Proposition 3.1, p. 12 — measurable selection theorem for closed-valued measurable multifunctions into ℝᵐ
ProvedDupacovaWets.Consistency.exists_measurable_selectionLet be a measurable space and a multifunction such that every is closed and is measurable. Then
and admits a measurable selector: a measurable with for every .
This is the measurable selection theorem (Kuratowski–Ryll-Nardzewski) quoted by Dupačová and Wets; in their consistency theorem it produces -measurable estimators from the measurable solution multifunctions.
Formalization Note The paper states it for on the data space ; it is stated here over an arbitrary measurable space, because the proof of Theorem 3.9 applies it on . The selector is a total measurable function that selects on ; since is measurable this is equivalent to the paper's (extend by a constant). is EuclideanSpace ℝ (Fin m).
import Mathlib import Definitions.Def_DupacovaWets_Consistency_Multifunction open MeasureTheory
namespace DupacovaWets.Consistency
/-- Dupačová–Wets (WP-86-41), p. 12, Proposition 3.1 (measurable selections), over an arbitrary
measurable space `T`: if `Γ : T ⇉ ℝᵐ` is closed-valued and measurable, then
`dom Γ = {t | Γ(t) ≠ ∅}` is measurable and `Γ` has a measurable selector on `dom Γ`
(here a measurable `x : T → ℝᵐ` with `x(t) ∈ Γ(t)` whenever `Γ(t) ≠ ∅`). -/
theorem exists_measurable_selection {T : Type*} [MeasurableSpace T] {m : ℕ}
(Γ : T → Set (EuclideanSpace ℝ (Fin m))) (hclosed : ∀ t, IsClosed (Γ t))
(hmeas : IsMeasurableMultifunction Γ) :
MeasurableSet {t | (Γ t).Nonempty} ∧
∃ x : T → EuclideanSpace ℝ (Fin m), Measurable x ∧ ∀ t, (Γ t).Nonempty → x t ∈ Γ t := by sorry
end DupacovaWets.Consistency
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.