eigenvalues and spectra of composition operators acting on weighted bergman spaces of infinite order...

8
Mediterr. J. Math. 7 (2010), 565–572 DOI 10.1007/s00009-010-0060-1 1660-5446/10/040565-8, published online April 22, 2010 © 2010 Springer Basel AG Mediterranean Journal of Mathematics Eigenvalues and Spectra of Composition Operators Acting on Weighted Bergman Spaces of Infinite Order on the Unit Polydisk Elke Wolf To Klaus D. Bierstedt Abstract. We give the spectra of bounded composition operators acting on the weighted Bergman spaces of infinite order on the unit polydisk defined for a weight v which is radial in each component, when the symbol of the operator has a fixed point in the unit polydisk. Mathematics Subject Classification (2010). Primary 47B33; Secondary 47B38. Keywords. Composition operator, spectrum, weighted Bergman space of infinite order, eigenvalues. 1. Introduction Let v be a strictly positive continuous function (weight ) on the unit polydisk D N := {z =(z 1 ,...,z N ) C N ; |z i | < 1 1 i N }, N 1. We are interested in the weighted Bergman spaces of infinite order H v := {f H (D N ); f v := sup zD N v(z )|f (z )| < ∞} and H 0 v := {f H v ; lim |zi|→1 v(z )|f (z )| =0 1 i N } endowed with norm . v where H (D N )= {f : D N C; f holomorphic}. In the one-dimensional case spaces of this type have been extensively studied in [2], [4], [11]. Let ϕ : D N D N be an analytic self-map of D N . Each such map induces through composition a linear composition operator C ϕ (f )= f ϕ acting on this type of spaces.

Upload: elke-wolf

Post on 10-Jul-2016

216 views

Category:

Documents


0 download

TRANSCRIPT

Mediterr. J. Math. 7 (2010), 565–572DOI 10.1007/s00009-010-0060-11660-5446/10/040565-8, published online April 22, 2010© 2010 Springer Basel AG

Mediterranean Journalof Mathematics

Eigenvalues and Spectra ofComposition Operators Acting onWeighted Bergman Spaces of Infinite Orderon the Unit Polydisk

Elke Wolf

To Klaus D. Bierstedt

Abstract. We give the spectra of bounded composition operators actingon the weighted Bergman spaces of infinite order on the unit polydiskdefined for a weight v which is radial in each component, when thesymbol of the operator has a fixed point in the unit polydisk.

Mathematics Subject Classification (2010). Primary 47B33; Secondary47B38.

Keywords. Composition operator, spectrum, weighted Bergman spaceof infinite order, eigenvalues.

1. Introduction

Let v be a strictly positive continuous function (weight) on the unit polydiskDN := {z = (z1, . . . , zN ) ∈ CN ; |zi| < 1 ∀1 ≤ i ≤ N}, N ≥ 1. We areinterested in the weighted Bergman spaces of infinite order

H∞v := {f ∈ H(DN ); ‖f‖v := sup

z∈DN

v(z)|f(z)| < ∞} and

H0v := {f ∈ H∞

v ; lim|zi|→1

v(z)|f(z)| = 0 ∀1 ≤ i ≤ N}

endowed with norm ‖.‖v where H(DN ) = {f : DN → C; f holomorphic}. Inthe one-dimensional case spaces of this type have been extensively studied in[2], [4], [11].Let ϕ : DN → DN be an analytic self-map of DN . Each such map inducesthrough composition a linear composition operator Cϕ(f) = f ◦ ϕ acting onthis type of spaces.

566 Elke Wolf Mediterr. J. Math.

Composition operators have been studied on various spaces of analytic func-tions, for an introduction and general information we refer the reader to theexcellent monographs [7] and [15]. In [3], [4], [14] and [6] boundedness andcompactness of composition operators acting on weighted Bergman spacesof infinite order were treated in the case N = 1, see also [13] for the Blochspace case. The spectrum of a composition operator has recently been inves-tigated in [8], [17] and [12] as well as in [1] and [5]. More precisely, in thesetting of N = 1 Aron and Lindstrom determined the spectrum of a boundedweighted composition operator acting on weighted Banach spaces H∞

vpwhere

vp(z) = (1 − |z|2)p, p > 0 (see [1]), z ∈ D. This was generalized by Bonet,Galindo and Lindstrom in the sense that they were able to characterize thespectrum of a bounded composition operator defined on a weighted Banachspace H∞

v , where v is a typical weight. The aim of this article is to generalizea part of the results of [1] and [5].

