Fixed Point Theory, Volume 4, No.

2, 2003, 143-158
http://www.math.ubbcluj.ro/∼ nodeacj/sfptcj.htm

SOME COMMON FIXED POINT THEOREMS

FOR SEQUENCES OF NONSELF MULTIVALUED
OPERATORS IN METRICALLY CONVEX METRIC
SPACES

B. C. DHAGE1 , U. P. DOLHARE2 AND ADRIAN PETRUŞEL3

1
Kasubai, Gurukul Colony, Ahmedpur-413 515
Dist: Latur, Maharashtra, India
e-mail: bcd20012001yahoo.co.in
2
Department of Mathematics
Dnyanopasak Mahavidyalaya, Jintur-431 509
Maharashtra, India
3
Department of Applied Mathematics
Babeş-Bolyai University
3400, Cluj-Napoca, Romania
e-mail: [email protected]

Abstract. In this paper some common fixed point theorems for sequences of nonself multi-
valued operators defined on a closed subset of a metrically convex metric space are proved.
Our results extend some fixed point theorems of Dhage [4] to a sequence of nonself multi-
maps and include the fixed point result of Huang and Cho [6].
2000 Mathematics Subject Classification: 47H10, 54H25.
Key Words and Phrases: metrically convex metric space, multivalued operator, common
fixed point.

1. Introduction

Fixed point theorems for nonself contraction multifunctions have been dis-
cussed in the literature, among many others, by Assad [1], Assad and Kirk [2],
Ćirić and Ume [3]. Itoh [7] extended these results to a more general class of
contraction multifunctions while Rhoades obtained a generalization of Itoh’s
fixed point theorem (see [7]) for the case of a multivalued operator F defined
on a subset K of a metrically convex metric space X. Common fixed point
143
144 B. C. DHAGE, U. P. DOLHARE AND ADRIAN PETRUŞEL

theorems for a sequences {Fn } of non-self multivalued operators in metrically

convex metric space have been also proved by Huang and Cho [6]. All these
results use a kind of boundary condition with respect to the multivalued op-
erator F and the subset K of the metric space X, namely F (∂K) ⊂ K, where
∂K denotes the boundary of K. In a recent paper [4], one of the present au-
thors proved some fixed point theorems for the non-self multivalued operators
on a metrically convex metric space, satisfying slightly stronger condition than
Rhoades [8], but under a weaker boundary condition than that in the above
mentioned papers.
The purpose of the present paper is to prove some common fixed point the-
orems for a sequence of non-self multivalued operators on a metrically convex
metric space satisfying certain contraction type conditions and under a weaker
boundary condition. Our results extend some theorems of Dhage [4] (to a se-
quence of multivalued operators) and include the result of Huang and Cho [6]
under a slightly stronger contraction condition.

2. Main Results

Let (X, d) denote a metric space and let CB(X) denote the class of all
non-empty closed and bounded subsets of X.

Definition 2.1. A metric space (X, d) is said to be metrically convex if for

any x, y ∈ X with x 6= y, there is a z ∈ X, x 6= z, y 6= z such that

d(x, z) + d(z, y) = d(x, y).

We need the following lemma in the sequel.

Lemma 2.1. (Assad and Kirk [2]) If K is a non-empty closed convex subset
of a complete and metrically convex metric space (X, d), then for any x ∈ K
and y ∈/ K, there exists a point z ∈ ∂K (the boundary of K) such that

d(x, z) + d(z, y) = d(x, y).

For any A, B ∈ CB(X) denote:

D(A, B) = inf{d(a, d) | a ∈ A, b ∈ B},

δ(A, B) = sup{d(a, b) | a ∈ A, b ∈ B}
 SOME COMMON FIXED POINT THEOREMS 145

and n o
H(A, B) = max sup D(a, B), sup D(A, b) .
a∈A b∈B
The following properties of the functional δ are well-known (see for example
Fisher [5] and Petruşel [11]) :
(i) δ(A, B) = 0 if and only if A = B = {x∗ }
(ii) δ(A, B) = δ(B, A) and
(iii) δ(A, B) ≤ δ(A, C) + δ(C, B)
for A, B, C ∈ CB(X).
We need the following lemma in the sequel.

