Strong Convergence Theorems by Shrinking Projection Methods for Class T Mappings

Volume 2011, Article ID 681214,7pages doi:10.1155/2011/681214

Research Article

Strong Convergence Theorems by Shrinking Projection Methods for Class T ^Mappings

Qiao-Li Dong,

Songnian He,

and Fang Su

1College of Science, Civil Aviation University of China, Tianjin 300300, China

2Tianjin Key Laboratory for Advanced Signal Processing, Civil Aviation University of China, Tianjin 300300, China

3Department of Mathematics and Systems Science, National University of Defense Technology, Changsha 410073, China

Correspondence should be addressed to Qiao-Li Dong,[email protected] Received 9 December 2010; Accepted 2 February 2011

Academic Editor: S. Al-Homidan

Copyrightq2011 Qiao-Li Dong et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

We prove a strong convergence theorem by a shrinking projection method for the class ofT mappings. Using this theorem, we get a new result. We also describe a shrinking projection method for a nonexpansive mapping on Hilbert spaces, which is the same as that of Takahashi et al.2008.

1. Introduction

LetHbe a real Hilbert space with inner product·,·and norm · , and letCbe a nonempty closed convex subset ofH. Recall that a mappingT :H → His said to be nonexpansive if Tx−Ty ≤ x−yfor allx, y∈H. The set of fixed points ofTis FixT:{x∈H:Txx}.

T : H → His said to be quasi-nonexpansive if FixTis nonempty andTx−p ≤ x−pfor allx∈Handp∈FixT.

Givenx, y∈H, let H

x, y :

z∈H:

z−y, x−y

≤0

1.1 be the half-space generated byx, y. A mappingT :H → His said to be the classTor a cutterifT ∈T{T :H → H|domT H and FixT⊂Hx, Tx, for allx∈H}.

Remark 1.1. The class T is fundamental because it contains several types of operators commonly found in various areas of applied mathematics and in particular in approximation and optimization theorysee1for details.

Combettes 2, Bauschke, and Combettes 1 studied properties of the class T mappings and presented several algorithms. They introduced an abstract Haugazeau method in1as follows: startingx0∈H,

xn1PHx0,xn∩Hxn,Tnxnx0. 1.2 Using Lemma 1.2 given below and the fact that a nonexpansive mapping is quasi- nonexpansive, one can easily obtain hybrid methods introduced by Nakajo and Takahashi 3for a nonexpansive mapping.

Recently, Takahashi et al.4proposed a shrinking projection method for nonexpan- sive mappingsTn:C → C. Letx0 ∈H,C1C,x1PC1x0, and

ynαn 1−αnTnxn, C_n1

z∈C_n:y_n−z ≤ x_n−z , xn1PCn1x0, n1,2, . . . ,

1.3

where 0≤α_n ≤a <1,P_Kdenotes the metric projection fromHonto a closed convex subset KofH.

Inspired by Bauschke and Combettes 1 and Takahashi et al. 4, we present a shrinking projection method for the class ofTmappings. Furthermore, we obtain a shrinking projection method for a nonexpansive mapping on Hilbert spaces, which is the same as presented by Takahashi et al.4.

We will use the following notations:

1for weak convergence and → for strong convergence;

2ω_wx_n {x:∃x_nj x}denotes the weakω-limit of{x_n}.

We need some facts and tools in a real Hilbert spaceHwhich are listed below.

Lemma 1.2see1. LetHbe a Hilbert space. LetIbe the identity operator ofH.

iIf domT H, then 2T−Iis quasi-nonexpansive if and only ifT ∈T.

iiIfT ∈T, thenλI 1−λT ∈T, for allλ∈0,1.

Definition 1.3. LetTn ∈Tfor eachn. The sequence{Tn}is called to be coherent if, for every bounded sequence{v_n}inH, there holds

∞ n0

vn1−vn²<∞, ∞

n0

v_n−T_nv_n²<∞,

⇒ωwvn⊂ ^∞

n0

FixTn. 1.4

Definition 1.4. T is called demiclosed aty ∈ Hif Tx ywhenever{x_n} ⊂ H,x_n xand Txn → y.

Next lemma shows that nonexpansive mappings are demeiclosed at 0.

Lemma 1.5Goebel and Kirk5. LetCbe a closed convex subset of a real Hilbert space H, and letT :C → Cbe a nonexpansive mapping such that FixT/∅. If a sequence{x_n}inCis such that xn zandxn−Txn → 0, thenzTz.

Lemma 1.6see6. LetKbe a closed convex subset ofH. Let{xn}be a sequence inHandu∈H.

LetqP_Ku. Ifx_nis such thatω_wx_n⊂Kand satisfies the condition

xn−u ≤ u−q, ∀n, 1.5

thenxn → q.

