On the stability of some functional equations in Menger φ-normed spaces
Transcript of On the stability of some functional equations in Menger φ-normed spaces
��
DOI: 10.2478/s12175-013-0197-z
Math. Slovaca 64 (2014), No. 1, 209–228
ON THE STABILITY
OF SOME FUNCTIONAL EQUATIONS
IN MENGER ϕ-NORMED SPACES
Dorel Mihet* — Reza Saadati**
(Communicated by Anatolij Dvurecenskij )
ABSTRACT. Recently, the authors [MIHET, D.—SAADATI, R.—VAEZ-POUR, S. M.: The stability of an additive functional equation in Menger prob-abilistic ϕ-normed spaces, Math. Slovaca 61 (2011), 817–826] considered the
stability of an additive functional in Menger ϕ-normed spaces. In this paper,we establish some stability results concerning the cubic, quadratic and quarticfunctional equations in complete Menger ϕ-normed spaces via fixed point theory.
c©2014Mathematical Institute
Slovak Academy of Sciences
1. Introduction
The study of stability problems for functional equations originates from aquestion of Ulam [40] concerning the stability of group homomorphisms, affir-matively answered by Hyers [8] and further generalized by Aoki [1] for additivemappings and by Rassias [34] for linear mappings. For some information onHyers-Ulam-Rassias stability for different functional equations the interestedreader is referred to [7,9,10,13,17,37].
In the recent papers [2], [23]–[30] the stability in the settings of probabilisticand fuzzy normed spaces is studied. Radu [33] pointed out that a fixed point al-ternative method can be successfully used to solve the Ulam problem. The fixedpoint method in the study of the probabilistic stability of functional equationsfirstly appears in [25] (see also [28]). In this paper we employ this method to
2010 Mathemat i c s Sub j e c t C l a s s i f i c a t i on: Primary 54E40; Secondary 39B82, 46S50,46S40.Keywords: probabilistic normed space, Hyers-Ulam stability, cubic functional equation, qua-dratic functional equation, quartic functional equation.
DOREL MIHET — REZA SAADATI
investigate the stability of the cubic and quartic functional equations in Mengerprobabilistic ϕ-normed spaces introduced by Golet in [11].
A function F : R → [0, 1] is called a distribution function if it is non-decreasingand left continuous with sup
t∈R
F (t) = 1 and inft∈R
F (t) = 0. The class of all distribu-
tion functions F with F (0) = 0 is denoted by D+. ε0 is the specific distributionfunction defined through
ε0(t) =
{0, t ≤ 0
1, t > 0.
���������� 1.1� A mapping T : [0, 1] × [0, 1] → [0, 1] is a triangular norm(briefly, a t-norm) if T satisfies the following conditions:
(i) T is commutative and associative;
(ii) T (a, 1) = a for all a ∈ [0, 1];
(iii) T (a, b) ≤ T (c, d) whenever a ≤ c and b ≤ d (a, b, c, d ∈ [0, 1]).
Three typical examples of continuous t-norms are TP (a, b) = ab, TL(a, b) =max(a+ b− 1, 0) and TM (a, b) = min(a, b).
Let ϕ be a function defined on the real field R into itself, with the followingproperties:
(a) ϕ(−t) = ϕ(t) for every t ∈ R;
(b) ϕ(1) = 1;
(c) ϕ is strictly increasing and continuous on [0,∞), ϕ(0) = 0 and limα→∞
ϕ(α)= ∞.
Examples of such functions are:
ϕ(α) = |α|; ϕ(α) = |α|p, p ∈ (0,∞); ϕ(α) =2α2n
|α|+ 1, n ∈ N.
���������� 1.2� ([11]) A Menger probabilistic ϕ-normed space is a triple(X, ν, T ), where X is a real vector space, T is a continuous t-norm, and ν isa from X into D+ such that the following conditions hold:
(PN1) νx(t) = ε0(t) for all t > 0 if and only if x = 0;
(PN2) ναx(t) = νx(t
ϕ(α) ) for all x in X, α �= 0 and t > 0;
(PN3) νx+y(t+ s) ≥ T (νx(t), νy(s)) for all x, y ∈ X and t, s ≥ 0.
We note that if ϕ(t) = |t| one obtains the definition of a random normedspace (RN space) or Menger probabilistic normed space, see [38]. An importantexample is given by the triple (X, ν, TM), where (X, ‖ · ‖) is a normed space and
νx(t) =t
t+ ‖x‖p , p ∈ (0, 1].
210
STABILITY OF SOME FUNCTIONAL EQUATIONS IN MENGER ϕ-NORMED SPACES
If (X, ν, T ) is a probabilistic ϕ-normed space under a continuous t-norm Tsuch that T ≥ TL, then
V ={V (ε, λ) : ε > 0, λ ∈ (0, 1)
}, V (ε, λ) =
{x ∈ X : νx(ε) > 1− λ
}is a complete system of neighborhoods of null vector for linear topology τ on Xgenerated by the ϕ-norm ν.
���������� 1.3� Let (X, ν, T ) be a Menger probabilistic ϕ-normed space.
(i) A sequence {xn} in X is said to be convergent to x in X in the topologyτ if for every t > 0 and ε > 0, there exists positive integer N such thatνxn−x(t) > 1− ε whenever n ≥ N .
(ii) A sequence {xn} in X is called Cauchy if for every t > 0 and ε > 0, thereexists positive integer N such that νxn−xm
(t) > 1− ε whenever n,m ≥ N .
(iii) A Menger probabilistic ϕ-normed space (X, ν, T ) is said to be complete ifevery Cauchy sequence in X is convergent to a point in X.
The notion of generalized metric space has been introduced by Luxemburgin [22], by allowing the value +∞ for the distance mapping. The followinglemma (Luxemburg-Jung theorem) will be used in the proof of Theorem 2.1.
