Applied Mathematical Sciences: Springer

Applied Mathematical Sciences
Volume 44
Editors
JoEo Marsden L. Sirovich F. John (deceased)
Advisors
Mo Ghil 1.K. Hale T. Kambe
1. Keller K. Kirchgiissner
B.Jo Matkowsky C.So Peskin
JoT. Stuart
Springer
New York
Berlin
Heidelberg
Barcelona
Budapest
Hong Kong
London
Milan
Paris
Santa Clara
Singapore
Tokyo
1. John: Partial Differential Equations, 4th ed. 34. KevorkianiCole: Perturbation Methods in
2. Sirovich: Techniques of Asymptotic Analysis. Applied Mathematics.
3. Hale: Theory of Functional Differential 35. Carr: Applications of Centre Manifold Theory.
Equations, 2nd ed. 36. BengtssoniGhillKlillen: Dynamic Meteorology:
4. Percus: Combinatorial Methods. Data Assimilation Methods.
5. von MisesiFriedrichs: Fluid Dynamics. 37. Saperstone: Semidynamical Systems in Infinite
6. Freiberger/Grenander: A Short Course in Dimensional Spaces.
Computational Probability and Statistics. 38. Lichtenberg/Lieberman: Regular and Chaotic
7. Pipkin: Lectures on Viscoelasticity Theory. Dynamics, 2nd ed.
8. Giacoglia: Perturbation Methods in Non-linear 39. PicciniiStampacchialVidossich: Ordinary
Systems. Differential Equations in Rn.
9. Friedrichs: Spectral Theory of Operators in 40. Naylor/Sell: Linear Operator Theory in
Hilbert Space. Engineering and Science.
10. Stroud: Numerical Quadrature and Solution of 41. Sparrow: The Lorenz Equations: Bifurcations,
Ordinary Differential Equations. Chaos, and Strange Attractors.
11. Wolovich: Linear Multivariable Systems. 42. Guckenheimer/Holmes: Nonlinear Oscillations,
12. Berkovitz: Optimal Control Theory. Dynamical Systems and Bifurcations of Vector
13. BlumaniCole: Similarity Methods for Fields.
Differential Equations. 43. OckendonlI'aylor: Inviscid Fluid Flows.
14. Yoshizawa: Stability Theory and the Existence 44. Pazy: Semi groups of Linear Operators and
of Periodic Solution and Almost Periodic Applications to Partial Differential Equations.
Solutions. 45. Glashoff/Gustafson: Linear Operations and
15. Braun: Differential Equations and Their Approximation: An Introduction to the
Applications, 3rd ed. Theoretical Analysis and Numerical Treatment
16. Lefschetz: Applications of Algebraic Topology. of Semi-Infinite Programs.
17. CollatzIWetterling: Optimization Problems. 46. Wilcox: Scattering Theory for Diffraction
18. Grenander: Pattern Synthesis: Lectures in Gratings.
Pattern Theory, Vol. I. 47. Hale et al: An Introduction to Infinite
19. MarsdeniMcCracken: Hopf Bifurcation and Its Dimensional Dynamical Systems--Geometric
Applications. Theory.
20. Driver: Ordinary and Delay Differential 48. Murray: Asymptotic Analysis.
Equations. 49. Ladyzhenskaya: The Boundary-Value Problems
21. Courant/Friedrichs: Supersonic Flow and Shock of Mathematical Physics.
Waves. 50. Wilcox: Sound Propagation in Stratified Fluids.
22. RouchelHabets/Laloy: Stability Theory by 51. Golubitsky/Schaeffer: Bifurcation and Groups in
Liapunov's Direct Method. Bifurcation Theory, Vol. I.
23. Lamperti: Stochastic Processes: A Survey of the 52. Chipot: Variational Inequalities and Flow in
Mathematical Theory. Porous Media.
24. Grenander: Pattern Analysis: Lectures in Pattern 53. Maida: Compressible Fluid Flow and System of
Theory, Vol. II. Conservation Laws in Several Space Variables.
25. Davies: Integral Transforms and Their 54. Wasow: Linear Turning Point Theory.
Applications, 2nd ed. 55. Yosida: Operational Calculus: A Theory of
26. Kushner/Clark: Stochastic Approximation Hyperfunctions.
Methods for Constrained and Unconstrained 56. Chang/Howes: Nonlinear Singular Perturbation
Systems. Phenomena: Theory and Applications.
27. de Boor: A Practical Guide to Splines. 57. Reinhardt: Analysis of Approximation Methods
28. Keilson: Markov Chain Models-Rarity and for Differential and Integral Equations.
Exponentiality. 58. DwoyermussainWoigt (eds): Theoretical
29. de Veubeke: A Course in Elasticity. Approaches to Turbulence.
30. Shiatycki: Geometric Quantization and Quantum 59. SandersIVerhulst: Averaging Methods in
Mechanics. Nonlinear Dynamical Systems.
31. Reid: Sturmian Theory for Ordinary Differential 60. GhillChiidress: Topics in Geophysical
Equations. Dynamics: Atmospheric Dynamics, Dynamo
32. MeisiMarkowitz: Numerical Solution of Partial Theory and Climate Dynamics.
Differential Equations.
33. Grenander: Regular Structures: Lectures in
Pattern Theory, Vol. III. . (continued following index)
A. pazy
Semigroups of Linear Operators

and Applications to
Partial Differential Equations
, Springer
Amnon Pazy
Council for Higher Education
Planning and Budgeting Committee
Jerusalem 91040
Israel
Editors
Jerrold E. Marsden L. Sirovich
Control and Dynamical Systems, 104-44 Division of Applied Mathematics
California Institute of Technology Brown University
Pasadena, CA 91125 Providence, RI02912
USA USA
Mathematics Subject Classification: H7D05, 35FlO, 35F25, 35G25
Library of Congress Cataloging in Publication Data

Pazy, A.
Semigroups of linear operators and applications to
partial differential equations.
(Applied mathematical sciences; v. 44)
Includes bibliographical references and index.
1. Differential equations, Partial. 2. Initial
value problems. 3. Semigroups of operators.
1. Title. II. Series: Applied mathematical sciences
(Springer-Verlag, New York Inc.); v.44.
Printed on acid-free paper.

© 1983 Springer-Verlag New York, Inc.
Softcover reprint of the hardcover Ist edition 1983
All rights reserved. This work may not be translated or copied in whole or in part without the
written permission of the publisher (Springer-Verlag New York, Inc., 175 Fifth Avenue, New
York, NY 1O0lO, USA), except for brief excerpts in connection with reviews or scholarly
analysis. Use in connection with any form of information storage and retrieval, electronic
adaptation, computer software, or by similar or dissimilar methodology now known or
hereafter developed is forbidden.
The use of general descriptive names, trade names, trademarks, etc., in this publication, even
if the former are not especially identified, is not to be taken as a sign that such names, as
understood by the Trade Marks and Merchandise Marks Act, may accordingly be used freely
by anyone.
Production coordinated by Brian Howe and managed by Francine Sikorski; manufacturing

supervised by Robert Paella.
Media conversion by Science Typographers, Inc., Medford, NY.
Printed and bound by Edwards Brothers, Inc., Ann Arbor, M1.
Printed in the United States of America.
9 8 7 6 5 4 3
ISBN-13: 978-1-4612-5563-5 e-ISBN-13: 978-1-4612-5561-1

001: 978-1-4612-5561-1
Preface to the Second Printing
This second printing of the book contains a few minor changes and
corrections. It is a pleasure for me to thank Peter Hess, Gunter Lumer,
R. de Roo, and Hans Sager for drawing my attention to many misprints
and some errors.
I am especially indebted to Shinnosuke Oharu, who went through the
whole book and recommended many valuable clarifications, modifications,
and corrections.
A. PAZY
Preface to the First Printing
The aim of this book is to give a simple and self-contained presentation of

the theory of sernigroups of bounded linear operators and its applications to
partial differential equations.
The book is a corrected and expanded version of a set of lecture notes
which I wrote at the University of Maryland in 1972-1973. The first three
chapters present a short account of the abstract theory of sernigroups of
bounded linear operators. Chapters 4 and 5 give a somewhat more detailed
study of the abstract Cauchy problem for autonomous and nonautonomous
linear initial value problems, while Chapter 6 is devoted to some abstract
nonlinear initial value problems. The first six chapters are self-contained
and the only prerequisite needed is some elementary knowledge of func-
tional analysis. Chapters 7 and 8 present applications of the abstract theory
to concrete initial value problems for linear and nonlinear partial differen-
tial equations. Some of the auxiliary results from the theory of partial
differential equations used in these chapters are stated without proof.
References where the proofs can be found are given in the bibliographical
notes to these chapters.
I am indebted to many good friends who read the lecture notes on which
this book is based, corrected errors, and suggested improvements. In partic-
ular I would like to express my thanks to H. Brezis, M.G. Crandall, and P.
Rabinowitz for their valuable advice, and to Danit Sharon for the tedious
work of typing the manuscript.
A. PAZY
Contents
Preface to the Second Printing v

Preface to the First Printing vii
Chapter 1
Generation and Representation
1.1 Uniformly Continuous Semigroups of Bounded Linear Operators 1
1.2 Strongly Continuous Semigroups of Bounded Linear Operators 4
1.3 The Hille-Yosida Theorem 8
1.4 The Lurner Phillips Theorem 13
1.5 The Characterization of the Infinitesimal Generators of Co Semigroups 17
1.6 Groups of Bounded Operators 22
1. 7 The Inversion of the Laplace Transform 25
1.8 Two Exponential Formulas 32
1.9 Pseudo Resolvents 36
1.10 The Dual Semigroup 38
Chapter 2
Spectral Properties and Regularity 42
2.1 Weak Equals Strong 42
2.2 Spectral Mapping Theorems 44
2.3 Semigroups of Compact Operators 48
2.4 Differentiability 51
2.5 Analytic Semigroups 60
2.6 Fractional Powers of Closed Operators 69
Chapter 3
Perturbations and Approximations 76
3.1 Perturbations by Bounded Linear Operators 76
3.2 Perturbations of Infinitesimal Generators of Analytic Semigroups 80
3.3 Perturbations of Infinitesimal Generators of Contraction Semigroups 81
3.4 The Trotter Approximation Theorem 84
x Contents
3.5 A General Representation Theorem 89

3.6 Approximation by Discrete Semigroups 94
Chapter 4
The Abstract Cauchy Problem 100
4.1 The Homogeneous Initial Value Problem 100
4.2 The Inhomogeneous Initial Value Problem 105
4.3 Regularity of Mild Solutions for Analytic Semigroups 110
4.4 Asymptotic Behavior of Solutions 115
4.5 Invariant and Admissible Subspaces 121
Chapter 5
Evolution Equations 126
5.1 Evolution Systems 126
5.2 Stable Families of Generators 130
5.3 An Evolution System in the Hyperbolic Case 134
5.4 Regular Soiutions in the Hyperbolic Case 139
5.5 The Inhomogeneous Equation in the Hyperbolic Case 146
5.6 An Evolution System for the Parabolic Initial Value Problem 149
5.7 The Inhomogeneous Equation in the Parabolic Case 167
5.8 Asymptotic Behavior of Solutions in the Parabolic Case 172
Chapter 6
Some Nonlinear Evolution Equations 183
6.1 Lipschitz Perturbations of Linear Evolution Equations 183
6.2 Serni1inear Equations with Compact Semigroups 191
6.3 Semilinear Equations with Analytic Semigroups 195
6.4 A Quasilinear Equation of Evolution 200
Chapter 7
Applications to Partial Differential Equations- Linear Equations 206
7.1 Introduction 206
7.2 Parabolic Equations-L2 Theory 208
7.3 Parabolic Equations-LP Theory 212
7.4 The Wave Equation 219
7.5 A SchrOdinger Equation 223
7.6 A Parabolic Evolution Equation 225
Chapter 8
Applications to Partial Differential Equations-Nonlinear Equations 230
8.1 A Nonlinear SchrOdinger Equation 230
8.2 A Nonlinear Heat Equation in R 1 234
8.3 A Semilinear Evolution Equation in R 3 238
8.4 A General Class of Serni1inear Initial Value Problems 241
8.5 The Korteweg-de Vries Equation 247
Bibliographical Notes and Remarks 252

Bibliography 264
Index 277
CHAPTER 1
Generation and Representation
1.1. Uniformly Continuous Semigroups of Bounded

Linear Operators
Definition 1.1. Let X be a Banach space. A one parameter family T(t),
o ::;;
t < 00, of bounded linear operators from X into X is a semigroup of
bounded linear operators on X if
(i) T(O) = I, (I is the identity operator on X).
(ii) T(t + s) = T(t)T(s) for every t, s ;::: 0 (the semigroup property).
A semigroup of bounded linear operators, T( t), is uniformly continuous if
limIlT(t) - III = O. (1.1 )
I~O
The linear operator A defined by
D(A) = {x E X: lim T(t)x - x eXists} (1.2)

no t
and
Ax = lim T(t)x - x = d+T(t)x I for x E D(A) (1.3)
1 ~o t dt 1=0
is the infinitesimal generator of the semigroup T(t), D(A) is the domain

of A.
This section is devoted to the study of uniformly continuous semigroups
of bounded linear operators. From the definition it is clear that if T( t) is a
uniformly continuous semigroup of bounded linear operators then
limIIT(s) - T(t)11 = O. ( 1.4)
s .... 1
2 Semigroups of Linear Operators
Theorem 1.2. A linear operator A is the infinitesimal generator of a uniformly

continuous semigroup if and only if A is a bounded linear operator.
PROOF. Let A be a bounded linear operator on X and set
T{t) = etA = f
n=O
(tAr.
n.
(l.5)
The right-hand side of (1.5) converges in norm for every t ~ 0 and defines,
for each such t, a bounded linear operator T( t). It is clear that T(O) = I and
a straightforward computation with the power series shows that T( t + s) =
T(t)T(s). Estimating the power series yields
IIT{t) - III ::s; tliAlletllAIl
and
T(t) - I II
II t -A ::s;IIAII'O~~tIIT(s)-III
which imply that T(t) is a uniformly continuous sernigroup of bounded

linear operators on X and that A is its infinitesimal generator.
Let T(t) be a uniformly continuous semigroup of bounded linear opera-
tors on X. Fix p > 0, small enough, such that II I - p - I ftT( s) ds II < 1.
This implies that p-1ftT(s)ds is invertible and therefore ftT(s)ds is
invertible. Now,
h - I ( T{ h) - I) f T{ s ) ds = h - I (loP T{ s + h) ds - loP T( s ) ds )
=h-l(~P+hT{S)dS- fohT(s)dS)
and therefore
h - I (T( h) - I) = ( h- I ~P + hT{ s ) ds - h - 110h T( s ) ds )( foP T{ s ) ds ) - I
(l.6)
Letting h ~ 0 in (1.6) shows that h - I (T( h) - I) converges in norm and
therefore strongly to the bounded linear operator (T( p) - 1)( f tT( s) ds) - I
which is the infinitesimal generator of T(t). 0
From Definition 1.1 it is clear tQat a sernigroup T( t) has a unique

infinitesimal generator. If T(t) is uniformly continuous its infinitesimal
generator is a bounded linear operator. On the other hand, every bounded
linear operator A is the infinitesimal generator of a uniformly continuous
sernigroup T( t). Is this sernigroup unique? The affirmative answer to this
question is given next.
I Generation and Representation 3
Theorem 1.3. Let T(t) and S(t) be uniformly continuous semigroups of

bounded linear operators. If
lim T( t) - I = A = lim S( t) - I ( 1.7)

t,j,O t t,j,O t
then T(t) = S(t) for t ~ O.

PROOF. We will show that given T> 0, S(t) = T(t) for 0 ~ t ~ T. Let
T > 0 be fixed, since t --+ II T(t) II and t --+ II S( t) II are continuous there is a
constant C such that II T(t)1I IIS(s)1I ~ C for 0 ~ s, t ~ T. Given f > 0 it
follows from (1.7) that there is a 8 > 0 such that
for 0 ~ h ~ 8. (1.8)
Let 0 ~ t ~ T and choose n ~ 1 such that tin < 8. From the semigroup
property and (1.8) it then follows that
IIT(t) - S(t)1I =IIT( n~) - s( n~ )11

11 - I II
~ k~O T ((n - k) -;;t) S (kt) n 1) t ) II
--;; - T ( (n - k - 1) -;;t) S ( k +
~ :~JT( (n - k- 1)~ )IIIIT( ~) - s( ~ )lllls( ~ )11 ~ Cn ;C ~~ f.
Since f> 0 was arbitrary T(t) = S(t) for 0 ~ t ~ T and the proof is
complete. 0
Corollary 1.4. Let T( t) be a uniformly continuous semigroup of bounded linear

operators. Then
a) There exists a constant w ~ 0 such that II T(t)1I ~ e wt .
b) There exists a unique bounded linear operator A such that T(t) = etA.
c) The operator A in part (b) is the infinitesimal generator of T(t).
d) t --+ T(t) is differentiable in norm and
dT(t) = AT(t) = T(t)A ( 1.9)

dt
PROOF. All the assertions of Corollary 1.4 follow easily from (b). To prove
(b) note that the infinitesimal generator of T(t) is a bounded linear operator
A. A is also the infinitesimal generator of etA defined by (1.5) and therefore,
by Theorem 1.3, T(t) = etA. 0
1.2. Strongly Continuous Semigroups of Bounded

Linear Operators
Throughout this section X will be a Banach space.
Definition 2.1. A semigroup T(t), 0 =::; t < 00, of bounded linear operators
on X is a strongly continuous semigroup of bounded linear operators if
lim T(t}x = x for every x E X. (2.1)
tiO
A strongly continuous semigroup of bounded linear operators on X will be

called a semigroup of class Co or simply a Co semigroup.
Theorem 2.2. Let T(t) be a Co semigroup. There exist constants w ~ 0 and

M ~ 1 such that
IIT(t}1I =::; Me wt for 0 =::; t < 00. (2.2).
PROOF. We show first that there is an '1/ > 0 such that II T(t) II is bounded
for 0 =::; t =::; '1/. If this is false then there is a sequence {tn} satisfying tn ~ 0,
limn .... ootn = 0 and II T(tn) II ~ n. From the uniform boundedness theorem
it then follows that for some x E X, II T(tn)xll is unbounded contrary
to (2.1). Thus, II T(t)1I =::; M for 0 =::; t =::; '1/. Since II T(O) II = 1, M ~ 1. Let
W = '1/-1 log M ~ O. Given t ~ 0 we have t = n'l/ + l) where 0 =::; l) < '1/ and
therefore by the semigroup property
IIT(t}1I = IIT(c5}T('l/rll =::; M n+ 1 =::; MM t/." = Me wt . D
Corollary 2.3. If T(t) is a Co semigroup then for every x EX, t -. T(t)x is a

continuous function from R ~ (the nonnegative real line) into X.
PROOF. Let t, h ~ O. The continuity of t -. T(t)x follows from

IIT(t + h}x - T(t}xlI =::; IIT(t}IIIIT(h}x - xII =::; MewtIlT(h}x - xII
and for t ~ h ~ 0
II T( t - h} x - T( t } x II =::; II T( t - h)ll II x - T( h } x II
=::; Me"'tll x - T(h}xll· D
Theorem 2.4. Let T( t) be a Co semigroup and let A be its infinitesimal

generator. Then
a) For x E X,
. 1
lim -h
ft+h T(s}x ds = T(t}x. (2.3)
h .... O t
b) For x E X, {T(S)X ds E D(A) and
A(foIT{S)Xds) = T{t)x - x. (2.4)
c) For x E D(A), T(t)x E D(A) and

d
dt T{t)x = AT{t)x = T{t)Ax. (2.5)
d) For x E D(A),
T{t)x - T{s)x = [T{T)AxdT = [AT(T)xdT. (2.6)

s s
PROOF. Part (a) follows directly from the continuity of t -+ T(t)x. To prove
(b) let x E X and h > o. Then,
T(h) - III III
h 0 T( S ) x ds = h 0 (T( s + h) x - T( s ) x ) ds
= -I fl+h T( s ) x ds - -11hT( s ) x ds
h I h 0
and as h!O the right-hand side tends to T(t)x - x, which proves (b). To
prove (c) let x E D(A) and h > o. Then
T( h) - I ( T( h) - I )
h T(t)x = T(t) h x -+ T(t)Ax as h !O.
(2.7)
Thus, T(t)x E D(A) and AT(t)x = T(t)Ax. (2.7) implies also that
d+
YtT(t)x = AT(t)x = T(t)Ax,
i.e., that the right derivative of T(t)x is T(t)Ax. To prove (2.5) we have to
show that for t > 0, the left derivative of T(t)x exists and equals T(t)Ax.
This follows from,
!i~ [ T( t ) x - ~(t - h) x _ T( t ) AX]
= limT(t - h)[ T(h); - x - AX] + lim (T(t - h)Ax - T(t)Ax),

h lO h lO
and the fact that both terms on the right-hand side are zero, the first since
x E D(A) and II T(t - h)1I is bounded on 0 ~ h ~ t and the second by the
strong continuity of T(t). This concludes the proof of (c). Part (d) is
obtained by integration of (2.5) from s to t. 0
Corollary 2.5. If A is the infinitesimal generator of a Co semigroup T(t) then

D(A), the domain of A, is dense in X and A is a closed linear operator.
PROOF. For every x E X set XI = ljtf:T(s)x ds. By part (b) of Theorem

2.4, XI E D~r t > 0 and by part (a) of the same theorem XI -+ X as
t !O. Thus D(A), the closure of D(A), equals X. The linearity of A is
evident. To prove its closedness let xn E D(A), xn -+ X and AXn -+ Y as
n -+ 00. From part (d) of Theorem 2.4 we have
T(t)x n - xn = {T(s )Axn ds. (2.8)
o
The integrand on the right-hand side of (2.8) converges to T(s)y uniformly
on bounded intervals. Consequently letting n -+ 00 in (2.8) yields
T(t)x - X = {T(s)yds. (2.9)
Dividing (2.9) by t > 0 and letting t ! 0, we see, using part (a) of Theorem
2.4, that X E D(A) and Ax = y. 0
Theorem 2.6. Let T( t) and S( t) be Co semigroups of bounded linear operators

with infinitesimal generators A and B respectively. If A = B then T(t) = S(t)
for t ~ O.
PROOF. Let X E D(A) = D(B). From Theorem 2.4 (c) it follows easily that
the function s -+ T(t - s)S(s)x is differentiable and that
d
ds T(t - s )S(s)x = -AT(t - s )S(s)x + T(t - s )BS(s)x
= -T(t - s)AS(s)x + T(t - s)BS(s)x = O.

Therefore s -+ T(t - s)S(s)x is constant and in particular its values at
s = 0 and s = t are the same, i.e., T(t)x = S(t)x. This holds for every
x E D(A) and since, by Corollary 2.5, D(A) is dense in X and T(t), S(t)
are bounded, T( t)x = S(t)x for every x E X. 0
If A is the infinitesimal generator of a Co semigroup then by Corollary

2.5, D( A) = X. Actually, a much stronger result is true. Indeed we have,
Theorem 2.7. Let A be the infinitesimal generator of the Co semigroup T(t). If

D(An) is the domain of An, then n ::"=ID(An) is dense in X.
PROOF. Let 6j) be the set of all infinitely differentiable compactly supported
complex valued functions on ]0, 00[. For x E X and cp E 6j) set
y = x(cp) = {X)cp(s)T(s)xds. (2.10)
o
Ifh>Othen
T(h) - I I (00
h Y = YiJo cp(s)[T(s + h)x - T(s)x] ds
(00 1
= Jo Yi[cp(s - h)- cp(s)]T(s)xds. (2.11 )
The integrand on the right-hand side of (2.11) converges as h ~ 0 to

-q/(s)T(s)x uniformly on [0,00[. Therefore y E D(A) and
Ay = lim T(hl-I y = -l°Ocpl(s)T(s)xds.

h ~o 0
Clearly, if cp E Gj) then cp(n), the n-th derivative of cp, is also in Gj) for
n = 1,2, .... Thus, repeating the previous argument we find that y E D( An)
Any=(-lrf~cp(n)(s)T(s)xds for n= 1,2, ...
o
and consequently y E n~~lD(An). Let Y be the linear span of {x(q:»:
x E X, q:> E Gj)}. Y is clearly a linear manifold. From what we have proved
so far it follows that Y ~ n~ ~ 1D(A n ). To conclude the proof we will show
that Y is dense in X. If Y is not dense in X, then by Hahn-Banach's
theorem there is a functional x* E X*, x* =1= 0 such that x*(y) = 0 for
every y E Y and therefore .
10
00
cp(s )x*(T(s)x) ds = x*(1oOO cp(s )T(s)x dS) = 0 (2.12)
for every x E X, cp E Gj). This implies that for x E X the continuous

function s --+ x*(T(s)x) must vanish identically on [0, oo[ since otherwise, it
would have been possible to choose cp E Gj) such that the left-hand side of
(2.12) does not vanish. Thus in particular for s = 0, x*(x) = O. This holds
for every x E X and therefore x* = 0 contrary to the choice of x*. 0
We conclude this section with a simple application of Theorem 2.4.
Lemma 2.8. Let A be the infinitesimal generator of a Co semigroup T( t )satis-

fying IIT(t)11 ::s; M for t ~ o. If x E D(A2) then
IIAxl12 ::s; 4M211A 2xII Ilxll. (2.13)
PROOF. Using (2.6) it is easy to check that for x E D(A2)
T(t)x - x = tAx + [(t - s )T(s )A 2x ds.
o
Therefore,
IIAxll ::s; t-I(IIT{t)xll + Ilxll) +t-I{(t - s)IIT{s)A 2X llds
::s; 2~ Ilxll + ~t IIA 2xll. (2.14)
Here we used that M ~ 1 (since II T(O) II = 1). If A 2x = 0 then (2.14) im-

plies Ax = 0 and (2.13) is satisfied. If A 2x =1= 0 we substitute t =
2I1xlll/21IA2XIl-I/2 in (2.14) and (2.13) follows. 0
EXAMPLE 2.9. Let X be the Banach space of bounded uniformly continuous

functions on 1- 00, oo[ with the supremum norm. For f E X we define
(T{t)f){s) = f{t + s).
8 Scmigroups of Linear Operator,
It is easy to check that T(t) is a Co semigroup satisfying II T(t)11 :s; I for

t ~ O. The infinitesimal generator of T( t) is defined on D( A) = {f: f E X,
/' exists, /' E X} and (Af)(s) = /'(s) for f E D(A). From Lemma 2.8 we
obtain Landau's inequality
(supl/'(s)1)2 :s; 4(suplf"(s)I)(suplf(s)l) (2.15)

where the sup are taken over I - 00,00[. Example 2.9 can be easily modified
to the case where X = LP( - 00, 00), I < P < 00.
1.3. The Hille-Yo sid a Theorem

Let T(t) be a Co semigroup. From Theorem 2.2 it follows that there are
constants w ~ 0 and M ~ ·1 such that II T(t)11 :s; Me W ( for t ~ O. If w = 0,
T( t) is called uniformly bounded and if moreover M = I it is called a Co
semigroup of contractions. This section is devoted to the characterization of
the infinitesimal generators of Co semigroups of contractions. Conditions on
the behavior of the resolvent of an operator A, which are necessary and
sufficient for A to be the infinitesimal generator of a Co semigroup of
contractions, are given.
Recall that if A is a linear, not necessarily bounded, operator in X, the
resolvent set p( A) of A is the set of all complex numbers A for which
AI - A is invertible, i.e., (A I - A) - I is a bounded linear operator in X. The
family RCA: A) = (A I - A) - I, A E p( A) of bounded linear operators is
called the resolvent of A.
Theorem 3.1 (Hille-Yosida). A linear (unbounded) operator A is the infini-

tesimal generator of a Co semigroup of contractions T( t), t ~ 0 if and only If
(i) A is closed and D( A) = X.
(ii) The resolvent set p(A) of A contains R + and for every A > 0
I
IIR(A: A)II :s; X. (3.1 )
PROOF OF THEOREM 3.1 (Necessity). If A is the infinitesimal generator of a

Co semigroup then it is closed and D( A) = X by Corollary 2.5. For A > 0
and x E X let
(3.2)
Since t --+ T(t)x is continuous and uniformly bounded the integral exists as
an improper Riemann integral and defines a bounded linear operator R(A)
satisfying
(3.3)
1 Generation and Representation 9
°
*10
Furthermore, for h >
T(hl- I R(A)X =
00 e-A1(T(t + h)x - T(t)x) dt
=
e Ah h- 1 100 e Ah
0 e-A1T(t)xdt - h
1he-AT(t)xdt.
0
(3.4)
As h ~ 0, the right-hand side of (3.4) converges to AR( A)X - x. This im-
plies that for every x E X and A> 0, R(A)X E D(A) and AR(A) =
AR(A) - I, or
(AI - A)R(A) = I. (3.5)
For x E D(A) we have
R(A)Ax = 1o e-A1T(t)Ax dt = 1 e-A1AT(t)x dt
00
0
00
= A (1000 e-A1T( t)x dt) = AR( A)x. (3.6)
Here we used Theorem 2.4 (c) and the closedness of A. From (3.5) and (3.6)
it follows that
R(A)(AI - A)x = x for x E D(A). (3.7)
Thus, R(A) is the inverse of AI - A, it exists for all A > and satisfies the
desired estimate (3.1). Conditions (i) and (ii) are therefore necessary. 0
°
In order to prove that the conditions (i) and (ii) are sufficient for A to be
the infinitesimal generator of a Co semigroup of contractions we will need
some lemmas.
Lemma 3.2. Let A satisfy the conditions (i) and (ii) of Theorem 3.1 and let
R (A : A) = (AI - A) - 1. Then
lim AR(A: A)x = x for x E X. (3.8)
A-+ 00
PROOF. Suppose first that x E D( A). Then

IIAR(A: A)x - xii = IIAR(A: A)xll
= IIR(A: A)Axll :S
1
XIIAxl1 -+ ° as A -+ 00.
But D(A) is dense in X and IIAR(A: A)II :S 1. Therefore AR(A: A)x -+ x

as A -+ 00 for every x E X. 0
We now define, for every A > 0, the Yosida approximation of A by

AA = AAR(A: A) = A2R(A: A) - AI. (3.9)
Ax is an approximation of A in the following sense:
Lemma 3.3. Let A satisfy the conditions (i) and (ii) of Theorem 3.1. If Ax is
the Yosida approximation of A, then
lim Axx = Ax for x E D{A). (3.1O)
x--> 00
PROOF. For x E D(A) we have by Lemma 3.2 and the definition of Ax that
lim Axx = lim AR{A: A)Ax = Ax. 0
x--> 00 x--> 00
Lemma 3.4. Let A satisfy the conditions (i) and (ii) of Theorem 3.1. If Ax is
the Yosida approximation of A, then Ax is the infinitesimal generator of a
uniformly continuous semigroup of contractions e tAA . Furthermore, for every
x E X, A, JL > 0 we have
(3.11)
PROOF. From (3.9) it is clear that Ax is a bounded linear operator and thus
is the infinitesimal generator of a uniformly continuous semigroup e tAA of
bounded linear operators (see e.g., Theorem 1.2). Also,
(3.12)
and therefore e tAA is a seroigroup of contractions. It is clear from the
definitions that etAA, etA~, Ax and AI' commute with each other. Conse-
quently
lIetAAx - etA'xll = Ilf ! (etsAAet(l-s)A.x) dsll

s ftlletsAAet(l-S)A'(AxX - A"x) lids s tllAxx - A.xll.
o
o
PROOF OF THEOREM 3.1 (Sufficiency). Let x E D(A). Then
lIetAAx - e'A~xll S tllAxx - A"xll s tllAxx - Axil + tllAx - A"xll.
(3.13)
From (3.13) and Lemma 3.3 it follows that for x E D(A), etAAx converges
as A -+ ex:> and the convergence is uniform on bounded intervals. Since
D(A) is dense in X and lIetAAIl S 1, it follows that
lim etAAx = T{t)x for every x E X. (3.14)
X--> 00
The limit in (3.14) is again uniform on bounded intervals. From (3.14) it

follows readily that the limit T(t) satisfies the semigroup property, that
T(O) = I and that II T( t) II s 1. Also t -+ T( t)x is continuous for t ~ 0 as a
uniform limit of the continuous functions t -+ etAAx. Thus T(t) is a Co
semigroup of contractions on X. To conclude the proof we will show that A

is the infinitesimal generator of T(t). Let x E D(A). Then using (3.14) and
Theorem 2.4 we have
T(t)x-x= lim (etA~x-x)= lim ltesA~AAxds=ltT(s)Axds.
A-+oo A-+OO 0 0
(3.15)
The last equality follows from the uniform convergence of etA~AAx to
°
T(t)Ax on bounded intervals. Let B be the infinitesimal generator of T(t)
and let x E D( A). Dividing (3.15) by t > and letting t ! we see that
x E D(B) and that Bx = Ax. Thus B ;;2 A. Since B is the infinitesimal
°
generator of T(t), it follows from the necessary conditions that 1 E p(B).
On the other hand, we assume (assumption (ii» that 1 E p(A). Since
B ;;2 A, (I - B)D(A) = (I - A)D(A) = X which implies D(B) =
(I - B)-IX = D(A) and therefore A = B. 0
Theorem 3.1 and its proof have some simple consequences which we now
state.
Corollary 3.5. Let A be the infinitesimal generator of a Co semigroup of

contractions T(t). If AA is the Yosida approximation of A, then
T(t)x = lim etA~x for x E X. (3.16)
A-+oo
PROOF. From the proof of Theorem 3.1 it follows that the right-hand side of
(3.16) defines a Co semigroup of contractions, S(t), whose infinitesimal
generator is A. From Theorem 2.6 it then follows that T(t) = S(t). 0

contractions T( t). The resolvent set of A contains the open right half-plane,
i.e., p(A) ;;2 {X: Re X > o} and for such X
1
IIR(X: A) II ::; Re X . (3.17)
PROOF. The operator R(X)x = ioOOe-AtT(t)x dt is well-defined for X satisfy-

ing Re X > 0. In the proof of the necessary part of Theorem 3.1 it was
shown that R(X) = (AI - A)-I and therefore p(A) ;;2 {X: Re X > a}. The
estimate (3.17) for R( X) is obvious. 0
The following example shows that the resolvent set of the infinitesimal
generator of a Co semigroup of contractions need not contain more than the
open right half-plane.
EXAMPLE 3.7. Let X = BU(O, 00), that is, the space of all bounded uni-
formly continuous functions on [0, 00[. Define
(T(t)f)(s) = f(t + s). (3.18)
12 Sernigroups of Linear Operators
T(t) is a Co semigroup of contractions on X. Its infinitesimal generator A is

given by
D(A) = {/:f and J' E X} (3.19)

and
(Af)(s)=J'(s) for fED(A). (3.20)
From Corollary 3.6 we know that p(A) ~ {A: Re A> O}. For every complex
A the equation (A - A)fPA = 0 has the nontrivial solution fPA(S) = e AS . If
Re A ~ 0, fPA E X and therefore the closed left half-plane is in the spectrum
a(A) of A.
Let T(t) be a Co semigroup satisfying II T(t)1I ~ e"" (for some w ;:::: 0).
Consider S(t) = e-W'T(t). S(t) is obviously a Co semigroup of contractions.
If A is the infinitesimal generator of T( t) then A - wI is the infinitesimal
generator of S(t). On the other hand if A is the infinitesimal generator of a
Co semigroup of contractions S(t), then A + wI is the infinitesimal genera-
tor of a Co semigroup T(t) satisfying II T(t)1I ~ e W'. Indeed, T(t) = eW'S(t).
These remarks lead us to the characterization of the infinitesimal generators
of Co semigroups satisfying II T(t)1I ~ e W'.
Corollary 3.8. A linear operator A is the infinitesimal generator of a Co

semigroup satisfying II T(t)11 ~ e W' if and only if
(i) A is closed and D(A) = X
(ii) The resolvent set p(A) of A contains the ray {A: 1m A = 0, A> w} and
for such A
1
IIR(A: A)II ~ A - w· (3.21)
We conclude this section with a result that is often useful in proving that
a given operator A satisfies the sufficient conditions of the Hille- Yosida
theorem (Theorem 3.1) and thus is the infinitesimal generator of a Co
semigroup of contractions.
Let X be a Banach space and let X* be its dual. We denote the value of
x* E X* at x E X by (x*, x) or (x, x*). If A is a linear operator in X its
numerical range S( A) is the set
S(A) = {(x*, Ax):x E D(A), IIxll = 1,
x* E X*, IIx*1I = I, (x*, x) = I}. (3.22)
Theorem 3.9. Let A be a closed linear operator with dense domain D( A) in X.

Let S(A) be the numerical range of A and let r: be the complement of S(A) in
C. If A E r: then AI - A is one-to-one and has closed range. Moreover, ifr: o
r:
is a component of satisfying p(A) (\ r:o
=1= 0 then the spectrum of A is
contained in the complement So 0/ Lo and
IIR(A: A)II ~ d(A :S(A») (3.23)
where d(A:S(A» is the distance o/A/rom SeA).

PROOF. Let A E L. If x E D(A), Ilxll = 1, x* E X*, IIx*1I = 1 and
(x*, x) = 1 then
0< d(A :S(A») ~ IA - (x*, Ax)1 = I(x*, AX - Ax)1 ~ IIAX - Axil

t 3.24)
and therefore AI - A is one-to-one and has closed range. If moreover
A E peA) then (3.24) implies (3.23) and
d(A:S(A») ~ IIR(A:A)II-I. (3.25)
It remains to show that if Lo is a component of E which has a nonempty

intersection with the resolvent set peA) of A then a(A) ~ So. To this end
consider the set peA) () Eo. This set is obviously open in Eo. But it is also
closed in Eo since An E peA) () Eo and An --+ A E Eo imply for n large
enough that d(An:S(A» > -!-d(A:S(A» > 0 and consequently for n
large enough IA - Ani < deAn :S(A». From (3.25) it then follows that for
large n, A is in a ball of radius less than IIR(An: A)II-I centered at An which
implies that A E peA) and therefore peA) () Eo is closed in Eo. The
connectedness of Eo then implies that peA) () Eo = Eo or peA) ;;2 Eo which
is equivalent to a(A) ~ So and the proof is complete. 0
1.4. The Lumer Phillips Theorem

In the previous section we saw the Hille-Yosida characterization of the
infinitesimal generator of a Co semigroup of contractions. In this section we
will see a different characterization of such infinitesimal generators. In order
to state and prove the result we need some preliminaries.
Let X be a Banach space and let X* be its dual. We denote the value of
x* E X* at x E X by (x*, x) or (x, x*). For every x E X we define the
duality set F(x) ~ X* by
F(x) = {x*: x* E X* and (x*, x) = IIxll2 = IIx*1I2}. (4.1)
From the Hahn-Banach theorem it follows that F(x) +- 0 for every x E X.
Definition 4.1. A linear operator A is dissipative if for every x E D(A) there

is a x* E F(x) such that Re(Ax, x*) ~ O.
A useful characterization of dissipative operators is given next.

Theorem 4.2. A linear operator A is dissipative if and only if

II(AI - A)xll 2: Allxll forall x E D(A) and A> o. (4.2)
PROOF. Let A be dissipative, A> 0 and x E D(A). If x* E F(x) and
Re (Ax, x*) .:5; 0 then
IIAX - Axllllxll 2: I(AX - Ax, x*)1 2: Re(Ax - Ax, x*) 2: Allxl12
and (4.2) follows at once. Conversely, let x E D(A) and assume that
Allxll .:5; IIAX - Axil for all A > O. IfyX' E F(AX - Ax) andzX' = yX'1 IlyX' II
then lizX'II = I and
Allxll .:5; IIAX - Axil = (AX - Ax, zX)
= ARe(x, zX) - Re(Ax, zX).:5; Allxll - Re(Ax, zX)
for every A > O. Therefore
I
Re(Ax, zX) .:5; 0 and Re(x, zX) 2: Ilxll - XIIAxll· (4.3)
Since the unit ball of X* is compact in the weak-star topology of X* the net
zX', A --+ 00, has a weak-star cluster point z* E X*, Ilz*11 .:5; 1. From (4.3) it
follows that Re(Ax, z*) .:5; 0 and Re(x, z*) 2: IIxll. But Re(x, z*) .:5;
I(x, z*) I .:5; IIxll and therefore (x, z*) = IIxll. Taking x* = ·llxllz* we have
x* E F(x) and Re (Ax, x*) .:5; O. Thus for every x E D(A) there is an
x* E F(x) such that Re (Ax, x*) .:5; 0 and A is dissipative. 0
Theorem 4.3 (Lumer-Phillips). Let A be a linear operator with dense domain

D(A) in X.
(a) If A is dissipative and there is a AO > 0 such that the range, R(AoI - A),
ofAoI - A is X, then A is the infinitesimal generator of a Co semigroup of
contractions on X.
(b) If A is the infinitesimal generator of a Co semigroup of contractions on X
then R(AI - A) = X for all A > 0 and A is dissipative. Moreover, for
every x E D(A) and every x* E F(x), Re (Ax, x*) .:5; O.
PROOF. Let A > 0, the dissipativeness of A implies by Theorem 4.2 that
IIAX - Axil 2: Allxll for every A> 0 and x E D(A). (4.4)
Since R(AoI - A) = X, it follows from (4.4) with A = AO that (AoI - A)-I
is a bounded linear operator and thus closed. But then AoI - A is closed
and therefore also A is closed. If R(AI - A) = X for every A > 0 then
p(A) :2]0, oo[ and IIR(A: A)II .:5; A- I by (4.4). It then follows from the
Hille-Yosida theorem that A is the infinitesimal generator of a Co semigroup
of contractions on X.
To complete the proof of (a) it remains to show that R(AI - A) = X for
all A > O. Consider the set
A = {A: 0 < A < 00 and R(AI - A) = X}.
Let A E A. By (4.4), A E p(A). Since p(A) is open, a neighborhood of A is

in p(A). The intersection of this neighborhood with the real line is clearly in
A and therefore A is open. On the other hand, let An E A, An --+ A > 0. For
every y E X there exists an xn E D(A) such that
AnXn - AXn = y. (4.5)
From (4.4) it follows that Ilxnll ::;; A; lilyll ::;; C for some C > 0. Now,
Amllxn - xmll ::;; IIAm(xn - xm) - A(xn - xm)1I
= IAn - Amlllxnli ::;; ClAn - Ami· (4.6)
Therefore {xn} is a Cauchy sequence. Let xn --+ x. Then by (4.5) AX n --+
AX - y. Since A is closed, X E D(A) and AX - Ax = y. Therefore
R(AI - A) = X and A E A. Thus A is also closed in ]O,oo[ and since
AO E A by assumption A =F 0 and therefore A = ]0,00[. This completes the
proof of (a).
If A is the infinitesimal generator of a Co sernigroup of contractions, T(t),
on X, then by the Hille- Yosida theorem p (A) ;2]0, 00 [ and therefore R (A I
- A) = X for all A> 0. Furthermore, if x E D(A), X* E F(x) then
I(T(t)x,x*)I::;; IIT(t)xllllx*ll::;; IIxII2
and therefore,
Re(T(t)x - x, x*) = Re(T(t)x, x*) - Ilx112::;; 0. (4.7)
Dividing (4.7) by t > °and letting t ~ °yields
Re(Ax, x*) ::;; 0. (4.8)
This holds for every X* E F(x) and the proof is complete. 0
Corollary 4.4. Let A be a densely defined closed linear operator. If both A and
A* are dissipative, then A is the infinitesimal generator of a Co semigroup of
contractions on x.
PROOF. By Theorem 4.3(a) it suffices to prove that R(I - A) = X. Since
° °
A is dissipative and closed R(I - A) is a closed subspace of X. If
R(I - A) =F X then there exists x* E X*, x* =F such that (x*, x - Ax)
= for x E D(A). This implies x* - A*x* = 0. Since A* is also dissipative
it follows from Theorem 4.2. that x* = 0, contradicting the construction of
x*. 0
We conclude this section with some properties of dissipative operators.
Theorem 4.5. Let A be a dissipative operator in X.

(a) If for some AO > 0, R(AoI - A) = X then R(AI - A) = X for all A > 0.
(b) If A is closable then ::4, the closure of A, is also dissipative.
(c) If D(A) = X then A is closable.
PROOF. The assertion (a) was proved in the proof of part (a) of Theorem
4.3. To prove (b) let x E DCA), y = Ax. Then there is a sequence {x n}
Xn E D(A), such that xn -+ x and AXn -+ Y = Ax. From Theorem 4.2 it
follows that IIAX n - Axnll ~ Allxnll for A> 0 and letting n -+ 00 we have
IIAX - Axil ~ Alixil for A> O. (4.9)
Since (4.9) holds for every x E D(A), A is dissipative by Theorem 4.2. To
prove (c) assume that A is not closable. Then there is a sequence {xn} such
that xn E D(A), xn -+ 0 and AXn -+ y with lIyll = 1. From Theorem 4.2 it
follows that for every t > 0 and x E D( A)
lI(x + r1x n ) - tA(x + r1xn)1I ~ IIx + r1xnll·
Letting n -+ 00 and then t -+ 0 yields II x - y II ~ II x II for every x E D( A).
But this is impossible if D( A) is dense in X and therefore A is closable. 0
Theorem 4.6. Let A be dissipative with R(1- A) = X. If X is reflexive then

D(A) = X.
PROOF. Let x* E X* be such that (x*, x) = 0 for every x E D(A). We

will show that x* = O. Since R(1 - A) = X it suffices to show that
(x*, x - Ax) = 0 for every x E D(A) which is equivalent to (x*, Ax) = 0
for every x E D(A). Let x E D(A) then by Theorem 4.5 (a) there is an xn
such that x = xn - (l/n)Axn. Since AXn = n(xn - x) E D(A), xn E
D(A2) and Ax = AXn - (l/n)A2xn or AXn = (1- (l/n)A)-IAx. From
Theorem 4.2 it follows that 11(1- (l/n)A)-11i ~ 1 and therefore IIAxnll ~
IIAxli. Also, IIxn - xII ~ l/nllAxnll ~ l/nllAxll and therefore Xn -+ x.
Since IIAxnll ~ C and Xis reflexive there is a subsequence AXnk ofAxn such
that AXnk -+ y weakly. Since A is closed (see Theorem 4.3 (a» it follows that
y = Ax. Finally, since (x*, z) = 0 for every z E D(A), we have
(4.10)
Letting n k -+ 00 in (4.10) yields (x*, Ax) = O. This holds for every x E

D(A) and thereforex* = 0 and D(A) = X. 0
The next example shows that Theorem 4.6 is not true for general Banach
spaces.
EXAMPLE 4.7. Let X = C([O, I)), i.e., the continuous functions on [0, I] with
the sup norm. Let D(A) = {u: u E CI([O, I)) and u(O) = O} and Au = - u'
for u E D(A). For every f E X the equation AU - Au = f has a solution u
given by
(4.11 )
This shows that R(AI - A) = X. From (4.11) it also follows that

Alu{x)1 ~ (I - e-Ax)lIfll ~ IIAu - Aull. (4.12)
Taking the sup over x E [0,1] of the left-hand side of (4.12) we find that
Allull ~ IIAu - Aull and therefore A is dissipative by Theorem 4.2. But
D (A) = {u: u E X and u(o) = o} X = C([O, 1D.+
1.5. The Characterization of the Infinitesimal

Generators of Co Semigroups
In the previous two sections we gave two different characterizations of the
infinitesimal generators of Co semigroups of contractions. We saw at the end
of Section 1.3 that these characterizations yield characterizations of the
infinitesimal generators of Co semigroups of bounded operators satisfying
II T(t)1I ~ e"'t. We tum now to the characterization of the infinitesimal
generators of general Co semigroups of bounded operators. From Theorem
2.2 it follows that for such semigroups there exist real constants M ~ 1 and
'" such that IIT(t)1I ~ Me"'t. Using arguments similar to those used at the
end of Section 1.3, we show that in order to characterize the infinitesimal
generator in the general case it suffices to characterize the infinitesimal
generators of uniformly bounded Co semigroups. This will be done by
renorming the Banach space X so that the uniformly bounded Co semigroup
becomes, in the new norm, a Co semigroup of contractions and then using
the previously proved characterizations of the infinitesimal generators of Co
semigroups of contractions.
We start with a renorming lemma.
Lemma 5.1. Let A be a linear operator for which p(A) ::J ]0,00[. If
IIAnR{A: A)"II ~ M for n = 1,2, ... , A> 0, (5.1)
then there exists a norm I . I on X which is equivalent to the original norm

II . II on X and satisfies:
for x E X (5.2)
and
IAR{A: A)xl ~ Ixl for x E X, A > 0. (5.3)
PROOF. Let p. > ° and

IIxll" = sup IIp.nR{p. : A) nxII. (5.4)
n~O
Then obviously,
(5.5)
and
(5.6)
We claim that
for 0 < A ::; p.. (5.7)
Indeed, if y = R(A: A)x then y = R(p.: A)(x + (p. - A)Y) and by (5.6),
IlyllJ£ ::; ; IIxllJ£ + (1 - ;) lIyllJ£

whence AllyllJ£ ::; II X 111£ as claimed. From (5.5) and (5.7) it follows that
IIAnR(A: Arxll ::; IIAnR(A: ArxlIJ£::; IlxllJ£ for 0 < A::; p..
(5.8)
Taking the sup over n ~ 0 on the left-hand side of (5.8) implies that
IIxll~ ::; IlxllJ£ for 0 < A ::; JL. Finally, we define
Ixl = lim IIxIIJ£. (5.9)

1£--> 00
Then, (5.2) follows from (5.5). Taking n = 1 in (5.8) we have

IIAR(A: A)xIIJ£::; IIxllJ£
and (5.3) follows upon letting p. --+ 00. o
Lemma 5.1 is closely related to the following observation. Let {By}, Y E f
be a family of uniformly bounded commuting linear operators. Then there
exists an equivalent norm on X for which all the By are contractions, if and
only if there is a constant M such that
(5.10)
for every finite subset {YI' Y2' ... ' Ym } of f. Indeed, it is clear that if there is
such an equivalent norm then (5.l0) is satisfied. On the other hand if (5.10)
is satisfied we define
(5.11)
where the sup is taken over all finite subsets of f (including the empty set),
and I . I is the desired equivalent norm. The weaker condition
IIBynxll ::; Mllxll for every Y E f and n ~ 0 (5.12)
is not sufficient, in general, to insure the existence of an equivalent norm on
X for which all By are contractions. In the special case where f = R + and
By = R ( y: A) for some fixed linear operator A, the previous lemma shows
that the weaker condition (5.12) suffices to insure such an equivalent norm.
Theorem 5.2. A linear operator A is the infinitesimal generator of a Co

semigroup T(t), satisfying II T(t)11 ~ M (M;::: 1), if and only if
(i) A is closed and D(A) is dense in X.
(ii) The resolvent set p(A) of A contains IR + and
for A> 0, n = 1,2,.... (5.13)

PROOF. Let T(t) be a Co semigroup on a Banach space X and let A be its
infinitesimal generator. If the norm in X is changed to an equivalent norm,
T(t) stays a Co sernigroup on X with the new norm. The infinitesimal
generator A does not change nor does the fact that A is closed and densely
defined change when we pass to an equivalent norm on X. All these are
topological properties which are independent of the particular equivalent
norm with which X is endowed.
Let A be the infinitesimal generator of a Co sernigroup satisfying
II T(t)1I ~ M. Define,
Ix I = sup II T{ t ) x II. (5.14)

(;:0,0
Then
(5.15)
and therefore I . I is a norm on X which is equivalent to the original norm

II . lion X. Furthermore,
IT{t)xl = supIIT{s)T{t)xll ~ supIIT{s)xll = Ixi (5.16)

s;:o,o s;:o,O
and T(t) is a Co semigroup of contractions on X endowed with the norm

I . I· It follows from the Hille-Y osida theorem and the remarks at the
beginning of the proof, that A is closed and densely defined and that
IR(A: A)I ~ A-I for A> 0. Therefore by (5.15) and (5.16) we have
IIRCA: A(xil ~ IRCA: A(xi ~ A-nixi ~ MA -nllxll

and the conditions (i) and (ii) are necessary.
Let the conditions (i) and (ii) be satisfied. By Lemma 5.1 there exists a
norm I . I on X satisfying (5.2) and (5.3). Considering X with this norm, A
is a closed densely defined operator with p(A) ::) ]0, oo[ and IR(A: A)I ~ A-I
for -;.. > 0. Thus by the Hille-Yosida theorem, A is the infinitesimal genera-
tor of a Co sernigroup of contractions on X endowed with the norm I . I·
Returning to the original norm, A is again the -infinitesimal generator of
T(t) and,
II T{t)xlI ~ I T{t)xl ~ Ixl ~ Mllxll
so II T(t)11 ~ M as required. The conditions (i) and (ii) are therefore also
sufficient. 0
20 Semi groups of Linear Operators
If T( t) is a general Co seroigroup on X then, by Theorem 2.2, there are

constants M ~ 1 and w such that
IIT(t)1I s Me w' . (5.17)
Consider the Co semigroup S( t) = e - W'T( t) then II S(t) II s M and A is the
infinitesimal generator of T(t) if and only if A - wI is the infinitesimal
generator of S(t). Using these remarks together with Theorem 5.2 we obtain
Theorem 5.3. A linear operator A is the infinitesimal generator of a Co

semigroup T(t) satisfying II T(t)1I s Me w' , if and only if
(i) A is closed and D(A) is dense in X.
(ii) The resolvent set p(A) of A contains the ray jw, oo[ and
IIR(A: Arll s M/(A - wr for A> w, n = 1,2, .... (5.18)
Remark 5.4. The condition that every real A, A > W, is in the resolvent set
of A together with the estimate (5.18) imply that every complex A satisfying
Re A > w is in the resolvent set of A and
forReA>w, n = 1,2, ....
(5.19)
PROOF. We define
R{A)X = fOe-A'T(t)xdt.
o
Since II T(t)11 s Me w' , R(A) is weB-defined for every A satisfying Re A > w.
An argument, identical to the argument used in the proof of Theorem 3.1,
shows that R(A) = R(A: A). To prove (5.19) we assume that Re A > w, then
ddA R(A:A)x = ddA fOo e-A1T(t)x dt = -l°Ote-IIIT(t)x dt.

0
Proceeding by induction we obtain
dn 00
dAnR(A:A)x = (-lrfo tne-A1T(t)xdt. (5.20)
On the other hand, from the resolvent identity

R(A: A) - R(p.: A) = (p. - A)R(A: A)R(p.: A)
it foBows that for every A E p(A), A -+ R(A: A) is holomorphic and
d 2
dAR(A:A)= -R(A:A). (5.21 )
Proceeding again by induction we find
dnR~~n:A) = (-lrn!R(A:Ar+1. (5.22)

Comparing (5.20) and (5.22) yields

R(A:A)"x =
(n - I)!
1 1 tn-1e-XtT(t)xdt
0
00
(5.23)
whence
IIR(A: A)"xil ~ M
(n - I)!
1 tn-1e(w-ReX)tllxlldt = (Re AM-
0
00
w)
n Ilxll.
We conclude this section by extending the representation formula of

Corollary 3.5 to the general case.
Theorem 5.5. Let A be the infinitesimal generator of a Co semigroup T( t) on

X. If Ax is the Yosida approximation of A, i.e.,A x = AAR(A: A) then
T(t)x = lim etA~x. (5.24)
A--> 00
PROOF. We start with the case where II T(t)11 ~ M. In the proof of Theorem
5.2 we exhibited a norm III . III on X which is equivalent to the original norm
II . lion X and for which T(t) is a Co semigroup of contractions. From
Corollary 3.5 it then follows that IlIetA~x - T(t)xlIl --+ 0 as A --+ 00 for every
x E X. Since III . III is equivalent to II . II (5.24) holds in X. In the general
case where II T( t) II ~ Me wt we have for w ~ 0, II T(t) II ~ M and therefore
by what we have just proved, the result holds. It remains to prove the result
for w > O. Let w > 0 and note that A --+ lIetA~11 is bounded for A> 2w.
Indeed,
lIetA~11 = e-XtlleX2R(X :A)tll
< e -'t i...

1\
A2k t k IlR(A: A)kll
~ ----''---:''':-----'--''- ~ Me(AW/X-W)t ~ Me 2wt .
- k=O k!
(5.25)
Next we consider the uniformly bounded semigroup S(t) = e-wtT(t) whose
infinitesimal generator is A - wI. From the first part of the proof we have
T(t)x = lim et(A-w[h+ wtx for x E X. (5.26)
A--> 00
A simple computation shows that
(A - wIh + wI = Ax+w + H(A)
where
H(A) = 2wI - w(w + 2A)R(A + w: A)
= W[WR(A + w: A) - 2AR(A + w: A)].
It is easy to check that IIH(A)II ~ 2w + (2w + A- l ( 2)M and that for
22 Scmigroups or Linear Operators
x E D(A) IIH(~)xll ~ M~-I(",2I1xll + 2"'IIAxll) --+ 0 as ~ -+ 00. There-

fore H(~)x -+ 0 as ~ -+ 00 for every x E X. Since
lIe 'H (A)x - xII ~ te/IlIl(A)IIIIH{~)xll
we have
for x EX. (5.27)
Finally, since H(~) and AA+w commute we have

lIe'A·x - T{t)xll ~ lIe ,A .+ 1H(A-..,)x - T{t)xll + lIe IA 'lIl1e ' If(A-W)x - xii.
(5.28) •
As ~ --+ 00 the first term on the right-hand side tends to zero by (5.26) while
the second term tends to zero by (5.25) and (5.27). Therefore
lim e'A·x = T{t)x for x EX
A-oo
and the proof is complete. o
1.6. Groups of Bounded Operators

Definition 6.1. A one parameter family T(t), - 00 < t < 00, of bounded
linear operators on a Banach space X is a Co group of bounded operators if it
satisfies
(i) T(O) = I,
(ii) T(t + s) = T(t)T(s) for - 00 < I, s < 00.
(iii) lim/_oT(t)x = x for x E X.
Definition 6.2. The infinitesimal generator A of a group T(t) is defined by
Ax = lim T{/)x - x (6.1)

1-0 I
whenever the limit exists; the domain of A is the set of all elements x E X
for which the limit (6.1) exists.
Note that in (6.1) t -+ 0 from both sides and not only t --+ 0+ as in the
case of the infinitesimal generator of a Co semigroup.
Let T( t) be a Co group of bounded operators. It is clear from the
definitions that for t ~ 0, T(t) is a Co semigroup of bounded operators
whose infinitesimal generator is A. Moreover, for t ~ 0, S(t) = T( - t) is
also a Co semigroup of bounded operators with ,the infinitesimal generator
- A. Thus if T( t) is a Co group of bounded operators on X, both A and - A
are infinitesimal generators of Co semigroups which are denoted by T +( t)
and Tjt} respectively. Conversely, if A and -A are the infinitesimal
generators of Co semigroups T +( t) and T _( t) then we will see that A is the
infinitesimal generator of a Co group T( t) given by

for t ~ 0
(6.2)
for t ::; 0
Theorem 6.3. A is the infinitesimal generator of a Co group of bounded

operators T(t) satisfying II T(t) II ::; Me w1tl if and only if
(i) A is closed and D( A) = X.
(ii) Every real A, IAI > W, is in the resolvent set peA) of A and for such A
n = 1,2, .... (6.3)
PROOF. The necessity of the conditions follows from the fact that both A
and - A are the infinitesimal generators of Co semigroups of bounded
operators satisfying the estimate II T(t)1I ::; Me wt . Since A is the infinitesi-
mal generator of such a semigroup it follows from Theorem 5.3 that A is
closed, D( A) = X and (6.3) is satisfied for A > w. Moreover, since -A is
also the infinitesimal generator of such a semigroup and clearly R(A: A) =
- R( - A : -A) it follows that (T( -A) = - (T(A) and that (6.3) is satisfied
for - A < - w. The conditions (i) and (ii) are therefore necessary.
If the conditions (i) and (ii) are satisfied it follows from Theorem 5.3 that
A and - A are the infinitesimal generators of Co semigroups T +( t) and
T -<t) respectively and that II T +(t)11 ::; Me wt . The semigroups T +(t) and
T -<t) commute since clearly etA, and e- tA ., where Ap is the Yosida
approximation of A, commute and by Theorem 5.5, T +(t)x = limA~ooetA,x
and L(t)x = liml'--+ooe-tA.x. If Wet) = T +(t)L(t) then Wet) is a Co
semigroup of bounded operators for t ~ O. For x E D(A) = D( - A) we
have
W(t)x - x = T (t) T+(t)x - x + _T__(-,-t,-)x_-_x

t - t t
~ Ax - Ax = 0 as t LO. ( 6.4)
Therefore, for x E D( A) we have W( t)x = x. Since D( A) is dense in X and
Wet) is bounded we have Wet) = lor L(t) = (T+(t))-I. Defining
fort ~ 0
(6.5)
fort ::; 0
we obtain a Co group of bounded operators satisfying IIT(t)11 ::; Me w1tl .

The conditions (i) and (ii) are therefore sufficient and the proof is complete.
D
Lemma 6.4. Let T(t) be a Co semigroup of bounded operators. If for every

t > 0, T(t)-I exists and is a bounded operator then S(t) = T(t)-I is a Co
semigroup of bounded operators whose infinitesimal generator is -A.
Moreover if
T(t) for t ~ 0
{
U( t) = T( _ t)-I
for t ~ 0
then U( t) is a Co group of bounded operators.
PROOF. The semi group property for Set) is obvious since
S(t + s) = T(t + S)-I = (T(t)T(s))-1 = T(S)-IT(t)-1 = S(s)S(t).
We prove the strong continuity of S(t). For s > 0 the range of T(s) is all of
X. Let x E X and let s > 1. There exists ay E X such that T(s)y = x. For
t < I we then have
= IIT(s - t)y - T(s)YII ~ 0 as t ! O.

Therefore, Set) is strongly continuous. Finally, for x E D(A) we have
· T(t)-IX - x
I1m I· T() T(t)-I X - x I· x - T(t)x A
= 1m t = 1m = - x
1.0 t 1.0 t uo t
and therefore - A is the infinitesimal generator of T( t) - I. The rest of the
proof is obvious. 0
Theorem 6.S. Let T(t) be a Co semigroup of bounded operators. If 0 E

p(T(to» for some to > 0 then 0 E p(T(t» for all t > 0 and T(t) can be
embedded in a Co group.
PROOF. In view of Lemma 6.4 it is sufficient to show that 0 E p(T(t» for
all t> o. Since 0 E p(T(to», T(to)n = T(nto) is one-to-one for every
n ~ 1. Let T(t)x = o. Choosing n such that nto > t we have T(nto)x =
T(nt o - t)T(t)x = 0 which implies x = o. Thus T(t) is one-to-one for
every t > o. We show next that R(T(t» = X for every t > O. This is clear
for t ~ to since by the semi group property R(T(t» => R(T(to» for t ~ to.
For t> to let t = kto + tl with 0 ~ tl < to. Then T(t) = T(tO)kT(t l ) and
therefore, again, R(T(t» = x. Thus T(t) is one-to-one and R(T(t» = X for
every t > 0 and by the closed graph theorem 0 E p(T(t» for all t > O. 0
Theorem 6.6. Let T(t) be a Co semigroup of bounded operators. If for some

So > 0, T(so) - I is compact, then T(t) is invertible for every t > 0 and T(l)
can be embedded in a Co group.
PROOF. In view of Theorem 6.5 it suffices to prove that T(so) is invertible.
If T(so) is not invertible then 0 E a(T(so» but by our assumption T(so) - I
is compact and so 0 is an eigenvalue of T(so) with finite multiplicity.
Let x f 0 be such that T(so)x = O. Set Sl = so/2 then T(sl)T(sl)x =
T(so)x = 0 and 0 is an eigenvalue of T(sl). Proceeding by induction we
define a sequenc". I 0 such that 0 is an eigenvalue of T(sn). If N(T(t» is
the null space of T(t) then clearly N(T(s» c N(T(t» for s :5: t. Let Qn =
N(T(sn» n {x: IIxll = I}. Qn is a decreasing sequence of closed nonempty
subsets of X. Since N(T(so» has finite dimension, Qo IS compact and
consequently n;;O~oQn =F 0. If x E n;;O~oQn then
II T(sn)x - xii = Ilxll = 1 for all sn' (6.6)
But sn ..... 0 as n ..... 00 and therefore (6.6) contradicts the strong continuity
of T(t). This contradiction shows that T(so) must be invertible and the
proof is complete. 0
1.7. The Inversion of the Laplace Transform

One of the fundamental problems in the theory of semigroups of operators
is the relation between the semigroup and its infinitesimal generator. Given
a semigroup T( t) one obtains its infinitesimal generator, by definition, as
Ax = lim T(t)x - x for x E D(A).

t~O t
A different way of obtaining A, or rather the resolvent of A, is by Remark

5.4. There we showed that if II T(t)11 :5: Me wt then
R(A.:A)x = Iaooe-AtT(t)Xdt for x E X, ReA. > w. (7.1)
From the point of view of applications to partial differential equations it is

more interesting to obtain T( t) from its infinitesimal generator. The reason
for this is that for x E D( A), T( t)x is the solution of the initial value
problem
du
- - Au = 0 u(O) = x.
dt '
This section and the next one !ire dedicated to the problem of represent-
ing T( t) in terms of its infinitesimal generator. One way of doing this has
already been exhibited in Theorem 5.5. Here we will use a different method.
If T(t) satisfies II T(t)11 :5: Me wt then the resolvent of A satisfies (7.1), i.e.,
the resolvent of A is the Laplace transform of the semigroup. We therefore
expect to obtain the semigroup T( t) from the resolvent of A by inverting the
Laplace transform. This will be done in this section. We start with some
preliminaries.
Lemma 7.1. Let B be a bounded linear operator. Ify > IIBII then
e tB = - .
I jY+iOO e"tR(A:
, B) dA. (7.2)
2'1T1 y-ioo
The convergence in (7.2) is in the uniform operator topology and uniformly in t

on bounded intervals.
PROOF. Let y> IIBII. Choose r such that y > r > IIBII and let Cr be the
circle of radius r centered at the origin. For IAI > r we have
00 Bk
R(A: B) = L k+l(7.3)
A k~O
where the convergence is in the uniform operator topology uniformly for
IAI ~ r. Multiplying (7.3) by (l/2'ITi)e At and integrating over Cr term by
term yields.
e tB = 1
-2 .j eAtR(A: B) dA.
'IT I Cr
(7.4)
Here we used the identities
_l_j A-k-leAt dA = !..!:.... for k=0,1,2, .... (7.5)

2'ITi ~ k!
Since outside Cr the integrand of (7.4) is analytic and IIR(A: B)II :s; CiAI- 1,
we can shift the path of integration from Cr to the line Re z = y, using
Cauchy's theorem. 0
Lemma 7.2. Let A be the infinitesimal generator of a Co semigroup T(t)

satisfying II T(t)11 :s; Me wt . Let /l be real, /l > W ~ 0, and let
AIL = /lAR(/l: A) = /l2R(/l: A) - /lI (7.6)
be the Yosida approximation of A. Then for Re A > WJ.L/(J.L - w) we have
R(A: A,.) = (A + /l)-I(/lI - A)R( /l/l: A : A) (7.7)
and
(7.8)
For Re A > E + WJ.L/(J.L - w) and J.L > 2w, there is a constant C depending
only on M and E such that for every x E D(A)
C
IIR(A: AI')xlI :s; 1II(lIx ll + IIAxll)· (7.9)
PROOF. Multiplying the right-hand side of (7.7) from the right or from the
left by AI - AI' and using the commutativity of A and its resolvent, one
obtains the identity, thus proving (7.7). To prove (7.8) we note that AI' is the
infinitesimal generator of etA. and that by (5.25)
IletA·11 :s; Mexp {t(/l ~ w)}

which implies (7.8) by Theorem 5.3. Finally for Re A > E + WJ.L/(J,L - w) it
follows from (7.8) that IIR(A: A,,)II s Me-I. If x E D(A) and IL> 2w
then,
IIA"xll = IIILR(IL: A)Axll s ILM

p.-w IIAxll s 2MIIAxii
and therefore
IIR( A: A,,)xll = II + R( A: ~")A,,X lis*, (11xll + 2~2 IIAXII)

c o
s 1XT(llxll + IIAxll).
Lemma 7.3. Let A be as in Lemma 7.2, A = Y + i1] where y > w + e is fixed.

For every x E X we have
lim R(A: A,,)x = R(A: A)x (7.10)

,,---+ 00
and for every Y > 0, the limit is uniform in 1] for 11] I s Y.
PROOF. Set v = f.LA/(f.L + A). From (7.7) we then have for f.L large enough
R(A: A,,) - R(A: A)
= (IL + A)-I[(IL! - A)R(v: A) - (IL + A)R(A: A)]

= (IL + A)-I(IL! - A)R(v: A)[(A! - A)(IL! - A)
- (IL + A)(V! - A)] R(IL: A)R(A: A)
= (IL + A) -I(IL! - A)R(v: A)A 2R(IL: A)R(A: A)
= (IL + A)-I A 2R(v: A)R(A: A).
For y > w + e Theorem 5.3 implies IIR(A: A)II s Me-I. Given Y> 0, we
can find ILo depending on Y and y such that if A = y + i1], 11] I s Y and
f.L > f.Lo then Re f.LA /(f.L + A) > W + e /2. Thus, for f.L > f.Lo we have
IIR(v: A)II s 2Me- l . Therefore if x E D(A 2) and f.L > f.Lo, we have
IIR(A: A,,)x - R(A: A)xlI s IlL! AI IIR(v: A) II IIR(A: A)IIIIA 2xli
s! 2M2 IIA2xII
IL e2
and (7.10) follows for x E D(A 2). Since D(A2) is dense in X (Theorem 2.7),
and since by Lemma 7.2, IIR(A: A,,)II is uniformly bounded for Re A>
w + e provided that p. > w + w2 /e and by Theorem 5.3 the same is true for
II R( A: A) II, (7.10) follows for every x E X. 0
Theorem 7.4. Let A be the infinitesimal generator of a Co semigroup T( t)

satisfying II T(t)1I ~ Me w1 and let 'I > max (0, w). If x E D(A) then
1o T(s)xds
t
= -.
1 f'l+i oo dA
eAtR(A: A)x-, (7.11)
27Tl '1-ioo A
and the integral on the right converges uniformly in t for t in bounded intervals.
PROOF. Let IL > 0 be fixed and let 8 > IIAI1II. Set
_ 1
Pk(s)--2.
'TTl 6-ik
1
6 + ik As ( .
eRA.A l1 ) xdA. (7.12)
Integrating both sides of (7.12) from 0 to t and interchanging the order of

integration we find
i l Pk(s)ds--
o
- 1 1
6+ ik At ( . ) dA 1
2 'TTl. 6-ik e R A.A I1 x~--2.
f\
6+ ik ( .
'TTl 6-ik
R A.A I1 ) x~.
dA
f\
1
(7.13)
Letting k ~ 00 it follows from Lemma 7.1 that Pk(t) ~ etAp.x uniformly
on 0 ~ t ~ T. Also,
hm
.
k-+oo
16+~
6-ik
R(A: AI1)x~
dA
f\
= O. (7.14)
This can be seen by integrating A-IR(A: A I1 )x on the path r k composed of

rp) = {'I + i1/: - k ~ 1/ ~ k} and the semi circle rf)
= {'I + kei'P: - 'TT/2
~ cp ~ 'TT /2}. From Cauchy's theorem the integral around r k is zero. As
k ~ 00 the integral along q2) tends to zero since IIR(A:AI1)1I ~ CI1IAI-1
for IAI :2: 8. Therefore passing to the limit as k ~ 00 in (7.13) we find
(7.15)
If 'I > max( w, 0) it is clear from Lemma 7.2 that there is a ILo > 0 such that
for IL :2: ILo {A: Re A :2: 'I} is in p(A I1 ) and for x E D(A)
C
IIR(A: AIL)xll ~ W(lIxll + IIAxIl) (7.16)
where C depends only on M and y. Therefore for IL :2: ILo we can shift the
path of integration in (7.15) from Re A = 8 to Re A = 'I and obtain
itesA.x ds
o
= _1_.
2'TTl
1 Y + ioo eA1R(A:
y-ioo
AI1)x dA.
A
(7.17)
From (7.16) it follows that for x E D(A) the integral
jX e"YIIIR( 'I + i1/: A I1 )xll d1/ (7.18)

-x ';'1 2 +1/2
converges uniformly for IL :2: ILo and t on bounded intervals. For x E D(A)
the integral
(7.19)
also converges uniformly in t on bounded intervals since for Re"A > w

IIR("A :A)xll ::::; q"AI-1(lIxll + IIAxll). Finally, using Theorem 5.5, it is
clear that as p. -+ 00 the left-hand side of (7.17) converges to fdT(s)x ds
whereas by Lemma 7.3, (7.18) and (7.19) the right-hand side converges to
the right-hand side of (7.11) and the proof is complete. 0
Corollary 7.5. Let A be the infinitesimal generator of a Co semigroup T( t)

satisfying II T(t)11 ::::; Me w' . Let y > max (0, w). If x E D(A2), then
T(t}x = -.
1 jY+iOO A
e 'R("A: A)x d"A (7.20)
2'IT1 y-ioo
and for every 8 > 0, the integral converges uniformly in t for t E [8,1/8].
PROOF. If x E D(A2), then Ax E D(A). Using Theorem 7.4 for Ax we find
T(t)x - x = 1o 1
T(s)Axds = -2'
1 jY+iOO
'IT I
. eA'R("A:A)Ax~
Y-IOO
d"A
f\
_ 1.
--2 jY+iOO AI( ( . )
. e R"A.Ax-, X) d"A. (7.21)
'TTl Y-IOO f\
But
-
1 jY+iOO Al d"A
2'ITi y-ioo
e x-=x
"A
for t> ° (7.22)
and (7.22) converges uniformly in t for t E [8,1/8]. Combining (7.21) and

(7.22) gives (7.20). 0

satisfying II T( t) II ::::; Me w' . Let y > max (0, w). For every x E X we have
1 o
1
(t - s )T(s)x ds
1 jY+iOO A
= -2'
'IT I
d"A
e 'R("A: A)x-
y-ioo
2
"A
(7.23)
and the convergence is uniform in t on bounded intervals.

