This article was downloaded by:[Tuan, Nguyen Minh]
On: 23 January 2008
Access Details: [subscription number 789785478]
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954
Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Complex Variables and Elliptic
Equations
An International Journal
Publication details, including instructions for authors and subscription information:
/>
On a class of singular integral equations with the linear
fractional Carleman shift and the degenerate kernel
Le Huy Chuan a; Nguyen Van Mau b; Nguyen Minh Tuan b
a
Department of Applied Physics, Graduate School of Engineering, Osaka
University, Japan
b
Faculty of Mathematics Mechanics and Informatics, Department of Analysis,
University of Hanoi, Hanoi, Vietnam
Online Publication Date: 01 February 2008
To cite this Article: Chuan, Le Huy, Van Mau, Nguyen and Tuan, Nguyen Minh
(2008) 'On a class of singular integral equations with the linear fractional Carleman shift and the degenerate kernel',
Complex Variables and Elliptic Equations, 53:2, 117 - 137
To link to this article: DOI: 10.1080/17476930701619782
URL: />
PLEASE SCROLL DOWN FOR ARTICLE
Full terms and conditions of use: />This article maybe used for research, teaching and private study purposes. Any substantial or systematic reproduction,
re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expressly
forbidden.
The publisher does not give any warranty express or implied or make any representation that the contents will be
complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should be
independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings,
demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or
arising out of the use of this material.
Downloaded By: [Tuan, Nguyen Minh] At: 18:21 23 January 2008
Complex Variables and Elliptic Equations
Vol. 53, No. 2, February 2008, 117–137
On a class of singular integral equations with the
linear fractional Carleman shift and the
degenerate kernel
LE HUY CHUANy, NGUYEN VAN MAUz and NGUYEN MINH TUAN*z
yDepartment of Applied Physics, Graduate School of Engineering, Osaka University, Japan
zFaculty of Mathematics Mechanics and Informatics, Department of Analysis,
University of Hanoi, 334, Nguyen Trai Str., Hanoi, Vietnam
Communicated by R. P. Gilbert
(Received 8 March 2007; in final form 9 June 2007)
This article deals with the solvability, the explicit solutions of a class of singular integral
equations with a linear-fractional Carleman shift and the degenerate kernel on the unit circle
by means of the Riemann boundary value problem and of a system of linear algebraic
equations. All cases about index of the coefficients in the equations are considered in detail.
Keywords: Integral operators; Singular Integral equations; Riemann boundary value problems
AMS Subject Classifications: 47G05; 45G05; 45E05
1. Introduction
Singular integral equations with a shift (SIES) have been studied for a long time (see [1,2]
and references therein). Many papers devoted to singular integral operators with a shift
(SIOS) are given to the construction of the Fredholm theory. Once M. G. Krein called the
Fredholm theory of linear operators a rough theory, and the theory describing its defect
subspaces a delicate theory [3]. However, the Fredholm theory of these operators brings
about only one thing, the defect of dimensions of kernel of the operator and its dual
operator. In other words, it is only the defect of the numbers of linear independent
solutions of the homogeneous equations reduced by the operator and the corresponding
dual operator. So the question of solving (and even of estimating the numbers of
solutions) of the corresponding equations actually remains open [4]. There are only a few
special types of SIES for which it is possible to answer this question to some extent [2,5].
Among the SIES of this type not reducible to two-term boundary value problems,
the most general and important is the class of singular integral equations with a
*Corresponding author. Email:
Complex Variables and Elliptic Equations
ISSN 1747-6933 print/ISSN 1747-6941 online ß 2008 Taylor & Francis
/>DOI: 10.1080/17476930701619782
Downloaded By: [Tuan, Nguyen Minh] At: 18:21 23 January 2008
118
L. H. Chuan et al.
linear-fractional Carleman shift. In our view, the singular integral equations with a
linear-fractional Carleman shift in the unit circle, in addition, deserve the interest.
Factorization is the main method used by some authors to investigate Fredholm and
solvability theory for SIES (see [3,4] and references therein). In [6], two of us gave a
general formula of linear-fractional Carleman shifts on the unit circle and solved by
means of Riemann boundary value problem for a class of singular integral equations with
a linear-fractional Carleman shift on the unit circle. In this article, we study the
solvability for a class of singular integral equations with a linear-fractional Carleman
shift and with the degenerate kernels on the unit circle. In genaral, one knows that the
singular integral operator of Cauchy’s type (denoted by S) do not commute with
Carleman shift opetaor (denoted by W), but the difference beetwen them WS À SW is a
compact operator [2]). In section 2, we obtain some identities relating to those operators.
The scheme of our investigation is divided into two parts: first, we move the degenerate
kernels to the right-side hand of the equation. Based on the identity W n ¼ I, we construct
the orthogonal projectors and reduce the equation to a system of singular integral
equations without shift and solve this system by means of Riemann boundary value
problem. Second, we reconstruct the solution of the orginal equation from the solutions
of system of equations that can be solved, but its solution depends on some unknown
parameters. As indicated below, the equations of the type (1.1) can be solved by means of
Riemann boundary value problem and by of a system of linear algebraic equations.
Let À ¼ {t 2 C, jtj ¼ 1} be the unit circle on the complex plane C and let X :¼ H(À),
0551. Let
! ðtÞ ¼
t þ
,
t þ
ð À
¼ 1;
6¼ 0Þ
be a Carleman linear-fractional function shift of order of n on À, i.e.,
8
! : À ! À;
>
>
>
<
!n ðtÞ ¼ t; for every t 2 À ð!k ðtÞ :¼ !ð!kÀ1 ðtÞÞ; !0 ðtÞ tÞ;
>
for some t 2 À; k ¼ 1, 2, . . . , n À 1,
!
k ðtÞ 6 t;
>
>
:
! positive orientation of À
The such functions ! (t) might be of the form
8
tÀ
>
, if
n ¼ 2;
<
" À1
! ðtÞ ¼ t
>
: ei t À , if n > 2,
" À1
t
" cos2 ðk=nÞ, for some k 2 {1, 2, . . . , n À 1},
where jj51; cos ¼ 1 À 2ð1 À Þ
(k, n) ¼ 1 [6].