2. Notations and definitions

We refer the reader to [7] and [15] for notation on composition operators. Theclosed unit ball of H∞

v resp. H0v is denoted by B∞

v resp. B0v . Many results on

weighted spaces of analytic functions and on operators between them haveto be given in terms of the so-called associated weights (see [2]) and in termsof the weights. For a weight v the associated weight v is defined as follows

v(z) :=1

sup{|f(z)|; f ∈ B∞v } =

1‖δz‖H∞

v′, z ∈ DN ,

where δz denotes the point evaluation of z. The associated weights are alsocontinuous and v ≥ v > 0 (see [2]). Furthermore, for each z ∈ DN there isfz ∈ B∞

v , such that |fz(z)| = 1v(z) . A weight v is called essential if there is a

constant C > 0 with

v(z) ≤ v(z) ≤ Cv(z) for every z ∈ DN .

For examples of essential weights and conditions when weights are essentialsee [2], [4] and [14]. We say that a weight v is radial in each component,if v(z1, . . . , zi−1, zi, zi+1, . . . , zN) = v(z1, . . . , zi−1, |zi|, zi+1, . . . , zN) for ev-ery z = (z1, . . . , zN) ∈ DN and every 1 ≤ i ≤ N . Every weight v whichis non-increasing in each component i with respect to |zi| and such thatlim|zi|→1 v(z) = 0 for every 1 ≤ i ≤ N is called typical in each compo-nent. In the sequel every radial weight is assumed to be non-increasing. Forweights v which are typical in each component the polynomials lie dense inH0

v . Moreover a weight v which is radial in each component satisfies (L1) ineach component if for every 1 ≤ i ≤ N and every z ∈ DN the following holds

(L1) infk∈N

v(z1, . . . , zi−1, 1 − 2−k−1, zi+1, . . . , zN )v(z1, . . . , zi−1, 1 − 2−k, zi+1, . . . , zN)

> 0

Vol. 7 (2010) Eigenvalues and Spectra of Composition Operators 567

3. Results

The results in this section were obtained in [5] in case N = 1. In this sectionwe assume that ϕ(0) = 0, which implies the boundedness of the operatorCϕ : H∞

v → H∞v . First, we restrict our attention to a special setting.

Proposition 3.1. Let v be a weight on DN , N ≥ 1, which is typical in eachcomponent and ϕ : DN → DN be an analytic map such that ϕ(z1, . . . , zN ) =(ϕ1(z1), . . . , ϕN (zN )). Moreover, let ϕ(0) = 0 and 0 < |∂ϕl

∂zl(0)| < 1 for every

1 ≤ l ≤ N . We assume that ϕ is no automorphism. Then {∂ϕl

∂zl(0)n, 1 ≤ l ≤

N, n ∈ N0} ⊂ σH∞v

(Cϕ).

Proof. We use an argument of [10] to show that (zn1 +· · ·+zn

N) ∈ H∞v , n ∈ N0,

is not in the range of Cϕ − ∂ϕl

∂zl(0)nI on H∞

v . Let us assume that f ∈ H∞v

and f(ϕ(z)) − f(z) = N for every z ∈ DN . Hence 0 = f(ϕ(0)) − f(0) =f(0) − f(0) = N , which is a contraction. Finally, 1 ∈ σH∞

v(Cϕ).

Next, fix 1 ≤ l ≤ N and let us suppose that f ∈ H∞v and f(ϕ(z)) −

∂ϕl

∂zl(0)f(z) = z1 + · · · + zN . Then obviously f(0) − ∂ϕl

∂zl(0)f(0) = 0. Since

0 < |∂ϕl

∂zl(0)| < 1 we have f(0) = 0. Now differentiation on both sides of the

equation yields

f ′(ϕ(z))ϕ′(z) − ∂ϕl

∂zl(0)f ′(z) = (1, . . . , 1).

With z = 0, this gives f ′(0)ϕ′(0) − ∂ϕl

∂zl(0)f ′(0) = (1, . . . , 1). Finally,

(∂f

∂z1(0), . . . ,

∂f

∂zN(0)

)⎛⎜⎜⎜⎝

∂ϕ1∂z1

(0) 0 · · · 00 ∂ϕ2

∂z2(0) · · · 0

· · · · · · · · · · · ·0 · · · 0 ∂ϕN

∂zN(0)

⎞⎟⎟⎟⎠

−∂ϕl

∂zl(0)

(∂f

∂z1(0), . . . ,

∂f

∂zN(0)

)= (1, · · · , 1).

