Minghui Zhong and Xianming Hou
School of Mathematical Sciences,Xiamen University,Xiamen 361005,China.
Abstract.Let 1<p<∞and ω ∈Ap.The space CMO(Rn)is the closure in BMO(Rn)of the set of (Rn).In this paper,an equivalent characterization of CMO(Rn)with Ap weights is established.
Key words:BMOω(Rn),CMO(Rn),Ap,John-Nirenberg inequality.
The goal of this paper is to provide an equivalent characterization ofCMO(Rn),which is useful in the study of compactness of commutators of singular integral operator and fractional integral operator.
The spaceBMO(Rn)is defined by the set of functions(Rn)such that
where
The spaceCMO(Rn)is the closure inBMO(Rn)of the set of(Rn),which is a proper subspace ofBMO(Rn).
In fact,it is known thatCMO(Rn)=VMO0(Rn),whereVMO0(Rn)is the closure ofC0(Rn)inBMO(Rn),see[2,3,9].HereC0(Rn)is the set of continuous functions on Rnwhich vanish at infinity.Neri[8]gave a characterization ofCMO(Rn)by Riesz transforms.Meanwhile,Neri proposed the following characterization ofCMO(Rn)and its proof was established by Uchiyama in his remarkable work[11].
Theorem 1.1.Let f ∈BMO(Rn).Then f ∈CMO(Rn)if and only if f satisfies the following three conditions
Recently,Guo,Wu and Yang[6]established an equivalent characterization of spaceCMO(Rn)by local mean oscillations.Lots of works about spaceCMO(Rn)have been studied,see[4]for example. Muckenhoupt and Wheeden[7,Theorem 5]showed the norm ofBMOω(Rn)(see Definition 1.2)is equivalent to the norm ofBMO(Rn),where the weight functionωis MuckenhouptApweight.So it is natural to consider equivalent characterizations ofCMO(Rn)associated toApweights.
To state our main results,we first recall some relevant notions and notations.
The following class ofApwas introduced in[1,5].
Definition 1.1.Let ω(x)≥0and ω(x)(Rn).For1<p <∞,we say that ω(x)∈Ap if there exists a constant C>0such that for any cube Q,
Also,for p=1,we say that ω(x)∈A1if there is a constant C>0such that
where M is the Hardy-Littlewood maximal operator.For p ≥1,the smallest constant appearing in(1.1)and(1.2)is called the Ap characteristic constant of ω and is denoted by[ω]Ap.
Definition 1.2.Let ω∈Ap.For a cube Q inRn,we say a function f ∈(Rn)is in BMOω(Rn)if f satisfies
where
Letω∈Ap(p≥1),q>1,ThenBMOω,q(Rn)is defined by
where
Now,we can formulate our main results as follows.
Theorem 1.2.Let p ≥1,1<q <∞.Suppose f ∈BMO(Rn)and ω ∈Ap.Then the following conditions are equivalent:
(1)f ∈CMO(Rn);
(2)f satisfies the following three conditions:
(3)f satisfies the following three conditions:
Throughout this paper,the letterC,will stand for positive constants,not necessarily the same one at each occurrence,but independent of the essential variables.Iff ≤Cg,we writefgorgf;and iffgf,we writef ~g.A dyadic cubeQon Rnis a cube of the form
Rjmeans{x ∈Rn:|xi|<2j,i=1,2,···,n}.Forλ >0,λQdenotes the cube with the same center asQand side-lengthλtimes the side-length ofQ.
This section is devoted to the proof of Theorem 1.2.To do this,we firstly recall some auxiliary lemmas.Note that[7,Theorem 3]impiles the following weighted John-Nirenberg inequalities,also see[1,10].
Lemma 2.1.(John-Nirenberg)Let p∈[1,∞),ω∈Ap and f ∈BMOω(Rn).For every α>0and cube Q,there exist constants c1and c2such that
Next,we recall some useful properties ofApweights.
Lemma 2.2([5]).Let ω∈Ap and1≤p<∞.
1.There exist0<δ<1and C>0that depending only on the dimension n,p,and[ω]Ap such that for any cube Q and any measurable subset S of Q we have
2.There exist constants C and γ >0that depending only on the dimension n,p,and[ω]Ap such that for every cube Q we have
3.For all λ>1,and all cubes Q,
Now,we are in position to prove the Theorem 1.2.
