This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
0
denote by
Em;
since
IIS(:)vnil
with
t
Given
i s t h e supremum of t h e sequence
m(Z)
a dense sequence i n t h e unit sphere of Moreover,
S ( t ) E m E Em for
t h e norm of t h e r e s t r i c t i o n of
i s separable,
with
i s t h e closed subspace generated by t h e
i s separable and
complement of t h e n u l l s e t
En
and a separable subspace
hence i f Em
n’
en
to
El, {vn]
Em and is i t s e l f measurable.
em we have
m(s
+ t)
=
u E Em, hdl 5 11 5 SUP f lb(s>vli; v E Em, b l l < _ m ( t > l Accordingly, a c o n t r a d i c t i o n r e s u l t s from:
1/s(s)S(t)uli;
< - m(s)m(t).
LEW 1.2.
t
defined i n
Let
1 0
m( 0, t ,d e ) ,
m(s)m(t) (0,
c a< B
_ we have eU d U ( a - d) 2 [ O , a ] s o t h a t Id] + ] a - dl >_ a , where 1.1 indic a t e s Lebesgue measure. But Id1 = la - d l , hence Id1 >_ a/2. Assume now t h a t m($) is unbounded i n [ a , f31, so that t h e r e e x i s t s Proof.
Let
a sequence
[anl
a > 0.
If
t h e r e with
s
= a, t
m(an) +
m.
.
Applying t h e argument above we
5
FIRST ORDER EQUATIONS
deduce t h e existence of a measurable set
dn
[0, p ]
in
with
.Jm(an)
i n dn, which c o n t r a d i c t s t h e a/2 and m ( t ) 2 /dnl 1 an/2 f a c t t h a t m(t) i s everywhere f i n i t e . This completes t h e proof of
Lemma 1.2. End of proof of Theorem 1.1. Let 0
I h / 0.
bargained f o r i n t h e d e f i n i t i o n .
I n t h e uniform
Co exp ( & )
un(0)
t 20.
exp ( - u t ) u ( t )
4
u(0)
so then
When t h e Cauchy +
exp ( - & ) u ( t )
This is considerably more t h a n we
6
FIRST ORDER EQUATIONS
The Cauchy problem i n
$1.2
my.
(-m,
We d e c l a r e t h e Cauchy problem f o r (1.1)w e l l posed i n
-a
< t
a ) ) .
(-m,
i s s t r o n g l y continuous f o r a l l t. t ? 0.
The proof of (2.1) is t h e same a s that f o r t h e case
To show
(iii)we note that t h e assumptions imply that t h e Cauchy problems f o r
both (1.1)and U ' ( t ) = -Au(t) a r e w e l l posed i n ' t
2
0,
(2.3)
u(t^)
s i n c e t h e correspondence
~ ( - 2 ) maps
-t
s o l u t i o n s of (1.1) i n t o s o l u t i o n s of (2.3) and v i c e versa. be t h e propagator of (2.3) i n
t 2 0.
s (t)= Applying Theorem l . l ( i i i ) t o h
we deduce t h a t
S(t)
S(t^)
kt
~-(t)
We v e r i f y e a s i l y t h a t
S(-t)
and t o
(t 2 0)
.
is s t r o n g l y continuous i n t < 0
t > 0
in
S ( - c ) = S-(t^)
and
t > 0.
However,
S(h)u = S ( t + h)S(-t)u,
follows.
In p a r t i c u l a r , t h e Cauchy problems f o r (1.1)and ( 2 . 3 ) a r e
thus s t r o n g c o n t i n u i t y a t
t
= 0
FIRST ORDER EQUATIONS
7
t L 0 which y i e l d s ( 2 . 2 ) by i n t e r v e n t i o n o f Obviously, t h i s i n e q u a l i t y implies t h a t t h e Cauchy
uniformly w e l l posed i n Theorem l . l ( i i ) .
problem i s w e l l posed i n
(-m,
thus completing t h e proof of
m),
Theorem 2.1.
$1.3
The Hille-Yosida theorem. Assume that t h e Cauchy problem f o r (1.1) i s w e l l posed i n
that
is closed.
A
number with
Re
X>
Let
w
R(X)u =
(1.9), X
be t h e constant i n
E
Define an operator i n
w.
[
e
-Xt
S(t)u d t
and
t 2 0
a complex
by
(u
E
.
E)
(3.1)
I
Since t h e norm of t h e integrand is bounded by
ColIu/I exp ( w
is a bounded operator i n
u
Assume now t h a t
E.
s o l u t i o n of (l.l), s o t h a t
Then
D.
E
-
Re X)t, R ( X ) h
S(t)u
is a
s l ( t ) u = AS(t)u. (l) Using a w e l l known r e s u l t
on i n t r o d u c t i o n of closed operators under t h e i n t e g r a l s i g n (HILLE-PHILLIPS
[1957:1,p. 831) we obtain A
LT
-
e-‘%(t)u
dt =
L~
.,
= e -XT S(T)u
for
T > 0.
R(X)u
E
ment of
Letting
D(A)
E,
and
T
-t
00
e-XtAS(t)u d t = JOT e-XtSt ( t ) u d t
-
u
+ X L T
e-XtS(t)u d t
and using closedness of
AR(X)u = XR(X)u
choose a sequence
- u.
{unj
-
-
A ) R ( X )= I
I n p a r t i c u l a r , (3.2) shows t h a t
w e l l one-to-one. Then
(XI
-
-u
(XI
-
X
E
p(A),
A)-l = R(X), The f a c t t h a t
p(A)
A):D(A)
t h e resolvent s e t of p(A)
u. Then
-t
i s onto.
E
u
A
E D(A)
It is a s with
Au = Xu.
/lu(t)II =
u = 0.
W e have then
and t h a t
contains t h e half plane
R(X;A) = Re
X>
w.
i s nonempty allows f o r t h e complete i d e n t i f i c a -
t i o n of t h e subspace Do of “admissible i n i t i a l d a t a ” c o n s i s t i n g of a l l uo such t h a t X€p(A),
S($)uo
and
R(X)E C_ D ( A )
(3.2)
i s a s o l u t i o n of (1.1)with
s o that
-t
.
(exp (Re X)t)(/ull which c o n t r a d i c t s (1.9) u n l e s s proved t h a t
un
so that
To s e e t h i s assume t h e r e e x i s t s
u ( t ) = ( e m (Xt))u
i s an a r b i t r a r y e l e -
such t h a t
D
R(X)un -+ R(X)u, AR(X)un= XR(X)un u n + XR(X)u (XI
u
Now, i f
in
we deduce that
A
is a s o l u t i o n of (1.1). This is done a s follows.
E
E
If
R(X;A)S(t)u is a s o l u t i o n of (l.l), h e n c e i t f o l l o w s f r o m ( 1 . 6 ) t h a t
a
FIRST ORDER EQUATIONS
(t > 0)
R(X;A)S(t)u = S(t)R(X;A)u
u
This e q u a l i t y extends t o a l l
E by denseness of
E
(3.3) f o r u
Write
(XI
=
-
v E D(A)
A)v,
D
(3.3)
and implies t h a t
.
( t 2 0)
S(t)D(A) C_ D(A)
.
(3.4) (XI
and then apply
-
A)
to
The r e s u l t is
both s i d e s ,
AS(t)u = S(t)Au
u
valid f o r
E
Jb
S(t u - u =
for
u
E
D.
Apply
(3.5 we obtain
Making use of
D(A).
R(X;A)
t
t S'(s)u
"
t o both s i d e s and u s e
R(X;A)(S(t)u
u
E;
E
(3.3):
the r e s u l t i s
t
-
U)
S(S)AR(X;A)U
=
Since both s i d e s a r e bounded operators of to all
S(s)Au ds
ds =
u
.
ds
t h e e q u a l i t y can be extended
using (3.3) again and applying
(XI
-
t o both s i d e s ,
A)
we obtain
s(t)u
-
u
=
rt
AS(s)u ds
(3.6)
JO
for a l l
u
E
D(A).
(1.1)f o r any
u
Obviously, t h i s implies t h a t E
A,
D(, = D(A) We can obtain estimates for R ( X ; A ) formula
= (-l)nn!R(X;A)nfl
R(X;A)(II)
S(;)u
i s a s o l u t i o n of
hence
.
(3.7)
and i t s powers using t h e well known and d i f f e r e n t i a t i n g
(3.1) under
t h e i n t e g r a l s i g n ( t h i s can be e a s i l y j u s t i f i e d on t h e b a s i s of t h e dominated convergence theorem).
results for
where ities
u
E
E
and
Re
X>
The formula
0.
Using (1.9) we obtain
Co and w a r e t h e constants i n (1.9). We show next t h a t inequal(3.9) a r e as well s u f f i c i e n t f o r uniform well posedness of t h e
9
FIRST ORDER EQUATIONS
Cauchy problem for (1.1).
3A
THEOREM 3 . 1 (Hille-Yosida).
f o r (1.1)i s 5
and
w
The Cauchy problem
t 2 0 with propagator S ( t ) u(A) i s contained i n t h e half-space
uniformly w e l l posed i n
s a t i s f y i n g ( 1 . 9 ) i f and only i f Re X
b e closed.
R(X;A)
(3.9).
s a t i s f i e s inequalities
The proof of t h e other h a l f below
We have a l r e a d y proved one h a l f .
i s perhags not t h e s h o r t e s t a v a i l a b l e but can be adapted with minor changes t o equations o t h e r than (1.1) ( s e e Chapter 11). We begin by constructing s o l u t i o n s of (1.1)given s u f f i c i e n t l y "smooth" i n i t i a l conditions.
u
E
D(A 3 ),
> w, 0.
w'
Let
Define ( 2 )
N
u(t;u) = u
+
tAu
t2 2 +A u 2 w'+im
+
(3.10)
eXt
Ju'-im
&i
3u ax - R(X;A)A
( t 5 0)
,3
.
A deformation of contour shows t h a t t h e i n t e g r a l i n (3.10) vanishes for
t
5 .O,
thus N
u(0,u) = u ~ ~ ~ ( t=; O u ()e q ~ (~w ' t ) )
Obviously,
as
t
.
(3.11)
+
m.
An e a s i l y j u s t i f i a b l e
d i f f e r e n t i a t i o n under t h e i n t e g r a l s i g n shows that continuous d e r i v a t i v e in
t > 0;
;(E;u)
on t h e o t h e r hand
A
has a can be introduced
under t h e i n t e g r a l s i g n with convergence of t h e r e s u l t i n g i n t e g r a l , s o that
_ 0
E-valued continuous f u n c t i o n defined i n
f ( t ) = O ( e q (at))
such that
(Sf)(:)
f(?)
b e t h e Laplace transform of
as
t
+
f o r some
w
CY
and l e t
t,
(Sf)(X) =
e-Xtf(t)
dt
.
Then
t > 0.
uniformly on compacts of
The proof i n t h e s c a l a r c a s e can b e found i n WIDDER [1946:1, Chapter V I I ] .
The extension t o v e c t o r valued functions i s immediate.
End of proof of Theorem 3.1.
We apply t h e inversion formula (3.14)
in (3.10); i n view of (3.13) we obtain
t o t h e function
(3.15)
s o t h a t , u s i n g i n e q u a l i t i e s (3.9) w e o b t a i n IllY(t;u)ll
5 Collu/I lim
n
Define N
N
S(t)u = u(t;u) for
u
E
ment of
D ( A 3 ),
u(t)
s(%) we
operator such t h a t
(t
0)
t h e solution i n (3.10).
can extend each E ( t )
g($)
.
Arguing as i n t h e t r e a t -
t o a bounded everywhere defined
i s s t r o n g l y continuous i n
t 2 0
and
C l e a r l y , t h e proof of Theorem 3 . 1 w i l l be complete i f we show t h a t any solution
u(t)
of (1.1)admits t h e r e p r e s e n t a t i o n U(t) = iqt)U(O)
.
(3.18)
11
FIRST ORDER EQUATIONS
This is done as follows.
It r e s u l t s from t h e d e f i n i t i o n of for all
R(X;A)S(t)u = g ( t ) R ( X ; A ) u
u
thus f o r a l l
D(A 3 )
E
g(t) u
E
that
E.
Ace ord i n g l y ,
-S ( t ) A u = A i ( t ) u m(t^)
On t h e o t h e r hand,
(u
t 2 0)
D(A),
E
.
(3.19)
i s a continuously d i f f e r e n t i a b l e
= S(i)R(h;A)3
W(t)
o p e r a t o r valued f u n c t i o n t h a t s a t i s f i e s
Accordingly, i f
= Am(t).
i s a n a r b i t r a r y s o l u t i o n of (1.1)we have
u(^t)
d as m ( t
-
+ m ( t - s)Au(s)
(3.19), hence R(X;A)3u(t)
after
-
(XI
Applying
3.3.
= 0
(0
5 s 5 t)
(3.20)
= m(O)u(t) = h ( t ) u ( O ) = R(X;A)3S(t)u(0).
u(A)
(-m,
be c l o s e d .
A
uniformly w e l l posed i n i f and only i f
s)u(s)
t o b o t h s i d e s (3.18) r e s u l t s .
A)3
Corresponding t o t h e c a s e
THEORFM
-
s ) u ( s ) = -Am(t
(-m,
m)
we have
a)
The Cauchy problem f o r (1.1)&
w i t h propagator
i s contained i n t h e s t r i p
satisfying (2.2)
S(%) /Re
XI 5 w and R(X;A)
w,
n
satisfies
5
IIR(X;A)n(l Proof.
Co( /Re
XI -
( IRe XI
=
0,1,. . . )
.
(3.21)
We have a l r e a d y observed t h a t i f t h e Cauchy problem f o r (1.1)
i s w e l l posed i n
t h e n t h e Cauchy problem f o r ( 2 . 3 ) ( o f course
m)
(-m,
a l s o f o r (1.1))i s w e l l posed i n
t 2 0.
i n e q u a l i t i e s (3.21) f o l l o w from Theorem
Since
3.1.
R(X;-A)
= -R(-X;A),
Conversely, we deduce from
t h e same theorem t h a t i f (3.21) holds t h e n t h e Cauchy problem f o r (1.1) and (3.2) i s w e l l posed i n
t 2 0.
t 2 0 with
S
and
S- s a t i s f y i n g (1.9) i n
T h i s i s e a s i l y s e e n t o imply t h a t t h e Cauchy problem f o r (1.1)i s
w e l l posed i n
(-a,
m),
t h u s completing t h e proof of Theorem 2.1.
For t h e s a k e of b r e v i t y i n f u t u r e s t a t e m e n t s , w e i n t r o d u c e t h e A closed, densely defined operator A i s s a i d t o
following n o t a t i o n .
1
belong t o & + ( C 0 , w ) posed i n
t 2 0
i f t h e Cauchy problem f o r (1.1)i s uniformly w e l l
and t h e s o l u t i o n o p e r a t o r
S(c)
satisfies
(1.9).
We
12
F I S T ORDER EQUATIONS
1 1 &+(w) = U {&+(Co,w); C > 13 ( n o t e t h a t Co < 1 i s h p o s s i 01 1 The n o t a t i o n s f o r t h e case b l e ) and &+ = U {Q+(w); -m c w < a].
a l s o write
-m
c t
(Re
e-XtS(t)u d t = R(X;A)u
Then -
contains t h e
an operator valued f u n c t i o n s t r o n g l y con-
S(t)
wt lls(t)lI
W,
n
s
0,1,...)
.
(3.25)
Although t h i s form i s l e s s p r a c t i c a l , s i m i l a r i t y with generation theorems f o r other equations becomes more apparent ( s e e Section 11.2).
13
FIRST ORDER EQ,UATIONS
91.4
Semigroup t h e o r y . Given two Banach spaces
a l l l i n e a r bounded o p e r a t o r s from (uniform o p e r a t o r ) topology.
A
E-valued f u n c t i o n
F
and
E
we denote by
into
E
F
(E;E)
t 2 0
defined i n
t h e space of
endowed with i t s usual
We u s u a l l y a b b r e v i a t e
S(^t)
(E;F)
to
(E).
i s c a l l e d a semigroup
i f (1.7) h o l d s , that i s i f
s(s +
s(0) = I ,
(s, t 2 0)
t ) = S(S)S(t)
(4.1)
*
Equations ( 4 . 1 ) a r e o f t e n c a l l e d t h e e x p o n e n t i a l ( f u n c t i o n a l ) equations. We have s e e n i n t h e preceding s e c t i o n s t h a t t h e s o l u t i o n o p e r a t o r of a w e l l posed Cauchy problem i s a s t r o n g l y continuous semigroup.
As t h e
following r e s u l t shows, t h e converse is as w e l l t r u e . THEORFM 4.1.
t L 0.
S(-)
b e a semigroup s t r o n g l y continuous i n
Then t h e r e e x i s t s a unique c l o s e d , densely d e f i n e d o p e r a t o r
such t h a t
E
1
E+
i s t h e e v o l u t i o n o p e r a t o r of
S(t)
Proof.
A
We d e f i n e t h e i n f i n i t e s i m a l g e n e r a t o r
A
of
S(t)
by t h e
formula 1 AU = lim I;(S(h) kt0+
The domain of
A
a r b i t r a r y and
a > 0
u
c o n s i s t s of a l l
E
E
-
such t h a t
r a S(S)U
=
ds
a Jo
The second e q u a t i o n ( 4 . 1 ) implies t h a t
hence
ua
E
a+ h
-
I)ua =
D(A)
S ( S ) U ds
ua -+ u
as
(4.3) e x i s t s .
For u
-
.
+
(4.4)
h
S(s)u ds),
(with AU& =
But
(4.3)
define ua
g 1( S ( h )
.
I)u
a
( a f o r t i o r i , D(A))
-+
0,
1 g(s(a) - I)U)
.
t h u s t h e s e t of a l l elements of t h e form
is dense i n
E.
(4.5) ua
FIRST ORDER EQUATIONS
14 We prove next t h a t
t 2 0
for h, 0 ,
i s closed.
A
thus if
u
E
D(A)
u
If
then
u
after (4.5). Au -+ v. n
E
Let
E
as w e l l and
D(A)
.
(4.6)
E,
{u n
D(A)
be a sequence i n
un + u,
such t h a t
Then 1 j-(S(h)
1 1 ) u = l i m h(S(h) n+
-
h + O+
Taking limits a s
Let now
u
E
-
we see t h a t
shows t h e closedness of
ht
E we have
S(t)u
AS(t)u = S(t)Au Hence, for any
E
h
1 ) u = l i m (Au,) n n+m
u
E
D(A)
.
h
= v
and t h a t
Au = v,
which
A.
D(A).
Integrating
(4.6) i n
0
ds
S(s)u
S(s)Au ds =
5
s
5t
we o b t a i n
tAut = S ( t ) u
=
-
u
h
using
(4.6) i n t h e l a s t e q u a l i t y .
differentiable i n
t
>0
This shows t h a t
with
,
S ' ( t ) u = S(t)Au = AS(t)u
so that
t >_ 0.
S(^t)u i s a s o l u t i o n of (4.2) i n
This proves ( a ) i n t h e
d e f i n i t i o n of uniformly w e l l posed Cauchy problem. dence p r o p e r t y ( b ) i s checked a s f o l l o w s . for a l l
u,
llS(t)ull
t 2 0.
0
A~(Z)
D(A),
E
( 0
(r)
S(i)
More g e n e r a l l y , we can c o n s t r u c t
lib+,
i n such a way t h a t
where
h >0
i s a r b i t r a r i l y preassigned.
See ZAECZYK
[1975:11;
e a r l i e r (although l e s s elementary) example can be found i n HILLEPHILLIPS
[1957:1, p . 6651 where
EXERCISE
14.
n(A)
3.:?
Prove Lemma
i s i n f a c t empty ( 1 )
so t h a t
u s e the f a c t that
(Hint:
with
(6.18) EXERCISE 15.
Prove t h a t t h e “ r e d ” i n e q u a l i t i e s (3.24) imply t h e i r
complex c o u n t e r p a r t s ( 3 . U ) .
(Hint:
if
Re h >
f~
express
R(A;A)
by
means of i t s Taylor s e r i e s
with
p
real.
EXERCISE
R(A;A)
=
Then l e t
p
16. Let
i n f i n i t e s i m a l generator, Show t h a t i f
u
E
D(A)
be a s t r o n g l y continuous semigroup,
S(t)
t h e niimber defined by
(11
(6.15),
its
A
w’ >w,
0.
then
(6.20) t h e l i m i t being uniform on compact s u b s e t s of
l i m i t is
EXERCISE 17.
Let
E
such t h a t
h(h ; A ) / / 5 u
E
D(A)
For
t = 0
the
be a ( n o t n e c e s s a r i l y densely defined) l i n e a r
A
operator i n a Banach space
Show t h a t i f
t > 0.
(6.3)).
(compare with
$u
C/h
R(X;A) exists for
( h > w)
.
X >
w
(6.21)
then
l i m
A-
+m
hR(h;A)u = u
.
(6.22)
and
FIRST ORDER EQUATIONS
23
FOOTNOTES TO CHAPTER I (1) We d e n o t e h e r e by
not t h e d e r i v a t i v e of
( E ) ) applied t o valued funct i on s .
u.
S'(;)u S(;)
t h e d e r i v a t i v e of t h e f u n c t i o n
S(t^)u
(which may f a i l t o e x i s t i n t h e norm of
The same o b s e r v a t i o n a p p l i e s t o o t h e r o p e r a t o r The c o r r e c t n o t a t i o n
(S(i)u)'
becomes cumbersome
later. ('2)
Roughly speaking,
R( i;A)u.
u(t)
i s t h e i n v e r s e Laplace t r a n s f o r m of
However, t h e c o r r e s p o n d i n g i n t e g r a l may n o t be c o n v e r g e n t
-1
-2
...
t h u s we u s e t h e w e l l known formula R(h;A)u = h u + A Au + + A- mAm -1 u + hmmR(h;A)Amu f o r m = 3, which a l l o w s d i f f e r e n t i a t i o n w i t h respect t o
t.
24
CXIIPTER I1
THL CAUCHY PROELEM FOR SECOND ORDER EQUATIONS COSINE FUIVCTION THEORY
The Cauchy problem f o r second o r d e r e q u a t i o n s .
$11.1
A t h e o r y f o r t h e equation
u"(t)
(1-11
Au(t)
=
t h a t p a r a l l e l s c l o s e l y t h a t of (1.1.1)can be developed without undue difficulty.
We prove t h e fundamental r e s u l t s i n t h i s chapter.
s o l u t i o n of (1.1)in t E-valued f u n c t i o n u(:)
30
A
i s a twice c o n t i n u o u s l y d i f f e r e n t i a b l e
such t h a t
u(t)
E
D(A)
and (1.1)h o l d s t h e r e ;
s o l u t i o n s i n d i f f e r e n t i n t e r v a l s a1.e defined accordingly.
The Cauchy
problem i s now t h a t of f i n d i n g s o l l t i o n s of (1.1)t h a t s a t i s f y t h e i n i t i a l conditions u(0)
=
u0
, u'(0)
=
u1
(1.2)
The Cauchy problem i s w e l l posed o r p r o p e r l y posed i n
2
t
i f and only
0
if (a)
uo,ul
E
D
There e x i s t s a dense subspace there e x i s t s a solution
D
of
u ( 7 ) of
E
such t h a t for any
(1.1)2
t
2
satisfying
0
(1.2). (b)
t
2
0
There e x i s t s a nonnegative, f i n i t e f u n c t i o n
C(ll>
The d e f i n i t i o n s of well posed problem in posed problem i n
[0,m)
and
(-my=)
(-m,m)
(t
2
(1.3)
0)
and uniformly well
a r e obvious analogues of t h e f i r s t
o r d e r case and we omit t h e d e t a i l s . We d e f i n e now two p r o p a g a t o r s o r s o l u t i o n o p e r a t o r s
c(
L
0)
- c .
The checking t h a t
w,O.
