Volume 2012, Issue 1 438023
Research Article
Open Access

Strong Convergence of the Viscosity Approximation Process for the Split Common Fixed-Point Problem of Quasi-Nonexpansive Mappings

Jing Zhao

Corresponding Author

Jing Zhao

College of Science, Civil Aviation University of China, Tianjin 300300, China cauc.edu.cn

Search for more papers by this author
Songnian He

Songnian He

College of Science, Civil Aviation University of China, Tianjin 300300, China cauc.edu.cn

Search for more papers by this author
First published: 25 March 2012
Citations: 9
Academic Editor: Yonghong Yao

Abstract

Very recently, Moudafi (2011) introduced an algorithm with weak convergence for the split common fixed-point problem. In this paper, we will continue to consider the split common fixed-point problem. We discuss the strong convergence of the viscosity approximation method for solving the split common fixed-point problem for the class of quasi-nonexpansive mappings in Hilbert spaces. Our results improve and extend the corresponding results announced by many others.

1. Introduction and Preliminary

Throughout this paper, we always assume that H is a real Hilbert space with inner product 〈·, ·〉 and norm ∥·∥. Let I denote the identity operator on H. Let C and Q be nonempty closed convex subset of real Hilbert spaces H1 and H2, respectively. The split feasibility problem (SFP) is to find a point
()
where A : H1H2 is a bounded linear operator. The SFP in finite-dimensional Hilbert spaces was first introduced by Censor and Elfving [1] for modeling inverse problems which arise from phase retrievals and in medical image reconstruction [2]. The SFP attracts many authors′ attention due to its application in signal processing. Various algorithms have been invented to solve it (see [39] and references therein).
Note that the split feasibility problem (1.1) can be formulated as a fixed-point equation by using the fact
()
that is, x* solves the SFP (1.1) if and only if x* solves the fixed point equation (1.2) (see [10] for the details). This implies that we can use fixed-point algorithms (see [1113]) to solve SFP. A popular algorithm that solves the SFP (1.1) is due to Byrne′s CQ algorithm [2] which is found to be a gradient-projection method (GPM) in convex minimization. Subsequently, Byrne [3] applied KM iteration to the CQ algorithm, and Zhao and Yang [14] applied KM iteration to the perturbed CQ algorithm to solve the SFP. It is well known that the CQ algorithm and the KM algorithm for a split feasibility problem do not necessarily converge strongly in the infinite-dimensional Hilbert spaces.

The split common fixed-point problem (SCFP) is a generalization of the split feasibility problem (SFP) and the convex feasibility problem (CFP); see [15]. In this paper, we introduce and study the convergence properties of a viscosity approximation algorithm for solving the SCFP for the class of quasi-nonexpansive operators S such that IS is demiclosed at the origin.

Now let us first recall the definition of quasi-nonexpansive operators which appear naturally when using subgradient projection operator techniques in solving some feasibility problems, and also some definitions of classes of operators often used in fixed-point theory and which are commonly encountered in the literature.

Let T : HH be a mapping. A point xH is said to be a fixed point of T provided that Tx = x. In this paper, we use F(T) to denote the fixed-point set and use → and ⇀ to denote the strong convergence and weak convergence, respectively. We use stand for the weak ω-limit set of {xk}.
  • (i)

    A mapping T : HH belongs to the general class ΦQ of (possibly discontinuous) quasi-nonexpansive mappings if

    ()

  • (ii)

    A mapping T : HH belongs to the set ΦN of nonexpansive mappings if

    ()

  • (iii)

    A mapping T : HH belongs to the set ΦFN of firmly nonexpansive mappings if

    ()

  • (iv)

    A mapping T : HH belongs to the set ΦFQ of firmly quasi-nonexpansive mappings if

    ()

It is easily observed that ΦFN ⊂ ΦN ⊂ ΦQ and that ΦFN ⊂ ΦFQ ⊂ ΦQ. Furthermore, ΦFN is well known to include resolvents and projection operators, while ΦFQ contains subgradient projection operators (see, e.g., [16] and the reference therein).

