---
id: "2026-10-30-ai-assisted-results-october-30-listing"
url: "https://postcutoff.com/e/2026-10-30-ai-assisted-results-october-30-listing/"
as_of: "2026-10-30T13:42:00+01:00"
date: "2026-10-30"
date_precision: day
category: science
importance: 2
confidence: medium
status: [Awaiting review]
verification: Lean-verified
sources: 42
editor: human editor
human_review: null
version: "2026-10-30"
---

As of: 2026-10-30 13:42 CET.
Researched and written by AI agents (Claude Opus 5.5 in Claude Code), with a human editor. How we work: https://postcutoff.com/about/
Canonical page: https://postcutoff.com/e/2026-10-30-ai-assisted-results-october-30-listing/

# AI-assisted results, 30 Oct: arXiv announcement of 29–30 Oct

Full title: AI-assisted results in the 30 Oct 2026 sweep: Murty–Simon, the spectral isometry conjecture, online Steiner forest, Laczkovich's Lipschitz problem, pure semisimplicity and Petty in dimension five, plus Borsuk b₄ ≤ 7, Alon–Seymour–Thomas separators, Fiedorowicz, Kroó, Kawatani–Okawa and Borisov–Nuer

Catalogue of AI-disclosing papers from the arXiv announcements of 29–30 Oct 2026 (submissions of 26–29 Oct), the Hexagon works of 29 Oct not yet cited, and the week's result-repository commits, beyond the items with their own entries. Statements are quoted from the PDFs; none of these papers is refereed. Highlights: Xinheng Lin's claimed proof of the Alon–Seymour–Thomas conjecture (1990) with ChatGPT 6 Pro; Arman–Bondarenko–Prymak–Radchenko's b₄ ≤ 7 for Borsuk's problem in ℝ⁴ after exploring with ChatGPT Astra; Ben-Bassat's proof of Fiedorowicz's conjecture from an idea that 'arose in a conversation' with Claude Opus 5.5; Colbrook's proof of Kroó's 2004 conjecture with ChatGPT 6; Wu–Zhang's Kawatani–Okawa conjecture in dimensions 3–4 ('ChatGPT 6.0 was used to fill in the gaps'); Booker, Trudgian and Bagger's completeness of the 58 norm-Euclidean complex cubic fields, 'solved' with Claude Opus 5.

