Open Access

Some fixed point-type results for a class of extended cyclic self-mappings with a more general contractive condition

Fixed Point Theory and Applications20112011:59

https://doi.org/10.1186/1687-1812-2011-59

Received: 16 May 2011

Accepted: 28 September 2011

Published: 28 September 2011

Abstract

This article discusses a more general contractive condition for a class of extended (p ≥ 2) -cyclic self-mappings on the union of a finite number of subsets of a metric space which are allowed to have a finite number of successive images in the same subsets of its domain. If the space is uniformly convex and the subsets are non-empty, closed and convex, then all the iterates converge to a unique closed limiting finite sequence which contains the best proximity points of adjacent subsets and reduces to a unique fixed point if all such subsets intersect.

1. Introduction

A general contractive condition of rational type has been proposed in [1, 2] for a partially ordered metric space. Results about the existence of a fixed point and then its uniqueness under supplementary conditions are proved in those articles. The general rational contractive condition of [3] includes as particular cases several of the existing ones [1, 412] including Banach's principle [5] and Kannan's fixed point theorems [4, 8, 9, 11]. The general rational contractive conditions of [1, 2] are applicable only on distinct points of the considered metric spaces. In particular, the fixed point theory for Kannan's mappings is extended in [4] by the use of a non-increasing function affecting to the contractive condition and the best constant to ensure that a fixed point is also obtained. Three fixed point theorems which extended the fixed point theory for Kannan's mappings were proved in [11]. On the other hand, important attention has been paid during the last decades to the study of standard contractive and Meir-Keeler-type contractive cyclic self-mappings (see, for instance, [1322]). More recent investigation about cyclic self-mappings is being devoted to its characterization in partially ordered spaces and to the formal extension of the contractive condition through the use of more general strictly increasing functions of the distance between adjacent subsets. In particular, the uniqueness of the best proximity points to which all the sequences of iterates converge is proven in [14] for the extension of the contractive principle for cyclic self-mappings in uniformly convex Banach spaces (then being strictly convex and reflexive [23]) if the p subsets A i X of the metric space (X, d), or the Banach space (X, || ||), where the cyclic self-mappings are defined are non-empty, convex and closed. The research in [14] is centred on the case of the cyclic self-mapping being defined on the union of two subsets of the metric space. Those results are extended in [14] for Meir-Keeler cyclic contraction maps and, in general, for the self-mapping T : i p ̄ A i i p ̄ A i be a p(≥ 2) -cyclic self-mapping being defined on any number of subsets of the metric space with p ̄ : = 1 , 2 , . . . , p .

Other recent researches which have been performed in the field of cyclic maps are related to the introduction and discussion of the so-called cyclic representation of a set M, decomposed as the union of a set of non-empty sets as M = i = 1 m M i , with respect to an operator f: MM[24]. Subsequently, cyclic representations have been used in [25] to investigate operators from M to M which are cyclic φ-contractions, where φ: R0+R0+ is a given comparison function, M X and (X, d) is a metric space. The above cyclic representation has also been used in [26] to prove the existence of a fixed point for a self-mapping defined on a complete metric space which satisfies a cyclic weak φ-contraction. In [27], a characterization of best proximity points is studied for individual and pairs of non-self-mappings S, T: AB, where A and B are non-empty subsets of a metric space. In general, best proximity points do not fulfil the usual "best proximity" condition x = Sx = Tx under this framework. However, best proximity points are proven to jointly globally optimize the mappings from x to the distances d(x, Tx) and d(x, Sx). Furthermore, a class of cyclic φ-contractions, which contain the cyclic contraction maps as a subclass, has been proposed in [28] to investigate the convergence and existence results of best proximity points in reflexive Banach spaces completing previous related results in [14]. Also, the existence and uniqueness of best proximity points of p(≥ 2) -cyclic φ-contractive self-mappings in reflexive Banach spaces has been investigated in [29].

In this article, it is also proven that the distance between the adjacent subsets A i , Ai+1 X are identical if the p(≥ 2) -cyclic self-mapping is non-expansive [16]. This article is devoted to a generalization of the contractive condition of [1] for a class of extended cyclic self-mappings on any number of non-empty convex and closed subsets A i X, i p ̄ . The combination of constants defined the contraction may be different on each of the subsets and only the product of all the constants is requested to be less than unity. On the other hand, the self-mapping can perform a number of iterations on each of the subsets before transferring its image to the next adjacent subset of the p(≥ 2) -cyclic self-mapping. The existence of a unique closed finite limiting sequence on any sequence of iterates from any initial point in the union of the subsets is proven if X is a uniformly convex Banach space and all the subsets of X are non-empty, convex and closed. Such a limiting sequence is of size qp (with the inequality being strict if there is at least one iteration with image in the same subset as its domain) where p of its elements (all of them if q = p) are best proximity points between adjacent subsets. In the case that all the subsets A i X intersect, the above limit sequence reduces to a unique fixed point allocated within the intersection of all such subsets.

2. Main results for non-cyclic self-mappings

Let (X, d) be a metric space for a metric d: X × XR0+ with a self-mapping T: XX which has the following contractive condition proposed and discussed in [1]:
d T x , T y α d x , T x d y , T y d x , y + β d x , y , x , y x X
(2.1)
for some real constants α, β R0+ and α + β < 1 where R0+ = {r R: r ≥ 0}. A more general one involving powers of the distance is the following:
d s x , y T x , T y α d σ x , y x , T x d r x , y y , T y d σ x , y x , y + β d t x , y x , y , x , y x X ,
(2.2)

where s, σ, r, t: X × XR+ = {r R: r > 0} are continuous and symmetric with respect to the order permutation of the arguments x and y. It is noted that if x = y then (2.1) has a sense only if x is a fixed point, i.e. x = y = Tx = Ty implies that (2.1) reduces to the inequality "0 ≤ 0". The following result holds:

Theorem 2.1: Assume that the condition (2.2) holds for some symmetric continuous functions subject to 0 < r(x, y) ≤ s(x, y)+ln(P-d(Tx, Ty)) if r(x, y) ≠ s(x, y) and 0 < t x , y s x , y + l n Q - d x , y if t(x, y) ≠ s(x, y) for some real constants α , β , P , Q 0 , subject to the constraint α P + β Q < 1 . Then, d(Tn+1x, T n x) → 0 as n → ∞; x X. Furthermore, T n x n N 0 is a Cauchy sequence.