Lemma 1.7Goebel and Kirk5. LetKbe a closed convex subset of real Hilbert spaceH, and let PKbe the (metric or nearest point) projection fromHontoK(i.e., forx∈H,PKxis the only point inKsuch thatx−PKxinf{x−z:z∈K}). Givenx∈Handz∈K, thenzPKxif and only if there holds the relation

x−z, y−z

≤0, ∀y∈K. 1.6

2. Main Results

In this section, we will introduce a shrinking projection method for the class ofTmappings and prove strong convergence theorem.

Theorem 2.1. LetT_n∈Tfor eachnsuch thatF:_∞

n1FixT_n/∅. Suppose that the sequence{T_n} is coherent. Letx0∈H. ForC1Handx1x0, define a sequence{xn}as follows:

x_n1P_Cn1x0, n1,2, . . . ,

Cn1{z∈Cn:z−Tnxn, xn−Tnxn ≤0}. 2.1

Then,{x_n}converges strongly toP_Fx0.

Proof. We first show by induction thatF ⊂ Cn for all n ∈ N.F ⊂ C1 is obvious. Suppose F ⊂C_kfor somek∈N. Note that, by the definition ofT_k∈T, we always haveF ⊂FixT_k⊂ Hx_k, T_kx_k, that is,

z−Tkxk, xk−Tkxk ≤0, ∀z∈ F. 2.2

From the definition ofCk1andF ⊂Ck, we obtainF ⊂Ck1. This implies that

F ⊂Cn, ∀n∈N. 2.3

It is obvious thatC1His closed and convex. So, from the definition,Cnis closed and convex for alln∈N. So we get that{xn}is well defined.

Sincex_nis the projection ofx0ontoC_nwhich containsF, we have

x0−xn ≤ x0−y, ∀y∈Cn. 2.4

TakingyPFx0 ∈ F, we get

x0−xn ≤ x0−PFx0. 2.5

The last inequality ensures that{x0−xn}is bounded. FromxnPCnx0andxn1PCn1x0∈ Cn1⊂Cn, usingLemma 1.7, we get

xn1−xn, x0−xn ≤0. 2.6

It follows that

x0−xn1²x0−xn−xn1−xn²

x0−xn²−2x0−xn, xn1−xnxn1−xn²

≥ x0−xn²xn1−xn²

≥ x0−xn².

2.7

Thus{xn−x0}is increasing. Since{xn−x0}is bounded, lim_{n→ ∞}xn−x0exists. From 2.7, it follows that

xn1−xn²≤ x0−xn1²− x0−xn², 2.8 and_∞

n1xn1−xn²<∞. On the other hand, byxn1PCn1x0∈Cn1, we have

xn1−Tnxn, xn−Tnxn ≤0. 2.9

Hence,

x_n1−x_n²x_n1−T_nx_n−x_n−T_nx_n²

x_n1−T_nx_n²−2x_n1−T_nx_n, x_n−T_nx_nx_n−T_nx_n²

≥ xn1−T_nx_n²x_n−T_nx_n².

2.10

We therefore get _∞

n1x_n−T_nx_n² < ∞. Since the sequence {T_n} is coherent, we have ωwxn⊂ F. FromLemma 1.6and2.5, the result holds.

Remark 2.2. We takeC1Hso thatF ⊂C1is satisfied.

Theorem 2.3. LetTn∈Tfor eachnsuch thatF:_∞

n1FixTn/∅. Suppose that the sequence{Tn} is coherent. Letx0∈H. ForC1Handx1x0, define a sequence{x_n}as follows:

y_nα_nx_n 1−α_nT_nx_n, Cn1

z∈Cn:

z−yn, xn−yn

≤0 , xn1PCn1x0, n1,2, . . . ,

2.11

where 0≤α_n≤a <1. Then,{x_n}converges strongly toP_Fx0.

Proof. SetS_n α_nI 1−α_nT_n. ByLemma 1.2ii, we have thatS_n ∈T. Fromx_n−S_nx_n 1−αnxn −Tnxn, it follows that1−axn−Tnxn ≤ xn−Snxn ≤ xn −Tnxnwhich implies that the sequence{S_n}is coherent. It is obvious that FixS_n FixT_n, for alln∈N.

HenceF_∞

n1FixS_{n ∞}

n1FixT_n. UsingTheorem 2.1, we get the desired result.

3. Deduced Results

In this section, usingTheorem 2.3, we obtain some new strong convergence results for the class of T mappings, a quasi-nonexpansive mapping and a nonexpansive mapping in a Hilbert space.

Theorem 3.1. LetT ∈Tsuch that FixT/∅and satisfying thatI−Tis demiclosed at 0. Letx0∈H.

