R E S E A R C H
Open Access
Note on the strong convergence of a
weighted sum
Xiaochun Li
*, Yu Miao and Xiaoman Zhang
*Correspondence:
L.xiaochun@tom.com College of Mathematics and Information Science, Henan Normal University, Xin Xiang, China
Abstract
In this paper, we consider an interesting weighted sums
μ
n:=(log1n)1–pn
j=2
Xn+2–j
j(logj)p, 0 <p< 1, where{X,Xn,n≥1}is a sequence of independent and identically
distributed random variables withEX= 0, and the equivalence of the almost sure and complete convergence of
μ
nis proved.Keywords: strong convergence; complete convergence; weighted sums
1 Introduction
Hsu and Robbins [] introduced the concept of complete convergence as follows: a se-quence{Yn,n≥}converges completely to the constantcif
∞
n=
P|Yn–c|>ε
<∞ for allε> .
They also show that the sequence of arithmetic means of i.i.d. random variables converges completely to the expected value if the variance of the summands is finite. The converse was proved by Erdös [, ]. Furthermore, ifYn→ccompletely, then the Borel-Cantelli lemma trivially implies thatYn
a.s.
−→c, almost sure convergence asn→ ∞. The converse statement is generally, or even ‘typically’, not true.
There are numerous publications in the literature studying the almost sure convergence and complete convergence for the weighted sums of a sequence of random variables. Gut [] provided necessary and sufficient conditions for the complete convergence of the Cesáro means of i.i.d. random variables. Liet al.[] obtained some results on complete convergence for weighted sums of independent random variables. Cuzick [] proved a strong law for weighted sums of i.i.d. random variables. Miao and Xu [] established a general result for the weighted sums of stationary sequence.
Throughout this paper, assume that {X,Xn,n≥}is a sequence of independent and identically distributed random variables withEX= . Chow and Lai [] established the following result.
Theorem .[, Theorem ] Let X,X,X, . . .be i.i.d.random variables such thatEX= .
Forα> ,the following statements are equivalent: . Eexp(t|X|α) <∞ ∀t> ;
. limn→∞(logn)–/αX
n= a.e.;
. limn→∞(logn)–/αni=cn–iXi= a.e.for some(or equivalently for every)nonvoid
sequence of real numbers(cn,n≥)such thatcn=O(n–β),whereβ=α/(α+ ).
As an interesting particular case of Theorem ., we consider the following weighted sums:
μn:= (logn)–p
n
j=
Xn+–j
j(logj)p, <p< .
The aim of this paper is to prove the following.
Theorem . We haveμn
a.s.
−→if and only ifμn→completely.
Remark . For the casep= , Wu [] discussed the following weighted sums:
νn:=
log logn
n
j=
Xn+–j
jlogj
and proved the equivalence of the almost sure and complete convergence of the se-quenceνn. On the one hand, because of the limitation ofαin Theorem ., we here only discuss the case ofp> . On the other hand, in order to prove the equivalence of the al-most sure and complete convergence ofμnfor the casep= , we need the exponential integrability ofX, but from Theorem ., this does not hold.
LetSj=
j
k=
Xk
k(logk)p; then we have the following.
Corollary . Ifμn
a.s. −→,then
(logn)–pmax≤j≤nSj→ completely. 2 Proofs of main results
Let [·] denote the usual integer part of ‘·’ and assume thatn>n= [e]. The constantC
in the proofs below depends only on the distribution of the underlying random variableX
and may denote different quantities at different appearances.
2.1 Proof of Theorem 1.2
The proof of Theorem . can be derived from Lemma . and Lemma ..
Lemma . The following estimate holds: n
j=[logn]
log
+C
logn j(logj)p
<Clogn.
Proof Sinceg(j) =log( +C(j(loglogjn)p)) decreases inj,
n
j=[logn]
log
+C
logn j(logj)p
≤
n
logn
log
+C
logn x(logx)p
≤logn
n
logn
log
+C
t(logt+log logn)p
dt+O()
≤logn
∞
log
+C
t(logt+log logn)p
dt+O()
≤Clogn,
which yields the desired result.
Lemma . Ifμn
a.s. −→,then
μ()n := (logn)–p
n
j=[logn]
Xn+–j
j(logj)p →
completely and
(logn)–p
[log[nlogn]]
j=[logn]+
Xn+–j
j(logj)p →
completely.
