Hindawi Publishing Corporation Journal of Inequalities and Applications Volume 2011, Article ID 479576, 14 pages doi:10.1155/2011/479576 Research Article Hypersingular Marcinkiewicz Integrals along Surface with Variable Kernels on Sobolev Space and Hardy-Sobolev Space Wei Ruiying and Li Yin School of Mathematics and Information Science, Shaoguan University, Shaoguan 512005, China Correspondence should be addressed to Wei Ruiying, weiruiying521@163.com Received 30 June 2010; Revised 5 December 2010; Accepted 20 January 2011 Academic Editor: Andrei Volodin Copyright q 2011 W. Ruiying and L. Yin. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Let α ≥ 0, the authors introduce in this paper a class of the hypersingular Marcinkiewicz integrals along surface with variable kernels defined by μΦ Ω,α f x ∞ 1/2 2 n−1 32α ∞ n q n−1 0 | |y|≤t Ωx, y/|y| f x − Φ|y|y dy| dt/t , where Ωx, z ∈ L Ê × L Ë with q > max{1, 2n − 1/n 2α}. The authors prove that the operator μΦ Ω,α is bounded from p p n p n Sobolev space Lα Ê to L Ê space for 1 < p ≤ 2, and from Hardy-Sobolev space Hα Ên to p n 2 n L Ê space for n/n α < p ≤ 1. As corollaries of the result, they also prove the L̇α R − L2 Rn boundedness of the Littlewood-Paley type operators μΦ and μ∗,Φ which relate to the Lusin Ω,α,λ Ω,α,S ∗ area integral and the Littlewood-Paley gλ function. 1. Introduction Let Ên n ≥ 2 be the n-dimensional Euclidean space and Ën−1 be the unit sphere in Ên equipped with the normalized Lebesgue measure dσ dσ·. For x ∈ Ên \ {0}, let x x/|x|. Before stating our theorems, we first introduce some definitions about the variable kernel Ωx, z. A function Ωx, z defined on Ên × Ên is said to be in L∞ Ên × Lq Ën−1, q ≥ 1, if Ωx, z satisfies the following two conditions: 1 Ωx, λz Ωx, z, for any x, z ∈ Ên and any λ > 0; 1/q 2 ΩL∞ Ên×Lq Ën−1 supr≥0, y∈Ên Ën−1 |Ωrz y, z |q dσz < ∞. In 1955, Calderón and Zygmund 1 investigated the Lp boundedness of the singular integrals TΩ with variable kernel. They found that these operators connect closely with the 2 Journal of Inequalities and Applications problem about the second-order linear elliptic equations with variable coefficients. In 2002, Tang and Yang 2 gave Lp boundedness of the singular integrals with variable kernels associated to surfaces of the form {x Φ|y|y }, where y y/|y| for any y ∈ Ên \ {0} n ≥ 2. That is, they considered the variable Calderón-Zygmund singular integral operator TΩΦ defined by TΩΦ f x Ω x, y p·v· n f x − Φ y y dy. Ên y 1.1 On the other hand, as a related vector-valued singular integral with variable kernel, the Marcinkiewicz singular with rough variable kernel associated with surfaces of the form {x Φ|y|y } is considered. It is defined by μΦ Ω f x ∞ 2 dt 1/2 Φ , FΩ,t x 3 t 0 1.2 where Φ FΩ,t x |y|≤t Ë n−1 Ω x, y n−1 f x − Φ y y dy, y Ω x, z dσ z 0. 