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!
=
, since
=
. It follows that contradicting
P(D)
again contradicting
m ¢ N ,
!~erefore,
n Pt D P(D)t ~ O
Lemma 2 (i/x m) ~ C~(R I) + D'(R I) ,
P(D)x = ~ . Then
< t 2 , • > ~ D'(R I)
be such that
7/0 c S~ . One
the hypothesis
WV
97
Proof Assume, it is false. Then
(50)
i/x m = @ + T
for certain (51)
tp ~ C°~(R1) and T t
for certain
T e D' (R1)
Hence
0
=X
RI\ { 0}
X e C°°(RI\{0})
. But, a c c o r d i n g t o the d e f i n i t i o n
of
D ' (RI)
it follows
that
(52)
S
=
~
%tOrT ¢ D~(R 1)
o~r~-q for certain
q c N
S1
(53)
,
~r ~ C1
S ] I RI\{0}
=
% q ~ 0 . Then.
,
¢ D~(RI\{0})
Now, (51-53) imply
(54)
S 1 = P(D)x e C#°(RI\{0})
where
P(D) =
SI = 0
E X D r . As o~r~q r
6¢~ n D~ = {0}
the relations (53)
(54) result in
~
which together with (50-52) gives
P(D)(1/x TM) = P(D)~
on
RI\{0}
Computing the derivative in the left side, one obtains
Z ( - 1 ) r (m+r-l)! ( m - l ~ X x q - r = xm+q P(D)~(x) ! r o~r~q Taking the l i m i t f o r (-1)q
x + 0 , one o b t a i n s
(m+q-l)!
(m-l):
X
q
=
0
contradicting the assumption that
§6.
g x ¢ RI\{0}
l
q
# 0
VVV
A PROPERTY OF THE DERIVATIVE IN THE ALGEBRAS
In the present section, the case of derivative algebras containing
D'(R I)
will be con
sidered. According to the general result in I) in Theorem 3, chap. 1 , §8, the derivative mappings within the algebras Dqp+q : Ap+q ÷ Ap , coincide on
C~(R I)
p ~ N ,
q e N
with the usual derivatives
will be strengthened in Theorem 7.
Dq of smooth functions. That result
98
T Z c T , the same I) in Theorem 3, chap. i,
First, we notice that due to the inclusion
§ 8, implies that the derivative mappings within the algebras coincide on C ~ ( R I ) Q D ~ ( R I)
with the usual distribution derivatives.
Theorem 7 Given any distribution
(V,TQS1) tive
T • D ' ( R 1) \
such that
within
mappings coincide on
rivatives, where
MT
oo 1 (C ( R ) + D ~ ( R 1 ) )
the corresponding
algebras
C~(R I) + D~ (RI) + M T
is the vector subspace in
{ T , DT , D2T . . . .
there
exist
regularizations
Ap , p ~ N , t h e d e r i v a -
with the usual distribution deD' (RI)
generated by
} .
Proof Assume
(UQS~)
T =
for
a certain
t e S
n P t = 0 . But, obviously
. Then, according
o S~ =
VoQT z , for
Z ~ g~ , one can choose a vector subspace
(55)
VoQ~ZQs
(s6)
uGp t c sI
Taking
in
So
§5,
Z ¢ Z~ • Therefore, given
any
such that
I = so
T : T~ , the relations
(V,TQSI)
SI
t o Lemma l ,
(55), (56) will imply (16) and (17), therefore
will be a regularization. Noticing that
HT:{<s,'>
I s ~pt }
and taking into account i) in Theorem 3, chap. i, §8, the proof is completed
§7.
THE EXISTENCE OF THE SEQUENCES IN
In order
to prove that z
x
*
~
(see ,
Z
s
v
o
belonging to
Z°
o
§3) x
o
~R n
,
it is obviously sufficient to show that ces
Z
~ ~ . In this respect, a class of sequeno will be constructed by a proper generalization to n > 1
dimensions of the method in (7) and (8). Suppose
(57)
~ e D(R n)
such that
s~(V) (x) = UI(V)
~n ~(x)dx = I
" ...
" ~n(~)
and define
• ~(l~l(u)x I V
N ~
÷
U(~) = ( U I 0 ) ) , - - . , U n ( U ) )
so • W
.....
by
Un(U)x n)
,
V c N , x = (x I ,..., x n) ~ Rn
where the mapping (58)
~V
c Nn
99
is constructed
in (59-62).
First,
N ~ V ÷ k(~) e N ,
define
(59)
k(0)
Define
also
= 0
N By
and
÷ h(v)
k(v+l)
h(0)
D e f i n e now
N ~ ~ ÷ e(v) E N n , by
(61)
e(V)
Finally,
define
and
(62.2)
{ p(k(v)+l)
= (1,...,1)
= h(v)
+ ~ + 1 ,
~
V
9 E N .
,
V
~ E N .
V e N .
V
c Nn
....
, #(k(V+l))
can be written
,
} = P(n,v+l)
in Fig.
in terms
M4 = { p ( k ( 3 ) + l )
Lemma
+ £(n,v+l)
,h(v)))
(58) is illustrated
M4
= k(v)
(58), by
~(0)
The mapping
h(v+l)
= (h(v) ....
(62.1)
noted by
(see §3):
E N , by
(60)
= 0
by
...
+ e(v+l)
4) in the case of
of (62.2),
, p(k(4))
,
V
v e N .
n = 2 . There,
the set de-
as
} = P(2,4)
+ e(4)
3 s~
satisfies
(4) and (5).
Proof It follows
from the fact that lim pi(~__ = + k~,oo
The basic property Proposition
l_
of the sequences
s~
VVV
defined
in (57-62)
is given in:
4
The following
three conditions
*)
s~ ~ Z °
**)
W(s~)(0)
***)
V
J
DP~(o)
are equivalent:
is column wise nonsingular
~ 0 ,
v
p ~ Nn
(see(8))
Proof Taking into account
the definition
and **) are obviously **) and ***).
equivalent.
First) we compute
Dqs~(v)(0)
of Z in §3 as well as Lemma 3, the conditions *) o It only remains to establish the equivalence between
W(s~) (0) . The relation
= (p(~))q+e
D%(0)
,
V
q ¢ Nn
(57) will give easily j
~ ¢ N
)
100
;
\ \
\ \
i,
P
,
;
/
\\
q
F
iIy
•V
•
10
Fig./.,.
15
M4
101
where
e = (1,...,1)
~ Nn
W(s~) (0)
(63)
. Therefore
: A • B
where A
while
B
=
(
(1J(~>))P(CY)+e I x;,cY c N )
is a diagonal
(64)
DP(V)~b(O)
matrix with the diagonal
,
V
~
N
Now, according to Theorem 8, below, ma 4, below, choose
A
it suffices to show that
m ~ N , such that Vo = k ( m )
+
Then, the conditions
elements
is columnwise nonsingular. A
a = £(n,m+l)
satisfies - 1 ~ q
and
gular, only if
E N . We
..... ~q = k ( m + l )
1
*) and **) in (65) are obviously
(63),
~
k(m) ~ ~ . Now, we choose
orem 8, will directly imply ***) in (65). Therefore Now, the relations
Indeed, due to Lem-
(65). Assume given
(641 and Lemma 4, imply that
DP(~)~(0)
~ 0 ,
V v c N
satisfied,
while
(62.2) and The-
A is column wise nonsingular. W(s~)(0)
is column wise nonsin-
WV
And now, the main result of the present section Corollary
1
Zx
~
0
,
x o ~ Rn
V
o
Proof According to Proposition f ~(x)dx = 1 Rn Define
and
a : Rn ÷ R 1
such that
B ~ 0 K
Defining
by
and
=
4, it suffices to show the existence of
DP~(0)
~ 0 ,
~(x I , ... ,xn) = exp (x I +...+ x n) ~ = 1
in a certain neighbourhood
[~ ~(x)S(x)dx
of
such that
and assume
S c D(R n)
0 ¢ Rn . Then
> 0
~ = ~ • S / K , one obtains the required function,
V p ¢ Nn
~ ¢ D(R n)
V p ¢ Nn •
since
DP~(0)
= I/K > 0
WV
In case arbitrary positive powers of the Dirac ~ distributions in the algebras constructed
in the present chapter
chap. 4) one needs the result given in: Corollary
2 n W+
Zx O
~
~
,
V
x ° ~ Rn
are to be defined with-
(see Theorem 4, chap.
i, g8 and
102
Proof Choosing fore
8 c C~(R n)
in the proof of Corollary i, one obtains
oo n @ ~ C+(R )
and there-
VVV
st~ E Z ° n W+
Lemma 4 The
infinite matrix of c o m p l e x
numbers
A = (auo
[ V 0
E N)
is column wise non-
singular, only if V
~,#cN: o c N , Vo ,..., Vo ~ N :
**)
-
6
***)
(65)
%
o
<
o
< v
...
............
o
%
o
o
,
a M o 0
............
o
a.ooG
Proof I t follows easily from the definition in §3
VVV
And now, the theorem on generalized Vandermonde determinants
(for notations,
see §3),
whose present form, as well as proof was offered by R.C. King. Theorem 8 Suppose given Then, for each
nc
N,
n>
a c Nn
1
.
, a > e :
(1 .....
1)
~ Nn
and
£ ~ N , .% > 1
, the
rela-
tion holds (a+p(1))
p(1)
...............
(a+p(1))
p(~)
=--~-l~i~n (a+p ( £ ) ) pC1)
where
p(j)
...............
= (pl(j),...,pn(j))
(a+p(~)) p(£)
,
for
1 < j
< .% .
--~--~ (pi(j))! > 0 l~j~i
103
Remark The Value
on
of
the
depends only on
determinant
n,R,p(l),...,p(R)
and does not depend
a .
Proof Let
US
consider
the determinant
A, = det ((a+p(o))P(')) For
1 S T S R , the r-th
, where
a = (al
For
,...,
u , T < R
column in Al is (a 11+P (1))
if
15
P,(T)
x . . . x (a,+p,Ul
I
p,(r)
an) .
Is-cQ, , consider
the column p(l)p(r)
I where
$&)P(r)
0' = 1 whenever it
We obtain
occurs.
then
() P(T)
(66)
C,(T)
= Cl CT) + i
(-a)p(T)-p(x)
C (p(X)) 1
P(X)
where the sum i Introducting
is taken
for
follows
We shall
1 2 X 5 R such that
ip(x)I
y
l
.
the determinant A2 = det (pi)
it
all
3
from
, where
(66) that
A2 = Al , since
A2 with
now simplify
1 F(h,k)
if
C,(T)
is the -r-th
the help of the function k=O
= h(h-l)...(h-k+l)
which obviously
1 2 o,~ 5 9. ,
satisfies
the conditions
if
kxl
column in
A2 .
F : N xN + N defined
by
104
(67)
F(h,k)
= 0 ~ = ~ h - k + 1 _< 0 * = * h < k
(68)
F(h,h)
= h !
Now, for
1 -< T <- ~ , we define the column (Pl(1)
c30)
.... × .F(Pn(1) , pn(T))
, PIG))
x...
=
F(Pl(~) Then,
• pl(T))
it f o l l o w s
× F(pn(~,) , p n U ) )
that p ( 1 ) q (~)
(69)
C3(T ) = C2('r ) + Z ( - W ~ - - ( - j ) )
,
V
1
<£
jcJ (£)q(~) where (69o!)
the
sum
~
Ji
c
with
(69.2)
q(T) the
2]~e relation (70)
where
is
taken
for
all
{1,2,...,pi(~)-l}
= p(Z)
-
(IJ1E,...,iJn])
number of elements
in
J.
1
u ...
o Jn @ ~
for
1 <_ i
-< n ,
, with
IJi[
denoting
.
(69) can obviously be written under the form C3(T ) = C2(T ) + Z ( - - [ ~ - - ( - j ) ) jeJ
(69.1)
J = J1
•
and (69.2)
are
still
C2 ( X ( T , J 1 , . . . ,
valid
and
V
Jn) ) ,
X(T,J 1 ,...,
Jn)
e N
1 -< T - <
~
,
is uniquely defined
by p(X(z,J 1 ,...,
therefore
1 <- X ( T , J 1 . . . .
Denoting by implies
A3
A3
, Jn)
and
(EJll
.....
