A Weighted Spectral Quantum Fidelity
Cong Trinh Le, The Khoi Vu, Minh Toan Ho, Trung Hoa Dinh · 2026
Read A Weighted Spectral Quantum Fidelity by Cong Trinh Le, The Khoi Vu, Minh Toan Ho, Trung Hoa Dinh (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-17167.txt.
Introduction
Le and colleagues introduce a one-parameter family of quantum fidelities built on the weighted spectral geometric mean. The family bridges trivial overlap at the endpoints and Uhlmann fidelity at the midpoint. They prove invariance, multiplicativity and related structural properties, and find violations of data-processing for most parameter values. Closed forms cover pure states and qubits, plus a first Fuchs-van de Graaf inequality.
Catalog record
| CREATOR | Cong Trinh Le, The Khoi Vu, Minh Toan Ho, Trung Hoa Dinh |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 41823 |
| LICENSE | http://creativecommons.org/publicdomain/zero/1.0/ |
| RIGHTS | CC0 |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.17167
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A Weighted Spectral Quantum Fidelity Cong Trinh Le, The Khoi Vu, Minh Toan Ho, Trung Hoa Dinh Abstract We introduce and study a one-parameter family of fidelity-type quantities based on the weighted spectral geometric mean, which we call the weighted spectral fidelity 𝖥 t spec ( ρ , σ ) := Tr [ ρ ( ρ − 1 ♯ σ ) 2 t ] , t ∈ [ 0 , 1 ] . \mathsf{F}_{t}^{\mathrm{spec}}(\rho,\sigma):=\operatorname{Tr}\!\big[\rho(\rho^{-1}\sharp\sigma)^{2t}\big],\ t\in[0,1]. This family interpolates smoothly between the trivial overlap ( t = 0 , 1 t=0,1 ) and the Uhlmann (root) fidelity at t = 1 2 t=\tfrac{1}{2} , and it is distinct from the sandwiched Rényi family except at this midpoint. We establish core structural features-unitary invariance, tensor stabilization and multiplicativity, flip symmetry, endpoint behavior, and a orthogonality criterion. We further show explicit violations of DPI for generic t ≠ 1 2 t\neq\tfrac{1}{2} . For concavity in the state variables we obtain concavity in each variable separately. Closed forms are obtained for pure states and for qubits in Bloch coordinates. We also extend the first Fuchs–van de Graaf inequality to 𝖥 t spec \mathsf{F}_{t}^{\mathrm{spec}} for all t ∈ [ 0 , 1 ] t\in[0,1] , while the second inequality fails away from the midpoint. Keywords: Uhlmann fidelity, Matsumoto fidelity, spectral geometric mean, concavity, log-convexity, qubits 1 Introduction Quantifying the similarity between quantum states is a fundamental task in quantum information theory, with applications ranging from quantum hypothesis testing and error correction to quantum cryptography and quantum algorithms. A variety of fidelity-type measures have been introduced to formalize this notion of similarity. Among them, the Uhlmann fidelity [ 29 ] , also called the transition probability, has emerged as one of the most widely used. It admits multiple equivalent formulations, such as \mathsf{F}_{\mathrm{U}}(\rho,\sigma)=\operatorname{Tr}\sqrt{\rho^{1/2}\sigma\rho^{1/2}}, and satisfies a desirable list of axioms including symmetry, unitary invariance, monotonicity under quantum channels, and joint concavity [ 18 ] . These properties make it an indispensable tool in modern quantum theory. Beyond Uhlmann’s fidelity, other measures have been studied. Holevo introduced an alternative quantity in the 1970s, sometimes called the just-as-good fidelity or Holevo fidelity [ 17 ] , which was later rediscovered under different names including the pretty-good fidelity [ 7 ] , the A -fidelity [ 10 ] , and the affinity [ 26 ] . A comprehensive overview of such fidelity measures is provided by Liang et al. [ 21 ] . Each fidelity encapsulates a different operational or geometric perspective on quantum state similarity. For instance, the Uhlmann fidelity is linked to quantum-to-classical measurement scenarios, while Holevo’s fidelity connects to certain recovery tasks [ 30 ] . In 2010, Matsumoto introduced yet another fidelity function, which can be expressed compactly in terms of the matrix geometric mean: \mathsf{F}_{\mathrm{M}}(\rho,\sigma)=\operatorname{Tr}(\rho\sharp\sigma),\qquad\rho\sharp\sigma:=\rho^{1/2}\!\left(\rho^{-1/2}\sigma\rho^{-1/2}\right)^{1/2}\!\rho^{1/2}, for invertible density operators [ 22 , 23 , 24 ] . This measure, now known as the Matsumoto fidelity , satisfies many of the same axioms as Uhlmann’s fidelity and Holevo’s fidelity, but offers a dual interpretation: it can be viewed as the maximum classical fidelity associated with classical-to-quantum preparation procedures [ 11 ] . Despite its elegance, the Matsumoto fidelity remained relatively underexplored until the more recent analysis of Cree and Sikora [ 11 ] , who studied it through the lens of semidefinite programming and provided a geometric interpretation in terms of the Riemannian metric on the cone of positive definite matrices. These developments highlight the central role of the matrix geometric mean [ 5 , 9 ] in bridging linear algebra, operator theory, and quantum information. It appears not only in the definition of Matsumoto fidelity, but also in alternative characterizations of Uhlmann fidelity (e.g. Alberti’s variational principle [ 1 ] ) and in inequalities fundamental to quantum information [ 4 ] . Building on this framework, we introduce and analyze a spectral fidelity family : \mathsf{F}_{t}^{\mathrm{spec}}(\rho,\sigma)=\operatorname{Tr}\!