数学物理学报  2018, Vol. 38 Issue (4): 716-727   PDF    
扩展功能
加入收藏夹
复制引文信息
加入引用管理器
Email Alert
RSS
本文作者相关文章
王立伟
束立生
本征平方函数在变指数Herz及Herz-Hardy空间上的有界性
王立伟1, 束立生2     
1. 安徽工程大学数理学院 安徽 芜湖 241000;
2. 安徽师范大学数学计算机科学学院 安徽 芜湖 241003
摘要:该文证明了本征平方函数 $S_\beta$ 在变指数 Herz 空间 $\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)$ 及 Herz-Hardy 空间 $H\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)$ 上的有界性, 其中 $\alpha(\cdot)$$p(\cdot)$ 均为变指数. 特别地, 当 $\alpha(\cdot)\equiv\alpha$ 为常数时, 相应结果也是新的.
关键词本征平方函数    变指数Herz空间    变指数Herz-Hardy空间    
Boundedness of the Intrinsic Square Function on Variable Exponent Herz and Herz-Hardy Spaces
Wang Liwei1, Shu Lisheng2     
1. School of Mathematics and Physics, Anhui Polytechnic University, Anhui Wuhu 241000;
2. School of Mathematics and Computer Science, Anhui Normal University, Anhui Wuhu 241003
Abstract: In this paper, we show that the intrinsic square function $S_\beta$ is bounded on the variable exponent Herz spaces $\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)$ and Herz-Hardy spaces $H\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)$, where $\alpha(\cdot)$ and $p(\cdot)$ are variable. In particular, when $\alpha(\cdot)\equiv \alpha$ is constant, the corresponding main results are also new.
Key words: The intrinsic square function     Variable Herz spaces     Variable Herz-Hardy spaces    
1 引言

对任意的$0<\beta\leq 1$, 设${\cal C}_\beta$是由函数$\phi: {\Bbb R}^n\rightarrow{\Bbb R}$构成的函数族, 其中$\phi$满足${\rm supp}\phi\subset\{x\in{\Bbb R}^n: |x|\leq 1\}$, $\int_{{\Bbb R}^n}\phi(x){\rm d}x=0$且对任意的$x_1, x_2\in{\Bbb R}^n$, 有

$ \begin{equation}\label{eq:a1} |\phi(x_1)-\phi(x_2)|\leq |x_1-x_2|^\beta. \end{equation} $ (1.1)

对于$(y, t)\in{\Bbb R}^{n+1}_+={\Bbb R}^n\times(0, \infty)$$f\in L^1_{\rm loc}({\Bbb R}^n)$, 记

$ \begin{equation}\label{eq:a2} A_\beta(f)(y, t)=\sup\limits_{\phi\in{\cal C}_\beta}|f\ast\phi_t(y)|=\sup\limits_{\phi\in{\cal C}_\beta} \bigg|\int_{{\Bbb R}^n}\phi_t(y-z)f(z){\rm d}z\bigg|. \end{equation} $ (1.2)

本征平方函数$S_\beta$定义如下

$ \begin{equation}\label{eq:a3} S_\beta(f)(x)=\bigg(\int\!\!\!\int_{\Gamma(x)}(A_\beta(f)(y, t))^2\frac{ {\rm d}y{\rm d}t}{t^{n+1}}\bigg)^{\frac{1}{2}}, \end{equation} $ (1.3)

其中$\phi_t(x)=\frac{1}{t^n}\phi(\frac{x}{t})$, $\Gamma(x)=\{(y, t)\in{\Bbb R}^{n+1}_+: |x-y|<t\}$.

Wilson[1]首先定义了这类新的本征平方函数$S_\beta$为了解决Fefferman和Stein在文献[2]中提出的关于经典Lusin面积函数$S$在加权Lebesgue空间上有界性的猜想.随后, 许多作者还研究了利用这类本征平方函数$S_\beta$刻画一般的函数空间, 如Hardy空间, Musielak-Orlicz Hardy空间等, 见文献[3-4].关于本征平方函数$S_\beta$的近期进展及广泛应用, 见文献[5-6]等.

另一方面, 由于Kováčik和Rákosník[7]的基础性工作, 具有变指数Lebesgue空间及其它的一些函数空间得到了广泛的研究, 见文献[8-13]. 2012年, Almeida和Drihem[14]定义了变指数Herz空间$\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)$$K_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)$, 其中$\alpha$$p$均为变指数.当指数$\alpha$$p$满足一些自然的正则性条件如log-Hölder连续条件时, 他们建立了一类次线性算子(包含Hardy-Littlewood极大算子, Riesz位势算子和Calderón-Zygmund算子)在这些变指数Herz空间上的有界性. 2015年, Dong和Xu[15]定义了变指数Herz-Hardy空间$H\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)$$HK_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)$, 并建立了它们的原子分解.利用这些分解, 他们证明了一些奇异积分算子, 如Calderón-Zygmund算子等的有界性.关于其它的变指数函数空间, 如变指数Morrey空间, 变指数Hardy空间等, 见文献[16-21].

变指数函数空间与经典的函数空间有许多本质的区别, 如变指数Lebesgue空间不具有平移不变性, 即若$p(\cdot)$${\Bbb{ R}}^n$中不恒为常数, 则总是存在$f\in L^{p(\cdot)}({\Bbb{ R}}^n)$$h\in {{\Bbb{ R}}^n}$, 使得$f(x+h)\notin L^{p(\cdot)}({\Bbb{ R}}^n)$.因而, 在经典Lebesgue空间中常用的good-$\lambda$方法及一些重排不等式在变指数Lebesgue空间中不再适用, 见文献[22-23].因为这个原因, 近二十年来将经典函数空间中一些重要结果推广到变指数情形吸引了越来越多学者的兴趣.最近, Wang[24-25]证明了本征平方函数$S_\beta$在经典Herz及Herz-Hardy空间上的有界性, 受此启发, 该文将进一步在具有变指数的Herz及Herz-Hardy空间上考虑这些有界性质.

在该文中, 定义$B:=B(x, r)=\{y\in{\Bbb R}^n: |x-y|<r\}$. $f_B$表示$f$$B$上的积分平均, 即$f_B=|B|^{-1}\int_Bf(x){\rm d}x$. $p'(\cdot)$$p(\cdot)$的共轭指数, 定义为$1/p(\cdot)+1/p'(\cdot)=1$. ${\cal S}(\mathbb{R} ^n)$${\cal S'}(\mathbb{R} ^n)$分别表示Schwartz函数空间和缓增广义函数空间.对于$x\in{\Bbb R}$, $[x]$表示不超过$x$的最大整数.字母$C$表示正常数, 但在不同的地方可能取值不同.表达式$f\lesssim g$意思为$f\leq Cg$, $f\thickapprox g$意思为$f\lesssim g\lesssim f$.