Lemma 2.2. Fisher [5] Let {An } and {Bn } be two sequences in CB(X) con-
verging in CB(X) to the sets A and respectively B. Then
lim δ(An , Bn ) = δ(A, B).
n→∞

If T : X → CB(X) is a multivalued operator, then FT := {x ∈ X| x ∈ T (x)}

denotes the fixed point set T , while (SF )T := {x ∈ X| {x} = T (x)} is the
strict fixed point set of T .
Now we prove our first main result.

Theorem 2.1. Let (X, d) be a complete and metrically convex metric space,
K a non-empty, closed, convex and bounded subset of X. Let {Fn }∞ n=1 be a
sequence of multivalued operators of K into CB(X) satisfying for i 6= j,
δ(Fi (x), Fj (y)) ≤ α max{d(x, y), D(x, Fi (x)), D(y, Fj (y))}
(2.1)
+β[D(x, Fj (y) + D(y, Fi (x)]
for all x, y ∈ K, where α ≥ 0, β ≥ 0 and 2α + 3β < 1.
If Fn (x) ∩ K 6= ∅ for each x ∈ ∂K and each n ∈ N, then FFn = (SF )Fn =
{z}, for each n ∈ N . Moreover, for each n ∈ N, Fn is continuous in z with
respect to the Hausdorff-Pompeiu metric on X.

Proof. Let x ∈ K be arbitrary and define a sequence {xn } ⊂ K as follows.

Let x0 = x and take a point x1 ∈ F1 (x0 ) ∩ K if F1 (x0 ) ∩ K 6= ∅, otherwise
choose a point x2 ∈ ∂K such that
d(x0 , x1 ) + d(x1 , y1 ) = d(x0 , y1 )
for some y1 ∈ F x0 ⊂ X\K.
146 B. C. DHAGE, U. P. DOLHARE AND ADRIAN PETRUŞEL

Similarly choose a point x2 ∈ F2 (x1 )∩K if F2 (x1 )∩K 6= 0, otherwise choose

a point x2 ∈ ∂K, such that

d(x1 , x2 ) + d(x2 , y2 ) = d(x1 , y2 )

for some y2 ∈ F2 (x1 ) ⊂ X\K.

Continuing in this way, choose xn ∈ Fn (xn−1 ) ∩ K if Fn (xn−1 ) ∩ K 6= ∅,
otherwise select xn ∈ ∂K such that

d(xn−1 , xn ) + d(xn , yn ) = d(xn−1 , yn )

for some yn ∈ Fn (xn−1 ) ⊂ X\K. Denote by

P = {xn ∈ {xn } | xn ∈ Fn (xn−1 ), n ∈ N }

and

Q = {xn ∈ {xn } | xn ∈ ∂K, xn ∈ Fn (xn−1 ), n ∈ N }.

Clearly

{xn } = P ∪ Q ⊂ K.

Then for any two consecutive terms xn , xn−1 of the sequence {xn }, there are
only the following three possibilities:
(i) xn , xn−1 ∈ P,
(ii) xn ∈ P, xn+1 ∈ Q,
(iii) xn ∈ Q and xn+1 ∈ P.
We will prove that {xn } is a Cauchy sequence in K.
 SOME COMMON FIXED POINT THEOREMS 147

Case I: Suppose that xn , xn+1 ∈ P. Then by(2.1), we have,

d(xn , xn+1 ) ≤ δ(Fn (xn−1 ), Fn+1 (xn ))

≤ α max{d(xn−1 , xn ), D(xn−1 , Fn (xn−1 )), D(xn , Fn+1 (xn ))}
+β[D(xn−1 , Fn+1 (xn )) + D(xn , Fn (xn−1 ))]
≤ α max{d(xn−1 , xn ), d(xn−1 , xn ), d(xn , xn+1 )}
+β[d(xn−1 , xn+1 ) + d(xn , xn )]
≤ α max{d(xn−1 , xn ), d(xn , xn+1 )}
+β[d(xn−1 , xn ) + d(xn , xn+1 )]
= max{αd(xn−1 , xn ) + β[d(xn−1 , xn ) + d(xn , xn+1 )],
αd(xn , xn+1 ) + β[d(xn−1 , xn ) + d(xn , xn+1 )]}
= max{(α + β)d(xn−1 , xn ) + βd(xn , xn+1 ),
βd(xn−1 , xn ) + (α + β)d(xn , xn+1 )},

i.e.,
d(xn , xn+1 ) ≤ kd(xn−1 , xn ), (2.2)
where  
α+β β
k = max , <1
1 − β 1 − (α + β)
since 2α + 3β < 1.
Case II: Let xn ∈ P and xn+1 ∈ Q. Then

d(xn , xn+1 ) + d(xn+1 , yn+1 ) = d(xn , yn+1 )

for some yn+1 ∈ Fn+1 (xn ) ⊂ X\K.

