ON RANDOM

467

ON A PAIR OF RANDOM GENERALIZED NON-LINEAR CONTRACTIONS

J. ACHARI

Science College Nanded (Maharashta), INDIA

(Received March 9, 1981 and in revised form March 8, 1982)

ABSTRACT. A fixed point theorem for a pair of random generalized non-linear contrac- tion mappings involving four points of the space under consideration is proven. It is shown that this result e!udes the result of Lee and Padgett [i]. ^Also an application of the result is given.

KEY WORDS AND PHRASES. Complete probability measure space, Banach spaces, -algebras, Borel subsets, random variable, random operator, separable Banach space, random gen- eralized

nonlinear

ntraion, upper semicontinuous functns, Bochner integral, unique fixed ^point.

1980 MATHEMATICS SUBJECT CLASSIFICATION CODES. Primary 60H99, 4ZHO; Secondary 60H20.

1. INTRODUCTION.

The idea of fixed points plays a very important role in solving deterministic operator equations. Recently the idea of random fixed point theorems which are the stochastic generalization ^of the classical fixed point theorems has become a very ira- portant part of the theory of some operator equations which can be regarded as random operator equations. Many interesting results have been established by various authors

(see for example Bharucha-Reid

[2],

^Hans

[3],

^Padgett

[4],

^Tsokos

[5],

Tsokos and Pad- gett

[6],

Lee and Padgett

[7])

in this area.

Recently, a fixed point theorem for a pair of generalized non-linear contraction mappings involving four points of the space under consideration, which includes many well known results as special cases has been established by Achari

[8]

(see also Achari

[9],

Pittanuer [i0]).

The object of this paper is to study a stochastic version of a pair of generalized non-linear contraction mappings of Achari

[8].

Also it has been shown that this result

generalizes the result of Lee and Padgett [i]. It is interesting to note that with suitable modification of the conditions of the theorem, we can easily obtain stochastic generalizations of the results of different classical fixed points. Finally, we apply Theorem 2 to prove the existence of a solution in a Banach space of a random nonlinear integral equation of the form

x(t;) h(t;)

+ k(t,s;)f(s,x(s;))d(s)

(I.i)

S

where S is a locally compact metric space with metric d defined on S S, D is a corn- plete -finite measure defined on the collection of Borel subsets of S and the inte- gral is a Bochner integral.

2 PRELIMINARIES

In this section, we state some definitions as used by Lee and Padgett [i]. Let

(,S,P)

^{be a} complete probability measure space, and let

(X,)

and

(Y,C)

be two mea- surable spaces, where X and Y are Banach spaces and B and C are -algebras of Borel subsets of X and Y, respectively. First, we state the usual definitions of a Banach space-valued random variable and of a random operator.

DEFINITION i. A function V:

-

^{X is said to be an} X-valued random variable (Random element in

X,

or generalized random variable) if

{ :

V()

B}

S for each B

.

DEFINITION 2. A mapping T(): X Y is said to be a random operator if y() T(m)x is a Y-valued random variable for every x X.

DEFINITION 3. Any X-valued random variable x(o) which satisfies the condition

P({:

r()x(Io)

y()})

i

is said to be a random solution of the random operator equation T()x y().

DEFINITION 4. Let

yj(),

^{j 1,2, n} be second order real valued random vari- ables on a probability space

(f,S,P),

that is

yj() ^L2(,S,P).

^{The collection of all}

n-component random vectors

y’

()

(YI(I)’ Yn

⁽⁾⁾ constitutes a linear vector space if all equivalent random vectors are identified. Define the norm of y by

1/2 max

;IYjl

^max

lyjl2dp)

L2 l<j-<n l<j<n

The space of all n-component random vectors y with second-order components and norm given by

I’ll

_n ^{above is separable Banach space and will be denoted by}

Lm(,S,P

n ^or

L2

simply Ln

2. Let S be locally compact metric space with metric d defined on S S and let be a complete o-finite measure defined on the Borel subsets of S.

n(,S,P))

to be the space of all continu- DEFINITION 5. We define the space

C(S,L

2

n(,S

P) with the topology of uniform convergence on com- ous functions from S into L

2

pacta. It may be noted that

C(S,L2(,S,P))

n ^{is locally convex space whose topology is} defined by a countable family of seminorms given by

sup

Ix(t;)II _n’

^{j 1,2}

Ix(t;) lJ

tecj ^e

2

DEFINITION 6. Let B and D be Banach spaces. The pair (B,D) ^{is said} to be ad- missible with respect to a random operator U() if U()(B) D.

