Height in splittings of hyperbolic groups. (Q1431071)

Let \(H\) be a hyperbolic subgroup of the hyperbolic group \(G\). Call the elements \(\{g_i\mid i=1,\dots,n\}\) essentially distinct if \(Hg_i\neq Hg_j\) whenever \(i\neq j\). Conjugates of \(H\) by essentially distinct elements are called essentially distinct conjugates. The height of an infinite subgroup \(H\) in \(G\) is \(n\) if there is a collection of \(n\) essentially distinct conjugates of \(H\) such that the intersection of the collection is infinite and \(n\) is maximal with respect to this property. Finite groups have height \(0\) by definition. The main result of \textit{R. Gitik}, \textit{M. Mitra}, \textit{E. Rips}, and \textit{M. Sageev} [Trans. Am. Math. Soc. 350, No. 1, 321-329 (1998; Zbl 0897.20030)] is the following: if \(H\) is a quasiconvex subgroup of a hyperbolic group \(G\), then \(H\) has finite height. The paper under review proves the converse of that theorem for hyperbolic groups \(G\) which split as \(G=G_1*_HG_2\) or \(G=G_1*_H\) with hyperbolic edge and vertex groups where the inclusions are quasi-isometric embeddings, giving an affirmative answer to a question of Swarup. The main result states: Let \(G\) be a hyperbolic group splitting over \(H\) with hyperbolic edge and vertex groups; assume the inclusions of \(H\) are quasi-isometric embeddings; then \(H\) is of finite height in \(G\) if and only if it is quasiconvex in \(G\). The paper includes details regarding consequences and questions related to the main theorem including malnormality, more general graphs of hyperbolic groups as well as more geometric applications.