Divergence and quasimorphisms of right-angled Artin groups. (Q663221)

Let \(X\) be a geodesic metric space. Let \(\alpha\colon\mathbb R\to X\) be a bi-infinite geodesic in \(X\) and let \(\rho\) be a linear function \(\rho(r)=\delta r-\lambda\) with \(0<\delta<1\) and \(\lambda\geq 0\). Define \(\text{div}(\alpha,\rho)(r)=\) length of the shortest path from \(\alpha(-r)\) to \(\alpha(r)\) that stays outside the ball of radius \(\rho(r)\) about \(\alpha(0)\) (or \(\text{div}(\alpha,\rho)(r)=\infty\) if no such path exists). It is said that \(\alpha\) has linear divergence if for some choice of \(\rho\), \(\text{div}(\alpha,\rho)(r)\) is bounded by a linear function of \(r\), and \(\alpha\) has super-linear divergence otherwise. The space \(X\) is said to be wide if no asymptotic cone of \(X\) has cut-points. In the case that \(X\) is the Cayley graph of a group \(G\), then \(G\) is said to be wide. Let \(\Gamma\) be a finite, simplicial graph with vertex set \(V\). The right-angled Artin group associated to \(\Gamma\) is the group \(A_\Gamma\) with presentation \(A_\Gamma=\langle V\mid vw=wv\) if \(v\) and \(w\) are connected by an edge in \(\Gamma\rangle\). Associated to any right-angled Artin group \(A_\Gamma\) is a CAT(0) cubical complex \(X_\Gamma\) with free action of \(A_\Gamma\). Let \(\Gamma_1\) and \(\Gamma_2\) be graphs. The join of \(\Gamma_1\) and \(\Gamma_2\) is the graph formed by joining every vertex of \(\Gamma_1\) to every vertex of \(\Gamma_2\) by an edge. The Artin group associated to such a graph splits as a direct product \(A_\Gamma=A_{\Gamma_1}\times A_{\Gamma_2}\) and \(X_\Gamma\) splits as a metric product \(X_\Gamma=X_{\Gamma_1}\times X_{\Gamma_2}\). If \(J\) is a complete subgraph of \(\Gamma\) which decomposes as a non-trivial join, then the Artin group \(A_J\) is called a join subgroup of \(A_\Gamma\). For an element \(g\in A_\Gamma\) the join length of \(g\) is defined to be \(\ell_J(g)=\min\{k\mid g=a_1\cdots a_k\) where \(a_i\) lies in a join subgroup of \(A_\Gamma\}\). If \(\alpha\) is a finite geodesic in \(X_\Gamma\), it is associated with an element \(g_\alpha\in A_\Gamma\) and the join length of \(\alpha\) is defined as \(\ell_J(\alpha)=\ell_J(g_\alpha)\). If \(\alpha\) is a bi-infinite geodesic and \(\alpha_n\) denotes the restriction of \(\alpha\) to the interval \([-n,n]\), the join length of \(\alpha\) is \(\ell_J(\alpha)=\lim_{n\to\infty}\ell_J(\alpha_n)\). Now we are in the position to quote some of the results of the present paper. A. (Theorem 4.4) Let \(\Gamma\) be a connected graph with at least 2 vertices and let \(\alpha\) be a bi-infinite geodesic in \(X_\Gamma\). Then \(\alpha\) has linear divergence if and only if the join length of \(\alpha\) is finite. B. (Corollary 4.8) Let \(\Gamma\) be a connected graph with at least 2 vertices. Then the following are equivalent. (1) \(\Gamma\) is a join, (2) \(A_\Gamma\) is wide, (3) \(A_\Gamma\) has linear divergence. If \(\Gamma\) is not a join, then \(A_\Gamma\) has quadratic divergence. A hyperbolic isometry of a proper CAT(0) space is called a rank-one isometry if some axis of that isometry does not bound a half-plane. C. If \(G\leq A_\Gamma\) is any subgroup that is not contained in a conjugate of a join subgroup, then \(G\) contains an element which acts as a rank-one isometry of \(X_\Gamma\).