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

Area of Science:

• Computational Chemistry
• Statistical Mechanics

Background:

• The Flying Gaussian method enhances molecular simulations by using multiple replicas.
• A dynamic bias potential in this method raises concerns about free-energy surface accuracy.
• Reweighing methods are sensitive to the nature of the bias potential.

Purpose of the Study:

• To address concerns about the accuracy of free-energy surfaces predicted by the Flying Gaussian method.
• To investigate the nature of the bias potential from a multisystem perspective.
• To develop reliable methods for predicting free-energy surfaces in enhanced sampling simulations.

Main Methods:

• Simulating multiple replicas of a system using the Flying Gaussian method.
• Analyzing the bias potential from a multisystem viewpoint.
• Developing and applying two novel equations for free-energy surface prediction.

Main Results:

• The bias potential can be treated as static when considering the simulation from a multisystem perspective.
• Two new equations were derived for predicting the free-energy surface.
• The convergence of the derived equations was demonstrated, validating the approach.

Conclusions:

• The Flying Gaussian method's free-energy predictions are accurate when the bias potential is viewed as static across replicas.
• The presented equations offer a robust way to calculate free-energy surfaces in enhanced sampling simulations.
• This work resolves concerns regarding the dynamic nature of the bias potential in Flying Gaussian simulations.