Clearly
(
d(xn , xn+1 ) ≤ d(xn , yn+1 ),
(2.3)
d(xn , yn+1) ≤ δ(Fn (xn−1 ), Fn+1 (xn )).
Now following the arguments similar to that in Case I,

d(xn , yn+1 ) ≤ kd(xn−1 , xn ), (2.4)

where again
 
α+β β
k = max , < 1.
1 − β 1 − (α + β)
148 B. C. DHAGE, U. P. DOLHARE AND ADRIAN PETRUŞEL

From (2.3) and(2.4), it follows that

d(xn , xn+1 ) ≤ kd(xn−1 , xn ).

Case III: Suppose that xn ∈ Q and xn+1 ∈ P. We note that then xn−1 ∈ P.
By definition of {xn }, there is a point yn ∈ Fn (xn−1 ) such that

d(xn−1 , xn ) + d(xn , yn ) = d(xn−1 , yn ). (2.5)

We have successively:
d(xn , xn+1 ) ≤ d(xn , yn ) + d(yn , xn+1 ) ≤
≤ d(xn , yn ) + δ(Fn (xn−1 ), Fn+1 (xn )) ≤ d(xn , yn )+
+α max{d(xn−1 , xn ), D(xn−1 , Fn (xn−1 )), D(xn , Fn+1 (xn−1 ))}
+β[D(xn−1 , Fn+1 (xn ) + D(xn , Fn (xn−1 ))] = d(xn , yn )+
+α max{d(xn−1 , xn ), d(xn−1 , yn ), d(xn , xn+1 )}+
+β[d(xn−1 , xn+1 ) + d(xn , yn )] ≤ d(xn , yn )+
+α max{d(xn−1 , yn ), d(xn , xn+1 )}+
+β[d(xn−1 , xn ) + d(xn , xn+1 ) + d(xn , yn )] = d(xn−1 , yn )+
+α max{d(xn−1 , yn ), d(xn , xn+1 )} + β[d(xn−1 , yn ) + d(xn , xn+1 )].
From (2.4) of Case II applied to n − 1, we have

d(xn−1 , yn ) ≤ kd(xn−2 , xn−1 ) (2.6)

and hence

d(xn , xn+1 ) ≤ kd(xn−2 , xn−1 ) + α max{kd(xn−2 , xn−1 ), d(xn , xn+1 )}

+β[kd(xn−2 , xn−1 ) + d(xn , xn+1 )]
= max{(1 + α + β)kd(xn−2 , xn−1 ) + βd(xn , xn+1 ),
(1 + β)kd(xn−2 , xn−1 ) + (α + β)d(xn , xn+1 )}.

This further implies that

 
(1+α+β)k (1+β)k
d(xn , xn+1 ) ≤ max 1−β , 1−(α+β) d(xn−2 , xn−1 )
(2.7)
= qd(xn−2 , xn−1 ),
 SOME COMMON FIXED POINT THEOREMS 149

where
 
(1 + α + β)k (1 + β)k
q = max ,
1−β 1 − (α + β)
 
1+α+β 1+β
= k max ,
1−β 1 − (α + β)
 
1+β
= k < 1.
1 − (α + β)
Now for any n ∈ N, we have

d(x2n , x2n+1 ) ≤ qd(x2n−2 , x2n−1 )

≤ q n d(x0 , x1 ).

Since n is arbitrary, one has

d(xn .xn+1 ) ≤ q n d(x0 , x1 ).

