Volume 2013, Issue 1 381715
Research Article
Open Access

Some Convergence Theorems for Contractive Type Mappings in CAT(0) Spaces

Kyung Soo Kim

Corresponding Author

Kyung Soo Kim

Graduate School of Education, Mathematics Education, Kyungnam University, Changwon, Kyungnam 631-701, Republic of Korea kyungnam.ac.kr

Search for more papers by this author
First published: 20 November 2013
Citations: 2
Academic Editor: Sehie Park

Abstract

We establish theorems of strong convergence, for the Ishikawa-type (or two step; cf. Ishikawa, 1974) iteration scheme, to a fixed point of a uniformly L-Lipschitzian asymptotically demicontractive mapping and a uniformly L-Lipschitzian hemicontractive mapping in CAT(0) space. Moreover, we will propose some open problems.

1. Introduction

Let (X, d) be a metric space. One of the most interesting aspects of metric fixed point theory is to extend a linear version of known result to the nonlinear case in metric spaces. To achieve this, Takahashi [1] introduced a convex structure in a metric space (X, d). A mapping W : X × X × [0,1] → X is a convex structure in X if
()
for all x, yX and λ ∈ [0,1]. A metric space together with a convex structure W is known as a convex metric space. A nonempty subset K of a convex metric space is said to be convex if
()
for all x, yK and λ ∈ [0,1]. In fact, every normed space and its convex subsets are convex metric spaces but the converse is not true, in general (see, [1]).

Example 1 (see [2].)Let X = {(x1, x2) ∈   2 : x1 > 0, x2 > 0}, for all x = (x1, x2), y = (y1, y2) ∈ X, and λ ∈ [0,1]. We define a mapping W : X × X × [0,1] → X by

()
and define a metric d : X × X → [0, ) by
()
Then we can show that (X, d, W) is a convex metric space, but it is not a normed linear space.

A metric space X is a CAT(0) space (the term is due to Gromov [3] and it is an acronym for E. Cartan, A. D. Aleksandrov, and V. A. Toponogov) if it is geodesically connected and if every geodesic triangle in X is at least as “thin” as its comparison triangle in the Euclidean plane (see, e.g., [4], page 159). It is well known that any complete, simply connected Riemannian manifold nonpositive sectional curvature is a CAT(0) space. The precise definition is given below. For a thorough discussion of these spaces and of the fundamental role they play in various branches of mathematics, see Bridson and Haefliger [4] or Burago et al. [5].

Let (X, d) be a metric space. A geodesic path joining xX to yX (or, more briefly, a geodesic from x to y) is a mapping c from a closed interval [0, l] ⊂ to X such that c(0) = x, c(l) = y, and d(c(t), c(t)) = |tt|, for all t, t ∈ [0, l]. In particular, c is an isometry and d(x, y) = l. The image α of c is called a geodesic (or metric) segment joining x and y. When it is unique, this geodesic is denoted by [x, y]. The space (X, d) is said to be a geodesic space if every two points of X are joined by a geodesic, and X is said to be uniquely geodesic if there is exactly one geodesic joining x and y for each x, yX. A subset YX is said to be convex if Y includes every geodesic segment joining any two of its points.

A geodesic triangle  Δ(x1, x2, x3) is a geodesic metric space (X, d) that consists of three points x1, x2, x3X (the vertices of Δ) and is a geodesic segment between each pair of vertices (the edges of Δ). A comparison triangle for the geodesic triangle Δ(x1, x2, x3) in (X, d) is a triangle in   2 such that for i, j ∈ {1,2, 3}. Such a triangle always exists (see, [4]).

A geodesic metric space is said to be a CAT(0)  space if all geodesic triangles of appropriate size satisfy the following CAT(0) comparison axiom.

Let Δ be a geodesic triangle in X and let be a comparison triangle for Δ. Then Δ is said to satisfy the CAT(0)  inequality if for all x, y ∈ Δ and all comparison points ,
()
Complete CAT(0) spaces are often called Hadamard spaces (see, [6]). If x, y1, y2 are points of a CAT(0) space and if y0 is the midpoint of the segment [y1, y2], which we will denote by y1y2/2, then the CAT(0) inequality implies
()
This inequality is the (CN) inequality of Bruhat and Tits [7]. In fact, a geodesic space is a CAT(0) space if and only if it satisfies the (CN) inequality (cf. [4], page 163). The previous inequality has been extended by Khamsi and Kirk [8] as
for any α ∈ [0,1] and x, y, zX. The inequality (CN*) also appeared in [9].
Let us recall that a geodesic metric space is a CAT(0) space if and only if it satisfies the (CN) inequality (see, [4], page 163). Moreover, if X is a CAT(0) metric space and x, yX, then for any α ∈ [0,1], there exists a unique point αx ⊕ (1 − α)y ∈ [x, y] such that
()
for any zX and [x, y] = {αx ⊕ (1 − α)y : α ∈ [0,1]}. In view of the previous inequality, CAT(0) space has Takahashi′s convex structure W(x, y, α) = αx ⊕ (1 − α)y. It is easy to see that for any x, yX and λ ∈ [0,1],
()
As a consequence,
()
Moreover, a subset K of CAT(0) space X is convex if for any x, yK, we have [x, y] ⊂ K.

