A notion analogous to the discriminant for transcendental elements in certain extensions of local fields (Q776977)

Let \((K,|\; |)\) be a local field, that is, there is \(c\in\; ]0,1[\) and a discrete valuation \(v\) such that \(\forall x\in K\;|x|=c^{v(x)}\) (\(|\; |\) is called a discrete absolute value), and \(K\) is complete with respect to the topology defined by \(|\; |\). Denote by \(\overline{K}\) its separable algebraic closure and \(\Omega=\widetilde{\overline{K}}\) the completion of \(\overline{K}\) (here, \(|\; |\) is assumed to be extended to \(\overline{K}\), then to \(\Omega\)). If \(\alpha \in \overline{K}\), then its discriminant is \(\mathrm{disc}_K(\alpha)=\prod_{\alpha_i\neq\alpha_j}|\alpha_i-\alpha_j|\), where the \(\alpha_i\)'s are the \(\sigma_i(\alpha)\), with \(\sigma_i\in\mathrm{Gal}(K(\alpha)|K)\). This paper attempt to generalize the definition of the discriminant to a transcendental element \(T\) of \(\Omega\). Denote by \(C_K(T)=\{\sigma(T),\; \sigma\in \mathrm{Gal}_{\mathrm{cont}}(\Omega|K)\}\) the orbit of \(T\) under the subgroup of continuous elements of \(\mathrm{Gal}(\Omega|K)\). By properties of local fields, \(C_K(T)\) is compact. Hence for every \(\varepsilon> 0\) there is a finite number \(N(\varepsilon)\) a balls of radius \(\varepsilon\) covering \(C_K(T)\). Again by properties of local fields, this covering is a partition, and every element of a ball is the center of this ball. So, we can denote by \(T=\sigma_1(T),\sigma_2(T),\dots,\sigma_{N(\varepsilon)}(T)\) the centers of these balls. Note that since the \(\sigma_i\)'s are isomorphisms, for \(0<\varepsilon'<\varepsilon\), every ball of radius \(\varepsilon\) is covered by the same number of balls of radius \(\varepsilon'\). Hence \(N(\varepsilon)\) divides \(N(\varepsilon')\). We get a sequence of integers \(1=N_1<N_2<\cdots\) (which is finite if \(T\) is algebraic). Now, for every integer \(n\geq 2\) we set \(x_n=\prod_{i=2}^{N_n}|T-\sigma_i(T)|\). The authors show that the sequence \(\left(x_n^{1/N_n}\right)\) converges in \(\Omega\), and that its limit \(\Delta_K(T)\) is an invariant of \(T\) and of \(C_K(T)\). In the case where \(v\) is a rank \(1\) valuation, for every closed subfield \(L\) of \(\Omega\) containing \(K\) there is a sequence \((\alpha_n)\) of elements of \(L\cap \overline{K}\) which converges to an element \(T\in \Omega\backslash \overline{K}\) and such that \(L\cap \overline{K}= \bigcup_n K(\alpha_n)\), \(L=\widetilde{K(T)}\) and \(\Delta_K(T)\) is the limit of the sequence \(\left(\mathrm{disc}_K(\alpha_n)^{d_n(d_n-1)}\right)\), where \(d_n=[K(\alpha_n):K]\). Let \(K=\mathbb{Q}_p\), \(|\; |_p\) be the \(p\)-adic absolute value (i.e., \(v\) is the \(p\)-adic valuation and \(c=\frac{1}{p}\)), and let \(T\), \(U\) be elements of \(\mathbb{C}_p\backslash \mathbb{Q}_p\) which satisfy \(|T|_p<1\), \(|U|_p<1\) and \(\widetilde{\mathbb{Z}}_p[T]=\widetilde{\mathbb{Z}}_p[U]\). Then we have \(\Delta_{\mathbb{Q}_p}(T)=\Delta_{\mathbb{Q}_p}(U)\), that is, \(\Delta_{\mathbb{Q}_p}(T)\) is an invariant of the closed ring \(\widetilde{\mathbb{Z}}_p(T)\). Finally, the authors provide an example where \(K=\mathbb{Q}_p\), \(|\;|=|\;|_p\) and \(\Delta_K(T)\) is nonzero.