�� 1.4� ([19]) Let (X, d) be a complete generalized metric space andT : X → X be a strict contraction with the Lipschitz constant k, such thatd(x0, T (x0)) < +∞ for some x0 ∈ X. Then T has a unique fixed point inthe set Y :=
{y ∈ X : d(x0, y) < ∞}
and the sequence (Tn(x))n∈N convergesto the fixed point x∗ for every x ∈ Y . Moreover, d(x0, T (x0)) ≤ δ impliesd(x∗, x0) ≤ δ
1−k .
Let X be a linear space, (Y, ν, TM) be a complete Menger probabilisticϕ-normed space andG be a mapping fromX×R into [0, 1], such thatG(x, ·)∈D+
for all x. Consider the set E :={g : X → Y : g(0) = 0
}and the mapping dG
defined on E × E by
dG(g, h) = inf{u ∈ R
+ : νg(x)−h(x)(ut) ≥ G(x, t) for all x ∈ X and t > 0}
where, as usual, inf φ = +∞. The following lemma can be proved as in [23]:
�� 1.5� (cf. [23,25]) dG is a complete generalized metric on E.
211
DOREL MIHET — REZA SAADATI
2. Probabilistic stability of the equation 1af(bx) = f(x)
Our first result is the following stability theorem in random ϕ-normed spaces:
���� � 2.1� Let X be a real linear space, a, b ∈ R, a �= 0, f be a mappingfrom X into a complete Menger probabilistic ϕ-normed space (Y, ν, TM) withf(0) = 0 and let G : X → D+ be a mapping with the property
∃α ∈ (0, ϕ(a)) ∀x ∈ X ∀t ∈ (0,∞) : G(bx, αt) ≥ G(x, t). (2.1)
If
ν 1a f(bx)−f(x)(t) ≥ G(x, t) (x ∈ X, t > 0),
then there exists a unique mapping g : X → X such that g(bx) = ag(x), for allx ∈ X and
νg(x)−f(x)(t) ≥ G
(x,
ϕ(a)− α
ϕ(a)t
)(x ∈ X, t > 0).
Moreover,
g(x) = limn→∞
f(bnx)
an(x ∈ X).
P r o o f. Consider the set
E :={g : X → Y : g(0) = 0
}and the mapping dG defined on E ×E by
dG(g, h) = inf{u ∈ R+ : νg(x)−h(x)(ut) ≥ G(x, t) for all x ∈ X and t > 0
}.
By Lemma 1.5, (E, dG) is a complete generalized metric space. Now, let usconsider the linear mapping J : E → E,
Jg(x) :=1
ag(bx).
We show that J is a strictly contractive self-mapping of E with the Lipschitzconstant k = α/ϕ(a).
Indeed, let g, h in E be such that dG(g, h) < ε. Then
∀x ∈ X ∀t ∈ (0,∞) : νg(x)−h(x)(εt) ≥ G(x, t),
whence
νJg(x)−Jh(x)
(α
ϕ(a)εt
)= ν 1
a (g(bx)−h(bx))
(α
ϕ(a)εt
)= νg(bx)−h(bx)(αεt) ≥ G(bx, αt) (x ∈ X, t > 0).
Since G(bx, αt) ≥ G(x, t), then νJg(x)−Jh(x)(α
ϕ(a) εt) ≥ G(x, t), that is,
dG(g, h) < ε =⇒ dG(Jg, Jh) ≤ α
ϕ(a)ε.
212
STABILITY OF SOME FUNCTIONAL EQUATIONS IN MENGER ϕ-NORMED SPACES
This means that
dG(Jg, Jh) ≤ α
ϕ(a)dG(g, h),
for all g, h in E.
Next, from
νf(x)− 1a f(bx)(t) ≥ G(x, t)
it follows that dG(f, Jf) ≤ 1. Using the Luxemburg-Jung theorem (Lemma 1.4),we deduce the existence of a fixed point of J , that is, the existence of a mappingg : X → Y such that g(bx) = ag(x), for all x ∈ X.
Since for any x ∈ X and t > 0,
dG(u, v) < ε =⇒ νu(x)−v(x)(t) ≥ G
(x,
t
ε
),
from dG(Jnf, g) → 0, it follows that lim
n→∞f(bnx)
an = g(x), for any x ∈ X.
Also, dG(f, g) ≤ 11−kd(f, Jf) implies the inequality dG(f, g) ≤ 1
1− αϕ(a)
from
which it immediately follows νg(x)−f(x)(ϕ(a)
ϕ(a)−α t) ≥ G(x, t) for all t > 0 and
x ∈ X. This means that
∀x ∈ X ∀t ∈ (0,∞) : νg(x)−f(x)(t) ≥ G
(x,
ϕ(a)− α
ϕ(a)t
).
The uniqueness of g follows from the fact that g is the unique fixed point ofJ with the property: there is C ∈ (0,∞) such that νg(x)−f(x)(Ct) ≥ G(x, t) forall x ∈ X and t > 0. �
Remark 2.2� For a similar result in the setting of Menger spaces, the reader isreferred to the paper [26].
3. Probabilistic stability for the Jensen, quadratic, cubicand quartic functional equations
In the following we will appeal to Theorem 2.1 in order to prove the prob-abilistic stability for the Jensen, the quadratic, cubic and quartic functionalequations.
Recall that the functional equation
f
(1
2(x+ y)
)=
1
2
(f(x) + f(y)
)(3.1)
is called Jensen functional equation. A mapping f between two linear spaceswith f(0) = 0 satisfies the Jensen equation if and only if it is additive; cf. [32].
213
DOREL MIHET — REZA SAADATI
Stability of Jensen equation has been studied at first by Kominek [21] and thenby several other mathematicians, see [4], [16] and references therein.
The functional equation
f(2x+ y) + f(2x− y) = 2f(x+ y) + 2f(x− y) + 12f(x) (3.2)
is called a cubic functional equation since the function f(x) = cx3 is a solution.Every solution of the cubic functional equation is said to be a cubic mapping.The stability problem for the cubic functional equation was firstly considered in[12] for mappings f : X → Y , where X is a real normed space and Y is a Banachspace. Later, a number of mathematicians have worked on the stability of sometypes of the cubic equation [14,34].
