Iterative Unbiasing of Quasi-Equilibrium Sampling.

Area of Science:

• Computational Chemistry
• Materials Science
• Statistical Mechanics

Background:

• Atomistic modeling of rare events faces long simulation times due to high free energy barriers.
• Bias potentials accelerate phase space exploration but distort microstate distributions.
• Efficient reweighting is challenging for time-dependent biases in adaptive sampling methods.

Purpose of the Study:

• To develop an efficient iterative unbiasing scheme for adaptive sampling techniques.
• To enable accurate recovery of unbiased distributions from biased molecular dynamics simulations.
• To overcome limitations of grid-based reweighting methods, especially for high-dimensional biases.

Main Methods:

• An iterative unbiasing scheme utilizing all trajectory data.
• A novel approach that avoids grid-based distribution evaluation.
• Benchmarking against existing reweighting schemes on model systems.

Main Results:

• The proposed iterative unbiasing scheme efficiently recovers unbiased distributions.
• The method is effective even with high-dimensional bias potentials.
• Demonstrated performance across model systems of varying complexity and dimensionality.

Conclusions:

• The iterative unbiasing scheme offers a robust and efficient solution for analyzing biased molecular dynamics simulations.
• This method significantly enhances the study of rare events in atomistic modeling.
• The approach is versatile and applicable to complex systems and high-dimensional biases.