Machine learning for molecular simulation

TL;DR

This paper reviews recent machine learning methods applied to molecular simulation, highlighting neural networks for energy prediction, coarse-grained dynamics, free energy extraction, and generative models, while discussing open challenges in the field.

Contribution

It provides a comprehensive overview of ML techniques in molecular simulation, emphasizing recent advances and identifying key open problems for future research.

Findings

Neural networks effectively predict quantum energies and forces.

ML accelerates sampling of molecular structures and thermodynamics.

Identification of open challenges in integrating ML with molecular physics.

Abstract

Machine learning (ML) is transforming all areas of science. The complex and time-consuming calculations in molecular simulations are particularly suitable for a machine learning revolution and have already been profoundly impacted by the application of existing ML methods. Here we review recent ML methods for molecular simulation, with particular focus on (deep) neural networks for the prediction of quantum-mechanical energies and forces, coarse-grained molecular dynamics, the extraction of free energy surfaces and kinetics and generative network approaches to sample molecular equilibrium structures and compute thermodynamics. To explain these methods and illustrate open methodological problems, we review some important principles of molecular physics and describe how they can be incorporated into machine learning structures. Finally, we identify and describe a list of open challenges for the interface between ML and molecular simulation.