Neural Mode Jump Monte Carlo

TL;DR

This paper introduces a neural network-based Monte Carlo method that enhances sampling efficiency by enabling direct jumps between metastable states, addressing slow convergence issues in complex systems.

Contribution

It presents a novel neural mode jump Monte Carlo method that improves convergence by connecting metastable states with generative neural networks for efficient sampling.

Findings

Increased convergence speed in systems with many metastable states.

Effective direct mode jumps via neural network proposals.

Theoretical foundation and demonstration on example systems.

Abstract

Markov chain Monte Carlo methods are a powerful tool for sampling equilibrium configurations in complex systems. One problem these methods often face is slow convergence over large energy barriers. In this work, we propose a novel method which increases convergence in systems composed of many metastable states. This method aims to connect metastable regions directly using generative neural networks in order to propose new configurations in the Markov chain and optimizes the acceptance probability of large jumps between modes in configuration space. We provide a comprehensive theory and demonstrate the method on example systems.