On the Uniform Continuity in G-Metric Spaces
Baekjin Kim
designeralice@daum.net
A thesis submitted to the committee of Graduate School, Kongju National University in partial fulfillment of the requirements for the degree of Master of Arts Conferred in February 2009.
RISS: http://www.riss.kr/link?id=T11546441
Cite this thesis as:
Baekjin Kim (2009). On the Uniform Continuity in G-Metric Spaces (Master's thesis, Kongju National University, Korea). Retrieved from Research Information Sharing Service: http://www.riss.kr/link?id=T11546441.
ABSTRACT
In 2006, Z. Mustafa and B. Sims introduced G-metric which is a new generalized metric. In this paper, we define the uniform continuity in G-metric spaces and prove some properties of the uniform continuity in G-metric spaces.
Keywords: G-metric space, uniform continuity, Cauchy sequence
I. INTRODUCTION
In 1963, the notion of 2-metric space was introduced by S. Gähler [6, 7] as a generalization of usual notion of metric space (X,d). The definition of 2-metric space is as follows.
DEFINITION 1.1. Let X be a nonempty set. A function f:X3→[0,+∞) is said to be a 2-metric on X if it satisfies the following conditions.
- Given distinct elements x,y∈X, there exists an element z∈X such that d(x,y,z)≠0.
- d(x,y,z)=0 when at least two of x, y, z are equal.
- d(x,y,z)=d(x,z,y)=d(y,z,x) for all x,y,z∈X.
- d(x,y,z)≤d(x,y,a)+d(x,a,z)+d(a,y,z) for all x,y,a∈X.
But different authors proved that there is no relation between these two functions, for instance, K. S. Ha et al. [8] showed that the 2-metric is not always a continuous function, further there is no easy relationship between results obtained from the two metrics.
In 1992, B. Dhage in his Ph. D. thesis [1] introduced a new class of generalized metric space called D-metric spaces.
DEFINITION 1.2. Let X be a nonempty set. A D-metric on X is a function D:X3→[0,+∞) that satisfies the following conditions for each x, y, z, a∈X.
- D(x,y,z)≥0.
- D(x,y,z)=0 if and only if x=y=z.
- D(x,y,z)=D(p{x,y,z}) where p is a permutation function.
- D(x,y,z)≤D(x,y,a)+D(a,z,z).
In a subsequent series of papers [2, 3, 4], B. Dhage attempted to develop topological structures in such spaces. He claimed that D-metrics provide a generalization of ordinary metric functions.
But in 2004, Z. Mustafa and B. Sims demonstrated that most of the claims concerning the fundamental topological structure of D-metric space are incorrect (See [10, 13]), and in 2006, they introduced more appropriate notion of generalized metric space as follows (See [11, 12]).
DEFINITION 1.3. Let X be a nonempty set. A G-metric on X is a function G:X3→[0,+∞) that satisfies the following conditions for each x, y, z, a∈X.
- (G1) G(x,y,z)=0 if x=y=z.
- (G2) G(x,y,z)>0 if x≠y.
- (G3) G(x,x,y)≤G(x,y,z) if z≠y.
- (G4) G(x,y,z)=G(p{x,y,z}) where p is a permutation function.
- (G5) G(x,y,z)≤G(x,a,a)+G(a,y,z).
If G is a G-metric on X then the pair (X,G) is called a G-metric space.
EXAMPLE 1.4. Immediate examples of such a function are G(x,y,z):=max and G(x,\,y,\,z) := d(x,\,y) + d(y,\,z) + d(z,\,x) where d is an ordinary metric function.
EXAMPLE 1.5. If X = \mathbb{R}^n then a function G defined by G(x,\,y,\,z) := ( \lVert x-y \rVert ^p + \lVert y-z \rVert^p + \lVert z-x \rVert^p )^{\frac{1}{p}} is a G-metric for p > 0.
EXAMPLE 1.6. If X = \mathbb{R}^{+} then a function G defined by G(x,\,y,\,z) := \begin{cases} 0 \quad & \mathrm{if} \,\, x=y=z, \\[8pt] \max\left\{x,\,y,\,z\right\} \quad & \text{otherwise} \end{cases} is a G-metric.
II. PRELIMINARIES
We introduce some concepts and properties of sequences and functions on G-metric spaces as preliminaries. To find the proofs of propositions omitted in this chapter, see [11, 12, 14].
