The repository lists the 79‑page paper and a massive Lean codebase, yet omits the verification certificate.
*A 79‑page paper by L. Alpöge and T. Buckmaster claims a finite‑time blow‑up for the Navier‑Stokes equations. The authors paired the claim with a Lean 4 formalization hosted on GitHub. The repository’s gaps raise immediate questions about reproducibility and verification.*
A new GitHub repository claims to back a bold mathematical assertion: a finite‑time blow‑up for the Navier‑Stokes equations. The paper, authored by L. Alpöge and T. Buckmaster, spans 79 pages and was dated 20 November 2026. Alongside the manuscript, the authors released a Lean 4 formalization intended to certify the core theorems.
The repository’s structure is transparent: a PDF, a Lean package built on v4.33.1, README mappings, and a NOTICE file. Yet the repo omits critical components—no raw data, no certificate hash, and no source of the paper itself. The formal proofs remain incomplete, marked with sorry placeholders, and the heavy library (about 2,270 files, 600,000 lines) demands significant compute resources to compile. These gaps have sparked immediate questions about the claim’s reproducibility and the community’s ability to verify it.
The repo contains the 79‑page PDF titled “Finite-time blow‑up for the Navier–Stokes equations” (version dated 20 November 2026). It also includes a Lean 4 package built on v4.33.1 with Mathlib, a README that maps formal statements to Theorem 1.1, 7.1, Corollary 7.2 and Appendix B, and a NOTICE file describing proved files. The lean/ folder holds about 2,270 files and 600,000 lines of code. Build instructions are provided (lake exe cache get; lake build). No external data, no certificate hash, and no source of the paper are present.
Statement files at the top of lean/ define theorems but leave proofs as sorry placeholders. Solution.lean supplies proofs drawn from the library lean/Breakdown/. Comparator*.json files enforce agreement between statements and solutions and restrict axioms to propext, Quot.sound and Classical.choice. The library’s size demands substantial memory and time for a full build. While the formalization covers many statements, the presence of sorry blocks means the proof is not fully mechanized.
The formalization contains no data. The certificate whose hash the paper cites is absent and slated for future release. Verification of the certificate’s conditions in Appendix B relies on a rigorous interval‑arithmetic computation at 128‑bit precision using the Arb library through FLINT—neither the computation nor its results are in the repository. The source code of the paper itself is also missing. Moreover, a patch applied on 20 November 2026 to cell‑layer statements was never rebuilt before the repo was assembled.
The repository has attracted 218 stars, 7 watchers and 12 forks despite lacking a description, website or topic tags. The activity log shows two commits and a modest watch/fork count, indicating niche interest among formal methods and PDE researchers. The absence of a rebuilt patched package and missing verification artifacts have prompted scrutiny, but the star count suggests the claim still draws attention.
The paper version is dated 20 November 2026.
The library comprises about 2,270 files and 600,000 lines of code.
The verification uses 128 bits of precision with the Arb library through FLINT.
The release puts a high‑stakes claim into the open, but the missing certificate, unrebuilt patch, and incomplete proofs leave the result unverified. Until the authors publish the promised data and the rigorous interval‑arithmetic verification, the mathematical community must treat the blow‑up claim with caution. Future updates to the repo will determine whether the proof can move from speculative to certified.
Sources: [GitHub - tristanbuckmaster/navier-stokes]