Abstract
In this paper, we present some results on the existence of random coincidence points of expansive type completely random operators. Some applications to random fixed point theorems and random equations are given.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
Let \((\Omega , {\mathcal {F}}, P)\) be a probability space, \(X, Y\) be separable metric spaces and \(f:\Omega \times X\rightarrow Y\) be a random operator in the sense that for each fixed \(x\) in \( X\), the mapping \(f(.,x):\omega \mapsto f (\omega ,x)\) is measurable. The random operator \(f\) is said to be continuous if for each \(\omega \) in \(\Omega \), the mapping \(f(\omega ,.):x\mapsto f(\omega ,x)\) is continuous. An \(X\)-valued random variable \(\xi \) is said to be a random fixed point of the random operator \(f: \Omega \times X\rightarrow X\) if \( f(\omega ,\xi (\omega ))=\xi (\omega )\quad \text{ a.s. }\) and an \(X\)-valued random variable \(\xi \) is said to be a random coincidence point of the random operators \(f,g: \Omega \times X\rightarrow X\) if \( f(\omega ,\xi (\omega ))= g(\omega ,\xi (\omega ))\,\,\text{ a.s. }\)
The theory of random fixed points and random coincidence points is an important topic of the stochastic analysis and has been investigated by various authors (see, e.g. [2–5, 14–18]).
In this paper, we are concerned with mapping \(\Phi :L_0^X(\Omega ) \rightarrow L_0^Y(\Omega )\). Since a random operator \(f\) can be viewed as an action which transforms each deterministic input \(x\) in \( X\) into a random output \(f(x) \) in \( L_0^Y(\omega )\) while \(\Phi :L_0^X(\Omega ) \rightarrow L_0^Y(\Omega )\) can be viewed as an action which transforms each random input \(u \) in \(L_0^X(\Omega )\) into a random output \(\Phi u\), we call \(\Phi \) a completely random operator. In the Sect. 2, we present some properties of completely random operators. Section 3 deals with the notion of random coincidence points of completely random operators and gives some conditions ensuring the existence of a random coincidence point of expansive type completely random operators. It should be noted that the existence of a random coincidence point of completely random operators does not follow from the existence of corresponding deterministic coincidence point theorem as in the case of the random operator. In the Sect. 4, some applications to random fixed point theorems and random equations are presented.
2 Some Properties of Completely Random Operators
Let \((\Omega , {\mathcal {F}}, P)\) be a complete probability space and \(X\) be a separable Banach space. A mapping \(\xi :\Omega \rightarrow X\) is called an \(X\)-valued random variable if \(\xi \) is \(({\mathcal {F}},{\mathcal {B}}(X))\)-measurable, where \({\mathcal {B}}(X)\) denotes the Borel \(\sigma \)-algebra of \(X\). The set of all (equivalent classes) \(X\)-valued random variables is denoted by \(L_0^X(\Omega )\) and it is equipped with the topology of convergence in probability. For each \(p>0\), the set of \(X\)-valued random variables \(\xi \) such that \(E\Vert \xi \Vert ^p<\infty \) is denoted by \(L_p^X(\Omega )\).
At first, recall that (see, e.g. [22])
Definition 2.1
Let \(X,Y\) be two separable Banach spaces.
-
(1)
A mapping \(f:\Omega \times X\rightarrow Y\) is said to be a random operator if for each fixed \(x\) in \( X\), the mapping \(\omega \mapsto f (\omega ,x)\) is measurable.
-
(2)
The random operator \(f: \Omega \times X\rightarrow Y\) is said to be continuous if for each \(\omega \) in \( \Omega \) the mapping \(x\mapsto f(\omega , x)\) is continuous.
-
(3)
Let \(f,g: \Omega \times X\rightarrow Y\) be two random operators. The random operator \(g\) is said to be a modification of \(f\) if for each \(x \) in \( X\), we have \(f(\omega , x)=g(\omega , x)\ \text{ a.s. }\) Noting that the exceptional set can depend on \(x\).
The following is the notion of the completely random operator.
Definition 2.2
Let \(X,Y\) be two separable Banach spaces.
-
(1)
A mapping \(\Phi :L_0^X(\Omega ) \rightarrow L_0^Y(\Omega )\) is called a completely random operator.
-
(2)
The completely random operator \(\Phi \) is said to be continuous if for each sequence \((u_n) \) in \( L_0^X(\Omega )\) such that \(\lim u_n =u\) a.s., we have \(\lim \Phi u_n =\Phi u\) a.s.
-
(3)
The completely random operator \(\Phi \) is said to be continuous in probability if for each sequence \((u_n)\) in \( L_0^X(\Omega )\) such that \(\lim u_n =u\) in probability, we have \(\lim \Phi u_n =\Phi u\) in probability.