Thus, we get 0 = ∂f∂zl

(0)∂ϕl

∂zl(0) − ∂ϕl

∂zl(0) ∂f

∂zl(0) = 1, which is a contradiction.

In case n = 2, we suppose f ∈ H∞v and f(ϕ(z))−∂ϕl

∂zl(0)2f(z) = (z2

1+· · ·+z2N).

With z = 0 f(0) − ∂ϕl

∂zl(0)2f(0) = 0 and thus f(0) = 0. Now, differenti-

ation on both sides of the equation yields f ′(ϕ(z))ϕ′(z) − ∂ϕl

∂zl(0)2f(z) =

(2z1, . . . , 2zN). Putting in z = 0 gives

(∂f

∂z1(0), . . . ,

∂f

∂zN(0)

) ⎛⎜⎜⎜⎝

∂ϕ1∂z1

(0) 0 · · · 00 ∂ϕ2

∂z2(0) 0 0

· · · · · · · · · · · ·0 0 · · · ∂ϕN

∂zN(0)

⎞⎟⎟⎟⎠

−∂ϕl

∂zl(0)2

(∂f

∂z1(0), . . . ,

∂f

∂zN(0)

)= (0, . . . , 0).

568 Elke Wolf Mediterr. J. Math.

Hence ∂ϕl

∂zl(0) ∂f

∂zl(0) − ∂ϕl

∂zl(0)2 ∂f

∂zl(0) = 0 and thus ∂f

∂zl(0) = 0. Again by

differentiation

f ′′(ϕ(z))ϕ′(z)2 + f ′(ϕ(z))ϕ′′(z)− ∂ϕl

∂zl(0)2f ′′(z) =

⎛⎜⎜⎝

2 0 · · · 00 2 · · · 0· · · · · · · · · · · ·0 · · · 0 2

⎞⎟⎟⎠.

Hence with z = 0⎛⎜⎜⎜⎜⎜⎝

∂2f

∂z21

(0) ∂2f∂z2∂z1

(0) · · · ∂2f∂zN ∂z1

(0)

· · · · · · · · · · · ·· · · · · · · · · · · ·

∂2f∂z1∂zN

(0) ∂2f∂z2∂zN

(0) · · · ∂2f

∂z2N

(0)

⎞⎟⎟⎟⎟⎟⎠

⎛⎜⎜⎜⎜⎜⎝

∂ϕ1∂z1

(0)2 0 · · · 0

0∂ϕ2∂z2

(0)2 0 0

· · · · · · · · · · · ·0 · · · 0

∂ϕN∂zN

(0)2

⎞⎟⎟⎟⎟⎟⎠

− ∂ϕl

∂zl

(0)2

⎛⎜⎜⎜⎜⎜⎜⎜⎝

∂2f

∂z21

(0) ∂2f∂z1∂z2

(0) · · · ∂2f∂z1∂zN

(0)

∂2f∂z2∂z1

(0) ∂2f

∂z22

(0) · · · ∂2f∂z2∂zN

(0)

· · · · · · · · · · · ·∂2f

∂z2∂zN(0) · · · · · · ∂2f

∂z2N

(0)

⎞⎟⎟⎟⎟⎟⎟⎟⎠

=

⎛⎜⎜⎝

2 0 · · · 00 2 · · · 0

· · · · · · · · · · · ·0 0 · · · 2

⎞⎟⎟⎠

Thus, 0 = ∂2f∂zl

(0)∂ϕl

∂zl(0)2 − ∂2f

∂zl(0)∂ϕl

∂zl(0)2 = 2, which is a contradiction.

For n > 2, suppose f ∈ H∞v and f(ϕ(z)) − ∂ϕl

∂zl(0)nf(z) = (zn

1 + · · · + znN).

By repeated differentiation on both sides of the equation we get for all k < n

that ∂kf∂zk

l

(0) = 0. Then for k = n and z = 0 we obtain the contradiction

0 = n!. This means that ∂ϕl

∂zl(0)n ∈ σH∞

v(Cϕ) for all n ≥ 0. �

Proposition 3.2. Let v be a weight on DN , N ≥ 1, which is typical in eachcomponent. Let ϕ : DN → DN be an analytic map such that ϕ(z1, . . . , zN ) =(ϕ1(z1), . . . , ϕN (zN )). Moreover, let ϕ(0) = 0 and 0 < |∂ϕl

∂zl(0)| < 1 for every

1 ≤ l ≤ N . We assume that ϕ is no automorphism. If λ is an eigenvalue ofCϕ, then λ ∈ {∂ϕl

∂zl(0)n, 1 ≤ l ≤ N, n ∈ N0}.

