Relatively projective pro-\(p\) groups (Q6584660)

A subgroup of a free profinite group need not be free, that is the Schreier theorem does not hold for free profinite groups (subgroups of free profinite groups are groups of cohomological dimension \(1\); they are called projective, as they satisfy the same universal property with the projective modules). Similarly, the Kurosh subgroup theorem [\textit{A. Kurosch}, Math. Ann. 109, 647--660 (1934; Zbl 0009.01004)] does not hold for free products of profinite groups. The first author [Proc. Lond. Math. Soc., III. Ser. 55, 266--298 (1987; Zbl 0666.20015)] introduced the notion of a profinite group projective relative to a family \(\mathcal{G}\) of its subgroups, closed under conjugation. In the same paper, he proved that a subgroup \(G\) of a free profinite product \(H=\coprod_{x\in X} H_{x}\) is projective relative to the family \(\mathcal{G}=\{ H_{x}^{h} \cap G \mid h \in H, \, x \in X \}\) of subgroups. An analogous result holds for subgroups of free pro-\(p\) products.\N\NIn the paper under review, the authors prove the converse of this fact (that is, a profinite group \(G\), projective relative to a continuous family \(\mathcal{G}\) of its subgroups, is a subgroup of a free product in the above manner) in the case of pro-\(p\) groups.