Topology and geometry of the Berkovich ramification locus for rational functions. II (Q2376893)

This paper builds upon earlier work of the author in [\textit{X. Faber}, ``Topology and geometry of the Berkovich ramification locus for rational functions'', Preprint, \url{arXiv:1102.1432v3}] on understanding the Berkovich ramification locus for a rational map \(\varphi: \mathbb{P}^1_{\mathrm{Berk}} \to \mathbb{P}^1_{\mathrm{Berk}}\) over a non-archimedean field \(k\). This locus is the subset of the source \(\mathbb{P}^1_{\mathrm{Berk}}\) where \(f\) is not locally injective. Give \(\mathbb{P}^1_{\mathrm{Berk}}\) a metric in the standard way. Let \(C \subseteq \mathbb{P}^1_{\mathrm{Berk}}\) be the hull of the critical locus of \(\varphi\). The paper has three main theorems. Loosely stated, they are: (1) If \(k\) has characteristic zero, then a point in the ramification locus is not further than \(1/(p-1)\) away from \(C\) (and lies on \(C\) if the residue characteristic is zero or if \(\deg(\varphi) < p\)). (2) If \(k\) has characteristic \(p\), then a theorem of type (1) holds if and only if \(\varphi\) is tamely ramified. (3) Given a small enough Berkovich neighborhood \(U\) of a critical point, the intersection of \(U\) with the Berkovich ramification locus contains the intersection of \(U\) with \(C\) (the precise containment relation is given, and depends on the multiplicity of the point and the characteristic of \(k\)). The proof proceeds by explicit methods involving Newton polygons of polynomials. In particular, an explicit function (called the visible ramification) is defined, which essentially measures, for a point in the Berkovich ramification locus and a direction away from critical points of \(\varphi\), how far one is from the border of the Berkovich ramification locus. This function is defined in two different ways, and careful study of it leads to the theorems above. The theorems are applied to give, among other things, a non-Archimedean version of Rolle's theorem.