The functional equation
f(x+ y) + f(x− y) = 2f(x) + 2f(y) (3.3)
is called a quadratic functional equation. F. Skof [39] was the first one whoinvestigated the stability of this equation. She showed that, if f is a mappingfrom a normed space X into a Banach space Y satisfying ‖f(x+ y)+ f(x− y)−2f(x) − 2f(y)‖ ≤ ε for some ε > 0, then there is a unique quadratic functiong : X → Y such that ‖f(x) − g(x)‖ ≤ ε
2 . P. W. Cholewa [5] extended Skof’stheorem by replacing X by an abelian group. Skof’s result was later generalizedby S. Czerwik [6] in the spirit of Hyers-Ulam-Rassias. The stability problem ofthe quadratic equation has been extensively investigated by a number of othermathematicians, see e.g. [3,7,15,18,20,31,35] and references therein.
The functional equation
f(2x+ y) + f(2x− y) = 4f(x+ y) + 4f(x− y) + 24f(x)− 6f(y) (3.4)
is called a quartic functional equation, since the function f(x) = cx4 is a solution.Every solution of the quartic functional equation is called a quartic mapping.The stability problem for the quartic functional equation first was considered byTh. M. Rassias [36] for mappings f : X → Y , where X is a real normed spaceand Y is a Banach space.
Our first theorem in this section is a stability result for the Jensen functionalequation.
���� � 3.1� Let X be a real linear space, (Y, ν, TM) be a complete Mengerprobabilistic ϕ-normed space and Φ be a mapping from X2 to D+ (Φ(x, y) isdenoted by Φx,y) such that, for some 0 < α < ϕ(1/2),
Φ 12x,
12y(αt) ≥ Φx,y(t) (x, y, z ∈ X, t > 0) (3.5)
and
limn→∞αnϕ(2n) = 0. (3.6)
214
STABILITY OF SOME FUNCTIONAL EQUATIONS IN MENGER ϕ-NORMED SPACES
Suppose that f : X → Y has the properties: f(0) = 0 and
νf( 12 (x+y))−( 1
2 (f(x)+f(y))(t) ≥ Φx,y(t) (x, y ∈ X, t > 0). (3.7)
Then there exists a unique additive mapping g : X → Y such that
νf(x)−g(x)(t) ≥ Φx,0(Mt) (x ∈ X, t > 0), (3.8)
where M = ϕ(1/2)−αϕ(1/2)ϕ(2) .
P r o o f. By setting y = 0 in (3.7) we obtain that
νf( 12x)− 1
2 f(x)(t) ≥ Φx,0(t) (x ∈ X, t > 0),
that is,
ν2f( 12x)−f(x)(ϕ(2)t) = νf( 1
2x)− 12 f(x)
(t) ≥ Φx,0(t) (x ∈ X, t > 0).
It follows that
ν2f( 12x)−f(x)(t) ≥ G(x, t),
where G(x, t) := Φx,0
(t
ϕ(2)
).
From Theorem 2.1 we deduce that g(x) = limn→∞ 2nf( 1
2nx) (x ∈ X) is the
unique mapping g : X → Y such that g(12x
)= 1
2g(x), for all x ∈ X and
νg(x)−f(x)(t) ≥ Φx,0
(ϕ(1/2)− α
ϕ(1/2)ϕ(2)t
)(x ∈ X, t > 0).
To show that g is a solution for the Jensen equation, it suffices to prove thatthe mapping g is Cauchy additive. We have:
νg(x)+g(y)−g(x+y)(t)
≥ Min{νg(x)−2nf( 1
2n x)(t4
), ν
g(y)−2nf(
12n y
)( t4
), ν
g(x+y)−2nf(
12n (x+y)
) ( t4
),
ν2nf( 12n (x+y))−2nf( 1
2n x)−2nf( 12n y)
(t4
) }(x, y ∈ X, t > 0, n ∈ N).
The first three terms on the right-hand side of the above inequality tend to 1 asn → ∞. Also, let us note from (3.5) it immediately follows by induction on nthat
Φ 12n x, 1
2n y (αnt) ≥ Φx,y(t) (x, y ∈ X, t > 0),
hence
Φ 12n x, 1
2n y(t) ≥ Φx,y
(t
αnt
)(x, y ∈ X, t > 0). (3.9)
215
DOREL MIHET — REZA SAADATI
Then, by using (3.7), we obtain
ν2nf( 12n (x+y))−2nf( 1
2nx )−2nf( 12n y)
(t
4
)= νf( 1
2n (x+y))−f( 12n x)−f( 1
2n y)
(t
4ϕ(2n)
)
≥ Φ 12n x, 1
2n y
(t
4ϕ(2n)
)
≥ Φx,y
(t
4αnϕ(2n)
).
From (3.9) we deduce that the fourth term also tends to 1 when n tends to ∞,concluding that g is additive. �
In what follows we obtain similar results on the stability of cubic, quadraticand quartic functional equations.
���� � 3.2� Let X be a real linear space, let f be a mapping from X into acomplete Menger probabilistic ϕ-normed space (Y, ν, TM) with f(0) = 0 and letΦ: X2 → D+ be a mapping with the property
∃α ∈ (0, ϕ(8)) ∀x, y ∈ X ∀t ∈ (0,∞) : Φ2x,2y(αt) ≥ Φx,y(t). (3.10)
If
νf(2x+y)+f(2x−y)−2f(x+y)−2f(x−y)−12f(x)(t) ≥ Φx,y(t) for all x, y ∈ X,(3.11)
and
limn→∞
αnϕ
(1
23n
)= 0, (3.12)
then there is a unique cubic mapping g : X → Y such that
νg(x)−f(x)(t) ≥ Φx,0 (Mt) for all x ∈ X, t > 0, (3.13)
where
M =ϕ(8)− α
ϕ(8)ϕ(1/16).
