Applied Mathematics Letters 24 (2011) 17911794

Contents lists available at ScienceDirect

Applied Mathematics Letters
journal homepage: www.elsevier.com/locate/aml
Fixed point theory for set-valued quasi-contraction maps in metric
spaces
A. Amini-Harandi

Department of Mathematics, University of Shahrekord, Shahrekord, 88186-34141, Iran

School of Mathematics, Institute for Research in Fundamental Sciences (IPM), P.O. Box: 19395-5746, Tehran, Iran
a r t i c l e i n f o
Article history:
Received 5 September 2010
Received in revised form 1 March 2011
Accepted 22 April 2011
Keywords:
Fixed points
Set-valued quasi-contraction maps
a b s t r a c t
Inthis paper, we give a fixed point theoremfor set-valuedquasi-contractionmaps inmetric
spaces. Our main result improves some well-known results from the literature.
2011 Elsevier Ltd. All rights reserved.
1. Introduction and preliminaries
Let (X, d) be a metric space. We denote the family of all nonempty closed and bounded subsets of X by CB(X). A mapping
T : X X is said to be a quasi-contraction if there exists k < 1 such that
d(Tx, Ty) k max{d(x, y), d(x, Tx), d(y, Ty), d(x, Ty), d(y, Tx)},
for any x, y X. In 1974, iri [1] introduced these maps and proved an existence and uniqueness fixed point theorem. In
1969, Nadler [2] extended the Banach Contraction Principle to set-valued mappings. Since then many authors have studied
fixed points for set-valued maps. Among many other works, see for example [35], and the references therein. Our aim in
this paper is to prove a set-valued version of the above mentioned iri theorem.
2. The main results
Now, we introduce the concept of a set-valued quasi-contraction in metric spaces.
Definition 2.1. Let (X, d) be a metric space. The set-valued map T : X CB(X) is said to be a k-set-valued quasi-contraction
if
H(Tx, Ty) k max{d(x, y), d(x, Tx), d(y, Ty), d(x, Ty), d(y, Tx)},
for any x, y X, where 0 k < 1 and H denotes the Hausdorff metric on CB(X) induced by d, that is,
H(A, B) = max{sup
xA
d(x, B), sup
yB
d(y, A)},
for all A, B CB(X).

Corresponding address: Department of Mathematics, University of Shahrekord, Shahrekord, 88186-34141, Iran. Tel.: +98 381 4424419; fax: +98 381
4424419.
E-mail address: [email protected].
0893-9659/$ see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.aml.2011.04.033
1792 A. Amini-Harandi / Applied Mathematics Letters 24 (2011) 17911794
Now, we are ready to state our main result.
Theorem 2.2. Let (X, d) be a complete metric space. Let T : X CB(X) be a k-set-valued quasi-contraction with k <
1
2
. Then,
T has a fixed point.
Proof. Notice first that for each A, B CB(X), a A and > 0 with H(A, B) < , there exists b B such that d(a, b) < .
Now, let c > 0 be such that k < c <
1
2
. Then
H(Tx, Ty) < c max{d(x, y), d(x, Tx), d(y, Ty), d(x, Ty), d(y, Tx)},
for any x = y X. Suppose that x
0
X and that x
1
X. Then by induction and by the above observation, we can find a
sequence {x
n
} in X such that x
n+1
Tx
n
and
d(x
n+1
, x
n
) < c max{d(x
n
, x
n1
), d(x
n
, Tx
n
), d(x
n1
, Tx
n1
), d(x
n
, Tx
n1
), d(x
n1
, Tx
n
)},
for each n N (note that if x
n
= x
n+1
for some n N, then x
n
= x
n+1
Tx
n
, that is, x
n
is a fixed point of T and we are
finished. So, we may assume that d(x
n+1
, x
n
) > 0 for each n N). Hence from the above, we get (note that x
n
Tx
n1
and
x
n+1
Tx
n
)
d(x
n+1
, x
n
) < c max{d(x
n
, x
n1
), d(x
n
, Tx
n
), d(x
n1
, Tx
n1
), d(x
n
, Tx
n1
), d(x
n1
, Tx
n
)}
= c max{d(x
n
, x
n1
), d(x
n1
, x
n+1
)}.
Thus for each n N,
d(x
n+1
, x
n
) < c max{d(x
n
, x
n1
), d(x
n1
, x
n+1
)}. (2.1)
Now, we show that {x
n
} is bounded. To prove the claim, we modify the technique in [6]. Put
1
= d(x
0
, Tx
0
). From (2.1) we
have two cases:
d(x
1
, x
2
) cd(x
0
, x
1
) or d(x
1
, x
2
) cd(x
0
, x
2
).
Let us assume that d(x
1
, x
2
) cd(x
0
, x
1
). Since d(x
0
, x
2
) d(x
0
, x
1
) +d(x
1
, x
2
), we then have
d(x
0
, x
2
) (1 +c)d(x
0
, x
1
)
_
1 +
c
1 c
_
d(x
0
, x
1
).
If d(x
1
, x
2
) cd(x
0
, x
2
), then from the inequality d(x
0
, x
2
) d(x
0
, x
1
) +d(x
1
, x
2
), we get
d(x
0
, x
2
)
1
1 c
d(x
0
, x
1
)
1
1
c
1c
d(x
0
, x
1
).
Put
2
=
1+
c
1c
1
c
1c
d(x
0
, x
1
). Note that in both cases