To avoid overusing the prefix 'G-' we adopt the convention that the prefix is not denoted at convergence, continuity, completeness if it doesn't make any confusions.
DEFINITION 2.1. Let (X,\,G) be a G-metric space, and let \left\{x_n \right\} be a sequence of points of X. A point x\in X is said to be the limit of the sequence \left\{ x_n \right\} if for any \epsilon > 0, there exists N \in \mathbb{N} such that G(x,\,x_n ,\, x_m ) < \epsilon for all n,\,m \ge N .
PROPOSITION 2.2. (Equivalent Conditions for Convergence of a Sequence)
Let (X,\,G) be a G-metric space, then the following four statements are equivalent.
- \left\{ x_n \right\} is G-convergent to x.
- G(x_n ,\,x_n ,\,x ) \to 0 as n \to +\infty .
- G(x_n ,\,x,\,x) \to 0 as n \to +\infty .
- G(x_m ,\, x_n ,\, x) \to 0 as m,\,n \to +\infty .
PROPOSITION 2.3. (Uniqueness of the Limit of a Sequence)
Let (X,\,G) be a G-metric space. If sequence \left\{ x_n \right\} in X
converges to x, then x is unique.
Proof. Let x_n \to y and y \neq x. Since \left\{ x_n \right\} converges to x and y, for each \epsilon > 0 there exist N_1 \in \mathbb{N} and N_2 \in \mathbb{N} such that for every n \ge N_1 , G(x,\,x_n ,\,x_n ) < \frac{\epsilon}{2} holds and for every n \ge N_2 , G(y,\,y,\,x_n ) < \frac{\epsilon}{2} holds. If N := \max \left\{ N_1 ,\,N_2 \right\} then for every n \ge N we have G(x,\,y,\,y) \le G(x,\,x_n ,\,x_n ) + G(x_n ,\,y,\,y) < \frac{\epsilon}{2} + \frac{\epsilon}{2} = \epsilon . Hence G(x,\,y,\,y) =0, which is a contradiction. So, x=y.
DEFINITION 2.4. Let (X,\,G) be a G-metric space. A sequence \left\{ x_n \right\} is said to be G-Cauchy if for every \epsilon > 0, there exists N\in \mathbb{N} such that G(x_n ,\,x_m ,\,x_p ) < \epsilon for all n,\,m,\,p \ge N.
PROPOSITION 2.5. (A Condition for a Sequence to be Cauchy)
If (X,\,G) is a G-metric space, then the following statements are equivalent.
- The sequence \left\{ x_n \right\} is G-Cauchy.
- For every \epsilon > 0, there exists N \in \mathbb{N} such that G(x_n ,\,x_m ,\, x_m ) < \epsilon for all n,\,m \ge N.
PROPOSITION 2.6. (Relation between Convergent and Cauchy)
Let (X,\,G) be a G-metric space. If a sequence \left\{ x_n \right\} in X converges to x, then the sequence \left\{ x_n \right\} is a Cauchy sequence.
Proof. Since x_n \to x , for each \epsilon > 0 there exists N_1 ,\, N_2 \in \mathbb{N} such that n \ge N_1 \,\, \text{implies} \,\, G(x_n ,\, x,\,x) < \frac{\epsilon}{2} and n \ge N_2 \,\, \text{implies} \,\, G(x ,\, x_m ,\,x_m ) < \frac{\epsilon}{2} . If N := \max \left\{ N_1 ,\,N_2 \right\} , then for every n,\,m \ge N we have G(x_n ,\, x_m ,\, x_m ) \leq G(x_n ,\, x,\,x ) + G(x,\, x_m ,\, x_m ) < \frac{\epsilon}{2} + \frac{\epsilon}{2} = \epsilon . Hence \left\{ x_n \right\} is a Cauchy sequence.
DEFINITION 2.7. A G-metric space (X,\,G) is said to be G-complete (or complete G-metric) if every G-Cauchy sequence in (X,\,G) is G-convergent in (X,\,G).