Moreover,
g(x) = limn→∞
f(2nx)
23n.
P r o o f. By setting y = 0 in (3.11), we obtain
ν2f(2x)−16f(x)(t) ≥ Φx,0(t),
for all x ∈ X, whence
ν 18 f(2x)−f(x)(t) = ν 1
16 (2f(2x)−16f(x))(t) = ν2f(2x)−16f(x)
(t/ϕ(
1
16)
)
≥ Φx,0
(t/ϕ(
1
16)
), for all x ∈ X, t > 0.
216
STABILITY OF SOME FUNCTIONAL EQUATIONS IN MENGER ϕ-NORMED SPACES
Let
G(x, t) := Φx,0
(t
ϕ( 116 )
).
By Theorem 2.1 it follows the existence of a unique mapping g : X → Y suchthat g(2x) = 8g(x), for all x ∈ X and
νg(x)−f(x)(t) ≥ Φx,0
(ϕ(8)− α
ϕ(8)ϕ(1/16)t
)for all x ∈ X, t > 0.
Moreover, limn→∞
f(2nx)23n = g(x), for any x ∈ X.
The proof of the fact that g is a cubic mapping is similar to the proof oflinearity in the preceding theorem. �
In a similar way we can obtain stability results for the quadratic and thequartic functional equations, which are stated in the following.
���� � 3.3� Let X be a real linear space, let f be a mapping from X into acomplete Menger probabilistic ϕ-normed space (Y, ν, TM) with f(0) = 0 and letΦ: X2 → D+ be a mapping with the property
∃α ∈ (0, ϕ(4)) ∀x, y ∈ X ∀t ∈ (0,∞) : Φ2x,2y(αt) ≥ Φx,y(t).
If
νf(x+y)+f(x−y)−2f(x)−2f(y)(t) ≥ Φx,y(t) for all x, y ∈ X,
and
limn→∞αnϕ
(1
22n
)= 0,
then the formula g(x) = limn→∞
f(2nx)22n defines a unique quadratic mapping g : X→ Y
such that
νg(x)−f(x)(t) ≥ Φx,x (Mt) , for all x ∈ X, t > 0,
where
M =ϕ(4)− α
ϕ(4)ϕ(1/4).
P r o o f. (Sketch) By setting x = y in (3.4), we obtain
νf(2x)−4f(x)(t) ≥ Φx,x(t),
for all x ∈ X, whence
ν 14 f(2x)−f(x)(t) ≥ G(x, t), for all x ∈ X, t > 0,
where G(x, t) := Φx,x
(t/ϕ(14 )
). �
217
DOREL MIHET — REZA SAADATI
���� � 3.4� Let X be a real linear space, let f be a mapping from X into acomplete Menger probabilistic ϕ-normed space (Y, ν, TM) with f(0) = 0 and letΦ: X2 → D+ be a mapping with the property
∃α ∈ (0, ϕ(16)) ∀x, y ∈ X ∀t ∈ (0,∞) : Φ2x,2y(αt) ≥ Φx,y(t).
If
νf(2x+y)+f(2x−y)−4f(x+y)−4f(x−y)−24f(x)+6f(y)(t) ≥ Φx,y(t) for all x, y ∈ X,
and
limn→∞αnϕ
(1
24n
)= 0,
then there is a unique quartic mapping g : X → Y , g(x) = limn→∞
f(2nx)24n , such that
νg(x)−f(x)(t) ≥ Φx,0 (Mt) for all x ∈ X, t > 0,
where
M =ϕ(16)− α
ϕ(16)ϕ(1/32).
4. Particular cases
For concrete choices of ϕ, Φ or ν, one can obtain stability theorems for differ-ent functional equations in random normed spaces or in linear normed spaces.
���� � 4.1� Let X be a real linear space, (Y, ν, TM) be a complete probabilisticnormed space and Φ be a mapping from X2 to D+ such that, for some 0 < α< 1/2,
Φ 12x,
12y(αt) ≥ Φx,y(t) (x, y ∈ X, t > 0).
Suppose that f : X → Y has the properties: f(0) = 0 and
νf( 12 (x+y))−( 1
2 (f(x)+f(y))(t) ≥ Φx,y(t) (x, y ∈ X, t > 0).
Then there exists a unique additive mapping g : X → Y such that
νf(x)−g(x)(t) ≥ Φx,0((1/2− α)t) (x ∈ X, t > 0).
P r o o f. The conclusion follows by considering ϕ(α) = |α| in Theorem 3.1 (we
note that in this case ϕ(1/2)−αϕ(1/2)ϕ(2) = ϕ(1/2− α).
The conditionlim
n→∞αnϕ(2n) = 0
is satisfied, as it reduces tolimn→∞
(2α)n = 0.
�
218
STABILITY OF SOME FUNCTIONAL EQUATIONS IN MENGER ϕ-NORMED SPACES
���� � 4.2� Let (X, ‖ · ‖) be a real normed linear space, (Y, ν, TM) be acomplete RN-space and p be nonnegative real number. If f : X → Y is a mappingsuch that
ν f(x)+f(y)2 −f( x+y
2 )(t) ≥ t
t+ ‖x‖p + ‖y‖p (x, y ∈ X, t > 0), (4.1)
and p < 1, then there exists a unique additive mapping g : X → Y such that
νf(x)−g(x)(t) ≥ (1− 2p−1)t
(1− 2p−1)t+ 2p‖x‖p . (4.2)
for all x ∈ X and t > 0.
P r o o f. Consider the mapping Φ: X2 → D+ defined through
Φx,y(t) =t
t+ ‖x‖p + ‖y‖pand let ϕ(t) = |t|p (t ∈ R), where 0 < p < 1, α = 1/2.
