16 April 2026 · 8 min read

One Identity, and Where It Stops

The energy identity gives a finite dissipation budget. Turning that into an enstrophy bound needs hypotheses, and explicit smooth fields show where they fail.

2ν ∫₀^∞ Ω(t) dt ≤ E(0)

The total viscous dissipation over all time is bounded by the initial kinetic energy.

Every smooth solution of the 3D incompressible Navier–Stokes equations satisfies one exact identity. With E the kinetic energy and Ω the enstrophy,

dE/dt = −2νΩ.

Integrate it and you have the line above. The fluid starts with a finite budget and viscosity spends it at rate 2νΩ. The enstrophy is integrable in time. That is what the identity gives, and it is the one thing about the problem that is not in doubt.

What the Identity Gives

It gives a time-integrated bound: Ω is in L¹ of time. It does not give a bound on Ω at any particular moment, and regularity needs the second kind. A uniform-in-time enstrophy bound puts u in L^∞(H¹), which embeds in L^∞(L⁶), the Prodi–Serrin class, and the solution is smooth for all time. The gap between “Ω integrable” and “Ω bounded” is the whole problem. Nothing in the identity prevents Ω from being large for a short time, which is exactly what a singularity would do.

A Conditional Bound on the Active Shells

In the Energy Argument note I take the identity as far as it goes under explicit cascade hypotheses. Decompose the flow into wavenumber shells K, with energy E_K and flux Π_K through each shell. Suppose three things:

Uniform intra-shell dispersion. The energy in each shell is spread across its modes with a dispersion constant that does not depend on K.
Pointwise Kolmogorov closure. The flux through every shell is bounded by the local turnover: |Π_K| ≤ αK E_K^3/2 for every K.
Cascade contiguity. Energy reaches a shell only through the shells below it.

Under these three hypotheses the enstrophy carried by the active shells satisfies

Ω_A ≤ C · E(0)^5/3 / ν^4/3.

The cascade cannot reach beyond the dissipation scale, the shells it does reach carry bounded enstrophy, and any uniform-in-time enstrophy bound gives regularity by Prodi–Serrin. That is a correct conditional statement. The whole question is whether the hypotheses are properties of smooth flows.

Where It Stops

They are not. Explicit smooth divergence-free fields violate them at t = 0, before any dynamics has happened.

A spectral gap

Take a datum with no energy in shell 10, E₁₀ = 0. The flux through that shell at t = 0 is Π₁₀ = −1.04. The closure requires |Π₁₀| ≤ αK · 0. It fails outright.

Vortex blobs

Localised vortex blobs of size ℓ carry flux at their own scale that exceeds the closure by a factor of order ℓ⁻³: 13.6× at ℓ = 0.25, rising to 55× at ℓ = 0.15. The smaller the structure, the worse the closure.

The dissipation range

For generic data in the exponential dissipation range the closure is exceeded by 7.5×, 30× and 98× at K = 40, 48, 52. The ratio grows with K.

A shear flow

For u = f(x₁)e₂ every shell has exactly two modes, so the dispersion constant is n_K/2: 9, 31, 175, 625, 2517 at K = 1, 2, 5, 10, 20. It is not uniform in K. It grows like K².

None of these fields is exotic. A spectral gap, a compact vortex, a plane shear: these are the first things one writes down. The vortex-blob case is the informative one. A closure of the form |Π_K| ≤ αK E_K^3/2 says the flux through scale K is controlled by the energy already at scale K. A concentrated structure of size ℓ puts flux at scale 1/ℓ with energy that is small at that scale, by a factor of the volume ℓ³. The closure is, precisely, the statement that energy does not concentrate.

What This Clarifies

A pointwise-in-K spectral closure excludes concentration a priori. Concentration is the only way a smooth solution could form a singularity. So a closure of this kind cannot be a hypothesis of a regularity theorem: it assumes the conclusion in spectral clothing. The same holds for uniform intra-shell dispersion, which the shear flow breaks exactly, shell by shell.

That fixes where a regularity argument has to work. The energy identity supplies the budget. The lattice supplies the set of interactions. Neither controls concentration. What remains is the covariance between strain and the vorticity direction at the flux scale, and the concentration branch itself, where a first singularity zooms to an ancient Navier–Stokes solution (Type I) or an ancient Euler solution (Type II) and the question becomes a Liouville statement for those limit objects. A regularity proof has to handle concentration directly, not by hypothesis.