0 views

Uploaded by Ravindra

THIS is a good book

save

- BANACH SPACES AND RAMSEY THEORY: SOME OPEN PROBLEMS
- euclid.pjm.1102708251
- S26
- Topology Course Lexture Notes.ps
- smoothing-continuous-flows-on-two-manifolds.pdf
- SINGULAR SETS AND MAXIMAL TOPOLOGIES
- An Optimization Approach to Kinetic Model Reduction For
- Banach Contraction Principle
- G02107476.pdf
- Lect-20-R
- e_4
- LectureNotes8U
- pdf
- 10. IJAMSS - On Convergence Properties of Szasz Type Positive Linear Operator
- S. V. Astashkin- Disjointly Strictly Singular Inclusions of Symmetric Spaces
- zero lesson plan
- 1--Maths - IJMCAR - COMMON - Vishal Gupta - Paid
- Equilateral Flat Unit Folding Instructions
- qp2
- Sophia
- mathgen-87198825
- arc
- 1MJM080
- Geometria Plana and Space - Aula 11-524
- geometria violino
- Comcor
- A History of Topology
- Digital_Project_v1r4_quick_start_guide.pdf
- ss_maths_f3c_2010
- Katarzyna Pietruska-Paluba- The Reflected Brownian Motion on the Sierpinski Gasket
- Prachin Bharat Ka Itihaas OLD NCERT Ancient History of INDIA in HINDI ]_2
- BSER_Raj_Adhyan-1.pdf.pdf
- Rivers of India
- Prachin Bharat Ka Itihaas OLD NCERT Ancient History of INDIA in HINDI
- Ch6_presN
- Prachin Bharat ka Itihaas OLD NCERT Ancient History of INDIA in HINDI ]_2.pdf

You are on page 1of 87

American Mathematical Society

Number 974

Composition Operators

on Hardy-Orlicz Spaces

Pascal Lefèvre

Daniel Li

Hervé Queffélec

Luis Rodrı́guez-Piazza

September 2010 • Volume 207 • Number 974 (fourth of 5 numbers) • ISSN 0065-9266

**American Mathematical Society
**

Number 974

Composition Operators

on Hardy-Orlicz Spaces

Pascal Lefèvre

Daniel Li

Hervé Queffélec

Luis Rodrı́guez-Piazza

**September 2010 • Volume 207 • Number 974 (fourth of 5 numbers) • ISSN 0065-9266
**

Library of Congress Cataloging-in-Publication Data

Composition operators on Hardy-Orlicz spaces / Pascal Lefèvre ... [et al.].

p. cm. — (Memoirs of the American Mathematical Society, ISSN 0065-9266 ; no. 974)

“Volume 207, number 974 (fourth of 5 numbers).”

Includes bibliographical references.

ISBN 978-0-8218-4637-7 (alk. paper)

1. Composition operators. 2. Hardy spaces. I. Lefèvre, Pascal, 1969-.

QA329.2.C654 2010

515.7246—dc22 2010022778

**Memoirs of the American Mathematical Society
**

This journal is devoted entirely to research in pure and applied mathematics.

Publisher Item Identiﬁer. The Publisher Item Identiﬁer (PII) appears as a footnote on

the Abstract page of each article. This alphanumeric string of characters uniquely identiﬁes each

article and can be used for future cataloguing, searching, and electronic retrieval.

Subscription information. Beginning with the January 2010 issue, Memoirs is accessi-

ble from www.ams.org/journals. The 2010 subscription begins with volume 203 and consists of

six mailings, each containing one or more numbers. Subscription prices are as follows: for paper

delivery, US$709 list, US$567 institutional member; for electronic delivery, US$638 list, US$510 in-

stitutional member. Upon request, subscribers to paper delivery of this journal are also entitled to

receive electronic delivery. If ordering the paper version, subscribers outside the United States and

India must pay a postage surcharge of US$65; subscribers in India must pay a postage surcharge of

US$95. Expedited delivery to destinations in North America US$57; elsewhere US$160. Subscrip-

tion renewals are subject to late fees. See www.ams.org/customers/macs-faq.html#journal for

more information. Each number may be ordered separately; please specify number when ordering

an individual number.

Back number information. For back issues see www.ams.org/bookstore.

Subscriptions and orders should be addressed to the American Mathematical Society, P. O.

Box 845904, Boston, MA 02284-5904 USA. All orders must be accompanied by payment. Other

correspondence should be addressed to 201 Charles Street, Providence, RI 02904-2294 USA.

Copying and reprinting. Individual readers of this publication, and nonproﬁt libraries

acting for them, are permitted to make fair use of the material, such as to copy a chapter for use

in teaching or research. Permission is granted to quote brief passages from this publication in

reviews, provided the customary acknowledgment of the source is given.

Republication, systematic copying, or multiple reproduction of any material in this publication

is permitted only under license from the American Mathematical Society. Requests for such

permission should be addressed to the Acquisitions Department, American Mathematical Society,

201 Charles Street, Providence, Rhode Island 02904-2294 USA. Requests can also be made by

e-mail to reprint-permission@ams.org.

**Memoirs of the American Mathematical Society (ISSN 0065-9266) is published bimonthly (each
**

volume consisting usually of more than one number) by the American Mathematical Society at

201 Charles Street, Providence, RI 02904-2294 USA. Periodicals postage paid at Providence, RI.

Postmaster: Send address changes to Memoirs, American Mathematical Society, 201 Charles

Street, Providence, RI 02904-2294 USA.

c 2010 by the American Mathematical Society. All rights reserved.

Copyright of individual articles may revert to the public domain 28 years

after publication. Contact the AMS for copyright status of individual articles.

This publication is indexed in Science Citation Index R , SciSearch

R , Research Alert

R,

**CompuMath Citation Index R , Current Contents
**

R /Physical, Chemical & Earth Sciences.

**Printed in the United States of America.
**

∞ The paper used in this book is acid-free and falls within the guidelines

established to ensure permanence and durability.

Visit the AMS home page at http://www.ams.org/

10 9 8 7 6 5 4 3 2 1 15 14 13 12 11 10

Contents

Chapter 1. Introduction 1

Chapter 2. Notation 5

1. Growth conditions 5

2. Some speciﬁc functions 7

Chapter 3. Composition operators on Hardy-Orlicz spaces 9

1. Hardy-Orlicz spaces 9

2. Preliminary results 13

3. Composition operators 15

4. Order bounded composition operators 17

5. Weakly compact composition operators 21

6. p-summing operators 26

Chapter 4. Carleson measures 33

1. Introduction 33

2. Compactness on H Ψ versus compactness on H 2 36

3. General measures 38

4. Characterization of the compactness of composition operators 55

Chapter 5. Bergman spaces 63

1. Bergman-Orlicz spaces 63

2. Compact composition operators on Bergman-Orlicz spaces 66

Acknowledgement 71

References 73

iii

Abstract

**We investigate composition operators on Hardy-Orlicz spaces when the Orlicz
**

function Ψ grows rapidly: compactness, weak compactness, to be p-summing, order

bounded, . . . , and show how these notions behave according to the growth of Ψ. We

introduce an adapted version of Carleson measure. We construct various examples

showing that our results are essentially sharp. In the last part, we study the case

of Bergman-Orlicz spaces.

**Received by the editor June 26, 2006.
**

Article electronically published on March 17, 2010; S 0065-9266(10)00580-6.

2000 Mathematics Subject Classiﬁcation. Primary: 47B33, 46E30.

Key words and phrases. Bergman-Orlicz space, Carleson measure, composition operator,

Hardy-Orlicz space.

2010

c American Mathematical Society

v

CHAPTER 1

Introduction

**Composition operators on the classical Hardy spaces H p have been widely
**

studied (see [39], and [10], and references therein; see also [18] and [19], and [6],

[9], [17], [31], [36], [40], [41] for some more recent works), but it seems that one

has not paid much attention to the Hardy-Orlicz spaces (in [43] and [44], J.-O.

Strömberg studied Hardy-Orlicz spaces in the case when the Orlicz function Ψ

increases smoothly; see also [29] for composition operators). We shall investigate

what happens when the Orlicz function grows more rapidly than a power function.

Recall that, given an analytic self-map φ : D → D of the unit disk D, the

composition operator associated to φ is the map Cφ : f → f ◦ φ. This map may

operate on various Banach spaces X of analytic functions on D (Hardy spaces,

Bergman spaces, . . . , and their weighted versions (see [45] for instance), Bloch

spaces B and B0 , BM OA and V M OA, Dirichlet spaces (see [1]), or some more

general spaces as Nevanlinna or Smirnov classes: [7], [8], [16], [20]; see also [2],

[3], [13] for composition operators on H p spaces of Dirichlet series, though they

are not induced by an analytic self-map of D). The goal is to link properties of the

composition operator Cφ : X → X (compactness, strong or weak, for example) to

properties of the symbol φ (essentially its behaviour near the frontier of D). For

that study, one can roughly speaking (see [10], Chapter 4, though their notions are

diﬀerent from ours) distinguish two kind of spaces.

1) The small spaces X; those spaces are in a sense close to the Hardy space

H ∞ : the compactness of Cφ : X → X is very restrictive and it imposes severe

restrictions on φ. For example, if X = H ∞ , a theorem of H.J. Schwartz ([37])

implies that Cφ : H ∞ → H ∞ is compact if and only if φ∞ < 1 (weakly compact

suﬃces: see [23]), which in turn implies reinforced compactness properties for Cφ .

For example, Cφ : H ∞ → H ∞ is nuclear and 1-summing as soon as it is compact.

2) The large spaces X; those spaces are in a sense close to the Hardy space H 1 :

the compactness of Cφ : X → X can take place fairly often, and in general implies

no self-improvement. For example, for X = H 2 , Cφ : H 2 → H 2 can be compact

without being Hilbert-Schmidt, even if φ is injective ([42], Theorem 6.2). Another

formulation (which lends better to generalizations in the non-Hilbertian case) is

that Cφ can be non-order bounded (see [15], and our Chapter 3) and yet compact.

In this paper, we shall rather be on the small space side, since we shall work

in spaces associated to a very large Orlicz function Ψ (typically: Ψ(x) = Ψ2 (x) =

2

ex − 1), and the previous situation will not take place: our operators will be e.g.

order-bounded as soon as they are (weakly) compact, even if the situation is not

so extreme as for H ∞ . However, for slightly smaller Orlicz functions (for instance

2

Ψ(x) = exp log(x + 1) − 1), the situation is closer to the H 2 case: the compo-

sition operators may be compact on H Ψ , but not order-bounded (Theorem 4.22).

1

2 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

**This paper is divided into ﬁve parts. Chapter 1 is this Introduction. In Chapter
**

2, which is essentially notational, we recall some more or less standard facts on

Orlicz functions Ψ, on associated Orlicz spaces LΨ , and the “little” Orlicz space

M Ψ , and their Banachic properties, associated with various slow growth conditions

(indicated by subscripts: Δ1 , Δ2 , . . . ) or fast growth conditions (indicated by

superscripts: Δ0 , Δ1 , Δ2 , . . . ).

In Chapter 3, we introduce the Hardy-Orlicz space H Ψ , and its (or his) “little

brother” HM Ψ . These spaces have already been studied (see [14], [34]), but rather

for slowly growing functions Ψ (having Δ2 most of the time), and their deﬁnition

is not so clearly outlined, so we give a detailed exposition of the equivalence of the

two natural deﬁnitions that one has in mind (if one wants to extend the case of

Hardy spaces H p associated to Ψ(x) = xp ), as well as of the automatic boundedness

(through the Littlewood subordination principle and the case of inner self-maps of

the disk) of composition operators on those spaces. Two of the main theorems

are Theorem 3.24 and Theorem 3.27. Roughly speaking, Theorem 3.24 says the

following: if Ψ is very fast growing (having Δ2 more precisely), H Ψ is a small

space, the (weak) compactness of Cφ is very restrictive, and even if the situation

is not so extreme as for H ∞ (φ∞ < 1), φ has to tend to the boundary very

slowly, and Cφ is automatically order-bounded into M Ψ . However, Theorem 3.27

shows the limits of this self-improvement: Cφ may be order-bounded into M Ψ (and

hence compact), but p-summing for no ﬁnite p. We also show that, when Ψ has

Δ2 growth, there are always symbols φ inducing compact composition operators on

H 2 (even Hilbert-Schmidt), but not compact on H Ψ .

Chapter 4 is devoted to the use of Carleson measures. The usefulness of those

measures in the study of composition operators is well-known (see [11], [10], [5])

and, to our knowledge, ﬁrst explicitly used for compactness in [30]. In particular,

we recall the following necessary and suﬃcient condition for Cφ : H 2 → H 2 to be

compact: if h > 0 and w ∈ ∂D, consider the Carleson window :

W (w, h) = {z ∈ D ; |z| ≥ 1 − h and | arg(z w̄)| ≤ h}.

If φ is an analytic self-map of D with boundary values φ∗ , and μφ = φ∗ (m) denotes

the image under φ∗ of the normalized Lebesgue measure (Haar measure) on T = ∂D,

the measure μφ is always a Carleson measure, i.e.:

sup μφ W (w, h) = O (h).

w∈∂D

Now, we can state:

Theorem 1.1 (B. MacCluer [30]). The composition operator Cφ : H 2 → H 2 is

compact if and only if μφ satisﬁes the “ little-oh” condition, i.e. if and only if:

(1.1) sup μφ W (w, h) = o (h) as h → 0.

w∈∂D

**There is another famous necessary and suﬃcient compactness condition, due
**

to J. Shapiro ([38]): Let us denote by Nφ the Nevanlinna counting function of φ,

i.e.: 1

log |z| if w = φ(0) and w ∈ φ(D)

Nφ (w) = φ(z)=w

0 if w = φ(0) or w ∈/ φ(D).

By Littlewood’s inequality, one always has (see [10], page 33):

Nφ (w) = O (1 − |w|).

1. INTRODUCTION 3

Now, Shapiro’s Theorem reads:

**Theorem 1.2 (J. Shapiro [38]). The composition operator Cφ : H 2 → H 2 is
**

compact if and only if Nφ satisﬁes the “ little-oh” condition, i.e.:

(1.2) Nφ (w) = o (1 − |w|) as |w| → 1− .

**Theorem 1.2 is very elegant, and probably more “popular” than Theorem 1.1.
**

Yet, it is diﬃcult to apply because the assumption (1.2) is diﬃcult to check. Here,

we shall appeal to Theorem 1.1 to prove (Theorem 4.1) that the compactness of

Cφ : H 2 → H 2 cannot be read on |φ∗ | when φ is not ﬁnitely valent; more precisely,

there are two analytic self-maps φ1 and φ2 : D → D such that: |φ∗1 | = |φ∗2 | m-a.e.,

but Cφ1 : H 2 → H 2 is not compact, though Cφ2 : H 2 → H 2 is compact.

We show then that every composition operator which is compact on H Ψ is

necessarily compact on H p for all p < ∞. However, there exist (see above, or

Chapter 3) symbols φ inducing compact composition operators on H 2 but which

are not compact on H Ψ , when Ψ has Δ2 growth. Hence condition (1.1) does not

suﬃce to characterize the compact composition operators on H Ψ . We have to

replace Carleson measures and condition (1.1) by what we may call “Ψ-Carleson

measures”, and an adaptated “little-oh” condition, which allows us to characterize

compactness for composition operators. It follows that if Ψ ∈ Δ0 , then the weak

compactness of Cφ : H Ψ → H Ψ implies its compactness.

We also show that the above example φ2 induces, for an Orlicz function Ψ which

does not satisfy Δ2 , but which satisﬁes Δ1 , a composition operator on H Ψ which

is compact, but not order bounded into M Ψ (T) (Theorem 4.22), showing that the

assumption that Ψ ∈ Δ2 in Theorem 3.24 is not only a technical assumption.

In Chapter 5, we introduce the Bergman-Orlicz spaces. Let us remind that, in

the Hilbertian case, the study of compactness of composition operators is simpler

for the Bergman space B 2 than for the Hardy space H 2 . For example, we have the

following:

Theorem 1.3 (see [39]).

i) Cφ : B 2 → B 2 is compact if and only if

1 − |φ(z)|

(1.3) lim = +∞.

|z|→1− 1 − |z|

**ii) (1.3) is always necessary for Cφ : H 2 → H 2 to be compact, and it is suﬃcient
**

when φ is injective, or only ﬁnitely valent.

iii) There are Blaschke products φ satisfying (1.3) for which Cφ : H 2 → H 2 is

(in an obvious manner) non-compact.

**We perform here a similar study for the Bergman-Orlicz space B Ψ , and com-
**

pare the situation with that of the Hardy-Orlicz space H Ψ . We are naturally led

to a reinforcement of (1.3) under the form:

1

Ψ−1

(1 − |φ(a)|)2

(1.4) −→ 0

−1

1 |a|→1−

Ψ

(1 − |a|)2

4 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

**(always necessary, and suﬃcient when Ψ is Δ2 ), which reads, in the case Ψ(x) =
**

2

Ψ2 (x) = ex − 1:

(1.5) 1 − |φ(z)| ≥ cε (1 − |z|)ε for all ε > 0.

In [39], the construction of a Blaschke product satisfying (1.3) is fairly delicate, and

appeals to Frostman’s Lemma and Julia-Caratheodory’s Theorem on non-angular

derivatives at the boundary. Here, we can no longer use these tools for the rein-

forcement (1.5), so we do make a direct construction, using the Parseval formula

for ﬁnite groups. In passing, the construction gives a simpler proof for (1.3). Oth-

erwise, the theorem which we obtain is similar to Shapiro’s one, if one ignores some

technical diﬃculties due to the non-separability of B Ψ : we have to “transit” by the

smaller Bergman-Morse-Transue space BM Ψ , which is the closure of H ∞ in B Ψ ,

is separable, and has B Ψ as its bidual.

CHAPTER 2

Notation

**Let D be the open unit disk of the complex plane, that is the set of complex
**

numbers with modulus strictly less than 1, and T the unit circle, i.e. the set of

complex numbers with modulus 1.

We shall consider in this paper Orlicz spaces deﬁned on a probability space

(Ω, P), which will be the unit circle T, with its (normalized) Haar measure m

(most often identiﬁed with the normalized Lebesgue measure dx/2π on the interval

[0, 2π]), or the open unit disk D, provided with the normalized area measure A .

By an Orlicz function, we shall understand that Ψ : [0, ∞] → [0, ∞] is a non-

decreasing convex function such that Ψ(0) = 0 and Ψ(∞) = ∞. To avoid patholo-

gies, we shall assume that we work with an Orlicz function Ψ having the following

additional properties: Ψ is continuous at 0, strictly convex (hence increasing), and

such that

Ψ(x)

−→ ∞.

x x→∞

This is essentially to exclude the case of Ψ(x) = ax.

If Ψ

is the left (or instead, if one prefers, the right) derivative of Ψ, one has

x

Ψ(x) = 0 Ψ (t) dt for every x > 0.

The Orlicz space LΨ (Ω) is the space of all (equivalence classes of) measurable

functions f : Ω → C for which there is a constant C > 0 such that

|f (t)|

Ψ dP(t) < +∞

Ω C

and then f Ψ (the Luxemburg norm) is the inﬁmum of all possible constants C

such that this integral is ≤ 1. The Morse-Transue space M Ψ (Ω) is the subspace

generated by L∞ (Ω), or, equivalently, the subspace of all functions f for which the

above integral is ﬁnite for all C > 0.

To every Orlicz function is associated the complementary Orlicz function Φ =

Ψ∗ : [0, ∞] → [0, ∞] deﬁned by:

Φ(x) = sup xy − Ψ(y) ,

y≥0

**The extra assumptions on Ψ ensure that Φ is itself strictly convex.
**

When Φ satisﬁes the Δ2 condition (see the deﬁnition below), LΨ is (isomorphi-

cally, if LΦ is itself normed by the Luxemburg norm) the dual space of LΦ , which

is, in turn, the dual of M Ψ .

1. Growth conditions

We shall have to use various growth conditions for the Orlicz function Ψ. These

conditions are usually denoted as Δ-conditions. Our interest is in Orlicz functions

which have a somewhat fast growth. Usually, some of these conditions are deﬁned

5

6 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

**through a moderate growth condition on the complementary function Φ of Ψ, and
**

the condition Δ for the Orlicz function is translated as a ∇-condition for the com-

plementary function. So we shall distinguish between moderate growth conditions,

that we shall deﬁne for the complementary Orlicz function, and fast growth condi-

tions. To emphasize this distinction, we shall denote, sometimes in changing the

usual notation (see [22, 33]), the moderate growth conditions with a subscript, and

the fast growth conditions with a superscript.

Moderate growth conditions

• The Orlicz function Φ satiﬁes the Δ1 -condition (Φ ∈ Δ1 ) if, for some

constant c > 0, one has:

Φ(xy) ≤ c Φ(x)Φ(y)

for x, y large enough.

This is equivalent to say that

Φ(axy) ≤ Φ(x)Φ(y)

for some constant a > 0 and x, y large enough.

This condition is usually denoted by Δ (see [33], page 28).

• Φ satisﬁes the Δ2 -condition (Φ ∈ Δ2 ) if

Φ(2x) ≤ K Φ(x)

for some constant K > 1 and x large enough.

One has:

Φ ∈ Δ1 ⇒ Φ ∈ Δ2 .

Fast growth conditions

• The Orlicz function Ψ satisﬁes the Δ0 -condition (Ψ ∈ Δ0 ) if (see [25]),

for some β > 1:

Ψ(βx)

lim = +∞.

x→+∞ Ψ(x)

A typical example is Ψ(x) = exp log(x + 2) log log(x + 2) − 2log log 2 ;

another is Ψ(x) = exp log(x + 1)3/2 − 1.

• The Orlicz function Ψ satisﬁes the Δ1 -condition (Ψ ∈ Δ1 ) if there is some

β > 1 such that:

xΨ(x) ≤ Ψ(βx)

for x large enough.

Note that this latter condition is usually written as Δ3 -condition, with a subscript

(see [33], §2.5). This notation ﬁts better with our convention, and the superscript

1 agrees with the fact that this Δ1 -condition is between the Δ0 -condition and the

following Δ2 -condition. Ψ ∈ Δ1 implies that

Ψ(x) ≥ exp α (log x)2

for some α > 0 and x large enough (see [33], Proposition 2, page 37). A typical

2

example is Ψ(x) = e(log(x+1)) − 1.

2. SOME SPECIFIC FUNCTIONS 7

**• The Orlicz function Ψ : [0, ∞) → [0, ∞) is said to satisfy the Δ2 -condition
**

(Ψ ∈ Δ2 ) if there exists some α > 1 such that:

Ψ(x)2 ≤ Ψ(αx)

for x large enough.

This implies that

Ψ(x) ≥ exp(xα )

for some α > 0 and x large enough ([33], Proposition 6, page 40). A typical example

2

is Ψ(x) = Ψ2 (x) = ex − 1.

Conditions of regularity

• The Orlicz function Ψ satisﬁes the ∇2 -condition (Ψ ∈ ∇2 ) if its comple-

mentary function Φ satisﬁes the Δ2 -condition.

This is equivalent to say that for some constant β > 1 and some x0 > 0,

one has Ψ(βx) ≥ 2β Ψ(x) for x ≥ x0 , and that implies that Ψ(x)

x −→ ∞.

x→∞

In particular, this excludes the case LΨ = L1 .

• The Orlicz function Ψ satisﬁes the ∇1 -condition (Ψ ∈ ∇1 ) if its comple-

mentary function Φ satisﬁes the Δ1 -condition.

This is equivalent to say that

Ψ(x)Ψ(y) ≤ Ψ(bxy)

for some constant b > 0 and x, y large enough.

All power functions Ψ(x) = xp satisfy ∇1 , but Ψ(x) = xp log(x + 1) does

not.

One has (see [33], page 43):

+3 Ψ ∈ Δ0

7? ∈ Δ

1

Ψ FFFFF

x

xxxxx FFFF

x

xx FFF

xxxxx &

Ψ ∈ Δ SS

2

:2 Ψ ∈ ∇2

SSS

SSSS lllll

ll

SS %- lll

Ψ ∈ ∇1

But Δ1 does not imply ∇1 . That ∇1 does not even imply Δ0 is clear since any

power function Ψ(x) = xp (p ≥ 1) is in ∇1 .

**2. Some speciﬁc functions
**

In this paper, we shall make a repeated use of the following functions:

• If Ψ is an Orlicz function, we set, for every K > 0:

(2.1) χK (x) = Ψ KΨ−1 (x) , x > 0.

For example, if Ψ(x) = ex − 1, then Ψ−1 (x) = log(1 + x), and χK (x) =

(1 + x)K − 1.

Note that:

χβ (u)

– Ψ ∈ Δ0 means that −→ +∞, for some β > 1.

u u→∞

– Ψ ∈ Δ means that χβ (u) ≥ uΨ−1 (u) for u large enough, for some

1

β > 1.

– Ψ ∈ Δ2 means that χα (u) ≥ u2 for u large enough, for some α > 1.

8 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

**– Ψ ∈ ∇1 means that χA (u) ≥ (Ψ(A)/b) u for u large enough and for
**

every A large enough, for some b > 0.

• For |a| = 1 and 0 ≤ r < 1, ua,r is the function deﬁned on the unit disk D

by:

1 − r
2

(2.2) ua,r (z) = , |z| < 1.

1 − ārz

Note that ua,r ∞ = 1 and ua,r 1 ≤ 1 − r.

CHAPTER 3

Composition operators on Hardy-Orlicz spaces

**1. Hardy-Orlicz spaces
**

It is well-known that the classical H p spaces (1 ≤ p ≤ ∞) can be deﬁned in

two equivalent ways:

1) H p is the space of analytic functions f : D → C for which, setting fr (t) =

f (reit ):

f H p = sup fr p

0≤r<1

**is ﬁnite (recall that the numbers fr p increase with r). When f ∈ H p , the Fatou-
**

Riesz Theorem asserts that the boundary limits f ∗ (t) = limr→1− fr (t) exist almost

everywhere and f H p = f ∗ p . One has f ∗ ∈ Lp ([0, 2π]), and its Fourier coeﬃ-

cients f∗ (n) vanish for n < 0.

2) Conversely, for every function g ∈ Lp ([0, 2π]) whose Fourier coeﬃcients

vanish nfor n < p0, the analytic extension P [g] : D →∗ C deﬁned by P [g](z) =

ĝ(n)

n≥0 ĝ(n)z is in H and g is the boundary limit (P [g]) of P [g].

Hardy-Orlicz spaces H Ψ are deﬁned in a similar way. However, we did not ﬁnd

very satisfactory references, and, though the reasonings are essentially the same

as in the classical case, the lack of homogeneity of Ψ and the presence of the two

spaces M Ψ and LΨ gives proofs which are not so obvious and therefore we shall

give some details.

It should be noted that our deﬁnition is not exactly the same as the one given

in [33], § 9.1.

We shall begin with the following proposition.

Proposition 3.1. Let f : D → C be an analytic function. For every Orlicz

function Ψ, the following assertions are equivalent:

1) sup0≤r<1 fr Ψ < +∞, where fr (t) = f (reit );

2) there exists f ∗ ∈ LΨ ([0, 2π]) such that f∗ (n) = 0 for n < 0 and for which

f (z) = n≥0 f∗ (n)z n , z ∈ D.

When these conditions are satisﬁed, one has f ∗ Ψ = sup0≤r<1 fr Ψ .

