arXiv subjects

Jiaxin Zhang

Publications and source records attributed to Jiaxin Zhang.

At least 145 records · Page 8Linked to original sources

arXiv2020

Thermodynamic Consistent Neural Networks for Learning Material Interfacial Mechanics

For multilayer materials in thin substrate systems, interfacial failure is one of the most challenges. The traction-separation relations (TSR) quantitatively describe the mechanical behavior of a material interface undergoing openings, which is critical to understand and predict interfacial failures under complex loadings. However, existing theoretical models have limitations on enough complexity and flexibility to well learn the real-world TSR from experimental observations. A neural network can fit well along with the loading paths but often fails to obey the laws of physics, due to a lack of experimental data and understanding of the hidden physical mechanism. In this paper, we propose a thermodynamic consistent neural network (TCNN) approach to build a data-driven model of the TSR with sparse experimental data. The TCNN leverages recent advances in physics-informed neural networks (PINN) that encode prior physical information into the loss function and efficiently train the neural networks using automatic differentiation. We investigate three thermodynamic consistent principles, i.e., positive energy dissipation, steepest energy dissipation gradient, and energy conservative loading path. All of them are mathematically formulated and embedded into a neural network model with a novel defined loss function. A real-world experiment demonstrates the superior performance of TCNN, and we find that TCNN provides an accurate prediction of the whole TSR surface and significantly reduces the violated prediction against the laws of physics.

cs.CE↗

arXiv2020

Scalable Deep-Learning-Accelerated Topology Optimization for Additively Manufactured Materials

Topology optimization (TO) is a popular and powerful computational approach for designing novel structures, materials, and devices. Two computational challenges have limited the applicability of TO to a variety of industrial applications. First, a TO problem often involves a large number of design variables to guarantee sufficient expressive power. Second, many TO problems require a large number of expensive physical model simulations, and those simulations cannot be parallelized. To address these issues, we propose a general scalable deep-learning (DL) based TO framework, referred to as SDL-TO, which utilizes parallel schemes in high performance computing (HPC) to accelerate the TO process for designing additively manufactured (AM) materials. Unlike the existing studies of DL for TO, our framework accelerates TO by learning the iterative history data and simultaneously training on the mapping between the given design and its gradient. The surrogate gradient is learned by utilizing parallel computing on multiple CPUs incorporated with a distributed DL training on multiple GPUs. The learned TO gradient enables a fast online update scheme instead of an expensive update based on the physical simulator or solver. Using a local sampling strategy, we achieve to reduce the intrinsic high dimensionality of the design space and improve the training accuracy and the scalability of the SDL-TO framework. The method is demonstrated by benchmark examples and AM materials design for heat conduction. The proposed SDL-TO framework shows competitive performance compared to the baseline methods but significantly reduces the computational cost by a speed up of around 8.6x over the standard TO implementation.

cs.CE↗

arXiv2020

Fast and stable deep-learning predictions of material properties for solid solution alloys

We present a novel deep learning (DL) approach to produce highly accurate predictions of macroscopic physical properties of solid solution binary alloys and magnetic systems. The major idea is to make use of the correlations between different physical properties in alloy systems to improve the prediction accuracy of neural network (NN) models. We use multitasking NN models to simultaneously predict the total energy, charge density and magnetic moment. These physical properties mutually serve as constraints during the training of the multitasking NN, resulting in more reliable DL models because multiple physics properties are correctly learned by a single model. Two binary alloys, copper-gold (CuAu) and iron-platinum (FePt), were studied. Our results show that once the multitasking NN's are trained, they can estimate the material properties for a specific configuration hundreds of times faster than first-principles density functional theory calculations while retaining comparable accuracy. We used a simple measure based on the root-mean-squared errors (RMSE) to quantify the quality of the NN models, and found that the inclusion of charge density and magnetic moment as physical constraints leads to more stable models that exhibit improved accuracy and reduced uncertainty for the energy predictions.

physics.comp-ph↗

arXiv2020

Modern Monte Carlo Methods for Efficient Uncertainty Quantification and Propagation: A Survey

