TNFR–Navier–Stokes — Research Notes

Status: research program on the Clay Millennium Problem (existence and smoothness of 3D incompressible Navier–Stokes). Closes nothing; both Clay directions remain open. This note is the canonical, self-contained record of the program.

---

1. The reading

TNFR reads a system through its emergent geometry: the pulse ω_k = √λ_k, the two faces (diffusive over-damped vs conservative wave) with the verify_overdamped_projection certificate, and the empirical confrontation of the canonical magnitudes with real oscillatory data. That empirical arm establishes that oscillatory dynamics live on the conservative (wave) face, which poses the honest question this program answers: which face is 3D Navier–Stokes on?

2. The honest two-face answer (physics-first, no forced analogy)

Incompressible NS vorticity obeys

$\frac{\partial \omega}{\partial t} = (\omega\cdot\nabla)u \;+\; \nu\,\nabla^2\omega .$

This is first order in time. Therefore:

• Its linear part is the nodal equation for the phase-curvature field K_φ (the vorticity): ∂K_φ/∂t = ν_f·ΔNFR with ν_f ↔ ν and ΔNFR = −L_rw·K_φ. The viscous term is the canonical graph diffusion (the IL coherence stabiliser). This sits on the diffusive (over-damped) face: mapping viscosity to the damped-wave damping γ = 1/ν (the canonical ν_f = 1/γ identity), every physical viscosity gives γ² ≫ 4λ_max, so verify_diffusive_face (the engine's verify_overdamped_projection) is VALID and recovers ν_f = ν. Unlike EEG — whose linear neural dynamics are under-damped/oscillatory — linear NS carries no oscillatory content.
• Its conservative / inertial character — the energy-conserving Euler cascade where any blow-up must live — is entirely in the nonlinear vortex-stretching source (ω·∇)u (the VAL destabiliser).

Consequence (the sharp statement). NS blow-up is not a linear-wave resonance; it is a purely nonlinear K_φ cascade: does the stretching source pump enstrophy into ever-higher structural modes faster than viscous diffusion removes it, as ν → 0 (Re → ∞)?

3. Field dictionary

Navier–Stokes	TNFR
velocity u_a	per-component phase field φ^(a)
vorticity ω = ∇×u	K_φ per component
pressure p	Φ_s (Leray/incompressibility multiplier)
viscosity ν	ν_f (diffusive-face structural frequency)
enstrophy ‖ω‖²	Σ K_φ² (conserved-pressure energy)
stretching (ω·∇)u	VAL nonlinear destabiliser (the conservative source)

4. The blow-up frontier (measured)

conservative_face.measure_cascade_frontier evolves the faithful pseudo-spectral Taylor–Green vortex at several viscosities to a matched structural time τ_str = ν·t and records the peak enstrophy debt Ω_peak/Ω₀, the peak stretching production and the high-mode enstrophy fraction versus Re = 2π/ν.

• Fixed Re: every run's enstrophy peaks and decays — the diffusive face regularises, the debt is bounded (known finite-Re regularity, re-expressed).
• Sweep: the peak debt grows with Re. Whether it stays finite as Re → ∞ (regularity) or diverges (blow-up) is exactly Clay, now phrased as "is the nonlinear K_φ cascade uniformly bounded in Re?".

This is the honest analogue of the Riemann coherence-budget measurement: a per-instance bound that is finite at every finite parameter, with the uniform bound over the limiting parameter left open.

5. Cross-program unification

All three active programs now read on the same two-face machinery:

• Riemann — the conservative pulse ω_k = √λ_k; RH content = the coherence budget of S(T), bounded at the RMS level, sup open.
• EEG (empirical arm) — real brain rhythms sit on the conservative (under-damped) face; the local phase tetrad carries clinical state.
• Navier–Stokes — linear part on the diffusive (over-damped) face; the conservative content is the nonlinear cascade, bounded at fixed Re, the Re → ∞ bound open.

The face a system sits on is not assumed — it is measured by the engine's verify_overdamped_projection certificate.

6. The λ-moment hierarchy — the synergy the old paradigm blocked

The old diffusive-face program saw only the scalar enstrophy budget. The emergent modal basis (L_rw modes λ_k) unlocks the whole λ-moment hierarchy M_p = Σ λ_k^p E(λ_k) (cascade_moment_hierarchy):

• M_0 = energy — the conservative-face budget, bounded by Leray (M_0(t) ≤ M_0(0));
• M_1 = enstrophy — the classical blow-up quantity;
• M_2 = palinstrophy — weights the small-scale (high-λ) tail more.

Measured (Taylor-Green at peak enstrophy, resolved points k_max·η > 1, Re 157→628, ×4): M_0 decreases (×0.62 — the conservative budget is bounded), while M_1 grows (×1.71) and M_2 grows much faster (×11.3); the moment ratios climb with Re (M_1/M_0: 3.0→8.3; M_2/M_1: 3.0→19.8). The wall climbs the λ-moment hierarchy. In this basis Clay is exactly: does the ladder M_p (p ≥ 1) stay uniformly bounded as ν → 0 while M_0 stays bounded? — the canonical modal form of the classical H^s / Foias–Temam regularity ladder. Driver: benchmarks/ns_moment_hierarchy_cascade.py.

