Volume 2010, Article ID 897123,14pages doi:10.1155/2010/897123
Research Article
On the Superstability of the Pexider Type
Trigonometric Functional Equation
Gwang Hui Kim
Department of Mathematics, Kangnam University, Yongin, Gyeonggi 446-702, South Korea
Correspondence should be addressed to Gwang Hui Kim,[email protected]
Received 23 October 2009; Accepted 3 January 2010
Academic Editor: Yeol Je Cho
Copyrightq2010 Gwang Hui Kim. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
We will investigate the superstability of thehyperbolictrigonometric functional equation from the following functional equations:fxy±gx−y λfxgyandfxy±gx−y
λgxfy, which can be considered the mixed functional equations of the sine function and cosine function, the hyperbolic sine function and hyperbolic cosine function, and the exponential functions, respectively.
1. Introduction
Baker et al. in1stated the following: iff satisfies the inequality|E1f−E2f| ≤ ε, then eitherfis bounded orE1f E2f. This is frequently referred to as superstability.
The superstability of the cosine functional equation also called the d’Alembert equation
fxyfx−y2fxfy, C
and the sine functional equation
fxfyf xy
2
2
−f x−y
2
2
, S
were investigated by Baker2and Cholewa3, respectively. Their results were improved by Badora4, Badora and Ger5, G˘avrut¸a6, and Kimsee7,8.
The superstability of the Wilson equation
fxyfx−y2fxgy, Cfg
The superstability of the trigonometric functional equation concerned with the sine and the cosine equations
fxy−fx−y2fxfy, T
fxy−fx−y2fxgy, Tfg
fxy−fx−y2gxfy, Tgf
fxy−fx−y2gxhy, Tgh
was investigated by Kim10,11, Kim and Lee12.
The hyperbolic cosine function, hyperbolic sine function, hyperbolic trigonometric function, and some exponential functions also satisfy the above mentioned equations; thus they can be called by the hyperbolic cosine sine, trigonometric, exponential functional equations, respectively.
For example,
coshxycoshx−y2 coshxcoshy,
sinhxysinhx−y2 sinhxcoshy,
sinh2
xy
2
−sinh2
x−y
2
sinhxsinhy,
caxycax−y2cax
2
aya−y2cexaya−y
2 ,
exyex−y2ex
2
eye−y2excoshy,
nxycnx−yc2nxc : Jensen equation,forfx nxc,
1.1
whereaandcare constants.
The aim of this paper is to investigate the superstability of the hyperbolic sine functional equationSfrom the following functional equations:
fxygx−yλfxgy, Cfgfg
fxygx−yλgxfy, Cfggf
fxy−gx−yλfxgy, Tfgfg
on the abelian group. As corollaries, we obtain the superstability ofSfrom the following functional equations:
fxygx−yλfxfy, Cfgff
fxygx−yλgxgy, Cfggg
fxy−gx−yλfxfy, Tfgff
fxy−gx−yλgxgy. Tfggg
Furthermore, the obtained results can be extended to the Banach space.
In this paper, letG,be a uniquely 2-divisible Abelian group,Cthe field of complex numbers, and Rthe field of real numbers. Whenever we deal withC, G,only needs Abelian which is not 2-divisibility.
We may assume thatfandgare nonzero functions,λ, εis a nonnegative real constant, andϕ : G → Ris a mapping. For simplicity, we will form the notations of the equation as follows:
fxyfx−yλfxfy, Cλ
fxyfx−yλfxgy, Cλfg
gxygx−yλgxgy, Cλg
gxygx−yλgxfy. Cλgf
2. Superstability of the Functional Equations
In this section, we will investigate the superstability of the hyperbolic sine functional equationSfrom the functional equationsCfgfg,Cfggf,Tfgfg, andTfggfunder the conditions from which the differences of each equation are bounded byϕxandϕy.
Theorem 2.1. Suppose thatf, g:G → Csatisfy the inequality f
xygx−y−λfxgy≤ϕx ∀x, y∈G. 2.1
Then, eithergwithf0 0is bounded orfsatisfiesS. Particularly, ifgsatisfiesCλ, then fandgare the solutions of the Wilson type equationCλfg.
