Entrée Pandipedia
Mis à jour le 21 Nov 20264 sourcesParcourir Pandipedia
Rapport de recherche

What would verify a claimed proof of the Riemann hypothesis?

A computer can check that a proof establishes the theorem written in its code. To verify a Riemann hypothesis claim, mathematicians also need to check that the code states the same problem and relies on sound assumptions. [1] [2]

The claim and the question

On 20 November, mathematician Levent Alpöge announced that he believed he had found a proof while working with Claude. He released material in Lean, a language that can express a mathematical statement and check a proof of it. The author’s note was the starting point for verification. [3] [4] [5]

The Riemann zeta function takes complex numbers as inputs. Its zeros are the inputs that make its output zero. The hypothesis says all of its nontrivial zeros lie on the vertical line where the real part is ½. Clay Mathematics Institute describes why the location of those zeros matters for understanding how prime numbers are distributed. [6]

Three things to check

Imagine a result proved for one million examples. Checking the proof might settle all one million. The Riemann hypothesis is a claim about every nontrivial zero, so a statement restricted to a finite set would still be a different theorem. [7] [8]

CheckWhat it would establish
Read the exact statementThe definitions match Clay’s problem and the claim covers all nontrivial zeros. [9] [10]
Inspect its assumptionsThe result follows from the assumptions it actually uses. An unproved assumption that supplies the key result would not settle the original problem. [11]
Replay the checkA separate checker can check the delivered proof with the declared dependencies. That tests the proof of the encoded statement. [12]

Lean’s reference describes the checks and the confidence each can support. [13]

What an independent check has found

In an update dated 22:05 UTC on 20 November, MIT mathematician Andrew Sutherland reported that the submitted theorem used Mathlib’s existing definition of the Riemann hypothesis and depended on only Lean’s three standard axioms. Two independently written proof-checking engines accepted the export, and a separate statement comparison matched the target theorems. [14]

Two further verification runs were still in progress at the time of that update. The results report what the formal checks accepted. Clay Mathematics Institute’s problem page is a separate source for the mathematical question being tested. [15] [16]

What to look for next

Look for a precise statement of the theorem, a record of its assumptions and a reproducible check. Sutherland’s log is useful because it reports which tests passed and which were still running, each with a timestamp. [17] [18]

Continuer l’exploration
Tout parcourir