Proof.To prove(1)?(2)in Theorem 1.2.Assume thatf ∈CMO(Rn).If(Rn),then(i)?(iii)hold.It is obvious that(i)holds for uniformly continuous functionsf.Without loss of generality,we assumesupp(f)?Q0.Then for eachQ?Rn,there existsh∈Rn,for|x|>|h|,we haveQ0∩(Q+x)=?,(iii)holds.
Note that
For(Rn),we have
On the other hand,Q(0,r)denotes the closed cube centered at 0 with side-lengthr.For anyx0∈Q(0,r),there exists a cubeQcentered atx0such thatQ(0,r)?Q,by(2.1),we get
which tends to 0 as|Q|tends to+∞,(ii)holds.
Iffor any givenε >0,there existsfε ∈(Rn)satisfying(i)?(iii)andThen by Lemma 2.1 and(2.2),forω∈Ap,1<p<∞,it is easy to see
The detailed proof of(2.4)also can be found in[1,7].By(2.4)and the triangle inequality,we deduce that(i)?(iii)hold forf.
The proof of(2)?(3).By the H?lder inequality,we get
where 1/q+1/q′=1.
It follows from(2.5)that iffsatisfies(i)?(iii)thenfsatisfies(i′)?(iii′).
The proof of(3)?(1).Now we show that iffsatisfies(i′)?(iii′)then for allε>0,there existsgε ∈BMO(Rn)such that
We prove(2.6)and(2.7)by the following two steps.
Step IBy(i′)and(ii′),there existiεandkεsuch that
By(iii′),there existsjε >kεsuch that
Now for eachx ∈Rjε,we take dyadic cubeQxwith side-length 2iεcontainingx;ifmeans a dyadic cube of side-length 2iε+m?jε.Set=m(f,Qx)ω,by(ii′),there existsmε >jεsuch that
To see this,by(ii′),letmε>jε+kε?iεbe large enough such that when
for some positive constantC1.
For,it is obvious that
This together with(2.12)and(2.3)imply that
Sinceby(2.3)and(2.12),we have
By(2.13),(2.14)and(2.12),we conclude that for anyQxwith
For anyby(2.15),we get
Step IIDefinewhenx∈RmεandwhenNotice that
By the definition ofiε,jεandmε,iforx,,there existsC2>0 such that
In fact,assume that|x|<|y|.Firstly,we show that ifx,,then(2.17)holds.By noting thatx,,we get
and(2.17)holds.Next,if,we deduce from(2.11)that
Thirdly,ifand,(2.15)indicates that
by(2.8),(2.17)holds.Similarly,ifandby(2.8),(2.17)also holds.Ifnotice thatQ∩Rjε=?and by(2.10),
Similarly,we have
Hence
Combining these cases,(2.17)holds.
We turn to prove thatgεsatisfies(2.6).Set
By the definition ofgε,we get
Selectt<2iε,then
where in the second inequality we make the change of variabley=ut.
Sinceu∈B(0,1)andt<2iε,?x∈Rn,
By(2.17),ifx,x?tu∈Rmε,,hence
If one ofxandx?tuin,the other must be in,we also have
Therefore
We obtain that(2.6)holds.
Now we prove(2.7).By the definitioniεandjεagain,we obtain that for anyx∈Rmε,
Indeed,
IfQx∩Rjε=?,by(2.10),(2.18)holds.Ifusing(2.8),(2.18)holds.
LetQbe a arbitrary cube in Rn.In order to prove(2.7)holds,it suffices to show
We consider the following four cases:
Case(i):Q ?Rmεand max{diam Qx:Qx∩Q?}>4diam Q,by(2.16),the number ofis finite.Ifand,by(2.17),
Moreover,if,then,by(2.8),we haveM(f,Q)ω <ε;ifQ∩Rjε=?,by(2.10),we also obtainM(f,Q)ω <ε.Hence
Case(ii):Q?Rmεanddiam Q,we have
Invoking(2.18),we get
Case(iii):,thenQ∩Rjε=?andM(f,Q)ω <ε.Using(2.17),
Hence
Case(iv):andLetPQbe a smallest positive number such thatQ?RPQ.Then
Moreover,
On the one hand,by(2.18),we have
On the other hand,it is easy to prove that
Combining with(2.9)and(2.18)and the fact thatgε(x)=gε(y)for any,we obtain
This implies(2.7)and completes the proof of Theorem 1.2.
Journal of Mathematical Study2020年1期