Fractional Gray-Scott Model: Well-posedness, Discretization, and Simulations

TL;DR

This paper introduces a fractional Laplacian into the Gray-Scott reaction-diffusion model, proving well-posedness, developing numerical schemes, and analyzing pattern formation differences caused by anomalous diffusion.

Contribution

It extends the classical Gray-Scott model by incorporating fractional diffusion, providing stability analysis, convergence results, and numerical verification of pattern differences.

Findings

Different patterns form at various fractional orders

Numerical schemes achieve second-order convergence

Steady pattern scaling law in terms of fractional order

Abstract

The Gray-Scott (GS) model represents the dynamics and steady state pattern formation in reaction-diffusion systems and has been extensively studied in the past. In this paper, we consider the effects of anomalous diffusion on pattern formation by introducing the fractional Laplacian into the GS model. First, we prove that the continuous solutions of the fractional GS model are unique. We then introduce the Crank-Nicolson (C-N) scheme for time discretization and weighted shifted Gr\"unwald difference operator for spatial discretization. We perform stability analysis for the time semi-discrete numerical scheme, and furthermore, we analyze numerically the errors with benchmark solutions that show second-order convergence both in time and space. We also employ the spectral collocation method in space and C-N scheme in time to solve the GS model in order to verify the accuracy of our numerical solutions. We observe the formation of different patterns at different values of the fractional order, which are quite different than the patterns of the corresponding integer-order GS model, and quantify them by using the radial distribution function (RDF). Finally, we discover the scaling law for steady patterns of the RDFs in terms of the fractional order $1<\alpha \leq 2 $.