-
(4)
The completely random operator \(\Phi \) is said to be an extension of a random operator \(f:\Omega \times X\rightarrow Y\) if for each \(x \) in \( X\)
$$\begin{aligned} \Phi x(\omega )=f(\omega ,x)\ \text{ a.s. } \end{aligned}$$where for each \(x \) in \( X\), \(x\) denotes the random variable \(u \) in \( L_0^X(\Omega )\) given by \(u(\omega )=x\quad \text{ a.s. }\)
For later using, we list some following results.
Theorem 2.3
[24, Theorem 2.3] Let \(f:\Omega \times X\rightarrow Y\) be a random operator admitting a continuous modification. Then, there exists a continuous completely random operator \(\Phi :L_0^X(\Omega ) \rightarrow L_0^Y(\Omega )\) such that \(\Phi \) is an extension of \(f\).
Proposition 2.4
[24, Proposition 2.4] Let \(\Phi :L_0^X(\Omega ) \rightarrow L_0^Y(\Omega )\) be a completely random operator. Then, the continuity of \(\Phi \) implies the continuity in probability of \(\Phi \).
3 Random Coincidence Points of Completely Random Operators
Let \(f,g: \Omega \times X\rightarrow X\) be random operators. Recall that (see, e.g. [1, 3, 18]), an \(X\)-valued random variable \(\xi \) is said to be a random fixed point of the random operator \(f\) if
An \(X\)-valued random variable \(u^*\) is said to be a random coincidence point of two random operators \( f,g\) if
Assume that \(f,g \) are continuous. Then, by Theorem 2.3 the mappings \(\Phi ,\Psi : L_0^X(\Omega ) \rightarrow L_0^X(\Omega )\) defined respectively by
are completely random operators extending \(f\) and \(g\), respectively. For each random fixed point \(\xi \) of \(f\), we get
and for each random coincidence point \(u^*\) of two random operators \( f,g\), we have
This leads us to the following definition.
Definition 3.1
-
(1)
Let \(\Phi : L_0^X(\Omega ) \rightarrow L_0^X(\Omega )\) be a completely random operator. An \(X\)-valued random variable \(\xi \) in \( L_0^X(\Omega ) \) is called a random fixed point of \(\Phi \) if
$$\begin{aligned} \Phi \xi =\xi . \end{aligned}$$ -
(2)
Let \(\Phi _1,\Phi _2,...,\Phi _n:L_0^X(\Omega ) \rightarrow L_0^X(\Omega )\) be completely random operators. An \(X\)-valued random variable \(u^*\) in \( L_0^X(\Omega )\) is called a random coincidence point of \(\Phi _1,\Phi _2,...,\Phi _n\) if
$$\begin{aligned} \Phi _1 u^* =\Phi _2 u^*= ...=\Phi _n u^*. \end{aligned}$$(3.1)
In this section, we present some conditions ensuring the existence of a random coincidence point of completely random operators.
Theorem 3.2
Let \(\Phi ,\Psi ,\Theta :L_0^X(\Omega ) \rightarrow L_0^X(\Omega )\) be continuous in probability completely random operators, \(\Phi ,\Psi \) be surjective and \(f:[0,\infty )\rightarrow [0,\infty )\) be a mapping such that for each \(t>0\),
Assume that for any random variables \(u,v \) in \( L_0^X(\Omega )\) and \(t>0\), we have
Then, \(\Phi ,\Theta \) have a random coincidence point and \(\Psi ,\Theta \) have a random coincidence point if there exist random variables \(u_0,v_0 \) in \( L_0^X(\Omega )\) and \(p >0\) such that \(\Phi v_0=\Theta u_0\) and
Proof
Suppose that \(E\Vert \Theta v_0-\Theta u_0\Vert ^p<\infty \) for random variables \(u_0,v_0 \) in \( L_0^X(\Omega )\) such that \(\Phi v_0=\Theta u_0\) and \(p>0\). Because \(\Phi ,\Psi \) are surjective, there exists a random variable \(u_1\) in \( L_0^X(\Omega )\) such that \(\Phi u_1=\Theta u_0,u_1=v_0\). Again, there exists a random variable \(u_2\) in \( L_0^X(\Omega )\) such that \(\Psi u_2= \Theta u_1\). By induction, there exists a sequence \((u_n)\) in \( L_0^X(\Omega )\) such that
We will show that \((\xi _n)\) given by \(\xi _{n}=\Theta u_{n-1}\) \((n=1,2,...)\) in (3.5) is a Cauchy sequence in \( L_0^X(\Omega )\). Define the function \(g(t), t>0\) by
So, we have
Since \( f(t)>0\quad \forall t>0\), we get \( g(t)>1\quad \forall t>0\). For any random variables \(u,v \) in \( L_0^X(\Omega )\), we have
Equivalently,
Fixed \(t>0\). For each \( s\ge t>0\), we have
Since \(g(t)>1\), we get
Hence,
Put \(q=q(t)\), noting that \(q>1\) since \(h(t)>0\).
