• Tidak ada hasil yang ditemukan

Hiroshima Math. J. volume 34:1

N/A
N/A
Protected

Academic year: 2024

Membagikan "Hiroshima Math. J. volume 34:1"

Copied!
10
0
0

Teks penuh

(1)

34 (2004), 117–126

On products of distributions involving delta function and its partial derivatives

Sandra Monica Molina

(Received June 6, 2003) (Revised September 30, 2003)

Abstract. Li Banghe and Li Yaqing defined in [1] the product ST as a hyper- distribution (we call hyperdistributions to the complex linear functional of DðRnÞ in

rC0) forSandT inD0ðRnÞand they calculateddd. In this paper, we shall obtain expressions for the productsdqxq

idand q

qxidqxq

idfori¼1;. . .;n, and for evenn, these

products have the Hadamard finite part nonzero.

1. Introduction

In this section we shall define the product ‘‘’’. This definition involves a harmonic representation of distributions and Non-Standard Analysis. Let C and R be nonstandard models for the complex field and the real field respec- tively, and let r be a positive infinitesimal. Let Y denote the set of all infin- itesimal in C. Let

rC ¼ fxAC: for some finite integer n;jxj<rng; rC0¼rC=Y:

We call hyperdistributions the complex linear functional of DðRnÞ in rC0. Definition 1.1. Let TAD0ðRnÞ and uðx;yÞ be a harmonic function in Rþnþ1¼ fðx;yÞ:xARn;y>0g so that limy!0þ uðx;yÞ ¼T in the sense of D0ðRnÞ. Then uðx;yÞ is called a harmonic function associated with T or har- monic representation of T.

Lemma 1.2. Let S;T AD0ðRnÞ and SS;^ TT be harmonic representations of S^ and T respectively, and SS;^ TT denote the nonstandard extensions of^ SS;^ TT respec-^ tively. For any fADðRnÞ,

hSSðx;^ rÞTT^ðx;rÞ;fðxÞi ArC:

Lemma 1.3. Let S;T AD0ðRnÞ. If SS^1;SS^2 and TT^1;TT^2 are two harmonic representations of S and T respectively, then for any fADðRnÞ, we have

2000 Mathematics Subject Classification. 46F05, 46F10, 03H05, 46F20.

Keywords and phrases. nonstandard analysis, distribution, products.

(2)

hSS^2ðx;rÞTT^2ðx;rÞ;fðxÞihSS^1ðx;rÞTT^1ðx;rÞ;fðxÞiA0 Let F:rC! rC0 be the homomorphism modulo Y. We define

Definition 1.4. Let S;TAD0ðRnÞ and SS;^ TT be harmonic representations^ of S and T respectively. We call the complex linear functional of DðRnÞ !rC0

hST;fi¼FðhSSðx;^ rÞTT^ðx;rÞ;fðxÞiÞ the product of S and T.

The product ST is a generalization of the usual product for continuous func- tions onRn. In the following, we enunciate the fundamental properties of the product:

I. Commutativity: ST ¼TS, with S;TAD0ðRnÞ.

II. Bilinearity:

ða1S1þa2S2Þ ðb1T1þb2T2Þ ¼ X2

k;m¼1

akbmðSkTmÞ with ak;bmAC and Sk;TmAD0ðRnÞ for k;m¼1;2.

III. Localizability: Let U be an open subset of Rn and let Sk;TmA D0ðRnÞ for k;m¼1;2 so that S1nU ¼S2nU and T1nU ¼T2nU. Then ðS1T1ÞnU ¼ ðS2T2ÞnU.

IV. Let f ACy, TAD0ðRnÞ and f T be the usual multiplication.

Then f T ¼ f T.

V. Let f and g be continuous functions on Rn. Then f g¼ fg.

VI. The Leibnitz formula: Let S;TAD0ðRnÞ. Then q

qxi

ðSTÞ ¼ qS qxi

TþSqT qxi

:

2. The product dqxqd

i

Li Banghe and Li Yaqing obtained in [1], the following formula for dd with dAD0ðRnÞ:

dd¼X½n=2

j¼0

X

siANtfog Tsi¼j

r2jnCs1;...;sn

q2jd qx2s11. . .qxn2sn

; ð1Þ

where

Cs1;...;sn¼ 2

Gn2þjYn

i¼1

Gsiþ12 ð2siÞ!

