J. Math. Anal. Appl.

Contents lists available at SciVerse ScienceDirect

Journal of Mathematical Analysis and

Applications
journal homepage: www.elsevier.com/locate/jmaa

Composition operators acting on weighted Dirichlet spaces

Jordi Pau , Patricio A. Prez
Departament de Matemtica Aplicada i Analisi, Universitat de Barcelona, 08007 Barcelona, Spain

article

abstract

info

Article history:
Received 22 August 2012
Available online xxxx
Submitted by Thomas Ransford

We study composition operators acting on weighted Dirichlet spaces. We obtain estimates

for the essential norm, describe the membership in SchattenVon Neumann ideals and
characterize the composition operators with closed range.
2012 Elsevier Inc. All rights reserved.

Keywords:
Composition operators
Dirichlet spaces
Schatten classes

1. Introduction
Let D denote the open unit disk in the complex plane, and let : D D be analytic. The function induces a composition
operator C acting on H (D), the space of all analytic functions on D, by the formula
C f (z ) = f ((z )).
It is an interesting problem to describe the operator properties of C in terms of the function properties of the symbol
when the operator C acts on several spaces of analytic functions in D. In this paper, we are going to study the composition
operator C acting on weighted Dirichlet spaces D , > 0, so let us proceed to introduce these spaces.
For 0, the Dirichlet type space D consists of those analytic functions f on D with

f D = |f (0)| +
2

1/2
|f (z )| dA (z )
< ,

where
dA (z ) = (1 + ) (1 |z |2 ) dA(z ),
and dA(z ) = 1 dx dy is the normalized area measure on D. Observe that f D if and only if f belongs to the Bergman
space A2 . Let > 1, the weighted Bergman space A2 consists of those functions f H (D) with

f A2 =

1/2
|f (z )| dA (z )
< .
2

It is well known that D1 = H 2 , the classical Hardy space, and D = A22 if > 1. The results we are going to obtain about
composition operators on the spaces D are well known for the Hardy and Bergman spaces. Therefore we will focus on the
case 0 < < 1 (the reason why we exclude the case = 0 will be explained later). We are going to estimate the essential
norm of the composition operator acting on D , obtain a description of the membership in the SchattenVon Neumann
ideal Sp of the composition operator on D , and finally, we characterize the composition operators with closed range. All of
this will be accomplished with the aid of the generalized Nevanlinna counting function, which we study in the next section.

Corresponding author.
E-mail addresses: jordi.pau@ub.edu (J. Pau), patricioeada@yahoo.com.br (P.A. Prez).

0022-247X/$ see front matter 2012 Elsevier Inc. All rights reserved.
doi:10.1016/j.jmaa.2012.12.052

J. Pau, P.A. Prez / J. Math. Anal. Appl. (

2. The generalized Nevanlinna counting function

Suppose that : D D is analytic. The (classical) Nevanlinna counting function of is

N (z ) =

log

w:(w)=z

|w|

z D \ {(0)},

where the sum is interpreted as being zero if z (D). An important property of the counting function N is that, being not
necessarily subharmonic, it satisfies the submean value property (see [22] or [24]).
Submean value property for N . Suppose that : D D is analytic. Then
1

N (w) dA(w),
|B| B
where z D \ {(0)} and B is any euclidean disk centered at z contained in D \ {(0)}. Here |B| denotes the area of the
N (z )

disk B.
Let > 0 and : D D be analytic. The generalized Nevanlinna counting function of is

(1 |w|2 ) ,

N, (z ) =

z D \ {(0)},

w:(w)=z

where, as before, the last sum is interpreted as being zero if z (D).

The following change of variables formula can be found, for example, in [4, Proposition 2.1] or [6, Theorem 2.32].
Change of variables formula. If f is nonnegative on D and : D D is analytic, then

f (z ) | (z )|2 (1 |z |2 ) dA(z ) =

f (z ) N, (z ) dA(z ).
D

For a D, denote by a the disk automorphism defined by

a (z ) =

az
1 a z

z D.

A key tool for our results is the submean value property for the generalized counting function N, . This fact is deduced
from the corresponding property for the classical Nevanlinna counting function together with the following formula due to
A. Aleman (see Lemma 2.3 in [4]) that relates the function N, with the classical Nevanlinna counting function N .
Alemans formula. Let 0 < < 1 and : D D be analytic and nonconstant. Then for every (D),
N, ( ) =

(z ) Nz ( ) dA(z ),

(2.1)

where (z ) = (1 |z |2 ) , and denotes the standard Laplace operator.

Proposition 2.1. Let 0 < < 1 and 0 < p < . Suppose that : D D is analytic. Then there is a positive constant
C = Cp < such that
N, ( )
p

|B|

N, (w)p dA(w),
B

where D \ {(0)} and B is any euclidean disk centered at contained in D \ {(0)}. Moreover, one can take C = 1 if p 1.
Proof. We begin with the case p = 1. This case has been proved recently in [10]. Let (z ) = (1 |z |2 ) . An application of
Alemans formula together with the submean value property for the classical Nevanlinna counting function, and taking into
account that (z ) 0, gives
N, ( )

1
2

(z )

|B|

Nz (w) dA(w)

dA(z ),

for any euclidean disk B centered at contained in D \ {(0)}. Now, we continue by using Fubinis theorem together with
another application of Alemans formula to obtain
N, ( )

|B|
1

|B|

1
2

(z ) Nz (w) dA(z ) dA(w)

N, (w) dA(w),

which proves the desired result for p = 1 with C = 1. This special case together with Hlders inequality then implies that
the desired inequality holds with C = 1 for all p 1.