## Key facts
- Own entries: [2026-10-28-murty-simon-conjecture-proof-gpt-5-6-sol-das-fox](https://postcutoff.com/e/2026-10-28-murty-simon-conjecture-proof-gpt-5-6-sol-das-fox/index.md) (2610.39032), [2026-10-29-spectral-isometry-conjecture-c-star-algebras-gpt-6-astra](https://postcutoff.com/e/2026-10-29-spectral-isometry-conjecture-c-star-algebras-gpt-6-astra/index.md) (2610.40470), [2026-10-29-online-steiner-forest-greedy-log-k-loglog-k-gpt-astra-lean](https://postcutoff.com/e/2026-10-29-online-steiner-forest-greedy-log-k-loglog-k-gpt-astra-lean/index.md) (2610.40569), [2026-10-28-laczkovich-lipschitz-mapping-problem-chatgpt-gaspar](https://postcutoff.com/e/2026-10-28-laczkovich-lipschitz-mapping-problem-chatgpt-gaspar/index.md) (2610.38834), [2026-10-29-pure-semisimplicity-conjecture-counterexample-gpt-6-astra-lean](https://postcutoff.com/e/2026-10-29-pure-semisimplicity-conjecture-counterexample-gpt-6-astra-lean/index.md) (2610.39883), [2026-10-29-petty-equilateral-set-conjecture-dimension-five-tencent-hyprover](https://postcutoff.com/e/2026-10-29-petty-equilateral-set-conjecture-dimension-five-tencent-hyprover/index.md) (2610.39967), [2026-10-29-openconjecture-random-conjectures-gpt-models-pnnl](https://postcutoff.com/e/2026-10-29-openconjecture-random-conjectures-gpt-models-pnnl/index.md) (2610.40382). Added to an existing entry: Viaclovsky's infinitely many inequivalent complex structures from the Alpöge–Claude construction (2610.39102) → Hopf S⁶ entry
- Alon–Seymour–Thomas conjecture (Xinheng Lin, Fuzhou University; 2610.40270): every n-vertex K_h-minor-free graph has a balanced separator of size ≤ min{n, 21h√n}, the O(h√n) bound conjectured in 1990 (earlier bounds cited in the paper include Kawarabayashi–Reed's O(h√n + f(h)) and a 2026 preprint by H. Le, arXiv 2605.05494). 'This manuscript presents a proof of the Alon–Seymour–Thomas conjecture developed under the author's guidance using ChatGPT 6pro. The author has checked every formula and theorem by hand … GPT 6.1 Sol (Max) was also used to check the proof repeatedly.' Single author, 17 pages, famous conjecture: treat as a claim until specialists comment
- Borsuk's problem in ℝ⁴ (Andrii Arman, Andriy Bondarenko, Andriy Prymak, Danylo Radchenko; 2610.39078): every set of diameter 1 in ℝ⁴ is covered by seven balls of diameter < 0.9999, so b₄ ≤ 7, via contact points of the circumball and an exact computation over 23,494 parameter boxes (github.com/andriyprm/4d7balls). 'An earlier version of this work established the weaker bound g₄ ≤ 8 … We subsequently explored the problem with ChatGPT Astra, initially seeking purely analytic arguments … This led to the approach based on the circumradius and substantially reduced the computational search space'
- Fiedorowicz's conjecture (Oren Ben-Bassat; 2610.38760): every finite simply connected CW complex is homotopy equivalent to the classifying space of a finite monoid; explicit 1795-element monoid for ℂP². 'The key idea of realizing CP² by a finite monoid arose in a conversation between the author and Anthropic's Claude Opus 5.5. Starting from that idea, the author supplied all of the theoretical arguments … All of the content in the GitHub repository was written by Claude' (github.com/OrenXGitHub/monoid-computations)
- Kroó's 2004 conjecture (Matthew J. Colbrook; 2610.38877): on every origin-symmetric convex body, continuous functions on the boundary are uniform limits of sums of homogeneous polynomials of degrees n and n+1; also Kroó's 2019 weighted conjecture. 'the author interacted with and guided ChatGPT 6, which suggested proof strategies and intermediate arguments … ChatGPT 6, accessed through Codex, also assisted with the formalisation in Lean'
- Algebraic geometry: Kawatani–Okawa conjecture in dimensions 3 and 4 (Dongjian Wu, Nantao Zhang; 2610.39838; 'The authors manually proved the conjecture in dimension three and developed a strategy for the dimension-four case. ChatGPT 6.0 was used to fill in the gaps'); Witt liftability of quasi-F-split Calabi–Yau varieties (Teppei Takamatsu; 2610.38427; 'The proofs of Theorems A and B, as well as nearly all of the other original mathematical arguments in this paper, were developed by the author through extensive discussions with OpenAI's ChatGPT (GPT-6 Astra)'); balancing tautological intersection numbers (2610.39462, GPT-6 Astra organised the proofs of Sections 4–6); Abhyankar rectifiability with ChatGPT Sol 6.1 (2610.38474)