ForC1Handx1x0, define a sequence{xn}as follows:

y_nα_nx_n 1−α_nTx_n, Cn1

z∈Cn:z−yn, xn−yn ≤0 , xn1PCn1x0, n1,2, . . . ,

3.1

where 0≤α_n≤a <1. Then,{x_n}converges strongly toP_FixTx0.

Proof. LetTn T in2.11for alln ∈ N. Following the proof ofTheorem 2.1, we can easily get2.5and_∞

n1x_n−Tx_n² <∞. By2.5, we obtain that{x_n}is bounded andω_wx_nis nonempty. For anyx ∈ω_wx_n, there exists a subsequence{x_nj}of the sequence{x_n}such thatxnj x. From _∞

n1xn−Txn² < ∞, it follows thatxn−Txn → 0. Since I−T is demiclosed at 0, we getx ∈FixT. Thusω_wx_n ⊂FixTwhich together withLemma 1.6 and2.5implies thatxn → PFixTx0.

Theorem 3.2. LetH be a Hilbert space. Let Sbe a quasi-nonexpansive mapping on H such that FixS/∅and satisfying thatI−Sis demiclosed at 0. Letx0∈H. ForC1Handx1x0, define a sequence{xn}as follows:

u_nα_nx_n 1−α_nSx_n, C_n1{z∈C_n:z−u_n ≤ x_n−z},

xn1PCn1x0, n1,2, . . . ,

3.2

where 0≤α_n≤a <1. Then,{x_n}converges strongly toP_FixSx0.

Proof. ByLemma 1.2i,SI/2 ∈ T. Substitute T in 3.1bySI/2. Then yn 1 α_n/2x_n 1−α_n/2Sx_n. Setu_n2y_n−x_nα_nx_n 1−α_nSx_n, theny_n u_nx_n/2. So, we have

Cn1

z∈Cn:

z−yn, xn−yn

≤0 {z∈C_n:2z−x_nu_n, x_n−u_n ≤0}

{z∈C_n:z−u_n ≤ x_n−z}.

3.3

SinceI−Sis demiclosed at 0,I−SI/2 I−S/2 is demiclosed at 0. So we can obtain the result by usingTheorem 3.1.

Since a nonexpansive mapping is quasi-nonexpansive, using Lemma 1.5 and Theorem 3.2, we have following corollary.

Corollary 3.3. LetHbe a Hilbert space. LetSbe a nonexpansive mappingHsuch that FixS/∅.

Letx0∈H. ForC1Handx1x0, define a sequence{x_n}as follows:

unαnxn 1−αnSxn, C_n1{z∈C_n: z−u_n ≤ x_n−z},

xn1PCn1x0, n1,2, . . . ,

3.4

where 0≤αn≤a <1. Then,{xn}converges strongly toPFixSx0.

Remark 3.4. Corollary 3.3is a special case of Theorem 4.1 in4whenC1H.

Acknowledgments

The authors would like to express their thanks to the referee for the valuable comments and suggestions for improving this paper. This paper is supported by Research Funds of Civil Aviation University of China Grant08QD10Xand Fundamental Research Funds for the Central Universities GrantZXH2009D021.

References

1 H. H. Bauschke and P. L. Combettes, “A weak-to-strong convergence principle for Fej´er-monotone methods in Hilbert spaces,” Mathematics of Operations Research, vol. 26, no. 2, pp. 248–264, 2001.

2 P. L. Combettes, “Quasi-Fej´erian analysis of some optimization algorithms,” in Inherently Parallel Algorithms in Feasibility and Optimization and Their Applications, vol. 8, pp. 115–152, North-Holland, Amsterdam, The Netherlands, 2001.

3 K. Nakajo and W. Takahashi, “Strong convergence theorems for nonexpansive mappings and nonexpansive semigroups,” Journal of Mathematical Analysis and Applications, vol. 279, no. 2, pp. 372–

379, 2003.

4 W. Takahashi, Y. Takeuchi, and R. Kubota, “Strong convergence theorems by hybrid methods for families of nonexpansive mappings in Hilbert spaces,” Journal of Mathematical Analysis and Applications, vol. 341, no. 1, pp. 276–286, 2008.

5 K. Goebel and W. A. Kirk, Topics in Metric Fixed Point Theory, vol. 28 of Cambridge Studies in Advanced Mathematics, Cambridge University Press, Cambridge, UK, 1990.

6 C. Martinez-Yanes and H.-K. Xu, “Strong convergence of the CQ method for fixed point iteration processes,” Nonlinear Analysis. Theory, Methods & Applications, vol. 64, no. 11, pp. 2400–2411, 2006.