In the recent years, there have been many papers focused on studying the composition operator in function spaces (say, for 1-dimensional case see [3-7], for n-dimensional case see [8-11]). More recently, Contreras and Hernandez-Diaz in [6] and Montes-Rodriguez in [7] discuss the weighted composition operators on weighted banach spaces of analyticfunctions (for 1-dimensional case), respectively. From those, they get some sufficient and necessary conditions for composition operator to be bounded and compact.

In this paper, we discuss the boundedness and compactness of the weighted composition operators from $F(p,q,s)$  to $\alpha$  - $Bloch$  space $\beta \alpha$  .

As far as the boundedness of $W\psi ,\phi$  is concerned, we can have the following result:

Theorem 1. For $0<p,s<+\infty ,-n-1<q<+\infty ,q+s>-1,\alpha \ge 0,$  $\phi$  be a holomorphic self-map of $B$  , and $\psi$  a holomorphic function on $B$  . Then the weighted composition operator $W\psi ,\phi :F(p,q,s)\to \beta \alpha$  is bounded if and only if

$$\begin{array}{c}\begin{array}{c}sup\stackrel{u\in Cn-\left\{0\right\}}{\omega \in Bn}\left|\psi \right(\omega \left)\right|\frac{(1-|w\left|2\right)\alpha }{(1-|\phi \left(w\right)\left|2\right)\frac{q+n+1}{p}}\left\{\frac{(1-|\phi \left(w\right)\left|2\right)\left|J\phi \right(w\left)u\right|2+|<\phi (w),J\phi (w)u>|2}{(1-|w\left|2\right)\left|u\right|2+|<w,u>|2}\right\}\frac{1}{2}<+\infty \end{array}\end{array}$$

(5)

and

$$\begin{array}{c}{sup}_{\omega \in B}G\frac{q+n+1}{p}\left(\phi \right(\omega \left)\right)|\nabla \psi (\omega \left)\right|(1-|\omega \left|2\right)\alpha <+\infty ,\end{array}$$

(6)

where $J_{\phi }$  is the Jabocian of $\phi$  , and

$$\begin{array}{c}\begin{array}{c}{G}_{\frac{n+1+q}{p}}\left(\omega \right)=\{\begin{array}{cc}1,& 0<\frac{n+1+q}{p}<1,\\ log\frac{2}{1-|\omega {|}^2},& \frac{n+1+q}{p}=1,\\ {\left(\frac{1}{1-|\omega {|}^2}\right)}^{\frac{n+1+q}{p}-1},& \frac{n+1+q}{p}>1.\end{array}\end{array}\end{array}$$

(7)

On the other hand, the compactness of $W\psi ,\phi$  is much more complicated than the case of boundedness. The result varies sharply by the choosen of the real numbers $p$  and $q$  .

Theorem 2. For $0<p,s<+\infty ,-n-1<q<+\infty ,q+s>-1,$  and $\alpha \ge 0,$  $\phi$  be a holomorphic self-map of $B$  , $\psi$  a holomorphic function on $B$  . If $\frac{q+n+1}{p}\in (0,1)$  , then $W\psi ,\phi :F(p,q,s)\to \beta \alpha$  is compact if and only if $W\psi ,\phi$  is bounded and

$$\begin{array}{c}\begin{array}{c}{sup}_{u\in {\text{C}}^n-\left\{0\right\}}\left|\psi \right(\omega \left)\right|\frac{(1-|w\left|2\right)\alpha }{(1-|\phi \left(w\right)\left|2\right)\frac{q+n+1}{p}}\left\{\frac{(1-|\phi \left(w\right)\left|2\right)\left|J\phi \right(w\left)u\right|2+|<\phi (w),J\phi (w)u>|2}{(1-|w\left|2\right)\left|u\right|2+|<w,u>|2}\right\}\frac{1}{2}\to 0\end{array}\end{array}$$

(8)

when $\phi \left(\omega \right)\to \partial B$  .

Theorem 3. If $\frac{q+n+1}{p}\ge 1,$  then $W\psi ,\phi :F(p,q,s)\to \beta \alpha$  is compact if and only if $W\psi ,\phi$  is bounded and

$$\begin{array}{c}\begin{array}{c}{sup}_{u\in {\text{C}}^n-\left\{0\right\}}\left|\psi \right(\omega \left)\right|\frac{(1-|w\left|2\right)\alpha }{(1-|\phi \left(w\right)\left|2\right)\frac{q+n+1}{p}}\left\{\frac{(1-|\phi \left(w\right)\left|2\right)\left|J\phi \right(w\left)u\right|2+|<\phi (w),J\phi (w)u>|2}{(1-|w\left|2\right)\left|u\right|2+|<w,u>|2}\right\}\frac{1}{2}\to 0\end{array}\end{array}$$

(9)

$$\begin{array}{c}\begin{array}{c}G\frac{q+n+1}{p}\left(\phi \right(\omega \left)\right)|\nabla \psi (\omega \left)\right|(1-|\omega \left|2\right)\alpha \to 0\end{array}\end{array}$$

(10)

when $\phi \left(\omega \right)\to \partial B$  .

Latter, in Lemma 2.1 and Lemma 2.5, we will show that $F(p,q,s)$  is just ${\beta }^{\frac{n+1+q}{p}}$  if $s>n$  , therefore by the theorems above, we get Corollary 1. Let $\phi =({\phi }_1,\cdots ,{\phi }_n)$  be a holomorphic self-map of $B$  and $\psi$  a holomorphic function of $B,p>0,q>0$  , then $W\psi ,\phi :{\beta }^p\to {\beta }^q$  is bounded if and only if

$$\begin{array}{c}\begin{array}{c}{sup}_{\stackrel{u\in {\text{C}}^n-\left\{0\right\}}{\omega \in B_n}}\left|\psi \right(\omega \left)\right|\frac{(1-|w{|}^2{)}^q}{(1-|\phi \left(w\right){|}^2{)}^p}{\left\{\frac{(1-|\phi \left(w\right){|}^2\left)\right|J_{\phi }\left(w\right)u{|}^2+|<\phi (w),J_{\phi }(w)u>{|}^2}{(1-|w{|}^2\left)\right|u{|}^2+|<w,u>{|}^2}\right\}}^{\frac{1}{2}}<\infty \end{array}\end{array}$$

(11)

and

$$\begin{array}{c}{sup}_{\omega \in B}{G}_p\left(\phi \right(\omega \left)\right)|\nabla \psi (\omega \left)\right|(1-|\omega {|}^2{)}^q<\infty .\end{array}$$

(12)

Corollary 2. Let $\phi =({\phi }_1,\cdots ,{\phi }_n)$  be a holomorphic self-map of $B$  and $\psi \left(z\right)$  a holomorphic function of $B,0<p<1,q>0$  , then $W\psi ,\phi :{\beta }^p\to {\beta }^q$  is compact if and only if $W\psi ,\phi$  is bounded and

$$\begin{array}{c}\begin{array}{c}{sup}_{u\in {\text{C}}^n-\left\{0\right\}}\left|\psi \right(\omega \left)\right|\frac{(1-|w{|}^2{)}^q}{(1-|\phi \left(w\right){|}^2{)}^p}{\left\{\frac{(1-|\phi \left(w\right){|}^2\left)\right|J_{\phi }\left(w\right)u{|}^2+|<\phi (w),J_{\phi }(w)u>{|}^2}{(1-|w{|}^2\left)\right|u{|}^2+|<w,u>{|}^2}\right\}}^{\frac{1}{2}}\to 0\end{array}\end{array}$$

(13)

when $\phi \left(\omega \right)\to \partial B$  .

Corollary 3. Let $\phi =({\phi }_1,\cdots ,{\phi }_n)$  be a holomorphic self-map of $B$  and $\psi \left(z\right)$  a holomorphic function of $B,p\ge 1,q>0$  , then $W\psi ,\phi :{\beta }^p⟶{\beta }^q$  is compact if and only if $W\psi ,\phi$  is bounded and

$$\begin{array}{c}\begin{array}{c}{sup}_{u\in {\text{C}}^n-\left\{0\right\}}\left|\psi \right(z\left)\right|\frac{(1-|w{|}^2{)}^q}{(1-|\phi \left(w\right){|}^2{)}^p}{\left\{\frac{(1-|\phi \left(w\right){|}^2\left)\right|J_{\phi }\left(w\right)u{|}^2+|<\phi (w),J_{\phi }(w)u>{|}^2}{(1-|w{|}^2\left)\right|u{|}^2+|<w,u>{|}^2}\right\}}^{\frac{1}{2}}\to 0\end{array}\end{array}$$

(14)

$$\begin{array}{c}\begin{array}{c}{G}_p\left(\phi \right(\omega \left)\right)|\nabla \psi (\omega \left)\right|(1-|\omega {|}^2{)}^q\to 0\end{array}\end{array}$$

(15)

when $\phi \left(\omega \right)\to \partial B$  .

2 Some Lemmas

In the following, we will use the symbol $c$  to denote a finite positive number which does not depend on variables $z,a,\omega$  and may depend on some norms and parameters $p,q,n,\alpha ,x,f$  etc, not necessarily the same at each occurrence.

In order to prove the main result, we will give some Lemmas first.

Lemma 2.1 If $0<p,s<+\infty ,-n-1<q<+\infty ,q+s>-1$  , then $F(p,q,s)\subset {\beta }^{\frac{n+1+q}{p}}$  and $\exists c>0$  s.t. for $\forall f\in F(p,q,s),$  $\parallel f{\parallel }_{{\beta }^{\frac{n+1+q}{p}}}\le c\parallel f{\parallel }_{F(p,q,s)}.$  Proof Suppose $f\in F(p,q,s).$  Fixed $0<r_0<1,$  since $\left(Rf\right)\circ {\phi }_a\in H\left(B\right),$  so $\left|\right(Rf)\circ {\phi }_a{|}^p$  is subharmonic in $B$  . That is

$$\begin{array}{c}\begin{array}{cc}\left|Rf\right(a){|}^p=& \left|\right(Rf)\circ {\phi }_a(0){|}^p\le \frac{1}{r_0^{2n}}{\int }_{r_{0}B}|\left(Rf\right)\circ {\phi }_a\left(\omega \right){|}^{p}dv\left(\omega \right)\\ & =\frac{1}{r_0^{2n}}{\int }_{r_{0}B}\left|\right(Rf\left){|}^p\frac{(1-|a|2{)}^{n+1}}{|1-<z,a>{|}^{(2n+2)}}dv\right(z).\end{array}\end{array}$$

(16)

From (5) in [17], we have $$\frac{1-r_0}{1+r_0}(1-|a\left|2\right)\le (1-|z\left|2\right)\le \frac{1+r_0}{1-r_0}(1-|z\left|2\right)$$  as $z\in {\phi }_a\left(r_{0}B\right)$  . Thus

$$\begin{array}{c}\frac{(1-|a|2{)}^{n+1}}{|1-<z,a>{|}^{2n+2}(1-\left|z\right|2{)}^q{g}^s(z,a)}\le \frac{4^{n+1}}{(1-|a|2{)}^{n+1+q}}(\frac{1+r_0}{1-r_0}{)}^{\left|q\right|}{log}^{-s}\frac{1}{r_0}.\end{array}$$

(17)

From (16) and (17), we get

$$\begin{array}{cc}& \left|Rf\right(a){|}^p\le \frac{1}{r_0^{2n}}{\int }_{{\phi }_a\left(r_{0}B\right)}|Rf\left(z\right){|}^p\frac{(1-|a|2{)}^{n+1}}{|1-<z,a>{|}^{2n+2}}dv\left(z\right)\end{array}$$

$$\begin{array}{cc}& =\frac{1}{r_0^{2n}}{\int }_{{\phi }_a\left(r_{0}B\right)}\left|Rf\right(z\left){|}^p\right(1-\left|z\right|2{)}^{q}gs(z,a)\frac{(1-|a|2{)}^{n+1}}{|1-<z,a>{|}^{2n+2}(1-\left|z\right|2{)}^{q}gs(z,a)}dv\left(z\right)\end{array}$$

$$\begin{array}{cc}& \le \frac{4^{n+1}{r}_0^{-2n}}{(1-|a|2{)}^{n+1+q}}(\frac{1+r_0}{1-r_0}{)}^{\left|q\right|}{log}^{-s}\frac{1}{r_0}\parallel f{\parallel }_{F(p,q,s)}^p.\end{array}$$

This shows that $f\in {\beta }^{\frac{n+1+q}{p}}$  and $\parallel f{\parallel }_{{\beta }^{\frac{n+1+q}{p}}}\le c\parallel f{\parallel }_{F(p,q,s)}.$  Lemma 2.2 Let $p>0$  , then there is a constant $c>0,$  for $\forall f\in {\beta }^p$  and $\forall z\in B,$  the estimate

$$\begin{array}{c}\left|f\right(z\left)\right|\le c{G}_{\alpha }\left(z\right)\parallel f{\parallel }_{{\beta }^p},\end{array}$$

(18)

holds, where the function $G_{\alpha }$  has been defined at ( 7 ).

