Accelerating Molecular Dynamics through Informed Resetting.

Area of Science:

• Computational Chemistry
• Statistical Mechanics
• Biophysics

Background:

• Molecular dynamics simulations are crucial for studying biological processes.
• Many important phenomena occur over long timescales, posing challenges for standard simulations.
• Stochastic resetting has shown promise in accelerating simulations with broad transition times.

Purpose of the Study:

• To develop and demonstrate an informed stochastic resetting procedure for enhanced molecular dynamics sampling.
• To improve upon standard stochastic resetting by incorporating information about the reaction progress.
• To accelerate simulations of complex systems like protein folding and crystal nucleation.

Main Methods:

• Implementing an informed stochastic resetting protocol in molecular dynamics and Metadynamics simulations.
• Utilizing a condition-based resetting strategy (e.g., based on reaction coordinate distance).
• Applying the method to a model system and the chignolin peptide in explicit water.

Main Results:

• Informed resetting significantly outperforms standard stochastic resetting in accelerating simulations.
• Achieved speedups of 2-3 orders of magnitude for direct transit time calculations using Metadynamics with informed resetting.
• Maintained relatively low error margins (∼35-70%) despite substantial acceleration.

Conclusions:

• Informed stochastic resetting is a powerful technique for enhancing molecular simulation sampling.
• This method extends the applicability of stochastic resetting to a wider range of complex molecular processes.
• The approach enables more efficient inference of kinetic observables from simulations.