IJn[)
/X(T,J 1 ,...,
Jn) l < lp(z)l
C3(I ),...,C3(£ ) , the relation
=
A1
(67) gives for any
--I [-- F(Pi(O) l_
1 -< ~,T <- £
the equivalences
, pi(T)) -- 0 ~=* (3 l-
taking into account (68) and the form of the colturms
the relation
(70)
and thus
Now, the relation
]herefore,
-< ~
-
the determinant with the columns
A3 = A2
(71)
Jn) ) = p(z)
(71) will end the proof
VW
C3(T ) , with
I-
105 §8.
STRONGER RELATIONS CONTAINING
The stronger version of
PRODUCTS WITH DIRAC DISTRIBUTIONS
t he relations
in 2) and 3) in Theorem i, §4, obtained in the
present section is of the following type: Given a locally finite subset family of Dirac distribution mily
derivatives
(Ca I a ~ X) of functions
high order vanish in in§§2-4,
(72)
Z ~a(X-a) aeX
qa c N~ , and a faup to a sufficiently
the relations: • Dqa6a(X)
Dirac algebras
dure in §§2-4.
whose derivatives
0 ¢ R n , one obtains within a subclass of the Dirac algebras
constructed
~le mentioned
(Dqa~ a I a c X) , with
Ca c C~(R n)
X c Rn , a
= 0 ,
x e Rn .
are constructed
through a particularization
Nkmely, the family of Dirac classes
Z Z , with
will be replaced by a smaller family which possesses First, we restrict the representatians
of the proce-
Z c Z 6 , defined in §3,
stronger properties.
of the Dirac 6 distributions
given by
Z~
in
§3. Denote by
g~ V
(73)
the set of all X c Rn ,
(73.1)
X
V
x
V V
o
E
Rn
I x c X)
the condition
is locally finite,
V
v c N ,
of
x° ,
~ ~ N
:
~ ~ p :
Vn l _ _ l xcX\{xo} The analog of Corollaries Proposition
satisfying
:
neighbourhood
~ c N ,
I x c R n) e Z~
locally finite:
(supp Sx(V)
(73.2)
Z = (sx
supp
Sx(~)
i and 2 in
=
7 is obtained in:
S
g~ ~ ~
and there exist
Z = (s x I x ~ R n) c g ~
such that
VxcR
sx c ~/+
n
Proof It results from the proof of Corollary
Now, for
Z
=
(s x I x c R n) e
Z~
2, §7
, denote by
WV
TE
the vector subspace
in
SO
genera-
ted by all the sums Z x~X
Z X Dqs qcN n xq x q4P x
where
X c Rn ,
X
locally finite,
Px c N n
and
~
c C I . One can notice that due xq
106
to (73.1), the definition of
TE
is correct.
And now, the analog of Proposition 2 in §3: Proposition 6 For each ideal
E • Zg , TZ
is a Dirac class which is compatible with the Dirac
I~
Proof First, we prove that
~
is a Dirac class, that is, it satisfies (17.1), (17.2) and
(24) in chap. 2. Indeed, the conditions (17.1) and (24) result easily. Assume now t • TZ , t % O , then (74)
t =
E x•X
E ~ Dqs qcN n xq x q<-Px
where
X c Rn ,
(75)
X
locally finite, nonvoid,
x° • X '
~
qo ~ Nn '
We show that (17.2) is satisfied for (76)
p c N : V
~
Px ¢ Nn
and
lxq ~ CI" Moreover
qo -< Px o : ~ Xoqo :~ 0 x
o
given in (75). Assume, it is false and
~ • N , ~ -> P : t(~)(x o) = 0
then (74) gives (77)
E x¢ X
qEENn %xqDqsx(U)(Xo)
= 0 ,
¥
~ e N , v >-
q-
E h Dqs (U)(Xo) = 0 , q•N n Xoq xo
P
%; ~ • N ,
such that (77) will result in u >-
q-
TE
and
Z~
TE
is a Dirac class. It remains to
are compatible. Since, obviously
TE c TE
and
16 , T E
are
compatible according to Proposition 2, §3, it suffices to prove that (79)
I~ n TE = O .
Assume therefore But
t • I~
t • I ~ n T 7' , t % 0 , then (74) and [75] hold again, since
t • TZ .
and [2) will again give (76), thus the reasoning above, contradicting (75),
will end the proof of (79)
VVV
Based on the above result we proceed to construct the Dirac algebras in which relations of type (72) are valid.
107
Suppose given
E ~ Z~ .
For a n y ideal
I
vector subspace
W , I ~ 16 , compatible vector subspace
in V
in
T
in
SO , T ~ TE
I n V
and vector subspace S 1 in S satisfying (16) and o o (17), one obtains a regularization ( V , T Q S I ) . Then, for any admissible property Q,
one can define Dirac algebras according to (18). The analog of Theorem 1 in §4, stating the validity of (72) within the algebras defined above is obtained in Theorem 9. First, we shall specify within the algebras the meaning of expressions more general, (80)
as in (72), or
of the form
E ~a(X-a) a~X
• ( E h qcN n aq Dq~a)
q~Pa where
X c Rn ,
Suppose
X
locally
H = ( h a [ a e X)
(81)
Y
a c X :
]
Va c R n ' ha = 1
(82)
V
finite, is
Va
of functions
neighbourhood
of
a
and
Xaq ¢ C 1
h a ¢ D(R n)
such that
:
a $ b :
h a o supp h b =
The existence of such families
H
results from the fact that
X
is locally finite.
one can define ~(x)
and then
a family
' Pa c Nn
on V a
a,b c X , supp
Obviously,
~ a ¢ C~(Rn)
=
E ha(X ) • ~a(X-a) aeX
~ e C~(R n)
(83)
T =
Z aCX
. Define
xE
,
W ¢ D,(R n)
R1
by
Z l Dq~ qCN n aq a q<Pa
Lemma 5 Within the algebras provided
that
(19)
Ap , p ~ ~n , the product
~-T
does not depend on
H ,
is valid.
Proof Assume
H' = (h a I a ¢ X)
~' ¢ c~(R n)
is an other family satisfying
by
~ ' (X) =
Y: h a ( X ) • ~Pa(X-a) x£X
,
x
R1
(81) and (82) and define
108
We shall prOve that (84)
~b' • T = @ • T
holds within the algebras relation
%
• P ¢ ~n . I n d e e d ,
(83) above, the inclusion T=
E
Z
a¢X
qCN n
X
TE c T
Dqs
aq
a
assume
E = (s x ] x ¢ Rn)
a n d 3) i n T h e o r e m 2, c h a p .
, then the
1, §8, g i v e s
IQ(v(p),T(~SI)e %
+
q~Pa since
t =
E acX
E q¢N n
Dqs
aq
c 7E
a
and
T =
q-<Pa Therefore,
to • T = u(to) 4''
• t + IQ(v(p),TQS1)¢
T = u(@')"
Ap
IQ(v(p),TQS1)E%
t +
hence (85)
IQ(v(p),T(~SI)
tO" T - ~ • T = u(to'-to) • t +
But, due to (73.2), ¥
x
t
satisfies
c
Ap
the condition
ERn\X: O
(86)
~
V
V
x) ~ N ,
neighbourhood ~) _> p
of
xo ,
~ c N :
:
V n supp t(~) = while, due to (81),
(87)
u(to'-to).t
V
XoCX:
~
U
satisfies
neighbourhood
of
x
the condition
: O
(l~'-to) • t(v) = 0 Now, the relations
U ,
V
~) c N
(86) and (87) will imply
Dr(u(to'-to)-t) therefore,
on
¢ f 6 o V° ,
(19) and (85) will give
V
r ¢ h~
(84)
VVV
Lemma 6 Within the algebras ( E
%
ha(X)'toa(X-a))
aCY
for any
, P ¢ ~n , t h e r e l a t i o n s • ( E
aeY
y c X ,
Y
E
finite, provided
Proof Assume
a,b
¢ Y ,
a + b
and
X
qcNn q~Pa
q ¢ Nn , t h e n
aq
that
hold
Dq6a) =
E
toa(X-a)( E
acY
(19) is valid.
qeNn q~Pa
~ a q Dq6a)
109
(88)
ha(X)
Indeed,
" ~a(X-a)
• DC~b = 0 e A P
assume
~ = (s =< I x c Rn ) , t h e n , and the inclusion Z c T , one obtains (89)
ha(X)
But,
(81)
• ~0a(X-a)
and (82)
(90) taking
into
account
will imply (88)
• Dq6 b = U ( h a )
imply that
Dr(u(ha)
t o 3) i n T h e o r e m 2, c h a p .
• U(~a('-a))
DqSb) • ira n V° ,
1
§8,
IQ(V(p),TQ)S1) • Ap
Dqs b +
b $ supp h a , therefore,
• u(tPa(--a)) that
according
one obtains ¥
s b e Zb . Now, t h e r e l a t i o n s
r • Nn (19)
a n d (90)
together
with
(89)
777
The Lemmas 5 and 6 suggest within the algebras
Ap • p c ~n
the following definition
of the expressions in (80) Z ~a(X-a) • ( Z ~ Dq~a) = ( 2 -~a(X-a)) • ( Z a•X qcN n aq aeX h a ( X ) a•X
(91)
q~Pa where
H = (h a ] a • X)
Z X Dq~a) q•N n aq q~Pa
is any family
of functions
h a ¢ D(R n)
satisfying
(81)
and
(82).
Theorem 9
In case (19) is valid, the following relations hold in the algebras p e ~n : I)
2
~a(X-a) • Dqa~a(X) = 0 • A
a•X for
Ap , with
P ' each
X c Rn ,
X
locally
finite,
qa ~ Nn
and
~ a e ~ ( R n)
satisfying the condition Dr~a(0)
(92)
=
In particular• if 2)
Z
%/ a e X •
0 •
p • Nn
r • Nn
,
r
<
p
or
r
-< q a
"
then
(x-a) r a . Dqa~ (x) = 0 • A
a•X
a
for eack
X c Rn
p "
X
locally finite•
ra • qa • Nn •
ra ~ p
and
ra ~ qa "
Proof i)
Assume h
(93)
a
2 = (s x I x e Rn)
• D(R n) Z aeX
and
H = (h a I a c X)
is a family of functions
satisfying (81) and (82). Then, according to (91) ~a(X_a).
Dqa~a(X ) = ( Z a•X
ha(X)-~a(X_a)).
( Z aeX
Dqa~a(X))
110
We shall p r o v e that the right side of the above r e l a t i o n , d e n o t e d b y in
A
. Indeed,
S ~ D' (Rn)
S , vanishes
, thus, taking into account 3) in T h e o r e m 2, chap.
P §8, and d e n o t i n g v = u ( Z h • ~a(--a)) a£ X a
•
Z Dqas ae X a
one obtains
(94)
S = v + IQ(V(p),TC)Sl) since
But
Z Dqasa • ~ a•X v • V
E Ap ,
c T .
due to the fact that
(92) holds for
r • ~
, r -< _ qa
" Therefore
o
(95)
Dry • V° , A s s u m e now ther w i t h
r ~ N~ .
r ~ Nn , r < p , then
(90) and (5) imply that
Dry ~ /o
w h i c h toge-
(95) and (19) will give
Dry
2)
V
That
relation
shes
in
¢ /
n Vo c V ,
implies
¥
v • V(p)
r
• Nn
, hence
,
r
due to
. (94)
the
expression
A P
It follows f r o m I) taking
T ~ . ~barX~ = x a ,
%/
a • X ,
x • Rn
WV
in
(93)
vani-
I,
C h a p t e r
6
LINEAR INDEP£NDENT FAMILIES OF DIRAC DISTRIBUTIONS
§i.
INTRODUCTION
The representations of the Dirac ~ distribution used in chapters 4 and 5, were given by weakly convergent sequences of smooth functions satisfying a condition of strong local presence (see (6) in chap. 5, §3). A first consequence of that condition was the nonsymmetry of these representations• implying that the Dirac distribution derivatives
D°~
of any order
q e N n , are not invariant within the algebras under the
transformation of coordinates Rn9 {see pct.
x
+
a • xc
Rn ,
2 i n Remark 2, c h a p .
5,
a = -1
,
§5).
In the present chapter the following stronger result is proved within the algebras containing the distributions in
D' (Rn)
: Applying to any given derivative
D q~ ,
q ~ N n, of the Dirac distribution the transformations of coordinates Rn B x one o b t a i n e s
÷
for pair
a " x c Rn ,
ac
wise different
a°
Dq6 ( a o x ) . . . .