If, in addition, (X, d) is complete then T n xz as n → ∞, for some z X. If, furthermore, T: XX is continuous, then z = Tz is the unique fixed point of T: XX.

Proof: If y = Tx, then the above given constraints on the symmetric functions become 0 < r(x, x) ≤ s(x, x)+ln(P-d(Tx, T2x)) if r(x, x) ≠ s(x, x) and 0 < t x , x s x , x + ln Q - d x , T x if t(x, x) ≠ s(x, x). If y = x = Tx, then d(Tn+1x, T n x) → 0 as n → ∞; x X follows directly from (2.2) since ds(x, x)(Tn+1x, T n x) = 0. Now, take y = Tx so that for any xTx for x, Tx X and note that the conditions 0 < r(x, x) ≤ s(x, x)+ln(P-d(Tx, T2x)) if r(x, x) ≠ s(x, x) and 0 < t x , x s x , x + l n Q - d x , T x if t(x, x) ≠ s(x, x) are identical to
d r x , x x , T x P d s ( x , x ) x , T x ; d t x , x x , T x Q d s ( x , x ) x , T x
(2.3)
Thus, one gets from (2.1):
d s ( x , x ) T x , T 2 x α d r x , x T x , T 2 x + β d t x , x x , T x α P d s ( x , x ) T x , T 2 x + β Q d s ( x , x ) x , T x
(2.4)
so that, since k : = β Q 1 - α P < 1 , one gets from (2.4) proceeding by complete induction for n N0
0 d s ( x , x ) T n + 2 x , T n + 1 x β Q 1 - α P d s ( x , x ) T n + 1 x , T n x k n d s ( x , x ) T x , x 0 as n
(2.5)
what implies d(Tn+1x, T n x) ≤ kn/s(x, x)d(Tx, x) → 0 as n → ∞; x X. Taking n, m(≥ n+2) N0, one can get from (2.5):
d s ( x , x ) T m x , T n + 1 x j = n m - 1 k j d s ( x , x ) T x , x k n 1 - k d s ( x , x ) T x , x 0 a s n
(2.6)
so that
d T m x , T n + 1 x j = n m - 1 k j 1 s x , x d T x , x k n 1 - k 1 s ( x , x ) d T x , x 0 a s n , m
(2.7)
what proves that T n x n N 0 is a Cauchy sequence. Such a Cauchy sequence has a limit z = lim n T n x in X if (X, d) is complete from the convergence property of Cauchy sequences to points of the space X. If, in addition, T: XX is continuous then T z = T lim n T n x = lim n T n + 1 x = z so that the limit of the sequence is a fixed point. The uniqueness of the fixed point is now proven (i.e. z is not dependent on of x) by contradiction. Assume that there exists two distinct fixed points y = Ty and z = Tz in X. Then, from (2.5):
d T y , y = d T z , z = 0 0 < d y , z d y , T y + d T y , T z + d T z , z = d T y , T z
so that
0 < d s ( y , z ) T y , T z α d σ y , z y , T y d r y , z z , T z d σ y , z y , z + β d t y , z y , z = β d t y , z T y , T z β Q d s y , . z T y , T z

what implies β Q 1 if d(Ty, Tz) = d(y, z) > 0 contradicting d T y , T z = d y , z > 0 . Thus, y = z and hence the theorem. □

A simpler contractive condition leads to a close result to Theorem 2.1 as follows:

Corollary 2.2: Assume that the condition (2.2) is modified as follows:
d s T x , T y α d s x , T x d s y , T y d s x , y + β d s x , y
(2.8)

for some real constants s R+, α, β R0+, subject to α+β < 1. Then, Theorem 2.1 holds.

Proof: Taking P = Q = 1 then (2.3)-(2.7) hold by replacing r(x), s(x), t(x) → s R+. Thus, Theorem 2.1 holds for this particular case. Hence, the corollary. □

3. Main results for p(≥ 2) -cyclic self-mappings and extended p-cyclic self-mappings

Let T : i p ̄ A i i p ̄ A i be an extended p(≥ 2) -cyclic self-mapping where A i A i+kp X; i p ̄ : = 1 , 2 , . . . . , p , k N subject to the constraints T(A i ) A i Ai+1, T (A i ) Ai+1; j i - 1 ¯ and T j i A i A i + 1 for some finite integers j i ≥ 1; i p ̄ (this implies that q : = i = 1 p j i p with equality standing if and only if j i ≥ 1; i p ̄ , i.e. if the cyclic mapping is of standard type) with T k = T T k - 1 and T0id. It is noted that the extended p(≥ 2) -cyclic self-mapping T : i p ̄ A i i p ̄ A i is characterized by the p-tuple of integers j i : i p ̄ , where i = 1 p j i = q p and if, in particular, j i = 1; i p ̄ then T : i p ̄ A i i p ̄ A i is the standard p-cyclic self-mapping. It is also noted that the self-mappings T j i : i p ̄ A i i p ̄ A i and the composed mappings T q + j i : i p ̄ A i i p ̄ A i satisfying the extended inclusion constraint T(A i ) A i Ai+1, subject to T (A i ) A i , T j i A i A i + 1 ; j i - 1 ¯ ; i p ̄ , are not q-cyclic self-mappings [1317], except if q = p, since T q + j A i A i + fails for i , i p ̄ unless j j i . The contractive condition (2.1) becomes modified as follows:
d s x , y T x , T y α i d σ x , y x , T x d r x , y y , T y d σ x , y x , y + β i d t x , y x , y + γ i D s x , y
(3.1)
for x, y A i Ai+1, Tx A i Ai+1, Ty Ai+1 Ai+2and some real constants γ i R0+ while Tx, Ty are not both in the same subset A j for j = i, i+1, i+2 for any i p ̄ , and
d s x , y T x , T y α i d σ x , y x , T x d r x , y y , T y d σ x , y x , y + β i d t x , y x , y if x , y A i , T x , T y A i
(3.2)
or if x, y Ai+1, Tx, Ty Ai+1for any i p ̄ , where D: = dist(A i , Ai+1) being zero if A i A i + 1 ; i p ̄ . Fix y = Tx then, one can get from (3.2) for x A i :
1 - α i P i d s ( x , x ) T x , T 2 x β i d t x , x x , T x + 1 - γ i D β i Q i d s ( x , x ) x , T x + γ i D s ( x , x )
(3.3)
if Tx Ai+1, i p ̄ , and
d s ( x , x ) T x , T 2 x α i d r x , x T x , T 2 x + β i d t x , x x , T x α i P i d s ( x , x ) T x , T 2 x + β i Q i d s ( x , x ) x , T x
(3.4)
if x, Tx A i provided that the following upper-bounding conditions hold:
d r x , x x , T x P i d s ( x , x ) x , T x ; d t x , x x , T x Q i d s ( x , x ) x , T x
(3.5)
α i , β i , P i , Q i 0