Uncertainty quantification (UQ) includes the characterization, integration, and propagation of uncertainties that result from stochastic variations and a lack of knowledge or data in the natural world. Monte Carlo (MC) method is a sampling-based approach that has widely used for quantification and propagation of uncertainties. However, the standard MC method is often time-consuming if the simulation-based model is computationally intensive. This article gives an overview of modern MC methods to address the existing challenges of the standard MC in the context of UQ. Specifically, multilevel Monte Carlo (MLMC) extending the concept of control variates achieves a significant reduction of the computational cost by performing most evaluations with low accuracy and corresponding low cost, and relatively few evaluations at high accuracy and correspondingly high cost. Multifidelity Monte Carlo (MFMC) accelerates the convergence of standard Monte Carlo by generalizing the control variates with different models having varying fidelities and varying computational costs. Multimodel Monte Carlo method (MMMC), having a different setting of MLMC and MFMC, aims to address the issue of uncertainty quantification and propagation when data for characterizing probability distributions are limited. Multimodel inference combined with importance sampling is proposed for quantifying and efficiently propagating the uncertainties resulting from small datasets. All of these three modern MC methods achieve a significant improvement of computational efficiency for probabilistic UQ, particularly uncertainty propagation. An algorithm summary and the corresponding code implementation are provided for each of the modern Monte Carlo methods. The extension and application of these methods are discussed in detail.

stat.ME↗

arXiv2020

Monte Carlo simulation of order-disorder transition in refractory high entropy alloys: a data-driven approach

High entropy alloys (HEAs) are a series of novel materials that demonstrate many exceptional mechanical properties. To understand the origin of these attractive properties, it is important to investigate the thermodynamics and elucidate the evolution of various chemical phases. In this work, we introduce a data-driven approach to construct the effective Hamiltonian and study the thermodynamics of HEAs through canonical Monte Carlo simulation. The main characteristic of our method is to use pairwise interactions between atoms as features and systematically improve the representativeness of the dataset using samples from Monte Carlo simulation. We find this method produces highly robust and accurate effective Hamiltonians that give less than 0.1 mRy test error for all the three refractory HEAs: MoNbTaW, MoNbTaVW, and MoNbTaTiW. Using replica exchange to speed up the MC simulation, we calculated the specific heats and short-range order parameters in a wide range of temperatures. For all the studied materials, we find there are two major order-disorder transitions occurring respectively at $T_1$ and $T_2$, where $T_1$ is near room temperature but $T_2$ is much higher. We further demonstrate that the transition at $T_1$ is caused by W and Nb while the one at $T_2$ is caused by the other elements. By comparing with experiments, {\color{black} the results provide insight into the role of chemical ordering in the strength and ductility of HEAs.

cond-mat.mtrl-sci↗

arXiv2020

A Novel Evolution Strategy with Directional Gaussian Smoothing for Blackbox Optimization

We propose an improved evolution strategy (ES) using a novel nonlocal gradient operator for high-dimensional black-box optimization. Standard ES methods with $d$-dimensional Gaussian smoothing suffer from the curse of dimensionality due to the high variance of Monte Carlo (MC) based gradient estimators. To control the variance, Gaussian smoothing is usually limited in a small region, so existing ES methods lack nonlocal exploration ability required for escaping from local minima. We develop a nonlocal gradient operator with directional Gaussian smoothing (DGS) to address this challenge. The DGS conducts 1D nonlocal explorations along $d$ orthogonal directions in $\mathbb{R}^d$, each of which defines a nonlocal directional derivative as a 1D integral. We then use Gauss-Hermite quadrature, instead of MC sampling, to estimate the $d$ 1D integrals to ensure high accuracy (i.e., small variance). Our method enables effective nonlocal exploration to facilitate the global search in high-dimensional optimization. We demonstrate the superior performance of our method in three sets of examples, including benchmark functions for global optimization, and real-world science and engineering applications.

math.OC↗

arXiv2020

On the quantification and efficient propagation of imprecise probabilities with copula dependence

