FUNCTIONAL CONTRACTIONS IN

LOCAL BRANCIARI METRIC SPACES

arXiv:1208.4610v1 [math.GN] 22 Aug 2012

MIHAI TURINICI

Abstract. A fixed point result is given for a class of functional contractions

over local Branciari metric spaces. It extends some contributions in the area
due to Fora et al [Mat. Vesnik, 61 (2009), 203-208].

1. Introduction
Let X be a nonempty set. By a symmetric over X we shall mean any map
d : X × X → R+ with (cf. Hicks [11])
(a01) d(x, y) = d(y, x), ∀x, y ∈ X (d is symmetric);
the couple (X, d) will be referred to as a symmetric space. Further, let T : X → X
be a selfmap of X, and put Fix(T ) = {z ∈ X; z = T z}; any such point will be
called fixed under T . According to Rus [24, Ch 2, Sect 2.2], we say that x ∈ X is a
Picard point (modulo (d, T )) if 1a) (T n x; n ≥ 0) is d-convergent, 1b) each point of
limn (T n x) is in Fix(T ). If this happens for each x ∈ X then T is referred to as a
Picard operator (modulo d); if (in addition) Fix(T ) is singleton (x, y ∈ Fix(T ) =⇒
x = y), then T is called a strong Picard operator (modulo d). [We refer to Section
2 for all unexplained notions]. Sufficient conditions for these properties to be valid
require some additional conditions upon d; the usual ones are
(a02) d(x, y) = 0 iff x = y (d is reflexive sufficient)
(a03) d(x, y) ≤ d(x, z) + d(z, y), ∀x, y, z ∈ X (d is triangular);
when both these hold, d is called a (standard) metric on X. In this (classical)
setting, a basic result to the question we deal with is the 1922 one due to Banach
[3]; it says that, whenever (X, d) is complete and (for some λ in [0, 1[)
(a04) d(T x, T y) ≤ λd(x, y), ∀x, y ∈ X,
then, T is a strong Picard operator. This result found various applications in
operator equations theory; so, it was the subject of many extensions. A natural
way of doing this is by considering ”functional” contractive conditions like
(a05) d(T x, T y) ≤ F (d(x, y), d(x, T x), d(y, T y), d(x, T y), d(y, T x)), ∀x, y ∈ X;
5
where F : R+ → R+ is an appropriate function. For more details about the pos-
sible choices of F we refer to the 1977 paper by Rhoades [23]; see also Turinici
[28]. Another way of extension is that of conditions imposed upon d being mod-
ified. For example, in the class of symmetric spaces, a relevant paper concerning
the contractive question is the 2005 one due to Zhu et al [29]. Here, we shall be

2010 Mathematics Subject Classification. 47H10 (Primary), 54H25 (Secondary).

Key words and phrases. Symmetric, polygonal inequality, local Branciari metric, conver-
gent/Cauchy sequence, Matkowski function, contraction, periodic and fixed point.
1
2 MIHAI TURINICI

interested in fixed point results established over generalized metric spaces, intro-
duced as in Branciari [5]; where, the triangular property (a03) is to be substituted
by the tetrahedral one:
(a06) d(x, y) ≤ d(x, u) + d(u, v) + d(v, y),
whenever x, y, u, v ∈ X are distinct to each other.
Some pioneering results in the area were given by Das [7], Miheţ [21], and Samet
[25]; see also Azam and Arshad [2]. In parallel to such developments, some technical
problems involving these structures were considered. For example, Sarma et al [27]
observed that Branciari’s result may not hold, in view of the Hausdorff property
for (X, d) being not deductible from (a06). This remark was followed by a series of
results founded on this property being ab initio imposed; see, in this direction, Chen
and Sun [6] or Lakzian and Samet [18]. However, in 2011, Kikina and Kikina [16]
noticed that such a regularity condition is ultimately superfluous for the ambient
space; so, the initial setting will suffice for these results being retainable. It is our
aim in the present exposition to confirm this, within a class of ”local” Branciari
metric spaces. Further aspects will be delineated elsewhere.