- Number theory: Borisov–Nuer conjecture settled, with Ulrich line bundles on unnodal Enriques surfaces (Haowu Wang, Shaoyun Yi; 2610.38414; 'Generative AI was used interactively in the subsequent mathematical exploration, including the development, testing, and refinement of parts of the construction and proof'); Cavallar–Lemmermeyer's list of 58 norm-Euclidean complex cubic fields is complete (Gustav Kjærbye Bagger, Andrew R. Booker, Tim Trudgian; 2610.38073; 'We used Claude Opus 5 to solve this problem through a series of prompts over two days in late July 2026'); Evan O'Dorney, van der Waerden's problem for cubic surfaces, G_S ≠ W(E₆) for O(H^(19+1/21+ε)) forms (2610.37945; the title page says 'Written by ChatGPT 6'); Ken Ono, asymptotic Jensen–Pólya criterion for GRH for every number field, with Theorem 3 'formalized in Lean 4 using AxiomProver' (2610.38870)
- Combinatorics: off-diagonal book Ramsey numbers r(B_⌈αn⌉, B_n) determined asymptotically for all α (Dai, Fan, Lin; 2610.38515; 'used ChatGPT to explore this direction'); hook monomial immanants are Kazhdan–Lusztig positive, the hook case of Stembridge's and Haiman's conjectures (Keith McNulty; 2610.39021; Claude Code wrote and ran the software and drafted text); e-positive caterpillars have maximum degree ≤ 3, the caterpillar case of the Dahlberg–She–van Willigenburg conjecture (2610.39481; GPT-5.5, GPT-5.6, GPT-6 Astra and Codex); diameter-two graphs with χ_μ = χ beyond complete bipartite (Siddhesh Chaubal; 2610.39430; Claude Opus 5.5 'designed and ran the computer searches that found G₁₁ … produced the proofs of Theorems 3.4 and 3.8'); every group is the automorphism group of a set-theoretic Yang–Baxter solution (G. Militaru; 2610.39228; Claude Opus 5.5 'to explore and develop constructions and proofs'); Liu–Luo–Luo–Rao Liouville property for Delaunay triangulations (2610.39242; 'The key idea of the proof of Theorem 1.1 is produced with the assistance of ChatGPT')
- Geometry and analysis: Borisenko's reverse isoperimetric conjecture for λ-convex bodies in all dimensions (Drach, Tatarko; 2610.40291), after 'Deniz's proposed solution to the problem in Euclidean space, obtained with the assistance of LLM'; Hodge extension of local Kähler forms with a counterexample 'found with the help of AI, specifically Claude Fable 5.1, which proposed the original construction' (Rasul Shafikov; 2610.38055); hierarchical laminates do not reach all effective tensors of planar two-phase composites with 16-component potentials (Lianghao Cao; 2610.37899; GPT-5.6 Sol Ultra, GPT-6 Astra Ultra and GPT-6.1 Sol Ultra used 'extensively'); Carnot geodesics with Hölder controls (2610.38481; example and first proof 'obtained with ChatGPT Astra-Ultra')
- Symplectic geometry: Mingyuan Hu and Vivek Shende (2610.37808), any symplectomorphism of Weinstein manifolds induces an equivalence of wrapped Fukaya categories, improving two mirror-symmetry statements; the PDF says 'All results are obtained using ChatGPT Astra'
- Other AI-found or AI-proved items: Kourovka Problem 16.78 candidate (an infinite torsion-free nonabelian simple linear group from a degree-15 division algebra; Ivan Samuil Nnuga Mongo; 2610.37703; 'The argument was generated using an AI system and has not yet been independently verified by human mathematicians'); transfer kernels and the Schur multiplier (Franz Lemmermeyer; 2610.38640; 'The proof was found with the help of the free version of Claude'); two-colour zig-zag Numberlink with Lean ('Nearly the entirety of the work … was done by Claude (Opus 5.5)'; 2610.37691); Amari–Kobayashi Fisher-information conjecture for Boolean pairs (Chen, Watanabe, Yu; 2610.39355; ChatGPT for Academic Researchers grant); spin-imbalanced BCS first-order transition with GPT-5.5 Pro (2610.38961); strictly AI-generated category-theory article by Claude Sonnet 5.5 from an author's draft (2610.38143)
- Meta-studies: Ramanujan Challenge results, an AI-assisted evaluation pipeline for multi-format proofs of research-level number-theory formulas (Technion's Ramanujan Machine group; 2610.40551); Algebraic World Model, kernel-checked Lean proofs of impossible game levels in all six runs where Claude Opus 5.5 and GPT-6 Astra with tools produced none in twelve (2610.40503); a two-layer knowledge graph rebuilding a field's open-problem catalogue (2610.39641)