Then for any positive integer p,

n+p+1
X
d(xn , xn+p ) ≤ d(xi , xi+1 )
i=1
n+p+1
X
≤ q i d(x0 , x1 )
i=1
(1 − q n+p−1 )
= qn d(x0 , x1 ) → 0 as n → ∞.
1−q
This shows that {xn } is a Cauchy sequence in K. Since K is closed, it is
complete and there is a point z ∈ K such that lim xn = z exists. We show
that z is a fixed point of Fn . Without loss of generality, we may assume that
xn+1 ∈ Fn+1 (xn ) for some n ∈ N. Then,

δ(z, Fj (z)) ≤ δ(z, xn+1 ) + δ(xn+1 , Fj (z))

= δ(z, xn+1 ) + δ(Fj (z), Fn+1 (xn ))
= δ(z, xn+1 ) + α max{d(z, xn ), D(z, Fj (z)), D(xn , Fn+1 (xn ))}
+β[D(z, Fn+1 (xn )) + D(xn , Fj (z))]
≤ δ(z, xn+1 ) + α max{d(z, xn ), δ(z, Fj (z)), d(xn , xn+1 )}
+β[d(z, xn+1 ) + δ(xn , Fj (z))].
150 B. C. DHAGE, U. P. DOLHARE AND ADRIAN PETRUŞEL

Taking limit as n → ∞ in above inequality yields that

δ(z, Fj (z)) ≤ 0 + α max{0, δ(z, Fj (z), 0, βδ(z, Fj (z))}

= αδ(z, Fj (z)),

which implies that δ(z, Fj (z)) = 0 since α < 1, i.e., Fj (z) = {z} for each
j ∈ N.
To prove uniqueness, let z ∗ (6= z) be another common fixed point of {Fn }.
Then by(2.1) we get

d(z, z ∗ ) ≤ δ(Fi (z), Fj (z ∗ ))

≤ α max{d(z, z ∗ ), D(z, Fi (z), D(z ∗ , Fj (z ∗ ))}
+β[d(z, Fj (z ∗ )) + D(z ∗ , Fi (z))]
= (α + 2β)d(z, z ∗ ),

which is a contradiction since α + 2β ≤ 2α + 3β < 1. Hence z = z ∗ .

Finally we prove the continuity of Fn for each n ∈ N. Let {zn } be any
sequence in K converging to the unique common fixed point z of {Fn }∞ n=1 . To
conclude, it, is enough to prove that limn H(Fj (zn ), Fj z) = 1 for each i ∈ N.
We know that

(∗) H(Fj (zn ), Fj (z)) ≤ δ(Fi (zn ), Fi (z)).

Now for any i 6= j,

δ(Fi (zn ), Fi (z)) = δ(Fi (zn ), Fj (z))

≤ α max{d(zn , z), D(zn , Fi (zn ), D(z, Fj (z))}
+β[D(zn , Fj (z)) + D(z, Fi (zn ))]
≤ α max{d(zn , z), δ(zn , Fi (zn )), 0}
+β[δ(zn , Fj (z)) + δ(z, Fi (zn ))].

Taking limit as n → ∞, we get

lim δ(Fi (zn ), Fi (z)) ≤ α max{0, lim δ(Fi (zn ), Fi (z)), 0}

n n
+β[0 + lim δ(Fi (zn ), Fi (z))]
n
= (α + β) lim δ(Fi (zn ), Fi (z)),
n
 SOME COMMON FIXED POINT THEOREMS 151

which is possible only when limn δ(Fi (zn ), Fi (z)) = 0.

From (∗) it follows that lim H(Fi (zn ), Fi (z)) = 0. This completes the proof.
n→∞


Remark 2.1. With respect to condition (2.1), the following implications hold:
i) (2.1) and (x ∈ FFi ∩ FFj , i 6= j) ⇒ Fi (x) = Fj (x) = {x}.
ii) (2.1) and (x ∈ FFi , y ∈ FFj , i 6= j) ⇒ δ(Fi (x), Fj (y)) ≤
α max{d(x, y), D(x, Fi (x)), D(y, Fj (y))} + β[D(x, Fj (y) + D(y, Fi (x)] ≤
≤ (α + 2β) · δ(Fi (x), Fj (y)).
Hence δ(Fi (x), Fj (y)) = 0 and so Fi (x) = Fi (y) = {z}. In conclusion
z = x = y.
iii) (2.1) and (x ∈ FFi , y = x) ⇒ δ(Fi (x), Fj (x)) ≤ (α + β) ·
δ(Fi (x), Fj (x)).
Hence δ(Fi (x), Fj (x)) = 0 and so Fi (x) = Fj (x) = {z}. In conclusion
z = x = y.