J. Pau, P.A. Prez / J. Math. Anal. Appl. (

To deal with the case p (0, 1) we use an idea from [19]. Since the result holds for p = 1 with C = 1, a standard regularization argument from potential theory (see p. 51 of [21]) produces a nonnegative subharmonic function u on D \ {(0)}
such that N, u everywhere in D \ {(0)} and N, = u almost everywhere on D \ {(0)} (the function u is just
the upper semicontinuous regularization of N, ). Now, by an inequality due to Hardy and Littlewood (see [7, Lemma 2] or
Lemma 3.7 in Chapter III of [8]), there is a positive constant C = Cp such that
u ( )
p

|B|

u (w)p dA(w),
B

for any euclidean disk B centered at contained in D \ {(0)}. Observe that the inequality in [7] or [8] is stated for the
absolute value of a harmonic function, but its proof actually works for nonnegative subharmonic functions. Thus
N, ( )p u ( )p

|B|

u (w)p dA(w) =

completing the proof of the proposition.

|B|

N, (w)p dA(w)
B

Remark. When considering the composition operator acting on the classical Dirichlet space D , one should replace the
function N, (z ) by the counting function n (z ) := #{w : (w) = z }. However, the function n does not satisfy the analogue
of Proposition 2.1, that is, the counting function n (z ) does not necessarily satisfies the generalized submean value property
(see [25] for an example). For that reason, and due to the fact that Proposition 2.1 plays a prominent role in the proof of our
results, we exclude the case = 0 of our study.
3. The essential norm of the composition operator
The essential norm of a bounded linear operator T is defined to be the distance to the compact operators, that is,

T e = inf{T K : K is compact}.
It is our aim in this section to obtain estimates for the essential norm of the composition operator C acting on D . Before
doing that, we describe the symbols inducing bounded composition operators on D . This result can be found in [10].
For completeness we offer here the proof. A different description in terms of Carleson measure properties of N, dA can be
found in [25].
Theorem 3.1. Let 0 < < 1 and : D D be analytic. The composition operator C is bounded on D if and only if
sup
aD

N, (a)
< .
(1 |a|2 )

(3.1)

Proof. Suppose first that the composition operator C is bounded on D , that is, C f D C f D for any f D . For
a D, apply the previous inequality to the functions
2 1+
2

f a ( z ) = ( 1 | a| )

d
,
(1 a )2+

z D.

The use of Lemma 3.10 in [24] gives supaD fa 2D C . Also, the change of variables formula yields

C fa D = |fa ((0))| + (1 + )
2

|(fa ) (z )|2 N, (z ) dA(z ),

D

and therefore, we have

(1 |a|2 )2+
N, (z ) dA(z ) C ,
|1 a z |4+2

(3.2)

with the constant C not depending on a. Now, for a D with |a| > 12 (1 + |(0)|), let D(a) = z D : |z a| < 12 (1 |a|) .
Observe that (0) D(a). Then, Proposition 2.1, the well known fact that |1 a z | (1 |a|2 ) for z D(a), and (3.2),
finally gives

N, (a)

N, (z ) dA(z ) C

(1 |a|2 )2+2
N, (z ) dA(z )
|1 a z |4+2

(1 |a|) D(a)
D(a)

2 2+2
(1 |a| )
C
N, (z ) dA(z ) C (1 |a|2 ) ,
4+2

|
1

a
z
|
D
2

J. Pau, P.A. Prez / J. Math. Anal. Appl. (

where C does not depend on the point a. This gives

N, (a)

sup
0)|
|a|> 1+|(
2

(1 |a|2 )

< .

(3.3)

On the other hand, we know from the proof of Proposition 2.1 that there is a subharmonic function u such that N, u .
Since any upper semi-continuous function is bounded above on compact sets, we get
N, (a)

sup
0)|
|a| 1+|(
2

(1 |a|2 )

(1 |(0)|)

sup
0)|
|a| 1+|(
2

u (a) < .

This, together with (3.3) proves (3.1).

Conversely, assume that (3.1) holds. Then

|f (z )|2 N, (z ) dA(z ) C f 2D .
D

Also, the well known pointwise estimate for functions in D gives

|f ((0))|2

(1 |(0)|2 )

f 2D .

Since, by the change of variables formula, we have

C f D = |f ((0))| + (1 + )
2

|f (z )|2 N, (z ) dA(z ),
D

then, by the previous inequalities, the composition operator C is bounded on D .

It is now a classical result of J.H. Shapiro (see [22]) that the essential norm of the composition operator C acting on the
Hardy space H 2 equals

lim sup
|z |1

N (z )
log |1z |

In particular, C is compact on H 2 if and only if lim|z |1

N (z )
log |1z |

= 0 (see [22] or [6, Theorem 3.20]). Concerning the essential

norm of C acting on D we have the following result (see also [23]).

