Documentation

LeanPool.CaffarelliKohnNirenberg.Core.Step3.GradientSlotDuhamelTested

Gradient Slot Duhamel Tested #

Part of the Caffarelli–Kohn–Nirenberg partial regularity proof.

Gradient Slot Duhamel Transfers #

Part of the Caffarelli–Kohn–Nirenberg partial regularity proof.

Scalar and pressure forms of the localized gradient transfers #

The paper label lem:local-equation records the localized form of the Caffarelli–Kohn–Nirenberg equation tested against a product cutoff φ · ψ. The two statements below extract from it the two ingredients used when the cutoff is frozen in the time variable and only the spatial slot structure matters:

Scalar form of the diffusion transfer of lem:local-equation: testing the vector localized equation with the vector field whose i-th component is a scalar cutoff ψ and whose other components vanish collapses the transfer identity to the scalar identity in the i-th coordinate. No divergence information is used beyond what IsSuitableWeakSolutionIntegrable already provides.

Pressure transfer of lem:local-equation against the product cutoff: the pairing of the pressure with the spatial derivative of φ · ψ equals the pairing of the weak pressure gradient with φ · ψ. The pressure enters the localized equation only through its weak gradient, so this transfer carries no solution hypothesis: all it needs is the weak-gradient pairing rule on the support box of φ.

Gradient Slot Duhamel Split #

Part of the Caffarelli–Kohn–Nirenberg partial regularity proof.

Regrouped integrands for the local equation #

The tested form of the localized equation (lem:local-equation) pairs the divergence-form sources localizedDivergenceG, localizedDivergenceH with a test field and its first spatial derivative. The paper writes the same expression after moving the derivative of the product φ * ψ onto the cutoff factor, which is what the identities below record.

gradientSlot_divergence_integrand is the divergence-form regrouping: the source integrand together with the derivative slot equals the time, force, convection, derivative and pressure contributions collected on the right. gradientSlot_gradient_integrand is the corresponding regrouping for the pressure-gradient slot of localizedGradientSourceG, where the second spatial derivatives of the cutoff appear explicitly; it needs no smoothness hypotheses because the product rule is not used.

theorem CKN.Core.Step3.gradientSlot_divergence_integrand {φ ψ : Foundation.Parabolic.Vec3 × ℝ → ℝ} (hφd : ContDiff ℝ (↑⊤) φ) (hψd : ContDiff ℝ (↑⊤) ψ) {u : Foundation.Parabolic.ParabolicPoint → Foundation.Parabolic.Vec3} {Du : Foundation.Parabolic.ParabolicPoint → Fin 3 → Foundation.Parabolic.Vec3} {p : Foundation.Parabolic.ParabolicPoint → ℝ} {f : Foundation.Parabolic.ParabolicPoint → Foundation.Parabolic.Vec3} (i : Fin 3) (z : Foundation.Parabolic.ParabolicPoint) :
localizedDivergenceG φ u Du p f z i * ψ z + ∑ j : Fin 3, localizedDivergenceH φ u p j z i * spatialPartial ψ j z = u z i * (timePartial φ z * ψ z) + f z i * (φ z * ψ z) + ∑ j : Fin 3, u z i * u z j * spatialPartial (fun (w : Foundation.Parabolic.ParabolicPoint) => φ w * ψ w) j z + ∑ j : Fin 3, (-(Du z i j * (spatialPartial φ j z * ψ z)) + u z i * (spatialPartial φ j z * spatialPartial ψ j z)) + p z * spatialPartial (fun (w : Foundation.Parabolic.ParabolicPoint) => φ w * ψ w) i z

The divergence-form tested integrand of the local equation, regrouped onto the cutoff product φ * ψ (lem:local-equation). The only analytic input is the product rule spatialPartial_mul_full for the smooth cutoff factors.

The pressure-gradient tested integrand of the local equation, regrouped onto the cutoff product φ * ψ (lem:local-equation). No derivative of the test field is moved, so no smoothness hypothesis is needed.

Trading the divergence-form pressure slot for the gradient slot #

Paper label lem:local-equation. The cutoff-tested identity for the localized velocity is first obtained with the pressure sitting in the divergence-form slot, as p ∂ᵢφ together with the diagonal entry δᵢⱼ p φ. The estimates of Step 4 instead need the pressure as φ Dp in the heat slot, the convection in the form φ (u · ∇) u, and the second cutoff derivative Δφ u explicit. The theorem below performs that exchange once and for all at the level of the tested identity: the divergence-form right-hand side and the gradient-slot right-hand side agree for every space-time test function. The three inputs are the convection transfer (∑ⱼ ∫ uᵢuⱼ ∂ⱼ(φψ) = −∫ φψ (u · ∇)uᵢ), the diffusion transfer (one integration by parts in ∂ⱼφ ψ) and the weak pressure gradient tested against the product cutoff φψ.

theorem CKN.Core.Step3.gradientSlot_tested_transfer_of_sws {Ω : Set Foundation.Parabolic.Vec3} {I : Set ℝ} {q : ℝ} {u : Foundation.Parabolic.ParabolicPoint → Foundation.Parabolic.Vec3} {Du : Foundation.Parabolic.ParabolicPoint → Fin 3 → Foundation.Parabolic.Vec3} {p : Foundation.Parabolic.ParabolicPoint → ℝ} {f : Foundation.Parabolic.ParabolicPoint → Foundation.Parabolic.Vec3} (hsol : IsSuitableWeakSolutionIntegrable Ω I q u Du p f) {φ : Foundation.Parabolic.Vec3 × ℝ → ℝ} (hφ : φ ∈ spaceTimeTestFunction Ω I) {Ω' : Set Foundation.Parabolic.Vec3} {J : Set ℝ} (hbox : localBox Ω I Ω' J) (hφbox : tsupport φ ⊆ Ω' ×ˢ J) {Dp : Foundation.Parabolic.ParabolicPoint → Foundation.Parabolic.Vec3} (hDpInt : ∀ (i : Fin 3), MeasureTheory.Integrable (fun (z : Foundation.Parabolic.ParabolicPoint) => Dp z i) (MeasureTheory.volume.restrict (spaceTimeSet Ω' J))) (hDpweak : ∀ (i : Fin 3), ∀ χ ∈ spaceTimeTestFunction Set.univ Set.univ, tsupport χ ⊆ Ω' ×ˢ J → ∫ (z : Foundation.Parabolic.ParabolicPoint), p z * spatialPartial χ i z = -∫ (z : Foundation.Parabolic.ParabolicPoint), Dp z i * χ z) {ψ : Foundation.Parabolic.Vec3 × ℝ → ℝ} (hψ : ψ ∈ spaceTimeTestFunction Set.univ Set.univ) (i : Fin 3) :

The divergence-form and gradient-slot right-hand sides of the tested local equation agree, for every space-time test function ψ. Paper label lem:local-equation: this is the passage from the raw tested identity to the displayed equation eq:local-equation, in which the pressure enters through its weak gradient Dp.