We collect, improve, and generalize very recent results due to Mongkolkeha et al. (2014) in three directions: firstly, we study g-best proximity points; secondly, we employ more general test functions than can be found in that paper, which lets us prove best proximity results using different kinds of control functions; thirdly, we introduce and handle a weak version of the P-property. Our results can also be applied to the study of coincidence points between two mappings as a particular case. As a consequence, the contractive condition we introduce is more general than was used in the mentioned paper.
1. Introduction
Fixed point theory is a branch of nonlinear analysis which has attracted much attention in recent times due to its possible applications. After the appearance of the pioneering Banach contractive mapping principle in 1922, many mathematicians have intensively investigated sufficient conditions to ensure that certain contractive mappings have a fixed point. Some of the most well-known generalizations are due to Zabreĭko and Krasnoselĭ [1], Edelstein [2], Browder [3], and Caristi [4].
When a mapping from a metric space into itself has no fixed points, it could be interesting to study the existence and uniqueness of some points that minimize the distance between an origin and its corresponding image. These points are known as best proximity points and they were introduced by Fan [5] and modified by Sadiq Basha in [6]. The study of this kind of points and their properties has become one of the newest branches of fixed point theory, and many interesting results, generalizing the notion of fixed point, have been presented. In fact, many theorems in fixed point theory have been very useful so as to introduce their corresponding extensions to this new field of study (see also [7–13] and references therein).
On the other hand, in the past years, fixed point theorems in partially ordered metric spaces have also attracted much attention, especially after the works of Ran and Reurings [14], Nieto and Rodríguez-López [15], Gnana Bhaskar and Lakshmikantham [16], Berinde and Borcut [17, 18], Karapınar and Berinde [19, 20], Berzig and Samet [21], and Roldán et al. [22–24], among others. Their results were extended to more general contractivity conditions in which altering distance functions play a key role. Very recently, Alghamdi and Karapınar [25] used a similar notion in G-metric spaces, and Berzig and Karapınar [26] also considered a more general kind of contractivity conditions using a pair of generalized altering distance functions.
In order to consider a contractive condition on the whole metric space that can be particularized to partially ordered metric spaces, some advances have been done in recent times (see, for instance, [25–27] and references therein). This subject has been extended by Mongkolkeha et al. [28] to the field of determining best proximity points, describing a wide class of contractive mappings and using very general control functions. The main aim of this paper is to collect, generalize, and improve their results using contractive conditions and control functions that can be particularized in a wide kind of different results applicable to several frameworks.
2. Preliminaries
Let ℕ={0,1,2,…} denote the set of all nonnegative integers. Throughout this paper, let (X,d) be a metric space, let A and B two nonempty subsets of X, and let T:A→B, g:A→A, and α:X×X→[0,∞) be three mappings. Define
(1)ΔAB=dist(A,B)=inf({d(a,b):a∈A,b∈B}),A0={a∈A:∃b∈Bsuchthatd(a,b)=ΔAB},B0={b∈B:∃a∈Asuchthatd(a,b)=ΔAB}.
Notice that, if a∈A and b∈B verify d(a,b)=ΔAB, then a∈A0 and b∈B0. Therefore, A0 is nonempty if, and only if, B0 is nonempty. Thus, if A0 is nonempty, then A, B, and B0 are nonempty subsets of X. It is clear that, if A∩B≠∅, then A0 is nonempty. In [29], the authors discussed sufficient conditions in order to guarantee the nonemptiness of A0. In general, if A and B are closed subsets of a normed linear space such that ΔAB>0, then A0 is contained in the boundary of A (see [6]).
The main aim of this paper is to study sufficient conditions to ensure the existence and, in some cases, the unicity of the following kind of points.
Definition 1.
One will say a point x∈A is a g-best proximity point of T if d(gx,Tx)=ΔAB and x is a best proximity point of T if d(x,Tx)=ΔAB.
If A=B, a g-best proximity point of T is called a coincidence point ofT andg (i.e., Tx=gx), and if g is the identity mapping on A, then x is a fixed point ofT (i.e., Tx=x).
We describe the families of functions that we will use henceforth.
Definition 2.
(i) One will denote by Ψ the family of all functions φ:[0,∞)→[0,∞) such that, for all t>0, the series ∑n≥1φn(t) converges (functions in Ψ are called (c)-comparison functions).
(ii) One will denote by Φ the family of all functions ϕ:[0,∞)→[0,∞) such that ϕ(t)<t and limr→t+ϕ(r)<t for all t>0.
(iii) One will denote by Θ the family of all continuous mappings θ:[0,∞)4→[0,∞) such that θ(a,b,c,d)=0 if one or more arguments take the value zero (i.e., if abcd=0).
(iv) One will denote by Ω the family of all mappings θ:[0,∞)4→[0,∞) such that θ(a,b,c,d)=0 if one or more arguments take the value zero (i.e., if abcd=0).
(v) One will denote by Ω′ the family of all mappings θ:[0,∞)4→[0,∞) such that θ(0,b,c,d)=0.
(vi) One will denote by Ω′′ the family of all mappings θ:[0,∞)4→[0,∞) such that limn→∞θ(tn1,tn2,tn3,tn4)=0 whatever the sequences {tn1},{tn2},{tn3},{tn4}⊂[0,∞) such that, at least one of them, is convergent to zero (i.e., there exists i∈{1,2,3,4} verifying {tni}→0).
Remark 3.
(1) It is easy to see that, if φ∈Ψ, then φ(t)<t for all t>0.
(2) We point out that we do not impose any monotone condition on the control function we will use.
(3) Clearly Θ⊂Ω⊂Ω′ and Θ⊂Ω⊂Ω′′. Notice that functions in Ω, Ω′ and Ω′′ have not to be continuous.
Example 4.
Examples of functions in Θ are the following ones (where λ>0):
(2)θ1(t1,t2,t3,t4)=λt1β1t2β2t3β3t4β4,whereβ1,β2,β3,β4>0;θ2(t1,t2,t3,t4)=λln(1+t1t2t3t4)β,whereβ>0;θ3(t1,t2,t3,t4)=λmin(t1,t2,t3,t4).
The mappings of Φ have been very useful in the framework of fixed point theory (see [30–32]). The following lemma can be found in the literature but we recall it here for the sake of completeness.
Lemma 5.
Let ϕ∈Φ be a mapping and let {am}⊂ℝ0+ be a sequence. If am+1≤ϕ(am) and am≠0 for all m, then {am}→0.
In the following result, 𝒫4 denotes the family of all permutations σ:{1,2,3,4}→{1,2,3,4}.
Lemma 6.
Given λ>0 and θ∈Θ, define θλ′:[0,∞)4→[0,∞) by
(3)θλ′(t1,t2,t3,t4)=λmax(θ(tσ(1),tσ(2),tσ(3),tσ(4)):σ∈𝒫4)∀t1,t2,t3,t4∈[0,∞).
Then θλ′∈Θ and θλ′ is symmetric. Furthermore, if λ≥1, then θ≤θλ′.
Definition 7.
If ℛ is a binary relation on X, one will consider the mapping αℛ:X×X→[0,∞) given, for all x,y∈X, by
(4)αℛ(x,y)={1,ifxℛy,0,otherwise.
Definition 8.
A preorder (or a quasiorder) ≼ on X is a binary relation on X that is reflexive (i.e., x≼x for all x∈X) and transitive (if x,y,z∈X verify x≼y and y≼z, then x≼z). In such a case, we say that (X, ≼) is a preordered space (or a preordered set). If a preorder ≼ is also antisymmetric (x≼y and y≼x implies x=y), then ≼ is called a partial order.