2 预备知识与主要结果

$p(\cdot): {\Bbb{ R}}^n\rightarrow [1, \infty)$为可测函数.变指数Lebesgue空间$L^{p(\cdot)}({\Bbb{ R}}^n)$表示${\Bbb{ R}}^n$上所有满足以下条件的可测函数$f$的集合

$ I_{p(\cdot)}(f):=\int_{{\Bbb{ R}}^n}|f(x)|^{p(x)}{\rm d}x<\infty. $

当被赋予如下Luxemburg-Nakano范数时, $L^{p(\cdot)}({\Bbb{ R}}^n)$为Banach函数空间

$ \|f\|_{L^{p(\cdot)}({\Bbb{ R}}^n)}:=\mbox{inf}\{\mu>0: I_{p(\cdot)}(f/\mu)\leq 1 \}. $

显然, 当$p(\cdot)\equiv p$时, $L^{p(\cdot)}({\Bbb{ R}}^n)=L^p({\Bbb{ R}}^n)$为经典的Lebesgue空间.给定一个开集$\Omega\subset {\Bbb{ R}}^n$, 局部$L^{p(\cdot)}_{\rm loc}(\Omega)$空间定义为

$ L^{p(\cdot)}_{\rm loc}(\Omega):=\{f: f\in L^{p(\cdot)}(K)\ {\mbox{对所有紧子集 }}K\subset \Omega\}. $

为了简便, 我们使用以下记号

$ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ p_-:=p_-({\Bbb{ R}}^n)=\mbox{ess inf}\{p(x): x\in {\Bbb{ R}}^n\}, \\ \ \ \ \ \ \ \ \ p_+:=p_+({\Bbb{ R}}^n)=\mbox{ess sup}\{p(x): x\in {\Bbb{ R}}^n\}, \\ \ \ \ \ \ \ \ \ \ \ {\cal P}({\Bbb{ R}}^n):=\{p(\cdot): \ p_->1 \ \mbox{且}\ p_+<\infty\}, \\ {\cal B}({\Bbb{ R}}^n):=\{p(\cdot)\in {\cal P}({\Bbb{ R}}^n): \ \ M \ {\mbox{在}} L^{p(\cdot)}({\Bbb{ R}}^n){\mbox{上有界}} \}, $

其中$M$表示Hardy-Littlewood极大算子, 定义为

$ Mf(x):=\sup\limits_{x\in {\Bbb{ R}}^n, r>0}r^{-n}\int_{B(x, r)}|f(y)|{\rm d}y. $

$p(\cdot)\in{\cal P}({\Bbb{ R}}^n)$, $f\in L^{p(\cdot)}({\Bbb{ R}}^n)$$g\in L^{p'(\cdot)}({\Bbb{ R}}^n)$, 则有以下广义Hölder不等式成立

$ \begin{equation}\label{eq:a4} \int_{{\Bbb R}^n}|f(x)g(x)|{\rm d}x\leq r_p\|f\|_{L^{p(\cdot)}({\Bbb R}^n)}\|g\|_{L^{p'(\cdot)}({\Bbb R}^n)}, \end{equation} $ (2.1)

其中$r_p :=1+1/p_--1/p_+$, 见文献[7, 定理2.1].

下面的引理2.1和2.2可见文献[17](或者[26]).

引理2.1  设$p(\cdot)\in{\cal B}({\Bbb{R}}^n)$, 则$ |B|^{-1}\|\chi_B\|_{L^{p(\cdot)}({\Bbb R}^n)}\|\chi_B\|_{L^{p'(\cdot)}({\Bbb R}^n)}\leq C. $

引理2.2  设$p(\cdot)\in{\cal B}({\Bbb{ R}}^n)$, 则对所有可测子集$E\subset B$, 有

$ \frac{\|\chi_E\|_{L^{p(\cdot)}({\Bbb R}^n)}}{\|\chi_B\|_{L^{p(\cdot)}({\Bbb R}^n)}} \leq C\bigg(\frac{|E|}{|B|}\bigg)^{\delta_1}, \quad \frac{\|\chi_E\|_{L^{p'(\cdot)}({\Bbb R}^n)}}{\|\chi_B\|_{L^{p'(\cdot)}({\Bbb R}^n)}}\leq C\bigg(\frac{|E|}{|B|}\bigg)^{\delta_2}, $

其中$\delta_1, \delta_2$为常数, 且$0<\delta_1, \delta_2<1$.

$B_l := \{x \in {\mathbb{R} } ^n: |x| \leq 2^l\}$, $R_l := B_l\backslash B_{l-1}$, 这里$l\in {\Bbb Z}$, $\chi_l := \chi_{R_l}$表示$R_l$的特征函数.

定义2.1  设$0< q\leq\infty$, $p(\cdot)\in {\cal P}({\Bbb{ R}}^n)$$\alpha(\cdot)\in L^\infty({\Bbb{ R}}^n)$.齐次Herz空间$\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)$定义为

$ {\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}}({\Bbb{ R}}^n) := \bigg\{f\in L_{\rm {loc}}^{p(\cdot)}({{\Bbb{ R}}^n}\backslash\{0\}): \|f\|_{{\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}}({\Bbb{ R}}^n)}<\infty\bigg\}, $

其中

$ \|f\|_{{\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}}({\Bbb{ R}}^n)}:= \bigg(\sum\limits_{k\in{\Bbb Z}}\|2^{k\alpha(\cdot)} f\chi_k\|^q_{L^{p(\cdot)}({\Bbb{ R}}^n)}\bigg)^{1/q}. $

$q=\infty$时, 作通常意义下的修改.

我们称函数$\alpha(\cdot):{\Bbb{ R}}^n \rightarrow \Bbb R$在原点处log-Hölder连续, 如果存在常数$C_{\rm {log}}>0$使得

$ \begin{equation} |\alpha(x)-\alpha(0)|\leq \frac{C_{\rm {log}}}{\mbox{log}(e+1/|x|)}, \quad x\in {\Bbb{ R}}^n. \end{equation} $ (2.2)

如果存在常数$\alpha_\infty\in {\Bbb{R}}$$C_{\rm {log}}>0$, 使得

$ \begin{equation} |\alpha(x)-\alpha_\infty|\leq \frac{C_{\rm {log}}}{\mbox{log}(e+|x|)}, \quad x\in {\Bbb{ R}}^n, \end{equation} $ (2.3)

则称$\alpha(\cdot)$在无穷处log-Hölder连续. Almeida和Drihem[14]证明了以下结果.

命题2.1  设$0< q\leq\infty$, $p(\cdot)\in {\cal P}({\Bbb{ R}}^n)$$\alpha(\cdot)\in L^\infty({\Bbb{ R}}^n)$.若$\alpha(\cdot)$既在原点也在无穷处log-Hölder连续, 则

$ \|f\|_{\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)}\approx\bigg(\sum\limits_{k=-\infty}^{-1}2^{k\alpha(0) q}\| f\chi_k\|^q_{L^{p(\cdot)}({\Bbb{ R}}^n)}\bigg)^{\frac{1}{q}}+\bigg(\sum\limits_{k=0}^{\infty}2^{k\alpha_\infty q}\| f \chi_k\|^q_{L^{p(\cdot)}({\Bbb{ R}}^n)}\bigg)^{\frac{1}{q}}. $

$G_Nf$$f$的grand极大函数, 定义为

$ G_N(f)(x):=\sup\limits_{\phi\in{\cal A}_N}|\phi^{\ast}_\nabla(f)(x)|, \quad \forall x\in{\Bbb R}^n, $

其中

$ {\cal A}_N :=\Big\{\phi\in{\cal S}({\Bbb{ R}}^n): \sup\limits_{|\alpha|\leq N, |\beta|\leq N}|x^{\alpha}D^{\beta}\phi(x)|\leq1\Big\}, \quad N>n+1, $

$\phi _\nabla ^ * (f)(x): = \mathop {\sup }\limits_{|y - x| < t} |{\phi _t} * f(y)|$, $\phi_t(x) :=t^{-n}\phi(\frac{x}{t})$.