, Dq6 ( a m X )
.
In §4,
result
extended to
that
is
include
R1 \
{0}
,
• . . . , am c R I \ { 0 }
also
generalized
, linear
independent
Dirac elements
of the form
lim n~ (ax) a->Co The a b o v e p r o b l e m s arbitrary
§2.
are approached within
coordinate
transformations
a general
within
framework established
the algebras
are
i n §2, w h e r e
studied.
COMPATIBLE ALGEBRAS AND TRANSFORMATIONS
Suppose given G i v e n an i d e a l a vector
Z ~ Z6 I
in
subspace
V
{see chap.
§§2-4).
W , I = I ~ , a compatible in
(i)
VoQTQS
(2)
U c V(p) Q T Q S
it follows that
S,
I
n Y
and a vector
1 ,
V
o
vector
subspace
T
subspace S1
in
S
o
in
So , T = T Z ,
such that
1 = SO
(V,TQSI)
p c ~n •
is a regularization, therefore, one can define for any ad
112
missible p r o p e r t y (3)
Q
the Dirac algebras
AQ(v, T Q S
1 , p)
,
~n
p
denoted for the sake of simplicity b y
A P
It will be assumed throughout (4)
§§2 and 3 that
v = f6 n v 0
Given a m a p p i n g
e : R n ÷ Rn • a e C ~
called transformation,
we define
~ : W ÷ W
by (~)(~)(x)
= s(~)(~(x))
obtaining thus a homomorphism An algebra
A
,
v
s
of the algebra
and the transformation
~
w,
~
~N
•
x
~ R1
,
W .
~ are called compatible,
only if
P
IQ(v(p),T+(~SI )A and AQ(v(p),T+(~SI)~ are In that case,
invariant
of
one can define the algebra h o m o m o r p h i s m
~ : W ÷ W . ~ : A
÷ A P
~(s+IQ(v(p),TGSI)) = A transformation -I
Proposition
~ : Rn ÷ R n : Rn ÷ R n
~(s) +
a e C~
exists and
IQ(v(p),TGSI) , V
is called invertible, -I
given by P s
~
AQ(v(p),TQSI)
only if
E C~
i
An algebra
and an invertible
A
transformation
is an invariant of
AQ (V ( p ) , T Q S ~ )
a
~ are compatible•
only if
: W ÷ W .
Proof The
necessity
is obvious.
Now, the sufficiency.
IQ(v(p),T~SI)~
is an invariant
AQ(v(p),TGSI)
generated by
ant of
a
: W ÷ W
of
V(p)
~
One needs only to show that
IQ(V(p),TGSI)
is the ideal in
due to Lemma 1 below,
it is an invari
: W ÷ W . But,
, therefore,
WV
Lemma 1 If ~ is an invertible of
~
transformation,
: W ÷ W .
Proof Since ~ is invertible•
(s) The relation
~ ( v o) c V°
one obtains
easily
then
V(p)
, with
p E ~n
, are invariant
113
(6)
a(I ~) c I ~
is also valid. satisfies Then
w
V e N But
Indeed,
in
x°
the conditions
xI
satisfy
and
c R n given. We have to show that aw o (2) and (3) in chap. S, §2. Denote x I = a(Xo) .
(2) in chap.
w
and
e N
xI
also satisfy
V
5, §2, thus
(3) in chap.
is a neighbourhood
big enough.
for
v c N
of
(aw)(~)(x o) = w(~)(Xl)
= 0 , for
to
{x o}
and (6) will obviously sy to see that
The result Suppose
a(V) c V 1 . Now,
then also
assume that of
~-i
V n supp
imply
~(V(p))
~(V) c V , since
c V(p)
,
V p ~ An
1 above,
has the property
A
= ~
shrinks
for to
{x I} ,
it will follow that V n supp(~w)(v) The relations
(4) was assumed valid,
the following
transformations.
PM ' only if
of
lhen,
(5)
it is ea-
WV
justifies
is a set of invertible
V1
if
(~w)(~)
V 1 n supp w(v)
v ÷ ~ , and the proof of (6) is completed.
in Proposition
M
such that
big enough,
On the other side,
, when
S, §2, for a certain neighbourhood
x°
~ ÷ ~ . Now, due to the continuity
shrinks
W
x
big enough.
V 1 n supp w(~) = ~
in
w c I~
and
x I . Assume
when
assume
definitions.
We shall
is an invariant
say that a subalgebra
of each
~
A
: W ÷ W , with
~ ~ M . Obviously,
PM
The algebras
is an admissible
property
(see chap.
(3) will be called M-transform
I, §6).
algebras,
only if
Q
is stronger
than
PM "
Corollary
1
An M-transform
algebra
A
and a transformation
c M
are compatible.
P
Proof The subalgebra
AQ(v(p),T
+ SI)
is invariant
of
~ : W ÷ W
since
A
is an N-transP
form algebra and
§3.
LINEAR
Denote by defined by
Theorem
e c M
WV
INDEPENDENT
M
FAMILIES
OF DIRAC DISTRIBUTIONS
the set of invertible aa(X)
= ax
,
transformations
a
: R n ÷ R n , with
a
~
RI\{0}
V x e Rn .
1
The Dirac distribution
derivative
q ~ N n , are linear independent
transforms
within
the
Dq~(aoX),... ,DqG (amX) Mo-transform
algebras
with given Ap , p ~ A n ,
,
114
provided
that
m <- Ipl
and
a ° ,..., a m ¢ RI\{0}
are pair wise different.
Proof Assume•
k ° ,.. ., k m c C I
it is false and
(7)
koDq6(aoX)
(8)
]
are such that
+ ... + kmDq@(amX)
= 0 ~ A P
0 ~ i ~ m : k. $ 0 i
Assume
E = (s x
TZ c T
give
I x ¢ R n)
then 3) in Theorem
Dq6 = Dqs ° + I Q ( v ( p ) , T Q S I )
2, chap.
I, §8, and the inclusion
( Ap '
therefore kiDq@ (aix) = ki~ai Dqs ° + I Q ( V ( p ) , T G S I )
(9)
since
A
p
is an
(i0)
then
v =
(7) and
v =
with
vj ¢ V(p)
chap.
E o~i<_m
and
wj ~ A Q ( v ( p ) , T G S I ) (4) above,
re
Nn ,
ueN
(12) V
~¢
0 _< i _< m ,
hence
v. • w. J J
S, §2, one obtains V
¥
Denote
v c IQ(v(p),TQSI)
~ o~j~h
into account
algebra.
,
k.~ Dqs • ai o
(9) imply
(ii)
Taking
M -transform o
¢ %
.
as well as condition
(2) in the definition
of
16
in
from (II) the relation r
:
~)>U
:
: N,
Drv(~) (0) = 0 Since
m-<
Ipl
, the relations
IL o
(13)
(i0) and (12) will give
ki(ai )Irl]
D q+r So(V)(0)
= 0 ,
V
r e Nn , Irl _< m
for a suitable But
(14)
so
r < p , •
~ e N ,
U' e N .
Z o ) therefore V
re
@
V ¢ N ,
Nn ,
Drs
(V)(0)
oe
N:
V >- g
:
~ 0
O
since the matrix Now, the relations
W(s o) 60] is column wise nonsingular (13) and (14) result
in
(see chap.
5, §3).
~ >- U'
)
115
ki(ai )£ = 0 ,
V
Z c N ,
Z ~ ]pl
o~i~m Since
m ~ Ipl
and
a° ,..., am
are pair wise different,
Vandermonde determinant will imply
the known property of the
k ° = ... = k m = 0 , contradicting
(8)
W?
Corollary 2
§4.
The family
(Dq6 (ax)
with given
q e N n , is linear independent within the
GENERALIZED
Within
I a c RI\{0})
derivative
Mo-transform
transforms algebras
DIRAC ELEMENTS
D' (Rn) , the operation
(15)
lim an~ S a_+Oo a
has a meaning for certain distributions K = D/n f(x)dx thus
of Dirac distribution
, then
(iS) gives
(15) will give
an ~
lira a+oo
algebras
S = f c LI(R n)
S = ~ , one obtains
due to Corollary
A class of algebras,
anea S = S
,
x c Rn ,
2 in §3.
similar to the ones used in §§2 and 3 will be constructed
section and it will be shown in Theorem 2 that within those algebras, (16) exists and it is different of The mentioned
I~
w(~)
a e RI\{0}
the limit in
. I~
defined in chap. S
condition
vanishes outside of a bounded subset of
~) e N
R n , provided
that
is big enough.
It is easy to see that
PropOsition
, with
in this
, consisting of all the sequences of smooth functions
which satisfy the additional
(17)
an6(ax)
algebras are constructed by replacing the ideal
§2, with the smaller one w c I~
and
Ap , p c ~n , Ipl = ~ , the problem of the limit
= lira an~(ax) a_>=o
a
becomes nontrivial
K~ . In case
if
S = ~ .
Within the M - t r a n s f o r m (16)
S . For instance,
16
is indeed an ideal in
W , actually a Dirac ideal.
2
For each
E e Z6 , the Dirac class
TE
and the Dirac ideal
I~
are compatible.
Proof It follows from the inclusion
I~ c 16
and Proposition
2 in chap. 5, §3
VW
116 Suppose now given
Z ~ Z6 .
Given an ideal
in
I
a vector subspace
W , I = 16
V
in
I n V°
(IS)
VoQTGs 1 :
(19)
U c V(p)QTQS
it follows
ble property
Q
vector subspace
and a vector subspace
S1
in
T
in
So
So , T = TE ,
such that
SO
1
(V,TQSI)
that
, a compatible
V
p c An
is a regularization,
thus,
one can define for any admissi
the Dirac algebras
AQ(v,TQS 1 ,p] , p c ~n ,
(20) denoted
for simplicity by
A P
We shall assume in the sequel that V = I~ n V °
(21)
Within the above algebras
~
, p c ~n , the limit in (16) will be obtained
as given
b y a relation (22) with b
lim t c W
an6(ax)
IQ(v(p),T(~SI)~ c PA
= t +
, t(v) = ( b ) n s(v) ( b x)
> 0 , limb
V w ~ N , x ~ R n , where
,
The entities of type
(22) will be called generalized
Using a p r o o f similar to the ones in Proposition Proposition
A
and an invertible
AQ(v (p) ,T Q S ~ )
given
homomorphism
is an invariant
b = (b v I v c N)
transformation of
, with
a
: W + W
b w c RI\{0}
Ibvl n w(v)(b x)
We shall only be interested
~ are compatible,
only if
•
• Then,
one can define the algebra
,
V
w ~ W ,
~ ~ N ,
x ~ Rn .
in the case when
lim b~ = +
Since in the definition a : Rn ÷ R n
definition
of compatibility
between
given in §2, the transformation
ra h o m o m o r p h i s m ned b e t w e e n
besides
Dirac elements.
1 and Lemma 1 in §2, one obtains:
~b : W + W , where (c~w) (v) (x) =
(23)
and
3
An algebra
Suppose
s c go
=
~ : W ÷ W , it follows
the algebras
A
of M - t r a n s f o r m
transformations
an algebra A and a transformation P appears only through the generated algeb-
that the compatibility
and algebra homomorphisms
P algebras
given in §2, will
also algebra homomorphisms
of
of
has actually been defi-
W . In the same way, the
still be correct W .
if
M
contains
117 Proposition 4 The algebra
c~
and the algebra homomorphism
A
AQ ( Y ( p ) , T Q S 1
is an invariant of
p
are
compatible, only if
0% .
Proof S e e the proof of Proposition i, §2 and the following:
Lemma 2 V(p) , with
Nn
p E
, are invariant of
~b "
Proof Assume
(24)
lim by = + = 12+oo
First we prove
(25)
% 4 % ) c f6
Indeed, assume
w E I~
and
quence of smooth functions
x ° E Rn ~b w
given. First, we notice that due to (24), the se
will obviously satisfy (17), since
that condition. We shall now show that (3) in chap. S, §2. Indeed, in case (24), while for
x° = 0
~b w
satisfies in
xo
w E f~
satisfies
the conditions (2) and
x ° ~ 0 , the two conditions result easily from
they are obvious.