In this article, we study the solvability of the equations of the form
Z
nÀ1
bðtÞ X
’ ðÞ
nÀk 1
d
"
a ðtÞ ’ ðtÞ þ
n k¼0 ‘ i À À !k ðtÞ
Z
m
X
1
þ
aj ðtÞbj ðÞ’ ðÞd ¼ fðtÞ,
i À
j¼1
where 1
‘
n À 1, "1 ¼ eð2i=nÞ, "‘ ¼ "‘1 :
ð1:1Þ
Downloaded By: [Tuan, Nguyen Minh] At: 18:21 23 January 2008
Singular integral equations
119
2. Some identities of singular integral operator of Cauchy’s type and linear-fractional
shift operator on unit circle
Consider the following operators in X:
Z
1
’ ðÞ
d,
ðS’Þ ðtÞ ¼
i À À t
ðW’Þ ðtÞ ¼ ’ð! ðtÞÞ,
n
1X
Pk ¼
"nÀ1Àj W jþ1 ,
n j¼1 k
ð2:1Þ
k ¼ 1, 2, . . . :
In the sequel, we shall need the following identities [7,8]:
8
n
P
>
k
>
W
¼
"kj Pj ; k ¼ 1; 2, . . . , n,
>
>
>
j¼1
>
<
Pk Pj ¼ kj Pj ; k; j ¼ 1; 2, . . . , n,
>
n
>
X
>
>
>
Pj ¼ I;
>
:
ð2:2Þ
j¼1
where kj is the Kronecker symbol. For every a 2 X we write (K a ’) (t) ¼ a (t) ’ (t).
Lemma 2.1 Let a 2 X be fixed. Then for every (k, j ), k, j 2 {1, 2, . . . , n} there exists
an element b 2 X such that Kb Xj & Xk and Pk Ka Pj ¼ Kb Pj , where xk :¼ P( X ).
The Function b(t) will be denoted by akj (t) and determined as follows
akj ðtÞ ¼
n
1X
" jÀk að!þ1 ðtÞÞ:
n ¼1 þ1
ð2:3Þ
By using (2.2) we get
Proof
Pk Ka Pj ¼
n
n
1X
1X
þ1
"nÀ1À
W
K
P
¼
"nÀ1À að!þ1 ðtÞÞW þ1 Pj
a
j
n ¼1 k
n ¼1 k
n
n
n
X
1X
1X
þ1
"Àðþ1Þ
að!
ðtÞÞ
"
P
P
¼
"Àðþ1Þ
að!þ1 ðtÞÞ"þ1
Pj
þ1
j
j
k
n ¼1 k
n
¼1
¼1
!
n
n
1X
1X
jÀk
Àk j
¼
" " að!þ1 ðtÞÞPj ¼
" að!þ1 ðtÞÞ Pj ¼ akj ðtÞPj ;
n ¼1 þ1 þ1
n ¼1 þ1
¼
where
ak j ðtÞ ¼
n
1X
" jÀk að!þ1 ðtÞÞ:
n ¼1 þ1
Putting akj(t) :¼ b(t), we obtain b 2 X and Pk Ka Pj ¼ Kb Pj.
Lemma 2.2
Let a 2 X be fixed. Then for any k, j 2 {1, 2, . . . , n}, we have
Pk Kak j ¼ Kak j Pj ,
ð2:4Þ
g
Downloaded By: [Tuan, Nguyen Minh] At: 18:21 23 January 2008
120
L. H. Chuan et al.
where akj(t) are determined by (2.4).
Proof
For any ’ 2 X we have
n
1X
"nÀ1À Wþ1 akj ðtÞ’ðtÞ
n ¼1 k
!
!
n
n
1X
1X
Àðþ1Þ þ1
jÀk
"
W
" að!þ1 ðtÞÞ ’ðtÞ
¼
n ¼1 k
n ¼1 þ1
"
#
n
n
Á
1X
1X
Àðþ1Þ À
jÀk jÀk
"þ1 "þ1 að!þ1þþ1 ðtÞÞ "kÀj
W þ1 ’ ðtÞ
¼
þ1 "k
n ¼1 n ¼1
ðPk Kakj ’ÞðtÞ ¼ Pk akj ðtÞ’ðtÞ ¼
n
À þ1 Á
1X
Àk
akj ðtÞ"kÀj
’ ðtÞ
þ1 "þ1 W
n ¼1
"
#
n
1X
nÀ1À
þ1
"
W
’ðtÞ ¼ akj ðtÞðPj ’ÞðtÞ ¼ ðKakj Pj ÞðtÞ:
¼ akj ðtÞ
n ¼1 j
¼
g
Thus Pk Kakj Kakj Pj :
Lemma 2.3
Let ’ 2 X. Then for every z 2 C we have
(1) ðSW ’ÞðzÞ ¼ ðW k S’ÞðzÞ À ðW kÀ1 S’Þ ð=
Þ; k ¼ 1; 2, . . . , W 0 ¼ I:
(2) ðPk S’ÞðzÞ ¼ ðSPk ’ÞðzÞ þ ð1="k À 1ÞðSPk ’Þ ð=
Þ; k ¼ 1; 2, . . . , n À 1,
k
where and
are
!ðzÞ ¼ ððz þ
Þ=ð
z þ ÞÞ.
the
coefficients
of
the
linear-fractional
function
Proof
(1) By induction on k. For k ¼ 1 we have
Z
1
’ð!ðÞÞ
d:
ðSW’ÞðzÞ ¼
i À À z
Put
¼ !À1 ðxÞ ¼
x À
;
À
x þ
then d ¼
1
dx:
ðÀ
x þ Þ2
Therefore,
Z
’ðxÞð1=ð
x À Þ2 Þ
dx
À ðx À
= À
x þ Þ À z
Z
1
’ðxÞ
dx
¼
i À ðÀ
x þ Þðx À
À zðÀ
x þ ÞÞ
Z
1
1
1
À
¼
’ðxÞdx
i À x À ðz þ
=
z þ Þ x À ð=
Þ
Z
Z
1
’ðxÞ
1
’ðxÞ
dx À
dx ¼ ðWS’ÞðzÞ À ðS’Þ
:
¼
i À x À !ðzÞ
i À x À ð=
Þ
ðSW’ÞðzÞ ¼
1
i
121
Downloaded By: [Tuan, Nguyen Minh] At: 18:21 23 January 2008
Singular integral equations
So, we obtain
;
ðSW’ÞðzÞ ¼ ðWS’ÞðzÞ À ðS’Þ
for any z 2 C:
ð2:5Þ
Suppose that (i) is true for k ¼ m. For k ¼ m þ 1 we find
ðSW mþ1 ’ÞðzÞ ¼ ½SWðW m ’ÞðzÞ
m
m
¼ W ½ðSW ’ÞðzÞ À ðSW ’Þ
¼ ½WSðW ’ÞðzÞ À ½SðW ’Þ
!
