Steinitz classes of Galois extensions with Galois group a 2-group (Q351237)

Let \(K/k\) be an extension of number fields and let \(\mathcal O_K\) and \(\mathcal O_k\) be the corresponding rings of integers. Then, as an \(\mathcal O_k\)-module, \(\mathcal O_K\cong\mathcal O_k^{[K:k]-1}\oplus I\), where \(I\) is an ideal of \(\mathcal O_k\), which is well-defined up to principal ideals. The ideal class of \(I\) is denoted by \(\mathrm{cl}_k(\mathcal O_K)\) and is called the Steinitz class of \(K/k\). Given a number field \(k\) and a finite group \(\Gamma\), the set \(R_m(k,\Gamma)=\{c\in\mathrm{Cl}(k):\;\exists\, K/k\text{ tame Galois, }\mathrm{Gal}(K/k)\cong \Gamma, \mathrm{cl}_k(\mathcal O_K)=c\}\) has been conjectured to be a subgroup of \(\mathrm{Cl}(k)\) for all choices of \(k\) and \(\Gamma\). This conjecture is known to hold for abelian groups, as a consequence of the main result in [\textit{L. R. McCulloh}, J. Reine Angew. Math. 375/376, 259--306 (1987; Zbl 0619.12008)], which however does not give a very explicit description for \(R_m(k,\Gamma)\). If the order of the abelian group \(\Gamma\) is odd, then an explicit description of \(R_m(k,\Gamma)\) has been given by L. P. Endo in his unpublished PhD thesis. An analogous result when the class number of \(k\) is of odd order and the order of the abelian group \(\Gamma\) is arbitrary is implicit in Endo's work (some of Endo's results are recalled in [\textit{A. Cobbe}, J. Théor. Nombres Bordx. 22, No. 3, 607--628 (2010; Zbl 1267.11112)]). Endo's results have been generalized by different authors and the conjecture about \(R_m(k,\Gamma)\) is known to hold also for a lot of non-abelian groups. The first main result proved in the paper under review is that if \(\Gamma\) is any finite \(2\)-group and the base field \(k\) has odd class number, then \(R_m(k,\Gamma)=\mathrm{Cl}(k)\). As discussed above, the assumption about the class number is quite natural in this context, since it simplifies calculations already in the abelian case. The main idea in the proof of the theorem is to consider a subextension of the compositum of an arbitrary tame Galois extension \(N/k\) with Galois group \(\Gamma\) (which exists by a well-known theorem of Shafarevich) and some extensions \(k(\sqrt{r})/k\) of degree \(2\). By varying \(r\) this construction gives tame \(\Gamma\)-extensions with any Steinitz class. This idea has been further developed by the same author in [J. Number Theory 133, No. 2, 611--619 (2013; Zbl 1311.11106)], where some more general results have been obtained. The second main theorem proved in this paper states that given a number field \(k\) and a class \(c\in\mathrm{Cl}(k)\), there exists a tamely ramified quadratic extension of \(k\) with Steinitz class \(c\), which can be embedded in a Galois extension \(N/K\) of Galois group \(\Gamma\), where \(\Gamma\) belongs to a certain family of \(2\)-groups defined in the paper. The motivation for this kind of results comes from the study of realizable Galois module classes.