1. Trang chủ
  2. » Thể loại khác

DSpace at VNU: Strong laws of large numbers for random fields in martingale type p Banach spaces

8 211 0

Đang tải... (xem toàn văn)

THÔNG TIN TÀI LIỆU

Thông tin cơ bản

Định dạng
Số trang 8
Dung lượng 268,33 KB

Các công cụ chuyển đổi và chỉnh sửa cho tài liệu này

Nội dung

Contents lists available atScienceDirectStatistics and Probability Letters journal homepage:www.elsevier.com/locate/stapro Strong laws of large numbers for random fields in martingale ty

Trang 1

Contents lists available atScienceDirect

Statistics and Probability Letters journal homepage:www.elsevier.com/locate/stapro

Strong laws of large numbers for random fields in martingale type p

Banach spaces

Le Van Dunga,∗, Nguyen Duy Tienb

aFaculty of Mathematics, Danang University of Education, 459 Ton Duc Thang, Lien Chieu, Danang, Viet Nam

bFaculty of Mathematics, Hanoi University of Science, 334 Nguyen Trai, Thanh Xuan, Hanoi, Viet Nam

a r t i c l e i n f o

Article history:

Received 17 June 2009

Received in revised form 10 January 2010

Accepted 11 January 2010

Available online 29 January 2010

MSC:

60B11

60B12

60F15

60G42

a b s t r a c t

We extend Marcinkiewicz–Zygmund strong laws for random fields{Vn;n ∈ Nd}with

values in martingale type p Banach spaces Our results are more general and stronger than

the result ofGut and Stadtmüller(2009) and some other ones

© 2010 Elsevier B.V All rights reserved

1 Introduction

Let Nd be the positive integer d-dimensional lattice points, where d is a positive integer For m= (m1, ,m d)and n= (n1, ,n d) ∈ Nd , notation mn means that m in i,1 ≤id, |nα|is used forQd

i= 1nαi

i ,[m,n) = Qd

i= 1[m i,n i)is a

d-dimensional rectangle andWd

i= 1(m i <n i)means that there is at least one of m1 <n1, m2 <n2, ., m d<n dholds We

write 1= (1, ,1) ∈Nd

Consider a random field{V n,n∈ Nd}of random elements defined on a probability space(Ω,F,P)taking values in a

real separable martingale type p(1≤p≤2)Banach spaceXwith normk · k In the current work, we establish strong laws

of large numbers (SLLN) for|nα|− 1maxknkS kk This can be done by studying convergence of sums of type

X

n

|n|−1P{max

kn

kS kk > |nα|} for every >0.

Many authors have investigated the Marcinkiewicz type strong laws of large numbers for random fields{X n,n ∈ Nd}

of random variables For example,Fazekas and Tómács(1998) studied strong laws of large numbers|n|−1/r S n(for some

0 < r < 1) for pairwise independent random variables,Czerebak-Mrozowicz et al.(2002) studied Marcinkiewicz type strong laws of large number|n|− 1 /p(S nES n)(for some 1<p<2) for pairwise independent random fields Recently,Gut and Stadtmüller(2009) studied Marcinkiewicz–Zygmund laws of large numbers for random fields of i.i.d random variables

In this paper, we not only extend these results to random fields in martingale type p Banach spaces but also bring more

general and stronger ones

Throughout this paper, the symbol C will denote a generic constant (0<C< ∞) which is not necessarily the same one

in each appearance

∗Corresponding author.

E-mail addresses:lvdunght@gmail.com (L.V Dung), nduytien2006@yahoo.com (N.D Tien).

0167-7152/$ – see front matter © 2010 Elsevier B.V All rights reserved.

Trang 2

2 Preliminaries

Technical definitions relevant to the current work will be discussed in this section

Scalora(1961) introduced the idea of the conditional expectation of a random element in a Banach space For a random

element V and sub-σ-algebraGofF, the conditional expectation E(V|G)is defined analogously to that in the random variable case and enjoys similar properties

A real separable Banach spaceXis said to be martingale type p(1 ≤p≤2)if there exists a finite positive constant C

such that for all martingales{S n;n≥1}with values inX,

sup

n≥ 1

EkS nkpC

∞ X

n= 1

EkS nS n− 1kp.

