• 検索結果がありません。

Oscillation For Higher Order Sublinear Delay Difference Equation ∗

N/A
N/A
Protected

Academic year: 2022

シェア "Oscillation For Higher Order Sublinear Delay Difference Equation ∗ "

Copied!
9
0
0

読み込み中.... (全文を見る)

全文

(1)

Oscillation For Higher Order Sublinear Delay Difference Equation

Mustafa Kemal Yildiz

and ¨ Ozkan ¨ Ocalan

Received 22 November 2007

Abstract

In this paper, we shall consider higher order nonlinear delay difference equation of the type

mxn+pnm1xn+qnxαn−k= 0, n≥0

wherem >2,{pn}and{qn}are sequences of nonnegative real numbers, 0≤pn<

1 forn≥n0 ≥0,kis a positive integer andα∈(0,1) is a ratio of odd positive integers. We obtain sufficient conditions for the oscillation of all solutions of this equation.

1 Introduction

We consider the following higher order nonlinear delay difference equation:

mxn+pnm1xn+qnxαn−k= 0, m >2, (1) where ∆ is the usual forward difference operator defined by ∆xn=xn+1−xn, kis a positive integer,{pn}and{qn}are sequences of nonnegative real numbers, 0≤pn <1 for n≥n0 ≥0 and α∈(0,1) is a ratio of odd positive integers. If 0 < α <1, then equation (1) is called sublinear equation, whenα >1, it is called superlinear equation.

Recently, there are many studies concerning the behavior of the oscillatory difference equations, see [1]–[11] and the reference cited therein. In particular, in [8], Tang and Liu have investigated the oscillatory behaviour of the first order sublinear and superlinear delay difference equations of the form

xn+1−xn+pnxαn−k= 0, (2)

where{pn}is a sequence of nonnegative numbers,kis a positive integer andα∈(0,∞) is a quotient of odd positive integers. Tang and Liu have discussed the oscillation for

Mathematics Subject Classifications: 39A10

Afyon Kocatepe University, Faculty of Science and Arts, Department of Mathematics, ANS Cam- pus, 03200, Afyonkarahisar, Turkey

63

(2)

equation (2) in the two cases where α∈(0,1) andα∈(1,∞). For the former case, it is shown that every solution of sublinear equation (2) oscillates if and only if

X

n=0

pn=∞.

For the latter case, it is shown that, if there exists aλ > k1lnαsuch that lim inf

n→∞

pnexp(−eλn)

>0,

then every solution of superlinear equation (2) oscillates. When α= 1, (2) reduces to linear delay difference equation and the oscillatory behavior of this equation has been investigated in the literature, which we refer to [1, 5, 6, 7].

In [3], Agarwal and Grace investigated that oscillatory properties of equation (1) in the caseα= 1 andmis even.

By a solution of (1), we mean a real sequence {xn} which is the defined for all n≥ −k and satisfies equation (1) forn≥0.A solution of (1) is said to be oscillatory of it is neither eventually positive nor eventually negative.

In this paper our aim is to obtain sufficient conditions for all solutions of equation (1) whenm is even or odd.

The following lemmas are needed in the proof of our results, which Lemma 1 and Lemma 2 are obtained in [1], Lemma 3 is obtained in [11].

LEMMA 1 [Discrete Kneser’s Theorem]. Letzn be defined forn≥n0, andzn >0 with ∆mzn of constant sign forn≥n0 and not identically zero. Then, there exists an integer j, 0≤j≤mwith (m+j) odd for ∆mzn ≤0, and (m+j) even for ∆mzn ≥0 such that

(i) j≤m−1 implies (−1)j+iizn>0, for alln≥n0,j≤i≤m−1, (ii) j≥1 implies ∆izn >0, for all largen≥n0, 1≤i≤j−1.

