On the exponent of Bogomolov multipliers (Q2414566)

Let \(G\) be a finite group and recall that the \textit{Bogomolov multiplier} \(\operatorname{B}_0(G)\) of \(G\) is the intersection of the kernels of the restriction maps \[ \operatorname{H}^2(G,\,\mathbb{Q}/\mathbb{Z})\longrightarrow\operatorname{H}^2(A,\, \mathbb{Q}/\mathbb{Z}), \] where \(A\) runs through all (two-generator) abelian subgroups of \(G\) (in particular, it is isomorphic to a subgroup of the Schur multiplier \(\operatorname{M}(G)\simeq \operatorname{H}^2(G,\,\mathbb{Q}/\mathbb{Z})\) of \(G\)). A classical problem about the Schur multiplier asks what is the relation between the exponent of a finite group and its Schur multiplier. It is known that \(\operatorname{exp}(\operatorname{M}(G))\) always divides the order of \(G\) and that it often (but not always) divides \(\operatorname{exp}(G)\); moreover, \(\operatorname{exp}(\operatorname{M}(G))\) can be bounded in terms of \(\operatorname{exp}(G)\) only, but no good bound in this sense has yet been found. It is not difficult to see that the exponent of the Schur multiplier divides the product of the exponent of the group by the exponent of the Bogomolov multiplier; so good bounds on \(\operatorname{exp}(\operatorname{B}_0(G))\) will produce good bounds on \(\operatorname{exp}(\operatorname{M}(G))\). The aim of the paper under review is to show that the exponent of \(\operatorname{B}_0(G)\) divides the exponent of \(G\) in many more cases than those known for the Schur multiplier. These cases are the following ones: (i) \(G\) is metabelian; (ii) \(\operatorname{exp}(G)=4\); (iii) \(G\) is nilpotent of class \(\leq5\); (iv) \(G\) is a \(4\)-Engel group. In order to prove the result in the above four cases, the author observes that it is possible to assume \(G\) being a \(p\)-group; he then set up a \textit{\(\operatorname{CP}\) cover} \(H\) of \(G\), which is essentially an analogue to a Schur cover but in which the Bogomolov multiplier takes the place of the Schur multiplier and in which any commutator law satisfied by \(G\) is also valid; and, finally, he attacks each of the four cases separately through careful commutator calculations in \([H,\,H]\). The author poses the following conjecture: does \(\operatorname{exp}(\operatorname{B}_0(G))\) always divides \(\operatorname{exp}(G)\)? A positive answer to such a question yields also that \(\operatorname{exp}(\operatorname{M}(G))\) always divides \((\operatorname{exp}(G))^2\).