High-contrast homogenization: #
Provider.PolynomialHomogenization.Root.Dirichlet.ExplicitScheduledRate
Imported from the Apache-2.0 HighContrastHomogenization development at commit
7a13dbcd8d6609264a713373f5c69ceeac870472.
(δ), the core: the scheduled rate at an explicit C₀ #
the flux scheduled rate at the witness
(UniformFluxScheduledRateExtraction:36) selects its constant as A.toReal,
where
A = ofReal outputFactor * ofReal Cdual *
((Cbesov² · ofReal Kflux)^(1/2) · hardyConstant^(1/2) +
(Cbesov² · ofReal Kflux)^(1/2) · hardyConstant^(1/2))
and hides it behind an ∃. Because the witness is opaque, the module can only
conclude ∃ C₀, 0 ≤ C₀ s ρ Rad ∧ … with C₀ bound inside ∀ abar U s ρ Rad
— which is exactly the defect the restated Dirichlet clause forbids.
This module repeats that proof with the head named and bounded:
scheduledRateHead is A, and rowConvertedFluxScheduledRateAtWitness_of_head_le
delivers RowConvertedFluxScheduledRateAtWitness d C₀ … for any C₀
dominating the head. scheduledRateHead_le_ofReal then converts a real-valued
bound on the four factors into that domination, so a law-free C₀ can be built
from the ρ-paid output factor, the flux slot split, that module (the
energy collapse) and the uniform Hardy constant.
The Hardy-top premise is not needed here: it served only to know A ≠ ⊤ so that
A.toReal could be formed. With the head bounded rather than reified, the
hypothesis drops.
(δ), the core. The converted flux rate holds at any C₀ that
dominates the scheduled head — in particular at a law-free one.
The real-valued bridge #
The head, bounded in real terms. All four factors of the scheduled head
are real (the Hardy constant through any real dominating it), so a law-free C₀
built from the modules named above dominates it.