Zero biasing and a discrete central limit theorem

DOI10.1214/009117906000000250MaRDI QIDQ858982zbMATH OpenOpenAlexFDO

Authors Larry Goldstein, Aihua Xia

Publication date 12 January 2007

Published in The Annals of Probability (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/math/0509444

zbMATH Keywords

integer-valued random variablesStein's methodtotal variation

Mathematics Subject Classification ID

Central limit and other weak theorems (60F05) Sums of independent random variables; random walks (60G50)

Abstract: We introduce a new family of distributions to approximate $mathbbP(WinA)$ for $Asubset\mathrm{...},-2,-1,0,1,2,\mathrm{...}$ and $W$ a sum of independent integer-valued random variables $x{i}_1$, $x{i}_2$, $\mathrm{...},$ $x{i}_n$ with finite second moments, where, with large probability, $W$ is not concentrated on a lattice of span greater than 1. The well-known Berry--Esseen theorem states that, for $Z$ a normal random variable with mean $mathbbE(W)$ and variance $operatornameVar(W)$, $mathbbP(ZinA)$ provides a good approximation to $mathbbP(WinA)$ for $A$ of the form $(-infty,x]$. However, for more general $A$, such as the set of all even numbers, the normal approximation becomes unsatisfactory and it is desirable to have an appropriate discrete, nonnormal distribution which approximates $W$ in total variation, and a discrete version of the Berry--Esseen theorem to bound the error. In this paper, using the concept of zero biasing for discrete random variables (cf. Goldstein and Reinert [J. Theoret. Probab. 18 (2005) 237--260]), we introduce a new family of discrete distributions and provide a discrete version of the Berry--Esseen theorem showing how members of the family approximate the distribution of a sum $W$ of integer-valued variables in total variation.

Review text

The authors propose a class of approximating distributions which have carrier space \(\mathbb Z\), thus avoiding truncation and integerization problems. These new distributions are uniquely determined by parameters \(\mu\) and \({\sigma}^2\), similarly to how the approximating normal distribution is determined in the classical central limit theorem. It is expected that any such approximating family of discrete distributions be related to the Poisson, a distribution characterized by the property of being equal to its own reduced Palm distribution. As this property is intrinsic in the study of certain Poisson approximations, and since the Palm distribution involves only the first moment of the distribution, it is of interest to determine whether there exists any counterpart to the Poisson also involving the second moment, which gives additional flexibility in approximation. One appropriate counterpart can be uncovered through the concept of zero biasing. Based on the continuous normal case, it is expected that the class of approximating distributions should arrive naturally as the unique candidates which equal their zero-biased distribution.

Recommendations

• Stein's method using approximate zero bias couplings with applications to combinatorial central limit theorems under the Ewens distribution
• Some optimal bounds in the central limit theorem using zero biasing
• Centered Poisson approximation via Stein's method
• Stein's method and the zero bias transformation with application to simple random sampling
• Poisson approximations for sequences of random variables

Cites work

• L^1 bounds in normal approximation
• A non-uniform bound for translated {P}oisson approximation
• Berry-Esseen bounds for combinatorial central limit theorems and pattern occurrences, using zero and size biasing
• Centered Poisson approximation via Stein's method
• Distributional transformations, orthogonal polynomials, and Stein characterizations
• Exponential and uniform ergodicity of Markov processes
• Fourier analysis of distribution functions. A mathematical study of the Laplace-Gaussian law
• scientific article; zbMATH DE number 3866301 (Why is no real title available?)
• scientific article; zbMATH DE number 3951715 (Why is no real title available?)
• scientific article; zbMATH DE number 4104086 (Why is no real title available?)
• scientific article; zbMATH DE number 50805 (Why is no real title available?)
• scientific article; zbMATH DE number 52632 (Why is no real title available?)
• scientific article; zbMATH DE number 3549966 (Why is no real title available?)
• scientific article; zbMATH DE number 3438144 (Why is no real title available?)
• scientific article; zbMATH DE number 205896 (Why is no real title available?)
• scientific article; zbMATH DE number 1834589 (Why is no real title available?)
• Normal approximation for hierarchical structures
• On the binary expansion of a random integer
• On the Number of Successes in Independent Trials
• Poisson approximation for dependent trials
• Poisson perturbations
• Stein's method and birth-death processes
• Stein's method and point process approximation
• Stein's method and the zero bias transformation with application to simple random sampling
• Stein's method, Palm theory and Poisson process approximation.
• The Accuracy of the Gaussian Approximation to the Sum of Independent Variates
• Variational formulas of Poincaré-type inequalities for birth-death processes.

Cited in

(14)

• Poisson type approximations for the Markov binomial distribution
• Error bounds in local limit theorems using Stein's method
• Palm theory, random measures and Stein couplings
• Geometric sums, size biasing and zero biasing
• Local limit theorems via Landau-Kolmogorov inequalities
• Translated Poisson approximation using exchangeable pair couplings
• Discretized normal approximation by Stein's method
• Poisson process approximation for dependent superposition of point processes
• On moderate deviations in Poisson approximation
• Normal approximation in total variation for statistics in geometric probability
• Linking population-size-dependent and controlled branching processes
• Generalized gamma approximation with rates for urns, walks and trees
• Kendall's tau-type rank statistics in genome data.
• A polynomial birth-death point process approximation to the Bernoulli process

This page was built for publication: Zero biasing and a discrete central limit theorem

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q858982)