Diffeological statistical models, the Fisher metric and probabilistic mappings
Abstract
In this note we introduce the notion of a -diffeological statistical model, which allows us to apply the theory of diffeological spaces to (possibly singular) statistical models. In particular, we introduce a class of almost 2-integrable -diffeological statistical models that encompasses all known statistical models for which the Fisher metric is defined. This class contains a statistical model which does not appear in the Ay-Jost-L\^e-Schwachh\"ofer theory of parametrized measure models. Then we show that for any positive integer the class of almost 2-integrable -diffeological statistical models is preserved under probabilistic mappings. Furthermore, the monotonicity theorem for the Fisher metric also holds for this class. As a consequence, the Fisher metric on an almost 2-integrable -diffeological statistical model is preserved under any probabilistic mapping that is sufficient w.r.t. . Finally we extend the Cram\'er-Rao inequality to the class of 2-integrable -diffeological statistical models.
Keywords
Cite
@article{arxiv.1912.02090,
title = {Diffeological statistical models, the Fisher metric and probabilistic mappings},
author = {Hông Vân Lê},
journal= {arXiv preprint arXiv:1912.02090},
year = {2020}
}
Comments
16 p., final version, accepted to Journal Mathematics/MDPI