Counting versions for restricted and somewhat-restricted 3-term arithmetic progressions

Investigate the counting versions of restricted and somewhat-restricted 3-term arithmetic progression problems in F_p^n by determining whether a subset A ⊆ F_p^n of constant density δ > 0 must contain a constant (2^{Ω(1)}) fraction of all somewhat-restricted 3-term arithmetic progressions and a constant (2^{Ω(1)}) fraction of all restricted 3-term arithmetic progressions.

Background

The paper considers restricted variants of 3-term arithmetic progressions in finite fields, including ‘restricted’ (a ∈ {0,1}n \ {0}) and ‘somewhat-restricted’ (a ∈ {0,1,2}n \ {0}) patterns. While density upper bounds have been obtained using inverse-correlation techniques (Theorems 5.2 and 5.3), the counting versions—establishing that dense sets contain a constant fraction of such progressions—remain unresolved.

These counting questions interface with product-space models of correlation and could require new analytic or combinatorial tools beyond current inverse theorems.

References

It also makes sense to consider the counting versions of these problems: if A has density & > 0 (thought of as a constant), must it contain 22%(1) fraction of the somewhat-restricted 3-APs? Must it contain 22@(1) fraction of the restricted 3-APs? The counting versions, for which the density Hales-Jewett does not apply, have been raised in [40] and in [35] and remain completely open.

The Lens of Abelian Embeddings  (2602.22183 - Minzer, 25 Feb 2026) in Section 5.1.1