DEFINITION 7. A random operator

T()

on a Banach space X with domain D(T()) is said to be a random generalized nonlinear contraction if there exists non-negative real-valued upper semicontinuous functions

i()’

^{i 1,2} 5 satisfying

i

^()(r)< r for r > 0,

i()(0)

0 and such that

IIT(oJ)x

_I

T(oo)x211 ^-< ql(llx

I

x211) ^{+ q2(I Ix}

I

r(o)XllI) ⁺ q3(Ilx

2

+ 4(I Ix

I

T()x21 ^{I) +} 5(I ^Ix

2

T()Xll ^I)

for all

Xl,X

2 D(T()).

3. A FIXED POINT THEOREM FOR A PAIR OF RANDOM GENERALIZED NONLINEAR CONTRACTIONS.

THEORY.! i. Suppose

AI(

^{and A}2() are a pair of random operators ^{from a} separ- able Banach space X ^into itself such that

IAl()x

I

Am()x211

^<

@l(IIx

I

x211) ⁺ 2(IIx

I

Al()x31 ^{I) +} 3(Ilx

2

Am()x411)

+ 4(I Ix

I

A2()x41 ^{I) +} 5(I Ix

2

Al()x31 ^I)

^(3.i)

where

.

^{(), i 1,2}

^,5,

are non-negative real-valued upper semicontinuous functions satisfying

i()(r)

^< r ^{for r > 0,}

i=O()(0)

and for all

Xl,X2,X3,X

4 X. Then there exists an X-valued random variable N() which is the unique ^common fixed point of A

I()

^{and A}

2().

PROOF. Let x,y X and we define

xI

A2()y

^x₂

Al()x

^x3 x, x 4 y,

Then (3.1) takes the form

IAI()A2()Y A2()AI()xll ^-< i(I IAI()

^x

A2()Yl ^{I) +} 2(I IA _l()x A2()y ^I)

+ $3(I IAl()x A2()Yl ^I)-

^(3.2)

Let x

O X be arbitrary and construct a sequence

{x

defined by n

Al()Xn_

_I

Xn, A2()x

n

Xn+l, Al()Xn+

I

Xn+2,

^{n 1,2}

Let us put x x_1 and y ^{x in}

(3.1),

then we have

n+/- n

IA

_I^()A₂

()Xn-A

2()A

I

()Xn_

_I

II

^<

_i ^(I IA l()xn_l-A

₂

()Xnll) 2 ^{(I IA}

I

()Xn_l-A

2()x n

If)

or

+ 3(I IA

₁

()Xn_

1 A 2

Xn+

²

Xn+ II

^<

^{i(I Ix}

ⁿ

^Xn+ II ^{I) + 2(I Ix}

ⁿ

^Xn+ II ^{I) + 3(I Ix}

ⁿ

Xn+ iI ^I)

^(3.3)

We take n to be even and set

n IXn-i ^{Xn II-}

^Then

n+2 II _{Xn+ Xn+}

₂

II

^<

_i (llXn _{Xn+l I)} ⁺ 2 (IIxn Xn+i I) + 3 ^{(I Ix}

ⁿ

^Xn+

¹

<

i ^{(5,+i) +} 2 ^{(n+l) +} @3 (an+l)"

^(3.4)

From (3.4) it is clear that a decreases with n and hence a a as n 0%

n n

Then since

i

^{is upper} semicontinuous, ^{we obtain in} the limit as n a ^_<

l(a) ⁺ 2

⁽⁾

⁺ 3

^{() <} 3 ^e which is impossible unless

O.

LetS> 0.

