Generic hyperbolicity of aubry sets on surfaces (Q2342179)

The authors answer positively, for dimension 2, the following question posed by \textit{M. Herman} [Doc. Math., J. DMV, 797--808 (1998; Zbl 0910.58036)]: Is it true that the Aubry set is hyperbolic? Recall that the Aubry set is a compact subset, invariant by the flow of a Tonelli Hamiltonian (fiberwise convex and superlinear) \(H:T^{*}M \rightarrow \mathbb{R}\), \(M\) a closed manifold. The Aubry set contains the Mather minimizing measures and it is a graph over a compact subset of \(M\). Genericity means the existence of an open and dense subset \(\mathcal{U} \subset C^{2}(M)\) such that for every potential \(V \in \mathcal{U}\), the Aubry set associated to the Hamiltonian \(H+V\) is hyperbolic. This concept of genericity was introduced by \textit{R. Mañé} [Nonlinearity 9, No. 2, 273--310 (1996; Zbl 0886.58037)], who also conjectured that generically the Aubry set is either an equilibrium point or a periodic orbit. Related to this conjecture, the following is also proven in this paper: for every subset \(U \subset T^{*}M\) containing the Aubry set and for every neighborhood \(\mathcal{V} \subset C^{2}(M)\) of \(0\), there exist \(\theta \in U\) and a potential \(V \in \mathcal{V} \) such that the orbit of the Hamiltonian flow of \(H+V\) is periodic and hyperbolic. The proof of the main result uses Green bundles along minimizing orbits, which in the two-dimensional case, either are transverse (leading to hyperbolicity) or coincide. In this last case, the crucial step is to prove a closing lemma for minimal (every orbit is dense) Aubry sets. This is done by refining the techniques developed in [the second and the third author, Commun. Pure Appl. Math. 68, No. 2, 210--285 (2015; Zbl 1321.37061); ibid. 68, No. 3, 345--412 (2015; Zbl 1377.37090)].