Acta Mathematica Scientia 2018,38B(6):1861–1880

http://actams.wipm.ac.cn

AN INTEGRAL ESTIMATE AND THE EQUIVALENT

NORMS ON F (p, q, s, k) SPACES IN THE UNIT BALL∗

Xuejun ZHANG ( ÜÆ
) †

Shenlian LI ( iê)
Qingli SHANG ( ÿ˜w) Yuting GUO ( H…x)
College of Mathematics and Statistics, Hunan Normal University, Changsha 410006, China
E-mail : [email protected]; [email protected];
[email protected]; [email protected]

Abstract In this article, the authors give a typical integral’s bidirectional estimates for all
cases. At the same time, several equivalent characterizations on the F (p, q, s, k) space in the
unit ball of Cn are given.
Key words F (p, q, s, k) space; integral estimate; equivalent norm; unit ball
2010 MR Subject Classification 32A37

1 Introduction
Let B denote the unit ball in Cn . Suppose D denotes the unit disc of C. The class of all
holomorphic functions on B is denoted by H(B). Let dv be the normalized volume measure on
B such that v(B)=1, and dσ be the normalized measure on ∂B such that σ(∂B)=1.
Definition 1.1 Let p > 0 and α > −1. The space Apα (B) = Lpα (B) ∩ H(B), where
( Z  p1 )
p p
Lα (B) = f : f is measurable on B and ||f ||p,α = |f (z)| dvα (z) <∞ ,
B

dvα (z) = cα (1 − |z| ) dv(z), the constant cα such that vα (B) = 1. When α = 0, Apα (B) is just
2 α

the Bergman space Ap (B).

Let ϕa be the involution automorphism of B, with ϕa (0) = a and ϕa (a) = 0.
Definition 1.2 Let p > 0, s ≥ 0, q + s > −1, q + n > −1, k ≥ 0. f is said to belong to
the logarithm weight general function space F (p, q, s, k) if f ∈ H(B) and
Z
e
p
I = sup |∇f (z)|p (1 − |z|2 )q g s (z, a) logk dv(z) < ∞,
a∈B B 1 − |z|2
where complex gradient ∇ and weight function g are defined by
 
∂f ∂f 1
∇f (z) = (z), · · · , (z) and g(z, a) = log , respectively.
∂z1 ∂zn |ϕa (z)|
∗ Received May 17, 2017; revised April 9, 2018. The research was supported by the National Natural

Science Foundation of China (11571104) and the Hunan Provincial Innovation Foundation for Postgraduate
(CX2017B220); Key Laboratory of High Performance Computing and Stochastic Information Processing, Col-
lege of Mathematics and Computer Science, Hunan Normal University, Changsha, Hunan 410081, P.R. China;
Supported by the Construct Program of the Key Discipline in Hunan Province.
† Corresponding author: Xuejun ZHANG.
1862 ACTA MATHEMATICA SCIENTIA Vol.38 Ser.B

We use the notation ||f ||F (p,q,s,k) = |f (0)| + I. The space F (p, q, s, k) is a Banach space
with the norm || ||F (p,q,s,k) when p ≥ 1. If 0 < p < 1, the space F (p, q, s, k) is a complete metric
space with the following distance:

ρ(f, g) = ||f − g||pF (p,q,s,k) .

In this article, z = (z1 , · · · , zn ), w = (w1 , · · · , wn ), hz, wi = z1 w1 + · · · + zn wn . Radial

derivative Rf (z) = h∇f (z), zi for f ∈ H(B). We use the symbols c, c1 , · · · to denote positive
constants independent of variables z, w, u, a. But they may depend on some parameters with
different values in different cases. We call E and F are equivalent (denoted by E ≈ F in the
following ) if there exist two positive constants A1 and A2 such that A1 E ≤ F ≤ A2 E.
Suppose r ≥ 0, t ≥ 0, δ > −1, k ≥ 0. In this article, we consider the integral
Z
(1 − |z|2 )δ e
Jw,a = t |1 − hz, ai|r
logk dv(z) (w, a ∈ B).
B |1 − hz, wi| 1 − |z|2
This integral has a very wide range of applications. Therefore, many mathematicians
were very interested in its exact and complete estimates. On December 2015, we discussed
the integral estimation problem with Professors Zhu, Zhao, Xiao and so on at the Tsinghua
Sanya international mathematics forum. In 2016, we communicated some results at the several
complex variables annual conference of China and the fifth function space conference of China,
respectively. For “ ≤ ” part, Ortega and Fabrega gave some estimates in [1]. In 2016, Li, Zhang
and Xu considered the equivalent problem in [2] when k = 0. In this article, we give and prove
the integral’s bidirectional estimates for all cases.
In this article, as an application of the integral estimates, we consider the equivalent norms
on the F (p, q, s, k) space. If k = 0, then the space F (p, q, s, k) is just the general function space
F (p, q, s). The space F (p, q, s) was first introduced on the unit disk D by Zhao in [3]. Later,
it was studied on D or B by many mathematicians. A lot of function spaces associating with
F (p, q, s) were discussed (for example, [2–20, 23, 26]). It is known that the space F (p, q, s)
contains a lot of classic function spaces (for example, Bergman spaces, Besov spaces, Dirichlet
type spaces, H 2 space, BMOA space, Bloch type spaces, Lipschitz type spaces, Qs spaces). We
are interested in studying the space F (p, q, s, k) since it is an extension of F (p, q, s). However,
we always encounter some obstacles because of the varieties of parameters p, q, s and k. In
order to overcome these obstacles, we need a variety of means or techniques. The equivalent
characterization is one of them. If n > 1, then there are several kinds of derivatives to choose.
In addition, there are several weight functions that we may use. Therefore, we may use several
ways to define the norm of F (p, q, s, k). Are these norms equivalent? In this article, another
goal is to prove that these norms on the F (p, q, s, k) space are equivalent.

2 Some Lemmas
In order to prove the main results, we first give some lemmas.
Lemma 2.1 Let c be real and δ > −1. Then the integrals
Z Z
1 (1 − |w|2 )δ
I(z) = n+c
dσ(ξ) and J(z) = n+1+δ+c
dv(w) (z ∈ B)
∂B |1 − hz, ξi| B |1 − hz, wi|

have the following asymptotic properties.

No.6 X.J. Zhang et al: AN INTEGRAL ESTIMATE AND THE EQUIVALENT NORMS 1863

(1) If c < 0, then I(z) ≈ J(z) ≈ 1.

e
(2) If c = 0, then I(z) ≈ J(z) ≈ log 1−|z| 2.
2 −c
(3) If c > 0, then I(z) ≈ J(z) ≈ (1 − |z| ) .
Proof This is Proposition 1.4.10 in [21]. 
Lemma 2.2 Let k be real, δ > −1 and c ≥ 0. Then the integral
Z 1
(1 − r)δ e
Iρ = δ+1+c
logk dr (0 ≤ ρ < 1)
0 (1 − rρ) 1 − r
has the following asymptotic properties.
(1) If c = 0 and k > −1, then Iρ ≈ logk+1 e
1−ρ .
1 k e
(2) If c > 0, then Iρ ≈ (1−ρ) c log 1−ρ .
Proof This is Lemma 2.2 in [26]. 
Lemma 2.3 Let p > 0, s ≥ 0, q + s > −1, q + n > −1, k ≥ 0. If f ∈ H(B) and
Z
e
p
I = sup |Rf (z)|p (1 − |z|2 )q (1 − |ϕa (z)|2 )s logk dv(z) < ∞,
a∈B B 1 − |z|2
then there exists a constant c > 0 such that
  kp
2
q+n+1 e
sup (1 − |z| ) p log |Rf (z)| ≤ cI.
z∈B 1 − |z|2
Proof For z ∈ B and r > 0, we let D(z, r) denote the Bergman metric ball of radius of
r at z. By Lemma 2.24 in [22], (2.3) in [27], we have
Z
c
|Rf (z)|p ≤ |Rf (w)|p dv(w)
(1 − |z|2 )n+1 D(z,1)
 −k  2 2s
c e e +1
≤ log ·
(1 − |z|2 )q+n+1 1 − |z|2 2e
Z
e
× |Rf (w)|p (1 − |z|2 )q (1 − |ϕz (w)|2 )s logk dv(w)
D(z,1) 1 − |z|2
 −k  2 2s
ce2|q| e e +1
≤ log ·
(1 − |z|2 )q+n+1 1 − |z|2 2e
Z
e3
× |Rf (w)|p (1 − |w|2 )q (1 − |ϕz (w)|2 )s logk dv(w)
D(z,1) 1 − |w|2
 −k  2 2s
ce2|q| 3k e e +1
≤ log · I p.
(1 − |z|2 )q+n+1 1 − |z|2 2e
This proof is completed. 
Lemma 2.4 Suppose α > −1, β is a real constant and g ∈ L1α (B). If
Z
g(w)
f (z) = β
dvα (w) (z ∈ B),
B (1 − hz, wi)
then Z
√ 1 |g(w)|
e (z)| ≤
|∇f 2 |β| (1 − |z|2 ) 2 1 dvα (w),
B |1 − hz, wi|β+ 2
e (z) = ∇(f ◦ ϕz )(0) is complex invariant gradient of f .
where ∇f
Proof It comes from Lemma 3.3 in [22]. We replace g(w) by g(w)cα (1 − |w|2 )α . 
1864 ACTA MATHEMATICA SCIENTIA Vol.38 Ser.B