Now,

we shall show that

{x

is a Cauchy sequence. If not, then there is an e > 0 and n

for all positive integers k, there exist

{m(k)}

^and

{n(k)}

with m(k) > n(k) ^>_ k, such that

dk

lXm(k)

^xn(k)

ll

^{> E.}

We may assume that

lXm(k)_l

^xn(k)

ll

^{< g,}

by choosing m(k) to be the smallest number exceeding n(k) for which (3.5) holds.

(3.5)

Then we have

<

]Xm(k) Xm(k)_iI ⁺ lXm(k)_

1

Xn(k) ll

<

m(k) ⁺

^<

⁺

^E

which implies that d

k ^{g as} k

.

^Now the following cases are to be considered.

(i) m is even and n is odd, (ii) m and n are both odd,

(iii) m is odd and n is even, (iv) m and n are both even.

"e (i)

dk

lx

_m

Xnll

^<

^lx

^m

Xm+lll ⁺ lXm+

¹

Xn+iIl ^{+ fix}

ⁿ

<

r+l ⁺ n+l ⁺ IIAl()Xm-

By putting x

I

Xn,

^x₂

Xm,

^x₃

Xn_l,

^x₄

Xm_

₁ ^{in (3.1), we have}

<

m+l ⁺ n+l ⁺ }i(I IXm- ^{Xnll) + 2(I} IXn ^{Al()Xn-iI I) + 3(I} IXm ^{A2()Xm-ll I)}

+ }4(I Ix

_n

Am()Xm_ll ^{I) +} 5(I Ix

_m

Al()Xn_ll ^])

-< m+l ⁺ n+l ⁺ 91() ⁺ 94(dk) ⁺ 5 ^(dk)"

Letting k we have

g

<g.

3

This is a contradiction if g > 0. In the Case (ii), we have

dk

lx

_m ^x

nll< ^lx

^m

Xm+lll ⁺ lXm+

2

Xm+iII ⁺ lXm+

²

Xn+lll ^{+ lx}

ⁿ

Xn+lll

<

m+2 ⁺ m+l ⁺ n+l ⁺ IIA2()Xm+l- Al()Xnl ^I.

By putting x

I ^x

n,

^x2

Xm+

^{I, x}3

Xn_

I, x

4

Xm

^{in (3.1), we get}

<

em+2 ⁺ m+l ⁺ n+l ⁺ 91(I IXm+

¹

^{XnIl) + 2(I Ix}

^{n A}

^{l()xn_ ilI)}

+ 3(llXm+

I

A2()Xmll) ⁺ 4(IIx

n

A2()Xmll) ⁺ 5(llx

m

Al()Xn_ll ^I)

<

m+2 ⁺ m+l ⁺ n+l ⁺ i ^{(dk + am+l) +} 4 ^{(dk + am+l) + 5(dk + am+l)"}

Letting k in the above inequality we obtain < 3

g,

which is a contradiction if g > 0. Similarly, the cases (iii) and (iv) may be disposed of. This leads us to con- clude that

{x

is a Cauchy sequence. Let N() ^{be the limit} of the sequence. We shall

n

now show that

Al(00)(

⁽⁾

A2()(

^{). Putting x}_I

Xn_l,

^x₂ ^{(), x}₃

Xn+l,

x4 x

n,

ⁱⁿ

(3.1),

^we get

l()Xn_l A2()N()I

^<

i(I IXn_

₁

^{n()ll) +} 2(I IXn_

_I

Al()Xn+ll ^I)

+ 3(I ^In() A2()Xnl ^{I) +} 4(I IXn_

_I

A2()Xnl ^{I) +} 95(I ^In() Al(m)Xn+ll ^I).

Letting n o% ^we get

{N(m) A2<)N<)I

^{< 0 which is a} contradiction and hence N()

A2()N().

^{In the same way, it is possible to show that N()}

AI()N().

Thus

N()

is a common fixed point of

AI()

^and

A2().

^{Suppose there is another fixed}

point

() #

() ^{of A}

I()

^and

A2(

^). Then putting x

I ^x4 () and x

2 ^x4

()

in

(3.1),

we have

If:i() ()II

^<

i(I ^{I() ()I I) +} 2(I ^{I() ()I I) +} 3( ^{() () )}

3

which is a contradiction. Hence ()

().