Let us note that, since

Ψ is convex and increasing, Ψ(a|f |) is subharmonic on

D, and hence the numbers T Ψ(a|fr |) dm increase with r, for every a > 0.

This proposition leads to the following deﬁnition.

Definition 3.2. Given an Orlicz function Ψ, the Hardy-Orlicz space H Ψ as-

sociated to Ψ is the space of analytic functions f : D → C such that one of the

equivalent conditions of the above proposition is satisﬁed. The norm of f is deﬁned

by f H Ψ = f ∗ Ψ . We shall denote by HM Ψ the Hardy-Morse-Transue space,

i.e. the subspace {f ∈ H Ψ ; f ∗ ∈ M Ψ (T)}.

9

10 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

**In the sequel, we shall make no distinction between f and f ∗ , unless there may
**

be some ambiguity, and shall write f instead of f ∗ for the boundary limit. Hence

we shall allow ourselves to write f (eit ) instead of f ∗ (t), or even f ∗ (eit ). Moreover,

we shall write f Ψ instead of f H Ψ .

It follows that H Ψ becomes a subspace of LΨ (T) and HM Ψ = H Ψ ∩ M Ψ (T).

These two spaces are closed (hence Banach spaces) since Proposition 3.1 gives:

Corollary 3.3. H Ψ is weak-star closed in LΨ = (M Φ )∗ . When Ψ satisﬁes

∇2 , it is isometrically isomorphic to the bidual of HM Ψ .

Proof. The weak-star closure of H Ψ is obvious with Proposition 3.1, 2).

Suppose now that Φ satisﬁes ∇2 , it is plainly seen that (HM Ψ )⊥ is the closed

subspace of LΦ = (M Ψ )∗ generated by all characters en with n < 0, where

en (t) = eint (for convenience, we deﬁne the duality between f ∈ LΨ and g ∈ LΦ by

integrating the product f ǧ, where ǧ(t) = g(−t)). As H Ψ ⊆ LΨ = (LΦ )∗ = (M Φ )∗

is the orthogonal of this latter subspace. So, we have (HM Ψ )⊥⊥ = H Ψ .

Proof of Proposition 3.1. Assume that 1) is satisﬁed. Since fr 1 ≤ CΨ fr Ψ ,

one has f ∈ H 1 , and hence, by Fatou-Riesz Theorem, f has almost everywhere a

boundary limit f ∗ ∈ L1 (m). If C = sup0≤r<1 fr Ψ , one has:

|fr |

Ψ dm ≤ 1

T C

for every r < 1; hence Fatou’s lemma implies:

∗

|f |

Ψ dm ≤ 1,

T C

i.e. f ∗ ∈ LΨ and f ∗ Ψ ≤ C.

Conversely, assume that 2) is satisﬁed. In particular f ∗ ∈ L1 (m); hence f ∈ H 1

and f ∗ = limr→1 fr almost everywhere. One has fr = f ∗ ∗ Pr , where Pr is the

Poisson kernel at r. Hence, using Jensen’s inequality for the probability measure

Pr (θ − t) 2π

dt

, we get:

2π ∗ 2π 2π ∗

|(f ∗ Pr )(θ)|
dθ |f (t)| dt dθ

Ψ ≤ Ψ Pr (θ − t)

0 f ∗ Ψ 2π 0 0 f ∗ Ψ 2π 2π

2π 2π ∗

|f (t)| dt dθ

≤ Ψ Pr (θ − t)

0 0 f ∗ Ψ 2π 2π

2π 2π ∗

dθ |f (t)|
dt

= Pr (θ − t) Ψ

0 0 2π f ∗ Ψ 2π

2π ∗

|f (t)|
dt

= Ψ ≤ 1,

0 f ∗ Ψ 2π

so that fr Ψ ≤ f ∗ Ψ . Hence we have 1), and f H Ψ ≤ f ∗ Ψ .

The two parts of the proof actually give f H Ψ = f ∗ Ψ .

**Proposition 3.4. For every f ∈ HM Ψ , one has fr − f ∗ Ψ −→ 0. Therefore
**

r→1

the polynomials on D are dense in HM Ψ .

Equivalently, on T = ∂D, the analytic trigonometric polynomials are dense in

HM Ψ .

1. HARDY-ORLICZ SPACES 11

LΨ

Proof. Let f ∈ HM Ψ and ε > 0. Since M Ψ = C(T) , there exists a continuous

function h on T such that f − hΨ ≤ ε. We have, for every r < 1:

Pr ∗ f − f Ψ ≤ Pr ∗ (f − h)Ψ + Pr ∗ h − hΨ + h − f Ψ ≤ 2ε + Pr ∗ h − hΨ

because Pr ∗ gΨ ≤ gΨ , for every r < 1 and every g ∈ LΨ .

But now, Pr ∗ h −→ h uniformly. The conclusion follows.

r→1

**Remark. We do not have to use a maximal function to prove the existence of
**

boundary limits because we use their existence for functions in H 1 . However, as in

the classical case, the Marcinkiewicz interpolation Theorem, or, rather, its Orlicz

space version ensures that the maximal non-tangential function is in LΨ . This result

is undoubtedly known, but perhaps never stated in the following form. Recall that

Nα is deﬁned, for every f , say in L1 (T), as

(Nα f )(eiθ ) = sup |(f ∗ Pr )(eit )| = sup |f (z)|,

reit ∈Sθ z∈Sθ

**where Sθ is the Stolz domain at eiθ with opening α (see [4], page 177); here f
**

deﬁnes a harmonic function in D.

Proposition 3.5. Assume that the complementary function Φ of the Orlicz

function Ψ satisﬁes the Δ2 condition (i.e. Ψ ∈ ∇2 ). Then every linear, or sublinear,

operator which is of weak-type (1, 1) and (strong) type (∞, ∞) is bounded from LΨ

into itself. In particular, for every f ∈ LΨ (T), the maximal non-tangential function

Nα f is in LΨ (T) (0 < α < 1).

Proof. If Ψ ∈ ∇2 , then ([33], Theorem 3, 1 (iii), page 23), there exists some

β > 1 such that xΨ (x) ≥ βΨ(x) for x large enough. Integrating between u and

u β

v, for u < v large enough, we get Ψ(u)

Ψ(v) ≥ v . Hence, for s, t large enough

Ψ−1 (s)

Ψ−1 (s/t)≤ t1/β . This means (see [4], Theorem 8.18) that the upper Boyd index

of L is ≤ 1/β < 1. Hence ([4], Theorem 5.17), Nα is bounded on LΨ (it is well-

Ψ

known that Nα f is dominated by the Hardy-Littlewood maximal function M f ).

**The following, essentially well-known, criterion for compactness of operators
**

will be very useful.

Proposition 3.6. 1) Every bounded linear operator T : H Ψ → X from H Ψ

into a Banach space X which maps every bounded sequence which is uniformly

convergent on compact subsets of D into a norm convergent sequence is compact.

2) Conversely, if T : H Ψ → X is compact and weak-star to weak continuous, or

if T : H Ψ → Y ∗ is compact and weak-star continuous, then T maps every bounded

sequence which is uniformly convergent on compact subsets of D into a norm con-

vergent sequence.

Though well-known (at least for the classical case of H p spaces), the link with

the weak (actually the weak-star) topology is usually not highlighted. Indeed, the

criterion is an easy consequence of the following proposition. Note that Proposi-

tion 3.6 will apply to the composition operators on H Ψ since they are weak-star

continuous.

Proposition 3.7. On the unit ball of H Ψ , the weak-star topology is the topology

of uniform convergence on every compact subset of D.

12 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

**Proof. First we notice that the topologies are metrizable. Indeed, this is known
**

for the topology of uniform convergence on every compact subset of D and, on the

other hand, M Φ is separable, so that the weak-star topology is metrizable on the

unit ball of its dual space LΨ , and a fortiori on that of H Ψ . Now, it is suﬃcient to

prove that the convergent sequences in both topologies are the same.

Let (fk )k≥1 be in the unit ball of H Ψ and weak-star convergent to f ∈ H Ψ .

Let us ﬁx a compact subset K of D. We may suppose that K is the closed ball of

center 0 and radius r < 1. First, testing the weak-star convergence on characters,

we have fk (n) −→ fˆ(n) for every n ∈ Z. Then:

k→∞

sup |fk (z) − f (z)| = sup |fk (z) − f (z)| ≤ r n |fk (n) − fˆ(n)|.

|z|≤r |z|=r n≥0

**The last term obviously tends to zero when k tends to inﬁnity. The result follows.
**

Conversely, let (fk )k≥1 be in the unit ball of H Ψ and converging to some holo-

morphic function f uniformly on every compact subset of D. We ﬁrst notice that

f actually lies in the unit ball of H Ψ (by Fatou’s lemma). Fix h ∈ M Φ and ε > 0.

There exists some r < 1 such that Pr ∗ h − hΦ ≤ ε/8, where Pr is the Poisson

kernel with parameter r. Then (see [33], page 58, inequality (4) for the presence of

the coeﬃcient 2):

|h, fk − f | ≤ |Pr ∗ h − h, fk − f | + |Pr ∗ h, fk − f |

≤ 2 Pr ∗ h − hΦ fk − f Ψ + |h, Pr ∗ (fk − f )|

ε

≤ + 2 hΦ [fk ]r − fr Ψ

2

ε

≤ + 2α hΦ [fk ]r − fr ∞

2

ε

= + 2α hΦ sup |fk (z) − f (z)|.

2 |z|=r

**where α is the norm of the injection of L∞ into LΨ .
**

Now, by uniform convergence on the closed ball of center 0 and radius r, there

exists kε ≥ 1 such that for every integer k ≥ kε , one has

hΦ sup |fk (z) − f (z)| ≤ ε/4.

|z|=r

**It follows that (fk )k weak-star converges to f .
**

However, we shall have to use a similar compactness criterion for Bergman-

Orlicz spaces, and it is worth stating and proving a general criterion. We shall

say that a Banach space X is a Banach space of holomorphic functions on an

open subset Ω of the complex plane if X is continuously embedded (through the

canonical injection) in H (Ω), the space of holomorphic functions on Ω, equipped

with its natural topology of compact convergence. We shall say that a Banach

space X of holomorphic functions on an open set Ω ⊆ C, has the Fatou property

if: for every bounded sequence (fn )n in X which converges uniformly on compact

subsets of Ω to a function f , one has f ∈ X. Then:

Proposition 3.8 (Compactness criterion). Let X, Y be two Banach spaces of

analytic functions on an open set Ω ⊆ C, X having the Fatou property. Let φ be

an analytic self-map of Ω such that Cφ = f ◦ φ ∈ Y whenever f ∈ X.

2. PRELIMINARY RESULTS 13

**Then Cφ : X → Y is compact if and only if for every bounded sequence (fn )n in
**

X which converges to 0 uniformly on compact subsets of Ω, one has Cφ (fn )Y → 0.

Note that Hardy-Orlicz H Ψ and Bergman-Orlicz B Ψ (see Chapter 5) spaces

trivially have the Fatou property, because of Fatou’s Lemma.

Proof. Assume that the above condition is fulﬁlled. Let (fn )n≥1 be in the unit

ball of X. The assumption on X implies that (fn )n is a normal family in H (Ω).

Montel’s Theorem allows us to extract a subsequence, that we still denote by (fn )n

to save notation, which converges to some f ∈ H (Ω), uniformly on compact subsets

of Ω. Since X has the Fatou property, one has f ∈ X. Now, since (fn − f )n is a

bounded sequence in X which converges to 0 uniformly on compact subsets of Ω,

one has Cφ (fn ) − Cφ (f )Y = Cφ (fn − f )Y → 0. Hence Cφ is compact.

Conversely, assume that Cφ is compact. Let (fn )n be a bounded sequence in

X which converges to 0 uniformly on compact subsets of Ω. By the compactness

of Cφ , we may assume that Cφ (fn ) → g ∈ Y . The space Y being continuously

embedded in H (Ω), (fn ◦ φ)n converges pointwise to g. Since (fn )n converges to

0 uniformly on compact subsets of Ω, the same is true for (fn ◦ φ)n . Hence g = 0.

Therefore, since Cφ is compact, we get Cφ (fn )Y → 0.

2. Preliminary results

Lemma 3.9. Let (Ω, P) be any probability space. For every function g ∈ L∞ (Ω),

one has:

g∞

gΨ ≤ −1 ·

Ψ (g∞ /g1 )

Proof. We may suppose that g∞ = 1.

Since Ψ(0) = 0, the convexity of Ψ implies Ψ(ax) ≤ aΨ(x) for 0 ≤ a ≤ 1. Hence,

for every C > 0, one has, since |g| ≤ 1:

Ψ(|g|/C) dP ≤ |g|Ψ(1/C) dP = g1 Ψ(1/C).

Ω Ω

But g1 Ψ(1/C) ≤ 1 if and only if C ≥ 1/Ψ−1 (1/g1 ), and that proves the lemma.

**Corollary 3.10. For |a| = 1 and 0 ≤ r < 1, one has:
**

1

ua,r Ψ ≤ ·

Ψ−1 ( 1−r

1

)

Proof. One has ua,r ∞ = 1, and:

2π

1 − r 2

ua,r 1 = dm(t)

0 1 − āreit

+∞

(1 − r)2 1−r

= (1 − r)2 r 2n = = ·

n=0

1 − r 2 1 +r

Hence ua,r Ψ ≤ 1/Ψ−1 ((1 + r)/(1 − r)), by using Lemma 3.9, giving the result

since (1 + r)/(1 − r) ≥ 1/1 − r.

Remark. We hence get actually ua,r Ψ ≤ 1/Ψ−1 ((1 + r)/(1 − r)); the term 1 + r

has no important meaning, so we omit it in the statement of Corollary 3.10, but

14 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

**sometimes, for symmetry of formulae, or in order to be in accordance with the
**

classical case, we shall use this more precise estimate.

**For every f ∈ L1 (T) and every z = r eiθ ∈ D, one has
**

2π

(P [f ])(z) = f (eit )Pz (t) dm(t),

0

**where Pz is the Poisson kernel:
**

1 − r2 1 − |z|2

Pz (t) = = it ,

1 − 2r cos(θ − t) + r 2 |e − z|2

and f (z) = (P [f ])(z) when f is analytic on D. Since Pz ∈ L∞ (T) ⊆ LΦ (T), it

follows that the evaluation in z ∈ D:

δz (f ) = f (z)

Ψ

is a continuous linear form on H . The following lemma explicits the behaviour of

its norm.

Lemma 3.11. For |z| < 1, the norm of the evaluation functional at z is:

1

δz (HM Ψ )∗ = δz (H Ψ )∗ ≈ Ψ−1 ·

1 − |z|

More precisely:

1 −1 1 + |z|
1 + |z|

Ψ ≤ δz (H Ψ )∗ ≤ 2Ψ−1 ·

4 1 − |z| 1 − |z|

Remark. In particular:

1 −1 1
1

Ψ ≤ δz (H Ψ )∗ ≤ 4Ψ−1 ,

4 1 − |z| 1 − |z|

which often suﬃces for our purpose.

**Proof. The ﬁrst equality δz (HM Ψ )∗ = δz (H Ψ )∗ comes from the fact that fr ∈
**

HM Ψ , for every f ∈ H Ψ and r < 1 (thus f (rz) −→ f (z), when z ∈ D and f ∈ H Ψ ).

r→1

On the one hand, we have, when |z| = r, using [33], inequality (4) page 58,

and Lemma 3.9, since Pz 1 = 1 and Pz ∞ = 1−r

1+r

:

1+r 1

δz (H Ψ )∗ ≤ 2 Pz Φ ≤ 2 1+r ,

1−rΦ −1

1−r

**which is less than 2Ψ−1 ((1+r)/(1−r)), by using the inequality (see [33], Proposition
**

1 (ii), page 14, or [22], pages 12–13):

Ψ−1 (x)Φ−1 (x) ≥ x , x > 0.

On the other hand, one has, using Corollary 3.10, with r = |z| and āz = r:

|ua,r (z)| 1/(1 + r)2 1 1 + r

δz (H Ψ )∗ ≥ ≥ 1+r ≥ Ψ−1 ,

ua,r Ψ 1/Ψ−1 1−r 4 1−r

and that ends the proof.

3. COMPOSITION OPERATORS 15

3. Composition operators

We establish now some estimations for the norm of composition operators.

Proposition 3.12. 1) Every analytic self-map φ : D → D induces a bounded

composition operator Cφ : H Ψ → H Ψ by setting Cφ (f ) = f ◦ φ. More precisely:

1 + |φ(0)|

Cφ ≤ ·

1 − |φ(0)|

In particular, Cφ ≤ 1 if φ(0) = 0.

2) One has:

1 −1 1 + |φ(0)|

Cφ ≥ Ψ .

8Ψ−1 (1) 1 − |φ(0)|

3) When Ψ ∈ ∇1 globally: Ψ(x)Ψ(y) ≤ Ψ(bxy) for all x, y ≥ 0, we also have:

1 + |φ(0)|

Cφ ≤ bΨ−1 .

1 − |φ(0)|

4) Moreover, Cφ maps HM Ψ into HM Ψ . Hence, if Ψ ∈ ∇2 , then Cφ : H Ψ →

H is the bi-adjoint of the composition operator Cφ : HM Ψ → HM Ψ .

Ψ

**Note that when Ψ(x) = xp for 1 ≤ p < ∞, then Ψ ∈ ∇1 globally, with b = 1.
**

Proof. 1) Assume ﬁrst that φ(0) = 0. Let f ∈ H Ψ , with f Ψ = 1. Since Ψ is

convex and increasing, the function u = Ψ ◦ |f | is subharmonic on D, thanks to

Jensen’s inequality. The condition φ(0) = 0 allows to use Littlewood’s subordina-

tion principle ([11], Theorem 1.7); for r < 1, one has:

2π 2π

dt dt

Ψ |(f ◦ φ)(reit )| ≤ Ψ |f (reit )| ≤ 1.

0 2π 0 2π

Hence f ◦ φ ∈ H Ψ and f ◦ φΨ ≤ 1.

Assume now that φ is an inner function, and let a = φ(0). It is known that

(see [32], Theorem 1) that

φ∗ (m) = Pa .m ,

where φ∗ (m) is the image under φ∗ (the boundary limit of φ) of the normalized

Lebesgue measure m, and Pa .m is the measure of density Pa , the Poisson kernel

at a. Therefore, for every f ∈ H Ψ with f Ψ = 1, one has for 0 ≤ r < 1, setting

1+|a|

Ka = Pa ∞ = 1−|a| :

2π

|(f ◦ φ)(reit )|
dt |(f ◦ φ)∗ |

Ψ ≤ Ψ dm

0 Ka 2π T Ka

∗

|f ◦ φ∗ |

= Ψ dm (recall that |φ∗ | = 1)

T Ka

∗
∗

|f | |f |

(3.1) = Ψ dφ∗ (m) = Ψ Pa dm

Ka Ka

T T

1

≤ Ψ(|f ∗ |) Pa dm , since Ka > 1

T K a

1

≤ Ψ(|f ∗ |) Pa ∞ dm

K

T a

= Ψ(|f ∗ |) dm ≤ 1.

T

16 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

**Hence (f ◦ φ)r Ψ ≤ Ka , and therefore f ◦ φΨ ≤ Ka .
**

Then, for an arbitrary φ, let a = φ(0) again, and let φa be the automorphism

z → 1−āz

z−a ,

whose inverse is φ−a . Since φ = φ−a ◦(φa ◦φ), one has Cφ = Cφa ◦φ ◦Cφ−a .

But φ−a is inner and, on the other hand, (φa ◦ φ)(0) = 0; hence parts a) and b) of

the proof give:

1 + |a|

Cφ ≤ Cφ−a ≤ Ka = ,

1 − |a|

which gives the ﬁrst part of the proof.

2) By Lemma 3.11, we have for every f ∈ H Ψ with f Ψ ≤ 1:

|(f ◦ φ)(0)| ≤ δ0 (H Ψ )∗ f ◦ φΨ ≤ 2Ψ−1 (1) Cφ .

In other words:

|f φ(0) | ≤ 2Ψ−1 (1)Cφ

for every such f ∈ H Ψ . Hence:

δφ(0) (H Ψ )∗ ≤ 2Ψ−1 (1) Cφ ,

giving

1 −1 1 + |φ(0)|

Cφ ≥ Ψ ,

8Ψ−1 (1) 1 − |φ(0)|

by using Lemma 3.11 again, but the minoration.

3) When Ψ ∈ ∇1 globally, we go back to the proof of 1). We have only to

modify inequalities (3.1). Setting Ka = Ψ−1 (Ka ), and writing Pa = Ψ Ψ−1 (Pa ) ,

we get, for every f ∈ H Ψ with f Ψ = 1:

2π ∗

|(f ◦ φ)(reit )|
dt |f |

Ψ

≤ Ψ Pa dm

bKa 2π bK

0

T ∗a

|f |

= Ψ

Ψ Ψ−1 (Pa ) dm

bK

T ∗a

|f | −1

≤ Ψ Ψ (P a ) dm

Ka

T

≤ Ψ(|f ∗ |) dm ≤ 1,

T

**since Ψ−1 (Pa ) ≤ Ψ−1 (Pa ∞ ) = Ka , giving f ◦ φΨ ≤ bKa .
**

4) Suppose now that f ∈ HM Ψ . As before, when φ(0) = 0, Littlewood’s

subordination principle gives, for every C > 0:

2π 2π

dt dt

Ψ C|(f ◦ φ)(eit )| = sup Ψ C|(f ◦ φ)(reit )|

0 2π r<1 0 2π

2π 2π

dt dt

≤ sup Ψ C|f (reit )| = Ψ C|f (eit )| < +∞ ;

r<1 0 2π 0 2π

hence f ◦ φ ∈ HM Ψ . When φ is inner, the same computations as in 1) above, using

that φ∗ (m) = Pa .m, where a = φ(0), give, for every C > 0:

2π

dt

Ψ C|(f ◦ φ)(reit )| ≤ Ψ(C|f ∗ |)Pa ∞ dm

2π T

0

= Ka Ψ(C|f ∗ |) dm < +∞,

T

4. ORDER BOUNDED COMPOSITION OPERATORS 17

**and f ◦ φ ∈ HM Ψ again. The general case follows as in 1) above, since f ∈ HM Ψ
**

implies f ◦φ−a ∈ HM Ψ , because φ−a is inner, and then f ◦φ = (f ◦φ−a )◦(φa ◦φ) ∈

HM Ψ since (φa ◦ φ)(0) = 0.

**4. Order bounded composition operators
**

Recall that an operator T : X → Z from a Banach space X into a Banach

subspace Z of a Banach lattice Y is order bounded if there is some positive y ∈ Y

such that |T x| ≤ y for every x in the unit ball of X.

Before studying order bounded composition operators, we shall recall the fol-

lowing, certainly well-known, result, which says that order boundedness can be seen

as stronger than compactness.

Proposition 3.13. Let T : L2 (μ) → L2 (μ) be a continuous linear operator.

Then T is order bounded if and only if it is a Hilbert-Schmidt operator.

The proof is straightforward: if B is the unit ball of L2 (μ), and (ei )i is an

2 1/2

orthonormal basis, one has supf ∈B |T f | = i |T ei | . Hence supf ∈B |T f | ∈

L2 (μ) if and only if i |T ei | 2

dμ = i T ei 2

< +∞, i.e. if and only if T is

Hilbert-Schmidt.

J. H. Shapiro and P. D. Taylor proved in [42] that there exist composition

operators on H 2 which are compact but not Hilbert-Schmidt. We are going to see

that, when the Orlicz function Ψ grows fast enough, the compactness of composition

operators on H Ψ is equivalent to their order boundedness.

Proposition 3.14. The composition operator Cφ : H Ψ → HΨ is order bounded

(resp. order bounded into M Ψ (T)) if and only if Ψ−1 1−|φ|1

∈ LΨ (T) (resp.

Ψ−1 1−|φ|

1

∈ M Ψ (T)). Equivalently, if and only if

1

(OB1) χA ∈ L1 (T) for some A > 0,

1 − |φ|

respectively:

1

(OB2) χA ∈ L1 (T) for every A > 0,

1 − |φ|

In other words (recall that χA (x) = Ψ AΨ−1 (x) ), if and only if:

1

−1

Ψ AΨ dm < +∞ for some (resp. every) A > 0.

T 1 − |φ|

**Remark For a sequence (gn )n of elements of LΨ (Ω), one has gn Ψ −→ 0 if and
**

n→+∞

only if

|g |

n

(3.2) Ψ dP −→ 0 for every ε > 0.

Ω ε n→+∞

**In fact, if (3.2) holds, then for n ≥ nε , the above integrals are ≤ 1, and hence
**

gn Ψ ≤ ε. Conversely, assume that gn Ψ −→ 0, and let ε > 0 be given.

n→+∞

Let 0 < δ ≤ 1. Since gn /(εδ)Ψ −→ 0, one has gn /(εδ)Ψ ≤ 1, and hence

n→+∞

18 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

Ω

Ψ |gεδn | dP ≤ 1, for n large enough. Then, using the convexity of Ψ:

|g |
|g |
|g |

n n n

Ψ dP = Ψ δ dP ≤ δ Ψ dP ≤ δ ,

Ω ε Ω εδ Ω εδ

for n large enough.

Remark If Cφ : H Ψ → H Ψ is order bounded into M Ψ (T), then the composition

operator Cφ : H Ψ → H Ψ is compact.

Indeed, let (gn )n≥0 be a sequence in the unit ball of H Ψ . Thank s to Propo-

sition 3.7, there exists a subsequence, which we still denote by (gn )n≥0 , uniformly

converging to some g (lying in the unit ball of H Ψ ) on every compact subset of D.

Since (OB2) implies that |φ∗ | < 1 almost everywhere, we have for every > 0,

|g ◦ φ∗ |

n

Ψ −→ 0 a.e.

ε n→+∞

**On the other hand, the order boundedness hypothesis on Cφ implies that, on T:
**

|g ◦ φ∗ |
G

n

Ψ ≤Ψ a.e.,

ε ε

where G ∈ M Ψ (T). Using Lebesgue’s dominated convergence Theorem and the

previous remark, the conclusion follows.

Proof of Proposition 3.14. As HM Ψ is separable, there exists a sequence

(fn )n≥1 in the unit ball of HM Ψ such that

supfn ◦ φ(reiθ ) = δφ(reiθ ) (HM Ψ )∗ .

n

**Now, suppose that Cφ is order bounded into LΨ (T) (resp. into M Ψ (T)). Then
**

there exists some g in LΨ (T) (resp. in M Ψ (T)) such that g ≥ |Cφ (f )| a.e., for every

f in the unit ball of H Ψ . Using Lemma 3.11, we have a.e.

1

Ψ−1 ≤ 4δφ(reiθ ) (H Ψ )∗ = 4 supfn ◦ φ(reiθ ) ≤ 4g.

1 − |φ(re )|

iθ

n

The result hence follows letting r ↑ 1.

The converse is obvious.

**Theorem 3.15. If the composition operator Cφ : H Ψ → H Ψ is order bounded,
**

then:

1

(OB3) m(1 − |φ| < λ) = O , as λ → 0, for some A > 0.

χA (1/λ)

and if it is order bounded into M Ψ (T), then:

1

(OB4) m(1 − |φ| < λ) = O , as λ → 0, for every A > 0.

χA (1/λ)

Under the hypothesis Ψ ∈ Δ1 , the converse holds.

**Proof. The necessary condition follows from Proposition 3.14 and Markov’s in-
**

equality:

1
1 1

m >t ≤ χA dm.