1

2
and d(x
0
, x
2
)
2
.
Now by induction we show that for each n 2 there exists 1 m n such that
d(x
n
, x
n1
)
_
c
1 c
_
n1
d(x
0
, x
m
). (2.2)
If n = 2, then
d(x
1
, x
2
) c max{d(x
0
, x
1
), d(x
0
, x
2
)} = cd(x
0
, x
m
),
for some 1 m 2. Thus (2.2) holds for n = 2. Suppose that (2.2) holds for each k < n and we showthat it holds for k = n.
From (2.1), we have
d(x
n
, x
n1
) < c max{d(x
n1
, x
n2
), d(x
n2
, x
n
)}.
It is trivial that (2.2) holds if d(x
n
, x
n1
) < cd(x
n1
, x
n2
). Now suppose that d(x
n
, x
n1
) < cd(x
n2
, x
n
). Then by our
assumption, we have
d(x
n2
, x
n
) d(x
n
, x
n1
) +d(x
n1
, x
n2
) cd(x
n2
, x
n
) +
_
c
1 c
_
n2
d(x
0
, x
m
),
for some 1 m n 1. Hence
d(x
n2
, x
n
)
1
1 c
_
c
1 c
_
n2
d(x
0
, x
m
),
A. Amini-Harandi / Applied Mathematics Letters 24 (2011) 17911794 1793
and so
d(x
n
, x
n1
)
_
c
1 c
_
n1
d(x
0
, x
m
),
for some 1 m n 1. Now, by induction we define the nondecreasing sequence {
n
} such that
max{d(x
0
, x
m
) : 1 m n}
n
.
From (2.2) we have
d(x
0
, x
n
) d(x
0
, x
n1
) +d(x
n1
, x
n
) d(x
0
, x
n1
) +
_
c
1 c
_
n1
d(x
0
, x
m
),
for some 1 m n. If n > m, then
d(x
0
, x
n
) d(x
0
, x
n1
) +
_
c
1 c
_
n1
d(x
0
, x
m
)
_
1 +
_
c
1 c
_
n1
_

n1
.
If n = m, then (note that d(x
0
, x
n
) d(x
0
, x
n1
) +
_
c
1c
_
n1
d(x
0
, x
n
))
d(x
0
, x
n
)
1
1
_
c
1c
_
n1
d(x
0
, x
n1
)
1
1
_
c
1c
_
n1

n1
.
Suppose that

n
=
1 +
_
c
1c
_
n1
1
_
c
1c
_
n1

n1
.
Note that
n1

n
and in both cases
d(x
0
, x
n
)
n
.
The sequence {x
n
} is bounded if and only if
= lim
n

