The Evolution of Machine Learning Potentials for Molecules, Reactions and Materials

TL;DR

This review traces the development and recent advances of machine learning potentials in chemistry, physics, and materials science, highlighting their ability to enable accurate, large-scale atomistic simulations from ab initio data.

Contribution

It provides a comprehensive overview of the evolution, recent state-of-the-art methods, applications, and future challenges of machine learning potentials over the past two decades.

Findings

MLPs enable accurate, efficient atomistic simulations.

Recent MLPs achieve universal applicability across systems.

Open challenges include improving transferability and scalability.

Abstract

Recent years have witnessed the fast development of machine learning potentials (MLPs) and their widespread applications in chemistry, physics, and material science. By fitting discrete ab initio data faithfully to continuous and symmetry-preserving mathematical forms, MLPs have enabled accurate and efficient atomistic simulations in a large scale from first principles. In this review, we provide an overview of the evolution of MLPs in the past two decades and focus on the state-of-the-art MLPs proposed in the last a few years for molecules, reactions, and materials. We discuss some representative applications of MLPs and the trend of developing universal potentials across a variety of systems. Finally, we outline a list of open challenges and opportunities in the development and applications of MLPs.