Short-Time Infrequent Metadynamics for Improved Kinetics Inference.

Area of Science:

• Computational Chemistry
• Statistical Mechanics
• Molecular Dynamics

Background:

• Infrequent Metadynamics is widely used for accelerated simulations of long time-scale processes.
• Current inference methods rely on rescaling first-passage times, but are limited to Poisson kinetics and require slow bias deposition and optimal collective variables.

Purpose of the Study:

• To develop an improved inference scheme for Infrequent Metadynamics that overcomes limitations of existing methods.
• To enhance the trade-off between simulation speedup and accuracy, particularly with suboptimal collective variables.

Main Methods:

• The proposed scheme leverages two key observations: time-independent rates can be estimated from short trajectories, and short trajectories experience minimal bias.
• The inference is based on short time scales, utilizing rescaled first-passage times from these trajectories.

Main Results:

• The improved scheme provides a better speedup-accuracy balance at no additional computational cost.
• It is particularly effective when using suboptimal collective variables.
• The method was successfully demonstrated on a model system and two molecular systems.

Conclusions:

• The novel inference scheme offers a more robust and efficient approach for analyzing accelerated simulations.
• This method expands the applicability of Infrequent Metadynamics, especially in scenarios with suboptimal collective variables or limited prior knowledge.