Exploration vs Convergence Speed in Adaptive-Bias Enhanced Sampling.

Area of Science:

• Computational Chemistry and Physics
• Statistical Mechanics
• Materials Science

Background:

• Adaptive-bias enhanced sampling methods utilize bias potentials to facilitate transitions between metastable states.
• These methods modify the bias potential based on collective variables and the free energy surface.
• A key challenge is the exploration-convergence tradeoff when collective variables are suboptimal.

Purpose of the Study:

• To address limitations of existing methods where fast exploration is preferred over rapid convergence.
• To introduce a novel variant of On-the-fly Probability Enhanced Sampling (OPES) prioritizing efficient escape from metastable states.
• To evaluate the performance of this new OPES variant against established methods like metadynamics.

Main Methods:

• Development of a new OPES variant focused on accelerating transitions out of metastable states.
• The new method balances exploration efficiency with convergence speed.
• Application and testing of the method on prototypical systems.

Main Results:

• The new OPES variant demonstrates improved performance in systems where rapid exploration is crucial.
• It effectively escapes metastable states, albeit at a potentially slower convergence rate.
• Comparative studies show superior performance over the popular metadynamics method in specific scenarios.

Conclusions:

• The developed OPES variant offers a valuable alternative for enhanced sampling when fast exploration is paramount.
• It provides a flexible approach to manage the exploration-convergence tradeoff in complex systems.
• This method enhances the ability to accurately estimate free energy landscapes by facilitating phase space exploration.