2. Preliminaries
Let N = {0, 1, ...} denote the set of all natural numbers. For each n ≥ 1 in N ,
let N (n, >) := {0, ..., n − 1} stand for the initial interval (in N ) induced by n. Any
set P with P ∼ N (in the sense: there exists a bijection from P to N ) will be
referred to as effectively denumerable; also denoted as: card(P ) = ℵ0 . In addition,
given some n ≥ 1, any set Q with Q ∼ N (n, >) will be said to be n-finite; and we
write this: card(Q) = n; when n is generic here, we say that Q is finite. Finally,
the (nonempty) set Y is called (at most) denumerable iff it is either effectively
denumerable or finite.
(A) Let (X, d) be a symmetric space. Given k ≥ 1, any ordered system C =
(x1 , ..., xk ) in X k will be called a k-chain of X; the class of all these will be re-
denoted as chain(X; k). Given such an object, put [C] = {x1 , ..., xk }. If card([C]) =
k, then C will be referred to as a regular k-chain (in X); denote the class of all these
as rchain(X; k). In particular, any point a ∈ X may be identified with a regular
1-chain of X. For any C ∈ chain(X; k), where k ≥ 2, denote
Λ(C) = d(x1 , x2 ) + ... + d(xk−1 , xk ), whenever C = (x1 , ..., xk )
(the ”length” of C). Given h ≥ 1 and the h-chain D = (y1 , ..., yh ) in X, let (C; D)
stand for the (k + h)-chain E = (z1 , ..., zk+h ) in X introduced as
zi = xi , 1 ≤ i ≤ k; zk+j = yj , 1 ≤ j ≤ h;
it will be referred to as the ”product” between C and D. This operation may be
extended to a finite family of such objects.
Having these precise, let us say that the symmetric d is a local Branciari metric
when it is reflexive sufficient and has the property: for each effectively denumerable
M ⊆ X, there exists k = k(M ) ≥ 1 such that
(b01) d(x, y) ≤ Λ(x; C; y), for all x, y ∈ M , x 6= y, and
all C ∈ rchain(M ; k), with (x; C; y) ∈ rchain(M ; 2 + k)
(referred to as: the (2 + k)-polyhedral inequality). Note that, the triangular in-
equality (a03) and the tetrahedral inequality (a06) are particular cases of this one,
corresponding to k = 1 and k = 2, respectively. On the other hand, (b01) is not
 FUNCTIONAL CONTRACTIONS IN LOCAL BRANCIARI METRIC SPACES 3