!
À ðW m S’Þ
À ðW mÀ1 S’Þ
¼ W ðW m S’ÞðzÞ À ðW mÀ1 S’Þ
!
À ðW m S’Þ
þ ðW mÀ1 S’Þ
:
¼ ðW mþ1 S’ÞðzÞ À W ðW mÀ1 S’Þ
m
m
Hence W ½ðW mÀ1 S’Þð=
Þ ¼ ðW mÀ1 S’Þð=
Þ; provided ðW mÀ1 S’Þð=
Þ is a constant.
Therefore
mþ1
mþ1
m
:
ðSW ’ÞðzÞ ¼ ðW S’ÞðzÞ À ðW S’Þ
The first part of the lemma is proved.
(2) Rewrite the equality in (1) in the form
ðW k S’ÞðzÞ ¼ ðSW k ’ÞðzÞ þ ðW kÀ1 S’Þ
:
We find
!
n
n
1X
1X
nÀiÀ1
iþ1
nÀiÀ1
iþ1
i
"
ðW S’ÞðzÞ ¼
"
ðSW ’ÞðzÞ þ ðW S’Þ
ðPk S’ÞðzÞ ¼
n i¼1 k
n i¼1 k
"
! #
"
! #
n
n
1X
1
1X
nÀiÀ1
iþ1
nÀi
i
"k
W
"k W S’
¼ S
’ ðzÞ þ
n i¼1
"k
n i¼1
1
¼ ðSPk ’ÞðzÞ þ ðPk S’Þ
:
"k
Substituting z ¼ ð=
Þ in formula (2.6), we get
1
ðPk S’Þ
¼ ðSPk ’Þ
:
1À
"k
ð2:6Þ
ð2:7Þ
If k 2 {1, 2, . . . , n À 1} then "k 6¼ 1. In this case, substituting (2.7) in (2.6) we receive
1
ðSPk Þ’
:
g
ðPk S’ÞðzÞ ¼ ðSPk ’ÞðzÞ þ
"k À 1
Downloaded By: [Tuan, Nguyen Minh] At: 18:21 23 January 2008
122
L. H. Chuan et al.
Comment From the identity (2.5), one can say that the operators S and W do not
commute to each other, but the difference of WS and SW at the a function ’(t) always
equals ðS’Þð=
Þ:
3. Reducing equation (1.1) to a system of singular integral equations
We now represent the equation (1.1) in the following form
a ðtÞ’ ðtÞ þ bðtÞðP‘ S’ÞðtÞ þ
Z
m
X
1
aj ðtÞbj ðÞ’ðÞd ¼ f ðtÞ;
i À
j¼1
ð3:1Þ
where a, b, f, a1, . . . , am, b1, . . . , bm 2 X are given, and S, P‘ (1 ‘ n À 1) are the
operators defined by (2.1). Suppose that a(t) is a non-vanishing function on À.
Denote by Mbj ; j ¼ 1, . . . , m, the linear functionals on X defined as follows
Mbj ð’Þ ¼
1
i
Z
bj ðÞ’ðÞd;
for any ’ 2 X:
À
Put j ¼Mbj(’), j=1, . . . , m. We reduce equation (3.1) to the following problem:
Find solutions ’ of equation
aðtÞ’ðtÞ þ bðtÞðP‘ S’ÞðtÞ ¼ fðtÞ À
m
X
j aj ðtÞ
ð3:2Þ
j¼1
depended on the parameters 1, . . . , m, and fulfilled the following conditions
Mbj ð’Þ ¼ j ; j ¼ 1, . . . :
ð3:3Þ
Lemma 3.1 Let ’ 2 X. Then ’ is a solution of (3.2) if and only if {’k ¼ Pk’,
k ¼ 1, 2, . . . , n} is a solution of the following system
Ã
a ðtÞ ’k ðtÞ þ
bÃk‘ ðtÞðS’‘ ÞðtÞ
bÃk‘ ðtÞ
ðS’‘ Þ
þ
¼ fkà ðtÞ,
"‘ À 1
k ¼ 1, 2, . . . , n,
ð3:4Þ
where
aà ðtÞ ¼
n
Y
a ð! j þ1 ðtÞÞ;
j¼1
bÃk‘ ðtÞ ¼
n
n
Y
1X
jþ1
"‘Àk
bð!
ðtÞÞ
að! þ1 ðtÞÞ;
n j¼1 jþ1
¼1
ð3:5Þ
6¼j
f Ãk ðtÞ
"
#
n
m
n
Y
X
1X
nÀ1Àj
jþ1
jþ1
¼
"k
f ð! ðtÞÞ À
a ð! ðtÞÞ
að! þ1 ðtÞÞ:
n j¼1
¼1
¼1
6¼j
123
Downloaded By: [Tuan, Nguyen Minh] At: 18:21 23 January 2008
Singular integral equations
Proof
Suppose that ’ 2 X is a solution of (3.2). We then have
n
Y
að! þ1 ðtÞÞ’ðtÞ þ bðtÞ
¼1
"
¼ fðtÞ À
m
X
#
j aj ðtÞ
j¼1
n
Y
að!þ1 ðtÞÞðP‘ S’ÞðtÞ
¼1
6¼nÀ1
n
Y
að!þ1 ðtÞÞ:
¼1
6¼nÀ1
Applying the projections Pk, k ¼ 1, 2, . . . , n to both sides of above equation and using
the Lemmas 2.1 and 2.2, we obtain
2
3
n
n
X
Y
61
7
"‘Àk bð! jþ1 ðtÞÞ
að! þjþ2 ðtÞÞ5ðP‘ S’ÞðtÞ
aà ðtÞðPk ’ÞðtÞ þ 4
n j¼1 jþ1
¼1
6¼nÀ1
"
#
n
m
n
Y
X
1X
nÀjÀ1
jþ1
jþ1
¼
"k
f ð! ðtÞÞ À
a ð! ðtÞÞ
að! þjþ2 ðtÞÞ:
n j¼1
¼1
¼1
ð3:6Þ
6¼nÀ1
It is easy to see that
n
Y
að! þjþ2 ðtÞÞ
¼1
6¼nÀ1
n
Y
að!þ1 ðtÞÞ
for any j 2 f1; 2, . . . , ng:
¼1
6¼j
Hence, (3.6) is equivalent to the following system
aà ðtÞðPk ’ÞðtÞ þ bÃk‘ ðtÞðP‘ S’ÞðtÞ ¼ f Ãk ðtÞ; k ¼ 1; 2, . . . , n:
ð3:7Þ
Using Lemma 2.3, we rewrite the system (3.7) in the form
bà ðtÞ
aà ðtÞðPk ’ÞðtÞ þ bÃk‘ ðtÞðSP‘ ’ÞðtÞ þ k‘ ðSP‘ ’Þ
¼ f Ãk ðtÞ, k ¼ 1, 2, . . . , n:
"‘ À 1
Thus (P1’, P2’ , . . . , Pn’) is a solution of (3.4).
