Physics-informed machine learning for functional materials
ABSTRACT :
In recent years, artificial
intelligence, and machine learning (AI/ML) methods are being rapidly adapted in physical sciences to gain comprehensive understandings of material structures, properties, system evolutions over spatial-temporal resolution, processes involving phase transitions across various time- and length scales. However, the in-built correlative nature of traditional ML techniques fails to capture the causal mechanisms driving any physical phenomena. Therefore, the challenge lies in development of causal ML methods combined with ML/deep learning (DL) techniques to integrate physics-based prior knowledge, aiding materials design, discovery while using knowledge from both high-performance computing (HPC) simulations, and automated experiments. This presentation will showcase how causal learning can be employed to understand fundamental atomistic mechanisms behind A-site cation ordering exhibited by polar double perovskite oxides. Details on physics-informed active learning workflows to bridge between experiments with atomistic simulations for appropriate feature finding, structure-property predictions for a few instances of graphene physics will also be included in the presentation. Finally, we aim towards designing physics-informed ML schemes that can exploit knowledge from physical models along with observational data to explore causal mechanisms underpinning materials structure and functionality.
Acknowledgements: This effort (machine learning) is based upon work supported by the U.S. Department of Energy (DOE), Office of Science, Office of Basic Energy Sciences Data, Artificial Intelligence and Machine Learning at DOE Scientific User Facilities (A.G.). Part of this research was conducted at the Center for Nanophase Materials Sciences, which is a DOE Office of Science User Facility.