Documentation

TauCeti.Analysis.PDE.MaximumPrinciple.GlobalBoundedness

Global boundedness of weak subsolutions (Stampacchia) #

Let Ω ⊆ ℝⁿ be open with finite measure, let a be measurable and uniformly elliptic on Ω with constants 0 < λ ≤ Λ, and let c be bounded, measurable and nonnegative. This file proves the weak maximum principle with a right-hand side: if u ∈ H¹(Ω) is a weak subsolution of

-∂ᵢ(aⁱʲ ∂ⱼu) + c u ≤ f in Ω, with f ∈ Lʳ(Ω), r > 1 and r > n/2,

meaning a(u, v) ≤ ∫_Ω f v for every nonnegative v ∈ H¹₀(Ω), and u ≤ k on ∂Ω for a level k ≥ 0, in the sense that (u - k)⁺ ∈ H¹₀(Ω), then

u ≤ k + C ‖f‖_{Lʳ(Ω)} / λ · |Ω| ^ (2/n - 1/r) almost everywhere in Ω,

with C depending only on n, r and the normalization of the Haar measure. In particular the weak solution u ∈ H¹₀(Ω) of the Dirichlet problem -∂ᵢ(aⁱʲ ∂ⱼu) + c u = f with f ∈ Lʳ(Ω), meaning a(u, v) = ∫_Ω f v for every v ∈ H¹₀(Ω), is bounded, with |u| ≤ C ‖f‖_{Lʳ} / λ · |Ω| ^ (2/n - 1/r). No regularity of ∂Ω and no regularity of the coefficients beyond measurability is used, and the power of |Ω| makes the estimate invariant under dilations.

This is the bound that lets De Giorgi's interior estimates absorb a right-hand side: on a ball B ⊆ Ω, a solution of -∂ᵢ(aⁱʲ ∂ⱼu) = f differs from a solution of the homogeneous equation by the solution of the Dirichlet problem on B with data f, and the bound above controls that difference by R ^ (2 - n/r) ‖f‖_{Lʳ}.

Stampacchia's method #

For a level l ≥ k, the truncation w = (u - l)⁺ lies in H¹₀(Ω) and is a nonnegative test function, and its gradient is ∇u on {u > l} and zero elsewhere. Testing the subsolution inequality against it, ellipticity and c ≥ 0 give λ ‖∇w‖² ≤ ∫ f w. A Sobolev inequality ‖w‖_{Lˢ} ≤ S ‖∇w‖_{L²} on H¹₀(Ω) and Hölder's inequality on the superlevel set {u > l}, off which w vanishes, turn this into

‖(u - l)⁺‖_{Lˢ} ≤ S² ‖f‖_{Lʳ} / λ · |{u > l}| ^ (1 - 1/r - 1/s).

Chebyshev's inequality on {u > h}, where (u - l)⁺ ≥ h - l, then gives, for φ l = |{u > l}| and M = S² ‖f‖_{Lʳ} / λ,

φ h ≤ (M / (h - l)) ^ s φ l ^ β for k ≤ l < h, where β = s (1 - 1/r - 1/s).

When s (1 - 1/r) > 2, the exponent β exceeds 1, and Stampacchia's lemma (TauCeti.apply_add_eq_zero_of_le_div_sub_rpow_mul_rpow) shows that φ vanishes at a finite level above k. In every dimension the Sobolev inequality on sets of finite measure (TauCeti.W1p.eLpNorm_value_le_mul_measure_rpow_mul_enorm_gradient_of_lt) provides such an exponent s whenever r > 1 and r > n/2, with S a multiple of |Ω| ^ (1/n - 1/2 + 1/s). Only the energy estimate for ‖(u - l)⁺‖_{Lˢ} depends on the right-hand side; the passage from it to the bound on u is shared by both kinds of data below.

