Documentation

LeanPool.EllipticPDE.Regularity.DifferentiatedWkInfty

Moving a derivative onto the solution under Guo's coefficient hypothesis #

EllipticPdes.Regularity.principal_move, transport_move and zeroth_move move ∂_ℓ from the test function onto the solution in the three terms of the equation, each by one application of the Leibniz rule for a C¹ weight. Guo, Partial Differential Equations (Course Lecture Notes), Theorem VIII.3.2 (p. 65) asks only for W^{k,∞} coefficients, which have no classical derivative, and this file repeats the three with HasWeakDerivOn.mul_isWkInfty_left in place of HasWeakDerivOn.mul_contDiff_left.

The statements differ from their C¹ counterparts in one place: where those write partialD ℓ (fun y => A.a y i j) for the derivative of a coefficient, these write the chosen representative hA.D [ℓ] i j that IsWkInftyCoeff supplies. Everything else is unchanged, so a consumer that reaches for the commutator by name sees the same shape.

Main declarations #

Reading the order-one data off a W^{k,∞} bundle #

The coefficient entry is measurable, read off the order-zero member of the family.

theorem EllipticPdes.Regularity.IsWkInftyCoeff.hasWeakPartial_D {d : ℕ} {A : Sobolev.EllipticCoeff d} {k : ℕ} (hA : IsWkInftyCoeff A (k + 1)) (ℓ i j : Fin d) :
HasWeakPartial ℓ (fun (x : EuclideanSpace ℝ (Fin d)) => A.a x i j) (hA.D [ℓ] i j)

The order-one member of the family is a weak partial derivative of the coefficient entry.

The order-one member of the family is measurable.

theorem EllipticPdes.Regularity.IsWkInftyCoeff.ae_abs_D_singleton_le {d : ℕ} {A : Sobolev.EllipticCoeff d} {k : ℕ} (hA : IsWkInftyCoeff A (k + 1)) (ℓ i j : Fin d) :
∀ᵐ (x : EuclideanSpace ℝ (Fin d)), |hA.D [ℓ] i j x| ≤ hA.bound 1

The order-one member of the family is essentially bounded by bound 1.

theorem EllipticPdes.Regularity.IsWkInftyCoeff.hasWeakPartial_D_singleton {d : ℕ} {A : Sobolev.EllipticCoeff d} {k : ℕ} (hA : IsWkInftyCoeff A (k + 2)) (m ℓ i j : Fin d) :
HasWeakPartial m (hA.D [ℓ] i j) (hA.D [m, ℓ] i j)

The order-two member of the family is a weak partial derivative of the order-one member.

theorem EllipticPdes.Regularity.IsWkInftyCoeff.measurable_D_pair {d : ℕ} {A : Sobolev.EllipticCoeff d} {k : ℕ} (hA : IsWkInftyCoeff A (k + 2)) (m ℓ i j : Fin d) :
Measurable (hA.D [m, ℓ] i j)

The order-two member of the family is measurable.

theorem EllipticPdes.Regularity.IsWkInftyCoeff.ae_abs_D_pair_le {d : ℕ} {A : Sobolev.EllipticCoeff d} {k : ℕ} (hA : IsWkInftyCoeff A (k + 2)) (m ℓ i j : Fin d) :
∀ᵐ (x : EuclideanSpace ℝ (Fin d)), |hA.D [m, ℓ] i j x| ≤ hA.bound 2

The order-two member of the family is essentially bounded by bound 2.

The function is measurable, read off the order-zero member of the family.

The function is essentially bounded by bound 0.

theorem EllipticPdes.Regularity.IsWkInfty.hasWeakPartial_D {d : ℕ} {f : EuclideanSpace ℝ (Fin d) → ℝ} {k : ℕ} (hf : IsWkInfty f (k + 1)) (ℓ : Fin d) :
HasWeakPartial ℓ f (hf.D [ℓ])

The order-one member of the family is a weak partial derivative of the function.

theorem EllipticPdes.Regularity.IsWkInfty.measurable_D_singleton {d : ℕ} {f : EuclideanSpace ℝ (Fin d) → ℝ} {k : ℕ} (hf : IsWkInfty f (k + 1)) (ℓ : Fin d) :
Measurable (hf.D [ℓ])

The order-one member of the family is measurable.

theorem EllipticPdes.Regularity.IsWkInfty.ae_abs_D_singleton_le {d : ℕ} {f : EuclideanSpace ℝ (Fin d) → ℝ} {k : ℕ} (hf : IsWkInfty f (k + 1)) (ℓ : Fin d) :
∀ᵐ (x : EuclideanSpace ℝ (Fin d)), |hf.D [ℓ] x| ≤ hf.bound 1

The order-one member of the family is essentially bounded by bound 1.

