Electronic Journal of Dierential Equations, Vol. 2008(2008), No. 03, pp. 16. ISSN: 1072-6691. URL: http://ejde.math.txstate.edu or http://ejde.math.unt.

edu ftp ejde.math.txstate.edu (login: ftp)

SUFFICIENT CONDITIONS FOR THE OSCILLATION OF SOLUTIONS TO NONLINEAR SECOND-ORDER DIFFERENTIAL EQUATIONS
AHMED BERKANE

Abstract. We present sucient conditions for all solutions to a second-order ordinary dierential equations to be oscillatory.

1. Introduction Kirane and Rogovchenko [4] studied the oscillatory solutions of the equation r(t)(x(t))x (t) + p(t)x (t) + q(t)f (x(t)) = g(t),
1

t t0 ,

(1.1)

where t0 0, r(t) C ([t0 , ); (0, )), p(t) C([t0 , ); R), q(t) C([t0 , ); (0, )), q(t) is not identical zero on [t , ) for some t t0 , f (x), (x) C(R, R) and (x) > 0 for x = 0. Their results read as follows Theorem 1.1. Case g(t) 0: Assume that for some constants K, C, C1 and for all x = 0, f (x)/x K > 0 and 0 < C (x) C1 . Let h, H C(D, R), where D = {(t, s) : t s t0 }, be such that (i) H(t, t) = 0 for t t0 , H(t, s) > 0 in D0 = {(t, s) : t s t0 } (ii) H has a continuous and non-positive partial derivative in D0 with respect to the second variable, and H = h(t, s) H(t, s) s for all (t, s) D0 . If there exists a function C 1 ([t0 , ); (0, )) such that lim sup
t+

1 H(t, t0 )

[H(t, s)(s)
t0

C1 (s)r(s)Q2 (t, s)]ds = , 4

where 1 1 p2 (t) , C C1 4r(t) p(s) (s) Q(t, s) = h(t, s) + H(t, s), C1 r(s) (s) (t) = (t) Kq(t)
2000 Mathematics Subject Classication. 34C15. Key words and phrases. Nonlinear dierential equations; oscillation. c 2008 Texas State University - San Marcos. Submitted September 7, 2007. Published January 2, 2008.
1

A. BERKANE

EJDE-2008/03

then (1.1) is oscillatory. Theorem 1.2. Case g(t) = 0: Let the assumptions of theorem 1 be satised and suppose that the function g(t) C([t0 , ); R) satises

(s)|g(s)|ds = N < . Then any proper solution x(t) of (1.1); i.e, a non-constant solution which exists for all t t0 and satises suptt0 |x(t)| > 0, satises lim inf |x(t)| = 0.
t

Note that localization of the zeros is not given in the work by Kirane and Rogovchenko [4]. Here we intend to give conditions that allow us to localize the zeros of solutions to (1.1). Observe that in contrast to [4] where a Ricatti type transform, v(t) = r(t)(x(t))x (t) , x(t)

is used, here we simply use a usual Ricatti transform. 2. Main Results Dierential equation without a forcing term. Consider the second-order differential equation r(t)(x(t))x (t) + p(t)x (t) + q(t)f (x(t)) = 0, t t0 (2.1)

where t0 0, r(t) C 1 ([t0 , ); (0, )), p(t) C([t0 , )); R), q(t) C([t0 , )); R), p(t) and q(t) are not identical zero on [t , ) for some t t0 , f (x), (x) C(R, R) and (x) > 0 for x = 0. The next theorem follows the ideas in Nasr [6]. Assume that there exists an interval [a, b], where a, b t , such that e(t) 0. Theorem 2.1. Assume that for some constants K, C, C1 and for all x = 0, f (x) K 0, x 0 < C (x) C1 . (2.2) (2.3)

Suppose further there exists a continuous function u(t) such that u(a) = u(b) = 0, u(t) is dierentiable on the open set (a, b), a, b t , and
b

Kq(t)
a

p2 (t) u2 (t) 2C1 r(t)(u )2 (t) dt 0 . 2Cr(t)

(2.4)

Then every solution of (2.1) has a zero in [a, b]. Proof. Let x(t) be a solution of (2.1) that has zero on [a, b]. We may assume that x(t) > 0 for all t [a, b] since the case when x(t) < 0 can be treated analogously. Let x (t) v(t) = , t [a, b]. (2.5) x(t)

