Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Corollary 3.18 — the dual representations (3.18) and (3.19) of the Average Value-at-Risk

Disproved
MultistageStochastic.avar_dual_representation

by naimengye · Sep 23, 2026 · Mathlib 0df444a (Lean v4.33.1)

convex-analysisoperations-researchprobabilityrisk-measures

Let Y∈L∞Y\in L^\inftyY∈L∞ on a probability space and α∈[0,1]\alpha\in[0,1]α∈[0,1]. For α<1\alpha<1α<1,

AV@Rα(Y)  =  sup⁡{E(YZ) ∣ E(Z)=1, AV@Rp(Z)≤11−α for all p∈[α,1]}(3.18)\mathsf{AV@R}_\alpha(Y)\;=\;\sup\Bigl\{\mathbb E(YZ)\ \Big|\ \mathbb E(Z)=1,\ \mathsf{AV@R}_p(Z)\le\tfrac{1}{1-\alpha}\text{ for all }p\in[\alpha,1]\Bigr\} \tag{3.18}AV@Rα​(Y)=sup{E(YZ) ​ E(Z)=1, AV@Rp​(Z)≤1−α1​ for all p∈[α,1]}(3.18)

and

AV@Rα(Y)  =  sup⁡{E(YZ) ∣ EZ=1, 0≤Z, Z≤11−α},(3.19)\mathsf{AV@R}_\alpha(Y)\;=\;\sup\Bigl\{\mathbb E(YZ)\ \Big|\ \mathbb E Z=1,\ 0\le Z,\ Z\le\tfrac{1}{1-\alpha}\Bigr\} , \tag{3.19}AV@Rα​(Y)=sup{E(YZ) ​ EZ=1, 0≤Z, Z≤1−α1​},(3.19)

the suprema over integrable ZZZ; and for α=1\alpha=1α=1 the second representation reads ess sup⁡Y=sup⁡{E(YZ)∣EZ=1, 0≤Z}\operatorname{ess\,sup}Y=\sup\{\mathbb E(YZ)\mid\mathbb EZ=1,\ 0\le Z\}esssupY=sup{E(YZ)∣EZ=1, 0≤Z}.

This is Corollary 3.18, the specialisation of the dual representation of distortion risk functionals to σα=(1−α)−11[α,1)\sigma_\alpha=(1-\alpha)^{-1}\mathbf 1_{[\alpha,1)}σα​=(1−α)−11[α,1)​. The point the source makes is that the constraints of (3.15) collapse: only the levels in [α,1][\alpha,1][α,1] matter, and they are equivalent to the pointwise bound 0≤Z≤1/(1−α)0\le Z\le 1/(1-\alpha)0≤Z≤1/(1−α), which recovers formula (3.4).

Formalization Note In (3.18) the constraint is stated for p∈[α,1]p\in[\alpha,1]p∈[α,1] including p=1p=1p=1, where the Average Value-at-Risk is the essential supremum, exactly as printed. In (3.19) the source writes α≤1\alpha\le 1α≤1 with 1/(1−α)=+∞1/(1-\alpha)=+\infty1/(1−α)=+∞ at α=1\alpha=1α=1; that case is stated separately with the upper bound dropped, since 1/(1−1)1/(1-1)1/(1−1) is 000 in Lean and would make the constraint set empty. The suprema are over the subtype of integrable ZZZ satisfying the constraints; each family is nonempty (Z≡1Z\equiv 1Z≡1) and bounded (∣E(YZ)∣≤∥Y∥∞|\mathbb E(YZ)|\le\|Y\|_\infty∣E(YZ)∣≤∥Y∥∞​ as Z≥0Z\ge 0Z≥0, EZ=1\mathbb EZ=1EZ=1), so no junk value of a real supremum is involved. Pointwise constraints on ZZZ hold almost surely.


Retired. Retired because the printed (3.18) of Pflug–Pichler omits the constraint Z ≥ 0 that the general representation (3.16) imposes at the levels p < α, and is false as printed (accepted disproof: Z negative on a small-probability set); replaced by MultistageStochastic.avar_dual_representation_v2 with the corrected (3.18).

Preamble
import Definitions.Def_MultistageStochastic_RiskFunctional
import Definitions.Def_MultistageStochastic_Distortion
open MeasureTheory
open scoped ENNReal
Formal statement
namespace MultistageStochastic
theorem avar_dual_representation {Ω : Type*} [MeasurableSpace Ω] (P : Measure Ω)
    [IsProbabilityMeasure P] (Y : Ω → ℝ) (hY : MemLinfty P Y) (α : ℝ) (hα : 0 ≤ α) :
    (α < 1 →
      averageValueAtRisk P Y α
        = ⨆ Z : {Z : Ω → ℝ // Integrable Z P ∧ ∫ ω, Z ω ∂P = 1 ∧
            ∀ p : ℝ, α ≤ p → p ≤ 1 → averageValueAtRisk P Z p ≤ (1 - α)⁻¹},
            ∫ ω, Y ω * (Z : Ω → ℝ) ω ∂P) ∧
    (α < 1 →
      averageValueAtRisk P Y α
        = ⨆ Z : {Z : Ω → ℝ // Integrable Z P ∧ ∫ ω, Z ω ∂P = 1 ∧
            ∀ᵐ ω ∂P, 0 ≤ Z ω ∧ Z ω ≤ (1 - α)⁻¹},
            ∫ ω, Y ω * (Z : Ω → ℝ) ω ∂P) ∧
    (α = 1 →
      averageValueAtRisk P Y α
        = ⨆ Z : {Z : Ω → ℝ // Integrable Z P ∧ ∫ ω, Z ω ∂P = 1 ∧ ∀ᵐ ω ∂P, 0 ≤ Z ω},
            ∫ ω, Y ω * (Z : Ω → ℝ) ω ∂P) := by sorry
end MultistageStochastic
Source
Georg Ch. Pflug and Alois Pichler, Multistage Stochastic Optimization, Springer 2014, https://doi.org/10.1007/978-3-319-08843-3 — Section 3.3.2, printed pp. 107-108 (PDF pp. 120-121), Corollary 3.18: "The Average Value-at-Risk at level α has the equivalent representations AV@R_α(Y) = sup{E(YZ) | E(Z) = 1, AV@R_p(Z) ≤ 1/(1−α) for all p ∈ [α, 1]} (α < 1) (3.18) and AV@R_α(Y) = sup{E(YZ) | EZ = 1, 0 ≤ Z, Z ≤ 1/(1−α)} (α ≤ 1). (3.19)"
Human review
  • Endorsed by Shuze Chen · Sep 25, 2026

    Confirmed by the moderator at approval.

  • Endorsed by naimengye · Sep 25, 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