Right-angled Artin groups and a generalized isomorphism problem for finitely generated subgroups of mapping class groups (Q1934646): Difference between revisions

Let \(\text{Mod}_{g,p}\) the mapping class group of the surface of genus \(g\), i.e., the group consisting of isotopy classes of self-homeomorphisms which preserve a set of \(p\) points. For the main result of the paper, we start with a finite collection of elements of \(\text{Mod}_{g,p}\) that are either Dehn twists or pseudo-Anosov homeomorphisms supported on connected subsurfaces of the surface and that they are irredundant (i.e., if a subcollection generate an abelian collection, then they there is no linear relation between them that vanishes). Then, the main result is that there is a positive integer \(N\), such that for all \(n\geq N\), the \(n\)-th powers of the elements generate a right-angled Artin group. The isomorphism type of the Artin group is determined by the coincidence graph \(\Gamma\) of the original collection. The coincidence graph is the graph which vertices are the given elements, and where two elements are connected if they are supported by disjoint subsurfaces. For the inverse problem, given a right-angled Artin group \(A({\Gamma})\), there is a number \(g\) so that \(\text{Mod}_{g,0}\) contains \(A({\Gamma})\). Furthermore, the vertices of \(\Gamma\) correspond to Dehn twists about simple closed curves or pseudo-Anosov homeomorphisms supported on disjoint subsurfaces. For the other inclusion, it is proved that \(\text{Mod}_{g,p}\), for \(g>2\), or \(g=2\) and \(p > 1\), does not virtually inject into a right-angled Artin group. For the analogue of the Tit's alternative for finitely generated linear groups, it is proved that a subgroup \(G\) of \(\text{Mod}_{g,p}\) generated by finitely many elements of infinite order, is either a virtually torsion-free abelian group or there is an \(N\) such that for all \(n \geq N\), the \(n\)-th powers of the generators generate a subgroup of \(G\) that is enveloped by a right-angled Artin group. For decision problems, it should be noted that the isomorphism question for right-angled Artin groups has been settled. The author shows that if \(G\) is a finitely generated group that contains the product of two non-abelian free groups, then there is no algorithm for determining if two finitely generated subgroups are isomorphic. For analogues with lattices in semisimple Lie groups, the author proves the following results. Start with a finite collection of irredundant elements that generate a discrete subgroup of \(\mathrm{SO}(K, 1)\). Then there is a positive integer \(N\), such that for all \(n\geq N\), the \(n\)-th powers of the elements generate a right-angled Artin group that is the free product of finitely generated free subgroups. That shows that a lattice of \(\mathrm{SO}(k, 1)\), with \(k \geq 3\), can not be commensurable with a right-angled Artin group. For the complex hyperbolic case, an extension of the above methods, shows that a similar result holds for subgroups of \(\mathrm{SU}(k, 1)\) where the last group is a right-angled Artin group which is a free product of finitely generated torsion-free nilpotent groups. The author states a number of open questions. First, he asks if \(\text{Out}(F_3)\) embeds into a right-angled Artin group. That will show that \(\text{Out}(F_3)\) is linear. Also he asks whether the number \(N\) that appears in the main theorem is uniform, i.e., that it depends only on the surface. Also, the author closes with the conjecture that \(\text{Mod}_{2,0}\) and the braid groups \(B_n\) are virtually special.