Sie sind auf Seite 1von 6

# Int. Journal of Math. Analysis, Vol. 1, 2007, no.

## Some Coincidence Point Theorems

in Ultra Metric Spaces

## Dept. of Applied Mathematics

Acharya Nagarjuna University-Nuzvid Campus
Nuzvid-521 201, A.P., India
kprrao2004@yahoo.com
Abstract
Two coincidence point theorems for three and four self maps in a
spherically complete ultra metric space are obtained. Our theorems
generalize the theorem of Gajic.
Mathematics Subject Classification: 47H10, 54H25

## 1 Introduction and Preliminaries

Van Roovij [1] introduced the concept of ultra metric space as follows:
Definition 1.1 Let (X, d) be a metric space. If the metric d satisfies strong
triangle inequality
d(x, y) ≤ max{d(x, z), d(z, y)} for all x, y, z ∈ X,
then d is called an ultra metric on X and (X, d) is called an ultra metric space.
Definition 1.2 An ultra metric space (X, d) is said to be spherically com-
plete if every shrinking collection of balls in X has a non empty intersection.
Recently Lj Gajic [3] generalized the result of Petalas and Vidalis [2] as
follows :
Theorem 1.3 (Theorem 1, [3]): Let (X, d) be a spherically complete ultra
metric space. If T : X → X is a mapping such that
d(T x, T y) < max{d(x, y), d(x, T x), d(y, T y)} for every x, y ∈ X, x = y
then T has a unique fixed point.
In this paper we extend this theorem for three maps which are pairwise
commuting.
898 K. P. R. Rao, G. N. V. Kishore and T. Ranga Rao

2 Main Results
Theorem 2.1 Let (X, d) be an ultra metric space, f, S, T : X → X
satisfying
(1) f (X) is spherically complete,
(2) d(Sx, T y) < max{d(f x, f y), d(f x, Sx), d(f y, T y)} for x, y ∈ X, x = y,
(3) f S = Sf, f T = T f, ST = T S,
(4) S(X) ⊆ f (X), T (X) ⊆ f (X).
Then either f w = Sw or f w = T w for some w ∈ X.

Proof. For a ∈ X, let Ba = (f a; max{d(f a, Sa), d(f a, T a)}) denote the closed
sphere centered at f a with the radius max{d(f a, Sa), d(f a, T a)}.
Let A be the collection of all the spheres for all a ∈ f (X).
Then the relation Ba ≤ Bb iﬀ Bb ⊆ Ba is a partial order on A.
Now, consider a totally ordered sub family A1 of A.

Since f (X) is spherically complete, we have Ba = B = φ.
Ba ∈A1
Let f b ∈ B where b ∈ f (X) and Ba ∈ A1 . Then f b ∈ Ba . Hence
d(f b, f a) ≤ max{d(f a, Sa), d(f a, T a)} · · · · · ·(i)
If a = b then Ba = Bb . Assume that a = b.
Let x ∈ Bb . Then
d(x, f b) ≤ max{d(f b, Sb), d(f b, T b)}
≤ max{d(f b, f a), d(f a, T a), d(T a, Sb), d(f b, f a), d(f a, Sa),
d(Sa, T b)}
< max{d(f b, f a), d(f a, T a), d(f a, Sa), max{d(f b, f a), d(f b, Sb),
d(f a, T a)}, max{d(f a, f b), d(f a, Sa), d(f b, T b)}} from (2)
= max{d(f a, Sa), d(f a, T a)} from (i).
Thus d(x, f b) < max{d(f a, Sa), d(f a, T a)} · · · · · · (ii)
Now d(x, f a) ≤ max{d(x, f b), d(f b, f a)}
≤ max{d(f a, Sa), d(f a, T a)} from (i) and (ii).
Thus x ∈ Ba . Hence Bb ⊆ Ba for any Ba ∈ A1 .
Thus Bb is an upper bound in A for the family A1 and hence by Zorn’s Lemma,
there is a maximal element in A, say Bz , z ∈ f (X).There exists w ∈ X
 z = f w.
Suppose f w = Sw and f w = T w.
d(Sf w, T Sw) < max{d(f 2 w, f Sw), d(f 2w, Sf w), d(f Sw, T Sw)}
= d(f 2 w, f Sw) · · · · · · (iii) since f S = Sf .
d(ST w, T f w) < max{d(f T w, f 2 w), d(f T w, ST w), d(f 2w, T f w)}
= d(f 2 w, f T w) · · · · · · (iv) since f T = T f .
d(Sf w, S 2w) ≤ max{d(Sf w, T Sw), d(T Sw, T f w), d(T fw, S 2w)}
< max{d(f 2 w, f Sw), d(f 2w, f T w), max{d(f Sw, f 2w),
d(f Sw, S 2w), d(f 2w, T f w)}} from (iii), (iv) and ST = T S.
Some coincidence point theorems 899