DEFINITION 2.8. Let (X,\,G) and (Y,\,H) be G-metric spaces, and let f : (X,\,G) \to (Y,\,H) be a function. Then f is said to be G-continuous at a\in X if and only if, given \epsilon > 0, there exists \delta > 0 such that x,\,y \in X and G(a,\,x,\,y) < \delta implies H(f(a),\,f(x),\,f(y)) < \epsilon . A function f is said to be G-continuous on X if and only if it is G-continuous at all a\in X.
PROPOSITION 2.9. (Relation between Continuity and Sequential Continuity)
Let (X,\,G) and (Y,\,H) be G-metric spaces. Then a function
f : (X,\,G) \to (Y,\,H) is continuous at a point x\in X if and only if it is
G-sequentially continuous at x, i.e., whenever \left\{ x_n \right\} is G-convergent to x, \left\{ f(x_n ) \right\} is G-convergent to f(x).
DEFINITION 2.10. A G-metric space (X,\,G) is said to be symmetric G-metric space if G(x,\,y,\,y)=G(y,\,x,\,x) for all x,\,y\in X.
PROPOSITION 2.11. (Continuity of a Symmetric G-metric)
Let (X,\,G) be a G-metric space, then the function G(x,\,y,\,z)
is jointly continuous in all three of its variables.
PROPOSITION 2.12. (Relation between D-Metric and G-Metric)
Every G-metric space (X,\,G) defines a metric space (X,\,d_G ) by
d_G (x,\,y) := \frac{1}{2} (G(x,\,y,\,y) + G(y,\,x,\,x)) \tag{2.1}
for all x,\,y\in X.
Note that if (X,\,G) is a symmetric G-metric space, then d_G (x,\,y) = G(x,\,y,\,y) \tag{2.2} for all x,\,y\in X . However, if (X,\,G) is not symmetric, then it holds by the G-metric properties that \frac{3}{4} G(x,\,y,\,y) \leq d_G (x,\,y) \leq \frac{3}{2} G(x,\,y,\,y) \tag{2.3} for all x,\,y\in X and in general these inequalities cannot be improved.
REMARK 2.13. The original definition of d_G by Z. Mustafa and B. Sims was d_G (x,\,y) := G(x,\,x,\,y) + G(x,\,y,\,y) . However, since metric d_D induced by D-metric is defined by d_D (x,\,y) := D(x,\,x,\,y) = D(x,\,y,\,y) , it makes sense to define d_G by (2.1). (Note that every D-metric is symmetric.)
PROPOSITION 2.14. (Relation between Completeness of G-Metric and induced D-Metric Spaces)
A G-metric space (X,\,G) is G-complete if and only if (X,\,d_G )
is a complete metric space.
DEFINITION 2.15. Let (X,\,G) be a G-metric space. For r \ge 0 we define B_G (x,\,r) := \left\{ y \in X \,\vert\, d_G (x,\,y) < r \right\} and B_G ' (x,\,r ) := \left\{ y\in X \,\vert\, d_G (x,\,y) < r \right\} \setminus \left\{ x \right\} . B_G (x,\,r) is called an open ball with center x \in X and radius r, and B_G ' (x,\,r) is called a deleted(or punctured) open ball with center x\in X and radius r.
EXAMPLE 2.16. Let X = \mathbb{R} . Denote G(x,\,y,\,z) = \lvert x-y \rvert + \lvert y-z \rvert + \lvert z-x \rvert for all x,\,y,\,z \in \mathbb{R}. Thus \begin{eqnarray} B_G (1,\,2) &=& \left\{ y\in \mathbb{R} \,\vert\, G(1,\,y,\,y) + G(y,\,1,\,1) < 4 \right\} \\[8pt] &=& \left\{ y\in \mathbb{R} \,\vert\, \lvert y-1 \rvert < 1 \right\} = (0,\,2). \end{eqnarray} This example shows that an open ball in a G-metric space is a generalization of an open interval in a usual metric space.
Let \tau be the set of all A\subseteq X with x\in A if and only if there exists r > 0 such that B_G (x,\,r) \subseteq A. Then \tau is a topology on X (induced by the G-metric G).
PROPOSITION 2.17. (The Openness of Open Balls)
Let (X,\,G) be a G-metric space. If r > 0 , then a ball B_G (x,\,r) with center x\in X and radius r is an open set.