reducible to (a03) or (a06); because, aside from k > 2 being allowed, the index in
question depends on each effectively denumerable subset M of X.
Suppose that we introduced such an object. Define a d-convergence structure
over X as follows. Given the sequence (xn ) in X and the point x ∈ X, we say that
d
(xn ), d-converges to x (written as: xn −→ x) provided d(xn , x) → 0; i.e.,
(b02) ∀ε > 0, ∃i = i(ε): n ≥ i =⇒ d(xn , x) < ε.
(This concept meets the standard requirements in Kasahara [14]; we do not give
details). The set of all such points x will be denoted limn (xn ); when it is nonempty,
(xn ) is called d-convergent. Note that, in this last case, limn (xn ) may be not a
singleton, even if (a06) holds; cf. Samet [26]. Further, call the sequence (xn ),
d-Cauchy when d(xm , xn ) → 0 as m, n → ∞, m < n; i.e.,
(b03) ∀ε > 0, ∃j = j(ε): j ≤ m < n =⇒ d(xm , xn ) < ε.
Clearly, a necessary condition for this is
d(xm , xm+i ) → 0 as m → ∞, for each i > 0;
referred to as: (xn ) is d-semi-Cauchy; but the converse is not in general true. Note
that, by the adopted setting, a d-convergent sequence need not be d-Cauchy. even
if d is tetrahedral; see the quoted paper for details. Despite of this, (X, d) is called
complete, if each d-Cauchy sequence is d-convergent.
(B) As already precise, the (nonempty) set of limit points for a convergent
sequence is not in general a singleton. However, in the usual (metric) fixed point
arguments, the convergence property of this sequence comes from the d-Cauchy
property of the same. So, we may ask whether this supplementary condition upon
(xn ) will suffice for such a property. Call (X, d), Cauchy-separated if, for each
d-convergent d-Cauchy sequence (xn ) in X, limn (xn ) is a singleton.
Proposition 1. Assume that d is a local Branciari metric (see above). Then,
(X, d) is Cauchy-separated.
Proof. Let (xn ) be a d-convergent d-Cauchy sequence. Assume by contradiction
that limn (xn ) has at least two distinct points:
d d
(b04) ∃u, v ∈ X with u 6= v, such that xn −→ u, xn −→ v.
i) Denote A = {n ∈ N ; xn = u}, B = {n ∈ N ; xn = v}. We claim that both A
and B are finite. In fact, if A is effectively denumerable, then A = {n(j); j ≥ 0},
where (n(j); j ≥ 0) is strictly ascending (hence n(j) → ∞ as j → ∞) and xn(j) = u,
∀j ≥ 0. Since, on the other hand, xn(j) → v as j → ∞, we must have d(u, v) = 0; so
that, u = v, contradiction. An identical reasoning is applicable when B is effectively
denumerable; hence the claim. As a consequence, there exists p ∈ N , such that:
xn 6= u, xn 6= v, for all n ≥ p. Without loss, one may assume that p = 0; i.e.,
{xn ; n ≥ 0} ∩ {u, v} = ∅ [xn 6= u and xn 6= v, for all n ≥ 0]. (2.1)
ii) Put h(0) = 0. We claim that the set S0 = {n ∈ N ; xn = xh(0) } is finite.
For, otherwise, it has the representation S0 = {m(j); j ≥ 0}, where (m(j); j ≥ 0)
is strictly ascending (hence m(j) → ∞ as j → ∞) and xm(j) = x0 , ∀j ≥ 0.
Combining with (b04) gives x0 = u, x0 = v; hence, u = v, contradiction. As a
consequence of this, there exists h(1) > h(0) with xh(1) 6= xh(0) . Further, by a very
similar reasoning, S0,1 = {n ∈ N ; xn ∈ {xh(0) , xh(1) }} is finite too; hence, there
4 MIHAI TURINICI

exists h(2) > h(1) with xh(2) ∈/ {xh(0) , xh(1) }; and so on. By induction, we get a
subsequence (yn := xh(n) ; n ≥ 0) of (xn ) with
d d
yi 6= yj , for i 6= j; yn −→ u, yn −→ v as n → ∞. (2.2)
The subset M = {yn ; n ≥ 0} ∪ {u, v} is effectively denumerable; let k = k(M ) ≥ 1
stand for the natural number assured by the local Branciari metric property of d.
From the (2 + k)-polyhedral inequality (b01) we have, for each n ≥ 0,
d(u, v) ≤ d(u, yn+1 ) + ... + d(yn+k , v).
(The possibility of writing this is assured by (2.1) and (2.2) above). On the other
hand, (yn ) is a d-Cauchy sequence; because, so is (xn ); hence d(ym , ym+1 ) → 0 as
m → ∞. Passing to limit in the above relation gives d(u, v) = 0; whence, u = v,
contradiction. So, (b04) is not acceptable; and this concludes the argument. 