This completes the proof. If in Theo- rem

lwe

^{put A}

I()

^A

2()

^{A() and x}_I ^x3 x, ^x

2 ^x4 y then we have the fol- lowing theorem which we only state without proof.

THEOREM 2. If A() is a random generalized nonlinear contraction from a separ- able Banach space X into itself, then there exists an X-valued random variable which is the unique fixed point of A().

We now have the following corollary of Theorem 2.

COROLLARY i. If

Ab()

is a random generalized contraction from X into itself for some positive integer b, then A() has a unique fixed point () which is an X-valued random variable.

PROOF. Since

Ab()

is a random generalized nonlinear contraction operator on

X,

by Theorem 2, there exists a unique X-valued random variable () ^such that

A

D()N()

N().

We claim that A()() (). If not, consider

3 ^{IB (m)-#}

^(m)(m)

⁾⁴ 11A(m)(m)-#

(m) (m)

)5 11 (m)-#+l

^{(m) (m) (3.6)}

Moreover,

the left hand side of (3.6) is

From (3.6)

and

(3.7),

we have

(3.7)

IA()()

()

-< : ^{(I IA()()-()I I+3(I} IA()()-() I)+

₅

(I

which is a contradiction and hence A()() ().

We remark that under the conditions A

I()

^A

2()

^{A() and x}_I ^x3 x, x

2 x

4 ^y,

l ’ ^j()(r)

^{0, j 2,3,4,5,} the Theorem i reduces to the following corollary.

COROLLARY 2. (Lee and Padgett [i]). If

A()

is a random nonlinear contraction operator from a separable Banach space X into itself, then there exists an X-valued random variable () which is the unique random fixed point of A().

4. APPLICATION TO A RANDOM NONLINEAR INTEGRAL EQUATION.

In this section we give an application of Theorem 2 to a random nonlinear integral equation. To do so we have followed the steps of Lee and Padgett

[I]

with necessary modifications as required for the more general settings. We shall assume the following conditions concerning the random kernel k(t,s;). The function k(.,.;.): S ^x

Sx-R

is such that

(i)

k(t,s;00):

S ^x S

L(,S,P)

^{such that}

II Ik(t,s;)lll ^n(,S,p)) lx(s;)ll

where for each^{Ln2 is}

-

integrable with respect to s S for each t S and x C(S,L 2

(t,s)

S ^x S

II ^Ik(t,s;) II _{Lo (fl, S;P)}

is the norm in

Loo(,S,P)

(ii) for each s S,

k(t,s;)

is continuous in t S from S into

Lo(,S,P);

^for

each t e S,

k(t,s;)

is continuous in s S from S into

L(fl,S,P);

^and

(iii) there exists a positive real-valued function H on S such that

n( S,P))

and such that for each t s S

H(s)[[x(s;)l

^is -integrable for x C(S,L 2

111k(t,u;

)

k(s,u;)IIl Ix(u;)[l

_n ^<

H(u)[Ix(u;)l[

_n

L2 L

2

n(,S,P)

We now define the Thus, .for ^each

(t,s)

S x S, ^we have

k(t,s;)x(s;)

L

2

n(,S,P))

by random integral operator T() on

C(S,L

2

[T

(m)

x]

(t

;) k(t,s;)x(s;)d(s)

^(4.1)

S

where the integral is a Bochner integral.

Moreover,

^we have that for each t e

S, [T(m)x](t;)

L

2n(,S,P)

^{and that is} a continuous linear operator from

C(S,L2(fl,S,P))

n into itself. We now have the following theorem.