Now, we prove that (26)
~b4V)
Assume,
indeed
show that
(27)
c V
v E V , then
~b v E V°
lira [
, which
v ~ I6 is
n Vo
therefore,
equivalent
due to
425),
it
will
suffice
to
to proving
v4~)4x)~(x/bv)dx = 0 ,
V
~ E D(R n)
First, we notice that 428) for such
supp v(~) suitable that
K c Rn X = 1
429)
on
lira [ %~+0o
But, for any
c K , , K
v E N ,
bounded
K . Due t o
and (28),
V >-
~ E N , since the
relation
v(v)(x)x(x)~(x/b )dx = 0 ,
~ ¢ D(R n) , the sequence
V x E Rn , is convergent in
vEY=f~nVo
V
c yo
D(R n)
to
(~
427)
. A s s u m e now
will
hold
only
X ~ D(Rn)
if
~ E D(R n)
I v E N)
~(0)
hence, the sequence
V
v ¢ I~
with
~(x)
X • Therefore,
(v(~)) [ ~ ¢ N)
= X(X) ° ~4x/bv)
converges in
Thus, the relation (26) is proved. It follows then easily that
,
(29) holds, since N'(R n) to
0
118
~b(V(p))c V(p) The c a s e
when
lira b
= -~
,
V
p ( ~n .
is similar
WV
k>+¢o
Suppose given
b = (b~ ] ~ c N) , with
= M o u {~b }
where
Mo
by c RI\{0}
(23) and denote
was defined in §3.
As noticed above, one can define Mb-transform to Propositions
, satisfying
algebras
A p • with p c ~n . According 3 and 4, these algebras will be compatible with any B ~ M b •
And now, the result concerning generalized Dirac elements. Theorem 2 The Dirac distribution derivative
transforms
neralized Dirac element derivative
Dq~b6(X)
Dq6(aoX),...,Dq6(a~nX) , with given
and the ge
q E Nn • are linear
independent within the Mb-transform algebras Ap , p c ~n , provided that R 1 \ {0} are pair wise different. and a° ,..., a m c
m ~ Ip]
Proof Assume,
k ° ,. • ., k m ,
it is false and
(30)
koDq~(aoX)
(31)
k = 0
~
k c CI
+ -.- + kmDq~(amX) ~
Z = (s x l x c R n)
TZ c T
imply
(32)
is an ~b-transform kiDqS(aix)
= 0 ~ Ap
then 3) in Theorem 2, chap. I, §8, and the inclusion
~ Ap
Dq8 = Dqs ° + I Q ( v ( p ) , T Q S I ) Ap
+ k~bDq6(x)
0 ~ i ~ m : ki $ 0
Assume
Since
are such that
algebra,
it follows that
= ki~a. Dqs ° + I Q ( v ( p ) , T Q S I )
c Ap ,
V
i
(33)
kDq~bs(X)
= k • t + I Q ( v ( p ) , T Q S I) ~ Ap
where (34)
t(~)(x) = Lb~l n Dqso(~)(b~x)
,
V
~ ~ N ,
Denote
(35)
v =
then (30),
Z k.~ Dqs + k • t 1 a. o o_
(32) and (33) imply
(36)
v =
with
vj. ~ V(p)
v c I Q ( v ( p ) , T Q S I)
Z v. • w. o<j
w.j c A Q ( v ( p ) , T Q S I )
•
thus
x ~ Rn •
0 ~ i ~ m ,
119
Taking into account chap.
(21) above as well as condition
5, §2, the relation
(2)
in the definition
of
I~
in
(36) will result in
V
r~
Nn ,
r
:
V
~e
N
~>-~
:
(37) Dry(v) (0) = 0 But
m ~ Ip[
, therefore
(35) and (37) will give
Irl
Z ki(ai) o~i~m
(38)
+
klbv[ n + l r l )
V r e # for a suitable Now,
that
Z ki(ai )~ + klb~ o~i~m
must v a n i s h s i n c e
a ° ,..., am
terminants
Corollary
£ ~ N ,
~ =
0 ,
are constant
(31)
~o
~)c
N,
O->~
' ,
V
£ e N ,
k ° = ... = k m = 0
~ < m _
(38) results
in
% -< m
''''' ~E e
N ,
in (39). Therefore,
above can be a r b i t r a r y .
are pair wise different,
will imply
will contradict
V
o c N , one can find suitable
Z ki(ai )~ o~i~m Since
Irl ~ m ,
r ~ p ,
In+£ = 0 ,
k ° ,..., k m , k , a ° ,..., a m
k
0
(14) in the proof of T h e o r e m 1 in §3, the relation
where for any given But
=
~' ~ N .
according to
(39)
,
Dq+rso (V) (O)
Then,
~o ''''' ~
~ o
(23) will imply
(39) becomes
.
the known property of the Vandermonde
which together with
k = 0
obtained
de-
above,
VVV
3
The family
(Dq6(ax)
I a e RI\{0})
together with the generalized
Dirac element derivative
q e N n , are linear independent p e N n , ]pL = ~
of Dirac distribution
within the M b - t r a n s f o r m
derivative
~b Dq
(x)
algebras
transforms
with given Alp ,
C h a p t e r SUPPORT,
§i.
7
LOCAL PROPERTIES
INTRODUCTION
An important property of the distribution [II], [61],
[66-69], V
multiplication,
[783, which can be formulated
S,S',T,T' •)
¢ D'(R n)
S • T ,
• *)
S = S'
,
E c Rn •
S' • T' ,
T = T'
on
will be proved valid for the multiplication butions.
E # ~ ,
exist in
namely
D'CRn)
E
open ]
: ~
S • T = S' • T'
as well as several
on
E
J
within the algebras
An extension of the notion of support of a distribution
ing the above,
its local character,
as follows
other local properties
containing
the distri-
is used in establish-
of the elements
in the algeb-
ras.
§2.
THE EXTENDED NOTION OF SUPPORT
Suppose
P
is an admissible property,
sible property, Given
such that
S c AQ(v,S ',p)
Q -< P
and
*) **)
for a certain
~eN S
S
vanishes
on
E , only if
S = s + IQ(v(p),S ') s(x)) = 0
on
E , with
v ~ N ,
~) >-I/
,
. will be the closed subset
of
(2)
supp S = R n \ {x c R n
Proposition
! in
C°~(Rn )
I S
vanishes
on a n e i g h b o u r h o o d
Proof from (20.2)
of
x}
the above notion of support is identical
usual one.
It results
Q is an admis-
:
The support
For functions
is a P-regularization,
p ~ NP .
E c R n , we say that
s c AQ(v(p),S') (I)
and
(V,S')
in §7 and Theorem 2 in §8, chap.
i
WV
.
with the
121
The local character of the multiplication the distributions
is established
and addition within the algebras containing
in the following two theorems.
Theorem 1 If
S,T c AQ(v,S',p)
, then
I)
supp (S + T) c supp S u supp T
2)
supp
(S • T) c supp S n supp T
Proof It follows from a direct verification
Given
S,S' c AQ(v,S',p)
vanishes
on
and
of the definitions
E c Rn , we say that
WV
S = S'
on
E , only if
S - S'
E .
Theorem 2 Suppose on
S,S',T,T'
E AQ(v,S',p)
and
Ec
R n . If
S = S'
on
E
and
T = T
E , then
I)
S + T = S' + T'
on
E
2)
S • T = S' • T'
on
E
Proof It follows directly form the definitions
An important
WV
feature of the above notion of support is pointed out in the case of the
algebras where the Dirac ~ function is represented smooth functions
satisfying
by weakly convergent
the condition of strong local presence
sequences of
(see chap. 5, §3
and also chap. 4, §6). Theorem 3 In the case of the algebras constructed any order
q ~ N n of the Dirac ~ distribution,
l)
s u p p Dq~ = (0}
2)
Dq~
v a n i s h e s on any
3)
Dq6
d o e s n o t v a n i s h on
c h a p . 5,
*)
cl
E
E c Rn
such that
possess the properties:
0 + cl
E
*)
Rn\(o} , provided t h a t the c o n d i t i o n
§4 i s v a l i d .
is the topological
closure of
Dq6
in chapter 5, the derivatives
E c Rn
(19) i n
of
122
4)
Dq6
does not vanish on
chap.
{0} , provided
that the condition
(27) in
5, §4 is valid.
Proof Assume
Z = (s x [ x ~ R n)
TZ c T
gives
(3)
for
Dq6
3)
We notice
C4) be
I, §8, and the inclusion
I) •
A Q ( v , 2 ~ ) S 1 ,Pl •
(5) in chap.
p c ~n
5, §3.
that in any representation
Dq 6 = t will
easily from
2, chap.
representation
Dq6 = Dqs o + I Q ( v ( P ) , T Q S
Then i) and 2) result
t
, then 3) in Theorem
the following
IQ(v(p),TQS1) c AQ(v,T(~S 1
+
a sequence
of
smooth
functions.
,p)
Therefore,
p •
,
if
Dq6
y~n , vanishes
on
Rn\{0}
,
Thus,
due
then
(5)
t(V)
for certain
= 0
on
Rn
V
~) c N
U • N . But (5) obviously
V >
implies
Drt e 16 n V
V r e Nn O
to (191 in chap.
5, §4, one obtains
Dq6 = 0 • A Q ( v , T Q S 4)
Assume,
(6)
I)
t(~)(o)
for certain
V p • ~n
,
V p ~ ~n
, contradicting
'
. Now,
(4) will
11 in Theorem
a representation
imply
i, chap.
5, §4.
(4) with
such that =
o
,
v
~
c N
,
v
~'
>
U' e N . Denote
(7)
v = t - Dqs
O
(3) and (4) imply v =
with
,
it is false and there exists
t • AQ(v(p),TQS
then
1 ,p)
t • Y(p)
Z o_<j<_m
v •
IQ(v(p),TQS I)
therefore
v. • w. J 3
vi • V(p I , w i • A Q ( v ( p ) , T Q S I )
. Now, the condition
(27) in chap.
5, §4, will
give v. c V(p) c V c 16 ] which together
(2) in chap.
vj(~)(0)
(8)
for certain Now,
with
(6),
dition
=
0
,
V
V
0 _< j -< m
5, §2 results 0 -< j -< m
,
~
in •
N
,
v > U"
U" ~ N .
(81 and (71 imply
(91 with
,
11 ~ N
Dqso(~)(0)
= 0 ,
V
suitably
chosen.
But,
(61 in chap.
5, §3
WV
9 c N ,
~ >-
s o • Z o ' while
C9) ohviously
contradicts
the con
123 Remark 1 The p r o p e r t y
i n 4 ) , Theorem 3, t h a t t h e D i r a c d i s t r i b u t i o n
x ° c R n , q ¢ N n , do not vanish on strong local presence The
and it is proper
'delta s e q u e n c e s '
generally used,
137], [ 1 6 2 ] , do n o t n e c e s s a r i l y
A characterization
{x o}
sequences
Dq~
[4],
[35-41],
[53],
prevent the vanishing of
[68-69], Dq6
X
in algebras
of smooth functions
, with
X
of the con__tion°o_g~#
for the algebras used in chapters
of the support of the elements
of the representing
derivatives
is a consequence
is presented
4, 5 and 6.
[105-110],
on
[136-
{x } . O
O
in terms of the supports in:
Theorem 4 Suppose
S ~ AQ(v,S',p)
supp S = n cl
lim supp s(v)
where the intersection
(10)
then
is taken over all the representations
S = s + I Q ( V ( p ) , S ')
AQ(F,S ' , p )
with
s e A Q [ v ( p ) , S ')
Proof The inclusion c
. Assume
s
V n supp s(V) = ~ , for a certain neighbourhood Conversely, such that
assume S
V
V of
~ c N , x
and
vanishes
on
. Then
v ~ ~ ,
~ ~ N . Now, obviously
x e Rn\supp S , then there exists
V n lim supp s(v) = ~ v+oo
x % supp S .
an open n e i g h b o u r h o o d
V . Hence, one can obtain a representation
V
of
in chap.
6, §4, an additional
result on the
can be obtained.
Theorem 5 Suppose given S = sI +
S e
AQ(v,TQS 1
IQ(v(p),TQS I)
Then, the subsets
in
lim supp Sl(~)
,p)
and two representations
= s2 + IQ(v(p),TQS
~
Proof sI - s2 e
,P)
•
supp s2(~)
differ in at most a finite number of points,
Obviously
I) e AQ(v,TQS 1
Rn ,
IQ(F(p),TQS I) , hence
provided
that
x ,
(i0), such that
WV
In the case of the algebras constructed support
x e R n \ cl lim supp s(V) v+oo
given in (i0) and
V c 16 n V ° .
