Perturbations of normally hyperbolic manifolds with applications to the Navier-Stokes equations (Q5929060)

One of the major challenges arising in the study of nonlinear dynamics is the problem of trying to explain the robustness of chaotic dynamics. During the last several decades, a dynamical theory based on the study of certain hyperbolic structures within nonlinear dynamical systems, has proven to be a useful tool for addressing this problem. The oldest aspect of this theory goes back to the time of Lyapunov and Poincaré, and it involves the persistence of the saddle point property for a hyperbolic fixed-point under small perturbations. At a later time, a similar result was shown in the case of a hyperbolic periodic orbit. However, in these classical situations, one does not really encounter chaotic dynamics. Chaotic dynamics can appear when one tries to extend these earlier perturbations theories to the study of perturbations of normally hyperbolic, compact, invariant manifolds. The primary goal of the paper is to derive a theory of persistence for perturbations of normally hyperbolic, compact, invariant manifolds in a infinite-dimensional setting. The authors prove, that, under reasonable conditions, if \(M\) is a normally hyperbolic, compact, invariant manifold for a semiflow \(S_{0}\) generated by a given evolutionary equation \(\partial _{t}u+Au=F(u)\) on a Banach space \(W\), then for every small perturbation \(G\) of the given evolutionary equation, there is a homeomorphism \(h_{G}:M\to W\) such that \(M^{G}=h_{G}(M)\) is a normally hyperbolic, compact, invariant manifold for the perturbed semiflow \(S_{G}\), and that \(h_{G}\) converges to the identity mapping on \(M\), as \(G\) converges to \(0\). The proof closest to the theory presented in this paper can be found in the paper of \textit{P. W. Bates, K. Lu} and \textit{C. Zeng} [Mem. Am. Math. Soc. 645, 1-129 (1998; Zbl 1023.37013)] whereas the proof of the existence of the perturbed manifold \(M^{G}\) differs significantly from that of Bates, Lu and Zeng (1998). This offers several advantages: the proof is shorter, less complex, and, most importantly, the methodology developed therein may be more suitable for applications. In particular, the authors seek to build their theory by using the physical parameters of the problem. Of special interest in this work is the role that the theory of persistence of invariant manifolds plays in the context of the numerical analysis of the longtime dynamics of solutions to partial differential equations. So, the second objective of the paper is to develop a methodology which is rich enough to show that this theory can be easily applied to a wide range of applications, including the Navier-Stokes equations. It is noteworthy in this regard that, in order to be able to apply this theory in the analysis of numerical schemes used to study discretizations of partial differential equations, one needs to use a new measure of norm of the perturbation term \(G\) that arises in these schemes. In section 5, the authors examine two applications of the dynamical theory presented in the preceeding four sections. First, they develop a rigorous foundation for the dynamical features seen in the classical Couette-Taylor flows. The second application arises in the numerical analysis of fluid flows. The authors show, that numerical methods based on the Bubnov-Galerkin approximants can lead to good information on the longtime dynamics of the underlying problem. The paper is self-contained: In section 2, the authors examine a number of classical issues arising in the study of the dynamics of linear and nonlinear evolution equations. Under the assumption of \(A\) to be a sectorial operator and using fractional power spaces, the authors provide an existence theory for mild and strong solutions to linear and nonlinear evolutionary equations. In order to study linearized equations, they recall the notion of skew-product semiflows. In section 3, they study the hyperbolic structure and present the basic theory of exponential trichotomies on Banach spaces.