Boundedness of Multilinear Calderón-Zygmund Operators on Grand Variable Herz Spaces
Abstract
In this paper, we prove the boundedness of multilinear Calderón-Zygmund operators on product of grand variable Herz spaces. These results generalize the boundedness of multilinear Calderón-Zygmund operators on product of variable exponent Lebesgue spaces and variable Herz spaces.
1. Introduction
There has been increased interest in the study of multilinear singular integral operators in recent years. The class of multilinear singular integrals with standard Calderón-Zygmund kernels provides the foundation and starting point of research of the theory. Such a class of multilinear Calderón-Zygmund operators was introduced and first studied by Coifman and Meyer [1–3] and later by Grafakos and Torres [4]. For the boundedness and other properties of multilinear fractional integrals, we refer to, e.g., [5–8].
Variable Lebesgue spaces were introduced in [9], but stayed under the radar for a considerable amount of time. Apart from some previous sporadic episodes, the research boom on such spaces can be traced back to the foundational paper [10]. Since then, these spaces have attracted much attention of mathematicians, not only because of their connection with harmonic analysis but also due to their usefulness in application to a wide range of problems, see, e.g., [11]. The standard references to the general theory of variable Lebesgue spaces are [12, 13].
The classical definition of Herz spaces was introduced in [14]. Many studies can be found related to these spaces and its variations, which include variable Herz spaces, continual Herz spaces, and Herz spaces with variable smoothness and integrability. For details, see [15–21] and references therein.
Grand Lebesgue spaces on bounded sets, which proved to be useful in application to partial differential equations, were introduced in [22, 23]. In the last years, various operators of harmonic analysis have been intensively studied on grand spaces, see for instance [20, 24–32]. Grand Lebesgue sequence spaces were introduced recently in [33], where several operators of harmonic analysis were studied, e.g., maximal, convolutions, Hardy, Hilbert, and fractional operators.
In this paper, we prove the boundedness of multilinear Calderón-Zygmund operators on grand variable Herz spaces which were introduced in [34]. The present paper is organized in the following way. Apart from the introduction, in Section 2, we recall some definitions and results related to variable exponent spaces. Section 3 contains some details about multilinear Calderón-Zygmund kernels and the proof of the main result.
Notations.
- (i)
ℕ is the set of natural numbers and ℕ0≔ℕ ∪ {0}
- (ii)
ℤ is the set of integers
- (iii)
ℤ− is the set of negative integers
- (iv)
for n, m ∈ ℤ and n < m
- (v)
B(x, r) is the ball of radius r center at the point x
- (vi)
Bk≔{x ∈ ℝn : |x| ≤ 2k} for all k ∈ ℤ
- (vii)
Rt,τ≔Bτ\Bt = {x : t<∣x∣<τ} is a spherical layer
- (viii)
Rk≔Bk\Bk−1
- (ix)
- (x)
f≲g means that f ≤ Cg and f≃g means that f≲g≲f
- (xi)
constants (often different constant in the same chain of inequalities) will mainly be denoted by c or C
2. Function Spaces with Variable Exponent
In this section, we recall definitions and results related to variable exponent Lebesgue spaces, variable Herz spaces, and grand variable Herz spaces.
2.1. Lebesgue Space with Variable Exponent
By q′, we denote the conjugate exponent of q, defined by q′(x) = q(x)/(q(x) − 1). In the sequel, we use
For the following lemma, we refer to, e.g., [12].
Lemma 1 (Generalized Hölder’s Inequality). Given , define by 1/r(x) = 1/p(x) + 1/q(x). Then, there exists a constant c such that for all f ∈ Lp(·)(ℝn) and g ∈ Lq(·)(ℝn), fg ∈ Lr(·)(ℝn) and
The following lemma appears in [21].
Lemma 2. Let D > 1 and Then
2.2. Herz Spaces with Variable Exponent
The classical Herz spaces were first introduced in [14]. We recall the definition of variable exponent Herz spaces.
2.3. Grand Lebesgue Sequence Space
In this section, we recall the definition of grand Lebesgue sequence space. For the following definition and statements, see [33]. In what follows, stands for one of the sets ℤn, ℤ, ℕ, and ℕ0.