EJDE-2008/03

OSCILLATION OF SOLUTIONS

Multiplying this equality by r(t)(x(t)) and dierentiate the result. Using (2.1) we obtain (r(t)(x(t))v(t)) = (r(t)(x(t))x (t)) + r(t)(x(t))v 2 (t) x(t) f (x(t)) + r(t)(x(t))v 2 (t) = p(t)v(t) + q(t) x(t) (r(t)(x(t)) 2 r(t)(x(t)) 2 p(t) = v (t) + v (t) 2 2 2 r(t)(x(t))v(t) f (x(t)) + q(t) x(t) (r(t)(x(t)) 2 r(t)(x(t)) p(t) 2 = v (t) + v(t) 2 2 (r(t)(x(t)) f (x(t)) p2 (t) + q(t) . 2r(t)(x(t)) x(t)

Using (2.2)-(2.3) and the fact that p(t) (r(t)(x(t)) v(t) 2 r(t)(x(t)) we have (r(t)(x(t))v(t)) (r(t)(x(t)) 2 p2 (t) v (t) + Kq(t) 2 2Cr(t) (2.6)
2

0,

Multiplying both sides of this inequality by u2 (t) and integrating on [a, b]. Using integration by parts on the left side, the condition u(a) = u(b) = 0 and (2.3), we obtain
b

0
a

r(t)(x(t)) 2 v (t)u2 (t)dt + 2 2

r(t)(x(t))v(t)u(t)u (t)dt
a

+
a b

Kq(t)u2 (t)dt
a

p2 (t) 2 u (t)dt 2Cr(t)

r(t)(x(t)) 2 (v (t)u2 (t) + 4v(t)u(t)u (t))dt 2

+
a b

Kq(t)u2 (t)dt
a

p2 (t)u2 (t) dt 2Cr(t)

r(t)(x(t)) [v(t)u(t) + 2u (t)]2 dt 2 2

r(t)(x(t))u 2 (t)dt
a

+
a b

Kq(t)u2 (t)dt
a 2

p2 (t) 2 u (t)dt 2Cr(t)

a b

Kq(t) +
a

p (t) u2 (t) 2r(t)(x(t))u 2 (t) dt 2Cr(t)

r(t)(x(t)) [v(t)u(t) + 2u (t)]2 dt . 2

A. BERKANE

EJDE-2008/03

Now, from (2.3) we have

0
a b

Kq(t) +
a

p2 (t) u2 (t) 2r(t)C1 u 2 (t) dt 2Cr(t)

r(t)(x(t)) [v(t)u(t) + 2u (t)]2 dt. 2

If the rst integral on the right-hand side of the inequality is greater than zero, then we have directly a contradiction. If the rst integral is zero and the second is also zero then x(t) has the same zeros as u(t) at the points a and b; (x(t) = ku2 (t)), which is again a contradiction with our assumption. Corollary 2.2. Assume that there exist a sequence of disjoint intervals [an , bn ], and a sequence of functions un (t) dened and continuous an [an , bn ], dierentiable on (an , bn ) with un (an ) = un (bn ) = 0, and satisfying assumption (2.4). Let the conditions of Theorem 2.1. hold. Then (2.1) is oscillatory. Dierential equation with a forcing term. Consider the dierential equation r(t)(x(t))x (t) + p(t)x (t) + q(t)f (x(t)) = g(t), t t0 (2.7)

where t0 0, g(t) C([t0 , ); R) r(t) C 1 ([t0 , ); (0, )), p(t) C([t0 , )); R), q(t) C([t0 , )); R), p(t) and q(t) are not identical zero on [t , [ for some t t0 , f (x), (x) C(R, R) and (x) > 0 for x = 0. Assume that there exists an interval [a, b], where a, b t , such that g(t) 0 and there exists c (a, b) such that g(t) has dierent signs on [a, c] and [c, b]. Without loss of generality, let g(t) 0 on [a, c] and g(t) 0 on [c, b]. Theorem 2.3. Let (2.3) hold and assume that f (x) K, x|x| (2.8)

for a positive constant K and for all x = 0. Furthermore assume that there exists a continuous function u(t) such that u(a) = u(b) = u(c) = 0, u(t) dierentiable on the open set (a, c) (c, b), and satises the inequalities
c

Kq(t)g|(t)|
a b

p2 (t) u2 2C1 r(t)(u )2 (t) d(t) 0, 2Cr(t)

(2.9) (2.10)

Kq(t)g|(t)|
c

p2 (t) u2 2C1 r(t)(u )2 (t) d(t) 0 . 2Cr(t)

Then every solution of equation (2.7) has a zero in [a, b]. Proof. Assume to the contrary that x(t), a solution of (2.7), has no zero in [a, b]. Let x(t) < 0 for example. Using the same computations as in the rst part, we