(B) Let F (R+ ) stand for the class of all functions ϕ : R+ → R+ . Denote
(b05) Fr (R+ ) = {ϕ ∈ F (R+ ); ϕ(0) = 0; ϕ(t) < t, ∀t > 0};
each ϕ ∈ Fr (R+ ) will be referred to as regressive. Note that, for any such function,
∀u, v ∈ R+ : v ≤ ϕ(max{u, v}) =⇒ v ≤ ϕ(u). (2.3)
Call ϕ ∈ Fr (R+ ), strongly regressive, provided
(b06) ∀γ > 0, ∃β ∈]0, γ[, (∀t): γ ≤ t < γ + β =⇒ ϕ(t) ≤ γ;
or, equivalently: 0 ≤ t < γ + β =⇒ ϕ(t) ≤ γ.
Some basic properties of such functions are given below.
Proposition 2. Let ϕ ∈ Fr (R+ ) be strongly regressive. Then,
i) for each sequence (rn ; n ≥ 0) in R+ with rn+1 ≤ ϕ(rn ), ∀n, we have rn → 0
[we then say that ϕ is iteratively asymptotic]
ii) in addition, for each sequence (sn ; n ≥ 0) in R+ with sn+1 ≤ ϕ(max{sn , rn }),
∀n we have sn → 0.
Proof. i) Let (rn ; n ≥ 0) be as in the premise of this assertion. As ϕ is regressive,
we have rn+1 ≤ rn , ∀n. The sequence (rn ; n ≥ 0) is therefore decreasing; hence
γ := limn (rn ) exists in R+ . Assume by contradiction that γ > 0; and let β ∈]0, γ[
be the number indicated by the strong regressiveness of ϕ. As rn ≥ γ > 0 (and
ϕ=regressive), one gets rn+1 < rn , ∀n; hence, rn > γ, ∀n. Further, as rn → γ, there
exists some rank n(β) in such a way that (combining with the above) n ≥ n(β) =⇒
γ < rn < γ + β. The strong regressiveness of ϕ then gives (for the same ranks, n)
γ < rn+1 ≤ ϕ(rn ) ≤ γ; contradiction. Consequently, γ = 0; and we are done.
ii) Let (rn ; n ≥ 0) and (sn ; n ≥ 0) be as in the premise of these assertions.
Denote for simplicity tn := max{sn , rn }, n ≥ 0. As ϕ is regressive, one has (for
each n) rn+1 ≤ rn ≤ tn , sn+1 ≤ tn ; hence (for the same ranks) tn+1 ≤ tn . The
sequence (tn ; n ≥ 0) is therefore descending; wherefrom, t := limn (tn ) exists in R+
and tn ≥ t, ∀n. Assume by contradiction that t > 0; As rn → 0, there must be
some rank n(t) such that n ≥ n(t) =⇒ rn < t. Combining with the above, one gets
tn ≥ t > rn , for all n ≥ n(t); whence tn = sn , for all n ≥ n(t). But then, the choice
of (sn ; n ≥ 0) gives sn+1 ≤ ϕ(sn ), for all n ≥ n(t). This, along with the first part of
the proof, gives sn → 0; hence tn → 0; contradiction. Consequently, tn → 0; and,
from this, the conclusion follows. 
 FUNCTIONAL CONTRACTIONS IN LOCAL BRANCIARI METRIC SPACES 5

Now, let us give two basic examples of such functions.

B1) Suppose that ϕ ∈ Fr (R+ ) is a Boyd-Wong function [4]; i.e.
(b07) lim supt→s+ ϕ(t) < s, for all s > 0.
Then, ϕ is strongly regressive. The verification is immediate, by definition; so, we
do not give details.
B2) Suppose that ϕ ∈ Fr (R+ ) is a Matkowski function [20]; i.e.
(b08) ϕ is increasing; ϕn (t) → 0 as n → ∞, for all t > 0.
[Here, for each n ≥ 0, ϕn stands for the n-th iterate of ϕ]. Then, ϕ is strongly
regressive. The verification of this assertion is to be found to Jachymski [12];
however, for completeness reasons, we shall provide it, with some modifications.
Assume by contradiction that ϕ is not strongly regressive; that is (for some γ > 0)
∀β ∈]0, γ[, ∃t ∈ [γ, γ + β[: ϕ(t) > γ (hence, γ < t < γ + β).
As ϕ=increasing, this yields ϕ(t) > γ, ∀t > γ. By induction, we get ϕn (t) > γ, for
all n, and all t > γ. Fixing some t > γ, we have (passing to limit as n → ∞) 0 ≥ γ,
contradiction; hence the claim.