Proof. Let z\in B_G (x,\,r) , hence G(x,\,z,\,z) + G(z,\,x,\,x) < 2r . If we set \delta := G(x,\,z,\,z) + G(z,\,x,\,x) \quad \text{and} \quad r ' := r-\delta then we prove that B_G (z,\,r ' ) \subseteq B_G (x,\,r). Let y\in B_G (z,\,r ' ), by definition of open ball, we have G(y,\,z,\,z) + G(z,\,y,\,y) < 2r ' . By (G5), we have \begin{eqnarray} G(y,\,x,\,x) + G(x,\,y,\,y) & \leq & G(y,\,z,\,z) + G(z,\,x,\,x) + G(x,\,z,\,z) + G(z,\,y,\,y) \\[8pt] & < & \delta + 2r ' = \delta + 2r - 2 \delta \leq 2r . \end{eqnarray} Hence B_G (z,\,r ' ) \subseteq B_G (x,\,r). That is, the ball B_G (x,\,r) is an open set.
DEFINITION 2.18. Let (X,\,G) be a G-metric space and S be a subset of X.
- The intersection of every closed set containing S is called closure of S and denoted by \overline{S} .
- If every open ball with center p\in X and radius r > 0 contains both some points of S and some points of X \setminus S, then p is called boundary point of S. A set of every boundary point of S is called boundary of S and denoted by \partial S .
- If every deleted open ball with center p\in X and radius r > 0 contains some points of S, then p is called cluster point of S. A set of every cluster point of S is called derived set of S and denoted by S ' .
PROPOSITION 2.19. (Properties of a Closure)
Let S be a subset of a metric space X. Then following relations hold.
- \overline{S} \setminus S \subseteq \partial S .
- \overline{S} \setminus S \subseteq S ' .
- \overline{S} = S \cup S ' .
Proof. See [9] $17.
PROPOSITION 2.20. (Relation between Convergence of a Sequence and Splitted Sequences)
Let \left\{ c_n \right\} be a sequence in a G-metric space (X,\,G), M \cup K = \mathbb{N}, M and K be disjoint and infinite, M = \left\{ m_k \right\}, K = \left\{ n_k \right\}, both \left\{ m_k \right\} and \left\{ n_k \right\} be strictly increasing sequences, and the limits of both \left\{ c_{n_k} \right\} and \left\{ c_{m_k} \right\} converge to a same point \lambda as k \to +\infty . Then \left\{ c_n \right\} converges to \lambda as n \to + \infty .
Proof. Let \epsilon > 0 be given. There exists N_1 \in \mathbb{N} and N_2 \in \mathbb{N} such that G(c_{n_k} ,\, \lambda ,\, \lambda ) < \epsilon \,\,\text{and}\,\, G(c_{m_k} ,\,\lambda ,\,\lambda ) < \epsilon for all k \ge N_1 and k \ge N_2 . Set N := \max \left\{ n_{N_1} ,\, m_{N_2} \right\} and let n > N then n=n_k or n=m_k for some k . Besides, we have k \ge N_1 and k \ge N_2 by monotonicity of \left\{ m_k \right\} and \left\{ n_k \right\}. Thus one of G(c_n ,\,\lambda,\,\lambda ) = G(c_{n_k} ,\, \lambda ,\,\lambda ) < \epsilon or G(c_n ,\,\lambda,\,\lambda ) = G(c_{m_k} ,\, \lambda ,\,\lambda ) < \epsilon holds and it completes the proof.
III. THE MAIN RESULT
One of the well-known properties of continuous functions is sequential continuity. That is, iff:D \to E is a continuous function and \left\{ x_n \right\} is a convergent sequence in D where D and E are metric spaces then the sequence \left\{ y_n \right\} defined by y_n := f(x_n ) is also convergent. If D is a first-countable space, then the converse also holds: any function preserving sequential limit is continuous. When D and E are further complete, by definition of completeness, f : D \to E is continuous if and only if \left\{ f(x_n )\right\} becomes a Cauchy sequence whenever \left\{ x_n \right\} is a Cauchy sequence in D. Revising this equivalent statement, it is possible to derive a sequential condition for uniform continuity of f.
