The congruence subgroup problem for low rank free and free metabelian groups (Q1702696)

The classical congruence subgroup problem asks whether all subgroups of finite index in an arithmetic group can be described by congruence relations: for example, in the most classical setting, whether any finite-index subgroup of \(\mathrm {SL}_2(\mathbb Z)\) contains the subgroup of matrices which reduce to the identity modulo \(n\), for some integer \(n\). Arithmetic groups are themselves described as automorphisms of lattices in vector spaces (with additional structures) and another formulation of the problem is: if \(L\) is a lattice in a real vector space and \(G = \mathrm{Aut}(L)\), is the natural surjective map \(\widehat G \to \mathrm{Aut} (\widehat{L})\) injective (where \(\,\widehat\cdot\,\) denotes profinite completion), and if not what is its kernel? The latter formulation makes sense for any finitely generated group \(L\). In the first part of this paper, the authors give a new proof of the result (originally proven by \textit{M. Asada} [J. Pure Appl. Algebra 159, No. 2--3, 123--147 (2001; Zbl 1045.14013)]) that this generalised congruence subgroup problem has a positive answer for the free group \(F_2\) on two generators. The proof is completely elementary and relies on a virtual splitting of \(\mathrm{Aut}(F_2)\). In the second part of the paper, the authors deal with the free metabelian group \(\Phi_n\), which is the largest group on \(n\) generators, the derived series of which stops at the second step. In other words, it is isomorphic to \(F_n / [F_n', F_n']\). The authors prove that the generalised congruence subgroup problem has a negative solution for \(\Phi_2\) and \(\Phi_3\) in a very strong sense: in each case the congruence kernel contains an infinitely generated profinite free group. They also announce results of the first author [``The congruence subgroup problem for the free metabelian group on \(n \geq 4\) generators'', Preprint, \url{arXiv:1701.02459}] that the congruence kernel for \(\mathrm{Aut}(\Phi_n)\) is trivial for \(n \geq 4\).