From this, for each \(n\), we obtain
and
By induction and Chebyshev inequality, we get
Let \(r\) be a number in \((1,q)\). Then, \(r>1\) and \((r-1)(\frac{1}{r}\,+\,\frac{1}{r^2}\,+\,...\,+\,\frac{1}{r^m})\,+\,\frac{1}{r^m}=1\ \ \ \forall m\ge 1\).
Thus, for any \(t>0\), \(n\ge 2\) and \( m\) in \( N\), we have
which tends to \(0\) as \(n\rightarrow \infty \). It implies that \((\xi _n)\) is a Cauchy sequence in \( L_0^X(\Omega )\). Hence, there exists \(\xi \) in \( L_0^X(\Omega ) \) such that p-\(\lim \xi _n=\xi \). Because \(\Phi \) is surjective, there exists \(u^* \) in \( L_0^X(\Omega )\) such that \(\Phi u^*=\xi \). So, we have
Let \(n\rightarrow \infty \), we receive \( P\left( {\left\| {\xi - \Theta {u^*}} \right\| > t} \right) = 0\) implying \(\Theta u^*=\xi \,\,a.s.\) Then, \(\Phi ,\Theta \) have a random coincidence point \(u^*\).
By the same argument, \(\Psi ,\Theta \) have a random coincidence point \(v^*\).\(\square \)
Corollary 3.3
Let \(\Phi ,\Theta :L_0^X(\Omega ) \rightarrow L_0^X(\Omega )\) be continuous in probability completely random operators, \(\Phi \) be surjective and \(f:[0,\infty )\rightarrow [0,\infty )\) be a mapping such that for each \(t>0\),
Assume that for each pair \(u,v\) in \( L_0^X(\Omega )\) and \(t>0\), we have
Then \(\Phi ,\Theta \) have a random coincidence point if and only if there exist random variables \(u_0,v_0 \) in \( L_0^X(\Omega )\) and \(p >0\) such that \(\Phi v_0=\Theta u_0\)
Proof
Put \(\Psi v=\Phi v\), then all the conditions in the Theorem 3.2 are satisfied.\(\square \)
Corollary 3.4
Let \(\Phi , \Theta \) be completely random operators satisfying the conditions stated in the Corollary 3.3. Assume that there exists a number \(q>1\) such that
for all random variables \(u,v\) in \( L_0^X(\Omega )\) and \(t>0\). Then \(\Phi ,\Theta \) have a random coincidence point if and only if there exist random variables \(u_0,v_0 \) in \( L_0^X(\Omega )\) and \(p >0\) such that \(\Phi v_0=\Theta u_0\) and (3.9) holds.
Proof
Consider the function \(f(t)=(q-1)t\) and \( h(t)= q-1 >0\). Then \(f(t)\) satisfies the conditions stated in the Corollary 3.3.\(\square \)
Remark
The following simple example shows that the random coincidence point of \(\Phi \) and \(\Theta \) in the Corollary 3.3 need not be unique.
Example 3.5
Define two completely random operators \(\Phi , \Theta : L_0^{ R}(\Omega )\rightarrow L_0^{ R}(\Omega )\) by
where \(\eta \) is a positive random variable, \(q>1\).
It is easy to check that \(\Phi ,\Theta \) satisfy all assumptions of Corollary 3.3 with \(f(t)=(q-1)t \). On the other hand, \(\Phi \) and \(\Theta \) have two random coincidence points \(u^* _1 = \frac{1}{{q - 1}}\eta ,u^* _2 = - \frac{1}{{q - 1}}\eta \).
Theorem 3.6
Let \(\Phi , \Psi ,\Theta : L_0^X(\Omega ) \rightarrow L_0^X(\Omega )\) be continuous in probability completely random operators, \(\Phi ,\Psi \) be surjective and \(f: [0,\infty )\rightarrow [0,\infty )\) be a continuous, increasing function such that \(f (0)=0,\lim _{t\rightarrow \infty } f(t)=\infty \) and \(q>1\). Assume that for any random variables \(u,v\) in \(L_0^X(\Omega )\) and \(t>0\), we have
If there exist random variables \(u_0,v_0 \) in \( L_0^X(\Omega )\) and \(p >0\) such that \(\Phi v_0=\Theta u_0\) and
Then,
-
(1)
Assume that there exists a number \( c>1/q\) such that
$$\begin{aligned} \sum _{n=1}^\infty f( c^n)<\infty . \end{aligned}$$(3.13)Then, the condition (3.12) is sufficient for \(\Phi ,\Theta \) have a random coincidence point and \(\Psi ,\Theta \) have a random coincidence point.
