On the cohomological dimension of kernels of maps to \(\mathbb{Z}\) (Q6866341)

A group \(G\) is called residually finite rationally solvable (RFRS) if there is a chain \(G=G_{0} \geq G_{1} \geq \cdots \) of\Nfinite-index subgroups such that \(\bigcap _{i\geq 0}G_{1}=\{1 \} \) and \(ker (G_{1} \longrightarrow G_{i}^{ab}\otimes \mathbb{Q})\leq G_{i+1}\) for all \(i>0\).\N\NA coherent group is a group where every finitely generated subgroup is finitely presented.\N\NA \(2\)-complex \(X\) has non-positive immersions if for every immersion \(Y \looparrowright X\) of a compact connected complex \(Y\), either \(\mathcal{X}(Y)\leq 0\) or \(\pi _{1}(Y)\)1 is trivial.\N\NIn [Indiana Univ. Math. J. 71, No. 1, 409--438 (2022; Zbl 1551.20086)] \textit{D. Wise} conjectured (Conjecture 2.6) that non-positive immersions should be equivalent to the vanishing of the second \(\ell ^{2}\)-Betti number. Therefore the author situate the following question:\N\N\textbf{Question} Let \(G\) be a finitely generated group of geometric dimension at most two. Are the following properties equivalent?\N\N(1) \(b_{2}^{(2)}(G) = 0\).\N\N(2) \(G\) is coherent.\N\N(3) \(G\) has a finite-index subgroup that is the fundamental group of a \(2\)-complex with non-positive immersions.\N\NHere the author focus on the class of RFRS groups and his main result is:\N\N\textbf{Theorem A.} Let \(G\) be a finitely generated RFRS group. Then \(G\) is virtually free-bycyclic if and only if \(b_{2}^{(2)}(G) = 0\) and \(cd _{\mathbb{Q}}(G) \leq 2\).\N\NIf \(G\) is free-by-cyclic, then \(G\) is coherent by a result in [ Ann. of Math. 149:3 (1999) 1061-1077 ], so being virtually free-by-cyclic is a strong form of coherence.\N\N\textbf{Corollary B.} Let \(G\) be a RFRS group with \( cd _{\mathbb{Q}}(G)\leq 2\),then\N\N(1) \(G\) is coherent,\N\N(2) the group algebra \(k[G]\) is a coherent ring for any field \(k\),\N\N(3) if \(G\) is finitely generated, then \(G\) is virtually the fundamental group of an aspherical \(2\)-complex with non-positive immersions.\N\NTherefore it is obtained the implications \((1) \Longrightarrow (2)\) and \((1) \Longrightarrow (3)\) of the Question in the class of RFRS groups.\N\NAs an application of Theorem A it is obtained sufficient conditions for a RFRS group to be virtually free-by-cyclic.\N\N\textbf{Corollary C.} Let \(G\) be a RFRS group and additionally suppose that one of the following holds:\N\N(1) \(G\) admits an elementary hierarchy [\textit{M. Hagen} and \textit{D. T. Wise}, Groups Geom. Dyn. 4, No. 3, 597--603 (2010; Zbl 1220.20016), Theorem A].\N\N(2) \(G\) is the fundamental group of a finite \(2\)-complex \(X\) such that \(b_{2}(X) = 0\).\N\N(3) \(G\) G is a one-relator group.\N\NThen \(G\) is virtually free-by-cyclic.\N\NThe results of Theorem A have been obtained earlier by \textit{D. Kielak} and \textit{M. Linton} [Geom. Funct. Anal. 34, No. 5, 1580--1608 (2024; Zbl 1552.20155)] under the additional hypotheses that \(G\) be hyperbolic and virtually compact special. Here this additional hypotheses have been dropped. The proof given here is entirely homological and it is not needed the geometric assumptions of hyperbolicity and specialness, but only the more algebraic RFRS condition.\N\NTheorem A (and the similar Theorem of Kielak and Linton) are both special cases of results that take as input a group of rational cohomological dimension \(n\) and produce a virtual map to \(\mathbb{Z}\) with kernel of rational cohomological dimension \(n-1\). More precisely it is proved:\N\N\textbf{Theorem 3.6} Let \(G\) be a nontrivial finitely generated RFRS group such that there is a projective resolution\N\N\(0 \longrightarrow P_{n} \longrightarrow \cdots \longrightarrow P_{0} k \longrightarrow 0\)\N\Nof the trivial \(k[G]\)-module \(k\) and \(P_{n}\) is finitely generated. The following are equivalent:\N\N(1) \(b^{(2)}_{n}(G;k) = 0\).\N\N(2) There is a finite-index subgroup \(H\leq G\) and an epimorphism \(\chi : H \longrightarrow \mathbb{Z}\) such that \(cd_{k}(\ker \chi\leq n-1\).\N\N\((3)\)\quad There is a finite-index subgroup \(H\leq G\) and an epimorphism \(\chi : H \longrightarrow \mathbb{Z}\) such that \(hd_{k}(\ker \chi)\leq n-1\).\N\NIf any of the above conditions hold and additionally there is an integer \(m\) such that \(G\) is of type \(FP_{m}(k)\) and \(b^{(2)}_{i}(G:k) = 0\) for all \(i\leq m\), then there is a finite-index subgroup \(H\leq G\) and an epimorphism \(\chi : H \longrightarrow \mathbb{Z}\) such that \(\ker \chi\) is of type \(FP_{m}(k)\) and \(cd_{k}(\ker \chi)\leq n-1\).