Proof of Theorem 4.2, p. 13 — under Assumption 4.1, (12a) is an equality for every ε > 0
ProvedWassDDRO.Reduction.worstCase_eq_program12c_of_posdistributionally-robust-optimizationmoment-problemp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1strong-dualitywasserstein
Let , , and the samples be as in Theorem 4.2 (, each measurable), and suppose Assumption 4.1 holds: is convex and closed, each is proper, convex and lower semicontinuous, and no is identically on . Then for every
that is, the max-min inequality (12a) holds with equality and (10) equals the optimal value of (12c).
This is the strong-duality step of the proof, an instance of strong duality for moment problems.
Preamble
import Mathlib import Definitions.Def_WassDDRO_Reduction_Setting open MeasureTheory Filter Topology
Formal statement
namespace WassDDRO.Reduction
/-- Proof of Theorem 4.2, p. 13: under Assumption 4.1 the inequality (12a) is an equality
for any `ε > 0`, i.e. (10) equals the optimal value of (12c). -/
theorem worstCase_eq_program12c_of_pos {E : Type*} [NormedAddCommGroup E] [NormedSpace ℝ E] [FiniteDimensional ℝ E]
[MeasurableSpace E] [BorelSpace E]
{K N : ℕ} (hK : 0 < K) (hN : 0 < N) (Ξ : Set E) (ℓ : Fin K → E → EReal)
(hmeas : ∀ k, Measurable (ℓ k)) (ξhat : Fin N → E) (hξ : ∀ i, ξhat i ∈ Ξ)
(ε : ℝ) (hεpos : 0 < ε) (hA : Assumption41 Ξ ℓ) :
worstCaseExpectation ε Ξ ξhat (maxLoss ℓ) = program12cValue ε Ξ ξhat (maxLoss ℓ) := by sorry
end WassDDRO.Reduction
Source
Mohajerin Esfahani & Kuhn, arXiv:1505.05116v3, proof of Theorem 4.2, p. 13 (equality in (12a) for ε > 0)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.