Thus, the following technical result holds which does not require completeness of the metric space, uniform convexity assumption on some associated Banach space or particular properties of the non-empty subsets A i ; i p ̄ . The result will be then used to obtain the property of convergence of the sequences of iterates to best proximity points allocated in the various subsets.

Theorem 3.1: Let (X, d) a metric space and A i Ai+kp X; i p ̄ . Assume that T : i p ̄ A i i p ̄ A i is an extended (p ≥ 2) p-cyclic map, subject to the extended contractive condition (3.1), with T(A i ) A i Ai+1, T (A i ) Ai+1; j i - 1 ¯ and T j i A i A i + 1 for some finite integers j i ≥ 1 and q : = i = 1 p j i p ; i p ̄ . Define k i : = β i Q i 1 - α i P i j i , subject to k : = i = 1 p k i < 1 , and γ i = 1-k i ; i p ̄ . Assume also that σ(x, x) > 0, s(x, x) > 0, 0 < r(x, x) ≤ s(x, x)+ln(P-d(Tx, T2x)) if r(x, x) ≠ s(x, x) and 0 < t x , x s x , x + ln Q - d x , T x if t(x, x) ≠ s(x, x); x i p ̄ A i . Then, the following properties hold:
  1. (i)
    lim n d s ( x , x ) T n q + j i x , T n q x = D s ( x , x ) ; lim n d T n q + j i x , T n q x = D x A i , i p ̄ ;
    (3.6a)
    lim n d T n q + j i + j i + 1 x , T n q + j i x = D ; lim n d T n q + = i m j x , T n q + j i x = D
    (3.6b)
    lim n d T n q + = i m j x , T n q + = i m j x = D ; x A i i p ̄ ,
    (3.6c)
     
with im' < m < p+i, j p+i = j p ; i p ̄ , and similarly:
limsup n d s ( x , x ) T n q + x , T n q x β i Q i 1 - α i P i D s ( x , x ) ; x A i ; j i - 1 ¯ , i p ̄
(3.7)
limsup n d T n q + x , T n q x β i Q i 1 - α i P i s ( x , x ) D ; x A i ; j i - 1 ¯ , i p ̄
(3.8)
  1. (ii)
    lim n d s ( x , x ) T ( n + m ) q + j i x , T n q x = D s ( x , x ) ; lim n d T ( n + m ) q + j i x , T n q x = D ; x A i , m N , i p ̄
    (3.9)
    limsup n d s ( x , x ) T ( n + m ) q + x , T n q x β i Q i 1 - α i P i D s ( x , x ) ; x A i ; j i - 1 ¯ , m N , i p ̄
    (3.10)
    limsup n d T ( n + m ) q + x , T n q x β i Q i 1 - α i P i s ( x , x ) D ; x A i ; j i - 1 ¯ , m N , i p ̄
    (3.11)
     
Proof: The proof of Property (i) follows from the following inequalities which follow by recursion from (3.3) to (3.5):
d s ( x , x ) T + 1 x , T x β i Q i 1 - α i P i d s ( x , x ) x , T x
(3.12)
d s ( x , x ) T j i x , T j i - 1 x k i d s ( x , x ) x , T x + 1 - k i D s ( x , x )
(3.13)
d s ( x , x ) T q + 1 x , T q x k d s ( x , x ) x , T x + 1 - k D
(3.14)
x A i , T x A i , T j i x A i + 1 ; j i - 1 ¯ , i p ̄ since q : = i = 1 p j i p . One can get from (3.12) and (3.14) and, respectively, from (3.13) to (3.14):
d s ( x , x ) T q + x , T q x β i Q i 1 - α i P i d s ( x , x ) T q + 1 x , T q x β i Q i 1 - α i P i k d s ( x , x ) x , T x + 1 - k D
(3.15)
d s ( x , x ) ( T q + j i x , T q x ) k i d s ( x , x ) ( T q + 1 x , T q x ) + ( 1 k i ) D s ( x , x ) k i [ k d s ( x , x ) ( x , T x ) + ( 1 k ) D ] + ( 1 k i ) D s ( x , x )
(3.16)
Proceeding recursively with (3.14) for n N, one can get:
d s ( x , x ) T n q + x , T n q x β i Q i 1 - α i P i k n d s ( x , x ) x , T x + 1 - k n D s ( x , x ) ; x , T x A i
(3.17)
for j i - 1 ¯ , and
D s ( x , x ) d s ( x , x ) T n q + j i x , T n q x k i k n d s ( x , x ) x , T x + 1 - k n D s ( x , x ) + 1 - k i D s ( x , x )
(3.18)
; x Ai; T nq x A i ; T n q + j i x A i + 1 ; i p ̄ . One can get (3.6a) from (3.18) and (3.7)-(3.8) from (3.17), respectively, since k < 1 by taking limits as n → ∞. Equations 3.6b and 3.6c follow directly from (3.6a) as follows:
lim n sup d T n q + j i + j i + 1 x , T n q + j i x liminf n k i d T n q + j i x , T n q x + 1 - k i D = lim n k i d T n q + j i x , T n q x + 1 - k i D = D
(3.19)

with T n q + j i + j i + 1 x A i + 2 ; i p ̄ since T n q + j i x A i + 1 . Then, lim n d T n q + j i + j i + 1 x , T n q + j i x = D .