Proof From Theorem ., we knowEe|X|–p
<∞, which impliesEe|X|<∞. Hence by the
elementary inequality
eηx– –ηx ≤ηe|x| for allη∈[, ],
we have
EeηX≤ +ηEe|X|= +ηC.
Notice that ifn>nandn≥j≥[logn], then j(lognlogj)p < . Hence, by Lemma ., for any
ε> ,
∞
n=n
Pμ()n >ε≤
∞
n=n
n–ε(logn)(–p)
Eelogn
n j=[ logn]
Xn+–j j(logj)p
=
∞
n=n
n–ε(logn)(–p) n j=[logn]
Ee
logn j(logj)pXn+–j
≤ ∞
n=n
n–ε(logn)(–p) n j=[logn]
+C
logn j(logj)p
≤ ∞
n=n
n–ε(logn)(–p)eClogn
Similarly, we can obtain∞n=nP(μ()n < –ε) <∞, then
∞
n=nP(|μ
()
n |>ε) <∞. The first statement now follows if we combine the two inequalities. The same technique yields the
second statement.
Lemma . If
μ()n := (logn)–p
[logn]
j=
Xn+–j
j(logj)p
a.s. −→,
thenμ()n →completely.
Proof For anyε> we have =Plim sup
n→∞
μ()n >ε
≥Plim sup
m→∞
μ()n(m) >ε
,
wheren(m) = [mlogm],m∈N. Since
n(m+ ) + –
logn(m+ ) >n(m)
for m> , the random variables μ()n(m), m= , , . . . are independent. Notice also that (logn(m))–p< (logm)–p. So by the Borel-Cantelli lemma and the second statement of Lemma ., we have∞m=nP(|μ()n(m)|>ε) <∞, proving the lemma.
2.2 Proof of Corollary 1.1
We easily see that
ESn=EX
n
j=
(j(logj)p) <C<∞.
Now we apply the inequality
Pmax
≤j≤nSj>η
≤PSn>η–
ES
n
(cf.Chow and Teicher [], p.) by takingη= (logn)–pεforε> . Sinceη–ES
n>η
for allnsufficiently large,
P
max
≤j≤n
Sj (logn)–p >ε
≤P
(logn)–p
n
j=
Xn+–j
j(logj)p ≥ ε
.
Therefore,
∞
n=n
P
max
≤j≤n
Sj (logn)–p >ε
≤
∞
n=n
P
μn>
ε
<∞.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
All authors contributed equally to the manuscript, read and approved the final manuscript.
Acknowledgements
This work is supported by HASTIT (No. 2011HASTIT011), NSFC (No. 11001077), NCET (NCET-11-0945), and Plan For Scientific Innovation Talent of Henan Province (124100510014).
Received: 11 January 2014 Accepted: 21 April 2014 Published:12 May 2014
References
1. Hsu, PL, Robbins, H: Complete convergence and the law of large number. Proc. Natl. Acad. Sci. USA33, 25-31 (1947) 2. Erdös, P: On a theorem of Hsu and Robbins. Ann. Math. Stat.20, 286-291 (1949)
3. Erdös, P: Remark on my paper ‘On a theorem of Hsu and Robbins’. Ann. Math. Stat.21, 138 (1950)
4. Gut, A: Complete convergence and Cesáro summation for i.i.d. random variables. Probab. Theory Relat. Fields97, 169-178 (1993)
5. Li, DL, Rao, MB, Jiang, TF, Wang, XC: Complete convergence and almost sure convergence of weighted sums of random variables. J. Theor. Probab.8, 49-76 (1995)
6. Cuzick, J: A strong law for weighted sums of i.i.d. random variables. J. Theor. Probab.8, 625-641 (1995) 7. Miao, Y, Xu, SF: Almost sure convergence of weighted sum. Miskolc Math. Notes14, 173-181 (2013)
8. Chow, YS, Lai, TL: Limiting behavior of weighted sums of independent random variables. Ann. Probab.1, 810-824 (1973)
9. Wu, WB: On the strong convergence of a weighted sum. Stat. Probab. Lett.1, 19-22 (1999) 10. Chow, YS, Teicher, H: Probability Theory, 3rd edn. Springer, New York (1978)
10.1186/1029-242X-2014-179
Cite this article as:Li et al.:Note on the strong convergence of a weighted sum.Journal of Inequalities and Applications