Event and complementary event marginal laws imply independent event indicator
ProvedAvramDividend.Classical.indicator_indepFun_of_both_event_marginalscomplement-eventevent-independenceprobabilitystrong-markov
For a probability measure, if both an event A and its complement satisfy the expected factorisation with each measurable inverse image under a random element Y, then the indicator of A is independent of Y.
Preamble
import Mathlib.Probability.Independence.Basic import Mathlib.MeasureTheory.Measure.Restrict open MeasureTheory ProbabilityTheory Set
Formal statement
theorem AvramDividend.Classical.indicator_indepFun_of_both_event_marginals
{Ω β : Type*} [MeasurableSpace Ω] [MeasurableSpace β]
{P : Measure Ω} [IsProbabilityMeasure P]
(A : Set Ω) (Y : Ω → β)
(hA : MeasurableSet A) (hY : Measurable Y)
(hFirst : ∀ B : Set β, MeasurableSet B →
P (A ∩ Y ⁻¹' B) = P A * P (Y ⁻¹' B))
(hSecond : ∀ B : Set β, MeasurableSet B →
P (Aᶜ ∩ Y ⁻¹' B) = P Aᶜ * P (Y ⁻¹' B)) :
IndepFun (Set.indicator A (fun _ : Ω => (1 : ℝ))) Y P := by
sorrySource
Bridges the proved restricted-map event-marginal identity f2cff3c4-e6ba-4ab0-92c3-ede656bfd82d and the complement stopped-history event lemma 14adf425-f38f-4621-b8d2-fdc1e495b583 to the stochastic full-path joint-law theorem 61930a38-11f1-449e-bbb6-67d59c6d5ca8.