PROOF. Integrating (7.11) from ° to t we obtain
1o (t -
1 1 . lljY+iooeASR("A:A)xd"A"A ds
s)T(s)xds = -2
'IT I 0 Y-IOO
= -1. jY+iOO (e A1 _ I)R("A: A) xd"A

-2 .
2 'TTl y-ioo "A
But
1 lY+ioo dA
-2. 2 =0
R(A:A)x-
WI y-ioo A
and therefore (7.23) follows for x E D(A). The right-hand side of (7.23)
converges in the uniform operator topology and therefore defines a bounded
linear operator. Since D(A) is dense in X, (7.23) holds for every x E X. 0
We conclude this section with an important sufficient (but not necessary)

condition for an operator A to be the infinitesimal generator of a Co
semigroup. In contrast to the conditions of Theorem 5.2 and 5.3, the
conditions of Theorem 7.7 below, are often rather easy to check for concrete
examples.
Theorem 7.7. Let A be a densely defined operator in X satisfying the following

conditions.
(i) For some 0 < 8 < w/2, p(A) :::) Ea = {A: larg AI < w/2 + 8} U {O}.
(ii) There exists a constant M such that
M
IIR(A: A)II :-;; IXT for A E Ea , A =f= O. (7.24)
Then, A is the infinitesimal generator of a Co semigroup T(t) satisfying

II T(t)1I :-;; C for some constant C. Moreover
T(t) = -1-.1e"'R(A: A) dA (7.25)
2m r
where f is a smooth curve in E" running/rom ooe- i{} to ooe i{} for w/2 < -tt <
w/2 + 8. The integral (7.25) converges for t > 0 in the uniform operator
topology.
PROOF. Set
U(t) = 21 .1 e J"R(p,: A) dp,. (7.26)

m r
From (7.24) it follows easily that for t > 0 the integral in (7.26) converges in
the uniform topology. Moreover, since R(A: A) is analytic in E" we may
shift the path of integration in (7.26) to f, where f, = fl U f2 U f3 and
fl = {re-i{}:r l :-;; r < oo}, f2 = {t-1ei'l':- -tt:-;; <p:-;; -tt} and f3 = {rei{}:t- I
:-;; r < oo} without changing the value of the integral. But,
11
-1 -.1
2m r 3
eJ"R(p,: A) dP,ll:-;; _1 foo e-rlsin({}-fT/2)Mr-1 dr
2w I-I
The integral on fl is estimated similarly and on f2 we have
_1 ·1r el'tR(IL: A) dlLll < M fl) ecos,!, dcp ~ C

11 2'TTI
2 2'TT -I)
2•
Therefore there is a constant C such that II U(t)1I ~ C for 0 < t < 00. Next
we show that for ;\. > 0
R(;\.: A) = [Xl
o
e->..tU(t) dt. (7.27)
To this end we multiply (7.26) by e->..t and integrate from 0 to T. Using

Fubini's theorem and the residue theorem we find
[Te->..tU(t) dt
10
= _1_.1_ 1_(e(I'->..)T -
2m r IL - ;\.
I)R(IL: A) dlL
-1-1
= R("\1\ .• A) + 2 . e (I'->..)TR(IL: A)
'TTl r
"\ dIL·
IL - 1\
(7.28)
But,
as T --+ 00.
Therefore, passing to the limit as T --+ 00 in (7.28) we obtain (7.27). Since

II U(t)11 ~ C we can differentiate (7.27) n - I times under the integral sign
to find
d n- I 00
--R(;\.: A) = (-lr- 1 [ tn-1e->..tU(t) dt.
d;\.n-I 10
Since by (5.22)
d n- I R(;\.:A) = (-lr- l (n - 1)!R(;\.:Ar

d;\.n-I
we obtain
(7.29)
Therefore by Theorem 5.2, A is the infinitesimal generator of a Co semi-

group T(t) satisfying II T(t)1I ~ C. It remains to prove (7.25). Let x E
D(A 2 ). From Corollary 7.5 it follows that
T(t)x = -I. jY+ioo e>.. tR(;\': A)xd;\'. (7.30)

2 'TTl y-ioo
Using (7.24) we can shift the path of integration in (7.30) to r and so
T(t)x = -l-.le AtR(X: A)x dX (7.31)

2Wl r
holds for every x E D(A 2 ). Since by the first part of the proof the integral
freXtR(X:A) dX converges in the uniform operator topology and since
D(A2) is dense in X (Theorem 2.7) it follows that (7.31) holds for every
x E X whence the result. 0
1.8. Two Exponential Formulas

As we have already mentioned a Co semigroup T( t) is equal in some sense
to etA where A is the infinitesimal generator of T(t). Equality holds if A is a
bounded linear operator. In the case where A is unbounded Theorem 5.5
gives one possible interpretation to the sense in which T( t) "equals" etA. In
this section we give two more results of the same nature.
Theorem 8.1. Let T(t) be a Co semigroup on X. If

A(h)x= T(h)x-x (8.1)
h
then for every x E X we have
T(t)x = limetA(h)x (8.2)
h,J.O
and the limit is uniform in t on any bounded interval [0, T).

PROOF. Let II T(t)1I :s; Me w1 with CAl ~ 0 and let A be the infinitesimal
generator of T(t). Since for every h > 0 A(h) is bounded, e'A(h) is well-
defined. Furthermore, sinceA(h) and T(t) commute, so do etA(h) and T(t).
Also,
Therefore, for 0 < h :s; 1 we have

lIetA(h)1I :s; Met(eW-l).
It is easy to verify that for x E D(A), e(t-s)A(h)T(s)x is differentiable in s

and that
~ (e(t-S)A(h)T(s)x) = - A( h )e(t-s)A(h)T(s)x + e(t-s)A(h)AT(s)x
= e(t-s)A(h)T(s)(Ax - A(h)x).
Consequently, for 0 < h ~ I and x E D(A) we have
IIT{t)x - etA(h)xll =II{ !(e(t-S)A(h)T{S)X)dsll
~ flle(t-S)A(h)IIIIT{s)IIIIAx - A(h)xllds
o
~ tM 2e t(e w +w-l)IIAx - A(h)xll. (8.3)
Letting h ~O in (8.3) yields (8.2) for x E D(A). Since both lIetA(h)11 and
II T(t)11 are uniformly bounded on any finite t-interval and since D(A) is
dense in X, (8.2) holds for every x E X. 0
EXAMPLE 8.2. Let X = BU(R), i.e., X is the space of uniformly continuous

bounded functions on R. Let,
(T(t)f){x) = f(x + t) for - 00 < x < 00, 0 ~ t < 00. (8.4)
T(t) is a Co semigroup of contractions on X. Its infinitesimal generator A
has the domain
D(A) = {f: f EX, /' exists and /' E X}
and on D(A), Af = /'. For this semigroup we have
(A(h)f)(x) = f(x + hl- f(x) = (Ahf)(X).
It is easy to verify that

k
(A(h)kf)(X) = ~ L (_l)k-m(~)f(x + mh) = (A~f)(x).
h m=O
Using Theorem 8.l we obtain

tk
L F(A~f)(x).
00
f(x + t) = lim (8.5)
hJ.O k=O .
The limit in (8.5) exists uniformly with respect to x in R and uniformly with
respect to t on any finite interval. Formula (8.5) is a generalization of
Taylor's formula for an f which is merely continuous. Note that if f has k
continuous derivatives then
Theorem 8.3 (The exponential formula). Let T(t) be a Co semigroup on X. If

A is the infinitesimal generator of T(t) then
T(t)x= lim (I_!...A)-nX= lim [!!.R(!!. :A)]nX for xEX

n--+oo n n--+oo t t
(8.6)
and the limit is uniform in t on any bounded interval.
PROOF. Assume that II T(t)1I ::s; Me"'!. We have seen that for Re A > w,
R(A: A) is analytic in A and
R(A:A)x=lOOe-~ST(s)xds for xEX. (8.7)
o
Differentiating (8.7) n times with respect to A, substituting s = vt and taking
A = nit we find
R( ~ : A t\ = (-lrtn+lfo'~"( ve-vrT(tv)x dv.
But
and therefore
[ tn R ( t:
n
A
)] n+ I nn+ I
x = n!
100 (ve-vrT(tv)xdv.
0
Noting that
we obtain
[~R( ~:A)r+IX - T(t)x = n:~1 fooo(ve-Vr[T(vt)X - T(t)x] dv.
(8.8)
Given e > 0, we choose 0 < a < I < b < 00 such that t E [0, toJ implies
II T( tv ) x - T( t ) x II < e for a::S; v ::s; b.
Then we break the integral on the right-hand side of (8.8) into three
integrals II' 12 , 13 on the intervals [0, aJ, [a, bJ and [b, oo[ respectively. We
have
nn+1 1a
IIItil ::s; - , (ae-ar II T( vt)x - T(t)xlldv,
n. 0
111211
nn+1
::s; e-,-
fb (ve-vr dv < e
n. a
111311 = 11-,
nn+ I 100 (ve-V)n(T(tv)x - T(t)x) dvll·
n. b
Here we used the fact that ve- V ~ 0 is monotonically non decreasing for
o ::s; v ::s; 1 and non increasing on v ~ 1. Since furthermore ve- V < e- I for
v f 1, 111111 -+ 0 uniformly in t E [0, toJ as n -+ 00. Choosing n > wt in 13 ,
we see that the integral in the estimate of 13 converges and that 111311 -+ 0
uniformly in t E [0, tol as n ----> 00. Consequently,
li~s~pll[~R(~ :A)r+ IX
- T(t)xll~ e
and since e > 0 was arbitrary we have
lim
n~oo
[-nt R (n-t : A )]"+1 x = T( t ) X.
But by Lemma 3.2
lim
ll~OC
!!.t R(!!.t : A) x = X
and (8.6) follows. o

Remark 8.4. In section 7 we saw that T(t) can be obtained from the
resolvent of its infinitesimal generator by inverting the Laplace transform.
Theorem 8.3 gives us also an inversion of the Laplace transform which is
related to the Post-Widder real inversion formula, namely
f(t) = lim (- ~)k (~)k+IJ(k)(~)

k~'XJ k. t t
where Jis the Laplace transform of f.
Remark 8.5. The formula (8.6) has another interesting interpretation. Let A
be the infinitesimal generator of a Co semigroup T(t). Suppose we want to
solve the initial value problem
du
dt = Au, u(O) = x. (8.9)
A standard way of doing this is to replace (8.9) by
u
n
()t) _ u
n n
((j -n I) t ) _ ()t)
-~:"":"'---'---~-.!.- - AU n - , (8.10)
t n
n
which is an implicit difference approximation of (8.9). The equations (8.10)
can be solved explicitly and their solution u,,(t) is given by
un(t)= (/-~ArnX
un(t) is an approximation of the solution of (8.9) at t. Theorem 8.3 implies
that as n ----> 00, un(t) ----> T(t)x. From what we know already it is not
difficult to deduce that if x E D(A), T(t)x is the unique solution of (8.9).
Thus the solutions of the difference equations (8.10) converge to the solution
of the differential equation (8.9). If x rt:. D( A) then (8.9) need not have a
solution at all. The solutions of the difference equations do, nevertheless,
converge to T(t)x which should be considered as a generalized solution of

(8.9) in this case.
1.9. Pseudo Resolvents

We have seen that the characterization of the infinitesimal generator of a Co
semigroup on X is usually done in terms of conditions on the resolvent of A
(see e.g. Theorems 3.1 and 5.3). This is not an exceptional situation.
Indeed, in the study of unbounded linear operators on X it is often more
convenient to deal with their resolvent families which consist of bounded
linear operators. This short section is devoted to the characterization of
the resolvent family of an operator A in X by means of its main
properties.
Let A be a closed and densely defined operator on X and let R(A: A) =
(AI - A)-I be its resolvent. If IL and A are in the resolvent set p(A) of A,
then we have the resolvent identity
R(A: A) - R(IL: A) = (IL - A)R(A: A)R(IL: A). (9.1)
This identity motivates our next definition.
Definition 9.1. Let fl be a subset of the complex plane. A family J(A),

A E fl, of bounded linear operators on X satisfying
J(A) - J(IL) = (IL - A)J(A)J(IL) for A, IL E fl (9.2)
is called a pseudo resolvent on fl.
Our main objective in this section is to determine conditions under which

there exists a densely defined closed linear operator A such that J(A) is the
resolvent family of A.
Lemma 9.2. Let fl be a subset ofC (the complex plane). If J(A) is a pseudo
resolvent on fl, then J(A)J(IL) = J(IL)J(A). The null space N(J(A» and the
range, R(J(A», are independent of A E fl. N(J(A» is a closed subspace of X.
PROOF. It is evident from (9.2) that J(A) and J(IL) commute for A, IL E fl.
Also rewriting (9.2) in the form
J(A) = J(IL)[I + (IL - A)J(A)]
it is clear that R(J(IL» :J R(J(A» and, by symmetry, we have equality.
Similarly N(J(A» = N(J(IL». The closedness of N(J(A» is evident. D
Theorem 9.3. Let fl be a subset of C and let J( A) be a pseudo resolvent on fl.

Then, J(A) is the resolvent of a unique densely defined closed linear operator A
if and only if N(J(A» = {O} and R(J(A» is dense in X.
PROOF. Clearly if J(A) is the resolvent of a densely defined closed operator

A, we have N(J(A» = {O} and R(J(A» = D(A) is dense in X. Assume
now that N(J(A» = {O} and R(J(A» is dense in X. From N(J(A» = {O} it
follows that J(A) is one-to-one. Let Ao E Ll and define
A = AoI - J(Ao)-I. (9.3)
The operator A thus defined is clearly linear, closed and D(A) = R(J(Ao»
is dense in X. From (9.3) it is clear that
(AoI - A)J(Ao) = J(Ao)(AoI - A) = I, (9.4)
and therefore J(Ao) = R(Ao: A). If A E Ll then
(AI - A)J(A) = (A - Ao)I + (AoI - A))J(A)
= (A - Ao)I + (AoI - A))J(Ao)[I - (A - Ao)J(A)]
= I + (A - Ao)[J(Ao) - J(A) - (A - Ao)J(A)J(Ao)]
=1
and similarly J(A)(AI - A) = I. ThereforeJ(A) = R(A: A) for every A E Ll.
In particular A is independent of Ao and is uniquely determined by J(A). D
We conclude this section with two useful sufficient conditions for a

pseudo resolvent to be a resolvent.
Theorem 9.4. Let Ll be an unbounded subset of C and let J(A) be a pseudo

resolvent on Ll. If R(J(A» is dense in X and there is a sequence An E Ll such
that IAnl -+ 00 and
(9.5)
for some constant M then J( A) is the resolvent of a unique densely defined
closed linear operator A.
PROOF. From (9.5) it follows that IIJ(An)11 -+ 0 as n -+ 00. Let p. E Ll.
From (9.2) we deduce that
as n -+ 00. (9.6)
Therefore, if x is in the range of J(p.) we have
as n -+ 00. (9.7)
Since R(J(p.» is dense in X and AnJ(An) are uniformly bounded, we have
(9.7) for every x E X. If x E N(J(A» then AnJ(An)X = 0 and from (9.7) we
deduce that x = O. Thus N(J(A» = {O} and J(A) is the resolvent of a
densely defined closed operator A by Theorem 9.3. D
Corollary 9.5. Let Ll be an unbOunded subset of C and let J(A) be a pseudo

resolvent on Ll. If there is a sequence An E Ll such that IAnl -+ 00 as n -+ 00
and
lim AnJ{An)X = x forall x E X (9.8)
n-+oo
then J(A) is the resolvent of a unique densely defined closed operator A.

PROOF. From the uniform boundedness theorem and (9.8) it follows that
(9.5) holds. From Lemma 9.2 we know that R(J(A)) is independent of
A E !l and therefore (9.8) implies that R(J(A)) is dense in X. Thus, the
conditions of Theorem 9.4 hold and J(A) is the resolvent of an operator A.
o
1.10. The Dual Semigroup

We start with a few preliminaries. Let X be a Banach space with dual X* .
. We denote by (x*, x) or (x, x*) the value of x* E X* at x E X. Let S be a
linear operator with dense domain, D(S), in X. Recall that the adjoint S* of
S, is a linear operator from D(S*) c X* into X* defined as follows: D(S*)
is the set of all elements x* E X* for which there is a y* E X* such that
(x*,Sx) = (y*,x) forall xED{S) (1O.1)
and if x* E D(S*) theny* = S*x* wherey* is the element of X* satisfying
(10.1). Note that since D(S) is dense in X there is at most one y* E X* for
which (10.1) can hold.
Lemma 10.1. Let S be a bounded operator on X then S* is a bounded operator
on X* and IISII = IIS*II.
PROOF. For every x* E X*, (x*, Sx) is a bounded linear functional on X
and so it determines a unique element y* E X* for which (y*, x) =
(x*, Sx) and so D(S*) = X*. Moreover,
IIS*II = sup IIS*x*1I = sup sup I(S*x*, x)1
IIx'II~1 IIx'II~1 IIxll~1
= sup sup I(x*, Sx)1 = sup IISxll = IISII. 0

IIxll~1 IIx'II~1 IIxll~1
Lemma 10.2. Let A be a linear densely defined operator in X. If A E p(A)

then A E p(A*) and
R{A:A*) = R{A:A)*. (1O.2)
PROOF. From the definition of the adjoint we have (AI - A)* = AI* - A*
where 1* is the identity in X*. Since R(A: A) is a bounded operator
R (A : A)* is a bounded operator on X* by Lemma 10.1. We will prove that
R(A: A*) exists and that it equals R(A: A)*. First we show that AI* - A* is
one-to-one. If for some x* f 0, (AI* - A*)x* = 0 then 0 = «AI* -
A*)x*, x) = «AI - A)x, x*) for all x E D(A). But since A E p(A),
R(AI - A) = X and therefore x-*

x E X, x* E D(A*) then
= ° and A1* - A* is one-to-one. Now if
(x*,x) = (x*,(AI-A)R(A:A)x) = «AI* -A*)x*,R(A:A)x)

and therefore
R(A: A)*(AI* - A*)x* = x* for every x* E D(A*). (10.3)
On the other hand if x* E X* and x E D(A) then
(x*, x) = (x*, R(A: A)(AI - A)x) = (R(A: A)*x*,(AI - A)x)
which implies
(AI* - A*)R(A:A)*x* = x* for every x* E X*. (lOA)
From (10.3) and (lOA) it follows that A E p(A*) and that R(A: A*) =
R(A: A)*. 0
Let T( t), t ~ 0, be a Co semigroup on X. For t > let T(t)* be the °

adjoint operator of T(t). From the definition of the adjoint operator it is
clear that the family T*( t), I ~ 0, of bounded operators on X*, satisfies the
semigroup property. This family is therefore called the adjoint semigroup of
T(t). The adjoint semigroup however, need not be a Co semigroup on X*
since the mapping T( t) --+ T( t)* does not necessarily conserve the strong
continuity of T(t). Before we state and prove the main result of this section
concerning the relations between the semigroups T( t), T(t)* and their
infinitesimal generators we need one more definition.
Definition 10.3. Let S be a linear operator in X and let Y be a subspace of

X. The operator S defined by D(S) = {x E D(S) n Y: Sx E Y} and Sx =
Sx for x E D(S) is called the part of Sin Y.
Theorem 10.4. Let T( t) be a Co semigroup on X with the infinitesimal

generator A and let T(t)* be its adjoint semigroup. If A* is the adjoint of A
and Y* is the closure of D( A*) in X* then the restriction T( t) + of T( t)* to y*
is a Co semigroup on Y*. The infinitesimal generator A + of T( t) + is the part of
A* in Y*.
PROOF. Since A is the infinitesimal generator of T(t), there are constants w

and M such that for all real A, A > w, A E p(A) and
n = 1,2, .... ( 10.5)
This is a consequence of Theorem 5.3. From Lemma 10.2 and Lemma 10.1
it follows that if A > w, A E p(A*) and
IIR(A: A*rll ~ M n n=I,2, .... (1O.6)

(A - w)
Let J(A) be the restriction of R(A: A*) to Y*. Then obviously we have
IIJ(Arll :::;; (A ~wr' (10.7)
J(]")-J(p,) = (p,-]")J(]")J(p,) for ]..,p,>w (10.8)

and by Lemma 3.2
lim ]"J(A)X* = x* for every x* E Y*. (10.9)
A --> 00
From (10.8), (10.9) and Corollary 9.5 it follows that J(]") is a resolvent of a
closed densely defined operator A + in Y*. From (10.7) and Theorem 5.3
it follows that A + is the infinitesimal generator of a Co semigroup T(t)+
on Y*. For x E X and x* E Y* we have by the definitions
n = 1,2, ....
(10.10)
Letting n ~ 00 in (10.10) and using Theorem 8.3 we obtain
(x*, T(t)x) = (T(t)+ x*, x) (10.11)
and so for x* E Y*, T(t)*x* = T(t)+ x* and T(t)+ is the restriction of

T(t)* to Y*.
To conclude the proof we have to show that A + is the part of A* in Y*.
Let x* E D(A*) be such that x* E y* and A*x* E Y*. Then (]"I* - A*)x*
E y* and
(10.12)
Therefore x* E D(A+) and applying A1* - A+ on both sides of (10.12)

yields (]"I* - A*)x* = (]"I* - A +)x* and therefore A + x* = A*x*. Thus
A + is the part of A* in Y*. 0
In the special case where X is a reflexive Banach space we have,
Lemma 10.5. If S is a densely defined closed operator in X then D(S*) is

dense in X*.
PROOF. If D(S*) is not dense in X* then there is an element Xo E X such
that Xo =1= 0 and (x*, x o ) = 0 for every x* E D( S*). Since S is closed its
graph in X X X is closed and does not contain (0, x o ). From the Hahn-
Banach theorem it follows that there are xj, xi E X* such that (xj, x) -
(xi, Sx) = 0 for every x E D(S) and (xj,O) - (xi, x o ) =1= O. From the
second equation it follows that xi =1= 0 and that (xi, x o ) =1= O. But from the
first equation it follows that xi E D(S*) which implies (xi, x o ) = 0, a
contradiction. Thus D(S*) = X*. 0
As a consequence of Theorem 10.4 and Lemma 10.5 we have,
Corollary 10.6. Let X be a reflexive Banach space and let T( t) be a Co

semigroup on X with infinitesimal generator A. The adjoint semigroup T( t)* of
T(t) is a Co semigroup on X* whose infinitesimal generator is A* the adjoint
of A.
We conclude this section with a result in Hilbert space.
Definition 10.7. Let H be a Hilbert space with scalar product (,). An

operator A in H i~ symmetric if D(A) = H and A c A*, that is, (Ax, y) =
(x, Ay) for all x, y E D(A). A is self-adjoint if A = A*. A bounded operator
U on H is unitary if U* = U-I.
We recall that any adjoint operator is closed and that U is unitary if and
only if R(U) = Hand U is an isometry. Both these facts are easy to prove
and are left as exercises to the reader.
Theorem 10.8 (Stone). A is the infinitesimal generator of a Co group of

unitary operators on a Hilbert space H if and only if iA is self-adjoint.
PROOF. If A is the infinitesimal generator of a Co group of unitary operators
U(t), then A is densely defined (Corollary 2.5) and for x E D(A)
-Ax = limt-I(U( -t)x - x) = limt-I(U(t)*x - x) = A*x
t!O t!O
which implies that A = -A* and therefore iA = (iA)* and iA is self-

adjoint.
If iA is self adjoint then A is densely defined and A = - A*. Thus for

every x E D( A) we have
(Ax, x) = (x, A*x) = - (x, Ax) = -(Ax,x)
and therefore Re (Ax, x) = 0 for every x E D( A), i.e., A is dissipative.
Since A = -A* also Re(A*x, x) = 0 for every x E D(A*) = D(A) and
also A* is dissipative. By the remarks preceding the theorem it follows that
A and A* are closed and since A** = A, both A and A* = - A are the
infinitesimal generators of Co semigroups of contractions on H by Corollary
4.4. If U +( t) and U-(I) are the semigroups generated by A and A*
respectively we define
for t ~ 0
(10.13)
for t :::; O.
Then U(t) is a group (see Section 1.6) and since U(t)-I = U( - t), II U(t)11
:::; 1, II U( - t)11 :::; 1 it follows that R(U(t)) = X and U(t) is an isometry for
every t and thus U( t) is a group of unitary operators on H as desired. D
CHAPTER 2
Spectral Properties and Regularity
2.1. Weak Equals Strong

Let T( t) be a Co semigroup of bounded linear operators on a Banach space
X. Let A be its infinitesimal generator as defined in Definition 1.1.1. We
consider now the operator
Ax = w - lim T{h)x - x (1.1)

h!O h
where w - lim denotes the weak limit in X. The domain of A is the set of all
x E X for which the weak limit on the right-hand side of (l.l) exists. Since
the existence of a limit implies the existence of a weak limit, it is clear that A
extends A. That this extension is not genuine follows from Theorem 1.3
below. In the proof of this theorem we will need the following real variable
results.
Lemma 1.1. Let the real valued function w be continuous and differentiable
from the right on [a, b[. Let D+ w be the right derivative of w. If w(a) = 0 and
D+w(t) sOon [a, b[ then w(t) sOon [a, b[.
PROOF. Assume first that D + w (t) < o. If the result is false then there is a
tl EJa, b[ for which W(tl) > O. Let to = inf{t: w(t) > O}. By the continuity
of w, w(to) = 0 and by the definition of to we have a sequence Un} such that
tn ~ to and w(tn) > O. Therefore,
D+w{t o ) = lim w{tn) - w{to ) ~0

t n ! to tn - to
in contradiction to our assumption that D + w (t) < 0 and thus w (t) sOon
[a, b[.
2 Spectral Properties and Regularity 43
Returning to the general case where D + w (t) ::; 0 we consider for every
e > 0 the function w,(t) = w(t) - e(t - a). For w,(t) we have w,(a) = 0
and D + W, ::; - e < O. Therefore, by the first part of the proof, w,( t) ::; 0 on
[a, b[, i.e., w(t) ::; e(t - a). Since e > 0 is arbitrary, w(t) ::; 0 on [a, b[. D
Corollary 1.2. Let cp be continuous and differentiable from the right on [a, b[.
If D+cp is continuous on [a, b[ then cp is continuously differentiable on [a, b[.
PROOF. Let 1/; = D+cp and define X(t) = cp(a) + f:H'T) d'T. The function X
thus defined is clearly continuously differentiable on [a, b[. Let w(t) = X(t)
- cp(t) then w(a) = 0 and D+ w(t) = 0 on [a, b[. From Lemma 1.1 it then
follows that w(t) ::; 0 on [a, b[. Similarly -w(t) also satisfies the conditions
of Lemma 1.1 and therefore w(t) ~ O. Hence w(t) = 0 on [a, b[, i.e.,
cp (t) = X(t) and the proof is complete. D
Theorem 1.3. Let T(t) be a Co semigroup of bounded operators and let A be

its infinitesimal generator. If A is the operator defined by (1.1) then A = A.
PROOF. From the definitions of A and A it is clear that A :::) A. Let
x E D(A). Since bounded linear operators are weakly continuous, we have
W - l1m
. T( t + h) x - T( t ) x
=W-lm
1. T() ( T( h ) x - x )
t
h to h h to h
= T(t)( W -li~ T(h): - x) = T(t)Ax.
( l.2)
Therefore, if x E D( A) and x* E X* then
D+(x*, T(t)x) = (x*, T(t)Ax), (l.3)
i.e., the right derivative of (x*, T( t)x) exists on [0, oo[ and equals
(x*, T(t)Ax). But t ~ (x*, T(t)Ax) is continuous in t and so by Corollary
l.2 (x*, T( t)x) is continuously differentiable on [0, oo[ and its derivative is
(x*, T(t)Ax). Furthermore,
(x*, T(t)x - x) = (x*, T(t)x) - (x*, x) = {(x*, T(s)Ax) ds

o
=\X*,{T(S)AXdsj. (1.4)
Since (1.4) holds for every x* E X*, it follows from the Hahn-Banach
theorem that
T( t ) x - x = {T( s ) Ax ds. (l.5)
o
Dividing (1.5) by t > 0 and letting t ! 0, we obtain
· T(t)x - x
1Im A-
= x. (l.6)
t to t
Therefore x E D( A) and Ax = Ax. This implies A ::J A and thus A = A. D
Another result in which weak implies strong is the next theorem which we
state here without proof.
Theorem 1.4. If T(t) is a semigroup of bounded linear operators on a Banach

space X (Definition 1.1.1) satisfying
w - lim T( t ) x = X for every x E X (1.7)

t~O
then T(t) is a Co semigroup of bounded linear operators.
2.2. Spectral Mapping Theorems

Let T(t) be a Co semigroup on a Banach space X and let A be its
infinitesimal generator. In this section we will be interested in the relations
between the spectrum of A and the spectrum of each one of the operators
T(t), t ~ O. From a purely formal point of view one would expect the
relation a(T(t» = exp{(a(A)}. This, however, is not true in general as is
shown by the following example.
EXAMPLE 2.1. Let X be the Banach space of continuous functions on [0, 1]

which are equal to zero at x = 1 with the supremum norm. Define
(T(t)f)(x) = {f(x 0+ t)
if x+t::;1
if x+t>l
T(t) is obviously a Co semigroup of contractions on X. Its infinitesimal
generator A is given by
D(A) = {t:fE C1([0, 1]) () X,I' E X}

and
Af=1' for fE D(A).
One checks easily that for every A E C and g E X the equation Af - I' = g
has a unique solution f E X given by
f(t) = te~(t-S)g(s) ds.

t
Therefore a(A) = cpo On the other hand, since for every t ~ 0, T(t) is a
bounded linear operator, a(T(t» =F cp for all t ~ 0 and the relation a(T(t»
= exp{ ta( A)} does not hold for any t ~ O.
Lemma 2.2. Let T(t) be a Co semigroup and let A be its infinitesimal

generator. If
B}..{t)x = [eAU-S)T(s)x ds (2.1 )
o
then
for every x E X (2.2)
and
forevery x E D(A). (2.3)
PROOf. For every fixed A and t, BA(t) defined by (2.1) is a bounded linear
operator on X. Moreover, for every x E X we have
T(h)-I
--'--'---BA(t)x =
e Ah - I l eAU-')T(s)xds + -eHf'+heAU-S)T(s)xds
l .
h h II h I
(2.4)
As h to the right-hand side of (2.4) converges to ABA(t)x + T(t)x - eA1x

and consequently BA(t)x E D{A) and
ABA(t)x = ABA{t)x + T{t)x - eA1x (2.5)
which implies (2.2). From the definition of BA(t) it is clear that for
x E D(A), ABA(t)x = BA(t)Ax and (2.3) follows. 0

generator. Then,
o{T(t)):J e1o(A) for t ~ o. (2.6)
PROOf. Let eAI E p(T(t» and let Q = (eA1I - T(t))-I. The operators BA(t),
defined by (2.1), and Q clearly commute. From (2.2) and (2.3) we deduce
for every x E X (2.7)
and
for every x E D(A). (2.8)
Since BA(t) and Q commute we also have
BA(t)Q(AI - A)x = x for every x E D(A). (2.9)
Therefore, A E peA), BA{t)Q = (AI - A)-I = R(A: A) and p(T(t» c
exp{tp(A)} which implies (2.6). 0
We recall that the spectrum of A consists of three mutually exclusive

parts; the point spectrum op (A) the continuous spectrum oc( A) and the
residual spectrum orCA). These are defined as follows: A E op{A) if AI - A
is not one-to-one, A E oAA) if AI - A is one-to-one, AI - A is not onto

but its range is dense in X and finally A E O"r(A) if AI - A is one-to-one and
its range is not dense in X. From these definitions it is clear that O"p(A),
O"c(A) and O"r(A) are mutually exclusive and that their union is O"(A). In the
rest of this section we will study the relations between each part of the
spectrum of A and the corresponding part of the spectrum of T(t). We start
with the point spectrum.

generator. Then
(2.1O)
More precisely if A E O"p(A) then eXt E O"p(T(t)) and if eXt E O"p(T(t)) there
exists a k, k E 7L such that Ak = A + 27Tik/t E (Tp(A).
PROOF. If A E O"/A) then there is an Xo E D(A), Xo =F 0, such that (AI -
A)xo = O. From (2.3) it then follows that (eXtI - T(t))x o = 0 and therefore
eAt E O"p(T(t)) which proves the first inclusion. To prove the second inclu-
sion let eXt E O"p(T(t)) and let Xo =F 0 satisfy (eXtI - T(t))xo = O. This
implies that the continuous function s --+ e-xST(s)x o is periodic with period
t and since it does not vanish identically one of its Fourier coefficients must
be different from zero. Therefore there is a k, k E 7L such that
(2.ll)
We will show that Ak = A + 2'11'ik/t is an eigenvalue of A. Let II T(t)11 .::;;

Me wt • For Re JL > W we have
R{JL: A)xo = 1o e-P.sT{s )xo ds = L l(n+

00 00
n=O nl
1)1
e-P.sT{s )x o ds
L
00
= en(X-P.)t[e-P.ST{s )x o ds
n=O 0
= {I - e(X-P.)t)-I[e-P.ST{s)xods (2.12)
o
where we used the periodicity of e-xST(s)x o' The integral on the right-hand
side of (2.12) is clearly an entire function and therefore R(JL: A)xo can be
extended by (2.12) to a meromorphic function with possible poles at
An = A + 27Tin/t, n E 7L. Using (2.12) it is easy to show that .
(2.13)
and
(2.l4)
From the closedness of A and (2.l3), (2.14) it follows that Xk E D(A) and
that (AJ - A)Xk = 0, i.e., Ak E C1p (A). 0
We tum now to the residual spectrum of A.
Theorem 2.5. Let T(t) be a Co semigroup and let A be its infinitesimal

generator. Then,
(i) If A E a/A) and none of the An = A + 27Tin/t, n E Z is in O"p(A)
then eAt E O"r(T(t)).
(in If eAt E 0",(T(t) then none of the An = A + 27Tin/t, n E Z is in O"p(A)
and there exists a k, k E Z such that Ak E O",(A).
PROOF. If A E C1r(A) then there is an x* E X*, x* =f- 0, such that
(x*,(AI - A)x) = 0 for all x E D(A). From (2.2) it then follows that
(x*, (eAtI - T(t»x) = 0 for all x E X and therefore the range of eAtI -
T(t) is not dense in X. If eAtI - T( t) is not one-to-one then by Theorem 2.4
there is a k E Z such that Ak E O"/A) contradicting our assumption that
An fl C1p (A). Therefore eAtI - T(t) is one-to-one and eAt E C1,(T(t» which
concludes the proof of (i).
To prove (ii) we note first that if for some k, Ak = A + 27Tik/t E C1p(A)
then by Theorem 2.4 eAt E C1,,(T(t» contradicting the assumption that
eAt E 0",(T(t). It suffices therefore to show that for some k E Z, Ak E
O"r(A). This follows at once if we show that {An} C peA) U O"c(A) is
impossible. From (2.3) we have
for x E D(A) n E Z.
(2.15)
Since by our assumption eAt = e Ant E C1,(T(t» the left hand side of (2.15)
belongs to a fixed nondense linear subspace Y of X. On the other hand if
An E peA) U C1c (A) then the range of AnI - A is dense in X which implies
by (2.15) that the range of BA (t) belongs to Y for every n E Z. Writing
the Fourier series of the conti~uous function e-AsT(s)x we have
e-At
L
00 .
e-AST(s)x ~ -t- e(27rIn/t)SBAn(t)x (2.16)

n =-00
and each term on the right-hand side of (2.16) belongs to Y. As in the

classical numerical case the series (2.16) is (C.I) summable to e-AST(s)x for
o < s < t and therefore for 0 < S < t, e-AST(s)x E Y. Letting s ~O it
follows that every x E D(A) satisfies x E Y which is impossible since Y is a
proper closed subspace of X and D(A) is dense in X. 0
Theorem 2.6. Let T(t) be a Co semigroup and let A be its infiniteSimal

generator. If A E C1c (A) and if none of the An = A + 27Tin/t is in C1p (A) U
C1r (A) then eAt E C1c (T(t».
PROOF. From Theorem 2.3 it follows that if A E O'AA) then eAt E O'(T(t)).
If eAt E O'p(T(t)) then by Theorem 2.4 some Ak E O'p(A) and therefore
eAt ~ O'p(T(t)). Similarly if eAt E O'r(T(t)) then some Ak E O'r(A) and again
eAt ~ O'r(T(t)). 0
Remark. The converse of Theorem 2.6 does not hold. It is possible that
= A + 2'1Tin/t are in p(A).
eAt E O'c(T(t)) while all An
2.3. Semigroups of Compact Operators

Definition 3.1. A Co semigroup T(t) is called compact for t > to if for every
t > to, T(t) is a compact operator. T(t) is called compact if it is compact for
t> O.
Note that if T(t) is compact for t ~ 0, then in particular the identity is
compact and X is necessarily finite dimensional. Note also that if for some
to > 0, T(to) is compact, then so is T(t) for every t ~ to since T(t) =
T(t - t6)T(to) and T(t - to) is bounded.
Theorem 3.2. Let T(t) be a Co semigroup. If T(t) is compact for t > to, then
T(t) is continuous in the uniform operator topology for t > to.
PROOF. Let IIT(s)11 ~ M for 0 ~ s ~ 1 and let e> 0 be given. If t> to

then the set 0, = {T(t)x: IIxll ~ I} is compact and therefore, there exist
XI' x 2 , ... , x N such that the open balls with radius e/2(M + I) centered at
T(t)x j , I ~j ~ N cover 0,. From the strong continuity of T(t) it is clear
that there exists an 0 < ho ~ I such that
IIT(t + h)x j - T(t)xJ < e/2 for 0 ~ h ~ ho and 1 ~j ~ N.
(3.1)
Let x E X, II x II ~ 1, then thereis an indexj, 1 ~ j ~ N (j depending on x)
such that
IIT(t)x - T(t)xjll < e/2(M + 1). (3.2)
Thus, for 0 ~ h ~ ho and IIxll ~ 1, we have
IIT(t+ h)x - T(t)xlI ~ IIT(h)III1T(t)x - T(t)xjll
+ IIT(t + h)xj - T(t)xjll + IIT(t)xj - T(t)xlI < e (3.3)
which proves the continuity of T(t) in the uniform operator topology for
I > 10 , 0

generator. T(t) is a compact semigroup if and only if T(t) is continuous in the
uniform operator topology for I> 0 and R(A: A) is compact for A E p(A).
PROOF. Let II T(t)11 ::0:; Me w1 • If T(t) is compact for t > 0, then by Theorem
3.2, T(t) is continuous in the uniform operator topology for t > 0. There-
fore,
for Re A> W {3.4}
and the integral exists in the uniform operator topology. Let e > 0, Re A > W
and
R,{A} = 1• e-AST(s} ds.
00
(3.5)
Since T( s) is compact for every s > 0, R ,( A) is compact. But
IIR(A: A} - R.(A}II ::0:; Ilfe-AST(S} dsll::o:; eMe w , ~ ° as dO
and therefore R(A: A) is compact as a uniform limit of compact operators.

From the resolvent identity
R(A: A} - R(p.: A} = (p. - A}R(A: A}R(p.: A) A, p. E p(A}
it follows that if R(J.t: A) is compact for some J.t E p(A), R(A: A) is
compact for every A E p(A). The conditions of the theorem are therefore
necessary.
Assume now that R(A: A) is compact for A E p(A) and that T(t) is
continuous in the uniform operator topology for t > 0. It follows that (3.4)
holds and that
AR{A: A}T{t} - T{t} = A1000 e-As(T(t + s} - T(t)) ds. (3.6)
If A is real, A > w, then for every I) > °we have

IIAR(A: A}T{t} - T(t}11 ::0:; fJAe-ASIIT(t + s} - T(t}llds
+ fsooAe-ASIIT(t + s} - T(t}llds
::0:; sup II T( t + s} - T( t) II
which implies
limsupIIAR(A:A}T(t) - T(t)II::o:; sup IIT(t +s) - T(t)11
°
,\-->00
for every > (3.7)

°
{j
Since I) > is arbitrary we have

lim IIAR(A: A}T(t} - T(t}11 = 0.
A--+ 00
But AR(A: A)T(t) is compact for every A > wand therefore T(t) is
compact. o
Corollary 3.4. Let T(t) be a Co semigroup and let A be its infinitesimal

generator. If R( A : A) is compact for some "A E p( A) and T( t) is continuous in
the uniform operator topology for t > to, then T(t) is compact for t > to.
PROOF. From our assumptions it follows that R("A: A) is compact for every
A E p(A) and that (3.6) holds for every t > to. The rest of the proof is
identical to the end of the proof of Theorem 3.3. 0
Corollary 3.5. Let T( t) be a uniformly continuous semigroup (Definition

1.1.1). T( t) is a compact semigroup if and only if R( A: A) is compact for every
A E p(A).
The characterization of compact semigroups in Theorem 3.3 is not

completely satisfactory since it does not characterize the compact semi group
T(t) solely in terms of properties of its infinitesimal generator A. The reason
for this is that so far, there are no known necessary and sufficient condi-
tions, in terms of A or the resolvent R(A: A), which assure the continuity
°
for t > of T(t) in the uniform operator topology. A necessary condition
for T(t) to be continuous, in the uniform operator topology, for t > is
given next.
°
generator. If T( t) is continuous in the uniform operator topology for t > 0,
then there exists a function IjJ : [0, oo[ -+ [0, oo[ such that
p(A) ~ {A: A = a + iT, ITI ~ 1jJ(lal)}, (3.8)

and
lim
ITI-oo
IIR(a + iT:A)11 = ° for every real a. (3.9)
PROOF. We will assume without loss of generality that p(A) ~ {"A: Re"A > o}
and that IIT(t)11 :5 M. Otherwise, we consider S(t) = e~WT(t) with w
the uniform operator topology for t > °

chosen so that these conditions are satisfied. Obviously T(t) is continuous in
if and only if S( t) has this
property.
If a > ° then by our assumption A = a + iT E p(A). Substituting x =
R("A: A)y in (2.3), we obtain
eAtR("A: A)y - T(t)R(A: A)y = BA(t)y for y EX (3.10)
which implies
(eo t - M)IIR(A: A)II :5 eotll{e~iTSe~OST(s) dsll

Choosing t > a ~ I log M yields
IIR( a + iT: A) II :5 cll{e-iTSe-O'T(s) dsll (3.11 )

for some constant C independent of T. The right-hand side of (3.11) tends

to zero as ITI -+ 00 by the lemma of Riemann-Lebesgue. For a ~ 0 we
write
00
R(A: A) = L R(l + iT: A)k+l{1 + iT - A)k (3.12)

k=O
and set
cp(ITI) = max IIR{1 + it:A)II·
It I ;0, ITI
By what we have already proved above, cp(ITI) -+ 0 as ITI -+ 00. The series
(3.12) clearly converges (in the uniform operator topology) for 11 - a I ~
1/2cp(ITJ), which implies (3.8). Moreover, for any fixed a satisfying 11 - al
~ 1/2cp(ITI) we have
IIR(a + iT:A)11 ~ 211R(1 + iT:A)11 ~ 2cp(lTI)

and therefore (3.9) holds and the proof is complete. D
Corollary 3.7. Let T(t) be a compact Co semigroup and let A be its infinitesi-
mal generator. For every - 00 < a ~ f3 < 00, the intersection of the strip
a ~ Re A ~ f3 with a( A) contains at most a finite number of eigenvalues of A.
PROOF. The compactness of the semigroup T( t) implies the compactness of
R(A: A) for A E p(A) (Theorem 3.3). Therefore the spectrum of R(A: A)
consists of zero and a sequence, which may be finite or even empty, of
eigenvalues converging to zero if the sequence is infinite. This implies that
a(A) consists of a sequence of eigenvalues with 00 as the only possible limit
point. From Theorem 3.6 it follows that the intersection of a(A) and the
strip a ~ Re A ~ f3 is compact and hence can contain only a finite number
of eigenvalues of A. D
Note that in Corollary 3.7 we have proved that if T(t) is a compact Co

semigroup the spectrum a(A) of its infinitesimal generator consists solely of
eigenvalues.
2.4. Differentiability
Definition 4.1. Let T(t) be a Co semigroup on a Banach space X. The
semigroup T(t) is called differentiable for t> to if for every x E X, t -+
T(t)x is differentiable for t > to. T(t) is called differentiable if it is differen-
tiable for t > O.
We have seen in Theorem 1.2.4 (c) that if T(t) is a Co semigroup with

infinitesimal generator A and x E D(A) then t -+ T(t)x is differentiable for
t ~ O. If T( t) is moreover differentiable then for every x E X, t -+ T( t)x is
differentiable for t > O. Note that if t --+ T( t)x is differentiable for every
x E X and t ~ 0 then D(A) = X and since A is closed it is necessarily
bounded.
Example 2.1 provides a simple example of a Co semigroup which IS

differentiable for t > 1.
Lemma 4.2. Let T(t) be a Co semigroup which is differentiable for t > to and
let A be its infinitesimal generator, then
(a) For t> nto, n = 1,2, ... , T(t): X --+ D(An) and T(n)(t) = AnT(t) is a
bounded linear operator.
(b) For t > nt o, n = 1,2, ... , T(n- 1)( t) is continuous in the uniform operator
topology.
PROOF. We start with n = 1. By our assumption t - T(t)x is differentiable
for t> to and all x E X. Therefore T(t)x E D(A) and T'(t)x = AT(t)x
for every x E X and t> to. Moreover, since A is closed and T(t) is
bounded, AT(t) is closed. For t> to, AT(t) is defined on all of X and
therefore, by the closed graph theorem, it is a bounded linear operator. This
concludes the proof of (a) for n = 1. To prove (b) let II T(t)11 s MI for
o s t s I and let to < tl S t2 S tl + I then,
T(t2)X - T(tl)x = {2AT(s)xds = {2T(s - tl)AT(tl)xds (4.l)
/\ /\
and therefore
IIT(t2)x - T(tl)xlI S (t2 - tl)MtlIAT(tl)llllxll
which implies the continuity of T(t) for t > to in the uniform operator
topology.
We now proceed by induction on n. Assume that (a) and (b) are true for n
and let t > (n + I)t o. Choose s > nto such that t - s > to' Then
T(n)(t)x=T(t-s)AnT(s)x forevery xEX. (4.2)
The right-hand side of (4.2) is differentiable since t - s > to and therefore
T(t)x is (n + I)-times differentiable and T(n+ 1)(t)X = A n+ IT(t)x for every
x E X and t > (n + I)t o. This implies like in the case n = 1 that T(t): X
--+ D(An+l) and that An+IT(t) is a bounded linear operator for t > (n +
I)t o. This concludes the proof of (a). The continuity of T(n)(t) for t > (n +
I)to in the uniform operator topology is proved exactly as for the case
n = 1, using the fact that AnT(t) is bounded for t > (n + I)t o. 0
Corollary 4.3. Let T(t) be a Co semigroup which is differentiable for t > to. If
t > (n + 1)t 0 then T( t) is n-times differentiable in the uniform operator
topology.
PROOF. From part (b) of Lemma 4.2 it follows that for t> (n + l)t o,
AkT(t), 1 s k s n is continuous in the uniform operator topology. There-
fore if t > (n+ I)t o• we have

T(k-I)(t + h) - T(k-I)(t} = j
l+h
AkT(s} ds for I ~ k ~ n,
I
which implies the differentiability of T(k-I)(t) in the uniform operator

topology for I ~ k ~ nand t > (n + I)t 0 and thus T( t) is n-times differen-
tiable in the uniform operator topology. D
Corollary 4.4. If T(t) is a differentiable Co semigroup, then T(t) is differen-

tiable infinitely many times in the uniform operator topology for t > o.
Lemma 4.5. Let T(t) be a differentiable Co semigroup and let A be its

infinitesimal generator. Then
n = 1,2, .... (4.3)
PROOF. The lemma is proved by induction on n. For n = I the result has

been proved in Lemma 4.2. If (4.3) holds for nand t 2 s then
Differentiating (4.4) with respect to t we find
Substituting s = ntln + I in (4.5) yields the result for n + 1. D
We turn now to the characterization of the infinitesimal generator of a Co

semi group which is differentiable for t > to. Before turning to the main
result we will need one more preliminary.
Lemma 4.6. Let T( t) be a Co semigroup and let A be its infinitesimal

generator. If T(t) is differentiable for t > to and A E a(A), t> to then
Ae A1 E a(AT(t».
PROOF. We define
BA(t}x = [eA(I-S)T(s}xds.
o
BA(t)x is clearly differentiable in t and differentiating it we find
B;"(t}x = T(t}x + ABA(t}x.

B;'(t) is a bounded linear operator in X. Assuming now that t > to and
differentiating (2.2) with respect to t, we obtain
AeA1x - AT(t}x = (AI - A}B;,,(t}x for every x E X. (4.6)

Let
C{t)x = Aelllx - AT{t)x.
For t > to, C(t) is a bounded linear operator. It is easy to check that B~(t)
and C(t) commute and that for x E D(A), AB~(t)x = B~(t)Ax. If Ae lll E
p{AT(t» then C(t) is invertible and from (4.6) it follows that
x = (AI - A)B~{t)C{t)-IX for every x E X,
i.e., B~(t)C(t)-1 is a right inverse of AI - A. Multiplying (4.6) from the left
by C(t)-I we have
x = C{t)-I{AI - A)B~{t)x.
Choosing x E D{A) we can commute B~(t) and AI - A. Then using the
commutativity of B~ (t) and C( t) and therefore also of B~ (t) and C( t) - I,
we obtain
x = B~{t)C{t)-I{AI - A)x for every x E D{A).
Therefore, B~(t)C(t)-1 is the inverse of AI - A, A E p(A) and the result

follows. D

generator. If II T{t)11 :$ Me wl then the following two assertions are equivalent:
°
(i) There exists a to > such that T(t) is differentiable for t > to.
(ii) There exist real constants a, band C such that b > 0, C > 0,
p{A) ::> ~ = {A: Re A;;::: a - blog 11m AI} (4.7)

and
IIR{A: A)II :$ qlm AI for AE ~, Re A :$ w. (4.8)
PROOF. We may assume without loss of generality that w < 0. Otherwise we

consider the sernigroup TI(t) = e-(w+E)T(t) satisfying II T1(t)11 :$ Me- El
and for which (i) or (ii) hold if and only if they hold for T( t). We will
therefore assume that w < 0.
We start by showing that (ii) implies (i). Let f be a path in ~ composed of
°
three parts; fl given by Re A = 2a - b log ( - 1m A) for - 00 < 1m A :$
- L = _e 2a / b , f2 is given by Re A = for - L :$ 1m 'A :$ Land f3 is given
by Re A = 2a - blog(lm A) for L :$ 1m A < 00. f is oriented so that 1m A
increases along f. By changing the constant C in (4.8) we can assume that
(4.8) holds for A E fi' j = 1,3. Let fn = f n {A: 11m AI < n}. Since A ~
eAtR(A: A) is a continuous function from p{A) c C into B(X) (the space
of all bounded linear operators on X) the integrals
1
Sn(t) = -2 .1rnelllR{A: A) d'A
'TTl,
are well defined. If Sn(t) converge in B(X) as n --+ 00 we define the limit to
be the improper integral
S(/) = _1_. [eAtR(A: A) dA (4.9)

2'TT/ Jr
and say that the integral (4.9) converges in B( X), i.e., in the uniform
operator topology. Moreover, it is easy to see that Sn(t) are differentiable in
B(X) and their derivatives S~(/) are given by
S~(/) = - 12.
'TTl
jr.AeAtR(A: A) dA. (4.lO)
If Sn(t) and S~(t) converge in B(X) uniformly, say for I ~ II' as n -+ 00

then, for I ~ I I the limit S'( I) of S~ (t) is obviously the derivative of S( I) in
B(X). We will show that (4.9) converges in B(X) for I> 21b and that
S'(/) = _1_. [AeAtR(A: A) dA (4.11)

2'TT1 Jr
converges in B(X) for I> 31b. Moreover (4.9) and (4.11) converge uni-
formly in I for I ~ 21b + 8 and t ~ 31b + 8 respectively, for every 8 > O.
To prove these claims we set rj,n = rj II {A: 11m AI < n} j = 1,2,3,
S
J.n
(/)=~j
'TTl ~ .•
eAtR(A:A)dA )= 1,2,3, (4.12)
and
}=1,2,3. (4.13)
Taking n > L it is clear that r 2 • n = r 2 and S2, net) = S2(t) and thus S2(t) is
well defined for every t ~ O. To prove the convergence of the integrals
Sj,n(t),} = 1,3, we estimate their integrands on the respective paths of
integration and find for A = a + iT E rj , } = 1,3
IleAtR(A:A)1I = leAtIIIR(A:A)11 ~ e2atlTI-btqTI = Ce2atITII-bt.
(4.14)
Therefore, for n > m ~ L we have,
~ Cle2atjnld-bl dl'Tl ( 4.15)

m
where C 1 is a constant independent of t. Thus for t > 21b, Sj,n(t)j = 1,3

converge in B(X) and the convergence is uniform in t on every compact
subinterval of (2Ib, (0). This concludes the proof of the convergence of
(4.9). To prove the convergence of (4.11) we proceed similarly. First, we
note that S;(t) exists for t ~ O. Then we estimate the integrands of SjjO,
j= 1, 3 on their respective paths of integration.
(4.16)
where C2 is a constant independent of t. The convergence of S;, net),} = 1,3
for t> 31b now follows exactly as the convergence of Sj, n(t)} = 1,3 for
t > 21b. Thus S(t) exists for t > 21b and is differentiable for t > 31b. To
conclude that T(t) is differentiable for t > 31b we will now show that for
t> 21b, S(t) = T(t).
Let x E D(A 2 ). From Corollary 1.7.5 it follows that
. 1 jY+iT A
T(t)x = hm -2' e tR(A: A)x dA for every y > O.
ITI--->oo 7Tl y-iT
( 4.17)
But for x E D(A2) we have
R( ' . A) = ~ Ax R(A: A)A2X (4.18)

1\ . x A + A2 + A2 .
From (4.9) it follows that for every x E X and t > 21b
S(t)x = 1 . (eAtR(A: A)x dA.

-2 (4.19)
m Jr
Taking x E D(A2) in (4.19), using the estimates (4.18) and (4.8), we observe
that one can shift the path of integration in (4.19) from r to the line y + iT,
- 00 < T < 00. Therefore, for t> 21b and x E D(A2), T(t)x = S(t)x.
Since for t> 21b both Set) and T(t) are bounded operators and since
D(A2) is dense in X it follows that Set) = T(t) for t> 21b and conse-
quently T(t) is differentiable for t > 31b even in the uniform operator
topology. Thus (ii) implies (i).
Next we show that (i) implies (ii). If t] > to then AT(t]) is a bounded
linear operator. Set IIAT(t])11 = M(t]). From Lemma 4.6 it follows that
a(A) C {A: Ae At \ E a(AT(t]))} C {A: IAeAt\1 :::; M(t])}. (4.20)
Consequently
p ( A) :::) {A : Re A > t]] log M ( t]) - t]] log 11m AI}.
Set
~ = {A : Re A > t]] log (1 + c5) M ( t]) - t]] log 11m AI}
for some c5 > O. (4.21)
Obviously ~ C peA), which proves (4.7). To prove (4.8) substitute R(A: A)x
for x in (4.6). The result is,
AeAtR(A: A)x = AT(t)R(A: A)x + T(t)x + A[eA(t-S)T(s)x ds.

o
(4.22)
Estimating (4.22) with t = tl and A = a + iT E ~ we find

IIR(A: A)xlI ~ ITI-Ie-all(IIAT(tl)llIlR(A: A)II
+ II T(tl)II)lIxll + Ilfot1e->'ST(S)X dsll·
But for A E~, ITI-Ie-atIIIAT(tl)11 ~ (1 + 8)-1. Choosing ITI ;::: 1 and

a ~ w < 0 we find
IIR(A:A)xlI ~ 1 ; 8 [ITI-Ie(W-allIMIIXIl +11{'e->'ST(S)Xdsll]
~ (1; 8)Me(W-alll(I'TI-I + tl)lIxll

M(1 + tl)e Wl1
~ 81I AT(t l )1I ITlllxlI = ClTlllxlI·
Thus, for A E ~, Re A ~ w, IIR(A: A)II ~ ClIm AI and the proof is com-

~~ 0
From the proof of Theorem 4.7 it follows that if T(t) is a Co semigroup

satisfying (4.7) and (4.8) then T(t) is differentiable for t > to = 31b, and if
T(t) is differentiable for t > to then for every tl > to the constant b in (4.7)
can be taken as b = 1/t I. These remarks enable us to give the following
characterization of the infinitesimal generator of a differentiable semigroup.
Theorem 4.8. Let T(t) be a Co semigroup satisfying II T(t)11 ~ Me wl and let

A be its infinitesimal generator. T(t) is a differentiable semigroup if and only if
for every b > 0 there are constants ab real and Cb positive such that
p(A) ::J ~b = {A: Re A> ab - blog 11m AI}, (4.23)
and
for A E ~b' Re A ~ w. (4.24)
Our next theorem is a simple consequence of Theorem 4.7.

satisfying II T(t)1I ~ Me wl . If for some JL ;::: w
lim sup log ITIIIR(JL + iT: A)II = C < 00 (4.25)
ITI-'oo
then T( t) is differentiable for t > 3C.

PROOF. We will show that (4.25) implies condition (ii) of Theorem 4.7.
Developing the resolvent R(A: A) into a Taylor series around the point
p. + iT we obtain
00
R(A: A) = .E R(p. + iT: A)k+I(p. + iT -

A)k. (4.26)
k=O
This series converges in the uniform operator topology as long as IIR(p. +
iT: A)II Ip. + iT - AI < l. Let e > 0 be fixed and let TO be such that for
ITI > TO'
. C + e/2
IIR(p. + 1T: A)II ~ loglTI
holds. Choosing A = a + iT we see that (4.26) converges in the region

ITI > TO' la - p.1 < (C + e)-l log ITI, i.e., the resolvent exists for
(J> Co - (C + e)-llogITI, ITI > TO' (4.27)
where Co = max(p., w + (e + C- 1) log TO). Moreover, in this region
( CI
IIR A: A)II ~ log 11m AI ~ C2 •
From the remarks following Theorem 4.7 we have that T(t) is differentiable
for t> 3(C + e) and since e> 0 was arbitrary, T(t) is differentiable for
t> 3C. 0
Corollary 4.10. Let A be the infinitesimal generator of a Co semigroup T(t)

satisfying II T(t)11 ~ Me wt . If for some p. ~ w
lim sup log ITIIIR(p. + iT: A)II = 0, (4.28)
ITI-oo
then T(t) is a differentiable semigroup.
We conclude this section with some results that give the connection
between the differentiability of a Co semigroup for t > to and the behavior
of II T(t) - III as t --+ o. We already know (see Theorem l.l.2) that if
II T(t) - III --+ 0 as t ! 0 then T( t) is a differentiable semigroup. In this case
T(t) is differentiable in the uniform operator topology for t ~ 0 and its
generator A is a bounded linear operator. Our next theorem is a consider-
able generalization of this result.
Theorem 4.11. Let T(t) be a Co semigroup satisfying II T(t)1I ~ Me wt . If

there are constants C > 0 and 8e > 0 such that
IIT(t) - III ~ 2 - Ctlog(llt) for 0 < t < 8e , (4.29)
then T( t) is differentiable for t > 3MI C.
PROOF. We first prove the result for uniformly bounded semigroups, that is,
II T(t)11 ~ M. If a is real and x E D(A) then by (2.3) we have
T(t)x - eiatx = {eia<t-S)T(s)(A - iaI)x ds. (4.30)

o
This implies
IIT(t)x - eiatxll :5: tMII(A - iaI}xll.
Substituting a = ± '1TIt we obtain
From (4.29) it follows that

11(1 + T(t)xlI ~ 211xll - 11(1 - T(t))xlI ~ (Ctlog (1lt)) IIxll
and therefore,
for ° < t < 8e ,
(4.31 )
where T = ±'1Tlt.
Thus, for IT I sufficiently large A - iT! is one-to-one and has closed range.
We will show that the range of A - iT! is all of X. Since A is the
infinitesimal generator of a uniformly bounded Co semigroup T( t), it
follows from Remark 1.5.4, that for every p> 0, (A - (p + iT)I)-1 is a
bounded linear operator in X and its norm is bounded by Mp - I. Let I E X
and set
(A - (p + iT )I}xp = I.
Then Ilxpll :5: Mp-III/II and therefore,
II(A - iTI}Xpll :5: pllxpll + 11/11 :5: (M + 1)11/11. (4.32)
From (4.31) and (4.32) it follows that IIxpll is bounded as p ~ 0. Therefore,
II (A - iTI}X p - III :5: pllxpll ~ ° as p ~ 0. (4.33)
Using (4.31) again we deduce from (4.33) that xp converges to some x as
p ~ 0. Since A is closed it follows that x E D(A) and (A - iTI)x = I and
r
therefore A - iT! is onto and from (4.31) we have
II(A - iTI}-11i :5: ~ (log 1:1

l
,
which implies
lim sup log ITI II(A - iTI}-11i :5: ~

ITI ---+ 00
and the result follows from Theorem 4.9. This concludes the proof for the
case II T(t)1I :5: M. If II T(t)1I :5: Me wt , W > 0, we consider S(t) = e-wtT(t).
Then IIS(t)1I :5: M and
IIS(t) - 111:5: e-wtIIT(t) - III + Ie-wI - 11

1 1
:5: 2 - Ct log - + Ie-wI - 11:5: 2 - CIt log-,
t t
for every C> C I > 0 and 0 < t < 8c- Therefore, S(t) is differentiable for
t > 3M/C I by the first part of the proof. Since T(t) = ewtS(t), T(t) is
differentiable for t > 3M/C I and since C I < C is arbitrary, T(t) is differen-
tiable for t > 3M/C. 0
Corollary 4.12. Let T(t) be a Co semigroup satisfying 1/ T(t) - III ~ 2 -

Cllog 1/ t for 0 < t < 8c- If T( t) can be extended to a group, then its
infinitesimal generator is necessarily bounded.
PROOF. From Theorem 4.11 it follows that for t large enough, T(t) is
differentiable and therefore, by Lemma 4.2, A T(t) is bounded. Since A =
T( - t)AT(t) it follows that A is bounded as a product of two bounded
operators. 0
Corollary 4.13. Let T( t) be a Co group and let A be its infinitesimal generator.

If A is unbounded, then
lim sup 1/1 - T{t)1/ ~ 2. (4.34)

t!O
PROOF. From Corollary 4.12 it follows that for every C > 0 and a > 0
sup
OStSa
(1/ T( t) - II/ + Cllog 1.)t ~ 2. (4.35)
Letting a to in (4.35), (4.34) follows. o
2.5. Analytic Semigroups

Up to this point we dealt with semigroups whose domain was the real
nonnegative axis. We will now consider the possibility of extending the
domain of the parameter to regions in the complex plane that include the
nonnegative real axis. It is clear that in order to preserve the semigroup
structure, the domain in which the complex parameter should vary must be
an additive semigroup of complex numbers. In this section however, we will
restrict ourselves to very special complex domains, namely. angles around
the positive real axis.
Definition 5.1. Let ~ = {z: !PI < arg z < fP2' !PI < 0 < !P2} and for z E .1
let T( z) be a bounded linear operator. The family T( z), z E .1 is an ana~}'tic
semigroup in .l if
(i) z -> T( z) is analytic in ~.
(ii) T(O) = I and lim T(z)x = x for every x E X.
:-0
:E.l
(iii) T(zl + zz) = T(ZI)T(Z2) for Zl' Z2 E ~.
A semigroup T( t) will be called analytic if it is analytic in some sector Il

containing the nonnegative real axis.
Clearly, the restriction of an analytic semigroup to the real axis is a Co

semigroup. We will be interested below in the possibility of extending a
given Co semigroup to an analytic semigroup in some sector Il around the
nonnegative real axis.
Since multiplication of a Co semigroup T(t) by e W1 does not effect the
possibility or impossibility of extending it to an analytic semigroup in some
sector Il, we will restrict ourselves in many of the results of this section to
the case of uniformly bounded Co semigroups. The results for general Co
semigroups follow from the corresponding results for uniformly bounded Co
semigroups in an obvious way. For convenience we will also often assume
that 0 E P(A) where A is the infinitesimal generator of the semigroup T( t).
This again can be always achieved by multiplying the uniformly bounded
semi group T(t) bye-a for f > O.
We start the discussion by recalling Theorem 1.7.7 which claims that a
densely defined operator A in X satisfying
7T
p(A):::) ~ = {,\: larg,\1 < ~ + 8} U (O) for some 0 < 8 < 2'
(5.1 )
and
IIR('\:A)II ~M/I'\I for ,\ E ~,'\ =f 0 (5.2)
is the infinitesimal generator of a uniformly bounded Co semigroup T( t).

More is actually true. The semigroup T( t) generated by a densely defined A
satisfying (5.1) and (5.2), can be extended to an analytic semigroup in
the sector ~o = {z: I arg z I < 8} and in every closed subsector LS o' =
{z: larg zl ~ 8' < 8}, II T(z)11 is uniformly bounded, This and much more
follow from our next theorem.
Theorem 5.2. Let T(t) be a uniformly bounded Co semigroup. Let A be the

infinitesimal generator of T(t) and assume 0 E peA). The following state-
ments are equivalent:
(a) T( t) can be extended to an anazvtic semigroup in a sector ~o =
{z: Iarg z I < 8} and II T( z) II is uniformzy bounded in every' closed sub-
sector LS o" 8' < 8, of Il o'
(b) There exists a constant C such that for every a > O. T f 0
C
IIR(a+iT:A)II~~. (5.3)
(c) There exist 0 < 8 < 7T/2 and M > 0 such that
p( A) :::) ~ = { A : Iarg ,\ I < ~ + 8} U {O} ( 5 .4 )

and
M
IIR(A: A)II ::; W for A E ~, A =1= o. (5.5)
(d) T( t) is differentiable for t > 0 and there is a constant C such that
IIAT(t)1I ::; tC for t > O. (5.6)
PROOF. (a) = (b). Let 0 < 8' < 8 be such that II T(z)11 ::; C I for z E ~6' =
{z: larg zl ::; 8'}. For x E X and a> 0 we have
R(a + i'T:A)x = Iaooe-(a+iT)IT(t)Xdt. (5.7)
From the analyticity and the uniform boundedness of T(z) in ~6' it follows
that we can shift the path of integration in (5.7) from the positive real axis
to any ray pei~, 0 0, shifting the path of
integration to the ray pew and estimating the resulting integral we find
I R( + i'T: A)x II::; [Oe-p(UCOS IJ'+Tsin IJ')C1 II x II dp
(T
o
::; C111xll. ::; C Ilxll.
(T cos 5' + 'T sm 5' 'T
Similarly for 'T < 0 we shift the path of integration to the ray pe- i6 ' and
obtain IIR(a + i'T: A)II ::; -CI'T and thus (5.3) holds.
(b) = (c). Since A is by assumption the infinitesimal generator of a Co
semigroup we have IIR(A: A)II ::; MI/Re A for Re A> o. From (b) we have
for Re A> 0, IIR(A: A)II ::; CI 11m AI and therefore, IIR(A: A)II ::; CII
IAI for Re A> o. Let a > 0 and write the Taylor expansion for R(A: A)
around A = a + i'T
00
R(A:A) = L R(a + i'T:Ar+l(a + i'T - Ar. (5.8)

n=O
This series converges in B(X) for IIR(a + i'T:A)llla + i'T - AI::; k < l.
Choosing A = Re A + i'T in (5.8) and using (5.3) we see that the series
converges uniformly in B(X) for la - Re AI ::; kl'TI/C since both a> 0
and k < 1 are arbitrary it follows that p(A) contains the set of all A with
Re A ::; 0 satisfying IRe AII 11m AI < II C and in particular
p(A):::> {A: largAI ::;i+8} (5.9)
where 8 = k arctan IjC, 0 < k < l. Moreover, in this region
C 1 ';C 2 + 1 1 M
IIR(A:A)II::; l-k·"GT::; (l-k) W=W· (5.10)
Since by assumption 0 E p(A), A satisfies (c).

(c) = (d). If A satisfies (c) it follows from Theorem 1.7.7 that
T(t) = _1_. (e~tR(X: A) dX (5.11)

2.", Jr
where r is the path composed from the two rays pe iD and pe- iD , 0 o. Differentiating (5.11) with
respect to t (first just formally) yields
T'(t) = 21 . (Xe~tR(X: A) dX. (5.12)

.",Jr
But, the integral (5.12) converges in B(X) for every t > 0 since
IIT'(t)1I ~ .!.l°OMe-pcosDtdp = (~).!.. (5.13)

." 0 ."cos & t
Therefore the formal differentiation of T( t) is justified, T( t) is differentiable
for t > 0 and
IIAT(t)1I = II T'(t)11 ~ Cit for t> o. (5.14)
(d) = (a). Since T(t) is differentiable for t > 0 it follows from Lemma 4.5
that II T(n)(t)1I = II T'(t/nnl ~ II T'(t/n) II n. Using this fact together with
(5.14) and n!e n ~ nn we have
J,IIT(n)(t)1I ~ (ce)n. (5.15)

n. t
We consider now the power series
T( z) = T( t) + L
00
n=\
T(n)(t)
, (z -
n.
t r. (5.16)
This series converges uniformly in B(X) for Iz - tl ~ k(t/eC) for every

k < 1. Therefore T( z) is analytic in tl = {z: Iarg z I < arctan 1/Ce}. Since
obviously for real values of z, T(z) = T(t), T(z) extends T(t) to the sector
tl. By the analyticity of T(z) it follows that T(z) satisfies the semigroup
property and from (5.16) one sees that T(z)x -+ x as z -+ 0 in tl. Finally,
reducing the sector tl to every closed subsector liE = {z: larg zl ~
arctan(1/Ce) - E} we see that II T(z)1I is uniformly bounded in liE and the
There are several relations between the different constants that appear in
the statement of Theorem 5.2. These relations can be discovered by check-
ing carefully the details of the proof. In particular, as we have mentioned
before the statement of the theorem, the 8 in (5.4) implies the same 8 in part
(a) of the theorem. This follows easily by checking the regions of conver-
gence of the integrals (5.11) and (5.12).
In the part (d) = (a) of the theorem we saw that if IIAT(t)1I ~ Cit then
T( t) can be extended to an analytic semigroup in a sector around the
positive real axis. If the constant C is small enough the opening angle of the
sector becomes greater than 2'17 and T(t) is analytic in the whole plane
which implies in particular that A is bounded. More precisely we have
Theorem 5.3. Let T( t) be a Co semigroup which is differentiable for t > O. Let

A be the infinitesimal generator of T(t). If
limsuptIIAT(t)1I <1. (5.17)

/-+0 e
then A is a bounded operator and T( t) can be extended analytically to the
whole complex plane.
PROOF. From (5.17) it follows that
li~s~p;IIT'(;)II< lje
and therefore the series
converges in the uniform operator topology for Iz - tilt < 1 + 8 for some
8 > O. But this domain contains the origin as an interior point. Therefore
liml-+oll T(t) - III = 0 which by Theorem 1.1.2 implies that A is bounded.
But a bounded linear operator clearly generates a seroigroup T(t) which can
be extended analytically to the whole plane. 0
EXAMPLE 5.4. Let X = 12 and for every a = {an} E 12 let
(5.18)
It is clear that T(t) defined by (5.18) is a Co seroigroup on X. Its infinitesi-

mal generator A is defined on D(A) = {{an} :{na n} E 12 } and for a E D(A),
A{a n} = {-nan}'
Also,
1 1
IIAT(t)1I = - sup (nte- nl ) =-
I n Ie
and therefore
limsuptllAT(t)ll = 1..
/-+0 e
Since A is unbounded, this example shows that the constant 1/e in Theorem
5.3 is the best possible one.
The characterization of the infinitesimal generator A of a Co seroigroup

involves usually only estimates of R(A: A) for real values of A (see e.g.,
Theorems 1.3.1 and 1.5.2) while in the characterization of the infinitesimal

generator of an analytic semigroup we used, in Theorem 5.2 estimates of
R(A: A) for complex values of A. Our next theorem gives a characterization
of the infinitesimal generator of an analytic semigroup in terms of estimates
of R(A: A) for only real values of A.

satisfying II T(t)11 :::;; MeW'. Then T(t) is analytic if and only if there are
constants C > 0 and A ;;::: 0 such that
for A> nA n = 1,2, .... (5.19)
PROOF. We note first that from Theorem 5.2 it foHows easily that T(t) is
analytic if and only if it is differentiable for t > 0 and there are constants
C I > 0 and WI > 0 such that
C
IIAT(t)1I :::;; ---;-e w1 ' for t> O. (5.20)
If A satisfies (5.19) then for A> nA and x E D(A) we have

'\ )n+1 C
liAR ( 1\: A xii = IIR ('\1\: A )n+1 Axil :::;; nAn IIxll· (5.21 )
Choosing t < 1/A and substituting A = n/t in (5.21) we find
for x E D(A).
Letting n --+ 00 it follows from Theorem 1.8.3 and the closedness of A that
IIAT(t)xlI :::;; tC II xII for x E D(A), 0 < t < I/A. (5.22)
Since D( A) is dense in X and AT( t) is closed, it follows that (5.22) holds for
every x E X. Therefore there are constants C I > 0 and WI > 0 such that
(5.20) holds and T(t) is analytic.
For the converse, we differentiate the formula
R(A: A)x = 1o e-"A'T(t)x dt
00
n times with respect to A and find

R(A: A)(n)x = (-I)nn!R(A: Ar+ IX = (-Ir 1o tne-"A'T(t)x. dt.
00
(5.23)
Operating with A on both sides of (5.23) and estimating the right-hand side
using (5.20) yields
n! IIAR(A: Ar+lxll:::;; C I (10 00

tn-Ie-("A-wl)' dt) IIxll = (A ~~I)n (n- I)! IIxll
and therefore, for A> nA
The characterizations of analytic semigroups given so far in this section

are based on conditions on the infinitesimal generator A of the semigroup or
on conditions on the resolvent R(A: A) of A. A different type of characteri-
zation of analytic semigroups based on the behavior of T( t) near its spectral
radius is the subject of our next theorem.
Theorem 5.6. For a uniformly bounded Co semigroup T(t) the following

conditions are equivalent:
(a) T(t) is analytic in a sector around the nonnegative real axis.
(b) For every complex r, r =1= I, Irl ;: : I there exist positive constants 8 and K
r
such that E p(T(t» and
for 0 < t < 8. (5.23)

(c) There exist a complex number r, Irl = I, and positive constants K and 8
such that
I
II (rl - T(t))xlI ;::: K IIxll for every x E X,O < t < 8. (5.24)
PROOF. (a) = (b). Let T(t) be analytic in a sector around the nonnegative
real axis. From Theorems 5.2 and 1.7.7, it follows that for t > 0
1
T{t) = -2
7Tl
.1 elltR{A: A) dA
f'
where f' is a path composed from two rays p ei{\ 7T /2 < {} I < 7T, P ;::: I and
pea;" -7T < {}2 < -7T /2, p ;::: I and a curve p = p( {}) joining e iif \ to e iif ,
inside the resolvent set. f' is oriented in a way that 1m A increases along f'.
Changing variables to z = At we obtain
1 .1e Z (zl- tA)-ldz.

