Global convergence towards pushed travelling fronts for parabolic gradient systems

TL;DR

This paper proves that solutions of certain parabolic systems with coercive potentials converge globally to pushed travelling fronts under specific initial conditions, using variational methods without maximum principle.

Contribution

It introduces a variational approach to establish global convergence to pushed fronts for parabolic gradient systems, extending previous methods to systems lacking maximum principle.

Findings

Solutions invade the equilibrium at speeds greater than a threshold

Existence of pushed fronts is characterized by a variational criterion

A Poincaré inequality is used to analyze the variational landscape

Abstract

This article addresses the issue of global convergence towards pushed travelling fronts for solutions of parabolic systems of the form \[ u_t = - \nabla V(u) + u_{xx} \,, \] where the potential $V$ is coercive at infinity. It is proved that, if an initial condition $x\mapsto u(x,t=0)$ approaches, rapidly enough, a critical point $e$ of $V$ to the right end of space, and if, for some speed $c_0$ greater than the linear spreading speed associated with $e$, the energy of this initial condition in a frame travelling at the speed $c_0$ is negative $\unicode{x2013}$ with symbols, \[ \int_{\mathbb{R}} e^{c_0 x}\left(\frac{1}{2} u_x(x,0)^2 + V\bigl(u(x,0)\bigr)- V(e)\right)\, dx < 0 \,, \] then the corresponding solution invades $e$ at a speed $c$ greater than $c_0$, and approaches, around the leading edge and as time goes to $+\infty$, profiles of pushed fronts (in most cases a single one) travelling at the speed $c$. A necessary and sufficient condition for the existence of pushed fronts invading a critical point at a speed greater than its linear spreading speed follows as a corollary. In the absence of maximum principle, the arguments are purely variational. The key ingredient is a Poincar\'e inequality showing that, in frames travelling at speeds exceeding the linear spreading speed, the variational landscape does not differ much from the case where the invaded equilibrium $e$ is stable. The proof is notably inspired by ideas and techniques introduced by Th. Gallay and R. Joly, and subsequently used by C. Luo, in the setting of nonlinear damped wave equations.