Conversely, suppose that there exists 2 X such that (P1’, P2’ , . . . , Pn’) is a solution
of (3.4). Summing by k from 1 to n, we obtain
! X
n
n
X
1
Ã
Ã
ðSP‘ ’Þ
bk‘ ðtÞ ðSP‘ ’ÞðtÞ þ
f Ãk ðtÞ:
ð3:8Þ
¼
a ðtÞ ’ ðtÞ þ
"
À
1
‘
k¼1
k¼1
From (3.5), we get
n
X
k¼1
n
n
X
1X
jþ1
"‘Àk
ðtÞÞ
jþ1 bð!
n
Y
að! þ1 ðtÞÞ
n
¼1
j¼1
k¼1
6¼j
"
#
n
n
n
X 1X
Y
jþ1
¼
"‘Àk
ðtÞÞ
að! þ1 ðtÞÞ
bð!
jþ1
n
¼1
j¼1
k¼1
bÃk‘ ðtÞ ¼
6¼j
¼ bðtÞ
n
Y
¼1
6¼nÀ1
að! þ1 ðtÞÞ:
ð3:9Þ
Downloaded By: [Tuan, Nguyen Minh] At: 18:21 23 January 2008
124
L. H. Chuan et al.
Similarly,
n
X
"
f
Ã
k ðtÞ
¼
f ðtÞ À
m
X
#
j¼1
k¼1
n
Y
j aj ðtÞ
að! þ1 ðtÞÞ:
ð3:10Þ
¼1
6¼nÀ1
Therefore, (3.8) is equivalent to the following equality
aà ðtÞ’ðtÞ þ bðtÞ
n
Y
að! þ1 ðtÞÞ ðSP‘ ’ÞðtÞ þ
¼1
6¼nÀ1
¼ fðtÞ À
m
X
j¼1
!
1
ðSP‘ ’Þ
"‘ À 1
!Y
n
j aj ðtÞ
að! þ1 ðtÞÞ:
¼1
6¼nÀ1
This implies
aðtÞ’ðtÞ þ bðtÞðP‘ S’ÞðtÞ ¼ fðtÞ À
m
X
g
j aj ðtÞ:
j¼1
Lemma 3.2 If (’1, ’2, . . . , ’n) is a solution of system (3.4) then (P1’1, P2’2 , . . . , Pn’n)
is also its solution.
Proof Suppose (1, 2, . . . ,n) is a solution of the system (3.4). Applying the
projections Pk to both sides of k-th equation of (3.4) we get
aà ðtÞðPk ’k ÞðtÞ þ Pk bÃk‘ ðtÞðS’‘ ÞðtÞ þ
!
bÃk‘ ðtÞ
ðS’‘ Þ
¼ Pk ðf Ãk ðtÞÞ:
"‘ À 1
ð3:11Þ
It is easy to see that
Pk ðf Ãk ðtÞÞ ¼ f Ãk ðtÞ and
Pk bÃk‘ ðtÞ ¼ bÃk‘ ðtÞP‘ :
ð3:12Þ
Substituting (3.12) into (3.11), we obtain
Ã
a ðtÞðPk ’k ÞðtÞ þ
bÃk‘ ðtÞðP‘ S’‘ ÞðtÞ
bÃk‘ ðtÞ
P‘ ðS’‘ Þ
þ
¼ f Ãk ðtÞ:
"‘ À 1
ð3:13Þ
Provided ðS’‘ Þð=
Þ is a constant function, then
P‘
n
1X
nÀjÀ1
jþ1
ðS’‘ Þ
"
W
ðS’‘ Þ
¼
n j¼1 ‘
X
1
n nÀjÀ1
¼ ðS’‘ Þ
"
¼ 0:
n
j¼1 ‘
ð3:14Þ
Using Lemma 2.3, (3.13) is equivalent to the following equation
aà ðtÞðPk ’k ÞðtÞ þ bÃk‘ ðtÞðSP‘ ’‘ ÞðtÞ þ
bÃk‘ ðtÞ
ðSP‘ ’‘ Þ
¼ f Ãk ðtÞ;
"‘ À 1
Thus, (P1’1, P2’2, . . . , Pn’n) is a solution of (3.4).
k ¼ 1; 2, . . . , n:
Downloaded By: [Tuan, Nguyen Minh] At: 18:21 23 January 2008
Singular integral equations
Theorem 3.1
125
The equation (3.2) has solutions in X if and only if the following equation
aà ðtÞ’‘ ðtÞ þ bÑ‘ ðtÞðS’‘ ÞðtÞ þ
bÑ‘ ðtÞ
ðS’‘ Þ
¼ f Ñ ðtÞ
"‘ À 1
ð3:15Þ
has solutions. Moreover, if ’‘(t) is a solution of equation (3.15) then equation (3.2) has a
solution given by formula
’ðtÞ ¼
f ðtÞ À
Pm
j¼1
j aj ðtÞ À bðtÞðP‘ S’‘ ÞðtÞ
:
aðtÞ
ð3:16Þ
Proof Suppose that ’ 2 X is a solution of equation (3.2). By Lemma 3.1,
(P1’, P2’, . . . , Pn’) is a solution of system (3.4). Hence, P‘’ is a solution of (3.15).
Conversely, suppose that ’‘(t) is a solution of (3.15). In this case, system (3.4) has
solution (’1, ’2, . . . , ’n) determined by the formula
f Ãk ðtÞ À bÃk‘ ðtÞðS’‘ ÞðtÞ À ððbÃk‘ ðtÞÞ=ð"‘ À 1ÞÞðS’‘ Þð=
Þ
;
aà ðtÞ
k ¼ 1, 2, . . . , n; k 6¼ ‘:
’k ðtÞ ¼
ð3:17Þ
By Lemma 3.2, we have (P1’1, P2’2, . . . , Pn’n) is also a solution of (3.4). Put
’¼
n
X
Pk ’k :
ð3:18Þ
k¼1
It is clear that Pk’ ¼ Pk’k. This means that (P1’, P2’ , . . . , Pn’) is a solution of (3.4).
