Skip to main content

A fixed point approach to the Hyers-Ulam stability of an AQ functional equation on β-Banach modules

Abstract

In this paper, we establish the general solution and investigate the generalized Hyers-Ulam stability of the following mixed additive and quadratic functional equation: f(kx+ly)+f(kxly)=f(kx)+f(x)+ 1 2 (k1)[(k+2)f(x)+kf(x)]+ l 2 [f(y)+f(y)] (k,lZ{0}) in β-Banach modules on a Banach algebra. In addition, we show that under some suitable conditions, an approximately mixed additive-quadratic function can be approximated by a mixed additive and quadratic mapping.

MSC:39B82, 39B52, 46H25.

1 Introduction and preliminaries

Starting with the article of Hyers [1] and continuing with those of Rassias [2], Găvruţa [3], Czervick [4] and so on, the authors used the ‘direct method’ to prove the stability properties for functional equations. Namely, the exact solution of the functional equation is explicitly constructed as a limit of a sequence starting from the given approximate solution.

On the other hand, Baker [5] used the Banach fixed point theorem to give a Hyers-Ulam stability result for a nonlinear functional equation in a single variable. In 2002, Cădariu and Radu delivered a lecture entitled ‘On the stability of the Cauchy functional equation: a fixed point approach’ in ‘The 14th European Conference on Iteration Theory - ECIT 2002, Evora, Portugal, 2002’. Their idea was to obtain, in β-normed spaces, the existence of the exact solution and the error estimations by using the fixed point alternative theorem [6]. This new method was used in two successive papers [7, 8] in 2003 to obtain the properties of generalized Hyers-Ulam stability for Jensen’s functional equation. Also, the lecture from ECIT 2002 was materialized in [9]. After that, a lot of papers used the ‘fixed point alternative’ to obtain generalized Hyers-Ulam stability results for different functional equations in various spaces. The reader is referred to the following books and research papers which provide an extensive account of the progress made on Ulam’s problem during the last seventy years (see, for instance, [1025]).

The motivation of this paper is to present a new mixed additive and quadratic (‘AQ’ for short) functional equation. We obtain the general solution of the AQ-functional equation. Moreover, we prove the generalized Hyers-Ulam stability of the AQ-functional equation in Banach modules on a Banach algebra using the fixed point method.

The functional equation

f(x+y)+f(xy)=2f(x)+2f(y)
(1.1)

is related to a symmetric bi-additive function [20]. It is natural that such an equation is called a quadratic functional equation. In particular, every solution of the quadratic equation (1.1) is said to be a quadratic function. It is well known that a function f between real vector spaces is quadratic if and only if there exists a unique symmetric bi-additive function B such that f(x)=B(x,x) for all x (see [20]). The bi-additive function B is given by B(x,y)= 1 4 (f(x+y)+f(xy)). In [4], Czerwik proved the Hyers-Ulam stability of the quadratic functional equation (1.1). A Hyers-Ulam stability problem for the quadratic functional equation (1.1) was proved by Skof for functions f: E 1 E 2 , where E 1 is a normed space and E 2 is a Banach space (see [26]). Cholewa [27] noticed that the theorem of Skof is still true if the relevant domain E 1 is replaced by an Abelian group. Grabiec in [28] has generalized the above mentioned results. The quadratic functional equation and several other functional equations are useful to characterize inner product spaces (see, for instance, [10, 17, 18, 21, 29]).

Now, we consider a mapping f:XY that satisfies the following general mixed additive and quadratic functional equation:

f ( k x + l y ) + f ( k x l y ) = f ( k x ) + f ( x ) + 1 2 ( k 1 ) [ ( k + 2 ) f ( x ) + k f ( x ) ] + l 2 [ f ( y ) + f ( y ) ] ,
(1.2)

where k,lZ{0}. It is easy to see that the function f(x)=a x 2 +bx is a solution of the functional equation (1.2).

Let β be a real number with 0<β1, and let K denote either R or C. Let X be a linear space over K. A real-valued function β is called a β-norm on X if and only if it satisfies

(βN1) x β =0 if and only if x=0;

(βN2) λ x β = | λ | β x β for all λK and all xX;

(βN3) x + y β x β + y β for all x,yX.

The pair (X, β ) is called a β-normed space (see [30]). A β-Banach space is a complete β-normed space.

For explicit later use, we recall the following result by Diaz and Margolis [6].

Theorem 1.1 Let (Ω,d) be a complete generalized metric space and J:ΩΩ be a strictly contractive mapping with the Lipschitz constant L<1, that is,

d(Jx,Jy)Ld(x,y),x,yΩ.

Then for each given xΩ, either

d ( J n x , J n + 1 x ) =,n0,

or there exists a non-negative integer n 0 such that

  1. (1)

    d( J n x, J n + 1 x)< for all n n 0 ;

  2. (2)

    the sequence { J n x} converges to a fixed point y of J;

  3. (3)

    y is the unique fixed point of J in the set Ω ={yΩd( J n 0 x,y)<};

  4. (4)

    d(y, y ) 1 1 L d(y,Jy) for all y Ω .

2 General solution

Throughout this section, X and Y will be real vector spaces. Before proceeding with the proof of Theorem 2.3, which is the main result in this section, we shall need the following two lemmas:

Lemma 2.1 If an odd mapping f:XY satisfies (1.2) for all x,yX, then f is additive.

