Bull. Korean Math. Soc. 40 (2003), No. 4, pp.

565576

ON THE GENERAL SOLUTION OF A QUARTIC FUNCTIONAL EQUATION Jukang K. Chung and Prasanna K. Sahoo
Abstract. In this paper, we determine the general solution of the quartic equation f (x+2y )+f (x2y )+6f (x) = 4[f (x+y )+f (xy )+ 6f (y )] for all x, y R without assuming any regularity conditions on the unknown function f . The method used for solving this quartic functional equation is elementary but exploits an important result due to M. Hossz u [3]. The solution of this functional equation is also determined in certain commutative groups using two important results due to L. Sz ekelyhidi [5].

1. Introduction In this paper, we determine the general solution of the functional equation (1.1) f (x + 2y ) + f (x 2y ) + 6f (x) = 4 [f (x + y ) + f (x y ) + 6f (y )] for all x, y R (the set of reals). We will solve the above functional equation using an elementary technique but without using any regularity conditions. Recently, J. M. Rassias [6] investigated the Hyers-Ulam stability of the functional equation (1.1). It was mentioned in [6] that f (x) = x4 is a solution of the above functional equation because of the identity (x + 2y )4 + (x 2y )4 + 6x4 = 4 (x + y )4 + (x y )4 + 6y 4 . For the obvious reason, he called the above functional equation a quartic functional equation and any solution of the above equation a quartic function. He proves the following result: Let X be a normed linear
Received September 24, 2002. 2000 Mathematics Subject Classication: Primary 39B22. Key words and phrases: additive function, dierence operator, Fr echet functional equation, n-additive function, quartic map, and quartic functional equation.

566

Jukang K. Chung and Prasanna K. Sahoo

space and Y be a real complete normed linear space. If f : X Y satises the inequality ||f (x + 2y ) + f (x 2y ) + 6f (x) 4 [f (x + y ) + f (x y ) + 6f (y )] || for all x, y X with a constant 0 (independent of x and y ), then there exists a unique quartic function F : X Y such that ||F (x) 17 for all x X . f (x)|| 180 A function A : R R is said to be additive if A(x + y ) = A(x) + A(y ) for all x, y R (see [2]). Let n N (the set of natural numbers). A function An : Rn R is called n-additive if it is additive in each of its variable. A function An is called symmetric if An (x1 , x2 , . . . , xn ) = An (x(1) , x(2) , . . . , x(n) ) for every permutation { (1), (2), . . . , (n)} of {1, 2, . . . , n}. If An (x1 , x2 , . . . , xn ) is a n-additive symmetric map, then An (x) will denote the diagonal An (x, x, . . . , x). Further the resulting function after substitution x1 = x2 = = x = x and x +1 = x +2 = = xn = y in An (x1 , x2 , . . . , xn ) will be denoted by A ,n (x, y ). A 2-additive map is said to be biadditive map. The diagonal of a biadditive map B is the map x B (x, x). A function Q : R R is said to be quadratic if Q(x + y ) + Q(x y ) = 2Q(x) + 2Q(y ) for all x, y R. It is well known (see [1]) that a quadratic function from R into R is the diagonal of a symmetric biadditive map. For f : R R, let h be the dierence operator dened as follows: h f (x) = f (x + h) f (x) for h R.

1 n Further, let 0 h f (x) = f (x), h f (x) = h f (x) and h h f (x) = n+1 n h f (x) for all n N and all h R. Here h h denotes the composition of the operators h and n h . For any given n N {0}, the functional equation +1 n h f (x) = 0

for all x, h R is well studied. It is known (see Kuczma [4]) that in the case where one deals with functions dened in R the last functional equation is equivalent to the Fr echet functional equation (1.2) h1 ,...,hn+1 f (x) = 0

where x, h1 , . . . , hn+1 R and h1 ,...,hk = hk h1 for k = 2, 3, . . . , n + 1.