From Lemma 3.1 it follows that is a solution of (3.2). Moreover, from (3.17) and
(3.18) we get
’ðtÞ ¼
n
X
k¼1
¼
Pk ’k ¼
n
X
k¼1
Pk
f Ãk ðtÞ À bÃk‘ ðtÞðS’‘ ÞðtÞ À ðbÃk‘ ðtÞÞ=ð"‘ À 1ÞðS’‘ Þð=
Þ
aà ðtÞ
!
n
bÃk‘ ðtÞ
1 X
Ã
Ã
P
f
ðtÞ
À
b
ðtÞðP
S’
ÞðtÞ
À
ðS’
Þ
:
‘
‘
‘
‘
k‘
aà ðtÞ k¼1 k
"‘ À 1
ð3:19Þ
Substituting (3.9), (3.10), (3.14) into (3.19) we obtain
P
f ðtÞ À m
j¼1 j aj ðtÞ À bðtÞðP‘ S’‘ ÞðtÞ
’ðtÞ ¼
:
aðtÞ
g
The proof is complete.
4. The solvability of equation (3.15)
We set
Dþ ¼ fz 2 C : jzj51g;
DÀ ¼ fz 2 C : jzj41g
Downloaded By: [Tuan, Nguyen Minh] At: 18:21 23 January 2008
126
L. H. Chuan et al.
Denote by H(Dþ),H(DÀ) the sets of the analytic functions in Dþ and DÀ respectively.
Consider the equation (3.15)
bà ðtÞ
aà ðtÞ’‘ ðtÞ þ bÑ‘ ðtÞðS’‘ ÞðtÞ þ ‘‘ ðS’‘ Þ
¼ f Ñ ðtÞ:
"‘ À 1
Put
È‘ ðzÞ ¼
1
2i
Z
’‘ ðÞ
d;
ÀÀz
z 2 CnÀ:
According to Sokhotski–Plemelij formula, we have [1]
À
’‘ ðtÞ ¼ Èþ
‘ ðtÞ À È‘ ðtÞ,
ð4:1Þ
À
ðS’‘ ÞðtÞ ¼ Èþ
‘ ðtÞ þ È‘ ðtÞ:
ð4:2Þ
Moreover, ðS’‘ Þð=
Þ ¼ 2È‘ ð=
Þ: Put 0 ¼ È‘ ð=
Þ. We reduce equation (3.15) to the
following boundary problem: find a sectionally analytic function È‘(z) (È(z) ¼ Èþ(z)
for z 2 Dþ, È(z) ¼ ÈÀ(z) for z 2 DÀ) vanishes at infinity, È‘ ð=
Þ ¼ 0 and satisfies the
following linear relation on À
À
Èþ
‘ ðtÞ ¼ GðtÞÈ‘ ðtÞ þ gðtÞ;
t 2 À;
ð4:3Þ
where
GðtÞ ¼
aà ðtÞ À bÑ‘ ðtÞ
f à ðtÞ À 0 ðð2bÑ‘ ðtÞÞ=ð"‘ À 1ÞÞ
; gðtÞ ¼ ‘
:
Ã
Ã
a ðtÞ þ b‘‘ ðtÞ
aà ðtÞ þ bÑ‘ ðtÞ
ð4:4Þ
Suppose that aà ðtÞ Æ bÑ‘ ðtÞ are the non-vanishing functions on À. Then G(t), g(t) 2 X and
G(t) 6¼ 0 for any t 2 À. Put
Z
1
d ln GðtÞ,
{ ¼ GðtÞ ¼
2i À
Z
1
ln½ À{ GðÞ
d,
ÀðzÞ ¼
2i À
Àz
þ
Xþ ðzÞ ¼ eÀ
ðzÞ
,
À
XÀ ðzÞ ¼ zÀ{ eÀ
ðzÞ
:
ð4:5Þ
Using the results in [2] (p. 16–20) we get the following cases:
(1) { ! 0. The equation (4.3) has general solution is given by formula
!
Z
1
gðÞ d
È‘ ðzÞ ¼ XðzÞ
þ P{À1 ðzÞ
2i À Xþ ðÞ À z
Â
Ã
¼ XðzÞ ÉðzÞ À 0 BðzÞ þ P{À1 ðzÞ ,
ð4:6Þ
where
ÉðzÞ ¼
1
2i
Z
f‘Ã ðÞ
d
,
Ã
þ
Ã
À X ðÞða ðÞ þ b‘‘ ðÞÞ À z
ð4:7Þ
127
Downloaded By: [Tuan, Nguyen Minh] At: 18:21 23 January 2008
Singular integral equations
1
BðzÞ ¼
2i
Z
ð2bÑ‘ ðÞ="‘ À 1Þ
d
,
þ ðÞðaà ðÞ þ bà ðÞÞ À z
X
À
‘‘
P{À1 ðzÞ 0,
ð4:8Þ
if { ¼ 0,
and
P{À1 ðzÞ ¼ p1 þ p2 z þ Á Á Á þ p{ z{À1 ,
if { ! 1,
ð4:9Þ
which is a polynomial of degree { À 1 with arbitrary complex coefficients. The
function È‘(z) determined in (4.6) is a solution of problem (4.3) if È‘ ð=
Þ ¼ 0 ,
that is
!
É
À 0 B
þ P{À1
¼ 0 :
X
This implies
!
!
0 1 þ X
B
¼X
É
þ P{À1
:
ð4:10Þ
(i) If 1 þ Xð=
ÞBð=
Þ 6¼ 0 : from (4.10), we get
0 ¼
Xð=
Þ½Éð=
Þ þ P{À1 ð=
Þ
:
1 þ Xð=
ÞBð=
Þ
ð4:11Þ
In this case, the general solution of problem (4.3) is given by formula
!
Xð=
Þ½Éð=
Þ þ P{À1 ð=
Þ
BðzÞ þ P{À1 ðzÞ ,
È‘ ðzÞ ¼ XðzÞ ÉðzÞ À
1 þ Xð=
ÞBð=
Þ
ð4:12Þ
where X(z), É(z), B(z) are determined by (4.5), (4.7), (4.8) and P{À1(z) is a
polynomial of degree { À 1 with arbitrary complex coefficients.
