Proof of Theorem 5, p. 40 โ the zฬ-marginal of every lifted distribution lies in ๐ซโฒ
ProvedDRConvexOpt.Lifting.marginal_subset_momentSetambiguity-setdistributionally-robust-optimizationmarginal-distributionp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1
Let be measurable, , a proper cone, and , , measurable sets with bounds . Then
where is the lifted ambiguity set of distributions of with , almost surely and , and is the set of distributions of with and the same probability bounds.
This is the reverse inclusion of the Lifting Theorem: the auxiliary vector can always be projected out without violating the conic moment bound.
Formalization Note The lifted set requires to be integrable (an addition to the page; see the goal theorem). Expectations and probabilities of functions of under and under its marginal coincide because the marginal is the pushforward by the measurable first projection.
Preamble
import Mathlib import Definitions.Def_DRConvexOpt_Lifting_Setting
Formal statement
namespace DRConvexOpt.Lifting
open MeasureTheory
/-- Proof of Theorem 5, p. 40: ฮ _zฬ๐ซ โ ๐ซโฒ. -/
theorem marginal_subset_momentSet {nP nM nI : โ} (g : (Fin nP โ โ) โ (Fin nM โ โ))
(hg : Measurable g) (f : Fin nM โ โ) (K : Set (Fin nM โ โ)) (hK : DRConvexOpt.Reform.IsProperCone K)
(C : Fin nI โ Set (Fin nP โ โ)) (hC : โ i, MeasurableSet (C i)) (plo phi : Fin nI โ โ) :
marginal (liftedSet g f K C plo phi) โ momentSet g f K C plo phi := by sorry
end DRConvexOpt.Lifting
Source
Wiesemann, Kuhn & Sim, Distributionally Robust Convex Optimization, Optimization Online preprint 3757 (version of September 22, 2013), p. 40, proof of Theorem 5, second inclusion ๐ซโฒ โ ฮ _zฬ๐ซ
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.