§3 after (3.17) — is a Kuhn–Tucker vector for iff is optimal for and (P) is normal
ProvedPenaltyLag.Exact.kt_vector_iffLet be a nonempty convex subset of a real vector space and convex functions on , defining the convex program (P). Let , let be the penalty Lagrangian (2.3), the objective of the dual problem , and let . Then is a Kuhn–Tucker vector for (P) relative to , that is,
if and only if is an optimal solution to and (P) is normal (the dual optimal value equals the optimal value of (P)).
Together with the fact that and have the same optimal solutions, this shows that the notion of Kuhn–Tucker vector does not depend on . It is the step that turns the hypotheses of Theorem 3.5 (normality and dual optimality) into a Kuhn–Tucker vector.
Formalization Note The paper states the standing assumption (convexity of and the , ) once on p. 358; here it is a hypothesis. "Optimal solution to " means with (the convention of p. 361). Infima and suprema are taken in the extended reals; indices are 0-based (Fin m).
import Mathlib import Definitions.Def_PenaltyLag_Exact_Basic
namespace PenaltyLag.Exact
/-- §3 after (3.17), pp. 361–362: for r > 0, ȳ is a Kuhn–Tucker vector relative to L_r
iff ȳ is an optimal solution to (D_r) and (P) is normal. -/
theorem kt_vector_iff {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) :
IsKTVector X f₀ f r ybar ↔ (IsDualOptimal X f₀ f r ybar ∧ IsNormal X f₀ f) := by sorry
end PenaltyLag.Exact
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.