0 < p 1 - Journal of King Saud University - Science" /> 0 < p 1 - Journal of King Saud University - Science" />
7.2
CiteScore
3.7
Impact Factor
Generic selectors
Exact matches only
Search in title
Search in content
Post Type Selectors
Search in posts
Search in pages
Filter by Categories
ABUNDANCE ESTIMATION IN AN ARID ENVIRONMENT
Case Study
Correspondence
Corrigendum
Editorial
Full Length Article
Invited review
Letter to the Editor
Original Article
Retraction notice
REVIEW
Review Article
SHORT COMMUNICATION
Short review
7.2
CiteScore
3.7
Impact Factor
Generic selectors
Exact matches only
Search in title
Search in content
Post Type Selectors
Search in posts
Search in pages
Filter by Categories
ABUNDANCE ESTIMATION IN AN ARID ENVIRONMENT
Case Study
Correspondence
Corrigendum
Editorial
Full Length Article
Invited review
Letter to the Editor
Original Article
Retraction notice
REVIEW
Review Article
SHORT COMMUNICATION
Short review
View/Download PDF

Discrete Hardy’s inequalities with 0<p1

Department of Mathematics and Information Technology, The Education University of Hong Kong, 10 Lo Ping Road, Tai Po, Hong Kong, China
Disclaimer:
This article was originally published by Elsevier and was migrated to Scientific Scholar after the change of Publisher.

Peer review under responsibility of King Saud University.

Abstract

We generalize the famous discrete Hardy inequality to 0<p1 . We obtain this generalization by using the atomic decompositions of the discrete Hardy spaces.

1

1 Introduction

The main theme of this paper is the following discrete Hardy inequalities.

Let 0<p1 . There exists a constant C>0 such that for any sequence a=a(n)nZ with a(-n)=0,nN{0} , we have

(1)
mN1mj=1ma(j)p1pCmN|a(m)|p+minZjma(j)m-jp1p.

The family of Hardy inequalities, which consists of the discrete form and the integral form, is one of the most important inequalities in analysis. For the history of the Hardy inequalities, the reader is referred to Kufner et al. (2006, 2007). For the applications and further developments of this famous inequality, the reader may consult (Kufner et al., 2007; Kufner and Persson, 2003; Opic and Kufner, 1990).

Recently, the integral form Hardy inequality had been extended to Hardy spaces on R . In Ho (2016), the Hardy inequalities in Hardy spaces are established by using the atomic decompositions of Hardy spaces. This method is also used in Ho (2016, 2017a,b) to study the Hardy inequalities on Hardy-Morrey spaces with variable exponents and weak Hardy-Morrey spaces.

In this paper, we use the idea from Ho (2016) to obtain the above generalization (1) of Hardy’s inequality to 0<p1 .

To apply the method in Ho (2016), we need to consider the discrete analogue of the classical Hardy spaces on R . The discrete Hardy spaces had been introduced in Boza and Carro (1998, 2002). Moreover, the atomic decompositions of the discrete Hardy spaces were also obtained in Boza and Carro (1998, 2002). These atomic decomposition are precisely what we need to establish the discrete Hardy inequality with 0<p1 .

Thus, on one hand, we extend the classical discrete Hardy inequalities to 0<p1 , on the other hand, the main result of this paper gives an application of the atomic decompositions established in Boza and Carro (1998, 2002).

This paper is organized as follows. In Section 2, we recall the definition of the discrete Hardy spaces. The atomic decompositions for the discrete Hardy spaces are also presented in this section. The discrete Hardy inequalities with 0<p1 are established in Section 3.

2

2 Discrete Hardy spaces

In this section, we first recall the definition of the discrete Hardy spaces by discrete Hilbert transform on Z .

For any sequence a=a(n)nZ , the discrete Hilbert transform of a is defined by (Hda)(m)=nma(n)m-n.

For any BZ , let |B| denote the cardinality of B.

We use the definition of discrete Hardy spaces from (Boza and Carro, 1998, Definition 3.1).

Definition 2.1

Let 0<p1 . The discrete Hardy spaces Hp(Z) consists of those sequence a=a(n)nZ satisfying aHp(Z)=alp(Z)+Hdalp(Z)<.

In view of the above definition, we see the reason why the second summation on the left hand side of (1) is taking over Z .

We now present the atomic characterization of Hp(Z) , we begin with the definition of Hp(Z) -atom (Boza and Carro, 1998, Definition 3.9).