Theorem 2.2. Let (X, d) be a metrically convex complete space, K a non-

empty, closed, convex and bounded subset of X. Let {Fn }∞n=1 be a sequence of
mappings from K into CB(X) satisfying for i 6= j,
n
δ(Fi (x), Fj (y)) ≤ αd(x, y) + β max 12 [D(x, Fi (x)) + D(y, Fj (y))],
o (2.8)
1
2 [D(x, Fj (y)) + D(y, Fi (x))]

for all x, y ∈ K, where α ≥ 0 and β ≥ 0 satisfying α2 + α + α · β + 3β

2 < 1.
If Fn (K) ∩ K 6= ∅ for each x ∈ ∂K and n ∈ N, then FFn = (SF )Fn = {x∗ },
for each n ∈ N . Moreover, for each n ∈ N, Fn is continuous in x∗ with respect
to the Hausdorff-Pompeiu metric.

Proof. Let x ∈ K be arbitrary and define a sequence {xn } ⊂ K as in the

previous proof. So {xn } = P ∪ Q, where

P = {xn ∈ {xn } | xn ∈ Fn (xn−1 ), n ∈ N }

and
Q = {xn ∈ {xn } | xn ∈ ∂K, xn ∈ Fn (xn−1 ), n ∈ N }.
We show that {xn } is a Cauchy sequence. Now for any two consecutive
terms xn , xn+1 ∈ {xn }, there are following three cases:
152 B. C. DHAGE, U. P. DOLHARE AND ADRIAN PETRUŞEL

Case I: Suppose that xn , xn+1 ∈ P. Then by (2.9) we get

d(xn , xn+1 )
≤ δ(Fn (xn−1 )Fn+1 (xn )) n
1
≤ αd(xn−1 , xn ) + β max 2 [D(xn−1 , Fn (xn−1 )) + D(xn , Fn+1 (xn ))],
o
1
2 [D(xn−1 , Fn+1 (xn−1 ))
n
+ D(xn , Fn (xn ))]
≤ αd(xn−1 , xn ) + β max 21 [d(xn−1 , xn ) + d(xn , xn+1 )]
o
1
2 max[d(x , x
n−1 n+1 ) + d(x ,
n nx )]
n
1
≤ αd(xn−1 , xn ) + β max 2 [d(xn−1 , xn ) + d(xn , xn+1 )],
o
1
2 [d(xn−1 , xn ) + d(xn , xn+1 )]
β β
≤ αd(x
 βn−1 , xn ) + 2 d(xn−1 , xn ) + 2 d(xn , xn+1 )
α+ 2
= β d(xn−1 , xn ).
1− 2
(2.9)
Case II: Suppose that xn ∈ P and xn+1 ∈ Q. Then there is a point yn+1 ∈
Fn+1 (xn ) ⊂ X\K such that

d(xn , xn+1 ) + d(xn+1 , yn+1 ) = d(xn , yn+1 ),

which further implies that

d(xn , yn+1 ) ≤ d(xn , yn+1 )

and

d(xn , yn+1 ) ≤ δ(Fn (xn−1 ), Fn+1 (xn )). (2.10)

Now
n+1 ) ≤ δ(Fn (xn−1 ), Fn+1 (xn )) ≤ αd(xn−1 , xn )+
d(xn , yn
β max 21 [D(xn−1 , Fn (xn−1 )) + D(xn , Fn+1 (xn ))], 12 [D(xn−1 , Fn+1 (xn ))+
o
+D(xn , Fn (xn−1 ))] ≤ αd(xn−1 , xn )+
n o
+β max 21 [d(xn−1 , xn ) + d(xn , xn+1 )], 21 [d(xn−1 , yn+1 ) + d(xn , xn )]
n
= αd(xn−1 , xn ) + β max 21 [d(xn−1 , xn )+
o
+d(xn , xn+1 )], 12 [d(xn−1 , xn ) + d(xn , yn+1 )] = αd(xn−1 , xn )+
+ β2 d(xn−1 , xn ) + β2 d(xn , yn+1 ) = k · d(xn−1 , xn ),
 SOME COMMON FIXED POINT THEOREMS 153

where
β
α+ 2
k= β
< 1.
1− 2
From (3) it follows that

d(xn , xn+1 ) ≤ kd(xn−1 , xn ).

