§2 (4) — Moore's theorem: a positive matrix is the kernel of one and only one Hilbert space of functions
ProvedAronszajnRK.Sum.moore_existence_uniquenessLet be a set and a positive matrix in the sense of E. H. Moore: for every , all points and all ,
Then:
- (existence) there is a complex Hilbert space of functions on with continuous point evaluations whose reproducing kernel is ;
- (uniqueness) if and are two such spaces with reproducing kernel , they consist of the same functions, and every function has the same norm in as in .
Hence determines its class of functions together with its quadratic form . This is what makes "the class with kernel " a well-defined object, which every other statement of the mission uses.
Formalization Note A positive matrix is (Matrix.of K).PosSemidef (Hermitian with non-negative finitely supported quadratic form , no finiteness of ). The two spaces in the uniqueness half are quantified over arbitrary universes; the existence witness lives in the universe of . Equality of norms gives equality of scalar products by polarization. Existence alone is Mathlib's RKHS.OfKernel construction (after converting the scalar kernel to an operator-valued one); uniqueness is not in Mathlib.
import Mathlib import Definitions.Def_AronszajnRK_Sum_kernelFn open scoped ComplexOrder
namespace AronszajnRK.Sum
universe u v w
/-- **Moore's theorem** (Aronszajn, *Theory of Reproducing Kernels*, Trans. Amer. Math. Soc. 68
(1950), §2 (4), p. 344 (PDF 8)): to every positive matrix `K(x, y)` there corresponds one and only
one class of functions with a uniquely determined quadratic form in it, forming a Hilbert space and
admitting `K(x, y)` as a reproducing kernel.
A positive matrix (§2 (3), p. 344) is `(Matrix.of K).PosSemidef`: `K` is Hermitian and
`∑ᵢⱼ K(yᵢ, yⱼ) ξ̄ᵢ ξⱼ ≥ 0` for every finite family. Existence: some complex RKHS on `X` has scalar
kernel `K`. Uniqueness: any two complex RKHSs on `X` with scalar kernel `K` contain the same
functions, and an element of either has the same norm (quadratic form) as the element of the other
with the same function. -/
theorem moore_existence_uniqueness {X : Type u} (K : X → X → ℂ)
(hK : (Matrix.of K).PosSemidef) :
(∃ (H : Type u) (_ : NormedAddCommGroup H) (_ : InnerProductSpace ℂ H) (_ : CompleteSpace H)
(_ : RKHS ℂ H X ℂ), kernelFn H = K) ∧
∀ (H₁ : Type v) [NormedAddCommGroup H₁] [InnerProductSpace ℂ H₁] [CompleteSpace H₁]
[RKHS ℂ H₁ X ℂ] (H₂ : Type w) [NormedAddCommGroup H₂] [InnerProductSpace ℂ H₂]
[CompleteSpace H₂] [RKHS ℂ H₂ X ℂ],
kernelFn H₁ = K → kernelFn H₂ = K →
Set.range (fun f : H₁ => (f : X → ℂ)) = Set.range (fun f : H₂ => (f : X → ℂ)) ∧
∀ (f₁ : H₁) (f₂ : H₂), (f₁ : X → ℂ) = (f₂ : X → ℂ) → ‖f₁‖ = ‖f₂‖ := by sorry
end AronszajnRK.Sum
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.