\(C^1\) density of stable ergodicity (Q2220468)

In this paper the \(C^1\)-version of the Pugh-Shub conjecture [\textit{C. Pugh} and \textit{M. Shub}, J. Complexity 13, No. 1, 125--179 (1997; Zbl 0883.58025)] is proven: among volume-preserving \(C^{1+}\) partially hyperbolic diffeomorphisms, ergodic ones form a \(C^1\) open and dense subset. The result comes from a recent breakthrough by the authors [Publ. Math., Inst. Hautes Étud. Sci. 124, 319--347 (2016; Zbl 1362.37017)] establishing genericity of nonuniform hyperbolicity of volume as well as an improvement of Hopf's argument for measurable homoclinic classes [\textit{F. Rodriguez Hertz} et al., Duke Math. J. 160, No. 3, 599--629 (2011; Zbl 1290.37011)] and the fact that accessible partially hyperbolic systems are \(C^1\) dense [\textit{D. Dolgopyat} and \textit{A. Wilkinson}, Astérisque 287, 33--60 (2003; Zbl 1213.37053)]. Here the authors need to implement this strategy to provide a very deep result about the creation of (super)-blenders. This can can be thought of as the main technical contribution of the paper (Theorem C, which in turn uses Theorem B providing a linearization procedure in the \(C^1\)-topology of horseshoes with large entropy). Even if the paper is a culmination of many works in the subject, it contains a pretty self-contained indication of the full proof in Sections 2--4, which are very nice to read. Then, Sections 5--7 and the appendix are devoted to the proofs of Theorems B and C. The authors express their indebtedness to a recent work of \textit{C. G. Moreira} and \textit{W. C. Silva} [``On the geometry of horseshoes in higher dimensions'', Preprint, \url{arXiv:1210.2623}] which proposes a different approach to understand blenders and leads the authors to introduce the new notion of superblenders and the proof of Theorem C. As for the need of the \(C^1\)-topology (besides the fact that they use generic results such as their previous result and the density of accessibility) the use of the \(C^1\)-topology is crucial in their proof of Theorem B. It allows them to work with affine iterated function systems. This simplifies the study considerably and it is an assumption in their Theorem C. However, the statement of Theorem C is done for the \(C^r\)-topology and some stronger form of Theorem C may be true in higher regularity (see in particular Theorem 3.2 of the paper).