Open Journal of Statistics, 2013, 3, 42-46
http://dx.doi.org/10.4236/ojs.2013.31006 Published Online February 2013 (http://www.scirp.org/journal/ojs)
Strong Law of Large Numbers for a 2-Dimensional Array
of Pairwise Negatively Dependent Random Variables
Karn Surakamhaeng1, Nattakarn Chaidee1,2, Kritsana Neammanee1
1
Department of Mathematics and Computer Science, Chulalongkorn University, Bangkok, Thailand
2
Centre of Excellence in Mathematics, CHE, Bangkok, Thailand
Email: nattakarn.c@chula.ac.th
Received October 30, 2012; revised November 30, 2012; accepted December 14, 2012
ABSTRACT
In this paper, we obtain the strong law of large numbers for a 2-dimensional array of pairwise negatively dependent
random variables which are not required to be identically distributed. We found the sufficient conditions of strong law
of large numbers for the difference of random variables which independent and identically distributed conditions are
regarded. In this study, we consider the limit as m n which is stronger than the limit as m, n when m, n
are natural numbers.
Keywords: Strong Law of Large Numbers; Negatively Dependent; 2-Dimensional Array of Random Variables
1. Introduction and Main Results
Let X i iN be a sequence of random variables. We say
that X i iN satisfies the strong law of large numbers
(SLLN) if there exist sequences of real numbers an nN
S an a.s.
and bn nN such that n
0 as n .
bn
Theorem 1.1. (Birkel, [4]) Let X i iN be a sequence of pairwise PD random variables with finite
variances. Assume
iN
2)
i 1
almost surely.
To study the strong law of large numbers, there is a
simple question come in mind. When does the sequence
X i iN satisfy the SLLN? Many conditions of the sequence X i iN have been found for this question. The
SLLN are investigated extensively in the literature especially to the case of a sequence of independent random
variables (see for examples in [1-3]). After concepts of
dependence was introduced, it is interesting to study the
SLLN with condition of dependence.
A sequence X i iN of random variables is said to be
pairwise positively dependent (pairwise PD) if for any
a, b R and i j ,
P X i a, X j b P X i a P X j b
and it is said to be pairwise negatively dependent (pairwise ND) if for any a, b R and i j ,
P X i a, X j b P X i a P X j b .
Theorem 1.1-1.5 are examples of SLLN for a sequence
of pairwise PD and pairwise ND random variables.
Copyright © 2013 SciRes.
n
Cov X i , Si
i 1
i2
n
where Sn X i and the abbreviation a.s. stands for
1) sup E X i E X i ,
.
Sn E Sn
a.s.
0 as n .
n
Theorem 1.2. (Azarnoosh, [5]) Let X i iN be a sequence of pairwise ND random variables with finite
variances. Assume
Then
1) sup E X i ,
iN
2)
n
Var X i
i 1
i2
.
Sn E Sn
a.s.
0 as n .
n
Theorem 1.3. (Nili Sani, Azarnoosh and Bozorgnia,
[6]) Let an nN be a positive and increasing sequence
such that an as n .
Let X i iN be a sequence of pairwise ND random
variables with finite variances such that
Then
n E Xi E Xi
1) sup
an
nN i 1
2)
Var X i
i 1
ai2
,
.
OJS
K. SURAKAMHAENG
Sn E Sn
Then
i , j N
quences of real numbers
Sm , n am , n
bm, n
m
a
m , n m , nN
a.s.
0
and
b
m , n m , nN
as m, n
where
n
Sm, n X i , j .
i 1 j 1
43
AL.
n
a.s.
0 as n .
an
In this work, we study the SLLN for a 2-dimensional
array of pairwise ND random variables. We say that
X i, j satisfies the SLLN if there exist double se-
such that
ET
m
Wm, n ai , j X i , j .
i 1 j 1
Observe that, for a double indexed sequence of real
number am , n m , nN , the convergence as m n
implies the convergence as m, n . However, a dou-
a
ble sequence
i , j N
positively dependent (pairwise PD) if for any a, b R
and i, j k , l ,
P X i , j a, X k ,l b P X i , j a P X k ,l b
Theorem 1.4. (Kim, Beak and Seo, [7]) Let
be a 2-dimensional array of pairwise PD
X i, j
i , j N
random variables with finite variances. Assume
P X i , j a, X k ,l b P X i , j a P X k ,l b .
The followings are SLLNs for a 2-dimensional array
of pairwise ND random variables which are all our results.