3 Main Results
First, we prove following integral estimates.
Theorem 3.1 Suppose r ≥ 0, t ≥ 0, δ > −1, k ≥ 0. Let
Z
(1 − |z|2 )δ e
Jw,a = t r
logk dv(z) (w, a ∈ B).
B |1 − hz, wi| |1 − hz, ai| 1 − |z|2
Then following results hold.
(1) When t + r − δ < n + 1, Jw,a ≈ 1.
(2) When t + r − δ = n + 1, t > 0 and r > 0,
e
Jw,a ≈ logk+1 .
|1 − hw, ai|
(3) When t + r − δ > n + 1, t − δ < n + 1 and r − δ < n + 1,
1 e
Jw,a ≈ logk .
|1 − hw, ai|t+r−δ−n−1 |1 − hw, ai|
(4) When t − δ = n + 1 > r − δ and r > 0,
1 e e
Jw,a ≈ r
logk 2
log .
|1 − hw, ai| 1 − |w| |1 − hw, ϕw (a)i|
(5) When t − δ = n + 1 = r − δ,
1 e e
Jw,a ≈ logk log
|1 − hw, ai|δ+n+1 1 − |w|2 |1 − hw, ϕw (a)i|
1 e e
+ δ+n+1
logk 2
log .
|1 − hw, ai| 1 − |a| |1 − ha, ϕa (w)i|
(6) When t − δ > n + 1 > r − δ,
1 e
Jw,a ≈ logk .
(1 − |w|2 )t−δ−n−1 |1 − hw, ai|r 1 − |w|2
(7) When t − δ > n + 1 = r − δ,
1 e
Jw,a ≈ logk
(1 − |w|2 )t−δ−n−1 |1− hw, ai|δ+n+1 1 − |w|2
1 k e e
+ log log .
|1 − hw, ai|t 1 − |a|2 |1 − ha, ϕa (w)i|
(8) When t − δ > n + 1 and r − δ > n + 1,
1 e
Jw,a ≈ logk
(1 − |w|2 )t−δ−n−1 |1− hw, ai|r 1 − |w|2
1 e
+ logk .
(1 − |a|2 )r−δ−n−1 |1 − hw, ai|t 1 − |a|2
(9) When t − δ = n + 1 and r = 0, Jw,a ≈ logk+1 e
1−|w|2 .
Proof If there exists a constant 0 < r0 < 1 such that |w| ≤ r0 or |a| ≤ r0 , then these
bidirectional estimates are obvious. Without loss of generality, we let |w| → 1− and |a| → 1− .
Otherwise, we use the result: (p + q)σ ≈ pσ + q σ (σ > 0) for all p ≥ 0 and q ≥ 0.
If l > 0 and σ > 0, then it is clear that
(  l )
σ l e σ−l l σ l e
sup x log = max e , 2 log . (3.1)
0<x<2 x σ 2
No.6 X.J. Zhang et al: AN INTEGRAL ESTIMATE AND THE EQUIVALENT NORMS 1865

If t + r − δ < n + 1, then we may choose 0 < σ < min{δ + 1, δ + 1 + n − t − r} when k > 0

or σ = 0 when k = 0. By (3.1) and Lemma 2.1, we have
Z Z 
(1 − |z|2 )δ−σ (1 − |z|2 )δ−σ
2−t−r c−1
δ ≤ Jw,a ≤ c t+r
dv(z) + t+r
dv(z) ≤ c1 .
B |1 − hz, wi| B |1 − hz, ai|

This shows that result (1) is true.

By Lemma 1.8 and Theorem 1.12 in [22], if t + r − δ = n + 1 and r = 0, then
Z 1
(1 − ρ)δ e
Jw,a ≈ ρn−1 2 )δ+1
logk dρ.
0 (1 − ρ|w| 1 − ρ
It is clear that
Z 1 Z 1
(1 − ρ)δ e (1 − ρ)δ e
ρn−1 logk dρ ≈ logk dρ.
0 (1 − ρ|w|2 )δ+1 1−ρ 0 (1 − ρ|w|2 )δ+1 1−ρ
In fact, we know that
Z 1
(1 − ρ)δ e
ρn−1 2 δ+1
logk dρ
0 (1 − ρ|w| ) 1−ρ
 Z 1
1 T1 (1 − ρ)δ e
≥ min n−1
, 2 δ+1
logk dρ,
2 T0 0 (1 − ρ|w| ) 1−ρ
where
Z 1 Z 1
2 1 e 2 e
T0 = logk dρ, T1 = ρn−1 (1 − ρ)δ logk dρ.
0 1−ρ 1−ρ 0 1−ρ
Therefore, Lemma 2.2 shows that result (9) is true.
In the following, we give the proof for cases (2)–(8).
We first prove that the left-hand side is “ ≥ ” the right-hand side in each case.
Cases (6) and (8). Let z = ϕw (u). By Lemma 1.3 and Lemma 1.7 in [22], we have
Z
(1 − |w|2 )δ+n+1−t (1 − |u|2 )δ |1 − hu, ϕw (a)i|−r e
Jw,a = logk dv(u). (3.2)
|1 − hw, ai|r B |1 − hw, ui|2δ+2n+2−t−r 1 − |ϕw (u)|
2

If |u| ≤ 1/2, then

(1 − |u|2 )δ |1 − hu, ϕw (a)i|−r e e
logk ≈ logk (|w| → 1− ).
|1 − hw, ui|2δ+2n+2−t−r 1 − |ϕw (u)|2 1 − |w|2
This means that
c e
Jw,a ≥ logk . (3.3)
(1 − |w|2 )t−δ−n−1 |1 − hw, ai|r 1 − |w|2
This shows that “ ≥ ” part of (6) is true.
If we make a change of variables z = ϕa (u), then we may get
c e
Jw,a ≥ logk . (3.4)
(1 − |a|2 )r−δ−n−1 |1 − hw, ai|t 1 − |a|2
(3.3) and (3.4) show that “ ≥ ” part of (8) is true.
Cases (2)–(3). In order to estimate Jw,a when t + r − δ ≥ n + 1, we first estimate
Z
dσ(ξ)
t |1 − hξ, ai|r
,
∂B |1 − hw, ξi|
when t + r > n.
1866 ACTA MATHEMATICA SCIENTIA Vol.38 Ser.B

We consider the function

Z
dσ(ξ)
f (w, a) = (w, a ∈ B).
∂B (1 − hw, ξi)t (1 − hξ, ai)r
By a change of variables ξ = ϕw (η), Lemmas 1.3 and (4.7) in [22], we have
Z
1 dσ(η)
f (w, w) = .
(1 − |w|2 )t+r−n ∂B (1 − hw, ηi)n−t (1 − hη, wi)n−r
According to the uniqueness theorem in several complex variables (see p.326 in [24] or
p.1022 in [25]), we have
g(w, a)
f (w, a) = ,
(1 − hw, ai)t+r−n
where Z
dσ(η)
g(w, a) =
∂B (1 − hw, ηi)n−t (1
− hη, ai)n−r
is holomorphic about w and a respectively.
By (1.21) and Lemma 1.11 in [22], if t < n and r < n, then we have
∞
X (n − 1)!Γ(k + n − t)Γ(k + n − r)hw, aik
g(w, a) = .
Γ(n − t)Γ(n − r)Γ(k + n)Γ(k + 1)
k=0

Further, by the Stirling’s formula, if t + r > n and max{t, r} < n, then we have
∞
X ∞
X
Γ(k + n − t)Γ(k + n − r) 1
≈ ≈ 1.
Γ(k + n)Γ(k + 1) k t+r−n+1
k=0 k=1