Proof. Letgbe the unbounded solution of the inequality2.1. Then, there exists a sequence
{yn}inGsuch that 0/|gyn| → ∞asn → ∞.
Takingy yn in the inequality2.1, dividing both sides by|λgyn|,and passing to
the limit asn → ∞,we obtain the following:
fx lim
n→ ∞
fxyngx−yn
Using2.1, we have
f
xyyngx−yyn−λfxgyyn
fx−yyngx−−yyn−λfxg−yyn≤2ϕx, 2.3
and thus,
fxyyngxy−yn λgyn
f
x−yyngx−y−yn
λgyn −λfx·
gyyng−yyn λgyn
≤ 2ϕx λgyn
2.4
for allx, y∈G.
We conclude that, for everyy∈G, there exists a limit function
k1
y: lim
n→ ∞
gyyng−yyn
λgyn , 2.5
where the functionk1:G → Csatisfies the equation
fxyfx−yλfxk1
y ∀x, y∈G. 2.6
Applying the casef0 0 in2.6, which implies thatf is odd and keeping this in mind, by means of2.6, we infer the equality
fxy2−fx−y2λfxk 1
yfxy−fx−y
fxfx2y−fx−2y
fxf2yxf2y−x
λfxf2yk1x.
2.7
Puttingyxin2.6, we obtain the equation
f2x λfxk1x, x∈G. 2.8
This, in return, leads to the equation
fxy2−fx−y2f
being valid for allx, y∈G,which, in the light of the unique 2-divisibility ofG, states nothing else butS.
Particularly, ifgsatisfiesCλ, the limitk1states nothing else butg; thus,2.6validates the required equationCλfg.
Corollary 2.2. Suppose thatf, g:G → Csatisfy the inequality
fxygx−y−λfxfy≤ϕx ∀x, y∈G. 2.10
Then, eitherfwithf0 0is bounded orfsatisfiesS.
Proof. Replacinggybyfyin2.1ofTheorem 2.1, an obvious slight change in the proof steps applied inTheorem 2.1allows us to show thatfsatisfiesS.
Namely, forfbe unbounded, there exists a sequence{yn}inGsuch that 0/|fyn| →
∞as n → ∞. Takingy yn in the inequality 2.1, dividing both sides by |λfyn|,and
passing to the limit asn → ∞,we obtain
fx lim
n→ ∞
fxyngx−yn
λfyn , x∈G. 2.11
A similar procedure to that applied after formula2.2yields the required result by using of
2.6.
Theorem 2.3. Suppose thatf, g:G → Csatisfy the inequality
fxygx−y−λgxfy≤ϕx ∀x, y∈G. 2.12
Then, eitherfwithg0 0is bounded orgsatisfiesS. Particularly, iffsatisfiesCλ, then gandfare the solutions of equationCλgf.
Proof. For the unboundedf, we can choose a sequence{yn}inGsuch that 0/|fyn| → ∞
asn → ∞.
The same reasoning to the proof applied inTheorem 2.1for2.12withyyngives
gx lim
n→ ∞
fxyngx−yn
λfyn , x∈G. 2.13
Substitutingyynand−yynforyin2.12, and dividing by|λfyn|,then it gives
us the existence of a limit function
k2
y: lim
n→ ∞
fyynf−yyn
where the functionk2:G → Csatisfies the equation
gxygx−yλgxk2
y ∀x, y∈G. 2.15
Applying the caseg0 0 in2.15, it implies thatgis odd.
A similar procedure to that applied after formula 2.6 allows us to show that g satisfiesS.
Particularly, iffsatisfiesCλ, the limitk2states nothing else butf; thus, the required equationCλgfholds from2.15.
Corollary 2.4. Suppose thatf, g:G → Csatisfy the inequality
fxygx−y−λgxgy≤ϕx ∀x, y∈G. 2.16
Then, eithergwithg0 0is bounded orgsatisfiesS.