Case III: Suppose that xn ∈ Q and xn+1 ∈ P. Note that xn−1 ∈ P. Then
there is a point yn ∈ Fn (xn−1 ) ⊂ X\K such that

d(xn−1 , xn ) + d(xn , yn ) = d(xn−1 , yn ),

which further implies that

d(xn , yn ) ≤ d(xn−1 , yn ). (2.11)

Now
d(xn , xn+1 )
≤ d(xn , yn ) + d(yn , xn+1 )
= d(xn , yn ) + δ(Fn (xn−1 ), Fn+1 (xn ))n
1
≤ d(xn , yn ) + αd(xn−1 , xn ) + β max 2 [D(xn−1 , Fn (xn−1 ))
o
+D(xn , Fn+1 (xn ))], 21 [D(xn−1 , Fn+1 (xn )) + D(xn , Fn (xn−1 ))]
n
= d(xn , yn ) + αd(xn−1 , xn ) + β max 21 [d(xn−1 , yn ) + d(xn , xn+1 )],
o
1
2 [d(x n−1 , x n+1 ) + d(x n , y n )]
n
≤ d(xn−1 , yn ) + αd(xn−1 , xn ) + β max 21 [d(xn−1 , yn ) + d(xn , xn+1 )],
o
1
2 [d(x n−1 , x n ) + d(x n , x n+1 ) + d(xn , y n )]
n
= d(xn−1 , yn ) + αd(xn−1 , xn ) + β max 21 [d(xn−1 , yn ) + d(xn , xn+1 )],
o
1
2 [d(x n−1 , y n ) + d(xn , xn+1 )]
= d(xn−1 , yn ) + αd(xn−1 , xn ) + β2 d(xn−1 , yn ) + β2 d(xn , xn+1 )
= kd(xn−2 , xn−1 ) + kαd(xn−2 , xn−1 ) + β2 kd(xn−2 , xn−1 )
+ β2 d(xn , xn+1 ).
  (2.12)
k+kα+k β2
It follows that d(xn , xn+1 ) ≤ d(xn−2 , xn−1 )
1− β2
 
1+α+ β2
=k β d(xn−2 , xn−1 )
1− 2
154 B. C. DHAGE, U. P. DOLHARE AND ADRIAN PETRUŞEL

= qd(xn−2 , xn−1 ), where

 β   β  β 
1+α+ 2 α+ 2 1+α+ 2
q=k = < 1.
1 − β2 1− β
2 1− β
2
Thus in all three cases one has
d(xn , xn+1 ) ≤ qd(xn−2 , xn+1 ).
Therefore
d(x2n , x2n+1 ) ≤ qd(x2n−2 , x2n−1 )
≤ q n d(x0 , x1 )
for all n ∈ N. Since n is arbitrary, we have
d(xn , xn+1 ) ≤ q n d(x0 , x1 ).
This shows that {xn } is a Cauchy sequence in K. As K is closed, it is
complete and there is a point z ∈ K such that limn xn = z.
We show that z is a fixed point of Fn . Without loss of generality, we may
assume that xn+1 ∈ Fn+1 (xn ). Then
δ(zn , Fj (z))
≤ δ(z, xn+1 ) + δ(xn+1 , Fj (z))
= δ(z, xn+1 ) + δ(Fn+1 (xn ), Fj (z))
n1
≤ d(z, xn+1 ) + αd(xn , z) + β max [D(xn , Fn+1 (xn )) + D(z, Fj (z))],
2
1 o
[D(xn , Fj (z)) + D(z, Fn+1 (xn ))]
2
n1
= d(z, xn+1 ) + αd(xn , z) + β max [d(xn , xn+1 ) + δ(z, Fj (z))],
2
1 o
[δ(xn , Fj (z)) + d(z, xn+1 ))] .
2
Taking limit as n → ∞, we get
n1 1 o
δ(z, Fj (z)) ≤ 0 + β max [0 + δ(z, Fj (z))], [δ(z, Fj (z)) + 0]
2 2
β
= δ(z, Fj (z)),
2
which is possible only when Fj (z) = {z} for j ∈ N. Again the uniqueness of
z follows from the condition (2.9). Finally we prove the continuity of Fn for
 SOME COMMON FIXED POINT THEOREMS 155