Theorem 3.2. Let 0 < < 1, and let : D D be analytic inducing a bounded composition operator on D . Then, the
following estimate for the essential norm of the composition operator C acting on D holds:

C 2e lim sup
|a|1

N, (a)

(1 |a|2 )

Proof. The upper estimate

C 2e C1 lim sup
|a|1

N, (a)
,
(1 |a|2 )

for some positive constant C1 can be found in p. 136 of CowenMacCluers book [6]. So let us proceed to show that the lower
estimate

C 2e C2 lim sup
|a|1

N, (a)
(1 |a|2 )

(3.4)

holds for some positive constant C2 . To this end, for each a D consider the functions

fa (z ) =

(1 |a|2 )1 2
,
1 a z

z D.

Using Lemma 3.10 in [24] it is easy to see that supaD fa D C . It is also clear that the functions fa converge to zero as
|a| 1 uniformly on compact subsets of D. Therefore, lim|a|1 K (fa )D = 0 for every compact operator K on D . This
implies that

C K C lim sup C fa K (fa )D

|a|1

C lim sup C fa D .
|a|1

J. Pau, P.A. Prez / J. Math. Anal. Appl. (

Since this holds for any compact operator K , and

C e = inf{C K : K is compact},
we have

C e C lim sup C fa D .

(3.5)

|a|1

Now, by the change of variables formula,

C fa 2D = |fa ((0))|2 + (1 + )

|(fa ) (z )|2 N, (z ) dA(z )

N, (z )
(1 |a|2 )2
2
2 2
+ (1 + )|a| (1 |a| )
dA(z ).
=
|1 a (0)|2
|
1
a z |4
D
D

It is clear that the first term tends to zero as |a| 1 . Thus

lim sup C fa 2D (1 + ) lim sup |a|2 (1 |a|2 )2
|a|1

|a|1

N, (z )

D(a)

|1 a z |4

dA(z ),

where, as before,

D(a) =

z D : |z a| <

1
2

(1 |a|) .

Observe that (0) D(a) if |a| is close enough to 1, and so, after using that |1 a z | (1 |a|) for z D(a), we can apply
Proposition 2.1 in order to obtain
lim sup C fa 2D C lim sup
|a|1

|a|1

N, (a)
.
(1 |a|2 )

This together with (3.5) proves (3.4) finishing the proof of the theorem.

Since C e = 0 if and only if C is compact, as an immediate consequence of the previous theorem, we obtain the
following description of compact composition operators on Dirichlet type spaces, result that has been obtained recently by
Kellay and Lefevre in [10].
Corollary 3.3. Let : D D be analytic and 0 < < 1. Then C is compact on D if and only if
N, (a)
0 as |a| 1 .
(1 |a|)
Corollary 3.4. Let : D D be univalent and 0 < < 1. Then C is compact on D if and only if
1 |(a)|2

lim

|a|1

1 |a|2

= .

(3.6)

Proof. Since is univalent, then N, (a) = (1 | 1 (a)|2 ) , and the result is an immediate consequence of the previous
corollary. 
We remark here that, by the JuliaCaratheodory Theorem, the condition (3.6) is equivalent to the fact that does not
have finite angular derivatives at any point of the circle.
4. Schatten class composition operators
Schatten classes. If H and K are separable Hilbert spaces, a compact operator T from H to K is said to belong to the Schatten
class Sp = Sp (H , K ) if its sequence of singular numbers is in the sequence space p . Recall that the singular numbers of a
compact operator T are the square root of the eigenvalues of the positive operator T T , where T denotes the adjoint of T .
Also, the compact operator T admits a decomposition of the form
T =

n , en H fn ,

J. Pau, P.A. Prez / J. Math. Anal. Appl. (

where {n } are the singular numbers of T , {en } is an orthonormal set in H, and {fn } is an orthonormal set in K . Moreover, if
the singular values {n } are ordered in a decreasing order, then

n = n (T ) = inf{T K : rank K n}.

For p 1, the class Sp is a Banach space with the norm

1/p

T p =

|n |

while for 0 < p < 1 one has the inequality

S + T pp S pp + T pp .
We refer the reader to [24, Chapter 1] for a brief account on the theory of Schatten p-classes.
Schatten class Toeplitz operators on Bergman spaces. Let > 1 and let be a positive function in L1 (D, dA ). The Toeplitz
operator T with symbol acting on the Bergman space A2 is
T f (z ) =

(w) f (w)
dA (w),
(1 w
z )2+

f A2 .