Definition 4. Let 1 ≤ p < ∞ and θ > 0. The grand Lebesgue sequence space lp),θ is defined by the norm
2.4. Grand Variable Herz Space
Following [34], we now introduce the grand variable Herz spaces.
Definition 5. Let α ∈ L∞(ℝ), 1 ≤ p < ∞, q : ℝn⟶[1, ∞), and θ > 0. The homogeneous grand variable Herz space is defined by
The following lemma, see [15], is helpful to estimate the norm of characteristics functions.
Lemma 6. Let and let R = B(0, r)\B(0, (r/2)). If |R| ≤ 2−n, then
The left-hand side equivalence remains true for every |R| > 0 if we assume, additionally, that
3. Boundedness of Multilinear Calderón-Zygmund Operators
- (i)
The related kernel belongs to m-CZK(c, ε) class
- (ii)
T is bounded from to Lq for some 1 < q1, q2, ⋯, qm < ∞, and 1/q = 1/q1 + ⋯+1/qm
Grafakos and Torres [4] proved the boundedness of T from to Lq,∞ for some 1 < q1, q2, ⋯, qm < ∞, and 1/q = 1/q1 + ⋯+1/qm, and from L1 × ⋯×L1 to L1/m,∞.
The boundedness of the multilinear Calderón-Zygmund operator T on variable exponent Lebesgue spaces was proved in [37], as stated below.
Lemma 7. Let , , with 1/q(x) = 1/q1(x) + ⋯+1/qm(x), and for some 0 < s < q−. Then, the m-linear Calderón-Zygmund operator T is bounded on the product of variable exponent Lebesgue spaces. Moreover,
We now state and prove the boundedness of multilinear Calderón-Zygmund operator on grand variable Herz spaces.
Theorem 8. Let , , such that qi(0) ≤ qi(∞), 1/q(x) = 1/q1(x) + ⋯+1/qm(x), , and for some 0 < s < q−. Let θ > 0, 1 < pi < ∞, and αi ∈ L∞(ℝ) be log-Hölder continuous both at the origin and at infinity for with
Suppose that and Then, the m-linear Calderón-Zygmund operator T is bounded on the product of grand variable Herz spaces. Moreover,
Proof. We restrict ourselves to m = 2, the general case following in a similar manner. Defining and , we decompose the component functions of as
For future usage, we divide ℤ into the following sets
From Definition 5 and (26), we have
It is necessary to estimate I1, I2, I3, I5, I6, and I9, since I4, I7, and I8 can be obtained in a similar manner as I2, I3, and I6, respectively. Estimation for I1: splitting ℤ = ℤ− ∪ ℕ0 and by the asymptotic 2kα(x)≃2kα(0) (x ∈ Rk and k < 0) and 2kα(x)≃2kα(∞) (x ∈ Rk and k ≥ 0), we get
For φ, σ ∈ Lk, x ∈ Rk, y1 ∈ Rφ, and y2 ∈ Rσ, we have
From estimate (35), Hölder’s inequality, 1/q(x) = 1/q1(x) + 1/q2(x), and Lemma 2, we obtain
Taking into account the previous estimate for νk(Lk, Lk), the equality 1p = 1p1 + 1p2, and Hölder’s inequality, we have
By Hölder’s inequality, Fubini’s theorem for series, , and defining b1≔n/q1(0) − n + α1(0) < 0, we obtain
The estimate is obtained, mutatis mutandis, via the estimation for I111. With this estimates at hand, we obtain
To estimate I12, we split as follow
The estimate I122 follows in similar manner as in I11 with simply replaced αi(0) by αi(∞) and used the fact qi(0) ≤ qi(∞).
