(14), Sect. 5.1.1, p. 14 — ℙᴺ{S_k} = C(N,k) ℙᴺ{A} by exchangeability
ProvedWaitJudge.Convex.prob_Sk_eq_choose_mulexchangeabilityp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1scenario-optimizationsupport-constraints
Let be i.i.d. with law , and let the measurability conditions hold. Fix and a threshold . Let
and let be the event (15) where, in addition, the first constraints are of support. Then
The identity groups the samples of by the index set of their support constraints; the groups have the same probability because the sample is i.i.d. and the solution does not depend on the order of the constraints.
Formalization Note "The first constraints" are those with indices . The measurability hypotheses (joint measurability of and of the solution maps) make both events measurable; the paper takes measurability for granted. No convexity and no Assumption 1 or 2 is needed.
Preamble
import Mathlib import Definitions.Def_ScenarioApproach_Generalization_violation import Definitions.Def_WaitJudge_Convex_Setting open MeasureTheory ScenarioApproach.Generalization
Formal statement
namespace WaitJudge.Convex
theorem prob_Sk_eq_choose_mul {d p : ℕ} {Δ : Type*} [MeasurableSpace Δ] (P : Measure Δ) [IsProbabilityMeasure P]
(c : E d) (X : Set (E d)) (Xδ : Δ → Set (E d)) (tb : Fin p → E d → ℝ)
(hmeas : MeasurabilityPins c X Xδ tb) (ε : ℕ → ℝ) (N k : ℕ) (hk : k ≤ N) :
(Measure.pi fun _ : Fin N => P) {ω | ε k < violation P Xδ (xstar c X Xδ tb ω) ∧ sstar c X Xδ tb ω = k} =
((N.choose k : ℕ) : ENNReal) * (Measure.pi fun _ : Fin N => P)
{ω | ε k < violation P Xδ (xstar c X Xδ tb ω) ∧ sstar c X Xδ tb ω = k ∧
∀ i : Fin N, i.val < k → i ∈ supportSet c X Xδ tb ω} := by sorry
end WaitJudge.Convex
Source
Campi & Garatti, Wait-and-judge scenario optimization, Math. Program. (2018), doi:10.1007/s10107-016-1056-9 (accepted manuscript), PDF pp. 13–14, Sect. 5.1.1, (14) and (15)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.