Introduction

In 2005, Echenique started combining fixed point theory and graph theory by giving a short constructive proof for the Tarski fixed point theorem using graphs [1]. Afterwards, Espinola and Kirk applied fixed point results in graph theory [2]. A considerable contribution was made by Jachymski [3] and Beg et al. [4]. More recently, the authors, by providing a new notion of (P)-graphs and using arguments similar to those of Reich et al. [58], presented some iterative scheme results for G-contractive and G-nonexpansive maps on graphs [9]. In this paper, we obtain some fixed point results on subgraphs of directed graphs. As some consequences of our results, we obtain the Caristi fixed point theorem and Knaster-Tarski fixed point theorem.

Let (X,d) be a metric space and G a directed graph G such that V(G)=X and the set E(G) of its edges contains all loops. We denote the conversion of a graph G by G−1, that is, the graph obtained from G by reversing the direction of the edges. A mapping f:XX preserves the edges of G whenever (x,y)∈E(G) implies (f x,f y)∈E(G) for all x,yX[3]. Since G is a directed graph, the direction of edge (x,y) is the inverse of the direction of edge (y,x), that is, (x,y)≠(y,x). Let G be the directed graph. A finite path of length n in G from x to y is a sequence { x i } i = 0 n of distinct vertices such that x0=x, x n =y, and (x i ,xi+1)∈E(G) for i=0,1,...,n−1 [9]. In fixed point theory, we like to deal with infinite graphs (see [9]). For this reason, we consider infinite paths. In fact, BE(G) is an infinite path whenever there is a finite path between any of its two vertices. Throughout this paper, a path could be finite or infinite, and the vertices of the path are pairwise distinct. Also, we consider cycles as finite paths. We denote by [x] G the set of all vertices in G wherein there is a (finite or infinite) path from those to x.

Let G be a subgraph of the directed graph G and xG. We emphasize that [ x] G denotes the set of all vertices in G wherein there is a path from those to x via the edges in G. Also, we remind here that V(G)⊆V(G) and E(G)⊆E(G). Let G be a subgraph of the directed graph G. We say that bG is an upper bound for G whenever g∈ [ b] G for all gG. Also, we say that cG is a supremum of G whenever c∈ [ b] G for all upper bounds b. In fact, c is a least upper bound in a sense.

Example 1.1

Let G be the directed graph via the vertices V(G) = {a,b,c,d} and the edges E(G) = {(a,b),(b,c),(c,d),(d,a)}. Suppose that G is a subgraph of G denied by V(G) = {a,b,c} and E(G) = {(a,b),(b,c)}, then c,d are upper bounds of G. Thus, an upper bound is not unique in a subgraph necessarily.

Example 1.2

Let G be the directed graph via the vertices V(G)={0,2, 1 n :n1} and the edges E(G)={( 1 n , 1 n + 1 ):n1}{( 1 n ,0)}(0,2){( 1 n ,2):n1}. If V( G )={0, 1 n :n1} and E( G )={( 1 n , 1 n + 1 )}, then 0 and 2 are the supremum of G. Thus, a supremum is not unique in a subgraph necessarily.

Let G be a directed graph and x0G. We say that x0 is an end point whenever there is no xG such that (x0,x)∈G and xx0. There are many directed graphs via end points. In the following result, we give a class of directed graphs which have end points. The proof of this result is straightforward.

Lemma 1.1

Let G be the directed graph, X = V (G), φ:XR a function, E (G) = {(x,y) : d(x,y) ≤ φ(x) − φ(y)}, and d a metric on X. If there exists x0X such that φ(x0) = infxX φ(x), then x0 is an end point of G.

Main results

Now we are ready to state and prove our main results. Let G be the directed graph and M the set of all paths in G. Then ⊆ is a partial order on M. By using Hausdorff’s maximum principle, M has a maximal element. This means that G has a maximal path. We use this subject in our results.

Theorem 2.1

Let G be a directed graph such that every path in G has an upper bound. Then G has an end point or a cycle.

Proof

Suppose that G has no cycle. Let B be the maximal path in G and u an upper bond of B. If u is not an end point, there exists xG such that xu and (u,x)∈E(G). Thus, B{x} is a path in G and BB{x}. This contradiction shows that u is an end point of G. □

Let G be a directed graph and T a selfmap on G. We say that T is a self-path map whenever x∈ [ T x] G for all xG.

Theorem 2.2

Let G be a directed graph. Then G has an end point if and only if each self-path map on G has a fixed point.