n
=

n=1
_
1 +
_
c
1c
_
n1
_

n=1
_
1
_
c
1c
_
n1
_
< .
Since the series

n=1
(1 +(
c
1c
)
n1
) and

n=1
(1 (
c
1c
)
n1
) are convergent (note that
c
1c
< 1), we then have

n=1
_
1 +
_
c
1 c
_
n1
_
< and

n=1
_
1
_
c
1 c
_
n1
_
> 0,
and so < . Now, we show that {x
n
} is a Cauchy sequence. Suppose that M = sup{d(x
m
, x
n
) : m, n N}. From (2.2), we
have
d(x
n
, x
n1
)
_
c
1 c
_
n1
M.
Thus for m < n, we have
d(x
n
, x
m
)
n1

k=m
d(x
k
, x
k+1
)
n1

k=m
_
c
1 c
_
k
M < ,
for sufficiently large m, n (note that the series

k=1
_
c
1c
_
k
is convergent). Thus {x
n
} is a Cauchy sequence. Since (X, d) is
complete, then there exists a z X such that lim
n
x
n
= z. Then (note that x
n+1
Tx
n
)
d(z, Tz) = lim
n
d(x
n+1
, Tz) lim
n
H(Tx
n
, Tz) lim
n
c max{d(x
n
, z), d(x
n
, Tx
n
), d(z, Tz), d(x
n
, Tz), d(z, Tx
n
)}
lim
n
c max{d(x
n
, z), d(x
n
, x
n+1
), d(z, Tz), d(x
n
, Tz), d(z, x
n+1
)}
= cd(z, Tz).
Therefore d(z, Tz) = 0 and so z Tz (note that Tz is closed).
1794 A. Amini-Harandi / Applied Mathematics Letters 24 (2011) 17911794
The following question naturally arises:
Question 2.3. Does the conclusion of Theorem 2.2 remain true for any
1
2
k < 1?
Remark 2.4. Many authors have studied fixed point theory for the following kinds of set-valued maps:
The set-valued map T : X CB(X) is said to be weakly contractive if there exists 0 k < 1 such that
H(Tx, Ty) k max
_
d(x, y), d(x, Tx), d(y, Ty),
d(x, Ty) +d(y, Tx)
2
_
,
for any x = y X. Notice that since
d(x, Ty) +d(y, Tx)
2
max{d(x, Ty), d(y, Tx)},
we thenhave that every weak contractive set-valuedmapis quasi-contractive. So, for k <
1
2
, Theorem2.2improves Corollary
3.3 of [4] and Theorem 3.3 of [5].
Remark 2.5. Using arguments in [79], coincidence and common fixed point results can be obtained and also invariant
approximation results can be derived for k-set-valued quasi-I-contractions.
Acknowledgments
The author is grateful to the referee for his/her helpful comments leading to improvement of the presentation of the
work. This work was supported by the University of Shahrekord. The author would like to express thanks for this support.
The author was also partially supported by the Center of Excellence for Mathematics, University of Shahrekord. This research
was in part supported by a grant from IPM (No. 89470016).
References
[1] L.B. iri, A generalization of Banachs contraction principle, Proc. Amer. Math. Soc. 45 (2) (1974) 267273.
[2] S.B. Nadler, Multi-valued contraction mappings, Pacific J. Math. 30 (1969) 475488.
[3] N. Mizoguchi, W. Takahashi, Fixed point theorems for multivalued mappings on complete metric spaces, J. Math. Anal. Appl. 141 (1989) 177188.
[4] B. Djafari Rouhani, S. Moradi, Common fixed point of multivalued generalized -weak contractive mappings, Fixed Point Theory and Applications,
vol. 2010, Article ID 708984, 13 pages.
[5] P.Z. Daffer, H. Kaneko, Fixed points of generalized contractive multi-valued mappings, J. Math. Anal. Appl. 192 (1995) 655666.
[6] Sh. Rezapour, R.H. Haghi, N. Shahzad, Some notes on fixed points of quasi-contraction maps, Appl. Math. Lett. 23 (4) (2010) 498502.
[7] M.A. Al-Thagafi, N. Shahzad, Coincidence points, generalized I-nonexpansive multimaps, and applications, Nonlinear Anal. 67 (2007) 21802188.
[8] D. ORegan, N. Shahzad, Invariant approximations for generalized I-contractions, Numer. Funct. Anal. Optim. 26 (2005) 565575.
[9] D. ORegan, N. Shahzad, Coincidence points and invariant approximation results for multimaps, Acta Math. Sin. (Engl. Ser.) 23 (2007) 16011610.