1 − |φ| χA (t) T 1 − |φ|

For the converse, we shall prove a stronger result, and for that, we deﬁne the

weak-LΨ space as follows:

4. ORDER BOUNDED COMPOSITION OPERATORS 19

**Definition 3.16. The weak-LΨ space LΨ,∞ (Ω) is the space of measurable func-
**

tions f : Ω → C such that, for some constant c > 0, one has, for every t > 0:

1

P(|f | > t) ≤ ·

Ψ(ct)

For subsequent references, we shall state separately the following elementary

result.

Lemma 3.17. For every f ∈ LΨ (Ω), one has, for every t > 0:

t

f Ψ ≥ ·

Ψ−1 1

P(|f |>t)

**Proof. By Markov’s inequality, one has, for t > 0:
**

t
|f |

Ψ P(|f | > t) ≤ Ψ dP ≤ 1;

f Ψ Ω f Ψ

and that gives the lemma.

**Since Lemma 3.17 can be read:
**

1

P(|f | > t) ≤ ,

Ψ(t/f Ψ )

we get that LΨ (Ω) ⊆ LΨ,∞ (Ω).

The converse of Theorem 3.15 is now an immediate consequence of the following

proposition.

Proposition 3.18.

1) a) If Ψ ∈ Δ1 , then LΨ (Ω) = LΨ,∞ (Ω).

b) If LΨ (Ω) = LΨ,∞ (Ω), then Ψ ∈ Δ0 .

2) If LΨ (T) = LΨ,∞ (T), then condition (OB3) implies that Cφ : H Ψ → H Ψ is

order bounded, and condition (OB4) implies that it is order bounded into M Ψ (T).

Lemma 3.19. The following assertions are equivalent

i) LΨ (Ω) = LΨ,∞ (Ω).

∞ +∞

Ψ (u) 1

ii) du ≡ dx < +∞, for some B > 1.

1 Ψ(Bu) Ψ(1) χ B (x)

Proof of the Lemma. Assume ﬁrst that 1/χB is integrable on (Ψ(1), ∞). For

every f ∈ LΨ,∞ (Ω), there is a c > 0 such that P(|f | > t) ≤ 1/Ψ(ct). Then

|f |
+∞

Ψ c dP = P(|f | > Bt/c) Ψ (t) dt

Ω B 0

+∞

≤ Ψ(1) + P(|f | > Bt/c) Ψ (t) dt

1

+∞ +∞

Ψ (t) du

≤ Ψ(1) + dt = Ψ(1) + < +∞,

1 Ψ(Bt) Ψ(1) χ B (u)

so that f ∈ LΨ (Ω).

Conversely, assume that LΨ (Ω) = LΨ,∞ (Ω). Since 1/Ψ is decreasing, there is

a measurable function f : Ω → C such that P(|f | > t) = 1/Ψ(t), when t ≥ Ψ−1 (1).

20 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

**Such a function is in LΨ,∞ (Ω); hence it is in LΨ (Ω), by our hypothesis. Therefore,
**

there is a B > 1 such that

Ψ(|f |/B) dP < +∞.

Ω

But

+∞

+∞

Ψ (t)

Ψ(|f |/B) dP = P(|f | > Bt) Ψ (t) dt ≥ dt

Ω 0 1

BΨ

−1 (1) Ψ(Bt)

+∞

Ψ (t) +∞

du

≥ dt = ,

Ψ−1 (1) Ψ(Bt) 1 χB (u)

and hence 1/χB is integrable on (1, ∞).

Proof of Proposition 3.18. 1)a) We ﬁrst remark that for every Orlicz function

Ψ, one has Ψ(2x) ≥ xΨ (x) for every x > 0, because, since Ψ is positive and

increasing:

2x 2x

Ψ(2x) = Ψ (t) dt ≥ Ψ (t) dt ≥ xΨ (x).

0 x

Assume now that Ψ ∈ Δ1 : xΨ(x) ≤ Ψ(βx) for some β > 0 and for x ≥ x0 > 0. By

Lemma 3.19, it suﬃces to show that 1/χ2β is integrable on (Ψ(x0 ), +∞). But

+∞ +∞ +∞ +∞

dx Ψ (u) Ψ(2u)/u du

= du ≤ du = < +∞.

Ψ(x0 ) χ2β (x) x0 Ψ(2βu) x0 2uΨ(2u) x0 2u2

1)b) Suppose now that LΨ (Ω) = LΨ,∞ (Ω). By the preceding lemma, there

exists some B > 1 such that

2x

Ψ (u)

lim du = 0.

x→+∞ x Ψ(Bu)

By convexity, Ψ(2x) ≥ 2Ψ(x) and Ψ is nonnegative, so that

2x 2x

Ψ (u) 1 Ψ(2x) − Ψ(x) Ψ(x)

du ≥ Ψ (u) du ≥ ≥ ·

x Ψ(Bu) Ψ(2Bx) x Ψ(2Bx) Ψ(2Bx)

Therefore Ψ satisﬁes Δ0 .

2) Assume that LΨ (T) = LΨ,∞ (T) and that condition (OB3) (resp. (OB4))

holds. By Lemma 3.19, there is a B > 0 such that 1/χB is integrable on (1, +∞).

We get, using condition (OB3) (resp. (OB4)) and setting C = A/B:

1
+∞

1

χC dm = m > t χC (t) dt

T 1 − |φ| 0 1 − |φ|

+∞

= m(1 − |φ| < 1/t) χC (t) dt

0

+∞

χC (t)

≤ χC (1) + K dt.

1 χA (t)

But, if we set u = χC (t), i.e. u = Ψ CΨ−1 (t) , then Ψ−1 (u) = CΨ−1 (t), and hence

χA (t) = Ψ AΨ−1 (t) = Ψ BΨ−1 (u) = χB (u).

Therefore:

1
+∞

du

χC dm ≤ χC (1) + K du < +∞.

T 1 − |φ| χC (1) χB (u)

5. WEAKLY COMPACT COMPOSITION OPERATORS 21

**It follows from Proposition 3.14 that Cφ is order bounded (resp. into M Ψ (T)).
**

Remark. The condition Ψ ∈ Δ1 is not equivalent to LΨ (Ω) = LΨ,∞ (Ω). For

example, we can take:

3/2

Ψ(x) = exp log(x + 1) − 1.

Then, as x tends to inﬁnity,

3

Ψ(Kx) ∼ Ψ(x) exp (log K)(log x)1/2 ,

2

1/2

and hence Ψ ∈ Δ1 . On the other hand, Ψ (x) ∼ 32 (log xx) Ψ(x); hence:

+∞ +∞

du Ψ (t)

= dt

Ψ(1) χK (u) 1 Ψ(Kt)

+∞ +∞

3u2 3u2

∼ 3 du = du < +∞

0 exp( 2 (log K)u) 0 K 3u/2

for K > 1. Therefore LΨ (Ω) = LΨ,∞ (Ω) by Lemma 3.19.

**5. Weakly compact composition operators
**

We saw in Lemma 3.10 that

1

ua,r Ψ ≤ ·

Ψ−1 ( 1−r

1

)

The next result shows that the weak compactness of Cφ transforms the “big-oh”

into a “little-oh”, when Ψ grows fast enough.

Theorem 3.20. Assume that Ψ ∈ Δ0 . Then the weak compactness of the

composition operator Cφ : H Ψ → H Ψ implies that:

1

(W) sup Cφ (ua,r )Ψ = o 1 , as r → 1.

a∈T Ψ−1 1−r

**Proof. Actually, we only need to use that the restriction of Cφ : H Ψ → H Ψ to
**

HM Ψ is weakly compact. We proved in [25], Theorem 4, that, under the hypothesis

Ψ ∈ Δ0 , the operator Cφ : HM Ψ → HM Ψ is weakly compact if and only if for every

ε > 0, we can ﬁnd Kε > 0 such that, for every f ∈ HM Ψ , one has:

Cφ (f )Ψ ≤ Kε f 1 + ε f Ψ .

Using Corollary 3.10, we get, for every ε > 0, since ua,r 1 ≤ 1 − r:

1

Cφ (ua,r )Ψ ≤ Kε (1 − r) + ε ·

Ψ−1 ( 1−r

1

)

Ψ(x)

But −→ +∞.

x x→+∞ Hence

1

1−r =o as r → 1,

Ψ−1 ( 1−r

1

)

and that proves the theorem.

22 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

**We shall see in Chapter 4 that the converse holds when Ψ satisﬁes Δ0 and more-
**

over that Cφ is then compact. That will use some other techniques. Nevertheless,

we can prove, from now on, the following result.

Theorem 3.21. If Ψ ∈ Δ2 , then condition:

1

(W) sup Cφ (ua,r )Ψ = o 1 , as r → 1,

a∈T Ψ−1 1−r

in Theorem 3.20 implies condition:

1

(OB4) m(1 − |φ| < λ) = O , as λ → 0, for every A > 0,

χA (1/λ)

in Theorem 3.15.

Remark that when Ψ ∈ Δ0 (and in particular when Ψ ∈ Δ2 ), one has, for any

B > βA (where β is given by the deﬁnition of Δ0 ): Ψ(Bx)

Ψ(Ax) → +∞ as x → +∞;

hence 1/χA (x) = o 1/χB (x) and the “big-oh” condition in (OB4) may be replaced

by a “little-oh” condition.

Before proving this theorem, let us note that:

Proposition 3.22. Condition

1

(W) sup Cφ (ua,r )Ψ = o as r → 1

a∈T Ψ−1 1−r

1

implies that m(|φ| = 1) = 0.

Proof. Otherwise, one has m(|φ| = 1) = δ > 0. Splitting the unit circle T into N

parts, we get some a ∈ T such that:

m(|φ − a| ≤ π/N ) ≥ δ/N.

But, the inequality |φ − a| ≤ π/N implies, with r = 1 − 1/N (since |φ| ≤ 1):

|1 − ārφ| ≤ |1 − āφ| + (1 − r)|āφ| = |a − φ| + (1 − r)|φ| < 5/N = 5(1 − r).

Hence:

m |Cφ (ua,r )| > 1/25) = m |1 − ārφ| < 5(1 − r)

≥ m |φ − a| ≤ π(1 − r) ≥ δ(1 − r) ,

and therefore, by Lemma 3.17:

1/25 δ/25

Cφ (ua,r )Ψ ≥ ≥ ·

Ψ−1 1/δ(1 − r) Ψ−1 1/(1 − r)

Since r can be taken arbitrarily close to 1, that proves the proposition.

**Proof of Theorem 3.21. Assume that condition (W) is satisﬁed, and ﬁx A > 1.
**

Let ε > 0 to be adjusted later. We can ﬁnd rε < 1 such that rε ≤ r < 1 implies:

ε

Cφ (ua,r )Ψ ≤ −1 1 , ∀a ∈ T.

Ψ 1−r

Now, Lemma 3.17 also reads:

1

m(|f | > t) ≤ t ,

Ψ f Ψ

5. WEAKLY COMPACT COMPOSITION OPERATORS 23

**so that, if one sets B = 1/9ε:
**

1

m |1 − ārφ| < 3(1 − r) = m |Cφ (ua,r )| > 1/9 ≤ 1 ·

Ψ BΨ−1 1−r

We claim that this implies a good upper bound on m(|φ| > r), even if we lose

a factor 1/(1 − r), due to the eﬀect of a rotation on φ. For that, we shall use the

following lemma.

Lemma 3.23. Let φ : D → D be an analytic function. Then, for every r with

0 < r < 1, there exists a ∈ T such that:

1−r

m |1 − ār φ| < 3(1 − r) ≥ m(|φ| > r).

8

Admitting this for a while, we are going to ﬁnish the proof.

Fix an r such that rε ≤ r < 1, and take an a ∈ T as in Lemma 3.23. We get,

from the preceding, in setting λ = 1 − r:

m(1 − |φ| < λ) = m(|φ| > r)

8

≤ m |1 − ārφ| < 3(1 − r)

1−r

8 1

≤ 1 ,

1 − r Ψ BΨ−1 1−r

i.e., setting x = Ψ−1 (1/1 − r) = Ψ−1 (1/λ):

Ψ(x)

m(1 − |φ| < λ) ≤ 8 ·

Ψ(Bx)

2

But Ψ satiﬁes the Δ2 -condition: Ψ(y) ≤ Ψ(αy) for some α > 1 and y large

enough. Then, adjusting now ε > 0 as ε = 1/9αA, in order that B = αA, we get,

for x large enough, since A > 1:

Ψ(x)Ψ(Ax) ≤ [Ψ(Ax)]2 ≤ Ψ(Bx).

Therefore, for r close enough to 1:

8 8

m(1 − |φ| < λ) ≤ = ·

Ψ(Ax) χA (1/λ)

We hence get condition (OB4), and that proves Theorem 3.21.

**Proof of Lemma 3.23. Let λ = 1 − r, and let δ > 0 be a number which we shall
**

specify later. Consider the set:

Cδ = {z ∈ D ; |z| ≥ 1 − λ and |arg z| ≤ δ}

(for δ = λ, Cδ is a closed Carleson window). It is geometrically clear that (1 − λ)Cδ

is contained in the closed disk of center 1 and whose edge contains (1 − λ)2 eiδ ;

hence, for every z ∈ Cδ , one has:

|1 − (1 − λ)z|2 ≤ |1 − (1 − λ)2 eiδ |2 = 2(1 − λ)2 (1 − cos δ) + λ2 (2 − λ)2

≤ (1 − λ)2 δ 2 + λ2 (2 − λ)2 ≤ 9λ2

if δ ≤ λ.

By rotation, one has, for every a ∈ T:

|z| ≥ 1 − λ and |arg (āz)| ≤ δ ⇒ |1 − (1 − λ)āz| ≤ 3λ.

24 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

**Let now N ≥ 2 be the integer such that:
**

π π

≤λ< ,

N N −1

and take δ = π/N .

One has, by the previous inequalities, setting ak = e2ikδ :

{z ∈ D ; |z| ≥ 1 − λ} = āk Cδ ⊆ {z ∈ D ; |1 − (1 − λ)āk z| ≤ 3λ}.

1≤k≤N 1≤k≤N

it

Hence, with z = φ(e ) (remark that, by Proposition 3.22, we have only to consider

the values of eit for which |φ(eit )| < 1; however, in this lemma, we may replace D

by D), and get:

m(|φ| ≥ 1 − λ) ≤ N sup m(|1 − (1 − λ)āk φ| ≤ 3λ).

1≤k≤N

**Therefore, we can ﬁnd some k such that:
**

1 λ

m(|1 − (1 − λ)āk φ| ≤ 3λ) ≥

m(|φ| ≥ 1 − λ) ≥ m(|φ| ≥ 1 − λ),

N 8

since λ ≤ 2π/N ≤ 8/N . That proves Lemma 3.23.

**Since the Δ2 -condition implies the Δ1 -condition, which, in its turn, implies the
**

0

Δ -condition, we get, from Theorem 3.15, Theorem 3.20 and Theorem 3.21 that

the weak compactness of Cφ implies its order boundedness into M Ψ (T), and thanks

to the Remark after Proposition 3.14, its compactness. We get

Theorem 3.24. If Ψ satisﬁes the Δ2 -condition, then the following assertions

for composition operator Cφ : H Ψ → H Ψ are equivalent:

1) Cφ is order bounded into M Ψ (T);

2) Cφ is compact;

3) Cφ is weakly

compact;

4) Ψ−1 1−|φ|

1

∈ M Ψ (T) (i.e.: χB (1/1 − |φ|) ∈ L1 (T) for every B > 0);

5) m(1 − |φ| < λ) = O χA (1/λ)

1

as λ → 0, for every A > 0;

1

6) sup Cφ (ua,r )Ψ = o 1 as r → 1 (W )

a∈T Ψ−1 1−r

Moreover these conditions

imply

7) φn p = o n1q , for every q ≥ 1 and every p ≥ 1.

8) Cφ is a nuclear operator on every H p , where p ≥ 1.

Remark. We shall see in the next Chapter (Theorem 4.22) that the assumption

Ψ ∈ Δ2 cannot be removed in general: Theorem 3.24 is not true for the Orlicz

2

function Ψ(x) = exp (log(x + 1) − 1 (which nevertheless satisﬁes Δ1 ).

Proof. Only the two last assertions are not proved yet.

First, we are going to show that 4) ⇒ 7).

Since Ψ ∈ Δ2 , there exists α > 1 such that Ψ(αx) ≥ [Ψ(x)]2 for x large enough.

Hence there exists a > 0 such that Ψ(αx) ≥ a[Ψ(x)]2 for every x > 0, and, for every

integer k ≥ 1, one has Ψ(αk x) ≥ a2 −1 [Ψ(x)]2 for every x > 0.

k k

**Let q ≥ 1 and take k ≥ 1 such that 2 k−1
**

≤ q < 2k .

5. WEAKLY COMPACT COMPOSITION OPERATORS 25

**We shall use the following inequality:
**

1 n n/2q

≥ t for 0 ≤ t < 1, n ≥ 1,

1−t 2q

which follows from the fact that the map t → td (1 − t) is less than 1

d+1 on [0, 1].

Take B = αk in 4) and set

1

C = Cq = Ψ αk Ψ−1 dm.

T 1 − |φ|

Since

1
n
2 k

Ψ αk Ψ−1 ≥ a2 −1

k k

|φ|2 n/2q ,

1 − |φ| 2q

k−1

we get, since |φ| ≤ 1 and q ≥ 2k−1 imply that |φ|2 n/q ≥ |φn |:

n
2k

C ≥ a2 −1

k

φn 1 ,

2q

so:

C(2q)2 1

k 1

φn 1 ≤ 2k −1 2k = o q ,

a n n

since q < 2k .

Now, by interpolation, we actually have φn p = o n1q , for every q, p ≥ 1,

since φ∞ ≤ 1, and that proves 7).

To see that Cφ is nuclear on each H p , it suﬃces to notice that, for every f ∈ H p :

∞

∞

Cφ (f ) = fˆ(n)φn = e∗n (f )φn

n=0 n=0

**where e∗n (f ) = fˆ(n) is a norm one functional on H p . Moreover
**

∞

e∗n (H p )∗ .φn p < ∞

n=0

and that ends the proof.

xp

If one specializes this corollary to the case where Ψ(x) = Ψp (x) = e −1

(p > 0), which veriﬁes the Δ2 -condition, we get, using Stirling’s formula:

Corollary 3.25. The following assertions are equivalent:

1) Cφ : H Ψp → H Ψp is order bounded into M Ψp (T);

2) Cφ : H Ψp → H Ψp is compact;

3) Cφ : H Ψp → H Ψp is weakly compact;

1

4) 1−|φ| ∈ Lr (T), ∀r ≥ 1;

5) ∀q ≥ 1 ∃Cq > 0: m(1 − |φ| < λ) ≤ Cq λ q

;

−1/p

6) sup Cφ (ua,r )Ψp = o log(1 − r) as r → 1

a∈T

7) ∀q ≥ 1 φn 2 =√o (n−q );

8) φn Ψp = o (1/ p log n).

Remark. Observe that conditions 4), 5) and 7) do not depend on p. Hence this is

equivalent to the same properties of Cφ acting on H Ψs for another s > 0.

As a consequence of Theorem 3.24, we obtain the following:

26 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

**Corollary 3.26. Assume that Ψ ∈ Δ2 . Then there exist compact composition
**

operators Cφ : H p → H p for 1 ≤ p < ∞ (even Hilbert-Schmidt on H 2 ) which are

not compact as operators Cφ : H Ψ → H Ψ .

Remark. We shall see in Theorem 4.3 that compactness on H Ψ implies com-

pactness on H p for p < ∞. Note that this shows that, though H Ψ is an inter-

polation space between H 1 and H ∞ (see [4], Theorem V.10.8), the compactness

of Cφ : H 1 → H 1 with the continuity of Cφ : H ∞ → H ∞ does not suﬃce to have

compactness for Cφ : H Ψ → H Ψ .

Remark. Corollary 3.26 will be improved in [27], Theorem 4.1, where we show

that whenever Ψ ∈ / Δ2 , there exists a composition operator which is compact on

H 2 , but not on H Ψ .

Proof. As the Δ2 condition implies the Δ0 condition, we have x = o χβ (x) as

x → ∞, for some β > 1. It follows that we can ﬁnd a positive function a : T → R+

such that a ≥ 2, a ∈ L1 but χβ (a) ∈ / L1 . Set h = 1 − a1 . One has 1/2 ≤ h ≤ 1 and

in particular log h ∈ L (T). Then the outer function φ : D → C deﬁned for z ∈ D

1

by:

u+z

φ(z) = exp log h(u) dm(u)

T u−z

is analytic on D and its boundary limit veriﬁes |φ| = h ≤ 1 on T. By [42],

Theorem

6.2, Cφ : H 2 → H 2 is Hilbert-Schmidt, and hence compact, since T 1−|φ| 1

dm =

T

a dm < +∞.

It is then

1 compact

from H p

to H p

for every p < ∞ ([42], Theorem

6.1). However, T χβ 1−|φ| dm = T χβ (a) dm = +∞, and hence, by our Theorem

3.24, Cφ is not compact on H Ψ .

6. p-summing operators

Recall that an operator T : X → Y between two Banach spaces is said to be

p-summing (1 ≤ p < +∞) if there is a constant C > 0 such that, for every choice

of x1 , . . . , xn ∈ X, one has:

n

n

T xk p ≤ C p sup |x∗ (xk )|p .

x∗ ∈X ∗

k=1 x∗ ≤1 k=1

**In other terms, T maps weakly unconditionaly p-summable sequences into
**

norm p-

summable sequences. When X ⊆ Y = LΨ , this implies that whenever n≥1 |gn | ∈

LΨ , then n≥1 T gn pΨ < +∞.

For 1 ≤ p < +∞, J. H. Shapiro and P. D. Taylor proved in [42], Theorem 6.2,

that the condition:

dm

(3.3) < +∞

T 1 − |φ|

**implies that the composition operator Cφ : H p → H p is p-summing (and condition
**

(3.3) is necessary for 1 ≤ p ≤ 2; in particular, for p = 2, it is equivalent to say that

Cφ is Hilbert-Schmidt). Actually, they proved that (3.3) is equivalent to the fact

that Cφ is order bounded on H p , and (acknowledging to A. Shields, L. Wallen, and

J. Williams) every order bounded operator into an Lp -space is p-summing. The

6. p-SUMMING OPERATORS 27

**counterpart of (3.3) in our setting, are conditions (OB1) and (OB2)
**

1

χA dm < +∞

T 1 − |φ|

in Proposition 3.14. We are going to see that, if Ψ grows fast enough, order bound-

edeness does not imply that Cφ is p-summing. Note that, for composition operators

on H ∞ , being p-summing is equivalent to being compact ([23], Theorem 2.6), but

H ∞ corresponds to the very degenerate Orlicz function Ψ(x) = 0 for 0 ≤ x ≤ 1

and Ψ(x) = +∞ for x > 1, which does not match in the proof below.

Theorem 3.27. If Ψ ∈ Δ2 , then there exists a composition operator Cφ : H Ψ →

H which is order bounded into M Ψ (T), and hence compact, but which is p-

Ψ

summing for no p < ∞.

Note that every p-summing operator is Dunford-Pettis (it maps the weakly

convergent sequences into norm convergent sequences); therefore, when it starts

from a reﬂexive space, it is compact. However, when Ψ ∈ Δ2 , being Dunford-Pettis

implies compactness for composition operators on H Ψ , though H Ψ is not reﬂexive,

thanks to the next proposition and Theorem 3.24. Later (see Theorem 4.21), we

shall see that, under condition Δ0 , every Dunford-Pettis composition operator is

compact.

Proposition 3.28. When Ψ ∈ ∇2 , every Dunford-Pettis composition operator

satisﬁes condition (W ).

Proof. Let ga,r = Ψ−1 1/(1 − r) ua,r . If condition (W) were not satisﬁed, we

could ﬁnd a sequence (an )n≥1 in T and a sequence of numbers (rn )n≥1 tending to

1 such that Cφ (gan ,rn )Ψ ≥ δ > 0 for all n ≥ 1. But (1 − r)2 Ψ−1 1/(1 − r) −→ 0.

r→1

Therefore gan ,rn (z) = (1 − rn )2 Ψ−1 1/(1 − rn ) /(1 − ān rn z) tends to 0 uniformly

on compact sets of D. Hence, by Proposition 3.7, (gan ,rn )n≥1 tends weakly to 0

(because gan ,rn ∈ HM Ψ and, on HM Ψ, the weak-star Ψ

topology of H is the weak

topology). Since Cφ is Dunford-Pettis, Cφ (gan ,rn ) n≥1 tends in norm to 0, and we

get a contradiction, proving the proposition.

Proof of Theorem 3.27. We shall begin with some preliminaries. First, since

2

Ψ ∈ Δ2 , there exists α > 1 such that Ψ(x) ≤ Ψ(αx) for x large enough. Hence:

Ψ(x) x

≥Ψ −→ +∞.

Ψ(x/α) α x→+∞

Therefore, there exists, for every n ≥ 1, some xn > 0 such that:

Ψ(x)

≥ 2n , ∀x ≥ xn .

Ψ(x/α)

Then

x
1 x
1

n

Ψ ≤ n Ψ(x) + Ψ ≤ n Ψ(x) + Ψ(xn ) , ∀x > 0.

α 2 α 2

For convenience, we shall assume, as we may, that Ψ(xn ) ≥ 1.

Remark also that, setting a = Ψ−1 (1), one has, for every f ∈ L∞ :

|f |

Ψ a dm ≤ 1,

T f ∞

28 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

so that:

1

f Ψ ≤f ∞ .

a

We are now going to start the construction.

For n ≥ 1, let Mn = log(n + 1). Choose positive numbers βn which tend to 0

fast enough to have:

βk ≤ βn , ∀n ≥ 1,

k>n

and

Ψ−1 (8/βn )

tn = −→ +∞,

Mn n→+∞

**with (tn )n≥1 non-decreasing and Ψ(tn ) > 1.
**

Set:

1

rn = 1 − ·

Ψ(tn )

One has rn −→ 1 and:

n→+∞

1
8

χM n = ·

1 − rn βn

Actually, for the end of the proof, we shall have to choose the βn ’s decreasing

so fast that: n

t1 + · · · + tn−1
2 tn 1

Ψ + Ψ(xn ) ≤ n·

α Ψ(tn ) 2

This is possible, by induction, since t/Ψ(t) −→ 0. Note that, since Ψ(xn ) ≥ 1,

t→+∞

2n tn

one has, in particular: +∞

n=1 Ψ(tn ) < +∞.

** measurable subsets of T with measure m(Bn ) = cβn (where
**

Let Bn be disjoint

c ≥ 1 is such that n≥1 βn = 1/c), and whose union is T. Deﬁne h : T → C by:

h= rn IBn .

n≥1

**has log h ∈ L (T), since h does not vanish, rn ≥ 1/2 for n large enough, and
**

1

One

n m(Bn ) = 1 < +∞. We can deﬁne the outer function:

u+z

φ(z) = exp log h(u) dm(u) , |z| < 1.

T u−z

**φ is analytic on D and its boundary limit veriﬁes |φ| = h ≤ 1 on T. Hence φ deﬁnes
**

a composition operator on H Ψ .

