeasurementBased Estimation of Dynamic System State Matrix in Ambient Conditions ; In this paper, a purely measurementbased method is proposed to estimate the dynamic system state matrix by applying the regression theorem of the multivariate OrnsteinUhlenbeck process. The proposed method employs a recursive algorithm to minimize the required computational effort, making it applicable to the realtime environment. One main advantage of the proposed method is model independence, i.e., it is independent of the network model and the dynamic model of generators. Among various applications of the estimated matrix, detecting and locating unexpected network topology change is illustrated in details. Simulation studies have shown that the proposed measurementbased method can provide an accurate and efficient estimation of the dynamic system state matrix under the occurrence of unexpected topology change. Besides, various implementation conditions are tested to show that the proposed method can provide accurate approximation despite measurement noise, missing PMUs, and the implementation of higherorder generator models with control devices.

AlignFlow Cycle Consistent Learning from Multiple Domains via Normalizing Flows ; Given datasets from multiple domains, a key challenge is to efficiently exploit these data sources for modeling a target domain. Variants of this problem have been studied in many contexts, such as crossdomain translation and domain adaptation. We propose AlignFlow, a generative modeling framework that models each domain via a normalizing flow. The use of normalizing flows allows for a flexibility in specifying learning objectives via adversarial training, maximum likelihood estimation, or a hybrid of the two methods; and b learning and exact inference of a shared representation in the latent space of the generative model. We derive a uniform set of conditions under which AlignFlow is marginallyconsistent for the different learning objectives. Furthermore, we show that AlignFlow guarantees exact cycle consistency in mapping datapoints from a source domain to target and back to the source domain. Empirically, AlignFlow outperforms relevant baselines on imagetoimage translation and unsupervised domain adaptation and can be used to simultaneously interpolate across the various domains using the learned representation.

Oneelement Batch Training by Moving Window ; Several deep models, esp. the generative, compare the samples from two distributions e.g. WAE like AutoEncoder models, setprocessing deep networks, etc in their cost functions. Using all these methods one cannot train the model directly taking small size in extreme one element batches, due to the fact that samples are to be compared. We propose a generic approach to training such models using oneelement minibatches. The idea is based on splitting the batch in latent into parts previous, i.e. historical, elements used for latent space distribution matching and the current ones, used both for latent distribution computation and the minimization process. Due to the smaller memory requirements, this allows to train networks on higher resolution images then in the classical approach.

Improving the Detection of Burnt Areas in Remote Sensing using Hyperfeatures Evolved by M3GP ; One problem found when working with satellite images is the radiometric variations across the image and different images. Intending to improve remote sensing models for the classification of burnt areas, we set two objectives. The first is to understand the relationship between feature spaces and the predictive ability of the models, allowing us to explain the differences between learning and generalization when training and testing in different datasets. We find that training on datasets built from more than one image provides models that generalize better. These results are explained by visualizing the dispersion of values on the feature space. The second objective is to evolve hyperfeatures that improve the performance of different classifiers on a variety of test sets. We find the hyperfeatures to be beneficial, and obtain the best models with XGBoost, even if the hyperfeatures are optimized for a different method.

Higherspin gravity and torsion on quantized spacetime in matrix models ; A geometric formalism is developed which allows to describe the nonlinear regime of higherspin gravity emerging on a cosmological quantum spacetime in the IKKT matrix model. The vacuum solutions are Ricciflat up to an effective vacuum energymomentum tensor quadratic in the torsion, which arises from a Weitzenbocktype higher spin connection. Torsion is expected to be significant only at cosmic scales and around very massive objects, and could behave like dark matter. A nonlinear equation for the torsion tensor is found, which encodes the YangMills equations of the matrix model. The metric and torsion transform covariantly under a higherspin generalization of volumepreserving diffeomorphisms, which arises from the gauge invariance of the matrix model.