Proof

Suppose that G has an end point x0 and T is a self-path map. We prove that x0 is a fixed point of T. Since x0∈ [ Tx0] G , there is a (finite or infinite) path {λ i }i≥0 between x0 and Tx0. Since x0 is the end point of G and λ0 = x0, we have x0 = λ1. By continuing this process, it is easy to see that x0 = λ i for all i. Thus, x0=Tx0. Now assume that G is a directed graph and each self-path map on G has a fixed point but has no end point. Then for each xG, there exists yG such that yx and (x,y)∈E(G). By using the selection principle, we can define a selfmap T on G by T x=y. Note that T is a self-path map which has no fixed point. □

Example 2.1

Let G be the directed graph via the vertices V(G)={0, 1 n :n1} and the edges E(G)={( 1 n , 1 n + 1 ):n1}{( 1 n ,0)}(0,1). Define the selfmap T on G by T0=1 and T 1 n = 1 n + 1 for all n≥1. Then x∈ [ T x] G for all xG; T has no fixed point and G has no end point.

Theorem 2.3

Let G be a directed graph such that every path in G has a supremum and T a selfmap on G such that T x∈ [ T y] G for all x∈ [ y] G , G = {xG:x∈ [ T x] G }≠, and G has no cycle. Then T has a fixed point in G.

Proof

Suppose that B is a path in G and b is the supremum of B in G. Since c∈ [ b] G for all cB, T c∈ [ T b] G and so c∈ [ T b] G . It follows that Tb is an upper bound for B. Since b is the supremum, bT b. Thus, bG. By using Theorem , G has an end point. Since x∈ [ T x] for all xG, T x∈ [ T2x] G and so T is a self-path map on G. Now by using Theorem , T has a fixed point in G. □

Now we show that a version of Knaster-Tarski fixed point theorem is a consequence of Theorem .

Theorem 2.4

Let (X,≼) be a partially ordered set such that each chain in X has a supremum and T a monotone selfmap on X. Assume that there exists aX such that aT a. Then T has a fixed point.

Proof

Define the graph G by V(G)=X and E(G) = {(x,y):xy and xy}. Then T x∈ [ T y] G for all x∈ [ y] G . Since G = {xG:x∈ [ T x] G }≠ and G has no cycle, by using Theorem , T has a fixed point. □

Let X be a set and φ:X→(−,) a map. Suppose that G is the directed graph defined by V(G)=X and E(G) = {(x,y):d(x,y)≤φ(x)−φ(y)}. We say that φ is lower semi-continuous whenever φ(x)≤φ(x n ) for all sequence {x n } in X with x n x.

Lemma 2.5

Let X be a complete metric space and φ:X→(−,) a map bounded from below. Suppose that G is the directed graph defined by V(G)=X and E(G) = {(x,y):d(x,y)≤φ(x)−φ(y)}. If φ is lower semi-continuous, then G has an end point.

Proof

First we prove that G has no cycle. If G has a cycle, then there exists a path { λ i } i = 1 n in G such that λ1=λ n . It is easy to check that d(λ1,λ i )≤φ(λ1)−φ(λ i ) and d(λ i ,λ n )≤φ(λ i )−φ(λ n ) for all i=2,3,...,n−1, and so λ i =λ1 for i≥2. This contradiction shows that G has no cycle. Now we prove that each path in G has an upper bound. Let {x α }αΩ be a path in G. Then {φ(x α )} Ω is a decreasing net of real numbers. Since φ is bounded from below, there is an increasing sequence {α n }n≥1 in Ω such that lim n φ( x α n )= inf α Ω φ( x α ). One can easily show that { x α n } n 1 is a Cauchy sequence and so converges to some xX. Since φ is lower semi-continuous, x α n [ x ] G for all n≥1. Thus, x is an upper bound for { x α n } n 1 .

Now we show that x is an upper bound for {x α }αΩ. If there exists βΩ such that x α n [ x β ] G for all n≥1, then φ( x β )φ( x α n ) for all n≥1 which implies that φ(x β )= infαΩ φ(x α ). Since d( x α n , x β )φ( x α n )φ( x β ), we get x α n x β which implies that x β =x. Hence, φ(x)= infαΩ φ(x α ). Now we claim that x α ∈ [ x] G , and so x is an upper bound for { x α n } n 1 . In fact if there is αΩ such that x∈ [ x α ] G , then d(x,x α )≤φ(x)−φ(x α )≤φ(x α )−φ(x α )=0, and so x=x α . Since {x α }αΩ is a path in G, if the last case does not hold, then for each αΩ there exists n≥1 such that x α [ x α n ] G . Hence, x α ∈ [ x] G for all αΩ. Thus, x is an upper bound for {x α }αΩ. Now by using Theorem , G has an end point. □

Now we can consequent the Caristi fixed point theorem.

Theorem 2.5

Let X be a complete metric space, φ:X→(−,) a map bounded from below and lower semi-continuous, and T:XX a selfmap satisfying d(x,T x)≤φ(x)−φ(T x) for all xX. Then T has a fixed point.

Proof

Suppose that G is the directed graph via the vertices V(G)=X and the edges E(G) = {(x,y):d(x,y)≤φ(x)−φ(y)}. By using Lemma , G has an end point. It is easy to see that T is a self-path map on G. Now by using Theorem , T has a fixed point. □