Proof. Let λ be an eigenvalue and f �= 0 be the corresponding eigenvector.Then we have

f(ϕ(z)) = λf(z) for every z ∈ DN .

Thus f(0) = λf(0). If f(0) �= 0, then we obtain λ = 1. If f(0) = 0, thendifferentiation on both sides of the equation yields

f ′(ϕ(z))ϕ′(z) = λf ′(z) for every z ∈ DN .

Thus, for z = 0

(∂f

∂z1(0), . . . ,

∂f

∂zN(0)

) ⎛⎜⎜⎜⎝

∂ϕ1∂z1

(0) 0 · · · 00 ∂ϕ2

∂z2(0) · · · 0

· · · · · · · · · · · ·0 · · · 0 ∂ϕN

∂zN(0)

⎞⎟⎟⎟⎠

= λ

(∂f

∂z1(0), . . . ,

∂f

∂zN(0)

).

Vol. 7 (2010) Eigenvalues and Spectra of Composition Operators 569

If ∂f∂zl

(0) �= 0, then λ = ∂ϕl

∂zl(0), 1 ≤ l ≤ N . If ∂f

∂zl(0) = 0, differentiation on

both sides of the equation yields

f ′(ϕ(z))ϕ′′(z) + f ′′(ϕ(z))ϕ′(z)2 = λf ′′(z) for every z ∈ DN .

Hence, for z = 0 we get f ′′(0)ϕ′(0)2 = λf ′′(0), i.e.⎛⎜⎜⎜⎜⎜⎜⎜⎝

∂2f

∂z21

(0) ∂2f∂z2∂z1

(0) · · · ∂2f∂zN ∂z1

(0)

∂2f∂z1∂z2

(0) ∂2f

∂z22

(0) · · · ∂2f∂zN ∂z2

(0)

· · · · · · · · · · · ·∂2f

∂z1∂zN(0) · · · · · · ∂2f

∂z2N

(0)

⎞⎟⎟⎟⎟⎟⎟⎟⎠

⎛⎜⎜⎜⎝

∂ϕ1∂z1

(0)2 0 · · · 0

0∂ϕ2∂z2

(0)2 · · · 0

0 · · · · · · ∂ϕN∂zN

(0)2

⎞⎟⎟⎟⎠

= λ

⎛⎜⎜⎜⎜⎜⎜⎜⎝

∂2f

∂z21

(0)∂2f

∂z2∂z1(0) · · · ∂2f

∂zN ∂z1(0)

∂2f∂z1∂z2

(0) ∂2f

∂z22

(0) · · · ∂2f∂zN ∂z2

(0)

· · · · · · · · · · · ·∂2f

∂z1∂zN(0) · · · · · · ∂2f

∂z2N

(0)

⎞⎟⎟⎟⎟⎟⎟⎟⎠

.

Now, if ∂2f∂z2

l(0) �= 0, then λ = ∂ϕl

∂zl(0)2, 1 ≤ l ≤ N . If ∂2f

∂zl(0) = 0, then we

proceed by differentiation until we find ∂kf∂zk

l

(0) �= 0 which must exist. Thiscompletes the proof. �

Now, we want to consider the problem in a more difficult setting. Forsimplicity we restrict ourselves to the bidisk D2, but all the methods alsowork for arbitrary DN , N ≥ 1.

We put An = ϕ′(0)n =

(a(n)11 a

(n)12

a(n)21 a

(n)22

)and

a(n) := a(n)11 +a

(n)22

2 ±√(

a(n)11 +a

(n)22

2

)2

− detAn for every n ∈ N.

Proposition 3.3. Let ϕ be an analytic self-map of D2, such that ϕ is noautomorphism and has fixed point 0. Let v be a weight on D2 which is typicalin each component. If λ is an eigenvalue of the composition operatorCϕ : H∞

v → H∞v , then

λ = 1 or λ = a(n)22 or λ = a

(n)11 +a

(n)22

2 ±√(

a(n)11 +a

(n)22

2

)2

− det An.

Proof. Let λ be an eigenvalue and f �= 0 be the corresponding eigenvector.Then we have

f(ϕ(z)) = λf(z) for every z ∈ D2.

