Remarks on oscillation and nonoscillation for second-order linear difference equations

Remarks on oscillation and nonoscillation for second-order linear difference equations

Applied Mathematics Letters 21 (2008) 578–580 www.elsevier.com/locate/aml Remarks on oscillation and nonoscillation for second-order linear differenc...

156KB Sizes 0 Downloads 15 Views

Applied Mathematics Letters 21 (2008) 578–580 www.elsevier.com/locate/aml

Remarks on oscillation and nonoscillation for second-order linear difference equationsI Yong Zhou a,∗ , B.G. Zhang b , C.F. Li a a Department of Mathematics, Xiangtan University, Hunan 411105, PR China b Department of Applied Mathematics, Ocean University of China, Qingdao 266003, PR China

Received 10 August 2005; received in revised form 9 February 2006; accepted 14 February 2006

Abstract In this remark, we shall show the main results of the earlier work [W.T. Li, S.S. Cheng, Remarks on two recent oscillation theorems for second-order linear difference equations, Appl. Math. Lett. 16 (2003) 161–163] are incorrect. c 2007 Elsevier Ltd. All rights reserved.

Keywords: Oscillation; Nonoscillation; Difference equations

In this work, we consider the second-order linear difference equation ∆2 xn−1 + pn xn = 0,

n ≥ n0

(1)

where the forward difference ∆ is defined as usual, i.e., ∆xn = xn+1 − xn , ∆2 xn = ∆(∆xn ), and { pn }∞ n=1 is a real sequence with pn ≥ 0 for n ≥ n 0 , n 0 ∈ N, N is the set of natural numbers. A solution {xn } of (1) is said to be oscillatory if the terms xn of the sequence {xn } are neither eventually all positive nor eventually all negative. Otherwise, the solution is called nonoscillatory. Oscillation and nonoscillation of Eq. (1) have been investigated intensively; see [1–5]. Some typical results about the oscillatory and nonoscillatory properties of Eq. (1) are the following. Theorem A ([2]). If lim sup n n→∞

∞ X i=n+1

pi <

1 , 4

(2)

then Eq. (1) is nonoscillatory.

I This research was supported by the National Natural Science Foundation of PR China (10371103) and the Research Fund of Hunan Provincial Education Department (04A055). ∗ Corresponding author. E-mail addresses: [email protected] (Y. Zhou), [email protected] (B.G. Zhang), [email protected] (C.F. Li).

c 2007 Elsevier Ltd. All rights reserved. 0893-9659/$ - see front matter doi:10.1016/j.aml.2006.02.035

Y. Zhou et al. / Applied Mathematics Letters 21 (2008) 578–580

579

Theorem B ([4]). If ∞ X

lim inf n n→∞

pi >

i=n+1

1 , 4

(3)

then Eq. (1) is oscillatory. In [5], Zhang and Zhou proved oscillation and nonoscillation criteria for Eq. (1). They proved the following two theorems. Theorem C ([5]). If there exists n 0 ≥ n 0 such that for every m ∈ N, 2m+1 n 0 −1 X

pi ≤

i=2m n 0

α0 , 2m+1 n 0

(4)

√ then Eq. (1) is nonoscillatory, where α0 = 3 − 2 2. Theorem D ([5]). If there exists n 0 ≥ n 0 and α > α0 such that for every m ∈ N, 2m+1 n 0 −1 X

pi ≥

i=2m n 0

α 2m n

0

,

(5)

then Eq. (1) is oscillatory. Recently, Li and Cheng [3] purported to prove that Theorems C and D are special cases of Theorems A and B. In the following, we will show that their proofs are incorrect. So, they have not proved the above proposition. In fact, by the argument of [3, pages 162,163], we can obtain from (4) and (5) lim sup N0 N0 →∞

∞ X

pi <

1 , 4

(6)

pi >

1 , 4

(7)

i=N0 +1

and lim inf N0 N0 →∞

∞ X i=N0 +1

respectively, where N0 = [n 0 /2n 0 ], where [y] denotes the greatest integral part of the number y. From the proof in [3], we note that N0 should be [n 0 /2m 0 ], m 0 ∈ N. Their argument that (6) and (7) imply (2) and (3), respectively, is incorrect. It is easy to see that the set {N0 } is a subset of the set {n 0 , n 0 + 1, n 0 + 2, . . .}. Hence lim sup n n→∞

∞ X i=n+1

pi ≥ lim sup N0 N0 →∞

∞ X

pi ,

i=N0 +1

and lim inf n n→∞

∞ X i=n+1

pi ≤ lim inf N0 N0 →∞

∞ X

pi .

i=N0 +1

Therefore, (6) and (7) do not imply (2) and (3). There are several mistakes in [3] as follows: (i) lim sup in [3, page 162, line 5 and page 163, line 4] should be lim inf. (ii) Their proof assumed that conditions (4) and (5) hold for all n 0 , i.e., n 0 can be arbitrary, whereas Theorems C and D hypothesized that only one such n 0 need exist. In addition, we can obtain the following oscillation and nonoscillation criteria which generalize Theorems C and D by replacing 2 with λ (λ > 1).

580

Y. Zhou et al. / Applied Mathematics Letters 21 (2008) 578–580

Theorem E. Let λ > 1, λ ∈ N. If for some n 0 ∈ N and every positive integer m, λm+1 n 0 −1 X

α , (λ − 1)λm+1 n 0 i=λm n 0 √ where α ≤ k0 (λ) = ( λ − 1)2 , then Eq. (1) is nonoscillatory. pi ≤

(8)

Theorem F. Let λ > 1, λ ∈ N. If for some n 0 ∈ N and every positive integer m, λm+1 n 0 −1 X

α , mn (λ − 1)λ 0 i=λm n 0 √ where α > k0 (λ) = ( λ − 1)2 , then Eq. (1) is oscillatory. pi ≥

(9)

By using the method of [5], we prove easily Theorems E and F. We omit the details. √ Clearly Theorems C and D are special cases of Theorems E and F with λ = 2. In particular k0 (2) = α0 = 3 − 2 2. Acknowledgments The authors thank the referee and editor for useful comments and suggestions. References [1] R.P. Agarwal, Difference Equations and Inequalities, second edn, Marcel Dekker, New York, 2000. [2] S.S. Cheng, T.C. Yan, H.J. Li, Oscillation criteria for second order difference equation, Funkcial. Ekvac. 34 (1991) 223–239. [3] W.T. Li, S.S. Cheng, Remarks on two recent oscillation theorems for second-order linear difference equations, Appl. Math. Lett. 16 (2003) 161–163. [4] B.G. Zhang, S.S. Cheng, Oscillation criteria and comparison theorems for delay difference equations, Fasc. Math. 25 (1995) 13–32. [5] B.G. Zhang, Yong Zhou, Oscillation and nonoscillation of second order linear difference equation, Comput. Math. Appl. 39 (1–2) (2000) 1–7.