On the general solution of a quartic functional equation

567

2. Some preliminary results In this section, we will prove a couple of lemmas that will be needed to solve the quartic functional equation. Lemma 2.1. If any function f satises the quartic functional equation (1.1) for all x, y R, then it also satises the functional equation (2.1) f (x + 2y ) + f (x 2y ) = f (2x + y ) + f (2x y ) + 30f (y ) 30f (x) for all x, y R. Proof. Letting x = 0 = y in (1.1), we obtain f (0) = 0. Next, letting x = 0 and y = x in (1.1), we have (2.2) f (2x) + f (2x) 4f (x) 28f (x) = 0

for all x R. Similarly, letting x = 0 and y = x in (1.1) and then using f (0) = 0, we obtain (2.3) f (2x) + f (2x) 4f (x) 28f (x) = 0.

Hence the equations (2.2) and (2.3) imply f (x) = f (x) for all x R, that is f is an even function. Next, we interchange x with y in (1.1) and using the fact that f is even, we have (2.4) f (2x + y ) + f (2x y ) + 6f (y ) = 4 [f (x + y ) + f (x y ) + 6f (x)] . From (1.1) and (2.4), we have the asserted result and the proof of the lemma is now complete. In the following lemma, we reduce the functional equation (2.1) to the functional equation (1.2) when n = 4. Lemma 2.2. If the function F : R R satises the functional equation (2.5) F (x + 2y ) + F (x 2y ) = F (2x + y ) + F (2x y ) + 30F (y ) 30F (x) for all x, y R, then F also satises the Fr echet functional equation (2.6) for all x0 , x1 , . . . , x5 R. Proof. Let us write (2.5) as (2.7) F (x + 2y ) + F (x 2y ) = F (2x + y ) + F (2x y ) + F1 (x) + F2 (y ) x1 ,...,x5 F (x0 ) = 0

568

Jukang K. Chung and Prasanna K. Sahoo

where F1 (x) = 30F (x) and F2 (y ) = 30F (y ). Now we substitute x0 = 1 x + 2y and y1 = x 2y that is x = 2 (x0 + y1 ) and y = 1 4 (x0 y1 ) in (2.7) to get F (x0 ) + F (y1 ) = F (2.8) 5 3 3 5 x0 + y1 + F x0 + y1 4 4 4 4 1 1 (x0 + y1 ) + F2 (x0 y1 ) . + F1 2 4

Replacing x0 by x0 + x1 in (2.8), we obtain F (x0 + x1 ) + F (y1 ) 3 5 3 5 = F (x0 + x1 ) + y1 + F (x0 + x1 ) + y1 4 4 4 4 1 1 (x0 + x1 + y1 ) + F2 (x0 + x1 y1 ) . + F1 2 4

(2.9)

Subtracting (2.8) from (2.9), we have F (x0 + x1 ) F (x0 ) 5 3 5 3 = F (x0 + x1 ) + y1 F x0 + y1 4 4 4 4 3 5 3 5 +F (x0 + x1 ) + y1 F x0 + y1 4 4 4 4 1 1 + F1 (x0 + x1 + y1 ) F1 (x0 + y1 ) 2 2 1 1 + F2 (x0 + x1 y1 ) F2 (x0 y1 ) . 4 4

(2.10)

3 4 5 Letting y2 = 5 4 x0 + 4 y1 (that is, y1 = 3 y2 3 x0 ) in (2.10), we see that

(2.11)

F (x0 + x1 ) F (x0 ) 5 = F y2 + x1 F (y2 ) 4 5 4 3 5 4 +F y2 x0 + x1 F y2 x 0 3 3 4 3 3 2 1 1 2 1 + F1 y2 x0 + x1 F1 y2 x0 3 3 2 3 3 1 2 1 1 2 + F2 y2 + x0 + x1 F2 y2 + x0 . 3 3 4 3 3

