A Gaussian kernel Monte Carlo resampling method to construct smooth free energy surface from discrete simulation...

Area of Science:

• Computational Chemistry
• Biophysics
• Statistical Mechanics

Background:

• Constructing free energy surfaces from molecular simulation data is crucial in chemistry and biophysics.
• Traditional histogram-based methods struggle with sparse data, leading to rough surfaces.
• Existing histogram-free methods often sacrifice local details for global smoothness.

Purpose of the Study:

• To develop a robust method for constructing smooth free energy surfaces from discrete molecular simulation data.
• To address the limitations of existing methods in accurately depicting both global and local free energy features.
• To introduce the Gaussian kernel Monte Carlo (GKMC) resampling method.

Main Methods:

• Developed and applied the Gaussian kernel Monte Carlo (GKMC) resampling method.
• Mapped the free energy surface as a sum of local Gaussian basis functions.
• Utilized Monte Carlo resampling to determine Gaussian basis function heights from simulation data.
• Illustrated the method using data from a generalized orthogonal space tempering simulation of deca-alanine peptide.

Main Results:

• GKMC resampling robustly generated smooth free energy surfaces accurately representing simulated probability distributions.
• The method effectively removed data noise, enabling informative display of local free energy features.
• Local free energy smoothness could be adjusted without altering the global free energy shape by modifying the Gaussian kernel width.

Conclusions:

• GKMC resampling is a robust and effective approach for high-quality free energy surface construction.
• The method overcomes limitations of traditional techniques, providing accurate global and local free energy landscape representations.
• GKMC offers a valuable tool for computational studies requiring detailed free energy surface analysis.