A right-hand side in divergence form #

The same holds for weak subsolutions of -∂ᵢ(aⁱʲ ∂ⱼu) + c u ≤ ∂ᵢFⁱ with F ∈ L^q(Ω; ℝⁿ), q > n and q > 2, meaning a(u, v) ≤ -∫_Ω F · ∇v for every nonnegative v ∈ H¹₀(Ω):

u ≤ k + C ‖F‖_{L^q(Ω)} / λ · |Ω| ^ (1/n - 1/q) almost everywhere in Ω.

Testing against w = (u - l)⁺ now gives λ ‖∇w‖² ≤ ∫ ‖F‖ ‖∇w‖, and since ∇w vanishes off {u > l}, Hölder's inequality bounds the right side by ‖F‖_{L^q} ‖∇w‖_{L²} |{u > l}| ^ (1/2 - 1/q). After cancelling one factor of ‖∇w‖_{L²}, the Sobolev inequality gives ‖(u - l)⁺‖_{Lˢ} ≤ S ‖F‖_{L^q} / λ · |{u > l}| ^ (1/2 - 1/q), and Stampacchia's lemma applies once s (1/2 - 1/q) > 1. On a ball B of radius ρ this bounds the solution of the Dirichlet problem on B with datum ∂ᵢFⁱ by a multiple of ρ ^ (1 - n/q) ‖F‖_{L^q}. That is the divergence-form counterpart of the bound above, needed when De Giorgi's interior estimates absorb a divergence-form right-hand side.

Main declarations #

References #

theorem TauCeti.PDE.UniformlyEllipticOn.ae_value_le_add_of_forall_eLpNorm_le {ι : Type u_1} [Fintype ι] [DecidableEq ι] {mu : MeasureTheory.Measure (EuclideanSpace ℝ ι)} [mu.IsAddHaarMeasure] {Omega : TopologicalSpace.Opens (EuclideanSpace ℝ ι)} {a : EuclideanSpace ℝ ι → Matrix ι ι ℝ} {c : EuclideanSpace ℝ ι → ℝ} {lam Lam gamma : ℝ} (h : UniformlyEllipticOn (↑Omega) a lam Lam) (ha : MeasureTheory.AEStronglyMeasurable a (mu.restrict ↑Omega)) (hc : MeasureTheory.AEStronglyMeasurable c (mu.restrict ↑Omega)) (hc_bound : ∀ x ∈ ↑Omega, ‖c x‖ ≤ gamma) (hc_nonneg : ∀ x ∈ ↑Omega, 0 ≤ c x) (hOmega : mu ↑Omega ≠ ⊤) {r s S : NNReal} (hr : 0 < r) (hrs : 2 < ↑s * (1 - (↑r)⁻¹)) (hS : ∀ v ∈ w1p0Submodule mu Omega 2, MeasureTheory.eLpNorm (↑↑(W1p.value v)) (↑s) (mu.restrict ↑Omega) ≤ ↑S * ‖W1p.gradient v‖ₑ) {f : EuclideanSpace ℝ ι → ℝ} (hf : MeasureTheory.MemLp f (↑r) (mu.restrict ↑Omega)) {u : ↥(W1p mu Omega 2)} (hu : ∀ (v : ↥(W1p0 mu Omega 2)), (∀ᵐ (x : EuclideanSpace ℝ ι) ∂mu.restrict ↑Omega, 0 ≤ ↑↑(W1p.value ↑v) x) → energyFormH1 a 0 c u ↑v ≤ ∫ (x : EuclideanSpace ℝ ι) in ↑Omega, f x * ↑↑(W1p.value ↑v) x ∂mu) {k : ℝ} (hk : 0 ≤ k) (hboundary : W1p.posPartAbove ae_value_le_add_of_forall_eLpNorm_le._proof_1 hk u ∈ w1p0Submodule mu Omega 2) :
∀ᵐ (x : EuclideanSpace ℝ ι) ∂mu.restrict ↑Omega, ↑↑(W1p.value u) x ≤ k + 2 ^ ((↑s * (1 - (↑r)⁻¹) - 1) / (↑s * (1 - (↑r)⁻¹) - 2)) * ↑S ^ 2 * (MeasureTheory.eLpNorm f (↑r) (mu.restrict ↑Omega)).toReal / lam * mu.real ↑Omega ^ (1 - (↑r)⁻¹ - 2 / ↑s)