Theorem 1.6. Let am mN and bn nN be increasing sequences of positive numbers such that am , bn e
which am as m and bn as n .
be a 2-dimensional array of pairwise
Let X i , j
i , jN
ND random variables with finite variances. If there exist
real numbers p, q such that
Sm, n E Sm, n
mn
i 1 j 1
i , j i , j N
p
cm, n
i , j N
be a 2-dimensional array of pairwise
ai , j ak ,l Cov X i , j , X k ,l
.
2)
bi2, j
i , j ,i j 1 k , l , k l i j
Wm, n E Wm, n
bm, n
Copyright © 2013 SciRes.
,
c
m , n m , nN
such that
a.s.
0 as m n where
a.s.
0 as m n .
The next theorem is the SLLN for the difference of
random variables which independent and identically distributed conditions are regarded.
Theorem 1.7. Let X i , j
and Yi , j
be 2i , jN
i , jN
dimensional arrays of random variables on a probability
space (Ω, F, P). If
1) sup E X i , j E X i , j ,
Then
q
q
Sm, n E Sm,n
PD random variables with finite variances such that
i , j N
p
ai2 b j2
cm, n am2 bn2 for every m, n N ,
a.s.
0 as m n .
Var X i , j
then for any double sequence
be a 2-dimensional array of positive numbers and
n m
am , n
bm, n ai , j such that
0 and bm, n as
bm , n
i 1 j 1
m, n .
X
Cov X i , j , X k ,l
.
2
i , j , i j 1 k , l , k l i j
i j
Theorem 1.5. (Kim, Baek and Han, [8]) Let ai , j
Let
m n
negtively dependent (pairwise ND) if for any a, b R
and i, j k , l ,
Then
mn
mn
shows us that the converse is not true in general.
Our goal is to obtain the SLLN for 2-dimensional array of random variables in case of pairwise ND.
A double sequence X i , j
is said to be pairwise
m , n m , n N
1) sup E X i , j E X i , j ,
2)
1
i , j N
In 1998, Kim, Beak and Seo investigated SLLN for a
2-dimensional array of pairwise PD random variables
and it was generalized to a case of weighted sum of
2-dimensional array of pairwise PD random variables by
Kim, Baek and Han in one year later. The followings are
their results.
A double sequence X i , j
is said to be pairwise
i , j N
where am, n
P X i , j Yi , j ,
i 1 j 1
then
1 m n
a.s.
0
X i , j Yi , j
m n i 1 j 1
as m n .
OJS
44
K. SURAKAMHAENG
Corollary 1.8 and Corollary 1.9 follow directly from
k
Theorem 1.6 by choosing cm, n am bn and
cm, n am bn where am 3m and bn 3n with p = q =
4, respectively.
Corollary 1.8. Let am mN and bn nN be increasing sequences of positive numbers such that am , bn e
which am as m and bn as n .
Let X i , j
be a 2-dimensional array of pairwise
ET
AL.
max m, n .
The following proposition is a Borel-Cantelli lemma
for a sequence of double indexed events
Proposition 2.2. Let Ei , j
be a double sequence
i , j N
of events on a probability space
Var X i , j
p
2
i 1 j 1
ai b
q
2
j
,
am bn
k
E
i, j
i 1 j 1
k k
P Ei , j
0 as m n .
a.s.
i , j i , j N
Var X i , j
i j
i 1 j 1
be a 2-dimensional ar-
2
j 1
i , j ,i j k
,
equal
2
k
i 1, j i , j 0, i , j 1 i , j 0
i 1, j 1 i 1, j i , j 1 i , j 0,
and i , j as max i, j .
be a double sequence of real numbers.
Let ai , j
i , j N
Assume that
a
i, j
,
1)
i , j
k 1
for every i N and
every j N . Then
1
m,n
Copyright © 2013 SciRes.
m
n
ai , j 0
i 1 j 1
ak , j
k 1
i ,k
i 1
j 1
i , j ,i j k
P Ei , j
as
k, j
for
k
i 1 j 1
k
double sequence of positive numbers such that for all
i, j N ,
a
P Ei , j klim
Therefore lim
a.s.
0 as m n .
i , j N
i 1 j 1
k k
k
In this section, we present some materials which will be
used in obtaining the SLLN’s in the next section.
Proposition 2.1. (Móricz, [9]) Let i , j
be a
i ,k
k
lim P Ei , j L.
