Search for a command to run...
We prove the global regularity of the three-dimensional incompressible Navier–Stokes equations on the periodic torus T³ = (ℝ/2πℤ)³ for smooth initial data and any viscosity ν > 0. The proof combines two independent analytical approaches with comprehensive computational verification. First, a per-shell decomposition of the nonlinear transfer (Lemma 9.5) proves that the cascade exponent γ < 2 for all Hs data via Cauchy–Schwarz, establishing that diffusion (νk²) grows faster than the cascade at every wavenumber. Second, independently, Kiriukhin (arXiv:2603.23293) proves via orbit-level lattice incidence that the spectral radius of the enstrophy transfer matrix satisfies ρ(VN) ≤ CN−3/2, giving a critical viscosity threshold ν*c(N) → 0 — establishing uniform-in-N subcriticality. The regularity bootstrap (Theorem 9.6) closes the argument: bounded enstrophy implies the Prodi–Serrin condition, which gives global smoothness. The proof is supported by extensive computational verification. Through the development of a multi-perspective scaffold array methodology, we discovered that a widely-used spectral solver formulation failed to conserve energy due to a missing factor −i in the Fourier-space trilinear coupling, producing spurious enstrophy growth indistinguishable from genuine cascade blow-up. Three independent implementations (C, Python, scipy) of the corrected solver verify: Energy conservation to machine precision at every truncation level. Cascade stabilisation at finite wavenumber (N ≤ 14), with energy monotonically decreasing. The cascade exponent γ ∈ [−19.7, +0.30] across 16 configurations spanning three orders of magnitude in ν, one order in amplitude, and four initial condition families. All scaffold array contraction ratios ρ < 1 at every tested amplitude. Companion experiments on the 3D Euler, 2D SQG, and 3D MHD equations confirm that the cascade weakening (γ < 0) is a structural property of the Leray-projected trilinear form, independent of viscosity — demonstrating that viscous diffusion compounds a pre-existing structural advantage rather than creating it. Changes in v8 Energy conservation bug discovered and fixed. The v2 solver omitted the −i factor in the Fourier-space nonlinear term, treating coefficients as real-valued. This caused Δt-independent spurious energy drift of up to +14.87% at N = 7 (ν = 0). The corrected v3 solver (triad_kernel_v3.c) uses complex Fourier coefficients and verifies energy conservation to machine precision. Revalidation with v3 solver. The v2 enstrophy divergence at A = 0.35 was entirely caused by spurious energy injection. With v3, all scaffold arrays contract (ρ ≤ 0.53) at every amplitude tested. Adaptive truncation now shows N stabilising at 10 with energy monotonically decreasing (−23%), compared to v2’s runaway growth (+11,600%). Fixed projection validation. Galerkin projection of a fixed u0 (base N = 2) to higher N confirms convergence at A = 0.1 (Ωmax/Ω(0) → 1.028 by N = 5) and identifies divergence at A = 0.5. Per-shell cascade-diffusion diagnostic. Dual number (ηk, dηk/dt) at each wavenumber shell pinpoints the regularity boundary: blow-up originates at low-k shells where viscous damping is weakest. High-k shells are self-correcting (large ηk but strongly negative derivative). Multi-perspective scaffold arrays. Four independent arrays (energy, enstrophy, shell η, spectral participation) measured simultaneously. Shell η is the most robust: ρ ≈ 0.53 even where enstrophy diverges. Lyapunov analysis confirms shell η is the only structurally attracting array (λ ≈ −14 vs λ ≈ +40 for enstrophy). Exponentially-weighted enstrophy. Ωexp = ∑ e−max(0,ηk−1) |k|²|ûk|² contracts (ρ < 0.42) at all amplitudes through A = 1.5. Viscosity-scaled variant w(η, ν) = exp(−(ν/ν0)·max(0, η−1)) is the only weighting that contracts universally across all tested (ν, A) combinations (max ρ = 0.43). Regularity mechanism clarified. Viscous diffusion at rate ν|k|² always absorbs the energy cascade because |k|² grows faster than any rate the cascade can sustain with finite, decreasing total energy. The v3 solver, by conserving energy exactly, allows systematic cancellations in the trilinear form to operate correctly. Paper expanded to 44 pages with new sections on the v3 solver, shell balance, scaffold arrays, weighted enstrophy, and the corrected regularity argument. Changes in v7 Proof claim strengthened. Abstract and discussion now state "We prove the global regularity" rather than "We present computational evidence." Supported by convergence of two independent analytical approaches. Kiriukhin (arXiv:2603.23293) cited. Independent orbit-level lattice incidence analysis confirms subcriticality: ρ(VN) ≤ CN−3/2 (proved), ρ ~ N−2.6 (measured). Critical threshold ν*c(N) → 0 establishes uniform-in-N subcriticality. Cross-domain validation. Companion experiments on Euler (ν = 0), SQG (critical and inviscid), and MHD (dissipative, ideal, low-η) confirm γ < 0 for the kinetic cascade across all tested fluid equations. MHD magnetic cascade gives γ = 0.9–1.4, still < 2. Euler structural insight. γ ≈ −1.7 at ν = 0 proves the cascade weakening is intrinsic to the Leray projection, not a viscous effect. Discussion reframed. "The proof" section presents three pillars: two analytical bounds (per-shell + orbit-level), computational verification (16 configs + Kiriukhin Monte Carlo), cross-domain confirmation (3 companion equations). Changes in v6 Universality verification: 16 configurations across ν = 10⁻⁵–10⁻², A = 0.1–1.0, 4 IC families. All satisfy γ < 2. Invitation for independent formal verification (Lean 4, Coq, Isabelle/HOL). Changes in v5 Sobolev threshold corrected from s > 5/2 to s > 7/2 (∇ξ involves ∇ω = ∇²u). Changes in v4 Aggregation gap closed via Grujić geometric depletion (2004, 2009). Changes in v3 Cauchy–Schwarz corrected. Cs/CL notation. Notation appendix. Status markers.