VibeMathedMath problems solved with AI

Seymour's second neighborhood conjecture

An oriented graph is a finite digraph with no loops, multiple arcs or pairs of opposite arcs. For a vertex vv let N1+(v)N_1^+(v) be its out-neighbours and N2+(v)N_2^+(v) the vertices at directed distance exactly two. Seymour conjectured that every nonempty oriented graph has a vertex with ∣N1+(v)∣≤∣N2+(v)∣|N_1^+(v)|\le|N_2^+(v)|. The tournament case (Dean's conjecture) was proved by Fisher (1996); Kaneko and Locke proved it for minimum outdegree at most six, and later work treated tournaments missing a matching, star or clique, dense cases, and random orientations, with a best universal ratio of about 0.7155 (Huang-Peng). Does every nonempty finite oriented graph have such a vertex?

Result
Proved(see note)
Status
Candidate (review pending)
AI contribution
AI-discovered
Method
Argument
Field
Directed graphs; oriented graphs and tournaments
Posed by
P. Seymour; recorded by N. Dean and B. J. Latka, Squaring the tournament - an open problem, Congressus Numerantium 109 (1995)
Year posed
1995
Years open
31y
Solved
2026-09-23
Model
Unreleased internal OpenAI model
Vendor
OpenAI
Collaborators
—
Verification
Lean-checked, statement unaudited
Publication
Announced
Collection
OpenAI math release (October 2026), version adc7f12
Significance
38 / 100
Disclosed cost
—
Wikipedia
No dedicated article

What was actually shown

Theorem 1.1: every nonempty finite oriented graph has a vertex vv with ∣N1+(v)∣≤∣N2+(v)∣|N_1^+(v)|\le|N_2^+(v)|, with no connectivity, density or degree assumption. Via Seacrest's equivalences it also gives the vertex- and arc-weighted forms, and it yields short-cycle consequences, for example a directed triangle whenever minimum in- and outdegree are at least one third of the order. It does not address infinite digraphs or digraphs with 2-cycles, which the conjecture excludes.

What the AI did

Produced by an unreleased internal OpenAI model as part of an OpenAI evaluation on open research problems, published in the openai/math release (pinned commit adc7f12). The release README says the vast majority of results used one fixed procedure, averaging about three hours of ChatGPT Pro thinking compute per result; this result is not among the README's stated exceptions (the Riemann zeta zero-free region work and the Hodge conjecture for CM abelian varieties). The manuscript is authored as OpenAI with no human author named. The main theorem has a Lean formalization in the release (Comparator challenge SeymourSecondNeighborhood).

Verification

No independent mathematician has checked this yet. Checked here: abstract, introduction and Theorem 1.1 of the TeX source, read against the conjecture as recorded by Dean and Latka. Lean-checked on the release's Comparator challenge SeymourSecondNeighborhood (declaration OAI.SeymourSecondNeighborhood.exists_goodVertex, listed in lean/formalization.yaml). Its statement was read here: for every nonempty finite type with a loopless asymmetric relation, some vertex has at most as many first out-neighbours as second neighbours, where second neighbours exclude the vertex itself and its out-neighbours and are reached by a two-step path. That is exactly the headline claim, with no connectivity or degree hypothesis. Not rebuilt here. The minimal-counterexample and pruning argument was not refereed.

Sources

Changelog1 change

Discussion