Henselian elements (Q406347)

First recall some definitions. For an extension \(A\subseteq B\) of rings with unity, an element \(b\) of \(B\) such that \(h(b)=0\) and \(h'(b)\) is a unit of \(B\), for some polynomial \(h(X)\) in \(A[X]\), is called a Henselian element over \(A\). A local ring is regular if its maximal ideal is generated by \(d\) elements, where \(d\) is the maximal lenght of a chain of prime ideals. A valued function field \((F|K, v)\) is said to admit local uniformization if, for every finite subset \(Z\) of the valuation ring \({\mathcal O}_F\) of \(F\), there exists a subring \(R\) of \(F\), of finite type over \(K\), such that the field of fractions \(\text{frac}(R)\) of \(R\) is equal to \(F\), \(R_{{\mathfrak p}}\) is regular (where \(\{{\mathfrak p}= \{x\in R\mid v(x)> 0\}\)) and \(Z\subseteq R\). Let \(A\) be an integrally closed domain with quotient field \(L\), \(I(A)\) be the integral closure of \(A\) in the separable-algebraic closure \(L^{\text{sep}}\) of \(L\), and \({\mathfrak m}\) be a maximal ideal of \(I(A)\). The inertia field of \(L^{\text{sep}}|L\) with relation to \({\mathfrak m}\) is the subfield \(L^i\) of \(L^{\text{sep}}\) fixed by the group \(\{\sigma\in \text{Gal}(L^{\text{sep}}|L)\mid\forall x\in I(A): x- \sigma(x)\in F{\mathfrak m}\}\); we know that if \(A\) is the valuation ring of a valuation \(v\), then \(v(L^i)= v(L)\) and if \(L\subseteq F\subseteq L^i\) then the extension \((F|L, v)\) is said to be unramified. Now let \(F\subseteq L^i\) such that \(F|L\) is a finite extension, \(A^*= F\cap I(A)\) and \(B= A^*_{{\mathfrak m}\cap A^*}\). First the authors prove that there is \(\eta\in B\) such that \(F= L(\eta)\), the minimal polynomial \(h(\eta)\) of \(\eta\) over \(L\) lies in \(A[X]\) and \(\eta\) and \(h'(\eta)\) are units of B. Next they prove that \(B= A[\eta]_{{\mathfrak n}}\), where \({\mathfrak n}= A[\eta]\cap{\mathfrak m}\). In particular, for every finite \(Z\subseteq B\) there exists a unit \(u\) of \(B\) in \(A[\eta]\) such that \(Z\subseteq A[\eta,1/\mu]\). With the extra assumption that \(A\) is a valuation ring, for every \(b\in B\) there exist Henselian elements \(r\), \(s\) in \(B\) such that \(b\in A[\eta,r,s]\). They prove that there exists a finite valued field extension \((F|L,v)\), where \(F\subseteq L^i\), such that \({\mathcal O}_F\) is not a finitely generated \({\mathcal O}_L\) algebra. They give a list of equivalent conditions under which \({\mathcal O}_F\) is a finitely generated \({\mathcal O}_L\) algebra. For example, both of: there exists a unit \(u\) of \(B\) in \({\mathcal O}_L[\eta]\) such that \({\mathcal O}_F={\mathcal O}_L[\eta,1/u]\), and: there are Henselian elements \(r\), \(s\) in \({\mathcal O}_F\) such that \({\mathcal O}_F={\mathcal O}_L[\eta,r,s]\). The other conditions are related to \(\text{Spec}({\mathcal O}_L[\eta])\), and they hold if \(\text{Spec}({\mathcal O}_L)\) is well-ordered by inclusion. Now, let \((F|K, v)\) be a valued function field. If there exists a transcendence basis \(T\) of \(F|K\) such that \((K(T)|K,v)\) admits local uniformization, \(F\subseteq K(T)^i\) and \(v\) is trivial on \(K\), then \((F|K, v)\) admits local uniformization. In the appendix, they prove that if \(R\) is an integrally closed domain and \(R\subseteq S\) is an integral extension such that \(\text{frac}(S)|\text{frac}(R)\) is finite and \(\text{Spec}(R)\) is the union of finitely many chains, then \(\text{Spec}(S)\) is the union of finitely many chains.