This paper is a survey dedicated to the following question: given a group acting on a CAT(0) cube complex, how to exploit this action to determine whether or not the group is Gromov / relatively / acylindrically hyperbolic? As much as possible, the different criteria we mention are illustrated by applications. We also propose a model for universal acylindrical actions of cubulable groups, and give a few applications to Morse, stable and hyperbolically embedded subgroups. arXiv:1709.08843v2 [math.GR] 26 Apr 2019
PreliminariesA cube complex is a CW complex constructed by gluing together cubes of arbitrary (finite) dimension by isometries along their faces. It is nonpositively curved if the link of any of its vertices is a simplicial flag complex (ie., n + 1 vertices span a n-simplex if and only if they are pairwise adjacent), and CAT(0) if it is nonpositively curved and simply-connected. See [BH99, page 111] for more information. Proposition 2.2. [Gen16b, Proposition 2.9] Let X be a CAT(0) cube complex and A, B ⊂ X two convex subcomplexes. The projection proj B (A) is a geodesic subcomplex of B. Moreover, the hyperplanes intersecting proj B (A) are precisely those which intersect both A and B. Lemma 2.3. [HW08, Lemma 13.8] Let X be a CAT(0) cube complex, Y ⊂ X a convex subcomplex and x ∈ X a vertex. Any hyperplane separating x from its projection onto Y must separate x from Y . Lemma 2.4. [Gen16c, Proposition 2.6] Let X be a CAT (0) cube complex, C ⊂ X a convex subcomplex and x, y ∈ X two vertices. The hyperplanes separating the projections of x and y onto C are precisely the hyperplanes separating x and y which intersect C.Lemma 2.5. [HW08, Corollary 13.10] Let X be a CAT(0) cube complex and Y 1 , Y 2 ⊂ X two convex subcomplexes. If x ∈ Y 1 and y ∈ Y 2 are two vertices minimising the distance between Y 1 and Y 2 , then the hyperplanes separating x and y are precisely the hyperplanes separating Y 1 and Y 2 .Cycles of subcomplexes. Given a CAT(0) cube complex, a cycle of subcomplexes is a sequence of subcomplexes (C 1 , . . . , C r ) such that, for every i ∈ Z/rZ, the subcomplexes C i and C i+1 intersects.Proposition 2.7. Let (A, B, C, D) be a cycle of four convex subcomplexes. There exists a combinatorial isometric embedding [0, p] disjoint from B and D (resp. A and C).Proof. First of all, let us record a statement which is contained into the proof of [Gen17b, Proposition 2.111] (in the context of quasi-median graphs, a class of graphs including median graphs, i.e., one-skeleta of CAT(0) cube complexes).Fact 2.8. If a is a vertex of A ∩ D minimising the distance to B ∩ C and if b (resp. c, d) denotes the projection of a onto B (resp. B ∩ C, C), then there exists a combinatorial isometric embedding [0, p] × [0, q] → X such that (0, 0) = a, (p, 0) = b, (p, q) = c and (0, q) = d.By convexity of A, B, C and D, this implies that [0, p] Let J be a hyperplane intersecting [0, p] × {0}. We know from Lemma 2.3 that J must be disjoint from B. Moreover, if J intersects D, then it follows from Helly's property, satisfied by...