1.3 1.4 If Φ|y| |y|, we put μΦ Ω μΩ . Historically, the higher dimension Marcinkiewicz integral operator μΩ with convolution kernel, that is Ωx, z Ωz, was first defined and studied by Stein 3 in 1958. See also 4–6 for some further works on μΩ with convolution kernel. Recently, Xue and Yabuta 7 studied the L2 boundedness of the operator μΦ Ω with variable kernel. Theorem 1.1 see 7. Suppose that Ωx, y is positively homogeneous in y of degree 0, and satisfies 1.4 and 1/q 2 supy∈Ên Ën−1 |Ωy, z |q dσz < ∞, for some q > 2n − 1/n. Let Φ be a positive and monotonic (or negative and monotonic) C1 function on 0, ∞ and let it satisfy the following conditions: i δ ≤ |Φt/tΦ t| ≤ M for some 0 < δ ≤ M < ∞; ii Φ t is monotonic on 0, ∞. Then there is a constant C such that μΦ Ω f2 ≤ Cf2 , where constant C is independent of f. Since the condition 2 implies 2 , so the L2 Ên boundedness of μΦ Ω holds if Ω ∈ L Ê × Lq Ën−1 with q > 2n − 1/n. Our aim of this paper is to study the hypersingular Marcinkiewicz integral μΦ Ω,α along surfaces with variable kernel Ω, and with index α ≥ 0, on the homogeneous Sobolev space ∞ n Journal of Inequalities and Applications 3 Lα Ên for 1 < p ≤ 2 and the homogeneous Hardy-Sobolev space Hα Ên for some n/nα < Φ x be as above, we then define the operators μΦ p ≤ 1. Let FΩ,t Ω,α by p p μΦ Ω,α f x ∞ 2 dt 1/2 Φ , FΩ,t x 32α t 0 α ≥ 0. 1.5 Our main results are as follows. Theorem 1.2. Suppose that α ≥ 0, Ωx, y satisfies 1.4 and Ω ∈ L∞ Ên × Lq Ën−1 with q > max{1, 2n − 1/n 2α}. Let Φ be a positive and increasing C1 function on 0, ∞ and let it satisfy the following conditions: i Φt tΦ t; ii 0 ≤ Φ t ≤ W on 0, ∞. Then there is a constant C such that μΦ Ω,α fL2 Ên ≤ CfL2α Ên , where constant C is independent of f. Theorem 1.3. Suppose 0 < α < n/2, and that Ω ∈ L∞ Ên × Lq Sn−1 , with q > max{1, 2n − 1/n 2α}, and satisfies 1.4. Let Φ be a positive and increasing C1 function on 0, ∞ and let it satisfy the following conditions: i Φt tΦ t; ii 0 < Φ t ≤ 1, Φ0 0. p Then, for n/n α < p ≤ 1, there is a constant C such that μΦ Ω,α fLp Ên ≤ CfHα Ên , where constant C is independent of any f ∈ Hα Ên ∩ SÊn . p Furthermore, our result can be extended to the