Proceeding recursively:
lim n sup d ( T n q + = i m j x , T n q + j i x ) lim inf n ( = i 1 m 1 [ k ] d ( T n q + j i x , T n q x ) + ( 1 = i 1 m 1 [ k ] i ) D ) = lim n ( k i d ( T n q + j i x , T n q x ) + ( 1 k i ) D ) = D
(3.20)
with i < m( N) < p+i, j p+i = j p , T n q + = i m j x A i + m + 1 ; i p ̄ so that lim n d T n q + = i m j x , T n q + j i x = D . In the same way, one can get:
lim n sup d T n q + = i m j x , T n q + = i m - m j x liminf n = m - m m - 1 k d T n q + = i m - m j x , T n q x + 1 - = m - m m - 1 k D = lim n k i d T n q + j i x , T n q x + 1 - k i D = D
(3.21)
with im'+i ≤ ( N) ≤ p+i, jp+i= j p , T n q + = i m - m 1 j A i + m - m 1 + 1 ; i p ̄ . Then, lim n d T n q + = i m j x , T n q + = i m - m j x = D . Property (i) has been proven. Now, note from (3.16), (3.15) and (3.18) that
D s ( x , x ) d s ( x , x ) ( T ( n + m ) q + x , T n q x ) ( β i Q i 1 α i P i ) d s ( x , x ) ( T ( n + m ) q x , T n q x ) k i [ k n + m d s ( x , x ) ( x , T x ) + ( 1 k n + m ) D s ( x , x ) ] + ( 1 k i ) D s ( x , x )
(3.22)
D s ( x , x ) d s ( x , x ) ( T ( n + m ) q + j i x , T ( n + m ) q x ) k i d s ( x , x ) ( T ( n + m ) q x , T n q x ) k i [ k n + m d s ( x , x ) ( x , T x ) + ( 1 k n + m ) D s ( x , x ) ] + ( 1 k i ) D s ( x , x )
(3.23)

; x A i , T x A i , T j i x A i + 1 ; j i - 1 ¯ , m N , i p ̄ . Hence, Property (ii). □

Remark 3.2. It is noted that if A i A = and x A j for some i , i p ̄ and j < j i then d T n + 1 q + x , T n q + x D as n → ∞ for all < j i . □

The following result is concerned with the proved property that distances of iterates obtained through the composed self-mapping T q : i p ̄ A i i p ̄ A i starting from a point x in any of the subsets, and located within two distinct of such subsets for all the iteration steps, asymptotically converge to the distance D between such subsets in uniformly convex Banach spaces, with at least one of them being convex. It is also obtained a convergence property of the iterates of the composed self-mapping T q : i p ̄ A i i p ̄ A i to limit points within each of the subsets.

Lemma 3.3: Let (X, || ||) be a uniformly convex Banach space endowed with the norm || || and let d: X × X →R0+ be a metric induced by such a norm || || so that (X, d) is a complete metric space. Assume that the non-empty subsets A i of X and the extended p(≥ 2) -cyclic self-mapping T : i p ̄ A i i p ̄ A i fulfil the constraints of Theorem 3.1 and, furthermore, one subset is closed and another one is convex and closed in each pair (A i , Ai+1) of adjacent subsets, i p ̄ . Then, the following properties hold:
  1. (i)
    lim n d T n + n q + j i x , T n q + j i x = lim n d T n + n q + = i m j x , T n q + = i m j x = 0
    (3.24a)
    lim n d T n + n q + j i x , T n q x = lim n d T n + n q + = i m j x , T n q + = i m j x = D ; x A i , i p ̄
    (3.24b)
     
; x A i , im' < m < p+i, jp+i= j p , m , m p ̄ , n N , i p ̄ . Furthermore, if Ai+1is convex, then T n q + j i x z i + 1 A i + 1 as n → ∞. Also, T n q + = 1 m j x z i + m + 1 = T = 1 m j z i A i + m + 1 as n → ∞ with zi+m+1zi+m+1-p, Ai+m+1Ai+m+1-pif m > p+1-i. Furthermore, if all the subsets A i i p ̄ are closed and convex, then T q n x z i p ̄ A i = T q z as n → ∞ if D = 0, that is if i p ̄ A i , so that z i p ̄ A i is the unique fixed point of T q : i p ̄ A i i p ̄ A i in i p ̄ A i .
  1. (ii)
    If A i or A i+m+1is convex then
    lim n d T n + n q + = i m j x , T n q + = i m j x = 0
    (3.25)
     

; x A i , i m < p + i , j p + i = j p , n N 0 : = N 0 , i p ̄

Proof: Note from (3.6a) that
d T n q + j i x , T n q x D d T n + n q + j i x , T n q x D d T n + n q + j i x , T n q + j i x 0 as n
(3.26a)
; x A i , jp+i= j p , n' N, i p ̄ with T nq x A i , T n q + j i x , T n + n q + j i x A i + 1 with Ai+1Ai+1-pif i > p-1, since (X, || ||) is a uniformly convex Banach space, d: X × X → R0+be a metric induced by the norm || ||, so that (X, d) is a complete metric space, and A i and Ai+1are non-empty closed subsets of X and at least one of them is convex (see Lemma 3.8 of [13]). Then lim n d T n + n q + j i x , T n q + j i x = 0 . On the other hand, lim n d T n + n q + = i m j x , T n q + = i m j x = 0 is proven by replacing (3.26a) by
d T n q + = i m j x , T n q x D d T n + n q + = i m j x , T n q x D d T n + n q + = i m j x , T n q + = i m j x 0  as  n
(3.26b)
with im < p+i, n' N. Thus, x A i implies T nq x A i , T n q + = i m j x , T n + n q + = i m j x A i + m + 1 . The identities (3.24a) have been proven. To prove (3.24b), note from Equation 3.24a of Property (i) and the triangle inequality that the following holds:
lim n d ( T ( n + n ) q + j i x , T n q x ) lim n d ( T n q + j i x , T n q x ) + lim n d ( T ( n + n ) q + j i x , T n q + j i x ) = lim n d ( T n q + j i x , T n q x ) = D
(3.27)
lim n d ( T ( n + n ) q + = i m j x , T n q + = i m j x ) lim n d ( T n q + = i m j x , T n q + = i m j x ) + lim n d ( T n q + = i m j x , T ( n + n ) q + = i m j x ) = lim n d ( T n q + = i m j x , T ( n + n ) q + = i m j x ) = D
(3.28)

