Let {Xni,i≥1,n≥1} be an array of rowwise ρ˜-mixing random
variables. Some strong law of large numbers for arrays of rowwise ρ˜-mixing
random variables is studied under some simple and weak conditions.
1. Introduction
Let {X,Xn,n≥1} be a sequence of independent and identically distributed random variables. The Marcinkiewicz-Zygmund strong law of large numbers states that 1nα∑i=1n(Xi-EXi)⟶0a.s.for1≤α<2,1nα∑i=1nXi⟶0a.s.for0<α<1
if and only if E|X|α<∞. In the case of independence, Hu and Taylor [1] proved the following strong law of large numbers.
Theorem 1.1.
Let {Xni:1≤i≤n,n≥1} be a triangular array of rowwise independent random variables. Let {an,n≥1} be a sequence of positive real numbers such that 0<an↑∞. Let g(t) be a positive, even function such that g(|t|)/|t|p is an increasing function of |t| and g(|t|)/|t|p+1 is a decreasing function of |t|, respectively, that is,
g(|t|)|t|p↑,g(|t|)|t|p+1↓,as|t|↑
for some nonnegative integer p. If p≥2 and
EXni=0,∑n=1∞∑i=1nEg(|Xni|)g(an)<∞,∑n=1∞(∑i=1nE(Xnian)2)2k<∞,
where k is a positive integer, then
1an∑i=1nXni⟶0a.s.asn⟶∞.
Zhu [2] generalized and improved the result of Hu and Taylor [1] for triangular arrays of rowwise independent random variables to the case of arrays of rowwise ρ̃-mixing random variables as follows.
Theorem 1.2.
Let {Xni:i≥1,n≥1} be an array of rowwise ρ̃-mixing random variables. Let {an,n≥1} be a sequence of positive real numbers such that 0<an↑∞. Let g(t) be a positive, even function such that g(|t|)/|t| is an increasing function of |t| and g(|t|)/|t|p is a decreasing function of |t|, respectively, that is,
g(|t|)|t|↑,g(|t|)|t|p↓,as|t|↑
for some nonnegative integer p. If p≥2 and
EXni=0,∑n=1∞∑i=1nEg(|Xni|)g(an)<∞,∑n=1∞(∑i=1nE(Xnian)2)v/2<∞,
where v is a positive integer, v≥p, then
∀ε>0,∑n=1∞P(max1≤j≤n|1an∑i=1jXni|>ε)<∞.
In the following, we will give the definitions of a ρ̃-mixing sequence and the array of rowwise ρ̃-mixing random variables.
Let {Xn,n≥1} be a sequence of random variables defined on a fixed probability space (Ω,ℱ,P). Write ℱ𝒮=σ(Xi,i∈S⊂ℕ). For any given σ-algebras ℬ, ℛ in ℱ, let ρ(B,R)=supX∈L2(B),Y∈L2(R)|EXY-EXEY|(VarXVarY)1/2.
Define the ρ̃-mixing coefficients by ρ̃(k)=sup{ρ(FS,FT):S,TarefinitesubsetsofNsuchthatdist(S,T)≥k},k≥0.
Obviously, 0≤ρ̃(k+1)≤ρ̃(k)≤1 and ρ̃(0)=1.
Definition 1.3.
A sequence {Xn,n≥1} of random variables is said to be a ρ̃-mixing sequence if, there exists k∈ℕ such that ρ̃(k)<1.
An array of random variables {Xni,i≥1,n≥1} is called rowwise ρ̃-mixing if for every n≥1, {Xni,i≥1} is a ρ̃-mixing sequence of random variables.
The ρ̃-mixing random variables were introduced by Bradley [3], and many applications have been found. ρ̃-mixing is similar to ρ-mixing, but both are quite different. Many authors have studied this concept providing interesting results and applications. See, for example, Zhu [2], An and Yuan [4], Kuczmaszewska [5], Bryc and Smoleński [6], Cai [7], Gan [8], Peligrad [9, 10], Peligrad and Gut [11], Sung [12], Utev and Peligrad [13], Wu and Jiang [14], and so on. When these are compared with the corresponding results of independent random variable sequences, there still remains much to be desired.
The main purpose of this paper is to further study the strong law of large numbers for arrays of rowwise ρ̃-mixing random variables. We will introduce some simple conditions to prove the strong law of large numbers. The techniques used in the paper are inspired by Zhu [2].
2. Main Results
Throughout the paper, let I(A) be the indicator function of the set A. C denotes a positive constant which may be different in various places.
The proofs of the main results of this paper are based upon the following lemma.
Lemma 2.1 (Utev and Peligrad [13, Theorem 2.1]).