Proof. Substitutinggyforfyin2.12ofTheorem 2.3, the next of the proof runs along that of the above theorem.
Theorem 2.5. Suppose thatf, g:G → Csatisfy the inequality f
xygx−y−λfxgy≤ϕy ∀x, y∈G. 2.17
Then, eitherfwithg0 0is bounded orgsatisfiesS. Particularly, iffsatisfiesCλ, then gandfare solutions of the Wilson type equationCλgf.
Proof. For the unboundedfof the inequality2.17, we can choose a sequence{xn}inGsuch
that 0/|fxn| → ∞asn → ∞.
Takingxxnin the inequality2.17, dividing both sides by|λfxn|, and passing to
the limit asn → ∞,we obtain
gy lim
n→ ∞
fxnygxn−y
λfxn , x∈G. 2.18
In2.17, replacingxbyxnyandxn−y, replacingybyx, and dividing by|λfxn|,
it then gives us the existence of a limit function
k3x:nlim→ ∞
fxnyfxn−y
λfxn , 2.19
where the functionk3:G → Csatisfies the equation
gxygx−yλgxk3
y ∀x, y∈G. 2.20
Applying the caseg0 0 in2.20, it implies thatg is odd. Since2.20equals to
thatg satisfiesS. Particularly, iff satisfiesCλ, then the limitk3states nothing else butf, thus, the required equationCλfgholds from2.20.
Corollary 2.6. Suppose thatf, g:G → Csatisfy the inequality
fxygx−y−λfxfy≤ϕy, ∀x, y∈G. 2.21
Then, eitherfwithf0 0is bounded orfsatisfiesS.
Proof. Substitutingfyforgyin2.17of Theorem 2.5, asCorollary 2.4, we then obtain the required result from the above theorem.
Theorem 2.7. Suppose thatf, g:G → Csatisfy the inequality
fxygx−y−λgxfy≤ϕy, ∀x, y∈G. 2.22
Then, eithergwithf0 0is bounded orfsatisfiesS. Particularly, ifgsatisfiesCλ, then fandgare the solutions of equationCλfg.
Proof. For the unboundedg, we can choose a sequence{xn}inGsuch that 0/|gxn| → ∞
asn → ∞.
For2.22withxxn, the same reasoning as the proof applied inTheorem 2.1gives
us
fy lim
n→ ∞
fxnygxn−y
λgxn , x∈G. 2.23
In2.22, replacingxbyxnyandxn−y,replacingxbyy, and dividing by|λgxn|,
it then gives us the existence of a limit function
k4
y: lim
n→ ∞
gxnygxn−y
λgxn , 2.24
where the functionk4:G → Csatisfies the equation
fxyfx−yλfxk4
y, ∀x, y∈G. 2.25
Since2.25is the same as2.6, the next proof runs along that ofTheorem 2.1.
Corollary 2.8. Suppose thatf, g:G → Csatisfy the inequality
fxygx−y−λgxgy≤ϕy, ∀x, y∈G. 2.26
Proof. Substitutinggyforfyin2.22ofTheorem 2.7, the next proof runs along that of the above theorem.
The casesϕx,ϕyin the following result follow the procedure applied in Theorems
2.1and2.5, respectively. In the obtained result, applying the casesλ2 andϕx ϕy ε, then they are founded in2,4–6.
Corollary 2.9. Suppose thatf:G → Csatisfy the inequality
f
xyfx−y−λfxfy≤
⎧ ⎨ ⎩
ϕx,
ϕy, ∀x, y∈G. 2.27
Then, eitherfis bounded orfsatisfiesCλ.
Proof. Forf being unbounded, we can choose two sequences{xn}and{yn}inGsuch that
0/|fxn|and|fyn| → ∞asn → ∞.
(i) case
ϕ
x
Takingyynin inequality2.27, dividing it by|λfyn|, and passing to the limit asn → ∞,
we obtain
fx lim
n→ ∞
fxynfx−yn
λfyn , x∈G. 2.28
In2.27, replacingybyyyn and−yyn,and dividing by|λfyn|, it then gives,
with the application of2.28, thatfsatisfiesCλ.
