Abstract
In this paper, we propose an iterative algorithm and, by using the proposed algorithm, prove some strong convergence theorems for finding a common element of the set of solutions of a finite family of split equilibrium problems and the set of common fixed points of a countable family of nonexpansive mappings in Hilbert spaces. An example is given to illustrate the main result of this paper. As an application, we construct an algorithm to solve an optimization problem.
Similar content being viewed by others
1 Introduction
Throughout this paper, let \(\mathbb{R}\) denote the set of all real numbers, \(\mathbb{N}\) denote the set of all positive integer numbers, H be a real Hilbert space and C be a nonempty closed convex subset of H. A mapping \(S: C\to C\) is said to nonexpansive if
for all \(x,y\in C\). The set of fixed points of S is denoted by \(\operatorname {Fix}(S)\). It is known that the set \(\operatorname {Fix}(S)\) is closed and convex.
Let \(F: C\times C\to\mathbb{R}\) be a bifunction. The equilibrium problem for F is to find \(z\in C\) such that
for all \(y\in C\). The set of all solutions of the problem (1.1) is denoted by \(\operatorname {EP}(F)\), i.e.,
From the problem (1.1), we can consider some related problems, that is, variational inequality problems, complementarity problems, fixed point problems, game theory and other problems. Also, many problems in physics, optimization, and economics can be reduced to finding a solution of the problem (1.1) (see [1–4]).
In 1997, Combettes and Hirstoaga [5] introduced an iterative scheme of finding a solution of the problem (1.1) under the assumption that \(\operatorname {EP}(F)\) is nonempty. Later on, many iterative algorithms were considered to find a common element of the set of \(\operatorname {Fix}(S)\cap \operatorname {EP}(F)\) (see [6–11]).
Recently, some new problems called split variational inequality problems were considered by some authors. Especially, Censor et al. [12] initially studied this class of split variational inequality problems.
Let \(H_{1}\) and \(H_{2}\) be two real Hilbert spaces. Given the operators \(f: H_{1}\to H_{1}\) and \(g: H_{2}\to H_{2}\), bounded linear operator \(A:H_{1}\to H_{2}\), and nonempty closed convex subsets \(C\subset H_{1}\) and \(Q\subset H_{2}\), the split variational inequality problem is formulated as follows:
Find a point \(x^{*}\in C\) such that
for all \(x\in C\) and such that
for all \(y\in Q\).
After investigating the algorithm of Censor et al., Moudafi [13] introduced a new iterative scheme to solve the following split monotone variational inclusion:
Find \(x^{*}\in H_{1}\) such that
and such that
where \(B_{1}: H_{i}\to2^{H_{i}}\) is a set-valued mappings for \(i=1,2\).
In 2013, Kazmi and Rizvi [14] considered a new class of split equilibrium problems. Let \(F_{1}: C\times C\to\mathbb{R}\) and \(F_{2}: Q\times Q\to\mathbb{R}\) be two bifunctions and \(A: H_{1}\to H_{2}\) be a bounded linear operator. The split equilibrium problem is as follows:
Find \(x^{*}\in C\) such that
for all \(x\in C\) and such that
for all \(y\in Q\). The set of all solutions of the problems (1.2) and (1.3) is denoted by Ω, i.e.,
For more details as regards the split equilibrium problems, refer to [15, 16], in which the author gave an iterative algorithm to find a common element of the sets of solutions of the split equilibrium problem and hierarchical fixed point problem.
In this paper, inspired by the results in [14] and [16], we propose an iterative algorithm to find a common element of the set of solutions for a family of split equilibrium problems and the set of common fixed points of a countable family of nonexpansive mappings. In particular, we use some new methods to prove the main result of this paper. As an application, we propose an iterative algorithm to solve a split variational inequality problem.
2 Preliminaries
Let H be a Hilbert space and C be a nonempty closed subset of H. For each point \(x\in H\), there exists a unique nearest point of C, denoted by \(P_{C}x\), such that
for all \(y\in C\). Such a \(P_{C}\) is called the metric projection from H onto C. It is well known that \(P_{C}\) is a firmly nonexpansive mapping from H onto C, i.e.,
for all \(x,y\in H\). Further, for any \(x\in H\) and \(z\in C\), \(z=P_{C}x\) if and only if
for all \(y\in C\).
A mapping \(B:C\to H\) is called α-inverse strongly monotone if there exists \(\alpha>0\) such that
for all \(x,y\in H\). For each \(\lambda\in(0, 2\alpha]\), \(I-\lambda B\) is a nonexpansive mapping of C into H (see [17]).
Consider the following variational inequality for an inverse strongly monotone mapping B:
Find \(u\in C\) such that
for all \(v\in C\). The set of solutions of the variational inequality is denoted \(\operatorname {VI}(C,B)\). It is well known that
for any \(\lambda>0\). By this property, we can use a simple method to show that \(u\in \operatorname {VI}(C,B)\). In fact, let \(\{x_{n}\}\) be a sequence in C with \(x_{n}\rightharpoonup u\). If \(x_{n}-P_{C}(I-\lambda B)x_{n}\to0\), then, by the demiclosedness principle, it follows that \(u=P_{C}(I-\lambda B)\), i.e., \(u\in \operatorname {VI}(C,B)\). In Section 3, we use this method to show the conclusions of our main results in this paper.
Let \(S: C\to C \) be a mapping. It is well known that S is nonexpansive if and only if the complement \(I-S\) is \(\frac {1}{2}\)-inverse strongly monotone (see [18]). Assume that \(\operatorname {Fix}(S)\neq\emptyset\). Then we have
for all \(x\in C\) and \(\hat{x}\in \operatorname {Fix}(S)\), which is obtained directly from
Let F be a bifunction of \(C\times C\) into \(\mathbb{R}\) satisfying the following conditions:
-
(A1)
\(F(x, x)=0\) for all \(x\in C\);
-
(A2)
F is monotone, i.e., \(F(x,y)+F(y,x)\leq0\) for all \(x,y\in C\);
-
(A3)
for each \(x,y,z\in C\), \(\lim_{t\downarrow0}F(tz+(1-t)x,y)\leq F(x,y)\);
-
(A4)
for each \(x\in C\), \(y\mapsto F(x,y)\) is convex and lower semi-continuous.
