Long-time shadow limit for reaction-diffusion-ODE systems

Abstract

Shadow systems are an approximation of reaction-diffusion-type problems obtained in the infinite diffusion coefficient limit. They allow reducing complexity of the system and hence facilitate its analysis. The quality of approximation can be considered in three time regimes: (i) short-time intervals taking account for the initial time layer, (ii) long-time intervals scaling with the diffusion coefficient and tending to infinity for diffusion tending to infinity, and (iii) asymptotic state for times up to $T = \infty$. In this paper we focus on uniform error estimates in the long-time case. Using linearization at a time-dependent shadow solution, we derive sufficient conditions for control of the errors. The employed methods are cut-off techniques and $L^p$-estimates combined with stability conditions for the linearized shadow system. Additionally, we show that the global-in-time extension of the uniform error estimates may fail without stronger assumptions on the model linearization. The approach is presented on example of reaction-diffusion equations coupled to ordinary differential equations (ODEs), including classical reaction-diffusion system. The results are illustrated by examples showing necessity and applicability of the established conditions.