Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Proof of Theorem 3.5 — at a dual optimal yˉ\bar yyˉ​, a minimizer xˉ\bar xxˉ of Lr(⋅,yˉ)L_r(\cdot, \bar y)Lr​(⋅,yˉ​) has ∇yLr(xˉ,yˉ)=0\nabla_y L_r(\bar x, \bar y) = 0∇y​Lr​(xˉ,yˉ​)=0 and maximizes in yyy

Proved
PenaltyLag.Exact.stationary_at_dual_optimal

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

augmented-lagrangiandualityp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1

Let XXX be a nonempty convex subset of a real vector space EEE and f0,f1,…,fmf_0, f_1, \dots, f_mf0​,f1​,…,fm​ convex functions on XXX, defining the convex program (P). Let r>0r > 0r>0, let yˉ∈Rm\bar y \in \mathbb R^myˉ​∈Rm be an optimal solution to the dual problem (Dr)(D_r)(Dr​) (maximize gr(y)=inf⁡x∈XLr(x,y)g_r(y) = \inf_{x \in X} L_r(x, y)gr​(y)=infx∈X​Lr​(x,y) over Rm\mathbb R^mRm), and let xˉ∈X\bar x \in Xxˉ∈X minimize Lr(⋅,yˉ)L_r(\cdot, \bar y)Lr​(⋅,yˉ​) over XXX. Then the function y↦Lr(xˉ,y)y \mapsto L_r(\bar x, y)y↦Lr​(xˉ,y) is differentiable at yˉ\bar yyˉ​ with

∇yLr(xˉ,yˉ)=0,\nabla_y L_r(\bar x, \bar y) = 0,∇y​Lr​(xˉ,yˉ​)=0,

and yˉ\bar yyˉ​ maximizes Lr(xˉ,y)L_r(\bar x, y)Lr​(xˉ,y) over all y∈Rmy \in \mathbb R^my∈Rm.

In the paper this is the step of the proof of Theorem 3.5 that follows from Theorem 3.2: ∇yLr(xˉ,yˉ)=∇gr(yˉ)=0\nabla_y L_r(\bar x, \bar y) = \nabla g_r(\bar y) = 0∇y​Lr​(xˉ,yˉ​)=∇gr​(yˉ​)=0, and the concavity of Lr(xˉ,⋅)L_r(\bar x, \cdot)Lr​(xˉ,⋅) turns stationarity into maximality, so that (xˉ,yˉ)(\bar x, \bar y)(xˉ,yˉ​) is a saddle point.

Formalization Note The standing assumption of p. 358 is a hypothesis. "Optimal solution to (Dr)(D_r)(Dr​)" means gr(yˉ)=sup⁡ygr(y)>−∞g_r(\bar y) = \sup_y g_r(y) > -\inftygr​(yˉ​)=supy​gr​(y)>−∞. The gradient statement is HasGradientAt … 0 ȳ in EuclideanSpace ℝ (Fin m), which asserts differentiability as well as the value of the gradient. No sign restriction is placed on yˉ\bar yyˉ​.

Preamble
import Mathlib
import Definitions.Def_PenaltyLag_Exact_Basic
Formal statement
namespace PenaltyLag.Exact

/-- Proof of Theorem 3.5, pp. 362–363: if ȳ is an optimal solution to (D_r), r > 0, and
x̄ ∈ X minimizes L_r(·, ȳ) over X, then ∇_y L_r(x̄, ȳ) = 0 and ȳ maximizes L_r(x̄, ·) over ℝ^m. -/
theorem stationary_at_dual_optimal {E : Type*} [AddCommGroup E] [Module ℝ E] {m : ℕ}
    (X : Set E) (hX : Convex ℝ X) (hXne : X.Nonempty)
    (f₀ : E → ℝ) (f : Fin m → E → ℝ) (hf₀ : ConvexOn ℝ X f₀) (hf : ∀ i, ConvexOn ℝ X (f i))
    (r : ℝ) (hr : 0 < r) (ybar : PenaltyLag.Asymptotic.Mult m) (hybar : IsDualOptimal X f₀ f r ybar)
    (xbar : E) (hxbar : xbar ∈ X) (hmin : ∀ x ∈ X, PenaltyLag.Asymptotic.Lr f₀ f r xbar ybar ≤ PenaltyLag.Asymptotic.Lr f₀ f r x ybar) :
    HasGradientAt (fun y => PenaltyLag.Asymptotic.Lr f₀ f r xbar y) 0 ybar ∧ ∀ y, PenaltyLag.Asymptotic.Lr f₀ f r xbar y ≤ PenaltyLag.Asymptotic.Lr f₀ f r xbar ybar := by sorry

end PenaltyLag.Exact
Source
Rockafellar, A Dual Approach to Solving Nonlinear Programming Problems by Unconstrained Optimization, Math. Programming 5 (1973), proof of Theorem 3.5, pp. 362–363
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