极大算子$G_N$首先由Fefferman和Stein[27]为了研究经典的Hardy空间而引入.关于经典的Hardy空间及其相应的变指数函数空间研究, 见文献[28-30]等.

定义2.2  设$0<q\leq\infty$, $p(\cdot)\in{\cal P}({\Bbb{ R}}^n)$, $\alpha(\cdot)\in L^\infty({\Bbb{ R}}^n)$$N>n+1$.齐次Herz型Hardy空间$H\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)$定义为

$ H\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n):=\bigg\{f\in{\cal S'}({\Bbb{ R}}^n): G_Nf\in\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)\bigg\} $

$\|f\|_{H\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)}=\|G_Nf\|_{\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)}$.

定义2.3  设$p(\cdot)\in{\cal P}({\Bbb{ R}}^n)$, $\alpha(\cdot)\in L^\infty({\Bbb{ R}}^n)$既在原点也在无穷处log-Hölder连续.若非负整数$s\geq [\alpha_r-n\delta_2]$, 其中$n\delta_2\leq \alpha_r<\infty$, $\delta_2$同引理2.2中所定义.这里$\alpha_r=\alpha(0)$$0<r<1$, $\alpha_r=\alpha_\infty$$r\geq 1$.称函数$a(\cdot)$为一个中心$(\alpha(\cdot), p(\cdot))$ -原子, 如果满足

$ \begin{eqnarray*} &&(1)\ \ \ {\rm supp}a\subset B(0, r):=\{x\in{\Bbb R}^n: |x|<r\}.\\ &&(2)\ \ \ \|a\|_{L^{p(\cdot)}({\Bbb{ R}}^n)}\leq|B(0, r)|^{-\frac{\alpha(0)}{n}}, \ 0<r<1.\\ &&(3)\ \ \ \|a\|_{L^{p(\cdot)}({\Bbb{ R}}^n)}\leq|B(0, r)|^{-\frac{\alpha_\infty}{n}}, \ r\geq 1.\\ &&(4)\ \ \ \int_{{\Bbb{ R}}^n}a(x)x^{\beta}{\rm d}x=0, \ |\beta|\leq s. \end{eqnarray*} $

显然, 若$p(\cdot)\equiv p$, $\alpha(\cdot)\equiv\alpha$为常数, 则取$\delta_2=1-\frac{1}{p}$我们即可得到经典的情形, 见文献[28]. Dong和Xu在文献[15, 定理2.1]中给出了$H\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)$空间的原子刻画.

命题2.2  设$0<q<\infty$, $p(\cdot)\in{\cal B}({\Bbb{ R}}^n)$, $\alpha(\cdot)\in L^\infty({\Bbb{ R}}^n)$既在原点也在无穷处log-H$\ddot{\rm o}$lder连续.若$n\delta_2\leq\alpha(0), \alpha_\infty<\infty$, 其中$\delta_2$同引理2.2中所定义.则$f\in H\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)$当且仅当在分布意义下它可以分解为$f=\sum\limits_{k=-\infty}^{\infty}\lambda_ka_k, $其中$a_k$是中心$(\alpha(\cdot), p(\cdot))$ -原子, $\mbox{supp}a_k\subset B_k$$\sum\limits_{k=-\infty}^{\infty}|\lambda_k|^q<\infty$.进一步, $ \|f\|_{H\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{R}}^n)} \approx \inf\Big(\sum\limits_{k=-\infty}^{\infty}|\lambda_k|^q \Big)^{1/q}, $其中下确界取遍$f$所有上述分解.

下面给出我们的主要结果.

定理2.1  设$0<\beta\leq 1$, $0<q\leq\infty$, $p(\cdot)\in{\cal B}({\Bbb{ R}}^n)$, $\alpha(\cdot)\in L^\infty({\Bbb{ R}}^n)$既在原点也在无穷处log-H$\ddot{\rm o}$lder连续, 使得$-n\delta_1<\alpha(0)\leq\alpha_{\infty}<n\delta_2$, 其中$0<\delta_1, \delta_2<1$同引理2.2中所定义.则对任意的$f\in\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)$$ \|S_\beta(f)\|_{\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)} \leq C\|f\|_{\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)}. $

注意到在定理2.1中, 我们限制$\alpha(0), \alpha_{\infty}\in(-n\delta_1, n\delta_2)$, 下面的结果进一步将定理2.1推广到了$\alpha(0), \alpha_{\infty}\geq n\delta_2$时情形.

定理2.2  设$0<\beta\leq 1$, $0<q<\infty$, $p(\cdot)\in{\cal B}({\Bbb{ R}}^n)$, $\alpha(\cdot)\in L^\infty({\Bbb{ R}}^n)$既在原点也在无穷处log-Hölder连续.若$n\delta_2\leq\alpha(0), \alpha_\infty<n\delta_2+\beta$, 其中$\delta_2$同引理2.2中所定义.则对任意的$f\in H\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)$$ \|S_\beta(f)\|_{\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)} \leq C\|f\|_{H\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)}. $

注2.1  正如文献[15]中指出, 若$p(\cdot)\in{\cal B}({\Bbb{ R}}^n)$$\alpha(\cdot)\in L^\infty({\Bbb{ R}}^n)$既在原点也在无穷处log-Hölder连续, 则当$-n\delta_1<\alpha(0)\leq\alpha_{\infty}<n\delta_2$时, $H\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)\bigcap L_{\rm {loc}}^{p(\cdot)}({{\Bbb{R}}^n}\backslash\{0\})=\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)$.当$n\delta_2\leq\alpha(0), \alpha_{\infty}<\infty$时, $H\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)\bigcap L_{\rm {loc}}^{p(\cdot)}({{\Bbb{ R}}^n}\backslash\{0\})\subsetneqq\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)$.比较上述结果可以发现在定理2.2中, $H\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)$成为$\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)$的合适替代.特别地, 当$\alpha(\cdot)\equiv\alpha$为常数时, 即在变指数Herz型Hardy空间$H\dot{K}_{p(\cdot)}^{\alpha, q}({\Bbb{ R}}^n)$中(定义见文献[20]), 定理2.2的结果也是新的.

3 定理2.1和2.2的证明

以下我们仅证明$0<q<\infty$情形.当$q=\infty$时,证明过程只需作简单修改.

定理2.1的证明  记

$ \begin{eqnarray*} f(x)=\sum\limits_{i=-\infty}^\infty f(x)\chi_i(x)\equiv\sum\limits_{i=-\infty}^{\infty} f_i(x). \end{eqnarray*} $

由命题2.1及Minkowski不等式, 我们有

$ \begin{eqnarray*} \|S_\beta(f)\|_{\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)}&\approx &\bigg\{\sum\limits_{k=-\infty}^{-1}2^{\alpha(0) kq} \|S_\beta(f)\chi_k\|^q_{L^{p(\cdot)}({\Bbb R}^n)}\bigg\}^{1/q}\\ &&+\bigg\{\sum\limits_{k=0}^{\infty}2^{\alpha_\infty kq} \|S_\beta(f)\chi_k\|^q_{L^{p(\cdot)}({\Bbb R}^n)}\bigg\}^{1/q}\\ &\leq& C(H_1+H_2+H_3+H_4+H_5+H_6), \end{eqnarray*} $