Therefore, |g(w, a)| ≈ 1 when b = |1 − hw, ai| → 0+ in above case. This means that there exist
constants c0 > 0 and 0 < r0 < 1 such that
Z
dσ(ξ) c0
t |1 − hξ, ai|r
≥ |f (w, a)| ≥ t+r−n ,
∂B |1 − hξ, wi| b
when b ≤ r0 , t + r > n and max{t, r} < n.
If b ≥ r0 , then we have
Z
dσ(ξ) 1 2−t−r r0t+r−n
t r
≥ t+r ≥ when t + r > n.
∂B |1 − hξ, wi| |1 − hξ, ai| 2 bt+r−n
This shows that Z
dσ(ξ) c
t r
≥ , (3.5)
∂B |1 − hξ, wi| |1 − hξ, ai| |1 − hw, ai|t+r−n
when t + r > n and max{t, r} < n.
If t ≥ n, by a change of variables ξ = ϕw (η), Lemma 1.3 and (4.7) in [22], (1.22) in [22],
then we can obtain
Z Z
dσ(ξ) (1 − |w|2 )n−t |1 − hϕw (a), ηi|−r
t r
= dσ(η)
∂B |1 − hξ, wi| |1 − hξ, ai| |1 − hw, ai|r ∂B |1 − hw, ηi|2n−t−r
( ∞
)
2−r (1 − |w|2 )n−t X (n − 1)! α2 · · · (α − l + 1)2 2l
≥ 1+ |w|
|1 − hw, ai|r l! (n − 1 + l)!
l=1
2−r
≥ , (3.6)
(1 − |w| ) |1 − hw, ai|r
2 t−n

t+r
where α = 2 − n.
No.6 X.J. Zhang et al: AN INTEGRAL ESTIMATE AND THE EQUIVALENT NORMS 1867

(3.6) implies that

Z
dσ(ξ) 2n−t−r
≥ , (3.7)
∂B |1 − hξ, wi|t |1 − hξ, ai|r |1 − hw, ai|t+r−n
when t ≥ n.
Similarly, we may obtain
Z
dσ(ξ) 2n−t−r
t r
≥ , (3.8)
∂B |1 − hξ, wi| |1 − hξ, ai| |1 − hw, ai|t+r−n
when r ≥ n.
Therefore, by (3.5) and (3.7)–(3.8), if t + r > n, then we always have
Z
dσ(ξ) c
t r
≥ .
∂B |1 − hξ, wi| |1 − hξ, ai| |1 − hw, ai|t+r−n
This means that
Z
dσ(ξ) c
≥ (3.9)
∂B |1 − hξ, ρwi|t |1 − hξ, ρai|r |1 − ρ2 hw, ai|t+r−n
for all 0 ≤ ρ < 1 and w, a ∈ B.
If t + r > n, by Lemma 1.8 in [22], (3.9), then we may obtain
Z 1 Z 
e dσ(ξ)
Jw,a = 2n ρ2n−1 (1 − ρ2 )δ logk dρ
0 1 − ρ2 t
∂B |1 − hρξ, wi| |1 − hρξ, ai|
r
Z 1
c(1 − ρ)δ e
≥ t+r−n
logk dρ
0 |1 − ρhw, ai| 1 − ρ
Z 1
c(1 − ρ)δ e
≥ t+r−n
logk dρ
0 {b + |hw, ai|(1 − ρ)} 1−ρ
Z 1
c(b + |hw, ai|)n−t−r (1 − ρ)δ e
= −1 |hw, ai|ρ}t+r−n
logk dρ.
0 {1 − (b + |hw, ai|) 1−ρ
By Lemma 2.2 and 1 ≤ b + |hw, ai| < 3, if t + r = δ + n + 1(> n), then we obtain
e
Jw,a ≥ c1 logk+1 . (3.10)
|1 − hw, ai|
(3.10) shows that “ ≥ ” part of (2) is true.
If t + r > δ + n + 1(> n), then we have
c1 e
Jw,a ≥ logk . (3.11)
|1 − hw, ai|t+r−δ−n−1 |1 − hw, ai|
(3.11) shows that “ ≥ ” part of (3) is true.
Cases (4)–(5) and (7). It is clear that
2(1 − |w|2 ) 2(1 + |w|) 4 8
2
≤ < <
b + 2(1 − |w| ) 3 + 2|w| 5 9
and
ak+1
1 − ak+1
2 ≥ (a1 − a2 )k+1 (a1 > a2 > 0).

In the following integral, we make changes of variables

|w|2 (1 − ρ) 1 − |w|2
x= and y= .
1 − ρ|w|2 b + 1 − |w|2 − bx
1868 ACTA MATHEMATICA SCIENTIA Vol.38 Ser.B

√
If b < 1/8 and |w| > 1/ 2, then we have
Z 1
(1 − ρ)δ e
2 δ+1 |1 − ρhw, ai|r
logk dρ
0 (1 − ρ|w| ) 1 − ρ
Z |w|2  −r
1 xδ  hw, ai(1 − |w|2 )x  2
k e|w| (1 − x)
= 1 − hw, ai + log dx
|w|2(δ+1) 0 1−x  |w|2 (1 − x)  (1 − |w|2 )x
Z |w|2  −r
1 1 − |w|2 e(1 − x)
≥c b+ logk dx
1
2
1 − x 1 − x 1 − |w|2
Z b+11
c (1 − y)r−1 e(1 − y)
= r logk dy
b 2(1−|w|2 ) y by
b+2(1−|w|2 )
Z 8
c1 9 1 e
≥ r logk dy
b 2(1−|w|2 )
2
y 9by
b+2(1−|w| )

c2 e 4b + 8(1 − |w|2 ) c2 k+1 4b + 8(1 − |w| )

2
≥ r logk log + log
b 9b 9(1 − |w|2 ) br 9(1 − |w|2 )
2
c3 4b + 8(1 − |w| ) 4e
≥ r log logk .
b 9(1 − |w|2 ) 81(1 − |w|2 )
When 2b ≤ 9(1 − |w|2 ), we take c0 = log 9e
2
(10/9), then
 c0
4b + 8(1 − |w|2 ) 10 9e eb
log ≥ log = log ≥ c0 log .
9(1 − |w|2 ) 9 2 1 − |w|2
When 2b > 9(1 − |w|2 ), we take c0 = log2 9e − 1, then
  c0 
4b eb 2c0 +1
2 2
> = 1.
9(1 − |w| ) 1 − |w| 9e
This means that
4b + 8(1 − |w|2 ) eb 4b eb
c0 log 2
− log 2
> c0 log 2
− log > 0.
9(1 − |w| ) 1 − |w| 9(1 − |w| ) 1 − |w|2
This shows that
Z 1
(1 − ρ)δ e
2 δ+1 r
logk dρ
0 (1 − ρ|w| ) |1 − ρhw, ai| 1−ρ
c4 eb e
≥ r log logk (|w| → 1− ). (3.12)
b 1 − |w|2 1 − |w|2
If b ≥ 1/8, then we have
Z 1
(1 − ρ)δ e
2 δ+1 r
logk dρ
0 (1 − ρ|w| ) |1 − ρhw, ai| 1−ρ
Z b+1
1 Z 1
c (1 − y)r−1 k e(1 − y) c1 b+1 1 e
≥ r log dy ≥ r logk dy
b 2(1−|w|2 ) y by b 2(1−|w|2 ) y (b + 1)y
b+2(1−|w|2 ) b+2(1−|w|2 )
 2

c1 k+1 e(b + 2 − 2|w| ) 1 e
= r
log 2
− 1 ≈ r logk+1 (|w| → 1− )
(k + 1)b 2(b + 1)(1 − |w| ) b 1 − |w|2
1 e eb
≈ r logk log (|w| → 1− ). (3.13)
b 1 − |w|2 1 − |w|2
By Lemma 1.8 in [22], (3.6) and (3.12)–(3.13), if t − δ = n + 1, then we have
Z 1 Z 
(1 − ρ2 )δ k e
Jw,a = 2n ρ2n−1 t r
log dσ(ξ) dρ
0 ∂B |1 − hρξ, wi| |1 − hρξ, ai| 1 − ρ2
No.6 X.J. Zhang et al: AN INTEGRAL ESTIMATE AND THE EQUIVALENT NORMS 1869
Z 1
(1 − ρ)δ e
≥c 2 )δ+1 |1 − ρhw, ai|r
logk dρ
0 (1 − ρ|w| 1 − ρ
c1 e e
≥ r
logk 2
log . (3.14)
|1 − hw, ai| 1 − |w| |1 − hw, ϕw (a)i|
Similarly, if r − δ = n + 1, then we have
c e e
Jw,a ≥ logk log . (3.15)
|1 − hw, ai|t 1 − |a|2 |1 − ha, ϕa (w)i|
(3.4) and (3.14) show that “ ≥ ” part of (4) is true. (3.14) and (3.15) show that “ ≥ ” part
of (5) is true. (3.3) and (3.15) show that “ ≥ ” part of (7) is true.
Next, we prove the other directions of the estimates, that is the “ ≤ ” part.
By a partition of B, there exist Ω1 , Ω2 , Ω3 , Ω4 such that B = Ω1 ∪ Ω2 ∪ Ω3 ∪ Ω4 , where Ωj
are mutually disjoint. By Lemma 3.3 in [1], we have

|1 − hz, wi| ≤ c|1 − hw, ai| ≤ c1 |1 − hz, ai| (z ∈ Ω1 );

|1 − hz, ai| ≤ c|1 − hw, ai| ≤ c1 |1 − hz, wi| (z ∈ Ω2 );
|1 − hw, ai| ≤ c|1 − hz, wi| ≤ c1 |1 − hz, ai| (z ∈ Ω3 );
|1 − hw, ai| ≤ c|1 − hz, ai| ≤ c1 |1 − hz, wi| (z ∈ Ω4 ).

