Turing patterns in a network-reduced FitzHugh-Nagumo model

Abstract

Reduction of a two-component FitzHugh-Nagumo model to a single-component model with long-range connection is considered on general networks. The reduced model describes a single chemical species reacting on the nodes and diffusing across the links of a multigraph with weighted long-range connections that naturally emerge from the adiabatic elimination, which defines a new class of networked {dynamical} systems with local and nonlocal Laplace matrices. We study the conditions for the instability of homogeneous states in the original and reduced models and show that Turing patterns can emerge in both models.