Path Integral Stochastic Optimal Control for Sampling Transition Paths in Molecular Dynamics

Explore a comprehensive lecture on sampling transition paths in molecular systems using path integral stochastic optimal control. Delve into the challenges of sampling transition paths between metastable conformational states of molecules and learn about a novel method that considers the entire geometry of molecules without relying on collective variables. Discover the connection between this problem and recent work on the Schrodinger bridge problem and stochastic optimal control. Examine how the proposed method accounts for important characteristics of molecular systems, including second-order dynamics and invariance to rotations and translations. Follow along as the speaker demonstrates the application of this method to various molecular systems, including Alanine Dipeptide, Polyproline, and Chignolin. Gain insights into the background of molecular dynamics, transition path ensembles, and prior work in the field before diving into the specifics of the proposed Path Integral Stochastic Optimal Control for Transition Path (PIPS) method.

- Intro
- Contributions/Outline
- Background: Molecular Dynamics
- Transition Path Ensemble
- Prior Work
- Bias Enhanced Transition Path TP Ensemble: The Problem
- Stochastic Optimal Control SOC for Sampling the TP Ensemble
- SOC Solves the Bias Enhanced TP Problem
- PIPS: Path Integral SOC for Transition Path
- Results: Alanine Dipeptide
- Polyproline Helix
- Chignon
- Summary
- Q&A