Definition 2.2

Let 0<p1 . A sequence a=a(k)kZ is an Hp(Z) -atom if it satisfies

  • supp a is contained in a ball in Z of cardinality 2n+1,n1 .

  • al(Z)2n+1-1/p .

  • nZnαa(n)=0 for every αN with α1p-1 .

To present the atomic decomposition of Hp(Z) , we recall the atomic version of Hp(Z) from (Boza and Carro, 1998, p.43).

Definition 2.3

Let 0<p1 . The atomic discrete Hardy space Hatp(Z) consists of those sequence a=a(n)nZ such that a=jNλjaj where aj are Hp(Z) -atoms and aHatp(Z)=infjN|λj|p1p where the infimum is taken over all possible representations of a in terms of Hp(Z) -atoms.

The following result gives the atomic decomposition of Hp(Z) .

Theorem 1

Let 0<p1 . Then, there exist constants B,C>0 such that for any sequence a=a(n)nZ , we have BaHp(Z)aHatp(Z)CaHp(Z).

The reader is referred to (Boza and Carro, 1998, Theorems 3.10 and 3.14) for the proof of the above theorem.

We can also characterize discrete Hardy spaces by using Poisson integral and area functions, see (Boza and Carro, 1998, Theorems 3.4 and 3.8). The reader is also referred to Boza (2012) and Kanjin and Satake (2000), Komori (2002) for the factorization theorem and the molecular characterizations of discrete Hardy spaces, respectively.

The reader is also referred to Herz (1973), Ho (2009, 2012), Jiao et al. (2017), Weisz (1994) for some other applications of the atomic decompositions such as the characterizations of BMO and martingale BMO.

3

3 Hardy’s inequalities

We establish the main result of this paper in this section. We first introduce the Hardy operators in order to simplify our presentation. For any αN{0} and μ>0 , define (Tα,μa)(m)=1mα-μ+1j=1mjαa(j),mN.

Notice that when α=μ=0 , we have

(2)
(T0,0a)(m)=1mj=1ma(j).

It is precisely the Hardy-Littlewood average for the sequence a=a(k)k=1 .

We are now ready to present the main result of this paper, the mapping properties of Tα,μ on discrete Hardy spaces Hp(Z) .

Theorem 2

Let 0<p1 and 0μ<1 . Suppose that αN{0} satisfies α1p-1 . If

(3)
1p=1r+μ, then there exists a constant C>0 such that for any sequence aHp(Z) with support contained in N , we have Tα,μalr(N)CaHp(Z).

When 0<p1 and α=μ=0 , we have p=r and the above theorem yields T0,0alp(N)CaHp(Z).

In view of (2), it establishes the discrete Hardy inequality (1).

We need several supporting results to obtain the proof of Theorem 2. We start with the mapping property of Tα,μ on Hp(Z) -atoms.

Lemma 3

Let 0<p1 , 0μ<1 and αN{0} with α1p-1 . Suppose that 1q-1r<μ1q for some q>1 . If a=a(n)nZ satisfies

  • suppa is contained in a ball B in N{0} of cardinality 2n+1,n1 ,

  • al(N)2n+1-1/p ,

  • n=0nαa(n)=0 ,

then, we have Tα,μalr(N)C|B|μ+1r-1p.

Proof

Let B={iN:MiN},M,NN . We have |B|=N-M+1 . For any i<M , we have (Tα,μa)(i)=1iα-μ+1j=1ijαa(j)=0.

Similarly, for any i>N , Items (1) and (3) assure that (Tα,μa)(i)=1iα-μ+1j=1ijαa(j)=1iα-μ+1jNjαa(j)=0.

Therefore, we find that supp(Tα,μa)B .

The Hölder inequality yields |j=1mjαa(j)|j=1m|a(j)|q1/qj=1mjαq1/qCalqmα+1q for some C>0 . Therefore, |(Tα,μa)(m)|C1mα-μ+1alqmα+1q=Cmμ-1qalq.

Furthermore, as supp(Tα,μa)B and μ1q , we obtain Tα,μalrrCalqrm=MNmrμ-rqCMN+1xrμ-rqdxCalqr(Nrμ-rq+1-Mrμ-rq+1).

Since 1q-1r<μ1q , we find that 0<rμ-rq+11 . Consequently, (Nrμ-rq+1-Mrμ-rq+1)N-Mrμ-rq+1|B|rμ-rq+1.