Blank Language Models ; We propose Blank Language Model BLM, a model that generates sequences by dynamically creating and filling in blanks. The blanks control which part of the sequence to expand, making BLM ideal for a variety of text editing and rewriting tasks. The model can start from a single blank or partially completed text with blanks at specified locations. It iteratively determines which word to place in a blank and whether to insert new blanks, and stops generating when no blanks are left to fill. BLM can be efficiently trained using a lower bound of the marginal data likelihood. On the task of filling missing text snippets, BLM significantly outperforms all other baselines in terms of both accuracy and fluency. Experiments on style transfer and damaged ancient text restoration demonstrate the potential of this framework for a wide range of applications.

Timeaware Large Kernel Convolutions ; To date, most stateoftheart sequence modeling architectures use attention to build generative models for language based tasks. Some of these models use all the available sequence tokens to generate an attention distribution which results in time complexity of On2. Alternatively, they utilize depthwise convolutions with softmax normalized kernels of size k acting as a limitedwindow selfattention, resulting in time complexity of Okcdotn. In this paper, we introduce Timeaware Large Kernel TaLK Convolutions, a novel adaptive convolution operation that learns to predict the size of a summation kernel instead of using a fixedsized kernel matrix. This method yields a time complexity of On, effectively making the sequence encoding process linear to the number of tokens. We evaluate the proposed method on largescale standard machine translation, abstractive summarization and language modeling datasets and show that TaLK Convolutions constitute an efficient improvement over other attentionconvolution based approaches.

A Model of Fast Concept Inference with ObjectFactorized Cognitive Programs ; The ability of humans to quickly identify general concepts from a handful of images has proven difficult to emulate with robots. Recently, a computer architecture was developed that allows robots to mimic some aspects of this human ability by modeling concepts as cognitive programs using an instruction set of primitive cognitive functions. This allowed a robot to emulate human imagination by simulating candidate programs in a world model before generalizing to the physical world. However, this model used a naive search algorithm that required 30 minutes to discover a single concept, and became intractable for programs with more than 20 instructions. To circumvent this bottleneck, we present an algorithm that emulates the human cognitive heuristics of object factorization and subgoaling, allowing humanlevel inference speed, improving accuracy, and making the output more explainable.

Predicting drug properties with parameterfree machine learning ParetoOptimal Embedded Modeling POEM ; The prediction of absorption, distribution, metabolism, excretion, and toxicity ADMET of small molecules from their molecular structure is a central problem in medicinal chemistry with great practical importance in drug discovery. Creating predictive models conventionally requires substantial trialanderror for the selection of molecular representations, machine learning ML algorithms, and hyperparameter tuning. A generally applicable method that performs well on all datasets without tuning would be of great value but is currently lacking. Here, we describe ParetoOptimal Embedded Modeling POEM, a similaritybased method for predicting molecular properties. POEM is a nonparametric, supervised ML algorithm developed to generate reliable predictive models without need for optimization. POEMs predictive strength is obtained by combining multiple different representations of molecular structures in a contextspecific manner, while maintaining low dimensionality. We benchmark POEM relative to industrystandard ML algorithms and published results across 17 classifications tasks. POEM performs well in all cases and reduces the risk of overfitting.

BlackBox Optimization with Local Generative Surrogates ; We propose a novel method for gradientbased optimization of blackbox simulators using differentiable local surrogate models. In fields such as physics and engineering, many processes are modeled with nondifferentiable simulators with intractable likelihoods. Optimization of these forward models is particularly challenging, especially when the simulator is stochastic. To address such cases, we introduce the use of deep generative models to iteratively approximate the simulator in local neighborhoods of the parameter space. We demonstrate that these local surrogates can be used to approximate the gradient of the simulator, and thus enable gradientbased optimization of simulator parameters. In cases where the dependence of the simulator on the parameter space is constrained to a low dimensional submanifold, we observe that our method attains minima faster than baseline methods, including Bayesian optimization, numerical optimization, and approaches using score function gradient estimators.