Littlewood-Paley type operators μΦ Ω,α,S μ∗,Φ Ω,α,λ and with variable kernels and index α ≥ 0, which relate to the Lusin area integral and Φ x be as above, we then define the the Littlewood-Paley gλ∗ function, respectively. Let FΩ,t ∗,Φ n operators μΦ Ω,α,S and μΩ,α,λ for f ∈ SÊ , respectively by μΦ Ω,α,S f x ⎛ ∗,Φ μΩ,α,λ f x ⎝ with λ > 1, where Γx {y, t ∈ have the following conclusion. Φ 2 dydt FΩ,t y n32α t Γx λn 1/2 , ⎞1/2 Φ 2 dydt ⎠ , FΩ,t y n32α t 1.6 Ê t t x − y Ên1 : |x − y| < t}. As an application of Theorem 1.2, we n1 Theorem 1.4. Under the assumption of Theorem 1.2, then Theorem 1.2 still holds for μΦ Ω,α,S and μ∗,Φ Ω,α,λ . By Theorems 1.2 and 1.3 and applying the interpolation theorem of sublinear operator, p we obtain the Lα − Lp boundedness of μΦ Ω,α . 4 Journal of Inequalities and Applications Corollary 1.5. Suppose 0 < α < n/2, and that Ω ∈ L∞ Ên × Lq Sn−1 , q > max{1, 2n − 1/n 2α}, and satisfies 1.4. Let Φ be given as in Theorem 1.3. Then, for 1 < p ≤ 2, there exists an absolute positive constant C such that Φ μΩ,α f Ê Lp n ≤ Cf Lpα Ên , 1.7 for all f ∈ Lα Ên ∩ SÊn . p Remark 1.6. It is obvious that the conclusions of Theorem 1.2 are the substantial improvements and extensions of Stein’s results in 3 about the Marcinkiewicz integral μΩ with convolution kernel, and of Ding’s results in 8 about the Marcinkiewicz integral μΩ with variable kernels. Remark 1.7. Recently, the authors in 9 proved the boundedness of hypersingular p Marcinkiewicz integral with variable kernels on homogeneous Sobolev space Lα Rn for 1 < p ≤ 2 and 0 < α < 1 without any smoothness on Ω. So Corollary 1.5 extended the results in 9, Theorem 5. Throughout this paper, the letter C always remains to denote a positive constant not necessarily the same at each occurrence. 2. The Bounedness on Sobolev Spaces Before giving the definition of the Sobolev space, let us first recall the Triebel-Lizorkin space. Fix a radial function ϕx ∈ C∞ satisfying suppϕ ⊆ {x : 1/2 < |x| ≤ 2} and 0 ≤ ϕx ≤ 1, and ϕx > c > 0 if 3/5 ≤ |x| ≤ 5/3. Let ϕj x ϕ2j x. Define the function ψj x by Fψj ξ ϕj ξ, such that Fψj ∗ fξ Ffξϕj ξ. α,q For 0 < p, q < ∞, and α ∈ Ê, the homogeneous Triebel-Lizorkin space Ḟp is the set of all distributions f satisfying ⎫ ⎧ 1/q ⎪ ⎪ ⎬ ⎨ q α,q n n −αk Ḟp Ê f ∈ S Ê : f Ḟpα,q <∞ . 