T{t) =-2 (5.25)
7Tl f
Given r +- 1, Irl ;: : I, the path of integration f can be chosen to be

independent of t, for 0 < t < 8 and such that e Z =1= for all z on and to the r
left of f. Having chosen such 8 and f,we define
B{t) = 1 .1e Z (e Z
-2 - n-1(zl - tA)-1 dz 0<t<8. (5.26)
7Tl f
This integral converges in the uniform operator topology and thus defines a
bounded linear operator B (t). Since on f we have II (zl - tA) - I II s
2 Spectral Properties and Regulari ty 67
MI Iz I - I, it follows that
IIB(t)11 ~ ~; Irlz-le=(e Z - n-Illdzl ~ M". (5.27)
Since e Z and e Z ( e Z - n
-I are analytic on and to the left of r and tend
rapidly to zero as Re z -- - 00 and since e Z • e Z ( e Z - I = eZ + n-
~e:(eZ - n-
I , we obtain from the Dunford-Taylor operator calculus that
T( t ) B ( t) = T( t) + ~ B ( t ),
which implies
(I - B(t»)(O - T(t» = (0 - T(t»(I - B(t» = 0. (5.28)
Therefore,
(0 - T( t » - I = ~ - I (I - B (t »
and
(5.29)
(b) = (c) is trivial.
(c) = (a). Substituting A = - ia with a real into (2.3) we have
e- II "T(t)x - x = [e-ifaT(s)(A - iaJ)xds for x E D(A).
o
(5.30)
Since II T(t)11 ~ M, (5.30) implies
II(T(t) - ell"I) xii ~ Mtll(A - iaJ)xlI. (5.31)
If e ± II" = e iD = ~, ~ > 0, then by assumption (c) we have
K-Illxll ~ II(T(t) - e±i"lI)xll ~ Mtll(A ± iaJ)xlI

~ M~lal-III(A ± iaJ)xll, (5.32)
which implies
(5.33)
and therefore, A ± iaI is one to one and has closed range. We will now
show that it is onto. Since A is the infinitesimal generator of a uniformly
bounded Co sernigroup, (A - (e ± ia)l)-I exists and II(A - (e ±
ia)l)-III ~ Me-I. Forevery/E X let
(A -(e±ia)J)x,=I. (5.34)
Then Ilx,11 ~ Me-III/II and ellx,11 ~ MII/II. This together with (5.34) imply
II(A ± ial)x,11 ~ (M + 1)11/11. From (5.33) we then deduce that Ilx,11 ~ C
and therefore,
II (A ± iaJ)x, - III ~ ellx,11 -- 0 as e! O. (~.. 35)

Combining (5.35) with (5.33), we see that x f is a Cauchy net as e -+ 0 and

therefore x f -+ x. From the closedness of A it then follows that x E D(A)
and (A ± iaI)x = f. Thus the range of (A ± iaI) is all of X and (5.33)
implies
MK~
II(A ± iaI)
-1
II ~ Tal (5.36)
which implies that T( t) can be extended analytically to a sector /). around

the nonnegative real axis by Theorem 5.2 (b). 0
Corollary 5.7. Let T(t) be a Co semigroup. If

lim sup III - T(t}1I < 2 (5.37)
1.1.0
then T(t) is analytic in a sector around the nonnegative real axis.
PROOF. From (5.37) it follows that there exist 8 > 0 and e > 0 such that
111- T(t}1I ~2-e for 0<t<8. (5.38)
But then
II( -I - T(t)}xll ~ 2I1xll-ll(I - T(t)}xll ~ Ellxll for 0 < t < 8.
This implies by Theorem 5.6 (c) with r = - I that T(t) is analytic. 0
Corollary 5.7 shows that a certain behavior of III - T(t)1I at t = 0 can be

translated into the analyticity of T(t) in an angle around the nonnegative
real axis. It is natural to ask whether the analyticity of T(t) implies (5.37).
In general the answer to this is negative. Under certain restricting assump-
tions however, it is positive. A simple result in this direction is,
Corollary 5.8. If T(t) is an analytic semigroup of contractions on a uniformly

convex Banach space X, then
lim sup 111- T(t}1I < 2. (5.39)
1.1.0
PROOF. Since T(t) is a semigroup of contractions III - T(OIi :;:; 2. If
lim sup II I - T( t } II = 2
1.1.0
then there exist sequences Xn and tn such that IIxnll = I, tn -+ 0 and
11(1 - T(tn})xnll ~ 2 - lin. (5.40)
Since II T(tn)xnll ~ I it follows from (5.40) and the uniform convexity of X
that
as n -+ 00. (5.4l)
But this contradicts Theorem 5.6 (b) and therefore (5.39) must hold. 0
2.6. Fractional Powers of Closed Operators

In this section we define fractional powers of certain unbounded linear
operators and study some of their properties. We concentrate mainly on
fractional powers of operators A for which - A is the infinitesimal generator
of an analytic semigroup. The results of this section will be used in the study
of solutions of semilinear initial value problems.
For our definition we will make the following assumption.
Assumption 6.1. Let A be a densely defined closed linear operator for which
p(A):::) ~+= {A: 0 < Co) < largAI 5 'IT} U V (6.1)
where V is a neighborhood of zero, and
M
IIR(A: A)II 5 1 + IAI for AE ~+. (6.2)
If M = 1 and Co) = 'IT/2 then - A is the infinitesimal generator of a Co

semigroup. If Co) < 'IT/2 then, by Theorem 5.2, - A is the infinitesimal
generator of an analytic semigroup. The assumption that 0 E p(A) and
therefore a whole neighborhood V of zero is in p(A) was made mainly for
convenience. Most of the results on fractional powers that we will obtain in
this section remain true even if 0 $ p(A).
For an operator A satisfying Assumption 6.1 and 0: > 0 we define
A-a = _l_.jz-a(A - zI)-1 dz (6.3)

2m c
where the path C runs in the resolvent set of A from ooe-il} to ooeil} ,
Co)< f} < 'IT, avoiding the negative real axis and the origin and z-a is taken
to be positive for real positive values of z. The integral (6.3) converges in the
uniform operator topology for every 0: > 0 and thus defines a bounded
linear operator A-a. If 0: = n the integrand is analytic in ~+ and it is easy
to check that the path of integration C can be transformed to a small circle
around the origin. Then using the residue theorem it follows that the
integral equals A -" and thus for positive integral values of 0: the definition
(6.3) coincides with the classical definition of (A -I)".
For 0 < 0: < I we can deform the path of integration C into the upper
and lower sides of the negative real axis and obtain
0<0:<1. (6.4)
If Co) < 'IT/2, i.e., if - A is the infinitesimal generator of an analytic

semigroup T(t) we obtain still another representation of A -a. This represen-
tation turns out to be very useful and therefore in the rest of this section we
will assume, unless we state explicitly otherwise, that w < 'IT /2. In this case
since by Assumption 6.1 0 E p(A) there exists a constant 8 > 0 such that
- A + 8 is still an infinitesimal generator of an analytic semigroup. This
implies the following estimates;
II T( t)ll :5; Me- B1 (6.5)
IIAT(t)1I :5; M1r1e- B1 (6.6)
IIAmT(t)1I :5; Mmt-me- B1 . (6.7)
The two first estimates are simple consequences of the results of Section 5
while (6.7) follows from (6.6) since
:5; (Mlr1e-B1/m)m = Mmt-me- B1 .

Furthermore, we know that
(tI + A)-l = 1000e-SIT(s) ds (6.8)
converges uniformly for t ~ 0 in the uniform operator topology, by (6.5).

Substituting (6.8) into (6.4) and using Fubini's theorem, we have
A -a = sin;a 10 00
r a (looo e-sIT(s) dS) dt
= sin'IT'lTa 10 00
T(s )(10 00
t-ae- SI dt) ds
Since
we finally obtain
A-a = -1-
r(a)
1 ta-1T(t) dt
0
00
(6.9)
where the integral converges in the uniform operator topology for every
a > O. In the case where w < 'IT /2, i.e., - A is the infinitesimal generator of
an analytic semigroup T(t) we will use (6.9) as the definition of A -a for
a > 0 and we define A - 0 = I.
Lemma 6.2. For a, f3 ~ 0

(6.10)
PROOF.
Lemma 6.3. There exists a constant C such that

for 0 s a s 1. (6.11)
PROOF. For a = 0 and a = 1, (6.11) is obvious. For 0 < a < 1 we use (6.4)
to obtain
IIA -ails [[ Sin;a ft-a(tI + Ar1dt [[ + [[ Sin;a ()t-1-at(tI + A)-l dt II.

Let II(tI + A)-III s Co for 0 s t s 1. From Assumption 6.1 we have
Ilt(tI + A)-III s C I for t ~ 1 and therefore
IIA-all sISin'IT(1-a)lc +ISin'ITal c s C. 0

'IT(1 - a) 0 'ITa I
Lemma 6.4. For every x E X we have

lim A-ax = x. (6.12)
a---+O
PROOF. Assume first that x E D(A). Since 0 E p(A) we have x = A -Iy for
some y E X. Therefore,
Using (6.5) we therefore have

a
IIA-ax - xII s Clfok[ f(l t + a) - 1[e- 8l dt + C2~ te- 8l dt (6.13)
00
for every k > 0 and 0 s a s 1. Given e > 0 we first choose k so large that
the second term on the right of (6.13) is less than e/2 and then choose a so
small that the first term on the right of (6.13) is less than e/2. Thus for
x E D(A) we have A-ax -+ x as a -+ O. Since D(A) is dense in X and by
Lemma 6.3, A -a are uniformly bounded (6.12) follows for every x E X. 0
Combining the previous results we have
Corollary 6.5. If A satisfies Assumption 6.1 with w < 7T /2 then A - t is a Co

semigroup of bounded linear operators.
Lemma 6.6. A -a defined by (6.9) is one-to-one.

PROOF. It is clear that A - I is one-to-one. Therefore. for every integer n ;? I.
A-"isone-to-one.LetA-ax = O.Taken;? athenA-nx =A-n+aA-"x = O.
This implies x = 0 and thus A -a is one-to-one. 0
Definition 6.7. Let A satisfy Assumption 6.1 with w < 7T /2. For every a > 0
we define
(6.14)
For a = 0, A" = I.
In the rest of this section we assume that A satisfies Assumption 6.1 with
W < 7T /2 and collect some simple properties of A" in our next theorem.
Theorem 6.S. Let A" be defined by Definition 6.7 then,

(a) A" is a closed operator with domain D(A") = R(A -") = the range of
A -".
(b) a ;? /3
> 0 implies D(A") c D(AP).
(c) D(Aa) = X for every a ;? O.
(d) If a, /3 are real then
(6.15)
for every x E D(AY) where y = max(a, /3, a + /3).
PROOF. For a s 0, A" is bounded and (a) is clear. If a > 0, A" is invertible
and therefore 0 E p(A"). This implies that A" is closed. For a ;? /3 we have
by Lemma 6.2, A -" = A -fJ . A -{"-fJ) and therefore R(A -") c R(A -fJ)
which implies (b). Since by Theorem 1.2.7, D(A") = X for every n = 1,2, ...
and for as n D(A"):::> D(An) by (b), we have (c). Finally, (d) is again a
simple consequence of the definition of A" and Lemma 6.2, for example if
a > 0 and /3 > 0 and x E D(A"AfJ) then x E D(AfJ) and AfJx E D(A").
Lety = A"APx thenAPx = A-"y and x = A-PA-"y = A-("+fJ)y. Therefore
x E D(Aa+fJ) and A"+fJx = y = A"APx . Similarly if x E D(A"+P) we have
x E D(A"AfJ) and A"+fJ = A" . AfJ. 0
In Definition 6.7 we defineAa in an indirect way. For x E D(A) c D(A"),

o< a < 1, we have the following explicit formula for A"x.
Theorem 6.9. Let 0 < a < 1. If x E D(A) then

sin -
A"x = - 7Ta 1 ta-1A(tI + A)
00 - I
x dt. (6.16)
7T 0
PROOF. We have 0 < 1 - a < 1. Therefore by (6.4) we have
A",-I X = - 7Ta
sin - 1 00
t",-I(tI + A) _ I
xdt. (6.17)
7T 0
The integrand on the right-hand side of (6.17) is in D(A) for every t > 0 and
t",-IA(tI + A)-IX is integrable on [0, oo[ since near t = 0 IIA(tI + A -1)11 is
uniformly bounded and near t = 00 1it"'-IA(tI + A)-IXIl ~ t a- 2MIIAxll.
Finally if x E D(Aa), Aa-Ix E D(A) and from the closedness of A we
deduce
Aax = A(Aa-lx) = - 7Ta

sin - 1 00
t"'-~(tI + A) - I
xdt. D
7T 0
Remark. From the proof of Theorem 6.9 it is clear that (6.16) holds for
every x E D(AY) with y > a.
Theorem 6.10. Let 0 < a < 1. There exists a constant Co> 0 such that for
every x E D(A) and p > 0, we have
IIAax11 ~ Co(pallxll + pa-IIIAxll) (6.18)
and
(6.19)
PROOF. By our assumptions on A there exists a constant M satisfying
II(tI + A)-III ~ Mit for every t > O. If x E D(A) then by Theorem 6.9,
IIA"'xll ~I sin7T7Ta Ifopta-IIIA(tI + A)-lllllxlidt
+ I sin7T7Ta I~OOta-lll(tI + A)-IIIIIAxlidt
~ 1si::a 1(1 + M)pallxll + ISi:(l(~ ~) IMpa-IIIAxll
~ Co(p"'lIxll + pa-IIIAxll)·
For x = 0, (6.19) is obvious. For x 1= 0, (6.19) follows from (6.18) taking
p = IIAxll1 IIxli. D
Corollary 6.11. Let B be a closed linear operator satisfying D(B) ~ D(Aa),

o< a ~ 1 then
IIBxll ~ qAaxll for every x E D(Aa) (6.20)
and there is a constant C I such that for every p > 0 and x E D(A)
IIBxll ~ Cl(pallxll + p",-IIIAxll). ( 6.21)
PROOF. Consider the closed operator BA -a. Since D(B) ~ D(Aa), BA -a is

defined on all of X and by the closed graph theorem it is bounded. This
proves (6.20). For x E D(A), (6.21) is a direct consequence of (6.20) and

(6.18). D
A sufficient condition for D(B) :J D(Aa) is given in our next theorem.
Theorem 6.12. Let B be a closed linear operator satisfying D(B) :J D(A). If

for some y, 0 < y < 1, and every P ~ Po > 0 we have
IIBxll .::; C(pYllxll + py-IIIAxll) for x E D(A) (6.22)
then
D(B) :J D(Aa) for every y < a .::; 1. (6.23)
PROOF. Let x E D(AI-a) then A -ax E D(A) c D(B). Since B is closed
BA -ax = _1_100
r(a)
ta-IBT( t)x dt
0
provided that the integral is convergent. But
IIBA-axll .::; r/a) (Ia t"-IIIBT(t)X ll dt + ~oota-IIIBT(t)Xlldt).

8
(6.24)
Since T(t) is an analytic semigroup T(t)x E D(A) for every t > O. Choos-
ing 8 = Po I and using (6.22) with p = t- I in the first integral on the
right-hand side of (6.24) and with p = Po in the second integral and making
use of (6.5) and (6.6) we find IIBA-axll.::; CIIxll. This is true for all
x E D(A I- a ). Since BA- a is closed and D(A I-,,) is dense in X, IIBA-axll
.::; Cllxll for every x E X and therefore D(B) :J D(A"). D
It can be shown that if A satisfies the Assumption 6.1 without any

restriction on w, then - Aa with a .::; 1/2 is the infinitesimal generator of a
Co semigroup of bounded linear operators. If w < 7T/2 as we assume, - Aa
is the infinitesimal generator of an analytic semigroup for all 0 < a .::; 1.
We conclude this section with some results relating A a and the analytic
semigroup T(t) generated by - A.
Theorem 6.13. Let - A be the infinitesimal generator of an analytic semigroup

T(t). If 0 E p(A) then,
(a) T(t): X --+ D(Aa) for every t > 0 and a ~ O.
(b) For every x E D(Aa) we have T(t)Aax = A"T(t)x.
(c) For every t > 0 the operator AaT(t) is bounded and
IIA aT(t)1I .::; M"t-ae- IJ1 • (6.25)
(d) Let 0 < a .::; 1 and x E D(Aa) then
IIT(t)x - xii.::; C"t"IIAaxll· (6.26)
PROOF. Our assumptions on A imply that it satisfies Assumption 6.1 with

W < '!T12 and therefore we have the existence of Aa for a ~ O. Since T(t) is
analytic we have T(t): X -+ n ~_oD(An) C D(Aa) for every a ~ 0 which
proves (a).
Let x E D(Aa) then x = A -ay for some y E X and
T(t)x = T(t)A-ay = f(a)

1 1 sa-IT(s)T(t)yds
0
00
= A-aT(t)y = A-aT(t)Aax
and (b) follows.
Since A a is closed so is AaT(t). By part (a) AaT(t) is everywhere defined
and therefore by the closed graph theorem AaT(t) is bounded. Let n - 1 <
a ~ n then using (6.7) we have
IIAaT(t)11 = IIAa-nAnT(t)11 ~ f(n 1_ a) 1a00sn-a-IIIAnT(t + s)llds
< Mn 1°Osn-a-l(t+s)-ne-li(t+S)ds
- f(n - a) 0
Me-lit 00 M
~f( n )a1 un-a-I(I+u)-ndu=~e-lit.
n- a t o t
Finally,
II T(t)x - xii = II{AT(S)X dsll = II{AI-aT(S )Aax dsll
~ C [sa-lIIAaxllds = CutallAaxll. o
o
CHAPTER 3
Perturbations and Approximations
3.1. Perturbations by Bounded Linear Operators

Theorem 1.1. Let X be a Banach space and let A be the infinitesimal
generator of a Co semigroup T(t) on X, satisfying II T(t)11 :s: Me W1 • If B is a
bounded linear operator on X then A + B is the infinitesimal generator of a Co
semigroup S(t) on X, satisfying IIS(t)11 :s: Me(w+MIIBII)I.
PROOF. From Lemma 1.5.1 and Theorem 1.5.3 it follows that there exists a
norml·lonXsuchthatllxll:s: Ixl :s:MllxllforeveryxEX,IT(t)1 :s:e w1
and IR(A: A)I :s: (A - W)-I for real A satisfying A > w. Thus, for A> W +
IBI the bounded operator BR(A: A) satisfies IBR(A: A)I < 1 and therefore
1- BR(A: A) is invertible for A > W + IBI. Set
00
R = R(A: A)(I - BR(A: A))-l = L R(A: A)[BR(A: A)]k (1.1)

k~O
then
(AI - A - B)R = (I - BR(A: A))-I
-BR(A: A)(I - BR(A: A))-l = I
and
R· (AI - A - B)x = R(A: A)(AI - A - B)x
00
+ L R(A: A)[BR(A: A)] k(AI - A - B)x

k~l
00
= x - R ( A: A) Bx + L [R (A : A) B] kX
k~l
00
- L [R(A:A)B]kX = x
k=2
3 Perturbations and Approximations 77
for every x E D(A). Therefore, the resolvent of A + B exists for A > W +

IBI and it is given by the operator R. Moreover,
I(AI-A -B)-II =lk~oR(A:A)[BR(A:A)]kl

~ (A - w) - 1(1 - IBR (A : A) I) - I ~ (A - W - IB I) - I.
From Corollary 1.3.8 it follows that A + B is the infinitesimal generator of a
Co semigroup Set), satisfying IS(t)1 ~ e(w+ IBI)'. Returning to the original
norm II II on X we have,
IIS(t)1I ~ Me(w+MIIBII>'. 0
We are now interested in the relations between the semigroup T(t)

generated by A and the semigroup S( t) generated by A + B. To this end we
consider the operator H(s) = T(t - s)S(s). For x E D(A) = D(A + B),
s -+ H(s)x is differentiable and H'(s)x = T(t - s)BS(s)x. Integrating
H'(s)x from 0 to t yields
S(t)x = T(t)x + 1o'T(t - s)BS(s)xds for x E D(A}. (1.2)
Since the operators on both sides of (1.2) are bounded. (1.2) holds for every
x E X. The semigroup S(t) is therefore a solution of the integral equation
(1.2). For such integral equations we have:
Proposition 1.2. Let T(t) be a Co semigroup satisfying II T(t)1I ~ MeW'. Let

B be a bounded operator on X. Then there exists a unique family V( t), t ~ 0 of
bounded operators on X such that t -+ V(t)x is continuous on [0. oo[ for every
x E X and
V(t)x = T(t}x + [T(I - s)BV(s)xds for x E X. (1.3)

o
PROOF. Set
( 1.4)
and define Vn(t) inductively by
v" + I ( I ) x = 10'T( t - s} B v" (s ) x ds for x E X, n ~ O. (1.5)
From this definition it is obvious that I --+ Vn ( t)x is continuous for x E X,

I ~0 and every n ~ O. Next we prove by induction that,
( 1.6)
Indeed, for n = 0, (1.6) holds by our assumptions on T(/) and the definition
of Vo(t). Assume (1.6) holds for n then by (1.5) we have

t Mn/lBllnsn
11v,,+l(t)xll~ (Mew(t-S)/iB/IMe ws I /lx/lds
Jo n.
Mn+l/1B/ln+ltn+l
= Me wt /lx/l
(n+l)!
and thus (1.6) holds for n > O. Defining
00
V( t) = L v" (t ), (1.7)
n=O
it follows from (1.6) that the series (1.7) converges uniformly in the uniform
operator topology on bounded intervals. Therefore t --+ V(t)x is continuous
for every x E X and moreover by (1.4) and (1.5) it follows that for every
x E X, V(t)x satisfies the equation (l.3). This concludes the proof of the
existence statement. To prove the uniqueness let U( t), t ~ 0 be a family of
bounded operators for which t --+ U(t)x is continuous for every x E X and
U(t)x = T(t)x + fotT(t - s)BU(s)xds for x E X. (1.8)
Subtracting (1.8) from (1.2) and estimating the difference yields
II(V(t) - u(t))xll ~ {MeW(I-S)/lBIIII(V(s) - U(s))x II ds. (1.9)
But (1.9) implies, for example by Gronwall's inequality, that II(V(t)-

U(t»xlI = 0 for every t ~ 0 and therefore Vet) = U(t). D
From Proposition 1.2 and the fact that the semigroup S(t) generated by
A + B satisfies the integral equation (1.2) we immediately obtain the
following explicit representation of Set) in terms of T(t):
(1.10)
where So(t) = T(t),
Sn+1(t)x = tT(t - s)BSn(s)xds XEX (1.11)

o
and the convergence in (LlO) is in the uniform operator topology.
For the difference between T(t) and S(t) we have:

satisfying II T(t)/I ~ Me w1 • Let B be a bounded operator and let S(t) be the Co
semigroup generated by A + B. Then
(1.12)
PROOF. From the integral equation (l.2) and Theorem 1.1 we have
IIS(t)x - T(t)xll S; [IIT(t - s)IIIIBIIIIS(s)llllxllds

o
S; M 2e wt liBII [eMIlBllsllxllds
o
= Mewt(eMIIBllt - 1)lIxll· D
The main result of this section, Theorem 1.1, shows that the addition of a
bounded linear operator B to an infinitesimal generator A of a Co semi-
group, does not destroy this property of A. It is natural to ask which special
properties of the semigroup T( t) generated by A are preserved when A is
perturbed by a bounded operator B. It is not difficult to show that if A is the
infinitesimal generator of a compact or analytic semigroup so is A + B. We
will prove here the statement about compact semigroups. The statement
about analytic semigroups will follow from the results of the next section
(see Corollary 2.2).
Proposition 1.4. Let A be the infinitesimal generator of a compact Co semi-

group T(t). Let B be a bounded operator, then A + B is the infinitesimal
generator of a compact Co semigroup S( t).
PROOF. From Theorem 2.3.3 it follows that T(t) is continuous in the
uniform operator topology for t > 0 and that R("A: A) is compact for
"A E p(A). Since
00
R("A:A+B)= LR("A:A)[BR("A:A)]k (1.13)

k~O
and IIR("A: A)II S; M("A - W)-I for "A> w, it follows that for "A> w +
MIIBII + I, (1.13) converges in B(X) and since each one of the terms on
the right-hand side of (1.13) is compact so is R("A: A + B) for "A > w +
MIIBII + l. From the resolvent identity it follows that R("A: A + B) is
compact for every "A E p(A + B). To show that S(t) is a compact semi-
group it is therefore sufficient, by Theorem 2.3.3., to show that S(t) is
continuous in the uniform operator topology for t > O. To show this we
note that if T( t) is continuous in the uniform operator topology for t > 0
then each one of the operators Sn(t) defined by (1.11) is continuous in the
uniform operator topology for t> O. Since S(t) is the uniform limit (on
bounded t-sets) in the uniform operator topology of L'J~o5.J(t) it follows
that S(t) is continuous in the uniform operator topology for t > o. D
Not all the properties of the semigroup T( t) are preserved by a bounded

perturbation of its infinitesimal generator. For example it is known that if A
is the infinitesimal generator of a semigroup T(t) which is continuous in the
uniform operator topology for t ~ to > 0, or is differentiable for t ~ to > 0
or is compact for t ~ to > 0 then S(t), the semigroup generated by A + B

where B is a bounded operator need not have the corresponding property.
3.2. Perturbations of Infinitesimal Generators of

Analytic Semigroups
Theorem 2.1. Let A be the infinitesimal generator of an analytic semigroup.
Let B be a closed linear operator satisfying D(B) ::J D(A) and
IIBxll :::; allAxl1 + bllxll for x E D(A). (2.1)
There exists a positive number 8 such that if 0 :::; a :::; 8 then A + B is the
infinitesimal generator of an analytic semigroup.
PROOF. Assume first that the semigroup T( t) generated by A is uniformly
bounded. Then p(A)::J ~ = {A: larg AI :::; 'TT/2 + w} for some w > 0 and
in ~, IIR(A:A)II:::; MIAI- 1• Consider the bounded operator BR(A:A).
From (2.1) it follows that for every x E X
IIBR(A: A)xlI :::; aIIAR(A: A)xlI + bIIR(A: A)xlI
bM
:::; a(M + 1)lIxll + !XI Ilxll· (2.2)
Choosing 8 = ~(1 + M)-l and IAI > 2bM we have IIBR(A: A)II < 1 and
therefore the operator 1- BR(A: A) is invertible. A simple computation
shows that
(AI - (A + B))-l = R(A: A)(I - BR(A: A))-l. (2.3)
Thus for IAI > 2bM and Iarg AI :::; 'TT/2 + w we obtain from (2.3) that
IIR(A:A + B)II :::; M'IAI- 1 (2.4)
which implies that A + B is the infinitesimal generator of an analytic
semigroup.
If T(t) is not uniformly bounded, let II T(t)11 :::; Me w' . Consider the
semigroup e-W'T(t) generated by Ao = A - wI. From (2.1) we have
IIBxl1 :::; allAoxl1 + (aw + b)lIxll for x E D(A).
Therefore, by the first part of the proof if 0 :::; a :::; 8, Ao + B = A + B - wI
is the infinitesimal generator of an analytic semigroup which implies that
A + B is also the infinitesimal generator of an analytic semigroup. 0
Remark. In Theorem 2.l, the semigroup S(t) generated by A + B satisfies

IIS(t)11 :::; Me(w+A(b»1 where limb~oA(b) = o.
From the case a = 0 in Theorem 2.1 we obtain,
Corollary 2.2. Let A be the infinitesimal generator of an analytic semigroup. If

B is a bounded linear operator then A + B is the infinitesimal generator of an
analytic semigroup.
From the proof of Theorem 2.1 one deduces easily the following corollary.
Corollary 2.3. Let A be the infinitesimal generator of a uniformly bounded

analytic semigroup. Let B be a closed operator satisfying D(B) :::) D(A) and
IIBxll s allAxl1 for x E D(A). (2.5)
Then there exists a positive constant 8 such that for 0 s a s 8, A + B is the
infinitesimal generator of a uniformly bounded analytic semigroup.
Corollary 2.4. Let A be the infinitesimal generator of an analytic semigroup.
Let B be closed and suppose that for some 0 < a < I, D(B) :::) D(Aa) then
A + B is the .infinitesimal generator of an analytic semigroup.
PROOF. Since D(B):::) D(Aa) we obviously have D(B):::) D(A). From
Corollary 2.6.11 it follows that
IIBxll s C(pallxll + pa-tIIAxll) for x E D(A) and p > o.
(2.6)
Choosing p > 0 so large that Cpa-t < 8 where 8 is the constant given in the
statement of Theorem 2.1, the result follows readily from Theorem 2.1. D
3.3. Perturbations of Infinitesimal Generators of

Contraction Semigroups
We start with a definition.
Definition 3.1. A dissipative operator A for which R(l- A) = X, is called

m-dissipative.
If A is dissipative so is /LA for all /L > 0 and therefore if A is m-dissipative

then RCA.I - A) = X for every A > O. In terms of m-dissipative operators
the Lumer-Phillips theorem can be restated as: A densely defined operator
A is the infinitesimal generator of a Co semigroup of contractions if and only
if it is m-dissipative.
The main result of this section is the following perturbation theorem for
m-dissipative operators.
Theorem 3.2. Let A and B be linear operators in X such that D(B) :::) D(A)
and A + tB is dissipative for 0 S t S 1. If
IIBxl1 s allAxl1 + ,Bllxll for x E D(A) (3.1)
where 0 ~ a < 1, P ~ 0 and for some to E [0,1], A + toB is m-dissipative

then A + tB is m-dissipative for all t E [0, 1].
PROOF. We will show that there is a ~ > 0 such that if A + toB is m-dis-
sipative, A + tB is m-dissipative for all t E [0,1] satisfying It - tol ~ ~.
Since every point in [0, 1] can be reached from every other point by a finite
number of steps of length ~ or less this implies the result.
Assume that for some to E [0, 1] A + toB is m-dissipative. Then I -
(A + toB) is invertible. Denoting (I - (A + toB»-1 by R(to) we have
IIR(to)1I ~ 1. We show now that the operator BR(to) is a bounded linear
operator. From (3.1) and the triangle inequality we have for x E D(A)
IIBxll ~ all(A + toB)xlI + atollBxll + Pllxll
~ all(A + toB)xlI + allBxll + Pllxll
and therefore
IIBxl1 ~ 1 ~ a II(A + toB)xlI + 1 ~ a IIxll· (3.2)
Since R(to): X --+ D(A) and (A + toB)R(to) = R(to) - I it follows from

(3.2) that
IIBR(to)xll ~ 1 ~ a II(R(to) - I}xll + 1 ~ a IIR(to)xlI

for all x E X (3.3)
and so BR{to) is bounded. To show that A + tB is m-dissipative we will

show that I - (A + tB) is invertible and thus its range is all of X. We have
I - (A + tB) = I - (A + toB) + (to - t)B
= {I + (to - t)BR(to))(I - (A + toB)). (3.4)
Therefore I - (A + tB) is invertible if and only if I + (to - t)BR{to) is
invertible. But I + (to - t)BR{to) is invertible for all t satisfying It - tol <
(1 - a)(2a + P)-l ~ IIBR(t0)1I- 1 and we can therefore choose ~ =
(1 - a)(4a + 2P)-1 to conclude the proof. 0
Theorem 3.2 is usually used through the following simple corollary.

contractions. Let B be dissipative and satisfy D(B) :l D(A) and
IIBxll ~ allAxll + Pllxll for x E D(A) (3.5)
where 0 ~ a < 1 and P ~ O. Then A + B is the infinitesimal generator of a Co
semigroup of contractions.
PROOF. By Lumer-Phillips' theorem (Theorem 1.43), D(A) = X and A is
m-dissipative. Therefore A + tB is dissipative for every 0 ~ t ~1. This
°
follows from the fact that if A is m-dissipative Re (Ax, x*) ::s; for every
x* E F(x). Indeed, if B is dissipative with D(B) ::J D(A), then for every
x E D(A) there is an x* E F(x) such that Re (Bx, x*) ::s; and for this°
same x*, Re (Ax + tBx, x*) ::s; 0. From Theorem 3.2 it follows that A + tB
is m-dissipative for all t E [0, 1] and in particular A + B is m-dissipative.
Since D(A + B) = D(A) is dense in X, A + B is the infinitesimal generator
of a Co sernigroup by Lumer-Phillips' theorem. 0
Note that Corollary 3.3 can be stated in a slightly more symmetric form
as follows: If A + tB is dissipative for t E [0,1], D(B) ::J D(A), D{A) = X
and (3.5) holds then either both A and A + Bare m-dissipative or neither A
nor A + Bare m-dissipative.
Theorem 3.2 and Corollary 3.3 do not hold in general if a < 1 in (3.1) is
replaced by a = 1. One of the reasons for this is that in this case it is no
more true that A + B is necessarily closed. If A + B is not closed it cannot
be the infinitesimal generator of a Co sernigroup. A simple example of this
kind of situation is provided by a self adjoint operator iA in a Hilbert space.
If iA is self adjoint both A and -A are infinitesimal generators of Co
semigroups of contractions (see Theorem 1.10.8). Taking B = -A in Theo-
rem 3.2 we have the estimate (3.1) with a = 1 and f3 = 0, but A + B
restricted to D(A) is not closed. In this simple example however, the closure
of A + B, i.e., the zero operator on the whole space, is the infinitesimal
generator of a Co sernigroup of contractions. Our next theorem shows that
under a certain additional assumption this is always the case.
Theorem 3.4. Let A be the infinitesimal generator of a Co semigroup of

contractions. Let B be dissipative such that D(B) ::J D(A) and
IIBxl1 ::s; IIAxl1 + f3llxll for x E D{A) (3.6)

°
where f3 :?: is a constant. If B*, the adjoint of B, is densely defined then the
closure A + B of A + B is the infinitesimal generator of a Co semigroup of
contractions.
PROOF. A + B is dissipative and densely defined since A is m-dissipative
and B is dissipative with D(B)::J D(A). Therefore, by Theorem 1.4.5,
A + B is closable and its closure A + B is dissipative. To prove that A + B
is the infinitesimal generator of a Co semigroup of contractions it is
therefore sufficient to show that R(I - (A + B)) = X. Since A + B is
dissipative and closed, it follows from Theorem 1.4.2. that R(I - (A + B))
is closed and therefore it suffices to show that R(I - (A + B)) is dense in X.
°
Let y* E X* be "orthogonal" to the range of I - (A + B), that is,
(y*, z) = for every z E R(I - (A + B)). Lety E Xbe such that Ily*11 ::s;
°
(y*, y). From Corollary 3.3 it follows that A + tB is m-dissipative for
::s; t < 1 and therefore the equation
(3.7)
has a unique solution x t for every 0 ~ t < 1. Moreover, since A + tB is

dissipative Ilxtll ~ lIyll. From (3.6) it follows that
IIBxtll ~ IIAxtll + .Bllxtll ~ II(A + tB)xtll + tllBxtl1 + .Bllxtll
~ IIY - xtll + tllBxtl1 + .Bllxtll
and therefore,
(1 - t)IIBxtll ~ IIY - xtll + .Bllxtll ~ (2 + Mllyll. (3.8)
Let z* E D(B*) then
l(z*,(I - t)Bxt)l= (I - t)I(B*z*,xt)1

~ (I - t)IIB*z*11 Ilyll --+ 0 as t --+ 1. (3.9)
Since D(B*) is dense in X* and since by (3.8) (1 - t)Bx t is uniformly
bounded it follows from (3.9) that (l - t)Bx t tends weakly to zero as t --+ 1.
In particular by the choice of y* we have
IIY*II ~ (y*, y) = (y*, x t - AX t - tBxt)
= (y*,(1 - t)Bx t ) --+ 0 as t --+ 1

which implies y* = 0 and the range of I - (A + B) is dense in X. 0
Let X be a reflexive Banach space and let T be a closable densely defined

operator in X. Then it is well known that T* is closed and D(T*) is dense in
X* (see Lemma 1.10.5). Therefore, for reflexive Banach spaces we have:
Corollary 3.5. Let X be a refleXive Banach space and let A be the infinitesimal
generator of a Co semigroup of contractions in X. Let B be dissipative such that
D(B) ::) D(A) and
IIBxl1 ~ IIAxll + .BllxlI for x E D(A)
where .B ~ O. Then A + B, the closure of A + B, is the infinitesimal generator

of a Co semigroup of contractions in X.
3.4. The Trotter Approximation Theorem

In this section we study, roughly speaking, the continuous dependence of a
semigroup T(t) on its infinitesimal generator A and the continuous depen-
dence of A on T(t). We show that the convergence (in an appropriate sense)
of a sequence of infinitesimal generators is equivalent to the convergence of
the corresponding semigroups. We start with a lemma.
Lemma 4.1. Let A and B be the infinitesimal generators of Co semigroups T( t)

and Set) respectively. For every x E X and lI. E peA) () p(B) we have
R(lI.: B)[T(t) - S(t)]R(lI.: A)x
= fS(t - s )[R(lI.: A) - R(lI.: B)] T(s)x ds. (4.1)

o
PROOF. For every x E X and lI. E peA) () p(B) the X valued function
s ~ S(t - s)R(lI.: B)T(s)R(lI.: A)x is differentiable. A simple computa-
tion yields
d
ds [S(t - s )R(lI.: B)T(s)R(lI.: A)x]
= S(t - s)[ -BR(lI.: B)T(s) + R(lI.: B)T(s)A]R(lI.: A)x

= S(t - s)[R(lI.: A) - R(lI.: B)]T(s)x
where we have used the fact that R(A: A)T(s)x = T(s)R(lI.: A)x. Integrat-
ing the last equation from 0 to t yields (4.1). 0
In the sequel we will use the notation A E G (M, w) for an operator A

which is the infinitesimal generator of a Co semigroup T(t) satisfying
II T(t)11 :$; Me w1 •
Theorem 4.2. Let A, An E G(M, w) and let T(t) and Tn(t) be the semigrnups
generated by A and An respectively then the following are equivalent:
(a) For every x E X and A with Re lI. > w. R(A: An)x ~ R(A: A)x as
n --> 00.
(b) For every x E X and t 2 0, Tn(t)x --> T(t)x as n --> 00.
Moreover, the convergence in part (b) is uniform on bounded t-intervaL~.
PROOF. We start by showing that (a) = (b). Fix x E X and an interval

o :$; t :$; T and consider
II(Tn(t) - T(t))R(A: A)xll:$; IITn(t)(R(lI.: A) - R(lI.: An))xll

+ IIR(lI.: An)(Tn(t) - T(t))xll
+ II(R(lI.: An) - R(A: A))T(t)xll
(4.2)
Since II Tn(t)11 :$; Me wT for O:$; t :$; T it follows from (a) that D, --> 0 as
n --> 00 uniformly on [0, T]. Also, since t --> T( t)x is continuous the set
{T(t)x: 0 :$; t :$; T} is compact in X and therefore D) --> 0 as n --> 00
uniformly on [0, T]. Finally, using Lemma 4.1 with B = An we have

IIR{A: An)(Tn{t) - T{t))R{A: A)xll
:::; {II Tn{t - s) IIII(R{A: A) - R(A: An))T{s )xll ds
o
:::; tllTn(t -s)IIII(R(A:A) -R(A:An))T(s)xlldx. (4.3)
o
The integrand on the right hand side of (4.3) is bounded by 2M3 e2wt(Re A
- w)-'lIxll and it tends to zero as n ~ 00. By Lebesgue's bounded
convergence theorem the right hand side of (4.3) tends to zero and therefore
lim IIR{A: An)(T,,{t) - T{t))R{A: A)xll = 0 (4.4)
n~oo
and the limit in (4.4) is uniform on [0, T]. Since every x E D(A) can be
written as x = R(A: A)z for some z E X it follows from (4.4) that for
x E D(A), D2 ~ 0 as n ~ 00 uniformly on [0, T]. From (4.2) it then
follows that for x E D( A 2 )
lim II(Tn(t) - T{t))xll = 0 (4.5)
n~oo
and the limit in (4.5) is uniform on [0, T]. Since II T,,(t) - T(t)11 are
uniformly bounded on [0, T] and since D(A2) is dense in X (see Theorem
1.2.7) it follows that (4.5) holds for every x E X uniformly on [0, T] and
(a) = (b).
Assume now that (b) holds and Re A > w then
IIR{A: An)x - R{A: A)xll :::; 1o e-ReAtll(Tn{t) -

00
T{t))xll dt. (4.6)
The right-hand side of (4.6) tends to zero as n ~ 00 by (b) and Lebesgue's

dominated convergence theorem and therefore (b) = (a). 0
Remark. From the proof of Theorem 4.2 it is clear that a weaker version of
(a) namely, for all x E X and some AO with Re AO > w, R(AO: An)x ~
R(Ao: A)x as n ~ 00, still implies (b).
We say that a sequence of operators An' r-converges to an operator A if
for some complex A, R(A: An)x ~ R(A: A)x for all x E X. In Theorem 4.2
we assumed the existence of the r-limit A of the sequence An and further-
more assumed that A E G( M, w). It turns out that these assumptions are
unnecessary. This will follow from our next theorem.
Theorem 4.3. Let An E G( M, w). If there exists a Ao with Re AO > w such

that
(a) for every x E X, R(Ao: An)x ~ R(Ao)X as n ~ 00 and
(b) the range of R(AO) is dense in X,
then there exists a unique operator A E G (M, w) such that R (A 0) =
R(Ao: A).
PROOF. We will assume without loss of generality that w = 0 and start by

proving that R(A: An)x converges as n -+ 00 for every A with Re A> O.
Indeed, let S = {A: Re A> 0, R(A: An)x converges as n -+ oo}. S is open.
To see this expand R(A: An) in a Taylor series around a point p. at which
R(p.: An)x converges as n -+ 00. Then
00
R(A:An) = L (IL -A)kR(IL:An)k+l. ( 4.7)

k=O
Since by Remark 1.5.4 IIR(p.: An)kll :;;; M(Rep.)-k, the series (4.7) con-
verges in the uniform operator topology for all A satisfying Ip. - AI(Re p.) - I
< 1. The convergence is uniform in A for A satisfying Ip. - AI(Re p. ) - I :;;; {f
< 1 and the series of constants r:r:=oM{fk+ I is majorant to the series
Lk=olp. - AlkIIR(p.: An)k+lli. This implies the convergence of R(A: An)x
as n -+ 00 for all A satisfying Ip. - AI(Re p.) - I :;;; {f < 1, and the set S is
open as claimed. Let A be a cluster point of S with Re A > O. Given
0< {f < 1 there exists a point p. E S such that Ip. - AI(Rep.)-1 :;;; {f < 1
and therefore by the first part of the proof R(A: An)x converges as n -+ 00,
i.e., A E S. Thus S is relatively closed in Re A > O. Since by assumption
AO E S we conclude that S = {A: Re A> O}.
For every A with Re A > 0 we define a linear operator R( A) by
R(A)X = lim R(A: An)x. (4.8)
n--- 00
Clearly,
R(A) - R(p.) = (p. - A)R(A)R(p.) for Re A > 0 and Re p. > 0
(4.9)
and therefore R(A) is a pseudo resolvent on Re A> 0 (see Definition 1.9.1).
Since for a pseudo resolvent the range of R(A) is independent of A (see
Lemma 1.9.2.) we have by (b) that the range of R(A) is dense in X. Also,
from the definition of R(A) it is clear that
IIR(A)kll :;;;M(ReA)-k for ReA>O,k= 1,2, .... (4.10)
In particular for real A, A > 0
IIAR(A)II :;;; M for all A > O. (4.11 )
It follows from Theorem 1.9.4 that there exists a unique closed densely
defined linear operator A for which R(A) = R(A: A). Finally, from (4.10)
and Theorem 1.5.2 it follows that A E G( M, 0) and the proof is complete. D
A direct consequence of Theorems 4.2 and 4.3 is the following theorem.
Theorem 4.4 (Trotter-Kato). Let An E G(M, w) and let Tn(t) be the semi-
group whose infinitesimal generator is An. If for some AO with Re AO > W we
have:
(a) As n -+ 00, R(AO: An)x -+ R(AO)X for all x E X and
(b) the range of R(AO) is dense in X,
then there exists a unique operator A E G (M, w) such that R (A 0) = R (A 0 :

A). If T(t) is the Co semigroup generated by A then as n -+ 00, Tn(t)x ->
T( t)x for all t ~ 0 and x E X. The limit is uniform in t for t in bounded
intervals.
A somewhat different consequence of the previous results is the following

theorem.
Theorem 4.5. Let An E G(M, w) and assume

(a) As n -+ 00, Anx -> Ax for every xED where D is a dense subset of X.
(b) There exists a Ao with Re AO > W for which (A01 - A)D is dense in X,
then the closure X of A is in G(M, w). If Tn(t) and T(t) are the Co
semigroups generated by An and X respectively then
lim Tn ( t ) x = T( t }X for all t ~ 0, X E X (4.12)
and the limit in (4.12) is uniform in t for t in bounded intervals.

PROOF. Let y E D, x = (A01 - A)y and Xn = (A01 - An)Y' Since AnY ->
Ay, xn -> x as n -+ 00. Also since IIR(AO:An)11 ::;: M(ReAo - W)-I it
follows that
lim R(Ao: An}x = lim (R(Ao: AJ(x - x n ) + y) = y (4.13)
n-oo n-oo
i.e., R(Ao: An) converges on the range of A01 - A. But by (b) this range is
dense in X and by our assumptions IIR(Ao: An)1I are uniformly bounded.
Therefore R(Ao: An)x converges for every x E X. Let
lim R( AO : An}x = R( Ao}X. (4.14)
n ..... 00
From (4.13) it follows that the range of R(AO) contains D and is therefore
dense in X. Theorem 4.3 implies the existence of an operator A'E G(M, w)
satisfying R(Ao) = R(Ao: A'). To conclude the proof we show that X = A'.
Let xED then
lim R(Ao: An){A01 - A}x = R(Ao: A'}(AoI - A}x. (4.15)
n-oo
On the other hand as n -> 00
R(AO: An){AoI - A)x = R(Ao: An){A01 - An}x

+R(Ao: An){An - A}x
= x + R(Ao: An){An - A}x -> x,
since IIR(Ao: An)1I are uniformly bounded and for xED, Anx -> Ax.
Therefore
R(AO: A'}(A01 - A}x = x for xED. ( 4.16)
But (4.16) implies A'X = Ax for xED and therefore A' ::> A. Since A' is
closed, A is closable. Next we show that A:J A'. Let y' = A'x'. Since
(Ao! - A)D is dense in X there exists a sequence xn E D such that
Yn = (Ao! - A')x n = (Ao! - A)x n -+ Aox' - y'
= (Ao! - A')x' as n -+ 00. (4.17)
Therefore,
xn = R(Ao: A'}Yn -+ R(Ao: A')(A o! - A')x' = x' as n -+ 00 (4.18)
and
as n -+ 00. (4.19)
From (4.18) and (4.19) it follows that y' = Ax' and A:J A'. Thus A = A'.
The rest of the assertions of the theorem follow now directly from Theorem
« 0
3.5. A General Representation Theorem

Using the results of the previous section we will obtain in this section a
representation theorem which generalizes considerably the results of Section
1.8. We start with a preliminary estimate.
Lemma 5.1. Let T be a bounded linear operator satisfying

k = 1,2, ... (5.1)
where N ~ 1. Then for every n ~ 0 we have
Ile(T-l)n x - Tnxll :s; MNn-1e(N-l)n [n 2 (N - If + nN r/211x - Txll.
(5.2)
PROOF. Let k, n ~ 0 be integers. If k ~ n then
k-l
IITkX - Tnxll = L (Tj+lX - Tjx)
j=n
k-l
:s; Mllx - Txll L Nj:s; (k - n)MNk-1Ilx - Txll
j=n
(5.3)
From the symmetry of the estimate (5.3) with respect to k and n it is clear
that (5.3) holds also for n > k. For k = n we have equality and therefore
(5.3) is valid for all integers k, n ~ O. Now
lIet(T-l)x - Tnxll =lle- t f

k=ok.
t~ (TkX - T nx)ll::5; e- t f
k=ok.
t~ IITkx - Tnxll
00 (tN)k
::5; MNn-111x - TXlle-tk~o ~Ik - nl· (5.4)
Using the Cauchy-Schwartz inequality we have
I {t:t In - kl ,; L~o {t:t rl~o {t~?, k)'r {n -
= etN(n _ Nt)2 + Nt)1/2. (5.5)

Combining (5.4) and (5.5) we obtain
lIet(T-l)x - Tnxll ::5; MNn-1e(N-l)t[(n - Nt)2 + Ntr/2l1x - Txll.
(5.6)
Substituting t = n in (5.6) we get (5.2). o
Remark. The function et(T-l)x is the solution of the differential equation
du/dt = (T - I)u with u(O) = x. The elements Px are the polygonal
approximations with steps of length 1 of the solution of this equation, i.e.,
the solution of the difference equations
u(j + 1) - u(j) = (T - I)u(j), u(O) = x.
Corollary 5.2. If T is nonexpansive on X, i.e., II Til ::5; 1, then for every n ~ 0

we have
lIe(T-I)n X - Tnxll ::5; ..;n IIx - Txll. (5.7)
We now turn to the representation theorem.
Theorem 5.3. Let F(p), p ~ 0 be a family of bounded linear operators

satisfying
k = 1,2, ... (5.8)
for some constants w ~ 0 and M ~ 1. Let D be a dense subset of X and let
limp-I(F(p)x - x) = Ax for xED. (5.9)
p ..... O
If for some Ao with Re Ao > w, (AoI - A)D is dense in X then A is closable

and A~ the closure of A, satisfies A E G(M, w). Moreover, if T(t) is the Co
semigroup generated by A then for every sequence of positive integers k n ~ 00
satisfying knPn ~ t we have
lim F(Pn)k nx = T(t)x for x E X. (5.10)
n ..... 00
Choosing Pnkn = t for every n, the limit in (5.10) is uniform on bounded t

intervals.
PROOF. For P > 0 consider the bounded operators Ap = p-I(F(p) - /).
These operators are the infinitesimal generators of uniformly continuous
semigroups Sp (t) satisfying:
IISp(t)11 ~ e- t/ p
k=O
f (~)k IIF~)kll ~ Mexp {~(eWp -
p. P
I)}.
Let e > 0 be such that Re AD > W + e and let Po > 0 be such that for
o< P ~ Po, (e WP - I)p - I < W + e. Then
for 0 <P ~ Po.
From Theorem 4.5 it follows that A is closable and that X E G(M, w + e).
If T(t) is the semigroup generated by X then Theorem 4.5 implies further
that
as p ~ 0 (5.11)
uniformly on bounded t intervals.
On the other hand, it follows from Lemma 5.1 that
IISpJPnkJx - F(Pn)knxll ~ Mexp {wPn(k n - 1) + (e wPn - I)k n }
. [k;(e WPn _ 1)2 + knewPnr/2. Pnll F(Pn~~ - x II.
Choosing xED, Pn ~ 0, k n ~ 00 such that Pnkn ~ t it is obvious that

Pnkn' (e WPn - I)k n and Pn-IIIF(Pn)x - xII stay bounded as n ~ 00. There-
fore we have
as n ~ 00. (5.12)
If Pn = tjkn one can choose the constant C independent of t for 0 ~ t ~ T,

which implies, in this case, uniform convergence on bounded intervals in
(5.12). For xED we have
IIT(t)x - F(pJk nx11 ~ IIT(t)x - Spn(t)xll + IISpn(t)x - Sp,,{knPn)xll
+ IISp,,(knpn)x - F(Pn)k"xll = 11 + 12 + 13 •
From (5.11) and (5.12) it follows that II ~ 0 and 13 ~ 0 as n ~ 00. To
show that 12 ~ 0 as n ~ 00 we observe that for xED, 0 ~ t ~ T we have
for large values of n
IISp,,(t)x - Sp,,(PnkJxll ~ Me(w+elTlt - Pnkn'll F(Pn~~ - x II ~ 0
as n ~ 00.
If Pn = tjk n then 12 == O. This concludes the proof of (5.10) for xED.

92 Scmigroups of Linear Operators
Since D is dense in X and IIT(t) - F(Pnlnll are uniformly bounded (5.10)

holds for every x E X.
Finally, the semi group T(t) generated by X satisfies IIT(t)11 ~ Me(W+f)1
for every small enough e > 0 and therefore it also satisfies II T(t)11 ~ Me w1
and X E G(M, w). D
Corollary 5.4. Let F( p), p ;::: 0 be a family of bounded linear operators

sati:,jying
k = 1,2, ... , (5.13 )
for some constants w ;::: 0 and M ;::: I. Let A be the infinitesimal generator of a
Co semigroup T(t). If
p'-I(F(p)x - x) --+ Ax as p --+ 0 for every xED ( A ) (5.14)
then,
T( t) x = lim
Ii-OO
F(!..)
n
nX for x EX (5.15 )
and the limit is uniform on bounded t-intervals.

PROOF. Since A is the infinitesimal generator of a Co semi group it is closed
and for every real A. large enough the range of A. I - A is all of X. Therefore,
our result follows readily from Theorem 5.3. D
As a simple consequence of Corollary 5.4, we can prove the exponential

formula
T(t)x = lim
II-OC
(I - !"A)-n
n
X for x E X (5.16)
where T( t) is a Co semi group and A is its infinitesimal generator. This

formula has already been proved in Theorem 1.8.3 by a different method.
To prove (5.16) assume that A E G( M, w) and set F( p) = (/ - pA) - I =
(ljp)R(ljp: A), for 0 < p < Ijw. From Theorem 1.5.3. it follows that
IIF(p)"l1 ~ M(I - pw)-n ~ Me 2wpfl
for p small enough. Also from Lemma 1.3.2 it follows that if x E D( A) then
~(F(P) - 1)x = A( ~R( ~ : A)X) --+ Ax.
Therefore F(p) = (I - pAr I satisfies the conditions of Corollary 5.4 and

(5.16) is then a direct consequence of this corollary.
Corollary 5.5. Let Ai E G( Mi , w), j = 1,2, ... , k and let S)(t) be the semi-
group generated by Aj' Let n J_tD(A) be dense in X and
n = 1,2, ... (5.17)

for some constants M ~ 1 and w ~ O. If for some A with Re A > w the range
of AI - (AI + A z + ... + A k ) is dense in X then Al + A z + ... + Ak E
G(M, w). If S(t) is the semigroup generated by Al + A z + ... + Ak we have
for x E X (5.18)
and the limit is uniform on bounded t intervals.
PROOF. Set F(t) = SI(t)SZ(t) ... Sk(t) and n{=ls;(t) = SI(t)SZ(t) ...
~(t). For x E n :=ID(A) and t -+ 0 we have
F(t)x - x j~1 OJ S;(t)) Sit); - x -+ (AI + A z + ... + Ak)X.
(5.19)
The result follows now directly from Theorem 5.3. o
Corollary 5.5 is an abstract version of the method of fractional steps

which is used in solving partial differential equations. The idea behind this
method is the following; in order to solve the initial value problem
du
dt = (A I + A z + . . . + A k ) u, u(O) = x, (5.20)
we have only to solve the k simpler problems
j=1,2, ... ,k (5.21 )
and obtain the solution of (5.20) by combining the solutions of (5.21)

according to (5.18).
The method of "alternating directions" is also a special case of this
general abstract result.
We conclude this section with the analogue of Corollary 5.5 for the
backwards difference approximations of (5.20).
Corollary 5.6. Let Aj E G(~, w), j = 1,2, ... , k. If Al + A z + ... + Ak

E G(M, w) and
n = 1, 2, ... ( 5 .22 )
then the semigroup S(t) generated by Al + A z + ... + Ak is given by
S(t)x= }~[(I-~Alrl(I-~Azrl ... (I-~AkrlrX

for every x E X. (5.23)
PROOF. Set F(t) = (/ - tA1)-I(/ - tA 2 )-1 ... (/ - tAk)-1 = n~=I(/

tAi)-I. For x E n ;=ID(A) = D(AI + ... + Ad and t ~ 0 we have
F() k(j-l )(I-tA)-I X - x

t x - x = j~1 }] (I - tA i ) - I jt
(5.24)
Here we used that if A E G(M, w) then (/ - tA)-ly ~ Y as t ~ 0 for every
y E X and rl«(/ - tA)-I X - x) ~ Ax as t ~ 0 for x E D(A). The result
follows now directly from Corollary 5.4. 0
3.6. Approximation by Discrete Semigroups

In this section we show by means of an example how the results of the
previous sections can be applied to obtain solutions of initial value prob-
lems for partial differential equations via difference equations.
The results that we present here are rather special in the sense that
stronger results of similar nature may be obtained even under somewhat
weaker assumptions. Since our goal here is only to demonstrate the method
we preferred to make some superfluous assumptions (e.g., part (iv) of
Assumption 6.1 below) in order to avoid some of the technicalities.
Let X and Xn be Banach spaces with norms II . II and II lin respectively.
We shall make the following assumption.
Assumption 6.1. For every n ~ 1 there exist bounded linear operators Pn : X

~ Xn and En : Xn ~ X such that
(i) IIPnl1 .:S N, II En II .:S N', Nand N' independent ofn.

(ii) IIPnxli n ~ IIxll as n ~ 00 for every x E X.
(iii) IIEnPnx - xII ~ 0 as n ~ 00 for every x E X.
(iv) PnEn = In where In is the identity operator on X n.
EXAMPLE 6.2. Let X = BU([ - 00,00]) be the space of all bounded uni-
formly continuous real valued functions defined on R I. Let Xn = b be the
space of all bounded real sequences {c n }::"= -00' Both spaces BU([ - 00, 00])
and b are normed with the usual supremum norm. We define Pnf(x) =
{f(kln)}'k=-oo' Then Pn is obviously linear and IIPnll .:S 1. From the
definitions of the norms and the uniform continuity of the elements of X it
is also clear that (ii) is satisfied. Taking for En the linear operator which
assigns to a sequence {Ck}'k= -00 the functionf(x) which is equal to ck at the
point x = kin and is linear between any two consecutive points jln and
j + ljn we obtain IIEnll.:S 1. Obviously PnEn = In and (iii) follows
from the uniform continuity of the elements of X and the definitions of En
and Pn •
Definition 6.3. A sequence xn E Xn converges to x E X if

IIPnx - xnll n --+ 0 as n --+ 00. (6.1 )
This type of convergence will be denoted, without danger of confusion, by

Xn --+ x or limn->ooxn = x.
Note that the limit of such a convergent sequence is unique. Indeed, if
xn --+ x and xn --+ y then
Ilx - yll = lim IIPn(x - y} lin
n-> 00
~ lim
n~oo
IIPnx - xnlln + n--+oo
lim Ilxn - Pnylln = O.
Definition 6.4. A sequence of linear operators An' An: Xn --+ Xn converges

to an operator A, A : X --+ X if
and
Ax = lim AnPnx for x E D(A}. (6.3)
n-> 00
We will denote this type of convergence by An --+ ~ A.

Note that An --+ --+ A means that for every x E D(A)
as n--+oo. (6.4)
Lemma 6.5. Let {X~}k= 1 be a Cauchy sequence in Xn. If for every fixed
k, x~ ~ xk E X as n --+ 00 (in the sense of Definition 6.3) and x~ --+ Xn E Xn
uniformly with respect to n as k --+ 00 then the double limit exists and
lim Xn = lim xk = x E X. (6.5)
n-+oo k--+oo
PROOF. We first prove that Xk is a convergent sequence in X. We have
IIPnx k - Pnx/lin ~ IIPnx k - x~lln + Ilx~ - x~lln + Ilx~ - Pnx/lln·

Given e > 0 we choose k, I ~ K(e) such that Ilx~ - x~lln < e/6 for all n.
Then we choose no = no(k, I) so large that IIPnx k - x~lIn < e/6, IIPnx ' -
x~lln < e/6 and IIXk - xiii ~ IIPn(x k - x/)lln + e/2 for all n ~ no' So for
k, I ~ K(e) we have, by choosing n ~ no(k, I), Ilxk - xiii < e and there-
fore xk --+ x as k --+ 00. Next, we show that limn->ooxn = x. Indeed,
IlPnx - xnll n ~ IIPnx - Pnxklln + IIPnx k - x~lIn + Ilx~ - xnll n

~ Nllx - xkll + IIPnx k - x~lln + Ilx~ - xnll n·
Given e > 0 we first choose and fix k so large that Nllx - xkll < e/3 and
Ilx~ - xnll n < e/3 for all n. Then we choose no(k) so large that IIPnx k -
x~lln < e/3 for all n > no' Thus IIPnx - xnll n < e for n > no and Xn --+ x.
D
Lemma 6.6. Let An be a sequence of bounded linear operators An: Xn ~ Xn·

If IIAnll ::;;; M, An ~ ~ A and D(A) is dense in X, then D(A) = X and
IIAII ::;;;M.
PROOF. Let x E X. Since D(A) is dense in X there is a sequence Xk E D(A)
such that Xk ~ X. Now,
implies AnPnxk ~ AnPnx as k ~ 00 uniformly in n. Moreover, since Xk E

D(A) AnPnxk ~ AXk as n ~ 00 for each k. Applying Lemma 6.5 to the
sequence x! = AnPnxk it follows that AnPnx converges as n ~ 00 which
implies x E D(A). Thus D(A) = X. Finally,
IIAxll = lim IIPnAxli n = lim IIAnPnxlln::;;; M lim IIPnxlin = Mllxll

n~oo n~oo n ...... oo

Theorem 6.7. Let F(Pn) be a sequence of bounded linear operators from Xn
into Xn satisfying
k = 1,2, ... (6.7)
and
(6.8)
If D(A) is dense in X and if there is a Ao with Re Ao > '" such that the range
of AoI - A is dense in X then A, the closure of A, is the infinitesimal generator
of a Co semigroup S(t) on X. Moreover, if knPn ~ t as n ~ 00 then
(6.9)
where D(S(t» = x.
PROOF. An is a bounded linear operator on Xn and therefore generates a
uniformly continuous semigroup Sn(t) on Xn. It is easy to check (see, e.g.,
the proof of Theorem 5.3) that
(6.10)
where "'n = Pn-l(e"'Pn - 1). Given E > 0, "'n ::;;; '" + E for all Pn small enough.
Set An = EnAnPn' An is a bounded linear operator on X and therefore
generates a semigroup Sn(t) on X. Using Assumption 6.1, (iv), we have
tk _ tk k ( t k k) _
L L L
_ 00 00 00
Sn(t) = k! A~ = k! EnAnPn = En k! An Pn - EnSn(t)Pn·
k-O k=O k=O
(6.11)
Therefore, for n large enough we have
IISn(t)1I ::;;; MNN'e(",+e)' = M1e(",+e)t. (6.l2)
If X E D(A) then as n --+ 00
II.Anx - Axil = IIEnAnPnx - Axil

::5: IIEnAnPnx - EnPnAx11 + IIEnPnAx - Axil
::5: N'IIAnPnx - PnAxlln + IIEnPnAx - Axil --+ 0 (6.13)
where the first term on the right of (6.13) tends to zero as n --+ 00 since
x E D(A) and An --+ --+ A, and the second term tends to zero as n --+ 00 by
Assumption 6.1 (iii). From Theorem 4.5 it follows that A E G(M1 , W + d.
Since e > 0 was arbitrary we actually have A E G(M1 , w), that is, A is the
infinitesimal generator of a Co semigroup S(t) satisfying IIS(t)11 ::5: M1e wt •
From Theorem 4.5 we also have
as n --+ 00. (6.14)
Therefore,
IISn(t)Pnx - PnS(t)xlln = IIPnSn(t)x - PnS(t)xlln
::5: NIISn(t)x - S(t)xlI --+ 0 as n --+ 00.
( 6.15)
From Lemma 5.1 we have a constant C such that
IISn(Pnkn)Pnx - F(Pn)knPnxlin
::5: CpY211 F(P~ - I PnXIIn = Cp~/2I1AnPnxlin' (6.16)
The estimate (6.16) follows from Lemma 5.1 similarly to the way (5.12)
follows from this lemma. Choosing x E D( A) we have
Finally,
IIF(Pn)k npnx - PnS(t)xlln ::5:IIF(Pn)k npnx - Sn(Pnkn)Pnxlln

+ IISn(Pnkn)Pnx - Sn(t)Pnxlin + IISn(t)Pnx - PnS(t)xlln
::5: (Cpy2 + IPnkn - tl)IIAnPnxlin + IISn(t)Pnx - PnS(t)xlln.
(6.18)
Combining (6.15), (6.17) and (6.18) and letting Pn --+ 0 such that Pnkn --+ t
we obtain for every x E D( A),
as n --+ 00. (6.19)
Since II F( Pn)kn II are uniformly bounded, (6.19) holds for every x E X, i.e.,
F(Pn)k n --+ --+ S(t). 0
Remark. From the proof of Theorem 6.7 it follows that if k n = [tIPn] the
convergence of F(Pn)k n to S(t) is uniform on bounded t intervals.
We now turn to a concrete example. Let X = BU([ - 00, 00]) be the space
of all bounded uniformly continuous real valued functions on IR I, and
consider the following initial value problem for the classical heat equation
au a2 u
{ at = ax 2 for - 00 < x < 00, t > 0
(6.20)
u(O,x} =f(x} for -00 < x < 00
with f E X. We intend to prove the existence of a solution u(t, x) of

(6.20). Furthermore, we will also obtain a numerical approximation of the
solution. This will be done by replacing the differential equation in (6.20)
by a difference equation.
In order to reduce the differential equation to a difference equation we
consider for each given n and Tn functions defined on the lattice (kin, ITn),
k = 0, ± I, ±2, ... , 1= 0,1,2, ... in the (x, t) plane. We set u(kln, ITn) =
Uk .1' A reasonable difference equation that will correspond to the differen-
tial equation in (6.20) is
Tn-l(u k . l+I - Uk . l } = n 2( Uk +1./ - 2u k . 1 + u k -l.I}· (6.21 )
Rearranging (6.21) we have
Uk . I + 1 = (I - 2n 2Tn )U k . 1 + n2Tn(Uk+I.1
+ Uk - I. I ). (6.22)
Thus if u k • O = fk are given we can compute all Uk. I by the recursion formula
(6.22).
In order to use our previous results we consider the Banach space Xn = b
(i.e., the space of all bounded real sequences {c n }::"= _ 00 with the supremum
norm) and define operators Pn and En as in Example 6.2. We then define an
operator F( Tn) mapping Xn into Xn as follows
F( Tn){ Uk. I} = {U k • l +I} (6.23)
where {Uk. 1+ I} is obtained from {uk,I} by (6.22). Set an = 2n2Tn and choose
Tn such that an < I. Then
IIF( Tn){ Uk./} II = SUp lu k • l + II

k
::; (1 - an) sup IUk./1 + an sup IUk./1 = sup IUk./I·

k k k
(6.24)
Therefore II F( Tn)1I ::; 1 and the stability condition (6.7) of Theorem 6.7
holds with w = 0 and M = 1. Let
It is clear that D is dense in X. For fED we have

II 'Tn- I (F( 'Tn) - I) Pnf - Pnf" II n
= s~pln2(f( k: 1) _2f(~) + f( k: 1)) - f"(~)I.

(6.25)
Since fED, f"( x) is uniformly continuous on ~ I and therefore the right-
hand side of (6.25) tends to zero as n -> 00. The assumption (6.8) of
Theorem 6.7 is thus satisfied with the operator A defined on D by Af = f".
Finally, to apply Theorem 6.7 to our problem we have to show that for
some A> 0 the range of AI - A is dense in X. Set A = 1. We then have to
show that for a dense set of elements hEX the differential equation
f-f"=h (6.26)
has a solution fED. We will show that this is true for any hEX. Let
f:
hEX and consider the function
f(x) = ~ ( eX {OO h (ne-~ d~ + e- x 00 h (~) e~ d~ )

= l foo h(~)e-I~-xld~. (6.27)
2 -00
It is easy to show thatf ED, Ilfll ::;; Ilhll and thatfis indeed the solution of
(6.26).
Thus all the conditions of Theorem 6.7 are satisfied and we deduce that
the closure of A is the infinitesimal generator of a Co sernigroup of
contractions S( t) on X. In our particular case it is not difficult to show that
A is closed and therefore A itself is the infinitesimal generator of S( t). This
semigroup as we shall see in more detail in the next chapter, is the solution
of the initial value problem (6.20).
Also choosing a sequence k n such that 'Tnkn -> t and 2n 2'Tn = an ::;; 1/ < 1,
we obtain from Theorem 6.7 that
as n -> 00 (6.28)
that is, the values that are computed recursively by the difference equation
(6.22) at the points (kin, l'Tn) converge to the solution of the heat equation
(6.20) at (x, t) where kin -> x, l'Tn -> t as n -> 00.
CHAPTER 4
The Abstract Cauchy Problem
4.1. The Homogeneous Initial Value Problem

Let X be a Banach space and let A be a linear operator from D( A) c X into
X. Given x E X the abstract Cauchy problem for A with initial data x
consists of finding a solution U(/) to the initial value problem
dUd~/) = Au(t), t> 0

{ (1.1)
u(o) = x
where by a solution we mean an X valued function u( I) such that u( I) is
continuous for I ~ 0, continuously differentiable and u(/) E D(A) for I> 0
and (1.1) is satisfied. Note that since u(t) E D(A) for I> 0 and u is
continuous at I = 0, (1.1) cannot have a solution for x ~D(A).
From the results of Chapter I it is clear that if A is the infinitesimal
generator of a Co semigroup T( I), the abstract Cauchy problem for A has a
solution, namely u(t) = T(t)x, for every x E D(A) (see e.g. Theorem 1.2.4).
It is not difficult to show that for x E D( A), u(t) = T(t)x is the only
solution of (1.1). Actually, uniqueness of solutions of the initial value
problem (1.1) follows from much weaker assumptions as we will see in
Theorem 1.2 below.
Lemma 1.1. LeI U(/) be a conli!luous X valued function on [0, T]. If
IlfoTenSu(s) dsll:$; M for n = 1,2, ... (1.2)
Ihen U(/) == 0 on [0, T].

4 The Abstract Cauchy Problem 101
PROOF. Let x* E X* and set cp(t) = (x*, u(t» then cp is clearly continuous
on [0, T] and
I10Tenscp(s ) dsl = I(x*, 1oTenSu(s) dS) I ~ IIx* II . M = MI

for n=I,2, .... (1.3)
We will show that (1.3) implies that cp(t) == 0 on [0, T] and since x* E X*
was arbitrary it follows that u(t) == 0 on [0, T].
Consider the series
This series converges uniformly in 'T on bounded intervals. Therefore,
i E (_I~k-I
T
lo k=1 k.
ekn(I-T+s)cp(s) dsl
~ k~l
00 I ekn(I-T) I10T eknscp(s) ds I ~ M1(exp{en(I-T)} -
k! I). (1.4)
For t < T the right-hand side of (1.4) tends to zero as n -+ 00. On the other
hand we have
10k=IL
TOO (_I)k-I
k.
,ekn(I-T+S)cp(s) ds = 1 (I - exp{ _en(t-T+S)})cp(s) ds.
0
T
(1.5)
Using Lebesgue's dominated convergence theorem we see that the right-hand
side of (1.5) converges to fJ-ICP(S) ds as n -+ 00. Combining this together
with (1.4) we find that for every 0 ~ t < T, fJ-ICP(S) ds = 0 which implies
cp(t) == 0 on [0, T]. D
Theorem 1.2. Let A be a densely defined linear operator. If R(A: A) exists for
all real A ~ Ao and
limsupA -llogIIR(A: A) II ~ 0 (1.6)
,\--->00
then the initial value problem (1.1) has at most one solution for every x E X.
PROOF. Note first that u(t) is a solution of (1.1) if and only if eZ1u(t) is a
solution of the initial value problem
dv
dt = (A + zI)v, v(O) = x.
Thus we may translate A by a constant multiple of the identity and assume
that R(A: A) exists for all real A, A ~ 0 and that (1.6) is satisfied.
Let u(t) be a solution of (1.1) satisfying u(O) = O. We prove that

u(t) == O. To this end consider the function t --> RCA: A)u(t) for A > O.
Since u( t) is a solution of (l.l) we have
d
dt R (A : A) u (t) = R (A : A) Au (t) = AR ( A: A) u ( t) - u (t )
which implies
R(A: A)u(t) = - [eAU-T)u( T) d-r. ( \.7)
o
From the assumption (1.6) it follows that for every 0 > 0
lim e-OhIIR(A: A)II = 0

'\-00
and therefore it follows from (1.7) that
(1.8 )
From Lemma l.l we deduce that u( T) == 0 for 0 $ T $ t - o. Since t and 0

were arbitrary, u( t) == 0 for t ~ O. 0
From Theorem 1.2 it follows that in order to obtain the uniqueness of the
solutions of the initial value problem (l.l) it is not necessary to assume that
A is the infinitesimal generator of a Co semi group or equivalently, that for
some wEIR!, p(A)::l]w,oo[ and II(A-W)nR(A:Anl $M for A>W,
much less than this suffices for the uniqueness. Also to obtain existence of
solutions of (I.l) for some dense subsets D of initial values it is not
necessary to assume that A is the infinitesimal generator of a Co semigroup.
Depending on the set D of initial values, existence results can be obtained
under weaker assumptions. However, in order to obtain existence and
uniqueness for all x E D(A) as well as differentiability of the solution on
[0, oo[ one has to assume that A is the infinitesimal generator of a Co
semigroup. This is the contents of our next theorem.
Theorem 1.3. Let A be a densely defined linear operator with a nonempty

resolvent set p( A). The initial value problem (1.1) has a unique solution u( t),
which is continuously differentiable on [0, 00[, for every initial value x E D( A)
if and only if A is the infinitesimal generator of a Co semigroup T( t).
PROOF. If A is the infinitesimal generator of a Co semigroup T(t) then from

Theorem 1.2.4 it follows that for every x E D(A), T(t)x is the unique
solution of (1.1) with the initial value x E D(A). Moreover, T(t)x is
continuously differentiable for 0 $ t < 00.
On the other hand, if (1.1) has a unique continuously differentiable
solution on [0, oo[ for every initial data x E D(A) then we will see that A is
the infinitesimal generator of a Co semi group T(t). We now assume that for
every x E D(A) the initial value problem (1.1) has a unique continuously
differentiable solution on [0, oo[ which we denote by u(t; x).
For x E D(A) we define the graph norm by IxlG = Ilxll + IIAxll. Since
peA) -=1= 0 A is closed and therefore D(A) endowed with the graph norm is
a Banach space which we denote by [D(A)]. Let X/ o be the Banach space of
continuous functions from [0, to] into [D(A)] with the usual supremum
norm. We consider the mapping S:[D(A)]-+ X/ o defined by Sx = u(t; x)
for 0 ::; t ::; to. From the linearity of (1.1) and the uniqueness of the
solutions it is clear that S is a linear operator defined on all of [D(A)]. The
operator S is closed. Indeed, if Xn -+ x in [D(A)] and SXn -+ v in X/ o then
from the closedness of A and
u{t; xJ = xn + fAu{ T; x n ) dT
o
it follows that as n -+ 00
V{t)=x+ fAv{T)dT
o
which implies vet) = u(t: x) and S is closed. Therefore, by the closed graph
theorem, S is bounded, and
sup lu{t; x)IG::; qXlc. (1.9)
0';/';/0
We now define a mapping T(t) :[D(A)]-+ [D(A)] by T(t)x = u(t; x).

