Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Lemma 2 — c‾≤δ≤c‾\underline c \le \delta \le \overline cc​≤δ≤c if λ∗>0\lambda^* > 0λ∗>0; δ≥c‾=c‾\delta \ge \overline c = \underline cδ≥c=c​ if λ∗=0\lambda^* = 0λ∗=0

Proved
ModelRiskOT.WorstProb.lemma_2

by mikedeng1 · Oct 6, 2026 · Mathlib 0df444a (Lean v4.33.1)

optimal-transportp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1worst-case-probability

Let SSS be a Polish space, ccc a cost satisfying (A1), μ\muμ a probability measure on SSS, δ>0\delta > 0δ>0, and A⊆SA \subseteq SA⊆S a nonempty closed set. Let c‾\underline cc​ and c‾\overline cc be the quantities (14) defined in terms of λ∗\lambda^*λ∗.

  1. If λ∗∈(0,∞)\lambda^* \in (0,\infty)λ∗∈(0,∞) attains the infimum in (13), then
c‾≤δ≤c‾.\underline c \le \delta \le \overline c.c​≤δ≤c.
  1. If the infimum in (13) is attained at λ∗=0\lambda^* = 0λ∗=0, then
δ≥c‾=c‾.\delta \ge \overline c = \underline c.δ≥c=c​.

The lemma locates the budget δ\deltaδ between the transport costs of the two extreme ways of moving mass into AAA at level 1/λ∗1/\lambda^*1/λ∗: moving all points with c(x,A)<1/λ∗c(x,A) < 1/\lambda^*c(x,A)<1/λ∗, or all points with c(x,A)≤1/λ∗c(x,A) \le 1/\lambda^*c(x,A)≤1/λ∗. It is the reason why, when c‾=c‾\underline c = \overline cc​=c, the worst case is attained by moving exactly the inflated set {c(x,A)≤1/λ∗}\{c(x,A) \le 1/\lambda^*\}{c(x,A)≤1/λ∗}.

Formalization Note With the multiplied-form thresholds, at λ∗=0\lambda^* = 0λ∗=0 both sets in (14) are SSS, so c‾=c‾=∫c(x,A) dμ(x)\underline c = \overline c = \int c(x,A)\,d\mu(x)c​=c=∫c(x,A)dμ(x), as in the paper's proof. Values are in [0,∞][0,\infty][0,∞] and δ\deltaδ enters as a nonnegative extended real.

Preamble
import Mathlib
import Definitions.Def_ModelRiskOT_WorstProb_WorstCaseProb

open MeasureTheory
open scoped ENNReal
Formal statement
namespace ModelRiskOT.WorstProb

/-- Lemma 2, p. 10. Under (A1), for a nonempty closed `A`: if `λ* ∈ (0, ∞)` attains the infimum in
(13), then `c̲ ≤ δ ≤ c̄`; if the infimum in (13) is attained at `λ* = 0`, then `δ ≥ c̄ = c̲`
(`c̲`, `c̄` as in (14)). -/
theorem lemma_2 {S : Type*} [TopologicalSpace S] [PolishSpace S] [MeasurableSpace S]
    [BorelSpace S] (c : S → S → ℝ) (hc : ModelRiskOT.Duality.AssumptionA1 c) (μ : Measure S) [IsProbabilityMeasure μ]
    (δ : ℝ) (hδ : 0 < δ) (A : Set S) (hA : IsClosed A) (hAne : A.Nonempty)
    (lamStar : ℝ) (hlam : AttainsInf13 c μ δ A lamStar) :
    (0 < lamStar →
      cLower c μ A lamStar ≤ ENNReal.ofReal δ ∧ ENNReal.ofReal δ ≤ cUpper c μ A lamStar) ∧
    (lamStar = 0 →
      cUpper c μ A lamStar ≤ ENNReal.ofReal δ ∧ cUpper c μ A lamStar = cLower c μ A lamStar) := by sorry

end ModelRiskOT.WorstProb
Source
Blanchet & Murthy, Quantifying Distributional Model Risk via Optimal Transport, arXiv:1604.01446v2, p. 10, Lemma 2
Human review
  • Endorsed by Shuze Chen · Oct 6, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 6, 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