## = max{d(f 2 w, f Sw), d(f 2w, f T w)} · · · · · · (v)

From (iii), (v) we have
max{d(Sf w, T Sw), d(Sf w, S 2w)} < max{d(f 2 w, f Sw), d(f 2w, f T w)} · · ·(vi)
d(T f w, T 2w) ≤ max{d(T f w, T Sw), d(T Sw, Sf w), d(Sfw, T 2w)}
< max{d(f 2 w, f T w), d(f 2w, f Sw), max{d(f 2w, f T w),
d(f 2w, Sf w), d(f T w, T 2w)}} from (iii), (iv)
= max{d(f 2 w, f T w), d(f 2w, f Sw)} · · · (vii)
From (iv), (vii) we have
max{d(ST w, T f w), d(T f w, T 2w)}
< max{d(f 2 w, f T w), d(f 2w, f Sw)} · · · (viii)
If max{d(f 2 w, f T w), d(f 2w, f Sw)} = d(f 2w, f Sw) then
from (vi), max{d(Sf w, T Sw), d(Sf w, S 2w)} < d(f 2 w, f Sw)
which gives f 2 w ∈
/ BSw . Hence f z ∈ / BSw . But f z ∈ Bz . Hence Bz ⊆ BSw .
It is a contradiction to the maximality of Bz in A, since Sw ∈ S(X) ⊆ f (X).
If max{d(f 2 w, f T w), d(f 2w, f Sw)} = d(f 2w, f T w)
then from (viii), max{d(ST w, T f w), d(T f w, T 2w)} < d(f 2 w, f T w)
which gives f 2 w ∈
/ BT w . Hence f z ∈/ BT w . But f z ∈ Bz . Hence Bz ⊆ BT w .
It is a contradiction to the maximality of Bz in A, since T w ∈ T (X) ⊆ f (X).
Hence either f w = Sw or f w = T w.

Corollary 2.2 Let (X, d) be a spherically complete ultra metric space and
S, T : X → X be commuting maps such that
d (Sx, Ty) < max{d (x, y), d (x, Sx), d (y, Ty)}
for all x, y ∈ X, with x = y.
Then either S or T has a fixed point in X.

Corollary 2.3 Let (X, d) be a spherically complete ultra metric space and
f, T : X → X be such that T(X) ⊆ f(X),
(2.3.1) d (Tx, Ty) < max{d (fx, fy), d (fx, Tx), d (fy, Ty)}
for all x, y ∈ X, with x = y.
Then there exists z ∈ Xsuch that f z = T z.
Further if f and T are coincidentally commuting at z then z is unique common
fixed point of f and T.

## Proof. Let Ba = (f a; d(f a, T a)) denote the closed sphere centered at f a

with radius d(f a, T a) and let A be collection of these spheres for all a ∈ X.
Proceeding as in Theorem 2.1 we can conclude that A has maximal element,
say Bz , z ∈ X.
Suppose f z = T z.
Since T z ∈ T (X) ⊆ f (X), there exists w ∈ X such that T z = f w.
Clearly w = z. From (2.3.1) we have
d(f w, T w) = d(T z, T w) < max{d(f z, f w), d(f z, T z), d(f w, T w)}
= d(f z, f w).
900 K. P. R. Rao, G. N. V. Kishore and T. Ranga Rao

## Thus f z ∈ / Bw . Hence Bz ⊆ Bw . It is a contradiction to the maximality of

Bz . Hence f z = T z.
Further assume that f and T are coincidentally commuting at z.
Then f 2 z = f (f z) = f T z = T f z = T (T z) = T 2 z.
Suppose f z = z. From (2.3.1) we have
d(T f z, T z) < max{d(f 2 z, f z), d(f 2 z, T f z), d(f z, T z)}
= d(T f z, T z). It is a contradiction.
Hence f z = z. Thus z = f z = T z.
Uniqueness of z follows easily from (2.3.1).

Remark 2.4 Theorem 1 of [3] follows from Corollary 2.3 with f = Identity
map.

## Finally we prove the following theorem for four maps.

Theorem 2.5 Let (X,d) be an ultra metric space and f, g, S and T be self
maps on X satisfying
(1) d (Sx, Ty ) < max{d (fx, gy), d (fx, Sx), d (gy, Ty)}
for all x, y ∈ X, with fx = gy,
(2) fS = Sf, fg = gf, fT = Tf, gS = Sg, gT = Tg, ST = TS,
(3) S(X) ⊆ f(X), T(X) ⊆ g(X),
(4) fg(X) is spherically complete and fg is one - one mapping.
Then f and S have a coincident point or g and T have a coincident point in X.