From the uniqueness of the solutions of (1.1) it follows readily that T(t) has
the semi group property. From (1.9) it follows that for 0 ::; t ::; to, T(t) is
uniformly bounded. This implies (see, e.g., the proof of Theorem 1.2.2)
that T(t) can be extended by, T(t)x = T(t - nto)T(toYx for nto ::; t <
(n + 1)10 to a semi group on [D(A)] satisfying 1T(t)xIG ::; Mew/lxlc.
Next we show that
T{t)Ay = AT{t)y (1.10)
Setting
V{t) = y + {u(s; Ay) ds (1.11)
we have
V'{t) = u{t; Ay) = Ay + 1dsU{S; Ay) ds

0
1 d
= A(y + {u(s; Ay) dS) = Av{t). (1.12)
Since v(O) = y we have by the uniqueness of the solution of (1.1), vet) =

u( t; y) and therefore Au( t; y) = v'( t) = u( t; Ay) which is the same as
(1.10).
Now, since D(A) is dense in X and by our assumption peA) -=1= 0 also
D(A2) is dense in X. Let Ao E peA), Ao -=1= 0, be fixed and lety E D(A 2). If
x = (11.01 - A)y then, by (1.10), T(t)x = (AoI - A)T(t)y and therefore

IIT{t)xll = II{AoI - A)T{t)YII ~ CiT{t)YIG ~ Clew11YIG' (1.13)
But
which implies
( 1.14)
Therefore T(t) can be extended to all of X by continuity. After this

extension T(t) becomes a Co seruigroup on X. To complete the proof we
have to show that A is the infinitesimal generator of T( t). Denote by A I the
infinitesimal generator of T(t). If x E D(A) then by the definition of T(t)
we have T( t)x = u(t. x) and therefore by our assumptions
d
dt T{t)x = AT{t)x for t ~ 0
which implies in particular that (d/dt)T(t)xll=o = Ax and therefore

AI::) A.
Let Re A > wand let y E D(A 2 ). It follows from (1.10) and from A I ::) A
that
e-X1AT{t)y = e-X1T{t)Ay = e-X1T{t)Aly· ( 1.15)
Integrating (1.15) from 0 to 00 yields
AR{A: AI)y = R{A: A1)Aly. (1.16)
But A,R(A: A1)y = R(A: AI)Aly and therefore AR(A: A,)y = AIR(A:
A I ) Y for every y E D( A 2). Since A I R (A : A I) are uniformly bounded, A is
closed and D(A2) is dense in X, it follows that AR(A: A1)y = A1R(A: AI)y
for every y E X. This implies D(A) ::) Range R(A: AI) = D(A I) and A ::)
A I' Therefore A = A I and the proof is complete. 0
Our next theorem describes a situation in which the initial value problem
(1.1) has a unique solution for every x E X.
Theorem 1.4. If A is the infinitesimal generator of a differentiable semigroup

then for every x E X the initial value problem (1.1) has a unique solution.
PROOF. The uniqueness follows from Theorem 1.2. If x E D(A) the ex-
istence follows from Theorem 1.3. If x E X then by the differentiability of
T( t)x and the results of Section 2.2.4 it follows that for every x E X,
(d/dt)T(t)x = AT(t)x for t> 0 and AT(t)x is Lipschitz continuous for
t > O. Thus T( t)x is the solution of (1.1). 0
Corollary 1.5. If A is the infinitesimal generator of an analytic semigroup then

for every x E X the initial value problem (1.1) has a unique solution.
If A is the infinitesimal generator of a Co semigroup which is not

differentiable then, in general, if x$. D(A), the initial value problem (1.1)
does not have a solution. The function t ~ T(t)x is then a "generalized
solution" of the initial value problem (1.1) which we will call a mild solution.
There are many different ways to define generalized solutions of the initial
value problem (1.1). All lead eventually to T(t)x. One such way of defining
a generalized solution of (1.1) is the following: A continuous function u on
[O,oo[ is a generalized solution of (1.1) if there are xn E D(A) such that
xn ~ u(O) as n ~ 00 and T(t)x n ~ u(t) uniformly on bounded intervals. It
is obvious that the generalized solution thus defined is independent of the
sequence {x n }, is unique and if u(O) E D(A) it gives the solution of (1.1).
Clearly, with this definition of generalized solution, (1.1) has a generalized
solution for every x E X and this generalized solution is T(t)x.
4.2. The Inhomogeneous Initial Value Problem

In this section we consider the inhomogeneous initial value problem
dUd~t) = Au(t) + Jet) t> 0

{ (2.1 )
u(O) = x
where j:[O, T[ ~ X. We will assume throughout this section that A is the

infinitesimal generator of a Co semigroup T( t) so that the corresponding
homogeneous equation, i.e., the equation with j == 0, has a unique solution
for every initial value x E D(A).
Definition 2.1. A function u:[O, T[ ~ X is a (classical) solution of (2.1) on

[0, T[ if u is continuous on [0, T[, continuously differentiable on )0, T[,
u(t) E D(A) for 0 < t < T and (2.1) is satisfied on [0, T[.
Let T( t) be the Co semigroup generated by A and let u be a solution of
(2.1). Then the X valued function g( s) = T(t - s) u( s) is differentiable for
o < s < t and
dg
-AT(t - s)u(s) + T(t - s)u'(s)
ds
-AT(t - s)u(s) + T(t - s)Au(s) + T(t - s)j(s)
= T(t - s )j(s). (2.2)
If j E LI(O, T: X) then, T(t - s)j(s) is integrable and integrating (2.2)
from 0 to t yields
u(t) = T(t)x + {T(t - s)j(s) ds. (2.3)

Consequently we have
Corollary 2.2. 1/ / E L'(O, T: X) then for every x E X the initial value

problem (2.1) has at most one solution. 1/ it has a solution, this solution is
given by (2.3).
For every / E L'(O, T: X) the right-hand side of (2.3) is a continuous

function on [0, T]. It is natural to consider it as a generalized solution of
(2.1) even if it is not differentiable and does not strictly satisfy the equation
in the sense of Definition 2.1. We therefore define,
Definition 2.3. Let A be the infinitesimal generator of a Co semigroup T( t).

Let x E X and / E L'(O, T: X). The function u E C([O, T]: X) given by
u(t) = T(t)x + fT(t - s)/(s) ds, 0:::; t :::; T,

o
is the mild solution of the initial value problem (2.1) on [0, T].
°
The definition of the mild solution of the initial value problem (2.1)
coincides when / == with the definition of T( t)x as the mild solution of the
corresponding homogeneous equation. It is therefore clear that not every
mild solution of (2.1) is indeed a (classical) solution even in the case / == 0.
For / E L'(O, T: X) the initial value problem (2.1) has by Definition 2.3 a
unique mild solution. We will now be interested in imposing further
conditions on / so that for x E D( A), the mild solution becomes a (classical)
solution and thus proving, under these conditions, the existence of solutions
of (2.1) for x E D(A).
We start by showing that the continuity of /, in general, is not sufficient
to ensure the existence of solutions of (2.1) for x E D( A). Indeed, let A be
the infinitesimal generator of a Co semigroup T( t) and let x E X be such
that T(t)x rE D(A) for any t z 0. Let /(s) = T(s)x. Then /(s) is continu-
ous for s z 0. Consider the initial value problem
dUd~t) = Au(t) + T(t)x

°.
{ (2.4)
u(O) =
We claim that (2.4) has no solution even though u(O)

the mild solution of (2.4) is
= ° E D(A). Indeed,
u(t) = [T(t - s)T(s)xds = tT(t)x,
°
o
but tT(t)x is not differentiable for t > and therefore cannot be the
solution of (2.4).
Thus in order to prove the existence of solutions of (2.1) we have to
require more than just the continuity of f. We start with a general criterion
for the existence of solutions of the initial value problem (2.1).
Theorem 2.4. Let A be the infinitesimal generator of a Co semigroup T( t), let

fE LI(O, T: X) be continuous on ]0, T] and let
vet) = [T(t - s)f(s) ds, o .s; t .s; T. (2.5)

o
The initial value problem (2.1) has a solution u on [0, T[for every x E D( A) if
one of the following conditions is satisfied;
(i) v(t) is continuously differentiable on ]0, T[.
(ii) vet) E D(A) for 0 < t < T and Av(t) is continuous on ]0, T[.
If (2.1) has a solution u on [0, T[ for some x E D( A) then v satisfies hoth (i)
and (ii).
PROOF. If the initial value problem (2.1) has a solution u for some x E D( A)
then this solution is given by (2.3). Consequently vet) = u(t) - T(t)x is
differentiable for t > 0 as the difference of two such differentiable functions
and v'(t) = u'(t) - T(t)Ax is obviously continuous on ]0, T[. Therefore (i)
is satisfied. Also if x E D( A) T( t)x E D( A) for t ::::: 0 and therefore v(t) =
u(t) - T(t)x E D(A) for t> 0 and Av(t) = Au(t) - AT(t)x = u'(t) -
f(t) - T(t)Ax is continuous on ]0, T[. Thus also (ii) is satisfied.
*
On the other hand, it is easy to verify for h > 0 the identity
T(hl - I vet) = v(t + hl - vet) - F+hT(t + h - s)f(s) ds. (2.6)

I
From the continuity of f it is clear that the second term on the right-hand
side of (2.6) has the limit f(t) as h -> O. If v( t) is continuously differentiable
on ]0, T[ then it follows from (2.6) that v(t) E D( A) for 0 < t < T and
Av(t) = v'(t) - f(t). Since v(O) = 0 it follows that u(t) = T(t)x + vet) is
the solution of the initial value problem (2.1) for x E D( A). If v( t) E D( A)
it follows from (2.6) that vet) is differentiable from the right at t and the
right derivative D+v(t) of v satisfies D+v(t) = Av(t) + f(t). Since D+v(t)
is continuous, vet) is continuously differentiable and v'(t) = Av(t) + f(t).
Since v(O) = 0, u(t) = T(t)x + vet) is the solution of (2.1) for x E D(A)
and the proof is complete. 0
From Theorem 2.4 we draw the following two useful corollaries.
Corollary 2.5. Let A be the infinitesimal generator of a Co semigroup T(t). If

f( s) is continuous(v differentiable on [0, T] then the initial value problem (2.1)
has a solution u on [0, T[ for every x E D( A).
PROOF. We have
vet) = llT(t - s )f(s) ds = [T(s )f(t - s) ds. (2.7)

o 0
It IS clear from (2.7) that v (t) is differentiable for t > 0 and that its
derivative
v'{t) = T{t)f{O) +lIT{s)/,{t - s) ds = T{t)f(O) + [T(t - s)/,{s) ds

o 0
is continuous on ]0, T[. The result therefore follows from Theorem 2.4 (i). 0
Corollary 2.6. Let A be the infinitesimal generator of a Co semigroup T( t). Let

f E LI(O, T: X) be continuous on ]0, T[. If f(s) E D(A) for < S < T and °
Af(s) E LI(O, T: X) then for every x E D(A) the initial value problem (2.1)
has a solution on [0, T[.
PROOF. From the conditions it follows that for s > 0, T(t - s)f(s) E D(A)
and that AT(t - s)f(s) = T(t - s)Af(s) is integrable. Therefore vet)
defined by (2.5) satisfies v (t) E D( A) for t > and °
Av{t) = A {T(t - s)f{s) ds = {T(t - s)Af{s) ds
is continuous. The result follows now from Theorem 2.4 (ii). o

As a consequence of the previous results we can prove,
Theorem 2.7. Let fE LI(O, T: X). Ifu is the mild solution of (2.1) on [0, T]
then for every T' < T, u is the uniform limit on [0, T'] of solutions of (2.1).
PROOF. Assume that II T(t)11 ~ Me w1 . Let x1/ E D(A) satisfy xn ~ x and let
/" E CI([O, T]: X) satisfy /" ~ f in LI(O, T: X). From Corollary 2.5 it
follows that for each n ~ 1 the initial value problem
dU/1{t) () ()
{ ~ = AU/1 t + /" t (2.8)
un{O) - xn
has a solution u/1(t) on [0, T[ satisfying
u 1/ (t) = T{ t ) X /1 + 101T{ t - s) /" (s ) ds.
If u is the mild solution of (2.1) on [0, T] then
Ilu/1(t) - u(t)11 ~ MeW11lxn - xII + [Mew(l-S)llfn(s) - f(s)llds

o
~ Me WT ( IIxn - xli + laTllfn(s) - f{s )lldS) (2.9)
and the result follows readily from (2.9). 0
We conclude this section with a few remarks concerning still another

notion of solution of the initial value problem (2.1) namely the strong,
solution.
Definition 2.S. A function u which is differentiable almost everywhere on

[0, T] such that u' E LI(O, T: X) is called a strong solution of the initial
value problem (2.1) if u(O) = x and u'(t) = Au(t) + f(t)a.e. on [0, T].
We note that if A = 0 and f E L 1(0, T: X) the initial value problem (2.1)

has usually no solution unless f is continuous. It has however always a
strong solution given by u(t) = u(O) + idf(s). It is easy to show that if u is
a strong solution of (2.1) and f E LI(O, T: X) then u is given by (2.3) and
therefore is a mild solution of (2.1) and is the unique strong solution of
(2.1). A natural problem is to determine when is a mild solution a strong
solution. It is not difficult to show, essentially with the same proof as the
proof of Theorem 2.4 that we have:
Theorem 2.9. Let A be the infinitesimal generator of a Co semigroup T( t), let

f E LI(O, T: X) and let
vet) = [T(t - s)f(s) ds, 0::; t::; T.

o
The initial value problem (2.1) has a strong solution u on [0, T] for every
x E D( A) if one of the following conditions is satisfied;
(i) vet) is differentiable a.e. on [0, T] and v'(t) E LI(O, T: X)
(ii) vet) E D(A) a.e. on [0, T] and Av(t) E LI(O, T: X).
If (2.1) has a strong solution u on [0, T] for some x E D( A) then v satisfies
both (i) and (ii).
As a consequence of Theorem 2.9 we have:
Corollary 2.10. Let A be the infinitesimal generator of a Co semigroup T(t). If
f is differentiable a.e. on [0, T] and f' E LI(O, T: X) then for every x E D( A)
the initial value problem (2.1) has a unique strong solution on [0, T].
In general, the Lipschitz continuity of f on [0, T] is not sufficient to assure

the existence of a strong solution of (2.1) for x E D( A). However, if X is
reflexive and f is Lipschitz continuous on [0, T] that is
Ilf(t l ) - f(t 2 )!! ::; Cltl - t21 for t l , t2 E [0, T]
then by a classical result f is differentiable a.e. and f' E LI(O, T: X).
Corollary 2.10 therefore implies:
Corollary 2.11. Let X be a reflexive Banach space and let A be the infinitesi-
mal generator of a Co semigroup T(t) on X. If f is Lipschitz continuous on
[0, T] then for every x E D(A) the initial value problem (2.1) has a unique
solution u on [0, T] given by
u(t) = T(t)x + {T(t - s)f(s) ds.
PROOF. From the previous remarks it is obvious that (2.1) has a strong
solution. Therefore by Theorem 2.9, v(t) given by (2.5), is differentiable
a.e. on [0, T] and
v'(t) = T(t)f(O) + [T(t - s)f'(s) ds = get).

o
It is easy to verify that get) is continuous on [0, T] and the result follows
from Theorem 2.4. 0
4.3. Regularity of Mild Solutions for Analytic

Semigroups
Let A be the infinitesimal generator of a Co semigroup T( t) and let
f E LI(O, T: X). In the previous section we defined the mild solution of the
initial value problem
duj/) = Au(t) + f(t)
{ (3.1 )
u(O) = x
to be the continuous function
u(t) = T(t)x + [T(t - s)f(s) ds. (3.2)
o
We saw that if one imposes further conditions on f, e.g., f E CI([O, T]: X)
then the mild solution (3.2) becomes a (classical) solution, i.e., a continu-
ously differentiable solution of (3.1). If A is the infinitesimal generator of an
analytic semigroup we have stronger results. For example we will see
(Corollary 3.3) that in this case Holder continuity of f already implies that
the mild solution (3.2) is a solution of (3.1).
We start by showing that if T(t) is an analytic semigroup and f E
LP(O, T: X) with p > I then the mild solution (3.2) is Holder continuous.
More precisely we have:
Theorem 3.1. Let A be the infinitesimal generator of an analytic semigroup
T(t) and let f E LP(O, T: X) with I 0. If moreover x E D(A) then u is Holder continuous with the same
exponent on [0, T].
PROOF. Let II T(t)11 ::;; M on [0, T). Since T(t) is analytic there is a constant
C such that IIAT(t) II ::;; Ct- I on ]0, T]. This implies that T( t)x is Lipschitz
continuous on [e, T] for every e> 0. If x E D(A), T(t)x is Lipschitz
continuous on [0, T). It suffices therefore to show that if f E LP(O, T: X),
°
I we have
vet + h) - vet) = j t+h
T(t +h - s)f(s) ds
t
+ [(T(t + h - s) - T(t - s»f(s) ds = II + 12 ,

o
We estimate II and 12 separately. For II we use Holder's inequality to

obtain,
IIIIII50M j T+h Ilf(s)llds5o Mh(p-I)/p (jT+h Ilf(s)IIPds)I/P50 Mlflph(P-I)/P
T T
(3.3)
°
where Iflp = <ioTllf(s) liP dS)I/p is the norm offin LP(O, T: X). In order to
estimate 12 we note that for h >
IIT(t + h) - T(t)1I 50 2M for t, t +h E [0, T]
and
IIT(t + h) - T(t)1I 50 C~ for t, t + h E ]0, T].
Therefore,
II T( t + h) - T( t) II 50 CIJ.L( h, t) = C I min ( 1, ~ )
for t, t + h E [0, T] (3.4)
where C I is a constant satisfying C I ~ max (2M, C). Using (3.4) and
Holder's inequality we find
iT (iT
111211 50C I oJ.L(h,t-s)llf(s)llds5oCllfl p /(h,t-s)P P- ds
/( I) )(P-I)/P
.
°
(3.5)
But since J.L ~ we have
1o J.L(h, t-s)P/(P-1) ds= fJ.L(h, rV/(p-l) dr50 l°OJ.L(h, rV/(p-l) dr=ph

T
0 0
and combining (3.5) with the last inequality we find 111211 50 const . h(p-I)/p.
o
We turn now to conditions on f that will ensure that the mild solution of
(3.1) is a strong solution.
liT > °
T(t). Let f E LI(O, T: X) and assume that for every < t < T there is a
and a continuous real value function ~(r):
°
[0, oo[ ~ [0, oo[ such that
Ilf(t) - f(s)1I 50 ~(It - sl) (3.6)
and
1o - - d r <
81 ~(r)
r
00. (3.7)
Then for every x E X the mild solution of (3.1) is a classical solution.

PROOF. Since T(t) is an analytic sernigroup, T(t)x is the solution of the
homogeneous equation with initial data x for every x E X. To prove
°
the theorem it is therefore sufficient, by Theorem 2.4, to show that vet) =
f;T(t - s)f(s) ds E D(A) for < t < T and that Av(t) is continuous on
this interval. To this end we write

V(I) = VI(I) + V2 (1)
= [IT(I - s)([(s) - /(1» ds + {T(t - S)/(I) ds. (3.8)
10 0
From Theorem 1.2.4 (b) it follows that V2 (1) E D(A) and that AV2(t) =
(T( t) - 1)/( I). Since the assumptions of our theorem imply that / is
continuous on ]0. T[ it follows that Av 2 (t) is continuous on ]0. T[. To prove
the same conclusion for v I we define
for 1~ E (3.9)
and
V1 .• (I) = 0 for 1 < E. (3.10)
From this definition it is clear that v I .• ( I) -+ V I (t) as E -+ O. It is also clear
that vI.,(t) E D(A) and for 1 ~ E
AVI..(t) = fal-'AT(t - s)([(s) - /(1» ds. (3.11)
From (3.6) and (3.7) it follows that for I > 0 Avl .•(t) converges as E -+ 0
and that
limAvl..(I) = [IAT(I - s)([(s) - /(1» ds.
f~O 10
The closed ness of A then implies that v I (t) E D( A) for t > 0 and
AVI(/) = falAT(1 -- s)([(s) - /(t» ds. (3.12)
To conclude the proof we have to show that AvI(t) is continuous on ]0. T[.
For 0 < 8 < t we have
AV1(1) = fo6AT(t - s)(/(s} - /(/» ds + ~IAT(I - s)([(s) - /(/» ds.
(3.13)
For fixed 8 > 0 the second integral on the right of (3.13) is a continuous
function of t while the first integral is 0(8) uniformly in I. Thus. Avl(t) is.
continuous and the proof is complete. 0
Let I be an interval. A function /: I -+ X is Holder continuous with

exponent {}, 0 < {} < I on I if there is a constant L such that
11/(/) - /(s}ll ~ LII - sl" for S,I E I. (3.14)
I t is locally Holder continuous if every l E I has a neighborhood in which /

is Holder continuous. It is easy to check that if 1 is compact then/is Holder
continuous on I if it is locally Holder continuous. We denote the family of
all Holder continuous functions with exponent {} on I by C"(l: X).
An immediate consequence of Theorem 3.2 is,
Corollary 3.3. Let A be the infinitesimal generator 01 an analytic semigroup

T(t). II IE LI(O, T: X) is locally Holder continuous on ]0, T] then lor every
x E X the initial value problem (3.1) has a unique solution u.
More can be said on the regularity of the solution u under the assump-
tions of Corollary 3.3. This will be seen in Theorem 3.5. In the proof of
Theorem 3.5 we will need the following lemma.
Lemma 3.4. Let A be the infinitesimal generator 01 an analytic semigroup T( t)

and let IE CII([O, T]: X). II
VI{t) = fotT{t - s)(f{s) - I{t)) ds (3.15)
then VI(t) E D(A) lor ° ~ t ~ T and Avl(t) E CII([O, T]: X).
PROOF. The fact that VI(t) E D(A) for 0 ~ t ~ T is an immediate conse-

quence of the proof of Theorem 3.2, so we have only to prove the Holder
continuity of Avl(t). Assume that II T(t)1I ~ M on [0, T] and that
IIAT(t)II~Ct-1 for O<t~T. (3.16)
Then, for every 0 < S < t ~ T we have
IIAT(t) -AT(s)11 =11[A 2 T(T) dTIl ~ [IIA 2 T(T)lldT

~ 4C [T- 2 dT = 4Ct- IS- I(t - s). (3.17)
s
Let t ~ 0 and h > 0 then
Av I (t + h) - A v I ( t) = A [( T( t + h - s) - T( t - s)) (f (s) - I (t )) ds
o
+ A fot T( t + h ~ s )(f( t) - I( t + h)) ds
+A j t+h
T(t + h - s)(/(s) - I(t + h)) ds
t
= II + 12 + 13 , (3.18)
We estimate each of the three terms separately. First from (3.14) and (3.17)
we have
111111 ~ [IIAT(t + h - s) - AT(t - s)llll/(s) - l(t)lIds
o
~ 4CLh[ ds I II ~ Clhll. (3.19)

o (t-s+h)(t-s)-
To estimate 12 we use Theorem 1.2.4(b) and (3.14),
11/211 = II(T(t + h) - T(h))(f(t) - l{t + h))11
~ IIT(t + h) - T(h)lIll/(t) - I(t + h)11 ~ 2MLhll. (3.20)
Finally, to estimate 13 we use (3.16) and (3.14) to find
111311 :s; J
t
t+h
IIAT(t + h - s)IIII/(s) - I(t + h)lIds
:s; CL[+\t + h - s)D-lds:s; C2 hD. (3.21)

t
Combining (3.18) with the estimates (3.19), (3.20) and (3.21) we see that
Avl(t) is Holder continuous with exponent 1Jo on [0, T]. 0
The main regularity result for the case where A generates an analytic
semigroup and I is Holder continuous comes next.
Theorem 3.5. Let A be the infinitesimal generator 01 an analytic semigroup

T( t) and let IE CD([O, T] : X). II u is the solution 01 the initial value problem
(3.1) on [0, T] then,
(i) For every 8 > 0, Au E C D([ 8, T] : X) and du/ dt E C D([ 8, T] : X).
(ii) II x E D(A) then Au and du/dt are continuous on [0, T].
(iii) II x = ° °
and 1(0) = then Au, du/ dt E CD([O, T] : X).
PROOF. We have,
u(t) = T(t)x + fT(t - s)/(s) ds = T(t)x + v(t).
o
8> °
Since by (3.17) AT(t)x is Lipschitz continuous on 8 :s; t:s; T for every
it suffices to show that Av(t) E C D([8, T]: X). To this end we
decompose v as before to
v ( t) = VI( t) + v 2 ( t) = {T( t - s ) (f ( s) - I ( t)) ds + {T( t - s ) I ( t) ds.
From Lemma 3.4 it follows that Av 1(t) E Ci}([O, T]: X) so it remains

only to show that Av 2(t) E Ci}([8, T]: X) for every 8 > O. But Av 2(t) =
(T(t) - 1)1(1) and since IE Ci}([O, T]: X) we have only to show that
T(t)/(t) E Ci}([8, T]: X) for every 8 > O. Let t ~ 8 and h > 0 then
IIT(t + h)/(t + h) - T(t)/(t)1I
:s; IIT(t + h)lllI/(t + h) - l(t)1I + IIT(t + h) - T(t)llll/(t)1I
(3.22)
where we used (3.4), (3.14) and denoted 11/1100 = maxO$/$T II/(t)lI. This
completes the proof of (i). To prove (ii) we note first that if x E D(A) then
AT(t)x E C([O, T]: X). By Lemma 3.4 Avl(t) E CD([O, T]: X) and sincel
is continuous on [0, T] it remains only to show that T(t)/(t) is continuous
°
on [0, T]. From (i) it is clear that T(t)I(t) is continuous on ]0, T]. The
continuity at t = follows readily from,
IIT(t)/(t) - 1(0)11 :s; IIT(t)/(O) - 1(0)11 + MII/(t) - 1(0)11
and this completes the proof of (ii). Finally. to prove (iii) we have again only
to show that in this case T(t )f(l) E C~([O. T 1: X) and this follows from
IIT(I + h)f(1 + h) - T(I)f(I)1I
~ IIT(t + h)llllf(t + h) - f(t)1I + II(T(t + h) - T(t»f(1)11
~ MLh'~ +IIF+hAT(T)f(t) dTIl
~ MLh~ + r;hIIAT(T)(f(t) - f(O»lIdT
I
~ MLh~ + CL f l+hT-Itl} dT ~ MLh'~ + CL fl+h TI}-I dT ~ ChI}

I I
We conclude this section with a result which is analogous to Theorem 3.2

in which the condition on the modulus of continuity of f is replaced by
another regularity condition.

T(t) and let 0 E peA). If f(s) is continuous, f(s) E D«-A)a), 0 < a ~ I
and II ( - A )"f( .1') II is bounded, then for every x E X the mild solution of (3.1)
is a classical solution.
PROOF. As in the proof of Theorem 3.2 it suffices to show that vet) E D(A)
for t > 0 and that Av( t) is continuous for t > O. Since T( t) is analytic
T(t - s)f(s) E D(A) for t > s and by Theorem 2.6.13(c),
IIAT(t - s)f(s)1I = II( -A)I-aT(t - .1')( -A)af(s)11
~ Cjt - sla-III(-Arf(s)lI.
Therefore AT(t - s)f(s) is integrable and vet) E D(A) as well as
AV(/) = [AT(t - s)f(s) ds.

o
The continuity of Av(t) for t > 0 is proved exactly as the continuity of
Alv(/) is proved in Theorem 3.2. 0
4.4. Asymptotic Behavior of Solutions

In this section we intend to study the asymptotic behavior of solutions of
the initial value problem
due I)
----;}{ = Au(t) + f{t), u(O) = x. (4.1 )
We start with the solutions of the homogeneous problem i.e.,f == 0 and look
for conditions that guarantee their exponential decay.
Theorem 4.1. Let A be the infinitesimal generator of a Co semigroup T(t). If

for some p, 1 ::; P < 00
for every x E X ( 4.2)
then there are constants M ;::: 1 and p. > 0 such that II T(t)11 ::; Me-P.I.
PROOF. We start by showing that (4.2) implies the boundedness of t ~
IIT(t)ll. Let IIT(t)11 ::; M1e wl where MI ;::: 1 and w;::: O. If w = 0 there is
nothing to prove so we assume w > O. From (4.2) it then follows that
T( t)x ~ 0 as t ~ 00 for every x E X. Indeed, if this were false we could
find x E X, 8 > 0 and t) ~ 00 such that II T(t))xll ;::: 8. Without loss of
generality we can assume that t) + 1 - t) > w - I. Set 11) = [t) - w - I, t)], then
m(l1) = w- I > 0 and the intervals 11) do not overlap. For t E 11) we have
II T(t)xll ;::: 8(M 1e)-1 and therefore
contradicting (4.2). Thus T(t)x ~ 0 as t ~ 00 for every x E X and the

uniform boundedness theorem implies II T(t)11 ::; M for t ;::: O.
Next, consider the mapping S; X ~ LP(~ + ; X) defined by Sx = T(t)x.
From (4.2) it follows that S is defined on all of X. It is not difficult to see
that S is closed and therefore, by the closed graph theorem, S is bounded,
i.e.,
[~)IIT(t)XIIPdt::; Mfllxll P. (4.3)
o
Let 0 < p < M- 1 where IIT(t)11 ::; M. Define tAp) by
tx(p) = sup{t; II T(s )xll ;::: pllxll for 0::; s ::; t}.
Since IIT(t)xll ~ 0 as t ~ 00, tAp) is finite and positive for every x E X.
Moreover,
fO
t)p )pPllxll P ::; [x(Pl ll T(t)xII P dt ::;
o 0
II T(t)xII P dt ::; MfllxliP
and therefore tAp)::; (M2 /p)P = to. For t > to we have
IIT(t)xll ::; IIT(t - tx(p))IIIIT(tx(p))xll ::; Mpllxll ::; f3llxll
where 0 ::; f3 = Mp < l. Finally, let tl > to be fixed and let t = nt l + S,
o ::; s < t I' Then
IIT(t)ll::; IIT(s)IIIIT(ntl)11 ::; MIIT(tl)ll n ::; Mf3n::; M'e-P.I
where M' = M{3-1 and JL = -(1/t l )log {3 > O. o
Theorem 4.1 shows that if T(t)x E LP(R + : X) for every x E X then

II T(t)11~ Me-I't for some M ~ 1 and J1. > 0. We are now interested in
conditions on the infinitesimal generator A of T(t) which will insure a
similar behavior.
For a Banach space X of finite dimension it is well known that if
°
sup {Re A: A E o(A)} = a < then II T(t)11 decays exponentially. This be-
havior is a consequence of the fact that linear operators in finite dimen-
sional Banach spaces have only point spectrum. Since this is not the case in
general Banach spaces one does not expect this result to be true in general
Banach spaces.
EXAMPLE 4,2. For a measurable function I on [0, oo[ set
III1 = IaooeSI/{s} Ids
and let E be the space of all measurable functions I on [0, oo[ for which
III1 < 00. Let X = Ell LP(O, 00), 1 < P < 00. X endowed with the norm
11/11 = III1 + 11/11 LP is easily seen to be a Banach space. In X we define a
semigroup {T(t)} by;
T{t}/{x} = I{x + t} for t ~ 0. {4.4}

It follows readily from its definition that {T(t)} is a Co semigroup on X and
that II T(t)11 ~ 1. Choosing I E X to be the characteristic function of the
interval [t, t + f P ), f > 0, and letting f J,O shows that II T(t)11 ~ 1 and thus
IIT(t)11 = 1 for t ~ 0.
The infinitesimal generator A of {T(t)} is given by
D(A} = {u: u is absolutely continuous, u' E X} (4.5)
and
Au = u' for u E D(A}. (4.6)
Let I E X and consider the equation
AU - Au = AU - u' = I. (4.7)
A simple computation shows that
u(t} = 1o e-~Sl(t + s}
00
ds = e~ljoo e-~SI(s} ds
t
(4.8)
is a solution of (4.7). We will show that if A satisfies Re A > -1 then u,

given by (4.8), is in D(A) and thus {A: Re A > -I} c p(A). To show that
u E D(A) it suffices by (4.7) to show that u E X and this follows from
lu(t)1 ~ eRe~tjooe-(Re~+I)SeSI/(s}lds ~ e-tjooeSI/(s}lds ~ e-tl/l l

t t
which implies that u E LP(O, 00), and
lull :s; 1o jooe(Re

00
t
11.+ 1)(t-S)eSI/( s) Ids dt
= fooo({e(Rd.+I)(I-S) dt )esl/(s)ldS
= (ReA + 1)-110 (1 - e-(Re~+I)S)eSI/(s)lds

00
:s; (Re A + 1) - 1III1 .

The set {A: Re A > - I} is therefore a subset of p(A), C1 = sup {Re A: A E
C1(A)}:s; -1 while II T(t)11 does not decay exponentially. D
From Example 4.2 we conclude that in order to obtain exponential decay

of II T(t)11 from the spectral condition sup{Re A: A E C1(A)} = C1 < 0 one
has to supplement it with some further conditions on T(t) or A. There are
many possible assumptions that imply the result. We choose here a simple
but rather useful such assumption, namely that A is the infinitesimal
generator of an analytic semigroup.
Theorem 4.3. Let A be the infinitesimal generator 01 an analytic semigroup

T(t). II
C1 = sup {Re A : A E C1 (A)} < 0
then there are constants M ~ 1 and p. > 0 such that II T(t)1I :s; Me-Ill.
PROOF. From the results of Section 2.5 it follows easily that there are
constants w ~ 0, M ~ 1, 8 > 0 and a neighborhood U of A = w such that
p(A):::)~={A:larg(A-w)I<~+8}UU (4.9)
and
M
IIR(A: A)II :s; IA - wi for AE ~. (4.10)
<Moreover,
T(t) = _1_. [e~IR(A: A) dA (4.11 )

2m Jr
where f consists of the two rays fl = {A = pei~ + w: p ~ 0, 77/2 < it <
77/2 + 8} and f2 = {A = pe-i~ + w: p ~ 0, 77/2 < it < 77/2 + 8} and is
oriented such that 1m A increases along f. The convergence in (4.11) for
t> 0 is in the uniform operator topology. By our assumption R(A: A) is
analytic in a neighborhood of the triangle
,;l = {A : Re A > C1 1 , Iarg( A - w) I ~ it}
where 0> (JI > (J. From Cauchy's theorem it follows that f in (4.11) can be
shifted without changing the value of the integral to the path f' where f' is
w-a}
composed of
f' =
1
{=A pe iit + w'• P >- 1
Icos 1'J-1 '
f2 = {Re A = (JI : 11m AI ::;; (w - (JI)ltan it!),
f' = {A = pe- iit + w: p ;::: w - (JI }

3 Icos itl
and is oriented so that 1m A increases along f'. Thus
T(t) = -1-.1 e~IR(A: A) dA.

2m r'
Estimating II T(t)II, on f; i = 1,2,3 one finds easily that for t ;::: 1 and some
constant M I, II T(t)11 ::;; MI ea,l. Since II T(t)11 ::;; M2 for 0 ::;; t ::;; I we have
II T(t)11 ::;; Mea,l for t ;::: 0 and the proof is complete. 0
We turn now to some simple results on the asymptotic behavior of mild

solutions of the inhomogeneous initial value problem (4.1).
Theorem 4.4. Let p. > 0 and let A be the infinitesimal generator of a Co

semigroup T(t) satisfying II T(t)1I ::;; Me-P.l. Let f be bounded and measurable
on [0, 00[. If
lim f(t) = fo (4.12)
I~OO
then, u( t), the mild solution of (4.1) satisfies

lim u( t) = -A - I/o. (4.13)
I~OO
PROOF. Since II T(t)11 ::;; Me-P.l it follows that 0 E peA) (see Theorem 1.5.3)
and II T(t)xlI-+ 0 as t -+ 00. Now
v(t) = llT(t - s)f(s) ds = fT(t - s}[f(s) - fo] ds

o 0
Clearly, (see proof of Theorem 1.3.1),
lim v 2(t) =l°OT(t)fodt = R(O:A)fo = -A-Ifo'

I~oo 0
°as t
°
To complete the proof we have to show that VI(t)
Given E > we choose to such that for t > to
-+ -+ 00.
IIf(t) - foil < 2E~. (4.14)

Then, setting IIflloo = sUP,~o 11/(1)11 we have,

Ilv,(t)1I ~ l'°IlT(t - s)U Ilf(s) - follds
o
Choosing t > to large enough, the first term on the right becomes less than
e/2 and thus for t large enough IIvl(l)11 < e and the proof is complete. 0
A result of similar nature is the following:
Theorem 4.5. Let JL > 0 and let A be the infinitesimal generator of a Co

semigroup T(t) satisfying II T(t)1I ~ Me-"'. Let f be continuous and bounded
on [0,00[. If u,(t) is the mild solution of
du,( t)
e-;]( = Au.(t) + f(t), u.(O) = x, e>O (4.15)
then
(4.16)
and the Iimit'is uniform on every interval [8, T] where 0 < 8 < T.
PROOF. The operator e-IA is clearly the infinitesimal generator of the Co
semigroup T.(t) = T(t/e). We have
u,(t) = T.(t)x + e-I{T.(t - s)f(s) ds. (4.17)
Since 11T.(t)11 ~ Me-("/')' it follows that 11T.(t)xll -+ 0 as e -+ 0 uni-

formly on every interval [8, T]. Now,
v,(t) = e-I{T.(t - s)f(s) ds
= e- l l'T.(t - s)[f(s) - f(t)] ds + e- I (1T.(t - s)f(t) ds

o 10
= V'I ( t) + V E2 (t ).
For V,I{t) we have
~ Me-1fo1e-I'T/'llf(t - 'T) - f(t)lId'T
~ Me-I ~('e-I'T/'llf(t - 'T) - f(t)lId'T + 211fll MJL-Ie-I'r

00
= M (e-""lIf(t - eo) - /(t)lIdo + 2e-,.rMIl/1l JL- 1

10 00
where Ilflloo = sUPt;"ollf(t)11 and r> O. Given p > 0 we first choose r so

large that the second term on the right-hand side becomes less than p/2 and
then choose e so small that, by the continuity of f, the first term on the
right-hand side is less than p/2. Thus v,,(t) ~ 0 as e ~ O. Finally,
v E2 {t) = e-I[1'.{t - s)f{t) ds = e-I[1'.{ T)f{t) dT
o 0
= e- 1 1o 1'.{
OO
T)f{t) dT - e-1joo1'.{ T)f{t) dT
t
= -A-1f{t) + T.{t)A-lf{t).
Letting € ~ 0 we therefore have vel (t) ~ - A - If( t) uniformly on [I), T]. 0
Remark. In Theorem 4.5 if x E D( A) and f is continuously differentiable

on [0, oo[ then it is not difficult to show that
du f ( t) ~ _A- If' ( t ) as € ~0 (4. 18 )

dt
and the limit is uniform on compact subsets of ]0, T[.
4.5. Invariant and Admissible Subspaces
Let X be a Banach space, Ya subspace (not necessarily closed) of X and let

s: DeS) c X ~ X be a linear operator in X. The subspace Y of X is an
invariant subspace of S if S: D(S) n Y ~ Y.
Given a Co semigroup T( t) on X we will be interested in conditions for a
subspace Y of X to be an invariant subspace of T( t) for all t 2 O. Such a
subspace will be called an invariant subspace of the semigroup T( t). If Y is a
closed subspace of X we have:
Theorem 5.1. Let T( t) be a Co semigroup on X with infinitesimal generator A.

If Y is a closed subspace of X then Y is an invariant subspace of T( t) if and
only if there is a real number w such that for every A > w, Y is an invariant
subspace of R(A: A), the resolvent of A.
PROOF. From the results of Chapter 1 it follows that there is an w such that
RCA: A)x = [Oe-AtT(t)xdt (5.1)

o
for A > w. Thus, AR(A: A)x is in the closed convex hull of the trajectory
{T( t)x: t 2 O}. If T( t) Y c Y for every t 2 0 it follows from (5.1) that
AR(A: A)Y c Y for every A > w.
Conversely, by the exponential formula (Theorem 1.8.3) we have
T{t)x = lim ('2R('2 : A))"X for x E X. (5.2)

I/~OC t t
If RCA: A) Y c Y for A > w then for all n large enough «n/t )R(n/t: A»" Y
c Y and (5.2) implies that T(t) Y c Y for every t > 0. 0
Remark. From the proof of Theorem 5.1 it is clear that the result holds also
if Y is a closed convex cone with vertex at zero, rather than a closed
suhspace of X.
In the sequel we will be interested in invariant subspaces Y which are not

closed in X. In order to state such results we need some preliminaries. We
start by recalling (see Definition 1.l0.3) that if s: D(S) c X -> X and Y is
a subspace of X then the part of S in Y is the linear operator S with the
domain D(S) = {x E D(S) () Y: Sx E Y} and for x E D(S). Sx = Sx.
The restriction SlY of S to _Y clearly satisfies SlY ::l S. If Y is an invariant
subspace of S then SI Y = S.
Lemma 5.2. Let S: D( S) c X -> X be invertible and let Y be a subspace of

X. If S-Iy c Y then S, the part of Sin Y is invertible and S-I = (S-I)IY'
PROOF. Let x E Y and z = S--IX. Then z E D(S) () Y and Sz = x E Y.
Therefore z E D(S) and Sz = Sz = x. This shows that the range of S is all
of Y and that S -I is well defined and S- IX = S- IX for all x E Y, I.e.,
S-I=(S-I)IY' 0
In the rest of this section we will assume that X is a Banach space, Y is a

subspace of X which is closed with respect to a norm II II y (and hence is
itself a Banach space). We will further assume that the norm II II y is
stronger than the original norm II II of x. This means that there is a
constant C such that
lIyll ~ Cilyll Y for y E Y. (5.3)
Note that by assumption Y is closed in the norm II II y but in general it is
not closed in the norm II II.
Definition 5.3. Let T(t) be a Co semigroup and let A be its infinitesimal

generator. A subspace Y of X is called A-admissible if it is an invariant
subspace of T( t), I ::2: 0, and the restriction of T( t) to Y is a Co semigroup in
Y (i.e., it is strongly continuous in the norm II II y).
EXAMPLE 5.4. Let X be the space of bounded uniformly continuous real

valued functions on [0, oo[ with the usual supremum norm and let Y' =
X () CI([O, oo[). Set
T( t ) f (x) = f (x + t) for f EX, t z 0. (5.4)
T(t) is obviously a Co semigroup of contractions on X. Its infinitesimal

generator A is given by D( A) = {f E Y': f' E X} and Af = f' for f E D( A).
Denoting the norm in X by II II, we consider the space Y of elements
g E Y' for which g' E X. We equip Y with the norm Ilgll y = IIgll + Ilg'll
for g E Y. The norm II II y is stronger than II II, Y is closed in the norm
II II y and it is easy to see that the semigroup T(t) defined by (5.4) leaves Y
invariant and is a Co semigroup in Y. Thus Y is A-admissible.
Theorem 5.5. Let T( t) be a Co semigroup on X and let A be its infinitesimal

generator.
A subspace Y of X is A -admissible if and only if
(i) Y is an invariant subspace of R(A: A) for all A > w.

(ii) A, the part of A in Y, is the infinitesimal generator of a Co semigroup
on Y.
Moreover, if Y is A-admissible then A is the infinitesimal generator of the

restriction of T( t) to Y.
PROOF. Assume that Y is A-admissible. Since T(t)Y c Y for t ~ 0 and

since II II y is stronger than II II and the restriction of T(t) to Y is a Co
semigroup in Y it follows from (5.1) that there is an w such that for A> w
R(A: A)Y c Y. Let Al be the infinitesimal generator of the restriction of
T( t) to Y. From the definition of the infinitesimal generator it follows
readily that D(AI) c D(A) () Yand that for x E D(AI)' Alx = Ax and so
A :J AI. On the other hand if x E D(A) then Ax E Yand the equality
T(t)x - x = {T(s )Ax ds (5.5)
holds in Y. Dividing (5.5) by t > 0 and letting t ! 0 it follows that x E D( A I)

and so D(A 1) :J DC'I). Thus A = Al and A is the infinitesimal generator
of a Co semigroup on Y, namely, the restriction of T(t) to Y.
Conversely assume that (i) and (ii) are satisfied and denote by S(t) the Co
semigroup generated by A on Y. From the assumption (i) and Lemma 5.2 it
follows that R(A: A)x = R(A: A)x for every x E Yand therefore also
(5.6)
for all n large enough and x E Y. Passing to the limit as n -+ 00 it follows

from the exponential formula (Theorem 1.8.3) that the left-hand side of
(5.6) converges in Y, and hence also in X, to S(t)x while the right-hand side
converges in X to T(t)x. Therefore S(t)x = T(t)x for every x E Y which
implies both that Y is an invariant subspace for T( t) and that T( t) is a Co
semigroup on Y. 0
Corollary 5.6. Y is A -admissible if and only if
(i) For sufficient~y large A, Y is an invariant subspace of RCA. : A).

(ii) There exist constants M and /3 such that
IIR(A: Arllt·:s; M(A. - /3)-n, A > {j, n = 1,2,.... (5.7)

(iii) For A > /3, R( A: A) Y is dense in Y.
PROOF. Condition (i) is the same as in Theorem 5.5. From (i) and Lemma
5.2 it follows that R(A: A)x = R(A: A)x for x E Y and A> w. Therefore
we can replace A by A in (5.7) and in Condition (iii). Condition (iii) is then
equivalent to the fact that D(A) = R(A: A)Y = R(A: A)Y is dense in Y
and from Theorem 1.5.3 it follows that A generates a Co semigroup on Y if
and only if (ii) and (iii) are satisfied. From Theorem 5.5 it then follows that
Y is A-admissible if and only if (i)-(iii) are satisfied. 0
Remark 5.7. If in Corollary 5.6 Y is reflexive the Condition (iii) follows

from (i) and (ii). Indeed, for A, p. E peA) we have the resolvent identity
R(A: A) - R(p.: A) = (p. - A)R(A: A)R(p.: A)
which implies directly that D = R(A: A)Y is independent of A E peA).

From (5.7) with n = I it follows that for x E Y, AR(A: A)x is bounded in
Y as A -> 00. The reflexivity of Y then implies that there is a sequence
An -> 00 such A" R( An: A)x converges weakly in Y to some y E Y. Since A
is the infinitesimal generator of a Co semigroup on X, AR(A: A)x -> x
strongly in X as A -> 00 (Lemma 1.3.2) so y = x. Since for large values
A", A"R(A,,: A)x E D we conclude that the weak closure of D in Y is all of
Y. But the weak and strong closures of a linear subspace of a Banach space
are the same and so D is dense in Y.
We conclude this section with a useful criterion for a subspace Y of X to

be A-admissible.
Theorem 5.S. Let Y be the closure of Y in the norm of X. Let S be all

isomorphism of Y onto Y. Y is A -admissible if alld only if A I = SA S - I is the
infinitesimal generator of a Cu semigroup on Y. In this case we have in Y
where TI (t) is the semigroup generated by A I'

PROOF. Let A be the part of A in Y. From the definition of A I we have
D(A I ) = {x E Y: S-I X E D(A),AS-Ix E Y}
= {x E Y: S-I X E D(A)} = SD(A).
It follows that D( A I) is dense in Y if and only if D( A) is dense in Y.
Moreover for x E D( A I) we have

(AI - AI)x = (AI - SAS-I)x = S(AJ - A)S-I X = S(AI - A,)S-I X •
(5.8)
By assumption for A > w, R(A: A) is a bounded operator on X. We claim
that R (A : A I) exists as a bounded operator on f if and only if R (A : A) Y c Y
and then
R(A:A I ) = SR(A:A)S-I = SR(A:A)S-I (5.9)
in Y. Indeed, if R(A: A)Y C Y, SR(A: A)S-I is a bounded linear operator
on f which is the inverse of S(AJ - A)S-I and (5.9) follows from (5.8). On
the other hand if R(A: AI) exists in Y, S(AI - A)S-I is invertible and its
inverse SR(A: A)S-I is a bounded linear operator satisfying (5.9) and
therefore also S-IR(A: AI) = R(A : A)S-I which implies R(A: A)Y c Y.
Now if Al is the infinitesimal generator of a Co semigroup on f, D(AI) is
dense in f and therefore D(A) is dense in Y. Moreover for A> w R(A: AI)
exists and therefore by the first part of the proof R(A: A)Y c Yand (5.9)
holds. Theorem 1.5.3 then implies that A generates a Co semigroup on Yand
by Theorem 5.5, Y is A-admissible. On the other hand if Y is A-admissible,
R(A: A)Y c Y (Theorem 5.5) and by the first part of the proof (5.9) holds.
Since D(A) is then dense in Y, D(AI) is dense in f and Theorem 1.5.3
implies that AI is the infinitesimal generator of a Co semigroup on Y.
Finally, (5.9) together with the exponential formula (Theorem 1.8.3) imply
that TI (t) = ST( t )S- I and the proof is complete. 0
CHAPTER 5
Evolution Equations
5.1. Evolution Systems

Let X be a Banach space. For every I, 0 ~ 1 ~ T let A (I): D( A (t» c X --> X
be a linear operator in X and let f(l) be an X valued function. In this
chapter we will study the initial value problem
duj/) = A(I)u(t) + f(t) for s< t ~ T

{ ( 1.1)
u(s)=x.
The initial value problem (l.l) is called an evolution problem. An X valued

function u: [s, T] --> X is a classical solution of (l.l) if u is continuous on
[s, T], u(t) E D(A(t» for s < t ~ T, u is continuously differentiable on
s < t ~ T and satisfies (l.l).
The previous chapter was dedicated to the special case of (1.1) where
A(I) = A is independent of t. We saw that in this case, the solution of the
inhomogeneous initial value problem, i.e., the problem with f $- 0, can be
represented in terms of the solutions of the homogeneous initial value
problem via the formula of " variations of constants"
u(t) = T(I - s)u(s) + j'T(1 - T)f(T) dT (1.2)

s
where T(t)x is the solution of the initial value problem
du ( t) = Au ( t ) u(O) = x. ( 1.3)
dl '
We will see later that a similar result is also true when A(t) depends on t.
5 Evolution Equations 127
Therefore we concentrate at the beginning on the homogeneous initial value

problem:
du(t) = A(t)u(t) 05,s<t5,T

{ dt (1.4)
u(s) = x.
In order to obtain some feeling for the behavior of the solutions of (1.4)
we consider first the simple case where for 0 :s; t :s; T, A(t) is a bounded
linear operator on X and t --+ A(t) is continuous in the uniform operator
topology. For this case we have:
Theorem 5.1. Let X be a Banach space and for every t, 0 :s; t :s; T let A ( t) be
a bounded linear operator on X. If the function t --+ A(t) is continuous in the
uniform operator topology then for every x E X the initial value problem (1.4)
has a unique classical solution u.
PROOF. The proof of this theorem is standard using Picard's iterations
method. Let a = max O:5t:5T IIA(t)1I and define a mapping S from
C([s, T] : X) into itself by
(Su)(t) = x + fA(T)U(T) dT. (1.5)

s
Denoting Ilulloo = max s :5t:5T lIu(t)11 it is easy to check that
IISu(t) - Sv(t)1I :s; a(t - s)lIu - vll oo ' s 5, t 5, T. (1.6)

Using (1.5) and (1.6) it follows by induction that
an(t - sr
Ilsnu(t) - snv(t)1I :s; ,
n.
Ilu - vll oo ' s :s; t :s; T,
and therefore,
IISnu - snvlloo 5,
an(T - s
oo ,
r lIu - vll '
n.
For n large enough an(T - s)n/n! < 1 and by a well known generalization
of the Banach contraction principle, S has a unique fixed point u in
C([s, T]: X) for which
u(t) = x + fA( T)u( T) dT. (1.7)
s
Since u is continuous, the right hand side of (1.7) is differentiable. Thus u is
differentiable and its derivative, obtained by differentiating (1.7), satisfies
u'(t) = A(t)u(t). So, u is a solution of the initial value problem (1.4). Since
every solution of (1.4) is also a solution of (1.7), the solution of (1.4) is
unique. 0
We define the "solution operator" of the initial value problem (1.4) by

U(t,s)x=u(t) for O:s;s5,t5,T (1.8)
where u is the solution of (1.4). U(t, s) is a two parameter family of

operators. From the uniqueness of the solution of the initial value problem
(1.4) it follows readily that if A(t) = A is independent of t then U(t, s) =
U( t - s) and the two parameter family of operators reduces to the one
parameter family U( t), t ~ 0, which is of course the semigroup generated by
A. The main properties of U(t, s), in our special case where A(t) is a
bounded linear operator on X for 0 ~ t ~ T and t ~ A(t) is continuous in
the uniform operator topology, are given in the next theorem.
Theorem 5.2. For every 0 ~ s ~ t ~ T, U(t, s) is a bounded linear operator

and
(i) IIU(t, s)11 ~ exp(j/IIA(T)lldT).

(ii) U(t, t) = I, U(t, s) = U(t, r)U(r, s) for 0 ~ s ~ r ~ t ~ T.
(iii) (t, s) ~ U(t, s) is continuous in the uniform operator topology for 0 ~
s~t~T.
(iv) aU(t, s)/at = A(t)U(t, s)for 0 ~ s ~ t ~ T.
(v) aU(t, s)/as = - U(t, s)A(s) for 0 ~ s ~ t ~ T.
PROOF. Since the problem (1.4) is linear it is obvious that U(t, s) is a linear
operator defined on all of X. From (1.7) it follows that
Ilu(t)1I ~ Ilxll + [IIA(T)llllu(T)1I dT

s
which by Gronwall's inequality implies
II U(t, s)xll = Ilu(t)1I ~ Ilxll exp ({IIA( T)II dT) (l.9)
and so U(t, s) is bounded and satisfies (i).

From (1.8) it follows readily that U(t, t) = I and from the uniqueness of
the solution of (1.4) the relation U(t, s) = U(t, r)U(r, s) for 0 ~ s ~ r ~
t ~ T follows. Combining (i) and (ii), (iii) follows. Finally, from (1.7) and
(iii) it follows that U(t, s) is the unique solution of the integral equation
U(t, s) = 1+ [A( T)U( T, s) dT (1.10)

s
in B( X) (the space of all bounded linear operators on X). Differentiating

(1.1 0) with respect to t yields (iv). Differentiating (1.10) with respect to s we
find
a
asU(t,s)= -A(s)+ sA(T)asU(T,S)dT.
fl a (1.11)
From the uniqueness of the solution of (1.10) it follows that

a
asU(t,s)= -U(t,s)A(s) (1.12)

The two parameter family of operators U(t, s) replaces in the non-

autonomous case, i.e., in the case where A(t) depends on t, the one
parameter semigroup U(t) of the autonomous case. This motivates the
following definition.
Definition 5.3. A two parameter family of bounded linear operators U( t, s),

o :::; s :::; t :::; T,
on X is called an evolution system if the following two
conditions are satisfied:
(i) U(s, s) = I, U(t, r)U(r, s) = U(t, s) for 0:::; s :::; r:::; t :::; T.
(ii) (t, s) ~ U(t, s) is strongly continuous for 0:::; s :::; t :::; T.
Note that by analogy to the autonomous case, since we are not really
interested in uniform continuity of solutions, we have replaced the continu-
ity of U(t, s} in the uniform operator topology by strong continuity.
In the next sections we will give conditions on a given family of linear,
usually unbounded, operators {A(t}}, 0:::; t:::; T that guarantee the ex-
istence of a unique classical solution of the initial value problem
dUd~t) = A(t)u(t), u(s) = x (1.13)

for a dense set of initial values x E X. The existence of such a unique
solution will provide us with an evolution system associated with the family
{A(t}}, 0:::; t :::; T. The uniqueness of the solution of (1.13) will imply the
property (i) of evolution systems while the continuity of the solution at the
initial data will imply the property (ii). The relations between A(t) and
U(t, s) will be determined by some generalized versions of the equations
au~;s) =A(t)U(t,s) (1.14)
au~; s) = _ U(t, s )A(s). (1.15)
We conclude this section with a remark concerning the inhomogeneous

initial value problem (1.1) where f E LI(O, T: X). If there is an evolution
system U(t, s) associated with this initial value problem such that for every
v E D(A(s», U(t, s)v E D(A(t» and U(t, s)v is differentiable both in t
and s satisfying
a
at U(t, s)v = A(t)U(t, s)v (1.16)
a
as U( t , s ) v = - u( t, s ) A (s ) v (1.17)
then every classical solution u of (1.1) with x E D(A(s» is given by
u(t)= U(t, s)x + fU(t, r)f(r) dr. (1.18)

s
Indeed, in this case the function r -4 U(t, r)u(r) is differentiable on [s, T)

and
a
ar U(I, r}u(r} = - U(I, r}A(r}u(r} + U(I, r}A(r}u(r} + U(t, r}f(r)
= U(I, r}f(r). (1.19)
Integrating (1.19) from s to t yields (1.18). Thus, in this case, the inhomoge-
neous initial value problem (1.1) has at most one classical solution u which,
if it exists, is given by (1.18). However, for any evolution system U(t, s) and
f E LI(O, T: X) the right-hand side of (1.18) is a well defined continuous
function satisfying u(s) = x. As in the autonomous case (Section 4.5.2) we
will often consider this function as a generalized solution of the initial value
problem (1.1).
5.2. Stable Families of Generators

This section is devoted to some preliminaries that will be needed in the
construction of an evolution system for the initial value problem
{
du(t) = A(t)u(t)
dt
o~ s ~ t
(2.1 )
u(s} = x
in the "hyperbolic" case.
Definition 2.1. Let Xbe a Banach space. A family {A(I)}IE[O, T] of infinitesi-

mal generators of Co semi groups on X is called stable if there are constants
M ~ I and w (called the stability constants) such that
p(A(t}) ::l ]w, oo[ for t E [0, T] (2.2)
and
for A> W (2.3)
and every finite sequence 0 ~ II ~ 12" " , tk ~ T, k = 1,2, ....
Note that in general the operators R(A: A(t) do not commute and
therefore the order of terms in (2.3) is important. In (2.3) and in the sequel
products containing (t) will always be "time-ordered", i.e., a factor with a
larger I j stands to the left of ones with smaller I j .
From the definition of stability it is clear that the stability of a family of
infinitesimal generators {A(t)} is preserved when the norm in X is replaced
by an equivalent norm. The constants of stability however, depend on the
particular norm in X.
If for t E [0, Tl, A(t) E G(1, w), i.e., A(t) is the infinitesimal generator of
a Co seruigroup St(s), s ~ 0, satisfying IISt(s)11 ~ e WS then the family
{A(t)}tE[O. TJ is clearly stable with constants M = 1 and w. In particular any
family {A (t )}t E [0, TJ of infinitesimal generators of Co semigroups of contrac-
tions is stable.
Theorem 2.2. For t E [0, Tl let A(t) be the infinitesimal generator of a

Co semigroup St(s) on the Banach space X. The family of generators
{A(t)}tE[O, TJ is stable if and only if there are constants M ~ 1 and w such that
p(A(t» ::) lw, oo[ for t E [0, Tl and either one of the following conditions is
satisfied
(2.4)
and any finite sequence °~ tl ~ t2 ~ ... ~ tk ~ T, k = 1,2, ... or
for Aj> w (2.5)
and any finite sequence °~ tl ~ t2 ~ ... ~ tk ~ T, k = 1,2, ....

PROOF. From the statement of the theorem it is clear that it suffices to prove
that for a family {A(t)}tE[O, TJ of infinitesimal generators for which p(A(t»
::) lw, oo[ the estimates (2.3), (2.4) and (2.5) are equivalent.
Assume that (2.3) holds and let Sj' 1 ~ j ~ k be positive rational num-
bers. Let A = N be a positive integer such that NSj = mj is a positive integer
for 1 ~j ~ k. In (2.3) we take m = L)~lmj terms and subdivide them into
k subsets containing m j , 1 ~ j ~ k, terms. All values of t in the j-th subset
are taken to be equal to t j . After dividing both sides of the inequality by N m
we find
Letting N --> 00, such that Nsj , 1 ~ j ~ k, stay integers, each one of the mj
tends to infinity and by the exponential formula (Theorem 1.8.3) we obtain
and therefore (2.4) holds for all positive rationals Sj' The general case of
non-negative real Sj follows from the strong continuity of St(s) in s and thus
(2.3) implies (2.4).
In Chapter 1 we saw that
R(Aj:A(tJ)x =l°Oe-A,SSt (s)xds for Aj>W. (2.7)
o '
Iterating (2.7) a finite number of times yields
D R(Aj : A(tJx = {Xl .. '1000 exp { - j~1 AjSj } J] St/Sj)X dS I ... dsk·
(2.8)
Using (2.4) to estimate the norm of the right-hand side of (2.8) we find
Ii Ii
II J=I R(A j : A(tj))xll:::;; Mllxll J=I lOOe(W-~j)Sjdsj
0
= Mllxll Ii
J=I
(Aj - w r l
and therefore (2.4) implies (2.5). Finally, choosing all Aj equal to A in (2.5)
shows that (2.5) implies (2.3) and the proof is complete. 0
We have noted above that if {A(t)}tE[O. T] is a family of infinitesimal

generators satisfying A(t) E G(l, w) for t E [0, T] then it is a stable family.
In general however, it is not always easy to decide whether or not a given
family of infinitesimal generators is stable. The following perturbation
theorem is a useful criterion for this.
Theorem 2.3. Let {A(t)}tE[O. T] be a stable family of infinitesimal generators

with stability constants M and w. Let B(t), 0:::;; t :::;; T be bounded linear
operators on X. If IIB(t)11 :::;; K for all 0:::;; t :::;; T then {A(t) + B(t)}tE[O. T] is
a stable family of infinitesimal generators with stability constants M and
w+KM.
PROOF. From Theorem 3.l.l it follows that for every t E [0, T], A(t) + B(t)
is the infinitesimal generator of a Co semigroup. It is easy to check that if
A > w + KM then A is in the resolvent set of A(t) + B(t) and
00
R(A: A(t) + B(t)) = L R(A: A(t))[B(t)R(A: A(t))] n.
n=O
Therefore,
l~ R ( A : A ( tj) + B ( tj)) = J] (1l~0 R ( A : A ( tj) ) [ B ( tj) R ( A : A ( ti ) ) ] n) .

(2.9)
Expanding the right-hand side of (2.9) we find a series whose general term is
of the form
R(A: A(tk))[B(tk)R(A: A(tk ))] nk
... R(A: A(tl))[B(tl)R(A: A(tl))r'
where nj ~ O. If I:Y=
In j = n then estimating this term, using the stability of
the family {A(t)}tE[O.T]' yields the estimate Mn+IKn(A-W)-n-k. The
number of terms in which EJ~ 1n j = n in this series is (n ~ ~ ~ 1) and

therefore
Ill~R(A: A(t) + B(t)) 11:0; M(A - w)-k n~o (n Z~ ~ 1 )(MK(A - w)-l)n
= M (A - w - MK)-k
Let X and Y be Banach spaces and assume that Y is densely and
continuously embedded in X. ~et {A(t)}IE[O. T) be a stable family,?f infinites-
imal generators in X and let {A(t)}'E[O. T) be the family of p~rts A(t) of A(t)
in Y. Our last result gives a useful sufficient condition for {A(t)}IE[O. T) to be
stable in Y.
Theorem 2.4. Let Q(t), ° ::s; t :0; T, be a family of isomorphisms of Yonto X

with the following properties.
(i) IIQ(t)IIY~x and IIQ(t)-llIx~Y are uniformly bounded by a constant C.
(ii) The map t ~ Q(t) is of bounded variation in the B(Y, X) norm II . II Y~X'
Let {A(t)}'E[O. T) be a stable family of infinitesimal generators in X and let
AI(t) = Q(t)A(t)Q(t)-I. If {AI(t)}IE[O. T) is a stable family qf infinitesimal
generators in X then Y is A(t)-admissible for t E [0, T] and {A(t)}IE[O. T) is a
stable family of infinitesimal generators in Y.
PROOF. From Theorem 4.5.8 it follows readily that Y is A(t)-admissible for
every t E [0, T] and therefore by Theorem 4.5.5, A(t) the part of A(t) in Y
is the infinitesimal generator of a Co semi group in Y. From the definition of
A I(t) it follows that
D(AI(t)) = {x E X: Q(t)-I X E D(A(t)),A(t)Q(t)-l x E Y}
= {x E X: Q(t)-I X E D(A(t))} = Q(t)D(A(t))

and thereforeAI(t) = Q(t)A(t)Q(t)-I. This implies that for large enough
real >..
and thus,
k k
TI R(>..: A(tj )) = TI Q(tjrIR(>": AI(tJ)Q(tJ. (2.10)
j-I j~1
Setting lJ = (Q(t) - Q(tj_I»Q(tj_I)-1 the right-hand side of (2.10) be-

comes
Q(tk)-I{R(>..: AI(tk))(I + Pk )··· (I + P2 )R(>..: AI(tl))}Q(t l )·
(2.11 )
Let MI and WI be the stability constants of {AI(t)}'E[O. TI' Expanding the

expression in the curly bracket into a polynomial in the Pj and noting that
similarly to the proof of Theorem 2.3, only m + I factors of M, are needed
to estimate a term involving m of the~, we can estimate the X norm of this
expression by
n (I + M,II~II).
k
M I (;\ - WI)-k {2.12}
J~2
From the definition of Pj we have II~II :os; CIIQ(t) - Q(t;-,)IIY~x and

therefore
{2.13}
Denoting by V the total variation of t -> Q(t) in the B(Y, X) norm and
estimating the Y norm of (2.11) using (2.12) and (2.13) yields
IIl~ R(;\: A(tJL :os; C M 2

I (;\ - wl)-k
·exp {M,Cjt21IQ(tj) - Q(t;_,)11 Y~X}
< C 2M I eCM,V(;\ _ W I )-k

-
and thus {A(t)}'E[O. T] is stable in Y. o
5.3. An Evolution System in the Hyperbolic Case

This section is devoted to the construction of an evolution system for the
dUd~t) = A ( t ) U( t ) o :os; s :os; t :os; T

{ {3.1 }
u(s) = v
where the family {A(t )}'E[O. TI satisfies the conditions (HI )-( H 3 ) below. The
set of conditions (HI )-( H 3 ) is usually referred to as the" hyperbolic" case
in contrast to the" parabolic" case in which each of the operators A (t).
t ~ 0 is assumed to be the infinitesimal generator of an analytic semigroup.
The reason for these names lies in the different applications of the abstract
results to partial differential equations.
Let X and Y be Banach spaces with norms II II and II II Y respectively.
Throughout this section we will assume that Y is densely and continuously
imbedded in X, i.e., Y is a dense subspace of X and there is a constant C
such that
for WE Y. (3.2)
Let A be the infinitesimal generator of a Co semigroup S(s), s ~ 0, on X.
Recall (Definition 4.5.3) that Y is A-admissible if Y is an invariant subspace
of S( s) and the restriction S( s) of S( s) to Y is a Co semi group on Y.
Moreover, A the part of A in Y is, in this case, the infinitesimal generator of
the semigroup S( s) on Y.
For t E [0, T] let A(t) be the infinitesimal generator of a Co semigroup
SI(S), s ~ 0, on X. We will make the following assumptions.
(HI) {A(t)}IE[O,Tl is a stable family with stability constants M,., w.
(H2 ) Y is "!(t}-admissible for t E [0, Tj and the family {A(t)}IE[O, T) of
parts A(t) of A( t) in Y, is a stable family in Y with stability constants
M,w.
(H) For t E [0, Tj, D(A(t»::) Y, A(t) is a bounded operator from Y into
X and t -+ A(t) is continuous in the B(Y, X) norm II II Y-x.
The principal result of this section, Theorem 3.1, shows that if {A(t)}IE[O. Tl
°
satisfies the conditions (HI )-( H) then one can associate a unique evolution
system U(t, s), ~ s ~ t ~ T, with the initial value problem (3.1).
Theorem 3.1. Let A(t), °

~ t ~ T, be the infinitesimal generator of a Co
semigroup SI(S), s ~ 0, on X. If the family {A(t)}IE[O. T) satisfies the conditions
(HI )-( H 3 ) then there exists a unique evolution system U( t, s), ~ s ~ t ~ T,
in X satisfying
°
IIU(t,s}1I ~ Mexp{w(t - s}} for 0 ~ s ~ t ~ T.
a+
TtU(t, s}v I=s I = A(s}v for v E Y, 0 ~ s ~ T.
a for v Y, 0 s t T.
asU(t,s}v= -U(t,s}A(s}v E ~ ~ ~
Where the derivative from the right in (£2) and the derivative in (£3) are in
the strong sense in X.
PROOF. We start by approximating the family {A(t)}IE[O, T) by piecewise
constant families {A,,(t)}IE[O, T)' n = 1,2, ... , defined as follows: Let ti: =
(k/n)T, k = 0,1, ... , n and let
A,,(t} = A{t;:) k = 0, 1, ... , n - 1

{
A,,(T} = A(T} .
(3.3)
Since t -+ A(t) is continuous in the B(Y, X) norm it follows that
as n -+ 00 (3.4)
l36 Semi groups of Linear Operators
uniformly in t E [0, T]. From the definition of An(t) and the conditions of
the theorem it follows readily that !or n :2: I, {An(t)}IE[O, T] is a stable family
in X with ~onstants M, wand {An(t)}IE[O, T] is a stable family in Y with
constants M, W.
N ext we define a two parameter family of operators Un (t, s), 0 ::s; s ::s; t ::s;
Tby,
It is easy to verify that Un (t, s) is an evolution system, that is
for O::s; s ::s; r ::s; t ::s; T

(3.6)
and
(t, s) ~ Un(t, s) is strongly continuous on 0 ::s; s::s; t::s; T. (3.7)
From Theorem 2.2 it follows that
IIUn(t,s)II::S;Mew(l-s) for O::S;s::s;t::s;T (3.8)
and from (H2 ) we have
Un(t,s)YcY for O::s;s::s;t::s;T. (3.9)
Since D(A(t»:::J Y for t E [0, T], the definition of Un(t, s) implies that for
v E Y
a
aiUn(t,s)v=An(t)Un(t,s)v for tftj',j=O,I, ... ,n (3.10)
a
asUn(t,s)v= -Un(t,s)An(s)v for sftj,j=O,I, ... ,n.
(3.11)
Moreover, (H 2 ) together with Theorem 2.2 imply
IIUn(t,s)lly::s;Mew(l-s) for O::S;s::s;t::s;T. (3.12)
Let v E Y and consider the map r ~ Un(t, r)Um(r, s)v. From (3.10) and
(3.11) it follows that except for a finite number of values of r, this map is
differentiable in r, s ::s; r ::s; t, and
Un(t,S)V- Urn(t,S)V= - f arU,,(t,r)Um(r,s)vdr

S
a l
= tUn ( t, r ) ( An ( r) - Am (r ) ) Urn ( r, s ) v dr.

s
(3.l3)
Denoting y = max(w, w), (3.13) implies

II Un(t, s)v - Um(t, s )vlI .:s; MMey(t-s)llvll Y [IIAn(r) - Am(r )11 Y-+X dr.
s
(3.14)
From (3.14) and (3.4) it follows that u,,(t, s)v converges in X, uniformly on
o .:s; s .:s; t .:s; T, as n --+ 00. As Y is dense in X, this convergence of u,,(t, s)v
together with (3.8) imply that Un(t, s) converges strongly in X, uniformly on
o .:s; s .:s; t .:s; T, as n --+ 00. Let
U(t, s)x = lim u,,(t, s)x for x E X,O .:s; s .:s; t .:s; T. (3.15)
n --+ 00
From (3.6) and (3.7) it is clear that U(t, s) is an evolution system in X and
from (3.8) it follows that (E I) is satisfied.
To prove (E2 ) and (E3) consider the function r -+ Un(t, r)ST(r - s)v for
v E Y. This function is differentiable except for a finite number of values of
r and we have
Un(t, S)V - ST(t - S)V = - f ara Un(t, r )ST(r -

s
l
S)V dr
= [Un ( t, r ) ( An ( r) - A ( T )) ST (r - s ) v dr
s (3.16)
and therefore,
IIUn(t,s)v - ST(t - s)vll .:s; MMeY(I-S)lIvlly[IIAn(r) -A(T)lly-+x dr.
s
Passing to the limit as n --+ 00 this yields

II U(t, s}v - ST(t - s )vlI .:s; MMey(t-s)lIvll y[IIA(r) - A( T) II y-+x dr.
s (3.17)
Choosing T = s in (3.17), dividing it by t - s > 0 and letting t ~ s we find
lim sup _1_11 U(t, s)v - Ss(t - s)vll = 0 (3.18)
I~S t-s
where we used the continuity of t --+ A(t) in the B(Y, X) norm. Since
Ss(t - s)v is differentiable from the right at t = s, it follows from (3.18)
that so is U(t, s)v and that their derivatives from the right at t = s are the
same. This implies (E 2 ).
Choosing T = t in (3.17), dividing it by t - s > 0 and letting s t t we find
lim sup _1_11 U(t, s)v - SI(t - s )vll = 0 (3.19)

sil t-s
which implies, as above, that
a- U(t, s}v I
-a = -A(t)v. (3.20)
S s-t
For s < t, (E2 ) together with the strong continuity of U(t, s) in X imply
a+ 1
-a
s
U(t, s)v = lim -h {U(t, s + h)v - U(t, s)v}
h ~o
=!i~U(t,s+h){V- U(sh+h,s)V} = -U(t,s)A(s)v
(3.21 )
and for s ::; t we have by (3.20)
a- 1
-a u( t, s ) v = lim -h {U( t, s) v - U( t, s - h) v}
S h ~O
= !i~ U(t, s) { h
V- U(s S-h)V}
= - U(t, s )A(s )v.
(3.22)
Combining (3.21) and (3.22) shows that U(t, s) satisfies (E3)'
To complete the proof it remains to show that U(t, s), °: ;
s ::; t ::; Tis
the only evolution system satisfying (E 1 ), (E2 ), (E3)' Suppose V( t, s) is an
evolution system satisfying (EI)-(E3)' For v E Y consider the function
r -+ V(t, r)Un(r, s)v. Since V(t, s) satisfies (E3) it follows from the con-
struction of Un(t, s) that this function is differentiable except for a finite
number of values of r. Integrating its derivative yields
V( t, s) v - Un (t, s) v = [V( t, r)( A (r) - An (r)) 0" (r, s) v dr
s
and therefore,
IIV(t,s)v - Un(t,s)vll ::; MMeYU-S)lIvlly[IIA(r) -An(r)IIY~x dr.
s
(3.23)
Letting n -+ 00 in (3.23) and using (3.4) implies V(t, s)v = U(t, s)v for
v E Y. Since Y is dense in X and both U(t, s) and V(t, s) satisfy (E 1 ),
U( t, s) = V( t, s) and the proof is complete. D
The assumption that the family {A(t)}tE[O. T] satisfies (H2 ) is not always
easy to check. A sufficient condition for (H2 ) which can be effectively
checked in many applications is given in Theorem 2.4 above. It states that
(H2 ) holds if there is a family {Q(t)} of isomorphisms of Y onto X for
which II Q(t)1I y~X and II Q(t)-III x~ yare uniformly bounded and t -+ Q(t)
is of bounded variation in the B( Y, X) norm.
Remark 3.2. If condition (H3) in Theorem 3.2 is replaced by the weaker

condition:
(H3 ), For t E [0, T], D(A(t)) ~ Y and A(t) E LI(O, T: B(Y, X)) we can
still construct a unique evolution system U(t, s) for the initial value
problem (3.1). Indeed, if (H3)' is satisfied, there exists a sequence of

partitions (tn~~n? of [0, T] for which 8n= max{t k+ 1 - ti:) --+ 0 as n ~ 00
and the corresponding operators An(t), constructed as In the proof of
Theorem 3.1, satisfy
lim
n--+oo
l
0
T
IIA n (r) -A(r)lly--+xdr= O. (3.24)
Constructing Un(t, s) as in the proof of Theorem 3.1, replacing of course the

partition {(k/n)TYk~1 by the partition {tn~~n? it follows from (3.14)
together with (3.24) that Un(t, s)v converge uniformly on 0 ~ s ~ t ~ T to
U(t, s)v and thus U(t, s) exists and satisfies (E I ). Moreover, in this case,
(3.19) holds a.e. on [0, T] and hence we have
(E 2 )'
a+
TtU(t, I
s)v t~s = A(s)v for v E Yand a.e. on 0 ~ s ~ t ~ T
and similarly,
( E 3 )'
a
asU(t,s)v= -U(t,s)A(s)v for v E Yand a.e. on 0 ~ s ~ t ~ T.
The properties (E 2 ), and (E3)' together with (E I ) and the strong continu-
ity of U(t, s) suffice to ensure the uniqueness of U(t, s).
5.4. Regular Solutions in the Hyperbolic Case

Let X and Y be Banach spaces such that Y is densely and continuously
imbedded in X and let {A(t)}tE[O. T] be a family of infinitesimal generators of
Co semigroups on X satisfying the assumptions (HI)' (H2 ), (H3) of the
previous section. Let I E C([s, T] : X) and consider the initial value prob-
lem
dUd~t) = A(t)u(t) + I(t) for 0 ~ s ~ t ~ T

{ (4.1 )
u(s)=v.
A function u E C([ s, T] : X) is a classical solution of (4.1) if u is continu-

ously differentiable in X on ]s, T], u(t) E D(A(t)) for s < t ~ T and (4.1) is
satisfied in X. Unfortunately we do not know any simple conditions that
guarantee the existence of classical solutions of the initial value problem
(4.1) in the hyperbolic case even if I == O. In order to obtain classical
solutions of (4.1) under reasonable conditions, we will restrict ourselves in
this section to a rather strong and therefore quite restricted notion of
solutions of (4.1) namely the Y-valued solutions.
Definition 4.1. A function u E C([s, T]: Y) is a Y-valued solution of the

initial value problem (4.1) if u E CI(]s, T]: X) and (4.1) is satisfied in X.
A Y-valued solution u of (4.1) differs from a classical solution by

satisfying for s.::;; t .::;; T, u(t) EYe D(A(t» rather than only u(t) E
D(A(t» and by being continuous in the stronger Y-norm rather than
merely in the X-norm. For Y-valued solutions we have:
Theorem 4.2. Let {A(t)}tE[O,T] be a family of infinitesimal generators of Co

semigroups on X satisfying the condition (HI)' (H2 ), (H3) of Theorem 3.1 and
let f E C([s, T]: X). If the initial value problem (4.1) has a Y-valued solution
u then this solution is unique and moreover
u(t) = U(t, s)v + [U(t, r)f(r) dr (4.2)
s
where U(t, s) is the evolution system provided by Theorem 3.1.
PROOF. Let Un(t, s), 0 .::;; s .::;; t .::;; T be the evolution system constructed in
the proof of Theorem 3.1 (see (3.5» and let u be a Y-valued solution of (4.1).
From the properties of Un(t, s) and u it follows that the function r -+
Un(t, r)u(r) is continuously differentiable in X except for a finite number of
values of rand
a
ar Un(t, r)u(r) = - lJ,,(t, r)An(r)u(r)
+ Un(t, r)A(r)u(r) + Un(t, r)f(r). (4.3)

Integrating (4.3) from s to t we find
u(t) = Un(t, s)v + [Un(t, r)f(r) dr

s
+ [Un(t, r)(A(r) - An(r))u(r) dr. (4.4)

s
Denoting C = maxs:s;rS:Tllu(r)lly and using (3.8) to estimate (4.4) we find
lIu(t) - lJ,,(t, s)v - {Un(t, r )f(r) drll
.::;; Mew(t-s)C [IIA(r) - An(r )lIy .... xdr. (4.5)

s
Letting n -+ 00 in (4.5) and using (3.4) and (3.15) we find (4.2). The
uniqueness of u is a consequence of the representation (4.2). 0
We tum now to the problem of the existence of Y-valued solutions of the

homogeneous initial value problem
du(t) = A(t)u(t) for 0.::;; s < t.::;; T

{ dt (4.6)
u(s) = v.
From Theorem 4.2 it follows that if the family {A(t)}tE[O. TI satisfies the
conditions of Theorem 3.1 and the initial value problem (4.6) has a Y-valued
solution, this solution is given by u(t) = U(t, s)v where U(t, s), 0 ~ s ~
t ~ T, is the evolution system associated with the family {A(t)}tE[O. T] by
Theorem 3.1. In general however, u(t) = U(t, s)v is not a Y-valued solution
of (4.6) even if v E Y. The reason for this is twofold, Y need not be an
invariant subspace for U(t, s) and even if it is such an invariant subspace,
U(t, s)v for v E Y need not be continuous in the Y-norm. Both these
properties of U(t, s) are needed for u(t) = U(t, s)v to be a Y-valued
solution of (4.6). Our next result shows that they are also sufficient for this
purpose.
Theorem 4.3. Let {A(t)}tE[O. T] satisfy the conditions of Theorem 3.1 and let
U(t, s), 0 ~ s ~ t ~ T be the evolution system given in Theorem 3.1. If
( E4 ) U( t, s) Y C Y for 0~s ~ t ~ T
and
(E5) For v E Y, U(t, s)v is continuous in Y for 0 ~ s ~ t ~ T
then for every v E Y, U(t, s)v is the unique Y-valued solution of the initial
value problem (4.6).
PROOF. The uniqueness of Y-valued solutions of the initial value problem
(4.6) is an immediate consequence of Theorem 4.2. It suffices therefore to
prove that if v E Y then u(t) = U(t, s)v is a Y-valued solution of (4.6).
From (E4) and (E5) it follows that u(t) E Y for s ~ t ~ T and that it is
continuous in the Y-norm for s ~ t ~ T. To complete the proof it remains
to show that u satisfies the differential equation in (4.6). Since u(t) =
U(t, s)v E Y for s ~ t ~ T we have by (E 2 ) that
a+ U( ) -1' U(t + h, s)v - U(t, s)v
-at
t, s v - 1m
h,J.O
h
= lim U(t +~' t) - I U(t, s)v = A(t)U(t, s)v. (4.7)

h,J.O
The right-hand side of (4.7) is continuous in X since t ~ U(t, s)v is

continuous in the Y-norm and t ~ A(t) is continuous in B(Y, X). There-
fore, the right-derivative of U(t, s)v is continuous in X and as a conse-
quence U(t, s)v is continuously differentiable in X and by (4.7)
a
at U(t, s)v = A(t)U(t, s)v for s ~ t ~ T. o
From Theorem 4.3 it follows that if U(t, s), the evolution system given by
Theorem 3.1 also satisfies (E4) and (E5) then for every v E Y the initial
value problem (4.6) has a unique Y-valued solution given by U(t, s)v. In
order to get an evolution system U(t, s) that satisfies (E))-(E5) we will
replace the condition (H2 ) of Theorem 3.1 by the following condition:
(H2 ) + There is a family {Q(t)}'E[O, T] of isomorphisms of Y onto X such that

for every v E Y, Q(t)v is continuously differentiable in X on [0, T] and
Q(t)A(t)Q(t)-1 = A(t) + B(t) (4.8)
where B(t), O:$; t:$; T, is a strongly continuous family of bounded opera-
tors on X.
In the proof of our main result, Theorem 4.6, we will need the following
two technical results.
Lemma 4.4. The conditions (HI) and (H2 )+ imply the condition (H 2 ).
PROOF. From (H2 )+ it follows that for every v E Y, t ---> dQ(t)v/dt is
continuous in X on [0, T] and therefore IIdQ(t)/dtll y~X is bounded on
[0, T]. This implies that t ---> Q(t) is Lipschitz continuous and hence of
bounded variation on [0, T] in the B(Y, X) norm and II Q(t)11 y~X is
bounded on [0, T]. The Lipschitz continuity of t ---> Q(t) in B(Y, X) also
implies the continuity of t ---> Q( t) - I in B( X, Y) and therefore
IIQ(t)-'llx~Y is bounded on [0, T]. Since by (HI) {A(t)}'E[O,T] is stable in
X it follows from Theorem 2.3 that {A(t) + B(t)}tE[O, T] is a stable family in
X. From Theorem 2.4 it then follows that Y is A(t)-admissible for every
t E [0, T] and {A(t)}tE[O, T] is a stable family in Y. 0
° °
Lemma 4.5. Let U( t, s), :$; s :$; t :$; T be an evolution system in a Banach
space X satisfying II U(t, s) II :$; M for :$; s :$; t :$; T. If H( t) is a strongly
°
continuous family of bounded linear operators in X then there exists a unique
family of bounded linear operators V( t, s), :$; s :$; t :$; T on X such that
V(t,s)x = U(t,s)x + {V(t,r)H(r)U(r,s)xdr for x E X

s
and V(t, s)x is continuous in s, t for ° :$; s :$; t :$; T.

