Dually flat structure of von Mises-Fisher distributions

The von Mises--Fisher (vMF) distribution is the natural exponential family analogue of the normal distribution on the unit hypersphere. We bring together previous results of its dually flat structure and establish new results on the information geometry of the family of vMF distributions. In particular, we characterize the exponential and mixture autoparallel submanifolds, and give the classification of doubly autoparallel submanifolds. Furthermore, we derive explicit formulas for the unique minimizers of the Kullback-Leibler divergence on these submanifolds and establish the corresponding projection theorems.

Authors

Publication Details

Journal
arXiv (Cornell University)
Published
2026-10-29
DOI
https://doi.org/10.48550/arxiv.2610.39249
Primary Topic
Advanced Differential Geometry Research
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
preprint

Dually flat structure of von Mises-Fisher distributions

Behzad Najafi, Ryu Ueno, Morteza Sharifi-Najjar
arXiv (Cornell University)
Advanced Differential Geometry Research
preprint

Dually flat structure of von Mises-Fisher distributions

Behzad Najafi, Ryu Ueno, Morteza Sharifi-Najjar
preprint en

Abstract

The von Mises--Fisher (vMF) distribution is the natural exponential family analogue of the normal distribution on the unit hypersphere. We bring together previous results of its dually flat structure and establish new results on the information geometry of the family of vMF distributions. In particular, we characterize the exponential and mixture autoparallel submanifolds, and give the classification of doubly autoparallel submanifolds. Furthermore, we derive explicit formulas for the unique minimizers of the Kullback-Leibler divergence on these submanifolds and establish the corresponding projection theorems.

arXiv (Cornell University)
Advanced Differential Geometry Research
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon