Diffeomorphism stability and codimension three (Q2238542)

Given \(k\in \mathbb{R}\), \(n\in \mathbb{N}\), and \(\nu, D>0\), let \(\{M_{\alpha}\}_{\alpha=1}^{\infty}\) be a sequence of Riemannian \(n\)-manifolds with sectional curvature \(\geq k\), volume \(>\nu\), and diameter \(\leq D\) which converges in the Gromov-Hausdorff topology to an Alexandrov space \(X\). Perelman's Stability Theorem then implies that \(M_{\alpha}\) is homeomorphic to \(X\) for all large \(\alpha\). Motivated by Perelman's Stability Theorem the \emph{Diffeomorphism Stability Question} asks whether all but finitely many of \(M_{\alpha}\)'s are diffeomorphic. The present article addresses this question and provides an affirmative answer to it in a special case when all of the singularities of \(X\) occur along Riemannian manifolds of codimension \(\leq 3\) as elucidated below. To state the main result, the authors first define two notions: Let \(\mathcal{M}_{k, \nu, 0}^{\infty, \infty, D}(n)\) denote the class of closed Riemannian \(n\)-manifolds \(M\) with sectional curvature \(\geq k\), volume \(>\nu\), and diameter \(\leq D\). Then Definition A. A space \(X\in \text{closure}\, (\mathcal{M}_{k, \nu, 0}^{\infty, \infty, D}(n))\) is called diffeomorphically stable if for any sequence \(\{M_{\alpha}\}_{\alpha=1}^{\infty}\subset \mathcal{M}_{k, \nu, 0}^{\infty, \infty, D}(n)\) converging in the Gromov-Hausdorff topology to \(X\), all but finitely many of \(M_{\alpha}\)'s are diffeomorphic. Definition B. Let \(X\) be an Alexandrov space and \((S, g)\) a Riemannian manifold. Let \(d^{X}\) be the distance of \(X\) and \(d^{S} \) the distance on \(S\) induced by \(g\). An embedding \( \iota \colon (S, g)\to X\) is infinitesimally isometric if and only if for every \(\varepsilon> 0\) there is a \(\delta>0\) so that for distinct \(a,b \in S\) with \(d^{S} (a,b) < \delta,\) \begin{align*} \left | \dfrac{d^{S}(a, b)}{d^{X}(\iota(a), \iota(b))}-1\right |<\varepsilon. \end{align*} The main theorem of the article then reads as follows: Theorem C. There is a \(\delta(k,\nu,D,n) > 0 \) so that \(X\in \text{closure}\, (\mathcal{M}_{k, \nu, 0}^{\infty, \infty, D}(n))\) is diffeomorphically stable provided \(X\) contains a finite collection \(\mathcal{S}=\{S_{j}\}_{j\in J} \) of infinitesimally isometrically embedded, pairwise disjoint, Riemannian manifolds \(S_{j}\) without boundary that have the following properties: \begin{itemize} \item[1.] Every point of \(X\setminus\cup_{j\in J} S_{j}\) is \((n,\delta)\)-strained. \item[2.] No point of any \(S \in \mathcal{S}\) is \((\dim (S) + 1, \delta)\)-strained. \item[3.] \(S\) is the union of two subcollections \(\mathcal{K} \) and \(\mathcal{N}\) . \item[4.] Elements of \(\mathcal{K} \) are compact and have codimension \(\leq 3\). \item[5.] Elements of \(\mathcal{N}\) are not compact and have codimension \(\leq 2\). \item[6.] The closure of an element \(N\in \mathcal{N}\) is a union of elements of \(\mathcal{S}\). \end{itemize} As the authors point out in Corollary D, a special case when the condition of Theorem C is satisfied is when \(X\) is the quotient of an isometric group action on a Riemannian manifold provided that all its singular strata have codimension \(\leq 3\). By using Theorem C, the authors are able to reprove the previous results by the first author et al. [Commun. Anal. Geom. 25, No. 1, 243--267 (2017; Zbl 1380.53043); Adv. Geom. 17, No. 2, 231--245 (2017; Zbl 1387.53049)] and even improve Theorem 1 in [the first author, J. Geom. Anal. 26, No. 4, 2955--2983 (2016; Zbl 1354.53049)], where ``homeomorphic'' can be replaced with ``diffeomorphic''. Moreover, Theorem C generalizes Theorem 6.1 in [\textit{K. Kuwae} et al., Math. Z. 238, No. 2, 269--316 (2001; Zbl 1001.53017)], which is also utilized in some first steps of the proof of Theorem C.