The 2D Gray-Scott system of equations: constructive proofs of existence of localized stationary patterns
Matthieu Cadiot, Dominic Blanco
TL;DR
The paper tackles the constructive existence of stationary localized solutions for autonomous systems of semi-linear PDEs, focusing on the planar Gray-Scott reaction-diffusion system in $\mathbb{R}^2$. It develops a general framework using a $D_4$-symmetric Hilbert space $H_{D_4}$ to host the problem, leveraging square-domain Fourier representations to obtain a square-supported approximate solution. A Newton-Kantorovich scheme is implemented by building an explicit approximate inverse $\mathbb{A}$ to $D\mathbb{F}(\mathbf{u}_0)$ and applying radii-polynomial–based contraction arguments to prove a local, unique solution near $\mathbf{u}_0$, with symmetry and localization guaranteed. The method yields four distinct, non-radial, $D_4$-symmetric stationary patterns in the 2D Gray-Scott system, with complete computer-assisted proofs and implementation details on GitHub, and it connects these localized states to unbounded branches of spatially periodic solutions as the period tends to infinity, extending prior scalar results to systems.
Abstract
In this article, we present a comprehensive framework for constructing smooth, localized solutions in systems of semi-linear partial differential equations, with a particular emphasis to the Gray-Scott model. Specifically, we construct a natural Hilbert space $\mathcal{H}$ for the study of systems of autonomous semi-linear PDEs, on which products and differential operators are well-defined. Then, given an approximate solution $\mathbf{u}_0$, we derive a Newton-Kantorovich approach based on the construction of an approximate inverse of the linearization around $\mathbf{u}_0$. In particular, we derive a condition under which we prove the existence of a unique solution in a neighborhood of $\mathbf{u}_0$. Such a condition can be verified thanks to the explicit computation of different upper bounds, for which analytical details are presented. Furthermore, we provide an extra condition under which localized patterns are proven to be the limit of an unbounded branch of (spatially) periodic solutions as the period tends to infinity. We then demonstrate our approach by proving (constructively) the existence of four different localized patterns in the 2D Gray-Scott model. In addition, these solutions are proven to satisfy the $D_4$-symmetry. That is, the symmetry of the square. The algorithmic details to perform the computer-assisted proofs are available on GitHub.