Compound Poisson Approximation via Information Functionals
Abstract
An information-theoretic development is given for the problem of compound Poisson approximation, which parallels earlier treatments for Gaussian and Poisson approximation. Let be the distribution of a sum of independent integer-valued random variables . Nonasymptotic bounds are derived for the distance between and an appropriately chosen compound Poisson law. In the case where all have the same conditional distribution given , a bound on the relative entropy distance between and the compound Poisson distribution is derived, based on the data-processing property of relative entropy and earlier Poisson approximation results. When the have arbitrary distributions, corresponding bounds are derived in terms of the total variation distance. The main technical ingredient is the introduction of two "information functionals," and the analysis of their properties. These information functionals play a role analogous to that of the classical Fisher information in normal approximation. Detailed comparisons are made between the resulting inequalities and related bounds.
Cite
@article{arxiv.1004.3692,
title = {Compound Poisson Approximation via Information Functionals},
author = {A. D. Barbour and Oliver Johnson and Ioannis Kontoyiannis and Mokshay Madiman},
journal= {arXiv preprint arXiv:1004.3692},
year = {2019}
}
Comments
27 pages