EJDE-2008/03

OSCILLATION OF SOLUTIONS

obtain: (r(t)(x(t))v(t)) = (r(t)(x(t))x (t)) g(t) + r(t)(x(t))v 2 (t) x(t) x(t) g(t) f (x(t)) = p(t)v(t) + q(t) + r(t)(x(t))v 2 (t) x(t) x(t) (r(t)(x(t)) 2 r(t)(x(t)) 2 p(t)v(t) = v (t) + v (t) 2 2 2 r(t)(x(t)) f (x(t)) g(t) + q(t) x(t) x(t) p(t) (r(t)(x(t)) 2 r(t)(x(t)) v(t) = v (t) + 2 2 r(t)(x(t))v(t) p2 (t) f (x(t)) g(t) + q(t) 2r(t)(x(t)) x(t) x(t)

For t [c, b] we have (r(t)(x(t))v(t)) = r(t)(x(t)) 2 r(t)(x(t)) p(t) v (t) + v(t) 2 2 r(t)(x(t)) p2 (t) f (x(t)) |g(t)| + q(t) |x(t)| + 2r(t)(x(t)) x(t)|x(t)| |x(t)|
2

From (2.8), and using the fact that r(t)(x(t)) p(t) v(t) 2 r(t)(x(t)) we deduce (r(t)(x(t))v(t)) p2 (t) |g(t)| (r(t)(x(t)) 2 v (t) +Kq(t)|x(t)|+ . (2.11) 2 2r(t)(x(t)) |x(t)|
2

Using the Hlder inequality in (2.11) we obtain o (r(t)(x(t))v(t)) (r(t)(x(t)) 2 v (t) + 2 Kq(t)|g(t)| p2 (t) . 2r(t)(x(t)) (2.12)

Multiplying both sides of this inequality by u2 (t) and integrating on [c, b], we obtain after using integration by parts on the left-hand side and the condition u(c) = u(b) = 0,
b

0
c

r(t)(x(t)) 2 v (t)u2 (t)dt + 2

Kq(t)|g(t)|u2 (t)dt
c

p2 (t)u2 (t) dt + 2 2r(t)(x(t))

r(t)(x(t))v(t)u(t)u (t)dt
c b

r(t)(x(t)) [v(t)u(t) 2u (t)]2 dt 2 2

r(t)(x(t))u 2 (t)dt
c

+
c

Kq(t)|g(t)|u2 (t)dt
c

p2 (t)u2 (t) dt. 2r(t)(x(t))

A. BERKANE

EJDE-2008/03

Assumption (2.3) allows us to write

0
c

r(t)(x(t)) [v(t)u(t) + 2u (t)]2 dt 2 2

C1 r(t)(u )2 (t)dt
c

+
c b

Kq(t)|g(t)|u2 (t)dt
c

p2 (t)u2 (t) dt 2Cr(t)

r(t)(x(t)) [v(t)u(t) + 2u (t)]2 dt 2

p2 (t) u2 (t) 2C1 r(t)(u )2 (t) dt. 2Cr(t) c This leads to a contradiction as in Theorem 2.1; the proof is complete. + Kq(t)g|(t)| Corollary 2.4. Assume that there exist a sequence of disjoint intervals [an , bn ] a sequences of points cn (an , cn ), and a sequence of functions un (t) dened and continuous on [an , bn ], dierentiable on (an , cn ) (cn , bn )with un (an ) = un (bn ) = un (cn ) = 0, and satisfying assumptions (2.9)-(2.10). Let the conditions of Theorem 2.3 hold. Then (2.7) is oscillatory. References
[1] M. A. El-Sayed, An oscillation criterion for forced second order linear dierential equation, Proc. Amer. Math. Soc. 33 (1972), 613-817. [2] A. G. Kartsatos, On the maintenance of oscillation of n-th order equation under the eect of a small forcing terms, J. Di. Equations 10 (1971), 355-363. [3] A. G. Kartsatos, On the maintenance of oscillation of a periodic forcing term, Proc. Amer. Math. Soc. 33 (1972), 377-382. [4] M. Kirane and Y. Rogovchenko, Oscillations results for the second order damped dierential equation with non-monotonous nonlinearity, J. Math. Anal. Appl. 250, 118-138 (2000). [5] V. Komkov, On boundedness and oscillation of the dierential equation x A(t)g(x) = f (t) in Rn , SIAM J. Appl. Math. 22 (1972), 561-568. [6] A. H. Nasr, Sucient conditions for the oscillation of forced super-linear second order dierential equations with oscillatory potential, Proc. Amer. Math. Soc. 126, 1, (1998) 123-125. [7] S. M. Rankin, Oscillation theorems for second order nonhomogeneous linear dierential equations, J. Math. Anal. 53 (1976), 550-556. [8] J. S. W. Wong, Second order nonlinear forced oscillations, SIAM J. Math. Anal. 19, 3 (1998). Ahmed Berkane Faculte des Sciences, Departement de Mathematiques, Universite Badji Mokhtar, 23000 Annaba, Algeria E-mail address: aberkane20052005@yahoo.fr