其中

$ H_1 :=\bigg\{\sum\limits_{k=-\infty}^{-1}2^{\alpha(0) kq}\bigg(\sum\limits_{i=-\infty}^{k-2} \|S_\beta(f_i)\chi_k\|_{L^{p(\cdot)}({\Bbb R}^n)}\bigg)^q\bigg\}^{1/q}, \\ H_2 :=\bigg\{\sum\limits_{k=-\infty}^{-1}2^{\alpha(0) kq}\bigg(\sum\limits_{i=k-1}^{k+1} \|S_\beta(f_i)\chi_k\|_{L^{p(\cdot)}({\Bbb R}^n)}\bigg)^q\Biggr\}^{1/q}, \\ H_3 :=\bigg\{\sum\limits_{k=-\infty}^{-1}2^{\alpha(0) kq}\bigg(\sum\limits_{i=k+2}^{\infty} \|S_\beta(f_i)\chi_k\|_{L^{p(\cdot)}({\Bbb R}^n)}\Biggr)^q\bigg\}^{1/q}, \\ \ \ \ H_4 :=\bigg\{\sum\limits_{k=0}^{\infty}2^{\alpha_\infty kq}\bigg(\sum\limits_{i=-\infty}^{k-2} \|S_\beta(f_i)\chi_k\|_{L^{p(\cdot)}({\Bbb R}^n)}\bigg)^q\bigg\}^{1/q}, \\ \ \ \ H_5 :=\bigg\{\sum\limits_{k=0}^{\infty}2^{\alpha_\infty kq}\bigg(\sum\limits_{i=k-1}^{k+1} \|S_\beta(f_i)\chi_k\|_{L^{p(\cdot)}({\Bbb R}^n)}\bigg)^q\Biggr\}^{1/q}, \\ \ \ \ H_6 :=\bigg\{\sum\limits_{k=0}^{\infty}2^{\alpha_\infty kq}\bigg(\sum\limits_{i=k+2}^{\infty} \|S_\beta(f_i)\chi_k\|_{L^{p(\cdot)}({\Bbb R}^n)}\Biggr)^q\bigg\}^{1/q}. $

首先估计$H_1$.注意到当$x\in R_k$, $(y, t)\in\Gamma(x)$, $z\in R_i\bigcap\{z: |y-z|\leq t\}$$i\leq {k-2}$, 则有

$ \begin{equation}\label{eq:a5} t\geq \frac{1}{2}(|x-y|+|y-z|)\geq \frac{1}{2}|x-z|\geq \frac{1}{2}(|x|-|z|)\geq\frac{1}{4}|x|. \end{equation} $ (3.1)

因此, 由广义Hölder不等式(2.1), 我们有

$ \begin{eqnarray}\label{eq:a6} |S_\beta(f_i)(x)| &=&\bigg(\int\!\!\!\int_{\Gamma(x)}\sup\limits_{\phi\in{\cal C}_\beta} \bigg|\int_{{\Bbb{ R}}^n}\phi_t(y-z)f_i(z){\rm d}z\bigg|^2\frac{{\rm d}y{\rm d}t}{t^{n+1}}\bigg)^{\frac{1}{2}}\\ &\leq& C\bigg(\int_{\frac{|x|}{4}}^{\infty}\int_{|x-y|<t} \bigg|t^{-n}\int_{{R_i}\bigcap\{z: |y-z|\leq t\}}f_i(z){\rm d}z\bigg|^2\frac{{\rm d}y{\rm d}t}{t^{n+1}}\bigg)^{\frac{1}{2}} \\ &\leq &C\bigg(\int_{R_i}|f_i(z)|{\rm d}z\bigg)\bigg(\int_{\frac{|x|}{4}}^{\infty}\frac{{\rm d}t}{t^{2n+1}}\bigg)^{\frac{1}{2}} \\ &\leq &C2^{-kn}\|f_i\|_{L^{p(\cdot)}({\Bbb R}^n)} \|\chi_{B_i}\|_{L^{p'(\cdot)}({\Bbb R}^n)}. \end{eqnarray} $ (3.2)

结合(3.2)式, 引理2.1和引理2.2, 我们有

$ \begin{eqnarray}\label{eq:a7} \|S_\beta(f_i)(x)\chi_k\|_{L^{p(\cdot)}({\Bbb R}^n)} &\leq& C2^{-kn}\|f_i\|_{L^{p(\cdot)}({\Bbb R}^n)} \|\chi_{B_i}\|_{L^{p'(\cdot)}({\Bbb R}^n)}\|\chi_{B_k}\|_{L^{p(\cdot)}({\Bbb R}^n)}\\ &\leq& C\|f_i\|_{L^{p(\cdot)}({\Bbb R}^n)}\frac{\|\chi_{B_i}\|_{L^{p'(\cdot)}({\Bbb R}^n)}}{\|\chi_{B_k}\|_{L^{p'(\cdot)}({\Bbb R}^n)}}\\ &\leq &C2^{(i-k)n\delta_2}\|f_i\|_{L^{p(\cdot)}({\Bbb R}^n)}. \end{eqnarray} $ (3.3)

由(3.3)式可得

$ \begin{eqnarray*} H_1&\leq &C\bigg\{\sum\limits_{k=-\infty}^{-1}2^{k\alpha(0) q} \bigg(\sum\limits_{i=-\infty}^{k-2}\|f_i\|_{L^{p(\cdot)}({\Bbb R}^n)}2^{(i-k)n\delta_2}\bigg)^q\bigg\}^{1/q}\\ &=& C\bigg\{\sum\limits_{k=-\infty}^{-1} \bigg(\sum\limits_{i=-\infty}^{k-2}2^{i\alpha(0)}\|f_i\|_{L^{p(\cdot)}({\Bbb R}^n)}2^{(i-k)(n\delta_2-\alpha(0))}\bigg)^q\bigg\}^{1/q}. \end{eqnarray*} $

我们分以下两种情况考虑.

情形Ⅰ  $0<q\leq 1$.由已知不等式

$ \begin{equation}\label{eq:a8} \bigg(\sum\limits_{j=1}^\infty \theta_j\bigg)^q\leq \sum\limits_{j=1}^\infty \theta_j^q, \quad (\theta_j>0, \ j\in{\Bbb N}) \end{equation} $ (3.4)

$n\delta_2-\alpha(0)>0$, 我们得到以下估计

$ \begin{eqnarray*} H_1&\leq& C\bigg\{\sum\limits_{k=-\infty}^{-1} \sum\limits_{i=-\infty}^{k-2}2^{i\alpha(0) q}\|f_i\|^q_{L^{p(\cdot)}({\Bbb R}^n)}2^{(i-k)(n\delta_2-\alpha(0))q}\bigg\}^{1/q}\\ &=&C\bigg\{\sum\limits_{i=-\infty}^{-3}2^{i\alpha(0) q} \|f_i\|^q_{L^{p(\cdot)}({\Bbb R}^n)}\sum\limits_{k=i+2}^{-1} 2^{(i-k)(n\delta_2-\alpha(0))q}\bigg\}^{1/q}\\ &\leq& C\bigg\{\sum\limits_{i=-\infty}^{-3}2^{i\alpha(0) q} \|f_i\|^q_{L^{p(\cdot)}({\Bbb R}^n)}\bigg\}^{1/q} \\ &\leq &C\|f\|_{\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)}. \end{eqnarray*} $

