There are no \(C^{1}\)-stable intersections of regular Cantor sets (Q642120)

A \(C^s\)-regular Cantor set is the maximal invariant set of an expanding piecewise \(C^s\) map of the unit interval. A classical result of \textit{S. E. Newhouse} [Publ. Math., Inst. Hautes Étud. Sci. 50, 101--152 (1979; Zbl 0445.58022)] states that, in the \(C^2\) topology, there are persistent non-hyperbolic surface diffeomorphisms. His proof relies on an analysis of the intersection of two \(C^2\)-regular Cantor sets that appear in the unfolding of homoclinic tangencies. These sets have the bounded distortion property, meaning small parts of it resemble the Cantor set itself. In his previous paper [Ann. Inst. Henri Poincaré, Anal. Non Linéaire 13, No. 6, 747--781 (1996; Zbl 0865.58035)], the author defined the notion of stable intersection of Cantor sets and extended Newhouse's results. Later on, together with \textit{J.-C. Yoccoz} [Ann. Math. (2) 154, No. 1, 45--96 (2001; Zbl 1195.37015); corrigendum ibid. 154, No. 2, 527 (2001)] he proved that, indeed, on the space where the sum of the Hausdorff dimensions is greater than 1, the set of pairs of \(C^2\)-regular Cantor sets with \(C^2\)-stable intersection is open and dense. In the paper under review, the author shows that, contrary to these classical results, the set of pairs of \(C^1\)-regular Cantor sets without \(C^1\)-stable intersection is open and dense. Since intersections of Cantor sets are the main known obstructions for hyperbolicity of surface diffeomorphisms, this result is an indication that Smale's conjecture on \(C^1\)-denseness of hyperbolic surface diffeomorphisms might be true. The main property exploited by the author is the lack of bounded distortion for general \(C^1\)-regular Cantor sets. The proof is as follows. Step 1: by a dimension argument, for large \(r\), typical intersection of \(r\) regular Cantor sets is empty. Step 2: on advanced steps of the construction of a regular Cantor set, small \(C^1\) perturbations can generate regular Cantor sets with distorted geometry. Step 3: by backwards induction, if a typical intersection of \(r\) regular Cantor sets is empty, Step 2 implies that also a typical intersection of \(r-1\) regular Cantor sets is empty.