It can be shown using classical methods from martingale theory that ifXis of martingale type p, then for all 1r< ∞

there exists a finite constant C such that

E sup

n≥ 1

kS nkrCE

∞ X

n= 1

kS nS n− 1kp

!r p

. Clearly every real separable Banach space is of martingale type 1 and the real line (the same as any Hilbert space) is of

martingale type 2 If a real separable Banach space of martingale type p for some 1<p2 then it is of martingale type r for all r∈ [1,p)

It follows from theHoffmann-Jørgensen and Pisier(1976) characterization of Rademacher type p Banach spaces that if a Banach space is of martingale type p, then it is of Rademacher type p But the notion of martingale type p is only superficially similar to that of Rademacher type p and has a geometric characterization in terms of smoothness For proofs and more

details, the reader may refer toPisier (1975,1986)

To prove the main result we need the following lemma which was proved byDung et al.(2009) in the case d=2 If d is

arbitrary positive integer, then the proof is similar and so is omitted

Lemma 2.1 Let 1p2 and let {V k,kn}be a collection of |n|random elements in a real separable martingale type p Banach space with E(V k|Fk) = 0 for all kn, whereFk is theσ-field generated by the family of random elements

{V l: Wd

i= 1(l i<k i)},F1= {∅ ,Ω} Then

E max

kn

kS kkpCX

kn

EkV kkp,

where S k= P

ik V i In the case of p=1, the hypothesis that E(V k|Fk) =0 for all kn is superfluous.

Lemma 2.2 Let 1<p2 Letα1, , αd be positive constants satisfying 1/p<min{ α1 , αd} <1, let q be the number of

integers s such thatαs=min{ α1 , αd} If E kVkr(log+kVk )q− 1

< ∞then we have

(i) X

n

1

|nα|

Z ∞

||

P{kVk ≥t}dt< ∞,

(ii) X

n

1

|nα|p

Z ||p

0

P{kVkpt}dt< ∞.

Proof Without loss of generality, we may assume min{ α1, , αd} = α1= · · · = αq< αq+ 1≤ αd

We first prove(i) We have by Lemma 3 ofStadtmüller and Thalmaier(2009) that

1 ≤n1 n q.nαq+ 1 /α 1

q+ 1 nαd/α 1

dj

1 ∼ C j(log j)q− 1

(q−1)! as j→ ∞ . Denote∆g(j) =g(j) −g(j−1)we get

X

n

1

|nα|

Z ∞

||

P{kVk ≥t}dt

∞ X

k= 1

1

kα 1∆g(k) Z

1

P{kVk ≥t}dt=

∞ X

k= 1

1

kα 1∆g(k)

∞ X

i=k

Z (i+ 1 )α1

1

P{kVk ≥t}dt

∞ X

k= 1

1

kα 1∆g(k)

∞ X

i=k

Z (i+ 1 ) α 1

1

P{kVk ≥iα 1}dt

kα 1∆g(k)

∞ X

iα 1 − 1P{kVk ≥iα 1}

Trang 3

∞ X

k= 1

1

kα 1∆g(k)

∞ X

i=k

iα 1 − 1

∞ X

j=i

P{jα 1 ≤ kVk < (j+1)α 1}

∞ X

k= 1

1

kα 1∆g(k)

∞ X

j=k

P{jα 1 ≤ kVk < (j+1)α 1}

j

X

i= 1

iα 1 − 1

C

∞ X

k= 1

1

kα 1∆g(k)

∞ X

j=k

jα 1P{jα 1 ≤ kVk < (j+1)α 1}

C

∞ X

j= 1

jα 1P{jα 1≤ kVk < (j+1)α 1}

j

X

k= 1

1

kα 1∆g(k)

= C

∞ X

j= 1

jα 1P{jα 1≤ kVk < (j+1)α 1}

j

X

k= 1

1

kα 1∆g(k)

= C

∞ X

j= 1

jα 1P{jα 1≤ kVk < (j+1)α 1}

j− 1

X

k= 1

 1

kα 1

(k+1)α 1



g(k)