Three terms #

theorem EllipticPdes.Regularity.principal_move_wkInfty {d : ℕ} {V : Set (EuclideanSpace ℝ (Fin d))} {A : Sobolev.EllipticCoeff d} {k : ℕ} (hA : IsWkInftyCoeff A (k + 1)) (ℓ : Fin d) (Du : Fin d → ↥(MeasureTheory.Lp ℝ 2 (MeasureTheory.volume.restrict V))) (D2 : Fin d → Fin d → ↥(MeasureTheory.Lp ℝ 2 (MeasureTheory.volume.restrict V))) (hD2 : ∀ (i : Fin d), HasWeakDerivOn V ℓ (Du i) (D2 ℓ i)) (aDu : Fin d → Fin d → ↥(MeasureTheory.Lp ℝ 2 (MeasureTheory.volume.restrict V))) (haDu : ∀ (i j : Fin d), ↑↑(aDu i j) =ᵐ[MeasureTheory.volume.restrict V] fun (x : EuclideanSpace ℝ (Fin d)) => A.a x i j * ↑↑(Du i) x) (comm : Fin d → Fin d → ↥(MeasureTheory.Lp ℝ 2 (MeasureTheory.volume.restrict V))) (hcomm : ∀ (i j : Fin d), ↑↑(comm i j) =ᵐ[MeasureTheory.volume.restrict V] fun (x : EuclideanSpace ℝ (Fin d)) => hA.D [ℓ] i j x * ↑↑(Du i) x + A.a x i j * ↑↑(D2 ℓ i) x) {φ : EuclideanSpace ℝ (Fin d) → ℝ} (hφc : ContDiff ℝ (↑⊤) φ) (hφcs : HasCompactSupport φ) (hφV : tsupport φ ⊆ V) :
∑ i : Fin d, ∑ j : Fin d, ∫ (x : EuclideanSpace ℝ (Fin d)) in V, ↑↑(aDu i j) x * Sobolev.partialD ℓ (Sobolev.partialD j φ) x = -∑ i : Fin d, ∑ j : Fin d, ∫ (x : EuclideanSpace ℝ (Fin d)) in V, ↑↑(comm i j) x * Sobolev.partialD j φ x

Principal term for a W^{1,∞} coefficient. For every direction pair the weighted first derivative a_{ij}·∂ᵢu has weak ℓ-derivative (∂_ℓ a_{ij})·∂ᵢu + a_{ij}·∂_ℓ∂ᵢu, with the coefficient derivative read off the family rather than taken classically. Testing against ∂ⱼφ and summing gives ∑ ∫_V a_{ij}(∂ᵢu) ∂_ℓ∂ⱼφ = -∑ ∫_V [(∂_ℓ a_{ij})(∂ᵢu) + a_{ij}(∂ₗ∂ᵢu)] ∂ⱼφ.

theorem EllipticPdes.Regularity.transport_move_wkInfty {d : ℕ} {V : Set (EuclideanSpace ℝ (Fin d))} (ℓ : Fin d) {bi : EuclideanSpace ℝ (Fin d) → ℝ} {k : ℕ} (hbi : IsWkInfty bi (k + 1)) (Du_i D2_ℓi : ↥(MeasureTheory.Lp ℝ 2 (MeasureTheory.volume.restrict V))) (hD2 : HasWeakDerivOn V ℓ Du_i D2_ℓi) (bDu : ↥(MeasureTheory.Lp ℝ 2 (MeasureTheory.volume.restrict V))) (hbDu : ↑↑bDu =ᵐ[MeasureTheory.volume.restrict V] fun (x : EuclideanSpace ℝ (Fin d)) => bi x * ↑↑Du_i x) (comm : ↥(MeasureTheory.Lp ℝ 2 (MeasureTheory.volume.restrict V))) (hcomm : ↑↑comm =ᵐ[MeasureTheory.volume.restrict V] fun (x : EuclideanSpace ℝ (Fin d)) => hbi.D [ℓ] x * ↑↑Du_i x + bi x * ↑↑D2_ℓi x) {φ : EuclideanSpace ℝ (Fin d) → ℝ} (hφc : ContDiff ℝ (↑⊤) φ) (hφcs : HasCompactSupport φ) (hφV : tsupport φ ⊆ V) :
∫ (x : EuclideanSpace ℝ (Fin d)) in V, ↑↑bDu x * Sobolev.partialD ℓ φ x = -∫ (x : EuclideanSpace ℝ (Fin d)) in V, ↑↑comm x * φ x

