 Research
 Open Access
 Published:
Bounds for the approximation of Poissonbinomial distribution by Poisson distribution
Journal of Inequalities and Applications volume 2013, Article number: 30 (2013)
Abstract
Let ({X}_{nk}, k=1,2,\dots ,n; n=1,2,\dots) be a rowwise triangular array of independent Bernoulli random variables with success probabilities P({X}_{nk}=1)=1P({X}_{nk}=0)={p}_{nk}\in [0,1], k=1,2,\dots ,n; n=1,2,\dots . For every n=1,2,\dots , the random variables {S}_{n}={\sum}_{k=1}^{n}{X}_{nk} have probability distributions with complicated structure and therefore they are used to being approximated by Poisson distribution. Wellknown Le Cam’s inequality is established for providing information on the quality of a Poisson approximation. The main aim of this paper is to reestablish the Le Camtype inequalities via a linear operator. The operator method used in this paper is quite elementary and it also could be applied for the probability distributions of random sums {S}_{{N}_{n}}={\sum}_{k=1}^{{N}_{n}}{X}_{nk} in the Poisson approximation, where {N}_{n}, n=1,2,\dots , are positive integervalued random variables, independent of all {X}_{nk}, k=1,2,\dots ,n; n=1,2,\dots .
MSC:60F05, 60G50, 41A36.
1 Introduction
Throughout this paper, let ({X}_{nk}, k=1,2,\dots ,n; n=1,2,\dots) be a rowwise triangular array of independent Bernoulli random variables with success probabilities P({X}_{nk}=1)=1P({X}_{nk}=0)={p}_{nk}\in [0,1], k=1,2,\dots ,n; n=1,2,\dots . The random variables {S}_{n}={\sum}_{k=1}^{n}{X}_{nk}, n=1,2,\dots , are often called the Poissonbinomial random variables. And it is easily seen that the mean, variance, and characteristic function of {S}_{n}, n=1,2,\dots , are E({S}_{n})={\sum}_{k=1}^{n}{p}_{nk}, D({S}_{n})={\sum}_{k=1}^{n}{p}_{nk}(1{p}_{nk}), and {f}_{{S}_{n}}(t)=E({e}^{it{S}_{n}})={\prod}_{k=1}^{n}(1{p}_{nk}+{p}_{nk}{e}^{it}), respectively.
The probability distributions of {S}_{n}, n=1,2,\dots , have many applications in various areas of mathematics and statistics such as reliability, survival analysis, survey sampling, econometrics, and so on (the reader is referred to [1, 2] and [3] for full development). However, since the probability distributions of {S}_{n}, n\ge 1, have the complicated structure (see, for instance, [3]), they are used to being approximated by the distribution of Poisson random variables {Z}_{{\lambda}_{n}} with a positive parameter {\lambda}_{n}=E({S}_{n})={\sum}_{k=1}^{n}{p}_{nk}. More specifically, assume that
then
where, and from now on, the notation \stackrel{d}{\to} means the convergence in distribution (see, for instance, [4]). Moreover, remarkable Le Cam’s inequality for the Poissonbinomial distribution [5] is widely considered in literature as follows:
(we refer the reader to the results of Le Cam [5], Barbour, Holst, and Janson [6], Steele [7], Chen [8], Chen and Liu [1], Neammanee [9], and Ross [10] for more details).
It should be noted that in [6, 7], and [9] various powerful tools (such as the method of matrix analysis, the semigroup method, the coupling method, and the ChenStein method) for providing Le Cam’s inequality have been demonstrated. The main objective of this paper is to obtain the bounds for wellknown Le Cam’s inequality in (3) using the operator method, introduced by Renyi [4]. In the third section, we use the operator method from [4] to establish the bounds for the approximation of Poissonbinomial distribution by Poisson distribution. The operator method in this paper is quite elementary and it also could be applied for random sums {S}_{{N}_{n}}={\sum}_{k=1}^{{N}_{n}}{X}_{nk}, {S}_{0}=0, where {N}_{n}, n=1,2,\dots are positive integervalued random variables, independent of all {X}_{nk}, k=1,2,\dots ,n; n=1,2,\dots . This will be taken up in the last section. We refer the reader to the works of Trotter [11], Renyi [4], and Hung [12] for a deeper discussion of this operator method. Based on the operator method, the received results of this paper are analogues of Le Cam’s inequality in classical literature (we refer the reader to Steele [7], Le Cam [5], Chen [8], Neammanee [9], and Wang [3] for a complete treatment of the problem).