Lemma 2.1
[19]
Let C be a nonempty closed convex subset of a Hilbert space H and \(F: C\times C\to\mathbb{R}\) be a bifunction which satisfies the conditions (A1)-(A4). For any \(x\in H\) and \(r >0\), define a mapping \(T_{r}:H\to C\) by
Then \(T_{r}^{F}\) is well defined and the following hold:
-
(1)
\(T_{r}^{F}\) is single-valued;
-
(2)
\(T_{r}^{F}\) is firmly nonexpansive, i.e., for any \(x,y\in H\),
$$\bigl\Vert T_{r}^{F}x-T_{r}^{F}y \bigr\Vert ^{2}\leq\bigl\langle T_{r}^{F}x-T_{r}^{F}y,x-y \bigr\rangle ; $$ -
(3)
\(\operatorname {Fix}(T_{r}^{F})=\operatorname {EP}(F)\);
-
(4)
\(\operatorname {EP}(F)\) is closed and convex.
Lemma 2.2
[20]
Let \(F: C\times C\to\mathbb{R}\) be a bifunction satisfying the conditions (A1)-A(4). Let \(T^{F}_{r}\) and \(T^{F}_{s}\) be defined as in Lemma 2.1 with \(r,s>0\). Then, for any \(x,y\in H\), one has
Remark 2.1
In [20], some other conditions are required besides the conditions (A1)-(A4). In fact, the conditions (A1)-(A4) are enough for Lemma 2.2. For the proof, refer to [9, 20].
Lemma 2.3
[9]
Let \(F: C\times C\to\mathbb{R}\) be a functions satisfying the conditions (A1)-(A4) and \(T^{F}_{s}\), \(T^{F}_{t}\) be defined as in Lemma 2.1 with \(s,t>0\). Then the following holds:
for all \(x\in H\).
Lemma 2.4
[21]
Let \(\{a_{n}\}\) be a sequence in \([0,1]\) such that \(\sum_{n=1}^{\infty}a_{n}=1\). Then we have the following:
for any bounded sequence \(\{x_{n}\}\) in a Hilbert space H.
Lemma 2.5
(Demiclosedness principle)
Let T be a nonexpansive mapping on a closed convex subset C of a real Hilbert space H. Then \(I-T\) is demiclosed at any point \(y\in H\), that is, if \(x_{n}\rightharpoonup x\) and \(x_{n}-Tx_{n}\to y\in H\), then \(x-Tx=y\).
Lemma 2.6
[22]
Assume that \(\{a_{n}\}\) is a sequence of nonnegative numbers such that
for each \(n\geq0\), where \(\{\gamma_{n}\}\) is a sequence in \((0,1)\) and \(\{\delta_{n}\}\) is a sequence in \(\mathbb{R}\) such that
-
(1)
\(\sum_{n=1}^{\infty}\gamma_{n}=\infty\);
-
(2)
\(\limsup_{n\to\infty}\delta_{n}/\gamma_{n}\leq0\) or \(\sum_{=1}^{\infty} \vert \delta_{n}\vert <\infty\).
Then \(\lim_{n\to\infty}a_{n}=0\).
Lemma 2.7
Let U and V be nonexpansive mappings. For \(\sigma\in(0,1)\), define \(S=\sigma U+(1-\sigma)V\). Suppose that \(\operatorname {Fix}(U)\cap \operatorname {Fix}(V)\neq \emptyset\). Then \(\operatorname {Fix}(U)\cap \operatorname {Fix}(V)=\operatorname {Fix}(S)\).
From [24] we can see that Lemma 2.7 holds whenever U and V are self or non-self mappings.
Lemma 2.8
[24]
Let C be a nonempty closed convex subset of a Hilbert space H and \(T: C\to H\) be a nonexpansive mapping with \(\operatorname {Fix}(T)\neq\emptyset\). Let \(P_{C}\) be the metric projection from H onto C. Then \(\operatorname {Fix}(P_{C}T)=\operatorname {Fix}(T)=\operatorname {Fix}(TP_{C})\).
Remark 2.2
Let \(S_{1},S_{2}: C\to H\) be two nonexpansive mappings with \(\operatorname {Fix}(S_{1})\cap \operatorname {Fix}(S_{2})\neq\emptyset\). Let \(\sigma\in(0,1)\) and define the mapping \(S: C\to H\) by \(S=\sigma S_{1}+(1-\sigma)S_{2}\). By Lemmas 2.7 and 2.8, it is easy to see that \(\operatorname {Fix}(P_{C}S)=\operatorname {Fix}(P_{C}S_{1})\cap \operatorname {Fix}(P_{C}S_{2})\).
From Remark 2.2, we get the following result.
Lemma 2.9
Let \(\{B_{i}\}_{i=1}^{N}\) be a finite family of inverse strongly monotone mappings from C to H with the constants \(\{\beta_{i}\}_{i=1}^{N}\) and assume that \(\bigcap_{i=1}^{N}\operatorname {VI}(C,B_{i})\neq\emptyset\). Let \(B=\sum_{i=1}^{N} \alpha_{i}B_{i}\) with \(\{\alpha_{i}\}_{i=1}^{N}\subset(0,1)\) and \(\sum_{i=1}^{N}\alpha_{i}=1\). Then \(B: C\to H\) is a β-inverse strongly monotone mapping with \(\beta=\min\{\beta_{1},\ldots, \beta_{N}\}\) and \(\operatorname {VI}(C,B)=\bigcap_{i=1}^{N} \operatorname {VI}(C,B_{i})\).
