§5.4, Corollary 2 — Existence of a norming functional
ProvedVectorSpaceOpt.norming_functionaldualityfunctional-analysishahn-banachnorming-functional
Let be a nontrivial real normed space and fix . There exists a nonzero continuous linear functional attaining equality in the operator-norm estimate at x:
Such an f is aligned with x and is a norming functional for that vector. The explicit nontriviality hypothesis is needed for the source's nonzero conclusion when x = 0; it is automatic in the usual nonzero ambient spaces. The theorem supplies pointwise dual witnesses for norms and is the local certificate later applied to approximation residuals.
Preamble
import Mathlib
Formal statement
namespace VectorSpaceOpt
theorem norming_functional
{X : Type*} [NormedAddCommGroup X] [NormedSpace ℝ X] [Nontrivial X]
(x : X) :
∃ f : X →L[ℝ] ℝ, f ≠ 0 ∧ f x = ‖f‖ * ‖x‖ := by sorry
end VectorSpaceOptSource
David G. Luenberger, Optimization by Vector Space Methods, §5.4, Corollary 2, book pp. 112–113 (PDF physical pp. 132–133), https://sites.science.oregonstate.edu/~show/old/142_Luenberger.pdf