Proof For $f\in {\beta }^p\left(B_n\right),$  since $\left|\right|f|{|}_{{\beta }^p}=|f\left(0\right)|+{sup}_{z\in B_n}(1-\left|z{|}^2{)}^p\right|\nabla f\left(z\right)|$  , we have $$\left|f\right(0\left)\right|\le \left|\right|f|{|}_{{\beta }^p},\text{and}|\nabla f\left(z\right)|\le \frac{\left|\right|f|{|}_{{\beta }^p}}{(1-|z{|}^2{)}^p}.$$  but $$f\left(z\right)=f\left(0\right)+{\int }_0^1<z,\overline{\nabla f\left(tz\right)}>dt.$$  therefore

$$\begin{array}{ccc}& & \left|f\right(z\left)\right|\le \left|f\right(0\left)\right|+{\int }_0^1\left|z\right||\nabla f(tz\left)\right|dt\end{array}$$

$$\begin{array}{ccc}& \le & \left|\right|f|{|}_{{\beta }^p}+|\left|f\right|{|}_{{\beta }^p}{\int }_0^1\frac{1}{(1-|tz{|}^2{)}^p}dt\le \left|\right|f|{|}_{{\beta }^p}\left(1+{\int }_0^{\left|z\right|}\frac{dt}{(1-t^2{)}^p}\right).\end{array}$$

when $p=1$  , ${\int }_0^{\left|z\right|}\frac{dt}{1-t^2}=\frac{1}{2}ln\frac{1+\left|z\right|}{1-\left|z\right|}\le \frac{1}{2}ln\frac{4}{1-|z{|}^2}$  , therefore $$\left|f\right(z\left)\right|\le \left(1+\frac{1}{2}ln\frac{4}{1-|z{|}^2}\right)\left|\right|f|{|}_{{\beta }^p}.$$  If $p\ne 1$  , then $${\int }_0^{\left|z\right|}\frac{dt}{(1-t^2{)}^p}={\int }_0^{\left|z\right|}\frac{dt}{(1-t{)}^p(1+t{)}^p}\le {\int }_0^{\left|z\right|}\frac{dt}{(1-t{)}^p}=\frac{1-(1-|z|{)}^{1-p}}{1-p}$$  therefore when $0<p<1$  , notice that ${\int }_0^{\left|z\right|}\frac{dt}{(1-t^2{)}^p}\le \frac{1}{1-p}$  we get $$\Rightarrow \left|f\right(z\left)\right|\le \left(1+\frac{1}{1-p}\right)\left|\right|f|{|}_{{\beta }^p}.$$  and when $p>1$  $${\int }_0^{\left|z\right|}\frac{dt}{(1-t^2{)}^p}\le \frac{1-(1-|z|{)}^{1-p}}{1-p}=\frac{1-(1-|z|{)}^{p-1}}{(p-1)(1-|z|{)}^{p-1}}\le \frac{2^{p-1}}{(p-1)(1-|z{|}^2{)}^{p-1}}$$  so $$\left|f\right(z\left)\right|\le \left(1+\frac{2^{p-1}}{(p-1)(1-|z{|}^2{)}^{p-1}}\right)\left|\right|f|{|}_{{\beta }^p}.$$  Lemma 2.3 Let $\phi$  be a holomorphic self-map of $B$  and $\psi$  a holomorphic function on $B$  , $K$  is an arbitrary point set. Then $W\psi ,\phi :F(p,q,s)\to \beta \alpha$  is compact if and only if for any uniformly bounded sequence $\left\{f_{j,u}\right\}(j\in \text{N},u\in K)$  in $F(p,q,s)$  which converges to zero uniformly for $u\in K$  and $z$  on compact subsets of $B$  when $j\to \infty$  , $\parallel W\psi ,\phi f_{j,u}{\parallel }_{\alpha }\to 0.$  Proof Suppose that $W\psi ,\phi$  is compact and $\left\{f_{j,u}\right\}$  is a uniformly bounded sequence in $F(p,q,s)$  which converges to zero uniformly for $u\in K$  and $z$  on compact subsets of $B$  when $j\to \infty$  . Then $\{W\psi ,\phi (f_{j,u}\left)\right\}$  has a subsequence $\{W\psi ,\phi (f_{j_m,u_k}\left)\right\}$  converges to $g\in \beta \alpha$  . And by lemma 2.2, $\forall$  compact subset $M\subset B,$  there is a positive constant $C_M$  independent of $f_{j,u}$  such that $$\left|\psi \right(z\left)f_{j_m,u_k}\right(\phi \left(z\right))-g(z\left)\right|\le C_M\parallel \psi f_{j_m,u_k}\circ \phi -g{\parallel }_{\alpha },$$  for all $z\in M.$  Therefore $\left\{\psi \right(z\left)f_{j_m,u_k}\right(\phi \left(z\right))-g(z\left)\right\}$  converges to 0 uniformly on M. Notice that there is a constant $c>0$  , such that $\left|\psi \right(z\left)\right|<c,\forall z\in M$  , and that $\phi \left(M\right)$  is compact in B, we have $\left|\psi \right(z\left)f_{j_m,u_k}\right(\phi \left(z\right)\left)\right|\le c\left|f_{j_m,u_k}\right(\phi \left(z\right)\left)\right|\to 0,$  uniformly on $M$  . And for thearbitrariness of $M$  , we have $g\equiv 0$  . Since it is true for arbitrary subsequence of $\left\{f_{j,u}\right\}$  , we see that $W\psi ,\phi f_{j,u}\to 0$  in $\beta \alpha$  , when $j\to +\infty .$  On the other hand, fixed a point $b$  in $K$  . Suppose $\left\{g_{j,b}\right\}\subset K_r=B_{F(p,q,s)}(0,r)$  , where $B_{F(p,q,s)}(0,r)$  is a ball in $F(p,q,s),$  then by lemma 2.2, $\left\{g_{j,b}\right\}$  is uniformly bounded in arbitrary compact subset M of $B$  . By Montel's Lemma, $\left\{g_{j,b}\right\}$  is regular, therefore there is a subsequence $\left\{g_{j_m,b}\right\}$  which converges uniformly to $g_b\in H\left(B\right)$  on compact subsets of $B$  . It follows that $\nabla g_{j_m,b}\to \nabla g_b$  uniformly on compact subsets of B. Denote $B_k=B(0,1-\frac{1}{k})\subset {\text{C}}^n$  , then

$$\begin{array}{cc}& {\int }_B|\nabla g_b{|}^p(1-\left|z\right|2{)}^{q}gs(z,a)dv\left(z\right)\end{array}$$

$$\begin{array}{cc}& ={lim}_{k\to +\infty }{\int }_{B_k}{lim}_{m\to +\infty }|\nabla g_{j_m,b}{|}^p(1-\left|z\right|2{)}^{q}gs(z,a)dv\left(z\right)\end{array}$$

$$\begin{array}{cc}& ={lim}_{k\to +\infty }{lim}_{m\to +\infty }{\int }_{B_k}|\nabla g_{j_m,b}{|}^p(1-\left|z\right|2{)}^{q}gs(z,a)dv\left(z\right)\end{array}$$

But $\left\{g_{j_m,b}\right\}\subset B_{F(p,q,s)}(0,r)$  , then ${\int }_{B_k}|\nabla g_{j_m,b}{|}^p(1-\left|z\right|2{)}^{q}gs(z,a)dv\left(z\right)<r^p$  , therefore $${\int }_B|\nabla g_b{|}^p(1-\left|z\right|2{)}^{q}gs(z,a)dv\left(z\right)\le r^p.$$  So $\parallel g_b{\parallel }_{F(p,q,s)}\le r,$  and $g_b\in F(p,q,s).$  Hence the sequence $\{g_{j_m,b}-g_b\}$  is such that $\parallel g_{j_m,b}-g_b\parallel \le 2r<\infty$  and converges to $0$  on compact subsets of B, by the hypothesis of this lemma, we have that $\psi g_{j_m,b}\circ \phi \to \psi g_b\circ \phi$  in $\beta \alpha$  . Thus the set $W\psi ,\phi \left(K_r\right)$  is relatively compact, finishing the proof.

Lemma 2.4 Let $0<p<\infty ,f\in H\left(B_n\right),$  then $f\in {\beta }^p\left(B_n\right)$  if and only if $${sup}_{\stackrel{u\in {\text{C}}^n-\left\{0\right\}}{z\in B_n}}\frac{(1-|z{|}^2{)}^p|\nabla f(z\left)u\right|}{\sqrt{(1-|z{|}^2\left)\right|u{|}^2+|<z,u>{|}^2}}<\infty .$$  furthermore $$\left|\right|f|{|}_{{\beta }^p}\approx |f\left(0\right)|+{sup}_{\stackrel{u\in {\text{C}}^n-\left\{0\right\}}{z\in B_n}}\frac{(1-|z{|}^2{)}^p|\nabla f(z\left)u\right|}{\sqrt{(1-|z{|}^2\left)\right|u{|}^2+|<z,u>{|}^2}}.$$  Proof By Bergman metric in [1], there are constants $A_1>0$  and $A_2>0$  , such that

$$\begin{array}{cc}& {sup}_{u\in {\text{C}}^n-\left\{0\right\}}\frac{A_1|\nabla f(z\left)u\right|}{\sqrt{(1-|z\left|2\right)|u{|}^2+|<z,u>{|}^2}}\le |\nabla f(z\left)\right|\end{array}$$

$$\begin{array}{cc}& \le {sup}_{u\in {\text{C}}^n-\left\{0\right\}}\frac{A_2|\nabla f(z\left)u\right|}{\sqrt{(1-|z\left|2\right)|u{|}^2+|<z,u>{|}^2}}.\end{array}$$

Let $C_1=min\{A_1,1\}$  , $C_2=min\{A_2,1\}$  , we have

$$\begin{array}{c}\begin{array}{cc}& C_1\left\{\right|f\left(0\right)|+{sup}_{u\in {\text{C}}^n-\left\{0\right\}}\frac{(1-|z\left|2{)}^p\right|\nabla f\left(z\right)u|}{\sqrt{(1-|z\left|2\right)|u{|}^2+|<z,u>{|}^2}}\}\le \parallel f{\parallel }_{{\beta }^p}\\ & C_2\left\{\right|f\left(0\right)|+{sup}_{u\in {\text{C}}^n-\left\{0\right\}}\frac{(1-|z\left|2{)}^p\right|\nabla f\left(z\right)u|}{\sqrt{(1-|z\left|2\right)|u{|}^2+|<z,u>{|}^2}}\}\end{array}\end{array}$$

(19)

This proof is completed.

