On the integral cohomology of wreath products (Q1375980)

Spectral sequence theoretic methods are used in this paper to study the integral cohomology of the wreath product of \(G\) by \(\mathbb{Z}/p\) or \(S_p\). Its flavour is provided by the description in section six of the kernel of the restriction map to the sum of the invariant part of the cohomology of \(G^p\) and that of \(G\times \mathbb{Z}/p\). This kernel is trivial with \(\text{mod } p\) coefficients, but with integral coefficients and \(p\) equal to an odd prime, each cyclic summand of \(H^i(G,\mathbb{Z})\) of order divisible by \(p\) gives rise to one cyclic summand of the kernel in each degree \(pi,pi-2,\dots, p(i-1)+3\), and these summands exhaust the kernel. Surprisingly, the situation is more straightforward for wreath products with \(S_p\). The author is motivated by the problem of finding an exponent for the integral cohomology of a \(p\)-group, well-known to be in general larger than for the group itself. Waving hands one makes the comparison with representation theory, for which corestriction up from proper subgroups -- classically closely related to the wreath product construction -- suffices to construct all the irreducible representations of dimension greater than one. Cohomologically it is possible to show that if \(G\) does not provide a counterexample to a certain conjectured value then neither does the wreath product of \(G\) with \(\mathbb{Z}/p\). However an upper bound for the (eventual) exponent, based on generalized Frattini subgroups, is not the best possible (Section 9). Given the importance of wreath products to both topologists and group theorists, this paper is to be strongly recommended. Its methods are likely to find application for a long time to come.