Proof
It is easy to show that B is a β-inverse strongly monotone mapping. In fact, for all \(x,y\in C\), by Lemma 2.4, we have
which implies that B is a β-inverse strongly monotone mapping.
Next, we prove that \(\operatorname {VI}(C,B)=\bigcap_{i=1}^{N} \operatorname {VI}(C,B_{i})\). Obviously, we have
Now, for any \(w\in \operatorname {VI}(C,B)\), we show that \(w\in\bigcap_{i=1}^{N} \operatorname {VI}(C,B_{i})\). Take a constant \(\lambda\in(0,2\beta]\). Then \(I-\lambda B\) is nonexpansive. Note that \(I-\lambda B=\sum_{i=1}^{N}\alpha_{i}(I-\lambda B_{i})\) and each \(I-\lambda B_{i}\) is nonexpansive. From Remark 2.2, it follows that
Thus we have
for each \(i=1,\ldots, N\). Therefore, \(w\in\bigcap_{i=1}^{N} \operatorname {VI}(C,B_{i})\). This completes the proof. □
3 Main result
Now, we give the main results of this paper.
Theorem 3.1
Let \(H_{1}\), \(H_{2}\) be two real Hilbert spaces and \(C\subset H_{1}\), \(Q\subset H_{2}\) be nonempty closed convex subsets. Let \(A_{i}: H_{1}\to H_{2}\) be a bounded linear operator for each \(i=1,\ldots ,N_{1}\) with \(N_{1}\in\mathbb{N}\) and \(B_{i}: C\to H_{1}\) be a \(\beta_{i} \)-inverse strongly monotone operator for each \(i=1,\ldots,N_{2}\) with \(N_{2}\in\mathbb{N}\). Assume that \(F:C\times C\to\mathbb{R}\) satisfies (A1)-(A4), \(F_{i}: Q\times Q\to\mathbb{R}\) (\(i=1,\ldots, N_{1}\)) satisfies (A1)-(A4). Let \(\{S_{n}\}\) be a countable family of nonexpansive mappings from C into C. Assume that \(\Theta=\Gamma \cap\Omega\cap \operatorname {VI}\neq\emptyset\), where \(\Gamma=\bigcap_{n=1}^{\infty} \operatorname {Fix}(S_{n})\), \(\Omega=\{z\in C: z\in \operatorname {EP}(F)\ \textit{and}\ A_{i}z\in \operatorname {EP}(F_{i}), i=1,\ldots ,N_{1} \}\) and \(\operatorname {VI}=\bigcap_{i=1}^{N_{2}}\operatorname {VI}(C,B_{i})\). Let \(\{\gamma_{1},\ldots, \gamma_{N_{2}}\}\subset (0,1)\) with \(\sum_{i=1}^{N_{2}}\gamma_{i}=1\). Take \(v,x_{1}\in C\) arbitrarily and define an iterative scheme in the following manner:
for each \(i=1,\ldots,N_{1}\) and \(n\in\mathbb{N}\), where \(\{r_{n}\}\subset (r,\infty)\) with \(r>0\), \(\{\lambda_{n}\}\subset(0,2\beta)\) with \(\beta =\min\{\beta_{1},\ldots,\beta_{N_{2}}\}\) and \(\gamma\subset(0,1/L^{2}]\), \(L=\max\{L_{1},\ldots,L_{N_{1}}\}\) and \(L_{i}\) is the spectral radius of the operator \(A_{i}^{*}A_{i}\) and \(A_{i}^{*}\) is the adjoint of \(A_{i}\) for each \(i\in\{ 1,\ldots, N_{1}\}\), and \(\{\alpha_{n}\}\subset(0,1)\) is a strictly decreasing sequence. Let \(\alpha_{0}=1\) and assume that the control sequences \(\{\alpha_{n}\}\), \(\{\lambda_{n}\}\), \(\{r_{n}\}\) satisfy the following conditions:
-
(1)
\(\lim_{n\to\infty}\alpha_{n}=0\) and \(\sum_{n=1}^{\infty}\alpha _{n}=\infty\);
-
(2)
\(\sum_{n=1}^{\infty} \vert r_{n+1}-r_{n}\vert <\infty\) and \(\sum_{n=1}^{\infty} \vert \lambda_{n+1}-\lambda_{n}\vert <\infty\);
-
(3)
\(\lim_{n\to\infty} \lambda_{n}=\lambda>0\).
Then the sequence \(\{x_{n}\}\) defined by (3.1) converges strongly to a point \(z=P_{\Theta}v\).
Proof
We first show that, for each \(i=1,\ldots, N_{1}\) and \(n\in \mathbb{N}\), \(A_{i}^{*}(I-T_{r_{n}}^{F_{i}})A_{i}\) is a \(\frac {1}{2L_{i}^{2}}\)-inverse strongly monotone mapping. In fact, since \(T_{r_{n}}^{F_{i}}\) is (firmly) nonexpansive and \(I-T_{r_{n}}^{F_{i}}\) is \(\frac {1}{2}\)-inverse strongly monotone, we have
for all \(x,y\in H_{1}\), which implies that \(A_{i}^{*}(I-T_{r_{n}}^{F_{i}})A_{i}\) is a \(\frac {1}{2L_{i}^{2}}\)-inverse strongly monotone mapping. Note that \(\gamma\in (0,\frac{1}{L_{i}^{2}}]\). Thus \(I-\gamma A_{i}^{*}(I-T_{r_{n}}^{F_{i}})A_{i}\) is nonexpansive for each \(i=1,\ldots, N_{1}\) and \(n\in\mathbb{N}\).
Now, we complete the proof by the next steps.
Step 1. \(\{x_{n}\}\) is bounded.
