Elementary equivalence versus isomorphism. II (Q1686477)

The elementary equivalence versus isomorphism problem asks the following : given any two finitely generated fields, does elementary equivalence between them in the language of rings (i.e. they satisfy exactly the same first-order sentences) imply that they are isomorphic? Recall the definition of the Kronecker dimension of a field: \(\mathrm{krdim}(\mathbb F)=0\) if \(\mathbb F\) is a finite prime field, \(\mathrm{krdim}(\mathbb Q) =1\), and if \(k\) is the prime field of a field \(K\) then \(\mathrm{krdim}(K)= \mathrm{krdim}(k) + \mathrm{trdeg}(K/k)\), where \(\mathrm{trdeg}\) denotes transcendence degree. In a previous paper on this same problem [Inv. Math. 150, 1011--1046 (2002; Zbl 1162.12302)], the author showed, amongst other things, that for every integer \(d\geq 0\) there exists a first-order sentence \(\varphi_d\) in the language of rings such that for each finitely generated field \(K\), we have \(\mathrm{krdim}(K)=d\) iff \(\varphi_d\) holds in \(K\). In this paper, the author gives a positive answer to the above problem for fields of Kronecker dimension strictly less than 3, but in a strong sense. Namely, he proves that for every finitely generated field \(K\) such that \(\mathrm{krdim}(K) < 3\), there exists a first-order sentence \(\vartheta_K\) in the language of rings such that if \(L\) is a finitely generated field and \(\vartheta_K\) holds in \(L\), then \(L\) is isomorphic to \(K\). This generalizes a result of Rumely for global fields (i.e. fields of Kronecker dimension 1; [\textit{R. S. Rumely}, Trans. Am. Math. Soc. 262, 195--217 (1980; Zbl 0472.03010)]). The proof relies on strong auxiliary uniform definability results in finitetly generated fields, in particular concerning valuations of function fields of curves over global fields. More recently, the author has announced the result for Kronecker dimension equal to 3 and characteristic different from 2, and some further results in characteristic zero.