For any A > 0, one has, when n is large enough to ensure Mn ≥ A, and when

rn ≤ r < rn+1 :

8c

m(|φ| > r) = m(|φ| = rk ) = c βk ≤ c βn =

k>n k>n

χ M n

1/(1 − r n )

8c 8c

≤ ≤ ·

χA 1/(1 − rn ) χA 1/(1 − r)

Since rn −→ 1, it follows from Theorem 3.15 that Cφ : H Ψ → H Ψ is order-

n→+∞

bounded into M Ψ (T) (and hence is compact).

6. p-SUMMING OPERATORS 29

**We are Ψ now going
**

to construct a sequence of functions gn ∈ H Ψ such that

p

n |gn | ∈ L , but n Cφ (gn )Ψ = +∞ for all 1 ≤ p < ∞. That will prove that

Cφ is p-summing for no p < ∞.

Since

m(|φ| ≥ rn ) ≥ m(|φ| = rn ) = c βn ≥ βn ,

we can apply Lemma 3.17 and Lemma 3.23 (which remain valid with non-strict

inequalities instead of strict ones), and we are able to ﬁnd, for every n ≥ 1, some

an ∈ T such that:

1/9

Cφ (uan ,rn )Ψ ≥
·

8

Ψ−1

(1 − rn )βn

But:

8

= Ψ(tn )Ψ(Mn tn ).

(1 − rn )βn

Since now Ψ satisﬁes Δ2 : [Ψ(x)]2 ≤ Ψ(αx), for x large enough and one has, for n

large enough, since Mn ≥ 1 (for n ≥ 2):

2

Ψ(tn )Ψ(Mn tn ) ≤ Ψ(Mn tn ) ≤ Ψ(αMn tn ).

Therefore:

1/9

Cφ (uan ,rn )Ψ ≥ ·

αMn tn

Taking now:

1

gn = Ψ−1 uan ,rn = tn uan ,rn ,

1 − rn

one has gn Ψ ≤ 1 (by Corollary 3.10), and

1/9 1/9

Cφ (gn )Ψ ≥ = ·

αMn α log(n + 1)

Therefore

+∞

Cφ (gn )pΨ = +∞

n=1

for every p ≥ 1.

It remains to show that g = n |gn | ∈ LΨ .

We shall follow the lines of proof of Theorem II.1 in [24].

By Markov’s inequality, one has:

2n tn

m(|gn | > 2−n ) ≤ 2n tn uan ,rn 1 ≤ 2n tn (1 − rn ) = ·

Ψ(tn )

Set:

An = {|gn | > 2−n } ; Ãn = An \ Aj ,

j>n

and

ğn = gn I{|gn |>2−n } .

Since:

+∞

1

+∞

1 1

+∞

1

gn − ğn Ψ ≤ gn − ğn ∞ ≤ n

= < +∞,

n=1

a n=1 a n=1 2 a

it suﬃces to show that ğ = n |ğn | ∈ L .

Ψ

30 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

But ğ vanishes out of n≥1 Ãn ∪ lim supn An , and m lim supn An = 0, since

+∞

+∞

2n tn

m(An ) ≤ < +∞ .

n=1 n=1

Ψ(tn )

Therefore:

|ğ|
+∞

|ğ|

Ψ dm = Ψ dm.

T 2α n=1 Ãn

2α

Since ğj = 0 on Ãn for j > n, we get:

+∞

|ğ | + · · · + |ğ |

|ğ| 1 n

Ψ dm = Ψ dm.

T 2α n=1 Ãn

2α

Now, by the convexity of Ψ:

|ğ | + · · · + |ğ |
1 |ğ | + · · · + |ğ
|ğ |

1 n 1 n−1 | n

Ψ ≤ Ψ +Ψ ·

2α 2 α α

But:

|ğ |
1

n

Ψ ≤ n Ψ(|ğn |) + Ψ(xn ) ,

α 2

and

|ğ | + · · · + |ğ
t + · · · + t

1 n−1 | 1 n−1

Ψ ≤Ψ ;

α

α

therefore, using that T Ψ(|ğn |) dm ≤ T Ψ(|gn |) dm ≤ 1:

+∞

|ğ| 1 t1 + · · · + tn−1

Ψ dm ≤ Ψ m(Ãn )

T 2α n=1

2 α

1

+ n Ψ(|ğn |) dm + Ψ(xn )m(Ãn )

2 T

+∞

1 t1 + · · · + tn−1
2n t

n 1

≤ Ψ + Ψ(xn ) +

n=1

2 α Ψ(tn ) 2n

+∞

1 1 1

≤ + = 1,

n=1

2 2n 2n

**which proves that ğ ∈ LΨ , and ğΨ ≤ 2α.
**

The proof is fully achieved.

Remark. In the above proof, we chose Mn = log(n + 1). This choice was only

used to conclude that n Cφ (gn )pΨ = +∞ for every p < ∞. Therefore, the above

proof shows that, given any increasing function Υ : (0, ∞) → (0, ∞) tending to

∞, we can ﬁnd, with a suitable choice of a slowly increasing

sequence (Mn )n≥1 ,

n n | ∈ L , although

|g

Ψ Ψ

a

symbol φ and a sequence (g )

n n≥1 in H such that

n Υ(C (g

φ n Ψ) ) = +∞.

**Remark. Proposition 3.14 obviously implies that, if the composition operator
**

Cφ : H Ψ → H Ψ is order bounded into M Ψ (T), then Cφ : H 2 → H 2 is order bounded

(i.e. Hilbert-Schmidt). One could expect that under such an hypothesis with a fast

growth condition for Ψ, Cφ : H 2 → H 2 is actually in a “better” Schatten class,

6. p-SUMMING OPERATORS 31

**nuclear for instance. This is the case when Ψ ∈ Δ2 : see 8) in Theorem 3.24. Is it
**

the case too under a weaker growth condition for Ψ?

In the same spirit, if the composition operator Cφ : H Ψ → H Ψ is compact and

Ψ veriﬁes a very fast growth condition (such as Δ2 for instance and Cφ is even

order bounded into M Ψ (T) in that situation), we could expect that Cφ : H Ψ → H Ψ

is actually nuclear as this is the case when H Ψ is replaced by H ∞ . But, this is false

as shown by Theorem 3.27: the composition operator cannot be nuclear since it is

not an absolutely summing operator.

CHAPTER 4

Carleson measures

1. Introduction

B. MacCluer ([30]; see also [10], Theorem 3.12) has characterized compact

composition operators on Hardy spaces H p (p < ∞) in term of Carleson measures.

In this chapter, we shall give an improvement of this result for Hardy-Orlicz spaces

H Ψ , but in terms of “Ψ-Carleson measures”. Indeed, Carleson measures do not

characterize the compactness of composition operators when Ψ grows too quickly,

as it follows from Corollary 3.26.

Before that, we shall recall some deﬁnitions (see for example [10], pages 37–38,

or [11], page 157).

Let ξ ∈ T and h ∈ (0, 1). Deﬁne

(4.1) S(ξ, h) = {z ∈ D ; |ξ − z| < h}.

The Carleson window W (ξ, h) is the following subset of D:

(4.2) W (ξ, h) = {z ∈ D ; 1 − h < |z| ≤ 1 and ¯ < h}.

|arg(z ξ)|

It is easy to show that we have for every ξ ∈ T and h ∈ (0, 1):

S(ξ, h/2) ⊆ W (ξ, h) and W (ξ, h/2) ⊆ S(ξ, h) ,

**so that, in the sequel, we may work equivalently with either S(ξ, h) or W (ξ, h).
**

Recall that a positive Borel measure μ on D (or D) is called a Carleson measure if

there exists some constant K > 0 such that:

μ S(ξ, h) ≤ Kh , ∀ξ ∈ T , ∀h ∈ (0, 1).

**Carleson’s Theorem (see [10], Theorem 2.33, or [11], Theorem 9.3) asserts that, for
**

0 < p < ∞, the Hardy space H p is continuously embedded into Lp (μ) if and only

if μ is a Carleson measure.

Given an analytic self-map φ : D → D, we deﬁne the pullback measure μφ on

the closed unit disk D (which we shall denote simply μ when this is unambiguous)

as the image of the Haar measure m of T = ∂D under the map φ∗ (the boundary

limit of φ):

(4.3) μφ (E) = m φ∗ −1 (E) ,

**for every Borel subset E of D.
**

The automatic continuity of composition operators Cφ on the Hardy space H p ,

combined with Carleson’s Theorem means that μφ is always a Carleson measure.

33

34 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

**B. MacCluer ([30], [10], Theorem 3.12) showed that:
**

The composition operator Cφ is compact on H 2

if and only if

(MC) μφ S(ξ, h) = o (h) as h → 0, uniformly for ξ ∈ T.

if and only if

μφ W (ξ, h) = o (h) as h → 0, uniformly for ξ ∈ T.

While the Shapiro’s compactness criterion, via the Nevanlinna counting func-

tion ([39]), deals with the behavior of φ inside the open unit disk, the character-

ization (M C) deals with its boundary values φ∗ . It is natural to wonder whether

the modulus of φ∗ on T = ∂D suﬃces to characterize the compactness of Cφ . This

leads to the following question: if two functions φ1 and φ2 have the same modulus

on T, are the compactness of the two associated composition operators equivalent?

We have seen in Theorem 3.24 that the answer is positive on H Ψ when Ψ ∈ Δ2 .

However, on H 2 it turns out to be negative. We give the following counterexample.

Theorem 4.1. There exist two analytic functions φ1 and φ2 from D into itself

such that |φ∗1 | = |φ∗2 | on T but for which the composition operator Cφ2 : H 2 → H 2

is compact, though Cφ1 : H 2 → H 2 is not compact.

Remark. Let Ψ be an Orlicz function which satisﬁes Δ2 . We shall see in Theorem

4.3 that every composition operator Cφ : H 2 → H 2 is compact as soon as Cφ : H Ψ →

H Ψ is compact. Hence, in the above theorem, Cφ1 : H Ψ → H Ψ is not compact.

It follows hence from Theorem 3.24, since φ∗1 and φ∗2 have the same modulus,

that Cφ2 : H Ψ → H Ψ is not compact (and, even, not weakly compact), though

Cφ2 : H 2 → H 2 is compact. We have already seen such a phenomenon in Corollary

3.26. However, the results ofthe next section

will allow us to conclude (Theorem

4.22) that, when Ψ(x) = exp log(x+1)2 −1, which does not satisfy condition Δ2 ,

but satisﬁes conditions Δ1 and ∇1 , the composition operator Cφ2 : H Ψ → H Ψ is

compact, but not order bounded into M Ψ (T). That will show that our assumption

that Ψ ∈ Δ2 in Theorem 3.24 is not only a technical one. Nevertheless, it will be

shown in [27] that Theorem 3.24 holds for a larger class of Orlicz functions.

We shall improve Theorem 4.1 in [26], Theorem 4.2, where we show that more-

over, for every p > 2, the composition operator Cφ2 can be taken in the Schatten

class Sp .

Proof. We start simply with φ1 (z) = 1+z 2 . It is well known that Cφ1 is not

compact on H 2 (this was ﬁrst observed in H. J. Schwartz’s thesis [37]: see [42],

page 471). Now, let:

1 + z

M (z) = exp −

1−z

and

φ2 (z) = φ1 (z) M (z).

For simplicity, we shall write φ = φ2 , and we are going to show that Cφ is a compact

operator on H 2 , using the criterion (MC).

Let ξ = eiα ∈ T, with |α| ≤ π. We are going to prove that:

μφ W (ξ, h) = O (h3/2 ).

For θ ∈ (−π, π), one has |φ(eiθ )| = |φ1 (eiθ )| = cos(θ/2), and so the condition

1 − h < |φ(eiθ )| ≤ 1 is equivalent to 1 − h < cos(θ/2) < 1, which implies, since

1. INTRODUCTION 35

**cos t = 1 − 2 sin2 (t/2) ≤ 1 − 2t2 /π 2 ≤ 1 − t2 /5 for 0 ≤ t ≤ π/2 (because sin t ≥ π2 t),
**

√

that 1 − h < 1 − (θ/2)

2

/5, i.e. θ 2 ≤ 20h and so |θ| ≤ 6 h. On the other

hand, M (eiθ ) = exp − i cot(θ/2) ; hence arg φ(eiθ ) = θ/2 − cot(θ/2), modulo 2π.

Therefore, for h small enough:

√

μφ W (ξ, h) ≤ m({|θ| ≤ 6 h ; | − α + θ/2 − cot(θ/2)| ≤ h, mod 2π})

√

≤2 m({|t| ≤ 3 h ; | − α + t − cot t + 2πn| ≤ h}).

n∈Z

√

We have to majorize both m({0 < t ≤ 3 h ; | − α + t − cot t + 2πn| ≤ h}) and

√ n∈Z

m({0 < t ≤ 3 h ; |α + t − cot t + 2πn| ≤ h}).

n∈Z

The function

F (t) = t − cot(t)

is increasing , and we deﬁne an , bn ∈ (0, π) by:

F (an ) = α − 2π(nh + n) − hand F (bn ) = α − 2π(nh + n) + h,

√

where the integer nh is given large enough to ensure that a0 ≤ 3 h. Of course,

a√ √ . Observe that 2h = F (b0 ) − F (a0 ) ≥ b0 − a0 , and then b0 ≤

n < bn < an−1

3 h + 2h ≤ 4 h for h small enough. One has:

√ ∞

m({0 < t ≤ 3 h ; | − α + t − cot t + 2πn| ≤ h}) ≤ (bn − an ).

n∈Z n=0

1 1

Since F (t) = 1 + 2 ≥ t2 , one has, on the one hand:

sin t

bn

bn − an

(4.4) 2h = F (bn ) − F (an ) = F (t) dt ≥ ;

an an bn

hence:

(4.5) bn − an ≤ 2han bn ≤ 2hb2n , for all n ≥ 0.

On the other hand, let us ﬁrst point out that, for 0 ≤ t ≤ 1,

1 π 2 /4 4

F (t) = 1 + 2 ≤ 1 + t2 ≤ 2·

sin t t

Hence, for h small enough:

bn

bn − bn+1

2π = F (bn ) − F (bn+1 ) = F (t) dt ≤ 4 ,

bn+1 bn+1 bn

and we get:

2

b2n+1 ≤ bn+1 bn ≤ (bn − bn+1 ).

π

Hence, using the fact that (4.4) gives b0 − a0 ≤ 2ha0 b0 ≤ 24 h2 , we get, from (4.5):

∞ ∞

4h

(bn − an ) ≤ (b0 − a0 ) + (bn − bn+1 )

n=0

π n=0

4h 4h √ √ 16
3/2

≤ 24 h2 + b0 ≤ 24 h2 + 4 h ≤ 24 h + h ≤ 6 h3/2 ,

π π π

for h small enough.

36 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

**In the same way, we have
**

∞ √

m({0 < t ≤ 3 h ; |α + t − cot t + 2πn| ≤ h}) ≤ 6 h3/2 .

n=0

We can hence conclude that μφ W (ξ, h) ≤ Ch3/2 , where C is a numerical

constant.

Remark. Cφ actually maps continuously H 2 into H 3 , and compactly H 2 into H p ,

for any p < 3 (see [15] or Theorem 4.10 and 4.11).

However, in some cases, the behaviour of |φ∗ | on the boundary ∂D suﬃces.

Proposition 4.2. Let φ1 and φ2 be two analytic self-maps of D such that

|φ∗1 | ≤ |φ∗2 | on ∂D. Assume that they are both one-to-one on D, and that there

exists a ∈ D such that φ2 (a) = 0. Then the compactness of Cφ2 : H 2 → H 2 implies

that of Cφ1 : H 2 → H 2 .

Proof. By composing φ1 and φ2 with the automorphism of D which maps 0 into

a, we may assume that a = 0. We can hence write φ2 (z) = zφ(z), with φ : D → C

analytic in D. φ does not vanish in D because of the injectivity of φ2 (this is obvious

for z = 0, and for z = 0, follows from the fact that the injectivity of φ2 implies

φ2 (0) = 0).

Then there is some δ > 0 such that |φ(z)| ≥ δ for every z ∈ D. In fact,

by continuity, there is some α > 0 and some 0 < r < 1 such that |φ(z)| ≥ α

and |φ2 (z)| ≤ α for |z| ≤ r. But being analytic and non constant, φ2 is an open

map, so there is some ρ > 0 such that ρD ⊆ φ2 (rD). Injectivity of φ2 shows that

φ2 (D \ rD) ∩ ρD = ∅, that is to say that |φ2 (z)| ≥ ρ for |z| > r. A fortiori |φ(z)| ≥ ρ

for |z| > r. The claim is proved with δ = min(α, ρ).

φ∗ φ∗ 1 (z)

Then φ1 ∈ H ∞ , as well as φφ1 . Since φ1∗ = φ1∗ ≤ 1, one has φφ(z) ≤ 1 for

2

every z ∈ D. Hence:

1 − |φ1 (z)| 1 − |φ(z)| 1 − |φ2 (z)|

≥ = − |φ(z)|.

1 − |z| 1 − |z| 1 − |z|

Now ([39], Theorem 3.5), the compactness of Cφ2 : H 2 → H 2 implies that

1 − |φ2 (z)|

lim− = +∞;

z→1 1 − |z|

we get:

1 − |φ1 (z)|

lim = +∞,

1 − |z|

z→1−

which implies the compactness of Cφ1 : H 2 → H 2 , thanks to the injectivity of φ1

([39], Theorem 3.2).

**2. Compactness on H Ψ versus compactness on H 2
**

The equivalence (MC) holds actually for every H p space (with p < ∞) instead

of H 2 . We are going to see in this chapter that for Hardy-Orlicz spaces H Ψ ,

one needs a new notion of Carleson measures, which one may call Ψ-Carleson

measures. Before that, we are going to see that condition (MC) allows to get that

the compactness of composition operators on H Ψ always implies that on H p for

p < ∞. Recall that, when Ψ ∈ Δ2 , we have seen in Corollary 3.26 that the converse

is not true.

2. COMPACTNESS ON H Ψ VERSUS COMPACTNESS ON H 2 37

**Theorem 4.3. Let φ : D → D be an analytic function. If one of the following
**

conditions:

i) Cφ is a compact operator on H Ψ

ii) Ψ ∈ Δ0 and Cφ is a weakly compact operator on H Ψ

is satisﬁed, the composition operator Cφ is compact on H 2 .

Note that we have proved in Theorem 3.24 that the weak compactness of

Cφ : H Ψ → H Ψ implies its compactness only when Ψ satisﬁes the Δ2 condition.

Nevertheless, we shall show in Theorem 4.21 that when Ψ ∈ Δ0 , the weak compact-

ness of Cφ is equivalent to its compactness. This is obviously false (in particular

when LΨ is reﬂexive) without any assumption on Ψ.

Proof. We are going to use the characterization (MC) for compact composition

operators on H 2 .

Suppose that the condition on μφ is not fulﬁlled. Then there exist β ∈ (0, 1),

ξn ∈ T, and hn ∈ (0, 1), with hn −→ 0, such that:

n→+∞

μφ S(ξn , hn ) ≥ βhn .

We are now going to use the functions:

h2n

vn (z) = ,

(1 − an z)2

where:

an = (1 − hn )ξn .

Of course, vn is actually nothing but uξn ,1−hn . We have by Corollary 3.10:

1

vn Ψ ≤ −1 ·

Ψ (1/hn )

Deﬁne gn = Ψ−1 (1/hn )vn , which is in the

unit

Ψ

ball of HM . We have assumed at

the beginning of the paper that x = o Ψ(x) as x → ∞; hence Ψ−1 (x) = o (x) as

x → ∞, and so h2n Ψ−1 (1/hn ) → 0. Therefore (gn )n converges uniformly to zero

on compact subsets of D and gn 1 → 0, because gn 1 ≤ hn Ψ−1 (1/hn ). Then, in

both cases, we should have Cφ (gn )Ψ → 0. Indeed, in case i), this follows from

Proposition 3.6, and in case ii), this follows from [25], Theorem 4.

We are going to show that this is not true. Indeed:

4 4

Ψ |gn ◦ φ| dm ≥ Ψ Ψ−1 (1/hn )|vn (z)| dμφ

T β β

D 4

≥ Ψ Ψ−1 (1/hn )|vn (z)| dμφ .

S(ξn ,hn ) β

But when z ∈ S(ξn , hn ), one has |vn (z)| ≥ 1/4, because

|1 − an z| ≤ |1 − an ξn | + |an (ξn − z)| = hn + (1 − hn )hn ≤ 2hn .

We obtain that by convexity (since β < 1):

4 1 −1

Ψ |gn ◦ φ| dm ≥ Ψ 4Ψ (1/hn )|vn (z)|) dm

T β T β

1

≥ μφ S(ξn , hn ) ≥ 1.

βhn

This implies that Cφ (gn )Ψ ≥ β/4 and proves the theorem.

38 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

3. General measures

We used several times the criterion (MC) for compactness on H 2 via Carleson

measures. The fact that this provides such a useful tool leads to wonder if the

boundedness and the compactness on Hardy-Orlicz spaces can be expressed in such

a pleasant manner. Theorem 4.10 and Theorem 4.11 are the Orlicz version of

Carleson’s Theorem for H p spaces ([10], Theorem 2.35).

The key for our general characterization is the use of the following functions

(see (4.2) for the deﬁnition of the Carleson’s window W (ξ, h)).

Definition 4.4. For any positive ﬁnite Borel measure μ on the unit disk D (or

on D), we set, for h ∈ (0, 1]:

(4.6) ρμ (h) = sup μ W (ξ, h) ,

ξ∈T

ρμ (t)

(4.7) Kμ (h) = sup

0<t≤h t

Hence μ W (ξ, t) ≤ Kμ (h)t for t ≤ h.

The measure μ is a Carleson measure if and only if Kμ (h) is bounded by a

constant K, for h ∈ (0, 1) and this happens as soon as Kμ (h0 ) is ﬁnite for some

h0 ∈ (0, 1).

In Theorem 4.11, we shall use the following condition.

Definition 4.5. We say that Ψ satisﬁes the ∇0 condition if for some x0 > 0,

C ≥ 1 and every x0 ≤ x ≤ y, one has:

Ψ(2x) Ψ(2Cy)

(4.8) ≤ ·

Ψ(x) Ψ(y)

This is a condition on the regularity of Ψ. It is satisﬁed if

Ψ(2x) Ψ(2y)

≤C ·

Ψ(x) Ψ(y)

Before stating our main theorems, let us give some results about this condition.

Proposition 4.6. The following assertions are equivalent

i) Ψ satisﬁes the ∇0 condition.

ii) There exists some x0 > 0 satisfying: for every β > 1, there exists Cβ ≥ 1

such that

Ψ(βx) Ψ(βCβ y)

≤ , for every x0 ≤ x ≤ y.

Ψ(x) Ψ(y)

iii) There exist x0 > 0, β > 1 and Cβ ≥ 1 such that

Ψ(βx) Ψ(βCβ y)

≤ , for every x0 ≤ x ≤ y.

Ψ(x) Ψ(y)

Proof. We only have to prove i) ⇒ ii), since iii) ⇒ i) is similar and ii) ⇒ iii) is

trivial.

If β ∈ (1, 2], it is easy, taking Cβ = 2C/β. Now, if β ∈ (2b , 2b+1 ] for some

integer b ≥ 1, we write for every x0 ≤ x ≤ y:

Ψ(βx) Ψ(2b+1 x) Ψ(2b+1 x) Ψ(2x)

≤ = b

··· ·

Ψ(x) Ψ(x) Ψ(2 x) Ψ(x)

3. GENERAL MEASURES 39

**But we have for every integer j ≥ 1: 2j−1 x ≤ (2C)j−1 x ≤ (2C)j−1 y, so:
**

Ψ(2j x) Ψ((2C)j y)

j−1

≤ ,

Ψ(2 x) Ψ((2C)j−1 y)

and we obtain:

Ψ(βx) Ψ((2C)b+1 y) Ψ(βCβ y)

≤ ≤ ,

Ψ(x) Ψ(y) Ψ(y)

where Cβ = 2C b+1 .

Examples. It isimmediately α seen that the following functions satisfy ∇0 : Ψ(x) =

α

xp , Ψ(x) = exp log(x + 1) − 1, Ψ(x) = ex − 1, α ≥ 1.

Note that when Ψ ∈ ∇0 , with constant C = 1, i.e. Ψ(βx)/Ψ(x) is increasing

for x large enough, then we have the dichotomy: either Ψ ∈ Δ2 , or Ψ ∈ Δ0 .

We shall say that ∇0 is uniformly satisﬁed if there exist C ≥ 1 and x0 > 0 such

that, for every β > 1:

Ψ(βx) Ψ(Cβy)

(4.9) ≤ for x0 ≤ x ≤ y.

Ψ(x) Ψ(y)

One has:

Proposition 4.7.

1) Condition Δ2 implies condition ∇0 uniformly.

2) If Ψ ∈ ∇0 uniformly, then Ψ ∈ ∇1 .

3) The function κ(x) = log Ψ(ex ) is convex on (x0 , +∞) if and only if, in ( 4.9),

∇0 is satisﬁed with constant C = 1.

We shall say that Ψ is κ-convex when κ is convex at inﬁnity. Note that Ψ is

κ-convex whenever Ψ is log-convex. In the above examples Ψ is κ-convex; it also

the case of Ψ(x) = x2 / log x, x ≥ e; but, on the other hand, if Ψ(x) = x2 log x for

x ≥ e, then Ψ is not κ-convex. Nevertheless, for β 2 ≤ x ≤ y, one has:

Ψ(βx) log β
3β 2 3 Ψ(βy)

= β2 1 + ≤ ≤ ,

Ψ(x) log x 2 2 Ψ(y)

and hence Ψ ∈ ∇0 .

We do not know whether Ψ ∈ ∇0 uniformly implies that Ψ is equivalent to an

Orlicz function for which the associated function κ is convex.

2

Proof. 1) Since Ψ ∈ Δ2 , one has Ψ(u) ≤ Ψ(αu) for some α > 1 and x ≥ x0 .

We may assume that Ψ(x0 ) ≥ 1. Then, for y ≥ x ≥ x0 and every β > 1:

2

Ψ(βx)Ψ(y) ≤ Ψ(βy) ≤ Ψ(αβy) ≤ Ψ(αβy)Ψ(x),

which is (4.9).

2) Suppose that Ψ satisﬁes condition ∇0 uniformly. We may assume that

Ψ(x0 ) ≥ 1. Let x0 ≤ u ≤ v; we can write u = βx0 for some β ≥ 1. Then

condition (4.9) gives:

Ψ(u)Ψ(v) = Ψ(βx0 )Ψ(v) ≤ Ψ(x0 )Ψ(Cβv) ≤ Ψ Ψ(x0 )Cβv = Ψ(buv),

with b = CΨ(x0 )/x0 .

3) Assume that κ is convex on (x0 , +∞). For every β > 1, let κβ (t) =

κ(t log β) = log(Ψ(β t )), which is convex on (x0 / log(β), +∞). Taking y ≥ x ≥ ex0 ,

40 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

write x = β θ and y = β θ with θ ≤ θ . Convexity of κ gives, since θ ≥ θ ≥

x0 / log(β):

κβ (θ + 1) − κβ (θ) ≤ κβ (θ + 1) − κβ (θ ),

which means that:

Ψ(βx) Ψ(βy)

≤ ·

Ψ(x) Ψ(y)

Assume that (4.9) is fulﬁlled for every β > 1, with C = 1. Then, taking y = βx,

one has:

2

Ψ(βx) ≤ Ψ(x)Ψ(β 2 x).