Lemma 2.5 For $0<p,s<+\infty ,-n-1<q<+\infty ,q+s>-1$  , there exists $c>0$  such that

$$\begin{array}{c}\begin{array}{c}{sup}_{a\in B}{\int }_B\frac{(1-|\omega |2{)}^p}{|1-<z,w>{|}^{n+1+q+p}}(1-|z\left|2{)}^{q}gs\right(z,a\left)dv\right(z)\le c,\end{array}\end{array}$$

(20)

for every $\omega \in B$  . Furthermore when $p+s>n$  , then the inequility

$$\begin{array}{c}\begin{array}{c}{sup}_{a\in B}{\int }_B\frac{(1-|\omega |2{)}^p}{|1-<z,w>{|}^p}(1-|z\left|2{)}^{-n-1}gs\right(z,a\left)dv\right(z)\le c\end{array}\end{array}$$

(21)

holds.

Proof It's easy to verify that if $0<x<\frac{1}{2}$  , then $-log(1-x)<4x$  . If we let $x=1-\left|{\phi }_a\right(z){|}^2$  , then $-log(1-(1-\left|{\phi }_a\right(z\left){|}^2\right))<4(1-\left|{\phi }_a\right(z\left){|}^2\right)$  , or $g(z,a)<2(1-|{\phi }_a\left(z\right){|}^2),$  when $\left|{\phi }_a\right(z){|}^2>\frac{1}{2}$  . Note that $1-\left|{\phi }_a\right(z){|}^2=\frac{(1-|a\left|2\right)(1-|z\left|2\right)}{|1-<z,a>{|}^2}$  . Therefore when $s>n$ 

$$\begin{array}{c}\begin{array}{cc}& {\int }_{1>\left|{\phi }_a\right(z){|}^2>\frac{1}{2}}\frac{(1-|\omega |2{)}^p}{|1-<z,w>{|}^{n+1+q+p}}(1-|z\left|2{)}^{q}gs\right(z,a\left)dv\right(z)\\ & \le {\int }_{1>\left|{\phi }_a\right(z){|}^2>\frac{1}{2}}\frac{(1-|\omega \left|2{)}^p\right(1-\left|z\right|2{)}^q{2}^s(1-|a\left|2{)}^s\right(1-\left|z\right|2{)}^s}{|1-<z,w>{|}^{n+1+q+p}|1-<z,a>{|}^{2s}}dv\left(z\right)\\ & ={\int }_{1>\left|{\phi }_a\right(z){|}^2>\frac{1}{2}}\frac{2^s(1-|\omega \left|2{)}^p\right(1-\left|a\right|2{)}^s(1-|z|2{)}^{q+s}}{|1-<z,w>{|}^{n+1+q+p}|1-<z,a>{|}^{2s}}dv\left(z\right)\\ & \le {\int }_{B}c\frac{(1-|a\left|2{)}^s\right(1-\left|z\right|2{)}^{s-n-1}}{|1-<z,a>{|}^{2s}}dv\left(z\right)<c\end{array}\end{array}$$

(22)

where the last inequality follows by theorem 1.4.10 in [16] and the fact that $|1-<z,w>|\ge 1-\left|z\right|$  .

And if $s\le n$  , we choose constants $x,x^{\prime },\lambda$  satisfying $$max\{1,\frac{n}{q+n+1}\}<x<\frac{n}{n-s}$$  (when s=n, just consider $x$  such that $x>max\{1,\frac{n}{q+n+1}\})$  , and let $$\lambda =q+n+1-\frac{n+1}{x},\frac{1}{x}+\frac{1}{x^{\prime }}=1,$$  then $\lambda x>-1$  and $(q+s-\lambda )x^{\prime }>-1$  . Thus

$$\begin{array}{c}\begin{array}{cc}& {\int }_{1>\left|{\phi }_a\right(z){|}^2>\frac{1}{2}}\frac{(1-|\omega |2{)}^p}{|1-<z,w>{|}^{n+1+q+p}}(1-|z\left|2{)}^{q}gs\right(z,a\left)dv\right(z)\\ & \le {\int }_{1>\left|{\phi }_a\right(z){|}^2>\frac{1}{2}}\frac{(1-|\omega \left|2{)}^p\right(1-\left|z\right|2{)}^q{2}^s(1-|a\left|2{)}^s\right(1-\left|z\right|2{)}^s}{|1-<z,w>{|}^{n+1+q+p}|1-<z,a>{|}^{2s}}dv\left(z\right)\\ & \le 2^s{\int }_B\frac{(1-|\omega \left|2{)}^p\right(1-\left|z\right|2{)}^{\lambda }}{|1-<z,w>{|}^{n+1+q+p}}\frac{(1-|z\left|2{)}^{q+s-\lambda }\right(1-\left|a\right|2{)}^s}{|1-<z,a>{|}^{2s}}dv\left(z\right)\\ & \le c\left\{{\int }_B\frac{(1-|\omega \left|2{)}^{px}\right(1-\left|z\right|2{)}^{\lambda x}}{|1-<z,w>{|}^{(n+1+q+p)x}}dv\right(z\left){\}}^{\frac{1}{x}}\right\{{\int }_B\frac{(1-|z\left|2{)}^{(q+s-\lambda )x^{\prime }}\right(1-\left|a\right|2{)}^{s{x}^{\prime }}}{|1-<z,a>{|}^{2s{x}^{\prime }}}dv\left(z\right){\}}^{\frac{1}{x^{\prime }}}\\ & \le c\end{array}\end{array}$$

(23)

At the same time

$$\begin{array}{c}\begin{array}{cc}& {\int }_{\left|{\phi }_a\right(z){|}^2\le \frac{1}{2}}\frac{(1-|\omega |2{)}^p}{|1-<z,w>{|}^{n+1+q+p}}(1-|z\left|2{)}^{q}gs\right(z,a\left)dv\right(z)\\ & ={\int }_{\left|u\right|\le \frac{1}{2}}\frac{(1-|\omega \left|2{)}^p\right(1-\left|{\phi }_a\right(u\left){|}^2{)}^q\right(1-\left|a\right|2{)}^{n+1}}{|1-<{\phi }_a(u),\omega >{|}^{n+1+q+p}|1-<u,a>{|}^{2n+2}}{log}^s\frac{1}{\left|u\right|}dv\left(u\right)\\ & \le c{\int }_{\left|u\right|\le \frac{1}{2}}\frac{1}{|1-<{\phi }_a(u),\omega >{|}^{n+1}|1-<u,a>{|}^{n+1}}{log}^s\frac{1}{\left|u\right|}dv\left(u\right)\\ & =c{\int }_{\left|u\right|\le \frac{1}{2}}\frac{1}{(1-|u|2{)}^{n+1}}{log}^s\frac{1}{\left|u\right|}dv\left(u\right)\\ & \le c{\int }_B{log}^s\frac{1}{\left|u\right|}dv\left(u\right)<c\end{array}\end{array}$$

(24)

.

Combine ( 22 ),( 23 ),and ( 24 ), we have $${sup}_{a\in B}{\int }_B\frac{(1-|\omega |2{)}^p}{|1-<z,w>{|}^{n+1+q+p}}(1-|z\left|2{)}^{q}gs\right(z,a\left)dv\right(z)\le c.$$  We can use the same method to prove that ( 21 ) holds, we omit the proof here. $♯$  Remark: we can also prove that when $s>n$  , then $\beta \alpha \subset F(p,q,s)$  , if we use the same method used in ( 22 ) and ( 24 ). Therefore, combine lemma 2.1, we conclude that ${\beta }^{\frac{q+1+n}{p}}=F(p,q,s)$  .

Lemma 2.6 If $W\psi ,\phi :F(p,q,s)\to \beta \alpha$  is bounded, then for every $\omega \in B$  satisfies $\left|\phi \right(\omega \left)\right|>\sqrt{\frac{2}{3}}$  and every $u\in {\text{C}}^n-\left\{0\right\},$  there is a function $f\omega ,u\in F(p,q,s)$  such that i) $\exists c>0,$  independent of $\omega$  and $u$  , s.t. $$\parallel f\omega ,u\parallel F(p,q,s)<c$$  $ii)$  $\{f\omega ,u\}$  converges to zero uniformly for $u\in {\text{C}}^n-\left\{0\right\}$  and z on compact subsets of $B$  , when $\phi \left(\omega \right)\to \partial B$  .

$iii)$  There is a constant $c>0$  , for $\forall \omega$  and $u$ 

$$\begin{array}{c}\begin{array}{c}\parallel W\psi ,\phi f\omega ,u{\parallel }_{\beta \alpha }\ge c\left|\psi \right(\omega \left)\right|\frac{(1-|w\left|2\right)\alpha }{(1-|\phi \left(w\right)\left|2\right)\frac{q+n+1}{p}}\left\{\frac{(1-|\phi \left(w\right)\left|2\right)\left|J\phi \right(w\left)u\right|2+|<\phi (w),J\phi (w)u>|2}{(1-|w\left|2\right)\left|u\right|2+|<w,u>|2}\right\}\frac{1}{2}\end{array}\end{array}$$

(25)

holds Proof First we suppose $\phi \left(\omega \right)=r\omega e_1,$  where $r\omega =\left|\phi \right(\omega \left)\right|$  , $e_1$  is the vector $(1,0,0,\cdots ,0)$  .

If $\left|\phi \right(\omega \left)\right|>\sqrt{2/3}$  and $\sqrt{1-\left|\phi \right(\omega \left)\right|2}\left|J\phi \right(\omega \left)u\right|\le |<\phi (\omega ),J\phi (\omega )u>|$  .

Take

$$\begin{array}{c}f\omega ,u\left(z\right)=\frac{(z_1-r\omega )(1-r2\omega )}{(1-r\omega z_1{)}^{\frac{n+1+q}{p}+1}}\end{array}$$

(26)

Then $$\frac{\partial f\omega ,u\left(z\right)}{\partial z_1}=\frac{1-r2\omega }{(1-r\omega z_1{)}^{\frac{n+1+q}{p}+1}}\left(1+\frac{n+1+q}{p}\frac{(z_1-r\omega )r\omega }{1-r\omega z_1}\right)$$  and $$\frac{\partial f\omega ,u\left(z\right)}{\partial z_k}=0,k=2,\cdots ,n.$$  Therefore

$$\begin{array}{c}\begin{array}{cc}|\nabla f\omega ,u(z\left)\right|& =\frac{1-r2\omega }{|1-r\omega z_1{|}^{\frac{n+1+q}{p}+1}}|1+\frac{n+1+q}{p}\frac{(z_1-r\omega )r\omega }{1-r\omega z_1}|\\ & \le (1+\frac{n+1+q}{p})\frac{1-r2\omega }{|1-r\omega z_1{|}^{\frac{n+1+q}{p}+1}}\end{array}\end{array}$$

(27)

By Lemma 2.5, $f\omega ,u\in F(p,q,s)$  , and exists $c>0$  independent of $\omega$  and $u$  , s.t.

$$\parallel f\omega ,u\parallel \le c.$$  or $i)$  holds.