Let {Xn,n≥1} be a ρ̃-mixing sequence of random variables, EXi=0, E|Xi|p<∞ for some p≥2 and for every i≥1. Then, there exists a positive constant C depending only on p such that
E(max1≤j≤n|∑i=1jXi|p)≤C{∑i=1nE|Xi|p+(∑i=1nEXi2)p/2}.
As for arrays of rowwise ρ̃-mixing random variables {Xni:i≥1,n≥1}, we assume that the constant C from Lemma 2.1 is the same for each row throughout the paper. Our main results are as follows.
Theorem 2.2.
Let {Xni:i≥1,n≥1} be an array of rowwise ρ̃-mixing random variables and let {an,n≥1} be a sequence of positive real numbers. Let {gn(t),n≥1} be a sequence of positive, even functions such that gn(|t|) is an increasing function of |t| and gn(|t|)/|t| is a decreasing function of |t| for every n≥1, respectively, that is,
gn(|t|)↑,gn(|t|)|t|↓,as|t|↑.
If
∑n=1∞∑i=1nEgn(|Xni|)gn(an)<∞,
then, for any ε>0,
∑n=1∞P(max1≤j≤n|1an∑i=1jXni|>ε)<∞.
Proof.
For fixed n≥1, define
Xi(n)=XniI(|Xni|≤an),i≥1,Tj(n)=1an∑i=1j(Xi(n)-EXi(n)),j=1,2,…,n.
It is easy to check that for any ε>0,
(max1≤j≤n|1an∑i=1jXni|>ε)⊂(max1≤i≤n|Xni|>an)⋃(max1≤j≤n|1an∑i=1jXi(n)|>ε),
which implies that
P(max1≤j≤n|1an∑i=1jXni|>ε)≤P(max1≤i≤n|Xni|>an)+P(max1≤j≤n|1an∑i=1jXi(n)|>ε)≤∑i=1nP(|Xni|>an)+P(max1≤j≤n|Tj(n)|>ε-max1≤j≤n|1an∑i=1jEXi(n)|).
Firstly, we will show that
max1≤j≤n|1an∑i=1jEXi(n)|⟶0,asn⟶∞.
Actually, by conditions gn(|t|)/|t|↓ as |t|↑ and (2.3), we have that
max1≤j≤n|1an∑i=1jEXi(n)|=max1≤j≤n|1an∑i=1jEXniI(|Xni|≤an)|≤1an∑i=1nE|Xni|I(|Xni|≤an)≤∑i=1nEgn(|Xni|)I(|Xni|≤an)gn(an)≤∑i=1nEgn(|Xni|)gn(an)⟶0,asn⟶∞,
which implies (2.8). It follows from (2.7) and (2.8) that for n large enough,
P(max1≤j≤n|1an∑i=1jXni|>ε)≤∑i=1nP(|Xni|>an)+P(max1≤j≤n|Tj(n)|>ε2).
Hence, to prove (2.4), we only need to show that
∑n=1∞∑i=1nP(|Xni|>an)<∞,∑n=1∞P(max1≤j≤n|Tj(n)|>ε2)<∞.
The conditions gn(|t|)↑ as |t|↑ and (2.3) yield that
∑n=1∞∑i=1nP(|Xni|>an)≤∑n=1∞∑i=1nEgn(|Xni|)gn(an)<∞,
which implies (2.11).
By Markov's inequality, Lemma 2.1 (for p=2), Cr's inequality, gn(|t|)/|t|↓ as |t|↑ and (2.3), we can get that∑n=1∞P(max1≤j≤n|Tj(n)|>ε2)≤C∑n=1∞E(max1≤j≤n|Tj(n)|2)≤C∑n=1∞1an2∑i=1nE|Xi(n)-EXi(n)|2≤C∑n=1∞1an2∑i=1nE|Xi(n)|2=C∑n=1∞∑i=1nE|Xni|2I(|Xni|≤an)an2≤C∑n=1∞∑i=1nE|Xni|I(|Xni|≤an)an≤C∑n=1∞∑i=1nEgn(|Xni|)I(|Xni|≤an)gn(an)≤C∑n=1∞∑i=1nEgn(|Xni|)gn(an)<∞,
which implies (2.12). This completes the proof of the theorem.
Corollary 2.3.
Under the conditions of Theorem 2.2,
1an∑i=1nXni⟶0a.s.asn⟶∞.
Theorem 2.4.
Let {Xni:i≥1,n≥1} be an array of rowwise ρ̃-mixing random variables and let {an,n≥1} be a sequence of positive real numbers. Let {gn(t),n≥1} be a sequence of nonnegative, even functions such that gn(|t|) is an increasing function of |t| for every n≥1. Assume that there exists a constant δ>0 such that gn(t)≥δt for 0<t≤1. If
∑n=1∞∑i=1nEgn(Xnian)<∞,
then for any ε>0, (2.4) holds true.
Proof.
We use the same notations as that in Theorem 2.2. The proof is similar to that of Theorem 2.2.
