Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Corollary 3.4 (r>0r > 0r>0) — LrL_rLr​ and L0L_0L0​ have the same Kuhn–Tucker vectors and saddle points; saddle points are the Kuhn–Tucker pairs

Proved
PenaltyLag.Exact.corollary_3_4

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

augmented-lagrangiandualitykuhn-tuckerp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1saddle-point

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, xˉ∈E\bar x \in Exˉ∈E and yˉ∈Rm\bar y \in \mathbb R^myˉ​∈Rm. Then:

  1. yˉ\bar yyˉ​ is a Kuhn–Tucker vector relative to LrL_rLr​ if and only if it is a Kuhn–Tucker vector relative to the ordinary Lagrangian L0L_0L0​;
  2. (xˉ,yˉ)(\bar x, \bar y)(xˉ,yˉ​) is a saddle point of LrL_rLr​ on X×RmX \times \mathbb R^mX×Rm if and only if it is a saddle point of L0L_0L0​ on X×RmX \times \mathbb R^mX×Rm;
  3. (xˉ,yˉ)(\bar x, \bar y)(xˉ,yˉ​) is a saddle point of LrL_rLr​ if and only if the ordinary Kuhn–Tucker conditions hold:
(i)  yˉi≥0,  fi(xˉ)≤0,  yˉifi(xˉ)=0  (i=1,…,m);(ii)  xˉ∈X minimizes f0+∑i=1myˉifi over X.\text{(i)}\ \ \bar y_i \ge 0,\ \ f_i(\bar x) \le 0,\ \ \bar y_i f_i(\bar x) = 0 \ \ (i = 1, \dots, m); \qquad \text{(ii)}\ \ \bar x \in X \text{ minimizes } f_0 + \textstyle\sum_{i=1}^m \bar y_i f_i \text{ over } X.(i)  yˉ​i​≥0,  fi​(xˉ)≤0,  yˉ​i​fi​(xˉ)=0  (i=1,…,m);(ii)  xˉ∈X minimizes f0​+∑i=1m​yˉ​i​fi​ over X.

Comparing every r>0r > 0r>0 with r=0r = 0r=0 gives the paper's "relative to the Lagrangians LrL_rLr​, r≥0r \ge 0r≥0, one has the same Kuhn–Tucker vectors and saddle points"; the case r=0r = 0r=0 of item 3 is a separate statement of this mission. The corollary shows what LrL_rLr​, r>0r > 0r>0, has in common with L0L_0L0​.

Formalization Note The standing assumption of p. 358 is a hypothesis. A Kuhn–Tucker vector relative to L0L_0L0​ is (3.17) with L0L_0L0​ in place of LrL_rLr​; since L0(x,y)=−∞L_0(x, y) = -\inftyL0​(x,y)=−∞ for y≱0y \not\ge 0y≥0, it is automatically nonnegative. Saddle points of L0L_0L0​ are compared in the extended reals.

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

/-- Corollary 3.4, p. 362 (r > 0 against r = 0): L_r and L₀ have the same Kuhn–Tucker
vectors and the same saddle points, and (x̄, ȳ) is a saddle point of L_r iff the ordinary
Kuhn–Tucker conditions hold. -/
theorem corollary_3_4 {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) (xbar : E) (ybar : PenaltyLag.Asymptotic.Mult m) :
    (IsKTVector X f₀ f r ybar ↔ IsKTVector0 X f₀ f ybar) ∧
    (IsSaddle X f₀ f r xbar ybar ↔ IsSaddle0 X f₀ f xbar ybar) ∧
    (IsSaddle X f₀ f r xbar ybar ↔ KuhnTuckerConditions X f₀ f xbar ybar) := by sorry

end PenaltyLag.Exact
Source
Rockafellar, A Dual Approach to Solving Nonlinear Programming Problems by Unconstrained Optimization, Math. Programming 5 (1973), p. 362, Corollary 3.4 (case r > 0)
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