It is immediate that0 < 1/2 < ϕ(1/2),
Φ 12x,
12y(αt) ≥ Φx,y(t) (x, y, z ∈ X, t > 0)
andlim
n→∞αnϕ(2n) = lim
n→∞2(p−1)n = 0.
Now the conclusion follows from Theorem 3.1. �
�� ��� � 4.3� ([16: Theorem 1], with δ = 0, θ = 2) Let f : X → Y be amapping between real Banach spaces and let 0 ≤ p < 1 be fixed. If f satisfies theinequality ∥∥∥∥f
(x+ y
2
)− f(x) + f(y)
2
∥∥∥∥ ≤ ‖x‖p + ‖y‖p
for all x, y ∈ X, then there exists a unique additive mapping F : X → Y suchthat
‖f(x)− F (x)‖ ≤ 2p
1− 2p−1‖x‖
for all x ∈ X.
P r o o f. Consider the induced RN space (X, ν, TM), where νx(t) =t
t+‖x‖p . Then
(4.1) is equivalent to∥∥∥∥f(x+ y
2
)− f(x) + f(y)
2
∥∥∥∥ ≤ ‖x‖p + ‖y‖p,
while (4.2) is equivalent to
‖f(x)− g(x)‖ ≤ 2p
1− 2p−1‖x‖. �
219
DOREL MIHET — REZA SAADATI
���� � 4.4� Let X be a real linear space, (Y, ν, TM) be a complete probabilisticnormed space and Φ be a mapping from X2 to D+ such that, for some 0 < α < 8,
Φ2x,2y(αt) ≥ Φx,y(t) (x, y ∈ X, t > 0).
Suppose that f : X → Y has the properties: f(0) = 0 and
νf(2x+y)+f(2x−y)−2f(x+y)−2f(x−y)−12f(x)(t) ≥ Φx,y(t) (x, y ∈ X, t > 0).
Then there exists a unique cubic mapping g : X → Y such that
νf(x)−g(x)(t) ≥ Φx,0(2(8− α)t) (x ∈ X, t > 0).
P r o o f. The conclusion follows by considering ϕ(α) = |α| in Theorem 3.2 (we
note that in this case ϕ(8)−αϕ(8)ϕ(1/16) = ϕ(2(8− α)).
The condition
limn→∞αnϕ
(1
23n
)= 0
is satisfied, as it reduces to
limn→∞
(α8
)n
= 0. �
���� � 4.5� Let (X, ‖ · ‖) be a real normed linear space, (Y, ν, TM) be acomplete RN-space and p be nonnegative real number. If f : X → Y is a mappingsuch that
νf(2x+y)+f(2x−y)−2f(x+y)−2f(x−y)−12f(x)(t) ≥ t
t+ ‖x‖p + ‖y‖p(x, y ∈ X, t > 0),
(4.3)
and 1 < p < 3, then there exists a unique cubic mapping g : X → Y such that
νf(x)−g(x)(t) ≥ (8p − 8)t
(8p − 8)t+ 2−p‖x‖p . (4.4)
for all x ∈ X and t > 0.
P r o o f. Consider the mapping Φ: X2 → D+ defined through
Φx,y(t) =t
t+ ‖x‖p + ‖y‖pand let ϕ(t) = |t|p (t ∈ R), where 1 < p < 3, α = 8.
It is immediate that0 < 8 < ϕ(8),
Φ2x,2y(αt) ≥ Φx,y(t) (x, y, z ∈ X, t > 0)
and
limn→∞αnϕ
( 1
8n
)= lim
n→∞ 8(1−p)n = 0.
Now the conclusion follows from Theorem 3.2. �
220
STABILITY OF SOME FUNCTIONAL EQUATIONS IN MENGER ϕ-NORMED SPACES
���� � 4.6� Let (X, ‖ · ‖) be a real normed linear space, (Y, ν, TM) be acomplete RN-space and p be nonnegative real number. If f : X → Y is a mappingsuch that
νf(2x+y)+f(2x−y)−2f(x+y)−2f(x−y)−12f(x)(t) ≥ t
t+ ‖x‖p + ‖y‖p(x, y ∈ X, t > 0),
(4.5)
and 13 < p < 1, then there exists a unique cubic mapping g : X → Y such that
νf(x)−g(x)(t) ≥ (8p − 2)t
(8p − 2)t+ 2−p‖x‖p . (4.6)
for all x ∈ X and t > 0.
P r o o f. Consider the mapping Φ: X2 → D+ defined through
Φx,y(t) =t
t+ ‖x‖p + ‖y‖pand let ϕ(t) = |t|p (t ∈ R), where 1
3 < p < 1, α = 2.
It is immediate that0 < 2 < ϕ(8),
Φ2x,2y(αt) ≥ Φx,y(t) (x, y, z ∈ X, t > 0)
and
limn→∞αnϕ
(1
8n
)= lim
n→∞ 2(1−3p)n = 0.
Now the conclusion follows from Theorem 3.2. ��� ��� � 4.7� ([12: Theorem 1], with δ = 0, θ = 2) Let f : X → Y be amapping between real Banach spaces and let 1/3 ≤ p < 1 be fixed. If f satisfiesthe inequality
‖f(2x+ y) + f(2x− y)− 2f(x+ y)− 2f(x− y)− 12f(x)‖ ≤ ‖x‖p + ‖y‖pfor all x, y ∈ X, then there exists a unique cubic mapping g : X → Y such that
‖f(x)− g(x)‖ ≤ 2−p
8p − 2‖x‖
for all x ∈ X.
P r o o f. Consider the induced RN space (X, ν, TM), where νx(t) =t
t+‖x‖p . Then
(4.5) is equivalent to
‖f(2x+ y) + f(2x− y)− 2f(x+ y)− 2f(x− y)− 12f(x)‖ ≤ ‖x‖p + ‖y‖p,while (4.6) is equivalent to
‖f(x)− g(x)‖ ≤ 2−p
8p − 2‖x‖. �
221
DOREL MIHET — REZA SAADATI
���� � 4.8� Let X be a real linear space, (Y, ν, TM) be a complete probabilisticnormed space and Φ be a mapping from X2 to D+ such that, for some 0 < α < 4,
Φ2x,2y(αt) ≥ Φx,y(t) (x, y, z ∈ X, t > 0).