Stampacchia's global bound. Let Ω have finite measure, let a be measurable and uniformly elliptic on Ω with constants 0 < λ ≤ Λ, and let c be bounded, measurable and nonnegative. Suppose that W^{1,2}_0(Ω) satisfies a Sobolev inequality ‖v‖_{Lˢ} ≤ S ‖∇v‖_{L²}, and let f ∈ Lʳ(Ω) with s (1 - 1/r) > 2. If u ∈ H¹(Ω) is a weak subsolution of

-∂ᵢ(aⁱʲ ∂ⱼu) + c u ≤ f in Ω,

that is a(u, v) ≤ ∫_Ω f v for every nonnegative v ∈ H¹₀(Ω), and u ≤ k on ∂Ω for a level k ≥ 0, in the sense that (u - k)⁺ ∈ H¹₀(Ω), then almost everywhere in Ω

u ≤ k + 2 ^ (β / (β - 1)) S² ‖f‖_{Lʳ} / λ · |Ω| ^ (1 - 1/r - 2/s), where β = s (1 - 1/r) - 1.

No regularity of ∂Ω or of the coefficients beyond measurability is needed.

theorem TauCeti.PDE.exists_ae_value_le_add_mul_div_mul_rpow {ι : Type u_1} [Fintype ι] [DecidableEq ι] {mu : MeasureTheory.Measure (EuclideanSpace ℝ ι)} [mu.IsAddHaarMeasure] [Nonempty ι] {r : NNReal} (hr : 1 < r) (hrn : ↑(Fintype.card ι) < 2 * ↑r) :
∃ (C : ℝ), 0 < C ∧ ∀ {Omega : TopologicalSpace.Opens (EuclideanSpace ℝ ι)} {a : EuclideanSpace ℝ ι → Matrix ι ι ℝ} {c : EuclideanSpace ℝ ι → ℝ} {lam Lam gamma : ℝ} {f : EuclideanSpace ℝ ι → ℝ} {u : ↥(W1p mu Omega 2)} {k : ℝ} (hk : 0 ≤ k), mu ↑Omega ≠ ⊤ → UniformlyEllipticOn (↑Omega) a lam Lam → MeasureTheory.AEStronglyMeasurable a (mu.restrict ↑Omega) → MeasureTheory.AEStronglyMeasurable c (mu.restrict ↑Omega) → (∀ x ∈ ↑Omega, ‖c x‖ ≤ gamma) → (∀ x ∈ ↑Omega, 0 ≤ c x) → MeasureTheory.MemLp f (↑r) (mu.restrict ↑Omega) → (∀ (v : ↥(W1p0 mu Omega 2)), (∀ᵐ (x : EuclideanSpace ℝ ι) ∂mu.restrict ↑Omega, 0 ≤ ↑↑(W1p.value ↑v) x) → energyFormH1 a 0 c u ↑v ≤ ∫ (x : EuclideanSpace ℝ ι) in ↑Omega, f x * ↑↑(W1p.value ↑v) x ∂mu) → W1p.posPartAbove UniformlyEllipticOn.ae_value_le_add_of_forall_eLpNorm_le._proof_1 hk u ∈ w1p0Submodule mu Omega 2 → ∀ᵐ (x : EuclideanSpace ℝ ι) ∂mu.restrict ↑Omega, ↑↑(W1p.value u) x ≤ k + C * (MeasureTheory.eLpNorm f (↑r) (mu.restrict ↑Omega)).toReal / lam * mu.real ↑Omega ^ (2 / ↑(Fintype.card ι) - (↑r)⁻¹)

