On the length of cohomology spheres (Q2661722)

A general numerical cohomological index for compact Lie groups was developed by \textit{T. Bartsch} [Topological methods for variational problems with symmetries. Berlin: Springer-Verlag (1993; Zbl 0789.58001)]. Let \(G\) be a compact Lie group, \(\mathcal{A}\) a set of \(G\)-spaces, \(h^*\) a multiplicative equivariant cohomology theory, and \(I\subseteq h^*(\{pt\})\) an ideal. The \((\mathcal{A}, h^*, I)\)-length of a \(G\)-space \(X\) is defined to be the smallest integer \(k\geq 1\) such that there exist \(A_1, \ldots, A_k \in \mathcal{A}\) with the property that for any \(\alpha_i \in I \cap \ker\big (h^*(\{pt\}) \to h^*(A_i)\big)\), \(1 \leq i \leq k\), the cup product \[p_X^*(\alpha_1) \cdots p_X^*(\alpha_k) =0,\] where \(p_X \colon X \to \{pt\}\). If such a \(k\) does not exist, we say that the \((\mathcal{A}, h^*, I)\)-length of \(X\) is \(\infty\). Like various other index theories, the length is a powerful tool for studying equivariant maps between \(G\)-spaces. It has been used to prove various versions of Borsuk-Ulam and Bourgin-Yang type theorems and also in the study of critical points of functions with symmetries. \par Computing the length is a difficult problem, and as expected, depends on the group and the space under consideration. In the paper under review, the authors compute the length of \(G\)-spaces which are cohomology spheres, where \(G\) is a torus or a \(p\)-torus with \(p\) a prime. As a consequence, they obtain Borsuk-Ulam and Bourgin-Yang type theorems in this context. A sharper version of the Bourgin-Yang theorem for topological manifolds is also proved. The authors conclude by giving some upper and lower bounds for the length for the torus and \(p\)-torus.