Definition 2. Let C be a nonempty subset of a metric space (X, d). Let F(T) denote the fixed point set of T. Let F(T) ≠ .

  • (1)

    A mapping T : CC is said to be k-strict asymptotically pseudocontractive with sequence {an} if lim nan = 1 for some constant k, 0 ≤ k < 1 and

    ()

  • for all x, yC,  n.

  • If k = 0, then T is said to be asymptotically nonexpansive with sequence {an}, that is,

    ()

  • (2)

    A mapping T : CC is said to be asymptotically demicontractive with sequence {an} if lim nan = 1 for some constant k, 0 ≤ k < 1, and

    ()

  • for all xC,  n.

  • If k = 0, then T is said to be asymptotically quasi-nonexpansive with sequence {an}, that is,

    ()

  • (3)

    A mapping T : CC is said to be asymptotically pseudocontractive with sequence {an} if lim nan = 1 and

    ()

  • for all x, yC,  n.

  • (4)

    A mapping T : CC is said to be asymptotically hemicontractive with sequence {an} if lim nan = 1 and

    ()

  • for all xC,  n.

  • (5)

    A mapping T : CC is said to be uniformly  L-Lipschitzian if for some constant L > 0,

    ()

  • for all n.

Liu [10] has proved the convergence of Mann and Ishikawa iterative sequence for uniformly L-Lipschitzian asymptotically demicontractive and hemicontractive mappings in Hilbert space (cf. [11]). The existence of (common) fixed points of one mapping (or two mappings or family of mappings) is not known in many situations. So the approximation of fixed points of one or more nonexpansive, asymptotically nonexpansive, or asymptotically quasi-nonexpansive mappings by various iterations have been extensively studied in Banach spaces, convex metric spaces, CAT(0) spaces, and so on (see, [2, 6, 8, 9, 1227]).

In this paper, we establish theorems of strong convergence for the Ishikawa-type (or two step, cf. [28]) iteration scheme to a fixed point of a uniformly L-Lipschitzian asymptotically demicontractive mapping and a uniformly L-Lipschitzian asymptotically hemicontractive mapping in CAT(0) space. Moreover, we will propose some open problems.

2. Preliminaries

We introduce the following iteration process.

Let C be a nonempty convex subset of a CAT(0) space (X, d) and let T : CC be a given mapping. Let x1C be a given point.

Algorithm 3. The sequences {xn} and {yn} defined by the iterative process

()
is called an Ishikawa-type iterative sequence (cf. [28]).

If βn ≡ 0, then Algorithm 3 reduces to the following

Algorithm 4. The sequence {xn} defined by the iterative process

()
is called a Mann-type iterative sequence (cf. [29]).

Lemma 5 (see [10].)Let sequences {an}, {bn} satisfy that

()
an ≥ 0, for all n, is convergent, and {an} has a subsequence converging to 0. Then, we must have
()

3. Convergence Theorems

Lemma 6. Let (X, d) be a CAT(0) space and let C be a nonempty convex subset of X. Let T : CC be an uniformly L-Lipschitzian mapping and let {αn}, {βn} be sequence in [0,1]. Define the iteration scheme {xn} as Algorithm 3. Then

()
for all n ≥ 1.

Proof. Let Cn = d(xn, Tnxn). We have

()
From (22), we get
()
From (22) and (23), we get
()
This completes the proof of Lemma 6.

Theorem 7. Let (X, d) be a complete CAT(0) space, let C be a nonempty bounded closed convex subset of X, and let T : CC be a completely continuous and uniformly L-Lipschitzian and asymptotically demicontractive with sequence {an}, an ∈ [1, ), , εαn ≤ 1 − kε, for all n and some ε > 0. Given x0C, define the iteration scheme {xn} by

()
Then {xn} converges strongly to some fixed point of T.