Suppose that f : X → Y has the properties: f(0) = 0 and
νf(x+y)+f(x−y)−2f(x)−2f(y)(t) ≥ Φx,y(t) (x, y ∈ X, t > 0).
Then there exists a unique quadratic mapping g : X → Y such that
νf(x)−g(x)(t) ≥ Φx,x((4− α)t) (x ∈ X, t > 0).
P r o o f. The conclusion follows by considering ϕ(α) = |α| in Theorem 3.3 (we
note that in this case ϕ(4)−αϕ(4)ϕ(1/4) = ϕ(4− α).
The condition
limn→∞αnϕ
(1
22n
)= 0
is satisfied, as it reduces to
limn→∞
(α4
)n
= 0. �
���� � 4.9� Let (X, ‖ · ‖) be a real normed linear space, (Y, ν, TM) be acomplete RN-space and p be nonnegative real number. If f : X → Y is a mappingsuch that
νf(x+y)+f(x−y)−2f(x)−2f(y)(t) ≥ t
t+ ‖x‖p + ‖y‖p (x, y ∈ X, t > 0), (4.7)
and 1 < p < 2, then there exists a unique quadratic mapping g : X → Y suchthat
νf(x)−g(x)(t) ≥ (4p − 4)t
(4p − 4)t+ 2‖x‖p . (4.8)
for all x ∈ X and t > 0.
P r o o f. Consider the mapping Φ: X2 → D+ defined through
Φx,y(t) =t
t+ ‖x‖p + ‖y‖pand let ϕ(t) = |t|p (t ∈ R), where 1 < p < 2, α = 4.
It is immediate that
0 < 4 < ϕ(4),
Φ2x,2y(αt) ≥ Φx,y(t) (x, y, z ∈ X, t > 0)
and
limn→∞αnϕ
(1
4n
)= lim
n→∞ 4(1−p)n = 0.
Now the conclusion follows from Theorem 3.3. �
222
STABILITY OF SOME FUNCTIONAL EQUATIONS IN MENGER ϕ-NORMED SPACES
���� � 4.10� Let (X, ‖ · ‖) be a real normed linear space, (Y, ν, TM) be acomplete RN-space and p be nonnegative real number. If f : X → Y is a mappingsuch that
νf(x+y)+f(x−y)−2f(x)−2f(y)(t) ≥ t
t+ ‖x‖p + ‖y‖p (x, y ∈ X, t > 0), (4.9)
and 12 < p < 1, then there exists a unique quadratic mapping g : X → Y such
that
νf(x)−g(x)(t) ≥ (4p − 2)t
(4p − 2)t+ 2‖x‖p . (4.10)
for all x ∈ X and t > 0.
P r o o f. Consider the mapping Φ: X2 → D+ defined through
Φx,y(t) =t
t+ ‖x‖p + ‖y‖pand let ϕ(t) = |t|p (t ∈ R), where 1
2 < p < 1, α = 2.
It is immediate that
0 < 2 < ϕ(4),
Φ2x,2y(αt) ≥ Φx,y(t) (x, y, z ∈ X, t > 0)
and
limn→∞
αnϕ
(1
4n
)= lim
n→∞2(1−2p)n = 0.
Now the conclusion follows from Theorem 3.3. �
�� ��� � 4.11� ([39: Theorem 1], with δ = 0, θ = 2) Let f : X → Y be amapping between real Banach spaces and let 1/2 ≤ p < 1 be fixed. If f satisfiesthe inequality
‖f(x+ y) + f(x− y)− 2f(x)− 2f(y)‖ ≤ ‖x‖p + ‖y‖pfor all x, y ∈ X, then there exists a unique quadratic mapping g : X → Y suchthat
‖f(x)− g(x)‖ ≤ 2
4p − 2‖x‖
for all x ∈ X.
P r o o f. Consider the induced RN space (X, ν, TM), where νx(t) =t
t+‖x‖p . Then
(4.9) is equivalent to
‖f(x+ y) + f(x− y)− 2f(x)− 2f(y)‖ ≤ ‖x‖p + ‖y‖p,while (4.10) is equivalent to
‖f(x)− g(x)‖ ≤ 2
4p − 2‖x‖. �
223
DOREL MIHET — REZA SAADATI
���� � 4.12� Let X be a real linear space, (Y, ν, TM) be a complete proba-bilistic normed space and Φ be a mapping from X2 to D+ such that, for some0 < α < 16,
Φ2x,2y(αt) ≥ Φx,y(t) (x, y, z ∈ X, t > 0).
Suppose that f : X → Y has the properties: f(0) = 0 and
νf(2x+y)+f(2x−y)−4f(x+y)−4f(x−y)−24f(x)+6f(y)(t) ≥ Φx,y(t)
(x, y ∈ X, t > 0).
Then there exists a unique quadratic mapping g : X → Y such that
νf(x)−g(x)(t) ≥ Φx,0(2(16− α)t) (x ∈ X, t > 0).
P r o o f. The conclusion follows by considering ϕ(α) = |α| in Theorem 3.4 (we
note that in this case ϕ(16)−αϕ(16)ϕ(1/32) = ϕ(2(16− α)).
The condition
limn→∞αnϕ
(1
24n
)= 0
is satisfied, as it reduces to
limn→∞
( α
16
)n
= 0. �
���� � 4.13� Let (X, ‖ · ‖) be a real normed linear space, (Y, ν, TM) be acomplete RN-space and p be nonnegative real number. If f : X → Y is a mappingsuch that
νf(2x+y)+f(2x−y)−4f(x+y)−4f(x−y)−24f(x)+6f(y)(t) ≥ t
t+ ‖x‖p + ‖y‖p(x, y ∈ X, t > 0),
(4.11)
and 1 < p < 4, then there exists a unique quartic mapping g : X → Y such that
νf(x)−g(x)(t) ≥ (16p − 16)t
(16p − 16)t+ 2−p‖x‖p . (4.12)
for all x ∈ X and t > 0.