; x A i , im' < m < p+i, jp+i= j p , m , m p ̄ , n N , i p ̄ . lim n d T n + n q + = i m j x , T n q + = i m j x = lim n d T n + n q + j i x , T n q x = D . On the other hand, note that T n q + j i x n N is a Cauchy sequence since lim n n d T n + n q + = i m j x , T n q + = i m j x = 0 from (3.24a) which then has a limit in X which is also in the closed and convex subset Ai+1of X. The proof of T n q + = i m j x z i + m + 1 A i + m + 1 with Ai+m+1A i , since m < p+i, as n → ∞ follows from similar arguments since one of the subsets in each adjacent pair of subsets is convex and both of them are closed by assumption so that T n q + = i m j x = T = i + 1 m j T n q + j i x z i + m + 1 = T = i + 1 m j z i + 1 as n → ∞; x A i , i p ̄ . Finally, if the subsets intersect and are closed and convex then the composed self-mapping T q : i p ̄ A i i p ̄ A i is contractive, then continuous everywhere in its definition domain, so that it converges to a unique fixed point in the non-empty, closed and convex set i p ̄ A i . Hence, Property (i).

To prove Property (ii), note from (3.6d) with m = i and m1 = 0 that
( [ d ( T ( n + n ) q + = i m j x , T n q x ) D ] [ d ( T n q + = i m j x , T n q x ) D ] ) ( d ( T ( n + m ) q + j i x , T n q + j i x ) 0 )  as  n
(3.29)

; x A i , im < p+i, j p+i = j p , n' N; i p ̄ with T nq x A i , T n q + = i m j x , T n + n q + = i m j x A i + m + 1 A i ; i p ̄ with Ai+m+1Ai+m+1-pif m > p-i-1; i p ̄ , since (X, || ||) is a uniformly convex Banach space (and then (X, d) is a complete metric space) and A i and Ai+m+1are non-empty closed subsets of X and A i or Ai+m+1is convex. Then, (3.25) follows in the same way as Property (i). □

The following result concerning to convergence of the iterates to closed finite sequences--eventually to unique fixed points if all the subsets intersect--is supported by Theorem 3.1 and Lemma 3.3.

Theorem 3.4: Let A i be non-empty closed and convex subsets of a uniformly convex Banach space (X, || ||); i p ̄ . Assume that T : i p ̄ A i i p ̄ A i is an extended (p ≥ 2) -cyclic map, subject to the extended contractive condition (3.1), with T(A i ) A i Ai+1, T (A i ) Ai+1; j i - 1 ¯ and T j i A i A i + 1 for some finite integers j i ≥ 1, i p ̄ and q : = i = 1 p j i p . Then, the following properties hold:
  1. (i)
    T qn xz i A i , x A i as n → ∞ and there is a q-tuple:
    z ^ i : = ( T z i = T q + 1 z i , ... , ω i + 1 = T j i z i , T j i + 1 z i , ... , ω i + 2 = T j i + j i + 1 z i , T j i + j i + 1 + 1 z i , ... , ω i + p = ω i = z i = T q z i )
    (3.30)
     
; i p ̄ which is the unique limit sequence of limit points of any q-tuple of sequences:
x ^ q n : = T q n + 1 x , . . . , T q n + j i x , T q n + j i + 1 x , . . . , T q n + j i + j i + 1 x , T q n + j i + j i + 1 + 1 x , . . . , T q + 1 n x
(3.31)
; x A i , where T k z i A i ; k j i - 1 ¯ 0 ; i p ̄ , ωi+ℓ Ai+is the unique best proximity point in Ai+; p ̄ such that D = dist(A i , Ai+1) = d(ω i , ωi+1); i p ̄ .
  1. (ii)

    Assume that i p ̄ A i . Then, the self-mapping T : i p ̄ A i i p ̄ A i has a unique fixed point z i p ̄ A i . Then, any q-tuple of sequences (3.31) converges to a unique limit q-tuple (3.30) of the form : = z , . . . . , z for any x i p ̄ A i and for any i p ̄ .

     
Proof: To keep a coherent treatment with the previous part of the manuscript and, since (X, || ||) is a Banach space with norm || ||, we can use a norm-induced metric d: X × XR0+ which is equivalent to any other metric and then apply Theorem 3.1 to the metric space (X, d) which is complete since (X, || ||) is a Banach space. Assume the following cases:
  1. (A)

    D = 0 so that A i A j for i , j i p ̄ ; i.e. all the subsets have a non-empty intersection. Then, d T n q + j i x , T n q x 0 , d(T (n+1)q x, T nq x) → 0 and d(T nq+ℓ x, T nq x) → 0; x A i ; k = i m + i - 1 j k : m p ¯ , i p ̄ as n → ∞ from Theorem 3.1, Equations 3.6 and 3.8, with A j A j-pfor 2pj > p. Thus, T qn x( A i ) → z, since {T qn x}n Nis a Cauchy sequence (and also T qn+ℓ x( A i+ℓ) → z) for some z i p ̄ A i , since i p ̄ A i is non-empty, convex and closed from Banach contraction principle since k < 1. Since k < 1, the composed self-mapping T q : i p ̄ A i i p ̄ A i is contractive, and then continuous, and since (X, d) is complete, since the associated (X, || ||) is a Banach space, z i p ̄ A i is a unique fixed point of T q : i p ̄ A i i p ̄ A i . Thus, again the continuity of T q : i p ̄ A i i p ̄ A i and the fact that it has a unique fixed point z leads to the identities T q (Tz) = T q+1 z = T(T q z) = Tz = T q (T q z) = T q z so that z = Tz and then z is also a fixed point of T : i p ̄ A i i p ̄ A i . Furthermore, z is also the unique fixed point of T : i p ̄ A i i p ̄ A i as follows by contradiction. Assume that z is not unique. Then, such that z is the unique fixed point of T q : i p ̄ A i i p ̄ A i and y z i p ̄ A i such that y and z are both fixed points of T : i p ̄ A i i p ̄ A i . Then, T q y = T(T q y) = Ty = y which contradicts that z is the unique fixed point of T q : i p ̄ A i i p ̄ A i . Finally, as a result of the uniqueness of the fixed point, it follows directly that any q-tuple (3.30) converges to a unique q-tuple : = z , . . . . , z = i ; i p ̄ for any x i p ̄ A i . Hence, Property (ii).