(4.9)
PROOF. Let V(O)(t, s) = U(t, s) and define

v(m)(t, s)x = {v(m-I)(t, r)H(r)U(r, s)xdr for x E X.
s
(4.10)
°
The integrand in (4.10) is continuous on :$; s :$; r :$; t :$; T as is easily seen
°
by induction on m. From the uniform boundedness principle it follows that
there is an H > such that IIH(t)11 :$; H for t E [0, T] and by induction on
m one verifies easily the estimate
IIv(m)(t,s)ll :$; Mm+IHm(t - ;)m

m.
The series
00
Vet,s) = L v(m)(t,s) (4.11 )

m~O
therefore converges in the uniform operator topology on X and V(t, s) thus

defined, is strongly continuous on 0 :$ s :$ t :$ T. Moreover it follows from
(4.10) and (4.11) that V( t, s) satisfies (4.9). To complete the proof it remains
to prove the uniquness of V(t, s). Let VI(t, s) satisfy (4.9) and set W(t, s)
V(t, s) - VI(t, s) then
W(t,s)x={W(t,r)H(r)U(r,s)xdr for xEX. (4.12)

s
Estimating (4.12) yields
IIW(t,s)xll:$ MH{II W (t,r)xlldr for x E X

s
which by Gronwall's inequality implies W(t, s)x = 0 for 0 :$ s :$ t :$ T and
xE X whence V(t, s) = VI(t, s) and the proof is complete. 0
The main result of this section is:
Theorem 4.6. Let A(t), 0:$ t :$ T be the infinitesimal generator of a Co

semigroup on X. If the family {A(t)}IE[O. TI satisfies the conditions (HI)'
( H 2) + and ( H 3 ) then there exists a unique evolution system U(t, s), 0 :$ s :$
I :$ T, in X salisfying (E 1)-( E 5 ).
PROOF. From Lemma 4.4 it follows that {A(t)}IE[O. TJ satisfies the conditions
(HI)' (H 2 ), (H 3 ) and therefore, by Theorem 3.1, there exists a unique
evolution system U(/, s) satisfying (EI )-(E3)'
Let v E Y and denote the derivative of Q(t)v by Q(t)v. Set
C(t) = Q(t)Q(t)-I. (4.13)
C( I), 0 :$ t :$ T, is clearly a strongly continuous family of bounded opera-
tors on X. Let W( t, s) be the unique solution of the integral equation
W(t, s)x = U(t, s)x + {W(t, r)[B(r) + C(r)]U(r, s)xdr
.<
for xEX. (4.14)

The existence, uniqueness and properties of W( t, s) follow from Lemma
4.5. Below we will prove
U(t,s) = Q(t)-IW(t,S)Q(s). (4.15)
From (4.15) it follows that U(t,s)Yc Y since W(t,s)EB(X). Thus
U(t, s) satisfies (E4)' Moreover, from the continuity of W(t. s)x on 0 :$
s :$ I :$ T and the properties of Q( s) and Q(t) - I it follows that U( t, s) is
strongly continuous in Y for 0 :$ s :$ t :$ T and therefore satisfies (Es)·
We turn now to the proof of (4.15). First we note that from our
assumptions on Q(t) it follows easily that for every x E X Q( t) - IX is
differentiable in Y and
(4.16)
Set
Q(t, r) = U(t, r)Q(r)-I. ( 4.17)
From (E3) and (4.16) it follows that for every x E X, r ~ Q(t, r)x is
differentiable in X and
;r Q(t, r)x = - U(t, r)A(r)Q(r)-lx - U(t, r)Q(r)-IQ(r)Q(r)-lx
= - U(t, r)A(r)Q(r)-'x - Q(t, r)C(r)x.

But for every v E Y we have by (H 2 )+
A(r)Q(r)-lv = Q(r)-I(A(r) + B(r))v

and therefore for v E Y
8
8r Q(t, r)v = -Q(t, r)[A(r) + B(r) + C(r)]v. (4.18)
Let v,,(t, s) be the operators constructed in the proof of Theorem 3.1 (see
(3.5» then by (3.10)
for v E Y (4.19)
where (4.19) holds for all s::s; r except for a finite number of values of r.
Combining (4.18) and (4.19) we find
8
8r Q(t, r)Un(r,s)v
= - Q(t, r )(A(r) + B(r)+ C(r) - An(r ))Un(r, s )v. (4.20)
Integrating (4.20) from r = s to r = t yields
Q(t)-IUn(t,S)v - Q(t,s)v
= - [Q(t, r )(A(r) + B(r)+ C(r) - An(r ))Un(r, s)v dr. (4.21)

s
From (E I ) and (3.12) we deduce
II{Q(t, r)(A(r) - An(r))Un(r, S)Vdrll
I
::s; MMey(t-s) sup Q-l(r) Ilx~yllvllY[II A(r) - An(r) IIY~x dr (4.22)
r s
where y = max (w, w). Passing to the limit as n ~ 00 in (4.21) and using
(4.22) and (3.15) we obtain for v E Y
Q( t) - I U( t, s) v - Q( t, s) v = - [Q( t, r)( B( r) + C( r)) U( r, s) v dr.

s
(4.23)
Sin"e all operators in (4.23) are bounded in X and since Y is dense in X,

(4.23) holds for every v E X and hence after rearrangement we have
Q(t,s)x = Q(t)-IU(t,S)x + [Q(t,r){B(r) + C(r))U{r,s)xdr.
s
(4.24)
On the other hand, multiplying (4.14) from the left by Q(t)-I yields
Q(t)-IW(t, s)x = Q(t)-IU(t, s)x + [Q(t)-IW(t, r){B(r)
s
+C(r))U(r, s)xdr. (4.25)

From (4.24) and the uniqueness of the solution of (4.25) it follows that
U(t,s)Q(S)-1 = Q(t,s) = Q(t)-IW(t,S)
which implies (4.15) and the proof is complete. o
From Theorems 4.6 and 4.3 we obtain,
Corollary 4.7. Let {A(t)tE[O, TJ be a family of infinitesimal generators of Co

semigroups on X. If {A(t)tE[O, TJ satisfies the conditions (HI)' (H2 ) + and (H3 )
then for every v E Y the initial value problem
dUd~t) = A (t ) u(t ) for s < t :::; T

{ (4.26)
u(s) = v
has a unique Y-valued solution u on s :::; t :::; T.
One special case in which the conditions of Theorem 4.6 can be easily
verified is the case where D(A(t)) = D is independent of t. In this case we
define on D a norm II II y by
Ilvll y = IIvll + IIA(O)vll for v E Y = D (4.27)
and it is not difficult to see, using the closedness of A(O), that D equipped
with this norm is a Banach space which we denote by Y. This Y is clearly
densely and continuously imbedded in X and we have:
Theorem 4.8. Let {A(t)tE[O. TJ be a stable family of infinitesimal generators of

Co semigroups on X. If D(A(t)) = D is independent of t and for v E D, A(t)v
is continuously differentiable in X then there exists a unique evolution system
U( t, s), 0 :::; s :::; t :::; T, satisfying (E 1)-( E 5 ) where Y is D equipped with the
norm II II y given by (4.27).
PROOF. We will show that {A(t)}tE[O. TJ satisfies the conditions (HI)' (H2 )+
and (H3)' Condition (HI) is explicitly assumed in our theorem. The
continuous differentiability of A(t)v in X clearly implies that t ---> A(t) is
continuous in the B(Y, X) norm so (H3) is satisfied. To prove (H2 )+ note

that for Ao > w the operator Q(t) = Aol - A(t) is an isomorphism of Y
onto X and by our assumption on A(t)u it follows that Q(t)v is continu-
ously differentiable in X for every v E Y. Finally,
Q(t)A(t)Q(t)-1 = A(t)
and therefore (4.8) is satisfied with B(t) == 0, so (H2 )+ holds and the proof
is complete. 0
5.5. The Inhomogeneous Equation in

the Hyperbolic Case
This section is devoted to a few remarks concerning the solutions of the
inhomogeneous initial value problem
du~t) = A(t)u(t) + f(t) for 0::;; s < t ::;; T

{ (5.1)
u(s) = v
in the hyperbolic case. In Section 5.3 we have considered the corresponding

homogeneous initial value problem and under the assumptions (HI)' (H2 ),
(H3 ) we have constructed (Theorem 3.1) a unique evolution system U(t, s),
o ::;; s ::;; t ::;; T, satisfying the properties (EI)-(E3)' Motivated by the auton-
omous case (see Section 4.2) we make the following definition.
Definition 5.1. Let {A(t)}/E[O..Tl satisfy the conditions of Theorem 3.l.and

let U(t, s), 0::;; s ::;; t ::;; T be the evolution system given by Theorem 3.l.
For every f E LI(S, T: X) and v E X the continuous function
u(t) = U(t, s)v + [U(t, r)f(r) dr (5.2)
s
is called the mild solution of the initial value problem (5.1).
From the concluding remarks of Section 5.1 it follows that if the

evolution system U(t, s) is regular enough and f E CI([s, T]: X) then the
initial value problem (5.1) has a unique classical solution for every v E
D(A(s» and this solution coincides with the mild solution (5.2). A similar
result (Theorem 4.2) holds for Y-valued solutions of (5.1).
Existence of Y-valued solutions for the inhomogeneous initial value
problem is provided by:
Theorem 5.2. Let {A(t)}/E[o.Tl satisfy the condition of Theorem 4.3. If

f E C([ s, T]: Y) then for every v E Y the initial value problem (5.1) pos-
sesses a unique Y-valued solution u given by (5.2).
PROOF. It has been shown in Theorem 4.3 that U( t, s) V 1S a Y-valued

solution of the homogeneous initial value problem
du ( t) = A ( t ) u ( t ) for 0 ~ s < t ~ T
{ dt (5.3)
u{s) = v.
To prove that u given by (5.2) is a Y-valued solution of (5.1) we will show
that
w{ t) = {U( t, r) f{ r ) dr (5.4)
s
is a Y-valued solution of (5.1) with the initial value w(s) = v = O. From our
assumptions on f and (£4) it follows readily that w( t) E Y for s ~ t ~ T.
From (£5) it follows that r ~ U(t, r)f(r) is continuous in Y which implies
that t ~ w(t) is continuous in Yand that r ~ A(t)U(t, r)f(r) is continu-
ous in X for s ~ t ~ T. The continuity of r ~ A(t)U(t, r)f(r) implies that
w( t) is continuously differentiable in X and that
d
dt w{t) = A{t)w{t) + f{t) for s ~ t ~ T
holds in X as desired. Finally, the uniqueness of Y-valued solutions of (5.1)

is a direct consequence of Theorem 4.2. 0
Theorem 5.2 shows that if the family {A(t)};E[O. TJ of infinitesimal genera-

tors of Co semigroups on X satisfies the conditions (HI)' (H 2) + and (H3 )
then for every v E y and f E C([s, Tj: Y) the initial value problem 5.1
possesses a unique Y-valued solution u given by (5.2). This result is
reminiscent of Corollary 4.2.6.
Our next result, for the special case where all the operators A (t),
o ~ t ~ T, have a common domain D independent of t is reminiscent of
Corollary 4.2.5.
Theorem 5.3. Let {A( t )}CE[O. TJ be a stable family of infinitesimal generators of

Co semi groups on X such that D( A (t» = D is independent of t and for every
v E D, A(t)v is continuously differentiable in X. Iff E CI([s, Tj: X) then for
every v E D the initial value problem (5.1) has a unique classical solution u
given by
u(t) = U{t, s)v + {U(t, r)f{r) dr. (5.5)
s
PROOF. As in Theorem 4.8 we endow D with the graph norm of A(O) and
denote this Banach space by Y. From our assumptions it then follows that
for ;\.0 large enough and every t E [0, Tj, Q(t) = ;\.01 - A(t) is an isomor-
phism of Y onto X such that Q( t) v is continuously differentiable in X for
every v E Y. We denote the derivative of Q(t)v by Q(t)v and note that
Q(O E B(Y, X) and that {IIQ(Olly .... x} is uniformly bounded. From Theo-
rem 4.8 it follows that vet, s)v is the Y-valued solution of the homoge-
neous initial value problem (5.3). To show that u given by (5.5) is a
classical solution of (5.1) it is, therefore sufficient to show that
w(t)= [V(t,r)f(r)dr
s
is a classical solution of (5.1) satisfying w(s) = O. To this end we note first

that Q(r)-If(r) is differentiable in Yand that
:r(Q(r)-lf(r)) = -Q(r)-IQ(r)Q(r)-lf(r) + Q(r)-If'(r)
=Q(r)-lg(r) (5.6)
where f'(r) is the derivative of fer) and g(r) = f'(r) - Q(r)Q(r)-lf(r).
Differentiating Vet, r)Q(r)-lf(r) with respect to r using (E3) and (5.6)
we find
:r Vet, r)Q(r)-lf(r)
= - Vet, r)A(r)Q(r)-lf(r) + Vet, r)Q(r)-lg(r)

= Vet, r)f(r) + Vet, r)Q(r)-I(g(r) - Aof(r)).
Integrating this equality from r = s to r = t we obtain after rearrangement
wet) = Q(t)-If(t)
- [V(t, s)Q(S)-lf(S) + [vet, r)Q(r)-I(g(r) - Aof(r)) dr]
= Q(t)-If(t) - vet) (5.7)

where v(t) is defined by the second equality of (5.7). Since Q(s)-If(s) E Y
and r ---> Q(r)-I(g(r) - Aof(r)) is continuous in Yon [s, T] it follows
from Theorem 5.2 that
dVd~t) = A(t)v(t) + Q(t)-I(g(t) - Aof(t)) for 0 ~ s ~ t ~ T.
(5.8)
Therefore, using (5.7) we have
dw(t) = !!"-(Q(t)-If(t)) _ dv(t)

dt dt dt
= Q(t)-Ig(t) - Q(t)-I(g(t) -Aof(t)) -A(t)v(t)
= A(t)w(t) + AoQ(t)-lf(t) - A(t)Q(t)-lf(t)
= A(t)w(t) + f(t).
Since dv(t)/dt and Q(t)-Ig(t) are continuous in Xit follows that dw(t)/dt
5 Evolution Equatiom 149
is continuous in X and w is a classical solution of (5.1) with v = O. To prove

the uniqueness of the classical solution u, let v, be a classical solution of
(5.1). From our assumptions and the properties of U( t, s) (see Theorem 4.8)
it follows that' -> U(t, ,) v, (,) is continuously differentiable in X and that
a,a U(t, ,)v,(,) = U(t, ,)f(,)·

Integrating this equality from s to t yields v, (t) = u(t). o
5.6. An Evolution System for the Parabolic Initial

Value Problem
This section starts the second part of Chapter 5 in which we study the initial
value problem
{
du d/) + A ( t ) u ( t) = f ( t ) (6.1 )
u(s) = x
in the parabolic case.' The results of this part are independent of the results
of Sections 5.2-5.5 in which the corresponding hyperbolic case was treated.
The evolution system for the parabolic initial value problem
{
dUd~t) + A ( t ) U( t) = 0 O~s<t~T
(6.2)
u(s) = x
will be constructed below by a method which is entirely different from the
method used in the hyperbolic case (Section 5.3). The present section is
devoted to this construction and to the study of the main properties of the
resulting evolution system.
We start with a formal computation that will lead us to the method of
construction of the evolution system. Suppose that for each t E [0, Tj,
- A(t) is the infinitesimal generator of a Co semigroup SI(S), s z 0 on the
r
Banach space X and let U( t, s) be an evolution system for (6.2). Set
U ( t, s) = Ss (t - s) + W ( t , s) = Ss (t - s) + ST (t - 7") R ( 7", S) d 7".

s
( 6.3)
Then (formally)
ata U( t, s) il
= - A (s ) Ss (t - s) + R (t, s) - s A ( 7" ) ST (t - 7") R ( 7", s) d7"
I In the parabolic case it is customary to write the term A(t)u(t) on the left-hand side of the
equation. This is done to overcome some notational difficulties related to the use of fractional
powers of A (I).
and
a
-a U(t,s)+A(t)U(t,s)=R(t,s)-RI(t,s)- ft RI(t,r)R(r,s)dT
t s
( 6.4)
where
RI{t, s) = (A(s) - A{t))Ss{t - s). (6.5)
Since U(t, s) is an evolution system for (6.2) it follows from (6.4) that the
integral equation
R(t,s)=RI(t,s)+ [RI(t,r)R(r,s)dT (6.6)
s
must be satisfied. Given RI(t, s) we can try to reverse the argument, i.e.,
solve the integral equation (6.6) for R(t, s) and then define U(t, s) by
(6.3). This will indeed be the method of constructing U(t, s) below. To carry
out this program rigorously we will need the following assumptions:
(PI) The domain D(A(t» = D of A(t), O:s t :S T is dense in X and
independent of t.
°
(P2 ) For t E [0, T], the resolvent R(A: A(t» of A(t) exists for all A with
Re A :S and there is a constant M such that
M
IIR(A: A(t))11 :S IAI + 1 for Re A :S 0, t E [0, T]. (6.7)
(P3 ) There exist constants Land °< a :S 1 such that
II(A(t)-A(s))A(r)-11i :sLlt-sl a for s,t,rE[O,T].

(6.8)
We note that the interval for which (PI )-( P3 ) are satisfied was chosen as
[0, T] only for notational convenience. All the results that will be proven
°
below, hold for any interval [a, b], :S a < b < 00, on which these assump-
tions are satisfied.
The main result of this section is:
°
Theorem 6.1. Under the assumptions (P I)-(P3 ) there is a unique evolution
system U( t, s) on :S s :S t :S T, satisfying:
(E I ), II U(t, s)11 :S C for O:s s :S t :S T.
(E2 )+ For O:s s < t :S T, U(t, s): X -+ D and t -+ U(t, s) is strongly
°
differentiable in X. The derivative (a/at)u(t, s) E B(X) and it is
strongly continuous on :S s < t :S T. Moreover,
a
at U(t, s) + A(t)U(t, s) = 0 for O:s s < t:S T, (6.9)
II %t U(t, s)11 = IIA{t)U(t, s)11 :S t:- s (6.10)

and
IIA(t)U(t, s)A(S)-111 :S C for O:s s :S t :S T. ( 6.11)
to s on °:; ;
(E3)+ For every v E D and t E]O, T], U(t, s)v is differentiable with respect
s :;;; t :;;; T and
a
as U(t, s)v = U(t, s )A(s )v. ( 6.12)
The proof of Theorem 6.1 will occupy most of this section. It will be
divided into three main parts. In the first part we will construct U(t, s) by
solving the integral equation (6.6) and using (6.3) to define U(t, s). In the
second part we will prove that U(t, s) satisfies the properties stated in
(E 2 )+ and in the third part the uniqueness of U(t, s) together with
U(t, s) = U(t, r)U(r, s), for °:; ;
s :;;; r :;;; t:;;; T and (E3)+ will be proved.
Before starting with the proof we derive some direct consequences of the
assumptions (PI)-(P3)' First we note that (P2 ) and the fact that D is dense
in X imply that for every t E [0, T], - A(t) is the infinitesimal generator of
an analytic semigroup S/(s), s ~ 0, satisfying (see Theorem 2.5.2)
IIS/(s)1I :;;; C for s ~ ° (6.13)
IIA(t)S/(s)1I :;;; C
s
for s> ° (6.l4)
where here, and in the sequel, we denote by C a generic constant. From (P2 )
it also follows that there exists an angle 1'Jo E ]0, 'TT12[ such that
p(A(t)) :::> }: = {A: larg AI ~ 1'Jo} U {O} (6.15)
and that (6.7) holds for all A E }:, possibly with a different constant M.
Some more consequences of the assumptions (P I )-(P3 ) are collected in
the next lemma.
Lemma 6.2. Let (P I )-(P3 ) be satisfied then

C
II(A(t l ) - A(t 2))ST(S)11 :;;; -It I - t 21a
s
for s E]O, T], t l , t2 E [0, T] (6.16)
C
IIA(t)(ST(S2) - ST(sl))11 :;;; -IS2 - Sll
SlS2
for °< Sl' S2 E ]0, T], t, 7 E [0, T] (6.17)
C
IIA(t)(ST(S)-ST(S))1I
I 2
:;;;-17
S 1 - 72I a
for s E]O, T], t, 7 1 , 72 E [0, T]. (6.18)
Moreover, A(t)ST(S) E B(X) for s E]O, T], 7, t E [0, T] and the B(X)
valued function A(t)ST(S) is uniformly continuous in the uniform operator
topology for s E fe, T], t,7 E [0, T] for every e > 0.
PROOF. Since ST(S): X ~ D for s > °

(Theorem 2.5.2) it follows from the
closed graph theorem that A(t)ST(S) is a bounded linear operator for
t, r E [0, Tj, s EjO, Tj. From (6.8) and (6.14) we have
II ( A ( t I) - A ( t 2) ) ST (s ) II :s; II ( A ( t I) - A ( t 2» A ( r ) - I II II A ( r ) ST ( S) II
:s; C It2 - tda
° SI
S
which proves (6.16). To prove (6.17) let < :s; S2 and x E X. From the
results of Section 2.5 we have
and therefore by (6.14)
Since by (6.8), IIA(t)A( r)-III :s; C for t, r E [0, Tj, we obtain
IIA(t)(ST(SI) - ST(s2»)1I :s; IIA(t)A(r)-IIIIIA(r}{ST(sl) - ST(S2»11

C
:s; - I s - s I
S2 s 1 2 I
which proves (6.17).

To prove (6.18) note that from (P2) it follows that
IIA(t)R(A:A(t»11 :s; C for O:s; t:s; T
and therefore
IIA(t}{R(A: A( r l ») - R(A: A( r2»))11
= IIA(t)R(A:A(rl»(A(r l ) -A(r2»)R(A:A(r2»1I
:s; IIA(t)A(rl)-IIIIIAh)R(A:A(rl»)11
. II(A(r l ) -A(r2 »)A(r2)-IIIIIA(r2 )R(A:A(r2»1I

:s; Cirl - r2la • (6.19)
From the results of Section 2.5 we get the following representation

A(t)ST'(S)X - A(t)ST2(S)X
1 . (e-ASA(t)(R(A:A(r l ») -R(A:A(r2»))xdA
=-2 (6.20)
'fI'l Jr
where f is a smooth path in ~ connecting ooe- dJ to ooe ,lJ • Using (6.19) to

estimate (6.20) we find
IIA{t)ST,{S)X - A{t)ST,{S)xll ~ CiT, - T21"llxlI! le-AslldAI

T
whence (6.18).
Finally, to prove the uniform continuity of A(t)ST(S) in the uniform
operator topology for t, T E [0, Tj, s E [f, Tj we have only to combine
(6.16), (6.17), (6.18) and use the triangle inequality. 0
Corollary 6.3. The operator R1(t, s) = (A(s) - A(t»Ss(t - s) is uniformly

continuous in the uniform operator topology on 0 ~ s ~ t - f ~ T for every
f > 0 and
for 0 ~ s < t ~ T. (6.21 )

PROOF. The first part of the claim is a direct consequence of the uniform
continuity of A(t)ST(S) in B(X) while (6.21) follows from
IIR1{t,s)ll ~ II{A{t) -A{s))A{s)-'IIIIA{s)Ss{t - s)1I

o
We are now ready to start the construction of U( t, s).
I. Construction of the Evolution System
We begin by solving the integral equation (6.6) for R(t,s). If Rt(t, s)

satisfies (6.21) then (6.6) can be solved by successive approximations as
follows: For m z I we define inductively
Rm+l{t,s)= [Rt{t,T)Rm{T,S)dT. ( 6.22)

s
Then we prove by induction that Rm(t, s) is continuous in the uniform

operator topology for 0 ~ s < t ~ T and that
{ (Cf{a))m{ )ma-I
IIRmt,s)ll~ r{ma) t-s (6.23 )
where f(·) is the classical gamma function. In the inductive proof of (6.23)
we use the well known identity
j '{.t-T )"-I{ T-S )p-I d

s
T=
( t-s )a+p-I f(a)f(,B)
f(a+,B)
(6.24)
which holds for every a, {3 > O. We note that the integral defining Rm+ l(t, s)
is an improper integral whose existence is an immediate consequence of
(6.23). The continuity of Rm+1(s, t) also follows easily from the continuity
of Rm{t, s), R1(t, s) and (6.23).
The estimates (6.23) imply that the series
00
R{t, s) = L Rm{t, s)
m=l
converges uniformly in the uniform operator topology for 0 ~ s ~ t - E ~ T
and every E > O. As a consequence R(t, s) is uniformly continuous in B(X)
for 0 ~ s ~ t - E ~ T and every E > O.
00 00
R(t,s)= L Rm(t,s)=R1{t,s)+ L [R1{t,r)Rm{r,s)dr.

m=l m=l S
(6.25)
The continuity of Rm(t, s), m ~ 1, (6.21) and (6.23) imply that one can
interchange the summation and integral in (6.25) and thus see that R(t, s) is
a solution of the integral equation (6.6). Moreover, using Stirling's formula
we have
00
IIR{t, s)11 ~ L r{ma)-I(Cr(a»m(t - s)ma-l

n=1
~ C~l r(ma)-I(Cr{a»mTa(m-l) )(t - st- 1
~ C{t - st- 1• (6.26)
Defining U(t, s) by (6.3) it follows readily from the strong continuity of

Ss( r), (6.13) and (6.26) that U(t, s) is strongly continuous for 0 ~ s ~ t ~ T
and that
II U{t, s)ll ~ IISs{t - s)11 + [IIST{t - T)IIIIR( r, s)ll dT
s
~ C1 + l
C2 l{ r - s )a-l
s
~ c. (6.27)
Therefore (E I)' is satisfied. In order to show that U(t, s), 0 ~ s ~ t ~ T is

an evolution system it remains to show that U(t, s) = U(t, r)U(r, s) for
s ~ r ~ t. This will follow from the uniqueness of the solution of the initial
value problem (6.2) that will be proved below (Theorem 6.8), and the fact
that by (E2 )+ the solution of (6.2) is U(t, s)x.
II. Differentiability of U( t, s).
We turn now to the proof that U(t, s), constructed above, has the properties
stated in (£2)+. For this we need a few preliminaries.
Lemma 6.4. For every P, 0 < P :s; IX, there is a constant Cp such that
IIRt(t,s) - RI(T,s)1I :s; Cp(t - T)P(T - s),,-p-t
for O:s; s < T < t :s; T. (6.28)
PROOF. We have
R t ( t , s) - R I (-r, s) = (A ( T) - A ( t ) ) Ss (t - s)
+ (A ( s) - A ( T) )( S, (t - s) - Ss ( T - s».
II(A(T) -A(t))Ss(t - s)1I :s; C(t - T)"(t - s)-t
:s; C(t - T)"(T - S)-I.

Also,
II(A(s) - A( T»(Ss(t - s) - Ss( T - s»)11

:s; II ( A (s) - A ( T) ) A (s ) - I II . II A (s )( S, (t - s) - Ss ( T - s» II.
Estimating the right-hand side of the last inequality using (6.8) and (6.17)
we find that it is bounded by C( T - S),,-2(t - T) while estimating it using
(6.8) and (6.14) we find that it is bounded by C(T - S),,-I. Therefore,
II(A(s) -A(T»(S,(t - s) - Ss(T - s»11
:s; C[(T - s)"-\t - T)]"[(T - s)"-T-a
= C(t - T)"(T - srI

and thus
IIRI(t, s) - R I( T, s)1I :s; C(t - T)"( T - S)-I.
On the other hand we have by (6.21)
IIRt(t,s)-Rt(T,S)II:S; IIRt(t,s)1I + IIRI(T,S)II
:s; C((t - S)"-I + (T - S)"-I):s; C(T - s)a-I.
Interpolating the two estimates for II R I (t, s) - R I ( T, s) II we find
IIRI(t, s) - R t ( T, s)1I :s; C[(t - T)"( T - S)-I]P/"[( T - s)"-T-p/a
o
Corollary 6.5. For every {3, 0 < {3 < a, there is a constant Cf3 such that
II R ( t , s) - R ( T, S) II .::; Cp (t - T) P ( T - S )'" - p- 1
for 0.::; s < T < t .::; T. (6.29)
PROOF. From the integral equation (6.6) we have
R(t, s) - R( T, s) = R1(t, s) - R 1( T, s) + [R1(t, o)R(o, s) do

T
+ j"( R 1( t , 0) - R 1( T, 0 ) ) R ( 0, s ) do.
s
The estimates (6.21) and (6.26) imply
IlfR1(t, o)R(o, s) doll

.::; C [(t - 0)"'-1(0 - S)"'-I do
T
.::; C( T - S)"'-Ij\t - 0)",-1 do

T
.::; C(T - S)",-I(t - T)"''::; C(T - S)"'-P-I(t - T)P

while (6.28), (6.26) and (6.24) imply
lit( R 1( t, 0) - R 1( T, 0») R ( 0 , s ) do I
.::; C(t - T)P {(T - O)"'-P-I(O - S)"'-I do
s
.::; C(t - T)P(T - s)2a- p -I.::; C(t - T)P(T - S)"'-P-I.
The estimate (6.29) is now an immediate consequence of (6.28) and the two
last inequalities. D
Lemma 6.6. For every x E X we have

lim S/ (e)x = x uniformly in 0.::; t .::; T. (6.30)
,-+0
PROOF. For xED we have
Therefore,
Ilx - S/(e)xll .::; {IISt(O)IIIIA(t)A(O)-IIIIIA(O)Xlldo.::; eCIIA(O)xll
and (6.30) holds for every xED. Since D is dense in X and IISt(s)1I .::; C
the result for every x E X follows by approximation. D
We turn now to prove the differentiability of U(t, s). Since Ss(t - s) is

differentiable for t > sand (J/Jt)Ss(t - s) = -A(s)Ss(t - s) is a bounded
linear operator which is continuous in B( X) for t > s it suffices to prove the

differentiability of W(t, s). To this end we set
W.(t, s) = [-'STet - T)R( T, s) dT
s
for 0 < E < t - s. (6.31)
As E -+ 0, W,(t, s) -+ Wet, s). Moreover, w.(t, s) is differentiable in t and

a W.(t, s)
at = St_,(E)R(1 - E, s) -., jl-'A( T)ST(I - T)R( T, s) dT.
(6.32)
Using the equality A(t)St(t - T) = (a/aT)St(t - T) we can rewrite the last
equation as
a
at W.(t, s) = St_,(E)R(t - E, s)
+ [-'(A(/)St(t - T) - A( T)ST(t - T»)R( T, s) dT

,.
+ [-'A (t ) St (t - T)( R (I, s) - R ( T, s» dT
s
+ (St(1 - s) - St(E))R(/, s). (6.33)

From (6.13) and (6.26) it follows that the first and the last terms on the
right-hand side of (6.33) are bounded in norm by C(t - s - E)a-I while
from (6.16) and (6.18) we deduce easily that
IIA(/)St(1 - T) -A(T)ST(I - T)II ~ C(I - Tt- I
and therefore,
II { -,( A ( I ) St (I - T) - A ( T) ST (I - T») R ( T, S ) d Til
~ C [-E(I _ T)a-I( T - s)a-I dT ~ C(I _ s)2a-1

s
~ C(t - S t(t - S - Et- t ~ C(I - s - Et- I.

Finally, from (6.14) and (6.29) we have
11{-'A(/)St(1 - T)(R(I, s) - R( T, s» dT11
~ C[-'(t - T)P-t(T - s)a-P-l dT

s
(
~CI-s ) a-I (
~CI-S-E
)a-I .
Combining these estimates we find
II : I w. (I , s ) I ~ (t-S-E)
C I- a (6.34)
where C is a constant which is also independent of e > O. Letting e -+ 0 on

the right-hand side of (6.33) and using Lemma 6.6 we see that (a/ at)w.(t, s)
converges strongly as e -+ o. Denoting its limit by W'(t, s) we have,
W'(/, s) = SI(I - s)R(/, s)
+ [( A ( I ) SI (I - 'T) - A ( 'T ) ST (I - 'T») R ( 'T, s) d'T
s
+ [A(/)SI(I - 'T )(R(t, s) - R( 'T, s)) d'T (6.35)

s
which implies that W'(/, s) is strongly continuous for 0 ~ s < I ~ T. Pass-

ing to the limit as e -+ 0 in (6.34) yields moreover
(6.36)
Now, letting e -+ 0 in
W.(/ 2, s) - W.(/I' s) = {2 :'T w.( 'T, s) d'T

yields
W(/2' s) - W(/I' s) = [2 W'( 'T, s) d'T

I,
where 12 > II > s + e. Since W'( I, s) is strongly continuous for 0 ~ s < I ~

T, it follows that W(/, s) is strongly continuously differentiable with respect
to I and that
%IW(/,S) = W'(/,S).
Therefore, U(t, s) is strongly continuously differentiable,
a
al U(/, s) = -A(s)Ss(1 - s) +
a
al W(/, s),
and by (6.14) and (6.36)
Setting
U.(t,s) =Ss(t-s) + w.(t,s) for /3 > 0, t - s > 0,
it follows readily that lJ.(t, s): X -+ D and by (6.31), (6.32),
a
at lJ.(/, s) + A(/)lJ.(/, s)
= SI_,(e}R(1 - e, s} - RI(/, s) - [-'RI(t, 'T }R( 'T, s} d'T. (6.37)
s
Passing to the limit as e -+ 0, the right hand side of (6.37) tends strongly to
zero. Since (a/at)lJ.(/, s) -+ (a/at)u(t, s) strongly, it follows from (6.37)
that A(/)lJ.(/, s) converges strongly as e -+ O. Let x E X, the closedness of
A(t) together with U.(t, s)x -+ U(t, s)x imply that U(t, s)x E D and that
A(t)U.(t, s)x -+ A(t)U(t, s)x. Thus passing to the strong limit as f -+ 0 in
(6.37) yields
a
at U(t, s) + A(t)U(t, s) = 0 for t > s.
This concludes the proofs of (6.9) and (6.10).

To prove (6.11) we will need:
Lemma 6.7. Let cp(t, s) ~ 0 be continuous on 0 ~ s< t ~ T. If there are

positive constants A, B, 0' such that
cp(t, s) ~ A + B [(t - or-1cp(0, s) do for 0 ~ s< t ~ T
.\.
(6.38)
then there is a constant C such that cp( t, s) ~ C for 0 ~ s< t ~ T.
PROOF. Iterating (6.38) n - I times using the identity (6.24) and estimating
t - s by T we find
"~I(BTa)j (Bf(O'»)"f'( )"a-I ( )

cp ( t,s ) ~Aj':O ~ + [(nO') s t-o cp a,s do.
Choosing n sufficiently large so that nO' > I and estimating (t - o)na-I by

T"a-I we get
cp(t, s) ~ c 1 + c2 [cp(0, s) do
s
which by Gronwall's inequality implies cp(t, s) ~ c1ecz(t-S) ~ c1e czT ~ C.

Since c 1 and C 2 do not depend on s this estimate holds for 0 ~ s < t ~ T. 0
We turn now to the proof of (6.11). Let x E X and consider the function
!/I(s) = S,(t - slUes, T)A(T)-I X for 0 ~ T < S < t ~ T. It is easy to see
that !/I is differentiable with respect to s and that
!/I'(s) = S,(t - s)[A(t) -A(s)]U(S,T)A(T)-I X.

Integrating !/I' from T to t and applying A (t) to the result we find
Z(t, T)X = A(t)S,(t - T)A(T)-I X + {Y(t, s)Z(s, T)xds

T
(6.39)
where
Y( t , s) = A (t ) S, (t - s)[ A ( I) - A (s )] A (s ) - I
and
From (6.8) and (6.13) we have
IIA(t)St(t - r)A(r)-111 = IISt(t - r)A(t)A(r)-11i
~ IISt(t - r)IIIIA(t)A(r)-111 ~ CI
and from (6.8) and (6.14),
IIY(t,s)11 ~ IIA(t)St(t - s)IIII(A(t) -A(s))A(s)-111 ~ C2 (t - S)"-I.

Estimating now (6.39) we find
IIZ(t, r )xll ~ Clllxli + c2 J\t - s ),,-IIIZ(s, r )xlI ds

T
which implies by Lemma 6.7, IIZ(t, r)xll ~ CIIxll whence
IIZ(t,r)11 = IIA(t)U(t,r)A(r)-11i ~ C
as desired. This completes the proof of (E 2 ) +.
III. Uniqueness
The uniqueness of the evolution system U(t, s) satisfying (E I )" (E 2 )+ and
(E3)+ will be a simple consequence of (E3)+ as we will see below.
We start by proving (E3)+ under the supplementary assumption that for
every v E D, A(t)v is continuously differentiable on [0, T]. This assumption
and the uniform boundedness theorem imply that (a/at)A(t)A(O)-l =
A'(t)A(O)-l is uniformly bounded on [0, T]. It also implies that for every
A E I, R(A : A(s» is differentiable with respect to s and that
J
JSR(A: A(s)) = R(A: A(s ))A'(s )R(A: A(s )). (6.40)
From (6.7) and (6.40) we deduce that
for AE };. (6.41)
The assumptions (PI) and (P2 ) imply (see Section 2.5) that
Ss(t - s) = 1 . (e-A(t-S)R(A: A(s)) dA

-2
7Tzlr
where r is a smooth path in }; connecting ooe- i " to ooe i ". From our
supplementary assumption it now follows that if t - s > 0 then Ss (t - s) is
strongly differentiable in sand
a (t
-s
as S
1.
- s) = -1. Ae-A(t-s)R(A: A(s)) dA
27Tl r
+ _1_. (e-A(I-S)~R(A: A(s)) dA

27Tl Jr as
= -~S(t-s)+_1_. (e-A(t-S)~R(A:A(s))dA.
at S 2m Jr as
To prove (£3)+ we construct an operator valued function V(t, s) satisfying
{:s V(t, s)v

V(t, t) = I
= V(t, s)A(s)v for 0::; s ::; t ::; T, v E D
(6.42)
and prove later that V(t, s) = U(t, s). The construction of V(t, s) follows
the same lines as the construction of U( t, s) above. We set
Q I ( t, s ) -- (~
at + ~)S( 1 (-A(I-S)~
as S t - s ) -_27TiJ/
_ as R ('"1\. A ())d"
s 1\.
Using (6.41) and estimating QI(t, s) as in the proof of Theorem 1.7.7 we

find
IIQI(t,s)1I =112~ilre-A(I-S) :sR(A:A(S))dAII::; c.

N ext we solve by successive approximations the integral equation
Q(t, s) = QI (t , s) + [Q (t, T ) QI ( T, s) d T. (6.43)

S
This is done in exactly the same way as the solution of the integral equation
(6.6). Since in this case QI(t, s) is uniformly bounded the solution Q(t, s) of
(6.43) will satisfy
IIQ(t, s)11 ::; C.
Setting
V( t, s) = Ss (t - s) + [Q ( t, T) Ss ( T - s) d T
s
we find that II V(t, s)11 ::; C and for v E D, V(t, s)v is differentiable in s.
Differentiating V( t, s) v with respect to s yields
a
as V(t, s)v - V(t, s)A(s)v
= QJt,s)v + [Q(t,T)QI(T,S)vdT-Q(t,S)v = O.
s
From the definition of V(t, s) it follows that V(t, t) = I and so V(t, s) is a

solution of (6.42).
For x E X and s < r < t the function r --+ V(t, r)U(r, s)x is differentia-
ble in rand
a
ar V(t, r)U(r, s)x = V(t, r)A(r)U(r, s)x - v(t, r)A(r)U(r, s)x = O.
This shows that V(t, r)U(r, s)x is independent of r for s < r < t. Letting
r t sand r i t we find V(t, s)x = U(t, s)x for every x E X. Therefore
U(t, s) = V(t, s) and U(t, s) satisfies
a
as U(t, s)v = U(t, s )A(s)v for v ED (6.44)
as desired.
We continue by showing the validity of (6.44) in general, that is, without
assuming the continuous differentiability of A(t)A(O)-1 which was assumed
above. To do so we approximate A(t) by a sequence of operators An(t) for
which An(t)An(O)-1 is continuously differentiable. This is done as follows:
Let p(t);::: 0 be a continuously differentiable real valued function on R
satisfying p(t) = 0 for It I ;::: 1 and f~oop(t) dt = 1. Let Pn(t) = np(nt) and
extend A(t) to all of R by defining A(t) = A(O) for t < 0 and A(t) = A(T)
for t ;::: T. Let v E D and set
An(t)v = J'X)-00
Pn(t - (J)A«(J)vd(J = foo
-00
Pn«(J)A(t - (J)vd(J.
An(t)v thus defined is continuously differentiable on [0, T]. We will now

show that An(t) satisfy the conditions (P I )-(P3 ).
By defintion we have D(An(t» = D and therefore (PI) is satisfied. For
X E ~ we have
x - (X - An(t»R(X: A(t»x = - (A(t) - An(t»)R(X: A(t»x
= foo Pn(t - r )(A( r) - A(t»R(X: A(t»x dr.

-00
For It - rl ::;; ljn, (6.7) and (6.8) imply

II ( A ( t) - A(r » (X :
R A ( t ) ) II ::;; Cn - a .
Therefore,
IIx - (X - An(t»R(X: A(t»xlI ::;; Cn-allxll (6.45)
and in particular taking X = 0 we have
II (A(t) - An(t »A(t) -III ::;; Cn-a. ( 6.46)
From (6.45) it follows easily that for v E D, we have
(1 - Cn-a)II(X - A(t»vlI ::;; II(X - An(t»)vlI
::;; (1 + Cn-a)II(X - A(t»vlI (6.47)
and therefore if n is sufficiently large so that Cn-a < 1 and X E ~,
'AI - AI/(t) is closed, R('AJ - An(t» = X and 'AI - An(t) has a bounded
inverse R ('A: An (t» satisfying
M
IIR('A: AI/(t))11 ::; I'AI + 1 for 'A E L
and so (P2) is satisfied.
Choosing n in (6.46) such that Cn-a < 1 we obtain
00
A(t)AI/(t)-l = L [1 - AI/(t)A(t)-lr
k~O
and IIA(t)AI/(t)-lll ::; C. From the definition of AI/(t) and (6.8) it follows
that
JJ(An(t) -An(s))A(lT)-luJJ
::; JOO PnCr)JJ(A(t - IT) -A(s - IT))A(T)-luJkr::; Cit - sl<>llull

-00
therefore
II(An(t) - An(s))AnCr)-lll
::; II(An(t)-An(s))A(T)-lll·IIA(T)An(T)-lll::; qt-sl"
and so (P3 ) is also satisfied by An (t).
From the first part of the proof it follows that there is an operator valued
function U,,(t, s) satisfying II Un(t, s)11 ::; C, where C is independent of n,
and
for 0::; s < t ::; T.
Since An(t)v is continuously differentiable in t for v E D, it follows that
for v E D. ( 6.48)
From (6.48) and the properties of U(t, s) it follows that for every v E D the
function r --> Un(t, r)U(r, s)v is differentiable and
U(t,s)v- Un(t,s)v= 1 ar{U,,(t,r)U(r,s)v}dr

s
a 1
= {Un(t,r)[An(r)-A(r)]U(r,s)vdr
s
= {Un(t, r)[An(r) - A(r)]A(r)-lA(r)

s
x U(r,s)A(s)-lA(s)udr. (6.49)
Using (6.46) and (6.11) to estimate (6.49) we find
II U{t, s)v - Un(t, S )vll ::; Cn-"(t - s) IIA(s )vll ::; Cn-aIIA(O)vll
and therefore Un (t, s) V --> U( t, s) V uniformly in t and s. Since D is dense in
X it follows that Vn(t, s)x --+ V(t, s)x uniformly in t and s for every x E X.
For v E D we have
II Vn(t, s)An(s)v - V(t, s)A(s)vlI
:s; II U,,(t, s )(An(s)v - A(s)v)1I + II( Vn{t, s) - V(t, s »)A(s) vii
:s; Cn-"IIA(O)vll + II(Vn(t, s) - V(t, s»)A(s)vll
and therefore Vn(t, s)An(s)v --+ V(t, s)A(s)v uniformly in t and s. For
r < s < t and v E D we have
a
Vn(t,s)v- U,,(t,r)v= 1, aoVn(t,o)vdo=
s
1, Vn(t,o)An(o)vdo s
which in the limit as n --+ 00 yields
V(t,s)v - V(t, r)v = {V(t,o)A(o)vdo

r
and therefore (6.44) holds in general. This concludes the proof of (E)+. To
conclude the proof of Theorem 6.1 we still have to show the uniqueness of
Vet, s) and that it satisfies V(t, s) = Vet, r)V(r, s) for 0 :s; s :s; r :s; t :s; T.
Both these claims follow from:
Theorem 6.S. Let A(t), 0 :s; t:s; T satisfy the conditions (p\)-(p). For every
o :s; s <
T and x E X the initial value problem
dUd~t) + A(t)u(t) = O. s<t:s;T

{ ( 6.50)
u(s) = x
has a unique solution u given by u(t) = V(t, s)x where Vet, s) is the evolution
system constructed above.
PROOF. From (6.9) it follows readily that u(t) = Vet, s)x is a solution of
(6.50). To prove its uniqueness let v(t) be a solution of (6.50). Since for
every r> s, vCr) ED, it follows from (6.12) and (6.50) that the function
r --+ Vet, r)v(r) is differentiable and
a
ar V(t, r)v(r) = V(t, r)A(r)v(r) - V(t, r)A(r)v(r) = O.
Therefore V(t, r)v(r) is constant on s < r < t. Since it is also continuous

on s :s; r:s; t we can let r --+ t and r --+ S to obtain V(t, s)x = vet) and the
uniqueness of the solution of (6.50) follows. 0
From Theorem 6.8 it follows readily that for x E X,

V(t, s)x = V(t, r)V(r, s)x for O:s; s:s; I:S; T
and V(t, s) is therefore an evolution system satisfying (E I )" (E 2 )+ and
( E) +. If V( t, s) is an evolution system satisfying (E I)' and (E 2 ) + then
V(/, s)x is a solution of (6.50) and from Theorem 6.8 it follows that
v(t, s)x = U(t, s)x and so V(t, s) = U(t, s) and U(t, s) is the unique
evolution system satisfying (E])', (E 2 )+ and (E3)+. This concludes the
proof of Theorem 6.1. 0
°
In Theorem 6.1 we proved that -(a/at)U(t, s) = A(t)U(t, s) is strongly
continuous on ~ s < t ~ T. Much more is actually true. Indeed we have:
E °
Theorem 6.9. Let the assumptions of Theorem 6.1 be satisfied. Then for every
°
> the map t -> A(t)U(t, s) is Holder continuous, with exponent (3 < a, in
the uniform operator topology, for < s + E ~ t ~ T.
PROOF. We recall that
U( t, s) = Ss (t - s) + W( t, s).
Since (a/at)Ss(t - s) = A(s)Ss(t - s) is Lipschitz continuous in t for
° °
t E [s + E, T] it remains to show that (a/at)W(t, s) is Holder continuous
as claimed. We fix E > and assume that ~ s < S + E ~ 7 ~ t ~ T. From
(6.35) we have
aW(t, s) aw( 7, s)
at at
= [s/ (t - s) R ( t, s) - ST ( 7 - S) R ( 7, S )]
+ [(A(t)S/(t - a) - A( a )S,,(t - a ))R( a, s) doo

T
+ [( A ( t ) S/ (t - a) - A ( a ) s" (t - a) - A ( 7 ) ST ( 7 - a)
s
+A(oo)S,,(7 - oo))R(oo, s) doo
+ [A(t)S/(t - oo)(R(t, s) - R(oo, s)) doo

T
+ [[A(t)S/(t - oo)(R(t, s) - R(oo, s))

s
- A ( 7) ST ( 7 - 00)( R ( 7, s) - R (a, s))] doo

= I] + 12 + 13 + 14 + 15 •
We will now estimate each one of the terms Ii' 1 ~ j ~ 5 separately. The
generic constants appearing in these estimates will usually depend on the
E > 0, chosen above.
II IIiI ~ II (S/ (t - s) - S/ ( 7 - s)) R (t, s) II

+ II(S/(7 - s) - ST(7 - s))R(t,s)11
+ II ST ( 7 - s)( R ( t, s) - R ( 7, S ) ) II
~ C] (t - 7) + C2 (t - 7 r + C3 (t - 7 r ~ C( t - 7 r·
Here we used the Lipschitz continuity of S/(s) for s > 0, (6.18) and
Corollary 6.5 with a = (3. The second term is estimated as follows.

111211 ~ [11(A(/)St{1 - a) -A(a)Sa(1 - a))R(a, s)11 da
i
T
~ C t( 1 - a ) a - I(
a - s) a - I da
T
~ C[(I - ar-Ida ~ C(I - Tr

T
where we used (6.26) and

IIA(/)St(1 - a) -A(T)ST(I - a)11 ~ C(I - Tr{1 - a)-I (6.51)
which is a simple consequence of (6.16) and (6.18).
To estimate 13 we note first that from (6.18) and (6.14) we have
IIA(a)Sa(P)-A(T)ST(p)1i ~ C(T-ar
P
and therefore,
IIA(a) 2Sa(P) -A(T)2 ST (p)11
~ IIA (a ) Sa ( I)( A( a) Sa ( I)- A( ST ( I))IIT)
+ II ( A ( a ) Sa ( I)- A ( T) ST ( I))A ( T) ST ( I)II
~~(T-ar. (6.52)
P
We rewrite the integrand of 13 as follows,
[ A ( 1) St (I - a) - A ( a ) Sa (I - a) - A ( T ) ST ( T - a)
+ A ( a ) Sa ( T - a)] R ( a, s)
= [A (a )( Sa ( T - a) - Sa (I - a)) - A ( T)( ST ( T - a) - ST (I - a))
+ A ( 1) St (I - a) - A ( T ) ST (I - a)] R ( a, s)
= [~~-aa(A(a)2Sa(P) -A(T)2ST(P)) dp + A(/)St(1 - a)
- A ( T ) ST (I - a)] R ( a, s ).
Estimating this integrand using (6.52), (6.51) and (6.26) we find for 0 <
{3<a
111311 ~ C[[(I - T)(T - ar-I(I - a)-I
s
+ (I - Tr(1 - a)-I](a - sr-Ida

~C(/-T)
{3i (T-a)
T (a-{3)-1
(a-s) a-I da~C(/-T) {3
s
where we used that 1 - T ~ 1 - a and 1 - a ~ T - a and therefore for

For /4 we have
11/411 :s; [IIA(t)SI(t - a)IIIIR(t, s) - R(a, s)lI da
T
:s; C[(t - a)-I(t - a)"(a - s)-I da

T
:s; C [(t - a)"-I da:s; C(t - T)".

T
Here we used (6.l4) and Corollary 6.5. Finally to estimate /5 we rewrite its
integrand as follows
[A(t)SI(t - a)(R(t, s) - R(a, s»
a)( R ( T , s) - R ( a, s
- A ( T ) ST ( T - »]
= A(t)SI(t - a)(R(t, s) - R( T, s»
+ [A ( t ) SI (t - a) - A ( T ) ST (t - a)
+ A ( T ) ( ST (t - a) - ST ( T - a»] ( R ( T, s) - R ( a, s ) )
Taking {3 < y < a and estimating the integrand of /5 using Corollary 6.5,
(6.14), (6.51) and (6.17) we find
lI/sll:s; C[(t - a)-I{(t - Tr(T - s),,-.B- I
s
+ [(t - T)Y + (T - a)-I(t - T)](T - a)Y(a - s)"-.B- I} da

:s; C(t - T)Y + C(t - Tr[( T - a)"-I(a - s)(,,-.Bl- I da
s
+c(t - T).B j\T - a)(Y-.Bl-l(a - s)(,,-.Bl- I da

s
:s; C(t - T).B

5.7. The Inhomogeneous Equation in

the Parabolic Case
Let {A(t)}IE[O. T] satisfy the conditions (PI)-(P3) of the previous section and
consider the inhomogeneous initial value problem
dUd~t) + A(t)u(t) = J(t) for O:s; s < t :s; T

{ (7.1)
u(s) = x
in the Banach space X. From Theorem 6.1 it follows that there is a unique
evolution system U(t, s), 0::;; s ::;; t ::;; T, associated with the family
{A(t)}tE[O. TJ' As we have already done in the autonomous and hyperbolic
cases, the continuous function
u(t}= U(t,s}x+ [U(t,a}f(a}da (7.2)

s
withf E Ll(S, T: X) and x E X will be called the mild solution of the initial
value problem (7.1). Thus for every f E Ll(S, T: X) and x E X the initial
value problem (7.1) possesses a unique mild solution given by (7.2). In this
section we will be interested in classical solutions of (7.1) i.e., functions
u :[s, T]---+ X which are continuous for s ::;; t ::;; T, continuously differentia-
ble for s < t::;; T, u(t) ED for s < t ::;; T, u(s) = x and u'(t) + A(t)u(t) =
f(t) holds for s < t ::;; T. We will call a function u a solution of the initial
value problem (7.1) if it is a classical solution of this problem. In order to
obtain (classical) solutions of (7.1) we will have to impose further conditions
on the function f. The situation here is completely analogous to the
situation in the autonomous case with - A being the infinitesimal generator
of an analytic semigroup (see Corollary 4.3.3).
Theorem 7.1. Let {A(t)}tE[O. TJ satisfy the conditions (Pl)-(P3) of section 5.6
and let U( t, s) be the evolution system provided by Theorem 6.1. If f is Holder
continuous on [s, T] then the initial value problem (7.1) has, for every x E X,
a unique solution u given by;
u(t}= U(t,s}x+ [U(t,a}f(a}da.

s
PROOF. From Theorem 6.8 it follows that U(t, s)x is the unique solution of
the homogeneous initial value problem
du(t}
-;Jt + A(t}u(t} = 0, u(s}=x. (7.3)
To prove the existence of a solution of the initial value problem (7.1) it is

therefore sufficient to show that
v(t}= [U(t,a}f(a}da
s
is a solution of the initial value problem
{
dVd~t} + A(t}v(t} = f(t} (7.4)
v(s} = 0.
To this end we set
ve(t} = [-'U(t, a}f(a} da. (7.5)
s
From Theorem 6.1 it follows that vE(t) is differentiable in t and

a a jl-E
atVE(t)=7ii U(t,a)f(a)da
s
= U(t, t - e)f(t - e) - r-EA(t)U(t, a)f(a) da

s
and therefore
a
7iiVE(t) + A(t)vE(t) = U(t, t - e)f(t - e). (7.6)
The proof will be concluded by letting e ! 0 in (7.6). In order to justify this
passage to the limit, we will first show that (a/at) V.( t) has a limit as e ! O.
From the results of Section 6 we have
U( t, s) = Ss (t - s) + W( t, s) (7.7)
where S/ s) is the analytic semigroup generated by - A (t) and W( t, s) is
strongly continuously differentiable in t for 0 ::; s < t ::; T (see proof of
Theorem 6.1) and
:t [-'W(t, a )f(a) da = W(t, t - e)f(t - e)
+ js
l-' a
7ii W(t,a)f(a)da.
Moreover, W(t, t) = 0 and (see (6.36)) II(a/at)W(t, s)11 ::; C(t - S),,-I.
Therefore,
lim aa jl-'W(t, a)f(a) da = jl aa W(t, a)f(a) da.

dO t s s t
Next,
:t [-'So(t - a )f( a) da = SI_E(e)f(t - e) + [-' :t So(t - a )f( a) da
= St_,(e)f(t - e)
+ r-'[A(t)St(t - a) -A(a)S,,(t - a)]f(a) da
s
- r-'A(t)SJt - a)f(a) da. (7.8)

s
To show the existence of the limit as e ! 0, we treat each of the three terms
on the right-hand side of (7.8) separately.
From Lemma 6.6 it follows that
limSt_,(e)f(t - e) = f(t).
dO
The assumptions (PI )-( P3 ) imply (see Lemma 6.2) that
IIA(t)St(t - a) - A(a)So(t - a)11 ::; C(t - a),,-I
170 Sernigroups of Linear Operators
and therefore,
limjt-'[A(t)St(t - 0) - A( 0 )S,,(t - 0)] f(o) do

dO s
= t[A(t)St(t - 0) -A(o)S,,(t - o)]f(o) do.

s
Finally, for the last term we have
t-'A(t)St(t - o)f(o) do
s
= t-'A(t)St(t - o)(f(o) - f(t» do + t-fA(t)St(t - o)f(t) do

s s
= t-'A(t)St(t - o)(f(o) - f(t» do - [St(t - s) - St(e)]f(t).

s
(7.9)
The Holder continuity of f and the estimate IIA(t)St{t - 0)11 ~ C(t - 0 )-1
imply that the integrand on the first term on the right-hand side of (7.9) is
integrable and therefore
limjt-'A(t)St(t - o)f(o) do = jtA(t)St(t - o)(f(o) - f(t» do

,~o s s
-St(t - s)f(t) + f(t).

Combining the last results we conclude that (a/at)!st-'S,,{t - o)f(o) do
converges as dO and since we saw that (a/at)!.r-'w(t, o)f(o) do con-
verges as e~O it follows that lim,~o(a/at)v,(t) exists. Reviewing the proof
it is not difficult to see that this limit is uniform on [s + e, T] for every
e > O.
Returning to (7.6) we can now conclude that A (t) v.( t) converges as e ! 0
and since by (7.5) v,(t) -+ v(t) as e ~O it follows from the closedness of A(t)
that v(t) E D(A(t» = D and A(t)v,(t) -+ A(t)v(t) for s < t ~ T. In-
tegrating (7.6) from t to t + h we have
v,(t+h)-v,(t)= - f t+hA(T)v,(T)dT+ ft+h U(T,T-e)f(T-e)dT.

t t
(7.10)
Passing to the limit as e ~ 0 yields
v(t+h)-v(t)= - f t+hA(T)v(T)dT+ ft+h f(T)dT. (7.11)

t t
Finally, dividing (7.11) by h and letting h!O we find that v(t) satisfies (7.4).
The uniqueness of the solution of the initial value problem (7.1) is an
immediate consequence of the uniqueness of the solution of the homoge-
neous initial value problem (7.3) which was proved in Theorem 6.8. This
concludes the proof of Theorem 7.1. 0
Remark 7.2. Let f satisfy

Ilf(t) - fCr)11 =0; Lit - TI~, 0 < it < 1, T, t E [0, T).
I t can be shown that if u is the solution of the initial value problem (7.1)
then u'(t) is Holder continuous with exponent /3, /3 < min(a, it) on [s +
E, T] for every E > o. Moreover, if f(s) = 0 and x = 0 then u' is Holder
continuous with exponent /3 on [s, T].
The proof of remark 7.2 is rather long and tedious and we will not give it
here. Instead, we will show now how to use Theorem 7.1 in order to obtain
higher differentiability of the solution of (7.1) under stronger assumptions.
A typical result of this kind is:
Theorem 7.3. Let {A(t)}tE[O, T] satisfy the conditions (PI) and (P2 ) and
assume that t ~ A(t)A(O)-1 is differentiable and its derivative A'(t)A(O)-1
satisfies a Holder condition
IIA'{t)A(O)-1 - A'{T)A(O)-III =0; Cit - Tl a
for T, t E [0, TJ, ° < a =0; l.

Let f(t) be differentiable on [0, T] and assume that its derivative I' is HOlder
continuous on [0, T] i.e.,
III' ( t) - I' ( T ) II =0; Ci t - T I,B for T, t E [0, T), °< /3 =0; 1.
If u is the solution of the initial value problem
{
dUd~t) + A{t)u{t) = f{t) for 0 < t =0; T
(7.12)
u(O) = x
then u is twice continuously differentiable on ]0, T].
PROOF. Let u be the solution of (7.12) and set
=u{t+h)-u(t) r( )=f(t+h)-f(t)
uh ( t ) h , Jh t h
then
dU~; t) + A ( t ) Uh( t) = fh ( t) _ A (t + h~ - A ( t) u (t + h) (7.13)
The right-hand side of (7.13) is Holder continuous in t and therefore by

Theorem 7.1 we have for T> 0
Uh (t) = U( t, T) U h ( T)
+ {U(t, o)[fh(O) - A(o + h~ - A(o) u(o + h)] do.

(7.14)
Our assumptions on I(t) imply that Ih(t) --+ /'(t) as h --+ 0 uniformly on
[0, T]. From the differentiability of A(t)A(O)-I, the strong continuity of
A(~)A(t)-1 and the continuity of A(t)u(t) for t > 0 it follows that
A(o+h)-A(o) ( h)
h u 0+
= A(o + h~ - A(o) A(O)-IA(O)A(o + h)-IA(O + h)u(o + h)
°
converges as h --+ 0 uniformly on [7, T - e] for e > 0, to A'(u )A(O)-l
A(O)u(u). Therefore we can let h --+ in (7.14) and obtain
u'(t} = U(t, T)u'(-r} + [U(t, o}[/,(o) - A'(o)A(O)-IA(O}U(o}] do.

T
From our assumptions and the fact that u is C I on [T, t] it follows that
g(o) = /'(0) - A'(o)A(O)-IA(O)u(o) satisfies a Holder condition on [T, T]
and therefore by Theorem 7.1 u'(t) is continuously differentiable on ]T, t]
and satisfies the equation
dVd~t) + A(t)v(t) = /'(1) - A'(t}A(O)-IA(O}U(t} for T < t .::; T.
(7.15)
Since T > 0 is arbitrary it follows that u'( t) is continuously differentiable on
]0, T] as claimed. 0
Remark 7.4. If in the previous theorem we assume that A(t)A(O)-1 is k

times continuously differentiable and its k-th derivative is Holder continu-
ous and if we also assume that I is k times continuously differentiable and
. its k-th derivative is Holder continuous, then we can proceed as in the proof
of Theorem 7.3 and show that u is k + 1 times continuously differentiable
on ]0, T]. In particular if A(t)A(O)-1 and I(t) are COO functions on [0, T]
then u is cae on ]0, T].
5.8. Asymptotic Behavior of Solutions in

the Parabolic Case
°
Let A(t). t ~ be a family of operators in a Banach space X satisfying for
every T> 0 the assumptions (PI )-(P3 ), of Section 5. If 1:[0, oa[ --+ X is
Holder continuous on [0, oa[ then the initial value problem,
dUd~t} + A(t}u(t) = I(t}, t> 0

(8.1)
{
u(O} = x
has a unique solution u on [0, 00[. The asymptotic behavior of this solution,
as t ~ 00, is the subject of the present section. In order to study this
asymptotic behavior we will assume that the conditions (PI)-(P3) hold
uniformly on [0, oo[ i.e., that they hold for every T> 0 with constants M, L
and a which are independent of T. We will further assume:
(P4 ) The operators A(t)A(s)-1 are uniformly bounded for 0 :$ s, t < 00
and there exists a closed operator A (00) with domain D such that
lim II(A(t) -A(oo))A(O)-111 = o. (8.2)
t---+ 00
Under these assumptions the solutions of the homogeneous evolution

equation decay exponentially to zero. This follows from:
Theorem 8.1. Let {A(t)}C200 satisfy the conditions (P I )-(P3 ) uniformly on

[0,00[. If {A(t)}C200 satisfies also (P4 ) and U(t, s) is the evolution system for
{A (t)} (see Theorem 6. I) then there exist constants C ~ 0 and {t > 0 such that
IIU(t,s)II:$Ce-~(t-S) for O:$S:$t. (8.3)
PROOF. We note first that from the results of Section 2.5, the denseness of D
and the fact that (P2) holds for all t ~ 0 it follows that - A (t) is the
infinitesimal generator of an analytic semigroup Sc(s), s ~ 0 for every t ~ 0
and that there are constants C ~ 0 and l) > 0 (independent of t) such that;
IISr(s) II :$ Ce- Ils for s~ 0 (8.4)
IIA(t)Sc(s)1I :$ C e- Ils for s> O. (8.5)

S
Set
P(Il) = sup {II (A (t) - A (s)) A (r) - III : 0 :$ Il :$ s, t < 00, 0 :$ r < oo}.
(8.6)
From (P3 ) and (P4 ) it follows that p(ll) is finite for Il ~ 0 and p(p.) ~ 0 as
Il-> 00. Combining (8.6) with the assumption (P3 ) we find
for Il:$ s, t (8.7)
where throughout the rest of the proof C will denote a generic constant.
From (8.7) we deduce
IIRI(t,s)11 = II(A(s) -A(t))Ss(t - s)11

:$ II(A(s) - A(t))A(s)-IIIIIA(s)Ss(t - s)11
:$ cr( ~ )Vp(,u) (t - S)"/2-1 e - 8(t-s) for ,u:$ s :$ t,
(8.8)
and by induction on m it follows easily that
II R '" ( 1• s ) II = II { R 1( I. T) Rm_ 1 ( T, S ) d T II
r::t.:\
e-8(t-q ( Cvp(p.) (I - s)'" /2) m
<-- (8.9)
- 1 - s r( n;a)
Since for 0 < f3 .$ 1 there is a constant Cp such that
for x ~ 0,
we have for p. .$ s .$ 1
oc
IIR(I,s)lI.$ L IIR",(t,s)II.$C/p(p.)(r-s),,/2-'
",=1
and thus, for every 0 < ~' < 6 there is a P.o > 0 such that for t ~ s ~ p. ~ P.o
IIR(I, s)11 .$ C/p(p.) (I - S),,/2-1 exp{ -~'(I - s». (8.10)
r
Finally fixing ~, ~ < ~' < 6, we find
II U ( 1, s ) II .$ II Ss (I - s) II + II ST (I - T) II II R ( T, S ) II d T
s
.$ Ce- 8(I-s) + C{P(p.) e-II'(I-S)(t - s ),,/2 .$ Ce-lI(t-S)
for t ~ s ~ p. ~ P.o' (8.11)

By modifying, if necessary, the constant C, the inequality (8.11) holds for all
t ~ s ~ 0 and the proof is complete. 0
We turn now to the inhomogeneous initial value problem (8.1), and make
the following additional assumption:
(F). The function f:[O, oo[ --+ X satisfies a uniform Holder condition on
[0,00[, i.e., there are constants C ~ 0 and 0 < y .$ 1 such that
IIf(t) - f(s)1I .$ Cil - slY for 0.$ s, t < 00
and there is an element f( 00) E X for which

lim IIf( t) - f( 00) II = O. (8.12)
IT CXl
We have now,
Theorem 8.2. Let {A(t)}t~O satisfy (PI)-(P3) uniformly on [0,00[. Assume

further that {A(t)}I~o satisfy (P4 ) and that f satisfies (F). If u is the solution
of the initial value problem (1) then there is an element u( (0) E X, indepen-
dent of x E X, such that
lim u(t) = u(oo), (8.13)
t --+ 00
u(oo) E D, A(oo)u(oo) = f(oo) (8.14)

and
lim u'(t) = O. (8.15)
t--+ 00
PROOF. We start by showing that if f(oo) = 0 then u(t) -+ 0 as t -+ 00.

From Theorem 7.1 it follows that
u(t) = U(t,O)x + [U(t, a)f(a) da.

o
Since by Theorem 8.1 II U(t, O)xll ::s; Ce-~tllxll for some iJ> > 0 it suffices to
prove that f;U(t, a)f(a) da -+ 0 as t -+ 00. Let II U(t, s)II ::s; Ce-~(t-s) and
IIfil 00 = SUPt~O IIf(t) II . Given e > 0 choose T> 0 so that for a ~ T, IIf( a) II
::s; (e/2)(iJ>/C) and choose TI > T such that for t ~ T I, e-~(t-T)::s;
(e/2)(iJ>/C)(IIfiloo + 1)-1. Then for t > TI
II {U( t, a )f( a) dall ::s; Ilfo T U( t, a )f( a) dall + II f; U( t, a )f( a) dall

e e
<2"+2"=e
and therefore u(t) -+ 0 as t -+ 00.
Next we show that the assumptions (PI)-(P4) imply that A(oo) is
invertible. Indeed, from (P4 ) it follows that
III - A(oo)A(t)-lli ::s; II(A{t) -A{oo))A{O)-lli IIA{O)A{t)-lli -+ 0
as t -+ 00.
Therefore, for t large enough A( 00 )A(t)-I is invertible. Denoting its inverse
by C(t), it is easy to see that A(t )-IC(t) is the inverse of A( (0).
We can now prove (8.13) and (8.14) as follows: Let u(t) be the solution of
(8.1) and set u(oo) = A(OO)-If(oo). If v(t) = u(t) - u(oo) then
dVd{tt) + A{t)v{t) = f{t) - A{t)u{oo) = g(t)

{ (8.16)
v(O) = x - u(oo).
But g(t) is obviously Holder continuous on [0, oo[ and

IIg(t)II ::s; IIf(t) - f(oo) II + II{A(oo) - A(t))A(oo)-lf(OO)II
::s; IIf(t) - f(oo)II + II(A(oo) -A(t))A(O)-lli IIA(O)A(oo)-lf(OO)II -+ 0
(8.17)
as t -+ 00. Therefore by the first part of the proof IIv(t)II -+ 0 as t -+ 00
and hence u(t) -+ u(oo) as t -+ 00 and the proof of (8.13) and (8.14) is
complete.
To prove the last part of the claim, i.e., u'(t) ~ 0 as t ~ 00 we need

further estimates which we state and prove in the next lemma.
Lemma 8.3. Let the conditions of Theorem 8.2 be satisfied. Then for /L .::; s <
'T, t,
IIA(t)St(t - s) - A(s )Ss(t - s) II .::; C/p(Jl) (t - s r /2 - l e- 8 (t-s)
(8.18)
and
II R (t, s) - R ( 'T, s) II .::; C/P(/L) (t - 'T) 13 ( 'T - s r /2 - 13 - Ie-il(r-s)
(8.19)
where 0 < & < S, 0 < f3 < aj2 and as before
P(Jl) = sup {II (A (t) - A (s )) A (r) - III : Jl .::; s, t < 00, 0 .::; r < oo}.
PROOF OF LEMMA 8.3: The first estimate, (8.18), is obtained similarly to
(6.18) by contour integration. Let f = {A: I arg AI = &} and f8 = {A: A -
S E f} then
A ( t ) St (t - s) - A ( s ) Ss (t - s)
= 1
-2 .1 Ae-i\(t-S)[R(A: A(t)) - R(A: A(s))] dA
7TI f8
= _1-. e - 8 (t-S)1(A + S)e-i\(t-s)

2m f
. [R (A + S : A (t )) - R (A + S : A (s ))] d A. (8.20)
But,
IIR(A + S: A(t)) - R(A + S; A(s ))11
.::; IIR(A+S:A(t))IIII(A(t)-A(s))A(s)-111
. IIA(s)R(A + S: A(s))11
.::; q(A(t) -A(S))A(s)-IIIIA! SI
.::; C/ P (/L) (t - s r /2 IA! SI (8.21 )
where we used (8.7) and (P2). Using (8.21) to estimate (8.20) yields (8.18).
To prove (8.19) we note that if we use (8.7) to estimate II(A(t)-
A(s»A( 'T)-III and follow the proof of Lemma 6.4 we find that for Jl .::; s <
'T .::; t
II RI (t, s) - RI ( 'T, s) II .::; C/P(/L) (t - 'T) 13 ( 'T - s r /2 - 13 - 1 e- 8 (r-s)

(8.22)
where 0 < f3 < aj2. Recalling that (see (8.8))
IIR1(t, s)1I .::; C/p(J-L) (t - S)"/2-1 e - O(t-S)
.::; Cp.{}VP(J-L) (t - s)p-1e-{}(t-S) (8.23)

and
we find,
IlfR1(t, O')R(O', s) dO'II.::; Cp(J-L)e-{}(t-S)( T - S)0</2-1~\t - O')P-l dO'
.::; CVP(J-L) (t - T)P( T - S)"/2-1 e -{}(t-S). (8.25)
Estimating IIR(t, s) - R(T, s)11 as in the proof of Corollary 6.5, using

(8.22), (8.24) and (8.25) yields (8.19). 0
We now return to the proof of Theorem 8.2. For every t ;:::: s > 0 the
solution u of the initial value problem (8.1) can be written as
u(t)= U(t,s)u(s)+ fU(t,O')/(O') dO'.

