Documentation

LeanPool.HighContrastHomogenization.Provider.PolynomialHomogenization.Root.Corrector.C1ExactGaugeFrameTransport

High-contrast homogenization: #

Provider.PolynomialHomogenization.Root.Corrector.C1ExactGaugeFrameTransport

Imported from the Apache-2.0 HighContrastHomogenization development at commit 7a13dbcd8d6609264a713373f5c69ceeac870472.

Exact gauge-frame transport for the C1 estimate #

The normalized-root frame uses the exact centered coefficient and the canonical joint corrector. An estimate in that frame returns to the physical ellipsoids without introducing a rounded coefficient family.

theorem HCPolySupport.HighContrast.Root.c1DecayBound_weaken_exponent (A eta etaFast r R : ℝ) (hA : 0 ≤ A) (hr : 0 < r) (hR : 0 < R) (hrR : r ≤ R) (heta : eta ≤ etaFast) {X Y : ENNReal} (hbound : X ≤ ENNReal.ofReal (A * (r / R) ^ etaFast) * Y) :
X ≤ ENNReal.ofReal (A * (r / R) ^ eta) * Y

A decay estimate at a larger exponent implies the same estimate at a smaller exponent on a nested-radius interval.

theorem HCPolySupport.HighContrast.Root.delayedPhysicalC1_of_exactRootGaugeFrame {d : ℕ} [NeZero d] {abar : Mat d} (hS : Matrix.PosDef (symmPart abar)) (a : CoeffSpace d) (gradPhi : Vec d → CoeffSpace d → Vec d → Vec d) (PhiRef : Vec d → NormalizedLocalH1Carrier d) (hpullback : ∀ (e : Vec d), (fun (y : Vec d) => matVecMul (Selection.normalizedRoot (symmPart abar)) (e + gradPhi e a (matVecMul (Selection.normalizedRoot (symmPart abar)) y))) =ᵐ[MeasureTheory.volume] fun (y : Vec d) => matVecMul (Selection.normalizedRoot (symmPart abar)) e + (PhiRef (matVecMul (Selection.normalizedRoot (symmPart abar)) e)).globalGradientRepresentative y) (L : ℕ) (eta A x R : ℝ) (Du : Vec d → Vec d) :
(have rStart := 3 ^ (Quenched.triadicCeilingIndex x + L); rStart ≤ R) → (let q := Selection.normalizedRoot (symmPart abar); have hq := ⋯; have bRef := affineCoefficient q hq ↑↑(normalizedCenteredCoeff a abar hS); have rStart := 3 ^ (Quenched.triadicCeilingIndex x + L); ∃ (eRef : Vec d), ∀ r ∈ Set.Icc rStart R, (weightedGradNorm bRef {y : Vec d | vecNormSq y ≤ r ^ 2} fun (y : Vec d) => matVecMul q (Du (matVecMul q y)) - (eRef + (PhiRef eRef).globalGradientRepresentative y)) ≤ ENNReal.ofReal (A * (r / R) ^ eta) * weightedGradNorm bRef {y : Vec d | vecNormSq y ≤ R ^ 2} fun (y : Vec d) => matVecMul q (Du (matVecMul q y))) → ∃ (e : Vec d), ∀ r ∈ Set.Icc (3 ^ (Quenched.triadicCeilingIndex x + L)) R, (weightedGradNorm (fun (y : Vec d) => ↑↑a y) (ellipsoid abar r) fun (y : Vec d) => Du y - (e + gradPhi e a y)) ≤ ENNReal.ofReal (A * (r / R) ^ eta) * weightedGradNorm (fun (y : Vec d) => ↑↑a y) (ellipsoid abar R) Du

A delayed C1 estimate for the exact normalized coefficient and canonical joint corrector transports to the same delayed estimate on physical ellipsoids.