Hence, Tα,μalrCalq|B|μ-rq+1rCal|B|1q|B|μ-1q+1rC|B|μ+1r-1p for some C>0 .  □

Lemma 3 applies to those Hp(Z) -atom a=a(n)nZ with support in N{0} . On the other hand, for any aHp(Z) with support in N{0},a does not necessarily possess an atomic decomposition with all the Hp(Z) -atoms supported in N{0} . To tackle this difficulty, we consider the odd and the even extensions of aHp(Z) .

For any sequence a=a(n)nZ , the even part of a and the odd part of a, denoted by ae and ao , are defined by ae=a(n)+a(-n)2nZandao=a(n)-a(-n)2nZ, respectively.

As the even part is a “reflection” about k=0 , the term a(0) will be counted twice in the even part. Thus, for the case when α equals to zero, we need some further modifications. The details of these modifications and the uses of the even and odd parts are given in the subsequent results.

Lemma 4

Let 0<p1 and αN{0} with α1p-1 . If aHp(Z) with suppaN{0} , then there exist a sequence of scalars λjjN and sequences ajjN satisfying

  • supp aj is contained in a ball in N{0} of cardinality 2n+1,n1 ,

  • ajl(Z)22n+1-1/p ,

  • n=0nαaj(n)=0 ,

such that a=jNλjaj and jN|λj|p1/pCaHp(Z).

Proof

In view of Theorem 1, we have a sequence of scalars λjjN and Hp(Z) -atoms ajjN such that a=jNλjaj and jN|λj|p1/pCaHp(Z) .

We split the proof into three cases, α is a positive even integer, α is a positive odd integer and α equals zero.

  1. α is a positive even integer. We consider the even part of a, ae(k)=jNλjaj(k)+aj(-k)2.

    Since suppaN{0} , we have a=2aeχN{0} . That is, a(k)=jNλj(aj(k)+aj(-k))χN{0}(k).

    Write Aj=Aj(k)kZ where Aj(k)=(aj(k)+aj(-k))χN{0}(k) . We are going to show that Aj fulfills Item (1)–(3).

    As α is even, we find that kZkαaj(k)=kZkαaj(-k)=0.

    If suppajN{0},Aj=aj satisfies Item (1)–(3). If suppajZ(N{0}) , then Aj(·)=aj(-·) also satisfies Item (1)–(3).

    If 0suppaj , we find that n=0nαAj(n)=n=0nα(aj(n)+aj(-n))=nZnαaj(n)=0 because α is a positive even integer.

    We have Ajl(N)2ajl(N) . Moreover, suppAj(suppajsuppaj(-·))(N{0}) . Hence, |suppAj||suppaj| .

    Consequently, Aj fulfills Item (1)–(3).

  2. α is a positive odd integer. We consider the odd part of a, ao(k)=jNλjaj(k)-aj(-k)2.

    Since suppaN{0} , we have a=2aoχN{0} . That is, a(k)=jNλj(aj(k)-aj(-k))χN{0}(k).

    Write Aj=Aj(k)kZ where Aj(k)=(aj(k)-aj(-k))χN{0}(k) .

    As α is odd, we find that kZkαaj(k)=-kZkαaj(-k)=0.

    If suppajN{0},Aj=aj satisfies Item (1)–(3). If suppajZ(N{0}) , then Aj(·)=aj(-·) also satisfies Item (1)–(3).

    If 0suppaj , we find that n=0nαAj(n)=n=0nα(aj(n)-aj(-n))=nZnαaj(n)=0 because α is a positive odd integer.

    Therefore, Aj fulfills Item (1)–(3).

  3. α equals to zero. In this case, we also consider the even part of a, ae(k)=jNλjaj(k)+aj(-k)2. Write bj(k)=aj(k),k0,aj(0)2,k=0. When k0 , we obviously have ae(k)=jNλjaj(k)+aj(-k)2=jNλjbj(k)+bj(-k)2. Since suppaN{0} , we have a(0)=jNλjaj(0)=0. Therefore, for k=0 , we have ae(0)=a(0)=0=jNλjaj(0)=jNλjbj(0)+bj(0)2. As suppaN{0} , we have a=2aeχN{0} . That is, a(k)=jNλj(bj(k)+bj(-k))χN{0}(k).