2 ψk ∗ f ⎪ ⎪ k ⎭ ⎩ 2.1 p For p ≥ 1, the homogeneous Sobolev spaces Lα Ên is defined by Lα Ên Ḟpα,2 Ên , namely fLpα fḞpα,2 . From 10 we know that for any f ∈ L2α Ên p f ∼ L2α Ên Ên p 2 2α F f ξ |ξ| dξ 1/2 , 2.2 and if α is a nonnegative integer, then for any f ∈ Lα Ên p f p ∼ Ên Lα D τ f |τ|α Ên . Lp 2.3 Journal of Inequalities and Applications 5 For 0 < p ≤ 1, we define the homogeneous Hardy-Sobolev space Hα Ên by Hα Ên Ê It is well known that H p Ên Ḟp0,2 Ên for 0 < p ≤ 1, one can refer 10 for the details. Next, let us give the main lemmas we will use in proving theorems. p p Ḟpα,2 n . Lemma 2.1 see 11. Suppose that n ≥ 2 and f ∈ L1 Ên ∩ L2 Ên has the form fx f0 |x|P x where P x is a solid spherical harmonic polynomial of degree m. Then the Fourier transform of f has the form Ffx F0 |x|P x, where F0 r 2πi−m r −n2m−2/2 ∞ f0 sJn2m−2/2 2πrssn2m/2 ds, 2.4 0 and r |ξ|, Jm s is the Bessel function. Lemma 2.2 see 12. For λ n − 2/2, and −λ ≤ α ≤ 1, there exists C > 0 such that for any h ≥ 0 and m 1, 2, . . ., h J t C mλ dt ≤ λα . m 0 tλα 2.5 Lemma 2.3. Let α ≥ 0, λ n − 2/2, Φ is a C1 function on 0, ∞ and let it satisfy the conditions (i) and (ii) in Theorem 1.2. ∞ 1/2 Denote gα fx 0 |Nε fx|2 dε/ε12α , if F Nε f ξ Φε|ξ| 0 Jmλ t dt · F f ξ. tλ1 2.6 Then there exists a constant C independent of m, such that gα fL2 ≤ C/mλ1α fL2α for every integer m ∈ Æ , m > α. Proof. Let η|x| |x| 0 Jmλ t/tλ1 dt, then we have 2 g α f 2 Ên ∞ 0 ∞ Nε fx2 dε dx ε12α ηΦε|ξ|F f ξ2 dξ dε ε12α 0 Ê ∞ 2 η Φ β |ξ| F f ξ |ξ|2α dξ dβ |ξ| β12α 0 Ên ∞ 2 η Φ β |ξ| dβ F f ξ2 |ξ|2α dξ. 12α |ξ| β Ên 0 n ∞ 2 ηΦβ/|ξ||ξ|2 dβ/β12α ≤ C/mλ1α . ∞ m/2 ∞ Decompose this integral into two parts 0 0 m/2 : I1 I2 . So it suffices to show 0 2.7 6 Journal of Inequalities and Applications For I2 , by using Lemma 2.2 and Φt tΦ t, we can get I2 ∞ Φβ/|ξ||ξ| m/2 ≤ ≤ 0 C m2λ2 ∞ m/2 Jmλ t dt tλ1 2 dβ β12α dβ β12α 2.8 C . m2λ22α For the other part I1 , applying Stirling’s formula, we have √ √ 2πxx−1/2 e−x ≤ Γx ≤ 2 2πxx−1/2 e−x . 2.9 Also in 13, the authors proved the following inequality |Jν t| ≤ t/2ν . Γν 1 2.10 So by 2.9 and 2.10, 0 ≤ α < α 1 ≤ m, and noting that Φt ≤ Wt, we have I1 m/2 Φβ/|ξ||ξ| 0 ≤ 0 m/2 Φβ/|ξ||ξ| 0 0 Jmλ t dt tλ1 2 |Jmλ t| dt tλ1 1 ≤ 2m2λ 2 2 Γ m λ 1 dβ β12α 2 dβ β12α m/2 Φβ/|ξ||ξ| 0 0 tmλ dt tλ1 2 dβ β12α m/2 2m dβ β Φ |ξ| β12α 0 m/2 2m dβ β e2m2λ2 Φ ≤ 12α−2m 2m2λ1 |ξ| β 2π22m2λ m λ 1 0 ≤ 1 22m2λ Γ2 m λ 1 ≤C ≤C ≤ m/2 e2m2λ2 2π22m2λ m e 2m2λ2 4 2m2λ1 λ 1 0 dβ β12α−2m 1 m2α2λ2 C . m2α2λ2 So far we can deduce the desired conclusion of Lemma 2.3. 