Machine learning improves molecular transition rate predictions
Machine learning kinetics from molecular dynamics data
Machine Learning
Summary
Many important changes in molecules happen too slowly to watch directly with computer simulations. The authors explain how to use machine learning to estimate the chance that a molecule will change from one state to another, called the committor. These estimates help predict reaction rates without needing to track every possible step. They review different machine learning methods that learn from the molecular movements themselves, making it easier to understand and predict chemical reactions. Their work also highlights ways to handle complex situations and suggests future research directions.
What this means in practice
- •For computational chemists: Use advanced machine learning techniques to better estimate molecular transition probabilities and reaction rates from simulation data without relying on labeled paths.
- •For pharmaceutical modelers: Improve prediction of drug binding/unbinding kinetics by learning transition probabilities directly from molecular simulation trajectories.
A survey. It maps existing work.
Authors
Jonathan Weare, Aaron R. Dinner
Abstract
Most molecular transitions occur on timescales far beyond direct molecular dynamics simulations. The committor, the probability that a configuration reaches a product state before a reactant state, is a central kinetic statistic, providing a mechanism-independent reaction coordinate and a foundation for transition path theory and the calculation of rates. This review surveys modern approaches for estimating the committor and related kinetic statistics from molecular simulations, with an emphasis on self-supervised methods that learn solutions of their defining dynamical equations rather than relying on labeled shooting data. We develop a common operator viewpoint connecting generator-based partial differential equations, variational principles, Markov state models, dynamical Galerkin approximation, and neural networks. Empirical and theoretical evidence points to the efficiency of these methods. We provide theoretical and practical guidance for realizing their full potential in applications, including strategies for treating non-Markovian effects and for sampling. We conclude by identifying opportunities for further research, including connections to reinforcement learning and generative modeling.