Turing instabilities and patterns near a Hopf bifurcation (Q556053)

A two-component nonlinear reaction-diffusion system (S) is considered under vanishing Neumann boundary conditions on a \(k\)-dimensional cube. The reduced system (T) obtained from (S) by neglecting the diffusion term is assumed to have an equilibrium \(E\). The linearization of (S) at \(E\) is solved by separation and superposition with time-dependent coefficients such that the boundary conditions are satisfied. The coefficients are determined by infinitely many two-dimensional linear homogeneous first order differential sytems. By definition, system (S) has a Turing instability or a diffusion-driven instability, if the maximum spectral abscissa of these homogeneous systems belongs to a real positive eigenvalue. On the other hand, a non-constant time-independent solution of (S) is called a Turing pattern. The author establishes necessary and sufficient conditions for (S) to have a Turing instability. The results are applied to the Brusselator model and the Ginzburg-Landau model, whose corresponding vector fields (T) admit Hopf bifurcation. Here, bifurcation diagrams are obtained that separate regions with oscillatory solutions from regions with asymptotically stable patterns. It turns out that Turing patterns and oscillatory behavior can coexist and, moreover, that Turing instability is neither necessary nor sufficient for the existence of Turing patterns.