3. Main result
Let X be a nonempty set; and d(., .) be a local Branciari metric over it, with
(c01) (X, d) is complete (each d-Cauchy sequence is d-convergent).
Note that, by Proposition 1, for each d-Cauchy sequence (xn ) in X, limn (xn ) is a
(nonempty) singleton, {z}; as usually, we write limn (xn ) = {z} as limn (xn ) = z.
Let T : X → X be a selfmap of X. We say that x ∈ X is a Picard point (modulo
(d, T )) if 3a) (T n x; n ≥ 0) is d-Cauchy (hence d-convergent), ii) limn (T n x) is in
Fix(T ). If this happens for each x ∈ X then T is referred to as a Picard operator
(modulo d); if (in addition) Fix(T ) is a singleton, then T is called a strong Picard
operator (modulo d).
Now, concrete circumstances guaranteeing such properties involve functional
contractive (modulo d) conditions upon T . Precisely, denote for x, y ∈ X:
(c02) M (x, y) = max{d(x, y), d(x, T x), d(y, T y)}.
It is easy to see that
M (x, T x) = max{d(x, T x), d(T x, T 2 x)}, ∀x, y ∈ X. (3.1)
Given ϕ ∈ Fr (R+ ), we say that T is (d, M ; ϕ)-contractive if
(c03) d(T x, T y) ≤ ϕ(M (x, y)), ∀x, y ∈ X.
The main result of this note is
Theorem 1. Suppose that T is (d, M ; ϕ)-contractive, where ϕ ∈ Fr (R+ ) is strongly
regressive. Then, T is a strong Picard operator (modulo d).
Proof. First, we check the singleton property. Let z1 , z2 ∈ Fix(T ) be arbitrary
fixed. By this very choice,
M (z1 , z2 ) = max{d(z1 , z2 ), 0, 0} = d(z1 , z2 ).
Combining with the contractive condition yields
d(z1 , z2 ) = d(T z1 , T z2 ) ≤ ϕ(d(z1 , z2 ));
6 MIHAI TURINICI

wherefrom d(z1 , z2 ) = 0; hence z1 = z2 ; so that, Fix(T ) is (at most) a singleton.

It remains now to establish the Picard property. Fix some x0 ∈ X; and put
xn = T n x0 , n ≥ 0. There are several steps to be passed.
I) If xn = xn+1 for some n ≥ 0, we are done. So, it remains to discuss the
remaining situation; i.e. (by the reflexive sufficiency of d)
(c04) ρn := d(xn , xn+1 ) > 0, for all n ≥ 0.
By the contractive property and (3.1), ρn+1 ≤ ϕ(max{ρn , ρn+1 }), for all n ≥ 0; so
that (taking (2.3) into account)
ρn+1 ≤ ϕ(ρn ), ∀n ≥ 0. (3.2)
Combining with (c04) one gets that (ρn ; n ≥ 0) is strictly descending; moreover, by
Proposition 2, ρn → 0 as n → ∞.
II) Fix i ≥ 1, and put (σni := d(xn , xn+i ); n ≥ 0). Again by the contractive
condition, we get the evaluation
i
σn+1 = d(T xn , T xn+i ) ≤ ϕ(M (xn , xn+i )) = ϕ(max{σni , ρn , ρn+i }), ∀n ≥ 0;
wherefrom, by (3.2)
i
σn+1 ≤ ϕ(max{σni , ρn }), ∀n ≥ 0. (3.3)
This yields (again via Proposition 2) σni → 0, for each i ≥ 1; that is,
d(xn , xn+i ) → 0 as n → ∞, for each i ≥ 1; (3.4)
or, in other words: (xn ) is d-semi-Cauchy.
III) Suppose that
(c05) there exists i, j ∈ N such that i < j, xi = xj .
Denoting p = j −i, we thus have p > 0 and xi = xi+p ; so that (by the very definition
of our iterative sequence)
xi = xi+np , xi+1 = xi+np+1 , for all n ≥ 0.
By the introduced notations this yields (via (c04) and (3.4))
0 < ρi = ρi+np → 0 as n → ∞;
contradiction. Hence, (c05) cannot hold; wherefrom, we must have
for all i, j ∈ N : i 6= j implies xi 6= xj . (3.5)
IV) As a consequence of this fact, the map n 7→ xn is injective; so that, Y :=
{xn ; n ≥ 0} is effectively denumerable. Let k = k(Y ) ≥ 1 be the natural number
attached to it, by the local Branciari property of d. Also, let γ > 0 be arbitrary
fixed; and β ∈]0, γ[ be given by the strong regressivity of ϕ. By the d-semi-Cauchy
property (3.4), there exists j(β) ∈ N such that
d(xn , xn+i ) < β/4k (< β/2 < γ + β/2), ∀n ≥ j(β), ∀i ∈ {1, ..., k + 1}. (3.6)
We now claim that
(∀p ≥ 1) : d(xn , xn+p ) < γ + β/2, ∀n ≥ j(β); (3.7)
and, from this, the d-Cauchy property for (xn ; n ≥ 0) follows. The case of p ∈
{1, ..., k + 1} is clear, via (3.6). Assume that (3.7) holds, for some p ≥ k + 1; we
 FUNCTIONAL CONTRACTIONS IN LOCAL BRANCIARI METRIC SPACES 7

