Documentation

LeanPool.HighContrastHomogenization.Provider.Entry.AdapterCellMean

High-contrast homogenization: Provider.Entry.AdapterCellMean #

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

The annealed bound on an aligned adapted cell #

Averaging the Euclidean filling of z + โ‹„_r^q against the Euclidean cell bound E[๐€(z+โ–ก_i)] โ‰ค (1+9K_{ฮจ_S}^23^{-i})^g๐„ and summing the geometric series gives E[๐€(z+โ‹„_r^q)] โ‰ค C_d(1-g)^{-1}(1+9K_{ฮจ_S}^23^{-r})^g๐„ at every generation and every aligned centre, with C_d = 126d^{3/2}.

This is the bound e.coarse.ellipticity does not give directly: the dagger speaks of standard cubes, and an adapted cell is only reached through its own Euclidean filling. The single factor (1-g)^{-1} is the geometric series of that filling; it is what the composed reading of the reverse comparison spends over the printed boundary decomposition.

There are no definitions in this file.

theorem HCPolySupport.HighContrast.Entry.annealedBlock_adaptedCellAt_le {d : โ„•} [NeZero d] {P : MeasureTheory.Measure (CoeffSpace d)} [MeasureTheory.IsProbabilityMeasure P] {g : โ„} {E : BlockMat d} {ฮจ : โ„ โ†’ โ„} {K : โ„} {S : CoeffSpace d โ†’ โ„} (hd : 1 โ‰ค d) (hstat : IsStationaryLaw P) (hdag : CoarseEllipticityDagger P g E ฮจ K S) {q : Mat d} (hq : Matrix.PosDef q) (hqinv : โ€–qโปยนโ€– โ‰ค 101 / 100) (r : โ„ค) (w : Fin d โ†’ โ„ค) :
BlockMatLoewnerLE (annealedBlock P (adaptedCellAt q r w)) (blockScale (126 * โ†‘d * โˆšโ†‘d * (1 - g)โปยน * (1 + 9 * K ^ 2 * 3 ^ (-r)) ^ g) E)

The annealed bound on an aligned adapted cell. Averaging the Euclidean filling of z + โ‹„_r^q against the Euclidean cell bound and summing the geometric series gives E[๐€(z+โ‹„_r^q)] โ‰ค C_d(1-g)^{-1}(1+9K_{ฮจ_S}^23^{-r})^g๐„, at every generation and every aligned centre, with C_d = 126d^{3/2}. The one factor (1-g)^{-1} is the geometric series of the filling.