For a function between Euclidean spaces, uniform continuity can be defined in terms of how the function behaves on sequences (See [5]). More specifically, let A be a subset of \mathbb{R} ^n . A function f : A \to \mathbb{R}^m is uniformly continuous if and only if for every pair of sequences \left\{ x_n \right\} and \left\{ y_n \right\} such that \lim_{n\to\infty} \lVert x_n - y_n \rVert =0 we have \lim_{n\to\infty} \lVert f(x_n ) - f(y_n ) \rVert =0 . Being motivated from Fitzpatrick's alternative definition of uniform continuity, an equivalent condition for uniform continuity related Cauchy sequence can be suggested: If f is defined on a subset of a compact set and preserves every Cauchy sequence to a Cauchy sequence, then f is uniformly continuous. We observe this property of uniform continuity in G-metric spaces.
DEFINITION 3.1. Let (X,\,G) and (Y,\,H) be G-metric spaces, S be a subset of X, and f : X \to Y be a function. If for every \epsilon > 0 there exists \delta > 0 such that H(f(s),\, f(t) ,\, f(u)) < \epsilon holds whenever G(s,\,t,\,u) < \delta and s,\,t,\,u \in S, then f is said to be uniformly G-continuous on S . If f is uniformly G-continuous on its domain, then f is said to be uniformly G-continuous.
THEOREM 3.2. (A Condition for Uniform Continuity)
Let (X,\,G) and (Y,\,H) be G-metric spaces, then the following
statements are equivalent.
- f : X \to Y is uniformly G-continuous.
- For every \epsilon > 0 there exists \delta > 0 such that H(f(s) ,\, f(t) ,\, f(t)) < \epsilon holds whenever G(s,\,t,\,t) < \delta .
Proof. It is trivial that the first statement implies the second one. Now we prove that the second statement implies the first one. Let \epsilon > 0 be given. Then there exists \delta > 0 such that H(f(s) ,\, f(t) ,\, f(t)) < \frac{\epsilon}{2} whenever G(s,\,t,\,t) < \delta . Let G(s,\,t,\,u) < \delta then we have G(s,\,t,\,t) \leq G(s,\,t,\,u) < \delta and G(t,\,t,\,u) \leq G(s,\,t,\,u) < \delta . Thus \begin{eqnarray} H(f(s),\,f(t),\,f(u)) & \leq & H(f(s),\,f(t),\,f(t)) + H(f(t),\,f(t),\,f(u)) \\[8pt] & < & \frac{\epsilon}{2} + \frac{\epsilon}{2} = \epsilon . \end{eqnarray} It completes the proof.
We need a lemma of Heine-Cantor's theorem for G-metric spaces.
THEOREM 3.3. (Heine-Cantor, G-Metric Version)
Let (X,\,G) and (Y,\,H) be G-metric spaces, X be compact, f : X \to Y be G-continuous. Then f is uniformly G-continuous.
Proof. Let \epsilon > 0 be given. Since f is continuous, we can associate to each point s\in X a positive number \delta _s such that t\in X,\, G(t,\,s,\,s) < \delta_s \,\,\text{implies}\,\, H(f(t),\,f(s),\,f(s)) < \frac{\epsilon}{2} . \tag{3.1} Let J_s be the set of all t\in X for which G(t,\,s,\,s) < \frac{1}{4} \delta_s . Since s\in J_s , the collection of all the sets J_s is an open cover of X. Since X is compact, there is a finite set of points s_1 , s_2 , \cdots , s_n in X such that X \subseteq J_{s_1} \cup J_{s_2} \cup \cdots \cup J_{s_n} . \tag{3.2} We put \delta := \frac{1}{4} \min \left\{ J_{s_1} ,\, J_{s_2} ,\, \cdots ,\, J_{s_n} \right\} . Then \delta > 0 . Now let t and s be points of X such that G(t,\,s,\,s) < \delta . By (3.2), there is an integer m such that 1 \le m \le n and s\in J_{s_m} . Hence G(s_k ,\,s,\,s ) \leq \frac{1}{4} \delta_{s_m} and we also have \begin{eqnarray} G(t,\,s_m ,\, s_m ) & \leq & G(t,\,s,\,s) + G(s,\,s_m ,\, s_m ) \\[8pt] & \leq & G(t,\,s,\,s) + 3G(s,\,s,\,s_m ) \\[8pt] & < & \delta + \frac{3}{4} \delta_{s_m} \leq \delta_{s_m} . \end{eqnarray} Finally (3.1) shows that H(f(t),\,f(s),\,f(s)) \leq H(f(s) ,\, f(s) ,\,f(s_m )) + H(f(t),\,f(s_m ),\, f(s_m )) < \epsilon . This completes the proof.
