Valuations on free resolutions and higher geometric invariants of groups (Q1108380)

This is a continuation of work on what has become known as the ``Bieri- Strebel invariant''. This was first introduced for metabelian groups by \textit{R. Bieri} and \textit{R. Strebel} [in Proc. Lond. Math. Soc., III. Ser. 41, 439-464 (1980; Zbl 0448.20029)] and then extended to general groups by \textit{R. Bieri}, \textit{R. Strebel} and \textit{W. Neumann} [in Invent. Math. 90, 451-477 (1987; Zbl 0642.57002)]. This paper makes another large step forward in the theory. The geometric invariant of Bieri and Strebel now becomes the dimension zero term of a sequence of geometric invariants. More precisely, given a finitely generated group G and a G- module A, they define a series \(S^{n-1}\supseteq \Sigma\) \(0\supseteq \Sigma\) \(1\supseteq..\). of subsets of the sphere. Here n is the torsion- free rank of the abelianisation of G. The invariant \(\Sigma\) k gives complete information as to which subgroups U of G containing the derived group G' have the property that A is of type \(FP_ k\) over U. The term \(\Sigma\) 0 is the invariant of Bieri and Strebel whereas the invariant of Bieri, Strebel and Neumann is related to the \(\Sigma\) 1 term. The first main theorem of the paper shows that the \(\Sigma\)-sets are open subsets of the sphere; this requires an alternative characterization, of some considerable depth, of the \(\Sigma\)-subsets. Identify the (n-1)- sphere with equivalence classes of characters \(G\to {\mathbb{R}}\) where two such characters are equivalent if one is a positive real multiple of the other. If U is a subgroup of G containing G' define S(G,U) to be the subset of the sphere consisting of all characters of which the restriction to U is zero. Then the second main theorem of the paper states that a G-module A is of type \(FP_ k\) over U if and only if \(\Sigma\) k contains S(G,U). The proofs of these theorems involve a detailed technical discussion of alternative definitions of the invariants. The paper also contains a neat proof, using this discussion, of a description, due originally to \textit{K. Brown} [Trees, valuations and the Bieri-Strebel invariant'' (preprint)], of \(\Sigma\) k(G;\({\mathbb{Z}})\) in the case that G is a one relator group. It also shows that \(\Sigma\) \(k(G;{\mathbb{Z}})=\Sigma\) 1(G;\({\mathbb{Z}})\) for all \(k>1\); a result pointed out originally by Walter Neumann.