• Nie Znaleziono Wyników

On some application of algebraic quasinuclei to the determinant theory

N/A
N/A
Protected

Academic year: 2022

Share "On some application of algebraic quasinuclei to the determinant theory"

Copied!
12
0
0

Pełen tekst

(1)

ON SOME APPLICATION OF ALGEBRAIC QUASINUCLEI TO THE DETERMINANT THEORY

Grażyna Ciecierska

Faculty of Mathematics and Computer Science, University of Warmia and Mazury in Olsztyn Olsztyn, Poland

grac@matman.uwm.edu.pl

Abstract. In the paper we apply the modified powers of algebraic quasinuclei to construc- tion of determinant systems for quasinuclear perturbations of Fredholm operators. Given two pairs (Ξ,Χ), (Ω ,Υ) of conjugate linear spaces, an algebraic quasinucleus

(Ξ Χ Υ)

an ,

F and a determinant system for the Fredholm operator

(Ξ Χ Υ)

op ,

S , we obtain algebraic formulas for terms of a determinant system for S +TF.

Keywords: determinant system, Fredholm operator, quasinucleus, quasinuclear operator

Introduction

Determinant systems for operators acting in infinite dimensional Banach spaces provide important tools for solving linear equations. The determinant system for linear operator A gives full information on solving the equation Ax =y0, where y0 belongs to the range of A. The Sikorski’s and Buraczewski’s formulas [1-3]

for the solution are generalizations of the famous Cramer’s rule for solving finite systems of linear equations.

The first theory of determinants in arbitrary Banach spaces was developed by A.F. Ruston [4] and A. Grothendieck [5] and another one by T. Leżański [6], R. Sikorski [1] and A. Buraczewski [2]. A general approach to the theory of determi- nants was proposed by A. Pietsch [7], I. Gohberg, S. Goldberg and N. Krupnik [8].

The study of determinant systems leads to the study of concrete classes of Fred- holm operators. In this approach we consider the class of quasinuclear perturba- tions of Fredholm operators. Algebraic quasinuclei play an important role in the theory of determinant systems; if ( )D is a determinant system for a Fredholm n operator S and T is the quasinuclear operator determined by an algebraic F quasinucleus F , then we can obtain effective formulas for a determinant system for the operator S +TF in Banach spaces. The purpose of this paper is to give purely algebraic formulas for terms of the mentioned determinant system. The for-

(2)

mulas were first given by Plemelj [9] for endomorphisms of the form I + , where T T is an integral endomorphism, in the space C[a,b]. These formulas were obtained on the basis of the Fredholm theory of integral equations. They were modified by Smithies [10], also in the case of endomorphisms I + , where T is integral. T R. Sikorski [1] generalized the formulas over the endomorphisms I + , where T T is quasinuclear. A. Buraczewski [2] made further generalization of these formulas in the case of operators of the form S + , where S is a fixed Fredholm operator of T order zero and T is quasinuclear. Later contribution was made by D.H.U. Marchetti [11], who presented an alternative to Plemelj-Smithies formulas in the case of endomorphisms I + , where T belongs to the trace class of endomorphisms in T a separable Hilbert space. In this paper we generalize Plemelj-Smithies formulas over the operators of the form S + , where S is an arbitrary Fredholm operator T and T is quasinuclear. The result is formulated by means of the modified powers of quasinuclei.

1. Terminology and notation

We begin with a brief review on the terminology used in the determinant theory.

We follow the notation of [1-4].

Let (Ξ,Χ), (Ω ,Υ ), (Λ,Z) denote pairs of conjugate linear spaces over K

(K=Ror K=C). A bilinear functional A:×Χ K, whose value at a point

(ω,x)×X is denoted by Axω , satisfying the condition ωAx=ω( ) (Ax = ωA)x, where Ax ∈ and Y ωAΞ, is called (Ξ ,Υ)-operator on Ω ×Χ; the space of all

(Ξ ,Υ)-operators on Ω ×Χ is denoted by op( Ξ,X Y). For fixed non-zero elements x ∈0 X, ω ∈0 , x0ω0 denotes the bilinear functional on

Y

Ξ× , defined by ξ(x0ω0)y=ξx0ω0y for (ξ,y)∈ Ξ ×Y. An operator

( ,Y X)

op

B Ξ such that ABA =A, BAB =B is said to be a generalized inverse of an operator Aop( Ξ,X Y). The value of a (µ+m)-linear functional D:Ξµ×Υ mK, µ,mN{ }0 , at a point

