Polynomial gain on symmetric intersecting vector families

Status open medium confidence

The conjecture asks for constants $c, \delta > 0$ (possibly depending on $k$) such that every symmetric intersecting $A \subset [k]^n$ satisfies $\log_k |A| \leq n - cn^{\delta}$, a polynomial improvement over the qualitative $o(k^n)$ bound of Theorem 1.1 for fixed $k \geq 3$. Keller, Lifshitz, and Marcus (arXiv:2307.01356, 2023) prove the first quantitative upper bound in the transitive-symmetric vector-intersecting setting via sharp hypercontractivity, but their result applies only for $k$ growing with $n$ (specifically $2\log n \leq k \leq \sqrt{n}\log n$), leaving the fixed-$k$ case open. No paper resolving the polynomial improvement for fixed $k$ was found in the indexed literature.

Cited literature (1)

• partial Sharp Hypercontractivity for Global Functions (2023)
  Keller, Lifshitz, Marcus · arXiv preprint · arXiv:2307.01356

  Proves the first quantitative upper bound on transitive-symmetric vector-intersecting families (Theorem 1.6): for $k \in [2\log n, \sqrt{n}\log n]$, any such family satisfies $|\mathcal{A}|/k^n \leq c_1 \exp(-c_2 k/\log n)$; this is a polynomial improvement in $n$ for $k$ polynomial in $n$, but does not address the fixed-$k$ regime targeted by the conjecture.

Reviewer notes. The conjecture targets fixed $k \geq 3$ with $n \to \infty$; arXiv:2307.01356 makes partial progress for $k$ growing with $n$ via sharp hypercontractivity (Theorem 1.6), but this regime does not overlap with fixed $k$. The qualitative $o(k^n)$ bound from the source paper may also follow from earlier work of Dinur, Friedgut, and Regev (as noted in the paper's revised version). Author names for 2307.01356 are from PDF metadata extraction and may need independent verification.

Auto-reviewed 2026-05-15 with claude-sonnet-4-6 (web search enabled).