On cyclicity-one elliptic islands of the standard map (Q371052)

The author studies the one-parameter standard family of area-preserving diffeomorphisms in the torus \(\mathbb{T}^2\), given by \[ f_{\kappa} : (x,y) \mapsto\biggl(y, -x + 2y + \kappa \psi(y)\biggr) \mod \mathbb{Z}^2, \] where the parameter \( \kappa \) is a real number and \(\psi\) is a differentiable map on the circle with critical points. He focuses the research on large values of the parameter \(\kappa\), for which the map \(f_{\kappa}\) is non-integrable, has horseshoes, hence positive topological entropy, but still exhibits elliptic islands. He considers a \textit{critical set} \(C(\kappa) \subset \mathbb{T}^2\), which is given by thin vertical strips containing the critical points of \(\psi\). From the dynamical point of view, any orbit of \(f_{\kappa}\) that does not intersect \(C(\kappa)\) is hyperbolic. For non-hyperbolic periodic orbits of period \(N\), the \textit{cyclicity} is the number of different points of the periodic orbit in the critical set. The main theorem states: (a) For almost all large values of \(\kappa\), the map \(f_{\kappa}\) has only finitely many elliptic islands of cyclicity-one. (b) The set of values of \(\kappa\) for which \(f_{\kappa}\) exhibits infinitely many elliptic islands of cyclicity one (that accumulate on a locally maximal hyperbolic set) is dense in an interval of sufficiently large values, and besides it has Hausdorff dimension larger than \(1/4\). The proof of the main theorem is constructive showing where in the phase space, and also in the parameter space, the elliptic islands of cyclicity one appear. In spite of the length and denseness of lemmata, and of the necessarily complicated notation, the proofs are clear and very well explained. To give an idea of the constructive method that is used in the proofs, let us describe the route to arrive to part (a) of the main theorem. It has three parts: the first one is for fixed \(\kappa\), the construction of a dynamical structure through which the periodic orbits of cyclicity one can be characterized. The second part is the study of how that dynamical structure changes when the parameter \(\kappa\) changes. The third and last part is the conclusion of part (a) of the main theorem. The first part consists of three sub-parts. First, for a fixed (large enough) value of \(\kappa\), the author defines a dynamically adapted covering of the torus \(\mathbb{T}^2\), which is similar to a Markov partition. Roughly speaking this covering is composed by hyperbolic rectangles and critical pieces. Second, he studies the admissible itineraries with respect to the Markov-like covering, and uses the symbolic dynamics to encode different types of orbits, according to the appearance of certain symbols or words in their itineraries. Third, he defines \textit{the critical domains} (Definition 3.15) and proves that periodic points of period \(N\) and cyclicity one belong to some critical domain (Corollary 3.18). The second part also can be divided into three sub-parts. First, the author proves that the admissibility condition is monotone with respect to the parameter. Namely, if \(\kappa_1 \leq \kappa_2\), then the set of admissible words in the symbolic dynamics of \(f_{\kappa_1}\) is contained in the set of admissible words for \(f_{\kappa_2}\) (Lemma 3.20 and Corollary 3.21). Second, he cuts the parameter line into intervals \({\mathcal{K}}_{\omega}\), such that for all \(\kappa \in {\mathcal{K}}_{\omega}\), the word \(\omega\) persists being admissible for the symbolic dynamics of \(f_{\kappa}\). Then, he constructs, inside the interval \({\mathcal{K}}_{\omega}\) the subset \({\mathcal{E}}_1(N, \omega)\) such that for all \(\kappa \in {\mathcal{E}}_1 (N, \omega)\) the map \(f_{\kappa}\) has a cyclicity-one elliptic periodic point of period at least \(N\). Third, the author proves (up to rough simplifications in my description) that the Lebesgue measure of the set \({\mathcal{E}}_1 (N, \omega) \subset \mathbb{R}\) is smaller than \(K e^{-4N}\) for some constant \(K\) (Proposition 4.2). Finally, in the third part of the proof the author considers the following argument (up to simplifications in my description): \[ \sum _{N \geq 1} \text{Leb}({\mathcal{E}}_1(N, \omega)) \leq K \sum_{N \geq 1} e^{-4N} < + \infty . \] So, from Borel-Cantelli Lemma, and after adding in the finite number of words \(\omega\), the author ends the proof of part a) of his main theorem.