+C

∞ X

j= 1

P{jα 1 ≤ kVk < (j+1)α 1}g(j)

C

∞ X

j= 1

jα 1P{jα 1≤ kVk < (j+1)α 1}

j− 1

X

k= 1

 1

kα 1

(k+1)α 1



k(log k)q− 1

+C

∞ X

j= 1

P{jα 1 ≤ kVk < (j+1)α 1}j(log j)q− 1

C

∞ X

j= 1

jα 1P{jα 1≤ kVk < (j+1)α 1} (log j)q− 1

j− 1

X

k= 1

 1

kα 1

(k+1)α 1



k

+C

∞ X

j= 1

P{jα 1 ≤ kVk < (j+1)α 1}j(log j)q− 1

C

∞ X

j= 1

jα 1P{jα 1≤ kVk < (j+1)α 1} (log j)q− 1

j

X

k= 1

1

kα 1

+C

∞ X

j= 1

P{jα 1 ≤ kVk < (j+1)α 1}j(log j)q− 1

C

∞ X

j= 1

P{jα 1≤ kVk < (j+1)α 1}j(log j)q− 1< ∞.

Now we prove (ii)

X

n

1

|nα|p

Z ||p

0

P{kVkpt}dt = X

n

1

|nα|p

Z 1

0

P{kVkpt}dt+ X

n

1

|nα|p

Z ||p

1

P{kVkpt}dt

n

1

|nα|p+CX

n

1

|nα|p

Z ||p

1

P{kVkpt}dt. Noting that the first term on the right-hand side is finite, it remains to prove that

X

n

1

|nα|p

Z ||p

1

P{kVkpt}dt.

Denote d(k) = Pn1 n q=k1, we have by Lemma 3.1 ofGut(2001) that

Xd(j)

j pα 1 ∼ C(log k)q− 1

k pα 1 − 1 .

Trang 4

Hence, we have

X

n

1

|nα|p

Z ||p

1

P{kVkpt}dt

∞ X

k,n q+ 1 , ,n d= 1

k pα 1.n pαq+ 1

q+ 1 .n pαd

d

[1nαq+ 1

q+ 1 nαd] X

j= 1

E(kVkp I(j≤ kVk <j+1)) (where[x]denotes the greatest integer not exceeding x)

C

∞ X

k,n q+ 1 , ,n d= 1

k pα 1.n pαq+ 1

q+ 1 .n pαd

d

[1nαq+ 1

q+ 1 nαd] X

j= 1

j p P(j≤ kVk <j+1)

C

∞ X

k,n q+ 1 , ,n d= 1

k pα 1.n pαq+ 1

q+ 1 .n pαd

d

[1nαq+ 1

q+ 1 nαd] X

j= 1

[j p− (j−1)p]P(kVk ≥j)

C

∞ X

k,n q+ 1 , ,n d= 1

k pα 1.n pαq+ 1

q+ 1 .n pαd

d

[1nαq+ 1

q+ 1 nαd] X

j= 1

pj p−1P{kVk ≥j}

=C

X

n q+ 1 , ,n d= 1

1

n pαq+ 1

q+ 1 .n pαd

d

∞ X

k= 1

d(k)

k pα 1

[1nαq+ 1

q+ 1 nαd] X

j= 1

pj p−1P{kVk ≥j}

C

X

n q+ 1 , ,n d= 1

1

n pαq+ 1

q+ 1 .n pαd

d

[nαq+ 1

q+ 1 nαd] X

j= 1

pj( 1 /α 1 − 1 )j(p− 1 /α 1 )P{kVk ≥j} X∞

k= 1

d(k)

k pα 1

+

∞ X

n q+ 1 , ,n d= 1

1

n pαq+ 1

q+ 1 .n pαd

d

∞ X

i=[nαq+ 1

q+ 1 nαd]+ 1

pj p−1P{kVk ≥j}

∞ X

k=



j/nαq+ 1

q+ 1 nαd 1 /α 1

d(k)