Proof Since f satisfies the functional equation (1.2), putting x=y=0 in (1.2), we get (1 k 2 2 l 2 )f(0)=0, or f(0)=0 if k 2 +2 l 2 1. Note that, in view of the oddness of f, we have f(x)=f(x) for all xX. If x=0 in f(x)=f(x), one gets f(0)=f(0), or f(0)=0 without assuming the condition k 2 +2 l 2 1. Yet, it is assumed that f is odd. Hence, (1.2) implies the following equation:

f(kx+ly)+f(kxly)=f(kx)+kf(x)
(2.1)

for all x,yX. Letting y=0 in (2.1), we get

f(kx)=kf(x)
(2.2)

for all xX. Replacing x and y by lx and kx, respectively, in (2.1) and using (2.2), we have

f(2lx)=2f(lx)
(2.3)

for all xX. Replacing y by x l in (2.3), we get

f(2x)=2f(x)
(2.4)

for all xX. Replacing y by k y l in (2.1) and using (2.2), we obtain

f(x+y)+f(xy)=2f(x)
(2.5)

for all x,yX. Replacing x and y by x + y 2 and x y 2 , respectively, in (2.5) and using (2.4), we get

f(x+y)=f(x)+f(y)
(2.6)

for all x,yX. Therefore, the mapping f:XY is additive. □

Lemma 2.2 If an even mapping f:XY satisfies (1.2) for all x,yX, then f is quadratic.

Proof Since f satisfies the functional equation (1.2), putting x=y=0 in (1.2), we get f(0)=0. Note that, in view of the evenness of f, we have f(x)=f(x) for all xX. Hence, (1.2) implies the following equation:

f(kx+ly)+f(kxly)=f(kx)+ k 2 f(x)+2 l 2 f(y)
(2.7)

for all x,yX. Letting y=0 in (2.7) and using f(0)=0, we get

f(kx)= k 2 f(x)
(2.8)

for all xX. Replacing x and y by 0 and x, respectively, in (2.7) and using f(0)=0, we have

f(lx)= l 2 f(x)
(2.9)

for all xX. Replacing x by lx in (2.7) and using (2.9), we get

f(kx+y)+f(kxy)=f(kx)+ k 2 f(x)+2f(y)
(2.10)

for all xX. Replacing y by ky in (2.10) and using (2.10), we obtain

f(x+y)+f(xy)=2f(x)+2f(y)
(2.11)

for all x,yX. Therefore, the function f:XY is quadratic. □

Now, we are ready to find the general solution of (1.2).

Theorem 2.3 A mapping f:XY satisfies (1.2) for all x,yX if and only if there exist a symmetric bi-additive mapping B:X×XY and an additive mapping A:XY such that f(x)=B(x,x)+A(x) for all xX.

Proof If there exist a symmetric bi-additive function B:X×XY and an additive function A:XY such that f(x)=B(x,x)+A(x) for all xX, it is easy to show that

f ( k x + l y ) + f ( k x l y ) = 2 k 2 B ( x , x ) + 2 l 2 B ( y , y ) + 2 k A ( x ) = f ( k x ) + f ( x ) + 1 2 ( k 1 ) [ ( k + 2 ) f ( x ) + k f ( x ) ] + l 2 [ f ( y ) + f ( y ) ]

for all x,yX. Therefore, the function f:XY satisfies (1.2).

Conversely, we decompose f into the odd part and the even part by putting

f o (x)= f ( x ) f ( x ) 2 and f e (x)= f ( x ) + f ( x ) 2

for all xX. It is clear that f(x)= f o (x)+ f e (x) for all xX. It is easy to show that the functions f o and f e satisfy (1.2). Hence, by Lemmas 2.1 and 2.2, we achieve that the functions f o and f e are additive and quadratic, respectively. Therefore, there exists a symmetric bi-additive function B:X×XY such that f e (x)=B(x,x) for all xX (see [10]). So,

f(x)=B(x,x)+A(x)

for all xX, where A(x)= f o (x) for all xX. □

3 Approximate mixed additive and quadratic mappings

In this section, we prove the generalized Hyers-Ulam stability of the mixed additive and quadratic functional equation (1.2) using the fixed point method introduced by Radu in [7] (see also [8, 9, 3135]).

Throughout this section, let B be a unital Banach algebra with norm B , B 1 ={bB b B =1}, X be a β-normed left B-module and Y be a β-normed left Banach B-module, and let k,lZ{0} be fixed integers. For a given mapping f:XY, we define the difference operators

D b f ( x , y ) = f ( k b x + l b y ) + f ( k b x l b y ) b f ( k x ) b f ( x ) 1 2 ( k 1 ) b [ ( k + 2 ) f ( x ) + k f ( x ) ] l 2 b [ f ( y ) + f ( y ) ]

and

D ˜ b f ( x , y ) = f ( k b x + l b y ) + f ( k b x l b y ) b 2 f ( k x ) b 2 f ( x ) 1 2 ( k 1 ) b 2 [ ( k + 2 ) f ( x ) + k f ( x ) ] l 2 b 2 [ f ( y ) + f ( y ) ]

for all x,yX and b B 1 .

Theorem 3.1 Let φ: X 2 [0,) be a function such that

lim n 1 | k | n β φ ( k n x , k n y ) =0
(3.1)

for all x,yX. Let f:XY be an odd mapping such that

D b f ( x , y ) β φ(x,y)
(3.2)

for all x,yX and all b B 1 . If there exists a Lipschitz constant 0<L<1 such that

φ(kx,0) | k | β Lφ(x,0)
(3.3)

for all xX, then there exists a unique additive mapping A:XY such that

f ( x ) A ( x ) β 1 | k | β ( 1 L ) φ(x,0)
(3.4)

for all xX. Moreover, if f(tx) is continuous in tR for each fixed xX, then A is B-linear, i.e., A(bx)=bA(x) for all xX and all bB.

