Baumslag-Solitar groups and some other groups of cohomological dimension two (Q1173776)

This paper concerns groups which are finitely generated of cohomological dimension two (fgcd2). Let \(\mathfrak X\) be the class of fgcd2-groups \(G\) having a subgroup \(H\simeq\mathbb{Z}\) such that \(H\cap H^ g\neq 1\) for all \(g\) in \(G\). Any non-abelian fgcd2-group with non-trivial centre lies in \(\mathfrak X\), for by a result of \textit{R. Bieri} [Queen Mary College Math. Notes (1976; Zbl 0357.20027), \S 8.3] the centre is infinite cyclic. Any Baumslag-Solitar group \(\langle x,y: (x^ m)^ y=x^ n\rangle\), \(m,n\neq 0\), lies in \(\mathfrak X\): take \(H=\langle x\rangle\) and recall a classic theorem of \textit{R. Lyndon} [Ann. Math., II. Ser. 52, 650-655 (1950; Zbl 0039.02302)] stating that this group is fgcd2. In particular any soluble fgcd2-group lies in \(\mathfrak X\), for by a result of \textit{D. Gildenhuys} [Math. Z. 166, 21-25 (1979; Zbl 0414.20032)] such a group is isomorphic to \(\langle x,y: x^ y=x^ n\rangle\), \(n\neq 0\), and so is of Baumslag-Solitar type. The author proves that a non-cyclic group lies in \(\mathfrak X\) if and only if it is the fundamental group of a finite graph of infinite cyclic groups. The lengthy proof uses a finiteness criterion of \textit{R. Strebel} [Math. Z. 151, 263-275 (1976; Zbl 0325.16023)], ideas of \textit{M. Dunwoody} on cutting up graphs [Combinatorica 2, 15-23 (1982; Zbl 0504.05035)], and finally a generalization to \((FP)_ 2\)-groups (``almost finitely presented'' groups) of a result of \textit{M. Bestvina} and \textit{M. Feighn} [Invent. Math. 103, No. 3, 449-469 (1991; Zbl 0724.20019)] on the complexity of simplicial group actions on trees. Among the corollaries to the main theorem are (a) a finitely generated subgroup of an \(\mathfrak X\)- group is finitely presented; (b) if \(G\) lies in \(\mathfrak X\) then \(G''\) is free; (c) a torsion-free finite extension of an \(\mathfrak X\)-group is again an \(\mathfrak X\)-group.