124
(8)
SI - S2 =
with
vj c V(p)
~ V. " w. o~j~m ~ J
, wj ~ A Q ( v ( p ) , T Q S I )
vj ~ V(p) c V c I~ therefore Then,
vj
, with
due to (8),
Corollary
,
. Now,
V
in chap.
S, §4 implies
0 ~ j ~ m
0 ~ j ~ m , satisfy
Sl - ~2
(27)
(3) in chap.
will also satisfy
5, §2 and (17) in chap.
those two conditions
6, §4.
VVV
I
Under the conditions Sin= 0 for certain
in Theorem
5, if
AQ(v , T Q S 1 , P) m e N , m e i , then
supp S
is a finite
subset of points
in
Rn .
Proof Assume
given
a representation
S = s + IQ(V(p),T~SI)~_~
, with
s c AQ(V(p),T~SI),~
then S m = s m + IQ(v(p),TQSI) s m ~ IQ(v(p),TQSI)
therefore
Using an argument s
m
~ I~
similam
therefore
.
to the one in the proof of Theorem
lim supp sm(w) v+oo
supp s(w) = supp sm(~)
,
= 0 c A Q ( v , T Q S 1 , p)
V v ~ N
is a finite
5, it follows
subset of points
in
Rn
that Finally,
VVV
Remark 2 In the case of the algebras
constructed
lary 1 will still be valid, provided
§3.
5, the results is replaced
in Theorem
by locally
5 and Corol-
finite.
LOCALIZATION
Given S
in chap.
that finite
S ~ AQ(v,S',p)
vanishes
on
denote by
ES
the set of all open subsets
E c Rn
such that
E .
The relation ~-J
E
=
R n \ supp S
EcE S is obvious.
In case
stributions
are used,
S e D'(R n)
and the usual notions
the corresponding
set
E$
of vanishing
and support
has the known property
for di-
that any union
125
of sets in
ES
is again a set in
ES .
In particular R n \ supp S = ~ J E e ES E•E S A first p r o b l e m approached ty to the algebras the structure
of
in the present
containing ES
section is the extension of the above proper
the distributions.
In this respect,
several results on
will be given.
Theorem 6 Suppose
and
S • AQ(v,S',p)
EI , E~ • E S . I f
d ( E l k E 2 , E2\E 1) > 0
*)
then
E 1 u E2 • E S .
Proof Assume
such that
s I • s 2 • AQ(V(p),S ')
(9)
S = s i + IQ(v(p),S ') • AQ(v,s',p)
(i0)
si(v ) = 0
for certain
(n)
on
Ei ,
> • N . Denote
V
1 < i < 2 ,
one obtains
(12)
for
v(V) (x) =
According
~ • N ,
v ~ N ,
~ _> ~ ,
~ = 1/2
on
~ ->
if
x • Rn \ ( ~
s 1 (v) (x)
if
x • E2 \ E1
- s 2 ( V ) (x)
if
x • E1 \ E2
0
if
x • E1 n E2
to Lemma 1 below,
thus denoting
- s2(, )(x)
e C = ( R n)
there exists
~ / E 2 . Denote
s = (Sl+S2)/2
+ w
w = u(~)
S = s + IQ(v(p),S '3 • AQ(v,S',p)
=
0
on
(14)
s(~)
Now,
(13) and (14) will
E1 u E2 , imply
, such that
• v , then
, the relation
But, due to (12), it follows obviously
*)
1 < i -< 2 ,
v = s I - s 2 , then (9) implies
Sl(X; ) ( x )
(13)
V
v • IQ(v(p),S ')
Further,
and
,
(ii) implies
(9) will give
,
s ~ AQ(v,S',p)
that V
~
E1 u E2 • E S
~ N
,
~
~
VVV
d( , ) is the Euclidian distance on R n and Rn d(E,F) = inf {d(x,y) [ x • E , y 6 F) for E,F c
u E 2)
= -1/2
on
E2\E 1
w c IQ(v(p),S ') ,
126 Lemma i Suppose
F c G c Rn
such that
d(F,Rn~G) > 0 ,
then there exists
t~ e U°°(Rn)
with the properties I)
0 ~ ~ ~ 1
on
Rn
2)
~ = 1
on
F
3)
~ = 0
on
Rn\G
4)
@ ~ D(R n)
if
F
bounded
Proof Define
X : Rn
(0,~) ÷ R I
x
K ×(x,s)
by " exp (e2/(iixll2-e2))
0 where
if
llxl]
< e
if
llxll
~ ¢
=
Ks = 1 /
exp
(c2/(llx[12-e2))dx
llxl <~ Assume
0 < E < d(F,Rn\G)/2 ~(x)
=
J
and define
~ : Rn + R I
by
X(X-y,e)dy
F(~) where
F(¢)
= {y ¢ Rn [ d ( y , F )
It can easily
Corollary
be seen that
~ e}
~ is the required
function
WV
2
Suppose
S c AQ(v,S',p)
i)
cl E n cl F =
2)
F
then
E u F c ES
and
E,F E E S . If
bounded
PrOof The
sets
Theorem
E
and
F
satisfy
the conditions
in Theorem
6
VW
7
Suppose then
S ~
AQ(V,S',p)
E[ u E 2 ~ E S .
and
E 1 , E 2 E E S . If
E~ c E 1
and
d(Ei
, Rn\E1 ) > 0
127
Proof We shall use the notations According
in the proof of Theorem 6.
to Lemma I, there exists
= 1/2
on
(15) therefore,
Corollary
it can be seen that the relation
E i . Then, s(~) = 0
such that
~ e Cm[R n]
on
the relation
Ei u E2 , (13) will
V
~ ~ N ,
imply
~ = -1/2
on
E2/E 1
and
(14) becomes
~ a p ,
E i u E2 e E S
VW
3
Suppose
S c AQ(v,S',p)
i)
cl E n supp S =
2)
E
then
E ~ ES
and
E c Rn \ s u p p
S ,
E
open.
If
bounded
Proof Assume
K c R n \ supp S ,
x ° .... , x m ¢ K
(16) If
K c m = 0
Assume
[ J o~i~m
K = E . It follows
then
pair wise different, B(xi
from
(2) that
*)
such that
, ¢x./2) i
E c K c B(Xo
, gx /2) o
and the proof is completed.
m = 1 . Denote E1 = B(Xl
then
compact,
x ~ K : ~ ex > 0 : B ( x , ex ) c E S
V Assume
K
E1 , E2
and
' gXl ) '
E2 = B(x°
fulfil
the conditions
E~
and due to (16) the proof Assume
' gXo ) ,
E i = B(x 1 , eXl / 2 )
in Theorem
7, therefore
~
o E2 e ES
is again completed.
m = 2 . Denote E 1 = B ( x 2 , ex2 )
,
E2 = B(x °
, eXo ) u B(x 1 , eXl / 2 )
,
E~ = B ( x 2 , ¢x2 / 2 ) then
S
fulfil
vanishes
on
the conditions
is completed
B(x,e)
and as seen
in Theorem
above,
7, therefore
also
on
E2 . Moreover,
Ei u E2 ~ ES
again.
The a b o v e p r o c e d u r e
*)
E1
= { y ~ Rn
can be used
for
I ][y-xl[
< e}
any
for
m ¢ N , m > 3
x
~ Rn
,
VW
e > 0
E 1 , E2
and
E~
and due to (16) the proof
128 Corollary 4 Suppose
(17)
S ~ AQ(_v,s ' ,p]
V
Kc
~
E £ ES :
Rn\
H ,
and K
Hc
Rn
then
supp S c H
only if
compact:
KcE
Proof Assume (17) and x % supp S
x c Rn\H
since
E
into account that
then,
x c E
for a certain
E c E S . Therefore
is open. The converse results directly from Corollary 3 taking
supp S
is closed
W7
Corollary 5 Suppose V
S c AQ(v,S',p) E c Rn ,
E
, then
supp S = $ , only if
open, bounded:
E e ES
Proof It follows directly from Corollary 3
VW
Theorem 8 Suppose
S c AQ(v,s ' ,p)
,[ V
and
F c Rn ,
$ E D(Rn\F) n:
F
closed. Then
] supp S c F
$.
o ~ A~(V,S ' , p )
s:
In the case of sectional algebras (see chap. i, §6) the converse implication is also valid.
Proof Assume
x ( Rn\F • Since
and a neighbourhood
V
$ • S = 0 (AQ(v,S',p) (18)
F of
is closed, it follows that there exists x
such that
@ = 1
on
, therefore, given any representation
S = s + IQ(v(p),S
')
(AQ(v,s',p)
• with
one obtains
(193
u($) • s c IQ(v(p),S ')
But~ (18) and (19) imply (203
S = 11(1-$) • s + IQ(v(p),S') c AQ(v,S',p)
$ (D(RnXF)
V . Then, due to the hypothesis
s c AQ(v(p),S ')
129
Denoting imply
t = u(1-~)
x ~ supp S
• s , it follows that
t(v) = 0
on
V ,
and the first part of T h e o r e m 8 is proved.
Assume now, in the case of sectional algebras the inclusion • D(Rn\F)
W ~ ¢ N . Then (20) will
. Since
supp ~
is compact,
supp S c F
and
Corollary 4 implies the existence of
E ¢ ES
such that (21)
supp ~ c E
Due to the fact that
S
vanishes on
E , one can assume that
s
in (18) satisfies the
condition (22)
s(w) = 0
for c e r t a i n me
~ = 0
imply
on
E ,
p e N . Now, the p r e s e n c e of sectional in (22). Then,
Two d e c o m p o s i t i o n r u l e s supports
algebras makes" it p o s s i b l e to assu-
(21) and (22) will result in
~ • S = 0 • AQ(v,S',p)
of their
~ m P ,
w ¢ N ,
V
u(~)
• s c 0
hence
(18) will
VW
for the elements of the algebras,
corresponding
to components
a r e g i v e n now.
Theorem 9 In t h e c a s e o f s e c t i o n a l with
F
closed,
K
algebras
suppose
compact and
S e AQ(v,S',p)
. If
supp S = F u K
F n K = ~ , then the decomposition holds
S = SF + SK for certain
S F , S K ¢ AQ(v,S',p)
I)
supp S F n supp S K =
2)
K n supp S F =
3)
F n s u p p SK = ~
and
satisfying the c o n d i t i o n s
s u p p SK
compact
Proof Assume
G1 , G2 , G3 , G4 c R n
cl G 3 c G4 • K1
cl G 4 n F = ~
is compact and
such that
such that
and
G4
K c GI ,
is bounded.
cl G 1 c G 2 •
Denote
cl G 2 a G 3 ,
K 1 = (cl G4) \ G 1 , then
K 1 n supp S = ¢ . A c c o r d i n g to C o r o l l a r y 4, there exists
E eE S
K1 c E . Then, for a certain r e p r e s e n t a t i o n
S = s + IQ(V(p),S ')
e AQ(v,S',p)
s(v) = 0
V
,
with
s e AQ(v(p),S ')
,
one obtains
w i t h suitably chosen
on
E ,
on
Rn\G4 )
~ >- p
~ ¢ N . But the case of sectional algebras allows t h e choice of
H = 0 . Now, Lemma 1 g r a n t s ~F -- 1
~ ¢ N ,
~F = 0
the existence on
c l G3 ,
of
~F ~ C~(Rn)
~K = 1
on
c l G1
and and
~K ~ D(Rn) OK = 0
such that on
Rn\G2 •
130
Then, obviously S K : U(~K)
u(~ F + ~ K ) • s = s . Defining
• s + IQ[v(p),S')
and 3) will be satisfied
In the one dimensional First,
S = SF + S K
, one obtains
• s + IQ(v(p),S ')
and the properties
and i), 2)
VVV
case
n = 1 , a stronger decomposition
a notion of sepamation
for pairs of subsets
are called finitely separated, vering
S F = U(@F)
in
can be obtained.
R 1 . Two subsets
F , L c R1
only if there exists a finite number of intervals
co-
F u L (-~
, CO )
such that no interval
•
(C O
,
C1
contains
.....