Stampacchia's global bound, in every dimension. Let n ≥ 1 be the dimension and let r > 1 with r > n/2. There is C > 0, depending only on n, r and the normalization of the additive Haar measure mu, with the following property. Let Ω have finite measure, let a be measurable and uniformly elliptic on Ω with constants 0 < λ ≤ Λ, let c be bounded, measurable and nonnegative, and let f ∈ Lʳ(Ω). If u ∈ H¹(Ω) is a weak subsolution of

-∂ᵢ(aⁱʲ ∂ⱼu) + c u ≤ f in Ω

with u ≤ k on ∂Ω for a level k ≥ 0, in the sense that (u - k)⁺ ∈ H¹₀(Ω), then almost everywhere in Ω

u ≤ k + C ‖f‖_{Lʳ(Ω)} / λ · |Ω| ^ (2/n - 1/r).

The bound is invariant under dilations of Ω. The Sobolev inequality behind it is the one on sets of finite measure, TauCeti.W1p.eLpNorm_value_le_mul_measure_rpow_mul_enorm_gradient_of_lt, which holds in every dimension.

theorem TauCeti.PDE.exists_ae_abs_value_le_mul_div_mul_rpow {ι : Type u_1} [Fintype ι] [DecidableEq ι] {mu : MeasureTheory.Measure (EuclideanSpace ℝ ι)} [mu.IsAddHaarMeasure] [Nonempty ι] {r : NNReal} (hr : 1 < r) (hrn : ↑(Fintype.card ι) < 2 * ↑r) :
∃ (C : ℝ), 0 < C ∧ ∀ {Omega : TopologicalSpace.Opens (EuclideanSpace ℝ ι)} {a : EuclideanSpace ℝ ι → Matrix ι ι ℝ} {c : EuclideanSpace ℝ ι → ℝ} {lam Lam gamma : ℝ} {f : EuclideanSpace ℝ ι → ℝ} {u : ↥(W1p0 mu Omega 2)}, mu ↑Omega ≠ ⊤ → UniformlyEllipticOn (↑Omega) a lam Lam → MeasureTheory.AEStronglyMeasurable a (mu.restrict ↑Omega) → MeasureTheory.AEStronglyMeasurable c (mu.restrict ↑Omega) → (∀ x ∈ ↑Omega, ‖c x‖ ≤ gamma) → (∀ x ∈ ↑Omega, 0 ≤ c x) → MeasureTheory.MemLp f (↑r) (mu.restrict ↑Omega) → (∀ (v : ↥(W1p0 mu Omega 2)), energyFormH1 a 0 c ↑u ↑v = ∫ (x : EuclideanSpace ℝ ι) in ↑Omega, f x * ↑↑(W1p.value ↑v) x ∂mu) → ∀ᵐ (x : EuclideanSpace ℝ ι) ∂mu.restrict ↑Omega, |↑↑(W1p.value ↑u) x| ≤ C * (MeasureTheory.eLpNorm f (↑r) (mu.restrict ↑Omega)).toReal / lam * mu.real ↑Omega ^ (2 / ↑(Fintype.card ι) - (↑r)⁻¹)

Global boundedness of weak Dirichlet solutions, in every dimension. Let n ≥ 1 be the dimension and let r > 1 with r > n/2. There is C > 0, depending only on n, r and the normalization of mu, such that on every Ω of finite measure, for measurable, uniformly elliptic a with constants 0 < λ ≤ Λ and bounded, measurable, nonnegative c, the weak solution u ∈ H¹₀(Ω) of the Dirichlet problem -∂ᵢ(aⁱʲ ∂ⱼu) + c u = f with f ∈ Lʳ(Ω), that is a(u, v) = ∫_Ω f v for every v ∈ H¹₀(Ω), satisfies, almost everywhere in Ω,

|u| ≤ C ‖f‖_{Lʳ(Ω)} / λ · |Ω| ^ (2/n - 1/r).