ðy

0

t2jþn1 cn2ð1þt2Þnþ1 dt and cn¼pðnþ1Þ=2

Gð Þnþ12 , r a positive infinitesimal.

(3)

We can simplify the expression for the coe‰cients Cs1;...;sn by taking into account the following formula for the Beta function:

bðp;qÞ ¼ ð1

0

xp1ð1xÞq1dx¼ ðy

0

yp1 ð1þyÞpþq dy

¼2 ðy

0

t2p1

ð1þt2Þpþq dt ð2Þ

where p;qARþ. Next, we obtain by taking into account commutativity, the Leibnitz formula and (2), the following expression for dqxqdi:

d qd qxi¼1

2 q

qxifddg

¼X½n=2

j¼0

X

siANtfog Tsi¼j

r2jnDs1;...;sn

q2jþ1d

qx12s1. . .qx2si iþ1. . .qxn2sn

; ð3Þ

where

Ds1;...;sn¼Gn2 jþ1 Qn

i¼1Gsiþ12 2Qn

i¼1ð2siÞ!n!cn2 : ð4Þ We observe that the Hadamard finite part of dqxqd

i is nonzero for even n, because it is obtained for j¼n2 in the second term of the formula (3), so then

Hpf d qd qxi

¼ X

siANtfog Tsi¼n=2

Ds1;...;sn

qnþ1d

qx12s1. . .qxi2siþ1. . .qx2snn

; ð5Þ

where

Ds1;...;sn¼ Qn

i¼1Gsiþ12 2Qn

i¼1ð2siÞ!n!cn2: ð6Þ

3. The product qxqd

iqxqdi

An expression for the product qxqd

iqxqdi can be obtained by evaluating the following integral:

ð

Rn

cqd qxi qd

qxiðx;rÞ qdc qxi qd

qxiðx;rÞ fðxÞdx; ð7Þ here, we consider

c qd qxi

qd qxi

¼ rðnþ1Þxi

cnðjxj2þr2Þðnþ3Þ=2: ð8Þ

(4)

a harmonic representation for qxqd

i

AD0ðRnÞ, fADðRnÞ andr the positive infin- itesimal considered in section 1. We know that

fðxÞ ¼Xnþ2

j¼0

X

ðk1;...;knÞAN0n Tki¼j

xk11. . .xnkn k1! . . .kn!

qjf

qx1k1. . .qxnknð0Þ þ jxjnþ3cðxÞ ð9Þ

is the Taylor expansion of ftonþ2 th order, where cis continuous forx00 and bounded near x¼0. Then, we obtain in (7), by taking into account (8), (9) and suppfHfx:jxj<ag that

ð

Rn

qdc qxi

qd qxi

ðx;rÞ cqd qxi

qd qxi

ðx;rÞ fðxÞdx

¼ ð

jxj<a

r2ðnþ1Þ2xi2 c2nðjxj2þr2Þnþ3

Xnþ2

j¼0

X

ðk1;...;knÞAN0n Tki¼j

x1k1. . .xnkn k1! . . .kn!

qjf qx1k1. . .qxnkn

ð0Þ 8>

>>

<

>>

>:

9>

>>

=

>>

>; dx

þ ð

jxj<a

r2ðnþ1Þ2x2i

cn2ðjxj2þr2Þnþ3jxjnþ3cðxÞdx: ð10Þ The last term of the second member of the equality (10) is an infinitesimal number and we can omit it, in fact:

Let M¼supfcðxÞ:jxj<ag and we consider the polar coordinate in Rn: x1 ¼ rcosy1

x2 ¼ rsiny1 cosy2 ...

...

