Proposition 3, p. 9 — a fixed point ν̄ of ℚ_μ makes μ(·) the conditional probability of ν̄ given ℱ₁
ProvedCondRiskMap.Repr.proposition_3Let be paired locally convex spaces over satisfying (C) and (C′), and assume contains the indicator of every . Let be a positively homogeneous, lower semicontinuous conditional risk mapping with risk envelope , let be a weakly* -measurable selection of , and let be a fixed point of , i.e. for all . Then is the conditional probability of with respect to : for every the function is -measurable, and
Thus each measurable selection of the risk envelope is a regular conditional probability of some probability measure, which is how becomes a conditional expectation.
Formalization Note The hypothesis that contains all indicators , (the paper's example making (C) hold, p. 4), is added: the proof derives the -measurability of from weak* measurability, which needs ; without it the integral in (4.4) could be of a non-measurable function.
import Mathlib import Definitions.Def_CondRiskMap_Repr_Setting
namespace CondRiskMap.Repr
theorem proposition_3 {Ω : Type*} [MeasurableSpace Ω] {𝒳 𝒴 : Type*}
[AddCommGroup 𝒳] [Module ℝ 𝒳] [TopologicalSpace 𝒳]
[IsTopologicalAddGroup 𝒳] [ContinuousSMul ℝ 𝒳] [LocallyConvexSpace ℝ 𝒳]
[AddCommGroup 𝒴] [Module ℝ 𝒴] [TopologicalSpace 𝒴]
[IsTopologicalAddGroup 𝒴] [ContinuousSMul ℝ 𝒴] [LocallyConvexSpace ℝ 𝒴]
(S : PairedSpaces Ω 𝒳 𝒴) (hC : CondC S) (hC' : CondC' S) (hI : HasIndicators S)
(ρ : 𝒳 → 𝒳) (hρ : IsCondRiskMapping S ρ) (hph : PosHomogeneous ρ) (hlsc : IsLsc S ρ)
(κ : Ω → 𝒴) (hκ : IsSelection S ρ κ) (ν : 𝒴) (hν : ν ∈ Prob S)
(hfix : IsFixedPoint S κ ν) :
IsCondProb S ν κ := by sorry
end CondRiskMap.Repr
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.