THEOREM 3. We consider the stochastic integral equation (!.i) subject to the following conditions

(a) B and D are Banach spaces stronger (cf. [i]) then

C(S,L2Q,S,P))

n ^{such that}

(B,D)

is admissible with respect to the integral operator defined by

(4.1);

(b) x(t;)

- ^f(t,x(t;))

Q(O)

{x(t;m):

^{is an operator}

x(t;)

^{from the}

D, l[x(t;m)

^set

l[

_D

0}

into the space B satisfying

<

i()(I ^Ix(t;

⁾

y(t;)llD ⁺ 2()(I ^Ix(t;)

lf(t,x(t;)) f(t,y(t;))ll

B

f(t,x(t;))l ID)+3()(] ]Y(t;)-f(t,y(t;))ll D) ⁺ l()(] Ix(t;)-f(t,y(t;))l D)

+ 5()(I lY(t;)-f(t,x(t;))l D)

for

x(t;),

y(t;) Q(O), where

i ^()’

^{i 1,2 5 are} non-negative real-valued upper semicontinuous functions satisfying

i()(r)

^< r ^{for r > 0 and}

i()(0)

^0;

(c) h(t;) D.

Then there exists a unique random solution of (i.I) in Q(0), provided c() < i and

+ 2c() If(t;0)

_B ^{< 0(i- c()) where c() is the norm of T().}

PROOF. Define the operator U() from Q(p) into D by

[U()x](t;) h(t;) +

|

k(t,s;)f(s;))d(s).

S Now

<

lh(t;)l ⁺ c()llf(t x(t;))ll

_B

II [U()x](t;)]l

_{D D}

<

lh(t;)llD ⁺ c()llf(t;0)ll

_B

Then from the conditions of the theorem

+ c()llf(t,x(t;)) ^f(t;0) IIB.

c()llf(t,x(t;))-f(t,0)ll

B ^< c()[

I() (I Ix(t;)ll

D

+ 2()(I Ix(t;)-f(t,x(t;))ll D)

Hence i.e.

+ 3(m)(llf(t;0)l ID) ⁺ 4()(llx(t;)llD) ⁺ 5()(llf(t,x(t;))llD),

3 < 3 3

c()

If(t,x(t;)

f(t’0)

B c()p

+ c()l ]f(t;0) 1B

<

lh(t;)ll + 2c()llf(t;0)ll

_B

+

c()p

l[g()x](t;m)ll

D D

< p(1 c())

+

c()p

<

Hence

[U()x](t;)

e Q(O).

Now,

for

x(t;),

y(t;) e Q(O) we have by condition (b)

l[U()x](t;) [U(o)y](t;o)II

D

Ill k(t,s;oo)[f(s,x(s;o)) f(s,y(s;o))ldD(s)ll

S D

<

c()[ I()(I ^{Ix(t’) Y(t;)l]} D)

<

c()l If(t,x(t;)) f(t,y(t;))ll

B

+ 2()(llx(t;) ^f(t,x(t;)) D) ⁺ 3()([ ^lY(t;

^{) f(t,y(t;))}

D)

+ 4()(] ^Ix(t;) f(t,y(t;))l + 5()(lly(t;) f(t,x(t;))llD)

<- i()(I ^Ix(t;) -y(t;)llD ⁺ 2()(llx(t;) f(t,x(t;))l

D

+ 3 ^{()(I ly(t;)} f(t’y(t;))I ID) ⁺ 4 ^{()(I Ix(t;}

⁾

f(t,y(t;))I D)

+ 5()(I ^ly(t;) f(t,x(t;))l D)

since c()

-<

i. Thus U() is a random nonlinear contraction operator on Q(p). Hence, by Theorem 2 there exists a unique X-valued random variable x*(t;) e Q(p) which is a fixed point of U(), that is x*(t;) is the unique random solution of the Equation (1.1).

ACKNOWLEDGEMENT. The author is thankful to the referee for his valuable comments.

This work was supported by UGC Grant No. F. 25-9

(11658)/80.

REFERENCES

i. LEE, ARTHUR C.H. and

PADGETT,

W.J. On random nonlinear contraction, Math.

Sstems

^{Theory ii}

^(1977),

^77-84.

2. BHARUCHA-REID, A.T. ^Random

Integral.

Equations, ^Academic Press, New York, (1972).

3. HANS, O. Random fixed point theorems, Trans. First Prague Conf. on Information

Theory,

Statist. Decision Functions on Random Process, (1957), 105-125.

4.

