A rotational hyperbolic theory for surface homeomorphisms
Pierre-Antoine Guihéneuf
TL;DR
This work builds a rotationally hyperbolic framework for surface homeomorphisms by translating rotational data from ergodic measures into hyperbolic-like rotational classes, then organizing these through forcing theory to produce a network of rotational horseshoes. It defines five heteroclinic-type relations, proves them equivalent, and encodes the global rotational structure in graphs $G$ and $\mathcal{T}$ to describe how rotation vectors assemble into $rot(f)$. The theory yields realizability and convex-structure results for rotation sets, and constructs invariant open sets tied to the class graph, laying the groundwork for a rotational Axiom A-type viewpoint with a companion applications paper for big rotation sets.
Abstract
We develop a rotational hyperbolic theory for surface homeomorphisms. We use the equivalence relation on ergodic measures that have nontrivial rotational behaviour defined in arXiv:2312.06249 to define a rotational counterpart of homoclinic classes. These allows to produce a network of horseshoes representing the whole rotational behaviour f the homeomorphism. We also study the counterpart of heteroclinic connections and give 5 different characterizations of such connections. The main technical tool is the forcing theory of Le Calvez and Tal arXiv:1503.09127, arXiv:1803.04557, and in particular a result of creation of periodic points that can also be seen as a statement of homotopically bounded deviations [GT25a]. This theoretical article is followed by a paper focused of some applications of it to the case of homeomorphisms with big rotation set [Gui25].