     
  2. (B)
    D ≠ 0 so that A i A j for i , j i p ̄ . One can get from (3.6) to (3.8):
    lim n d T n q + j i x i , T n q x i = D , limsup n d T n q + x i , T n q x i β i Q i 1 - α i P i s ( x , x ) D ,
    (3.32)
     
; x i A i ; j i - 1 ¯ , i p ̄ . Thus:
lim n d T n + 1 q + j i x i , T n q x i = D , limsup n d T n + 1 q + x i , T n q x i β i Q i 1 - α i P i s ( x , x ) D
(3.33)
; x i A i , j i - 1 ¯ , i p ̄ . One has from Lemma 3.3, Equation 3.24b and Proposition 3.2 of [14]:
lim m k > n k d T m k q + = i m j x i , T n k q + = i m j x i = d ω i + m + 1 , ω i + m + 1 = D
(3.34)
, that is, the distance between the subsets Ai+m+1(≡ Ai+m+1-pif m > p+1-i) and Ai+m'+1(≡ Ai+m'+1-pif m > m' > p+1-i) of X equalizes that of two corresponding best proximity points, for some convergent subsequences T m k q + = i m j x i A i + m + 1 and T n k q + = i m j x i A i + m + 1 and two best proximity points: ωi+m+1 Ai+m+1; ωi+m'+1 Ai+m'+1. Then, again from Lemma 3.3 and (3.34), one can get
lim n d ( T ( n + n ) q + = i m j x i , T n q + = i m j x i ) = lim n d ( T ( n + n ) q + j i x i , T n q x i ) = d ( ω i + m + 1 , ω i + m + 1 ) = D
(3.35)

Also, one has from Lemma 3.3, that lim n d T n q + = i m j x i , T n + 1 q + = i m j x i = 0 so that, by taking into account (3.35), T n q + = i m j x i ω i + m + 1 A i + m + 1 A i + m + 1 - p if m > p + 1 - i as n → ∞; x A i , i , m i p ̄ with j = jℓ-pfor any ℓ > p, n' N0, since T n q + = i m j x i A i + m + 1 . That is, T n q + = i m j x i converges to a best proximity point of Ai+m+1; x i A i ; i p ̄ .

Now, take a sequence T n q + = i m j + j x i where j j m + 1 - 1 ¯ . Then, T n q + = i m j + j x i A i + m + 1 ; j j m + 1 - 1 ¯ . Assume that T n q + = i m j + j x i n N 0 does not converge in Ai+m+1so that one can get from Theorem 3.1, Equation 3.8 and Lemma 3.3 (ii):
0 < lim inf n d ( T ( n + 1 ) q + = i m j + j x i , T n q + = i m j x i ) ( β i Q i 1 α i P i ) j / s ( x , x ) lim n d ( T ( n + 1 ) q + = i m j x i , T n q + = i m j x i ) = 0 ; x i A i ; j i 1 ¯ , j j m + 1 1 ¯ , i p ¯
(3.36)
which is a contradiction. Then, lim n d T n + 1 q + = i m j + j x i , T n q + = i m j x i = 0 and T n q + = i m j + j x i n N 0 is a Cauchy sequence with a limit in the closed and convex Ai+m+1 X for any j j m + 1 - 1 ¯ , x i A i and i p ̄ . It has been proven that x ^ q n i ; i p ̄ and the set ω i : i p ̄ is a set of best proximity points with ω i + 1 = T j i ω i = T j i + q ω i ; i p ̄ and ωp+1= ω1. It remains to prove that the elements of the limit sequence i are not dependent on the initial point x i A i ; i , m i p ̄ to construct any sequence of iterates except, perhaps, in the order that the limiting points are allocated within such a limiting sequence. Proceed by contradiction by assuming that there are two distinct best proximity points ω i , z i A i for some i p ̄ so that:
D = d T j i ω i , ω i < d T j i ω i , z i = d T j i + n q ω i , T n q z i
(3.37)
since ω i , z i (≠ ω i ) A i are best proximity points and T j i ω i A i + 1 . Since that the above property holds irrespective of the integers i , m i p ̄ and n N0, the following contradiction follows from (3.37), and Theorem 3.1, Equation 3.6a:
D = d ( T j i ω i , ω i ) = d ( T j i z i , z i ) < d ( T j i ω i , z i ) = d ( ω i , T j i z i ) = lim n d ( T j i + n q ω i , T n q z i ) = D
(3.38)

irrespective of i p ̄ . Therefore, the best proximity points are unique within each of the subsets. Furthermore, since T : i p ̄ A i i p ̄ A i , the limit sequence (3.30) is unique by successive iterations from any of the best proximity points. Since there is a convergence to it from any initial point in i p ̄ A i , any q-sequence of iterates converges to such a limit sequence, irrespective of the initial point, except for the order of the elements. □

Remark 3.5. Concerning with Theorem 3.4 (ii), note that T : i p ̄ A i i p ̄ A i is not necessarily contractive when D = 0 although T q : i p ̄ A i i p ̄ A i be contractive. However, the contractive property of T q : i p ̄ A i i p ̄ A i leads to that of T : i p ̄ A i i p ̄ A i possessing also a unique fixed point in i p ̄ A i which is that of T q : i p ̄ A i i p ̄ A i . It is also noted that the limit sequence is unique except in the order of the elements in the sense that if a sequence of iterates x ^ q n Equation 3.31 converges to i Equation 3.30 for any initial point x A i , then for x A k the limit sequence being asymptotically reached will be:
z ^ k : = ( T z k = T q + 1 z k , ... , ω j + 1 = T j k z k , T j k + 1 z k , ... , ω k + 2 = T j k + j k + 1 z i , T j k + j k + 1 + 1 z k , ... , ω k + p = ω k = z k = T q z k )
(3.39)