(ii) If 1 þ Xð=
ÞBð=
Þ ¼ 0 : from (4.10) we get
þ P{À1
¼ 0:
ð4:13Þ
É
Then, the general solution of problem (4.3) is given by formula
È‘ ðzÞ ¼ XðzÞ½ÉðzÞ À 0 BðzÞ þ P{À1 ðzÞ,
ð4:14Þ
where X(z), É(z), B(z) are determined by (4.5), (4.7), (4.8), 0 is arbitrary and
P{À1(z) is a polynomial of degree { À 1 with complex coefficients satisfying
condition (4.13).
(2) {50. The necessary condition for the problem (4.3) to be solvable is that
Z
gðÞ {À1
d ¼ 0, k ¼ 1, . . . , À{:
þ
À X ðÞ
Downloaded By: [Tuan, Nguyen Minh] At: 18:21 23 January 2008
128
L. H. Chuan et al.
This condition can be written as follows
Z
f Ñ ðÞ {À1
d ¼ 0
Ã
þ
Ã
À X ðÞða ðÞ þ b‘‘ ðÞÞ
Z
ðð2bÑ‘ ðÞÞ=ð"‘ À 1ÞÞ kÀ1
d;
Ã
þ
Ã
À X ðÞða ðÞ þ b‘‘ ðÞÞ
k ¼ 1, . . . , À{: ð4:15Þ
In this case, we have P{À1(z) 0. So we receive
(i) If 1 þ Xð=
ÞBð=
Þ 6¼ 0 : from (4.11) we get
0 ¼
Xð=
ÞÉð=
Þ
:
1 þ Xð=
ÞBð=
Þ
Hence, (4.15) becomes the following condition
Z
f Ñ ðÞ kÀ1
Xð=
ÞÉð=
Þ
d ¼
þ ðÞðaà ðÞ þ bà ðÞÞ
1
þ
Xð=
ÞBð=
Þ
X
À
‘‘
Z
ð2bÑ‘ ðÞ="‘ À 1Þ kÀ1
d;
Ã
þ
Ã
À X ðÞða ðÞ þ b‘‘ ðÞÞ
ð4:16Þ
k ¼ 1, . . . , À {:
If the condition (4.16) is satisfied then the solution of the problem (4.3) is given
by formula
!
Xð=
ÞÉð=
Þ
BðzÞ ;
È‘ ðzÞ ¼ XðzÞ ÉðzÞ À
1 þ Xð=
ÞBð=
Þ
ð4:17Þ
where X(z), É(z), B(z) are determined by (4.5), (4.7), (4.8).
(ii) If 1 þ Xð=
ÞBð=
Þ ¼ 0 : from (4.13) we get
É
¼ 0:
ð4:18Þ
If the conditions (4.15) and (4.18) are satisfied then the solution of the problem
(4.3) is given by formula
Â
Ã
È‘ ðzÞ ¼ XðzÞ ÉðzÞ À 0 BðzÞ ;
ð4:19Þ
where X(z), É(z), B(z) are determined by (4.5), (4.7), (4.8) and 0 is determined
from condition (4.15).
Now we can formulate the main results about solutions of the equation (3.15)
in the following form
Theorem 4.1
Suppose that the functions aà ðtÞ Æ bÑ‘ ðtÞ does not vanish on À.
(1) If 1 þ Xð=
ÞBð=
Þ 6¼ 0 and { ! 0 then equation (3.15) has solutions ’‘ which
satisfy the following formula
!
Xð=
Þ½Éð=
Þ þ P{À1 ð=
Þ þ
B ðtÞ þ P{À1 ðtÞ
S’‘ ðtÞ ¼ Xþ ðtÞ Éþ ðtÞ À
1 þ Xð=
ÞBð=
Þ
!
Xð=
Þ½Éð=
Þ
þ P{À1 ð=
Þ À
þ X À ðtÞ ÉÀ ðtÞ À
B ðtÞ þ P{À1 ðtÞ ,
ð4:20Þ
1 þ Xð=
ÞBð=
Þ
Downloaded By: [Tuan, Nguyen Minh] At: 18:21 23 January 2008
Singular integral equations
129
where X(z), É(z), B(z) are determined by (4.5), (4.7), (4.8) and P{ À 1(z) is a
polynomial of degree { À 1 with arbitrary complex coefficients.
(2) If 1 þ Xð=
ÞBð=
Þ 6¼ 0 and {50 then equation (3.15) is solvable if the condition
(4.16) is satisfied. In this case, equation (3.15) has unique solution which satisfies the
formula (4.20), where P{À1(z) 0.
(3) If 1 þ X=
ÞBð=
Þ ¼ 0 and { ! 0 then equation (3.15) has solutions ’‘ which
satisfy the following formula
Â
Ã
S’‘ ðtÞ ¼ Xþ ðtÞ Éþ ðtÞ À 0 Bþ ðtÞ þ P{À1 ðtÞ
þ XÀ ðtÞ½ÉÀ ðtÞ À 0 BÀ ðtÞ þ P{À1 ðtÞ,
ð4:21Þ
where X(z), É(z), B(z) are determined by (4.5), (4.7), (4.8), 0 is arbitrary and
P{À1(z) is a polynomial of degree { À 1 with complex coefficients satisfying the
condition (4.13).
(4) If 1 þ Xð=
ÞBð=
Þ ¼ 0 and {50 then the equation (3.15) is solvable if the
condition (4.15) and (4.18) are satisfied. In this case, equation (3.15) has unique
solution which satisfies the formula (4.21), where P{À1(z) 0 and 0 is determined
from the condition (4.15).
Proof (1) From assumption it follows that the problem (4.3) has a solution È‘(z)
determined by (4.12). Therefore, equation (3.15) has a solution ’‘(t) determined by
(4.1). Moreover, from (4.2) we get
À
S’‘ ðtÞ ¼ Èþ
‘ ðtÞ þ È‘ ðtÞ
!
Xð=
Þ ½Éð=
Þ þ P{À1 ð=
Þ þ
B ðtÞ þ P{À1 ðtÞ
1 þ Xð=
ÞBð=
Þ
!