2 Preliminaries
In the sequel we will need the operator method, which has been used for a long time in various studies of classical limit theorems for sums of independent random variables (see Trotter [11], Renyi [4], and Hung [12] for the complete bibliography).
We recall some definitions and notations. We denote by K the set of all realvalued bounded functions f(x), defined on the set of nonnegative integers {Z}_{+}=\{0,1,2,\dots \}. The norm of a function f\in K is defined by \parallel f\parallel ={sup}_{x\in {Z}_{+}}f(x).
Definition 2.1 We define a linear operator associated with a positive discrete random variable X, {A}_{X}:K\to K, by setting
It is to be noticed that the linear operator defined in (4) is actually a discrete form of Trotter’s operator (we refer the readers to Trotter [11], Renyi [4], and Hung [12] for a more general and detailed discussion of this operator method).
We will need some properties of the operator in (4) in the sequel. Let {A}_{X}, {A}_{Y} be operators associated with two discrete random variables X and Y for f,g\in K. Suppose that α and β are two real numbers, then we easily get the following linear property of the operator in (4):
We define the operator ({A}_{X}+{A}_{Y}) by ({A}_{X}+{A}_{Y})f={A}_{X}f+{A}_{Y}f, \mathrm{\forall}f\in K, and the product of two operators {A}_{X} and {A}_{Y} is ({A}_{X}{A}_{Y})f={A}_{X}({A}_{Y}f), \mathrm{\forall}f\in K.
It is obvious that

1.
\parallel {A}_{X}f\parallel \u2a7d\parallel f\parallel for all f\in K.

2.
\parallel {A}_{X}f+{A}_{Y}f\parallel \u2a7d\parallel {A}_{X}f\parallel +\parallel {A}_{Y}f\parallel for all f\in K.

3.
Suppose that {A}_{X} and {A}_{Y} are operators associated with two independent random variables X, Y and f\in K. Then {A}_{X}{A}_{Y}f={A}_{Y}{A}_{X}f={A}_{X+Y}f.
In fact, for all f\in K and x\in {Z}_{+},
by an argument analogous to that used for the proof of {A}_{Y}{A}_{X}f={A}_{X+Y}f.

4.
Suppose that {A}_{{X}_{1}},{A}_{{X}_{2}},\dots ,{A}_{{X}_{n}} are the operators associated with the independent random variables {X}_{1},{X}_{2},\dots ,{X}_{n}. Then {A}_{{S}_{n}}={A}_{{X}_{1}}{A}_{{X}_{2}}\cdots {A}_{{X}_{n}} is the operator associated with the partial sum {S}_{n}={X}_{1}+{X}_{2}+\cdots +{X}_{n}.

5.
Suppose that {A}_{{X}_{1}},{A}_{{X}_{2}},\dots ,{A}_{{X}_{n}} and {A}_{{Y}_{1}},{A}_{{Y}_{2}},\dots ,{A}_{{Y}_{n}} are operators associated with independent random variables {X}_{1},{X}_{2},\dots ,{X}_{n} and {Y}_{1},{Y}_{2},\dots ,{Y}_{n}. Moreover, assume that all {X}_{i} and {Y}_{j} are independent for i,j=1,2,\dots ,n. Then, for every f\in K,
\parallel {A}_{{\sum}_{k=1}^{n}{X}_{k}}f{A}_{{\sum}_{k=1}^{n}{Y}_{k}}f\parallel \u2a7d\sum _{k=1}^{n}\parallel {A}_{{X}_{k}}f{A}_{{Y}_{k}}f\parallel .(5)
Clearly,
It deduces that

6.
It is to be noticed that \parallel {A}_{X}^{n}f{A}_{Y}^{n}f\parallel \le n\parallel {A}_{X}f{A}_{Y}f\parallel.

7.
Suppose that {X}_{1},\dots ,{X}_{n} and {Y}_{1},\dots ,{Y}_{n} are independent random variables (in each group), and let \{{N}_{n},n=1,2,\dots \} be a sequence of positive integervalued random variables independent of all {X}_{k} and {Y}_{k}, k=1,2,\dots . Then, for every f\in K,
\parallel {A}_{{\sum}_{k=1}^{{N}_{n}}{X}_{k}}f{A}_{{\sum}_{k=1}^{{N}_{n}}{Y}_{k}}f\parallel \le \sum _{n=1}^{\mathrm{\infty}}P({N}_{n}=n)\sum _{k=1}^{n}\parallel {A}_{{X}_{k}}f{A}_{{Y}_{k}}f\parallel .