-
(2)
Assume that for each \(t,s>0\)
$$\begin{aligned} f(t+s)\ge f(t)+f(s). \end{aligned}$$(3.14)Then, the condition (3.12) is also sufficient for \(\Phi ,\Theta \) have a random coincidence point and \(\Psi ,\Theta \) have a random coincidence point.
Proof
Let \(g=f^{-1}\) be the inverse function of \(f\). Then, \(g: [0,\infty )\rightarrow [0,\infty )\) is increasing with \(g (0)=0,\lim _{t\rightarrow \infty } g(t)=\infty \). The condition (3.11) is equivalent to the following
Let \(u_0\) be a random variable in \( L_0^X(\Omega )\) such that (3.12) holds. Because \(\Phi ,\Psi \) are surjective, there exists a random variable \(u_1\) in \(L_0^X(\Omega )\) such that \(\Phi u_1=\Theta u_0, u_1=v_0\). Again, there exists a random variable \(u_2\) in \(L_0^X(\Omega )\) such that \(\Psi u_2=\Theta u_1\). By induction, there exists a sequence \( (u_n ) \) in \( L_0^X(\Omega )\) by
Put \(\xi _{n}=\Theta u_{n-1}, \quad n=1,2,...\). From (3.15), for each \(n\), we obtain
and
By induction, we obtain for each \(n\)
Then,
-
(1)
From (3.12), we have
$$\begin{aligned} P\left( g(\Vert \Phi u_0- \Theta u_{0}\Vert )>s\right) = P\left( \Vert \Phi u_0- \Theta u_{0}\Vert >f(s)\right) \le \frac{M}{s^p}. \end{aligned}$$(3.18)From (3.17) and (3.18), we get
$$\begin{aligned} P\left( g\left( \Vert \xi _{n + 1} -\xi _n \Vert \right) > t \right) \le \frac{M}{q^{{(n-1)}p}t^p}. \end{aligned}$$(3.19)Taking \( t=c^{n}\), from (3.19), we get
$$\begin{aligned} P\left( g\left( \Vert \xi _{n + 1} - \xi _n \Vert \right) > c^n \right) \le M\frac{1}{q^{{(n-1)}p}c^{np}}, \end{aligned}$$(3.20)i.e.
$$\begin{aligned} P\left( \Vert \xi _{n + 1} - \xi _n \Vert > f(c^n) \right) \le M\frac{1}{q^{{(n-1)}p}c^{np}}. \end{aligned}$$(3.21)Since
$$\begin{aligned} \sum _{n=1}^\infty P\left( \Vert \xi _{n + 1} - \xi _n \Vert > f(c^n) \right) \le M\sum _{n=1}^\infty \frac{1}{q^{{(n-1)}p}c^{np}}<\infty , \end{aligned}$$by the Borel-Cantelli Lemma, there is a set \(D\) with probability one such that for each \(\omega \) in \( D\) there is \(N(\omega )\)
$$\begin{aligned} \Vert \xi _{n + 1}(\omega ) - \xi _n(\omega ) \Vert \le f(c^n) \quad \forall n>N(\omega ). \end{aligned}$$By (3.13), we conclude that \( \sum _{n=1}^\infty \Vert \xi _{n + 1}(\omega ) - \xi _n(\omega ) \Vert <\infty \) for all \(\omega \) in \( D\), which implies that there exists \(\lim \xi _n(\omega )\) for all \( \omega \) in \( D\). Consequently, the sequence \((\xi _n)\) converges a.s. to \(\xi \) in \( L_0^X(\Omega )\). Because \(\Phi \) is surjective, there exists \(u^* \) in \( L_0^X(\Omega )\) such that \(\Phi u^*=\xi \). So, we have
$$\begin{aligned} \begin{array}{ll} P\left( {\left\| {\xi - {\xi _{2n }} } \right\| > f(qt)} \right) &{}= P\left( {\left\| {\xi _{2n} - \Phi u^* } \right\| > f(qt)} \right) \\ &{}= P\left( {\left\| {\Psi u_{2n} - \Phi u^* } \right\| >f(qt)} \right) \\ &{}\ge P\left( {\left\| {\Theta u_{2n} - \Theta u^* } \right\| > f(t)} \right) \\ &{}\ge P\left( {\left\| {\xi _{2n+1} - \Theta u^* } \right\| > f(t)} \right) . \end{array} \end{aligned}$$Let \(n\rightarrow \infty \), we receive \( P\left( {\left\| {\xi - \Theta u^* } \right\| > f(t)} \right) = 0 \) for all \(t>0\) implying \( \Theta u^* =\xi \) a.s. Then, \(\Phi ,\Theta \) have a random coincidence point \(u^*\). By the same argument, \(\Psi ,\Theta \) have a random coincidence point \(v^*\).