(1.1) i s done j u s t a s i n Theorem
t h e same way t h a t
dh
A5
llL(t;u)ll
= O(exp
-u ( ~ , u )
= U,
(w't))
-
u'(O,U)
u(t;u)
is actually a
1.3.1; a l s o , we prove i n
as = 0
t
-
m,
(2.13)
and
L m e - h t L ( t ; u ) d t = hR(h2 ;A)u
for
Re7 >
W'.
It follows from t h i s e q u a l i t y and from Lemma 1.3.2 t h a t
31
SECOND (\WF:Y FCUATIONS
(2.14) hence, using i n e q u a l i t i e s
(2.1),
l l h t ; ~ ) l l5 C o I I ~ I / l i m n-
p-y,1
-(n+l)
wt
wt
= cOllulle
ft 1 0 )
.
(2.15)
m
Condition ( a ) i n the d e f i n i t i o n of uniformly well posed problem f o r (1.1) i s v e r i f i e d a s follows:
if u
u
0’ 1
E
D(A3),
i s given by
(1.2)
u(t)
G(t;u )
=
0
-1
Lt-
@(t)U
(E)
u
E
D(A3)
- valued
We s h a l l show t h a t i f 0
we define
(2.16)
= i(t;u)
E,
ohtaining a
s t r o n g l y continuous function with
-S ( t ) u 2
(b)
a n d extend it by c o n t i n u i t y t o all of
A second operator valued function
t
.
u ( s ; u ~ )d S
To check t h e continuous dependence statement
for
a solution satisfying
u(;)
-8 ( t )
=Lt
i s defined by
.
“@s)u d s
i s an a r b i t r a r y s o l u t i o n of
(1.1) i n
we must have
u(t)
=
E(t)u(O) + Z(t)u’(O).
(2.18)
This i s done a s i n the f i r s t order case and we only sketch t h e d e t a i l s . The f i r s t step i s t o show t h a t and consequently with functions
h(t)
A;
c(t)
= “@t)R(h;A)3 and
=
“(t)R(A;AP
commute with
h ( t ) = i(t)R(A;A)3
continuously d i f f e r e n t i a b l e and s a t i s f y h’(t)
z(t)
and
R(X;A)
t h e second i s t o note t h a t t h e operator valued
h’(t)
= AZ(t)R(A;Af’=A h ( t ) .
=
a r e (twice)
h ( t ) , and the equality
Accordingly
32
SECOND OFTIER EQUATIONS
and
(2.18)
h ( 0 ) = R(A;A)3,
follows noting t h a t
h(0)
=
This com-
0.
p l e t e s t h e proof of Theorem 2.1.
REMARK 2.2. Theorem 2.1 shows i n p a r t i c u l a r t h a t we have t h e r e l a t i o n 2 2 t h e region t o t h e l e f t o(A) 5 ;Reb 5 w } = {A;Reh 5 b) - (Imh)2/4m23, 2 I n p a r t i c u l a r , if o f a p a r a b o l a p a s s i n g through t h e p o i n t s U2, 2 2iw
{w
.
a(A)
o=O,
i s contained i n t h e n e g a t i v e r e a l axis.
A n o t a t i o n s i m i l a r t o t h a t f o r t h e f i r s t o r d e r c a s e w i l l be u s e f u l here.
@(t) s a t i s f i e s
g2(Co,o) w
2
for
The following analogue of Theorem
&
THEOREM 2.3. t h e h a l f plane continuous i n
t
2
km
u
E
2 Q
(Co,w)
wt Cge
@(
=
for
< 0 by Theorem 2 . 1 ) .
1.3.4 holds:
(t
L
:1 1 )
hR(h2 ;A)u
2 R( A ; A )
exists i n
(2.20)
0).
(Reh
> ho).
(2.a)
i s t h e s o l u t i o n o p e r a t o r of
The proof i m i t a t e s t h a t o f Theorem
REMARK 2.4.
8(u)
E
eht@(t)u d t
F
05
and such t h a t
0
Assume t h a t , f o r each
A
if
-
I)U
a
0
it r e s u l t s t h a t
i s closed we n o t e t h a t
s,t;
=
(3.1)
D(A)
i s dense i n
implies t h a t
C(s)C(t)
E. =
T o show
C(t)C(s)
34
SECOND ORDER EQUATIONS 2 2 ( @ ( h ) -I)@(t)ll = @ ( t7 ) ( @ ( h ) -I)U
h
h2 for
h
#
Accordingly,
0.
u
if
D(A)
E
@(t)u
then
D(A)
E
and
A@(t)II = @(t)Au
f o r any
u
E.
E
Hence
after (3.6). Let Aun v. Then
{u }
-
2
(1.7)
n
( @ ( h )- 1)u
be a sequence i n
= l i m
so t h a t t a k i n g l i m i t s as
h
2 ( @ ( h )2
h
n-m
h
+
D(A)
h
I ) u n = l i m (Au ) n n- m
u
we deduce t h a t
0
D(A)
E
-
un
such t h a t
=
h
v
u,
,
and
Au = v
a s claimed.
If u
E
D(A)
we have
r t ( t- s ) @ ( s ) A u d s
=
' 0 = A
rt(t-
s ) @ ( s ) u ds =
Lt t2
( t - s ) A C ( s ) u ds
Au
t
=
C(t)u
- u.
(3.9)
0
Hence
@(G)u i s twice continuously d i f f e r e n t i a b l e i n @"(t)u
so t h a t
=
@(:, = 7 t2n+1hn/(2n + l)!). Show t h a t (6.4) and (6.5) imply
40
SECOND ORDER EQUATIONS
; A
0 = SuprRe A’/‘
(i)
Alternately,
E
.
~ ( A ) I
(6.8)
i s t h e l e a s t p o s i t i v e number such t h a t
(,lo
a(A)
i s con-
t a i n e d i n t h e closed r e g i o n t o t h e l e f t of t h e parabola
5
(d
>0
(with
$/4‘,’2
2
p a s s i n g through t h e p o i n t s with
(? -
=
, -+ S i O i2 .
(0
,
(6.9)
I n p a r t i c u l a r , (6.6)and(6.7)hold
n a t u r a l l y depending on
C
( 6 . 6 ) and (6.7) do not n e c e s s a r i l y hold w i t h 8(;)
5;
0’
A
s(5)
and
s(t^)i s
@(;)
Show t h a t
(1)
holomorphic
@(;)
i n particular,
continuous i n t h e norm of (E)(of course, t h e norm of ( E )
w =
@(i)
can be extended t o f u n c t i o n s
(as (E)-valued f u n c t i o n s ) f o r all
is
Produce a n example t o show t h a t
contained i n t h e negative r e a l a x i s . and
a(A)
if
(11)
is
always continuous i n
l/@(;)ll
due t o formula ( 1 . 5 ) and boundedness of
on
compact s u b s e t s ) . n
EXERCISE 2. topology of
@ ( t )be a c o s i n e f u n c t i o n continuous i n t h e
Let
(E)
t
(continuity a t
suffices).
= 0
A
use formula ( 2 . 1 1 )
n = 1 and c o n t i n u i t y of
for
- I I/
Ilh2R( h2;A)
show t h a t 2 2 h R(h ; A )
of
< 1 for
h
h a s a bounded i n v e r s e ) .
r e p r e s e n t a t i o n s ( 6 . 2 ) and
EXERCISE 3.
Let
-
1x1 -,
as
0
t
at
=
0
to
@(;)
admits the
(6.4).
E = C
m
@(;)
(Hint:
s u f f i c i e n t l y l a r g e so t h a t
Show t h a t
be t h e Banach space of all (complex-
(-m,m)
0 valued) continuous f u n c t i o n s u ( x ) u(x)
Show t h a t t h e
@(o
A
-
D1
moment's c o n s i d e r a t i o n shows i h a t
(1.8)
i s a Eanach space equipped
11. 1l0.
with
THEORFM 1 . 3 .
The space
i s a phase space for
C?
5
x E 1 with = E 0 < t < ffi; moreover,
= E
m
(1.1)
-m
-
-
and
Eo = D1
(1.9)
I11 = D1'
6(
/l[u,vl\\,
\/U!le =
Em we have
5 (co +
IlG(t)l!
I f , say, t h e norm
0 .
Since
C/lAl
Ih- h0 1 < %/Cl
0.
Then
a r c s i n (l/C1)
=
e x i s t s a constant
lIR(h;A)(/
replaced by (cp-).
or
Cp ( r e s p .
c+(cp-)w i t h
a, t h e r e
we w r i t e
T,
5
i s d e f i n e d i n t h e same way w i t h
exists i n the sector
f o r every
Cp,
b e l o n g s t o any s e c t o r
THEORD4 3.1. Assume t h a t
R(A;A)
Given
+ i n d i c a t e s e x c l u s i o n of
subindex
(2.1)
i n a sector containing t h e p o s i t i v e r e a l axis.
We i n t r o d u c e some n o t a t i o n s .
R(h;A)
is the
A'
C
=
C
Z+h' 1 ) .
Cp'
such t h a t
(3.1)
CLpO
]/R(hO;A)l] 5 5 l/llR(hO;A)ll
it f o l l o w s t h a t
and c a n be e x p r e s s e d
there by t h e power series m
R(A;A)
=
T j=0
Since
(pO
- A)~R(A@;A)~+'.
(3.2)
57
PHASE SPACES
with 0
0. I n a
arg p =
p l a n e minus t h e l i n e
R(A;-(-A)1'2)
(3.13) y i e l d s a n a n a l y t i c e x t e n s i o n t o t h e h a l f p l a n e 1 t h u s R(k;-(-A)1/2) exists i n (q' + T I ) ) .
symmetric f a s h i o n ,
> 0,
R e ei"'2p
It remains t o estimate t h e r e s o l v e n t . so t h a t
with
For
c > 0,
p E
$
1 55 (n +
+ T ) ) , Im p
5
0
0
p
w e use (3.13) i n s t e a d o b t a i n i n g t h e
Gf
t h e mere f a c t t h a t
and
-p
p
b e a complex number s u c h t h a t
belong t o
p(
-( -A)1/2)
w e have
= ~ ( ~ ; - ( - A ) " ~ ) R ( - ~ ; - ( - A )l/2)E = ~ ( -2p; A ) E = D(A).
that
p((-A)'"
improvement of a ( p a r t i c u l a r c a s e ) of Theorem
Proof: L e t both
( ( -A)1'2)2
Im p 20
Then
T h i s completes t h e proof of Theorem 3.2.
An i m p o r t a n t consequence fGllOWing
IJ. E
TI).
so t h a t
x+( 2-1 (ql
same e s t i m a t e .
Let
5
p = (p\eiJi w i t h
c+( c+(12(ql + n)),
C
-A,
-p2
D( ( (
f
#
is the
2.5:
E
p(A).
Since =
Since (2.22) i m p l i e s
(3.14) s t a n d s proved.
5111.4 T r a n s l a t i o n of g e n e r a t o r s o f c o s i n e f u n c t i o n s . If A i s t h e i n f i n i t e s i m a l g e n e r a t o r of a s t r o n g l y c o n t i n u o u s
60
PIIASE SPACES
semigroup
then
S(:)
A-bI
any complex number) i s th:
(b
infinitesimal
The S ( G ) = e-btS(f). b corresponding r e s u l t for cosine functions i s somewhat harder to prove.
generator of t h e s t r o n g l y continuous semigroup
LEMMA 4.1.
be t h e i n f i n i t e s i m a l generator of a strongly
A
~ ( ts a t i s f y i n g
continuous cosine f u n c t i o n
Il@(t111 5 coeb It1 and l e t
b
(-m
be a n a r b i t r a r y complex number.
0,
+
a)
[Rehl -h))-(n+m)) (5.25)
69
PHASE SPACES
Proof:
Using
(5.22)
i n formula
(1.3.8) a l l
result instantly.
The corresponding formula f o r
estimates
when
(5.25)
is a group.
S(t)
a r e a consequence of Lemma 1.3.2. =
( - l ) n n ! R(h;A)n
-A
(5.23)
t a k e s care of
The opposite i m p l i c a t i o n s
In f a c t , since
we o b t a i n from formula
inequalities
R(A;A)(~)
=
(1.3.14) ( s e e a l s o
(1.3.15))
that
Ils(t>llIc0 Colimtmn-n(n + 1)
... (n + m > ( l - - wt )-(n+m+l) n 5
t
The corresponding e s t i m a t e i n way.
w t -(n+l) + l i m (1- F )
assumed for
h >
co(l + tm)ewt. (t 2
0)
for groups follows i n t h e same and
(5.25)
need o n l y be
real.
End of proof of Theorem
%(.)
0
(5.23)
We observe i n passing t h a t
=
satisfying
(5.20).
IIR(~;%)~II(~)
5.5.
g e n e r a t e s a group
5.6 we o b t a i n
5 c ( ( I h l - r ~ - b ) -+~n ( l h I (111
%
The o p e r a t o r
Applying Lemma
- w - b)-(”+l))
,...
+ b , n = 0, 1
).
(5.26)
Consider t h e s e r i e s
.. . R( h;\)(R(
R( h ; $ ) ( R ( A;\)b’B+
for
kl,k
*,...
=
t h u s each term of
where
of
k =
(5.26)
O,l,Z
,..., Ihl
(5.27)
>
W
h;%)bp?”
(5.27)
+ 2b. It i s e a s y t o see t h a t
can be w r i t t e n i n t h e form
and p + q = k + n. We make use of t h i s r e l a t i o n and j t o deduce t h a t t h e g e n e r i c term i n t h e series (5.27) i s k
bounded i n norm by a n expression of t h e form
PHASE SPACES
70
Cbk
( Ih( - fIi
1 - b)k+n
+
Cb
k
k + n (/Al-fIl-b)
(k + n ) ( k + n + 1)
+ mk
( [ A ] - 0 1 - b)k+n+2
k+n+l
' (5.30)
*
We observe next t h a t
111 >
for
U
..
+
where it must be remembered t h a t
2b,
k =
..
cki
and t h a t
.,k assume independently all t h e v a l u e s 0,1,. We d i f f e r e n t i a t e n next (5.31) r e p e a t e d l y w i t h r e s p e c t t o Ihl o b t a i n i n g t h e e q u a l i t i e s
kl,
1
+
k
n
n)b ( , h / - u - b ) k+n+l
1 Accordingly,
k
k
+
n)(k
+n +
( / h i - u - 2 b ) n+l
-
l)b
k
- (
( ( ] A / -td-b)k+n+2
the series
(5.27)
n(n
(5.32)
'
+ 1)
I A l - u - 2b)n+2
(5.33) '
i s convergent i n t h e norm of t h e space
( 3 ) and we can e s t i m a t e t h e norm o f t h e sum by ( a constant t i m e s ) t h e
sum of
(5.31), (5.32) and (5.33).
We observe f i n a l l y t h a t
(2.57)
is
nothing b u t
and check (by d i r e c t a p p l i c a t i o n of t h e d e f i n i t i o n ) t h a t R(A;%)
(R(h;\)bp)j
=
R(A;%
(5.34)
+ bB) = R(X;%).
The end r e s u l t i s t h e sequence o f i n e q u a l i t i e s
l ] ~ ( X ; % ) ~ l / ( ~y) C ( l h l - ~ - 2 b ) -+ ~Cn(1Al - u - 2 b ) - ( n + 1 )
+
Cn(n
+ 1)( I A / -
U s i n g (a s l i g h t Il'$(t)ll
5 c(1
+
- 2b)-(n+2)
( / A 1 > w + 2b, n
=
... )
0,1,
(5.35)
71
PHASE SPACES
-00
C
t
C
(5.17).
which completes t h e proof of
m,
W e a t t e n d f i n a l l y t o t h e l a s t statement i n t h e proof o f Theorem
5.5. with u(0)
u(
0,
1< p
0.
Lp( (-T,T);E)
t h a t i s , converges i n
hp(E),
f o r every
Passing if necessary t o a subsequence we can t h e n i n s u r e t h a t
0. that
Applying t h e n r e p e a t e d l y e =
(-co,~),
graph theorem
e
- e 5 e,
t h u s by a
c o n t a i n s an i n t e r v a l
t h e second r e l a t i o n
S b ( t ) E _C D(A;l2)
so t h a t
$12Sb(t)
e
for a l l
(6.20)
t;
(-a,a),
we deduce
by t h e closed
i s a bounded o p e r a t o r .
Consider now t h e group
l$(ll 5 co Here
c(:)
(-02
Proof:
If
b
Im
5
0
(ii)
0
(iii)
0.
2W
a d m i t s a n extension
such t h a t
Lj(t)
(i)
i s an
l+(t)
to
(E)-valued a n a l y t i c
There e x i s t a constant
then t h e operator
%(i)
i s strongly C > 0
satisfies
-Ab
such t h a t
(2.1)
It follows t h e n from Theorem 3.2 t h a t t h e r e e x i s t s (see (4.11)). $ > 0 such t h a t R ( h ; (-%) 1/ 2 ) e x i s t s i n larg A \ < ~i + ~ r / 2 and satisfies
-
Thus Theorem 7.1 a p p l i e s t o show t h a t
-(-%)1/2
generator of a s t r o n g l y continuous semigroup
is the infinitesimal
bb(S)
analytic i n
larg 51 < q, 5 f 0. Since Theorem 3.2 does not provide d i r e c t i n f o r mation on t h e growth of b b ( t ) f o r t r e a l we s h a l l o b t a i n t h i s information by means o f an e x p l i c i t r e p r e s e n t a t i o n for b b ( t ) .
Define
To show t h a t t h e l i m i t e x i s t s we perform a n i n t e g r a t i o n by p a r t s , obtaining t h e equivalent express ion
rm
Tb(t) =
/
'- 0
h(t,h)R(h;-%)2
dh
,
(7.19)
84
PHASE SPACES
there e x i s t s a constant
-< c,h1/2
for t
2
(7.20)
C,
5
(t
+
2
may i n p r i n c i p l e depend on
C'
need only use
(7.20)
t >0
continuous i n
h> 0
in
b > w),
i s c e r t a i n l y the case i f
E
(7.20)
O),
(7.19)
h
at
= 1
(7.U) we o b t a i n t h e estimate
and
ll'b(t)ll where
2
Dividing t h e domain of i n t e g r a t i o n i n
6.
and using
(A
such t h a t
C6
' > '1, %'
(if
6.
so t h a t
(7.22) e x i s t s we
in
C' = 0
(7.22):
xb(t)
W e prove e a s i l y t h a t
i n t h e norm of
(E).
this
is
On t h e o t h e r hand, i f
D ( A ) we have Tb(t)u-u
=
f a s i n thlp(R(A;-%)u-$
l i m
a + ~0
and t h i s expression tends t o zero when
t
+
0.
Although
a c t u a l l y strongly continuous a t t h e o r i g i n ( i . e .
for every
u
E
E)
u ) dh
TJO
xb(t)u
xb(i) +
u
as
is
t
+
r e s u l t w i l l be obtained below.
If we t a k e
u
E
D(A)
then t h e previous
s t e p s show t h a t t h e following computation i s j u s t i f i e d :
A1/2
R(h;-%)u
=
d h = R(p;-(-%)1/2)~.
Since t h e same Laplace transform r e l a t i o n must of needs hold for bb(;)u,
-
where
0
we need not prove t h i s d i r e c t l y , a s a f a r stronger
bb(;)
we have
i s t h e a n a l y t i c semigroup generated by 3,(t)u
=
Irb(t)u
(by uniqueness of Laplace
(7.24)
PHASE SPACES
u
transforms) f o r We extend
u
a f o r t i o r i for
D(A),
E
85
E
l+( t oit) h e upper h a l f plane
E.
z2
by mea.ns of
0
t h e formula
%(C)
=
% ( t+
\(;)
Since
i'I) = % ( t ) b b ( z )
i s s t r o n g l y continuous i n
s t r o n g l y continuous i n upper h a l f plane
2
0,
l+(c)
group and
m
...
and t
S(c)
S(t)E
5 D(Am)
(7.26)
complex a s w e l l ) .
t > 0;
be an a r b i t r a r y element of
Lrb('c)u
E
D(-(-%)'12)
=
D(i(-%)'/*)
i n f i n i t e s i m a l g e n e r a t o r of with r e s p e c t t o
t
E.
t
-
- ( - A b ) 1/2*
T > 0,
Then, i f
is the
Since
=
\(:),
obviously,
W e apply t h i s obser-
v a t i o n (for m = 1) t o t h e a n a l y t i c semigroup generated by
u
for
R(A;A) dh
e
( 7 . 2 6 ) can be extended t o t Let
is
i s a n a n a l y t i c semi-
AmS(t) i s (E)-continuous i n
(so t h a t , i n c i d e n t a l l y ,
bb(q)
0 and
1
AmS(t) =
(7.25)
i s closed, t h e r e s o l v e n t e q u a t i o n
A
i t s i n f i n i t e s i m a l generator t h e n
1,2,
and
m
0).
i s s t r o n g l y continuous i n t h e
and Cauchy's formula t h a t i f A
=
m Em,k
( t ) - m j
as
j
h
m
.
(8.28)
F i s t h e E l b e r t sum of a l l t h e f i n i t e dimensional spaces 2 t h u s i s a r e f l e x i v e Banach space; s i n c e t h e
F. = E J m ( j ,m(j+1)' norm i n 1x1 5 TT
i s dominated by
6
L
times t h e supremum norm i n t h e
same i n t e r v a l , F i s a subspace of t h e space Lo (-,a) of a l l odd, 2lr 2 ~ - p e r i o d i c f u n c t i o n s which a r e square i n t e g r a b l e i n 1x1 5 TT endowed The cosine f u n c t i o n @(;) i s defined by 2 i t s i n f i n i t e s i m a l g e n e r a t o r A i s d /ax2 with maximal domain.
with t h e corresponding norm.
(8.1);
Assume t h a t f o r some for
%(;).
shows t h a t
LL,(;)
i s bounded i n , say, defined b y (8.28)
b
we can f i n d a group decomposition
(8.9)
must obey (8.9) and t h a t i n case Ill.+(t)/lF It1 51 t h e same must be t r u e of ~ ~ ~ o (b,(t) t ) ~ ~ ,
with
b = 0.
( t . ) i s t h e sequence i n
But i f
J
then
.sup
3 21ym( j),m(
thus
(8.U)
An argument very s i m i l a r t o t h a t p e r t a i n i n g t o t h e space
j+l)(tk)
~ ~ b o ( t .+ k )m~ ~by v i r t u e of
ym(k),m(k+l)(tk)J
(8.28),
(8.29)
and a c o n t r a d i c t i o n i s
obtained. The following r e s u l t shows t h a t t h e problem of f i n d i n g a group
94
PHASE SPACES
decomposition of a cosine function becomes m d i c a l l y simpler
if one
i s allowed t o enlarge t h e underlying space.
THEOREM 8.4.
Let c ( ; )
i n t h e Banach space_ E
be a s t r o n g l y continuous cosine function
satisf'ying
wltl Then t h e r e e x i s t s a Banach space /Iu/IE5 (C
(u
E
E)
(8.30)
m).
5 F,
E
F such t h a t
l I ~ l5l ~Cllu/lE
(8.90))
t h e constant i n
0) i s
< m ) )
(b21
@(t^)
1
i n v e r t i b l e and
KV denotes t h e Macdonald f u n c t i o n defined by
where
for
v
#
..
+3,~2,.
and extended lyi c o n t i n u i t y t o a l l values of
(WATSON [1944:1, p. 781).
of a well known i n t e g r a l formula (GRADSTEIN-RLDZYK [l963 :1, p.
EXFRCISE
3.
Let
A,
v
We note t h a t (9.4) i s a vector-valued analogue S(t)
b e as i n Exercise 1.
Given
u
763 ] ). E
E
we say
97
PHASE SPACES
that
S($)u
L
i s c o n t i n u o u s l y d i f f e r e n t i a b l e of o r d e r
and o n l y i f t h e r e e x i s t s
p >
sal/fs(s)(l i n t e g r a b l e i n
s >_ 0
fB(g)
and a f u n c t i o n
w
t
2
0
if
continuous, w i t h
and such t h a t
i m
(t > - 0).
