MCAMC: An Advanced Algorithm for Kinetic Monte Carlo Simulations: from Magnetization Switching to Protein Folding

TL;DR

The paper introduces MCAMC, an efficient algorithm for long kinetic Monte Carlo simulations that significantly accelerates computations without altering system dynamics, especially useful for low-temperature models with discrete states.

Contribution

It presents the MCAMC algorithm, a novel method that greatly speeds up kinetic Monte Carlo simulations for systems with discrete states, applicable to fields like protein folding and nanoscale magnet dynamics.

Findings

MCAMC speeds up simulations by many orders of magnitude.

Simple MCAMC can be further accelerated with more complex implementations.

Applicable to models resembling protein folding and nanoscale magnetic domain motion.

Abstract

We present the Monte Carlo with Absorbing Markov Chains (MCAMC) method for extremely long kinetic Monte Carlo simulations. The MCAMC algorithm does not modify the system dynamics. It is extremely useful for models with discrete state spaces when low-temperature simulations are desired. To illustrate the strengths and limitations of this algorithm we introduce a simple model involving random walkers on an energy landscape. This simple model has some of the characteristics of protein folding and could also be experimentally realizable in domain motion in nanoscale magnets. We find that even the simplest MCAMC algorithm can speed up calculations by many orders of magnitude. More complicated MCAMC simulations can gain further increases in speed by orders of magnitude.