Navier–Stokes existence and smoothness

At least 91 years old · documented by

Let ν>0\nu>0. The unknowns are the velocity field u:R3×[0,∞)→R3u:\mathbb{R}^3\times[0,\infty)\to\mathbb{R}^3 and the pressure p:R3×[0,∞)→Rp:\mathbb{R}^3\times[0,\infty)\to\mathbb{R}. Given an external force ff, the incompressible Navier-Stokes equations are

∂u∂t+(u⋅∇)u=νΔu−∇p+f,∇⋅u=0,\frac{\partial u}{\partial t}+(u\cdot\nabla)u =\nu\Delta u-\nabla p+f, \qquad \nabla\cdot u=0,

with initial condition

u(x,0)=u0(x).u(x,0)=u_0(x).

Here Δ=∑j=13∂2/∂xj2\Delta=\sum_{j=1}^3\partial^2/\partial x_j^2, and the initial velocity must be divergence-free:

∇⋅u0=0.\nabla\cdot u_0=0.

Admissible data and solutions

In the whole-space formulation, u0u_0 is smooth and rapidly decreasing: for every multi-index α\alpha and every K≥0K\geq 0 there is a constant Cα,KC_{\alpha,K} such that

∣∂xαu0(x)∣≤Cα,K(1+∣x∣)−K.\left|\partial_x^\alpha u_0(x)\right| \leq C_{\alpha,K}(1+|x|)^{-K}.

When a force is allowed, it is smooth and rapidly decreasing in space and time: for every α,m,K\alpha,m,K,

∣∂xα∂tmf(x,t)∣≤Cα,m,K(1+∣x∣+t)−K.\left|\partial_x^\alpha\partial_t^m f(x,t)\right| \leq C_{\alpha,m,K}(1+|x|+t)^{-K}.

A physically reasonable whole-space solution is globally smooth,

u,p∈C∞ ⁣(R3×[0,∞)),u,p\in C^\infty\!\left(\mathbb{R}^3\times[0,\infty)\right),

and has uniformly bounded kinetic energy:

sup⁡t≥0∫R3∣u(x,t)∣2 dx<∞.\sup_{t\geq 0}\int_{\mathbb{R}^3}|u(x,t)|^2\,dx<\infty.

In the periodic formulation, space is the three-torus T3=R3/Z3\mathbb{T}^3=\mathbb{R}^3/\mathbb{Z}^3. The fields are 11-periodic in each coordinate,

u(x+ej,t)=u(x,t),p(x+ej,t)=p(x,t),f(x+ej,t)=f(x,t)u(x+e_j,t)=u(x,t),\qquad p(x+e_j,t)=p(x,t),\qquad f(x+e_j,t)=f(x,t)

for j=1,2,3j=1,2,3. The initial velocity is smooth, periodic, and divergence-free. When a force is allowed, all its space-time derivatives decrease faster than any power of time:

∣∂xα∂tmf(x,t)∣≤Cα,m,K(1+t)−K.\left|\partial_x^\alpha\partial_t^m f(x,t)\right| \leq C_{\alpha,m,K}(1+t)^{-K}.

The problem

The Clay problem is solved by proving at least one of the following four statements.

(A) Global regularity on R3\mathbb{R}^3. For every smooth, rapidly decreasing, divergence-free u0u_0, with f≡0f\equiv 0, there exist global smooth uu and pp satisfying the equations and the bounded-energy condition.

(B) Global regularity on T3\mathbb{T}^3. For every smooth, periodic, divergence-free u0u_0, with f≡0f\equiv 0, there exist global smooth periodic uu and pp satisfying the equations.

(C) Breakdown on R3\mathbb{R}^3. There exist smooth, rapidly decreasing, divergence-free initial data u0u_0 and a smooth rapidly decreasing force ff for which no global smooth bounded-energy solution exists.