Let u < v be large enough. Taking x = u2 and β = v/u, we get:

2

Ψ(uv) ≤ Ψ(u2 )Ψ(v 2 ),

which means that κ is convex.

Remark. The growth and regularity conditions for Ψ can be expressed in the

following form:

• Ψ ∈ Δ0 iﬀ κ(x + β ) − κ(x) −→ +∞, for some β > 0.

x→∞

• Ψ ∈ Δ1 iﬀ for some β > 0, one has x + κ(x) ≤ κ(x + β ), for x large

enough.

• Ψ ∈ Δ2 iﬀ for some α > 0, one has 2κ(x) ≤ κ(x + α ), for x large enough.

• Ψ ∈ ∇1 iﬀ κB (x)+κB (y) ≤ κB (x+y) for x, y large enough, with B = e−b .

• Ψ ∈ ∇0 iﬀ for some c ≥ 1 and A > 1, one has κA (θ + 1) − κA (θ) ≤

κA (θ + c ) − κA (θ ) for θ ≤ θ large enough.

Before proving the main results of this chapter, let us collect some basic facts

on the compactness of the embedding of H Ψ1 into LΨ2 (μ). First:

Lemma 4.8. Let Ψ1 , Ψ2 be two Orlicz functions and μ a ﬁnite Borel measure

on D. Assume that the identity maps H Ψ1 into LΨ2 (μ) compactly. Then, μ(T) = 0.

Proof. The sequence (z n )n≥1 is weakly null in M Ψ1 by Riemann-Lebesgue’s

Lemma. Its image by a compact operator is then norm null. This implies that

for every ε ∈ (0, 1), we have, for n large enough,

Ψ2 |z|n dμ ≤ ε.

D

Fatou’s Lemma yields Ψ2 (1)μ(T) ≤ ε.

**Now, we summarize what is true in full generality about compactness for canon-
**

ical embeddings.

Proposition 4.9. Let Ψ1 , Ψ2 be two Orlicz functions and μ a ﬁnite Borel

measure on D. The following assertions are equivalent

i) The identity maps H Ψ1 into LΨ2 (μ) compactly.

ii) Every sequence in the unit ball of H Ψ1 , which is convergent to 0 uniformly

on every compact subset of D, is norm-null in LΨ2 (μ).

iii) The identity maps H Ψ1 into LΨ2 (μ) continuously and

lim− Ir = 0, where Ir (f ) = f ID\rD .

r→1

3. GENERAL MEASURES 41

**Proof. i) ⇒ ii): let (fn )n≥1 be a sequence in the unit ball of H Ψ1 , which is
**

uniformly convergent to 0 on every compact subset of D. In particular, fn (z)

converges to 0 for every z ∈ D. This means that (fn )n≥1 converges to 0 μ-almost

everywhere, since μ(T) = 0 by the preceding lemma. If the conclusion did not hold,

we could assume, up to an extraction, that limfn Ψ2 > 0. Thus by compactness

of the embedding, up to a new extraction, (fn )n≥1 is norm-convergent to some

g ∈ LΨ2 (μ). Necessarily g = 0. A subsequence of (fn )n≥1 would be convergent to

g μ-almost everywhere. This gives a contradiction.

ii) ⇒ iii): if not, there exist a sequence (fn )n≥1 in the unit ball of H Ψ1 and

δ > 0 with fn ID\(1− 1 )D Ψ2 > δ, for every n ≥ 1. Let us introduce the sequence

n

gn (z) = z n fn (z) for z ∈ D. The sequence (gn )n≥1 lies in the unit ball of H Ψ1 and

is convergent to 0 uniformly on every compact subset of D. But

1
n 1
n

gn Ψ2 ≥ z n fn ID\(1− 1 )D Ψ2 ≥ 1 − fn ID\(1− 1 )D Ψ2 ≥ 1 − δ.

n n n n

This contradicts (ii).

iii) ⇒ ii) is very easy.

ii) ⇒ i) follows from Proposition 3.6.

**We can now state some deeper characterizations.
**

Theorem 4.10. Let μ be a ﬁnite Borel measure on the closed unit disk D and

let Ψ1 and Ψ2 be two Orlicz functions. Then:

1) If the identity maps H Ψ1 into LΨ2 (μ) continuously, there exists some A > 0

such that:

1

(R) ρμ (h) ≤ for every h ∈ (0, 1].

Ψ2 AΨ−11 (1/h)

**2) If there exists some A > 0 such that:
**

1/h

(K) Kμ (h) ≤ for every h ∈ (0, 1],

Ψ2 AΨ−1

1 (1/h)

then the identity maps H Ψ1 into LΨ2 (μ) continuously.

**Theorem 4.11. Let μ be a ﬁnite Borel measure on the closed unit disk D and
**

let Ψ1 and Ψ2 be two Orlicz functions. Then:

1) If the identity maps H Ψ1 into LΨ2 (μ) compactly, then

1

(R0 ) ρμ (h) = o −1

as h → 0, for every A > 0.

Ψ2 AΨ1 (1/h)

2) If μ(T) = 0 and

1/h

(K0 ) Kμ (h) = o −1

as h → 0 , for every A > 0,

Ψ2 AΨ1 (1/h)

**then the identity maps H Ψ1 into LΨ2 (μ) compactly.
**

3) When Ψ1 = Ψ2 = Ψ satisﬁes condition ∇0 , then the above conditions are

equivalent: the identity maps H Ψ1 into LΨ2 (μ) compactly if and only if condition

(R0 ) is satisﬁed and if and only if condition (K0 ) is satisﬁed.

42 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

Remarks.

1. a) If ρμ (h) ≤ C/Ψ2 AΨ−1 1 (1/h) with C > 1, then the convexity of Ψ2 gives

Ψ2 (t/C) ≤ Ψ2 (t)/C and hence:

1

ρμ (h) ≤ A −1 ·

Ψ2 C Ψ1 (1/h)

b) If ρμ (h) ≤ 1/Ψ2 AΨ−1 1 (1/h) only for h ≤ hA , one can ﬁnd some C =

CA > 0 such that ρμ (h) ≤ C/Ψ2 AΨ−1 1 (1/h) for every h ∈ (0, 1]. In fact, ρμ (h) ≤

μ(D) and 1/Ψ2 AΨ−1 (1/h) ≥ 1/Ψ AΨ −1

(1/h ) for h ≥ hA ; hence ρμ (h) ≤

−1

1 2

1−1 A

C/Ψ2 AΨ1 (1/h) with C = μ(D)/Ψ2 AΨ1 (1/hA ) .

The same remark applies for Kμ .

2. a) In the case where Ψ1 = Ψ2 = Ψ, one has: Ψ AΨ−1 (t) ≤ Ψ Ψ−1 (t) = t

when A ≤ 1 and Ψ AΨ−1 (t) ≥ Ψ Ψ−1 (t) = t when A ≥ 1. On the other hand,

if Ψ ∈ Δ2 , one has, for some constant C = CA > 0: Ψ(Ax) ≤ CΨ(x), when A ≥ 1

and Ψ(Ax) ≥ (1/C)Ψ(x) when A ≤ 1. Hence, when Ψ ∈ Δ2 , one has, for every

A > 0:

1 1

= ≈h

−1

Ψ AΨ (1/h) χ (1/h)

A

and Theorem 4.10 is nothing but Carleson’s Theorem.

b) If Ψ1 (x) = xp and Ψ2 (x) = xq with p < q < ∞, then:

Ψ2 AΨ−11 (t) = A t

q q/p

,

and condition (R) means that μ is a β-Carleson measure, with β = q/p (see [11],

Theorem 9.4).

Ψ1 (x)

c) If, for ﬁxed A > 0, the function x → is non increasing, at least for

Ψ2 (Ax)

x large enough, conditions (R) and (K) (resp. conditions (R0 ) and (K0 ) below)

are clearly equivalent. This is the case in the framework of classical Hardy spaces:

Ψ1 (x) = xp and Ψ2 (x) = xq , with q ≥ p.

When Ψ1 = Ψ2 = Ψ, this is equivalent, if A > 1, to the convexity of the

function κ(x) = log Ψ(ex ) (see Proposition 4.7).

3. a) When Ψ1 = Ψ2 = Ψ, the condition μ(T) = 0 is automatically fulﬁlled (and

so can be removed from (K0 )). This follows on the one hand from the majorization

in (K0 ), which implies that Kμ (h) → 0 (when h → 0); and on the other hand from

the inequality:

2π

μ D \ rD ≤ ρμ (1 − r) ≤ 2πKμ (1 − r).

1−r

Indeed, D \ rD can be covered by less than 1−r 2π

Carleson’s windows of “size” 1 − r.

b) Nevertheless, the condition μ(T) = 0 cannot be removed in full generality in

Theorem 4.11. Indeed, if we consider the identity j from H 4 into L2 (D, m̃), where

m̃ is 0 on D and its restriction to the torus is the normalized Lebesgue measure. It

1

is easily seen that K(h) is bounded and so less than 2 1/2 , for h small enough.

A h

But j is not compact.

4. In the case where Ψ1 = Ψ2 = Ψ and μ is a Carleson measure, then Kμ is

bounded, by say K ≥ 1, and condition (K) is satisﬁed for A = 1/K,

since A ≤ 1

implies, by the convexity of Ψ: Ψ AΨ−1 (1/h) ≤ AΨ Ψ−1 (1/h) = A/h. Hence

the canonical embedding H Ψ → LΨ (μ) is continuous. We get hence, by Carleson’s

Theorem ([11], Theorem 9.3):

3. GENERAL MEASURES 43

**Proposition 4.12. Let μ be a positive ﬁnite measure on D. Assume that the
**

canonical embedding jμ : H p → Lp (μ) is continuous for some 0 < p < ∞. Then

jμ : H Ψ → LΨ (μ) is continuous.

Note that this is actually a consequence of the fact that H Ψ is an interpolation

space for H 1 and H ∞ (see [4], Theorem V.10.8).

When μ = μφ is the image of the Haar measure m under φ∗ , where φ is an

analytic self-map of D, we know (Proposition 3.12) that the composition operator

Cφ : H Ψ → H Ψ is always continuous. This can be read as the continuity of H Ψ →

LΨ (μφ ). Hence condition (R) must be satisﬁed, for some A > 0. Note that for

A ≤ 1, 1/χA (1/h) ≥ h, and so condition (R) is implied by the fact that μφ is a

Carleson measure.

5. To have the majorization in condition (K0 ), it suﬃces that: for every A > 0,

there exists hA ∈ (0, 1] such that

1/h

(4.10) Kμ (h) ≤ , for every h ∈ (0, hA ].

Ψ2 AΨ−1

1 (1/h)

In fact, ﬁxing A > 0 and ε ∈ (0, 1), we have, by convexity:

Ψ2 (Ax) ≤ εΨ2 (Ãx),

with Ã = A/ε. Since we have (4.10) with Ã, when h is small enough (depending

on A and ε), we get, for x = Ψ−1

1 (1/h), condition (K0 ).

**To prove both Theorem 4.10 and Theorem 4.11, we shall need some auxiliary
**

results. The following is actually the heart of the classical Theorem of Carleson,

though it is not usually stated in this form. The maximal (non-tangential) function

Mf (which is essentially the same as Nα f in the previous chapter) will be deﬁned

by:

Mf eiθ = sup{|f (z)|; z ∈ Gθ },

where

Gθ = {z ∈ D ; |eiθ − z| < 3(1 − |z|)}.

Theorem 4.13 (Carleson’s Theorem). For every f ∈ H 1 and every ﬁnite pos-

itive measure μ on the closed unit disk D, one has, for every h ∈ (0, 1] and every

t > 0:

μ {z ∈ D ; |z| > 1 − h and |f (z)| > t} ≤ 2πKμ (h) m({Mf > t}).

As this theorem is not usually stated in such a way, we shall give a few words

of explanations.

Proof. For convenience , we shall denote, when I is a subarc of T:

z

W (I) = {z ∈ D ; |z| > max(0, 1 − |I|/2) and ∈ I}.

|z|

Obviously, when |I| ≤ 2, we have W (I) = W (ξ, |I|/2), where ξ is the center of I.

Since it is clear that z ∈ Gθ when z = |z| eiθ , the condition |f (z)| > t implies

z

that Mf |z| > t. Hence it suﬃces to majorize the measure of the set

z

M = {z ∈ D ; |z| > 1 − h and Mf > t}.

|z|

The open set {Mf > t} is the disjoint union of a countable family of open arcs

Ij of T. So every z ∈ M belongs to some W (Ij ).

44 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

**Now, when |Ij | > 2h, we may cover Ij in such a way that we can write Ij ⊂
**

Jj,1 ∪ · · · ∪ Jj,Nj with the arcs Jj,1 , . . . , Jj,Nj satisfying: |Jj,k | = 2h for every k =

Nj

1, . . . , Nj and 2|Ij | ≥ |Ij | + 2h ≥ |Jj,k |.

k=1

When |Ij | ≤ 2h, we simply deﬁne: Nj = 1 and J1,Nj = Ij .

We then notice that, for every j, we have by deﬁnition of Kμ (h):

|Jj,k |

(4.11) μ W (Jj,k ) ≤ Kμ (h) ·

2

Given E ⊆ D, let us denote by Eh the set of points z ∈ E such that |z| > 1 − h.

Since, for every j:

W (Ij ) h ⊆ W (Jj,k ),

1≤k≤Nj

we get, using (4.11):

N N

j

j

1

μ W (Ij )h ≤ μ W (Jj,k ) ≤ Kμ (h) |Jj,k | ≤ Kμ (h) |Ij |.

2

k=1 k=1

On the other hand, M ⊆ W (Ij ) h .

j

It follows that:

μ(M) ≤ μ W (Ij )h ≤ Kμ (h) |Ij |

j j

= 2πKμ (h) m(Ij ) = 2πKμ (h) m({Mf > t}),

j

**remembering that {Mf > t} is the disjoint union of the Ij ’s.
**

We obtain

μ {z ∈ D ; |z| > 1 − h and |f (z)| > t} ≤ 2πKμ (h) m({Mf > t}) ,

as announced.

The following estimation will be useful for the study both of boundedness and

compactness.

Lemma 4.14. Let μ be a ﬁnite Borel measure on D. Let Ψ1 and Ψ2 be two

Orlicz functions. Suppose that there exists A > 0 and hA ∈ (0, 1) such that

1/h

Kμ (h) ≤ , for every h ∈ (0, hA ).

Ψ2 AΨ−1

1 (1/h)

**Then, for every f ∈ H Ψ1 such that f Ψ1 ≤ 1 and every Borel subset E of D, we
**

have: A

π

Ψ2 |f | dμ ≤ μ(E)Ψ2 (xA ) + Ψ1 (Mf ) dm

E 8 2 T

A

where xA = Ψ−1 (1/hA ).

2 1

Proof. For every s > 0 and every z ∈ D, the inequality |f (z)| > s implies that the

norm of the evaluation at z is greater than s; hence by Lemma 3.11:

1

s < 4Ψ−1 ,

1

1 − |z|

3. GENERAL MEASURES 45

i.e.:

1

|z| > 1 − ·

Ψ1 (s/4)

Obviously, this inequality still holds if z ∈ T.

Carleson’s Theorem (Theorem 4.13) gives:

1

μ {|f (z)| > s} ≤ 2πKμ m({Mf > s})

Ψ1 (s/4)

when Ψ1 (s/4) ≥ 1. Hence:

A
∞

Ψ2 |f | dμ = Ψ2 (t) μ({|f | > 8t/A} ∩ E) dt .

E 8 0

But our hypothesis means that, when Ψ1 (s/4) > 1/hA :

1
Ψ1 (s/4)

Kμ ≤ ·

Ψ1 (s/4) Ψ2 (As/4)

We have then

Ψ1 (2t/A)

μ {|f (z)| > 8t/A} ≤ 2π m({Mf > 8t/A}).

Ψ2 (2t)

So:

A
xA

Ψ2 |f | dμ ≤ Ψ2 (t)μ(E) dt

E 8 0

+∞

Ψ1 (2t/A)

+ 2π Ψ2 (t) m({Mf > 8t/A}) dt

xA Ψ2 (2t)

≤ Ψ2 (xA )μ(E)

+∞

Ψ2 (t)

+ 2π Ψ1 (2t/A) m({Mf > 8t/A}) dt .

xA Ψ2 (2t)

For the second integral, note that one has Ψ(x) ≤ xΨ (x) ≤ Ψ(2x), for any

Orlicz function Ψ. This leads to:

∞

Ψ2 (t)

Ψ1 (2t/A) m({Mf > 8t/A}) dt

xA Ψ2 (2t)

∞

Ψ1 (2t/A)

≤ m({Mf > 8t/A}) dt

t

0

2 ∞

≤ Ψ1 (2t/A) m({Mf > 8t/A}) dt

A 0

∞ 1

= Ψ1 (x) m({Mf > 4x}) dx = Ψ1 Mf dm

T 4

0

1

≤ Ψ1 (Mf ) dm.

4 T

which leads to the desired result.

**For the proofs of Theorem 4.10 and Theorem 4.11, we may restrict ourselves to
**

the case of functions Ψ1 and Ψ2 satisfying ∇2 . Indeed, suppose that Ψ1 and Ψ2 are

Orlicz functions and deﬁne Ψ j (t) = Ψj (t2 ), for j ∈ {1, 2}. The functions Ψ

1 and

**Ψ2 are Orlicz functions satisfying ∇2 since, with β = 2, we have for every t ≥ 0:
**

j (βt) = Ψj (4t2 ) ≥ 4Ψj (t2 ) = 2β Ψ

Ψ j (t).

46 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

Now, we claim that μ satisﬁes (R), (K), (R0 ) or (K0 ) for the couple Ψ1 , Ψ2

if and only if μ satisﬁes it for the couple Ψ 2 . This is simply due to the fact

1 , Ψ

that for every A > 0 and t ≥ 0, we have

2 AΨ

Ψ 1 −1 (t) = Ψ2 A2 Ψ−1 (t) .

1

**Moreover, notice that, writing f = Bg 2 (where B is a Blaschke product), we have
**

f ∈ H Ψ if and only if g ∈ H Ψ ; thus gLΨ = f LΨ . It is then clear that

2

Id : H Ψ1 −→ LΨ2 (μ) = Id : H Ψ1 −→ LΨ2 (μ) ,

so that the canonical embedding is bounded (resp. compact) for the couple Ψ1 , Ψ2

if and only if it is so for the couple Ψ1 , Ψ

2 , thanks to Proposition 4.9.

**Proof of Theorem 4.10. 1) Let C be the norm of the canonical embedding
**

j : H Ψ1 → LΨ2 (μ), and let ξ ∈ T and h ∈ (0, 1). It suﬃces to test the continuity of

j on f = Ψ−1 1 (1/h)uξ,1−h , which is in the unit ball of HM

Ψ1

, by Corollary 3.10.

But, when z ∈ W (ξ, h) one has, with a = (1 − h)ξ:

z z

|1 − āz| ≤ |1 − āξ| + |āξ − āz| ≤ h + (1 − h) ξ − + − z

|z| |z|

≤ h + (1 − h)[h + (1 − |z|)] ≤ h + (1 − h)[h + h] ≤ 3h;

hence |uξ,1−h (z)| ≥ 1/9 and |f (z)| ≥ (1/9)Ψ−11 (1/h); therefore:

|f |
1

1≥ Ψ2 dμ ≥ Ψ2 Ψ−1

1 (1/h) μ W (ξ, h) ,

D C 9C

which is (R).

2) By Proposition 3.5, the maximal (non-tangential) function M is bounded

on LΨ1 (T): there exists a constant C ≥ 1 such that Mf Ψ1 ≤ Cf Ψ1 for every

f ∈ LΨ1 (T). We ﬁx f in the unit ball of H Ψ1 (note that f /CΨ1 remains ≤ 1)

and use Lemma 4.14, with E = D and f replaced by f /C (here hA = 1)). Writing

π

C̃ = + μ(D)Ψ2 (xA ), we get:

2

A
A

1

Ψ2 |f | dμ ≤ Ψ2 |f | dμ

D 8C C̃ C̃ D 8C

1 π 1

≤ μ(D)Ψ2 (xA ) + Ψ1 Mf dm

C̃ 2 T C

1 π

≤ μ(D)Ψ2 (xA ) + = 1,

C̃ 2

8C C̃

which means that f LΨ2 (μ) ≤ ·

A

**Proof of Theorem 4.11. 1) Suppose that the embedding is compact, but that
**

condition (R0 ) is not satisﬁed. Then there exist ε0 ∈ (0, 1), A > 0, a sequence of

positive numbers (hn )n decreasing to 0, and a sequence of ξn ∈ T, such that:

ε0

μ W (ξn , hn ) ≥ ·

Ψ2 AΨ−1

1 (1/hn )

3. GENERAL MEASURES 47

**Consider the sequence of functions
**

h2n

fn (z) = Ψ1 −1 (1/hn ) = Ψ1 −1 (1/hn ) uξn ,|an | ,

(1 − ān z)2

where an = (1−hn )ξn . By Corollary 3.10, fn is in the unit ball of HM Ψ1 . Moreover,

it is plain that (fn )n converges to 0 uniformly on every compact subset of D. By the

compactness criterion (Proposition 3.6), (fn )n is norm-converging to 0 in LΨ2 (μ).

But, as above (proof of Theorem 4.10), for every n ≥ 1, one has |fn (z)| ≥

(1/9)Ψ−1 1 (1/hn ) when z ∈ W (ξn , hn ); hence:

9A
A

Ψ2 |fn | dμ ≥ Ψ2 Ψ−1

1 (1/hn ) μ W (ξn , hn )

D ε0 ε0

A
ε0

≥ Ψ2 Ψ−1 (1/h n ) ≥ 1,

ε0 1 Ψ2 AΨ−1

1 (1/hn )

**by the convexity of Ψ2 . This implies that fn LΨ2 (μ) ≥ ε0 /9A and gives a contra-
**

diction.

2) We have to prove that for every ε > 0, there exists an r ∈ (0, 1) such

that the norm of the injection Ir : H Ψ1 −→ LΨ2 (D \ rD, μ) is smaller than ε (see

Proposition 4.9).

Let C ≥ 1 be the norm of the maximal operator, as in the proof of Theo-

rem 4.10: Mf Ψ1 ≤ Cf Ψ1 for every f ∈ LΨ1 (T), and set A = 16C/ε. Condition

(K0 ) gives us hA ∈ (0, 1) such that:

1 1/h

Kμ (h) ≤

2 Ψ2 AΨ−1

1 (1/h)

when h ≤ hA .

Let f in the unit ball of H Ψ1 and r ∈ (0, 1). By Lemma 4.14:

|f |
A
A

1

Ψ2 dμ = Ψ2 |f | dμ ≤ Ψ2 |f | dμ

D\rD ε D\rD 16C 2 D\rD 8C

1 π Mf

≤ μ D \ rD Ψ2 (xA ) + Ψ1 dm

2 2 T C

π

≤ + Ψ2 (xA )μ D \ rD .

4

As μ(T) = 0, there exists some r0 ∈ (0, 1) such that π4 + Ψ2 (xA )μ D \ rD ≤ 1, for

every r ∈ (r0 , 1). This ends the proof of 2).

3) Assume that Ψ satisﬁes condition ∇0 and that condition (R0 ) is fulﬁlled:

for every A > 0 and every h ∈ (0, hA ) (hA small enough), we have:

1

ρμ (h) ≤ ·

Ψ AΨ−1 (1/h)

This implies that:

ρμ (s) 1/s Ψ(x)

Kμ (h) = sup ≤ sup = sup ·

s −1

0<s≤h 0<s≤h Ψ AΨ (1/s) x≥Ψ−1 (1/h) Ψ(A x)

**Fix an arbitrary β > 1 and choose A = βCβ > 1, where Cβ is given by the
**

∇0 condition for Ψ and Proposition 4.6. We have, for h small enough and x ≥

48 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

Ψ−1 (1/h):

Ψ βΨ−1 (1/h) Ψ(βCβ x) Ψ(A x)

≤ = ;

−1

Ψ Ψ (1/h) Ψ(x) Ψ(x)

we get hence, for h small enough:

Ψ Ψ−1 (1/h) 1/h

Kμ (h) ≤ −1 = ,

Ψ βΨ (1/h) Ψ βΨ−1 (1/h)

and condition (K0 ) is fulﬁlled. Keeping in mind that μ(T) = 0, by Remark 3.a.,

the two ﬁrst part 1) and 2) imply the result.

**Remark. Actually, the proof of Theorem 4.10, 1) gives, for every measure μ on D
**

and every Orlicz function Ψ:

|uξ,1−h |
1

1≥ Ψ dμ ≥ Ψ μ W (ξ, h) ,

D uξ,1−h LΨ (μ) 9 uξ,1−h LΨ (μ)

and hence:

1

(4.12) μ W (ξ, h) ≤
·

1

Ψ

9 uξ,1−h LΨ (μ)

In particular, if μ = μφ is the image of the Haar measure m under a self-map φ of

D, one has:

1

(4.13) μ W (ξ, h) ≤
·

1

Ψ

9 Cφ (uξ,1−h )Ψ

Condition (4.12) allows to have an upper control for the μ-measure of Carleson

windows, with uξ,1−h LΨ (μ) . It is possible, conversely, to majorize these norms.

Definition 4.15. We shall say that a measure μ on D is a Ψ-Carleson measure

if there exists some A > 0 such that

1

μ W (ξ, h) ≤ −1

, for every ξ ∈ T and every h ∈ (0, 1).

Ψ AΨ (1/h)

We shall say that a measure μ on D is a vanishing Ψ-Carleson measure if, for

every A > 0,

lim Ψ AΨ−1 (1/h) .ρμ (h) = 0.

h→0

Equivalently if, for every A > 0, there exists hA ∈ (0, 1) such that

1

μ W (ξ, h) ≤ −1

, for every ξ ∈ T and every h ∈ (0, hA ).

Ψ AΨ (1/h)

Remarks. 1) Let us precise how the second assertion implies the ﬁrst one: this

follows from the convexity of Ψ and the fact that the condition involves arbitrary

A > 0. Indeed, ﬁxing A > 0, we have for every ε ∈ (0, 1):

1 ε

ρμ (h) ≤ A −1 ≤ , for every h ∈ (0, h A ).

Ψ ε Ψ (1/h) Ψ AΨ−1 (1/h) ε

**2) Let us point out that the fact that μ is a Ψ-Carleson measure (resp. a
**

vanishing Ψ-Carleson measure) means that μ satisﬁes condition (R) in Theorem

4.10 (resp. condition (R0 ) in Theorem 4.11), with Ψ1 = Ψ2 = Ψ.

3. GENERAL MEASURES 49

We have the following characterizations:

**Proposition 4.16. 1) μ is a Ψ-Carleson measure on D if and only if there
**

exists some constant C ≥ 1 such that :

C

uξ,1−h LΨ (μ) ≤ , for every ξ ∈ T and every h ∈ (0, 1).

Ψ−1 (1/h)

2) μ is a vanishing Ψ-Carleson measure on D if and only if

**lim sup Ψ−1 (1/h)uξ,1−h LΨ (μ) = 0.
**

h→0 ξ∈T

**Proof. The suﬃciency (both for 1) and 2)) follows easily from (4.12) in the pre-
**

ceding remark.

The converse is an obvious consequence of the following lemma.

Lemma 4.17. Suppose that there exist A > 0 and h0 ∈ (0, 1) such that:

1

ρμ (h) ≤ , for every h ∈ (0, h0 ).

