Proof of Theorem 4.4, p. 17 — for every feasible point of (13), Q = (1/N)ΣΣ α_ik δ_{ξ_ik} lies in B_ε(P̂_N)
ProvedWassDDRO.Extremal.discreteQ_mem_balldistributionally-robust-optimizationoptimal-transportp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1wasserstein
Let , let , and let be feasible for program (13): , , , and (with when ). Then the discrete distribution
belongs to the Wasserstein ball : it is a probability measure supported on with .
This is the first half of the second claim of Theorem 4.4: every sequence of feasible decisions of (13) yields distributions inside the ambiguity set.
Formalization Note The ball is the published ambiguitySet ε 1 Ξ around the published empiricalDistribution. When the Lean atom is and carries weight .
Preamble
import Mathlib import Definitions.Def_WassDDRO_Extremal_Setting
Formal statement
namespace WassDDRO.Extremal
/-- Proof of Theorem 4.4, second claim, p. 17: for any feasible point (α, q) of (13), the
discrete distribution Q = (1/N) Σᵢ Σ_k α_ik δ_{ξ̂ᵢ − q_ik/α_ik} lies in the Wasserstein ball
B_ε(P̂_N); the coupling (1/N) Σᵢ Σ_k α_ik δ_{(ξ_ik, ξ̂ᵢ)} has transport cost (1/N) Σᵢ Σ_k ‖q_ik‖ ≤ ε. -/
theorem discreteQ_mem_ball {E : Type*} [NormedAddCommGroup E] [NormedSpace ℝ E]
[FiniteDimensional ℝ E] [MeasurableSpace E] [BorelSpace E]
{K N : ℕ} (hN : 0 < N) (Ξ : Set E) (ξhat : Fin N → E) (hξ : ∀ i, ξhat i ∈ Ξ) (ε : ℝ)
(α : Fin N → Fin K → ℝ) (q : Fin N → Fin K → E) (h : Feasible13 ε Ξ ξhat α q) :
discreteQ ξhat α q ∈ WassersteinDRO.Duality.ambiguitySet ε 1 Ξ
(WassersteinDRO.Duality.empiricalDistribution ξhat) := by sorry
end WassDDRO.Extremal
Source
Mohajerin Esfahani & Kuhn, arXiv:1505.05116v3, proof of Theorem 4.4 (second claim), p. 17
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.