Proof Letting b=1 and y=0 in (3.2), we get

f ( k x ) k f ( x ) β φ(x,0)
(3.5)

for all xX. Consider the set Ω:={gg:XY,g(0)=0} and introduce the generalized metric on Ω:

d(g,h)=inf { C [ 0 , ) g ( x ) h ( x ) β C φ ( x , 0 ) , x X } .
(3.6)

It is easy to show that (Ω,d) is a complete generalized metric space (see the Theorem 2.1 of [13]). We now define a function J:ΩΩ by

(Jg)(x)= 1 k g(kx),gΩ,xX.
(3.7)

Let g,hΩ and C[0,] be an arbitrary constant with d(g,h)<C; by the definition of d, it follows

g ( x ) h ( x ) β Cφ(x,0),xX.
(3.8)

By the given hypothesis and the last inequality, one has

1 k g ( k x ) 1 k h ( k x ) β CLφ(x,0),xX.
(3.9)

Hence, it holds that d(Jg,Jh)Ld(g,h). It follows from (3.5) that d(Jf,f) 1 | k | β <. Therefore, by Theorem 1.1, J has a unique fixed point A:XY in the set Ω ={gΩd(f,g)<} such that

A(x):= lim n ( J n f ) (x)= lim n 1 k n f ( k n x )
(3.10)

and A(kx)=kA(x) for all xX. Also,

d(A,f) 1 1 L d(Jf,f) 1 | k | β ( 1 L ) .
(3.11)

This means that (3.4) holds for all xX.

Now, we show that A is additive. By (3.1), (3.2) and (3.10), we have

D 1 A ( x , y ) β = lim n 1 | k | n β D 1 f ( k n x , k n y ) β lim n 1 | k | n β φ ( k n x , k n y ) =0,

that is,

f(kx+ly)+f(kxly)=f(kx)+f(x)+ 1 2 (k1) [ ( k + 2 ) f ( x ) + k f ( x ) ] + l 2 [ f ( y ) + f ( y ) ]

for all x,yX. Therefore, by Lemma 2.1, we get that the mapping A is additive. To prove the uniqueness assertion, let us assume that there exists an additive function T:XY which satisfies (3.4). Since d(f,T)1/[ | k | β (1L)] and T is additive, we get T Ω and (JT)(x)= 1 k T(kx)=T(x) for all xX, i.e., T is a fixed point of J. Since A is the unique fixed point of J in Ω , then T=A.

Moreover, if f(tx) is continuous in tR for each fixed xX, then, by the same reasoning as in the proof of [2], A is R-linear. Since A is additive, A(rx)=rA(x) for any rational number r. Fix x 0 X and ρ Y (the dual space of Y). Consider the mapping φ:RR, φ(t):=ρ(A(t x 0 )), tR. Then φ(t+s)=φ(t)+φ(s), t,sR, i.e. φ is a group homomorphism. Moreover, φ is a Borel function because of the following reasoning. Let φ(t)= lim n ρ(f( k n t x 0 ))/ k n and put φ n (t)=ρ(f( k n t x 0 ))/ k n . Then φ n (t) are continuous functions. But φ(t) is the pointwise limit of continuous functions, thus φ(t) is a Borel function. It is a known fact that if φ:RR is a function such that φ is a group homomorphism, i.e. φ(t+s)=φ(t)+φ(s) and φ is a measurable function, then φ is continuous. Therefore, φ(t) is a continuous function. Let aR. Then a= lim n r n , where { r n } is a sequence of rational numbers. Thus

φ(at)=φ ( t lim n r n ) = lim n φ(t r n )= ( lim n r n ) φ(t)=aφ(t).

Therefore, φ(at)=aφ(t) for any aR. And then, A(ax)=aA(x) for any aR. Hence, the additive mapping A is R-linear.

Letting y=0 in (3.2), we get

2 f ( k b x ) b f ( k x ) k b f ( x ) β φ(x,0)
(3.12)

for all xX and all b B 1 . By definition of A, (3.1) and (3.12), we obtain

2 A ( k b x ) b A ( k x ) k b A ( x ) β = lim n 1 | k | n β 2 f ( k n + 1 b x ) b f ( k n + 1 x ) k b f ( k n x ) β lim n 1 | k | n β φ ( k n x , 0 ) = 0

for all xX and all b B 1 . So, 2A(kbx)bA(kx)kbA(x)=0 for all xX and all b B 1 . Since A is additive, we get A(bx)=bA(x) for all xX and all b B 1 {0}. Now, let aB{0}. Since A is R-linear,

A ( b x ) = A ( b B b b B x ) = b B A ( b b B x ) = b B b b B A ( x ) = b A ( x )

for all xX and all bB. This proves that A is B-linear. □

Corollary 3.2 Let 0<p<1, δ,θ[0,), and let f:XY be an odd mapping for which

D b f ( x , y ) β δ+θ ( x β p + y β p )
(3.13)

for all x,yX and b B 1 . Then there exists a unique additive mapping A:XY such that

f ( x ) A ( x ) β 1 | k | β | k | β p δ+ 1 | k | β | k | β p θ x β p

for all xX. Moreover, if f(tx) is continuous in tR for each fixed xX, then A is B-linear.

Proof The proof follows from Theorem 3.1 by taking φ(x,y)=δ+θ( x β p + y β p ) for all x,yX. We can choose L= | k | β ( p 1 ) to get the desired result. □

The generalized Hyers-Ulam stability problem for the case of p=1 was excluded in Corollary 3.2. In fact, the functional equation (1.2) is not stable for p=1 in (3.13) as we shall see in the following example, which is based on the example given in [36] (see also [34]).

Example 3.3 Let ϕ:CC be defined by

ϕ(x)={ x , for  | x | < 1 , 1 , for  | x | 1 .

