Distributions, Integral T r a n s f o r m s

a n d Applications
 A N A L Y T I C A L M E T H O D S A N D S P E C I A L F U N C T I O N S

An International Series of Monographs in Mathematics

F O U N D I N G EDITOR: A.P. Prudnikov (Russia)

S E R I E S E D I T O R S : C F . D u n k l (USA), H . - J . Glaeske (Germany) and M . Saigo (Japan)

Volume 1
Series of Faber Polynomials
RK. Suetin

Volume 2
I n v e r s e S p e c t r a l P r o b l e m s for D i f f e r e n t i a l O p e r a t o r s a n d T h e i r A p p l i c a t i o n s
V.A. Yurko

Volume 3
Orthogonal Polynomials in Two Variables
P.K. Suetin

Volume 4
Fourier Transforms and Their Approximations
A.M. Sedletskii

Volume 5
Hyper singular Integrals and Their Applications
S. G. Samko

Volume 6
M e t h o d s of the Theory of Generalized Functions
V.S. Vladimirov

Volume 7
Distributions, Integral Transforms and Applications
W. Kierat and U. Sztaba

T h i s book is p a r t of a series. T h e publisher w i l l accept continuation orders which may be

c a n c e l l e d at a n y t i m e a n d w h i c h p r o v i d e for a u t o m a t i c b i l l i n g a n d s h i p p i n g of e a c h title i n the
series u p o n p u b l i c a t i o n . P l e a s e w r i t e for details.
D i s t r i b u t i o n s , I n t e g r a l T r a n s f o r m s

a n d A p p l i c a t i o n s

Wladyslaw Kierat
a n d

Urszula Sztaba

T a y l o r &. F r a n c i s

Taylor &. Francis G r o u p

L O N D O N A N D NEW YORK
Published i n 2003 b y
C R C Press
Taylor & Francis Group
6000 B r o k e n Sound Parkway N W , Suite 300
B o c a Raton, F L 33487-2742

© 2003 by Taylor & Francis Group, L L C

C R C Press is an imprint of Taylor & Francis Group

N o c l a i m to original U . S . Government works

Printed i n the United States of A m e r i c a on acid-free paper
10 9 8 7 6 5 4 3 2

International Standard B o o k Number-10: 0-415-26958-X (Hardcover)

International Standard B o o k Number-13: 978-0-415-26958-2 (Hardcover)

This book contains information obtained from authentic and highly regarded sources. Reprinted material is
quoted with permission, and sources are indicated. A wide variety of references are listed. Reasonable efforts
have been made to publish reliable data and information, but the author and the publisher cannot assume
responsibility for the validity of a l l materials or for the consequences of their use.

N o part of this book may be reprinted, reproduced, transmitted, or utilized i n any form by any electronic,
mechanical, or other means, now k n o w n or hereafter invented, including photocopying, microfilming, and
recording, or in any information storage or retrieval system, without written permission from the publishers.

Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only
for identification and explanation without intent to infringe.

Library of Congress Cataloging-in-Publication Data

Catalog record is available from the Library of Congress

Visit the Taylor & Francis Web site at

T & F i n f o r m a http://www.taylorandfrancis.com

&
T a y l o r F r a n c i s G r o u p
and the C R C Press Web site at
is the Academic Division of T & F Informa plc. http://www^rcpressxom
 CONTENTS

Preface vii

Chapter 1. Definitions and preliminaries 1

1.1. The spaces V and V 1
1.2. Approximation lemmas 3
1.3. Regularizations of functions 5
1.4. D u Bois-Rymond's lemma 6
1.5. Some density theorem 7
1.6. Distributional convergence 8
1.7. Algebraic operations on distributions 9
1.8. Linear transformations i n the space of the independent variables 10
1.9. Differentiation of distributions 11
1.10. W e a k derivatives of locally integrable functions 12
1.11. L o c a l Sobolev spaces 16
1.12. Sobolev spaces 19
1.13. Differential equations of the second order with measures
as coefficients 21

Chapter 2. Local properties of distributions 27

2.1. S m o o t h partitions of unity 27
2.2. A p p r o x i m a t i o n of functions belonging to W m
' p
( í í ) by smooth
functions 29
2.3. Restrictions of distributions 31
2.4. Support of distributions 33
2.5. Distributions of finite order 35
2.6. Cartesian products of B a n a c h spaces 37
2.7. Some local representations of distributions 38

Chapter 3. Tensor products and convolution products 41

3.1. Regularization of distributions 41
3.2. A characterization of convolution operators 44
3.3. Tensor product of distributions 45
3.4. Differentiation and support of tensor product 47
3.5. The theorem of kernels 48
3.6. Connection between tensor product and convolution product
of distributions 54
3.7. Differentiation and support of convolution product 58

Chapter 4. Differential equations 59

4.1. Fundamental solutions of differential equations 59
vi CONTENTS

4.2. The C a u c h y p r o b l e m for the wave equation with distribution

data 60
4.3. Fundamental solutions of the Laplace operator and
the heat operator 63
4.4. The Hormander inequalities 66
4.5. L -solvability
2
70
4.6. Regularity properties of differential operators 72

Chapter 5. Particular types of distributions and Cauchy transforms 75

5.1. Integrable distributions 75
5.2. R e g u l a r iz at i o n of integrable distributions 77
5.3. Tensor product of integrable distributions 78
5.4. T>LP a n d V ' L P spaces 81
5.5. Convolution product 82
5.6. Cauchy transforms of integrable distributions 84
5.7. Cauchy transforms of distributions belonging to V LP 87
5.8. Cauchy transforms of some distributions 91

Chapter 6. Tempered distributions and Fourier transforms 95

6.1. T h e spaces S a n d S' 95
6.2. Tensor product of tempered distributions 97
6.3. Fourier transforms of integrable functions 98
6.4. F o r m a l properties of Fourier transforms 100
6.5. Fourier transforms of functions i n S 102
6.6. Fourier transforms of functions in L 2
( R n
) 103
6.7. Fourier transforms of the Hermite functions 107
6.8. Fourier transforms of tempered distributions 108
6.9. F o r m a l properties of Fourier transforms of tempered distributions 109
6.10. Fourier transforms of integrable distributions 110
6.11. Fourier transforms of square integrable distributions 112
6.12. Determining Fourier transforms of square integrable distributions 114
6.13. Hubert transforms 116
6.14. C a r l e m a n transforms 117
6.15. The Paleya-Wiener type theorems 119
6.16. The Cauchy semigroup 121
6.17. The C a u c h y p r o b l e m for the heat e q u a t i o n 123

Chapter 7. Orthogonal expansions of distributions 127

7.1. The Poisson summation formula 127
7.2. Periodic distributions 128
7.3. T h e spaces A a n d A! 131
7.4. C a u c h y transforms of elements o f A' 134

7.5. T h e Wiener expansion of square integrable distributions 136

A P P E N D I X . Sequential completeness of some spaces 139

Subject index 143

N o t e s a n d references to the literature 145

Bibliography 147

Index of symbols 149

 PREFACE

This book has been w r i t t e n as an approachable introduction to the theory of dis­

tributions and integral transforms. The principal intention of our book is t o em­
phasize the remarkable connections between distributional theory, classical analysis
and theory of differential equations. W e use the theory of the Lebesgue integral as
a fundamental tool i n the proofs of m a n y theorems. The theory is d e v e l o p e d from
its b e g i n n i n g to the point where many of the fundamental and deep theorems are
proved, for example the Schwartz kernel theorem and the Malgrange-Ehrenpreis
theorem. W e particularly want to show some practical means of using the theory of
distributions i n the cases (for e x a m p l e F o u r i e r a n a l y s i s ) , w h e n m e t h o d s o f c l a s s i c a l
analysis are insufficient. This book naturally covers general theory, examples and
applications. W e also try to answer the m a i n questions related to the topics pre­
sented i n the text. T h e material presented i n our b o o k will be understandable to a
student w h o has completed a course of advanced calculus. T h e b o o k has been made
self-contained b y inclusion of the necessary facts f r o m f u n c t i o n a l analysis i n the A p ­
pendix. T h e r e are elementary approaches to the theory of distributions considering
distributions as limits of sequences of functions. The theoretical foundations in
these approaches are usually presented only i n a simplified manner insufficient for
applications. I n t h i s v o l u m e w e g i v e S c h w a r t z ' s d e s c r i p t i o n o f d i s t r i b u t i o n s as linear
continuous forms o n topological vector spaces. This presentation of distributions is
flexible in applications in various branches of mathematics.
T h i s b o o k is d i v i d e d i n t o s e v e n chapters.
In Chapter 1 we define distributions and introduce some fundamental notions
as r e g u l a r d i s t r i b u t i o n s , w e a k d e r i v a t i v e s o f l o c a l l y i n t e g r a b l e f u n c t i o n s , derivatives
of d i s t r i b u t i o n s a n d the convergence of distributions.
Chapter 2 treats the question of local properties of distributions.
Chapter 3 is d e v o t e d to tensor products and convolution products of distribu­
tions.
In Chapter 4 we give some applications of the theory of distributions to differ­
ential equations.
In the next two chapters we deal with integral transforms s u c h as the Cauchy
transforms, Fourier transforms and Hubert transforms.
In the last chapter we present some aspects of the theory of orthogonal expan­
sions of some classes of distributions.
The basic definitions and theorems of measure theory are taken from Rudin's
book: Real and Complex Analysis ([23]). Theorems, lemmas, definitions a n d for­
m u l a s are n u m b e r e d separately i n each section, for e x a m p l e T h e o r e m 2 i n S e c t i o n 10
of C h a p t e r 1 will be denoted by Theorem 1.10.2.

Wladyslaw Kierat and Urszula Sztaba

 C H A P T E R 1

DEFINITIONS A N D PRELIMINARIES

1.1. T h e spaces V and V

We will use the following s y m b o l s :

N n
— set o f a l l n o n - n e g a t i v e i n t e g e r p o i n t s o f K n
,
Ci — ( 0 , . . . , 0 , 1 , 0 , . . . , 0 ) , , w h e r e n u m b e r 1 is o n i - t h p o s i t i o n ,
a < (3, a, (3 eW 1
means a ¿ < 0i for i = 1,..., n,
a < (3 m e a n s a < f3 a n d OLÍ < fa for s o m e i = 1,..., n,

|a| ai + a
= h a, 2 n

a\ = ai\a2^ • • • a l , n
(Oi\ Oil

s u p p ip = {x : <p(x) ^ 0} .
Let fl be an o p e n set in R n
. W e will denote b y V(Q) the set of all infinitely
differentiable functions w i t h c o m p a c t supports contained i n O . These functions will
be called smooth. If Ü = R n
we will write V instead of V(W ). l
(In another cases,
when functions constituting a functional space are d e f i n e d o n w h o l e lZ n
1 then we
will proceed analogously.) O b v i o u s l y , V(Q) is a v e c t o r space.

EXAMPLE 1 . 1 . 1 . T h e following function

- i
for \x\ < 1, |x| = (xj -j h x 2
) 2

for \x\ > 1

belongs to V.

We introduce a convergence of sequences in V(Q).

DEFINITION 1.1.1. W e say that a sequence (<¿v), yv G V(Q) converges to

if e V(Q) iff

(i) t h e r e e x i s t s a c o m p a c t set K C ft such that supp p v C K for each i / G N

and

(ii) the sequence d (pa

u converges to d (p
a
uniformly on K for e a c h a G N . n

Of course, the sequence (<¿v) c o n v e r g e s t o (p i n V(Q) iff the sequence (ipu — (p)
converges to zero i n V(Q).

DEFINITION 1 . 1 . 2 . A linear form defined o n V(il) continuous w i t h respect to

the a b o v e c o n v e r g e n c e is s a i d t o b e a d i s t r i b u t i o n .

The set o f d i s t r i b u t i o n s w i l l b e d e n o t e d b y V(Çl).

 1. DEFINITIONS AND PRELIMINARIES
2

THEOREM 1 . 1 . 1 . A linear form A defined on D ( 0 ) is continuous iff for every

compact set K C O there exist a constant C > 0 and an index m such that

|A(v?)| <C|M| ,K, M (1.1.1)

w/iere | M U , x = m a x | a | < m s u p ^ ^ ip{x) and s u p p (p C K.

PROOF. T h e sufficiency is evident. W e shall n o w prove the necessity. Assume

that A G V(Q) and (1.1.1) d o e s n o t h o l d for s o m e c o m p a c t set K C ft. Therefore

for each v G N there exists a function p v G V(fl) such that s u p p <p u C K and

|A((^ )|
/ y > v\\ipv\\v,K- Obviously, one can choose (p u such that A((p„) = 1 for each

v G N. F r o m this we obtain

1 = HVv) > v\\<Pv\\v,K,

consequently < ¿. Hence

1 1

V + /i V

From this it follows that H ^ v ^ H ^ / r 0 as z/ —> o o for fixed fi. Therefore (pu

c o n v e r g e s t o z e r o as v —» c o i n V{ft). This contradicts the continuity of A . Thus,

t h e p r o o f is a c h i e v e d . •

EXAMPLE 1 . 1 . 2 . Let / G LJoc. W e take

A / M : = j f<P>

Of course, \Af(p)\ < H/HL^K)IMIO,K" if suppy> C K. Therefore Af is i n V. The

d i s t r i b u t i o n A / is s a i d t o b e regular.

EXAMPLE 1 . 1 . 3 . Put ô((p) := <p(0) for (p G V. It is v e r y e a s i l y seen t h a t (1.1.1)

holds. W e shall now show that ô is n o t a regular distribution. Indeed, let (p be

as i n E x a m p l e 1.1.1. Put p (x)

u := p{yx). Obviously, <¿v(0) = e" . 1
Moreover, for

each x / 0, l i m ^ - ^ o o <Pv{x) = 0. Assume that S(<p) = J Rn ftp for s o m e / G L]oc and

p e V. Obviously, e"1
= S(p ) u = f Rn f<p . u Note that f Rn f<p v tends to zero as

v —» o o . T h i s c o n t r a d i c t s o u r a s s u m p t i o n . T h e r e f o r e S is n o t a r e g u l a r d i s t r i b u t i o n .

The d i s t r i b u t i o n S is c a l l e d D i r a c m e a s u r e concentrated i n the zero point.

EXAMPLE 1 . 1 . 4 . Let p be in V and suppcp C [ - a , a]. Put

A(y>) = TW) := lim Í ^ d * -

|*|>e

Let us take a n e w function

t m ^ m t o r l # 0 ,
\ip'(0) fora; = 0.
 1.2. APPROXIMATION LEMMAS 3

Clearly, ip is c o n t i n u o u s a n d (p(x) = ip(Q) + xip(x) for x G IL O f course, ip(x)

ip'(Q x • x), 0 < <d x < 1 for x ^ 0. This implies that

0 ) | < max |</0)| < |b||l,[-a,a]

xG[-a,aj

for x G l Note that

/
|rr|>e e<|x|<a
/ da;

e<|x|<a
/ «
B y the Lebesgue dominated convergence theorem we have

A(y>) = l i m j ^ - d x = J r/>(x)dx.
\x\>e \x\<a

Hence we obtain the following inequality

\A((p)\ < 2a||<p||i _ j [ a j a ]

i f s u p p if C [—a, a]. T h u s , w e h a v e s h o w n t h a t A is a d i s t r i b u t i o n . T h e d i s t r i b u t i o n

is c a l l e d t h e Cauchy finite part of the integral f R -pj.

1.2. Approximation lemmas

Suppose that ip G L 1
and J Rn ip = 1. Let ip (x)
e = e~ i/j ( f ) , n
e > 0 and / G L,
p

1 < p < oo. Put

(f*^e)(y) := J f(x - y)ip {x)dx. e

We prove the following

LEMMA 1.2.1. If f e L , p
1 < p < oo, then f * ip -ï e f as e -> 0 in L , p
if
f G L°° and f is uniformly continuous on an open set V C W , 1
then f * tp e almost
uniformly converges to f as e —> 0 on V.

PROOF. W e n o w recall Minkowski's inequality

j (j\F(x,y)\dy) dx < J I J' \F(x,y)f dx) dy, (1.2.1)

LG \n n \G

where i î c M n
, C c M m
and 1 < p < oo. It is e a s y t o c h e c k t h a t f Rn ip (x) dx — 1
e

for a l l e > 0. Hence

(/ * - /(*) = / (/(* - !/) - / ( * ) ) & ( î > ) dj/

= J (f(x-ey)-f(x))ip(y)dy.
 1. DEFINITIONS AND PRELIMINARIES
4

B e c a u s e o f (1.2.1) we have

{f(x - ey) - f(x))ip(y) dy dx

< I J I j\(f(x - ey) - f(x))tl>(y)\dy \ dx

- / (/ l f { x
~ ~~ ^ ^ )
ey) f p pdx dy

= / Mv)\ I / \f(x-ey)-f(x)\»dx dy

=J Mv)\\\f(--ey)-f(-)h*dy.

Note that ||/(-- ey) - f(-)\\ „L < 2\\f\\ , L a n d ||/(- - ey) - f(-)\\ LP -> 0 as e 0

for e a c h y G W. 1
Hence, b y the Lebesgue dominated convergence theorem we infer

that / * ip e / as e - » 0 i n L P
. Thus, the proof of the first part of our theorem is

finished. O n the other hand suppose that / G L°° and / is u n i f o r m l y continuous

on an open s e t V. Having given ô > 0, choose a compact set W C W'

1
so that

f ^ Rn w < à. T h e n we have

SUp \(f*^ ){x) e ~ f(x)\

x£V 0

< sup / \ f(x - ey) - f(x)\\i/j(y)\dy + sup / - ey) - f(x)\\^(y)\dy

xeV J 0 xEVo J
W R -W n

<rjj \il>{y)\dy + 26\\f\\ -L

for VQ (E V , where this symbol means that the closure VQ o f V b i s c o m p a c t a n d

V b C V. N o t e t h a t 77 i s s m a l l f o r s m a l l e. T h i s finishes the proof. •

REMARK 1 . 2 . 1 . I n the preceding proof, the appeal to the Lebesgue dominated

convergence theorem m a y seem to be illegitimate since t h e Lebesgue dominated

convergence t h e o r e m deals only w i t h countable sequences of functions. However, it

does enable us to conclude that

/ ^(y)\\\f(--e,y)-f(-)\\ pdy^0 L

as e v —> 0 f o r e v e r y s e q u e n c e (e ) u w h i c h converges to zero a n d it says exactly that

my)\\\f(-~ey)-f(-)\\ dy^0 LP
 1.3. REGULARISATIONS OF FUNCTIONS
5

as e - > 0. W e shall encounter similar situations again and shall apply convergence

theorems to t h e m without further comment.

COROLLARY 1 . 2 . 1 . If ^ e L, 1
J Rn ip = 1 and ip (x)
€ = e' tpn
(f ) ; then
¡ R n i¡j (p -» <p(0) as e
e 0 for ip eV.

PROOF. P u t <f>{x) := ip(-x). Then

Jip <p = j
e Cp{-x)i¡) (x) e dx = (tp* ip )(0) - > 0 ( 0 ) = p ( 0 )
e

a s e —> 0 . T h i s s t a t e m e n t finishes the proof. •

A v e r y s i m i l a r l e m m a t o L e m m a 1.2.1 is t r u e for a r b i t r a r y o p e n sets i n W. 1

W e
shall now introduce some notations. Let

B (Q)r = {x : d i s t ( x , Ü) < r},

in particular, B (0)
r = B ({0})
r — {x : \x\ < r}. L e t Î/J b e i n L,
1
supp^ C #i(0),
/ R n V = 1 and / G L (ft). p
P u t by definition

We shall now show that f e G and supp / e C j B ( 0 ) for e <

r r. Indeed, let us

extend / on W 1
putting f(x) := f(x) for x G O and f(x) = 0 foi x ^ Ü. Take by

definition ip (x) = €~ ip ( f
e
n
). T h e n we have f e = ipe* f • B y the Y o u n g inequality

(see (5.5.1)) w e infer t h a t fe G L (R ).

p n
Therefore f e G L (ü).
p
It is easy t o verify

t h a t fe{x) — 0 for x £ J5 (0). R B y v i r t u e o f L e m m a 1.2.1 we get

II/€-/||LP(R«) ->O

a s e —>• 0 . F r o m t h i s , w e i n f e r t h a t ||/ e — /||LP(0) —>- 0 a s e — » 0 .

O u r consideration leads to the following

LEMMA 1 . 2 . 2 . If f e L (Q), P
ip G L , 1
supp^ C B^O), J ip = 1 and

fe(x) = e- j^^—l^f{y)dy
n

then f e G L (ü),
p
fe tends to f as e 0 in L (Q) P
and supp fe C Br(il) for e < r.

1.3. Regularizations of functions

Let / be in L\ oc and i\) b e in V.

DEFINITION 1 . 3 . 1 . The function = ip * / is s a i d t o b e t h e r e g u l a r i z a t i o n o f

/ b
y

THEOREM 1 . 3 . 1 . If f G LJ oc andiP eV then f^ is in C°° and d f^ = a

d° ip*f
i

for a G N . Moreover, n
if f G IP, 1 < p < oo, f Rn ip(x)dx = 1 and i¡) (x) e =
e~ ip ( ^ ) then f^
n
e tends to f as e 0 in IP.
 1. DEFINITIONS AND PRELIMINARIES
6

PROOF. Since (/ * ip)(x) = f Rn ip(x — y)f(y) dy therefore

U(x + 0 - U(x) = J(t/>(x + Í - y) - ^{x - y))f(y) dy.

MP

N o t e t h a t t h e s u p p o r t s oí ip(x + t; —-)— ip(x — •) a r e c o n t a i n e d i n a c o m p a c t s e t when

x is fixed and ]£| < 5. Since / is i n L} , o c we conclude by the Lebesgue dominated

convergence theorem that f^(x + £) — fip(x) t e n d s t o z e r o as £ - » 0. Therefore

is continuous. O f course,

—tp(x + 6/ie ? ; - y)f(y) dy,

MP

where e¿ = ( 0 , . . . 0,1, 0 , . . . 0). Repeating the previous consideration we infer that

(f\ b(x
f + hei) — f^{x))h~ 1
tends to {j^r^ * f^j (x) as h 0. N o w we proceed by

induction. T h e second part o f t h e l e m m a is a c o n s e q u e n c e o f L e m m a 1.2.1. •

DEFINITION 1 . 3 . 2 . W e say that the function <p with a compact support belongs

to C (ft)
c i f it is c o n t i n u o u s a n d s u p p <p C ft.

THEOREM 1 . 3 . 2 . Let ip £ V, supp V c #i(0) and J Rn ip = 1. / / / ¿5 i n C (ft),

c

then there exists eo > 0 such that fy e G V(ft) for e < eo a n d f^ e tends to f
uniformly in ft.

PROOF. W i t h o u t loss of generality w e assume that / G C ( R c

n
) . Namely, one

can take f(x) = 0 for x ^ supp/. Let x £ £? (supp/), e then \x — y\ > e for each y

in supp / and tp (x — y)
e = 0- H e n c e w e h a v e * f)(x) — 0. This implies that

supp U e C £ (supp/).
e

Therefore i f eo < dist(supp /, 9 0 ) then G P(iî) for e < eo- In view of

L e m m a 1.2.1, f^ e converges u n i f o r m l y to / i n ft. This finishes the proof. •

1.4. D u Bois-Rymond's l e m m a

We k n o w that every locally integrable function / determines a regular distribution

Aj b y e q u a l i t y A(ip) = J f(p (see S e c t i o n 1.1). W e shall n o w show that the mapping
/ —)- Af is a n injection from L} {ft)
oc into V'{ft). This fact will be an immediate
consequence of the following

LEMMA 1 . 4 . 1 (du Bois-Rymond). / / / is in L} (ft) oc and f Q f(p = 0 for ip G V(fL)

then f(x) = 0 almost everywhere in ft.

PROOF. L e t K <s ft. P u t

if f(x) ¿ 0
sgn/(x) :=
if / ( * ) = 0.

Note that \f(x)\ = \f \f

f f
(x = f (x) sgn f (x). Let K <s ft. Denote by XK the

c h a r a c t e r i s t i c f u n c t i o n o f K. F u r t h e r , p u t g(x) := XK sgn/(x). Obviously, |#(x)| <

 1.5. SOME DENSITY THEOREM 7

1 for x G fl. Because of Lusin's l e m m a (see [23]) for a r b i t r a r y S > 0 there exists

ip G C (ft) c such that \ip(x)\ < 1 for x G ft a n d ¿¿(.A) < £, where

A = {x : x e ft, ip(x) ^ #(x)}.

O f course, the following equalities

y \f(x)\dx = JXKÍ(x)sgnf(x) dx
K Q

= J f(x)ip(x) dx + J f(x)(xK sgn/(x) - ip(x))dx

hold. Assuming that f f<p

n = 0 for </? G and taking into account Theo-

rem 1.3.2, w e i n f e r t h a t J fip

Q = 0 for ip G C (ft). c F r o m this it follows that

J \f(x)\dx = J f(x)(xK sgn/Or) - ip{x))dx,

K A

where ¡JL(A) < Ô. Since K <& ft and / G Lj (ft) oc therefore for every e > 0 one

can choose Ô > 0 so t h a t J A Í(X)(XK sgn/(#) - ip(x))dx < e if /x(A) < 5. This

statement finishes the proof. •

1.5. Some density theorem

THEOREM 1 . 5 . 1 . V(ft) is dense in L (ft) p

if 1 < p < oo.

PROOF. B e c a u s e of T h e o r e m 1.3.2 we only need to show that C (ft) c is dense

inL (ft). p
Let / be i n L (ft) a n d e
p
> 0. W e s h a l l s h o w t h a t t h e r e e x i s t s a function

p G C (ft) c such that \\f -(f\\ p(Q) <L e. S e t t i n g / = / x -f2 + i(h~h) where each/ i 5

j — 1, 2 , 3 , 4 i s r e a l v a l u e d a n d n o n n e g a t i v e , we can restrict ourselves without loss

of generality to p r o v i n g a r e d u c e d form of the theorem. Namely, we assume that

/ is i n L (ft)
p
a n d it takes nonnegative values. Choose a monotonically increasing

sequence (s ) u of nonnegative simple functions converging point wise to / i n ft. Since

0 < s {x)
u < f(x)

therefore su G L (ft)
p
and

(f(x) - a (x)r v < F{x).

F r o m this, b y the Lebesgue d o m i n a t e d convergence theorem it follows that s u tends

to / in L (ft).
p
Hence we have \\s u — / | | L P < § for v > VQ. Since su is a simple

function therefore the supp of s u must have finite volume. B y Lusin's lemma there

exists p G C (ft) c such that

\<p(x)\ < ||¿V || oo 0 L

and

vol({x :xeSl, s {x) ¿ ip(x)})

UQ < (\\s \\ oo)- y
l/0 L
i
.
 1. DEFINITIONS AND PRELIMINARIES
8

Hence we obtain

ll^o _
PWLP < \\sv 0 - (P\\L™(VOI{X : x eft, s (x)Vo ^ p(x)})p

< 2 | | S L / 0 | | ^ | ( | | S L / 0 | | ^ ) - 1
= |

F i n a l l y w e h a v e | | / — <P\\LP < e. Thus, the t h e o r e m is p r o v e d . •

1.6. Distributional convergence

Let A a , A b e i n V'{ft), a G M.

DEFINITION 1 . 6 . 1 . W e say that A a d i s t r i b u t i o n a l l y c o n v e r g e s t o A as a —> ŒQ

if l i m _ ^
a a o A ((p)
a = A(p) for e a c h p G V(ft).

In particular, if v G N then we take lim^oo A v = limi 0 A u = A if

lim^^oo A ((p) = A(ip)

u for e a c h ip G V(ft).

EXAMPLE 1 . 6 . 1 . The following sequences

\v*~ vXA
(Picard), (1.6.1)

v
(Stieltjes), (1.6.2)

1 V
(Cauchy), (1.6.3)
7T 1 + (vx) 2

Z = e ~ ^ 2
(Gauss) (1.6.4)

27T
distributionally c o n v e r g e t o S. S e e C o r o l l a r y 1.2.1.

We s h a l l n o w s h o w t h a t f o r e a c h (p i n V

lim f dx =-m8{ip) + ^{<p) (1.6.5)

and

lim / ^ ^ = ¿ 7 r ^ ) + -\(^), (1.6.6)

e->o+ J/ xx -—i e ie (•)

where (7y ( v ) 9
= l i m ^ o I\ \>e^x^
x ^x (see Example 1.1.4). The equalities (1.6.5)

and (1.6.6) are s a i d to be the Sochocki-Plemelj formulas.

Let p be in V and supp p C [—a, a ] . W e k n o w t h a t <¿? m a y b e w r i t t e n a s follows

p(x) — p(0) + xip(x), where ^ is a c o n t i n u o u s function. Moreover, ^y(<p) = i- ^
a

(see Example 1.1.4). H e n c e w e have

f ^ d x = [ A d x + f ^ m d x .
J X -\-16 J X + ie J X + l€
—a —a
 1.7. ALGEBRAIC OPERATIONS ON DISTRIBUTIONS 9

Note that

a a a

—a —a —a

= — 7TÍ(p(0)— j
7T J t 2
+ l

Finally we have

\[ m Í ^(^) ^ x _ _^^(O) = — 7ciS((p).

R
Of course,
a a

lim / d i = / dx.
e-+oJ x + ie J '
—a —a

This statement finishes t h e p r o o f o f (1.6.5). A n a l o g o u s l y o n e c a n p r o v e (1.6.6).

EXAMPLE 1 . 6 . 2 . A n easy computation shows that f R 2x (l2

+ x )~
2 2
dx — ir.
Take ip(x) = 2 K~ X (1/ 1 2
+ x )~2 2
and ip (x)
u — vi\)(yx). T h e sequence {ip )v converges

to ô (see C o r o l l a r y 1.2.1). O n the other hand ip (x)

v tends t o zero as v —>• o o f o r

each x G M. This example indicates that the pointwise convergence is n o t t h e

distributional convergence i n general.

EXAMPLE 1 . 6 . 3 . L e t f, v f be in Lj (fl)
oc a n d the sequence (f ) u converges to /

a s v —y o o i n t h e s e n s e o f Lj (Q).
oc It is easily seen that f n f (pu converges to f n ftp

a s v - » o o f o r e a c h Lp G V(Q). This means that the sequence (f )

u distributionally

converges to / .

1.7. Algebraic operations o n distributions

L e t S a n d T b e i n V'(Q).

DEFINITION 1 . 7 . 1 . T h e expression S +T defined i n the following w a y

(S + T)(<p):=S(<p)+T{<p), ^ e P ( f i )

is c a l l e d t h e s u m o f t h e d i s t r i b u t i o n s S a n d T.

Obviously, S+T is also a continuous linear f o r m o n U(Sl). Moreover, if S¡ a n d

T G are regular distributions corresponding to the functions / a n d g, t h e n Sf + T g

is a l s o a r e g u l a r d i s t r i b u t i o n a n d i t c o r r e s p o n d s t o t h e f u n c t i o n / + g.
T h e r e is n o t a n y n a t u r a l w a y t o define t h e p r o d u c t of two arbitrary distribu-
tions. Nevertheless, i t is possible t o define t h e p r o d u c t o f a n y d i s t r i b u t i o n A b y a n
i n f i n i t e l y d i f f e r e n t i a b l e f u n c t i o n g. Note that the product of a n infinitely differen-
tiable function g a n d a function p from V(ft) belongs to T>(Q). Moreover, if the
sequence (<¿v), Lp G V(ü) v converges to zero i n V(ft) then the sequence (gip )
u also
converges to zero i n V{Q).
Let A b e a d i s t r i b u t i o n . T h e n t h e m a p p i n g Lp - » A(gp) defines a n e w d i s t r i b u -
tion o n V(il).
10
1. DEFINITIONS AND PRELIMINARIES
DEFINITION 1 . 7 . 2 . The distribution p A(gip) is s a i d t o be the product of

the distribution A a n d the function g.

For the regular distribution A / corresponding to the locally integrable function

/, multiplication by g corresponds to multiplication of / by g i n the usual sense.

1.8. Linear transformations in the space of the independent variables

Let i be a linear nonsingular transformation in R n

and A ' 1
be its inverse. As-
suming / to be locally integrable and l e t t i n g ip b e a function in D , by changing
variables i n the integral fRn f(A~ x)p(x) 1
dx we obtain

J fiA-'xMx) dx = \A\ f f(yMAy) dy, (1.8.1)

R" R"

where y = A~ x x
and |^4| i s t h e absolute value of the determinant of the matrix of
the transformation A. E q u a l i t y (1.8.1) c a n b e w r i t t e n as follows

A -,{p)
foA = \A\A {p>oA),
f (1.8.2)

where Aj -i oA and A / are the regular distributions corresponding to the locally

integrable functions f o A~ l
and /, respectively. Taking into account (1.8.2) one
can define the linear t r a n s f o r m a t i o n A ~ x
for a n y d i s t r i b u t i o n A G V i n the following
way

A~ A{p) 1
: = \A\A(tpoA), (1.8.3)

where \A\ is t h e a b s o l u t e v a l u e o f t h e d e t e r m i n a n t of the m a t r i x of the transforma-

t i o n A. I f Af is t h e r e g u l a r d i s t r i b u t i o n c o r r e s p o n d i n g t o t h e f u n c t i o n / from L] OC

then A~ Af
x
— A ¿ Q A - \ .
We n o w describe some examples of very useful linear transformations.

EXAMPLE 1 . 8 . 1 (Reflection i n the origin). T a k i n g x — Ay —\ —y, ip(y) = p(—y)

we have A(p) = A(p). I f A y is t h e regular distribution corresponding to / G L]OC

we get A (ip)
f = f Rn fp = J Rn fp. Therefore A f = A¿.

EXAMPLE 1 . 8 . 2 (Similarity transformation). Put x = Ay := ay, a > 0. A c -

cording to (1.8.3) s i m i l a r i t y t r a n s f o r m Si_ is d e f i n e d by

S±A((p) :=a A(p(a-)).

n

If A / is t h e regular distribution corresponding to the function / belonging to Lj oc

then

SxAf(p) = a n
J f(y)p(ay) dy = J f [^j (p(x) dx.
R n
R n

Finally we have S±A¡ = A^^y

EXAMPLE 1 . 8 . 3 (Orthogonal transformation). Let A be an orthogonal trans-

f o r m a t i o n i n W. 1
F u r t h e r , let A be i n V then taking into account (1.8.3) w e have

A~ A(p>) 1
= A(ipoA)
 1.9. DIFFERENTIATION OF DISTRIBUTIONS 11
I f Af is t h e regular d i s t r i b u t i o n c o r r e s p o n d i n g t o t h e f u n c t i o n / f r o m L] oc then we

have

A~ A1
f = A f o A - i = A f o A T,

where A T
is t h e transpose m a t r i x o f A.

Although translation through the vector h of independent variables is n o t a

linear t r a n s f o r m a t i o n i n W, 1
w e shall define t h e translation operator i n the same
way as t h e previous examples.

E X A M P L E 1 . 8 . 4 ( T r a n s l a t i o n t h r o u g h t h e v e c t o r h). T h e translation operator

r_/ . w e d e f i n e b y (1.8.3)
7 putting

T-hM<f) :=Mr (ip))=A(<p(--h)).

h

O n e c a n v e r i f y t h a t i f Af is t h e regular d i s t r i b u t i o n c o r r e s p o n d i n g t o / G L} , o c then
T
-h.A f = A_ .
r hf

1.9. Differentiation of distributions

Let ip a n d / b e i n V(Q) and C^ m

^ ( 0 ) respectively, i ] c l n
. Integrating b y parts
a n d r e c a l l i n g t h a t p i s i n V{0) we arrive at

j i^f(x)ip(x)dx = - j f(x)-^<p(x) dx.

By induction we have

for | a | < m. O f course, the m a p p i n g

0 M
(-1) H
j f(x)g^ip{x)dX

is a l i n e a r c o n t i n u o u s f o r m o n D ( O ) . F r o m ( 1 . 9 . 1 ) i t f o l l o w s t h a t t h e a b o v e mapping
is t h e r e g u l a r distribution corresponding to the function <9 /, where a
df
a
denotes
the a - p a r t i a l d e r i v a t i v e o f / . W e s h a l l n o w use e q u a t i o n (1.9.1) t o define t h e p a r t i a l
derivatives of a distribution.

DEFINITION 1 . 9 . 1 . Let p be i n D ( O ) , then the linear form defined b y

(-l)WA(d p) a

will be called the a-derivative of A . This form will be denoted by D a

A .

F r o m n o w o n , f o r s i m p l i c i t y o f n o t a t i o n s w e w r i t e / i n s t e a d o f Af i f Af is t h e reg­
ular d i s t r i b u t i o n corresponding t o t h e function / . N o t e that i f / is i n ( 7 ( ) ( 0 ) , m
then
Df a
= df a
f o r | a | < m. Moreover, for each distribution A G £>'(0), D a
A = D&A,
where the operators D a
a n d D@ h a v e o n l y t h e d i f f e r e n t o r d e r o f d i f f e r e n t i a t i o n .
12
1. DEFINITIONS AND PRELIMINARIES
EXAMPLE 1 . 9 . 1 . Let v be i n Lj (R) oc a n dput u(x, y) = v(x + y). W e shall now

s h o w t h a t t h e f u n c t i o n u fulfils t h e w a v e equation

£>(2,0) _ (0,2) w D u = o ( L 9 > 2 )

in R . 2
W e have to show that

JJ v
( x J
r y) (^~^(x y y) - 2 / ) ^ dxdy = 0 (1.9.3)
R 2

for (p G £ > ( R ) . 2
I n order to d o it w e change the variables i n the above integral

p u t t i n g £ = x — y, rj — x + y . Therefore

A n easy c o m p u t a t i o n shows that

a 2

A f t e r c h a n g i n g variables i n t h e integral (1.9.3) w e have

id 2
d \ 2

v(x + y) yo^<P( > x

V) ~ g^2^( > 2/)J x d x d
V

d 2
ft + V V-t
= 2 ¡ ¡ V { V )

dridf

R \ R /

N o t e t h a t t h e f u n c t i o n <£> ( - i p , ^ p ) v a n i s h e s w i t h o u t a b o u n d e d i n t e r v a l depending

o n r¡. T h i s i m p l i e s t h a t t h e i n t e r i o r i n t e g r a l i s e q u a l t o z e r o f o r e a c h rj. Therefore

(1.9.3) holds.

We shall n o wshow that differentiation o f distributions is a linear continuous

operation i n V(Q).

THEOREM 1 . 9 . 1 . If A v and A are in V'(Q) and A u tends to A as v -ï oo in

V(Q) then D A a
u also tends to D A a
as v - > o o .

PROOF. T h e t h e o r e m is a n i m m e d i a t e consequence of the definition of the

derivatives o f distributions. •

1.10. W e a k derivatives of locally integrable functions

DEFINITION 1 . 1 0 . 1 . L e t/ be i n L} (Q).
oc I f there exists a function v belonging
to Lj (Cl)
oc such that

f f <9H
/ v(x)cp(x) dx = (-l) a
/ f(x) — ip(x) dx

f o r (p G then w e say that v is a-weak derivative of / .

 1.10. WEAK DERIVATIVES. 13
In this case we shall write df a
= v instead of D f a
= v. That means that
the distributional derivative D a
f of the locally integrable function / is t h e regular
distribution corresponding t o v. In some situations, when a G N , the weak or
distributional derivatives will be denoted traditionally.

THEOREM 1 . 1 0 . 1 . If the locally integrable functions v\ andv<i are a-weak deriva-

tives of f £ L} (ft) oc then Vi(x) — 2 TJ(X)
almost everywhere on ft (we write V\ — v 2

a.e.).

PROOF. T h e theorem is a n immediate consequence of d u Bois-Rymond's

lemma. •

EXAMPLE 1 . 1 0 . 1 . L e t H(x) = 0 if x < 0 and H(x) = 1 if x > 0 (Heaviside

step function). Therefore D H((p)

1
= - H(x)p> (x) f
dx = (p(0) = 6((p). T h e

distributional derivative DH X
of H is n o t a weak derivative o f H.

EXAMPLE 1 . 1 0 . 2 . P u t f(x,y) = 0 if x > y and f(x,y) = y - x i f x < y for

(x, y) G M 2
. T h e function / has the classical derivatives outside t h e set A = {(#,y) :
x = y , x,y £ R } . Namely

d d

~f(x,y)=0 i f x> y a n d — f(x, y) = - 1 i f x < y.

Similarly

d d
—f(x, y) = 0 if x > y and ^ ~ / ( ^ , y) = 1 if x < y.
dy dy
We shall n o w show that f = <9 ( 1
' 0 )
/- Indeed, note that

oo oo oo / y \

- J J f{x,y)-^-ip(x,y)dxdy = - ^ I ^ ( y - x)-j^<p(x,y) dx J
-oo \— oo
oo oo
d
= - / / rtx,v)éc\dy= f f — f(x,y)p(x,y) dx dy.
-oo \-oo

Analogously one c a n show that D ^ f = ô ( 0

' 1 }
/ .

EXAMPLE 1 . 1 0 . 3 . L e t / be absolutely continuous o n [a, ¡3}. Then f(x) = /(a) +

/* f'(t) dt a n d f G Note that

p p

•ff(t)<p'(t)dt =J f(t)<p(t)dt

i f ip i s i n V(a,/3). This implies that / ' is the weak derivative o f / .

Further, we shall need the following

LEMMA 1 . 1 0 . 1 . Let f be a continuous function defined on the interval (o¡,/3).

// /(*) dt = 0 for every Lp G V(a /3) i then f is a constant function.

 1. DEFINITIONS AND PRELIMINARIES
14

PROOF. I n the first place we show that the function 6 G V(a,/3) is t h e deriva-

t i v e o f s o m e f u n c t i o n ip b e l o n g i n g t o V(a, ¡3) i f f 0(t) dt = 0.

N e c e s s i t y is e v i d e n t . A s s u m e t h a t 0 G î > ( a , / 3 ) a n d f£ 6(t) dt = 0 . P u t ip(x) : =

dt. W e s h a l l n o w s h o w t h a t s u p p ip C (a + e, /3 - e) f o r s o m e e > 0. Indeed,

let supp 6 C (a -f e, /3 - e ) . F i r s t o f a l l ip(x) = 0 i f x G ( a , a + e ] . F o r x G [/3 - e, ¡3)

we have ^(x) = J^~" 0(t) dt + f*_ 6(t)dt. e This implies that f*6{t)dt = 0 for

x G [/3 — e , / 3 ) . Therefore s u p p ^ C (a + e,(3 — e). O f c o u r s e , ?// = #.

Further, each function p in T>(a,(3) can be represented as follows

p(x) = tpo(x) J Lp(t)dt + 6{x),

where ip 0l 0 G £>(o!, ¡3) and f^p 0 = l,f^6 = 0. I n fact, one c a n put

6>(x) : = p(x) - <¿> (» y

0 <¿>,

where J^f <£>o = 1- W e see a t once that f£ 6 — 0. In accordance w i t h the previous

c o n s i d e r a t i o n , t h e r e e x i s t s a f u n c t i o n ip i n (3) s u c h that

<p(x) = v?o(x) J <p + il>'(x).

Assume that / is a c o n t i n u o u s f u n c t i o n , t h e n fp — p fcp +0 ip f.

f
F r o m

t h e a s s u m p t i o n , w e h a v e f£ ip'f = 0. T h e r e f o r e fp = c p, w h e r e c = fp .
0

Because of the continuity of / i n view of d u B o i s - R y m o n d ' s l e m m a it follows that

f{x) = c for x G (a,/3). •

We are n o w i n a p o s i t i o n to prove the following

THEOREM 1 . 1 0 . 2 . Iff is a continuous function on (a, [3) but it is not absolutely

continuous, then it has no weak derivative.

PROOF. I n fact, assume that the locally integrable function v is the weak

derivative o f / , t h e n for each f u n c t i o n p in V(a /3)

1 we have

p p

vp.
- f « - f

Put V(x) = / * v(t) dt, then - /f v<p = f a V<p'. Therefore (/ - V)tp' = 0. B y

virtue of L e m m a 1.10.1, f = V + c. This contradicts our assumption. •

REMARK 1 . 1 0 . 1 . Let p a n d A b e i n X>(cx,/3) a n d V'(a,l3) respectively, (a, fi) C

R. A s s u m e that the d e r i v a t i v e DA o f A is e q u a l t o zero, t h i s m e a n s A{p') — 0 for

 1.10. WEAK DERIVATIVES. 15
each p G D(a, /3). W e k n o w t h a t f o r e a c h p i n V(a, (3) t h e r e e x i s t t w o f u n c t i o n s tpo

and 9 in V(a,(3) such that

0 (3 ¡3

P = Po Jp+0 and j p0 = 1, J9— 0.

a a: a

The f u n c t i o n (¿? c a n b e c h o s e n i n d e p e n d e n t l y
0 o n p. Hence we have

P
A(p) = A(p ) 0 J <p + A(0).
a

Let ip G V(a,(3) and ip' = 9. Since A(ip') = 0 therefore A(ip) = A(p )p.
0 From

this, it follows that A is a r e g u l a r d i s t r i b u t i o n g e n e r a t e d by the constant function

A(po). This d i s t r i b u t i o n fulfils the differential equation Dx = 0. Therefore each

distributional solution of the d i f f e r e n t i a l e q u a t i o n Dx = 0 is a r e g u l a r distribution

generated by a constant function. N o t e t h a t it satisfies this e q u a t i o n at e a c h point

x o f ( a , (3). O n e can replace the constant f u n c t i o n A(po) by another function, for

e x a m p l e p u t t i n g f(x) : = 0 i f x is a r a t i o n a l n u m b e r , b u t for x w h i c h is n o t rational

it takes f(x) := A(p ). 0 O f course,

p P

From this, it follows that the function / is a d i s t r i b u t i o n a l s o l u t i o n o f d i f f e r e n t i a l

equation Dx — 0. Obviously, the function / is n o t differentiate in any point of

( a , ¡3). The above observation indicates that there exists an essential difference

between two statements: the function / satisfies a differential e q u a t i o n everywhere

or almost everywhere on a set (I and it fulfils this differential equation in the

d i s t r i b u t i o n a l sense o n 0.

EXAMPLE 1 . 1 0 . 4 . W e shall now show that D l n | • | = Indeed, let p be in

V. T h e n we have

co 0 oo
D i n I • \(p) = — J In \x\(p'(x) dx = — J hi(—x)p'(x)dx — Jhix(p'(x)dx.

Of course,

CO oo
Inxp'(x) dx = l i m / laxp'(x) dx.
e-+0+ J
/
Note that

—e —(

J ln(—x)p'(x) dx = — p{—e) l n e + J x
oo
 1. DEFINITIONS AND PRELIMINARIES
16

and
oo oo
— I hix p'(x) dx = Lp(e) l n e -f- / dx.
e e

It is e a s i l y s e e n t h a t l i m _ e > 0 + (p(e) — p(—e)) lne = 0. F r o m t h i s i t f o l l o w s that

oo
— f Yn.\x\p'(x) dx = l i m f ^ 0 * 0 ^
J e-+0+ J X
-OO \x\>€

F i n a l l y we have D In | • | = -pj.

1.11. Local Sobolev spaces

In this section there will be i n t r o d u c e d l o c a l S o b o l e v s p a c e s as a n a t u r a l genera-

lization of the spaces Lf (fl).
oc Assume that / has the weak derivatives df a
up to
order m o n O . W e shall n o w define a family of s e m i n o r m s i n this way

\f\\m^K=[ E \\dV\\l P { K ) ) if 1 < P < 0 C , (1.11.1)

\\a\<m J

||/||m,oo,x = max \\d f\\ ~>

a
L {K) if p = oo, (1.11.2)
|a j <m
where K is a c o m p a c t s e t c o n t a i n e d i n ft. Let

W£f(il) := {/ : df
a
e L (Ü) p
oc for \a\ < m, dfa
is t h e w e a k d e r i v a t i v e o f / } .

O f course, W^ (ft)
p
is a v e c t o r space.

DEFINITION 1.11.1. Vector spaces W¡^f(ft) equipped with the appropriate

seminorms (1.11.1) o r (1.11.2), where K goes t h r o u g h the family of all compact

s e t s c o n t a i n e d i n ft are called local S o b o l e v spaces over ft.

Obviously, W^ (ft)
p
= L (ft). p
oc The purpose of our considerations is t o make
a description of functions belonging to W ^ ' ( í í ) , w h e n 1 < p < c
p
o o . L e t tp b e i n V.
If / is i n L (ft),
P
oc then the following integral

j f(x-y)f(y)
(
dy

is m e a n i n g f u l f o r e a c h x G l n
. The function , defined by

U(x) : = J (p(x-y)f(y) dy
n

is c a l l e d t h e regularization of / b y m e a n s o f p. T a k i n g f(x) = f(x) if x G ft and

f(x) = 0 if x ft we obtain

U(x) = (ip * f)(x) = j <p(x - y)f(y) dy.