P r o o f. Consider the mapping Φ: X2 → D+ defined through
Φx,y(t) =t
t+ ‖x‖p + ‖y‖pand let ϕ(t) = |t|p (t ∈ R), where 1 < p < 4, α = 16.
224
STABILITY OF SOME FUNCTIONAL EQUATIONS IN MENGER ϕ-NORMED SPACES
It is immediate that0 < 16 < ϕ(16),
Φ2x,2y(αt) ≥ Φx,y(t) (x, y, z ∈ X, t > 0)
and
limn→∞αnϕ
( 1
16n
)= lim
n→∞ 16(1−p)n = 0.
Now the conclusion follows from Theorem 3.4. �
���� � 4.14� Let (X, ‖ · ‖) be a real normed linear space, (Y, ν, TM) be acomplete RN-space and p be nonnegative real number. If f : X → Y is a mappingsuch that
νf(2x+y)+f(2x−y)−4f(x+y)−4f(x−y)−24f(x)+6f(y)(t) ≥ t
t+ ‖x‖p + ‖y‖p(x, y ∈ X, t > 0),
(4.13)
and 14 < p < 1, then there exists a unique quartic mapping g : X → Y such that
νf(x)−g(x)(t) ≥ (16p − 2)t
(16p − 2)t+ 2−p‖x‖p . (4.14)
for all x ∈ X and t > 0.
P r o o f. Consider the mapping Φ: X2 → D+ defined through
Φx,y(t) =t
t+ ‖x‖p + ‖y‖pand let ϕ(t) = |t|p (t ∈ R), where 1
4 < p < 1, α = 2.
It is immediate that0 < 2 < ϕ(16),
Φ2x,2y(αt) ≥ Φx,y(t) (x, y, z ∈ X, t > 0)
and
limn→∞αnϕ
(1
16n
)= lim
n→∞ 2(1−4p)n = 0.
Now the conclusion follows from Theorem 3.4. �
�� ��� � 4.15� ([36: Theorem 1], with δ = 0, θ = 2) Let f : X → Y be amapping between real Banach spaces and let 1/4 ≤ p < 1 be fixed. If f satisfiesthe inequality
‖f(2x+ y)+ f(2x− y)− 4f(x+ y)− 4f(x− y)− 24f(x)+ 6f(y)‖ ≤ ‖x‖p+ ‖y‖pfor all x, y ∈ X, then there exists a unique quartic mapping g : X → Y such that
‖f(x)− g(x)‖ ≤ 2−p
16p − 2‖x‖
for all x ∈ X.
225
DOREL MIHET — REZA SAADATI
P r o o f. Consider the induced RN space (X, ν, TM), where νx(t) =t
t+‖x‖p . Then
(4.13) is equivalent to
‖f(2x+ y)+ f(2x− y)− 4f(x+ y)− 4f(x− y)− 24f(x)+6f(y)‖ ≤ ‖x‖p+ ‖y‖p,while (4.14) is equivalent to
‖f(x)− g(x)‖ ≤ 2−p
16p − 2‖x‖.
�
Recently (see [41]) it was shown that if (X,µ, T ) is a Menger probabilisticϕ-normed space with T = TM (as in our paper), then ϕ(t) = |t|p, with p ∈ (0, 1].Nevertheless, Menger probabilistic ϕ-normed spaces still represent a rich enoughstructure, allowing one to unify many known stability results, as it have seen inSections 2 and 3. We also mention here the paper [42], where complementaryresults in the case α > ϕ(a) are provided.
Acknowledgement� The authors would like to thank the referee and man-aging editor Professor Anatolij Dvurecenskij for giving useful suggestions andcomments for the improvement of this paper.
REFERENCES
[1] AOKI, T.: On the stability of the linear transformation in Banach spaces, J. Math. Soc.Japan 2 (1950), 64–66.
[2] BAKTASH, E.—CHO, Y. J.—JALILI, M.—SAADATI, R.—VAEZPOUR, S. M.: On thestability of cubic mappings and quadratic mappings in random normed spaces, J. Inequal.Appl. 2008 (2008), Article ID 902187, 11 pp.
[3] CADARIU, L.—RADU, V.: The alternative of fixed point and stability results for func-tional equations, Int. J. Appl. Math. Stat. 7 (2007), 40–58.
[4] CADARIU, L.—RADU, V.: Fixed points and the stability of Jensen’s functional equation,JIPAM. J. Inequal. Pure Appl. Math. 4 (2003), Article 4.
[5] CHOLEWA, P. W.: Remarks on the stability of functional equations, Aequationes Math.
27 (1984), 76–86.[6] CZERWIK S.: On the stability of the quadratic mapping in normed spaces, Abh. Math.
Sem. Univ. Hamburg 62 (1992), 59–64.[7] CZERWIK, S.: Functional Equations and Inequalities in Several Variables, World Scien-
tific, River Edge, NJ, 2002.[8] HYERS, D. H.: On the stability of the linear functional equation, Proc. Natl. Acad. Sci.
USA 27 (1941), 222–224.[9] HYERS, D. H.—ISAC, G.—RASSIAS, Th. M.: Stability of Functional Equations in Sev-
eral Variables, Birkhauser, Basel, 1998.
[10] GAVRUTA, P.: A generalization of the Hyers-Ulam-Rassias stability of approximatelyadditive mappings, J. Math. Anal. Appl. 184 (1994), 431–436.
[11] GOLET, I.: Some remarks on functions with values in probabilistic normed spaces, Math.Slovaca 57 (2007), 259–270.
