A local minimum of a convex function is global
ProvedVectorSpaceOpt.convex_local_min_globalLet be a convex subset of a real normed space and let be convex, represented by ConvexOn ℝ C f. If is a local minimum relative to , then
The local-minimum hypothesis uses the subspace neighborhood filter, so no openness of is assumed and boundary minimizers are included. This is the first global consequence of convexity in Luenberger's chapter. As a mission milestone it establishes the local-to-global optimization interface that later lets dual certificates be interpreted as certificates for the entire feasible domain.
import Mathlib
namespace VectorSpaceOpt
/-- Luenberger, Chapter 7, §7.8, Proposition 1. -/
theorem convex_local_min_global
{X : Type*} [NormedAddCommGroup X] [NormedSpace ℝ X]
(C : Set X) (f : X → ℝ) (x₀ : X)
(hf : ConvexOn ℝ C f) (hx₀ : x₀ ∈ C)
(hlocal : IsLocalMinOn f C x₀) :
∀ x ∈ C, f x₀ ≤ f x := by sorry
end VectorSpaceOptRead-back
What the Lean code literally says, in plain math · gpt-5
For every real normed vector space , set , function , and point , if is convex on , , and is a local minimum of relative to , then for every . The membership hypothesis makes nonempty; no closedness, openness, finite-dimensionality, or attainment statement for any other point is assumed.
Confirmed by the mission captain (proposal self-audit).