 1.11. LOCAL SOBOLEV SPACES 17
By virtue of T h e o r e m 1.3.1 we infer that is s m o o t h and

d ^ u ( x ) =
J - y ) f i y ) d y
- ( L 1 L 3 )

Let t{> > 0 be in V, s u p p i/> C i ? i ( 0 ) a n d J> = 1. Put V (z) e = e~ ip

n
(f), e > 0. In
the sequel, we shall need the regularization f £ given by

Mx)~U.{*) = j Mx-v)f(v)dy = JMx-y)f(y)dy. (1.11.4)

At the beginning, we present two lemmas.

LEMMA 1 . 1 1 . 1 . If f is in L] (ÇÎ) and d f is its weak derivative,

oc
a
then for the
regularizations fe and (d f) given by ( 1 . 1 1 . 4 ) we have
a
e

|^/e(z) = (d fUx),
a
(1.11.5)

if x G íl e — {x : x G O, dist(x, 9 0 ) > e}.

PROOF. In accordance w i t h the definition of weak derivative we get

(d f) (x)
a
£ = JA(x-y)d f(y)dy a

= (-i) H
f~M*-y)f(y)dy
n

= J Q^Mx-y)f(y)dy-
n

By (1.11.3) w e get (1.11.5) for x G iî . c T h u s , t h e p r o o f is finished. •

LEMMA 1 . 1 1 . 2 . 1/ / is in L f o c (0) and 1 < p < oo, then f , e given by (1.11.4)

tends to f as e —>• 0 in L (Q).
p
oc

PROOF. After a change of variables i n the integral fRn ip (x

e — y)f(y) dy we
obtain

f (x)
e = J ip(z)f(x - ez) dz.

Bi(0)
18 1. D E F I N I T I O N S A N D P R E L I M I N A R I E S

Let K C O be a compact set. Using Minkowski's inequality we get the following

estimation

< J I r/>(z)\f(x-ez)\dz
\Bi(0) )

<j 1>(z) J \f(x ~ ez)\ p

dx\ dz

£i(0) \K )

< \\f\\LP(B (K))-

€

A s s u m e t h a t e < d — d i s t ( i ^ , dCt). T h e n we get

Without loss o f generality one c a n a s s u m e t h a t K = O ' , ft' (s Çl. Because of the

density of C (Q!) C in L (Çl'),

v
for fixed rj > 0 there exists a function g in C (Q ) c
f
such

that 11/ — g\\LP(K) < f - B y v i r t u e o f T h e o r e m 1.3.2, w e infer t h a t \\g — g \\Lp(K) <

e \
f o r s m a l l e, w h e r e g e is defined b y (1.11.4). Note that f e —g e = (f — g)e- B y (1.11.6)

we have

||/e - 9C\\LP(K) < 11/ - 9\\LP{B {K))

C < \\f ~ 9\\LP(K) + \\f\\LP(B (K)-K)'
£

Since / G L (Çl), p
oc therefore

\\f\\LP(B (K)-K)
E < ^

for s m a l l e. N o w , o n e c a n m a k e t h e f o l l o w i n g estimation

11/ - fe\\LP(K) < 11/ - 9\\LP(K) + lb - 9C\\LP(K) + \\9e ~ fe\\LP{K) < V

for s m a l l e. T h u s , t h e p r o o f o f t h e l e m m a i s finished. •

THEOREM 1 . 1 1 . 1 . Let f be in L (ft) p

oc and let D f a
be its distributional deriva-
tive. D f a
is weak derivative vf f and belongs to L (il) p
oc iff there exists a sequence
(/',.,), f u G C°°(Q) convergent to f in L (Vt) p
oc such that the sequence (d f ) a
v is a
Cauchy sequence in L (ft).
p
oc

PROOF OF SUFFICIENCY. A s s u m e that the sequence f v G C°°(ft) con-

verges to / in L (ft)
p
oc and that the sequence (d f ) a
u is a Cauchy sequence in

L (Q).
p
oc B y continuity of differentiation, df a
u tends to Df a
as v —>• o o i n D'(il).
Since (d f ) a
u is a Cauchy sequence i n L f o c (0), therefore there exists a function

v a G L (íl)
p
oc such that

lim [ 8 Uy a
= [ vp a - D f(<p) a

n Q

for ip G V(ü). F i n a l l y w e get

D f(<p)
a
= (-l)W J fd v a
= j v . aV
 1.12. SOBOLEV SPACES 19
T h i s m e a n s t h a t Df a
= v a is t h e w e a k d e r i v a t i v e o f / . •

PROOF OF NECESSITY. L e t K CO be a compact set a n d let d — dist(üf, diï).

A s s u m e t h a t df a
is t h e w e a k d e r i v a t i v e o f / i n Q. Take the sequence ^fi j, i / G N ,

fi defined by (1.11.4). T h e n b y L e m m a 1.11.1 w e have

a n

for x G K, if \ < d. F r o m this and L e m m a 1.11.2, i t f o l l o w s t h a t fi and (d f)±

a

tend to / and d a
f a s v —> o o , r e s p e c t i v e l y u n d e r t h e n o r m ||- | | L P ( Í Í ) - This statement

finishes the p r o o f of necessity. •

F r o m our consideration the following corollaries follow.

COROLLARY 1 . 1 1 . 1 . C°°{tt) is dense in W%¿ (íl) p

ifl<p<oo.

COROLLARY 1 . 1 1 . 2 .
The space W[£¿ (Sl), 1 < p < o o is complete in the fol- p

lowing sense: if (/^), f G W^¿ (ü) is a Cauchy sequence in W¡^¿ (íí),

v then therep p

exists a function f in Wj^¿ (ü) such that l i m ^ o o fv = fin the sense ofp
W^ (Q). P

PROOF. C o m p a r e w i t h T h e o r e m 1 . 1 2 . 1 . •

COROLLARY 1 . 1 1 . 3 . / / / is in W^f(ft), then

( i ) DP(d f) = d^ ^f
a
if \a + /3| < m, a+

( i i ) d f — d f if \a\ < m, where d and d@ have only the different

a 13 a
order of
differentiation.

1.12. Sobolev spaces

W e shall now introduce Sobolev spaces of integer order and establish some of their
properties. Similarly as i n the previous section we denote by W m , p
( f 2 ) the set
{/ : / G L (ü),
p
1 < p < oo, df a
G L (ü)
p
for \a\ < ra, where df a
is t h e weak
derivative of / } . Obviously, W ' (Q)
m p
is a v e c t o r space.

DEFINITION 1 . 1 2 . 1 . T h e vector spaces V F m

' (Q)
p
equipped w i t h the norms

ll/lkp = ( E W fWUn) da
] i f
1<P<™ (1.12.1)
\oc\<ra
or

||/|| m > 0 0 = max ||5 /|| oo

a
L (Q) if p = oo (1.12.2)
\a\<m
are c a l l e d S o b o l e v spaces over O.

Clearly, W°> (tt) p

= L (ü). p
W e first prove the following

THEOREM 1 . 1 2 . 1 . W ' (Q) is m p

a Banach space.

PROOF. L e t (f ) b e a C a u c h y u sequence i n W m , p
( 0 ) . T h e n for each |a| < m
there exists a function f a in L (Q) P
such that \\d f a
u — f \\Lp(Q)
a tends to zero as

v - » oo. Since the sequence (f )

u converges to / i n L ( 0 ) , therefore
P
the sequence
 1. DEFINITIONS AND PRELIMINARIES
20

{d a
fv) converges to D f a
i n t h e d i s t r i b u t i o n a l sense for each a G N n
. Therefore

Df
a
= f a i n V'{0) if|a| < ra. This means that

-1) H
j fd"tp = J f <pa

for Lp G V(fl). Therefore fa is the weak derivative d f a

of/ if \a\ < ra. T h u s , t h e

p r o o f is finished. •

In the next chapter w e shall show that C°°(ü) is dense i n W ' (il)m p
i f O is a n
arbitrary open set a n d 1 < p < oo. I n this section w erestrict ourselves t o presenting
the fact t h a t V(Vt) is dense i n W ^(ü), w h e n ft = W
m 1
a n d1 < p< oo. Moreover,
we show that V{Q) is n o t d e n s e i n W ^{ü) iîtt^W m 1
i n general.

THEOREM 1 . 1 2 . 2 . If1 < p< oo ; then V is dense in W™^.

PROOF. L e t xp b e as i n t h e previous section a n d / G W . m,p

T h e n t h e regular-

ization f =ip *fisa,

e e smooth function and df
a
e = d ip oc
e */ . B y virtue of Young's

inequality, it follows that df a

e is i n L p
for each a ^ f f .

Since / is i n W , m,p
therefore the weak derivatives d a
f of / belong to L p
for

|a| < ra. T a k i n g into account the definition o fweak d e r i v a t i v e , b y T h e o r e m 1.3.1

we get

= Jip±(x- y)d f(y)dy

a
for vG N.

Moreover, from this theorem it also follows that d f±

a
tends to df a
as v —>> o o i n

L .
p
This means that C°° is dense i n W '. m p

W h a t is left t o s h o w i s t h a t t h e f o l l o w i n g l e m m a h o l d s . •

LEMMA 1.12.1. / / / is smooth and d f a

G L , \a\ < m, then there exists a
p

sequence (f ), f u v G V such that d f a

v tends to d f a
in L for \a\ < ra.
p

PROOF. L e t XB V b e t h e characteristic function of the ball B (Q),

U v G N . P u t

9u(x) = (XB U * ^){x) = JXB ( u

x
- y)^(y) v- d

N o t e t h a t g {x) v = 1 forx G B I / _i(0). Assume now that / is a s m o o t h function a n d

df
a
belongs to L p
for \a\ < ra. W eshall now show that d (g f)
a
l/ tends to dfa
as

v - »oo i n L p
. O f course,

d (g„f)
a
= g d°f v + E (f\&9vff*-Pf.

0</3<a
It is e a s y t o c h e c k that

-Q^p9Á ) x
= 0
for x G £i/_i(0) a n d /3 > 0 .

Therefore

d^(9,f)(x)=g^f(x) for i e B . i ( f l ) .
 1.13. DIFFERENTIAL EQUATIONS OF THE SECOND ORDER. 21

It is e a s i l y seen that

Hence we have

I
f\d (9 f)-d f\*]
a
v
a
< J E (^dlg^-Pf
\R n
—B„

<
0</3<a

Since df a
G L p
for | a | < m , therefore t h e right side o f t h e above inequality is small

f o r l a r g e v. This statement finishes the proof. •

EXAMPLE 1 . 1 2 . 1 . W e s h a l l n o w s h o w t h a t D ( G , 1) is n o t dense i n W 1 , 2
( 0 , 1).
Indeed, let p be i n î>(0,1), then

a:r

/ V /
2
(t ^
2

L (0,1)
2
= / ^(y)^ da: < I 11 \<p'(y)\dy\ dx
J
0 y vo \o

< ( y i i ^ i i i ^ o , ! ) ^ ) ^II^IIL (O,I)- 2

Therefore

IMIL (O,I)2
^ ll^'IU (o,i) 2
(1.12.3)

for p G £ > ( 0 , 1 ) . S u p p o s e t h a t f o r s o m e p G D ( 0 , 1 ) , ||<¿> — X ( o , i ) I I 1,2 < § > t h e n

1 1

/ k-X(o,i)| + j 2
W?<\-
0 0

Hence ||V?'||L (O,I)

2
< \. Consequently

||^'IU (0,1)
2
<l-\\p~ X(0,l)||l,2 < |MU (0,1)-2

This statement gives us a contradiction w i t h (1.12.3).

Let W™*(Q) denote the closure of V(Q) i n the space W ^(Ü).

m
T h e above
example shows that W™' V(Q) P
is different f r o m W m
' p
( i l ) i n general.

1.13. Differential equations of the second order with measures

as coefficients

We shall consider the boundary value problem

—11" + \1\U — /i 2

(1.13.1)
u(a) = u(b) = 0,

where \i\ a n d ¡12 a r e r e a l B o r e l measures, ¡i\ > 0. I f ¡i\ a n d /¿2 a r e i n t e g r a b l e

functions w i t h respect t o t h e Lebesgue measure, then t h e R i t z - G a l e r k i n m e t h o d is
 1. DEFINITIONS AND PRELIMINARIES
22

often used to investigate P r o b l e m (1.13.1). Here we shall show that this method
may also b e a p p l i e d t o solving P r o b l e m (1.13.1) u n d e r the above assumptions. W e
are l o o k i n g for a c o n t i n u o u s f u n c t i o n u v a n i s h i n g at t h e e n d p o i n t s a, b a n d fulfilling
equation (1.13.1) i n t h e d i s t r i b u t i o n a l sense. This means that

b b b

— J up" dx j + uipdfii(x) = J ip dfi2(x) for p G £>(a, 6), (1.13.2)

a a a
where f^ipdfi2(x) := pqd\¡i2\(x), \^\ is t h e v a r i a t i o n o f fi 2 and \q(x)\ = 1 a.e.

(see Section 2.7). O n e c a n show that

14Q 7 , 2
( a , b) = {u : u is absolutely continuous, u(a) = u(b) = 0 a n d u' G L (a,6)} .
2

For simplicity of notation we put

b b
a(p,ip) :— j pip' dx -f- j pip dfii(x),

for p , ip G W 0
1 , 2
( a , b). It is easy to check that a is a bilinear s y m m e t r i c positive

definite form on W 0
1 , 2
( a , b) a n d f3 i s a l i n e a r f o r m o n W 0
1 , 2
(a,6). If we are looking

for a s o l u t i o n u i n W^ (a, 2
6), t h e n equation (1.13.2) is equivalent t o t h e equation

6 6 6

Ju'p dx f
+ j updfii(x) — J pd¡i2(x), p G D ( a , b).
a a a
or, u s i n g t h e a b o v e n o t a t i o n , also

a(u,p) = /%), (p G V(a,b). (1.13.3)

In the sequel we shall need the following t w o norms

\\u\\ := [a(u,u)]î
a and \\u\\ := D \\u'\\ 2
L

for u i n W¿' (a b).

¿
7 N o wwe are i n a position to state

THEOREM 1 . 1 3 . 1 . The following norms || • 11x,2^ || • |D arid || • \\a are equivalent

on Wo ' 1 2
(a, 6).

PROOF. S i m i l a r l y t o E x a m p l e 1.12.1 o n e c a n s h o w that

\\<p\\v<(b-a)\\<p\\ . D (1.13.4)

It i m p l i e s t h a t t h e n o r m s || • a n d || • |jr> are equivalent o n the space W^ , 2

(a, 6).

It is e a s y t o see t h a t

\p(x)\ < (b-a)i\\p\\ D

f o r x G [ a , b] a n d p G W 0
1 , 2
( a , 6). Therefore

< (6-a)*|| |b. V (1.13.5)

 1.13. DIFFERENTIAL EQUATIONS OF THE SECOND ORDER. 23
Note that

IMU < 1Mb + IMU2/H

for p> G W 0
1,2
(a,6). Therefore we have

1Mb < IMU < 1Mb + I M b v i

for ip G W 0
1)2
(a,6). B y (1.13.5) w e obtain

IMIi^ < ll^||ioo//i([a,6]) < (b- a)|Mli>Mi([M])-

Finally we get

1Mb < IMU < [l +((6-a) i([a, M &]))*] \\<p\\ . D (1.13.6)

Thus, the proof of our theorem is finished. •

Let (•, - ) L 2
and (•, -)¿2 denote the ordinary inner product o n the space L 2
( a , b)

and L 2
lli ( a , b). W e s e t

(<p,il>) D := ( p ' ^ V , ^ € W 0
1 , 2
(a,6),

(<p, ip) := (ip, zp) D + (</?, ?/;) 2, L y?, ^ G W ¿ ' ( a , &)> 2

and thus we have

MI

for ip, i> e W 0

1 , 2
( o , 6). W e know that (W 0
1 , 2
( a , 6), (•, • ) ) i s a H u b e r t space.

COROLLARY 1.13.1. The spaces (Wo' (a,6), 2

(•, -) )D and (WQ' (a,b),a(-, 2
•)) are
Hubert spaces, too.

Now, w e are i n a p o s i t i o n to prove the main

THEOREM 1.13.2. Problem (1.13.1) has exactly one solution in WQ , 2

(a, b).

PROOF. B y t h e d e f i n i t i o n o f W Q ' ( a , 6), t h e s e t T>(a 2

b) is d e n s e i n W 1 , 2
(a, 6) s o
(1.13.1) i n
1 0

there exists at m o s t one solution of P r o b l e m W^ , 2

(a, b). Since the space

(Wo A , i !
( a , 6 ) , ( • , • ) ) i s a H u b e r t s p a c e a n d /3 i s a c o n t i n u o u s l i n e a r f o r m o n WQ ,2
(a, 6),

there exists a function u in W j ' ( a , 2

6) such that (1.13.3) h o l d s . This finishes the

proof. •

In general, there exist no more regular solutions of P r o b l e m (1.13.1), apart

from those belonging to WQ' (a, 6). 2

EXAMPLE 1.13.1. L e t us consider the differential equation

-x" + ÔXX = f

w i t h the boundary condition

x(0) = x(l) = 0,

where Si is t h e Dirac measure concentrated at the p o i n t t = \ and / G L (0,1).1

It

is e a s y t o see t h a t t h i s p r o b l e m h a s n o c l a s s i c a l s o l u t i o n s ( b e l o n g i n g t o WQ (0, 1)). ,2

 1. DEFINITIONS AND PRELIMINARIES
24

Now, we use the R i t z - G a l e r k i n m e t h o d to determine approximate solutions o f

Problem (1.13.1). W e b e g i n with a formulation of the Ritz theorem.
Let E be a real vector space and a : E x E ^ W be a, b i l i n e a r symmetric
positive definite form. M o r e o v e r , l e t /3 : E —» M b e a l i n e a r f o r m . L e t us consider
the quadratic form

F(x) := ^a(x,x) - (3(x).

THEOREM 1.13.3 (Ritz). The following conditions are equivalent:

( i ) a(x,y) = (3(y) for y G E,
(ii) F(x)= míl F (y).

We introduce the norm ||x|| a = [ a ( x , x)]i i n t h e s p a c e E. If we assume that

(E, II • | | ) i s c o m p l e t e , t h e n
a (E, a ( » , •)) i s a H i l b e r t s p a c e . Let ( x n ) , n = 1, 2,... be
a sequence of elements x n belonging to E such that

lm{x : n = 1,2,...} = E, (1.13.7)

1,
n

where lin{x n : n = 2,...} d e n o t e s t h e vector space spanned b y the elements x. n

Let E n be the space spanned b y elements x\ . . .x . n L e t x * be a n element in En

such that F ( x * ) = i n f ^ ^ F(y). It is k n o w n that — tends to zero, when

F ( o ; ) - i n f y e £ ; J F ( y ) .
The above information will be used in determining approximate solutions of
Problem (1.13.1) i n t h e s p a c e WQ' (a,
2
b). Let E be n the vector space spanned by
the functions /2,..., f . n T h e quadratic form F takes the following form

F(y) = G(\ ,...,\ )

2 n

- . i n n ^ n n \ \ n ^
= 2 E E A
^ - / ^ i à
+ E E A
^ / M ^ I W -^ A
*/ ^ ^ W '
\i=2 j=2
(1.13.8)
where y = A2/2 \ fn-
n Formula (1.13.8) m a y b e rewritten i n matrix form

1
G(A) - ^ A T
( r + A ) A - A T
P ,

where

"A" 2
"(/2,/2>i ••• (f2ifn)D

A= r =
_A_ N
(fn, $2)0 (fni fn )D_
fa f2fndfl (x) 1

P =
jl fnÍ2dfJ. (x) 1 fa fnfndfl (x)_
1 Jafndn {x)_ 2

It is e a s y t o c h e c k that
G(A*) = inf G ( A ) ,
AGE-- 1

+ A)A* = p. (1.13.9)
when
(r
 1.13. DIFFERENTIAL EQUATIONS OF THE SECOND ORDER.. 25
Now, we construct a sequence ( / ) i n the space
n W 0
1 , 2
(0,1) which has property
(1.13.7). F o r m = 2 w e p u t f {t) 2 := t f o r 0 < t < § , f {t) 2 = 1 -1 for f < t < 1. I n
the general case w e take

f 2 i t - f f e _ f f o r < t <

/ m ( í ) : = < - 2 * í + ^ f o r f ^ < t < ^ r

[o for other t i n [0,1],

where m = 2 n
+ k, 1 < k < 2 n
, n = 0,1, 2,...

THEOREM 1.13.4. T/ie functions fnj n — 2,3,... constitute a complete or-

thonormal system in ( W Q ( 0 , 1 ) , (•, -)D)- , 2

PROOF. Note that the weak derivatives of the functions fn are k n o w n Haar

functions % . F o r / n G W 0
1 , 2
( 0 , 1 ) we have

_)dt
f{x) = J g(t)<
o
for s o m e g G £ ( 0 , 1 ) , x G [0,1]. T h e f u n c t i o n g h a s t h e F o u r i e r
2
representation

\ oo
g= g(x)dx + ^ c n x n (1.13.10)

o "= 2

w i t h respect t o t h e H a a r functions. It is clear that g(x) dx — 0. S i n c e t h e space

(WQ' (0,
2
1 ) , (•, •)D) is c o m p l e t e , therefore there exists a function / i n WQ' (0,
2
1)

such that

J^c f n n = f. (1.13.11)
71 —2

Series (1.13.11) converges i n t h e d i s t r i b u t i o n a l sense. Hence w e have

oo

$> Xn = f. n (1.13.12)
n=2
It is k n o w n t h a t series (1.13.12) converges to / ' i n the space L ( 0 , 1 ) (also
2
a.e.).

This implies that f — g a.e. o n [0,1]. From this we obtain that f(x) — f(x) for
each x G [0,1]. F i n a l l y w e have

oo

f(x) = J2(fJn) fn- D

n=2
This completes the proof of the theorem. •

COROLLARY 1.13.2. lin{/ n : ra = 2 , 3 , . . . } = W 0 ' (0,1) with respect to the

norm \\ • || . a

For the differential equation

-x" + 5ix = 1
; N (1.13.13)
26
1. D E F I N I T I O N S A N D P R E L I M I N A R I E S

we obtain the following m a t r i x equation

"A • 2
a 2

\2m-\-l Qj2 + m 1

where a-> = \ , a k2 +l = 2 ï 2
for I— 1, 2 , . . . , 2, k =
k
1,..., m. T h e exact solution
of (1.13.13) is

x { t ) = {- i
t
+ ¥ f o r O < i < |

l - T +i*-è for I <t < 1.

The following graphs c o m p a r e x a n d x* for n = 2 , 4 , 8.

0 0.025 0.5 0.75 1

0.125 0.875
 C H A P T E R 2

LOCAL PROPERTIES OF DISTRIBUTIONS

2.1. S m o o t h partitions of unity

We have a l r e a d y seen t h a t d i s t r i b u t i o n s c a n n o t b e assigned values at i s o l a t e d p o i n t s .

O n e c a n n o t , for i n s t a n c e , say t h a t " a d i s t r i b u t i o n A is e q u a l t o zero at X Q " • H o w e v e r ,
t h e s t a t e m e n t t h a t " A i s e q u a l t o z e r o i n n e i g h b o u r h o o d U o f a?o" c a n b e meaningful.
This will mean, namely, that for every p? G V(ft) with support in U, we have
A(p) — 0. For example, the D i r a c m e a s u r e is e q u a l t o zero i n the neighbourhood
of every point x ^ 0. W e say that a distribution A G V (ft)
(
vanishes in an open
set ftf
C ft i f f o r e v e r y (p G V(ft) with support i n f i ' w e h a v e A(p>) = 0. A typically
local property is t h e following one. A distribution vanishes i n a n o p e n set í í ' c í í
iff it e q u a l s zero i n the neighbourhood of every point i n this open set. This fact
can be used to construct a d i s t r i b u t i o n i n its e n t i r e l y w h e n it is d e f i n e d everywhere
only locally. W e shall prove this property i n S e c t i o n 2.3.
Now, we start w i t h presenting the following

LEMMA 2 . 1 . 1 .
Let ft be an open set in W and K C il be a compact set. 1
Then
there exists a function f in V(fl) such that 0 < f(x) < 1 for x E ft and f(x) — 1
for x G K.

PROOF. Take a function ip in V such that ip > 0, s u p p ip C Bi(0) and f ip = 1

and put

^ e ( x ) : = e - n
^ ( | ) , € > 0 .

Further, a s s u m e 2e < d i s t ( Ü T , dft). The characteristic function \B (K) ° ^ B (K)

e is

integrable. W e shall n o w show that the regularization Xe,

Xe(x) = J 1p (x - y)x-BjK)(y)
e
d
V

may be taken as a required function /. Indeed, the function Xe is s m o o t h (see

Section 1.11). A n easy computation s h o w s t h a t s u p p Xe C ft, Xe(#) = 1 for x G K
and 0 < Xe( )
x
< 1 f o r x G ft. T h u s , the p r o o f is finished. •

T h e following theorem is i m p o r t a n t for applications.

THEOREM 2 . 1 . 1 . Let i9 = {ft : a G A} be a collection of open sets.

a Assume
that p G V(ft) and ft C A then there U<EA
exists a finite subcollection O i , . . . , ftk
of d and a finite collection of functions p p G V(ft ) v = 1 , . . . , k such that VJ v v J

k
<P = '52<P». (2.1.1)
28
2. LOCAL PROPERTIES OF DISTRIBUTIONS
PROOF. S i n c e supp ip is c o m p a c t , therefore there exists a finite subcollection
O i , . . . , ft m C i9 s u c h t h a t s u p p (p C U j L i % • Let x G s u p p 99, t h e n t h e r e exists an
open neighbourhood U x of x c o n t a i n e d i n flj for s o m e j. Obviously, we choose U x

so t h a t U x C ftj. B e c a u s e o f t h e c o m p a c t n e s s o f s u p p (p o n e c a n c h o o s e a finite set

of the neighbourhoods U X l ,..., U , Xn X{ G s u p p <¿> s o t h a t s u p p ip C U¿=i Í4 ¿ and
C íí¿ for s o m e j. Put ÜTj : = Ul=i U. Xi In view of L e m m a 2.1.1 there exists a
function ijjj G T>(ftj) such that 0 < ipj{x) < 1, x G f2j a n d ipj(x) = 1 for x G üfj.
We s h a l l n o w define n e w f u n c t i o n s as follows:

V?2 ~ ^2(1 - ^l),

We are able to s h o w that

k k

j=l j=l
Of c o u r s e , (2.1.2) is t r u e for k = 1. Suppose

j=i i=i

Therefore

3=1 3=1
The e x p r e s s i o n o n t h e r i g h t h a n d o f t h i s e q u a l i t y c a n b e w r i t t e n as follows

i i i+l
H(i - ^ ) + W i + i - = i r * -
j=i i=i i=i

Thus, (2.1.2) is s h o w n . Note that (p(x) Y[ (l k

j=1 - ipj(x)) = 0 for x G O. Indeed, if

x G s u p p </?, t h e n 1 — ipj(x) — 0 for s o m e j b u t if x £ supp <p t h e n = 0. Thus,

we have p r o v e d (2.1.1). •

DEFINITION 2 . 1 . 1 . A collection = {ft a : a £ A} oí o p e n sets fta c W 1

such

that C UaGA is c a l l e d a n o p e n c o v e r i n g o f A

THEOREM 2 . 1 . 2 Let A be an arbitrary set in W

(Smooth partition of unity). 1

and d be an open covering of A. Then there exists a family ^ of functions I/J in V

'which have the following properties:
(i) 0 < I¡J(X) < 1 for x G W, 1

( i i ) if K (& A, then all except for possibly a large finite number of ip G VP

vanish identically on K,
( i i i ) for each function ip in \& there exists ft G i9 s t ¿ c / ¿ í f t a í s u p p xp C ft,
(iv) / o r every x e A, J2^ * e ^0*0 = !•
DEFINITION 2 . 1 . 2 . A c o l l e c t i o n \I> o f f u n c t i o n s ^ from D having the properties
listed i n t h e a b o v e t h e o r e m is c a l l e d a s m o o t h p a r t i t i o n o f u n i t y for A subordinate
to tf.
 2.2. APPROXIMATION OF FUNCTIONS BELONGING TO W > (n). m p
29
PROOF OF THEOREM 2 . 1 . 2 . T h e p r o o f f a l l s n a t u r a l l y i n t o t h r e e p a r t s . A t the
beginning, suppose t h a t A i s a c o m p a c t s e t . L e t $ b e a n o p e n c o v e r i n g o f A. Then
there e x i s t s e t s Í7¿ G i = 1,..., k such that A C U¿=i ^ « : =
^- B y virtue of
L e m m a 2.1.1 there exists a f u n c t i o n ip i n such that ip(x) = 1 for x £ A.
T a k i n g i n t o a c c o u n t T h e o r e m 2.1.1 w e infer t h a t t h e r e exist f u n c t i o n s ipi i n V(fti)
such that ip — Y2i=i ^i- T h u s , the proof is finished if A is a c o m p a c t set.
We shall n o w consider t h e case, w h e n A is a n o p e n set. P u t

Aj = \x G A :j - 1 < \x\ < j , d i s t ( x , dA) > *

for j = 1,2,... and A-i = A 0 = 0, the empty set. Then Aj is c o m p a c t and

A = Ujli A r F o r each j , the collection

^ := { ¡ 7 H interior A j + 1 D ( A _ i 2 ) c
: Í7 G i ? }

covers A ^ . In accordance with the above statement, there exist functions

Gj,i . .. Qj,kj which are subordinated to the covering tij and ®jv( ) x
— 1 for

x G Aj. P u t

oo fc¿

j = l u=l
Of c o u r s e , o~(x) > 0 for x E A. F o r e a c h a; G i the above s u m consists o f only a
finite number of summands. Let

* = j ^ v : = j = 1,2,..., z/ = 1...

N o t e t h a t t h e c o l l e c t i o n \I> h a s t h e p r o p e r t i e s (i), (ii), (iii) a n d (iv).

F i n a l l y , i f A i s a n a r b i t r a r y set, t h e n A C JB, w h e r e 5 is open. A n y partition
o f u n i t y f o r B w i l l d o f o r A, as well. •

2.2. Approximation of functions belonging to W ' (ft)

rn p
by smooth
functions

In Section 1.12 i t w a s p r o v e d that V(ft) is dense i n W *{Ü)

m
if O W a =
1
nd
1 < p < oo. Example 1.12.1 i n d i c a t e s that D(ft) i s n o t dense i n W > (ft)
m p
i n
general. W e shall show that C°°(ft) is dense i n W ' (ft)
m p
f o r a n y o p e n s e t Í2 a n d
1 < p < oo.

THEOREM 2 . 2 . 1 ( M e y e r s - S e e r i n ) . C°°(f2) zs dense in W ^ (Q) rn p

for any open
set ft and 1 < p < oo.

PROOF. L e t / be i n W (ft)m,p
and e > 0, w e h a v e to show that there exists
I> G
(
C°°(ft) such that ||/ — $|| m } P ) n < e. C o n s i d e r t h e c o l l e c t i o n o f t h e o p e n sets

ft k = í^x E ft : \x\ < k a n d d i s t ( x , dft) > ^ j

for A: = 1, 2 , F o r k — —1,0 we take 0 _ x = O 0 = 0. T h e collection

= v fc = í i f c + i n ( n _ i ) , fc = f c
c
l,2,...}

of o p e n sets T 4 covers O . L e t ^ b e a s m o o t h p a r t i t i o n o f u n i t y f o r ft s u b o r d i n a t e to
d. B y i/jjç, w e denote the s u m o f a large finite number of functions ip, ip E^ whose
 2. LOCAL PROPERTIES OF DISTRIBUTIONS
30

supports are contained i n V. k Then xjj k G T>(V ) k a n d ] C & L i ^Pkix) = 1 for x £ ft.
Since ^ / G W > (ft),
m p
by L e m m a 1.11.1 w e h a v e

0(V> f c /) (aO
e = ( 8 ° W ) ) e ( a O

i f x G ft a n d d i s t ( x , dft) > e. N o t e that

SUpp(^fc/) C ft + € k 2 H (iîjfe_ ) = Vfc <E ÎÎ 2

C

if 0 < e < •^ 2 =
k+ï ~~ • F o r e > 0 i n v i e w o f L e m m a 1.11.2 w e c a n c h o o s e

ek) 0 < e k < •^ 2 s o t n a t

||(^(^/)k-^(^/)|U y 2 ( f c ) <¿.

S i n c e d i s t ( T 4 , <9f2) < e a n d s u p p ( ( ^ / ) e - ^fc/) C V * , therefore

\\d (lP f)e

a
k k -&*iS> f)\\ k LP{a) = \\(d ^ f)) -d (M)\\
a
k ek
a
LP{Q)

= \\d (M)e -d (M)\\

a
k
a
LP(Vk)<^

i f | a | < rn. W e c a n choose e k such that

\\Wkf)e h - < M L , P , n = UW-k/k - ^/IU,„,v f e < ¿ (2.2.1)

because

\\(ipkf)e k ~ ipkf\\m, y P k = m a x ||9 (^/)e a

f c -d (x/; f)\\
a
k LP{Vky
I OÍ [ <m
Let # = J]^ ( 0jt/)e . • O
=1
/
A n any O' O this s u m involves only a large finite number
of n o n z e r o terms. Therefore # is i n C°°(ft). F o rx G 0 ^ w e have

k+2 k+2
f(x) = J2tl>,ix)f(x) a n d * ( x ) = £ ( ^ / ) C i ( x ) .

i=i j=i

Thus
fc+2

Taking into account (2.2.1) w e get

ll/-*l|m > P l n f c < c.

Finally, b y the Lebesgue monotone convergence theorem

11/ - * l k p , n < e. •

REMARK 2.2.1. Let ft = (-1,1), / ( x ) = \x\. O f course, / G W^°°(-l,l).

N o t e t h a t f'(x) = - 1 for x G ( - 1 , 0 ) a n d f'(x) = 1 for x G (0,1). Obviously, there
is n o f u n c t i o n # G C ° ° ( - l , 1) s u c h t h a t ||$' - f \\L°°{-i,i)
f
< e ife < | .
 2.3. RESTRICTIONS OF DISTRIBUTIONS 31
2.3. Restrictions of distributions

L e t ft' be a n o p e n subset of t h e o p e n set ft. F o r A G V'(ft) p u t

A| ,(y>) ~A(<p)
n if peV(tt').

O b v i o u s l y A|Q/ is a d i s t r i b u t i o n o n ft'. T h i s d i s t r i b u t i o n w i l l be called the r e s t r i c t i o n

of A t o t h e o p e n set ft'.

THEOREM
2.3.1. If A £ V(ft) and for each x e ft there exists an open neigh-
bourhood U such that A\u — 0, then A = 0 on ft.
x x

PROOF. L e t p i n V(ft) a n d x G ft, t h e n i n accordance w i t h t h e a s s u m p t i o n

there exists a n o p e n n e i g h b o u r h o o d U X such t h a t A\JJ X — 0. Since supp p is a
c o m p a c t set, therefore one c a n choose a finite n u m b e r of sets U ,..., Xl U ,
Xk ^ e i l
so t h a t supp p C [J U . i=1 Xi B y T h e o r e m 2.1.1 there exist functions pi G V{U ) Xi

such t h a t p — ^2 i==1 Pi- Hence we o b t a i n

k
A(p) = J2A\ (p> )=0. Uxi i •
i=l

W e say t h a t t h e d i s t r i b u t i o n s A i a n d A coincide i n a n e i g h b o u r h o o d of x 2 0

if their difference A i — A vanishes i n this n e i g h b o u r h o o d . W e have p r o v e d t h a t

2

d i s t r i b u t i o n s w h i c h coincide i n the n e i g h b o u r h o o d of every p o i n t of ft are equal

to each other o n ft. T h u s , every d i s t r i b u t i o n is u n i q u e l y d e t e r m i n e d b y i t s l o c a l
properties. T h i s fact c a n be used t o construct a d i s t r i b u t i o n i n i t s entirely o n ft,
w h e n i t is defined o n ft , a G A a n d ft = [j ^
a ft - ae a

THEOREM 2.3.2. Let -d be a collection of open sets ft , a G A and a ft =

{J ft .
aeA a Assume that for each a G A is given a distribution A G V(ft ) a a such
that for two indices a± and c¿2, A (p) = A (p) for p G V(ft f l ft ),
aithen a2 ai a2 there
exists exactly one distribution A G V(ft) such that A\n = A for every a G a a A.

PROOF OF UNIQUENESS. Suppose t h a t there are two d i s t r i b u t i o n s A i a n d A 2

b e l o n g i n g t o D'(ft) satisfying our assumptions. L e t supp p C (Jt=i ®*cx Because of r

T h e o r e m 2.1.1 there exist functions pj G V(ft .) a such t h a t p = Yl^i Vj- Therefore

T h u s , the p r o o f of uniqueness is finished. •

PROOF OF EXISTENCE. P u t as i n the above c o n s i d e r a t i o n

k
A
(^) = E « > ; ) > A
( - )
2 3J

where pj G V(ft ). 0ij W e have t o show t h a t A(p) does not d e p e n d o n representations

of p. T a k e two representations p — Yl^Li a n
d p = Y^jLi • W i t h o u t loss o f
generality we c a n assume t h a t k\ — /c = k. N o t e t h a t 2

k k
A(0) = A aj (p) - tf) = Y, A ,(
Q V j ),
32 2. L O C A L P R O P E R T I E S O F D I S T R I B U T I O N S

where ipj = Lp* — Lp . W h a t is left to show is t h a t the following c o n d i t i o n

2

'3

k k

3=1 3=1

holds. L e t F = U j L i supp y>j y Lpj G V(ü .). a B y L e m m a 2.1.1 there exists a

f u n c t i o n ip e V ( U j L i such t h a t ip(x) = 1 o n F. T h e f u n c t i o n ip c a n be

w r i t t e n as follows ip — Y^j=i w n e r e
rfj £ W e proceed to show t h a t
the i m p l i c a t i o n (2.3.2) holds. A s s u m e t h a t X ^ j L i <Pj — 0- Since F = U ¿ = i supp (pj,
therefore T/^- = Lpj for j = 1 , . . . /c. Hence we have
fc fc k / k \

E (p¿) = E om) = E A
«i (E ^ )
j=i j=i j=i VÎ=I /
fe fe

= E E « i ( ^ ) - A

j=ii=i
N o t e t h a t supp ipiLpj C fi¿ f l Q j . I n accordance w i t h the a s s u m p t i o n we have
A a , - ( ^ ^ j ) = Aa^ipupj), therefore

k k k k

/ fe \
" =0.

It is easily seen t h a t A given b y (2.3.1) is a linear f o r m over V(fl). It r e m a i n s t o

prove t h a t A is continuous. Indeed, let K C O be a c o m p a c t set a n d K C U j L i ^ a - • 3

B y v i r t u e of L e m m a 2.1.1 there exists ip e V ^ U j L i ^ a ¿ ) such t h a t ip(x) = 1 for

x G K a n d 0 < f(x) < 1 if x G U¿=i T a k i n g i n t o account T h e o r e m 2.1.1 we
can w r i t e ip = ^j,^ G V(Q ). aj F i n a l l y we get

( E
k

^
\

=
E
k

A
W w ) =
E
k

A
« i ( ^ ) - ( - - )
2 3 3

3=1 J 3=1 3=1

F r o m this, it follows i m m e d i a t e l y t h a t i f supp Lp C K a n d the sequence (d (p ) v
a
l/

u n i f o r m l y converges to zero for a G N , t h e n ipjLp —» 0 as v -> oo i n P ( O . ) . I n

n
v a j

view of (2.3.3) we infer t h a t k(jpv) —> 0 as z/ —>» oo. T h u s , the p r o o f of existence is

complete. •
E X A M P L E 2.3.1. L e t O = R - {0}. P u t A(<p) = f eiip{x) Q dx for Lp e V(Q).
W e s h a l l now show t h a t there is no d i s t r i b u t i o n A i n V such t h a t A | Q = A . Indeed,
let Lp G £>(0, 2), Lp > 0 a n d f ip = 1. P u t ^ ( x ) = e~^z/(^(z/x), supp 9 9 ^ C (0, 2) C
(-2,2). It is easy to check t h a t tp v —> 0 i n V(—2, 2). T o o b t a i n a c o n t r a d i c t i o n
suppose t h a t there exists a d i s t r i b u t i o n A i n V such t h a t its r e s t r i c t i o n to Q is
 2.4. S U P P O R T O F D I S T R I B U T I O N S 33

equal to A , t h e n we have
2

A(p ) u = j e^~^vp(yx) dx.

o
A n easy c o m p u t a t i o n shows t h a t
2 2

A(p )v = j e< ^p(y)dy> J p(y)dy = 1.

o o
T h i s c o n t r a d i c t s the c o n t i n u i t y of A .

2.4. S u p p o r t of d i s t r i b u t i o n s

Let T G X>'(0). L e t ÎÎ' be a u n i o n of o p e n sets ft C ft, a e A such t h a t T vanishes

a

i n ft . A c c o r d i n g t o T h e o r e m 2.3.1, T vanishes i n ÎÎ'.

a

T h i s justifies t a k i n g the following

D E F I N I T I O N 2.4.1. Iabeld2.4.1 L e t A G V'{ft). T h e n the s u p p o r t of A is the

c o m p l e m e n t of the largest o p e n set ft' C ft w i t h respect t o ft i n w h i c h A vanishes.

N o t e , t h a t under this definition the s u p p o r t of a d i s t r i b u t i o n is always closed.

W e s h a l l u s u a l l y take ft = R . I n this case, the s u p p o r t of a d i s t r i b u t i o n m a y be
n

whole W or it m a y be the e m p t y set. F o r example, supp

1
= R a n d supp 6 = {0}.
Some d i s t r i b u t i o n s are sensible for p b e l o n g i n g to larger test spaces t h a n V(ft).
For e x a m p l e 8{p) is m e a n i n g f u l for p G C ( K ) . W e s h a l l now show t h a t a n a r b i t r a r y
n

d i s t r i b u t i o n A G V'(ft) m a y be extended o n a subspace of C°°(ft). L e t

C^(ft) :={p:p>e C ° ° ( 0 ) , supp pi) supp A (Ê ft}.

N o t e t h a t C^(ft) is a vector subspace of C ° ° ( i î ) .

W e are n o w i n a p o s i t i o n to prove the following

T H E O R E M2.4.1. If A £ V(ft), then there exists exactly one linear form A :

C^(ft) -> C such that
(i) A(p) = A(p) ifpeV{Ü),
(ii) A{p) = 0ifipe C°°(ft) and supp p D s u p p A = 0.

P R O O F . L e t supp A fl supp p = K (s ft- B y v i r t u e of L e m m a 2.1.1 there exists

a f u n c t i o n ip G V(ft) such t h a t 0 < ip(x) < 1 i f x G fl a n d ip(x) = 1 for x G B (K),
e

B (K)
e <s ft for some e > 0 (if K = 0 we take ip = 0). E a c h f u n c t i o n p> G C^(ft)
may be represented as follows p — p\ + <¿>2, where p\ = ipp a n d p 2 = (1 — ip)p>-
N o t e t h a t s u p p A fl supp p 2 — 0- Since y? G C ^ ° ( i î ) , therefore <pi a n d ^ 2 are i n
C ^ ° ( 0 ) , as well. Suppose now t h a t there exists a linear f o r m A : C™(ft) —> C
fulfilling (i) a n d (ii), t h e n A(p) — A(pi) + A(p ) 2 = A{ipp). P u t by definition

\(<p) := A ( ^ ) . (2.4.1)

W e have to show t h a t A(ipp) does not d e p e n d o n ip i f ip G C™(ft). Indeed, assume

t h a t ipi has the some properties as ip, t h e n A(p) — A(ip\p). N o t e t h a t supp[('0i —
ip)p] fl supp A = 0. Therefore A(ipip>) — A(ipp). O n e c a n prove t h a t A is a linear
form o n C^(ft) satisfying (i) a n d ( i i ) . •
 2. LOCAL PROPERTIES OF DISTRIBUTIONS
34

If supp A is c o m p a c t , t h e n one c a n take K = supp A . O f course, C ^ ° ( f î ) —

C°°(ft).

DEFINITION 2 . 4 . 2 . A sequence ((fv), y G C ° ° ( Q ) is s a i d to be convergent i n

v

C°°(ft) i f the sequences (d tp ) converge u n i f o r m l y i n every c o m p a c t set K C ft

a
u

for each a G N . 7 1

O n e c a n observe t h a t i f A has a c o m p a c t s u p p o r t t h e n the linear f o r m A :

C°°(ft) —> C is continuous under the above defined convergence (see T h e o r e m 1 . 1 . 1 ) .
T h u s , we have s h o w n t h a t a n a r b i t r a r y d i s t r i b u t i o n A G V {il) w i t h a c o m p a c t
s u p p o r t m a y be extended b y ( 2 . 4 . 1 ) o n whole C°°(ft) a n d the linear f o r m A is
continuous. M o r e o v e r , it m a y be extended o n C°° (W ) i n the same way. 1
Since
A((f) — A(ip) i f (p G D(fl), the d i s t r i b u t i o n A is also continuous i n the above sense.
N o t e t h a t a sequence ( y v ) , (p u G C°°(ft) converges t o zero i n C°°(ft) iff for each
c o m p a c t set K C ÎÎ a n d m G N, H ^ H m ^ —» 0 as v —» oo (see (1.1.1)). Let

K u — | x : x G ft, \x\ < v a n d d i s t ( x , dft) > —

t h e n ft — U ^ L i T h i s indicates to us t h a t the convergence i n C°°(ft) m a y be

defined b y means of the countable f a m i l y (|| • | | x ) m = 1, 2 , . . . of the s e m i n o r m s
m ) m 5

II • l U ^ m -
N o w , we give a c h a r a c t e r i z a t i o n of the linear continuous forms o n C°°(ft).

THEOREM 2 . 4 . 2 . A linear form A : C ° ° ( 0 ) C is continuous with respect to

C°° (Q)-convergence iff there exist a compact set K C ft, m G N and M > 0 such
that
|AfoO| < M\\<p\\ mtK (2.4.2)
for ip G C ° ° ( f t ) .

PROOF. T h e sufficiency is evident. W e s h a l l n o w prove the necessity. A s s u m e

that A G V'(Vt) a n d t h a t there is not any c o m p a c t set, m G N a n d M > 0 s u c h
that ( 2 . 4 . 2 ) holds. Therefore for each v G N there exists (p G C°°(ft) such t h a t u

|A(<¿v)| > vWtpvWv^Kv- O n e c a n choose Lp so t h a t A ( ^ ) = 1. T h u s , we have

v

ll^i/IU,^ < ¿ for z/ = 1, 2 , . . N o t e t h a t v

+ ~ + + +M <
< ~-

T h i s i m p l i e s t h a t the sequence [}p ) converges to zero i n C ° ° ( 0 ) . T h i s is a c o n t r a -

v

d i c t i o n since A is continuous. •

W e are able to give a c h a r a c t e r i z a t i o n of d i s t r i b u t i o n s w i t h c o m p a c t s u p p o r t s .

T H E O R E M 2 . 4 . 3 . A distribution A e V'(ft) has a compact support iff (2.4.2)

holds for some K, m, M and (p G V(ft).

P R O O F . T a k i n g i n t o account the previous consideration, we o n l y need to show

the sufficiency. Indeed, suppose t h a t supp (p D K = 0, t h e n b y ( 2 . 4 . 2 ) , A((p) = 0 .
Therefore supp A C K. T h i s proves the theorem. •

T h e following t h e o r e m is i m p o r t a n t for a p p l i c a t i o n s .

THEOREM 2 . 4 . 4 . V(ft) is dense in C°°(ft).

 2.5. DISTRIBUTIONS OF FINITE ORDER 35
PROOF. B y L e m m a 2.1.1 there exists a f u n c t i o n ip G V(fl) such t h a t ip (x) = 1 v u

for x G K , v = 1 , 2 , . . . . T a k e / „ = ip f.
v O f course, ) G V(fl). u Note that v

\\f u - fWm.Km = 0 for z/ > m . Therefore f converges to / i n C°°(fl). v •

COROLLARY 2.4.1. If A £ V (fl) has a compact support, then the linear form
A : C°°(fï) —> C, defined by (2.4.1) may be obtained by means of the continuous
extension A from V(fl) onto C°°(fl).

2.5. D i s t r i b u t i o n s of finite o r d e r

Let A G V'(fï). I n accordance w i t h T h e o r e m 1.1.1, for every c o m p a c t set K C fl

there exist a nonnegative integer ra a n d C > 0 such t h a t

|A(y>)| < C\\<p\\ ,K m (2.5.1)

if p G V(fl) a n d supp <p C K. F o r some d i s t r i b u t i o n s the i n d e x ra does not d e p e n d
on K.

DEFINITION2.5.1. A d i s t r i b u t i o n A is s a i d to be of finite order i f there exists

a nonnegative integer m (independent o n K) s u c h t h a t (2.5.1) holds for (p G V(fl),
supp p C K ( i n general C depends o n K). T h e smallest integer m > 0 s u c h t h a t
(2.5.1) holds is called the order of A .