LEMMA 2. Letzn be defined forn≥n0, andzn >0 with ∆mzn ≤0 forn≥n0

and not identically zero. Then, there exists a large integer n1≥n0 such that zn≥ 1

(m−1)!(n−n1)m1m1z2m−j−1n n≥n1, where jis defined in Lemma 1. Further, ifzn is increasing, then

zn ≥ 1 (m−1)!

n 2m−1

m1

m−1zn, n≥2m−1n1. LEMMA 3. Assume that for largen,

(pn, pn+1, . . . , pn+k1)6= 0.

Then

xn+1−xn+pnxαn−k= 0, n= 0,1,2, . . .

has an eventually positive solution if and only if the corresponding inequality xn+1−xn+pnxαn−k≤0, n= 0,1,2, . . .

has an eventually positive solution.

(3)

2 Oscillation of Equation (1)

THEOREM 1. Let ∆pn ≤0, for n≥n0 ≥0,α∈(0,1), lim infn→∞qn >0,m is an even positive integer,k >1 and

Cn= min{Qn, Rn}> pn−k, for all large n, where

Qn= θ

(m−2)!

α

n1

X

j=nk

(j−k)α(m2)qj

, θ= 1 2(m2)2 and

Rn=

n1

X

j=n−k

qj

n1

X

s=j−k

1

(m−3)!(s+m+k−j−3)(m3)

α

. If the difference equation

∆zn+cnznαk= 0, (3)

where

cn = min

Cn−pnk,

1 (m−1)!

1 2(m1)2

α

(n−k)α(m1)qn

(4) is oscillatory, then (1) is oscillatory.

PROOF. Let{xn}be a nonoscillatory solution of (1), sayxn >0 forn≥n0 ≥1.

First, we claim that

m−1xn is eventually of one sign. To this end, we assume that ∆m1xn is oscillatory. There existsN≥n0+ksuch that ∆m1xN <0. Letn=N in (1) and the multiply the resulting equation by ∆m1xN, to obtain

mxNm1xN =−pN(∆m1xN)2−qNxαNkm1xN ≥ −pN(∆m1xN)2 or

m1xN+1m1xN ≥(1−pN)(∆m1xN)2>0, which implies that

m1xN+1<0.

By induction, we obtain ∆m−1xn <0 for n ≥N, contradicting the assumption that ∆m−1xn is oscillatory.

Next, suppose there existsN ≥n0+k such that ∆m−1xN = 0. Let n=N in (1) leads to

mxN =−qNxαN−k≤0, which implies that

m−1xN+1≤∆m−1xN = 0.

As in the above case, we have seen that this contradicts the assumption that

m1xn

is oscillatory.

Now, we consider the following two cases either ∆m1xn < 0 or ∆m1xn > 0 eventually.

(4)

CASE A. Suppose that ∆m1xn <0 for n≥n1 ≥n0. By Lemma 1, there exists ann2≥n1such that forn≥n2

m2xn>0, and either (i) ∆xn>0 or (ii) ∆xn<0.

(i) Suppose that ∆xn > 0 forn ≥n2. By applying Lemma 2, there exists n3 ≥ 2m−1n2such that

xn≥ 1 (m−2)!

n 2m2

m−2

m2xn, forn≥n3. There exists n4≥n3such that

xαnk≥ 1

(m−2)!

1 2(m2)2

α

(n−k)α(m2)m2xn−kα

, forn≥n4. (5) Using (5) in (1) yields

2zn+pn∆zn+ θ

(m−2)!

α

(n−k)α(m−2)qnzαnk≤0, forn≥n4, (6) where zn = ∆m2xn,n≥n4 andθ= 1/2(m2)2 . Summing up both sides of (6) from n−kton−1, we have

∆zn−∆znk+

n1

X

j=n−k

pj∆zj+ θ

(m−2)!

α n1 X

j=n−k

(j−k)α(m−2)qjzαjk≤0 (7) or

0≥∆zn+

pnzn−pnkznk

n1

X

j=n−k+1

zj∆pj1

+ θ

(m−2)!

α

n−1

X

j=n−k

(j−k)α(m2)qj

znαk.

Since ∆zn < 0, zn > 0, limn→∞zn = c ∈ [0,∞) exists. We claim that c = 0.

