§5.12, Theorem 2 — Supporting hyperplane
ProvedVectorSpaceOpt.supporting_hyperplaneLet be a real normed space and a convex set whose interior is nonempty. Fix a point . Then there is a nonzero continuous linear functional whose level through bounds all of :
The point may lie on the boundary or outside the closure; the statement does not add closedness or compactness. It converts geometric exclusion from the interior into a scalar support inequality and provides the one-set separation interface used in the capstone's distance certificate.
import Mathlib
namespace VectorSpaceOpt
theorem supporting_hyperplane
{X : Type*} [NormedAddCommGroup X] [NormedSpace ℝ X]
(K : Set X) (hK : Convex ℝ K) (hKint : (interior K).Nonempty)
(x : X) (hx : x ∉ interior K) :
∃ f : X →L[ℝ] ℝ, f ≠ 0 ∧ ∀ k ∈ K, f k ≤ f x := by sorry
end VectorSpaceOptRead-back
What the Lean code literally says, in plain math · gpt-5
Let be any real normed vector space, let be convex with nonempty interior, and let be any point outside the interior of . Then there exists a nonzero continuous real-linear functional such that for every . The point is not assumed to belong to or even to its closure, and is not assumed closed; no completeness, finite-dimensionality, or nontriviality assumption is present.
Confirmed by the mission captain (proposal self-audit).