Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Proof of Theorem 4.4, p. 17 — for Q = (1/N)ΣΣ α_ik δ_{ξ_ik}, E^Q[ℓ(ξ)] = (1/N)ΣΣ α_ik ℓ(ξ_ik) ≥ the objective of (13) at a feasible point

Proved
WassDDRO.Extremal.objective13_le_expect

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

distributionally-robust-optimizationp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1worst-case-distribution

Let N≥1N \ge 1N≥1, let ℓ1,…,ℓK:E→R‾\ell_1,\dots,\ell_K : E \to \overline{\mathbb R}ℓ1​,…,ℓK​:E→R be measurable and never +∞+\infty+∞, and let ℓ=max⁡kℓk\ell = \max_k \ell_kℓ=maxk​ℓk​. For every feasible point (αik,qik)(\alpha_{ik}, q_{ik})(αik​,qik​) of program (13), with ξik=ξ^i−qik/αik\xi_{ik} = \hat\xi_i - q_{ik}/\alpha_{ik}ξik​=ξ^​i​−qik​/αik​ and Q=1N∑i,kαikδξik\mathbb Q = \frac1N\sum_{i,k}\alpha_{ik}\delta_{\xi_{ik}}Q=N1​∑i,k​αik​δξik​​,

EQ[ℓ(ξ)] = 1N∑i=1N∑k=1Kαik ℓ(ξik)and1N∑i=1N∑k=1Kαik ℓk(ξik) ≤ EQ[ℓ(ξ)].\mathbb E^{\mathbb Q}[\ell(\xi)] \ =\ \frac1N\sum_{i=1}^N\sum_{k=1}^K \alpha_{ik}\,\ell(\xi_{ik}) \qquad\text{and}\qquad \frac1N\sum_{i=1}^N\sum_{k=1}^K \alpha_{ik}\,\ell_k(\xi_{ik}) \ \le\ \mathbb E^{\mathbb Q}[\ell(\xi)].EQ[ℓ(ξ)] = N1​i=1∑N​k=1∑K​αik​ℓ(ξik​)andN1​i=1∑N​k=1∑K​αik​ℓk​(ξik​) ≤ EQ[ℓ(ξ)].

The equality is the expectation of ℓ\ellℓ under a discrete distribution (the middle equality of the paper's chain on p. 17), and the inequality follows the paper's estimate ℓ≥ℓk\ell \ge \ell_kℓ≥ℓk​. Together with the previous milestone it gives the lower bound sup⁡Q∈Bε(P^N)EQ[ℓ(ξ)]≥lim sup⁡rEQr[ℓ(ξ)]\sup_{\mathbb Q \in \mathbb B_\varepsilon(\widehat{\mathbb P}_N)} \mathbb E^{\mathbb Q}[\ell(\xi)] \ge \limsup_r \mathbb E^{\mathbb Q_r}[\ell(\xi)]supQ∈Bε​(PN​)​EQ[ℓ(ξ)]≥limsupr​EQr​[ℓ(ξ)] chain in the proof of Theorem 4.4.

Formalization Note The expectation is the extended expectation of p. 5 (the positive and negative parts are integrated separately, with ∞−∞=∞\infty - \infty = \infty∞−∞=∞). The ikikik-th term on either side is 000 when αik=0\alpha_{ik} = 0αik​=0 (the convention of p. 14).

Preamble
import Mathlib
import Definitions.Def_WassDDRO_Extremal_Setting
Formal statement
namespace WassDDRO.Extremal

/-- Proof of Theorem 4.4, p. 17: for any feasible point (α, q) of (13), the expectation of
ℓ = max_k ℓ_k under Q = (1/N) Σᵢ Σ_k α_ik δ_{ξ_ik} equals (1/N) Σᵢ Σ_k α_ik ℓ(ξ_ik) (the term
being 0 when α_ik = 0), and by ℓ ≥ ℓ_k this dominates the objective (1/N) Σᵢ Σ_k α_ik ℓ_k(ξ_ik)
of (13). -/
theorem objective13_le_expect {E : Type*} [NormedAddCommGroup E] [NormedSpace ℝ E]
    [FiniteDimensional ℝ E] [MeasurableSpace E] [BorelSpace E]
    {K N : ℕ} (hN : 0 < N) (Ξ : Set E) (ℓ : Fin K → E → EReal)
    (hmeas : ∀ k, Measurable (ℓ k)) (hne_top : ∀ k ξ, ℓ k ξ ≠ ⊤)
    (ξhat : Fin N → E) (ε : ℝ)
    (α : Fin N → Fin K → ℝ) (q : Fin N → Fin K → E) (h : Feasible13 ε Ξ ξhat α q) :
    DupacovaWets.Consistency.expect (discreteQ ξhat α q) (maxLoss ℓ) =
        ((1 / (N : ℝ) : ℝ) : EReal) * ∑ i, ∑ k, term13 (maxLoss ℓ) (ξhat i) (α i k) (q i k) ∧
      objective13 ξhat ℓ α q ≤
        DupacovaWets.Consistency.expect (discreteQ ξhat α q) (maxLoss ℓ) := by sorry

end WassDDRO.Extremal
Source
Mohajerin Esfahani & Kuhn, arXiv:1505.05116v3, proof of Theorem 4.4 (second claim), p. 17
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