‘-1 f p ( s ) d s
=
e-%(t)u
&
0
(9.6)
h
The f u n c t i o n
of
c1
i s c o n t i n u o u s l y d i f f e r e n t i a b l e of o r d e r
S(c)u
that
i s t h e d e r i v a t i v e of o r d e r
f,(:)
-
u E Ea = D ( ( b 1 ( n o t e t h a t b y Lemma 4.2,
D((b1
a
- A) )
e-@u. c1
Show
i f and o n l y i f
(9.7)
A)‘)
does n o t depend on
b).
The
r e s u l t shows, i n p a r t i c u l a r t h a t t h e d e f i n i t i o n of c o n t i n u o u s d i f f e r e n t i a b i l i t y of o r d e r EXERCISE 4.
2.
Given
E
a &
order
function at
u
t m
E -X
fe($)
does not depend on t h e
~1
[1966:1 3 or
KOMATSU
( t h e author,
[1983:3]).
we say t h a t
@(;)
i f and o n l y i f there e x i s t s
continuous i n
0 and u
0, 6
i s c o n t i n u o u s l y d i f f e r e n t i a b l e of o r d e r
EXERCISE
with
‘= D((b21 - A ) a ) .
For t h e e x c e p t i o n a l v a l u e s of EXERCISE
2 ‘
Show
then u
E x e r c i s e 2.
e-p%(t^)u.
[1983:3]).
A, @(;)
b e as i n
E E
then
‘+by
@(t)u
201
Let
E
b e a space s a t i s f y i n g
98
EVSE SPACES
@(z)
(6.1) w i t h 1 < r < m, A,
i's i n E x e r c i s e 2.
c o n t i n u o u s l y d i f f e r e n t i a b l e of order
2a w i t h
a
Then
>
0
@(t^)u i s i f and o n l y i f
(9.9) h o l d s . A
EXERCISE 7.
0
- 0.
a in
Let
EXERCISE 0
_
t
Assume t h a t 0.
i s c o n t i n u o u s l y d i f f e r e n t i a b l e of o r d e r
S($)u
Show t h a t
on compact s u b s e t s of
b e a s t r o n g l y c o n t i n u o u s semigroup,
c(t^)u
i s Kdlder c o n t i n u o u s w i t h exponent c1
< t < m.
( t h e author,
[1369:2]).
Let
@($) a s t r o n g l y c o n t i n u o u s c o s i n e f u n c t i o n i n 2.
Banach s p a c e ,
t h a t Assumption 5 . 1 h o l d s i f and o n l y i f s t r o n g l y continuous f u n c t i o n i n
i
> - 0,
V(Z)E C - D(A)
=
/ ii
space,
~ ( 2 )a
AV(F)
Show
is a
-
log s ( S ( s + t ) - S(s
t))ds.
(9.10)
0
use t h e e x p r e s s i o n for
EXERCISE 10.
and
where
r l
V(t)
(Hint:
b e an a r b i t r a r y
E
(b21
-
o b t a i n e d i n E x e r c i s e 2).
( t h e a u t h o r [1169:1]).
Let
E
b e a n a r b i t r a r y Banach
strongly cosine function satisfying
Using f o r m u l a (6.18) show t h a t
@ ( s ) u ds
TI
_
E.
v.p.
i n d i c a t e s l i m i t as
Formula ( 9 . B ) i s a n o p e r a t o r
a n a l o g u e o f t h e scalar f o r m u l a
v a l i d for
a
>_
0
(9.Q)
(GWSTEIN-RIDZYK [1963:1, p. 4211)
99
PIUSE SPACES
EXERCISE 11.
Under t h e assumptions i n E x e r c i s e 10, show u s i n g
formula (6.23) t h a t
u
for e v e r y
D(A).
E
Formula
(9.14) i s a n o p e r a t o r a n a l o g u e of t h e
s c a l a r formula
valid for
a
>
0.
(GRADSTEIN-RIDZYK [I963:1,p. 4201).
EXERCISE l2. Using E x e r c i s e s 5 and 8 show t h a t f o r m u l a (?.l2), as < t < M for
w e l l as i t s more g e n e r a l v e r s i o n (6.18) h o l d i n
ucE,y>O. Y EXERCISE
13. Using E x e r c i s e s 5 and 8 show t h a t formula (9.14), as < t < m f o r u c: E (6.23), h o l d i n
w e l l as i t s more g e n e r a l v e r s i o n f o r any
y
> 1/2.
EXERCISE 14. n o t bounded i n
F?(t^)
equals
Y
Show t h a t t h e s i n g u l a r i n t e g r a l o p e r a t o r (8.3) i s
C271(-m,m).
EXERCISE 15. of
-M
Prove Theorem
R(h;U)
1.3 showing t h a t t h e Laplace t r a n s f o r m
and a p p l y i n g Theorem
1.3.4.
FOOTNOTES TO CHAPTER I11
(1) Elements of
Eo x El
and similar p r o d u c t s p a c e s w i l l b e d e n o t e d
as "row v e c t o r s " or "column v e c t o r s " a c c o r d i n g t o convenience. (2)
3
T h i s e s t i m a t e c a n b e c o n s i d e r a b l y improved (see Chapter VI, E x e r c i s e s
t o 8).
(3) (4) (5) (6) (7)
See f o o t n o t e ( 2 ) . See f o o t n o t e ( 2 ) . See f o o t n o t e ( 2 ) . See f o o t n o t e ( 2 ) . See f o o t n o t e ( 2 ) .
100
CHAFTER I V APPLICATIONS TO PARTIAL DIFFERENTIAL EQUATIONS
Wave equations:
5IV.l
t h e D i r i c l l e t boundary condition.
W e consider i n t h e f i r s t s i x s e c t i o n s of t h i s chapter t h e equation U"(t) = A ( @ ) u ( t ) .
(1.1)
Here
m
m
Au =
D j ( a . (x)Dju) Jk
with
x = (xl, . . . , x m ) , D ' = a/ax
j
~ ( x )a r e defined i n a domain 0 A(B)
denotes t h e r e s t r i c t i o n of
condition
@
r
a t t h e boundary
or
bj(x)Dju
-e
+
c(x)u
j =1
j=1 k = l
and t h e c o e f f i c i e n t s a . ( x ) , b . ( x ) , Jk J of m-dimensional Euclidean space Rm A
obtained by means of a boundary
of t h e form
D " u ( ~ ) = y ( ~ ) ~ ( ~( X)
-
E
r),
(1.4)
D" denotes t h e conormal d e r i v a t i v e t o be defined below ( s e e (4.1)). 1 Since t o r e p l a c e a j k ( x ) by ( a j k ( x ) + a k j ( x ) ) does not change t h e
where
a c t i o n of
A
on smooth f u n c t i o n s we s h a l l assume from now on t h a t
a
jk
We r e q u i r e t h e valued. A
If t h e
a
jk a
(x)
= a
kj
(x).
t o be real-valued; t h e
b. J
and
c
can b e complex-
have f i r s t order p a r t i a l d e r i v a t i v e s , we can w r i t e
jk
i n t h e more n a t u r a l form
m
Au =
m
m A
a . (x)DjDku + r b . ( x ) D j u j=1 k = l j=1 J
+
c(x)u
,
(1.5)
101
PARTIAL DIFFEREWIAL EQUATIONS
where
The passage from ( 1 . 2 ) t o (1.5) and v i c e v e r s a i s no longer p o s s i b l e i f
the
a
assume
(1.5) represent
a r e not d i f f e r e n t i a b l e ; i n t h i s case ( 1 . 2 ) and
jk
[1974:1]).
q u i t e d i f f e r e n t e n t i t i e s ( s e e PUCCI-TALENT1
We s h a l l always
i s w r i t t e n i n t h e form ( l . 2 ) , c a l l e d t h e divergence o r
A
v a r i a t i o n a l form.
The c o e f f i c i e n t s
a
b c w i l l be required t o be jk’ j’ merely measurable and bounded; we p o s t u l a t e i n a d d i t i o n t h a t A be uniformly e l l i p t i c i n t h e sense t h a t
(1.6) f o r some
ic
> 0.
Our f i r s t r e s u l t concerns t h e D i r i c h l e t boundary condition (1.3). No assumptions whatsoever w i l l be placed on t h e domain boundary
r.
or the
0
In t h i s high l e v e l of g e n e r a l i t y , it i s obvious t h a t
(as well as t h e boundary condition
u
=
0)
Au
will have t o be understood
i n a s u i t a b l y generalized sense; for instance, i n view of t h e l a c k of smoothness o f t h e
a
jk’
i t i s not c l e a r whether
can be applied t o
A
any nonzero function. The b a s i c space i n our treatment i s supporting r o l e w i l l be played by
a(n)
( o f Schwartz t e s t f b n c t i o n s )
+(O), in
H
2 = L (0).
An important
t h e c l o s u r e of t h e space
$(n);
m(X)
c o n s i s t s of all i n f i n i t e l y d i f f e r e n t i a b l e functions in
0, and t h a t t h e space
a;LI
functions
u
$(n)
(k
an i n t e g e r
and
+(n)
2
L2(n).
with support
1) c o n s i s t s o f
having p a r t i a l d e r i v a t i v e s of order
understood i n t h e sense of d i s t r i b u t i o n s ) i n
8(n)
we r e c a l l t h a t
5k
(derivatives
The spaces
( f o r all necessary f a c t s on t h e Sobolev spaces
ADAMS b975 :1I )
Hk(n), #(n)
consult
.
The f i r s t stage of o u r argument w i l l be t h e c o n s t r u c t i o n of operator
d(0)
a r e f i l b e r t spaces equipped with t h e s c a l a r product
Ao(B),
where
A0
i s the
self a d j o i n t p a r t ol
A,
the
102
PARTIAL DIFFEREmIAL EQUATIONS
With t h i s i n mind, we introduce a new s c a l a r product i n
#,(fi)
by t h e
formula
(u,v),
=
[(a- c ) c v dx
ifi
where
dx
=
... dxm
dxl
(u,~),
(and conjugate l i n e a r i n C > c
constants
h ( y y ajk&Dkv)
dx
(1.9)
and
cY>v It i s obvious t h a t
+
=
e s s . sup c
-
and t h a t
= (V,U)~
u).
.
(1.10) (U,V)~
i s linear i n
v
Moreover, we see e a s i l y t h a t t h e r e e x i s t
such t h a t
0
( w e u s e t h e uniform e l l i p t i c i t y assumption for t h e f i r s t i n e q u a l i t y ) . Accordingly, t h e norm
(1.12) corresponding t o t h e s c a l a r product norm of
4
defined by
(1.7);
(1.9) i s
equivdent t o the original
t h u s we s h a l l assume from now on
&(n)
endowed with (1.12) ( o f cotirse, t h e same arguments apply t o t h e
space
$(",
a f a c t t h a t w i l l be used i n 61V.4).
A function
u
E
$(n)
D(AO(f3)) i f and only i f
belongs t o
(1-13 1
i s continuous i n t h e norm of
L
2
(i)):
i f t h i s i s t h e case, w e extend
( s i n c e €$(a) i s dense L2(Q) 2 i n t h e topology of L (n) t h i s extension i s unique). L e t
t h e l i n e a r f u n c t i o n a l (1.1)) t o a l l of in v
L E
2
(Q)
2
L (0) be such t h a t
(1.14) Define A0(B)" (Motivation i s obvious:
= azI
-v .
i f the coefficients a
(1.15) jk
and t h e boundary
r
103
P A R T I A L DIFFEREIWIAL EQUATIONS
u
are smooth and
u
and
w
=
r,
on
0
=
r).
on
0
-A u = v in n 0 f o l l o w s f o r any smooth w such t h a t
i s a smooth f u n c t i o n such t h a t then
(1.14)
We check e a s i l y t h a t t h e d e f i n i t i o n of A (f3)
above does
0
a.
not depend on
We wish t o show t h a t t h e o p e r a t o r
j u s t defined i s s e l f a d j o i n t .
AO(B)
We b e g i n by proving t h a t
( U - AO(B))D(AO(f3)) h > v.
f o r any
In fact, l e t
(1.16)
= L2P)
be a n a y b i t r a r y element o f
v
L2(Q)>.
Define a l i n e a r f u n c t i o n a l by
w
(1.17) i s
Since
4(0),
L * ( ~ I it i s a s well continuous i n
continuous i n
u
thus there exists
E
$(n)
(1.16)
such t h a t =
( U , d A
hence
(1.17)
*
(V,.>X
-+
(1.18)
(v,w>, u
note t h a t o u r c o n s t r u c t i o n of
follows:
yields the
estimate
Rewriting ( 1 . 1 4 ) i n t h e form
w = u
and t a k i n g
we s e e t h a t
(1.19)we deduce t h a t from
(1.18) t h a t i f
R(h;Ao(p))
u, v
( h- AO(f3)~,v)= so t h a t
Ao(B)
defined.
A1
E
D(A,(B))
(U,V),
i s symmetric.
u
E
4(Q)i n
D(AO(@)).
X-D(A0(B))
To prove t h a t
result.
h > v.
combining with
It a l s o f o l l o w s
then =
=
( u , hv
- A(B)v) i s densely
AO(B)
We f i n a l l y prove t h a t
In o r d e r t o d o t h i s it i s sufficient t o show that
i s dense i n all
i s one-to-one;
exists for
t h e topology of
c a s e , t h e r e would e x i s t an element
to
- AO(B)
w
E
d(". $(."I
I f t h i s were not t h e with
( U , W ) ~=
I n view of (1.20) t h i s i m p l i e s t h a t which, d u e t o
A (B) 0
(1.16),
D(AO(B))
shows t h a t
w
0
for
i s orthogonal
w = 0.
i s s e l f a d j o i n t we make u s e of t h e following
104
PARTIAL DIFFERENTIAL EQUATIONS
Let
LEMMA 1.1. X l b e r t space
number
A.
Proof: -
Assume t h a t t h e resolvent
H.
Then
Let
be a. densely defined symmetric operator i n t h e
A
u,v
be two a r b i t r a r y elements of
( R( A;A)u,v) =
so t h a t
R()\;A)
( ( AI
p(A)
contains a r e a l
i s s e l f adjoint.
A
=
(R( A ; A ) U ,( AI
- A ) R (A;A)u,
R( A;A)v)
H.
Then
- A ) R ( A;A)V) =
(u, R( A, A ) V )
i s symmetric; t h u s
where t h e interchange of i n v e r s e s a n d a d j o i n t s i s e a s i l y j u s t i f i e d (see
RIESZ-SZ.-NAGY [1955:1I ) .
This ends t h e proof.
i s s e l f a d j o i n t and bounded above (by Ao(B) generates a strongly continuous cosine f u n c t i o n A0 ( @ ) @( 0 3
b-,
s ~m =; 0 3 ~ .
E
A l i t t l e use w i l l be made i n t h e following l i n e s of t h e Sobolev spaces
&’P(n)
c o n s i s t i n g of a l l f i n c t i o n s
LP(n);
p a r t i a l derivatives i n
t h e space
Also, we s h a l l employ t h e spaces u in
-
continuous i n
n,
C(l)(F)
having f i r s t
i s normed with
c o n s i s t i n g of a l l f u n c t i o n s
having continuous f i r s t p a r t i a l d e r i v a t i v e s
each d e r i v a t i v e admitting a continuous extension t o
THEOREM
1 f p c
u r LP(fi)
&”(n)
m.
4.1
& n
men
(a)
be a bounded domain of c l a s s
if
Dju
a. and l e t
114
PARTIAL DII’FERENTLAL EQUATIONS
t h e r e e x i s t s a constant
u
f o r everx
then
E
(depending only on 0, p , q ) s u c h t h a t
C
C(’)(F).
if
(b)
(4.3) holds f o r everx
q
2
1.
For t h e proof of a c o n s i d e r a b l y more g e n e r a l r e s u l t see ADAMS We n o t e t h a t Theorem 4.1 h o l d s a s w e l l f o r domains
[1975:1, p. 1141.
which a r e “piecewise of c l a s s c y l i n d e r s whose base i s a
(in
C (’”’
such a s , say, p a r a l l e l e p i p e d o n s or
- 1)- d i m e n s i o n d
also, t h e boundedness h y p o t e s i s i s not e s s e n t i a l : r e s u l t holds i f
THEOREM 4.2. 8
( b u t not
Let fl
n)
f o r instance, t h e
i s bounded.
be a domain o f c l a s s
, 15p
C( 0 )
(or, r a t h e r , t h e s e t of r e s t r i c t i o n s of f u n c t i o n s of
dense i n
exists i n
cy.
i s symmetric s o t h a t , using Lemma 1.1 we show t h a t
i s s e l f a d j o i n t and bounded above by
depending not only on t h e cosine f u n c t i o n
OIV.5
(4.12)
L2W,
a, where
v but also on t h e c o e f f i c i e n t
@,(t)
generated by
Ao(@)
ff
y.
i s a constant Accordingly,
i s t h i s time given by
The phase space.
The arguments i n s I V . 2 have an obvious counterpart h e r e . c o n s t r u c t i o n of t h e square r o o t
B of
Ao(B)
The
proceeds i n t h e same way,
as does t h e proof of THEORFM 5.1 D(E) =
d(n).
(5.1)
The phase space f o r t h e equation
u " ( t ) = Ao(B)u(t) i s now
(5.3)
El
=
$(".
(5.4)
117
PARTIAL DIFFEREWIAL EQUATIONS
Again, t h e phase space The group
Go(;)
( 5 . 3 ) i s t h e same one provided by Theorem 111.1.3.
propagating t h e s o l u t i o n s of ( 5 . 2 ) i s given by (2.11)
with i n f i n i t e s i m a l g e n e r a t o r D(210(f3)) = D ( A O ( f 3 ) )
(2.12), i t s domain being i d e n t i f i e d by
x €$(D).
To t a k e c a r e of t h e f i r s t order terms we
use Theorem 2.3 a p p l i e d t o t h e bounded p e r t u r b a t i o n o p e r a t o r (2.13).
I n t h i s way we o b t a i n ;
Let
THEOREM 5.1.
r,
A
0
be a domain of c l a s s
t h e operator ( l . 2 ) ,
(3
measurable and bounded on
I-.
with domain
D(A(f3))
=
CiLi
t h e boundary c o n d i t i o n
w i t h bounded boundary
( 1 . 4 ) with y
Let -
D(AO(f3)).
d(n)
Then t h e space
X L'(0)
is a
phase space f o r t h e e q u a t i o n
u"(t)
Q1v.6
=
.
A(B)u(t)
(5.6)
The Cauchy problem.
A l l t h e r e s u l t s i n S e c t i o n IV.3 have a n immediate c o u n t e r p a r t h e r e ; we
d e f i n e t h e semigroup B O ( i ) given by ( 3 . 1 ) i n t h e product space 2 2 = L (0) X L ( a ) ; again, depends on t h e p a r t i c u l a r square r o o t
z0(t)
of
Ao(B)
chosen.
B
However, we need
This can be achieved by r e p l a c i n g l a r g e i n t h e d e f i n i t i o n of
Ao(f3);
t o have a bounded i n v e r s e .
c ( x ) by
m Pu = C b . ( x ) D J u + j =1 J TmOREM boundary
y
2
6.1.
r, A
fi
t h e o p e r a t o r (1.2), B
D(A(B))
i s w e l l posed i n
-m
= D(Ao(f3)).
< t
’ ;X
a(x)) < m,
(7.8)
N
a(x), t h e spectrum of
where
E
i s e a s i l y i d e n t i f i e d as
A,
As proved i n E x e r c i s e 11.5, (7.8) i s e q u i v a l e n t t o t h e f a c t t h a t
~(x)
i s contained i n a r e g i o n of t h e form Re h < - w2 - ( I m h ) > / 4 3 . LEMMA 7.2.
-
A
2
i f and o n l y i f
E
p
(a)
(7.9) i s even ,(b)
a
is P -
r e a l with (-l)p’zap (c) j
is r e a l i f
aj
i s odd,
j
>
j
i s even
>
p/z,
,
(7.10)
(d) aj i s imaginary i f
p/2.
Assume t h a t ( a ) , ( b ) , ( c ) and ( d ) hold.
Proof:
P(t) = where
j
p,
Since
a n i n e q u a l i t y of t h e type of
51.
not hold i n t h i s c a s e f o r l a r g e
This ends t h e proof
of Lemma 7.1. We note t h e c u r i o u s consequences o f Lemma
belongs t o
2
,
d
-
A =
+
-
=
(-$I8
6
+(-&)5 (d/dx) 5
does not, i n s p i t e of t h e f a c t t h a t (d/dx)8
although t h e o p e r a t o r
t h e operator A
of
(dx)
7.1:
than
(d/dx)
6
i s a “tamer” p e r t u r b a t i o n
.
I n t h e following s e c t i o n we s h a l l attempt a t h e o r y of t h e equation
(7.l),
b u t only i n t h e c a s e where
t h e D i r i c h l e t boundary c o n d i t i o n . c o e f f i c i e n t s of
O1v.8
A
of o r d e r > p/2
B
i s t h e h i g h e r order v e r s i o n of
Lemma
7.1 i n d i c a t e s
that the
w i l l have t o be s u i t a b l y r e s t r i c t e d .
Higher o r d e r e q u a t i o n s ( c o n t i n u a t i o n )
We study here t h e e q u a t i o n
(7.1) w i t h
an operator
A
of t h e form
121
PARTIAL DIFFERFNTIAL EQUATIONS
c
7
Au =
(-l)Ial-'D"(a+(x)D
Bu ) +
I4 5 k
la1 5 k The c o e f f i c i e n t s
101 T
am, ba
k
a r e r e a l and defined i n a bounded domain
Rm.
in-dimensional Euclidean space
of
W e s h a l l assume t h a t t h e c o e f f i c i e n t s
of t h e p r i n c i p a l p a r t of t h e operator
a
(8.1)
bo/(x)Dau.
A,
OB
c
(-l)ial-lDw(a~D')
,
(8.2)
Ictl=k [BI=k
a r e continuous i n
-
n;
t h e r e s t of t h e
simply measurable and bounded i n r e s t r i c t i o n of
A
R.
a
*'
a s well as t h e
A(B)
The operator
obtained by imposition a t t h e boundary
b,
are
denotes t h e
r
of t h e
D i ric h l et b ound a ry cond it ion
... =
u = D"u =
(Dw)k-l~ = 0
(x
E
r)
(8.3)
(8.3) w i l l be s a t i s f i e d only i n a generalized sense t o be
(although
c l a r i f i e d l a t e r ) . We assume t h a t
and t h a t
A
f o r some
K
i s u n i f o r d y e l l i p t i c , which i n t h i s case means t h a t
> 0.
The following r e s u l t (Ggrding's i n e q u a l i t y ) w i l l be b a s i c .
To s t a t e
it w e introduce t h e Sobolev spaces wk'p(fi) (1 5 p < m ) c o n s i s t i n g of u defined i n fl and having p a r t i a l d e r i v a t i v e s of
all f u n c t i o n s
5k
order
For
p
(understood i n t h e sense of d i s t r i b u t i o n s ) i n
LP(R);
the
w k ~ p ( n >i s
norm of
=
2
wky2(n) =
( t h e only case of i n t e r e s t t o u s ) we s h a l l w r i t e
$(Q).
( i n t h e norm of
The space
Hk(n)).
$(n)
The statement t h a t
v e r s i o n o f t h e boundary conditions THEORail
8.1 L e t
L
i s t h e c l o s u r e of u E %(a)
(8.3).
be a d i f f e r e n t i a l operator;
d)(n)
in
$(n).
i s t h e weak
122
PARTIAL DIFFERENTIAL EQUATIONS
i n a bounded domain
7
c
bI5k
lPl5k
Q
5 Rm.
(-l)lN(-lDTY(aOIT;Dpu)
Assume t h a t a l l t h e c o e f f i c i e n t s
a r e measurable and bounded and t h a t
ICY~ =
=
k.
i s continuous i n
a$
Then t h e r e e x i s t constants
x
14 5 k Is I I
0
when
such t h a t
C,CY
JaaM(x)D?DBu
-
dx
2
k
For a proof see FRIEDMAN [1969:1,p.321. W e proceed t o t h e c o n s t r u c t i o n of a phase space f o r t h e equation
where
A.
i s t h e s e l f a d j o i n t p a r t of A, A~ =
7 Ao(B)
The d e f i n i t i o n of renorm t h e space
where
CY
$(n)
Y
(8.8)
(-i)lml-lDm(aOIT;Dpu).