Let \(p\in\Theta\). Then \(p=T_{r_{n}}^{F_{i}}p\) and \((I-\gamma A_{i}^{*}(I-T_{r_{n}}^{F_{i}})A_{i})p=p\). Thus we have
Let \(B=\sum_{i=1}^{N_{2}}\gamma_{i} B_{i}\). Then B is a β-inverse strongly monotone mapping. Since \(\{\lambda_{n}\}\subset(0,2\beta)\), \(I-\lambda_{n} B \) is nonexpansive. Thus from (3.2), we have
Thus from (3.3), it follows that
for all \(n\in\mathbb{N}\), which implies that \(\{x_{n}\}\) is bounded and so are \(\{u_{i,n}\}\) (\(i=1,\ldots,N_{1}\)) and \(\{y_{n}\}\).
Step 2. \(\lim_{n\to\infty} \Vert x_{n+1}-x_{n}\Vert =0\) and \(\lim_{n\to\infty }\Vert u_{i,n+1}-u_{i,n}\Vert =0\) for each \(i=1,\ldots, N_{1}\).
Since the mappings \(I-\gamma A^{*}(I-T_{r_{n}}^{F_{i}})A\) are nonexpansive, by Lemmas 2.2 and 2.3, we have
where
and
Note that
where \(w_{n}=\frac{1}{N_{1}}\sum_{i=1}^{N_{1}}u_{i,n}\). Let \(M_{1}= \sup_{n\in\mathbb{N}}\Vert Bw_{n}\Vert \). By (3.1), (3.4), and (3.5), we have
Since \(\{\alpha_{n}\}\) is strictly decreasing, by using (3.6), we have
where \(M_{2}=\sup\{\Vert S_{n}y_{n}\Vert +\Vert v\Vert :n\in\mathbb{N}\}\). By (i) and (ii) and Lemma 2.6, we conclude that
Further, by (3.4) and (3.6), we have
Step 3. \(\lim_{n\to\infty} \Vert S_{i}x_{n}-x_{n}\Vert \to0\) for each \(i\in \mathbb{N}\).
First, we show that \(\lim_{n\to\infty} \Vert u_{i,n}-x_{n}\Vert =0\) for each \(i\in \{1,\ldots,N_{1}\}\). Since each \(A_{i}^{*}(I-T_{r_{n}}^{F_{i}})A_{i}\) is \(\frac{1}{ 2L_{i}^{2}}\)-inverse strongly monotone, by (3.1), we have
From Lemma 2.4 and (3.9), it follows that
Since \(\gamma<\frac{1}{L^{2}}=\max\{\frac{1}{L_{1}^{2}},\ldots, \frac {1}{L_{N_{1}}^{2}}\}\), we have
Since \(\alpha_{n}\to0\), by (3.7), we have
for each \(i\in\{1,\ldots, N_{1}\}\), which implies that
for each \(i\in\{1,\ldots, N_{1}\}\). Since \(T_{r_{n}}^{F }\) is firmly nonexpansive and \(I-\gamma A_{i} ^{*}(I-T_{r_{n}}^{F_{i}})A_{i} \) is nonexpansive, by (3.1), we have
which implies that
Now, from (3.1) and (3.12), it follows that
and so
Since \(\alpha_{n}\to0\), both \(\{u_{i,n}\}\) and \(\{x_{n}\}\) are bounded, by (3.7) and (3.10), we have
for each \(i\in\{1,\ldots,N_{1}\}\).
Next, we show that \(\lim_{n\to\infty} \Vert y_{n}-u_{n}\Vert =0\), where \(u_{n}=\frac {1}{N_{1}}\sum_{i=1}^{N_{1}}u_{i,n}\). Note that \(p={P_{C} (I-\lambda_{n}B)p}\). By (3.1), we have
and so
Since \(\alpha_{n}\to0\) and \(0<\lim_{n\to\infty} \lambda_{n}=\lambda<2\beta \), by (3.7), we have
Since \(P_{C}\) is firmly nonexpansive and \((I-\lambda_{n} B)\) is nonexpansive, by (3.1), we have
and so
From (3.1) and (3.15), we have
Therefore, we have
Since \(\lim_{n\to\infty}\alpha_{n}=0\) and both \(\{y_{n}\}\) and \(\{u_{n}\}\) are bounded, by (3.7) and (3.14), we have
Further, from (3.7), (3.13), (3.16), and
it follows that
Now, from (3.1), it follows that
Since \(\{\alpha_{n}\}\) is strictly decreasing, for each \(i\in\mathbb{N}\), by (2.1) and (3.18), we have
Since \(\lim_{n\to\infty}\alpha_{n}=0\) and \(\{y_{n}\}\) is bounded, by (3.17), one has
for all \(i\in\mathbb{N}\). Further, since
by (3.7), (3.17), and (3.19), we obtain
for all \(i\in\mathbb{N}\).
Step 4. \(\limsup_{n\to\infty}\langle v-z,x_{n}-z\rangle\leq0\).
Let \(z=P_{\Theta}v\). Since \(\{x_{n}\}\) is bounded, we can choose a subsequence \(\{x_{n_{j}}\}\) of \(\{x_{n}\}\) such that
Since \(\{x_{n_{j}}\}\) is bounded, there exists a subsequence \(\{ x_{n_{j_{i}}}\}\) of \(\{x_{n_{j}}\}\) converging weakly to a point \(w\in C\). Without loss of generality, we can assume that \(x_{n_{j}}\rightharpoonup w\).
Now, we show that \(w\in\Theta\). First of all, we prove that \(w\in \Gamma=\bigcap_{i=1}^{\infty} \operatorname {Fix}(S_{i})\). In fact, since \(x_{n}-S_{i}x_{n}\to0\) for each \(i\in\mathbb{N}\) and \(x_{n_{j}}\rightharpoonup w\), by Lemma 2.5, we obtain \(w\in\bigcap_{i=1}^{\infty} \operatorname {Fix}(S_{i})=\Gamma\).