On the other hand, by ( 2 ), ( 21 ) and ( 27 ), notice that $$<\phi \left(\omega \right),J\phi \left(\omega \right)u>=r\omega e_{1}J\phi \left(\omega \right)u,$$  and use lemma 2.4,

$$\begin{array}{c}\begin{array}{cc}\parallel W\psi ,\phi f\omega ,u{\parallel }_{\beta \alpha }& \ge (1-|\omega \left|2{)}^{\alpha }\right|\nabla (W\psi ,\phi f\omega ,u)\left(\omega \right)|\\ & =(1-|\omega \left|2{)}^{\alpha }\right|\psi \left(\omega \right)\nabla (f_{\omega }\circ \phi )\left(\omega \right)|\\ & \ge c(1-|\omega \left|2{)}^{\alpha }\right|\psi \left(\omega \right)\left|\right|\frac{\nabla f\omega ,u\left(\phi \right(\omega \left)\right)J\phi \left(\omega \right)u}{\sqrt{(1-|\omega \left|2\right)|u{|}^2+|<\omega ,u>{|}^2}}|\\ & =c(1-|\omega \left|2{)}^{\alpha }\right|\psi \left(\omega \right)\left|\frac{1-r2\omega }{(1-r2\omega {)}^{\frac{n+1+q}{p}+1}}\right|\frac{e_{1}J\phi \left(\omega \right)u}{\sqrt{(1-|\omega \left|2\right)|u{|}^2+|<\omega ,u>{|}^2}}|\\ & =c\left|\psi \right(\omega \left)\right|\frac{(1-|\omega |2{)}^{\alpha }}{r\omega (1-r2\omega {)}^{\frac{n+1+q}{p}}}\left|\frac{<\phi \left(\omega \right),J\phi \left(\omega \right)u>}{\sqrt{(1-|\omega \left|2\right)\left|u\right|2+|<\omega ,u>|2}}\right|\\ & \ge c\left|\psi \right(\omega \left)\right|\frac{(1-|\omega |2{)}^{\alpha }}{(1-r2\omega {)}^{\frac{n+1+q}{p}}}\frac{\sqrt{|<\phi (\omega ),J\phi (\omega )u>{|}^2+|<\phi \left(\omega \right),J\phi \left(\omega \right)u>}|}{\sqrt{(1-|\omega \left|2\right)\left|u\right|2+|<\omega ,u>|2}}\\ & =c\left|\psi \right(\omega \left)\right|\frac{(1-|w\left|2\right)\alpha }{(1-|\phi \left(w\right)\left|2\right)\frac{q+n+1}{p}}\left\{\frac{(1-|\phi \left(w\right)\left|2\right)\left|J\phi \right(w\left)u\right|2+|<\phi (w),J\phi (w)u>|2}{(1-|w\left|2\right)\left|u\right|2+|<w,u>|2}\right\}\frac{1}{2}\end{array}\end{array}$$

(28)

If $\sqrt{1-\left|\phi \right(\omega \left)\right|2}\left|J\phi \right(\omega \left)u\right|\le |<\phi (\omega ),J\phi (\omega )u>|$  . Write $J_{\phi \left(\omega \right)}u=({\xi }_1,\cdots ,{\xi }_n{)}^T.$  For $j=2,\cdots ,n,$  let ${\theta }_j=arg{\xi }_j$  and $a_j=e^{-i{\theta }_j}$  (if ${\xi }_j=0$  , let $a_j=0.$  ) Let

$$\begin{array}{c}f\omega ,u\left(z\right)=\frac{(a_2{z}_2+\cdots +a_n{z}_n)(1-r2\omega {)}^{3/2}}{(1-r\omega z_1{)}^{\frac{n+1+q}{p}+1}}.\end{array}$$

(29)

then $$\frac{\partial f\omega ,u\left(z\right)}{\partial z_1}=\frac{(\frac{n+1+q}{p}+1)r\omega (1-r2\omega {)}^{3/2}}{(1-r\omega z_1{)}^{\frac{n+1+q}{p}+2}}(a_2{z}_2+\cdots +a_n{z}_n)$$  and $$\frac{\partial f\omega ,u\left(z\right)}{\partial z_k}=\frac{a_k(1-r2\omega {)}^{3/2}}{(1-r\omega z_1{)}^{\frac{n+1+q}{p}+1}},k=2,\cdots ,n.$$  Therefore

$$\begin{array}{c}\begin{array}{cc}|\nabla f\omega ,u(z\left)\right|& =\sqrt{|\frac{\partial f\omega ,u\left(z\right)}{\partial z_1}{|}^2+|\frac{\partial f\omega ,u\left(z\right)}{\partial z_2}{|}^2+\cdots +|\frac{\partial f\omega ,u\left(z\right)}{\partial z_1}{|}^2}\\ & =\sqrt{\frac{(\frac{n+1+q}{p}+1{)}^2{r}^2\omega (1-r2\omega {)}^3|a_2{z}_2+\cdots +a_n{z}_n{|}^2}{|1-r\omega z_1{|}^{2(\frac{n+1+q}{p}+2)}}+\frac{(n-1)(1-r2\omega {)}^3}{|1-r\omega z_1{|}^{2(\frac{n+1+q}{p}+1)}}}\\ & \le \sqrt{\frac{(n-1)(\frac{n+1+q}{p}+1{)}^2{r}^2\omega (1-r2\omega {)}^3\left(\right|z_2{|}^2+\cdots +\left|z_n{|}^2\right)}{|1-r\omega z_1{|}^{2(\frac{n+1+q}{p}+2)}}+\frac{(n-1)(1-r2\omega {)}^3}{|1-r\omega z_1{|}^{2(\frac{n+1+q}{p}+1)}}}\\ & \le \frac{\sqrt{(n-1)}(1-r2\omega {)}^{3/2}}{|1-r\omega z_1{|}^{\frac{n+1+q}{p}+1}}\sqrt{\frac{(\frac{n+1+q}{p}+1{)}^2{r}^2\omega (1-\left|z_1{|}^2\right)}{|1-r\omega z_1{|}^2}+1}\\ & \le \frac{\sqrt{(n-1)}(1-r2\omega )}{|1-r\omega z_1{|}^{\frac{n+1+q}{p}+1}}|1-r2\omega +r^2\omega (\frac{n+1+q}{p}+1{)}^2){|}^{\frac{1}{2}}\\ & \le c\frac{1-r2\omega }{|1-r\omega z_1{|}^{\frac{n+1+q}{p}+1}}\end{array}\end{array}$$

(30)

so, by lemma 2.5, we know that $f\omega ,u\in F(p,q,s),$  and $f\omega ,u$  satisfies $\left(i\right)$  . Next we will extimate $\parallel W\psi ,\phi f\omega ,u{\parallel }_{\alpha }$  .

Since $\left|\phi \right(\omega \left)\right|>\sqrt{2/3}$  and $\sqrt{1-\left|\phi \right(\omega \left)\right|2}\left|J\phi \right(\omega \left)u\right|>|<\phi (\omega ),J\phi (\omega )u>|$  , we can get $\sqrt{3(1-|\phi \left(\omega \right)\left|2\right)\left(\right|{\xi }_2{|}^2+\cdots +\left|{\xi }_n{|}^2\right)}>\left|{\xi }_1\right|.$  Then $|{\xi }_2{|}^2+\cdots +|{\xi }_n{|}^2\ge \frac{1}{2}\left(\right|{\xi }_1{|}^2+\cdots +\left|{\xi }_n{|}^2\right)$  . Therefore

$$\begin{array}{c}\begin{array}{cc}\parallel W\psi ,\phi f\omega ,u{\parallel }_{\alpha }& \ge c(1-|\omega \left|2{)}^{\alpha }\right|\psi \left(\omega \right)\left|\right|\frac{\nabla f\omega ,u\left(\phi \right(\omega \left)\right)J\phi \left(\omega \right)u}{\sqrt{(1-|\omega \left|2\right)|u{|}^2+|<\omega ,u>{|}^2}}|\\ & =c(1-|\omega \left|2{)}^{\alpha }\right|\psi \left(\omega \right)|\frac{(1-|\omega \left|2{)}^{\alpha }\right(1-r2\omega {)}^{\frac{1}{2}}\left(\right|{\xi }_2|+\cdots +|{\xi }_n\left|\right)}{(1-r2\omega {)}^{\frac{n+1+q}{p}}\sqrt{(1-|\omega \left|2\right)\left|u\right|2+|<\omega ,u>|2}}\\ & \ge c(1-|\omega \left|2{)}^{\alpha }\right|\psi \left(\omega \right)|\frac{(1-|\omega \left|2{)}^{\alpha }\right(1-r2\omega {)}^{\frac{1}{2}}\sqrt{|{\xi }_2{|}^2+\cdots +|{\xi }_n{|}^2}}{(1-r2\omega {)}^{\frac{n+1+q}{p}}\sqrt{(1-|\omega \left|2\right)\left|u\right|2+|<\omega ,u>|2}}\\ & \ge c(1-|\omega \left|2{)}^{\alpha }\right|\psi \left(\omega \right)|\frac{(1-|\omega \left|2{)}^{\alpha }\right(1-r2\omega {)}^{\frac{1}{2}}\sqrt{|{\xi }_1{|}^2+|{\xi }_2{|}^2+\cdots +|{\xi }_n{|}^2}}{(1-r2\omega {)}^{\frac{n+1+q}{p}}\sqrt{(1-|\omega \left|2\right)\left|u\right|2+|<\omega ,u>|2}}\\ & \ge c(1-|\omega \left|2{)}^{\alpha }\right|\psi \left(\omega \right)|\frac{(1-|\omega \left|2{)}^{\alpha }\right(1-r2\omega {)}^{\frac{1}{2}}\left|J\phi \right(\omega \left)u\right|}{(1-r2\omega {)}^{\frac{n+1+q}{p}}\sqrt{(1-|\omega \left|2\right)\left|u\right|2+|<\omega ,u>|2}}\\ & \ge c\left|\psi \right(\omega \left)\right|\frac{(1-|w\left|2\right)\alpha }{(1-|\phi \left(w\right)\left|2\right)\frac{q+n+1}{p}}\left\{\frac{(1-|\phi \left(w\right)\left|2\right)\left|J\phi \right(w\left)u\right|2+|<\phi (w),J\phi (w)u>|2}{(1-|w\left|2\right)\left|u\right|2+|<w,u>|2}\right\}\frac{1}{2}.\end{array}\end{array}$$

(31)

By our discussion above, for every $\omega \in B$  satisfies $\left|\phi \right(\omega \left)\right|>\sqrt{2/3}$  and $u\in {\text{C}}^n-\left\{0\right\}$  , there is a $f\omega ,u\in F(p,q,s)$  satisfies $i)$  and $iii)$  . And that $f\omega ,u$  has the property $ii)$  is clearly, since it is very easy to verify that for every $\omega$  and $u$  $$|f\omega ,u(z\left)\right|\le c\frac{1-r2\omega }{(1-|z|2{)}^{\frac{n+1+q}{p}+1}}.$$  In general situation, or if $\phi \left(\omega \right)\ne \left|\phi \right(\omega \left)\right|e_1,$  we use the unitary transformation $U\omega$  to make $\phi \left(\omega \right)=r\omega e_{1}U\omega$  , where $r\omega =\left|\phi \right(\omega \left)\right|>\sqrt{2/3}$  . Then $g\omega ,u=f\omega ,u\circ U_{\omega }^{-1}$  is the desired function.