If z ∈ Ω1 ∪ Ω3 , then there exists constant c > 0 such that |1 − hz, ai| ≥ c|1 − hw, ai|. By
Lemma 1.8 in [22] and Lemmas 2.1–2.2, if t − δ > n + 1, then we have
Z
(1 − |z|2 )δ e
t |1 − hz, ai|r
logk dv(z)
Ω1 ∪Ω3 |1 − hz, wi| 1 − |z|2
Z
c (1 − |z|2 )δ e
≤ r t
logk dv(z)
|1 − hw, ai| B |1 − hz, wi| 1 − |z|2
Z 1
c1 (1 − ρ)δ e
≤ r t−n
logk dρ
|1 − hw, ai| 0 (1 − ρ|w|) 1−ρ
c2 e
≤ logk . (3.16)
|1 − hw, ai|r (1 − |w|2 )t−δ−n−1 1 − |w|2
If z ∈ Ω2 ∪ Ω4 , then there exists constant c > 0 such that |1 − hz, wi| ≥ c(b + |1 − hz, ai|) ≥
c|b + 1 − hz, ai|. By Lemma 1.8 and (1.13) in [22], the monotone increasing of integral mean
value of analytic function, if n > 1, then we have
Z
(1 − |z|2 )δ e
t |1 − hz, ai|r
logk dv(z)
Ω2 ∪Ω4 |1 − hz, wi| 1 − |z|2
Z
(1 − |z|2 )δ e
≤c t |1 − hz, ai|r
logk dv(z)
B |b + 1 − hz, ai| 1 − |z|2
Z 1 Z 
2n−1 2 δ k e (1 − |η|2 )n−2 dA(η)
≤ 2n(n − 1)c ρ (1 − ρ ) log dρ
0 1 − ρ2 t
D |b + 1 − ρη| |1 − ρη|
r
Z 1 ( Z )
δ k e (ρ − |u − 1|)n−2
≤ c1 (1 − ρ) log dA(u) dρ
0 1−ρ |u−1|<ρ |b + u|t |u|r
Z 1 (Z )
1 Z π 2 {R − (1 − ρ)}n−2
δ k e
≤ c2 (1 − ρ) log t dRdθ dρ
0 1−ρ 1−ρ − π 2
(b2 + R2 ) 2 Rr−1
Z 1 (Z )
1+ρ Z π
δ k e 2 ρn−2
+c2 (1 − ρ) log t dRdθ dρ
0 1−ρ 1 −π 2
(b2 + R2 ) 2 Rr−1
1870 ACTA MATHEMATICA SCIENTIA Vol.38 Ser.B

Z 1 Z 1 
δ k e Rn−1−r
≤ c3 (1 − ρ) log dR dρ + c3
0 1−ρ 1−ρ (b + R)t
Z 1 δ Z 1b
(1 − ρ) k e xn−1−r
≤ c4 r+t−n
log dρ dx + c3
1−ρ 1−ρ (1 + x)t
0 b b
Z 1b n−1−r Z 1
c4 x e
= r+t−n t
dx (1 − ρ)δ logk dρ + c3
b 0 (1 + x) 1−bx 1 − ρ
Z 1b n+δ−r
c5 x e
≤ r+t−δ−n−1 t
logk dx + c3
b 0 (1 + x) bx
Z 1
(1 − y)n+δ−r e
≤ c6 −1 y}t
logk dy + c3 .
0 {1 − (b + 1) 1 − y
By Lemma 2.2, if t + r − δ > n + 1 > r − δ, then we have
Z
(1 − |z|2 )δ e c7 e
t r
logk 2
dv(z) ≤ r+t−δ−n−1 logk . (3.17)
Ω2 ∪Ω4 |1 − hz, wi| |1 − hz, ai| 1 − |z| b b
If t + r − δ = n + 1 and t > 0, then we have
Z
(1 − |z|2 )δ e e
t r
logk 2
dv(z) ≤ c7 logk+1 . (3.18)
Ω2 ∪Ω4 |1 − hz, wi| |1 − hz, ai| 1 − |z| b
The proof of case n = 1 is similar to the proof of Lemma 1 in [23].
Similarly, if t + r − δ > n + 1 > t − δ, then we may obtain
Z
(1 − |z|2 )δ e c e
t |1 − hz, ai|r
logk 2
dv(z) ≤ r+t−δ−n−1 logk . (3.19)
Ω1 ∪Ω3 |1 − hz, wi| 1 − |z| b b
If t + r − δ = n + 1 and r > 0, then we have
Z
(1 − |z|2 )δ e e
t r
logk 2
dv(z) ≤ c logk+1 . (3.20)
Ω1 ∪Ω3 |1 − hz, wi| |1 − hz, ai| 1 − |z| b
(3.18) and (3.20) show that “ ≤ ” part of (2) is true. (3.17) and (3.19) show that “ ≤ ”
part of (3) is true. (3.16)–(3.17) imply that “ ≤ ” part of (6) is true.
By result (2), if r − δ < n + 1 and r > 0, then we have
Z
(1 − |u|2 )δ e
δ+n+1−r r
logk dv(u)
B |1 − hw, ui| |1 − hu, ϕw (a)i| 1 − |ϕw (u)|2
Z
(1 − |u|2 )δ e
≤c δ+n+1−r
dv(u) logk
B |1 − hw, ui| |1 − hu, ϕw (a)i|r 1 − |w|2
Z
(1 − |u|2 )δ e
+c δ+n+1−r r
logk dv(u)
B |1 − hw, ui| |1 − hu, ϕw (a)i| 1 − |u|2
e e e
≤ c1 logk log + c1 logk+1
1 − |w|2 |1 − hw, ϕw (a)i| |1 − hw, ϕw (a)i|
e e
≤ c2 logk log . (3.21)
1 − |w|2 |1 − hw, ϕw (a)i|
(3.2) and (3.21) show that “ ≤ ” part of (4) is true.
If z ∈ Ω1 , then there exists a constant c > 0 such that |1−hz, ai| ≥ c|1−hw, ai|. If z ∈ B\Ω1 ,
then there exists a constant c > 0 such that |1 − hz, wi| ≥ c|1 − hw, ai|. By Lemma 1.8 in [22],
Lemmas 2.1–2.2, if t − n > δ + 1 and r − n > δ + 1, then we have
Z Z
c1 (1 − |z|2 )δ k e c1 (1 − |z|2 )δ e
Jw,a ≤ t r
log 2
dv(z) + t r
logk dv(z)
Ω1 |1 − hz, wi| b 1 − |z| B\Ω1 b |1 − hz, ai| 1 − |z|2
No.6 X.J. Zhang et al: AN INTEGRAL ESTIMATE AND THE EQUIVALENT NORMS 1871
Z 1 Z 1
(1 − ρ)δ k e (1 − ρ)δ e
≤ c2 r (1 − ρ|w|)t−n
log dρ + c 2 t (1 − ρ|a|)r−n
logk dρ
0 b 1 − ρ 0 b 1 − ρ
c3 (1 − |w|2 )δ+n+1−t k e c3 (1 − |a|2 )δ+n+1−r e
≤ r
log 2
+ t
logk .
|1 − hw, ai| 1 − |w| |1 − hw, ai| 1 − |a|2
This shows that “ ≤ ” part of (8) is true.
If t − δ = n + 1 = r − δ, then we take r1 = r − α and t1 = t − α, where 0 < α < δ + n + 1.
This means that t − δ = n + 1 > r1 − δ and r − δ = n + 1 > t1 − δ, t1 > 0 and r1 > 0. By result
(4), we obtain
Z
c1 (1 − |z|2 )δ e
Jw,a ≤ t |1 − hz, ai|r1 bα
logk dv(z)
Ω1 |1 − hz, wi| 1 − |z|2
Z
c1 (1 − |z|2 )δ e
+ α |1 − hz, wi|t1 |1 − hz, ai|r
logk dv(z)
B\Ω1 b 1 − |z|2
Z
c1 (1 − |z|2 )δ e
≤ t r α
logk dv(z)
B |1 − hz, wi| |1 − hz, ai|
1 b 1 − |z|2
Z
c1 (1 − |z|2 )δ e
+ α t1 |1 − hz, ai|r
logk dv(z)
B b |1 − hz, wi| 1 − |z|2
c2 e e
≤ logk log
|1 − hw, ai|δ+n+1 1 − |w|2 |1 − hw, ϕw (a)i|
c2 e e
+ logk log .
|1 − hw, ai|δ+n+1 1 − |a|2 |1 − ha, ϕa (w)i|
This shows that “ ≤ ” part of (5) is true.
If t − δ > n + 1 = r − δ, then we take t1 = t − α, where t − δ − n − 1 < α < t. This means
that r − δ = n + 1 > t1 − δ and t1 > 0. By the results (4) and (6), we have
Z
c1 (1 − |z|2 )δ e
Jw,a ≤ t |1 − hw, ai|δ+n+1
logk dv(z)
Ω1 |1 − hz, wi| 1 − |z|2
Z
c1 (1 − |z|2 )δ e
+ α |1 − hz, wi|t1 |1 − hz, ai|r
logk dv(z)
B\Ω1 |1 − hw, ai| 1 − |z|2
Z
c1 (1 − |z|2 )δ e
≤ t δ+n+1
logk dv(z)
B |1 − hz, wi| |1 − hw, ai| 1 − |z|2
Z
c1 (1 − |z|2 )δ e
+ α t1 |1 − hz, ai|r
logk dv(z)
B |1 − hw, ai| |1 − hz, wi| 1 − |z|2
c2 e
≤ logk
(1 − |w|2 )t−δ−n−1 |1 − hw, ai|δ+n+1 1 − |w|2
c2 e e
+ logk log .
|1 − hw, ai|t 1 − |a|2 |1 − ha, ϕa (w)i|
This shows that “ ≤ ” part of (7) is true.
This proof is completed. 
As an application of Theorem 3.1, we consider the equivalent norms on F (p, q, s, k) spaces.
Theorem 3.2 Let p > 0, s ≥ 0, q + s > −1, q + n > −1, k ≥ 0. The following quantities
are equivalent for all f ∈ F (p, q, s, k).
(1) ||f ||pF (p,q,s,k) .
Z
e
(2) I1 = |f (0)|p + sup |∇f (z)|p (1 − |z|2 )q (1 − |ϕa (z)|2 )s logk dv(z).
a∈B B 1 − |z|2
1872 ACTA MATHEMATICA SCIENTIA Vol.38 Ser.B
Z
e
(3) I2 = |f (0)|p + sup |Rf (z)|p (1 − |z|2 )q (1 − |ϕa (z)|2 )s logk dv(z).
a∈B B 1 − |z|2
Z
e
p
(4) I3 = |f (0)| + sup |Rf (z)|p (1 − |z|2 )q g s (z, a) logk dv(z).
a∈B B 1 − |z|2
Proof First, for any a ∈ B, if a 6= z ∈ B, then we have
 
