On the indecomposability of a remarkable new family of modules appearing in Galois theory (Q2667909)

Let \(K\) be a field, \(K_{sep}\) a separable closure of \(K\), and \(G_K\) the absolute Galois group of \(K\), namely, the Galois group \(\mathcal{G}(K_{sep}/K)\). Fix a prime number \(p\) and denote by \(G_K(p)\) the absolute \(p\)-Galois group of \(K\), defined to be the maximal \(p\)-quotient of \(G_K\) (equivalently, the Galois group of the maximal \(p\)-extension \(K(p)\) of \(K\) in \(K_{sep}\)). One of the major problems in Galois theory is to find properties of absolute Galois groups and of absolute \(p\)-Galois groups which distinguish them among general profinite groups and pro-\(p\)-groups, respectively. The research presented in the paper under review is motivated by a powerful new perspective in the analysis of the considered problem, which has recently emerged from the study of decompositions of Galois modules related to classical parameterizing spaces of certain Galois extensions. For example, various Galois modules over the field \(\mathbb{Q}\) of rational numbers have successfully been used in: [\textit{L. Bary-Soroker} et al., Adv. Math. 284, 186--212 (2015; Zbl 1334.12003)], to make substantial progress towards determining the explicit structure of the Sylow pro-\(p\) subgroups of \(G_{\mathbb{Q}}\), for \(p > 2\). A simple example of the considered spaces is the group \(J(K)\) of period \(p\) whose nontrivial cyclic subgroups parameterize the degree \(p\) extensions of \(K\) in \(K(p)\); by Kummer theory, \(J(K) = K^{\ast}/K^{\ast p}\) if \(K\) contains a primitive \(p\)-th root of unity, and in case char\((K) = p\), by the Artin-Schreier theorem, \(J(K) = K/\rho (K)\), where \(\rho (K) = \{\alpha^p - \alpha : \alpha \in K\}\). Under the hypothesis that \(K\) is a Galois extension of a field \(F\), the action of the Galois group \(\mathcal{G}(K/F)\) upon \(K\) induces on \(J(K)\) a structure of a module over the group algebra \(\mathbb{F}_p[\mathcal{G}(K/F)]\) (as usual, \(\mathbb{F}_p\) denotes the field with \(p\) elements). The cohomology groups \(H^m(G_K(p), \mathbb{F}_p)\), \(m \in \mathbb{N}\), can also be viewed as \(\mathbb{F}_p[\mathcal{G}(K/F)]\)-modules, and for each \(m \in \mathbb{N}\), the quotient group \(K^{\ast}/K^{\ast p^{m}}\) has a structure of a \((\mathbb{Z}/p^m\mathbb{Z})[\mathcal{G}(K/F)]\)-module. For technical and mathematical reasons (explained in the Introduction to the reviewed paper), most of the research in this area restricts to the special case where \(\mathcal{G}(K/F)\) is a cyclic \(p\)-group. As it turns out, most of the summands in the decompositions of the considered Galois modules are free over some appropriately chosen quotient ring; this can then be translated into properties that distinguish absolute \(p\)-Galois groups from arbitrary pro-\(p\) groups. Another fact of particular interest is the appearance of ``exceptional'' summands, in many respects; frequently non-free, they are far more scarce than other summand types, and they help control certain arithmetic conditions which is critical to ensure the other summands are free. In this way, these exceptional summands are sources of new information on various specific properties of absolute Galois groups. The paper under review deals with the study of the module that appears as a single exceptional summand in the decomposition of \(K^{\ast}/K^{\ast p^{m}}\), under the assumption that \(\mathcal{G}(K/F)\) is a cyclic group of order \(p^n\); the structure of \(K^{\ast}/K^{\ast p^{m}}\) as a \((\mathbb{Z}/p^m\mathbb{Z})[\mathcal{G}(K/F)]\)-module has been obtained by the authors in the preprint [\textit{J. Mináč} et al., ``Galois module structure of the units modulo \(p^m\) of cyclic extensions of degree \(p^n\)'', Preprint, \url{arXiv:2105.13216}]. The authors observe that the isomorphism class of the exceptional module belongs to a series of additive \((\mathbb{Z}/p^m\mathbb{Z})[\mathbb{Z}/p^n\mathbb{Z}]\)-modules, each of which is presented in terms of generators and defining relations; the latter depend on \(m\), an integer \(d\), and an \(m\)-tuple \(\bar a = (a_0, \dots , a_{m-1}) \in \{-\infty , 0, 1, \dots , n\}^m\). The module is denoted by \(X_{\bar a,d,m}\), and the main result of the reviewed paper states that, under certain conditions, it is indecomposable. In addition, it is proved that if \(m\), \(d\) and \(\bar a\) satisfy these conditions, then the \(m\)-tuple \(\bar a\) is determined by module theoretic invariants of \(X_{\bar a,d,m}\). Finally, the authors consider two of the imposed conditions and give examples which show that the violation of any of them could yield decomposability of \(X_{\bar a,d,m}\). As a whole, the paper is well-written and informative of various aspects of the discussed topic. The list of references contains 44 items.