(ii) case
ϕ
y
For the chosen sequence{xn}, the procedure asiimplies
fy lim
n→ ∞
fxnyfxn−y
λfxn , x∈G. 2.29
In2.27, replacingxbyxnyandxn−yand replacingybyx, the other procedure is
the same asi.
Since the proofs of the following results Theorems 2.10–2.16 and Corollaries 2.11–
2.17for the functional equationsTfgfg,Tfggf,Tfgff, andTfgggare, respectively, the same processes those of Theorems2.1–2.7and Corollaries2.2–2.8, as we will skip their proofs.
Theorem 2.10. Suppose thatf, g:G → Csatisfy the inequality f
Then, eithergwithf0 0is bounded orfsatisfiesS. Particularly, ifgsatisfiesCλ, then fandgare solutions of the Wilson type equationCλfg.
Corollary 2.11. Suppose thatf, g:G → Csatisfy the inequality
fxy−gx−y−λfxfy≤ϕx ∀x, y∈G. 2.31
Then, eitherfwithf0 0is bounded orfsatisfiesS.
Theorem 2.12. Suppose thatf, g:G → Csatisfy the inequality
f
xy−gx−y−λgxfy≤ϕx ∀x, y∈G. 2.32
Then, eitherfwithg0 0is bounded orgsatisfiesS. Particularly, iffsatisfiesCλ, then gandfare solutions of the Wilson type equationCλgf.
Corollary 2.13. Suppose thatf, g:G → Csatisfy the inequality
fxy−gx−y−λgxgy≤ϕx ∀x, y∈G. 2.33
Then, eithergwithg0 0is bounded orgsatisfiesS.
Theorem 2.14. Suppose thatf, g:G → Csatisfy the inequality
fxy−gx−y−λfxgy≤ϕy ∀x, y∈G. 2.34
Then, eitherfwithg0 0is bounded orgsatisfiesS. Particularly, iffsatisfiesCλ, then gandfare solutions of the Wilson type equationCλgf.
Corollary 2.15. Suppose thatf, g:G → Csatisfy the inequality
fxy−gx−y−λfxfy≤ϕy ∀x, y∈G. 2.35
Then, eitherfwithf0 0is bounded orfsatisfiesS.
Theorem 2.16. Suppose thatf, g:G → Csatisfy the inequality
fxy−gx−y−λgxfy≤ϕy ∀x, y∈G. 2.36
Corollary 2.17. Suppose thatf, g:G → Csatisfy the inequality
fxy−gx−y−λgxgy≤ϕy ∀x, y∈G. 2.37
Then, eithergwithg0 0is bounded orgsatisfiesS.
The cases ϕx and ϕy in the following result follow the procedure applied in Theorems 2.10and 2.14, respectively. In the obtained result, applying the casesλ 2 and
ϕx ϕy ε, then they are founded in10–12.
Corollary 2.18. Suppose thatf :G → Csatisfy the inequality
fxy−fx−y−λfxfy≤
⎧ ⎨ ⎩
ϕx,
ϕy, ∀x, y∈G. 2.38
Thenfis bounded.
Proof. Forf being unbounded, we can choose two sequences{xn}and{yn}inGsuch that
0/|fxn|and|fyn| → ∞asn → ∞.
(i) case
ϕ
x
First, going though the same process of the caseiofCorollary 2.9, then we obtain thatf satisfiesCλ.
Secondly, from the chosen sequence{yn}, we obtain
fx lim
n→ ∞
fxyn−fx−yn
λfyn , x∈G. 2.39
Replacingxbyxyn andx−ynin2.38, then we obtain, from their difference, the
inequality
fxyyn−fxy−yn λfyn
−f
x−yyn−fx−y−yn λfyn −λ·
fxyn−fx−yn λfyn ·f
y
≤ 2ϕx λfyn,
2.40
which gives, with the application of2.39, the equation
Thus, since the functionfsatisfies two equationsCλandTλ, the functionf states nothing else but zero. It is a contradiction assuming thatfis nonzero. Thusfis bounded.