(ξ ,K, ξµ,y ,K, ym)Ξµ×Υ m

1

1 is denoted by 



ym

, , y

, D ,

K K

1

1 ξµ

ξ . A (µ+m)-

-linear functional D on Ξµ×Υm is said to be bi-skew symmetric if it is skew symmetric in variables from both Ξ , and Y. A (µ+m)-linear functional

K :

D Ξµ×Υ m is said to be(Ω ,Χ)-functional on Ξµ×Υm, if for any fixed elements ξ ,K,ξi ,ξi+ ,K,ξµΞ

1 1

1 (i=1,K,µ), y ,K,ymΥ

1 there exists an ele-

ment xiΧ such that 



= +

m i

i

i y , , y

, , , , , D , x

K K K

1

1 1

1 ξ ξ ξ ξµ

ξ ξ for every ξ ∈Ξ and

(3)

for any fixed elements ξ1,K,ξµΞ , y1,K,yj1,yj+1,K,ymΥ (j=1,K,m) there exists an element ω ∈j such that

=

+

j m

j

j y , , y ,y,y , , y

, D ,

y

K K

K

1 1 1

1 ξµ

ω ξ

for every yΥ .

A sequence ( )Dn n∈N{ }0 is called a determinant system for operator

( ,X Y)

op

A Ξ , if for n∈ N{ }0

D is bi-skew symmetric n (,Χ)- -functional on Ξµn ×Υmn where n,m m n

n

n =µ0+ = 0+

µ , min µ( 0,m0)=0, there exists r∈ N{ }0 such that Dr 0 and the following identities hold:

( )

= +

+

=

n

n n

n n

m

j j j m

n j j m

n y , , y ,y , , y

, D ,

y y , , y

, , , D A

0 0 1 1

1 0

1

1 1

K K

K K

K µ ξ ξµ

ξ ω ξ

ω ,

( )

=

+

+

=

n

n n

n n

i m

i i n

i i m

n y , , y

, , , , D ,

y x , , y , Ax

, D ,

µ

µ

µ ξ ξ ξ ξ

ξ ξ ξ

0 1

1 1 0

1 0

1 1

K K K

K K

, where

( µ ) Υ ( ) Χ ω

Ξ

ξi i=1,K, n ,yj j=1,K,mn , x , . The least r∈ N{ }0 , such that Dr 0 is called the order of determinant system ( )Dn n∈N{ }0 . The integer

0

0 m

µ is called the index of determinant system ( )Dn n∈N{ }0 . If A ∈op(Ξ,Χ Υ ) is a Fredholm operator of order r=min{n,m} and index d =nm, Bop(Ξ ,Υ Χ) is a generalized inverse of A,

{z1,K,zn }

, {ς1,K,ς }

m are complete systems of solutions of the homogenous equations Ax=0 and ωA=0, respectively, then the sequence ( )Dn n∈N{ }0 defined by the formula

,

y y

y y

z z

By By

z z

By By

y , , y

, D ,

r m n m m

r m n

n r n n r

n n r m n r n n r

n n

n r

m n

r m n

r n n n

0 0

0 0

1

1 1

1

1 1

1 1

1 1

1 1

1 1

K K

M M

M M

K K

K K

M M

M M

K K

K K

+

+

+

+

+

+

+

+

+

+ =



ς ς

ς ς

ξ ξ

ξ ξ

ξ ξ

ξ ξ

ξ ξ

for ξiΞ (i=1,K,n+nr),yjΥ (j=1,K,n+mr), is a determinant system for the operator A.

A linear functional F:op(Ξ ,Υ Χ)K is said to be an algebraic quasinucleus on op(Ξ ,Υ Χ), if there exists TFop(Ξ,Χ Υ)

such that F(xω =) ωTFx for (ω,x)×Χ. T is called a quasinuclear operator F

(4)

determined by F. The space of all algebraic quasinuclei on op(Ξ ,Υ Χ) is denoted by an(Ξ,X Y). If Υ ξ Ξ

0

0 ,

y are fixed, then algebraic quasinucleus ξ0y0an( Ξ,Χ Υ ) such that (ξ0y0)( )B =ξ0By0 for

(Ξ Υ Χ)

op ,

B is called one-dimensional. Every finite sum

=

n

i

i

i y

1

ξ of one-dimensional quasinuclei is called finitely dimensional quasinucleus. By the trace of an algebraic quasinucleus Fan(Ξ Ξ,Χ X) we understand the number TrF =F( )I . For Fan(Ξ,Χ Υ ) and Cop(Λ,Υ Ζ)

we define CFan(ΛΞ,Χ Ζ):