In fact, by $\nabla g\omega ,u\left(z\right)=\nabla (f\omega ,u\cdot U_{\omega }^{-1})\left(z\right)=(\nabla f\omega ,u)\left(z{U}_{\omega }^{-1}\right)(U_{\omega }^{-1}{)}^T$  and $\left|z{U}_{\omega }^{-1}\right|=\left|z\right|$  ,we have

$$\begin{array}{cc}& {\int }_B|\nabla g\omega ,u(z\left){|}^p\right(1-\left|z\right|2{)}^{q}gs(z,a)dv\left(z\right)\end{array}$$

$$\begin{array}{cc}& ={\int }_B\left|\right(\nabla f\omega ,u\left)\right(z{U}_{\omega }^{-1}\left)\right(U_{\omega }^{-1}{)}^T{|}^p(1-|z\left|2{)}^{q}gs\right(z,a\left)dv\right(z)\end{array}$$

$$\begin{array}{cc}& ={\int }_B|\nabla f\omega ,u(z\left){|}^p\right(1-\left|z\right|2{)}^{q}gs(z,a)dv\left(z\right)\end{array}$$

where the last equation we use the linear coordinate translation $z=z{U}_{\omega }^{-1}$  and the fact that $gs(z,a)$  is invariant under $m\ddot{o}bious$  translation. So, $\parallel f\omega ,u\parallel F(p,q,s)=\parallel g\omega ,u\parallel F(p,q,s)$  .

Then we can prove the same result by the same way, so we omit the details. $♯$  Lemma 2.7 Let $0\le p<1,$  then for every $f\in {\beta }^p$  and $\forall z,w\in B$  ,

$$\begin{array}{c}\left|f\right(z)-f(w\left)\right|\le \frac{2\parallel f{\parallel }_p}{1-p}|z-w{|}^{1-p}\end{array}$$

(32)

especially when $0<\frac{n+1+q}{p}<1,$  there is a constant $c>0$  such that

$$\begin{array}{c}\left|f\right(z)-f(w\left)\right|\le c\frac{\parallel f{\parallel }_{F(p,q,s)}}{1-p}|z-w{|}^{1-p}\end{array}$$

(33)

$\forall f\in F(p,q,s)$  .

Proof Denote $F\left(t\right)=f(tz+(1-t\left)w\right)$  . Then

$$\begin{array}{cc}f\left(z\right)-f\left(w\right)& =F\left(1\right)-F\left(0\right)\end{array}$$

$$\begin{array}{cc}& ={\int }_0^1{F}^{\prime }\left(t\right)dt\end{array}$$

$$\begin{array}{cc}& ={\int }_0^1{\sum }_{k=1}^n\frac{\partial f(tz+(1-t)w}{\partial {\xi }_k}(z_k-w_k)dt\end{array}$$

$$\begin{array}{cc}& ={\int }_0^1<\nabla f(tz+(1-t\left)w\right),\overline{z-w}>dt\end{array}$$

so

$$\begin{array}{cc}\left|f\right(z)-f(w\left)\right|& =|{\int }_0^1<\nabla f(tz+(1-t)w),\overline{z-w}>dt|\end{array}$$

$$\begin{array}{cc}& \le {\int }_0^1|\nabla f(tz+(1-t)w\left)\right||z-w|dt\end{array}$$

Since $f\in {\beta }^p,$  it's easy to show that $$|\nabla f(z\left)\right|\le \frac{\parallel f{\parallel }_{{\beta }^p}}{(1-|z|2{)}^p}\le \frac{\parallel f{\parallel }_{{\beta }^p}}{(1-|z|{)}^p},\forall z\in B.$$  Note that if two numbers $a,b$  , satisfy $|a+b|<1,\left|a\right|<1$  and $\left|b\right|<1$  , then $|a+b|+\left|a\right|+\left|b\right|<2$  and so $1-|a+b|>\left|\right(1-\left|a\right|-\left|b\right|\left)\right|$  . Therefore

$$\begin{array}{cc}\left|f\right(z)-f(w\left)\right|& \le {\int }_0^1\frac{\parallel f{\parallel }_{{\beta }^p}|z-w|}{(1-|w+t(z-w)|{)}^p}dt\end{array}$$

$$\begin{array}{cc}& \le {\int }_0^1\frac{\parallel f{\parallel }_{{\beta }^p}|z-w|}{|1-|w|-t|z-w|{|}^p}dt\end{array}$$

If $1-\left|w\right|-|z-w|\ge 0$  , then

$$\begin{array}{c}\begin{array}{cc}\left|f\right(z)-f(w\left)\right|& \le {\int }_0^1\frac{\parallel f{\parallel }_{{\beta }^p}|z-w|}{(1-|w|-t|z-w|{)}^p}dt\\ & \le \frac{\parallel f{\parallel }_{{\beta }^p}}{1-p}(1-|w|-t|z-w|{)}^{(1-p)}{|}_1^0\\ & =\frac{\parallel f{\parallel }_{{\beta }^p}}{1-p}\left\{\right(1-\left|w\right|{)}^{1-p}-(1-|w|-|z-w\left|{)}^{1-p}\right\}\end{array}\end{array}$$

(34)

Write $\phi \left(x\right)=x^{1-p}-b^{1-p}-(x-b{)}^{1-p},x\in (a,b)$  , then $${\phi }^{\prime }\left(x\right)=(1-p)x^{-p}-(1-p)(x-b{)}^{-p}=(1-p\left)\right(\frac{1}{x^p}-\frac{1}{(x-b{)}^p})<0,$$  since $b<x<a$  . Therefore $\phi \left(a\right)<\phi \left(b\right)=0$  , or $a^{1-p}-b^{1-p}-(a-b{)}^{1-p}<0$  , if $b<a$  . If we let $a=1-\left|w\right|,b=|z-w|,$  then $$(1-|w|{)}^{1-p}<|z-w{|}^{1-p}+(1-|w|-|z-w|{)}^{1-p}.$$  Combine (34), we have $$\left|f\right(z)-f(w\left)\right|\le \frac{\parallel f{\parallel }_p}{1-p}|z-w{|}^{1-p}.$$  If $1-\left|w\right|-|z-w|<0$  , then there is a $\xi \in (0,1),$  s.t. $$1-\left|w\right|-\xi |z-w|=0.$$  so

$$\begin{array}{cc}\left|f\right(z)-f(w\left)\right|& \le {\int }_0^{\xi }\frac{\parallel f{\parallel }_{{\beta }^p}|z-w|}{(1-|w|-t|z-w|{)}^p}dt+{\int }_{\xi }^1\frac{\parallel f{\parallel }_{{\beta }^p}|z-w|}{\left(\right|w|+t|z-w|-1{)}^p}dt\end{array}$$

$$\begin{array}{cc}& \le \frac{\parallel f{\parallel }_{{\beta }^p}}{1-p}\left\{\right(1-\left|w\right|-t|z-w|{)}^{(1-p)}{|}_{\xi }^0+\left(\right|w|+t|z-w\left|{)}^{(1-p)}{|}_{\xi }^1\right\}\end{array}$$

$$\begin{array}{cc}& =\frac{\parallel f{\parallel }_{{\beta }^p}}{1-p}\left\{\right(1-\left|w\right|{)}^{1-p}+\left(\right|w\left|\right|z-w|-1{)}^{1-p}\}\end{array}$$

$$\begin{array}{cc}& \le \frac{2}{1-p}\parallel f{\parallel }_{{\beta }^p}|z-w{|}^{1-p}.\end{array}$$

Therefore for all $z,w\in B,$  $$\left|f\right(z)-f(w\left)\right|\le \frac{2\parallel f{\parallel }_{{\beta }^p}}{1-p}|z-w{|}^{1-p}.$$  The second result of this lemma follows quickly by lemma 2.1. $♯$ 

3 The proof of theorem 1.

We'll give the sufficiency first. If ( 5 ) and ( 6 ) hold, by lemma 2.1,( 5 ) and ( 6 ), For every $f\in F(p,q,s)$  , a direct calculation gives

$$\begin{array}{c}\begin{array}{cc}(1-|\omega |2{)}^{\alpha }& |\nabla W\psi ,\phi f(\omega \left)\right|\\ & =(1-|\omega \left|2{)}^{\alpha }\right|\psi \left(\omega \right)\nabla (f\circ \phi )\left(\omega \right)+f\left(\phi \right(\omega \left)\right)\nabla \psi \left(\omega \right)|\\ & \le (1-|\omega \left|2{)}^{\alpha }\right|\psi \left(\omega \right)\nabla (f\circ \phi )\left(\omega \right)|+(1-\left|\omega \right|2{)}^{\alpha }\left|f\right(\phi \left(\omega \right))\nabla \psi (\omega \left)\right|\\ & \le c{sup}_{u\in {\text{C}}^n-\left\{0\right\}}\frac{(1-|\omega \left|2{)}^{\alpha }\right|\phi \left(\omega \right)\left|\right|\nabla f\left(\right(1-\left|\omega \right|2\left)\right)J\phi \left(\omega \right)u|}{\sqrt{(1-|\omega \left|2\right)\left|u\right|2+|<\omega ,u>|2}}\\ & +c(1-|\omega \left|2\right)\alpha G_{\frac{n+1+q}{p}}\left(\phi \right(\omega \left)\right)\parallel f{\parallel }_{F(p,q,s)}|\nabla \psi (\omega \left)\right|\\ & =c{sup}_{u\in {\text{C}}^n-\left\{0\right\}}\frac{(1-|\omega \left|2\right)\alpha \left|\psi \right(\omega \left)\right|}{(1-|\phi \left(\omega \right)\left|2\right)\frac{n+1+q}{p}}\frac{\sqrt{(1-|\phi \left(w\right)\left|2\right)\left|J\phi \right(w\left)u\right|2+|<\phi (w),J\phi (w)u>|2}}{\sqrt{(1-|\omega \left|2\right)\left|u\right|2+|<\omega ,u>|2}}\times \\ & \frac{(1-|\phi \left(\omega \right)\left|2\right)\frac{n+1+q}{p}\parallel \nabla f\left(\phi \right(\omega \left)\right)\cdot J\phi \left(\omega \right)u|}{\sqrt{(1-|\phi \left(w\right)\left|2\right)\left|J\phi \right(w\left)u\right|2+|<\phi (w),J\phi (w)u>|2}}+\\ & c(1-|\omega \left|2\right)\alpha G_{\frac{n+1+q}{p}}\left(\phi \right(\omega \left)\right)\parallel f{\parallel }_{F(p,q,s)}|\nabla \psi (\omega \left)\right|\\ & \le c{sup}_{u\in {\text{C}}^n-\left\{0\right\}}\left|\psi \right(\omega \left)\right|\frac{(1-|w\left|2\right)\alpha }{(1-|\phi \left(w\right)\left|2\right)\frac{q+n+1}{p}}\left\{\frac{(1-|\phi \left(w\right)\left|2\right)\left|J\phi \right(w\left)u\right|2+|<\phi (w),J\phi (w)u>|2}{(1-|w\left|2\right)\left|u\right|2+|<w,u>|2}\right\}\frac{1}{2}\times \\ & \parallel f{\parallel }_{{\beta }^{\frac{n+1+q}{p}}}+c(1-|\omega \left|2\right)\alpha G_{\frac{n+1+q}{p}}\left(\phi \right(\omega \left)\right)\parallel f{\parallel }_{F(p,q,s)}|\nabla \psi (\omega \left)\right|\\ & \le c{sup}_{u\in {\text{C}}^n-\left\{0\right\}}\left|\psi \right(\omega \left)\right|\frac{(1-|w\left|2\right)\alpha }{(1-|\phi \left(w\right)\left|2\right)\frac{q+n+1}{p}}\left\{\frac{(1-|\phi \left(w\right)\left|2\right)\left|J\phi \right(w\left)u\right|2+|<\phi (w),J\phi (w)u>|2}{(1-|w\left|2\right)\left|u\right|2+|<w,u>|2}\right\}\frac{1}{2}\times \\ & \parallel f{\parallel }_{F(p,q,s)}+c(1-|\omega \left|2\right)\alpha G_{\frac{n+1+q}{p}}\left(\phi \right(\omega \left)\right)|\nabla \psi (\omega \left)\right|\parallel f{\parallel }_{F(p,q,s)}\\ & \le c\parallel f{\parallel }_{F(p,q,s)}\end{array}\end{array}$$

(35)

notice that $$|W\psi ,\phi f(0\left)\right|=\left|\psi \right(0\left)\right|\left|f\right(\phi \left(0\right)\left)\right|\le \left|f\right(\phi \left(0\right)\left)\right|\le G_{\frac{n+1+q}{p}}\left(\phi \right(0\left)\right)\parallel f{\parallel }_{F(p,q,s)},$$  therefore, we have

$$\begin{array}{c}\parallel W\psi ,\phi f{\parallel }_{\beta \alpha }\le c\parallel f{\parallel }_{F(p,q,s)}.\end{array}$$

(36)

Since the constant $c$  is independent of $f$  , $W\psi ,\phi :F(p,q,s)\to \beta \alpha$  is bounded.