Proof. For a ∈ X, let Ba = (f ga; max{d(f ga, Sga), d(f ga, T f a)}) denote the
closed sphere with centre f ga and radius max{d(f ga, Sga), d(f ga, T f a)}.
Let A be the collection of all the spheres for all a ∈ f g(X).
Then the relation Ba ≤ Bb iﬀ Bb ⊆ Ba is a partial order on A.
Consider a totally ordered sub family A1 of A. Since f g(X) is spherically

complete, we have Ba = B = φ.
Ba ∈A1
Let f gb ∈ B where b ∈ f g(X) and Ba ∈ A1 .
Then f gb ∈ Ba . Hence
d(f gb, f ga) ≤ max{d(f ga, Sga), d(f ga, T f a)}
If a = b then Ba = Bb . Assume that a = b.
Since f g is one - one, we have f ga = f gb.
Let x ∈ Bb . Then
d(x, f gb) ≤ max{d(f gb, Sgb), d(f gb, T f b)}
≤ max{d(f gb, f ga), d(f ga, T f a), d(T f a, Sgb), d(f gb, f ga),
d(f ga, Sga), d(Sga, T f b)}
< max{d(f ga, Sga), d(f ga, T f a),
max{d(f gb, gf a), d(f gb, Sgb), d(gf a, T f a)},
max{d(f ga, gf b), d(f ga, Sga), d(gf b, T f b)}}
Some coincidence point theorems 901

= max{d(f ga, Sga), d(f ga, T f a)} from (i), (1) and (2).
Now d(x, f ga) ≤ max{d(x, f gb), d(f gb, f ga)}
≤ max{d(f ga, Sga), d(f ga, T f a)}
Thus x ∈ Ba . Hence Bb ⊆ Ba for every Bb ∈ A1 .
Thus Bb is an upper bound in A for the family A1 and hence by Zorn’s lemma,
there is a maximal element in A, say Bz , z ∈ f g(X).There exists w ∈ X such
that z = f gw.
Suppose S(gf gw) = f (gf gw) and T (f f gw) = g(f f gw).
From (1) we have
d(Sgf gw, T Sf gw) < max{d(f gf gw, gSf gw), d(f gf gw, Sgf gw),
d(gSf gw, T Sf gw)}
= d(f gf gw, gSf gw) · · · · · · (ii)
d(ST f gw, T f f gw) < max{d(f T f gw, gf f gw), d(f T fgw, ST f gw),
d(gf f gw, T f f gw)}
= d(f T f gw, gf f gw) · · · · · · (iii)
d(f ggSw, SggSw) ≤ max{d(f ggSw, T Sf gw), d(T Sf gw, T ff gw),
d(T f f gw, SggSw)}
< max{d(f gf gw, gSf gw), d(f T fgw, gf fgw),
max{d(f ggSw, gf f gw), d(f ggSw, SggSw),
d(gf f gw, T f f gw)}} from (ii), (iii), (1),(2).
= max{d(f gf gw, gSf gw), d(f T fgw, gf fgw)} · · · · · · (iv)
d(T f f gw, T f f T w) ≤ max{d(T f f gw, ST f gw), d(ST f gw, Sgf gw),
d(Sgf gw, T f f T w)}
< max{d(f T f gw, gf f gw), d(f gf gw, gSf gw),
max{d(f gf gw, gf f T w), d(f gf gw, Sgf gw),
d(gf f T w, T f f T w)}} from (ii),(iii),(1),(2)
= max{d(f T f gw, gf f gw), d(f gf gw, gSf gw)} · · · · · · (v)
From (ii), (iv) we have
max{d(Sgf gw, T Sf gw), d(f ggSw, SggSw)}
< max{d(f gf gw, gSf gw), d(f T f gw, gf fgw)} · · · · · · (vi)
From (iii), (v) we have
max{d(ST f gw, T f f gw), d(T f f gw, T ff T w)}
< max{d(f gf gw, gSf gw), d(f T f gw, gf fgw)} · · · · · · (vii)
If max{d(f gf gw, gSf gw), d(f T f gw, gf fgw)} = d(f gf gw, gSf gw),
then from (vi), f gf gw ∈ / BgSw ⇒ f gz ∈ / BgSw .
Hence Bz ⊂ BgSw . It is a contradiction to the maximality of Bz in A,
since gSw ⊆ gf (X) = f g(X).
If max{d(f gf gw, gSf gw), d(f T f gw, gf fgw)} = d(f T f gw, gf f gw)
then from (vii), f gf gw ∈ / Bf T w ⇒ f gz ∈ / Bf T w . Hence Bz ⊂ Bf T w . It is a
contradiction to the maximality of Bz in A , since f T w ⊆ f g(X).
Hence either S(gf gw) = f (gf gw) or T (f f gw) = g(f f gw).
Thus either f and S or g and T have a coincident point in X.
902 K. P. R. Rao, G. N. V. Kishore and T. Ranga Rao

Remark 2.6 In Theorem 2.1 we are not using the condition f is one - one.
So naturally we have the following open problem.

## Problem 2.7 : Is Theorem 2.5 true if the condition “f g is one - one” is

dropped?

References
[1] A.C.M.Van Roovij, Non-Archimedean functional analysis,Marcel Dekker,
Newyork, 1978.

## [2] C.Petalas, F.Vidalis, A ﬁxed point theorem in non Archimedaen vector

spaces, Proc.Amer.Math.Soc., 118 (1993), 819-821.

[3] Ljiljana Gajic, On ultra metric spaces, Novi Sad J.Math., 31, 2 (2001),
69-71.