EXAMPLE 2.5.1. L e t A : V(ü) -> C be a l i n e a r f o r m fulfilling the f o l l o w i n g

condition
p>0=> A(p) > 0. (2.5.2)
W e s h a l l n o w show t h a t A is a d i s t r i b u t i o n of finite order. Indeed, let K C fl be
a c o m p a c t set. I n v i e w of L e m m a 2.1.1 there exists ip G V(fl), 0 < ip(x) < 1 a n d
ip(x) — 1 for x G K. L e t p G V(fl) a n d supp p C K, t h e n we have

- | M l o , / r V > ( a O < <p(aO < ||^||O,K^(^)

for x £ fl. F i n a l l y we get
|A(vO| < C7|b||o,K-,
where C — A(ip). N o t e t h a t the D i r a c measure ô satisfies (2.5.2). I n t h i s case,
C — 1 for each c o m p a c t set K CW . 1

EXAMPLE 2.5.2. P u t A = J2„ez D"5„, where 8 (p) = p{y)- It is easily seen u

t h a t the above series converges i n V'(R). T h e d i s t r i b u t i o n A is not of finite order.

EXAMPLE 2.5.3. E v e r y d i s t r i b u t i o n w i t h a c o m p a c t s u p p o r t is of finite order

(see T h e o r e m 2.4.3).

W e are i n a p o s i t i o n to prove the following

THEOREM 2.5.1. Let A be a distribution with the compact support K of order

m. If p £ C°°(W 1
) and its derivatives up to order m vanish on K, then A{p) = 0.

PROOF. T a k e b y definition

77(e) \— m a x sup \d p(x)\.

a

\ \<™>xeB (K)
a
e

Note that if d p a
vanishes o n K, \a\ < ra, t h e n 77(e) —» 0 as e —» 0. L e t x be the
characteristic f u n c t i o n of B¿(K). P u t Xe — X * ^ ! 5 where ip (x) — e~~ ip ( ^ ) , e > 0,
e
n

ip G V(Bi(0)), ip > 0 a n d f ip = 1. It is easy to see t h a t

Xe{x) = 1 if x G B (K)
i and Xe(x) = 0 if x i BM(K).
36
2. LOCAL PROPERTIES OF DISTRIBUTIONS
A n easy c o m p u t a t i o n shows t h a t

M,
d e(x)\
a
X < - ~ for x GR. n
(2.5.3)
e
W e s h a l l n o w estimate \d@(p\. N o t e t h a t

(ei,x)

d <p(x) =
f>
J d <p(T,x ,...,x )dT,
l3+ei
2 n (2.5.4)

(ei,xo)

where x G 9iíT, e i = ( 1 , 0 , . . . , 0) a n d (x,y)

0 = ( x i y i , . . . , x y ). n n I n accordance
w i t h the a s s u m p t i o n we have

d^ (p(x)
+ei
= 0 if xe K

and
| ^ + e
V W | < 77(e) if xeB (K)-K
e when |/3 + e i | < m .
T a k i n g i n t o account (2.5.4) we get

\d (p(x)\
fi
< n(e)e if xe B (K)
e and \(3 + e | < ra.
x

T h e same estimate is o b t a i n e d i f we integrate over Xj, 1 < j < ra, p r o v i d e d t h a t

|/3 + e j | < ra. R e p e a t i n g this process w i t h respect to the second v a r i a b l e we get

\&*(p(x)\ <n(e)e 2
if xeB (K)
e

and |/3 + e i -f- e21 < ra. B y i t e r a t i n g the process a n d i n t e g r a t i n g ra — \(3\ times, we
o b t a i n the estimate

\df*<p(x)\ < r)(e)e -W m

for x G B (K).
e

B y c o m b i n i n g this estimate w i t h (2.5.3) a n d u t i l i z i n g the L e i b n i z f o r m u l a we get

|9*( W e ) ( z ) | < 77(e) ( ^ ) e m

- | / 3 |
M a _ ^ - ^ =M (e)e™-W
V (2.5.5)

for x G M , where M = Y^o</3<a {p) a-{3-

n M
F r o m this i t follows

H X c P l k A < Mr7(e)

if 0 < e < 1 a n d A D B (K) is a n a r b i t r a r y c o m p a c t set.

e Since A is of order m
w i t h the c o m p a c t s u p p o r t K, therefore for some A D B (K) e we have

\A(<pXe)\ < C\\xeV\\m,A < CMrj(e).

In accordance w i t h the definition of A we get

\li(<p)\ < CM (e).

V

Since 77(e) —> 0 as e 0, therefore A(<p) = 0. T h u s , the p r o o f is finished. •

A s a c o r o l l a r y we o b t a i n

THEOREM 2.5.2. A distribution A with the point support {xo} is a finite linear
combination of Dirac's measure concentrated in XQ and its derivatives.
 2.6. CARTESIAN PRODUCTS OF BANACH SPACES 37
PROOF. Suppose t h a t A is of order ra. L e t p G C°°. T h e f u n c t i o n p m a y be
w r i t t e n as follows

<p{*)= £ ^~^(x-x r 0 + R (x) m

\a\<m
and d Rm(xo)
a
= 0 i f \a\ < ra. I n accordance w i t h the above t h e o r e m we infer t h a t

A(vO = £ a (-l)l l9>(^o)

a
a

I a j <m

for tp £V, where a a = — ^ — A ( ( - — x ) ) . F i n a l l y we have

i
0
a

A= £ a £><M .
a Xo •
ja I <ra
2.6. C a r t e s i a n p r o d u c t s o f B a n a c h spaces

Let Ej, j — 1 , . . . , n be a B a n a c h space over C . W e r e c a l l t h a t the C a r t e s i a n

product E — X " Ej of the vector spaces Ej consists of a l l vectors x — ( x \ , x ) ,
= 1 n

where Xj G Ej. T h e element Xj is called the j - t h c o m p o n e n t of x. W e i n t r o d u c e

i n the set E a d d i t i o n a n d m u l t i p l i c a t i o n b y scalar of elements b e l o n g i n g t o E as
follows
x + y = (x 1 + y u ...,x n + y)
n

if x = O i , . . . ,x ),n y = (2/1,... ,2/ ), n

Ax = ( A x i , . . . , A x ) , n

A G C . O b v i o u s l y I£ is a vector space under these operations. T h e m a p ITj f r o m E

onto Ej defined b y Hj(x) — Xj is called the j - t h p r o j e c t i o n . T a k e

Ej — {XJ : Xj = ( 0 , . . . , 0, Xj, 0 , . . . , 0), Xj G Ej}.

N o t e t h a t 11^ establishes a n i s o m o r p h i s m from Ej onto Ej (Iij[xj) — Xj). E v e r y

element x G E m a y be w r i t t e n i n the f o r m x = x\ + • • • + x . T h e representation n

of each element x G E i n t h i s f o r m is u n i q u e . T h i s means t h a t the space E is the

direct s u m of the subspaces Ej j = 1 , . . . , n. W e s h a l l w r i t e E = Ej- ^ E n

one c a n i n t r o d u c e two types of n o r m s

W M\P = I N I ? ) , 1 <P < oo,

P

(ii) Halloo = m a x j i . . = v ) n \\xj\\j, p — oo,

where is the n o r m of Xj i n Note that

Pill? = Wttj^xMp = \\ j\\j, x

1 < p < oo.
T h i s means t h a t the m a p II j is a linear i s o m e t r y from Ej onto Ej. O b v i o u s l y , i f Ej
(j = 1 , . . . , n) are the B a n a c h spaces, t h e n E — X ™ Ej is also the B a n a c h space = 1

under the n o r m s || • 1 < p < oo.

LEMMA 2.6.1. Let K be a linear continuous form on E — X " Ej with respect = 1

to the norm \\ • \\ , 1 < p < oo, then there exist continuous linear forms Aj on Ej
p

such that
n
A(s) = $>,•(*,•), (2.6.1)
3=1
where x = ( a ? i , . . . , x ) , Xj G Ej. n
 2. LOCAL PROPERTIES OF DISTRIBUTIONS
38

PROOF. Since E = 0 " = 1 E h therefore A(x) = £" = 1 A(XJ). N o t e t h a t A(XJ) =

A(UJ (xj)).
1
Since A is a continuous linear f o r m o n E, therefore the m a p Ej 3 Xj —>
(A o UJ )(xj)
1
is a continuous linear f o r m o n Ej. W e take A j = A o UJ . 1
Finally
we o b t a i n (2.6.1). •

2.7. S o m e l o c a l r e p r e s e n t a t i o n s of d i s t r i b u t i o n s

W e start w i t h presenting some i n f o r m a t i o n c o n c e r n i n g c o m p l e x measures.

DEFINITION 2.7.1. L e t ft be a n o p e n set i n R . W e say t h a t cp G C ( O ) i f it is

n
0

continuous i n ÍÍ, a n d for every e > 0 there exists a c o m p a c t set K C ft such t h a t

\<p(x)\ < e for x G ft - K.

It is easily seen t h a t Co(ft) is a vector space over the field of the c o m p l e x

numbers. P u t
IMICo(O) •= SUp|</?(z)|.

A n easy c o m p u t a t i o n shows t h a t Co(ft) is complete under the n o r m || • ||c (fi)- 0

W e shall n o w recall t h a t C [ft) denotes the c o l l e c t i o n of a l l continuous c o m p l e x

c

functions o n ft whose s u p p o r t is c o m p a c t (see S e c t i o n 1.3).

W e are able to prove the following

THEOREM 2.7.1. C (ft)

c is dense in C (ft).
0

PROOF. G i v e n / G Co(ft) a n d e > 0, t h e n there exists a c o m p a c t set K C ft

such t h a t | / ( x ) | < e outside K. L e m m a 2.1.1 gives us a f u n c t i o n ip G C (ft) s u c h c

t h a t 0 < ip(x) < 1 for x G Ü a n d ip(x) = 1 o n K. P u t g = fip. T h e n g G C (ft) c

a n d 11/ — g\\c (Q) < - T h u s , the p r o o f is

0
E
finished. •

L e t A I be a cr-algebra i n ft ([23]). A countable c o l l e c t i o n {E„} of the m e m b e r s

of M is called a p a r t i t i o n of E i f E{ D Ej = 0 whenever i / j a n d E = ( J ^ L i

DEFINITION 2.7.2. A f u n c t i o n \i : M -> C such t h a t

oo

for every p a r t i t i o n {E } v of E is called a c o m p l e x measure o n A4.

W e shall n o w describe some general properties of the c o m p l e x measures. Since

the u n i o n of the sets E is not changed i f the subscripts v are p e r m u t e d , therefore
v

the series (2.7.1) must be convergent absolutely. W e are n o w i n a p o s i t i o n t o define

a finite p o s i t i v e measure o n M, p u t t i n g
oo
M(£?):=sup£|Ai(2U, (2-7.2)
I/=l

the s u p r e m u m b e i n g t a k e n over a l l p a r t i t i o n s {E } u of E.

DEFINITION 2.7.3. T h e measure is c a l l e d the t o t a l v a r i a t i o n of the m e a -

sure ¡i.
 2.7. SOME LOCAL REPRESENTATIONS OF DISTRIBUTIONS 39
It is k n o w n t h a t there exists a measurable f u n c t i o n h w i t h respect to cr-algebra
M such t h a t \h(x)\ = 1 for x G a n d ¡JL(E) — f hd\fi\ for E G M. It is therefore
E

reasonable to define i n t e g r a t i o n w i t h respect to a c o m p l e x measure ¡i b y the f o r m u l a

Jfd¡i =J fhd\n\.
E E

W e s h a l l n o w remember the R i e s z representation t h e o r e m for linear c o n t i n u o u s f o r m

on CQ(^), where Q is a n o p e n set.

THEOREM 2 . 7 . 2 . To each linear continuous form A on C o ( í í ) there corresponds

a unique complex regular measure ¡i such that

A ( / ) = J fdp = J fhd\»\

n n
for f G Co(íl), where is a positive finite regular Borel measure and h is a Borel
function such that \h(x)\ = 1 for x E ft.

PROOF.. See [23]. •

THEOREM 2 . 7 . 3 . Let Q' Ü and A <E
V(fl), G then there exist m G N and
complex regular Borel measures fi on Q ' such that a

\oc\<m ,n

if supp If C Q . F

PROOF. L e t ffêi] a n d A G V(Çl), t h e n a c c o r d i n g to T h e o r e m 1.1.1, ra GN

a n d C > 0 exist such t h a t
_ | A M | < <7|ML,ÏF
if supp LÇ C Í2'. C o n s i d e r the C a r t e s i a n p r o d u c t

c 0
w
(ir)= x co(fi'), JV = ]T i-
|a|<m ^

T h e elements of CQ {Q!) w i l l ,be denoted b y (</? ). T h e c o m p o n e n t of (v?«) corre-

a

s p o n d i n g to subscript a w i l l be denoted by ip . L e t us endow the C a r t e s i a n p r o d u c t

a

C 7 ^ ( Q ' ) w i t h the n o r m
ll(^«)llc "(fi') = ll(<A*)||oo
0
:

(see (ii) i n S e c t i o n 2 . 6 ) . A c c o r d i n g to the previous considerations we infer t h a t

OQ (£1') is a B a n a c h space w i t h respect to this n o r m . W e s h a l l n o w define a m a p
1

JC f r o m D ( Q ' ) i n t o Ctf as follows

If we endow the set V(Q!) w i t h the n o r m || • | | M QT, t h e n the m a p K is a n i s o m o r p h i s m

from V(n') onto K{V(Ü')) C C ^ ( í í ' ) , moreover

I M L J F = II ( ( - i ) H
9>) \\c»w

C o n s i d e r a new f u n c t i o n a l A : JC(V(Q')) —> C defined b y means of the f o r m u l a

 ( ( ( - l ) H d » ) :=A(<p).
40
2. L O C A L P R O P E R T I E S O F D I S T R I B U T I O N S

T h e definition of Á is meaningful, because t h e m a p K is a n i n j e c t i o n of V(Q!) i n t o

CQ (ft'). Since JC is a n isometry i s o m o r p h i s m from V{Q!) onto K,(V(Q,')), therefore
A is a continuous linear f o r m o n JC(V(ft')). B y the H a h n - B a n a c h theorem the
linear f o r m A m a y be extended from fc(V(fl')) o n CQ so t h a t

\k(<p ))\a < c\\(<p )\\

a cm

for (ip ) G 0^(0.').

a B y v i r t u e of L e m m a 2.6.1 a n d T h e o r e m 2.7.2 we o b t a i n

\a\<m , n

where fi a are c o m p l e x regular B o r e l measures o n ÎÎ'. Hence, for ( ( - l ) H d > ) G

JC(V(n')) we have A ( ( - l ) l ¡ < 9 » = A(<p). T h u s , finally we o b t a i n
a

\a\<m Qf

T h i s statement finishes t h e proof. •

 CHAPTER 3

TENSOR PRODUCTS AND CONVOLUTION

PRODUCTS

3.1. R e g u l a r i z a t i o n of d i s t r i b u t i o n s

In this chapter we s h a l l assume, for s i m p l i c i t y of consideration, t h a t the d i s t r i b u -

tions are defined o n whole W . L e t p be i n £>(M x W ) a n d A G £ > ' ( R ) . L e t us
1 m 1 m

consider the following m a p

R n
3y-^ AM-,?/)).

P u t b y definition

THEOREM 3 . 1 . 1 . If A e £ > ' ( R ) and <p G V(W

m n
x W ), 1
then ip is in V(R ).
n

PROOF. Let supp p C I m x J , where I

n m and J n are c o m p a c t intervals i n
W n
a n d R , respectively.
n
Since the f u n c t i o n cÇV('> •) is u n i f o r m l y continuous o n
M M
x R , therefore
n

#1*1
p>{x,y + h) - -—p{x,y) aeN 71

converges u n i f o r m l y to zero as h 0 w i t h respect to x G I m for fixed y. F r o m this,

it follows t h a t A(p(-,y + /&)) tends to A(<p(-,2/)) as h -> 0. It is easily seen t h a t
supp ip C I - n It remains to prove t h a t V is s m o o t h . T a k i n g i n t o account the above
c o n s i d e r a t i o n one c a n r e m a r k t h a t \h~ x
(p{-,y + hei) — p(-,y)) — tends
to zero as h - » 0 i n £ > ( R ) for m
fixed Therefore

as h - » 0. T h i s gives us

T h e c o n t i n u i t y of ô * ^ is obvious. B y i n d u c t i o n we have
6

T h u s , the p r o o f is finished.
•
L e t (p be i n V a n d A be i n V'. P u t b y definition

A^x) := A(<p(x - •)) (3.1.1)

42 3. TENSOR PRODUCTS AND CONVOLUTION PRODUCTS
for x G W . W e s h a l l now present the following t h e o r e m w h i c h is p a r a l l e l w i t h
1

T h e o r e m 3.1.1.

THEOREM 3.1.2. If A eV and tp eV then A^ is a smooth function and

Q\a\ / Q\a\

Since the p r o o f of this t h e o r e m is the same as the p r o o f of T h e o r e m 3.1.1 we

s h a l l not give it here.

DEFINITION 3.1.1. T h e f u n c t i o n A<¿> given b y (3.1.1) w i l l be c a l l e d the regular-

i z a r o n of A .

O f course, i f A has a c o m p a c t s u p p o r t t h e n A ^ is i n V for each ip i n V.

EXAMPLE 3.1.1. F o r A = Ô, where S is the D i r a c measure c o n c e n t r a t e d at zero

we have 6 (x) =
ip (p(x).

EXAMPLE 3.1.2. If A = / a n d / G L¡ (R ),
oc
n
then

A (x)
(p = J f(y)(p(x-y)dy = (f*<p)(x).

T h i s justifies the following n o t a t i o n A *<£>:— A ^ .

T h e following t h e o r e m is n o w a n easy consequence of the preceding t h e o r e m .

THEOREM 3.1.3. If A is in V and tp belongs to V, then

(D Aa
* tp){x) = ( A * d ip)(x) a
= d (A a
* (p)(x). (3.1.2)

PROOF. I n accordance w i t h T h e o r e m 3.1.2 we have

d {A
a
* tp)(x) = A{d ip(x a
- •)) = ( A * d cp){x).a
(3.1.3)

Note that

(D A
a
* <p)(x) = D°A(<p(x - •)) = ( - l ) | a |
A, (f^(* - V))

= A y (f^(* - </)) = A(d^(x - •)). (3.1.4)

In v i e w of (3.1.3) a n d (3.1.4) we have (3.1.2). •

T h e next t h e o r e m is very i m p o r t a n t for a p p l i c a t i o n s .

THEOREM 3.1.4. If (p and ip belong to V but A is in V, then

(A *<£>)* ip — A * (<p * ip).

PROOF. P u t
a (t)
h := ] T <p(t - hv)iP(hv)h , n
(3.1.5)
 3.1. REGULARIZATION OF DISTRIBUTIONS 43
where v = ( z / . . . , i / ) , z/¿ G Z , h > 0 a n d ZJJ = {zv : hv G supp?/;}. N o t e t h a t
l 5 n

cTh(t) is a R i e m a n n s u m of the c o n v o l u t i o n <p * ip at p o i n t t. D i f f e r e n t i a t i n g a - t i m e s

e q u a l i t y (3.1.5) we o b t a i n

d a (t)
a
h = ] T 9 <P(t - a
hiy)ip(hu)h . n
(3.1.6)

B y v i r t u e of the m e a n value t h e o r e m we get

g\ \
a
$ H
— ^-r)^(r)ar= £ — - 8»^(e»fc , (3.1.7)
n

where 0 < 6 „ < 1. B y (3.1.6) a n d (3.1.7) we get

—<p(t - hv)ip(hv) - —(p(¿ - © hv)ip(@ hv)

u v

P u t b y definition g(t,r) := ^ ( t — r)ip(r). It is easy t o check t h a t

supp g C (supp <¿> + supp ^ ) x supp ^ .

Therefore the f u n c t i o n g is u n i f o r m l y continuous i n R n

x R . M o r e o v e r , there exists
n

M > 0 such t h a t X ^ e Z £ h n
< M for h > 0. F r o m this, it follows t h a t

$ H Q\C*\
< € (3.1.8)
â F ^ ( ' ) - â F ^ ^ ) C )

for s m a l l / i a n d í G R . It means t h a t ^^cr (t)

n
h tends u n i f o r m l y t o §pr(<¿> * ip)(t)
as / i —» 0 i n R . O f course, ^~(Jh(x
n
- t) also tends u n i f o r m l y to f ^ r ( ^ * 0)(x - ¿) /

as h —y 0 w i t h respect to ¿ G R n
a n d fixed x . It is easily seen t h a t

supp<Jh{% — ') C x — (suppip + supp-0).

T a k i n g i n t o account these statements we infer t h a t

A(a (x
h - •)) -> A((v? * ^)(ar - •)) = ( A * (p * ^ ) ) ( x )

as / i —» 0. O f course,

K{a {x h - •)) = E A
( M * - •- hv))^{hv)h . n

N o t e t h a t the expression o n the right side of this e q u a l i t y is a R i e m a n n s u m of the

integral

j (A * p){x — r)ip(r)dr.

Therefore A(ŒH(X — •)) tends to ( ( A *<£>)* ^ ) ( x ) as / i - » 0. T h i s completes the

proof. •
 3. TENSOR PRODUCTS AND CONVOLUTION PRODUCTS
44

W e now conclude our c o n s i d e r a t i o n w i t h a n interesting a p p l i c a t i o n of the pre-

ceding theorem. A s usually, let ^ G V, supp ip C Bi(0) a n d J ip = 1. P u t

THEOREM 3.1.5. If A e Vthen A*ip e tends to A as e 0 in V.

PROOF. W e have to show t h a t

l i m / ( A * ip )(x)(f(x) dx — A(tp)
J
e

for ip G V. Indeed, i f we p u t Lp(-) :— </?( — •), t h e n

j ( A * 0 )(a:)(p(x)
,
e = J (A* ip )(x)(p(0 e — x ) dx

= ((A*^ )*^)(0). E

B y T h e o r e m 3.1.4 we have

J(A * V; )(x)<¿>(x) d x = A ( ( ^ * £ ) ( 0 - • ) ) •
e e

Because d (<p * ip) = d (p * ^ , t a k i n g into account T h e o r e m 1.3.2, we infer t h a t

a a

ip * (p tends to
e as e —> 0 i n V. Therefore A(ip e *</?)--> A(<£( —•)) = A(y?) as e -> 0.
F i n a l l y we have

J (A* ip )(x)<p(x) dx -> A(<p)

e

as e -> 0. •

T h i s t h e o r e m tells us t h a t the space C ° ° ( R ) is dense i n n

V'(R ).
n

3.2. A c h a r a c t e r i z a t i o n of c o n v o l u t i o n o p e r a t o r s

L e t A be i n V a n d (p be i n V. W e have p r o v e d t h a t A * (p is infinitely d i f f e r e n t i a t e
in R . P u t by definition
n

U\(jp) := A* <p.

THEOREM 3.2.1. The map U\ : V —> C°° is a continuous linear operator and

A*(T p)h = T (A*Lp)

h (3.2.1)

for every h G W . 1

PROOF. O f course, UA is a linear operator from V into C°°. Note that

( A * ( r ^ ) ) ( x ) = ( A * <p(. - h))(x) = A{<p{x h))

= r {A((p(x
h - •))) = r {A h * tp){x).

Therefore, we o n l y need to prove t h a t UA is continuous. L e t cp be i n V. Assume

t h a t supp (p C B (0).r F i x a closed b a l l BR(0). Obviously

B (0)
r + B (0)cB {0).
R r+R
 3.3. TENSOR PRODUCT OF DISTRIBUTIONS 45
Since A is i n V, therefore there exist a nonnegative integer m a n d M > 0 s u c h t h a t

\d (A
a
* <p)(x)\ = | A ( 0 < V ( * - - ) ) l < M | | ô > | | m ^

for x G BR(0). T h i s i m p l i e s t h a t U\ is continuous. T h u s , the p r o o f is finished. •

W e s h a l l n o w show t h a t this t h e o r e m a d m i t s a converse t h e o r e m .

THEOREM 3.2.2. If U : V —» C°° is a continuous linear operator and Ur x —

rU
x for x G MP, then there exists a distribution A such that Up> = A * p for p> G V.

PROOF. Define a new m a p as follows

ip - > (L^)(0),

where p(x) = p(—x). It is easily seen t h a t this m a p is a d i s t r i b u t i o n . P u t A(p) :=

(Up)(0). It remains to prove t h a t (Up)(x) = ( A * p)(x). Indeed

(U<p)(-h) = Th(U<p)(0) = U(r <p)(0) h = Afar) - A(^(-- h))

= A(<p(---h)) = (A*<p)(-h).

P u t x — — / i , t h e n (Up)(x) — ( A * p)(x). T h i s completes the proof. •

3.3. T e n s o r p r o d u c t of d i s t r i b u t i o n s

L e t / a n d g be i n C(W ) n
a n d C ( R ) , respectively. P u t b y definition
n

if®g){x,y) := f{x)g(y)

for x G M a n d y G W . T h e f u n c t i o n / (g) # is called the tensor p r o d u c t of / a n d

m 1

r;. F o r a r b i t r a r y functions and b e l o n g i n g to V(W ) and D ( R ) , respectively n n

we have

J (f ® g)(x,y)p (x)p (y)dxdy

1 2 = J f(x)p (x)dx
1 J g{y)p (y)
2 dy. (3.3.1)
l m
x l n
Rm
Rn

E q u a l i t y (3.3.1) defines a d i s t r i b u t i o n i n R m
x R . T h i s o b s e r v a t i o n suggests to us
n

the following

THEOREM 3.3.1. 7 / 5 G 7 > ' ( R ) and T G V(R ), m n

then there exists exactly one
distribution K in £ > ' ( R ) such that m + n

K(p ®p )
1 2 = S(p )T(p )
1 2 (3.3.2)

for p 1 G P ( R ) and p> G £ > ( R ) .

m
2
n
Moreover,

K(<p(., •)) = S (T (p(x,y))) x y = T (S (p(x,y)))

y x (3.3.3)

forpeV(W n
xR ). n

PROOF. W e start w i t h the p r o o f of uniqueness. N o t e t h a t we o n l y need t o

show t h a t i f
K{p ®p )
1 2 = S{p )T{p )=Q
1 2
46
3. T E N S O R P R O D U C T S A N D C O N V O L U T I O N PRODUCTS

for <p G V(W )x

n
a n d ip G £ > ( R ) , t h e n K(ip) = 0 for a l l ip G î > ( R
2
n m
x R ) . L e t ip
n
t

and ^ 2 be i n £ > ( R ) a n d D ( R ) , respectively a n d /

m n
R m ^ i = / „ fa = 1. P u t b y
R

definition

F u r t h e r , suppose t h a t K belongs to D ' ( R m + n

) a n d (3.3.2) holds. O f course,
€ -m^ i (^ij G D ( R ) for each x G Rm m
and e ~ ^ n
2 ( V ) G
^ ( ^ ) for each y G R .
n n

Because of the a s s u m p t i o n , note t h a t

(K * ^ ) ( x , 2/) = K (e~ fa
e
m
( — ) e""^ ^^
e / \ e

= 5 (c-^x [ )T [e-> 2 ( ^ — ) ) = 0

for x G R m
a n d y G R . I n v i e w of T h e o r e m 3.1.5
n

0 = J (K * fa)(x, y)<p(x, y)dxdy •))

as e —>• 0. T h i s statement finishes the p r o o f of uniqueness.

W e are left w i t h the task of d e t e r m i n i n g K. I n order to do it assume t h a t </?(•, •)
is i n D ( R m
x R ) a n d supp(/?(•, •) C I
n
m x (see S e c t i o n 3.1). P u t b y definition

^ ( x ) := T(y>(a;, •))

and

My) ==%>(•>»))•
T a k i n g i n t o account the p r o o f of T h e o r e m 3.1.1 one c a n observe t h a t ipi is i n V(I ) m

a n d ip2 is i n V(I ). n Moreover,

d\*\
« M * ) = T [ Q ¿ ; < P ( X , . ) ) and — y ) = , ^ ^ y ) J
dx'
M S

Since S e V ( R ) a n d T e V ( R ), therefore there exist non-negative integers m i

m n

a n d ra 2 as w e l l as C\ a n d C 2 such t h a t

|S(tfi)l<C'iM'i||,» ,/ 1 m (3-3.4)

for G P ( K ) i f supp V i C J
m
m and

| r ( V ) | < C U \\ ,i
2 2 2 m2 n (3.3.5)

for (p G D ( R ) i f supp ^ 2 C I - F i n a l l y we have

2
n
n

\S (T (ip{x,y))\
x y < CiC ||^(-,-)llm
2 1 + m ,/ x/
2 m n (3-3.6)

and
\T {S {<p(x,y))\
y x < C C |b(T)IU
1 2 1 + m ,/ x/
2 m n (3.3.7)
for ip G V(W n
x R ) i f supp<¿>(-, •) C I
n
m x J . O f course, the m a p s
n

—> S'a; ( T „ ( ( x , y ) ) )
v and <p^T {S {ip(x,y)))
y x (3.3.8)
 3.4. DIFFERENTIATION AND SUPPORT. 47
are linear forms o n V(R m
x l n
) . B y v i r t u e of (3.3.6) a n d (3.3.7) we infer t h a t the
forms (3.3.8) are continuous o n V(R m
x R ).
n
Let

K { ^ ) )x ^ S {T (ip{x,y)))
x y

and

K (p(- )):=T (S (p(x,y)))

2 r y x

for p G V(R m
xR ). n
Note that

K^px <g> p> ) = K ((pi 2 2 <g> p ).

2

A n a l y s i s s i m i l a r to t h a t i n the p r o o f of uniqueness shows t h a t for a n a r b i t r a r y

p G V{R m
x R ) , (K
n
x - K )((p(-,
2 •)) = 0. P u t

K~K!= K. 2

T h u s , we have p r o v e d (3.3.2) a n d (3.3.3). •

A f t e r the above considerations we s h a l l n o w give a definition of the tensor

p r o d u c t of two d i s t r i b u t i o n s .

DEFINITION 3.3.1. L e t S a n d T be i n V'(R ) n

a n d £ > ' ( R ) , respectively. T h e n
m

the linear f o r m K defined above belongs to V (R f m

x R ) a n d fulfils b o t h (3.3.2)
n

a n d (3.3.3). T h e d i s t r i b u t i o n K is called the tensor p r o d u c t of S a n d T .

T h e tensor p r o d u c t of S a n d T is u s u a l l y denoted b y S 0 T . F r o m (3.3.2) a n d

(3.3.3) we o b t a i n i m m e d i a t e l y the following theorems:

THEOREM 3.3.2. If S G D ' ( R ) and T G V'(R ),

m n
then

5®T = T®SGD'(lR m + r i
).

THEOREM 3.3.3. If Q G £ > ' ( R ) , S G V(R ) n i n2

andT e V'(R *), n
then

(Q®S)®T = Q®(S®T)e x> (R , ni+n2+riB

).

3.4. Differentiation and support of tensor product

L e t a = ( a i , . . . , a ) a n d ¡3 = ( f t , f 3 ) . P u t (a,/3) = (a
m n u a , m ft,/3 ).
n We
know that if / G C ° ° ( R ) and # G C ° ° ( R ) , then
m n

for every a G N a n d /3 G N . W e s h a l l now show t h a t the same f o r m u l a holds for

m n

the tensor p r o d u c t S ® T of the a r b i t r a r y d i s t r i b u t i o n s S a n d T.

THEOREM 3.4.1. If S G £ > ' ( R ) andT m

e V'(R ), n
then

D^Dl(S ^T )(p(x,y))
x y = (D^S 0DlT )(p(x,y)).
x y (3.4.1)
 3. TENSOR PRODUCTS AND CONVOLUTION PRODUCTS
48

PROOF. Indeed, we have

D5DÇ(S ®T )<p(x,y)
x y = D° ((-l) (S ®T ) w
x y ( ~ ^ ^ v )

= D° ( ( - 1 ) ^ [ S x ( r y [ ^ ^ y )

= D S (D^T (^(x,y)))
a
x y

= (
" 1 ) H
^(l^ ( j D
^ ^' ( y ) ) )

H - i ) | a |
î . ( ^ ( ^ , y ) ) ^

= D^S (D^T (^(x,y)))

x y

= (D S®DPT)ip(-,-).
a

T h u s , the p r o o f is finished. •

THEOREM 3.4.2. If S and T are in VCR ) 171

and V'(R ) n
respectively, then we
have
s u p p ( S <g> T ) = supp S x supp T . (3.4.2)

PROOF. F i r s t of a l l we show t h a t supp S x s u p p T C supp(S ® T). Indeed,

suppose t h a t (xo, yo) G supp S x supp T a n d t h a t U(x$, yo) is a n o p e n n e i g h b o u r -
h o o d of (#o, 2/o ) i n R m
x R , t h e n there exist o p e n neighb our ho o ds UI(XQ)
n
of XQ i n
R M
a n d J7 (yo) of y
2 0 in R n
such t h a t U^XQ) X í / ( y o ) C U(x ,y ). 2 0 0 T a k e the t w o
functions <pi G P ( L ^ i ( x ) ) a n d </? G V(U2(yo))0 2 so t h a t S ' ( ^ i ) T ( ^ ) ^ 0. O f course,
2

( 5 <g> T)(v?i 0 <¿>) / 0. Therefore supp 5 x s u p p T C s u p p ( S <g> T ) .

2

W e s h a l l now show the converse i n c l u s i o n . L e t </?(•, •) be i n P ( R m

x R ) and
n

suppose t h a t supp ip D (supp S x supp T) = 0. T h e n there exists a n o p e n set U i n

R M
such t h a t supp S C U a n d supp <¿?(x, •) D supp T = 0 for x G (7. Therefore

:=TWx,.))=0

for x G U. T h u s , we have 5 ( ^ ) = S (T (ip(x,y))) = 0. It means t h a t ( S (g>T)<¿?(-, •)

X y

vanishes for 99 i n D ( R x R - (supp S x supp T ) ) . T h i s statement finishes the

m n

proof. •

3.5. T h e t h e o r e m of kernels

Let g be i n C ( R m
x R ). n
P u t b y definition

(Kip)(x) := Jg(x,y)ip(y)dy

for ip G V(R ). It is easy to verify t h a t the m a p p i n g (p -> Kip f r o m £ > ( R ) i n t o

n n

C(R ) is continuous. W e w i l l consider the b i l i n e a r f o r m (Kip, ip) o n £ > ( R ) x

m
V(R ) n rn

defined b y

(Kip,ip) := J(K(p)(x)ip(x)dx = J Jg(x,y)ip(y)ip(x)dy dx, (3.5.1)

 3.5. THE THEOREM OF KERNELS 49
where p G V(R ) a n d ip G V(R ).
n
It is clear t h a t this b i l i n e a r f o r m is continuous.
m

S i m u l t a n e o u s l y we c a n observe t h a t the m a p p i n g

p®ip-+ {KLp,ip) (3.5.2)

is the r e s t r i c t i o n of the regular d i s t r i b u t i o n K generated b y g o n the set of a l l g

functions p ® ip, where p G T > ( R ) a n d ip G V(R ). These statements give us the

N m

following r e l a t i o n
{K<p,iP) = K (p®iP) g (3.5.3)
for p G V(R ) a n d ip G V(R ).
n
O n e c a n ask whether for every linear continuous
rn

operator JC : V(R ) -> £ > ' ( R ) there exists a d i s t r i b u t i o n K G £ > ' ( R

n M
) such t h a t M + N

(JCp,ip) = K(p®iP) (3.5.4)

for p G V(R ) a n d ^ G £ > ( R ) , where (/Cp,V>> denotes the value of JCp for the
n M

argument ip. T h e next t h e o r e m w i l l t e l l us yes.

DEFINITION 3 . 5 . 1 . We shall call formula (3.5.4) the Schwartz representation of

the operator JC. T h e d i s t r i b u t i o n K w i l l be called the kernel of K.

THEOREM 3 . 5 . 1 (Schwartz kernel theorem).

(i) Every distribution K in P ' ( R ) determines a linear operator M + N
JC :
V(R ) —>• V^R™) given by ( 3 . 5 . 4 ) . This operator is continuous
n
in the
following sense: if p - > 0 in V(R ) then K<p 0 in
v V(R ). n
u
m

(ii) For every continuous operator JC : V(R ) —» V(R ) n M

¿/¿ere exists exactly
one distribution K in V'(R x R ) 5 0 ¿/ia¿ ( 3 . 5 . 4 ) m N
holds.

PROOF OF PART (i). L e t K be i n £ > ' ( R M + N

). It is easily seen t h a t the m a p p i n g
ip — » K(p ® ip) is a continuous linear f o r m o n X > ( R ) for fixed </? i n £ > ( R ) . M N
We
are now i n a p o s i t i o n to define a linear operator K f r o m V(R ) n
into X> ( R ) t a k i n g
7 M

JCp(ip) = K(p <g) ip). W e have to show t h a t the m a p p i n g p —>> /C^> f r o m P ( R ) i n t o N

'D^R 7 7 1
) is continuous. I n order to do it we have to verify t h a t i f p v —» 0 i n D ( R ) , N

t h e n for fixed ip i n P ( R ) , K(p M

u (g) -0) - > 0 , too. T h i s fact is evident. T h i s finishes
the p r o o f of p a r t (i). •

N o t e t h a t the m a p p i n g

X > ( R ) x V{R )
N m
3 (ip,iP) -> K(ip®iP)

is a b i l i n e a r continuous f o r m (see T h e o r e m 1 . 1 . 1 a n d A p p e n d i x ) .

PROOF OF PART ( i i ) . W e s h a l l now prove t h a t for every continuous linear op-

erator JC : V(R ) n
-> V (R )
f m
there exists e x a c t l y one d i s t r i b u t i o n K i n £ > ' ( R M + N
)
such t h a t ( 3 . 5 . 4 ) holds. A n a l y s i s s i m i l a r to t h a t of the p r o o f of T h e o r e m 3 . 3 . 1
shows t h a t there exists at most one d i s t r i b u t i o n K i n D ' ( R M + N
) satisfying ( 3 . 5 . 4 ) .
Therefore, it remains to prove the existence of such d i s t r i b u t i o n . It is easy to see
t h a t the m a p p i n g
V(R ) n
x V(R ) m
3 (p,ip) (JCip^ip)
is b i l i n e a r a n d separately continuous. L e t Qi (Ë R M
and 0 2 <s R . L e t
N
CQ°(ÍII)
a n d C7Q°(Í2 ) be defined as follows
2

Cg°(ííi) = W € C°°(ííi) : d p a
G Co(fii), a EN }
11
50 3. T E N S O R PRODUCTS ANDCONVOLUTION PRODUCTS

and
c%°{n ) 2 = We C°°(tt ) 2 : d <pG C o ( í í ) ,
a
2 OLGN }. 71

O f course, the sets CQ°(QI) a n d CQ°(Ü ) 2 are vector spaces. L e t us equip these
vector spaces w i t h the norms || • | | ^ m a n d || • | | Q , rn = 0 , 1 , 2 , . . . , respectively
m 2

(see S e c t i o n 1.1). C Q ° ( O I ) a n d C§°(fl ) 2 are easily seen to be complete. B y v i r t u e of

T h e o r e m 2.7.1 we infer t h a t P ( Q i ) a n d V{tt ) 2 are dense i n C £ ° ( f i i ) a n d C £ ° ( Q ) , 2

respectively. E v e r y linear f o r m A G D ' ( i î i ) m a y be b y c o n t i n u i t y p r o l o n g e d f r o m

£>(Qi) onto Q ^ O i ) . T h u s , we c a n regard the operator JC : V(W ) N
-> V'(R ) N
as a
linear continuous operator from V(ft ) 2 into (CQ°(ÜI))\ where (CQ°(ÜI))' consists
of a l l linear continuous forms o n CQ°(ÍÍI). Since ( C o ° ( í í i ) ) ' is sequentially c o m p l e t e
(see T h . A . l ) , therefore the operator JC : V(Ç1 ) 2 —> ( C o ° ( O i ) ) m a y be extended b y
/

c o n t i n u i t y from V(Q ) 2 onto CQ°(Í1 ). 2 T h i s extension w i l l be denoted b y JC. (JC -, •)

is easily seen to be a b i l i n e a r f o r m o n CQ°(ÜI) X CQ°(Í1 ). It is simple m a t t e r t o
2

show t h a t (JC-, •) is separately continuous. T h i s implies t h a t it is j o i n t l y continuous,

too (see T h e o r e m A . 3 ) . Therefore there exist non-negative integers m i , m 2 and
C > 0 such t h a t

\{K^)\ <C||^|| m a i n a |Hl m i > n i , (3.5-5)

if tp G £>(Oi) a n d ip G V(Vt ). 2

T a k e two compact sets X C Cli a n d Y C iï a n d assume t h a t d i s t ( X , dfli) > e 2 0

a n d d i s t ( Y , < 9 0 ) > e . L e t ^ i a n d ip be i n £ > ( R ) a n d V(R ),

2 0 respectively.
2
m N

F u r t h e r , we require t h a t suppip C 5 f (0) C R , s u p p ^ C B?(0) C M , /

r ^ i = m
2
n
R m

f
Rn ip = 1. P u t b y definition
2

K (x,y)
e := (3.5.6)

It is easily seen t h a t V>2 (^f ) 1

G £>(Q ) a n d ^ 2 (^) G £>(iîi) i f y G F , x G X a n d
6 < 6Q. I n v i e w of (3.5.5) we have

\K (x,y)\<C
e Vi (3.5.7)
T7?,o .fio mi ,Qi

if x G X a n d y G Y. A n easy c o m p u t a t i o n shows t h a t 7Í(.)(-, •) is continuous o n

A = X x r x { e : 0 < e < e }.
0

M o r e o v e r , this c o n t i n u i t y is u n i f o r m under the variables x a n d y. Suppose t h a t

there exists a d i s t r i b u t i o n K i n £ > ' ( M m + n
) fulfilling (3.5.4), t h e n e q u a l i t y (3.5.6)
m a y be w r i t t e n i n the following c o n v o l u t i o n f o r m

K (x,y)
t = (K*i¡> )(x,y), t (3-5.8)

where

(see S e c t i o n 3.3). T h e above equality indicates to us a way of d e t e r m i n i n g the

required d i s t r i b u t i o n K. W e o n l y need to show t h a t K (-, e •) converges to a d i s t r i -
b u t i o n K i n V(R ) RN+N
a n d t h a t it fulfils (3.5.4). A s t a n d a r d c o m p u t a t i o n gives us
 3.5. THE THEOREM OF KERNELS 51

the following equalities

JLK.fr,) = ( A ( -^ ( ^ ) ) )e (£^j), ( 3 ,, 0 )

where y = (y ,. .. ,y )
± n a n d i / = 1,. . . , n ,

= (A: ( . - » ( « ^ ) ). £ ( « - • . ( ^ ) ) ) , (3-5.11)

where x — ( x i , . . . , x m ) a n d ¿¿ = 1 , . . . , ra. O f course, the functions d K^(-, e •),

d K^(-y)
Vu a n d d K^(',
Xfx •) are continuous i n A O n e c a n show t h a t #(.)(•, •) is
infinitely differentiable i n A. T a k i n g i n t o account (3.5.5), one c a n show t h a t

\K (x,y)\
€ < Me~ s
(3.5.12)

if (x, y) G X x y , where s = m + n + m i + m 2 (compare (2.5.3)). W e s h a l l n o w

prove the c r u c i a l e q u a l i t y for o u r p r o o f

!(«-*(!))=££('-*<(!))• (»•»•")

for T/; G C ( 1
) ( R ) , where ipj(x)
N
= ~x ijj(x).
j Note that

d ^j(x)
e
= -iP{x) - Xjd J<iP(x).
e

Hence we get

0«^i = (I) - ^ S e
^ (I) • (3.5.14)
O n the other h a n d we have

x^
= - n e " " " V (I) - e""" ¿ 1
9 "V ( | )
e

e/ e

F r o m this b y (3.5.14) we o b t a i n

I (!)) - (!)+«--1 (* (f ) + ( f ) )
i/=l

- - • ¿ ^ ( f ) - t ¿ ( ^ ( ! ) ) -
52
3. T E N S O R P R O D U C T S A N D C O N V O L U T I O N PRODUCTS

T h u s , (3.5.13) is proved. I n v i e w o f (3.5.13), equality (3.5.9) m a y be w r i t t e n as

follows

where ^2,(0,^(2/) = -Vvfa{y) a n d ^ i , ( / i , o ) ( ^ ) = -x^i{x). Hence, b y (3.5.10) a n d

(3.5.11) we get

F o r s i m p l i c i t y o f notations p u t

£ ,i,(e o)(z,y) := (lC (e~ ih

e M>
n
( ~ ~ ) ) » c _ r o
^.(^,o) ( ^ )

£ ,l,(0,e„)(z,2/) :=
£ (^ (6-^2,(0,,) (^7")) > e - i
>l

O n e c a n observe t h a t these functions satisfy (3.5.12). F i n a l l y we have

d m
d n
d

R e p e a t i n g the above a l g o r i t h m a g a i n for the functions £ ,i,(e ,o) a n d I/ ,i (o, ) o n e e M e 5 eiy

can show t h a t •§^K (x y) e J is a finite s u m of Q^dyP¿ ,2,(a,/3)? e |(»,/3)| = 2, where t h e

functions L ^,( ,(3) E A satisfy (3.5.12). Generally, -^K (x, € y), k = 0,1, 2 . . . is a finite
s u m of d dPLz^^afi),
a
\(a,/3)\ = k, w h i c h are infinitely differentiable i n A a n d fulfil
(3.5.12). W e s h a l l now e x p a n d t h e f u n c t i o n K^(x,y) i n the Taylor formula around
some p o i n t 5, 0 < S < eo- T h u s , we have

JL t¡L (x,y)^ _ ^ . ( - S)'* f

^ ^ -J
1

) = g ^ p \ e - ô y
Ks e
K y
^ ( x , y + ^K 5 + t i e _ s ) ( X ) y)(l-tydt.
i=o 0

A s s u m e t h a t ip £ V(R ), m+n
supp <¿? C X xY, t h e n we have

j J K (x,y)(p(x,y)
€ dxdy = ^j ^ J J ^- ô(x,y)(f{x,y)
K
dx dy
X Y J
j=0
- ° ' y
X Y
x

(e - 5) s+1
1
}ff <9 S + 1

5!
0 X Y
 3.5. THE THEOREM OF KERNELS 53

Since djl+i 7f(.)(-, •) is a finite s u m of d d^L ^ j ^ ^(', a

e s rl a •), where |(a,/3)| = s + 1
we o n l y need t o show t h a t there exists
i

/ / ~ ^) S + l L
^ ( € - 5 ) , + i , ( a ^ ) ( ^ , 2 / ) ( l - t) (p(x,y)
s
s
dxdydt=: M(<p)
0 X Y

for (£> G £ > ( R m + n

) , where supp <¿> C X x Y . B y (3.5.12) we get

|(e - o) L _s),s+i,(a,{3)(x,
s+1
s+t{e y)(l - t) p(x, s
y)\

< C5~ (5 s
- e) s+1
< C (l - |)* (S - e) < Co

for O < e < 6 < 1 a n d 0 < t < 1. Since I/(.) +i,( ,/3) (•> *) is continuous i n A, )S a

therefore by. the Lebesgue d o m i n a t e d convergence t h e o r e m we have

+1 1

M(p) = ^—^ j J J L __ + ^ (x,y)(l-t) p(x,y)

0{1 t)jS 1Áa )
s
dx dy dt.
0 x Y
F i n a l l y we o b t a i n

lim / / K (x y)(f(x y)dxdy

e 1 1 = —— / / -Q- ô( ,y)v( ,y) K x x
dx dy
X Y 3 = 0
X Y
1
") s + 1
i f f d s + 1

s! / / / g^ ^i-t)( ^y) f( ^y)( - )

K x ( x 1 t s
dxdydt.
s\
X Y 0
for ip G £ > ( R m + n
) w i t h supp p C X xY.
O f course, for a r b i t r a r y c o m p a c t sets X C R m
and Y C R n
there exist O i (s R m

and £l d R2
n
such t h a t X C Í2i, 7 a n d (3.5.5) holds ( i n general 5 depends o n
X a n d Y ) . Since £ > ' ( R m + n
) is sequentially complete ( T h e o r e m A . 2 , E x a m p l e A . l ) ,
therefore there exists K G P ( ^ , m + n
) such t h a t ÜT - > i f as e - > 0 i n P ' ( R e
m + n
).
W e s h a l l now show t h a t (3.5.4) holds for ip G V(R ) n
a n d 7/; G V(R ). m
Put

^ ( x ) :=

and

Hence, b y (3.5.6)

K (x,y)(p(y)ip(x)
e dxdy = J J(/C(^ ,e(- 2 - 2 / ) ) , ^ i , ( - - x))<p(y)i/;(x) c dxdy.