Note that the operator T is well defined on H , the algebra of bounded analytic functions on D, that is dense on A2 , and
therefore the Toeplitz operator T is densely defined on A2 . A description of those positive symbols for which T is
bounded or compact on A2 can be found, for example, in Chapter 7 of K. Zhus book [24], where one can also found the following description, essentially due to D. Luecking [17], of Toeplitz operators with positive symbols belonging to the Schatten
class Sp .
Theorem A. Let L1 (D, dA ) be a nonnegative function on D, and let > 1 and 0 < p < . Then T Sp (A2 ) if and
only if the function

r (z ) =

(1 |z |2 )2+

(z ,r )

(w) dA (w),

0 < r < 1,

is in Lp (D, d).
Here, d(z ) = (1 |z |2 )2 dA(z ) is the hyperbolic measure in D, and (z , r ) = {w : |z (w)| < r } is the pseudohyperbolic disk with center z and radius r. We are going to use this result for our study of the Schatten class composition
operators on Dirichlet type spaces.
Composition operators belonging to the Schatten ideals. A characterization of the membership in the Schatten class Sp for
0 < p < of the composition operator in the Hardy space H 2 was obtained by D. Luecking and K. Zhu [19]. His result reads
as follows: let 0 < p < and let : D D be analytic. Then C Sp (H 2 ) if and only if
N (z )
log |1z |

Lp/2 (D, d).

Next, we are going to extend this result to all Dirichlet type spaces D with 0 < < 1.
Theorem 4.1. Let : D D be analytic, 0 < < 1 and 0 < p < . Then C Sp (D ) if and only if
N, (z )
Lp/2 (D, d).
(1 |z |2 )
In order to prove this result we use a well known relationship of composition operators and Toeplitz operators in order
to deduce the result from Theorem A and Proposition 2.1. Before going to the proof, we need some preliminaries in order to
make clear the connection mentioned above.
be the closed subspace of D consisting of functions f with f (0) = 0. If : D D is analytic and (0) = 0, then
Let D
invariant, and we can think of C as an operator on D . In the case that (0) = 0,
the composition operator C leaves D
D instead of C : D D . Also, when discussing
it turns out that it is more convenient to work with C : D
the membership in Schatten classes of the composition operator C , without loss of any generality, we can always assume
(0) = 0 and consider the operator acting on D . Indeed, if : D D is analytic with a = (0), then the function
= a satisfies (0) = 0, and = a , so that C = C Ca . Since Ca is invertible (the inverse being itself), it
has codimension
follows that C is in the Schatten class Sp (D ) if and only if C is in the same Schatten class Sp (D ). Since D

1 in D , the membership in Sp of C : D D is the same as that of C : D D .

A2 defined by U f (z ) = f (z ). It is obvious that U is a unitary operator
Now, consider the linear mapping U : D
onto A2 . We now consider the action of U on a composition operator. If : D D is analytic with (0) = 0, and
from D

J. Pau, P.A. Prez / J. Math. Anal. Appl. (

, then it follows easily from elementary calculations that U C U = D on A2 ,

C is the composition operator acting on D
where D is the weighted composition operator on A2 given by
D f (z ) = f (z ) (z ).

D is the same as that of D : A2 A2 . The following result

Therefore, the study of the membership in Sp of C : D
exhibits the relationship between D and a certain Toeplitz operator acting on the Bergman space A2 .
Lemma 4.2. Assume that : D D is analytic with (0) = 0, and let

(z ) =

N, (z )
,
(1 |z |2 )

z D.

(4.1)

Then D D = T on A2 .
Proof. Let , be the inner product in A2 . Using the properties of the reproducing kernel of A2 it is easy to see that, for any
f and g in A2 , we have

T f , g = (1 + )

f (w) g (w) N, (w) dA(w).

D

On the other hand, by the change of variables formula,

D D f , g = D f , D g

=
f (z ) g (z ) | (z )|2 dA (z )
D

= (1 + ) f (w) g (w) N, (w) dA(w).

D

It follows that T f , g =

D D f , g

for all f and g in A2 and hence D D = T on A2 .

Proof of Theorem 4.1. By the previous observations we may assume that (0) = 0. Also, by the remarks made and
Lemma 4.2, we have C Sp (D ) if and only if D Sp (A2 ) if and only if the Toeplitz operator T Sp/2 (A2 ), where
r is in Lp/2 (D, d). Now, it is clear
is the function defined in (4.1). By Theorem A, T Sp/2 (A2 ) if and only if the function
that the pseudo-hyperbolic disk (z , r ) contains an euclidean disk centered at z of radius (1 |z |) with depending only
on r. Therefore, since (1 |w|) (1 |z |) for w (z , r ), using Proposition 2.1 we see that, for 0 < q < , there is a
positive constant C such that

(z )q

(1 |z |2 )2+

(z ,r )

(w)q dA (w),

|z | > r .

(4.2)

r (z ). Hence Lp/2 (D, d) if

r is in Lp/2 (D, d) proving the necessity part
In particular, the case q = 1 gives (z ) C
of the theorem.
r is in Lp/2 (D, d). From the case q = p/2 of (4.2) and the well known
Conversely, if Lp/2 (D, d) we will show that
2
2
estimates (1 |w| ) (1 |z | ) |1 w
z | for w (z , r ), we deduce

r (z )p/2 C sup (w)p : w (z , r )

1
(u)p/2 dA (u)
C sup
2 2+
w(z ,r ) (1 |w| )
(w,r )

p/2
(
u
)
C (1 |z |2 )2+
dA (u).
z |4+2
D |1 u
Therefore, by Fubinis theorem and Lemma 3.10 of [24],

r (z )p/2 d(z ) C

(u)

D
C

p/2

(u)

(1 | z |2 )
dA(z ) dA (u)
|1 u z |4+2

d(u).

This completes the proof of the theorem.