Firstly, we will show that (2.8) holds true. In fact, by the conditions gn(t)≥δt for 0<t≤1 and (2.16), we have thatmax1≤j≤n|1an∑i=1jEXi(n)|=max1≤j≤n|1an∑i=1jEXniI(|Xni|≤an)|≤∑i=1nE(|Xni|anI(|Xni|≤an))≤1δ∑i=1nEgn(Xnian)I(|Xni|≤an)≤1δ∑i=1nEgn(Xnian)⟶0,asn⟶∞,
which implies (2.8).
According to the proof of Theorem 2.2, we only need to prove that (2.11) and (2.12) hold true.
When |Xni|>an>0, we have gn(Xni/an)≥gn(1)≥δ, which yields thatP(|Xni|>an)=EI(|Xni|>an)≤1δEgn(Xnian).
Hence,
∑n=1∞∑i=1nP(|Xni|>an)≤1δ∑n=1∞∑i=1nEgn(Xnian)<∞,
which implies (2.11).
By Markov's inequality, Lemma 2.1 (for p=2), Cr's inequality, gn(t)≥δt for 0<t≤1 and (2.16), we can get that∑n=1∞P(max1≤j≤n|Tj(n)|>ε2)≤C∑n=1∞∑i=1nE|Xni|2I(|Xni|≤an)an2≤C∑n=1∞∑i=1nE|Xni|I(|Xni|≤an)an≤C∑n=1∞∑i=1nEgn(Xnian)I(|Xni|≤an)≤C∑n=1∞∑i=1nEgn(Xnian)<∞,
which implies (2.12). This completes the proof of the theorem.
Corollary 2.5.
Under the conditions of Theorem 2.4,
1an∑i=1nXni⟶0a.s.asn⟶∞.
Corollary 2.6.
Let {Xni,i≥1,n≥1} be an array of rowwise ρ̃-mixing random variables and let {an,n≥1} be a positive real numbers. If there exists a constant β∈(0,1] such that
∑n=1∞∑i=1nE(|Xni|β|an|β+|Xni|β)<∞,
then (2.4) holds true.
Proof.
In Theorem 2.4, we take
gn(t)=|t|β1+|t|β,0<β≤1,n≥1.
It is easy to check that {gn(t),n≥1} is a sequence of nonnegative, even functions such that gn(|t|) is an increasing function of |t| for every n≥1. And
gn(t)≥12tβ≥12t,0<t≤1,0<β≤1.
Therefore, by Theorem 2.4, we can easily get (2.4).
Acknowledgments
The authors are most grateful to the editor Carlo Piccardi and anonymous referee for careful reading of the paper and valuable suggestions which helped to improve an earlier version of this paper. This work is supported by the National Natural Science Foundation of China (10871001) and the Academic innovation team of Anhui University (KJTD001B).
HuT.-C.TaylorR. L.On the strong law for arrays and for the bootstrap mean and variance199720237538210.1155/S01611712970004831444739ZBL0883.60024ZhuM.-H.Strong laws of large numbers for arrays of rowwise ρ˜-mixing random variables200720076742962272417BradleyR. C.On the spectral density and asymptotic normality of weakly dependent random fields199252355373115799010.1007/BF01046741ZBL0787.60059AnJ.YuanD. M.Complete convergence of weighted sums for ρ˜-mixing sequence of random variables2008781214661472245380110.1016/j.spl.2007.12.020KuczmaszewskaA.On Chung-Teicher type strong law of large numbers for ρ˜-mixing random variables20082008101405482403201BrycW.SmoleńskiW.Moment conditions for almost sure convergence of weakly correlated random variables1993119262963510.2307/21599501149969ZBL0785.60018CaiG.-H.Strong law of large numbers for ρ˜-mixing sequences with different distributions200620067276482244296GanS. X.Almost sure convergence for ρ˜-mixing random variable sequences2004674289298206012810.1016/j.spl.2003.12.011PeligradM.On the asymptotic normality of sequences of weak dependent random variables199693703715140059510.1007/BF02214083ZBL0855.60021PeligradM.Maximum of partial sums and an invariance principle for a class of weak dependent random variables199812641181118910.1090/S0002-9939-98-04177-X1425136ZBL0899.60044PeligradM.GutA.Almost-sure results for a class of dependent random variables19991218710410.1023/A:10217446267731674972ZBL0928.60025SungS. H.Complete convergence for weighted sums of ρ˜-mixing random variables20102010136306082611046UtevS.PeligradM.Maximal inequalities and an invariance principle for a class of weakly dependent random variables200316110111510.1023/A:10222784046341956823ZBL1012.60022WuQ.JiangY.Some strong limit theorems for ρ˜-mixing sequences of random variables20087881017102310.1016/j.spl.2007.09.0612418919