...

xn1 ¼ rsiny1 siny2. . .sinyn2 cosyn1

xn ¼ rsiny1 siny2. . .sinyn2 sinyn1

where

0ay1;y2;. . .;yn2ap; 0ayn1a2p: ð11Þ

We denote dWthe volume element of the unitary sphere Sn1. Then we have dx¼rn1drdW with

dW¼sinn2y1sinn3y2. . .sinyn2dy1. . .dyn1: Let Bn ¼2pn=2

Gð Þn2 be the ðn1Þ-dimensional measure of the ðn1Þ-sphere unitary Sn1. We obtain

(5)

ð

jxj<a

r2ðnþ1Þ2xi2

cn2ðjxj2þr2Þnþ3jxjnþ3cðxÞdx

ar2ðnþ1Þ2

c2n M ð

jxj<a

jxjnþ5 ðjxj2þr2Þnþ3 dx

¼r2ðnþ1Þ2 cn2 M

ð

Sn1

dW ða

0

rnþ5

ðr2þr2Þnþ3rn1dr

r¼rt¼ rM0

ða=r 0

t2nþ4

ð1þt2Þnþ3dt; ð12Þ

where M0¼ðnþ1Þ2MBn

cn2 , M0AR.

Now, we have ða=r

0

t2nþ4

ð1þt2Þnþ3 dt¼ ð1

0

t2nþ4

ð1þt2Þnþ3 dtþ ða=r

1

t2nþ4

ð1þt2Þnþ3 dt: ð13Þ Moreover

ða=r 1

t2nþ4 ð1þt2Þnþ3 dt

<

ða=r 1

1

t2 dt<y: ð14Þ Next, from (12) we obtain:

rM0 ða=r

0

t2nþ4

ð1þt2Þnþ3 dtA0

since rA0. Then, we consider in (10) the terms in the Taylor expansion of f until nþ2 th order

ð

jxj<a

r2ðnþ1Þ2xi2 cn2ðjxj2þr2Þnþ3

Xnþ2

j¼0

X

ðk1;...;knÞAN0n Tki¼j

x1k1. . .xnkn k1! . . .kn!

qjf qx1k1. . .qxnkn

ð0Þ 8>

>>

<

>>

>:

9>

>>

=

>>

>; dx

¼Xnþ2

j¼0

X

ðk1;...;knÞAN0n Tki¼j

r2ðnþ1Þ2 cn2k1! . . .kn!

ð

jxj<a

x1k1. . .xikiþ2. . .xnkn ðjxj2þr2Þnþ3 dx

qjf qx1k1. . .qxnkn

ð0Þ: ð15Þ

We consider for each janþ2, ðk1;. . .;knÞAN0n with k1þ þkn¼ j, the following terms:

(6)

r2ðnþ1Þ2 cn2k1! . . .kn!

ð

jxj<a

x1k1. . .xikiþ2. . .xnkn

ðjxj2þr2Þnþ3 dx: ð16Þ First, we consider i¼n. For that, we calculate the following terms:

r2ðnþ1Þ2 cn2k1! . . .kn!

ð

jxj<a

x1k1. . .xiki. . .xnknþ2

ðjxj2þr2Þnþ3 dx1. . .dxn: ð17Þ

Applying polar coordinate in Rn and by taking into account that Pn

i¼1ki¼ j, we obtain in (17) the following formula:

r2ðnþ1Þ2 cn2k1! . . .kn!

ð

jxj<a

x1k1. . .xiki. . .xnknþ2

ðjxj2þr2Þnþ3 dx1. . .dxn

¼ r2ðnþ1Þ2 cn2k1! . . .kn!

ða 0

rk1þþknþ2

ðr2þr2Þnþ3rn1dr ðp

0

cosk1y1sinjk1þ2þn2 y1dy1

ðp

0

cosk2y2sinjðk1þk2Þþ2þn3y2dy2

ðp 0

coskn2yn2sinjðk1þk2þþkn2Þþ2þnðn2þ1Þ

yn2dyn2

ð2p

0

coskn1 yn1sinjðk1þk2þþkn1Þþ2þnðn1þ1Þyn1dyn1: ð18Þ The integrals of the type Ðp

0 coskysinrydy are zero for k odd, then we consider even k1;. . .;kn2. Moreover, the integrals of typeÐ2p

0 cosk ysinrydy are nonzero only for even k and r. Then we consider even ki for i¼ 1;. . .;n. Let ki¼2si, then Pn

i¼1ki¼Pn

i¼12si¼ j, we obtain in (18) r2ðnþ1Þ2

c2nð2s1Þ! . . .ð2snÞ!