Definition 9 (Raj [33]).
Let A and B be two subsets of a metric space (X,d) such that A0 is nonempty. We say that the pair (A,B) has the P-property if
(5)a1,a2∈A0,b2,b2∈B0d(a1,b1)=ΔABd(a2,b2)=ΔAB}⟹d(a1,a2)=d(b1,b2).
In [28], the authors introduced the following find of contractive mappings and succeed in proving the following result.
Definition 10 (Mongkolkeha et al. [28], Definition 3.1).
Let A and B be nonempty subsets of a metric space (X,d). A mapping T:A→B is said to be a generalized almost(φ,θ)α contraction if
(6)α(x,y)d(Tx,Ty)≤φ(M(x,y))+θ(d(y,Tx)-ΔAB,d(x,Ty)-ΔAB,d(x,Tx)-ΔAB,d(y,Ty)-AB),
for all x,y∈A, where α:A×A→[0,∞), φ∈Ψ, θ∈Θ, and
(7)M(x,y)=max(d(x,y),d(x,Tx)-ΔAB,d(x,Ty)+d(y,Tx)2d(y,Ty)-ΔAB,d(x,Ty)+d(y,Tx)2-ΔAB).
Theorem 11 (Mongkolkeha et al. [28], Theorem 3.2).
Let A and B be nonempty closed subsets of a complete metric space X such that A0 is nonempty and the pair (A,B) has the P-property. Let T:A→B satisfy the following conditions:
Tis an α-proximal admissible and generalized almost (φ,θ)α-contraction;
T is continuous;
there exist elements x0,x1∈A0 such that d(x1,Tx0)=ΔAB and α(x0,x1)≥1;
T(A0)⊆B0.
Then there exists an element x∈A such that
(8)d(x,Tx)=ΔAB.
Moreover, for any fixed x0∈A0, the sequence {xn}, defined by
(9)d(xn+1,Txn)=ΔAB,
converges to the element x.
3. Existence of g-Best Proximity Points under Different Conditions
The main aim of this paper is to study the following kind of mappings and to ensure that, under some conditions, they have a g-best proximity point.
Definition 12.
Let T:A→B, g:A→A, φ:[0,∞)→[0,∞), θ:[0,∞)4→[0,∞), and α:X×X→[0,∞) be five mappings. One will say that T is a (φ,θ,α,g)-contraction if, for all x,y∈A0 such that d(gy,Tx)=ΔAB and α(gx,gy)≥1, we have that
(10)α(gx,gy)d(Tx,Ty)≤φ(Mg(x,y))+θ(d(gy,Tx)-ΔAB,d(gx,Ty)-ΔAB,d(gx,Tx)-ΔAB,d(gy,Ty)-ΔAB),
where
(11)Mg(x,y)=max(d(gx,gy),d(gx,Tx)-ΔAB,d(gx,Ty)+d(gy,Tx)2d(gy,Ty)-ΔAB,d(gx,Ty)+d(gy,Tx)2-ΔAB).
In the previous definition, we have not supposed that φ∈Ψ or θ∈Θ because the main aim of the present paper is to introduce sufficient conditions on the involved mappings (φ, θ, α, and g) and on the ambient space to ensure the existence and, in some cases, the unicity of g-best proximity points of T.
Remark 13.
(1) Some other authors used to impose that their contractive condition must be verified for all x,y∈A. However, our condition (10) must only be satisfied for all x,y∈A0. Later, we will discuss when it is necessary to assume that this property holds for all x,y∈A.
(2) The mapping θ need not be symmetric. However, if θ∈Θ and T is a (φ,θ,α,g)-contraction, then T is also a (φ,θ1′,α,g)-contraction, where θ1′ is defined as in Lemma 6. In such a case, when θ∈Θ, without loss of generality, we can consider that θ is symmetric; that is, in this case, the order of the arguments of θ in (10) is not important.
The following definitions are very useful in order to establish weaker conditions than the P-property (see also [34]) or the notion of α-proximal admissible mapping.
Definition 14.
Let A and B be two subsets of a metric space (X,d) such that A0 is nonempty, and let T:A→B and g:A→A be two mappings. One will say that the quadruple (A,B,T,g) has the following:
the weakP-property of the first kind if
(12)a1,a2,a3,a4∈A0d(ga1,Ta3)=ΔABd(ga2,Ta4)=ΔAB}⟹d(ga1,ga2)≤d(Ta3,Ta4);
the weakP-property of the second kind if
(13)a1,a2,a3,a4∈A0d(ga1,Ta3)=ΔABd(ga2,Ta4)=ΔAB}⟹d(ga1,ga2)=d(Ta3,Ta4);
the weakP-property of the third kind if
(14)a1,a2∈A,b1,b2∈Bd(ga1,b1)=ΔABd(ga2,b2)=ΔAB}⟹d(ga1,ga2)≤d(b1,b2).
Lemma 15.
If the pair (A,B) has the P-property, then the quadruple (A,B,T,g) has the weak P-property of the first, the second, and the third kind, whatever the mappings T:A→B and g:A→A.
Remark 16.
Obviously, if (X,d) is a metric space, then the pair (X,X) has the P-property. Therefore, the quadruple (X,X,T,g) has the weak P-property of the first, the second, and the third kinds whatever the mappings T:A→B and g:A→A.
Definition 17.
Let A and B be two subsets of a metric space (X,d) such that A0 is nonempty, and let T:A→B, g:A→A, and α:X×X→[0,∞) be three mappings. One will say that T is (α,g)-proximal admissible if
(15)a1,a2,b1,b2∈A0α(gb1,gb2)≥1d(ga1,Tb1)=ΔABd(ga2,Tb2)=ΔAB}⟹α(ga1,ga2)≥1.
Lemma 18.
If T is α-proximal admissible, then T is (α,g)-proximal admissible, whatever g:A→A.
Definition 19.
Let g:A→A and α:X×X→[0,∞) be two mappings and let N∈ℕ and N≥2. We will say that α is (N,g)-transitive onA0 if
(16)x1,x2,…,xN+1∈A0α(gxi,gxi+1)≥1,∀i∈{1,2,…,N}⇓α(gx1,gxN+1)≥1.
Indeed, one will only use the notion of (2,g)-transitive mapping onA0; that is,
(17)x1,x2,x3∈A0α(gx1,gx2)≥1α(gx2,gx3)≥1}⟹α(gx1,gx3)≥1.
Next we prove our first main result.
Theorem 20.
Let A and B be two closed subsets of a complete metric space (X,d) and let T:A→B, g:A→A, φ:[0,∞)→[0,∞), θ:[0,∞)4→[0,∞), and α:X×X→[0,∞) be five mappings. Assume that the following conditions hold:
∅≠A0⊆gA0 and T(A0)⊆B0;
the quadruple (A,B,T,g) has the weak P-property of the first kind;
T is a (α,g)-proximal admissible (φ,θ,α,g)-contraction;
if {zn}⊆A0 is a sequence such that {gzn}⊆A0 is Cauchy, then {zn} also is Cauchy;
there exists (x0,x1)∈A0×A0 such that d(gx1,Tx0)=ΔAB and α(gx0,gx1)≥1;
g is a continuous mapping;
T is a continuous mapping;
φ∈Ψ and θ∈Ω′.
Then there exists a convergent sequence {xn}n≥0⊆A0 verifying
(18)d(gxn+1,Txn)=ΔAB∀n≥0,
whose limit is a g-best proximity point of T.
