The Bogomolov multiplier of rigid finite groups (Q2447684)

Let \(G\) be a finite group, and let \[ B_0(G)=\mathrm{Ker} \left( H^2(G,\mathbb{Q}/\mathbb{Z}) \to \bigoplus_{A \subset G} H^2(A,\mathbb{Q}/\mathbb{Z}) \right) \] be the \textit{Bogomolov multiplier} of \(G\), where \(A\) runs over all abelian subgroups of \(G\). Let \(G\) act on itself by conjugation, and let \(H^1(G,G)\) be the first cohomology pointed set. Denote by \(TS(G)\) the subgroup of \(H^1(G,G)\) consisting of the cohomology classes becoming trivial after restricting to every cyclic subgroup of \(G\); it is called the Shafarevich-Tate group of \(G\). The group \(G\) is called \textit{rigid} if \(TS(G)\) is trivial. The aim of this paper is to show that \(B_0(G)=0\) for many rigid groups. More precisely, the authors first give a list (Proposition 1.1) of finite groups which are known to be rigid. Then the main result (Theorem 1.2 and \S 3.Remark) states that \(B_0(G)\) is trivial for all the groups of the list. The seven first cases were already known (and the authors give the relevant references), two cases constitute the main body of the paper, and the two last cases were answered in the affirmative by Michailov after the submission of the paper (see \S 3.Remark). A last part contains some general remarks and open question around the connection between the rigid groups and groups whose Bogomolov multiplier is trivial. In particular, there is an example of a group \(G\) which is rigid but satisfies \(B_0(G) \neq 0\).