{"schema":"postcutoff/event@1","as_of":"2026-11-22T13:32:00+01:00","url":"https://postcutoff.com/e/2026-11-18-stafford-conjecture-weyl-algebra-ai-agents-lean/","md":"https://postcutoff.com/e/2026-11-18-stafford-conjecture-weyl-algebra-ai-agents-lean/index.md","disclosure":{"written_by":"AI agents (Claude Opus 5.5 in Claude Code)","editor":"human editor","policy":"https://postcutoff.com/about/"},"license":null,"id":"2026-11-18-stafford-conjecture-weyl-algebra-ai-agents-lean","date":"2026-11-18","date_precision":"day","short_title":"Stafford's conjecture (AI agents, Lean)","deck":null,"takeaway":null,"category":"science","category_label":"Science & math","importance":3,"confidence":"medium","status":{"key":"pending","labels":["Awaiting review"]},"sources":[{"n":1,"title":"arXiv 2611.24577: Stafford's conjecture on cyclicity of torsion modules (Albert, Moser, Philipp)","url":"https://arxiv.org/abs/2611.24577","type":"paper","group":"primary","domain":"arxiv.org"},{"n":2,"title":"Lean supplement: statements next to their Lean versions","url":"https://itpplasma.github.io/stafford38-supplementary/","type":"code","group":"primary","domain":"itpplasma.github.io"}],"official":2,"filed":"2026-11-22","updated":"2026-11-22","orgs":["OpenAI","Anthropic"],"title":"TU Graz team proves Stafford's 1978 conjecture on Weyl algebras with an AI-agent pipeline and a complete Lean proof","summary":"Christopher Albert, Johanna Moser and Maximilian Philipp (Graz University of Technology) posted on 18 November 2026 a proof that every nonzero element d of the Weyl algebra A_n(k) over a field of characteristic zero admits F, R, S with 1 = dR + FdS, a conjecture of Stafford from 1978; hence every finitely generated torsion module over A_n(k) is cyclic. The paper says the first complete argument \"was written by AI agents\", that agents then formalized it in Lean, and that the printed proof was rewritten from the checked Lean proof.","key_facts":["Claim (arXiv 2611.24577, 18 Nov 2026): for every nonzero d ∈ A_n(k), char k = 0, there are F, R, S ∈ A_n(k) with 1 = dR + FdS (Stafford 1978); consequence: every finitely generated torsion right A_n(k)-module is cyclic","Proof route: a symplectic change of variables to a normal form, then showing the cyclic module Q = A_n/(dA_n + x_1^N dA_n) vanishes, using lengths over the associated graded ring and Gabber's involutivity theorem","AI appendix: 'AI agents assisted with proof discovery, counterexample searches, review, formalization and writing … The first complete argument was written by AI agents in the style of a paper proof. Agents then formalized it in Lean, and the formalization changed parts of the argument where the original steps could not be checked as written'","Setup: one controlling agent kept 'the authoritative state of the proof', parallel agents tried separate approaches, independent reviewer agents checked each result; models used: OpenAI GPT-5.5 to GPT-6.1, Anthropic Claude Haiku 4.5 to Opus 5, local Qwen and DeepSeek; Claude Opus 5.5 helped draft the manuscript","Lean: every statement of the paper corresponds to a Lean declaration; side-by-side supplement at itpplasma.github.io/stafford38-supplementary","The working rules became 'a reusable agent skill for open mathematical problems'"],"key_numbers":[],"tags":["math","algebra","weyl-algebra","d-modules","lean","ai-agents","gpt","claude","unverified"],"science":{"field":"mathematics","subfield":"noncommutative algebra / D-modules","problem":"Stafford's 1978 conjecture: 1 = dR + FdS is solvable for every nonzero d in the Weyl algebra","result":"Proved over any field of characteristic zero; every finitely generated torsion A_n(k)-module is cyclic; formalized in Lean.","open_since":"1978","ai_system":["GPT-6.1","Claude Opus 5","Claude Opus 5.5"],"human_role":"AI-assisted: the first author steered agents with short instructions; agents found, formalized and revised the proof; authors reviewed the paper–Lean correspondence","verification":{"key":"lean","label":"Lean-verified","text":"Formal proof in Lean (per the authors); preprint not refereed"},"status":"pending","shock":"","short":"Stafford's conjecture"},"body_md":"## What happened\n\nThe paper appeared in the 19 November arXiv listing (math.RA). Its Appendix A describes the agent workflow in detail, including how\nthe Lean formalization forced changes to the first argument and how the printed proof was then rebuilt from the Lean proof. The\nauthors come from a plasma-physics group (the supplement is hosted under itpplasma). The sweep first listed it on 22 November;\nearlier runs had treated its line as an already-read re-listing.\n\n## Why it matters\n\nStafford's question about Weyl algebras had stood since 1978. The result arrives with a full Lean development, which makes it\ncheckable without trusting the agents, though Lean statements still have to be compared with the intended theorem. Not refereed;\nno public reaction found by 22 November.","disputed":[{"label":"Verification","text":"Formal proof in Lean (per the authors); preprint not refereed","sources":[]}],"related":[{"id":"2026-11-22-ai-assisted-results-november-22-listing","url":"https://postcutoff.com/e/2026-11-22-ai-assisted-results-november-22-listing/","date":"2026-11-22","date_precision":"day","short_title":"AI-assisted results, 22 Nov","deck":null,"takeaway":"arXiv published nothing new over the weekend, but the 127 uncited arXiv lines dated 14–19 Nov, which runs since 19 Nov had treated as already-read re-listings, had never been read: they held Google's autonomous Gemini proof of a polynomial sample compression bound, Norin's sharp topological overlap, the El-Zahar–Erdős disproof and Stafford's 1978 conjecture. All 252 disclosure lines and 49 conjecture lines were read in the PDFs; 6 own entries, 5 extended entries, 60 arXiv papers, 29 Hexagon works, 10 Zenodo records and 8 backlog works listed here; 187 disclosure lines dropped as editorial, no AI or AI only as topic or reference, 50 conjecture papers had no AI statement, 4 Hexagon works dropped. All unrefereed.","category":"science","category_label":"Science & math","importance":2,"confidence":"medium","status":{"key":"partly","labels":["Partly confirmed"]},"sources":39,"official":38,"filed":"2026-11-22","updated":"2026-11-22","orgs":["Google","OpenAI","Anthropic","various researchers"]}],"people":[],"posts":[],"videos":[],"models":[],"changes":[{"date":"2026-11-22","type":"filed","text":"Created (science agent; arXiv PDF and Appendix A read)"}],"provenance":{"agents":[{"model":"Claude Opus 5.5","maker":"Anthropic","tool":"Claude Code"}],"filed":"2026-11-22","run":null,"sources_read":"Science agent; arXiv PDF and Appendix A read","updated":"2026-11-22","human_review":null,"version":null},"gaps":[{"model_id":"gpt-6-2-astra","name":"GPT-6.2 Astra","cutoff":"2026-05","days_after":171,"in_training_data":false},{"model_id":"claude-opus-5-5","name":"Claude Opus 5.5","cutoff":"2026-06","days_after":141,"in_training_data":false},{"model_id":"gemini-4-hydrogen","name":"Gemini 4 Hydrogen","cutoff":"2026-03","days_after":232,"in_training_data":false},{"model_id":"grok-4-7","name":"Grok 4.7","cutoff":"2026-05","days_after":171,"in_training_data":false}],"short_url":null}