(D) Breakdown on T3\mathbb{T}^3. There exist smooth periodic, divergence-free initial data u0u_0 and a smooth periodic force ff with the stated time decay for which no global smooth periodic solution exists.

Thus the question is whether arbitrary smooth three-dimensional incompressible data remain smooth for all time, or whether some admissible data develop a finite-time singularity.

References

Additional references

  1. Charles L. Fefferman, Existence and Smoothness of the Navier-Stokes Equation, the official Clay Mathematics Institute problem description, including its appended errata.
  2. Clay Mathematics Institute, Navier-Stokes Equation, Millennium Prize Problems overview.
  3. J. Leray, "Sur le mouvement d'un liquide visqueux emplissant l'espace," Acta Mathematica 63 (1934), 193-248.
  4. L. Caffarelli, R. Kohn, and L. Nirenberg, "Partial regularity of suitable weak solutions of the Navier-Stokes equations," Communications on Pure and Applied Mathematics 35 (1982), 771-831.

Progress summary

Refreshed
Claimed solved

OpenAI announced a counterexample claiming that some three-dimensional fluid flows become singular in finite time, but the claim has not been independently confirmed.

The Clay problem asks whether every admissible smooth three-dimensional incompressible flow remains smooth for all time, or whether some admissible data produce finite-time breakdown. The formulation permits either global regularity or breakdown in the whole space or on the three-torus.

Known results and September 2026 claim

  • Leray, 1934: global finite-energy weak solutions and local smooth solutions; small-data global smoothness is also known.
  • Caffarelli, Kohn, and Nirenberg, 1982: suitable weak solutions have a singular set of parabolic measure zero.
  • Escauriaza, Seregin, and Šverák: an Lt∞Lx3L^\infty_tL^3_x bound excludes singularities.
  • On September 8, 2026, OpenAI claimed an analytical construction satisfying breakdown alternatives CC and DD, with formalization in Lean by GPT-6 Astra. Terence Tao reported on September 11 that the problem remained open; related forced-blowup results concern Euler and other equations, not Navier–Stokes.

Community submission (unverified), September 11, 2026

A submitted argument develops vorticity amplitude-direction identities, angular decompositions, and dyadic estimates, but explicitly asserts no global-regularity conclusion.

Current status (as of September 2026): The Navier–Stokes problem remains open; OpenAI’s claimed counterexample for CC and DD is unverified.

  • AstraOpenAIsolved2026-09-08evidence

    OpenAI credits GPT-6 Astra with the subsequent Lean formalization and verification of its announced Navier-Stokes result. A separate unreleased internal model produced the original proof. This records Astra's assistance with the claimed solution, not independent discovery or independent mathematical confirmation.

Sources

Solutions 1

ProofQuadrupolar Structure, Directional Decomposition, and Critical Cancellation in the Three-Dimensional Navier–Stokes EquationsSee full solutionHide full solution

We develop an exact structural formulation of the vorticity dynamics for the three-dimensional incompressible Navier–Stokes equations. Writing ω=ρξ with |ξ|=1, we derive the amplitude–direction decomposition, the exact viscous splitting, the directional stretching identity, and the trace-free quadrupole tensor formulation. We then identify the signed angular structure of the Biot–Savart strain and derive the corrected mixed amplitude–direction pairing. The angular mean-zero and L² identities isolate a finite-dimensional, quadrupolar interaction channel. A dyadic formulation is given using Littlewood–Paley projectors. The results are structural identities and proved estimates; no global-regularity conclusion is asserted.

  • Article_I_Quadrupolar_Structure_DeWatson-Ledetsambali.pdf286,450 bytesOpen
  • Article_II_Critical_Quadrupole_Endpoint_DeWatson-Ledetsambali.pdf275,753 bytesOpen
  • navier-1-compressed.pdf6,679,767 bytesOpen
  • Article_III_Conditional_Global_Regularity_DeWatson-Ledetsambali.pdf268,163 bytesOpen