Abstract
Necessary and sufficient conditions are given for the complete convergence of maximal sums of identically distributed negatively associated random variables. The conditions are expressed in terms of integrability of random variables. Proofs are based on new maximal inequalities for sums of bounded negatively associated random variables.
1. Introduction
The paper by Hsu and Robbins [1] initiated a great interest to the complete convergence of sums of independent random variables. Their research was continued by Erdös [2, 3], Spitzer [4], and Baum and Katz [5]. Kruglov et al. [6] proved two general theorems that provide sufficient conditions for the complete convergence for sums of arrays of row-wise independent random variables. In the paper of Kruglov and Volodin [7], a criterion was proved for the complete convergence of sums of independent identically distributed random variable in a rather general setting. Taylor et al. [8] and Chen et al. [9, 10] demonstrated that many known sufficient conditions for complete convergence of sums of independent random variables can be transformed to sufficient conditions for the complete convergence of sums of negatively associated random variables. Here we give necessary and sufficient conditions for the complete convergence of maximal sums of negatively associated identically distributed random variables. They resemble the criterions presented by Baum and Katz [5] and by Kruglov and Volodin [7] for the complete convergence of sums of independent identically distributed random variables. Theorems 2.3 and 2.5 are our main results. Theorem 2.3 is new even for independent random variables.
In what follows we assume that all random variables under consideration are defined on a probability space We use standard notations, in particular, denotes the indicator function of a set Recall the notion of negatively associated random variables and some properties of such random variables.
Definition 1.1. Random variables are called negatively associated if
for any pair of nonempty disjoint subsets and of the set and for any bounded coordinate-wise increasing real functions and Random variables are negatively associated if for any random variables are negatively associated.
In this definition the coordinate-wise increasing functions and may be replaced by coordinate-wise decreasing functions. Indeed, if and are coordinate-wise decreasing functions, then and are coordinate-wise increasing functions and the covariance (1.1) coincides with the covariance for and
Theorem A. Let be negatively associated random variables. Then for every the random variables are negatively associated. For every and the inequalities hold.
Proof. It can be found in Taylor et al. [8].
Theorem B. Let be negatively associated random variables. Let be independent random variables such that and are identically distributed for every Then If and for all and for some then
Proof. It can be found in Qi-Man Shao [11].
2. Main Results
Our basic theorems will be stated in terms of special functions. They were introduced in Kruglov and Volodin [7].
Definition 2.1. A nonnegative function belongs to the class for some if it is nondecreasing, is not equal to zero identically, and
The class contains all nondecreasing nonnegative functions slowly varying at infinity which are not equal to zero identically, and in particular, with The functions and with and are also in
Remark 2.2. If a nonnegative function is nondecreasing and satisfies condition (2.1), then for all greater than some and for some and
Proof. We may assume that for all greater than some From condition (2.1), it follows that Choose a number such that If then for some and Inequality (2.3) holds for all with .
Theorem 2.3. Let be negatively associated identically distributed random variables, Let be a function which is nondecreasing, is not equal to zero identically, and satisfies condition (2.1). Then the following conditions are equivalent:
Corollary 2.4. Let be negatively associated identically distributed random variables, The following conditions are equivalent:
A part of Theorem 2.3 can be generalized to a larger range of under additional restrictions on functions
Theorem 2.5. Let be negatively associated identically distributed random variables, Then the following conditions are equivalent:
Corollary 2.6. Let be negatively associated identically distributed random variables, The following conditions are equivalent:
Proof of Theorem 2.3. The theorem is obvious if the random variable is degenerate, that is, From now on we suppose that the random variable is not degenerate. Denote
where is the same as in (2.3). Define the function
where is a fixed number. By Theorem A the random variables are negatively associated. Put Note that and as If then by the dominated convergence theorem we have
Prove that (2.4)(2.5). Assume that The probability can be estimated as follows:
We intend to use Lemma 3.1 from the third part of the paper. Put From (2.13) it follows that the inequality holds for all greater than some By inequality (3.1), we get, for all
These inequalities and (2.3) imply that
Since we obtain
The last inequality holds by Lemma 3.2. From (2.1) it follows that
Condition (2.4) and inequalities (2.14)–(2.17) imply (2.5) for
Now assume that First we consider the case Note that
By Lemma 3.4, we have as and hence
for all greater than some The probability on the right-hand side can be estimated as follows:
In order to apply Lemma 3.1, put From (2.13) it follows that the inequality holds for all greater than some By inequality (3.2) we get, for
These inequalities and (2.3) imply
From this inequality and (2.17) with instead of (2.20) and (2.21), it follows that (2.5) holds for The case can be considered in the same way. In order to apply Lemma 3.1, put From (2.13) it follows that the inequality holds for all greater than some By inequality (3.2), we get, for
These inequalities and (2.3) imply
From this inequality and (2.17) with instead of (2.20) and (2.21), it follows that (2.5) holds for
Prove that (2.5)(2.6) Note that
The last series can be estimated as follows:
Condition (2.5) and these inequalities imply (2.6).