-
(2)
It is easy to see that for each \(t,s>0\)
$$\begin{aligned} g(s+t)\le g(t)+g(s). \end{aligned}$$Hence, for \( a\ge \sum _{i=1}^m s_i\), we have
$$\begin{aligned} \begin{array}{ll} P\left( g (\Vert \xi _{n+m}-\xi _n\Vert )>a\right) &{}\le P\left( g\left( \sum _{i=1}^{m}\Vert \xi _{n+i}-\xi _{n+i-1}\Vert \right) >a\right) \\ &{}\le P\left( \sum _{i=1}^{m} g(\Vert \xi _{n+i}-\xi _{n+i-1}\Vert )>\sum _{i=1}^m s_i\right) \\ &{}\le \sum _{i=1}^m P\left( g(\Vert \xi _{n+i}-\xi _{n+i-1}\Vert )>s_i\right) . \end{array} \end{aligned}$$From (3.12), we have
$$\begin{aligned} P\left( g\left( \Vert \xi _{n + i} - \xi _{n+i-1} \Vert \right) > s_i \right) \le \frac{Mq^{(n+i-1)p}}{s_i^p}. \end{aligned}$$(3.22)Put \(r\) be a number in \((1,q)\) and \(s_i=s(r-1)/r^i \). An argument similar to that in the forward proof yields
$$\begin{aligned} \lim _{n\rightarrow \infty } P(g(\Vert \xi _{n+m}-\xi _n\Vert )>s)=0 \quad \forall s>0, \end{aligned}$$so
$$\begin{aligned} \lim _{n\rightarrow \infty } P(\Vert \xi _{n+m}-\xi _n\Vert >f(s)) =0\quad \forall s>0. \end{aligned}$$Thus, we obtain
$$\begin{aligned} \lim _{n\rightarrow \infty } P(\Vert \xi _{n+m}-\xi _n\Vert > t)=0\quad \forall t>0. \end{aligned}$$Consequently, the sequence \((\xi _n)\) converges in probability to \(\xi \) in \( L_0^X(\Omega )\). Because \(\Phi \) is surjective, there exists \(u^* \) in \( L_0^X(\Omega )\) such that \(\Phi u^*=\xi \). So, we have
$$\begin{aligned} \begin{array}{ll} P\left( {\left\| {\xi - {\xi _{2n }} } \right\| > f(qt)} \right) &{}= P\left( {\left\| {\xi _{2n} - \Phi u^* } \right\| > f(qt)} \right) \\ &{}= P\left( {\left\| {\Psi u_{2n} - \Phi u^* } \right\| >f(qt)} \right) \\ &{}\ge P\left( {\left\| {\Theta u_{2n} - \Theta u^* } \right\| > f(t)} \right) \\ &{}\ge P\left( {\left\| {\xi _{2n+1} - \Theta u^* } \right\| > f(t)} \right) . \end{array} \end{aligned}$$Let \(n\rightarrow \infty \), we receive \( P\left( {\left\| {\xi - \Theta u^* } \right\| > f(t)} \right) = 0 \) for all \(t>0\) implying \( \Theta u^* =\xi \) a.s. Then, \(\Phi ,\Theta \) have a random coincidence point \(u^*\). By the same argument, \(\Psi ,\Theta \) have a random coincidence point \(v^*\).\(\square \)
Corollary 3.7
Let \(\Phi , \Theta : L_0^X(\Omega ) \rightarrow L_0^X(\Omega )\) be continuous in probability completely random operators, \(\Phi \) be surjective and \(f: [0,\infty )\rightarrow [0,\infty )\) be a continuous, increasing function such that \(f (0)=0,\lim _{t\rightarrow \infty } f(t)=\infty \) and \(q>1\). Assume that for any \(u,v\) in \(L_0^X(\Omega )\) and \(t>0\), we have
If there exist random variables \(u_0,v_0 \) in \( L_0^X(\Omega )\) and \(p >0\) such that \(\Phi v_0=\Theta u_0\) and
Then,
-
(1)
Assume that there exists a number \( c>1/q\) such that
$$\begin{aligned} \sum _{n=1}^\infty f( c^n)<\infty . \end{aligned}$$(3.25)Then, the condition (3.24) is sufficient for \(\Phi ,\Theta \) to have a random coincidence point.
-
(2)
Assume that for each \(t,s>0\)
$$\begin{aligned} f(t+s)\ge f(t)+f(s). \end{aligned}$$(3.26)Then, the condition (3.24) is also sufficient for \(\Phi ,\Theta \) to have a random coincidence point.
Proof
It is easy to receive the corollary when we take \(\Psi v=\Phi v\) in Theorem 3.6.\(\square \)
4 Applications to Random Fixed Point Theorems and Random Equations
In this section, we present some applications to random fixed point theorems and random equations.