One the other hand, suppose $W\psi ,\phi$  is bounded, with $$\parallel W\psi ,\phi f{\parallel }_{\beta \alpha }\le c\parallel f{\parallel }_{F(p,q,s)}.$$  for all $f\in F(p,q,s).$  It's very easy to show that the functions $f_l\left(z\right)=z_l,l=1,\cdots ,n$  and $f\left(z\right)=1$  are in $F(p,q,s)$  . Therefore, $W\psi ,\phi f_l$  and $W\psi ,\phi f$  must in $\beta \alpha$  too, in an other word, $\psi {\phi }_l,l=1,\cdots ,n$  and $\psi$  lies in $\beta \alpha$  .

1) If $\left|\phi \right(\omega ){|}^2\le 2/3$  , then

$$\begin{array}{c}\begin{array}{cc}& \left|\psi \right(\omega \left)\right|\frac{(1-|w\left|2\right)\alpha }{(1-|\phi \left(w\right)\left|2\right)\frac{q+n+1}{p}}\left\{\frac{(1-|\phi \left(w\right)\left|2\right)\left|J\phi \right(w\left)u\right|2+|<\phi (w),J\phi (w)u>|2}{(1-|w\left|2\right)\left|u\right|2+|<w,u>|2}\right\}\frac{1}{2}\\ & \le c\left|\psi \right(\omega \left)\right|(1-|\omega \left|2\right)\alpha \frac{\left|J\phi \right(\omega \left)u\right|}{\sqrt{(1-|\omega \left|2\right)\left|u\right|2+|<\omega ,u>|2}}\\ & \le c\left|\psi \right(\omega \left)\right|(1-|\omega \left|2\right)\alpha \left\{{\sum }_{l=1}^n\frac{|\nabla {\phi }_l(\omega \left)u\right|}{\sqrt{(1-|\omega \left|2\right)\left|u\right|2+|<\omega ,u>|2}}\right\}\\ & \le c(1-|\omega \left|2\right)\alpha \left\{{\sum }_{l=1}^n\right|\psi \left(\omega \right)\nabla {\phi }_l\left(\omega \right)\left|\right\}\\ & =c(1-|\omega \left|2\right)\alpha \left\{{\sum }_{l=1}^n\right[|\nabla (\psi \cdot {\phi }_l\left)\right(\omega )-{\phi }_l(\omega )\cdot \nabla \psi (\omega \left)\right|\left]\right\}\\ & \le c(1-|\omega \left|2\right)\alpha \left\{{\sum }_{l=1}^n\right[|\nabla (\psi \cdot {\phi }_l\left)\right(\omega \left)\right|+\left|{\phi }_l\right(\omega )\cdot \nabla \psi (\omega \left)\right|\left]\right\}\\ & \le c{\sum }_{l=1}^n\parallel \psi {\phi }_l{\parallel }_{\beta \alpha }+\parallel \psi {\parallel }_{\beta \alpha }\\ & \le c.\end{array}\end{array}$$

(37)

where the last inequality follows by the fact that $\psi \cdot {\phi }_l,(l=1,\cdots ,n)$  and $\psi \left(z\right)$  lies in $\beta \alpha$  .

If $\left|\phi \right(\omega ){|}^2>2/3$  , then by lemma 2.6, for every $\omega \in B$  satisfies $\left|\phi \right(\omega ){|}^2>2/3$  and every $u\in {\text{C}}^n-\left\{0\right\},$  exists a function $f\omega ,u\in F(p,q,s)$  such that $\parallel f\omega ,u\parallel F(p,q,s)<c$  , and

$$\begin{array}{cc}& \parallel W\psi ,\phi f\omega ,u{\parallel }_{\beta \alpha }\end{array}$$

$$\begin{array}{cc}& \ge c\left|\psi \right(\omega \left)\right|\frac{(1-|w\left|2\right)\alpha }{(1-|\phi \left(w\right)\left|2\right)\frac{q+n+1}{p}}\left\{\frac{(1-|\phi \left(w\right)\left|2\right)\left|J\phi \right(w\left)u\right|2+|<\phi (w),J\phi (w)u>|2}{(1-|w\left|2\right)\left|u\right|2+|<w,u>|2}\right\}\frac{1}{2}\end{array}$$

but $\parallel W\psi ,\phi f\omega ,u{\parallel }_{\beta \alpha }\le \parallel W\psi ,\phi \parallel \parallel f\omega ,u{\parallel }_{F(p,q,s)},$  we get

$$\begin{array}{c}\left|\psi \right(\omega \left)\right|\frac{(1-|w\left|2\right)\alpha }{(1-|\phi \left(w\right)\left|2\right)\frac{q+n+1}{p}}\left\{\frac{(1-|\phi \left(w\right)\left|2\right)\left|J\phi \right(w\left)u\right|2+|<\phi (w),J\phi (w)u>|2}{(1-|w\left|2\right)\left|u\right|2+|<w,u>|2}\right\}\frac{1}{2}\le c.\end{array}$$

(38)

By ( 37 ) and ( 38 ), we know that ( 5 ) holds.

2) If $0<\frac{n+1+q}{p}<1$  , it's obviously that ( 6 ) holds, since $\psi$  lies in $\beta \alpha$  . By the previous discussion, we can suppose that $\phi \left(\omega \right)=r\omega e_1$  .

If $\frac{n+1+q}{p}>1,$  take

$$\begin{array}{c}f\omega \left(z\right)=\frac{1-r2\omega }{(1-r\omega z_1{)}^{\frac{n+1+q}{p}}}\end{array}$$

then $$\frac{\partial f\omega \left(z\right)}{\partial z_1}=\frac{n+1+q}{p}\frac{r\omega (1-r2\omega )}{(1-r\omega z_1{)}^{\frac{n+1+q}{p}+1}}$$  and $$\frac{\partial f\omega \left(z\right)}{\partial z_k}=0,k=2,\cdots ,n.$$  so, by lemma 2.5, $f\omega \in F(p,q,s)$  , and $$\parallel f\omega {\parallel }_{F(p,q,s)}\le c,$$  for every $\omega$  in $B$  . Therefore

$$\begin{array}{c}\begin{array}{cc}(1-|\omega \left|2\right)\alpha & |\nabla \psi (\omega \left)\right|G_{\frac{n+1+q}{p}}\left(\phi \right(\omega \left)\right)\\ & =\frac{(1-|\omega \left|2\right)\alpha |\nabla \psi (\omega \left)\right|}{(1-r2\omega {)}^{\frac{n+1+q}{p}-1}}\\ & =(1-|\omega \left|2\right)\alpha |\nabla \psi (\omega \left)\right|\left|f\omega \right(\phi \left(\omega \right)\left)\right|\\ & \le (1-|\omega \left|2\right)\alpha \left\{\right|\nabla (\psi \cdot f\omega \circ \phi )\left(\omega \right)|+|\psi \left(\omega \right)\nabla (f\omega \circ \phi )\left(\omega \right)\left|\right\}\\ & \le \parallel W\psi ,\phi f\omega {\parallel }_{\beta \alpha }+(1-|\omega \left|2\right)\alpha \left|\psi \right(\omega )\nabla (f\omega \circ \phi \left)\right(\omega \left)\right|\\ & \le c\parallel f\omega {\parallel }_{F(p,q,s)}+c{sup}_{u\in {\text{C}}^n\backslash \left\{0\right\}}\frac{(1-|\omega \left|2\right)\alpha \left|\psi \right(\omega \left)\right||\nabla f\omega (\phi \left(\omega \right))\cdot J\phi (\omega \left)u\right|}{\sqrt{(1-|\omega \left|2\right)\left|u\right|2+|<\omega ,u>|2}}\\ & =c\parallel f\omega {\parallel }_{F(p,q,s)}+\\ & c\frac{n+1+q}{p}{sup}_{u\in {\text{C}}^n\backslash \left\{0\right\}}\left|\psi \right(\omega \left)\right|\frac{(1-|\omega \left|2\right)\alpha }{(1-r2\omega {)}^{\frac{n+1+q}{p}}}\frac{\sqrt{|<\phi (\omega ),J\phi (\omega )u>{|}^2}}{\sqrt{(1-|\omega \left|2\right)\left|u\right|2+|<\omega ,u>|2}}\\ & \le c\parallel f\omega {\parallel }_{F(p,q,s)}\\ & +csup\stackrel{u\in Cn-\left\{0\right\}}{\omega \in Bn}\left|\psi \right(\omega \left)\right|\frac{(1-|w\left|2\right)\alpha }{(1-|\phi \left(w\right)\left|2\right)\frac{q+n+1}{p}}\left\{\frac{(1-|\phi \left(w\right)\left|2\right)\left|J\phi \right(w\left)u\right|2+|<\phi (w),J\phi (w)u>|2}{(1-|w\left|2\right)\left|u\right|2+|<w,u>|2}\right\}\frac{1}{2}\\ & \le c.\end{array}\end{array}$$

(39)

If $\frac{n+1+q}{p}=1$  , for every $\omega \in B$  and $k\in \text{N}$  , take $$f\omega ,k\left(z\right)=log\frac{1-r^k\omega z_1}{1-r\omega z_1},$$  then $$\frac{\partial f\omega ,k\left(z\right)}{\partial z_1}=\frac{r\omega -r^k\omega }{(1-r\omega z_1)(1-r^k\omega z_1)}$$  and $$\frac{\partial f\omega ,k\left(z\right)}{\partial z_l}=0,l=2,\cdots ,n.$$  Therefore $$|\nabla f\omega ,k(z\left)\right|=\frac{r\omega -r^k\omega }{|1-r\omega z_1||1-r^k\omega z_1|}\le \frac{2(1-(r^2\omega {)}^k)}{(1-|z{|}^2\left)\right|1-r^k\omega z|}.$$  so, by lemma2.5, $f\omega \in F(p,q,s)$  , and there is a constant c, s.t.

