materials-science

Overview

Our research focuses on finding and generating natural structures in high-dimensional, multi-model, and multi-scale Material datasets using Physics-based Machine Learning and Deep Learning to accelerate Material discovery and close the gap in the processing-structure-properties (PSP) relationship. Each time getting data from physical experiments are time-consuming and costly. Therefore numerical simulation tools are used to generate synthetic data. However, results from existing physics-based numerical simulation models are not up to mark and computationally infeasible at high resolution in space and time. We are designing neural network-based generative models using physics to generate synthetic materials from given properties and structures.