The following theorem shows our main result.
THEOREM 3.4. (Relation between G-Cauchy Sequence and Its Image)
Let (X,\,G) and (Y,\,H) be G-complete metric spaces, S be a subset of X, \overline{S} be compact and f : S \to Y be a function. Then f is uniformly G-continuous on S if and only if \left\{ f(s_n ) \right\} becomes a G-Cauchy sequence whenever \left\{ s_n \right\} is a G-Cauchy sequence in S.
Proof of the necessary condition. Assume that f is uniformly continuous on S. Suppose further that \left\{ s_n \right\} is a Cauchy sequence in S.
Let \epsilon > 0 be given. The uniform continuity of f implies the existence of \delta > 0 such that for all s,\,t,\,u\in S, H(f(s),\,f(t),\,f(u)) < \epsilon holds whenever G(s,\,t,\,u) < \delta . Since \left\{ s_n \right\} is a Cauchy sequence, there exists a positive integer N such that G(s_m ,\,s_n ,\,s_p ) < \delta for all m,\,n,\,p > N. Thus we have H(f(s_m ),\, f(s_n ),\, f(s_p )) < \epsilon for all m,\,n,\,p > N and we conclude that \left\{ f(s_n )\right\} is a Cauchy sequence.
Proof of the sufficiency condition. Assume that \left\{ f(s_n ) \right\} becomes a Cauchy sequence whenever \left\{ s_n \right\} is a Cauchy sequence in S. Assumption of theorem implies continuity of f. Suppose contrarily that f is not uniformly continuous on S. We need to prove the following claim.
CLAIM. Limit of f diverges at some point a\in \partial S .
Proof of Claim. Suppose contrarily that the limit of f converges at arbitrary point a\in \partial S. Define \overline{f} : \overline{S} \to Y by \overline{f} (x) = \begin{cases} f(x) \quad & \text{if} \,\, x\in S \\[8pt] \lim_{t\to x} f(t) \quad & \text{if} \,\, x\in\overline{S} \setminus S . \end{cases} \tag{3.3} It is guaranteed that the limit of f at x\in \overline{S} \setminus S in (3.3) converges, for \overline{S} \setminus S \subseteq \partial S .
Let b\in \overline{S} be given. We will prove that \overline{f} is continuous at b. Note that the variable t of the limit in (3.3) is taken over S, not \overline{S} . Thus continuity of \overline{f} is not yet guaranteed.
CASE 1. If b is an isolated point of \overline{S}, then \overline{f} is continuous at b by definition of continuity.
CASE 2. If b is not an isolated point of \overline{S} , then b is a cluster point of \overline{S}.
Let \left\{ a_n \right\} be a sequence in S that converges to b. Continuity of f and definition of \overline{f} together imply that \left\{ \overline{f} (a_n ) \right\} converges to f(b).
Let \left\{ b_n \right\} be a sequence in \overline{S} \setminus S that converges to b. By definition of \overline{f} , for arbitrary positive integer n, there exists a positive real number \delta_n such that \delta_n < \frac{1}{n} and H ( \overline{f} (t_n ) ,\, \overline{f} ( b_n ) ,\, \overline{f} ( b_n )) < \frac{1}{n} \tag{3.4} whenever G(t_n ,\,b_n ,\, b_n ) < \delta_n and t_n \in S. By the axiom of choice, t_n can be chosen for every n, and a sequence \left\{t_n \right\} is well-defined. (3.4) yields \begin{eqnarray} H(\overline{f} (b) ,\, \overline{f} (b_n ),\, \overline{f} (b_n )) & \leq & H(\overline{f} (b) ,\, \overline{f} (t_n ),\, \overline{f} (t_n )) + H( \overline{f} (t_n ),\, \overline{f} (b_n ),\, \overline{f} (b_n )) \\[8pt] & < & \frac{1}{n} + H(\overline{f} (b) ,\, \overline{f} (t_n ),\, \overline{f} (t_n )) \tag{3.5} \end{eqnarray} for every n. Here, since \left\{ t_n \right\} converges to b and t_n \in S for every n, we have \frac{1}{n} + H(\overline{f} (b) ,\, \overline{f} (t_n ),\, \overline{f} (t_n )) \to 0 \tag{3.6} as n \to +\infty . Thus (3.5) and (3.6) together imply \lim_{n\to\infty} H(\overline{f} (b) ,\, \overline{f} (b_n ),\, \overline{f} (b_n )) =0 , which implies \overline{f} (b_n ) \to \overline{f} (b) as n \to + \infty .