ð

jxj<a

x2s11. . .xi2si. . .xn2snþ2

ðjxj2þr2Þnþ3 dx1. . .dxn

¼ r2ðnþ1Þ2 cn2k1! . . .kn!

ða 0

rjþnþ1 ðr2þr2Þnþ3 dr

G2s12þ1

Gj2s1þ2þn2þ12 Gjþ2þn22 þ1

G2s22þ1

Gjð2s1þ2s22Þþ2þn3þ1 Gj2s1þ2þn32 þ1

G2sn22þ1

G jð2s1þ2s2þþ2sn2Þþ2þnðn2þ1Þþ1 2

G jð2s1þ2s2þþ2sn3Þþ2þnðn2þ1Þ

2 þ1

(7)

2

G2sn12þ1

G jð2s1þ2s2þþ2sn1Þþ2þnðn1þ1Þþ1 2

G jð2s1þ2s2þþ2sn2Þþ2þnðn1þ1Þ

2 þ1

¼ r2ðnþ1Þ2 cn2k1! . . .kn!

ða 0

rjþnþ1 ðr2þr2Þnþ3 dr

2

Qn1

i¼1 Gsiþ12 G j

Pi

r¼12srþnðiþ1Þþ3 2

G jþ2þn2

2 þ1

Qn2

i¼1 G j

Pi

r¼12srþnðiþ1þ1Þþ2

2 þ1

: ð19Þ

Now,

ða 0

rjþnþ1

ðr2þr2Þnþ3 drr¼rt¼ rjn4 ða=r

0

tjþnþ1

ð1þt2Þnþ3 dt ð20Þ and,

Qn1

i¼1 Gsiþ12 G j

Pi

r¼12srþnðiþ1Þþ3 2

Gjþ2þn22 þ1 Qn2

i¼1 G j

Pi

r¼12srþnðiþ1þ1Þþ2

2 þ1

¼snþ12 Qn

i¼1Gsiþ12

Gjþn2 þ1 : ð21Þ

Next, we obtain from (17) and the above calculations:

r2ðnþ1Þ2 cn2ð2s1Þ! . . .ð2snÞ!

ð

jxj<a

x12s1. . .x2si i. . .xn2snþ2

ðjxj2þr2Þnþ3 dx1. . .dxn

¼ r2ðnþ1Þ2

c2nð2s1Þ! . . .ð2snÞ!rjn4 ða=r

0

tjþnþ1

ð1þt2Þnþ3 dt2snþ12 Qn

i¼1Gsiþ12

Gjþn2 þ1 : ð22Þ We observe that

ða=r 0

tjþnþ1 ð1þt2Þnþ3dt¼

ðy 0

tjþnþ1 ð1þt2Þnþ3 dt

ðy a=r

tjþnþ1 ð1þt2Þnþ3 dt;

and,

rjn4 ðy

a=r

tjþnþ1 ð1þt2Þnþ3 dt

arjn4 ðy

a=r

tjn5dt¼ ajn4 njþ4:

(8)

Therefore, this last term is a finite number. Next, we obtain from (22), r2ðnþ1Þ2

cn2ð2s1Þ! . . .ð2snÞ!

ð

jxj<a

x12s1. . .x2si i. . .xn2snþ2

ðjxj2þr2Þnþ3 dx1. . .dxn

¼ ðnþ1Þ2

c2nð2s1Þ! . . .ð2snÞ! r2rjn4 ðy

0

tjþnþ1

ð1þt2Þnþ3 dtr2rjn4 ðy

a=r

tjþnþ1 ð1þt2Þnþ3 dt

" #

2snþ12 Qn

i¼1Gsiþ12

Gjþn2 þ1 ð23Þ

and

r2rjn4 ðy

a=r

tjþnþ1

ð1þt2Þnþ3 dtA0:

Next,

r2ðnþ1Þ2 cn2ð2s1Þ! . . .ð2snÞ!

ð

jxj<a

x12s1. . .x2si i. . .xn2snþ2

ðjxj2þr2Þnþ3 dx1. . .dxn

¼ ðnþ1Þ2

c2nð2s1Þ! . . .ð2snÞ!rjn2 ðy

0

tjþnþ1

ð1þt2Þnþ3 dt2snþ12 Qn

i¼1Gsiþ12

Gjþn2 þ1 : ð24Þ By taking into account the formula (2), we obtain:

ðy

0

tjþnþ1

ð1þt2Þnþ3 dt¼1 2

Gjþnþ22

Gn2jþ4

Gðnþ3Þ : ð25Þ

We obtain from (24) and (25) that:

r2ðnþ1Þ2 cn2ð2s1Þ! . . .ð2snÞ!

