Searcharxiv⌕ Search

arXiv subjects

Hiroshi Matsuzoe

Publications and source records attributed to Hiroshi Matsuzoe.

9 recordsLinked to original sources
arXiv2025

Divergence and Deformed Exponential Family

The Kullback--Leibler divergence together with exponential families establishes the foundation of information geometry and is widely generalized. Among the generalization, we focus on the $(h,τ)$-divergence and $(h,τ)$-exponential families. We present a sufficient condition for the $(h,τ)$-divergence to induce a Hessian structure on an $(h,τ)$-exponential family. We also define the $(h,τ)$-dependence of random variables and prove a kind of the law of large numbers.

math.DG↗
arXiv2023

Huygens' equations and the gradient-flow equations in information geometry

We revisit the relation between the gradient-flow equations and Hamilton's equations in information geometry. By regarding the gradient-flow equations as Huygens' equations in geometric optics, we have related the gradient flows to the geodesic flows induced by the geodesic Hamiltonian in an appropriate Riemannian geometry. The original evolution parameter $\textit{t}$ in the gradient-flow equations is related to the arc-length parameter in the associated Riemannian manifold by Jacobi-Maupertuis transformation. As a by-product, it is found the relation between the gradient-flow equation and replicator equations.

cs.IT↗
arXiv2020

Invariant metric under deformed Markov embeddings with overlapped supports

Due to Čencov's theorem, there exists a unique family of invariant symmetric $(0,2)$-tensor fields on the space of positive probability measures on a set of $n$-points indexed by $n\in \mathbb{N}$ under Markov embeddings. We deform Markov embeddings keeping sufficiency, and prove existence and uniqueness of invariant families under the embeddings.

math.DG↗
arXiv2020

Gauge freedom of entropies on $q$-Gaussian measures

A $q$-Gaussian measure is a generalization of a Gaussian measure. This generalization is obtained by replacing the exponential function with the power function of exponent $1/(1-q)$ ($q\neq 1$). The limit case $q=1$ recovers a Gaussian measure. For $1\leq q <3$, the set of all $q$-Gaussian densities over the real line satisfies a certain regularity condition to define information geometric structures such as an entropy and a relative entropy via escort expectations. The ordinary expectation of a random variable is the integral of the random variable with respect to its law. Escort expectations admit us to replace the law to any other measures. A choice of escort expectations on the set of all $q$-Gaussian densities determines an entropy and a relative entropy. One of most important escort expectations on the set of all $q$-Gaussian densities is the $q$-escort expectation since this escort expectation determines the Tsallis entropy and the Tsallis relative entropy. The phenomenon gauge freedom of entropies is that different escort expectations determine the same entropy, but different relative entropies. In this note, we first introduce a refinement of the $q$-logarithmic function. Then we demonstrate the phenomenon on an open set of all $q$-Gaussian densities over the real line by using the refined $q$-logarithmic functions. We write down the corresponding Riemannian metric.

math.QA↗
arXiv2019

Advantages of $q$-logarithm representation over $q$-exponential representation from the sense of scale and shift on nonlinear systems

Addition and subtraction of observed values can be computed under the obvious and implicit assumption that the scale unit of measurement should be the same for all arguments, which is valid even for any nonlinear systems. This paper starts with the distinction between exponential and non-exponential family in the sense of the scale unit of measurement. In the simplest nonlinear model ${dy}/{dx}=y^{q}$, it is shown how typical effects such as rescaling and shift emerge in the nonlinear systems and affect observed data. Based on the present results, the two representations, namely the $q$-exponential and the $q$-logarithm ones, are proposed. The former is for rescaling, the latter for unified understanding with a fixed scale unit. As applications of these representations, the corresponding entropy and the general probability expression for unified understanding with a fixed scale unit are presented. For the theoretical study of nonlinear systems, $q$-logarithm representation is shown to have significant advantages over $q$-exponential representation.

math-ph↗
arXiv2019

On the canonical distributions of a thermal particle in the weakly confining potential of special type

We consider a thermal particle which is diffusing in velocity-space and in a weakly confining potential characterized by the inverse hyperbolic sine function of the particle velocity $v$ and the control parameter $v_c$. The stationary state of the Fokker-Planck equation is shown to be a canonical probability distribution. Furthermore an appropriate re-parametrization relates this stationary state with the $κ$-deformed Gaussian.

cond-mat.stat-mech↗
arXiv2010

Dually flat structure with escort probability and its application to alpha-Voronoi diagrams

This paper studies geometrical structure of the manifold of escort probability distributions and shows its new applicability to information science. In order to realize escort probabilities we use a conformal transformation that flattens so-called alpha-geometry of the space of discrete probability distributions, which well characterizes nonadditive statistics on the space. As a result escort probabilities are proved to be flat coordinates of the usual probabilities for the derived dually flat structure. Finally, we demonstrate that escort probabilities with the new structure admits a simple algorithm to compute Voronoi diagrams and centroids with respect to alpha-divergences.

cond-mat.stat-mech↗