Positively curved manifolds with symmetry (Q2500437)

Only little is known about which simply connected compact manifolds admit Riemannian metrics with positive sectional curvature \(K > 0\). There are only few examples (in dimension \(> 24\) only the compact rank 1 symmetric spaces) and very few known general obstructions (Gromov's Betti number theorem for manifolds with \(K \geq 0\)). The paper under review studies compact simply connected manifolds with \(K > 0\) and large isometry groups. This subject came to life with \textit{M. Berger}'s paper [Ann. Sc. Norm. Super. Pisa 15, 179--246 (1961; Zbl 0101.14201)], which was the starting point for the classification of the homogeneous spaces of positive sectional curvature by \textit{N. R. Wallach} [Ann. Math. 96, 277--295 (1972; Zbl 0261.53033)] and \textit{L. Berard-Bergery} [J. Math. Pur. Appl. 55, 47--68 (1976; Zbl 0289.53037)]. This classification revealed examples in dimensions 6, 7, 12, 13, and 24 (these were later supplemented by the inhomogeneous biquotients of \textit{J.-H. Eschenburg} [Invent. Math. 66, 469--480 (1982; Zbl 0484.53031)] and \textit{Ya. V. Bazaikin} [Sib. Math. J. 37, 1068--1085 (1996; Zbl 0874.53034)]). The natural question is now which new examples occur when the symmetry assumptions are reduced. \textit{K. Grove} suggested to do this reduction in various ways. Together with \textit{C. Searle}, he classified manifolds with \(K > 0\) and maximal symmetry rank [J. Pure Appl. Algebra 91, 137--142 (1994; Zbl 0793.53040)] and fixed point homogeneous manifolds with \(K>0\) [J. Differ. Geom. 47, 530--559 (1997; Zbl 0929.53017)]. Among the several subsequent results in this spirit, we only mention \textit{L. Verdiani}'s classification of even dimensional cohomogeneity one manifolds with \(K > 0\) [J. Differ. Geom. 68, 31--72 (2004; Zbl 1100.53033)]. All these classifications did not uncover new examples so far. The present paper and its dizygotic twin [\textit{B. Wilking}, Acta Math. 191, 259--297 (2003; Zbl 1062.53029)] make statements about such classifications in general. For example, it is not possible to find new manifolds with \(K > 0\) and cohomogeneity \(\leq k\) for fixed \(k\) in arbitrary high dimensions. More precisely, the main results derived in this paper are the following: If the dimension of the isometry group of a compact \(n\)-dimensional manifold \(M\) with \(K > 0\) is \(\geq 2n-6\) then \(M\) is tangentially homotopically equivalent to a rank one symmetric space or to a homogeneous space with \(K > 0\). If the rank of the isometry group is \(>3(\text{cohom}(M)+1)\) then \(M\) is homogeneous or tangentially homotopically equivalent to a rank one symmetric space. If \(\text{cohom}(M) = k > 1\) and \(\text{dim}(M) > 18(k+1)^2\) then \(M\) is tangentially homotopically equivalent to a rank one symmetric space. At least as important as the results are the new tools developed in this paper. Like in its above mentioned twin a central role is played by the connectedness lemma of the author: A totally geodesic embedded submanifold \(N^{n-h} \to M^n\) is \((n-2h+1)\)-connected if \(M\) has positive sectional curvature. This and the several other tools have already found applications in a forthcoming paper by Grove, Wilking, and Ziller where the three authors classify the odd dimensional cohomogeneity one manifolds with \(K > 0\) up to a short list of \(7\)-dimensional promising candidates.