Transport term for a W^{1,∞} coefficient. ∫_V b_i(∂ᵢu) ∂_ℓφ = -∫_V [(∂_ℓ b_i)(∂ᵢu) + b_i(∂ₗ∂ᵢu)] φ, with the coefficient derivative read off the family.

theorem EllipticPdes.Regularity.zeroth_move_wkInfty {d : ℕ} {V : Set (EuclideanSpace ℝ (Fin d))} (ℓ : Fin d) {c : EuclideanSpace ℝ (Fin d) → ℝ} {k : ℕ} (hc : IsWkInfty c (k + 1)) (u_V Du_ℓ : ↥(MeasureTheory.Lp ℝ 2 (MeasureTheory.volume.restrict V))) (hDu : HasWeakDerivOn V ℓ u_V Du_ℓ) (cu : ↥(MeasureTheory.Lp ℝ 2 (MeasureTheory.volume.restrict V))) (hcu : ↑↑cu =ᵐ[MeasureTheory.volume.restrict V] fun (x : EuclideanSpace ℝ (Fin d)) => c x * ↑↑u_V x) (comm : ↥(MeasureTheory.Lp ℝ 2 (MeasureTheory.volume.restrict V))) (hcomm : ↑↑comm =ᵐ[MeasureTheory.volume.restrict V] fun (x : EuclideanSpace ℝ (Fin d)) => hc.D [ℓ] x * ↑↑u_V x + c x * ↑↑Du_ℓ x) {φ : EuclideanSpace ℝ (Fin d) → ℝ} (hφc : ContDiff ℝ (↑⊤) φ) (hφcs : HasCompactSupport φ) (hφV : tsupport φ ⊆ V) :
∫ (x : EuclideanSpace ℝ (Fin d)) in V, ↑↑cu x * Sobolev.partialD ℓ φ x = -∫ (x : EuclideanSpace ℝ (Fin d)) in V, ↑↑comm x * φ x

Zeroth-order term for a W^{1,∞} coefficient. ∫_V c·u·∂_ℓφ = -∫_V [(∂_ℓ c)·u + c·(∂ₗu)] φ, with the coefficient derivative read off the family.

Principal commutator #

theorem EllipticPdes.Regularity.commutator_move_wkInfty {d : ℕ} {V : Set (EuclideanSpace ℝ (Fin d))} {A : Sobolev.EllipticCoeff d} {k : ℕ} (hA : IsWkInftyCoeff A (k + 2)) (ℓ i j : Fin d) (Du_i D2_ji : Sobolev.L2D V) (hD2 : HasWeakDerivOn V j Du_i D2_ji) {φ : EuclideanSpace ℝ (Fin d) → ℝ} (hφc : ContDiff ℝ (↑⊤) φ) (hφcs : HasCompactSupport φ) (hφV : tsupport φ ⊆ V) :
∫ (x : EuclideanSpace ℝ (Fin d)) in V, hA.D [ℓ] i j x * ↑↑Du_i x * Sobolev.partialD j φ x = -∫ (x : EuclideanSpace ℝ (Fin d)) in V, (hA.D [j, ℓ] i j x * ↑↑Du_i x + hA.D [ℓ] i j x * ↑↑D2_ji x) * φ x

Moving ∂ⱼ off the principal commutator for a W^{2,∞} coefficient. The coefficient derivative ∂_ℓ a_{ij} is itself a W^{1,∞} weight, so the product (∂_ℓ a_{ij})·∂ᵢu has a weak j-derivative and testing against φ moves ∂ⱼ onto the product: ∫_V (∂_ℓ a_{ij})(∂ᵢu) ∂ⱼφ = -∫_V [(∂ⱼ∂_ℓ a_{ij})(∂ᵢu) + (∂_ℓ a_{ij})(∂ⱼ∂ᵢu)] φ.

