§5.7, Theorem 1 — Closed subspace as its double annihilator
ProvedVectorSpaceOpt.closed_subspace_eq_preannihilator_annihilatorannihilatorclosed-subspacefunctional-analysisseparation
Let be a real normed space and let be a closed real linear subspace. Consider all continuous linear functionals on that vanish on . A point is annihilated by every one of those functionals exactly when already belongs to :
This is the concrete preannihilator-of-annihilator identity, stated without introducing a separate dual-submodule definition. It shows that the continuous dual separates each point outside a closed subspace and provides a reusable closedness/separation principle for later duality statements and quotient-space characterizations.
Preamble
import Mathlib
Formal statement
namespace VectorSpaceOpt
theorem closed_subspace_eq_preannihilator_annihilator
{X : Type*} [NormedAddCommGroup X] [NormedSpace ℝ X]
(M : Submodule ℝ X) (hM : IsClosed (M : Set X)) :
{x : X | ∀ f : X →L[ℝ] ℝ, (∀ m ∈ M, f m = 0) → f x = 0} = (M : Set X) := by sorry
end VectorSpaceOptSource
David G. Luenberger, Optimization by Vector Space Methods, §5.7, Theorem 1, book p. 118 (PDF physical p. 138), https://sites.science.oregonstate.edu/~show/old/142_Luenberger.pdf