This paper addresses the problem of quantification and propagation of uncertainties associated with dependence modeling when data for characterizing probability models are limited. Practically, the system inputs are often assumed to be mutually independent or correlated by a multivariate Gaussian distribution. However, this subjective assumption may introduce bias in the response estimate if the real dependence structure deviates from this assumption. In this work, we overcome this limitation by introducing a flexible copula dependence model to capture complex dependencies. A hierarchical Bayesian multimodel approach is proposed to quantify uncertainty in dependence model-form and model parameters that result from small data sets. This approach begins by identifying, through Bayesian multimodel inference, a set of candidate marginal models and their corresponding model probabilities, and then estimating the uncertainty in the copula-based dependence structure, which is conditional on the marginals and their parameters. The overall uncertainties integrating marginals and copulas are probabilistically represented by an ensemble of multivariate candidate densities. A novel importance sampling reweighting approach is proposed to efficiently propagate the overall uncertainties through a computational model. Through an example studying the influence of constituent properties on the out-of-plane properties of transversely isotropic E- glass fiber composites, we show that the composite property with copula-based dependence model converges to the true estimate as data set size increases, while an independence or arbitrary Gaussian correlation assumption leads to a biased estimate.

stat.CO↗

arXiv2020

PBGen: Partial Binarization of Deconvolution-Based Generators for Edge Intelligence

This work explores the binarization of the deconvolution-based generator in a GAN for memory saving and speedup of image construction. Our study suggests that different from convolutional neural networks (including the discriminator) where all layers can be binarized, only some of the layers in the generator can be binarized without significant performance loss. Supported by theoretical analysis and verified by experiments, a direct metric based on the dimension of deconvolution operations is established, which can be used to quickly decide which layers in the generator can be binarized. Our results also indicate that both the generator and the discriminator should be binarized simultaneously for balanced competition and better performance. Experimental results based on CelebA suggest that directly applying state-of-the-art binarization techniques to all the layers of the generator will lead to 2.83$\times$ performance loss measured by sliced Wasserstein distance compared with the original generator, while applying them to selected layers only can yield up to 25.81$\times$ saving in memory consumption, and 1.96$\times$ and 1.32$\times$ speedup in inference and training respectively with little performance loss.

cs.CV↗

arXiv2020

Antiferromagnetic quantum spin Hall states in iron halogenide

It is widely known that quantum spin Hall (QSH) insulator can be viewed as two copies of quantum anomalous Hall (QAH) insulator with opposite local magnetic moments. However, nearly every QSH insulator discovered so far is a nonmagnetic semiconductor. Due to the vanishing local magnetic moment of each copy, the QAH states only conceptually exist in these QSH insulators. In this work, we show a realistic construction of QSH states with finite local magnetic moment by staking bilayer QAH insulators. Our explicit construction benefits from an effective QAH model with a large topological gap and is further supported by a class of two-dimensional ferromagnetic materials. Our work not only validates the conceptual relationship of QSH and QAH but also provides an ideal material platform for realizing antiferromagnetic QSH state which is highly tunable between QAH and QSH states as a function of the number of layers.

cond-mat.mtrl-sci↗

arXiv2019

Machine Learning the Effective Hamiltonian in High Entropy Alloys

The development of machine learning sheds new light on the problem of statistical thermodynamics in multicomponent alloys. However, a data-driven approach to construct the effective Hamiltonian requires sufficiently large data sets, which is expensive to calculate with conventional density functional theory (DFT). To solve this problem, we propose to use the atomic local energy as the target variable, and harness the power of the linear-scaling DFT to accelerate the data generating process. Using the large amounts of DFT data sets, various complex models are devised and applied to learn the effective Hamiltonians of a range of refractory high entropy alloys (HEAs). The testing $R^2$ scores of the effective pair interaction model are higher than 0.99, demonstrating that the pair interactions within the 6-th coordination shell provide an excellent description of the atomic local energies for all the four HEAs. This model is further improved by including nonlinear and multi-site interactions. In particular, the deep neural networks (DNNs) built directly in the local configuration space (therefore no hand-crafted features) are employed to model the effective Hamiltonian. The results demonstrate that neural networks are promising for the modeling of effective Hamiltonian due to its excellent representation power.

cond-mat.mtrl-sci↗

arXiv2019

Robust data-driven approach for predicting the configurational energy of high entropy alloys

