• Corpus ID: 53770688

Physics-aware Deep Generative Models for Creating Synthetic Microstructures

  title={Physics-aware Deep Generative Models for Creating Synthetic Microstructures},
  author={Rahul Dheerendra Singh and Viraj Shah and Balaji Sesha Sarath Pokuri and Soumik Sarkar and Baskar Ganapathysubramanian and Chinmay Hegde},
A key problem in computational material science deals with understanding the effect of material distribution (i.e., microstructure) on material performance. The challenge is to synthesize microstructures, given a finite number of microstructure images, and/or some physical invariances that the microstructure exhibits. Conventional approaches are based on stochastic optimization and are computationally intensive. We introduce three generative models for the fast synthesis of binary… 

Figures from this paper

A Conditional Generative Model for Predicting Material Microstructures from Processing Methods

This work proposes a deep learning based approach that models the processing-structure relationship as a conditional image synthesis problem and develops an auxiliary classifier Wasserstein GAN with gradient penalty (ACWGAN-GP) to synthesize microstructures under a given processing condition.

Three-dimensional microstructure generation using generative adversarial neural networks in the context of continuum micromechanics

Virtual microstructure design for steels using generative adversarial networks

The prediction of macro‐scale materials properties from microstructures, and vice versa, should be a key part in modeling quantitative microstructure‐physical property relationships. It would be

Microstructure Generation via Generative Adversarial Network for Heterogeneous, Topologically Complex 3D Materials

Using a large-scale, experimentally captured 3D microstructure data set, we implement the generative adversarial network (GAN) framework to learn and generate 3D microstructures of solid oxide fuel

Microstructure Generation via Generative Adversarial Network for Heterogeneous, Topologically Complex 3D Materials

The ability of the generative machine learning model to recreate microstructural parameters with high fidelity suggests that the essence of complex microstructures may be captured and represented in a compact and manipulatable form.

A Survey on Evaluation Metrics for Synthetic Material Micro-Structure Images from Generative Models

A variety of methods are evaluated on scanning electron microscope (SEM) images of graphene-reinforced polyurethane foams to encourage the machine learning community to consider enhancements in metrics for assessing quality of synthetic images in the material science domain.

Interpretable deep learning for guided microstructure-property explorations in photovoltaics

A data-driven approach for mapping the microstructure to photovoltaic performance using deep convolutional neural networks, characterized in terms of two critical metrics, its generalizability, and its intepretability.

Physics-consistent deep learning for structural topology optimization

A deep learning-based framework for performing topology optimization for three-dimensional geometries with a reasonably fine (high) resolution is explored, able to achieve this by training multiple networks, each trying to learn a different aspect of the overall topology optimize methodology.

Physics-guided neural networks (PGNNs) to solve differential equations for spatial analysis

This publication explains the concept of physicsguided neural network (PGNN), makes an overview of already proposed solutions in the field and describes possibilities of implementing physics-based loss functions for spatial analysis and presents two applications of PGNNs.

Algorithmically-consistent deep learning frameworks for structural topology optimization

A non-linear dimension reduction methodology for generating data-driven stochastic input models

Reconstructing random media

We formulate a procedure to reconstruct the structure of general random heterogeneous media from limited morphological information by extending the methodology of Rintoul and Torquato [J. Colloid

Generative Adversarial Nets

We propose a new framework for estimating generative models via an adversarial process, in which we simultaneously train two models: a generative model G that captures the data distribution, and a

Microstructures to control elasticity in 3D printing

This work builds on work in the area of metamaterials, using numerical optimization to design tiled microstructures with desired properties, but with the key difference that this method designs families of related structures that can be interpolated to smoothly vary the material properties over a wide range.

Statistical reconstruction of three-dimensional porous media from two-dimensional images

A method of modeling the three-dimensional microstructure of random isotropic two-phase materials is proposed. The information required to implement the technique can be obtained from two-dimensional