(ii) case
ϕ
y
For the chosen sequence {xn}, the same procedure as in the above case i gives us the
required result.
Remark 2.19. aSubstitutingf forgof the second term of the stability inequalities in all the results ofSection 2, then we obtain the same number of corollaries, which are the stability of the hyperbolic cosine type functional equations Cλfg, Cλgf, and the hyperbolic trigonometric type functional equationsfx y−fx−y λ fxgy, fx y−
fx−y λ gxfy.
bApplying the caseλ 2 in all the results ofSection 2anda’s application, then we obtain the same number of corollaries. Some of their stabilities were founded in papers
6,7,9–11.
cApplyingϕx ϕy εin all the results ofSection 2anda’s application, then we obtain the same number of corollaries. Some of their stabilities were founded in papers
6,7,9–12.
dApplyingλ2 andϕx ϕy εin all the results ofSection 2,a’s,b’s, and
c’s applications, then we obtain the same number of corollaries. Some of their stabilities were founded in papers5–7,9–12.
3. Extension to the Banach Space
In all the results presented inSection 2, the range of functions on the abelian group can be extended to the Banach space. For simplicity, we will only prove the plus case of3.1 of
Theorem 3.1. The other cases are similar to this, thus their proofs will be omitted.
Theorem 3.1. LetE, · be a semisimple commutative Banach space. Assume thatf, g :G → E
satisfy one of each inequalities
fxy±gx−y−λfxgy ≤ϕx, 3.1
fxy±gx−y−λgxfy ≤ϕx 3.2
for allx, y∈G.For an arbitrary linear multiplicative functionalx∗∈E∗, then,
icase3.1, eitherx∗◦g withf0 0is bounded orfsatisfiesS. Particularly, ifg satisfiesCλ, thenfandgare the solutions of the Wilson type equationCλfg.
iicase3.2, eitherfwithg0 0is bounded orgsatisfiesS. Particularly, iffsatisfies
Cλ, thengandfare the solutions of the Wilson type equationCλgf.
Assume that3.1holds and arbitrarily fixes a linear multiplicative functionalx∗∈E∗. As is well known, we havex∗1,hence, for everyx, y∈G, we have
ϕx≥fxygx−y−λfxgy
sup
y∗1
y∗fxygx−y−λfxgy
≥x∗fxyx∗gx−y−2x∗fxx∗gy,
3.3
which states that the superpositionsx∗◦fandx∗◦gyield a solution of inequality2.1. Since, by assumption, the superpositionx∗◦gwithf0 0 is unbounded, an appeal toTheorem 2.1
shows that the two results hold.
First, the functionx∗◦fsolvesS. In other words, bearing the linear multiplicativity ofx∗in mind, for allx, y∈G, the differenceDx, y:G×G → Cdefined by
DSx, y:f
xy
2
2
−f x−y
2
2
−fxfy, 3.4
falls into the kernel ofx∗.Therefore, in view of the unrestricted choice ofx∗,we infer that
DSx, y∈kerx∗:x∗ is a multiplicative member of E∗ 3.5
for allx, y∈G.Since the algebraEhas been assumed to be semisimple, the last term of the above formula coincides with the singleton{0},that is,
DSx, y0, ∀x, y∈G, 3.6
as claimed.
Second, in particular, ifx∗◦gsatisfiesCλ, thenx∗◦fandx∗◦gare solutions of the Wilson type equationCλfg. This means that
DCλ fg
x, y:fxyfx−y−λfxgy, 3.7
falls into the kernel ofx∗. Through the above process, we obtain
DCλ fg
x, y0, ∀x, y∈G, 3.8
as claimed. The minus case is also similar.
