Virtual specialness of certain graphs of special cube complexes (Q6144248)

Suppose \(\Sigma\) is a surface of finite type with infinite fundamental group, and draw a family of (essential) disjoint simple closed curves on the surface. Lifting to the universal cover \(\tilde \Sigma\), we see in the preimage of these curves a pattern of disjoint lines, each of which divides \(\tilde \Sigma\) into two pieces. We may form a graph with vertices the complementary components of these lines, connecting two vertices by an edge when their closures intersect in one of our lines. This graph turns out to be a tree, and \(G = \pi_1(\Sigma)\) acts on this tree, which we say is a \textit{splitting} of \(G\). If the curves are allowed to (self-)intersect, the lifted pattern of lines will have intersections and this ``dual'' graph is no longer a tree, even though the lines, being codimension one in \(\tilde\Sigma\), separate it into two pieces. It was \textit{M. Sageev} [Proc. Lond. Math. Soc. (3) 71, No. 3, 585--617 (1995; Zbl 0861.20041)] who showed that the proper ``dual'' object is a \textit{CAT(0) cube complex,} and that one may be constructed any time one has a ``codimension-one subgroup'' of a group \(G\). CAT(0) cube complexes and splittings of groups lie at the heart of geometric group theory. Particularly of interest when one has a nonpositively curved cube complex (that is, the quotient, or ``downstairs'' object analogous to \(\Sigma\) rather than to \(\tilde\Sigma\)) is the behavior of its \textit{hyperplanes.} These hyperplanes may exhibit certain pathologies; when they are absent, one says that one has a \textit{special cube complex.} Fundamental groups of (virtually) compact special cube complexes are highly constrained (for example, they are linear over \(\mathbb{Z}\)), and yet a rich and varied family (for example, containing the fundamental groups of all finite-volume hyperbolic \(3\)-manifolds). Verifying that some finite cover of a given nonpositively curved cube complex is special thus becomes a thorny and very interesting problem. In this paper, the authors state a conjectural characterization of virtual specialness: a compact nonpositively curved cube complex \(X\) is virtually special if and only if \(\pi_1(X)\) has \textit{finite stature} (a term introduced by the authors in a previous paper) with respect to the collection of all hyperplane stabilizers in the action of \(\pi_1(X)\) on \(\tilde X\). They verify the ``only if'' direction of this conjecture, and provide evidence in its favor by studying splittings of groups as graphs of nonpositively curved cubical groups. The notion of \textit{finite stature} is related (as the name implies) to the notion of \textit{finite height}, see [\textit{R. Gitik} et al., Trans. Am. Math. Soc. 350, No. 1, 321--329 (1998; Zbl 0897.20030)], but is weaker. Here is the main theorem: suppose \(X\) is a compact nonpositively curved cube complex which splits as a graph of nonpositively curved cube complexes. If the fundamental group of each vertex space is word-hyperbolic (notice that \(\pi_1(X)\) emphatically need not be) and \(\pi_1(X)\) has finite stature with respect to the collection of these vertex groups, then \(X\) is virtually special. (This is in fact an ``if and only'' if by the ``only if'' direction of the main conjecture.)