Next, we show that \(w\in\Omega\), i.e., \(w\in \operatorname {EP}(F )\) and \(A_{i}w\in \operatorname {EP}(F_{i})\) for each \(i=1,\ldots, N_{1}\).
Let \(w_{i,n}=(I-A_{i}^{*}(I-T_{r_{n}}^{F_{i}}))A_{i}x_{n}\) for each \(i=1,\ldots, N_{1}\). By (3.10) and (3.13) we see that \(w_{i,n}-x_{n}\to0\) and \(T_{r_{n}}^{F}w_{i,n}-w_{i,n}\to0\) as \(n\to\infty\). By Lemma 2.2 we see that \(\Vert T_{r_{n}}^{F}w_{i,n}-T_{r}^{F}w_{i,n}\Vert \leq \vert 1-\frac{r}{r_{n}}\vert \Vert T_{r_{n}}^{F}w_{i,n}-w_{i,n}\Vert \to0\) as \(n\to\infty\). Hence \(T_{r}^{F}w_{i,n}-w_{i,n}\to0\) as \(n\to\infty\) for each \(i=1,\ldots,N_{1}\). Since \(T_{r}^{F}\) is non-expansive and \(\{w_{i,n}\}\) converges weakly to w, by Lemma 2.5 we get \(w=T_{r}^{F} w\), i.e., \(w\in \operatorname {EP}(F)\). On the other hand, since \((I-\gamma A^{*}_{i}(I-T_{r_{n}}^{F_{i}})A_{i} )x_{n}-x_{n}\to0 \) (by (3.13)) and \(I-\gamma A^{*}_{i}(I-T_{r_{n}}^{F_{i}})A_{i} \) is non-expansive, from Lemmas 2.2 and 2.5 it follows that \(w=(I-\gamma A^{*}_{i}(I-T_{r}^{F_{i}})A_{i} )w\), i.e., \(w=T_{r}^{F} A_{i}w\). Therefore, \(w\in\Omega\).
Finally, we prove that \(w\in \operatorname {VI}=\bigcap_{i=1}^{N_{2}} \operatorname {VI}(C,B_{i})\) by demiclosedness principle. Obviously, we only need to show that \(w=P_{C}(w-\lambda B_{i}w)\), where \(\lambda=\lim_{n\to\infty}\lambda_{n}\). By (3.1) and (3.16), one has \(\Vert u_{n}-P_{C}(I-\lambda_{n} B)u_{n}\Vert \to0\), where \(u_{n}=\frac{1}{N_{1}}\sum_{i=1}^{N_{1}}u_{i,n}\). Then we have
Since \(\lambda_{n}\to\lambda>0\), \(\{Bu_{n}\}\) is bounded and \(\Vert u_{n}-P_{C}(I-\lambda B)u_{n}\Vert \to0\), we have
On the other hand, since \(\{\lambda_{n}\}\subset(0,2\beta)\), one has \(\lambda\in(0,2\beta]\). Thus \(I-\lambda B\) is nonexpansive and, further, \(P_{C}(I-\lambda B)\) is nonexpansive. Noting that \(u_{n_{j}}\rightharpoonup w\) as \(j\to\infty\), by Lemma 2.5, we obtain \(w=P_{C}(I-\lambda B)w\). By Lemma 2.8, we get \(w\in \operatorname {VI}=\bigcap_{i=1}^{N_{2}}\operatorname {VI}(C,B_{i})\). Therefore, \(w\in\Theta\). By the property on \(P_{C}\), we have
Step 5. \(x_{n}\to z=P_{\Theta}v \) as \(n\to\infty\).
By (3.1), we have
which implies that
By Lemma 2.6 and (3.21), we can conclude that \(\lim_{n\to\infty} \Vert x_{n}-z\Vert =0\). This completes the proof. □
The following results follow directly from Theorem 3.1.
Corollary 3.2
Let \(H_{1}\), \(H_{2}\) be two real Hilbert spaces and \(C\subset H_{1}\), \(Q\subset H_{2}\) be nonempty closed convex subsets. Let \(A: H_{1}\to H_{2}\) be a bounded linear operator and \(B: C\to H_{1}\) be a β-inverse strongly monotone operator. Assume that \(F :C\times C\to\mathbb{R}\), \(F_{1}: Q\times Q\to\mathbb{R} \) are bifunctions satisfying the conditions (A1)-(A4). Let \(\{S_{n}\}\) be countable family of nonexpansive mappings from C into C. Assume that \(\Theta =\Gamma\cap\Omega\cap \operatorname {VI}(C,B)\neq\emptyset\), where \(\Gamma=\bigcap_{n=1}^{\infty} \operatorname {Fix}(S_{n})\) and \(\Omega=\{z\in C: z\in \operatorname {EP}(F)\ \textit{and}\ Az\in \operatorname {EP}(F_{1}) \}\). Take \(v\in C\) arbitrarily and define an iterative scheme in the following manner:
for all \(n\in\mathbb{N}\), where \(\{r_{n}\}\subset(r,\infty)\) with \(r>0\), \(\{\lambda_{n}\}\subset(0,2\beta)\), and \(\gamma\subset(0,1/L^{2}]\), L is the spectral radius of the operator \(A^{*}A\) and \(A^{*}\) is the adjoint of A, \(\alpha_{0}=1\), and \(\{\alpha_{n}\}\subset(0,1)\) is a strictly decreasing sequence. Assume that the control sequences \(\{\alpha_{n}\}\), \(\{\lambda_{n}\}\), and \(\{r_{n}\}\) satisfy the following conditions:
-
(1)
\(\lim_{n\to\infty}\alpha_{n}=0\) and \(\sum_{n=1}^{\infty}\alpha _{n}=\infty\);
-
(2)
\(\sum_{n=1}^{\infty} \vert r_{n+1}-r_{n}\vert <\infty\) and \(\sum_{n=1}^{\infty} \vert \lambda_{n+1}-\lambda_{n}\vert <\infty\);
-
(3)
\(\lim_{n\to\infty} \lambda_{n}=\lambda\in(0,2\beta)\).