)
elements
(c m , ~
F
of both
vals do not contain elements of the same set
)
F
,
and
or
me L
N, while successive
inter-
L .
T h e o r e m 10 In the case of sectional with
F , L
disjoint,
algebras
S ~ AQ(v,S',p)
suppose
closed and finitely
separated,
. If
supp S : F u L
then the decomposition
holds S = SF + SL for certain
satisfying
S F , S L ~ AQ(v,S',p)
I)
supp S F n supp S L =
2)
L n supp S F = F n supp S L =
the conditions
Proof F , L
Since
are closed, F u L c
there exists
(-~,Co-e)
u
g > 0
(Co+e,Cl-e)
such that
u ... u (Cm+e,~)
Denote K = ~ J [ci-~,ci+e] o_<"
K
compact
such that
and
K c R 1 \ supp S . According to Corollary
4, there exists
E E ES
K c E . Assume a representation S = s + IQ(v(p),S ') ~ AQ(v,s',p)
,
with
s e AQ(v(p),S ')
such that s(V) = 0 for certain gebras.
on
E ,
V
~ c N ,
D e N . But, one can assume
Further,
~ ->
D = 0 , due to the presence
Lemma 1 implies the existence of
~o = 1
on (-oO,Co-e/21 ,
~i = 1 ....
~o
''" "' ~m+l c C°°(RI)
on [ c i + ~ / 2 , c 2 - e / 2 1 , ~el
: 1
of sectional
on
[c + e / 2 , °~]
...
al-
such that
131
and
~o
= 0
on [c o-~/3,+ m)
'
~i = on (-~, Cl+E/3] o [c2-~/3 , m) ....
~m+l = 0
on
(-~,
, ...
Cm+~/3]
Denote J
o = (-~'
Co]
'
J1 = [Co
'Cl]
.....
Jm+l
= [Cm ' ~ )
and define ~F =
It
can
easily
Z ~i ogi~m+l FnJ.= ~ I
be
seen
that
and
S L = U(~L)
§4.
THE EQUIVALENCE
If
S = 0 c AQ(v,S',p)
~L =
'
Z ~i l~i~n+l LnJ.= i
u(~ F + ~L)S = s • Defining
S F = U(~F)
• s + IQ(v(p),S ') , the required properties
BETWEEN
S = 0
supp S = ~ . A result on the converse
then, obviously
WV
will follow easily
=~
supp s
AND
• s + IQ(v(p),S ')
impli-
cation is given in: Cor011ary 6 In the case of sectional i)
supp S =
2)
S
then
S = 0 e AQ(v,S',p)
vanishes
algebras
S e AQ(v,s ',p)
suppose
outside of a bounded
subset of
. If
Rn ,
.
Proof Assume
B a Rn ,
B
bounded,
such that
vanishes
S
on
R n \ B . Then,
there exists a
representation S = s + IQ(v(p),S t] c AQ(v,s ',p)
,
with
s ~ AQ(v(p),S ') ,
such that s(~) for a certain Assume (23) But
= 0
on
Rn \
B ,
V
~
~
N ,
v e ~
p c N . Due to the presence of sectional
~ ~ D ( R n)
such that
~ = 1
on
, algebras,
B , then obviously
u(~)
one can assume P = 0 • s = s , that is
~ • S = S E AQ(v,S',p) supp ~ n supp S = ~
and
• S = 0 e A Q • The relation
supp S
is closed,
(23) completes
therefore
the p r o o f
WV
Theorem 8 will imply
Chapter
8
NECESSARY STRUCTURE OF THE DISTRIBUTION MULTIPLICATIONS
§i.
INTRODUCTION
The algebras containing completions A
the distributions
of the smooth functions
is a subalgebra in
That construction
on
in
Rn
W = N ÷ C~{R n) , while
functions,
noti
their connection with certain zero-filters I
A
, where
.
itself is a subalgebra in
an essential
feature of the ideals
generated by subsets of
specific connections
in the case of the quotient
the distributions.
between
N
×R n
I
is
on which
ideals and zero-filters
constructions
will be
giving the algebras containing
In that way, an information on the necessary
stribution multiplications
§2.
is an ideal in
as sequential A/I
vanish.
In the present chapter, established
I
W
C ° ( N × R n , C I) . As known in that context,
in
were constructed
can naturally be placed within the framework of the theory of alge-
bras of continuous
the functions
D'(R n)
and resulted as quotients
structure of the di-
will be obtained.
ZERO SETS AND FAMILIES
For a sequence of smooth functions Z(w) = { (~,x) £ N and call it the zero set of
w 6 W
denote
x R n I w(~)(x)
w .
For a set of sequences of smooth functions Z(H) = { Z
] t ~H
and call it the zero family of
A standard argument
= 0 }
H
H c W
denote
} .
in algebras of functions
gives:
Theorem 1 Suppose
(V,S')
ted by NxR
n
V . Then .
is a regularization Z(1)
and denote by
and in particular,
Z(V)
I
the ideal in
W
genera-
are filter generators on
133
Proof First we p r o v e the r e l a t i o n (1)
Z(v)
Assume it w ¢ W which
is
by
~ ~ ,
false,
then
w(w)(x)
results
in
v
v ~ z v(v)(x)
= 1/v(w)(x) I(V,W)
= I(V#) $ 0 ,
,
V v ¢ N , x ~ Rn
g v ~ N , x ~ Rn
= W . Then
(20.3)
and
, therefore,
. Thus
(20.1)
in
u(1) chap.
one can define
= v • w ¢ 1,
will
I(V,W)
contradict
each other. Now, we can p r o v e (2)
Z(Vo)
Indeed, define jugate of (3) But,
n ...
w ¢ W
n Z(Vh)
by
~ ~ ,
w = v
o
. v
V
v ° ,..., v h ( I ( V , W )
* o + ... + v h • v h
where
V*i
is the complex con-
v i . Then, o b v i o u s l y Z(w) = Z(Vo)
w c I(V,W)
, since
n
...
n Z(Vh)
v ° ,..., v h ( I ( V , W )
. Therefore,
(3) and (i) will imply
(2)
VVV
The f o l l o w i n g three theorems r e f o r m u l a t e in terms of zero sets the results in Theorems 6, 7 and 8 in chap. i, §i0. Theorem 2 S u p p o s e given a local type r e g u l a r i z a t i o n
(V,S') , with
VQS'
sectional in-
variant. T h e n Z(v) for any
n ({~,U+I,...}
v ( V ,
U c N
×G) and
is infinite G c Rn ,
G $ ~ , open.
Theorem 3 Suppose given a r e g u l a r i z a t i o n ant. Then, the zero set Z(v)
n [ J ((v} I/_<~)
Z(v)
(V,S')
VQS'
such that
of each
v c V
x supp t(w)) infinite,
is sectional
invari-
has the property:
V
t ~ V(~S'
,
t~V,
U(N
Theorem 4 Suppose
(V,S')
is a r e g u l a r i z a t i o n .
Then, the zero set
Z(v)
of each
has the property: Z(v)
n U
~(N
({x~}
× supp
t(,))
~ ~ ,
W
t ~ VQS'
,
t ~ V
v ~ V
134
§3
ZERO SETS AND FAMILIES AT A POINT
The results
in Theorems
I, 2, 3, and 4 in §2, will be strengthened
in the present
sec-
tion. Given
w ~ W
,
H c W
and
x c R n , call
Zx(W ) = { , c N [ w ( , ) ( X ) the zero set
of
w
Zx~)
at
x
and c a l l
= { Zx(t)
the zero family of
H
at
I t ¢H
, the zero families
First,
several
Suppose
E
}
x .
It will be shown in Corollaries
(V,S')
= 0 }
i, 2 and 3 that for a large class of P-regularizations
Zx( ~
of
V
at any
x c Rn
are filter generators
on
N
notations.
is a set of subsets
WE={W{W
in
Rn
and denote
%/ xE c~EE : :
t
f Call
E
hereditary, %/
only if
x E E c E :
E \ {x} [ ~ Denote by sets in
Ef
and
E
c
~
E \ {x}
c E
the set of all nonvoid
Rn . O b v i o u s l y
Ef
and
Ec
and finite,
respectively
countable
are hereditary.
Theorem 5 Suppose given a vector
subspace
V
in
W , satisfying
the condition
vc~ Then, f o r any
vo
%/ E ~ E
:
xcE
:
Zx(Vo) n . . . If
E
is hereditary, %/
E ~E:
@
E'c E :
vh e V
, . . . ,
the property holds
n Zx(V h) # then the stronger
I)
car E' = car E
2)
Zx(Vo) n . . .
n Zx(Vh) ~ ~ ,
property
V
results
x c E'
for a certain
E c Ec
sub-
135 Proof It follows
Corollary
WV
from Len~aa 1 b e l o w
1
Suppose (4)
@ ~
Zx(V)
v
v E v
Then, the zero f a m i l y
Zx(Y )
on
(V,S')
the P-regularization ,
,
x
of
c
satisfies
the condition
Rn
g
at any
x c Rn
is a filter generator
N .
Proof It is easy to notice that
Corollary
(4] is equivalent with
V c W~ f
VVV
2
If under the conditions
in Corollary i,
zero family
V
Zx(V )
subsets of
of
at any
Y
is sectional
x c R n , generates
invariant,
a filter of infinite
N .
Proof Assume it is false and
(s)
v'o ,..., v
(6)
~ W
v:
¢ V , since
P c N
such that
by 1
if
v -
vi(v) (y)
if
~_>p+l
v. ~ V
and
V
is sectional
i
Zx(Vo) n . . . n Zx(V~) ~ 0
due t o C o r o l l a r y But,
and
v~(~] (y) =
i
(7)
v ° ,..., v h ~ Y
Zx(V o) n ... n Zx(V h) ~ {0 ..... ~}
Define
Then
x ~ Rn ,
1.
(5) implies that ¥
v ~ N ,
~ -> p + i :
i~ ¢ {0,...,h}
:
vi (~))(x) ~ 0 w h i c h together with
Call a subset
M
{~+I
for a certain
(6) is contradicting
of
N
, ~+2
p ~ N .
sectional, ,
...}
c
M
(7)
only if
VVV
then the
invariant.
Therefore
136
Corollary 3 If u n d e r the conditions in C o r o l l a r y i,
V
is subsequence invariant
i, §6), then the zero family
V
at any
ter of sectional subsets of
Zx(V )
of
(see chap.
x e R n , generates a fil-
N .
Proof It suffices to show that the sets nal in
Zx(V ) , with
N . Assume it is false and
tional in
v • V
and
x ~ Rn
N . Then, there exists an infinite subset
(8)
and
v c V
x • Rn
such that M
of
are all sectioZx(V )
is not sec-
N , such that
Zx(V) n M =
Assume
M = {Po ' Pl " ' ' }
(9)
v'(v)(y)
Then
v' ~ V , since
(10)
Zx(V' ) #
and d e f i n e
= v(~ v) (y)
g
•
v • V and
v ' c W by v c N
,
y e Rn
V i s subsequence i n v a r i a n t .
Therefore
due to C o r o l l a r y I. But,
(8) implies that V
~ c N
:
v(pv) (x) * 0 which t o g e t h e r w i t h (9) i s c o n t r a d i c t i n g
(10)
VW
And now, a lemma of a general interest, used in the'proof of T h e o r e m 5, is presented. Given a n o n v o i d set X , denote by For
w c W
and
x c X
W
the set of all functions
w
: N x X + C1
denote
Zx(W ) = { v • N [ w(v,x) = 0 } For a set
Y
of n o n v o i d and countable subsets in
Wy = { w ~ W
g~
yY •cY Y :
X , denote
t
Zy(W) ~
Lemma 1 Suppose
V
is a v e c t o r subspace in
v ° ,..., v h ~ V , the relation holds
(II)
V
Y~y
~
y c Y : Zy(Vo]
:
n ... n Zy(Vh) #
W
and
V c ~
. Then, for any
/37
If Y
is hereditary, V
Y~Y
:
y'c
(12)
then (ii) obtains the stronger form
y :
(12.1)
car
Y'
= car
Y
(12.2)
Zy(Vo)
n ...
n Zy(Vh)
$ ~ ,
V
y ~ Y'
Proof First, we prove
Y c Y
(ii). Assume it is false and
Zy(Vo) n ... n Zy(Vh) = ~ ,
V
such that
ycY
Then V
~c
N,
ye
Y :
i~,y c {O,...,h}
(13)
v.
