Navier-Stokes can blow up with no outside force, Buckmaster and Alpöge say

The Next Web
Read full story· 3 min
Tristan Buckmaster and Levent Alpöge posted a proof showing the Navier-Stokes equations can blow up in finite time without external force, addressing a Millennium Prize Problem. Their AI-assisted proof includes a Lean formalisation and follows a computer-assisted approach. They also released a proof for the Euler equations without viscosity, with mathematician Terence Tao noting the need for further refinement.

What happened

Tristan Buckmaster of New York University and Levent Alpöge of Anthropic released a proof on GitHub showing that the Navier-Stokes equations can develop a blow-up in finite time without any external force. Their work, posted on November 20, 2023, uses AI tools and includes a formal verification in the Lean proof assistant. The same day, they also published a proof for the Euler equations without viscosity, again with no external force. Mathematician Terence Tao reviewed the Navier-Stokes proof and noted that while the strategy is clear, the write-up needs further refinement.

Key details

  • The Navier-Stokes proof paper is 79 pages long and was posted on November 20, 2023.
  • The Lean formalisation for Navier-Stokes includes about 2,270 files and 600,000 lines of code.
  • A certificate of numerical data referenced in the paper has not yet been released.
  • The Euler equations proof, co-authored with Matei Coiculescu, is 55 pages and also comes with a Lean formalisation.
  • Buckmaster and Alpöge used Anthropic's Claude and OpenAI's Codex models throughout their work.

Why it matters

Mathematicians and researchers working on fluid dynamics and the Millennium Prize Problems now have a new AI-assisted approach to the Navier-Stokes blow-up question, specifically for the case with no external force. The release of a formalised proof and supporting code allows others to check and build on the result, though some components, such as the numerical certificate, are still pending. Those in the mathematical community must now evaluate the completeness and correctness of the proof, as further refinement is needed before it can be fully accepted.


More in AI Research