show that it holds as well for p + 1. So, let n ≥ j(β) be arbitrary fixed. By the
inductive hypothesis and (3.6),
d(xn+k , xn+p ) < γ + β/2 < γ + β
d(xn+k , xn+k+1 ) < β/4k < β < γ + β
d(xn+p , xn+p+1 ) < β/4k < β < γ + β;
whence, by definition,
M (xn+k , xn+p ) < γ + β.
This, by the contractive condition and (b06), gives
d(xn+k+1 , xn+p+1 ) ≤ ϕ(M (xn+k , xn+p )) ≤ γ.
Combining with the (2+k)-polyhedral inequality (applied to C = (xn+2 , ..., xn+k+1 ))
d(xn , xn+p+1 ) ≤ d(xn , xn+2 ) + ... + d(xn+k , xn+k+1 ) + d(xn+k+1 , xn+p+1 )
< β(k + 1)/4k + γ ≤ β/2 + γ;
and the assertion follows. As (X, d) is complete, we have
d
xn −→ x as n → ∞, for some z ∈ X; (3.8)
moreover, by Proposition 1, z is uniquely determined by this relation. We claim that
this is our desired point. Assume by contradiction that z 6= T z; or, equivalently,
ρ := d(z, T z) > 0.
V) Denote A = {n ∈ N ; xn = z}, B = {n ∈ N ; xn = T z}. If A is effectively
denumerable, we have A = {m(j); j ≥ 0}, where (m(j); j ≥ 0) is strictly ascending
(hence m(j) → ∞). As xm(j) = z, ∀j ≥ 0, we have xm(j)+1 = T z, ∀j ≥ 0.
d
Combining with xm(j)+1 −→ z as j → ∞, we must have d(z, T z) = 0; hence
z = T z, contradiction. On the other hand, if B is effectively denumerable, we have
B = {n(j); j ≥ 0}, where (n(j); j ≥ 0) is strictly ascending (hence n(j) → ∞). As
d
xn(j) = T z, ∀j ≥ 0, one gets (via xn(j) −→ z as j → ∞) d(z, T z) = 0; whence
z = T z, again a contradiction. It remains to discuss the case of both A and B
being finite; i.e.,
(c06) there exists h ≥ 0 such that: {xn ; n ≥ h} ∩ {z, T z} = ∅.
The subset Y := {xn ; n ≥ h} ∪ {z, T z} is therefore effectively denumerable. Let
k = k(Y ) ≥ 1 be the natural number attached to it, by the local Branciari property
of d. We have, for each n ≥ k (by the (2 + k)-polyhedral inequality applied to
C := (xn+2 , ..., xn+k+1 ))
ρ ≤ d(z, xn+2 ) + ... + d(xn+k , xn+k+1 ) + d(xn+k+1 , T z) (3.9)
By (3.4) and (3.8), there exists j(ρ) ≥ h in such a way that
n ≥ j(ρ) =⇒ d(xn , z), d(xn , xn+1 ) < ρ/2.
As a consequence, we must have
M (xn+k , z) = ρ, ∀n ≥ j(ρ).
so that, by the contractive condition,
d(xn+k+1 , T z) ≤ ϕ(ρ), ∀n ≥ j(ρ).
Replacing in (3.9), we get an evaluation like
ρ ≤ d(z, xn+2 ) + ... + d(xn+k , xn+k+1 ) + ϕ(ρ), ∀n ≥ j(ρ).
8 MIHAI TURINICI