Consider the function f:CC be defined by

f(x)= m = 0 α m ϕ ( α m x )

for all xC, where α>max{|k|,|l|}. Let

D b f ( x , y ) = f ( k b x + l b y ) + f ( k b x l b y ) b f ( k x ) b f ( x ) 1 2 ( k 1 ) b [ ( k + 2 ) f ( x ) + k f ( x ) ] l 2 b [ f ( y ) + f ( y ) ]

for all x,yC and bT:={λC|λ|=1}. Then f satisfies the functional inequality

| D b f ( x , y ) | α 2 ( k 2 + 3 + 2 l 2 ) α 1 ( | x | + | y | )
(3.14)

for all x,yC, but there do not exist an additive mapping A:CC and a constant d>0 such that |f(x)A(x)|d|x| for all xC.

It is clear that f is bounded by α α 1 on C. If |x|+|y|=0 or |x|+|y| 1 α , then

| D μ f ( x , y ) | α 2 ( k 2 + 3 + 2 l 2 ) α 1 ( | x | + | y | ) .

Now, suppose that 0<|x|+|y|< 1 α . Then there exists an integer n1 such that

1 α n + 1 |x|+|y|< 1 α n .
(3.15)

Hence,

α m |kbx±lby|<1, α m |kx|<1, α m |x|<1, α m |y|<1

for all m=0,1,,n1. From the definition of f and (3.15), we obtain that

| D b f ( x , y ) | = | m = n α m ϕ ( α m ( k b x + l b y ) ) + m = n α m ϕ ( α m ( k b x l b y ) ) b m = n α m ϕ ( α m k x ) b m = n α m ϕ ( α m x ) 1 2 ( k 1 ) b [ ( k + 2 ) m = n α m ϕ ( α m x ) + k m = n α m ϕ ( α m x ) ] + l 2 b [ m = n α m ϕ ( α m y ) + m = n α m ϕ ( α m y ) ] | α 2 ( k 2 + 3 + 2 l 2 ) α 1 ( | x | + | y | ) .

Therefore, f satisfies (3.14). Now, we claim that the functional equation (1.2) is not stable for p=1 in Corollary 3.2. Suppose, on the contrary, that there exist an additive mapping A:CC and a constant d>0 such that |f(x)A(x)|d|x| for all xC. Then there exists a constant cC such that A(x)=cx for all rational numbers x. So, we obtain that

| f ( x ) | ( d + | c | ) |x|
(3.16)

for all rational numbers x. Let sN with s+1>d+|c|. If x is a rational number in (0, α s ), then α m x(0,1) for all m=0,1,,s, and for this x, we get

f(x)= m = 0 ϕ ( α m x ) α m m = 0 s ϕ ( α m x ) α m =(s+1)x> ( d + | c | ) x,

which contradicts (3.16).

Corollary 3.4 Let t,s>0 such that λ:=t+s<1 and δ, θ be non-negative real numbers, and let f:XY be an odd mapping for which

D b f ( x , y ) β δ+θ [ x β t y β s + ( x β λ + y β λ ) ]

for all x,yX and b B 1 . Then there exists a unique additive mapping A:XY such that

f ( x ) A ( x ) β 1 | k | β | k | β λ δ+ 1 | k | β | k | β λ θ x β λ

for all xX. Moreover, if f(tx) is continuous in tR for each fixed xX, then A is B-linear.

Proof The proof follows from Theorem 3.1 by taking φ(x,y)=δ+θ[ x β r y β s +( x β λ + y β λ )] for all x,yX. We can choose L= | k | β ( λ 1 ) to get the desired result. □

The generalized Hyers-Ulam stability problem for the case of λ=1 was excluded in Corollary 3.4. Similar to Theorem 3.1, one can obtain the following theorem.

Theorem 3.5 Let φ: X 2 [0,) be a function such that

lim n | k | n β φ ( x k n , y k n ) =0

for all x,yX. Let f:XY be an odd mapping such that

D b f ( x , y ) β φ(x,y)

for all x,yX and all b B 1 . If there exists a Lipschitz constant 0<L<1 such that φ(x,0) | k | β Lφ(kx,0) for all xX, then there exists a unique additive mapping A:XY such that

f ( x ) A ( x ) β L | k | β ( 1 L ) φ(x,0)

for all xX. Moreover, if f(tx) is continuous in tR for each fixed xX, then A is B-linear.

As applications of Theorem 3.5, one can get the following Corollaries 3.6 and 3.7.

Corollary 3.6 Let r>1 and θ be a non-negative real number, and let f:XY be an odd mapping such that

D b f ( x , y ) β θ ( x β r + y β r )

for all x,yX and b B 1 . Then there exists a unique additive mapping A:XY such that

f ( x ) A ( x ) β 1 | k | β r | k | β θ x β r

for all xX. Moreover, if f(tx) is continuous in tR for each fixed xX, then A is B-linear.

Corollary 3.7 Let t,s>0 such that λ:=t+s>1 and θ be a non-negative real number, and let f:XY be an odd mapping such that

D b f ( x , y ) β θ [ x β t y β s + ( x β λ + y β λ ) ]

for all x,yX and b B 1 . Then there exists a unique additive mapping A:XY such that

f ( x ) A ( x ) β 1 | k | β λ | k | β θ x β λ

for all xX. Moreover, if f(tx) is continuous in tR for each fixed xX, then A is B-linear.

