Reversible molecular simulation for training classical and machine learning force fields

TL;DR

This paper introduces a reversible molecular simulation method that efficiently computes gradients for training classical and machine learning force fields, improving accuracy and enabling training from scratch.

Contribution

It presents a reversible simulation approach that explicitly calculates gradients with constant memory cost, enhancing training of force fields from data and simulations.

Findings

Reversible simulation improves gradient accuracy over ensemble reweighting.

Method successfully trains water, gas diffusion models, and a diamond potential.

Reversible approach matches time-dependent observables more effectively.

Abstract

The next generation of force fields for molecular dynamics will be developed using a wealth of data. Training systematically with experimental data remains a challenge, however, especially for machine learning potentials. Differentiable molecular simulation calculates gradients of observables with respect to parameters through molecular dynamics trajectories. Here we improve this approach by explicitly calculating gradients using a reverse-time simulation with effectively constant memory cost and a computation count similar to the forward simulation. The method is applied to learn all-atom water and gas diffusion models with different functional forms, and to train a machine learning potential for diamond from scratch. Comparison to ensemble reweighting indicates that reversible simulation can provide more accurate gradients and train to match time-dependent observables.