Probing the Seesaw Mechanism with Cosmological Data ; We investigate cosmological consequences of an inflationary model which incorporates a generic seesaw extension types I and II of the Standard Model of Particle Physics. A nonminimal coupling between the inflaton field and the Ricci scalar is considered as well as radiative corrections at one loop order. This connection between the inflationary dynamics with neutrino physics results in a predictive model whose observational viability is investigated in light of the current cosmic microwave background data, baryon acoustic oscillation observations and type Ia supernovae measurements. Our results show that the nonminimal coupled seesaw potential provides a good description of the observational data when radiative corrections are positive. Such result favours the type II seesaw mechanism over type I and may be an indication for physics beyond the Standard Model.

Scalable and Practical Natural Gradient for LargeScale Deep Learning ; Largescale distributed training of deep neural networks results in models with worse generalization performance as a result of the increase in the effective minibatch size. Previous approaches attempt to address this problem by varying the learning rate and batch size over epochs and layers, or ad hoc modifications of batch normalization. We propose Scalable and Practical Natural Gradient Descent SPNGD, a principled approach for training models that allows them to attain similar generalization performance to models trained with firstorder optimization methods, but with accelerated convergence. Furthermore, SPNGD scales to large minibatch sizes with a negligible computational overhead as compared to firstorder methods. We evaluated SPNGD on a benchmark task where highly optimized firstorder methods are available as references training a ResNet50 model for image classification on ImageNet. We demonstrate convergence to a top1 validation accuracy of 75.4 in 5.5 minutes using a minibatch size of 32,768 with 1,024 GPUs, as well as an accuracy of 74.9 with an extremely large minibatch size of 131,072 in 873 steps of SPNGD.

Autonomous UnknownApplication Filtering and Labeling for DLbased Traffic Classifier Update ; Network traffic classification has been widely studied to fundamentally advance network measurement and management. Machine Learning is one of the effective approaches for network traffic classification. Specifically, Deep Learning DL has attracted much attention from the researchers due to its effectiveness even in encrypted network traffic without compromising neither user privacy nor network security. However, most of the existing models are created from closedworld datasets, thus they can only classify those existing classes previously sampled and labeled. In this case, unknown classes cannot be correctly classified. To tackle this issue, an autonomous learning framework is proposed to effectively update DLbased traffic classification models during active operations. The core of the proposed framework consists of a DLbased classifier, a selflearned discriminator, and an autonomous selflabeling model. The discriminator and selflabeling process can generate new dataset during active operations to support classifier update. Evaluation of the proposed framework is performed on an open dataset, i.e., ISCX VPNnonVPN, and independently collected data packets. The results demonstrate that the proposed autonomous learning framework can filter packets from unknown classes and provide accurate labels. Thus, corresponding DLbased classification models can be updated successfully with the autonomously generated dataset.

Deep Gaussian Markov Random Fields ; Gaussian Markov random fields GMRFs are probabilistic graphical models widely used in spatial statistics and related fields to model dependencies over spatial structures. We establish a formal connection between GMRFs and convolutional neural networks CNNs. Common GMRFs are special cases of a generative model where the inverse mapping from data to latent variables is given by a 1layer linear CNN. This connection allows us to generalize GMRFs to multilayer CNN architectures, effectively increasing the order of the corresponding GMRF in a way which has favorable computational scaling. We describe how wellestablished tools, such as autodiff and variational inference, can be used for simple and efficient inference and learning of the deep GMRF. We demonstrate the flexibility of the proposed model and show that it outperforms the stateoftheart on a dataset of satellite temperatures, in terms of prediction and predictive uncertainty.

A General Pairwise Comparison Model for Extremely Sparse Networks ; Statistical inference using pairwise comparison data is an effective approach to analyzing largescale sparse networks. In this paper, we propose a general framework to model the mutual interactions in a network, which enjoys ample flexibility in terms of model parametrization. Under this setup, we show that the maximum likelihood estimator for the latent score vector of the subjects is uniformly consistent under a nearminimal condition on network sparsity. This condition is sharp in terms of the leading order asymptotics describing the sparsity. Our analysis utilizes a novel chaining technique and illustrates an important connection between graph topology and model consistency. Our results guarantee that the maximum likelihood estimator is justified for estimation in largescale pairwise comparison networks where data are asymptotically deficient. Simulation studies are provided in support of our theoretical findings.