which is identical to (3.30) except in the order of its elements. □

An example is given below

Example 3.6: Take p = 2 and subsets A1C(a, 0, a-a0) and A2C(-a, 0, a-a0) of R2 are circles of centre in (a, 0) and (-a, 0), respectively, and radius a-a0 with a > a0 which are defined by
A 1 : = { ( x , y ) R 2 : ( x a ) 2 + y 2 ( a a 0 ) 2 } ; A 2 : = { ( x , y ) R 2 : ( x + a ) 2 + y 2 ( a a 0 ) 2 }
(3.40)
We consider the complete metric space (R2, d) with the Euclidean metric. It is clear that such a space being considered as the Banach space (R2, || ||), endowed with the Euclidean norm, is uniformly convex, then strictly convex and reflexive [23]. It is noted that D = 2a0 = dist(A1, A2) and now consider the constraint (3.1) with functions s, x, r, t: (A1 A2) × (A1 A2) → R+ being constant identically unity. Assume that T is some extended 2-cyclic self-mapping on A1 A2 with j1 = 2, j2 = 1. Thus, one can get
g = x , y A 1 T g A 1 , T 2 g A 2 , T 3 g A 1 , T 4 g A 1 , T 5 g A 2 ,
(3.41a)
g = x , y A 2 T g A 1 , T 2 g A 1 , T 3 g A 2 , T 4 g A 1 , T 5 g A 1 ,
(3.41b)
Now, define also the family of parameterized circles A ^ a 1 : = x , y R 2 : x - a 2 + y 2 a 1 2 A 1 being with A1 and contained in it (with proper or improper set inclusion) of radius 0 < a1a-a' for a' [0, a). It is noted that A ^ a A 1 . Next, we define constructively a self-mapping which is an extended cyclic one and which verifies the properties. Now, consider a self-mapping T: A1 A2A1 A2 defined as follows:
g 0 = x 0 , y 0 A 1 ; g 1 = x 1 , y 1 = T g 0 = k ^ 1 x 0 , y 0 x 0 , k ^ 1 x 0 , y 0 y 0 F r A ^ a 1 A 1 = x , y : k ^ 1 x 0 - a 2 + k ^ 1 2 y 0 2 = a 1 2 a - a 2
(3.42)
The positive solution in k ^ 1 of the equality defining a circle A ^ a 1 for a fixed a1 = a1(x0, y0) ≤ a-a' is defined below together with available point-dependent lower and upper-bounds:
a 2 a + a 0 a x 0 x 0 2 + y 0 2 k ^ 1 = k ^ 1 x 0 , y 0 = 2 a x 0 + 4 a 2 x 0 2 - 4 a 2 - a 1 2 x 0 2 + y 0 2 2 x 0 2 + y 0 2 2 a x 0 x 0 2 + y 0 2 a x 0 k ̄ 1 : = a a 0
(3.43)
It is noted that such a positive solution always exists everywhere in A1 since a a 1 a 1 - x 0 2 x 0 2 + y 0 2 . Thus, the constraint (3.1) is fulfilled by any self-mapping T: A1 A2A1 A2 with 1 > α1 ≥ 0, β 1 = a a 0 1 - α 1 > 1 - α 1 and k 1 : = k ̄ 1 2 = a a 0 2 = β 1 1 - α 1 2 > 1 . It is noted that the condition (3.1) is not guaranteed to be contractive for any point of A1. It is also noted that if (x0, y0) Fr(A1) then k ^ 1 x 0 , y 0 = 1 with a0 = 0 so that g1 = Tg0 = g0. However, it can be noticed that g1 is not a fixed point since g2 = Tg1g1. Next, define:
g 2 = x 2 , y 2 = T g 1 = T 2 g 0 = min - k ^ 1 x 1 , - k ^ 1 a 0 , - k ^ 1 y 1 A 2 x 1 = a 0 , y 1 = 0 k ^ 1 = 1 , x 2 = - a 0 , y 2 = 0 d g 1 , g 2 = D
(3.44)
g 3 = x 3 , y 3 = T g 2 = T 2 g 1 = T 3 g 0 = max - k 2 x 2 , - k 2 a 0 , - k 2 y 2 A 1 x 2 = - a 0 , y 2 = 0 x 3 = a 0 , y 3 = 0 d g 3 , g 2 = D
(3.45)
under the contractive constant k2 defined as follows subject to constraints:
k 2 : = β 2 1 - α 2 < k 1 - 1 = a 0 a 2 = 1 - α 1 β 1 2
(3.46)

so that the composed extended cyclic self-mapping T3: A1 A2A1 A2 is subject to a the contractive condition (3.1) of contractive constant k = k1k2 < 1 with s, x, r, t: (A1 A2) × (A1 A2) → R+, i.e. for the Euclidean distance. Consider initial points in A2 as g0 = (x0, y0) A2. Thus, (a) first apply (3.45) with the replacements (x3, y3) → (x1, y1) A2 and g3g1(x1, y1) A1; (b) then apply (3.42) with the replacement (x3, y3) → (x1, y1) and g1 →g2 = (x2, y2) A1; (c) later on apply (3.44) with the replacement (x2, y2) → (x3, y3) and g2g3(x3, y3) A2. Theorems 3.1 and 3.4 are fulfilled and there is a limiting repeated sequence of three points S*: = ((-a0, 0)), (a0, 0), (a0, 0)) which are two best proximity points, one located in A2 and another one in A1, repeated. It is noted that the repeated value of the best proximity point (a0, 0) in the limiting sequence is due to the fact that the circumference being the boundary of A1 is invariant under T, although infinitely many different cyclic self-mappings can be defined on the same two subsets A1, 2 so that a limiting sequence is reached having a common (perhaps only) the best proximity points with S*. It is also noted that if a0 ≠ 0 then (a0, 0) is not a fixed point of T: A1 A2A1 A2 since T(a0, 0) = (a0, 0) T2(a0, 0) = (-a0, 0). However, if a0 = D = 0 then the intersection of both circles is {(0, 0)} so that the limiting sequence consists of a repeated fixed point.

Declarations

Acknowledgements

The authors wish to thank the Spanish Ministry of Education for their support through Grant DPI2009-07197. They are also grateful to the Basque Government for their support through Grants IT378-10, and SAIOTEK SPE09UN12. The authors thank the referees for their useful suggestions.

