§4.3, proof of Theorem 3, p. 25, fourth paragraph — 𝔼[z*(2ξ̄ − ξ)] = 𝔼[z*(ξ)] under symmetry
ProvedPDScenRed.Mean.integral_zsel_reflectp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1scenario-reductionsymmetry
Let be a nonempty bounded polytope and let be a measurable selection of minimizers, for every . Let be an integrable random vector with law and mean , and suppose and are equal in distribution (Assumption 5c). Then
Combined with the previous step, this makes the two parts and of the subdifferential of at cancel, so that .
Formalization Note Measurability of is needed for to be meaningful; since is bounded, is then bounded and both expectations are genuine (not the Bochner integral's junk value).
Preamble
import Mathlib import Definitions.Def_PDScenRed_Mean_Setting open scoped InnerProductSpace open MeasureTheory
Formal statement
namespace PDScenRed.Mean
theorem integral_zsel_reflect {d r : ℕ} (P : Matrix (Fin r) (Fin d) ℝ) (qv : Fin r → ℝ)
(hne : (polytope P qv).Nonempty) (hbdd : Bornology.IsBounded (polytope P qv))
(zsel : EuclideanSpace ℝ (Fin d) → EuclideanSpace ℝ (Fin d))
(hsel : ∀ ξ, zsel ξ ∈ SmartPTO.Fisher.Wstar (polytope P qv) ξ) (hmeas : Measurable zsel)
(μ : Measure (EuclideanSpace ℝ (Fin d))) [IsProbabilityMeasure μ]
(hint : Integrable (fun ξ : EuclideanSpace ℝ (Fin d) => ξ) μ)
(hsym : SmartPTO.Fisher.CentrallySymmetric μ) :
∫ ξ, zsel ((2 : ℝ) • (∫ x, x ∂μ) - ξ) ∂μ = ∫ ξ, zsel ξ ∂μ := by sorry
end PDScenRed.Mean
Source
Bertsimas & Mundru, Optimization-based Scenario Reduction for Data-Driven Two-stage Stochastic Optimization, author manuscript (MIT DSpace), p. 25, §4.3, proof of Theorem 3, fourth paragraph, first sentence
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.