1 − |ϕa (z)|2 1 1 1 − |ϕa (z)|2 1 − |ϕa (z)|2
< log = log 1 + < . (3.22)
2 |ϕa (z)| 2 |ϕa (z)|2 2|ϕa (z)|2
By |Rf (z)| ≤ |∇f (z)| and (3.22), if f ∈ F (p, q, s, k), then we have

I1 ≤ 2s ||f ||pF (p,q,s,k) and I2 ≤ 2s I3 ≤ 2s ||f ||pF (p,q,s,k) .

Next, we prove ||f ||pF (p,q,s,k) ≤ cI1 .

For any j ∈ {1, 2, · · · , n}, let Dj denote partial derivatives to the j-th variable. Replacing
R by Dj in the proof of Lemma 2.3, we may obtain
 −k
c e
|Dj f (z)|p ≤ log I1 for all z ∈ B.
(1 − |z|2 )n+1+q 1 − |z|2
This means that
 −k
cI1 e
|∇f (z)|p ≤ log for all z ∈ B.
(1 − |z|2 )n+1+q 1 − |z|2
Therefore, by Theorem 2.2.2 in [21], we have
Z
e
|∇f (z)|p (1 − |z|2 )q g s (z, a) logk dv(z)
|ϕa (z)|≤ 21 1 − |z|2
Z
cI1 1
≤ logs dv(z) ≤ c1 I1 . (3.23)
|ϕa (z)|≤ 12 (1 − |z|2 )n+1 |ϕa (z)|
From (3.22), we have
Z
e
|∇f (z)|p (1 − |z|2 )q g s (z, a) logk dv(z)
1
2 <|ϕa (z)|<1
1 − |z|2
Z
e
≤ 2s |∇f (z)|p (1 − |z|2 )q (1 − |ϕa (z)|2 )s logk dv(z) ≤ 2s I1 . (3.24)
1
2 <|ϕa (z)|<1
1 − |z|2
By (3.23)–(3.24) and |f (0)|p ≤ I1 , we have ||f ||pF (p,q,s,k) ≤ cI1 .
Finally, we prove I1 ≤ cI2 .
In the following, we may suppose that α is a sufficiently large positive number for some
calculation requirements.
By Lemma 2.3, if α > (q + n + 1)/p − 1, then Rf ∈ A1α (B). By Theorem 2.2 in [22], then
Z
Rf (w)
Rf (z) = dvα (w) (z ∈ B).
B (1 − hz, wi)n+1+α
R
Since 0 = Rf (0) = B Rf (w)dvα (w), by the Fubini’s theorem, we have
Z 1 Z
Rf (tz) − Rf (0)
f (z) − f (0) = dt = Rf (w) L(z, w)dvα (w),
0 t B

where Z  
1
1 1
L(z, w) = −1 dt.
0 (1 − thz, wi)n+1+α t
No.6 X.J. Zhang et al: AN INTEGRAL ESTIMATE AND THE EQUIVALENT NORMS 1873

This shows that

Z Z 1 
∂f (n + 1 + α)wj
(z) = Rf (w) dt dvα (w).
∂zj B 0 (1 − thz, wi)n+2+α
Therefore, we may obtain
Z
|Rf (w)|
|∇f (z)| ≤ c dvα (w). (3.25)
B |1 − hz, wi|n+α+1
1 1
(i) Case p > 1. Let G(z) = |Rf (z)|, H(z, w) = |1−hz,wi| n+α+1 , h(z) = (1−|z|2 )σ (σ > 0) and
Z
T G(z) = H(z, w) G(w)dvα (w).
B
′
By Lemma 2.1, if α > p σ − 1 = pσ/(p − 1) − 1, then we have
Z Z ′
′ cα (1 − |w|2 )α−p σ
H(z, w) hp (w)dvα (w) = n+α+1
dv(w)
B B |1 − hw, zi|
c ′
≤ = c hp (z).
(1 − |z|2 )p′ σ
By Hölder inequality, we have
Z  p′ Z 
p
p p′ −p p
|T G(z)| ≤ H(z, w)h (w)dvα (w) H(z, w)h (w) G (w)dvα (w)
B B
Z 
p −p p
≤ c h (z) H(z, w)h (w)G (w)dvα (w) . (3.26)
B

If 0 ≤ s < q + n + 1, then we may choose

 
q+n+1−s q+s+1
0 < σ < min , .
p p
Therefore, we have 0 ≤ s < q + n + 1 − pσ. As long as α > q + s − pσ, by (3.25)–(3.26),
the Fubini’s theorem and Theorem 3.1(6), we have
Z
e
|∇f (z)|p (1 − |z|2 )q (1 − |ϕa (z)|2 )s logk dv(z)
B 1 − |z|2
Z
e
≤c |T G(z)|p (1 − |z|2 )q (1 − |ϕa (z)|2 )s logk dv(z)
B 1 − |z|2
Z Z 
≤ c1 hp (z) H(z, w) h−p (w) Gp (w)dvα (w)
B B
e
×(1 − |z| ) (1 − |ϕa (z)|2 )s logk
2 q
dv(z)
1 − |z|2
Z Z (1 − |z|2 )q+s−pσ (1 − |a|2 )s logk e
!
−p p 1−|z|2
= c1 h (w) G (w) dv(z) dvα (w)
B B |1 − hz, wi|n+α+1 |1 − hz, ai|2s
Z (1 − |a|2 )s logk e
1−|w|2
≤ c2 h−p (w) Gp (w) dvα (w) ≤ c2 cα I2 .
B (1 − |w|2 )α−q−s+pσ |1 − hw, ai|2s
If s ≥ q + n + 1, then we may choose
 
q+n+1 q+s+1
0 < σ < min , .
p p
Therefore, we have s > q + n + 1 − pσ. As long as α > max{q + s − pσ, q + n − pσ}, by
(3.25)–(3.26), the Fubini’s theorem and Theorem 3.1(8), Lemma 2.3, Lemma 1.8 in [22] and
1874 ACTA MATHEMATICA SCIENTIA Vol.38 Ser.B

