Convexity of the finite restricted conjugate
ProvedVectorSpaceOpt.restrictedConvexConjugate_convexLet be a nonempty subset of a real normed space and let . Define to consist of those continuous linear functionals for which is bounded above. Then
No convexity hypothesis on is needed for this assertion: the conjugate is a supremum of affine functions of . The explicit boundedness predicate makes the statement faithful to the finite conjugate domain in §7.10. This milestone provides a reusable convex dual objective and its natural domain before any closedness or duality qualification is imposed.
import Definitions.Def_VectorSpaceOpt_restrictedConvexConjugate
namespace VectorSpaceOpt
/-- Luenberger, Chapter 7, §7.10, Proposition 1. -/
theorem restrictedConvexConjugate_convex
{X : Type*} [NormedAddCommGroup X] [NormedSpace ℝ X]
(C : Set X) (f : X → ℝ) (hC : C.Nonempty) :
Convex ℝ {ℓ : X →L[ℝ] ℝ |
BddAbove ((fun x : X ↦ ℓ x - f x) '' C)} ∧
ConvexOn ℝ
{ℓ : X →L[ℝ] ℝ |
BddAbove ((fun x : X ↦ ℓ x - f x) '' C)}
(restrictedConvexConjugate C f) := by sorry
end VectorSpaceOptRead-back
What the Lean code literally says, in plain math · gpt-5
For every real normed vector space , nonempty set , and arbitrary function , let be the set of continuous real-linear functionals for which is bounded above. The theorem asserts both that is a convex subset of the real vector space of continuous linear functionals and that the function is convex on . Because is nonempty and members of satisfy boundedness, the supremum on is the ordinary finite real supremum. No convexity assumption on or is present, no claim is made about convexity of the conjugate outside , and itself is not assumed nonempty.
Confirmed by the mission captain (proposal self-audit).