$$\parallel f\omega ,k{\parallel }_{F(p,q,s)}\le c,$$  for every $\omega \in B$  and $k\in \text{N}$  . By the same calculation as ( 39 )

$$\begin{array}{cc}(1-|\omega \left|2\right)\alpha |\nabla & \psi \left(\omega \right)\left|\right|f\omega ,k\left(\phi \right(\omega \left)\right)|\end{array}$$

$$\begin{array}{cc}& =(1-|\omega \left|2\right)\alpha |\nabla \psi (\omega \left)\right|log\frac{1-r^{k+1}\omega }{1-r^2\omega }\end{array}$$

$$\begin{array}{cc}& \le \parallel W\psi ,\phi f\omega ,k{\parallel }_{\beta \alpha }+(1-|\omega \left|2\right)\alpha \left|\psi \right(\omega )\nabla (f\omega ,k\circ \phi \left)\right(\omega \left)\right|\end{array}$$

$$\begin{array}{cc}& \le c\parallel f\omega ,k{\parallel }_{F(p,q,s)}+c{sup}_{u\in {\text{C}}^n-\left\{0\right\}}\frac{(1-|\omega \left|2\right)\alpha \left|\psi \right(\omega \left)\right||\nabla f\omega ,k(\phi \left(\omega \right))\cdot J\phi (\omega \left)u\right|}{\sqrt{(1-|\omega \left|2\right)\left|u\right|2+|<\omega ,u>|2}}\end{array}$$

$$\begin{array}{cc}& =c\parallel f\omega ,k{\parallel }_{F(p,q,s)}+\end{array}$$

$$\begin{array}{cc}& c\frac{n+1+q}{p}{sup}_{u\in {\text{C}}^n-\left\{0\right\}}\left|\psi \right(\omega \left)\right|\frac{(1-|\omega \left|2\right)\alpha (1-r^{k-1}\omega )}{(1-r2\omega )(1-r^{k+1}\omega )}\frac{\sqrt{|<\phi (\omega ),J\phi (\omega )u>{|}^2}}{\sqrt{(1-|\omega \left|2\right)\left|u\right|2+|<\omega ,u>|2}}\end{array}$$

$$\begin{array}{cc}& \le c\parallel f\omega ,k{\parallel }_{F(p,q,s)}+\end{array}$$

$$\begin{array}{cc}& c{sup}_{u\in {\text{C}}^n-\left\{0\right\}}\left|\psi \right(\omega \left)\right|\frac{(1-|\omega \left|2\right)\alpha }{1-r2\omega }\frac{\sqrt{|<\phi (\omega ),J\phi (\omega )u>{|}^2}}{\sqrt{(1-|\omega \left|2\right)\left|u\right|2+|<\omega ,u>|2}}\end{array}$$

$$\begin{array}{cc}& \le c\parallel f\omega ,k{\parallel }_{F(p,q,s)}+\end{array}$$

$$\begin{array}{cc}& c{sup}_{u\in {\text{C}}^n-\left\{0\right\}}\left|\psi \right(\omega \left)\right|\frac{(1-|\omega \left|2\right)\alpha }{1-r2\omega }\frac{\sqrt{(1-|\phi \left(w\right)\left|2\right)\left|J\phi \right(w\left)u\right|2+|<\phi (w),J\phi (w)u>|2}}{\sqrt{(1-|\omega \left|2\right)\left|u\right|2+|<\omega ,u>|2}}\end{array}$$

$$\begin{array}{cc}& \le c\parallel f\omega ,k{\parallel }_{F(p,q,s)}\end{array}$$

$$\begin{array}{cc}& \le c.\end{array}$$

take $k\to \infty ,$  we get $$(1-|\omega \left|2\right)\alpha |\nabla \psi (\omega \left)\right|log\frac{1}{1-r2\omega }<\infty$$  notice that $\psi \in \beta \alpha$  , so we have $$(1-|\omega \left|2\right)\alpha |\nabla \psi (\omega \left)\right|log\frac{2}{1-r2\omega }<\infty$$  or $$(1-|\omega \left|2\right)\alpha |\nabla \psi (\omega \left)\right|G_1\left(\phi \right(\omega \left)\right)<\infty$$  Now the proof of Theorem 1 is completed.

4 the Proof of Theorem 2

If $W\psi ,\phi$  is bounded and ( 8 ) holds. $f_j$  is a sequence in $F(p,q,s)$  which converges to zero uniformly on compact subsets of $B$  and $\parallel f_j{\parallel }_{F(p,q,s)}\le 1$  for every $j\in \text{N}$  . $\forall \varepsilon >0,$  suppose $\delta ={\varepsilon }^{\frac{1}{1-\frac{n+1+q}{p}}}$  , there is a $k_0\in \text{N},$  for $k>k_0,$  and $\forall w\in \overline{B_{1-\delta }}=\{z:|z|\le 1-\delta \}$  have $$\left|f_k\right(w\left)\right|<\varepsilon .$$  One the other hand, by lemma 2.7, if $|z-w|<2\delta$  and $1-\left|w\right|<\delta$ 

$$\begin{array}{c}\left|f_k\right(z)-f_k(w\left)\right|\le c\frac{\parallel f_k{\parallel }_{F(p,q,s)}}{1-p}|z-w{|}^{1-\frac{n+1+q}{p}}\le c\varepsilon \end{array}$$

(40)

holds for every $k\in \text{N}$  . For every $w\in B\backslash \overline{B_{1-\delta }}$  or in a other word, $1-\delta <\left|w\right|<1$  , we can always find a point $z_w\in B_{1-\delta }$  s.t. $|z-w|<2\delta$  , then if $k>k_0$ 

$$\begin{array}{cc}\left|f_k\right(w\left)\right|& \le \left|f_k\right(w)-f_k(z_w\left)\right|+\left|f_k\right(z_w\left)\right|\le c\varepsilon +\varepsilon <c\varepsilon .\end{array}$$

Then, $\forall \varepsilon >0$  , there is a $k_0>0,$  s.t. for all $z\in B$  and $k>k_0$  , $\left|f_k\right(z\left)\right|<\varepsilon$  holds. So, the sequence $\left\{f_j\right\}$  converges to zero uniformly on $B$  .

Since ( 8 ) holds, therefore, there is a $\delta >0$  , such that if $1-\left|\phi \right(\omega \left)\right|>\delta$ 

$$\begin{array}{c}\begin{array}{c}{sup}_{u\in {\text{C}}^n\backslash \left\{0\right\}}\left|\psi \right(\omega \left)\right|\frac{(1-|w\left|2\right)\alpha }{(1-|\phi \left(w\right)\left|2\right)\frac{q+n+1}{p}}\left\{\frac{(1-|\phi \left(w\right)\left|2\right)\left|J\phi \right(w\left)u\right|2+|<\phi (w),J\phi (w)u>|2}{(1-|w\left|2\right)\left|u\right|2+|<w,u>|2}\right\}\frac{1}{2}<\varepsilon .\end{array}\end{array}$$

(41)

So, when $1-\left|\phi \right(\omega \left)\right|>\delta$  ,

$$\begin{array}{c}\begin{array}{cc}& (1-|\omega \left|2\right)\alpha |\nabla W\psi ,\phi f_k(\omega \left)\right|\\ & \le (1-|\omega \left|2\right)\alpha |\nabla \psi (\omega \left)f_k\right(\phi \left(\omega \right)\left)\right|+(1-|\omega \left|2\right)\alpha \left|\psi \right(\omega )\nabla (f_k\circ \phi \left)\right(\omega \left)\right|\\ & \le \parallel \psi {\parallel }_{\beta \alpha }\left|f_k\right(\phi \left(\omega \right)\left)\right|+c{sup}_{u\in {\text{C}}^{-}\left\{0\right\}}(1-|\omega \left|2\right)\alpha \left|\psi \right(\omega \left)\right|\frac{|\nabla f_k(\phi \left(\omega \right))\cdot J\phi (\omega \left)u\right|}{\sqrt{(1-|\omega \left|2\right)\left|u\right|2+|<\omega ,u>|2}}\\ & \le c\varepsilon +c{sup}_{u\in {\text{C}}^n-\left\{0\right\}}\frac{(1-|\omega \left|2\right)\alpha \left|\psi \right(\omega \left)\right|}{(1-|\phi \left(\omega \right)\left|2\right)\frac{n+1+q}{p}}\frac{\sqrt{(1-|\phi \left(w\right)\left|2\right)\left|J\phi \right(w\left)u\right|2+|<\phi (w),J\phi (w)u>|2}}{\sqrt{(1-|\omega \left|2\right)\left|u\right|2+|<\omega ,u>|2}}\times \\ & \frac{(1-|\phi \left(\omega \right)\left|2\right)\frac{n+1+q}{p}\parallel \nabla f\left(\phi \right(\omega \left)\right)\cdot J\phi \left(\omega \right)u|}{\sqrt{(1-|\phi \left(w\right)\left|2\right)\left|J\phi \right(w\left)u\right|2+|<\phi (w),J\phi (w)u>|2}}\\ & \le c\varepsilon +c{sup}_{u\in {\text{C}}^n-\left\{0\right\}}\left|\psi \right(\omega \left)\right|\frac{(1-|w\left|2\right)\alpha }{(1-|\phi \left(w\right)\left|2\right)\frac{q+n+1}{p}}\left\{\frac{(1-|\phi \left(w\right)\left|2\right)\left|J\phi \right(w\left)u\right|2+|<\phi (w),J\phi (w)u>|2}{(1-|w\left|2\right)\left|u\right|2+|<w,u>|2}\right\}\frac{1}{2}\\ & \times \parallel f{\parallel }_{F(p,q,s)}\\ & \le c\varepsilon ,\end{array}\end{array}$$

(42)

When $\left|\phi \right(\omega \left)\right|\le \delta ,$  let $E=\{\omega :|\omega \le \delta \left|\right\}$  , then E is compact and $\{\omega \in B:\phi (\omega )\le \delta \}\subset E$  . Therefore

$$\begin{array}{c}\begin{array}{cc}& (1-|\omega \left|2\right)\alpha |\nabla W\psi ,\phi f_k(\omega \left)\right|\\ & \le (1-|\omega \left|2\right)\alpha |\nabla \psi (\omega \left)f_k\right(\phi \left(\omega \right)\left)\right|+(1-|\omega \left|2\right)\alpha \left|\psi \right(\omega )\nabla (f_k\circ \phi \left)\right(\omega \left)\right|\\ & \le (1-|\omega \left|2\right)\alpha |\nabla \psi (\omega \left)f_k\right(\phi \left(\omega \right)\left)\right|+(1-|\omega \left|2\right)\alpha \left|\psi \right(\omega )\nabla f_k(\phi \left(\omega \right)\left)\right|\left|J_{\phi }\right(\omega \left)\right|\\ & \le \psi {\parallel }_{\beta \alpha }\left|f_k\right(\phi \left(\omega \right)\left)\right|+{\sum }_{l=1}^n(\parallel \psi {\phi }_l{\parallel }_{\beta \alpha }+\parallel \psi {\parallel }_{\beta \alpha })|\nabla f_k(\phi \left(\omega \right)\left)\right|\\ & <c\varepsilon .\end{array}\end{array}$$

(43)

where the last inequality follows by the fact that the sequence $\{\nabla f_k\}$  converges to zero uniformity on $E$  .

Therefore $(1-|\omega \left|2\right)\alpha |\nabla W\psi ,\phi f_k(\omega \left)\right|$  converges to zero uniformly on $B$  .