J. Pau, P.A. Prez / J. Math. Anal. Appl. (

Some consequences. It is known that the results on compactness and membership in Schatten classes for composition
operators on Hardy spaces can be reformulated in terms of the function

(z ) =

N (z )
1 |z |2

where
N (z ) :=

(1 |w|2 ).

(w)=z

Then Shapiros compactness theorem can be restated by saying that C is compact in H 2 if and only if lim|z |1 (z ) = 0, and

the necessary and sufficient condition for the membership in C in Sp in LueckingZhus theorem is then Lp/2 (D, d).
Also, since is a bounded function, one has

(z ) C (z ) C

N, (z )

(1 |z |2 )

0 < < 1.

Taking all of this into account and bearing in mind Theorems 3.1 and 4.1 it follows that, for 0 < < 1, the compactness of
the composition operator C in D implies the compactness of C in H 2 , and the membership of C in the Schatten class Sp of
D implies that C belongs to Sp (H 2 ). Also, when is univalent (or even with bounded valence) C is compact on the Hardy
space H 2 if and only if C is compact on D , 0 < < 1. Therefore, several existing examples of composition operators on
the Hardy space may be used to produce relevant examples of composition operators on D . In particular, Lottos simpler
example [15] of a compact composition operator on H 2 not being HilbertSchmidt in H 2 is actually an example of a compact
composition operator on D that is not HilbertSchmidt in D . Also, Akeroyds example [1] provides a compact composition
operator on D that does not belong to any Schatten ideal Sp of D .
5. Composition operators between different Dirichlet type spaces
One can also consider the composition operator C acting from one Dirichlet type space D to another Dirichlet type
space D with = . The proof of the result stated below is the same as the case = considered above, and therefore
it is omitted.
Theorem 5.1. Let : D D be analytic. Let 0 and > 0. Then
N, (z )

(a) C : D D is bounded if and only if supz D (1|z |2 ) < .

(b) The essential norm of C : D D is comparable to the quantity
lim sup
|z |1

N, (z )
.
(1 |z |2 )
N, (z )

(c) C : D D is compact if and only if lim|z |1 (1|z |2 ) = 0.

(d) Let 0 < p < . Then C Sp (D , D ) if and only if
N, (z )

(1 |z |2 )

Lp/2 (D, d).

We want to remark here that in the previous result the case = 0 is included, that is, when the starting space is the
classical Dirichlet space.
6. Composition operators with closed range
In this section, we are going to describe those bounded composition operators acting on D with closed range. The
corresponding result for the Hardy space H 2 was obtained by N. Zorboska [26] (see [5] for a different description): a
composition operator C has closed range in H 2 if and only if there are positive constants , and 0 < r < 1 so that if
G = {w : (w) > }, then
Area(G (z , r )) (1 |z |2 )2
for all pseudo-hyperbolic disks (z , r ) with z D. Here (z ) = N (z )/(1 |z |). A similar result for Bergman spaces was
also obtained there [26] (see also [2,3] for some variants). A key ingredient for the proof of that result (that we also will use
in order to characterize the closed range composition operators on D ) is the following result due to D. Luecking [16].
Theorem B. Let G be a measurable subset of D. The following conditions are equivalent.
(a) There is a positive constant C such that for every g in A2 , > 1,

|g (z )|2 dA (z ) C
D

|g (z )|2 dA (z ).
G

J. Pau, P.A. Prez / J. Math. Anal. Appl. (

(b) There exists > 0 and 0 < r < 1 such that

Area(G (z , r )) (1 |z |2 )2
for every z D.
We need a modified version of this theorem whose proof can be deduced from Theorem B in the same way as in Corollary
3.34 of [6], so that the proof will be omitted.
Proposition 6.1. Let (z ) be a bounded, measurable non-negative function on D. The following conditions are equivalent.
(a) There exists M > 0 such that for every f D , 0,

|f (z )| dA (z ) M

|f (z )|2 (z ) dA (z ).

(b) There exists > 0, > 0 and 0 < r < 1 such that
Area(G (z , r )) (1 |z |2 )2
for every z D, where G = {w D : (w) > }.
We also need the following well known result (see [6, Proposition 3.30] for example) saying that the closed range
composition operators are precisely those operators that are bounded below.
Lemma C. A bounded composition operator C acting on a Hilbert space H of analytic functions in the disk has closed range if
and only if there exists m > 0 such that

C f H mf H for all f H .
For : D D analytic and > 0, consider the function

, (z ) =

N, (z )

(1 |z |2 )

Note that this is the same function that appears in (4.1). Now we are prepared to state and prove our description of the
closed range composition operators acting on D .
Theorem 6.2. Let 0 < < 1 and let : D D be analytic inducing a bounded composition operator C on D . Then
the composition operator C has closed range on D if and only if there are positive constants , and 0 < r < 1 so that if
G = {w : , (w) > }, then
Area(G (z , r )) (1 |z |2 )2
for all pseudo-hyperbolic disks (z , r ) with z D.
Proof. Since, by Theorem 3.1, , (z ) is a bounded function, the result follows from a direct application of Proposition 6.1
and the following claim: a bounded composition operator C has closed range in D if and only if there exists a constant
M > 0 so that

|f (z )|2 , (z ) dA (z ) M
D

|f (z )|2 dA (z )

(6.1)

whenever f is in D . In order to prove that claim, we first assume that (0) = 0. In this case, we can think as C acting on
D , the closed subspace of D consisting of functions f with f (0) = 0. By Lemma C, the composition operator C has closed
range if and only if there exists a constant m > 0 so that

f 2D mC f 2D
. By the change of variables formula, we have
for all f D
2

|f (z )|2 , (z ) dA (z ),

C f D =
D

that proves the claim in that case.

