Signed logarithmic classes of number fields (Q5939720)

The author defines a signed logarithmic 2-class group of a number field as the actual analogue in the logarithmic context of the usual restricted ideal 2-class group. Logarithmic classes were introduced by the author [J. Théor. Nombres Bordx. 6, 301-325 (1994; Zbl 0827.11064)] in connection with the wild kernel of \(K\)-theory: as a matter of fact, when the number field \(K\) contains the \(2p\)th roots of unity for an odd prime \(p\), there exists a canonical isomorphism \(\mu_p \otimes\widetilde{\text{Cl}}_K\simeq WK_2(K)/ W(K_2(K)^p\) between the twisted logarithmic \(p\)-class group and the \(p\)-quotient of the wild kernel [Acta Arith. 67, 335-348 (1994; Zbl 0835.11042)]. And since the logarithmic class groups are easily computable, as explained by \textit{F. Diaz y Diaz} and the reviewer [J. Number Theory 76, 1-15 (1999; Zbl 0930.11079)], this gives rise to an algorithmic approach to the Hilbert kernel. For \(p=2\) however, the isomorphism above is not valid without the assumption \(i\in K\) (in fact when \(i\) is not contained in the cyclotomic \(\mathbb{Z}_2\)-extension \(K^c\) of \(K\)) and the situation is more intricate. In the article under review, the author defines a logarithmic signature that involves both the real places and certain 2-adic ones (namely the 2-adic places \({\mathfrak p}\) such that the cyclotomic \(\mathbb{Z}_2\)-extension \(K_{\mathfrak p}^c\) of the local field \(K_{\mathfrak p}\) does not contain the 4th root of unity \(i\)). By this way he extends a previous generalization of the reviewer [Manuscr. Math. 93, 409-420 (1997; Zbl 0887.11044)]. Unfortunately there is a misprint in Lemma 8 that affects some formulas in Section 3 of a factor equal to 1 or 2. This point will be corrected in a Corrigendum to appear. Finally applications to the wild kernel will also follow in a forthcoming article.