A connection between cellularization for groups and spaces via two-complexes (Q2481806)

Let \(M\) be a two-dimensional Moore space (so \(H_2(M,\mathbb Z)=0\)), with fundamental group \(G\). The \(M\)-cellular spaces are those one can build from \(M\) using wedges, push-outs, and telescopes (and hence all pointed homotopy colimits). Denote by \(J\) the set of primes \(p\) for which \(G_{ab}\) is uniquely \(p\)-divisible, and define \(R=\mathbb Z_{(J)}\), the integers localized at \(J\), if \(G_{ab}\) is torsion, and \(R=\oplus_{p\in J}\mathbb Z/p\) otherwise. The purpose of the text is to characterize \(M\)-cellular spaces in terms of \(G\). For a group \(G\), the class of \(G\)-cellular groups is the smallest class of groups containing \(G\) and closed under colimits. Theorem. If \(G\) is either finite abelian or a subring of \(\mathbb Q\), then a space \(X\) is \(M\)-cellular if and only if \(\pi_1X\) is \(G\)-cellular and \(X\) is \(H\, R\)-acyclic. In particular a simply connected space \(X\) is \(M\)-cellular if and only if \(X\) is \(H\,R\)-acyclic. In general it is not possible to characterize the class of \(M\)-cellular spaces by means of algebraic properties derived from the group \(G\) and homological information, and the authors are forced to study a certain universal extension. Consider for instance the group \(G=\mathbb Z[1/p]*\mathbb Z/p\), and \(X=K(\mathbb Z, 2)\). Then, as shown by the authors, \(K(\mathbb Z,2)\) is \(H\, R\)-acyclic but not \(M\)-cellular.