Lemmas 2.1–2.2, we have

Z
e
|∇f (z)|p (1 − |z|2 )q (1 − |ϕa (z)|2 )s logk dv(z)
B 1 − |z|2
Z Z (1 − |z|2 )q+s−pσ (1 − |a|2 )s logk e
!
1−|z|2
≤c h−p (w) Gp (w) dv(z) dvα (w)
B B |1 − hz, wi|n+α+1 |1 − hz, ai|2s
Z (1 − |a|2 )s logk e
1−|w|2
≤ c1 h−p (w) Gp (w) dvα (w)
B (1 − |w|2 )α−q−s+pσ |1 − hw, ai|2s
Z (1 − |a|2 )s logk 1−|a|
e
2
+c1 h−p (w) Gp (w) 2 s−q−n−1+pσ
dvα (w)
B (1 − |a| ) |1 − hw, ai|n+α+1
Z
(1 − |w|2 )α−q−n−1+pσ (1 − |a|2 )q+n+1−pσ
≤ c2 I2 + c2 I2 dv(w)
n+α+1 logk e −k e
B |1 − hw, ai| 1−|w|2 log 1−|a|2
Z 1
(1 − ρ)α−q−n−1+pσ (1 − |a|2 )q+n+1−pσ e
≤ c2 I2 + c3 I2 −k e
log−k dρ
0 (1 − ρ|a|) α+1 log 1−|a|2 1 − ρ
≤ c4 I2 .
(ii) Case 0 < p ≤ 1. We take α such that α′ = p(n + 1 + α) − n − 1 > max{q + s, q + n}.
Let
Rf (w)
G(w) = .
(1 − hw, zi)n+α+1
Then we have G ∈ Apα′ (B). By Lemma 2.15 in [22], we can obtain
Z
||G||p,α′
|G(w)|(1 − |w|2 )α dv(w) ≤ .
B cα′
Therefore, by (3.25), we have
Z p
|Rf (w)|
|∇f (z)|p ≤ c n+α+1
dvα (w) ≤ c||G||pp,α′
B |1 − hz, wi|
Z Z
c|Rf (w)|p c|Rf (w)|p
= pn+pα+p
dvα′ (w) =
n+α′ +1
dvα′ (w). (3.27)
B |1 − hw, zi| B |1 − hw, zi|
By (3.27), the Fubini’s theorem and Theorem 3.1(6), if 0 ≤ s < q + n + 1, then
Z
e
|∇f (z)|p (1 − |z|2 )q (1 − |ϕa (z)|2 )s logk dv(z)
B 1 − |z|2
Z Z |Rf (w)|p logk e
1−|z|2 2 q
≤c n+α ′ +1 dvα′ (w) (1 − |z| ) (1 − |ϕa (z)|2 )s dv(z)
B B |1 − hw, zi|
Z Z (1 − |z|2 )q+s (1 − |a|2 )s logk e !
p 1−|z|2
=c |Rf (w)| dv(z) dvα′ (w)
B B |1 − hw, zi|n+α′ +1 |1 − hz, ai|2s
Z (1 − |a|2 )s logk 1−|w|
e
2
p
≤ c1 |Rf (w)| ′
2 )α −q−s |1 − hw, ai|2s
dvα′ (w) ≤ c1 cα′ I2 .
B (1 − |w|
Similarly, by (3.27), the Fubini’s theorem, Theorem 3.1(8) and Lemma 2.3, Lemma 1.8 in [22],
Lemmas 2.1–2.2, if s > q + n + 1, then
Z
e
|∇f (z)|p (1 − |z|2 )q (1 − |ϕa (z)|2 )s logk dv(z)
B 1 − |z|2
Z (1 − |a|2 )s logk 1−|w|
e
2
≤c |Rf (w)|p 2 α′ −q−s dvα′ (w)
B (1 − |w| ) |1 − hw, ai|2s
No.6 X.J. Zhang et al: AN INTEGRAL ESTIMATE AND THE EQUIVALENT NORMS 1875
Z (1 − |a|2 )s logk e
1−|a|2
+c |Rf (w)|p dvα′ (w)
B (1 − |a|2 )s−q−n−1 |1 − hw, ai|n+α′ +1
Z ′
(1 − |a|2 )q+n+1 (1 − |w|2 )α −q−n−1
≤ c1 I2 + c1 I2 dv(w)
B |1 − hw, ai|n+α′ +1 log−k e
1−|a|2 logk e
1−|w|2
≤ c2 I2 .

If s = q + n + 1, then we take 0 < σ < q + n + 1. By (3.27) and the Fubini theorem,

Theorem 3.1(7), (3.1), Lemma 1.8 in [22] and Lemmas 2.1–2.3, we have
Z
e
|∇f (z)|p (1 − |z|2 )q (1 − |ϕa (z)|2 )s logk dv(z)
B 1 − |z|2
Z  
(1 − |a|2 )q+n+1 k e e|1 − hw, ai|
≤ cI2 + c |Rf (w)|p log log dvα′ (w)
B |1 − hw, ai|n+α′ +1 1 − |a|2 1 − |a|2
Z  
(1 − |a|2 )q+n+1−σ k e
≤ cI2 + c1 |Rf (w)|p log dvα′ (w)
B |1 − hw, ai|n+α′ +1−σ 1 − |a|2
Z ′
c2 I2 (1 − |w|2 )α −q−n−1 (1 − |a|2 )q+n+1−σ e
≤ cI2 + k
logk dv(w) ≤ c3 I2 .
B |1 − hw, ai|
n+α ′ +1−σ
log e(1 − |w| )2 −1 1 − |a|2
The above proofs show that
Z
e
|∇f (z)|p (1 − |z|2 )q (1 − |ϕa (z)|2 )s logk dv(z) ≤ cI2 for all cases.
B 1 − |z|2
On the other hand, |f (0)|p ≤ I2 , this means that I1 ≤ cI2 .
The proof is completed. 
Further, we have following results.
Theorem 3.3 Let n > 1, s ≥ 0, q + s > −1, q + n > −1 and k ≥ 0. If 0 < p < min{2(q +
s + 1), 2(q + n + 1)}, then the following six quantities are equivalent for all f ∈ F (p, q, s, k):

(1) ||f ||pF (p,q,s,k) .

Z
e
p
(2) I1 = |f (0)| + sup |∇f (z)|p (1 − |z|2 )q (1 − |ϕa (z)|2 )s logk dv(z).
a∈B B 1 − |z|2
Z
e
(3) I2 = |f (0)|p + sup |Rf (z)|p (1 − |z|2 )q (1 − |ϕa (z)|2 )s logk dv(z).
a∈B B 1 − |z|2
Z
e
(4) I3 = |f (0)|p + sup |Rf (z)|p (1 − |z|2 )q g s (z, a) logk dv(z).
a∈B B 1 − |z|2
Z
e (z)|p (1 − |z|2 )q−p (1 − |ϕa (z)|2 )s logk e
(5) I4 = |f (0)|p + sup |∇f dv(z).
a∈B B 1 − |z|2
Z
e (z)|p (1 − |z|2 )q−p g s (z, a) logk e
(6) I5 = |f (0)|p + sup |∇f dv(z).
a∈B B 1 − |z|2

Proof In Theorem 3.2, we prove ||f ||pF (p,q,s,k) ≈ I1 ≈ I2 ≈ I3 . By Lemma 2.13 and
e (z)|2 = (1 − |z|2 ){|∇f (z)|2 − |Rf (z)|2 }≥ (1 − |z|2 )2 |Rf (z)|2
Lemma 2.14 in [22], we have |∇f
⇒ I4 ≥ I2 . On the other hand, by (3.22), we have I5 ≥ 2−s I4 .
We need only prove I4 ≤ cI2 and I5 ≤ cI4 .
First, we prove I4 ≤ cI2 . By Lemma 2.3, Theorem 7.1 in [22], if f ∈ F (p, q, s, k) and
1876 ACTA MATHEMATICA SCIENTIA Vol.38 Ser.B

