Combination of universal spaces for proper actions (Q897096)

Let \(G\) be a group and let \(\mathcal{F}\) be a family of subgroups that is closed under conjugation and passage to subgroups (standard examples are the family consisting just of the trivial group, the family of all finite subgroup groups of \(G\), and the family of all virtually cyclic subgroups of \(G\)). A universal space for actions with stabilizers in \(\mathcal{F}\) is a \(G\)-CW complex such that {\parindent=6mm \begin{itemize} \item[-] the fixed point set of every \(H \in \mathcal{F}\) is contractible, \item [-] every cell-stabilizer is in \(\mathcal{F}\). \end{itemize}} A group is of type \(\mathcal{F}\)-\(F\) if such a space exists where the \(G\)-action is cocompact. The classical case is the one where \(\mathcal{F}\) consists of just the trivial group. Then an action with stabilizers in \(\mathcal{F}\) is a free action, a universal space is a model for \(EG\), and being of type \(F\) means to admit a compact classifying space. In the present article the author is mostly concerned with the family of finite subgroups. The main result is the following: if there is a simplicial complex on which \(G\) acts cocompactly and rigidly (i.e. the stabilizer of a cell fixes it) such that {\parindent=6mm \begin{itemize} \item[-] the fixed point set of every \(H \in \mathcal{F}\) is contractible, \item [-] every cell-stabilizer \(K\) is of type \(\mathcal{F}_K\)-\(F\), \end{itemize}} then \(G\) is of type \(\mathcal{F}\)-\(F\). Here \(\mathcal{F}_K = \{H \cap K \mid H \in \mathcal{F}\}\). Theorem 1 of the paper states this only for the case where \(\mathcal{F}\) is the family of finite groups but it is implicitly claimed in the paper (and has been confirmed by the author upon enquiry) that the proof should work for general \(\mathcal{F}\). The assumptions that the complex is simplicial (rather than an arbitrary CW-complex) and that the action is rigid come from the fact that the author uses complexes of groups to prove his result. It should be possible in many cases to generalize by taking subdivisions. The author uses his construction to remove a technical assumption from a previous paper, showing that a group that acts acylindrically and cocompactly on a CAT(-1)-complex with quasi-convex stabilizers is hyperbolic. There is a variant of the above statement for type \(F_n\) which is explicitly formulated for general families in the paper, as a Corollary (to the proof) on page 806. A group is of type \(\mathcal{F}\)-\(F_n\) if there is a universal space for actions with stabilizers in \(\mathcal{F}\) whose \(n\)-skeleton is cocompact. The Corollary states that if there is simplicial complex with a rigid, cocompact \(G\)-action such that {\parindent=6mm \begin{itemize} \item[-] the fixed point set of every \(H \in \mathcal{F}\) is contractible, \item [-] the stabilizer \(K\) of a \(\sigma\)-cell is of type \(\mathcal{F}_K\)-\(F_{n - \dim \sigma}\), \end{itemize}} then \(G\) is of type \(\mathcal{F}\)-\(F_n\). In this statement one cannot just take subdivisions because the finiteness properties of stabilizers in a subdivision are not as needed. For free actions, both statements can be obtained using the Borel construction, see for example [\textit{R. Geoghegan}, Topological methods in group theory. Graduate Texts in Mathematics 243. New York, NY: Springer (2008; Zbl 1141.57001), Section 6]. Working with free actions is technically easier because (among other things) one can easily go back and forth between the universal space \(X\) and its quotient \(G \backslash X\). Results for arbitrary stabilizers have been proven in the past by Lück and his school (for \(G\)-CW complexes rather than simplicial complexes). Notably [\textit{W. Lück} and \textit{M. Weiermann}, Pure Appl. Math. Q. 8, No. 2, 497--555 (2012; Zbl 1258.55011), Theorem 5.1] gives the main result of the present paper under the additional assumption that all stabilizers have contractible fixed point sets (taking \(\mathcal{G}\) to be the family of stabilizers) and the variant for \(F_\infty\) (that is, if all stabilizers are of type \(\mathcal{F}_K\)-\(F_n\) for all \(n\) then the group is of type \(\mathcal{F}\)-\(F_n\) for all \(n\)). If \(X\) is a universal space for \(G\) that is not cocompact, the finiteness properties of \(G\) are determined by the \textit{essential connectivity properties} of any cocompact filtration. This is a famous criterion due to [\textit{K. S. Brown}, J. Pure Appl. Algebra 44, 45--75 (1987; Zbl 0613.20033)] who proved it (for free actions) with homological methods. A topological proof is sketched in [\textit{R. Geoghegan}, loc. cit., Section 7]. The homological part has been generalized to arbitrary families in [\textit{M. G. Fluch} and \textit{S. Witzel}, Homology Homotopy Appl. 15, No. 2, 153--162 (2013; Zbl 1297.20053)]. The reviewer is not aware of a reference that would treat the \(\mathcal{F}\)-\(F_2\) case (which is the only missing case by the Hurewicz Theorem, see [\textit{W. Lück} and \textit{M. Weiermann}, loc. cit., Theorem 0.1(b)]), but it should be possible to deduce it from the (implicit) proof of the Corollary mentioned above.