Then the sequence \(\{x_{n}\}\) defined by (3.22) converges strongly to a point \(z=P_{\Theta}v\).
Corollary 3.3
Let \(H_{1}\), \(H_{2}\) be two real Hilbert spaces and \(C\subset H_{1}\), \(Q\subset H_{2}\) be nonempty closed convex subsets. Let \(A: H_{1}\to H_{2}\) be a bounded linear operator and \(B: C\to H_{1}\) be a β-inverse strongly monotone operator. Assume that \(F :C\times C\to\mathbb{R}\), \(F_{1}: Q\times Q\to\mathbb{R} \) are the bifunctions satisfying the conditions (A1)-(A4). Let \(S: C\to C\) be a nonexpansive mapping. Assume that \(\Theta=\operatorname {Fix}(S) \cap\Omega\cap \operatorname {VI}(C,B)\neq\emptyset\), where \(\Omega=\{z\in C: z\in \operatorname {EP}(F)\ \textit{and}\ Az\in \operatorname {EP}(F_{1}) \}\). Take \(v\in C\) arbitrarily and define an iterative scheme in the following manner:
for all \(n\in\mathbb{N}\), where \(\{r_{n}\}\subset(r,\infty)\) with \(r>0\), \(\{\lambda_{n}\}\subset(0,2\beta)\), and \(\gamma\subset(0,1/L^{2}]\), L is the spectral radius of the operator \(A^{*}A\) and \(A^{*}\) is the adjoint of A, \(\{\alpha_{n}\}\subset(0,1)\) is a sequence. Assume that the control sequences \(\{\alpha_{n}\}\), \(\{\lambda_{n}\}\), and \(\{r_{n}\}\) satisfy the following conditions:
-
(1)
\(\lim_{n\to\infty}\alpha_{n}=0\) and \(\sum_{n=1}^{\infty}\alpha _{n}=\infty\);
-
(2)
\(\sum_{n=1}^{\infty} \vert r_{n+1}-r_{n}\vert <\infty\) and \(\sum_{n=1}^{\infty} \vert \lambda_{n+1}-\lambda_{n}\vert <\infty\);
-
(3)
\(\lim_{n\to\infty} \lambda_{n}=\lambda\in(0,2\beta)\).
Then the sequence \(\{x_{n}\}\) defined by (3.23) converges strongly to a point \(z=P_{\Theta}v\).
Remark 3.4
Theorem 3.1 and Corollary 3.3 extend the corresponding one of Kazmi and Rizvi [14] from a nonexpansive mapping to a finite of family of nonexpansive mappings and from a split equilibrium problem to a finite of family of split equilibrium problems. It is a little simple to prove that \(w\in \operatorname {VI}\) by the demiclosedness principle in Theorem 3.1.
We give an example to illustrate Theorem 3.1 as follows.
Example 3.5
Let \(H_{1}=\mathbb{R}\) and \(H_{2}=\mathbb{R}^{2}\), \(C=[0,1]\), and \(Q=[0,1]\times[0,1]\). Let \(A_{1}: H_{1}\to H_{2}\) and \(A_{2}: H_{1}\to H_{2}\) defined by \(A_{1}x=(x,x)^{T}\) and \(A_{2}x=(\frac{x}{2},\frac {x}{2})^{T}\) for each \(x\in H_{1}\). Then \(A_{1}^{*}y=y_{1}+y_{2}\) and \(A_{2}^{*}y=\frac{y_{1}+y_{2}}{2}\) for each \(y=(y_{1},y_{2})^{T}\in H_{2}\). Then \(L_{1}=2\) and \(L_{2}=\frac{1}{2}\), where \(L_{1}\) and \(L_{2}\) are the spectral radius of \(A_{1}^{*}A_{1}\) and \(A_{2}^{*}A_{2}\), respectively.
Let \(B_{1}= 2(x-1)\) and \(B_{2}= -4\) for all \(x\in C\). Then it is easy to see that \(B_{1}\) and \(B_{2}\) are \(\frac{1}{2}\) and 1-inverse strongly monotone operators from C into \(H_{1}\). Find that \(\operatorname {VI}=\operatorname {VI}(C,B_{1})\cap \operatorname {VI}(C,B_{2})=\{1\}\). For each \(n\in\mathbb{N}\), let \(S_{n}:C\to C\) defined by \(S_{n}(x)=x+\frac{1}{3n}\) for each \(x\in[0,\frac{1}{2}]\) and \(S_{n}(x)=x\) for each \(x\in(\frac{1}{2},1]\). Then \(\{S_{n}\}\) is a countable family of nonexpansive mappings from C into C and it is easy to see that \(\Gamma=\bigcap_{n=1}^{\infty} \operatorname {Fix}(S_{n})=(\frac{1}{2},1]\). For each \(x,y\in C\), define the bifunction \(F: C\times C\to\mathbb{R}\) by \(F(x,y)=x-y\) for all \(x,y\in C\). For each \(u=(u_{1},u_{2})^{T}\) and \(v=(v_{1},v_{2})^{T}\in Q\), define \(F_{1}: Q\times Q\to\mathbb{R}\) and \(F_{2}:Q\times Q\to\mathbb{R}\) by
and
It is easy to check that the bifunctions F, \(F_{1}\), and \(F_{2}\) satisfy the conditions (A1)-(A4) and \(F_{1}\). Moreover, \(\Omega=\{1\}\), where \(\Omega=\{z\in C: z\in \operatorname {EP}(F), A_{1}z\in \operatorname {EP}(F_{1}) \mbox{ and } A_{2}z\in \operatorname {EP}(F_{2}) \}\). Therefore, \(\Theta=\Gamma\cap \mathrm{VI}\cap \Omega=\{1\}\).
