Proposition 1 — Projection of the Wasserstein ball onto the mean-covariance space
DisprovedWassersteinDRO.Gelbrich.projection_ambiguity_setoperations-researchprobabilityrobust-optimization
If has mean vector and covariance matrix , the image of the type-2 Wasserstein ambiguity set under is contained in ; the inclusion becomes an equality when and is an elliptical distribution.
Preamble
import Mathlib import Definitions.Def_WassersteinDRO_Gelbrich_ambiguitySet import Definitions.Def_WassersteinDRO_Gelbrich_meanVector import Definitions.Def_WassersteinDRO_Gelbrich_covarianceMatrix import Definitions.Def_WassersteinDRO_Gelbrich_meanCovarianceUncertaintySet import Definitions.Def_WassersteinDRO_Gelbrich_IsElliptical open MeasureTheory
Formal statement
namespace WassersteinDRO.Gelbrich
/-- Proposition 1 (Projection of `B_{ε,2}(P̂N)` onto the mean-covariance space), Kuhn et al.
2019, p. 16-17: if `P̂N` has mean vector `μ̂` and covariance matrix `Ŝ`, then the image of the
Wasserstein ambiguity set under `Q ↦ (E_Q[ξ], Cov_Q[ξ])` is contained in `U_ε(μ̂,Ŝ)`; the
inclusion becomes an equality if `Ξ = R^m` and `P̂N` is an elliptical distribution
`E_g(μ̂,Ŝ)`. -/
theorem projection_ambiguity_set {m : ℕ} (ε : ℝ) (Ξ : Set (EuclideanSpace ℝ (Fin m)))
(PN : Measure (EuclideanSpace ℝ (Fin m))) (μhat : EuclideanSpace ℝ (Fin m))
(SigmaHat : Matrix (Fin m) (Fin m) ℝ)
(hμ : meanVector PN = μhat) (hS : covarianceMatrix PN = SigmaHat) :
(∀ Q ∈ ambiguitySet ε 2 Ξ PN,
(meanVector Q, covarianceMatrix Q) ∈ meanCovarianceUncertaintySet ε μhat SigmaHat) ∧
(Ξ = Set.univ → (∃ g : ℝ → ℝ, IsElliptical PN g μhat SigmaHat) →
∀ pair ∈ meanCovarianceUncertaintySet ε μhat SigmaHat,
∃ Q ∈ ambiguitySet ε 2 Ξ PN, (meanVector Q, covarianceMatrix Q) = pair) := by sorry
end WassersteinDRO.Gelbrich
Source
Kuhn, Mohajerin Esfahani, Nguyen & Shafieezadeh-Abadeh, Wasserstein Distributionally Robust Optimization, INFORMS TutORials 2019, Proposition 1, p. 16-17
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.