Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 3.16 — the dual representation (3.15) of distortion risk functionals

Proved
MultistageStochastic.distortion_dual_representation

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

convex-analysisoperations-researchprobabilityrisk-measures

Let σ\sigmaσ be a distortion function — nonnegative, nondecreasing on [0,1)[0,1)[0,1), integrating to 111 — on a probability space (Ω,F,P)(\Omega,\mathcal F,P)(Ω,F,P), and let Y∈L∞Y\in L^\inftyY∈L∞. Then

Rσ(Y)  =  sup⁡{E(Y⋅Z) ∣ E(Z)=1, and AV@Rα(Z)≤11−α∫α1σ(u) du for all α∈[0,1)},(3.15)\mathcal R_\sigma(Y)\;=\;\sup\Bigl\{\mathbb E(Y\cdot Z)\ \Big|\ \mathbb E(Z)=1,\ \text{and } \mathsf{AV@R}_\alpha(Z)\le\frac{1}{1-\alpha}\int_\alpha^1\sigma(u)\,du\text{ for all }\alpha\in[0,1)\Bigr\} , \tag{3.15}Rσ​(Y)=sup{E(Y⋅Z) ​ E(Z)=1, and AV@Rα​(Z)≤1−α1​∫α1​σ(u)du for all α∈[0,1)},(3.15)

the supremum over integrable ZZZ. The constraint set is the relation Z≼σZ\preccurlyeq\sigmaZ≼σ of (3.16), and the theorem says that the distortion functional is the support function of that set: Rσ(Y)=sup⁡Z≼σE(YZ)\mathcal R_\sigma(Y)=\sup_{Z\preccurlyeq\sigma}\mathbb E(YZ)Rσ​(Y)=supZ≼σ​E(YZ), the Fenchel–Moreau representation with conjugate Rσ∗\mathcal R_\sigma^*Rσ∗​ equal to 000 on {Z≼σ}\{Z\preccurlyeq\sigma\}{Z≼σ} and +∞+\infty+∞ elsewhere (Remark 3.17).

This is Theorem 3.16, the dual representation from which the source derives the dual characterisation of the Average Value-at-Risk (Corollary 3.18), the co-monotone maximum (Corollary 3.19) and, together with Kusuoka's theorem, the dual representation of every version independent risk functional (Corollary 3.21). The content is the identification of the conjugate: by the rearrangement (Chebyshev) inequality E(YZ)≤∫01GY−1GZ−1\mathbb E(YZ)\le\int_0^1G_Y^{-1}G_Z^{-1}E(YZ)≤∫01​GY−1​GZ−1​, so Rσ∗(Z)=sup⁡Y∫01GY−1(u) (GZ−1(u)−σ(u)) du\mathcal R_\sigma^*(Z)=\sup_Y\int_0^1G_Y^{-1}(u)\,(G_Z^{-1}(u)-\sigma(u))\,duRσ∗​(Z)=supY​∫01​GY−1​(u)(GZ−1​(u)−σ(u))du, and testing with indicator-type YYY shows this is 000 exactly when the upper-tail averages of ZZZ are dominated by those of σ\sigmaσ.

Formalization Note The source writes the constraint for α∈[0,1]\alpha\in[0,1]α∈[0,1]. At α=1\alpha=1α=1 the right-hand side is 10∫11σ\frac{1}{0}\int_1^1\sigma01​∫11​σ, to be read as the limit σ(1−)\sigma(1^-)σ(1−); since the Average Value-at-Risk is continuous and nondecreasing in its level, the α=1\alpha=1α=1 constraint is the limit of the α<1\alpha<1α<1 ones and is implied by them, so the formal constraint set quantifies over [0,1)[0,1)[0,1), where every term is finite and no division by zero occurs. The supremum is a real supremum over a nonempty (Z≡1Z\equiv 1Z≡1 is feasible, as the average of σ\sigmaσ over [α,1][\alpha,1][α,1] is at least 111) and bounded family (Z≥0Z\ge 0Z≥0 follows from the constraints, so E(YZ)≤∥Y∥∞\mathbb E(YZ)\le\|Y\|_\inftyE(YZ)≤∥Y∥∞​), so it is not a junk value. No atomless hypothesis is imposed, matching the printed statement; the identity holds on atomic spaces as well, as the two-point instances in the mission notes verify.

Preamble
import Definitions.Def_MultistageStochastic_RiskFunctional
import Definitions.Def_MultistageStochastic_Distortion
open MeasureTheory
open scoped ENNReal
Formal statement
namespace MultistageStochastic
theorem distortion_dual_representation {Ω : Type*} [MeasurableSpace Ω] (P : Measure Ω)
    [IsProbabilityMeasure P] (σ : ℝ → ℝ) (hσ : IsDistortionFunction σ)
    (Y : Ω → ℝ) (hY : MemLinfty P Y) :
    distortionFunctional P σ Y
      = ⨆ Z : {Z : Ω → ℝ // DominatedByDistortion P σ Z},
          ∫ 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. 105-106 (PDF pp. 118-119), Theorem 3.16 (Dual Representation of Distortion Risk Functionals): "Let R_σ be a distortion risk functional. Then R_σ has the representation R_σ(Y) = sup{E(Y·Z) | E(Z) = 1, and AV@R_α(Z) ≤ (1/(1−α)) ∫_α^1 σ(u) du for all α ∈ [0, 1]}. (3.15)"
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