7
$\begingroup$
Definition. According to Shelah, a field $K$ does not have the independence property (i.e. is NIP) if for every first order formula $\varphi(x, \bar y)$ in the language of fields $(+,\times,0,1)$, the Vapnik–Chervonenkis dimension of the family of subsets$\{\varphi(K, \bar k) : \bar k \in K^n\}$ is finite, i.e. there do not exist arbitrarily large finite sets $A \subset K$ whose subsets are all of the form $A \cap \varphi(K, \bar k)$ for some $\bar k$ from $K^n$.

Examples of NIP fields include, in characteristic $p$ finite fields (easy), $\mathbf F^{alg}$, and in zero characteristic, $\mathbf C$, $\mathbf R$ and $\mathbf Q_p$.

A Theorem of I. Kaplan asserts that a NIP field has no Artin-Schreier extension. I found

Theorem (I. Kaplan, F. Wagner and T. Scanlon). A valued field of characteristic $p$ with perfect infinite NIP residue field, with $p$-divisible value group and which is algebraically maximal ($i.e.$ with no proper algebraic valued extension having both same residue field and same valued group) is NIP.

I am not familiar with valuation theory, so the above is of little help to me, but for those who know,

Question: would you have any concrete examples of NIP fields of characteristic $p>0$ ? (non separably closed)

I am looking for one which has a cyclic (Galois) extension.

$\endgroup$
2

2 Answers 2

4
$\begingroup$

Understanding concretely which fields (in the "pure" language of rings, $\mathcal{L} = \{0, 1, +, \cdot\}$) are NIP is a topic of current interest in model theory. The 2015 paper "Dp-minimal valued fields" by Jahnke, Simon, and Walsberg begins:

"Very little is known about NIP fields. It is widely believed that an NIP field is either real closed, separably closed or admits a definable henselian valuation. Note that even the stable case of this conjecture is open."

(http://arxiv.org/pdf/1507.03911v1.pdf)

So conjecturally, you may have to become familiar with a little valuation theory even to understand some interesting concrete examples.

To complement the 2/6/16 answer by Drike, we do have a fairly good idea of which fields are dp-minimal, where dp-minimal theories are a subclass of NIP theories in which one-variable definable sets are "indecomposable" in a certain sense (they generalize strongly minimal stable theories). A remarkable characterization of dp-minimal theories was given in 2015 by Will Johnson, and the answer involves a nontrivial amount of valued field theory:

"On dp-minimal fields," Will Johnson, http://arxiv.org/pdf/1507.02745v1.pdf

From that paper: "[the main theorem] almost says that all dp-minimal fields are elementarily equivalent to ones of the form $K((t^\Gamma))$ [AKA Malc'ev-Neumann fields] where $K$ is $\mathbb{F}^{alg}_p$ or a characteristic $0$ local field, and $\Gamma$ satisfies some divisibility conditions. The one exceptional case is the mixed characteristic case, which includes fields such as the spherical completion of $\mathbb{Z}^{un}_p(p^{1/{p^\infty}})$.''

$\endgroup$
6
$\begingroup$

For any field $K$ and ordered group $\Gamma$, the Malc'ev-Neumann field $K((\Gamma))$ is maximal by a Theorem of Krull, hence algebraically maximal.

Taking a perfect infinite NIP field $K$, for instance $\mathbf F_p^{alg}$ and a non-divisible $p$-divisible ordered group $\Gamma$, for instance the subgroup $\langle \frac{1}{p^i}:i\in\mathbf N\rangle\subset\mathbf R^+$ provides an example of a non separably closed infinite NIP field.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.