Steinitz classes of some abelian and nonabelian extensions of even degree (Q719137)

Let \(k\) be a number field, \(O_k\) its ring of integers and \(Cl(k)\) its class group. Let \(M\) be a finitely generated, torsion-free module of rank \(n\) over \(O_k\). Then there exists an ideal \(I\) of \(O_k\) such that \(M \simeq O_k^{n-1}\oplus I\) as an \(O_k\)-module. The class of \(I\) in \(Cl(k)\) is called the Steinitz class of \(M\). The structure of \(M\) as an \(O_k\)-module is determined up to isomorphism by its rank and its Steinitz class. One applies the previous discussion to \(M=O_K\) the ring of integers of \(K\), where \(K/k\) is an extension of number fields of degree \(n\). We also say that the Steinitz class of \(O_K\) is the Steinitz class of \(K/k\). Let \(G\) be a finite group. Let \(R_t(k, G)\) be the subset of \(Cl(k)\) consisting of those classes which are realizable as Steinitz classes of tame Galois extensions of \(k\) with Galois group isomorphic to \(G\). It follows from a theorem of \textit{L. R. McCulloh} in [J. Reine Angew. Math. 375/376, 259--306 (1987; Zbl 0619.12008)] that \(R_t(k, G)\) is a subgroup of \(Cl(k)\) when \(G\) is abelian. It is conjectured that \(R_t(k, G)\) is always a subgroup of \(Cl(k)\). The author recalls from his paper [J. Number Theory 130, No. 5, 1129--1154 (2010; Zbl 1215.11108)] his definition of a good group; for instance, abelian groups of odd order and the alternating group \(A_4\) are good; by definition the above conjecture is verified for good groups. Let \(\mathcal{G}\) be a good group of order \(m\). Let \(H\) be an abelian group of order \(n\) prime to \(m\). Let \(\mu : \mathcal{G}\rightarrow Aut(H)\) be an an action of \(\mathcal{G}\) on \(H\). Let \(G=H\rtimes_{\mu}\mathcal{G}\). In the above paper, the author supposes that \(n\) is odd and proves that \(R_t(k, G)\) is a subgroup of \(Cl(k)\). In the present paper the author considers the case of \(n\) even. The main result is a fairly explicit description of \(R_t(k, G)\) when \(H\) is a \(2\)-elementary group, from which it follows that it is a subgroup of \(Cl(k)\). Precisely \(R_t(k, G)=R_t(k, \mathcal{G})^nCl(k)^{m2^{k-2}}\), where \(n=2^k\). Furthermore \(G\) is a good group. Some examples of such groups \(G\) are: \(A_4\); \(\mathcal{G}\) cyclic of order \(m=2^k-1\) and \(H\) of order \(n=2^k\), \(G\) is then the group considered in [\textit{N. P. Byott, C. Greither} and the reviewer, J. Reine Angew. Math. 601, 1--27 (2006; Zbl 1137.11069)]. The arguments use essentially class field theory. Now suppose that \(G\) is abelian. From the theorem of McCulloh cited above, it is very hard to deduce an explicit description of \(R_t(k, G)\). In his PhD thesis (see a summary of its main results in [\textit{J. E. Carter} and the reviewer, J. Lond. Math. Soc., II. Ser. 76, No. 2, 331--344 (2007; Zbl 1130.11064)], Endo gave an explicit description of \(R_t(k, G)\) when \(G\) is abelian of odd order, or \(G\) cyclic of order \(2^l\) with \(Gal(k(\zeta_{2^l})/k)\) cyclic, where \(\zeta_{2^l}\) is a primitive \(2^l\)-th root of unity. In the last section of the present paper, the author shows that we can reduce the study of \(R_t(k, G)\) to the case of \(G\) cyclic of order \(2^l\). As a corollary of his investigation, the author proves that \(R_t(k, D_{2k})\) is a subgroup of \(Cl(k)\), where \(D_{2k}\) is the dihedral group of order \(4k\).