The Hilbert projection theorem
ProvedFamousTheorems.hilbert_projection_theoremThe Hilbert projection theorem. Let be a nonempty complete convex subset of a real inner product space (e.g. a closed convex subset of a Hilbert space). For every point there is minimising the distance to :
The minimiser is unique and is characterised by a variational inequality. This is the basis of orthogonal projections, the decomposition , the Riesz representation theorem, and least-squares approximation.
Formalization note. Mathlib's exists_norm_eq_iInf_of_complete_convex (existence only; uniqueness is a separate lemma).
Preamble
import Mathlib
Formal statement
namespace FamousTheorems
theorem hilbert_projection_theorem {F : Type*} [NormedAddCommGroup F] [InnerProductSpace ℝ F] {K : Set F} (hne : K.Nonempty)
(hc : IsComplete K) (hK : Convex ℝ K) (u : F) : ∃ v ∈ K, ‖u - v‖ = ⨅ w : K, ‖u - w‖ := by sorry
end FamousTheoremsSource
Listed in Mathlib's curated theorem manifests (docs/1000.yaml); formalized in Mathlib as `exists_norm_eq_iInf_of_complete_convex`. Proof here reduces to that Mathlib result.