Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 1, p. 909 — π¹(r⁰, r) of (21) is a tight upper bound: sup over 𝔽₁ of E_ℙ((r⁰ + r′ζ̃)⁺) = π¹(r⁰, r)

Proved
TractableDRO.MeanSupport.theorem_1

by mikedeng1 · Oct 7, 2026 · Mathlib 0df444a (Lean v4.33.1)

distributionally-robust-optimizationdualitymoment-boundsp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1

This is Theorem 1 of Goh and Sim (2010), the mean-and-support bound on the expected positive part of an affine function of a random vector.

Let V,V^⊆Rn\mathcal V, \widehat{\mathcal V} \subseteq \mathbb R^nV,V⊆Rn be closed convex sets such that V^\widehat{\mathcal V}V meets the interior of V\mathcal VV. Let F1\mathbb F_1F1​ be the family of all probability distributions P\mathbb PP of a random vector ζ~∈Rn\tilde\zeta \in \mathbb R^nζ~​∈Rn such that ζ~\tilde\zetaζ~​ has support in V\mathcal VV and its mean has support in V^\widehat{\mathcal V}V:

F1={P:ζ^=EP(ζ~)∈V^, P(ζ~∈V)=1}.\mathbb F_1 = \{\mathbb P : \hat\zeta = \mathbb E_{\mathbb P}(\tilde\zeta) \in \widehat{\mathcal V},\ \mathbb P(\tilde\zeta \in \mathcal V) = 1\}.F1​={P:ζ^​=EP​(ζ~​)∈V, P(ζ~​∈V)=1}.

Then for every r0∈Rr^0 \in \mathbb Rr0∈R and r∈Rnr \in \mathbb R^nr∈Rn,

sup⁡P∈F1EP((r0+r′ζ~)+)=π1(r0,r)=inf⁡s∈Rn(sup⁡ζ^∈V^{s′ζ^}+sup⁡ζ∈Vmax⁡{r0+r′ζ−s′ζ, −s′ζ}),\sup_{\mathbb P \in \mathbb F_1} \mathbb E_{\mathbb P}\big((r^0 + r'\tilde\zeta)^+\big) = \pi^1(r^0, r) = \inf_{s \in \mathbb R^{n}} \Big( \sup_{\hat\zeta \in \widehat{\mathcal V}} \{s'\hat\zeta\} + \sup_{\zeta \in \mathcal V} \max\{r^0 + r'\zeta - s'\zeta,\ -s'\zeta\} \Big),P∈F1​sup​EP​((r0+r′ζ~​)+)=π1(r0,r)=s∈Rninf​(ζ^​∈Vsup​{s′ζ^​}+ζ∈Vsup​max{r0+r′ζ−s′ζ, −s′ζ}),

with both sides in the extended reals [−∞,+∞][-\infty, +\infty][−∞,+∞].

The theorem replaces an optimization over an infinite-dimensional family of distributions by a finite-dimensional convex program over the given sets. In the paper it bounds the expected positive part of a segregated linear decision rule and feeds the deflected linear decision rules of §6.

Formalization Note Distributions are probability measures on Fin n → ℝ with integrable coordinates; the mean is MomentDRO.Conf.meanVec; P(ζ~∈V)=1\mathbb P(\tilde\zeta\in\mathcal V)=1P(ζ~​∈V)=1 is an almost-everywhere statement. Both sides are EReal, with the paper's convention that an infeasible maximization (minimization) has value −∞-\infty−∞ (+∞+\infty+∞); the paper's "inf" in (21) is ⨅. The sets are closed and convex (the paper's tractable conic representable sets). The constraint qualification V^∩int⁡V≠∅\widehat{\mathcal V} \cap \operatorname{int}\mathcal V \ne \emptysetV∩intV=∅ is added: as printed the theorem is false when F1=∅\mathbb F_1 = \emptysetF1​=∅ (for V={(a,b):a>0, b≥1/a}\mathcal V = \{(a,b) : a > 0,\ b \ge 1/a\}V={(a,b):a>0, b≥1/a}, V^={(a,0):a∈R}\widehat{\mathcal V} = \{(a, 0) : a \in \mathbb R\}V={(a,0):a∈R} the left side is −∞-\infty−∞ and π1≥0\pi^1 \ge 0π1≥0), and the condition matches the paper's full-dimensionality assumptions (pp. 905–906). Under it both suprema in (21) are over nonempty sets, so the EReal sum ⊥+⊤\bot + \top⊥+⊤ never arises.


Retired 2026-10-27: a Lean-vs-paper fidelity audit found this statement weaker than, or narrower than, the cited result. A corrected statement will be resubmitted.

Preamble
import Mathlib
import Definitions.Def_TractableDRO_MeanSupport_Model

open MeasureTheory
Formal statement
namespace TractableDRO.MeanSupport

theorem theorem_1 {n : ℕ} (V Vhat : Set (Fin n → ℝ))
    (hVconv : Convex ℝ V) (hVclosed : IsClosed V)
    (hVhconv : Convex ℝ Vhat) (hVhclosed : IsClosed Vhat)
    (hCQ : (Vhat ∩ interior V).Nonempty) (r0 : ℝ) (r : Fin n → ℝ) :
    worstCase (family1 V Vhat) r0 r = pi1 V Vhat r0 r := by sorry

end TractableDRO.MeanSupport
Source
Goh & Sim, Distributionally Robust Optimization and Its Tractable Approximations, Oper. Res. 58(4), 2010, p. 909, Theorem 1, (21)
Human review
  • Endorsed by Shuze Chen · Oct 7, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 7, 2026

    Confirmed by the mission captain (proposal self-audit).

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me