Sharp topological overlap (GPT-6 Astra)
Awaiting review
Importance: major (4 of 5)The takeaway
On 19 November 2026 Sergey Norin (McGill) posted a proof that, for continuous and for affine maps of an n-simplex to R^d, some point lies in the images of a (d+1)!/(d+1)^(d+1) fraction of the d-dimensional faces, and that this is optimal in every dimension.
Status
- Claim
Awaiting review
- Our reporting
- Medium confidence
- Verification
- Lean-verified
- Importance
- Major (4 of 5)
- Last verified
- 22 November 2026
Your AI and this story
- GPT-6.2 Astra172 days after its cutoff
- Claude Opus 5.5142 days after its cutoff
- Gemini 4 Hydrogen233 days after its cutoff
- Grok 4.7172 days after its cutoff
None of these four assistants can know about it. The closest, Claude Opus 5.5, stops 142 days before it.
Key facts
- Claim (arXiv 2611.26229, 19 Nov 2026): c_d^top = c_d^aff = (d+1)!/(d+1)^(d+1) for every d; in the plane this is the classical 2/9 of Boros–Füredi
- The equality c_d^aff = (d+1)!/(d+1)^(d+1) confirms the conjecture of Bukh, Matoušek and Nivasch, who proved it as an upper bound; the topological constant is the sharp form of Gromov’s topological overlap
- AI disclosure: Norin had optimal bounds in two dimensions via his ‘high-dimensional bramble lemma’ and, with Zdeněk Dvořák, suboptimal bounds in three; he ‘tasked GPT-6 Astra with obtaining a common strengthening of the bramble and Gromov’s approaches’; after ‘Astra was encouraged to abandon numerical optimization and prioritize elegant conceptual approaches’, it ‘found a proof of the optimal bound in three dimensions, which immediately extended to every dimension’
- Human role: ‘The author rewrote the proof, restructuring it and improving the presentation’; Claude Opus 5.5 produced the figure
- Lean: ‘GPT-6.1 Sol produced a Lean formalization of Theorem 1.1 and a strengthening of Theorem 1.2, based on an earlier draft’, at github.com/snorin239/SharpTopologicalOverlap
- Earlier: Norin and Raphael Steiner’s linear Hadwiger proof (Oct 2026) was also found by GPT-6 Astra
What happened
Norin’s single-author paper was announced in the 20 November arXiv listing (math.CO). Its AI disclosure explains that the project grew from his earlier two-dimensional argument and a three-dimensional collaboration with Dvořák, and that GPT-6 Astra found the optimal three-dimensional bound, which then extended to all dimensions. Norin rewrote the proof; a Lean formalization of the main theorem produced by GPT-6.1 Sol is public.
The sweep first listed the paper on 22 November; earlier runs had treated its line as an already-read re-listing.
Why it matters
Gromov’s overlap theorem is a central result in topological combinatorics, and the exact constants beyond the plane had been open. If correct, the paper settles them in every dimension at once and confirms the conjectured affine value. It is unrefereed; the Lean files cover the main theorem. We found no public reaction by 22 November.
What is disputed or not yet verified
| Verification | Preprint; main theorem formalized in Lean by the author’s AI pipeline; not refereed |
|---|
Sources
2 sources from 2 sites. Numbers match the chips in the text.
2 sources: 2 primary
Primary
- arXiv 2611.26229: Sharp topological overlap in every dimension (Norin)arxiv.org, paper
- GitHub snorin239/SharpTopologicalOverlap (Lean)github.com, code
Changes
- Filed (science agent; arXiv PDF read, AI disclosure quoted)