On the general solution of a quartic functional equation

569

Now replacing x0 by x0 + x2 in (2.11) and subtracting (2.11) from the resulting expression, we obtain F (x0 + x1 + x2 ) F (x0 + x1 ) F (x0 + x2 ) + F (x0 ) 5 4 3 5 4 3 = F y2 (x0 + x2 ) + x1 F y2 x0 + x1 3 3 4 3 3 4 5 4 5 4 F y2 (x0 + x2 ) + F y2 x0 3 3 3 3 2 1 1 2 1 1 (2.12) + F1 y2 (x0 + x2 ) + x1 F1 y2 x0 + x1 3 3 2 3 3 2 2 1 2 1 F1 y2 (x0 + x2 ) + F1 y2 x 0 3 3 3 3 1 2 1 1 2 1 + F2 y2 + (x0 + x2 ) + x1 F2 y2 + x0 + x1 3 3 4 3 3 4 2 1 2 1 F2 y2 + (x0 + x2 ) + F2 y2 + x0 . 3 3 3 3
4 3 4 Now we substitute y3 = 5 3 y2 3 x0 (that is y2 = 5 y3 + 5 x0 ) in (2.12) to get

F (x0 + x1 + x2 ) F (x0 + x1 ) F (x0 + x2 ) + F (x0 ) 4 3 3 4 = F y3 x2 + x1 F y3 + x1 F y3 x2 3 4 4 3 1 1 1 2 y3 + x0 + x1 x2 + F (y3 ) + F1 5 5 2 3 2 1 1 F1 y3 + x0 + x1 5 5 2 1 1 2 1 2 (2.13) y3 + x0 x2 + F1 y3 + x0 F1 5 5 3 5 5 1 2 1 2 + F2 y3 + x0 + x1 + x2 5 5 4 3 1 2 1 F2 y3 + x0 + x1 5 5 4 1 2 2 F2 y3 + x0 + x2 5 5 3 1 2 + F2 y3 + x0 . 5 5

570

Jukang K. Chung and Prasanna K. Sahoo

Again we replace x0 by x0 + x3 in (2.13) and then subtracting (2.13) from the resulting expression, we have F (x0 + x1 + x2 + x3 ) F (x0 + x1 + x2 ) F (x0 + x1 + x3 ) F (x0 + x2 + x3 ) + F (x0 + x1 ) + F (x0 + x2 ) + F (x0 + x3 ) F (x0 ) 2 1 1 1 = F1 y3 + (x0 + x3 ) + x1 x2 5 5 2 3 2 1 1 1 F1 y3 + x0 + x1 x2 5 5 2 3 1 1 2 1 1 2 y3 + (x0 + x3 ) + x1 + F1 y3 + x0 + x1 F1 5 5 2 5 5 2 2 1 1 F1 y3 + (x0 + x3 ) x2 5 5 3 1 1 2 y3 + x0 x2 + F1 5 5 3 2 1 + F1 y3 + (x0 + x3 ) 5 5 1 2 y3 + x0 F1 5 5 1 2 1 2 + F2 y3 + (x0 + x3 ) + x1 + x2 5 5 4 3 1 2 1 2 F2 y3 + x0 + x1 + x2 5 5 4 3 1 2 1 F2 y3 + (x0 + x3 ) + x1 5 5 4 1 2 1 + F2 y3 + x0 + x1 5 5 4 1 2 2 1 2 2 F2 y3 + (x0 + x3 ) + x2 + F2 y3 + x0 + x2 5 5 3 5 5 3 1 2 1 2 + F2 y3 + (x0 + x3 ) F2 y3 + x0 . 5 5 5 5
1 5 1 Letting y4 = 2 5 y3 + 5 x0 (so that y3 = 2 y4 2 x0 ) in the last equation, we obtain