Theorem 3.8 Let φ: X 2 [0,) be a function such that

lim n 1 | k | 2 n β φ ( k n x , k n y ) =0
(3.17)

for all x,yX. Let f:XY be an even mapping with f(0)=0 such that

D ˜ b f ( x , y ) β φ(x,y)
(3.18)

for all x,yX and all b B 1 . If there exists a Lipschitz constant 0<L<1 such that

φ(kx,0) | k | 2 β Lφ(x,0)
(3.19)

for all xX, then there exists a unique quadratic mapping Q:XY such that

f ( x ) Q ( x ) β 1 | k | 2 β ( 1 L ) φ(x,0)
(3.20)

for all xX. Moreover, if f(tx) is continuous in tR for each fixed xX, then Q is B-quadratic, i.e., Q(bx)= b 2 Q(x) for all xX and all bB.

Proof Letting b=1 and y=0 in (3.18), we get

f ( k x ) k 2 f ( x ) β φ(x,0)
(3.21)

for all xX. Consider the set Ω:={gg:XY,g(0)=0} and introduce the generalized metric on Ω:

d(g,h)=inf { C [ 0 , ) g ( x ) h ( x ) β C φ ( x , 0 ) , x X } .

It is easy to show that (Ω,d) is a complete generalized metric space (see Theorem 2.1 of [13]). We now define a function J:ΩΩ by

(Jg)(x)= 1 k 2 g(kx),gΩ,xX.

Let g,hΩ and C[0,] be an arbitrary constant with d(g,h)<C; by the definition of d, it follows

g ( x ) h ( x ) β Cφ(x,0),xX.

By the given hypothesis and the last inequality, one has

1 k 2 g ( k x ) 1 k 2 h ( k x ) β CLφ(x,0),xX.

Hence, it holds that d(Jg,Jh)Ld(g,h). It follows from (3.21) that d(Jf,f)1/ | k | 2 β <. Therefore, by Theorem 1.1, J has a unique fixed point Q:XY in the set Ω ={gΩd(f,g)<} such that

Q(x):= lim n ( J n f ) (x)= lim n 1 k 2 n f ( k n x )
(3.22)

and Q(kx)= k 2 Q(x) for all xX. Also,

d(Q,f) 1 1 L d(Jf,f) 1 | k | 2 β ( 1 L ) .

This means that (3.20) holds for all xX. The mapping Q is quadratic because it satisfies equation (1.2) as follows:

D ˜ 1 Q ( x , y ) β = lim n 1 | k | 2 n β D ˜ 1 f ( k n x , k n y ) β lim n 1 | k | 2 n β φ ( k n x , k n y ) =0

for all x,yX; therefore, by Lemma 2.2, it is quadratic. To prove the uniqueness assertion, let us assume that there exists a quadratic mapping S:XY which satisfies (3.20). Since d(f,S)1/[ | k | 2 β (1L)] and S is quadratic, we get S Ω and (JS)(x)= 1 k 2 S(kx)=S(x) for all xX, i.e., S is a fixed point of J. Since Q is the unique fixed point of J in Ω , then S=Q.

Moreover, if f(tx) is continuous in tR for each fixed xX, then by the same reasoning as in the proof of [2], Q is R-quadratic. Letting y=0 in (3.18), we get

2 f ( k b x ) b 2 f ( k x ) k 2 b 2 f ( x ) β φ(x,0)
(3.23)

for all xX and all b B 1 . By definition of Q, (3.17) and (3.23), we obtain

2 Q ( k b x ) b 2 Q ( k x ) k 2 b 2 Q ( x ) β lim n 1 k 2 n β φ ( k n x , 0 ) =0

for all xX and all b B 1 . So, 2Q(kbx) b 2 Q(kx) k 2 b 2 Q(x)=0 for all xX and all b B 1 . Since Q(kx)= k 2 Q(x), we get Q(bx)= b 2 Q(x) for all xX and all b B 1 {0}. Now, let bB{0}. Since Q is R-quadratic,

Q ( b x ) = Q ( b B b b B x ) = b B 2 Q ( b b B x ) = b B 2 ( b b B ) 2 Q ( x ) = b 2 Q ( x )

for all xX and all bB. This proves that Q is B-quadratic. □

Corollary 3.9 Let 0<p<2, δ,θ[0,), and let f:XY be an even mapping with f(0)=0 such that

D ˜ b f ( x , y ) β δ+θ ( x β p + y β p )

for all x,yX and b B 1 . Then there exists a unique quadratic mapping Q:XY such that

f ( x ) Q ( x ) β 1 | k | 2 β | k | β p δ+ 1 | k | 2 β | k | β p θ x β p

for all xX. Moreover, if f(tx) is continuous in tR for each fixed xX, then Q is B-quadratic.

Proof The proof follows from Theorem 3.8 by taking φ(x,y)=δ+θ( x β p + y β p ) for all x,yX. We can choose L= | k | β ( p 2 ) to get the desired result. □

The following example shows that the generalized Hyers-Ulam stability problem for the case of p=2 was excluded in Corollary 3.9. This example is a modified version of Czerwik [4].

Example 3.10 Let ϕ:CC be defined by

ϕ(x)={ x 2 , for  | x | < 1 , 1 , for  | x | 1 .

Consider the function f:CC be defined by

f(x)= m = 0 α 2 m ϕ ( α m x )

for all xC, where α>max{|k|,|l|}. Let

D ˜ b f ( x , y ) = f ( k b x + l b y ) + f ( k b x l b y ) b 2 f ( k x ) b 2 f ( x ) 1 2 ( k 1 ) b 2 [ ( k + 2 ) f ( x ) + k f ( x ) ] l 2 b 2 [ f ( y ) + f ( y ) ]

for all x,yC and bT:={λC|λ|=1}. Then f satisfies the functional inequality

| D ˜ μ f ( x , y ) | ( k 2 + 3 + 2 l 2 ) α 4 α 2 1 ( | x | 2 + | y | 2 )
(3.24)

for all x,yC, but there do not exist a quadratic mapping Q:CC and a constant d>0 such that |f(x)Q(x)|d | x | 2 for all xC.

