Navier-Stokes Blowup Proof Claimed With AI Help

Fluid flow equations on a blackboard illustrating a Navier-Stokes blowup proof built with AI

Three results landed in mathematics this week, and AI sits behind all of them. A Navier-Stokes blowup proof for the unforced equations in three dimensions is the headline, alongside a similar result for the Euler equations and a Lean-only proof of the Generalized Riemann Hypothesis. None has been through peer review yet.

The authors used AI heavily to produce the work, and each result sits at a different stage of being understood by other mathematicians. That ordering matters, so here is how they stack up.

The Euler equations result

Levent Alpöge, Tristan Buckmaster and Matei Coiculescu wrote a proof of finite-time blowup for the unforced incompressible Euler equations in three dimensions. They did it with heavy AI assistance.

This is the most digested of the three. The proof is human-readable, meaning mathematicians can follow the argument on the page instead of trusting a machine. It also comes with a formalization in Lean, a proof assistant that checks every logical step mechanically.

“Finite-time blowup” means a smooth starting state of the fluid develops a singularity, a point where the equations stop making sense, after a finite amount of time. The Euler equations describe an ideal fluid with no friction.

The Navier-Stokes blowup proof

Alpöge and Buckmaster, again using AI extensively, produced a proof of finite-time blowup for the unforced incompressible Navier-Stokes equations in three dimensions. Navier-Stokes adds viscosity to the Euler picture, and whether such solutions can blow up is one of the Clay Mathematics Institute’s Millennium Prize Problems.

This one also comes with a Lean formalization. The paper is technically human-readable, and the overall strategy is broadly clear. It follows one of the expected routes to the result:

  • Build a numerical approximation of a discretely self-similar solution.
  • Apply rigorous perturbation theory to convert that approximation into an exact solution.

There’s a caveat. More work is needed before a proper, professionally written proof exists, so the current paper is a working version rather than a polished one.

The Generalized Riemann Hypothesis claim

The third item is the boldest. Alpöge, using an internal AI model, generated a Lean-only proof of the Generalized Riemann Hypothesis, which extends the famous Riemann Hypothesis to a wider family of functions. The proof is described as enormous.

Unlike the other two, there’s no human-readable paper to lean on. Lean is the only artifact. Efforts to verify it and pull out the underlying strategy are running in a public discussion thread on the Lean community’s Zulip. Moderators there have since narrowed the conversation to the Lean-related aspects of the proof.

Early analysis suggests the proof draws on the full power of modern developments in the Langlands program and in arithmetic geometry. That’s a first impression, not a finding, and it may change as people read more of the code.

Why the Lean files matter

A Lean proof that compiles is a strong guarantee that the logic holds, given the stated definitions. It doesn’t say whether the statement Lean proved is the one mathematicians care about, or whether a human can learn anything from it. Checking both is slow work.

The Navier-Stokes blowup proof shows the useful middle case. Because the strategy is readable, specialists can compare it against the expected route and judge where the heavy lifting happens. The Generalized Riemann Hypothesis file offers no such shortcut. Reading it is closer to reverse engineering than to refereeing.

One thing helps. The author is available to answer questions, which isn’t always true when a proof comes out of an automated system. That should speed up the digestion process, though it will still take time.

What happens next

Expect the next few weeks to be about verification, not celebration. The Lean community is working through the Generalized Riemann Hypothesis files, and the Navier-Stokes authors still have a professionally written proof to produce.

The Euler paper is the one to watch first, since it is both formalized and readable. If it holds up, it gives the other two results some credibility by association, because the same people and the same AI-heavy workflow are behind them.

The Millennium Prize question for Navier-Stokes won’t be settled by a preprint. It needs independent checking, and that will take months, not days.

Source: What's new

Written by

Saksham Bhargava

Saksham Bhargava runs Alteroid and has been writing about consumer technology for over five years, covering iPhone, Android, Windows, AI tools and gaming. With an MCA and six years of hands-on web development experience, he tests the features he writes about on his own devices and focuses on clear, practical guides that solve real problems. Away from the screen, he enjoys travelling to places most tourists skip.