AAA Abstract and Applied Analysis 1687-0409 1085-3375 Hindawi Publishing Corporation 10.1155/2014/902616 902616 Research Article Oscillations in Difference Equations with Deviating Arguments and Variable Coefficients Chatzarakis G. E. 1 Péics H. 2 Stavroulakis I. P. 3 Wong Patricia J. Y. 1 Department of Electrical and Electronic Engineering Educators School of Pedagogical and Technological Education (ASPETE) 14121 N. Heraklio Athens Greece aspete.gr 2 Faculty of Civil Engineering University of Novi Sad 24000 Subotica Serbia uns.ac.rs 3 Department of Mathematics University of Ioannina 451 10 Ioannina Greece uoi.gr 2014 1882014 2014 19 05 2014 21 07 2014 18 8 2014 2014 Copyright © 2014 G. E. Chatzarakis et al. 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.

New sufficient conditions for the oscillation of all solutions of difference equations with several deviating arguments and variable coefficients are presented. Examples illustrating the results are also given.

1. Introduction

In this paper we study the oscillation of all solutions of difference equation with several variable retarded arguments of the form ( E R ) Δ x ( n ) + i = 1 m p i ( n ) x ( τ i ( n ) ) = 0 , n N 0 , and the (dual) difference equation with several variable advanced arguments of the form ( E A ) x ( n ) - i = 1 m p i ( n ) x ( σ i ( n ) ) = 0 , n N , where m N , ( p i ( n ) ) , 1 i m , are sequences of positive real numbers and ( τ i ( n ) ) , 1 i m , are sequences of integers such that (1) τ i ( n ) n - 1 , n N 0 , lim n τ i ( n ) = , 1 i m and ( σ i ( n ) ) , 1 i m , are sequences of integers such that (2) σ i ( n ) n + 1 , n N , 1 i m . Here, Δ denotes the forward difference operator Δ x ( n ) = x ( n + 1 ) - x ( n ) and denotes the backward difference operator x ( n ) = x ( n ) - x ( n - 1 ) .

Strong interest in ( E R ) is motivated by the fact that it represents a discrete analogue of the differential equation (see  and the references cited therein) (3) x ( t ) + i = 1 m p i ( t ) x ( τ i ( t ) ) = 0 , t 0 , where, for every i { 1 , , m } , p i is a continuous real-valued function in the interval [ 0 , ) and τ i is a continuous real-valued function on [ 0 , ) such that (4) τ i ( t ) t , t 0 , lim t τ i ( t ) = , while ( E A ) represents a discrete analogue of the advanced differential equation (see [1, 2] and the references cited therein) (5) x ( t ) - i = 1 m p i ( t ) x ( σ i ( t ) ) = 0 , t 1 , where, for every i { 1 , , m } , p i is a continuous real-valued function in the interval [ 1 , ) and σ i is a continuous real-valued function on [ 1 , ) such that (6) σ i ( t ) t , t 1 .

By a solution of ( E R ) , we mean a sequence of real numbers ( x ( n ) ) n - w which satisfies ( E R ) for all n 0 . Here (7) w = - min n 0 1 i m τ i ( n ) . It is clear that, for each choice of real numbers c - w ,   c - w + 1 , ,   c - 1 ,   c 0 , there exists a unique solution ( x ( n ) ) n - w of ( E R ) which satisfies the initial conditions x ( - w ) = c - w ,   x ( - w + 1 ) = c - w + 1 , ,    x ( - 1 ) = c - 1 ,   x ( 0 ) = c 0 .

By a solution of ( E A ) , we mean a sequence of real numbers ( x ( n ) ) n 0 which satisfies ( E A ) for all n 1 .

A solution ( x ( n ) ) n - w (or ( x ( n ) ) n 0 ) of ( E R ) (or ( E A ) ) is called oscillatory, if the terms x ( n ) of the sequence are neither eventually positive nor eventually negative. Otherwise, the solution is said to be nonoscillatory.

In the last few decades, the oscillatory behavior of the solutions of difference and differential equations with several deviating arguments and variable coefficients has been studied. See, for example,  and the references cited therein.

In 2006, Berezansky and Braverman  proved that if (8) limsup n i = 1 m p i ( n ) > 0 , liminf n i = 1 m j = τ ( n ) n - 1 p i ( j ) > 1 e , where τ ( n ) = max 1 i m τ i ( n ) , for all n 0 , then all solutions of ( E R ) oscillate.

Recently, Chatzarakis et al.  established the following theorems.

Theorem 1 (see [<xref ref-type="bibr" rid="B8">9</xref>]).

Assume that the sequences ( τ i ( n ) )    [ ( σ i ( n ) ) ] ,   1 i m , are increasing, (1) [ (2) ] holds, and (9) limsup n i = 1 m j = τ ( n ) n p i ( j ) [ limsup n i = 1 m j = n σ ( n ) p i ( j ) ] > 1 , where τ ( n ) = max 1 i m τ i ( n ) , for all n 0 , [ σ ( n ) = min 1 i m σ i ( n ) , for all n 1 ] , or (10) limsup n i = 1 m p i ( n ) > 0 , liminf n i = 1 m j = τ i ( n ) n - 1 p i ( j ) [ liminf n i = 1 m j = n + 1 σ i ( n ) p i ( j ) ] > 1 e , then all solutions of ( E R )   [ ( E A ) ] oscillate.

Theorem 2 (see [<xref ref-type="bibr" rid="B6">7</xref>, <xref ref-type="bibr" rid="B7">8</xref>]).