k pα 1

C

X

n q+ 1 , ,n d= 1

1

nβq+ 1

q+ 1 .nβd

d

[nαq+ 1

q+ 1 nαd] X

j= 1

pj( 1 /α 1 − 1 )P{kVk ≥j} X∞

k= 1

d(k)

k pα 1

+

∞ X

n q+ 1 , ,n d= 1

1

n pαq+ 1

q+ 1 .n pαd

d

∞ X

j=[nαq+ 1

q+ 1 nαd]+ 1

pj p−1P{kVk ≥j}

∞ X

k=



j/nαq+ 1

q+ 1 nαd 1 /α1

d(k)

k pα 1

C

X

n q+ 1 , ,n d= 1

1

nβq+ 1

q+ 1 .nβd

d

∞ X

j= 1

j r−1(log i)q− 1P{kVk ≥j}

which is finite if E(kVkrlog+kVk )q− 1< ∞and sinceβl= αl/α1>1 for q+1≤ld. 

The random field{V n,n∈Nd}is said to be weakly mean dominated by the random element V if, for some 0<C< ∞, 1

|n|

X

kn

P{kV kk ≥x} ≤CP{kVk ≥x}

for all n∈Nd and x>0

3 Main results

With the preliminaries accounted for, the main results may now be established In the following, we let{V n;n ∈ Nd}

be an array of random elements in a real separable Banach spaceX,Fkis theσ-field generated by the family of random elements{V l: Wd

i= 1(l i<k i)},F1= {∅ ,Ω}

The first theorem is a general a.s convergence one

Theorem 3.1 Letα1, , αd be positive constants Let{V n,n∈Nd}be a random field of random elements If

|n|P{max

kn

Trang 5

X

n

P



max

and, a fortiori, the SLLN

1

|nα|maxkn

obtains.

Conversely,(3.2)implies that(3.1)holds.

Proof ((3.1)⇒(3.2)) Fix >0, denote 2n= (2n1, ,2n d)and 2nα= (2n1 α 1, ,2n dαd) We have the inequalities X

n

P



max

l≺ 2n kS lk > |2nα|  ≤ X

n

min

k∈[ 2n, 2n+ 1 )P

 max

lk

kS lk > 

2α 1 +··· ,αd

|kα| 

n

X

k∈[ 2n, 2n+ 1 )

1

|2n|P

 max

lk

kS lk > 

2α 1 +···+ αd

|kα| 

n

X

k∈[ 2n, 2n+ 1 )

2d

|k|P

 max

lk

kS lk > 

2α 1 +···+ αd

|kα| 

≤2dX

n

1

|n|P

 max

ln

kS lk > 

2α 1 +···+ αd|nα|  < ∞. (by(3.1)) This implies by the Borel–Cantelli lemma that

1

Now for k∈ [2n,2n+1)we have

1

|kα|maxlk kS lk ≤ 1

|kα|l≺max2n+ 1

kS lk ≤ 1

|2nα|l≺max2n+ 1

kS lk = 2α 1 +···+ αd

|2(n+ 1 )α| l≺max2n+ 1

and so the conclusion(3.3)follows from(3.4)and(3.5)

((3.2)⇒(3.1)) Suppose that(3.2)holds, we easily to prove that for every >0,

X

n

1

|n|P



max

ln

kS lk > |nα|  ≤ X

n P

 max

l≺ 2n kS lk > 

2α 1 +···+ αd

|2nα|  ,

which implies that(3.1)holds The proof is completed 

The following theorem characterizes the martingale type p Banach spaces.

Theorem 3.2 Let 1p2 and letXbe a separable Banach space Then the following two statements are equivalent:

(i) The BanachXis of martingale type p.

(ii) For every random field{V n;n∈Nd}inXwith E(V n|Fn) =0 for all n∈Nd and for everyα = (α1, , αd)withαi>0 for

all 1id, the condition

X

n

EkV nkp

|nα|p < ∞

implies that, for every >0,

X

n

1

|n|P{max

kn

kS kk > |nα|} < ∞

and, a fortiori, the SLLN

1

|nα|maxkn

kS kk →0 a.s as|n| → ∞

obtains.