Generalisation error in learning with random features and the hidden manifold model ; We study generalised linear regression and classification for a synthetically generated dataset encompassing different problems of interest, such as learning with random features, neural networks in the lazy training regime, and the hidden manifold model. We consider the highdimensional regime and using the replica method from statistical physics, we provide a closedform expression for the asymptotic generalisation performance in these problems, valid in both the under and overparametrised regimes and for a broad choice of generalised linear model loss functions. In particular, we show how to obtain analytically the socalled double descent behaviour for logistic regression with a peak at the interpolation threshold, we illustrate the superiority of orthogonal against random Gaussian projections in learning with random features, and discuss the role played by correlations in the data generated by the hidden manifold model. Beyond the interest in these particular problems, the theoretical formalism introduced in this manuscript provides a path to further extensions to more complex tasks.

Coexistence of localized Gibbs measures and delocalized gradient Gibbs measures on trees ; We study gradient models for spins taking values in the integers or an integer lattice, which interact via a general potential depending only on the differences of the spin values at neighboring sites, located on a regular tree with d 1 neighbors. We first provide general conditions in terms of the relevant pnorms of the associated transfer operator Q which ensure the existence of a countable family of proper Gibbs measures. Next we prove existence of delocalized gradient Gibbs measures, under natural conditions on Q. This implies coexistence of both types of measures for large classes of models including the SOSmodel, and heavytailed models arising for instance for potentials of logarithmic growth.

On The Microscopic Modeling of Vehicular Traffic on General Networks ; We introduce a formalism to deal with the microscopic modeling of vehicular traffic on a road network. Traffic on each road is unidirectional, and the dynamics of each vehicle is described by a FollowtheLeader model. From a mathematical point of view, this amounts to define a system of ordinary differential equations on an arbitrary network. A general existence and uniqueness result is provided, while priorities at junctions are shown to hinder the stability of solutions. We investigate the occurrence of the Braess paradox in a timedependent setting within this model. The emergence of Nash equilibria in a nonstationary situation results in the appearance of Braess type paradoxes, and this is supported by numerical simulations.

Bayesian Multiscale Modeling of Factor Matrix without using Partition Tree ; The multiscale factor models are particularly appealing for analyzing matrix or tensorvalued data, due to their adaptiveness to local geometry and intuitive interpretation. However, the reliance on the binary tree for recursive partitioning creates high complexity in the parameter space, making it extremely challenging to quantify its uncertainty. In this article, we discover an alternative way to generate multiscale matrix using simple matrix operation starting from a random matrix with each column having two unique values, its Cholesky whitening transform obeys a recursive partitioning structure. This allows us to consider a generative distribution with large prior support on common multiscale factor models, and efficient posterior computation via Hamiltonian Monte Carlo. We demonstrate its potential in a multiscale factor model to find broader regions of interest for human brain connectivity.

MultiRepresentation Knowledge Distillation For Audio Classification ; As an important component of multimedia analysis tasks, audio classification aims to discriminate between different audio signal types and has received intensive attention due to its wide applications. Generally speaking, the raw signal can be transformed into various representations such as Short Time Fourier Transform and Mel Frequency Cepstral Coefficients, and information implied in different representations can be complementary. Ensembling the models trained on different representations can greatly boost the classification performance, however, making inference using a large number of models is cumbersome and computationally expensive. In this paper, we propose a novel endtoend collaborative learning framework for the audio classification task. The framework takes multiple representations as the input to train the models in parallel. The complementary information provided by different representations is shared by knowledge distillation. Consequently, the performance of each model can be significantly promoted without increasing the computational overhead in the inference stage. Extensive experimental results demonstrate that the proposed approach can improve the classification performance and achieve stateoftheart results on both acoustic scene classification tasks and general audio tagging tasks.