    Write Bj=Bj(k)kZ where Bj(k)=(bj(k)+bj(-k))χN{0}(k) . It remains to show that Bj satisfies Item (1)–(3).

    When suppajN{0},Bj=aj satisfies Item (1)–(3). When suppajZ(N{0}) , then Bj(·)=aj(-·) also satisfies Item (1)–(3).

    If 0suppaj , we find that n=0Bj(n)=n=0(bj(n)+bj(-n))=n=1aj(n)+n=--1aj(n)+2bj(0)=nZaj(n)=0.

Thus, Bj fulfills Item (1)–(3). □

Proof of Theorem 2.

In view of Lemma 4, for any aHp(Z) with suppaN{0} , there exist a sequence of scalars λjjN and ajjN satisfying.

  • supp aj is contained in a ball in N{0} of cardinality 2n+1,n1 ,

  • ajl(Z)22n+1-1/p ,

  • n=0nαaj(n)=0 ,

such that a=jNλjaj and
(4)
jN|λj|p1/pCaHp(Z).

When 0<r1 , Lemma 3 and (3) assure that Tα,μalr(N)rjN|λj|rTα,μajlr(N)rCjN|λj|r for some C>0 . Furthermore, (3) also guarantees that r>p . Thus, (4) yields Tα,μalr(N)CjN|λj|r1/rCjN|λj|p1/p=CaHp(Z) for some C>0 .

When r>1 , we have Tα,μalr(N)jN|λj|Tα,μajlr(N)CjN|λj|CjN|λj|p1/p=CaHp(Z) because 0<p1 .  □

Acknowledgment

The author would like to thank the referees for their helpful suggestions.

References

  1. , , . Discrete Hardy spaces. Stud. Math.. 1998;129:31-50.
    [Google Scholar]
  2. , , . Hardy spaces on Zn . Proc. R. Soc. Edinburgh. 2002;132A:25-43.
    [Google Scholar]
  3. , . Factorization of sequences in discrete Hardy spaces. Stud. Math.. 2012;209:53-69.
    [Google Scholar]
  4. Herz, C., 1973/74. Hp -spaces of martingales, 0<p1 . Z. Wahrscheinlichkeitstheorie und Verw. Gebiete 28, 189–205.
  5. , . Characterization of BMO in terms of Rearrangement-invariant Banach function spaces. Expo. Math.. 2009;27:363-372.
    [Google Scholar]
  6. , . Atomic decomposition of Hardy spaces and characterization of BMO via Banach function spaces. Anal. Math.. 2012;38:173-185.
    [Google Scholar]
  7. , . Hardy’s inequality on Hardy spaces. Proc. Jpn. Acad. Ser. A. 2016;92:125-130.
    [Google Scholar]
  8. Ho, K.-P., 2016. Hardy’s inequality on Hardy-Morrey spaces. Georgian Math. J. (accepted).
  9. , . Hardy’s inequality on Hardy-Morrey spaces with variable exponents. Mediter. J. Math. 2017
    [CrossRef] [Google Scholar]
  10. , . Atomic decompositions and Hardy’s inequality on weak Hardy-Morrey spaces. Sci. China Math.. 2017;60:449-468.
    [Google Scholar]
  11. , . The predual and John-Nirenberg inequalities on generalized BMO martingale spaces. Trans. Am. Math. Soc.. 2017;369:537-553.
    [Google Scholar]
  12. , , . Inequalities for discrete Hardy spaces. Acta Math. Hungar.. 2000;89:301-313.
    [Google Scholar]
  13. , . The Prehistory of the Hardy Inequality. Am. Math. Monthly. 2006;113:715-732.
    [Google Scholar]
  14. , . The Hardy Inequality, About its History and Some Related Results. Pilsen: Vydavatelsky Publishing House; .
  15. , , . Weighted inequalities of Hardy type. River Edge, NJ: World Scientific Publishing Co., Inc.; .
  16. , . The atomic decomposition of molecule on discrete Hardy spaces. Acta Math. Hungar.. 2002;95:21-27.
    [Google Scholar]
  17. Opic, B., Kufner, A., 1990. Hardy-type inequalities, Pitman Reserach Notes in Math. Series 219, Longman Sci. and Tech, Harlow.
  18. , . Martingale Hardy spaces and their applications in Fourier analysis.Lecture Notes in Mathematics. Vol vol. 1568. Berlin: Springer-Verlag; .
Show Sections