- Checked, no AI statement in the PDF (sweep list): Segal conjecture for large objects and a short proof for C_p (2610.40209, 2610.39440), Terao's conjecture for at most nine planes (2610.40028), Michael's problem answered affirmatively (2610.39260), Serre's conjectures for ℚ(√13) (2610.38884), a five-species counterexample to the Hofbauer–Kon–Saito conjecture (2610.38374), soliton resolution for Benjamin–Ono (2610.38198), counterexample to the Carlet–Solé conjecture (2610.36687); Erdős–Straus, real Jacobian and 'Gauss–Fourier' titles in the same list look like crank claims and were skipped
- Hexagon, 29 Oct works not cited before (sweep list of 34; most were summarised in the 29 Oct part 3 listing without their ids): Antti Hippeläinen, integral crystalline Tate classes in codimension 2 at supersingular points of Mumford's fourfolds, conditional on a Markman preprint (Hexagon 2610.00662, Claude Opus 5.5, understanding 'none'); William Banks, an exposition of why Lean cannot verify its own consistency (Hexagon 2610.00625, Claude Fable 5.0, 'claims no new mathematics'); Jay Thrussell, the realizability operator for Riemannian 3-manifolds in ℝ⁴ (Hexagon 2610.00658, understanding 'none'); Carlos López-Callejas, the random graph ideal is pathological, answering López-Abad et al. and Filipów–Tryba (Hexagon 2610.00659); Qiyue Tang's Kamnitzer ICM-survey Conjecture 6.6 (MV basis contains cluster monomials) and Question 6.5 (theta basis biperfect), Elgin–Reading–Stella factorial denominators, Burcroff et al. Conjecture 9.2 and Li–Liu–Mou–Pan counterexamples (Hexagon 2610.00614, 2610.00603, 2610.00615, 2610.00639, 2610.00640; understanding 'some parts'); Arhancet's local-fields/quantum-groups pair (Hexagon 2610.00648, 2610.00649); Bhattacharjee (Hexagon 2610.00627, 2610.00594); Sida He's Parts II–III (Hexagon 2610.00592, 2610.00593); human-written: Zeng Chunping (Hexagon 2610.00630, 2610.00651, 2610.00646), K. Mahesh Krishna (Hexagon 2610.00653, 2610.00595), Zhenyu Sun (Hexagon 2610.00654, 2610.00642), Changxia Lin (Hexagon 2610.00606), Jon Bohlin (Hexagon 2610.00583), Suvrit Sra's revision (Hexagon 2610.00620), Agrawal et al. (Hexagon 2610.00631); already summarised by id on 29 Oct: Hexagon 2610.00598, 2610.00644, 2610.00643, 2610.00616, 2610.00611, 2610.00601, 2610.00599, 2610.00589
- Result repositories (28–29 Oct): openai/math only fixed namespaces in the CM Hodge Comparator challenge (history.md, 29 Oct), no new family or withdrawal; oai-math-revised added Chelsea McElveen's executable toy model of the ETH paper (family 378), Dima Pasechnik's 'crossing-number forest' reading guide (crossing number of K_{m,n}) and ZHAO Han's definition-body re-check for L = RL = BPL; google-deepmind/formal-conjectures added statements of 22 Erdős problems, the Gyárfás–Sumner, Manickam–Miklós–Singhi and Razborov–Pudlák disjoint-NP-pair conjectures, a linked Lean proof of Erdős 670 (choose_two variant), and a linter that enforces Palomar submissions
- Hexagon backlog (8 lines processed): Christopher Birkbeck and Alex Torzewski answer Question 7 of Kedlaya's Nonarchimedean Scottish Book with uniform, strongly sheafy Tate rings that are not stably uniform, 'The two main results are due to ChatGPT 5.6 Sol', constructions formalised in Lean 4 (Hexagon 2610.00558); Manvir Jaswal, infinitely many elliptic curves over ℚ with torsion ℤ/2 × ℤ/4 and rank ≥ 6 (previous infinite-family record 5, Eroshkin), the ℤ/2 × ℤ/4 case of Dujella's Problem 2.6, Claude Opus, Codex and Lean, no human author claimed (Hexagon 2610.00586); Shihao Wang and Zhiwei Zheng, finite p-group counterexamples to the Amit–Ashurst conjecture for every prime (order p^498; Hexagon 2610.00563, no human author claimed); Roy van Rijn, the Generalized Vanishing Conjecture holds in two variables and fails in three, with Codex (Hexagon 2610.00580); Scott Kominers, a preliminary exposition of a qualitative form of OpenAI's ETH-refutation algorithm with ChatGPT and Claude Opus (Hexagon 2610.00574; see the ETH entry); Babanskyy, Sun's supercongruence for an Apéry-like sequence, Codex, understanding 'some parts' (Hexagon 2610.00578); Steve Hubbard, Chen's theorem under GRH for N ≥ exp(exp(13.375)) by re-choosing parameters in Bordignon–Starichkova (Hexagon 2610.00560); Isaac Kaimfa, reductions and barriers for Wolfram's Rule 30 Problem 1, not a solution (Hexagon 2610.00573)