IBl5k
l+k
follows t h a t f o r t h e second order case.
W e
by means of t h e s c a l a r product
i s t h e constant i n (8.6).
We have
(8.10) The second i n e q u a l i t y follows f r o m t h e boundedness of t h e c o e f f i c i e n t s
of A ; t h e first i s a consequence of Theorem 8.1.
u
E
4(.".)
belongs t o
D(AO(B))
i f and only i f t h e l i n e a r f u n c t i o n a l
w -, ( u , ~ ) , i s continuous i n t h e norm of element of
2
L (a)
L
2
(Q),
A,(@).
being t h e orily
that satisfies
W e show i n t h e same way as i n t h e case a d j o i n t and t h a t
An element
Ao(B)
k = 2
i s bounded above
g e n e r a t e s t h e cosine f u n c t i o n
(by
that o!),
Ao(B) so that
i s self Ao(B)
123
PARTIAL DIFFERENTIAL EQUATIONS
C(t) and a square r o o t
cash t A o ( B ) 1 / 2
=
=
(8.12)
can be defined as i n gIV.2: we have
B = A,(@)'/' D(B)
,
#(n)
=
D((U h >
t h e l a s t i n e q u a l i t y holding for
- A ~ ( B ) 1) /2 ),
(8.13)
Theorem 111.5.4, combined with
cy.
(8.13) i m p l i e s t h a t Q =
i s a state space for (8.7).
(8-14)
H p ) x L2(Q) To show t h a t
Gf
i s as well a s t a t e space
f o r the f u l l equation
we i n c o r p o r a t e t h e lower order terms i n
(8.1) through p e r t u r b a t i o n
(Theorem 2.3) d e f i n i n g
(8.16) and
:]
? = [ : We o b t a i n i n t h i s way: THEOREM 8.2.
Let A
(8.1), @
be t h e operator
the Dirichlet
boundary c o n d i t i o n (8.3), and l e t
(8.18)
A(B) = Ao(B) + P w i t h domain
Then t h e space
D(A(@)) = D(Ao(@)).
a phase space f o r t h e equation
$(Q)
x L2(n)
(8.15).
The t r e a t m e n t of t h e Cauchy problem f o r (8.15) f o l l o w s word by word t h a t f o r second order e q u a t i o n s i n pIV.3; THEOREM
8.3.
L A A
we only s t a t e t h e f i n a l r e s u l t .
be t h e operator (8.1),
boundary c o n d i t i o n ( 8 . 3 ) , and l e t
B the Dirichlet
124
PARTIAL DIFFERFNTIAL EQUATIONS
A ( B ) = Ao(B) with domain
D(A(B))
( 8 -19)
P
Then t h e Cauchy problem f o r t h e
= D(AO(B)).
equation (8.15) i s well posed i n
9IV.g
+-
0, e
=
5 ~ / 5f o r
T )- f ( t ) l l
such t h a t
6 = F(E/?)
[O,aI
i n t e r v a l s of l e n g t h
56
with
f i n i t e number
R
and p i c k r a t i o n a l s tl,
t h e s e i n t e r v a l s ; once t h i s i s done, determine
'G = ' G ( t )
t
be an a r b i t r a r y r e a l number. e
E
t
and a p o i n t
(-t, -t
i n the interval
It +
such t h a t
j
b(t + kn) - f ( t
T -t
and a l l
5
E/5
if
.I
r
of sub-
i n each of
such t h a t
j
=
1,2
,..., r.
Pick a t r a n s l a t i o n number
+ a)
J
. ..,t r
no
+ km)ll 5 ~ / 5 f o r m,n 3 no,
l\f(t. + kn)-f(tj Let
a11 t
\if(t)-f(t')II
a
( t h a t t h i s i s p o s s i t l e follows from Lemma 3.1 ( a ) ) .
6
Finally, cover t h e i n t e r v a l
J
e(E/5;f)
=
(so t h a t
.c 6 .
+ km)ll 5 / k ( t + k n >
0 < t + T
21%/
i n t e r v a l of l e n g h t of
Let
f(t).
Obviously,
h2
=
h2-hl
( 5 + bl)/2
of
1% 1 +
and d e f i n e
f( 2(
not containing any E - t r a n s l a t i o n number
Ih21) h
3
(a2,b2)
not containing any € - t r a n s l a t i o n number
= (a
2
+ b2)/2.
Then
h
3
-Ll and
h3-h;,
n - yh n - h 2 , . . . , h n - h n - l
{hn)
f(t).
which shows t h a t t h e sequence
Then, i f
(f(G
sequence uniformly convergent i n
+
-a
hn))
f(s)
ds
2
0
if
f
2
0
(-m
0
o+, [1958:1, p . 5671),
a s claimed.
This ends t h e proof
147
I N HILBERT SPACE
of Theorem
6.1.
The next r e s u l t i s a n exact c o u n t e r p a r t of Theoyem 2.2 for c o s i n e However, t h e method of proof i s somewhat d i f f e r e n t .
functions.
THEORZM 6.2. @(s
+ t) +
@(s
Let c(
=
e(E,T,u)
0, E
=
5
Ct;o 5 t 5
t h e c h a r a c t e r i s t i c f u n c t i o n of
t
Proof: Set equation
=
s
= u/2
cr
e
E
E
H,
(6 -12)
ll@(t)uII < ~ l l u 1 1 3 .
e.
i n t h e (second) cosine f u n c t i o n a l
0/2
=
@(u) + I
-
we have
Accordingly, i f
so t h a t
hence
(11.1.9). m e r e s u l t i s 2qu/2)2
Hence, i f
T,
(v);
using
(6.11).
+ l), u
1/(2C
m
&
f! e.
5 1/(2C + 1)
we deduce t h a t
It follows t h a t i f
shows t h a t t h e f u n c t i o n s
x (i)
and
u
x
E
f2;)
e
then
2a
#
e,
which
have d i s j o i n t support.
Hence
by
(i)
and
t i o n we o b t a i n
(v).
Taking t h e change-of-variable property i n considera-
(6.13),
thus ending t h e proof of Lemma 6.4.
149
I N HILBERT SPACE
Proof of Theorem 6.2.
m e operator
i s t h i s time defined by
P
ds
E y v i r t u e of Lemma
6.4
with
E =
1/(2C m
t h u s it f o l l o w s from t h e d e f i n i t i o n of
+ 1)
P
.
(6.14)
we have
that
On t h e o t h e r hand, it i s obvious t h a t
Accordirgly, i f inequalities Let now
i s t h e p o s i t i v e , s e l f a d j o i n t square r o o t of P,
Q
(6.8) h o l d . t
be a real number,
u,v
elements of
H.
Using t h e
c o s i n e f u n c t i o n a l e q u a t i o n s and Theorem 5.2 we deduce t h a t ( P @ ( t ) u , v ) = LIM
L T ( @ ( s ) C ( t ) u , @ ( s ) v d) s
T-‘M T L =
1LIM 2 T-m
$lo -T
( @ ( s + t)u,C(s)v) ds
$k
T
+ 1_
*
LIM T-m
+ 1_
LIM T-m
(@(s - t)u,@(s)v) d s
PT
‘
for
u,v
E
H.
$ j o ( @ ( s ) u , @ ( s+ t ) v ) d s
Accordingly,
(6.18) P r e - and p o s t - m u l t i p l y i n g by QC(t)Q-l
Q
-1
= Q-’@(t)*Q
we obta.in =
(Q@(t)Q-’)*.
(6.19)
150
I N HILBERT SPACE
This completes t h e proof of Theorem
6.2.
The coriiments following Theorem 2.2 apply h e r e a s w e l l :
replacing
t h e o r i g i n a l s c a l a r product by ttie ( t o p o l o g i c a l l y e q u i v a l e d c ) s c a l a r product
(2.12)
-
@ ( t )s e l f a d j o i n t .
r e n d e r s each
COROLLARY 6.5.
Assume i n a d d i t i o n t h a t
C(;)
Q
B
u
B
2
t
E.
Then t h e r e e x i s t s a s e l f ad-
tnd a bounded s e l f a d j o i n t o p e r a t o r
0
(6.8)
s a t i s f y i n g i n e q u a l i t i e s of t h e form
@ ( t=) Q - l c o s (tB1/')Q @(;)
Conversely, e v e r y
(-a
E 2 EI f o r some
(6.7).
and such t h a t
u , v )d s
m
(&( ; ) U , V ) and
so t h a t both almost p e r i o d i c f u n c t i o n s
By Theorem
have t h e same Fourier s e r i e s .
(@(;),u,v)
3.8 we must have
& ( s ) = @ ( s ) by a r b i t r a r i n e s s of u,v.
( & ( s ) ~ , v= ) ( @ ( s ) u , v ) , thus
This conclludes t h e p r o o f 3f Theorem 7.1. I n t h e language of a b s t r a c t d i f f e r e n t i a l equations, t h i s r e s u l t can be f o r n u l a t e d a s follows.
Let
THEOREM 7.2. t h e f i l b e r t space
H
A
be a closed, densely defined o p e r a t o r i n
such t h a t t h e Cauchy problem for
u”(t) i s w e l l posed i n
-m
0 ; cf.
holds i n Cgsaro mean,
o r i n Abel mean
u EXERCISE
=
-
m
l i m (1 A) F A ~ M 2u h-. 1k=O -(kp)
.
15. S t a t e and prove a s u i t a b l e converse
(8.22) of Exercise
14.
165
CHAPTER V I
THE PARAEOLIC SINGULAR PERTUREATION PROELEM
Vibrations of a membrane
4VI.l
i n a viscous medium.
r
Consider a uniform membrane fixed t o t h e boundary
of a two
R and immersed i n a viscous medium.
dimensional domain
The small
o s c i l a t i o n s of t h e membrane a r e described by t h e equation pvtt where brane
+ yt
UAV
=
(X
E
a) ,
(1.1)
i s t h e v e r t i c a l movement of t h e mem-
v = v ( x , t ) = v(xl,x2,t) and t h e c o n s t a n t s p,y,a
are, r e s p e c t i v e l y , t h e mass d e n s i t y
p e r u n i t a r e a of t h e membrane, t h e c o e f f i c i e n t of v i s c o s i t y of t h e medium and t h e t e n s i o n of t h e membrane.
v satisfies
The displacement
t h e boundary c o n d i t i o n
o
v ( x , t )=
r) ,
(.
(1.2)
and t h e i n i t i a l c o n d i t i o n s v(x,O) If we d e f i n e
=
v ( x , t ) = u(x, (u/y)t)
and t h e boundary c o n d i t i o n ( 1 . 2 ) ; u(x,0)
(~p)'/~/y
Setting
=
0
u
R)
.
(1.3)
s a t i s f i e s t h e equation
=
-Y 0u
JX)
(x
E
0)
.
(1.5)
w e can w r i t e (1.5) i n t h e form
= F
E U E
then
E
the i n i t i a l conditions are
u ( x ) , ut(x,O)
2
where
(x
u 0 (x),vt(x,O) = u,(x)
tt
+ u
t
= a u
(.
E
n>
9
w i l l be small i f the medium i s h i g h l y viscous
can write t h e second i n i t i a l c o n d i t i o n i n t h e form ut(x,O) = c-1(;)1k1(x)
.
(1.6) ( y >> 1). W e
166
PARABOLIC SINGULAR PERTWATION
This suggests t h e problem of studying t h e behavior of t h e s o l u t i o n of
(1.6) on
as
-
E
allowing f o r dependence of t h e i n i t i a l c o n d i t i o n s
0,
This w i l l be done i n t h e r e s t of t h e c h a p t e r for a n a b s t r a c t
E.
model encompassing e q u a t i o n s l i k e
E x p l i c i t s o l u t i o n of t h e perturbed
Singular perturbation.
sV1.2
(1.6).
equation.
w i l l be t h e i n f i n i t e s i m a l g e n e r a t o r
Throughout t h i s c h a p t e r A
c(:)
of a s t r o n g l y continuous c o s i n e f u n c t i o n space
(that is,
E
A
E
8;
i n t h e complex Banach
see: Exercise 2 ) .
We c o n s i d e r t h e
Cauchy problems
2
E U"(t;E)
+
=
U(0;E)
+
= AU(t;E)
U'(t;E)
f(t;E)
3 o),
(t
(2.1)
u
0
( t ) ,
U'(0,E)
= U1(E),
and
u'(t)
-
=
Au(t)
+
(t
f(t)
-
3 0),
~ ( 0 =) u 0'
-
(2.2)
Roughly speaking, t h e r e s u l t s i n t h e f o l l o w i n g s e c t i o n s e s t a b l i s h that
if
u(t;E)
u(t)
as
E
0
if
uO(E)
i s suitably restricted.
ul(E)
u,
f(t;E)
f(t)
and
R e s u l t s o f t h e same t y p e hold
f o r derivatives.
A s a f i r s t s t e p , we compute t h e e x p l i c i t s o l u t i o n of ( 2 . 1 ) using simple changes of dependent and independent v a r i a b l e . Writing -t/2&2 t/2E U(t;E) = e v(t/E;E) ( o r , equivalently, v(t;E) = e u(Et;E)) we e a s i l y show that
i s a s o l u t i o n of t h e i n i t i a l v d u e
v(t;E)
problem v(t;&) + e
V(O,E)
= UO(E),
Conversely, e v e r y s o l u t i o n u(;;E)
=
V'(0,E)
t/2E
1 U (E) 2E 0
f(&t;&),
(2.3)
+
EU1(E).
v ( t ^ ; ~ ) of ( 2 . 3 ) g i v e s r i s e t o a s o l u t i o n
of t h e i n i t i a l v a l u e proklem ( 2 . 1 ) .
The ( g e n e r a l i z e d )
s o l u t i o n of t h e i n i t i d value problem (2.3) i s g i v e n by V(t;E)
=
C(t;E)V(O;E)
+
s(t;E)v'(O;&)
+,JtS(t
-
s ; E)es/2E f ( E s ; E )
ds,
(2.4)
PARAEOLIC SINGULAR PERTURBATION
c(:p
where
S(t;E)u =
i s t h e cosine f u n c t i o n generated by
167
A + ( 2 ~ ) - ~and 1
W e know from Lemma 111.4.1 t h a t t h e
C(s;E)u d s .
Cauchy prtbpem f o r
However, t h e s e r i e s r e p r e s e n t a t i o n (111.4.7) i s not
i s well posed.
convenient here and we provide a d i f f e r e n t r e p r e s e n t a t i o n below which, A
i n c i d e n t a l l y , proves a f r e s h t h a t LEMMA 2.1.
A
kt
8, t h a t
E
be well posed, and l e t
b
(2E)-21
E
8.
i s , l e t t h e Cauchy problem f o r =
U"(t)
+
Au(t)
(2.6)
be an a r b i t r a r y complex number.
Then
t h e Cauchy problem f o r U"(t)
i s well posed; t h e propagators
=
(A
2
+
b I)u(t)
cb(;),
S b ( t ) of (2.6)
by Cb(t)u =
c(t)U
d t Il(b(t2
+
(2.7)
bt
(t2
-
-
('1
a r e given
s~)'/~) C(s)u d s , ( 2 . 8 )
s2)1/2
and
where
Io,
5
a r e t h e Bessel f u n c t i o n s defined by (2.10)
The proof can be c a r r i e d out i n ( a t l e a s t ) t h r e e ways, all of which we sketch b r i e f l y below.
In t h e f i r s t , one t a k e s advantage of
t h e d i f f e r e n t i a l equation
(2.11) s a t i s f i e d by if
u, v
then
E
u(f)
Iv(x)
D(A)
and of a l i t t l e i n t e g r a t i o n by p a r t s t o show t h a t
(so that
= cb(;)u
c(t*)u, S(t*)u a r e genuine s o l u t i o n s of ( 2 . 6 ) ) i s a genuine s o l u t i o n of (2.7): uniqueness
+ %(t*)u
of s o l u t i o n s of ( 2 . 7 ) i s proved transforming them back i n t o s o l u t i o n s of
(2.6) by means of formulas (2.8) and (2.9) with b replaced by i b .
168
PARAEOLIC SINGULAR PERTUREATION
For d e t a i l s on t h i s l i n e of approzch (&though t h e context t h e r e i s somewhat d i f f e r e n t ) see SOVA
[1970:4].
The second uses Theorem 11.2.3; w e
show t h a t t h e Laplace transform of C b ( c ) u given by ( 2 . 7 ) equdls 2 2 2 The t h i r d i s based on Lemma 111.4.1, h R ( A ; A + b I ) u = AR(A2 - b ;A)u.
en i n (111.4.7) combined
p l i c i t c o n s t r u c t i o n of each of t h e terms s e r i e s r e p r e s e n t a t i o n (2.10) f o r
I o ( x ) , I1(x).
ex-
with the
We omit t h e d e t a i l s .
(2.8), (2.9) t o t h e equation
Applying t h e t r a n s l a t i o n formulas (2.5) we o b t a i n C(tjE)U
+
= C(t)U
and
=L t
S(t;E)u
-
l o ( ( t 2 s')'/~/~E)C(S)U d s
both v a l i d f o r a r b i t r a r y
u
and
t.
c ( s ) u ds
(2.12)
,
(2.13)
Using t h e s e formulas i n ( 2 . 4 ) and
making a n obvious change of v a r i a b l e i n t h e i t e r a t e d i n t e g r a l r e s u l t i n g from t h e i n t e g r a l i n ( 2 . 4 ) we o b t a i n t h e f i n a l formula for u ( c ; & ) , we have used u(t;&) = e
+
(11.4.3) f o r t h e s o l l i t i o n of t h e nonhomogeneous equation:
-t/2E
2 ( t /& )uo ( E 1
-
t e - t / 2 E 2 ~ t / &T ( ( ( t / E ) 2 E2
((t/E)*
2 = e-t/2E
where
C(t/&)u,,(E)
-
s)ll2/2E)
c(s)(, u O ( E ) ) d s S2)1/2
1
+
cp(t,s;&)C(s)(~ uO(E)) d s ' 0
2 = e-t/2E
C(t/E)Uo(&)
+
R(t;E)($
U,(E))
+
qt;E)($
U O(E)
+
E2U,(&))
(2.14)
169
PARABOLIC SINGULAR PERTURBATION
where t h e d e f i n i t i o n s o f t h e s c a l a r f u n c t i o n s
Cp(t,s;&),
qt;&)a r e
R(t;&),
t h e o p e r a t o r valued f u n c t i o n s t h e formula (3).
$(t,s;E)
and
e a s i l y read off
The f o l l o w i n g r e s u l t i s a n immediate consequence of formula (2.14) and t h e correspondence between s o l u t i o n s of ( 2 . 1 ) and t h o s e of (2.3):
LEMMA 2.2.
Every g e n e r a l i z e d s o l u t i o n
t 10.
continuous i n
f u
0
then
(E) = O
of
u(t;E)
(1.1)
u ( t ; & ) i s continuously
differentiable. precisely, i t s
We examine now t h e i n i t i a l value problem (2.2),
r e l a t i o n with t h e second order i n i t i a l value problem f o r (2.6).
Let A
THEOREM 2.3.
(2.6) be w e l l posed.
8,
E
t h a t is, l e t t h e Cauchy problem f o r
Then t h e i n i t i a l value problem f o r
u ' ( t ) = Au(t)
i s w e l l posed i n t
2
(2.15)
t h e propagator of (2.15)
0,
g i v e n by t h e
a b s t r a c t W e i e r s t r a s s formula
S(t)u = The propagator in -
if
5 >
-
Re A
0;
m
2
e-'
/4tC(s)u
more p r e c i s e l y ,
then
A
E
G(rp)
(t > 0 ) .
ds
c a n be extended t o a ( E )
S(?(E)
-
In
h-(E) =
(1 + 4 u ~ ~ & ~ )>~1/ /~ ~/ ~ &W .e~ o b t a i n i n t h i s way t h e following r e s u l t :
where
Co,w
a r e t h e c o n s t a n t s i n (2.17).
To study convergence of
u(t;E)
we s h a l l use t h e asymptotic
s e r i e s f o r t h e Bessel f u n c t i o n s ,
where
( s e e WATSON [1948:1, pp. 203 and 1981). The asymptotic s e r i e s ( 3 . 5 ) w i l l be used with
x =
2E
((:)*
-
both t o c a l c u l a t e limits (as i n (3.10)-(3.11),
(3.25)-(3.26),
(4.14)-
(4.15)) and f o r e s t i m a t i o n purposes (as i n (3.14)-(3.15), (3.27)-(3.28), (4.16)). The second a p p l i c a t i o n deserves some comment. Since t h e functions x = 0 , (3.5) w i l l not provide good bounds
t o be estimated are r e g u l a r a t
near zero. However, t h i s i s not very s i g n i f i c a n t , s i n c e t h e s e functions converge t o limits t h a t do have s i n g u l a r i t i e s a t
x = 0. To improve t h e
estimations, we s h a l l r e p l a c e remainders of t h e form
O(x%)
by
o ( ( x + a)%) which a r e r e g u l a r a t t h e o r i g i n . It i s p l a i n t h a t t h e observations above can be a p p l i e d a s w e l l t o t h e asymptotic s e r i e s obtained from t h e s e r i e s (3.5) f o r functions of t h e form
173
PARAEOLIC SINGULAR PERTUREATION
x-'Iu(x),
where
B
and
a r e a r b i t r a r y r e a l numbers. Likewise, we s h a l l
u
( 3 . 5 ) term by term t o provide asymptotic s e r i e s for d e r i v a -
differentiate
t i v e s of a r b i t r a r y o r d e r . This can be easily j u s t i f i e d . To t h i s end we We proceed t o t h e e s t i m a t i o n of u ( t ; & ) - u ( t ) . d i v i d e t h e i n t e r v a l of i n t e g r a t i o n i n ( 2 . 1 4 ) i n
0
5
5
s
s(E),
where t h e asymptotic development ( 2 . 5 ) c a n be u s e d , and t h e "small" interval
s(&)
5 t 5 t / & where
rougher bounds w i l l s u f f i c e .
The
d i v i s o r y p o i n t between t h e " i n n e r " and " o u t e r " i n t e g r a l s i s defined by
where
q
< 1/2 w i l l be s p e c i f i e d l a t e r .
The l e n g t h of t h e second
interval is
We use t h e asymptotic development ( 3 . 5 ) t o o r d e r
0
5
s
5
S(E):
* X
IJX)
=
(1 +
o($))
m
=
in
0
.
(3.9)
(2lrx)l
A f t e r d i v i d i n g by
x
and performing a few c a n c e l l a t i o n s we o b t a i n
where =
X(t,S;E)
To o b t a i n (3.11) from
Obviously,
(1.10)
-
- q ( F )2 ) -3/4(1 + o($)).
(3.9) we
T(x),
-1
m
=
0.
e x i s t s a constant
have used t h e i n e q u a l i t y
(3.11) i s uniform i n t h e i n t e r v a l
To o b t a i n a n e s t i m a t e f o r
x
Since C
(3.11)
(lrt)
cp
we use formula
xmlI (x) i s r e g u l a r f o r 1
such t h a t
0
5
s
5
s(E).
(3.5) f o r t h e f u n c t i o n x
=
0,
there
PARABOLIC SINGULAR PERTURBATION
174
where
x; t h e choice of 2 i n t h e denominator of increasing, since ( 2 + x) -3/2 ex
does n o t depend on
C
(3.13) makes t h e f b n c t i o n (a
+
x)"
ex
is increasing f o r
l a t i o n s leading t o (3.10)-(3.11)
holds i n
0
5 s 5 t,
and t h e c o n s t a n t
C
a>
CY. Essentially
t h e same manipu-
reveal that t h e e s t i m a t e
where
does not depend on
E,t.
