Generalized Kuhn–Tucker theorem
ProvedVectorSpaceOpt.generalized_kuhn_tuckerLet and be normed real spaces, let be a convex cone in with nonempty interior, and let and have continuous-linear Gâteaux derivatives and at . Assume is a local minimizer of subject to , and assume Luenberger regularity: is feasible and some linearized direction is strictly feasible. Then a cone-positive continuous functional exists with
This capstone is the coordinate-free KKT necessary condition and reuses the cone order and dual positivity published by the earlier global-duality mission.
import Definitions.Def_VectorSpaceOpt_cone_optimization open Set
namespace VectorSpaceOpt
/-- Luenberger, Chapter 9, §9.4, Theorem 1. -/
theorem generalized_kuhn_tucker
{X Z : Type*}
[NormedAddCommGroup X] [NormedSpace ℝ X]
[NormedAddCommGroup Z] [NormedSpace ℝ Z]
(P : ConvexCone ℝ Z) (f : X → ℝ) (G : X → Z) (x₀ : X)
(f' : X →L[ℝ] ℝ) (G' : X →L[ℝ] Z)
(hPint : (interior (P : Set Z)).Nonempty)
(hf : HasGateauxDerivAt f f' x₀) (hG : HasGateauxDerivAt G G' x₀)
(hmin : IsLocalMinOn f {x | coneLE P (G x) 0} x₀)
(hregular : IsConeRegularAt P G G' x₀) :
∃ z₀ : Z →L[ℝ] ℝ, dualPositive P z₀ ∧
f' + z₀.comp G' = 0 ∧ z₀ (G x₀) = 0 := by
sorry
end VectorSpaceOptRead-back
What the Lean code literally says, in plain math · gpt-5
Let be real normed spaces, not assumed complete; let be a convex cone in with nonempty topological interior; let , , ; and let , be continuous real-linear maps. Assume that in every direction , the curves and have derivatives and at . Assume is a local minimum of relative to the set of satisfying , where this means . Also assume cone regularity at : and there exists with . Then there exists a continuous real-linear functional that is nonnegative on every element of , such that is identically zero and . No nonzeroness or uniqueness of is asserted.
Confirmed by the mission captain (proposal self-audit).