Stochastic Control Policies for Robust Molecular Transition Path Sampling

2026-08-13

AI Summary

This study addresses the performance instability and seed sensitivity inherent in control-based molecular transition path sampling by reformulating control strategies as proposal distributions within path space. We introduce two stochastic methods, FS-TPS and LaS-TPS, which optimize stochasticity through state-dependent Gaussian parameterization and latent variable decoding. These techniques significantly enhance sampling robustness and exploration efficiency. Experiments on multiscale biomolecular systems demonstrate that the proposed approach substantially reduces sensitivity to initialization while effectively improving both transition success rates and path quality. Collectively, these results validate the superiority of our method for sampling complex molecular systems, offering a reliable solution to longstanding challenges in transition path generation.

Abstract

Transition path sampling (TPS) aims to efficiently generate rare molecular transition trajectories between metastable states and is essential for understanding biomolecular mechanisms. Beyond traditional molecular dynamics (MD)-based sampling, machine learning has become central to state-of-the-art TPS. One major class of methods learns control forces during explicit MD rollouts. By preserving the underlying molecular dynamics, these methods tend to produce more physically plausible trajectories than endpoint-conditioned generators that construct paths directly. However, rollout-based control methods have been reported to exhibit unstable and strongly seed-dependent performance. We recast rollout-based control as learning a path-space proposal distribution and investigate stochasticity placement as a design choice for improving exploration and optimization robustness. We develop two stochastic policies: FS-TPS, which directly parameterizes a state-dependent Gaussian distribution over the control policy output, and LaS-TPS, which samples a compact latent control variable and decodes it into structured, cross-atom-correlated force variation. We conduct extensive multi-seed experiments on three biomolecular systems of increasing size: alanine dipeptide, chignolin, and BBL, a fast-folding protein. Stochastic policies consistently improve transition success and path quality over deterministic-policy baselines while substantially reducing sensitivity to random initialization.

Problem

• Transition Path Sampling
• Stochastic Control Policies
• Robustness
• Seed Dependence
• Molecular Dynamics

Innovation

• Stochastic Control Policies
• Transition Path Sampling
• Path-space Proposal Distribution
• Latent Control Variable
• Optimization Robustness