情形Ⅱ  $1<q<\infty$.利用Hölder不等式, 我们有

$ \begin{eqnarray*} H_1&\leq &C\bigg\{\sum\limits_{k=-\infty}^{-1} \bigg(\sum\limits_{i=-\infty}^{k-2}2^{i\alpha(0) q}\|f_i\|^q_{L^{p(\cdot)}({\Bbb R}^n)}2^{(i-k)(n\delta_2-\alpha(0))q/2}\bigg)\\ &&\times\bigg(\sum\limits_{i=-\infty}^{k-2} 2^{(i-k)(n\delta_2-\alpha(0))q'/2}\bigg)^{q/q'}\bigg\}^{1/q}\\ &\leq& C\bigg\{\sum\limits_{i=-\infty}^{-3}2^{i\alpha(0) q}\|f_i\|^q_{L^{p(\cdot)}({\Bbb R}^n)}\sum\limits_{k=i+2}^{-1} 2^{(i-k)(n\delta_2-\alpha(0))q/2}\bigg\}^{1/q}\\ &\leq &C\bigg\{\sum\limits_{i=-\infty}^{-3}2^{i\alpha(0) q} \|f_i\|^q_{L^{p(\cdot)}({\Bbb R}^n)}\bigg\}^{1/q} \\ &\leq &C\|f\|_{\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)}. \end{eqnarray*} $

对于$H_4$, 用$\alpha_\infty$代替$\alpha(0)$, 利用与$H_1$相同的估计方法可得

$ H_4\leq C\bigg\{\sum\limits_{k=0}^{\infty} \bigg(\sum\limits_{i=-\infty}^{k-2}2^{i\alpha_{\infty}}\|f_i\|_{L^{p(\cdot)}({\Bbb R}^n)}2^{(i-k)(n\delta_2-\alpha_{\infty})}\bigg)^q\bigg\}^{1/q}. $

$0<q\leq 1$, 注意到$\alpha(0)\leq\alpha_{\infty}<n\delta_2$, 由Minkowski不等式和(3.4)式, 我们有

$\begin{eqnarray*} H_4&\leq& C\bigg\{\sum\limits_{k=0}^{\infty} \bigg(\sum\limits_{i=-\infty}^{-2}2^{i\alpha_{\infty}}\|f_i\|_{L^{p(\cdot)}({\Bbb R}^n)}2^{(i-k)(n\delta_2-\alpha_{\infty})}\bigg)^q\bigg\}^{1/q}\\ &&+C\bigg\{\sum\limits_{k=0}^{\infty} \bigg(\sum\limits_{i=-1}^{k-2}2^{i\alpha_{\infty}}\|f_i\|_{L^{p(\cdot)}({\Bbb R}^n)}2^{(i-k)(n\delta_2-\alpha_{\infty})}\bigg)^q\bigg\}^{1/q} \\ &\leq& C\bigg\{\sum\limits_{i=-\infty}^{-2}2^{i\alpha_{\infty} q}\|f_i\|^q_{L^{p(\cdot)}({\Bbb R}^n)}\sum\limits_{k=0}^{\infty}2^{(i-k)(n\delta_2-\alpha_{\infty})q}\bigg\}^{1/q}\\ &&+C\bigg\{\sum\limits_{i=-1}^{\infty}2^{i\alpha_{\infty} q}\|f_i\|^q_{L^{p(\cdot)}({\Bbb R}^n)}\sum\limits_{k=i+2}^{\infty}2^{(i-k)(n\delta_2-\alpha_{\infty})q} \bigg\}^{1/q}\\ &\leq& C\bigg\{\sum\limits_{i=-\infty}^{-2}2^{i\alpha(0) q}\|f_i\|^q_{L^{p(\cdot)}({\Bbb R}^n)}\bigg\}^{1/q} +C\bigg\{\sum\limits_{i=-1}^{\infty}2^{i\alpha_{\infty} q}\|f_i\|^q_{L^{p(\cdot)}({\Bbb R}^n)} \bigg\}^{1/q}\\ &\leq &C\|f\|_{\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)}. \end{eqnarray*} $

$1<q<\infty$, 由Hölder不等式可得

$ \begin{eqnarray*} H_4&\leq& C\bigg\{\sum\limits_{k=0}^{\infty} \bigg(\sum\limits_{i=-\infty}^{-2}2^{i\alpha_{\infty} q}\|f_i\|^q_{L^{p(\cdot)}({\Bbb R}^n)}2^{(i-k)(n\delta_2-\alpha_{\infty})q/2}\bigg)\\ &\quad&\times\bigg(\sum\limits_{i=-\infty}^{-2} 2^{(i-k)(n\delta_2-\alpha_{\infty})q'/2}\bigg)^{q/q'}\bigg\}^{1/q}\\ &\quad&+C\bigg\{\sum\limits_{k=0}^{\infty} \bigg(\sum\limits_{i=-1}^{k-2}2^{i\alpha_{\infty} q}\|f_i\|^q_{L^{p(\cdot)}({\Bbb R}^n)}2^{(i-k)(n\delta_2-\alpha_{\infty})q/2}\bigg)\\ &\quad&\times\bigg(\sum\limits_{i=-1}^{k-2} 2^{(i-k)(n\delta_2-\alpha_{\infty})q'/2}\bigg)^{q/q'}\bigg\}^{1/q}\\ &\leq& C\bigg\{\sum\limits_{i=-\infty}^{-2}2^{i\alpha_{\infty} q}\|f_i\|^q_{L^{p(\cdot)}({\Bbb R}^n)}\sum\limits_{k=0}^{\infty}2^{(i-k)(n\delta_2-\alpha_{\infty})q/2}\bigg\}^{1/q}\\ &\quad&+C\bigg\{\sum\limits_{i=-1}^{\infty}2^{i\alpha_{\infty} q}\|f_i\|^q_{L^{p(\cdot)}({\Bbb R}^n)}\sum\limits_{k=i+2}^{\infty}2^{(i-k)(n\delta_2-\alpha_{\infty})q/2} \bigg\}^{1/q}\\ &\leq& C\bigg\{\sum\limits_{i=-\infty}^{-2}2^{i\alpha(0) q}\|f_i\|^q_{L^{p(\cdot)}({\Bbb R}^n)}\bigg\}^{1/q} +C\bigg\{\sum\limits_{i=-1}^{\infty}2^{i\alpha_{\infty} q}\|f_i\|^q_{L^{p(\cdot)}({\Bbb R}^n)} \bigg\}^{1/q}\\ &\leq& C\|f\|_{\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)}. \end{eqnarray*} $

对于$H_2$$H_5$, 由$S_\beta$$L^{p(\cdot)}({\Bbb R}^n)$有界性(见文献[5, 定理3.4])可得

$ H_2\leq C\bigg\{\sum\limits_{k=-\infty}^{-1}2^{k\alpha(0) q} \|f_k\|^q_{L^{p(\cdot)}({\Bbb R}^n)} \bigg\}^{1/q}\leq C\|f\|_{\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)}.\\ \ \ \ H_5\leq C\bigg\{\sum\limits_{k=0}^{\infty}2^{k\alpha_{\infty} q} \|f_k\|^q_{L^{p(\cdot)}({\Bbb R}^n)} \bigg\}^{1/q} \leq C\|f\|_{\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)}. $

下面我们估计$H_3$.注意到$x\in R_k$, $(y, t)\in\Gamma(x)$, $z\in R_i\bigcap\{z: |y-z|\leq t\}$$i\geq {k+2}$, 我们有

$ \begin{equation}\label{eq:a9} t\geq \frac{1}{2}(|x-y|+|y-z|)\geq \frac{1}{2}|x-z|\geq \frac{1}{2}(|z|-|x|)\geq\frac{1}{4}|z|. \end{equation} $ (3.5)