2.11 Journal of Inequalities and Applications 7 Proof of Theorem 1.2. The basic idea of proof can go back to 14, for recently papers, one see 8, 15. By the same argument as in 1, let {Ym,j } m ≥ 1, j 1, 2, . . . , Dm denote the complete system of normalized surface spherical harmonics. See 14 for instance, we can decompose Ωx, y as following: Dm ∞ Ω x, y am,j xYm,j y is a finite sum. 2.12 m1 j1 Denote ⎞1/2 ⎛ Dm 2 am x ⎝ am,j x ⎠ , bm,j x j1 am,j x , am x 2.13 then we get Dm 2 bm,j x 1, Dm ∞ am x bm,j xYm,j y . Ω x, y j1 m1 2.14 j1 Then, applying Hölder inequality twice, we have for any 0 < ε < 1 that 2 ∞ 2 ∞ Ym,j y dt Φ bm,j x n−1 f x − Φ y y dy 32α μΩ,α fx t y 0 |y|≤t m1 ≤ ∞ a2m xm−ε12α m1 ∞ mε12α m1 2 ∞ Dm Ym,j y dt × b f x − Φ y dy y x m,j n−1 t32α y 0 |y|≤t j1 ≤ ∞ a2m xm−ε12α m1 ∞ mε12α m1 ∞ 0 ⎛ ⎝ Dm ⎞ 2 bm,j x⎠ j1 2 Dm Ym,j y dt f x − Φ y y dy 32α × t |y|≤t yn−1 j1 ∞ a2m xm−ε12α m1 ∞ mε12α m1 2 ∞ Dm Ym,j y dt × f x − Φ y y dy 32α . t |y|≤t yn−1 0 j1 2.15 8 Journal of Inequalities and Applications By 14, page 230, equation 4.4, we can observe that the series in the first parenthesis on the right-hand side of the inequality above, for each x fixed, is equal to Ωx, ·2L2 Ën−1 , −γ where L2−γ Ën−1 is the Sobolev space on Ën−1 with γ ε1/2 α for any 0 < ε < 1. So if we take ε sufficiently close to 1, then by the Sobolev imbedding theorem Lq ⊂ L2−γ , we have 1/2 2 −ε12α am xm ≤ CΩL∞ Ên×Lq Ën−1 : CΩ 2.16 m with q > max{1, 2n − 1/n 2α}. By Fourier transform and 2.16, we get 2 Dm Ym,j y dt n−1 f x − Φ y y dy dx 32α t n 0 Ê j1 |y|≤t y m1 2 Dm ∞ ∞ y Y m,j 2 y y dy ξ dξ dt ≤ CΩ mε12α f · − Φ F n−1 t32α Ên |y|≤t y m1 j1 0 ∞ ∞ Φ 2 μΩ,α f ≤ CΩ2 mε12α 2 ∞ : CΩ2 mε12α m1 Dm Φ 2 μΩ,j,α f . j1 2 2.17 For μΦ Ω,j,α f, we have 2 Ym,j y −2πix·ξ dt y e f x − Φ y dx dy dξ 32α n−1 t 0 Ên |y|≤t Ên y ∞ Ym,j y −2πiΦ|y|y ·ξ e 0 Ên |y|≤t yn−1 2 ∞ Φ μΩ,j,α f 2 2 dt f x − Φ y y e−2πix−Φ|y|y ·ξ dx dy dξ 32α t n Ê 2 ∞ Ym,j y −2πiΦ|y|y ·ξ 2 dt dy F f ξ dξ 32α n−1 e t n 0 Ê |y|≤t y 2 ∞ Ym,j y −2πiΦ|y|y ·ξ dt 2 1 F f ξ dξ. e dy t Ên 0 t1α |y|≤t yn−1 × 2.18 Journal of Inequalities and Applications 9 For the integral on the right hand side of the above inequality, by changing of variable, we can get 1 t1α |y|≤t Ym,j y −2πiΦ|y|y ·ξ dy n−1 e y t 1 t1α Ën−1 0 Ym,j y e−2πiΦsy ·ξ dy ds Φt 1 t1α Ën−1 0 1 t1α |y|≤Φt Ym,j y e−2πiγ y ·ξ Φ−1 γ dy dγ 2.19 Ym,j y −2πiy·ξ −1 Φ y dy. n−1 e y So we have 2 Φ μΩ,j,α f 2 2 ∞ Ym,j y −2πiy·ξ −1 dt 2 1 F f ξ dξ. y Φ e dy t Ên 0 t1α |y|≤Φt yn−1 2.20 Φ,m,j Put Pm,j x Ym,j x |x|m and ϕt,α we can deduce from Lemma 2.1 that x Pm,j x · |x|−n−m1 χ|x|≤Φt xΦ−1 |x| t−1−α , Φ,m,j F ϕt,α ξ Pm,j |ξ| · F0 |ξ| Ym,j ξ · |ξ|m F0 |ξ|, 2.21 