Documentation

LeanPool.HighContrastHomogenization.Provider.Entry.AdapterAssembly

High-contrast homogenization: Provider.Entry.AdapterAssembly #

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

The adapted-to-Euclidean comparison constant #

The comparison between adapted and Euclidean cubes, in the shape the response and transfer proposition p.response.transfer binds: one constant, fixed before the law, with the two comparison estimates at every rounded grid, every alignment and every admissible triple of generations.

The two halves are assembled from AdapterForward and AdapterReverse. The cross-grid factors are the two absolute bounds of e.rounded.grid.bounds: |q^{-1}| ≤ 101/100 forward, so that half carries no eccentricity, and |q| ≤ (100/99)𝔢_q in reverse, which is where the printed 𝔢_q enters — exactly once, as the printed statement has it.

The constant. What is exhibited is C(d,g) = 15274d^3(1-g)^{-1}, not the printed dimension-only C_AE(d). The reverse half reads each boundary cell of the adapted filling through a Euclidean filling of its own, and that composition spends one more geometric series than the printed boundary decomposition does. The frozen consumer obligation binds its constant after the burn exponent g, so the dependence is admissible there; the printed dimension-only constant would need the boundary-shell reading of the source rather than the composed filling.

There are no definitions in this file.

theorem HCPolySupport.HighContrast.Entry.zpow_eq_rpow_neg_sub (i j : ℤ) :
3 ^ (-(↑j - ↑i)) = 3 ^ (i - j)

The integer power of three read as a real power of the negated difference.

theorem HCPolySupport.HighContrast.Entry.exists_euclidean_adapter (d : ℕ) (hd : 2 ≤ d) (g : ℝ) (hg : g ∈ Set.Ico 0 1) :
∃ (CAE : ℝ), 0 < CAE ∧ ∀ (P : MeasureTheory.Measure (CoeffSpace d)) (E : BlockMat d) (Ψ : ℝ → ℝ) (K : ℝ) (S : CoeffSpace d → ℝ), MeasureTheory.IsProbabilityMeasure P → IsStationaryLaw P → IsUnitRangeLaw P → CoarseEllipticityDagger P g E Ψ K S → ∀ (lAl : ℤ), ↑(kZero d) ≤ lAl → ∀ (mAl : Mat d), Matrix.PosDef mAl → ∀ (k n m : ℤ), 0 ≤ k → k < n → n < m → BlockMatLoewnerLE (blockSub (adaptedMean P (roundedGrid lAl mAl) n) (annealedBlock P (centeredCube d k))) (blockScale (CAE * witnessEccentricity mAl * transferGauge g K k * 3 ^ (-(↑n - ↑k))) E) ∧ (lAl ≤ n → BlockMatLoewnerLE (blockSub (annealedBlock P (centeredCube d m)) (adaptedMean P (roundedGrid lAl mAl) n)) (blockScale (CAE * witnessEccentricity mAl * transferGauge g K n * 3 ^ (-(↑m - ↑n))) E))

The comparison between adapted and Euclidean cubes, in the shape the response and transfer proposition binds: a constant fixed before the law, with the comparison in both directions, at every rounded grid and every alignment.