Documentation

LeanPool.HighContrastHomogenization.Provider.Transport.TransportCoefficients

High-contrast homogenization: Provider.Transport.TransportCoefficients #

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

The transport source coefficients #

The source residual the transport pays at a target level of the new grid is measured by the two source coefficients of the transport estimate: the continued one š–£_cont^{tr,š’®}(q,q') = C_dK(q,q')Īŗ_š„B_qB_{q'}χ_gš’°Ā², which the old-times-new orientation of the boundary rows produces, and the early one š–£_early^{tr,š’®}(q') = C_dĪŗ_š„B_{q'}Ā²š’°Ā², which the new-squared orientation of the whole-cell rows produces. The total coefficient š–£_src^{tr,š’®} = 1 + š–£_cont^{tr,š’®} + š–£_early^{tr,š’®} is what the transported source majorant is built from.

This file records the order relations between the three that every source estimate reads. Each factor of the two rows is nonnegative — the cross-grid factor is at least one, the reference ratio is an infimum of nonnegative numbers, the boundary constant is a nonnegative multiple of a square root, and the geometric series χ_g is positive below the critical exponent — so the two rows are nonnegative, the total is at least one, and it dominates each of them. Without the last two the printed majorant would not majorize the printed rows and the source estimate would be false as written.

The reference ratio is nonnegative, being an infimum of nonnegative numbers.

The geometric series χ_g = (3^{1-g} - 1)^{-1} is positive below the critical exponent.

theorem HCPolySupport.HighContrast.Transport.zero_le_transportContCoeff {d : ā„•} {Cd g : ā„} (hCd : 0 ≤ Cd) (hg : g < 1) (E : BlockMat d) (jStar : ℤ) (mu mu' : Mat d) :
0 ≤ transportContCoeff Cd g E jStar mu mu'

The continued transport coefficient š–£_cont^{tr,š’®}(q,q') = C_dK(q,q')Īŗ_š„B_qB_{q'}χ_gš’°Ā² is nonnegative.

theorem HCPolySupport.HighContrast.Transport.zero_le_transportEarlyCoeff {d : ā„•} {Cd g : ā„} (hCd : 0 ≤ Cd) (hg : g < 1) (E : BlockMat d) (mu' : Mat d) :

The early transport coefficient š–£_early^{tr,š’®}(q') = C_dĪŗ_š„B_{q'}Ā²š’°Ā² is nonnegative.

theorem HCPolySupport.HighContrast.Transport.one_le_transportSrcCoeff {d : ā„•} {Cd g : ā„} (hCd : 0 ≤ Cd) (hg : g < 1) (E : BlockMat d) (jStar : ℤ) (mu mu' : Mat d) :
1 ≤ transportSrcCoeff Cd g E jStar mu mu'

The total transport source coefficient is at least one, being one plus the two nonnegative rows.

theorem HCPolySupport.HighContrast.Transport.transportContCoeff_le_transportSrcCoeff {d : ā„•} {Cd g : ā„} (hCd : 0 ≤ Cd) (hg : g < 1) (E : BlockMat d) (jStar : ℤ) (mu mu' : Mat d) :
transportContCoeff Cd g E jStar mu mu' ≤ transportSrcCoeff Cd g E jStar mu mu'

The continued row is dominated by the total source coefficient.

theorem HCPolySupport.HighContrast.Transport.transportEarlyCoeff_le_transportSrcCoeff {d : ā„•} {Cd g : ā„} (hCd : 0 ≤ Cd) (hg : g < 1) (E : BlockMat d) (jStar : ℤ) (mu mu' : Mat d) :
transportEarlyCoeff Cd g E mu' ≤ transportSrcCoeff Cd g E jStar mu mu'

The early row is dominated by the total source coefficient.