High-contrast homogenization: Provider.Quenched.SmallContrastThresholdPolynomial #
Imported from the Apache-2.0 HighContrastHomogenization development at commit
7a13dbcd8d6609264a713373f5c69ceeac870472.
The thresholds are polynomial #
The near-reference comparison confines the reference aspect ratio to two scale
thresholds; the binder's sanctioned Π-slot is
3^{m_0} ≤ (2 + Π·K)^{C_delay}, so what must be checked is that three raised to
each threshold is polynomial in the data. Both are, and both by the same
elementary bound 3^{max 1 ⌈y⌉} ≤ 3·max 1 (3^y):
3^{euclideanEntryThreshold K cEnt sK} ≤ 3^{sK}·3·max 1 (M₂(K)/cEnt)— linear in the second source moment, i.e. polynomial inK_bar_S; this is's first input, which the rework was predicted to shrink, and it does;3^{tiltGap Cd g cEnt nu G} ≤ 3·max 1 ((2·boundaryConst·3^{gG}/cEnt)^{1/(1-g)})— and at the fusion's bindingnu = canonicalMetric 𝐄,G = canonicalGridEnlargement d Π, the base is polynomial inΠbywitnessEccentricity_canonicalMetric_le_aspectRatio, so the whole gap is. The exponent1/(1-g)is the printed(1-γ)^{-1}.
The elementary ceiling bound #
The Euclidean entry threshold #
's first input, at the rework's threshold. Three raised to the Euclidean entry threshold is linear in the second source moment.
The tilt gap #
Three raised to the tilt gap is polynomial in the boundary envelope,
with the printed (1-γ)^{-1} as the exponent.
The tilt gap at the fusion's binding is polynomial in Π #
The reference aspect ratio is at least one at a coarse-ellipticity reference.