No integrability of f beyond Lʳ is assumed; for f ∈ L²(Ω) the hypothesis on u is TauCeti.PDE.IsWeakSolutionDirichlet a 0 c f u, by TauCeti.PDE.isWeakSolutionDirichlet_iff. This is the two-sided form of TauCeti.PDE.exists_ae_value_le_add_mul_div_mul_rpow at the level k = 0, applied to u and to -u.

A right-hand side in divergence form #

theorem TauCeti.PDE.UniformlyEllipticOn.ae_value_le_add_of_forall_eLpNorm_le_divergence {ι : Type u_1} [Fintype ι] [DecidableEq ι] {mu : MeasureTheory.Measure (EuclideanSpace ℝ ι)} [mu.IsAddHaarMeasure] {Omega : TopologicalSpace.Opens (EuclideanSpace ℝ ι)} {a : EuclideanSpace ℝ ι → Matrix ι ι ℝ} {c : EuclideanSpace ℝ ι → ℝ} {lam Lam gamma : ℝ} (h : UniformlyEllipticOn (↑Omega) a lam Lam) (ha : MeasureTheory.AEStronglyMeasurable a (mu.restrict ↑Omega)) (hc : MeasureTheory.AEStronglyMeasurable c (mu.restrict ↑Omega)) (hc_bound : ∀ x ∈ ↑Omega, ‖c x‖ ≤ gamma) (hc_nonneg : ∀ x ∈ ↑Omega, 0 ≤ c x) (hOmega : mu ↑Omega ≠ ⊤) {q s S : NNReal} (hq : 2 ≤ ↑q) (hqs : 1 < ↑s * (2⁻¹ - (↑q)⁻¹)) (hS : ∀ v ∈ w1p0Submodule mu Omega 2, MeasureTheory.eLpNorm (↑↑(W1p.value v)) (↑s) (mu.restrict ↑Omega) ≤ ↑S * ‖W1p.gradient v‖ₑ) {F : EuclideanSpace ℝ ι → EuclideanSpace ℝ ι} (hF : MeasureTheory.MemLp F (↑q) (mu.restrict ↑Omega)) {u : ↥(W1p mu Omega 2)} (hu : ∀ (v : ↥(W1p0 mu Omega 2)), (∀ᵐ (x : EuclideanSpace ℝ ι) ∂mu.restrict ↑Omega, 0 ≤ ↑↑(W1p.value ↑v) x) → energyFormH1 a 0 c u ↑v ≤ -∫ (x : EuclideanSpace ℝ ι) in ↑Omega, inner ℝ (F x) (↑↑(W1p.gradient ↑v) x) ∂mu) {k : ℝ} (hk : 0 ≤ k) (hboundary : W1p.posPartAbove ae_value_le_add_of_forall_eLpNorm_le._proof_1 hk u ∈ w1p0Submodule mu Omega 2) :
∀ᵐ (x : EuclideanSpace ℝ ι) ∂mu.restrict ↑Omega, ↑↑(W1p.value u) x ≤ k + 2 ^ (↑s * (2⁻¹ - (↑q)⁻¹) / (↑s * (2⁻¹ - (↑q)⁻¹) - 1)) * ↑S * (MeasureTheory.eLpNorm F (↑q) (mu.restrict ↑Omega)).toReal / lam * mu.real ↑Omega ^ (2⁻¹ - (↑q)⁻¹ - (↑s)⁻¹)

Stampacchia's global bound for a right-hand side in divergence form. Let Ω have finite measure, let a be measurable and uniformly elliptic on Ω with constants 0 < λ ≤ Λ, and let c be bounded, measurable and nonnegative. Suppose that W^{1,2}_0(Ω) satisfies a Sobolev inequality ‖v‖_{Lˢ} ≤ S ‖∇v‖_{L²}, and let F ∈ L^q(Ω; ℝⁿ) with q ≥ 2 and s (1/2 - 1/q) > 1. If u ∈ H¹(Ω) is a weak subsolution of