Proof. Since T is a completely continuous mapping in a bounded closed convex subset C of complete metric space, from Schauder′s theorem, F(T) is nonempty. It follows from (CN*) inequality that

()
for all pF(T). Since T is a asymptotically demicontractive, we get
()
Since 0 < εαn ≤ 1 − kε, we have 1 − kαnε. Thus,
()
From (27), we have
()
for all pF(T). Since C is bounded and T is self-mapping in C, there exist some M > 0 so that d2(xn, p) ≤ M, for all n. Since 0 ≤ αn ≤ 1, it follows from (29) that
()
Therefore,
()
So
()
for all m. Since , we get
()
Therefore,
()
Since T is a uniformly L-Lipschitzian, it follows from Lemma 6 that
()
Since {xn} is a bounded sequence and T is completely continuous, there exist a convergent subsequence of {Txn}. Therefore, from (35), {xn} has a convergent subsequence . Let . It follows from the continuity of T and (35), we have q = Tq. Therefore, {xn} has a subsequence which converges to the fixed point q of T. Let p = q in the inequality (30). Since and , from (30) and Lemma 5, we have
()
Therefore,
()
This completes the proof of Theorem 7.

Corollary 8. Let (X, d) be a complete CAT(0) space, let C be a nonempty bounded closed convex subset of X, and let T : CC be a completely continuous and uniformly L-Lipschitzian and k-strict asymptotically pseudocontractive with sequence {an}, an ∈ [1, ), , and εαn ≤ 1 − kε, for all n and some ε > 0. Given x0C, define the iteration scheme {xn} by

()
Then {xn} converges strongly to some fixed point of T.

Proof. By Definition 2, T is k-strict asymptotically pseudocontractive; then T must be asymptotically demicontractive. Therefore, Corollary 8 can be proved by using Theorem 7.

Lemma 9. Let (X, d) be a CAT(0) space and let C be a nonempty convex subset of X. Let T : CC be an uniformly L-Lipschitzian and asymptotically hemicontractive with sequence {an}⊂[1, ), for all n, and F(T) is nonempty. Define the iteration scheme {xn} as follows:

()
Then the following inequality holds:
()
for all pF(T).

Proof. It follows from (CN*) inequality that

()
()
for all pF(T). Since T is asymptotically hemicontractive, we get
()
()
From (42) and (44), we have
()
From (CN*) inequality, we have
()
Substituting (45) and (46) into (43), we get
()
From (41) and (47), we obtain
()
Since T is uniformly L-Lipschitzian, we have
()
Substituting (49) into (48), we obtain
()
This completes the proof of Lemma 9.

Lemma 10. Let (X, d) be a CAT(0) space and let C be a nonempty bounded convex subset of X. Let T : CC be a uniformly L-Lipschitzian and asymptotically hemicontractive with sequence {an}⊂[1, ), for all n and . Let F(T) be nonempty. Given x1C, define the iteration scheme {xn} by

()
If εαnβnb for some ε > 0 and , then
()

Proof. First, we will prove lim nd(xn, Tnxn) = 0. From Lemma 9 and 0 ≤ αnβn, we have

()
Thus
()
Since , we have lim n(an − 1) = 0. Hence, {an} is bounded. By boundedness of C and 0 ≤ αnβn ≤ 1, we obtain that {αn(1 + anβn)d2(xn, p)} is bounded. Therefore, there exists a constant M > 0 such that
()
From (54) and (55), we get
()

Let D = 1 − 2bL2b2 > 0. Since lim nan = 1, there exists N such that

()
for all nN. Suppose that lim nd(xn, Tnxn) ≠ 0, then there exist a ε0 > 0 and a subsequence of {xn} such that
()
Without loss of generality, we let n1N. From (56), we have
()
so
()
From (57)–(60) and εαnβn, we obtain
()
Since and the boundedness of C, the right side of (61) is bounded. However, if we have i, then the left side of (61) is unbounded. This is a contradiction. Therefore,
()
Since T is a uniformly L-Lipschitzian, from Lemma 6, we get
()
This completes the proof of Lemma 10.

Theorem 11. Let (X, d) be a complete CAT(0) space, let C be a nonempty bounded closed convex subset of X, and let T : CC be a completely continuous and uniformly L-Lipschitzian and asymptotically hemicontractive with sequence {an}⊂[1, ) satisfying , for all n. Given x1C, define the iterative scheme {xn} by

()
If {αn}, {βn}⊂[0,1] with εαnβnb for some ε > 0 and , then {xn} converges strongly to some fixed point of T.

Proof. Since T is a completely continuous mapping in a bounded closed convex subset C of complete metric space, from Schauder′s theorem, F(T) is nonempty. Since T is completely continuous, there exist a convergent subset of {Txn}. Let

()
Since lim nd(xn, Txn) = 0, from Lemma 10, we have
()
On the other hand, from the continuity of T, (66), and Lemma 10, we have
()
This means that p is a fixed point of T. From (55), (57), and αnβn, we obtain Lemma 9 that
()
From (66), there exists a subsequence of {d2(xn, p)} which converges to 0. Therefore, from Lemma 5 and (68),
()
Hence,
()
This completes the proof of Theorem 11.