Ψ AΨ−1 (1/h)

Then there exists h1 ∈ (0, 1) such that:

24

uξ,1−h LΨ (μ) ≤ ,

AΨ−1 (1/h)

for every ξ ∈ T and every h ∈ (0, h1 ).

**Proof of the lemma. This is inspired from [12], Chapter VI, Lemma 3.3, page
**

239. We may assume that h ≤ h0 /4 ≤ 1/4.

First, writing a = (1 − h)ξ (where ξ ∈ T), we observe that, when |z − ξ| ≥ bh

for a b > 0, we have:

¯

|1 − āz|2 = 1 + |a|2 |z|2 − 2|a|Re (ξz)

= |a||ξ − z|2 + (1 − |a|) + |a|2 |z|2 − |a||z|2

≥ |a|b2 h2 + (1 − |a|)2 ≥ (|a|b2 + 1)h2 .

1

So, we have |uξ,1−h (z)| ≤ ≤ min(1, 2/b2 ), when |z − ξ| ≥ bh.

|a|b2 + 1

Now, deﬁne, for every n ∈ N and ξ ∈ T:

Sn = S(ξ, 2n+1 h) = {z ∈ D ; |z − ξ| < 2n+1 h} ⊂ W (ξ, 2.2n+1 h).

**Our observation implies that |uξ,1−h (z)| ≤ min(1, 2/4n ), for every z ∈ D \ Sn−1 .
**

For z ∈ S0 , one has simply |uξ,1−h (z)| ≤ 1.

There exists an integer N such that 2N +2 h ≤ h0 < 2N +3 h.

50 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

Let us compute:

N

Ψ M |uξ,1−h | dμ = Ψ M |uξ,1−h | dμ + Ψ M |uξ,1−h | dμ

D S0 n=1 Sn \Sn−1

+ Ψ M |uξ,1−h | dμ

D\SN

N 2M
2M

≤ Ψ(M )μ(S0 ) + Ψ n μ(Sn ) + Ψ N μ(D)

n=1

4 4

N

1 4M
2M

≤ Ψ n μ(Sn ) + Ψ N μ(D).

n=0

2n+1 2 4

**But for n ≤ N , we have 2.2n+1 h ≤ 2N +2 h ≤ h0 , so the hypothesis gives:
**

1

μ(Sn ) ≤ ·

Ψ AΨ−1 (1/2n+2 h)

Take now:

A −1 1

(4.14) M= Ψ ·

24 h

1

We have, using that Ψ−1 2n+2 h ≥ 2n+2 Ψ

1 −1 1

h ,

N

1 A −1 1

Ψ M |uξ,1−h | dμ ≤ + μ(D)Ψ Ψ

D n=0

2n+1 12.4N h

2

1 16A.h −1 1

≤ + μ(D)Ψ Ψ ,

2 3h20 h

2

1 8h

because N ≤ .

4 h0

We can choose h1 small enough to have:

16A.h2 −1 1
1

μ(D)Ψ 2 Ψ ≤

3h0 h 2

for every h ∈ (0, h1 ), since lim h2 Ψ−1 h1 = 0.

h→0

We get for such h

Ψ M |uξ,1−h | dμ ≤ 1,

D

so that

1 24 1

uξ,1−h LΨ (μ) ≤ = −1

,

M A Ψ (1/h)

as it was announced.

**Examples and counterexamples.
**

We are going to give some examples showing that we do not have the reverse

implications in general in Theorem 4.10 and Theorem 4.11.

1. Condition (R) is not suﬃcient in general to have a continuous embedding.

Let Ψ(x) = ex − 1 (note that this Orlicz function even fulﬁlls the Δ1 condition !).

Note that Ψ(AΨ−1 (1/h)) ∼ h−A , when h → 0.

3. GENERAL MEASURES 51

**a. Let ν be a probability measure on T, supported by a compact set L of
**

Lebesgue measure zero, such that ν(I) ≤ |I|1/2 , for each I. We can associate to ν

the measure on D deﬁned by ν̃(E) = ν(E ∩ T). Then the identity map from H Ψ to

LΨ (ν̃) is not even deﬁned. Nevertheless the condition (R) is clearly fulﬁlled with

A = 1/2.

b. Now, we exhibit a similar example (less artiﬁcial) on the open disk. Let ν be

as previously. By a standard argument: for every integer n, there exists a function

gn in the unit ball of the disk algebra such that |gn | = 1 on L and gn Ψ ≤ 4−n . As

L is compact, there exists some rn ∈ (1/2, 1) such that |gn (rn z)| ≥ 1/2 for every

z ∈ L. Now, deﬁne the measure μ by:

∞

1

μ(E) = ν (E),

n n

n=1

2

where:

νn (E) = ν {z ∈ T| rn z ∈ E} .

If W is a Carleson window of “size” h then, for each n ≥ 1, we have:

ν {z ∈ T| rn z ∈ W } ≤ ν(W ∩ T) ≤ (2h)1/2 .

Hence, μ(W ) ≤ (2h)1/2 and the condition (R) is fulﬁlled.

Nevertheless, the identity from H Ψ to L1 (μ) is not continuous: gn Ψ ≤ 4−n ;

but

1 1 1

gn 1 ≥ n |gn | dνn ≥ n |gn (rn w)| dν(w) ≥ n+1 ·

2 rn T 2 L 2

2. Condition (K) is not necessary in general to have a continuous embedding.

When Ψ satisﬁes Δ2 , the identity from H Ψ to LΨ (μ) is continuous if and only if μ

is a Carleson measure. So the conditions (R), (K) and the continuity are equivalent

in this case. Actually, when Ψ does not satisfy Δ2 , we construct below a measure

μ on D such that the identity from H Ψ to LΨ (μ) is continuous and order bounded,

but μ is not a Carleson measure (a fortiori does not verify (K)). Note that the

measure μ is then Ψ-Carleson but not Carleson. Here is the example:

We have assumed that Ψ does not satisfy Δ2 ; so there exists an increasing

Ψ(an ) Ψ(2an )

sequence (an )n≥1 such that is increasing and ≥ n2n . Now, deﬁne

n Ψ(an )

the discrete measure

∞

n n+1

μ= − δxn ,

n=1

Ψ(2an ) Ψ(2an+1 )

where:

1

xn = 1 − ·

Ψ(2an )

N ,

As μ [xN , 1] = the measure μ is not Carleson: it should be bounded

Ψ(2aN )

c

by c(1 − xN ) = , where c is some constant.

Ψ(2aN )

forevery f in the unit ball of H and every x ∈ (0, 1), we have

Ψ

We know that

1

|f (x)| ≤ 4Ψ−1 : see Lemma 3.11. So we only have to see that g ∈ LΨ (μ),

1−x

52 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

1

where g(x) = Ψ−1 . Indeed, we have:

1−x

∞

|g| n n + 1
|g(xn )|

Ψ dμ = − Ψ

D 2 n=1

Ψ(2an ) Ψ(2an+1 ) 2

∞

n 1 1

≤ Ψ Ψ−1

n=1

Ψ(2an ) 2 1 − xn

∞

∞

Ψ(an ) 1

≤ n ≤

n=1

Ψ(2an ) n=1 2n

≤1

so gΨ ≤ 2.

3. Condition (K0 ) is not necessary in general to have a compact embedding.

We can ﬁnd, for every Orlicz function Ψ not satisfying ∇0 , a measure μ such

that the identity from H Ψ to LΨ (μ) is compact but (K0 ) is not satisﬁed. Indeed:

since Ψ ∈/ ∇0 , we can select two increasing sequences (xn )n≥1 and (yn )n≥1 , with

1 ≤ xn ≤ yn ≤ xn+1 and Ψ(xn ) > 1, and such that lim xn = +∞ and:

Ψ(2xn ) Ψ(2n yn )

≥ ·

Ψ(xn ) Ψ(yn )

1

Deﬁne rn = 1 − and the discrete measure:

Ψ(yn )

∞

1

μ= δr .

n=1

Ψ(2n yn ) n

**The series converge since Ψ(2n yn ) ≥ 2n .
**

Thanks to Lemma 3.11, we have, for every f in the unit ball of H Ψ and every

n ≥ 1:

1

|f (rn )| ≤ 4Ψ−1 = 4yn .

1 − rn

Given r > r1 , there exists an integer N ≥ 1 such that rN < r ≤ rN +1 . Then, for

every f in the unit ball of H Ψ , we have f LΨ (D\rD,μ) ≤ 2−N +2 , since:

∞

|f | 1 |f (rn )|

Ψ −N +2 dμ = Ψ

D\rD 2 Ψ(2n yn ) 2−N +2

n=N +1

∞

Ψ(2N yn )

≤

Ψ(2n yn )

n=N +1

∞

1

≤ = 1.

2n−N

n=N +1

**This implies that lim sup f LΨ (D\rD,μ) = 0. By Proposition 4.9, the identity
**

r→1 f ≤1

Ψ

from H Ψ to LΨ (μ) is compact.

3. GENERAL MEASURES 53

1 1 ,

On the other hand, writing hn = and tn = we have:

Ψ(xn ) Ψ(yn )

∞

μ([1 − tn , 1]) 1 Ψ(yn )

Kμ (hn ) ≥ = Ψ(yn ) my )

≥ ny )

tn m=n

Ψ(2 m Ψ(2 n

Ψ(xn ) 1/hn

≥ = −1 ,

Ψ(2xn ) Ψ 2Ψ (1/hn )

and this shows that (K0 ) is not satisﬁed.

This shows that condition Ψ ∈ ∇0 is necessary and suﬃcient in order to have:

the identity from H Ψ to LΨ (μ) is compact if and only if μ satisﬁes (K0 ).

4. Condition (R0 ) is not suﬃcient in general to have a compact embedding.

We can ﬁnd an Orlicz function Ψ and a vanishing Ψ-Carleson measure (i.e. (R0 )

is satisﬁed) μ such that the identity from H Ψ to LΨ (μ) is not compact.

We shall use the Orlicz function introduced in [25]. The key properties of this

function Ψ are:

i) For every x > 0, Ψ(x) ≥ x3 /3.

ii) For every integer k ≥ 1, Ψ(k!) ≤ (k!)3 .

iii) For every integer k ≥ 1, Ψ(3(k!)) > k.(k!)3 .

Once again, the job is done by a discrete measure.

(k + 1)! 1 1

Deﬁne xk = k!; yk = 1/3

; rk = 1 − and ρk = 1 − · Of course,

k Ψ(yk ) Ψ(xk )

x2 < y2 < x3 < · · · .

Let ν be the discrete measure deﬁned by:

∞

ν= νk ,

k=2

where:

1

νk = δr k a .

Ψ (k + 1)! k2

a =1

2 2

k 3.k

Observe that νk ≤ ≤ so that the series converges. Note

Ψ (k + 1)! (k + 1)!3

that ν is supported in the union of the circles of radii rk and not in a subset of the

segment [0, 1] as in the preceding counterexamples.

In order to show that (R0 ) is satisﬁed, it is clearly suﬃcient to prove that, when

1 1

≤h< (with k ≥ 3), we have:

Ψ(yk ) Ψ(yk−1 )

1

ρν (h) ≤ k1/3
·

−1

Ψ Ψ (1/h)

2

1 1 , we have Ψ−1 (1/h) ≤ yk so

Supposing then ≤h<

Ψ(yk ) Ψ(yk−1 )

k1/3
1

Ψ Ψ−1 (1/h) ≤ Ψ (k + 1)! .

2 2

2

Therefore, it is suﬃcient to establish that ρν (h) ≤ ·

Ψ (k + 1)!

54 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

**A Carleson window W (ξ, h) (where ξ ∈ T) can contain at most one k2 -root of
**

2 6 6(k − 1) 2π

the unity, since 2h < ≤ 3 ≤ ≤ 2 · This implies that

Ψ(yk−1 ) yk−1 (k!)3 k

1

νk W (ξ, h) ≤ ·

Ψ (k + 1)!

**Besides, when j < k, the window W(ξ, h) cannot meet any circle of radius rj
**

(centered at the origin), so νj W (ξ, h) = 0. We obtain:

∞

1 j2

ν W (ξ, h) = νj W (ξ, h) ≤ +

j=k

Ψ (k + 1)! j>k

Ψ (j + 1)!

∞

1 3j 2

≤ +

Ψ (k + 1)! (j + 1)!3

j=k+1

∞ s

1 3 1

≤ +

Ψ (k + 1)! (k + 1)!3 s=1 k + 1

2

≤ ·

Ψ (k + 1)!

**This proves that (R0 ) is satisﬁed.
**

2

2 1 − ρk

Let us introduce the function fk (z) = xk u1,ρk z k = xk .

1 − ρk z k 2

It lies in the unit ball of H Ψ by Corollary 3.10:

x

fk Ψ = xk u1,ρk Ψ ≤ k 1 = 1.

Ψ−1 1−ρk

2

An easy computation gives rkk ≥ ρk , for every k ≥ 2. So, for every a ∈ T with

2

ak = 1, we have:

2

1 − ρk 1

fk (ark ) ≥ xk ≥ xk .

1 − ρ2k 4

So

k2

Ψ(12|fk |) dν ≥ Ψ(12|fk |) dνk ≥ Ψ(3xk )

D\rk−1 D D\rk−1 D Ψ((k + 1)!)

k 2

> k.(k!)3 ≥ 1.

Ψ((k + 1)!)

1

Therefore, we conclude that sup f LΨ (D\rk D,μ) ≥ , though rk → 1. By

f Ψ ≤1 12

Proposition 4.9, the identity from H Ψ to LΨ (μ) is not compact.

4. CHARACTERIZATION OF THE COMPACTNESS OF COMPOSITION OPERATORS 55

**4. Characterization of the compactness of composition operators
**

For composition operators, compactness can be characterized in terms of Ψ-

Carleson measures, as stated in the following result.

Theorem 4.18. For every analytic self-map φ : D → D and every Orlicz func-

tion Ψ, the composition operator Cφ : H Ψ → H Ψ is compact if and only if one

has:

1

(R0 ) ρμφ (h) = o as h → 0, for every A > 0.

Ψ AΨ−1 (1/h)

In other words, if and only if μφ is a vanishing Ψ-Carleson measure.

In order to get this result, we shall show that in Theorem 4.11, conditions (R0 )

and (K0 ) are equivalent for the pull-back measure μφ induced by φ. This is the

object of the following theorem.

Theorem 4.19. There exists a constant k1 > 0 such that, for every analytic

self-map φ : D → D, one has:

(4.15) μφ S(ξ, εh) ≤ k1 ε μφ S(ξ, h) ,

for every h ∈ (0, 1 − |φ(0)|), and every ε ∈ (0, 1).

Note that we prefer here to work with the sets:

S(ξ, h) = {z ∈ D ; |z − ξ| < h} , ξ ∈ T, 0 < h < 1,

instead of the Carleson windows W (ξ, h). Recall also that the pull-back measure

μφ is deﬁned by (4.3).

We are going to postpone the proof of Theorem 4.19, and shall give before some

consequences.

4.1. Some consequences. An immediate consequence of Proposition 4.16

and Theorem 4.18 is the following

Theorem 4.20. Let φ : D → D be analytic and Ψ be an Orlicz function.

The operator Cφ on H Ψ is compact if and only if

1

(W) sup Cφ (uξ,1−h )Ψ = o , as h → 0.

ξ∈T Ψ−1 1/h

We deduce:

Theorem 4.21. Let φ : D → D be analytic.

1) Assume that the Orlicz function Ψ satisﬁes condition Δ0 . Then, the operator

Cφ on H Ψ is weakly compact if and only if it is compact.

2) Assume that the Orlicz function Ψ satisﬁes condition ∇2 . Then, the operator

Cφ on H Ψ is a Dunford-Pettis operator if and only if it is compact.

Recall that (see Theorem 3.24), under condition Δ2 for Ψ, the weak compact-

ness of the composition operator Cφ is equivalent to its compactness, and even

to Cφ being order bounded into M Ψ (T). However, we shall see below, in Theo-

rem 4.22, that there exist Orlicz functions Ψ ∈ Δ0 (and even Ψ ∈ Δ1 ) for which

Cφ is compact, but not order bounded into M Ψ (T).

Proof. In both cases, the result follows from Theorem 4.20, since condition (W)

is satisﬁed. Indeed, if Cφ : H Ψ → H Ψ is weakly compact and Ψ ∈ Δ0 , we use

56 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

**Theorem 3.20. If Cφ : H Ψ → H Ψ is a Dunford-Pettis operator, this is due to
**

Proposition 3.28.

Now, we have:

Theorem 4.22. There exist an Orlicz function Ψ satisfying Δ1 , and an analytic

self-map φ : D → D such that the composition operator Cφ : H Ψ → H Ψ is not order

bounded into M Ψ (T), though it is compact.

Remark. It follows that our assumption that Ψ ∈ Δ2 in Theorem 3.24 is not only

a technical one, though it might perhaps be weakened.

Proof. Let: √

exp (log x)2 if x ≥ e,

Ψ(x) = √

e−1/4 x if 0 ≤ x ≤ e.

It is plain that Ψ ∈ Δ1 ∩ ∇0 .

Moreover, for every A > 0, one has, for h small enough:

1/h

= exp − (log A)2 − 2(log A) log(1/h) .

Ψ AΨ−1 (1/h)

Consider now φ = φ2 , the analytic self-map of D constructed in Theorem 4.1.

Then Cφ : H Ψ → H Ψ is not order bounded into M Ψ (T), by Theorem 3.15, since,

otherwise, Cφ1 : H Ψ → H Ψ would also be order bounded into M Ψ (T), which is

easily seen to be not the case (we may also argue as follows: Cφ1 : H Ψ → H Ψ

would be compact, and hence, by Theorem 4.3, Cφ1 would be compact from H 2

into H 2 , which is false).

On the other hand, we have proved that ρμφ (h) = O h3/2 . So the conclusion

follows from Theorem 4.11, 3) and the fact that for every c > 0:

1

−(log A)2 − 2(log A) log(1/h) ≥ −c log(1/h)

2

when h is small enough.

Remark. We saw that a compact composition operator on H Ψ is always compact

on H 2 , and even nuclear when Ψ ∈ Δ2 (see Theorem 3.24). A natural question is:

when Ψ ∈ Δ0 , is a compact composition operator on H Ψ automatically a Hilbert-

Schmidt operator on H 2 ? The answer is “no” in general: the function φ2 in the

previous proof gives the counterexample. Indeed, Cφ2 is compact on H Ψ . If it

was a Hilbert-Schmidt operator on H 2 , Cφ1 would be Hilbert-Schmidt as well since

|φ∗1 | = |φ∗2 | but Cφ1 is not compact on H 2 .

**4.2. Preliminary results. We shall use the radial maximal function N , de-
**

ﬁned for every harmonic function u on D by:

(4.16) (N u)(ξ) = sup |u(rξ)| , ξ ∈ T.

0≤r<1

**Recall that for every positive harmonic function u : D → C whose boundary
**

values u∗ are in L1 (T) (i.e. u ∈ h1 ), one has, for every ξ ∈ T:

(4.17) N u(ξ) ≤ M u∗ (ξ) ≤ πN u(ξ) ,

where M u∗ is the Hardy-Littlewood maximal function of u∗ (see [35], Theorem

11.20 and Exercise 19, Chapter 11).

4. CHARACTERIZATION OF THE COMPACTNESS OF COMPOSITION OPERATORS 57

**We shall denote by Π the right half-plane:
**

Π = {z ∈ C ; Re z > 0}

and by C the cone:

C = {z ∈ C ; −π/6 < Arg z < π/6}.

The next result follows from Kolgomorov’s Theorem, saying that the Hilbert

transform is a weak (1, 1) operator (in applying this theorem to the positive har-

monic function 2 Re g = g + ḡ, noting that Re g1 = Re g(0) = g(0)).

Lemma 4.23. There exists a constant c > 0 such that, if g : D → Π is an

analytic function with g(0) > 0, then:

g(0)

m({|g ∗ | > λ}) ≤ c , for all λ > 0.

λ

3

Applying Lemma 4.23 to g(z) = f (z) + f (0) , and taking into account that

|w1 + w2 | ≥ |w1 |, if w1 , w2 ∈ C , we get:

Lemma 4.24. Let f : D → C be an analytic function with values in the cone C ,

and write f = u + iv, with u, v real-valued. Then:

u(0)
3

m({|f ∗ | > λ}) ≤ 8c , for all λ > 0.

λ

The next proposition is one of the keys. We postpone its proof.

Proposition 4.25. There exists a constant k2 > 0 such that, for every analytic

function f = u + iv : D → C with values in the cone C , one has:

α
3

(4.18) m({|f ∗ | > λ} ∩ I) ≤ k2 m(I), for all λ > 0,

λ

where I is the arc I = {eit ; a < t < b}, with a, b ∈ R and α > 0 satisfying

b − a < π/2, α ≥ N u(eia ), and α ≥ N u(eib ).

As a corollary we obtain:

Proposition 4.26. Let f : D → C be an analytic function, and write f = u+iv,

as in Proposition 4.25. If α > 0 satisﬁes m({N u > α}) < 1/4, then:

α
3 2α

(4.19) m({|f ∗ | > λ}) ≤ k2 m({N u > α}), for all λ ≥ √ ·

λ 3

Proof. The set {N u > α} is open, and one can decompose it into a disjoint union

of open arcs {Ij }j . Each arc has measure m(Ij ) ≤ m({N u > α}) < 1/4, and so is

an arc of length less than π/2. We can then apply Proposition 4.25 and we obtain:

α
3

m({|f ∗ | > λ} ∩ Ij ) ≤ k2 m(Ij ), for every j.

λ

Summing up all these inequalities we get:

α
3

m {|f ∗ | > λ} ∩ {N u > α} ≤ k2 m({N u > α}) .

λ

The proposition follows since |f ∗ | ≤ √23 u∗ ≤ √23 N u, and then {|f ∗ | > λ} is

contained in {N u > α}, for λ ≥ √ 2α

3

·

We shall need one more result.

58 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

**Proposition 4.27. There exists a constant k3 > 0 such that for every analytic
**

function f : D → C with values in the cone C , one has, writing f = u + iv:

m({M u∗ > α} ≤ k3 m({u∗ > α/2}, for every α > 0.

In order to prove it, we shall ﬁrst prove the following lemma.

Lemma 4.28. There exists a constant k4 > 0 such that for every analytic func-

tion f : D → C with values in the cone C , one has, writing f = u + iv:

2

M (u∗ )2 (ξ) ≤ k4 M u∗ (ξ) , for all ξ ∈ T.

Proof. Observe that u2 − v 2 is a positive harmonic function since it is the real

part of f 2 . But f 2 belongs to H 1 (see [11], Theorem 3.2), so u2 − v 2 ∈ h1 and

we can use inequalities (4.17). We also have, since −π/3 < Arg (f 2 ) < π/3, that

u2 ≥ 3v 2 , and so u2 − v 2 ≥ 2u2 /3 and |f | ≤ √23 u. We get:

3 3π

M (u∗ )2 ≤ M (u∗ )2 − (v ∗ )2 ≤ N (u2 − v 2 )

2 2

2

3π 3π 3π 2

≤ 2

N (f ) = (N f ) ≤

2

√ Nu

2 2 2 3

≤ 2π (M u∗ )2 .

Proof of Proposition 4.27. Write A = {M u∗ > α} and B = {u∗ > α/2}. For

every ξ ∈ A, there exists an open arc Iξ centered at ξ such that:

1

p= u∗ dm > M u∗ (ξ)/2 , and p > α .

m(Iξ ) Iξ

We have, using Lemma 4.28:

1 2

(u∗ )2 dm ≤ M (u∗ )2 (ξ) ≤ k4 M u∗ (ξ) ≤ 4k4 p2 .

m(Iξ ) Iξ

Let L be the set L = {u∗ > p/2} ∩ Iξ . Let P be the probability measure: P(ω) =

1

m(ω ∩ Iξ ). We apply the Paley-Zygmund inequality (see [28], Proposition

m(Iξ )

III.3. page 22; or [21], page 8) to the random variable u∗ , with the probability P

and we obtain

1 (Eu∗ )2 1

P(L) ≥ · ≥

4 E(u∗ )2 16k4

since Eu∗ = p and E(u∗ )2 ≤ 4k4 p2 .

Therefore, 16k4 m(L) ≥ m(Iξ ), and, since L ⊆ B ∩ Iξ , we have, for every

ξ ∈ A, an arc Iξ containing ξ such that 16k4 m(B ∩ Iξ ) ≥ m(Iξ ). Applying the

Hardy-Littlewood covering lemma, we then obtain:

n

n

m(A) ≤ 3 m(Iξj ) ≤ 3 × 16 × k4 × m(Iξj ∩ B) ≤ k3 m(B).

j=1 j=1

**Proof of Proposition 4.25. Composing f with a suitable rotation, we can sup-
**

pose that a = −δ and b = δ, for 0 < δ < π/4. Let us call I − and I + the arcs

I − = {eit ; −δ < t < 0} , I + = {eit ; 0 < t < δ} .

4. CHARACTERIZATION OF THE COMPACTNESS OF COMPOSITION OPERATORS 59

**We shall prove that:
**

α
3

(4.20) m({|f ∗ | > λ} ∩ I + ) ≤ k2 m(I + ) ,

λ

using just the fact that N u(eiδ ) ≤ α.

In the same way one can prove:

α
3

(4.21) m({|f ∗ | > λ} ∩ I − ) ≤ k2 m(I − ) ,

λ

using just that N u(e−iδ ) ≤ α.

Then, summing up (4.20) and (4.21), Proposition 4.25 will follow.

Let Q be the right half-disc

Q = {z ∈ D ; Re z > 0},

and denote by ψ the (unique) homeomorphism from D to Q, which is a conformal

mapping from D onto Q and sends 1 to 1, i to i, and −i to −i.

We can construct ψ as the composition of a Moebius transformation T , with

the square root function and then with T −1 . Namely, let:

z+i

T z = −i ;

z−i

T maps D onto the upper-half plane, sending −i into 0, −1 into −1, 1 into 1, and

also 0 into i, and i into ∞. The square root function maps the upper-half plane

into the ﬁrst quadrant and T −1 :

z−i

T −1 z =

1 − iz

maps this quadrant onto the half-disk Q. √

It is not diﬃcult to see that ψ(−1) = 0, ψ(0) = 2 − 1, and that there

exist ρ ∈ (0,√π/2) such that ψ(eiρ ) = e√πi/4 and ψ(e−iρ ) = e−πi/4 (we must have

iρ

e = 1/3 + ( 8)i/3; hence ρ = arctan( 8)).

Let J be the arc:

J = {eit ; −ρ < t < ρ}.

The map ψ is regular on J, and so there exist two constants γ1 and γ2 > 0 such

that for every Borel subset E of J, one has:

γ1 m(E) ≤ m ψ(E) ≤ γ2 m(E).

If now β ∈ (0, 1), we put:

ψβ (z) = (ψ(z))β .

Then, it is easy to see that for every Borel subset E of J, one has:

m ψβ (E) = β m ψ(E) ,

and so

γ1 β m(E) ≤ m ψβ (E) ≤ γ2 β m(E).

In order to prove (4.20), consider the function F : D → C deﬁned by:

F (z) = f eiδ ψβ (z) , where β = 4δ/π.

Then: √

Re F (0) = u ( 2 − 1)β eiδ ≤ N u(eiδ ) ≤ α.

Let us call χ the map:

χ(z) = eiδ ψβ (z).

