Linear Programming Bounds for LCD Codes via Gauss Phases

2026-09-08

AI Summary

该研究通过将线性码的权重枚举器的根单位值转换为线性约束，并结合Gauss相位线性规划，来改进LCD码的Hamming界。

Abstract

For $q\in\set{2,3}$, we show that a $k$-dimensional linear code over the finite field $\F_q$ of order $q$ is linear complementary dual (LCD) exactly when one root-of-unity value of its weight enumerator has magnitude $q^{k/2}$. We convert the phase of this value, together with the parity type in the binary case, into exact linear constraints on the weight distribution and incorporate them into a Gauss-phase linear program. The resulting program uses only the ordinary weight distributions of the code and its dual and adds only a constant-size set of branch equations to the usual Hamming/MacWilliams constraints, so it remains close in size to the standard Hamming LP while retaining additional arithmetic information. Computations over the audited binary and ternary ranges show systematic strengthening of the Hamming LCD relaxation. In the binary case, comparison with the established mixed joint-weight-enumerator LP yields four strict improvements, lowering the benchmark upper bound by one in each case. Each strict comparison is verified exactly by rational feasibility witnesses and integer Farkas certificates.

Problem

• Linear Complementary Dual (LCD) Codes
• Gauss Phase Linear Programming
• Weight Enumerator
• Hamming LP
• Upper Bound

Innovation

• Gauss-phase linear program
• Linear Complementary Dual (LCD) Codes
• Weight Enumerator
• Hamming/MacWilliams constraints
• Rational feasibility witnesses