Small weight codewords of projective geometric codes II

TL;DR

This paper proves that for certain projective geometric codes over composite prime powers, all codewords below a specific weight threshold can be expressed as linear combinations of a limited number of rows, extending previous results.

Contribution

It establishes that for q ≥ 32, all codewords of weight up to O(q^k√q) are linear combinations of at most √q rows, generalizing prior findings to more complex codes.

Findings

Codewords of weight ≤ O(q^k√q) are linear combinations of ≤ √q rows.

Results apply to codes over composite prime powers q ≥ 32.

Generalization to codes involving j-spaces and k-spaces.

Abstract

The $p$-ary linear code $\mathcal C_{k}(n,q)$ is defined as the row space of the incidence matrix $A$ of $k$-spaces and points of $\text{PG}(n,q)$. It is known that if $q$ is square, a codeword of weight $q^k\sqrt{q}+\mathcal O \left( q^{k-1} \right) $ exists that cannot be written as a linear combination of at most $\sqrt{q}$ rows of $A$. Over the past few decades, researchers have put a lot of effort towards proving that any codeword of smaller weight does meet this property. We show that if $ q \geqslant 32 $ is a composite prime power, every codeword of $\mathcal C_k(n,q)$ up to weight $\mathcal O \left( {q^k\sqrt{q}} \right) $ is a linear combination of at most $\sqrt{q}$ rows of $A$. We also generalise this result to the codes $\mathcal C_{j,k}(n,q) $, which are defined as the $p$-ary row span of the incidence matrix of $k$-spaces and $j$-spaces, $j < k$.