On the congruence subgroup problem for integral group rings (Q401923)

Let \(A\) be a finite-dimensional semisimple \(\mathbb{Q}\)-algebra and let \(R\) be a \(\mathbb{Z}\)-order in \(A\). In the archetypal case \(A=\mathbb{Q}G\) is the rational group algebra of some finite group \(G\) and \(R = \mathbb{Z}G\) is its integral group ring. For a ring \(R\) denote by \(\mathcal{U}(R)\) the group of units of \(R\) and by \(U(R, m)\) the kernel of the natural homomorphism \(\mathcal{U}(R)\rightarrow \mathcal{U}(R/mR)\), \(m\in\mathbb{Z}\). Recall that the congruence kernel \(C(R)\) is defined as the kernel of the canonical map between completions of \(\mathcal{U}(R)\) taken with respect to profinite and congruence topologies, i.e., \[ C(R) = \mathrm{ker}\left(\varprojlim_{N \trianglelefteq_{\text{f.\,i.}} \mathcal{U}(R)}\left(\mathcal{U}(R) / N \right) \rightarrow \varprojlim_{m}\left(\mathcal{U}(R)/\mathcal{U}(R, m)\right)\right). \] An algebra \(A\) is said to satisfy \textit{property CSP} if the congruence kernel \(C(R)\) is finite for some \(\mathbb{Z}\)-order \(R\) in \(A\) (this definition does not depend on the choice of \(R\)). One of the main results of the paper is a description of the class \(\mathcal{C}\) of finite groups whose rational group algebra \(\mathbb{Q}G\) satisfies property CSP. More precisely, the authors determine 19 families of finite groups which can not be homomorphic images of \(G\) if \(G\) lies in \(\mathcal{C}\). Conversely, they give a sufficient condition for \(G\) to lie in \(\mathcal{C}\) formulated in terms of 22 families of finite groups (there are 3 more additional families). If neither of homomorphic images of \(G\) lies in any of these families then \(G\) is in \(\mathcal{C}\).