Definable (p,q)-conjecture
Let be a first-order theory, let be an NIP formula, and let be the VC-dimension of the family . For all integers , there should exist an integer such that, in every model , whenever is finite and every -element subset contains a -element subset with , there is a finite set with such that for every .
References
- Local definable (p,q)-theorem and a refinement — arXiv — Gareth John Boxall, Charlotte Kestner, Tsinjo Rakotonarivo
Source: arXiv
Progress summary
An unrefereed 2026 preprint claims to settle the definability conjecture for all NIP theories, but the claim has not been independently checked.
The conjecture asks for a definable form of the classical -theorem in NIP theories. Earlier work established important special cases, while a new preprint claims the full result together with a refinement.
Known results
- Kaplan, 2022: proved a definable and uniform version for NIP theories, according to the preprint record.
- Boxall and Kestner: proved the conjecture in distal NIP theories.
- Simon: proved cases in dp-minimal theories, including formulas with , and cases involving invariant types of dp-rank .
- Andújar Guerrero, 2023: proved a definable result under VC-codensity below , not the full conjecture.
October 2026 claimed extension
Gareth John Boxall, Charlotte Kestner, and Tsinjo Rakotonarivo claim that the definable theorem extends to arbitrary first-order theories satisfying NIP, with an additional refinement. The retrieved evidence is an unrefereed preprint and supplies no independent mathematical verification.
Current status (as of October 2026): The conjecture is claimed solved for NIP theories by the 2026 preprint, but that claim remains unverified; earlier partial cases are established in the cited literature.
Solutions 0
No solutions have been posted yet.