Documentation

LeanPool.HighContrastHomogenization.Support.Deterministic.ConstantCoefficientDirichletBesov.StandardProjectionBoundaryGap

Coarse-graining support: #

Support.Deterministic.ConstantCoefficientDirichletBesov.StandardProjectionBoundaryGap

Imported from the Apache-2.0 CoarseGraining development at commit c7ddd76c08ade64fed1b8d2ca51be14dfee8deb4.

Projection gaps from boundary-budgeted increments #

This file sits downstream of StandardProjectionBoundary: the boundary file controls one martingale increment by an abstract ancestor budget. Here we insert that one-increment estimate into the already-proved telescoping estimate for the projection gap P_j u - P_0 u.

theorem HCPolySupport.cubeBesovOverlappingPositiveVectorDepthSeminorm_cubeIncrementVec_le_sqrt_boundaryBudget {d : ℕ} (Q : TriadicCube d) (s : ℝ) (u : Vec d → Vec d) {j m : ℕ} (hmj : m ≤ j) (B : TriadicCube d → ℝ) (hB : ∀ R ∈ descendantsAtDepth Q (m + 1), 0 ≤ B R) (hbound : ∀ R ∈ descendantsAtDepth Q (m + 1), ∀ S ∈ descendantBoundaryLayerAtDepth R (j - m), overlapCubeLpNorm S 2 (overlapCubeFluctuationVec S (cubeIncrementVec Q (m + 1) u)) ^ 2 ≤ B R) :
cubeBesovOverlappingPositiveVectorDepthSeminorm Q s (cubeIncrementVec Q (m + 1) u) j ≤ Real.rpow 3 (s * ↑j) * √((↑((3 ^ d) ^ j))⁻¹ * (2 * ↑d * (3 ^ (d - 1)) ^ (j - m) * ∑ R ∈ descendantsAtDepth Q (m + 1), B R))

Seminorm form of the one-increment boundary-budget estimate.

theorem HCPolySupport.cubeBesovOverlappingPositiveVectorDepthSeminorm_gap_zero_le_sum_sqrt_boundaryBudget {d : ℕ} (Q : TriadicCube d) (s : ℝ) (u : Vec d → Vec d) (j : ℕ) (B : ℕ → TriadicCube d → ℝ) (hincLoc : ∀ S ∈ overlapCentersAtDepth Q j, ∀ m ∈ Finset.range j, MeasureTheory.MemLp (cubeIncrementVec Q (m + 1) u) 2 (normalizedOverlapCubeMeasure S)) (hB : ∀ m ∈ Finset.range j, ∀ R ∈ descendantsAtDepth Q (m + 1), 0 ≤ B m R) (hbound : ∀ m ∈ Finset.range j, ∀ R ∈ descendantsAtDepth Q (m + 1), ∀ S ∈ descendantBoundaryLayerAtDepth R (j - m), overlapCubeLpNorm S 2 (overlapCubeFluctuationVec S (cubeIncrementVec Q (m + 1) u)) ^ 2 ≤ B m R) :
cubeBesovOverlappingPositiveVectorDepthSeminorm Q s (cubeProjectionGapVec Q 0 j u) j ≤ ∑ m ∈ Finset.range j, Real.rpow 3 (s * ↑j) * √((↑((3 ^ d) ^ j))⁻¹ * (2 * ↑d * (3 ^ (d - 1)) ^ (j - m) * ∑ R ∈ descendantsAtDepth Q (m + 1), B m R))

The projection gap is controlled by the sum of square-root boundary budgets for its martingale increments.