Prove that (2.6)(2.4) The sequence decreases to zero for any Indeed, if the sequence converges to a number then
Note that as
With the help of (1.2), we obtain
Denote and Note that and Since and as then and for all greater than some By the inequalities for and for , we obtain
for all , and
It follows by Lemma 3.3 that
Now we will prove that provided that Assume that Since then
and hence
But this contradicts the convergence of the series on the right-hand side of the inequality. The equality is proved.
Proof of Theorem 2.5. Both theorems overlap. We need to consider only the case Prove that (2.8)(2.9) Define random variables By Theorem A the random variables as well as are negatively associated. Denote By Theorem B there exist independent random variables such that the random variables and are identically distributed for all and Similarly to (2.21), we can prove that
for all grater than some In the same way as (2.17), one can prove that
With the help of the Markov inequality, we obtain
From (2.34) and (2.35), it follows that (2.9) holds if the series converges. This series can be estimated as follows:
Rewrite the first summand on the right-hand side in the following way:
From (2.1) and (2.2), it follows that there exist numbers and such that
For any we have
If then
With the help of these estimates, we get
The last series can be estimated as follows:
As a result we get that Taking account of (2.35) and (2.37), we see that
Prove that (2.9)(2.8) for Note that
Denote and Note that With the help of (1.2), one can prove the inequality Since and as then and for all greater than some By the inequalities for and for we obtain
By Lemma 3.3, we have that In the same way as in the proof of the previous theorem, one can prove that if
3. Auxiliary Results
Let be random variables. Denote for and for some
Lemma 3.1. If negatively associated random variables are bounded by a constant then for any number
Proof. Prove Inequality (3.1). It is easily verified that where and With the help of Markov inequality, we get
for any By Theorem A, random variables as well as are negatively associated. By Theorem B, the inequality
holds where random variables are independent, and for any random variables and are identically distributed. It follows that
Further we can proceed as in Prokhorov [12], Fuk and Nagaev [13], and Kruglov [14]. Assume that The function increases and hence With the help of this inequality, we obtain
From this inequality and from (3.5), it follows
Put As a result we obtain (3.1).
Now we assume that By the Markov inequality we get
for any By Theorem B, the inequality
holds for any Denote Note that for any With these remarks, we obtain
and hence
By the inequalities and for we obtain
Put for and if and if It can be easily verified that the function is continuous, even, and increases on Note that for all With these remarks, we obtain
and hence
In the same way, one can prove the inequality
From these inequalities and from (3.11), it follows that
Put As a result we obtain (3.2).
Lemma 3.2. Let be a nondecreasing nonnegative function, be a nonnegative random variable, Then
Proof. It can be found in Kruglov and Volodin [7].
Lemma 3.3. Let be a nondecreasing nonnegative function possessing property (2.1), be a nonnegative random variable, and an unbounded nondecreasing sequence of positive numbers such that for all and for some number Then there exist numbers and such that
Proof. It can be found in Kruglov and Volodin [7].
The next lemma was proved in Kruglov and Volodin [7] under an additional restriction.
Lemma 3.4. Let be identically distributed random variables such that for some and if Define the function where is a fixed number. Then
Proof. Note that
It suffices to prove that
Suppose that For any there exist such that We get, for any
and hence
This implies (3.21), since can be chosen arbitrarily small.
If and then we get
Acknowledgments
The author is grateful to an anonymous referee for careful reading of the paper and useful remarks. This research was supported by the Russian Foundation for Basic Research, projects 08-01-00563-a, 08-01-00567-a, 08-01-90252 Uzbekistan.