Otherwise, limn→∞zn =c ∈(0,∞). So, if we take limit on both sides of (7) we get the contradiction ∞ ≤0. Thus, we must havec = 0 andzn < znα<1 for sufficiently largen. And we have

0≥∆zn+

pnzn−pn−kzn−kα

n1

X

j=n−k+1

zj∆pj−1

+ θ

(m−2)!

α

n−1

X

j=nk

(j−k)α(m2)qj

znαk.

(5)

Since ∆pn ≤0, we have

∆zn+

 θ

(m−2)!

α

n−1

X

j=nk

(j−k)(m2)αqj

−pn−k

znαk ≤0, (8) for somen5≥n4 and hence by (4), we have

∆zn+cnznαk ≤0, forn≥n5.

Therefore, by Lemma 3, (3) has an eventually positive solution, which is a contradiction.

(ii) Suppose that ∆xn<0 forn≥n2. By applying Lemma 3, we must havej = 0 and one can easily see that

(−1)iixn>0, fori= 0,1, . . . , m−1 andn≥n3≥n2. (9) By discrete Taylor’s formula [1, p. 26],xg can be expressed as

xg=

m−3

X

i=0

(z+i−1−g)(i)

i! (−1)iixz+ 1 (m−3)!

z−1

X

s=g

(s+m−g−3)(m3)m2xs, (10) for all g ∈N(n3, z) = {n3, n3+ 1, , z}, where z∈N(n3) ={n3, n3+ 1, . . .}. Using (9) in (10) and lettingz−1 =n−k, we have

xg≥ 1 (m−3)!

nk

X

s=g

(s+m−g−3)(m−3)m−2xs, n≥n3. Letting g=j−k, and using the fact that ∆m2xnis decreasing, we have

xαjk≥ 1

(m−3)!

α

n−k

X

s=jk

(s+m−j+k−3)(m3)

α

m2xn−k

α

. (11) Summing both sides of (1) fromn−kton−1, one can easily see that

m−1xn−pn−km−2xn−k+

n1

X

j=n−k

qjxαjk≤0, (12) using (11) in (12), we have

∆zn−pnkznαk+

n1

X

j=n−k

qj

1 (m−3)!

α

n1

X

s=j−k

(s+m−j+k−3)(m−3)

α

znαk≤0, and

∆zn+

n1

X

j=nk

qj

1 (m−3)!

α

n1

X

s=jk

(s+m−j+k−3)(m3)

α

−pnk

zn−kα ≤0,

(6)

where zn = ∆m2xn,n≥n4≥n3. Next, by (4), we see that

∆zn+cnznαk ≤0, forn≥n4.

Therefore, by Lemma 3, (3) has an eventually positive solution, which is a contradiction.

CASE B. Suppose that ∆m−1xn>0 forn≥n1. From (1) it follows that

mxn+qnxαn−k≤0, forn≥n1. (13) By applying Lemma 2, there existsn2≥2m1n1 such that

xn≥ 1 (m−1)!

n 2m−1

m1

m1xn, forn≥n2. There exists n3≥n2such that

xnk≥ 1 (m−1)!

1 2m1

m1

(n−k)m1m1xnk forn≥n3. There exists n4≥n3such that

xαnk≥ 1

(m−1)!

1 2(m1)2

α

(n−k)α(m−1)m−1xnk

α

, forn≥n4. (14) Using (14) in (13), we obtain

∆zn+ 1

(m−1)!

1 2(m−1)2

α

(n−k)(m−1)αqnznαk≤0, forn≥n4, where zn = ∆m−1xn, n≥n4. By (4), we see that

∆zn+cnznαk ≤0, forn≥n4. The rest of the proof is similar to that of the above case.

THEOREM 2. Let ∆pn ≤0, for n≥n0 ≥0, α∈(0,1), lim infn→∞qn >0,m is an odd positive integer, k >1 andQn> pn−k, for all largen, where

Qn= θ

(m−2)!