where −m −n/2−m1 F0 r 2πi Φt r t−1−α s−n−m1 Φ−1 s Jn/2m−1 2πrssn/2m ds 0 −m −n/2−m1 −1−α 2πi r Φt t 0 s−n/21 Jn/2m−1 2πrsd Φ−1 s Jn/2m−1 2πrΦ β 2πi r t dβ n/2−1 0 Φ β −α t J n/2m−1 2πrΦ β n/2 −m −m t i r dβ. 2π t 0 2πrΦβn/2−1 −m −n/2−m1 −1−α t 2.22 10 Journal of Inequalities and Applications Hence, we have Φ 2 μΩ,j,α f ∞ 0 ∞ 2 2 dt Φ,m,j ϕt,α ∗ fx dx t Ên 2 dt Φ,m,j F ϕt,α ∗ f ξ dξ t 0 Ên 2 ∞ −α t J m −m −m n/2m−1 2π|ξ|Φ β dt 2 n/2 t ≤ F f ξ dξ dβ Ym,j ξ |ξ| i |ξ| 2π n/2−1 t t 0 Ên 0 2π|ξ|Φ β 2 ∞ 1 t Jn/2m−1 2π|ξ|Φ β dt 2 ≤C dβ Ym,j ξ 12α F f ξ dξ. n/2−1 t t 0 Ên 0 2π|ξ|Φ β 2.23 By 14, we know that So we can get Dm 2 Φ μΩ,j,α f ≤ Cmn−2 j1 2 Dm j1 |Ym,j z |2 ∼ mn−2 . 2 ∞ t Jn/2m−1 2π|ξ|Φ β dt 2 1 dβ 12α F f ξ dξ. n/2−1 t Ên 0 t 0 2π|ξ|Φ β 2.24 Set λ n/2 − 1, ρ 2π|ξ|Φβ and note that Φt tΦ t, we can deduce that Jn/2m−1 2π|ξ|Φ β n/2−1 dβ 0 2π|ξ|Φ β 1 2π|ξ|Φt Jmλ ρ 1 dρ t 0 ρλ 2π|ξ|Φ Φ−1 ρ/2π|ξ| ρ 1 2π|ξ|Φt Jmλ ρ −1 dρ. Φ t 0 2π|ξ| ρλ1 1 U : t t 2.25 Noting that Φt is increasing, by using the second mean-value theorem, we get, for some 0 ≤ η < 2π|ξ|Φt, 2π|ξ|Φt 1 ρ J mλ dρ |U| ≤ Φ−1 Φt λ1 t ρ η 2π|ξ|Φt Jmλ ρ ≤ dρ. 0 ρλ1 2.26 Journal of Inequalities and Applications 11 From 2.26, it follows that Dm Φ 2 n−2 μm,j,α f ≤ Cm 2 j1 2 ∞ 2π|ξ|Φt dt Jmλ ρ dρ · F f ξ 12α dξ. t ρλ1 Ên 0 0 2.27 Thus using Lemma 2.3, we can deduce the desired conclusion of Theorem 1.2. λn ∗,Φ Proof of Theorem 1.4. First, we know that μΦ Ω,α,S fx ≤ 2 μΩ,α,λ fx. On the other hand, 2 ∗,Φ μΩ,α,λ f 2 2 λn dzdt 1 Ωx, z dz f x − Φ|z|z n12α dx n−1 t n1 t n Ê R |z|≤t |z| ⎛ 2 λn ⎞ ∞ dzdt 1 t Ωx, z 1 ⎠ ⎝ dz dx f x − Φ|z|z 12α n n−1 t t t n n 0 Ê Ê t x−y |z|≤t |z| t t x − y 2 ≤ CμΦ Ω,α f . 2 2.28 Thus, using Theorem 1.2, we can finish Theorem 1.4. 3. The Bounedness on Hardy-Sobolev Spaces In order to prove the boundedness for operator μΦ Ω,α on Hardy-Sobolev spaces and prove Theorem 1.3, we first introduce a new kind of atomic decomposition for Hardy-Sobolev space as following which will be used next. Definition 3.1 see 16. For α ≥ 0, the function ax is called a p, 2, α atom if it satisfies the following three conditions: 1 suppa ⊂ B with a ball B ⊂ Ên ; 2 aL2α ≤ |B|1/2−1/p ; 3 Ên axP x 0, for any polynomial P x of degree ≤ N n1/p − 1α. By 16, we have that every f ∈ Hα Ên can be written as a sum of p, 2, α atoms aj x, that is, p f λj aj . j 3.1 12 Journal of Inequalities and Applications Proof of Theorem 1.3. Similar to the argument of Lemma 3.3 in 17 and using above atomic decomposition, it suffices to show