Fields of covariances on non-commutative probability spaces in finite dimensions

TL;DR

This paper introduces the concept of fields of covariances in non-commutative probability spaces, providing a unified framework that generalizes classical and quantum information geometry and extends existing classifications to non-faithful states.

Contribution

It defines fields of covariances as a categorical analogue of covariance, classifies them in finite dimensions, and extends quantum metric classifications to non-faithful states.

Findings

Unified classical and quantum information geometry framework

Complete classification of fields of covariances in finite dimensions

Extension of quantum metric classifications to non-faithful states

Abstract

We introduce the notion of a field of covariances, a contravariant functor from non-commutative probability spaces to Hilbert spaces, as the natural categorical analogue of statistical covariance. In the case of finite-dimensional non-commutative probability spaces, we obtain a complete classification of such fields. Our results unify classical and quantum information geometry: in the tracial case, we recover (a contravariant version of) Cencov's uniqueness of the Fisher-Rao metric, while in the faithful case, we recover (a contravariant version of) the Morozova-Cencov-Petz classification of quantum monotone metrics. Crucially, our classification extends naturally to non-faithful states that are not pure, thus generalizing Petz and Sudar's radial extension.