226
STABILITY OF SOME FUNCTIONAL EQUATIONS IN MENGER ϕ-NORMED SPACES
[12] JUN, K. W.—KIM, H. M.: The generalized Hyers-Ulam-Rassias stability of a cubic func-tional equation, J. Math. Anal. Appl. 274 (2002), 867–878.
[13] JUN, K. W.—KIM, H. M.: On the stability of Euler–Lagrange type cubic mappings inquasi-Banach spaces, J. Math. Anal. Appl. 332 (2007), 1335–1350.
[14] JUN, K. W.—KIM, H. M.—CHANG, I. S.: On the Hyers–Ulam stability of an Euler-Lagrange type cubic functional equation, J. Comput. Anal. Appl. 7 (2005), 21–33.
[15] JUN, K. W.—LEE, Y. H.: On the Hyers-Ulam-Rassias stability of a pexiderized quadraticinequality, Math. Inequal. Appl. 4 (2001), 93–118.
[16] JUNG, S.M.: Hyers-Ulam-Rassias stability of Jensen’s equation and its application, Proc.Amer. Math. Soc. 126 (1998), 3137–3143.
[17] JUNG, S. M.: Hyers-Ulam-Rassias Stability of Functional Equations in Mathematical
Analysis, Hadronic Press, Palm Harbor, 2001.[18] JUNG, S. M.—SAHOO, P. K.: Hyers–Ulam stability of the quadratic equation of Pexider
type, J. Korean Math. Soc. 38 (2001), 645–656.[19] JUNG, C. F. K.: On generalized complete metric spaces, Bull. Amer. Math. Soc. 75
(1969), 113–116.[20] KIM, H. M.—RASSIAS, J. M. : Generalization of Ulam stability problem for Euler-
Lagrange quadratic mappings, J. Math. Anal. Appl. 336 (2007), 277–296.[21] KOMINEK, Z. : On a local stability of the Jensen functional equation, Demonstratio
Math. 22 (1989), 499–507.[22] LUXEMBURG, W. A. J.: On the convergence of successive approximations in the theory
of ordinary differential equations II, Koninklijke Nederlandse Akademie van Wetenschap-pen Proc. Ser. A 61; Indag. Math. (N.S.) 20 (1958), 540–546.
[23] MIHET, D.—RADU, V. : On the stability of the additive Cauchy functional equation inrandom normed spaces, J. Math. Anal. Appl. 343 (2008), 567–572.
[24] MIHET, D. : The stability of the additive Cauchy functional equation in non-Archimedean fuzzy normed spaces, Fuzzy Sets and Systems 161 (2010), 2206–2212.
[25] MIHET, D.: The fixed point method for fuzzy stability of the Jensen functional equation,Fuzzy Sets and Systems 160 (2009), 1663–1667.
[26] MIHET, D.: The probabilistic stability for a functional equation in a single variable, ActaMath. Hungar. 123 (2009), 249–256.
[27] MIHET, D.—SAADATI, R.—VAEZPOUR, S. M.: On the stability of an additive func-tional equation in Menger probabilistic ϕ-normed spaces via fixed points, Math. Slovaca61 (2011), 817–826.
[28] MIHET, D.—SAADATI, R.: On the stability of the additive Cauchy functional equation
in random normed spaces, Appl. Math. Lett. 24 (2011), 2005–2009.[29] MIRMOSTAFAEE, A. K.—MOSLEHIAN, M. S.: Fuzzy versions of Hyers-Ulam-Rassias
theorem, Fuzzy Sets and Systems 159 (2008), 720–729.[30] MIRMOSTAFAEE, A. K.: A fixed point approach to almost quartic mappings in quasi-
fuzzy normed spaces, Fuzzy Sets and Systems 160 (2009), 1653–1662.
[31] PARK, C.: On the Hyers-Ulam-Rassias stability of generalized quadratic mappings inBanach modules, J. Math. Anal. Appl. 291 (2004), 214–223.
[32] PERNAMI, J. C.—VASUDEVA, H. L.: On Jensen’s functional equation, AequationesMath. 43 (1992), 211–218.
[33] RADU, V.: The fixed point alternative and the stability of functional equations, FixedPoint Theory 4 (2003), 91–96.
[34] RASSIAS, Th. M.: On the stability of the linear mapping in Banach spaces, Proc. Amer.Math. Soc. 72 (1978), 297–300.
[35] RASSIAS, Th. M.: On the stability of the quadratic functional equation and its applica-tions, Studia Univ. Babes-Bolyai Math. 43 (1998), 89–124.
227
DOREL MIHET — REZA SAADATI
[36] RASSIAS, Th. M.: Solution of the Ulam stability problem for quartic mappings, Glas.Mat. Ser. III 34 (1999), 243–252.
[37] RASSIAS, Th. M.: Functional Equations, Inequalities and Applications, Kluwer Aca-demic Publishers, Dordrecht-Boston-London, 2003.
[38] SCHWEIZER, B.—SKLAR, A.: Probabilistic Metric Spaces, Elsevier, North Holand,New York, 1983.
[39] SKOF, F.: Local properties and approximations of operators, Rend. Sem. Mat. Fis. Milano53 (1983) 113–129.
[40] ULAM, S. M.: Problems in Modern Mathematics. Science Editions, Wiley, New York,1964 (Chapter VI).
[41] ZAHARIA, C.—MIHET, D.: On the probabilistic stability of some functional equations,
Carpathian J. Math. 29 (2013), 125–132.[42] ZAHARIA, C.—MIHET, D.: A note on the definition of a generalized fuzzy normed
space, Fuzzy Sets and Systems 238 (2014), 129–134.
Received 1. 9. 2011Accepted 10. 10. 2011
*Faculty of Mathematics and Computer ScienceWest University of TimisoaraBv. V. Parvan 4
RO–300223 TimisoaraROMANIA
E-mail : [email protected]
**Department of Mathematics
Iran University of Science and TechnologyTehranIRAN
E-mail : [email protected]
228