-∂ᵢ(aⁱʲ ∂ⱼu) + c u ≤ ∂ᵢFⁱ in Ω,

that is a(u, v) ≤ -∫_Ω F · ∇v for every nonnegative v ∈ H¹₀(Ω), and u ≤ k on ∂Ω for a level k ≥ 0, in the sense that (u - k)⁺ ∈ H¹₀(Ω), then almost everywhere in Ω

u ≤ k + 2 ^ (β / (β - 1)) S ‖F‖_{L^q} / λ · |Ω| ^ (1/2 - 1/q - 1/s), where β = s (1/2 - 1/q).

No regularity of ∂Ω or of the coefficients beyond measurability is needed.

theorem TauCeti.PDE.exists_ae_value_le_add_mul_div_mul_rpow_divergence {ι : Type u_1} [Fintype ι] [DecidableEq ι] {mu : MeasureTheory.Measure (EuclideanSpace ℝ ι)} [mu.IsAddHaarMeasure] [Nonempty ι] {q : NNReal} (hq : 2 < ↑q) (hqn : ↑(Fintype.card ι) < ↑q) :
∃ (C : ℝ), 0 < C ∧ ∀ {Omega : TopologicalSpace.Opens (EuclideanSpace ℝ ι)} {a : EuclideanSpace ℝ ι → Matrix ι ι ℝ} {c : EuclideanSpace ℝ ι → ℝ} {lam Lam gamma : ℝ} {F : EuclideanSpace ℝ ι → EuclideanSpace ℝ ι} {u : ↥(W1p mu Omega 2)} {k : ℝ} (hk : 0 ≤ k), mu ↑Omega ≠ ⊤ → UniformlyEllipticOn (↑Omega) a lam Lam → MeasureTheory.AEStronglyMeasurable a (mu.restrict ↑Omega) → MeasureTheory.AEStronglyMeasurable c (mu.restrict ↑Omega) → (∀ x ∈ ↑Omega, ‖c x‖ ≤ gamma) → (∀ x ∈ ↑Omega, 0 ≤ c x) → MeasureTheory.MemLp F (↑q) (mu.restrict ↑Omega) → (∀ (v : ↥(W1p0 mu Omega 2)), (∀ᵐ (x : EuclideanSpace ℝ ι) ∂mu.restrict ↑Omega, 0 ≤ ↑↑(W1p.value ↑v) x) → energyFormH1 a 0 c u ↑v ≤ -∫ (x : EuclideanSpace ℝ ι) in ↑Omega, inner ℝ (F x) (↑↑(W1p.gradient ↑v) x) ∂mu) → W1p.posPartAbove UniformlyEllipticOn.ae_value_le_add_of_forall_eLpNorm_le._proof_1 hk u ∈ w1p0Submodule mu Omega 2 → ∀ᵐ (x : EuclideanSpace ℝ ι) ∂mu.restrict ↑Omega, ↑↑(W1p.value u) x ≤ k + C * (MeasureTheory.eLpNorm F (↑q) (mu.restrict ↑Omega)).toReal / lam * mu.real ↑Omega ^ ((↑(Fintype.card ι))⁻¹ - (↑q)⁻¹)

Stampacchia's global bound for a right-hand side in divergence form, in every dimension. Let n ≥ 1 be the dimension and let q > n with q > 2. There is C > 0, depending only on n, q and the normalization of the additive Haar measure mu, with the following property. Let Ω have finite measure, let a be measurable and uniformly elliptic on Ω with constants 0 < λ ≤ Λ, let c be bounded, measurable and nonnegative, and let F ∈ L^q(Ω; ℝⁿ). If u ∈ H¹(Ω) is a weak subsolution of

-∂ᵢ(aⁱʲ ∂ⱼu) + c u ≤ ∂ᵢFⁱ in Ω

with u ≤ k on ∂Ω for a level k ≥ 0, in the sense that (u - k)⁺ ∈ H¹₀(Ω), then almost everywhere in Ω

u ≤ k + C ‖F‖_{L^q(Ω)} / λ · |Ω| ^ (1/n - 1/q).