Actually, every sequence {xn}n≥0⊆A0 verifying (18) and α(gx0,gx1)≥1 converges to a g-best proximity point of T.
Remark 21.
(1) Although the previous result seems to have too many hypotheses, actually, this is its best advantage. As we will see in Section 5, there are a lot of different ways to particularize this theorem which generate many independent results. For instance, our control functions do not need any kind of monotone property.
(2) This result improves the main theorem in [28] in several aspects: firstly, we introduce a mapping g:A→A which is not necessarily the identity mapping on A; secondly, (A,B) need not have the P-property; thirdly, the contractive condition on T is weaker; finally, we only suppose θ∈Ω′; that is, θ is not necessarily continuous.
(3) Taking into account the completeness of the ambient space X, the condition (d) can be interpreted as the continuity of the inverse mapping of g, if g is invertible. A simple way to guarantee this condition is to suppose that there are λ,n>0 such that d(x,y)≤λd(gx,gy)n for all x,y∈A. For instance, the condition d(x,y)≤d(gx,gy) for all x,y∈A can be found in [35].
(4) Notice that the second part of the thesis does not clarify whether the g-best proximity point of T is unique or not.
Proof.
Given x1∈A0, we know that Tx1∈T(A0)⊆B0. Then, there is z2∈A such that d(z2,Tx1)=ΔAB. Therefore, z2∈A0. Since A0⊆gA0, there is x2∈A0 such that gx2=z1, so d(gx2,Tx1)=d(z2,Tx1)=ΔAB. Repeating the same argument starting from x2∈A0, there is x3∈A0 such that d(gx3,Tx2)=ΔAB. By induction, we can consider a sequence {xn}⊆A0 such that
(19)d(gxn+1,Txn)=ΔAB∀n≥0.
If there exists some n0∈ℕ such that gxn0=gxn0+1, then d(gxn0,Txn0)=d(gxn0+1,Txn0)=ΔAB, so xn0 is a g-best proximity point of T. In such a case, if we define xm=xn0 for all m≥n0, we have that {xn}n≥n0 is constant, so {xn} converges to a g-best proximity point of T. In this case, the proof is finished.
On the contrary, suppose that
(20)d(gxn,gxn+1)>0∀n≥0.
Notice that, in particular, xn,gxn+1∈A0 and Txn∈B0 for all n≥0. We claim that
(21)α(gxn,gxn+1)≥1∀n≥0.
If n=0, then α(gx0,gx1)≥1 by hypothesis. Suppose that α(gxn,gxn+1)≥1 for some n≥0. Hence, taking into account that T is (α,g)-proximal admissible, we have that
(22)xn,xn+1,xn+2∈A0α(gxn,gxn+1)≥1d(gxn+1,Txn)=ΔABd(gxn+2,Txn+1)=ΔAB}⟹α(gxn+1,gxn+2)≥1.
This proves that (21) holds. Moreover, using the weak P-property of the first kind, for all n≥0,
(23)xn,xn+1,xn+2∈A0d(gxn+1,Txn)=ΔABd(gxn+2,Txn+1)=ΔAB⇓d(gxn+1,gxn+2)≤d(Txn,Txn+1).
Next we use (21), (23), and the (φ,θ,α,g)-contractive property of T to see that, for all n≥0,
(24)d(gxn+1,gxn+2)≤d(Txn,Txn+1)≤α(gxn,gxn+1)d(Txn,Txn+1)≤φ(Mg(xn,xn+1))+θ(d(gxn+1,Txn)-ΔAB,d(gxn,Txn+1)-ΔAB,d(gxn,Txn)-ΔAB,d(gxn+1,Txn+1)-ΔAB)=φ(Mg(xn,xn+1))
(the last equality holds since the first argument of θ is zero), where
(25)Mg(xn,xn+1)=max(d(gxn,gxn+1),d(gxn,Txn)-ΔAB,d(gxn,Txn+1)+d(gxn+1,Txn)2d(gxn+1,Txn+1)-ΔAB,d(gxn,Txn+1)+d(gxn+1,Txn)2-ΔAB)≤max(d(gxn,gxn+1),d(gxn,gxn+1)-ΔAB+d(gxn+1,Txn)-ΔAB,d(gxn+1,gxn+2)+d(gxn+2,Txn+1)-ΔAB,((d(gxn,gxn+1)+d(gxn+1,gxn+2)×2-1+d(gxn+2,Txn+1)+d(gxn+1,Txn))×2-1)-ΔAB)=max(d(gxn,gxn+1),d(gxn,gxn+1)+ΔAB-ΔAB,d(gxn+1,gxn+2)+ΔAB-ΔAB,((d(gxn,gxn+1)+d(gxn+1,gxn+2)2-1+ΔAB+ΔAB+ΔAB)×2-1)-ΔAB)=max(d(gxn,gxn+1),d(gxn+1,gxn+2),d(gxn+1)+d(gxn+2)2d(gxn,gxn+1)+d(gxn+1,gxn+2)2)=max(d(gxn,gxn+1),d(gxn+1,gxn+2)).
Joining (24) and (25), we have that
(26)d(gxn+1,gxn+2)≤φ(max(d(gxn,gxn+1),d(gxn+1,gxn+2)))∀n≥0.
Using (20) and the fact that φ(t)<t for all t>0, if there exists some n0∈ℕ such that
(27)max(d(gxn0,gxn0+1),d(gxn0+1,gxn0+2))=d(gxn0+1,gxn0+2),
then we have that d(gxn0+1,gxn0+2)≤φ(d(gxn0+1,gxn0+2))<d(gxn0+1,gxn0+2), which is impossible. Then max(d(gxn,gxn+1),d(gxn+1,gxn+2))=d(gxn,gxn+1) for all n≥0 and (26) yields to
(28)d(gxn+1,gxn+2)≤φ(d(gxn,gxn+1))∀n≥0.
In particular, for all n≥1,
(29)d(gxn,gxn+1)≤φ(d(gxn-1,gxn))≤φ2(d(gxn-2,gxn-1))≤⋯≤φn(d(gx0,gx1)).
Next we prove that {gxn} is a Cauchy sequence. Fix ɛ>0 arbitrary and consider t0=d(gx0,gx1)>0. Since φ∈Ψ, the series ∑n≥1φn(t0) converges. In particular, there exists m0∈ℕ such that
(30)∑k=m0∞φn(t0)<ɛ.
Therefore, if m>n≥m0, we have that
(31)d(gxn,gxm)≤∑k=nm-1d(gxk,gxk+1)≤∑k=nm-1φk(d(gx0,gx1))≤∑k=m0∞φn(t0)<ɛ.
This means that {gxn} is a Cauchy sequence. Using the hypothesis (d), {xn} also is a Cauchy sequence. By the completeness of (X,d), there exists z∈X such that {xn}→z. From xn∈A0⊆A for all n, we deduce that z∈A (because A is closed). Since T and g are continuous mappings, {Txn}→Tz and {gxn}→gz. Taking limit in (19) as n→∞, we conclude that z is a g-best proximity point of T.
Next we change the conditions on the control functions.
Theorem 22.
Theorem 20 also holds if one replaces condition (h) by the following one:
φ∈Φ, θ∈Ω′′, and α is (2,g)-transitive.
Proof.
Taking into account that φ(t)<t for all t>0 and following the lines of the proof of Theorem 20, we deduce that
(32)xn∈A0,d(gxn+1,Txn)=ΔAB,d(gxn,gxn+1)>0,α(gxn,gxn+1)≥1,d(gxn+1,gxn+2)≤φ(d(gxn,gxn+1))∀n≥0.
By Lemma 5, we have that
(33){d(gxn,gxn+1)}⟶0.
Next, we are going to prove that {gxn} is a Cauchy sequence reasoning by contradiction. Assume that {gxn} is not Cauchy. In this case (following, for instance, [27]), there exist ε0>0 and two subsequences {xm(k)}k∈ℕ and {xn(k)}k∈ℕ verifying that, for all k∈ℕ,
(34)k≤m(k)<n(k),d(gxm(k),gxn(k))>ε0,d(gxm(k),gxp)≤ε0∀p∈{m(k)+1,m(k)+2,…,n(k)-2,n(k)-1},limk→∞d(gxm(k)-1,gxn(k)-1)=ε0,limk→∞d(gxm(k),gxn(k)+p)=ε0∀p≥0.
Notice that
(35)0≤d(gxn(k),Txn(k))-ΔAB≤d(gxn(k),gxn(k)+1)+d(gxn(k)+1,Txn(k))-ΔAB=d(gxn(k),gxn(k)+1).
Therefore
(36)limk→∞[d(gxn(k),Txn(k))-ΔAB]=0.
Similarly,
(37)limk→∞[d(gxm(k),Txm(k))-ΔAB]=0.
Furthermore,
(38)ε0<d(xm(k),xn(k))≤Mg(xm(k),xn(k))∀k≥0,
where, for all k≥0,
(39)Mg(xm(k),xn(k))=max(d(gxm(k),gxn(k)),d(gxm(k),Txm(k))-ΔAB,d(gxm(k),Txn(k))+d(gxn(k),Txm(k))2d(gxn(k),Txn(k))-ΔAB,d(gxm(k),Txn(k))+d(gxn(k),Txm(k))2-ΔAB).
Notice that
(40)d(gxm(k),Txn(k))+d(gxn(k),Txm(k))2-ΔAB≤(2-1(2-1d(gxm(k),gxn(k)+1)+d(gxn(k)+1,Txn(k))+2-1d(gxn(k),gxm(k)+1)+d(gxm(k)+1,Txm(k)))×2-1)-ΔAB=(2-1(d(gxm(k),gxn(k)+1)+ΔAB+d(gxn(k),gxm(k)+1)+ΔAB)×2-1)-ΔAB=d(gxm(k),gxn(k)+1)+d(gxn(k),gxm(k)+1)2.
Taking limit as k→∞ and using (34),
(41)limk→∞(d(gxm(k),Txn(k))+d(gxn(k),Txm(k))2-ΔAB)≤ε0+ε02=ε0.
Taking limit as k→∞ in (39) and using (34), (36), (37), and (41), we deduce that
(42)limk→∞Mg(xm(k),xn(k))=max(ɛ0,0,0,ε0)=ε0.
This means that {Mg(xm(k),xn(k))}k∈ℕ is a sequence of real numbers converging to ε0 and whose terms are strictly greater than ε0. In particular, since φ∈Φ,
(43)limk→∞φ(Mg(xm(k),xn(k)))=limt→ε0+φ(t)<ε0.
From the fact that α(gxn,gxn+1)≥1 for all n≥0 and using that α is (2,g)-transitive, we deduce that
(44)α(gxm(k),gxn(k))≥1∀k≥0.
Since (A,B,T,g) has the weak P-property of the first kind, for all k≥0,
(45)xm(k),xm(k)+1,xn(k),xn(k)+1∈A0d(gxm(k)+1,Txm(k))=ΔABd(gxn(k)+1,Txn(k))=ΔAB⇓d(gxm(k)+1,gxn(k)+1)≤d(Txm(k),Txn(k)).
Therefore, from the (φ,θ,α,g)-contractivity condition on T, it follows that, for all k≥0,
(46)d(gxm(k)+1,gxn(k)+1)≤d(Txm(k),Txn(k))≤α(Txm(k),Txn(k))d(Txm(k),Txn(k))≤φ(Mg(xm(k),xn(k)))+θ(d(gxn(k),Txm(k))-ΔAB,d(gxm(k),Txn(k))-ΔAB,d(gxm(k),Txm(k))-ΔAB,d(gxn(k),Txn(k))-ΔAB).
Using (36), the third and the fourth arguments of θ converge to zero as k→∞. Since θ∈Ω′′, all the terms tend to zero as k→∞. Hence, letting k→∞ in (46) and using (34) and (43), we conclude that
(47)ε0=limk→∞d(gxm(k)+1,gxn(k)+1)≤limk→∞φ(Mg(xm(k),xn(k)))<ε0,
which is impossible. This contradiction proves that {gxn} is a Cauchy sequence. Then, the rest of the proof is similar to the proof of Theorem 20.
In the following theorem, we replace the continuity of T by another condition.
Theorem 23.
Theorem 20 also holds if one supposes that the contractive condition (10) is valid for all x∈A0 and all y∈A, and one replaces conditions (b) and (g) by the following ones:
the quadruple (A,B,T,g) has the weak P-property of the second kind;
if {xn}⊆A0 is a sequence verifying {xn}→x∈A and α(gxn,gxn+1)≥1 for all n≥0, then there exists a partial subsequence {xn(k)} of {xn} such that α(gxn(k),gx)≥1 for all k≥0.
Proof.
Following the lines of the proof of Theorem 20, we deduce that {gxn} and {xn} are Cauchy sequences, contained in the closed subset A, of the complete metric space (X,d). Then, there is x∈A such that {xn}→x and, using that g is continuous, {gxn}→gx. We are going to prove that x is a g-best proximity point of T.
Since (A,B,T,g) has the weak P-property of the second kind, for all n,m∈ℕ,
(48)xm,xm+1,xn,xn+1∈A0d(gxm+1,Txm)=ΔABd(gxn+1,Txn)=ΔAB⇓d(gxm+1,gxn+1)=d(Txm,Txn).
It follows that {Txn} is also a Cauchy sequence in the closed subset B. Hence, there is z∈B such that {Txn}→z. This means that
(49){d(gxn,gx)}⟶0,{d(Txn,z)}⟶0.
Since d(gxn+1,Txn)=ΔAB for all n≥0, we deduce that
(50)d(gx,z)=ΔAB;
that is, gx∈A0 and z∈B0. Using condition (g'), we deduce that there exists a partial subsequence {xn(k)} of {xn} such that
(51)α(gxn(k),gx)≥1∀k≥0.
Notice that
(52)0≤d(gxn(k),Txn(k))-ΔAB≤d(gxn(k),gxn(k)+1)+d(gxn(k)+1,Txn(k))-ΔAB=d(gxn(k),gxn(k)+1).
Therefore
(53)limk→∞[d(gxn(k),Txn(k))-ΔAB]=0.
The first and the second arguments of
(54)Mg(xn(k),x)=max(d(gx,gxn(k)),d(gxn(k),Txn(k))-ΔAB,d(gxn(k),Tx)+d(gx,Txn(k))2d(gx,Tx)-ΔAB,d(gxn(k),Tx)+d(gx,Txn(k))2-ΔAB)
tend to zero, and the last argument tends to
(55)limk→∞(d(gxn(k),Tx)+d(gx,Txn(k))2-ΔAB)≤limk→∞((×2-1)(d(gxn(k),Tx)+d(gx,gxn(k)+1)(gxn(k),Tx)×2-1)+d(gxn(k)+1,Txn(k)))×2-1)-ΔAB)=limk→∞(d(gxn(k),Tx)+d(gx,gxn(k)+1)+ΔAB2-ΔAB)=d(gx,Tx)+0+ΔAB2-ΔAB=d(gx,Tx)-ΔAB2.
Therefore,
(56)limk→∞Mg(xn(k),x)=d(gx,Tx)-ΔAB.
Next we are going to show that x is a g-best proximity point of T reasoning by contradiction. Suppose that d(gx,Tx)≠ΔAB; that is,
(57)t0=d(gx,Tx)-ΔAB>0.
Since the first and the second terms in the maximum in (54) tend to zero, and the fourth term tends to t0/2, then there exists k0∈ℕ such that
(58)Mg(xn(k),x)=d(gx,Tx)-ΔAB=t0>0∀k≥k0.
Using the contractivity condition (notice that xn(k)∈A0 but x∈A), for all k≥k0,
(59)d(Txn(k),Tx)≤α(gxn(k),gx)d(Txn(k),Tx)≤φ(Mg(xn(k),x))+θ(d(gx,Tx)-ΔABd(gx,Txn(k))d(gx,Txn(k))-ΔAB,d(gxn(k),Tx)-ΔAB,d(gxn(k),Txn(k))-ΔAB,d(gx,Tx)-ΔABd(gx,Txn(k)))=φ(d(gx,Tx)-ΔAB)+θ(d(gx,Txn(k))-ΔAB,d(gxn(k),Tx)-ΔAB,d(gxn(k),Txn(k))-ΔAB,d(gx,Tx)-ΔAB).
Since the third argument of θ in (59) tends to zero and θ∈Ω′′, its limit as k→∞ is zero. Therefore, letting k→∞ in (59), we have that
(60)d(z,Tx)=limk→∞d(Txn(k),Tx)≤φ(d(gx,Tx)-ΔAB).
As d(gx,Tx)-ΔAB>0, item (1) of Remark 13 guarantees that φ(d(gx,Tx)-ΔAB)<d(gx,Tx)-ΔAB. Thus,
(61)d(z,Tx)≤φ(d(gx,Tx)-ΔAB)<d(gx,Tx)-ΔAB≤d(gx,z)+d(z,Tx)-ΔAB≤ΔAB+d(z,Tx)-ΔAB=d(z,Tx),
which is impossible. This contradiction shows that x must verify d(gx,Tx)=ΔAB; that is, x is a g-best proximity point of T.
Remark 24.
When α is (2,g)-transitive, condition (g') is equivalent to the following one.
If {xn}⊆A0 is a sequence verifying {xn}→x∈A0 and α(gxn,gxn+1)≥1 for all n≥0, then α(gxn,gx)≥1 for all n≥0.
Remark 25.
Notice that, following the same sketch of proof with appropriate changes, Theorem 23 remains true under the hypothesis of Theorem 22.
4. Uniqueness of g-Best Proximity Points
In this section, we introduce a sufficient condition in order to demonstrate that the g-best proximity point, whose existence is guaranteed by the previous results, is unique.
Definition 26.
Let T:A→B, g:A→A, and α:X×X→[0,∞) be three mappings. One will say that T is (α,g)-regular if, for all x,y∈A0 such that α(gx,gy)<1, there exists z∈A0 such that α(gx,gz)≥1 and α(gy,gz)≥1.
Theorem 27.
Under the hypothesis of Theorem 20, assume that θ∈Θ and T is (α,g)-regular. Then for all g-best proximity points x and y of T in A0 One has that gx=gy.
In particular, if g is injective on the set of all g-best proximity points of T in A0, then T has a unique g-best proximity point.
Proof.
Let x,y∈A0 be two g-best proximity points of T in A0. Since d(gx,Tx)=d(gy,Ty)=ΔAB and T is a (α,g)-proximal admissible, we deduce that
(62)d(gx,gy)≤d(Tx,Ty).
We distinguish whether α(gx,gy)≥1 or α(gx,gy)<1. Firstly, assume that α(gx,gy)≥1. In such a case, the contractivity condition yields to
(63)d(gx,gy)≤d(Tx,Ty)≤α(gx,gy)d(Tx,Ty)≤φ(Mg(x,y))+θ(d(gy,Tx)-ΔAB,d(gx,Ty)-ΔAB,d(gx,Tx)-ΔAB,d(gy,Ty)-ΔAB)=φ(Mg(x,y)),
where the last equality holds since θ∈Θ and the last two arguments of θ are zero. Since
(64)d(gx,Ty)+d(gy,Tx)2-ΔAB≤d(gx,gy)+d(gy,Ty)+d(gy,gx)+d(gx,Tx)2-ΔAB=d(gx,gy)+ΔAB+d(gy,gx)+ΔAB2-ΔAB=d(gx,gy)+d(gy,gx)2=d(gx,gy),
it follows that
(65)Mg(x,y)=max(d(gx,Ty)+d(gy,Tx)2d(gx,gy),d(gx,Tx)-ΔAB,d(gy,Ty)-ΔAB,d(gx,Ty)+d(gy,Tx)2-ΔAB)=d(gx,gy).
Therefore
(66)d(gx,gy)≤φ(Mg(x,y))=φ(d(gx,gy)),
which is only possible when d(gx,gy)=0; that is, gx=gy.
Next, suppose that α(gx,gy)<1. In this case, by the (α,g)-regularity of T, there exists z0∈A0 such that α(gx,gz0)≥1 and α(gy,gz0)≥1. Based on z0, we are going to define a sequence {zn} such that {gzn} will converge, at the same time, to gx and to gy. By the unicity of the limit, this will prove that gx=gy. We only reason with x, but the same argument is valid for y.
Indeed, since Tz0∈TA0⊆B0, there is s0∈A0 such that d(s0,Tz0)=ΔAB, and since s0∈A0⊆gA0, there is z1∈A0 verifying gz1=s0. Therefore, d(gz1,Tz0)=ΔAB. Repeating this argument, there exists a sequence {zn}⊆A0 such that d(gzn+1,Tzn)=ΔAB for all n≥0. In particular, gzn+1∈A0 and Tzn∈B0.
Now we reason using x. We claim that
(67)α(gx,gzn)≥1∀n≥0.
If n=0, α(gx,gz0)≥1 by the choice of z0. Suppose that α(gx,gzn)≥1 for some n≥0. In such a case, taking into account that T is (α,g)-proximal admissible, we have that
(68)x,zn,zn+1∈A0α(gx,gzn)≥1d(gx,Tx)=ΔABd(gzn+1,Tzn)=ΔAB}⟹α(gx,gzn+1)≥1.
This concludes that (67) holds. Taking into account that, for all n≥0,
(69)d(gx,Tzn)+d(gzn,Tx)2-ΔAB≤d(gx,gzn+1)+d(gzn+1,Tzn)+d(gzn,gx)+d(gx,Tx)2-ΔAB=d(gx,gzn+1)+ΔAB+d(gzn,gx)+ΔAB2-ΔAB=d(gx,gzn+1)+d(gzn,gx)2≤max(d(gx,gzn),d(gx,gzn+1)),
it follows that, for all n≥0,
(70)Mg(x,zn)=max(d(gx,gzn),d(gx,Tx)-ΔAB,d(gx,Tzn)+d(gzn,Tx)2d(gzn,Tzn)-ΔAB,d(gx,Tzn)+d(gzn,Tx)2-ΔAB)≤max(d(gx,gzn),d(gx,gzn+1)).