A mapping T : HH is called demiclosed at the origin if any sequence {xn} weakly converges to x, and if the sequence {Txn} strongly converges to 0, then Tx = 0. A mapping f : HH is called a contraction of modulus ρ ∈ [0,1) if
()
In what follows, we will focus our attention on the following general two-operator split common fixed-point problem:
()
where A : H1H2 is a bounded linear operator, U : H1H1 and S : H2H2 are two quasi-nonexpansive operators with nonempty fixed-point sets F(U) = C and F(S) = Q, and denote the solution set of the two-operator SCFP by
()
Recall that F(U) and F(S) are nonempty closed convex subsets of H1 and H2, respectively. If Γ ≠ , we have Γ which is close convex subset of H1. To solve (1.8), Censor and Segal [15] proposed and proved, in infinite-dimensional spaces, the convergence of the following algorithm:
()
where γ ∈ (0, 2/λ), with λ being the largest eigenvalue of the matrix AtA (At stands for matrix transposition). Very recently, Moudafi [17] introduced the following relaxed algorithm:
()

where uk = xk + γβA*(SI)Axk, β ∈ (0,1), αk ∈ (0,1), and γ ∈ (0,1/λβ), with λ being the spectral radius of the operator A*A. Moudafi proved weak convergence result of the algorithm in Hilbert spaces.

Inspired by their work, we introduce the following viscosity approximation algorithm.

Algorithm 1. Initialization: Let x0H be arbitrary.

Iterative step: Set T = U(I + γA*(SI)A). For kN, let

()
where f : HH is a contraction of modulus ρ,   ωk ∈ (0,1/2),   γ ∈ (0,1/λ) with λ being the spectral radius of the operator A*A, and αk ∈ (0,1).

This paper establishes the strong convergence of the sequence given by (1.12) to the unique solution of the variational inequality problem VIP(If, Γ):

()
Now we give a series of preliminary results needed for the convergence analysis of algorithm (1.12).

Lemma 1.1. Let H be a real Hilbert space and T : HH a quasi-nonexpansive mapping. Then, the following properties are reached:

  • (i)

    ;

  • (ii)

    and .

Remark 1.2. Let F : = If, where f is the contraction defined in (1.7). It is a simple matter to see that the operator F is (1 − ρ) strongly monotone over H; that is,

()

The next result is of fundamental importance for the techniques of analysis used in this paper. It was established in [18], and its proof is given for the sake of completeness.

Lemma 1.3 (see [18], Lemma 1.3.)Let {δn} be a sequence of real numbers that does not decrease at infinity, in the sense that there exists a subsequence of {δn} which satisfies for all j ≥ 0. Also consider the sequence of integers defined by

()
Then is a nondecreasing sequence verifying lim nτ(n) = , and, for all nn0, it holds that δτ(n)δτ(n)+1 and one has
()

Proof. Clearly, we can see that {τ(n)} is a well-defined sequence, and the fact that it is nondecreasing is obvious as well as lim nτ(n) = and δτ(n)δτ(n)+1. Let us prove (1.16). It is easily observed that τ(n) ≤ n. Consequently, we prove (1.16) by distinguishing the three cases: (c1) τ(n) = n; (c2) τ(n) = n − 1; (c3) τ(n) < n − 1. In the first case (i.e., τ(n) = n), (1.16) is immediately given by δτ(n)δτ(n)+1. In the second case (i.e., τ(n) = n − 1), (1.16) becomes obvious. In the third case (i.e., τ(n) ≤ n − 2), by (1.15) and for any integer nn0, we easily observe that δjδj+1 for τ(n) + 1 ≤ jn − 1; namely,

()
which entails the desired result.

2. Main Results

Theorem 2.1. Given a bounded linear operator A : H1H2, let U : H1H1 and S : H2H2 be quasi-nonexpansive mappings with nonempty fixed-point set F(U) = C and F(S) = Q. Assume that UI and SI are demiclosed at origin. Let {xk} be the sequence given by (1.12) with γ ∈ (0,1/λ), ωk ∈ (0,1/2) such that 0 < lim  inf kωk ≤ lim  sup kωk < 1/2 and {αk}⊂(0,1) such that lim kαk = 0 and ∑kαk = . If Γ ≠ , then the sequence {xk} strongly converges to a split common fixed-point x* ∈ Γ, verifying x* = PΓf(x*) which equivalently solves the following variational inequality problem:

()

Proof. Set . Then .