The bound is invariant under dilations of Ω; on a ball of radius ρ the factor |Ω| ^ (1/n - 1/q) is a multiple of the Morrey gain ρ ^ (1 - n/q). The condition q > 2 matters only for n = 1; it comes from the Sobolev exponents s = (1/2 - δ)⁻¹, δ < 1/2, which the method needs to satisfy s (1/2 - 1/q) > 1.

theorem TauCeti.PDE.exists_ae_abs_value_le_mul_div_mul_rpow_divergence {ι : Type u_1} [Fintype ι] [DecidableEq ι] {mu : MeasureTheory.Measure (EuclideanSpace ℝ ι)} [mu.IsAddHaarMeasure] [Nonempty ι] {q : NNReal} (hq : 2 < ↑q) (hqn : ↑(Fintype.card ι) < ↑q) :
∃ (C : ℝ), 0 < C ∧ ∀ {Omega : TopologicalSpace.Opens (EuclideanSpace ℝ ι)} {a : EuclideanSpace ℝ ι → Matrix ι ι ℝ} {c : EuclideanSpace ℝ ι → ℝ} {lam Lam gamma : ℝ} {F : EuclideanSpace ℝ ι → EuclideanSpace ℝ ι} {u : ↥(W1p0 mu Omega 2)}, mu ↑Omega ≠ ⊤ → UniformlyEllipticOn (↑Omega) a lam Lam → MeasureTheory.AEStronglyMeasurable a (mu.restrict ↑Omega) → MeasureTheory.AEStronglyMeasurable c (mu.restrict ↑Omega) → (∀ x ∈ ↑Omega, ‖c x‖ ≤ gamma) → (∀ x ∈ ↑Omega, 0 ≤ c x) → MeasureTheory.MemLp F (↑q) (mu.restrict ↑Omega) → (∀ (v : ↥(W1p0 mu Omega 2)), energyFormH1 a 0 c ↑u ↑v = -∫ (x : EuclideanSpace ℝ ι) in ↑Omega, inner ℝ (F x) (↑↑(W1p.gradient ↑v) x) ∂mu) → ∀ᵐ (x : EuclideanSpace ℝ ι) ∂mu.restrict ↑Omega, |↑↑(W1p.value ↑u) x| ≤ C * (MeasureTheory.eLpNorm F (↑q) (mu.restrict ↑Omega)).toReal / lam * mu.real ↑Omega ^ ((↑(Fintype.card ι))⁻¹ - (↑q)⁻¹)

Global boundedness of weak Dirichlet solutions with a right-hand side in divergence form, in every dimension. Let n ≥ 1 be the dimension and let q > n with q > 2. There is C > 0, depending only on n, q and the normalization of mu, such that on every Ω of finite measure, for measurable, uniformly elliptic a with constants 0 < λ ≤ Λ and bounded, measurable, nonnegative c, the weak solution u ∈ H¹₀(Ω) of the Dirichlet problem -∂ᵢ(aⁱʲ ∂ⱼu) + c u = ∂ᵢFⁱ with F ∈ L^q(Ω; ℝⁿ), that is a(u, v) = -∫_Ω F · ∇v for every v ∈ H¹₀(Ω), satisfies, almost everywhere in Ω,

|u| ≤ C ‖F‖_{L^q(Ω)} / λ · |Ω| ^ (1/n - 1/q).

This is the two-sided form of TauCeti.PDE.exists_ae_value_le_add_mul_div_mul_rpow_divergence at the level k = 0, applied to u and to -u.