(37), pp. 25–26 — weak duality with M=N
ProvedWaitJudge.Generic.weak_duality_Ndualitymoment-problemp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1scenario-optimization
Fix N and thresholds ε(k)∈[0,1]. Let μ_k be finite nonnegative measures whose first N+1 moment equations agree with the truncated primal problem (35). Let λ_0,…,λ_N satisfy the pointwise dual inequalities (36) with M=N. Then
This is the finite weak-duality inequality used to pass from the distributional representation of the violation probability to the dual value γ*_N.
Formalization Note The dual variable is a function on natural numbers, but only indices at most N occur. The dual inequalities use t=1−v, so v∈(ε(k),1] becomes t∈[0,1−ε(k)). Finiteness of the measures ensures the displayed real integrals have their intended values.
Preamble
import Mathlib import Definitions.Def_WaitJudge_Generic_Setting
Formal statement
namespace WaitJudge.Generic
open MeasureTheory
theorem weak_duality_N
(N : ℕ) (ε : ℕ → ℝ)
(hε : ∀ k ≤ N, 0 ≤ ε k ∧ ε k ≤ 1)
(μ : ℕ → Measure ℝ)
(hfinite : ∀ k ≤ N, μ k Set.univ < ⊤)
(hmoment : ∀ m ≤ N,
∑ k ∈ Finset.range (m + 1), (m.choose k : ℝ) *
∫ v in Set.Icc (0 : ℝ) 1, (1 - v) ^ (m - k) ∂(μ k) = 1)
(lam : ℕ → ℝ) (hdual : IsFeasibleDualN N ε lam) :
∑ k ∈ Finset.range (N + 1), (N.choose k : ℝ) *
∫ v in Set.Ioc (ε k) 1, (1 - v) ^ (N - k) ∂(μ k) ≤
∑ m ∈ Finset.range (N + 1), lam m := by sorry
end WaitJudge.Generic
Source
Campi & Garatti, Wait-and-judge scenario optimization, Math. Program. (2018), doi:10.1007/s10107-016-1056-9 (accepted manuscript), PDF pp. 25–26, (35)–(37), specialized to M=N
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.