Extending binary linear codes to self-orthogonal codes

TL;DR

This paper introduces a new method for extending binary linear codes to self-orthogonal codes using a special matrix derived from Reed-Muller codes, leading to the construction of many optimal codes and partial refutation of a previous conjecture.

Contribution

It proposes a novel extension technique utilizing the self-orthogonality matrix from Reed-Muller codes, enabling the creation of numerous optimal self-orthogonal codes of dimensions 5 and 6.

Findings

Constructed many optimal self-orthogonal codes of dimensions 5 and 6.

Partially disproved the conjecture by Kim et al. (2021) for certain code lengths.

Identified conditions under which optimal self-orthogonal codes exist for various lengths.

Abstract

Kim et al. (2021) gave a method to embed a given binary $[n,k]$ code $\mathcal{C}$ $(k = 3, 4)$ into a self-orthogonal code of the shortest length which has the same dimension $k$ and minimum distance $d' \ge d(\mathcal{C})$. We extend this result by proposing a new method related to a special matrix, called the self-orthogonality matrix $SO_k$, obtained by shortening a Reed-Muller code $\mathcal R(2,k)$. Using this approach, we can extend binary linear codes to many optimal self-orthogonal codes of dimensions $5$ and $6$. Furthermore, we partially disprove the conjecture (Kim et al. (2021)) by showing that if $31 \le n \le 256$ and $n\equiv 14,22,29 \pmod{31}$, then there exist optimal $[n,5]$ codes which are self-orthogonal. We also construct optimal self-orthogonal $[n,6]$ codes when $41 \le n \le 256$ satisfies $n \ne 46, 54, 61$ and $n \not\equiv 7, 14, 22, 29, 38, 45, 53, 60 \pmod{63}$.