S
Since by (6.10)
II %t U(t, s)u(s)ll.::; IIA(t)U(t, s)llllu(s)11 .::; t ~s

it suffices to prove that the norm of
;t [U(t, 0')/(0') dO' = ;t[Scr(t-O')/(O')dO'+ ;t[W(t,O')/(O') dO'
(8.26)
can be made as small as we wish by choosing sand t > S large enough. To

prove this claim we treat each term on the right-hand side of (8.26)
separately.
In order to estimate the first term we denote
ll(J-L) = sup{ll/(t) - l(s)1I : J-L'::; s, t < oo}.

From the assumption (F) it follows that ll(J-L) ~ 0 as J-L ~ 00 and
III ( t) - I ( s ) II .::; VII (J-L) It - slY/2 for J-L'::; s, t < 00. (8.27)
A simple computation yields
aat fs \ (t - a) f( a ) da = f[ A (t) St (t - a) - A (a) S" (t - a)] f( a ) da

s
- fA(t)St(t - a)[f(a) - f(t)] da

s
(8.28)
and therefore estimating each of the three terms on the right-hand side of
(8.28), separately, using (8.18), (8.27) and (8.4) we get
II %t t s,,( t - a )f( a) dall s CIIfll 00 VP(fL) + CV{j(fL) + CIIfll ooe- 8 (t-s).
(8.29)
To estimate the second term on the right-hand side of (8.26) we note that
by the proof of Theorem 7.1 we have
aft W(/, a)f(a) da = ft -a
-a a W(t, a)f(a) da (8.30)
1 s s t
and from (6.35)
aW(t,s)
at =
ft[
s A(/)St(t-T)-A(T)ST(/-T) ] R(T,S)dT
+ fA (I) St (I - T)[ R (I, s) - R ( T, s)] dT + Sf (t - s) R (I, s).

s
Estimating each one of the terms on the right-hand side, using (8.18), (8.24),
(8.19), (8.5) and (8.4) yields
II aw~:, s) II s CVP(fL) (I - s r/ 2- 1 e-~(t-s) (8.31)
and therefore,
lit %t W(t, a )f( a) dall s CIIfllooVP(fL) . (8.32)
Combining (8.29) and (8.32) and noting that P(fL) ~ 0 and {j(fL) ~ 0 as
fL
~ 00 yields that for any e > 0 there is a fL such that if 1 > s ~ fL
II %1 t U( I, a) f( a) dall < e + CIIflloo e- 8 (t-s)
which concludes the proof of Theorem 8.2. o
Theorem 8.2 shows that if as 1 ~ 00, A(t) converges to A( 00) and f(t) to
f(oo), then the solution u(t) of the initial value problem (1) converges to a
limit u( 00) as t ~ 00. In order to get more detailed information on the
convergence of u(t) to u( 00) more must be known on the convergence of
A (t) to A ( 00) and f (t) to f (00). We conclude this section with one result in
this direction.
We will make the following assumptions:
(All) The operator A(t) has an expansion
A(t) = A +.!.A
o t '
+ ~A
t2 2
+ ... + ~A + ~B (t)
tn n tn n
(8.33)
where Ao is a densely defined closed linear operator for which the resolvent
set p(Ao) satisfies p(Ao):) {A: ReA'::;; O} and
Mo
IIR(A: Ao)1I .::;; IAI + I (8.34)
The operators A k , 1 .::;; k .::;; nand Bn(t) for t ;::: 0 are closed linear operators
satisfying D(Ak):) D(Ao) and D(Bn(t»:) D(Ao). Furthermore, the
bounded linear operators Bn( t )Ao' satisfy
(8.35)
for some 0 0 and
lim IIBn(t)Ao'li = 0 (8.36)

t --+ 00
and
( Fn) The function f (t) has the following expansion
(8.37)
where CfJn(t) is Holder continuous in t and
lim II CfJn( t ) II = O. (8.38)

t --+ 00
We note that if {A(t)}t>t - 0

satisfies (An) for some n ;::: 0 it also satisfies
(Ad with 0.::;; k .::;; n where
n
Bk(t) = L t-t+kA t + t-n+kBn(t).
t=k+l
Furthermore if A(t) satisfies (An) with n ;::: 1 so doesA(t) + (a/t)] where]
is the identity operator. Finally if f( t) satisfies (F,,) it also satisfies (Fk ) with
o .: ; k .::;; n with the appropriate definition of CfJk(t).
We proceed by showing that the assumptions (An) imply the existence of
a to > 0 such that the family {A(t)}t~to satisfies the necessary conditions for
the existence of a unique solution u(t) to the initial value problem
dUd~t) + A(t)u(t) = f(t) for t> to

(8.39)
{
u(t o ) = x
where f satisfies the condition (F). More precisely we have:
Lemma 8.4. If {A(t)}t>o satisfies (An) with n ~ 0 then there is a to> 0 such
that {A(t)}t>t
-
satisfies;
()
(i) For every t ~ to' the resolvent R(X: A(t)) of A(t) exists for all Re X:::; 0
and
M
IIR(X: A(t))11 :::; IXI + I for all X with Re X :::; O.
(ii) There exist constants Land 0 < a :::; I such that

for to:::; t, s, T.
(iii) The operators IIA(t)A(s)-11i are uniformly bounded for to :::; s, t < 00
and
lim II(A(t) -Ao)Aolli = o. (8.40)
t --+ 00
PROOF. (i) Set Q(t) = A(t) - A o' from the closed graph theorem it follows
that for every t > 0 and X E p(Ao), Q(t)R(X: Ao) is a bounded linear
operator. Furthermore, for X with Re X :::; 0 we have
II
IIQ(t)R(X: Ao)1I :::; L y,t-' + f3n(t)t- n (8.41)

'=1
where y, = (Mo + 1)IIA,Ao III and f3n(t) = (Mo + 1)IIBn(t)AO III. There-
fore, there is a to> 0 such that for t ~ to and Re X:::; 0, IIQ(t)R(X: Ao)11 <
1. Fix such a to > 0, let X be such that Re X :::; 0 and consider
AI - A(t) = [I - Q(t)R(X: Ao)](AI - Ao). (8.42)
For t ~ to the operator on the right-hand side of (8.42) is invertible and
IIR(X:A(t))II:::; IIR(X:AoH IIU- Q(t)R(X:Ao)rlll:::; I~:o I

for all X with Re X:::; O. In particular it follows that for t ~ to, A(t)-I exists.
(ii) Using the Holder continuity of Bn(t)AO I it follows easily that for
t ~ to > 0 the operator A(t)Ao I is Holder continuous with exponent 0 < P
:::; 1. For T ~ to, II Q( T)Ao III < 1 and consequently the operator I +
Q( T)Ao I is invertible and its inverse AoA( T)-I has norm less or equal to 2.
Therefore,
II(A(t) -A(s))A(T)-111 :::; II(A(t) -A(s))AoIIIIIAoA(T)-111
:::; Cit - siP.
(iii) For t, s ;;::: to we have

IIA(t)A(s)-111 :::; IIA(t)AoIIIIIAoA(s)-11i :::; 2I1A(t)A lll o
= 2111 + Q(t)AOIIi :::; 3.
Finally choosing A = 0 in (8.41) and letting t --+ 00 yields (8.40). 0
Lemma 8.4 implies that if A(t), t ;;::: 0, satisfies (An) with n ;;::: 0 then
{A(t)}t;,to satisfies on [to' oo[ the assumptions (PI)-(P4) with A(oo) = Ao.
Moreover, it is easy to check that if f satisfies (Fn) with n ;;::: 0 then it
satisfies the assumption (F) with f( 00) = fo and therefore under these
assumptions the initial value problem (8.39) has a unique solution u on
[to' 00[.
Theorem 8.5. Let A(t) satisfy the conditions (An) with some n > 0 and let f
satisfy the condition (Fn) with the same n > o.
If u is the solution of the initial
value problem (8.39) then for t ;;::: to,
1 1 1 1
u(t) = u o + -u
t 1 + -u
t2 2 + ... + -u
tn n + -
tnvn(t) (8.43)
where vn ( t) --+ 0 as t --+ 00 and

(8.44)
k
AOu k - (k - l)u k _ 1 + L A.uk _. = fk for 1:::; k :::; n. (8.45)
p= 1
PROOF. For n = 0 Theorem 8.5 coincides, with the obvious changes of

notations, with Theorem 8.2 and therefore it is true for n = O. Assume that
it is true for (m - 1) < n. Then the equations (8.43), (8.44) and (8.45) hold
with n replaced by m - 1. We will show that in this case the theorem is true
also for m. Set
111
u(t) = u o + -u
t 1 + ... + - -u
t"'- 1 ",-I
+ -w(t)
t'" (8.46)
where Uk' 0 :::; k :::; m - 1 are determined consecutively by (8.45). Substitut-

ing (8.46) into the differential equation
du(t)
-;It + A(t)u(t) = f(t) (8.47)
we get
~[dW + (A(t)
t'" dt
_ m
t
I)W]
1 [
= t'" (m - m 1
l)u",_1 - P~IA.U",_p + f",-B",(t)u o + CPm(t) + "(g(t) ]
(8.48)
where
n
Brn{t) = L t-l+rnA, + rn+rnBn{t),
'~rn+1
n
IPrn{t) = L t-'+rnl, + t-n+rnIPn{t)
and g(t) is a finite sum of terms of the form t-k(A,uj + Brn(t)uJ with
o :::; k, j
:::; m - 1 and 0 :::; i, I:::; m. It is easy to check that for t ~ to > 0,
t-Ig(t) is Holder continuous in t. Multiplying both sides of (8.48) by t m we
find
~; + (A{t) -7 !)w = (m - l)u m_ 1 - f Avum-v + 1m

v~1
+ [IPrn{t) - Bm{t)u o + t-Ig{t)].

The term depending on t on the right-hand side is Holder continuous and
tends to zero as t ~ 00. The operator A(t) - (mit)! clearly satisfies (Ao)
and therefore by our theorem with n = 0,
(8.49)
where
lim Ilvm{t)11 = O.
{--> 00
Substituting (8.49) into (8.46) gives the desired result for m. The theorem
follows by induction. D
CHAPTER 6
Some Nonlinear Evolution Equations
6.1. Lipschitz Perturbations of Linear

Evolution Equations
In this section we will study the following semilinear initial value problem:
dUd~/) + AU(/) = /(1, U(/», I> 10

{ (1.1)
U(/o) = Uo
where - A is the infinitesimal generator of a Co semigroup T( I), 1 ~ 0, on a

Banach space X and /: [to, T] x X --+ X is continuous in 1 and satisfies a
Lipschitz condition in u.
Most of the results of this and the following sections, in which A is
assumed to be independent of t can be easily extended to the case where A
depends on 1 in a way that insures the existence of an evolution system
U(t, s), 0 :::;; s :::;; t :::;; T, for the family {A(t)}IE[O, Tj' We will not deal with
these extensions here and as a consequence the following sections (Section
6.1-6.3) are independent of the results of Chapter 5.
The initial value problem (6.l) does not necessarily have a solution of any
kind. However, if it has a classical or strong solution (see Definition 4.2.1)
then the argument given at the beginning of Section 4.2 shows that this
solution u satisfies the integral equation
U(/) = T(I - 10)U o + fT(1 - s)/(s, u(s» ds. (1.2)

10
184 Seroigroups of Linear Operators
It is therefore natural to define.
Definition 1.1. A continuous solution u of the integral equation (1.2) will be

called a mild solution of the initial value problem (1.1).
We start with the following classical result which assures the existence
and uniqueness of mild solutions of (1.1) for Lipschitz continuous func-
tions f.
Theorem 1.2. Let f: [to. T) X X -+ X be continuous in t on [to. T) and

uniformly Lipschitz continuous (with constant L) on X. If - A is the infinitesi-
mal generator of a Co semigroup T( t). t <! 0. on X then for every U o E X the
initial value problem (1.1) has a unique mild solution u E C([t o• T): X).
Moreover. the mapping U o -+ u is Lipschitz continuous from X into
C([t o• T): X).
PROOF. For a given U o E X we define a mapping
F: C([to. T] : X) --+ C([to. T]: X)

by
(Fu)(t) = T(t - to)u o + fT(t - s)f(s, u(s» ds,
'0
(1.3)
Denoting by lIulioo the norm of u as an element of C([to' T): X) it follows
readily from the definition of F that
II(Fu)(t) - (Fv)(t)11 ~ ML(t - to)Uu - vII 00 (1.4)
where M is a bound of II T(t)1I on [to' T). Using (1.3), (1.4) and induction
on n it follows easily that
"
II(F"u)(t) - (F"v)(t)1I ~ ( ML ( t ~ to ») lIu - vll oc
n.
whence
(MLT)"
IIF"u - F"vll ~ ,
n.
Ilu - vll oo · ( 1.5)
For n large enough (MLT)"/n! < I and by a well known extension of the
contraction principle F has a unique fixed point u in C([to' T): X). This
fixed point is the desired solution of the integral equation (1.2).
The uniqueness of u and the Lipschitz continuity of the map U o --+ u are
consequences of the following argument. Let v be a mild solution of (1.1) on
6 Some Nonlinear Evolution Equations 185
[to' T] with the initial value vo' Then,

Ilu(t) - v(t)II .s IIT(t - to)u o - T(t - to)voll
+ rllT(t - s)(f(s, u(s)) - f(s, v(s») ds
10
.s Mllu o - voll + ML rllu(s) - v(s)II ds

10
which implies, by Gronwall's inequality, that
and therefore
IIu - vll oo .s MeML(T-Io)IIu o - voll
which yields both the uniqueness of u and the Lipschitz continuity of the
map U o -> u. 0
It is not difficult to see that if g E C([to' T]: X) and in the proof of

Theorem 1.2 we modify the definition of F to
(Fu)(t) = g(t) + tT(t - s)f(s, u(s» ds

10
we obtain the following slightly more general result.
Corollary 1.3. If A and f satisfy the conditions of Theorem 1.2 then for every
g E C([to' T]; X) the integral equation
w(t) = g(t) + tT(t - s )f(s, w(s» ds ( 1.6)

10
has a unique solution w E C([to' T]: X).
The uniform Lipschitz condition of the function f in Theorem 1.2 assures

the existence of a global (i.e. defined on all of [to, T]) mild solution of (1.1).
°
If we assume that f satisfies only a local Lipschitz condition in u, uniformly
in t on bounded intervals, that is, for every t' ~ and constant c ~ there
is a constant L( c, t') such that
°
IIf(/, u) - f(/, v)11 .s L(c, 1')IIu - vII ( 1.7)
holds for all u, v E X with IIuli .s c, IIvll .s c and I E [0, I'], then we have
the following local version of Theorem 1.2.
Theorem 1.4. Let f: [0, oc[ X X -> X be continuous in t for t ~ °

and locally
Lipschitz continuous in u, uniformly in t on bounded intervals. If - A is the
infinitesimal generator of a Co semigroup T(t) on X then for every U o E X
there is a t max ~ 00 such that the initial value problem
dUd~t) + Au(t) = f(t, U(t», t~O

{ ( 1.8)
U(O) = Uo
has a unique mild solution u on [0, t max [. Moreover, if t max < 00 then
lim Ilu(t)11 = 00.
ti (max
PROOF. We start by showing that for every to ~ 0, Uo E X, the initial value

problem (1.1) has, under the assumptions of our theorem, a unique mild
solution u on an interval [to, tJ1 whose length is bounded below by
8(t II II) . {I Iluoll } (I 9)

0' Uo = mm 'K(to)L(K(t o ), to + I) + N(t o) .
where L(c, t) is the local Lipschitz constant of f as defined by (1.7),
M(t o) = sup{IIT(t)lI: 0 ~ t ~ to + 1}, K(t o) = 21Iu oIIM(t o) and N(to) =
max {lIf(t, 0)11 : 0 ~ t ~ to + I}. Indeed, let t) = to + 8(to, Iluol!) where
8(to, Iluol!) is given by (1.9). The mapping F defined by (1.3) maps the ball
of radius K(to) centered at 0 of C([t o, tJ1: X) into itself. This follows from
the estimate
II(Fu)(t}ll
~ M(to}lluoll+ {IIT(t - s)II(llf(s, u(s» - f(s,O)1I + Ilf(s,O)II) ds
(0
~ M(to)lIuoll + M(to)K(to)L(K(t o ), to + I)(t - to)

+ M(to)N(to)(t - to)
~ M(to){lluoll + K(to)L(K(t o), to + I)(t - to)+N(to)(t - to)}
~ 2M(t o )lIu oll = K(t o )
where the last inequality follows from the definition of t). In this ball, F
satisfies a uniform Lipschitz condition with constant L = L(K(to)' to + I)
and thus as in the proof of Theorem 1.2 it possesses a unique fixed point u
in the ball. This fixed point is the desired solution of (1.1) on the interval
[to' ttl·
From what we have just proved it follows that if u is a mild solution of
(1.8) on the interval [0, r] it can be extended to the interval [0, r + 8] with
8> 0 by defining on ['T, r + 8], u(t) = w(t) where wet) is the solution of
the integral equation
w(t) = T(t - r)u( r) + fT(t - s)f(s, w(s» ds, r ~ t ~ 'T + 8.

T
(1.10)
Moreover, 8 depends only on Ilu(r)II, K(r) and N('T).
Let [0, t max [ be the maximal interval of existence of the mild solution u of
(1.8). If t max < 00 then liml .... lmax lIu(t)11 = 00 since otherwise there is a
sequence tn i t max such that lIu(tn)1I :::; C for all n. This would imply by
°
what we have just proved that for each tn' near enough to t max , u defined on
[0, t n ] can be extended to [0, tn + 8] where 8 > is independent of tn and
hence u can be extended beyond t max contradicting the definition of t max .
To prove the uniqueness of the local mild solution u of (1.8) we note that
if v is a mild solution of (1.8) then on every closed interval [0, to] on which
both u and v exist they coincide by the uniqueness argument given at the
end of the proof of Theorem 1.2. Therefore, both u and v have the same t max
and on [0, t max [, u == v. 0
It is well known that in general, if I just satisfies the conditions of

Theorem 1.2 or Theorem 1.4 the mild solution of (1.1) need not be a
classical solution or even a strong solution of (1.1). A sufficient condition for
the mild solution of (1.1) to be a classical solution is given next.
Theorem 1.5 (Regularity). Let - A be the infinitesimal generator 01 a Co

semigroup T(t) on X. II I: [to' T] X X --+ X is continuously differentiable Irom
[to, T] X X into X then the mild solution 01(1.1) with U o E D(A) is a classical
solution 01 the initial value problem.
PROOF. We note first that the continuous differentiability of I from [to, T]

X X into X implies that I is continuous in t and Lipschitz continuous in u,
uniformly in t on [to, T]. Therefore the initial value problem (1.1) possesses
a unique mild solution u on [to' T] by Theorem 1.2. Next we show that this
mild solution is continuously differentiable on [to, T]. To this end we set
B(s) = (a/au)/(s, u) and
g(t) = T(t - to)/(t o , u(t o )) - AT(t - to)u o
+ 1
I
1
o
a
T(t - s) as/(s, u(s)) ds. (1.11)
From our assumptions it follows that g E C([t o, T]: X) and that the
function h(t, u) = B(t)u is continuous in t from [to, T] into X and uni-
formly Lipschitz continuous in u since s --+ B(s) is continuous from [to, T]
into B(X). Let w be the solution of the integral equation
w(t) = g(t) + {T(t - s)B(s)w(s) ds. (1.12)

10
The existence and uniqueness of wE C([to' T]: X) follows from Corollary

1.3. Moreover, from our assumptions we have
I(s, u(s + h)) - I(s, u(s)) = B(s)(u(s + h) - u(s)) + w\(s, h)
(1.13)
and
/(s + h, u(s + h)) - /(s, u(s + h»
= (a / as) / ( s, u (s + h» . h + w2 ( s, h )
(1.14)
where h-11Iwi(s, h)11 ~ 0 as h---> 0 uniformly on [to, T] for i = 1,2. If
whet) = h-I(u(t + h) - u(t» - wet) then from the definition of u, (1.12),
(1.13) and (1.14) we obtain
wh{t) [h-I(T{t + h - to)u o - T{t - to)u o) +AT{t - to)uoJ
*
=
+ [T( t - s) ( W I (s, h) + W 2 (s, h») ds

to
+ {T(t - s)( :s/(s, u(s + h)) - :/(s, u(s») ds
+ [* {o+hT(t + h - s)/(s, u(s» ds - T(t - to)/(t o, U(t o»)]
+ [T(t - s )B{s )wh(s) ds. (1.15)

to
It is not difficult to see that the norm of each one of the four first terms on
the right-hand side of (l.15) tends to zero as h ~ o. Therefore we have
IIwh(t)11 ::; e(h) + M [lIwh(s)llds (l.l6)
to
where M = max{IIT(t - s)IIIIB(s)ll: to::; s::; T} and e(h) ~ 0 as h ~ O.
From (1.16) it follows by Gronwall's inequality that Ilwh(t)11 ::; e(h)e(T-to)M
and therefore Ilwh(t)11 ---> 0 as h ~ O. This implies that u(t) is differentiable
on [to' T] and that its derivative is wet). Since wE q[to' T]: X), u is
continuously differentiable on [to' T].
Finally, to show that u is the classical solution of (1.1) we note that from
the continuous differentiability of u and the assumptions on the differentia-
bility of / it follows that s ~ /(s, u(s» is continuously differentiable on
[to' T]. From Corollary 4.2.5 it then follows that
v(t) = T{t - to)u o + [T{t - s)/{s, u(s» ds (1.17)
to
is the classical solution of the inifial value problem
{
dVd~t) _+ Av{t) = /(t, u{t»
(1.18)
v{t o ) - Uo .
But, by definition, u is a mild solution of (1.18) and by the uniqueness of

mild solutions of (1.18) it follows that u = von [to, T]. Thus, u is a classical
solution of the initial value problem (1.1). D
In general if /: [to' T] X X --+ X is just Lipschitz continuous in both

variables i.e.,
tl' t2 E [to' T]
( 1.19)
the mild solution of (1.1) need not be a strong solution of the initial value
problem. However, if X is reflexive, the Lipschitz continuity of / suffices to
assure that the mild solution u with initial data U o E D( A) is a strong
solution. Indeed we have:
Theorem 1.6. Let - A be the infinitesimal generator 0/ a Co semigroup T( t)

on a reflexive Banach space X. 1/ /: [to' T] X X --+ X is Lipschitz continuous
in both variables, U o E D( A) and u is the mild solution 0/ the initial value
problem (1.1) then u is the strong solution 0/ this inilial value problem.
PROOF. Let II T(t)11 ~ M and 11/(/, u(t»11 ~ N for 10 ~ 1 ~ T and let /
satisfy (\.19). For 0 < h < 1 - to we have
u(t + h) - u(t) = T(I + h ~ to)u o - T(t - to)u o
+ {o+hT(t + h - s )/(s, u(s)) ds
10
+ {T( t - s)[ / (s + h, u (s + h» - / (s, u (s »] ds

10
and therefore,
lIu(t + h) - u(t)11 ~ hMIIAuoll + hMN
+MC{(h+ lIu(s+h)-u(s)ll)ds
10
~ Clh + MC {lIu(s + h) - u(s)llds

10
which by Gronwall's inequality implies

Ilu(t + h) - u(t)11 ~ Cle™Ch ( 1.20)
and u is Lipschitz continuous.
The Lipschitz continuity of u combined with the Lipschitz continuity of /
imply that t --+ /(t, u(t» is Lipschitz continuous on [to' T]. From Corollary
4.2.11 it then follows that the initial value problem
dv
{ dt + Av = /(/, u(t» (1.21 )
v(t o ) = Uo
has a unique strong solution v on [to' T] satisfying

v(t) = T(t - to)u o + {T(t - s)/(s, u(s» ds = u(t)
10
and so u is a strong solution of (1.1). o

We conclude this section with an application of Theorem 1.2 which

provides us with a classical solution of the initial value problem (1.1). Let
- A be the infinitesimal generator of the Co semigroup T(t) on X. We
endow the domain D(A) of A with the graph norm, that is, for x E D(A)
we define IxlA = Ilxll + IIAxll. It is not difficult to show that D(A) with
the norm I . IA is a Banach space which we denote by Y. The completeness
of Y is a direct consequence of the closedness of A. Clearly Y c X and since
T( t) : D( A) ~ D( A), T( t), t ;;::: 0 is a semigroup on Y which is easily seen to
be a Co semigroup on Y.
Theorem 1.7. Let f: [to, T] X Y ~ Y be uniformly Lipschitz in Y and for

each y E Y let f(t, y) be continuous from [to' T] into Y. If U o E D(A) then
the initial value problem (1.1) has a unique classical solution on [to, T].
PROOF. We apply Theorem 1.2 in Y and obtain a function U E C([to, T]:

Y) satisfying in Y and a fortiori in X,
u(t) = T(t - to)u o + FT(t - s)f(s, u(s)) ds. (1.22)

to
Let g(s) = f(s, u(s)). From our assumptions it follows that g(s) E D(A)
for s E [to' T] and that s ~ Ag(s) is continuous in X. Therefore it follows
from Corollary 4.2.6 that the initial value problem
dv
{ dt + Av = g(t) ( 1.23)
v(t o ) = U o
has a unique classical solution von [to' T]. This solution is then clearly also
a mild solution of (1.23) and therefore
v(t) = T(t - to)u o + FT(t - s)g(s) ds

to
= T(t - to)u o + FT(t - s)f{s, u{s)) ds = u(t)

to
and u is a classical solution of (1,1) on [to, T]. D
If in the previous theorem we assume only that f: [to, T] X Y ~ Y is

locally Lipschitz continuous in Y uniformly in [to, T] we obtain, using
Theorem 1.4, that for every U o E D(A) the initial value problem possesses a
classical solution on a maximal interval [to, t max [ and if t max < T then
lim (lIu(t)11 + IIAu(t)ll) = 00. ( 1.24)

tit max
We note that in this situation it may well be that Ilu(t)11 is bounded on

[to'
t max [ and only IIAu(t)11 ~ 00 as t i t max . This is indeed the case in
many applications to partial differential equations.
6.2. Semilinear Equations with Compact Semigroups

We continue our study of the semilinear initiai value problem
dUd~t) + Au{t) = f{t, u{t))' t> 0

{ (2.1 )
u{o) = Uo
In the previous section we have proved the existence of mild solutions

(Definition l.l) of the initial value problem (2.1) under the assumptions that
- A is the infinitesimal generator of a Co semigroup of operators while
f( t, x) is continuous in both its variables and uniformly locally Lipschitz
continuous in x. If the Lipschitz continuity of f in x is dropped then, as is
well known, the existence of a mild solution of (2.1) is no more guaranteed
even if A == O. In order to assure the existence of a mild solution of (2.1) in
this case, we have to impose further conditions on the operator A. Our main
assumption in this section will be that - A is the infinitesimal generator of a
compact Co semigroup (Definition 2.3.1). We note in passing that in
applications generators of compact semi groups occur often in the case
where - A has a compact resolvent and generates an analytic semi group
T( t), t ~ o. Indeed, in this case,
by Theorem 2.S.2(d), so T(t) is continuous in the uniform operator topology

for t > 0 and hence by Theorem 2.3.3 it is also compact for t > O.
The main result of this section is the following local existence theorem.
Theorem 2.1. Let X be a Banach space and U c X be open. Let - A be the

infinitesimal generator of a compact semigroup T( t), t ~ o. If 0 < a ::; 00 and
f: [0, a[ X U ~ X is continuous then for every Uo E U there exists a t I =
t I (u o ), 0 < t I < a such that the initial value problem (2.1) has a mild solution
u E C([O, td: U).
PROOF. Since we are interested here only in local solutions, we may assume
that a < 00. Let II T(t)11 ::; M for 0 ::; t ::; a and let t' > 0, p > 0 be such
that B/u o ) = {v: Ilv - uoll ::; p} c U and Ilf(s, v)11 ::; N for 0::; s ::; t'
and v E Bp(u o). Choose t" > 0 such that

IIT(t)u o -u olI<p/2 for O.:<=;t.:<=;t"
and let
tl =
.(,,,
mm P) .
t, t ,a, 2MN
Set Y = C([O, ttl: X) and Yo = {u: u E Y, u(O) = uo, u(t) E Bp(u o ) for
o .:<=; t .:<=; t l ).
Yo is clearly a bounded closed convex subset of Y. We define a
mapping F: Y ~ Yo by
(Fu)(t) = T(t)u o + fT(t - s)j(s, u(s)) ds. (2.3)
o
Since
II(Fu)(t) - uoll .:<=; IIT(t)u o - uoll +11{T(t - s)j(s, u(s)) dsll
.:<=; P/2 + t I MN .:<=; p,

F maps Yo into Yo. Also, from the continuity of j on [0, a[ X U it follows
easily that F is a continuous map of Yo into Yo- Moreover, F maps Yo into a
precompact subset of Yo. To prove this, we first show that for every fixed t,
o .:<=; t .:<=; t l, the set Yo(t) = {(Fu)(t): u E Yo) is precompact in X. This is
clear for t = 0 since Yo(O) = {u o). Let t > 0 be fixed. For 0 < e < t set
U:u)(t) = T(t)u o + [-ET(t - s)j(s, u(s)) ds
o
= T(t)u o + T(e) {-ET(t - s - e)j(s, u(s)) ds.
Since T(t) is compact for every t > 0, the set Y,(t) = {(F'.u)(t): u E Yo) is
precompact in X for every e, 0 < e < t. Furthermore, for u E Yo we have
II(Fu)(t) - (F'.u)(t)1I .:<=; r

/-E
IIT(t - s)j(s, u(s))llds.:<=; eMN
which implies that Yo(t) is totally bounded, i.e, precompact in X. We

continue and show that
F(Yo ) = Y= {Fu: u E Yo) (2.4)
is an equicontinuous family of functions. For t2 > tl > 0 we have
II(Fu)(t l ) - (Fu)(t 2 )11 .:<=; II(T(t l ) - (T(t 2 ))u oll
+N f'IIT(t 2 - s) - T(t l - s)llds + (t2 - tl)MN.
o
(2.5)
The right-hand side of (2.5) is independent of u E Yo and tends to zero as
t2 - tl ~ 0 as a consequence of the continuity of T(t) in the uniform
operator topology for t > 0 which in turn follows from the compactness of
T( t), t > 0 (Theorem 2.3.2).
It is also clear that Y is bounded in Y. The desired precompactness of

Y= F(Yo) is now a consequence of Arzela-Ascoli's theorem. Finally, it
follows from Schauder's fixed point theorem that F has a fixed point in Yo
and any fixed point of F is a mild solution of (2.1) on [0, t d satisfying
u(t) E UforO..::; t..::; t]. 0
We turn now to global existence. Here further assumptions must be made

since global existence fails quite commonly. We start with the following
result.
Theorem 2.2. Let - A be the infinitesimal generator of a compact semigroup

°
T(t), t 2 on X. Iff: [0, oo[ X X ~ X is continuous and maps bounded sets in
[0, oo[ X X into bounded sets in X then for every U o E X the initial value
problem (2.1) has a mild solution u on a maximal interval of existence
[0, t max [. If t max < 00 then
lim II u( t) II = 00. (2.6)
t i t max
PROOF. First we note that a mild solution u of (2.1) defined on a closed

interval [0, tJl can be extended to a larger interval [0, t] + 8], 8 > 0, by
defining u( t + t]) = w( t) where w( t) is a mild solution of
dWd~ t) + A w ( t) = f (t + t], W( t ) )
{ (2.7)
w(O) = u(t])
the existence of which on an interval of positive length 8 > is assured by

Theorem 2.1. Let [0, t maxl be the maximal interval to which the mild
°
solution u of (2.1) can be extended. We will show that if t max < 00 then
Ilu(t)11 ~ 00 as t i t max • To do so we will first prove that t max < 00 implies
limtjtmJlu(t)11 = 00. Indeed, if t max < 00 and limtitmaJu(t)11 < 00 we can
assume that II T(t)11 ..::; M and Ilu(t)1 ..::; K for 0..::; t < t max where M and K
are constants. By our assumption on the function f we also have a constant
Nsuch that Ilf(t, u(t»11 ..::; NforO..::; t < t max • NowifO < p < t < t' < t max
then
Ilu(t') - u(t)11 ..::; IIT(t')u o - T(t)uoll
+II({-p +!~J{T(t' -s) - T{t - s))f{s, u{s)) dsll

+llt'T{t' - s}f{s, u{s)) dsll
..::; IIT{t')u o - T{t)uoll + N {-PIIT(t' - s) - T{t - s)llds
+2MNp + {t' - t)MN. (2.8)

Since t > P > 0 is arbitrary and since T( t) is continuous in the uniform

operator topology for t ;::: p > 0, the right-hand side of (2.8) tends to zero as
t, t' tend to t max . Therefore lim liln", u(t) = u(tmax) exists and by the first
part of the proof the solution u can be extended beyond t max' contradict-
ing the maximality of t max' Therefore the assumption that t max < 00 implies
that lim litm" Ilu(t)11 = 00. To conclude the proof we will show that
limtjln", Ilu(t)11 = 00. If this is false then there is a sequence til i (max and a
constant K such that Ilu(tll)11 .:s; K for all n. Let II T(t)11 .:s; M for 0 .:s; t .:s;
t max and let N = sup{llf(t, x)11 : 0 .:s; t .:s; t max , Ilxll .:s; M(K + In. Since
t ~ Ilu(t)11 is continuous and limlilma< Ilu(t)11 = 00 we can find a sequence
{h ll } with the following properties: hll ~ 0 as n ~ 00, Ilu(t)1I .:s; M(K + 1)
for til .:s; t .:s; til + hll and Ilu(tll + hll)11 = M(K + 1). But then we have
M(K + 1) = Ilu(tll + hn)11 .:s; IIT(hll)u(t,J11
+ 11"+11,, liT (
I"
tn+hll-s ) f ( s,u (s) ) lids
.:s; MK+ hnNM
which is absurd as h" ~ O. Therefore we have limtjlm,Ju(t)11 = 00 and the

We conclude this section with two useful conditions for the existence of
global mild solutions of the initial value problem (2.1) under the assump-
tions of Theorem 2.2.
Corollary 2.3. Let - A be the infinitesimal generator of a compact Co semi-

group, T(t), t;::: 0 on X. Let f: [0, oo[XX ~ X be continuous and map
bounded sets in [0, oo[ X X into bounded sets in X. Then for every U o E X the
initial value problem (2.1) has a global solution u E C([O, oo[ X X) if either one
of the following conditions is satisfied:
(i) There exists a continuous function k o(s) :[0, oo[ ~ ]0, oo[ such that II u( t) II
.:s; k o( t) for every t in the interval of existence of u.
(ii) There exist two locally integrable functions k](s) and k2(S) such that
for O.:s; s < 00, X E X. (2.9)
PROOF. Part (i) is a trivial consequence of Theorem 2.2. To prove (ii) we

reduce it to (i) as follows: Assume that the solution u exists on the interval
[0, t[. Set II T(t)11 .:s; Me WI and
lJt(t) = Mlluoll + foIMe-WSk2(s) ds.

6 Some Nonlinear Evolution Equation, 195
The function I/; thus defined is obviously continuous on [0, oo[ and we have
Ilu(t)lle-wI:'S: e-WIIIT(t)uoll + e-WILIIT(I - 5)/(5, U(5»)lId5
o
(2.10)
and by Gronwall's inequality
Ilu(1 )lie-wI :'S: 1/;(1) + M {kl(s )I/;(s) exp { M {kt(r) dr} ds
which implies the boundedness of II u( I) II by a continuous function. D
6.3. Semilinear Equations with Analytic Semigroups

As we have noted briefly at the beginning of section 6.2, the initial value
problem
dUd~l) + Au ( I) = / ( I, u ( 1) ) , 1 > 10
{ (3.1 )
U(lo) = Xo
occurs often in the applications with an operator - A which is the infinitesi-

mal generator of an analytic semigroup on a Banach space X. In this case if
RCA.: - A) is compact for some A. E p( - A) and /(1, x) is continuous in
[10' T] X X the problem has a (possibly non unique) mild local solution by
Theorem 2.1. If we assume further, as we will do in the sequel, that / is
regular with respect to - A in some sense, we will be able to obtain unique
local strong solutions of the initial value problem (3.1).
Throughout this section we will assume that - A is the infinitesimal
generator of an analytic semigroup T(t) on the Banach space X. For
convenience we will also assume that T( I) is bounded, that is II T( I) II :'S: M
for 1 :2: 0, and that 0 E P ( - A), i.e. - A is invertible.
We note that if -A is the infinitesimal generator of an analytic semi··
group then - A - aI is invertible and generates a bounded analytic semi-
group for a > 0 large enough. This enables one to reduce the general case
where - A is the infinitesimal generator of an analytic semigroup to the case
where the semigroup is bounded and - A is invertible.
From our assumptions on A and the results of Section 2.2.6 it follows that
Aa can be defined for 0 :'S: a .:0; I and A a is a closed linear invertible operator
with domain D(Aa) dense in X. The closed ness of A a implies that D(A a )
endowed with the graph norm of Aa, i.e. the norm IlIxlli = Ilxll + IIAaxll, is
a Banach space. Since Aa is invertible its graph norm III . III is equivalent to
the norm Ilxlla = IIAaxli. Thus, D(Aa) equipped with the norm II lIa is a
Banach space which we denote by Xa' From this definition it is clear that
o< a < fJ implies X" :;) Xp and that the imbedding of Xp in X" is continu-
ous.
Our main assumption concerning the function f in (3.1) will be,
Assumption (F). Let U be an open subset of IR + X X". The function f: U -+ X

satisfies the assumption (F) if for every (t, x) E U there is a neighborhood
V c U and constants L ~ 0, 0 < it ::;; 1 such that
Ilf(t l , XI) - f(t 2 , x 2 )n ::;; L(lt l - t21~ + IlxI - x2 11,,) (3.2)

for all (ti' Xi) E V.
We can now state and prove the main existence result of this section.
Theorem 3.1. Let -Abe the infinitesimal generator of an analytic semi-

group T(t) satisfying 11T(t)11 ::;; M and assume further that 0 E p( -A). If,
o < ex < 1 and f satisfies the assumption (F) then for every initial data
(to, xo) E U the initial value problem (3.1) has a unique local solution
u E C([t o, t I[: X) n CI(]t o, t I[: X) where tl = tI(t o, x o) > to.
PROOF. From our assumptions on the operator A it follows (Theorem

2.6.13) that
IIA"T(t)1I ::;; c"r" for t > O. (3.3)
For the rest of the proof, we fix (to, xo) E U and choose ti > to' ~ > 0 such
that the estimate (3.2) with some fixed constants L and it holds in the set
V= {(l,x): to::;; t::;; ti, IIx - xolI,,::;; ~}. Let,
B= max Ilf(t,xo)1I (3.4)
105.(5.('1
and choose t I such that

II T(t - to)A"xo - A"xoll < ~/2 (3.5)
and
. { ti - to, [ "2(1
0< tl - to < mm ~ - a)C,,-I(B + ~L) -I] I/I-,,} .
(3.6)
Let Y be the Banach space C([t o, tJl: X) with the usual supremum norm
which we denote by II . II y. On Y we define a mapping F by
Fy(t) = T(t - to)A"xo + [A"T(t - s)f(s, A-"y(s)) ds. (3.7)

to
Clearly, F: Y -+ Y and for every y E Y, Fy(to) = A"x o. Let S be the

nonempty closed and bounded subset of Y defined by
S = {y: y E Y,y(t o ) = A"x o ' lIy(t) - A"xoll .:s; ~} (3.8)
For yES we have,

IIFy(t) - Aaxoll ~ II T(t - to)Aaxo - Aaxoll
+ II{AaT(t - s)[J(s, A -ay(s)) - f(s, x o)] dsll
+ II{AaT(t - s)f(s, x o) dsll

~ ~ + Ca(L8 + B) [(t - s)-a ds
to
= ~ + Ca(I - a)-I(L8 + B)(t l - to)l-a ~8

where we used (3.2), (3.3), (3.6) and (3.8). Therefore F: S ..... S. Further-
more, if YI' Y2 E S then
IIFYI(t) - FY2(t) II
~ [IIAaT(t - s)llllf(s, A-aYI(S)) - f(s, A- aY2(S))llds

to
which implies
for YI' Y2 E S. (3.9)

By the contraction mapping theorem the mapping F has a unique fixed
point yES. This fixed point satisfies the integral equation
y(t) = T(t - to)Aa xo + [AaT(t - s)f(s, A-ay(s)) ds

to
for to ~ t ~ tl. (3.10)

From (3.2) and the continuity of y it follows that t ..... f(t, A -ay(t)) is
continuous on [to' t d and a fortiori bounded on this interval. Let
(3.11)
Next we want to show that t ..... f(t, A -ay(t)) is locally Holder continuous
on lto' til. To this end we show first that the solution y of (3.10) is locally
Holder continuous on lto' td.
We note that for every f3 satisfying 0 < f3 < I - a and every 0 < h < I
we have by Theorem 2.6.13.
!1(T(h) - I)A"T(t - s)11 ~ CphPIIA,,+,BT(t - s)11 ~ ChP(t - s)-(a+.Bl.

(3.12)
If to < t < t+ h s t I' then

Ily(t + h) - y(t)11 s II(T(h) - J)AaT(t - to)xoll
+ [1I(T(h) - J)AaT(t - s)J(s, A-ay(s))lIds
10
+ f
I
l+h
IIAaT(t + h - s )J(s, A -ay(s)) lids
= II + 12 + 13 • (3.13)
Using (3.11) and (3.12) we estimate each of the terms of (3.13) separately.
II s C(t - to)-(a+P)hPllxoll S Mlh P (3.14)
12 s CNh Pret - s) -(a+p) ds s M 2h P (3.15)
10
(3.16)
Note that M2 and M3 can be chosen to be independent of t E [to' td while

MI depends on t and blows up at t t to. Combining (3.13) with these
estimates it follows that for every tb > to there is a constant C such that
Ily(t) - y(s)1I s Cit - siP for to < tb s t, s S tl (3.17)
and therefore y is locally Holder continuous on ]to' t]. The local Holder
continuity of t ~ J(t, A-ay(t)) follows now from
IIJ(s, A-ay(s)) - J(t, A-ay(t))11 s L(lt - sl~ + Ily(t) - y(s)ll)
s C I (It - sl~ + It - siP). (3.18)
Let y be the solution of (3.10) and consider the inhomogeneous initial
value problem
{duj,') _+ Au(t) ~ /(t, A-"y(t)) (3.\9)

u(t o ) - xo·
By Corollary 4.3.3. this problem has a unique solution u E CI(]to, tIl: X).
The solution of (3.19) is given by
u(t) = T(t - to)xo + [T(t - s )J(s, A -ay(s)) ds. (3.20)
10
For t > to each term of (3.20) is in D(A) and a fortiori in D(Aa). Operating
on both sides of (3.20) with Aa we find
Aau(t) = T(t - to)Aa xo + [A"T(t - s)J(s, A-ay(s)) ds. (3.21)
10
But by (3.10) the right-hand side of (3.21) equals y( t) and therefore

u(t) = A-ay(t) and by (3.20), u is a CI(]to, td: X) solution of (3.1). The
uniqueness of u follows readily from the uniqueness of the solutions of

(3.10) and (3.19) and the proof is complete. D
Theorem 3.1 states that under suitable conditions we have a continuously

differentiable solution of the initial value problem (3.1) on the interval
lto' t d. More is actually true. The derivative u' of the solution is locally
Holder continuous on lto' td. This is a consequence of the following
regularity result.
Corollary 3.2. Let A and f satisfy the assumptions of Theorem 3.1 and assume
further that f satisfies (3.2) for every (t, x) E U (i.e. the constants it and L are
uniform in U). If u is the solution of the initial value problem (3.1) on [to, t d
then du/ dt is locally Holder continuous on lto, t ,1 with exponent I' = min (it, (3)
for any 13 satisfying 0 < 13 < I - lX.
PROOF. Let 0 < 13 < I - lX. In the proof of Theorem 3.2 we showed that if
u is the solution of the initial value problem (3.1) then for every to < tb < t"
f(t, u(t» is Holder continuous on [tb, td with exponent I' = min(i}, (3).
From Theorem 4.3.5 it then follows that for every t~ > tb, du/dt is Holder
continuous on [t~, ttl with the same exponent 1'. D
We conclude this section with a result on the existence of global solutions

of (3.1).
Theorem 3.3. Let 0 E p( - A) and let - A be the infinitesimal generator of an

analytic semigroup T( t) satisfying II T( t) II :s; M for t ~ O. Let f: [to' oo[ X X"
~ X satisfy (F). If there is a continuous nondecreasing real valued function
k(t) such that
Ilf(t, x)11 :s; k(t)(1 + Ilxll,,) (3.22)
then for every X o E X" the initial value problem (3.1) has a unique solution u
which exists for all t ~ to.
PROOF. Applying Theorem 3.1 we can continue the solution of (3.1) as long
as Ilu(t)ll" stays bounded. It is therefore sufficient to show that if u exists on
[0, T[ then Ilu(t)ll" is bounded as tiT. Since
A"u(t) = A"T(t - to)x o + [A"T(t - s)f(s, u(s» ds

to
it follows that
which implies by Lemma 5.6.7 that lIu(t)ll" :s; Con [0, T[ and the proof is
complete. D
200 Scmigroup, of Linear Operator,
6.4. A Quasilinear Equation of Evolution

In this section we will discuss the Cauchy problem for the quasi linear initial
value problem
{
dUd~t) + A (t, u) u = ° for 0.:0; t .:0; T
(4.1 )
u(o) = Uo
in a Banach space X.
The initial value problem (4.1) differs from the semi Ii near initial value
problems that were treated in the previous sections by the fact that here the
linear operator A(t, u) appearing in the problem depends explicitly on the
solution u of the problem, while in the semilinear case the nonlinear
operator was the sum of a fixed linear operator (independent of the solution
u) and a nonlinear" function" of u.
In general the study of quasi linear initial value problems is quite com-
plicated. For the sake of simplicity we will restrict ourselves in this section
to a rather simple framework starting with mild solutions of the initial value
problem (4.1). We begin by indicating briefly the general idea behind the
definition and the existence proof of such mild solutions.
Let u E C([O, T] : X) and consider the linear initial value problem
dv
{ dt + A ( t , u) v = ° for 0.:0; t .:0; T
( 4.2)
v(o) = Uo
If this problem has a unique mild solution v E C([O, T]: X), for every given
u E C([O, T]: X), then it defines a mapping u -+ v = F(u) of C([O, T]: X)
into itself. The fixed points of this mapping are defined to be mild solution
of (4.1).
To prove the existence of a local mild solution of (4.1) we will show that
°
under suitable conditions, there exists always a T', < T' .:0; T such that the
restriction of the mapping F to C([O, T']: X) is a contraction which maps
some ball of C([O, T']: X) into itself. The contraction mapping principle
will then imply the existence of a unique fixed point u of F in this ball and u
is then, by definition, the desired mild solution of (4.1).
In order to carry out the program as indicated above, we will need some
preliminaries. We start with the existence of mild solutions of the linear
initial value problem (4.2). To this end we modify the assumptions (HI )-( H)
of section 5.3 so that they depend on an additional parameter.
°
Definition 4.1. Let B be a subset of X and for every .:0; t .:0; T and b E B
let A( t, b) be the infinitesimal generator of a Co semigroup St. b(S), S ? 0, on
X. The family of operators (A(t; b)}, (t, b) E [0, T] X B, is stable if there
are constants M 2 1 and w such that

p(A(t, b)) :) ]w, oo[ for (t,b)E[O,T]XB (4.3)
and
for A> w (4.4)
and every finite sequences °: :; t] :::; t2 :::; ... :::; tk :::; T, bj E B, 1 :::; j :::; k.
It is not difficult to show (see proof of Theorem 5.2.2) that the stability of
{A(t, b)}, (t, b) E [0, T] X B implies that
(4.5)
and any finite sequences 0:::; t] :::; t 2 :::; ... :::; tk :::; T, bj E B, 1 :::;j:::; k.
Let X and Y be Banach spaces such that Y is densely and continuously
imbedded in X. Let B c X be a subset of X such that for every (t, b) E
[0, T] X B, A(t, b) is the infinitesimal generator of a Co semigroup St.b(S),
S 2 0, on X. We make the following assumptions:
(H]) The family {A(t, b)}, (t, b) E [0, T] X B is stable.

(H2) YisA(t, b)-admissible for (t, b) E [0, T] X Band thefamily{A(t, b)},
(t, b) E [0, T] X B of parts A(t, b) of A(t, b) in Y, is stable in Y.
(H3) For (t, b) E [0, T] X B, D(A(t, b»:) Y, A(t, b) is a bounded linear
operator from Y to X and t --+ A(t, b) is continuous in the B(Y, X)
norm II II Y--X for every bE B.
(H4) There is a constant L such that
IIA(t, b]) - A(t, b2 )11 y--x:::; Lllb] - b2 11 (4.6)
holds for every b], b2 E Band °: :; t :::; T.
Lemma 4.2. Let Be X and let u E C([O, T]: X) have values in B. If

{A(t, b)}, (t, b) E [0, T] X B is a family of operators satisfying the assump-
tions (H])-(H4) then {A(t, U(t)}tE[O. T] is a family of operators satisfying the
assumptions (H])-(H3) of Theorem 5.3.1.
PROOF. From (H]) and (H2) it follows readily that ~A(t, U(t)}tE[O. T] satisfies
(H]) and (H2 ). Moreover it is clear from (H3) that for t E [0, T]
D(A(t, u(t» :) Yand that A(t, u(t» is a bounded linear operator from Y
to X. It remains only to show that t --+ A(t, u(t» is continuous in the
B(Y, X) norm. But by (H4) we have
IIA(t], u(t])) - A(t2' u(t 2 ))11 Y--X
:::; IIA(t], u(t])) - A(t2' u(t]))11 y->X + C!lu(t]) - u(t 2)1I. (4.7)
Since u(t) is continuous in X, the continuity of t --+ A(t, b) together with
(4.7) imply the continuity of t --+ A(t, u(t» in the B(Y, X) norm. 0
As a consequence of Lemma 4.2 and Theorem 5.3.1 we now have:
Theorem 4.3. Let Be X and let {A(t, b)}, (t, b) E [0, T] X B be afamity of
operators satisfying the conditions (HI )-(H4). If u E C([O, T]: X) has values
in B then there is a unique evolution system Uu(t, s), °:;
s :;; t :;; T, in X
satisfying
for 0:;; s :;; t :;; T (4.8)
for wE Y, 0:;; s:;; T (4.9)
a
as Uu(t, s)w = - Uu(t, s )A(s, u(s ))w
for wE Y, °:; s :;; t:;; T. (4.10)
For every function u E C([O, T]: X) with values in Band U o E X the

function v(t) = Uu(t,O)u o is defined to be the mild solution of the initial
value problem (4.2). From Theorem 4.3 it therefore follows that if the family
{A(t, b)},(t, b) E [0, T] X B, satisfies the conditions (HI)-(H4) then for
every U o E X and u E C([O, T] : X) with values in B the initial value
problem (4.2) possesses a unique mild solution v given by
(4.12)
In the sequel we will need also the following continuous dependence
result.
Lemma 4.4. Let Be X and let {A(t, b)}, (t, b) E [0, T] X B, satisfy the
conditions (H I )-(H4 ). There is a constant C I such that for every u, v E
C([O, T]: X) with values in B and every w E Y we have
II U)t, s)w - Uv(t, s )wlI :;; Cdlwll y[lIu( T) - v( T)lldT. (4.13)

s
PROOF. As in the proof of Theorem 5.3.1 we obtain for every w E Y the

estimate,
IIUu(t,s}w - Uv(t.s}wlI:;; Cllwlly[IIA(T,u(T}} -A(T,v(T})lly .... x dT

s
(4.14)
where C depends only on the stability constants of {A(t, b)} and {A(t. b)}.
Combining (4.14) with (H4) yields (4.13). D
We turn now to the existence of local mild solutions of the initial value
problem (4.1). In the first result the initial value U o will be assumed to be in
Y and B will be a ball of radius r in X centered at uo.
Theorem 4.5. Let U o E Yand let B = {x: IIx - uoll :s; r}, r > O. If{A(t, b)},
(t, b) E [0, T] X B satisfy the assumptions ( ill )-( i14 ) then there is a T,O <
T :s; T such that the initial value prohlem
du
f dt + A{t, u)u = 0, {J::::t::::1"
(4.15 )
\u(O) = U o
has a unique mild solution u E C([O, T]: X) with u(t) E B for 0 :s; t :s; T.
PROOF. We note first that the constant function u(t) == U o satisfies the
assumptions of Theorem 4.3 and there is therefore an evolution system
Uuo(t, s), 0 :s; s :s; t :s; T associated to u o. Let 0 < t \ :s; T be such that
r
max IIUu (t,O}u o - uoll <-2
O~t~tl 0
and choose
(4.16)
where C is the constant appearing in Lemma 4.4. On the closed subset ;i) of
C([O, T]: X) defined by
5i = {u: u E C([O, T] : X), u(O) = uo, lIu(t) - uoll :s; r for O:s; t:s; T}
(4.17)
we consider the mapping
Fu(t) = Uu(t,O)u o for O:s; t :s; T. (4.18)
By our assumptions and Theorem 4.3 it is clear that F is well defined on ~
and that its range is in C([O, T): X). We claim that F: ~ --+ ~. Indeed, for
u E ;i) we clearly have Fu(O) = U o and by Lemma 4.4 and (4.16),
IIFu(t} - uoll :s; IIUu(t,O)u o - UuJt,O)uoll + IIUuJt,O)u o - uoll
-<" C'· : T'

r"u();lY -t- !....--
2 - r.
Moreover, if u l , U2 E ;i) then by Lemma 4.4
IIFul(t) - Fu 2 (t)H = IIUu,(t,O)u o - Uu2 (t,0)u oll
:s; Ciluolly[IIU\(T} - u 2 (T)lldT

o
:s; CiluollyTllu l - u2 1100 :s; ~llul - u2 1100
where II 1100 is the usual supremum norm in C([O, T] : X). From the last
inequality it follows readily that
IIFu l - Fu 2 1100:S; ~lIul - u2 1100 (4.19)
so that F is a contraction. From the contraction mapping theorem it follows
that F has a unique fixed point u E ~ which is the desired mild solution of
(4.15)on[0,T]. 0
A different version of Theorem 4.5 which is often very useful in the

applications of the theory to partial differential equations, is obtained by
restricting the set B that appears in the conditions (HI )-( H 4 ) to a ball in Y
rather than a ball in X as we have assumed above. The price that we have to
pay for this relaxation of the conditions are the following further assump-
tions,
(Hs) For every u E C([O, T): X) satisfying u(t) E B for 0:::; t:::; T, we
have
0:::; s :::; t:::; T (4.20)
and Uu(t, s) is strongly continuous in Y for
(H6) Closed convex bounded subsets of Yare also closed in X.
°: :;
s :::; t :::; T.
We note that the condition (H6) is always satisfied if X and Yare

reflexive Banach spaces. We now have:
Theorem 4.6. Let U o E Y and let B = {y: Ily - uoll y :::; r}, r> 0. Let
{A(t, b)}, (t, b) E [0, T) X B be a family of linear operators satisfying the
assumptions (HI)-(H6). If A(t, b)u o E Yand
IIA{t, b)uoll y k for (t, b) E [0, T] X B (4.2l)
°
:::;
then there exists aT', < T' :::; T such that the initial value problem (4.15)
has a unique classical solution u E C([O, T') : B) n CI([O, T') : X).
PROOF. We start by showing the existence of a unique mild solution of
(4.15). We note first that from the construction of Uu(t, s) and (Hs) it
follows that
for 0:::; s :::; t :::; T (4.22)
and every u C([O, T') : X) with values in B. After choosing
-I}
E
r 1
T' = min { T, kC 1 ' "2( Clluolly + 1) ( 4.23)
where C is the constant appearing in Lemma 4.4, we consider the subset ~

of C([O, T') : X) defined by
~ = {u: u E C{[O, T'] : X), u{O) = un' u{t) E B for 0:::; t:::; T'}.
From (H6 ) it follows that ~ is a closed convex subset of C([O, T'): X). Next
we define on ~ the mapping
Fu{t) = Uu{t,O)uo 0:::; t:::; T' (4.24)
and show that F: ~ -+ ~. Clearly Fu(O) = U o and Fu(t) E C([O, T'): X).
From (Hs) it follows that Fu(t) E Y for 0:::; t :::; T' and it remains to show
that IIFu(t) - uoll v :::; r for 0:::; t:::; T'. Integrating (4.10) in X from s to t
we find
(4.25)
Estimating (4.25) in Yand using (4.21), (4.22) and (4.23) yields

liEu(t) - uoll y = II Uu(l, O)u o - uoll y:;; C.kT:;; r.
Therefore F: ~ --+ ~. Now, exactly as in the proof of Theorem 4.5, we also
have for any u.' u 2 E ~
liEu. - Fu 2 11co :;; ~lIu. - u2 11co
where II II 00 is the supremum norm in C([O, T'] : X). Thus by the contrac-
tion mapping theorem F has a unique fixed point u E ~ which is the mild
solution of (4.15) of [0, T']. But u(t) = Uu(t, O)u o and therefore by (H5) and
Theorem 5.4.3, u is the unique Y-valued solution of the linear evolution
equation
dv
{ dt + A ( t, u) v = 0 (4.26)
v(O) = U o
and thus u is a classical solution of (4.15) and u E C([O, T']: y) n
C·([O, T']: X). The uniqueness of u is obvious and the proof is complete. 0
CHAPTER 7
Applications to Partial Differential

Equations-Linear Equations
7.1. Introduction
The theory of semigroups of linear operators has applications in many
branches of analysis. Such applications to Harmonic analysis, approxima-
tion theory, ergodic theory and many other subjects can be found in the
general texts that are mentioned at the beginning of the bibliographical
remarks.
In the present and following chapter we will restrict our attention to
applications which are related to the solution of initial value problems for
partial differential equations. The different sections of these two chapters
are essentially independent of each other and each one of them describes a
special application. Basic results from the general theory of partial differen-
tial equations which will be used will be stated, without proof, when needed.
In the applications of the abstract theory, it is usually shown that a given
differential operator A is the infinitesimal generator of a Co semigroup in a
certain concrete Banach space of functions X. This provides us with the
existence and uniqueness of a solution of the initial value problem
au(t'X)_A{
a - u t, x )
{ (l.l)
u{o,~) = uo{x)
in the sense of the Banach space X. The solution u thus obtained may
actually be a classical solution of the initial value problem (1.1). If this is the
case, it is usually proved by showing that u is more regular than the
regularity provided by the abstract theory. A common tool in such regular-
ity proofs is the Sobolev's imbedding theorem, a version of which will be
stated at the end of this section.
7 Application;, to Partial Differential Equations-Linear Equations 207
We turn now to the description of the main concrete Banach spaces that
will be used in the sequel. In doing so we will use the following notations;
x = (x I' Xl" .. , XII) is a variable point in the n-dimensional Euclidean
space 1R". For any two such points x = (XI"'" XII)' Y = (YI"'" Yn) we set
X . Y = L;'~IXiYI and IxI2 = X· x.
An n-tuple of nonnegative integers a = (a" a 2 , . .. , an) is called a multi-
index and we define
n
and
for x=(x"x 2 , ... ,x lI ) .
Denoting Dk = a/ax k and D = (D" D 2, ... , Dn) we have
a'" a'" a""
D" = D"'D'" ... D a " = - - - - ... - - .
'2 n axf' ax~2 ax~"
Let Q be a fixed domain in ~n with boundary aQ and closure Q. We will
usually assume that aQ is smooth. This will mean that aQ is of the class C k
for some suitable k z 1. Recall that aQ is of the class C k if for each point
x E aQ there is a ball B with center at x such that aQ n B can be
represented in the form Xi = <p(x" . .. , x i -" Xi +', . .. , XII) for some i with <p
k-times continuously differentiable.
By C»l(Q) (C»l(Q)) we denote the set of all m-times continuously
differentiable real-valued (or sometimes complex-valued) functions in Q (Q).
C;'(Q) will denote the subspace of C»l(Q) consisting of those functions
which have compact support in Q.
For u E C»l(Q) and 1 .:s; p < 00 we define
'IP
lIullm,p =
( 1L
QI"I<;m
ID"uI P dx ) ( 1.2)
If p = 2 and u, v E Cm(Q) we also define
(u,v)m=l L D"uD"vdx. (1.3)

QI"'I<;m
Denoting by Cpm(Q) the subset of Cm(Q) consisting of those functions u for
which Ilull»l,p < 00 we define _Wrn.p(Q) and Worn,p(Q) to be the comple-
tions in the norm II . Ilm.p of Cpm(Q) and Com(Q) respectively.
It is well known that wm,p(Q) and Worn,p(Q) are Banach spaces and
obviously Wom,p(Q) c wrn,p(Q). For p = 2 we denote wrn,2(Q) = H»l(Q)
and Wom,2(Q) = HO'(Q). The spaces Hrn(Q) and Hom(Q) are Hilbert spaces
with the scalar product ( , )m given by (l.3).
The spaces W m , P(Q) defined above, consist of functions u E LP(Q)
whose derivatives D"'u, in the sense of distributions, of order k .:s; m are in
LP(Q).
If n is a bounded domain then the Holder inequality implies

for 1:::; r:::;p ( 1.4)
and the imbedding is continuous. Furthermore we have:
Theorem 1.1. Let n be a bounded domain in R n with a smooth boundary an

(e. g. an is of the class C 1) and let 1 :::; r, p < 00. If j, m are integers such that
O:::;j < m and
1 1 j m
->-+--- (1.5)
p r n n
then wm.r(n) C W}·p(n) and the imbedding is compact.
Some relations between the space Wm·p(n) and the spaces of continu-
ously differentiable functions Ck(n) and integrable functions U(n), r ~ 1
are given next.
Theorem 1.2 (Sobolev). Let n be a bounded domain in R n with a smooth

boundary an (e. g. an is of class C m) then,
wk·p(n) c up/(n-kp)(n) for kp < n (1.6)
and
n
for
0:::; m < k - - . (1.7)
p
MOre01)er, there exist constants C 1 and C2 such that for any u E wm·p(n)
for kp < n (1.8)
and
n
sup{ID"'u(x)l: lal :::; m, x E O}:::; C21lullk.p for 0:::; m < k - - .
P
(1.9)
The space Wok. p(n) is the subspace of elements of Wk. p(n) which vanish
in some generalized sense on an. It can be shown that if an is of class C k
and u E Ck-1(O) n wtp(n) then u and its first k - 1 normal derivatives
vanish on an and conversely, if u E Ck(O) and its k - 1 first normal
derivatives vanish on an then u E Wok. p(n).
7.2. Parabolic Equations-L 2 Theory

Let n be a bounded domain in R n with smooth boundary an. Consider the
differential operator of order 2m,
A(x,D)= L a",(x)D'" (2.1 )
1"'1 :s;2m
7 Applications to Partial Differential Equations- Linear Equations 209
where the coefficients a,,(x) are sufficiently smooth complex-valued func-

tions of x in n.
The principal part A'(x, D) of A(x, D) is the operator
A'(x, D) = L aa(x)Da. (2.2)
1"1 =2m
Definition 2.1. The operator A(x, D) is strongly elliptic if there exists a

constant c > 0 such that
Re (_l)m A'(x, ~) ;:::: cl~12m (2.3)
for all x E n and ~ E IRn.
For strongly elliptic operators we have the following important estimate.
Theorem 2.2 (Garding's inequality). If A(x, D) is a strongly elliptic operator

of order 2m then there exist constants Co > 0 and Ao ;:::: 0 such that for every
u E H2m(Q) n Hom(Q) we have
Re(Au, u)o;:::: collull~,2 - AolluIl6,2' (2.4)
The proof of Garding's inequality is usually based on the definition of

strong ellipticity and the use of the Fourier transform. For certain simple
cases it can be obtained through integration by parts. Consider for example
the operator - ~ given by
- ~ is clearly strongly elliptic and for every u E q'0(Q) we have
(-~u,u)o= -lu~udx=l\1u.
Q Q
\1udx= Ilu11T2-llu1162 (2.5)
"
where the second equality follows from integration by parts while the last
equality is a direct consequence of the definition of the norms II 11m 2' A
simple limiting argument shows that (2,5) stays true for every u E H2(Q) n
HJ(Q) and (2.4) holds for -~.
Let A(x, D) be a strongly elliptic operator of order 2m with smooth
coefficients aa(x) in n.
Using integration by parts (Au + AU, v)o can be
extended to a continuous sesquilinear form on Hf{'(Q) X Hom(Q) for every
complex A. If Re A ;:::: Ao then it follows from Garding's inequality (2.4) that
this form is coercive. We may therefore apply the classical Lax-Milgram
lemma and derive the existence of a unique weak solution u E Hom(Q) of the
boundary value problem
A(x, D)u + AU = f (2.6)
for every f E L2(Q) and Re A;:::: Ao. It can then be shown, but this is not
easy, that the weak solution of (2.6) actually satisfies u E H2m(Q) and
therefore we have,
Theorem 2.3. Let A(x, D) be strongly elliptic of order 2m. For every A
satisfying Re A ~ Ao and every f E L2(Q) there exists a unique u E H2m(Q)
n H(;'(Q) satisfying the equation
A{x, D}u + AU = f.
With a given strongly elliptic operator A(x, D) on a bounded domain

Q c IR n we associate an unbounded linear operator A acting in the Hilbert
space H = L2(Q). This is done as follows:
Definition 2.4. Let A(x, D) = L 1al <;2maa(x)DIX be strongly elliptic in Q

and set D(A) = H2m(Q) n Hom(Q). For every u E D(A) define
Au = A{x, D}u.
With this definition we have:
Theorem 2.5. Let H = L2(Q) and let A be the operator defined above. For
every A satisfying Re A ~ AO the operator - AA = - (A + AI) is the infinites-
imal generator of a Co semigroup of contractions on H = L2(Q).
PROOF. Clearly D(AA) = D(A) ::J qO(Q). Since CO"(Q) is dense in H it

follows that D(AA) is dense in H. Also, from Garding's inequality we have,
Re{ -AAU, u}o s -collull~.2 + (A O - Re A}ll u lll. 2·

Since ReA~Ao' Re(-AAu,u)osO and -AA is dissipative. Finally if
Re A ~ Ao then the range of ILl + AA is all of H for every IL > O. This is a
direct consequence of Theorem 2.3. From Theorem 1.4.3 it follows now that
- AA is the infinitesimal generator of a Co semi group of contractions on
H = L2(Q). 0
An immediate consequence of Theorem 2.5 is:
Corollary 2.6. Let A(x, D) be a strongly elliptic operator of order 2m on a

bounded domain Q with smooth boundary eJQ in IRn. For every U o E H2m(Q)
n Ho(Q) the initial value problem
au{t,x}
{ at +A{x,D}u{t,x}=O in Q
{2.7}
u{O,x}=uo{x}
has a unique solution u(t, x) E CI([O, 00[: H2m(Q) n Ho(Q».
Theorem 2.5 implies that if A(x, D) is a strongly elliptic operator then

- A, defined by definition 2.4, is the infinitesimal generator of a Co
7 Applications to Partial Differential Equations~ Linear Equations 211
semigroup on H = L2(Q). Actually more is true in this case. Indeed, we

have:
Theorem 2.7. If A(x, D) is a strongly elliptic operator of order 2m then the

operator - A (given by definition 2.4) is the infinitesimal generator of an
analytic semigroup of operators on H = L 2(Q).
PROOF. Let AAO = A + AoI. From Garding's inequality we have

(2.8)
A simple integration by parts yields for every u E D(AAO)
IIm(AAo u , u)ol ::;; I(AAO U , u)ol ::;; bllull~,2 (2.9)
for some constant b > O. From (2.8) and (2.9) it follows that the numerical
range S(AAO) of AAO satisfies
S(AAJ C SIf\ = {A: - tJ 1 < arg A < tJ 1} (2.10)
where tJ I = arctan( b/ co) < 'IT /2. Choosing tJ such that tJ I < tJ < 'IT /2 and
denoting ~If = {A: larg AI > tJ} there exists a constant elf such that
for all A E ~If (2.11)
where d(A: S) denotes the distance between A and the set SeC. From
Theorem 2.3 it follows that all real p" p, < 0 are in the resolvent set of AAO
and therefore ~If is contained in a component of the complement of S( AAJ
which has a nonempty intersection with P(AAO)' Theorem 1.3.9 then implies
that P(AA O) :=l ~If and that for every A E ~If'
(2.12)
and therefore - AAO is the infinitesimal generator of an analytic sernigroup

by Theorem 2.5.2 (c). This implies finally (see e.g., Corollary 3.2.2) that - A
is the infinitesimal generator of an analytic semigroup of operators on
L2(Q). D
As a direct consequence of Theorem 2.7 and Corollary 4.3.3 we have:
Corollary 2.S. Let A(x, D) be a strongly elliptic operator of order 2m in a

bounded domain Q c IR n and let f(t, x) E L2(Q) for every t :2: O. If
(2.13)
then for every uo(x) E L2(0) the initial value problem
~~ + A (x, D) u = f ( t, x) in 0 X R+
{ (2.14)
u(O, x) = U o in 0
has a unique solution u(t, x) E CI(]O, 00[: H 2m (0) () Hom(o».
Remark 2.9. It is worthwhile to note that if the operator A has constant

coefficients, Theorems 2.5 and 2.7 remain true for the domain 0 = Rn. The
proofs of this particular case can be carried out easily using the Fourier
transform.
7.3. Parabolic Equations-LP Theory

Let 0 be a bounded domain with smooth boundary in Rn. In the previous
section we considered seroigroups defined on the Hilbert space L 2 (0). It is
often useful to replace the Hilbert space L2(0) by the Banach space LP(O),
1 ~ p :$; 00. This is usually important if one wishes to obtain optimal
regularity results. In the present section we will discuss the theory of
semigroups associated with strongly elliptic differential operators in LP(O).
During most of the section we will restrict ourselves to the values 1 < p < 00.
Some comments on the cases p = 1 and p = 00 will be made at the end of
the section.
Let 1 < P < 00 and let 0 be a bounded domain with smooth boundary
ao in Rn. Let
A(x, D)u = L aa(x}Da u (3.1)
lal :s;2m
be a strongly elliptic differential operator in O. The operator

A*(x, D}u = L (_1) l a I Da (aa(x)u) (3.2)
lal :s;2m
is called the formal adjoint of A(x, D). From the definition of strong
ellipticity it is clear that if A(x, D) is strongly elliptic so is A*(x, D). The
coefficients aa(x) of A(x, D) are tacitly assumed to be smooth enough, e.g.
aa(x) E C 2m (Q) or aa(x) E Coo(Q). Many of the results however, hold
under the weaker assumptions that aa(x) E Loo(O) for 0 :$; lal < 2m and
aa(x) E C(Q) for lal = 2m. For strongly elliptic differential operators the
following fundamental a-priori estimates have been established.
Theorem 3.1. Let A be a strongly elliptic operator of order 2m on a bounded

domain 0 with smooth boundary ao in R n and let 1 < P < 00. There exists a
constant C such that

(3.3)
Using this a-priori estimate together with an argument of S. Agmon one

proves the following theorem.
Theorem 3.2. Let A be a strongly elliptic operator of order 2m on a bounded

n
domain
° °
with smooth boundary an in IR n and let 1 0, R ~ and < (t < '1T /2 such that
00. There exist
C
lIullo,p ~ III"P.I + A)ullo,p (3.4)
for every u E W 2m ,P(n) () Wom,P(n) and A E C satisfying IAI ~ Rand

{t - '1T < arg A < '1T - (t.
With a strongly elliptic operator A(x, D) we associate a linear (un-

bounded) operator Ap in LP(n) as follows:
Definition 3.3. Let A = A(x, D) be a strongly elliptic operator of order 2m

on a bounded domain n in IR n and let I <P< 00. Set
(3.5)
and
(3.6)
The domain D(Ap) of Ap contains Coo(n) and it is therefore dense in

LP(n). Moreover, from Theorem 3.1 it follows readily. that Ap is a closed
operator in LP(n).
With this definition we have:
Lemma 3.4. Let A(x, D) be a strongly elliptic operator of order 2m on nand

let A p' 1 < P < 00, be the operator associated with it by Definition 3.3. The
operator A:, q = p /( p - 1) associated by Definition 3.3 with the formal
adjoint A*(x, D) of A(x, D) on U(n) is the adjoint operator of Ap.
From Theorems 3.1 and 3.2 we deduce,

Theorem 3.5. Let A(x, D) be a strongly elliptic operator of order 2m on a
bounded domain n with smooth boundary aQ in IR n and let 1 < P < 00. If Ap
is the operator associated with A by Definition 3.3 then - Ap is the infinitesi-

mal generator of an analytic semigroup on LP(~).
PROOF. We have already noted that D(Ap) is dense in LP(~) and that Ap is
a closed operator in LP(~). From Theorem 3.2 it follows that for every
AE ~.., = {p.: i} - '1T < arg p. < '1T - i}, 1p.1 ~ R} (3.7)
the operator AI + Ap is injective and has closed range. Similarly, it follows

from Theorem 3.2, applied to the operator A: on U(~), that there are
constants R' ~ 0 and 0 < W < '1T/2 such that for every A E ~.." = {p.: i}' -
'1T < arg p. < '1T - i}', 1p.1 ~ R'} ')\1 + A: is injective. Let i}1 = min (i), i}')
and RI = max(R, R') then for every
AI + Ap is bijective. Indeed, we have already seen that it is injective so it

remains only to show that it is surjective. Let A E ~"'I' If v E Lq(~) satisfies
«AI + Ap)u, v) = 0 for all u ~ D(Ap) then it follows from Lemma 3.4
that v E D(A:) and that (u~ (AI + A:)v) = 0 for all u E D(Ap}- Since
D(Ap) is dense in LP(~), (AI + A:)v = 0 and the injectivity of AI + A:
implies v = O. Thus for A E ~"'I' AI + Ap is invertible and from (3.4) it
follows that
C
IXT
-1
II ( AI + Ap ) II ~
which, by Theorem 2.5.2 (c), implies that -Ap is the infinitesimal generator
of an analytic semigroup on LP(~). D
Theorem 3.5 is based on the deep a-priori LP estimate (3.3). For second
order strongly elliptic operators with real coefficients written in divergence
form, Theorem 3.5 can be proved directly without the use of Theorems 3.1
and 3.2. This will be done next.
Let ~ be a bounded domain with smooth boundary a~ in R n and let
A(x, D) be the symmetric second order differential operator given by
A(x, D)u = -
k,'=1
Ln axa (ak,,(x)-a
k
au ).
x,
(3.8)
We assume that the coefficients a k ,(x) = a, k(X) are real valued and
continuously differentiable in n
and 'that A(x: D) is strongly elliptic, i.e.
that there is a constant Co > 0 such that

n n
L a k ,(X)~k~' ~ Co L ~~ = Col~12 (3.9)
k./~ I k~ I
for all real ~k' 1 ~ k ~ n.