α

n1

X

j=n−k

(j−k)α(m−2)qj

, θ= 1 2(m−2)2. If the difference equation

∆zn+cnznαk= 0, (15)

where cn= min

(

Qn−pn−k,

1 (m−1)!

1 2(m−1)2

α

(n−k)α(m1)qn,

(m−2)(m−2) (m−2)!

α qn

)

(16)

(7)

is oscillatory, then (1) is oscillatory.

PROOF. Let{xn}be a nonoscillatory solution of (1), sayxn >0 forn≥n0 ≥1.

As in Theorem 1, we see that

m−1xn is eventually of one sign. To this end, we consider the two as in Theorem 1.

CASE A. Suppose that ∆m−1xn <0 for n≥n1 ≥n0. By Lemma 1, there exists ann2≥n1such that forn≥n2

m−2xn>0, and ∆xn>0.

Since ∆xn>0 forn≥n2. By applying Lemma 2, there existsn3≥2m1n2such that xn≥ 1

(m−2)!

n 2m2

m−2

m−2xn, forn≥n3. There exists n4≥n3such that

xαnk≥ 1

(m−2)!

1 2(m2)2

α

(n−k)α(m−2)m−2xnk

α

, forn≥n4. (17) Using (17) in (1) yields

2zn+pn∆zn+ 1

(m−2)!

1 2(m2)2

α

(n−k)α(m2)qnzn−kα ≤0, forn≥n4, (18) where zn = ∆m2xn,n≥n4 andθ= 1/2(m2)2. Summing up both sides of (18) from n−kton−1, we have

∆zn−∆zn−k+

n1

X

j=n−k

pj∆zj+ θ

(m−2)!

α n1 X

j=n−k

(j−k)α(m2)qjzjαk ≤0

and using the fact that ∆m2xn is decreasing, we have 0≥∆zn+

pnzn−pn−kznαk

n1

X

j=nk+1

zj∆pj−1

+ θ

(m−2)!

α

n−1

X

j=nk

(j−k)α(m2)qj

znαk.

Since ∆pn ≤0,we have

∆zn+

 θ

(m−2)!

α

n1

X

j=nk

(j−k)α(m2)qj

−pnk

zn−kα ≤0, (19) for somen5≥n4 and hence by (16), we have

∆zn+cnznαk ≤0, forn≥n5.

(8)

Therefore, by Lemma 3, (15) has an eventually positive solution, which is a contradic- tion.

CASE B. Suppose that ∆m1xn>0 forn≥n1. From (1) it follows that

mxn+qnxαn−k≤0, forn≥n1. (20) By Lemma 1, there exists an n2 ≥n1 such that ∆m−2xn <0 for n≥n2. Therefore either ∆xn>0 or ∆xn<0 forn≥n2. We consider these cases separately as follows:

(i) Suppose that ∆xn > 0 forn ≥n2. By applying Lemma 2, there exists n3 ≥ 2m−1n2such that

xn≥ 1 (m−1)!

n 2m−1

m1

m1xn, forn≥n3. There exists n4≥n3such that

xαnk≥ 1

(m−1)!

1 2(m1)2

α

(n−k)α(m1)m1xnk

α

, forn≥n4. (21) Using (21) in (1) yields

∆zn+ 1

(m−1)!

1 2(m1)2

α

(n−k)α(m−1)qnzn−kα ≤0, forn≥n4, (22) where zn = ∆m1xn, n≥n4. By (16), we have

∆zn+cnznαk ≤0, forn≥n5.

Therefore, by Lemma 3, (15) has an eventually positive solution, which is a contradic- tion.

(ii) Suppose that ∆xn<0 forn≥n2. By applying Lemma 1, we must havej = 0 and one can easily see that

(−1)iixn>0, fori= 0,1, . . . , m−1 andn≥n3≥n2. (23) By discrete Taylor’s formula [1, p. 26],xg can be expressed as

xg=

m2

X

i=0

(z+i−1−g)(i)

i! (−1)iixz+ 1 (m−2)!

z1

X

s=g

(s+m−g−2)(m2)m1xs, (24) for all g∈N(n3, z) ={n3, n3+ 1, . . . , z}, where z∈N(n3) ={n3, n3+ 1, . . .}. Using (23) in (24) and lettingz−1 =n−k, we have

xg≥ 1 (m−2)!

nk

X

s=g

(s+m−g−2)(m2)m1xs, n≥n3. Letting g=j−k, and using the fact that ∆m−1xn is decreasing, we have

xαj−k≥ 1

(m−2)!

