VectorSpaceOpt.dualPositive
DefinitionVectorSpaceOpt_dualPositivecontinuous-dualconvex-conedefinitionlagrange-multiplier
Let be a real normed space, a convex cone, and a continuous linear functional. The functional is dual-positive when
This is membership in the positive dual cone described by Luenberger, but the predicate avoids imposing a separate cone structure on the continuous dual. It applies to cones that are not closed or pointed and therefore matches the hypotheses of the main Lagrange-duality theorem. The definition is the common multiplier-feasibility interface for cone separation, saddle-point sufficiency, sensitivity bounds, weak duality, and dual attainment.
Definition code
import Mathlib
namespace VectorSpaceOpt
/-- A continuous linear functional that is nonnegative on a convex cone. -/
def dualPositive
{Z : Type*} [NormedAddCommGroup Z] [NormedSpace ℝ Z]
(P : ConvexCone ℝ Z) (zstar : Z →L[ℝ] ℝ) : Prop :=
∀ p ∈ P, 0 ≤ zstar p
end VectorSpaceOptSource
David G. Luenberger, Optimization by Vector Space Methods, Wiley, 1969, public scan https://sites.science.oregonstate.edu/~show/old/142_Luenberger.pdf, Chapter 8, §8.2, definition of a functional positive with respect to P, printed pp. 214–215; physical PDF pp. 234–235.