跳转至

文章背景与核心概要

扎兰基维奇问题(Zarankiewicz problem)是极端图论中的核心问题之一,旨在寻找包含特定完全二部子图(如 \(K_{s,t}\))的最大边数。长期以来,确定特定参数下的精确扎兰基维奇数(Zarankiewicz numbers)因计算复杂度和状态空间爆炸而极具挑战。

本文由 Koyar Afrasyab 撰写,聚焦于 \(K_{3,3}\) 配置,通过创新的“证书驱动”(certificate-based)计算机辅助证明方法,成功计算出了多个有限切片和前沿区域的精确值与严格界限。该研究的亮点在于结合了浮点优化(仅用于发现证书)与精确的有理数/整数算术验证(确保结果的绝对严谨性),为图论中的极值问题开辟了高可靠性的计算新路径。


两个有限前沿切片上的扎兰基维奇精确值

作者: Koyar Afrasyab
主要学科: 组合数学 (math.CO)
arXiv ID: 2608.08154
提交时间: 2026年8月8日
代码仓库: GitHub - KAVentures/z1322-exact

Author: Koyar Afrasyab
Primary Subject: Combinatorics (math.CO)
arXiv ID: 2608.08154
Submitted: August 8, 2026
Repository: GitHub - KAVentures/z1322-exact


📌 摘要

本文研究了扎兰基维奇数 \(Z(m, n, s, t)\),其定义为阶数分别为 \(m\)\(n\) 且不包含同构于 \(K_{s,t}\) 的子图的二部图中的最大边数。

📌 Summary

The paper investigates the Zarankiewicz number \(Z(m, n, s, t)\), defined as the maximum number of edges in a bipartite graph with parts of orders \(m\) and \(n\) that contains no subgraph isomorphic to \(K_{s,t}\).

通过采用组合式的、基于证书的计算机辅助证明方法,作者针对 \(K_{3,3}\) 配置,在多个有限切片和前沿区域上确立了精确值与严格的界限。该方法仅依赖浮点优化来发现证书,而所有最终的断言均通过标准库的 Python 工具利用精确的整数和有理数算术进行独立验证。

Using a combined, certificate-based computer-assisted proof approach, the author establishes exact values and rigorous bounds across multiple finite slices and frontier regions for \(K_{3,3}\) configurations. The methodology relies on floating-point optimization solely for discovering certificates, while all final claims are independently verified using exact integer and rational arithmetic via standard-library Python tools.


🔑 核心结果与贡献

  • 有限切片的广义公式: $\(Z(12, n, 3, 3) = 6n \quad \text{当 } 18 \le n \le 22 \text{ 时}\)$
  • 特定切片的精确边数:
  • \(Z(13, 22, 3, 3) = 137\)
  • \(Z(13, 18, 3, 3) = 116\)
  • \(Z(14, 18, 3, 3) = 124\)
  • \(Z(15, 18, 3, 3) = 132\)
  • \(Z(14, 17, 3, 3) = 118\)
  • \(Z(15, 17, 3, 3) = 126\)
  • 严格的区间界限:
  • \(132 \le Z(16, 17, 3, 3) \le 133\)(由于仅对 132 边的下界见证以及已发表的 133 上界进行了形式化认证,故以区间形式报告)。

🔑 Key Results & Contributions

  • Generalized Formula for Finite Slices: $\(Z(12, n, 3, 3) = 6n \quad \text{for } 18 \le n \le 22\)$
  • Exact Edge Counts for Specific Slices:
  • \(Z(13, 22, 3, 3) = 137\)
  • \(Z(13, 18, 3, 3) = 116\)
  • \(Z(14, 18, 3, 3) = 124\)
  • \(Z(15, 18, 3, 3) = 132\)
  • \(Z(14, 17, 3, 3) = 118\)
  • \(Z(15, 17, 3, 3) = 126\)
  • Rigorous Interval Bounds:
  • \(132 \le Z(16, 17, 3, 3) \le 133\) (reported as an interval since only the 132-edge lower witness and the published 133 upper bound are formally certified).

🛠️ 方法论亮点

  • 上界证书: 精确的 \(12 \times 18\)\(13 \times 18\) 证书包确立了支撑性的上界。轨道证书(Orbit certificates)系统性地排除了在下一个边数下的所有假设矩阵。
  • 单元闭包: 删除引理(Deletion lemmas)和显式见证成功封闭了前沿上的四个相邻单元。
  • 针对 \(13 \times 22\) 的简化策略:
  • 将问题简化为 83 个不同的度数分布(degree profiles)。
  • 有理数分离其中 77 个分布。
  • 利用先进技术消除了其余六个分布,包括标记行同余(marked-row congruences)、剩余枚举(leave enumeration)、模 Gram 测试(modular Gram tests)以及精确的法卡斯证书(exact Farkas certificates)。

🛠️ Methodological Highlights

  • Upper Bound Certificates: Exact \(12 \times 18\) and \(13 \times 18\) certificate packages establish load-bearing upper bounds. Orbit certificates systematically exclude every hypothetical matrix at the next edge count.
  • Cell Closure: Deletion lemmas and explicit witnesses successfully close four neighboring cells on the frontier.
  • Reduction Strategy for \(13 \times 22\):
  • Reduces the problem to 83 distinct degree profiles.
  • Rationally separates 77 of these profiles.
  • Eliminates the remaining six profiles using advanced techniques, including marked-row congruences, leave enumeration, modular Gram tests, and exact Farkas certificates.