Cohomological rigidity and the number of homeomorphism types for small covers over prisms (Q633031): Difference between revisions

\textit{M. W. Davis} and \textit{T. Januszkiewicz} [Duke Math. J.~62, No.~2, 417--451 (1991; Zbl 0733.52006)] introduced small covers as a real analogue of toric manifolds. A small cover is an \(n\)-manifold with an action of \((\mathbb{Z}/2)^n\) such that the quotient is a convex polytope. A small cover can be reconstructed from its quotient polytope \(P\) together with the characteristic function \(\lambda\) that assigns to each facet its associated stabilizer. The equivariant classification of small covers is well understood. Two small covers are equivariantly homeomorphic if and only if they have isomorphic equivariant mod 2 cohomology by a result of \textit{M. Masuda} [Adv. Math. 218, No.~6, 2005--2012 (2008; Zbl 1152.57032)], and the equivariant cohomology is given by the Stanley-Reisner face ring. In this paper the authors work towards a (non-equivariant) homeomorphism classification of manifolds that arise as small covers. They primarily address the case where the polytope \(P\) is the product of an interval with a regular \(m\)-gon for \(m>4\). The main result is that two small covers of such a prism are homeomorphic (as manifolds, disregarding the projection to the prism) if and only if their mod 2 cohomology rings are isomorphic. This was already known for \(m=4\), and it is known to be false for arbitrary polytopes. The first technical contribution of this paper involves constructing a new small cover from a given one by slicing the polytope \(P\) in half along a hyperplane and then gluing it back together with a twist by an automorphism of the cross-sectional polytope \(S\) and its induced characteristic function \(\lambda|_S\). The authors describe conditions on the automorphism under which the new small cover will be abstractly homeomorphic to the starting one. The second contribution is to introduce two invariants of the mod 2 (non-equivariant) cohomology ring of a small cover. By studying how these invariants behave under the above slicing and re-gluing construction, the authors are able to prove the main theorem and to compute the number of homeomorphism classes of small covers over an interval crossed with an \(m\)-gon.