超平面覆盖的稳定性
Stability for hyperplane covers
AI总结:
该研究刻画了超立方体几乎k-覆盖中最大权重超平面的结构并给出稳定性界,借此缩小整数线性规划的搜索空间,确定了Q^6几乎k-覆盖的精确最小规模,还改进了高维下的Clifton–Huang下界。
摘要:
超立方体$Q^n = \{0,1\}^n$的几乎$k$-覆盖是指一组避开原点、且将其余每个顶点至少覆盖$k$次的超平面集合。当$k$相对于维度$n$较大时,Clifton和Huang渐近确定了几乎$k$-覆盖的最小可能规模,其证明的核心是LYM不等式的一个推广,涉及超平面的加权计数。\n 本文完全刻画了具有最大权重的超平面,证明这类超平面的数量为$\inom{2n-1}{n}$。我们进一步给出了稳定性结果,对所有非最大权重超平面的权重进行了界的估计。这些结果使得我们在使用整数线性规划构造小规模覆盖时,能够有效缩小搜索空间,借此我们确定了对于大多数$k$值,$Q^6$的几乎$k$-覆盖的精确最小规模。我们还利用该稳定性结果,在每个足够大的维度$n$中,对无穷多组$k$的取值改进了Clifton--Huang下界。
英文摘要:
An almost $k$-cover of the hypercube $Q^n = \{0,1\}^n$ is a collection of hyperplanes that avoids the origin and covers every other vertex at least $k$ times. When $k$ is large with respect to the dimension $n$, Clifton and Huang asymptotically determined the minimum possible size of an almost $k$-cover. Central to their proof was an extension of the LYM inequality, concerning a weighted count of hyperplanes. In this paper we completely characterise the hyperplanes of maximum weight, showing that there are $\binom{2n-1}{n}$ such planes. We further provide stability, bounding the weight of all hyperplanes that are not of maximum weight. These results allow us to effectively shrink the search space when using integer linear programming to construct small covers, and as a result we are able to determine the exact minimum size of an almost $k$-cover of $Q^6$ for most values of $k$. We further use the stability result to improve the Clifton--Huang lower bound for infinitely many choices of $k$ in every sufficiently large dimension $n$.
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。