Polar manifolds and actions (Q1941580): Difference between revisions

An isometric action of a group \(G\) on a Riemannian manifold \(M\) is said to be ``polar'' if there exists an immersion \(\sigma:\Sigma\rightarrow M\), called a ``section'', which intersects all the orbits orthogonally. The stabilizer group \(\Pi\) (mod kernel) is called the ``polar group''; \(\Sigma\) is a totally geodesic immersed submanifold of \(M\) and \(\Pi\) is a discrete group of isometries acting properly discontinuously on \(\Sigma\) and \(M/G=\Sigma/\Pi\). The polar group contains a reflection group \(R\) which is a normal subgroup of \(\Pi\) with fundamental domain a chamber \(C\). The action is said to be a ``Coxeter polar action'' if there are no exceptional orbits and the stabilizer group \(\Pi_C\) is trivial. The authors show: Theorem A. A Coxeter polar action \((M,G)\) is determined by its polar data \((C,G)\). There is a canonical construction of a Coxeter polar \(G\) manifold from this data and any manifold with this polar data is equivariantly diffeomorphic to this example. The authors also show that any simply connected compact spherical polar or hyper polar manifold is rationally elliptic. Section 1 of the paper is devoted to preliminaries; Section 2 treats the polar group, reflection subgroups, and chambers; Section 3 deals with reconstruction for polar manifolds; Section 4 treats constructions, applications, and examples. The paper contains an extensive and very useful bibliography.