Geometric anti-concentration for nonzero vectors

Question 1 · arXiv:2106.04894

arXiv Question high confidence— first stated 2022-06-14

Authors: Jacob Fox, Matthew Kwan, Hunter Spink

Status partial high confidence

Question 1 from arXiv:2106.04894 asks, for arbitrary nonzero vectors $a_i$, which geometric constraints on $S \subset \mathbb{R}^d$ yield non-trivial uniform upper bounds on $\Pr(X \in S)$. The source paper itself answers this for o-minimal sets (those definable in an o-minimal structure) not containing a line segment, giving a bound of $n^{-1/2+o(1)}$. A 2025 paper by Grebennikov and Kwan (arXiv:2505.24699) extends this programme substantially, proving that sets in convex position and algebraic varieties of prescribed dimension and degree also yield such bounds, thereby resolving Conjecture 10.1 of Fox-Kwan-Spink as well as a conjecture of Kwan-Sauermann; the general question of characterising all admissible geometric constraints remains an open research direction.

Cited literature (1)

partial Geometric Littlewood–Offord problems via lattice point counting (2025)

Alexandr Grebennikov, Matthew Kwan · arXiv preprint · arXiv:2505.24699

Resolves Conjecture 10.1 of Fox-Kwan-Spink (convex position) and the Kwan-Sauermann conjecture on algebraic varieties, providing new classes of geometric constraints on S that imply anti-concentration bounds of the form asked for in Question 1.

Reviewer notes. Question 1 is an open-ended research programme rather than a single falsifiable conjecture; new classes of admissible geometric constraints continue to be identified. The 2025 Grebennikov-Kwan paper (2505.24699) is the main follow-up and explicitly cites Fox-Kwan-Spink [16] as the source of the resolved conjecture (their Conjecture 10.1). The general characterisation question remains open.

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

Question. Assuming only that each $a_i \neq 0$, what geometric constraints can be imposed on a set $S \subset \mathbb{R}^d$ which imply non-trivial uniform upper bounds on $\Pr(X \in S)$?

Context

The paper introduces a new generalisation of the Littlewood–Offord problem to $\mathbb{R}^d$ in which no assumption is made on the sizes of the nonzero vectors $a_1,\ldots,a_n$, only that they are nonzero. The authors observe that this makes the question much more about the geometry of $S$ than about spatial anti-concentration of $X = a_1\xi_1 + \cdots + a_n\xi_n$, giving it a very different flavour from previous Littlewood–Offord variants.

Source paper

Geometric and o-minimal Littlewood-Offord problems
Jacob Fox, Matthew Kwan, Hunter Spink · 2022-06-14
https://arxiv.org/abs/2106.04894 PDF source