Partial differential equations and fluid mechanics
Navier–Stokes existence and smoothness
Which admissible three-dimensional viscous flows remain smooth, and can a smooth force generate a finite-time singularity?
Clay: Active
Status and source review: 2026-11-19
State the mathematical question precisely.
For incompressible velocity u, solve ∂ₜu + (u·∇)u = −∇p + νΔu + f, with div u = 0 and ν > 0. Fefferman's alternatives A/B require global smooth solutions for every admissible smooth divergence-free initial velocity with f = 0 on R³ or the periodic torus. C/D permit a counterexample with forcing satisfying the stated smoothness and decay conditions.
Choose the exact alternative and retain its domain, initial-data, forcing, global-time and energy requirements.
For a forced breakdown construction, prove that f extends smoothly through and beyond the singular time, with every required derivative and decay bound.
Establish genuine blowup and exclude a globally smooth solution with the same data; a numerical spike or a singular forcing term does not suffice.
Trace the results that changed the problem.
Leray constructs global weak solutions.
Weak solutions and an energy inequality establish a durable existence framework; they do not by themselves establish global classical smoothness.
Partial regularity restricts possible singular sets.
Caffarelli, Kohn and Nirenberg show that the singular set of suitable weak solutions has zero one-dimensional parabolic Hausdorff measure. Isolated singularities are still compatible with that bound.
A forced resolution is announced and Clay changes its category.
OpenAI publishes a claimed C/D construction and Lean repository. Clay acknowledges an apparent resolution and describes a deliberately unhurried evaluation process; its current category is Active.
Analytic forcing exposes a precise structural boundary.
Constantin, Ignatova and Vicol study a related symmetry-and-scaling class with analytic forcing. Hou's subsequent commentary explains why this does not refute the smooth nonanalytic construction.
Separate established results from the remaining claim.
The current claim has explicit forcing hypotheses.
Theorem 1.1 in OpenAI's manuscript claims that, for every ν > 0, zero initial velocity and a force f in C∞c(R³ × (0,∞)) produce a flow smooth before t = 1, with bounded L² norm and unbounded L∞ norm as t approaches 1. A periodic corollary is claimed. This dossier inspected those statements and did not independently verify the 166-page proof or replay Lean certificates.
Analytic forcing is a stricter condition than smooth forcing.
The Constantin–Ignatova–Vicol preprint proves regularity with exact axisymmetry in a collapsing core, specified anisotropic Type II bounds, and spatially analytic forcing under its stated uniformity assumptions. It does not prove general unforced regularity, and its assumptions differ from the announced smooth nonanalytic forcing.
Clay's announcement does not announce an awarded prize or completed adjudication. OpenAI's announcement states that it does not intend to claim the prize. The source record therefore separates an announced resolution, published formalization, independent validation and prize decision.
A formal check must prove the intended analytic hypotheses rather than assuming them as axioms; matching the formal statement to C/D is part of the work.
Make each proposed approach answer for its obstacle.
These are research directions and diagnostic tests. Their inclusion does not establish novelty, feasibility at full scale, or a solution.
Audit one proof dependency at a time.
A concrete next research task is to isolate smooth force extension, summability of corrections and the comparison argument. Pin the manuscript and formalization revision before reconstructing the dependency graph.
Obstacle. A successful build alone cannot show that formal definitions match the physical PDE; conversely, an explanatory diagram does not check every estimate.
Try the next discriminating test.
Replay the published independent-checking procedure in an isolated approved environment, inspect theorem assumptions and identify the exact statement checked. That replay has not been performed here.
Continue the unforced regularity investigation with a stronger invariant.
Seek a scale-critical quantity whose evolution uses the actual nonlinearity and pressure structure. Compare any candidate estimate against known model counterexamples.
Obstacle. The energy inequality is too weak by itself, and the published analytic-forcing obstruction blocks a direct transfer of the announced mechanism within its specific structural class.
Try the next discriminating test.
Derive the candidate differential inequality before discretizing it; record which stretching or nonlocal pressure term prevents closure.
The basic energy norm loses control at small scales.
For the unforced equation on R³, use u_λ(x,t) = λu(λx,λ²t). A change of variables gives ||u_λ||²₂ = λ⁻¹||u||²₂ and ||u_λ||₃ = ||u||₃.
At λ = 1, 4, 16, the squared-energy scale factors are 1, 1/4, 1/16, while the L³ factor stays 1. Increasing concentration can therefore reduce the quantity controlled by the energy estimate.
This dimensional calculation identifies a limitation of the unforced energy-only route. It neither constructs blowup nor independently validates the announced forced solution.
If the announced forced C/D proof survives independent scrutiny.
It resolves an allowed breakdown alternative in the official prize formulation; it does not automatically decide unforced global smoothness or show that the constructed mechanism occurs in ordinary physical flows.
These equations are classical continuum dynamics. Connections to statistical mechanics, hydrodynamic limits and field-theory methods require explicit assumptions; quantum terminology does not repair an unclosed regularity estimate.
Terence Tao / arXiv; JAMS 2016 · Blowup in an averaged equation preserving energy cancellation; limitation on arguments using only that coarse structure.
Peter Constantin, Mihaela Ignatova and Vlad Vicol / arXiv · Regularity under exact core axisymmetry, specified anisotropic bounds and analytic forcing; a structural obstruction to carrying the announced forced mechanism into the unforced setting.
Thomas Y. Hou / SIAM News, 2 November 2026 · Author's expert commentary on forcing assumptions, proof-assessment obligations and physical interpretation.