Kullback divergence D(p ∥ q) for a couple of probability distributions {p, q} and Fisher information J(θ) for a parametric family of probability distributions {p θ } are very important notions in the classical theory of statistical inference. Quantum analogues of these notions are known as Umegaki's relative entropy D(ρ ‖ σ) for a couple of density operators {ρ, σ} and the quantum Fisher information J(θ) based on the pymmetric logarithmic derivatives for a parametric family of density operators {ρ Θ }. In the classical case, the relation holds, which connects the asymptotic theory of hypothesis testing and that of parameter estimation. In the quantum case, the relation defines , which is not equal to J(θ) in general and is another quantum analogue of Fisher information. We show the inequality and elucidate its meaning in connection with the statistical inference for quantum states.
In classical statistics, the Fisher information is unique in the sense that it is essentially the only monotone Riemannian metric on the space of probability densities. In quantum theory, this uniqueness breaks down, and there are many natural quantum analogues of the Fisher information, among which two particular versions distinguish themselves by their intuitive and informational significance: The first has its origin in the skew information introduced by Wigner and Yanase in 1963 in the context of quantum measurement, and is defined via the square root of the density operator. The second arises from Helstrom's study of quantum detection in 1967, and is defined via the symmetric logarithmic derivative. The aim of this paper is to compare these two versions of quantum Fisher information, and to establish two informational inequalities relating them.
In this paper it was proved that the quantum relative entropy D(k) can be asymptotically attained by Kullback Leibler divergences of probabilities given by a certain sequence of measurements. The sequence of measurements depends on , but is independent of the choice of . In classical statistical theory the relative entropy D(pkq) is an information quantity which means the statistical eÆciency in order to distinguish a probability measure p of a measurable space from another probability measure q of the same measurable space. The states correspond to measures on measurable space. When p; q are discrete probabilities, the relative entropy (called also information divergence) introduced by Kullback and Leibler is dened by [1]: D(pkq) := X i p i log p i q i : In general, when p; q are measures on measurable space the relative entropy is dened by: D(pkq) := Z log dq dp (!)p( d!); where dq dp (!) is Radon-Nikodym derivative of q with respect to p. In this paper, we co...
Quantum generalizations of the Fisher information are important in quantum information science, with applications in high energy and condensed matter physics and in quantum estimation theory, machine learning, and optimization. One can derive a quantum generalization of the Fisher information matrix in a natural way as the Hessian matrix arising in a Taylor expansion of a smooth divergence. Such an approach is appealing for quantum information theorists, given the ubiquity of divergences in quantum information theory. In contrast to the classical case, there is not a unique quantum generalization of the Fisher information matrix, similar to how there is not a unique quantum generalization of the relative entropy or the Rényi relative entropy. In this paper, I derive information matrices arising from the log-Euclidean,
α-
z, and geometric Rényi relative entropies, with the main technical tool for doing so being the method of divided differences for calculating matrix derivatives. Interestingly, for all non-negative values of the Rényi parameter
α, the log-Euclidean Rényi relative entropy leads to the Kubo-Mori information matrix, and the geometric Rényi relative entropy leads to the right-logarithmic derivative Fisher information matrix. Thus, the resulting information matrices obey the data-processing inequality for all non-negative values of the Rényi parameter
α even though the original quantities do not. Additionally, I derive and establish basic properties of
α-
z information matrices resulting from the
α-
z Rényi relative entropies. For parameterized thermal states and time-evolved states, I establish formulas for their
α-
z information matrices and hybrid quantum-classical algorithms for estimating them, with applications in quantum Boltzmann machine learning.
A new canonical divergence is put forward for generalizing an information-geometric measure of complexity for both classical and quantum systems. On the simplex of probability measures, it is proved that the new divergence coincides with the Kullback-Leibler divergence, which is used to quantify how much a probability measure deviates from the non-interacting states that are modeled by exponential families of probabilities. On the space of positive density operators, we prove that the same divergence reduces to the quantum relative entropy, which quantifies many-party correlations of a quantum state from a Gibbs family.
We derive the quantum Fisher information for entropy estimation in a Gibbs state and show that it equals the inverse of the heat capacity, which is dual to the temperature Fisher information given by the heat capacity divided by the square of the temperature. Their product is independent of the Hamiltonian and depends only on the temperature, leading to a metrological uncertainty relation between the variances of entropy and temperature estimators in which all system-specific quantities cancel. This relation arises from the dually-flat structure of the Gibbs exponential family expressed in thermodynamic coordinates, and holds for all standard thermodynamically conjugate pairs. We identify energy measurement as the optimal protocol for entropy estimation, analyse critical-point scaling where the entropy Fisher information vanishes, and connect it to the Ruppeiner metric in entropy coordinates. We lastly examine the distinguished role of the von Neumann entropy within the Rényi family. Generalisations to the grand canonical and generalised Gibbs ensembles are given.