Lemma 2.1 The equation {A}_{X}f(x)={A}_{Y}f(x) for f\in K, x\in {Z}_{+}, provided that X and Y are identically distributed random variables.
Let {A}_{{X}_{1}},{A}_{{X}_{2}},\dots ,{A}_{{X}_{n}},\dots be a sequence of operators associated with the independent discrete random variables {X}_{1},{X}_{2},\dots ,{X}_{n},\dots , and {A}_{X} be the operator associated with the discrete random variable X. The following lemma states one of the most important properties of the operator {A}_{X}.
Lemma 2.2 A sufficient condition for a sequence of random variables {X}_{1},{X}_{2},\dots ,{X}_{n}\dots converging in distribution to a random variable X is that
Proof Since {lim}_{n\to \mathrm{\infty}}\parallel {A}_{{X}_{n}}f{A}_{X}f\parallel =0, for all f\in K, we get
If we choose
Then
It follows that P({X}_{n}\u2a7dt)P(X\u2a7dt)\to 0 as n tends to +∞.
In other words, {X}_{n}\stackrel{d}{\to}X as n\to +\mathrm{\infty}. □
3 A bound of Poissonbinomial approximation
Let {A}_{{X}_{nk}}, k=1,\dots ,n; n=1,2,\dots be the operators associated with the random variables {X}_{nk}, k=1,\dots ,n; n=1,2,\dots , and let {A}_{{Z}_{{p}_{nk}}}, k=1,\dots ,n; n=1,2,\dots , be the operators associated with the Poisson random variables with parameters {p}_{nk}, k=1,\dots ,n; n=1,2,\dots . On the assumption that {Z}_{{\lambda}_{n}} is a Poisson random variable with a positive parameter {\lambda}_{n}={\sum}_{k=1}^{n}{p}_{nk}, we can perform that {Z}_{{\lambda}_{n}}\stackrel{d}{=}{\sum}_{k=1}^{n}{Z}_{{p}_{nk}}, where {Z}_{{p}_{n1}},{Z}_{{p}_{n2}},\dots ,{Z}_{{p}_{nn}} are independent Poisson random variables with positive parameters {p}_{n1},{p}_{n2},\dots ,{p}_{nn}, and the notation \stackrel{d}{=} denotes coincidence of distributions. We will now state an analogue of Le Cam’s inequality [5] via the linear operator in (4) as follows.
Theorem 3.1 Let ({X}_{nk}, 1\le k\le n; n=1,2,\dots) be a rowwise triangular array of independent, Bernoulli random variables with success probabilities P({X}_{nk}=1)=1P({X}_{nk}=0)={p}_{nk}, {p}_{nk}\in [0,1], k=1,2,\dots ,n; n=1,2,\dots . Let us write {S}_{n}={\sum}_{k=1}^{n}{X}_{nk} and {\lambda}_{n}={\sum}_{k=1}^{n}{p}_{nk}. We denote by {Z}_{{\lambda}_{n}} the Poisson random variable with the parameter {\lambda}_{n}. Then, for all realvalued bounded functions f\in K, we have
Proof Applying the inequality in (5), we have
Moreover, for all f\in K and for all x\in {Z}_{+}, we conclude that
Since {\sum}_{r=2}^{\mathrm{\infty}}\frac{{e}^{{p}_{nk}}{p}_{nk}^{r}}{r!}=1{e}^{{p}_{nk}}{p}_{nk}{e}^{{p}_{nk}}, for all f\in K and x\in {Z}_{+}, it may be concluded that
Therefore, applying (5), we can assert that
This completes the proof. □
Remark 3.1 According to Theorem 3.1 and assumption (1), using the definition of the norm of the operator A, we get following inequality:
The following corollaries are immediate consequences from Theorem 3.1.