On the general solution of a quartic functional equation

571

F (x0 + x1 + x2 + x3 ) F (x0 + x1 + x2 ) F (x0 + x1 + x3 ) F (x0 + x2 + x3 ) + F (x0 + x1 ) + F (x0 + x2 ) + F (x0 + x3 ) F (x0 ) 1 1 1 1 1 = F1 y4 + x1 x2 + x3 F1 y4 + x1 x2 2 3 5 2 3 1 1 1 F1 y4 + x1 + x3 + F1 y4 + x1 2 5 2 1 1 1 F1 y4 x2 + x3 + F1 y4 x2 3 5 3 1 + F1 y4 + x3 F1 (y4 ) 5 1 1 2 2 1 + F2 y4 + x0 + x1 + x2 + x3 2 2 4 3 5 1 1 1 2 F2 y4 + x0 + x1 + x2 2 2 4 3 1 1 1 2 1 1 F2 y4 + x0 + x1 + x3 + F2 y4 + x0 + 2 2 4 5 2 2 1 2 2 1 1 1 F2 y4 + x0 + x2 + x3 + F2 y4 + x0 + 2 2 3 5 2 2 1 1 2 1 1 + F2 y4 + x0 + x3 F2 y4 + x0 . 2 2 5 2 2

1 x1 4 2 x2 3

Now we replace x0 by x0 + x4 in last equation and then subtracting the last equation from the resulting expression to get F (x0 + x1 + x2 + x3 + x4 ) F (x0 + x1 + x2 + x3 ) F (x0 + x1 + x2 + x4 ) F (x0 + x1 + x3 + x4 ) F (x0 + x2 + x3 + x4 ) + F (x0 + x1 + x2 ) + F (x0 + x1 + x3 ) + F (x0 + x1 + x4 ) + F (x0 + x2 + x3 ) + F (x0 + x2 + x4 ) + F (x0 + x3 + x4 ) F (x0 + x1 ) F (x0 + x2 ) F (x0 + x3 ) F (x0 + x4 ) + F (x0 ) 1 1 1 2 2 = F2 y4 + (x0 + x4 ) + x1 + x2 + x3 2 2 4 3 5 1 1 1 2 2 F2 y4 + x0 + x1 + x2 + x3 2 2 4 3 5

572

Jukang K. Chung and Prasanna K. Sahoo

1 F2 y4 + 2 1 + F2 y4 + 2 1 F2 y4 + 2 1 + F2 y4 + 2 1 + F2 y4 + 2 1 F2 y4 + 2 1 + F2 y4 + 2 1 + F2 y4 + 2 1 + F2 y4 + 2 1 F2 y4 + 2

1 1 2 (x0 + x4 ) + x1 + x2 2 4 3 1 1 2 x0 + x1 + x2 2 4 3 1 1 2 (x0 + x4 ) + x1 + x3 2 4 5 1 1 2 x0 + x1 + x3 2 4 5 1 1 1 1 (x0 + x4 ) + x1 F2 y4 + x0 + 2 4 2 2 1 2 2 (x0 + x4 ) + x2 + x3 2 3 5 1 2 2 x0 + x2 + x3 2 3 5 1 2 1 1 (x0 + x4 ) + x2 F2 y4 + x0 + 2 3 2 2 1 2 1 1 (x0 + x4 ) + x3 F2 y4 + x0 + 2 5 2 2 1 1 1 (x0 + x4 ) + F2 y4 + x0 . 2 2 2

1 x1 4

2 x2 3 2 x3 5

1 First we let y5 = 1 2 y4 + 2 x0 and then replace x0 by x0 + x5 in the last equation. Further subtracting the last equation from the resulting equation, we obtain

F (x0 + x1 + x2 + x3 + x4 + x5 ) F (x0 + x1 + x2 + x3 + x4 ) F (x0 + x1 + x2 + x3 + x5 ) F (x0 + x1 + x2 + x4 + x5 ) F (x0 + x1 + x3 + x4 + x5 ) F (x0 + x2 + x3 + x4 + x5 ) + F (x0 + x1 + x2 + x3 ) + F (x0 + x1 + x2 + x4 ) + F (x0 + x1 + x2 + x5 ) + F (x0 + x1 + x3 + x4 ) + F (x0 + x1 + x3 + x5 ) + F (x0 + x1 + x4 + x5 ) + F (x0 + x2 + x3 + x4 ) + F (x0 + x2 + x3 + x5 ) + F (x0 + x2 + x4 + x5 ) + F (x0 + x3 + x4 + x5 ) F (x0 + x1 + x2 ) F (x0 + x1 + x3 ) F (x0 + x1 + x4 ) F (x0 + x1 + x5 ) F (x0 + x2 + x3 ) F (x0 + x2 + x4 )