Corollary 12. Let (X, d) be a complete CAT(0) space, let C be a nonempty bounded closed convex subset of X, and let T : CC be a completely continuous and uniformly L-Lipschitzian and asymptotically pseudocontractive with sequence {an}⊂[1, ) satisfying , for all n. Given x1C, define the iterative scheme {xn} by

()
If {αn}, {βn}⊂[0,1] with εαnβnb for some ε > 0 and , then {xn} converges strongly to some fixed point of T.

Proof. By Definition 2, T is an asymptotically pseudocontractive mapping, then T is an asymptotically hemicontractive mapping. Since an ∈ [1, ), we have . Obviously, . Therefore, Corollary 12 can be proved by using Theorem 11.

4. Some Remarks and Open Problems

Let S be a semigroup. We denote by B(S) the space of all bounded real-valued functions defined on S with supremum norm. For each sS, we define the left and right translation operators ls and rs on B(S) by
()
for each tS and fB(S), respectively. Let X be a subspace of B(S) containing 1. An element μ in the dual space X* of X is said to be a mean on X if ∥μ∥ = μ(1) = 1. For sS, we can define a point evaluation δs by δs(f) = f(s) for each fX. It is well known that μ is mean on X if and only if
()
for each fX. Each mean on X is the weak*-limit of convex combination of point evaluations.
Let X be a translation invariant subspace of B(S) (i.e., lsXX and rsXX for each sS) containing 1. Then a mean μ on X is said to be left invariant (resp., right invariant) if
()
for each sS and fX. A mean μ on X is said to be invariant if μ is both left and right invariant ([3034]). X is said to be left (resp., right) amenable if X has a left (resp., right) invariant mean. X is amenable if X is left and right amenable. In this case, we say that the semigroup S is an amenable semigroup (see [35, 36]). Moreover, B(S) is amenable when S is a commutative semigroup or a solvable group. However, the free group or semigroup of two generators is not left or right amenable.
A net {μα} of means on X is said to be asymptotically left (resp., right) invariant if
()
for each fX and sS, and it is said to be left (resp., right) strongly asymptotically invariant (or strong regular) if
()
for each sS, where and are the adjoint operators of ls and rs, respectively. Such nets were first studied by Day in [35] where they were called weak* invariant and norm invariant, respectively.

It is easy to see that if a semigroup S is left (resp., right) amenable, then the semigroup S = S ∪ {e}, where es = se = s for all sS is also left (resp., right) amenable and conversely.

A semigroup S is called left reversible if any two right ideals of S have nonvoid intersection, that is, aSbS for a, bS. In this case, (S, ⪯) is a directed system when the binary relation “⪯” on S is defined by ab if and only if {a} ∪ aS⊇{b} ∪ bS for a, bS. It is easy to see that tts for all t, sS. Further, if ts, then ptps for all pS. The class of left reversible semigroup includes all groups and commutative semigroups. If a semigroup S is left amenable, then S is left reversible. But the converse is not true ([31, 3741]).

Let S be a semigroup and F(T) denote the fixed point set of T. Then = {Ts : sS} is called a representation of S if Te = I and Tst = TsTt for each s, tS. We denote by F() the set of common fixed points of {Ts : sS}, that is,
()

Open Problem 1. It will be interesting to obtain a generalization of both Theorems 7 and 11 to commutative, amenable, and reversible semigroups as in the case of Hilbert spaces or some Banach spaces (cf. [8, 30, 32, 4245]).

For a real number κ, a CAT(κ) space is defined by a geodesic metric space whose geodesic triangle is sufficiently thinner than the corresponding triangle in a model space with curvature κ.

For κ = 0, the 2-dimensional model space is the Euclidean space   2 with the metric induced from the Euclidean norm. For κ > 0,   is the 2-dimensional sphere whose metric is length of a minimal great arc joining each two points. For κ < 0, is the 2-dimensional hyperbolic space with the metric defined by a usual hyperbolic distance. For more details about the properties of CAT(κ) spaces, see [4, 4648].

Open Problem 2. It will be interesting to obtain a generalization of both Theorems 7 and 11 to CAT(κ) space.

Conflict of Interests

The author declares that there is no conflict of interests regarding the publication of this paper.

    Acknowledgments

    The author would like to thank Professor A. T.-M Lau and Professor J. K. Kim for their helpful suggestions. Also, special thanks are due to the referees for their deep insight which improved the presentation of this paper. This research was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (2012R1A1A4A01010526).

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