Documentation

TauCeti.Analysis.PDE.Regularity.Logarithm

The logarithm of a positive weak supersolution #

Let a be measurable and uniformly elliptic on Ω ⊆ ℝⁿ with constants 0 < λ ≤ Λ, and let u ∈ H¹(Ω) be a weak supersolution of the divergence-form equation

-∂ᵢ(aⁱʲ ∂ⱼu) ≥ 0 in Ω,

meaning a(u, v) ≥ 0 for every nonnegative v ∈ H¹₀(Ω), which is bounded below by a constant ε > 0. Then log u is a Sobolev function with weak gradient ∇u / u (TauCeti.W1p.exists_value_gradient_ae_eq_comp_of_le), and this file shows that this gradient is controlled by the ellipticity constants alone:

The constants C depend only on the dimension; nothing depends on u or on the lower bound ε, which serves only to make u⁻¹ and log u Sobolev functions. No regularity of the coefficients beyond measurability is used.

The mean oscillation bound is the hypothesis of the John–Nirenberg inequality. In Moser's proof of the weak Harnack inequality, John–Nirenberg turns it into the crossover estimate ⨍_B u^p · ⨍_B u^{-p} ≤ C for a small p > 0, which links bounds on negative powers of a supersolution to bounds on positive ones.

The logarithmic Caccioppoli inequality is the case p = 0 of the Caccioppoli inequality for powers of u (TauCeti.PDE.UniformlyEllipticOn.setIntegral_sq_mul_rpow_mul_norm_gradient_sq_le), which tests the supersolution inequality against ψ² u^{p-1}, here ψ² u⁻¹.

Main declarations #

References #

theorem TauCeti.PDE.UniformlyEllipticOn.setIntegral_sq_mul_norm_gradient_div_value_sq_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} {lam Lam : ℝ} (h : UniformlyEllipticOn (↑Omega) a lam Lam) (ha : MeasureTheory.AEStronglyMeasurable a (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) → 0 ≤ energyFormH1 a 0 0 u ↑v) {ε : ℝ} (hε : 0 < ε) (hεu : ∀ᵐ (x : EuclideanSpace ℝ ι) ∂mu.restrict ↑Omega, ε ≤ ↑↑(W1p.value u) x) {ψ : EuclideanSpace ℝ ι → ℝ} (hψ : ContDiff ℝ (↑⊤) ψ) (hcpt : HasCompactSupport ψ) (hts : tsupport ψ ⊆ ↑Omega) :
∫ (x : EuclideanSpace ℝ ι) in ↑Omega, ψ x ^ 2 * (‖↑↑(W1p.gradient u) x‖ / ↑↑(W1p.value u) x) ^ 2 ∂mu ≤ (2 * Lam / lam) ^ 2 * ∫ (x : EuclideanSpace ℝ ι) in ↑Omega, ‖gradient ψ x‖ ^ 2 ∂mu

The logarithmic Caccioppoli inequality. Let a be measurable and uniformly elliptic on Ω with constants 0 < λ ≤ Λ, and let u ∈ H¹(Ω) be a weak supersolution of -∂ᵢ(aⁱʲ ∂ⱼu) ≥ 0, that is a(u, v) ≥ 0 for every nonnegative v ∈ H¹₀(Ω), with u ≥ ε almost everywhere for some ε > 0. Then for every smooth ψ compactly supported in Ω,

∫_Ω ψ² (‖∇u‖ / u)² ≤ (2Λ/λ)² ∫_Ω ‖∇ψ‖².

The integrand on the left is ψ² ‖∇ log u‖², and the right side depends on neither u nor ε.

