Structure theory of metric measure spaces with lower Ricci curvature bounds (Q2418888)

This paper studies the structures of metric measure spaces. There is a well-developed structure theory for Gromov-Hausdorff limits of Riemannian manifolds with lower Ricci curvature bounds, e.g., the work of \textit{J. Cheeger} and \textit{T. H. Colding} [Ann. Math. (2) 144, No. 1, 189--237 (1996; Zbl 0865.53037); J. Differ. Geom. 54, No. 1, 37--74 (2000; Zbl 1027.53043)], and references in them and \textit{T. H. Colding} and \textit{A. Naber} [Ann. Math. (2) 176, No. 2, 1173--1229 (2012; Zbl 1260.53067)], etc. In the last 10 years, a lot of progress has been made on general metric measure spaces satisfying generalised lower Ricci curvature bounds, which was pioneered by \textit{J. Lott} and \textit{C. Villani} [Ann. Math. (2) 169, No. 3, 903--991 (2009; Zbl 1178.53038)] and \textit{K.-T. Sturm} [Acta Math. 196, No. 1, 65--131 (2006; Zbl 1105.53035); Acta Math. 196, No. 1, 133--177 (2006; Zbl 1106.53032)], and has been modified considerably by \textit{K. Bacher} and \textit{K.-T. Sturm} [J. Funct. Anal. 259, No. 1, 28--56 (2010; Zbl 1196.53027)] and \textit{L. Ambrosio} et al. [Invent. Math. 195, No. 2, 289--391 (2014; Zbl 1312.53056); Duke Math. J. 163, No. 7, 1405--1490 (2014; Zbl 1304.35310)]. These generalised lower Ricci curvature bounds are compatible with the smooth Riemannian case and are stable with respect to measured Gromov-Hausdorff convergence. It is a great interest to have a structure theory on such metric measure spaces as in the case of smooth limits. The correct notion of lower Ricci curvature bound on a general metric measure space is the \((K,N)\)-curvature dimension condition, denoted as \(CD(K,N)\), which corresponds to the statement `X has \(N\)-dimensional Ricci curvature bounded from below by \(K\)' as in the work of \textit{D. Bakry} and \textit{M. Émery} [Lect. Notes Math. 1123, 177--206 (1985; Zbl 0561.60080)]. A variant is the notion of reduced curvature dimension bound, \(CD^*(K,N)\). \textit{L. Ambrosio} et al. [Duke Math. J. 163, No. 7, 1405--1490 (2014; Zbl 1304.35310)] introduced a more restrictive condition which rules out certain \(CD(K,N)\)-spaces never appearing as Gromov-Hausdorff limits in the Cheeger-Colding theory while retaining the stability properties under measured Gromov-Hausdorff convergence. The extra condition imposed is that the Sobolev space \(W^{1,2}(X)\) of functions on the metric measure space \(X\) is a Hilbert space, equivalently, the Laplace operator on \(X\) is linear. This Hilbertian condition is called the Riemannian Curvature Dimension (RCD) bound, and refinements of this have led to the notion of \(RCD^*(K,N)\)-spaces. This paper proves that a metric measure space \((X,d,m)\) satisfying the \(RCD^*(K,N)\)-condition with \(N>1\) is rectifiable, i.e., there exist countable \(m\)-measurable subsets, covering \(X\) up to an \(m\)-negligible set, such that each subset is bi-Lipschitz to a measurable subset of \(\mathbb{R}^{k_i}\) for some \(1\leq k_i\leq N\). A corollary is that for \(m\)-a.e. point, the tangent cone is unique and Euclidean of dimension at most \(N\). The proof is based on a maximal function argument combined with an original Almost Splitting Theorem via estimates on the gradient of the excess. The authors also show a sharp integral Abresch-Gromoll type inequality for the excess function and an Abresch-Gromoll-type inequality for the gradient of the excess. The argument is new even in the smooth setting.