Deep dive into machine learning density functional theory for materials science and chemistry

Abstract

Electronic structure simulations enable the calculation of a wide variety of fundamental materials properties. However, they consume a significant portion of scientific HPC resources worldwide. Artificial intelligence and machine learning, which have emerged as a powerful tool for analyzing complex datasets, have the potential to accelerate electronic structure calculations such as density functional theory. The combination of these two fields enables highly efficient simulations at unprecedented scales. In this review, the authors present a comprehensive analysis of research articles in chemistry and materials science that employ machine-learning techniques and outline the current trends at the intersection of these fields.