On the general solution of a quartic functional equation

573

F (x0 + x2 + x5 ) F (x0 + x3 + x4 ) F (x0 + x3 + x5 ) F (x0 + x4 + x5 ) + F (x0 + x1 ) + F (x0 + x2 ) + F (x0 + x3 ) + F (x0 + x4 ) + F (x0 + x5 ) F (x0 ) = 0 which is (2.6). The proof of the lemma is now complete. The following lemma is a special case of a more general result due to Hosszu [3], and will be instrumental in determining the general solution of (1.1). Lemma 2.3. The map F from R into R satises the functional equation (2.6) for all x0 , x1 , x2 , x3 , x4 , x5 R if and only if F is given by (2.14) F (x) = A4 (x) + A3 (x) + A2 (x) + A1 (x) + A0 (x), where A0 (x) = A0 is an arbitrary constant and An (x) is the diagonal of a n-additive symmetric function An : Rn R for n = 1, 2, 3, 4. 3. Solution of equation (1.1) on reals Now we are ready to prove our main theorem. Theorem 3.1. The function f : R R satises the quartic functional equation (1.1) for all x, y R, if and only if f is of the form f (x) = A4 (x), where A4 (x) is the diagonal of a 4-additive symmetric function A4 : R4 R. Proof. From Lemma 2.1 we see that the functional equation (1.1) implies equation f (x + 2y ) + f (x 2y ) = f (2x + y ) + f (2x y ) + 30f (y ) 30f (x). From Lemma 2.2 we see that f satises the Fr echet functional equation (3.1) x1 ,...,x5 f (x0 ) = 0 for all x0 , x1 , . . . , x5 R. The general solution of the equation (3.1) can be obtained from Lemma 2.3 as (3.2) f (x) = A4 (x) + A3 (x) + A2 (x) + A1 (x) + A0 (x), where A0 (x) = A0 is an arbitrary constant and An (x) is the diagonal of the n-additive symmetric map An : Rn R for n = 1, 2, 3, 4.

574

Jukang K. Chung and Prasanna K. Sahoo

A1 (x) (3.3)

Since f is an even function (see Lemma 2.1), we have A3 (x) 0 and 0. Hence from (3.2), we have f (x) = A4 (x) + A2 (x) + A0 .

Letting (3.3) into (1.1) and noting that A4 (2y ) = 16 A4 (y ), A2 (2y ) = 4 A2 (y ), A2,2 (x, 2y ) = 4 A2,2 (x, y ), A4 (x + y ) + A4 (x y ) = 2A4 (x) + 2A4 (y ) + 12A2,2 (x, y ), and A2 (x + y ) + A2 (x y ) = 2A2 (x) + 2A2 (y ) for all x, y R, we get (3.4) 24A2 (y ) + 24A0 = 0 for all y R. Hence A2 (y ) 0 and A0 = 0. Thus from (3.3) we have f (x) = A4 (x) and the proof of the theorem is now complete. 4. Solution of equation (1.1) on commutative groups In this section, we solve the functional equation (1.1) on commutative groups with some additional requirements. A group G is said to be divisible if for every element b G and every n N, there exists an element a G such that na = b. If this element a is unique, then G is said to be uniquely divisible. In a uniquely divisible b group, this unique element a is denoted by n . The equation na = b has a solution is equivalent to say that the multiplication by n is surjective. Similarly, the equation na = b has a unique solution is equivalent to say that the multiplication by n is bijective. Thus the notions of ndivisibility and n-unique divisibility refer, respectively, to surjectivity and bijectivity of the multiplication by n. The proof of Theorem 3.1 can be generalized to abstract structures by using the more general result of Hosszu [3] instead of Lemma 2.3. Since the proof of the following theorem is identical to the proof of Theorem 3.1, we omit its proof. Theorem 4.1. Let G and S be uniquely divisible abelian groups. The function f : G S satises the quartic functional equation (1.1) for all x, y G, if and only if f is of the form f (x) = A4 (x), where A4 (x) is the diagonal of a 4-additive symmetric function A4 : G4 S.