Let \(\alpha_{0}=1\), \(\gamma_{1}=\gamma_{2}=\frac{1}{2}\), and \(\gamma=\frac {1}{4}\). For each \(n\in\mathbb{N}\), let \(r_{n}= 2\), \(\lambda_{n}=\frac {1}{4}\), \(\alpha_{n}=\frac{1}{n}\). Then the sequences \(\{\alpha_{n}\}\), \(\{ \lambda_{n}\}\), \(\{r_{n}\}\) satisfy the conditions (1)-(3) in Theorem 3.1.
For each \(x\in C\) and each \(n\in\mathbb{N}\), we compute \(T_{r_{n}}^{F_{1}}A_{1}x\), i.e., \(T_{r_{n}}^{F_{1}}(x,x)\). Find \(z=(1,1) \) such that
for all \(y=(y_{1},y_{2})\in Q\). Thus, from Lemma 2.1(1), it follows that \(T_{r_{n}}^{F_{1}}A_{1}x=(1,1)\) for each \(x\in C\). Similarly, for each \(x\in [0,1]\), we can find \(z=(1,1)\) such that, for \(y=(\frac{1}{2}, \frac{1}{2})\),
for \(y=(1, 1)\),
for \(y\in Q\setminus\{(1,1),(\frac{1}{2},\frac{1}{2})\}\),
Thus \(z=(1,1)=T_{r_{n}}^{F_{2}}A_{2}x\) for all \(x\in C\) by Lemma 2.1(1).
Now, take \(v=\frac{1}{2}\) and \(x_{1}=\frac{1}{4}\) and define the sequence \(\{x_{n}\}\) defined by (3.1). Since each \(x_{n}\in C\), from the statement above we get \(T_{r_{n}}^{F_{i}}A_{i}x_{n}=(1,1)\) for each \(i=1,2\). Furthermore, we can get
Note that
for all \(y\in C\). Thus \(u_{1,n}=1\) by Lemma 2.1(1) for each \(n\in \mathbb{N}\). Similarly, we can conclude that \(u_{2,n}=1\) for each \(n\in \mathbb{N}\).
Next, we compute the sequence \(\{y_{n}\}\). By the definition of \(\{y_{n}\} \), we see that
for all \(n\in\mathbb{N}\).
Finally, we compute the sequence \(\{x_{n}\}\) by the following iteration:
as \(n\to\infty\) as shown by Theorem 3.1.
4 Applications
In this section, let \(H_{1}\), \(H_{2}\) be two real Hilbert spaces and C, Q be two nonempty closed convex subsets of \(H_{1}\) and \(H_{2}\), respectively. Let \(f: C\to \mathbb{R}\), \(g: Q\to \mathbb{R}\) be two operators and \(A: H_{1}\to H_{2}\) be a bounded linear operator.
We consider the following optimization problem:
We denote the set of solutions of (4.1) by Θ and assume that \(\Theta\neq\emptyset\). Let \(F(x,y)=f(y)-f(x)\) for all \(x,y\in C\) and \(F_{1}(x,y)=g(y)-g(x)\) for all \(x,y\in Q\). Then \(F(x,y)\) and \(G(x,y)\) satisfy the conditions (A1)-(A4) in Section 2 provided that f is convex and lower semicontinuous on C and g is convex and lower semicontinuous on Q. Let \(\Omega=\{z\in C: z\in \operatorname {EP}(F)\mbox{ and }Az\in \operatorname {EP}(F_{1}) \}\). Obviously, \(\Theta=\Omega\).
By Corollary 3.3 with \(B=I\) and \(S=I\), we have the following iterative algorithm, which strongly converges to a point \(z=P_{\Theta}v\), which solves the optimization problem (4.1):
where \(\{r_{n}\}\subset(r,\infty)\) with \(r>0\), \(\{\lambda_{n}\}\subset (0,2)\), and \(\gamma\subset(0,1/L^{2}]\), L is the spectral radius of the operator \(A^{*}A\) and \(A^{*}\) is the adjoint of A, \(\{\alpha_{n}\} \subset(0,1)\) is a sequence. Assume that the control sequences \(\{\alpha _{n}\}\), \(\{\lambda_{n}\}\), and \(\{r_{n}\}\) satisfy the following conditions:
-
(1)
\(\lim_{n\to\infty}\alpha_{n}=0\) and \(\sum_{n=1}^{\infty}\alpha _{n}=\infty\);
-
(2)
\(\sum_{n=1}^{\infty} \vert r_{n+1}-r_{n}\vert <\infty\), \(\sum_{n=1}^{\infty} \vert \alpha_{n+1}-\alpha_{n}\vert <\infty\), and \(\sum_{n=1}^{\infty} \vert \lambda _{n+1}-\lambda_{n}\vert <\infty\);
-
(3)
\(\lim_{n\to\infty} \lambda_{n}=\lambda\in(0,2)\).
For the special case with \(H_{1}=H_{2}\) and \(C=Q\), we consider the following multi-objective optimization problem:
We denote the set of solution of (4.3) by Γ and assume that \(\Gamma\neq\emptyset\). In (4.2), setting \(A=I\) we get the following algorithm, which strongly converges to the solution of multi-objective optimization problem (4.3):
where \(\gamma\subset(0,1/L^{2}]\), L is the spectral radius of the operator \(I^{*}I\) and \(I^{*}\) is the adjoint of I, other parameters such as \(\{\alpha_{n}\}\), \(\{\lambda_{n}\}\), and \(\{r_{n}\}\) satisfy the same conditions (1)-(3).