High entropy alloys (HEAs) have been increasingly attractive as promising next-generation materials due to their various excellent properties. It's necessary to essentially characterize the degree of chemical ordering and identify order-disorder transitions through efficient simulation and modeling of thermodynamics. In this study, a robust data-driven framework based on Bayesian approaches is proposed and demonstrated on the accurate and efficient prediction of configurational energy of high entropy alloys. The proposed effective pair interaction (EPI) model with ensemble sampling is used to map the configuration and its corresponding energy. Given limited data calculated by first-principles calculations, Bayesian regularized regression not only offers an accurate and stable prediction but also effectively quantifies the uncertainties associated with EPI parameters. Compared with the arbitrary determination of model complexity, we further conduct a physical feature selection to identify the truncation of coordination shells in EPI model using Bayesian information criterion. The results achieve efficient and robust performance in predicting the configurational energy, particularly given small data. The developed methodology is applied to study a series of refractory HEAs, i.e. NbMoTaW, NbMoTaWV and NbMoTaWTi where it is demonstrated how dataset size affects the confidence we can place in statistical estimates of configurational energy when data are sparse.

cond-mat.mtrl-sci↗

arXiv2019

Chemical complexity in high entropy alloys: A pair-interaction perspective

The recently proposed pair-interaction model is applied to study a series of refractory high entropy alloys. The results demonstrate the simplicity, robustness, and high accuracy of this model in predicting the configuration energies of NbMoTaW, NbMoTaWV and NbMoTaWTi. The element-element pair interaction parameters obtained from the linear regression of first-principle data provide a new perspective to understand the strengthening mechanism in HEAs, as revealed by comparing the effects of adding vanadium and titanium. Using the pair-interaction model, an expression for the intrinsic energy fluctuation is derived, which provides guidance on both theoretical modeling and first principles calculation.

cond-mat.mtrl-sci↗

arXiv2019

Learning nonlinear level sets for dimensionality reduction in function approximation

We developed a Nonlinear Level-set Learning (NLL) method for dimensionality reduction in high-dimensional function approximation with small data. This work is motivated by a variety of design tasks in real-world engineering applications, where practitioners would replace their computationally intensive physical models (e.g., high-resolution fluid simulators) with fast-to-evaluate predictive machine learning models, so as to accelerate the engineering design processes. There are two major challenges in constructing such predictive models: (a) high-dimensional inputs (e.g., many independent design parameters) and (b) small training data, generated by running extremely time-consuming simulations. Thus, reducing the input dimension is critical to alleviate the over-fitting issue caused by data insufficiency. Existing methods, including sliced inverse regression and active subspace approaches, reduce the input dimension by learning a linear coordinate transformation; our main contribution is to extend the transformation approach to a nonlinear regime. Specifically, we exploit reversible networks (RevNets) to learn nonlinear level sets of a high-dimensional function and parameterize its level sets in low-dimensional spaces. A new loss function was designed to utilize samples of the target functions' gradient to encourage the transformed function to be sensitive to only a few transformed coordinates. The NLL approach is demonstrated by applying it to three 2D functions and two 20D functions for showing the improved approximation accuracy with the use of nonlinear transformation, as well as to an 8D composite material design problem for optimizing the buckling-resistance performance of composite shells of rocket inter-stages.

math.FA↗

arXiv2019

Machine learning modeling of high entropy alloy: the role of short-range order

The development of machine learning sheds new light on the traditionally complicated problem of thermodynamics in multicomponent alloys. Successful application of such a method, however, strongly depends on the quality of the data and model. Here we propose a scheme to improve the representativeness of the data by utilizing the short-range order (SRO) parameters to survey the configuration space. Using the improved data, a pair interaction model is trained for the NbMoTaW high entropy alloy using linear regression. Benefiting from the physics incorporated into the model, the learned effective Hamiltonian demonstrates excellent predictability over the whole configuration space. By including pair interactions within the 6th nearest-neighbor shell, this model achieves an $R^2$ testing score of 0.997 and root mean square error of 0.43 meV. We further perform a detailed analysis on the effects of training data, testing data, and model parameters. The results reveal the vital importance of representative data and physical model. On the other hand, we also examined the performance neural networks, which is found to demonstrate a strong tendency to overfit the data.

