首页 > AI前沿 > Goemans's Unsplittable-Flow Conjecture: The Anatomy of AI-Assisted Counterexample Discovery

Goemans's Unsplittable-Flow Conjecture: The Anatomy of AI-Assisted Counterexample Discovery

arXiv机器学习 2026-11-19 13:00 9 阅读 查看原文
用户需要我作为一个专业的新闻排版工程师,将提供的一段纯文本新闻正文排版成语义化的HTML片段。 需要遵循的规则: 1. 只输出HTML片段本身,不要Markdown围栏,不要解释。 2. 每个自然段落用`

`包裹,长文用`

`/`

`划分小节。 3. 列表用`
  • `,引用用`
    `,代码用`
    `,重要句子可用``。
    4. 保留原文所有信息,不得增删改内容。
    5. 不要输出``/``/``等文档级标签。
    
    分析提供的正文:
    "Goemans conjectured that any fractional single-source flow could be converted into an unsplittable flow without increasing total cost and while increasing any arc load by at most the largest demand. Recently, Dmitry Rybin announced a seven-vertex counterexample found through an interaction with ChatGPT, which disproved this long-standing conjecture. However, the interaction did not reveal why this particular topology and its demand, fractional-flow, and cost choices work together, whether the example belongs to a broader family, or which structural features determine where counterexamples can occur. It also left unclear why the AI-assisted search succeeded where earlier human efforts had not. In this paper, we develop a forward counterexample construction pipeline. It starts from an arborescence on which failure is impossible and adds terminal-entering shortcut arcs. The resulting load constraints define a finite set of admissible route choices; their convex hull, the conflict polytope, yields exact conditions for demands, fractional flows, and costs. Starting from a simple arborescence, the pipeline selects Rybin's topology and characterizes all counterexamples it supports. Experiments further demonstrate the pipeline's transferability: when instructed to follow it while varying the initial arborescence and shortcut placement, ChatGPT produced counterexamples on topologies genuinely different from Rybin's and found valid constructions starting from 93 other rooted arborescence shapes. The theory also explains why the original example was difficult to discover through blind enumeration and suggests that Rybin's interaction lies between blind enumeration and a complete theory-guided construction."
    
    这段文字是一段学术新闻或摘要,分为几个自然段落(或者可以划分为几个小节以符合长文的要求)。
    原文实际上是一段连续的文字,但为了符合“长文用

    /

    划分小节”的要求,我可以将其合理划分为几个逻辑小节,或者将其分为几个自然段落。 不过,规则说“每个自然段落用

    包裹,长文用

    /

    划分小节”。原文没有明确的小标题,但逻辑上可以拆分: 小节1:背景与Rybin的反例 小节2:论文提出的构建流水线 小节3:实验与理论解释 但是规则4要求“保留原文所有信息,不得增删改内容”。如果我添加小标题,算不算增删改内容? 为了严格遵守“不得增删改内容”,我不应该凭空捏造原文没有的小标题文字。我可以直接将原文的几句话拆分成几个`

    `,如果必须用`

    `/`

    `,我可以看看原文是否有可以作为小节标题的句子,或者干脆只用`

    `,因为如果原文本身没有明显的章节标题,强行加标题可能会违反“不得增删改内容”的规则。 等等,规则说“长文用

    /

    划分小节”,这可以理解为“如果是长文,请使用h2/h3来划分”,并不是强制要求必须使用,或者我可以把第一句话或者核心主题作为h2/h3?不,添加标题文字属于“增”。 最安全的做法:将原文按句群拆分为几个`

    `,不添加任何原文没有的标题文字。或者,如果一定要用h2/h3,可以把原文中的某些句子作为标题?不,那会改变语义。 其实,原文可以自然分为三个段落(按逻辑): 段落1:Goemans conjectured ... earlier human efforts had not. 段落2:In this paper, we develop ... it supports. 段落3:Experiments further demonstrate ... theory-guided construction. 这样拆分为三个`

    `,完全保留原文,不增删改。 或者,我可以使用`

    `来包裹某些核心论点?不,h2通常用于标题。如果没有标题,就不使用,避免增加内容。 “长文用

    /

    划分小节”是一个条件或建议,如果原文没有小标题,直接用`

    `分段即可。 让我们仔细划分段落,确保不遗漏任何字符。 原文: Goemans conjectured that any fractional single-source flow could be converted into an unsplittable flow without increasing total cost and while increasing any arc load by at most the largest demand. Recently, Dmitry Rybin announced a seven-vertex counterexample found through an interaction with ChatGPT, which disproved this long-standing conjecture. However, the interaction did not reveal why this particular topology and its demand, fractional-flow, and cost choices work together, whether the example belongs to a broader family, or which structural features determine where counterexamples can occur. It also left unclear why the AI-assisted search succeeded where earlier human efforts had not. In this paper, we develop a forward counterexample construction pipeline. It starts from an arborescence on which failure is impossible and adds terminal-entering shortcut arcs. The resulting load constraints define a finite set of admissible route choices; their convex hull, the conflict polytope, yields exact conditions for demands, fractional flows, and costs. Starting from a simple arborescence, the pipeline selects Rybin's topology and characterizes all counterexamples it supports. Experiments further demonstrate the pipeline's transferability: when instructed to follow it while varying the initial arborescence and shortcut placement, ChatGPT produced counterexamples on topologies genuinely different from Rybin's and found valid constructions starting from 93 other rooted arborescence shapes. The theory also explains why the original example was difficult to discover through blind enumeration and suggests that Rybin's interaction lies between blind enumeration and a complete theory-guided construction. 排版:

    Goemans conjectured that any fractional single-source flow could be converted into an unsplittable flow without increasing total cost and while increasing any arc load by at most the largest demand. Recently, Dmitry Rybin announced a seven-vertex counterexample found through an interaction with ChatGPT, which disproved this long-standing conjecture. However, the interaction did not reveal why this particular topology and its demand, fractional-flow, and cost choices work together, whether the example belongs to a broader family, or which structural features determine where counterexamples can occur. It also left unclear why the AI-assisted search succeeded where earlier human efforts had not.

    In this paper, we develop a forward counterexample construction pipeline. It starts from an arborescence on which failure is impossible and adds terminal-entering shortcut arcs. The resulting load constraints define a finite set of admissible route choices; their convex hull, the conflict polytope, yields exact conditions for demands, fractional flows, and costs. Starting from a simple arborescence, the pipeline selects Rybin's topology and characterizes all counterexamples it supports.

    Experiments further demonstrate the pipeline's transferability: when instructed to follow it while varying the initial arborescence and shortcut placement, ChatGPT produced counterexamples on topologies genuinely different from Rybin's and found valid constructions starting from 93 other rooted arborescence shapes. The theory also explains why the original example was difficult to discover through blind enumeration and suggests that Rybin's interaction lies between blind enumeration and a complete theory-guided construction.

    检查规则: 1. 只输出HTML片段本身,不要Markdown围栏、不要解释。 -> 满足。 2. 每个自然段落用

    包裹,长文用

    /

    划分小节。 -> 原文没有明确的小节标题,如果强行加h2/h3会违反“不得增删改内容”的规则。所以只用

    包裹三个自然段是最优解。或者,我可以把第一句作为h2?不行,改变内容。就只用p。 3. 列表用

    • ,引用用
      ,代码用
      ,重要句子可用。 -> 原文没有列表、引用、代码。重要句子可以用,比如把核心发现加粗。比如 "Dmitry Rybin announced a seven-vertex counterexample found through an interaction with ChatGPT, which disproved this long-standing conjecture." 或者 "In this