The second order of the coefficient hypothesis is used only here: it supplies the mixed member hA.D [j, ℓ] of the family. The C² version commutator_move spends most of its length turning the Hessian bound into a bound on the mixed partial through two operator-norm steps, and the W^{2,∞} bundle has that bound outright.

Differentiated identity #

theorem EllipticPdes.Regularity.differentiated_weakForm_div_wkInfty {d : ℕ} {V : Set (EuclideanSpace ℝ (Fin d))} (Op : Sobolev.FullEllipticOp d) {k m : ℕ} (hA : IsWkInftyCoeff Op.toEllipticCoeff (k + 1)) (hbc : IsWkInftyLower Op (m + 1)) (ℓ : Fin d) (u_V : Sobolev.L2D V) (Du : Fin d → Sobolev.L2D V) (D2 : Fin d → Fin d → Sobolev.L2D V) (f_V Df : Sobolev.L2D V) (hDu_D2 : ∀ (i : Fin d), HasWeakDerivOn V ℓ (Du i) (D2 ℓ i)) (hu_Duℓ : HasWeakDerivOn V ℓ u_V (Du ℓ)) (hf_Df : HasWeakDerivOn V ℓ f_V Df) (hLoc : ∀ (v : EuclideanSpace ℝ (Fin d) → ℝ), ContDiff ℝ (↑⊤) v → HasCompactSupport v → tsupport v ⊆ V → ((∑ i : Fin d, ∑ j : Fin d, ∫ (x : EuclideanSpace ℝ (Fin d)) in V, Op.a x i j * ↑↑(Du i) x * Sobolev.partialD j v x) + ∑ i : Fin d, ∫ (x : EuclideanSpace ℝ (Fin d)) in V, Op.b x i * ↑↑(Du i) x * v x) + ∫ (x : EuclideanSpace ℝ (Fin d)) in V, Op.c x * ↑↑u_V x * v x = ∫ (x : EuclideanSpace ℝ (Fin d)) in V, ↑↑f_V x * v x) {φ : EuclideanSpace ℝ (Fin d) → ℝ} (hφc : ContDiff ℝ (↑⊤) φ) (hφcs : HasCompactSupport φ) (hφV : tsupport φ ⊆ V) :
(∑ i : Fin d, ∑ j : Fin d, ∫ (x : EuclideanSpace ℝ (Fin d)) in V, Op.a x i j * ↑↑(D2 ℓ i) x * Sobolev.partialD j φ x) + ∑ i : Fin d, ∑ j : Fin d, ∫ (x : EuclideanSpace ℝ (Fin d)) in V, hA.D [ℓ] i j x * ↑↑(Du i) x * Sobolev.partialD j φ x = ((∫ (x : EuclideanSpace ℝ (Fin d)) in V, ↑↑Df x * φ x) - ∑ i : Fin d, ∫ (x : EuclideanSpace ℝ (Fin d)) in V, ((hbc.bReg i).D [ℓ] x * ↑↑(Du i) x + Op.b x i * ↑↑(D2 ℓ i) x) * φ x) - ∫ (x : EuclideanSpace ℝ (Fin d)) in V, (hbc.cReg.D [ℓ] x * ↑↑u_V x + Op.c x * ↑↑(Du ℓ) x) * φ x

Differentiated weak formulation (divergence-datum form) for W^{1,∞} coefficients. Evans, Partial Differential Equations (2nd ed.), §6.3.1, Theorem 2, with every classical coefficient derivative replaced by the chosen representative the W^{k,∞} bundles supply. Given the localised weak identity hLoc for u on V together with the first and second weak derivatives, for a fixed direction ℓ and every test function φ with tsupport φ ⊆ V, ∑ ∫_V a_{ij}(∂ₗ∂ᵢu) ∂ⱼφ + ∑ ∫_V (∂_ℓ a_{ij})(∂ᵢu) ∂ⱼφ = ∫_V (∂_ℓf) φ - ∑ ∫_V [(∂_ℓ b_i)(∂ᵢu)+b_i(∂ₗ∂ᵢu)] φ - ∫_V [(∂_ℓ c)u + c(∂_ℓu)] φ.

