Uniform rates of the Glivenko-Cantelli convergence and their use in approximating Bayesian inferences

DOI10.3150/18-BEJ1077MaRDI QIDQ2325348zbMATH OpenFDO

Authors Emanuele Dolera, Eugenio Regazzini

Publication date 25 September 2019

Published in Bernoulli (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1712.07361, https://projecteuclid.org/euclid.bj/1568362049

dominated ergodic theorem empirical measure exchangeability Glivenko-Cantelli theorem law of the iterated logarithm posterior distribution predictive distribution Wasserstein distance

Mathematics Subject Classification ID

Convergence of probability measures (60B10) Bayesian inference (62F15)

Abstract: This paper deals with the problem of quantifying the approximation a probability measure by means of an empirical (in a wide sense) random probability measure, depending on the first n terms of a sequence of random elements. In Section 2, one studies the range of oscillation near zero of the Wasserstein distance

u d_{p m s}^{(p)}

between

p f r a k_{0}

and

h a t {p f r a k}_{n}

, assuming that the

x i t i l_{i}

's are i.i.d. with

p f r a k_{0}

as common law. Theorem 2.3 deals with the case in which

p f r a k_{0}

is fixed as a generic element of the space of all probability measures on

(d, m a t h s c r B (d))

and

h a t {p f r a k}_{n}

coincides with the empirical measure. In Theorem 2.4 (Theorem 2.5, respectively) pfrak_0 is a d-dimensional Gaussian distribution (an element of a distinguished type of statistical exponential family, respectively) and

h a t {p f r a k}_{n}

is another

d

-dimensional Gaussian distribution with estimated mean and covariance matrix (another element of the same family with an estimated parameter, respectively). These new results improve on allied recent works (see, e.g., [31]) since they also provide uniform bounds with respect to

n

, meaning that the finiteness of the p-moment of the random variable

s u p_{n g e q 1} b_{n} u d_{p m s}^{(p)} (p f r a k_{0}, h a t {p f r a k}_{n})

is proved for some suitable diverging sequence b_n of positive numbers. In Section 3, under the hypothesis that the

x i t i l_{i}

's are exchangeable, one studies the range of the random oscillation near zero of the Wasserstein distance between the conditional distribution--also called posterior--of the directing measure of the sequence, given

x i t i l_{1}, d o t s, x i t i l_{n}

, and the point mass at

h a t {p f r a k}_{n}

. In a similar vein, a bound for the approximation of predictive distributions is given. Finally, Theorems from 3.3 to 3.5 reconsider Theorems from 2.3 to 2.5, respectively, according to a Bayesian perspective.

Recommendations

Cites work

Cited in

(10)

This page was built for publication: Uniform rates of the Glivenko-Cantelli convergence and their use in approximating Bayesian inferences

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2325348)