The Bredon cohomology of subgroup complexes (Q555976)

This paper is concerned with the homological algebra of coefficient systems on a class of subgroups of a group \(G\). It is partly structured around the investigation of three sequences associated to some class \({\mathcal X}\) of subgroups of \(G\), in particular their cohomology. The three sequences of particular interest are: \[ \begin{alignedat}{2} L(G) \to &\prod_{\sigma\in ch_0({\mathcal X})/G} L(N_G(\sigma)) &\;\to &\prod_{\sigma\in ch_1({\mathcal X})/G} L(N_G(\sigma)) \to\cdots\\ L(G) \to &\prod_{\sigma\in ch_0({\mathcal X})/G} L(\sigma_b)^{N_G(\sigma)} &\;\to &\prod_{\sigma\in ch_1({\mathcal X})/G} L(\sigma_b)^{N_G(\sigma)}\to\cdots\\ L(G) \to &\prod_{\sigma\in ch_0({\mathcal X})/G} L(C_G(\sigma_t))^{N_G(\sigma)} &\;\to &\prod_{\sigma\in ch_1({\mathcal X})/G} L(C_G(\sigma_t))^{N_G(\sigma)}\to\cdots. \end{alignedat} \] Here \(ch_n({\mathcal X})\) denotes the set of chains in \({\mathcal X}\) (without repetition) of length \(n+ 1\). The smallest element of chain \(\sigma\) is denoted by \(\sigma_b\) and the largest by \(\sigma_t\). The first of these sequences was investigated by \textit{P. J. Webb} [Comment. Math. Helv. 66, 34--69 (1991; Zbl 0722.57019)] when \(L\) is a Mackey functor. The second sequence first appeared in work of \textit{S. Bouc} [Proc. Symp. Pure Math. 63, 31--83 (1998; Zbl 0897.19001)], again for Mackey functors. An infinite version of the third sequence is implied by results of \textit{S. Jackowski} and \textit{J. E. McClure} [Topology 31, 113--132 (1992; Zbl 0754.55014)]. Later \textit{W. G. Dwyer} [Topology 36, 783--804 (1997; Zbl 0872.55014)] had infinite versions of the second and third sequences. Finally, the version of the third sequence (and also the second) appeared in work of \textit{J. Grodal} [Ann. Math. (2) 155, 405--457 (2002; Zbl 1004.55008)], and of \textit{R. Villarroel-Flores} and \textit{P. Webb} [Topology 41, 483--494 (2002; Zbl 1002.55013)]. The aim of the this paper is to present a unified treatment of all the results entirely within the homological algebra of coefficient systems. The strategy of the paper is to show that each of these sequences is representable as the complex of homomorphisms from some complex of coefficient systems \(\widetilde C_\bullet\) to \(L\), in other words that the sequence is of the form \(\Hom(\widetilde C_\bullet,L)\). If then turns out that \(\widetilde C_\bullet\) is homotopy equivalent to a projective resolution of the trivial coefficient system \(\overline R\). Thus the cohomology of the sequence is \(\text{Ext}^*(\overline R,L)\), which is the definition of the higher limits of \(L\). When these sequences are not exact then their cohomology is usually equal to the higher limits of \(L\). As the author points out, the article is inspirated by the paper of \textit{J. Grodal} [op. cit.], the proof consisting mainily in reformulations in algebraic terms of the geometric proofs given there. However the paper is very consistent, containing beside these important interpretations a series of concise interpretations in terms of higher limits of some statements in group theory as: Robinson's reformulation of Alperin's weight conjecture and Quillen's conjecture on the contractibility of subgroups complexes. The techniques for coefficient systems and the use of adjoint functors of several forgetful functors that the author applies are also very interesting.