Embeddings of Kleinian groups with torsion (Q5936960)

A group \(G\) is called cohopfian if every one-to-one homomorphism \(h: G \to G\) is in fact an isomorphism. For torsion free cocompact geometrically finite Kleinian groups of the first kind it was shown in [ \textit{S-C. Wang} and \textit{Y-Q. Wu}, Proc. Lond. Math. Soc., III. Ser. 68, No. 1, 203--224 (1994; Zbl 0841.57028)] that they are cohopfian. In [\textit{L. Potyagailo} and \textit{S.-C. Wang}, St. Petersbg. Math. J. 11, No. 5, 861--881 (2000); translation from Algebra Anal. 11, No. 5, 194--220 (2000; Zbl 0977.57017)] it was shown that a torsion free non-elementary freely indecomposable geometrically finite Kleinian group of the second kind is cohopfian if it does not split over a cyclic group. For instance, if we consider \(G=\langle A,B\rangle\) a Schottky group of genus \(2\), then we can consider the one-to-one homomorphism \(h:G \to G\) such that \(h(A)=A^2\) and \(h(B)=B\), which fails to be surjective. The Rips conjecture asserts that a non-elementary word hyperbolic group is cohopfian if and only if it is freely indecomposable. In this paper, the authors consider the torsion situation. They observe that for torsion Kleinian groups cohopficity it is more complicated than in the conjecture above. In fact, they observe that in this case cohopficity cannot be determined by the number of ends and that such a property is not necessarily preserved by finite index subgroups. The authors use the following known results. Every geometrically finite Kleinian group containing no subgroup \(Z+Z\) is isomorphic to a word hyperbolic group and \(G\) is not cohopfian if either it is not freely indecomposable or it is an elementary Kleinian group. It is for such reason they consider the class \(\mathcal K\) consisting of all non-elementary, freely indecomposable, geometrically finite Kleinian groups containing no \(Z+Z\) subgroup. The authors show that in the above class there are groups which are not cohopfian. Some of the results of the paper are the following. Theorem 1a. (Algebraic version): \(K \in {\mathcal K}\) is cohopfian if \(K\) has one end. Theorem 2a. (Algebraic version): Suppose \(K \in {\mathcal K}\) and each finite subgroup of \(K\) is cyclic. Then \(K\) is cohopfian if and only if each hyperbolic element which commutes with an elliptic element can be conjugated into a vertex group associated with a minimal splitting of \(K\) over finite groups. There are also geometric versions of the above results.