A rare event sampling method for diffusion Monte Carlo using smart darting.

Area of Science:

• Computational Chemistry
• Statistical Mechanics
• Rare Event Sampling

Background:

• Classical parallel tempering and diffusion Monte Carlo (DMC) methods struggle with complex potential energy surfaces.
• Existing methods face inefficiencies in achieving convergence for stochastic simulations of rare events.

Purpose of the Study:

• To develop and validate new methods for enhancing the efficiency of diffusion Monte Carlo algorithms.
• To introduce a robust testbed, the decoupled Double Wells [(DDW)(n)] model, for evaluating rare event sampling techniques.

Main Methods:

• Mathematical modeling using linear combinations of decoupled Double Wells [(DDW)(n)].
• Deterministic calculation of thermodynamics and quantum dynamics for the (DDW)(n) model.
• Implementation and testing of the 'smart darting' method within classical and guided diffusion Monte Carlo frameworks.

Main Results:

• The (DDW)(n) model effectively challenges existing rare event sampling methods.
• Smart darting successfully reduces quasiergodicity in (DDW)(n) simulations for n ≫ 100 with minimal computational moves.
• Incorporating smart darting into DMC algorithms drastically improves convergence rates.

Conclusions:

• Smart darting offers a significant enhancement to diffusion Monte Carlo algorithms.
• This advancement extends the applicability of DMC to a wider range of complex computational systems.
• The (DDW)(n) model serves as a valuable benchmark for future rare event sampling method development.