Optimisation of Simulations of Stochastic Processes by Removal of Opposing Reactions

TL;DR

This paper introduces a simple, efficient algorithm that speeds up stochastic process simulations by replacing reversible reactions with net reactions, easily integrating with existing methods without complex implementation.

Contribution

The authors present a novel algorithm that modifies transition rates to remove opposing reactions, simplifying and accelerating simulations of stochastic models.

Findings

Significant speed-up in simulation times for reaction-diffusion systems.

Easy implementation compatible with existing algorithms.

Maintains essential features of original models during acceleration.

Abstract

Models invoking the chemical master equation are used in many areas of science, and, hence, their simulation is of interest to many researchers. The complexity of the problems at hand often requires considerable computational power, so a large number of algorithms have been developed to speed up simulations. However, a drawback of many of these algorithms is that their implementation is more complicated than, for instance, the Gillespie algorithm, which is widely used to simulate the chemical master equation, and can be implemented with a few lines of code. Here, we present an algorithm which does not modify the way in which the master equation is solved, but instead modifies the transition rates, and can thus be implemented with a few lines of code. It works for all models in which reversible reactions occur by replacing such reversible reactions with effective net reactions. Examples of such systems include reaction-diffusion systems, in which diffusion is modelled by a random walk. The random movement of particles between neighbouring sites is then replaced with a net random flux. Furthermore, as we modify the transition rates of the model, rather than its implementation on a computer, our method can be combined with existing algorithms that were designed to speed up simulations of the stochastic master equation. By focusing on some specific models, we show how our algorithm can significantly speed up model simulations while maintaining essential features of the original model.