I n e q u a l i t y (3.14)
l e a d s t o t h e estimates below.
holds, where
i s a constant t h a t does not depend on
C
s,t,E.
We make use of (3.14)-(3.15) observing t h a t 2 1/2 2 (Es/t) ) 5 1 ( E s / t ) /2 i n t h e exponent and t h a t (Es/t) 2 ) 1/2 >_ (1 (Es(E)/t)')1/2 = 2~ i n t h e denominator of
Proof.
(1 (1
-
-
(2.14);
t h e term
-
2 4E /t
i n s i d e t h e p a r e n t h e s i s i s p o s i t i v e and can
be dropped.
where t h e c o n s t a n t
C
does not depend of
Proof. We use a g a i n (3.14)-(:,.15)
hand s i d e of t h e i n e q u a l i t y
s,t,E.
keeping i n mind t h a t t h e r i g h t
(3.13) is a n increasing f i n c t i o n of
x.
Accordingly, we can e s t i m a t e t h e r i g h t hand s i d e of (3.14) by t h e value obtained i n s e r t i n g t h e highest p o s s i b l e value of (which i s t h e summand
(1
-
2
(Es(E)/t)2)1/2
4E /t
= 2q).
(1
-
(Es/t
2 1/2
)
Once t h i s is done we d i s c a r d
i n t h e o u t e r p a r e n t h e s i s o f (3.15).
The r e s u l t
i s (3.17). A s a n immediate consequence of (3.17) and of the estimation (3.8)
PARABOLIC SINGULAR PERTURBATION f o r the length of the interval
s(E)
t h e r e and i n o t h e r i n e q u a l i t i e s
C
s ,t ,&
0 . We s a y t h a t a family of f'unctions converges uniformly i n t > - t ( E ) t o a f u n c t i o n g(:) if and t(E)
> 0
f o r each
E
only if sup
1 h
-
llg(t;E)
g(t)lI = 0 .
Ed0 t)t(E)
I f t h e supremum i s t a k e n i n s a y that
t
2
t ( E ) 0
for
a r b i t r a r y we
uniformly on compacts of
g(:)
t(E). We prove below that f o r every
on compacts of
t ,t(E)
s u b s e t s of
as long as
E,
u
E
E, R(t;E)u-r S(t)u
uniformly w i t h r e s p e c t t o
t(E)/E2
4
(E
m
.+ 0 )
uniformly
u on bounded
.
(3.20)
I n f a c t , assume t h i s i s f a l s e . Then t h e r e e x i s t s a bounded sequence
[u,]
sequence
{t,]
For each
n
C
E,
a sequence 2
such that
we choose
tn/En
'n
-21-
{En]
*
with and
m
End
0
and a bounded
lIR(tn;En)un
- S(t,)unll
2 6 > 0.
such that 'in
-
WE
n
n --
(3.21)
n (note t h a t zero,
'n
-
n 1/2
< 1/2: moreover, s i n c e both as
n
+ m).
En
and
E n t n-1/2
tend t o
We d i v i d e t h e i n t e r v a l of i n t e g r a t i o n i n
(2.14) according t o t h e e q u a l i t y (3.7) with
q = q n'
We have
176
PARABOLIC SINGUMR PERTURBATION
The f i r s t i n t e g r a l tends t o zero as convergence theorem: that
(note that
1
-
(Es/t)2
uniform estimate
due t o the dominated
m
+
9
-
cp(t,,s;E,)
n
i n f a c t , t e asymptotic r e l a t i o n (3.10) shows e-'
/4tn-+ 0
2 211, -,
1. hence
(3.16).
as
-
n-,
Es/t
m
0)
for
s
fixed
and we have t h e
The second i n t e g r a l tends t o zero by (3.18).
A s f o r t h e t h i r d it i s e a s i l y seen t o telescope making the change of variable
ti1/'s
=
(J
tn a r e bounded. I n f a c t ,
and r e c a l l i n g t h a t t h e
Now, it follows from (3.8) and (3.20) that S(E
n
)
=
tn 2 1/2 (1 - 4q ) n n
t
2
s(En) >_ 2 t 1/4&-1/2 n
Thus
8n
-
m
(1
-
,3/4
2'in)l/2 2 2
as
n
-
m
n -
-
(3.24) El/2 n s o t h a t (3.23) tends
t o zero (we note t h a t i f
w = 0 t h e i n t e g r a l (3.23) tends t o zero 2 under the s o l e assumption that t,,/En -,m , where t h e tn may be
unbounded; t h i s f a c t bears on a resuLt below). a contradiction and j u s t i f i e d
OUT
claim about
We have then obtained sf.
We prove next t h e corresponding statement f o r
5(t;E).
The
estimates a r e obtained i n a s i m i l a r fashion, thus we only s t a t e the final results.
Formula (3,10)-(3.11) has t h e following counterpart:
with X(t,S;E) = (77t)-1/2(1 The estimate is uniform i n The inequality
holds i n
0 5 s 5 t, where
-
2 (y))-1/4(l +
0 5 s 5 s(E).
.
(I(%))(3.26)
PARABOLIC SINGULAR PERTURBATION
177
p(t,S;E) = t and t h e constant
(3.28)
does not depend on
C
we use t h e asymptotic formula (3.5) f o r
m = 0
E,t.
To o b t a i n (3.25)-(3.26)
m = 1; t h e same formula with
yields the inequality
Using t h e i n e q u a l i t y (3.26)-(3.27) we e a s i l y o b t a i n t h e following counterparts of L e m 3.2 and Lemma 3.3 :
holds, where the constant
where t h e constant Using
C
does not depend on
does not depend on
C
s,t,E.
s,t,E.
(3.31) and (3.8) we obtain
We prove t h a t
6(t;E)u uniformly i n of
t >_
S( t ) u
uniformly with respect t o
t(E)
i n e x a c t l y t h e same way used f o r
E
(3.33)
R;
u
i n bounded s e t s
d e t a i l s a r e omitted.
A f t e r a n elementary estimation of t h e f i r s t term i n (2.5) t h e proof of t h e following r e s u l t i s complete:
Let
THEOREM 3.6. UO(E)
and l e t
u(t;E)
-b
v,
uO(E),ul(E) 2
E UJE)
E
E
-. uo
be such that
- v
(E
+
0)
be the generalized s o l u t i o n of ( 3 . l ) ,
number such t h a t (3.20) holds.
,
(3.34) t(E) > 0
Then
U(itjE) -. u(Z)
(3.35)
178
PARABOLIC S INGUMR PERTURBATION
unifwmly i n compacts of s o l u t i o n of ( 3 . 2 ) w i t h respect t o
uo,v
REMARK 3.7.
if.
2
t
t(E)
u(%)
where
u(0) = uo.
i s t h e generalized
The convergence i s uniform with
(Iuo(I, /(v(I a r e bounded.
does not converge t o
uO(E)
thus t h e r e i s a "boundary l a y e r " near zero where approximation t o
t >_ 0 u
Obviously, uniform convergence i n
expected s i n c e i n g e n e r a l
u(^t)
cannot be
as
0
-
E
0,
i s not a good
u(;;&)
[1981:1] f o r a thorough
( s e e KEVORKIAN-COLE
treatment of t h e one dimensional case). Note a l s o t h a t m i f o r m -opt -w2t e u(t;E) t o e u ( t ) i n t >_ t ( E ) cannot be
convergence of
assured even i n t h e s c a l a r case.
To s e e t h i s , l e t
s o l u t i o n of t h e i n i t i a l value problem
-w
< w2
since ?(&),A*(&) e-',2tew2t = 1,
we have
e
L3t -2 1 with
0 l e t t ( E ) be such that 2 -, a , where 0 < a < m . Then we seE from (3.3) that
follows. 0 , 0 c E c 1/&)
independent of
u)
> 0
and of
5
116"(t;E)(E-1R(&-1;A))ll
C(w4
f
2 w 2 / t + l/t 2 ) eu) t
The proof i s straightforward b u t tedious. that
U E
D(A),
B"(t;E)u
s o that
.
(4.26)
Assume f o r t h e moment
e x i s t s ; a n e x p l i c i t formula f o r it
can be obtained from (4.1): -t/2&2 G,"(t;E)u =
E3
-t/2E2 C'(t/E)u
-
4
-t/2E C(t/E)U
2E
+
te
2 C(t/E)U
8E6
C(s)u ds
te
2 -t/2E f
t e 4E7
C(s)u ds 2
-t/E
Jo
Ii(((t/E)2
-
2 1/2
s )
/2&) C(s)u ds
Il(((t/E)2
-
S2)l/'/2E) C(s)u ds
PARABOLIC SINGULAR PERTURBATION
187
e 4E5
where terms a r e grouped t o g e t h e r as t h e y appear i n d i f f e r e n t i a t i n g
(4.1).
Note a l s o that t h e t h i r d and f o u r t h i n t e g r a l s a r e i n d i v i d u a l l y
divergent and must be combined i n t o one.
We t a k e a look f i r s t a t t h e
terms that l a y o u t s i d e of i n t e g r a l s . For t h e second we have
and t h e same estimate o b t a i n s for t h e t h i r d and t h e f o u r t h ,
SO
that
t h e y s a t i s f y (4.26) even without t h e i n t e r c e s s i o n of t h e mollifying operator then
E-1R(E-1;A).(4)
For t h e f i r s t term we note that i f
C(^t)v i s continuously d i f f e r e n t i a b l e w i t h
hence
C' ( t ) v = d(t)Av
Since
w
7
v
E
D(A)
C " ( t ) v = C(t)Av,
and we have
0, l/S(t)ii 5 C e x p ( ~ t ) ( ~ and ) t h e r i g h t hand s i d e of (4.29)
can be estimated i n t h e same way as (4.28). To e s t i m a t e t h e six i n t e g r a l s i n (4.27) we d i v i d e t h e domain of integration a t specified l a t e r .
s = s(E)
given by
(3.7), with q < 1/2 t o be
For t h e f i r s t o u t e r i n t e g r a l we t a k e advantage of
(3.17) f o r cp(t,s;E), divided by t E 2 ; f o r t h e i n t e r v a l of i n t e g r a t i o n we use (3.8). The r e s u l t is a bound of t h e form t h e estimate
The s e c o n d , f i f i h and s i x t h i n t e g r a l s a r e t r e a t e d i n t h e sane way: i n a l l c a s e s , due t o t h e a d d i t i o n a l f a c t o r e s t i m a t e of t h e form
t/E2
we end up with a n
188
PARABOLIC SINGULAR PERTURBATION
A s pointed out a f t e r (4.27) the t h i r d and f o u r t h i n t e g r a l s must be
combined i n t o one t o a m i d divergence a t
s = t/E
w r i t t e n s e p a r a t e l y only f o r typographical reasons).
( i n f a c t , they a r e The basis of the
r e s u l t i n g estimation w i l l be t h e asymptotic s e r i e s f o r the h n c t i o n obtained from (3.6): Q(x) = X-~(X-~I~(X))'
we deduce from it that
The combined integrand of the f o u r t h and f i f t h i n t e g r a l (including f a c t o r s outside of the i n t e g r a l ) i s
-
2 -t/2&2 t e Q ( ((t/E)2
16E~
-
~~)'/~/2E)C(s)u.
I n view of (3.30) we have
where p(t,S;E) = t
(3
(4.34)
is increasing w e can bound the r i g h t hand side of (4.34) by i t s value a t s = s ( E ) subsequently deleting the f a c t o r 6$/t from t h e outer parenthesis. The r e s u l t i s an upper bound f o r
Since
-t
x)-5/2ex
t h e combined integrand of the form
Therefore, the i n t e g r a l can be bounded by the following expression:
This completes the consideration of the outer i n t e g r a l s .
We look a t t h e inner i n t e g r a l s . t h e i n t e g r a l belaw:
We begin by grouping them i n t o
PARABOLIC SINGUL4R PERTURBATION
189
O(t,s;E)C(s)u ds.
Using t h e asymptotic developments (3.5) f o r
Io, I1 and Ii of
m = 1 i n t h e f i r s t and f o u r t h i n t e g r a l s and of order
order
i n t h e r e s t we o b t a i n f o r
B
m = 2
a n expression of t h e form
a l i n e a r combination of terms of t h e form
with X
with
(4.36)
j = 2,1,0,C
f o r each t e r m
expression f o r
J
>
We then use T a y l o r ' s formula of order 2 ) 'j, ending up w i t h t h e following
U . , ~ .0.
J
(1 - ( E s / t ) X:
o(&)) 2
where each
i s independent of
X,(t,s)
t > 0
cosine f u n c t i o n
and apply formula C(^s) = cos og,
(in fact,
E
(4.27) i n
where
t h e space
(4.39)
is a f i n i t e
X,
>_ 0 ) . We
s@t-* w i t h ct:B
l i n e a r c o m b i d t i o n of terms of t h e form then f i x
2j
E = C
t o the
i s a r e a l parameter.
0
Naturally, t h e r e s u l t must be t h e second d e r i v a t i v e of t h e s o l u t i o n of E 2t " ( t j E )
4-
< ' ( t ; E ) = w 2< ( t j E )
,
(4.40)
w i t h i n i t i a l conditions 0
t6'(t;E)
be such t h a t (3.20) holds.
-
tS'(t),
t*E'l(t;E) (E-$(E-l;A>)
uniformly on compacts o f
t >_
t(E)
4
Then (4.52)
t2S"( t )
,
(4.53)
i n t h e topology o f
(E).
Obviously, r e s u l t s of t h e type o f Lemma 4.4 and Theorem 4.6 can be obtained f o r d e r i v a t i v e s of any o r d e r o f
6.
We omit t h e d e t a i l s .
I n a s e n s e , t h e r e s u l t s above do not t e l l t h e whole s t o r y about
5" s i n c e t h e smoothing o p e r a t o r
E-lR(E-l;A)
only plays a r o l e i n
t h e f i r s t term on t h e r i g h t hand s i d e of (4.27); b e s i d e s , t h e f i r s t
t e r m a l s o makes it necessary to s e p a r a t e t h e c a s e t i o n purposes.
is :
A s t a t e m e n t on
@'(t;E)
u)
= 0
f o r estima-
t h a t avoid t h e s e inconvenients
192
PARABOLIC SINGULAR PERTURBATION
THEOREM 4.7 w
2
and of
0
IlG"(t;E)U
(a)
-
E
.There e x i s t s a constant
( t > 0, 0
0,
*
m(A) = ( E A Then m' (A)
(5.24)
+
A)
q2
(A,O)
is alternating.
The proof i s l e f t t o t h e reader (Exercise 1).
.
(5.25)
196
PARABOLIC SINGULAR PERTURBATION LEMMA 5.4.
t
+
Aza.
Let
f ( % ) be continuous i n
Assume t h e Laplace transform
m.
t >_ 0 , f ( t )
=
O(exp a t )
Pf(^A) i s a l t e r n a t i n g i n
Then f ( t ) >_ 0
( t >_ 0 ) .
(5.26)
The proof is an immediate consequence of Lemma 1.3.2 ( s e e (1.3.14)). End of proof of Theorem 5.1. We go back t o (5.18). The d e f i n i t i o n (5.15) of t h e flmction h ( t , s , A ; t ) , Lemma 5.3, Lemma 5.2 and the is comments preceding it show t h a t h, a s a function of A, a l t e r n a t i n g f o r any s , t 2 0 , E > 0. Since t h e f i n c t i o n k ( s ) defined i n (5.19) i s nonnegative, it follows from (5.18) t h a t the Laplace i s a l t e r n a t i n g . Thus, by Lemma 5.4, transform of r(ht;E) r ( t ; E ) >_ 0 ( t 2 0 , E > 0 ) . Taking i n t o account t h e a r b i t r a r i n e s s of
*
u and u
,
(5.7) follows, completing t h e proof of Theorem 5.1.
I n a l l of t h e r e s u l t s t h a t follow u(:;&)
u(z))
(resp.
is the
s o l u t i o n of t h e homogeneous i n i t i a l value problem (2.1) (resp. ( 2 . 2 ) ) .
and applying (5.6) and (5.7) to t h e f i r s t term on t h e r i g h t hand s i d e t o estimate t h e other summands we use (3.4) which implies
of (5.28): (taking u
0
(E)
= 0
or
Ilc(t;E)ll _ 0,
E
’0)
*
(5.29)
W e obtain a simpler but l e s s p r e c i s e bound noting t h a t @,(t;&), u?t (Lemma 3.1) and i n t e g r a t i n g ( 5 . 7 ) by p a r t s ; it r e s u l t s YW(t,E) 5 e t h a t Ow(t;E) 5 (1+ w2t)eat so t h a t (5.27) becomes 2 lju(t;E) u(t>li 5 c O2 (1 ~ + w 2t ) e w2t / l ~ u +o ~coew ~ t ~ l u o ( ~ )~ , I I +
-
-
197
PARABOLIC SINGULAR PERTURBATION 2 wLt e I/ul(~)I(
( t 2 0,
+ c0E
Theorem 5.5 implies t h a t when
t
D(A)
E
0
-
/lU(t;E)
uniformly on compacts of
u
u(t)ll
E
> 0)
.
(5.30)
we have 2
= O(E
1
(5.31)
if
0
I1uO(~)- uoIl = O(E
2
and
llu,(~>Il =
o(1).
(5.32)
Estimates of t h e same s o r t can be e a s i l y obtained f o r t h e d e r i v a t i v e u'(t;E)
if
u
0
E
2
D(A )
and
uO(E)
In f a c t ,
D(A).
E
v(%;E) = u ' ( % j E )
i s t h e s o l u t i o n of t h e i n i t i a l value problem (2.1) w i t h v(O;E)=U'(O;E)=U~(E),
~ ( 2 =)
On t h e o t h e r hand,
= u"(0;E) = E
v'(O,E)
-2
(AuO(E)
-
ul(E)).
(5.33)
i s t h e s o l u t i o n of (2.2) w i t h
u'(t)
(5.34)
~ ( 0= ) ~ ' ( 0 =) AuO. Accordingly, we have
THEOREM 5.6.
Assume t h a t
u
2
0
E
and
D(A )
uO(E) E
2 2 l ( u ' ( t ; E ) - u ' ( t ) l l 5 COE @ w ( t ; E ) l ( A uolI + CO+w(t;E)lIU1(E)
-
+ COYW(t;E)/lU1(E)
2
D(A).
-
Then
AuolI
AU,(.)ll
2
2 w t 2 (1 + w t ) e IIA uolI
5
COE
+
c0 ew t(I/U1(E)
2
- AUoll
.
( t 2 0) It follows from t h i s r e s u l t that i f
- Auo(E)/I)
IlU,(E)
+
uo
(5.35) 2
D(A )
E
uO(E) E D(A)
and
then llu'(t;E) uniformly on compacts of Ilu,(E)
-
t 2 0 2
- u'(t)l/ =
2 O(E
(5.36)
)
if
- Au~(E)II =
2
)
and
IIu~(E)
= Of& )
and
/ I A u O ( ~ ) AuoII = O ( E ).
AuoI/ =
O(E
O(E
> , (5.37)
or,e q u i v a l e n t l y , i f Ilu,(E)
- Auoll
Theorems 5.5 and
2
-
2
(5.38)
5.6 a r e e a s i l y s e e n t o i n p l y convergence r e s u l t s
v a l i d f o r a r b i t r a r y i n i t i a l conditions.
PARABOLIC SINGULAR PZRTURBATION
198
Let
5.7.
THEOREM
(resp. ( 2 . 2 ) w i t h u
t i o n of (2.1)
E E
0
uO(E)
~ ( 2 ) ) be
(resp.
u(^tjE)
-. uo,
arbitrary).
-. 0 &s
E'u1(E)
t h e generalized solu-
E
+
Assume t h a t
(5.39)
0.
Then U(tjE)
uniformly on compacts o f Proof.
u.
U(0) =
6> 0
Pick
r(^t)be
Let
u(t)
+
E
(5.40)
0
+
t >_ 0 .
u
and choose
E
D(A)
with
;1
- uoI/ 5 & .
t h e s o l u t i o n of t h e i n i t i a l value problem (2.2) with
Applying Theorem 5.5 and i n e q u a l i t i e s
(5.29), ( 5 . 3 0 ) , we
obtain
-
IIU(tF)
5
COE
5
U(t>ll
2 (1+
2 (u
-
IIU(tF)
+
t ) e w t/lAiiI
5
COE
2
2 w t
e
l l ~ ~ l +l
i ~-
uOl/
2
coew t l l u o ( ~ ) 2 w t
11u1(@)11+ C06 e
-
uo1l
.
(5.41)
> 0
s u f f i c i e n t l y small we c a n obviously make t h e r i g h t hand 2 2C0 & ew a i n 0 5 t 5 a , a > 0. This ends the proof.
E
side e wt
+ Taking
c0E 2ew
U(t)lI
-
C o e " tlluo(E)
2
t
-
II3t)
:(t>lI +
2
2
Concerning d e r i v a t i v e s , we have
Let
THEOREM 5.8.
that
u(t)
u(tjE),
u~,u~(E E )D(A) AuO(E) a Au, u l ( f ~ )
-
b e its i n Theorem 5.7.
Au0 -as
E
+
Assume
0.
(5.42)
-
Then
U'(tjE) uniformly on compacts of
U'(t)
4
as
E
(5.43)
0
+
t >_ 0 .
The proof follows t h e l i n e s of t h a t of t h e previous r e s u l t . 6 > 0,
,(^t)
and choose
u
E
2
D(A )
such that
llAu
- AuoI/ 5 6 .
Let
Then, i f
is a g a i n t h e s o l u t i o n of t h e i n i t i a l value problem (2.2) w i t h
U(0) = uo we a p p l y (5.29) and (5.35),
obtaining
199
PARABOLIC SINGULAR PERTURBATION
2
+ c 08 e w t
(t
0,
E
(5.44)
> 0).
T h i s completes t h e proof.
5.9.
Convergence i n (5.31) and (5.36) is uniform i n t ? 0 ( r a t h e r t h a n j u s t uniform on compacts of t ? 0 ) i f w = 0 . O f REMARK
course, t h e same observation applies t o a l l t h e other r e s u l t s i n t h i s section. For easy reference l a t e r w e c o l l e c t t h e s e p a r t i c u l a r cases of Theorems 5.5, 5.6 5.7 and 5.8 under a single heading. THEOREM 5.10.
cosine m c t i o n
Assume that A
c(Z>
with I M t ) II 5
Let
u(ht;E)
(2.1)
,
u(%)
generates a s t r o n g l y continuous
co
e w Let
u
f
Ha.
(t
2
0)
.
(5.53)
Define
IuI,
= ~~u~~ + sup ( w 2
+
l/t)a-le-W
2 tl/AS(t)ul/.
(5.54)
t>_o Obviously,
I *ICYi s
a norm i n
31
CY
(which, i n c i d e n t a l l y , makes
Ha
a
Banach space). The following two r e s u l t s a r e formal counterparts of Theorem 5.5 and 5.6.
The proof of both r e s u l t s i s based on a d i f f e r e n t estimation of t h e operator
a.
commute with A
Assume f i r s t t h a t
u
E
D(A);
since
S
and
6'
and with each other we can w r i t e
=L t
B(t;&)u
It follows from (4.1) that
V(t
-
s;E)AS(S)U d s .
(5.57)
202
PARABOLIC SINGULAR PERTURBATION
-
Take the
(1 cu)-th power of both s i d e s , take the
a-th
power of both
sides of (4.20) and multiply the i n e q u a l i t i e s thus obtained term by term.
The r e s u l t is ll5'(tF)Il
Hence, i f
u
(u)
2
+
2
l/tlffew
(t
2 01.
(5.59)
Za,
E
/b(tF)UII 5
w
t
2
(w
(t;E) 0
7.7.
such t h a t
t h e constant i n (4.20) and (4.26)) and define
=
o
for
t c 0.
