Measurable restriction of an independent first-coordinate law retains product factorisation
ProvedKyprianou.Ch01.mapRestrictedIndependentProductindependencekyprianou-ch01measure-restrictionproduct-measureworkstream-a
If f and g are measurable independent random elements, restrict their common probability measure to the event f lies in a measurable set A. The resulting joint pushforward law remains the product of the restricted f-marginal and the original g-marginal. This is a general measure-level independence preservation lemma with no assumption about a particular stochastic model.
Preamble
import Mathlib.Probability.Independence.Basic import Mathlib.MeasureTheory.Measure.Prod import Mathlib.MeasureTheory.Measure.Restrict import Mathlib.MeasureTheory.Measure.Map open MeasureTheory ProbabilityTheory Set open scoped ENNReal NNReal set_option autoImplicit false
Formal statement
theorem Kyprianou.Ch01.mapRestrictedIndependentProduct
{Ω α β : Type*} [MeasurableSpace Ω]
[MeasurableSpace α] [MeasurableSpace β]
(P : Measure Ω) [IsProbabilityMeasure P]
(f : Ω → α) (g : Ω → β)
(hf : Measurable f) (hg : Measurable g)
(hind : IndepFun f g P)
(A : Set α) (hA : MeasurableSet A) :
Measure.map (fun ω => (f ω, g ω))
(P.restrict (f ⁻¹' A)) =
(Measure.map f (P.restrict (f ⁻¹' A))).prod
(Measure.map g P) := by
sorrySource
Kyprianou (2006), Exercise 1.8(ii), conditioning on the strictly positive prefix path, worked solution page 304