On ramification in transcendental extensions of local fields (Q1682065)

Let \(F\) be a field, \(F _{\mathrm{sep}}\) its separable closure, \(\mathcal{G}_F := \mathcal{G}(F_{\mathrm{sep}}/F)\) the absolute Galois group of \(F\), and \(H ^1(F, \mathbb{Q}/\mathbb{Z})\) the (continuous) character group of \(\mathcal{G} _F\); in what follows the elements of \(H ^1(F, \mathbb{Q}/\mathbb{Z})\) are called abelian characters. The paper under review is devoted to the study of important aspects of the ramification properties of extensions \(L/K\) of complete discrete valued fields, which are encoded in the notion of a Swan conductor Sw\(\chi \) (a suitably defined integer \(\ge 0\)), for abelian characters \(\chi : \mathcal{G}(L/K) \to \mathbb{Q}/\mathbb{Z}\), and are reflected in the behaviour of \(\operatorname{Sw} \chi \) under the natural homomorphism of \(H^1(K, \mathbb{Q}/\mathbb{Z})\) into \(H ^1(L, \mathbb{Q}/\mathbb{Z})\). It is classically known (see \textit{J.-P. Serre} [Local fields. New York, NY: Springer (1979; Zbl 0423.12016)]): that if \(L/K\) is a finite Galois extension, the residue field \(\widehat K\) of \(K\) is perfect, \(\chi \in H ^1(K, \mathbb{Q}/\mathbb{Z})\) and \(\chi_L\) is its image in \(H^1(L, \mathbb{Q}/\mathbb{Z})\), then \[ \operatorname{Sw}\chi_L = \psi_{L/K}(\operatorname{Sw}\chi ) = e(L/K)\operatorname{Sw}\chi - D_{L/K}^{\log}, \tag{1.1} \] provided that \(\operatorname{Sw}\chi \) is sufficiently ramified (notation \(\operatorname{Sw}\chi > 0\)), where \(\psi_{L/K}\) is the classical \(\psi \)-function, \(e(L/K)\) is the ramification index, \(D_{L/K}\) is the different, and \(D_{L/K}^{\log} = D_{L/K} - e(L/K) + 1\) is the wild different. The basis of the present research is also formed by the following major contributions concerning the case where \(\widehat K\) is no longer assumed to be perfect: \textit{K. Kato}'s generalization of the Swan conductor \(\operatorname{Sw}\chi \), for abelian characters \(\chi \colon \mathcal{G}(L/K) \to \mathbb{Q}/\mathbb{Z}\) (see his paper in: [Contemp. Math. 83, 101--131 (1989; Zbl 0716.12006)]), and the generalization of the upper ramification filtration by \textit{A. Abbes} and \textit{T. Saito} [Am. J. Math. 124, No. 5, 879--920 (2002; Zbl 1084.11064)]. The main results of the paper under review are stated as three theorems. The former two theorems provide a formula resembling (1.1) (for Kato's Swan conductor), for possibly transcendental extensions \(L/K\), such that \(\widehat K\) is perfect but \(\widehat L\) is not necessarily perfect. They present separately the cases where \(\operatorname{char}(K) = (\widehat K) > 0\) and where \(\operatorname{char}(K) = 0 \neq \operatorname{char}(\widehat K)\), respectively. The method of proving the formula in the former case differs substantially from the one applied to the mixed-characteristic case. Specifically, in the former case, the author uses the Artin-Schreier-Witt theory; in the mixed-characteristic case, she uses \textit{M. Kurihara}'s exponential map [J. Reine Angew. Math. 498, 201--221 (1998; Zbl 0909.19001)] and a modified version of higher dimensional local class field theory. After proving the formula for Kato's Swan conductor, the author associates with \(L/K\) two \(\psi \)-functions in a real variable \(t \ge 0\) (with non-negative values), assuming as above only that \(\widehat K\) is perfect. Then she shows that the defined functions coincide with the classical \(\psi _{L/K}\) function, in the case where \(L/K\) is a finite extension. The third main theorem of the paper under review shows that its first two main results can be regarded as formulas for one of the defined \(\psi \)-functions, when \(t\) is sufficiently large.