R e p l a c i n g the i n t e g r a l b y R i e m a n n sums a n d p r o c e e d i n g t o l i m i t (see p r o o f of

T h e o r e m 3.1.4) we o b t a i n

j j K [x,e y)p{y)ip(x) dx dy = ( / C ( ^ 2 | C * <p), O0i,e * VO)-

 3. TENSOR PRODUCTS AND CONVOLUTION PRODUCTS
54

Since (/C(-), •) is continuous o n V(R ) rn

x £ > ( R ) , therefore by L e m m a 1.2.1
n
and
T h e o r e m 1.3.1 we get

lim / / K (x,y)ip(y)ip(x)dxdy
e = (£(limine * lim(^i,2 * ^ ) ) .

- J J
e—>
0 e—>
0
- e->0
R m
MP
F i n a l l y we have

EXAMPLE 3.5.1. N o w we s h a l l find the kernel K of the operator / C , /C(<p) = A*</?,

where
A = 5^ a ^ Z ^ , G C.
|/3|<m
N o t e t h a t the m a p p i n g

P(R n
x R ) 3 rç( ) - » K(rj) =
n
v
a
$ f D vfax)dx
im

\/3\<m R n

is linear a n d continuous. L e t ip G £ > ( R ) , t h e n n

aVi,) = (1^5 *¥>)(»).

P u t r](x,y) = ij)(x)tp{y), t h e n we get

K(<p®il>) = A
¡3 J ip(x)(D ô*i )(y)dy.
i3
P

\P\<m M n

W e k n o w t h a t Kip G C ° ° ( R ) (see T h e o r e m 3.2.1). T h u s we o b t a i n

n

K(ip®fa = (Kip^).
F o r ra = 0 we get K((p) = 5 * ip = ip. Therefore, the k e r n e l K of the i d e n t i t y
operator is given b y f o r m u l a

K(rj) =j 7](x, x) dx

for 7] G V(R n
x R ).n

3.6. C o n n e c t i o n b e t w e e n tensor p r o d u c t a n d c o n v o l u t i o n p r o d u c t
of d i s t r i b u t i o n s

L e t / a n d g be i n L ( R ) . 1 n
Suppose t h a t ip G P ( R ) , t h e n a n easy c o m p u t a t i o n
n

shows t h a t .

/ (/ ^ ~" ^ X y >g dy
I V( )
x dx =
jj f(x)g(y)(p(x ry)dxdy.
J

R'
& n
\R n
/ R n
R n

T h i s e q u a l i t y m a y be w r i t t e n as follows

j(/ * 0)(a?MaO = (f x ® + 2/). (3.6.1)

 3.6. CONNECTION BETWEEN TENSOR PRODUCT. 55

Let S a n d T be i n V(R ). A s s u m e t h a t p G V(R ).

n
If the expression n

(S (g) Ty)ip(x - f y) is sensible for each p i n D ( R ) , t h e n equality (3.6.1) justifies

x
n

t a k i n g the following e q u a l i t y

(S*T){p) := (S ®T )<p(x
x y + y) (3.6.2)

as a d e f i n i t i o n of the c o n v o l u t i o n p r o d u c t of S a n d T .

REMARK 3.6.1. I n v i e w of (3.6.2) we see at once t h a t i f there exist the c o n v o l u -

t i o n p r o d u c t s Q*T a n d 5 * T , t h e n there exist the c o n v o l u t i o n p r o d u c t s (Q + S)*T
a n d (Q + S) * T = Q * T + S * T .

EXAMPLE 3.6.1. P u t S = T = 5, a n d t h e n we have

(5 * 5){p) = (S 0 á ) ( ^ ( x + y) = p{0) =
x y 5(p).

W e s h a l l n o w give another example.

EXAMPLE 3.6.2. L e t S be a n a r b i t r a r y d i s t r i b u t i o n a n d T be a d i s t r i b u t i o n
w i t h a c o m p a c t s u p p o r t . W e s h a l l n o w show t h a t (S x ® T )p(x y - f y) is m e a n i n g f u l
for each p i n V(R ). n
I n order t o do it take by d e f i n i t i o n ip(x) := T (p(x y - f y)). We
have to show t h a t ip is i n V(R ). n
O n e c a n verify t h a t

supp ip C - supp T + supp <p € R . n

N o t e t h a t d?p>(x + •) tends to 9 ^ ^ ( x + •) as x -> x 0 0 i n C ° ° ( R ) . Therefore ip is

n

continuous i n R . A n easy c o m p u t a t i o n shows t h a t

n

F r o m the above consideration, it follows t h a t ip is i n D ( R ) . n

T h u s , i f S is a n
a r b i t r a r y d i s t r i b u t i o n a n d T has a c o m p a c t s u p p o r t , t h e n the c o n v o l u t i o n p r o d u c t
S * T m a y be defined b y (3.6.2). Since the m a p p i n g

V(R ) n
3p-+p(- + -)e C°°(R n
x R ) n

is continuous, therefore S * T G V'(R ). n

THEOREM 3.6.1. If S £ £ > ' ( R ) a n d a n

G V(R ),
n
then the convolution product
S*a in the above sense agrees with the regularization S a of S (see Definition 3.1.1).

PROOF. I n accordance w i t h D e f i n i t i o n 3.3.1 we have

where l(w) — 1, w G R . B y commutât i v i t y of tensor p r o d u c t we infer t h a t

n

(S x 0 a )p(x
y + y) = (l w <g> S )(a(w
x - x)p>(w)) = / <p(w)S (a(w x - x)) dw.
R 7

Therefore ( 5 * cx)(p) = S (p). (X T h i s statement finishes the proof.

•
 3. TENSOR PRODUCTS AND CONVOLUTION PRODUCTS
56

In general, equality (3.6.2) is not sensible, because </?(• + •) does not b e l o n g t o

V(W x R ) . F o r sensibility of (3.6.2) we have to take some restrictions c o n c e r n i n g
l n

the supports of S a n d T.

DEFINITION 3.6.1. If for each b o u n d e d set K CR n

the set A n (K - B) is also
b o u n d e d , t h e n we say t h a t the sets A a n d B are regular at infinity.

EXAMPLE 3.6.3.
(i) If A is a n a r b i t r a r y set i n R n
a n d B is b o u n d e d , t h e n A a n d B are regular
at infinity.
(ii) T h e sets .A = { x : x G R , x > a} a n d B = {x : x G R , x > b} are regular
at infinity.
(iii) T h e sets A = {(x,y) : (x,y) G R , x > 0} a n d B = {(x,y) : (x,y)
2
G
R , 2
x > 0, \y\ < x} are regular at infinity.
(iv) T h e sets A — B — { ( x , y) : (x, Î / ) E ! , X > 0} are not regular at infinity.
2

W e b e g i n w i t h presenting a l e m m a .

LEMMA 3.6.1. If any sets A and B are regular at infinity and K is bounded,
then the set
ft K = { ( x , y) : x e A, y G B and x + y G K}
is bounded in R n
x R . n

PROOF. Since the sets A a n d B are regular at infinity, therefore the set C =
A H (K - B) is b o u n d e d for every b o u n d e d K C R . F i x K, let x G A, z G K a n d n

y G B, t h e n x = z—y G C . N o t e t h a t for a i l pairs (x, y) i n i l ^ , their first c o o r d i n a t e s

are b o u n d e d . Since JFC is b o u n d e d , the second coordinates of (x, y) G Vt K are also
b o u n d e d . T h i s finishes the proof. •

L e t A^ — { ( x , y) : x + y G supp ^ } . P u t Q<¿, :— A ^ f l (supp S x supp T ) . It is

easily seen t h a t the sets A^ a n d supp S x supp T are closed i n R . T a k i n g i n t o 2 n

account T h e o r e m 3.4.2 a n d L e m m a 3.6.1 we infer t h a t the following two cases

Clip is a n o n - e m p t y c o m p a c t set, (3.6.3)

dist(A^, supp(5 0 T)) > 0 (3.6.4)

h o l d . W e are now i n a p o s i t i o n to give a d e f i n i t i o n of the c o n v o l u t i o n p r o d u c t

of two d i s t r i b u t i o n s h a v i n g regular supports at infinity. Suppose now t h a t (3.6.3)
holds. Therefore, by L e m m a 2.1.1 there exists a f u n c t i o n ip^ i n £ > ( R ) s u c h t h a t 2 n

VV(X,T/) = 1 for (x,y) G B ( 0 ^ ) , e > 0. O f course, p ( - + - ^ M ' , ") is i n P ( R ) . I f

e
2 n

(3.6.4) is satisfied, t h e n for any G G £ > ( R ) such t h a t supp 9 D s u p p ( 5 (g> T ) = 0

2 n

we have (S <S> T)(©<¿?(- - f •)) = 0 I n this case, i t is n a t u r a l to take ip^ = 0. P u t

(S * T)(<p) := (S x 0 T ) ( ^ ( x + y ) ^ ( x , y))
y (3.6.5)

for if G P ( R ) . n
O f course, (S * T ) ( ^ ) = 0 w h e n (3.6.4) holds. (Compare with
Section 2.4.)

THEOREM 3.6.2. The mapping

ip -> (S*T)(</?) (3.6.6)

gzi>en 6y (3.6.5) zs a linear continuous form on P ( R ) . n

 3.6. CONNECTION BETWEEN TENSOR PRODUCT. 57

PROOF. W e have to show t h a t the right side of (3.6.5) does not d e p e n d o n a

choice of the f u n c t i o n ip^. L e t ip^ a n d ip^ be i n D ( R 2 n
) and

^( iV) x
= ^i( ,y)x
= 1

for (x, y) G B (ft ). e (p Hence we infer t h a t

s u p p ( S ® T ) i l supp((</¿ - + •)) = 0-

Therefore

( S ® T ) ( ( ^ - < / > > ( • + ")) = 0.

T h i s means t h a t the right side of ( 3 . 6 . 5 ) does not d e p e n d o n ip^. W e shall now

show the l i n e a r i t y of ( 3 . 6 . 5 ) . P u t

^>i,v>2 — {( iV)
: x : x
£ supp 5, y G supp T , x + y £ supp pi U supp ip }.2

M o r e o v e r , assume t h a t ipcp lj(p2 eV(R ) 2n

a n d ip (pu(p2(x,y) = 1 for (x,y)eB (Çl € (pii(p2).
Note that

(S x 0 T )(<pi(x + y)i¡)<p <p (x, y)) = (S

y u 2 x 0 T )((p(x -h y)ip< (x, y)),
y Pi i = 1,2.

The details are left to the reader.

The o n l y t h i n g left to show is t h a t this m a p p i n g is continuous. If a sequence
(<¿v) converges to zero i n £ > ( R ) , t h e n there exists a c o m p a c t set K C R
n n
such that
supp (p C u for k = 1, 2 , . . . a n d (d p ) a
u converges u n i f o r m l y to zero i n K for each
a G N . I n v i e w of L e m m a 2 . 1 . 1 there exists a f u n c t i o n ipn
n
K i n V(R )2n
such that
'ipQ (x,y)
K = 1 for (x, y) G ^ ( f ^ ) . It is evident t h a t the sequence (ipçi p (- K v + •))
converges to zero i n V(R ). 2n
Since S ®T belongs to £ > ' ( R ) , therefore 5 * T is i n
2 n

D ( R ) . T h u s , the p r o o f is
7 n
finished. •

DEFINITION 3 . 6 . 2 . The mapping (3.6.6) is called the c o n v o l u t i o n p r o d u c t of S

and T .

REMARK 3 . 6 . 2 . L e t the functions / a n d g be continuous i n R . Suppose t h a t n

the supports of / a n d g are regular at infinity, t h e n i n t e g r a l ip(x) — f Rn f(x —

y)g(y)dy is finite for every x G R . n
M o r e o v e r , the f u n c t i o n ip is continuous. It
implies t h a t for each p i n V(R ) n
there exist the integrals

R 1
R'

where ip^ is defined as above. T h u s , we have s h o w n t h a t i f / , g G C ( R ) a n d t h e i r n

supports are regular at infinity, t h e n there exists the c o n v o l u t i o n p r o d u c t / * g i n

the d i s t r i b u t i o n a l sense.
58
3. TENSOR PRODUCTS AND CONVOLUTION PRODUCTS
3.7. D i f f e r e n t i a t i o n a n d s u p p o r t of c o n v o l u t i o n p r o d u c t

In this s e c t i o n we prove two i m p o r t a n t theorems for a p p l i c a t i o n s .

THEOREM 3.7.1. Let S and T be in V'(R ). n

If supp S and s u p p T are regular
at infinity, then there exists the convolution product 5 * T and

D (S
a
* T) = DS a
* T = S * DT a
(3.7.1)

for each a G N n
.

PROOF. T h e existence of the c o n v o l u t i o n p r o d u c t was s h o w n i n S e c t i o n 3.6.

It r e m a i n s t o prove t h a t (3.7.1) holds. Suppose n o w t h a t (3.6.3) is satisfied. In
accordance w i t h the definition of S * T we have

(S * T)(tp) = ~(S X 0 T )(%(x,

y y)d <p{x + y))
ei

= ~Sx (ry (J^(^{x,y)(p(x + y)) - ~>ip (x,y)(p(x

(p + y) j^S
.

A n a l y s i s s i m i l a r t o t h a t i n the previous section shows t h a t

(s x o T ) (^</v(z, y)<p(x +
y y)^j = 0.

Therefore

J 0 ( 5 * T ) ( ^ ) = -S
C i
x (ry ^ ( ^ ( ^ 2 , ) ^ +

= S x 0 (£) C i
T ) ( ^ ( x , y)<p(x + y))
y

= {S*D T)((p). €i

E q u a l i t y (3.7.1) is evident w h e n (3.6.4) holds. S i m i l a r l y , one c a n show t h a t D (S 6i

*
T) — D S €i
* T . F i n a l l y , b y i n d u c t i o n we have (3.7.1). It proves the t h e o r e m . •

THEOREM 3.7.2. Let S and T be in V(R ). n

If supp S and s u p p T are regular
at infinity, then
s u p p ( 5 * T ) C supp S + supp T . (3.7.2)

PROOF. It is easy to verify t h a t i f supp <p D s u p p S + s u p p T = 0, t h e n (3.6.4)

holds. H e n c e , (S * T)(<p) = 0. T h u s , the p r o o f is complete. •
 CHAPTER 4

DIFFERENTIAL EQUATIONS

4.1. F u n d a m e n t a l solutions o f differential e q u a t i o n s

T o b e g i n w i t h , we shall deal w i t h t h e p r o b l e m of the f u n d a m e n t a l s o l u t i o n o f dif-

ferential equations w i t h constant coefficients.

DEFINITION 4 . 1 . 1 . A d i s t r i b u t i o n E w h i c h satisfies the differential e q u a t i o n

aDE
a
a
= 5, (4.1.1)
\a\<m
where a a G C is called a fundamental s o l u t i o n o f t h e differential operator
Yl\a\<m a
ocD° . L

W e are interested i n finding a s o l u t i o n of the differential e q u a t i o n

aDu
a
a
= f, (4.1.2)
|a:|<ra

where / G V(R ). Suppose t h a t there exists t h e c o n v o l u t i o n p r o d u c t u := E * f.

n

T h e n b y T h e o r e m 3 . 7 . 1 a n d R e m a r k 3 . 6 . 1 we have

aDu a
a
= I aDE a
a
)*/ = «**/ = /.
|a|<ra \ |a|<m /

T h u s , t h e d i s t r i b u t i o n u = E * / is a s o l u t i o n of E q u a t i o n ( 4 . 1 . 2 ) .

EXAMPLE 4 . 1 . 1 . W e are l o o k i n g for a f u n d a m e n t a l s o l u t i o n of the differential

operator • = D^^ - D^°\ T o this end define a set ft = {(x,t) : t > 0, | x | < £ } .
Set E = \xn, where xn t h e characteristic f u n c t i o n of ft. W e prove t h a t E is a
1 S

f u n d a m e n t a l s o l u t i o n of • . W e have t o show t h a t

//
1 / d 2
d 2
\
ÔXQ(M) -^<p(x,t) - -—(p( ,t) x dxdt = ^ ( 0 , 0 )
2^ iy
' J
V<9* 2
' ;
dx 2

R 2

for if G £ > ( R ) . T h i s e q u a t i o n c a n b e w r i t t e n i n t h e equivalent f o r m

2

1 rr/d 2
d 2
\
2 JJ { ^ ^ x
^ ~ d x ^ ^ x
^ J dxdt = <p(0,0).
n
W e s h a l l now change variables x a n d t i n t h e above i n t e g r a l p u t t i n g

- and t = í ± l .
2
A n easy c o m p u t a t i o n shows t h a t
d 2
t-j + n t + v\ i / o 2
d 2
.
60 4. D I F F E R E N T I A L EQUATIONS

Therefore

where ft = { ( £ , 77) : £ > 0, 77 > 0}. It is easily seen t h a t

f

w h i c h was supposed t o be shown.

W e are interested i n finding a s o l u t i o n of the differential e q u a t i o n

D ^ u ~ D ^ u = h, (4.1.3)

where h is i n L} (R ) a n d supp h C IR x R . It is easily seen t h a t t h e suppü?,

oc
2 +

E — \xn, a n d supp h are regular at infinity. Therefore b y R e m a r k 3.6.2 we infer

t h a t t h e c o n v o l u t i o n p r o d u c t E * h exists i n t h e classical sense. I n fact, we have

(E*h)(x,t) = l jf h{x-Z,t-ri)dZdri,

where Q = { ( £ , 77) : |£| < 77, 0 < 77 < t}. P u t £' = x - £, rj = t - 77, t h e n
t
f

where 0 ¿ — — íí¿ - f (x, £). F i n a l l y we get

t i x + (t-rj)

(E *h)(x,t) = ±J y fc(í, 77) d f J drj. (4.1.4)

0 V-(í-r7)

4.2. T h e C a u c h y p r o b l e m for t h e wave e q u a t i o n w i t h d i s t r i b u t i o n data

T h e C a u c h y p r o b l e m for t h e wave e q u a t i o n i n one d i m e n s i o n a l space seeks for a

f u n c t i o n u t h a t satisfies
d 2
d 2

0*2 ^ ^ ~ d x * ^ =
^ ( 4
' '^
2

for (x, í ) e l x [ í , 0 0 ) a n d
0

u(x, t ) = f(x)
0 and J ^ ( x , ¿0) = g(x) (4.2.2)

for x G R , where / a n d g are given functions. L e t / a n d g be i n C(3R). Suppose t h a t

a function u belonging to (Rx [to, 0 0 ) ) satisfies (4.2.1) a n d (4.2.2) i n the classical
sense. A d i s t r i b u t i o n t h e o r e t i c a l f o r m u l a t i o n o f the C a u c h y p r o b l e m c a n be f o u n d
by b e g i n n i n g w i t h a classical s o l u t i o n u a n d c a l c u l a t i n g D^^u — D^ ^u i n t h e 2

d i s t r i b u t i o n a l sense. T o t h i s end i t is n a t u r a l t o e x t e n d the definition of u a n d h f r o m

R x [t , 0 0 ) o n a l l of R b y defining u(x,t) = h(x,t) = 0 for (x,y) G R x ( - 0 0 , t ).
0
2
0
 4.2. T H E C A U C H Y PROBLEM F O R T H EWAVE EQUATION... 61

W i t h this u n d e r s t a n d i n g , one m a y regard u a n d h as elements of V (W ). f 2

F r o m the
a s s u m p t i o n we have

oo oo 2 2 oo oo

J j (^j^u(x, t) - ^-^u(x, t)^j p(x, t) dx dt = Jj h(x,t)ip(x,t) dx dt (4.2.3)

— oo ¿o —oo to

for p e D ( M . ) . T a k i n g i n t o account (4.2.2) c o m p u t e the i n t e g r a l

2

oo oo
f f d 2
32

— OO to

O f course,
a (0,l)( (0,l)
9 u y ) ) = 0(O,2) U V , + Ô (0,1) U Ô (0,1) ,.
¥

F r o m this we o b t a i n

CO oo
f f d 2

/ / ~Q^ ( ^) P( ^)
U X ( x
d x
dt
— oo ¿o
oo oo oo

— —J d^ ^u(x,
0,1
to)p(x, to) dx — J j —u(x,t)—p(x,t)dxdt
— oo — oo to
oo oo oo

= - J g(x)p(x,t )dx 0 - I I —u(x,t)^ip(x,t) dx dt.

— oo —oo ¿o

A g a i n note t h a t

Hence we get
oo oo

— oo to

oo oo oo 2
= - y f{x)^p(x,t )dx 0 - J J u(x,t)^p{x,t) dx dt.
— oo — oo to

F i n a l l y we have

oo oo

/ / ~^ ( ^)^( ^)
u x x
dxdt
— oo to
oo oo oo oo ^

=- J g(x)p(x,t )dx 0 + y /(x)J^(x,¿)rár+ ^ J u(x,t)-^p>(x,t) 0 dx dt.

62 4. D I F F E R E N T I A L EQUATIONS

T h i s e q u a l i t y c a n be w r i t t e n i n the following d i s t r i b u t i o n a l form

00 oo

1 / ^2 ( )
U X
1) dx dt
— OO ¿o

= - ( 0 * ® ¿ t o M M ) - Ux ®DS )<p(x,t) t0 + £><°' >u(y>).

2
(4.2.4)
It is easily seen t h a t
OO CO
d 2

I! ^ u(x, t)(p(x, t) dxdt = D^ u((p).

0)
d 2

-oo t 0

F r o m this, (4.2.4) a n d (4.2.3) we o b t a i n

D^ u
2)
- D^ u 0)
= g®ô +f® to D5 tQ + h^:k. (4.2.5)
T h u s , we have s h o w n t h a t i f u satisfies (4.2.1) a n d (4.2.2) i n the classical sense,
t h e n u regarded as a d i s t r i b u t i o n fulfils (4.2.5). N o t e t h a t supp k C R x [t , oo). 0

Conversely, suppose t h a t / , g a n d h are i n V'(R) a n d supp h C R x [t , oo), 0

t h e n a d i s t r i b u t i o n u satisfying (4.2.5) is called a s o l u t i o n of the C a u c h y p r o b l e m

(4.2.1) a n d (4.2.2) for the d i s t r i b u t i o n a l d a t a / a n d g. W e k n o w t h a t E = \xn
is a f u n d a m e n t a l s o l u t i o n of the wave e q u a t i o n (see S e c t i o n 4.1). P u t to = 0 a n d
suppose t h a t supp h C R x [0, oo), t h e n the supp E a n d s u p p ( / 0 Do + p ® 5 + /i)
+

are regular at infinity. Therefore there exists the c o n v o l u t i o n p r o d u c t

u := E * ( / 0 Do + g 0 S + h). (4.2.6)
O f course, the d i s t r i b u t i o n u fulfils (4.2.5).
L e t / a n d g be i n C ( R ) . W e shall now show t h a t (4.2.6) gives us the d ' A l e m b e r t
formula

( / ( x - * ) + / ( * + *) + J
X+ t

g(v)dv + J
t X+(t-T))

J M£^)d£dJ
\

. (4.2.7)
X - t 0 X -(t-T]) /

Indeed, let H denote the H e a v i s i d e step function. N o t e t h a t

E*(f®DS)= D(°> (E 2)
* (/ 0 if)), (4.2.8)

E * for 0 J) = D^ (E X)
* (<? 0 jff)). (4.2.9)
Therefore, we are left w i t h the task of d e t e r m i n i n g the c o n v o l u t i o n p r o d u c t s E *
( / 0 H) a n d E * (# 0 L e t F be a f u n c t i o n i n C ^ ( R ) such t h a t F " = / . It is
easily seen t h a t
, , „w x if M if t > 0
{f®H){x,t) =<
[0 V
i f ¿ < 0.
A ;

B y (4.1.4) we have
(E*(/0ii))(x,t) = i J y /(Ode dr¡=±(F(x + t)+F(x-t)-2F(x)).
0 \x-(i-77) /
Hence we have
rj2 -1
— ( S * ( / ® # ) ) ( * , í ) = - (/(a; + Í) + f{x - t)). (4.2.10)
 4.3. F U N D A M E N T A L SOLUTIONS O FT H E LAPLACE OPERATOR... 63

T a k i n g G i n C ' ' ( R ) so t h a t G" = g, i n t h e same way one c a n show t h a t

(E * (g ® H))(x, t) = ±(G(x + Í) + G(a: - t) - 2G(a:)).

Hence

(E*(g®H))(x,t) = ±(G(x f
+ t)-G'(x-t))^^ j g(rj)dq. (4.2.11)
dt

F i n a l l y , b y (4.1.4), (4.2.10) a n d (4.2.11) we get (4.2.7).

4.3. Fundamental solutions o f t h e L a p l a c e o p e r a t o r a n d

the heat o p e r a t o r

W e s h a l l consider separately three cases

(i) Au = D u i f n = 1, 2

(ii) Au = D ^ u + D(°Mu i f n = 2,
(iii) Au = DV>°>- >U + £>(°'- ) + • • • + D ^ - ^ u 2
W i f n > 2.
It is easy t o verify t h a t D E 2
= 6 i f £^(x) = W e s h a l l now show t h a t t h e
f u n c t i o n E(x) = ~ In \x\ is a f u n d a m e n t a l s o l u t i o n for t h e operator A = £)( >°) + 2

L>(°' ). N o t e t h a t
2

^- j j In \J x r 2j
x\ dxi dx2 = J r In r dr < oo.

Bfl(O) 0

Therefore E is i n L /
1
O C ( R ) . W e have t o show t h a t AE = 6. I n order t o do t h i s set
2

by d e f i n i t i o n

E (x)
e := ^ - In ^x 2
+x + e. 2 2

O f course, E pointwise converges t o E i n R

e
2
— {0} as e —)> 0. M o r e o v e r

o < ^ ) - ^ ) = ¿ m ^ i + Í J .

Suppose t h a t e < 1. Hence we have

(E -E)(x)<(E -E)(x).
€ 1

It is easy t o check t h a t E\—EE L * ( R ) . B y the Lebesgue d o m i n a t e d convergence

o c
2

t h e o r e m we infer t h a t E tends t o E as 6 - » 0 i n L ^ ( R ). Therefore JE7 converges

e O C
2
€

to E i n £ > ' ( R ) . Since d i s t r i b u t i o n a l d e r i v a t i o n is continuous, w h a t is left is t o show

2

t h a t AE tends t o S i n V ( R ). A n easy c o m p u t a t i o n shows t h a t

e
2

-2
AE (x)c = -e\\x\ +e )- 2 2 2
= -e- 2
2+ 1
7T 7T

P u t ip(x) := 7 r ( | x | + 1 ) ~ . A s t a n d a r d c o m p u t a t i o n gives J ip — 1. I n v i e w of
- 1 2 2
R2

C o r o l l a r y 1.2.1 we infer t h a t AE converges t o 5 i n V(R ) . e

2

U s i n g t h e same m e t h o d we s h a l l show t h a t t h e N e w t o n i a n p o t e n t i a l

|2-n
(2 - n ) u ; n
64 4. D I F F E R E N T I A L EQUATIONS

where uj is the area of the unite sphere ({x : \x\ = 1}) i n R , is a f u n d a m e n t a l

n
n

s o l u t i o n of the L a p l a c e operator A . F i r s t of a l l we show t h a t E is i n L ( R ) . To i

1
o c
n

this end c o m p u t e the i n t e g r a l J ^ E(x)dx. W e change variables x b y i n t r o d u c i n g

B R

p o l a r variables
x = r cos t\ cos ¿2 * • • cos t -2 cos t _ i
1 n n

x = r sin ¿i COS ¿2 • • • cos t —2 cos t —i

2 n n

#3 = r sin ¿ • • • c o s t _
2 n 2 cos ¿ _ i n (4.3.1)

= rsin ¿ _i, n

where 0 < ¿i < 27r, —^ < U < ^ i f i = 2 , . . . , n — 1. W e k n o w t h a t the J a c o b i a n of

(4.3.1) equals
J ( r , ¿ i , . . . , t -i) n — r~ n x
cos t c o s t • • • c o s ~ t _ i
2
2
3
n 2
n

and
2TT 2 "2

üü = J n J ''' J cos ¿2 c o s £3 * • • c o s 2 n — 2
t — i dt — \ • • • dt\.
n n

0
- f -Î
A f t e r the above p r e l i m i n a r i e s we s h a l l c o m p u t e the i n t e g r a l J B R ^ E(x)dx. I n fact,
we have
R
— —|x| 2 _ n
dx — — — / roj dr = —— n -.
/ (2~n)uJ 1
( 2 - n K i 2(2 - n
n)

B {0) R 0
Therefore E is i n Lj (W ) oc
l
J n > 2. S i m i l a r l y as i n the previous case p u t

^ [ X )
' ( 2 - n K •
N o t e t h a t £ is i n C ° ° ( R ) . M o r e o v e r E (x) tends to E(x) as e -> 0 a n d - £ ( x ) <
n
e e

—E(x) if x ^ 0. B y the Lebesgue d o m i n a t e d convergence t h e o r e m we infer t h a t E E

converges to E i n L \ ( R ). T h e p r o o f is c o m p l e t e d b y s h o w i n g t h a t AE tends t o
O C
n
E

ô as e —» 0. O n e c a n show t h a t

ne-(|f| 2
+ l ) " +

AE (x)
e = ^ L . (4.3.2)

Set by d e f i n i t i o n
n(\x\ 2
+ l ) - ^
ip(x) :=

W e s h a l l now show t h a t j Rn tp(x)dx = 1. B y i n t r o d u c i n g p o l a r variables i n the

above i n t e g r a l we o b t a i n
71+2 0 0

fn(\x\ 2
+ l)- 2 f , 2 n ± 2 n - l j

/ dx = / n(r 2
+ 1) 2 R » 1
dr.
R n n
o
Set s = r 2
+ 1, t h e n
OO n - 2

/* ^ r / 1 \ 2
/ * M * = = / ( i - i ) '
5~ G?S. 2
 4.3. FUNDAMENTAL SOLUTIONS O F T H E L A P L A C E OPERATOR... 65

P u t t i n g a g a i n t = - j , one c a n verify t h a t j ip(x) dx = 1. T a k i n g into account Rn

(4.3.2), b y C o r o l l a r y 1.2.1 we conclude t h a t AE converges t o ô i n V'(R ). e

n

T h e same m e t h o d c a n be used t o find a f u n d a m e n t a l s o l u t i o n of t h e heat oper-

ator D i - » M ) - A , where A = ¿)(2„o,...,o,o) . . . jD(o,...,2,o)>
0
+ i + A n e a s y c o m p u t a t o r i

shows t h a t

(47rt) ? J exp
• |2 N
d i C
= L

P u t b y definition

E{ ,t):=
X I * M ) E * » x ( 0 > O o ) ( 4 3 . 3 )

(0 i f (x, t) G R n
x ( - o o , 0] - { ( 0 , 0 ) } .

It is easily seen t h a t E ( - , •) is i n L¡ (R ). W e s h a l l n o w show t h a t t h e f u n c t i o n

oc
n+1

E is a f u n d a m e n t a l s o l u t i o n of t h e heat operator. It is easy to verify t h a t E G

C°°(M - { 0 } ) . M o r e o v e r , (J¿ - A ) E(x,t) = 0 for (x,t) G ( R
n + 1
x - {0}). n + 1

Suppose t h a t ip is i n D ( R ) a n d e > 0, t h e n we have

n + 1

oo

0- J J <p(x,t) (J^-A ^J X E{x,t)dxdt

R n
e
oo oo

=
II ^^ '^~dt^^ '^ dxdt - J J E(x, t)A p(x,
X X
x i) dx dt.
R n
e R™ e
Therefore
• oo oo

J J p(x,t)~E(x,t) dxdt = J J E(x,t)A (p(x,t) x dx dt. (4.3.4)

Note that
^(o,...,o,i) E = ô (o o,i) ( B ^ _ j57 (o,..,o,iy ô

F r o m this i t follows t h a t
oo oo

/ J Wt C ^ f c ^ M M ) ) ^zcft - RJ je E{x,t)^p(x,t) n dxdt

oo

— J J E(x,t)A (p(x,t) x dx dt.

E q u i v a l e n t l y , we have

E(x, e)ip(x, e) dx

oo oo

j j E(x,t)—(p(x t) y dxdt - j j E(x,t)A p(x,t) x dx dt.

66 4. DIFFERENTIAL EQUATIONS
Since E is i n L ¿ o c (R n + 1
) therefore by the Lebesgue d o m i n a t e d convergence t h e o r e m
we o b t a i n
l i m /E(x,e)<p(x,€)dx = (i^ -» » ) - 0 0 1
A)E(<p).

e->0 J
It remains to prove t h a t

l i m / E(x, e)(p(x, e) dx = <¿>(0,0).

We know that

y* E ( x , e)<p(#, e) dx — (47re)~? ^ exp ^ — - ¡ — < ¿ > ( x , e) dx.

Putting x = we o b t a i n

y E(x, e)(p(x, e) dx — ( 4 7 r ) ^ ^ exp -

ip(\feüú, e) du.
MP MP

U s i n g a g a i n the Lebesgue d o m i n a t e d convergence t h e o r e m we get

l i m / E(x, e)(f(x, e) dx = (¿>(0, 0).

Therefore E is a f u n d a m e n t a l s o l u t i o n of the heat operator.

4.4. T h e H ô r m a n d e r inequalities

For s o l v a b i l i t y of differential equations i n a b o u n d e d open set the k n o w n Hôrman-

der i n e q u a l i t y is very i m p o r t a n t . T h i s section contains a p r o o f of this inequality.
To b e g i n w i t h , let us r e c a l l some notations. T h e n o r m a n d the inner p r o d u c t i n
L ( 0 ) w i l l be denoted respectively b y || • || a n d (•, •) or || • |Q a n d (•, -)Q i f some
2

a m b i g u i t y is possible. L e t
P(d) = a
- Q V

be a (non zero) linear differential operator w i t h constant coefficients of order ra.

THEOREM 4.4.1 (Hôrmander). For every bounded open set Ü in W 1

there exists
a constant C > 0 such that for every ip G V(Q)
\\P{d)<p\\ > C | M | . (4.4.1)
As constant C, one can take C — |-P| -Km,n> where | P | m m — s u p { | a j | : | j | = m} and
K ,n
m depends only on m and the diameter of ft.

PROOF. L e t ip be i n V(fl) and A — s u p n \x\. O f course, we have

(d Gj
(Xjip), if) = (Xjd '(p, e
ip) + (tp, ip).
I n t e g r a t i n g b y parts we o b t a i n
(d (xjip),ip)
ej
= -{xj(p,d *<p). e

Hence
(ip,ip) = -( d *(p,(p) Xj
e
- (xjip.d^ip).
Therefore
M l 2
< 2AM \\d vl ei
 4.4. T H E H Ô R M A N D E R INEQUALITIES 67

F i n a l l y we have
\\y\\<2A\\d ^\\-
e

T h e general p r o o f goes b y i n d u c t i o n o n order of P. If P(d) is a differential operator

of order m > 0 t h e n for p G V{0) t h e following f o r m u l a
P(d)(xjp) = XjP(d)p + Pj(d)p (4.4.2)
holds. T h e operator Pj(d) is zero i f a n d o n l y i f P(d) does n o t involve a n y differ­
e n t i a t i o n w i t h respect X j , a n d i f non-zero, Pj(d) is of order < ra. B y i n d u c t i o n o n
m , we prove t h a t for every p G T>(íl)

\\Pj(d)p\\ < 2mA\\P(d)<p\\. (4.4.3)

N o t e t h a t t h i s e q u a l i t y is true for m = 1. W e s h a l l n o w show t h a t , i f (4.4.3) holds,
t h e n t h e following i n e q u a l i t y

\\P(d)( <p)\\ < ( 2 m + l)A\\P(d)<p\\

Xj (4.4.4)
s h o u l d be true. Indeed, b y (4.4.2) a n d (4.4.3) we get (4.4.4). A s s u m e n o w t h a t
(4.4.3) is true for a l l operators P(d) of order < m — 1. L e t t h e operator P(d) be
of order m n o w . T h e expression (P(d)(xjp), Pj(d)p) c a n be represented i n t w o
different ways. F i r s t , i n accordance w i t h the definition of Pj(d) we have

P(d)(xjp) = XjP{d)p + Pj(d)p.

F r o m this we o b t a i n

(p{d){ v),p {d)v)

Xj 3 = (xjPidfaPjidfr) + WPjidM . 2
(4.4.5)
Second, a p p l y i n g i n t e g r a t i o n b y p a r t s t o (P(d)(xjp), Pj(d)p), we get

{P{d){x ),P {d)i )

jV j P = (P;(d)(x ),P*(d) ),
jV V (4.4.6)

where P*(d) = J2\u\<m — l ) ' " ' ^ . It is easily seen t h a t

\\P(d)p\\ = \\P*(d)<p\\. (4.4.7)

B y (4.4.5) a n d (4.4.6) we o b t a i n
\\PAdM 2
= (p;(d)(x ),p*(d) ) jV v - {xiPidfaPjidM. (4.4.8)
Since P(d) is o f order ra, therefore Pj(d) is of order < m — 1. Because (4.4.3)
implies (4.4.4), therefore b y (4.4.7) we get

\\P;(d)(x^)\\ < ( 2 m - 1)^11^(5)^11. (4.4.9)

B y v i r t u e o f the Schwarz i n e q u a l i t y we have

1 ( ^ ( 9 ) ^ ( ^ ) 1 < A\\P(d)<p\\ WPjidM (4.4.10)

and
\(p;(d)(x ),p*(d) )\jV v < \\p;(d)(x )\\ jip \\p*(d)<p\\. (4.4.H)
F r o m (4.4.10) i t follows the following i n e q u a l i t y

-{xjPid^Pjid)*) < A\\P{d) \\ V \\Pi{d)f\Y (4-4-12)

In v i e w o f (4.4.8), (4.4.12), (4.4.10) a n d (4.4.11) we get

\\Pi{dW < \\Pi{d){xjV)\\ \\P*{S)<P\\ +A\\P{d)<p\\ \\Pj{d)v\\

T a k i n g i n t o account (4.4.7) a n d (4.4.9) we o b t a i n

||P,-(dM| 2
< ( 2 m - 1)^11^(9)^11 \\P{d) \\ + A\\P(d)<p\\ V ma)?]].
 4. DIFFERENTIAL EQUATIONS
68

F i n a l l y , we have
WPjidM < 2mA\\P(d)<p\\.
T h u s , the p r o o f of (4.4.3)
is complete.
W e are n o w i n a p o s i t i o n to finish the proof. L e t P(d) be a n operator of order
1. It c a n be w r i t t e n as follows

P(d) = a i d e i
+--. + a d " n
e
+a . 0

N o t e t h a t P(d)(xj(p) = XjP(d)(p + cijip. Therefore Pj(d) = a,j. L e t us choose j

so t h a t I I = m a x ^ . . . \a \. = 1 ) O f course, \dj\ = \Pj\o > \P\i a n d | P j | o | M I
; n v <
2 A | | P ( 9 ) v ? | | - I n general, one c a n show t h a t i f P(d) is a n operator of order m > 1,
t h e n there exists j G {1,...
,n} so t h a t Pj(d) is of order m — 1 a n d | P j | _ i >
(4.4.3) (4.4.1)
m

| P | . T a k i n g i n t o account this r e m a r k a n d
m i m m e d i a t e l y we o b t a i n
by i n d u c t i o n o n ra. •

(4.4.1) is called Hôrmander's inequality.

Equality

EXAMPLE 4.4.1. L e t O be a n o p e n b o u n d e d set. N o t e that

n

Applying (4.4.1) for the operator d , Cu

v — 1 , . . . n we have

i~A(p,(f) > nC' (<p,tp).

n

T a k i n g CQ = nC' n we o b t a i n the Poincaré i n e q u a l i t y

(-A(p, ip) > CQII^H 2

(4.4.13)
for ip G V(Q).

In the sequel we s h a l l need a new i n e q u a l i t y w i t h a weight. S i m i l a r l y as above,

let P(d) be a linear differential operator w i t h constant coefficients.

THEOREM 4.4.2
(Hôrmander). For every bounded open set Q in R n
there exists
a constant M > 0 such that for all r¡ G M and ip G V(Q)

J e^\P(d)ip(x)\ dx 2
>M j e^\ip(x)\ dx, 2
(4.4.14)

where x = (xi,..., Xj,..., x ).

n

PROOF. Define a n operator Q(d) as follows

Q(d) —ei *P(d)e-i K

x x

In the first place we s h a l l show t h a t Q (d) is a constant coefficient operator w i t h the

same terms of higher degree as P(d). N o t e t h a t we c a n restrict our considerations
to o r d i n a r y differential operators. L e t
d \ ¿rn d " 171 1
d
^ r a m
d ^ + a m
- i
d ^ +
' ' ' + a i
d ^ + a
° ' a i € C
' a m
* ° -
Compute
 4.4. THE HÔRMANDER INEQUALITIES 69

Note that

Hence

7
/i=i x
' /

F r o m this, it is easily seen t h a t Q(d) is indeed a constant coefficient operator w i t h

the same terms of higher degree as P(d). W e s h a l l now a p p l y (4.4.1) t o the f u n c t i o n
ip — ez (p(x) w i t h respect to Q(d). T h e n we have
x

2
J e* ¿P(d)ip(x)
x 2
dx>C 2
J e* ¿ip{x)
x
dx.

F i n a l l y , i f we take M = C 2
we o b t a i n (4.4.14). •

COROLLARY 4.4.1. If ip e V(R ), n

P ^ 0 and P(d)ip(x) = 0 in H + = {x : ^ >
0} 7 t/ien (p(x) = 0 in n + , too.

PROOF. L e t O be a n o p e n set c o n t a i n i n g supp ip. Suppose, c o n t r a r y t o our

c l a i m , t h a t ip does not v a n i s h i n n + , t h e n one c a n choose rj > 0 so t h a t the
expression o n the right side of (4.4.14) w i l l be a r b i t r a r i l y b i g , b u t the expression
on the left side of (4.4.14) w i l l be b o u n d e d for r¡ > 0. T h i s gives a c o n t r a d i c t i o n .
T h u s , the p r o o f is finished. •

W e are n o w i n a p o s i t i o n to prove a more general

THEOREM 4.4.3. Let ip G V(R ), n

or more generally ip G L (R ), 2 n
ip has a
compact support. J / P ^ O and P(D)(p = 0 in R n
— B (0) r (if ip is not smooth, then
P(D)ip = 0 in the distributional sense), then supp ip is in B (0).
r

PROOF. W e s h a l l first prove the s m o o t h case. N o t e t h a t the b a l l B (0) r is the

intersection of half spaces

U-rh = {x : Xih ± H h xh n n > r, \h\ = 1}.

Suppose t h a t P(d)ip is e q u a l to zero i n R n

— jB (0), then by using translations and
r

rotations, we infer i n v i e w of C o r o l l a r y 4.4.1, t h a t supp ip C B (0). r T o t r e a t the

n o n - s m o o t h case, take
ip = ip * 1p
e ei

where ip (x) = e~ ip ( f ) , ip G V(B (0))

e
n
1 and J > = 1. T h e n <p G V(R ) e
n
and
suppip C P ( s u p p i p ) . N o t e t h a t
e e

P(d)ip (x)
€ = J P(d )iP {x-y)v(y)
x e dy = J P(-Ô )ip {x y e - y)ip(y) dy.

Let x J B ( 0 ) , t h e n suppip (x - •) C R - B (0).

e + r e B y the a s s u m p t i o n we get
n
r

P(d)(p (x)e = 0. T h i s reduces the p r o b l e m to the s m o o t h case. B y L e m m a 1.2.2,

ip -> ip i n L (R )
e as e —» 0. T h u s , the p r o o f is
2 n
finished. •
70 4. DIFFERENTIAL EQUATIONS
4.5. L -solvability
2

L e t P(D) be a linear differential operator i n the d i s t r i b u t i o n a l sense w i t h constant

coefficients. U s i n g the H o r m a n d e r i n e q u a l i t y we prove the following

THEOREM 4 . 5 . 1 .
If ft is an open bounded set in W , 1
then for every g G L (íl)
2

there exists u G L (Q) such that P(D)u — g in V'(Vt).

2

PROOF. Since ||P*(9)<p|| = \\P{d)p\\ (see ( 4 . 4 . 7 ) ) therefore b y ( 4 . 4 . 1 ) we have

\\P*(d)p\\ > C||<p|| for (p G V(Q). L e t
x = {p*(d)p:pev(tt)}.

C o n s i d e r the following m a p p i n g

X3P*(d)tp^(g,<p). (4.5.1)
Note that
\(g,<p)\ < \\g\\M\ < C-'WgWWP^dM.
T h i s means t h a t the m a p p i n g ( 4 . 5 . 1 ) is continuous o n X w i t h respect to the n o r m
II • ||. It c a n be extended o n X } the closure of X i n L (ft). 2
T h e n the R i e s z t h e o r e m
gives the existence of u G X, so t h a t (g, p) = (t¿, P*(d)p). T h i s means t h a t
P(D)u = g i n V'(Q). T h u s , the p r o o f of the t h e o r e m is finished. •

T o solve the differential equations i n W 1

we need the following a p p r o x i m a t i o n

THEOREM 4 . 5 . 2 . Let 0 <r<

r' < R. If v G L (B r(0)) and v 2
r satisfies
P(D)v = 0 on B t(0), then there exists a sequence {v ), v G L (B (0))
r v v
2
R so that
P{D)v v = 0 on BR(0) and v tends to v in L (B (0)) as v -ï oo.
v
2
r

PROOF. L e t
A = {a : a G L (B (0)), 2
R P(D)a = 0m V'(B {0))} R

and let A - denote the set of a l l g G L (B (0))

1
w h i c h are o r t h o g o n a l t o every a e A
2
r

i n the sense of L (B (0)). W e s h a l l first show t h a t i f g G A -, t h e n

2
r
1

\{>P,9)B \<C\\P{d) \\T V BR (4.5.2)

for p G D ( M ) , where || • \¡B '= || • | U ( B ( o ) ) - L e t us consider t w o cases, n a m e l y
n
r
2
r

P(d)p — 0 a n d P(d)p / 0 . O b v i o u s l y , we have (p,g)B = 0 w h e n P(d)p = 0. r 5

W e s h a l l n o w consider the case, w h e n P(d)p> ^ 0 . L e t X denote the closure o f

X = { P * ( 9 ) B : 6 G V(B (0))} i n L (B (0)).
R B y T h e o r e m 4 . 5 . 1 there exists i\) i n
2
R

X such t h a t
(TP,p*(d)e) BR = (p(d)<p,o) BR

for 9 e V(B (0)).R Hence, by (4.4.1) we have

\(1>,p*(d)e) \ Blt = \(P(d)<p,e) \ BR < ll^(9)^||B ||e||B H f l

zc^PidMBjp'weWB».
Therefore
\( p,p"(d)Q) \<
1 BR awpidMBjp'wewB*.
It is easily seen t h a t

mx)B \<C \\P(d) p\\ \\x\\B

R 1 i BR R

if X G Af. F i n a l l y we have

I M I B H ^ I P ^ I I B * - (4-5.3)
 4.5. L-SOLVABILITY 2
71
O f course,
(<P,9)B = (<P-ll>,9)B r r + (^^g)B - r

Since ip — ip is i n A, therefore

(<p,g)B = r {iP,g)B . r

H e n c e b y ( 4 . 5 . 3 ) we get
\(<P,g)B \<c\\p{d)<p\\ T BR

for ip G D ( R ) , where C — Ci\\g\\B -

n
R T h u s , i n e q u a l i t y ( 4 . 5 . 2 ) is proved. L e t
y = {P(d)ip:^eV(R )}. n

B y ( 4 . 5 . 2 ) the m a p p i n g y 3 P(d)ip —> (ip, g) is a continuous linear f o r m o n ^ Br

w i t h respect to the n o r m || • \\B - T h i s m a p p i n g c a n be extended b y c o n t i n u i t y

R

from y onto y (the closure of y i n L (BR(0)). T h e n the R i e s z t h e o r e m gives the

2

existence of w i n y so t h a t

(<P,g) Br = (w,P(d)<p) BR (4.5.4)

for ip G £ > ( R ) . Set n

, . \w(x) for x G -Biî(O),

W
[0 for x G R n
- B (0)
R

and

^ ' [0 for x G R n
- P R ( 0 ) .

Hence, e q u a l i t y ( 4 . 5 . 4 ) c a n be w r i t t e n as follows
(<¿>,¿/) n M = (w,P(d)<p)mri. (4.5.5)

T h i s means t h a t g = P*(D)w i n V(R ). Since supp û) is a c o m p a c t set a n d n

supp g C B (Q),R therefore by T h e o r e m 4 . 4 . 3 supptü C B (0). T h i s implies that r

w(x) = 0 if x G PÍÍ(O) - P ( 0 ) . R (4.5.6)

W e b e g i n n o w the last step of the proof. Suppose t h a t , v G L (R ), 2 n

supp v C B >(0) a n d P ( D ) u = 0 i n V(B ,(0)).

r T h e n P ( 9 ) v ( x ) = 0 for x G J 3 ( 0 ) ,
r e R

e < eo < m i n ( r — r,R — r ' ) , where v = ip * v (see the p r o o f of T h e o r e m 4 . 4 . 3 ) .