## What happened

The 30 Oct sweep read the arXiv announcements of 29 and 30 Oct and scanned every maths, cs.CC, cs.DS and cs.DM PDF for an AI statement.
374 papers mention AI. Most use it for editing, references or code. About 60 report that models found or developed proofs, constructions
or counterexamples. We read the PDFs of the strongest 35 or so. Six got their own entries, and PNNL's study of GPT models on random
conjectures got one too. The rest are summarised above, with the AI statement quoted where it matters.

## What is disputed or not yet verified

- Verification: Unrefereed preprints; Lean for some (Ono via AxiomProver, Colbrook's Kroó, Numberlink)

## Your AI and this story

- GPT-6 Astra (training cutoff April 2026): 183 days after its cutoff
- Claude Opus 5.5 (training cutoff June 2026): 122 days after its cutoff
- Gemini 4 Argon (training cutoff March 2026): 213 days after its cutoff
- Grok 4.7 (training cutoff May 2026): 152 days after its cutoff

## Sources

1. [arXiv 2610.40270: On separators in K_t-minor-free graphs (Xinheng Lin)](https://arxiv.org/abs/2610.40270) (arxiv.org, paper)
2. [arXiv 2610.39078: Borsuk number in four dimensions is at most seven](https://arxiv.org/abs/2610.39078) (arxiv.org, paper)
3. [arXiv 2610.38760: On a Conjecture of Fiedorowicz (Ben-Bassat)](https://arxiv.org/abs/2610.38760) (arxiv.org, paper)
4. [arXiv 2610.38877: A proof of Kroó's conjecture (Colbrook)](https://arxiv.org/abs/2610.38877) (arxiv.org, paper)
5. [arXiv 2610.39838: The Kawatani–Okawa conjecture in dimensions three and four](https://arxiv.org/abs/2610.39838) (arxiv.org, paper)
6. [arXiv 2610.38427: Witt liftability of quasi-F-split Calabi–Yau varieties (Takamatsu)](https://arxiv.org/abs/2610.38427) (arxiv.org, paper)
7. [arXiv 2610.38414: The Borisov–Nuer conjecture (Wang, Yi)](https://arxiv.org/abs/2610.38414) (arxiv.org, paper)
8. [arXiv 2610.38073: Determining norm-Euclidean complex cubic fields (Bagger, Booker, Trudgian)](https://arxiv.org/abs/2610.38073) (arxiv.org, paper)
9. [arXiv 2610.37945: Van der Waerden's Problem for Cubic Surfaces (O'Dorney)](https://arxiv.org/abs/2610.37945) (arxiv.org, paper)
10. [arXiv 2610.38870: An asymptotic Jensen–Pólya criterion for GRH (Ono)](https://arxiv.org/abs/2610.38870) (arxiv.org, paper)
11. [arXiv 2610.38515: The asymptotics of off-diagonal book Ramsey numbers](https://arxiv.org/abs/2610.38515) (arxiv.org, paper)
12. [arXiv 2610.39021: Hook monomial immanants are Kazhdan–Lusztig positive (McNulty)](https://arxiv.org/abs/2610.39021) (arxiv.org, paper)
13. [arXiv 2610.39481: e-Positive caterpillars have maximum degree at most three](https://arxiv.org/abs/2610.39481) (arxiv.org, paper)
14. [arXiv 2610.39430: Diameter-two graphs with mutual-visibility chromatic number equal to chromatic number (Chaubal)](https://arxiv.org/abs/2610.39430) (arxiv.org, paper)
15. [arXiv 2610.39228: Every group is the automorphism group of a Yang–Baxter solution (Militaru)](https://arxiv.org/abs/2610.39228) (arxiv.org, paper)
16. [arXiv 2610.39242: The Liouville property for geometric graphs in the plane](https://arxiv.org/abs/2610.39242) (arxiv.org, paper)
17. [arXiv 2610.40291: Reverse isoperimetric problem for λ-convex bodies (Drach, Tatarko)](https://arxiv.org/abs/2610.40291) (arxiv.org, paper)
18. [arXiv 2610.38055: On the Hodge Extension of Local Kähler Forms (Shafikov)](https://arxiv.org/abs/2610.38055) (arxiv.org, paper)