Theorem 4.1
Let \(\Phi : L_0^X(\Omega ) \rightarrow L_0^X(\Omega )\) be surjective, continuous in probability completely random operator and \(f: [0,\infty )\rightarrow [0,\infty )\) be a continuous, increasing function such that \(f (0)=0,\lim _{t\rightarrow \infty } f(t)=\infty \) and \(q>1\). Assume that for each pair \(u,v \) in \(L_0^X(\Omega )\)
If there exist random variables \(v_0 \) in \( L_0^X(\Omega )\) and \(p >0\) such that
Then
-
(1)
Assume that there exists a number \( c>1/q\) such that
$$\begin{aligned} \sum _{n=1}^\infty f( c^n)<\infty . \end{aligned}$$(4.3)Then, the condition (4.2) is sufficient for \(\Phi \) to have a unique random fixed point.
-
(2)
Assume that for each \(t,s>0\)
$$\begin{aligned} f(t+s)\ge f(t)+f(s). \end{aligned}$$(4.4)Then, the condition (4.2) is also sufficient for \(\Phi \) to have a unique random fixed point.
Proof
Consider the completely random operator \(\Theta \) given by \(\Theta u=u\). By Corollary 3.7, \(\Phi \) and \(\Theta \) have a random coincidence point \(\xi \) which is exactly the random fixed point of \(\Phi \).
Let \(\xi , \eta \) be two random fixed points of \(\Phi \). Then, for each \(t>0\), we have
By induction, it follows that
Since \(\lim _{n\rightarrow \infty } f(q^nt)=+\infty \), we conclude that \( P\left( \Vert \xi -\eta \Vert >f(q^nt)\right) =0\) for each \(t>0\). Hence, \( g(\Vert \xi -\eta \Vert )=0\quad \text{ a.s. }\), with \(g\) is the inverse function of \(f\). So, we have \(\xi =\eta \quad \text{ a.s. }\) as claimed. \(\square \)
Theorem 4.2
Let \(\Phi ,\Theta :L_0^X(\Omega ) \rightarrow L_0^X(\Omega )\) be continuous in probability completely random operators, \(\Phi \) be surjective and \(f:[0,\infty )\rightarrow [0,\infty )\) be a mapping such that for each \(t>0\),
Assume that for each pair \(u,v\) in \( L_0^X(\Omega )\) and \(t>0\), we have
If \(\Phi , \Theta \) commute i.e. \(\Phi \Theta u=\Theta \Phi u\) for any random variable \(u \) in \( L_0^X(\Omega )\) then \(\Phi \) and \(\Psi \) have a unique common random fixed point if there exist random variables \(u_0,v_0 \) in \( L_0^X(\Omega )\) and \(p >0\) such that \(\Phi v_0=\Theta u_0\) and
Proof
Suppose that (4.7) holds. By Corollary 3.3, there exists \(u^*\) such that \(\Phi u^*=\Theta u^*=\xi \). For \(t>0\), we have
By induction, it follows that \( P(\Vert \Phi \xi -\xi \Vert >t)\le P(\Vert \Phi \xi -\xi \Vert >q^n t)\) for any \(n\in N\). Let \(n\rightarrow \infty \), we have \( P(\Vert \Phi \xi -\xi \Vert >t)=0\) for any \(t>0\). Thus, \(\Phi \xi =\xi \) i.e. \(\xi \) is a random fixed point of \(\Phi \). We have \(\Theta \xi =\Theta \Phi u^*=\Phi \Theta u^*=\Phi \xi =\xi \). So \(\xi \) is also a random fixed point of \(\Theta \).
Let \(\xi _1\) and \(\xi _2\) be two common random fixed points of \(\Phi \) and \(\Theta \). For each \(t>0\), we have
Let \(n\rightarrow \infty \), we have \( P(\Vert \xi _1-\xi _2\Vert >t)=0\) for all \(t>0\). Hence, \(\xi _1=\xi _2\).\(\square \)
Corollary 4.3
Let \(\Phi : L_0^X(\Omega ) \rightarrow L_0^X(\Omega )\) be a surjective, continuous in probability and probabilistic \( q\)-expansive completely random operator in the sense that there exists a number \(q>1\) such that
for all random variables \(u,v\) in \( L_0^X(\Omega )\) and \( t>0\). Then, \(\Phi \) has a unique random fixed point if there exists a random variable \(v_0 \) in \( L_0^X(\Omega )\) and \(p >0\) such that
Proof
Consider \(\Theta : L_0^X(\Omega ) \rightarrow L_0^X(\Omega )\) given by \(\Theta u=u\), the function \(f(t)=(1-q)t\) and \( h(t)= 1-q >0\). Then \(\Phi ,\Theta \) and \( f(t)\) satisfy the conditions stated in the Theorem 4.2 and \(\Phi , \Theta \) commute. Thus, \(\Phi \) and \(\Theta \) have a common random fixed point \(\xi \) i.e. \(\Phi \) has a random fixed point \(\xi \).