For estimate I121, by Hölder’s inequality, Lemma 2, and the inequality qi(0) ≤ qi(∞), we obtain
From the estimate of νk(ℤ−, ℤ−), the equality 1/p = 1/p1 + 1/p2, and Hölder’s inequality, we have
Invoking the Hölder inequality and defining ξ1≔n/q1(0) − n + α1(∞) < 0, we have
Similar estimate, with the corresponding changes, is obtained for A2, from which we obtain Hence
Estimation for I2: as in the case of I1, we obtain the following estimate
Notice that, for x ∈ Rk, y1 ∈ Rφ, y2 ∈ Rσ, φ ∈ Lk, and σ ∈ Mk, we have
From the estimate for νk(Lk, Mk) and Hölder’s inequality, we get
Notice that I211 = I111. For the estimate I212, we reason as follows
The term I22 is estimated by
To estimate I221, by Hölder’s inequality, Lemma 2, and the inequality q1(0) ⩽ q1(∞), we have
Thus using (51) and by Hölder’s inequality, we have
The term B1 is equal to A1 and for B2 we use similar arguments as for I212, replacing α2(0) with α2(∞).
For the term I222, by Hölder’s inequality and Lemma 2, we have
Taking into consideration (53) and applying Hölder’s inequality, we get
Noting that and estimating in a similar fashion as I212, we obtain , from which we get
Estimation for I3: we have
For φ ∈ Lk, σ ∈ Nk, x ∈ Rk, y1 ∈ Rφ, and y2 ∈ Rσ, we get
By Hölder’s inequality, Lemma 2, q2(0)≲q2(∞), and k ∈ ℤ−, we obtain
To estimate the term I31, using (59) and the Hölder inequality, we obtain
We have , since I311 = I111.
As for I312, we split as follow
By Hölder’s inequality, Fubini’s theorem for series, the inequality n/q2(∞) + α2(0) > 0 and , we obtain
For D2, using the same argument as above and the inequality α2(0) ⩽ α2(∞), we obtain
So
The term I32 can be estimated as
For k ∈ ℕ0, φ ∈ ℤ−, σ ∈ Nk, q1(0)≲q1(∞), and Lemma 2, we obtain
By (65) and Hölder’s inequality, we get
Note that The estimate for can be obtained in a similar way as for I312, by replacing α2(0) with α2(∞).
For estimate I322, by Hölder’s inequality and Lemma 2, we have
Using (67) and Hölder’s inequality, we obtain
We have and the estimate for the term is similar to Taking all the estimates into account yields
Estimation for I5: We have
By the Lq(·)-boundedness of T, see Lemma 7, we obtain that
By Hölder’s inequality, we have
Similarly, we can obtain similar estimate for I52, replacing αi(0) by αi(∞) with k ∈ ℕ0. Therefore
Estimation for I6: we have
For k ∈ ℤ−, φ ∈ Mk, σ ∈ Nk, and x ∈ Rk, we have
By Hölder’s inequality, Lemma 2, k ∈ ℤ−, and inequality q2(0) ⩽ q2(∞), we obtain
By (77) and invoking Hölder’s inequality, we get
Note that the estimate I611 is equal to that of I51 and I612 is obtained by similar argument used in I312.
For k ∈ ℕ0, we have
Using (79) and Hölder’s inequality, we get
The estimate I621 is similar to I52 and Hence
Estimation for I9: we estimate I9 as usual
For φ ∈ Nk, y1 ∈ Rφ, and x ∈ Rk, we have ∣x − y1 | 2φ, whereas for σ ∈ Nk, y2 ∈ Rσ, and x ∈ Rk, we have ∣x − y2 | 2σ. Under the assumptions for the obtained inequalities, we have
When k ∈ ℤ−, by Hölder’s inequality, Lemma 2, and qj(0) ⩽ qj(∞), we obtain
Thus
The estimates for the terms I911 and I912 are obtained in the same way as for I312. Therefore
Finally, for the term I92, we have
The terms I921 and I922 can be estimated in a similar manner as for that I322. Thus
Taking into account the estimates (44), (55), (69), (73), (81), and (88), we get
Conflicts of Interest
The authors declare that there is no conflict of interest.
Authors’ Contributions
All the authors have contributed equally in preparation and finalization of the manuscript.
Acknowledgments
We gratefully acknowledge the constructive comments of the anonymous referees.
Open Research
Data Availability
No data were used to support this study.