Geometry and Duality of Alternating Markov Chains

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Mithal, Deven, Orecchia, Lorenzo
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918003103236096
author Mithal, Deven
Orecchia, Lorenzo
author_facet Mithal, Deven
Orecchia, Lorenzo
contents In this note, we realize the half-steps of a general class of Markov chains as alternating projections with respect to the reverse Kullback-Leibler divergence between convex sets of joint probability distributions. Using this characterization, we provide a geometric proof of an information-theoretic duality between the Markov chains defined by the even and odd half-steps of the alternating projection scheme.
format Preprint
id arxiv_https___arxiv_org_abs_2410_12721
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Geometry and Duality of Alternating Markov Chains
Mithal, Deven
Orecchia, Lorenzo
Probability
Information Theory
In this note, we realize the half-steps of a general class of Markov chains as alternating projections with respect to the reverse Kullback-Leibler divergence between convex sets of joint probability distributions. Using this characterization, we provide a geometric proof of an information-theoretic duality between the Markov chains defined by the even and odd half-steps of the alternating projection scheme.
title Geometry and Duality of Alternating Markov Chains
topic Probability
Information Theory
url https://arxiv.org/abs/2410.12721