Trang 6

Proof In order to prove[ (i) ⇒ (ii)]we show that

X

n

P{max

k≺ 2nkS kk > |2nα|} < ∞ for every >0.

Applying Markov’s inequality andLemma 2.2we have that

X

n

P{max

k≺ 2nkS kk > |2nα|} ≤ X

n

1

p|2nα|p E(max

k≺ 2nkS kkp)

n

1

|2nα|p

X

k≺ 2n

EkV kkpCX

k

EkV kkp

|kα|p < ∞.

Now we prove[(ii)⇒(i)] Assume that (ii) holds Let{W n1,Gn1;n1≥1}be an arbitrary sequence of martingale difference

inXsuch that

X

n1 = 1

EkW n1kp

n p1 < ∞

Set

V n1, ,n d =W n1 if n2= · · · =n d=1 otherwise V n1, ,n d=0.

Then{V n1, ,n d}is the random field inXsatisfies E(V n1, ,n d|Fn1 , ,n d) =0 for all(n1, ,n d) ∈Nd, and

X

n1 , ,n d= 1

EkV n1, ,n dkp

(n1 .n d)p =

∞ X

n1 = 1

EkW n1kp

n p1 < ∞.

By (ii),

1

n1 .n d

X

i1n1

idnd

V i1, ,i d →0 a.s as(n1 .n d) → ∞.

Taking n2= · · · =n d=1 and letting n1→ ∞we obtain

1

n1

n1

X

j= 1

Then by Theorem 2.2 ofHoffmann-Jørgensen and Pisier(1976),Xis of martingale type p. 

In the next two theorems, we obtain the Marcinkiewicz–Zygmund type laws of large numbers for random fields of random elements

Theorem 3.3 LetXbe a martingale type p Banach space with 1 < p2 Letα1, , αd be positive constants satisfying

1/p < min{ α1 , αd} < 1, let q be the number of integers s such thatαs = min{ α1 , αd}and let{V n,n ∈ Nd}be a random field satisfying E(V n|Fn) = 0 for all n ∈ Nd Suppose that {V n,n ∈ Nd}is weakly mean dominated by V such that

E kVkr(log+kVk )q− 1

< ∞with r= 1

min { α 1 ,αd} Then

X

n

1

|n|P{max

kn

and, a fortiori, the SLLN

1

obtains.

Proof For each n∈Nd, set

V k0=V k I(kV kk ≤ |nα| ),V k00=V k I(kV kk > |nα| ),

Y k0=V k0−E(V k0|Fk),Y k00 =V k00−E(V k00|Fk),

S0

n= P

kn Y0

k,S00

n= P

kn Y00

k

Since E(V k|Fk) = 0, it follows that V k = Y0

k+Y00

k Moreover, ifG0

kandG00

kare theσ-fields generated by the family of random elements{Y0

l : Wd

i= 1(l i<k i)}and{Y00

l : Wd

i= 1(l i<k i)}, respectively, thenG0

k⊂FkandG00

k⊂Fk for all kn, which

imply that E(Y0|G0) =E(Y00|G00) =0 for all kn.

Trang 7

We now begin the proof For every >0,

X

n

1

|n|P{max

kn

kS kk >2|nα|} ≤ X

n

1

|n|P{max

kn

kS k0k > |nα|} + X

n

1

|n|P{max

kn

kS00kk > |nα|} (3.8) First, we show that

X

n

1

|n|P{max

kn

kS00kk > |nα|} < ∞.

Applying Markov’s inequality andLemma 2.2, we obtain

X

n

1

|n|P{max

kn

kS00kk > |nα|} ≤ X

n

1

|n||nE(max

kn

kS00kk ) ≤CX

n

1

|nα|

1

|n| X

kn

EkY k00k

n

1

|nα|

1

|n| X

kn

EkV k00k =CX

n

1

|nα|

1

|n| X

kn

Z ∞

0

P{kV k00k ≥t}dt

n

1

|nα|

1

|n| X

kn

Z ||

0

P{kV kk ≥ |nα|}dt+CX

n

1

|nα|

1

|n| X

kn

Z ∞

||

P{kV kk ≥t}dt

n

1

|n| X

kn

P{kV kk ≥ |nα|} +CX

n

1

|nα|

Z ∞

||

1

|n| X

kn

P{kV kk ≥t}dt

n

P{kVk ≥ |nα|} +CX

n

1

|nα|

Z ∞

||

P{kVk ≥t}dt

n

1

|nα|

Z ∞

||

P{kVk ≥t}dt< ∞ (byLemma 2.2).