由(3.5)式和广义Hölder不等式(2.1)可得

$\begin{eqnarray}\label{eq:a10} |S_\beta(f_i)(x)| &=&\bigg(\int\!\!\!\int_{\Gamma(x)}\sup\limits_{\phi\in{\cal C}_\beta} \bigg|\int_{{\Bbb{ R}}^n}\phi_t(y-z)f_i(z){\rm d}z\bigg|^2\frac{{\rm d}y{\rm d}t}{t^{n+1}}\bigg)^{\frac{1}{2}} \\ &\leq &C\bigg(\int_{\frac{|z|}{4}}^{\infty}\int_{|x-y|<t} \bigg|t^{-n}\int_{{R_i}\bigcap\{z: |y-z|\leq t\}}f_i(z){\rm d}z\bigg|^2\frac{{\rm d}y{\rm d}t}{t^{n+1}}\bigg)^{\frac{1}{2}} \\ &\leq &C\bigg(\int_{R_i}|f_i(z)|{\rm d}z\bigg)\bigg(\int_{\frac{|z|}{4}}^{\infty}\frac{{\rm d}t}{t^{2n+1}}\bigg)^{\frac{1}{2}}\\ &\leq& C2^{-in}\|f_i\|_{L^{p(\cdot)}({\Bbb R}^n)} \|\chi_{B_i}\|_{L^{p'(\cdot)}({\Bbb R}^n)}. \end{eqnarray} $ (3.6)

结合(3.6)式, 引理2.1和引理2.2, 我们有

$ \begin{eqnarray}\label{eq:a11} \|S_\beta(f_i)(x)\chi_k\|_{L^{p(\cdot)}({\Bbb R}^n)} &\leq& C2^{-in}\|f_i\|_{L^{p(\cdot)}({\Bbb R}^n)} \|\chi_{B_i}\|_{L^{p'(\cdot)}({\Bbb R}^n)}\|\chi_{B_k}\|_{L^{p(\cdot)}({\Bbb R}^n)}\\ &\leq& C\|f_i\|_{L^{p(\cdot)}({\Bbb R}^n)}\frac{\|\chi_{B_k}\|_{L^{p(\cdot)}({\Bbb R}^n)}}{\|\chi_{B_i}\|_{L^{p(\cdot)}({\Bbb R}^n)}}\\ &\leq &C2^{(k-i)n\delta_1}\|f_i\|_{L^{p(\cdot)}({\Bbb R}^n)}. \end{eqnarray} $ (3.7)

由(3.7)式可得

$ H_3\leq C\bigg\{\sum\limits_{k=-\infty}^{-1} \bigg(\sum\limits_{i=k+2}^{\infty}2^{i\alpha(0)}\|f_i\|_{L^{p(\cdot)}({\Bbb R}^n)}2^{(k-i)(n\delta_1+\alpha(0))}\bigg)^q\bigg\}^{1/q}. $

$\alpha_\infty$代替$\alpha(0)$, 同理可得

$ H_6\leq C\bigg\{\sum\limits_{k=0}^{\infty} \bigg(\sum\limits_{i=k+2}^{\infty}2^{i\alpha_{\infty}}\|f_i\|_{L^{p(\cdot)}({\Bbb R}^n)}2^{(k-i)(n\delta_1+\alpha_{\infty})}\bigg)^q\bigg\}^{1/q}. $

我们略去$H_3$$H_6$的估计, 因为它们本质上分别与$H_4$$H_1$估计相似.从而定理2.1证明完毕.

定理2.2的证明  设$f\in H\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)$, 由命题2.2, 在分布意义下我们有$f=\sum\limits_{j=-\infty}^{\infty}\lambda_ja_j, $其中$\lambda_j\geq 0$, $a_j$是中心$(\alpha(\cdot), p(\cdot))$ -原子, $\mbox{supp}a_j\subset B_j$

$ \|f\|_{H\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)}\approx \inf \bigg(\sum\limits_{j=-\infty}^{\infty}|\lambda_j|^q\bigg)^{\frac{1}{q}}. $

因此, 我们有

$ \begin{eqnarray*} \|S_\beta (f)\|_{\dot{K}_{p(\cdot)}^{\alpha(\cdot), q}({\Bbb{ R}}^n)} &\approx&\bigg(\sum\limits_{k=-\infty}^{-1}2^{k\alpha(0) q}\| S_\beta(f)\chi_k\|^q_{L^{p(\cdot)}({\Bbb{ R}}^n)}\bigg)^{\frac{1}{q}}\\ &\quad& +\bigg(\sum\limits_{k=0}^{\infty}2^{k\alpha_\infty q}\| S_\beta(f)\chi_k\|^q_{L^{p(\cdot)}({\Bbb{ R}}^n)}\bigg)^{\frac{1}{q}}\\ \\ &\leq& C\bigg\{\sum\limits_{k=-\infty}^{-1}2^{k\alpha(0)q} \bigg(\sum\limits_{j=-\infty}^{k-2}|\lambda_j|\| S_\beta(a_j)\chi_k\|_{L^{p(\cdot)}({\Bbb{ R}}^n)}\bigg)^q\bigg\}^{\frac{1}{q}}\\ &\quad& +C\bigg\{\sum\limits_{k=-\infty}^{-1}2^{k\alpha(0)q} \bigg(\sum\limits_{j=k-1}^{\infty}|\lambda_j|\| S_\beta(a_j)\chi_k\|_{L^{p(\cdot)}({\Bbb{ R}}^n)}\bigg)^q\bigg\}^{\frac{1}{q}}\\ &\quad &+C\bigg\{\sum\limits_{k=0}^{\infty}2^{k\alpha_\infty q} \bigg(\sum\limits_{j=-\infty}^{k-2}|\lambda_j|\| S_\beta(a_j)\chi_k\|_{L^{p(\cdot)}({\Bbb{ R}}^n)}\bigg)^q\bigg\}^{\frac{1}{q}}\\ &\quad &+C\bigg\{\sum\limits_{k=0}^{\infty}2^{k\alpha_\infty q} \bigg(\sum\limits_{j=k-1}^{\infty}|\lambda_j|\| S_\beta(a_j)\chi_k\|_{L^{p(\cdot)}({\Bbb{ R}}^n)}\bigg)^q\bigg\}^{\frac{1}{q}}\\ &:=&J_1+J_2+J_3+J_4. \end{eqnarray*} $

对于$J_1$, 注意到若$x\in R_k$, $(y, t)\in\Gamma(x)$, $z\in B_j\bigcap\{z: |y-z|\leq t\}$$j\leq {k-2}$, 则有$t\geq\frac{1}{4}|x|$.利用$a_j$的消失性条件, 我们有

$ \begin{eqnarray}\label{eq:a12} |S_\beta(a_j)(x)| &=&\bigg(\int\!\!\!\int_{\Gamma(x)}\sup\limits_{\phi\in{\cal C}_\beta} \bigg|\int_{{\Bbb{ R}}^n}\phi_t(y-z)a_j(z){\rm d}z\bigg|^2\frac{{\rm d}y{\rm d}t}{t^{n+1}}\bigg)^{\frac{1}{2}}\\ &=&\bigg(\int\!\!\!\int_{\Gamma(x)}\sup\limits_{\phi\in{\cal C}_\beta} \bigg|\int_{{\Bbb{ R}}^n}(\phi_t(y-z)-\phi_t(y))a_j(z){\rm d}z\bigg|^2\frac{{\rm d}y{\rm d}t}{t^{n+1}}\bigg)^{\frac{1}{2}}\\ &\leq& C2^{j\beta}\bigg(\int_{B_j}|a_j(z)|{\rm d}z\bigg)\bigg(\int_{\frac{|x|}{4}}^{\infty}\int_{|x-y|<t} \frac{{\rm d}y{\rm d}t}{t^{3n+1+2\beta}}\bigg)^{\frac{1}{2}}\\ &\leq &C2^{(j-k)\beta-kn}\|\chi_{B_j}\|_{L^{p'(\cdot)}({\Bbb{R}}^n)}\|a_j\|_{L^{p(\cdot)}({\Bbb{R}}^n)}. \end{eqnarray} $ (3.8)

