Valuative trees over valued fields (Q2097241)

Given a valuation \(\nu\) on a commutative ring \(A\), it is natural to consider the set \([\nu]\) of all valuations which are equivalent to the given one. The set of these equivalence classes is called the valuative spectrum of \(A\), noted \(\mathrm{Spv}(A)\). Then, given a ring homomorphism \(A \rightarrow B\), a pull-back mapping is defined: \(\mathrm{Spv}(B) \rightarrow \mathrm{Spv}(A)\). For a valued field \((K,v)\), considering the natural embedding \(K \hookrightarrow K[x]\), the equivalence class \([v] \in \mathrm{Spv}(K)\) determines a set \(\mathcal{T}_v \in \mathrm{Spv}(K[x])\) called the valuative tree over \((K,v)\) (the word tree makes perfect sense, although it is not entirely trivial why). This paper deals with the precise description of this object for a generic case. The first part of the paper (Sections 1 to 5) focuses on the case where the valuation groups come correspond to a fixed extension of ordered abelian groups and involves deep technical arguments encompassing many previous results (including previous work from the authors). Finally, in sections 6 and 7 the desired description is reached. In particular, the object \(\mathcal{T}_v\) is described via a bijection with a newly defined object \(\mathcal{T}_{\mathrm{sme}}\), which arises from the so-called small extensions of the valuation group. Summing up, this paper produces a very precise and complicated theorem, with lots of intermediate results, many of which are interesting on their own. The level of sophistication and technical difficulty is quite high, but the paper is clearly written.