By(3.8), in order to complete the proof, we next show that

X

n

1

|n|P{max

kn

kS0kk > |nα|} < ∞.

Again applying Markov’s inequality, we find that

X

n

1

|n|P{max

kn

kS0kk > |nα|} ≤ X

n

1

|n|

1

|nα|p E(max

kn

kS k0k > |nα| )p

n

1

|n|

1

|nα|p E(max

kn

kS k0k > |nα|p) ≤CX

n

1

|nα|p

1

|n| X

kn

EkY k0kp

n

1

|nα|p

1

|n| X

kn

EkV k0kp=CX

n

1

|nα|p

1

|n| X

kn

Z ∞

0

P{kV k0kpt}dt

n

1

|nα|p

1

|n| X

kn

Z ||p

0

P{kV kkpt}dt

n

1

|nα|p

Z ||p

0

1

|n| X

kn

P{kV kkpt}dt

n

1

|nα|p

Z ||p

0

P{kVkpt}dt< ∞ (byLemma 2.2).

Remark Note that in the case of q< d, positive constantsα1, , αdare not upper bounded by 1, which is weaker than condition (2.1) of Theorem 2.1 ofGut and Stadtmüller(2009)

Theorem 3.4 Letα1, , αd be positive constants satisfying min{ α1 , αd} > 1, let q be the number of integers s such that

αs=min{ α1 , αd} Suppose that{V n,n∈Nd}is weakly mean dominated by V such that E kVk (log+kVk )q− 1

< ∞ Then

(3.3)holds and then, the SLLN(3.4)obtains.

Proof The proof is similar to that ofTheorem 3.2with p=1 and we use T n0 = P

kn V k0and T n00 = P

kn V k00are instead of

S0

n and S00

n, respectively 

Trang 8

Czerebak-Mrozowicz, E.B., Klesov, O.I., Rychlik, Z., 2002 Marcinkiewicz-type strong laws of large numbers for pairwise independent random fields Probab Math Statist 22 (Fasc 1), 127–139.

Hoffmann-Jørgensen, J., Pisier, G., 1976 The law of large numbers and the central limit theorem in Banach spaces Ann Probab 4 (4), 587–599 Fazekas, I, Tómács, T., 1998 Strong laws of large numbers for pairwise independent random variables with multidimensional indices Publ Math Debrecen.

53 (1–2), 149–161.

Dung, L.V., Ngamkham, Th., Tien, N.D., Volodin, A.I., 2009 Marcinkiewwcz-type law of large numbers for double arrays of random elements in Banach spaces Lobachevskii J Math 30 (4), 337–346.

Gut, A., 2001 Convergence rates in the central limit theorem for multidimensionally indexed random variables Studia Sci Math Hungar 37, 401–418 Gut, A., Stadtmüller, U., 2009 An asymmetric Marcinkiewicz–Zygmund LLN for random fields Statist Probab Lett 79, 1016–1020.

Pisier, G., 1975 Martingales with values in uniformly convex spaces Israel J Math 20 (3–4), 326–350.

Pisier, G., 1986 Probabilistic methods in the geometry of Banach spaces In: Probability and Analysis (Varenna) In: Lecture Notes in Math., vol 1206 Springer, Berlin, pp 167–241.

Scalora, F.S., 1961 Abstract martingale convergence theorems Pacific J Math 11, 347–374.

Stadtmüller, U., Thalmaier, M., 2009 Strong laws for delayed sums of random fields Acta Sci Math (Szeged) 75 (3–4), 723–737.

Ngày đăng: 16/12/2017, 09:07

TỪ KHÓA LIÊN QUAN

🧩 Sản phẩm bạn có thể quan tâm