Endpoint estimates for the commutators of multilinear Calderón-Zygmund operators with Dini type kernels
Li and Xue Journal of Inequalities and Applications (2016) 2016:252
DOI 10.1186/s13660-016-1201-2
RESEARCH
Open Access
Endpoint estimates for the commutators
of multilinear Calderón-Zygmund operators
with Dini type kernels
Zhengyang Li and Qingying Xue*
*
Correspondence:
School of Mathematical Sciences,
Laboratory of Mathematics and
Complex Systems, Ministry of
Education, Beijing Normal
University, Beijing, 100875, People’s
Republic of China
Abstract
Let Tb and Tb be the commutators in the jth entry and iterated commutators of the
multilinear Calderón-Zygmund operators, respectively. It was well known that the
commutators of linear Calderón-Zygmund operators were not of weak type (1, 1) and
(H1 , L1 ), but they did satisfy certain endpoint L log L type estimates. In this paper, our
aim is to give more natural sharp endpoint results. We show that Tb and Tb are
1
bounded from the product Hardy space H1 × · · · × H1 to weak L m ,∞ space, whenever
the kernel satisfies a class of Dini type condition. This was done by using a key lemma
given by Christ, a very complex decomposition of the integrand domains, and
carefully splitting the commutators into several terms.
Keywords: commutators; multilinear Calderón-Zygmund operator; C-Z kernel of ω
type; Dini type conditions; Hardy spaces
1 Introduction
1.1 Commutators of classical C-Z operators
In , Coifman, Rochberg, and Weiss [] first introduced and studied the commutator
of classical linear Calderón-Zygmund singular integrals, which was defined by
Tb f = [b, T]f = bT(f ) – T(bf ).
The Lp boundedness of Tb was given in [] for < p < ∞ when b ∈ BMO(Rn ). It is well
known that Tb fails to be of weak type (, ) and is not bounded from H (Rn ) to L (Rn ).
Counterexamples were given by Pérez [] and Paluszyński []. As an alternative result
of the weak (, ) estimate of Tb , Pérez [] obtained the following L(log L) type endpoint
estimate:
x ∈ Rn : Tb f (x) > λ ≤ C
|f (x)|
|f (x)|
+ log+
dx,
λ
λ
Rn
λ > .
Moreover, alternative results of the (H , L ) boundedness were also considered in the work
of Alvarez [], Pérez [], and Liang, Ky, and Yang [], which concerned with the boundedness of Tb on the subspace of atomic Hardy spaces, or concerned with the (Hw , Lw )
© 2016 Li and Xue. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License
(http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and
indicate if changes were made.
Li and Xue Journal of Inequalities and Applications (2016) 2016:252
Page 2 of 22
boundedness of Tb if b belongs to a subspace of BMO which is associated to the weight
function w.
On the other hand, another more reasonable and alternative result of weak type (, )
and (H , L ) estimate was given by Liu and Lu [] in . The authors [] showed that
Tb is bounded from H (Rn ) to L,∞ (Rn ) if b ∈ BMO(Rn ). We note that Tb also fails to be
bounded from H p (Rn ) to Lp,∞ (Rn ) for < p < by the generalized interpolation theorem
[], pp.. Therefore, the (H , L,∞ ) boundedness of Tb becomes a sharp endpoint estimate. Moreover, always L(log L)(Sn– ) H (Sn– ) if f vanishes on the unit sphere. However, there is no such inclusion relationship on Rn . Moreover, the inverse including relationship is still not true, since the following example shows that H (Rn ) L(log L)(Rn ).
Example . Let
f (x) =
χ[– , ]
aj (x) =
Thus, f (x) =
f (x)
{χ + χ[ , ] } × j ,
f ( j+
) [– j ,– j+ ]
j+ j
λj =
)
f ( j+
j
.
∞
∞
j= λj aj (x), and it is easy to verify that each aj is a (, ∞, )-atom. Notice that
∞
|λj | =
j=
for some ε > ,
x log+ε
|x|
j=
|f ( j+
)|
j
∞
≤
∞
·
=
< ∞,
+ε j+
j
j+ log
(j + )+ε
j=
j=
then we have f ∈ H (Rn ). Obviously, f ∈/ L(log L)(Rn ).
Thus, the (H , L,∞ ) boundedness and the L log L type estimate of Tb are independent in
the sense that one cannot cover the results of the other.
1.2 Commutators of multilinear operators
In recent years, the theory of multilinear Calderón-Zygmund operators with standard kernels have been developed very quickly and a lot of work has been done. Among such
achievements is the celebrated work of Coifman and Meyer [–], Christ and Journé
[], Kenig and Stein [], Grafakos and Torres [, ], and Lerner et al. []. In order to
state some well-known results, we need to introduce some definitions.
Definition . (C-Z kernel of ω type [, ]) Let ω(t) be a non-negative and nondecreasing function on R+ . Let K(x, y , . . . , ym ) be a locally integrable function defined away
from the diagonal x = y = · · · = ym in (Rn )m+ . Denote (x, y) = (x, y , . . . , ym ), we say K is an
m-linear Calderón-Zygmund kernel of ω type, if there exists a positive constant C such
that
K(x, y) ≤ m C
,
( j= |x – yj |)mn
|x – x |
K(x, y) – K x , y ≤ m C
,
ω
m
( j= |x – yj |)mn
j= |x – yj |
(.)
(.)
Li and Xue Journal of Inequalities and Applications (2016) 2016:252
Page 3 of 22
whenever |x – x | ≤ max≤j≤m |x – yj |, and
K(x, y , . . . , yi , . . . , ym ) – K x, y , . . . , y , . . . , ym
i
C
|yi – yi |
,
≤ m
ω m
( j= |x – yj |)mn
j= |x – yj |
(.)
whenever |yi – yi | ≤ max≤j≤m |x – yj |.
Definition . (Multilinear C-Z singular integral operators [, ]) Let K(x, y) be a CZ kernel of ω type. For any f = (f , . . . , fm ) ∈ S (Rn ) × S (Rn ) × · · · × S (Rn ) and all x ∈/
m
j= supp fj , we define the multilinear Calderón-Zygmund singular integral operators as
follows:
K(x, y , . . . , ym )f (y ), . . . , fm (ym ) dy · · · dym .
T(f)(x) =
(Rn )m
Definition . (Commutators of multilinear C-Z operators) Let bj ∈ BMO(Rn ) and T be
the operator defined in Definition .. The commutators in the jth entry and the iterated
commutators of T are defined by
Tb (f)(x) =
m
Tb (f)(x)
j
j=
m
bj (x)T(f , . . . , fj , . . . , fm )(x) – T(f , . . . , bj fj , . . . , fm )(x)
=
(.)
j=
and
Tb (f) = b , b , . . . bm– , [bm , T]m , m– · · · (f)
m
=
bj (x) – bj (yj ) K(x, y , . . . , ym )f (y ) · · · fm (ym ) dy.
(.)
(Rn )m j=
Remark . Obviously, in the special case, ω(t) = t ε for some ε > , then the operator T
defined in Definition . coincides with the standard multilinear Calderón-Zygmund operator defined and studied by Grafakos and Torres []. Moreover, if ω(t) = t ε , the weighted
j
strong and L(log L) type endpoint estimates for Tb (f , . . . , fm )(x) = m
j= Tb (f ) and Tb have
already been studied in [] and [], respectively.
Definition . (Dini(a) type conditions) Let ω(t) be a non-negative and non-decreasing
function on R+ . ω is said to satisfy the Dini( (...truncated)