PADGETT,

W.J. On a nonlinear stochastic integral equations of Hammerstein type, Proc.

Amer.

th. Soc. 38

(1973),

625-631.

5. TSOKOS, C.P. On a stochastic integral equation ^of Voltera type, Math.

Systems

Theory ³ (1969), 222-231.

6.

TSOKOS,

C.P. and PADGETT, W.J. Random Integral Equations with Applications in Life Sciences and Engin.eer.ing ^Academic Press, New York, (1974).

7. LEE, A.C.H. and PADGETT, W.J. On a heavily non-linear stochastic integral equa- tions, Utilitas Mathematica 9 (1976), 128-138.

8.

ACHARI,

J. Fixed point theorems in complete metric spaces, Resultate der Mathematik 3 (1980), 1-6.

9.

ACHARI,

J. A result on fixed points, Kyungpook Math. Journal 19 (1979), 245-248.

i0.

PITTNAUER,

F. A fixed point theorem in complete metric spaces, to appear in Periodica Math. Hungarica.

Special Issue on

Intelligent Computational Methods for Financial Engineering

Call for Papers

As a multidisciplinary field, financial engineering is becom- ing increasingly important in today’s economic and financial world, especially in areas such as portfolio management, as- set valuation and prediction, fraud detection, and credit risk management. For example, in a credit risk context, the re- cently approved Basel II guidelines advise financial institu- tions to build comprehensible credit risk models in order to optimize their capital allocation policy. Computational methods are being intensively studied and applied to im- prove the quality of the financial decisions that need to be made. Until now, computational methods and models are central to the analysis of economic and financial decisions.

However, more and more researchers have found that the financial environment is not ruled by mathematical distribu- tions or statistical models. In such situations, some attempts have also been made to develop financial engineering mod- els using intelligent computing approaches. For example, an artificial neural network (ANN) is a nonparametric estima- tion technique which does not make any distributional as- sumptions regarding the underlying asset. Instead, ANN ap- proach develops a model using sets of unknown parameters and lets the optimization routine seek the best fitting pa- rameters to obtain the desired results. The main aim of this special issue is not to merely illustrate the superior perfor- mance of a new intelligent computational method, but also to demonstrate how it can be used e

ff

ectively in a financial engineering environment to improve and facilitate financial decision making. In this sense, the submissions should es- pecially address how the results of estimated computational models (e.g., ANN, support vector machines, evolutionary algorithm, and fuzzy models) can be used to develop intelli- gent, easy-to-use, and/or comprehensible computational sys- tems (e.g., decision support systems, agent-based system, and web-based systems)

This special issue will include (but not be limited to) the following topics:

• Computational methods

: artificial intelligence, neu- ral networks, evolutionary algorithms, fuzzy inference, hybrid learning, ensemble learning, cooperative learn- ing, multiagent learning

• Application fields

: asset valuation and prediction, as- set allocation and portfolio selection, bankruptcy pre- diction, fraud detection, credit risk management

• Implementation aspects

: decision support systems, expert systems, information systems, intelligent agents, web service, monitoring, deployment, imple- mentation

Authors should follow the Journal of Applied Mathemat- ics and Decision Sciences manuscript format described at the journal site

http://www.hindawi.com/journals/jamds/.

Prospective authors should submit an electronic copy of their complete manuscript through the journal Manuscript Track- ing System at

http://mts.hindawi.com/, according to the fol-

lowing timetable:

Manuscript Due December 1, 2008 First Round of Reviews March 1, 2009 Publication Date June 1, 2009

Guest Editors

Lean Yu,

Academy of Mathematics and Systems Science, Chinese Academy of Sciences, Beijing 100190, China;

Department of Management Sciences, City University of Hong Kong, Tat Chee Avenue, Kowloon, Hong Kong;

[email protected]

Shouyang Wang,

Academy of Mathematics and Systems Science, Chinese Academy of Sciences, Beijing 100190, China; [email protected]

K. K. Lai,

Department of Management Sciences, City University of Hong Kong, Tat Chee Avenue, Kowloon, Hong Kong; [email protected]

Hindawi Publishing Corporation http://www.hindawi.com