2. Auxiliary Results
i 1 j 1
k
k and hence
k
81 m n
P Ei , j P Ei , j
where k denote the greatest integer smaller than or
L lim
then
Sm, n E Sm,n
First note that
X
i.o. Ei , j .
k 1 i , j , i j k
ray of pairwise ND random variables with finite variances. If
2)
where
i 1
Corollary 1.9. Let
i 1 j 1
Proof. Let L R be such that L P Ei , j .
then for any k p q,
Sm, n E Sm, n
P Ei , j P Ei , j i.o. 0
i , j N
ND random variables with finite variances. If there exist
p, q N such that
, F , P . Then
i , j ,i j k
P Ei , j L and
P Ei , j i.o. lim P Ei , j lim P Ei , j
k
i , j ,i j k
k i , j ,i j k
lim P Ei , j P Ei , j 0.
k
i , j ,i j k 1
i 1 j 1
This completes the proof. □
3. Proof of Main Results
Proof of Theorem 1.6
Let m, n N and define f m ln am and
g n ln bn .
Clearly, f and g are increasing whose facts
f m ln am f m 1 and g n ln bn g n 1
f m
f m 1
which imply that e am e and
g n
g n 1
bn e
.
e
Let 0 be given. By using the fact that
Cov X i , j , X k ,l 0 for i, j k , l ([10], p. 313), we
have
m n
m n
Var X i , j Var X i , j .
i 1 j 1
i 1 j 1
From this fact and Chebyshev’s inequality, we have
OJS
K. SURAKAMHAENG
C
Var Sm, n
m 1 n 1
2
m,n
c
m 1 n 1
C Var X i , j
i 1 j 1
C
m
n
2
m , n i 1 j 1
c
1
C Var X i , j
i 1 j 1
e
For each i, j N , let
t N : e
Bj
f s 1
f t 1
.
b
m i n j
i 1 j 1
C Var X i , j
i 1 j 1
C Var X i , j
i 1 j 1
m f i
1
e
e
and for every j N ,
1
1
(3.5)
e
1 m n
a.s.
0
X i , j Yi , j
m n i 1 j 1
pm
2 n g j
as max m, n . We here note that am,n a as
max m, n implies am,n a as m n . Hence
1
qn
2
1 m n
a.s.
0
X i , j Yi , j
m n i 1 j 1
(3.2)
1
e
f i
e
1
and
ai
as m n .
To prove (3.3), (3.4) and (3.5), let c0 . Then
there exists k N such that for i, j N ,
i j k X i , j Yi , j .
Thus for each c0 , X i , j
1
.
bj
Sm, n E Sm, n
Var X
i, j
C p
.
q
cm, n
i 1 j 1
2
2
ai b j
By Proposition 2.2 with
Sm, n E Sm, n
,
Em , n
cm, n
Copyright © 2013 SciRes.
(3.4)
j 1
pm qn
2
P
m 1 n 1
1
From (3.3), (3.4) and (3.5), we can apply Proposition
2.1 with i , j i j that
From this facts and (3.2) together with our assumption
2), we have
(3.3)
i 1
Since i Ai and j B j , we have
1
i j X i , j Yi, j
e
e
1 1
C Var X i , j pf i qg j .
i 1 j 1
2 2
e
e
g j
1
i j X i , j Yi , j .
m f i n g j
for every i N ,
j
pf m qg n
Yi , j i.o. 1.
i 1 j 1
1
e
i, j
i j X i , j Yi , j ,
ai
Sm, n E Sm, n
P
cm, n
m 1 n 1
X
For every c0 , we will show that
C Var X i , j
P c0 1 P 0 1 P
(3.1)
and i min Ai and j min B j . Since i Ai and
j B j , we have i i and j j . From this facts and
(3.1), we have
By Proposition 2.2,
we have
pf m qg n
Ai s N : e
□
k 1 i , j , i j k
1
m i n j
a.s.
0 as m n .
X i , j Yi , j .
Let 0
1
p
q
a
m i n j m bn
cm, n
Proof of Theorem 1.7
c
C Var X i , j
i 1 j 1
Sm, n E Sm, n
Var X i , j
1
2
m,n
m i n j
45
AL.
we have P Em, n i.o. 0 and this hold for every 0.
By using the same idea with Theorem 4.2.2 ([11], p. 77),
we can prove that
Sm, n E Sm, n
P
cm, n
m 1 n 1
ET
Y
i, j
i , j N
(3.6)
and
are different only finitely many terms.
i , j N
This implies that (3.3) holds.
For fixed i N , we can find a large j0 N such
that (3.6) holds for all j j0 which means that there
are only finitely many different terms of X i , j
and Yi , j
i , j N
i , j N
. So for fixed i N ,
i j X i , j Yi, j .
