Coarse-graining support: Support.Book.Ch05.Theorems.Section57.BadScaleTailCollapse #
Imported from the Apache-2.0 CoarseGraining development at commit
c7ddd76c08ade64fed1b8d2ca51be14dfee8deb4.
Collapse of synchronized bad-scale components to one tail #
This file contains the deterministic final step after the three bad-scale
components have been estimated with synchronized constants. The probabilistic
input is the selected-denominator component theorem from BadScaleTailJoint;
the remaining hypotheses are purely large-scale/prefactor inequalities.
theorem
HCPolySupport.Book.Ch05.Section57.measureReal_badScaleEvent_le_exp_tail_of_component_prefactor_gap
{Ω : Type u_1}
[MeasurableSpace Ω]
{μ : MeasureTheory.Measure Ω}
{H : ℕ → ℕ → Ω → ℝ}
{t α : ℝ}
{q : ℕ}
{Ctop ChighBottom CcrudeBottom A Acrude Alead Atail η : ℝ}
(hcomponent :
μ.real (badScaleEvent H t α q) ≤ Ctop * Real.exp (-A ^ η) + ChighBottom * Real.exp (-A ^ η) + CcrudeBottom * Real.exp (-Acrude ^ η))
(hAlead_A : Alead ≤ A ^ η)
(hAlead_Acrude : Alead ≤ Acrude ^ η)
(hpref : max 0 Ctop + max 0 ChighBottom + max 0 CcrudeBottom ≤ Real.exp (Alead - Atail))
:
Deterministic component collapse in the exact algebraic shape produced by the synchronized selected-denominator estimate.
theorem
HCPolySupport.Book.Ch05.Section57.measureReal_badScaleEvent_le_exp_tail_of_selected_component_sum
{Ω : Type u_1}
[MeasurableSpace Ω]
{μ : MeasureTheory.Measure Ω}
{H : ℕ → ℕ → Ω → ℝ}
{t α : ℝ}
{q : ℕ}
{S qPlus Cbottom wq Ktop Kbottom Kcrude A Acrude Alead Atail η : ℝ}
(hcomponent :
μ.real (badScaleEvent H t α q) ≤ S * (Real.exp (-A ^ η) * Ktop) + qPlus * (Cbottom * wq) * (Real.exp (-A ^ η) * Kbottom) + qPlus * (S * wq) * (Real.exp (-Acrude ^ η) * Kcrude))
(hAlead_A : Alead ≤ A ^ η)
(hAlead_Acrude : Alead ≤ Acrude ^ η)
(hpref :
max 0 (S * Ktop) + max 0 (qPlus * (Cbottom * wq) * Kbottom) + max 0 (qPlus * (S * wq) * Kcrude) ≤ Real.exp (Alead - Atail))
:
Same deterministic collapse, with the component sum written in the selected-denominator theorem's native factorization.