Construction of invariant whiskered tori by a parameterization method. I: Maps and flows in finite dimensions (Q1016463): Difference between revisions

This fundamental paper provides a new powerful method for proving existence of hyperbolic invariant tori (also called whiskered tori) in finite-dimensional exact symplectic maps and flows. The method is based on the study of a functional equation expressing the existence of an invariant torus. Let \(K:\mathbb T^l\to \mathcal M\) be an embedding of the standard \(l\)-torus \(\mathbb T^l\) into a symplectic manifold \(\mathcal M\). For a symplectic map \(F\) from \({\mathcal M}\) into itself and a frequency vector \(\omega\in \mathbb R^l\), the embedding \(K\) satisfies the functional equation (*) \((F\circ K)(\theta)= K(\theta+\omega)\), \(\theta\in \mathbb T^l \) if and only if the range of \(K\) is an invariant torus of \(F\), and the induced map is \(\theta\mapsto\theta+\omega\). The main result of this paper shows that if an embedding \(K\) solves equation (*) up to a sufficiently small correction and satisfies some non-degeneracy assumptions (hyperbolicity and twist condition), then there is a true solution of \((*)\) close to \(K\). The exact solution corresponds to a whiskered torus and is unique up to a translation. The authors call the results based on validating the approximate solution ``a posteriori'' results. The proofs involve KAM-type iterative methods. However, in contrast to the usual KAM-schemes, the methods in this paper are not based on phase-space transformations, but rather on successive corrections. The theorems do not assume that the system be close to integrable nor that it is written in action-angle variables, nor that the hyperbolic bundles be trivial, nor that the linearized hyperbolic motion be reducible to constant coefficients. The paper is very well written. All the necessary definitions are given, and the proofs are presented in great detail. The authors provide various applications and consequences of the main results.