Therefore, using the weak P-property of the first kind,
(71)x,zn,zn+1∈A0d(gx,Tx)=ΔABd(gzn+1,Tzn)=ΔAB}⟹d(gx,gzn+1)≤d(Tx,Tzn),
and, hence, by the contractivity condition, for all n≥0,
(72)d(gx,gzn+1)≤d(Tx,Tzn)≤φ(Mg(x,zn))+θ(d(gzn,Tx)-ΔAB,d(gx,Tzn)-ΔAB,d(gx,Tx)-ΔAB,d(gzn,Tzn)-ΔAB)≤φ(Mg(x,zn))≤φ(max(d(gx,gzn),d(gx,gzn+1))).
Suppose that there is n0∈ℕ such that gzn0=gx. In this case
(73)d(gx,gzn0+1)≤φ(max(d(gx,gzn0),d(gx,gzn0+1)))=φ(d(gx,gzn0+1)),
but this is only possible when d(gx,gzn0+1)=0; that is, gzn0+1=gx. Repeating this argument, we have that gzn=gx for all n≥n0, which proves that {gzn} is a sequence converging to gx. In this case, the proof is finished.
On the other hand, suppose that gzn≠gx for all n≥0; that is, d(gx,gzn)>0 for all n≥0. In this case, it is impossible that max(d(gx,gzn),d(gx,gzn+1))=d(gx,gzn+1) for some n, since (72) would yield to
(74)d(gx,gzn+1)≤φ(max(d(gx,gzn),d(gx,gzn+1)))=φ(d(gx,gzn+1))<d(gx,gzn+1).
Therefore, max(d(gx,gzn),d(gx,gzn+1))=d(gx,gzn)); that is, for all n≥0,
(75)d(gx,gzn+1)≤φ(Mg(x,zn))=φ(d(gx,gzn)).
Recursively, for all n≥0,
(76)d(gx,gzn)≤φ(d(gx,gzn-1))≤φ2(d(gx,gzn-2))≤⋯≤φn(d(gx,gz0)).
Next we prove that {gzn} converges to gx. Fix ɛ>0 arbitrary and consider t0=d(gx,gz0)>0. Since φ∈Ψ, the series ∑n≥1φn(t0) converges. In particular, there exists m0∈ℕ such that ∑k=m0∞φn(t0)<ɛ. More precisely, φn(t0)<ɛ for all n≥m0. Therefore, if n≥m0, we have that
(77)d(gx,gzn)≤φn(d(gx,gz0))=φn(t0)<ɛ.
This means that {gxn} converges to gx, and this finishes the proof.
Notice that if the regularity condition considered in Definition 26 holds for all x,y∈A, then we can deduce that gx=gy for all g-best proximity points x and y of T in A, but using the weak P-property of the third kind in A. We also point out that we could deduce the unicity of the g-best proximity point if g is injective on the set of all g-best proximity points of T (not necessarily on A).
5. Consequences
If g is the identity mapping on A, we deduce the following result.
Corollary 28.
Theorem 11 immediately follows from Theorem 20.
If there is k∈[0,1) such that φ(t)=kt for all t>0, one has the following result.
Corollary 29.
Let A and B be two closed subsets of a complete metric space (X,d). Let T:A→B, g:A→A, φ:[0,∞)→[0,∞), θ:[0,∞)4→[0,∞), and α:X×X→[0,∞) be five mappings. Assume that the following conditions hold:
∅≠A0⊆gA0 and T(A0)⊆B0;
the quadruple (A,B,T,g) has the weak P-property of the first kind;
T is a (α,g)-proximal admissible and there is k∈[0,1) verifying that for all x,y∈A0 such that d(gy,Tx)=ΔAB and α(gx,gy)≥1, one has that
(78)α(gx,gy)d(Tx,Ty)≤kMg(x,y)+θ(d(gy,Tx)-ΔAB,d(gx,Ty)-ΔAB,d(gx,Tx)-ΔAB,d(gy,Ty)-ΔAB);
if {zn}⊆A0 is a sequence such that {gzn}⊆A0 is Cauchy, then {zn} also is Cauchy;
there exists (x0,x1)∈A0×A0 such that d(gx1,Tx0)=ΔAB and α(gx0,gx1)≥1;
g is a continuous mapping;
T is a continuous mapping;
φ∈Ψ and θ∈Ω′.
Then there exists a convergent sequence {xn}n≥0⊆A0 verifying
(79)d(gxn+1,Txn)=ΔAB∀n≥0,
whose limit is a g-best proximity point of T. Actually, every sequence {xn}n≥0⊆A0 verifying (18) and α(gx0,gx1)≥1 converges to a g-best proximity point of T.
If θ(t1,t2,t3,t4)=0 for all t1,t2,t3,t4≥0, we deduce the following corollary.
Corollary 30.
Let A and B be two closed subsets of a complete metric space (X,d). Let T:A→B, g:A→A, φ:[0,∞)→[0,∞), θ:[0,∞)4→[0,∞), and α:X×X→[0,∞) be five mappings. Assume that the following conditions hold:
∅≠A0⊆gA0 and T(A0)⊆B0;
the quadruple (A,B,T,g) has the weak P-property of the first kind;
T is a (α,g)-proximal admissible and there is φ∈Ψ verifying that for all x,y∈A0 such that d(gy,Tx)=ΔAB and α(gx,gy)≥1, one has that
(80)α(gx,gy)d(Tx,Ty)≤φ(Mg(x,y));
if {zn}⊆A0 is a sequence such that {gzn}⊆A0 is Cauchy, then {zn} also is Cauchy;
there exists (x0,x1)∈A0×A0 such that d(gx1,Tx0)=ΔAB and α(gx0,gx1)≥1;
g is a continuous mapping;
T is a continuous mapping;
φ∈Ψ and θ∈Ω′.
Then there exists a convergent sequence {xn}n≥0⊆A0 verifying
(81)d(gxn+1,Txn)=ΔAB∀n≥0,
whose limit is a g-best proximity point of T. Actually, every sequence {xn}n≥0⊆A0 verifying (18) and α(gx0,gx1)≥1 converges to a g-best proximity point of T.
If the pair (A,B) has the P-property, we conclude the following particular version.
Corollary 31.
Let A and B be two closed subsets of a complete metric space (X,d). Let T:A→B, g:A→A, φ:[0,∞)→[0,∞), θ:[0,∞)4→[0,∞), and α:X×X→[0,∞) be five mappings. Assume that the following conditions hold:
∅≠A0⊆gA0 and T(A0)⊆B0;
the pair (A,B) has the P-property;
T is a (α,g)-proximal admissible (φ,θ,α,g)-contraction;
if {zn}⊆A0 is a sequence such that {gzn}⊆A0 is Cauchy, then {zn} also is Cauchy;
there exists (x0,x1)∈A0×A0 such that d(gx1,Tx0)=ΔAB and α(gx0,gx1)≥1;
g is a continuous mapping;
T is a continuous mapping;
φ∈Ψ and θ∈Ω′.
Then there exists a convergent sequence {xn}n≥0⊆A0 verifying
(82)d(gxn+1,Txn)=ΔAB∀n≥0,
whose limit is a g-best proximity point of T. Actually, every sequence {xn}n≥0⊆A0 verifying (18) and α(gx0,gx1)≥1 converges to a g-best proximity point of T.
If α is the mapping associated to a binary relation ≼ (a transitive relation, a preorder, or a partial order), we have the following result.
Corollary 32.
Let A and B be two closed subsets of a complete metric space (X,d) provided with a binary relation ≼. Let T:A→B, g:A→A, φ:[0,∞)→[0,∞), and θ:[0,∞)4→[0,∞) be four mappings. Assume that the following conditions hold:
∅≠A0⊆gA0 and T(A0)⊆B0;
the quadruple (A,B,T,g) has the weak P-property of the first kind;
T verifies the following properties:
if a1,a2,b1,b2∈A0, then
(83)gb1≼gb2d(ga1,Tb1)=ΔABd(ga2,Tb2)=ΔAB}⟹ga1≼ga2;
for all x,y∈A0 such that d(gy,Tx)=ΔAB and gx≼gy, one has that
(84)d(Tx,Ty)≤φ(Mg(x,y))+θ(d(gy,Tx)-ΔAB,d(gx,Ty)-ΔAB,d(gx,Tx)-ΔAB,d(gy,Ty)-ΔAB),
if {zn}⊆A0 is a sequence such that {gzn}⊆A0 is Cauchy, then {zn} also is Cauchy;
there exists (x0,x1)∈A0×A0 such that d(gx1,Tx0)=ΔAB and gx0≼gx1;
g is a continuous mapping;
T is a continuous mapping;
φ∈Ψ and θ∈Ω′.
Then there exists a convergent sequence {xn}n≥0⊆A0 verifying
(85)d(gxn+1,Txn)=ΔAB∀n≥0,
whose limit is a g-best proximity point of T. Actually, every sequence {xn}n≥0⊆A0 verifying (18) and α(gx0,gx1)≥1 converges to a g-best proximity point of T.
If α(x,y)=1 for all x,y∈X, then we conclude the following consequence.
Corollary 33.
Let A and B be two closed subsets of a complete metric space (X,d) provided with a binary relationship ≼. Let T:A→B, g:A→A, φ:[0,∞)→[0,∞), and θ:[0,∞)4→[0,∞) be four mappings. Assume that the following conditions hold:
∅≠A0⊆gA0 and T(A0)⊆B0;
the quadruple (A,B,T,g) has the weak P-property of the first kind;
for all x,y∈A0 such that d(gy,Tx)=ΔAB, one has that
(86)d(Tx,Ty)≤φ(Mg(x,y))+θ(d(gy,Tx)-ΔAB,d(gx,Ty)-ΔAB,d(gx,Tx)-ΔAB,d(gy,Ty)-ΔAB);
if {zn}⊆A0 is a sequence such that {gzn}⊆A0 is Cauchy, then {zn} also is Cauchy;
there exists (x0,x1)∈A0×A0 such that d(gx1,Tx0)=ΔAB;
g is a continuous mapping;
T is a continuous mapping;
φ∈Ψ and θ∈Ω′.
Then there exists a convergent sequence {xn}n≥0⊆A0 verifying
(87)d(gxn+1,Txn)=ΔAB∀n≥0,
whose limit is a g-best proximity point of T. Actually, every sequence {xn}n≥0⊆A0 verifying (18) and α(gx0,gx1)≥1 converges to a g-best proximity point of T.
If A=B, the notion of g-best proximity point is equivalent to the concept of coincidence point. In this case, the pair (A,A) has the P-property.
Corollary 34.
Let A be a closed subset of a complete metric space (X,d). Let T,g:A→A, φ:[0,∞)→[0,∞), θ:[0,∞)4→[0,∞), and α:X×X→[0,∞) be five mappings. Assume that the following conditions hold:
∅≠A0⊆gA0 and T(A0)⊆A0;
T is a (α,g)-proximal admissible (φ,θ,α,g)-contraction;
if {zn}⊆A0 is a sequence such that {gzn}⊆A0 is Cauchy, then {zn} also is Cauchy;
there exists (x0,x1)∈A0×A0 such that gx1=Tx0 and α(gx0,gx1)≥1;
g is a continuous mapping;
T is a continuous mapping;
φ∈Ψ and θ∈Ω′.
Then there exists a convergent sequence {xn}n≥0⊆A0 verifying
(88)gxn+1=Txn∀n≥0,
whose limit is a coincidence point of T and g. Actually, every sequence {xn}n≥0⊆A0 verifying (88) and α(gx0,gx1)≥1 converges to a coincidence point of T and g.
As we have just seen, combining the previous results, including the possibility of changing (h) by (h'), we could deduce a lot of different independent corollaries, for instance, the following well-known ones. Using A=B=X, g as the identity mapping on X, φ(t)=kt for all t≥0 and α(x,y)=1 for all x,y∈X, we deduce the following property.
Every contractive mapping from a complete metric space into itself has a unique fixed point.
The following results are also particular cases of our main result.
Corollary 36 (Ran and Reurings [14]).
Let (X, ≼) be an ordered set endowed with a metric d and let T:X→X be a given mapping. Suppose that the following conditions hold.
(X,d) is complete.
T is nondecreasing (with respect to ≼≼).
T is continuous.
There exists x0∈X such that x0≼Tx0.
There exists a constant k∈(0,1) such that d(Tx,Ty)≤kd(x,y) for all x,y∈X with x≽y.
Then T has a fixed point. Moreover, if for all (x,y)∈X2 there exists z∈X such that x≼z and y≼z, one obtains uniqueness of the fixed point.
Proof.
Consider A=B=X. Therefore A0=B0=X. Let g be the identity mapping on X. By Remark 16, the quadruple (X,X,T,g) has the P-property. Let define α:X×X→[0,∞) by(89)α(x,y)={1,ifx≼y,0,otherwise.
As T is ≼-nondecreasing, then T is (α,g)-proximal admissible:
(90)α(x,y)≥1⟹x≼y⟹Tx≼Ty⟹α(Tx,Ty)≥1.
Let x,y∈X. If α(x,y)=0, then the contractivity condition (10) is obvious. If α(x,y)>0, then x≼y. In particular, using item (e) to y≽x and φ(t)=kt for all t≥0, we have that
(91)α(x,y)d(Tx,Ty)=d(Ty,Tx)≤kd(y,x)=φ(d(x,y)),
so (10) also holds (choosing whatever θ∈Θ; for instance, see Example 4). In any case, T is a (g,α)-proximal admissible (φ,θ,α,g)-contraction. Starting from x0∈X such that x0≼Tx0, let x1=Tx0. Then d(gx1,Tx0)=ΔAB=0 and α(x0,x1)≥1. As g and T are continuous, all hypotheses of Theorem 20 are satisfied. Then T has a fixed point. Moreover, Theorem 27 guarantees that it is unique.
Nieto and Rodríguez-López [15] slightly modified the hypothesis of the previous result obtaining the following theorem.
Corollary 37 (Nieto and Rodríguez-López [15]).
Let (X, ≼) be an ordered set endowed with a metric d and let T:X→X be a given mapping. Suppose that the following conditions hold.
(X,d) is complete.
T is nondecreasing (with respect to ≼).
If a nondecreasing sequence {xm} in X converges to a some point x∈X, then xm≼x for all m.
There exists x0∈X such that x0≼Tx0.
There exists a constant k∈(0,1) such that d(Tx,Ty)≤kd(x,y) for all x,y∈X with x≽y.
Then T has a fixed point. Moreover, if for all (x,y)∈X2 there exists z∈X such that x≼z and y≼z, one obtains uniqueness of the fixed point.
Proof.
We can follow point by point the proof of the previous result, but using Theorem 23 rather than Theorem 20.
