No label defined (Q4702002)

For a quadratic number field \(k\), let \(\text{Cl}_2(k)\) denote its \(2\)-class group and \(k^1\) its Hilbert \(2\)-class field; also, let \((a, b, \ldots)\) denote the direct sum of cyclic groups of order \(a, b, \dots\). The author proves results about the \(2\)-class group of \(k^1\) such as the following: (1) If \(k\) is a quadratic number field such that \(\text{Cl}_2(k)\) is elementary abelian and has rank at least \(4\), then \(\text{Cl}_2(k^1)\) is not cyclic. (2) If \(k\) is a real quadratic number field with \(\text{Cl}_2(k) \simeq (2,2,4)\), and if its discriminant is a sum of two squares, then \(\#\text{Cl}_2(k^1) \geq 4\). (3) If \(k\) is a complex quadratic number field with \(\text{Cl}_2(k) \simeq (4,4)\), then \(\#\text{Cl}_2(k^1) \geq 8\). The proof of (3) proceeds as follows: According to \textit{E. Benjamin, F. Lemmermeyer} and \textit{C. Snyder} [J. Number Theory 67, 229-245 (1997; Zbl 0919.11074)] it is known that \(\text{Cl}_2(k^1)\) is not cyclic; if the claim were wrong, then \(\text{Cl}_2(k^1) \simeq (2,2)\). Thus \(G = \text{Gal}(k^1/k)\) is a group of order \(64\) with \(G' \simeq (2,2)\) and \(G/G' \simeq (4,4)\). Going through a list of groups of order \(64\) he finds that there are three possibilities, and these are then excluded by computing some transfers and using results on the capitulation of ideal classes to derive a contradiction. The other claims are proved in a similar way.