Theorem 4.4
Let \(\Phi ,\Theta : L_0^X(\Omega )\rightarrow L_0^X(\Omega )\) be continuous in probability completely random operators, \(\Phi \) be surjective and
for all \(u,v\) in \( L_0^X(\Omega )\), \( t>0\) and \(f: [0,\infty )\rightarrow [0,\infty )\) be a continuous, increasing function such that \(f (0)=0,\lim _{t\rightarrow \infty } f(t)=\infty \) satisfying either (4.3) or (4.4) and \(q>1\). Consider random equation of the form
where \(\lambda \) is a real number and \(\eta \) is a random variable in \(L_0^X(\Omega )\).
Assume that
where \(q'>1 \). Then the equation (4.11) has a unique random solution if there exists a random variable \(v_0\) in \(L_0^X(\Omega )\) and a number \(p>0\) such that
Proof
Suppose that the condition (4.13) holds. Define a completely random operator \(\Psi \) by
From (4.13) it follows that
Let \(g=f^{-1}\) be the inverse function of \(f\). Then, \(g: [0,\infty )\rightarrow [0,\infty )\) is continuous, increasing with \(g (0)=0,\lim _{t\rightarrow \infty } g(t)=\infty \). For each \(t>0\), there exists \(t'\) so that \(f(t')=|\lambda | f(t)\) i.e. \(t'=g(|\lambda | f(t))\). So, we have
From (4.12), we receive \(|\lambda | f\left( t \right) \le f\left( {\frac{q}{{q'}}t} \right) \). Then, we deduce \( g\left( {|\lambda | f\left( t \right) } \right) \le \frac{q}{{q'}}t \). So, \( t' \le \frac{q}{{q'}}t \) and \( \frac{{q't'}}{qt} \le 1\). Hence,
which implies
Consequently, \(\Theta \) and \( \Psi \) satisfy the conditions stated in the Corollary 3.7. Hence, \(\Theta \) and \(\Psi \) has a random coincidence point \(\xi \) i.e. the equation (4.11) has a random solution \(\xi \).\(\square \)
Corollary 4.5
Let \(\Phi : L_0^X(\Omega )\rightarrow L_0^X(\Omega )\) be a surjective, continuous in probability completely random operator satisfying the following condition
for all \(u,v\) in \( L_0^X(\Omega )\), \( t>0\) , where \(f: [0,\infty )\rightarrow [0,\infty )\) is a continuous, increasing function such that \(f (0)=0,\lim _{t\rightarrow \infty } f(t)=\infty \) satisfying either (4.3) or (4.4) and \(q>1\). Consider random equation of the form
where \(\lambda \) is a real number and \(\eta \) is a random variable in \(L_0^X(\Omega )\).
Assume that
where \(q'>1 \). Then the equation (4.16) has a unique random solution if and only if there exists a random variable \(v_0\) in \(L_0^X(\Omega )\) and a number \(p>0\) such that
Proof
Applying the Theorem 4.4 for the completely random operator \(\Theta \) given by \(\Theta u=u\).\(\square \)
Corollary 4.6
Let \(\Phi ,\Theta : L_0^X(\Omega )\rightarrow L_0^X(\Omega )\) be continuous in probability completely random operators, \(\Phi \) be surjective satisfying the following condition
for all \(u,v\) in \( L_0^X(\Omega )\) and a number \(q>1\). Consider the random equation
where \(\lambda \) is a real number and \(\eta \) is a random variable in \(L_p^X(\Omega )\), \(p>0\).