由(3.8)式, 引理2.1和2.2, 可得

$ \begin{eqnarray}\label{eq:a13} \|S_\beta(a_j)\chi_k\|_{L^{p(\cdot)}({\Bbb{R}}^n)} &\leq& C2^{(j-k)\beta-kn}\|a_j\|_{L^{p(\cdot)}({\Bbb{R}}^n)} \|\chi_{B_k}\|_{L^{p(\cdot)}({\Bbb{R}}^n)}\|\chi_{B_j}\|_{L^{p'(\cdot)}({\Bbb{R}}^n)}\\ &\leq &C2^{(j-k)\beta}\|a_j\|_{L^{p(\cdot)}({\Bbb{R}}^n)} \frac{\|\chi_{B_j}\|_{L^{p'(\cdot)}({\Bbb{R}}^n)}}{\|\chi_{B_k} \|_{L^{p'(\cdot)}({\Bbb{R}}^n)}}\\ &\leq &C2^{(j-k)(\beta+n\delta_2)}\|a_j\|_{L^{p(\cdot)}({\Bbb{R}}^n)}. \end{eqnarray} $ (3.9)

为了简洁, 以下我们记$\xi=\beta+n\delta_2-\alpha(0)>0$, $\eta=\beta+n\delta_2-\alpha_\infty>0$.若$0<q\leq 1$, 注意到$j\leq k-2$$k<0$时, 有$\|a_j\|_{L^{p(\cdot)}({\Bbb{R}}^n)}\leq 2^{-j\alpha(0)}$, 由(3.9)式和$n\delta_2\leq \alpha(0)<n\delta_2+\beta$, 我们有

$ \begin{eqnarray*} J_1^q&=&C\sum\limits_{k=-\infty}^{-1}2^{k\alpha(0)q} \bigg(\sum\limits_{j=-\infty}^{k-2}|\lambda_j|\| S_\beta(a_j)\chi_k\|_{L^{p(\cdot)}({\Bbb{ R}}^n)}\bigg)^q\\ &\leq &C\sum\limits_{k=-\infty}^{-1}2^{k\alpha(0)q} \bigg(\sum\limits_{j=-\infty}^{k-2}|\lambda_j| 2^{(j-k)(\beta+n\delta_2)-j\alpha(0)}\bigg)^q\\ &\leq &C\sum\limits_{k=-\infty}^{-1}2^{k\alpha(0)q} \bigg(\sum\limits_{j=-\infty}^{k-2}|\lambda_j|^q 2^{(j-k)(\beta+n\delta_2)q-j\alpha(0) q}\bigg)\\ &=&C\sum\limits_{j=-\infty}^{-3}|\lambda_j|^q\sum\limits_{k=j+2}^{-1}2^{(j-k)\xi q} \leq C\sum\limits_{j=-\infty}^{-1}|\lambda_j|^q. \end{eqnarray*} $

$1<q<\infty$, 由Hölder不等式可得

$ \begin{eqnarray*} J_1^q&=&C\sum\limits_{k=-\infty}^{-1}2^{k\alpha(0)q} \bigg(\sum\limits_{j=-\infty}^{k-2}|\lambda_j|\| S_\beta(a_j)\chi_k\|_{L^{p(\cdot)}({\Bbb{ R}}^n)}\bigg)^q\\ &\leq &C\sum\limits_{k=-\infty}^{-1}2^{k\alpha(0)q} \bigg(\sum\limits_{j=-\infty}^{k-2}|\lambda_j| 2^{(j-k)(\beta+n\delta_2)-j\alpha(0)}\bigg)^q\\ &\leq &C\sum\limits_{k=-\infty}^{-1}\bigg(\sum\limits_{j=-\infty}^{k-2}|\lambda_j|^q2^{(j-k)\xi q/2}\bigg) \bigg(\sum\limits_{j=-\infty}^{k-2}2^{(j-k)\xi q'/2}\bigg)^{q/q'}\\ &\leq& C\sum\limits_{j=-\infty}^{-3}|\lambda_j|^q\sum\limits_{k=j+2}^{-1}2^{(j-k)\xi q/2} \leq C\sum\limits_{j=-\infty}^{-1}|\lambda_j|^q. \end{eqnarray*} $

对于$J_3$, 若$0<q\leq 1$, 利用(3.9)式以及$a_j$的尺寸条件, 我们有

$ \begin{eqnarray*} J_3^q&=&C\sum\limits_{k=0}^{\infty}2^{k\alpha_\infty q} \bigg(\sum\limits_{j=-\infty}^{k-2}|\lambda_j|\| S_\beta(a_j)\chi_k\|_{L^{p(\cdot)}({\Bbb{ R}}^n)}\bigg)^q\\ &\leq &C\sum\limits_{k=0}^{\infty}2^{k\alpha_\infty q} \bigg(\sum\limits_{j=-\infty}^{-1}|\lambda_j|^q 2^{(j-k)(\beta+n\delta_2)q-j\alpha(0)q}\bigg)\\ &\quad&+C\sum\limits_{k=0}^{\infty}2^{k\alpha_\infty q} \bigg(\sum\limits_{j=0}^{k-2}|\lambda_j|^q 2^{(j-k)(\beta+n\delta_2)q-j\alpha_\infty q}\bigg)\\ &\leq &C\sum\limits_{k=0}^{\infty}2^{k(\alpha_\infty-\beta-n\delta_2)q}\sum\limits_{j=-\infty}^{-1}|\lambda_j|^q 2^{j\xi q} +C\sum\limits_{j=0}^{\infty}|\lambda_j|^q\sum\limits_{k=j+2}^{\infty}2^{(j-k)\eta q}\\ &\leq& C\bigg(\sum\limits_{j=-\infty}^{-1}|\lambda_j|^q+\sum\limits_{j=0}^{\infty}|\lambda_j|^q\bigg) =C\sum\limits_{j=-\infty}^{\infty}|\lambda_j|^q. \end{eqnarray*} $