/
e e

M o r e o v e r , i> —» v as e —>> 0 i n L ( P /
e ( 0 ) ) a n d i> G 2 > ( J 5 / ( 0 ) ) . T a k i n g i n t o
2
R + E O e R + E O

account ( 4 . 5 . 6 ) a n d p u t t i n g v i n ( 4 . 5 . 4 ) i n the place of ip we o b t a i n

e

(v ,g)
e Br = (w,P(d)v ) e Br = 0. (4.5.7)
Hence we have '
]hn{v ,g) e Br = (v,g) Br = 0.
e—»0
Since g is a n a r b i t r a r y element of A -, therefore v G (A -) - = A, where A denotes1 1 1

the closure of A i n L (B (0)). F r o m this it follows t h a t there exists a sequence

2
r

(v ), v G A such t h a t v tends t o v as v —> oo i n L ( P ( 0 ) ) . T h u s , the p r o o f is

v v v
2
R

complete. •
T h e next t h e o r e m gives us the existence of solutions of differential equations
w i t h constant coefficients i n R . n

THEOREM 4 . 5 . 3 . Let P(D) be a (non-zero) constant coefficient linear differ-

ential operator in distributional sense on R . Then for every g G L (R ) n 2
oc
n
there
exists u G L ( R ) such that P(D)u = g.
2
o c
n
72 4. DIFFERENTIAL EQUATIONS
P R O O F . B y T h e o r e m 4 . 5 . 1 there exists u\ G L (B (0)) such t h a t P(D)m = g. 2
2

Let tu be a n a r b i t r a r y f u n c t i o n i n L (Bs(0)) fulfilling P(D)w

2
— g. I n v i e w o f
T h e o r e m 4 . 5 . 2 there exists v i n L (B (0)) such t h a t P(D)v = 0 o n B (0)
2
3 and 3

1
ll^i - w - v\\ Bl < 2'
Put u 2 = w + t h e n P(D)u 2 = g on £ 3 ( 0 ) a n d
1
I N - ui\\ Bl < -.

Inductively, assuming t h a t u has been chosen i n L (B +\{fS))

v so t h a t P(D)u = g 2
v u

on P ^ ( 0 ) , one c a n choose t ^ + i i n _ L ( J 9 „ 2 ( 0 ) )
+ 1 the following way. L e t w be i n
2
+
m

L (B + (G))
2
V 2 a n d P(D)w = g o n ^ ( 0 ) . O n £ „ + i ( 0 ) one has P(D)(u„ - w) = 0.
+ 2

B y v i r t u e of T h e o r e m 4 . 5 . 2 there exists v G L ( B + 2 ( 0 ) ) such t h a t P(D)v — 0 o n 2

I /

B + (0)
u 2 and

IK - v|| B l / <

S i m i l a r l y as above, set t¿^+i = w + v. T h e n P{D)u j \ u r = g on £^+2(0) and

n 1

I K + i - UVWB,, < ^j-

O f course, the sequence (u^) converges to some u i n L ( E ) . T h i s i m p l i e s t h a t 2

o c
n

P(D)u„ converges to P(D)u i n V(R ). Since P{D)u = g i n £ > ' ( £ „ ( ( ) ) ) , therefore

n
u

P ( D ) w = # i n V'{R ). T h i s finishes the proof.

n
•

W e are able to prove the m a i n t h e o r e m of this section.

THEOREM 4 . 5 . 4 ( M a l g r a n g e - E h r e n p r e i s ) . Every nonzero linear differential op-

erator P(D) with constant coefficients on R has a fundamental n
solution.

PROOF. Let
H(x) = H(xi) • • -H(x ), n x = (x u ... , z ) ,n

(H is the H e a v i s i d e step function), t h e n D^^'^H — S. B y T h e o r e m 4 . 5 . 3 there

exists u G Lf (R ), oc P(D)u = H. T a k e E = D ^ - ^ u . T h e n we have
n

P(D)E = P(D)D^-^u = D^-^P(D)u = D^-^H = S.

T h i s ends the proof. •

4.6. R e g u l a r i t y p r o p e r t i e s of differential o p e r a t o r s

Let A be a c o m p a c t s u p p o r t d i s t r i b u t i o n a n d © G C ° ° ( R ) . P u t n

A (x)
e :=A(S(x-.)).

THEOREM 4 . 6 . 1 . The function A e is in C ° ° ( R ) . n

PROOF. It is easily seen t h a t Q(x — •) tends to 6 ( # o — •) as x — » XQ in C°°(R ). n

The rest of the p r o o f runs as i n the p r o o f of T h e o r e m 3 . 1 . 1 . •

DEFINITION 4.6.1. T h e f u n c t i o n A e is called the r e g u l a r i z a t i o n of A b y means

of the f u n c t i o n O .

THEOREM 4 . 6 . 2 . If A is a compact support distribution and 9 G C (R ),

00 n
then
the convolution product A * O is in C°° (R ) and A * O = A e • n
 4.6. REGULARITY PROPERTIES OP DIFFERENTIAL OPERATORS 73
PROOF. TO this e n d we have t o show t h a t
(A y 0 S )(p(x
x + y) = J A (u)ip(u) @ du
R 71

for (p G V(W ). l
I n accordance w i t h the definition of A 0 0 we have

(A y 0 S )if(xx + y) = A y ¡ J S(x)if(x + y)dx \ = Ay Í J (p(u)O(u - y) duj J .

Set b y definition
o-h(y) := ^ ( ^ ) e ( A i / - y)h , (4.6.1) n

where NjJ = { i / : / i i / C supp <¿>} (compare the p r o o f of T h e o r e m 3.1.4). O n e c a n show

t h a t <7 (y) tends t o J
n Rn ip(u)Q(u — y) du as h —» 0 i n the sense of C -convergence.
00

F r o m this i t follows t h a t A(a ) h tends to A y (J R n <p(a;)0(u/ - y) do;). B y (4.6.1) we

have
A
K ) = E ^ ( H A ( 6 ( f t i / - t/))/i .
y
n
(4.6.2)

T h e expression o n the r i g h t side of (4.6.2) is a R i e m a n n s u m of the i n t e g r a l

f
Rn ip(u)Ae(u)du. Therefore A(a ) tends t o J (p(u)A@(u) du as h - » 0. F r o m
n Rn

the above c o n s i d e r a t i o n i t follows t h a t

(A Y 0 Q )if(x x + y) - j (p(u)Ae(u) du.

T h i s statement finishes the proof. •

DEFINITION 4.6.2. T h e singular s u p p o r t of a d i s t r i b u t i o n A is defined t o be

the c o m p l e m e n t of the largest o p e n set o n w h i c h A is a C7°°-function.

T h e singular s u p p o r t of A w i l l be denoted b y sing supp A .

THEOREM 4.6.3.
Let A be in V'(R ). If A is a C°°-function on R n n
- {0} and
S is a compact support distribution, then sing supp A * S C supp S.

PROOF. W e o n l y need to show t h a t i f x does not b e l o n g to s u p p 5 , t h e n there

exists a n e i g h b o u r h o o d U of x such t h a t the r e s t r i c t i o n A * S\u of A * S belongs
x x

to C°°(U ). T o this e n d assume t h a t x £ s u p p S , therefore there exists e > 0 s u c h

X

t h a t B (x) H supp S = 0. T a k e ip G V(B± (0)), ip > 0 a n d ip(x) = 1 o n B (0), t h e n

e %

by T h e o r e m 3.7.2 it follows t h a t

supp((^A) * S ) C suppS + B±(0).

T h i s implies t h a t ((ipA) * S ^ l s ^ ) = 0. Hence we have

(A*5)| = ((l-V)A*5)| B i ( x ) B 4 ( x ) .
4 4
O b v i o u s l y (1 - VOA € C * ° ° ( R ) . B y T h e o r e m 4.6.2 n

(A*S)\ eC°°(B (x)).

Bi(x) i

Since x was a r b i t r a r i l y chosen from the c o m p l e m e n t of supp 5 , therefore A * S is a

s m o o t h f u n c t i o n o n the outside supp S. T h u s , the p r o o f is finished. •
 4. DIFFERENTIAL EQUATIONS
74

DEFINITION 4.6.3. A differential operator P(D) = J2\ \<m a

a
ocD ,oc
a a G C is
said t o be h y p o e l l i p t i c i f for any A G V ( R ), sing supp A C sing supp
n
P(D)A.

I n other words, P(D) is h y p o e l l i p t i c i f a n d only i f for any o p e n set i l c M n

and
A G V (ft) the following i m p l i c a t i o n
f

P(D)A G C°°(Q) A G C°°(0)

holds.

THEOREM 4.6.4. .Le¿ P(D) be a partial differential operator with constant co-
efficients. Then the following conditions are equivalent:
(i) P(D) is hypoelliptic,
(ii) every fundamental solution of P(D) is C°° in W — {0}, 1

(iii) at least one fundamental solution of P(D) is C°° in R — { 0 } . n

PROOF. A f u n d a m e n t a l s o l u t i o n E of the operator P ( D ) satisfies the e q u a t i o n

P(D)E = 5. O f course, i f P(D) is h y p o e l l i p t i c , t h e n every f u n d a m e n t a l s o l u t i o n
of P(D) is a C - f u n c t i o n i n R — {0}, therefore (i) i m p l i e s (ii). T h e i m p l i c a t i o n
0 0 n

(ii)=>(iii) is c o m p l e t e l y t r i v i a l . W e o n l y need to show t h a t (iii) i m p l i e s (i). L e t

E be a f u n d a m e n t a l s o l u t i o n of P(D). Suppose t h a t E is C°° i n R - { 0 } . T a k e n

u G D ' ( R ) a n d assume t h a t P(D)u G C°°(ft), where Ü C R is open. F i x x i n Í2,

n n

t h e n B (x) C O for some e > 0. T a k e a f u n c t i o n ip i n V(B (x))

e such t h a t ip(x) > 0,
e

ip(x) = 1 for x G B&(x). N o t e t h a t

P(D)(ipu) = ipP(D)u + v,
where v = 0 o n B±(x) a n d o n outside B (x). € O f course,
E * (P(D)(ipu)) = E * (ipP(D)u) + E *v.
Since ipP(D)u is i n D ( i ? ( ; r ) ) , therefore b y T h e o r e m 3.1.2 a n d T h e o r e m 3.6.1,
€

E * (ipP(D)u) belongs to C ° ° ( R ) . T a k i n g into account t h a t supp v is o n outside

n

B± ( x ) , b y T h e o r e m 4.6.3 we infer t h a t s i n g s u p p ( £ ' * v) f l B±(x) — 0. Therefore

E\P{D){ipu) is a C°°-function o n B±(x). Note that

E * P(D)(ipu)
1
= P(D)E * (ipu) = 5 * (ipu) = ipu.
B y the above observation ipu is a C°°-function o n B±(x). Therefore u is s m o o t h
on BJL (X). T h i s statement finishes the proof. •

REMARK 4.6.1. T h e L a p l a c e operator a n d the heat operator are h y p o e l l i p t i c

but the wave operator is not h y p o e l l i p t i c . F o r p r o o f see Sections 4.2 a n d 4.3.
C o m p a r e also E x a m p l e 1.9.1.
 CHAPTER. 5

P A R T I C U L A R T Y P E S OF DISTRIBUTIONS A N D
CAUCHY TRANSFORMS

W e s h a l l present i n this chapter new subspaces T>' , 1 < p < oo of the space V, LP

w h i c h are n a t u r a l g e n e r a l i z a t i o n of Z A s p a c e s . T h e d i s t r i b u t i o n s b e l o n g i n g to V' LP

are i m p o r t a n t for a p p l i c a t i o n s a n d are s i m p l y represented b y a n a l y t i c functions.

5.1. Integrable distributions

o
DEFINITION 5 . 1 . 1 . W e say t h a t ip is i n B i f ip is s m o o t h a n d d (p a
belongs t o
C (R )
0
n
for each a G N . n

DEFINITION 5 . 1 . 2 . L e t (p be i n B, v = 1 , 2 , . . . . W e say t h a t the sequence

v (ip )u

o
converges t o zero i n B i f for each nonnegative integer ra, the sequence (||<£>iy||m R ) }
n

tends t o zero as v —> oo.

o

THEOREM 5 . 1 . 1 . The space V is dense in B.

o
PROOF. L e t (p be i n B. P u t ip v — g <p, where g
u v is t a k e n as i n L e m m a 1 . 1 2 . 1 .
It is evident t h a t \\(p — y>|| ,Rn tends to zero as v - » oo for ra = 0 , 1 , 2 ,
u m Since
o
the functions <p are i n T>, V is dense i n B. T h i s finishes the proof.
u •
T h e following t h e o r e m gives us a representation f o r m u l a for the linear contin¬
o
nous forms o n B.
o
THEOREM 5 . 1 . 2 . For an arbitrary continuous linear form À on B there exist
complex regular Borel measures fjL on W a
1
such that

|a|<m R n

Since the p r o o f of this t h e o r e m is the same as the p r o o f of T h e o r e m 2 . 7 . 3 , we

w i l l not give it here.
o
W e s h a l l n o w consider a larger vector space t h a n B.

DEFINITION 5 . 1 . 3 . W e say t h a t a f u n c t i o n ip belongs to VL°° i f it is s m o o t h

a n d a l l of its derivatives d ip, a
a GN n
are b o u n d e d o n I R . n

O n the space V^oo ( [ 2 4 ] ) we i n t r o d u c e a convergence i n the following way.

DEFINITION 5 . 1 . 4 . A sequence (ipv), ip v G UL°° is s a i d t o be /3-convergent t o

zero 'm VLOO i f i t satisfies the following two c o n d i t i o n s :
76
5. PARTICULAR TYPES OF DISTRIBUTIONS AND CAUCHY TRANSFORMS
(i) there exists a p o s i t i v e r e a l n u m b e r M a such t h a t H c ^ ^ H ^ o o < M a for

(ii) the sequence {d p ) a

v u n i f o r m l y converges t o zero o n each c o m p a c t set
K C R . n

THEOREM 5 . 1 . 3 . The space V is dense in with respect to -convergence.

PROOF. Note that R n

= \J™ =1 B (0).
u O f course, B„(0), v = 1,2,... is c o m -
pact. Let g v be as i n L e m m a 1 . 1 2 . 1 . T h e n there exist constants K a such that
11 ¿ ^ g ^ 11 LOO < K a for v = 1 , 2 , . . . . F o r p i n £>L°° take p v = g p.v N o t e t h a t <p v

belongs to T> a n d

0<¡3<a

It is easily seen t h a t d@ g (x) u = 0 for x G B„-i(0) a n d \/3\ > 0. H e n c e we get

Q\<*\ Q\a\
——(g p)(x) v = ——<p(x)

for x G ^ _ i ( 0 ) . Therefore

\\d (9v<P) ~ d p\\ oo B (o))

a a
L { k = 0

if v > k + 1. T h i s i m p l i e s ( i i ) . O n the other h a n d we have

ii9 (^)iu-<
a
£ (")iidViiL~ir-<viu~
0</3<a

^ £ ( " ) ^ i r - ^ i u ~ = : M a .
0<(3<a

T h u s , the sequence {p ) u satisfies (i). T h i s statement finishes the proof. •

DEFINITION 5 . 1 . 5 . A linear form o n V oo continuous w i t h respect to /^-conver-

L

gence is s a i d to be a n integrable d i s t r i b u t i o n (exactly its r e s t r i c t i o n to V). The

space of a l l integrable d i s t r i b u t i o n s w i l l be denoted b y V' .
LL

In accordance w i t h T h e o r e m 2 . 4 . 2 a n d T h e o r e m 2 . 4 . 3 every d i s t r i b u t i o n w i t h
a c o m p a c t s u p p o r t is integrable. L e t A be a n integrable d i s t r i b u t i o n . W e s h a l l n o w
o
show t h a t the r e s t r i c t i o n A l o of A i n t o B is continuous under the n a t u r a l convergence
o o o
of B. Indeed, let a sequence (<¿v), p v G B, converge to zero i n B. According to
o
the definition of the convergence i n B there exists constant M , a G N such n

<
a

t h a t IICP^HL
00
M , v — 1, 2 ,
a M o r e o v e r , the sequences {d p ) \cx\ < m cc
u 1

u n i f o r m l y converge t o zero o n W . T h i s i m p l i e s t h a t the c o n d i t i o n s (i) a n d (ii) are

1

o
fulfilled. Since V C B C V^oo a n d V is dense i n V^oo, we conclude b y the Lebesgue
d o m i n a t e d convergence t h e o r e m t h a t the f o r m u l a ( 5 . 1 . 1 ) is true for Lp G T)L^ , t o o .
O u r c o n s i d e r a t i o n give us the following
 5.2. REGULARIZATION OF INTEGRABLE DISTRIBUTIONS 77
THEOREM 5 . 1 . 4 . If A G V' l1 , then there exist a nonnegative integer m and
complex regular Borel measures fi , a \a\ < m such that

A
(^)= E (-1)^ J^yd^ for <peV oo. L (5.1.2)
\a\<m R n

5.2. Regularization of i n t e g r a b l e d i s t r i b u t i o n s

DEFINITION 5 . 2 . 1 . W e s h a l l denote b y V q, L 1 < q < oo, the set of a l l s m o o t h

functions (p such t h a t d ip G L (R ) a q n
for a G N n
. Suppose t h a t (p e V , € Lq e > 0.
We say t h a t the ip converge to ip i n V^q
e i f the ^ ^ a
e t e n d to d ip a
as e —> 0 i n
L ( R ) for a l l a G N .
q n n

LEMMA 5 . 2 . 1 . // <£> G 1 < g < oo andip G I/ (R ) 1 n

; t/ien ip*ip is a smooth
function and d (ip * ^ ) — d ip *ip for all a G N
a a n
.

PROOF.. It is k n o w n t h a t cV q L 2 if 1 < g i < #2 < oo (see R e m a r k 5.4.1).

In p a r t i c u l a r , P ¿ g C i f g > 1. N o t e t h a t

/i ((</? *
_1
+ /ie¿) -(<£>* ip)(x)) ~ ( 9 V * ^ ) ( x ) e

= J (d i (x ei
P + Ghe -y)-d ip(x-y))iP(y)
i
ei
dy,
MP

where 0 < 0 < 1. B y the Lebesgue d o m i n a t e d convergence t h e o r e m we infer t h a t

l i m [ (d v(x ei
+ ehe -y)-d v(x-y))^(y)dy ei
= 0
h-+oj
i

MP

for fixed x G R . W e n o w proceed by i n d u c t i o n .

n
•

LEMMA 5 . 2 . 2 . Let ip be in L 1
and J ip Rn = 1. Set ip (x) e := e~ ip(~).
n
If
ip G T>Lq, 1 < q < oo, then the ip * ip tend to (p as e —ï Q in t V^q.

PROOF. T h i s l e m m a is a n i m m e d i a t e consequence of L e m m a s 5.2.1 and 1.2.1.

•
THEOREM 5 . 2 . 1 . If A G V' L1 and <p G V oo or ip e V i, L L then

A ( a ? ) := A(y>(x - •))
v

is a smooth function (compare with Theorem 3.1.2). Moreover, A^ is in D^oo if

ip G P^oo and A^ is in V i L if ip £ V i.
L

PROOF. B y v i r t u e of T h e o r e m 5.1.4,

M * ) = £ (- ) 1 l / 3 1
/ ^ ^ - - W e -
\<m

N o t e t h a t the (^(x — •) t e n d to (p(x 0 — •) as x - » XQ in £>L°O • This implies that A ^

is continuous i n R . O n e c a n s i m i l a r l y show, as i n the p r o o f of L e m m a 5 . 2 . 1 , t h a t
n

d A^(x)=
a
Yl / d ^(x--)dfi .
a+
p

\(3\<m R n
 5. PARTICULAR TYPES OF DISTRIBUTIONS AND CAUCHY TRANSFORMS
78

O f course, d A a
(p is continuous i n R . Suppose now t h a t p is i n V i. n
L I n order t o
prove t h a t d A a
(p is i n V i L we s h a l l estimate

\\d K\\»<
a
X J j \d +^{x~y)\dxd\^\{y)
a

< M M(R"),
\(3\<m

where M = m a x j ^ j < m ||<9 ^<¿?||£i. A n a l o g o u s l y one c a n show t h a t

a+

||9 A^|| oo < M i

a
L J2 M(R ), n

\(3\<m

where M 1 = max|^j< m ||<9 V||Loo.a+/

•

DEFINITION 5 . 2 . 2 . T h e f u n c t i o n A ^ is called the r e g u l a r i z a t i o n of A b y means

of (f.

5.3. T e n s o r p r o d u c t of i n t e g r a b l e d i s t r i b u t i o n s

L e t p be i n V oo(R x l ) . Assume that S G £ > ( R ) and T G £ > ( R ) .

L
m n
L l
m
L l
n
In
m u c h the same way as i n the p r o o f of T h e o r e m 5 . 2 . 1 one c a n show t h a t

il>(x):=T(tp(x,.))

belongs to V^oo ( R ) . Therefore the expression S (T (p(x,

m
y)) is sensible for p G x y

V oo(R
L x R ) . M o r e o v e r , a c c o r d i n g to T h e o r e m 5 . 1 . 4 we have
m n

S (T (p(x,y)))
x y

( r fí\ot+¡3\ \
j Q^<P(x,y)h (x)he(y)d\^\(y)\
a d\n \(x)
a

/
(5.3.1)

(see S e c t i o n 2 . 7 ) . N o t e t h a t the m a p p i n g

d\a+P\
» m
x R n
3 (x,y) -> ^-^<p(x,y)h (x)hp(y) a

is a b o u n d e d B o r e l f u n c t i o n o n R m
x R . Therefore, b y the F u b i n i t h e o r e m ( [ 2 3 ] )
n

we o b t a i n the following e q u a l i t y

f i r <9i l a+/3
\
J I J ^( ^y)ha(x)h/3(y)
dxady(3
x
d\fii3\(y) I d\fi \(x) a

i dx<*dy^ ' x
y) ^ ) ^y)
h x h
^(1/^*1 x
 5.3. TENSOR PRODUCT OF INTEGRABLE DISTRIBUTIONS 79
where | / i | x \¡±p\ denotes the p r o d u c t of the measures | / ¿ | a n d
a a It is convenient
to w r i t e the double i n t e g r a l f Rrn JRn QlQ X
+
y0 <p(x, y) dfi (x) a dfip(y) i n the place o f
/ r x r ^^ïï^^V^a^hpiy) d(\fi \ a x \¡i^\)(x,y). T h u s , we have

p p a|a+/3|

N o t e t h a t the m a p p i n g

2?¿oo(M m
x R " ) 3 <p —> S {T {<p(x,y)))
x y

is linear a n d continuous w i t h respect to /3-convergence (see D e f i n i t i o n 5.1.4). I n ac-

cordance w i t h the general definition of tensor p r o d u c t of two a r b i t r a r y d i s t r i b u t i o n s
put
( S ® T ) ( p ) := S (T (<p(x,y))).
x y

It follows f r o m the above c o n s i d e r a t i o n t h a t 5 0 T is i n V' Ll (E m

x l " ) . Obviously,
S ® T = T(g>S.

DEFINITION 5.3.1. T h e d i s t r i b u t i o n S ® T given by (5.3.1) is s a i d to be the

tensor p r o d u c t of S a n d T .

O f course, i f S a n d T b e l o n g to V (R ), L1
n
t h e n the m a p p i n g

£>L~(R ) 3 ^ - > n
(S ®T )<p(x
x y + y)

is a n integrable d i s t r i b u t i o n o n R . W e s h a l l denote this d i s t r i b u t i o n b y S * T a n d

n

c a l l the c o n v o l u t i o n p r o d u c t of S a n d T. T a k i n g i n t o account (5.3.1) one c a n verify

that
(S * T ) * V = S * (T * V)
for 5 , T a n d V b e l o n g i n g to V' (R ). Ll
a
W e are able to prove the following

THEOREM 5.3.1. If A G V' Ll and f G V i , L then

(A*/)(p) = j A (x)tp(x)f dx

for tp G V.

PROOF. W e have to show t h a t

(Ay (g) f )(p(x

x + y) = J Af(x)(p(x) dx

for (pEU. T h e following equalities

(A y ® f )<p(x
x + y) = A y f(x)tp(x + y) dx^j = A y f(u - y)<p(u)du)^

are evident. A s i n the p r o o f of T h e o r e m 3.1.4 we p u t

a
h(y)'= f(hv-y)<p(hv)h . n
 5. PARTICULAR TYPES OF DISTRIBUTIONS AND CAUCHY TRANSFORMS
80

where = {y : hv G supp p). O n e c a n show t h a t cr^y) tends t o

j f(cu — y)(p(u) duj as h —» 0

i n V °v.
L T h i s i m p l i e s t h a t the A ( a / ( - ) ) t e n d to
l

Ay I[ J/ /(CJ
f{u- — y)ip(u)
y)ip(u) du | as / i -> 0.

O n the other h a n d we have

A M - ) ) = E A(/(/H/ - -MhvW

T h e right side of this e q u a l i t y is a R i e m a n n s u m of the i n t e g r a l f Rn A(f(ou — •)

(^(CJ)) e/o;. F i n a l l y we o b t a i n

(Ay ® f )p(x
x + y) = y A (u)<p(u))duj.
f

T h u s , the t h e o r e m is proved. •

T h e next t h e o r e m is v e r y i m p o r t a n t for a p p l i c a t i o n s .

THEOREM 5.3.2. If AeV' , Ll ipeV i L and f Rn = then the f Rn A^ (x)p(x)

e dx
tend to A((p) as e - » 0 for p G V oo, L where ip (x) = e~ i/j e
n
(^).

PROOF. B y v i r t u e of T h e o r e m 5.2.1, A ^ E G L (R ).1 N

H e n c e for p> G V oo L we
have

J A^, (x)p(x)
e dx = f I f 9 /
V e ( z ~ y)df¿(3(y) j <£>(z)d£

= E i- ) ^
1 1 1
/ ( / ^ ( * - ^ M * ) ^ J d^(y).

A f t e r a change of variable i n the i n t e r i o r i n t e g r a l we get

J A^(x)p(x)dx= J2 (-l)W J I jiP(u)d^p(y + eu)du \ d^(y).

O n the other h a n d we have

H<P) = E (~ ) 1 l/31
/ (/ V-H^l d^ip(y)d^(y).
\P\<m R n /
 5.4. T> p A N D V' SPACES
81
L LP

Therefore

A^ (x)(p(x)dx
e - A((p)

N o t e t h a t for fixed UJ the expression d (p(y + eu;) - d (p(y) u n i f o r m l y converges t o

¡3 (3

zero as e -> 0 w i t h respect to y. O f course, this expression is b o u n d e d o n W x l 1 n

.
Because tp G L ( R ) , we c a n use the Lebesgue d o m i n a t e d convergence t h e o r e m .
1 n

Therefore the

J J ( dP<p{y + ew) - dP^y)) ^ ( w ) dfi (y) dcu p

R R n n

t e n d to zero as e —> 0. T h i s statement finishes the proof. •

5.4. £>P
L a n d £>¿ p spaces

In this section we s h a l l e x a m i n e a new class of d i s t r i b u t i o n s c o n t a i n i n g the set o f

integrable d i s t r i b u t i o n s . A s the test space for these d i s t r i b u t i o n s we take the space
V, Lq1 < q < oo g i v e n b y D e f i n i t i o n 5.2.1. T h e convergence i n V g m a y be defined L

by means of the following f a m i l y of n o r m s

i

m U = \ £ llalli,) , m = 0,1,2,.... (5.4.1)

DEFINITION 5.4.1. L e t 1 < g < oo a n d ^ + ± = 1. T h e set of a l l linear forms

defined o n V^ q a n d continuous w i t h respect t o the n o r m s (5.4.1) w i l l be d e n o t e d
by V' .
LP

W e s h a l l n o w a p p e a l to the w e l l - k n o w n facts c o n c e r n i n g Sob ole v spaces, w h i c h

w i l l be collected i n the following

REMARK 5.4.1. I n v i e w of the S o b o l e v i m b e d d i n g t h e o r e m ([1]), i t follows t h a t

™+i
W m c W m,, 2 ) i < g i < g 2 < oo a n d \\<p\\ miq2 < |M| m + 1 ) ( ? 1 for ip G W m+1
^.
T h i s means t h a t the n a t u r a l i m b e d d i n g of V^i into V g L 2 is continuous i f 1 < qi <
Q2 < oo. A s a consequence of this fact, we have V LP1 C T> f
LP2 for 1 < p\ < p 2 < oo
(this means t h a t the r e s t r i c t i o n A|x> L<?1 of A G V LP2 i n t o V^i, + ^ - — 1 belongs
to V' LP1 ). It w i l l be also s h o w n t h a t V' Ll C V' Lv for p > 1 (see R e m a r k 5.6.1).

In accordance w i t h L e m m a 1.12.1, V is dense i n V , 1 < g < oo. Therefore, Lq

a d i s t r i b u t i o n A w h i c h is continuous w i t h respect to some n o r m of f a m i l y (5.4.1)

m a y be extended b y c o n t i n u i t y o n D . W e are able t o formulate a r e p r e s e n t a t i o n
L 9

t h e o r e m for the elements b e l o n g i n g to V' . LP

THEOREM 5.4.1. Each element A belonging to D' , LP 1 < p < oo takes the
following form

where ip G V^, ^ 4 - ^ = 1 and f a is a fixed element of L. p

82
5. PARTICULAR TYPES OF DISTRIBUTIONS AND CAUCHY TRANSFORMS
PROOF. T h i s t h e o r e m m a y be p r o v e d i n m u c h the same w a y as T h e o r e m 2.7.3.

In this case we o n l y need to take the C a r t e s i a n p r o d u c t C (R ) N n

:= X | j <
a m L (R
q n
),
where N — Ys\ \<m a
1 m s t
e a d of C (Q ) N f
= X| |< a m C o ( f t ' ) a n d to replace T h e o ­
r e m 2 . 7 . 2 b y the R i e s z representation t h e o r e m for b o u n d e d linear forms o n L (W ).
q n

•
5.5. C o n v o l u t i o n p r o d u c t

W e b e g i n our c o n s i d e r a t i o n w i t h presenting a n i m p o r t a n t t h e o r e m c o n c e r n i n g c o n ­
v o l u t i o n p r o d u c t of functions.

THEOREM 5 . 5 . 1 (Young, [21, vol.2). ] If f G L (R ) p n

and g G L«(R ), n
1 < p,
q < oo and ^ + ~ — 1 > 0 , then the convolution product f * g is in U ( R ), where n

1
+ 1
- 1 = 1
and
p q r
< ||/||z>IMk- (5.5.1)

W e s h a l l now show t h a t a s i m i l a r t h e o r e m is true i f the d i s t r i b u t i o n s S a n d

T are i n V and V , LP respectively. It was s h o w n i n S e c t i o n 5 . 3 , t h a t i f S a n d T
Lq

belong to V' , t h e n their c o n v o l u t i o n p r o d u c t is also i n V' .

Ll N o w , suppose t h a t Ll

S is i n V' a n d T is i n V ,
Ll 1 < p < oo. W e define s i m i l a r l y to S e c t i o n 5 . 3 the
LP

c o n v o l u t i o n p r o d u c t S * T of S a n d T as follows

(S * T)(<p) = (S x ® T )<p(x + y).

y (5.5.2)

W e shall n o w show t h a t the above f o r m u l a is m e a n i n g f u l for p G PL<? 5 ^ -h ^ — 1.

M o r e o v e r , it w i l l be s h o w n t h a t the m a p p i n g

V 3ip-^(S*
Lq T)(v) (5.5.3)

is continuous. F o r m a l l y , b y v i r t u e of T h e o r e m s 5 . 1 . 4 a n d 5 . 4 . 1 we c a n w r i t e

(S x ®T )ip(xy + y)

= £ E / j K(x)fp(y)d«+^(x + y)dyd\^\{x),
| a | < m i |/3|<m 2 M n n R

where fp G L ( R ) , \h (x)\ = 1 for x G R a n d \fi \ is the t o t a l v a r i a t i o n of

p n
a
n
a ji a

(see S e c t i o n 2 . 7 ) , N o w , we have to prove t h a t there exists the double i n t e g r a l

lict\(x). (5.5.4)

Put

M*)~ J My)d ^(x a+

+ y)dy.

Since h a is b o u n d e d i n R , it suffices to show t h a t tpp is also b o u n d e d i n R . It is

n n

easy to verify t h a t ij)p(x) = fp * d a + /

V , where fp(-) = fp( — -). B y T h e o r e m 5 . 5 . 1
we have
 5.5. C O N V O L U T I O N P R O D U C T 83

Therefore there exists the i n t e g r a l f \h (x)ipp(x)\ d\fi \(x). T h i s i m p l i e s the ex-

Rn a a

istence of (5.5.4). F r o m this, b y F u b i n i ' s t h e o r e m we infer t h a t the e q u a l i t y

(S*T)(<p) = (S ®T )<p(x a y + y)

is sensible for ip G L q
and

\(S*T)( )\< V £ £ ||^|| L P | «|(K")||^ ^|| ,.

M
+
L

\oi\<mi |/3|<m 2

T h i s i m p l i e s t h a t (5.5.3) is continuous.
It remains t o consider the t h i r d case, w h e n 1 < p , q a n d ^ + ^ — 1 > 0. I n this
case we c a n w r i t e formally, by v i r t u e of T h e o r e m 5.4.1

(S*T)(*0

= £ £ (-l)l a +
"l J j f (x)g (y)d P<p(x
a 0
a+
+ y)dydx, (5.5.5)
| a | < m i \(3\<m 2 M n on

where f G L (R )
a a n d g$ G L (R ).
p n
W e s h a l l n o w prove t h a t the right side of
q n

(5.5.5) is sensible for (p G V^s, where ^ + ^ + ^ — 2. F u r t h e r , i t w i l l be s h o w n t h a t

the m a p p i n g
Vs L 3 ip — » ( S * T ) ( < p ) (5.5.6)
is continuous. S i m i l a r l y as i n the above case we set

ipp(x) := Jg (y)d + ip(x (3

a l3
+ y)dy.

Since gp is i n L (R ) a n d d ipq n a+i3

G L ( R ) , it follows b y T h e o r e m 5.5.1 t h a t
s n

ijjp G L ( I R ) , where ^ - f ^ — 1 =
í n
It is easily seen t h a t j = 1 — K H e n c e we have
j + ^ = 1. I n v i e w of H o l d e r ' s i n e q u a l i t y we o b t a i n

/ \f (x)lp{3(x)\
a dx < | | / | | L p | t t

T h i s i m p l i e s the existence of the double i n t e g r a l

fa{x)g (y)d ip{x (3

oc+(3
+ y) dx dy.

It is easy to see t h a t

\{s*T){ )\< v £ £ | | / 4 ^ M M | d Q + /
V l k * .
\a\<mi \(3\<m 2

T h i s means t h a t (5.5.6) is continuous. Therefore S * T is i n V , Lr where £

It is easy to check t h a t - + - — 1 = - .
J p q r
F i n a l l y we o b t a i n the following
THEOREM 5.5.2 ( L . S c h w a r t z ) . If S and T belong to V' LP and V' Lq respectively,
1 < p , q < oo and ~ + ^ — 1 > 0 ; then there exists the convolution product S *T
and *S * T G £ > ' , w/iere rr
1
+ 1
- 1 = 1
.
 5. PARTICULAR TYPES OF DISTRIBUTIONS AND CAUCHY TRANSFORMS
84

5.6. C a u c h y t r a n s f o r m s of i n t e g r a b l e d i s t r i b u t i o n s

It is easily seen t h a t the f u n c t i o n -—^ 7ZI7)> 3 z 7^ 0 is i n V oo(R). L This remark

makes the following s y m b o l

C A
^ :
= ¿ A
( ^ ) í - - )
5 6 1

sensible for A G V' Ll a n d Ssz ^ 0 .

DEFINITION 5 . 6 . 1 . T h e f u n c t i o n CA is s a i d to be the C a u c h y t r a n s f o r m of A.

W e s h a l l now prove the following

THEOREM 5 . 6 . 1 . If A G V' (R), L1 then CA is a holomorphic function in C — R

and

|cA(z) = ¿ A t ( ^ 5 5 ) = C A « ( z ) , (5.6.2)

'where A^ is the distributional derivative of A .

PROOF. TO this end we s h a l l show t h a t

l i m -^A t (h^ttt -z- h)- 1

-it- z)- ) 1
-{t- z)~ )2
= 0.
/i->0 27T1 V
'

T o prove this one we o n l y need to verify t h a t for fixed z a n d a G N , the -§^(h(t —

z)~ (t2
— z — h)~ ) x
t e n d u n i f o r m l y to zero o n R as h —ï 0 . F o r s i m p l i c i t y of n o t a t i o n
put
g (t)
hiZ :=h(t-z)- (t-z-h)-\ 2

B y the L e i b n i z f o r m u l a we get

J ^ S M ( t ) = (-l) «!ft a
+ 1)(* - z)- » (t { +2)
- z - A)-^ ""). 1

P u t t i n g z = x + iy, h = £ + Í77 we o b t a i n

|J^0M(*)I

< a!|h| $ > + i ) ( ( i - z ) + y ) - ^ ( ( i - z - £ ) + (j, + T ? ) ) " ^ .

2 2 2 2
(5.6.3)
i/=0

Set d = y — 5 iî y > 0 a n d d = \y - f í | i f y < 0 . Hence, for \h\ < ô < \y\ we have

((* - xf + y )-^
2
((t — x — 0 2
+ (y + V ) 2
) - ^ < d-"'3

for t G R . F i n a l l y we o b t a i n the following e s t i m a t i o n

a

así»...**) <C\h\d~ - , a 3
C = a ¡ 5 ] ) ( i / + 1).
i/=0

T h u s , we have p r o v e d t h a t the -§^gh,z(t) converge u n i f o r m l y to zero w i t h respect

to t G R as h —> 0 . Hence, it follows t h a t

é C i
< " - s î A
( r h ) î ) -
 5.6. CAUCHY TRANSFORMS OF INTEGRABLE DISTRIBUTIONS 85
Note that

2TTÍ \ ' ~ Z) 2TTÍ \ d t t - z j 2TTÍ \(. - z)2

/

F i n a l l y , we o b t a i n (5.6.2). •

T h e next t h e o r e m is of t h e o r e t i c a l interest a n d is i m p o r t a n t for a p p l i c a t i o n s .

THEOREM 5.6.2. If f e L ( R )1
7 then

Cf(- + ie) - Cf(- ~ ie) -» / as e 0 in L ( R ) . 1

(5.6.4)

PROOF. A n easy c o m p u t a t i o n shows t h a t

C/(- + i e ) - C / ( . - t e ) = / i * / , e (5.6.5)

where h = M ( f ) ,
e = ^ T ^ 2 - B y L e m m a 1.2.1 we infer t h a t C7/(- + i e ) - C / ( - -
ie) tends to / as e —» 0 i n I / ( R ) . T h i s finishes the proof.
1
•

COROLLARY 5.6.1. / / / is in V i L then

Cf{- + \e)-Cf{--\e)-> f ase->0 inV i(R).L (5.6.6)

T h i s c o r o l l a r y is a n i m m e d i a t e consequence of L e m m a 5.2.2.

EXAMPLE 5.6.1. P u t f(t) := JT^J> W e s h a l l a p p l y the t h e o r y of residues t o

d e t e r m i n i n g Cf. L e t us consider the following contour

-R R

It is easy to check t h a t
1 f 1
lim - — / — — at — 0.
R-+00 2TTÍ J (t2
+ l)(t- z)

Note that

Cf(z) — l i m ~~7 Í •dt

V J
R-+oo 27T1 J (t
2
+ l)(t-z) 2i(i-z)
R,R]UF
[-R,R]l R

if $z < 0. S i m i l a r l y one c a n show t h a t

Cf(z) = 1

2i(i + z)
if > 0. A c c o r d i n g t o T h e o r e m 5.6.2
-1 1 1
2i(- + i ( l + e)) 2 i ( . - i ( l + e)) (0 + l 2

^(R).
A s i m i l a r t h e o r e m to T h e o r e m 5.6.2 is true i f A is a n integrable d i s t r i b u t i o n .
 5. PARTICULAR TYPES OF DISTRIBUTIONS AND CAUCHY TRANSFORMS
86

THEOREM 5 . 6 . 3 . If A belongs to V' , Ll then the CA{- + ie) - C A ( - - ie) tend to

A as e —> 0 in V' . L1

PROOF. I n accordance w i t h T h e o r e m 5.1.4 the integrable d i s t r i b u t i o n A c a n

be w r i t t e n as follows
m p

A(*0 = £ ( - l ) t t
/

where ¡i a is a c o m p l e x regular B o r e l measure a n d p G • W e have to show t h a t

J(CA(x + ie) - C A ( x - ie))p(x) dx £ ( - l ) a

J dp a
dfi a

R a = 0
R
as e -> 0. N o t e t h a t

J(CA(x+ie)-CA(x-ie))p(x)dx = JT(--l) a
f í ^ d h ( t - x)dfi (t)
a
e a J <£>(z) d x ,

R a=0 R \ R /

where / i ( t ) = ^
e T 2 + E 2 • B y the F u b i n i t h e o r e m we have

j(CA(x + ¿e) - CA(x - ie))p(x) dx = ¿ ( - l ) a

^(9 /i a
e * dfi (t)
a

R a = 0
R
m p
= E ( - ! ) a
{h *d ){t)d {t).
í
a
V íía

a=0 J
R

Hence, i n v i e w of L e m m a 5.2.2
m „ m „

lim£)(-ir / * (t) df2 (t) = J ] ( - l )

a
a
/ 8 ip a
d¡i . a

T h i s statement finishes the proof. •

REMARK 5 . 6 . 1 . L e t A be i n V . W e s h a l l show now t h a t the r e s t r i c t i o n A\x>

f
Ll Lq

of A to V g, L q > 1 belongs to V , LP where ^ + ^ — 1. Indeed, because cV oo

L

and V^q is complete, therefore b y T h e o r e m A . l (see A p p e n d i x ) a n d T h e o r e m 5 . 6 . 3

we infer t h a t A\x> LQ belongs to V . LP

T h e following t h e o r e m gives a c o n n e c t i o n between the C a u c h y transforms o f

integrable d i s t r i b u t i o n s a n d h a r m o n i c functions defined i n the u p p e r h a l f plane.

THEOREM 5 . 6 . 4 . Let A be in V' , Ll then the function

(x, y) —> CA(x + iy) - CA(x - iy)

is a harmonic function in I l +
= { ( x , y) : x G M , y > 0 } a n d

l i m (CA(x + i y ) - C A ( z - iy)) = A in V' i. L

 5.7. CAUCHY TRANSFORMS OF DISTRIBUTIONS BELONGING TO V' LP 87
PROOF. Since C - R 3 z —> CK(z) is a h o l o m o r p h i c f u n c t i o n i n C - R ,
therefore the f u n c t i o n

n +
3 (a;, y) —> CA(x + iy) - CA(x - iy)

is a h a r m o n i c function. T h e o r e m 5 . 6 . 3 gives us the second p a r t of our t h e o r e m . •

T h i s t h e o r e m tells t h a t every integrable d i s t r i b u t i o n A is the d i s t r i b u t i o n a l

b o u n d a r y value of a h a r m o n i c f u n c t i o n . T h e p r o b l e m of finding a h a r m o n i c func-
t i o n , w h e n given b o u n d a r y values are assumed is k n o w n as the D i r i c h l e t p r o b l e m .

REMARK 5 . 6 . 2 . It is easily seen t h a t

i y
CA(x + iy) - CA(x - iy) = A t

^TT (t — x) + y 2 2

The formula
i y
u(x y) = A ( - J , (5.6.7)
(t - X) + y-
9 t u 2 | 2

7T
2
K

where y > 0 is a n a t u r a l generalization of the P o i s s o n i n t e g r a l

If y
u(x,y) = - / 2f(t)dt
7T J (t-x) 2
+ y 2

for the u p p e r h a l f plane.

EXAMPLE 5 . 6 . 2 . It is easily seen t h a t CS(z) = —3^, ^ / O , where S denotes

the D i r a c measure concentrated at zero. I n accordance w i t h T h e o r e m 5 . 6 . 4 the
function
(x, y) —> CS(x + iy) - C6(x - iy) =
ir x + y £ z

is h a r m o n i c i n I I +
and ^ (.)2+ 2 y tends to S as y -> 0 +
i n V' . Ll

5.7. C a u c h y t r a n s f o r m s of d i s t r i b u t i o n s b e l o n g i n g to VLP

We s h a l l n o w show t h a t (.l y z is i n 1 < q < 0 0 i f ^sz ^ 0. N o t e t h a t

j^(t-*)-* = (-!)««!(*-s)-»- . 1

A s i m p l e c o m p u t a t i o n shows t h a t

2
j \ t - x - i y \ - " ^ d t < \ y \ - ^ j {(çy-) 1 dt
R R ^

= ^(a+D+l y ( 2 >
a; + 1 ) -f(a+l)^

= C\y\- q { a + 1 ) +
\

where C = ¡ (UJ r
2
+ l)~f du. Therefore € P L g , 1 < q < 0 0 if ^ 0.

T h i s i m p l i e s t h a t the s y m b o l C A ( z ) = i~A {zz^) * s a

^ s o s e n s i
b l e if A G V ,
LP

1 < p < 00.

 5. PARTICULAR TYPES OF DISTRIBUTIONS AND CAUCHY TRANSFORMS
88

THEOREM 5.7.1. If A e V , LP 1 < p < oo, then the Cauchy transform CA is a

holomorphic function in C — M and

l C A
w = è A
( r ^ ) = C A
" ,
w .

where A^ is the distributional derivative of A.

PROOF. A n a l y s i s s i m i l a r t o t h a t i n the p r o o f of T h e o r e m 5.6.1 shows t h a t

(CA(z + h)~ CA(z))^ 1

- ~ A (p^jâ) = ¿ M ( í - z)- (t 2
- z - hr'h).

Since A belongs to V , LP it remains to prove t h a t

9hA') :=(--z)- (--z-h)- h 2 1

tends to zero as h -> 0 i n V^g, where ^ + ^ = 1- T h e same c o m p u t a t i o n as i n the

p r o o f of T h e o r e m 5.6.1 gives us

^g , (t)
h z = ( - l ) a ! f c f > + 1)(* " z ) - ^
a + 2
\t - z - h)~^ + 1
- \
v

P u t t i n g z = x + iy a n d h = £ + 177 we get

J\{t - z)-^ + 2
\ t -z- dt
R
(v + 2)q (a+l-t^)g

/
((t-:r) 2
+ d ) 2 2
((t - x - £ ) + ¿ ) 2 2 2
dt=:,4„
R
where d = y — 5 ií y > 0 a n d d = \y + 5\ if y < 0. T h e n for < ô < \y\, b y the
H o l d e r i n e q u a l i t y we get

J ((t-x) 2
+ d 2
y ^ - dt\ Ij ({t-x-0 2
+ d 2
y ^ - ^ dt\ ,
R / \ R /

where ± + ± = 1 a n d 1/ = 0 , 1 , 2 , . . . . N o t e t h a t i i d ^ l a l > 1 a n d > I for

u = 0 , 1 , 2 , . . . a . Therefore the integrals

(v + 2)q 2

(ft - x) 2
+ d ) 2 2
dt = : B u

/
R

and
_ ( + - )pg
a 1 iy

((i - ^ - 0 + d ) 2 2
dt = : a
R

are finite a n d A v < BC. V V F i n a l l y , we have

W9KML* < a\\h\ ¿ ( 1 / + l)Bfct . q

 5.7. CAUCHY TRANSFORMS OF DISTRIBUTIONS BELONGING TO V' LP 89
T h i s i m p l i e s t h a t | | 9 ^ , ( - ) I U ? tends t o zero as h —» 0. T h u s , we have s h o w n t h a t
a
z
<

dz v ;
2TTÍ V(- ~ ^ ) 2
,

T h e equality ^ ^ = C A J ^ Z ) is evident. T h i s finishes t h e proof. •

W e shall n o w show t h a t each A i n V' , 1 < p < oo m a y b e r e p r o d u c e d b y means LP

of its C a u c h y t r a n s f o r m . I n order t o get this result we s h a l l need t h e following

LEMMA 5.7.1. If A e V , LP 1 < p < oo, g e V i L and f Rn g — 1, then

lim j Ag (x)ip(x)
e dx = A((p)
R n

for <p e V , Lq ± + \ = 1, where g (x) = e~ g ( f ) and A e

n
9e (x) = A(g (x e - •))•

PROOF. F i r s t o f a l l we have t o show t h a t A 9e is i n V^p. I n view of Theo-

r e m 5.4.1 we have

A 9 « W = £ ( - i ) w
f~9Áx-y)fa(y)dy,
\^\<m R n

where f a is some fixed element of IP. B y t h e Y o u n g t h e o r e m we infer t h a t A 9e is

i n VLV for e > 0. F r o m this, i t follows t h a t t h e following equalities

j\ .{x)ip{x)dx
g = j<p{x) I £ (- )' 1 Q |
/ ^9e(x-y)f (y)dy a \ dx
R" R» \ N < m R" /

= £ ( /^9e(x-y)f (y)dy\ a dx
\a\<m R n /

h o l d . T h e last i n t e g r a l c a n be w r i t t e n as follows

( - l ) H J<p(x) l ~ fg (x-y)f (y)dy)

e a dx.
R n
\ R n
I

T h e a - f o l d i n t e g r a t i o n b y p a r t s (see L e m m a 1.12.1) shows t h a t

R n
\ R n
/

= J I /9Áx-v)f (v)dy a j dx.