On the general solution of a quartic functional equation

575

Theorem 4.1 can be further strengthened using two important results due to Sz ekelyhidi [5]. The results needed for this improvements are the followings (see [5], pp. 7072): Theorem 4.2. Let G be a commutative semigroup with identity, S a commutative group and n a nonnegative integer. Let the multiplication by n! be bijective in S. The function f : G S is a solution of Fr echet functional equation (4.1) x1 ,...,xn+1 f (x0 ) = 0 x0 , x1 , . . . , xn+1 G if and only if f is a polynomial of degree at most n. Theorem 4.3. Let G and S be commutative groups, n a nonnegative integer, i , i additive functions from G into G and i (G) i (G) (i = 1, 2, . . . , n + 1). If the functions f, fi : G S (i = 1, 2, . . . , n + 1) satisfy
n+1

(4.2)

f (x) +
i=1

fi (i (x) + i (y )) = 0

then f satises Fr echet functional equation (4.1). Using these two theorems, Theorem 4.1 can be further improved. Theorem 4.4. Let G and S be commutative groups. Let the multiplication by 2 be surjective in G and let the multiplication by 24 be bijective in S. The function f : G S satises the quartic functional equation (1.1) for all x, y G, if and only if f is of the form (4.3) f (x) = A4 (x), where A4 (x) is the diagonal of a 4-additive symmetric function A4 : G4 S. Proof. Using the unique divisibility of S by 24 we can rewrite the functional equation (1.1) in the form
5

(4.4)

f (x) +
i=1

fi (i (x) + i (y )) = 0

where f1 = f2 = 1 6 f, f3 3 (x) = 4 (x) = x, 5 (x)

= f4 = 2 3 f, f5 = 4f , 1 (x) = 2 (x) = = 0, and 1 (y ) = 2y, 2 (y ) = 2y, 3 (y ) = 5 (y ) = y, 4 (y ) = y . From these i and i we see that i (G) i (G) for i = 1, 2, 3, 4, 5. Hence by Theorem 4.3, f satises the Fr echet functional equation (4.1). By Theorem 4.2, f is a polynomial of degree at most 4, that is f is of the form f (x) = A0 (x) + A1 (x) + A2 (x) + A3 (x) + A4 (x),

576

Jukang K. Chung and Prasanna K. Sahoo

where A0 (x) = A0 is an arbitrary constant, A1 Hom(G, S), and An (x) is the diagonal of a n-additive symmetric function An : Gn S, n {2, 3, 4}. Interchanging y with y in (1.1), one obtains 24f (y ) = 24f (y ) and hence f is an even function. The same argument as used in the last ten lines of the proof of Theorem 3.1 shows that any function of the form (4.3) actually satises (1.1). References
[1] J. Acz el, J. K. Chung and C. T. Ng, Symmetric second dierences in product form on groups. Topics in mathematical analysis, World Sci. Publishing Co., 1989, 122. [2] J. Acz el and J. Dhombres, Functional equations in several variables, Cambridge University Press, Cambridge, 1989. [3] M. Hosszu, On the Fr echets functional equation, Bul. Isnt. Politech. Iasi 10 (1964), no. 1-2, 2728. [4] M. Kuczma, An introduction to the theory of functional equations and inequali ties, Pa nstwowe Wydawnictwo Naukowe-Uniwersytet Slaski, Warszawa-Krak owKatowice, 1985. [5] L. Sz ekelyhidi, Convolution type functional equation on topological abelian groups, World Scientic, Singapore, 1991. [6] J. M. Rassias, Solution of the Ulam stability problem for quartic mappings, Glasnik Matemati cki 34 (54) (1999), no. 2, 243252. Jukang K. Chung, Department of Applied Mathematics, South China University of Technology, Guangzhou, P. R. China Prasanna K. Sahoo, Department of Mathematics, University of Louisville, Louisville, Kentucky 40292, USA E-mail : sahoo@louisville.edu