Theorem 3.16 — every separable Hilbert space is of strong negative type
ProvedDistCov.Hilbert.theorem_3_16Let be a separable real Hilbert space with its norm distance and its Borel -field. Then is of strong negative type:
- has negative type: whenever ;
- for Borel probability measures on with finite first moments,
Combined with Theorem 3.11 of the paper, this shows that zero distance covariance characterizes independence for random variables taking values in separable Hilbert spaces, which answers a question of Kosorok (2009).
Formalization Note. Hilbert spaces are real, as the paper assumes (Errata (viii), p. 27); a complex Hilbert space has the same distances as its realification, so the real case is the theorem. Separability is SecondCountableTopology, which is equivalent for metric spaces. is the energy expansion of the Setting file, and strong negative type is defined by (3.1) and (3.3), not through an embedding or a barycenter map.
import Mathlib import Definitions.Def_DistCov_Hilbert_Setting
namespace DistCov.Hilbert
open MeasureTheory ProbabilityTheory
theorem theorem_3_16 (H : Type*) [NormedAddCommGroup H] [InnerProductSpace ℝ H] [CompleteSpace H]
[SecondCountableTopology H] [MeasurableSpace H] [BorelSpace H] :
DistCov.Indep.StrongNegType H := by sorry
end DistCov.Hilbert
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.