High-contrast homogenization: #
Provider.PolynomialHomogenization.Root.Duality.CubeProjectionStabilityDischarge
Imported from the Apache-2.0 HighContrastHomogenization development at commit
7a13dbcd8d6609264a713373f5c69ceeac870472.
Discharging the cube projection stability #
The positive-direction input of the cube Hodge estimate is proved from material:
- existence of the zero-trace identity-coefficient solution of
div (∇v − psi) = 0on the open cube (exists_isZeroTraceDirichletRhsWeakSolution_of_potentialZeroTraceClosureRealizationthrough the convex-domain closure realization); - the fractional Calderón--Zygmund estimate for that solution on an origin cube
(
centeredCubeH10ScalarDivergence_fractional_cz_full); - the two norm bridges
cubeEuclideanWspFullENorm_two_sq_eq_hsNormSqand the packaging of a finite-fractional field as a cubeWsp ∩ L²field.
noncomputable def
HCPolySupport.HighContrast.RowSupply.wspL2FieldOfHsNormSq
{d : ℕ}
(Q : TriadicCube d)
{s : ℝ}
(hs : 0 < s)
(hsHalf : s < 1 / 2)
(G : Vec d → Vec d)
(hGmeas : MeasureTheory.AEStronglyMeasurable G (MeasureTheory.volume.restrict (openCubeSet Q)))
(hGfinite : hsNormSq (openCubeSet Q) s G ≠ ⊤)
:
Packaging of a measurable field of finite normalized H^s norm as a cube
Wsp ∩ L² field. (The construction of the same content is private at
its file; it is reproved here rather than imported.)
Equations
- HCPolySupport.HighContrast.RowSupply.wspL2FieldOfHsNormSq Q hs hsHalf G hGmeas hGfinite = { toField := G, euclideanMemLp := ⋯, euclideanMemWsp := ⋯, euclideanMemL2 := ⋯ }
Instances For
theorem
HCPolySupport.HighContrast.RowSupply.exists_cubeGradientProjectionHsStability_originCube
(d : ℕ)
[NeZero d]
:
∃ (Cproj : ℝ),
0 ≤ Cproj ∧ ∀ (m : ℤ) {s : ℝ}, 0 < s → s < 1 / 2 → CubeGradientProjectionHsStability d (originCube d m) s Cproj
The cube projection stability is proved. Every admissible test on an
origin cube splits into a zero-trace potential part and a solenoidal remainder,
with the potential part H^s-bounded by the test, at a constant fixed before
the scale, the order and the test.