Xð=
Þ
½Éð=
Þ þ P{À1 ð=
Þ À
þ XÀ ðtÞ ÉÀ ðtÞ À
B ðtÞ þ P{À1 ðtÞ :
1 þ Xð=
ÞBð=
Þ
¼ Xþ ðtÞ Éþ ðtÞ À
Similarly, the cases (2), (3), (4) can be proved.
g
5. The solvability of equation (3.1)
Theorems 3.1 and 4.1 show that if aà ðtÞ Æ bÑ‘ ðtÞ 6¼ 0 on À then equation (3.2) is solvable
in a closed form. In this section, we study which solutions of (3.2) will be the solution of
(3.1), i.e., the solutions of (3.2) need to satisfy the condition (3.3). Consider the
following cases:
(1) 1 þ Xð=
ÞBð=
Þ 6¼ 0; { ! 0: By using Theorems 3.1 and 4.1, we have solutions
of (3.2) given by the following formula
’ ðtÞ ¼
f ðtÞ À
Pm
j¼1
j aj ðtÞ À bðtÞðP‘ S’‘ ÞðtÞ
;
aðtÞ
ð5:1Þ
Downloaded By: [Tuan, Nguyen Minh] At: 18:21 23 January 2008
130
L. H. Chuan et al.
where S’‘(t) is determined by (4.20). From (3.5) and (4.7) we get
Â
Ã
Z ð1=nÞ Pn "nÀ1Àj f ð! jþ1 ðÞÞ À Pm a ð!jþ1 ðÞÞ Qn¼1 að! þ1 ðÞÞ
j¼1 ‘
¼1
1
d
6¼j
ÉðzÞ ¼
Ã
þ
Ã
2i À
Àz
X ðÞða ðÞ þ b‘‘ ðÞÞ
m
X
¼ É1 ðzÞ À
A ðzÞ,
ð5:2Þ
¼1
where
Z ð1=nÞ Pn "nÀ1Àj f ð! jþ1 ðÞÞ Qn¼1 að!þ1 ðÞÞ
j¼1 ‘
d
6¼j
,
Ã
þ
Ã
Àz
X ðÞða ðÞ þ b‘‘ ðÞÞ
À
Z ð1=nÞ Pn "nÀ1Àj a ð! jþ1 ðÞÞ Qn¼1 að! þ1 ðÞÞ
j¼1 ‘
1
d
6¼j
:
A ðzÞ ¼
2i À
Àz
Xþ ðÞðaà ðÞ þ bÑ‘ ðÞÞ
1
É1 ðzÞ ¼
2i
ð5:3Þ
ð5:4Þ
Substituting (4.9), (5.2) into (4.20) we find
À
À
S’‘ ðtÞ ¼ ½Xþ ðtÞÉþ
1 ðtÞ þ X ðtÞÉ1 ðtÞ À
m
X
À
À
j ½Xþ ðtÞAþ
j ðtÞ þ X ðtÞAj ðtÞ
j¼1
Â
Pm
Xð=
Þ É1 ð=
Þ À j¼1 j Aj ð=
Þ þ
À
1 þ Xð=
ÞBð=
Þ
þ
{
X
P{
j¼1 pj ð=
Þ
jÀ1
Ã
½X þ ðtÞB þ ðtÞ þ X À ðtÞB À ðtÞ
pj t jÀ1 ½X þ ðtÞ þ X À ðtÞ:
j¼1
Since we can rewrite (5.1) in the form
’ðtÞ ¼
f ðtÞ À b ðtÞP‘ ½X þ ðtÞÉ1 þ ðtÞ þ X À ðtÞÉÀ
1 ðtÞ
aðtÞ
m
X
À
À
aj ðtÞ À bðtÞP‘ ½Xþ ðtÞAþ
j ðtÞ þ X ðtÞAj ðtÞ
aðtÞ
j¼1
Â
Ã
P
P{
jÀ1
Xð=
Þ É1 ð=
Þ À m
j¼1 j Aj ð=
Þ þ
j¼1 pj ð=
Þ
þ
1 þ Xð=
ÞBð=
Þ
À
j
bðtÞP‘ ½X þ ðtÞB þ ðtÞ þ X À ðtÞB À ðtÞ
aðtÞ
Ã
 jÀ1 þ
{
X bðtÞP‘ t ½X ðtÞ þ XÀ ðtÞ
,
À
pj
aðtÞ
j¼1
Â
ð5:5Þ
where X(z), B(z), É1(z), A1(z), . . . , Am(z) are determined by (4.5), (4.8), (5.3), (5.4), and
p1, . . . , p{ are arbitrary. The function ’ is a solution of the equation (3.1) if it satisfies
the condition (3.3), that is
Mbk ’ ¼ k ;
k ¼ 1, . . . , m:
Downloaded By: [Tuan, Nguyen Minh] At: 18:21 23 January 2008
Singular integral equations
131
Substituting (5.5) into the last condition, we obtain
"
X
jÀ1 #
m
m
{
{
X
X
X
k ¼ dk À
ekj j þ É1
j Aj
pj
gkj pj
fk À
À
þ
j¼1
j¼1
j¼1
j¼1
"
! X
!
jÀ1 #
m
{
X
¼ dk þ É1
ekj þ fk Aj
gkj À fk
pj ;
fk À
j À
j¼1
j¼1
k ¼ 1, 2, . . . , m,
ð5:6Þ
where
dk ¼ Mbk
ekj ¼ Mbk
À
À
f ðtÞ À bðtÞP‘ ½Xþ ðtÞÉþ
1 ðtÞ þ X ðtÞÉ1 ðtÞ
;
aðtÞ
!
À
À
aj ðtÞ À bðtÞP‘ ½Xþ ðtÞAþ
j ðtÞ þ X ðtÞAj ðtÞ
;
aðtÞ
Xð=
Þ
bðtÞP‘ ½X þ ðtÞB þ ðtÞ þ X À ðtÞBÀ ðtÞ
;
fk ¼ Mbk
1 þ Xð=
ÞBð=
Þ
aðtÞ
Ã
Â
bðtÞP‘ tjÀ1 ½Xþ ðtÞ þ XÀ ðtÞ
gkj ¼ Mbk
:
aðtÞ
ð5:7Þ
Put
1
0 1
B d1 þ É 1
f 1 C
p1
1
C
B
C
B
B . C
B .. C
.
C ,
B
B
C
.
¼@ . A , P¼@ . A , D¼B
..
C
C
B
A
@
m mÂ1
p{ {Â1
dm þ É 1
fm
mÂ1
0
1
B e11 þ f1 A1
Á Á Á e1m þ f1 Am
C
C
B
C
B
.. C
..
..
,
E¼B
C
B .
.
. C
B
@
A
em1 þ fm A1
Á Á Á emm þ fm Am
mÂm
0
0
{À1 1
C
B g11 À f1
Á Á Á g1{ À f1
C
B
C
B
.
.
.
C
B
G¼B
..
..
..