With the second order symmetric differential operator A(x, D) given by
(3.8) we associate the operator Ap on LP(O), 1 < p < 00, by Defintion 3.3.
We then have:
Theorem 3.6. Let 1 < P < 00, the operator - Ap is the infinitesimal generator
of an analytic semigroup of contractions on LP(g).
PROOF. Let 1 < p < 00 be fixed and let q = p/(p - 1). We denote the
pairing between LP(g) and Lq(g) by ( , ). If u E D(Ap) then the func-
tion u* = lul p- 2u is in U(O) and (u, u*) = IluIIC,p' Let 2 ~ p < 00,
integration by parts yields
(Apu, u*) = -1 L -aXa ( a ax,au ) ulul p- 2 dx

nk,/~1
n
k
k ,'-
= 1 L ak,l-a
nk./~1
au a
n
-a-(uluI P- 2 ) dx
XI Xk
But,
~ lul p- 2 = .!( p - 2) IU IP - 4( u au + u au ).
aX k 2 aX k aX k
Denoting, lul(P-4)j2 u(au/ax k ) = Oi k + i13 k we find after a simple computa-
tion that
1L
n
(Apu, u*) = ak,l((p - 1)) OikOi, + 13 k 13, + i(p - 2)Oi k 13,) dx.
nk.'~ I
(3.10)
Let 100 k ,(x)1 ~ M for I ~ k, I ~ n and x E Q and set
then it follows easily from (3.9) and (3.10) that
Re(Apu, u*) ~ Co((p - I)10i1 2 + 1131 2) ~ 0 (3.11)

and
Ip - 21M( ~ 10i1 2 + -h I/W)
(3.12)
Co((p - I)10i12 + 1131 2)
for every p > o. Choosing p = ~ in (3.12) yields
IIm(Apu,u*)1 Mlp-21
< . (3.13)
IRe ( Apu, u*)1 - 2Co~
From (3.11) it follows readily that for every A > 0 and u E D(A p ) we have
(3.14)
and therefore AI + Ap is injective and has closed range for every A > o.
Since (3.14) holds for every 2 :;;;. p < 00 it follows that for A > 0, AI + A p is
also surjective. Indeed, if v E U(Q) satisfies «AI + Ap)u, v) = 0 for all
u E D(Ap) then, since A(x, D) is formally self adjoint, Jt follows from
Lemma 3.4 that v E D(Aq), q = p/(p - 1), and that (~(AI + Aq)v) = 0
for every U E D(Ap). Since D(Ap) is dense in LP(Q), (AI + Aq)v = 0 and
(3.14), with p replaced by q, implies v = O. Thus, AI + Ap is bijective for
A > 0 and as a consequence of (3.14) we have
-1 1
II ( AI + Ap ) lIo.p :;;; I for A> O. (3.15)
The Hille-Yosida theorem (Theorem 1.3.1) now implies that -Ap is the
infinitesimal generator of a contraction semigroup on LP(Q) for every
2 :;;; p < 00. Finally, to prove that the semigroup generated by -Ap is
analytic we observe that by (3.11) and (3.13) the numerical range S( -Ap)
of -Ap is contained in the set S,'}j = {A: larg AI> 'Tf - ttl} where ttl =
arctan(Mlp - 21!2Co~), 0 < ttl < 'Tf /2. Choosing ttl < tt < 'Tf /2
and denoting
~I} = {A: larg AI < '1T - 1'J-} (3.16)
it follows that there is a constant CI} > 0 for which
for AE ~I}.
Since A> 0 is in the resolvent set p(-Ap) of -Ap by the first part of the
proof, it follows from Theorem 1.3.9 that p( -Ap) ::::> ~I} and that
( )
-1 1
II AI + Ap Ilo.p :;;; CI}IAI for AE ~I) (3.17)
whence by Theorem 2.5.2 (c), -Ap is the infinitesimal generator of an

analytic semi group on LP(fl) for 2:;;; p < 00. The case 1 < P < 2 is
obtained from the previous case 2 :;;; p < 00 by a duality argument. 0
7 Applications to Partial Differential Equations-Linear Equations 217
We turn now to the cases p = I and p = 00 and start with p = 00. Recall
that the norm in LOOm) is defined by
Ilullo.oo = esssup{lu(x)1 : x En}.
Let A(x, D) be the uniformly elliptic operator of order 2m given by (3.1)

and defined on a bounded domain n c Ill" with smooth boundary an. We
associate with A(x, D) an operator Aoo on LOO(n) as follows:
for all p > n,
on an for 0 ~ 1,81 < m} (3.18)
and
(3.19)
We note first that from Sobolev's theorem (Theorem 1.2) it follows that
D(Aoo) C C 2m - 1(n) and therefore, since by our assumptions an is smooth
the conditions DP u = 0 for 0 ~ 1,81 < m on an make sense. Moreover,
from the regularity of the boundary and the definition of D(Aoo) it follows
that D(A«,) c W 2m ,P(O) n Wom,P(O) = D(A p ) for every p > n. But -Aoo
is not the infinitesimal generator of a Co semigroup of operators on
LOO(O). The reason for this is that D(Aoo) is never dense in LOO(O). Indeed,
we have noted above that D(A oo ) c C(n) and therefore also D( Aoo) c
C(n), where D(Aoo) is the closure of D(A oo ) in the II 110,00 norm. Since
C(n) is not dense in LOO(O), D(A oo ) cannot be dense in LOO(O).
To overcome this difficulty we restrict ourselves to spaces of continuous
functions on n. We define
D(Ac) = {u: u E D(Aoo), A(x, D)u E C(n), A(x, D)u = 0 on an}
(3.20)
and
for u E D(AJ. (3.21)
The operator Ac thus defined is considered as an operator on the space:
C = {u: u E C(n), u = 0 on an} (3.22)
and we have:
Theorem 3.7. The operator - Ac is the infinitesimal generator of an analytic

semigroup on C.
The proof of Theorem 3.7 is based on a-priori estimates in the norms

II IIk.oo,
similar to the estimates (3.3). Since we have no such a-priori
estimates for the case p = 1, the results for this case will be derived in a
different way, which will exploit a duality between continuous functions and
L' functions. We start with a lemma.
Lemma 3.S. Let 0 be a bounded domain in IRn. For u E LI(O) we have
lIullo., = sup {fou(X)'P(X) dx: 'P E CO"(O), 1I'Pllo."" ~ I}. (3.23)
PROOF. Since for every 'P E Co""(O) satisfying 11'Pllo,oo ~ 1 we have
Ifou'Pdxl~ 1I'Pllo,ocllullo" ~ lIull o, ,
the sup on the right-hand side of (3.23) is clearly less or equal to Ilullo,I'
Since CO"(O) is dense in L'(O) it suffices to prove the result for U E CO"(O).
Let Pn(z) E COO(C) be such that Pn(O) = 0, IPn(z)1 ~ 1 and Pn(z) = i/lzl
for Izl ~ lin. Then Pn(u(x» E CO"(O) and IIPn(u(x»lIo,,,,, ~ 1. Also,
lim
n-+oo
1nu{x )Pn(U{X)) dx = lln u{x) Idx = lIullo"
and thus the sup on the right-hand side of (3.23) is larger or equal to lIullo,l'
o
We turn now to the definition of the operator A, associated with the
strongly elliptic operator A(x, D) given by (3.1), on the space L1(O).
Definition 3.9. Let A(x, D) be the strongly elliptic operator of order 2m on

the bounded domain 0 c Oil n with smooth boundary ao given by (3.1). Set,
D{A,) = {u: u E w2m-I,I(O) n Wom,I(O),A(x, D)u E LI(O)}
(3.24)
where A(x, D)u is understood in the sense of distributions. For u E D(A,),
A I is defined by
(3.25)
Theorem 3.10. The operator - A, is the infinitesimal generator of an analytic

semigroup on L'(O).
PROOF. Let
A(x, D)u = L aa(x)Da u
lals2m
and
A(x, D)u = L (-l)laIDa(aa(x)u).
lals2m
Let A(' be the operator associated with A(x, D) on the space C (given by
7 Applications to Partial Differential Equations-- Linear Equations 219
(3.22)). Since A(x, D) is strongly elliptic together with A(x, D) it follows

from Theorem 3.7 that - A, is the infinitesimal generator of an analytic
° °
semigroup on C. Theorem 2.5.2 then implies that there are constants M > 0,
R ~ and < {t < 'TT' /2 such that
(3.26)
for every,\ E ~,~ = {Jl: largJlI > {t, IJlI ~ R). Now, let U E D(AI)' From
Lemma 3.8 it follows that
Ilullo.1 = sUP{InU<PdX: <p E q'0(!:2), 11<pllo.",,::; I}. (3.27)
Since q'0(!:2) is contained in the range of ,\/ + A, for every ,\ E ~/} it

follows from (3.26) and (3.27) that
II ull o.1= sup {lnu(,\/ + AJ v dx : v E D( Ae), IIvllo.oo ::; MI,\I--I}

which implies that for every v E D(AJ, Ilvllo,oo ::; MI'\I-I we have
Ilull o.1::;lln u (,\/ + AJvdxl =11n('\/ + AI)UVdxl

::; II (,\ / + A I) u II o. I II v II 0.00 ::; MI'\ I - III (,\ / + A I) u II o. I'
Thus for every,\ E ~,~' ,\/ + AI is injective and has closed range. Moreover,
since D( A 2) c D( A I) the range of ,\ / + A I contains £2(!:2), which is dense
in LI(!:2), and therefore the range of,\/ + AI is all of LI(!:2) and
11('\/ + AI)-lllo.1 ::; MI'\I-I
for every ,\ E ~,~. From Theorem 2.5.2 it follows therefore that - A I is the
infinitesimal generator of an analytic semi group on LI(!:2). D
7.4. The Wave Equation

In this section we consider the initial value problem for the wave equation
in IR i.e., the initial value problem
11
a2~ = t:..u, for x E 1R11, t > °

{ at (4.1)
u(O, x) =
au
ul(x), Yt(O, x) = u 2 (x), for x E 1R11.
This problem is equivalent to the first order system:
for x E IR ", t > ° (4.2)

220 Semigroups of Linear Operator,
and
for x E 1R"
In order to use the theory of semigroups we are interested in showing that

the operator (~ ~) is the infinitesimal generator of a Co semigroup of
operators in some appropriately chosen Banach space of functions. It turns
out that the right space is the Hilbert space HI(lRn) X L2(1R").
The spaces Hk(IR") have been defined in Section 7.1. For the special case
where !2 = IR n they can be characterized in the following useful way: Let
f E L2(1R") and let
( 4.3)
be its Fourier transform. The function f E Hk(lRn) if and only if

(l + 1~12)k/2j(~) E L2(lRn). This characterization is a simple consequence
of ParsevaJ's identity and the elementary properties of the Fourier Trans-
form.
Given a vector U = [UI' u 2 ] E Cooo(lRn) X COOO(IR") we define the norm
It is easy to check that the completion of CO'(IR ") X COOO(IR ") with respect to
the norm III . III is the Hilbert space H = H I(IR ") X L 2(1R "). In this Hilbert
space we define the operator A associated with the differential operator
(~ ~) as follows;
Definition 4.1. Let

D(A) = H2(1R") X HI(IR") (4.5 )
and for U = lui' u 2 ] E D(A) let
AU = A[u l , u 2 ] = [U2' ~UI]' ( 4.6)
To prove that the operator A defined by (4.5), (4.6) is the infinitesimal
generator of a Co group of operators on H we will need the following simple
preliminaries.
Lemma 4.2. If p > 0 and f E Hk (IR "), k ~ 0, then there is a unique function
u E H k + 2 (1R") satisfying
u - p~u = f. (4.7)
PROOF. Let j(~) be the Fourier transform of f and let u(O =
(l+ pl~12)-?(n Since fE Hk(IR"), (J + 1~12)k/2ja) E L2(1R") and
therefore (l + 1~12)(k+2)/2u(~) E L2(Oin). If u is defined by
u(x) = (2?T)-n/2 ( eix'Eu(O d~

JR "
then u E Hk+2(Oi n ) and u is a solution of (4.7). The uniqueness of the
solution u of (4.7) follows from the fact that if wE Hk+2(Rn) satisfies
W - v ~W = 0 then w = 0 and therefore W = O. 0
Lemma 4.3. For every F = [/1' f2l E CO'(lR n) X q'''(lRn) and real A+-O the
equation
V-AAV=F (4.8)
has a unique solution V = [UI' u2l E Hk(lRn) X Hk-2(lRn) for every k ;;:: 2.
Moreover,
IIIVIII :::; (1-IAI- 1 )IIIFIII for 0 < IAI < 1. ( 4.9)
PROOF. Let A+-O be real and let WI' W 2 be solutions of
Wi - A2 ~Wi = /; i = 1,2. (4.10)
From Lemma 4.2 it is clear that such solutions exist and that Wi E Hk(lRn)
for every k ;;:: O. Set u l = WI + AW2, u 2 = w2 + A ~WI' It is easy to check
that V = [UI' u2l is a solution of (4.8) and therefore u l - AU 2 = fl and
u 2 - A ~UI = f2' Moreover, V E Hk(lRn) X Hk-2(lRn) for every k ;;:: 2. De-
noting (,)0 the scalar product in L2(lRn) we have
111£111 2 = (fl - ~fl' fl)o + (f2' f2)0
= (u l - AU 2 - ~UI + A ~U2' UI - AU 2 )0
+ (u 2 - A ~UI' U2 - A ~UI)O
;;:: (u l - ~UI' ul)o + Ilu 2115.2 - 21AI Re(u l , u 2 )0
;;:: (I - IAI)IIIVII1 2.
Therefore if 0 < IAI < 1,
III F 1112;;:: (1 - IAI)2111 V 1112. ( 4.11)
o
Lemma 4.3 shows that the range of the operator I - AA contains
C~(If~n) X C~(lRn) for all real A satisfying 0 < IAI < 1. Since the operator A
defined by Definition 4.1 is closed, the range of I - AA is all of H =
Hl(IJ~n) X L2(/R n) and we have
Corollary 4.4. For every FE Hl(IRn) X L2(/R n) and real A satisfying 0 <
IAI < 1 the equation
V- AAV= F (4.12)
III U III : : ; (1 - IAI) -1111 Fill. ( 4.13)
Theorem 4.5. The operator A, defined in Definition 4.1 is the infinitesimal

generator of a Co group on H = HI(Rn) X L2(Rn), satisfying
( 4.14)
PROOF. The domain of A, H2(1Rn) X HImn) is clearly dense in H. From

Corollary 4.4 it follows that (p'/ - A)-I exists for IJLI > 1 and satisfies
II(JLI-A)-Ill::::;; IJLI~ 1 for IJLI> 1. (4.15)
From Theorem 1.6.3 it follows that A is the infinitesimal generator of a

group T(t) satisfying (4.14). 0
Corollary 4.6. For every fl E H 2(R n), f2 E HI(Rn) there exists a unique
u(t, x) E CI([O, 00[; H2(Rn» satisfying the initial value problem
{ "~~:) :"/, (x)

u;(O, x) = f2(X).
(4.16)
PROOF. Let T(t) be the semigroup generated by A and set
then
a
at [u l ' u 2] = A[u l , U2] = [u 2' Ilu l ]
and UI is the desired solution. o
We conclude this section by showing that if the initial values fl' f2 in the
ini tial value problem (4.16) are smooth so is the solution. To this end we
note that Sobolev's theorem (Theorem 1.2) can be extended to the special
unbounded domain n = R II as follows:
Theorem 4.7. For 0::::;; m < k - nl2 we have

Hk(Rn) c C"'(Rn). ( 4.17)
PROOF. Let v E n
CO"(R ) then, as is well known, ~ava) E L2(Rn) for every
a and
Estimating Dav(x), by the Cauchy-Schwartz inequality, we find for every

N> n12,
IDav (x)12.::; (2'1T)-n1 (1 + 1~12rN d~11~12Ial(1 + 1~12tlt3(~)12d~

Rn Rn
1( t. 2)N+la l A(t.) 2 t.
.::; C l IR n 1 + 1.. 1 Iv .. I d .. .::; C2I1vII 2N+ lal,2 (4,18)
where C l and C2 are constants depending on Nand lal. Let u E Hk(lRn)

and let un E q'°(lRn) be such that un -> u in Hk(lRn). Then, from (4.18) it
follows that Dau n -> Dau uniformly in IR n for all a satisfying lal .::; m < k
- nl2 and therefore u E cm(lRn) as desired. 0
Consider now the initial value problem (4.16) with 11,12 E Co',o(lRn).
Clearly, [fl' 121 E D(Ak) for every k 2 1, where A is the operator defined in
Definition 4.1. Therefore [up u 2] = T(t)[fl' 12] E D(Ak) for every k 2 1
and in particular tiu l E L2(lRn) for all k 2 O. This implies that U l E
Hk(lRn) for every k 2 0 and from Theorem 4.8 it follows that u l , the
solution of (4.16), satisfies ul(t, x) E coo(lRn) for every t 2 O. With a little
more effort one can show that actually, ul(t, x) E COO(IR X IRn) and is a
classical smooth solution of (4.16), but we will not do this here.
7.5. A Schrodinger Equation

The Schrodinger equation is given by
I au
- - = llu - Vu (5.1 )
i at
where the function V is called the potential. We will consider this equation
in the Hilbert space H = L2(lRn). We start with the definition of the
operator Ao associated with the differential operator ill.
Definition 5.1. Let D(Ao) = H2(lRn) where the space H2(lRn) is defined in
Section 7.1. For u E D(Ao) let
Aou = i llu (5.2)
Lemma 5.2. The operator iAo is self adjoint in L2(lRn).

PROOF. Integration by parts yields
(-llu,v)o= -1 llu·vdx= -1 u·llvdx=(u,-llv)o

An Rn
and therefore iAo = - II is symmetric. To show that it is self adjoint it

suffices to show that for every A with 1m A =I' 0 the range of AI - iAo is
dense in L2(lRn). But, if f E q'''(lRn) then, using the Fourier transform, it

follows that
u(x) = (27T)-(n/2) r j(~)eiX'~ d~ (5.3)
JR " A+ 1~12
is in D(A o) = H2(lRn) and it is the solution of (AI - iAo)u = f. The range
of AI - iAo contains therefore CO'(lR n) and is thus dense in L2(lRn). 0
From Stone's theorem (Theorem l.l0.8) we now have:
Corollary 5.3. A o is the infinitesimal generator of a group of unitary operators

on L2(lRn).
Next we treat the potential V. To this end we define an operator V in

L2(lRn) by,
D(V) = {u: u E L2(lRn), V· u E I}(lR n )}
and for u E D(V), Vu = V(x)u(x).
Lemma 5.4. Let Vex) E LP(lRn). lfp > nl2 andp ~ 2 then for every e > 0
there exists a constant C( e) such that
(5.4)
where the norm II . II denotes the L 2 norm in IR n.
PROOF. If u E H2(lRn) then (1 + 1~12)a(~) E L2(lRn) and since p > nl2
we also have (1 + 1~12)-1 E LP(lRn). Using Holder's inequality and
Parseval's identity we have for q = 2 P1(2 + p)
::; (~n(I + 1~12rp d~r/p(~,,(I + ,~,2)2Ia(o,2d~r/2

::; Cp(IIAuil + IIull)·
Since p ~ 2, 1 ::; q ::; 2 and therefore by the classical theorem of Hausdorff
and Young we have IIullo.r::; IIallo,q where llr + l/q = 1. Thus,
IIullo,r::; Cp(IIAuil + IIull). (5.5)
Replacing the function u(x) in (5.5) by u(px), p > 0 and choosing an
appropriate p we can make the coefficient of IIAuil as small as we wish.
Given e > 0 we choose it so that
(5.6)
7 Applicatiom to Partial Differential Equations-Linear Equations 225
Finally, using Holder's inequality again we have
liVul1 2 = 1V HI!
2U 2 dx 1 I VIP dx )2/P( 1 lul dx )2/r
~ ( H'! H'!
r
and therefore by (5.6),

II Vull ~ II Vllo.pllullo.r ~ EII~ull + C(E)liuli
as desired. o
Theorem 5.5. Let Vex) be real, Vex) E LP(~"). If P > n12, p ~ 2 then
Ao - iV is the infinitesimal generator of a group of unitary operators on
L 2{~").
PROOF. We have already seen that the operator iAo is self adjoint (Lemma
5.2) and in particular ±Ao is m-dissipative. Since V is real the operator V is
symmetric and therefore iAo + V is a symmetric operator. To prove that it
is self adjoint we have to show that the range of I ± (Ao - iV) is all of
L2(~"). This follows readily from the fact that ±(Ao - iV) is m-dissipative
which in turn follows from the m-dissipativity of ±A o' the estimate
II Vull ~ E11Aou11 + C(e)llull for u E D(Ao)
and the perturbation Theorem 3.3.2. Thus, Ao - iV is self adjoint and by
Stone's theorem it is the infinitesimal generator of a group of unitary
operators on L 2{~ n). 0
Remark 5.6. Adding to V in Theorem 5.5 any real Vo such that Vo E

L""([Rn) will not change the conclusion of the theorem. This follows from
the fact that ± Vo is symmetric and bounded and therefore iAo + V + Vo
is again a self adjoint operator. The fact that the range of I ± (Ao - iV -
iVa) is all of L2([Rn) follows from the same fact for I ± (Ao - iV) and
Theorem 3.1.1.
7.6. A Parabolic Evolution Equation

In the previous sections we have applied the theory of semigroups to obtain
existence and uniqueness results for solutions of initial value problems for
partial differential operators. All these applications dealt with partial
differential operators which were independent of the t-variable. Once these
operators depend on t, the problem ceases to be autonomous and we have to
use the theory of evolution systems, as developed in Chapter 5, to obtain
similar results.
The use of the theory of evolution systems is technically more com-
plicated than the use of the semigroup theory. Therefore we will restrict
ourselves here only to one example of such an application which extends
some of the results of Section 7.3 to the non autonomous situation.
Let 1 < p < 00 and let Q be a bounded domain with smooth boundary
aQ in IR n. Consider the initial value problem
au
7ft + A(t, x, D)u = f(t, x) in Q X [0, T]
{ (6.1 )
Dau(t, x) = 0, lal < m, on aQ X [0, T]
u(O,x)=uo(x) in Q
where
A(t,x,D)= L aa(t,x)Da (6.2)
lal "",2m
with the notations introduced in Section 7.1. We will make the following
assumptions;
(HI) The operators A(t, x, D), t ~ 0, are uniformly strongly elliptic in Q
i.e., there is a constant c > 0 such that
(-I)mRe L aa(t, x)ga ~ clgl zm (6.3)
lal =2m
for every x E n, 0 ::; t::; T and g E IRn.
(Hz) The coefficients aa(t, x) are smooth functions of the variables x in n
for every 0::; t ::; T and satisfy for some constants C I > 0 and
0</3::;1
laa(t, x) - aa(s, x)1 ::; Cdt - siP ( 6.4)
for x E n, 0 ::; s, t ::; T and lal ::; 2m.
With the family A(t, x, D), t E [0, T], of strongly elliptic operators, we
associate a family oflinear operatorsA/t), t E [0, T], in LP(Q), I < P < 00.
This is done as follows:
and
Ap(t)u = A(t, x, D)u for u E D.
If U o E LP(Q) and f(t, x) E LP(Q) for every 0::; t::; T then a classical
solution u of the (abstract) initial value problem
du
{ dt + Ap(t)u = f
( 6.5)
u(O) = U o
in LP(Q) is defined to be a generalized solution of the initial value problem

(6.1). Recall that such a generalized solution u, if it exists, satisfies by its
definition; u(t, x) E WZm.p(Q) Ii Wom,P(Q) for every t > 0, du/dt exists, in
the sense of LP(Q) and is continuous on ]0, T], u itself is continuous on
[0, T] and satisfies (6.5) in LP(Q).
7 Applicatiom to Partial Diflercntial Equations-Linear Equations 227
The main result of this section is the existence and uniqueness of

generalized solutions of (6.1) under the assumptions (H I ),(H2 ) and the
Holder continuity of the function f. We start with the following technical
lemma.
Lemma 6.1. Under the assumptions (HI ), ( H 2) there is a constant k ;;:: 0 such
that the family of operators {Ap(t) + kJ}tE[O. T] satisfies the conditions
(PI )-( P3) of Section 5.6.
PROOF. From the definition of the operators Ap(t) given above it follows
readily that for every real k the domain of D(Ap(t) + kI) = D(Ap(t» = D
is independent of t and therefore, for any choice of k ;;:: 0, the family
{Ap(t) + kJ}tE[O.T] satisfies the condition (PI)'
Since the constant e in the a-priori estimate stated in Theorem 3.1
(equation (3.3» depends only on Q, n, m, p and the ellipticity constant c, we
have
(6.6)
for every u E D. The a-priori estimate (6.6) implies, via the argument of S.
Agmon, that
( 6.7)
for u E D and A satisfying Re A ;;:: 0 and IAI ;;:: R for some constant R ;;:: 0;
Choosing k > R, (6.7) implies that
Ilull o. p ~ IA ~lkl"(AI + (Ap(t) + kl))ull o. p

M
~ IAI + 1 II(AI + Ap(t) + kl)ullo.p ( 6.8)
holds for u E D and A satisfying Re A ;;:: O. Using Lemma 3.1, as in the

proof of Theorem 3.5, it can be shown that for Re A ;;:: 0, 0 ~ t ~ T the
operator AI + (Ap(t) + kI) is surjective and hence (6.8) implies
M
IIR(A:Ap(t)+kl)ullo.p~ 1 + IAlllullo.p (6.9)
for u E L p (Q) and A satisfying Re A ~ O. Therefore, fixing a k > R, as we

will now do, implies that the family {Ap( t) + knE[O. T] satisfies (P2 )·
Finally, for u E LP(Q) and w = (Ap( T) + kI)-l u we have wED and
II(Ap(t) + k!)w - (Ap(s) + k!)wll o.p
= 111"1~2m (a,,(t, x) - a,,(s, X))Dawllo.p
~ ellt - siP L IID"'wll o.p ~ e 1t -

2 sIPll wI1 2m . p· (6.10)
lal,,;2m
From (6.7) and (6.9) it follows that
IIwIl2m.p::::;; C(IIAp(T)(Ap(T) + k1r1ullo,p + II(Ap(T) + kIrlullo,p)

::::;; C(I + kM + M)lIullo,p' (6.11)
Combining (6.10) and (6.11) yields
( 6.12)
for every u E LP(O) and the family {Ap(t) + kI}tE[o, T] satisfies also the
condition (P3) of Section 5.6. 0
From Lemma 6.1 and Theorem 5.7.1 we now deduce our main result.
Theorem 6.2. Let the family A(t, x, D), 0::::;; t::::;; T, satisfy the conditions
(HI) and (H2 ) and let f(t, x) E LP(O) for 0::::;; t::::;; T satisfy
IIP
(
~If(t, x) - f(s, x)I Pdx ) ::; Cit - slY (6.13)
for some constants C > 0 and 0::::;; y < 1. Then for every uo(x) E LP(g) the
evolution equation (6.1) possesses a unique generalized solution.
PROOF. We note first that if f satisfies (6.13) so does e-ktf for every real k.
From Lemma 6.1 it follows that there are values of k ~ 0 such that the
family {Ap(t) + kI}tE[o, T] satisfies the assumptions (P I )-(P3 ) of Section
5.6. We choose and fix such a k.
Given uo(x) E LP(O), it follows from Theorem 5.7.1 that the initial value
problem
v(O) = U o ( 6.14)
has a unique (classical) solution v. A simple computation shows that the

function u = ektv is a solution of the initial value problem
u(O) = U o (6.15)
and therefore (by definition) it is a generalized solution of the initial value

problem (6.1).
The uniqueness of this generalized solution follows from the uniqueness
of the solution v of (6.14) combined with the fact that u is a solution of
(6.15) if and only if v = e-ktu is a solution of (6.14). 0
Remark 6.3. It can be shown that if the boundary an

of n is smooth enough
and the coefficients aa(t, x) and 1(1, x) are smooth enough then the
generalized solution of (6.1) is a classical solution of this initial value
problem. For example, if all the data is Coo i.e., the boundary an is of class
Coo, the coefficients aa(t, x) and /(1, x) are in Coo([O, T] X 0) then the
generalized solution u is in Coo(]O, T] X 0).
CHAPTER 8
Applications to Partial Differential

Equations-N onlinear Equations
8.1. A Nonlinear Schrodinger Equation

In this section we consider a simple application of the results of Section 6.1
to the initial value problem for the following nonlinear Schrodinger equa-
tion in jR 2
1. au _ flu + klul 2 u = 0 in ]0, oo[ X jR2

{ I al (1.1 )
u(x,O)=uo(x) inIR2
where u is a complex valued function and k a real constant. The space in

which this problem will be considered is L2(jR 2). Defining the linear
operator Ao by D(Ao) = H2(jR2) and Aou = -i flu for u E D(A o )' the
initial value problem (1.1) can be rewritten as
{
~~ + Aou + F( u) = 0 for I > 0
( 1.2)
u(O) = Uo
where F(u) = iklul2u.

From Corollary 7.5.3 it follows that the operator - Ao is the infinitesimal
generator of a Co group of unitary operators S(/), - 00 < I < 00, on
L 2(jR 2). A simple application of the Fourier transform gives the following
explicit formula for S(t);
1.
(S(/)U)(X)=-4 {exp{iIX~YI2}u(Y)dY,
1ft! J~2 I
(1.3)
8 Applications to Partial Differential Equations-Nonlinear Equations 231
Moreover, we have
°
Lemma 1.1. Let Set), t ~ be the semigroup given by (1.3). If 2:$ P :$ 00
and l/q + l/p = 1 then set) can be extended in a unique way to a
bounded operator from U([R2) into LP([R2) and
( 1.4)
PROOF. Since Set) is a unitary operator on L2(1R2) we have IIS(t)ull o2 =
lIullo 2 for u E L2(1R2). On the other hand it is clear from (1.3) 'that
S(t)': LI(1R2) -> Loo(1R2) and that for t > 0, IIS(t)ull o. oo :$ (47Tt)-lllullo,I'
The Riesz convexity theorem implies in this situation that Set) can be
extended uniquely to an operator from U([R2) into LP([R2) and that (1.4)
holds. D
In order to prove the existence of a local solution of the initial value

problem (1.2) for every U o E H2(1R2) we will use Theorem 6.1.7 and the
remarks following it. To do so we note first that the graph norm of
the operator Ao in L2(1R2) i.e., the norm IIlulll = Ilullo,2 + IIAoullo,2 for
u E D( Ao) is equivalent to the norm II' 112 2 in H2(1R 2). Therefore
D(Ao) equipped with the graph norm is the spa~e H2(1R2). Next we prove
the needed properties of the nonlinear operator F.
Lemma 1.2. The nonlinear mapping F( u) = ik Iu 12 u maps H2(IR 2) into itself

and satisfies for u, v E H2(1R 2),
IIF(u)II2,2:$ CII uI16,oolluI1 2.2 (1.5)

IIF(u) - F(v)II2,2:$ C(lI ullt2 + Ilvll~.2)llu - vll2,2' (1.6)
PROOF. From Sobolev's theorem in IR 2 (see Theorem 7.4.7) it follows that
H2(IR 2) C L OO(IR 2) and that there is a constant C such that
( 1.7)
Denoting by D any first order differential operator we have for every
u E H2(1R2)
and therefore
IlluI2uIl2,2:$ C(lIuI16.oollull2,2 + lIullo,,,oI!ulli,4)' (1.8)
From Gagliardo-Nirenberg inequalities we have
(1.9)
and combining (1.8) and (1.9) we obtain (1.5).

The inequality (1.6) is proved similarly using Leibnitz's formula for the
derivatives of products and the estimates (1.7) and (1.9). D
Denoting D(Ao) equipped with the graph norm of Ao by Y it follows

from Lemma 1.2 that F: Y -+ Yand that it is locally Lipschitz continuous
in Y. Therefore the remark following Theorem 6.1.7 implies:
Lemma 1.3. For every U o E H2(JR 2) there exists a unique solution u of the
initial value problem (1.2) defined for t E [0, Tma,[ such thaI
u E C'([O, Tma,[: L2(JR 2») n C([O, Tmax[: H2(JR 2 »)
with the property that either Tmax = 00 or Tmax < 00 and lim,_T",Jlu(t)lln
= 00.
From Lemma 1.3 it follows that the initial value problem (1.2) has a
unique local solution. To prove that this local solution is a global solution it
suffices, by Lemma 1.3, to prove that for every T > 0 if u is a solution of
(1.2) on [0, T[ then II u( t) 112.2 .::; C( T) for 0 .::; I < T and some constant
C(T). That this is indeed so in our case, at least if k ~ 0, is proved next.
Lemma 1.4. Let U o E H2(JR 2) and let u be Ihe solution of Ihe initial value
problem (1.2) on [0, T[. If k ~ Olhen Ilu(t)II2.2 is bounded on [0. T[.
PROOF. We will first show that Ilu(t)II,.2 is bounded on [0, T[. To this end
we multiply the equation
I au
fat - flu + klul 2 u = 0
(
1.10
)
by u and integrate over JR 2. Then, taking the imaginary part of the result
gives d/dtllull~.2 = 0 and therefore
IIU(/)lIo.2=lIu oll o. 2 for O.::;t<T. (1.11)
Next we multiply (1.10) by aU/aI, integrate over JR 2 and consider the real
part of the result. This leads to
~ ~ 21 V' u ( t, x) 12 dx + ~ ~ 21 it ( t, x) 14 dx
= ~~21V'uo(x)12dx + ~~2IUo(x)14dX. (1.12)
Therefore, since k ~ 0, lIull1.2 is bounded on [0, T[.

To prove that lIu(t)1I 2. 2 is bounded on [0, T[ we note first that from
Sobolev's theorem it follows that H'(JR2) c LP(JR2) for p > 2 and that
Ilvll o. p '::; CII vll,.2 (1.13)
Therefore if u is the solution of (1.2) on [0, T[ it follows from the bounded-

ness of II u(t) 111.2 on [0, T[ and 0.13) that
Ilu(t)ll o.p '::; C for p > 2, 0.::; I < T. (1.14)

8 Applications to Partial Differential Equations- Nonlinear Equations 233
Since u is the solution of (1.2) it is also the solution of the integral equation
u(t) = S(t)u o - [S(t - s)F(u(s)) ds. (1.15)
o
Denoting by D any first order derivative we have
Du(t) = S(t)Duo - {S(t - s)DF{u(s)) ds. (1.16)
We fix now p> 2 and let q = p/(p - 1) and r = 4p/(p - 2). Then
denoting by C a generic constant and using Lemma 1.1, (1.16) and the
Holder inequality we find
IIDu(t) Ilo,p ~ I/S(t)Duo//o,p + C [(I

o
- s)'-2/ lll u(s) 12IDu(s) Illo,q ds
q
~ Cllu oll2,2 + C {(t - s)'-2/ ll u(s) 11~,rIlDu(s) 110,2 ds

q
~ Cllu oll2,2 + C [(I

o
- S)'-2/ q ds ~ CCt)
where in the last inequality we used the facts that r > 2 and therefore
Ilu(s)llo,r ~ C by (1.14) and that IIDu(s)1I0.2 ~ qu(s)111,2 ~ C. There-
fore, lIu(t)III.p ~ C and since by Sobolev's theorem W 1• P (1R 2 ) c LOO(1R2)
for p > 2, it follows that lIu(t)llo.oo ~ C for ~ t < T. °
Finally, since S(t) is an isometry on L2(IRZ) it follows from (1.15) and
(1.5) that
Ilu(t)lIz.2 ~ IIS(t)u o Il 2.2 + {IIS(t - s)F{u(s))1I2.2ds
~ Ilu oll 2.2 + C[lIu(s)1I6.oollu(s)1I2.2 ds

o
which by Gronwall's inequality implies the boundedness of Ilu(t)112.2 on
[0, T[ as desired. 0
Combining Lemma 1.3 with Lemma 1.4 yields our main result,
Theorem 1.5. Let U o E H2(1R 2). If k ;;:: °then the initial value problem
-:- -at - t::.u + klul u °
1 au 2
=
{ I (1.17)
u{o, x) = uo(x)
has a unique global solution u E C([O, 00[: H2(R2» n CI([O, oa[: L2(R2».
In conclusion we make a few comments. First we note that the local
We can actually obtain global existence also for k < °

solution of (1.17) exists, by Lemma 1.3, also without the restriction k ;::: 0.
provided that
Ikl lIu oll6.2 < 2 since this condition together with (1.12) and (1.9) imply
that II u( t) 111.2 is bounded on [0, T [ and as in the proof of Lemma 1.4 this
implies the boundedness of lIu(t)112.2'
234 Semigroups of Linear Operator,
Also. it is not difficult to show that the initial value problem (1.2) has
local solutions for the more general F( u) defined by F( u) = k Iu IP t u with
p 2: I. Moreover. it can be shown that for k > 0 the solutions of (.1.2) with
F(u) = klul P - IU are actually global solutions for every p 2: I.
8.2. A Nonlinear Heat Equation in IR I

Consider the following initial value problem
au=a 2u+ f (U) O<x<l,t>O

at ax 2
{ (2.1 )
u(t.O) =_u(t.I), u',(t,O) = u',(t, I) t 2: 0
u(o.x) - uo(x).
This problem represents the heat flow in a ring of length one with a
temperature dependent "source". In this section we will prove. under
suitable conditions, the existence of local and global classical solutions of
this initial value problem and study the asymptotic behavior of the global
solutions as t -> 00.
We start by introducing a convenient abstract frame. Let X = Cp([O. I])
be the space of all continuous real valued periodic functions having period I
with the supremum norm II u II = max o <;, <; II u( x) I. X consists therefore of
continuous functions on [0, I] satisfying u(O) = u( I). Let A be the linear
operator in X defined by D(A) = {u: u, u', u" E X} where u' and u" are
the first and second derivatives of u respectively and for u E D( A), Au = u".
Lemma 2.1. The operator A defined above is the infinitesimal generator uf a

compact analytic semigroup T( t), t 2: 0 on X.
PROOF. Since the domain of A contains all the trigonometrical polynomials
(with period I) it is dense in X by the Weierstrass approximation theorem.
Let g E X and ,\ = pe iii with p > 0 and - Tr /2 < it < Tr /2. Consider the
boundary value problem
,\2U - u" = g
{ (2.2)
u(O) = u(l), u'(O) = u'(I).
A direct computation shows that this problem has a solution u which is
given by,
u(x) = ,\
2,\ sinh 2"
.[f cosh ,\ (x - y - ~ ) g ( y ) dy + 1. J cosh ,\ ( x - y + ~ ) g ( y) dy ]
(2.3)
X Applications to Partial DifTerential Equations- Nonlinear Equations 235
and that this solution is unique. Denoting Re'A = Ii = P cos it > 0 and
using the elementary inequalities
ISinh ~ I ~ sinh ~. ICOSh 'A ( x - y ± ~ ) I ~ cosh Ii ( x - y + ~)

we find
Ilgll
< -"";':';;"-"--
Iu( x) I
- 21'Alsinh ~
. [ {'COSh Ii (x - y - ~ ) dy + ! cosh Ii (x - y + ~ ) dY ]
I
II gil (2.4)
Fixing any 'IT /4 < ito < 'IT /2 we find that

p(A)::J ~(ito) = {'A: larg'Al < 2ito}
and
From Theorem 2.5.2 it follows now that A is the infinitesimal generator of

an analytic semigroup T( t). t ~ O.
Since T( t) is analytic it is continuous in the uniform operator topology
for t > O. This is an immediate consequence of the inequality
C
IIT(t+h)- T(r)11 ~hIIAT(t)11 ~-h (2.5)
t
which holds for analytic semigroups for every t > 0 and h ~ O. Further-
more, since for 'A E ~(ito). R('A: A) maps X into D(A) such that bounded
sets in X are mapped into bounded sets in D(A) which have also a uniform
bound on their first derivative. it follows from the Arzela-Ascoli theorem
that R( 'A: A) is a compact operator. From Theorem 2.3.2 it follows now
that T( t), t ~ 0 is a compact semigroup and the proof is complete. 0
From Lemma 2.1 in conjunction with Theorems 6.2.1 and 6.2.2 we now
have:
Theorem 2.2. For every continuous real valued function f and every U o EX
there exists a to > 0 such that the initial value problem (2.1) has a mild
solution u(t, x) on· [0, t o[ and either to = 00 or if to < 00 then
limsup t _ to Ilu{t, x)11 = 00.
If we assume further that f is Holder continuous then the mild solution
given by Theorem 2.2 is a classical solution. In this case we have:
Theorem 2.3. Iff is a Holder continuous real valued function then for every
u()(x) E X there is a to> 0 such that the initial value problem (2.1) has a
classical solution u(t, x) on [0, tor and either to = 00 or if to < 00 then

limsuPt---+to11u(t, x)1I = 00.
PROOF. From Theorem 2.2 it follows that the initial value problem (2.1)
has a mild solution u which by definition is continuous on [0, t o[ X [0, 1].
Therefore t ~ f(u(t; x» is continuous in X and by Theorem 4.3.1 u(t, x)
is HOlder continuous.. Since by our assumption, f is HOlder continuous it
follows that t ~ f(u(t, x» is Holder continuous on [0, tor. But then Corol-
lary 4.3.3 implies that u is a classical solution of the initial value problem
and the proof is complete. 0
We tum now to the study of global solutions of the initial value problem
(2.1) and start by noting that the conditions of Theorem 2.3 do not imply
the existence of a global solution of (2.1). Indeed, choosing for example
f(s) = S2 and uo(x) == 1 it is easy to see that the unique solution of (2.1) in
this case is u(t, x) = (I - t)-I which blows up as t ~ l.
Lemma 2.4. Let f be continuous and let u be a bounded mild solution of (2.1)
on [0, oa[ then the set {u(t, x): t ~ O} is precompact in X.
PROOF. Let Ilu(t)ll::s; K for t ~ O. The continuity of f implies that
Ilf(u(t»11 ::s; N for some constant N. Let T(t), t ~ 0 be the semigroup
generated by A and recall that, by Lemma 2.1, T(t) is compact for t > o.
Let 0 < 6 < 1, t ~ 1 and set
u(t} = T(E}U(t - E} + [u(t} - T(E}U(t - E}] = uE(t} + vE(t}.
The set {uE(t): t ~ I} is precompact in X since {u(t - E): t ~ I} is bounded

and T( E) is compact. Also,
IIvE(t}1I =llf~ET(t - s}f(u(s)) dsll

::s;
t-E
r
IIT(t - s}lIllf(u(s)}1I ds::s; EMN
where M = suP{1I T(t)1I : 0 ::s; t ::s; I}. Therefore {u(t): t ~ I} is totally

bounded i.e. precompact. Since {u(t): 0 ::s; t::s; I} is compact as the continu-
ous image of the interval [0, I] the result follows. 0
Lemma 2.5. Let f be Holder continuous. If for some Uo E X the initial value
problem (2.1) has a bounded global solution u( t, x) then there is a sequence
tk ~ 00 such that
lim U(tk' x} = cp(x} (2.6)
tk~oo
where cp ( x) is a solution of the boundary value problem
{
cp" + f(cp} = 0 (2.7)
cp(O} = cp(I}, cp'(O} = cp'(I}.
PROOF. Multiplying the equation

au-_-a-2 U
- + f()
u (2.8)
at ax 2
by au/at and integrating over x and t yields
j Tjll au 12 dxdt + ~jll au (T, x) 12 dx _jlP(u(T, x)) dx

o 0 at 2 0 ax 0
~ ~jllau(0'X)12
2 0 ax
dx-jl p (u(O,x))dx
0
(2.9)
where pes) = 1;f(r) dr. Since lu(t, x)1 ~ K for some constant K, we
deduce from (2.9) that
fO foIl ~~ 12 dx dt < 00.
Therefore, there exists a sequence t, ~ 00 for which limt,_oo(au(t" x)/at)

= 0 a.e. on [0, I], or (au(t" x)/at) ~ 0 in U(O, I). From Lemma 2.4
have lim t . _ oo U(tk' x) = cp(x) uniformly for °

it follows that for a subsequence of t, which we denote by t k , we
~ x ~ 1. Therefore,
limt._oof(u(t k , x» = f(cp(x» uniformly in x for x E [0, I]. Passing to the
limit as t ~ 00 through the sequence t k • in equation (2.8) in the sense of
°
L2(0, I) and using the closedness of the operator Au = u" as an operator in
L2(0, I) we find cp"(x) + f(cp(x» = in L2(0, I). Sincef(cp(x» is continu-
ous, this equation holds in a classical sense. Furthermore, the periodicity
conditions are satisfied by cp(x) since they are satisfied by u(t, x). 0
Corollary 2.6. Iff is Holder continuous and f(s) "* for all s
initial value problem (2.1) has no bounded global solutions.
° E IR, then the
PROOF. If f(s) "* 0 the boundary value problem (2.7) has no solution.
°
Indeed, integrating the equation cp" + f( cp) = over [0, 1] yields
cp'(I) - cp'(O) = ff(cp(s)) ds"* 0
and therefore the boundary conditions cannot be fulfilled. Thus by Lemma

2.5, no bounded solution of (2.1) can exist. 0
We conclude our discussion with the following result:
Theorem 2.7. Iff is Holder continuous and sf(s) < 0 for all s "* then all
solutions of the initial value problem (2.1) are bounded and moreover, all
°
solutions of (2.1) tend to zero as t ~ 00.
PROOF. The boundedness of the solution and, even more, the estimate:
max lu(t, x)1 ~ max lu(s, x)1 for t ~ s (2.10)
O~x~1 O~x~1
are immediate consequences of the maximum principle. Therefore all solu-

tions of the initial value problem (2.1) are bounded. Moreover from Lemma
2.5 we know that for some sequence 14 --. 00, U(/ 4 , x) --. cp(x) where cp(x)
is a solution of the boundary value problem (2.7). But the only solution of
this boundary value problem is cp == O. This can be seen by multiplying
cp" + J(cp) = 0 by cp, integrating over [0, I] and obtaining
to Icp'12 dx :5; 0
which implies cp' == 0 and cp = const. However the only solution of J( s) = 0

is s = 0 and therefore cp == O. Thus we have
lim U(/ 4 .X)=0. (2.11 )
'I.- x
Combining (2.10) and (2.11) yields u(t. x) -+ 0 as t --. 00. o
8.3. A Semilinear Evolution Equation in IR 3

Let {l be a bounded domain with smooth boundary a{l in R 3 and consider
the following nonlinear initial value problem
(
au
al
= flu + t u~
;=1 ax;
in ]0. T] X {l
(3.1 )
U(/. x) = 0 on [0. T] X a{l
u(O. x) = uo(x) in {l.
We will use the results of Section 6.3 to obtain a strong solution of the
initial value problem (3.1) in L 2 ({l).
In this section we will denote by (.) and II . II the scalar product and
norm in L 2 ({l). As in Section 7.2 we define an operator A by
D(A) = H2({l) () HJ({l), Au = -flu for u E D(A). (3.2)
The operator A is clearly symmetric and since - A is an infinitesimal
generator of a Co semi group on L 2 ({l) (e.g. by Theorem 7.2.5) it follows that
A is self adjoint. Moreover, from Theorem 7.2.7 it follows that -A is the
infinitesimal generator of an analytic semigroup on L 2 ({l). Therefore we can
use the results of Section 2.6 to define fractional powers of A. In particular
we have for some 8 > 0,
(Au, u) = (A 1/ 2U, AI/2u) = IIA'/2ull 2 = II Vull 2 ~ 811ull 2 (3.3)
where VU is the gradient of u and the inequality is a consequence of
Poincare's inequality. The domain of A consists of Holder continuous
functions. This follows from a version of Sobolev's imbedding theorem or
can be shown directly as follows:
H Applications to Partial Differential Equations-Nonlinear Equations 239
Lemma 3.1. D(A) consists of HOlder continuous functions with exponents ~

and there is a constant C such that
for u E D(A) (3.4)
where Xi E n, i = 1,2 and IXI - x 2 1 denotes the Euclidean distance be-
tween Xl and x 2 •
PROOF. For cp E Co(n) we have the classical identity
() 1 ~cp(y)
cp x = C n Ix - y I dy. (3.5)
r
From (3.5) and the Cauchy-Schwartz inequality we deduce
Icp(x J ) - cp(x 2 )1 2 ~ C2(~ ~cp(y)( IXI ~ yl - IX2 ~ yl ) dy
But,
~ C2~I~CPI2 dy . ~( IXI ~ yl - IX2 ~ yl r dy.
1 1)
~ ( IXI
2
- yl - IX2 - yl dy ~ ClXl - x 21
where C is a constant depending only on n. Therefore

Icp(x l ) - cp(x 2)1 ~ qAcplllx l - x211/2. (3.6)
Approximating u E D(A) in H2(n) n Hcl(n) by a sequence cp" E Ctf(n)
and passing to the limit yields (3.4) since H2(n) c C(Sl) by Theorem 7.1.2.
o
For functions u in the domain of A we will need the following estimate.
Lemma 3.2. There is a constant C such that

for u E D(A). (3.7)
PROOF. First we note that by Theorem 7.1.2 u E D(A) is in C(Sl) and since
an is assumed to be smooth it also follows that u vanishes on an. For u =0
(3.7) is trivial. Let Ilullo.co = L > O. From Lemma 3.1 we have
lu(x l ) - u(x 2 )1 ~ Klxl - x 211/ 2

where K = qAuli. Without loss of generality we assume that lu(O)1 = L
and let B R be an open ball of radius R = (L/ K)2 around O. In this ball we
have
L
lu(x)1 > lu(O)1 - lu(x) - u(O)1 ~ L - Klxll/2 > L - K K = O.
(3.8)
240 Semigroups oC Linear Operators
Since u vanishes on an we deduce from (3.8) that B R C n and for x E BR
lu(x)1 ~L-Klxll/2. (3.9)

Now,
= 47TL2R3 t(1-
o
TJ1!z('1zdTJ
= CL zR 3 = CL8K- 6
and (3.7) follows readily. o
Lemma 3.3. For y > 3/4 there is a constant C depending only on y and n
such that
for u E D(A). (3.10)
PROOF. Let 3/4 < y < 1. If w = AYu then (3.10) is equivalent to
IIA -Ywll o. 00 :s; Cllwll· In order to estimate IIA -Ywllo. 00 we use the definition
of A -Y given by formula (6.4) of Section 2.6. So,
A-Y w = sin'1TY1°Ot-Y(tI+A)-lwdt. (3.11)

'1T 0
From (3.3) it follows that IIA - III :s; 1)- I and that for every t ~ 0
lI(tI + A)-I WIl :s; (t + 1»-llIwll. (3.12)
Also since -A is dissipative in L 2(n) we have
(3.13)
and since (tl + A) - IW E D( A), Lemma 3.2 yields
lI(tl + A)-IWIl~.oo:S; CIIA(tI + A)-I WIl 3 11(tI + A)-I WII . (3.14)
Combining (3.11), (3.l2), (3.13) and (3.14) yields
IIA-Ywllo.oo:S; ClfoOOt-Y(1) + t)1/4I1wlldt. (3.l5)
For 3/4 < y < 1, the integral in (3.15) converges and we have IIA -Ywll o. 00
:s; q w II. For y ~ 1 the result follows from the result for 3/4 < y < 1 via
the estimate IIA -III :s; I)-I. 0
We tum now to the nonlinear term of (3.1) and start with the following
lemma.
Lemma 3.4. Let

3 au
f(u) = L u-. (3.16)
i-I aX i
Ify> 3/4 and u E D(A) then f(u) is well defined and

IIf(u)1I ~ CilAYuIlIlA 1/ 2ull. (3.17)
If u, V E D(A) then
IIf(u) - f(v)1I ~ C(IIAYuIiIiAI/2U - AI/2VIl + IIA 1/ 2vIlIIAYu - AYvll).
(3.18)
PROOF. Since D(A) C H2(n) it follows from Sobolev's theorem (Theorem
7.1.2) that u E LOO(n) and thereforef(u) E L 2(n) and is thus well-defined.
Moreover, from Lemma 3.3 we have
Ilf(u)11 ~ II ull o. II '\7ull ~ CilAYuli1i '\7ull = CIiA YuIiIIAI/2 U II·
00
Also,
Ilf(u)-f(v}ll ~ lI ull o.oo ll'\7(u-v)11 + lIu- vll o.oo lI'\7ulI
~ C(IIAYuIiIiAI/2U - AI/2VII + IIA 1/ 2vIlIlAYu - AYvll).
o
From (3.17) it follows that the mapping f can be extended by continuity
to D(AY) and that (3.17) and (3.18) hold for every u, v E D(AY). Therefore
the conditions of Theorem 6.3.1 are satisfied and we have:
Theorem 3.5. The initial value problem (3.1) has a unique local strong solution
for every Uo E D(AY) with y > 3/4.
We note that from the results of Section 6.3 it follows in the same way as
above that if
O~P<1
i
then the initial value problem
aau = lJ.u +
t
t
i-I
u aau + f( t, x)
X,
in ]0, T] X n
(3.19)
u(t,x)=O in[O,T]xan
u(O,x)=uo(x) inn
has a unique local strong solution for every initial value uo(x) E D(AY)
with y > 3/4.
8.4. A General Class of Semilinear Initial Value

Problems
The present section is devoted to a general class of semilinear initial value
problems which extends considerably the examples given in the previous
two sections. The main tool that will be used is Theorem 6.3.1. In order to
apply it we will have to use fractional powers of unbounded linear opera-

tors. We therefore start with some results concerning such fractional powers.
Recall that if - A is the infinitesimal generator of an analytic semigroup
in a Banach space X and 0 E peA) we can define fractional powers of A as
we have done in Section 2.6. For 0 < a ::s; 1, A a is a closed linear operator
whose domain D(Aa) ::::> D(A) is dense in X. We denote by Xa the Banach
space obtained by endowing D(Aa) with the graph norm of Aa. Since
o E peA), A a is invertible and the norm II Iia of Xa is equivalent to IIAaul1
for u E D(Aa). Also, for 0 < a < {3 ::s; 1, Xa::::> Xf3 and the imbedding is
continuous.
Let n c IR n be a bounded domain with smooth boundary and let an
A(x, D} = L aa(x}Da (4.1 )
lal,;2m
be a strongly elliptic differential operator in O. For the notations and

pertinent definitions see Sections 7.1 and 7.2. For 1 < P < 00 we associate
with A(x, D) and operator Ap in LP(O) by
D{Ap) = W 2m ,P(0} n Wom,P(O} (4.2)
and
(4.3)
We have seen in Section 7.3 (Theorem 7.3.5) that - Ap is the infinitesimal
generator of an analytic semigroup on LP(O). By adding to A(x, D), and
hence to A p ' a positive multiple of the identity we obtain an infinitesimal
generator - (A p + kI) of an analytic semigroup, which is invertible. In the
sequel we will tacitly assume that this has been done and thus assume
directly that Ap itself is invertible. From Theorem 7.3.1 we know the
following a-priori estimate
Since we assume now that Ap is invertible in LP(O) it follows readily that

Cliullo,p::S; IIApullo,p for some constant C > 0 and therefore we have
lIull2m,p::S; qApullo,p for u E D{Ap). (4.4)
Before we start describing some properties of the fractional powers of the
operator Ap we recall the well known Gagliardo-Nirenberg inequality.
Lemma 4.1. Let 0 be a bounded domain in IRn with boundary ao of class cm

and let u E w m, r(O) n Lq(O) where 1 ::s; r, q::s; 00. For any integer j,
o ::s; j < m and any j / m ::s; it ::s; 1 we have
( 4.5)
provided that
-1 = -j + it (1- - -m) + (1 - it)-1 (4.6)
P n r n q
X Application, to Partial Diflcrential Equations- Nonlinear Equations 243
and m - j - nlr is not a nonnegative integer. If m - j - nlr is a nonnega-

tive integer (4.5) holds with {j = jim.
The next lemma is our main working tool.
Lemma 4.2. Let I < p < 00 and let AI' be the operator defined above. For any
multi-index /3. 1/31 = j < 2m and any jl2m < a s; I we have
IIDflA;"ull o. p s; qUllo.p for u E D{A p }. (4.7)
PROOF. Set B = Dfl. Since 1,81 < 2m it is clear that D(B):::) D(Ap). From
the previous lemma we have
IIDfiulio.p s; quW(n~~~llull(\.f2m. (4.8)

Polarization of (4.8) together with the estimate (4.4) yield
IIDflullo.p s; C(p-l+I/2ntIIApullo.1' + pi/2mllullo.p) (4.9)
for p > 0 and u E D(Ap). From Theorem 2.6.12 it follows now that
D( B) :::) D( A~) for jl2m < a S; I i.e .. BA; " is bounded for these values of
a and the proof is complete. 0
Theorem 4.3. Let Q c Iflll be a bounded domain .....ith smooth boundarv aQ

and let AI' be as above. If 0 S; a S; I then
n n
for k - - < 2ma - - . ( 4.10)
q p
x" c C(Q) for 0< v < 2ma

- -!!.-
p (4.11 )
and the imbeddings are continuous.
PROOF. From Lemma 4.2 it follows readily that X" c WI.p(Q) provided
that j < 2ma and the imbedding is continuous. From Theorem 7.1.1 it
follows that WI.P(Q) is continuously imbedded in Wk.q(Q) provided that
k - nlq < j - nip and (4.10) follows. From Sobolev's theorem (Theorem
7.1.2) it follows that WI.p(Q) is continuously imbedded in C(Q) for
o S; v < j - nip and (4.11) follows. 0
We note in passing that Lemma 3.3 of the previous section is a special

case of Theorem 4.3 since it is a consequence of (4.10) taking k = O. q = 00.
n = 3. p = 2 and m = I.
We turn now to the applications of Theorem 6.3.1. But rather than stating
and proving a very general result, we prefer to restrict ourselves to a simple
example in Ifl J with p = 2 and a second order operator. which contains
already most of the ingredients of the general case and then comment
(without proof) on more general results at the end of the section.
Theorem 4.4. Let D be a bounded domain in R 3 with smooth boundary aD and

let A(x, D) be a strongly elliptic operator given by
a a
A(x, D) = - L
3
a-a k I(X)-a-
k,l=1 Xk ' XI
where ak,/(x) = al,k(x) are real valued and continuously differentiable in 0,
Let f( t, X, u, p), pER 3, be a locally Lipschitz continuous function of all its
arguments and assume further that there is a continuous function p( t, r) : R X
R --+ R + and a real constant I ~ Y < 3 such that
If(t, X, u, p)1 ~ pet, lul)(I + Ipj1') (4.12)
If(t,x,u,p)-f(t,x,u,q)1 ~p(t,lul)(1 + Iply-I + Iqly-I)lp-ql
(4.13)
If(t,x,u,p)-f(t,x,v,p)1 ~p(t,lul + Ivl)(1 + IpIY)lu-vl. (4.14)
Then for every U o E H2(D) n HJ(D) the initial value problem
{ :~
u( t , x)
+A(x,D)u=f(t,x,u,gradu)
= 0
inn
on an
( 4.15)
u(O, x) = uo(x) in n
has a unique local strong solution in L 2(D).
PROOF. We recall that with the strongly elliptic operator A(x, D) we

associate an operator A in L2(D) by D(A) = H2(D) n HJ(D) and Au =
A(x, D)u for u E D(A). From Theorem 7.3.6 it follows that -A is the
infinitesimal generator of an analytic sernigroup on L 2(D) and from the
strong ellipticity together with Poincare's inequality it follows readily that A
is also invertible. From Theorem 4.3 it follows that if a> 3/4 then
Xa c L OO(D) and if also 1/q > (5 - 4a)/6 then Xa C WI, q(D). Thus for
max(3/4,(5y - 3)/4y) < a < I we have
(4.16)
In order to apply Theorem 6.3.1 we have to show that the mapping
F(t, u)(x) = f(t, x, u(x), vu(x», xED (4.17)
is well defined on R + X Xa and satisfies a local Holder condition there.
From (4.12) and (4.16) we have for every u E Xa
liF(t, U)IIO,2 ~ 2p(t, lIullo.oo)(MI/2 + Ilulll.2y) (4.18)

where M is the measure of D. Therefore F is well defined on R + X Xa' To
show that F satisfies a local Holder condition we note that
II F( I, u) - F( I, v) 116. 2 ~ 211/(
0
I, x, u, V u) - /( I, x, u, vv) 11 dx
In
+ 2 1/( I, x, u, Vv) - /( I, x, v, vv) 11 dx
( 4.19)
and estimate each of the two terms on the right of (4.19) separately. From
(4.13) and (4.15) we have
11/(/, x, u, vu) - /(/, x, u, vv )1 1 dx

Q
~ C· p(t, Ilullo.oo)11(I + Ivul 2y - 1 + IvvI1y-2)IV(u - v)11dx

Q
~ C· p(/, Ilullo.oo)2(M, + IIvu16S;1 + IIvvI16:1-;1)IIV(u - v)116.1y

~ L(llull a, IIvllJllu - vllT.1 y ~ L(lIull a , IlvllJllu - vll~
where II Iia denotes the norm in Xa and L is a constant depending on lIuli a
and IIvlla' To obtain the second inequality we used Holder's inequality. The
last inequality is a consequence of the continuous imbedding of Xa in
WI.2 Y(O). Similarly for the second term we have by (4.14) and (4.16)
fl/(/,x,u,Vv) -/(t,x,v,Vv)1 2 dx
n
fn
~ Cp( t, Ilullo,oo + Ilvllo,a<i (1 + IVvI 2Y )lu - vl 2 dx
~ Cp( t, Ilullo. oo + Ilvllo. oo)2 1Iu - vI16,oo(1 + Ilvlli;2Y)

~ L(llulla, Ilvlla)llu - vii;
and therefore
IIF(I, u) - F(t, v)llo.2 ~ L(llull a, Ilvlla)llu - vila (4.20)
and the existence of the strong local solution of (4.15) is a direct conse-
quence of Theorem 6.3.1. 0
Before continuing we note that Theorem 3.5 is a special case of Theorem

4.4 since - ~ is obviously strongly elliptic and /( u, V u) = u . V u certainly
satisfies the conditions of Theorem 4.4. Furthermore, from Theorem 4.4 we
also obtain an extension of the existence results of Section 8.2 for 0 c ~ 3
assuming however that / is bounded and locally Lipschitz continuous in O.
From Theorem 4.4 we obtain a local strong solution in the sense of L2(0)
of the initial value problem (4.15). This solution satisfies, by Theorem 4.4
u E C([O, To[ : L2(0)) n C(]O, To[: H2(0) n HJ(O))

ncl(]o, To[: L2(0))
for some To > O. But in fact it is a classical solution of this initial value
problem for t > O. Indeed, since for 0 < t < To, u E D(A) c C(n) and, by
Corollary 6.3.2, t ~ du/dt E X" is locally Holder continuous for 0 < t <
To it follows that (t, x) ~ u(t, x) and (t, x) ~ (a/at)u(t, x) are continu-
ous on 0 < t < To, x E n.
To show that u is a classical solution of the
equation it remains to show that u(t, . ) E e 2 (O). From the fact that for
o < t < To, u(t, . ) E D(A) we have Vu E W I,Ql(O) c LPl(O) where ql =
2, PI = 6/(3 - 2) = 6. So, Au = F(t, u) - du/dt E L PliY(O) by (4.12)
whence by Theorem 7.3.1 u E W 2 ,6iY(O) and therefore Vu E W I,Q2(O)
with q2 = 6/y > 2. Repeating this process we find that Vu E WI/Qn(O)
where l/qn = y(1/qn-l - 1/3). It is easy to check that after a finite
number of steps (one step if 1 ~ Y < 2) qn > 3 and then Vu(t, . ) is
HOlder continuous in 0 and it follows that F(t, u) is Holder continuous in
O. Since a > 3/4 and (a/at)u(t, . ) E X" it follows that (a/at)u(t, . ) is
Holder continuous in O. But then Au = F(t, u) - du/dt is Holder con-
tinuous in 0 and by a classical regularity theorem for elliptic equations it
follows that u(t, . ) E e 2 +o(O) for some 0 > 0 that is, u has second order
Holder continuous derivatives in x and is thus the classical solution of
(4.15).
We conclude this section with some comments on more general existence
results. We assume that A(x, D) is a strongly elliptic differential operator
given by (4.1). We define an operator Ap in LP(n) by D(Ap) = w 2m ·p(n)
n Worn.P(n) and Apu = A(x, D)u for u E D(Ap). By adding a positive
multiple of the identity to Ap we can assume as we will tacitly do that Ap is
invertible. From Theorem 7.3.5 it follows that - Ap is the infinitesimal
generator of an analytic semigroup on LP(n). Let
F{t, u){x) = f(t, x, u, Du, D 2 u, ... , D2m- l u) (4.21 )
where Di stands for any j-th order derivative. Assume that f is a continu-
ously differentiable function of all its variables and consider the initial value
problem
{
~~ + Apu = F{t, u) (4.22)
u{O) = U o
in LP(n). From Theorem 4.3 it follows that if 1 - 1/2m < a < I and pis
sufficiently large, then X" is continuously imbedded in e 2m - I (n). This
implies that
IIF(t, A;"u) - F(s, A;"v)llo. p ~ e(lt - sl + Ilu - vll o. p ) (4.23)
where e is a constant which depends on IIDiA-"ullo.OO' IIDiA-ltvllo.oo for
o~ j < 2m - 1. Therefore if p is large enough the conditions of Theorem
6.3.1 are satisfied and we have
Theorem 4.5. Let n be a bounded domain in IR n with smooth boundary an and

let Ap be the operator defined above. Let F(t, u) be defined by (4.21) where f is
8 Applications to Partial Differential Equations- Nonlinear Equations 247
a continuously differentiable function of all its variables with the possible

exception of the x variables. If p > n then for every Uo E W 2m . P(~) n
w;r" P(~) the initial value problem (4.22) has a unique local strong solution.
If p < n (as is the case in Theorem 4.4) the argument leading to Theorem
4.5 fails since for no 0:;; a < I D 2m-'(A-ou) E LOO(~). In this case, in
order to obtain an existence result one has to assume that the function f
satisfies some further conditions similar in nature to the estimates
(4.12)-(4.14).
8.5. The Korteweg-de Vries Equation

In the present section we will use the results of Section 6.4 to obtain an
existence theorem of a local solution of the Cauchy problem for the
Korteweg-de Vries equation:
u, + u xn + uu< = 0 t ~ 0 -oo<x<oo
{ (5.1 )
u(O, x) = uo(x).
Throughout this section we will assume that all functions are real valued,
f J
denote by the integral over all of IR and denote by the Fourier transform
of f.
For every real s we introduce a Hilbert space HS(IR) as follows; Let
u E L2(1R) and set
(5.2)
The linear space of functions u E L 2(1R) for which II ull., is finite is a

pre-Hilbert space with the scalar product
(5.3)
The completion of this space with respect to the norm II II.. is a Hilbert
space which we denote by HS(IR).
It is clear that HO(IR) = L2(1R). The scalar product and norm in L2(1R)
will be denoted by ( , ) and II 110' Furthermore, it is easy to check that the
spaces H'(IR) with s = n coincide with the spaces Hn(IR), n ~ I, as defined
in Section 7.1 and the norms in the two different definitions are equivalent.
In the following lemma we collect some useful properties of the spaces
W(IR).
Lemma 5.1. (i) For t ~ s, W(IR)::> H'(IR) and lIull, ~ lIuliJor u E W(IR).
(ii) For s > !, W(IR) c C(IR) and for u E W(IR)
lIuli oo :;; qull.. (5.4)
where lIull oo = sup{lu(x)l: x E IR}.
PROOF. Part (i) is obvious from the definitions and the elementary inequal-
ity (l + e)1 ~ (1 + er
for t ~ s and ~ E ~.
From the Cauchy-Schwartz inequality we have,
lu(x)1 =I~ fejX~u(Od~1

r
~ ~ (f (1 :~2r /2 (f{1 + enu(~),2d~r/2 = qulls
so the integral defining u in terms of u converges uniformly and u is
continuous. Moreover Ilulloo ~ Cllulls. 0
Let X = L2(~) = HO(~) and Y = HS(~) with s ~ 3. We define an opera-
tor Ao by D(Ao) = H\~) and Aou = D3 U for u E D(Ao) where D = d/dx.
Lemma 5.2. Ao is the infinitesimal generator of a Co group of isometries on X.

PROOF. Ao is skew-adjoint i.e. iAo is self-adjoint or equivalently (Aou, u) = 0
for all u E D(Ao). This follows readily from
(Aou,u)= fD3U.udx= - fu.D3Udx= - (Aou,u)
where the second equality is achieved by integration by parts three times.

From Stone's theorem (Theorem 1.10.8) it follows that Ao is the infinitesi-
mal generator of a group of isometries on X = L2(~). 0
Next we define for every v E Y = HS(~), s ~ 3, an operator A)(v) by:

D(A)(v» = H)(~) and for u E D(A)(v», A)(v)u = vDu. We then have:
Lemma 5.3. For every v E Y the operator -A(v) = -(Ao + A 1(v» is the
infinitesimal generator of a Co semigroup T/t) on X satisfying
(5.5)
for every f3 ~ f3o(v) = collvlls where Co is a constant independent of v E Y.
PROOF. We note first that since v E HS(~), Dv E HS-)(~) and since s ~ 3
it follows from Lemma 5.1 that Dv E Loo(~) and that IIDvll 00 ~ qDvlls_)
~ qvlI s •
Now, for every u E H)(~) we have
(A) ( v ) u, u) = f v Du . u dx = ! f V Du 2 dx = -! f Dv u 2 dx
~ -! IIDvil00llul12 ~ -collvllslluI1 2.
Therefore -(A 1(v) + f3I) is dissipative for all f3 ~ f3iv) = collvll s. Since
Ao is skew-adjoint, Ao + A 1(v) + f31 is also dissipative for f3 ~ f3o(v).
Moreover,
R Applications to Partial Differential Equations- Nonlinear Equations 249
Using integration by parts it is not difficult to show that for every u E H\lRj
we have IIDul1 :::; lIuI1 2 / 31ID3 ull'/3 and by polarization we obtain for every
E> 0,
IIDul1 :::; EIID3 U II + C(E)lIull· (S.7)
Choosing E= 1/211vlloo and substituting (S.7) into (S.6) yields
II(A,(v) + /3I)ull :::; illAoul1 + Cliull for u E D(Ao) (S.8)
whence, by Corollary 3.3.3, Ao + A ,( v) + /31 = A( v) + /31 is the infinitesi-
mal generator of a Co semigroup of contractions of X for every /3 ;::: /30 ( v).
Therefore, A( v) is the infinitesimal generator of a Co semigroup T. (t)
satisfying (S.S). 0
We let now Br be the ball of radius r > 0 in Y centered at the origin and
consider the family of operators A( v), to E B r • We want to show that this
family satisfies the conditions of Theorem 6.4.6. Because of the special form
of the family A( v), v E Br , it follows that it suffices to prove the following
three conditions:
(A,) The family A( v), v E B r , is a stable family in X.

(A 2 ) There is an isomorphism of Y onto X such that for every v E Br
SA(v)S-' - A(v) is a bounded operator in X and
IISA{v)S-' - A{v)11 :::; C, forall vEBr • (S.9)

(A3) For each v E Br , D(A(v»:J y, A(v) is a bounded linear operator
from Y into X and
(S.IO)
We note that (A,) is the same as the condi tion (H, ) of Section 6.4 and
(A 2) implies both (H 2 ) and (Hs) as can be easily seen from Lemma S.4.4
and Theorem S.4.6. The condition (A 3 ) implies (H) and (H 4 ) while (HI» is
satisfied since both X and Y are reflexive. Finally, if Ilunll, < r and v E Br
then
IIA(v)uoll :::; IID3 uo II + IlvDuol1
:::; IID3 uo II + II vll oo llDu ol1
:::; Iluoll)(i + r):s; r{1 + r) = k (S. II)
and condition (4.21) of Theorem 6.4.6 is also satisfied.
In order to prove that the family A(v), v E Br , satisfies the conditions
(A, )-( A 3 ) we need one more preliminary result. Let
Nf = _1_
{2;
feiX~( 1 + e )'/2 j( 0 d~. (S .12)
It is not difficult to check that N is an isomorphism of Y = H'(IR) onto

X = e(IR). For a given function f E L2(1R) let Mr be the operator of
multiplication by the function f, i.e. Mfu = fu. We then have:
Lemma 5.4. Let f E HS(~), s > 3/2 and let T = (NMf - MfN)N -so Then
T is a bounded operator on X = L2(~) and
IITII :::; Cilgradflls-I' (5.13)

PROOF. The Fourier transform of T is the integral operator with kernel
k(~, 1/) given by
ilik(g,7]) = ((1 + e)'/2 - (1 + 7]2)'/2)/(g - 7])(1 + 7]2)(S-0/2.