There e x i s t constants
E
0'
B > 0
independent of
E
215
PARABOLIC SINGULAR PERTURBATION
-
(7 19)
Proof:
Write 6"(sjE)U
for
u
E
D(h),
of (4.27) and
where
5
0
=
X0 ( S ; E ) U +
is t h e f i r s t term on t h e r i g h t hand side
Xl i s t h e sum of t h e r e s t .
r a t h e r , Theorem 4.7)
(7.20)
Zl(S;E)U
Using Theorem 4.4 ( o r ,
we deduce 2
~ ~ ~ ( s ;5&c(w2 ) ~ -t ~ w2s-'
kt
t > 0, u
Kp(S;E)U-Kp(S
By Theorem
E
D(A).
-
tjE)U
+
s-2)eW
5
C's-2e ps
( s > 0)
.
(7.21)
We have =
4.1 we have
Putting together (7.21) and (7.23) we can estimate t h e integrand of t h e
f i r s t i n t e g r a l i n (7.22) by
Cb-2, thus t h e i n t e g r a l i t s e l f is
bounded by a constant times
1 s - t
- -s1 '
(7.24)
The second i n t e g r a l i n (7.22), a f t e r i n t e g r a t i o n by p a r t s , becomes
(7.25) A look at t h e integrand i n (7.25) makes p l a i n that it can be estimated
by a constant times
216
PARABOLIC SINGULAR PERTURBATION
thus t h e i n t e g r a l c o n t r i b u t e s another serving of (7.24).
Putting
t o g e t h e r a l l estimations and taking advantage of t h e f a c t t h a t
D(A)
is dense i n E we deduce t h a t
-p
c
+
~
2 -t),-( S-t)/2E eW( S-t)/E e
(5
-
,
E
(7 27)
t h e last two summands o r i g i n a t i n g from estimation of t h e boundary terms On t h i s basis, we proceed t o estimate t h e i n t e g r a l
i n (7.25).
(7.28) The i n t e g r a l of t h e f i r s t t e r m i n (7.27) i s computed as i n Lemma 6.3. The i n t e g r a l of t h e second term i n (7.27) i s
To compute t h e i n t e g r a l of t h e last term we make t h e change of variables
s
-
t =
t h e domain of i n t e g r a t i o n i s then
0;
< s +
(P
- 4
- b))1/2 - w ) I E 2 ( p -id2i-
A/
1x1
V/ 11-11
=
-
+ l)1’2(P/lUl)1’2)
(Re((U
2
UlAl-1’2}\U
0 < Re 1-1
i s t h e unique multiple of
h
- kdl
by
i s bounded away from zero i n t h e s t r i p t l y small, where
(p
(7.32) 1/2
]A\-’, setting u = E A we see t h a t it i s enough t o show t h a t
Multiplying numerator and denominator noting t h a t
217
(7.33)
+ 111’2 2 with
1.
sufficien-
E~
111 = P.
on t h e l i n e
p
and
We check e a s i l y t h a t (7.33) never vanishes, thus we only have t o show t h a t it i s bounded away from zero for 1 ~ +1 m. Note t h a t , f o r 1U1 = r a t t a i n s i t s minimum a t Re((u + 1)1’2(~/1~])1’2)
u
=
+ir, thus
(7.34) On t h e o t h e r
I
EOlUl -
1x1
hand,
q u + 111’2,
>
lhl-1/21U
so t h a t
Ei21~1
thus our claim holds f o r
Proof of Theorem 7.6.
T h a t the kernel
independent of
< 1, 1/2w.
K (t;E) P
satisfies
(a)
in
was shown i n Lemma 7.7, while (6.3), with B
E
likewise independent of
was t h e s u b j e c t of Lemma 7.8.
E
t h e operator
-Lt
f(;) i s bounded i n
-
_ 0 .
The function w(;;E)
= u(;jE)
- vo(G;E)
i s a s o l u t i o n of t h e i n i t i a l value problem ~
+
E2w"(tjE)
~ ( 0 ; s )=
1
w = w
U (E)
0
-
w'(t;E) = A w ( t ; E )
+ vo,
2 e-t/E Avo,
~ ' ( 0 j E ) = Ill(&)
- E -2
(8.10) To.
i s t h e s o l u t i o n of t h e homogeneous equation with t h e assigned i n i t i a l conditions and w2 is t h e s o l u t i o n
Write
-k
w2
where
w1
of the inhomogeneous equation with zero i n i t i a l conditions and 2 f ( t ; E ) = -e-t/E
We apply t o w1 while
w2
(8.W
AV0 '
Theorem 5.5 (with t h e simplified estimate (5.30)),
i s handled by means of Theorem 7.2 ( s p e c i f i c a l l y , t h e f i r s t
i n e q u a l i t y (7.4)). The final estimate i s
with t h e obvious modification i n t h e last term i f
w = 0.
This ends
t h e proof. For
N = 1 an a d d i t i o n a l c o r r e c t o r must be used, namely
2 vl(tjE) = -e -t/@
THEOREM 8.2.
1'
Assume t h a t ( 8 . 3 ) holds f o r
N = 1, t h a t i s
221
PARABOLIC SINGULAR PERTURBATION
u
0
u
=
(E)
= E - 2v 0
1 u , u 1, vo, v1
-~ and t h a t
!I
U(tjE) = u ( t ) +
D(A
U
2
0
t
u n i f o r m l y on compacts o f
O(E ),
f
+ d v l i-
o(1)
(8-131
1. Then
V0(tj&) f
0
2
t Eul
(E)
E(ul(t) +
(resp.
uo(i)
where
0,
+ O(E 2
Vl(tjE))
1
(8.14)
&
u,(t))
the s o l u t i o n o f (8.2) w i t h
u0 ( 0 ) = uo If -
i.
vo
(resp.
w = 0 , (8.14) holds uniformly i n
u,(o)
= u1 +
(8.15)
t >_ 0 .
We c o n s i d e r t h i s t i m e t h e f u n c t i o n
Proof.
w(^tjE) = U(ntjE)
-
-
vo(i;E)
= Aw(tjE)
E2w"(t;E) i w ' ( t ; E ) =
U
As i n Theorem
0
(E)
+
vo
C EV1,
2
-
W'(0,E) =
(8.16)
Evl(t;E)
t h a t s o l v e s t h e i n i t i a l v a l u e problem
w(O,E)
q.
e-t/E
U (E)
1
-
Av
0
-
2 @e-t/E Avl,
E -2V
(8.17)
-1
- E
y
1'
0
8.1, we write w as t h e sum o f a s o l u t i o n w1
(8.18)
of t h e
homogeneous e q u a t i o n t a k i n g t h e a s s i g n e d i n i t i a l c o n d i t i o n s and a solution
w2
of t h e nonhomogeneous e q u a t i o n w i t h z e r o i n i t i a l c o n d i t i o n s .
We a p p l y a g a i n t h e s e c o n d i n e q u a l i t y (7.4) t o
and
w2
(5.30) t o
W1,
obtaining
Obviously, a d i f f e r e n t t a c k must b e a d o p t e d f o r N >_ 2 , s i n c e t h e f i r s t term o n t h e r i g h t hand s i d e s of ( 8 . 1 2 ) and (8.19) c a n n o t b e squeezed smaller t h a n
level.
O(E2).
We p r o c e e d at first o n a p u r e l y f o r m a l
The a p p r o x i m a t i n g h n c t i o n w i l l b e of t h e form
u
N
(tjE)
= u (t) + EUl(t)
0
N
f
*-.
-t E UJt)
,
(8.20)
222
PAMBOLIC S I N G L U R PERTURBATION
a r e defined a s before and t h e where u0 (t), u1(^t) s a t i s f y t h e d i f f e r e n t i a l equations un' ( t ) = Aun(t)
-
~:-~(t)
u
( t >_ 0 )
(t), n 2 2, .
(8.21)
Noting t h a t t h e c o r r e c t o r s 2 vo, v1
used i n t h e cases N = 0,l a r e of 2 t h e form v O ( t j E ) = v (t/E ), vl(t;E) = v (t/E ), we s h a l l use a 0 1 combination of c o r r e c t o r s of t h e form II ( t ; E ) = v (t/E N 0
The
v
n'
n >_ 2
2
)
+
tVl(t/E
2
+
) +
N 2 VN(t/E ).
E
(8.22)
w i l l s a t i s f y t h e d i f f e r e n t i a l equations vn" ( t )
+
vA(t) = Avn-,(t)
(t
2
0)
,
(8.23)
and t h e decay condition vn(t)
-, o
as
t
4
m
and
(8.24)
n = 1,2
Note that t h e equation s a t i s f i e d by , ) : ( u U'n (
. is
(8.25)
t ) = Aun(t),
vn, n = 1 , 2 s a t i s f i e s v p )
+ vA(t)
= 0
.
(8.26)
Consider now t h e f'unction
N
I1=0 p )
= ( E 2U
N
+ &-2
N
c &"Vi(t/E2) + E-2 c E"V;l(t/E2)
+ E-2
17;O
+ up)) +
E ( E 2 u;l(t)
c E"(yll(t/&2) t VA(t/E2)) -2
+ up))
223
PARABOLIC SINGULAR PERTURBATION
+
N-2
c
EnAvn(t/E2)
n=0 N
=
c
EnA(un(t)
+
2
vn(t/E ) )
ti=O
The i n i t i a l conditions on u
Il’
u0 ( 0 ) = uo
-
E N - 1 AvNml(t/E2)
n = 0,l
+ vo,
ENAvN(t/E 2 ).
(8.28)
a r e those i n Theorem 8.1: = u1
u,(o)
+ v1
.
(8.29)
On t h e other hand, t h e i n i t i a l conditions on v n = 0 , 1 must be n’ 2 2 those t h a t insure t h a t v (t;E) = v (t/E ), v (t;E) = v,(t/E ), 0 0 1
v0 ( t ; E ) , v1(t;E)
where
a r e t h e correctors used i n Theorem 8.2.
Accordingly, $0)
= vo, vi(0) = vl,
hence, taking (8.26) and (8.24) i n t o account,
v ( t ) = -e 0
For
n
2
2,
-t vo, v,(t)
t h e i n i t i a l conditions f o r
= -e
un(i)
-t
v
1’
and
vn(t)
are,
respectively un(o) = un
thus for
un(E) tU(t;E)
-
must be constructed a f t e r
(8.32)
vn(o)
vn(t).
The i n i t i a l conditions
are obtained from (8.29) (8.30) ( 8 . 3 2 ) and (8.33):
PARABOLIC SINGULAR PERTUBBATION
224 lo ( 0 ; E ) = N
cN Enun(O) + cN Envn(o) = =O
Il=O
= u
0
- v
+
v
-
0
-
-
vn(o>>
G 2
+
EV
cN Envn(o) = cN n=2
0
=
c
+ N E n(un
“(9 + vl)
f
E
nun
(8.34)
=O
N
N
n=0
n=0
c Enu’n( 0 ) 4- c Enm2vn
N-2
c
E=O
EnU’(0) = n
c
n - 2 v n + E N-1 U&l(O)
E
I1;O
N
f
E
up)
(8.35)
Hence, i n view of (8.3), IlU(0F)
- mN(o;E)I/
= O(EPst1)
-
=
(8.36)
and IlU’(0;E)
lo$OjE)I/
O(EN-l)
.
(8.37)
We face now t h e problem of making a l l t h e s e computations valid. Roughly speaking, t h i s amounts t o :
(a) showing t h a t every d e r i v a t i v e w r i t t e n ( a s i n (8.171, (8.21), (8.24), e t c . ) a c t u a l l y e x i s t s . (b)
etc.),
showing t h a t every time we w r i t e
(as i n (8.17), (8.21),
Au
u a c t u a l l y belongs t o t h e domain of
A.
This w i l l be done by r e q u i r i n g “smoothness” conditions of varying degree on t h e c o e f f i c i e n t s
un’ vn
u o ( t > = S(t>(U0 + v&
i n (8.3). u,W
We begin with
= S(t)(U1+
a r e made e x p l i c i t i n (8.31).
“J
(8.38)
while
vo(t), vl(t)
v,(t),
v3(t) we solve (8.23) with t h e i n i t i a l condition (8.33) at and t h e decay condition (8.24) as t -. m:
t = 0
To construct
225
PARABOLIC S TNGULAR PERTURBATION
v;(o)
v2
=
-
v"(t) 3 v ~ ( o= ) v3
3
v2(t) -,o
u$o), f
-
vl(t) 3
= -e -tA v ~
-, o
u~(o), v 3 ( t )
t
as
=,
4
( t 2 0),
-
t
as
(8.40) m
.
Solving e x p l i c i t l y these e q u a t i o n s , v2(t)
= te
-t Avo
-
-t v ( t ) = t e Av
3
-t
(v2
- e -t (v3
1
- Au0 -
2Avo)
9
(8.41)
- AU1
avo)
9
(8.42)
-
A(uo + v,), U i ( 0 ) = A(U1 + vl). u s i n g t h e equation (8.21) and t h e
where w e have used t h e fact t h a t We compute next
e
u 2 ( t ) , u3(t)
U'
0
(0) =
i n i t i a l c o n d i t i o n (8.32) :
u;(t)
2
AU2(t) - S(t)A (u0 + v0)
=
( t >_ 0 ) , (8.43)
u2(o) =
?(t)
Au3(t)
=
-
U2
+ To
7
2
( t 2 0),
S(t)A (ul + vl)
(8.44) u ( 0 ) = u + v1 3 3
9
2 ) where we have used t h e f a c t s t h a t u " ( t ) = S(t)A (uo t v,), ~ " ( t = 0 1 2 (see (8.29) and (8.37)). = S(t)A (ul -t v,), v;)(O) = vo, v i ( 0 ) = v1
Hence
U,(t)
=
S(t)(U2 + v0)
=
S(t)(U2
f
v0)
-
u ( t ) = S(t)(u3
3
With
S(t
Lt
-
s)S(S)A
2
(u0 + v0) dS
tS(t)A 2 (u0 + v0) ,
-k
2
vl) - tS(t)A (ul
(8.45) -k
v~).
(8.46)
up(%),u 3 ( t ) ,
see that
v4(i),
y2(i), v3(t) already manufactured, we can e a s i l y v 5 ( t ) w i l l have t h e form v4(t)
=
e-tP4(t), v5 ( t )
=
e-tP5(t)
,
(8.47)
PARABOLIC SINGULAR PERTURBATION
226
where
is a polynomial of degree 2 whose c o e f f i c i e n t s a r e l i n e a r
P4(%)
combinations of
AJu
0’
AJvo ( j 5 j), Au2
and
Av2
uo, vo 7u2yv2 replaced by On t h e other hand, we have
t h e same polynomial with respectively.
U4(t)
- P4(0))
2tS(t)A3(uo
u (t) = S(t)(u
5
+
u
(i)
(resp. u5($))
-
vo)
- P5(0))
5
42
is
2
-2tS(t)A3(ul + vl) thus
P (t)
+ tS(t)A (u2 + v0) -
= S(t)(~4
-
and
5 u1,v1,u3,v3
t2 S(t)A4 (uo + v,),
+ tS(t)A2 (u3 i- vl)
(8.48)
-
- t2 S(t)A4(ul + v,),
(8.49)
can be constructed i f uo, vo E D(A 4 ), 4 2 D(A ), u3 E D(A ). However, i f we wish (8.47)
u E D(A ) (resp. ul,vl E 2 t o be a genuine solutions of (8.23) we a c t u a l l y need that
vo E D(A 5 ) and u E D(A 3 ), u4, P4(0) E D ( A ) ; i n view of our 0’ 2 previous comments about P4, it is s u f f i c i e n t f o r t h i s t h a t 4 3 2 uo, vo E D(A ), u2 E D(A ) v2 E D(A ) and u,, E D(A). Likewise, i f u
we wish (8.48) t o be a genuine s o l u t i o n of (8.21) we must a s k t h a t
3
2
E D(A5), u3 E D(A ) v3 E D(A ), u5 E D(A). It w i l l be of ul’ “1 i n t e r e s t l a t e r t o a s c e r t a i n t h a t u 4 ( t ) , u ( t ) a r e twice continuously 5 6 d i f f e r e n t i a b l e . This w i l l be t h e case i f u0’ V0Y U1’ v1 E D(A 1, u2, u3 E D(A 4), v2, v3 E D(A 3) and u4”-15 E D(A). From t h e s e observations we surmise t h e following r u l e s , v a l i d f o r
arbitrary m
2
1. I n t h e f i r s t place, we have
v,(t> where
(resp. Pml(%))
P,(t)
a r e l i n e a r combinations of (j
5
2m.-
A L ~A , J
3),
= e -tP*,(t)
= e-tP&),v;w,(t)
-
(8.50)
i s a polynomial whose c o e f f i c i e n t s
AJu ,AJ, 0
... Ajua-4,A3uh-4.(j
(j ~ 5~an
,
0
(j Y ~-2’~-ZjYvN-2’vN-3 u ~ , u ~ , vE ~ D(A , ~ ~).
THEOm8.3. Odd’
-4-
D(A )
3
N
,...,
N
N
n=O
n=O
(8.55) uniformly on compacts of
t 2
o
(uniformly i n
t
o
if
U)
= 0).
g
PARABOLIC SINGULAR PERTURBATION
228
i s even t h e same r e s u l t obtains under t h e assumption t h a t 2 Nt2
N>_ 2
uN>vN E D(A
1,
Proof. N = 2m
+
%-17%-29vN-19v~q-2
f
We consider f i r s t t h e case
1 and apply r u l e ( a ) .
N
odd
Avo(t),
...,AvN(t)
>_ 3;
we s e t here
Since conditions (8.51) a r e s a t i s f i e d
...
(with something t o spare) we deduce t h a t with
1.
D ( A ~ ) , . * * > U ~ > U O0, V E ~D(A >V
continuous.
vo(t), , v N ( t ) E D(A) Taking (8.50) i n t o account we
deduce t h a t
(8.56) This w i l l be used t o estimate the l a s t two terms on t h e r i g h t hand s i d e of (8.28):
f o r t h e f i r s t two terms we simply use t h e f a c t ,
u~-~(t)and
assured by (b), t h a t differentiable.
%(t)
a r e twice continuously
Using t h e f i r s t inequality (7.10) i n (8.28) we
obtain
where, i n v i e w of (8.56), the contribution of t h e l a s t two terms i s O(EW1),
This ends the proof.
The case
N
even >_ 2
i s handled
much i n t h e same way and we omit t h e d e t a i l s .
@?I. 9 E l l i p t i c d i f f e r e n t i a l equations. We apply the theory i n t h e lust eight sections t o t h e d i f f e r e n t i a l operator
m
m
i n a bounded domain R
.
m
of m-dimensional space w i t h boundary
T; here
229
PARABOLIC SINGULAR PERTURBATION
A(p)
d e n o t e s t h e r e s t r i c t i o n of
o b t a i n e d by means of t h e D i r i c h l e t
A
boundary c o n d i t i o n
o
=
U(X)
r),
(X E
(9.2)
or b y means of t h e v a r i a t i o n a l boundary c o n d i t i o n N
D ~ ~ ( X= )
The c o n s t r u c t i o n of
(x
y(x)u(x)
E
r).
(9.3)
w a s c a r r i e d o u t i n Chapter IV i n c o n s i d e r a b l e
A(@)
9IV.3 ( f o r t h e D i r i c h l e t boundary c o n d i t i o n ) and i n SN.6 (for t h e boundary c o n d i t i o n ( 9 . 3 ) ) t h a t A ( B )
d e t a i l ; i n p a r t i c u l a r , it w a s shown i n
g e n e r a t e s a s t r o n g l y continuous c o s i n e f u n c t i o n , t h u s a l l t h e r e s u l t s i n t h i s chapter apply automatically.
SVI. and u 0
5 , for 5.13.
i n s t a n c e t h e “ i n t e r m e d i a t e ” e s t i m a t i o n s i n Theorems 5.12 Combining Theorem
D ( ( b 2 1 - A(@))‘)
E
u n i f o r m l y on compacts of D((b21
-
5.u w i t h Lemma 5.15 w e deduce t h a t
-
u ( t ) I I = 0(E2’)
t > - 0.
p
Hl(n).
E
-o
(9.4)
The most i n t e r e s t i n g c a s e i s
D((b21 - A ( B ) ) if
as
can b e identified.
A(f3))‘)
if
l a r g e enough) t h e n
(b
jlu(t;E)
where
Of s p e c i a l i n t e r e s t a r e t h o s e i n
c1 =
1/2,
I n f a c t , w e s h a l l show t h a t
(9.5)
= Hi(Cl)
i s t h e D i r i c h l e t boundary c o n d i t i o n and
bl,
...,bm
belong t o
To show ( 9 . 5 ) we n o t e t h a t it h a s a l r e a d y b e e n proved t h a t
D ( ( b 2 1 - AO(@))1/2) especially
=
(see ( N . 2 . k ) ) a n d r e c a l l Theorem IV.2.2, HO(R) 1
(IV.2.6)). Thus, D((b21
We s k e t c h t h e p r o o f of
w e o n l y h a v e t o shod t h a t
- Ao(p))lb2) (9.6).
cosine function generated by used t o c o n s t r u c t
Let
- A(p))”I2)
.
(9.6)
C ( t ) = c o s h t Ao(@)1’2
Ao(p).
b e the 0 It f o l l o w s from t h e p e r t u r b a t i o n
e0(t) (or
6(t) from
cosine function generated by
= D((b21
A(p)lb
d i r e c t l y ) that
C(t),
the
c a n b e e x p r e s s e d b y means of t h e
perturbation series C ( t ) U = C0(t)U where domain
+
gTJF*Co(t)u
+ qTJF*qTJF*Co(t)u +
d e n o t e s t h e ( o n l y ) bounded e x t e n s i o n o f
O1 H (Q))
t o a l l of
* * *
,
So(t)P
(9.7) (with
L2( Q ) ; t h a t t h i s e x t e n s i o n e x i s t s follows
230
PARABOLIC SINGULAR PER’IURBATION
S ( t ) P i s bounded ( i n t h e norm o f 0
from t h e f a c t t h a t
1
L2(n))
in
~ ~ ( n )s,i n c e s o ( t ) P = (sinh t Ao(B)1/2)Ac(B)-1/2P, and
Using (1.5) and t h e “ r e c i p r o c a l ” series Co(t)u = we show t h a t
@(t)U
-
qqF*C(t)u
+
f40P*rn*C(t)u +
(9.8)
@ ( t ) u i s continuou.;ly d i f f e r e n t i a b l e i f and o n l y i f (9.6) follows f r o m Theorem
@,(ti) i s c o n t i n u o u s l y d i f f e r e n t i i b l e , t h u s
111.6.4. However, i n t h e p r e s e n t s i t u i t i o n , estimates on rates o f convergence l i k e (1.4) c a n b e o b t a i n e d under weaker assumptions b y more e l e m e n t a r y methods.
We s k e t c h below t h i s theeory i n a s u i t a b l y ” a b s t r a c t ” v e r s i o n .
Let
E = H
be a H i l b e r t s p a - e and
A.
a s e l f adjoint operator
such t h a t
with
K
>
0.
We c o n s i d e r t h e o p e r a t o r A = A.
where
P
+ P,
(9.10)
i s such that
m-l
(9.11)
i s bounded, where B = ( - A )1/2 d e f i n e d as i n srV.3. Using essentially 0 t h e same methods i n srV.4 we show t h a t A g e n e r a t e s a s t r o n g l y
c o n t i n u o u s c o s i n e f u n c t i o n , t h u s ill r e s u l t s i n t h i s c h a p t e r apply, i n particular those i n
sVI.5. We e x p l o i t t h e s e below.
Using t h e f u n c t i o n a l c a l c u l u s f o r s e l f a d j o i n t o p e r a t o r s we can d e f i n e f r a c t i o n a l powers
( -Ac)‘
of
-Ao
where
E = C1, A u = ku w i t h
A
E
= H
i s a H i l b e r t space and
It f o l l o w s from Fxercise 11.5 -that A
E
A
8(C,w)
g e n e r a t e s a s t r o n g l y continuous c o s i n e f u n c t i o n
3.1,
is a
(that
C(t)
satisfying
Ilc(t)ll 5 i f and only i f
cr(A),
C e U J l tI
t h e spectrum of
(-m
A,
- u0ll
THEOREM 5.6.
