Realizability and admissibility under extension of \(p\)-adic and number fields. (Q373994)

Let \(K\) be a field. An extension \(L\) of \(K\) is said to be \(K\)-adequate, if it is embeddable over \(K\) as a maximal subfield of a finite-dimensional central division \(K\)-algebra. A group \(G\) is called \(K\)-admissible, if there exists a \(G\)-extension \(L/K\), i.e. a Galois extension \(L/K\) with Galois group \(\mathcal G(L/K)\) isomorphic to \(G\), so that \(L\) is \(K\)-adequate. Equivalently, \(G\) is \(K\)-admissible, if there is a \(G\)-crossed product division \(K\)-algebra. Adequacy and admissibility were introduced by \textit{M. M. Schacher} who showed that every finite group \(G\) is admissible over some number field \(K\) [see J. Algebra 9, 451-477 (1968; Zbl 0174.34103)]. On the other hand, for many groups \(G\) it is an open problem to determine the number fields over which they are admissible. As a matter of fact, searching for an explicit description for all groups seems to be hopeless. Given a field \(K\) over which \(G\) is admissible, the paper under review considers the problem of finding those finite extensions of \(K\) over which \(G\) is still admissible. Assuming that \(M/K\) is a finite extension, the authors impose two conditions on its local behaviour. The first one is that \(M/K\) is non-sensitive. The condition is too technical to be stated here but it excludes certain configurations for the completions \(M_v/K_v\), one in case \(K_v=\mathbb Q_5\) and \(16\) when \(K_v=\mathbb Q_3\) (up-to a \(K_v\)-isomorphism). Secondly, the authors suppose that \(G\) is realizable as a Galois group over \(M\), and this can be achieved with prescribed local conditions; in other words, \(G\) admits the Grunwald-Neukirch (GN) property. Also, they assume that each Sylow subgroup of \(G\) is normal or metacyclic, and the Sylow \(2\)-subgroups of \(G\) are metacyclic. The main result of the paper states that under these three conditions \(G\) is \(M\)-admissible if and only if it satisfies Liedahl's condition over \(M\) (concerning this condition, see Theorem 28 of \textit{S. Liedahl}'s paper [J. Algebra 169, No. 3, 965-983 (1994; Zbl 0820.20024)]). The question of whether the non-sensitivity condition on \(M/K\) is essential seems to be open. The authors show that in many cases, including Sylow metacyclic and nilpotent groups of order prime to the number of roots of unity in \(M\), a \(K\)-admissible group is \(M\)-admissible if and only if it satisfies the (easily verifiable) Liedahl condition over \(M\). They provide examples illustrating the fact that, for certain \(K\)-admissible groups \(G\), the fulfillment of Liedahl's condition over \(M\) does not guarantee \(M\)-admissibility.