C
C
B
@
0
{À1 A
gm1 À fm
Á Á Á gm{ À fm
mÂ{:
0
1
0
ð5:8Þ
Now we write (5.6) in the form of matrix condition
ðI þ EÞ ¼ D À GP,
ð5:9Þ
where I is the unit matrix. So we can formulate that the function ’ determined by (5.5) is
a solution of (3.1) if and only if (1 , . . . , m) satisfy the condition (5.9).
Downloaded By: [Tuan, Nguyen Minh] At: 18:21 23 January 2008
132
L. H. Chuan et al.
(2) 1 þ Xð=
ÞBð=
Þ 6¼ 0; {50: From Theorems 3.1 and 4.1 it follows that the
equation (3.2) has solutions if and only if the condition (4.16) satisfied. If this is in
case, then P{À1 0. So, the solutions of (3.2) are given by as follows
’ðtÞ ¼
À
À
fðtÞ À bðtÞP‘ ½Xþ ðtÞÉþ
1 ðtÞ þ X ðtÞÉ1 ðtÞ
aðtÞ
À
À
m
X
aj ðtÞ À bðtÞP‘ ½Xþ ðtÞAþ
j ðtÞ þ X ðtÞAj ðtÞ
À
j
aðtÞ
j¼1
Â
Ã
Pm
Xð=
Þ É1 ð=
Þ À j¼1 j Aj ð=
Þ bðtÞP‘ ½Xþ ðtÞBþ ðtÞ þ XÀ ðtÞBÀ ðtÞ
:
þ
aðtÞ
1 þ Xð=
ÞBð=
Þ
ð5:10Þ
Therefore, the function determined by (5.10) is a solution of the equation (3.1) if and
only if (1, . . . , m) satisfy the following matrix condition
ðI þ E Þ ¼ D,
ð5:11Þ
where E and D are determined by (5.8). On the other hand, substituting (3.5), (5.2) into
(4.16) we get
m
m
X
0 X
0
0
0
dk À
ek ¼ É1
fk Aj
fk À
j , k ¼ 1, 2, . . . , À{ ,
ð5:12Þ
¼1
j¼1
where
Z ð1=nÞ Pn "nÀ1Àj f ð! jþ1 ðÞÞ Qn¼1 að! þ1 ðÞÞ
j¼1 ‘
6¼j
kÀ1 d
þ ðÞðaà ðÞ þ bà ðÞÞ
X
À
‘‘
Z ð1=nÞ Pn "nÀ1Àj a ð! jþ1 ðÞÞ Qn¼1 að! þ1 ðÞÞ
j¼1 ‘
6¼j
e0k ¼
kÀ1 d
Xþ ðÞðaà ðÞ þ bÑ‘ ðÞÞ
À
Z
Xð=
Þ
ðð2bÑ‘ ðÞÞ=ð"‘ À 1ÞÞ kÀ1
0
d:
fk ¼
1 þ Xð=
ÞBð=
Þ À Xþ ðÞðaà ðÞ þ bÑ‘ ðÞÞ
dk0
¼
ð5:13Þ
Put
1
0
0
À
É
f
d
1
B 1
1 C
C
B
C
B
C
B
0
.
D ¼B
C
..
C
B
C
B
A
@
0
0
dÀ{ À É1
fÀ{
À{Â1,
1
0
0
0
0
0
À
f
A
À
f
A
Á
Á
Á
e
e
1 1
1m
1 m
C
B 11
C
B
C
B
C
B
0
.
.
.
E ¼B
:
C
..
..
..
C
B
B
C
@
A
0
0
0
0
eÀ{1
À fÀ{
A1
À fÀ{
Am
Á Á Á eÀ{m
À{Âm
0
ð5:14Þ
133
Downloaded By: [Tuan, Nguyen Minh] At: 18:21 23 January 2008
Singular integral equations
We write (5.12) in the form of matrix condition
E 0 ¼ D 0:
ð5:15Þ
Combining (5.11) and (5.15) we can say that the function determined by
(5.10) is a solution of (3.1) if and only if (1 , . . . , m) satisfy the following matrix
condition
0
D
IþE
¼
:
ð5:16Þ
0
D ðmÀ{ÞÂ1
E
ðmÀ{ÞÂm
(3) 1 þ Xð=
ÞBð=
Þ ¼ 0, { ! 0: Then the solutions of the equation (3.2) are given by
the following formula
’ðtÞ ¼
À
À
fðtÞ À bðtÞP‘ ½Xþ ðtÞÉþ
1 ðtÞ þ X ðtÞÉ1 ðtÞ
aðtÞ
À
À
m
X
aj ðtÞ À bðtÞP‘ ½Xþ ðtÞAþ
j ðtÞ þ X ðtÞAj ðtÞ
À
j
aðtÞ
j¼1
bðtÞP‘ ½Xþ ðtÞBþ ðtÞ þ XÀ ðtÞBÀ ðtÞ
aðtÞ
Ã
 jÀ1 þ
{
X
bðtÞP‘ t ½X ðtÞ þ XÀ ðtÞ
,
À
pj
aðtÞ
j¼1
þ 0
ð5:17Þ
where X(z), B(z), É1(z), A1(z), . . . , Am(z) are determined by (4.5), (4.8), (5.3), (5.4), 0 is
an arbitrary complex number and p1, . . . , p{ satisfy the condition (4.13). Substituting
(5.2) in (4.13) we obtain
X
X
jÀ1
m
{
É1
j Aj
pj
¼ 0:
À
þ
ð5:18Þ
j¼1
j¼1
The function ’ is a solution of the equation (3.1) if it satisfies the condition (3.3).
Substituting (5.17) into (3.3) we get
k ¼ dk À
m
X
ekj j þ 0 fk À
j¼1
¼ dk þ 0 fk À
{
X
gkj pj
j¼1
m
X
ekj j À
j¼1
{
X
gkj pj ,
k ¼ 1, 2, . . . , m,
where dk, ekj, fk, gkj are determined by (5.7). Put
0 1
0 1
0 1
f1
1
d1
B . C
B .. C
B .. C
C
¼@ . A ,
D¼@ . A ,
F¼B
@ .. A ,
m mÂ1
dm mÂ1
fm mÂ1
0
1
0
1
g11 Á Á Á g1m
e11 Á Á Á e1m
B.
.. C
..
. . .. C
B.
C
E ¼ @ ..
,
G¼B
. A
.
. . A
@ ..
em1
ÁÁÁ
emm
mÂm
ð5:19Þ
j¼1
gmj
Á Á Á gm{
mÂ{
0
1
p1
B . C
C
P¼B
@ .. A ,
p{ {Â1
:
ð5:20Þ