Firstly, we prove that {xk} is bounded. Taking y ∈ Γ, that is, yF(U), AyF(S). We have

()
From the definition of , we get
()
On the other hand, we have
()
From the definition of λ, it follows that
()
Now, by using property (ii) of Lemma 1.1, we obtain
()
Combining (2.4)–(2.6), we have
()
From property (i) of Lemma 1.1, we have
()
From (2.3) and (2.8), we have
()
Combining (2.2), (2.3), and (2.9), it follows that
()
  It is obviously that
()
and hence {xk} is bounded. Let x* = PΓf(x*). We have
()
and hence
()
By (2.9) we obtain that
()
It follows from (2.13) that
()
and hence
()
Setting , we have
()
so that (2.16) can be rewritten as
()
Now using (2.12) again, we have
()
which yields
()
From (2.18) and (2.20), we obtain
()
It follows from Remark 1.2 that
()
and hence
()

The rest of the proof will be divided into two parts.

Case 1. Suppose that there exists k0 such that is nonincreasing. In this situation, {δk} is convergent because it is nonnegative, so that lim k(δk+1δk) = 0; hence, in light of (2.21) together with αk → 0, the boundedness of {xk}, and 0 < liminf kωk ≤ limsup kωk < 1/2, we obtain

()
From (2.21) again, we have
()
By ∑kαk = , we deduce that
()
and hence (as )
()
By (2.23) and (2.27), we have
()
recalling that lim kδk exists, we obtain
()
Now we prove that
()
It follows from (2.7) and (2.24) that
()
and hence
()
Taking yωw(xk), from the demiclosedness of SI at 0, we obtain
()
Now, by setting uk = xk + γA*(SI)Axk, it follows that yωw(uk). On the other hand,
()
which, combined with the demiclosedness of UI at 0, yields
()
Hence, yC and y ∈ Γ. We can take subsequence of {xk} such that as j → and
()
which leads to
()
By (2.29), we have lim kδk = 0, and hence {xk} converges strongly to x*.

Case 2. Suppose there exists a subsequence of {δk} such that for all j ≥ 0. In this situation, we consider the sequence of indices {τ(k)} as defined in Lemma 1.3. It follows that δτ(k)+1δτ(k) > 0, which by (2.21) amounts to

()
By the boundedness of {xk} and αk → 0, we immediately obtain
()
Similar to Case 1, we have
()
It follows from (2.38) that
()
which in the light of (2.23) yields
()
hence (as ατ(k)f(xτ(k)) − xτ(k)2 → 0) it follows that
()
From (2.40) we have limsup kδτ(k) = 0, so that lim kδτ(k) = 0, and hence lim kxτ(k)x*∥ = 0. On the other hand, it follows that
()
which, by (2.39), implies that
()
So we have
()
Then, recalling that δkδτ(k)+1 (by Lemma 1.3), we get lim kδk = 0, so that the sequence {xk} converges strongly to x*.

Theorem 2.2. Given a bounded linear operator A : H1H2, let U : H1H1 and S : H2H2 be quasi-nonexpansive mappings with nonempty fixed-point set F(U) = C and F(S) = Q. Assume that UI and SI are demiclosed at origin. Let x0H be arbitrary and {xk} the sequence given by

()
where T = U(I + γA*(SI)A), f : HH a contraction of modulus ρ,   γ ∈ (0,1/λ), ω ∈ (0,1/2), and {αk}⊂(0,1) such that lim kαk = 0 and ∑kαk = . If Γ ≠ , then the sequence {xk} strongly converges to a split common fixed-point x* ∈ Γ, verifying x* = PΓf(x*) which equivalently solves the following variational inequality problem:
()

Acknowledgments

The research was supported by Fundamental Research Funds for the Central Universities (Program No. ZXH2011D005); it was also supported by science research foundation program in Civil Aviation University of China (2011kys02).

      The full text of this article hosted at iucr.org is unavailable due to technical difficulties.