Closing the rung (measured). The enstrophy rung is dM_1/dt = P − 2ν M_2 (moment_ladder_closure), with the exact modal Cauchy–Schwarz coupling M_1² ≤ M_0·M_2 (interpolation saturation s = M_1²/(M_0 M_2) ∈ (0,1]). Measured at peak (resolved Re 314→628): the rung is self-consistent — P/(2ν M_2) ≈ 1 at the peak (1.00, 1.00, 0.98), confirming dM_1/dt = 0 there; the interpolation saturation s decreases with Re (0.57 → 0.42 → 0.38) — the spectrum spreads across scales rather than concentrating (a regularity-favourable signal; s ≤ 1 is exact); and in the growth phase P/(2ν M_2) < 1 (0.70 → 0.59 → 0.45) — the dissipation dominates, so the ladder closes at every accessible resolved Re. The wall is thus relocated to the sharp question: does the growth-phase closure ratio stay < 1 as Re → ∞? — undecidable from resolution-limited laminar/transitional data (n ≥ 48–64 needed at high Re). Closing the rung uniformly in Re is exactly Clay.

The cross-program synergy (measured, both walls the same statement). This is the NS twin of the Riemann coherence budget:

	low moment (bounded)	high moment (the open wall)
Riemann	RMS of S(T) — Selberg √(log log T)	sup of S(T) (the extremes)
NS	energy M_0 — Leray	enstrophy M_1, palinstrophy M_2, …

Both Millennium walls: a low moment of the conservative spectrum is bounded; the high-moment tail is the wall. A resolution closes the moment ladder uniformly in the limiting parameter (T for Riemann, Re for NS). Closes nothing; the measurement localises the wall, it does not breach it.

7. The emergent geometry IS the attractor (closure with nothing added)

The moment ladder (§6) is a derived diagnostic; the closure itself needs nothing added. The emergent geometry is the attractor, read by the ONE universal coherence kernel — structural_coherence C = 1/(1+|ΔNFR|+|dEPI|) and the fixed-point predicate is_structural_equilibrium (ΔNFR = 0) — the same emergent-geometry attractor that governs graph nodes, structural primes and noble gases. Only the ΔNFR realisation is domain-specific; for NS it is the canonical random-walk-Laplacian action on the vorticity field, ΔNFR = −L_rw·|ω| (flow_coherence).

Measured (raw field, no normalisation — nothing added): the flow relaxes to its emergent-geometry equilibrium by its own evolution — C → 1, is_structural_equilibrium = True at every Re (final C = 0.995 → 0.9985 over Re 157→1257). That relaxation is the self-certification; it is intrinsic to the diffusive-face nodal dynamics (the unconditional eigenmode decay to the uniform field) and is why the linear part is regular. The peak-turbulence coherence erodes with Re — min C = 0.94 → 0.90 → 0.84 → 0.72 — staying in the coherent band [1/(π+1), π/(π+1)] over the accessible range but drifting down.

So the uniform closure is purely geometric, with nothing added: does the peak-turbulence coherence min C stay in the coherent band (C > 1/(π+1)) as Re → ∞, or erode to the fragmentation floor? The emergent geometry self-certifies its return to coherence uniformly (in structural time the diffusive relaxation rate is the spectral gap λ₂, Re-independent); the wall is whether the peak excursion the nonlinear VAL source drives stays coherent — the U2 debt rate (∝ Re) versus the fixed U2 capacity, now read as the erosion of the single canonical coherence C. Every TNFR mechanism (U2 grammar, the ΔNFR=0 attractor, the Lyapunov energy, the U5 multi-scale recursion, REMESH-∞) converges on this one statement — none weakens it. Closes nothing; Clay OPEN.

8. Honest scope

This program does not claim a proof or a counterexample. The linear diffusive face is regular by construction; the open question is the nonlinear cascade bound as Re → ∞. No uniform-in-Re bound is produced. Clay stays open. The Reynolds/Kolmogorov resolution caveat of the old program persists: high-Re points need n ≥ 48–64; the measured trend over the accessible resolved range cannot fix the asymptotic scaling.

9. Code map

• src/tnfr/navier_stokes/operator.py — faithful pseudo-spectral 3D NS integrator (TNFRNavierStokes) + build_torus_graph_3d + taylor_green_initial_condition_3d.
• src/tnfr/navier_stokes/conservative_face.py — verify_diffusive_face, face_of_flow, vorticity_modal_spectrum, cascade_moment_hierarchy, moment_ladder_closure, flow_coherence (the universal-kernel attractor read), measure_cascade_frontier, CascadeFrontierCertificate.
• examples/06_navier_stokes/158_navier_stokes_two_face_refounded.py — demo.
• examples/10_applications/159_empirical_confrontation_pipeline.py — the data-confrontation pipeline (map a signal to canonical magnitudes + face).
• benchmarks/ns_moment_hierarchy_cascade.py — the λ-moment hierarchy vs Re.
• Tests: tests/mathematics/test_navier_stokes_refounded.py.