Transformation invariance in the combinatorial Nullstellensatz and nowhere-zero points of non-singular matrices (Q6091814)

In this paper, the authors describe a kind of transformation invariance in the quantitative combinatorial Nullstellensatz which is frequently used to prove list coloring theorems, illustrated by its usage in a new short proof of \textit{É. Balandraud} and \textit{B. Girard}'s theorem [Combinatorica 34, No. 6, 657--688 (2014; Zbl 1362.11038)] about zero-sum subsums. The transformation invariance is also used to study nowhere-zero points of non-singular matrices \(A \in \mathbb{F}^{n\times n}\), which are points \(x \in \mathbb{F}^{n}\) such that neither \(x\) nor \(Ax\) have zero entries. Utilizing the non-singularity of \(A\) in an elegant way, they give a new proof of \textit{N. Alon} and \textit{M. Tarsi}'s theorem [Combinatorica 9, No. 4, 393--395 (1989; Zbl 0717.05021)] about the existence of nowhere-zero points over fields \(\mathbb{F}\) that are not prime. Afterward, with other methods, the scope of Alon, Tarsi and Jaeger's conjecture was extended from fields to rings. Partially proving this extension, it was shown that over rings that are not fields, every invertible matrix has a nowhere-zero point. Moreover, over the integers modulo \(m\), a non-vanishing determinant suffices to guarantee nowhere-zero points, which was proved for all \(m\) that are not a prime power. Finally, the authors show that the four-color problem can be stated as an existence problem for nowhere-zero points over the field with three elements. More precisely, the four-color theorem is equivalent to the existence of nowhere-zero points of all matrices \(A\) that can be obtained from two planar binary trees that are glued together, to form a 3-regular planar graph. The four-color problem is equivalent to the problem of finding proofs for the existence of nowhere-zero points for those matrices \(A\).