-m
= o ( E CY
Let
E,A
< t
II
if = ~ ( E c y - ~ )a s
E
4
o
be as i n Theorem 5.5, and l e t
.
(5.37) u(t;&)
be a s o l u t i o n of t h e homogeneous problem (2.1) with uO(&) E D @ ) , u ( t ) a s o l u t i o n of t h e homogeneous problem (2.2) with uo E D((-AO)1- ). Then, i f (5.24) holds t h e r e e x i s t s a constant
C(U)
such t h a t
~ l u l ( t ; E )- u * ( t > l l sc b > E a ( i + I t l ) a / 2 e " l t l ~ ~ ( - ~ o ) 1 ~ u o ~ ~ (5.38)
258
OTHER PROELEMS
The proof c o n s i s t s i n applying Theorem t h e proof of Theorem
u0
4.2).
u0 (E)
D((-AO)lw),
E
E
llul(E)
-
iAuOII
~ ' ( t )( s e e
u'(t;E),
A s a consequence, we o b t a i n e a s i l y t h a t , i f D(A)
then
- u'(t)ll
IlUYtjE)
uniformly on compacts of
5.5 t o
(Hint:
0
t h i s f o l l o w s from a s t a n d a r d
theorem on p e r t u r b a t i o n of g e n e r a t o r s of c o s i n e f u n c t i o n s . a u t h o r [ 1971:11, SHIMIZU-MIYADNU
See t h e
[1978:I], TAKENAKA-OKAZAWA [1978:11
o r TMVIS-WEBB [1981:1].
4.
EXEBCISE
(Hint:
i s a s o l u t i o n of
(8.5) w e have
u(t^)
If
i s a s o l u t i o n of
6.
u(t)
and i n t e g r a t e ) .
(8.5) w e have
m u l t i p l y t h e e q u a t i o n (8.5) scalarly by
EXFXCISE
5;
u(t^)
m u l t i p l y t h e e q u a t i o n (8.5) scalarly b y
EXERCISE 5.
(Hint:
If
u' ( t ) and i n t e g r a t e ) .
The assumptions a r e t h e same i n E x e r c i s e s
3, 4, and
we r e q u i r e i n a d d i t i o n t h a t
(Hint:
m u l t i p l y (8.7) b y
and add t o (8.6).
Take r e a l p a r t s .
Combine t h e f i r s t t h r e e i n t e g r a l terms i n t o one and use t h e f a c t t h a t
for a l l
u
E
D(Q), v
EXERCISE
7.
E
H,
consequence of (8.8)).
Assume t h a t
268
OTHW PROBLEMS
Re(Bu,u)
_ 0 .
f o r any s o l u t i o n of (1.1) I f the function
t >_ 0
(1.3)
C(t)(llu(O)lI + l l U ~ ( 0 ) l l )
C(t)
in
(1.3) can be chosen nondecreasing i n
( o r , more g e n e r a l l y , bounded on compacts of
t
0)
then we
say t h a t t h e Cauchy problem f o r (1.1) i s uniformly w e l l posed ( o r
t >_ 0 .
uniformly properly posed) i n
The propagators o r s o l u t i o n operators of (1.1)a r e defined by
u(2)
where
(resp.
u ( 0 ) = u, u ' ( 0 ) = 0
C ( t ) (resp. of
D
Since both C(t)
and
0
a(t)
v(%))
i s t h e s o l u t i o n of (1.1)with v(0) = 0, v ' ( 0 ) = u).
(resp.
The d e f i n i t i o n of
s(t)) makes and
sense f o r u E D ( r e s p . f o r u E D1). 0 Dl a r e dense i n E we can extend (using (1.3))
t o all of
E
a s bounded operators; t h e s e operator-
valued functions r e s u l t s t r o n g l y continuous i n Iic(t)li
5 C ( t > , ils(t>iI 5 C ( t >
Moreover, by d e f i n i t i o n ,
C ( 0 ) = I,
S ( 0 ) = 0.
t >_ 0
U(t) =
c(t)u(o) +
and s a t i s f y
( t >_ 0 ) .
(1.5)
F i n a l l y , we prove e a s i l y
u(%) i s a n a r b i t r a r y s o l u t i o n of (1.1)i n
t h a t if
(1-4)
S ( t ) u = v(t),
@ ( t ) u =u ( t ) ,
t 2 0
then
S(t)Ul(O).
(1.6)
The proof i s t h e same a s t h a t of (11.1.6). We s h a U assume from now on that t h e operators
A
and
B
are
closed. $vIII.2
Growth of s o l u t i o n s and existence of phase spaces.
The d e f i n i t i o n of phase space i s , except f o r small modifications, t h e same i n $111.1. A phase space i n
t
0
f o r t h e equation (1.1)
equipped with any of i t s product n o r m , @ = Eo x E 1' El a r e Banach spaces s a t i s f y i n g t h e following assumptions:
i s a product space where
E 0 (a)
(-.
D1
dense i n (b)
and
E ,E
6 E with bounded inclusion; moreover, 1Do Eo El) is dense i n E i n t h e topology of Eo ( r e s p . i s
0
n
0
El i n t h e topology of
El).
There e x i s t s a s t r o n g l y continuous semigroup G ( t )
272
THE COMPLETE EQUATION
E = E
in
0
X
t 2
E
o
1
such that
with
for any s o l u t i o n u ( i )
u(0)
E
E ~ u , l(0)
E
E ~ .
The comments a f t e r t h e d e f i n i t i o n of phase space i n $111.1apply here:
we omit the d e t a i l s . We examine i n the rest of t h e s e c t i o n t h e r e l a t i o n of t h i s notion
w i t h t h a t of w e l l posed Cauchy problem i n the case where
E = Q2 is
t h e set of a l l sequences with
2 ~ ~ { u n= ] ~ Iu ~n
c
l2
u = [un jn >_ 1)= {un] of complex numbers and A , B a r e the operators c
ACunI = lanun),
B{ un] = Fbnun],
(2.2)
rb ) sequences of complex numbers t o be determined l a t e r : n t h e domain of A c o n s i s t s of a l l {u ) E E w i t h {a,.,) E E. The n domain of B i s s i m i l a r l y defined; observe that both A and B a r e
{an]
and
normal operators commuting w i t h each other.
u(%) = [un(%)] i s a s o l u t i o n of (1.1)then each u (%) s a t i s f i e s the s c a l a r equation n u''(t> + b n u ' ( t ) + a u ( t > = 0. On t h i s b a s i s , we deduce that the n?? propagators C(%), b ( t ) of (1.1) must be given by
where
+
hn,A,
If
a r e the r o o t s of t h e c h a r a c t e r i s t i c equation h2 + b A + a n = O ,
(2.5)
h = h- (a case t h a t we w i l l n n Obviously, a necessary condition f o r t h e Cauchy problem
w i t h the modifications de rigueur when
avoid here).
f
f o r (1.1) t o be w e l l posed i n
a(t) =
Ilc(t)II
t
2
= SUP
n>_l
0
i s that
-
A+ e n
n
A$
THE COMPLETE EQUATION
273
and
be bounded on compacts of
0
5 t
1 for a l l
Also, a,
m(n> >_
t >_ 0 ,
i s increasing i n
m(^t)
accordingly t h e f u n c t i o n on compacts.
n
thus bounded
n = wn
'n
(n
2 1)
.
(2.17)
Define yn = l o g I n view of t h e i n e q a l i t y
Cyn
+ Lwl/". n n
= log w;/n
log x i x
5
2-1/2ex,
valid for
(2.18) x
5.
0, we
h.ave yn We s e l e c t now
a n,bn i n
n
-
n'
(2.19)
(2.5) i n such a way that (2.20)
We have i
( h I = p > e . n n -
(2.21)
On t h e o t h e r hand, i n view of (2.19),
2
(8, thus the sequence
A =
{A'-]n /A/
-
2 1/2
VJ
>_ Yn>
i s contained i n t h e region e,
Accordingly t h e r e e x i s t c o n s t a n t s
0 C_
Re h
0> 8
7
5 I m A. 0
independent of
fl
such that (2.22)
275
THE COMPLETE EQUATION
I n view of t h e d e f i n i t i o n (2.14) of
m(t)
we obtain from (2.23) and
(2.24) that
e(m(t> -et> 5 a(t>5
5
e(m(t) - e t > in
t 2 0.
T(t)
o(m(t>
+ e
t
1,
(2.25)
5 o ( m ( t > + et >,
(2.26)
The i n e q u a l i t i e s on t h e right-hand s i d e s of (2.25), (2.26)
imply that t h e Cauchy problem f o r (1.1) i s w e l l posed i n only remains t o choose t h e sequence i n e q u a l i t i e s (2.10) a r e s a t i s f i e d .
0
To do t h i s , we assume ( a s we
Keeping i n mind that t h e
constant i n (2.22) i s independent of t h e choice of
Both conditions (2.U) a r e obvious. the greatest integer
5 t.
It
0.
i n such a way t h a t t h e
u(%) i s nondecreasing.
obviously may) t h a t
t
we s e t
fl,
On the other hand l e t
t
2
1, n = [ t ] ,
Then, taking (2.16) and (2.27) i n t o account,
we o b t a i n
whence t h e f i r s t i n e q u a l i t y (2.10) r e s u l t s from (2.25); t h e second follows i n a similar way from (2.26).
EXAMPLE 2.2.
Let
a > 0.
Then t h e r e e x i s t A , B of t h e form
(2.2)such t h a t t h e Cauchy problem f o r (1.1)i s w e l l posed i n 0
b u t not w e l l posed i n any i n - t e r v a l 0
2
0
0.
SUP
s20
e
-us
IFr( s ) u l l ,
(4.5)
Eo
=
(4.6)
D ( A ) _C E o .
i s a Banach space i s much t h e same as t h a t f o r t h e
(111.2.1) and w e omit i t .
THEORE3l 4.2.
t
2
W e obviously have Do
equation
c o n s i s t s of all
so l a r g e t h a t ( 3 . 2 ) , t h e f i r s t i n e q u a l i t y (3.57)
W'
and ( 4 . 2 ) h o l d .
The proof t h a t
Eo
is
Eo
W',
The space
i s continuously d i f f e r e n t i a b l e i n t
l I ~ 1 1=~ IIuII + where
(4.4)
Let t h e Cauchy problem f o r
and l e t Assumption 3.1 be s a t i s f i e d .
f o r (4.1). Proof:
We must show t h a t
( 4 . 1 ) be well posed i n Then
Em i s a phase space
29 1
THE COMPLETE EQUATION
We prove f i r s t t h a t each Em. q t ) i s a bounded operator i n Em. I n order t o do t h i s we t a k e u E D0 and f i x t > 0 . Due t o time invariance of ( 4 . 1 ) t h e f u n c t i o n i s a s t r o n g l y continuous semigroup i n
u(i)
=
C ( t + g)u
i s a s o l u t i o n of ( 4 . 1 ) t h u s it follows from formula
(1.5) that C(S
+
= C(s)C(t)u
t)U
This e q u a l i t y i s extended t o
+ S(s)C'(t)u
u
aJ_1
E
Eo
(s,t
as follows:
2
(4.8)
0).
integrate i n
0 5rzt,
-rote(. +
T)u dT = c ( s )
Lt
C(7)U dT
+
and extend (4.9) t o a r b i t r a r y
u E E by denseness of we d i f f e r e n t i a t e and o b t a i n (4.8). The analogue of ( 4 . 8 ) f o r S ( t ) i s S(S
+ t)u
= C(s)S(t)u
+
u(s) = 8(s
+
Do;
u
for
E
2
0),
(4.10)
t)u, u
E
D1;
since all
u
operators i n (4.10) are bounded we can extend t h e e q u a l i t y t o all We note i n passing t h a t (4.8) i t s e l f can be extended t o all
a( s ) C ' ( t )
modified form observing t h a t
Eo
(s,t
S(s)S'(t)u *I
and i s shown by applying (1.5) t o
(4.9)
S(s)(c(t)u-u),
u
f
E
E
E.
in a
must have a bounded extension.
We s h a l l not make use of t h i s i n what follows. We prove t h a t each
qt)
i s a bounded o p e r a t o r i n
Em. To do
t h i s , we m u s t show t h a t t h e o p e r a t o r s C ( t ) :Eo
-
c ' ( t ) : Eo
E 0 E
are bounded i n t h e spaces i n d i c a t e d .
8 ( t ) :E
-
8 ' ( t ) :E
Eo
-.
(4.11) E
This i s r a t h e r obvious f o r
c'(t)
E ) and f o r a t ( t ) (from Assumption 3.1). 0 Note a l s o t h a t it follows from Corollary 3.7 and Lemma 4.1 t h a t
(from t h e d e f i n i t i o n of
292
THE COMPLETE EQUATION
( h e r e and i n o t h e r i n e q u a l i t i e s
C
denotes an a r b i t r a r y constant, not
n e c e s s a r i l y t h e same i n d i f f e r e n t p l a c e s ) . Continuity of
C(t)
C(s)C(t)u = C ( s + t ) u - S ( s ) C ' ( t ) u j
form
and d i f f e r e n t i a t e with respect t o
apply t o an element
u
+ t ) u - 8'(s)CI(t)u.
i s a bounded operator from
C(t)
i n the of
E~
We obtain
s.
Cl(s)C(t)u = C l ( s If follows t h a t
Write ( 4 . 8 )
i s proved a s follows.
Eo
(4.13) into
and
Eo
(4.14) f o r some constant
Finally, boundedness of
C.
Write (4.10) i n t h e form
It follows t h a t
' 0
i s continuous i n
+ t ) u - 8t(s)Sl(t)u.
)
5
wt
(t
Ce
(4.15)
L
(4.16)
0)
W e have t h e n completed t h e proof t h a t each
C.
Em: moreover, t h e r e e x i s t s a constant
wt
l l ~ t I l l ~ 5~ Cme ) f o r some constant
W e then
The r e s u l t i s
i s a bounded operator and
8(t)
lls(t)ll(E.E f o r some constant
i s shown a s follows.
+ t)u-S(s)S'(t)u.
C(s>S(t)u = 8 ( s
d i f f e r e n t i a t e t h i s e q u a l i t y term by term. C f ( s ) S ( t ) u= S l ( s
S(t)
C,
t h e constant
(t
C
10 )
q t )
such t h a t (4.17)
being t h e same i n Corollary 3.7
w
and Lemma 4.1. The semigroup equation
follows from (4.8) and (4.10) and t h e i r d i f f e r e n t i a t e d versions (4.13) and (4.15). n
The next step i s t o show t h a t
q t )
i s s t r o n g l y continuous.
It
i s enough t o prove t h a t Ilqh)u as
h
qtk
-
O+.
- uII( Em)
+
(4.19)
0
However, we s h a l l skip t h i s step since we show below t h a t
has a derivative at t h e o r i g i n ( i n t h e norm of
s)
for u
in
293
THE COMPLETE EQUATION
Gm; t h i s , combined w i t h t h e uniform bound (4.28)
a dense subset of obviously i m plies
THEOREM 4.3.
(4.19),
q;)
since
i s a s t r o n g l y c o n t i n u o u s semigroup w i t h
8 given by
infinitesimal generator
8=
=
c l o s u r e of
71,
(4.20)
where
w i t h domain D(%) = D ( A )
The f u n c t i o n
i s a s o l u t i o n of
u(;)
u(t)
(4.22)
( D ( A ) fI D ( E ) ) .
X
(4.1) o n l y i f
= [u(t),u'(t)l
(4.2?)
i s a s o l u t i o n of
u'(t)
=
%u(t).
Proof: We b e g i n b y showing t h a t t o p o l o g y of
Eo.
i s dense i n
D(A)
To d o t h i s we s e l e c t a
(4.24)
"6-sequence"
Eo
{@,I
i n the
of scalar
f u n c t i o n s l i k e that used i n t h e proof o f C o r o l l a r y 3.5 ( b ) , and show that
u as
n
-
( f o r any
f o r each
m
u
E
E)
(4.13) we s e e that
u
n
= J$ , ( t ) c ( t ) u
E
E
0' i s obvious.
dt
That ( 4 . 2 6 )
-
(4.26)
u
h o l d s i n t h e t o p o l o g y of
Assume now t h a t
u
E
EO.
Then, using
E
294
THE COMPLETE EQUATION
and we check e a s i l y t h a t
e-WSC'(s)un converges uniformly i n
to
un
emWSCt(s)u, s o t h a t
-
u
in
20
t
EO'
W e show next t h a t
u
Em f o r each
in
l i m i t r e l a t i o n s as
E
h
-
D(3).
This i s equivalent t o t h e following f o u r
0+:
-
h'l(C(h)u for
for
u
u
u
for
E
E
E
u
E
h-lS(h)u
-
u
(4.28)
h-lC'(h)u
-
-Au
in
Eo
(4.29)
D(A) n D(B),
D(A),
D(A)
n D(B).
- u)
+
--. -Bu
in
0
(4.31)
E
To prove (14.28) we use (4.13) i n t h e form
s,
i s bounded i n norm by a constant
t h e constant described a f t e r
e W I S , m'
(4.32) as h
(4.30)
E
in
and
This expression, as a f u n c t i o n of
times
in Eo
D(A),
h-'(S1(h)u for
u ) -, 0
(4.5).
The l i m i t of
is
C"(~)U-S'(S)C"((~)U= C"(S)U + S'(S)AU = 0 after (3.11). C1(s)(h"S(h)u
To show (4.29) we w r i t e (4.10) in t h e form
- u ) = h''(St(s
+ h)u
- S t ( s ) u ) - g'(s)h''(8'(h)u - u ) - c ' ( s ) u , (4.33)
which i s bounded i n norm as well by a constant times
as
h
-
O+
eWts; i t s l i m i t
is
S"(S)U-S'(S)S"(O)-C'(S)U
=
S"(S)U + S'(S)BU
+
S(S)AU = 0
(4.34)
295
THE COMPLETE EQUATION
i n view of Corollary 3.5(d).
F i n a l l y , t h e two l i m i t r e l a t i o n s (4.30)
and (4.31) a r e obvious, since and
u
E
h-l(W'(h)u
- u) n D(E)
-D(A) 1
D
-
-
h-lC'(h)u
( s e e Corollary
3.5
%
=
=
-AC(O)u- EC'(0)u
= -Au
-Eu f o r
(c)).
q;)
Having proved (4.27), we know t h a t semigroup and t h a t , i f
C"(0)u
- BS'(0)u
S"(0)u = -A8(0)u
i s a s t r o n g l y continuous
i s i t s i n f i n i t e s i m a l generator, t h e n
(4.35)
U C B . T o improve (4.35) t o (4.20) it w i l l be s u f f i c i e n t t o prove t h a t Uh
f o r all u
E
D(9J)
s e l e c t a sequence
In f a c t , i f
{un]
(t)u d t
= kJhE
5 D(Z)
E
D(U)
(4.36)
(4.36) i s t r u e and
-
un
with
u
in
u Qm
E
(g, we may
(that
D(8)
is
dense i n ( u )h + 11
follows from (4.26) and f r o u Corollary 3.5 ( b ) ) . Then whereas u(un) h = 8 ( u n ) h = h -1(F(h)un- un) -+ h -1( S ( h ) u - u);
uhn€ D(E)
f o r any
u
E
Gh = Assume that tends t o 8 u
u
that
E
h > 0
Qm and any
and
-
h-'(5(h)u
.
u)
(4.37)
u E D(%). Taking i n t o account t h a t t h e r i g h t s i d e of (4.37) as h+O+ it follows from t h e fact t h a t i s closed
D(@
u
and
uu
= %u, which completes t h e proof of (4.20).
The i n c l u s i o n r e l a t i o n (4.36) c a n be reduced t o t h e f o u r r e l a t i o n s
(4.38)
(4.39) / g h C l ( t ) u d t = C(h)u
L h S t ( t ) u d t = S(h)u-u If
u
E
D(A)
E
-
u
D(A)
we have
E
D ( A ) fl D ( B )
n D(B)
(U E
(u
D(A)
=L
E
D(A)),
(4.40) (4.41)
D(E)).
h
AJOhC(t)u d t
so t h a t (4.38) h o l d s .
8( a + p l o g (1 + I h l ) I n Exercises ASSUMPTION
u
for all
of
E
{u
(5.8)
5 t o 11 we r e q u i r e p a r t ( a ) of Assumption 3.1, t h a t i s
5.1.
i s continuously d i f f e r e n t i a b l e i n t
S(i)u
E
5.
U s i n g Exercise 2 show t h a t t h e operator
0
Do Tl D(B); Bu
D1]
E
8(t)B
(with
h a s a bounded extension
t o all
given by = C(t)
EXERCISE 6.
Define
with
m(F)
-
Sf(t)
.
(5.9)
R(h;c~) as i n (3.19),
R( h;fn)u
Jbwe-htm(t)s(t)u
=
,
dt
a t e s t f u n c t i o n i d e n t i c d l y equal t o
(5.10)
1 near zero.
(a s l i g h t l y extended v e r s i o n o f ) (5.6) show t h a t if u such t h a t
Au, Bu
E
where
N(t;rp) = 2 v f ( t ) 3 ' ( t ) EXERCISE
Show t h a t for
7.
E
Do
Using
n D(B)
is
D1 t h e n (3.37) holds, that i s ,
+ $h;a)u
R(h;a)P(h)u = u
Define
-t
p"(t)s(t)
R(h)
as i n
+
,
(5.11)
a ' ( t ) W .
(3.41)
for
Reh
3 W,
w l a r g e enough.
u as i n Exercise 6 we have R(h)P(h)u = u
EXERCISE 8.
h(:;rp)
2
E E.
EXERCISE domain
.
For Reh
given by (3.46).
R( A). EXERCISE 9 .
Define
2
w,
W
.
l a r g e enough, d e f i n e
S(;)
Prove that t h e Laplace transform of
(5.12)
- (3.48),
by
8 equals
301
THE COMPLETE EQUATION
jta-'/r(a) ya ( c o n v o l u t i o n by
;lo
0)
(t < 0 )
Ya produces t h e " a n t i d e r i v a t i v e of order a " ) . Show
t h a t , m u l t i p l y i n g (5.12) by o b t a i n , u s i n g Exercise
for
2
(t
and i n v e r t i n g Laplace t r a n s f o r m s we
8,
as i n Exercise 6.
u
EXERCISE 10. Assume t h a t t h e s e t of d1 u
Au, Eu
E
D1 i s dense i n t h e space X
Do
E
n D(B)
= D(A)
n D(B)
such t h a t
endowed with t h e
norm
EXERCISE 11. Snow t h a t , i f
(Y1
€3 I
+ Y2
Combining (5.13) and
@ E
u
E
D1,
+ Y @A) * 3
SU = Y
€3 u
3
(t
2
(5.14)
0).
( 5 . 1 4 ) , prove t h a t , under t h e c o n d i t i o n s of
Exercise 10,
qt) = E(t). EXERCISE 12.
Under t h e c o n d i t i o n s of Exercise 1 0 , show t h a t
(3.53) ( r e s p . (3.54)) h o l d s for st(;)
formula
t h a t there exist constants
EXERCISE
u
E
E
SL(t)B).
c(t)u
Show
i s continuously d i f f e r e n t i a b l e i n
if
t 10
(4.3) h o l d s , so t h a t Ilcl(t)UII
C
(resp. f o r
such t h a t
C,W
13. Under t h e c o n d i t i o n s of Exercise 8 show t h a t ,
i s such t h a t
t h e n formula
with
(5.15)
( b u t not
w)
5
Ce
wt
may depend on
under t h e p r e s e n t hypotheses.
(t
u.
