Artin-Schreier extensions in NIP and simple fields (Q1758927): Difference between revisions

A classical result of Macintyre from early 70s is that every infinite \(\omega\)-stable field is algebraically closed, and many results establishing properties of fields with certain model-theoretic tameness assumptions had appeared since then. In this paper, the authors consider algebraic properties of fields definable and type-definable in simple and in NIP theories. A typical representative of the first class is a pseudo-finite field, while a real closed or a \(p\)-adically closed field is a typical representative of the second class. Specifically, they deal with Artin-Schreier extensions (recall that, for a field \(K\) of characteristic \(p\), its extension \(L/K\) is Artin-Schreier if \(L=K\left(\alpha\right)\) for some \(\alpha\in L\setminus K\) such that \(\alpha^{p}-\alpha\in K\)). In Section 2, using results from algebraic geometry, it is demonstrated (Corollary 2.9) that for a perfect field \(K\) of characteristic \(p\), given a tuple of algebraically independent elements \(\bar{a}=\left(a_{1},\ldots,a_{n}\right)\) from \(K\) and some large algebraically closed extension \(\mathbf{K}\), the group \(G_{\bar{a}}=\{ \left(t,x_{1},\ldots,x_{n}\right)\in\mathbf{K}^{n+1}:t=a_{i}\left(x_{i}^{p}-x_{i}\right) \text{ for } 1\leq i\leq n \} \) is algebraically isomorphic over \(K\) to \(\left(\mathbf{K},+\right)\). In Section 3 this fact is combined with a type-definable version of Schlichting's theorem, due to Wagner, showing (Theorem 3.2) that any field which is type-definable in a simple theory has only finitely many Artin-Schreier extensions. The result is optimal. In Section 4 it is shown (this time using the Baldwin-Saxl condition on intersections of subgroups in NIP) that every field definable in a NIP theory is Artin-Schreier closed. The type-definable case remains open, but Proposition 4.6 shows that either there are no Artin-Schreier extensions or there are unboundedly many. The final section provides a criterion for a valued field to be NIP. It combines some well-known results from the literature with some corollaries of the previous sections in order to show Theorem 5.9: An algebraically maximal valued field \(\left(K,v\right)\) of characteristic \(p\) with a perfect residue field \(k\) is NIP if and only if \(k\) is NIP and infinite, and the value group is \(p\)-divisible. Recently, some of these results were generalized to a wider class of NTP2 theories, see [\textit{A. Chernikov}, \textit{I. Kaplan} and \textit{P. Simon}, ``Groups and fields with NTP2'', Proc. Am. Math. Soc. (to appear), \url{arXiv:1212.6213}].