Passing to limit as n tends to infinity gives ρ ≤ ϕ(ρ); wherefrom (as ϕ is regressive)

ρ = 0; contradiction. Hence, z = T z; and the proof is complete. 
In particular, when the regressive function ϕ is a Boyd-Wong one, our main
result covers the one due to Das and Dey [8]; note that, by the developments in
Jachymski [13], it includes as well the related statements in Di Bari and Vetro [9].
On the other hand, when d(., .) is a standard metric, Theorem 1 reduces to the
statement in Leader [19]. Further aspects may be found in Kikina et al [17]; see
also Khojasteh et al [15].

4. Further aspects
A direct inspection of the proof above shows that conclusion of Theorem 1 is
retainable even if one works with orbital completeness of the ambient space. Some
conventions are in order. Let X be a nonempty set; and d(., .) be a reflexive
sufficient symmetric over it; supposed to be a local Branciari metric. Further, take
a selfmap T of X. Call the sequence (yn ; n ≥ 0) in X, T -orbital when yn = T n x,
n ≥ 0, for some x ∈ X. In this case, let us say that (X, d) is T -orbital complete
when each T -orbital d-Cauchy sequence is d-convergent.
The following extension of Theorem 1 is available. Let the general conditions
above be fulfilled; as well as
(d01) (X, d) is T -orbital complete.
Theorem 2. Suppose that T is (d, M ; ϕ)-contractive, where ϕ ∈ Fr (R+ ) is strongly
regressive. Then, T is a strong Picard operator (modulo d).
The proof mimics the one of Theorem 1; so, we omit it.
Call the regressive function ϕ ∈ F (R+ ), admissible provided
P
(d02) ϕ is increasing, usc and n ϕn (t) < ∞, ∀t > 0.
Clearly, ϕ is a Matkowski function; hence, in particular, a strongly regressive one.
This, in the particular case of d fulfilling the tetrahedral inequality, tells us that the
main result in Fora et al [10] is a particular case of Theorem 2 above. In addition,
we note that the usc condition posed by the authors may be removed. Note that,
the introduced framework may be also used to get an extension of the contributions
due to Akram and Siddiqui [1]; see also Moradi and Alimohammadi [22]. These
will be discussed elsewhere.

References
[1] M. Akram and A. A. Siddiqui, A fixed-point theorem for A-contractions on a class of gener-
alized metric spaces, Korean J. Math. Sci., 10 (2003), 1-5.
[2] A. Azam and M. Arshad, Kannan fixed point theorem on generalized metric spaces, J. Non-
linear Sci. Appl., 1 (2008), 45-48.
[3] S. Banach, Sur les opérations dans les ensembles abstraits et leur application aux équations
intégrales, Fund. Math. 3 (1922) 133-181.
[4] D. W. Boyd and J. S. W. Wong, On nonlinear contractions, Proc. Amer. Math. Soc., 20
(1969), 458-464.
[5] A. Branciari, A fixed point theorem of Banach-Caccioppoli type on a class of generalized
metric spaces, Publ. Math. Debrecen, 57 (2000), 31-37.
[6] C. M. Chen and W. Y. Sun, Periodic points and fixed points for the weaker (φ−ϕ)-contractive
mappings in complete generalized metric spaces, J. Appl. Math., Volume 2012, Article Id:
856974.
[7] P. Das, A fixed point theorem on a class of generalized metric spaces, Korean J. Math. Sci.,
9 (2002), 29-33.
 FUNCTIONAL CONTRACTIONS IN LOCAL BRANCIARI METRIC SPACES 9