Corollary 3.2. LetE, · be a semisimple commutative Banach space. Assume thatf, g:G → E
satisfy one of each inequalities
fxy±gx−y−λfxfy≤ϕx, 3.9
f
xy±gx−y−λgxgy≤ϕx 3.10
for allx, y∈G.For an arbitrary linear multiplicative functionalx∗∈E∗,
iin case3.9, eitherx∗◦fwithf0 0is bounded orfsatisfiesS.
iiin case3.10, eitherx∗◦gwithg0 0is bounded orgsatisfiesS.
Theorem 3.3. LetE, · be a semisimple commutative Banach space. Assume thatf, g :G → E
satisfy one of each inequalities
fxy±gx−y−λfxgy≤ϕy, 3.11
fxy±gx−y−λgxfy≤ϕy 3.12
for allx, y∈G.For an arbitrary linear multiplicative functionalx∗∈E∗, then,
iin case3.11, eitherx∗◦fwithg0 0is bounded orgsatisfiesS. Particularly, if fsatisfiesCλ, thengandfare solutions of the Wilson type equationCλgf.
iiin case3.12, eitherg withf0 0is bounded orfsatisfiesS. Particularly, ifg satisfiesCλ, thenfandgare solutions of the Wilson type equationCλfg.
Corollary 3.4. LetE, · be a semisimple commutative Banach space. Assume thatf, g:G → E
satisfy one of each inequalities
fxy±gx−y−λfxfy≤ϕy, 3.13
f
xy±gx−y−λgxgy≤ϕy 3.14
for allx, y∈G.For an arbitrary linear multiplicative functionalx∗∈E∗,
iin case3.13, eitherx∗◦fwithf0 0is bounded orfsatisfiesS.
iiin case3.14, eitherx∗◦gwithg0 0is bounded orgsatisfiesS.
Remark 3.5. We obtain the same number of corollaries on the Banach space for all the theorems mentioned inSection 2and all the results obtained by applying ofa,b,c, andd in
References
1 J. Baker, J. Lawrence, and F. Zorzitto, “The stability of the equationfxy fxfy,”Proceedings of the American Mathematical Society, vol. 74, no. 2, pp. 242–246, 1979.
2 J. A. Baker, “The stability of the cosine equation,”Proceedings of the American Mathematical Society, vol. 80, no. 3, pp. 411–416, 1980.
3 P. W. Cholewa, “The stability of the sine equation,”Proceedings of the American Mathematical Society, vol. 88, no. 4, pp. 631–634, 1983.
4 R. Badora, “On the stability of the cosine functional equation,”Rocznik Naukowo-Dydaktyczny, no. 15, pp. 5–14, 1998.
5 R. Badora and R. Ger, “On some trigonometric functional inequalities,” inFunctional Equations— Results and Advances, vol. 3 ofAdv. Math. (Dordr.), pp. 3–15, Kluwer Academic Publishers, Dordrecht, The Netherlands, 2002.
6 P. G˘avrut¸a, “On the stability of some functional equations,” inStability of Mappings of Hyers-Ulam Type, Th. M. Rassias and J. Tabor, Eds., Hadronic Press Collection of Original Articles, pp. 93–98, Hadronic Press, Palm Harbor, Fla, USA, 1994.
7 G. H. Kim, “The stability of d’Alembert and Jensen type functional equations,”Journal of Mathematical Analysis and Applications, vol. 325, no. 1, pp. 237–248, 2007.
8 G. H. Kim, “A stability of the generalized sine functional equations,”Journal of Mathematical Analysis and Applications, vol. 331, no. 2, pp. 886–894, 2007.
9 Pl. Kannappan and G. H. Kim, “On the stability of the generalized cosine functional equations,” Annales Academiae Paedagogicae Cracoviensis Studia Mathematica, vol. 1, pp. 49–58, 2001.
10 G. H. Kim, “On the stability of trigonometric functional equations,”Advances in Difference Equations, Article ID 90405, 10 pages, 2007.
11 G. H. Kim, “On the stability of the Pexiderized trigonometric functional equation,” Applied Mathematics and Computation, vol. 203, no. 1, pp. 99–105, 2008.