α > −1, then Z

h(w)
f (z) = dvα (w) ,
B (1 − hz, wi)n+γ+α
n o
c Rf (w)
where h(w) = α+γ 2 γ
cα (1 − |w| ) f (w) + n+α+γ , γ = q+n+1
p .
By Lemma 2.4, we have
√ Z
e 2 12 |h(w)|
|∇f (z)| ≤ 2 (n + γ + α)(1 − |z| ) n+γ+α+ 21
dvα (w)
B |1 − hz, wi|
Z
1 |f (w)| + |Rf (w)|
≤ c(1 − |z|2 ) 2 n+β+ 21
dvβ (w) (β = α + γ).
B |1 − hz, wi|

By simple computation, we can obtain

Z 1  Z
 Rf (ρw)  |Rf (u)|
|f (w) − f (0)| =  dρ ≤ c n+β
dvβ (u).
0 ρ B |1 − hw, ui|

Therefore, by the Fubini’s theorem and Theorem 3.1(3), we have

Z
|f (w) − f (0)|
n+β+ 12
dvβ (w)
B |1 − hz, wi|
Z Z 
(1 − |w|2 )β dv(w)
≤ c |Rf (u)| n+β+ 12 |1 − hw, ui|n+β
dvβ (u)
B B |1 − hz, wi|
Z Z
|Rf (u)| |Rf (u)|
≤ c1 n+β− 1 dvβ (u) ≤ 2c1
n+β+ 12
dvβ (u)
B |1 − hz, ui| 2 B |1 − hz, ui|
e (z)| = |∇[f
⇒ |∇f e (z) − f (0)]|
Z
1 |Rf (w)|
≤ c(1 − |z|2 ) 2 n+β+ 12 dv (w)
. (3.28)
B |1 − hz, wi| β

If min{2(q + s + 1), 2(q + n + 1)} > 1 and 1 < p < min{2(q + s + 1), 2(q + n + 1)}, then we
may take
 
q+s+1 1 q+n+1−s 1
0 < σ < min − , − when 0 ≤ s < q + n + 1 − p2
p 2 p 2
or we may take
 
q+s+1 1 q+n+1 1
0 < σ < min − , − when s ≥ q + n + 1 − p2 .
p 2 p 2
pσ
As long as β − p′ σ = β − p−1 > −1, by Hölder inequality, we have
Z p
|Rf (w)|
n+β+ 12
dvβ (w)
B |1 − hz, wi|
(Z ′
) p′ Z 
(1 − |w|2 )−p σ dvβ (w) (1 − |w|2 )pσ |Rf (w)|p
p

≤ n+β+1− p
dvβ (w)
B |1 − hz, wi|n+β+1 B |1 − hz, wi| 2

Z
c (1 − |w|2 )pσ |Rf (w)|p
≤ 2 pσ n+β+1− p
dvβ (w). (3.29)
(1 − |z| ) B |1 − hz, wi| 2

When 0 ≤ s < q + n + 1 − p2 , by the choice of σ, we have q + s − p2 − pσ > −1 and 2s −

p
(q + s − − pσ) < n + 1. As long as β + pσ > q + s, by (3.28)–(3.29) and Theorem 3.1(6), then
2
Z
e (z)|p (1 − |z|2 )q−p (1 − |ϕa (z)|2 )s logk e
|∇f dv(z)
B 1 − |z|2
No.6 X.J. Zhang et al: AN INTEGRAL ESTIMATE AND THE EQUIVALENT NORMS 1877
Z (Z p )
(1 − |z|2 )q+s− 2 −pσ (1 − |a|2 )s logk e
1−|z|2 dv(z)
2 pσ p
≤c (1 − |w| ) |Rf (w)| p dvβ (w)
B B |1 − hz, wi|n+β+1− 2 |1 − hz, ai|2s
Z
e
≤ c1 |Rf (w)|p (1 − |w|2 )q (1 − |ϕa (w)|2 )s logk dv(w)
B 1 − |w|2
≤ c1 I2 .

When s ≥ q + n + 1 − p2 , by the choice of σ, we have q + s − p2 − pσ > −1, q + n + 1 − pσ −

p
2 > 0 and 2s − (q + s − p2 − pσ) > n + 1. As long as β > max{q + s − pσ, q + n − pσ}, then
 
q+s+1 1 p
β > q + s − pσ > q + s − p − = − 1.
p 2 2
By (3.28)–(3.29), the Fubini theorem and Theorem 3.1(8), Lemma 2.3, Lemma 1.8 in [22] and
Lemmas 2.1–2.2, we have
Z
e (z)|p (1 − |z|2 )q−p (1 − |ϕa (z)|2 )s logk e
|∇f dv(z)
B 1 − |z|2
Z ( Z (1 − |z|2 )q+s− p2 −pσ (1 − |a|2 )s logk e dv(z) )
1−|z|2
≤ c (1 − |w|2 )pσ |Rf (w)|p n+β+1− p dvβ (w)
B B |1 − hz, wi| 2 |1 − hz, ai|2s
Z
e
≤ c1 |Rf (w)|p (1 − |w|2 )q (1 − |ϕa (w)|2 )s logk dv(w)
B 1 − |w|2
Z (1 − |w|2 )β+pσ−q−n−1 (1 − |a|2 )q+n+1−pσ− p2 logk e
1−|a|2
+c2 I2 n+β+1− p k e
dv(w)
B |1 − hw, ai| 2 log
1−|w|2
≤ c3 I2 .

This shows that I4 ≤ cI2 .

If min{2(q + s + 1), 2(q + n + 1)} > 1 ≥ p > 0, then we take
Rf (w)
G(w) = 1 .
(1 − hw, zi)n+β+ 2
Using (3.27)–(3.28), we may prove
Z
e (z)|p ≤ c(1 − |z|2 ) p2 |Rf (w)|p
|∇f ′ p dvβ ′ (w), (3.30)
B |1 − hz, wi|n+β +1− 2
where β ′ = p(n + β + 1) − n − 1 > q + n.
Next, by considering the cases 0 ≤ s < q + n + 1 − p2 , s > q + n+ 1 − p2 and s = q + n+ 1 − p2
separately, and using (3.30) and the proof method in Theorem 3.2, we may prove I4 ≤ cI2 .
If min{2(q + s + 1), 2(q + n + 1)} ≤ 1, then 0 < p < 1. We prove I4 ≤ cI2 .
In following, we prove I5 ≤ cI4 .
If min{2(q + s + 1), 2(q + n + 1)} > 1 and 1 < p < min{2(q + s + 1), 2(q + n + 1)}, then we
1−xδ
take 0 < σ < q+n+1
p − 12 and 0 < δ < 2n δ
s . Since lim− 1−x2 = 2 , then
x→1
 
s 1 1 s 1 − |ϕa (z)|δ (1 − |ϕa (z)|δ )s c(1 − |ϕa (z)|2 )s
log = s log 1 + ≤ ≤ . (3.31)
|ϕa (z)| δ |ϕa (z)|δ δ s |ϕa (z)|sδ |ϕa (z)|sδ
By (3.31), we have
Z
e (z)|p (1 − |z|2 )q−p g s (z, a) logk e
|∇f dv(z)
1
2 <|ϕa (z)|<1
1 − |z|2
1878 ACTA MATHEMATICA SCIENTIA Vol.38 Ser.B
Z
e (z)|p (1 − |z|2 )q−p (1 − |ϕa (z)|2 )s logk e
≤c |∇f dv(z) ≤ cI4 . (3.32)
B 1 − |z|2
Making a change of variables u = ϕa (z), from Lemmas 1.2–1.3 and Lemma 1.7 in [22], then
Z p
(1 − |z|2 )q− 2 −pσ (1 − |ϕa (z)|2 )s e
p logk dv(z)
sδ
|ϕa (z)|≤ 21 |ϕa (z)| |1 − hz, wi|
n+β+1− 2 1 − |z|2
Z p p
(1 − |a|2 )q+n+1− 2 −pσ (1 − |u|2 )q+s− 2 −pσ logk 1−|ϕea (u)|2 dv(u)
= p
|u|≤ 12 |u|sδ |1 − hu, ai|2q+2n+2−p−2pσ |1 − hϕa (u), wi|n+β+1− 2
Z p
(1 − |a|2 )q+n+1− 2 −pσ k e
≤c p log dv(u)
sδ
|u|≤ 12 |u| |1 − hϕa (u), wi|
n+β+1− 2 1 − |a|2
Z p
(1 − |a|2 )q+n+1− 2 −pσ k e
=c p log dv(u)
sδ
|u|≤ 12 |u| |1 − hϕa (u), ϕa [ϕa (w)]i|
n+β+1− 2 1 − |a|2
Z p p
(1 − |a|2 )q+n+1− 2 −pσ |1 − hu, ai|n+β+1− 2 logk 1−|a| e
2 dv(u)
=c p p
|u|≤ 12 |u|sδ |1 − hu, ϕa (w)i|n+β+1− 2 |1 − hw, ai|n+β+1− 2
Z p
(1 − |a|2 )q+n+1− 2 −pσ e
≤ c1 n+β+1− p
logk dv(u)
sδ
|u|≤ 21 |u| |1 − hw, ai| 2 1 − |a|2
p Z 12
2nc1 (1 − |a|2 )q+n+1− 2 −pσ k e
= p log r2n−sδ−1 dr
|1 − hw, ai|n+β+1− 2 1 − |a|2 0
p
c2 (1 − |a|2 )q+n+1− 2 −pσ e
≤ p logk . (3.33)
|1 − hw, ai| n+β+1− 2 1 − |a|2
As long as β + pσ > q + n, then we may obtain
 
