Sign in

Blog › Hacker News

Image: geralt / Pixabay
Hacker News · 3 min read

Lean Community Certifies Machine-Checked Riemann Hypothesis Proof

Published Nov 22, 2026

A 17-million-line Lean library claims to prove the Riemann Hypothesis. The Lean and Mathlib communities say independent checkers confirm it proves exactly Mathlib's own statement.

The Lean and Mathlib communities have issued a joint statement declaring that a newly released proof of the Riemann Hypothesis is correct. The proof, published by Levent Alpöge on November 20, 2026, is a Lean library of about 17 million lines of code that depends directly only on Mathlib. The statement is endorsed by the Lean FRO and the Mathlib Initiative.

If the claim holds up, it settles one of the most famous open problems in mathematics, and it does so in a form that does not rely on referees reading a paper. It relies on machines checking every logical step.

What the communities say they verified

The statement lays out four criteria, which it traces to a recent post on the Xena Project blog about how to trust a Lean-checked result. The communities say all four are met.

  • The statement is right. The Riemann Hypothesis is already formalized in Mathlib as RiemannHypothesis, a definition that has been vetted by Lean experts and number theorists.
  • The library compiles. Dozens of people have independently built the codebase without errors and reproduced the same raw proof export file.
  • The code proves the right theorem. A community-added file isolates the statement and imports only Mathlib. Two separate tools, comparator and con-leche in compare mode, confirmed that the theorem proved in Alpöge's library is exactly Mathlib's statement.
  • The proof is valid. The exported proof was run through several checkers, including Lean's official kernel and independent implementations written in different programming languages.

Why the checking setup matters

A proof of this size invites obvious worries, and the verification process is designed around them. One worry is that a proof assistant's own logic might have a flaw that a clever proof could exploit. The checkers con-leche and con-ron address this directly: according to the statement, they come with a formal proof, relative to set theory, that they cannot accept a false proof.

Another worry is malicious code hidden in the Lean sources themselves, since Lean is a programming language as well as a proof system. The checks therefore ran on an inert export of the proof, which is data rather than executable source. That keeps the verification out of reach of anything the proof's authors might have embedded in their code.

This layered approach, a trusted statement, reproducible builds, a statement-matching check and multiple independent kernels, is becoming a template for how large formal proofs are audited. An extensive overview of the checks is hosted on an MIT page, per the announcement.

Context: Lean and Mathlib

Lean is an open-source proof assistant and programming language. Mathlib is its community-maintained library of formalized mathematics, and it is the shared foundation that made this kind of verification practical. Because the Riemann Hypothesis statement lives in Mathlib rather than in the new codebase, reviewers did not have to trust the author's own definitions. They only had to confirm that the proof targets the library's established one.

The open-source angle is central here. The verification was a distributed effort in which many people donated compute and time to rebuild and check an enormous artifact, and the tooling they used was built by third parties in several languages.

What to watch next

A machine-checked proof tells mathematicians that the result is true, not why. The statement itself notes that the mathematical community now has work to do in explaining the new ideas behind the proof. With 17 million lines of Lean, extracting a human-readable argument will be a large project in its own right.

Developers should also watch the tooling. Proof export formats, independent kernels and statement-comparison tools were stress-tested at an unusual scale, and the lessons are likely to feed back into Lean and Mathlib. The proof's legitimacy now rests less on authority than on a process anyone can rerun, which is a notable shift for how mathematics is verified.

Original reporting: leanprover-community.github.io

Discussion

Loading comments …

The Repocharts newsletter

Once a week: the biggest repo clashes, the fastest risers and the AI and tech stories that matter. Free, no spam, unsubscribe with one click.

We send a confirmation link first (double opt-in). See our privacy policy.