1
(~,y)
:
~ o
~,y
Define Rh+l ~ X = (~o ''''' Xh)
+
v% = ~oVo + ... + XhV h E V
and N x y ~ It
is
easy
to
notice
(~,y)
+
that
Av,y
kv,y
= { X c Rh + l
, with
(v,y)
I vX(V,Y)
= 0 }
c N x y , are
vector
subspaces
i n Rh + l .
Moreover A v , y ~ Rh + l
(14) Indeed,
denoting
X = (0 . . . .
tion, one obtains Now,
,
X ~ k ,y
V
(V,y)
e N x y
,0,1,0,...,0)
e Rh + l
, with
1
in the
iv,y+l -th posi-
due to (13).
(14) and the Baire category argument will give
U
a , y ~ ah+l
>cN ycY since
Y
is
countable.
Assume therefore
% c R h+l \ [__J A ~6N v'Y y~Y then vx(V,y)
(is) But
v~ c V ¢ ~
Assume
Y
such that
$ 0 ,
therefore
V
V c N ,
y ~ Y
(15) is contradicted
is hereditary and take
Y c Y
and (ii) is proved.
• Then (ii) implies the existence of
Zy(Vo) n ... n Zy(vh) ¢ ~ . Denote
Y1 = Y \ {y} " If
Y1 = ~
y c Y
then taking
138
Y' = Y , the proof is completed. If
Y1 ~ ~
hereditary. Now, (ii) can be applied to
then
Y] , etc.
Y1 EY VW
due to the fact that
Y
is
139
REFERENCE
i.
Acker A., Walter W.: The quenching problem for nonlinear parabolic differential equations. Springer Lecture Notes in Mathematics, ~oi. 564, 1-12
2.
Alexandrov A.D.: Dirichlet's problem for the equation Vestinik Leningrad Univ. Ser. Mat. Mek. Astr. 13,1,1958
3.
Amrein W.O., Georgescu ?.~ Strong aaymptotik completeness of wave operators for highly singular potentials. Helv. Phys. Acta, 47,1974, 517-533
4.
Antosik P., Mikusinski J., Sikorski R.:
5.
Baumgarten D., Braunss G., Wagner 0.: An extension of the Gel'fand-Shilov regularization method and its application to the construction of causal solutions for nonlinear wave equations. Preprint, 1975
6.
Bogoliubov ~N., Parasiuk 0.S.: Ueber die Multiplikation der Kausalfunktionen in der Quantentheorie der Felder. Acta Math., 97,1957, 227-266
7.
Bellini C.G., Giambiagi J.J., Gonzales Dominguez A.: Analytic regularization and the divergencies of quantum field theories. Nuovo Cim., 31,3,1964, 550-561
8.
Braunss G.: 70-80
9.
Braunss G.: Causal functions of nonlinear wave equations. 1971, 86-92
det(zi.) = ~(z I .... ,zn) J
Theory of distributions. Elsevier, 1973
On the regularization of functional equations.
Math. Ann., 186,1970
J. Diff. 5q., 9,1,
i0.
Braunss G.: Weak solutions and junction conditions of a certain class of nonlinear partial differential equations. Preprint 1971
II.
Braunss G., Liese R.: Canonical products of distributions and causal solutions of nonlinear wave equations. J. Diff. Fxl., 16,1974, 399-412
12.
Bredimas A.: Sur une theorie electromagnetique non lineaire. 6,14,1973, 569-572
13.
Bredimas A.: Generalized convolution product and canonical product between distributions. Some simple applications to physics. Lett. Nuovo Cim., 13,16, 1975, 601-604
14.
Bredimas A.: La differentiation d'ordre complexe, le produit de convolution generalise et le produit canonique pour les distributions. C.R. Acad. Sc. Paris, 282,1976, A, 37-40
15.
Bredimas A.: Applications en Physique du theoreme d'inversion de la transformation spherique de Radon dans Rn . C.R. Acad. Sc. Paris, 282,1976, A, 1175-1178
16.
Bredimas A.: Extensions, proprietes complementaires et applications des operateurs de differentiation a gauche e t a droite d'ordre complexe. C.R. Acad. Sc. Paris, 283,1976, A, 3-6
17.
Bredimas A.: Applications a certaines equations differentielles des premier et second ordre a coefficients polynomiaux des operateurs de differentiation d'ordre complexe a gauche e t a droite. C.R. Acad. Sc. Paris, 283,1976, A, 337-340
Lett. Nuovo Cim.,
140
18.
Bredimas A.: La differentiation d'ordre complexe et les produits canoniques et la convolution generalise. C.R. Acad. Sc. Paris, 283,1976, A, 1095-1098
19.
Bremermann H.J., Durand L.: On analytic continuation, multiplication and Fourier transform of Schwartz distributions. J. Math. Phys., 2,2,1961, 240 258
20.
Bremermann H.J.: On finite renormalization constants and the multiplication of causal functions in perturbation theory. Preprint, 1962
21.
Bremermann H.J.: Some remarks on analytic representations and products of distributions. SIAM J. Appl. Math., 15,4,1967, 929-943
22.
Burrill C.W.:
23.
Carmignani R., Schrader K.: Subfunctions and distributional inequalities. SIAM J. Math. Anal., 8,1,1977, 52-68
24.
Conley C.C., Smoller J.A.: Shock waves as limits of progressive wave solutions of higher order equations. Comm. Pure Appl. Math., 24,1971, 459-472
25.
Conway E.D.: The formation and decay of shocks for a conservation law in several dimensions. Arch. Rat. Mech. Anal., 64,1,1977, 47-57
26.
Dafermos C.M.: Quasilinear hyperbolic systems that result from conservation laws. Leibovich S., Seebass A.R. (ed.): Nonlinear Waves, Cornell Univ. Press, 1974, 82-102
27.
Davies E.B.: Eigenfunetion expansions for singular Schroedinger operators. Arch. Rat. Mech. Anal., 63,3,1977, 261-272
28.
Delft P., Simon B.: On the decoupling of finite singularities from the question of asymptotic completeness in two body quantum systems. Preprint, 1975
29.
De Jager E.M.: Divergent convolution integrals in electrodynamics. Rep. Dept. Math. Univ. Calif. Berkeley, 1963
30.
De Mottoni P., Texi A.: On distribution solutions for nonlinear differential equations: nontriviality conditions. J. Diff. Fq., 24,1977, 355-364
31.
Di Perna R.J.: Singularities of solutions of nonlinear hyperbolic systems of conservation laws. Arch. Rat. Mech. Anal., 60,1976, 75-100
32.
Douglis A.: An ordering principle and generalized solutions of certain quasi linear partial differential equations. Comm. Pure Appl. Math., 12,1959, 87-112
33.
~wards R.E.:
34.
Ehrenpreis L.: Solutions of some problems of division. 1954, 883-903
35.
Fisher B.: The product of distributions. 291-298
36.
Fisher B.: The product of the distributions Camb. Phil. Soc. 71,1972, 123-130
x+-r-I/2
37.
Fisher B.: The product of the distributions Camb. Phil. Soc. 72,1972, 201-204
x
Foundations of Real Nhmbers.
Functional Analysis.
Mc Graw Hill, 1967
ONR, Tech.
Holt, 1965 Amer. J. Math., 76,
Quart. J. Math. Oxford, 22,2,1971,
-r
and
and
~
x -r-I/2 _
(r-l)
(x) .
Proc.
Proc.
141
38.
Fisher B.: Some results on the product of distributions. 73,1973, 317-325
39.
Fisher B.:
Singular products of distributions.
Proc. Camb. Phil. Soc
Math. Ann., 203,1973, 103-116
-r x+ 6(r-l)(x) . Stud. Sci. Math.
40.
Fisher B.: The neutrix distribution product Hung., 9,1974, 439-441
41.
Fisher B.: Distributions and the change of variable. Math. Rom., 19,1975, 11-20
42.
Fleishman B.A., Mahar T.J.: Boundary value problems for a nonlinear differential equation with discontinuous nonlinearities. Math. Balkanica, 3,1973, 98-108
43.
Fleishman B.A., Mahar T.J.: A perturbation method for free boundary problems of elliptic type. Preprint, 1976
44.
Flugge S.: Practical quantum mechanics I. Die Grundlehren der Mathematischen Wissenschaften. Band 177, Springer, 1971
45.
Fountain L., Jackson L.: A generalized solution of the boundary value problem. Pacif. J. Math., 12,1962, 1251-1272
46.
Fuchssteiner B.: Eine assoziative Algebra ueber einem Unterraum der Distributionen. Math. Ann., 178,1968, 302-314
47.
Galambos J.: Representation of real numbers by infinite series. ture Notes in Mathematics, vol. 502, 1976
48.
Garabedian P.R.:
49.
Garnir H.G.: Solovay's axiom and functional analysis. in Mathematics, vol. 399, 1974, 189-204
50.
Gel'fand J.M.: Some problems in the theory of quasilinear equations. Mat. N~uk, 14,1959, 87-158
51.
Glimm J., Lax P.D.: Decay of solutions of systems of nonlinear hyperbolic conservation laws. Memoirs of AMS, nr. 101, 1970
52.
Golubitsky M., Schaeffer D.G.: Stability of shock waves for a single conservation law. Adv. Math., 15,1975, 65-71
53.
Gonzales Dominguez A., Scarfiello R.: Nora sobre la formula Rev. Union Mat. Argen., I, 53-67, 1956
54.
Guettinger W.: Quantum field theory in the light of distribution analysis. Phys. Rev., 89,5,1953, 1004-1019
55.
Guettinger W.: Products of improper operators and the renormalization problem of quantum field theory. Progr. Theor. Phys., 13,6,1955, 612-.626
56.
Guettinger W.: Generalized functions and dispersion relations in physics. Fortschr. Phys., 14,1966, 483-602
57.
Guettinger W.: Generalized functions in elementary particle physics and passive system theory: recent trends and problems. SIAM J. Appl. Math., 15,4, 1967, 964-1000
58.
Guckenheimer J.:
An unsolvable equation.
Bull. Math. Soc. Sci.
Springer Lec-
Proc. AMS, 25,1970, 207-208 Springer Lecture Notes
Uspehi
v.p. 1 . 6 = 1 6' ~ -~
Solving a single conservation law. Preprint, 1973
142 59.
Harten A., Hyman J.M., Lax P.D., Keyfitz B.: ons and entropy conditions for shocks• 297-322
60.
Hepp K.: Proof of the Rogoliubov-Parasiuk theorem on renormalization. Math. Phys. 2,1966, 301-326
61.
Hirata y., Ogata H.: On the exchange formula for distributions. Univ., 22,1958, 147-152
62.
Hopf E.: The partial differential equation Math., 3,1950, 201-230
63.
Horvat J.: An introduction to distributions. 227-240
64.
Hoermander L.:
65.
Hoermander L.: An introduction to complex analysis in several variables. Van Nostrand, 1966
66.
Itano M.: On the multiplicative product of distributions. 29,1965, 51-74
67.
Itano M.: On the multiplicative products of 29,1965, 225-241
68.
Itano M.: On the theory of the multiplicative products of distributions. J. Sc. Hirosh. Univ., 30,1966, 151-181
69.
Itano M.: Remarks on the multiplicative product of distributions. Math. j., 6,1976, 365-375
70.
Jeffrey A.: The propagation of weak discontinuities in quasilinear symmetric hyperbolic systems. J. Appl. Math. Phys., 14,1963, 301-314
71.
Jeffrey A.:
72.
Jelinek J.: Sur le produit simple de deux distributions. rol., 5,4,1964, 209-214
73.
Kang H., Richards J.: Preprint, 1975
74.
Kang H.: A general definition of convolution for several distributions. Preprint, 1975
75.
Keisler H.J.:
76.
On finite difference approximatiComm. Pure Appl. Math., 29,1976,
u t + uu x = ~Uxx .
Comm.
J. Sc. Hirosh.
Comm. Pure Appl.
Amer. Math. Month., March 1970,
Linear Partial Differential Operators.
x~
and
Quasilinear hyperbolic systems and waves.
Springer, 1963
J. Sc. Hirosh. Univ.
B • J. Sc. Hirosh. Univ. x+
Hirosh.
Pitman, 1976 Comm. Math. Univ• Ca-
A general definition of convolution for distributions.
Element~ry Calculus.