Corollary 3.1 Under the stated assumptions of Theorem 3.1, for all k=0,1,2,\dots ,
Proof Choose the particular function f(x), x\in {Z}_{+}, such that
Set y=x+m. Since x,m\in {Z}^{+}, it follows that y\in {Z}_{+}. Then we have
Thus, according to Theorem 3.1, we conclude that
On the other hand, by choosing the function f(x) as above, we have
Applying (8) we can assert that
The proof is complete. □
Corollary 3.2 Let condition (1) hold. Under the hypotheses of Theorem 3.1, if moreover
then the distribution of {S}_{n} converges to the Poisson distribution with mean λ, i.e., {S}_{n}\stackrel{d}{\to}{Z}_{\lambda} as n\to \mathrm{\infty}.
Proof
The proof is based on the following observation:
According to the inequality in (6) for all f\in K and (9), we conclude that
As an argument analogous to the one used for the proof of Corollary 3.1, on account of Lemma 2.2, we get
Then, on account of (1), we have
Thus, the proof is straightforward. □
4 A bound of random Poissonbinomial approximation
Throughout this section, we begin with assuming that {N}_{n}, n=1,2,\dots , are positive integervalued random variables independent of all {X}_{nk}, k=1,2,\dots ,n; n=1,2,\dots , which are supposed to obey the relation
Here and subsequently, \stackrel{P}{\to} denotes the convergence in probability. For every n=1,2,\dots , we denote by {S}_{{N}_{n}} the random sums {S}_{{N}_{n}}={\sum}_{k=1}^{{N}_{n}}{X}_{nk} ({S}_{0}=0 by convention). Therefore, the random sums {S}_{{N}_{n}} could be said to be the random Poissonbinomial random variables. In this section, we establish Le Camtype inequalities related to the Poisson approximation for distributions of random Poissonbinomial variables. It is to be noticed that many various results concerning the random summations have already been included in the textbooks of probability theory; see, e.g., [4, 13, 14]).
Let {A}_{{X}_{n1}},{A}_{{X}_{n2}},\dots ,{A}_{{X}_{n{N}_{n}}} be operators associated with the independent triangular array of random variables {X}_{n1},{X}_{n2},\dots ,{X}_{n{N}_{n}}, and let {A}_{{Z}_{{p}_{n1}}},{A}_{{Z}_{{p}_{n2}}},\dots ,{A}_{{Z}_{{p}_{n{N}_{n}}}} be operators associated with the independent Poisson distributed random variables with positive parameters {p}_{n1},{p}_{n2},\dots ,{p}_{n{N}_{n}}. According to the properties of the linear operator in (4), we have {A}_{{S}_{{N}_{n}}}={A}_{{X}_{n1}}{A}_{{X}_{n2}}\cdots {A}_{{X}_{n{N}_{n}}} and {A}_{{Z}_{{\lambda}_{{N}_{n}}}}={A}_{{Z}_{{p}_{n1}}}{A}_{{Z}_{{p}_{n2}}}\cdots {A}_{{Z}_{{p}_{n{N}_{n}}}} are the respective operators associated with the random sums {S}_{{N}_{n}}={\sum}_{k=1}^{{N}_{n}}{X}_{nk} and {Z}_{{\lambda}_{{N}_{n}}}={\sum}_{k=1}^{{N}_{n}}{Z}_{{p}_{nk}}.
Theorem 4.1 Let ({X}_{nk}, k=1,2,\dots ,n; n=1,2,\dots) be a rowwise triangular array of independent, nonidentically distributed Bernoulli random variables with success probabilities P({X}_{nk}=1)=1P({X}_{nk}=0)={p}_{nk}, {p}_{nk}\in [0,1], k=1,2,\dots ,n; n=1,2,\dots . Moreover, we suppose that {N}_{n}, n=1,2,\dots are independent positive integervalued random variables, independent of all {X}_{nk}, k=1,2,\dots ,n; n=1,2,\dots . Then, for all realvalued bounded functions f\in K and for all x\in {Z}_{+}, we have
Proof According to the assumptions on the random variables {N}_{n}, {X}_{nk}, {Z}_{{\lambda}_{n}}, k=1,2,\dots ,n; n=1,2,\dots , we can write
and
Therefore, by an argument analogous to that used for the proof of Theorem 3.1, for all realvalued function f\in K, x\in {Z}_{+}, we have
The proof is complete. □
Note that the following remarks are immediate consequences from Theorem 4.1.
Remark 4.1 According to Theorem 4.1 and assumption (1), using the definition of the norm of the operator A, we conclude that
Remark 4.2 By an argument analogous to that used for the proof of Corollary 3.1, under the stated assumptions of Theorem 4.1, for all k=0,1,2,\dots , we have
When the success probability is identical, {p}_{nk}={p}_{n}\in [0,1], k=1,2,\dots ,n; for n=1,2,\dots , we obtain the following remark.