Let \left\{c_n \right\} be a sequence in \overline{S} that converges to b. Without loss of generality, we can assume that both \left\{ c_n \right\} \cap S and \left\{ c_n \right\} \cap (\overline{S} \setminus S) are infinite, for if one of them were finite it would be enough to consider the infinite one only. Let subsequence \left\{ c_{n_k} \right\} be constructed of all the terms of \left\{c_n \right\} which is in S, and subsequence \left\{ c_{m_k} \right\} be constructed of all the terms of \left\{c_n \right\} which is in \overline{S} \setminus S . By the result of previous discussion for \left\{ a_n \right\} and \left\{ b_n \right\} , both \left\{ \overline{f} (c_{n_k} )\right\} and \left\{ \overline{f} (c_{m_k} ) \right\} converge to \overline{f} (b) . Thus \left\{ \overline{f} (c_n )\right\} converges to \overline{f} (b) and \overline{f} is continuous at b.
Observing two cases, \overline{f} is continuous at arbitrary point b\in \overline{S}. Since \overline{S} is compact, \overline{f} is uniformly continuous on \overline{S} . Since f is a restriction of \overline{f}, f is also uniformly continuous on S. But it is contradict to assumption of f in the main proof. This proves Claim.
Now we go on to the main proof. By the claim, the limit of f diverges at some point a\in \partial S. Since a\in \partial S, a is a cluster point of S and there exists a sequence \left\{ s_n \right\} in S such that the limit of \left\{s_n \right\} converges to a and the limit of \left\{ f(s_n )\right\} diverges as n \to +\infty . Since X is complete, \left\{ x_n \right\} is a Cauchy sequence and \left\{ f(x_n )\right\} is also a Cauchy sequence. Completeness of Y implies the convergence of \left\{ f(s_n )\right\}. This contradiction yields uniform continuity of f on S.
REMARK 3.5. In Theorem 3.4, compactness of \overline{S} cannot be ommitted. For example, let G(x,\,y,\,z) = \lvert x-y \rvert + \lvert y-z \rvert + \lvert z-x \rvert and f : (\mathbb{R} ,\,G ) \to (\mathbb{R} ,\,G ) be defined by f(x) = x^2 . Then for every Cauchy sequence \left\{ s_n \right\}, \left\{ f(s_n ) \right\} is also a Cauchy sequence, while f is not uniformly continuous on \mathbb{R} .
Applying theorem 3.4 to functions in Euclidean spaces, we have the following result.
COROLLARY 3.6. (Cauchy-Sequential Condition for Uniform Continuity)
Let S be a bounded subset of \mathbb{R}^m and f : S \to \mathbb{R}^m be a
function. Then f is uniformly continuous on S if and only if \left\{ f(s_n ) \right\} becomes a
Cauchy sequence whenever \left\{ s_n \right\} is a Cauchy sequence.
Theorem 3.4 can be used to show a given function is not uniformly continuous.
EXAMPLE 3.7. Let f : (0,\,1) \to \mathbb{R} and \left\{ s_n \right\} be defined by f(x) = \cos \frac{1}{x} \,\,\text{and}\,\, s_n = \frac{1}{n \pi} . Then the limit of \left\{s_n \right\} converges to 0 and \left\{s_n \right\} is a Cauchy sequence. But \left\{ f(s_n ) \right\} is not a Cauchy sequence, for f(s_n ) = \cos n \pi = \begin{cases} 1 \quad & \text{if} \,\, n \,\,\text{is even},\\[8pt] -1 \quad & \text{if} \,\, n \,\,\text{is odd} \end{cases} diverges as n \to +\infty. Thus f is not uniformly continuous.
REMARK 3.8. As we see in Example 3.7, the condition of 'Cauchy sequence' in theorem 3.4 cannot be weaken as 'sequence.'