It is clear that f is bounded by α 2 α 2 1 on C. If | x | 2 + | y | 2 =0 or | x | 2 + | y | 2 1 α 2 , then

| D ˜ μ f ( x , y ) | ( k 2 + 3 + 2 l 2 ) α 4 α 2 1 ( | x | 2 + | y | 2 ) .

Now, suppose that 0< | x | 2 + | y | 2 < 1 α 2 . Then there exists an integer n1 such that

1 α 2 ( n + 2 ) | x | 2 + | y | 2 < 1 α 2 ( n + 1 ) .
(3.25)

Hence,

α m |kbx±lby|<1, α m |kx|<1, α m |x|<1, α m |y|<1

for all m=0,1,,n1. From the definition of f and the inequality (3.25), we obtain that

| D ˜ μ f(x,y)| ( k 2 + 3 + 2 l 2 ) α 4 α 2 1 ( | x | 2 + | y | 2 ) .

Now, we claim that the functional equation (1.2) is not stable for p=2 in Corollary 3.9. Suppose, on the contrary, that there exist a quadratic mapping Q:CC and a constant d>0 such that |f(x)Q(x)|d | x | 2 for all xC. Then there exists a constant cC such that Q(x)=c x 2 for all rational numbers x. So, we obtain that

| f ( x ) | ( d + | c | ) | x | 2
(3.26)

for all rational numbers x. Let sN with s+1>d+|c|. If x is a rational number in (0, α s ), then α m x(0,1) for all m=0,1,,s, and for this x, we get

f(x)= m = 0 ϕ ( α m x ) α 2 m m = 0 s ϕ ( α m x ) α 2 m =(s+1) x 2 > ( d + | c | ) x 2 ,

which contradicts (3.26).

Similar to Corollary 3.9, one can obtain the following corollary.

Corollary 3.11 Let t,s>0 such that λ:=t+s<2 and δ, θ be non-negative real numbers, and let f:XY be an even mapping with f(0)=0 such that

D ˜ b f ( x , y ) β δ+θ [ x β t y β s + ( x β λ + y β λ ) ]

for all x,yX and b B 1 . Then there exists a unique quadratic mapping Q:XY such that

f ( x ) Q ( x ) β 1 | k | 2 β | k | β λ δ+ 1 | k | 2 β | k | β λ θ x β λ

for all xX. Moreover, if f(tx) is continuous in tR for each fixed xX, then Q is B-quadratic.

Similar to Theorem 3.8, one can obtain the following theorem.

Theorem 3.12 Let φ: X 2 [0,) be a function such that

lim n k 2 n β φ ( x k n , y k n ) =0
(3.27)

for all x,yX. Let f:XY be an even mapping such that

D ˜ b f ( x , y ) β φ(x,y)
(3.28)

for all x,yX and all b B 1 . If there exists a Lipschitz constant 0<L<1 such that φ(x,0) k 2 β Lφ(kx,0) for all xX, then there exists a unique quadratic mapping Q:XY such that

f ( x ) Q ( x ) β L k 2 β ( 1 L ) φ(x,0)
(3.29)

for all xX. Moreover, if f(tx) is continuous in tR for each fixed xX, then Q is B-quadratic.

Remark 3.13 Let f:XY be a mapping for which there exists a function φ: X 2 [0,) satisfying (3.28). Let 0<L<1 be a constant such that φ(x,0) k 2 β Lφ(kx,0) for all xX. f(0)=0, since φ(0,0)=0.

We now prove our main theorem in this section.

Theorem 3.14 Let φ: X 2 [0,) be a function such that

lim n 1 | k | n β φ ( k n x , k n y ) =0
(3.30)

for all x,yX. Let f:XY be a mapping with f(0)=0 such that

D b f ( x , y ) β φ(x,y)and D ˜ b f ( x , y ) β φ(x,y)
(3.31)

for all x,yX and all b B 1 . If there exists a Lipschitz constant 0<L<1 such that

φ(kx,0) | k | β Lφ(x,0)
(3.32)

for all xX, then there exist a unique additive mapping A:XY and a unique quadratic mapping Q:XY such that

f ( x ) A ( x ) Q ( x ) β | k | β + 1 | k | 2 β ( 1 L ) [ φ ( x , 0 ) + φ ( x , 0 ) ]
(3.33)

for all xX. Moreover, if f(tx) is continuous in tR for each fixed xX, then A is B-linear and Q is B-quadratic.

Proof If we decompose f into the even and the odd parts by putting

f e (x)= f ( x ) + f ( x ) 2 and f o (x)= f ( x ) f ( x ) 2
(3.34)

for all xX, then f(x)= f e (x)+ f o (x). Let ψ(x,y)=[φ(x,y)+φ(x,y)]/ 2 β , then by (3.30)-(3.32) and (3.34), we have

Hence, by Theorems 3.1 and 3.8, there exist a unique additive mapping A:XY and a unique quadratic mapping Q:XY such that

f o ( x ) A ( x ) β 1 k β ( 1 L ) ψ(x,0), f e ( x ) Q ( x ) β 1 k 2 β ( 1 L ) ψ(x,0)

for all xX. Therefore,

f ( x ) A ( x ) Q ( x ) β f o ( x ) A ( x ) β + f e ( x ) Q ( x ) β 1 | k | β ( 1 L ) ψ ( x , 0 ) + 1 | k | 2 β ( 1 L ) ψ ( x , 0 ) = | k | β + 1 | k | 2 β ( 1 L ) [ φ ( x , 0 ) + φ ( x , 0 ) ]

for all xX. □

Corollary 3.15 Let 0<p<1 and δ, θ be non-negative real numbers, and let f:XY be a mapping with f(0)=0 such that

D b f ( x , y ) β δ+θ ( x β p + y β p ) and D ˜ b f ( x , y ) β δ+θ ( x β p + y β p )

for all x,yX and b B 1 . Then there exist a unique additive mapping A:XY and a unique quadratic mapping Q:XY such that

f ( x ) A ( x ) Q ( x ) β 2 ( | k | β + 1 ) | k | 2 β | k | β ( p + 1 ) [ δ + θ x β p ]

for all xX. Moreover, if f(tx) is continuous in tR for each fixed xX, then A is B-linear and Q is B-quadratic.