R n
\ R n
/

O n the other h a n d we have

jal <m
90 5. P A R T I C U L A R T Y P E S O F D I S T R I B U T I O N S A N D C A U C H Y TRANSFORMS

Hence we get t h e following e q u a l i t y

J A (x)(p(x)
9e dx - A((p)

= £ / Í J(9e(x-y)Uy)dy-f (x) a \ ^(x)dx.

\a\<m R n \ R n /

It is easy t o check t h a t

J9e(x - y)fa(y) dy - f (x) = j(f (x

a a - ey) - f (x))g(y)
a dy.

This implies that

J A 9e (x)ip(x)dx - A(p)

J f J{fa(x- ey) - f (x))g(y)

a dy j dx
\oi\<m

^ £ ( / ( / - cy) - J dxj í |
It remains t o b e proved t h a t t h e expression

J I y | J a ( z - ey) - f {x)\ a \g(y)\dy J d x

tends t o zero as e —> 0. U s i n g M i n k o w s k i ' s i n e q u a l i t y we get

/(/ ^ x
~ e y
ï ~ f<*( )\ \d(y)\dy^
x
dx

<j \g(y)\ I J \fa(x-ey)-f {x)\*dx\ a dy.

W e have t o show t h a t t h e expression o n the right side of this i n e q u a l i t y tends t o zero

as 6 -> 0. Indeed, | | / ( . - ey) - f\\ LP < 2\\f\\ LP for y G W a n d | | / ( . - ey) - f\\
1
LP -> 0
as e —y 0 for each y G W . Hence, b y t h e Lebesgue d o m i n a t e d convergence t h e o r e m
1

we o b t a i n t h e required fact. T h i s finishes the p r o o f of the l e m m a . •

N o w , we are able t o prove a r e p r o d u c e d t h e o r e m for A G V .

Lv
 5.8. CAUCHY TRANSFORMS OF SOME DISTRIBUTIONS 91

THEOREM 5 . 7 . 2 . If A e V' , LP 1 < p < oo, then

l i m / (CA(x + ie) - CA(x - ie))<p(x) dx = A(<p)

e-^OJ

for each <p G V^q, where ^ + ^ = 1 .

PROOF. Let A G V' L P . We know that

CA(x + ie) - CA(x - ie) = A (--

^7r(t-x) + e ; '
t
2 2

N o w , we see t h a t t h e t h e o r e m is a n i m m e d i a t e consequence of L e m m a 5 . 7 . 1 i f we
t*keg{x)= l Mx +iy •

5.8. C a u c h y transforms of some distributions

In this section we s h a l l b e concerned w i t h finite parts o f some divergent integrals.

Let <p be i n VLOO . P u t by definition

<P(0) + ^ ( 2 )
( 0 ) * 2
+ •••+ ( ¿ ^ ^ ^ M
( ° ) * 2 R O
- 2
) ) dx (5.8.1)

and

^ M : = / » - < ^ > ( ^ > 7 < - * >

R

- (±<p'(0)x + •••+ ( 2 M
1
_ 1 ) ! ^ ( 2 M
" 1 )
(0)^ 2 M
- ))1
dx (5.8.2)

for m = 1, 2, F o r m = 0 we have

<p(x) - <p(-ar)
dx (5.8.3)
2x
R

(compare w i t h E x a m p l e 1 . 1 . 4 ) . W e shall now show t h a t equalities ( 5 . 8 . 1 ) a n d ( 5 . 8 . 2 )

are m e a n i n g f u l for <p G VL°° i f m = 1, 2, W e restrict ourselves t o c o n s i d e r a t i o n
of ( 5 . 8 . 1 ) only. F o r m a l l y ( 5 . 8 . 1 ) c a n be w r i t t e n as follows

|x|<l

|a;j>l P l ^ l

It is easy t o verify t h a t t h e above integrals exist for (p G VL°°. M o r e o v e r , these

integrals determine linear continuous forms o n VL°O w i t h respect t o /3-convergence
92 5. P A R T I C U L A R T Y P E S O F D I S T R I B U T I O N S A N DC A U C H Y TRANSFORMS

(see S e c t i o n 5.1). Therefore is i n V' . Ll It is easy to show t h a t belongs

to V LP1 1 < p < oo. N o w , we d e t e r m i n e the C a u c h y t r a n s f o r m of ^ y . P u t t i n g
tp(x) = ^3^, $sx T¿ 0 i n t o (5.8.3) we o b t a i n

A n easy c o m p u t a t i o n shows t h a t

J - / " i f
^ > 0
(5 8 4 )
27TÍ 7 x 2
- ; ^ 1 JL i f ^ < 0 . l 5
- -
8 4 j

R ^ 2 2

To d e t e r m i n e the C a u c h y transforms of for A: > 1 we s h a l l need a r e c u r r e n t

formula

where k = 1,2, ••• a n d ^ ^ 0. F o r m u l a s (5.8.1) a n d (5.8.2) for tp(x) = —

can be w r i t t e n as follows

1 x 2
x (™~V\
2
,
(•) 2M V 7
27T1 J \X 2
- Z 2 ax

and

C T T ^ ( z ) = x-^+V ( + ^ + ^ + ••• + ^ _ 1 dx.

(.)2m+l ~ 2%
R

L e t us notice t h a t
2
7 2
T 2(m-lV

(•) 2 m V
' 27T1Z 7 V X
- Z Z
z2(m-l) J
R

T r 2(m-l)-l^
= - - i 2-2 _ z 2 ^2 z 2(m-l)
27TÏZ J

—1
*~ (.\2m2
(-)
v y

for ?n — 1,2; A n a l o g o u s l y we o b t a i n the following e q u a l i t y

C, J ^ (z) = I C T 4 - ( Z )

for m = 1, 2 F r o m t h i s a n d (5.8.4) we get

c l ( z ) = / s * i f
^ > ° (5.8.6)
(•)* 1-2^ ifS^<0, V ;

where /c — 1 , 2 , . . . .
 5.8. C A U C H Y TRANSFORMS OF SOME DISTRIBUTIONS 93

T h i s section w i l l be finished b y presenting t h e S o cho c k i - P l e m e l j formulas. W e

k n o w t h a t t h e C a u c h y t r a n s f o r m of D i r a c ' s measure S is given b y f o r m u l a

C6íz) = - ± ¿ , z * 0 .

M o r e o v e r , we have

¿ C Í W = C ^ )= * ^ , (5.8.7)

where 2 ^ 0 . H e n c e we o b t a i n

F r o m this, i t follows t h a t

1 1 \ TTi(-l)
2 \(x + ie) k
(x-ie) k
J (k - 1)

In v i e w of (5.8.6) we o b t a i n

\ +
T ^ w ) = C
W ( X + k )
" C
W { X
~ i e )
- ( 5
- '
8 1 0 )

B y v i r t u e o f (5.8.9) a n d (5.8.10) we get

1
-^ ~ -(CS^- \x +
i 1 ) k 1
i€)-CS^ - \x-ie))
k 1

(x + ie) k
(k - 1)!

C -^(x
~^(x 7 + ie)-C-^{x-ie)),
ie)-C-^ (5.8.11)

^ ^ ( C Ô ^ H x + ie)-C5^(x-ie))

+ (cj^(x + ie)-Cjj;(x-ie)y (5.8.12)

It was r e m a r k e d p r e v i o u s l y t h a t and are i n V' , Lv 1 < p < oo, k = 1, 2,

Therefore t h e expressions o n the right side of (5.8.11) a n d (5.8.12) a c c o r d i n g t o
T h e o r e m 5.7.2 t e n d t o

(*-l)! +
(-)fc a n d
(k-l)! d +
(-) f c

as e —> 0, respectively i n V . O n e c a n r e m a r k t h a t i f k > 2, t h e n the above

LP

convergence is i n V . F r o m t h i s we infer t h a t
f
Ll
94 5. P A R T I C U L A R T Y P E S O F D I S T R I B U T I O N S A N D C A U C H Y TRANSFORMS

as e —» 0 i n T>' , l < j 9 < o o i f &

LP = l and i n D f
L1 i f k > 2. F i n a l l y we o b t a i n the
following S o cho c k i - P l e m e l j formulas

1 _ TTi(-l)* 1
~ (fe^T)ï 5 +
( ^ ' ( }

1 7TÍ( — l ) f c
r.(i* i\ 1 ,

( ^ = - ( M ¿ +
Ô P ( 5
- -
8 1 4 )

(compare w i t h (1.6.5) a n d (1.6.6)).

 C H A P T E R 6

T E M P E R E D D I S T R I B U T I O N S A N D F O U R I E R

T R A N S F O R M S

6.1. T h e spaces S a n d S'

A space w h i c h v e r y often occurs i n a p p l i c a t i o n s is <S, t h e space of infinitely differen-

tiable functions defined o n E n
w h i c h , together w i t h t h e i r derivatives, a p p r o a c h zero
more r a p i d l y t h a n a n y power of j^y as \x\ -> oo (for instance exp( — | x | ) ) . 2
More
precisely, we assume t h a t p is i n S i f there exists a constant C p such t h a t a

\x^d Lp(x)\<c p
a
a

for x G K N
a n d a,(3 eN . n
O n e c a n i n t r o d u c e a f a m i l y of n o r m s i n S p u t t i n g

\\p\\m — :
max sup \x@ d p(x)\. a
(6.1.1)
0<\a+(3\<m x e Rn

DEFINITION 6.1.1. A sequence (p ) u is s a i d t o b e convergent t o p> i n <S, i f

\\Pv — p\\m tends t o zero as v —> oo for m — 0 , 1 , 2 , . . .

O b v i o u s l y , p tends to p as v —> oo i n S i f a n d o n l y i f (•)^d p

v converges a
L/

u n i f o r m l y t o (•) d p /3
i n W for a l l a,0 G N . W e also note t h a t i f <p ,p eV a n d
a 1 n
u

p —> p i n X>, t h e n clearly,

v —> p i n <S. I n other words t h e convergence i n V
is stronger t h a n convergence i n S. O n e c a n show t h a t S is a complete space w i t h
respect t o the f a m i l y of n o r m s ( 6 . 1 . 1 ) . W e see at once t h a t i f p> is i n <S, t h e n (•) d p (3 a

is also i n L ( R ) for a l l a,/3 G N .

1 - n n

THEOREM 6 . 1 . 1 . 25 dense m S.

PROOF. L e t g be as i n S e c t i o n 1 . 1 2 . W e s h a l l n o w show t h a t for a n y p i n S,

v

the sequence (g p) converges to Lp i n S. It is easy t o check t h a t

u

\d g„\ a
<M a

for y — 1 , 2 T h i s involves the i n c l u s i o n (g p) v i n S. M o r e o v e r , for fixed e > 0

there exists VQ G N such t h a t

\xPd«(g p-p)(x)\ u <e

for x ^ ^ o - i ( O ) a n d u > UQ. F u r t h e r , we have

dr(g„<p){x) = (g &*<p)(x) v = 9>(i)

for x G B -i(0)
u a n d z/ > UQ (see T h e o r e m 5 . 1 . 3 ) . F i n a l l y we o b t a i n

\x^d {g^-<p){x)\
a
<e

for x G R n
arid z/ > I/Q. T h i s statement finishes the proof. •
 6. TEMPERED DISTRIBUTIONS AND FOURIER TRANSFORMS
96

DEFINITION 6 . 1 . 2 . A linear f o r m defined o n S a n d continuous w i t h respect to

the convergence i n this space (see D e f i n i t i o n 6 . 1 . 1 ) is c a l l e d a t e m p e r e d d i s t r i b u t i o n
(exactly the r e s t r i c t i o n A\x> of A into V).
T h e set of a l l t e m p e r e d d i s t r i b u t i o n s w i l l be denoted b y S'.

DEFINITION 6 . 1 . 3 .
W e say t h a t / G O M i f / € C ° ° ( R ) a n d for each a e n
N n

there exist M > 0 a n d k G N such t h a t

—f{x) <M{l + \x\ ) . 2 k

REMARK 6 . 1 . 1 . It is easy to check t h a t i f ip G S a n d / G OM, t h e n fip is i n S .

M o r e o v e r , i f ip G S a n d ip converges to ip i n «S, t h e n fip converges t o f(p i n S.
v v u

T h i s statement shows us t h a t i f / is i n O M a n d T is i n <S', t h e n fT is also i n S'

(see S e c t i o n 1 . 7 ) .

THEOREM 6 . 1 . 2 . To each A in S' there correspond m G N and complex Borel

measures fi p a such that

HP) = E ( - l ) H
/ ( ' ) ^ > ^ (6.1.2)
0<|a+£|<m
/ o r (p E S.

PROOF. Since A is a linear continuous f o r m o n <S, therefore there exist ra G N

a n d M > 0 such t h a t
\A(ip)\ < M\\<p\\ m

for ip G S (see S e c t i o n 1 . 1 ) . T h e rest of the p r o o f runs as i n S e c t i o n 2 . 7 . L e t us

consider the C a r t e s i a n p r o d u c t

c *(R )=
0
n
x c 0 m , N= Y, L

W e s h a l l denote the elements of ( R ) by (ip p). T h e c o m p o n e n t of {<p p) cor-

n
a a

r e s p o n d i n g to subscript a/3 w i l l be denoted b y ip p. L e t us endow the C a r t e s i a n a

p r o d u c t C ^ ( R ) w i t h the n o r m
n

W&a^Wc" :
= _, m a
;* S U
P 1^(^)1-
0
0<|a+/3|<ma; ]Rn G

Set, b y d e f i n i t i o n

S™ := {(p : ip G C ° ° ( R ) , (•fd ip n a
G C ( M ) for | a + /3| <
0
n
ra}.
C o n s i d e r the m a p p i n g /C : <S - » C ^ ( R ) defined i n the following f o r m
m n

K(<p) = ((-l)W(-fd <p). a

N o t e t h a t , JC establishes the linear continuous i n j e c t i o n f r o m S M

i n t o CQ ( R ) . W e
n

s h a l l now define a linear continuous f o r m A o n JC(S ) m

taking

Â(/C(v>)) := A f a ) .
O f course,
\I(K(<p))\ = \A(<p)\<M\\K{<p)\\ . cSr

B y the H a h n - B a n a c h t h e o r e m there exists a continuous extension of A f r o m

JC(S )m
onto the whole space C ^ ( R ) . B y L e m m a 2 . 6 . 1 a n d T h e o r e m 2 . 7 . 2 we
n

obtain (6.1.2). •
 6.2. TENSOR PRODUCT OF TEMPERED DISTRIBUTIONS 97
6.2. T e n s o r p r o d u c t of t e m p e r e d d i s t r i b u t i o n s

L e t <p(-, •) b e l o n g to <S(R x R ) . L e t S a n d T be i n < S ' ( R ) a n d <S'(R ), respectively.

m n m n

Put
(S ®T )<p(x,y)
x y := S (T (p(x,y))).
x y

W e s h a l l now show t h a t the right side of this e q u a l i t y is sensible for p G <S(R x R ) . m n

In the first place we show t h a t ip belongs to <S(R ) i f ip(x) :— T (p(x^y)). Indeed, m

y

by T h e o r e m 6.1.2 we have

M*) = E (- )' '

1 7
/V Çï<P(*>v)dlh6(v)
S

0<\l+6\<l ¿ n
V

for <£>(•,•) G <S(R m

x R ) . Put n

^ô(x)
i i f
:= ( - l ) J y m
A^' ' 7

—p(x,y)dfi (y).l6

W e b e g i n by p r o v i n g the c o n t i n u i t y of IJJ S i n R . O f course, 1

m

i i f fd^ 9 I t I
\
ipjôix + O -ip-yâix) = (-1) 1 7 1
y y [^p{x + £,y)-—p{x,y)\ d^ (y).
ô

It is easily seen t h a t
9M Q\l\
y
dy^^ x +
^y)->y Q^v(x,y)

u n i f o r m l y w i t h respect to y G R n
as £ —y 0 for a r b i t r a r y fixed x G R m
. Since the
f u n c t i o n y^^^-p(x, y) is b o u n d e d i n R m
x R , therefore by the Lebesgue d o m i n a t e d
n

convergence t h e o r e m we infer t h a t ip^s is continuous. F u r t h e r , we have

h -\^s(x + he ) á - ^s(x)) = (-l)W J y ' H ^ T p{x + 6hej,y) d¡i {y),

l6

where 0 < 6 < 1. A passage to t h a t l i m i t s i m i l a r to the above i m p l i e s t h a t

_d_
-ip ô{x)
7 = (-l) l 7 i
j y 6
®^ dx Lp{x,y)dfi {y). l6

dx

W e n o w proceed b y i n d u c t i o n on a. F i n a l l y we get

<9'i ,
a
r g|a+7|
U l x

dx° /

It remains to prove t h a t the expression x@ ^zij) s{x) 1 is b o u n d e d i n R . I n fact m

have
g\ \ a
r ¿jl +7l a

^ d x ^ ^ - / , ™ d\^ô\(y)
dx dy^
X x ) P a

O X
J ( y)(=.Rrn
X:

<M a / 3 7 ¿|/i ¿|(R ), 7

n
98
6. TEMPERED DISTRIBUTIONS AND FOURIER TRANSFORMS
where | / ¿ ¿ | is t h e t o t a l v a r i a t i o n of / ¿ ¿ . A simple verification shows t h a t i f
7 7 •)
tends t o zero i n <S(R x R ) t h e n m n

(-1)' '/ 7
y ^A',y)d^s(y)^o
s

i n 5 ( R ) as v —> oo, as well. U s i n g a g a i n T h e o r e m 6.1.2 we o b t a i n

n

(S x <g> T )<p(x,y)
y

= E E (-!) | a + 7 1
/ ( / ^ y s
í ^ ^ y ) d ^ ( y ) )
\<*+0\<k\-r+6\<l Rm V " /
(6.2.1)
O f course, (6.2.1) is a linear continuous f o r m o n <S(R x R ) . m n

DEFINITION 6.2.1. T h e linear continuous form o n <S(R x R ) given b y (6.2.1) m n

is said t o b e t h e tensor p r o d u c t of S a n d T.

COROLLARY 6.2.1. Tensor product of tempered distributions is commutative.

(See Section 5.3.)

EXAMPLE 6.2.1. L e t l ( x ) = 1 for each x e R . A s s u m e t h a t T is i n <S'(R ), m n

t h e n we have
(l x (8) T )<p(x, y) = (T <g> l )(p(x,
y y x y).
T h i s equality c a n be w r i t t e n as follows

J T ((p(x,y))dx
y =T y j j ip(x,y) dx

6.3. Fourier transforms o f integrable functions

In this section, we s h a l l deal w i t h a concept w h i c h is of basic i m p o r t a n c e i n t h e

a p p l i c a t i o n s of d i s t r i b u t i o n s t o the theory of p a r t i a l differential equations, i n t h e
theory of p r o b a b i l i t y a n d m a n y other branches o f analysis. L e t / be i n L ( R ) . 1 n

Put
Tf{a)~ J e ^f(x)dx,
i
(6.3.1)

where (x, cr) = x\G\ + • • • - f xa.n n

DEFINITION 6.3.1. T h e f u n c t i o n Tf is called t h e F o u r i e r t r a n s f o r m o f / .

THEOREM 6.3.1. / / / ¿5 in L (W ) 1 l
f then Tf is continuous and

H ^ / l l i - < H/IUi. (6.3.2)

PROOF. W e see at once t h a t < > ° + ) tends pointwise t o e^ ^ as h -> 0 a n d
e
x h x

\e } f(x)\ = \f(x)\. N o w , we c a n a p p l y t h e Lebesgue d o m i n a t e d convergence

t{X)CT+h

t h e o r e m . T h e second p a r t of t h e t h e o r e m is evident. •

EXAMPLE 6.3.1. Let / = X [ - i , i ] - A n easy c o m p u t a t i o n shows t h a t

_ , x sin a
TX[-i,i}(cr) = 2——.

It is easy t o see t h a t does n o t b e l o n g t o L ( R ) . 1 n

 6.3. FOURIER TRANSFORMS OF INTEGRABLE FUNCTIONS 99
EXAMPLE 6 . 3 . 2 . T a k e f(x) = e~ 2 . L e t us consider the contour

—R 0 R

—R — icr R — ia

_
N o w , 2: —» e 2 is a n entire f u n c t i o n of the c o m p l e x variable z = x + iy, a n d
therefore its i n t e g r a l a r o u n d the rectangle w i t h vertices —R, R, —R — ia a n d R — ia
is zero (R > 0 ) . It is easy to check t h a t the integrals o n the t w o v e r t i c a l sides t e n d
to zero as R — » 0 0 . Therefore
dx.
R

O n the other h a n d we have

R R
j 2 > X 1 / \2 \ i 1 2

Since f R e~ï x
dx = v27r, therefore T ( e~ 2 (•) j (<j) = yj2ire~^ ' 0
. I n the same
m a n n e r we c a n see t h a t

T (c"* M 2
) (CT) = (27r)f e - 2 k l , 2
a € R . n
(6.3.3)
W e also define

T- f[o) x
= J e- ^f(x)dx. i
(6.3.4)

DEFINITION 6 . 3 . 2 . T h e f u n c t i o n T~ f x
is s a i d to be the inverse F o u r i e r trans-
f o r m of / .

THEOREM 6 . 3 . 2 (Parseval). / / / and g are in L ( R ) , 1 n

then

J Tf(a)g(a)da = J f(x)Tg(x)dx, (6.3.5)

Rn
R™

j F- î{a)g{a)do
l
= jf(x)T- g{x)dx. 1
(6.3.6)
R n
R n

PROOF. T h e formulas ( 6 . 3 . 5 ) a n d ( 6 . 3 . 6 ) are a n i m m e d i a t e consequence of

Fubini's theorem. •

W e are n o w i n a p o s i t i o n to prove some F o u r i e r t r a n s f o r m i n v e r s i o n t h e o r e m .

THEOREM 6 . 3 . 3 . / / / , Tf are in L ( M )
1 n
; then

(T- 1
o T)f = (T o T- )f X
= /. (6.3.7)
 6. TEMPERED DISTRIBUTIONS AND FOURIER TRANSFORMS
100

PROOF. N o t e t h a t i f / a n d Tf b e l o n g to Z ^ Q E T ) , t h e n T~ f x
belongs t o
L ( R ) , as well. P u t b y definition
1 n

G (a) xe : = e -i(^)-h \*\\ 2

W e s h a l l n o w a p p l y the F o u r i e r t r a n s f o r m a t i o n w i t h respect t o a. T h u s , we have

g (y) ex •= J e ^ e - ^ - ^ W 2
da = J e^v-^) ~h W\ e
2 2
d(7_
R n
R™

P u t t i n g cu = ea we o b t a i n

g (y)
ex = e- J n
e ^ - ^ e - ^ l 2
^ .
R 71

B y (6.3.3) we get

g (y) = nn _i
t- (^)ïe-î y — x

ex

It is easy t o check t h a t ( 2 7 r ) _ n
fRn g (y)dy
ex = 1 for x G R . Because of (6.3.5) we n

have

J Tf{a)G {a)da ex = J f{y)g {y) ex dy. (6.3.8)

R n
R n

P u t t i n g ô (x — y) — (27r)~ g (y)
e we o b t a i n J
n
ex Rn ô (x — y) dy = 1 for a l l x G R .
e
n

N o w , e q u a l i t y (6.3.8) c a n be w r i t t e n as follows

j Tf(a)G (a) ex do- = (2ir) (f n

* 6 )(x).
€

R n

B y v i r t u e of L e m m a 1.2.1 we infer t h a t f Rn Tf(o~)G (o~) ex da tends t o (2Tc) f n

as
e - » 0 i n L ( R ) . O n t h e other h a n d we have
1 n

JTf(a)G (a)da ex = J T f ( a ) e - ^ ~ ^ 2
da
R" R™

= (2^) (27r)- n n
J Tf(a)e- ^~i ^ i{x e2 2
da.
R n

.I ,2
Since J * / is i n L ( R ) a n d Tf(a)e~ < ^'~^
1 n
k l tends t o Tfi^e"^^ i( x
as e -> 0,
b y the Lebesgue d o m i n a t e d convergence t h e o r e m we deduce t h a t

j Ff(<T)G (a) ex da -+ ( 2 T T ) ^ ( ^ - O T)f(a 1

for x G R . T h i s i m p l i e s t h a t [T
n
o T)f{x) — f(x) almost everywhere i n R . I n 1 n

the same m a n n e r we c a n prove the second p a r t of (6.3.7). •

6.4. F o r m a l p r o p e r t i e s of F o u r i e r t r a n s f o r m s

F r o m n o w on, we s h a l l use a l t e r n a t e l y the n o t a t i o n s Tf or / as t h e F o u r i e r t r a n s f o r m

of / . W e list t h e f o r m a l properties of T i n t h e following
 6.4. FORMAL PROPERTIES OF FOURIER TRANSFORMS 101
THEOREM 6.4.1. / / / and (-) »f e
are inL ^ ), 1 71
then

(d »f)(a)e
= (7^f(a). (6.4.1)

/// and d »f e
are in L ( R ) , then x n

d^f = (-ia)^f(a). (6.4.2)

/// 25 in L ( M ) 1 n
7 then we have

:
f~f{<x) = e ^f(<r) i
(6.4.3)
for each x G l " (see Example 1.8.4)-
If f is in L ( R ) then we have
1 n
;

¿Mf(*)=r- f(a) X (6.4.4)

for x e R . n

If f is in L ( R ) and K : R
1 n n
—» R n
¿5 an orthogonal transformation, then

T¿K(<r) = (foK)(v). (6.4.5)

If / is in L ( R ) and S f{y)
x n
x = X f{Xy),
n
X>0, then

(^f(a) = f(^cry (6.4.6)

If f and g are in L (W ) 1 l
J then

f * g{o~) — f(o~) g(a) (Borel formula). (6.4.7)

PROOF. A d (6.4.1). N o t e t h a t

(f(a + he )v - f(a))h^ = h' 1

J ( e ^ ^ - é^)f(x) dx.

Since

• *
o
therefore |( *(*>^) - l ) ^ " 1 < \x \ for h ¿ 0. O n the other h a n d ( ^ ^ - l ) / * "
e
1
v e
x 1

tends to (ZÎC)" as / i -> 0. F r o m this, b y the Lebesgue d o m i n a t e d convergence

6

t h e o r e m we o b t a i n

lim(/(o- + /ie )-7(<j))h" == I/

1
f e ^(ix) ^f(x)dx.
i{x e

J
R
n

T h i s completes the p r o o f of (6.4.1).

 6. TEMPERED DISTRIBUTIONS AND FOURIER TRANSFORMS
102

A d (6.4.2). W e c a n w r i t e x a n d a i n t h e following f o r m X (*^ 7 Xu ) a n d

a — ( a ' , a v ) , where x' a n d a' are i n R n _ 1

. N o t e t h a t the following equalities

J e ^ d -f(x)dx=
i a) e
j e * '
i( # ff/)
I j e ^^ d ^f(x)dxA
i ) e
dx'
R71
Rn-l \R /

=j e***''*') I J(-io-) -e ^^f(x)dxA e i

dx'
R™- 1
\R J

are true. T h i s proves (6.4.2).

T h e formulas (6.4.3) a n d (6.4.4) c a n be o b t a i n e d b y direct s u b s t i t u t i o n i n t h e
definition f o r m u l a of F o u r i e r t r a n s f o r m .
T o prove (6.4.5) we c o m p u t e the i n t e g r a l

j e^'"\foK){y)dy = J f(Ky) dy.

R n
R n

T a k i n g cu = Ky we get y — K UJ, T
where K T
denotes the transpose m a t r i x of K.
We obtain

J7K((t) = J e ( "^f(uj)dw
i KT
= J e ^ ^f(co)duj i K
= (foK)(a).
R n
R n

F o r m u l a (6.4.6) c a n be d e t e r m i n e d b y a s i m p l e verification. T h e p r o o f of (6.4.7)

is a n a p p l i c a t i o n o f F u b i n i ' s theorem. •

6.5. Fourier transforms of functions i n S

It is easily seen t h a t the following formulas

d p(a)
a
= (7pï>((7) (6-5.1)

d^p{a) = (-ia) p(a) a

(6.5.2)

are true i f p is i n S. W e are able t o prove t h e following

THEOREM 6.5.1. If p is in S, then Tp is in S, too. Moreover, the mapping

' p^> Tp (6.5.3)

is continuous.

PROOF. O n account of (6.5.1) a n d (6.5.2) we have

^dPpia) = a (i^p{a)a
- i^^d^Yp){a). (6.5.4)

Since d {{'Yp)
a
is i n <S, therefore b y T h e o r e m 6.3.1 we conclude t h a t d@p> is c o n -
t i n u o u s a n d a d <p is b o u n d e d i n W . T h i s means t h a t p> belongs t o S. It r e m a i n s
a l3 1
 6.6. FOURIER TRANSFORMS OF FUNCTIONS IN L (M ) 2 N
103
to prove t h e c o n t i n u i t y of (6.5.3). I n accordance w i t h (6.5.4) for ¡3 > a we have

sup \a d^(a)\= a
sup \d¿(W<p)(*)\ < [ \d ((-)^)(x)\dx
a

aeR n
crÇR" J
R n

¿ 0<n<a k=l

^ K
E () a
f[Pk(Pk-i)---(fo-nk+i) sup((i+i^i n^-^ -^(^)i, 2 a

where K = J „ ( 1 + \x\ )~ dx, R

2 m
m > ^ . G e n e r a l l y , one c a n show t h a t there exists
M such that
I K - r ^ Ê l k - <M
Yl ll^0 ~'VIU~> a

0<fj,<a
where is a p o l y n o m i a l . T h i s i n e q u a l i t y i m p l i e s the c o n t i n u i t y of (6.5.3) a n d t h e
p r o o f is complete. •

REMARK 6.5.1. F o r m u l a (6.3.7) is true i f / is i n <S.

W e are able t o prove a stronger t h e o r e m t h a n T h e o r e m 6.3.1.

THEOREM 6.5.2 ( R i e m a n n - L e b e s g u e l e m m a ) . Iff is in L ( R )

1 n
7 then Tf is in
Co(R ). n

PROOF. I n v i e w of T h e o r e m 1.5.1 there exists a sequence ( / „ ) , fv G V con-

vergent i n L ( E ) t o / . I n accordance w i t h T h e o r e m 6.5.1, Tf
1 n
G S. T a k i n g i n t o v

account T h e o r e m 6.3.1 we infer t h a t Tf converges u n i f o r m l y i n R t o Tf. F o r

v
n

given e > 0 there exists VQ G N such t h a t

\rfA<r)-rf(*)\<¿

for a G R , v > v$. Since Tf

n
is i n S therefore \T f(o~)\ v u < § for \a\ > Q-Q. T h i s
implies t h a t \Tf(o~)\ < e i f \a\ > o~ T h u s , t h e p r o o f is 0 finished. •

6.6. F o u r i e r t r a n s f o r m s o f functions i n L ( R ) 2 n

L e t / be i n L ( R ) . P u t }{x) : = f{-x).
1 n
W e s h a l l n o w show t h a t

? = / . (6-6.1)
N o t e t h a t ab = ab for a, 6 G C . L e t / be a simple integrable f u n c t i o n . It is easily
seen t h a t

f(<r) = J e ^')f(z)dz i
=j e- ^f(z)dzi
= J e^f(-y)dy = ](a).

In t h e general case, we get (6.6.1) b y l i m i t p r o c e e d i n g i n L- i m

( Rn >) . n

W e are n o w i n a p o s i t i o n t o prove t h e following

THEOREM 6.6.1. If ip is in S, then

\\T<p\\ 2 = ( 2 T T ) * | M | 2
L L (6.6.2)
and

||^-V||L» - (2ir)-*\\<p\\ 2.
L (6.6.3)
 6. TEMPERED DISTRIBUTIONS AND FOURIER TRANSFORMS
104

PROOF. L e t (f b e l o n g t o S. T h e n the following equalities

MÛ* j \(f(x)\ dx 2
= J (f(x)(f(x)dx = J (f(x)(f(—x) dx = (p * (f)(0)

h o l d . O n t h e other h a n d we have

( 2 7 r ) - J f^(x)
n
dx = (27r)~ j n
f^(x)e~ ^ i(x
dx

F r o m this i t follows t h a t

M i l * == (27r) ^ (fTp(x)dx.
n

By (6.4.7) a n d ( 6 . 6 . 1 ) we o b t a i n ( 6 . 6 . 2 ) . P u t t i n g T~ p x
i n t o ( 6 . 6 . 2 ) we get (6.6.3).
T h e p r o o f is complete. •

L e t / b e i n L (R ). I n accordance w i t h T h e o r e m 1 . 5 . 1 there exists a sequence

2 n

(fu), Pv £T> such t h a t p tends t o / as v —> oo i n L (R v ). Because of ( 6 . 6 . 2 ) t h e 2 n

sequence (Tf) ) is a C a u c h y sequence i n L (R ).

v Suppose n o w t h a t (f G S a n d <ç
2 n
v u

converges t o / i n L (R ). T h e n Tf 2
converges t o some f u n c t i o n g also i n L (R ).
n
v
2 n

A f t e r t h e above p r e l i m i n a r i e s we shall i n t r o d u c e t h e following definition.

DEFINITION 6 . 6 . 1 . T h e f u n c t i o n g above defined is c a l l e d t h e F o u r i e r ( P l a n -

cher el) t r a n s f o r m of / .

To verify the consistence of this d e f i n i t i o n we have t o show t h a t i f the sequences

((fl) a n d ((fl) converge t o / , t h e n the sequence (T(f \ — T(f ) converges t o zero i n 2
v

L (R ).
2 n
T h i s fact is a consequence of ( 6 . 6 . 2 ) . S i m i l a r l y we c a n define t h e inverse
F o u r i e r t r a n s f o r m of any f u n c t i o n / b e l o n g i n g t o L (R ). 2 n

DEFINITION 6 . 6 . 2 . Suppose t h a t p u e S a n d the sequence (<p ) converges t o / u

i n L (R ). 2 n
T h e n we take
= l i m T~ (f T~ f x x
v

v—ïoo
To verify t h e consistency of this d e f i n i t i o n we a p p l y ( 6 . 6 . 3 ) .

THEOREM 6 . 6 . 2 ( P l a n c h e r e l ) . / / / is in L (R ), 2 n
then

(T- oT)f
1
= (ToT~ )f 1
= f, (6.6.4)

\\Tf\\ * = L (27r)î||/|| 2 L and f\\ 2 L = (2TT)-*||/|| 2. L (6.6.5)

PROOF.
L e t p be i n S. A s s u m e t h a t t h e p converge t o / i n L (R ).
v Then v
2 n

the sequences ((T~ o T)p ) a n d ((J o J ~~ )f )

x
v converge t o (J ~ o T)f a n d (T o
7 r 1
iy
7 1

T~ )fY
i n L (R ), 2
respectively. Since ( J
n
o T)<p = (T o T~ ))(f = (f , v =
7 - 1
v
x
v v

1 , 2 , . . . , therefore ( 6 . 6 . 4 ) is true. E q u a l i t y ( 6 . 6 . 5 ) is a n i m m e d i a t e consequence o f

the c o n t i n u i t y of the f u n c t i o n L 3 <p —y \ \ P W L - T h i s finishes t h e proof.
2
• 2

THEOREM 6 . 6 . 3 (Parseval). / / / and g are in L (R ), 2 n

then

¡ ^ 9 = / f^g. (6.6.6)
 6.6. FOURIER TRANSFORMS OF FUNCTIONS IN L(E ) 2 N
105
PROOF. L e t <p , ip , v = 1, 2 , . . . be i n S. A s s u m e t h a t the sequences {ip ) a n d
u u v

(?/v) converge to / a n d g, respectively i n L ( R ) . B y T h e o r e m 6 . 3 . 2 we have

2 N

W h a t is left to show is t h a t J Rn (p ip
u u and f Rn ipvijj v converge t o J Rn Tjg and
f Rn fTg respectively. I n fact, we have

< | | J 7 I U * h - ^ I U ' + M\\Ff - ^|| 2, L

where M — s u p ^ G N H ^ I U - T h e rest of the p r o o f is evident.

2
•

W e s h a l l n o w show t h a t i f / belongs b o t h t o L ( R ) a n d L ( R ) , t h e n D e f i n i ­ 1 N 2 N

tions 6 . 3 . 1 a n d 6 . 6 . 1 are equivalent. Indeed, assume t h a t / is i n L (R ) DL (M ). 2 N 1 n

F u r t h e r , let Tf denote the F o u r i e r t r a n s f o r m of / i n the sense of D e f i n i t i o n 6 . 6 . 1 .

Let us r e g a r d for a m o m e n t / as a n element of L ( R ), t h e n b y T h e o r e m 6 . 3 . 2 1 N

for each (p i n S. I n this case Tf denotes the F o u r i e r i n t e g r a l of / . O n the other

h a n d we c a n regard / as a n element of L ( R ), t h e n the above e q u a l i t y m a y be
2 N

u n d e r s t o o d i n the sense of T h e o r e m 6 . 6 . 3 . F r o m this we conclude t h a t Tf is the

F o u r i e r i n t e g r a l of / .

Let X[-v,v] denote the characteristic f u n c t i o n of the set

{x : x G R , —v < Xj < v, j = 1 , . . . , n } .
N

The next t h e o r e m is very i m p o r t a n t for a p p l i c a t i o n s .

THEOREM 6 . 6 . 4 . // / is in L (R ), 2 n
then X[-v,v]f converges to Tf in L ( R ) .
2 N

PROOF.. O f course, X[-u,u]f belongs to L ( R ) , 1 N

therefore

X^]f(c-) = J e^ ^ (x)f(x)dx.
X[ ]

N o t e t h a t (x\-v,v]f) is a C a u c h y sequence i n L (R ). 2 N
Put

/* = lim X[-v,v]f-

We have to show t h a t / * = Tf. L e t ip G <S, t h e n by T h e o r e m 6 . 3 . 2 we have

J X[-v,v]f<P = J X[-v,v]f$>

A passage t o l i m i t i m p l i e s t h a t

j ' / V = / /£• (6-6-7)

 6. TEMPERED DISTRIBUTIONS AND FOURIER TRANSFORMS
106

O n the other h a n d b y T h e o r e m 6 . 6 . 3 we have

ff$ = f F f<P- (6.6.8)

F r o m ( 6 . 6 . 7 ) a n d ( 6 . 6 . 8 ) i t follows t h a t / * = Tf. T h u s , the p r o o f is finished. •

T o determine F o u r i e r transforms of some functions we often use t h e following

LEMMA 6 . 6 . 1 ( J o r d a n ) . Let f be a continuous function in the upper or lower

half plane. If a > 0 and l i r r i | | _ f(z) = 0 or a < 0 and lim^^^
z > 0 0 f(z) = 0, then
%z<0
Sz>0
lim e i<TZ
dz = 0 or l iim
m / f(z) e l(TZ
dz = 0 ,
7 >00 -+00 J

where K+ = {z : z G C , \z\ = r , Sz > 0 } and K~ = {z : z G C , \z\ = r , %z < 0 } .

PROOF. W e o n l y shall prove t h e case, w h e n / is defined i n the u p p e r h a l f plane.

Put M(r) : = s u p t € [ 0 > 7 r ] \f{re )\.
u
T h e n we have

7T
dt.
j f{z)e iaz
dz < M(r)r j
KÍ

Since l i m ^ o o M(r) = 0 , therefore w h a t is left t o show is t h a t t h e i n t e g r a l

r

r ~e dt is b o u n d e d for fixed a > 0 . A n easy c o m p u t a t i o n shows t h a t

a r s x n t

7T 2
dt.
r J e~ (Trsint
dt = 2r J <

N o t e t h a t sin¿ > f for 0 < t < f . Therefore e - a r s

' m t
< e " ^ 1
for 0 < t < §.
Hence, we o b t a i n

^ . 2art , 7T
2r / e
7 <
~ di < - . •
EXAMPLE 6 . 6 . 1 . L e t f(x) = W e s h a l l now c o m p u t e the F o u r i e r t r a n s f o r m
of / . N o t e t h a t for a > 0 we have

1 / e ^ - ^ — d x + - ^ / e"*- dz= -e""

1+ z 2 2

27T2 . / 1+ X 2
27T2 y
*7
if r > 1. Hence, b y L e m m a 6 . 6 . 1 a n d T h e o r e m 6 . 6 . 4 we o b t a i n

Ff(a) = 7Tie~ a

for a > 0 . S i m i l a r l y , one c a n show t h a t

Tf(a) = -Trie* 7

if a < 0 .
 6.7. FOURIER TRANSFORMS OF THE HERMITE FUNCTIONS 107
EXAMPLE 6.6.2. P u t f(x) : = ¿ r ^ b y , ^ 0. F i r s t we suppose t h a t $sz > 0.
T h e n we have
r

27TZ J x —z 2m J £—z

for <J > 0 and

r
_ L f i<rx_J_
e dx + 1 Í e ^ _ L _ d ç = {)

2m J x —z 2m J í — z

for a < 0, where \z\ < r. F r o m this b y T h e o r e m 6.6.4 a n d L e m m a 6.6.1 we o b t a i n

if > 0, where H is H e a v i s i d e step f u n c t i o n . S i m i l a r l y , a s s u m i n g t h a t ^sz < 0 we

g ^ ( ¿ F i ) ) W = - f f
H e
' f f
'

6.7. Fourier transforms of the H e r m i t e functions

F i r s t we r e c a l l t h e d e f i n i t i o n of the H e r m i t e functions.

DEFINITION 6 . 7 . 1 . T h e function

H [x)
v = (-l)"e (e- )("\ x2 x2
v = 0,1... (6.7.1)
is called t h e H e r m i t e p o l y n o m i a l .

C a r r y i n g out differentiation i n ( 6 . 7 . 1 ) we find HQ(X) = 1, Hi(x) = 2 x , H (x) = 2

4x 2
— 2 , . . . A simple verification shows t h a t

2 k\y/ñ
k
i f /x = 1/ = fc.
R

Set
:= (2 k\y/Z)~ie-'*-H (x).
k
k

It is k n o w n t h a t t h e functions hk, k = 0 , 1 , 2 . . constitute a complete o r t h o n o r m a l

s y s t e m i n L ( R ) . It is called the H e r m i t e s y s t e m . T h e following differential o p e r a t o r
2

sL d _ x i d
A — e 2
—e = - 2
x
dx dx
w i l l be useful t o determine the F o u r i e r t r a n s f o r m of h . N a m e l y , we s h a l l show t h a t k

Ah k = -(2(fc + l ) ) * h f c + i, k = 0,1,2,... (6.7.2)

Indeed, i t is easy t o check t h a t

h' = (-l) (2 k\^)-i

k
k k
(xe^{e- ) V x2 {
+ e ^ ( e - x 2
) ^ ) .

O n t h e other h a n d we have

-xh (x)
k = (-l) (2 klViï)-i
k+1 k
(xe^{e- )^) x2
.

F i n a l l y we have (6.7.2). I n accordance w i t h (6.4.1) a n d (6.4.2) we have

T(Ay) = iAFip, (peS. (6.7.3)
 6. TEMPERED DISTRIBUTIONS AND FOURIER TRANSFORMS
108

W e are n o w . i n a p o s i t i o n to prove the following

THEOREM 6 . 7 . 1 .
Th k = 2ixi h , k
k k = 0,1,2,... (6.7.4)

PROOF. O f course, h = e * a n d Tho — 2irho. A s s u m e now, t h a t ( 6 . 7 . 4 ) is

0

true for n = k, k > 1. B y v i r t u e of ( 6 . 7 . 2 ) we get h +i = — (2(fc + Ah . In k k

accordance w i t h ( 6 . 7 . 3 )

Th k + 1 = -(2(fc + l))"niHfc.
F r o m this, b y the i n d u c t i o n a s s u m p t i o n we get

jr h k + 1 = _( (* + l))-ii
2
f c + 1
27rA^.

U s i n g ( 6 . 7 . 2 ) again we o b t a i n Th +i k = 2iri h j \. kJtl

k r T h i s statement finishes the
proof. •

6.8. F o u r i e r t r a n s f o r m s of t e m p e r e d d i s t r i b u t i o n s

W e r e c a l l n o w t h a t i f / is i n L ( E ) or L ( R ) , t h e n we have
1 n 2 n

R" R n

for p G 5 . T h i s leads us to

DEFINITION 6 . 8 . 1 . T h e F o u r i e r t r a n s f o r m TT ( T ) of the d i s t r i b u t i o n T G
<S'(R ) is the image of T under the transpose of the m a p T : <S(R ) -> <S(R ); i.e.
n n n

JTfa) = T ( ^ ) (6.8.2)
for p G 5 . S i m i l a r l y , one c a n define the inverse F o u r i e r t r a n s f o r m T~ T X
of t h e
tempered distribution T taking
F-\T{v))=T{T- 1
(y,)). (6.8.3)

It is easily seen t h a t i f T is i n L ( R ) or L ( R ) , t h e n the classical definitions

1 n 2 n

of TT a n d D e f i n i t i o n 6 . 8 . 1 are equivalent (see T h e o r e m s 6 . 3 . 2 a n d 6 . 6 . 3 ) .

THEOREM 6 . 8 . 1 . For a tempered distribution T the following equalities

(T'1
o T)T = (T o T~ )T X
= T (6.8.4)

hold.

PROOF. TO prove this statement we o n l y need to observe t h a t

{T~ l
o F)T( )V = {T- {TT)){v)
l
= - T{ToT- ){ )
x
9 = T{ ). v

T h e second p a r t of ( 6 . 8 . 4 ) c a n be proved i n the same way. •

THEOREM 6 . 8 . 2 . The Fourier transformation T : <S'(R ) - » <S'(R ) is

n n
contin-
uous.

PROOF. L e t T a n d T be i n <S'(R ), X¡, tends to T as ^

v oo i n <S'(R ). T h i s
n n

means t h a t T (p) —> T(<p) as z/ —» oo for each p G 5 . Therefore

v

= T„(£) T ( 0 ) = T(<p)
as z/ —» oo. T h i s finishes the proof. •
 6.9. FORMAL PROPERTIES OF FOURIER TRANSFORMS. 109
EXAMPLE 6.8.1. W e s h a l l n o w determine the F o u r i e r t r a n s f o r m T-^ of We
know that is i n V'L2 ( R ) . L e t C-^y be the C a u c h y t r a n s f o r m of ^ y . B y v i r t u e of
T h e o r e m 5.7.2 we infer t h a t

C^(- + i e ) - C ^ ( - - i e ) ^ - ^

as e -> 0+ i n V (R). f
L2 Therefore C ^ y (• + ie) - C ^ y ( - - ie) tends to ^y as e 0+ i n
<S (R). It remains to determine the F o u r i e r transforms of the functions C ^ y ( - + ie)
;

and — ie). I n accordance w i t h (5.8.6), C-^(z) = ¿ i f ^2: > 0 a n d C - ^ y ( z ) —

— 7^ i f 3 z < 0. U s i n g T h e o r e m 6.6.4 a n d L e m m a 6.6.1 we o b t a i n

'0 for a > 0

1 i e ^ - i— dx —
z + ie 1 -7rie for cr < 0
ecr
2 y

and
7rie~ e<T
for a > 0

i2 y/ - x - ie I0 for a < 0.
R V

T o o b t a i n the F o u r i e r t r a n s f o r m jF-^y of -^y we c o m p u t e

1 (^ 1 , , ^ 1 , A ivri i f cr > 0
lim - — (a) + F - — cr = i
6->o+ 2 V (- + i e ) (--ie) 7 v
[-Tri
; V
i f cr < 0.
F i n a l l y we have
JF—(cr) = 7ri sgn cr.

6 . 9 . F o r m a l p r o p e r t i e s of F o u r i e r t r a n s f o r m s of t e m p e r e d d i s t r i b u t i o n s

W e s h a l l now show t h a t formulas (6.4.1), (6.4.2), (6.4.3) a n d (6.4.4) are m e a n i n g f u l

for t e m p e r e d d i s t r i b u t i o n s .
THEOREM 6.9.1. If T is a tempered distribution, then the following equalities
hold:

D »fe
= ( i ^ T , (6.9.1)

D^T = (-i-) "f, e

(6.9.2)

TxT = e i M
f for each x GR , n
(6.9.3)

^MT = r_ T x for each x GR . n

(6.9.4)

PROOF. F o r m u l a s (6.9.1) a n d (6.9.2) are a n i m m e d i a t e consequence of (6.4.1)

a n d (6.4.2), respectively. Indeed, we have

D »T(ip)
e
= -T(d "<p) e
= ~T(d^p) = -T((-i.) "£) c

= (i-) "T(¡p)
e
= ( i O ^ H ,

W e s h a l l now show t h a t equalities (6.9.3) a n d (6.9.4) are also true. I n accordance

w i t h E x a m p l e 1.8.4, (6.4.3) a n d (6.4.4) we o b t a i n

^ T ( ^ ) = r T{(p) x = T(r_ £) = T ( e ^ V ) =
x e^fiip)
 6. TEMPERED DISTRIBUTIONS AND FOURIER TRANSFORMS
110

and

¿M~T(<p) = (e^T)^) = T(e^ip) = T(r^p) = T{T >)XV = r_ T(y)).

