On the Galois-Gauss sums of weakly ramified characters (Q6041085)

Let \(L/K\) be a Galois extension of number fields of odd degree, let \(G=\mathrm{Gal}(L/K)\) and write \(\mathcal{D}_{L/K}\) for the inverse different of \(L/K\). By Hilbert's classical formula for the valuation of \(\mathcal{D}_{L/K}\), there exists a unique \(G\)-stable fractional ideal \(\mathcal{A}_{L/K}\) of \(L\), such that \((\mathcal{A}_{L/K})^{2}=(\mathcal{D}_{L/K})^{-1}\). Further, under the hypotheses that \(L/K\) is weakly ramified (i.e.,\ the second ramification subgroup in \(G\) of every place of \(K\) is trivial), \textit{B. Erez} [Math. Z. 208, No. 2, 239--255 (1991; Zbl 0713.11078)] showed that \(\mathcal{A}_{L/K}\) is a projective \(\mathbb{Z}[G]\)-module, and thus defines a stable-isomorphism class \([\mathcal{A}_{L/K}]\) in the so-called reduced projective class group \(\mathrm{Cl}(\mathbb{Z}[G])\) of \(\mathbb{Z}[G]\). The square root of the inverse different \(\mathcal{A}_{L/K}\) and/or its class in (variants of) \(\mathrm{Cl}(\mathbb{Z}[G])\) have been studied in various situations in a number of further articles including [\textit{B. Erez} and \textit{M. J. Taylor}, Ann. Math. (2) 135, No. 2, 271--296 (1992; Zbl 0756.11035); \textit{S. Vinatier}, J. Théor. Nombres Bordx. 15, No. 1, 393--410 (2003; Zbl 1048.11088); \textit{S. Vinatier}, Acta Arith. 119, No. 2, 171--186 (2005; Zbl 1075.11071); \textit{E. J. Pickett} and \textit{S. Vinatier}, Compos. Math. 149, No. 7, 1175--1202 (2013; Zbl 1292.11128); \textit{E. J. Pickett} and \textit{L. Thomas}, Int. Math. Res. Not. 2016, No. 11, 3205--3236 (2016; Zbl 1404.11128)]. \textit{W. Bley} et al. [Ann. \(K\)-Theory 5, No. 1, 79--140 (2020; Zbl 1451.11121)] have formulated a conjecture that can be seen as a refinement of the many of the results of aforementioned articles, and is also motivated by a desired to link the theory of \(\mathcal{A}_{L/K}\) with the very general framework of the equivariant Tamagawa number conjecture of [\textit{D. Burns} and \textit{M. Flach} Doc. Math. 6, 501--570 (2001; Zbl 1052.11077)]. We now describe this conjecture in more detail. Let \(K_{0}(\mathbb{Z}[G],\mathbb{Q}[G])\) denote the relative algebraic \(K_{0}\)-group of the ring inclusion \(\mathbb{Z}[G] \subset \mathbb{Q}[G]\) and let \(\partial_{G} : K_{0}(\mathbb{Z}[G],\mathbb{Q}[G]) \rightarrow \mathrm{Cl}(\mathbb{Z}[G])\) denote the (surjective) connecting homomorphism of relative \(K\)-theory. Under the hypothesis that \(L/K\) is weakly ramified and of odd degree, Bley, Burns and Hahn defined certain elements \(\mathfrak{a}_{L/K}\) and \(\mathfrak{c}_{L/K}\) in \(K_{0}(\mathbb{Z}[G],\mathbb{Q}[G])\) and conjectured that they are in fact equal. Here \(\mathfrak{a}_{L/K}\) is a canonical pre-image of \([\mathcal{A}_{L/K}]\) under \(\partial_{G}\) and \(\mathfrak{c}_{L/K}\) is an `idelic twisted unramified characteristic' of \(L/K\). They showed that both elements in fact lie in the torsion subgroup of \(K_{0}(\mathbb{Z}[G],\mathbb{Q}[G])\) and enjoy good functorial properties under change of extension \(L/K\). They also showed that their conjecture holds when \(L/K\) is tamely ramified, and more generally, when every place \(v\) of \(K\) that is wildly ramified in \(L\) has the following three properties: (i) the decomposition group in \(G\) of any place above \(v\) is abelian; (ii) the inertia group in \(G\) of any place of \(L\) above \(v\) is cyclic; and (iii) the completion of \(K\) at \(v\) is absolutely unramified. The main purpose of the article under review is to provide further evidence for the conjecture of Bley, Burns and Hahn by proving the following result. Fix an odd prime \(p\) and suppose that \(L/K\) is a weakly ramified Galois extension of number fields of \(p\)-power degree. Suppose that \(p\) is unramified in \(K/\mathbb{Q}\) and let \(n(L/K)\) denote the maximal order of a decomposition group in \(G\) of a wildly ramified place. Then in \(K_{0}(\mathbb{Z}[G],\mathbb{Q}[G])\) one has \[ \frac{n(L/K)}{p} \cdot \mathfrak{a}_{L/K} = \frac{n(L/K)}{p} \cdot \mathfrak{c}_{L/K} = 0. \] This result is proven by combining the results described in the previous paragraph with a result on the structure of the torsion subgroup of \(K_{0}(\mathbb{Z}[G],\mathbb{Q}[G])\) in the case that \(G\) has prime power order; this latter result can be viewed as a refinement of a well-known result of \textit{S. V. Ullom} [J. Algebra 29, 124--132 (1974; Zbl 0278.20016)] on the kernel subgroup of \(\mathrm{Cl}(\mathbb{Z}[G])\) in this situation.