Sampling rare event energy landscapes via birth-death augmented dynamics

Area of Science:

• Computational Physics
• Computational Chemistry
• Statistical Mechanics

Background:

• Complex systems simulations face the rare event problem, hindering sampling of energy landscapes due to high kinetic barriers.
• Enhanced sampling methods are crucial for efficiently exploring metastable states on typical simulation timescales.

Purpose of the Study:

• To expand on a novel birth-death sampling algorithm augmenting Langevin dynamics.
• To demonstrate efficient sampling of rare event energy landscapes, independent of barrier height.
• To address shortcomings in barrier region sampling and establish theoretical properties of the modified algorithm.

Main Methods:

• Augmenting overdamped Langevin dynamics with a birth-death process.
• Introducing an alternative approximation for the birth-death term to correct barrier region sampling.
• Mathematical analysis to establish theoretical properties and convergence results.
• Numerical simulations to investigate parameter effects and computational efficiency.

Main Results:

• The birth-death sampling scheme efficiently samples rare event energy landscapes.
• Equilibration speed is independent of the energy barrier height.
• The modified algorithm corrects for incorrect sampling in barrier regions.
• The scheme accelerates sampling for both overdamped and underdamped Langevin dynamics.

Conclusions:

• The modified birth-death sampling scheme is a promising method for rare event simulations in computational physics and chemistry.
• The algorithm offers efficient exploration of energy landscapes, overcoming limitations of standard simulation techniques.
• Further investigation into parameter optimization can reduce computational effort.