Prindle,
1976
4 Kelemen P.J., Robinson A.: The nonstandard ~ : ~2(x] : model. I. The technique of nonstandard analysis in theoretical physics. II. The standard model from a nonstandard point of view. J. Math. Phys., 13,12,1972, 1870-1878
77.
Kelemen P.J.: Quantum mechanics, quantum field theory and hyper quantum mechanics. Springer Lecture Notes in Mathematics, vol. 369, 1974, 116-121
78.
Keller K.: Irregular operations in quantum field theory I, II, III, IV, V Preprint, 1977
79.
Keyfitz Quinn B.: Solutions with shocks: an example of an migroup. Comm. Pure Appl. Math., 24,1971, 125-132
L 1 contractive se-
143
80.
Knops R.J.: Comments on nonlinear elasticity and stability. Notes in Mathematics, vol. 564, 1976, 271-290
81.
Koenig H.:
82.
Koenig H.: Multiplikation und Variablentransformation in der Theorie der Distributionen. Arch. Math., 6,1955, 391-396
83.
Koenig H.: Multiplikationstheorie der verallgemeinerten Distributionen. Bayer. Akad. Wiss. Math. Nat. KI. Abh. (N.F.), 82,1957
84.
Larra Carrero L.: Stability of shock waves. matics, vol. 564, 1976, 316-328
85.
Laugwitz D.: Eine Einfuehrung der 6 Funktionen. turw. KI. 1959, 41-59
86.
Multiplikation yon Distributionen.
Springer Lecture
Math. Ann., 128,1955,
420-452
Springer Lecture Notes in Mathe-
Bayer. Akad. Wiss. Math. Na-
Laugwitz D.: Anwendungen unendlich kleiner Zahlen. I. Zur Theorie der Distributionen. II. Ein Zugang zur Operatorenrechnung von Mikusinski. J.f.d. reine und angew. Math., 207,1961, 53-60, 208,1961, 22-34
87.
Laugwitz D.: Bemerkungen zu Bolzano Groessenlehren. 1965, 398-409
88.
Laugwitz D.: Eine nichtarchimedische Erweiterung angeordneter Koerper. Nacht., 37,1968, 225-236
89.
Lax P.D.: Weak solutions of nonlinear hyperbolic equations and their numerical computation. Comm. Pure Appl. Math., 7,1954, 159-193
90.
Lax P.D.: Hyperbolic systems of conservation laws and the mathematical theory of shock waves. SIAM Regional Conference Series in Appl. Math., nr. II, 1973
91.
Leibovich S., Seebass A.R.
92.
Levine H.A., Payne L.E.: Nonexistence of global weak solutions for classes of nonlinear wave and parabolic equations. J. Math. Anal. Appl., 55,1976, 329-334
93.
Lewy H.: An example of a smooth linear partial differential equation without solution. Ann. Math., 66,2,1957, 155-158
94.
Ligeza J.: On generalized solutions of some differential nonlinear equations of order n . Ann. Pol. Math., 31,1975, 115-120
95.
Lightstone A.H.:
96.
Lightstone A.H., Kam Wong: Dirac delta functions via nonstandard analysis. Canad. Math. Bull., 81,5,1975, 759-762
97.
Lightstone A.H., Robinson A.: Non Archimedean fields and asymptotic expansions. North Holland, 1975
98.
Liverman T.P.G.: Physically motivated definition of distributions. Appl. Math., IS,1967, 1048-1076
99.
Lojasiewicz S.: Sur la valeur d'une distribution dans un point. Polon. Sc., 4,1956, 239-242
I00.
(ed.): Nonlinear Waves.
Infinitesimals.
Arch. Hist. Ex. Sc., 2,
Math.
Corneil Univ. Press, 1974
Amer. Math. Month., March 1972, 242-251
SIAM J.
Bull. Acad.
Lojasiewicz S.: Sur la valeur et la limite d'une distribution dans un point. Stud. Math., 16,1957, 1-36
144
i01.
Lojasiewicz S.: Sur la fixation des variables dans une distribution. Math., 17,1958, 1-64
Stud.
102.
Luxemburg W.A.J.: What is Nonstandard Analysis? part II, 38-67
103.
Malgrange B.: Existence et approximation des solutions des equations aux derivees partielles et des equations de convolution. Ann. Inst. Fourier Grenoble, 6,1955-56, p.271-375
104.
Meisters G.H.: Translation invariant linear forms and a formula for the Dirac measure. J. Funct. Anal., 8,1971, 173-188
i05.
Mikusinski J.: Sur la methode de generalisation de Laurent Schwartz et sur la convergence faible. Fund. Math., 35,1948, 235-239
106.
Mikusinski J.: 163-169
107.
Mikusinski J.: Criteria of the existence and of the associativity of the product of distributions. Stud. Math., 21,1962, 253-259
108.
Mikusinski J.: On the square of the Dirac delta distribution. Pol. S~., 14,9,1966, 511-513
109.
Mikusinski J.: 315-325
110.
Mikusinski J., Sikorski R.: The elementary theory of distributions I, II. Rozprawy Matematyczne, 12,1957 and 25,1961
Iii.
Moss W.F.: A remark on convergence of test functions. 20, A, 1975, 73-76
112.
Nokanishi N.: Complex dimensional invariant delta functions and lightcone singularities. Comm. Math. Phys., 48,1976, 97-118
113.
01einik O.A.: On the uniqueness of the generalized solution of the Cauchy problem for a nonlinear system of equations occuring in mechanics. Uspehi Mat. Nauk., 78,1957, 169-176
114.
Palamodov V.P.: Linear Differential Operators with Constant Coefficients. Springer, 1970
115.
Pearson D.B.: An example in potential scattering illustrating the breakdown in asymptotic completeness. Comm. Math. Phys., 40,1975, 125-146
116.
Pearson D.B.: General theory of potential scattering with absorption at local singularities. Preprint, 1975
117.
Plesset M.S., Prosperetti A.: Bubble dynamics and cavitation. Fluid Mech., 9,1977, 145-185
118.
Pogorelov A.V.: 1964
119.
Pogorelov A.V.: On a regular solution of the n dimensional Minkowski problem. Soviet Math. Dokl., 12,1971, 1192-1196
120.
Pogorelov A.V. : The Dirichlet problem for the n dimensional analogue of the Monge Ampere equation. Soviet Math. Dokl., 12,1971, 1727-1731
Amer. Math. Month., 80,6,
Irregular operations on distributions.
Germs and their operational calculus.
Stud. Math., 20,1961,
Bull. Acad.
Stud. Math., 26,1966,
Monge Ampere equation of elliptic type.
J. Austral. Math. Soc.
Ann. Rev.
Noordhoff, Groningen,
145
121.
Portnoy S.L.: On solutions to Appl., 55,1976, 291-294
u t = &u + u 2
122.
Reed M.C.: Abstract Nonlinear Wave Equations. matics, vol. 507, 1976
123.
Richtmyer R.D.: On the structure of some distributions discovered by Meisters. J. Funct. Anal., 9,1972, 336-348
124.
Robinson A.:
125.
Rosinger E.E.: Embedding the D'(R n) distributions into pseudotopological algebras. Stud. Cerc. Mat., 18,5,1966, 687-729
126.
Rosinger E.E.: Pseudotopological spaces. Embedding the D'(R n) into algebras. Stud. Cerc. Mat., 20,4,1968, 553-582
127.
Rosinger E.E.:
A distribution multiplication theory.
128.
Rosinger E.E.:
Division of distributions.
129.
Rosinger E.E.: An associative, commutative distribution multiplication I, II. Preprint, 1976
130.
Rosinger LE.: Nonsymmetric Dirac distributions in scattering theory. ger Lecture Notes in Mathematics, VOlo 564, 1976, 391-399
131.
Rosinger ~.E.: Nonlinear shock waves and distrihution multiplication. Preprint, 1977
132.
Sabharwal C.L.: Multiplication of singularity functions by discontinuous functions in Schwartz distribution theory. SIAM J. Appl. Math., 18,2,1970, 503-510
133.
Schaeffer D.G.: A regularity theorem for conservation laws. 1973, 368-386
134.
Schmieden C., Laugwitz D.: Eine Erweiterung der Infinitesimalrechnung. Zeitschr., 69,1958, 1-39
135.
Schwartz L.: Sur l'impossibilite de la multiplication des distributions. C.R. Aead. Sc. Paris, 239,1954, 847-848
136.
Shiraishi R., Itano M.: On the multiplicative products of distributions. J. Sc. Hirosh. Univ., 28,1964, 223-235
137.
Shiraishi R.: On the value of distributions at a point and the multiplicative products. J. Sc. Hirosh. Univ., 31,1967, 89-104
138.
Shiu Yuen Cheng, Shing Tung Yau: On the regularity of the Monge Ampere equation. Comm. Pure Appl. Math., 30,1977, 41-68
139.
Sikorsky R.: A definition of the notion of distribution. 2,5,1954, 209-211
140.
Simon B.: Quantum mechanics for Hamiltonian defined as quadratic forms. Princeton Univ. Press, 1971
141.
Slowikowski W.: A generalization of the theory of distributions. Pol. Sc., 3,1,1955, 3-6
Nonstandard Analysis.
in two dimensions.
J. Math. Anal.
Springer Lecture Notes in Mathe-
North Holland, 1966
distributions
Preprint, 1974
Pacif. J. Math., 66,1,1976, 257-263
Sprin-
Adv. Math., 11,3,
Math.
Bull. Acad. Pol. Sc.
Bull. Acad.
146
142.
Slowikowski W.: On the theory of operator systems. 1955, 137-142
143.
Speer LR.:
144.
Stroyan K.D., Luxemburg W.A.J.: Acad. Press, 1976
14S.
Struble R.A.: An algebraic view of distributions and operators. 37,1971, 103-109
146.
Stuart C.A.: Differential equations with discontinuous nonlinearities. Rat. Mech. Anal., 63,1,1976, 59-75
147.
Stuart C.A.: Boundary value problems with discontinuous nonlinearities. Springer Lecture Notes in Mathematics, vol. 564, 1976, 472-484
148.
Szaz A.: Convolution multipliers and distributions. 267-275
149.
Tavera G., Burnage H.: Sur les ondes de choc dans les ecoulements instationnaires. C.R. Acad. Sc. Paris, 284,1977, A, 571-573
150.
Temple G.: Theories and applications of generalized functions. Soc., 28,1953, 134-148
151.
Thurber J.K., Katz J.: Applications of fractional powers of delta functions. Springer Lecture Notes in Mathematics, vol. 369, 1974, 272-302
152.
Tillmann H.G.: Darstellung der Schwartz'schen Distributionen durch analytische Funktionen. Math. Z., 77,1961, 106-124
153.
Treves F.: Locally Convex Spaces and Linear Partial Differential Equations. Springer, 1967
154.
Treves F.: Applications of distributions to PDE theory. March 1970, 241-248
155.
Treves F.:
Linear Partial Differential Equations.
156.
Treves F.:
Basic Linear Partial Differential Equations.
157.
Van Osdol D.H.: rigorous.
158.
Van Rootselaar B.: Bolzano's t~eory of real numbers. 2,1964, 168-180
159.
Vol'pert A.J.: On differentiation and quasilinear operators in the space of functions whose generalized derivatives are measures. Soviet Math. Dokl. 7,6,1966, 1586-1589
160.
Vol'pert A.J.: The spaces BV and quasilinear equations. vol. 2,2,1967, 225-267
161.
Zel'dovich Ja. B.: Hydrodynamics of the universe. 1977, 215-228
162.
Walter H.F.: Ueber die Multiplikation von Distributionen in eine~ Folgenmodell Math. Ann., 189,1970, 211-221
163.
Whitham G.B.:
Analytic renormalization.
Bull. Acad. Pol. Sc., 3,3,
J. Math. Phys., 9,9,1968,
1404-1410
Introduction to the theory of infinitesimals.
Pacif. J
Stud. Math.,
Arch.
Math., 60,2,1975
J. London Math.
Amer. Math. Month.,
Gordon, Breach, 1970 Acad. Press, 1975
Truth with respect to an ultrafilter or how to make intuition Amer. Math. Month., April 1972, 355-363
Linear and Nonlinear Waves.
Arch. Hist. Exact Sc.,
Math. USSR Sbornik,
Ann. Rev. Fluid Mech., 9,
Wiley, 1974