High-contrast homogenization: Provider.Quenched.SmallContrastEntryRateMean #
Imported from the Apache-2.0 HighContrastHomogenization development at commit
7a13dbcd8d6609264a713373f5c69ceeac870472.
The mean slot's source, and hentry at named constants #
the corresponding argument.1 left one hole: hVmeanle : Vmean ⤠vmsrc + cVm · F j_b was consumed
everywhere and produced nowhere. This file closes it from the construction's own
mean-slot machinery and instantiates the leg-bound entry estimate so that hentry
stands fully discharged at named constants.
Where Vmean comes from. entry_supply_mean_slot_bound_at
(the cited theorem) converts the reference-normalized terminal
scale variance into the mean-normalized one:
scaleVariance P q (adaptedMean P q t) t ⤠ofReal (meanSlotConversion C_d g m_Al G_acc š Ā· Vā),
i.e. Vmean ⤠conv Ā· Vā with conv = meanSlotConversion ā¦. The slot family
already bounds Vā ā it is slotFamilyValue d Csub Msc delta F(j_b) J 0,
which is by definition slotSourceSeq ⦠J 0 + slotBaseCoefficient d · F(j_b).
So the mean slot splits into a source part and a base part with no new
estimate, only the conversion factor:
vmsrc = conv Ā· slotSourceSeq d Csub Msc delta J 0,
cVm = conv Ā· slotBaseCoefficient d.
Its decay is the same 3^{-J/2} as the variance leg, because
slotSourceSeq ⦠J 0 = slotSourceValue d Csub Msc J.
The hVmeanle hypothesis #
The conversion splits the mean slot. Any bound Vmean ⤠conv Ā· Vā with
Vā in source-plus-base shape gives hVmeanle's shape.
The mean slot's base coefficient.
Equations
Instances For
hVmeanle at the slot family. The Vā bound is definitional, so this
is the conversion step alone.