aiqbench analysis
This paper develops rigorous mathematical foundations for non-commutative probability spaces, establishing categorical analogues of statistical covariance within finite-dimensional quantum systems. By characterizing these fields through operator monotone functions, the work provides a precise geometric framework linking modular operators to metrics like the Fisher-Rao and Fubini-Study tensors on state manifolds. While highly theoretical, this research advances the mathematical underpinnings essential for understanding information geometry in quantum machine learning and quantum statistical inference.
Fields of covariances on non-commutative probability spaces in finite dimensions
arXiv:2510.24617v2 Announce Type: replace-cross Abstract: We introduce the notion of a field of covariances, a contravariant functor from non-commutative probability spaces to Hilbert spaces, as a categorical analogue of statistical covariance. In the finite-dimensional setting, we obtain a complete characterization of the fields of covariances satisfying an additional continuity property in terms of a strictly positive constant $\alpha$ and a continuous function $F\colon[0,+\infty)\rightarrow(0,+\infty)$ that is operator monotone on $(0,+\infty)$. In the commutative case, our result recovers a contravariant formulation of \v{C}encov's characterization of the Fisher--Rao metric tensor, and extends the corresponding uniqueness statement to tracial states on possibly noncommutative algebras. On faithful states, a subfamily satisfying a symmetry constraint recovers the regular Morozova--\v{C}encov--Petz monotone metric tensors. More generally, the covariance fields induce positive contravariant symmetric tensors on the homogeneous manifolds of states generated by the action of the invertible elements of an arbitrary finite-dimensional $C^*$-algebra. The geometry internal to the support is determined by the positive spectrum of the modular operator, whereas support-changing directions are governed by $F(0)$. For $F(t)=(1+t)/2$, and with a suitable normalization convention, the resulting tensors recover the inverse of the Riemannian metrics induced by the Jordan product. In particular, on pure states, every $F$ induces a multiple of the inverse Fubini--Study tensor, with coefficient $2F(0)$ in our normalization, in agreement after inversion with the Petz--Sud\'ar radial extension.
Abstract:We introduce the notion of a field of covariances, a contravariant functor from non-commutative probability spaces to Hilbert spaces, as a categorical analogue of statistical covariance. In the finite-dimensional setting, we obtain a complete characterization of the fields of covariances satisfying an additional continuity property in terms of a strictly positive constant $\alpha$ and a continuous function $F\colon[0,+\infty)\rightarrow(0,+\infty)$ that is operator monotone on $(0,+\infty)$. In the commutative case, our result recovers a contravariant formulation of Čencov's characterization of the Fisher--Rao metric tensor, and extends the corresponding uniqueness statement to tracial states on possibly noncommutative algebras. On faithful states, a subfamily satisfying a symmetry constraint recovers the regular Morozova--Čencov--Petz monotone metric tensors. More generally, the covariance fields induce positive contravariant symmetric tensors on the homogeneous manifolds of states generated by the action of the invertible elements of an arbitrary finite-dimensional $C^*$-algebra. The geometry internal to the support is determined by the positive spectrum of the modular operator, whereas support-changing directions are governed by $F(0)$. For $F(t)=(1+t)/2$, and with a suitable normalization convention, the resulting tensors recover the inverse of the Riemannian metrics induced by the Jordan product. In particular, on pure states, every $F$ induces a multiple of the inverse Fubini--Study tensor, with coefficient $2F(0)$ in our normalization, in agreement after inversion with the Petz--Sudár radial extension.
54 pages - Comments are welcome!
arXiv:2510.24617 [math-ph]
(or arXiv:2510.24617v2 [math-ph] for this version)
https://doi.org/10.48550/arXiv.2510.24617 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 311 words
Related coverage

Measurement incompatibility in Bayesian multiparameter quantum estimation
Quantum Journal · Aug 13, 2026
Cross-Calibrated Confidence Fields for Local Risk Updates
arXiv · Aug 17, 2026

Best MagSafe wallets of 2026: Expert tested and reviewed
ZDNet · Aug 3, 2026
Synthesis Superintelligence: from Semiconductors to Superconductors — Periodic Labs’ Liam Fedus and Ekin Dogus Cubuk
Latent Space · Oct 8, 2026