Next, assume that (0) = a = 0. Consider = a , so that (0) = 0. Since C = C Ca and Ca is invertible on D ,
then C has closed range if and only if C does. By the case already considered, C has closed range if and only if there exists
m > 0 so that

|f (z )|2 , (z ) dA (z ) m
D

|f (z )|2 dA (z ).
D

(6.2)

10

J. Pau, P.A. Prez / J. Math. Anal. Appl. (

If (6.1) holds, by the change of variables formula,

|f (z )|2 , (z ) dA (z ) =

|(f ) (z )|2 dA (z )
D

|(f a ) (z )|2 , (z ) dA (z )

M |(f a ) (z )|2 dA (z )
D

= (1 + )M |f (z )|2 (1 |a (z )|2 ) dA(z )

D

2
M2
(1 |a|2 ) |f (z )|2 dA (z ).

Hence (6.2) holds with m = M2

(1 |(0)| ) , and therefore C has closed range. Conversely, if C has closed range in
D , then (6.2) holds, and since = a we can use the same argument to show that (6.1) is satisfied. This completes the
proof. 
As an immediate consequence, we obtain the following alternate description of closed range composition operators in

D similar to the one obtained recently in [13, Theorem 5.1] for the Hardy space.
Corollary 6.3. Let 0 < < 1 and let : D D be analytic inducing a bounded composition operator C on D . Then C has
closed range in D if and only if there is 0 < r < 1 and a positive constant C such that

1
Area((z , r ))

(z ,r )

, (w) dA(w) C

(6.3)

for any pseudo-hyperbolic disk (z , r ), z D.

Proof. If C has closed range in D , by Theorem 6.2, there are constants > 0 and > 0 such that
Area(G (z , r )) Area((z , r ))
for any pseudo-hyperbolic disk (z , r ). Therefore,

1
Area((z , r ))

(z ,r )

, (w) dA(w)

Area((z , r ))

1
Area((z , r ))
1
Area((z , r ))

G
(z ,r )

G
(z ,r )

G
(z ,r )

Area(G (z , r ))
Area((z , r ))

Conversely, assume that (6.3) holds. Choose 0 < <

, (w) dA(w) +

C
.
2

, (w) dA(w)

Then
1

Area((z , r ))

(z ,r )\G

, (w) dA(w)

, (w) dA(w) + ,

so that
C
2

1
Area((z , r ))

G
(z ,r )

, (w) dA(w).

Since, by Theorem 3.1, M := supwD , (w) < , we deduce that

Area(G (z , r )) Area((z , r ))
with = C /(2M ). Thus, by Theorem 6.2, the composition operator C has closed range in D .

Corollary 6.4. Let 0 < < and : D D be analytic. Assume that C is bounded on D . If C has closed range on D ,
then C has closed range on D .
Proof. Since, for > , one has , (z ) C , (z ) for some positive constant C and, clearly, if C is bounded on D then
C is bounded on D (see Theorem 3.1), the result is an immediate consequence of Corollary 6.3. 

J. Pau, P.A. Prez / J. Math. Anal. Appl. (

11

Remark. It is standard to see that the results of Theorem 6.2 and Corollary 6.3 can be reformulated in terms of Carleson
sectors. Recall that, for a given arc I of the unit circle with normalized arc length |I |, the Carleson sector S (I ) based on I is
S (I ) = {z = reit D : 1 r < |I |; eit I }.
Then, in the condition in Theorem 6.2 one can replace the pseudo-hyperbolic disks by Carleson sectors, so that C has closed
range in D if and only if
Area(G S (I )) |I |2
for all arcs I. Also, the condition (6.3) in Corollary 6.3 can be replaced by the condition
1

|I |2

S (I )

, (w) dA(w) C

for all arcs I in the unit circle.

Some examples. Some consequences of Theorem 6.2 are listed below. All the assertions can be proved in a similar manner as
in [26], so that the proofs will be omitted here. Let > 0 and : D D be analytic. Then:

If the range of on D misses a neighborhood of a point on the unit circle then C does not have a closed range on D .
If the range of on D has a hole that includes a disk internally tangent to the unit disk, then C does not have a closed
range on D .

Assume that is univalent, and let : D D be analytic with (D) (D). If C does not have a closed range in D ,
then C does not have a closed range in D .

If (D) D \ [0, 1), then C does not have a closed range in D .