An analytic model of chiralinduced spin selectivity ; Organic materials are known to feature long spindiffusion times, originating in a generally small spinorbit coupling observed in these systems. From that perspective, chiral molecules acting as efficient spin selectors pose a puzzle, that attracted a lot of attention during the recent years. Here we revisit the physical origins of chiralinduced spin selectivity CISS, and propose a simple analytic minimal model to describe it. The model treats a chiral molecule as an anisotropic wire with molecular dipole moments aligned arbitrarily with respect to the wire's axes, and is therefore quite general. Importantly, it shows that helical structure of the molecule is not necessary to observe CISS and other chiral nonhelical molecules can also be considered as a potential candidates for CISS effect. We also show that the suggested simple model captures the main characteristics of CISS observed in experiment, without the need for additional constraints employed in the previous studies. The results pave the way for understanding other related physical phenomena where CISS effect plays an essential role.

Constantroll in the PalatiniR2 models ; We consider models of a scalar field coupled to quadratic RR2 gravity in the framework of the Palatini formulation. The resulting Einsteinframe generalized kinflation effective theory is analyzed assuming that the constantroll condition holds. We focus on a quartic selfinteraction potential, a case of particular appeal modelling Higgs inflation, considering the cases of minimal and nonminimal coupling of the inflaton to gravity. For an appropriate range of the model parameters in the large field domain the obtained values for the inflationary observables are found in agreement with current observations.

Non minimally coupled condensate cosmologies a phase space analysis ; We present an analysis of the phase space of cosmological models based on a non minimal coupling between the geometry and a fermionic condensate. We obtain that the strong constraint coming from the Dirac equations allows a detailed design of the cosmology of these models and at the same time guarantees an evolution towards a state indistinguishable from General Relativistic cosmological models. In this light, we show how the use of some specific potentials is able to reproduce naturally two de Sitter phases separated by a power law expansion which could be an interesting model for the unification of an inflationary phase and a dark energy era.

Manyserver heavytraffic limit for queues with timevarying parameters ; A manyserver heavytraffic FCLT is proved for the GtMstmathit GI queueing model, having timevarying arrival rate and staffing, a general arrival process satisfying a FCLT, exponential service times and customer abandonment according to a general probability distribution. The FCLT provides theoretical support for the approximating deterministic fluid model the authors analyzed in a previous paper and a refined Gaussian process approximation, using variance formulas given here. The model is assumed to alternate between underloaded and overloaded intervals, with critical loading only at the isolated switching points. The proof is based on a recursive analysis of the system over these successive intervals, drawing heavily on previous results for infiniteserver models. The FCLT requires careful treatment of the initial conditions for each interval.

Is the Higgs Boson Associated with ColemanWeinberg Dynamical Symmetry Breaking ; The Higgs mechanism may be a quantum phenomenon, i.e., a ColemanWeinberg potential generated by the explicit breaking of scale symmetry in Feynman loops. We review the relationship of scale symmetry, trace anomalies, and emphasize the role of the renormalization group in determining Coleman Weinberg potentials. We propose a simple phenomenological model with maximal visibility at the LHC containing a dormant Higgs doublet no VEV, coupled to standard model gauge interactions SU2times U1 with a mass of sim 380 GeV. We discuss the LHC phenomenology and UV challenges of such a model. We also give a schematic model in which new heavy fermions, with masses sim 230 GeV, can drive a ColemanWeinberg potential at twoloops. The role of the improved stress tensor is emphasized, and we propose a nongravitational term, analogous to the thetaterm in QCD, which generates it from a scalar action.

Reformulating the Situation Calculus and the Event Calculus in the General Theory of Stable Models and in Answer Set Programming ; Circumscription and logic programs under the stable model semantics are two wellknown nonmonotonic formalisms. The former has served as a basis of classical logic based action formalisms, such as the situation calculus, the event calculus and temporal action logics; the latter has served as a basis of a family of action languages, such as language A and several of its descendants. Based on the discovery that circumscription and the stable model semantics coincide on a class of canonical formulas, we reformulate the situation calculus and the event calculus in the general theory of stable models. We also present a translation that turns the reformulations further into answer set programs, so that efficient answer set solvers can be applied to compute the situation calculus and the event calculus.