60 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

**It is clear that I + is contained in χ(J). If A = {|f ∗ | > λ} ∩ I + , then E = χ−1 (A)
**

is a Borel subset of J, and for every ξ ∈ E, one has |F ∗ (ξ)| > λ. Then:

m(A) = m χ(E) = m ψβ (E) ≤ γ2 β m(E)

δ

≤ 8γ2 m({|F ∗ | > λ})

2π

= 8γ2 m(I + ) m({|F ∗ | > λ})

and using Lemma 4.24 for F ,

3

Re F (0)

≤ 8γ2 m(I + ) × 8c3

λ

α
3 α
3

≤ 64γ2 c3 m(I + ) = k2 m(I + ) .

λ λ

The proof of (4.20) is ﬁnished, and Proposition 4.25 follows.

4.3. Proof of Theorem 4.19. Using the fact h → μφ S(ξ, h) is nondecreas-

ing, it is enough to prove that there exist h0 > 0, and ε0 > 0, such that (4.15) is

true for 0 < h < h0 (1 − |φ(0)|), and 0 < ε < ε0 , because changing the constant k1 ,

if necessary, the theorem will follow.

We can also suppose that ξ = 1.

The real part of 1/ 1 − φ(z) is positive, in fact greater than 1/2, for every

z ∈ D. Take 0 < h < h0 (1 − |φ(0)|), and consider the analytic function f deﬁned

by:

1/3

h

f (z) = ,

1 − φ(z)

where the cubic root is taken in order that, for every z ∈ D, f (z) belongs to the

cone

C = {z ∈ C : −π/6 < Arg (z) < π/6 }.

√

Clearly μφ S(1, h) = m({|f ∗ | > 1}) and μφ S(1, εh) = m({|f ∗ | > 1/ 3 ε}). We

also have: 1/3

h

|f (0)| ≤ < 3 h0 .

1 − |φ(0)|

We shall write f = u + iv where u and v are real-valued harmonic functions.

Observe that:

1

|v(z)| < √ u(z), for every z ∈ D.

3

It is known that f ∈ H , for every p < 3 (see [11], Theorem 3.2), and so u and v

p

**are the Poisson integrals of u∗ and v ∗ (in particular √ u, v ∈ h1 ).
**

We are looking for a control of m({|f | > 1/ ε}) by ε times m({|f ∗ | > 1}).

∗ 3

**Proposition 4.26 provides this control if we replace m({|f ∗ | > 1}) by m({N u > 2}):
**

√

m({|f ∗ | > 1/ 3 ε}) ≤ 8k2 εm({N u > 2}),

when m({N u > 2}) < 1/4.

As f is valued in C , |f ∗ | is controlled by u∗ . Then what we need in fact is a

control of the measure of level sets of N u by the measure of level sets of u∗ . This

will be done by using Proposition 4.27.

Indeed, by Proposition 4.27, we have:

m({N u > 2}) ≤ m({M u∗ > 2}) ≤ k4 m({u∗ > 1}).

4. CHARACTERIZATION OF THE COMPACTNESS OF COMPOSITION OPERATORS 61

**We know that ||u∗ ||1 = u(0) ≤ |f (0)| ≤ h0 . Then, choosing h0 small enough,
**

1/3

1/3

we can have k4 h0 < 1/4, and so m({N u > 2}) < 1/4; therefore, we can use

Proposition

√ 4.26. Moreover we also have {u∗ > 1} ⊆ {|f ∗ | > 1}. Taking ε0 <

3 3/64, we have, for 0 < ε < ε0 , by Proposition 4.26:

√

m({|f ∗ | > 1/ 3 ε}) ≤ 8k2 ε m({N u > 2}) ≤ 8k4 k2 ε m({u∗ > 1})

≤ k1 ε m({|f ∗ | > 1}) ,

taking k1 = 8k4 k2 .

CHAPTER 5

Bergman spaces

**1. Bergman-Orlicz spaces
**

Definition 5.1. Let dA (z) = dxπdy (z = x + iy) be the normalized Lebesgue

measure on D. The Bergman-Orlicz space B Ψ denotes the space of analytic func-

tions f : D → C which are in the Orlicz space LΨ (D, dA ). The Bergman-Morse-

Transue space is the subspace BM Ψ = B Ψ ∩ M Ψ (D, dA ).

B Ψ , equipped with the induced norm of LΨ (D, dA ), is a Banach space, as an

obvious consequence of the following lemma, analogous to Lemma 3.11.

Lemma 5.2. For every a ∈ D, the norm of the evaluation functional δa , which

maps f ∈ B Ψ to f (a), is:

1

δa ≈ Ψ−1 ·

(1 − |a|)2

Proof. For every analytic function g : D → C, the mean-value property gives:

g(0) = g(z) dA (z).

D

**Hence if φa : D → D denotes the analytic automorphism
**

z−a

φa (z) = ,

1 − āz

whose inverse is φ−1

a = φ−a , one has, for every f ∈ B , using the change of variable

Ψ

formula:

f (a) = f ◦ φ−a (0) = f ◦ φ−a (z) dA (z) = f (w) |φa (w)|2 dA (w)

D D

= f (w)Ha (w) dA (w) ,

D

where:

(1 − |a|2 )2

Ha (w) = |φa (w)|2 = ·

|1 − āw|4

The kernel Ha plays for B Ψ the role that the Poisson kernel Pa plays for H Ψ : the

analytic reproducing kernel Ka for B 2 being Ka (z) = (1−āz)

1

2 ·

**We therefore have (using [33], Proposition 4, page 61, or [4], Theorem 8.14):
**

|f (a)| ≤ 2f Ψ Ha Φ ,

which proves the continuity of δa . To estimate its norm, we are going to majorize

Ha Φ , with the help of Lemma 3.9. Let us notice that, on the one hand, Ha 1 = 1

63

64 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

**(take f = I in the above identity); and on the other hand:
**

(1 − |a|2 )2 (1 + |a|)2

Ha ∞ = = ·

(1 − |a|) 4 (1 − |a|)2

We get, setting b = Ha ∞ , and using Lemma 3.9 for Φ :

b

Ha Φ ≤ ·

Φ−1 (b)

But b ≤ Φ−1 (b)Ψ−1 (b) (see [33], Proposition 1 (ii), page 14). Hence Ha Φ ≤

Ψ−1 (b). Now:

4

b≤ ·

(1 − |a|)2

We have Ψ−1 (4t) ≤ 4Ψ−1 (t) for all t > 0. It follows that

1

Ha Φ ≤ CΨ−1 ·

(1 − |a|)2

Since δa ≤ 2Ha Φ , we get the upper bound in Lemma 5.2.

For the lower bound, we simply observe that Ha = |Ga |, where Ga (z) =

(1 − |a|2 )2

, and by Lemma 3.9:

(1 − az)4

1

|Ga (a)| |Ha (a)| (1 − |a|2 )2 Ψ−1 (b)

δa ≥ = ≥ −1

=

Ga Ψ Ha Ψ b/Ψ (b) b(1 − |a|2 )2

Ψ−1 (b) 1 −1 1 −1 1

= ≥ Ψ (b) ≥ Ψ ,

(1 + |a|)4 16 16 (1 − |a|)2

since b ≥ 1/(1 − |a|)2 .

Proposition 5.3. We have the following properties

i) BM Ψ is the closure of H ∞ (D) in LΨ (D, A ) and actually the algebraic

polynomials are dense in BM Ψ .

ii) On the unit ball of B Ψ , the weak-star topology σ LΨ (D, A ), M Φ (D, A )

coincides with the topology of convergence on compact subsets of D.

iii) B Ψ is closed in LΨ (D, A ) for the weak-star topology.

iv) If Ψ ∈ ∇2 , B Ψ is (isometric to) the bidual of BM Ψ .

Proof. For the ﬁrst point, let us ﬁx f ∈ BM Ψ . Setting fr (z) = f (rz) for z ∈ D

and 0 ≤ r < 1, it suﬃces to show that fr − f Ψ −→ 0, since, being analytic in the

r→1

disk rD ⊂ D, fr can be uniformly approximated on D by its Taylor series. But the

norm of M Ψ is absolutely continuous (see [33], Theorem 14, page 84) and therefore,

for every ε > 0, there is some R > 0, with 1/3 ≤ R < 1, such that f ID\RD Ψ ≤ ε;

hence:

|f |
1 |f |
1

Ψ dA ≤ Ψ dA ≤ ·

D\RD 4ε D\RD 4 ε 4

When r ≥ 2R

R+1 ≥ 1/2, we therefore have:

|f |

r

Ψ dA ≤ 1,

D\ 2 D

1+R 4ε

1. BERGMAN-ORLICZ SPACES 65

and, by convexity of Ψ:

|fr − f |
|f − f |

r

Ψ dA ≤ Ψ dA

D 8ε 2 D

1+R 8ε

1 |fr |
|f |

+ Ψ +Ψ dA

D\ 1+R

2 D

2 4ε 4ε

|f − f |

r

≤ Ψ dA

2 D

1+R 8ε

|f |
|f |

1 r 1

+ Ψ dA + Ψ dA

2 D\ 1+R2 D

4ε 2 D\RD 4ε

≤ 1,

for r close enough to 1 since fr − f tends to 0 uniformly on 1+R2 D. Hence, for some

r0 < 1, one has fr − f Ψ ≤ 8ε for r0 ≤ r < 1. This was the claim.

ii) It suﬃces to use a sequential argument since the topologies are metrizable on

balls (the space M Φ (D, A ) is separable). Assume that f ∈ B Ψ (with f Ψ ≤ 1) is

the weak-star limit of a sequence of analytic functions fn ∈ B Ψ (with fn Ψ ≤ 1).

Testing this with the function hk (z) = (k + 1)z k , we obtain that the kth Taylor

coeﬃcient ak (n) of fn converges to the kth Taylor coeﬃcient ak of f :

ak (n) = fn hk dA −→ f hk dA = ak , for every k ≥ 0.

D D

Fix a compact K ⊂ D, there exists an r ∈ (0, 1) such that K ⊂ rD. We have:

sup |fn (z) − f (z)| ≤ |ak (n) − ak |r k −→ 0

z∈K

k≥0

**by the dominated convergence Theorem (observe that |ak − ak (n)| ≤ 2(k + 1) for
**

every k ≥ 0 and every n ≥ 0).

For the converse, suppose now that fn ∈ B Ψ (with fn Ψ ≤ 1) converges

uniformly on every compact subsets of D to f ∈ B Ψ (with f Ψ ≤ 1). Fixing

g ∈ M Φ (D, A ) and ε > 0, there exists an r ∈ (0, 1) such that gID\rD Φ ≤ ε. We

have:

(fn − f )gdA ≤ (fn − f )gdA + 2ε ≤ sup |fn (z) − f (z)|.g1 + 2ε.

D rD z∈rD

**By hypothesis sup |fn (z) − f (z)| −→ 0. The conclusion follows.
**

z∈rD

iii) By the classical Theorem of Banach-Dieudonné, it is suﬃcient to prove that

the balls are weak-star closed (equivalently weak-star compact) and by separability

of M Φ (D, A ), the weak-star topology is metrizable on balls. The previous fact ii)

shows that it is equivalent to prove that the unit ball of B Ψ is compact for the

topology of convergence on compact subsets. But this is easy: indeed, if fn in the

unit ball of B Ψ . This is a normal family thanks to Lemma 5.2. A subsequence

converges to an analytic function f on compact subsets of D and the Fatou Lemma

implies that f actually lies in the unit ball.

iv) Assume now that Ψ satisﬁes ∇2 . Since (M Ψ )∗∗ = LΨ (D, A ), we have

w∗

(BM Ψ )∗∗ = BM Ψ , in the space LΨ (D, A ). Hence it suﬃces to show that

w∗

BM Ψ = B Ψ . We already know that B Ψ is weak-star closed. Now, let f ∈ B Ψ .

Obviously, fr ∈ BM Ψ for every r ∈ (0, 1), where fr (z) = f (rz). Moreover fr Ψ ≤

66 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

**f Ψ . But it is clear that fr is uniformly convergent to f on compact subsets of
**

D, when r → 1− . By ii), the conclusion follows.

In the previous proof, we can see the points ii) and iii) in a slighlty diﬀerent

way: the unit ball B of B Ψ is compact for the topology τ of uniform convergence on

compact subsets. Then observe that the identity from B, equipped with the topol-

ogy τ , to B, equipped with the weak-star topology, is continuous (the sequential

argument is suﬃcient by metrizability). This implies, since the weak-star topology

is separated, that B is weak-star compact (hence closed) and that the topologies

coincide. Now, by Banach-Dieudonné’s theorem, the space B Ψ is weak-star closed.

**2. Compact composition operators on Bergman-Orlicz spaces
**

We shall begin with showing that, as in the Hardy-Orlicz case, every symbol φ

deﬁnes a bounded composition operator.

Proposition 5.4. Every analytic self-map φ : D → D induces a bounded com-

position operator Cφ : B Ψ → B Ψ . Moreover, Cφ f ∈ BM Ψ for every f ∈ BM Ψ ;

hence, when Ψ ∈ ∇2 , the former operator is the bi-transposed of Cφ : BM Ψ →

BM Ψ .

Proof. It suﬃces to follow the lines of Proposition 3.12, and to integrate the

integrals written there between 0 and 1, with respect to the measure 2rdr.

**Before stating and proving the main theorem, we are going to prove the follow-
**

ing auxiliary result, interesting in itself, and which reinforces an example of J. H.

Shapiro ([39], Example, page 185).

Proposition 5.5. There exists a Blaschke product B having angular derivative

at no point of T = ∂D, in the following sense:

(5.1) (∀ε > 0) (∃cε > 0) 1 − |B(z)| ≥ cε (1 − |z|)ε , ∀z ∈ D.

Proof. We shall take:

+∞

|zn | zn − z

B(z) = ,

z 1 − z̄n z

n=1 n

where:

{zn ; n ≥ 1} = An ,

n≥1

with:

An = {rn ωnj ; ωn = e2πi/pn , 0 ≤ j ≤ pn − 1} ,

where (rn )n≥1 is a (strictly) increasing sequence with 0 < rn < 1, and the integers

pn will have to be adjusted, satisfying the Blaschke condition:

+∞

+∞

(1 − |zn |) = pn (1 − rn ) < +∞.

n=1 n=1

One has:

+∞ z − z 2 +∞

n (1 − |z|2 )(1 − |zn |2 )

|B(z)|2 = = 1− ≤ exp − S(z) ,

n=1

1 − z̄n z n=1

|1 − z̄n z| 2

2. COMPACT COMPOSITION OPERATORS ON BERGMAN-ORLICZ SPACES 67

where:

+∞

(1 − |z|2 )(1 − |zn |2 )

S(z) = ·

n=1

|1 − z̄n z|2

We now proceed to minorize S(z). For this purpose, we shall need the following

simple lemma, whose proof will be temporarily postponed.

Lemma 5.6. For every positive integer p and every a ∈ D, one has, setting

ω = e2πi/p :

1

p−1

1 1 − |a|2p 1 1

= ≥ p |a|p .

p |1 − aω |

k 2 1 − |a| |1 − a |

2 p 2 4

k=0

**Then, setting r = |z|, we have:
**

+∞ n −1

p

1

S(z) ≥ (1 − r) (1 − rn )

n=1 k=0

|1 − rn ωn−k z|2

1−r 2

+∞

≥ p (1 − rn )(rn r)pn .

4 n=1 n

We shall take:

pn = (1 − rn )εn −1 + 1 ,

with:

1 1

εn = √ and rn = 1 − n

n 2

and where [ ] stands for the integer part. More explicitly:

√

pn = 2n− n + 1.

If r ≥ 1/2, let N ≥ 1 be such that rN < r ≤ rN +1 . One has:

1−r 1

S(z) ≥ (1 − rN )2εN −1 rN2pN ≥ (1 − rN )2εN rN2pN ,

4 8

since 1 − r ≥ 1 − rN +1 = (1 − rN )/2. Moreover:

pN ≤ 2(1 − rN )εN −1 = 2.2N (1−εN ) ≤ 2.2N ,

so that:

1

(1 − r)2εN (1 − 2−N )2

N +1

S(z) ≥ ≥ c(1 − r)2εN ,

8

where c is a positive numerical constant.

Hence, setting:

ε(z) = 2εN for |z| ≥ 1/2 and rN < r ≤ rN +1 ,

one has:

ε(z) −→− 0 ,

|z|→1

and we get:

1 − |B(z)|2 ≥ 1 − e−S(z) ≥ 1 − e−c(1−|z|) ≥ c (1 − |z|)ε(z) ,

ε(z)

**where c is another positive numerical constant. This gives condition (5.1), since
**

1 − |B(z)|2

1 − |B(z)| ≥ ·

2

68 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

**Finally, the Blaschke condition is satisﬁed, since:
**

√

pn (1 − rn ) ≤ 2(1 − rn )εn = 2.2− n

.

This ends the proof of Proposition 5.5.

Proof of Lemma 5.6. Let Gp be the ﬁnite group of pth roots of unity, equipped

with its normalized Haar measure. For u : Gp → C and 0 ≤ k ≤ p − 1, we denote

by û(k) the kth Fourier coeﬃcient of u, i.e.:

1

û(k) = u(z)z −k .

p

z∈Gp

**Then, the Plancherel-Parseval formula for Gp reads:
**

p−1

1

|û(k)|2 = |u(z)|2 .

p

k=0 z∈Gp

Applying this to

1

p−1

lp+k k

u(z) = = a z = û(k)z k ,

1 − az l≥0 k=0

0≤k≤p−1

with

ak

û(k) = alp+k = ,

1 − ap

l≥0

we get:

1 |a|2k

p−1 p−1

1 1 − |a|2p

= = ·

p |1 − aω |

k 2 |1 − a |

p 2 (1 − |a|2 )|1 − ap |2

k=0 k=0

To ﬁnish, we note that |1 − a | ≤ 2, and that, by the arithmetico-geometric in-

p

**equality, we have, with x = |a|2 :
**

1 − |a|2p

= 1 + x + · · · + xp−1

1 − |a|2

1/p p−1

≥ p x1+2+···+(p−1) = p x 2 ≥ p xp/2 = p |a|p .

Theorem 5.7. If the composition operator Cφ : B Ψ → B Ψ is compact, then

1

Ψ−1

(1 − |φ(a)|)2

(5.2) −→ 0.

−1

1 |a|→1−

Ψ

(1 − |a|)2

This condition is suﬃcient if Ψ ∈ Δ2 .

Before giving the proof of this theorem, let us note that in case where Ψ = Ψ2 ,

2

with Ψ2 (x) = ex − 1, it reads: the composition operator Cφ : B Ψ2 → B Ψ2 is

compact if and only if:

(5.3) (∀ε > 0) (∃cε > 0) 1 − |φ(z)| ≥ cε (1 − |z|)ε , ∀z ∈ D.

2. COMPACT COMPOSITION OPERATORS ON BERGMAN-ORLICZ SPACES 69

Indeed, Ψ2 ∈ Δ2 , and:

1 1

Ψ−1 log

2

(1 − |φ(a)|)2 (1 − |φ(a)|)2

= ;

−1 1 1

Ψ2 log

(1 − |a|)2 (1 − |a|)2

hence Cφ is compact if and only if

1

log

(1 − |φ(a)|)2

−→ 0.

1 |a|→1

log

(1 − |a|)2

Then, for every ε > 0, we can ﬁnd some cε > 0 such that, for all a ∈ D:

1 1

log ≤ ε log + cε ;

1 − |φ(a)| (1 − |a|)

which is equivalent to (5.3).

That allows to have compact composition operators on B Ψ2

which are not

compact on H Ψ2 . However it is very likely that this is the case for every Ψ ∈ Δ2 ,

but we have not tried to see this full generality.

Theorem 5.8. There exist symbols φ : D → D such that the composition op-

erators Cφ is compact from B Ψ2 into itself, but not compact from H Ψ2 into itself,

and even Cφ is an isometry onto its image.

A similar example is well-known for the Hilbert spaces B 2 and H 2 (see [39],

pages 180–186).

Proof. Let B be a Blaschke product verifying the condition of Proposition 5.5.

We introduce φ(z) = zB(z). The function still veriﬁes 1 − |φ(z)| ≥ Cε (1 − |z|)ε ,

since |φ(z)| ≤ |B(z)| on D. From Theorem 5.7, it follows, since Ψ2 ∈ Δ2 , that

Cφ : B Ψ2 → B Ψ2 is compact.

We are now going to see that Cφ : H Ψ2 → H Ψ2 is an isometry. Indeed, recall

the following well-known fact, that we already used (see [32], Theorem 1): since

φ is an inner function, the image φ(m) of the Haar measure m of T under φ is

equal to Pa .m, where a = φ(0) and Pa is the Poisson kernel at a. Here φ(0) = 0 so

φ(m) = m. It follows that for every f ∈ H Ψ2 , one has, for C > 0:

|f ◦ φ|
|f |

Ψ2 dm = Ψ2 dm

T C T C

so that f Ψ2 = f ◦ φΨ2 .

**We shall need also the following lemma, which completes Lemma 5.2.
**

Lemma 5.9. For every f ∈ BM Ψ , one has:

−1 1

f (a) = o Ψ as |a| −→ 1− .

(1 − |a|)2

Proof.

This is obvious

for the monomials en : z → z n since |en (a)| ≤ 1, whereas

−1

Ψ 1/(1 − |a|) −→ +∞. Since the evaluation δa is bounded on BM Ψ and

2

|a|→1

70 PASCAL LEFÈVRE, DANIEL LI, HERVÉ QUEFFÉLEC, LUIS RODRÍGUEZ-PIAZZA

−1

δa / Ψ (1/1 − |a|2 ) = O (1), it suﬃces to use that the polynomials are dense

in BM Ψ ; but this was already proved in Proposition 5.3.

**Proof of Theorem 5.7. If Cφ : B Ψ → B Ψ is compact, then so is the restriction
**

∗ ∗

Cφ : BM Ψ → BM Ψ and its adjoint Cφ∗ = Cφ : BM Ψ → BM Ψ . Since

w∗

Cφ∗ (δa ) = δφ(a) , Lemma 5.9 gives δa /δa −→ 0. Compactness of Cφ∗ now leads us

|a|→1

to

δ

Cφ∗

a

−→ 0.

δa |a|→1

That gives (5.2), in view of Lemma 5.2.

Conversely, assume that (5.2) is veriﬁed. Observe ﬁrst that, since Ψ ∈ Δ2 , one

has:

**(5.4) Ψ−1 (x2 ) ≤ αΨ−1 (x) for x large enough.
**

2

Indeed, let x0 > 0 be such that Ψ(αx) ≥ Ψ(x) for x ≥ x0 . For x ≥ y0 =

2

Ψ(αx0 ), with y = Ψ−1 (x2 ), one has x2 = Ψ(y) ≥ Ψ(y/α) , i.e. x ≥ Ψ(y/α),

and hence Ψ−1 (x2 ) = y ≤ αΨ−1 (x).

Therefore condition (5.2), which reads:

−1 1 −1 1

Ψ =o Ψ ,

(1 − |φ(z)|)2 (1 − |z|)2

reads as well, because of (5.4):

1

1

(5.5) Ψ−1 = o Ψ−1 , as |z| → 1.

1 − |φ(z)| 1 − |z|

**We have to prove that (5.5) implies the compactness of Cφ : B Ψ → B Ψ . So, by
**

Proposition 3.8, we have to prove that: for every sequence (fn )n in the unit ball of

B Ψ which converges uniformly to 0 on compact sets of D, one has fn ◦ φΨ −→ 0.

n→∞

But Lemma 5.2 and (5.4) imply that, for some C > 0:

1

(5.6) |fn (z)| ≤ C Ψ−1 , ∀z ∈ D .

1 − |z|

Let ε > 0 and set ε0 = ε/αC. Due to (5.5), we can ﬁnd some r with 0 < r < 1

such that:

⎧ √ 1

⎪ 1

⎪

⎪ 1−r ≤ ; Ψ−1 ≥ αx0 ;

⎨ 8 1−r

(5.7)
1

⎪

⎪ 1

⎪

⎩ Ψ−1 ≤ ε0 Ψ−1 if z ∈ D \ rD .

1 − |φ(z)| 1 − |z|

Then, since (fn )n converges uniformly to 0 on φ(rD), we have, for n large

enough (n ≥ n0 ):

|f ◦ φ|

n 1

Ψ dA (z) ≤ ·

rD ε 2

ACKNOWLEDGEMENT 71

**On the other hand, by (5.7):
**

|f ◦ φ|

n C −1 1

Ψ dA (z) ≤ Ψ Ψ dA (z)

D\rD ε D\rD ε 1 − |φ(z)|

ε0 C −1 1

≤ Ψ Ψ dA (z)

D\rD ε 1 − |z|

1 −1 1

= Ψ Ψ dA (z)

D\rD α 1 − |z|

$

1

≤ Ψ Ψ −1 dA (z)

D\rD 1 − |z|

1

since Ψ−1 ≥ αx0 ,

1−r

1

1 ρ dρ

= dA (z) = 2 √

D\rD 1 − |z| r 1−ρ

1

dρ √ 1

≤2 √ =4 1−r ≤ ·

r 1 − ρ 2

Putting together these two inequalities, we get, for n ≥ n0 :

|fn ◦ φ|
1 1

Ψ dA (z) ≤ + = 1,

D ε 2 2

and hence: fn ◦ φΨ ≤ ε, which ends the proof of Theorem 5.7.

Acknowledgement

Part of this work was made when the fourth-named author was Professeur

invité de l’Université d’Artois in May-June 2005 and during a visit of this author

in Lille and in Lens in March 2006. He wishes to thank both departements for their

hospitality. He was also partially supported by Spanish project MTM2006-05622. A

part was made too during a stay of the ﬁrst named author in the University of Sevilla

in September 2006. He is grateful to the departamento de Análisis Matemático.

References

**[1] J. Arazy, S. D. Fisher, and J. Peetre, Möbius invariant function spaces, J. Reine Angew.
**

Math. 363 (1985), 110–145. MR814017 (87f:30104)

[2] F. Bayart, Hardy spaces of Dirichlet series and their composition operators, Monatsh. Math.

136, no. 3 (2002), 203–236. MR1919645 (2003i:42032)

[3] F. Bayart, Compact composition operators on a Hilbert space of Dirichlet series, Illinois J.

Math. 47, no. 3 (2003), 725–743. MR2007233 (2004j:47049)

[4] C. Bennett and R. Sharpley, Interpolation of operators, Pure and Applied Math. 129, Aca-

demic Press (1988). MR928802 (89e:46001)

[5] O. Blasco de la Cruz and H. Jarchow, A note on Carleson measures for Hardy spaces, Acta

Sci. Math. (Szeged) 71, no. 1-2 (2005), 371–389. MR2160373 (2006d:30048)

[6] J. S. Choa and H. O. Kim, On function-theoretic conditions characterizing compact com-

position operators on H 2 , Proc. Japan Acad. Ser. A Math. Sci. 75, no. 7 (1999), 109–112.

MR1729855 (2000k:47027)

[7] J. S. Choa and H. O. Kim, Composition Operators between Nevanlinna-Type Spaces, J.