Notice that $\left\{f_k\right(0\left)\right\}$  converges to zero, and so we have $$\parallel W\psi ,\phi f_k{\parallel }_{\beta \alpha }\to 0.$$  as $\left|\phi \right(\omega \left)\right|\to 0$  On the other hand, if $W\psi ,\phi$  is compact, then clearly that $W\psi ,\phi$  is compact and by lemma 2.5 and lemma 2.3, it is easy to know that ( 8 ) holds. And so we are done.

5 The proof of theorem 3

Just as the proof of theorem 2, if $W\psi ,\phi$  is bounded and ( 9 )( 10 ) hold, $f_j$  is a sequence in $F(p,q,s)$  which converges to zero uniformly on compact subsets of $B$  and $\parallel f_j{\parallel }_{F(p,q,s)}\le 1$  for every $j\in \text{N}$  , then by the same discussion as lemma 2.6, we have

$$\begin{array}{c}\begin{array}{cc}& (1-|\omega \left|2{)}^{\alpha }\right|\nabla W\psi ,\phi f_j\left(\omega \right)|\\ & =(1-|\omega \left|2{)}^{\alpha }\right|\psi \left(\omega \right)\nabla (f_j\circ \phi )\left(\omega \right)+f_j\left(\phi \right(\omega \left)\right)\nabla \psi \left(\omega \right)|\\ & \le c{sup}_{u\in {\text{C}}^n-\left\{0\right\}}\left|\psi \right(\omega \left)\right|\frac{(1-|w\left|2\right)\alpha }{(1-|\phi \left(w\right)\left|2\right)\frac{q+n+1}{p}}\left\{\frac{(1-|\phi \left(w\right)\left|2\right)\left|J\phi \right(w\left)u\right|2+|<\phi (w),J\phi (w)u>|2}{(1-|w\left|2\right)\left|u\right|2+|<w,u>|2}\right\}\frac{1}{2}\parallel f_j{\parallel }_{F(p,q,s)}+\\ & c(1-|\omega \left|2\right)\alpha G_{\frac{n+1+q}{p}}\left(\phi \right(\omega \left)\right)|\nabla \psi (\omega \left)\right|\parallel f_j{\parallel }_{F(p,q,s)}\\ & \le c{sup}_{u\in {\text{C}}^n-\left\{0\right\}}\left|\psi \right(\omega \left)\right|\frac{(1-|w\left|2\right)\alpha }{(1-|\phi \left(w\right)\left|2\right)\frac{q+n+1}{p}}\left\{\frac{(1-|\phi \left(w\right)\left|2\right)\left|J\phi \right(w\left)u\right|2+|<\phi (w),J\phi (w)u>|2}{(1-|w\left|2\right)\left|u\right|2+|<w,u>|2}\right\}\frac{1}{2}+\\ & c(1-|\omega \left|2\right)\alpha \nabla \psi \left(\omega \right)\left|G_{\frac{n+1+q}{p}}\right(\phi \left(\omega \right)\left)\right|\\ \end{array}\end{array}$$

(44)

Therefore, $\forall \varepsilon >0,$  there is a $\delta >0,$  for $\forall j\in \text{N}$  and $1-\left|\phi \right(\omega \left)\right|<\delta$  ,

$$\begin{array}{c}(1-|\omega \left|2{)}^{\alpha }\right|\nabla W\psi ,\phi f_j\left(\omega \right)|\le c\varepsilon .\end{array}$$

(45)

follows.

And when $1-\left|\phi \right(\omega \left)\right|\le \delta$  , by the same discussion as the proof of Theorem 2, we know that there is a $k_0\in \text{N},$  s.t. when $j>k_0$  have

$$\begin{array}{c}(1-|\omega \left|2{)}^{\alpha }\right|\nabla W\psi ,\phi f_j\left(\omega \right)|\le c\varepsilon \end{array}$$

(46)

Combine ( 45 ) and ( 46 ), and notice that $\left\{f_k\right(0\left)\right\}$  converges to zero, we have $$\parallel W\psi ,\phi f_k{\parallel }_{\beta \alpha }\to 0,$$  as $\left|\phi \right(\omega \left)\right|\to 1$  , therefore by lemma 3, $W\psi ,\phi$  is compact.

If $W\psi ,\phi$  is compact, then $W\psi ,\phi$  is bounded, and by lemma 2.6 and lemma 2.3, it's clearly that ( 9 ) hold.

If $\frac{n+1+q}{p}=1$  and $\left|r\omega \right|>\frac{2}{3}$  , let

$$\begin{array}{c}f\omega ,k={\left(log\frac{1}{1-r2\omega }\right)}^{-1}{\left(log\frac{1-r^k\omega z_1}{1-r\omega z_1}\right)}^2,\end{array}$$

(47)

then $$\frac{\partial f\omega ,k\left(z\right)}{\partial z_1}=2{\left(log\frac{1}{1-r2\omega }\right)}^{-1}\left(log\frac{1-r^k\omega z_1}{1-r\omega z_1}\right)\frac{r\omega -r^k\omega }{(1-r\omega z_1)(1-r^k\omega z_1)}$$  and $$\frac{\partial f\omega ,k\left(z\right)}{\partial z_l}=0,l=2,\cdots ,n.$$  therefore

$$\begin{array}{cc}|\nabla f\omega ,k(z\left)\right|& =2\left|{\left(log\frac{1}{1-r2\omega }\right)}^{-1}\left(log\frac{1-r^k\omega z_1}{1-r\omega z_1}\right)\frac{r\omega -r^k\omega }{(1-r\omega z_1)(1-r^k\omega z_1)}\right|\end{array}$$

$$\begin{array}{cc}& \le 2{\left(log\frac{1}{1-r2\omega }\right)}^{-1}\left(log\frac{|1-r^k\omega z_1|}{|1-r\omega z_1|}+2\pi \right)\frac{r\omega -r^k\omega }{|1-r\omega z_1||1-r^k\omega z_1|}\end{array}$$

$$\begin{array}{cc}& \le 2{\left(log\frac{1}{1-r2\omega }\right)}^{-1}\left(log\frac{1}{|1-r\omega {|}^2}+2\pi \right)\frac{1-(r^2\omega {)}^k}{(1-|z{|}^2\left)\right|1-r^k\omega z_1|}\end{array}$$

$$\begin{array}{cc}& \le c\frac{1-(r^2\omega {)}^k}{(1-|z{|}^2\left)\right|1-r^k\omega z_1|}\end{array}$$

by lemma 2.5, $f\omega ,k\in F(p,q,s),$  and $\exists C>0$  , such that $\parallel f\omega ,k{\parallel }_{F(p,q,s)}<C$  . And it is clearly that $f\omega ,k$  converges to zero uniformly on compact subsets of $B$  and independent of $k\in \text{N}$  , by the same discussion as the proof of theorem 1 and use lemma 1, we have

$$\begin{array}{c}\begin{array}{cc}(1-|\omega \left|2\right)\alpha |\nabla & \psi \left(\omega \right)\left|\right|f\omega ,k\left(\phi \right(\omega \left)\right)|\\ & =(1-|\omega \left|2\right)\alpha |\nabla \psi (\omega \left)\right|log\frac{1-r^{k+1}\omega }{1-r^2\omega }\\ & \le c\parallel W\psi ,\phi f\omega ,k{\parallel }_{\beta \alpha }\\ & +c{sup}_{u\in {\text{C}}^{-}\left\{0\right\}}\left|\psi \right(\omega \left)\right|\frac{(1-|\omega \left|2\right)\alpha }{1-r2\omega }\frac{\sqrt{(1-|\phi \left(w\right)\left|2\right)\left|J\phi \right(w\left)u\right|2+|<\phi (w),J\phi (w)u>|2}}{\sqrt{(1-|\omega \left|2\right)\left|u\right|2+|<\omega ,u>|2}}\\ \end{array}\end{array}$$

(48)

but by lemma 2.3 and ( 9 ), we get $${lim}_{k\to \infty }(1-|\omega \left|2\right)\alpha |\nabla \psi (\omega \left)\right|log\frac{1-r^{k+1}\omega }{1-r^2\omega }=(1-|\omega \left|2\right)\alpha |\nabla \psi (\omega \left)\right|log\frac{1}{1-r^2\omega }\to 0,$$  as $\left|\phi \right(\omega \left)\right|\to 1.$  Notice that ( 6 ) holds, so $(1-|\omega \left|2\right)\alpha |\nabla \psi (\omega \left)\right|\to 0$  ( $\left|\phi \right(\omega \left)\right|\to 1.$  ) And then $(1-|\omega \left|2\right)\alpha |\nabla \psi (\omega \left)\right|log\frac{2}{1-r^2\omega }\to 0$  ( $\left|\phi \right(\omega \left)\right|\to 1$  .) If $\frac{n+1+q}{p}>1$  , just let $$f\omega =\frac{1-r^2\omega }{1-r\omega z_1}$$  and use the same method as the situation of $\frac{n+1+q}{p}<1$  , we can also prove that ( 10 ) holds, and we omit the details here. So, the proof of theorem 3 is completed.

References

1. R.Timoney, Bloch Function in several Complex Variables, I, Bull.London.Math.Soc. 12(1980), 241-267.
2. R.Timoney, Bloch Function in Several Complex Variables, II, J.Reine Angew.Math., 319(1980):1-22.
3. C.C.Cowen and B.D.MacCluer, Composition operators on spaces of analytic functions, CRC Press, Boca Raton, FL, 1995.
4. K.Madigan and A.Matheson, Compact composition operators on the Bloch space, Trans.Amer.Math.Sor. 347(1995):2679-2687.
5. K.Madigan, Compact composition operators on analytic Lipschitz space, Proc.Amer.Math.Soc. 119(1993)L465-473.
6. M.D.Contreras and A.G.Hernandez-Diaz, Weighted composition operators in weighted Banach spaces of analytic functions, J.Austral.Math.Soc(Series A), 69(2000):41-46.
7. Alfonso Montes-Rodriguez, Weighted composition operators on weighted Banach spaces of analytic functions, J.London Math.Soc.(2)61(2000):872-884.
8. J.H.Shi and L.Luo, Composition operators on the Bloch space of several complex variables, Acta Math.Sinica, English Series, 16(2000):85-98.
9. Z.H.Zhou and J.H.Shi, Compact composition operators on the Bloch space in polydisk, Science in China (Series A), 44(2001):286-291.
10. Z.H.Zhou, Composition operators on the Lipschitz spaces in polydisk. Science in China (Series A), 46(2003):33-38.
11. Z.H.Zhou and J.H.Shi, Compactness of composition operators on the Bloch space in classical bounded symmetric domains, The Michigan Mathematical Journal, 50(2002): 381-405.
12. Z.H.Zhou, Weighted Composition Operators between different Bloch-type spaces in polydisk, to appear.
13. J.P.D'Angelo, Several complex variables and geometry of real hypersurfaces, CRC Press, 1993.
14. Jiang L J, He Y Z. Composition operators form $\beta \alpha$  to $F(p,q,s)$  . Acta Math Scientia, 2003, 23B(2):252-260.
15. Yang W S, Ouyang C H. Exact location of $\alpha -Bloch$  spaces in $L_a^p$  and $H^p$  of a complex unit ball. Rocky Mountain J Math, 2000, 30:1151-1169.
16. Rudin W. Function theory in the unit ball of ${\text{C}}^n$  , Springer-Verlag. New York. 1980.
17. Zhuo W X, Ouyang C H. $M\ddot{o}bius$  invariant gradient and little $\alpha -Bloch$  functions, Acta Math Scientia, 2002, 22B(3):295-301.

^$1$  Supported in part by the National Natural Science Foundation of China (Grand No.10371091), and LiuHui Center for Applied Mathematics, Nankai University & Tianjin University