\big(\rho(\rho^{-1}\sharp\sigma)^{2t}\big),\qquad t\in[0,1], which interpolates between trivial overlaps and Uhlmann fidelity at t=\tfrac{1}{2} . This family inherits many desirable structural properties such as unitary invariance, tensor product multiplicativity, concavity, log-convexity in t , and a natural flip symmetry \mathsf{F}_{t}^{\mathrm{spec}}(\rho,\sigma)=\mathsf{F}_{1-t}^{\mathrm{spec}}(\sigma,\rho). Our aim in this work is to place \mathsf{F}_{t}^{\mathrm{spec}} in the broader landscape of quantum fidelity measures, to establish its structural properties rigorously, and to highlight connections to Ulhmann and Matsumoto fidelities. In particular, we show how the parameter t interpolates between different operational regimes. In Section 2 we recall basic definitions and properties of matrix and spectral geometric means. Section 3 introduces the weighted spectral fidelity \mathsf{F}_{t}^{\mathrm{spec}} and establishes its fundamental structural properties, including flip symmetry, unitary and tensor invariances, multiplicativity, endpoint behavior, and the orthogonality (zero-support) criterion. Section 4 derives closed forms of \mathsf{F}_{t}^{\mathrm{spec}} when one state is pure and presents Bloch-sphere expressions for qubit systems. Section 5 analyzes concavity and log-convexity using operator perspective techniques and Hadamard’s three-lines theorem, proving separate concavity in each state variable for all t\in[0,1] and log-convexity in the parameter t . We provide explicit counterexamples showing that the data processing inequality fails for all t\neq\tfrac{1}{2} . Section 6 extends the first Fuchs–van de Graaf inequality to all t\in[0,1] and shows the second fails away from the midpoint. 2 Preliminaries Let \mathbb{M}_{n} denote the algebra of n\times n complex matrices, and let \mathbb{P}_{n} denote the cone of positive definite matrices in \mathbb{M}_{n} . For A,B\in\mathbb{P}_{n} , the matrix geometric mean was introduced by Pusz and Woronowicz [ 27 ] in 1975 as A\sharp B\;=\;A^{1/2}\bigl(A^{-1/2}BA^{-1/2}\bigr)^{1/2}A^{1/2}. This mean is uniquely characterized as the positive solution X of the Riccati equation XA^{-1}X=B. In 1997, Fiedler and Pták [ 13 ] proposed the spectral geometric mean , defined by A\natural B\;=\;(A^{-1}\sharp B)^{1/2}\,A\,(A^{-1}\sharp B)^{1/2}. The terminology “spectral mean” originates from a fundamental eigenvalue property of this construction. A key feature is that (A\natural B)^{2} is similar to the product AB . Indeed, using the identity A^{-1}\sharp B=A^{-1/2}\,(A^{1/2}BA^{1/2})^{1/2}\,A^{-1/2}, one checks directly that (A\natural B)^{2}\sim A^{1/2}(A^{-1/2}BA^{-1/2})A^{1/2}=AB, where \sim denotes similarity of matrices. Since similarity preserves eigenvalues, it follows that \lambda\bigl((A\natural B)^{2}\bigr)=\lambda(AB). As A\natural B is positive definite, its eigenvalues are therefore the positive square roots of the eigenvalues of AB , namely \lambda(A\natural B)=\sqrt{\lambda(AB)}. This eigenvalue correspondence is the primary motivation for the term “spectral mean.” Moreover, the factor A^{-1}\sharp B has a variational characterization: it is the unique minimizer of F(X)=\operatorname{Tr}(AX)+\operatorname{Tr}(BX^{-1}),\qquad X>0, see, e.g., [ 9 ] . In 2015, Kim and Lee [ 20 ] introduced the weighted spectral mean , given by A\natural_{t}B\;=\;(A^{-1}\sharp B)^{t}\,A\,(A^{-1}\sharp B)^{t},\qquad t\in[0,1], which provides a continuous interpolation between A and B through spectral means. This family has subsequently been linked to the study of relative operator entropy and to several operator inequalities [ 14 , 20 , 19 ] . Definition 2.1 (Weighted spectral fidelity) . For density operators \rho,\sigma , we define the weighted spectral fidelity by \mathsf{F}_{t}^{\mathrm{spec}}(\rho,\sigma):=\operatorname{Tr}(\rho\natural_{t}\sigma)=\operatorname{Tr}\!\big(\rho\,(\rho^{-1}\sharp\sigma)^{2t}\big),\qquad t\in[0,1]. Remark 2.1 . Throughout the paper, the superscript “ \mathrm{spec} ” emphasizes that the quantity \mathsf{F}_{t}^{\mathrm{spec}}(\rho,\sigma) is a fidelity functional derived from the spectral geometric mean, and should not be confused with the operator mean \natural_{t} itself. The quantity \mathsf{F}_{t}^{\mathrm{spec}}(\rho,\sigma) should not be confused with the sandwiched Rényi divergence [ 25 , 8 ] . For \alpha>0 , \alpha\neq 1 , the sandwiched Rényi relative entropy is defined