Dp-finite fields. I(B): Positive characteristic (Q2662697)

This paper is the second part and a direct continuation of the author's paper [ibid. 172, No. 6, Article ID 102947, 39 p. (2021; Zbl 1473.03018)], which -- in turn -- is a vast generalisation of the author's proof of Shelah's conjecture for dp-minimal fields [J. Math Log. 18 No. 2, Article ID 1850007, 23 p. (2018; Zbl 06996241)]. Recall that Shelah's conjecture states (in one of its equivalent formulations) that any infinite dependent (i.e., NIP) field is either real closed, separably closed or supports a definable non-trivial valuation. In the present paper, the author proves the conjecture for dp-finite fields of positive characteristic. The main technical result of the paper is that any saturated enough unstable field of finite dp-rank admits an invariant (over a small set of parameters) non-trivial valuation ring. This suffices to prove Shelah's conjecture in positive characteristic using results from the first part of the paper. Thus, the paper focuses on the construction of non-trivial invariant valuation rings. The construction is lattice theoretic in nature, using techniques by \textit{E. R. Puczyłowski} [in: International symposium on ring theory. Proceedings of the 3rd Korea-China-Japan international symposium, held jointly with the 2nd Korea-Japan joint ring theory seminar, Kyongju, Korea, June 28 -- July 3, 1999. Boston, MA: Birkhäuser. 303--312 (2001; Zbl 1018.06007)]. Non-triviality of the construction is assured by results of Part (A) of the paper [the author, loc. cit.], specifically, the fact that the group of infinitesimals is a non-zero proper linear subspace (of the monster model, over the given model). The construction also needs some study of dp-finite abelian groups, e.g., Theorem 10.4 states that if \(H\) is a type-definable subgroup of a dp-finite abelian group \(G\), then \(|H/H^{00}|\) is bounded by a cardinal \(\kappa(G)\) depending only on \(G\). Reviewer's remark: In subsequent work the author gives a complete proof of Shellah's conjecture for dp-finite fields. This proof builds, rather, on the group of infinitesimals constructed in Part (A) of the paper [the author, loc. cit.], and avoids much of the machinery developed in the present paper.