Radiative Lepton Model and Dark Matter with Global U1' Symmetry ; We propose a radiative lepton model, in which the charged lepton masses are generated at oneloop level, and the neutrino masses are induced at twoloop level. On the other hand, tau mass is derived at tree level since it is too heavy to generate radiatively. Then we discuss muon anomalous magnetic moment together with the constraint of lepton flavor violation. A large muon magnetic moment is derived due to the vector like charged fermions which are newly added to the standard model. In addition, considering a scalar dark matter in our model, a strong gammaray signal is produced by dark matter annihilation via internal bremsstrahlung. We can also obtain the effective neutrino number by the dark radiation of the Goldstone boson coming from the imposed global U1' symmetry.

Cosmological Dynamics of Modified Gravity With a Nonminimal Curvaturematter Coupling ; We perform a phase space analysis of a nonminimally coupled modified gravity theory with the Lagrangian density of the form frac12 f1R1lambda f2RcalLm, where f1R and f2R are arbitrary functions of the curvature scalar R and calLm is the matter Lagrangian density. We apply the dynamical system approach to this scenario in two particular models. In the first model we assume f1R2R with a general form for f2R and set favorable values for effective equation of state parameter which is related to the several epochs of the cosmic evolution and study the critical points and their stability in each cosmic eras. In the second case, we allow the f1R to be an arbitrary function of R and set f2R2R. We find the late time attractor solution for the model and show that this model has a late time accelerating epoch and an acceptable matter era.

Phase diagram and sweep dynamics of a onedimensional generalized cluster model ; We numerically study quantum phase transitions and dynamical properties in the onedimensional cluster model with several interactions by using the timeevolving block decimation method for infinite systems and the exact diagonalization. First, boundaries among several quantum phases of the model are determined from energy gap and each phase is characterized by order parameters and the entanglement spectrum ES. We confirm that in the model with open boundary condition the degeneracy of the lowest levels in the ES corresponds to that of the ground states. Then, using the timedependent Bogoliubov transformation with open boundary condition, we investigate dynamical properties during an interaction sweep through the critical point which separates two topological phases involving fourfold degeneracy in the ground state. After a slow sweep across the critical point, we observe spatially periodic structures in the string correlation functions and the entanglement entropy. It is shown that the periodicities stem from the Bogoliubov quasiparticles generated near the critical point.

Deep Unsupervised Learning using Nonequilibrium Thermodynamics ; A central problem in machine learning involves modeling complex datasets using highly flexible families of probability distributions in which learning, sampling, inference, and evaluation are still analytically or computationally tractable. Here, we develop an approach that simultaneously achieves both flexibility and tractability. The essential idea, inspired by nonequilibrium statistical physics, is to systematically and slowly destroy structure in a data distribution through an iterative forward diffusion process. We then learn a reverse diffusion process that restores structure in data, yielding a highly flexible and tractable generative model of the data. This approach allows us to rapidly learn, sample from, and evaluate probabilities in deep generative models with thousands of layers or time steps, as well as to compute conditional and posterior probabilities under the learned model. We additionally release an open source reference implementation of the algorithm.

Explicit solutions for multilayered wide plates and beams with perfect and imperfect bonding and delaminations under thermomechanical loading ; Explicit expressions, for efficient application in engineering practice, are derived for generalized displacements and stresses in simply supported multilayered wide plates and beams subjected to steadystate thermal and mechanical loading. The expressions are general and apply to plates composed of an arbitrary number of layers, of arbitrary thickness and elasticthermal properties, and where the interfaces between the layers may be imperfect and allow relative sliding. The closedform solutions are obtained using a multiscale homogenized model which depends on only three displacement variables and overcomes limitations of current approaches based on computationally expensive discretelayer models. The accuracy of the expressions in predicting the highly complex and discontinuous fields, which characterize the response of thick and highly anisotropic plates with interlayer damage and delaminations, is verified using exact 2D thermoelasticity solutions. The asymptotic limits of the modelsolution correspond to the problems of an intact and a multiply delaminated plate. They are derived using a perturbation technique, which also explains the multiscale dependence of the model on the parameters.