Thanks to Remark 3.13, by a similar method to the proof of Theorem 3.14, one can obtain the following theorem.

Theorem 3.16 Let φ: X 2 [0,) be a function such that

lim n | k | n β φ ( x k n , y k n ) =0

for all x,yX. Let f:XY be a mapping such that

D b f ( x , y ) β φ(x,y)and D ˜ b f ( x , y ) β φ(x,y)

for all x,yX and all b B 1 . If there exists a Lipschitz constant 0<L<1 such that

φ(x,0) | k | 2 β Lφ(kx,0)

for all xX, then there exist a unique additive mapping A:XY and a unique quadratic mapping Q:XY such that

f ( x ) A ( x ) Q ( x ) β ( | k | β + 1 ) L | k | 2 β ( 1 L ) [ φ ( x , 0 ) + φ ( x , 0 ) ]

for all xX. Moreover, if f(tx) is continuous in tR for each fixed xX, then A is B-linear and Q is B-quadratic.

Corollary 3.17 Let p>2 and θ be a non-negative real number, and let f:XY be a mapping for which

D b f ( x , y ) β θ ( x β p + y β p ) and D ˜ b f ( x , y ) β θ ( x β p + y β p )

for all x,yX and b B 1 . Then there exist a unique additive mapping A:XY and a unique quadratic mapping Q:XY such that

f ( x ) A ( x ) Q ( x ) β 2 ( | k | β + 1 ) | k | p β | k | 2 β θ x β p

for all xX. Moreover, if f(tx) is continuous in tR for each fixed xX, then A is B-linear and Q is B-quadratic.

Remark 3.18 The generalized Hyers-Ulam stability problem for the cases of p=1 and p=2 were excluded in Corollaries 3.15 and 3.17, respectively.

References

  1. Hyers DH: On the stability of the linear functional equation. Proc. Natl. Acad. Sci. USA 1941, 27: 222–224. 10.1073/pnas.27.4.222

    MathSciNet  Article  Google Scholar 

  2. Rassias TM: On the stability of the linear mapping in Banach spaces. Proc. Am. Math. Soc. 1978, 72: 297–300. 10.1090/S0002-9939-1978-0507327-1

    Article  Google Scholar 

  3. Găvruţa P: A generalization of the Hyers-Ulam-Rassias stability of approximately additive mappings. J. Math. Anal. Appl. 1994, 184: 431–436. 10.1006/jmaa.1994.1211

    MathSciNet  Article  Google Scholar 

  4. Czerwik S: On the stability of the quadratic mapping in normed spaces. Abh. Math. Semin. Univ. Hamb. 1992, 62: 59–64. 10.1007/BF02941618

    MathSciNet  Article  Google Scholar 

  5. Baker JA: The stability of certain functional equations. Proc. Am. Math. Soc. 1991, 112: 729–732. 10.1090/S0002-9939-1991-1052568-7

    Article  Google Scholar 

  6. Diaz JB, Margolis B: A fixed point theorem of the alternative for the contractions on generalized complete metric space. Bull. Am. Math. Soc. 1968, 74: 305–309. 10.1090/S0002-9904-1968-11933-0

    MathSciNet  Article  Google Scholar 

  7. Radu V: The fixed point alternative and the stability of functional equations. Fixed Point Theory 2003, 4: 91–96.

    MathSciNet  Google Scholar 

  8. Cădariu L, Radu V: Fixed points and the stability of Jensen’s functional equation. J. Inequal. Pure Appl. Math. 2003, 4: 1–7.

    Google Scholar 

  9. Cădariu L, Radu V: On the stability of the Cauchy functional equation: a fixed points approach. Grazer Mathematische Berichte 346. In Iteration Theory (ECIT 2002) Edited by: Ramos JS, Gronau D, Mira C, Reich L, Sharkovsky AN. 2004, 43–52.

    Google Scholar 

  10. Aczél J, Dhombres J: Functional Equations in Several Variables. Cambridge University Press, Cambridge; 1989.

    Book  Google Scholar 

  11. Agarwal RP, Xu B, Zhang W: Stability of functional equations in single variable. J. Math. Anal. Appl. 2003, 288: 852–869. 10.1016/j.jmaa.2003.09.032

    MathSciNet  Article  Google Scholar 

  12. Brzdȩk J: On approximately additive functions. J. Math. Anal. Appl. 2011, 381: 299–307. 10.1016/j.jmaa.2011.02.048

    MathSciNet  Article  Google Scholar 

  13. Cădariu L, Radu V: Fixed point methods for the generalized stability of functional equations in a single variable. Fixed Point Theory Appl. 2008., 2008: Article ID 749392. doi:10.1155/2008/749392

    Google Scholar 

  14. Eskandani GZ, Găvruţa P, Rassias JM, Zarghami R: Generalized Hyers-Ulam stability for a general mixed functional equation in quasi- β -normed spaces. Mediterr. J. Math. 2011, 8: 331–348. 10.1007/s00009-010-0082-8

