Quantum statistics from classical simulations via generative Gibbs sampling

TL;DR

This paper introduces GG-PI, a novel framework that leverages generative modeling and Gibbs sampling to efficiently recover quantum statistics from classical simulations, significantly reducing computational costs in molecular modeling.

Contribution

The paper presents GG-PI, a new method combining generative modeling and Gibbs sampling to simulate nuclear quantum effects from classical data without retraining across temperatures.

Findings

Reduces wall clock time compared to PIMD

Works with standard classical simulations or existing data

Easily extends to various problems with similar Markov structure

Abstract

Accurate simulation of nuclear quantum effects is essential for molecular modeling but expensive using path integral molecular dynamics (PIMD). We present GG-PI, a ring-polymer-based framework that combines generative modeling of the single-bead conditional density with Gibbs sampling to recover quantum statistics from classical simulation data. GG-PI uses inexpensive standard classical simulations or existing data for training and allows transfer across temperatures without retraining. On standard test systems, GG-PI significantly reduces wall clock time compared to PIMD. Our approach extends easily to a wide range of problems with similar Markov structure.