(CF)( )A =F(AC) for Aop(Ξ Λ,Ζ Χ). (1) Let D be a bi-skew symmetric (,Χ)-functional on Ξµ×Υm, µ,mN, and

( Ξ Χ Υ)

an ,

F . Fixing all the variables ξ2,KξµΞ , Υ ym

, , y K

2

and interpreting D as the function of variables ξ1,y1 only, we define (µ+m2)- -linear functional Fξ1y1D on Ξµ1×Υm1 by

( )

F( )A

y , , y

, D ,

F

m

y =



K K

2 2 1 1

µ ξ

ξ

ξ ,

where





=

ym

, , y , y

, , D ,

Ay

K K

2 1

2 1 1

1

ξµ

ξ

ξ ξ for ξ1Ξ, y1Υ .

We can iterate the procedure k times, k=min{µ,m}, provided D F F F

, D F F , D

F y y y y y y

k

k 2 2 1 1

1 1 2 2 1

1 ξ ξ ξ ξ ξ

ξ K K

are (,Χ)-functionals [12]. By a reasoning similar to that in [6], since D

F

F y y

k

k τ τ1 τ1

τ ξ

ξ K does not depend on the choice of permutation τ of integers k

, ,K

1 , we denote by F F D

times k

43 42

1K





 the common value of all F y F y D

k

k τ τ1 τ1

τ ξ

ξ K . We also use the notation suitable for the formulation and the proof of the main theorem of the paper. A matrix

[ ]

m j ij i

a M

=

11 µ over the field K , is denoted by

(M ,K,Mµ)T

1 , where Mi=[ai1,K,aim],(i=1,K,µ), i.e. M is the i-th row of i M .

(5)

2. Algebraic formulas for terms of determinant systems for quasinuclear perturbations of Fredholm operators

We present the theorem, which gives a generalization of Plemelj-Smithies formulas for operators of the form S + , where S is Fredholm and T T is quasi- nuclear.

Theorem. Let Sop(Ξ,X Y) be a Fredholm operator of order

{n,m}

min

r= , index d =nm0 and determinant system ( )Dn n∈N{ }0 . Sup- pose that Uop(Ξ ,Υ Χ) is a generalized inverse of S and {z ,K,zn'}

1 ,

{ς ,K,ςm'}

1 are complete systems of solutions of the homogenous equations Sx=0 and ωS =0, respectively. Then for any Fan(Ξ,Χ Υ ) which determines

(Ξ Χ Υ )

=T op ,

TF the following formulas hold:

1 2 1

1 2

1 1

1 1 0

1 0 0 1

0 0 0

σ σ σ

σ

σ σ

σ σ

K K

M M M

M M

K K

43 42 1K

=

+

k k k n

k k k n

n n

times k

T T

k T

k T D F

F   n k (2)

for n,kN{ }0 , where

( )

[

UT UF

]

(m , k)

Tr m

m= 1 =1K

σ (3)

n

n D

T =0 and T is the nm (2n+n'+m'2r)-linear functional (4)

=

+

+

r m n

r n m n

n y , , y

, T ,

K K

1

1 ξ

ξ

( ) ( ) ( ) ( )

( ) ( ) ( ) ( )

( ) ( ) [ ( )]( ) [ ( )]( )

( ) ( ) [ ( )]( ) [ ( )]( )

,

z UT

T z

UT T y

TU y

TU

z UT

T z

UT T y

TU y

TU

z UT z

UT Uy

UT Uy

UT

z UT z

UT Uy

UT Uy

UT

n i ' m i

' m r m n i m i

m

n i i

r m n i i

n i r n n i

r n n r m n i r n n i

r n n

n i i

r m n i i

' m r ' n n '

m r ' n n '

m r ' n n '

m r ' n n

r ' n n r

' n n r

' n n r

' n n

r ' n n r

' n n r

' n n r

' n

n

+

+

+

+

+

+

+

+

+

+ +

+ +

+ +

+

+

+ +

+ +

+ +

+

+

+

+

+

=

ς ς

ς ς

ς ς

ς ς

ξ ξ

ξ ξ

ξ ξ

ξ ξ

K K

M M

M M

K K

K K

M M

M M

K K

1 1

1 1

1 1

1 1

1 1

1 1

1 1

1 1

1 1

1 1

1 1

1 1

where for s=1,Km', t=1,Kn'

(6)

( )

[ ]( )

( )

=

= =

+

+

+

+ +

+ +

+

m , i

if z UT

T

i z if

UT T

s r ' n n t

i s

s r ' n n t

i

s nn' r s

s r ' n n

1K 0 0

ς 1

ς

and is extended over all finite sequences of non-negative integers

' m r ' n

in

,

i K + +

1 , such that i i m

' m r ' n

n =

+ +K + +

1 ; i.e.

