Free-Energy Surface Prediction by Flying Gaussian Method: Multisystem Representation

Abstract

Flying Gaussian method simulates multiple replicas of the studied system and enhances sampling by disfavoring replicas to simultaneously sample similar states. The bias potential used for this enhancement is highly dynamic when looking at individual replica, which raises concerns about the accuracy of free-energy surfaces predicted by reweighing methods. Here we show that the bias potential can be considered as static when looking at the simulation from a multisystem perspective. We present two equations that can be used to predict the free-energy surface, and we demonstrate their convergence.