Math. Anal. Appl. 257, no. 2 (2001), 378–402. MR1827329 (2002c:47045)

[8] J. S. Choa, H. O. Kim, Hong Oh and J. H. Shapiro, Compact composition operators on

the Smirnov class, Proc. Amer. Math. Soc. 128, no. 8 (2000), 2297–2308. MR1653449

(2000k:47034)

[9] J. A. Cima and A. L. Matheson, Essential norms of composition operators and Aleksandrov

measures, Paciﬁc J. Math. 179, no. 1 (1997), 59–64. MR1452525 (98e:47047)

[10] C. C. Cowen and B. D. MacCluer, Composition operators on spaces of analytic func-

tions, Studies in Advanced Mathematics, CRC Press, Boca Raton, FL (1995). MR1397026

(97i:47056)

[11] P. L. Duren, Theory of H p spaces, Second edition, Dover Publications (2000).

[12] J. B. Garnett, Bounded Analytic Functions, Pure and Applied Math. 96, Academic Press

(1981). MR628971 (83g:30037)

[13] J. Gordon and H. Hedenmalm, The composition operators on the space of Dirichlet series

with square summable coeﬃcients, Michigan Math. J. 46, no. 2 (1999), 313–329. MR1704209

(2000f:47046)

[14] A. Hartmann, Interpolation and harmonic majorants in big Hardy-Orlicz spaces, J. Anal.

Math. 103 (2007), 197–219. Arxiv.org/abs/math.cv/0603624. MR2373268 (2008j:30051)

[15] H. Hunziker and H. Jarchow, Composition operators which improve integrability, Math.

Nachr. 152 (1991), 83–99. MR1121226 (93d:47061)

[16] N. Jaoua, Order bounded composition operators on the Hardy spaces and the Nevanlinna

class, Studia Math. 134, No.1 (1999), 35–55. MR1688214 (2000b:47069)

[17] N. Jaoua, Similarity to a contraction and hypercontractivity of composition operators, Proc.

Am. Math. Soc. 129, No.7 (2001), 2085–2092. MR1825921 (2002d:47038)

[18] H. Jarchow, Some functional analytic properties of composition operators, First International

Conference in Abstract Algebra (Kruger Park, 1993), Quaestiones Math. 18, no. 1-3 (1995),

229–256. MR1340480 (96f:47058)

[19] H. Jarchow, Compactness properties of composition operators, International Workshop on

Operator Theory (Cefalù, 1997), Rend. Circ. Mat. Palermo (2) Suppl., no. 56 (1998), 91–97.

MR1710825 (2000e:47047)

[20] H. Jarchow and J. Xiao, Composition operators between Nevanlinna classes and Bergman

spaces with weights, J. Operator Theory 46, no. 3, suppl. (2001), 605–618. MR1897157

(2003d:47034)

73

74 REFERENCES

**[21] J.-P. Kahane, Some random series of functions, Cambridge Studies in Adv. Math. 5. Cam-
**

bridge University Press, Cambridge, 1985. MR0833073 (87m:60119)

[22] M. A. Krasnosel’skiı̆ and Ja. B. Rutickiı̆, Convex functions and Orlicz spaces, P. Noordhoﬀ

Ltd., Groningen (1961). MR0126722 (23:A4016)

[23] P. Lefèvre, Some Characterizations of weakly compact operators on H ∞ and on the disk

algebra. Application to composition operators, J. Operator Theory 54:2 (2005), 229–238.

MR2186350 (2006k:47070)

[24] P. Lefèvre, D. Li, H. Queﬀélec, and L. Rodrı́guez-Piazza, Some translation-invariant Ba-

nach function spaces which contain c0 , Studia Math. 163 (2) (2004), 137–155. MR2047376

(2005b:46062)

[25] P. Lefèvre, D. Li, H. Queﬀélec, and L. Rodrı́guez-Piazza, A criterion of weak compactness for

operators on subspaces of Orlicz spaces, J. Funct. Spaces Appl. Vol. 6, No 3 (2008), 277–292.

MR2444526

[26] P. Lefèvre, D. Li, H. Queﬀélec, and L. Rodrı́guez-Piazza, Some examples of compact compo-

sition operators on H 2 , J. Funct. Analysis 255, No 11 (2008), 3098–3124. MR2464571

[27] P. Lefèvre, D. Li, H. Queﬀélec, and L. Rodrı́guez-Piazza, Compact composition operators on

H 2 and Hardy-Orlicz spaces, J. Math. Anal. Appl. 354 (2009), 360–371. MR2510446

[28] D. Li and H. Queﬀélec, Introduction à l’étude des espaces de Banach –Analyse et Probabilités,

Cours Spécialisés 12, Société Mathématique de France (2004). MR2124356 (2006c:46012)

[29] L. Liu, G. Cao, and X. Wang, Composition operators on Hardy-Orlicz spaces, Acta Math.

Sci. Ser. B Engl. Ed. 25, no. 1 (2005), 105–111. MR2119342 (2005k:47053)

[30] B. D. MacCluer, Compact composition operators on Hp (B N ), Michigan Math. J. 32 (1985),

no. 2, 237–248. MR783578 (86g:47037)

[31] V. Matache, A short proof of a characterization of inner functions in terms of the composition

operators they induce, Rocky Mountain J. Math. 35, no. 5 (2005), 1723–1726. MR2206032

(2007m:47052)

[32] E. A. Nordgren, Composition operators, Canad. J. Math. 20 (1968), 442–449. MR0223914

(36:6961)

[33] M. M. Rao and Z. D. Ren, Theory of Orlicz spaces, Pure and Applied Mathematics 146,

Marcel Dekker, Inc. (1991). MR1113700 (92e:46059)

[34] M. Rosenblum and J. Rovnyak, Topics in Hardy classes and univalent functions, Birkhäuser

Advanced Texts: Basler Lehrbücher, Birkhäuser Verlag, Basel (1994). MR1307384

(97a:30047)

[35] W. Rudin, Real and complex analysis, Third edition, McGraw-Hill Book Co., New York

(1987). MR924157 (88k:00002)

[36] D. Sarason, Composition operators as integral operators, Analysis and partial diﬀerential

equations, 545–565, Lecture Notes in Pure and Appl. Math., 122, Dekker, New York (1990).

MR1044808 (92a:47040)

[37] H. J. Schwartz, Composition operators on H p , Thesis, University of Toledo (1969).

[38] J. H. Shapiro, The essential norm of a composition operator, Annals of Math. 125 (1987),

375–404. MR881273 (88c:47058)

[39] J. H. Shapiro, Composition operators and classical function theory, Universitext. Tracts in

Mathematics, Springer-Verlag, New York (1993). MR1237406 (94k:47049)

[40] J. H. Shapiro, What do composition operators know about inner functions?, Monatsh. Math.

130, no. 1 (2000), 57–70. MR1762064 (2001a:47029)

[41] J. H. Shapiro, W. Smith, and D. A. Stegenga, Geometric models and compactness of com-

position operators, J. Funct. Anal. 127, no. 1 (1995), 21–62. MR1308616 (95m:47051)

[42] J. H. Shapiro and P. D. Taylor, Compact, Nuclear, and Hilbert-Schmidt Composition Oper-

ators on H 2 , Indiana Univ. Math. Journal 23 (6) (1973), 471–496. MR0326472 (48:4816)

[43] J.-O. Strömberg, Bounded mean oscillation with Orlicz norms and duality of Hardy spaces,

Bull. Amer. Math. Soc. 82, no. 6 (1976), 953–955. MR0419772 (54:7790)

[44] J.-O. Strömberg, Bounded mean oscillation with Orlicz norms and duality of Hardy spaces,

Indiana Univ. Math. J. 28, no. 3 (1979), 511–544. MR529683 (81f:42021)

[45] N. Zorboska, Compact composition operators on some weighted Hardy spaces, J. Operator

Theory 22, no. 2 (1989), 233–241. MR1043725 (91c:47062)

Editorial Information

To be published in the Memoirs, a paper must be correct, new, nontrivial, and sig-

niﬁcant. Further, it must be well written and of interest to a substantial number of

mathematicians. Piecemeal results, such as an inconclusive step toward an unproved ma-

jor theorem or a minor variation on a known result, are in general not acceptable for

publication.

Papers appearing in Memoirs are generally at least 80 and not more than 200 published

pages in length. Papers less than 80 or more than 200 published pages require the approval

of the Managing Editor of the Transactions/Memoirs Editorial Board. Published pages are

the same size as those generated in the style ﬁles provided for AMS-LATEX or AMS-TEX.

Information on the backlog for this journal can be found on the AMS website starting

from http://www.ams.org/memo.

A Consent to Publish and Copyright Agreement is required before a paper will be

published in the Memoirs. After a paper is accepted for publication, the Providence

oﬃce will send a Consent to Publish and Copyright Agreement to all authors of the

paper. By submitting a paper to the Memoirs, authors certify that the results have not

been submitted to nor are they under consideration for publication by another journal,

conference proceedings, or similar publication.

**Information for Authors
**

Memoirs is an author-prepared publication. Once formatted for print and on-line

publication, articles will be published as is with the addition of AMS-prepared frontmatter

and backmatter. Articles are not copyedited; however, conﬁrmation copy will be sent to

the authors.

Initial submission. The AMS uses Centralized Manuscript Processing for initial sub-

missions. Authors should submit a PDF ﬁle using the Initial Manuscript Submission form

found at www.ams.org/submission/memo, or send one copy of the manuscript to the follow-

ing address: Centralized Manuscript Processing, MEMOIRS OF THE AMS, 201 Charles

Street, Providence, RI 02904-2294 USA. If a paper copy is being forwarded to the AMS,

indicate that it is for Memoirs and include the name of the corresponding author, contact

information such as email address or mailing address, and the name of an appropriate

Editor to review the paper (see the list of Editors below).

The paper must contain a descriptive title and an abstract that summarizes the article

in language suitable for workers in the general ﬁeld (algebra, analysis, etc.). The descrip-

tive title should be short, but informative; useless or vague phrases such as “some remarks

about” or “concerning” should be avoided. The abstract should be at least one com-

plete sentence, and at most 300 words. Included with the footnotes to the paper should

be the 2010 Mathematics Subject Classiﬁcation representing the primary and secondary

subjects of the article. The classiﬁcations are accessible from www.ams.org/msc/. The

Mathematics Subject Classiﬁcation footnote may be followed by a list of key words and

phrases describing the subject matter of the article and taken from it. Journal abbrevi-

ations used in bibliographies are listed in the latest Mathematical Reviews annual index.

The series abbreviations are also accessible from www.ams.org/msnhtml/serials.pdf. To

help in preparing and verifying references, the AMS oﬀers MR Lookup, a Reference Tool

for Linking, at www.ams.org/mrlookup/.

Electronically prepared manuscripts. The AMS encourages electronically pre-

pared manuscripts, with a strong preference for AMS-LATEX. To this end, the Society

has prepared AMS-LATEX author packages for each AMS publication. Author packages

include instructions for preparing electronic manuscripts, samples, and a style ﬁle that gen-

erates the particular design speciﬁcations of that publication series. Though AMS-LATEX

is the highly preferred format of TEX, author packages are also available in AMS-TEX.

Authors may retrieve an author package for Memoirs of the AMS from www.ams.org/

journals/memo/memoauthorpac.html or via FTP to ftp.ams.org (login as anonymous,

enter your complete email address as password, and type cd pub/author-info). The

AMS Author Handbook and the Instruction Manual are available in PDF format from the

author package link. The author package can also be obtained free of charge by sending

email to tech-support@ams.org or from the Publication Division, American Mathematical

Society, 201 Charles St., Providence, RI 02904-2294, USA. When requesting an author

package, please specify AMS-LATEX or AMS-TEX and the publication in which your paper

will appear. Please be sure to include your complete mailing address.

After acceptance. The source ﬁles for the ﬁnal version of the electronic manuscript

should be sent to the Providence oﬃce immediately after the paper has been accepted for

publication. The author should also submit a PDF of the ﬁnal version of the paper to the

editor, who will forward a copy to the Providence oﬃce.

Accepted electronically prepared ﬁles can be submitted via the web at www.ams.org/

submit-book-journal/, sent via FTP, or sent on CD to the Electronic Prepress Depart-

ment, American Mathematical Society, 201 Charles Street, Providence, RI 02904-2294

USA. TEX source ﬁles and graphic ﬁles can be transferred over the Internet by FTP to

the Internet node ftp.ams.org (130.44.1.100). When sending a manuscript electronically

via CD, please be sure to include a message indicating that the paper is for the Memoirs.

Electronic graphics. Comprehensive instructions on preparing graphics are available

at www.ams.org/authors/journals.html. A few of the major requirements are given

here.

Submit ﬁles for graphics as EPS (Encapsulated PostScript) ﬁles. This includes graphics

originated via a graphics application as well as scanned photographs or other computer-

generated images. If this is not possible, TIFF ﬁles are acceptable as long as they can be

opened in Adobe Photoshop or Illustrator.

Authors using graphics packages for the creation of electronic art should also avoid the

use of any lines thinner than 0.5 points in width. Many graphics packages allow the user

to specify a “hairline” for a very thin line. Hairlines often look acceptable when proofed

on a typical laser printer. However, when produced on a high-resolution laser imagesetter,

hairlines become nearly invisible and will be lost entirely in the ﬁnal printing process.

Screens should be set to values between 15% and 85%. Screens which fall outside of this

range are too light or too dark to print correctly. Variations of screens within a graphic

should be no less than 10%.

Inquiries. Any inquiries concerning a paper that has been accepted for publication

should be sent to memo-query@ams.org or directly to the Electronic Prepress Department,

American Mathematical Society, 201 Charles St., Providence, RI 02904-2294 USA.

Editors

This journal is designed particularly for long research papers, normally at least 80 pages in

length, and groups of cognate papers in pure and applied mathematics. Papers intended for

publication in the Memoirs should be addressed to one of the following editors. The AMS uses

Centralized Manuscript Processing for initial submissions to AMS journals. Authors should follow

instructions listed on the Initial Submission page found at www.ams.org/memo/memosubmit.html.

Algebra, to ALEXANDER KLESHCHEV, Department of Mathematics, University of Oregon, Eu-

gene, OR 97403-1222; e-mail: ams@noether.uoregon.edu

Algebraic geometry, to DAN ABRAMOVICH, Department of Mathematics, Brown University,

Box 1917, Providence, RI 02912; e-mail: amsedit@math.brown.edu

Algebraic geometry and its applications, to MINA TEICHER, Emmy Noether Research Insti-

tute for Mathematics, Bar-Ilan University, Ramat-Gan 52900, Israel; e-mail: teicher@macs.biu.ac.il

Algebraic topology, to ALEJANDRO ADEM, Department of Mathematics, University of British

Columbia, Room 121, 1984 Mathematics Road, Vancouver, British Columbia, Canada V6T 1Z2; e-mail:

adem@math.ubc.ca

Combinatorics, to JOHN R. STEMBRIDGE, Department of Mathematics, University of Michigan,

Ann Arbor, Michigan 48109-1109; e-mail: JRS@umich.edu

Commutative and homological algebra, to LUCHEZAR L. AVRAMOV, Department of Math-

ematics, University of Nebraska, Lincoln, NE 68588-0130; e-mail: avramov@math.unl.edu

Complex analysis and harmonic analysis, to ALEXANDER NAGEL, Department of Mathemat-

ics, University of Wisconsin, 480 Lincoln Drive, Madison, WI 53706-1313; e-mail: nagel@math.wisc.edu

Diﬀerential geometry and global analysis, to CHRIS WOODWARD, Department of Mathemat-

ics, Rutgers University, 110 Frelinghuysen Road, Piscataway, NJ 08854; e-mail: ctw@math.rutgers.edu

Dynamical systems and ergodic theory and complex analysis, to YUNPING JIANG, Depart-

ment of Mathematics, CUNY Queens College and Graduate Center, 65-30 Kissena Blvd., Flushing, NY

11367; e-mail: Yunping.Jiang@qc.cuny.edu

Functional analysis and operator algebras, to DIMITRI SHLYAKHTENKO, Department of

Mathematics, University of California, Los Angeles, CA 90095; e-mail: shlyakht@math.ucla.edu

Geometric analysis, to WILLIAM P. MINICOZZI II, Department of Mathematics, Johns Hopkins

University, 3400 N. Charles St., Baltimore, MD 21218; e-mail: trans@math.jhu.edu

Geometric topology, to MARK FEIGHN, Math Department, Rutgers University, Newark, NJ

07102; e-mail: feighn@andromeda.rutgers.edu

Harmonic analysis, representation theory, and Lie theory, to E. P. VAN DEN BAN, De-

partment of Mathematics, Utrecht University, P.O. Box 80 010, 3508 TA Utrecht, The Netherlands;

e-mail: E.P.vandenBan@uu.nl

Logic, to STEFFEN LEMPP, Department of Mathematics, University of Wisconsin, 480 Lincoln

Drive, Madison, Wisconsin 53706-1388; e-mail: lempp@math.wisc.edu

Number theory, to JONATHAN ROGAWSKI, Department of Mathematics, University of Califor-

nia, Los Angeles, CA 90095; e-mail: jonr@math.ucla.edu

Number theory, to SHANKAR SEN, Department of Mathematics, 505 Malott Hall, Cornell Uni-

versity, Ithaca, NY 14853; e-mail: ss70@cornell.edu

Partial diﬀerential equations, to GUSTAVO PONCE, Department of Mathematics, South Hall,

Room 6607, University of California, Santa Barbara, CA 93106; e-mail: ponce@math.ucsb.edu

Partial diﬀerential equations and dynamical systems, to PETER POLACIK, School of Math-

ematics, University of Minnesota, Minneapolis, MN 55455; e-mail: polacik@math.umn.edu

Probability and statistics, to RICHARD BASS, Department of Mathematics, University of Con-

necticut, Storrs, CT 06269-3009; e-mail: bass@math.uconn.edu

Real analysis and partial diﬀerential equations, to DANIEL TATARU, Department of Mathe-

matics, University of California, Berkeley, Berkeley, CA 94720; e-mail: tataru@math.berkeley.edu

All other communications to the editors, should be addressed to the Managing Editor, ROBERT

GURALNICK, Department of Mathematics, University of Southern California, Los Angeles, CA 90089-

1113; e-mail: guralnic@math.usc.edu.

Titles in This Series

975 Javier Ribón, Topological classiﬁcation of families of diﬀeomorphisms without small

divisors, 2010

974 Pascal Lefèvre, Daniel Li, Hervé Queﬀélec, and Luis Rodrı́guez-Piazza,

Composition operators on Hardy-Orlicz spaces, 2010

973 Peter O’Sullivan, The generalised Jacobson-Morosov theorem, 2010

972 Patrick Iglesias-Zemmour, The moment maps in diﬀeology, 2010

971 Mark D. Hamilton, Locally toric manifolds and singular Bohr-Sommerfeld leaves, 2010

970 Klaus Thomsen, C ∗ -algebras of homoclinic and heteroclinic structure in expansive

dynamics, 2010

969 Makoto Sakai, Small modiﬁcations of quadrature domains, 2010

968 L. Nguyen Van Thé, Structural Ramsey theory of metric spaces and topological

dynamics of isometry groups, 2010

967 Zeng Lian and Kening Lu, Lyapunov exponents and invariant manifolds for random

dynamical systems in a Banach space, 2010

966 H. G. Dales, A. T.-M. Lau, and D. Strauss, Banach algebras on semigroups and on

their compactiﬁcations, 2010

965 Michael Lacey and Xiaochun Li, On a conjecture of E. M. Stein on the Hilbert

transform on vector ﬁelds, 2010

964 Gelu Popescu, Operator theory on noncommutative domains, 2010

963 Huaxin Lin, Approximate homotopy of homomorphisms from C(X) into a simple

C ∗ -algebra, 2010

962 Adam Coﬀman, Unfolding CR singularities, 2010

961 Marco Bramanti, Luca Brandolini, Ermanno Lanconelli, and Francesco

Uguzzoni, Non-divergence equations structured on Hörmander vector ﬁelds: Heat kernels

and Harnack inequalities, 2010

960 Olivier Alvarez and Martino Bardi, Ergodicity, stabilization, and singular

perturbations for Bellman-Isaacs equations, 2010

959 Alvaro Pelayo, Symplectic actions of 2-tori on 4-manifolds, 2010

958 Mark Behrens and Tyler Lawson, Topological automorphic forms, 2010

957 Ping-Shun Chan, Invariant representations of GSp(2) under tensor product with a

quadratic character, 2010

956 Richard Montgomery and Michail Zhitomirskii, Points and curves in the Monster

tower, 2010

955 Martin R. Bridson and Daniel Groves, The quadratic isoperimetric inequality for

mapping tori of free group automorphisms, 2010

954 Volker Mayer and Mariusz Urbański, Thermodynamical formalism and multifractal

analysis for meromorphic functions of ﬁnite order, 2010

953 Marius Junge and Javier Parcet, Mixed-norm inequalities and operator space Lp

embedding theory, 2010

952 Martin W. Liebeck, Cheryl E. Praeger, and Jan Saxl, Regular subgroups of

primitive permutation groups, 2010

951 Pierre Magal and Shigui Ruan, Center manifolds for semilinear equations with

non-dense domain and applications to Hopf bifurcation in age structured models, 2009

950 Cédric Villani, Hypocoercivity, 2009

949 Drew Armstrong, Generalized noncrossing partitions and combinatorics of Coxeter

groups, 2009

948 Nan-Kuo Ho and Chiu-Chu Melissa Liu, Yang-Mills connections on orientable and

nonorientable surfaces, 2009

947 W. Turner, Rock blocks, 2009

946 Jay Jorgenson and Serge Lang, Heat Eisenstein series on SLn (C), 2009

945 Tobias H. Jäger, The creation of strange non-chaotic attractors in non-smooth

saddle-node bifurcations, 2009

TITLES IN THIS SERIES

**944 Yuri Kifer, Large deviations and adiabatic transitions for dynamical systems and Markov
**

processes in fully coupled averaging, 2009

943 István Berkes and Michel Weber, On the convergence of ck f (nk x), 2009

942 Dirk Kussin, Noncommutative curves of genus zero: Related to ﬁnite dimensional

algebras, 2009

941 Gelu Popescu, Unitary invariants in multivariable operator theory, 2009

940 Gérard Iooss and Pavel I. Plotnikov, Small divisor problem in the theory of

three-dimensional water gravity waves, 2009

939 I. D. Suprunenko, The minimal polynomials of unipotent elements in irreducible

representations of the classical groups in odd characteristic, 2009

938 Antonino Morassi and Edi Rosset, Uniqueness and stability in determining a rigid

inclusion in an elastic body, 2009

937 Skip Garibaldi, Cohomological invariants: Exceptional groups and spin groups, 2009

936 André Martinez and Vania Sordoni, Twisted pseudodiﬀerential calculus and

application to the quantum evolution of molecules, 2009

935 Mihai Ciucu, The scaling limit of the correlation of holes on the triangular lattice with

periodic boundary conditions, 2009

934 Arjen Doelman, Björn Sandstede, Arnd Scheel, and Guido Schneider, The

dynamics of modulated wave trains, 2009

933 Luchezar Stoyanov, Scattering resonances for several small convex bodies and the

Lax-Phillips conjuecture, 2009

932 Jun Kigami, Volume doubling measures and heat kernel estimates of self-similar sets,

2009

931 Robert C. Dalang and Marta Sanz-Solé, Hölder-Sobolv regularity of the solution to

the stochastic wave equation in dimension three, 2009

930 Volkmar Liebscher, Random sets and invariants for (type II) continuous tensor product

systems of Hilbert spaces, 2009

929 Richard F. Bass, Xia Chen, and Jay Rosen, Moderate deviations for the range of

planar random walks, 2009

928 Ulrich Bunke, Index theory, eta forms, and Deligne cohomology, 2009

927 N. Chernov and D. Dolgopyat, Brownian Brownian motion-I, 2009

926 Riccardo Benedetti and Francesco Bonsante, Canonical wick rotations in

3-dimensional gravity, 2009

925 Sergey Zelik and Alexander Mielke, Multi-pulse evolution and space-time chaos in

dissipative systems, 2009

924 Pierre-Emmanuel Caprace, “Abstract” homomorphisms of split Kac-Moody groups,

2009

923 Michael Jöllenbeck and Volkmar Welker, Minimal resolutions via algebraic discrete

Morse theory, 2009

922 Ph. Barbe and W. P. McCormick, Asymptotic expansions for inﬁnite weighted

convolutions of heavy tail distributions and applications, 2009

921 Thomas Lehmkuhl, Compactiﬁcation of the Drinfeld modular surfaces, 2009

920 Georgia Benkart, Thomas Gregory, and Alexander Premet, The recognition

theorem for graded Lie algebras in prime characteristic, 2009

919 Roelof W. Bruggeman and Roberto J. Miatello, Sum formula for SL2 over a totally

real number ﬁeld, 2009

918 Jonathan Brundan and Alexander Kleshchev, Representations of shifted Yangians

and ﬁnite W -algebras, 2008

**For a complete list of titles in this series, visit the
**

AMS Bookstore at www.ams.org/bookstore/.

ISBN 978-0-8218-4637-7

9 780821 846377

MEMO/207/974

- BANACH SPACES AND RAMSEY THEORY: SOME OPEN PROBLEMSUploaded byWalner Mendonça Santos
- euclid.pjm.1102708251Uploaded byJiri Cepak
- S26Uploaded byandresgerardo123
- Topology Course Lexture Notes.psUploaded byMauricio Mallma
- smoothing-continuous-flows-on-two-manifolds.pdfUploaded byMarco Cotrina Teatino
- SINGULAR SETS AND MAXIMAL TOPOLOGIESUploaded byRichar Marcano
- An Optimization Approach to Kinetic Model Reduction ForUploaded byW00W
- Banach Contraction PrincipleUploaded byDaniel Pires
- G02107476.pdfUploaded byAJER JOURNAL
- Lect-20-RUploaded byinfiniti786
- e_4Uploaded byLuis Alberto Fuentes
- LectureNotes8UUploaded bySanjeev Shukla
- pdfUploaded byYoung Lion
- 10. IJAMSS - On Convergence Properties of Szasz Type Positive Linear OperatorUploaded byiaset123
- S. V. Astashkin- Disjointly Strictly Singular Inclusions of Symmetric SpacesUploaded byJutyyy
- zero lesson planUploaded byapi-252423002
- 1--Maths - IJMCAR - COMMON - Vishal Gupta - PaidUploaded byTJPRC Publications
- Equilateral Flat Unit Folding InstructionsUploaded byaddie_goodvibes
- qp2Uploaded byyogana
- SophiaUploaded byKayis Obispo
- mathgen-87198825Uploaded bySam
- arcUploaded byĐurđica Stanojković
- 1MJM080Uploaded byHo Nhat Nam
- Geometria Plana and Space - Aula 11-524Uploaded by4gen_5
- geometria violinoUploaded bybrunosls
- ComcorUploaded byagoszt
- A History of TopologyUploaded byvijayannv
- Digital_Project_v1r4_quick_start_guide.pdfUploaded byfraz
- ss_maths_f3c_2010Uploaded byChristian Lebon
- Katarzyna Pietruska-Paluba- The Reflected Brownian Motion on the Sierpinski GasketUploaded byIrokk

- Prachin Bharat Ka Itihaas OLD NCERT Ancient History of INDIA in HINDI ]_2Uploaded byRavindra
- BSER_Raj_Adhyan-1.pdf.pdfUploaded byRavindra
- Rivers of IndiaUploaded byRavindra
- Prachin Bharat Ka Itihaas OLD NCERT Ancient History of INDIA in HINDIUploaded byRavindra
- Ch6_presNUploaded byRavindra
- Prachin Bharat ka Itihaas OLD NCERT Ancient History of INDIA in HINDI ]_2.pdfUploaded byRavindra