Rank tests for corrupted linear models ; For some variants of regression models, including partial, measurement error or errorinvariables, latent effects, semiparametric and otherwise corrupted linear models, the classical parametric tests generally do not perform well. Various modifications and generalizations considered extensively in the literature rests on stringent regularity assumptions which are not likely to be tenable in many applications. However, in such nonstandard cases, rank based tests can be adapted better, and further, incorporation of rank analysis of covariance tools enhance their powerefficiency. Numerical studies and a real data illustration show the superiority of rank based inference in such corrupted linear models.

Exploring the effects of a double reconstruction on the geometrical parameters of coupled models, using observational data ; In this work we study the effects of the nongravitational exchange energy Q between dark matter DM fluid and dark energy DE fluid on the background evolution of the cosmological parameters. A varying equation of state EOS parameter, omega, for DE is proposed. Considering an universe spatially flat, two distinct coupled models were examined to explore the main cosmological effects generated by the simultaneous reconstruction of Q and omega on the shape of the jerk parameter, j, through a slight enhancement or suppression of their amplitudes with respect to noncoupled scenarios, during its evolution from the past to the near future. In consequence, j could be used to distinguish any coupled DE models. Otherwise, the observational data were used to put stringent constraints on Q and omega, respectively. In such a way, we used our results as evidences to search possible deviations from the standard concordance model LambdaCDM, examining their predictions and improving our knowledge of the cosmic evolution of the universe.

Dynamo Saturation in Rapidly Rotating SolarType Stars ; The magnetic activity of solartype stars generally increases with stellar rotation rate. The increase, however, saturates for fast rotation. The BabcockLeighton mechanism of stellar dynamos saturates as well when the mean tiltangle of active regions approaches ninety degrees. Saturation of magnetic activity may be a consequence of this property of the BabcockLeighton mechanism. Stellar dynamo models with a tiltangle proportional to the rotation rate are constructed to probe this idea. Two versions of the model treating the tiltangles globally and using Joy's law for its latitude dependence are considered. Both models show a saturation of dynamogenerated magnetic flux at high rotation rates. The model with latitudedependent tiltangles also shows a change in dynamo regime in the saturation region. The new regime combines a cyclic dynamo at low latitudes with an almost steady polar dynamo.

Leader Election and Shape Formation with SelfOrganizing Programmable Matter ; We consider programmable matter consisting of simple computational elements, called particles, that can establish and release bonds and can actively move in a selforganized way, and we investigate the feasibility of solving fundamental problems relevant for programmable matter. As a suitable model for such selforganizing particle systems, we will use a generalization of the geometric amoebot model first proposed in SPAA 2014. Based on the geometric model, we present efficient localcontrol algorithms for leader election and line formation requiring only particles with constant size memory, and we also discuss the limitations of solving these problems within the general amoebot model.

Majorana neutrinos with point interactions ; We propose a realistic model with Majorana neutrinos in the framework of unifying the three generations of fermions by point interactions in an extra dimension. This model can simultaneously explain the origin of fermion generations, fermion masses and mixing, and the smallness of the masses of Majorana neutrinos. We show that there are two mechanisms working together to suppress the neutrino masses significantly, so we do not have to introduce a very large extradimension cutoff scale. One is the typeI seesaw mechanism and the other is the overlap integration of localized lepton wave functions. A singlet scalar with an exponentiallike VEV plays a central role in these two mechanisms. For consistency in this model we introduce a U1' gauge symmetry, which will be broken by the singlet scalar. Parameters of our model can fit the masses and flavor mixing data well. These parameters can also predict all CP violating phases including the Majorana ones and accidentally rescue the proton from decay.

Identifiability of directed Gaussian graphical models with one latent source ; We study parameter identifiability of directed Gaussian graphical models with one latent variable. In the scenario we consider, the latent variable is a confounder that forms a source node of the graph and is a parent to all other nodes, which correspond to the observed variables. We give a graphical condition that is sufficient for the Jacobian matrix of the parametrization map to be full rank, which entails that the parametrization is generically finitetoone, a fact that is sometimes also referred to as local identifiability. We also derive a graphical condition that is necessary for such identifiability. Finally, we give a condition under which generic parameter identifiability can be determined from identifiability of a model associated with a subgraph. The power of these criteria is assessed via an exhaustive algebraic computational study on models with 4, 5, and 6 observable variables.