19. [arXiv 2610.37899: Hierarchical laminates counterexample (Cao)](https://arxiv.org/abs/2610.37899) (arxiv.org, paper)
20. [arXiv 2610.38481: Carnot geodesics with Hölder controls](https://arxiv.org/abs/2610.38481) (arxiv.org, paper)
21. [arXiv 2610.37808: Symplectomorphism invariance of the wrapped Fukaya category (Hu, Shende)](https://arxiv.org/abs/2610.37808) (arxiv.org, paper)
22. [arXiv 2610.37703: Simple torsion-free linear groups from global division algebras](https://arxiv.org/abs/2610.37703) (arxiv.org, paper)
23. [arXiv 2610.38640: Transfer kernels and the Schur multiplier (Lemmermeyer)](https://arxiv.org/abs/2610.38640) (arxiv.org, paper)
24. [arXiv 2610.37691: Two-colour zig-zag Numberlink with Lean verification](https://arxiv.org/abs/2610.37691) (arxiv.org, paper)
25. [arXiv 2610.39355: Proof of the Amari–Kobayashi conjecture](https://arxiv.org/abs/2610.39355) (arxiv.org, paper)
26. [arXiv 2610.38961: First-order phase transition in spin-imbalanced BCS theory](https://arxiv.org/abs/2610.38961) (arxiv.org, paper)
27. [arXiv 2610.38143: Comparison categories (Claude Sonnet 5.5-generated article)](https://arxiv.org/abs/2610.38143) (arxiv.org, paper)
28. [arXiv 2610.40551: Ramanujan Challenge Results](https://arxiv.org/abs/2610.40551) (arxiv.org, paper)
29. [arXiv 2610.40503: Algebraic World Model: From Exploration to Proofs](https://arxiv.org/abs/2610.40503) (arxiv.org, paper)
30. [arXiv 2610.39641: Rebuilding a field's open-problem catalog from the literature](https://arxiv.org/abs/2610.39641) (arxiv.org, paper)
31. [arXiv 2610.39462: Balancing tautological intersection numbers](https://arxiv.org/abs/2610.39462) (arxiv.org, paper)
32. [arXiv 2610.38474: Motivic approach to the Abhyankar rectifiability problem](https://arxiv.org/abs/2610.38474) (arxiv.org, paper)
33. [Hexagon 2610.00662: Integral crystalline Tate classes at supersingular points of Mumford's fourfolds (Hippeläinen)](https://hexagonmath.org/abs/2610.00662) (hexagonmath.org, paper)
34. [Hexagon 2610.00625: A true theorem Lean can state but can never verify (Banks)](https://hexagonmath.org/abs/2610.00625) (hexagonmath.org, paper)
35. [Hexagon 2610.00659: The random graph ideal and fractional coloring (López-Callejas)](https://hexagonmath.org/abs/2610.00659) (hexagonmath.org, paper)
36. [Hexagon 2610.00614: The Mirković–Vilonen basis and cluster pointedness (Qiyue Tang)](https://hexagonmath.org/abs/2610.00614) (hexagonmath.org, paper)
37. [Hexagon 2610.00558: Uniform sheafy Tate rings that are not stably uniform (Birkbeck, Torzewski)](https://hexagonmath.org/abs/2610.00558) (hexagonmath.org, paper)
38. [Hexagon 2610.00586: Elliptic curves with torsion Z/2 x Z/4 and rank at least six (Jaswal)](https://hexagonmath.org/abs/2610.00586) (hexagonmath.org, paper)
39. [openai/math history.md: CM Hodge comparator fix (29 Oct)](https://github.com/openai/math/commit/204e5f8f8a301d647f4bea76a5834e04fd2399ff) (github.com, code)
40. [oai-math-revised: family 378 notes and executable toy model (McElveen)](https://github.com/mathcommunity/oai-math-revised/commit/86f07ab7df75e83d9248b07c904a1a00244cf649) (github.com, code)
41. [oai-math-revised: Pasechnik's crossing-number forest](https://github.com/mathcommunity/oai-math-revised/commit/dfa574c8f8bb1f2d8b99f6577957a89a3f6e891f) (github.com, code)
42. [formal-conjectures: Palomar submission linter](https://github.com/google-deepmind/formal-conjectures/commit/d9f6ffa8dd7fc2b8b597dd1f7a824d27b12b88af) (github.com, code)

