Lemma 6, p. 33 — uniform positive bounds on E[ℓ′²]
ProvedOTDRO.StrongCvx.ell_deriv_sq_boundsdistributionally-robust-optimizationoptimal-transportp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1strong-convexity
Let be a probability law, nonempty, convex and compact, and let the loss satisfy Assumptions 2–3. There are positive constants and such that, for every ,
The bounds provide the constants that define the optimizer region and the strong-convexity radius.
Formalization Note Integrability of the squared derivative is stated along with its bounds, and nonemptiness of makes its radius and the uniform bound nonvacuous.
Preamble
import Mathlib import Definitions.Def_OTDRO_StrongCvx_Assumptions
Formal statement
namespace OTDRO.StrongCvx
open MeasureTheory
/-- Lemma 6, p. 33: the squared derivative has positive uniform lower
and finite upper expectation bounds over B. -/
theorem ell_deriv_sq_bounds {d : ℕ}
(P0 : Measure (EuclideanSpace ℝ (Fin d))) [IsProbabilityMeasure P0]
(ℓ : ℝ → ℝ) (h2 : OTDRO.Dual.Assumption2 P0 ℓ)
(B : Set (EuclideanSpace ℝ (Fin d))) (hB4 : Assumption4 B)
(hBne : B.Nonempty) (M : ℝ) (h3 : Assumption3 P0 ℓ B M) :
∃ Llow Lbar : ℝ, DerivSqBounds P0 ℓ B Llow Lbar := by sorry
end OTDRO.StrongCvx
Source
Blanchet, Murthy & Zhang, arXiv:1810.02403v3, Lemma 6, p. 33
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.