Closed range composition operators on the Dirichlet space. The counting function n (z ) does not necessarily satisfies the
submean value property, and the analogues of Theorem 6.2 and Corollary 6.3 for the Dirichlet space (replacing the function
, by the counting function n ) can fail. As far as we know, it is still an open problem to describe the composition operators
with closed range in the Dirichlet space D . We refer the reader to [9,18] for more information on closed range composition
operators on the Dirichlet space.
7. Pullback measures vs. counting functions
It is well known that the results on compactness or membership in Schatten classes of composition operators on Hardy
spaces can be restated in terms of pullback measures. More concretely, given : D D analytic, consider the measure m
defined on subsets E of the closed unit disk by

m (E ) = ( )1 (E )

where denotes the arc length measure on the unit circle and is the radial limit function of . Then, see [20], the
composition operator C is compact on H 2 if and only if m is a vanishing Carleson measure, that is, if
lim

m (S (I ))

|I |0

|I |

= 0,

or, equivalently, if
lim

|z |1

m (S (Iz ))
1 |z |

= 0,

where, for z = reit D, Iz denotes the arc centered at eit with |Iz | = 1 |z |. In view of Shapiros compactness theorem [22],
it is natural to expect some relationship between the quantities N (z ) and m (S (Iz )). More evidences on that is the fact that
the membership in the Schatten classes of the composition operator on H 2 can also be described in terms of the pullback
measure m (see [17,11]). Essentially, the result is that C belongs to Sp (H 2 ) if and only if the function m (S (Iz ))/(1 |z |) is
in Lp/2 (D, d) (compare this with LueckingZhus condition mentioned in Section 4). Recently, in [12], the relation between
N (z ) and m (S (Iz )) has been quantified in the following form: there exists positive constants C1 , c1 , C2 , c2 such that, for
z D with |z | close enough to 1, one has
N (z ) C1 m (S (c1 Iz )),

(7.1)

and
m (S (Iz ))

C2

(1 |z |)2

S (c2 Iz )

N (w) dA(w).

(7.2)

12

J. Pau, P.A. Prez / J. Math. Anal. Appl. (

Here, given a positive constant c, we denote by cI the arc with the same center as I and length c |I |. Of course, we can replace
N (z ) by N (z ) := N,1 (z ) in the previous inequalities.

A similar phenomenon occurs in the setting of the weighted Bergman spaces A2 with > 1. The pullback measure
that enters in action now is the measure m, defined on subsets E of the open unit disk D by
m, (E ) =

dA (z ).

1 (E )

The boundedness, compactness and membership in Schatten classes of the composition operator on A2 can be described in
terms of the properties of the Nevanlinna counting function N,2+ or in terms of properties of the function m, (S (Iz ))
(see [20,22] for boundedness and compactness, and [17,19] for the membership in Schatten classes). As before, the
relationship between N,2+ (z ) and m, (S (Iz )) can be quantified. The following result has been proved in [14] for = 0,
but the same proof works for all > 1. For the sake of completeness we provide the proof here.
Proposition 7.1. Let > 1 and : D D be analytic. There exists positive constants C1 , c1 , C2 , c2 such that, for z D with
|z | close enough to 1, one has
(a) N,2+ (z ) C1 m, (S (c1 Iz )),
(b) m, (S (Iz )) (1|2z |)2
C

S (c2 Iz )

N,2+ (w) dA(w).

Proof. As in [22, Proposition 6.6] one has

N,2+ (z ) = 4(2 + )(1 + )

N,1 (r , z ) (1 r 2 ) r dr ,

(7.3)

where

N,1 (r , z ) :=

(1 |w|2 /r 2 ) = Nr ,1 (z ),

0 < r 1,

(w)=z ,|w|<r

with r (z ) = (rz ). Then, from (7.3) and (7.1), we obtain

N,2+ (z ) = 4(2 + )(1 + )

Nr ,1 (z ) (1 r 2 ) r dr

0
1

mr (S (c1 Iz )) (1 r 2 ) r dr

0
1

= C

eit : (reit ) S (c1 Iz ) (1 r 2 ) r dr

= C m, (S (c1 Iz ))
which proves (a). Part (b) is proved in a similar manner bearing in mind that
m, (S (Iz )) =

mr (S (Iz )) (1 r 2 ) r dr

and using (7.2) and (7.3).

In the setting of Dirichlet type spaces, one considers the pullback measure , defined on subsets E of the open unit
disk D by

, (E ) =

1 (E )

| (z )|2 dA (z ).

The boundedness and compactness of the composition operator C on D can also be described in terms of the pullback
measure , [20]. Indeed, a change of variables gives

|(C f ) (z )|2 dA (z ) =
D

|f (w)|2 d, (w).
D

It follows that C is bounded on D if and only if , is a Carleson measure for A2 , that is (see [24, Theorem 7.4] for example), if
sup
z D

, (D(z ))
< ,
(1 |z |2 )2+

J. Pau, P.A. Prez / J. Math. Anal. Appl. (

where D(z ) = {w D : |w z | <

sup
z D

1
2

13

(1 |z |)}; or equivalently (using Carleson sectors) if

, (S (Iz ))
< .
(1 |z |2 )2+

Also C is compact in D if and only if , is a vanishing Carleson measure for A2 , that is (see [24, Theorem 7.7]), if
lim

|z |1

, (D(z ))
, (S (Iz ))
= 0 lim
= 0.
(1 |z |2 )2+
|z |1 (1 |z |2 )2+

Also, the membership of the composition operator C in the Schatten class Sp (D ) can be described in terms of properties of
the function , (D(z )) (see [17]). Thus, in view of Theorem 3.1, Corollary 3.3 and Theorem 4.1, it is natural to expect some
relationship between the quantities N, (z ) and , (D(z ))/(1 |z |)2 . This is quantified in the next result.
Proposition 7.2. Let > 0 and : D D be analytic. There is a constant C > 1 such that
C
, (D(z ))
N, (z )

C
(1 |z |)2
(1 |z |)2
1

D(z )

N, (w) dA(w).

