A Discrete Stochastic Formulation for Reversible Bimolecular Reactions via Diffusion Encounter

TL;DR

This paper introduces a discrete stochastic model for reversible bimolecular reactions via diffusion, unifying classical boundary conditions with simulation approaches and simplifying computational implementation.

Contribution

It develops a discrete stochastic framework that accurately models reversible reactions, clarifies the underlying stochastic process, and simplifies numerical discretization.

Findings

Recovers classical models and boundary conditions in the continuous limit.

Unifies back-reaction boundary condition with simulation methods.

Simplifies the complications in modeling reversible reactions.

Abstract

The classical models for irreversible diffusion-influenced reactions can be derived by introducing absorbing boundary conditions to over-damped continuous Brownian motion (BM) theory. As there is a clear corresponding stochastic process, the mathematical description takes both Kolmogorov forward equation for the evolution of the probability distribution function and the stochastic sample trajectories. This dual description is a fundamental characteristic of stochastic processes and allows simple particle based simulations to accurately match the expected statistical behavior. However, in the traditional theory using the back-reaction boundary condition to model reversible reactions with geminate recombinations, several subtleties arise: it is unclear what the underlying stochastic process is, which causes complications in producing accurate simulations; and it is non-trivial how to perform an appropriate discretization for numerical computations. In this work, we derive a discrete stochastic model that recovers the classical models and their boundary conditions in the continuous limit. In the case of reversible reactions, we recover the back-reaction boundary condition, unifying the back-reaction approach with those of current simulation packages. Furthermore, all the complications encountered in the continuous models become trivial in the discrete model. Our formulation brings to attention the question: With computations in mind, can we develop a discrete reaction kinetics model that is more fundamental than its continuous counterpart?