Putting in z = 0 it follows that f(0) = λf(0). If f(0) �= 0, thenλ = 1. If f(0) = 0 then differentiation on both sides of the equation givesf ′(ϕ(z))ϕ′(z) = λf ′(z) for every z ∈ D2. Hence(

∂f

∂z1(0),

∂f

∂z2(0)

) (∂ϕ1∂z1

(0) ∂ϕ1∂z2

(0)∂ϕ2∂z1

(0) ∂ϕ2∂z2

(0)

)= λ

(∂f

∂z1(0),

∂f

∂z2(0)

)

570 Elke Wolf Mediterr. J. Math.

and∂f

∂z1(0)

∂ϕ1

∂z1(0) +

∂f

∂z2(0)

∂ϕ2

∂z1(0) = λ

∂f

∂z1(0)

∂f

∂z1(0)

∂ϕ1

∂z2(0) +

∂f

∂z2(0)

∂ϕ2

∂z2(0) = λ

∂f

∂z2(0).

Thus, if ∂f∂z1

(0) �= 0, we get

∂ϕ1

∂z1(0) +

∂f∂z2

(0)∂f∂z1

(0)∂ϕ2

∂z1(0) = λ

∂f∂z2

(0)∂f∂z1

(0)

(λ − ∂ϕ2

∂z2(0)

)=

∂ϕ1

∂z2(0).

It follows that either λ = ∂ϕ2∂z2

(0) or λ �= ∂ϕ2∂z2

(0) and

∂ϕ1

∂z1(0) +

∂f∂z2

(0)∂f∂z1

(0)∂ϕ2

∂z1(0) = λ

∂f∂z2

(0)∂f∂z1

(0)=

∂ϕ1∂z2

(0)

λ − ∂ϕ2∂z2

(0).

Finally, we get

λ2 −(

∂ϕ1

∂z1(0) +

∂ϕ2

∂z2(0)

)λ +

(∂ϕ1

∂z1(0)

∂ϕ2

∂z2(0) − ∂ϕ1

∂z2(0)

∂ϕ2

∂z1(0)

)= 0

and in this case the claim follows. If ∂f∂z1

(0) = 0 and ∂f∂z2

(0) �= 0, then

∂f

∂z2(0)

∂ϕ2

∂z1(0) = 0

∂f

∂z2(0)

∂ϕ2

∂z2(0) = λ

∂f

∂z2(0).

Hence λ = ∂ϕ2∂z2

(0) = 0. If ∂f∂z2

(0) = 0 we proceed by differentiation until

we find k, l ∈ N such that ∂kf∂lz1∂k−lz2

(0) �= 0 which must exist. The proof iscomplete. �

Proposition 3.4. Let v be a weight on D2 which is typical in each component.Let ϕ : D2 → D2 be an analytic map such that ϕ(0) = 0. Moreover we assumethat a(n) �= a

(n)22 and a(n) �= 0 for every n ∈ N. Then{

a(n); n ∈ N

}∪ {1} ⊂ σH∞

v(Cϕ).

Proof. Let us assume that f ∈ H∞v and f(ϕ(z))− f(z) = 1 for every z ∈ D2.

Hence 0 = f(ϕ(0)) − f(0) = f(0) − f(0) = 1, which is a contraction. Thus,1 ∈ σH∞

v(Cϕ).

Vol. 7 (2010) Eigenvalues and Spectra of Composition Operators 571

Next, let us suppose that f ∈ H∞v and f(ϕ(z))− a(1)f(z) = z1. Then we ob-

tain f(0)−a(1)f(0) = 0. Since a(1) �= 0, we have f(0) = 0. Now differentiationon both sides of the equation yields

f ′(ϕ(z))ϕ′(z) − a(1)f ′(z) = (1, 0).

With z = 0, we get f ′(0)ϕ′(0) − a(1)f ′(0) = (1, 0). Therefore

(∂f

∂z1(0),

∂f

∂z2(0)

) (∂ϕ1∂z1

(0) ∂ϕ1∂z2

(0)∂ϕ2∂z1

(0) ∂ϕ2∂z2

(0)

)− a(1)

(∂f

∂z1(0),

∂f

∂z2(0)

)= (1, 0).

Thus, we obtain

∂f

∂z1(0)

∂ϕ1

∂z1(0) +

∂f

∂z2(0)

∂ϕ2

∂z1(0) − a(1) ∂f

∂z1(0) = 1

∂f

∂z1(0)

∂ϕ1

∂z2(0) +

∂f

∂z2(0)

∂ϕ2

∂z2(0) − a(1) ∂f

∂z2(0) = 0.