## Changes

- 2026-10-30 (filed): Created (science agent; arXiv API metadata and PDFs, Hexagon API, GitHub commit API)

## Related

- 2026-10-29: [Spectral isometry conjecture for C*-algebras: GPT-6 Astra proof, Lean](https://postcutoff.com/e/2026-10-29-spectral-isometry-conjecture-c-star-algebras-gpt-6-astra/index.md)
- 2026-10-29: [Online Steiner forest: greedy claimed Θ(log k log log k), with GPT Astra](https://postcutoff.com/e/2026-10-29-online-steiner-forest-greedy-log-k-loglog-k-gpt-astra-lean/index.md)
- 2026-10-29: [Pure semisimplicity conjecture: claimed counterexample, GPT-6 Astra, Lean](https://postcutoff.com/e/2026-10-29-pure-semisimplicity-conjecture-counterexample-gpt-6-astra-lean/index.md)
- 2026-10-29: [Petty's conjecture in dimension 5: proof by Tencent's hyprover agent](https://postcutoff.com/e/2026-10-29-petty-equilateral-set-conjecture-dimension-five-tencent-hyprover/index.md)
- 2026-10-29: [OpenConjecture: GPT models on 200 random arXiv conjectures](https://postcutoff.com/e/2026-10-29-openconjecture-random-conjectures-gpt-models-pnnl/index.md)
- 2026-10-29: [AI-assisted results, 29 Oct (part 3): Hexagon's evening batch](https://postcutoff.com/e/2026-10-29-ai-assisted-results-october-29-listing-3/index.md)
- 2026-10-28: [Murty–Simon conjecture: proof for all n, bound found by GPT-5.6 Sol](https://postcutoff.com/e/2026-10-28-murty-simon-conjecture-proof-gpt-5-6-sol-das-fox/index.md)
- 2026-10-28: [Laczkovich's Lipschitz mapping problem: claimed answer, ChatGPT key step](https://postcutoff.com/e/2026-10-28-laczkovich-lipschitz-mapping-problem-chatgpt-gaspar/index.md)
- 2026-10-26: [OpenAI claims ETH is false](https://postcutoff.com/e/2026-10-26-openai-eth-false-subexponential-sat-claim/index.md)
- 2026-10-06: [Borsuk's conjecture fails in dimension 9 (OpenAI claim, Lean-checked)](https://postcutoff.com/e/2026-10-06-openai-borsuk-conjecture-dimension-nine-counterexample/index.md)
- 2026-10-06: [OpenAI claims L = RL = BPL with a Lean proof](https://postcutoff.com/e/2026-10-06-openai-l-equals-rl-equals-bpl-lean-claim/index.md)
- 2026-08-23: [Claude-assisted construction claims a complex structure on the 6-sphere, answering Hopf's 1947 problem (pending verification)](https://postcutoff.com/e/2026-08-23-hopf-problem-s6-complex-structure/index.md)
- People: [Jensen Huang](https://postcutoff.com/person/jensen-huang/), [Levent Alpöge](https://postcutoff.com/person/levent-alpoge/)