Lemma 7.6 — the density of an image measure is the conditional expectation of the density
ProvedWeakMFG.Existence.lemma_7_6conditional-expectationmeasure-theoryp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1radon-nikodym
Let and be measurable spaces, let be probability measures on with , and let be measurable. Then
In the paper this transfers bounds on the density to the density of the law with respect to (Lemma 7.7).
Formalization Note Radon–Nikodym derivatives are Mathlib's rnDeriv, turned into real numbers; the conditional expectation is taken given the -field .
Preamble
import Mathlib open MeasureTheory ProbabilityTheory Filter Topology open scoped ENNReal NNReal Matrix
Formal statement
namespace WeakMFG.Existence
/-- Lemma 7.6 (Carmona–Lacker, arXiv:1307.1152v2, p. 24): let `(E, ℰ)`, `(F, ℱ)` be measurable
spaces, `μ, ν ∈ P(E)` with `ν ≪ μ`, and `X : E → F` measurable. Then
`d(ν ∘ X⁻¹)/d(μ ∘ X⁻¹) ∘ X = E^μ[dν/dμ | X]` `μ`-a.s.
Formalization Note: Radon–Nikodym derivatives are Mathlib's `rnDeriv` (in `ℝ≥0∞`, real-valued via
`toReal`); `E^μ[· | X]` is the conditional expectation given `σ(X) = ℱ.comap X`. -/
theorem lemma_7_6 {E F : Type*} [MeasurableSpace E] [mF : MeasurableSpace F]
(μ ν : Measure E) [IsProbabilityMeasure μ] [IsProbabilityMeasure ν] (hνμ : ν ≪ μ)
(X : E → F) (hX : Measurable X) :
(fun x => ((ν.map X).rnDeriv (μ.map X) (X x)).toReal) =ᵐ[μ]
μ[fun x => (ν.rnDeriv μ x).toReal | MeasurableSpace.comap X mF] := by sorry
end WeakMFG.Existence
Source
Carmona, Lacker, A probabilistic weak formulation of mean field games and applications, arXiv:1307.1152v2 (2014), Lemma 7.6, p. 24
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.