All Ur Frames Are Belong To Us
Red Dawn Academic Press Inc
Faithful Response Functionals Encode Exactly the Vorticity Supremum — Forthcoming
Running at tornado scale. Right now.
The iDNS solver is currently computing tornado vortex core dynamics at Re = 10⁸ on a single workstation — a Ryzen 9 9950X with 128 GB DDR5. Spectral Galerkin method with RK4 time integration in lifted computational time τ. Lamb-Oseen vortex tube with azimuthal perturbations modeling the suction vortices that spiral inside an EF5 tornado funnel — 10-to-100-meter structures that have never been computed from first principles.
The sigmoid controller engages. Energy is conserved. Enstrophy cascades. The solution does not diverge.
A conventional solver, given the same initial condition at the same Reynolds number, NaNs. Every time. The information exceeds the frame.
We build more frame.
When fluid dynamicists say “frame,” they mean one thing: which spatial coordinate system the equations are written in. Inertial frame — the standard form, balancing acceleration, advection, pressure, and viscous diffusion. Rotating frame — add Coriolis and centrifugal terms for weather and ocean currents. That’s the entire conversation. Spatial coordinates. Pick one, write your equations, run your solver.
Nobody talks about the temporal frame.
Every DNS code on Earth uses a fixed temporal frame. One timestep, one unit of computational resolution per unit of physical time. At low Reynolds numbers, this works. The fluid is well-behaved. The information generated per instant fits inside the frame you gave it.
Crank the Reynolds number to tornado scale — Re > 10⁸ — and the vorticity field generates more information per instant than that fixed frame can hold. The solver produces NaN. Not because the physics broke. Not because the numerics are wrong. Because the frame is too small.
This has been the state of the art for sixty years.
The iDNS (intelligent Direct Numerical Simulation) method introduces a temporal lifting φ that maps computational time τ to physical time t. The key object is φ'(τ) — not a rate controller, not a throttle. A ratio. A unit conversion between computational resolution and physical time.
When the vorticity is high, φ’ allocates more computational frames per unit of physical time. When it’s calm, fewer. Both solutions — the lifted solution in τ and the physical solution in t — co-exist from t = 0. The physics is unchanged. The Navier-Stokes equations are solved exactly as Leray, Hopf, and Fujita-Kato wrote them. What changes is how much frame each instant gets.
In differential geometry, this is a standard construction. Each sub-interval of dτ covering a piece of dt is a coordinate chart. The connected partition tiling [0, ∞) is an atlas. The sigmoid controller that drives φ’ responds to one input: the L∞ norm of vorticity, ‖∇ × u‖. When the flow demands more resolution, it gets it. When it doesn’t, computational effort is conserved.
The controller parameters — three numbers — are invariant from Re = 1,600 to Re = 10⁸. Same sigmoid. Same constants. Every Reynolds number.
The theory. Classical Navier-Stokes treats time as an unbounded external parameter while the spatial domain inherits the compact geometry of T³ = ℝ³/ℤ³. This asymmetry is artificial. Bounded temporal lifting embeds time into the torus geometry via a coupled system (U, φ) where the velocity field and a temporal diffeomorphism evolve simultaneously through a bounded vorticity-response functional Φ. The clock rate φ'(τ) = Φ(‖∇ × U(τ)‖_{L∞}) responds to instantaneous vorticity — giving the equations more computational resolution exactly where vortex stretching demands it. Physical time is always recoverable: t = φ(τ).
The invariance. The Beale-Kato-Majda and Prodi-Serrin regularity criteria are invariant under this lifting. The lifted solution visits the same states as the physical solution; only the speed of traversal differs. This is coordinate invariance, not approximation.
The information-carrier chain. What does Φ carry? For faithful (strictly monotone) response functionals, the answer is exact: a ceiling–floor pair proving Φ encodes precisely the vorticity supremum ‖∇ × U(τ)‖_{L∞} — no more, no less. Six counterexamples witness the sharpness of the faithfulness condition.
The verification. Every theorem is machine-verified in Lean 4 with Mathlib. 8,920 build jobs. Zero sorry. Zero custom axioms. Every #print axioms returns only [propext, Classical.choice, Quot.sound] — Lean’s own foundations.
The solver. iDNS uses a fixed sigmoid controller with three parameters (σ_min, α, c) — unchanged from Re = 10³ to Re = 10⁸. Five orders of magnitude, same controller, no retuning. The Taylor-Green vortex benchmark matches exact dissipation at R_ε = 1.000. True zero numerical dissipation: the temporal diffeomorphism is exact by construction.
Publications:
Global Regularity for Navier–Stokes on T³ via Bounded Vorticity–Response Functionals — SJPBE Vol 1 No 2, Dec 2025. Lean 4 verified.
Invariance of BKM and Prodi–Serrin Integrals under Bounded Temporal Lifting — SJPBE Vol 2 No 1, Mar 2026.
iDNS: True Zero-Dissipation DNS via Deterministic Bounded Temporal Lifting — SJPBE Vol 2 No 1, Mar 2026.
Temporal Lifting as Latent-Space Regularization for Continuous-Time Flow Models in AI Systems — arXiv:2510.09805.
Mathematical Algorithmic Conflation Error in AI Systems Regarding the Navier–Stokes Millennium Prize Proofs — SJPBE Vol 2 No 1, Sep 2026.
[Faithful Response Functionals Encode Exactly the Vorticity Supremum] — Forthcoming