Authors’ Affiliations

(1)
Instituto de Investigacion y Desarrollo de Procesos, Universidad del Pais Vasco, Campus of Leioa (Bizkaia)
(2)
Department of Mathematics, Texas A&M University- Kingsville

References

  1. Harjani J, Lopez B, Sadarangani K: A fixed point theorem for mappings satisfying a contractive condition of rational type of partially ordered metric space. Abstr Appl Anal 2010, 2010: 8. (Article Number 190701)MathSciNetView ArticleGoogle Scholar
  2. Harjani J, Sadarangani K: Fixed point theorems for weakly contractive mappings in partly ordered sets. Nonlinear Anal Theory Methods Appl 2009,71(7–8):3403–3410. 10.1016/j.na.2009.01.240MathSciNetView ArticleGoogle Scholar
  3. Bhardwaj R, Rajput SS, Yadava RN: Application of fixed point theory in metric spaces. Thai J Math 2007,5(2):253–259.MathSciNetGoogle Scholar
  4. Enjouji Y, Nakanishi M, Suzuki T: A generalization of Kannan's fixed point theorem. Fixed Point Theory Appl 2009, 2009: 10. (Article Number 192872)MathSciNetView ArticleGoogle Scholar
  5. Banach S: Sur les operations dans les ensembles abstracts et leur application aux équations intégrales. Fundamenta Mathematicae 1922, 3: 133181.Google Scholar
  6. Chatterjee SK: Fixed point theorems. Comptes Ren Acad Bulgaria Sci 1972, 25: 727730.Google Scholar
  7. Fisher B: A fixed point theorem for compact metric spaces. Publ Math Debrecen 1978, 25: 193194.Google Scholar
  8. Kannan R: Some results on fixed points. Bull Calcutta Math Soc 1968, 60: 7176.Google Scholar
  9. Kannan R: Some results on fixed points-II. Am Math Monthly 1969, 76: 405408.View ArticleGoogle Scholar
  10. Reich S: Some remarks concerning contraction mappings. Can Math Bull 1971, 14: 121124.View ArticleGoogle Scholar
  11. Kikkawa M, Suzuki T: Some similarity between contractions and Kannan mappings. Fixed Point Theory Appl 2008, 2008: 8. (Article ID 649749)MathSciNetView ArticleGoogle Scholar
  12. Subrahmanyam PV: Completeness and fixed points. Monatshefte für Mathematik 1975,80(4):325–330. 10.1007/BF01472580MathSciNetView ArticleGoogle Scholar
  13. Kirk WA, Srinivasan PS, Veeramani P: Fixed points for mappings satisfying cyclical contractive conditions. Fixed Point Theory 2003,4(1):79–89.MathSciNetGoogle Scholar
  14. Eldred AA, Veeramani P: Existence and convergence of best proximity points. J Math Anal Appl 2006,323(2):1001–1006. 10.1016/j.jmaa.2005.10.081MathSciNetView ArticleGoogle Scholar
  15. Karpagam S, Agrawal S: Best proximity point theorems for p-cyclic Meir-Keeler contractions. Fixed Point Theory Appl 2009, 2009: 9. (Article Number 197308)MathSciNetView ArticleGoogle Scholar
  16. Di Bari S, Suzuki T, Vetro C: Best proximity points for cyclic Meir-Keeler contractions. Nonlinear Anal.: Theory Methods Appl 2008,69(11):3790–3794. 10.1016/j.na.2007.10.014MathSciNetView ArticleGoogle Scholar
  17. De la Sen M: Linking contractive self-mappings and cyclic Meir-Keeler contractions with Kannan self-mappings. Fixed Point Theory Appl 2010, 2010: 23. (Article Number 572057)Google Scholar
  18. De la Sen M: Some combined relations between contractive mappings, Kannan mappings reasonable expansive mappings and T-stability. Fixed Point Theory Appl 2009, 2009: 25. (Article Number 815637)Google Scholar
  19. Kikkawa M, Suzuki T: Three fixed point theorems for generalized contractions with constants in complete metric spaces. Nonlinear Anal: Theory Methods Appl 2008,69(9):2942–2949. 10.1016/j.na.2007.08.064MathSciNetView ArticleGoogle Scholar
  20. Suzuki T: Some notes on Meir-Keeler contractions and L-functions. Bull Kyushu Inst Technol 2006, 53: 1213.Google Scholar
  21. Derafshpour M, Rezapour S, Shahzad N: On the existence of best proximity points of cyclic contractions. Adv Dyn Syst Appl 2011,6(1):33–40.MathSciNetGoogle Scholar
  22. Rezapour Sh, Derafshpour M, Shahzad N: Best proximity points of cyclic φ -contractions on reflexive Banach spaces. Fixed Point Theory Appl 2010, 2010: 7. (Article ID 946178)MathSciNetView ArticleGoogle Scholar
  23. Qin X, Kang SM, Agarwal RP: On the convergence of an implicit iterative process for generalized asymptotically quasi non-expansive mappings. Fixed Point Theory Appl 2010, 2010: 19. (Article Number 714860)MathSciNetGoogle Scholar
  24. Rus IA: Cyclic representations and fixed points. Ann T Popoviciu Seminar Funct Eq Approx Convexity 2005, 3: 171178.Google Scholar
  25. Pacurar M, Rus IA: Fixed point theory for cyclic φ -contraction. Nonlinear Anal: Theory Methods Appl 2010,72(3–4):1181–1187. 10.1016/j.na.2009.08.002MathSciNetView ArticleGoogle Scholar
  26. Karapinar E: Fixed point theory for cyclic weak φ -contraction. Appl Math Lett 2011, 24: 822825.Google Scholar
  27. Shazhad N, Sadiq Basha S, Jeyaraj R: Common best proximity points: global optimal solutions. J Optim Theory Appl 2011,148(1):69–78. 10.1007/s10957-010-9745-7MathSciNetView ArticleGoogle Scholar
  28. Al-Thagafi MA, Shahzad N: Convergence and existence results for best proximity points. Nonlinear Anal: Theory Methods Appl 2009,70(10):3665–3671. 10.1016/j.na.2008.07.022MathSciNetView ArticleGoogle Scholar
  29. Vetro C: Best proximity points: convergence and existence theorems for p-cyclic mappings. Nonlinear Anal: Theory Methods Appl 2010,73(7):2283–2291. 10.1016/j.na.2010.06.008MathSciNetView ArticleGoogle Scholar

Copyright

© De la Sen and Agarwal; licensee Springer. 2011

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.