    MathSciNet  Article  Google Scholar 

  15. Forti GL: Hyers-Ulam stability of functional equations in several variables. Aequ. Math. 1995, 50: 143–190. 10.1007/BF01831117

    MathSciNet  Article  Google Scholar 

  16. Hyers DH, Rassias TM: Approximate homomorphisms. Aequ. Math. 1992, 44: 125–153. 10.1007/BF01830975

    MathSciNet  Article  Google Scholar 

  17. Hyers DH, Isac G, Rassias TM: Stability of Functional Equations in Several Variables. Birkhäuser, Basel; 1998.

    Book  Google Scholar 

  18. Jung SM: Hyers-Ulam-Rassias Stability of Functional Equations in Nonlinear Analysis. Springer, New York; 2011.

    Book  Google Scholar 

  19. Jung SM: Hyers-Ulam stability of zeros of polynomials. Appl. Math. Lett. 2011, 24: 1322–1325. 10.1016/j.aml.2011.03.002

    MathSciNet  Article  Google Scholar 

  20. Kannappan P: Quadratic functional equation and inner product spaces. Results Math. 1995, 27: 368–372.

    MathSciNet  Article  Google Scholar 

  21. Kannappan P: Functional Equations and Inequalities with Applications. Springer, New York; 2009.

    Book  Google Scholar 

  22. Xu TZ, Rassias JM, Xu WX: Stability of a general mixed additive-cubic equation in F -spaces. J. Comput. Anal. Appl. 2012, 14: 1026–1037.

    MathSciNet  Google Scholar 

  23. Xu TZ, Rassias JM: Stability of general multi-Euler-Lagrange quadratic functional equations in non-Archimedean fuzzy normed spaces. Adv. Differ. Equ. 2012., 2012: Article ID 119

    Google Scholar 

  24. Kenary HA, Talebzadeh S, Rezaei H, Lee JR, Shin DY: Fuzzy Hyers-Ulam-Rassias approximation of a mixed AQ mapping. J. Inequal. Appl. 2012., 2012: Article ID 64

    Google Scholar 

  25. Kenary HA, Rezaei H, Ghaffaripour A, Talebzadeh S, Park C, Lee J: Fuzzy Hyers-Ulam stability of an additive functional equation. J. Inequal. Appl. 2011., 2011: Article ID 140

    Google Scholar 

  26. Skof F: Local properties and approximations of operators. Rend. Semin. Mat. Fis. Milano 1983, 53: 113–129. 10.1007/BF02924890

    MathSciNet  Article  Google Scholar 

  27. Cholewa PW: Remarks on the stability of functional equations. Aequ. Math. 1984, 27: 76–86. 10.1007/BF02192660

    MathSciNet  Article  Google Scholar 

  28. Grabiec A: The generalized Hyers-Ulam stability of a class of functional equations. Publ. Math. (Debr.) 1996, 48: 217–235.

    MathSciNet  Google Scholar 

  29. Czerwik S: Functional Equations and Inequalities in Several Variables. World Scientific, Singapore; 2002.

    Book  Google Scholar 

  30. Balachandran VK: Topological Algebras. Narosa Publishing House, New Delhi; 1999.

    Google Scholar 

  31. Kenary HA, Park C, Rezaei H, Jang SY: Stability of a generalized quadratic functional equation in various spaces: a fixed point alternative approach. Adv. Differ. Equ. 2011., 2011: Article ID 62

    Google Scholar 

  32. Gordji ME, Khodaei H, Rassias TM: Fixed points and stability for quadratic mappings in β -normed left Banach modules on Banach algebras. Results Math. 2012, 61: 393–400. 10.1007/s00025-011-0123-z

    MathSciNet  Article  Google Scholar 

  33. Kenary HA, Jang SY, Park C: A fixed point approach to the Hyers-Ulam stability of a functional equation in various normed spaces. Fixed Point Theory Appl. 2011., 2011: Article ID 67

    Google Scholar 

  34. Xu TZ, Rassias JM, Rassias MJ, Xu WX: A fixed point approach to the stability of quintic and sextic functional equations in quasi- β -normed spaces. J. Inequal. Appl. 2010., 2010: Article ID 423231

    Google Scholar 

  35. Xu TZ, Rassias JM, Xu WX: A fixed point approach to the stability of a general mixed additive-cubic equation on Banach modules. Acta Math. Sci. Ser. B 2012, 32: 866–892.

    MathSciNet  Article  Google Scholar 

  36. Gajda Z: On stability of additive mappings. Int. J. Math. Math. Sci. 1991, 14: 431–434. 10.1155/S016117129100056X

    MathSciNet  Article  Google Scholar 

Download references

Acknowledgements

The authors would like to thank the referees and the editors for their help and suggestions in improving this paper. The first author TZX was supported by the National Natural Science Foundation of China (NNSFC) (Grant No. 11171022).

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Tian Zhou Xu.

Additional information

Competing interests

The authors declare that they have no competing interests.

Authors’ contributions

All authors carried out the proof. All authors conceived of the study and participated in its design and coordination. All authors read and approved the final manuscript.

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 2.0 International License (https://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Reprints and Permissions

About this article

Cite this article

Xu, T.Z., Rassias, J.M. A fixed point approach to the Hyers-Ulam stability of an AQ functional equation on β-Banach modules. J Inequal Appl 2012, 204 (2012). https://doi.org/10.1186/1029-242X-2012-204

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/1029-242X-2012-204

Keywords

  • Hyers-Ulam stability
  • additive and quadratic equation
  • β-Banach module
  • fixed point method