Also, the same result holds for Carleson sectors, that is,

, (S (2Iz ))
C
N, (z )

C
(1 |z |)2
(1 |z |)2
1

S (2Iz )

N, (w) dA(w).

Proof. Denote by E the characteristic function of the set E. The definition of the measure , together with the change of
variables formula yields

, (D(z )) =

1 (D(z )) (w) | (w)|2 dA (w) =

D

= (1 + ) D(z ) ( )N, ( ) dA( ).

D(z ) ((w)) | (w)|2 dA (w)

From here the upper inequality is trivial, and the lower inequality is a consequence of the submean value property for N,
(see Proposition 2.1). The result in terms of Carleson sectors follows exactly the same proof. 
Acknowledgments
The first author is supported by SGR grant 2009SGR 420 (Generalitat de Catalunya) and DGICYT grant MTM2011-27932C02-01 (MCyT/MEC).
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]

J. Akeroyd, On Shapiros compactness criterion for composition operators, J. Math. Anal. Appl. 379 (2011) 17.
J. Akeroyd, P. Ghatage, Closed-range composition operators on A2 , Illinois J. Math. 52 (2008) 533549.
J. Akeroyd, P. Ghatage, M. Tjani, Closed-range composition operators on A2 and the Bloch space, Integral Equations Operator Theory 68 (2010) 503517.
A. Aleman, Hilbert spaces of analytic functions between the Hardy and the Dirichlet space, Proc. Amer. Math. Soc. 115 (1992) 97104.
J. Cima, J. Thomson, W. Wogen, On some properties of composition operators, Indiana Univ. Math. J. 24 (1974) 215220.
C. Cowen, B. MacCluer, Composition Operators on Spaces of Analytic Functions, CRC Press, Boca Raton, Florida, 1995.
C. Fefferman, E. Stein, H p spaces of several variables, Acta Math. 129 (1972) 137193.
J.B. Garnett, Bounded Analytic Functions, revised first ed., Springer-Verlag, New York, 2007.
M. Jovovic, B. MacCluer, Composition operators on Dirichlet spaces, Acta Sci. Math. (Szeged) 83 (1997) 229247.
K. Kellay, P. Lefvre, Compact composition operators on weighted Hilbert spaces of analytic functions, J. Math. Anal. Appl. 386 (2012) 718727.
P. Lefvre, D. Li, H. Queffelec, L. Rodrguez-Piazza, Some examples of compact composition operators on H 2 , J. Funct. Anal. 255 (2008) 30983124.
P. Lefvre, D. Li, H. Queffelec, L. Rodrguez-Piazza, Nevanlinna counting function and Carleson function of analytic maps, Math. Ann. 351 (2011)
305326.
P. Lefvre, D. Li, H. Queffelec, L. Rodrguez-Piazza, Some revisited results about composition operators on Hardy spaces, Rev. Mat. Iberoam. 28 (2012)
5776.
P. Lefvre, D. Li, H. Queffelec, L. Rodrguez-Piazza, Compact composition operators on BergmanOrlicz spaces, Preprint.
B.A. Lotto, A compact composition operator that is not HilbertSchmidt, Contemp. Math. 213 (1998) 9397.
D. Luecking, Inequalities on Bergman spaces, Illinois J. Math. 25 (1981) 111.
D. Luecking, Trace ideal criteria for Toeplitz operators, J. Funct. Anal. 73 (1987) 345368.
D. Luecking, Bounded composition operators with closed range on the Dirichlet space, Proc. Amer. Math. Soc. 128 (2000) 11091116.
D. Luecking, K. Zhu, Composition operators belonging to the Schatten ideals, Amer. J. Math. 114 (1992) 11271145.
B. MacCluer, J.H. Shapiro, Angular derivatives and compact composition operators on the Hardy and Bergman spaces, Canad. J. Math. 38 (1986)
878906.
T. Ransford, Potential Theory in the Complex Plane, Cambridge University Press, Cambridge, 1995.
J.H. Shapiro, The essential norm of a composition operator, Ann. of Math. 125 (1987) 375404.
S. Stevic, A.K. Sharma, Essential norm of composition operators between weighted Hardy spaces, Appl. Math. Comput. 217 (2011) 61926197.
K. Zhu, Operator Theory in Function Spaces, second ed., in: Math. Surveys and Monographs, vol. 138, American Mathematical Society, Providence,
Rhode Island, 2007.
N. Zorboska, Composition operators on weighted Dirichlet spaces, Proc. Amer. Math. Soc. 126 (1998) 20132023.
N. Zorboska, Composition operators with closed range, Trans. Amer. Math. Soc. 344 (1994) 791801.