REFERENCES
- B. C. Dhage (1992) Generalized metric space and mapping 「with fixed point」Bulletin of the Calcutta Mathematical Society, vol. 84, pp. 329-336.
- B. C. Dhage (1994) 「On generalized metric spaces and topological structure. II」Pure and Applied Mathematika Sciences, vol. 40, no. 1-2, pp. 37–41.
- B. C. Dhage (1994) 「On continuity of mappings in D-metric spaces」Bulletin of the Calcutta Mathematical Society, vol. 86, no. 6, pp. 503-508.
- Fitzpatrick, Patrick (2005) 《Advanced Calculus 2nd Ed》Brooks Cole.
- B. C. Dhage (2000) 「Generalized metric spaces and topological structure. I」Analele S¸ tiint¸ifice ale Universit˘at¸ii Al. I. Cuza din Ias¸i. Serie Nou˘a. Matematic˘a, vol. 46, no. 1, pp. 3–4.
- S. Gähler (1963) 「2-metrische Räume und ihre topologische Struktur」Mathematische Nachrichten, vol. 26, pp. 115-148.
- S. Gähler (1966) 「Zur geometric 2-metriche raume」Revue Roumaine de Mathématiques Pures et Appliquées, vol. 40, pp. 664-669.
- K. S. Ha, Y. J. Cho and A. White (1988) 「Strictly convex and strictly 2-convex 2-normed spaces」Mathematica Japonica, vol. 33, no. 3, pp. 375-384.
- J. R. Munkres (2000) 《TOPOLOGY 2nd Ed》Prentice Hall.
- Z. Mustafa and B. Sims (2004) 「Some remarks concerning D-metric spaces」in International Conference on Fixed Point Theory and Applications, pp. 189–-198, Yokohama Japan.
- Z. Mustafa and B. Sims (2006) 「A new approach to generalized metric spaces」Journal of Nonlinear and Convex Analysis, vol. 7, no. 2, pp. 289-297.
- Z. Mustafa, H. Obiedat, and F. Awawdeh (2008) 「Some Fixed Point Theorem for Mapping on Complete G-Metric Spaces」Fixed Point Theory and Applications, vol. 2008, p. 12.
- S. V. R. Naidu, K. P. R. Rao, and N. Srinivasa Rao (2004) 「On the topology of D-metric spaces and generation of D-metric spaces from metric spaces」International Journal of Mathematics and Mathematical Sciences, vol. 2004, no. 51, pp. 2719-2740.
- S. Sedghi and N. Shobe (2007) 「A common fixed point theorem in two M-fuzzy metric spaces」Communications of Korean Mathematical Society, vol. 22, no. 4, pp. 513-526.
국문초록
1963년 Gähler가 2-거리함수를 소개한 이후 거리공간의 일반화로서 2-거리함수가 사용되었다. 그러나 1988년 Gähler에 의하여 정의된 거리함수가 기존의 거리함수의 일반화로서 적절하지 않다는 것이 K. S. Ha 등에 의하여 밝혀졌다.
1992년 B. Dhage는 그의 박사학위 논문 [1]에서 기존 거리함수의 또 다른 일반화인 D-거리를 정의하였으며 그 후 발표한 일련의 논문 [2, 3, 4]에서 D-거리함수가 2-거리함수를 대체할만한 적절한 거리함수임을 주장하였다. 그러나 2004년 Z. Mustafa와 B. Sims는 B. Dhage가 주장했던 성질들이 사실이 아님을 밝히며 2006년의 논문 [11]에서 기존의 거리함수를 대체할 적절한 거리함수로서 G-거리함수를 정의하고, 2008년의 논문 [12]에서 G-거리공간에서의 고정점 정리를 증명하였다.
이 논문에서는 G-거리함수가 기존의 거리함수를 대체할 적절한 거리함수임을 밝히는 과정의 일환으로 G-거리공간 위에서 정의된 함수의 평등연속성을 정의하고 그와 관련된 성질을 밝힌다.
와! G-metric space에 대해 정말 간결하게 정리된 논문인것 같습니다!
다만 G-metric을 정의할때 2-metric의 삼각부등식이 어떤점에서 문제가 있어서
저런식의 정의로 바꾼것인지는 이 글에선 제가 알수가 없군요....
혹시 괜찮으시다면 간단한 이유를 알려주실수 있을까요..!?