Assume that \(0<|\lambda |<q\). Then, the random equation (4.20) has a solution if there exists a random variable \(v_0\) in \( L_0^ X(\Omega )\) such that
Proof
Suppose that there exists a random variable \(u_0 \) in \( L_0^ X(\Omega )\) such that (4.10) holds. So, \(\Phi \) and \(\Theta \) satisfy (4.15) where \(f(t)=t\). Take \( |\lambda |<s<q\), then \(q'=q/s>1\) and
Moreover, for each \(t>0\)
since
where \(C_p\) is a constant. Hence, the condition (4.13) is satisfied. By Theorem 4.4, we conclude that the equation (4.20) has a random solution.\(\square \)
Taking the completely random operator \(\Theta \) given by \(\Theta u=u\), we obtain
Corollary 4.7
Let \(\Phi : L_0^X(\Omega )\rightarrow L_0^X(\Omega )\) be a surjective, continuous in probability completely random operator satisfying the following condition
for all \(u,v\) in \( L_0^X(\Omega )\) and a number \(q>1\). Consider the random equation
where \(\lambda \) is a real number satisfying \(0<|\lambda |<q\) and \(\eta \) is a random variable in \(L_p^X(\Omega )\), \(p>0\). Then, the random equation (4.24) has a unique random solution if there exists a random variable \(v_0\) in \( L_0^ X(\Omega )\) such that
References
Beg, I., Shahzad, N.: Random fixed point theorems for nonexpansive and contractive-type random operators on Banach spaces. J. Appl. Math. Stoc. Anal. 7(4), 569–580 (1994)
Bharucha-Reid, A.T.: Random Integral Equations. Academic Press, New York (1972)
Bharucha-Reid, A.T.: Fixed point theorems in probabilistic analysis. Bull. Amer. Math. Soc. 82(5), 641–657 (1976)
Chouhury, B.S., Metiya, N.: The point of coincidence and common fixed point for a pair mappings in cone metric spaces. Comput. Math. Appl. 60, 1686–1695 (2010)
Fierro, R., Martnez, C., Morales, C.H.: Random coincidence theorems and applications. J. Math. Anal. Appl. 378(1), 213–219 (2011)
Hadzic, O., Pap, E.: Fixed Point Theory in Probabilistic Metric Spaces. Kluwer Academic Publisher, Dordrecht (2001)
Himmelberg, C.J.: Measurable relations. Fund. Math. 87, 53–72 (1975)
Jung, J.S., Cho, Y.J., Kang, S.M., Lee, B.S., Thakur, B.S.: Random fixed point theorems for a certain class of mappings in banach spaces. Czechoslovak Math. J. 50(2), 379–396 (2000)
Kang, S.M.: Fixed points for expansion mappings. Math. Jpn. 38, 713–717 (1993)
Khan, M.A., Khan, M.S., Sessa, S.: Some theorems on expansion mappings and their fixed points. Demonstr. Math. 19, 673–683 (1986)
Pathak, H.K., Kang, S.M., Ryu, J.W.: Some fixed points of expansion mappings. Intern. J. Math. Math. Sci. 19(1), 97–102 (1996)
Rhoades, B.E.: A comparison of various definitions of contractive mappings. Trans. Am. Math. Soc. 226, 257–290 (1977)
Shrivastava, R., Rajput, A., Singhd, S.K.: A common unique random fixed point theorem for expansive type mapping in hilbert space. Intern. J. Math. Math. Sci. Eng. Appl. 6(1), 425–430 (2012)
Shahzad, N.: Random approximations and random coincidence points of multivalued random maps with stochastic domain. New Zealand J. Math. 29(1), 91–96 (2000)
Shahzad, N.: Some general random coincidence point theorems. New Zealand J. Math. 33(1), 95–103 (2004)
Shahzad, N.: Random fixed points of discontinuous random maps. Math. Comput. Model. 41, 1431–1436 (2005)
Shahzad, N.: On random coincidence point theorems. Topol. Methods Nonlinear Anal. 25(2), 391–400 (2005)
Shahzad, N., Hussain, N.: Deterministic and random coincidence point results for f-nonexpansive maps. J. Math. Anal. Appl. 323, 1038–1046 (2006)
Skorokhod, A.V.: Random Linear Operators. Reidel Publishing Company, Dordrecht (1984)
Taniguchi, T.: Common fixed point theorems on expansion type mappings on complete metric spaces. Math. Jpn. 34, 139–142 (1989)
Tan, K.K., Yuan, X.Z.: On deterministic and random fixed points. Proc. Am. Math. Soc. 119(3), 849–856 (1993)
Thang, D.H., Cuong, T.M.: Some procedures for extending random operators. Random Oper. Stoch. Equ. 17(4), 359–380 (2009)
Thang, D.H., Anh, T.N.: On random equations and applications to random fixed point theorems. Random Oper. Stoch. Equ. 18(3), 199–212 (2010)
Thang, D.H., Anh, P.T.: Random fixed points of completely random operators Random Oper. Stoch. Equ. 21(1), 1–20 (2013)
Tsokos, Chris P., Padgett, W.J.: Random Integral Equations with Applications to Stochastic Sytems. Lecture notes in mathematics, vol. 233. Springer, Berlin-New York (1971)
Wang, S.Z., Li, B.Y., Gao, Z.M., Iseki, K.: Some fixed point theorems on expansion mappings. Math. Jpn. 29, 631–636 (1984)
Acknowledgments
This work is supported by the Vietnam National Foundation for Science Technology Development (NAFOSTED).
Author information
Authors and Affiliations
Corresponding author
Additional information
Communicated by Lee See Keong.
Rights and permissions
About this article
Cite this article
Anh, P.T. Random Coincidence Points of Expansive Type Completely Random Operators. Bull. Malays. Math. Sci. Soc. 38, 1609–1625 (2015). https://doi.org/10.1007/s40840-014-0094-9
Received:
Revised:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s40840-014-0094-9