B

T h e consequence of this reason is easily seen because of R e m a r k 6.1.1. •

6.10. Fourier transforms of integrable distributions

Put
E (x)
x ~ (27r)-tA- n
e -ilA| 2

(compare S e c t i o n 4.3). L e t T be in V , t h e n f
Ll

T (x) =
x (T*E )(x)=T(E (x--)) x x

is a r e g u l a r i z a t i o n o f T (see S e c t i o n 5.2). T h e first t h e o r e m of this section gives us

some g e n e r a l i z a t i o n o f the R i e m a n n - L e b e s g u e l e m m a .

THEOREM 6.10.1. IfT is in W , then L1

JT(or) = T ( e i ( ,
' < r )
). (6.10.1)
Moreover, TT is a continuous function and bounded by a polynomial.

PROOF. I t is k n o w n t h a t E is i n V i a n d f E\ = 1. I n v i e w o f T h e o -
1 L Rn

r e m 5.3.2, T\ tends t o T i n V as A —» 0. B y t h e c o n t i n u i t y of the F o u r i e r

f
Ll

t r a n s f o r m a t i o n i n S'(R ) i t follows t h a t TT tends t o TT i n <S'(R ) as A -> 0. W e

n
X
n

shall now show t h a t TT\(a) converges t o T(e^'^) as À - » 0 at each p o i n t a G R . n

In accordance w i t h T h e o r e m 5.1.4 we have

TT {a)
x = f e«*>°> I E (-l)W f ^ E x ( x - y)df, (y)\ a dx
M« \l«l<"* Rn /

= E / ( / e ^ ° X - i r ^ E x { x - y ) d x \ d» (y)
a

\a\<m R „ \ „ R /

= £ / W ( / e ^E (x-y)dx\ i
x dfi (y)
a

\a\<m p_ n \pn )

= E f{^) ^ '" a v )
( /e^E {uj)du\ x dti (y)
a

\a\<m R „ \ R „ /

= E (- )' 1 Q|
f{i*) e a i { y
' a )
e-^ 2
d» (y). a

\a\<m R n

Therefore TT\(a) converges t o

E ("1) H
f(ie) e ^d (y)=
a i
lia E /l^^^^)
\a\<m R „ |a|<m R „

= T(e (-' < <T)

).
Note that

\FTx(<r)\< E ki| ---kn| "|/ia|(R ).

ai a n
 6.10. FOURIER TRANSFORMS OF INTEGRABLE DISTRIBUTIONS 111
W e s h a l l now show t h a t

\?T (<T)<p(<T)\<M (l
X v + \<T\ )- ,
2 k

where k > T o prove this, it suffices to r e m a r k t h a t

\FT (v)<p(<r)\ < \<p(a)\(l + \<r\ ) (2n)- 5

x
2 k !
£ M " 1
• • • K | " ( l + W\ )~ .
a 2 k

\ot\<ra

T h e n we c a n set

M ^ : = s u p \ {a)\{l
V + \a\ ) {2^ 2 k
V ) | M A |(R") ••• K l " " -
\ct\<m

A f t e r these statements, b y the Lebesgue d o m i n a t e d convergence t h e o r e m we infer

that

J TT (a)ip(a)x da - » J T(e^' M<r) a )

as A —> 0 for each <p G S. T h e p r o o f w i l l be c o m p l e t e d b y s h o w i n g t h a t TT is

continuous. T h e c o n t i n u i t y of TT is easy to see from the representation f o r m u l a

TT(a) = J ] ) (-1)1°! J ^e^d^iy). •

|a|<~~

THEOREM 6 . 1 0 . 2 . // S and T belong to ^' i^ny L then S * T is in £>^ 1(Rn) and

the Borel formula
T(S * T)(a) = TS(a) TT(a) (6.10.2)

holds.

PROOF. Indeed, we have

T(S * T){a) = (S ® Ty)(e^ ^) x

x+y
= S ^ ^ S y ^ ^ )
= TS(a)TT(a).

T h i s finishes the proof. •

REMARK 6 . 1 0 . 1 . L e t S a n d T be i n T> ^ ^ f
Ll Rm a n d î ^ i ^ n ) , respectively a n d let
TS a n d F T be their F o u r i e r transforms. T h e n a s i m p l e verification shows t h a t

T{S®T)(t,r )=TS(£)TT{ ). 1 n

In other words, the F o u r i e r t r a n s f o r m of the tensor p r o d u c t of integrable d i s t r i b u -

tions is the tensor p r o d u c t of their F o u r i e r transforms.

EXAMPLE 6 . 1 0 . 1 . It is clear t h a t To = 1 for S G £>'(R). Therefore for 8 a G

D'(R )wehave
n

T6 a = F(6 ®--'®5 )
ai an = l,

where a = ( < 7 i , . . . , a ).n

 6. TEMPERED DISTRIBUTIONS AND FOURIER TRANSFORMS
112

6.11. F o u r i e r t r a n s f o r m s of square i n t e g r a b l e d i s t r i b u t i o n s

A d i s t r i b u t i o n b e l o n g i n g to V' from henceforth w i l l be c a l l e d a square integrable

L2

d i s t r i b u t i o n . D e n o t e by S' the vector space spanned b y the set of a l l functions

0

( • ) 7 , where / is i n
;
L (R ). 2 n

DEFINITION 6.11.1. A f u n c t i o n / b e l o n g i n g t o <S¿ w i l l be called slowly increas-

ing or t e m p e r e d f u n c t i o n .

THEOREM 6.11.1. TV' c S' and TS^ c L2 0 V' . L2

PROOF. L e t T G V , t h e n i n accordance L 2 w i t h T h e o r e m 5.4.1 we have

n<p)= E / ( E Da
f<*) (*>)
\a\<m R n \|a|<m /

where D f is the a - d i s t r i b u t i o n a l derivative of f

a
a a b e l o n g i n g to L (R ) a n d (p is
2 n

i n S. A c c o r d i n g t o D e f i n i t i o n 6.8.1

TT(<P) = i E Da
f«) @) = E jf*9 $= a
E / / « ( í ó ^ -
\|a|<m / H<m R n \a\<m R n

B y T h e o r e m 6.6.3 we have

f &)=
T
Y (" )' 1 a |
f(ix) Ffa{x)<p(x)àx.
a

\a\<m R n

T h i s i m p l i e s t h a t TT is i n SQ. Conversely, i f T belongs t o <S¿, t h e n

T(<p)= Y J X f (x)<p(x)dx,
a
a

\a\<m Rn

where f a G L (R )
2 n
a n d p> £ S. Therefore

TT(p) = Jx«f (x)p(x)dxa = Y Ji (-ix) fa(x)p(x)dx

lal a

\oc\<m Rn M< Rn m

Y J i fa(x)d^p(x)dx.
la]

|a|<m , R

U s i n g a g a i n T h e o r e m 6.6.3 we get

TT{v)= E (-1) H
J(-i) Ff (x)d <p(x)dx.
lal
a
a

\ot\<m RTX

Since f a is i n L (R2 n
), therefore J F T is i n V' . L2 T h i s statement finishes the proof. •

T h e next t h e o r e m is v e r y useful g e n e r a l i z a t i o n of T h e o r e m 6.10.2.

THEOREM 6.11.2. If S and T are in V' , then L2

S7T = ST. (6.11.1)

 6.11. F O U R I E R T R A N S F O R M S O F S Q U A R E I N T E G R A B L E D I S T R I B U T I O N S 113

P R O O F . L e t us b e g i n b y p r o v i n g (6.11.1) for S — / , T = g, where / a n d g are

i n L ( R ) . I n accordance w i t h D e f i n i t i o n 6.8.1 we have
2 n

for (f e S. F i r s t of a l l we s h a l l show t h a t / * g is i n C ( R ). Indeed, let f a n d 0

n
v

g be i n S for each v G N . A s s u m e t h a t t h e sequences (f ) a n d (g ) converge t o

v v v

/ a n d g, respectively i n L (R ). O f course, f * g is i n <S, therefore b y Y o u n g ' s

2 n
v v

i n e q u a l i t y we get

1 1 / < \\U - / I k 2
IMIL* + M | | ^ - ^|| 2,
L

where M = s u p ^ ^ H / ^ l ^ } . T h i s implies t h a t / * g G C ( R ) . F u r t h e r , / 0
n
R n (/„ *
g )(p tends t o J
v R n ( / * g)(p as z/ —» oo. B y (6.4.7) a n d T h e o r e m 6.6.3 we get

R n

O n the other h a n d we have

WUg, -?f?g\W < Wfv ~ ffh* W^gh* + { ^ M \ \ g v - r \\ .

g L2

In accordance w i t h T h e o r e m 6.6.2, f g v v tends t o TfTg in L ( R ) . 1 n

T h i s implies
that f Rn fi/g^ip tends t o f Rn fg<p as v -> oo. F i n a l l y we get

J(f *g)$ = j fg<p

R n
R n

for f,g i n L ( R ) . T h i s means t h a t (6.11.1) is true i f S a n d T are i n L ( R ) .

2 n 2 n

W e now t u r n t o the general case. B y (5.5.5) we have

(S*T)(<p)= J2 Jfa(x)g (y)d ^(x

0
a+
+ y)dxdy, (6.11.2)
\a\<k\¡3\<l Rn

where f a a n d gp are i n L ( R ) a n d ip e S. U s i n g t h e F u b i n i t h e o r e m we o b t a i n
2 n

J j fa(x)g(3(y)(p(x + y)dx dy = j f (x)

a I Jgp(y)<p(x + y) dy J dx
Rn R n
Rn \Rn /

da;
= J fa(x) Í Jgp(uj - x)(p(uj)duj \ dx = J I ^ ^ ( w - x)f (x)dx a j
R 7 1
\ R n
/ Rn \Rn /

= j {fa * gp){uj)ip(uj)duj.

Therefore e q u a l i t y (6.11.2) c a n be w r i t t e n as follows

(S * 2 > = E (-l) | a + / ? l
/ ( f a * g )(x)d ^(x)
0
a+
dx. (6.11.3)
 6. TEMPERED DISTRIBUTIONS AND FOURIER TRANSFORMS
114

F r o m this we get

T(S*T)(ip) = (S*T)(£) = E ^ ) ' ^ ' 1

f{fa*9fi)& ¡p(x)dx
a+P

\<*\<k\/3\<l ¿

H<k\/3\<1¿

T a k i n g i n t o account the first p a r t of the p r o o f we o b t a i n

^(S*T)te)= E E ÍU(cr)g^(a)(-iar ^(a)da +

\c\<k\0\<l¿

E E ÍD^Tai^D^siaM^da.
|a|<fc|/3|<ÍRn

Since 5 = S | a | < A ; D f a
a a n d T = !C|/3|<z D^9/3> therefore (6.11.1) generally h o l d s .
T h i s finishes the proof. •

A s i m p l e verification shows t h a t

r-ifr * V ) (v) = ( 2 T T ) " ( ^ - V ) ( t r ) ( ^ - V ) W

if <¿? a n d -0 are i n 5 . S i m i l a r arguments a p p l y i n g to the inverse F o u r i e r t r a n s f o r m
give us the following f o r m u l a

T- (S
X
* T ) = {27r) T- ST- Tn 1 1
(6.11.4)
for 5 a n d T i n £>^ . 2

T h u s , we o b t a i n the following

THEOREM 6.11.3. If S and T are tempered distributions in Lf (R ),

oc
n
then

S T = ^ - T - \ T S * TT). (6.11.5)

6.12. D e t e r m i n i n g F o u r i e r t r a n s f o r m s of s q u a r e i n t e g r a b l e d i s t r i b u t i o n s

F o r m u l a (6.10.1) cannot be used t o d e t e r m i n i n g the F o u r i e r t r a n s f o r m TT of T i f T

is i n T>' 2, because e ^'^ is on-the outside of V i .
L
1
W e c a n choose a new w a y t o d o
L

it. T h e o r e m 5.7.2 indicates such a m a n n e r . N a m e l y , assume t h a t T is i n X>^ , t h e n 2

CT(- + ie) - C T ( - - ie) tends to T as e —^ 0 i n V' . Therefore CT(-+ ie) - CT(- - ie) L2

converges t o T i n S'. B y the c o n t i n u i t y of the F o u r i e r t r a n s f o r m a t i o n i n S' we infer

t h a t C T ( - - f ie) — CT(- — ie) converges to T i n S'. T h u s , we o b t a i n the following

THEOREM 6.12.1. IfT is in V L2 and CT is its Cauchy transform, then

f = l i m ( C T ( - + ie) - CT(- - ie)). (6.12.1)

EXAMPLE 6.12.1. W e s h a l l now determine the F o u r i e r transforms of (. * ) + 0 fc and

(._^ )fc • A p p l y i n g the t h e o r y of residues one obtains the following results
Q

1 da) *- 1 1

T
J T W { ( l ) =
- ^ ( f c V " ^ ( 6
- 1 2
- 2 )
 6.12. DETERMINING FOURIER TRANSFORMS. 115
where H is the H e a v i s i d e f u n c t i o n . N o t e t h a t

\k-i)\ v J
(k-iy. J
(- + io) k

i n <S' as e —» 0. S i m i l a r l y ,

as e —» 0. W e are now i n a p o s i t i o n to d e t e r m i n e the F o u r i e r t r a n s f o r m of j ^ .

Since
1 1
2V(- + ¿0)* (.-i0)*
therefore

In p a r t i c u l a r , for k = 1 we have

J * - Q ( C T ) = Z7T sgn <j. (6.12.5)

N o w , as a n i m m e d i a t e consequence of (6.12.2), (6.12.3) a n d (6.11.1) we o b t a i n the

following f o r m u l a

(• + ¿ 0 ) " (• + iOy (/i - l ) ! ( i / - 1)! (• - f ¿ o ) ^ - 1 :

1 1 (u + v-2)\ 1 ,
(• - ¿0)^ * 7
(• - ¿^0Ñ) "- = (/i - l)!(z/ - 1)! (• - ¿ o )—^ -r ,
V M
¿ — — 7 1 ; (6.12.7)

7 — * 7 — = 0, (6.12.8)
(•-zQ)^ (-4-zO)^ v ;

where / i , € N.

If for some d i s t r i b u t i o n s 5 a n d T the r i g h t side of (6.11.5) is m e a n i n g f u l , t h e n

we c a n define their p r o d u c t b y means of (6.11.5).

EXAMPLE 6.12.2. It is easy t o check t h a t T{l}{cr) = 2nô. B y (6.9.1), =

-2TTÍO^ . W e k n o w t h a t T-^j(a) = in sgn a. N o t e t h a t

(i7rsgn(-) * ( - 2 T T Z ¿ ) ) ( 1 )
= (2TT s g n ( - ) )
2 (1)
= 4TT ¿.
2

H e n c e we have
insgn(-) * (-2TTZ(5) (1)
= 2TTF{1}.
Therefore ^ ( - ) = 1.

EXAMPLE 6.12.3. A c c o r d i n g to (6.12.3) we have

(jF, * * T, * 1 (a) = 4 7 r z ^ — - 2

V
(--¿0)/* (.-lO)"/ (¿HV ;

fzcr ) N /i+l/
~ 1
1
= 2TT • 2-Ki-r— 1
TT7 = 2 T T F 7 — — (cr).

F r o m t h i s , b y (6.11.5) we get
1 1 1
- i$y (• -io)» (. -ioy+v-
 6. TEMPERED DISTRIBUTIONS AND FOURIER TRANSFORMS
116

S i m i l a r l y one c a n show t h a t
1 1 1
(• + iQ)» (- + ¿0)^ (- + ¿ 0 ) ^ '
In p a r t i c u l a r , for \i = v — 1 we have

1
-iOJ (*-i0) ' 2

In v i e w of (5.8.14) we get

6.13. H i l b e r t transforms

In this section we define the H i l b e r t t r a n s f o r m of a d i s t r i b u t i o n a n d show t h a t the

H i l b e r t t r a n s f o r m a t i o n maps V' L2into itself. M o r e o v e r , it w i l l be also p r o v e n t h a t
Sobolev's spaces W ' m 2
are i n v a r i a n t w i t h respect to this t r a n s f o r m a t i o n .

DEFINITION 6.13.1. L e t T be a d i s t r i b u t i o n . T h e n the H i l b e r t t r a n s f o r m HT

of T is defined as follows
« r = - ~ * T . (6.13.1)

THEOREM 6.13.1. The Hilbert transformation is one to one mapping of V'L2

into itself.

PROOF. I n accordance w i t h (6.12.5), T ()

a
— - ¿ s g n c r . F r o m this, b y
(6.11.1) we get

T(HT)(a) = T * Tj (a) = - z s g n a ^ T ( o - ) .

Since T is i n P ^ , therefore b y T h e o r e m 6.11.1 we infer t h a t T(HT)

2 is i n Sf
0 and

H T belongs to V' . L2 Note that

for each a G R . O f course, t a k i n g i n t o account (6.13.1) we conclude t h a t the

following equalities

are true for T G V' . L2 P u t b y definition H" T X

:= ^ * T . T h e n equalities (6.13.2),
because of T h e o r e m 6.8.1 give us the following identities
{H o H~ )T X
= ( 7 - T o H)T
1
= T (6.13.3)

for T b e l o n g i n g to V' . L2 T h i s finishes the proof. •

It is easily seen t h a t / is i n W ' (R) (see S e c t i o n 1.12 a n d T h e o r e m 6.11.1) iff

m 2

(•) Ff
a
e L (R) for a = 0 , 1 , 2 , . . . , r a .
2

N o w , we are able to prove the following

 6.14. CARLEMAN TRANSFORMS 117
THEOREM 6.13.2. The Sobolev space W ' (R), m 2
m = 0 , 1 , 2 . . . is invariant with
respect to the Hubert transformation and

||/||m,2,R = ||W/||m,2,R (6.13.4)

for f G W ' (R).
m 2

PROOF. Note that

TT (•)

Hence, we have

T(D (Hf))a
= -isgno-T(D f)(a) a
= -isgna(-ia) Tf(a).
a

W e see at once t h a t F(D (Hf)) G L (R) i f a = 0 , 1 , . . . , ra. B y T h e o r e m 6.6.2 we

a 2

infer t h a t Hf G W ' ( R ) . E q u a l i t y (6.13.4) is evident.

m 2
•

6.14. C a r l e m a n transforms

T . C a r l e m a n defined i n [4] a t r a n s f o r m of a slowly increasing f u n c t i o n as follows

DEFINITION 6.14.1. L e t / be i n <S (R). Set 0

S i m i l a r l y , we p u t

W e s h a l l n o w prove the following

THEOREM 6.14.1. If f is in S' , 0 then

Tf(z) = (CoT)f(z) (6.14.1)

for z G C — R , w/iere t/ie symbols T and C denote the Fourier transformation and
the Cauchy transformation, respectively.

PROOF. F i r s t we prove a reduced f o r m of this theorem. N a m e l y , we assume

t h a t / is i n L ( R ) . D e t e r m i n e (C o T)f(z)
2
for z G C - R . A c c o r d i n g t o the
definitions of C a n d T we have

(it.
—Z
A p p l y i n g T h e o r e m 6.6.3 we get

It is k n o w n (see E x a m p l e 6.6.2) t h a t

1 T 1 ( T ) = f ff(i)e^ if > 0,
2TT¿ {--Z) [ ]
\-H(-t)e itz
if < 0.
 6. TEMPERED DISTRIBUTIONS AND FOURIER TRANSFORMS
118

F r o m this we o b t a i n

( C o F ) f ( ) - { ^ ^ z
f e l t Z d t i
^ >
f > 0

( )
l - J -oo
/
- J«W e ^ i
( )
i f ^ < 0 -

F i n a l l y we h a v e ( 6 . 1 4 . 1 ) for / G L (R).
-2/ 2

W e are able t o show t h a t t h e t h e o r e m is true i n general. It remains t o prove

t h a t ( 6 . 1 4 . 1 ) holds for f(x) = (ix) f (x), where f G L ( R ) a n d k = 0 , 1 , 2 , . . . I n
k
0 0
2

fact, we have

(Cof)((i.) f (.))(z)
k
0 = C((Ff ) )(z) 0
w
= ((CoT)fo)^(z) = ^Ffo(z)

\-fl J (it) fo{t)dt

00
iz k
if9*<0.

T a k i n g i n t o account D e f i n i t i o n 6 . 1 1 . 1 we get ( 6 . 1 4 . 1 ) for / G S' . I n the same w a y 0

one c a n prove t h e following e q u a l i t y

T- f{z)
1
= (CoT- )f{z) 1
(6.14.2)

for feS' . 0 •

A s a c o r o l l a r y from T h e o r e m 5 . 7 . 1 , T h e o r e m 5 . 7 . 2 a n d T h e o r e m 6 . 1 1 . 1 we
obtain

THEOREM 6 . 1 4 . 2 . If f £ S , then T f is a holomorphic

f
0 function i n C - R and

lim j (Tf(x + ie) - ff[x - ie))p(x) dx = Tf{p) (6.14.3)

for p G V 2.
L

F o r m u l a ( 6 . 1 4 . 3 ) m a y be used t o determine t h e F o u r i e r transforms o f s l o w l y

increasing functions.

EXAMPLE 6 . 1 4 . 1 . O f course, the H e a v i s i d e f u n c t i o n H is slowly increasing. I t

is easy t o see t h a t

= H ^ i
'f > 0

w
10 i f S z < 0.
Therefore
If 1 l
TH(p) — — lim - / p(x) dx = i- rip)-
K J
e-^o % J x + ie^ (- + ¿ 0 ) ^ K J v

REMARK 6 . 1 4 . 1 . If / G I? then

li (Tf(.
m + ie)-Tf(.-ie))=Tf
e—>0
in L.
2
 6.15. THE PALEYA-WIENER TYPE THEOREMS 119
6.15. T h e P a l e y a - W i e n e r type theorems

W e s h a l l n o w r e c a l l the classical P a l e y a - W i e n e r t h e o r e m ([23]).

THEOREM 6.15.1. Let f G L ( 0 , oo) and 2

CO

F(z)= I e f(t)dt,
izt
(6.15.1)
o
then F is a holomorphic function in I l +
= {z : *^sz > 0} and

sup / \F(x + iy)\ dx 2

= C < oo. (6.15.2)
0<y<oo J
E

Conversely, if F is a holomorphic function in I I and (6.15.2) holds then there +

exists f G £ ( 0 , oo) such that (6.15.1) is true for z G I T . Similarly,

2
if f G +

L (—oo, 0) and
2

o
F(z) = ~ J e f(t)dt,
izt
(6.15.3)
— CO

then F is a holomorphic function in I T = {z : ^sz < 0} a n d

-

sup J \F(x + % ) | da: = C < oo. 2

(6.15.4)

— co<y<0
E
Conversely, if F is a holomorphic function in IT" and (6.15.4) holds then there
exists f G L (—oo,0) 2
suc/i t/iai (6.15.3) is true for z G II"".
D e n o t e b y i J ( I I ~ U l I ) the vector space of a l l h o l o m o r p h i c functions i n Ü-UII" ".
+ 1

A s a c o r o l l a r y we o b t a i n the following

THEOREM 6.15.2. F e H{H~ U I I + ) and

sup [ \F(x + iy)\ 2

dx = C < oo (6.15.5)
E

iff there exists a function f in L (R) such that

2

F{z) = Tf{z)

for z G I I -
U I I / where Tf
+
is the Carleman transform of f.

W e are able to prove a s i m i l a r t h e o r e m for the C a u c h y t r a n s f o r m .

THEOREM 6.15.3. F is in H(U~ UU ) +

and (6.15.5) holds, iff there exists a
function f G L ( R ) such that
2

F(z) = Cf(z)
for z G ( I I U I I ) , where Cf
- +
is the Cauchy transform of f.

PROOF. B y T h e o r e m 6.15.2 there exists a f u n c t i o n g i n L ( R ) such t h a t F(z) — 2

Tg{z). T a k i n g into account (6.14.1) we get Tg(z) = ( C o T)g(z). Put / = Tg,

t h e n we have F(z) = Cf(z). T h i s finishes the proof. •

W e are now i n a p o s i t i o n to give a generalization of t h i s t h e o r e m .

 6. TEMPERED DISTRIBUTIONS AND FOURIER TRANSFORMS
120

THEOREM 6.15.4. A function F G H(Il~ U I1+) is the Cauchy transform of

some distribution A from T>' L2 iff
m d„

where F v G H(J\T U Ï Ï + ) and fulfils (6.15.5).

PROOF OF NECESSITY. L e t A e V' , L2 t h e n we have

CA(z) = ^ - A ' 1

2TTZ \(-- z) J '

In v i e w of T h e o r e m 5.4.1 we o b t a i n the following equality

v = z
° R

where / „ G L ( R ) . T h i s means t h a t CA(z) = Y^=o CD f (z),

2
where D " / * denotes v
u

/ / - d i s t r i b u t i o n a l derivative of I n accordance w i t h T h e o r e m 5.7.1 we have

m , u

P u t F (z)
u := Cf (z).
v B y v i r t u e of T h e o r e m 6.15.3, F v fulfils (6.15.5). •

PROOF OF SUFFICIENCY. Suppose t h a t F(z) = Y^=Q-£^ Á )^ where the F Z

f u n c t i o n F fulfils (6.15.5), v — l , . . . , m . I n v i e w of T h e o r e m 6.15.3 we infer

v

t h a t there exists a f u n c t i o n f G L (R) such t h a t F — Cf .

v
2
Therefore v v

m j
U m
F
& = E d ^ c
^ z
) = E C(D»U)(Z).

T h i s means t h a t F = C A , where A = E H o 1 7
^ ? ^ L ( R ) . T h i s implies that
2

A is i n V' . L2 •

W e are able to f o r m a s i m i l a r t h e o r e m for the C a r l e m a n t r a n s f o r m a t i o n .

THEOREM 6.15.5. A function F in H(IL~ U l l ) is the Carleman transform of +

a slowly increasing function f iff F may be written in the form (6.15.6), where the
function F fulfils (6.15.5).
v

PROOF. T h e t h e o r e m is a n i m m e d i a t e consequence of the previous t h e o r e m

a n d (6.14.1). •

EXAMPLE 6.15.1. L e t X [ - i , i ] be a characteristic f u n c t i o n of the i n t e r v a l [—1,1].

A n easy c o m p u t a t i o n shows t h a t

1 \ , x sinx

(see E x a m p l e 6.3.1) a n d

ÍÍ 1
V N i f 9f« > 0,
 6.16. THE CAUCHY SEMIGROUP 121
Hence we have
smx , i f 3 z > 0,
x • dx = < . â~
2%

liri J
2iri x -- z e 1
if < 0.
2«
0

R v

EXAMPLE 6.15.2. It is easy to verify t h a t the C a r l e m a n t r a n s f o r m of the H e a v -

iside f u n c t i o n is
if 3 z > 0,
[0 i f %z < 0.
In accordance w i t h T h e o r e m 6.14.2
i
(• + í o ) '
R
if (p G T>i2. Therefore

(• + ¿0) ;
[0 i f 3 z < 0.

6.16. T h e C a u c h y s e m i g r o u p

Let gt(x) :— ~ 2+ 2 ? x G IR. a n d t > 0. A s s u m e t h a t <p G <S(R) a n d p u t u ((p) :=

t x t

g * ip. B y Y o u n g ' s i n e q u a l i t y we have

t

\\u ((p)\\ 2
t L < \\gt\\ i I M U
L
2
= IMU -2

This implies that u t is a c o n t r a c t i o n operator o n L (R) 2

for ¿ > 0. W e k n o w t h a t
Tg (o-) t = e~^ ^. U s i n g the classical B o r e l f o r m u l a one c a n prove t h a t
a

(see (6.4.7)). F r o m this we get

^ i + t M = (dt! *9t ) * ^ = Píi * (#¿2 * V?) =

2 2 u {u {ip))
tl t2

for G 5 ( R ) . T h i s means t h a t u (-) t is a semigroup of the c o n t r a c t i o n s u tl t > 0 on

L ( R ) . T h i s semigroup is called the C a u c h y semigroup. O u r purpose is to find the
2

i n f i n i t e s i m a l operator of this semigroup. L e t us notice t h a t

T{t-\u {v) t - <p)){a) = t-\e-W - l)^(cr).

It is easy to see t h a t I t " ^ ^ 1 - 1
- 1)| < |<r|. Therefore

| - i ( - ^ l _ l)T(p(a)\
t e + H \T<p{a)\ < 2\a\ |^(a)|.

O f course, (-)T(p is i n L (R) i f ip belongs t o <S(R). O n the other h a n d t ~ ( e - l ¡ — 1)

2 1 t c r

tends p o i n t w i s e t o — | a | as t —Y 0 . T h i s i m p l i e s t h a t +

Hr\M<p) - H>))-\(-)\r<p
as t —» 0 i n L (R). U s i n g the P l a n c h e r e l t h e o r e m we infer t h a t £ (i¿t(<p) — tp)
2 -1

converges to some ip i n L (R). O b v i o u s l y , Tijj{cr) — — |cr|^(^(a). I n v i e w of T h e o -

2

rem 6.10.2 we have

O f course, ~T^2 * = ^ i n L (R). 2

F r o m this it follows t h a t the operator

AÍ \ 1 1
 6. TEMPERED DISTRIBUTIONS AND FOURIER TRANSFORMS
122

is the i n f i n i t e s i m a l operator of the C a u c h y semigroup i f we take S (R) as the d o m a i n

of t¿¿, t > 0.
T h e p r i n c i p a l goal of our c o n s i d e r a t i o n is t o show t h a t the C a u c h y s e m i g r o u p
m a y be r e p r o d u c e d b y means of the e x p o n e n t i a l f u n c t i o n expA(-) o n the dense
o
subspace S of L ( R ) . W e shall n o w define such subspace of L ( R ) . I n order to d o
2 2

o
it we use the H e r m i t e functions h (see S e c t i o n 6.7). L e t S denote the vector space v

spanned by the set {h„ : v — 1 , 2 , . . . } . Since H e r m i t e functions h c o n s t i t u t e a v

o
complete o r t h o n o r m a l s y s t e m i n L ( R ) , therefore the closure of S i n L ( R ) is the 2 2

whole space L (R). W e k n o w t h a t

2

Th = k i V27rh . k
k
k

For s i m p l i c i t y p u t x ~ r W - L e t x" =
X ** ''* i n
accordance w i t h Theo¬
v times
r e m 6.10.2 a n d (6.12.4) we have

HX V
*h ) k = {-\y\a\ i ^h (a). v k
k

F r o m this a n d the P l a n c h e r e l t h e o r e m it follows t h a t x" * h k is i n V^i. We shall

now show t h a t the series

h k + ^X^h k + |yx *h 2
k + •••+ ^ y x * h n
k + •• • (6.16.1)

converges t o u (h ) i n the sense of L ( R ) for k = 0 , 1 , . . . F i r s t let us find the

t k
2

F o u r i e r transforms of the p a r t i a l sums

t t n

S k
t = h k + -x*h k + -.. + —x n
* h k

of the above series. B y v i r t u e of the B o r e l f o r m u l a we o b t a i n

TS k
nt = i V^h (a)
k
k ( l - ±\a\ + ••• + ( - l ) n
^ M n
) •

Note that

\FS {a) - i V^h {a)e'^\ < 2irhl(\<r[ E (-D"SK , a

k k 2
nt k

< C k e - ^ 2
e 2 t
^ ^ GL^R),

for £ > 0 a n d some e > 0. O b v i o u s l y , FS (a) converges to 2rci h (a)e~' ^ for k

t
k
k
t

a G R . Therefore, by the Lebesgue d o m i n a t e d convergence t h e o r e m we c o n c l u d e

that S k
converges to \/2ni /i^e *'^'
t as n —> oo i n the sense of L .
k -
Because o f 2

the P l a n c h e r e l t h e o r e m the series (6.16.1) converges to u (h ) i n L . P u t t i n g for t k

2

simplicity

e A t
h k := l i m V * h
^

we finally arrive at the following

o

THEOREM 6.16.1. If <p e S, then u (<p) = e <p in the sense of t

At
L.
2
 6.17. THE CAUCHY PROBLEM FOR THE HEAT EQUATION 123
REMARK If we take W ( R ) as the d o m a i n of the operator A, t h e n
6.16.1.
1 , 2

it w i l l be a closed operator. W e o n l y need to verify t h a t i f ip are i n W ' ( R ) , v

1 2

ip -> (f i n L a n d A(p -> ip i n L , t h e n (p belongs to W (R)

v
2
u a n d Aip = ip. B y
2 1,2

T h e o r e m 6 . 1 1 . 2 we have

s 2 *<PV = -\(')Wv
W
In accordance w i t h T h e o r e m 6 . 6 . 2 , (¡5^ —^ <^ i n L 2
and — - > ^ i n L . One can
2

choose a subsequence (<p^) of the sequence ( ^ ) so t h a t —> — | ( ' ) I ^ T —>• ^

a n d - | ( - ) I ^ T — » —|(-)l<£ as z/ -> oo almost everywhere o n R . Therefore —| (•)!<? = V 7

i n L ( R ) . T h i s i m p l i e s t h a t <p is i n W ' ( R ) a n d ,4<¿> = ^ .

2 1 2

6.17. T h e C a u c h y p r o b l e m for the heat e q u a t i o n

A f o r m a l s o l u t i o n of the C a u c h y p r o b l e m

= &xu(x,t), (6.17.1)

u(x,0) = / ( * ) , (6.17.2)

where x G R , ¿ > 0 a n d A
n
x = + • • • + ^ 3 - , c a n be c o n s t r u c t e d by q u a d r a -
tures. T h e s o l u t i o n of e q u a t i o n ( 6 . 1 7 . 1 ) , w h i c h satisfies ( 6 . 1 7 . 2 ) is u s u a l l y w r i t t e n
as follows

u(x, t) = ( 4 T T Í ) - * j exp ± - l L / ( t / ) dy. (6.17.3)

T h e n f(x) — lim¿_ + i¿(a?,£). T h e f u n c t i o n

>0

E { X t t ) i ( ^ ) - e x p ( ^ ) if*>0,
[0 ift<0
is a d i s t r i b u t i o n a l s o l u t i o n of the differential e q u a t i o n

A
* - § i ) E =
~> 6
(- -)6 17 4

where ó denotes the D i r a c d i s t r i b u t i o n (compare w i t h S e c t i o n 4 . 3 ) . Equality

( 6 . 1 7 . 3 ) c a n be w r i t t e n i n the c o n v o l u t i o n f o r m

u(x,t).= (E(.,t)*f)(x). (6.17.5)

W e s h a l l now show t h a t t h i s f o r m u l a holds w h e n we s u b s t i t u t e a n a r b i t r a r y t e m -

pered d i s t r i b u t i o n A i n the place of / . I n t h i s case i¿(-, •) is also a s m o o t h f u n c t i o n
i n Q = {(x, t) : x G R , t > 0 } . A question, h o w the c o m p l i a n c e of ( 6 . 1 7 . 2 ) c a n
n

be u n d e r s t o o d , w h e n a t e m p e r e d d i s t r i b u t i o n A is t a k e n i n place of the f u n c t i o n / ,
n a t u r a l l y arises. I n this sequel we e x p l a i n the above question.
W e start w i t h the following

LEMMA 6 . 1 7 . 1 . Let f and g be in <S(R ). Put h(x,a) n

':= f(x - a) g (a), then
h G <S(R x R ) .
n n

PROOF. Note that

i—h(x,*)= Y { )^f(x-a)^- gfaV aeW .1

0<i/<a
124 6. T E M P E R E D D I S T R I B U T I O N S A N D F O U R I E R T R A N S F O R M S

Hence we o b t a i n
£i\ai ro¿2\ J
_ , / \ fk\oL2-\-v\ ^lai — v>\
, h(x,*)= Y a i
) x ^ - f — - f ( - a ) a ^ ^ x -g(a).

B y means of the T a y l o r f o r m u l a we get x@ — w(x — <r, cr), where w(-, •) is a p o l y - 2

n o m i a l . Therefore

dx^d^ X f { x
~ a )
= W { X
~ ' a a )
dx^d^ f { X
~ a )
-
It is easily seen t h a t

dx daa2 u v y K J
dy°
where y — x — a. T h e expression

w(y, a)a^ ( - l ) M — f{y)^ g(a)

is b o u n d e d i n R n
x R n
b y the a s s u m p t i o n . T h i s i m p l i e s t h a t the f u n c t i o n

1 rr&2 . -/i(:r, cr)

is b o u n d e d i n R n
x R . T h u s , the p r o o f is
n
finished. •

W e s h a l l n o w show t h a t the f u n c t i o n

u{x, a) := A {E{x a - cr, *)) (6.17.6)

satisfies e q u a t i o n (6.17.1) i n ft. Indeed, we have to show t h a t

// A (E(x
a - a,t)) (A X + I ) tp(x,t)dtdx = 0 (6.17.7)

for G I n v i e w of L e m m a 6.17.1 we infer t h a t

E(x-a,t) (^A +x <p(x,t) G <S(R x l n n

x R).

M o r e o v e r , the characteristic f u n c t i o n V of the set R x R belongs t o ¿>'(R x + n

R ) . E q u a l i t y (6.17.7) c a n b e w r i t t e n i n the f o r m of the tensor p r o d u c t of t h e

n

d i s t r i b u t i o n s F a n d A as follows

(T ( M ) ® A , ) (E(X - a, t) ( A x + ^ ^(x, t)) - 0 (6.17.8)

(see E x a m p l e 6.2.1). B y c o m m u t a t i v i t y of the tensor p r o d u c t of t e m p e r e d d i s t r i -
b u t i o n s we have

(L ( M ) <g> A „ ) (E{X - <J, t) (^A + ' y ( x , t)^j x

= K ¡ J J E{x-a,t) (^A + ^j x y{x,t) dt dx

\R+ R n
)

The integral

II E ( x - cr, t) ^ A + x ) <p(x, £) dt dx
 6.17. THE CAUCHY PROBLEM FOR THE HEAT EQUATION 125
c a n be w r i t t e n i n t h e following f o r m

j j ^A -^)E(x-a,t)ip(x,t)dtdx.
dt
x

R+ R n

B y (6.17.4) we have

j J (^A - x E(x-a,t)tp(x,t) dtdx = 0

R+ R n

for a G M . A c c o r d i n g t o T h e o r e m 4.6.4, i¿(-, •) is s m o o t h i n fl.

n

W e have proved t h e fact w h i c h c a n be stated as a separate

THEOREM 6.17.1. A function u(x, i) = A (E(x-a,t)), a A G t S ^ E ) is a solution

72

of (6.17.1) in ft.

T h e sense of satisfying (6.17.2) b y this s o l u t i o n w i l l be e x p l a i n e d b y t h e follow-

ing

THEOREM 6.17.2. u(-,t) converges to A in S'(R ) n

as t -> 0 . +

PROOF. Since A is i n <S'(lR ), therefore n

\u(x,t)\ < ¡AriEix - <T,t))\

<M y , s u p
<K .
da
At a v J At

M<mi |^|<m 2
iTGlB

W e shall now show t h a t

J u(x t)ip(x)dx 1 -> A(<p) (6.17.9)
R n

as t —> 0 for (p G <S(M ). I n accordance w i t h L e m m a 6.17.1, E(x — <r, t)<p(x)

+ n

is i n S(W x W ) for t > 0. C o m m u t a t i v i t y of the tensor p r o d u c t of t e m p e r e d

l 1

d i s t r i b u t i o n s i m p l i e s the following e q u a l i t y

J A (E(x
a - cr, t))cp(x) dx = ACT I J(E(x - cr, i)ip(x) dx) j

for <p G <S(R ). W e see at once t h a t

n

j E(x - cr, t)ip(x) dx = J E(a - x,t)ip(x) dx = {E(-,t) * (f)(cr).

R 71
R n

W e have t o show t h a t the expression £?(•, t) * tp converges t o (p as t -> 0 +

i n <S(R ) n

for each <p G <S(R ). E q u i v a l e n t l y ,

n

^ l ^ ^ ' ' í }
*^ x / 3
f ^ ( a ; ) ( 6
- 1 7
- 1 0 )

u n i f o r m l y as t —> 0 +
in R n
for each a,/3 G N . It is easy t o check t h a t 7 1

(•) 9"(£;(-,í)^) = (-)

/ 3 / 3
(^(-,í)*9>í

= £ ; ( • , * ) O<l/<0
E ( )((rE(;t))*((-)^d <p).
p
l
a
 6. TEMPERED DISTRIBUTIONS AND FOURIER TRANSFORMS
126

Hence, it follows t h a t
_ . f)\0\
{•)Pd"(E(-,t) * <p){a) ={~i) m
exp(-i|a| )^a^(a) 2

0O</3 v 7

O n e c a n observe t h a t
^ e x p ( - t H 2
) c ^ < 9 ^ ^ 0
as t -> 0+ i n L ( E ) , w h e n v > 0. O f course, T h e o r e m 6.3.1 is also true i f we p u t
1 n

T~x
i n the place of T. N o w , it is clear t h a t (6.17.10) holds. •

A s s u m i n g t h a t A belongs t o a s m a l l subspace of S' ( R ) we c a n o b t a i n a m o r e n

delicate i n f o r m a t i o n a b o u t the s o l u t i o n u near the b o u n d a r y of fl.

THEOREM 6.17.3. If A = f e £ ( R ) 1 < p < oo andu(t,x) p n

; = A(E(x--,t)),
then u(-, t) tends to A as t —» 0 in IP. +

PROOF. N o t e t h a t E(x,t) = t~^E 1^ a n d ¡ R n E(x, 1) dx = 1. Since

A(E(x - -, yft)) = (t~*E lj * (x), (6.17.11)

therefore it is a consequence of L e m m a 1.2.1. •

Since i¿(-, •) is s m o o t h i n Q , therefore u(-,t) does not converge t o / as t —» 0-f-

i n L ° ° ( R ) , i f we p u t / = H i n (6.17.11), where i f denotes the H e a v i s i d e f u n c t i o n .
n

It means t h a t the above t h e o r e m is not true, w h e n p = oo.

THEOREM 6.17.4. If A e V' , LP 1 < p < oo and u(x,t) = A(E(x - •,£)), then
u{-, t) converges to A in T>' as t Lv 0 .
+

PROOF. T h i s t h e o r e m is a n i m m e d i a t e consequence of T h e o r e m 5.3.2 a n d

L e m m a 5.7.1. •
 CHAPTER 7

O R T H O G O N A L E X P A N S I O N S O F D I S T R I B U T I O N S

7.1. T h e Poisson summation formula

L e t T — J2uei ^' F i r s t , we s h a l l show t h a t T is a t e m p e r e d d i s t r i b u t i o n . N o t e

Tv

t h a t T((p) = J2„ez O f course, the series ] T ^ ip(y) is convergent for each e Z

ip e S. Since the space S is complete (see S e c t i o n 6.1 a n d A p p e n d i x ) , therefore

f

T is i n S'. D e n o t e b y T the F o u r i e r t r a n s f o r m of T . W e are able t o prove the

following

THEOREM 7.1.1. ^ ^

PROOF. B y the c o n t i n u i t y of the F o u r i e r t r a n s f o r m a t i o n i n S' a n d (6.9.3) we

have

Let T | ( _ , 2 / r ) and T | (
7 r + 2 l / 7 r 7 r + , i ) be the restrictions of T to the o p e n sets
7 0 + 2 7 R I ) 2 7 R + 2 L / 7 R

(—7T + 2z/7T,7T + 2z/7r) a n d (0 - f 2TTU 2n + 2z/7r), respectively. W e s h a l l n o w show

:

that T | _( ) = 27rr ¿ a n d T\( + M,2n+2v*)

7 R + 2 L / 7 R Í 7 R + 2 L / 7 R = 0. W e s h a l l o n l y prove
27ri/ 0 2

the case w h e n v — 0. T h e p r o o f i n the rest cases runs i n the same m a n n e r . P u t

S (x) = J2\n\<v •
l/ It is easy t o check t h a t
e%[XX

sin(z/ - f ~)x
Sy(x) =

and S (x)dx
u = 2TT. Suppose t h a t ip G D(—7r, 7r). W e have to show t h a t

lim — / 5,y(x)(^(x) — y>(0)) = 0.

2TT J
— TT

In order to do it we o n l y need to show t h a t

TT

lim / S (x)(p(x)
u dx = 0
1/-+00 J
— TT

if 99(0) = 0. I n this case, the f u n c t i o n (p c a n be w r i t t e n as follows (p(x) = xip(x),

where ip is i n V(—7r,7r). F u r t h e r , note t h a t
TT TT 7T

sin(z/-f ^ ) x /* x cos f . /*
—(p(x) dx = / -r-ip(x) sinz/x -h / <p(x) cos z/x d x .
/ smf J smf i
128 7. ORTHOGONAL EXPANSIONS OF DISTRIBUTIONS
Since the f u n c t i o n / , f(x) = I/J(X) is continuous, therefore

lim / S {x)p(x)
v dx — 0.
v—ïoo J
— TT
w h e n p G V(—7r,7r) a n d p>(0) = 0 . I n the same m a n n e r we c a n see t h a t
2TT
sinfz/ + ^)x , s

lim / \ p(x)dx
x
2 >
i = 0
I/-+OO J

0
if p G £>(0, 2TT). F r o m this, b y T h e o r e m 2 . 3 . 2 , we o b t a i n ( 7 . 1 . 1 ) . T h u s , the p r o o f is
finished. •
( 7 . 1 . 1 ) is k n o w n as the P o i s s o n s u m m a t i o n f o r m u l a .

7.2. Periodic distributions

T h e following l e m m a w i l l be needed i n our considerations.

LEMMA 7 . 2 . 1 . There exists a function

3
ip in V( —" |4 ,' |)
4
3
such that 0 < ip(x) < 1,
'ip(x) = 1 for x G [—|, \] and

Y^(v -x) = 1

/ o r eac/i x G l .

PROOF. I n v i e w of L e m m a 2 . 1 . 1 there exists a f u n c t i o n 6 G V( — | , | ) such t h a t

0 < 6{x) < 1 a n d 6(x) = 1 for x G f ]. P u t

CJ(:C) := ])T 0 ( z / - a?) > 1

vez
and

u;(x)
It is easily seen t h a t X ^ e z ^(v — x) = 1 for each x G R . T h i s finishes the proof. •

DEFINITION 7 . 2 . 1 . W e say t h a t a d i s t r i b u t i o n T is a - p e r i o d i c , a > 0 i f r T a = T.

H a v i n g disposed of this p r e l i m i n a r y step, we c a n prove the f o l l o w i n g

THEOREM 7 . 2 . 1 . Every 1-periodic distribution can be represented as follows

in S', where c — T {i\)e~~ ^),

v tp is as in Lemma 7.2.1.
%2r:v
Moreover, there exist
M > 0 and m G N such that \c \ < M\v\ for v e Z - { 0 } .
u
m

PROOF. E q u a l i t y ( 7 . 1 . 1 ) c a n be w r i t t e n as follows

£ Ê ( I / ) = 2 * ^ ( 2 ™ ) (7.2.2)
vez vez
for p e S. R e p l a c i n g p by pe ^ lx
i n ( 7 . 2 . 2 ) , b y ( 6 . 9 . 4 ) we get

${x r^) J
= ^ Y l Vi^vy ™*. 2
(7.2.3)
vez vez
 7.2. P E R I O D I C DISTRIBUTIONS 129

L e t ip be as i n L e m m a 7.2.1, t h e n

W e s h a l l n o w show t h a t the series

27r^^(27rz^e* ™ - 2 ( )
(7.2.4)
i^GZ

is convergent i n V. Indeed, note t h a t

k

i=o
W e o n l y need t o show t h a t the series

Y, ^Txv){i2nvf-H^é ^\ 2
j = 0,..., k (7.2.5)

is u n i f o r m l y convergent i n R . T h i s fact is evident because </? is i n S. F r o m t h i s we

infer t h a t

T
( ^ + n
u = 2 7 T
Y, ^ V
) T
(V ^ ) 2 }
. (7.2.6)
\vei J ' vez
Put
c = T (yj ™^
u
2
= TiP{e ^).i27r

Since Tip has a c o m p a c t s u p p o r t , therefore b y T h e o r e m 6.10.1 there exist t w o

constants M > 0 a n d ra G N such t h a t

|c„| < M | i / | m
(7.2.7)

for v G Z — { 0 } . L e t ( p , ^ E S , / i = 1, 2, Assume that ^ -> <p i n <S. T h e n we

have
(1 + ( 2 7 r i / ) ) 2 f c + 1
| ^ ( 2 i r i / ) | < üf*
for v, fj, = 1, 2, Further,

i ^ " ( 2
^ i *
M
( i ¿lV)
+
fc+iJ
" fc
-
M X f c
îT(bp-
T h i s i m p l i e s t h a t the series X ^ e z ^ P^( ^ ) ( < 7riy 1 S
u n i f o r m l y convergent w i t h respect
to \i G N . Therefore

HH^X^ (V ™ ' ) 2 ( }
= J 2 T
(W ™ ) ^ ( 2 T T I / ) .
2 ()
(7.2.8)

N o w , suppose t h a t (p^ is i n D a n d t h a t (p^ converges to (p i n S. O f course, (p^ also

converges t o i n S. N o t e t h a t i f ip is i n P , t h e n

Vi^Z / i^GZ

Since T is a 1-periodic d i s t r i b u t i o n a n d ip £ V, therefore

V E Z / i/ez V^ez /
 7. ORTHOGONAL EXPANSIONS OP DISTRIBUTIONS
130

F r o m t h i s , b y (7.2.6)

T
(ív) = 2
^ E T
(^ 27ri/(,)
) <M 2 7 r z /
)-

T a k i n g i n t o account (7.2.8) we get

T(<p) = 27rJ2^ P( )- ( 27rP

vez
T h i s e q u a l i t y c a n be w r i t t e n as follows

T = 27r]Tc T2 ,¿. I/ 7RI (7.2.9)

vez
Since T\ = 2 K8, therefore b y (6.9.4) we have
/

vez
R e p l a c i n g —v b y v i n this e q u a l i t y we o b t a i n

T =Y c e v
i 2
^\ c = T(i;e- ^).
v
i27T
(7.2.10)
vez
F r o m (7.2.9) we infer t h a t c v does n o t d e p e n d o n ip. T h i s statement finishes t h e
proof. •

REMARK 7.2.1. I f T = / , where / is a 1-periodic continuous f u n c t i o n , t h e n

v —f \ f(x)e~~ 27TUX
dx. Indeed, we have

c„= J f(x)iP{x)e- i27TUX

dx =Y J f{x)^{x)e- dx i2lxux

2
—l Í2TTVX
]T Í f(uMcü + v)e-* ™"düü= 2
f f(x )e~ ^ vx
dx.

