Projective tilings and full-rank perfect codes
Denis S. Krotov
TL;DR
This work investigates tilings of vector spaces and their connection to $1$-perfect codes over finite fields, introducing semiprojective and projective tilings and proving new full-rank constructions. It provides explicit methods to realize full-rank tilings with projective components for various dimensions, including a ternary $1$-perfect code of length $13$, and analyzes how these tilings yield full-rank codes with controlled kernel dimensions. A key contribution is the demonstration that projective tilings correspond to factorizations of projective spaces, enabling a geometric lens on tilings and codes and raising questions about tile sizes and existence of full-rank factorizations. Overall, the results advance the theory of nonbinary perfect codes, expand the catalog of explicit constructions, and connect tiling theory with projective geometry.
Abstract
A tiling of a vector space $S$ is the pair $(U,V)$ of its subsets such that every vector in $S$ is uniquely represented as the sum of a vector from $U$ and a vector from $V$. A tiling is connected to a perfect codes if one of the sets, say $U$, is projective, i.e., the union of one-dimensional subspaces of $S$. A tiling $(U,V)$ is full-rank if the affine span of each of $U$, $V$ is $S$. For finite non-binary vector spaces of dimension at least $6$ (at least $10$), we construct full-rank tilings $(U,V)$ with projective $U$ (both $U$ and $V$, respectively). In particular, that construction gives a full-rank ternary $1$-perfect code of length $13$, solving a known problem. We also discuss the treatment of tilings with projective components as factorizations of projective spaces. Keywords: perfect codes, tilings, group factorization, full-rank tilings, projective geometry