If ∂f∂z1

(0) = ∂f∂z2

(0) = 0, then the first equation yields 0 = 1 which is acontradiction.If ∂f

∂z1(0) �= 0, we get

∂ϕ1

∂z1(0) +

∂ϕ2

∂z1(0)

∂f∂z2

(0)∂f∂z1

(0)− a(1) =

1∂f∂z1

(0)∂f∂z2

(0)∂f∂z1

(0)=

∂ϕ1∂z2

(0)

a(1) − ∂ϕ2∂z2

(0)

and hence

0 = a(1)2 −(

∂ϕ1

∂z1(0) +

∂ϕ2

∂z2(0)

)a(1) +

∂ϕ1

∂z1(0)

∂ϕ2

∂z2(0) − ∂ϕ1

∂z2(0)

∂ϕ2

∂z1(0)

=a(1) − ∂ϕ2

∂z2(0)

∂f∂z1

(0),

which is a contradiction. If ∂f∂z2

(0) �= 0, then the equations above give

∂f

∂z2(0)

∂ϕ2

∂z1(0) = 1(

a(1) − ∂ϕ2

∂z2(0)

)∂f

∂z2(0) = 0.

From the second of these equations we can deduce that ∂f∂z2

(0) = 0 whichleads to a contradiction.For n > 1, suppose f ∈ H∞

v and f(ϕ(z)) − a(n)f(z) = zn1 . By repeated

differentiation on both sides of the equation we get the desired result. �

572 Elke Wolf Mediterr. J. Math.

References

[1] R. Aron, M. Lindstrom, Spectra of weighted composition operators on weightedBanach spaces of analytic functions, Israel J. Math. 141 (2004), 263 - 276.

[2] K.D. Bierstedt, J. Bonet, J. Taskinen, Associated weights and spaces of holo-morphic functions, Studia Math. 127 (1998), 137-168.

[3] J. Bonet, P. Domanski, M. Lindstrom, Essential norm and weak compactness ofcomposition operators on weighted Banach spaces of analytic functions, Canad.Math. Bull. 42, no. 2, (1999), 139-148.

[4] J. Bonet, P. Domanski, M. Lindstrom, J. Taskinen, Composition operators be-tween weighted Banach spaces of analytic functions, J. Austral. Math. Soc. (Se-rie A) 64 (1998), 101-118.

[5] J. Bonet, P. Galindo, M. Lindstrom, Spectra and essential radii of compositionoperators on weighted Banach spaces of analytic functions, J. Math. Anal. Appl.340 (2008), no. 2, 884-891.

[6] M.D. Contreras, A.G. Hernandez-Dıaz, Weighted composition operators inweighted Banach spaces of analytic functions, J. Austral. Math. Soc. (SeriesA) 69 (2000), 41-60.

[7] C. Cowen, B. MacCluer, Composition Operators on Spaces of Analytic Func-tions, CRC Press, Boca Raton, 1995.

[8] C. Cowen, B. MacCluer, Spectra of some composition operators, J. Funct. Anal.125 (1994), 223-251.

[9] H.G. Heuser, Functional Analysis, John Wiley and Sons, 1982.[10] H. Kamowitz, The spectra of endomorphisms of the disc algebra, Pac. J. Math.

46, 433-440.[11] W. Lusky, On weighted spaces of harmonic and holomorphic functions, J. Lon-

don Math. Soc. 51 (1995), 309-320.[12] B. MacCluer, K. Saxe, Spectra of composition operators on the Bloch and

Bergman spaces, Israel J. Math. 128 (2002), 324-354.[13] A. Montes-Rodrıguez, The essential norm of a composition operator on Bloch

spaces, Pacific J. Math. 188 (1999), 339-351.[14] A. Montes-Rodrıguez, Weighted composition operators on weighted Banach

spaces of analytic functions, J. London Math. Soc. 61 (2000), 872-884.[15] J.H. Shapiro, Composition Operators and Classical Function Theory, Springer,

1993.[16] E. Wolf, Differences of composition operators between weighted Banach spaces

of holomorphic functions on the unit polydisk, Result Math. 51 (2008), 361-372.[17] L. Zheng, The essential norms and spectra of composition operators on H∞,

Pac. J. Math.203 (2003), no. 2, 503-510.

Elke WolfMathematical Institute, University of Paderborn, D-33095 Paderborn, Germanye-mail: [email protected]

Received: December 19, 2008.Revised: November 2, 2009.Accepted: November 5, 2009.