that p Φ μΩ,α a p ≤ C, 3.2 L with the constant C independent of any p, 2, α atom a. Assume suppa ⊂ B0, R. We first note that p Φ μΩ,α a p ≤ L |x|≤8R p Φ μΩ,α ax dx |x|>8R p Φ μΩ,α ax dx 3.3 : U1 U2 . For U1 , using Theorem 1.2, it is not difficult to deduce that p p n1−p/2 U1 ≤ CμΦ ≤ CaL2 Rn1−p/2 a 2R Ω,α L α 3.4 ≤ CRnp/2−1 Rn1−p/2 ≤ C. For U2 , we first consider the case n/n α < p < 1, according to 15, Lemma 5.5, for 0 < α < n/2 and p, 2, α atom a with support B B0, R, one has |ax|dx ≤ CRn−n/pα . 3.5 B Using Minkowski inequality and Hölder inequality for integrals, and 3.5, we can get U2 |x|>8R p Φ μΩ,α ax dx ⎞p/2 2 ∞ dt Ω x, y ⎝ n−1 a x − Φ y y dy 32α ⎠ dx t 0 |y|≤t y |x|>8R ⎛ p Ω x, y ≤ nα a x − Φ y y dy dx. y |x|>8R Ên 3.6 Journal of Inequalities and Applications 13 For the integral on the right hand side of the above inequality, by changing of variable and noting that 0 < Φ t ≤ 1, Φ0 0, we can get Ω x, y nα a x − Φ y y dy Ên y R Ω x, y a x − Φry dr dy 1α r Ën−1 0 ΦR Ω x, y a x − γ y 1 dγ dy 1α Φ Φ−1 γ Ën−1 0 Φ−1 γ ΦR Φ−1 γ Ω x, y dγ dy −1 1α a x − γ y γ Ën−1 0 Φ γ ΦR Ω x, y −1 α a x − γ y dγ dy Φ γ γ Ën−1 0 Ω x, y n α a x − y dy Φ−1 y |y|≤ΦR y Ω x, x − y α a y dy n −1 x − y |x−y|≤ΦR x − y Φ Ω x, x − y a y dy. ≤ x − ynα |x−y|≤ΦR 3.7 By 3.7, we can get U2 ≤ Ω x, x − y p a y dy dx x − ynα 2j R<|x|<2j1 R Ên ∞ j3 p |Ωx, x − y| ≤ nα a y dy dx 2j R<|x|<2j1 R Ên x − y j3 p ∞ n1−p Ω x, x − y j a y ≤ dx dy 2R x − ynα B 2j R<|x|<2j1 R j3 ∞ n1−p 2R j p ≤ CΩL∞ ×L1 a y dy 3.8 p ∞ −αp n1−p 2j R 2j R · . B j3 Thus by 3.5 and the condition p > n/n α, p U2 ≤ CΩL∞ ×L1 ∞ j3 2jn−np−αp ≤ C. 3.9 14 Journal of Inequalities and Applications As for p 1, similar to the argument of n/n α < p < 1, we can easily get U2 ≤ C. So far the proof of Theorem 1.3 has been finished. Acknowledgments This project supported by the National Natural Science Foundation of China under Grant no. 10747141, Zhejiang Provincial National Natural Science Foundation of China under Grant no. Y604056, and Science Foundation of Shaoguan University under Grant no. 200915001. References 1 A. P. Calderón and A. Zygmund, “On a problem of Mihlin,” Transactions of the American Mathematical Society, vol. 78, pp. 209–224, 1955. 2 L. Tang and D. Yang, “Boundedness of singular integrals of variable rough Calderón-Zygmund kernels along surfaces,” Integral Equations and Operator Theory, vol. 43, no. 4, pp. 488–502, 2002. 3 E. M. Stein, “On the functions of Littlewood-Paley, Lusin, and Marcinkiewicz,” Transactions of the American Mathematical Society, vol. 88, pp. 430–466, 1958. 