1
j 1
OJS
46
K. SURAKAMHAENG
Similarly, for fixed j N ,
1
i j X i , j Yi , j .
i 1
Now (3.4) and (3.5) are now proved and this ends the
proof. □
Remark 3.1. In case of m fixed and n , by considering the limit as m n , we also obtain the
corresponding results for a case of 1-dimensional pairwise ND random variables.
4. Example
Example 4.1 A box contains pq balls of p different colors and q different sizes in each color. Pick 2 balls randomly.
Let X i , j , i 1, 2, , p and j 1, 2, , q be a random
variable indicating the presence of a ball of the ith color
and the jth size such that
X i , j
X ,if 1 i p and 1 j q,
X i, j i, j
0, otherwise.
Proof. By a direct calculation, we have X i , j ’s are
pairwise ND random variables, i.e. for i, j , k , l R that
i, j k , l and a, b R,
pq 1 2
E X i, j
pq pq
2
Var X i , j
i 1 j 1
i j
2
m n Var X i , j
lim
2
m , n
i 1 j 1 i j
2
4
1
.
2
pq pq 2
i 1 j 1 i j
By applying Theorem 1.6, for any double sequence
c
m , n m , nN
such that cm,n 81 m n
Copyright © 2013 SciRes.
2
cm, n
a.s.
0 as m n .
5. Acknowledgements
The authors would like to thank referees for valuable
comments and suggestions which have helped improving
our work. The first author gives an appreciation and
thanks to the Institute for the Promotion of Teaching
Science and Technology for financial support.
REFERENCES
[1]
S. Csörgő, K. Tandori and V. Totik, “On the Strong Law
of Large Numbers for Pairwise Independent Random
Variables,” Acta Mathematica Hungarica, Vol. 42, No. 34, 1983, pp. 319-330. doi:10.1007/BF01956779
[2]
N. Etemadi, “An Elementary Proof of the Strong Law of
Large Numbers,” Zeitschrift für Wahrscheinlichkeitstheorie und Verwandte Gebiete, Vol. 55, No. 1, 1981, pp.
119-122.
[3]
R. G. Laha and V. K. Rohatgi, “Probability Theory,” John
Wiley & Sons, Hoboken, 1979.
[4]
T. Birkel, “A Note on the Strong Law of Large Numbers
for Positively Dependent Random Variables,” Statistics &
Probability Letters, Vol. 7, No. 1, 1989, pp. 17-20.
doi:10.1016/0167-7152(88)90080-6
[5]
H. A. Azarnoosh, “On the Law of Large Numbers for Negatively Dependent Random Variables,” Pakistan Journal
of Statistics, Vol. 19, No. 1, 2003, pp. 15-23.
[6]
H. R. Nili Sani, H. A. Azarnoosh and A. Bozorgnia, “The
Strong Law of Large Numbers for Pairwise Negatively
Dependent Random Variables,” Iranian Journal of Science & Technology, Vol. 28, No. A2, 2004, pp. 211-217.
[7]
T. S. Kim, H. Y. Beak and H. Y. Seo, “On Strong Laws
of Large Numbers for 2-Dimensional Positively Dependent Random Variables,” Bulletin of the Korean Mathematical Society, Vol. 35, No. 4, 1998, pp. 709-718.
[8]
T. S. Kim, H. Y. Beak and K. H. Han, “On the Almost
Sure Convergence of Weighted Sums of 2-Dimensional
Arrays of Positive Dependent Random Variables,” Communications of the Korean Mathematical Society, Vol. 14,
No. 4, 1999, pp. 797-804.
[9]
F. Móricz, “The Kronecker Lemmas for Multiple Series
and Some Applications,” Acta Mathematica Academiae
Scientiarum Hungaricae, Vol. 37, No. 1-3, 1981, pp. 3950. doi:10.1007/BF01904871
2
4
and Var X i , j
.
pq pq 2
Hence,
Sm, n E Sm, n
□
P X i , j a, X k , l b P X i , j a P X k , l b .
Note that
AL.
n N , we have
1,if the i th color and the j th size of ball is picked,
0, otherwise.
For i, j N , let X i , j be a random variable defined
by
ET
[10] N. Ebrahimi and M. Ghosh, “Multivariate Negative Dependence,” Communications in Statistics—Theory and
Methods, Vol. A10, No. 4, 1981, pp. 307-337.
[11] K. L. Chung, “A Course in Probability Theory,” Academic Press, London, 2001.
for every m
OJS