Stein Kernelized Molecular Dynamics for Active Learning of Interatomic Potentials

2026-06-02

AI Summary

This work addresses the limitation of machine-learned interatomic potentials, whose performance is constrained by the diversity and informativeness of training data, while conventional sampling methods remain inefficient. The authors propose Stein Kernelized Molecular Dynamics (SKMD), which for the first time integrates Stein variational gradient descent into molecular dynamics sampling. By combining a global atomic descriptor kernel, an asynchronous particle update mechanism, and a symmetry-aware configuration similarity metric, SKMD enables efficient active exploration of high-information configurations while rigorously preserving the Boltzmann distribution. Through an adaptive online selection strategy, the method significantly reduces the number of iterations and enhances model accuracy at equal sampling cost, as demonstrated on the Müller-Brown potential and MACE fine-tuning for alanine dipeptide, effectively balancing exploration and exploitation.

Abstract

Machine learning interatomic potentials (MLIPs) enable efficient and accurate atomistic simulations but depend critically on the quality and diversity of the training data. We introduce Stein kernelized molecular dynamics (SKMD), an enhanced sampling method that uses interacting particle dynamics to acquire informative training configurations for the active learning and fine-tuning of MLIPs. SKMD corresponds to a stochastic variant of Stein variational gradient descent that is adapted for molecular dynamics by incorporating asynchronous particle updates and a kernel of global atomic descriptors, which provides a symmetry-aware measure of configurational similarity. Unlike other enhanced samplers used in molecular dynamics, SKMD preserves the Boltzmann distribution as the asymptotic distribution of the dynamics. This property enforces a balance between the exploration of diverse configurations and attraction toward high-probability regions of the energy landscape. We further propose an approach to efficient online data acquisition using an adaptive stopping criterion that selects non-redundant training data over the course of simulation. We demonstrate SKMD for the active learning of a neural network model of the Müller-Brown potential and the fine-tuning of a MACE interatomic potential for alanine dipeptide. Compared to active learning baselines, our method achieves higher model accuracy in fewer training iterations with the same number of acquired training samples.

Problem

• machine learning interatomic potentials
• active learning
• training data diversity
• molecular dynamics
• enhanced sampling

Innovation

• Stein kernelized molecular dynamics
• active learning
• interatomic potentials
• enhanced sampling
• symmetry-aware kernel