Why is this valuation real (i.e., rank one)?

66 Views Asked by At

(skip to the part about theorem 4.8 if you know things about valuation theory on function fields and local uniformization).

Context:

For rings $R$, denote the quotient field by QF$(R)$. For valuations $v$, let $R_v$ denote the valuation ring, $M_v$ its maximal ideal, $D_v:= R_v/M_v$ the residue field, and $S_v$ the value group. If you are not familiar with quadratic transforms, you can probably ignore that part (I think my question is a basic valuation theoretic one).

Definition 3.10 ([Ramification Theoretic Methods in Algebraic Geometry - Shreeram Abhyanker], or briefly, [A]). If $(R,M)$ is a local domain so that there is a finite type algebra $A$ over a field $k$ s.t. $R = A_p$, where $p = M\cap A$, then we say that $R$ is $\textit{algebraic with ground field $k$}$ and we let $\text{k-rank}(R):= \text{tr.deg}_k R/M$. We have $\text{k-rank}(R) + \dim(R) = \text{tr.deg}_k \text{QF}(R)$.

Theorem 4.8 [A]. Let $L/k$ be a two-dimensional algebraic function field, and $K$ a finite algebraic extension of $L$. Let $v$ be a zero dimensional valuation of $K$ over $k$ (this means that $v\mid_k=0$ and tr.deg$_k D_v=0$). Let $(R,M)$ be a regular algebraic local domain with quotient field $K$ and ground field $k$ s.t.$^1$ $\dim(R) = 2$. Then for some quadratic transform $R^*$ of $R$ along $v$ there exists an algebraic local domain $S^*$ in $L$ with ground field $k$ lying below $R^*$.

(1: he also says: "k-rank$(R) = 2$" but I am fairly certain that is a typo, otherwise 3.10 together with our hypothesis would yield: $4 = 2 + 2 = \text{k-rank}(R) + \dim(R) = \text{tr.deg}_k K = \text{tr.deg}_k L + \text{tr.deg}_L K = 2 + 0 = 2$).

(I am not copying the proof here because I will state (have stated) below (above, resp.) everything that I think has a chance of being relevant to my question).

Question: Relatively early in the proof, Abhyanker uses the supposed fact of $v$ being a rank one valuation, but this is really not clear to me after a good several hours of contemplating.

Notes (what I tried so far): [A] has a characterization of valuations on a two dimensional function field (of which v is: since tr.deg$K/k = 2$), and by this characterization (and using some comments from Dale Cutkosky's $\textit{Resolution of Singularities}$ book (call this [R]) section 8.1) either we get what we want ($v$ has rank one, i.e. $S_v\leq \mathbb{R}$ as ordered abelian groups) or $S_v = (\mathbb{Z}^2)_{\text{lex}}$, but I really cant seem to find an issue with that being the case (I tried using 8.1.2 in [R], noting that, in the latter case, $v$ would be a zero dimensional valuation of rank two).