each n ∈ N. Let {zn } be any sequence in K converging to the unique common

fixed point z of {Fn }∞n=1 .
Now for any i 6= j,
δ(Fj (zn ), Fi (z))
= δ(Fi (zn ), Fj (z))
n1
≤ αd(zn , z) + β max [D(zn , Fi (zn )) + D(z, Fj (z))],
2
1 o
[D(zn , Fj (z)) + D(z, Fi (zn ))]
2
n1 1 o
≤ αd(zn , z) + β max [δ(zn , Fi (zn )) + 0], [δ(zn , Fj (z) + δ(z, Fi (zn ))] .
2 2
Taking limit as n → ∞,
β
lim δ(Fi (zn ), Fj (z)) ≤ lim δ(Fi (zn ), Fj (z)),
n 2 n
which is possible only when limn δ(Fi (zn ), Fi (z)) = 0. Since H(Fi (zn ), Fi (z)) ≤
δ(Fi (zn ), Fi (z)), we have limn H(Fi (zn ), Fi (z)) = 0 and so Fi is continuous at
z for each i ∈ N. This completes the proof. 
Now we will prove two results concerning the fixed point of sequence of
non-self maps on the subsets of a metrically convex metric space satisfying
a contraction condition more general than (2.1) and (2.9) and under certain
compactness type condition.

Theorem 2.3. Let (X, d) be a metrically convex metric space, K a compact

convex subset of X. Let {Fn }∞
n=1 be a sequence of continuous multivalued
operators from K into CB(X) satisfying for i 6= j,
δ(Fi (x), Fj (y)) < α max{d(x, y), D(x, Fi (x)), D(y, Fj (y))}
(2.13)
+β[D(x, Fj (y)) + D(y, Fi (x))]
for all x, y ∈ K with right hand side not zero, where α > 0, β > 0 and
2α + 3β ≤ 1.
If Fn (xn ) ∩ K 6= ∅ for each x ∈ ∂K and n ∈ N, then FFn = (SF )Fn = {x∗ },
for each n ∈ N .

Proof. First we note that if the sequence {Fn } of multivalued operators have
a common fixed point, then from (2.14) it follows that the common fixed point
is unique.
156 B. C. DHAGE, U. P. DOLHARE AND ADRIAN PETRUŞEL

As F is continuous and K is compact so both sides of the inequality (2.14)

are bounded on K. Now there are two cases:
Case I: Suppose that there exist some x, y ∈ K such that the right hand
side of (2.14) is zero. Then z = x = y is a common fixed point of {Fn } and so
it is unique.
Case II: Now we assume that the right hand of the inequality (2.14) is
positive on K. If 2α + 3β < 1, the desired conclusion follows from Theorem
2.1. Therefore we prove the result only in the case when 2α + 3β = 1.
Denote by

M (x, y) = α max{d(x, y), D(x, Fi (x)), D(y, Fj (y))}

+ β[D(x, Fj (y)) + D(y, Fi (x))].

Define a function T : K × K → R+ by
δ(Fi (x), Fj (y))
T (x, y) = , x, y ∈ K. (2.14)
M (x, y)

Clearly the function T is well defined, since M (x, y) 6= 0 for all x, y ∈ K.

Since each Fn is continuous, T is continuous and from compactness of K, it
follows that T attains its maximum on K × K at some point say (u, v) ∈ K 2 .
Call the value c. From (2.14), we get 0 < c < 1. By the definition of T we
obtain
δ(Fi (x), Fj (y))
≤ T (u, v) = c,
M (x, y)
i.e.,

δ(Fi (x), Fj (y)) ≤ cM (x, y)

= α0 max{d(x, y), D(x, Fi (x)), D(y, Fj (y))}, (2.15)
+β 0 [D(x, Fj (y)) + D(y, Fi (x))]

for all x, y ∈ K, where α0 ≥ 0, β 0 ≥ 0 satisfying 2α0 + 3β 0 = c(2α + 3β) < 1.