ð

jxj<a

x12s1. . .xi2si. . .xn2snþ2

ðjxj2þr2Þnþ3 dx1. . .dxn

¼

ðnþ1Þ2rjn2snþ12

Gnjþ42 Qn

i¼1Gsiþ12

c2nð2s1Þ! . . .ð2snÞ!ðnþ2Þ! ð26Þ Finally, we obtain from formulas (10) and (26) for the case i¼n that:

qd qxn

qd qxn

¼½ðnþ2Þ=2X

j¼0

r2jn2 X

siANUf0g Tsi¼j

Cs1;...;sn q2jd

qx12s1. . .qxn2sn

; ð27Þ

where

(9)

Cs1;...;sn¼ðnþ1Þ2Gn2jþ42 snþ12 Qn

r¼1Gsrþ12 Qn

r¼1ð2srÞ!cn2ðnþ2Þ! : ð28Þ

In the same way, the following formula is obtained for i¼1;. . .;n1:

qd qxi

qd qxi

¼½ðnþ2Þ=2X

j¼0

r2jn2 X

srANUf0g Tsr¼j

Cs1;...;sn q2jd

qx12s1. . .qxn2sn

; ð29Þ

where

Cs1;...;sn¼ðnþ1Þ2Gn2jþ42

siþ12 Qn

r¼1Gðsrþ12 Qn

r¼1ð2srÞ!cn2ðnþ2Þ! : ð30Þ

The Hadamard finite part is nonzero for evenn. In fact, we obtain in formula (29) that:

Hpf qd qxi

qd qxi

¼ X

siANUf0g Tsi¼ðnþ2Þ=2

Cs1;...;sn

q2jd qx12s1. . .qxn2sn

; ð31Þ

where

Cs1;...;sn ¼ðnþ1Þ2siþ12 Qn

r¼1Gsrþ12 Qn

r¼1ð2srÞ!cn2ðnþ2Þ! : ð32Þ

Acknowledgement

The author would like to express her thanks to Professor Li Bang-He for their helpful remarks and suggestions.

References

[ 1 ] Li Banghe and Li Yaqing, Nonstandard Analysis and Multiplication of Distributions in any dimension, Scientia Sinica (Series A) Vol. XXVIII No. 7, (1985), 716–726.

[ 2 ] Li Banghe and Li Yaqing, On the Harmonic and Analytic Representations of Distributions, Scientia Sinica (Series A). Vol. XXVIII No. 9, (1985), 923–937.

[ 3 ] Liu Shangping, Distributions in D0ðRnÞ as boundary values of harmonic functions, Scientia Sinica (Series A), Vol. XXVII No. 9, (1982).

[ 4 ] Li Banghe, Nonstandard Analysis and Multiplication of Distributions, Scientia Sinica, Vol.

XXI No. 5, (1978), 561–585.

[ 5 ] Li Ya-Qing, The harmonic product of dðx1;. . .;xnÞ and dðx1Þ and two combinatorial identities, Hiroshima Math. J.30 (2000), 371–376.

[ 6 ] A. Robinson, Nonstandard Analysis, North-Holland Publ., Amsterdam, 1966.

(10)

[ 7 ] W. A. J. Luxemburg, Non-Standard Analysis, Mathematics Department, California In- stitute of Technology, Pasadena, California, 1973.

[ 8 ] S. Albeverio, R. Hoegh-Krohn, Jens Erik Fenstad, Tom Lindstrom, Nonstandard Methods in Stochastic Analysis and Mathematical Physics, Academic Press, 1986.

[ 9 ] Laurent Schwartz, The´orie des Distributions, Ed. Hermann, Paris, 1966.

Departmento de Mathematica Facultad de Ciencias Exactas y Naturales

Universidad Nacional de Mar del Plata Funes 3350, CP. 7600

Mar. del Plata, Argentina e-mail: [email protected]

Referensi

Dokumen terkait