theorem TauCeti.PDE.exists_setIntegral_ball_norm_gradient_div_value_sq_le {ι : Type u_1} [Fintype ι] [DecidableEq ι] :
∃ (C : ℝ), 0 ≤ C ∧ ∀ (mu : MeasureTheory.Measure (EuclideanSpace ℝ ι)) [inst : mu.IsAddHaarMeasure] {Omega : TopologicalSpace.Opens (EuclideanSpace ℝ ι)} {a : EuclideanSpace ℝ ι → Matrix ι ι ℝ} {lam Lam : ℝ} {u : ↥(W1p mu Omega 2)} {ε : ℝ} {x₀ : EuclideanSpace ℝ ι} {r : ℝ}, UniformlyEllipticOn (↑Omega) a lam Lam → MeasureTheory.AEStronglyMeasurable a (mu.restrict ↑Omega) → (∀ (v : ↥(W1p0 mu Omega 2)), (∀ᵐ (x : EuclideanSpace ℝ ι) ∂mu.restrict ↑Omega, 0 ≤ ↑↑(W1p.value ↑v) x) → 0 ≤ energyFormH1 a 0 0 u ↑v) → 0 < ε → (∀ᵐ (x : EuclideanSpace ℝ ι) ∂mu.restrict ↑Omega, ε ≤ ↑↑(W1p.value u) x) → 0 < r → Metric.ball x₀ (2 * r) ⊆ ↑Omega → ∫ (x : EuclideanSpace ℝ ι) in Metric.ball x₀ r, (‖↑↑(W1p.gradient u) x‖ / ↑↑(W1p.value u) x) ^ 2 ∂mu ≤ C * (Lam / lam) ^ 2 * mu.real (Metric.ball x₀ r) / r ^ 2

The gradient of log u on balls. There is a constant C ≥ 0, depending only on the dimension, with the following property. Let a be measurable and uniformly elliptic on Ω with constants 0 < λ ≤ Λ, and let u ∈ H¹(Ω) be a weak supersolution of -∂ᵢ(aⁱʲ ∂ⱼu) ≥ 0 with u ≥ ε almost everywhere for some ε > 0. Then on every ball with B(x₀, 2r) ⊆ Ω,

∫_{B(x₀, r)} (‖∇u‖ / u)² ≤ C (Λ/λ)² |B(x₀, r)| / r².

So the average of ‖∇ log u‖² over B(x₀, r) is at most C (Λ/λ)² r⁻², the power of r dictated by scaling.

theorem TauCeti.PDE.exists_setAverage_abs_log_sub_setAverage_le {ι : Type u_1} [Fintype ι] [DecidableEq ι] :
∃ (C : ℝ), 0 ≤ C ∧ ∀ (mu : MeasureTheory.Measure (EuclideanSpace ℝ ι)) [inst : mu.IsAddHaarMeasure] {Omega : TopologicalSpace.Opens (EuclideanSpace ℝ ι)} {a : EuclideanSpace ℝ ι → Matrix ι ι ℝ} {lam Lam : ℝ} {u : ↥(W1p mu Omega 2)} {ε : ℝ} {x₀ : EuclideanSpace ℝ ι} {r : ℝ}, UniformlyEllipticOn (↑Omega) a lam Lam → MeasureTheory.AEStronglyMeasurable a (mu.restrict ↑Omega) → (∀ (v : ↥(W1p0 mu Omega 2)), (∀ᵐ (x : EuclideanSpace ℝ ι) ∂mu.restrict ↑Omega, 0 ≤ ↑↑(W1p.value ↑v) x) → 0 ≤ energyFormH1 a 0 0 u ↑v) → 0 < ε → (∀ᵐ (x : EuclideanSpace ℝ ι) ∂mu.restrict ↑Omega, ε ≤ ↑↑(W1p.value u) x) → 0 < r → Metric.ball x₀ (2 * r) ⊆ ↑Omega → ⨍ (x : EuclideanSpace ℝ ι) in Metric.ball x₀ r, |Real.log (↑↑(W1p.value u) x) - ⨍ (y : EuclideanSpace ℝ ι) in Metric.ball x₀ r, Real.log (↑↑(W1p.value u) y) ∂mu| ∂mu ≤ C * (Lam / lam)

The logarithm of a positive supersolution has bounded mean oscillation. There is a constant C ≥ 0, depending only on the dimension, with the following property. Let a be measurable and uniformly elliptic on Ω with constants 0 < λ ≤ Λ, and let u ∈ H¹(Ω) be a weak supersolution of -∂ᵢ(aⁱʲ ∂ⱼu) ≥ 0 with u ≥ ε almost everywhere for some ε > 0. Then on every ball with B(x₀, 2r) ⊆ Ω,

⨍_{B(x₀, r)} |log u - ⨍_{B(x₀, r)} log u| ≤ C Λ/λ.

The bound is uniform over all such balls, and independent of u and ε: this is the hypothesis of the John–Nirenberg inequality for log u.