Sofic groups and profinite topology on free groups. (Q958639)

Let \(H\) be a finite group with a bi-invariant metric \(d\). Let \(G\) be a group, \(\Phi\subset G\) be a finite set and \(\varepsilon>0\), \(\alpha>0\). A map \(\varphi\colon\Phi\to H\) is said to be a \((\Phi,\varepsilon,\alpha)\)-homomorphism if: (1) For any two elements \(a,b\in\Phi\) with \(a\cdot b\in\Phi\), \(d(\varphi(a)\cdot\varphi(b),\varphi(a\cdot b))<\varepsilon\). (2) If \(1_G\in\Phi\), then \(\varphi(1_G)=1_H\). (3) For any \(a\neq 1_G\), \(d(\varphi(a),1_H)>a\). A group \(G\) is called weakly sofic (w-sofic) if there exists an \(\alpha>0\) such that for any finite \(\Phi\subset G\) and for any \(\varepsilon>0\) there exists a finite group \(H\) with a bi-invariant metric \(d\) and a \((\Phi,\varepsilon,\alpha)\)-homomorphism to \((H,d)\). In case where the finite group \(H\) is a symmetric group \(S_n\) and the metric \(d\) the normalized Hamming metric, then the group \(G\) is called a sofic group. The property of a group to be a w-sofic group is closely related with the profinite topology of the group. For a group \(G\) and \(g\in G\) let \([g]^G\) denote the conjugacy class of \(g\) in \(G\). If \(A\) is a subset of the group \(G\) let \(\overline A\) denote the closure of \(A\) in the profinite topology on \(G\). The main theorem of the paper is the Theorem: Let \(F\) be a finitely generated free group and \(N\) be a normal subgroup of \(F\). Then \(F/N\) is a w-sofic group if and only if for any finite sequence \(g_1,g_2,\dots,g_k\) of elements of \(N\) one has \(\overline{[g_1]^F[g_2]^F\cdots[g_k]^F}\subseteq N\). An interesting corollary of this result is the equivalence of the following two conjectures. Conjecture 1. For a finitely generated free group \(F\), there exists a sequence \(g_1,g_2,\dots,g_k\in F\) such that \(\overline{[g_1]^F[g_2]^F\cdots[g_k]^F}\subseteq\langle g_1,g_2,\dots,g_k\rangle^F\), where \(\langle g_1,g_2,\dots,g_k\rangle^F\) denotes the normal subgroup of \(F\) generated by \(g_1,g_2,\dots,g_k\). Conjecture 2. There exists a non-w-sofic group. As the authors point out it is still an open question if there exists a non-sofic group.