Invariant manifolds in discrete and continuous dynamical systems (Q358877): Difference between revisions

As a matter of course, invariant manifolds are an indispensable and frequently used tool in the geometric theory of dynamical systems. Their history can be traced back to Hadamard (graph transform) and Perron. While the approach of Perron has a functional-analytical flavor, the graph transform is strongly based on geometric ideas and intuition. The present research monograph is not only an original and well-written account to invariant manifolds illustrating how useful and modern Hadamard's graph transform still is, but it also underlines applicability and provides a widely accessible reference for related existence and perturbation results. Indeed, the book is subdivided into three parts dealing with discrete dynamical systems, finite-dimensional ODEs and various applications with a focus on numerical dynamics. Starting point are flexible existence theorems for repulsive, attractive, resp.\ hyperbolic, invariant manifolds of Lipschitzian mappings defined on products of Banach spaces. Subsequently their behavior under perturbation (including often useful explicit estimates) and their smoothness based on the fiber-contraction principle is shown. Results on invariant foliations, in order to establish an asymptotic phase property, as well as a smoothness proof conclude the first part. By means of a time-\(T\) mapping the above results are transferred to the continuous time situation of autonomous ordinary differential equations; this approach based on the previous discrete time results clearly extends to more general semiflows on infinite-dimensional state spaces. The remaining half of the book is devoted to applications within the field of dynamical systems. The authors begin with the classical Hadamard-Perron theorem on the existence of stable/unstable manifolds of a hyperbolic fixed point; strongly stable manifolds (now in an ODE setting) are considered as well. One finds the interesting result (originally due to Kirchgraber) that for every linear multistep method there exists an asymptotically equivalent one-step methods for the numerical solution of ODEs. The ``slow'' invariant manifolds of singularly perturbed differential equations are constructed, both in an attractive, as well as a hyperbolic framework. Thereafter, (stiff) Runge-Kutta methods applied to singularly perturbed problems are investigated. Results on invariant curves of perturbed harmonic oscillators, blow-up in singular perturbations and Runge-Kutta methods in differential-algebraic problems conclude the presentation. The authors' intention is to provide easily applicable and yet largely quantitative results -- they succeed in both respects. The book will be a helpful tool for researchers in the field and moreover, the third part on applications might serve as basis for an advanced seminar.