Geometric stability analysis for periodic solutions of the Swift-Hohenberg equation (Q1379688)

The authors give a new proof of the nonlinear stability of certain space periodic equilibria of the Swift-Hohenberg equation \[ \partial_tu= (\varepsilon^2-(1+ \partial^2_x)^2)u- u^3.\tag{1} \] These equilibria are obtained from (1) via bifurcation arguments; they depend on \(\varepsilon\) and the wave number \(\omega\) and are denoted by \(u_{\varepsilon\omega}\). The proof that these solutions are asymptotically stable (in suitable function spaces) are based on a series of technical steps, too involved to be explained here. The authors rely on ideas of \textit{G. Schneider} who was first to prove the stability of the family \(u_{\varepsilon\omega}\) [Commun. Math. Phys. 178, 679-702 (1996; Zbl 0861.35107)]. The present proof differs in some essential points from that of G. Schneider. In fact, while the consideration take place in Bloch space, the authors avoid the use of renormalization methods. Instead, they apply a result of \textit{C. E. Wayne} [Arch. Ration. Mech. Anal. 138, 279-306 (1997; Zbl 0882.35061)] which states that under certain assumptions, center manifolds of PDE's on unbounded domains exist. These results are applied to a system of integro-differential equations, formulated in Bloch space, which follow by convolution from the equation \[ \partial_tv=- (1+\partial^2_x)^2v+ \varepsilon^2v- (u_{\varepsilon\omega})^2v- 3u_{\varepsilon\omega}v^2- v^3.\tag{2} \] Equation (2) is obtained from (1) by setting \(u_{\varepsilon\omega}+ v= u\). The authors' main result, Theorem 1.2, states that (2) (or rather its convolution counter-part in Boch space) admits a finite-dimensional center manifold which attracts all small solutions, and that these solutions tend to zero as \(t\to\infty\). The asymptotic form of these small solutions is described explicitely by the theorem. The center manifold, specifically its dimension, depends on an auxiliary Hilbert space \(H(n)\) which enters as a parameter.