Irredundant hyperplane covers

TL;DR

This paper proves a tight bound on the size of irredundant coset covers in abelian groups, resolving longstanding conjectures, and improves bounds for hyperplane covers in large prime fields, with applications to the Alon-Jaeger-Tarsi conjecture.

Contribution

It establishes the optimal bound for irredundant coset covers in abelian groups and enhances bounds for hyperplane covers in large prime fields, leading to a strengthened version of the Alon-Jaeger-Tarsi conjecture.

Findings

Bound |G : intersection of H_i| = 2^{O(k)} is tight and resolves conjectures.

Improved bounds for hyperplanes in elementary p-groups with many repetitions.

Strengthened Alon-Jaeger-Tarsi conjecture for large prime fields.

Abstract

We prove that if $G$ is an abelian group and $H_1x_1,\dots,H_{k}x_k$ is an irredundant (minimal) cover of $G$ with cosets, then $$|G:\bigcap_{i=1}^{k}H_{i}|=2^{O(k)}.$$ This bound is the best possible up to the constant hidden in the $O(\cdot)$ notation, and it resolves conjectures of Pyber (1996) and Szegedy (2007). We further show that if $G$ is an elementary $p$-group for some large prime $p$, and $H_1,\dots,H_k$ is a sequence of hyperplanes with many repetitions, then the bound above can be improved. As a consequence, we establish a substantial strengthening of the recently solved Alon-Jaeger-Tarsi conjecture: there exists $\alpha>0$ such that for every invertible matrix $M\in\mathbb{F}_p^{n\times n}$ and any set of at most $p^{\alpha}$ forbidden coordinates, one can find a vector $x\in\mathbb{F}_p^{n}$ such that neither $x$ nor $Mx$ have a forbidden coordinate.