4 X. X. Tao and R. Y. Wei, “Boundedness of commutators related to Marcinkiewicz integrals with variable kernels in Herz-type Hardy spaces,” Acta Mathematica Scientia, vol. 29, no. 6, pp. 1508–1517, 2009. 5 D. Fan and S. Sato, “Weak type 1, 1 estimates for Marcinkiewicz integrals with rough kernels,” The Tohoku Mathematical Journal, vol. 53, no. 2, pp. 265–284, 2001. 6 M. Sakamoto and K. Yabuta, “Boundedness of Marcinkiewicz functions,” Studia Mathematica, vol. 135, no. 2, pp. 103–142, 1999. 7 Q. Xue and K. Yabuta, “L2 -boundedness of Marcinkiewicz integrals along surfaces with variable kernels: another sufficient condition,” Journal of Inequalities and Applications, vol. 2007, Article ID 26765, 14 pages, 2007. 8 Y. Ding, C.-C. Lin, and S. Shao, “On the Marcinkiewicz integral with variable kernels,” Indiana University Mathematics Journal, vol. 53, no. 3, pp. 805–821, 2004. 9 J. Chen, Y. Ding, and D. Fan, “Littlewood-Paley operators with variable kernels,” Science in China Series A, vol. 49, no. 5, pp. 639–650, 2006. 10 M. Frazier, B. Jawerth, and G. Weiss, Littlewood-Paley Theory and the Study of Function Spaces, vol. 79 of CBMS Regional Conference Series in Mathematics, American Mathematical Society, Washington, DC, USA, 1991. 11 E. M. Stein and G. Weiss, Introduction to Fourier Analysis on Euclidean Spaces, Princeton University Press, Princeton, NJ, USA, 1971, Princeton Mathematical Series, No. 3. 12 Y. Ding and R. Li, “An estimate of Bessel function and its application,” Science in China Series A, vol. 38, no. 1, pp. 78–87, 2008. 13 N. E. Aguilera and E. O. Harboure, “Some inequalities for maximal operators,” Indiana University Mathematics Journal, vol. 29, no. 4, pp. 559–576, 1980. 14 A.-P. Calderón and A. Zygmund, “On singular integrals with variable kernels,” Applicable Analysis, vol. 7, no. 3, pp. 221–238, 1977-1978. 15 J. Chen, D. Fan, and Y. Ying, “Certain operators with rough singular kernels,” Canadian Journal of Mathematics, vol. 55, no. 3, pp. 504–532, 2003. 16 Y.-S. Han, M. Paluszyński, and G. Weiss, “A new atomic decomposition for the Triebel-Lizorkin spaces,” in Harmonic Analysis and Operator Theory (Caracas, 1994), vol. 189 of Contemporary Mathematics, pp. 235–249, American Mathematical Society, Providence, RI, USA, 1995. 17 S. Hofmann and S. Mayboroda, “Hardy and BMO spaces associated to divergence form elliptic operators,” Mathematische Annalen, vol. 344, no. 1, pp. 37–116, 2009.