Rényi-Induced Information Geometry and Hartigan's Prior Family

2025/05/22 by Rebecca Maria Kuntz, Kuntz, Rebecca Maria, Heinrich von Campe +3
Physics and Astronomy · #FOS: Mathematics #Statistical Mechanics and Entropy #Statistics Theory (math.ST)

paper · pdf · doi:10.48550/arxiv.2506.12028

openalex publication_date 2025/05/22 · openalex created_date 2025/10/13 · openalex updated_date 2026/07/28

Abstract

We derive the information geometry induced by the statistical Rényi divergence, namely its metric tensor, its dual parametrized connections, as well as its dual Laplacians. Based on these results, we demonstrate that the Rényi-geometry, though closely related, differs in structure from Amari's well-known α-geometry. Subsequently, we derive the canonical uniform prior distributions for a statistical manifold endowed with a Rényi-geometry, namely the dual Rényi-covolumes. We find that the Rényi-priors can be made to coincide with Takeuchi and Amari's α-priors by a reparameterization, which is itself of particular significance in statistics. Herewith, we demonstrate that Hartigan's parametrized (α_H) family of priors is precisely the parametrized (ρ) family of Rényi-priors (α_H = ρ).

Citations

• An elementary introduction to information geometry

Related

• On The Equivalence of Projections In Relative α-Entropy and Rényi Divergence
• Information Geometry and Statistical Manifold
• Pseudo-Riemannian geometry embeds information geometry in optimal transport
• Rényi Divergence and Majorization
• Logarithmic divergences from optimal transport and Rényi geometry