cond-mat.mtrl-sci↗

arXiv2019

The structures and decompositions of symmetries involving idempotents

Let $\mathcal{H}$ be a separable Hilbert space and $P$ be an idempotent on $\mathcal{H}.$ We denote by $$Γ_{P}=\{J: J=J^{\ast}=J^{-1} \hbox{ }\hbox{ and }\hbox{ } JPJ=I-P\}$$ and $$Δ_{P}=\{J: J=J^{\ast}=J^{-1} \hbox{ }\hbox{ and }\hbox{ } JPJ=I-P^*\}.$$ In this paper, we first get that symmetries $(2P-I)|2P-I|^{-1}$ and $(P+P^{*}-I)|P+P^{*}-I|^{-1}$ are the same. Then we show that $Γ_{P}\neq\emptyset$ if and only if $Δ_{P}\neq\emptyset.$ Also, the specific structures of all symmetries $J\inΓ_{P}$ and $J\inΔ_{P} $ are established, respectively. Moreover, we prove that $J\inΔ_{P}$ if and only if $\sqrt{-1}J(2P-I)|2P-I|^{-1}\inΓ_{P}.$

math.FA↗

arXiv2018

The minimal and maximal symmetries for $J$-contractive projections

In this paper, we firstly character the structures of symmetries $J$ such that a projection $P$ is $J$-contractive. Then the minimal and maximal elements of the symmetries $J$ with $P^{\ast}JP\leqslant J$(or $JP\geqslant0)$ are given. Moreover, some formulas between $P_{(2I-P-P^{\ast})^{+}}$ $(P_{(2I-P-P^{\ast})^{-}})$ and $P_{(P+P^{\ast})^-}$ $(P_{(P+P^{\ast})^+})$ are established.

math.FA↗

arXiv2018

Probability measure changes in Monte Carlo simulation

The objective of Bayesian inference is often to infer, from data, a probability measure for a random variable that can be used as input for Monte Carlo simulation. When datasets for Bayesian inference are small, a principle challenge is that, as additional data are collected, the probability measure inferred from Bayesian inference may change significantly. That is, the original probability density inferred from Bayesian inference may differ considerably from the updated probability density both in its model form and parameters. In such cases, expensive Monte Carlo simulations may have already been performed using the original distribution and it is infeasible to start again and perform a new Monte Carlo analysis using the updated density due to the large added computational cost. In this work, we discuss four strategies for updating Mote Carlo simulations for such a change in probability measure: 1. Importance sampling reweighting; 2. A sample augmenting strategy; 3. A sample filtering strategy; and 4. A mixed augmenting-filtering strategy. The efficiency of each strategy is compared and the ultimate aim is to achieve the change in distribution with a minimal number of added computational simulations. The comparison results show that when the change in measure is small importance sampling reweighting can be very effective. Otherwise, a proposed novel mixed augmenting-filtering algorithm can robustly and efficiently accommodate a measure change in Monte Carlo simulation that minimizes the impact on the sample set and saves a large amount of additional computational cost. The strategy is then applied for uncertainty quantification in the buckling strength of a simple plate given ongoing data collection to estimate uncertainty in the yield stress.

stat.CO↗

arXiv2018

The effect of prior probabilities on quantification and propagation of imprecise probabilities resulting from small datasets

This paper outlines a methodology for Bayesian multimodel uncertainty quantification (UQ) and propagation and presents an investigation into the effect of prior probabilities on the resulting uncertainties. The UQ methodology is adapted from the information-theoretic method previously presented by the authors (Zhang and Shields, 2018) to a fully Bayesian construction that enables greater flexibility in quantifying uncertainty in probability model form. Being Bayesian in nature and rooted in UQ from small datasets, prior probabilities in both probability model form and model parameters are shown to have a significant impact on quantified uncertainties and, consequently, on the uncertainties propagated through a physics-based model. These effects are specifically investigated for a simplified plate buckling problem with uncertainties in material properties derived from a small number of experiments using noninformative priors and priors derived from past studies of varying appropriateness. It is illustrated that prior probabilities can have a significant impact on multimodel UQ for small datasets and inappropriate (but seemingly reasonable) priors may even have lingering effects that bias probabilities even for large datasets. When applied to uncertainty propagation, this may result in probability bounds on response quantities that do not include the true probabilities.

stat.CO↗