The Navier–Stokes Problem Solved: How 10 000 Agents Proved Mathematical Singularity

The Smoothness Problem for Solutions of the 3D Navier–Stokes Equations: Historical Context

The Navier–Stokes equations, formulated in the nineteenth century, are the mathematical foundation of hydrodynamics, aerodynamics, climate-system modeling, and blood circulation. Despite their widespread applied numerical solution, a fundamental question remained open for about 90 years: do physically smooth initial conditions guarantee that solutions remain smooth over time, or can nonlinear flow self-amplification effects lead to a mathematical singularity (infinite velocity and vorticity in finite time)?

In the year 2000, the Clay Institute of Mathematics included the problem of the existence and smoothness of solutions to the Navier–Stokes equations on its list of seven “Millennium Prize Problems.” To date, no rigorous analytical answer has been obtained for the general three-dimensional case.

Research Swarm Architecture: 10 000 Interacting Agents

To tackle the fundamental mathematical problem, a distributed multi-agent environment was deployed based on an experimental next-generation research model with an extended reasoning window:

• Swarm size: approximately 10 000 specialized agents worked simultaneously in a closed-loop hypothesis-verification cycle;
• Subgroup specialization: parallel agent clusters tested various analytical approaches — the method of asymptotic expansions, energy inequalities, solution behavior in twisted spiral vortices, and the spectral properties of operators;
• Tool stack: agents had autonomous access to an indexed scientific-literature database and interactive Python/C++ environments for numerical modeling, and recursively exchanged intermediate lemmas through a central coordinator based on a specialized codebase.

Scale of Computation and Pipeline Time Requirements

The amount of hardware resources expended exceeded the figures for any classical research computations in theoretical mathematics:

Essence of the Discovered Solution: Explosive Vorticity Singularity

The resulting proof corresponds to variants C and D of the Clay Institute’s official formulation (disproof of global solution smoothness):

• Flow construction: in three-dimensional incompressible space, the existence of an initially smooth velocity vector field with finite kinetic energy was proved; under the action of a smooth external force of finite power, it collapses in finite time ^*$;
• Collapse mechanism: at its core lies the geometry of a contracting vortex ring (Swirling Vortex Filament), where the velocity gradient and vorticity ($\nabla u$ and $\\omega = \operatorname{rot} u$) tend to infinity according to a power law through the self-amplification of the vortex tube;
• Strictness of the conditions: the model proved that the loss of smoothness is caused by the internal nonlinear dynamics of momentum transport, rather than by artificial energy injection from an external singular source.

Formalization in Lean and the Role of Automated Proofs

The main obstacle in evaluating machine-generated discoveries has traditionally been reviewers’ distrust of possible hidden errors in lengthy mathematical derivations. To eliminate the human factor, the analytical draft was translated into the formal language Lean (an interactive system for verifying mathematical theorems).

During 17 hours of work by the formalizer, every lemma, energy-integral estimate, and limiting transition was checked by the Lean prover’s kernel without a single heuristic omission. This makes the proof mathematically incontrovertible at the level of machine syntax, opening a new era in proof-based science: multi-agent systems take on the generation and verification of hypotheses inaccessible to individual research teams.