From these results, it is also possible to prove many other fixed point results (see, for instance, [36]).
The main differences between our results and Theorem 11 are the following ones. (1) Theorem 11 can only ensure the existence of fixed points; however, we study the existence and uniqueness of coincidence points, involving a mapping g. (2) Our results use more general test functions in the contractivity condition, and our results guarantee existence and uniqueness of coincidence points under different contractivity conditions.
Example 38.
Let X=ℝ be provided with its Euclidean metric d and consider A=B=X and T,g:X→X are defined by Tx=(x/2)+1 and gx=2x for all x∈X. Let α:X×X→[0,∞) be the mapping given by. (92)α(x,y)={1,ifx≤y,0,otherwise.
Notice that A0=B0=X. If we take x0=-2 and x1=0, then all hypotheses of Theorem 20 are satisfied using φ(t)=t/2 for all t≥0. Therefore, T and g have a coincidence point, which is x=2/3. However, Theorem 11 cannot be applied because it only guarantees the existence of a fixed point of F (which, in this case, is x=2).
Conflict of Interests
The authors declare that there is no conflict of interests regarding the publication of this paper.
Acknowledgments
The authors would like to thank the referees for their useful comments and suggestions. The first author would like to thank the Commission on Higher Education, the Thailand Research Fund, and the King Mongkut's University of Technology Thonburi (Grant no. MRG5580213) for financial support during the preparation of this paper. The second author has been partially supported by Junta de Andalucía by Project FQM-268 of the Andalusian CICYE.
ZabreĭkoP. P.KrasnoselĭM. A.Iterations of operators, and fixed points197119610061009MR0315533EdelsteinM.On fixed and periodic points under contractive mappings1962377479MR0133102ZBL0113.16503BrowderF. E.Nonexpansive nonlinear operators in a Banach space19655410411044MR018712010.1073/pnas.54.4.1041ZBL0128.35801CaristiJ.Fixed point theorems for mappings satisfying inwardness conditions1976215241251MR039432910.1090/S0002-9947-1976-0394329-4ZBL0305.47029FanK.Extensions of two fixed point theorems of F. E. Browder1969112234240MR025160310.1007/BF01110225ZBL0185.39503Sadiq BashaS.Best proximity points: global optimal approximate solutions2011491152110.1007/s10898-009-9521-0MR2740511ZBL1208.90128MongkolkehaC.KumamP.Some common best proximity points for proximity commuting mappings2013781825183610.1007/s11590-012-0525-1MR3132148ZBL06252204MongkolkehaC.KumamP.Best proximity point theorems for generalized cyclic contractions in ordered metric spaces2012155121522610.1007/s10957-012-9991-yMR2983115ZBL1257.54041MongkolkehaC.KumamP.Best proximity points for asymptotic proximal pointwise weaker Meir-Keeler-type ψ-contraction mappings20132128790MR307304310.1016/j.joems.2012.12.002MongkolkehaC.ChoY. J.KumamP.Best proximity points for generalized proximal C-contraction mappings in metric spaces with partial orders20132013article 9410.1186/1029-242X-2013-94MR3037628MongkolkehaC.ChoY. J.KumamP.Best proximity points for Geraghty's proximal contraction mappings20132013article 18010.1186/1687-1812-2013-180MR3089869NashineH. K.KumamP.VetroC.Best proximity point theorems for rational proximal contractions20132013article 9510.1186/1687-1812-2013-95MR3057263SanhanW.MongkolkehaC.KumamP.Generalized proximal ψ-contraction mappings and best proximity points2012201219896912MR299100110.1155/2012/896912RanA. C. M.ReuringsM. C. B.A fixed point theorem in partially ordered sets and some applications to matrix equations200413251435144310.1090/S0002-9939-03-07220-4MR2053350ZBL1060.47056NietoJ. J.Rodríguez-LópezR.Contractive mapping theorems in partially ordered sets and applications to ordinary differential equations200522322323910.1007/s11083-005-9018-5MR2212687ZBL1095.47013Gnana BhaskarT.LakshmikanthamV.Fixed point theorems in partially ordered metric spaces and applications20066571379139310.1016/j.na.2005.10.017MR2245511ZBL1106.47047BerindeV.BorcutM.Tripled fixed point theorems for contractive type mappings in partially ordered metric spaces201174154889489710.1016/j.na.2011.03.032MR2810674ZBL1225.54014BorcutM.BerindeV.Tripled coincidence theorems for contractive type mappings in partially ordered metric spaces2012218105929593610.1016/j.amc.2011.11.049MR2873069ZBL1262.54017KarapınarE.Quadruple fixed point theorems for weak ϕ-contractions2011201115989423MR285605710.5402/2011/989423KarapınarE.BerindeV.Quadruple fixed point theorems for nonlinear contractions in partially ordered metric spaces2012617489MR2862544ZBL1264.47057BerzigM.SametB.An extension of coupled fixed point's concept in higher dimension and applications20126381319133410.1016/j.camwa.2012.01.018MR2913025ZBL1247.54048RoldánA.Martínez-MorenoJ.RoldánC.Multidimensional fixed point theorems in partially ordered complete metric spaces2012396253654510.1016/j.jmaa.2012.06.049MR2961247ZBL1266.54094RoldánA.KarapınarE.Some multidimensional fixed point theorems on partially preordered G*-metric spaces under ψ,φ- contractivity conditions20132013article 15823KarapınarE.RoldánA.Martínez-MorenoJ.RoldánC.Meir-Keeler type multidimensional fixed point theorems in partially ordered metric spaces201320139406026MR3039171ZBL1272.5403610.1155/2013/406026AlghamdiM. A.KarapınarE.G-β-ψ contractive-type mappings and related fixed point theorems20132013article 701610.1186/1029-242X-2013-70MR3037617ZBL06252157BerzigM.KarapınarE.Fixed point results for αψ,βφ-contractive mappings for a generalized altering distance20132013article 20518BerzigM.KarapınarE.RoldánA.Discussion on generalized-αψ,βφ-contractive mappings via generalized altering distance functions and related fixed point theorems201420141225976810.1155/2014/259768MongkolkehaC.KongbanC.KumamP.Existence and uniqueness of best proximity points for generalized almost contractions201420141110.1155/2014/813614813614KirkW. A.ReichS.VeeramaniP.Proximinal retracts and best proximity pair theorems2003247-885186210.1081/NFA-120026380MR2011594ZBL1054.47040AydiH.KarapınarE.PostolacheM.Tripled coincidence point theorems for weak φ-contractions in partially ordered metric spaces20122012article 4410.1186/1687-1812-2012-44MR2915801ChoudhuryB. S.KunduA.A coupled coincidence point result in partially ordered metric spaces for compatible mappings20107382524253110.1016/j.na.2010.06.025MR2674088ZBL1229.54051WangS.Coincidence point theorems for G-isotone mappings in partially ordered metric spaces20132013article 9610.1186/1687-1812-2013-96MR3056624RajV. S.A best proximity point theorem for weakly contractive non-self-mappings201174144804480810.1016/j.na.2011.04.052MR2810719ZBL1228.54046ZhangJ.SuY.ChengQ.A note on ‘a best proximity theorem for Geraghty contractions’20132013article 994AlghamdiM. A.ShahzadN.VetroF.Best proximity points for some classes of proximal contractions2013201310713252MR310864210.1155/2013/713252RoldánA.Martínez-MorenoJ.RoldánC.KarapınarE.Some remarks on multidimensional fixed point theoremsFixed Point Theory (Romanian). In press