q+n+1 1 p
β > q + n − pσ > q + n − p − = − 1.
p 2 2
By (3.28)–(3.29), (3.31) and the Fubini theorem, (3.33) and Lemma 2.3, Lemma 1.8 in [22]
and Lemmas 2.1–2.2, we have
Z
e (z)|p (1 − |z|2 )q−p g s (z, a) logk e
|∇f dv(z)
1
|ϕa (z)|≤ 2 1 − |z|2
Z e (z)|p (1 − |z|2 )q−p (1 − |ϕa (z)|2 )s
|∇f e
≤c sδ
logk dv(z)
1
|ϕa (z)|≤ 2 |ϕa (z)| 1 − |z|2
Z p
(1 − |w|2 )pσ |Rf (w)|p (1 − |a|2 )q+n+1− 2 −pσ e
≤ c1 n+β+1− p
logk dvβ (w)
B |1 − hw, ai| 2 1 − |a|2
Z (1 − |w|2 )pσ+β−q−n−1 (1 − |a|2 )q+n+1− p2 −pσ logk e dv(w)
1−|a|2
≤ c2 I2 n+β+1− p k e
B |1 − hw, ai| 2 log 1−|w|2
≤ c3 I4 . (3.34)

By (3.32) and (3.34), we have I5 ≤ cI4 .

If min{2(q + s + 1), 2(q + n + 1)} > 1 ≥ p > 0, then we may obtain
Z p
(1 − |z|2 )q− 2 (1 − |ϕa (z)|2 )s e
pn+pβ+ p
logk dv(z)
sδ
|ϕa (z)|≤ 12 |ϕa (z)| |1 − hz, wi| 2 1 − |z|2
p
c(1 − |a|2 )q+n+1− 2 k e
≤ ′ +1− p log . (3.35)
|1 − hw, ai| n+β 2 1 − |a|2
No.6 X.J. Zhang et al: AN INTEGRAL ESTIMATE AND THE EQUIVALENT NORMS 1879

By (3.30)–(3.31) and the Fubini theorem, (3.35) and Lemma 2.3, Lemma 1.8 in [22] and Lemmas
2.1–2.2, as long as β ′ > q + n, then
Z
e (z)|p (1 − |z|2 )q−p g s (z, a) logk e
|∇f dv(z)
|ϕa (z)|≤ 12 1 − |z|2
Z p
|Rf (w)|p (1 − |a|2 )q+n+1− 2 e
≤c ′ p logk dvβ ′ (w)
B |1 − hw, ai| n+β +1− 2 1 − |a|2
Z (1 − |w|2 )β ′ −q−n−1 (1 − |a|2 )q+n+1− p2 logk e dv(w)
1−|a|2
≤ c1 I2 n+β ′ +1− p k e
B |1 − hw, ai| 2 log
1−|w|2
≤ c2 I4 . (3.36)

From (3.32) and (3.36), we may obtain the result.

If min{2(q + s + 1), 2(q + n + 1)} ≤ 1, then 0 < p < 1. The result is proved.
The proof is completed. 

Remark 3.4 When n = 1, since |∇f e (z)| = (1 − |z|2 )|f ′ (z)|, then the six quantities in
Theorem 3.3 are equivalent for all p > 0.
Remark 3.5 When n > 1, if p ≥ min{2(q + s + 1), 2(q + n + 1)}, then, for f (z) = z1 , by
computation, we have I2 < ∞ and I4 = ∞. This means that I2 and I4 are not equivalent. But,
I4 and I5 are finite or ∞ at the same time for all p > 0.

References
[1] Ortega J, Fabrega J. Corona type decomposition in some Besov spaces. Math Scand, 1996, 78: 93–111
[2] Li S L, Zhang X J, Xu S. The Bergman type operators on the F (p, q, s) type spaces in Cn . Chin J Conte
Math, 2017, 38A(4): 303–316
[3] Zhao R H. On a general family of function spaces. Ann Acad Sci Math Diss, 1996, 105: 110–120
[4] Jiang L J, He Y Z. Composition operators from β α to F (p, q, s). Acta Math Sci, 2003, 23B: 252–260
[5] Zhou Z H, Chen R Y. Weighted composition operator from F (p, q, s) to Bloch type spaces on the unit ball.
Int J Math, 2008, 19: 899–926
[6] Zhang X J, He C Z, Cao F F. The equivalent norms of F (p, q, s) space in Cn . J Math Anal Appl, 2013,
401: 601–610
[7] Yang W S. Generalized weighted composition operators from the F (p, q, s) space to the Bloch-type space.
Appl Math & Comput, 2012, 218: 4967–4972
[8] Zhang X J, Xiao J B. Weighted composition operator between two analytic function spaces. Adv Math
(China), 2006, 35: 453–462
[9] Ye S L. Weighted composition operators from F (p, q, s) into logarithmic Bloch space. J Kore Math Soc,
2008, 45: 977–991
[10] Liang Y X, Zhou X H, Chen R Y. Product of extended Cesáro operator and composition operator from the
logarithmic Bloch-type space to F (p, q, s) space on the unit ball. J Comput Anal Appl, 2013, 15: 432–440
[11] Yang W. Volterra composition operators from F (p, q, s) spaces to Bloch-type spaces. Bull Malay Math Sci,
2011, 34: 267–277
[12] Liang Y X. On an integral-type operator from a weighted-type space to F (p, q, s) on the unit ball. Complex
Var Ellip Equ, 2015, 60: 282–291
[13] Zhang X J, Xiao J B, Hu Z H, Liu Y L, Xiong D H, Wu Y. Equivalent characterization and application of
F (p, q, s) space in Cn . Acta Math Sin, 2011, 54: 1029–1042 (in Chinese)
[14] Li S L, Zhang X J, Xu S. The equivalent characterization of F (p, q, s) space on bounded symmetric domains
of Cn . Acta Math Sci, 2017, 37B(6): 1791–1802
[15] Ouyang C H, Yang W S, Zhao R H. Characterizations of Bergman spaces and Bloch space in the unit ball
of Cn . Trans Amer Math Soc, 1995, 347: 4301–4319
1880 ACTA MATHEMATICA SCIENTIA Vol.38 Ser.B

[16] Zhang X J, Xi L H, Fan H X, Li J F. Atomic decomposition of µ-Bergman Space in Cn . Acta Math Sci,
2014, 34B(3): 779–789
[17] Wulan H S. Compactness of the composition operators from the Bloch space B to Qk spaces. Acta Math
Sin (English Series), 2005, 21: 1415–1424
[18] Li S L, Zhang X J, Xu S. The compact composition operator on the µ-Bergman space in the unit ball. Acta
Math Sci, 2017, 37B(2): 425–438
[19] Zhang X J, Li M, Guan Y, Li J F. Atomic decomposition for µ-Bloch space in Cn . Sci Sin Math, 2015, 45:
1677–1688 (in Chinese)
[20] Xiao J. Composition operators associated with Bloch type spaces. Complex Variables Theory Appl, 2001,
46: 109–121
[21] Rudin W. Function Theory in the Unit Ball of Cn . New York: Springer-Verlag, 1980
[22] Zhu K H. Spaces of Holomorphic Functions in the Unit Ball. New York: Springer-Verlag, 2005
[23] Zhao R H. Distances from Bloch functions to some Mobius invariant spaces. Annl Acad Sci Fen Math,
2008, 33: 303–313
[24] Krantz S. Function Theory of Several Complex Variables. John Wiley & Sons, 1982
[25] Zhu K H. Holomorphic Besov spaces on bounded symmetric domains (II). Indiana Univ Math J, 1995, 44:
1017–1031
[26] Zhang X J, Guo Y T, Shang Q L, Li S L. The Gleason’s problem on F (p, q, s) type spaces in the unit ball
of Cn , Complex Anal Oper Theory, 2018, 12(5): 1251–1265
[27] Zhang X J, Xiao J B, Hu Z J. The multipliers between the mixed norm spaces in Cn . J Math Anal Appl,
2005, 311(2): 664–674