Since
1(1 + e)'/2 - (1 + 1/ 2)'/21:::; sl~ - 1)1(1 + et- )/2 + (1 + 1/2t- )/2)
I I
we have
ililk(g,7])1 :::; s(l + et-0/ 2Ig - 7]1/U - 7])(1 + 7] 2 t- S
)/2
+ slg - 7]1/(g - 7]) = k U,7]) + k 2(g,7])·

j
To show that T is bounded it suffices to show that the operators TI and T2

wi th kernels k I(t 1/) and k 2 (t 1/) are bounded. Using the inverse Fourier
transform we find that
TI = sN-1MgN -s, T2 = sMg (5.14)
where Mg , is the multiplication operator by the function g for which
g(O = 1~lf(o. From Lemma 5.1 (ii) it follows that
Ilgllx :::; Cilgll,-1 :::; Cilgradflls_I' (5.15)

Now,
and
(5.17)
Therefore both TI and T2 are bounded operators in X. Combining (5.15)
with (5.14) and (5.17) yields the desired estimate (5.13). D
We now have:
Lemma 5.5. For every r > 0, the family of operators A( v), v E Br , satisfies
the conditions (AI)-(A3)'
PROOF. Let r > °
be fixed. From Lemma 5.3 it follows that if f3 2 cor. A(v)
is the infinitesimal generator of a Co semigroup Te(t) satisfying II T.(t)11 :::;
ef3 1 and therefore A( v), v E Br is a stable family in X (see Definition 6.4.1).
As we have mentioned above S = N is an isomorphism of Y = HS(~)
onto X = L2(~). A simple computation shows that for u, v E Y we have
(SA(V)S-I - A(v))u = (S(vD)S-1 - vD)u = (Sv - VS)S-I Du
and therefore by Lemma 5.4

II(SA{v)S-1 - A{v))ull = II {NMv - MvN)N-sA -I Dull
~ II {NMv - MvN)N-sIlIIA -I Dull
~ CIIgradvlls_dlull ~ CIIvllyllull·
Since Y is dense in X it follows that II SA( V)S-I - A( v )11 ~ CIIvll y ~ Cr
and (A 2 ) is satisfied.
Finally, since s ~ 3 D( A( v))::J Y for every v E Yand for v E Br
IIA{v)ulI ~ IID3 u II + IlvDul1 ~ IID3 u II + IlvlloollDul1
~ (1 + CIIvllJllull s ~ (1 + Cr)llull y
and therefore A( v) is a bounded operator from Y into X. Moreover if
v I' v 2 E Br , u E Y then
II(A{v l ) -A{v 2 ))ull = II{v l - v2 ) Dull
~ Ilvl - v211 IIDul1 00 ~ ClIVI - v211 Ilull y
From Lemma 5.5 it follows that the family A( v), v E Br satisfies the
conditions (AI)-(A3) stated above and therefore by the remarks following
these conditions all the assumptions of Theorem 6.4.6 are satisfied, provided
only that r > Iluolls. Consequently we have:
Theorem 5.6. For every

Uo E HS(IR), s ~ 3 there is aT> ° such that the
t ~ 0, - 00 < x < 00
u{O, x) = uo{x)
has a unique solution u E C([O, T] : W(IR)) n CI([O, T] : U(IR)).
Bibliographical Notes and Remarks
The abstract theory of semigroups of linear operators is a part of functional

analysis. As such it is covered to some extent by many texts of functional
analysis. The most extensive treatise of the subject is the classical book of
Hille and Phillips [1]. Other general references are the books of Butzer and
Berens [1]. Davies [1], Dunford and Schwartz [1], Dynkin [1], Friedman [1],
Ladas and Lakshmikantham [1], Kato [9], Krein [1], Martin [1], Reed and
Simon [1], Riesz and Nagy [1], Rudin [1], Schechter [4], Tanabe [6], Walker
[1], Yosida [7] and others.
A good introduction to the abstract theory as well as to some of its
applications is provided by the lecture notes of Yosida [3], Phillips [7] and
Goldstein [3].
The theory of semigroups of bounded linear operators developed quite
rapidly since the discovery of the generation theorem by Hille and Yosida in
1948. By now, it is an extensive mathematical subject with substantial
applications to many fields of analysis. Only a small part of this theory is
covered by the present book which is mainly oriented towards the applica-
tions to partial differential equations. We mention here briefly some themes
which are not touched at all in this book.
Most of the classical theory of semigroups of bounded linear operators on
a Banach space has been extended to equi-continuous semigroups of class
Co in locally convex linear topological spaces. The first work in this
direction was done by L. Schwartz [1]. Most of the classical results of the
theory were generalized to this case by K. Yosida [7]. Further results in a
more general set up are given in Komatsu [2], Dembart [1], Babalola [1],
Ouchi [1] and Komura [1].
The theory was also generalized to semigroups of distributions. The first
results in this direction are due to J. L. Lions [1]; see also Chazarin [1], Da
Prato and Mosco [1], Fujiwara [1] and Ushijima [1], [2].
In the present book we deal only with strongly continuous semigroups.

Different classes of continuity at zero were introduced and studied in
Hille-Phillips [1]. Some more recent results on semigroups which are not Co
semigroups can be found in Oharu [I]. Oharu and Sunouchi [I], Miyadera,
Oharu and Okazawa [1], Okazawa [2] and Miyadera [3].
The theory of semigroups of bounded linear operators is closely related to
the solution of ordinary differential equations in Banach spaces. Usually,
each "well-posed" linear autonomous initial value problem gives rise to a
semigroup of bounded linear operators. The book of S. G. Krein [1] studies
the theory of semigroups from this point of view. There are however
interesting results on differential equations in Banach spaces which are not
well posed. In this direction we mention the work of Agmon and Nirenberg
[1]; see also Lions [2], Lax [1], Zaidman [1], Ogawa [1], pazy [1], Maz'ja and
Plamenevskii [1] and Plamenevskii [1].
As we have just hinted semigroups of operators are obtained as solutions
of initial value problems for a first order differential equation in a Banach
space. Most of the theory deals with a single first order equation. The reason
for this is that higher order equations can be reduced to first order systems
and then by changing the underlying Banach space one obtains a first order
single equation. There are however results for higher order equations which
cannot be obtained by such a reduction and there are other results in which
it is just more convenient to treat the higher order equation directly. We
refer the interested reader to S. G. Krein [1] Chapter 3 for a discussion of
equations of order two. Further references are Fattorini [1], [2], Goldstein
[2], [4], Sova [1], Kisynski [3], [4], Nagy [1], [2], [3], Travis and Webb [1], [2],
Rankin [1] and others.
In recent years the theory of semigroups of bounded linear operators has
been extended to a large and interesting theory of semigroups of nonlinear
operators in Hilbert and Banach spaces. We mention here only a few
general references to the subject; Benilan, Crandall and pazy [I], Brezis [I],
Barbu [2], Crandall [1], Yosida [7], pazy [4], [8] and Pavel [3].
Before we tum to a somewhat more detailed bibliographical account on
the material presented in this book we note that no attempt has been made
to compile a complete bibliography even of those parts of the theory which
are covered by the present book. Most references given are only to indicate
sources of the material presented, or closely related topics, and sources for
further reading. An extensive bibliography of the subject was compiled by
J. A. Goldstein and will appear in a forthcoming book by him.
Section 1.1. The results on semigroups of bounded linear operators which

are continuous in the uniform operator topology at t = 0, or equivalently,
semigroups which are generated by bounded linear operators can be consid-
ered as results about the exponential function in a Banach algebra. This
approach was taken by M. N agumo [1] and K. Yosida [1], see also
Hille-Phillips [1] Chapter V. The representation of uniformly continuous
groups of operators as an exponential of a bounded operator was also

obtained by D. S. Nathan [1].
Section 1.2. Most of the results of this section are standard and can be
found in every text dealing with semigroups of linear operators e.g. all the
texts mentioned at the beginning of these bibliographical notes.
The proof of Theorem 2.7 follows a construction of I. Gelfand [1].
Lemma 2.8 is an extension of a classical inequality (Example 2.9) of E.
Landau. In the present form it is due to Kallman and Rota [1]. For the case
of a Hilbert space, T. Kato [12] proved that if T(t) is a semigroup of
contractions then 2 is the best possible constant in (2.13). For general
Banach spaces, the best possible constant seems to be unknown. More
details on related inequalities are given in Certain and Kurtz [1], see also
Holbrook [1].
Section 1.3. The main result of this section is Theorem 3.1 which gave the
first complete characterization of the infinitesimal generator of a strongly
continuous semigroup of contractions. This result was the starting point of
the subsequent systematic development of the theory of semigroups of
bounded linear operators. It was obtained independently by E. Hille [2] and
K. Y osida [2]. Our proof of the sufficient part of the theorem follows the
ideas of K. Yosida [2]. The bounded linear operator AA appearing in this
proof is called the Yosida approximation of A. Hille's proof is based on a
direct proof of the convergence of the exponential formula
( t)-n x
T( t ) x = lim I - - A
n .... oo n
for x E D(A 2 ), see e.g. Tanabe [6] Section 3.1.
Section 1.4. The results of this section for the special case where X = H is a
Hilbert space are due to R. S. Phillips [5]. The extension to the general case
was carried out by Lumer and Phillips [1]. We note in passing that the
characterization of the infinitesimal generator A of a semigroup of contrac-
tions as an m-dissipative operator i.e. a dissipative operator for which the
range of "-I - A, "- > 0 is all of X plays an essential role in the theory of
nonlinear semigroups.
Section 1.5. The main result of this section is Theorem 5.2 which gives a
complete characterization of the infinitesimal generator of a Co semigroup
of bounded linear operators and thus generalizes the Hille-Y osida theorem
which was restricted to the characterization of the generator of a Co
semigroup of contractions. Theorem 5.2 was obtained independently and
almost simultaneously by W. Feller [1], I. Miyadera [1] and R. S. Phillips [2].
Our proof of the theorem is a simplification of Feller's proof.
Another way to prove the sufficient part of Theorem 5.2 is to prove

Theorem 5.5 directly using a straightforward generalization of the proof of
the sufficient part of the Hille-Yosida theorem, see e.g. Dunford-Schwartz
[1] Chapter VIII.
Section 1.6. The study of semigroups of linear operators started actually

with the study of groups of operators. The first results were those for groups
generated by bounded linear operators (see Section 1.1). These works were
followed by M. Stone [1] and J. von-Neumann [1]. Theorem 6.3 is due to E.
Hille [1] and Theorem 6.6 is due to J. R. Cuthbert [1].
Section 1.7. The results about the inversion of the Laplace transform are
standard. Better results for the inversion of the Laplace transform can be
obtained by a somewhat more delicate analysis, see Hille-Phillips [1] Chapter
II. The conditions of Theorem 7.7 imply actually that A is the infinitesimal
generator of an analytic semigroup (see Section 2.5). Usually one proves for
such an A directly, using the Dunford-Taylor operator calculus, that U(t)
defined by (7.26) is a semigroup of bounded linear operators and that A is
its infinitesimal generator, see e.g. Friedman [1] Part 2 Section 2. Instead, we
prove that the condition (7.24) implies the conditions of Theorem 5.2 and A
is thus the infinitesimal generator of a Co semigroup.
Section I.S. Theorem 8.1 is due to E. Hille [1]. In this context see also
Dunford-Segal [1]. It is interesting to note that this paper of Dunford and
Segal stimulated K. Yosida strongly and led him to the characterization of
the infinitesimal generator Of a Co semigroup of contractions, Yosida [2].
Theorem 8.3 is due to E. Hille (see Hille-Phillips [ID. The exponential
formula given in Theorem 8.3 served as a base of Hille's proof of the
characterization of the infinitesimal generator of a Co semigroup of contrac-
tions. This formula was also the starting point of the theory of semigroups
of nonlinear contractions in general Banach spaces which started in 1971 by
the fundamental result of Crandall and Liggett. The proof of Theorem 8.3
that we give here follows Hille-Phillips [1]. A different proof of a more
general result is given in Section 3.5.
Section 1.9. The results of this section are based on Hille-Phillips [1]
Chapter V and Kato [3]. See also Kato [9] Chapter 8.
Section 1.10. In the definition of the adjoint semigroup we follow Phillips

[4], see also Hille-Phillips [1] Chapter XIV and K. Yosida [7] Chapter IX in
which an extension, by H. Komatsu [2], of the results of Phillips to locally
convex spaces is given.
A slightly different approach which leads however to the same strongly
continuous semigroups is taken by Butzer and Berens [1] Chapter I. Theo-
rem 10.8 is due to M. Stone [I] and was the first result concerning
semigroups generated by an unbounded linear operator.
Section 2.1. The algebraic semigroup property, T(t + s) = T(t)T(s), am-

plifies many topological properties of the semigroup T( t). Theorem 1.1
which is due to K. Yosida [3] is one example of such an effect. Another
example, Hille-Phillips [I] Chapter X, is;
Theorem. If T(t) is a semigroup of bounded linear operators which is strongly

measurable on ]0, oo[ then it is strongly continuous on ]0, 00[. Ifmoreover T(t)
is weakly continuous at t = 0 then T(t) is a Co semigroup.
Section 2.2. The results of this section cover most of the results of Chapter
XVI of Hille-Phillips [I]. While the proofs of the results in Hille-Phillips [1]
use the Gelfand representation theory, our proof of Theorem 2.3 is com-
pletely elementary and follows the approach taken in Hille [1]. Theorems
2.4, 2.5 and 2.6 also follow Hille [I].
A counter example to the converse of Theorem 2.6 is given in Hille-
Phillips [1] (page 469), see also Greiner, Voigt and Wolff [I]. Further results
on the spectral mapping theorem for Co semigroups of positive operators
can be found in Greiner [I], Derdinger [1] and Derdinger and Nagel [I].
Section 2.3. Theorem 3.2 is due to P. D. Lax (see Hille-Phillips [I] Chapter
X). Theorem 3.3 and Corollaries 3.4 and 3.5 are taken from pazy [3].
Theorem 3.6 comes from Hille-Phillips [I] Chapter XVI but, while the proof
there uses the Gelfand representation theory, our proof is elementary.
Theorem 3.6 gives a necessary condition for an infinitesimal generator A to
generate a Co semigroup which is continuous in the uniform operator
topology for t > O. It seems that a full characterization of the infinitesimal
generator of such semigroups in terms of properties of their resolvents is not
known.
Section 2.4. Some early results on the differentiability of Co sernigroups

were obtained by E. Hille [3] and K. Yosida [4]. The full characterization of
the infinitesimal generator of a differentiable semigroup, Theorem 4.7, is due
to pazy [3]. Theorem 4.7 was extended to semigroups of distributions by V.
Barbu [I] and to semigroups of linear operators on locally convex spaces by
M. Watanabe [I]. Corollary 4.10 is due to Yosida [4]. Theorem 4.11 and
Corollaries 4.12 and 4.14 come from pazy [5].
Section 2.5. Theorem 5.2 is due to E. Hille [1]. Our proof follows Yosida
[4]. Theorem 5.3 is due to E. Hille [3]. Theorem 5.5 is taken from Crandall,
pazy and Tartar [I] while Theorem 5.6 is due to Kato [10]. Corollary 5.7 is
due to J. Neuberger [1] and T. Kato [10]. Corollary 5.8 seems to be new. The
uniform convexity of the underlying space or a similar condition is neces-
sary since there are concrete examples of analytic semigroups of contrac-

tions for which limt-+ollI - T(t)1I = 2, G. Pisier (private communication).
Related to the results of this section are also the deep results of A.
Beurling [1] and M. Certain [1].
Section 2.6. Let A be the infinitesimal generator of a Co semigroup. The

fractional powers of - A were first investigated by S. Bochner [1] and R. S.
Phillips [1]. Later A. V. Balakrishnan [1], [2] gave a new definition of the
fractional powers of - A and extended the theory to a wider class of
operators. About the same time several other authors contributed to this
subject. Among them M. Z. Solomjak [1], K. Yosida [5], T. Kato [4], [5], [7],
Krasnoselskii and Sobolevskii [1], J. Watanabe [1]. Subsequently, H.
Komatsu gave a unified point of view in a series of papers Komatsu [3]-[7].
Our simplified treatment follows mainly Kato [4] and [5], see also
Friedman [1] Part 2 Section 14 and Tanabe [6] Section 2.3.
Section 3.1. The results of this section are due to R. S. Phillips [2]. For
related results see Hille-Phillips [1] Chapter XIII and Dunford-Schwartz [1]
Chapter 8. Phillips [2] also started the study of properties of Co semigroups
which are conserved under bounded perturbations (i.e. perturbations of the
infinitesimal generator by a bounded operator). Among other results he
showed that continuity in the uniform operator topology for t > 0 is
conserved while the same property for t > to > 0 is not conserved. The
problem whether or not the differentiability for t > 0 of a semigroup T( t) is
conserved under bounded perturbations of its generator seems to be still
open. For a result related to this problem see pazy [3].
Section 3.2. Theorem 2.1 is due to E. Hille [1], see also T. Kato [9] Chapter
9 and Hille-Phillips [1] Chapter XIII. A related result is given in Da Prato
[1].
Section 3.3. Corollary 3.3 was essentially proved by H. F. Trotter [2] for the
case a < t, see also Kato [9] Chapter 9. The general case of Corollary 3.3
with a < 1 was proved by K. Gustafson [1]. Theorem 3.2 is a consequence
of a more symmetric version of Corollary 3.3 proved in pazy [9].
Theorem 3.4 was proved by P. Chernoff [2]. Corollary 3.5 is due to P.
Chernoff [2] and N. Okazawa [1], it is a generalization of the result of R.
Wust [1] in Hilbert space.
Section 3.4. The main results of this section are due to H. F. Trotter [1]. J.
Neveu [1] has proved the convergence theorem (Theorem 4.5) for the special
case of semigroups of contractions independently. Convergence results of a
similar nature are also given in T. Kato [9] and T. Kurtz [1], [2]. In Trotter
[1] the proof that the limit of the resolvents R(A: An) of An is itself a
resolvent of some operator A is not clear. This was pointed out and
corrected by T. Kato [3]. In Theorem 4.5 the condition that (AoI - A)D is
dense in X assures that A (the closure of A) is an infinitesimal generator of a

Co semigroup. A different necessary and sufficient condition for this is given
in M. Hasegawa [l]. An interesting proof of Trotter's theorem was given by
Kisynski [2].
Trotter [1], treats also the question of convergence of Co semigroups
acting on different Banach spaces. Results of this nature are very useful in
proving the convergence of solutions of certain difference equations to the
solutions of a corresponding partial differential equation. An example of
this type is given in Section 3.6 below.
Convergence in a Banach space, of semigroups which are not Co semi-
groups was studied by I. Miyadera [2] and Oharu-Sunouchi [1].
The convergence results were also extended to semigroups on locally
convex spaces, see e.g. K. Yosida [7], T. Kurtz [2] and T. I. Seidman [l].
Section 3.5. Lemma 5.1 is a simple extension of Corollary 5.2 which is due
to P. Chernoff [l]. Theorem 5.3 and Corollary 5.4 are also extensions of the
results of Chernoff [1]. Corollary 5.5 is an extension of the Trotter product
formula, Trotter [2]. With regard to the conditions of this formula see Kurtz
and Pierre [l].
Section 3.6. The results of this section are relevant to the numerical
solutions of partial differential equations. They are siinilar in nature to the
results of Trotter [l] and Kato [9]. For results of similar nature see also
Lax-Richtmyer [1] and Richtmyer-Morton [l].
Section 4.1. The initial value problem (1.1) in the Banach space X is called
an abstract Cauchy problem. The systematic study of such problems started
with E. Hille [4]. The uniqueness theorem (Theorem 1.2) is due to Ljubic [1].
Theorem 1.3 is due to Hille [4], see also Phillips [3]. Sufficient conditions for
the existence of a solution of (1.1) for a dense subset D of X (not necessarily
equal to D(A)) of initial data are given in R. Beals [1].
A different way of defining a weak solution of (1.1) was given by J. Ball
[l], see remarks to the next section.
Section 4.2. Definition 2.3 defines a mild solution of (2.1) if A is the
infinitesimal generator of a semigroup T(t). J. M. Ball [1] defines a "weak
solution" of the equation
du
- = Au + f(t) (E)
dt
where A is a closed linear operator on X and f E LI(O, T; X) as follows:
Definition. A function u E C([O, T] : X) is a weak solution of (E) on [0, T]

if for every v* E D( A*) the function (u( t), v*) is absolutely continuous on
[0, T] and
d
dt (u(t), v*) = (u(t), A*v*) + U(t), v*) a.e. on [0, T].
He then proves,
Theorem (Ball). There exists for each x E X a unique weak solution u of (E)
on [0, T] satisfying u(O) = x if and only if A is the infinitesimal generator of a
Co semigroup T(t) of bounded linear operators on X, and in this case u is given
by
u{t} = T{t}x + fT{t - s}f{s} tis, 0:::;; t:::;; T.

o
Corollaries 2.5 and 2.6 are due to Phillips [2], see also T. Kato [9].
Theorem 2.9 is a straightforward generalization of Theorem 2.4.
Section 4.3. Theorem 3.1 is essentially due to A. pazy [7]. There it was only
proved that u is Holder continuous with exponent f3 satisfying f3 < 1 - l/p.
The fact that the result is true for f3 = 1 - l/p is due to L. Veron. Theorem
3.2 comes from Crandall and pazy [1]. Corollary 3.3, for the more general
situation where A depends on t (see Chapter 5) was proved by H. Tanabe
[2], P. E. Sobolevskii [4], E. T. Poulsen [1] and Kato [9]. Theorem 3.5 is due
to Kato [9], see also Da Prato and Grisvard [1]. Optimal regularity condi-
tions for this problem are given in E. Sinestrari [1].
Section 4.4. Theorem 4.1 was taken from pazy [6]. It is a simple generaliza-
tion of a previous result of R. Datko [1]. The idea of Example 4.2 is taken
from Greiner, Voigt and Wolff [1]. Other examples of this sort are also given
in Hille-Phillips [1] Chapter XXIII and Zabczyk [1].
A more general result than Theorem 4.3 was proved by M. Slemrod [1].
The same problem is also treated in Derdinger and Nagel [I] and Derdinger
[1].
Theorem 4.5 was taken from S. G. Krein [1] Chapter 4.
Section 4.5. The results of this section are technical and they are brought
here mainly as a preparation to the first sections of Chapter 5. In this
section we follow closely the results of T. Kato [11], see also H. Tanabe [6],
Chapter 4.
Section 5.1. The results of this section are completely elementary and their
sole aim is to motivate the rest of the results of this chapter and to
familiarize the reader with the notion and main properties of evolution-sys-
tems. The term "evolution-system" is not standard, some authors call it a
propagator, others a fundamental solution and still others an evolution-
operator.
Section 5.2. The results of this section follow those of Kato [11]. The notion
of stability defined here is stronger than the usual one used in the theory of
finite difference approximations. When A(t) is independent of t then the
stability condition coincides with the condition of Theorem 1.5.3 and
therefore we can renorm the space so that in the new norm A generates a
semigroup T(t) satisfying II T(t) II :::;; e"'t. If A(t) depends on t but D(A(t»
is independent of t and the operators A(t) commute for t ~ 0 then it is not

difficult to show that the stability of A(t) implies that X can be renormed so
that in the new norm IISt(s)1I :::::; eWS for every t E [0, T] where St(s) is the
semigroup generated by A(t).
Sections 5.3-5.5. The first construction of an evolution system for the

initial value problem (3.1) with unbounded operators A(t) was achieved by
T. Kato [1]. His main assumptions were that D(A(t» = D is independent of
t and that for each t ~ 0, A(t) is the infinitesimal generator of a Co
contraction semigroup on X together with some continuity conditions on the
family of bounded operators A(t)A(s )-1. The main result of Kato [l] is
essentially a special case of Theorem 4.8. In an attempt to extend the results
of Kato [1] and especially to remove the assumption that D(A(t» is
independent of t, several authors constructed evolution systems under a
variety of conditions, see e.g. Elliot [l], Goldstein [1], Heyn [1], Kato [2],
Kisynski [1], Yosida [7], [6] and others.
Our presentation follows closely that of T. Kato [11], [13] with a simplifi-
cation due to Dorroh [1], see also H. Tanabe [6] Chapter 4.
A different method of studying the evolution equations (3.1) directly in
the space LP(O, T: X), using a sum of operators technique is developed in
Da Prato and Iannelli [1], see also Da Prato and Grisvard [1] and Iannelli
[1].
Finally we note that the special partitions needed for Remark 3.2 are
constructed in the appendix of Kato [l3] or else in Evans [1].
Sections 5.6-5.7. The first evolution systems in the parabolic case were
constructed by H. Tanabe [l], [2], [3] and independently but by a similar
method by P. E. Sobolevskii [4]. In these works it was assumed that D(A(t»
is independent of t. This assumption was somewhat relaxed by T. Kato [6]
and P. E. Sobolevskii [1], [3] who assumed that D(A(t)Y) for some 0 < y < 1
is independent of t. Later, T. Kato and H. Tanabe [1], [2] succeeded in
removing the assumption that D(A(t» is independent of t. They replaced it
by some regularity assumptions on the function t --+ R(A: A(t». In this
context higher differentiability of the solution is obtained if one assumes
higher differentiability of t -+ R(A: A(t» see Suryanarayana [1]. Assuming
that the conditions hold in a complex neighborhood of [0, T] one obtains
solutions of (6.2) that can be extended to a complex neighborhood of ]0, T]
see Komatsu [1], Kato Tanabe [2] and K. Masuda [1]. K. Masuda [1]
showed further that in this particular situation the Kato-Tanabe conditions
are also necessary for the existence of an evolution system.
In Section 5.6 and 5.7 we deal only with the case where D(A(t» is
independent of t. We follow Tanabe [2], Sobolevskii [4] and Poulsen [1], see
also H. Tanabe [6] Chapter 5 where the case of variable D(A(t» is also
treated.
A different approach to the solution of the evolution equation (6.1) (with

D(A(t» independent of t) which is not based on a construction of an
evolution system for (6.2) is given in Da Prato and Sinestrari (1].
Section 5.S. Theorem 8.2 is due to H. Tanabe [4] and Theorem 8.5 is due to
pazy [2].
A subject which is related to the asymptotic behavior of solutions of the
evolution equation (8.1) and which has not been touched in this chapter is
singular perturbations, see e.g. Tanabe [5] and Tanabe and Watanabe {l].
Section 6.1. Theorems 1.2, 1.4 and 1.5 are due to I. Segal {l], see also T.
Kato [8]. An example in which f is Lipschitz continuous but the mild
solution of (1.1) is not a strong solution can be found in Webb [1].
Theorems 1.6 and 1.7 are simple but useful modifications of the previous
results.
We note that the Lipschitz continuity of f can be replaced by accretive-
ness and one still obtains, under suitable conditions, global solutions of the
initial value problem (1.1) see e.g. Kato [8], Martin [I] Chapter 8 and the
very general paper of N. Pavel [2].
Section 6.2. The results of this section are based on pazy [7]. Examples in
which (3.1) with A = 0 and f continuous does not have solutions are given
e.g. in Dieudonne [I] page 287 and J. Yorke [I]. It is known, in fact, that
with A = 0 the initial value problem (3.1) has a local strong solution for
every continuous f if and only if X is finite dimensional, Godunov [1].
The main existence result, Theorem 2.1, of this section was extended by
N. Pavel [I] as follows:
Theorem (Pavel). Let D c X be a locally closed subset of X, f:[to, tl[ -+ X

continuous and let S(t), t ~ 0 be a Co semigroup, with S(t) compact for t > O.
A necessary and sufficient condition for the existence of a local solution
u:[to' T(to: xo)] -+ D, where to < T(to: x o ) ~ t l , to (3.1) for every Xo E D
is
limh-ldist(S(h)z + hf(t, z): D) = 0
h .... O
for all t E [to' tl[ and zED.
Section 6.3. The main result of this section, Theorem 3.1, is motivated by
the work of H. Fujita and T. Kato {l]. Similar and more general existence
results of this type can be found in Sobolevskii [4], Friedman [I] Part 2
Section 16 and Kielhofer [3].
The treatise of D. Henri [I] "Geometric theory of semilinear parabolic
equations" contains along with an existence result similar to Theorem 3.1 an
extensive study of the dependence of the solutions on the data, their
asymptotic behavior and many interesting applications.
Results which are to some extent between those of this section and the
previous one are given in Lightboume and Martin [1] and in Martin [2]. In
these results f is assumed to be continuous (but not necessarily Lipschitz
continuous) with respect to some fractional power of A and S(t), the
semigroup generated by - A, is assumed to be compact for t > o.
The existence results of the previous sections were stated for the autono-
mous case (i.e. A independent of t) mainly for the sake of simplicity. They
can be extended to the nonautonomous case as is actually done in Segal [I]
and Pruss [1] for the results of Section 6.1, in Fitzgibbon [1] for those of
Section 6.2 and in Sobolevskii [4], Friedman [1] and Kielhofer [3] for those
of Section 6.3.
Some asymptotic results for nonautonomous semilinear evolution equa-
tions are given in Nambu [I].
Section 6.4. The results of this section follow closely Kato [14], see also
Kato [15]. A different method to treat similar equations was recently
developed by Crandall and Souganidis [I].
Section 7.1. As we have already mentioned in the introduction, the present

book's main aim is the applications of semigroup theory to partial differen-
tial equations. The purpose of this and the next chapter is to present some
examples of such applications.
A detailed study of Sobolev spaces is given in Adams [I], other references
are Necas [I], Friedman [I] and Lions-Magenes [1].
Section 7.2-7.3. In the applications presented in these sections we restrict

ourselves, for the sake of simplicity, to the Dirichlet boundary conditions.
All the results hold for more general boundary conditions see e.g. Agmon
[1], Stewart [2], Tanabe [6] Section 3.8 pazy [2] and others. The needed
a-priori estimates for the elliptic operators with general boundary conditions
are given in Agmon, Douglis and Nirenberg [I], Nirenberg [1], Schechter [I],
[2], [3] and Stewart [2], see also Lions-Magenes [1].
Theorem 2.2 is due to L. GArding [I], for a proof see e.g. Agmon [2],
Friedman [I], Yosida [7]. The regularity of solutions of elliptic boundary
value problems (Theorem 2.3) was proved for general boundary values and
I < p < 00 by Agmon, Douglis and Nirenberg [1] and for the Dirichlet
boundary values by Nirenberg [1], see also Agmon [2], Friedman [1] and
Lions-Megenes [I].
Theorem 3.1 is due to Agmon, Douglis and Nirenberg [1], Theorems 3.2
and 3.5 are due to Agmon [1] and Theorem 3.7 to Stewart [1].
Another interesting example of an operator that generates an analytic
semigroup is the classical Stokes operator. For details see Giga [1].
Section 7.4. In this section we follow the treatment of K. Yosida [3], [7] in
which more general hyperbolic equations are also treated.
Section 7.5. A proof of the classical Hausdorff-Young theorem used in this

section can be found e.g. in Stein and Weiss [1] Chapter V.
Section 7.6. Results similar to those presented in this section, with more
general boundary conditions can be found in Tanabe [6] Chapter 5 and
Friedman [1] Part 2 Sections 9, 10.
Section 8.1. The results of this section are due to Baillon, Cazenave and
Figueira [1] and to Ginibre and Velo [1]. Our presentation follows that of
Baillon et al. Related results can be found in Lin and Strauss [1], Pecher and
von Wahl [1] and Haraux [1].
Theorem 1.5 is also true in a bounded domain 0 in R2. The local
existence of the solution in this case is similar to the case on all of R 2 while
the global existence is more complicated since one cannot apply Sobolev's
imbedding theorem in a straightforward way. To prove the global existence
in this case a new interpolation-imbedding inequality is used, see Brezis and
Gallouet [1].
Section 8.2. The results of this section follow closely pazy [7].
Sections 8.3-8.4. In these two sections fractional powers of minus the

infinitesimal generators of analytic semigroups are used to obtain, via the
abstract results of Section 6.3, solutions of certain nonlinear initial value
problems for partial differential equations.
The results of Section 8.3 follow rather closely the ideas of Fujita and
Kato [1] in which the linear operator A is more complicated than in our
case. Lemma 3.3 is due to Fujita-Kato [1].
The results of Section 8.4 follow those of Sobolevskii 14] and Friedman
[1]. Results of similar nature in Holder spaces and for unbounded domains
can be found in Kielhofer [1], [2].
The Gagliardo-Nirenberg inequalities used in this section are proved e.g.
in Friedman [1] Part 1 Sections 9, 10.
In certain cases global solutions can be obtained, usually using some
further conditions, see e.g. Kielhofer [3] and von Wahl [1], [2]. For the
Navier-Stokes equations in R2 see Fujita-Kato [1] and Sobolevskii [2].
Finally we note that for the sake of simplicity we have chosen to take the
linear operator A to be independent of t. Similar results can be obtained
when A depends on t, see e.g. Friedman [1] and Kielhofer [3].
Section 8.5. The results of this section follow one of many examples given
in Kato [14]. For this particular example better results including a global
existence theorem are given in Kato [16].
Bibliography
R. A. Adams
[I] Sobo/ev spaces, Academic Press, New York (1975).
S. Agmon
[I] On the eigenfunctions and on the eigenvalues of general elliptic boundary
value problems, Comm. Pure Appl. Math. 15 (1962) 119-147.
[2] Elliptic boundary value problems, VanNostrand (1965).
S. Agmon and L. Nirenberg
[I] Properties of solutions of ordinary differential equations in Banach spaces,
Comm. Pure App/. Math. 16 (1963) 121-239.
S. Agmon, A. Douglis and L. Nirenberg
[1] Estimates near the boundary for solutions of elliptic partial differential
equations, Comm. Pure App/. Math 12 (1959) 623-727.
V. A. Babalola
[1] Semigroups of operators on locally convex spaces, Trans. A mer. Math. Soc.
199 (1974) 163-179.
J. B. Baillon, T. Cazenave, and M. Figueira
[I] Equation de Schrodinger non lineaire, C.R. A cad. Sc. Paris 284 A (1977)
869-872.
A. V. Balakrishnan
[I] An operator calculus for infinitesimal generators of semi-groups, Trans.
Amer. Math. Soc. 91 (1959) 330-353.
[2] Fractional pow~rs of closed operators and the semi-groups generated by
them, Pacific 1. Math. 10 (1960) 419-437.
J. M. Ball
[I] Strongly continuous semigroups, weak solutions and the variation of con-
stants formula, Proc. Amer. Math. Soc. 63 (1977) 370-373.
Bibliography 265
V. Barbu
[I) Differentiable distribution semi-groups, Anal. Scuola Norm. Sup. Pisa 23
(1969) 413-429.
(2) Nonlinear semigroups and differential equations in Banach spaces, Noordhoff
Int. Pub!. Leyden the Netherlands (1976).
R. Beals
[I) On the abstract Cauchy problem, J. Func. Anal. 10 (1972) 281-299.
Ph. Benilan, M. Crandall and A. pazy
[I) Nonlinear evolution governed by accretive operators (a book to appear).
A. Beurling
[I) On analytic extension of semi-groups of operators, J. Func. Anal. 6 (1970)
387-400.
S. Bochner
[I) Diffusion equations and stochastic processes, Proc. Nat. Acad. Sc. U.S.A. 35
(1949) 368-370.
P. L. Butzer and H. Berens
[I) Semi-groups of operators and approximation, Springer-Verlag New York
(1967).
H. Brezis
[I) Operateurs maximaux monotone et semigroups de contractions dans les espaces
de Hilbert, Math. Studies 5, North Holland (1973).
H. Brezis and T. Gallouet
[I) Nonlinear SchrOdinger evolution equations, Nonlinear Anal. TMA 4 (1980)
677-682.
M. Certain
[I) One-parameter semigroups holomorphic away from zero, Trans. Amer. Math.
Soc. 187 (1974) 377-389.
M. Certain and T. Kurtz
[I) Landau-Kolmogorov inequalities for semigroups and groups, Proc. A mer.
Math. Soc. 63 (1977) 226-230.
J. Chazarin
[I) Problemes de Cauchy abstrait et applications a quelques problemes mixtes, J.
Func. Anal. 7 (1971) 386-446.
P. Chernoff
[I) Note on product formulas for operator semi-groups, J. Func. Anal. 2 (1968)
238-242.
(2) Perturbations of dissipative operators with relative bound one, Proc. Amer.
Math. Soc. 33 (1972) 72-74.
M. G. Crandall
[I) An introduction to evolution governed by accretive operators, "Dynamical-
Systems-An international Symposium."(L. Cesari, J. Hale, J. LaSalle, Eds.)
Academic Press, New York (1976) 131-165.
M. G. Crandall and A. pazy
[I) On the differentiability of weak solutions of a differential equation in Banach
space, J. Math. and Mech. 18 (1969) 1007-1016.
M. G. Crandall, A. Pazy, and L. Tartar

[I] Remarks on generators of analytic semi groups, Israel J. Math 32 (1979)
363-374.
M. G. Crandall and P. E. Souganidis
[I] Nonlinear evolution equations MRC technical Rep. 2352 (1982).
1. R. Cu thbert
[I] On semi-groups such that ~ - I is compact for some t > O. Z. Whar. 18
(1971) 9- 16.
G. Da Prato
[I] Somma di genera tori infinitesimali di semigruppi analitici, Rend. Sem. Math.
Univ Padova 40 (1968) 151-161.
G. Da Prato and P. Grisvard
[I] Sommes d'operateurs lineaires et equations differentielles operationnelles, J.
Math. Pure et Appl. 54 (1975) 305-387.
G. Da Prato and M. Iannelli
[I] On a method for studying abstract evolution equations in the Hyperbolic
case, Comm. in Partial Diff. Eqs. 1 (1976) 585-608.
G. Da Prato and U. Mosco
[I] Semigruppi distribuzioni analitici, Ann. Scuola Norm. Sup. Pisa 19 (1965)
367-396.
G. Da Prato and E. Sinestrari
[I] Holder regularity for non autonomous abstract parabolic equations, Israel J.
Math. 42 (1982) 1-19.
R. Datko
[1] Extending a theorem of A. M. Liapunov to Hilbert space, J. Math. A nal. and
Appl. 32 (1970) 610-616.
E. B. Davies
[I] One-Parameter Semigroups, Academic Press, London (1980).
B. Dembart
[1] On the theory of semi groups of operators on locally convex spaces, J. Func.
Anal. 16(1974) 123-160.
R. Derdinger
[1] Uber das Spektrum positiver Generatoren, Math. Z. 172 (1980) 281-293.
R. Derdinger and R. Nagel
[I] Der Generator stark stetiger Verbandhalbgruppen auf C( X) und dessen
Spektrum, Math. Ann. 245 (1979) 159-177.
J. Dieudonne
[I] Foundation of modern analysis, Academic Press, New York (1960).
J. R. Dorroh
[1] A simplified proof of a theorem of Kato on linear evolution equations, J.
Math. Soc. Japan 27 (1975) 474-478.
N. Dunford and J. Schwartz
[I] Linear operators, Part I General theory, Interscience, New York (1958).
Bibliography 267
N. Dunford and 1. E. Segal

[I] Semi-groups of operators and the Weierstrass theorem, Bull. A mer. Math.
Soc. 52 (1946) 911-914.
E. B. Dynkin
[I] Markov Processes, Vol. 1. Springer-Verlag, Berlin (1965).
J. Elliot
[I] The equations of evolution in a Banach space, Trans. Amer. Math. Soc. 103
(1962) 470-483.
C. L. Evans
[I] Nonlinear evolution equations in an arbitrary Banach space, Israel J. Math.
26 (1977) 1-42.
H. Fattorini
[I] Ordinary differential equations in linear topological spaces 1. J. Diff. Eqs. 5
(1968) 72- \05.
[2] Ordinary differential equations in linear topological spaces II. J. Diff. Eqs. 6
(1969) 50-70.
W. Feller
[I] On the generation of unbounded semi-groups of bounded linear operators,
Ann. of Math. 58 (1953) 166-174.
W. E. Fitzgibbon
[I] Semilinear functional differential equations in Banach space, J. Diff. Eqs. 29
(1978) 1-14.
A. Freidman
[I] Partial differential equations, Holt, Reinhart, and Winston, New York (1969).
H. Fujita and T. Kato
[I] On the Navier-Stokes initial value problem 1. Arch. Rat. Mech. and Anal. 16
(1964) 269-315.
D. Fujiwara
[I] A characterization of exponential distribution semi-groups, J. Math. Soc.
Japan 18 (1966) 267-274.
L. Girding
[I] Dirichlet's problem for linear elliptic partial differential equations, Math.
Scan. 1 (1953) 55-72.
1. Gelfand
[I] On one-parameter groups of operators in normed spaces, Dokl. Akad. Nauk
SSSR. 25 (1939) 713-718.
Y. Giga
[I] Analyticity of the semigroup generated by the Stokes operator in Lr spaces,
Math. Z. 178 (1981) 297-329.
1. Ginibre and B. Velo
[I] On a class of nonlinear SchrOdinger equations, J. Func. Anal. 32 (1979) 1-71.
A. N. Godunov
[I] On Peano's theorem in Banach spaces, Func. Anal. and Appl. 9 (1975) 53-55.
J. A. Goldstein
[I] Abstract evolution equations, Trans. Amer. Math. Soc. 141 (1969) 159-185.
[2] Semi-groups and second order differential equations, J. Func. Anal. 4 (1969)
50-70.
[3] Semi-groups of operators and abstract Cauchy problems, Tulane Univ.
Lecture notes (1970).
[4] The universal addability problem for generators of cosine functions and
operators, Houston J. Math. 6 (1980) 365-373.
G. Greiner
[I] Zur Perron-Froebenius-Theorie stark stetiger Halbgruppen, Math. Z. 177
(1981) 401-423.
G. Greiner, J. Voigt and M. Wolff
[I] On the spectral bound of the generator of semigroups of positive operators,
J. Operator Theory 5 (1981) 245-256.
K. Gustafson
[I] A perturbation lemma, Bull. Amer. Math. Soc. 72 (1966) 334-338.
A. Haraux
[I] Nonlinear evolution equations-Global behavior of solutions, Lecture notes
in Math. 841 Springer-Verlag (1981).
M. Hasegawa
[I] On the convergence of resolvents of operators, Pacific J. Math. 21 (1967)
35-47.
D. Henry
[I] Geometric theory of semilinear parabolic equations, Lecture notes in Math.
840 Springer-Verlag (1981).
E. Heyn
[I] Die Differentialgleichung dT/dt = P(t)T fUr Operatorfunktionen, Math.
Nach. 24 (1962) 281-330.
E. Hille
[I] Representation of one-parameter semi-groups of linear transformations, Proc.
Nat. Acad. Sc. U.S.A. 28 (1942) 175-178.
[2] Functional analysis and semi-groups, Amer. Math. Soc. Colloq. Pub!. Vol. 31,
New York (1948).
[3] On the differentiability of semi-groups of operators, Acta Sc. Math. (Szeged)
12 (1950) 19-24.
[4] Une generalization du probleme de Cauchy, Ann. lnst. Fourier 4 (1952)
31-48.
E. Hille and R. S. Phillips
[I] Functional analysis and semi-groups, Amer. Math. Soc. Colloq. Publ. Vol. 31,
Providence R.I. (1957).
J. Holbrook
[I] A Kallman-Rota inequality for nearly Euclidean spaces, Adv. in Math. 14
(1974) 335-345.
M. Iannelli
[I] On the Green function for abstract evolution equations, Bul. U.M.I. 6
(1972) 154-174.
Bibliography 269
R. R. Kallman and G. C. Rota

[I) On the inequality II!' 112 ~ 411fll . II!" II, Inequalities, Vol. II. Academic Press
New York (1970) 187-192.
T. Kato
[I) Integration of the equation of evolution in Banach space, J. Math. Soc.
Japan 5 (I 953} 208-234.
[2) On linear differential equations in Banach spaces, Comm. Pure. Appl. Math.
9 (l956) 479-486.
[3) Remarks on pseudo-resolvents and infinitesimal generators of semi-groups,
Proc. Japan Acad. 35 (1959) 467-468.
[4) Note on fractional powers of linear operators, Proc. Japan Acad. 36 (I960)
94-96.
[5) Fractional powers of dissipative operators, J. Math. Soc. Japan 13 (l961)
246-274.
[6) Abstract evolution equations of parabolic type in Banach and Hilbert spaces,
Nagoya Math. J. 19 (I961) 93-125.
[7) Fractional powers of dissipative operators II, J. Math. Soc. Japan 14 (I 962}
242-248.
[8) Nonlinear evolution equations in Banach spaces, Proc. Symp. Appl. Math. 17
Amer. Math. Soc. (1965) 50-67.
[9) Perturbation theory for linear operators, Springer Verlag, New York (l966).
[10) A characterization of holomorphic semi-groups, Proc. A mer. Math. Soc. 25
(1970) 495-498.
[II) Linear evolution equations of "hyperbolic" type, J. Fac. Sc. Univ. of Tokyo
25 (1970) 241-258.
[I2) On an inequality of Hardy, Littlewood and Polya, Adv. in Math. 7 (1971)
107-133.
[13) Linear evolution equations of "hyperbolic" type II. J. Math. Soc. Japan 25
(1973) 648-666.
[14) Quasi-linear equations of evolution with applications to partial differential
equations, Lecture Notes in Math. 448, Springer Verlag (1975) 25-70.
[15) Linear and quasi-linear equations of evolution of hyperbolic type. C.I.M.E.
II CicIo (1976).
[I6) On the Korteweg-de Vries equation, Manuscripta Math. 28 (l979) 89-99.
T. Kato and H. Tanabe
[I) On the abstract evolution equation, Osaka Math. J. 14 (1962) 107-133.
[2) On the analyticity of solutions of evolution equations, Osaka J. Math 4
(1969) 1-4.
H. KielhOfer
[1) Halbgruppen und semilinear Anfangs-Randwertprobleme, Manuscripta Math.
12 (1974) 121-152.
[2) Existenz und Regularitiit von Losungen semilinearer parabolicher Anfangs-
Randwertprobleme, Math. Z. 142 (1975) 131-160.
[3) Global solutions of semilinear evolution equations satisfying an energy
inequality, J. DifJ. Eqs. 36 (l980) 188-222.
J. Kisynski
[I) Sur les operateures de Green des Problemes de Cauchy abstraits, Studia
Math. 23 (l963/4) 285-328.
[2] A proof of the Trotter-Kato theorem on approximation of semi-groups,

Colloq. Math. 18 (1967) 181-184.
[3] On second order Cauchy's problem in a Banach space, Bull. A cad. Pol. Sc.
18 (1970) 371-374.
[4] On cosine operator functions and one parameter groups of operators, Studia
Math. 144 (1972) 93-105.
H. Komatsu
[I] Abstract analyticity in time and unique continuation property of solutions of
a parabolic equation, J. Fac. Sc. Univ. of Tokyo 9 (1961) 1-11.
[2] Semi-groups of operators in locally convex spaces, J. Math. Soc. Japan 16
(1964) 230-262.
[3] Fractional powers of operators, Pacific J. Math. 19 (1966) 285-346.
[4] Fractional powers of operators II Interpolation spaces, Pacific J. Math. 21
(1967) 89-111.
[5] Fractional powers of operators III Negative powers, J. Math. Soc. Japan 21
(1969) 205-220.
[6] Fractional powers of operators IV Potential operators, J. Math. Soc. Japan
21 (1969) 221-228.
[7] Fractional powers of operators V Dual operators, J. Fac. Sc. Univ of Tokyo
17 (1970) 373-396.
T. Komura
[I] Semi-groups of operators in locally convex spaces, J. Func. Anal. 2 (1968)
258-296.
M. A. Krasnose1skii and P. E. Sobolevskii
[I] Fractional powers of operators defined on Banach spaces, Dokl. Akad. Nauk
SSSR 129 (1959) 499-502.
S. G. Krein
[I] Linear differential equations in Banach spaces, Translations Amer. Math. Soc.
29, Providence, RI (1971).
T. G. Kurtz
[I] Extensions of Trotter's operator semi-group approximation theorems, J.
Func. Anal. 3 (1969) 111-132.
[2] A general theorem on the convergence of operator semigroups, Trans. Amer.
Math. Soc. 148 (1970) 23-32.
T. G. Kurtz and M. Pierre
[1] A counterexample for the Trotter product formula, MRC Tech. Rep. 2091
( 1980).
G. E. Ladas and V. Lakshmikantham
[I] Differential equations in abstract spaces, Academic Press, New York (1972).
P. D. Lax
[I] On the Cauchy's problem for hyperbolic equations and the differentiability
of the solutions of elliptic equations, Comm. Pure Appl. Math. 8 (1955)
167-190.
P. D. Lax and R. D. Richtmyer
[I] Survey of the stability of linear finite difference equations, Comm. Pure Appl.
Math 9 (1956) 267-293.
Bibliography 271
J. H. Lightbourne and R. H. Martin

[I) Relatively continuous nonlinear perturbations of analytic semigroups, N on-
linear Anal. TMA 1 (1977) 277-292.
J. E. Lin and W. Strauss
[I) Decay and Scattering of solutions of nonlinear SchrOdinger equations,
J. Func. Anal. 30 (1978) 245-263.
J. L. Lions
[I) Les semi-groupes distributions, Portugal. Math 19 (1960) 141-164.
[2) Equations differentielles operationelles et problemes aux Iimites, Springer Verlag,
Berlin (1961).
J. L. Lions and E. Magenes
[I) Problemes aux limites non homogenes et applications, Vol. I, II Dunod Paris
(1968) Vol. III (1970).
Ju. I. Ljubic
[I) Conditions for the uniqueness of the solution of Cauchy's abstract problem,
Dokl. Akad. Nauk SSSR 130 (1960) 969-972.
G. Lumer and R. S. Phillips
[I) Dissipative operators in a Banach space, Pacific J. Math 11 (1961) 679-698.
R. H. Martin
[I) Nonlinear operators and differential equations in Banach spaces, John Wiley
and Sons, New York (1976).
[2) Invariant sets and a mathematical model involving semilinear differential
equations, Nonlinear equations in abstract spaces, Proc. Inter. Symp. Univ.
of Texas Arlington, Academic Press, New York (1978) 135-148.
K. Masuda
[I) On the holomorphic evolution operators, J. Math. Anal. Appl. 39 (1972)
706-711.
V. G. Maz'ja and B. A. Plamenevskii
[I) On the asymptotic behavior of solutions of differential equations in Hilbert
space, Math. USSR Izvestia 6 (1972) 1067-1116.
I. Miyadera
[I) Generation of a strongly continuous semi-group of operators, Tohoku Math.
J. 4 (1952) 109-114.
[2) Perturbation theory for semi-groups of operators (Japanese), Sugaku 20
(1968) 14-25.
[3) On the generation of semi groups of linear operators, Tohoku Math. J. 24
(1972) 251-261.
I. Miyadera, S. Oharu and N. Okazawa
[I) Generation theorems of semigroups of linear operators, Publ. R.I. M. S.
Kyoto Univ. 8 (1972/3) 509-555.
M. Nagumo
[I) Einige analytische Untersuchungen in linearen metrischen Rigen, Japan J.
Math. 13 (1936) 61-80.
B. Nagy
[I) On cosine operator functions in Banach spaces, Acta Sc. Math. 36 (1974)
281-289.
[2] Cosine operator functions and the abstract Cauchy problem, Periodica Math.
Hung. 7 (\976) 15-18.
[3] Approximation theorems for cosine operator functions, Acta Math. A cad. Sc.
Hung. 29 (1977) 69-76.
T. Nambu
[I] Asymptotic behavior of a class of nonlinear differential equations in Banach
space, SIAM J. Math. Anal. 9 (1978) 687-718.
D. S. Nathan
[I] One parameter groups of transformations in abstract vector spaces, Duke
Math. J. 1 (1935) 518-526.
J. Necas
[I] Les methodes directes en theorie des equations elliptiques, Masson and Cie Ed.
Paris (1967).
J. W. Neuberger
[I] Analiticity and quasi-analiticity for one-parameter semi-groups, Proc. Amer.
Math. Soc. 25 (1970) 488-494.
J. von-Neumann
[I] Uber die analytische Eigenschaften von Gruppen linearer Transformationen
und ihrer Darstellungen, Math. Z. 30 (1939) 3-42.
J. Neveu
[I] Theorie des semi-groupes de Markov, Univ. of Calif. Publ. in Statistics 2
(1958) 319-394.
L. Nirenberg
[I] On elliptic partial differential equations, Ann. Scuola Norm. Sup. Pisa 13
(1959) 1-48.
H. Ogawa
[I] Lower bounds for solutions of differential inequalities in Hilbert space, Proc.
A mer. Math. Soc. 16 (1965) 1241-1243.
[2] On the convergence of semi-groups of operators, Proc. Japan A cad. 42 (1966)
880-884.
S.Oharu
[I] Semigroups of linear operators in Banach spaces, Pub!. R.I.M.S., Kyoto
Univ. 7 (\971) 205-260.
S. Oharu and H. Sunouchi
[I] On the convergence of semigroups of linear operators, J. Func. Anal. 6 (1970)
292-304.
N.Okazawa
[I] A perturbation theorem for linear contraction semi-groups on reflexive
Banach spaces, Proc. Japan Acad. 47 (1971) 947-949.
[2] A generation theorem for semigroups of growth of order a, Tohoku Math. J.
26 (1974) 39-5\.
S.Ouchi
[I] Semi-groups of operators in locally convex spaces, J. Math. Soc. Japan 25
(1973) 265-276.
Bibliography 273
N. H. Pavel
[I] Invariant sets for a class of semi-linear equations of evolution, Nonlinear
Anal. TMA 1 (1977) 187-196.
[2] Global existence for nonautonomous perturbed differential equations and
flow invariance, Preprint Series in Math 22 (1981) INCREST, Bucharest.
[3] Analysis of some nonlinear problems in Banach spaces and applications,
Univ. "AI. I. Cuza" Iasi Facultatea de Matematica (1982).
A. pazy
[I] Asymptotic expansions of solutions of ordinary differential equations in
Hilbert space, Arch. Rat. Mech. and Anal. 24 (1967) 193-218.
[2] Asymptotic behavior of the solution of an abstract evolution equation and
some applications, J. Diff. Eqs. 4 (1968) 493-509.
[3] On the differentiability and compactness of semi-groups of linear operators,
J. Math. and Mech. 17 (1968) 1131-1141.
[4] Semigroups of nonlinear contractions in Hilbert space, CI.M.E. Varenna
1970 Ed. Cremonese (1971) 343-430.
[5] Approximation of the identity operator by semi-groups of linear operators,
Proc. Amer. Math. Soc. 30 (1971) 147-150.
[6] On the applicability of Lyapunov's theorem in Hilbert space, SIAM J. Math.
Anal. 3 (1972) 291-294.
[7] A class of semi-linear equations of evolution, Israel J. Math 20 (1975) 23-36.
[8] Semigroups of nonlinear contractions and their asymptotic behaviour, Non-
linear analysis and mechanics, Heriot-Watt Symp. Vol. III, R. J. Knops Ed.
Pitman Research notes in Math 30 (1979) 36-134.
[9] A perturbation theorem for linear m-dissipative operators, Memorias de
Matematica da U.F.R.J. 131 (1981).
H. Pecher and W. von-Wahl
[I] Tim<; dependent nonlinear SchrMinger equations, Manuscripta Math. 27
(1979) 125-157.
R. S. Phillips
[I] On the generation of semi-groups of linear operators, Pacific J. Math. 2
(1952) 343-369.
[2] Perturbation theory for semi-groups of linear operators, Trans. A mer. Math.
Soc. 74 (1953) 199-221.
[3] A note on the abstract Cauchy problem, Proc. Nat. A cad. Sci. U.S.A. 40
(1954) 244-248.
[4] The adjoint semi-group, PacificJ. Math. 5 (1955) 269-283.
[5] Dissipative operators and hyperbolic systems of partial differential equations,
Trans. Amer. Math. Soc. 90 (1959) 193-254.
[6] Dissipative operators and parabolic differential equations, Comm. Pure Appl.
Math. 12 (1959) 249-276.
[7] Semi-groups of contraction operators, Equazioni Differenzio Astratte,
C.I.M.E. (Rome) 1963.
B. A. Plamenevskii
[I] On the existence and asymptotic solutions of differential equations with
unbounded operator coefficients in a Banach space, Math. USSR Izvestia 6
(1972) 1327-1379.
E. T. Poulsen
[I] Evolutionsgleichungen in Banach Raumen, Math. Z. 90 (1965) 286-309.
J. Priiss
[I] On semilinear evolution equations in Banach spaces, J. Reine u. Ang. Mat.
303/4 (1978) 144-158.
S. M. Rankin
[I] A remark on cosine families, Proc. Amer. Math. Soc. 79 (1980) 376-378.
M. Reed and B. Simon
[I] Methods of modern mathematical physics II: Fourier analysis, Self-adjointness,
Academic Press, New York (1975).
F. Sz. Riesz and B. Nagy
[1] Functional Analysis, P. Ungar, New York (1955).
W. Rudin
[I] Functional analysis, McGraw-Hill, New York (1973).
R. D. Richtmyer and K. W. Morton
[I] Difference methods for initial value problems, Second ed. Interscience, New
York (1967).
M. Schechter
[I] Integral inequalities for partial differential operators and functions satisfying
general boundary conditions, Comm. Pure Appl. Math. 12 (1959) 37-66.
[2] General boundary value problems for elliptic partial differential equations,
Comm. Pure Appl. Math. 12 (1959) 457-486.
[3] Remarks on elliptic boundary value problems, Comm. Pure Appl. Math. 12
(1959) 561-578.
[4] Principles of functional ana(vsis, Academic Press, New York (1971).
L. Schwartz
[I] Lectures on mixed problems in partial differential equations and the repre-
sentation of semi-groups, Tata Inst. Fund. Research (1958).
1. Segal
[I] Non-linear semi-groups, Ann. of Math. 78 (1963) 339-364.
T. 1. Seidman
[I] Approximation of operator semi-groups, J. Func. Anal. 5 (1970) 160-166.
E. Sinestrari
[I] On the solutions of the inhomogeneous evolution equation in Banach space,
Rend. Acc. Naz. Lincei LXX (1981).
M. Slemrod
[I] Asymptotic behavior of Co semi-groups as determined by the spectrum of the
generator, Indiana Univ. Math. J. 25 (1976) 783-792.
P. E. Sobolevskii
[I] First-order differential equations in Hilbert space with variable positive
definite self-adjoint operator, a fractional power of which has a constant
domain of definition, Dokl. Akad. Nauk SSSR 123 (1958) 984-987.
[2] On non-stationary equations of hydrodinamics for viscous fluid, Dokl. Akad
Nauk SSSR 128 (1959) 45-48.
[3] Parabolic equations in a Banach space with an unbounded variable operator,
a fractional power of which has a constant domain of definition, Dokl. Akad.
Nauk SSSR 138 (1961) 59-62.
Bibliography 275
[4] On the equations of parabolic type in Banach space, Trudy Moscow Mat.
Obsc 10 (1961) 297-350, English trans. Amer. Math. Soc. Trans. 49 (1965)
1-62.
M. Z. Solomjak
[1] Applications of the semi-group theory to the study of differential equations
in Banach spaces, Dokl. Akad. Nauk SSSR 12 (1958) 766-769.
M. Sova
[1] Probleme de Cauchy pour equations hyperboliques operationelles a coeffi-
cients constants non-bornes, Ann. Scuola Norm. Sup. Pisa 22 (1968) 67-100.
E. M. Stein and G. Weiss
[1] Introduction to Fourier ana~vsis on Euclidean spaces, Princeton Univ. Press
(1971).
B. Stewart
[I] Generation of analytic semigroups by strongly elliptic operators, Trans.
Amer. Math. Soc. 199 (1974) 141-16l.
[2] Generation of analytic semigroups by strongly elliptic operators under gen-
eral boundary conditions. Trans. Amer. Math. Soc. 259 (1980) 299-310.
M. H. Stone
[1] On one-parameter unitary groups in Hilbert space, Ann. Math. 33 (1932)
643-648.
P. Suryanarayana
[1] The higher order differentiability of solutions of abstract evolution equations,
PacificJ. Math. 22 (1967) 543-56l.
H. Tanabe
[1] A class of the equations of evolution in a Banach space, Osaka Math. J. 11
(1959) 121-145.
[2] On the equations of evolution in Banach space, Osaka Math. J. 12 (1960)
363-376.
[3] Remarks on the equations of evolution in a Banach space Osaka Math. J. 12
(1960) 145-165.
[4] Convergence to a stationary state of the solution of some kind of differential
equations in a Banach space, Proc. Japan Acad. 37 (1961) 127-130.
[5] Note on singular perturbation for abstract differential equations, Osaka J.
Math. 1 (1964) 239-252.
[6] Equations of evolution (English translation) Pitman, London (1979).
H. Tanabe and M. Watanabe
[I] Note on perturbation and degeneration of abstract differential equations in
Banach space, Funkcialaj Ekvacioj 9 (1966) 163-170.
C. Travis and G. Webb
[I] Cosine families and abstract nonlinear second order differential equations,
Acta Math. A cad. Sc. Hung. 32 (1978) 75-96.
[2] Second order differential equations in Banach spaces, Nonlinear equations in
abstract spaces, Ed. V. Lakshmikantham, Academic Press, New York (1978)
331-36l.
H. F. Trotter
[I] Approximation of semi-groups of operators, Pacific J. Math. 8 (1958)
887-919.
[2] On the product of semi-groups of operators, Proc. Amer. Math. Soc. 10

(\ 959) 545-551.
T. Ushijima
[I] Some properties of regular distribution semi-groups, Proc. Japan A cad. 45
(\ 969) 224-227.
[2] On the generation and smoothness of semigroups of linear operators, J. Fac.
Sc. Univ. Tokyo 19 (\972) 65-127.
W. von-Wahl
[I] Instationary Navier-Stokes equations and parabolic systems, PacificJ. Math.
72 (1977) 557-569.
[2] Semilinear elliptic and parabolic equations of arbitrary order, Proc. Royal
Soc. Edinburgh A 78 (1978) 193-207.
J. A. Walker
[I] Dynamical systems and evolution equations, Plenum Press, New York (\980).
J. Watanabe
[I] On some properties of fractional powers of linear operators, Proc. Japan
Acad. 37 (1961) 273-275.
M. Watanabe
[I] On the differentiability of semi-groups of linear operators in locally convex
spaces, Sc. Rep. Niigata Univ. 9 (1972) 23-34.
G. Webb
[I] Continuous nonlinear perturbations of linear accretive operators in Banach
spaces, J. Func. Anal. 10 (1972) 191-203.
R. Wiist
(I] Generalization of Rellich's theorem of perturbations of (essentially) self
adjoint operators, Math. Z. 119 (\971) 276-280.
J. Yorke
[I] A continuous differential equations in Hilbert space without existence,
Funkcialaj Ekvacioj 12 (1970) 19-21.
K. Yosida
[I] On the group embedded in the metrical complete ring, Japan J. Math. 13
(1936) 61-80.
[2] On the differentiability and representation of one parameter semi-groups of
linear operators, J. Math. Soc. Japan 1 (1948) 15-21.
[3] Lectures on semi-group theory and its applications to Cauchy's problem in
partial differential equations, Tata Inst. Fund. Research (\ 957).
[4] On the differentiability of semi-groups of linear operators, Proc. Japan Acad.
34 (\ 958) 337-340.
[5] Fractional powers of infinitesimal generators and the analyticity of the
semi-groups generated by them, Proc. Japan Acad. 36 (1960) 86-89.
[6] Time dependent evolution equations in locally convex space, Math. Ann. 162
(1965) 83-86.
[7] Functional analysis (6th edition) Springer Verlag (1980).
J. Zabczyk
[I] A note on Co semigroups, Bull. A cad. Polan. Sc. (Math. Astr. Phys.) 23
(1975) 895-898.
S. D. Zaidman
[I] Abstract differential equations, Pitman Research Notes in Math. 36 (1979).
Index
A-admissible subspace 122 Decay of semigroups as t - 00 116

a-priori estimates 212 Decay of evolution systems as t - 00
Adjoint 173
operator 38 Differentiable semigroup 51
resolvent 38 Dissipative operator 13
semigroup 39 Distribution semigroup 252
strongly elliptic operator 212 Domain of infinitesimal generator
Alternating directions 93 1,5
Analytic semigroup 60 Dual semigroup 38
Duality set 13
Backward difference approximation

93 Embedding of semigroup in group 24
Equation
heat 98,234
Co-group 22 higher order 253
Co-semigroup 4 Korteweg-de Vries 247,252
Cauchy problem parabolic 225, 261
linear autonomous 100 ScrOdinger 238, 241
linear nonautonomous 105, 126, Equicontinuous semigroup 252
134 Evolution system 129
quasilinear 200 Exponential formula 21,32,33,92
semilinear 183
Compact semigroup 48
Contraction semigroup 8 Fractional powers of operator 72
Convergence of infinitesimal generators Fractional steps 93
88
Convergence of semigroups 85
Convergence of solutions as t - 00 Gagliardo-Nirenberg's inequality 242
119, 174 GArding's inequality 209
Graph norm 190 Operator

Group (see Co-group) 22 adjoint 38
dissipative 13
m-dissipative 81
Heat equation part of 39
linear 98 self adjoint 41
nonlinear 234 strongly elliptic 209
Hille-Yosida theorem 8 symmetric 41
Holder continuous function 112 unitary 41
Hyperbolic evolution system 135
Hyperbolic initial value problem 105,
134 Parabolic evolution system 150
Parabolic initial value problem 110,
167, 225
Implicit difference approximation 35 Part of operator in subspace 39
Inequality Perturbations
Gagliardo-Nirenberg's 242 by bounded operators 76
Glu"ding's 209 by Lipschitz operators 184
Hausdorf-Young's 224, 262 of generators of analytic semigroups
Kallman-Rota's 7, 253 80
Landau's 8 of generators of contraction
Initial value problem semigroups 81
linear autonomous 100 of stable families of generators 132
linear nonautonomous 105, 126, Positive semigroups 256
134 Post-Widder inversion formula 35
quasilinear 200 Pseudo resolvent 36
semilinear 183
higher order 253
non wen posed 253 Quasilinear initial value problem 200
Invariantsubspace 121
Infinitesimal generator I
r-convergence 86
Renorming 17, 18
Kallman-Rota's inequality 7,254 Representation of semigroups 90
Korteweg-de Vries equation 247 Resolvent
identity 20
operator 8
Landau's inequality 8 set 8
Laplace transform 25
Locally Lipschitz function 185
Lumer-Phillips theorem 14 ScrOdinger equation
linear 223
nonlinear 230
m-dissipative operator 81 Self adjoint operator 41
Mild solution 106, 146, 168, 184 Semilinear initial value problem 183
Semigroup
adjoint 39
Numerical range 12 analytic 60
Null space 36 bounded linear operators
Index 279
Co 4 Taylor's formula for continuous

compact 48 functions 33
contractions 8 Theorem
differentiable 51 Hille-Yosida 8, 19
distributions 252 Lumer-Phillips 14
equicontinuous 252 Sobolev 208, 222
nonlinear 253 Trotter-Kato 87
positive 256 Trotter-Kato theorem 87
strongly continuous (see Co) 4
strongly measurable 255
uniformly bounded 8
uniformly continuous 1 Uniformly bounded semigroups 8
weakly continuous 44, 255 Uniformly continuous semigroups I,
Sobolev's imbedding theorem 208, 25, 50
222 Uniqueness
Solution of initial value problem evolution systems 135, 145, 160
classical 100, 105, 139 infinitesimal generator 3, 6
generalized 36, 228 solution 101, 106
mild 106, 146, 168, 184 Y-valued solution 140, 146
strong 109 Unitary group 41
Y-valued 140 Unitary operator 41
weak 258
Spectrum
continuous 46 Weak
point 45 infinitesimal generator 42, 43
residual 46 solution 258
of Co semigroup 45 Weakly continuous semigroups 44,
of compact semigroup 51 255
of differentiable semigroup 53
Stable family of operators 130, 200
Strong solution 109
Strongly elliptic operator 209 Y-valued solution 140
Strongly measurable semigroup 255 Yosida approximation 9
(continued from page ii)
61. SattingerlWeaver: Lie Groups and Algebras with 89. O'Malley: Singular Perturbation Methods for
Applications to Physics, Geometry, and Ordinary Differential Equations.
Mechanics. 90. Meyer/Hall: Introduction to Hamiltonian
62. LaSalle: The Stability and Control of Discrete Dynamical Systems and the N-body Problem.
Processes. 91. Straughan: The Energy Method, Stability, and
63. Grasman: Asymptotic Methods of Relaxation Nonlinear Convection.
Oscillations and Applications. 92. Naber: The Geometry of Minkowski Spacetime.
64. Hsu: Cell-to-Cell Mapping: A Method of Global 93. Colton/Kress: Inverse Acoustic and
Analysis for Nonlinear Systems. Electromagnetic Scattering Theory.
65. Rand/Armbruster: Perturbation Methods, 94. Hoppensteadt: Analysis and Simulation of
Bifurcation Theory and Computer Algebra. Chaotic Systems.
66. HlavaceklHaslingerlNecasIlLov{sek: Solution of 95. Hackbusch: Iterative Solution of Large Sparse
Variational Inequalities in Mechanics. Systems of Equations.
67. Cercignani: The Boltzmann Equation and Its 96. Marchioro/Pulvirenti: Mathematical Theory of
Applications. Incompressible Nonviscous Fluids.
68. Temam: Infinite Dimensional Dynamical 97. Lasota/Mackey: Chaos, Fractals, and Noise:
Systems in Mechauics and Physics. Stochastic Aspects of Dynamics, 2nd ed.
69. Golubitsky/StewartlSchaeffer: Singularities and 98. de Boor/Hol/igiRiemenschneider: Box Splines.
Groups in Bifurcation Theory, Vol. II. 99. Hale/Lunel: Introduction to Functional
70. Constantin/FoiasINicolaenkolI'emam: Integral Differential Equations.
Manifolds and Inertial Manifolds for Dissipative 100. Sirovich (ed): Trends and Perspectives in
Partial Differential Equations. Applied Mathematics.
71. Catlin: Estimation, Control, and the Discrete 101. NusseIYorke: Dynamics: Numerical
Kalman Filter. Explorations.
72. LachaklMeunier: Multiphase Averaging for 102. Chossat//ooss: The Couette-Taylor Problem.
Classical Systems. 103. Chorin: Vorticity and Turbulence.
73. Wiggins: Global Bifurcations and Chaos. 104. Farkas: Periodic Motions.
74. MawhinIWil/em: Critical Point Theory and 105. Wiggins: Normally Hyperbolic Invariant
Hamiltonian Systems. Mauifolds in Dynamical Systems.
75. Abraham/Marsden/Ratiu: Manifolds, Tensor 106. CercignanillllnerlPulvirenti: The Mathematical
Analysis, and Applications, 2nd ed. Theory of Dilute Gases.
76. Lagerstrom: Matched Asymptotic Expansions: 107. Antman: Nonlinear Problems of Elasticity.
Ideas and Techniques. 108. Zeidler: Applied Functional Analysis:
77. Aldous: Probability Approximations via the Applications to Mathematical Physics.
Poisson Clumping Heuristic. 109. Zeidler: Applied Functional Analysis: Main
78. Dacorogna: Direct Methods in the Calculus of Principles and Their Applications.
Variations. 110. Diekmannlvan Gi/s/Verduyn LunellWalther:
79. Hernandez-Lerma: Adaptive Markov Processes. Delay Equations: Functional-, Complex-, and
80. Lawden: Elliptic Functions and Applications. Nonlinear Analysis.
81. BlumanIKumei: Symmetries and Differential 111. Visintin: Differential Models of Hysteresis.
Equations. 112. Kuznetsov: Elements of Applied Bifurcation
82. Kress: Linear Integral Equations. Theory.
83. BebernesiEberly: Mathematical Problems from 113. Hislop/Sigal: Introduction to Spectral Theory:
Combustion Theory. With Applications to SchrMinger Operators.
84. Joseph: Fluid Dynamics of Viscoelastic Fluids. 114. Kevorkian/Cole: Multiple Scale and Singular
85. Yang: Wave Packets and Their Bifurcations in Perturbation Methods.
Geophysical Fluid Dynamics. lIS. Taylor: Partial Differential Equations I, Basic
86. DendrinosiSonis: Chaos and Socio-Spatial Theory.
Dynamics. 116. Taylor: Partial Differential Equations II,
87. Weder: Spectral and Scattering Theory for Wave Qualitative Studies of Linear Equations.
Propagation in Perturbed Stratified Media. 117. Taylor: Partial Differential Equations III,
88. BogaevskilPovzner: Algebraic Methods in Nonlinear Equations.
Nonlinear Perturbation Theory.

Applied Mathematical Sciences: Springer

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Applied Mathematical Sciences: Springer

Uploaded by

Copyright:

Available Formats

Applied Mathematical Sciences

Semigroups of Linear Operators

Mathematics Subject Classification: H7D05, 35FlO, 35F25, 35G25

Library of Congress Cataloging in Publication Data

Printed on acid-free paper.

Production coordinated by Brian Howe and managed by Francine Sikorski; manufacturing

ISBN-13: 978-1-4612-5563-5 e-ISBN-13: 978-1-4612-5561-1

The aim of this book is to give a simple and self-contained presentation of

Preface to the Second Printing v

3.5 A General Representation Theorem 89

Bibliographical Notes and Remarks 252

Generation and Representation

1.1. Uniformly Continuous Semigroups of Bounded

The linear operator A defined by

D(A) = {x E X: lim T(t)x - x eXists} (1.2)

is the infinitesimal generator of the semigroup T(t), D(A) is the domain

Theorem 1.2. A linear operator A is the infinitesimal generator of a uniformly

which imply that T(t) is a uniformly continuous sernigroup of bounded

h - I (T( h) - I) = ( h- I ~P + hT{ s ) ds - h - 110h T( s ) ds )( foP T{ s ) ds ) - I

From Definition 1.1 it is clear tQat a sernigroup T( t) has a unique

Theorem 1.3. Let T(t) and S(t) be uniformly continuous semigroups of

lim T( t) - I = A = lim S( t) - I ( 1.7)

then T(t) = S(t) for t ~ O.

IIT(t) - S(t)1I =IIT( n~) - s( n~ )11

~ :~JT( (n - k- 1)~ )IIIIT( ~) - s( ~ )lllls( ~ )11 ~ Cn ;C ~~ f.

Corollary 1.4. Let T( t) be a uniformly continuous semigroup of bounded linear

dT(t) = AT(t) = T(t)A ( 1.9)

1.2. Strongly Continuous Semigroups of Bounded

A strongly continuous semigroup of bounded linear operators on X will be

Theorem 2.2. Let T(t) be a Co semigroup. There exist constants w ~ 0 and

IIT(t}1I =::; Me wt for 0 =::; t < 00. (2.2).

Corollary 2.3. If T(t) is a Co semigroup then for every x EX, t -. T(t)x is a

PROOF. Let t, h ~ O. The continuity of t -. T(t)x follows from

Theorem 2.4. Let T( t) be a Co semigroup and let A be its infinitesimal

b) For x E X, {T(S)X ds E D(A) and

A(foIT{S)Xds) = T{t)x - x. (2.4)

c) For x E D(A), T(t)x E D(A) and

T{t)x - T{s)x = [T{T)AxdT = [AT(T)xdT. (2.6)

!i~ [ T( t ) x - ~(t - h) x _ T( t ) AX]

= limT(t - h)[ T(h); - x - AX] + lim (T(t - h)Ax - T(t)Ax),

Corollary 2.5. If A is the infinitesimal generator of a Co semigroup T(t) then

PROOF. For every x E X set XI = ljtf:T(s)x ds. By part (b) of Theorem

Theorem 2.6. Let T( t) and S( t) be Co semigroups of bounded linear operators

= -T(t - s)AS(s)x + T(t - s)BS(s)x = O.

If A is the infinitesimal generator of a Co semigroup then by Corollary

Theorem 2.7. Let A be the infinitesimal generator of the Co semigroup T(t). If

The integrand on the right-hand side of (2.11) converges as h ~ 0 to

Ay = lim T(hl-I y = -l°Ocpl(s)T(s)xds.

for every x E X, cp E Gj). This implies that for x E X the continuous

We conclude this section with a simple application of Theorem 2.4.

Lemma 2.8. Let A be the infinitesimal generator of a Co semigroup T( t )satis-

::s; 2~ Ilxll + ~t IIA 2xll. (2.14)

Here we used that M ~ 1 (since II T(O) II = 1). If A 2x = 0 then (2.14) im-

EXAMPLE 2.9. Let X be the Banach space of bounded uniformly continuous

It is easy to check that T(t) is a Co semigroup satisfying II T(t)11 :s; I for

(supl/'(s)1)2 :s; 4(suplf"(s)I)(suplf(s)l) (2.15)

1.3. The Hille-Yo sid a Theorem

Theorem 3.1 (Hille-Yosida). A linear (unbounded) operator A is the infini-

PROOF OF THEOREM 3.1 (Necessity). If A is the infinitesimal generator of a

= A (1000 e-A1T( t)x dt) = AR( A)x. (3.6)

PROOF. Suppose first that x E D( A). Then

But D(A) is dense in X and IIAR(A: A)II :S 1. Therefore AR(A: A)x -+ x

We now define, for every A > 0, the Yosida approximation of A by

0< d(A :S(A») ~ IA - (x, Ax)1 = I(x, AX - Ax)1 ~ IIAX - Axil