Optimum bias for fast-switching free energy calculations

TL;DR

This paper derives an optimal bias function to minimize error in free energy calculations during fast-switching simulations, demonstrating improved sampling efficiency and reduced uncertainty.

Contribution

It introduces a new bias function optimized for fast-switching free energy calculations, enhancing sampling accuracy over existing methods.

Findings

Optimal bias function reduces statistical error.

Sampling both dominant and typical work values improves accuracy.

Method outperforms traditional bias functions in simulations.

Abstract

We derive the bias function that minimizes the statistical error of free energy differences calculated in work-biased fast-switching simulations. The optimum bias function is compared to other bias functions using a particle pulled through a viscous fluid as an illustrative example. Our analysis indicates that the uncertainty in the free energy is smallest if both dominant and typical work values are sampled with high frequency.