Lemma 2: conditional misclassification identity
ProvedQAlgorithms.Papers.Independent.Advance50.P50.lemma2Let be the conditional probability laws of a measurable exact model output in the two classes, sampled with equal prior probabilities. Suppose almost surely in class 1 and almost surely in class 0. Set . Then
The posterior is defined by the actual atom probabilities under the equal-prior mixture. Its conditioning event has mixture probability at least , so no conditional density or null-event convention is used.
import Definitions.Def_QAlgorithms_Papers_Independent_Advance50_P50
noncomputable section
namespace QAlgorithms.Papers.Independent.Advance50.P50
open MeasureTheory ProbabilityTheory
open scoped ComplexConjugate ENNReal
-- Exact accepted primitives; no paper-local redeclarations.
open QAlgorithms.Papers.Independent.Information100.P63 (Mat tensorPower)
open QAlgorithms.Papers.Independent.Gap48.P20 (Copies)
open QAlgorithms.Papers.Independent.Gap48.P38 (subsystemReduction)
open QAlgorithms.Papers.Independent.Information100.P32 (purity)
open QAlgorithms.Papers.Independent.Information100.P08 (IsPure)
-- Matrices carry the associative commutator bracket X*Y-Y*X.
attribute [local instance 100] LieRing.ofAssociativeRing
/-- PDF pp.26–27, Lemma 2, Eq.(E2): the omitted short-statement premise
that every class-1 output is c is retained from Eq.(E1). -/
theorem lemma2 {Omega : Type*} [MeasurableSpace Omega]
(mu0 mu1 : Measure Omega) [IsProbabilityMeasure mu0] [IsProbabilityMeasure mu1]
(X : Omega → ℝ) (hX : Measurable X) (c b1 b2 : ℝ)
(hclass1 : ∀ᵐ w ∂mu1, X w = c)
(hclass0 : ∀ᵐ w ∂mu0, X w ∈ Set.Icc b1 b2) :
posteriorAt mu0 mu1 X c =
(mu0 {w | X w = c}).toReal / (1 + (mu0 {w | X w = c}).toReal) := by sorry
end QAlgorithms.Papers.Independent.Advance50.P50
end
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
For every measurable space , probability measures , measurable function , and real numbers , assume that almost everywhere under and that almost everywhere under . Then, putting , regarded as a real number, the defined ratio satisfies
All masses in this formula are real conversions of measure values; they are finite because both measures are probabilities. The case is included. No separate assumption is made; if , the almost-everywhere interval hypothesis cannot hold for a probability measure.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.