L
0)
-
(5.17)
Show t h a t Theorem 4 . 2 i s v a l i d
THE COMPLETE EQUATION
302
EXERCISE w e l l posed i n
14. 0
5t 5
E
D(A)
n D(B)
i s well posed i n
2
0
u E.
i s dense i n
t
t h a t Assumption
a (a > 0),
t h e r e and t h a t t h e s e t of all
Bu
(5.3) i s 3.1 i s s a t i s f i e d
We suppose h e r e t h a t t h e Cauchy problem f o r
f
D ( A ) I- D(E)
such t h a t
Then t h e Cauchy problem for
(5.3)
and Assumption 3.1 i s s a t i s f i e d . Note t h a t 611
t h e assumptions i n t h i s Exercise a r e s a t i s f i e d f o r t h e incomplete equation
u"(t) + Au(t) = 0
(5.18)
under t h e only assumption t h a t t h e Cauchy problem for
(5.18) i s w e l l
posed; of course, t h e r e s u l t for (5.18) can be proved i n a more elementary way by ad hoc methods. FOGTNOTES TO CHAPTER VIII
(1) We note t h e i n c o n s i s t e n c y of n o t a t i o n involved i n w r i t i n g t h e incomplete e q u a t i o n
u" + Au
=
0,
and not
u"
=
Au
as i n Chapters I1
and 111. (2)
Although t h e argument could be completed using (3.25), t h e " l e f t -
(3.41) s i m p l i f i e s some of t h e arguments. ( 3 ) We might s e t h e r e W1 = min(U,wl): f o r i f W ' < U, R(A) c a n be ana'Lytically continued t o Reh > W ' by means of Q ( h ) . ( 4 ) Convolution by Y i s employzd h e r e t o avoid using convolution of 3
handed" r e p r e s e n t a t i o n
distributions.
303
BIBLIOGRAPHY
R. A. ADAMS [1975:1]
Sobolev Spaces. Academic Press, New York, 1975
L . AMERIO and G. PROUSE
[1971:1]
Almost-periodic Functions and Functional Equations. Van Nostrand - Reinhold, New York, 1971
A. V. BALAKRISHNAN [1960:1]
Fractional powers of closed operators and the semigroups generated by them. Pacific J. Math. 10 (1960) 419-437
S. BANACH [1932:11
Th6orie des OpCrations LinCaires. Monografje Matematyczne, Warsaw, 1932
H. BART [1977:1]
Periodic strongly continuous semigroups. Ann. Mat. Pura
m. 115 (1977) 311-318
H. BART and S. GOLDBERG [1978:1]
Characterizations of almost periodic strongly continuous groups and semigroups. Math. Ann. 236 (1978) 105-116
L. R. BRAGG and W. DETTMAN [1968:1]
An operator calculus for related partial differential equations. J. Math. Anal. Appl. 22 (1968) 261-271
P. L. BUTZER and H. BERENS [1967:1]
Semi-groups of Operators and Approximation. Springer, New York, 1967
J. CHAZARAIN [1971:1]
ProblGmes de Cauchy abstraits et applications a quelques problGmes mixtes. Jour. Functional Analysis 7 (1971) 386-445
304
I.
BIRLIOGRAPHY
CIORXNESCU
[1974:1]
Sur les solutions fondamentales d'ordre fini de croissance. Math. Ann. 211 (1974) 37-46
[1982:1]
On periodic distribution semigroups. Integral Equations and Operator Theory 7 (1982) 27-35
C. CORDUNEANU [1968:1]
Almost Periodic Functions. Interscience-Wiley, New York, 1968
G. DA PRATO [1968:1]
Semigruppi periodici. Ann. Mat. Pura Appl. 78 (1968) 55-67
G. DA PRATO and E. GIUSTI [1967:1]
Una caratterizzazione dei generatori di funzioni coseno astratte. Boll. Un. M,it. Italiana (3) 22 (1967) 357-362
E. M. DE JAGER [1975:1]
Singular perturbations of hyperbolic type. Nieuw Arch. Wisk. 23 (1975) 145-171
L. DE SIMON [1964:1]
Un'applicazione della teoria degli integrali singolari allo studio d e l l e equazioni differenziali lineari astratte del primo ordine. Renrl. Sem. Mat. Univ. Padova 34 (1964) 205-223
J. W. DETTMAN [1973:1]
Perturbation techniqws for related differential equations. J. Differential Equations 14 (1973) 547-558
J. DIXMIER [1950:1]
Les moyennes invariantes dans les semi-groupes et leurs applications. Acta Sci. Mat. Szeged 12 (1950) 213-227
N. DUNFORD and J. T. SCHWARTZ [1958:1]
Linear Operators, part I. Interscience, New York, 1958
[1963:1]
Linear Operators, part 11. Interscience-Wiley, New York, 1963
BIBLIOGRAPHY
305
H. 0. FATTORINI [ 1969:11
Sur quelques bquations diffgrentielles pour les distributions vectorielles. C. R. Acad. Sci. Paris 268 (1969) A707-A709
[ 1969:21
Ordinary differential equations in linear topological spaces, I. LDifferential Equations 5 (1969) 72-105
[ 1969:3]
Ordinary differential equations in linear topological spaces, 11. J. Differential Equations 6 (1969) 50-70
[ 1970: 11
On a class of differentia1 equations for vector valued distributions. Pacific J. Math. 32 (1970) 79-104
[ 1970:2.1
Extension and behavior at infinity of solutions of certain linear operational differential equations. Pacific J. Math. 33 (1970) 583-615
[ 1970:3]
Uniformly bounded cosine functions in Hilbert space. Indiana Univ. Math. J. 20 (1970) 411-425
[ 1971:13
Un teorema de perturbacibn para generadores de funciones coseno. Rev. Uni6n MatemBtica Argentina 25 (1971) 291-320
[1976:1]
Some remarks on convolution equations for vector valued distributions. Pacific J. Math. 66 (1976) 347-371
[ 1980:11
Vector valued distributions having a smooth convolution inverse. Pacific J. Math. 90 (1980) 347-372
[ 1981:1]
Some remarks on second order abstract Cauchy problems. Funkcialaj Ekvacioj 24 (1981) 331-344
[ 1983:1]
The Cauchy Problem. Encyclopedia of Mathematics and its Applications, vol. 18, Addison-Wesley, Reading, Mass. 1983
[ 1983:2]
Convergence and approximation theorems for vector valued distributions. Pacific J. Math. 105 (1983) 77-114
983:3]
A note on fractional derivatives of semigroups and cosine
functions. Pacific J. Math. 109 (1983) 335-347
983:4]
Singular perturbation and boundary layer for an abstract Cauchy problem. Jour. Math. Anal. Appl. 97 (1983) 529-571
985:1]
On the Schrodinger singular perturbation problem. To appear.
[ 1985:2]
On the growth of solutions of second order linear differential equations in Banach spaces. To appear.
[ 1985:3]
The hyperbolic singular perturbation problem: an operatortheoretical approach. To appear.
306
BIBLIOGRAPHY
W. FELLER [1953:1]
On the generation of unbounded semi-groups of bounded linear operators. Ann. of Math. 57 (1953) 287-308
S. R. FOGUEL [1964:1]
A conterexanlple to a problem of Sz.-Nagy. Proc. Amer. Math. SOC. 15 (1964) 788-790
A. FRIEDMAN [1969:1]
Partial Differential Equations. Holt-Rinehart-Winston, New York, 1969
[1969:2]
Singular perturbation for the Cauchy problem and for boundary value problems. J. Differential Equations 5 (1969) 226-261
R. GEEL [1978:1]
Singular Perturbations of Hyperbolic Type. Mathematical Centre Tracts no 98, Mathematisch Centrum, Amsterdam, 1978
J. GENET and M. MADAUNE
[1977:1]
Perturbations singulisres pour une classe de problsmes hyperboliques non lin6aires. Lecture Notes in Mathematics, vol. 594 (1977) 201-231, Springer, New York, 1977
E. GIUSTI [1967:1]
Funzioni coseno periodiche. B o l l . Un. Mat. Ital. (3) 22 (1967) 478-485
J. GOLDSTEIN [1969:1]
Semigroups and second order differential equations. J . Functional Analysis 4 (1969) 50-70
[1969:2]
An asymptotic property of solutions of wave equations. Proc. Amer. Math. SOC. 63 (1969) 359-363
[1970:1]
An symptotic property of solutions of wave equations, 11. J. Math. Anal. Appl. 32 (1970) 392-399
[1970:2]
On a connection between first and second order differential equations in Banach spaces. J. Math. Anal. Appl. 30 (1970) 246-251
[1972:1]
On the growth o f solutions of inhomogeneous abstract wave
equations. J. Math. Anal. Appl. 37 (1972) 650-654
307
[1974:1]
On the convergence and approximation of cosine functions. Aeq. Math. 10 (1974) 201-205
J. GOLDSTEIN and J . T. SANDEFUR [1976:1]
Asymptotic equipartition of energy for differential equations inHilbert space. Trans. Anier. Math. SOC. 219 (1976) 397-406
I. S. GRADSTEIN and I. M. RIDZYK [1963:1]
Tables of Integrals, Sums, Series and Derivatives. (Russian). Gostekhizdat, Moscow, 1963
F. P. GREENLEAF [1969:1]
Invariant Means on Topological Groups and Their Applications. Van Nostrand, New York, 1969
R. GRIEGO and R. HERSH [1971:1]
Theory of random evolutions with applications to partial .~ differential equations. Trans Amer. Math. SOC. 156 (1971) 405-418
J . HADAMARD
[1923:1]
Lectures on Cauchy's Problem in Linear Partial Differential Equations. Yale University Press, New Haven, 1923. Reprinted by Dover, New York, 1952
P. R. HALMOS [ 1967:11
A Hilbert Space Problem Book. Van Nostrand, New York, 1967
H. HELSON [1953:1]
Note on harmonic functions. Proc. Amer. Math. SOC. 4 (1953) 686-691
E. HILLE [ 1938:11
On semi-groups of linear transformations in Hilbert space. Proc. Nat. Acad. Sci. USA 24 (1938) 159-161
[ 1948:11
Functional Analysis and Semi-Groups. Amer. Math. SOC. Colloquium Pubs. vol. 31, New York, 1948
[ 1952:1]
On the generation of semi-groups and the theory of conjugate . functions. Comm. Sgm. Math. Univ. Lund Tome Supple'mentaiTe (1952) 122-134
BIBLIOGRAPHY
308 [ 1952:2]
A n o t e o n C a u c h y ' s p r o b l e m . Ann. SOC. P o l . Math. 2 5 ( 1 9 5 2 )
56-68 [ 1953: 11
S u r l e p r o b l s m e a b s t r a i t d e Cauchy. C. R . Acad. S c i . P& 236 ( 1 9 5 3 ) 1466-1467
[ 1953:2]
Le p r o b l 6 m e a b s t r a i t d e C a u c h y . U n i v . e P o l i t e c n i c o T o r i n o Rend. Sem. Mat. 1 2 ( 1 9 5 3 ) 95-105
[ 1 9 5 4 : 1]
Une g & n & r a l i s a t i o n d u problBme d e C a u c h y . Ann. I n s t . F o u r i e r G r e n o b l e 4 ( 1 9 5 4 ) 31-48
[ 1954:2]
T h e a b s t r a c t Cauchy p r o b l e m a n d C a u c h y ' s p r o b l e m f o r p a r a b o l i c p a r t i a l d i f f e r e n t i a l e q u a t i o n s . <J. A n a l y s e Math. 3 ( 1 9 5 4 ) 81-196
[ 1 9 5 7 : 11
Probl6me d e Cauchy: e x i s t e n c e e t u n i c i t 6 d e s s o l u t i o n s . B u l l . Mat. Soc. S c i . M,ith. P h y s . R . P. Roumaine ( N . S . ) 1 ( 1 9 5 7 ) 141-143
E. HILLE a n d R . S. PHILLIPS [1957:1]
F u n c t i o n a l A n a l y s i s a n d Semi-Groups. Amer. Math. SOC. Colloquium Pubs. v o l . 3 1 , P r o v i d e n c e , 1957
G . C. HSIAO a n d R . J . WEINACHT [1979:1]
A s i n g u l a r l y p e r t u r b e d Cauchy p r o b l e m . 7 1 (1979) 242-250
J. Math. A n a l . A p p l .
N . JACOBSON
[1953:1]
L e c t u r e s o n A b s t r a c t A l g e b r a . Van N o s t r a n d , N e w Y o r k , 1953
T. KATO [1976:1]
P e r t u r b a t i o n Theory for L i n e a r O p e r a t o r s . 2nd. e d . S p r i n g e r , N e w York, 1976
J. KEVORKIAN [1981:1]
a n d J . COLE P e r t u r b a t i o n Methods i n Applied Mathematics. S p r i n g e r , New Y o r k , 1981
J . KISYNSKI [1963:1]
Sur l e s Cquations hyperboliques avec p e t i t e paramktre. C o l l o q . Math. 10 (1963) 331-343
[1970:1]
On s e c o n d o r d e r C a u c h y ' s p r o b l e m i n a Banach s p a c e . B u l l . Acad. P o l o n . S c i . S e r . S c i . Math. A s t r o n o m . P h y s . 18 ( 1 9 7 0 ) 371-374
RIR1,TOGRAPHY
[1971:1]
309
On operator-valued solutions of D'Alembert's functional equation, I. Colloq. Math. 23 (1971) 107-114
[1972:1]
On operator-valued solutions of D'Alembert's functional equation, 11. Studia Math. 42 (1972) 43-66
[1972:2]
On cosine operator functions and one-parameter semigroups of operators. Studia Math. 44 (1972) 93-105
[ 1974: 1]
On M. Kac's probabilistic formula for the solution of the telegraphist's equation. Ann. Pol. Math. 29 (1974) 259-272
H. KOMATSU [1966:1]
s. G.
Fractional powers of operators. Pacific J. Math. 19 (1966) 285-346
KRE~N
[1967:1]
Linear Differential Equations in Banach Spaces (Russian). Izdat. "Nauka", Moscow, 1967
S. G. KREYN, J u . I. PETUNIN and E. M. SEMENOV [ 1978: 11
Interpolation of Linear Operators (Russian). Izdat. "Nauka", Moscow, 1978
S. KUREPA [ 1962: 1]
A cosine functional equation in Banach algebras. Acta Sci. Math. Szeged 23 (1962) 255-267
[ 1972: 11
Uniformly bounded cosine functions in a Banach space. Math. Balkanica 2 (1972) 109-115
[ 1973: 11
A weakly measurable selfadjoint cosine function. Glasnik Mat. Ser. I13 8 ( 2 8 ) (1973) 73-79
[ 1976: 11
Decomposition of weakly measurable semigroups and cosine operator functions. Glasnik Mat. Ser. 111 11 ( 3 1 ) (1976) 91-95
B. LATIL [ 1968: 11
Singular perturbations of Cauchy's problem. J. Math. Anal. 23 (1968) 683-698 &la
P. D. LAX and R. RICHTMYER [1956:1]
Survey of the stability of linear finite difference equations. Comm. Pure Appl. Math. 9 (1956) 267-293
310
BIB1,IOGRAPHY
J. L. LIONS [1957:1]
Un remarque slur les applications du th6orPme de Hille Yosida. J. Math. SOC. .lapan 9 (1957) 62-70
E. R. LORCH [1941:1]
The integral represent,ition of weakly almost periodic transformations in reflexive vector spaces. Trans. Amer. Math. SOC. 49 (1941) 18-40
D. LUTZ [ 1980: 11
Uber Operatorwertige Losungen der Funktionalgleichung des Cosinus. Math. Z. 171 (1980) 233-245
V. P. MIKHAILOV [1976:1]
Partial Differentia1 Equations (Russian). Izdat. "Nauka", Moscow, 1976.
B. C. MITJAGIN [1961:1[
Differential equations with small parameter in a Banach space. Izv. Akad. Nauk. Azerbaidzan, Ser. Fiz.-Matem. i Tehn. 1 (1961) 23-38
I. MIYADERA [1951:1]
On one-parameter semigroups of operators. J. Math. Tokyo 1 (1951) 23-26
[1952:1]
Generation of a strongly continuous semi-group of operators. Tahoku Math. J. (2) (1952) 109-114
B. NAGY [1976:1]
Cosine operator functions and the abstract Cauchy problem. Period. Math. Hungar. 7 ( 3 ) (1976) 15-18
H. S. NUR [1971:1]
Singular perturbations of differential equations in abstract spaces. Pacific J. Math. 36 (1971) 775-780
C. F. MUCKENHOUPT [1928/29:1] Almost periodic functions and vibrating systems. Jour Math. P h y s . M.I.T. (1928/29) 163-199
RlBLIOGRAPHY
311
E. W. PACKEL [1969:1]
A semigroup analogue of Foguel's counterexample. Proc. Amer. Math. Soc. 21 (1969) 240-244
R. S. PHILLIPS [1951:1]
On one-parameter semi-groups of linear transformations. Proc. Amer. Math. SOC. 2 (1951) 234-237
[1953:1]
Perturbation theory for semi-groups of linear operators. Trans. Amer. Math. Soc. 74 (1953) 199-221
[1954:1]
A note on the abstract Cauchy problem. Proc. Nat. Acad. Sci.
USA 40 (1954) 244-248
_ .
S. I. PISKAREV [1979:1]
Discretization of an abstract hyperbolic equation (Russian). Tartu Riikl. Uel. Toimetised N o 500 Trudy Mat. i Meh. 25 (1979) 3-23
[1982:1]
Periodic and almost periodic cosine operator functions. Mat. Sb. 118 N o 3 (1982) 386-398
C. PUCCI and G. TALENT1 [1974:1]
Elliptic (second-order) partial differential equations with measurable coefficients and approximating integral equations. Advances in Math. 19 (1976) 48-105
F. RIESZ and B. SZ.-NAGY [1955:1]
Functional Analysis. Ungar, New York, 1955
H. SLICHCHTING [1951:1]
Grenzschicht-Theorie. Braun, Karlsruhe, 1951
A. Y. SCHOENE [1970:11
Semi-groups and a class of singular perturbation problems. Indiana Univ. Math. J. 20 (1970) 247-263
M. SHIMIZU and I. MIYADERA [1978:1]
Perturbation theory for cosine families in Banach spaces. Tokyo J. Math. 1 (1978) 333-343
J. SMOLLER [ 1965:1]
Singular perturbations and a theorem of Kisy6ski. J. Math. Anal. A p p l . , 1 2 (1965) 105-114
312
[ 1965:2]
BIBLIOGRAPHY
S i n g u l a r p e r t u r b a t i o n s of Math. 28 (1965) 665-677
C a u c h y ' s p r o b l e m . Comm. P u r e A p p l .
M. SOVA
[1966:1] [1968:1]
C o s i n e o p e r a t - o r f u n c t i o n s . Rozprawy Mat. 49 (1966) 1-47 P r o b l s m e d e Cauchy poiir k q u a t i o n s h y p e r b o l i q u e s o p k r a t i o n n e l l e s Ann S c u o l a Norm. S u p . P i s a (3) 22 (1968) 67-100
2 c o e f f i c i e n t s c o n s t a i i t s non-born6s. [ 1968: 21
Unicitk des solutions exponentielles des kquations diffkrent i e l l e s o p 6 r a t i o n n e l l t s s l i n k a i r e s . B o l l . Un. Mat. I t a l . ( 4 ) 1 (1968) 629-699
[ 1968:3:
Solutions pkriodiques des kquations d i f f k r e n t i e l l e s o p k r a t i o n n e l l e s : Le miathode d e s d k v e l o p p e m e n t s d e F o u r i e r . C a s o p i s PeBt. Mat. 93 (1968) 380-421
[1968:4]
S e m i g r o u p s arid c o s i n e f u n c t i o n s of n o r m a l o p e r a t o r s i n H i l b e r t s p a c e s . C a s o p i s P e s t . Mat. 93 (1968) 437-458
[ 1970:1]
b q u a t i o n s d i f r f k r e n t i e l l e s o p k r a t i o n n e l l e s v l i n k a i r e s du s e c o n d o r d r e 2 c o e f f i c i e n t s t o n s t a n t s . Rozprawy C e s k o s l o v e n s k e Akad. Vgd. Rada Mat.. P r i r o d . Vgd. 80 (1970) s e z i t 7, 1-69
[ 1970:2]
kquations hyperboliquc,s a v e c p e t i t param6tre dans les e s p a c e s d e B a n a c h g k n 6 r a u x . C o l l o q . Math. 21 (1970) 303-320
[ 1970:3]
R k g u l a r i t k d e l ' e v o l u t i o n l i n k a i r e isochrone. Czechoslovak Math. J. 20 (1970) 251-302
[ 1970:4]
P e r t u r b a t i o n s nEmkriqiies d e s e v o l u t i o n s p a r a b o l i q u e s e t h y p e r b o l i q u e s . Casopi:; P e g t . Mat. 96 (1971) 406-407
[ 1972: 11
E n c o r e s u r l e s k a u a t i o n s h v o e r b o l i o u e s a v e c p e t i t param6tre d a n s l e s e s p a c e s d e Bzinach g 6 n k r a u x . C o l l o q . M a t h . 25 (1972) ,
I
135-161 Y
P . E. SOBOLEVSKII [ 1962:1]
On s e c o n d - o r d e r d i f f e r e n t i a l e q u a t i o n s i n a B a n a c h s p a c e 146 (1962) 774-777 ( R u s s i a n ) . D o k l a d y Akzid. Nauk.
[1964:11
On s e c o n d o r d e r e q u a t - i o n s w i t h a s m a l l p a r a m e t e r i n t h e h i g h e r d e r i v a t i v e . U s p e h i Mat. Nauk. 19 (1964) 217-219
SSSR
E. M. STEIN
[197O: 11
S i n g u l a r I n t e g r a l s a n d D i f f e r e n t i a b i l i t y P r o p e r t i e s of F u n c t i o n s . P r i n c e t o n l l n i v e r s i t y P r e s s , P r i n c e t o n , 1970
E. M. STEIN a n d G . WEISS [ 1971:1]
I n t r o d u c t i o n t o F o u r i e r A n a l y s i s on E u c l i d e a n S p a c e s . P r i n c e t o n U n i v e r s i t y P r e s s , P r i n c e t o n , 1971
BIBLIOGRAPHY
313
M. H. STONE [1932:1]
On one parameter unitary groups in HIlbert space. Ann. Math. 33 (1932) 643-648
B. SZ.-NAG1 [1938:1]
On semi-groups of self-adjoint transformations in Hilbert space. Proc. Nat. Acad. Sci. USA 24 (1938) 559-560
[1947:1]
On uniformly bounded linear transformations in Ililbert space. Acta Sci. Math. Szeged 11 (1947) 152-157
T. TAKENAKA and N. OKAZAWA [1978:1]
A Phillips-Miyadera type perturbation theorem for cosine functions of operators. Tahoku Math. J. 30 (1978) 107-115
C. C. TRAVIS and G. F. WEBB
[1981:1]
Perturbation of strongly continuous cosine family generators. Colloquium Math. 45 (1981) 277-285
G. N. WATSON [1948:1]
A treatise on the Theory of Bessel Functions. 2nd. ed. Cambridge University Press, Cambridge, 1948
B. W E I S S
[1967:1]
Abstract vibrating systems. J. Math. M g c L 17 (1967) 241-255
G. WHITHAM [1974:1]
Linear and Nonlinear Waves. Wiley-Interscience, New York, 1974
D. V . W I D D E R
[1931:1]
Necessary and sufficient conditions for the representation of a function as a Laplace integral. Trans Amer. Math. SOC. 33 (1931) 851-892
[1946:1]
The Laplace Transform. Princeton University Press, Princeton,
1946 K. YOSIDA
11948:ll
On the differentiability and the reuresentation of oneparameter semigroups of linear operators. J. Math. SOC. Japan 1 (1948) 15-21
[1978:1]
Functional Analysis. 5th. ed. Springer, Berlin 1978
314
BIBLIOGRAPHY
J . ZABCZYK
[1975:1]
A note on Co semigroups. B u l l . Acad. Polon. Sci. Ser. Sci. Math. Astronom. Phys. 23 (1975) 895-898
E. ZAUDERER
[1983:1]
Partial Differential Equations of A p p l i e d Mathematics. Wiley-Interscience, New York, 1983
A. ZYGMUND [1959:1]
Trigonometric Series, v o l I. Cambridge University Press, Cambridge, 1959