Heider balance, asymmetric ties, and gender segregation ; To remove a cognitive dissonance in interpersonal relations, people tend to divide our acquaintances into friendly and hostile parts, both groups internally friendly and mutually hostile. This process is modeled as an evolution towards the Heider balance. A set of differential equations have been proposed and validated Kulakowski it et al, IJMPC 16 2005 707 to model the Heider dynamics of this social and psychological process. Here we generalize the model by including the initial asymmetry of the interprersonal relations and the direct reciprocity effect which removes this asymmetry. Our model is applied to the data on enmity and friendship in 37 school classes and 4 groups of teachers in M'exico. For each class, a stable balanced partition is obtained into two groups. The gender structure of the groups reveals stronger gender segregation in younger classes, i.e. of age below 12 years, a fact consistent with other general empirical results.

A mathematical model of the discrete 3disk for the 3dimensional Universe ; A mathematical model of the distribution function for the discrete 3disk is proposed in order to utilize in the statistical evolution equation of the 3dimensional Universe. The model distribution is constructed based on analyses in known exact solutions of recursion equations for the generating functions of the discrete 2disk.The proposed distribution function is compared with numerical simulations of the dynamical triangulation with S3 , and D3 topologies.The model distribution exhibits three types of phases characterized by geometrical nature of the disk with either 1, 2, or 3 dimensional structure.Transitions between those phases are either crossover, 1st order, or 2nd order depending on the parameters, which reflect the type of discretization and matter fields coupled to space.

The impact of upper tropospheric friction and Gilltype heating on the location and strength of the Tropical Easterly Jet Idealized physics in a dry Atmospheric General Circulation Model ; An atmospheric general circulation model AGCM with idealized and complete physics has been used to evaluate the Tropical Easterly Jet TEJ jet. In idealized physics, the role of upper tropospheric friction has been found to be important in getting realistic upper tropospheric zonal wind patterns in response to heating. In idealized physics, the location and strength of the TEJ as a response to Gill heating has been studied. Though the Gill model is considered to be widely successful in capturing the lower tropospheric response, it is found to be inadequate in explaining the location and strength of the upper level TEJ. Heating from the Gill model and realistic upper tropospheric friction does not lead to the formation of a TEJ.

A Physicsbased Analytical Model for Perovskite Solar Cells ; Perovskites are promising nextgeneration absorber materials for lowcost and highefficiency solar cells. Although perovskite cells are configured similar to the classical solar cells, their operation is unique and requires development of a new physical model for characterization, optimization of the cells, and prediction of the panel performance. In this paper, we develop such a physicsbased analytical model to describe the operation of different types of perovskite solar cells, explicitly accounting nonuniform generation, carrier selective transport layers, and voltagedependent carrier collection. The model would allow experimentalists to characterize key parameters of existing cells, understand performance bottlenecks, and predict performance of perovskitebased solar panel the obvious next step to the evolution of perovskite solar cell technology.

SU5inspired double beta decay ; The shortrange part of the neutrinoless double beta amplitude is generated via the exchange of exotic particles, such as charged scalars, leptoquarks andor diquarks. In order to give a sizeable contribution to the total decay rate, the masses of these exotics should be of the order of at most a few TeV. Here, we argue that these exotics could be the light i.e weakscale remnants of some BL violating variants of SU5. We show that unification of the standard model gauge couplings, consistent with proton decay limits, can be achieved in such a setup without the need to introduce supersymmetry. Since these nonminimal SU5inspired models violate BL, they generate Majorana neutrino masses and therefore make it possible to explain neutrino oscillation data. The light coloured particles of these models can potentially be observed at the LHC, and it might be possible to probe the origin of the neutrino masses with Delta L2 violating signals. As particular realizations of this idea, we present two models, one for each of the two possible treelevel topologies of neutrinoless double beta decay.