Since K is compact, it is closed and bounded. Thus all the conditions of
Theorem 2.1 are satisfied and hence an application of it yields the desired
result. 

Theorem 2.4. Let (X, d) be a metrically convex metric space, K a compact

convex subset of X. Let {Fn } be a sequence of continuous multivalued operator
 SOME COMMON FIXED POINT THEOREMS 157

from K into CB(X) satisfying for each i 6= j the following condition:

n
δ(Fi (x), Fj (y)) < αd(x, y) + β max 12 [D(x, Fi (x) + D(y, Fj (y))],
o
1
2 [D(x, F j (y) + D(y, Fi (x))]
(2.16)
for all x, y ∈ K with right hand side not zero, where α > 0, β > 0 and
2α + αβ + β ≤ 1.
If Fn (x) ∩ K 6= ∅ for each x ∈ ∂K and n ∈ N, then FFn = (SF )Fn = {x∗ },
for each n ∈ N .

Proof. The proof is similar to the Theorem 2.3 with appropriate modifi-
cations. The result follows by an application of Theorem 2.2. The proof is
complete. 

Theorem 2.5. Let (X, d) be a metrically convex metric space, K a non-empty

compact convex subset of X. Let {Fn } be a sequence of continuous multivalued
operator of K into CB(X) satisfying for all i 6= j,
n
H(Fi (x), Fj (y)) < αd(x, y) + h max 12 [D(x, Fi (x)) + D(y, Fj (y))],
o
1
2 [D(x, F j (y)) + D(y, Fi (x))]
(2.17)
for all x, y ∈ K with right hand side not zero, where α ≥ 0, h ≥ 0 and
α + 23 h + αh
2 < 1,
If Fn (x) ⊂ K for each x ∈ ∂K and n ∈ N, then {Fn } have a common fixed
point z ∈ K.

Proof. The proof is similar to the Theorem 2.3 and now the desire conclusion
follows by an application of Theorem 3.1 of Hung and Cho [6]. 

References
[1] N. A. Assad, Fixed point theorems for set-valued transformations on compact sets, Boll.
U. M. I., 17(4), (1973), 1-7.
[2] N. A. Assad and W. A. Kirk, Fixed point theorem for set-valued mappings of contractive
type, Pacific J. Math., 43(1972), 553-562.
[3] Lj. B. Ćirić, J. S. Ume, Common fixed point theorems for multi-valued non-self mappings,
Publ. Math. Debrecen, 60(2002), 359-371.
158 B. C. DHAGE, U. P. DOLHARE AND ADRIAN PETRUŞEL

[4] N. J. Huang and Y. J. Cho, Common fixed point theorems for a sequence of set-valued
mappings, Korean J. Math. Sci., 4(1997), 1-10.
[5] S. Itoh, Multi-valued generalized contractions and fixed point theorems, Comment. Math.
Univ. Carolin., 18(1977), 247-248.
[6] B. Fisher, Common fixed point and contraction mappings satisfying a rational inequality,
Math. Sem. Notes, 6(1978), 29-35.
[7] B. C. Dhage, A fixed point theorem for non-self multimaps in metrically convex metric
space, Comment. Math. Univ. Carolin., 40(1999), 251-158.
[8] A. Muntean, Fixed point principles and applications to mathematical economics, Cluj
University Press, Cluj-Napoca, 2002.
[9] N. Negoescu, Common fixed point problems for pairs of ϕ-contractive mappings, ”Gh.
Asachi” Publ. House, Iaşi, 1999, (in Romanian).
[10] B. E. Rhoades, A fixed point theorem for a multi-valued non-self mappings, Comment.
Math. Univ. Carolin., 37(1996), 401-404.
[11] Adrian Petruşel, Operatorial inclusions, House of the Book of Science, Cluj-Napoca,
2002.
[12] I. A. Rus, Generalized contractions and applications, Cluj Univ. Press, 2001.
[13] I. A. Rus, A. Petruşel, G. Petruşel, Fixed point theory 1950-2000 : Romanian contribu-
tions, House of the Book of Science, Cluj-Napoca, 2002.
[14] A. Sı̂ntămărian, Common fixed point theorems for multivalued mappings, Semin. Fixed
Point Theory Cluj-Napoca, 1(2000), 93-102.