Proof of Proposition 5.3, (5.17)–(5.18), p. 8 — a measurable maximiser over a finite set of measurable functions exists
ProvedDelayImpulse.Optimal.proof_5_3_argmaximpulse-controlmeasurable-selectionp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1
Let be a measurable space, a measurable space in which singletons are measurable, a finite nonempty set, and a -measurable function for each . Then there is a -measurable with values in such that, for every ,
This is the construction of p. 8, stated for a general finite family: enumerating and putting on , where is the set on which attains the maximum. Applied with and the conditional expectations in (5.17), it produces the -measurable maximisers of Theorem 6.1.
Formalization Note The maximum is Finset.sup'.
Preamble
import Mathlib open MeasureTheory
Formal statement
namespace DelayImpulse.Optimal
theorem proof_5_3_argmax {Ω E : Type} [MeasurableSpace E] [MeasurableSingletonClass E]
(G : MeasurableSpace Ω) (U : Finset E) (hU : U.Nonempty) (X : E → Ω → ℝ)
(hX : ∀ β ∈ U, Measurable[G] (X β)) :
∃ b : Ω → E, Measurable[G] b ∧ (∀ ω, b ω ∈ U) ∧
∀ ω, X (b ω) ω = U.sup' hU (fun β => X β ω) := by sorry
end DelayImpulse.Optimal
Source
Djehiche, Hamadène, Hdhiri & Zaatra, Infinite Horizon Stochastic Impulse Control with Delay and Random Coefficients, arXiv:1904.11924v2, p. 8, proof of Proposition 5.3, (5.17)–(5.18) and the construction of βⁿ₀ via the sets 𝒜ᵢ
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.