$1<q<\infty$, 由Hölder不等式可得

$ \begin{eqnarray*} J_3^q&=&C\sum\limits_{k=0}^{\infty}2^{k\alpha_\infty q} \bigg(\sum\limits_{j=-\infty}^{k-2}|\lambda_j|\| S_\beta(a_j)\chi_k\|_{L^{p(\cdot)}({\Bbb{ R}}^n)}\bigg)^q\\ &\leq &C\sum\limits_{k=0}^{\infty}2^{k\alpha_\infty q} \bigg(\sum\limits_{j=-\infty}^{-1}|\lambda_j| 2^{(j-k)(\beta+n\delta_2)-j\alpha(0)}\bigg)^q\\ &\quad&+C\sum\limits_{k=0}^{\infty}2^{k\alpha_\infty q} \bigg(\sum\limits_{j=0}^{k-2}|\lambda_j| 2^{(j-k)(\beta+n\delta_2)-j\alpha_\infty}\bigg)^q\\ &\leq &C\sum\limits_{k=0}^{\infty}2^{k(\alpha_\infty-\beta-n\delta_2)q} \bigg(\sum\limits_{j=-\infty}^{-1}|\lambda_j| 2^{j\xi}\bigg)^q +C\sum\limits_{k=0}^{\infty}\bigg(\sum\limits_{j=0}^{k-2}|\lambda_j| 2^{(j-k)\eta}\bigg)^q\\ &\leq &C\bigg(\sum\limits_{j=-\infty}^{-1}|\lambda_j|^q 2^{j\xi q/2}\bigg) \bigg(\sum\limits_{j=-\infty}^{-1} 2^{j\xi q'/2}\bigg)^{q/q'}\\ &\quad&+C\sum\limits_{k=0}^{\infty}\bigg(\sum\limits_{j=0}^{k-2}|\lambda_j|^q 2^{(j-k)\eta q/2}\bigg)^q \bigg(\sum\limits_{j=0}^{k-2} 2^{(j-k)\eta q'/2}\bigg)^{q/q'} \\ &\leq& C\sum\limits_{j=-\infty}^{-1}|\lambda_j|^q 2^{j\xi q/2} +C\sum\limits_{j=0}^{\infty}|\lambda_j|^q\sum\limits_{k=j+2}^{\infty}2^{(j-k)\eta q/2} \\ &\leq& C\sum\limits_{j=-\infty}^{\infty}|\lambda_j|^q. \end{eqnarray*} $

对于$J_2$$J_4$, 利用$S_\beta$$L^{p(\cdot)}({\Bbb R}^n)$有界性以及文献[15, p338]中对$I_1$$I_3$相同的估计方法, 我们容易得到

$ J_2^q\leq C\sum\limits_{j=-\infty}^{\infty}|\lambda_j|^q \quad \mbox{及}\quad J_4^q\leq C\sum\limits_{j=0}^{\infty}|\lambda_j|^q. $

因此, 结合以上估计, 定理2.2证明完毕.

参考文献
[1] Wilson M. The intrinsic square function. Rev Mat Iberoam, 2007, 23: 771–791.
[2] Fefferman C, Stein E M. Some maximal inequalities. Amer J Math, 1971, 93: 107–115. DOI:10.2307/2373450
[3] Liang Y, Yang D. Intrinsic square function characterizations of Musielak-Orlicz Hardy spaces. Trans Amer Math Soc, 2015, 367: 3225–3256.
[4] Zhuo C, Yang D, Liang Y. Intrinsic square function characterizations of Hardy spaces with variable exponents. Bull Malays Math Sci Soc, 2016, 39: 1541–1577. DOI:10.1007/s40840-015-0266-2
[5] Ho K P. Intrinsic square functions on Morrey and Block spaces with variable exponents. Bull Malays Math Sci Soc, 2017, 40: 995–1010. DOI:10.1007/s40840-016-0330-6
[6] Wilson M. Weighted Littlewood-Paley Theory and Exponential-Square Integrability. Berlin: SpringerVerlag, 2007.
[7] Kováčik O, Rákosník J. On spaces Lp(x) and Wk, p(x). Czechoslovak Math J, 1991, 41: 592–618.
[8] Cruz-Uribe D, Fiorenza A, Neugebauer C. The maximal function on variable Lp spaces. Ann Acad Sci Fenn Math, 2003, 28: 223–238.
[9] Diening L. Maximal function on generalized Lebesgue spaces Lp(·). Math Inequal Appl, 2004, 7: 245–253.
[10] Harjulehto P, Hästö P, Lê Ú V, Nuortio M. Overview of differential equations with non-standard growth. Nonlinear Anal, 2010, 72: 4551–4574. DOI:10.1016/j.na.2010.02.033
[11] Nekvinda A. Hardy-littlewood maximal operator in $L^{p(x)}({\Bbb R}^n)$. Math Inequal Appl, 2004, 7: 255–265.
[12] Tan J, Liu Z, Zhao J. On some multilinear commutators in variable Lebesgue spaces. J Math Inequal, 2017, 11: 715–734.
[13] Zhang P, Wu J. Commutators for the maximal function on Lebesgue spaces with variable exponent. Math Inequal Appl, 2014, 17: 1375–1386.
[14] Almeida A, Drihem D. Maximal, potential and singular type operators on Herz spaces with variable exponents. J Math Anal Appl, 2012, 394: 781–795. DOI:10.1016/j.jmaa.2012.04.043
[15] Dong B, Xu J. Variable Herz-type Hardy spaces and their applications. Anal Theory Appl, 2015, 31: 321–353.
[16] Dong B, Xu J. New Herz type Besov and Triebel-Lizorkin spaces with variable exponents. J Funct Space Appl, 2012: Article ID:384593.
[17] Izuki M. Commutators of fractional integrals on Lebesgue and Herz spaces with variable exponent. Rend Circ Mat Palermo, 2010, 59: 461–472. DOI:10.1007/s12215-010-0034-y
[18] Lu Y, Zhu Y. Boundedness of multilinear Calderón-Zygmund singular operators on Morrey-Herz spaces with variable exponents. Acta Math Sin (Engl Ser), 2014, 30: 1180–1194. DOI:10.1007/s10114-014-3410-2
[19] Tao S, Wang L. Littlewood-Paley operators on Morrey spaces with variable exponent. The Scientific World Journal, 2014: Article ID:790671.
[20] Wang H, Liu Z. The Herz-type Hardy spaces with variable exponent and thier applications. Taiwanese J Math, 2012, 16: 1363–1389. DOI:10.11650/twjm/1500406739
[21] Yang D, Zhuo C, Yuan W. Besov-type spaces with variable smoothness and integrability. J Funct Anal, 2015, 269: 1840–1898. DOI:10.1016/j.jfa.2015.05.016
[22] Cruz-uribe D, Fiorenza A. Variable Lebesgue Spaces:Foundations and Harmonic Analysis, Applied and Numerical Harmonic Analysis. Boston: Birkhäuser Basel, 2013.
[23] Diening L, Harjulehto P, Hästö P, R·užička M. Lebesgue and Sobolev Spaces with Variable Exponents. Heidelberg: Springer, 2011.
[24] Wang H. The boundedness of intrinsic square functions on the weighted Herz spaces. J Funct Spaces, 2014: Article ID:274521.
[25] Wang H. Some estimates of intrinsic square functions on the weighted Herz-type Hardy spaces. J Inequal Appl, 2015, 2015(1): 62. DOI:10.1186/s13660-015-0583-x
[26] Diening L. Maximal function on Musielak-Orlicz spaces and generalized Lebesgue spaces. Bull Sci Math, 2005, 129: 657–700. DOI:10.1016/j.bulsci.2003.10.003
[27] Fefferman C, Stein E M. Hp spaces of several variables. Acta Math, 1972, 129: 137–193. DOI:10.1007/BF02392215
[28] Lu S, Yang D, Hu G. Herz Type Spaces and Their Applications. Beijing: Science Press, 2008.
[29] Nakai E, Sawano Y. Hardy spaces with variable exponents and generalized Campanato spaces. J Funct Anal, 2012, 262: 3665–3748. DOI:10.1016/j.jfa.2012.01.004
[30] Yan X, Yang D, Yuan W, Zhuo C. Variable weak Hardy spaces and their applications. J Funct Anal, 2016, 271: 2822–2887. DOI:10.1016/j.jfa.2016.07.006