Everything everywhere all at once: a probability-based enhanced sampling approach to rare events

Area of Science:

• Computational Chemistry
• Statistical Mechanics
• Molecular Dynamics

Background:

• Studying rare events in computer simulations is a significant challenge across various scientific domains.
• Previous work introduced a variational approach for computing the committor function to sample transition states.
• Efficiently characterizing rare events requires advanced simulation techniques.

Purpose of the Study:

• To significantly improve the computational efficiency and accuracy of rare event simulations.
• To develop a method that thoroughly samples free energy surfaces, including transition states and metastable basins.
• To provide a comprehensive characterization of rare events, including competing pathways and intermediates.

Main Methods:

• Integration of a metadynamics-like enhanced sampling approach with committor function calculation.
• Utilizing a logarithmic function of the committor as a collective variable for enhanced sampling.
• Iterative variational computation of the committor function for efficient sampling of the transition state ensemble.

Main Results:

• Accurate sampling of the free energy surface, enabling detailed study of transition states and metastable basins.
• Successful application to systems with competing reactive paths and metastable intermediates.
• Demonstration of obtaining physical insights from the optimized committor model and sampled data.

Conclusions:

• The enhanced sampling approach provides a powerful tool for the full characterization of rare events in computer simulations.
• This method offers a unified and thorough treatment of transition states, metastable basins, and competing reaction mechanisms.
• The approach yields valuable physical insights into the dynamics of rare events.