5 Conclusion
In this paper, we construct an iterative algorithm to find a common element of the set of solutions of a finite family of split equilibrium problems and the set of common fixed points of a countable family of nonexpansive mappings in Hilbert spaces. In the proof methods, we use the inverse strong monotonicity of each \(A^{*}(I-T_{r_{n}})A\), which is such that the proof is simple and is different from the ones given in [14–16]. Also, in the results of this paper, we do not assume that each \(F_{i}\) is upper semi-continuous in the first argument for each \(i=1,\ldots,N_{1}\), which is required in the result in [14–16]. As an application, we solve an optimization problem by the result of this paper.
References
Chang, SS, Lee, HWJ, Chan, CK: A new method for solving equilibrium problem fixed point problem and variational inequality problem with application to optimization. Nonlinear Anal. 70, 3307-3319 (2009)
Katchang, P, Kumam, P: A new iterative algorithm of solution for equilibrium problems, variational inequalities and fixed point problems in a Hilbert space. J. Appl. Math. Comput. 32, 19-38 (2010)
Plubtieng, S, Punpaeng, R: A general iterative method for equilibrium problems and fixed point problems in Hilbert spaces. J. Math. Anal. Appl. 336, 455-469 (2007)
Qin, X, Shang, M, Su, Y: A general iterative method for equilibrium problems and fixed point problems in Hilbert spaces. Nonlinear Anal. 69, 3897-3909 (2008)
Combettes, PL, Hirstoaga, SA: Equilibrium programming using proximal like algorithms. Math. Program. 78, 29-41 (1997)
Agarwal, RP, Chen, JW, Cho, YJ: Strong convergence theorems for equilibrium problems and weak Bregman relatively nonexpansive mappings in Banach spaces. J. Inequal. Appl. 2013, 119 (2013)
Tada, A, Takahashi, W: Weak and strong convergence theorems for a nonexpansive mapping and an equilibrium problem. J. Optim. Theory Appl. 133, 359-370 (2007)
Takahashi, S, Takahashi, W: Viscosity approximation methods for equilibrium problems and fixed point problems in Hilbert spaces. J. Math. Anal. Appl. 331, 506-515 (2007)
Takahashi, S, Takahashi, W: Strong convergence theorem for a generalized equilibrium problem and a nonexpansive mapping in a Hilbert space. Nonlinear Anal. 69, 1025-1033 (2008)
Bnouhachem, A, Chen, Y: An iterative method for a common solution of generalized mixed equilibrium problems, variational inequalities, and hierarchical fixed point problems. Fixed Point Theory Appl. 2014, 155 (2014)
Bnouhachem, A: An interactive method for system of generalized equilibrium problem and fixed point problem. Fixed Point Theory Appl. 2014, 235 (2014)
Censor, Y, Gibali, A, Reich, S: Algorithms for the split variational inequality problem. Numer. Algorithms 69, 301-323 (2012)
Moudafi, A: Split monotone variational inclusions. J. Optim. Theory Appl. 150, 275-283 (2011)
Kazmi, KR, Rizvi, SH: Iterative approximation of a common solution of a split equilibrium problem, a variational inequality problem and a fixed point problem. J. Egypt. Math. Soc. 21, 44-51 (2013)
Bnouhachem, A: Algorithms of common solutions for a variational inequality, a split equilibrium problem and a hierarchical fixed point problem. Fixed Point Theory Appl. 2013, 278 (2013)
Bnouhachem, A: Strong convergence algorithm for split equilibrium problems and hierarchical fixed point problems. Sci. World J. 2014, 390956 (2014)
Iiduka, H, Takahashi, W: Strong convergence theorems for nonexpansive mappings and inverse-strongly monotone mappings. Nonlinear Anal. 61, 341-350 (2005)
Takahashi, W: Nonlinear Functional Analysis. Yokohama Publishers, Yokohama (2000)
Combettes, PL, Hirstoaga, SA: Equilibrium programming in Hilbert spaces. J. Nonlinear Convex Anal. 6, 117-136 (2005)
Cianciaruso, F, Marino, G, Muglia, L, Yao, Y: A hybrid projection algorithm for finding solutions of mixed equilibrium problem and variational inequality problem. Fixed Point Theory Appl. 2010, 383740 (2010)
Yang, L, Zhao, F, Kim, JK: Hybrid projection method for generalized mixed equilibrium problem and fixed point problem of infinite family of asymptotically quasi-ψ-nonexpansive mappings in Banach spaces. Appl. Math. Comput. 218, 6072-6082 (2012)
Xu, HK: An iterative approach to quadratic optimization. J. Optim. Theory Appl. 116, 659-678 (2003)
Bruch, RE Jr.: Properties of fixed point sets of nonexpansive mappings in Banach spaces. Trans. Am. Math. Soc. 179, 251-289 (1973)
Zhou, HY, Wang, PY, Zhou, Y: Minimum-norm fixed point of nonexpansive mappings with applications. Optimization 64, 799-814 (2015)
Acknowledgements
This work is supported by the Natural Science Funds of Hebei (Grant Number: A2015502021), the Fundamental Research Funds for the Central Universities (Grant Number: 2014ZD44) and the Project-sponsored by SRF for ROCS, SEM. The fourth author thanks the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT and future Planning (Grant Number: 2014R1A2A2A01002100).
Author information
Authors and Affiliations
Corresponding author
Additional information
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
All authors read and approved the final manuscript.
Rights and permissions
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
About this article
Cite this article
Wang, S., Gong, X., Abdou, A.A. et al. Iterative algorithm for a family of split equilibrium problems and fixed point problems in Hilbert spaces with applications. Fixed Point Theory Appl 2016, 4 (2016). https://doi.org/10.1186/s13663-015-0475-2
Received:
Accepted:
Published:
DOI: https://doi.org/10.1186/s13663-015-0475-2