Proof of Theorem 4.2, (12a)–(12c), p. 12 — the worst-case expectation is at most the value of (12c)
ProvedWassDDRO.Reduction.worstCase_le_program12cdistributionally-robust-optimizationp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1wassersteinweak-duality
Let be a finite-dimensional real normed space with its Borel σ-algebra, , and let () be measurable, with . Let () be samples with empirical distribution , and let . Then, without any convexity assumption,
This is the weak-duality half of the reduction: the right-hand side is (12c), the epigraph form of (12b), obtained from (10) by the max-min inequality (12a).
Formalization Note The expectation follows the paper's extended-arithmetic convention (); the infimum is when (12c) is infeasible. The sample condition and measurability of the are the paper's standing assumptions (§2, p. 5; p. 11).
Preamble
import Mathlib import Definitions.Def_WassDDRO_Reduction_Setting open MeasureTheory Filter Topology
Formal statement
namespace WassDDRO.Reduction
/-- Proof of Theorem 4.2, (12a)–(12c), p. 12: without Assumption 4.1, the worst-case
expectation (10) is at most the optimal value of (12c). -/
theorem worstCase_le_program12c {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) (hε : 0 ≤ ε) :
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, (12a), (12b), (12c), p. 12
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.