α

nk

X

s=jk

(s+m+k−j−2)(m2)

α

m1xnk

α

. (25)

(9)

Using (25) in (20), we obtain

∆zn+qn

1 (m−2)!

α nk

X

s=nk

(s+m+k−n−2)(m−2)

!α

zαn−k≤0, forn≥n4

and

∆zn+qn

(m−2)(m2) (m−2)!

α

znαk ≤0, forn≥n4 (26) where zn = ∆m1xn, n≥n4≥n3. Next, by (16), we see that

∆zn+cnznαk≤0,forn≥n4.

Therefore, by Lemma 3, (15) has an eventually positive solution, which is a contradic- tion.

References

[1] R. P. Agarwal, Difference Equations and Inequalities, Marcel Dekker, New York, 2000.

[2] R. P. Agarwal, S. R. Grace and D. O’Regan, Oscillation Theory for Difference and Functional Differential Equations, Kluwer Academic Publishers, The Netherlands, 2000.

[3] R. P. Agarwal, S. R. Grace, Oscillation of higher-order difference equations, Appl.

Math. Lett., 13(2000), 81–88.

[4] R. P. Agarwal, S. R. Grace, Oscillation of certain difference equations, Mathemat- ical and Computer Modelling, 29(1999), 1–8.

[5] L. H. Erbe and B. G. Zhang, Oscillation of discrete analogues of delay equations, Differential and Integral Equations, 2(3)(1989), 300–309.

[6] I. Gy¨ori and G. Ladas, Oscillation Theory of Delay Differential Equations with Applications, Clarendon Press, Oxford, 1991.

[7] G. Ladas, Ch. G. Philos and Y.G. Sficas, Sharp conditions for the oscillation of delay difference equations, Journal of Applied Mathematics and Simulation, 2(1989), 101–111.

[8] X. H. Tang and Y. J. Liu, Oscillation for nonlinear delay difference equations, Tamkang J. Math., 32(2001), 275–280.

[9] E. Thandapani, R. Arul and P. S. Raja, Oscillation of first order netral delay difference equations, AMEN, 3(2003), 88–94.

[10] G. Zhang, Oscillation for nonlinear difference equations, AMEN, 2(2002), 22–24.

[11] G. Zhang and Y. Gao, Positive solution of higher nonlinear difference equation, Sys. Sci & Math. Sci., 19(1999), 157–167.

参照

関連したドキュメント

Saker, Oscillation criteria for second-order nonlinear neutral delay dynamic equations, Journal of Mathematical Analysis and Applications 300 (2004), no.. Peterson, Dynamic Equations

Saker, Oscillation criteria for second-order nonlinear neutral delay dynamic equations, Journal of Mathematical Analysis and Applications 300 (2004), no.. Peterson, Dynamic Equations

To the best of our knowledge, there is not yet any paper studying oscillation and asymptotic behaviour of higher-order dynamic equations, and therefore this paper is one of the

Mahfoud; Oscillation and asymptotic behavior of solutions of nth order nonlinear delay differential equations, J.. Sturm; M´ emoire sur les ´ equations diff´ erentielles lin´ eaires

Kusano, Oscillation properties of first order nonlinear functional differ- ential equations of neutral type, Differential Integral Equations 4 (1991), 425–436..

Oscillation criteria for two-dimensional difference and differential systems of first-order linear difference equations are generalized and extended to nonlinear dynamic equations

In this paper, we deal with the oscillation of bounded solutions of a class of second-order half-linear neutral delay dynamic equations with an oscillating coefficient on a time

Ladas, “Oscillation Theory of Delay Differential Equations”, Claren-.. don