=



+

+

r m n

r n m n

n y, , y

, T ,

K K

1

1 ξ

ξ

( ) ( ) [ ( )]( ) [ ( )]( )

= +

+ +

+ +

+

+

+ +

+

+

=

m i

i nm r n

i m

i i

r n n i

n r d n

m r n n r

n n r

n n

z ,

, z / y ,

, y

UT T , , UT

T / UT ,

, D UT

K

K K

K K

1

1 1

1 1

1

1 ξ ς ς

ξ

where

( ) ( ) [ ( )]( ) [ ( )]( ) =

+

+ + ++ ++

n r

m n

i m

i i

r n n i

n y , , y /z , , z

UT T , , UT

T / UT ,

, D UT

m r n n r

n n r

n n

K K

K K

1 1

1

1 1 ξ ς 1 ς

ξ

( ) ( )

, m i

i y if

, , y

UT ,

,

D UT n n r

r m n

i r n n i

n

r n

n + + =

=

+

+

+ +

K K

K

1 1

1 1 ξ

ξ

and if i i m

r n n <

+ +K +

1 , then

( ) ( ) [ ( )]( ) [ ( )]( ) det(D( ), ,D( ) ) ,

z ,

, z / y ,

, y

UT T , , UT T / UT ,

,

D UT n nnn r m T

n r

m n

i m i

i r n n i n

m r n n r

n n r

n n

+

+

+

+ =

+ ++ ++ ′′

K K

K

K K

1 1

1

1

1 1 ξ ς 1 ς

ξ

where

( )

[

( ) ( ) ( ) ( ) n

]

i j i

j r m n i j i

j n

j UT Uy, , UT Uy , UT z, , UT z

D =ξ j Kξ j + ξ j Kξ j

1

1 (j=1,K,n+nr)

( )

=

+ +

+ K 12K3

n r m n j j

n j r n

n y , , y , , ,

D 0 0

1 ς

ς if in+nr+j=0 (j=1,K,m),

( ) [ ( ) ( ) ( ) ( ) n]

i j i

j r m n i j i

j n

j r n

n TU y , , TU y , TUT z , , TUT z

D++ =ς n+nr+j 1Kς n+nr+j + ς n+nr+j1 1Kς n+nr+j1 if in+nr+j =1,K,m(j=1,K,m).

Proof. We shall use an induction argument on k to establish the formulas. If k = 0, then D =n Tn0 for any n∈ N{ }0 . Let k be any fixed positive integer.

Assume that for any n∈ N{ }0 the formulas (2) hold. For fixed n

Cytaty

Powiązane dokumenty

Wcześniej stypendysta Akademii w Bonn i Fundacji Tyssena, cieszył się sławą wy- bitnego specjalisty w zakresie historii nauki, którego głównym tematem zaintereso- wania

The real gnostic in Christ in Irenaeus had belief in the great orthodox doc- trines of unity: One God, who is the Father and Creator of all things, immate- rial and material, and

Podwójne widzenie rzeczywistości, choć nie do końca spójne, ale przecież analogiczne, tworzyło swoisty rodzaj wizualnego dialogu między różnymi osobami i przestrzeniami

B ie le ck i, Une remarque sur la méthode de Banach-Cacciopoli-Tihhonov dans la théorie des équations différentielles ordinaires,

We shall use, in this section, the admissibility theory of pairs of function spaces, in order to find existence (and uniqueness) results for some classes of nonlinear

Z Zakładu Matematycznych Metod Fizyki Zespołowej Katedry Matematyki Wydziału

The aim of this paper is to give a new existence theorem for a stochastic integral equation of the Volterra-Fredholm type of [9] and [10] (cf. also [13]) and to investigate

zrozumiałe, zważywszy na fakt, że początki rządów Augusta to przede wszyst- kim okres krystalizowania się pozycji princepsa w nowych realiach politycz- nych i