Remark 4.3 Suppose that the {N}_{n}, n=1,2,\dots are positive integervalued random variables independent of all independent identically distributed random variables {X}_{nk}, and assume that P({X}_{nk}=1)=1P({X}_{nk}=0)={p}_{n}\in [0,1], k=1,2,\dots ,{N}_{n}; n=1,2,\dots . Then, for all k=0,1,2,\dots , we get the following inequality:
It is worth noticing that when the positive integervalued random variables {N}_{n}, n=1,2,\dots take on the value n with probability one, i.e., P({N}_{n}=n)=1, the results concerning the probability distributions of the random sums {S}_{{N}_{n}} in the Poisson approximation in this section return to the ones in Section 3.
We conclude this paper with the following comments. The linear operator in this paper introduced by Renyi [4] essentially is a discrete form of Trotter’s operator [11] which has been used in the theory of limit theorems. The proofs of theorems in this paper by the operator method are very elementary and elegant. The received results in this article allow us to think about a new approach method to the Poisson approximation problems for the distributions of the sums of the discrete independent random variables like Poissonbinomial, geometric, and negative binomial variables.
References
Chen SX, Liu JS: Statistical applications of the Poissonbinomial and conditional Bernoulli distributions. Stat. Sin. 1997, 7: 875–892.
Tejada A, Dekker AJ: The role of Poisson’s binomial distribution in the analysis of TEM images. Ultramicroscopy 2011, 111: 1553–1556. 10.1016/j.ultramic.2011.08.010
Wang YH: On the number of successes in independent trials. Stat. Sin. 1993, 3: 295–312.
Renyi A: Probability Theory. Akad. Kiadó, Budapest; 1970.
Le Cam L: An approximation theorem for the Poisson binomial distribution. Pac. J. Math. 1960, 10(4):1181–1197. 10.2140/pjm.1960.10.1181
Barbour AD, Holst L, Janson S: Poisson Approximation. Clarendon, Oxford; 1992.
Steele JM: Le Cam’s inequality and Poisson approximations. Am. Math. Mon. 1994, 101(1):48–54. 10.2307/2325124
Chen LHY: On the convergence of Poisson binomial to Poisson distribution. Ann. Probab. 1974, 2(1):178–180. 10.1214/aop/1176996766
Neammanee K: A nonuniform bound for the approximation of Poisson binomial by Poisson distribution. Int. J. Math. Math. Sci. 2003, 2003(48):3041–3046. 10.1155/S0161171203212229
Ross SM: Introduction to Probability Models. 9th edition. Elsevier, Amsterdam; 2007.
Trotter HF: An elementary proof of the central limit theorem. Arch. Math. 1959, 10: 226–234. 10.1007/BF01240790
Hung TL: On a probability metric based on Trotter operator. Vietnam J. Math. 2007, 35(1):22–33.
Kruglov VM, Korolev VY: Limit Theorems for Random Sums. Moscow University Press, Moscow; 1990. (in Russian)
Gnedenko BV, Korolev VY: Random Summation: Limit Theorems and Applications. CRC Press, Boca Raton; 1996.
Acknowledgements
Dedicated to Professor Nguyen Duy Tien on the occasion of his 70th birthday.
The authors wish to express their gratitude to the referees for valuable remarks and comments improving the previous version of this paper. This work is supported by the Vietnam National Foundation For Science and Technology Development (NAFOSTED, Vietnam), grant 101.012010.02.
Author information
Authors and Affiliations
Corresponding author
Additional information
Competing interests
The authors declare that they have no completing interests.
Authors’ contributions
All authors read and approved the final manuscript.
Rights and permissions
Open Access This article is distributed under the terms of the Creative Commons Attribution 2.0 International License (https://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
About this article
Cite this article
Hung, T.L., Thao, V.T. Bounds for the approximation of Poissonbinomial distribution by Poisson distribution. J Inequal Appl 2013, 30 (2013). https://doi.org/10.1186/1029242X201330
Received:
Accepted:
Published:
DOI: https://doi.org/10.1186/1029242X201330
Keywords
 Poisson distribution
 Poissonbinomial random variable
 Le Cam’s inequality
 linear operator
 random sum
 convergence in distribution
 convergence in probability