The singularity obstruction for group splittings (Q5960417)

A group \(G\) is said to split over a subgroup \(H\) if it can be decomposed as a non-trivial amalgamated free product \(A*_HB\) or an HNN extension \(A*_H\). A first obstruction for such a splitting is the number of ends \(e(G,H)\) of the pair \((G,H)\) as studied by \textit{P. Scott} [J. Pure Appl. Algebra 11(1977), 179-198 (1978; Zbl 0368.20021)]. If \(G\) splits over \(H\) then \(e(G,H)\geq 2\) but the converse is not quite true. In the case \(e(G,H)\geq 2\) another obstruction to splitting \(G\) over \(H\) is studied in the present paper which the author calls the singularity obstruction \({\mathcal S}_A(G,H)\). This depends on the choice of a ``proper \(H\)-almost invariant'' subset \(A\) of \(G\), and if it vanishes for some choice of \(A\) then \(G\) splits over \(H\). ``Scott's original approach to the problem of producing a splitting from a pair \(H<G\) with \(e(G,H)\geq 2\) included the method of passing to a finite index subgroup \(G_1\) of \(G\) which contains the subgroup \(H\) and avoids all the elements of \({\mathcal S}_A(G,H)\)''. ``In this paper we take a somewhat different approach. Instead of trying to make the singularity obstruction empty we show that if \({\mathcal S}_A(G,H)\) is small in one of two technical senses then we obtain a splitting of \(G\) over a subgroup related to \(H\). This result is more closely related to Stallings' theorem (the characterization of groups with more than one end, or \(e(G,1)\geq 2\), as groups splitting over some finite subgroup) and the algebraic torus theorem, and potential applications include generalizations of those results''.