Elaborating Transition Interface Sampling Methods

TL;DR

This paper reviews advanced transition interface sampling methods for efficiently calculating rate constants in complex systems, introducing new algorithms to improve ergodicity and extend property calculations beyond rate constants.

Contribution

It introduces the transition interface sampling (TIS) method and its partial path version, along with new techniques to address ergodicity issues and compute additional properties from path ensembles.

Findings

TIS improves efficiency over standard TPS methods.

New algorithms alleviate ergodicity problems in path sampling.

Methods enable calculation of activation energies and reaction mechanisms.

Abstract

We review two recently developed efficient methods for calculating rate constants of processes dominated by rare events in high-dimensional complex systems. The first is transition interface sampling (TIS), based on the measurement of effective fluxes through hypersurfaces in phase space. TIS improves efficiency with respect to standard transition path sampling (TPS) rate constant techniques, because it allows a variable path length and is less sensitive to recrossings. The second method is the partial path version of TIS. Developed for diffusive processes, it exploits the loss of long time correlation. We discuss the relation between the new techniques and the standard reactive flux methods in detail. Path sampling algorithms can suffer from ergodicity problems, and we introduce several new techniques to alleviate these problems, notably path swapping, stochastic configurational bias Monte Carlo shooting moves and order-parameter free path sampling. In addition, we give algorithms to calculate other interesting properties from path ensembles besides rate constants, such as activation energies and reaction mechanisms.