arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2610.33976math.CO
收藏

所有满足4≤n≤108且n≡0 mod 4的n,以及所有1≤k≤n的显式 weighing 矩阵W(n,k)

Explicit weighing matrices $W(n,k)$ for all $n \equiv 0 \pmod 4$ with $4 \le n \le 108$, and all $1 \le k \le n$

Adrián Macías-Quintero

AI总结:

本研究构造了1512个满足4≤n≤108且n≡0 mod 4的显式weighing矩阵,涵盖所有1≤k≤n的权值,部分通过经典方法、部分通过穷举或模拟退火得到,还给出了奇偶性障碍及未解决参数矩阵的生成序列。

摘要:

我们针对所有满足4≤n≤108且n≡0 mod 4的阶数n,以及所有1≤k≤n的权值k,给出了显式 weighing 矩阵W(n,k),总计1512个矩阵,每个均通过精确整数算术校验。其中961个来自经典构造(包括单位矩阵、直和、克罗内克积、佩利矩阵及斜阿达马加倍),551个为戈瑟尔斯-赛德尔矩阵,通过穷举搜索(24个矩阵,n≤24)或基于周期自相关代价的模拟退火算法(527个矩阵,n≥28)得到。本注就所用方法而言是自包含的:它给出了所有使用的构造方法及在其中进行选择的算法,逐步解释该算法,证明构造所依赖的每个恒等式,并从目录中选取一个矩阵来说明每种方法。它未复现这1512个矩阵,这些矩阵可在仓库中获取。它还给出了针对m为偶数时戈瑟尔斯-赛德尔矩阵W(4m,4m-1)的一个简短奇偶性障碍,并列出了13个矩阵的生成序列,这些矩阵的参数在作者查阅的历史表格中显示为未解决;它们在当前文献中的状态仍在核查,且不主张优先权。

英文摘要:

We present an explicit weighing matrix $W(n,k)$ for every order $n\equiv 0\pmod 4$ with $4\le n\le 108$ and every weight $1\le k\le n$: 1512 matrices in total, each checked in exact integer arithmetic. Of these, 961 come from classical constructions (identity, direct sums, Kronecker products, Paley matrices and skew-Hadamard doubling) and 551 are Goethals-Seidel matrices, found by exhaustive search (24 matrices, $n\le 24$) or by simulated annealing on the periodic autocorrelation cost (527 matrices, $n\ge 28$). The note is self-contained as far as the methods are concerned: it presents all the construction methods used and the algorithm that selects among them, explains the algorithm step by step, proves every identity on which the constructions rely, and illustrates each method with a matrix taken from the catalogue. It does not reproduce the 1512 matrices, which are available in the repository. It also gives a short parity obstruction to Goethals-Seidel matrices $W(4m,4m-1)$ with $m$ even, and lists the generating sequences of thirteen matrices whose parameters appear as unresolved in historical tables consulted by the author; their status in the current literature is still being checked, and no claim of priority is made.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