Where differentiated_weakForm_div asks for a ∈ C² and b, c ∈ C¹, this asks for one weak derivative of each, which is Guo's hypothesis at the first order.

theorem EllipticPdes.Regularity.differentiated_weakForm_wkInfty {d : ℕ} {V : Set (EuclideanSpace ℝ (Fin d))} (Op : Sobolev.FullEllipticOp d) {k m : ℕ} (hA : IsWkInftyCoeff Op.toEllipticCoeff (k + 2)) (hbc : IsWkInftyLower Op (m + 1)) (ℓ : Fin d) (u_V : Sobolev.L2D V) (Du : Fin d → Sobolev.L2D V) (D2 : Fin d → Fin d → Sobolev.L2D V) (f_V Df : Sobolev.L2D V) (hDu_D2 : ∀ (i : Fin d), HasWeakDerivOn V ℓ (Du i) (D2 ℓ i)) (hD2_j : ∀ (i j : Fin d), HasWeakDerivOn V j (Du i) (D2 j i)) (hu_Duℓ : HasWeakDerivOn V ℓ u_V (Du ℓ)) (hf_Df : HasWeakDerivOn V ℓ f_V Df) (hLoc : ∀ (v : EuclideanSpace ℝ (Fin d) → ℝ), ContDiff ℝ (↑⊤) v → HasCompactSupport v → tsupport v ⊆ V → ((∑ i : Fin d, ∑ j : Fin d, ∫ (x : EuclideanSpace ℝ (Fin d)) in V, Op.a x i j * ↑↑(Du i) x * Sobolev.partialD j v x) + ∑ i : Fin d, ∫ (x : EuclideanSpace ℝ (Fin d)) in V, Op.b x i * ↑↑(Du i) x * v x) + ∫ (x : EuclideanSpace ℝ (Fin d)) in V, Op.c x * ↑↑u_V x * v x = ∫ (x : EuclideanSpace ℝ (Fin d)) in V, ↑↑f_V x * v x) {φ : EuclideanSpace ℝ (Fin d) → ℝ} (hφc : ContDiff ℝ (↑⊤) φ) (hφcs : HasCompactSupport φ) (hφV : tsupport φ ⊆ V) :
∑ i : Fin d, ∑ j : Fin d, ∫ (x : EuclideanSpace ℝ (Fin d)) in V, Op.a x i j * ↑↑(D2 ℓ i) x * Sobolev.partialD j φ x = (((∫ (x : EuclideanSpace ℝ (Fin d)) in V, ↑↑Df x * φ x) - ∑ i : Fin d, ∫ (x : EuclideanSpace ℝ (Fin d)) in V, ((hbc.bReg i).D [ℓ] x * ↑↑(Du i) x + Op.b x i * ↑↑(D2 ℓ i) x) * φ x) - ∫ (x : EuclideanSpace ℝ (Fin d)) in V, (hbc.cReg.D [ℓ] x * ↑↑u_V x + Op.c x * ↑↑(Du ℓ) x) * φ x) + ∑ i : Fin d, ∑ j : Fin d, ∫ (x : EuclideanSpace ℝ (Fin d)) in V, (hA.D [j, ℓ] i j x * ↑↑(Du i) x + hA.D [ℓ] i j x * ↑↑(D2 j i) x) * φ x

Differentiated weak formulation (Evans strong-datum form), for W^{2,∞} principal and W^{1,∞} lower-order coefficients. Moving ∂ⱼ off the principal commutator with commutator_move_wkInfty merges the second block of the left-hand side into the datum, leaving ∑ ∫_V a_{ij}(∂ₗ∂ᵢu) ∂ⱼφ = ∫_V f_ℓ · φ with f_ℓ = ∂_ℓf - ∑_i [(∂_ℓ b_i)(∂ᵢu)+b_i(∂ₗ∂ᵢu)] - [(∂_ℓ c)u + c(∂_ℓu)] + ∑_{i,j}[(∂ⱼ∂_ℓ a_{ij})(∂ᵢu)+(∂_ℓ a_{ij})(∂ⱼ∂ᵢu)] delivered as an explicit sum of integrals. Every derivative of a coefficient is read off the bundle, so nothing here asks a coefficient to be differentiable.