Independent checkers accept Lean proof of the Riemann Hypothesis
Original title: Verification files for the Lean proof of the Riemann Hypothesis
Five theorems depend only on three standard axioms; two kernel replays of the export were still running.
Andrew Sutherland published verification results for a Lean proof of the Riemann Hypothesis announced on November 20, with checks run using Claude Code. As of November 21, all 175,187 modules compiled on Lean v4.33.1, and independent kernels nanoda, con-leche and lean4lean accepted the proof or compiled modules. The five checked statements cover the Riemann Hypothesis and Dirichlet L-functions, not Dedekind, Hecke or general automorphic L-functions. Combining the proof with an independent Robin–Lagarias formalization yields two unconditional divisor-sum inequalities.
Unconfirmed: The lake comparator run and the Lean-kernel replay of the export were still in progress, with no verdict yet.
- Axiom audit found only propext, Classical.choice and Quot.sound.
- nanoda accepted the full export in 27 minutes 42 seconds on 80 threads.
- con-leche accepted 1,468,418 declarations and matched all five challenge theorems.
- Five lean4lean modules needed a raised step limit after deterministic timeouts.
- Older con-leche and con-ron binaries declined because of an unsupported inductive type, not a failed proof.
- Organizations
- MIT
- AI models
- Claude Code
Indicators of compromise
auto-extracted · verify before use · export allAll| Type | Indicator | Context |
|---|---|---|
| sha256 | 088a57c9e7454fa5279a86a825e551828ac8431694855aadd1f6e6a6eb37bb1d | ded189bf856e0c31176926f0d ; its challenge export has sha256 088a57c9e7454fa5279a86a825e551828ac8431694855aadd1f6e6a6eb37bb1d . These statements still rely on Mathlib's definitions of t |
| sha256 | 375cbee2fc3fc0f5088ef2e6dc9a5ac5530924310dc62422ec65aac9dc4c1d25 | 006d10bf5f873ddf796ad Manifest of all compiled files sha256 375cbee2fc3fc0f5088ef2e6dc9a5ac5530924310dc62422ec65aac9dc4c1d25 valid only for the build directory /home/ubuntu/grh/src/lea |
| sha256 | 44651dcc482a6e06e8d626994221aff638e422fded189bf856e0c31176926f0d | n proof The combined export is 61,752,134,938 bytes, sha256 44651dcc482a6e06e8d626994221aff638e422fded189bf856e0c31176926f0d ; its challenge export has sha256 088a57c9e7454fa5279a86a82 |
| sha256 | 5ea74b912147c5c93cc2b6d70cc727cf908df2f62ac6a20480a3fa02f6be244e | 0e3b954916af546d9004ade01c3555744a681b05381 unpacked sha256 5ea74b912147c5c93cc2b6d70cc727cf908df2f62ac6a20480a3fa02f6be244e The exported challenge: the five statements with sorry proo |
| sha256 | 61a8938ad5f5e57397c1c0e3b954916af546d9004ade01c3555744a681b05381 | e Main . Unpacks to 61,570,595,788 bytes. compressed sha256 61a8938ad5f5e57397c1c0e3b954916af546d9004ade01c3555744a681b05381 unpacked sha256 5ea74b912147c5c93cc2b6d70cc727cf908df2f62ac |
| sha256 | 9146dfb4148e24d2252e04c50742f3e49dd126dfcc5006d10bf5f873ddf796ad | t 5dc36dc1 , each built with Lean v4.36.0 Main.olean sha256 9146dfb4148e24d2252e04c50742f3e49dd126dfcc5006d10bf5f873ddf796ad Manifest of all compiled files sha256 375cbee2fc3fc0f5088ef |
| sha256 | 9631b4f020199e580ba86178cb56c6e9b09f690c2369a80dc4334aa93f49c7fc | the 14 parts from alpo.ge joined; 239,275,554 bytes sha256 9631b4f020199e580ba86178cb56c6e9b09f690c2369a80dc4334aa93f49c7fc Source mirror Deicyde/riemann-hypothesis , byte-identical t |
| sha256 | bd484ac9886a96f26b1d9a7a59ebd45b7a70d102c015d4f31f9302bbdd6ab2b9 | only Mathlib. Unpacks to 350,862,970 bytes. unpacked sha256 bd484ac9886a96f26b1d9a7a59ebd45b7a70d102c015d4f31f9302bbdd6ab2b9 Smaller files: comparator.json : Comparator config naming t |
From the original article
9 min read at math.mit.eduVerification files for the Lean proof of the Riemann Hypothesis Andrew Sutherland (MIT), checks run with Claude Code. Page last updated 2026-11-21 12:55 UTC. Results below are as of that time and this page will be updated as the remaining checks finish.
By Bluestein · math.mit.edu
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.