Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Conjugacy theorem on (U, Y): a proper convex weakly lsc φ satisfies φ** = φ

Proved
StochConvexProg.Duality.fenchel_moreau_U

by tianyipeng · Nov 6, 2026 · Mathlib 0df444a (Lean v4.33.1)

convex-dualityfenchel-moreaustochastic-programmingweak-topology

Let σ\sigmaσ be a probability measure on a measurable space (S,S)(S,\mathcal S)(S,S) and consider the paired spaces

U=Rm1×Lm2∞,Y=Rm1×Lm21,⟨u,y⟩=u1⋅y1+∫Su2(s)⋅y2(s) σ(ds)U=\mathbb R^{m_1}\times\mathcal L^\infty_{m_2},\qquad Y=\mathbb R^{m_1}\times\mathcal L^1_{m_2},\qquad \langle u,y\rangle=u_1\cdot y_1+\int_S u_2(s)\cdot y_2(s)\,\sigma(ds)U=Rm1​×Lm2​∞​,Y=Rm1​×Lm2​1​,⟨u,y⟩=u1​⋅y1​+∫S​u2​(s)⋅y2​(s)σ(ds)

(the pairing (1.6) of the paper). For φ:U→[−∞,+∞]\varphi:U\to[-\infty,+\infty]φ:U→[−∞,+∞] let φ∗(y)=sup⁡u∈U{⟨u,y⟩−φ(u)}\varphi^*(y)=\sup_{u\in U}\{\langle u,y\rangle-\varphi(u)\}φ∗(y)=supu∈U​{⟨u,y⟩−φ(u)} and φ∗∗(u)=sup⁡y∈Y{⟨u,y⟩−φ∗(y)}\varphi^{**}(u)=\sup_{y\in Y}\{\langle u,y\rangle-\varphi^*(y)\}φ∗∗(u)=supy∈Y​{⟨u,y⟩−φ∗(y)} be its conjugate and biconjugate with respect to this pairing.

Theorem (conjugacy theorem for the pairing (1.6)). Suppose φ\varphiφ is

  1. proper: φ(u)>−∞\varphi(u)>-\inftyφ(u)>−∞ for every uuu and φ(u)<+∞\varphi(u)<+\inftyφ(u)<+∞ for some uuu;
  2. convex: its epigraph {(u,α)∈U×R:φ(u)≤α}\{(u,\alpha)\in U\times\mathbb R : \varphi(u)\le\alpha\}{(u,α)∈U×R:φ(u)≤α} is convex;
  3. lower semicontinuous with respect to the weak topology σ(U,Y)\sigma(U,Y)σ(U,Y) on UUU induced by the pairing.

Then

φ∗∗=φ.\varphi^{**}=\varphi .φ∗∗=φ.

This is the final step of the proof of Theorem 3 of Rockafellar–Wets: once the perturbation function φ\varphiφ is known to be a proper convex function that is lower semicontinuous "in a topology compatible with the pairing between UUU and YYY", one has φ∗∗=φ\varphi^{**}=\varphiφ∗∗=φ. It is the Fenchel–Moreau theorem specialised to this dual pair.

Formalization Note Values are in EReal; the weak topology is weakU, the infimum over y∈Yy\in Yy∈Y of the topologies induced by u↦⟨u,y⟩u\mapsto\langle u,y\rangleu↦⟨u,y⟩, passed explicitly since UUU also carries its norm topology. Conjugates are conjU, biconjU from the shared definition file.

Preamble
import Mathlib
import Definitions.Def_StochConvexProg_Duality_Problem
import Definitions.Def_StochConvexProg_Duality_Dual

open MeasureTheory
Formal statement
namespace StochConvexProg.Duality

/-- Rockafellar–Wets (1976), p. 191, last step of the proof of Theorem 3 (the conjugacy theorem of
Rockafellar, *Conjugate Duality and Optimization*, 1974, Theorem 5, for the pairing (1.6)): a proper
convex function `φ : U → [−∞, +∞]` that is lower semicontinuous for the weak topology induced on
`U = R^{m₁} × ℒ^∞_{m₂}` by `Y = R^{m₁} × ℒ¹_{m₂}` satisfies `φ** = φ`. -/
theorem fenchel_moreau_U {S : Type*} [MeasurableSpace S] {σ : Measure S}
    [IsProbabilityMeasure σ] {m₁ m₂ : ℕ} (φ : USpace σ m₁ m₂ → EReal)
    (hproper : EProper φ) (hconv : EConvex φ)
    (hlsc : @LowerSemicontinuous (USpace σ m₁ m₂) EReal (weakU σ m₁ m₂) _ φ) :
    biconjU φ = φ := by sorry

end StochConvexProg.Duality
Source
Rockafellar and Wets, Stochastic convex programming: basic duality, Pacific J. Math. 62(1) (1976), p. 191, end of the proof of Theorem 3 ("Inasmuch as φ is a proper convex function on U which is lower semicontinuous in a topology compatible with the pairing between U and Y, we have φ** = φ"), invoking the conjugacy theorem of Rockafellar, Conjugate Duality and Optimization, SIAM 1974 (ref. [8]), Theorem 5. https://msp.org/pjm/1976/62-1/p20.xhtml

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