Sect. 5.1.1, pp. 14–15 — A = B up to a zero-probability set
ProvedWaitJudge.Convex.A_ae_eq_BLet Assumptions 1 and 2 hold, fix and a threshold , and write , for the solution and the number of support constraints of the program built on the first scenarios only. Let $$A={V(x^_N)>\epsilon(k)\ \wedge\ s^_N=k\ \wedge\ \text{the first constraints are of support}},
B={V(x^_k)>\epsilon(k)\ \wedge\ s^_k=k\ \wedge\ \text{the constraints with indexes are satisfied by }}.$$ Then and differ by a -null set:
This equality reduces the probability of , which involves the full program, to that of , which involves the first scenarios and an independent check of the remaining ; that is the route to formula (16).
Formalization Note Indices are -based: the first scenarios are and the remaining ones have index . The statement is an almost-everywhere equality of sets and needs no measurability hypothesis.
import Mathlib import Definitions.Def_ScenarioApproach_Generalization_violation import Definitions.Def_WaitJudge_Convex_Setting open MeasureTheory ScenarioApproach.Generalization
namespace WaitJudge.Convex
theorem A_ae_eq_B {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)
(hA1 : Assumption1 c X Xδ tb) (hA2 : Assumption2 c X Xδ tb P)
(ε : ℕ → ℝ) (N k : ℕ) (hk : k ≤ N) :
{ω : Fin N → Δ | 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}
=ᵐ[Measure.pi fun _ : Fin N => P]
{ω : Fin N → Δ | ε k < violation P Xδ (xstar c X Xδ tb (firstK hk ω)) ∧
sstar c X Xδ tb (firstK hk ω) = k ∧
∀ j : Fin N, k ≤ j.val → xstar c X Xδ tb (firstK hk ω) ∈ Xδ (ω j)} := by sorry
end WaitJudge.Convex
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.