F r o m T h e o r e m 7.2.1 i t follows t h e following

COROLLARY 7.2.1. Every periodic distribution is tempered.

W e s h a l l n o w present t h e inverse t h e o r e m t o t h e previous t h e o r e m .

THEOREM 7.2.2. If there exist constants M > 0 and m G N such that \c \ < u

M ¡vi™ for each v G Z — { 0 } , then the series

J2 » ™ '
c ei2 { ]
(7.2.11)
vez
is convergent in S'. Its sum is 1-periodic distribution.

PROOF. L e t
Sv=Y c ^ \
Note that

J Svv = c
^( 2 7 r
M)
 7.3. THE SPACES A AND A' 131
for p G S. T a k i n g i n t o account t h a t p is also i n S b y the a s s u m p t i o n we conclude
t h a t the series X ^ G Z c p{2irv) is convergent. T h i s means t h a t the series (7.2.11)
v

is convergent i n S'. It is easily seen t h a t the m a p p i n g S' 3 T — » r^T G S is f

continuous for each ft E R (see E x a m p l e 1.8.4). A n i m m e d i a t e consequence of this

r e m a r k are t h e following equalities

i/GZ z^GZ i/GZ

Therefore, t h e d i s t r i b u t i o n T , T = X ^ ç Z C i / ^ ™ ^ is a 1-periodic d i s t r i b u t i o n . T h i s
2

statement finishes the proof. •

7.3. T h e spaces A a n d A!

W e start w i t h the presentation o f some i n f o r m a t i o n a b o u t t h e W i e n e r o r t h o g o n a l

system. T h e W i e n e r functions

c o n s t i t u t e a complete o r t h o n o r m a l s y s t e m i n L (R).
2
T h e function pv is a n eigen-
f u n c t i o n of t h e differential operator A/",

Mu{t) == *(* ~ ((* +

c o r r e s p o n d i n g t o the eigenvalue — z/, v G Z . W e s h a l l n o w define a vector space A

of s m o o t h functions o n t h e r e a l line, w h i c h is v e r y i m p o r t a n t i n a p p l i c a t i o n s .

DEFINITION 7.3.1. W e say t h a t a s m o o t h f u n c t i o n p is i n A i f H p> k

G L (R)
2

a n d (M p,p„) k
= {pM py)
k
for each k G N a n d v G Z (see [27]).

Set b y d e f i n i t i o n

IMIfc := | | A / * y > | | 2 . L

O n e c a n show t h a t the set .A is a complete vector space w i t h t h e s y s t e m of t h e

s e m i n o r m s (|| • ||jfc, k G N ) ([27]). Because o f i d e n t i t y

( ( ^ + i ) v ^ > = è ( * ) i M i ï
v=0 ^ '

the s y s t e m (|| • ||fc, fc G N ) o f s e m i n o r m s || • ||¡t is equivalent t o t h e following s y s t e m

(¡II • | | | fc) k £ N ) of n o r m s , | | | p | | | * = ({Af + l) tp,tp).
2 k
O f course,

IIMIIo<IIMIIi<-"<IIMII f c <"-, keK (7.3.2)

W e s h a l l n o w show t h a t t h e n o r m s | | | • \ \ \ are c o m p a t i b l e i n the following sense

k

([9]):
If a sequence (<£v), p G A, converges t o zero w i t h respect t o the n o r m | | | • | | |
v m

a n d is a C a u c h y sequence i n t h e n o r m | | | • | | | , m < n , t h e n i t converges t o zero i n

n

the n o r m | | | • | | | . n
 7. ORTHOGONAL EXPANSIONS OF DISTRIBUTIONS
132

W i t h o u t loss of generality, one c a n assume t h a t m = 0. T h e n we have

\\M\l=((Af 2
+ l) ^,^)
n

=<(Af + l) {<p -2 n
v Vll ),ip ) v + ((Af + 2
*>„>

<((AT + 1 ) " ( ^ -
2
^ - p > * ( ( A T + 1 ) " ^ , <p )i
M
2
v

+ ||(Ar + 2
l ) > M | | L 2 | | ^ | | L 2

HII^ - ^ I I U H ^ I I L + \W 2
+ lr^lMlvvlU'-

For every e > 0 we c a n choose M > 0 a n d a p o s i t i v e integer z^ so t h a t 0

H l ^ l l l ' < e M + e||(A^ + l ) > | | 2

M L 2

for /¿, v > Z/Q. T h i s i m p l i e s t h a t the sequence ((¿7^) converges t o zero i n t h e n o r m

HI • | | | . Therefore t h e n o r m s | | | •
n G N are c o m p a t i b l e .
L e t Ak be t h e c o m p l e t i o n of A w i t h respect t o t h e n o r m 111 • 11 \ . E v e r y element k

/ i n Ak is defined b y t h e C a u c h y sequence (<p ), ip G «4, v = 1, 2 . . . , w i t h respect u v

to t h e n o r m | | | • | | | . A c c o r d i n g t o (7.3.2) this sequence is also a C a u c h y sequence

fc

w i t h respect t o t h e n o r m || • | | 2 = | | | • | | | a n d therefore defines a f u n c t i o n / i n

L 0

L ( E ) . I t is easy t o verify t h a t / is u n i q u e l y d e t e r m i n e d b y / . W e s h a l l n o w show

2

t h a t d i s t i n c t elements fi a n d / 2 from Ak cannot be m a p p e d i n t o the same element

/ G L ( M ) . It is sufficient to verify t h a t a nonzero element ip i n Ak cannot b e c a r r i e d
2

into the zero element of L (R). 2

Indeed, a s s u m i n g the contrary, for a sequence (</?*,),
iç v G A, w h i c h is a C a u c h y sequence w i t h respect t o t h e n o r m | | | • a n d defines
the element / ^ 0, we w o u l d have t h e following relations

l i m lllv^HU > 0, lim ||V^||L = 0. 2

v—>oo v—^oo

It w o u l d c o n t r a d i c t the c o m p a t i b i l i t y of the n o r m s | | | • a n d || • | | 2 . L

T h i s c o n s i d e r a t i o n indicates us t h a t t h e c o m p l e t i o n Ak of A m a y b e r e g a r d e d
as a subspace of L (R). A f u n c t i o n ip e L (R) is i n Ak iff there exists a sequence
2 2

(<¿>„), Lp G A such t h a t <p tends t o ip i n L as v —> oo a n d is a C a u c h y sequence

v u
2

w i t h respect t o the n o r m | | | • O f course, we have

L (R)=A
2
0 DÁ! D . . . (7.3.3)

B y continuity, one c a n e x t e n d t h e n o r m s | | | • \ \ \ from A onto Ak i n t h e following k

way: a s s u m i n g t h a t (p is i n Ak, t h e sequence ( y v ) , (p„ £ A converges t o (p i n L 2

and ((p„) is a C a u c h y sequence w i t h respect t o the n o r m | | | • \ \ \ , we take k

||M|| f c = l i m IH^III,.
v—>oo
A n a l o g o u s l y , the b i l i n e a r form (•, -)k : = ((A/" + l ) •, •) c a n be e x t e n d e d b y c o n t i n u i t y
2

from Ax A onto Ak x Ak b y t a k i n g

(ip,iP) — k l i m \U\\Wv ripA\\l

J
- 111^ - ^ l l l f c )

- i{\\\ Vv + ij, \\\l v - \\\<p - ixl>u\\\lj),

v (7.3.4)
 7.3. THE SPACES A AND A' 133
where the sequences (p ) u a n d (ip ) determine tp a n d ip i n A -
u k It is easy t o check
that

VO* = V> € A , (7.3.5)

(<P> < (<p, p)k+i> V £ A +i, k (7.3.6)

(p,iP) k = {(AT + 1 ) W ) ,
2
^ A ^ 4 (7.3.7)
T h e vector space A under the bilinear f o r m (•, - ) is a H i l b e r t space. E v e r y
k k

continuous linear form A o n A c a n be extended b y c o n t i n u i t y o n some space A . k

W e s h a l l n o w construct a n o r t h o n o r m a l s y s t e m i n A , A: = 1 , 2 , . . . . I n order k

to do this we p u t
Pvk = {v 2
+ l)' 1
Pu-

THEOREM 7 . 3 . 1 . The functions p,

uk v G Z constitute a complete orthonormal
system in A.
k

PROOF. F r o m ( 7 . 3 . 7 ) , it follows t h a t the functions p , v G Z c o n s t i t u t e a n vk

o r t h o n o r m a l s y s t e m i n A w i t h respect to the bilinear f o r m (•••)&• It r e m a i n s t o

k

prove t h a t the s y s t e m (p ) is complete i n A . Indeed, let p be i n A , t h e n

uk k k

vez
is i n L (R).
2
T h e series X ^ e z i v ^ Pvk)kpvk m a y be regarded as the F o u r i e r series of
p under the o r t h o n o r m a l s y s t e m (p ). O f course, uk

y\(PiPvk)k\ 2
< HMUfc-
vez
Since is a H i l b e r t space w i t h respect to the n o r m ||| • \ \ \ , therefore there exists k

ip i n Ak such t h a t ip = J^vezi^i Pvk)kPvk i n ^ U - Because of ( 7 . 3 . 7 ) we have

{p,Pvk)k = (v 2
+ l)*{p,Pv)- (7.3.8)
Hence we get ip = Y^vezi^^ P")P» Since the n o r m s ||| •
m
a n d || • \\ 2 are L

c o m p a t i b l e , therefore p = ip i n .4^. T h u s , the p r o o f of the t h e o r e m is finished. •

Because of Y,vez l ( ^ ¿ W / e | 2
< I I M I l L b y ( 7 . 3 . 8 ) we have

] T ( z / + l) \c \
2 k
v
2
< oo, c „ = (y>, ). Pl/ (7.3.9)
vez
Conversely, assume t h a t ( 7 . 3 . 9 ) is fulfilled, t h e n by the completeness of Ak we infer
t h a t the series Ylvez^ 2
+ c
vPvk converges to some p i n A k) therefore p is i n
A.
k

W e have p r o v e d the fact w h i c h c a n be stated as a separate

THEOREM 7 . 3 . 2 . A function p belonging to L (R) 2

is in Ak iff (7.3.9) holds.

COROLLARY 7 . 3 . 1 . A function <p belonging to L (R) 2

is in A iff (7.3.9) holds
for each k G N .
W e shall n o w describe the F o u r i e r e x p a n s i o n of the linear continuous forms
b e i n g i n the d u a l space A- of A . L e t A! denote the vector space of a l l continuous
k k

linear f o r m over A. E v e r y element A of A! c a n be extended o n some space A . I n k

this case we say t h a t A is i n A - . k

 7. ORTHOGONAL EXPANSIONS OF DISTRIBUTIONS
134

THEOREM 7.3.3. An element A from A f

is in A-k iff

E ? S ï < M
. Cv = A(p ). v (7.3.10)
vez ^ )

The n u m b e r A ( p ^ ) is called the F o u r i e r coefficient of A .

PROOF. L e t A be i n A-k, t h e n b y R i e s z t h e o r e m there exists a f u n c t i o n ip i n

Ak s u c h t h a t
A(jp) = (ip^) k (7.3.11)
for each (p e Ak- T h e f u n c t i o n ip has the following F o u r i e r representation

^ =
X c
»kPvk> c uk = (^, pi,fc)fc (7.3.12)

and
2 =
_ ¡ll^lll/c
||L/.|||2 = \ » \ '
C k 2
(7.3.13)
vEZ
In v i e w of (7.3.8) we o b t a i n

A ( ^ ) = Çy^vkpvki^k = y^c^/c(^ 2
+ I ) » (p„,y>).
vez vez
P u t t i n g (p = Pu in the previous e q u a l i t y we get

A(p ) = (/i + l ) * c
Ai
2
M f c . (7.3.14)
F i n a l l y we have
A
(^) = *52MPV)(PV><P)> V € A - (7.3.15)

F r o m (7.3.13) a n d (7.3.14) i t follows the following e q u a l i t y

vez v ;

Therefore (7.3.10) holds. It remains to prove t h a t (7.3.10) i m p l i e s the existence o f

a f u n c t i o n ip i n Ak such t h a t (7.3.11) holds. It is easy t o check t h a t the f u n c t i o n
ip — Ylvei (¡y2+i)fc Pv n a s
the r e q u i r e d properties. •

7.4. C a u c h y t r a n s f o r m s o f e l e m e n t s o f A!

We b e g i n b y d e t e r m i n i n g the F o u r i e r coefficients of the f u n c t i o n ¿ T r z ^ y , ^sz ^

0 w i t h respect t o the W i e n e r f u n c t i o n p , v v G Z. F i r s t let's assume t h a t z/ =
0,1, 2, W e s h a l l separately consider two cases: $sz < 0 a n d 9 z > 0. N o t e t h a t

P v W
v^(ii + |)" + 1

is h o l o m o r p h i c i f S t < | . Suppose t h a t Ssz < 0. T h e n , i t is easy to check t h a t

" V27f¿(--z); 2TT¿Í (¿Í + I ) - + i í - z ^ ( ^ + I ) - + i

R
and
 7.4. C A U C H Y T R A N S F O R M S OF ELEMENTS O F A' 135

O n the other h a n d for ^sz > 0 the f u n c t i o n ^^iy, +1 (t-z) 1 S

h°l° m o r
phi c
i f S t < 0.
This implies that

^ ( ¿ r h ) ) = ° -

In the same m a n n e r one c a n show t h a t i f v — — 1 , - 2 , . . . , t h e n

• i i \ i (iz - \y
if $z > 0
27Ti{--z)J (iz+ \) v + 1

and
1 1
= 0 if < 0.
v
\2m(--z)
It is easily seen t h a t

> 1 i f %z > 0. (7.4.2)

iz +
T h u s , we o b t a i n the following F o u r i e r e x p a n s i o n of ^ (--z) w
^ n
respect to the
W i e n e r functions

(7.4.3)

1 1 1 0 0
(¿z -
(7.4.4)

F r o m (7.4.2) a n d (7.4.1) it follows t h a t series (7.4.3) a n d (7.4.4) converge i n A i f

^sz > 0 a n d S z < 0, respectively. L e t A be i n A-k, t h e n we get the following
equalities
-1
1 1
A (7.4.5)
2TT¿ (• - z) J ~ ^/2TT
i / = — oo
oo
1 1
A = —T=YHPV)T: #T7 i f ^ < 0 . (7.4.6)
27TÍ (• - Z)

T h i s equalities are u n d e r s t o o d i n the sense of A-k- A c c o r d i n g t o (7.4.1) a n d (7.4.2)

we have
tz —
i r 3 z -> v:= \ e Bi(0)
IZ+Ï

and
iz 4- -
n +
9 z -> iu := - — \ G S i ( 0 )

Since A is i n A-k, therefore (7.3.10) holds. T h i s implies t h a t the series

1 0 0

— ^ A ^ K (7.4.7)
v
i/=0
and
1 00
 7. ORTHOGONAL EXPANSIONS OF DISTRIBUTIONS
136

are almost u n i f o r m l y convergent i n B\(Q). T a k i n g i n t o account (7.4.1) a n d (7.4.2)

we conclude t h a t series (7.4.5) a n d (7.4.6) also almost u n i f o r m l y converge i n I I +

and I I ~ , respectively. T h u s , we have o b t a i n e d the following

THEOREM 7.4.1. If A G A-k, then the function

1 1
CA(z) := A
2-iri (• - z) á

given by formulas (7.4.5) and (7.4.6) is holomorphic in H" U I T . +

COROLLARY 7.4.1. If A G A', then the function CA is holomorphic in XI"UII . +

DEFINITION 7.4.1. T h e f u n c t i o n CA defined b y (7.4.5) a n d (7.4.6) is called the

C a u c h y t r a n s f o r m of A .

7.5. T h e W i e n e r e x p a n s i o n of square i n t e g r a b l e d i s t r i b u t i o n s

It is clear t h a t V C S C A a n d t h a t these embedings are continuous.

THEOREM 7.5.1. A c V 2.
L

PROOF. Let <p be i n A. I n accordance w i t h D e f i n i t i o n 7.3.1, (p is i n L (R).

2

Note that

|P,WI = "¿6 2
+ ^) 2
- (7.5.1)

A n easy c o m p u t a t i o n shows t h a t

PÍ 1]
= - ( * ' + l)*Pi/+i + (2i/ + Vjipu - vipv-i. (7.5.2)

A c c o r d i n g to C o r o l l a r y 7.3.1 the f u n c t i o n (p c a n be represented as follows

<P = Y^vPv, C v = / <pp . v

F o r m a l l y differentiating this series we o b t a i n

Y c pi u
x)
= -i + l)c p„ u +1 + i ] P ( 2 z / + \)c p v v - i vc p - -
v v X

vez vez vez vez

B y v i r t u e of C o r o l l a r y 7.3.1 it follows t h a t the series ^ ] l y G Z (z/ + l)c , ^ ^ ^ ( 2 ^ 4 - 1 ) ^

i /

and J2vez v c
» a r e
absolutely convergent. Hence, b y (7.5.1) <pW is i n L (R). 2
In
general, it follows b y i n d u c t i o n t h a t ipW G L ( R ) for each k G N . T h u s , the p r o o f2

is finished. •
o
Let A be i n V' L2 ( R ) . D e n o t e b y A the r e s t r i c t i o n of A t o the set A.
o

THEOREM 7.5.2. A is in A!.

PROOF. Because of T h e o r e m 5.7.2, CA(- + ie) - CA(- - ie) tends to A i n V' . L2

Therefore, i n v i e w of T h e o r e m 7.5.1 we infer t h a t C A ( - - f ie) — CA(- — ie) converges

o
to A i n A!. T a k i n g i n t o account t h a t A is complete we conclude b y T h e o r e m A . l
o
t h a t A belongs to A!. •
 7.5. THE WIENER EXPANSION OF SQUARE INTEGRABLE DISTRIBUTIONS 137
COROLLARY 7.5.1. If A is in V' , then L2

A = Y C
vPv

in A ,
1
where c v — A(p ).
v

Since
CA(x + iy) - CA(x - iy) = A(g (x - •)) y

for y > 0, where g (x) y = i l y i (compare w i t h (5.6.1)), therefore

x 2

ix x -{-y
2

/ \ K i t w 1 Aí , (i(x + iy)-\r
i(x,y) := A(g (x y - •)) = —¡== g A
^ ( i ( a ! + i y ) + ly i+

V— — oo v v
2

1^
1 f A , , (i(x-zy)-lr , 7 „ .

F r o m this, for A b e l o n g i n g to V' L2 as a n i m m e d i a t e consequence of T h e o r e m 5.7.2

we o b t a i n the following

THEOREM 7.5.3. If A is in T>' , then the function given by (7.5.3) fulfils the
L2

Laplace equation Au — 0 in I l and l i m _ ^ + i'i y) = A in the sense of V'

+
y 0
u
L2

(compare with Theorem 5.6.4)-

 APPENDIX

SEQUENTIAL C O M P L E T E N E S S OF SOME SPACES

Let I? be a vector space over the field of c o m p l e x numbers. L e t us equip t h i s space

w i t h a f a m i l y of n o r m s || • | | fulfilling the inequalities
m

I M I m < I M L + i for (f e E

and m = 0 , 1 , 2 . . . .
A sequence (<p ) is called convergent t o <p i n E i f for m G N , \\ip — <p\\ tends
v v m

to zero as v oo. If for m G N , \\ip — ^ | | v goes to zero as v a n d fx t e n d t o

m

infinity, t h e n we say t h a t the sequence (tpv) is a C a u c h y sequence i n E.

A vector space is said t o be sequentially complete i f every C a u c h y sequence of
elements of E is convergent to a n element of E. A vector space e q u i p p e d w i t h a
f a m i l y of n o r m s || • | | , m G N , w h i c h is complete, is called a Fréchet space.
m

D e n o t e b y E' the space of a l l linear continuous forms o n E. W e say t h a t a

sequence ( A ^ ) , A G E' is a C a u c h y sequence i f for each ip i n E the n u m e r i c a l
v

sequence (A ((p)) is a C a u c h y sequence i n the u s u a l sense. T h e space E' is called

u

sequentially complete i f for every C a u c h y sequence ( A ^ ) the f u n c t i o n A defined b y

A{<p) = l i m A {<p)v (A.l)

for ip G E is a linear continuous f o r m o n E. A f t e r the above p r e l i m i n a r i e s we are

now i n a p o s i t i o n to prove the following

T H E O R E M A . l . If E is a Fréchet space, then its dual space E 1

is sequentially
complete.

P R O O F . It is easy to see t h a t the f u n c t i o n A defined by ( A . l ) is a linear f o r m o n

E. It remains to prove t h a t A is continuous. L e t (<Pv) be a sequence converging t o
zero. W e must show t h a t A((p ) —»> 0. L e t us assume the contrary. T h e n choosing,
u

if necessary, a subsequence we c a n assume t h a t l A ^ ^ ) ! > c > 0. N o w r e c a l l t h a t

convergence of a sequence (ip ) to zero i n E means t h a t ||<¿v|| —> 0 for m G N .
u m

A g a i n choosing a subsequence, we c a n assume t h a t

\WA\m < ¿ 7 , m = 0,1,2...,!/. (A.2)

P u t b y definition tjj = 2"(p , v G N . O f course, the sequence (?/v) is also convergent

u l/

to zero i n but
| A ( ^ ) | -> oo.

W e n o w define a subsequence (ipl) as follows: choose ip[ so t h a t l A ^ i ) ! > 1. N o w ,

since A (ip) —» A(ip) for ip G E, we c a n choose A ^ s u c h t h a t
u

| A i ( V i ) l > 1-
140 APPENDIX. SEQUENTIAL COMPLETENESS OF SOME SPACES

N o w suppose we have chosen K'- a n d i¡)'-, j — 1, 2 , . . . , v — 1. W e choose ip' u1 v > 1

to be one of the ip w i t h i n d e x so h i g h t h a t
v

| A i ( ^ ) | < 2¿fc, k = l,2,...,v-l, (A.3)

|A(^)l>El ^i)l A + I/
- ( -) A 4

T h e first is possible, because the ip converge to zero i n E a n d

u are i n E'. T h e
second is possible, because | A ( ^ ) | —> oo. Since A (ip) —>> A ( ^ ) for ip e E, we c a n u

choose AJ, from the (A¡,) sequence such t h a t

|AU<)|>X]I U^)I + ^ A
(A.5)

T h i s way we c a n go a n d construct the new infinite sequences (ip ) a n d (A' ). f

u u In
v i e w of ( A . 2 ) the sequence (s ), v s
u = Y^=i ^ s a
C a u c h y sequence i n E. E is a
Fréchet space, therefore there exists a ip i n E such t h a t ip = Y^jLi ^j- F u r t h e r , for
fixed v we have
v—1 oo

AU^) =E A
W + AU^)+ E A
W -
j=l j=v+l
T a k i n g i n t o account ( A . 3 ) a n d ( A . 5 ) we o b t a i n

v-1
\K(ip)\ > IAUOI E
j=l j=v+l

> \ K m \ -( $ > ' „ ( $ ) ! + E

Therefore | A J , ( ^ ) | —» oo as v —> oo. B u t this contradicts l i m ^ o o A' (ip) = A ( ^ ) . u

Consequently, A((p ) —>> 0 a n d the linear f o r m is therefore continuous. T h u s , the

u

p r o o f of the t h e o r e m is finished. •

W e shall n o w deal w i t h a more c o m p l i c a t e d p r o b l e m . L e t S — {E : 7 G F} 1

be a f a m i l y of vector subspaces of the vector space E, such t h a t E — \]^ _ E . Ç V 1

Suppose t h a t each of these spaces is a Fréchet space. W e i n t r o d u c e a convergence

i n the space E.
A sequence (ip ) is called convergent i n E i f it satisfies the following c o n d i t i o n s :
u

(i) (p„ E E^, v EN for some 7 G F ,

(ii) (ify) is a C a u c h y sequence i n E.
1

Sequences fulfilling (i) a n d (ii) w i l l be called o>convergent.

L e t E' denote the set of a l l linear forms defined o n E a n d continuous w i t h
respect to the a-convergence. W e u n d e r s t a n d C a u c h y sequences a n d the s e q u e n t i a l
completeness as i n the above case. W e s h a l l n o w formulate the following
 APPENDIX. SEQUENTIAL COMPLETENESS OF SOME SPACES 141

THEOREM A . 2. If E is a vector space equipped with an a-convergence, then

the space E' of all linear forms defined on E and continuous with respect to this
convergence is sequentially complete.

Since the p r o o f of this t h e o r e m is m u c h the same as the p r o o f of the p r e v i o u s

t h e o r e m we w i l l not give i t here.

EXAMPLE A . l . L e t Q be a n o p e n set i n R a n d K C Q be a c o m p a c t
N
set.
D e n o t e by V(K) the set of a l l ip b e l o n g i n g to V(Q) such t h a t supp ip C K. Of
course, V(K) is a vector subspace of V(Vt). It is clear t h a t

2>(fi) = (J V{K).
Ken
I n accordance w i t h D e f i n i t i o n 1.1.1, V(K) is a Fréchet space a n d the convergence
i n V(Q) is o>convergence. Therefore V(il) is sequentially complete. I n p r a c t i c e ,
K is always the closure of some o p e n set.

W e s h a l l n o w collect some i n f o r m a t i o n c o n c e r n i n g b i l i n e a r forms. Suppose t h a t

Ei a n d E are two vector spaces over the c o m p l e x field C a n d t h a t {•, •) is a m a p p i n g
2

of Ei x E i n t o C . T h e n , for each fixed ip G E , (•, •) defines a p a r t i a l m a p p i n g

2 2

(•, ip) of Ei into C ; a n d s i m i l a r l y , (ip, •) defines a p a r t i a l m a p p i n g of E i n t o C . If a l l 2

the p a r t i a l m a p p i n g s (•, ip) a n d (ip, •) are linear, t h e n (•, •) is c a l l e d a b i l i n e a r f o r m

o n Ei x E . L e t us equip these spaces w i t h countable families of n o r m s || • | | i a n d
2 m >

II • || 2 fulfilling the inequalities

M

m.l — IML+1,1
and
IM|m,2 < ll^llm+1,2
for (p G Ei a n d ip £ E , 2 respectively. M o r e o v e r , we assume t h a t E 2 is complete. If
the p a r t i a l linear m a p p i n g s (ip, •) a n d (•, ip) are continuous, t h e n we say t h a t the
b i l i n e a r f o r m (•, •) is separately continuous. W e are able to prove the following

THEOREM A . 3 . Assume that the spaces Ei and E have the properties

2 listed
above. If (•, •) is a bilinear form on Ei x E 2 separately continuous, then it is
continuous.

PROOF. W e have to show t h a t there exist a constant M > 0 a n d nonnegative

integers m a n d n such t h a t

|(^,^)| < M | | ^ | | , i | | ^ | |
m n | 2

for (p G Ei a n d ip £ E . O n the contrary, suppose t h a t our b i l i n e a r f o r m is not

2

continuous, therefore for every M > 0 a n d m , n G N there exist ^ , , M a n d ipm,n,M m n

such t h a t
)\>M\\
(pm,n,M IIm,l W^m^n.M ||n,2 •
Set m = n , M = n , t h e n we have
2

\(<p
n,n,n •> 1pn,n,n )| >
2 2 n
| | ^ n , n , n | | n , l | | ^n.n.n
2 2
||n,2-
F o r s i m p l i c i t y of n o t a t i o n we p u t
— 1 ^n,n,n 2
-7 1 ^n,n,n 2

:
= -TF Ñ) Wn = -TP?
:
n"
n\<Pn,n,n4 U ^n,n,n 2
142 APPENDIX. SEQUENTIAL COMPLETENESS OF SOME SPACES

Therefore |(y? , ip )\ > 1- W e s h a l l n o w show t h a t t h e <p converge t o zero i n E\

n n n

a n d the ip converge t o zero i n E .

n 2 Note that

\Wn+k\\k,i < \\p \\ +n,in+k k = & = 0 , . . . , n = 1,2,...

T h i s i m p l i e s t h a t ||^ +fc||fc,i tends t o 0 as n —> 0 0 . F r o m t h i s we infer t h a t tp , ip

n n n

t e n d t o zero i n E\ a n d E , respectively. T h e n , choosing i f necessary a subsequence,

2

we m a y assume t h a t

ll^i/llfc.i < ¿ 7 a n d
ll^i/IU.2 < ¿

for /c = 0 , 1 , . . . ,ZA
P u t ^ = 2 y? a n d V v = 2 " ^ . 1/
i/ T h e n the sequence ((p^^ip^)
converges t o ( 0 , 0 ) i n E\ x E . B u t 2

|(^,^)| >2 \ 2
i/ = 0 , l , . . . (A.6)
O f course, |(v?i, > 1- P u t (p[ — <pi a n d ip[ — ip\. B y the a s s u m p t i o n t h e
f u n c t i o n ((pi, •) is continuous o n E a n d ( A . 6 ) holds. M o r e o v e r , t h e ip converge t o
2 u

zero i n E . H e n c e i t follows t h a t one c a n choose ip f r o m the sequence (ip ) a n d ip'

2 2 u 2

from the sequence (ip ) so t h a t u

\(<Pi,1>2)\<\ and 1(^,^)1 > l < V i ^ 2 ) l + 2 (A.7)

N o w , suppose we have chosen p'- a n d ip'- for j = 1,..., 1/ — 1. W e choose ?// t o b e

one of t h e (ip ) sequence w i t h i n d e x so h i g h t h a t
u

K4-.#,)l<2¿7, j = 0,l-..,v-l. (A.8)

S u c h choice is possible because the functions (<¿/-, •) are continuous o n E 2 and the
ip u converge t o zero i n E . 2 T a k i n g i n t o account ( A . 6 ) we c a n choose ip' f r o m t h e v

sequence Lp so t h a t v

1(^^)1 > E l ( ^ ' ^ ) l + ^ (A.9)

Since ||Vv||fc,2 £ ¿ 7 for & = 0 , 1 , . . . , v a n d E 2 is complete therefore there exists ip

in E 2 such t h a t ip = Y^Li tPl- F u r t h e r , for fixed v we have
v— 1 00

<^,vo = E < ^ > + <^>^> + E <ri»^>-

i=i j=^+i
T a k i n g i n t o account ( A . 8 ) a n d ( A . 9 ) we o b t a i n

W^)\>\{v' ,<)\ v

j=l J=^+l

( i/ — 1 00 \

j=i ¿=«>+i /
Therefore | ^ ) | —» 0 0 as v —> 0 0 . B u t t h i s c o n t r a d i c t s t h e c o n t i n u i t y of (•, ip)
in E . x •
 SUBJECT INDEX

a n a l y t i c representation of an integrable — of finite order, 35

d i s t r i b u t i o n s , 84 — periodic, 128
of a d i s t r i b u t i o n i n T>' , 88
LP — regular, 2
— tempered, 96, 130
C a r l e m a n transform, 117, 120 — w i t h compact support, 34, 76, 129
C a r t e s i a n p r o d u c t , 37, 82, 96 — w i t h point support, 36
C a u c h y finite part, 3, 91 du B o i s - R y m o n d ' s l e m m a , 6, 14
— p r o b l e m , 60, 123
— semigroup, 121 E x t e n s i o n of distributions from V(Q) to
— transform of a d i s t r i b u t i o n i n > 120 C ° ° ( 0 ) , 33
a d i s t r i b u t i o n i n T>' , 87
LP

an element of A', 134 formula B o r e l ' s , 101, 111, 121

an integrable d i s t r i b u t i o n , 84 — d ' A l e m b e r t ' s , 62
Closed operator, 123 — Poisson's ( s u m m a t i o n formula), 127
convergence a-convergence, 140 — S o c h o c k i - P l e m e l j , 8, 94
— ^-convergence, 75 Fourier representation, 25, 134
— i n the sense of Wj™' , p
16 — transform a n d inverse Fourier transform
loc ' of a square integrable function, 103
of D , 1 of a tempered d i s t r i b u -
of £>', 44 t i o n , 108
of 5 , 95 of an integrable function,
of s', 109 98
of B, 75 of a square integrable d i s t r i b u t i o n ,
of 34, 72 114
of Co (ft), 38 of a tempered function, 102, 118
of W ' P ( a ) , 19
m

of an integrable d i s t r i b u t i o n , 110
of V' , 81
LV
of c o n v o l u t i o n p r o d u c t of integrable
of T>LP , 77, 81
distributions, 111
c o n v o l u t i o n operator, 44
of H e r m i t e function, 107
— product of distributions i n V' , 83
LP
of tensor product of integrable d i s t r i -
w i t h regular supports at infin¬
butions, 111
ity, 57
Fréchet space, 139
of functions i n L , 82 p

function B o r e l ' s , 39, 78

of square integrable distributions, 83,
— H a a r ' s , 25
112
— harmonic, 86
covergence i n the sense of T>', 8
— Heaviside's, 13, 62, 72, 107, 115, 126
differentiation of a d i s t r i b u t i o n , 11 — Hermite's, 107, 122
— of convolution product of distributions, — holomorphic, 84, 88, 136
58 — i n W™f(Q), 16

— of tensor product of distributions, 47 — i n A, 131

D i r i c h l e t p r o b l e m , 87 — i n Ak, 132
d i s t r i b u t i o n belonging to 2 ) ^ , 112, 136 — in V ,
LP 77, 81
2

to V' , 81 — i n 0 , 96
M
LP

— integrable, 75, 86 — i n B , 75
144 SUBJECT INDEX

— in C o ( O ) , 38 P a l e y a - W i e n e r t y p e theorems, 119
— i n C ( 0 ) , 6, 38
c
Parseval's theorem, 99, 104
— i n W 'P(fL),
Tn
19 Plancherel's theorem, 104, 121, 122
— i n W™' (Q), P
21 Poisson's integral, 87
— tempered, 112, 118 product of distributions, 96, 115
— W i e n e r ' s , 131, 134 — of the distributions, 10
fundamental solution, 59, 63, 65, 72, 74
regularization of a d i s t r i b u t i o n , 41
heat operator, 63, 74 — of a d i s t r i b u t i o n w i t h compact s u p p o r t ,
H e r m i t e ortogonal system, 107 72
— p o l y n o m i a l , 107 — of a function, 5, 17
H u b e r t transform, 116 — of an integrable d i s t r i b u t i o n , 78
h y p o e l l i p t i c operator, 74 restriction of a d i s t r i b u t i o n , 31
R i e m a n n - L e b e s g u e l e m m a , 103, 110
inequality H o l d e r ' s , 83, 88 Riesz's theorem, 39, 70, 82
— Hôrmander's, 66, 70 R i t z ' s theorem, 24
— M i n k o w s k i ' s , 3, 18, 90 R i t z - G a l e r k i n m e t h o d , 24
— Poincaré's, 68
— Y o u n g ' s , 5, 20, 82, 89, 113, 121 Schwartz kernel theorem, 49
infinitesimal operator, 121 — representation of an operator, 49
integration w i t h respect to a complex sequentially completeness, 139
measure, 39 sets regular at infinity, 56, 60
to product measure, 78 s m o o t h p a r t i t i o n of unity, 28
Sobolev i m b e d d i n g theorem, 81
Jordan's l e m m a , 106 — space, 19
support of a d i s t r i b u t i o n , 33
kernel of an operator, 49
— of c o n v o l u t i o n product of d i s t r i b u t i o n s ,
L - s o l v a b i l i t y , 70
2
58
Laplace operator, 63, 74 — of tensor product of d i s t r i b u t i o n s , 48
local representation of a d i s t r i b u t i o n , 38 — singular, 73
— Sobolev space, 16
tensor product of distributions, 45
Lusin's lemma, 7
of integrable d i s t r i b u t i o n s , 79
M a l g r a n g e - E h r e n p r e i s theorem, 72 of tempered d i s t r i b u t i o n s , 98
measure B o r e l ' s , 21, 39, 86
wave equation, 12, 60, 74
— D i r a c ' s , 2, 35, 36, 27, 42, 87, 93
weak derivative, 12, 19
— complex, 38
W i e n e r expansion, 136
M e y e r s - S e e r i n theorem, 29
— orthogonal system, 131
N e w t o n i a n p o t e n t i a l , 63
 NOTES AND REFERENCES TO THE LITERATURE

Chapter 1
Lemma 1.4.l and Theorem 1.5.l are taken from R.A. Adams's book: Sobolev
Spaces ([l]). Example 1.6.2 is taken from S.G. Krein's book: Functional Analysis
([18]). Example 1.12.1 appears in J.T. Marti's book: Introduction to Sobolev
Spaces and Finite Element Solution of Elliptic Boundary Value Problems ([19]).
The contents of Section 1.13 were originally produced by the authors in [16].

Chapter 2
The contents of Sections 2.1, 2.3 and 2.4 are a modification of some of L. Hor­
mander's book: The Analysis of Linear Partial Differential Operators 1 ([ll]). The
proof of Theorem 2.2.l appears here as in the paper: N. Meyers and J. Serrin,
H = W ([20]). Example 2.3.1 is originay due to V.C. Vladimirow in the book:
Generalized Functions in Mathematical Physics ([25]). The idea of applying the
representing theorem of linear continuous forms on Cartesian products of Banach
spaces to proving representation theorems on continuous linear forms in the theory
of distributions was earlier used in the books: J. Horwath, Topological Vector
Spaces and Distributions ([12]) and V.C. Vladimirow, Generalized Functions in
Mathematical Physics ([25]).
Chapter 3
Sections 3.1, 3.2, 3.3 and 3.4 are essentially an adaptation of some of L. Hor­
mander book: The Analysis of Linear Partial Differential Operators 1 ([ll]). The
idea of the sets regular at infinity is originally due to L. Schwartz: Theorie des
Distributions ([24]).
Chapter 4
The contents of Sections 4.1 and 4.2 are inspired by C.H. Wilcox's article: The
Cauchy Problem for the Wave Equation with Distribution Data ([26]). The results
of Sections 4.4 and 4.5 are taken from an article of J.P. Rosay: A Very Elementary
Proof of the Malgrange-Ehrenpreis Theorem ([22]). The material in Section 4.6
is taken from G.B. Folland's books: Introduction to Partial Differential Equations
( [6]) and Partial Differential Equations ([7]).

Chapter 5
The spaces 1J� p appear in L. Schwartz's book: Theorie des Distributions ([24]).
The results in Sections 5.1, 5.2, 5.3, 5.4 and 5.5 concerning these spaces are pre­
sented in another way than in [24], simply. The concept of the Cauchy transforma­
tions was introduced by H. Bremermann for distributions belonging to £' and 0�
in the book: Distributions, Complex Variables and Fourier transforms ([3]). The
146 NOTES AND REFERENCES TO THE LITERATURE

m a t e r i a l presented i n S e c t i o n 5.6, 5.7 a n d 5.8 is t a k e n f r o m the article: W . K i e r a t

a n d K . Skórnik, O n G e n e r a l i z e d F u n c t i o n s a p p r o a c h to the C a u c h y P r o b l e m for
H e a t E q u a t i o n ([14]).
Chapter 6
T h e o r e m 6.11.2 is o r i g i n a l l y due to E . J . B e l t r a m i a n d M . R . W o h l e r s : D i s ­
t r i b u t i o n s a n d the B o u n d a r y V a l u e s of A n a l y t i c F u n c t i o n s ([2]). T h e results of
S e c t i o n 6.12 are p a r t l y presented i n article: W . K i e r a t a n d U . Skórnik, O n some
R e l a t i o n s between the C a u c h y a n d F o u r i e r Transforms of D i s t r i b u t i o n s ([15]). T h e
m a t e r i a l i n c l u d e d i n Sections 6.13, 6.15 a n d p a r t l y i n 6.14 is p u b l i s h e d for the
first t i m e here. T h e results of S e c t i o n 6.14 are t a k e n from the article: W . K i e r a t
a n d U . S z t a b a , T h e G e n e r a l i z e d F o u r i e r Transforms of S l o w l y Increasing F u n c t i o n s
([17]). T h e contents of S e c t i o n 6.16 were p u b l i s h e d for the first t i m e i n the article:
W . K i e r a t , A R e m a r k a b o u t the Infinitesimal O p e r a t o r of the C a u c h y S e m i g r o u p
([13]).
Chapter 7
T h e t h e o r y of p e r i o d i c d i s t r i b u t i o n s i n c l u d e d i n Sections 7.1 a n d 7.2 is o r i g i n a l l y
due to L . Hôrmander, T h e A n a l y s i s of L i n e a r P a r t i a l Differential O p e r a t o r s 1 ([11]).
It is presented i n a more elementary f o r m here. T h e contents of Sections 7.3 a n d 7.4
are m o d i f i c a t i o n s of the article: A . C i c h o c k a a n d W . K i e r a t , A n A p p l i c a t i o n of the
W i e n e r F u n c t i o n s to the D i r i c h l e t P r o b l e m of the L a p l a c e E q u a t i o n ([5]).

Appendix
T h e p r o o f of T h e o r e m A . l is due to M . S . B r o d s k i a n d it is i n the book:
I . M . G e l ' f a n d a n d G . E . S h i l o v : G e n e r a l i z e d F u n c t i o n s , v o l u m e I ([8]).
 BIBLIOGRAPHY

R . A . A d a m s , Sobolev Spaces, A c a d e m i c Press, N e w Y o r k , L o n d o n , 1975.

E . J . B e l t r a m i and M . R . Wohlers, Distributions and the Boundary Values of Analytic Func­
tions, A c a d e m i c Press, N e w Y o r k , L o n d o n 1966.
H . B r e m e r m a n n , Distributions, Complex Variables and Fourier Transforms, Addison-
Wesley P u b l i s h i n g C o m p a n y I N C , R e a d i n g , Massachusetts 1965.
T . C a r l e m a n , L'intégrale de Fourier et questions qui s'y rattachent, A l m q v i s t a n d W i k s e l l ,
U p p s a l a , 1944.
A . C i c h o c k a a n d W . K i e r a t , An Application of the Wiener Functions to the Dirichlet
Problem of the Laplace Equation, Integral Transforms and Special Functions, V o l . 7, N o . 1—
2, pp. 13-20, 1998.
G . B . F o l l a n d , Introduction to Partial Differential Equations, P r i n c e t o n U n i v e r s i t y Press
and U n i v e r s i t y of T o k y o Press, P r i n c e t o n , N e w Jersey, 1976.
G . B . F o l l a n d , Partial Differential Equations, Springer-Verlag, B e r l i n , Heidelberg, N e w
Y o r k , Tokyo, 1983.
I. M . G e l ' f a n d and G . E . Shilov, Generalized Functions, V o l u m e 1, A c a d e m i c Press, N e w
Y o r k , L o n d o n , 1964.
I . M . G e l ' f a n d and G . E . Shilov, Generalized Functions, V o l u m e 2, A c a d e m i c Press, N e w
Y o r k , L o n d o n , 1968.
H . J . Glaeske, On the Wiener-Laguarre Transform of Generalized Functions, Generalized
Functions and Convergence, (Editors P. A n t o s i k and A . K a m i n s k i , M e m o r i a l V o l u m e for
Professor J a n M i k u s i n s k i ) W o r l d Scientific P u b l i s h i n g C o . P t e . L t d . , Singapore, N e w Jer­
sey, L o n d o n , H o n g K o n g , 1990.
L . Hôrmander, The Analysis of Linear Partial Differential Operators 1 ( Distribution
Theory and Fourier Analysis), Springer-Verlag, B e r l i n , Heidelberg, N e w Y o r k , Tokyo,
1983.
J . Horváth, Topological Vector Spaces and Distributions, V o l u m e 1, A d d i s o n - W e s l e y P u b ­
lishing C o m p a n y I N C , R e a d i n g , Massachusetts, 1966.
W . K i e r a t , A Remark about the Infinitesimal Operator of the Cauchy Semigroup, Integral
Transforms and Special Functions, V o l . 4, N o . 3, pp. 243-248, 1996.
W . K i e r a t a n d K . Skórnik, On Generalized Functions approach to the Cauchy Problem
for Heat Equation, Integral Transforms and Special Functions, V o l . 2, N o . 2, pp. 107-116,
1994.
W . K i e r a t and U . Skórnik, On some Relations between the Cauchy and Fourier Transforms
of Distributions, Integral Transforms and Special Functions, V o l . 3, N o . 4, pp. 269-278,
1995.
W . K i e r a t and U . Sztaba, Differential Equations of the Second Order with Measures as
Coefficients, Annales M a t h e m a t i c a e Silesianae 10, K a t o w i c e , pp. 79-85, 1996.
W . K i e r a t and U . Sztaba, The Generalized Fourier Transforms of Slowly Increasing Func­
tions, Demonstratio M a t h e m a t i c a , V o l . X X X , N o . 2, pp. 425-428, 1997.
S . G . K r e i n , Functional Analysis, (Russian), N a u k a , M o s c o w , 1964.
J . T . M a r t i , Introduction to Sobolev Spaces and Finite Element Solution of Elliptic Bound­
ary Value Problems, A c a d e m i c Press, L o n d o n , N e w Y o r k , Sydney, Tokyo, Toronto, 1986.
N . Meyers and J . Serrin, H=W, P r o c . N a t . A c a d . S c i . , U S A , 51 pp. 1055-1056, 1964.
 BIBLIOGRAPHY

M . R e e d and B . S i m o n , Methods of Modern Mathematical Physics, V o l u m e 2, A c a d e m i c

Press, N e w Y o r k , San Francisco, L o n d o n , 1975.
J.P. Rosay, A Very Elementary Proof of the Malgrange-Ehrenpreis Theorem, T h e A m e r i c a n
M a t h . M o n t h l y , v o l . 98, N o . 6, pp. 518-523, 1991.
W . R u d i n , Real and Complex Analysis, M c G r a w - H i l l B o o k Company, N e w Y o r k , Toronto,
Sydney, L o n d o n , 1966.
L . Schwartz, Théorie des Distributions, H e r m a n n , P a r i s , 1966.
V . C . V l a d i m i r o v , Generalized Functions in Mathematical Physics, (Russian), N a u k a ,
M o s c o w , 1976.
C . H . W i l c o x , The Cauchy Problem for the Wave Equation with Distribution Data: on
Elementary Approach, T h e A m e r i c a n M a t h . M o n t h l y , 98, pp. 401-410, 1991.
A . H . Z e m a n i a n , Generalized Integral Transformations, Interscience Publishers, N e w Y o r k ,
1968.
 INDEX OF SYMBOLS

e», 1 Di-, 75
supp (p, 1 V' , Ll 76
V(Ü), 1 Z>LS, 7 7
V'(Q), 1 V ,81
LP

<5, 2 CA, 8 4
v <s v,
0 4 S, 9 5
$', 96
B (Q),
r 5 O M , 96
C (0),
C 6 J^"f, 98
J - 1
/ , 99
D A,
a
11 f, 1 0 0
WC (íí), 1 6 P
S , 112
0

U, 16 ftT, 1 1 6
W M
' ( 0 ) ,
P
19 T' /,1
117
^ ' (n), 2 i
0
m p
117
supp A , 33
S, 122
' C ( O ) , 38
0
A 131
C ( Q ) , 38
A^, ^fc, 1 3 2
C

42
A-k, 133
A * <p, 42
.A', 133
5 <g> T , 47
€', 145
sing supp A , 73
145
B, 75