[8] P. Das and L. K. Dey, Fixed point of contractive mappings in generalized metric spaces,
Math. Slovaca, 59 (2009), 499-504.
[9] C. Di Bari and C. Vetro, Common fixed points in generalized metric spaces, Appl. Math.
Comput., 218 (2012), 7322-7325.
[10] A. Fora, A. Bellour and A. Al-Bsoul, Some results in fixed point theory concerning generalized
metric spaces, Mat. Vesnik, 61 (2009), 203-208.
[11] T. L. Hicks, Fixed-point theory in symmetric spaces with applications to probabilistic spaces,
Nonlin. Anal., 36 (1999), 331-344.
[12] J. Jachymski, Common fixed point theorems for some families of mappings, Indian J. Pure
Appl. Math., 25 (1994), 925-937.
[13] J. Jachymski, Equivalent conditions for generalized contractions on (ordered) metric spaces,
Nonlin. Anal., 74 (2011), 768-774
[14] S. Kasahara, On some generalizations of the Banach contraction theorem, Publ. Res. Inst.
Math. Sci. Kyoto Univ., 12 (1976), 427-437.
[15] F. Khojasteh, A. Razani, and S. Moradi, A fixed point of generalized TF-contraction map-
pings in cone metric spaces, Fixed Point Th. Appl., 2011, 2011:14.
[16] L. Kikina and K. Kikina, Fixed points on two generalized metric spaces, Int. J. Math. Anal.,
5 (2011), 1459-1467.
[17] L. Kikina, K. Kikina and K. Gjino, A new fixed point theorem on generalized quasimetric
spaces, ISRN Math. Analysis, Volume 2012, Article ID 457846.
[18] H. Lakzian and B. Samet, Fixed point for (ψ − ϕ)-weakly contractive mappings in generalized
metric spaces, Appl. Math. Lett., 25 (2011), 902-906.
[19] S. Leader, Fixed points for general contractions in metric spaces, Math. Japonica, 24 (1979),
17-24.
[20] J. Matkowski, Fixed point theorems for mappings with a contractive iterate at a point, Proc.
Amer. Math. Soc., 62 (1977), 344-348.
[21] D. Miheţ, On Kannan fixed point principle in generalized metric spaces, J. Nonlinear Sci.
Appl., 2 (2009), 92-96.
[22] S. Moradi and D. Alimohammadi, New extensions of Kannan fixed-point theorem on complete
metric and generalized metric spaces, Int. J. Math. Analysis, 5 (2011), 2313 - 2320.
[23] B. E. Rhoades, A comparison of various definitions of contractive mappings, Trans. Amer.
Math. Soc., 336 (1977), 257-290.
[24] I. A. Rus, Generalized Contractions and Applications, Cluj University Press, Cluj-Napoca,
2001.
[25] B. Samet, A fixed point theorem in a generalized metric space for mappings satisfying a
contractive condition of integral type, Int. J. Math. Anal., 3 (2009), 1265-1271.
[26] B. Samet, Discussion on ”A fixed point theorem of Banach-Caccioppoli type on a class of
generalized metric spaces” by A. Branciari, Publ. Math. Debrecen, 76 (2010), 493-494.
[27] I. R. Sarma, J. M. Rao and S. S. Rao, Contractions over generalized metric spaces, J.
Nonlinear Sci. Appl., 2 (2009), 180-182.
[28] M. Turinici, Fixed points in complete metric spaces, in ”Proc. Inst. Math. Iaşi” (Romanian
Academy, Iaşi Branch), pp. 179-182, Editura Academiei R.S.R., Bucureşti, 1976.
[29] J. Zhu, J. J. Cho and S. M. Kang, Equivalent contractive conditions in symmetric spaces,
Comput. Math. Appl., 50 (2005), 1621-1628.

”A. Myller” Mathematical Seminar; ”A. I. Cuza” University; 700506 Iaşi, Romania
E-mail address: [email protected]