Assume that the sequences ( τ i ( n ) )    [ ( σ i ( n ) ) ] ,   1 i m , are increasing and (1) [ (2) ] holds. Set (11) α = min { α i : 1 i m } , α i = liminf n j = τ i ( n ) n - 1 p i ( j ) [ liminf n j = n + 1 σ i ( n ) p i ( j ) ] . If 0 < α 1 / e , and (12) limsup n i = 1 m j = τ ( n ) n p i ( j ) [ limsup n i = 1 m j = n σ ( n ) p i ( j ) ] > 1 - ( 1 - 1 - α ) 2 , or (13) limsup n i = 1 m j = τ ( n ) n p i ( j ) [ limsup n i = 1 m j = n σ ( n ) p i ( j ) ] > 1 - 1 - α - 1 - 2 α - α 2 2 , then all solutions of ( E R )   [ ( E A ) ] oscillate.

The authors study further ( E R ) and ( E A ) and derive new sufficient oscillation conditions. These conditions are the improved and generalized discrete analogues of the oscillation conditions for the corresponding differential equations, which were studied in 1982 by Ladas and Stavroulakis . Examples illustrating the results are also given.

2. Oscillation Criteria 2.1. Retarded Difference Equations

We present new sufficient conditions for the oscillation of all solutions of ( E R ) .

Theorem 3.

Assume that ( τ i ( n ) ) , 1 i m , are increasing sequences of integers such that (1) holds and ( p i ( n ) ) , 1 i m , are sequences of positive real numbers and define α i , 1 i m , by (11). If α i > 0 , 1 i m , and (14) [ i = 1 m ( α i + j = 1 , j i m liminf n k = τ j ( n ) n - 1 p i ( k ) ) ] 1 / m > 1 e , then all solutions of ( E R ) oscillate.

Proof.

Assume, for the sake of contradiction, that ( x ( n ) ) n - w is a nonoscillatory solution of ( E R ) . Then it is either eventually positive or eventually negative. As ( - x ( n ) ) n - w is also a solution of ( E R ) , we may restrict ourselves only to the case where x ( n ) > 0 for all large n . Let n 1 - w be an integer such that x ( n ) > 0 for all n n 1 . Then, there exists n 2 n 1 such that (15) x ( τ i ( n ) ) > 0 , n n 2 , 1 i m . In view of this, ( E R ) becomes (16) Δ x ( n ) = - i = 1 m p i ( n ) x ( τ i ( n ) ) 0 , n n 2 , which means that the sequence ( x ( n ) ) is eventually decreasing.

Next choose a natural number n 3 > n 2 such that (17) x ( n ) < x ( τ i ( n ) ) , n n 3 , 1 i m . Set (18) z i ( n ) = x ( τ i ( n ) ) x ( n ) , n n 3 , 1 i m , φ i = liminf n z i ( n ) , 1 i m . It is obvious that (19) z i ( n ) > 1 , φ i 1 for i = 1,2 , , m . Now we will show that φ i < for i = 1,2 , , m . Indeed, assume that φ i = for some i , i = 1,2 , , m . For this i , by ( E R ) , we have (20) Δ x ( n ) + p i ( n ) x ( τ i ( n ) ) 0 , n n 3 .

At this point, we will establish the following claim.

Claim 1 (cf. ). For each n n 3 , there exists an integer n i * n for each i = 1,2 , , m such that τ i ( n i * ) n - 1 , and (21) j = n n i * p i ( j ) α i - ɛ 2 , (22) j = τ i ( n i * ) n - 1 p i ( j ) > α i - ɛ 2 , where ɛ is an arbitrary real number with 0 < ɛ < α i .

To prove this claim, let us consider an arbitrary real number ɛ with 0 < ɛ < α i . Then by (11) we can choose an integer n 3 n 2 such that (23) j = τ i ( n ) n - 1 p i ( j ) α i - ɛ , n n 3 , 1 i m . Assume, first, that p i ( n ) ( α i - ɛ ) / 2 and choose n i * = n . Then τ i ( n i * ) = τ i ( n ) n - 1 . Moreover, we have (24) j = n n i * p i ( j ) = j = n n p i ( j ) = p i ( n ) α i - ɛ 2 and, by (23), (25) j = τ i ( n i * ) n - 1 p i ( j ) = j = τ i ( n ) n - 1 p i ( j ) > α i - ɛ 2 . So, (21) and (22) are fulfilled. Next, we suppose that p i ( n ) < ( α i - ɛ ) / 2 . It is not difficult to see that (23) guarantees that j = 0 p i ( j ) = . In particular, it holds (26) j = n p i ( j ) = . Thus, as p i ( n ) < ( α i - ɛ ) / 2 , there always exists an integer n i * > n so that (27) j = n n i * - 1 p i ( j ) < α i - ɛ 2 and (21) holds. We assert that τ i ( n i * ) n - 1 . Otherwise, τ i ( n i * ) n . We also have τ i ( n i * ) n i * - 1 . Hence, in view of (27), we get (28) j = τ i ( n i * ) n i * - 1 p i ( j ) j = n n i * - 1 p i ( j ) < α i - ɛ 2 . On the other hand, (23) gives (29) j = τ i ( n i * ) n i * - 1 p i ( j ) α i - ɛ > α i - ɛ 2 . We have arrived at a contradiction, which shows our assertion that τ i ( n i * ) n - 1 . Furthermore, by using (23) (for the integer n i * ) as well as (27), we obtain (30) j = τ i ( n i * ) n - 1 p i ( j ) = j = τ i ( n i * ) n i * - 1 p i ( j ) - j = n n i * - 1 p i ( j ) > ( α i - ɛ ) - α i - ɛ 2 = α i - ɛ 2 and consequently (22) holds true. Our claim has been proved.

Now, summing up (20) from n to n i * , we find (31) x ( n i * + 1 ) - x ( n ) + j = n n i * p i ( j ) x ( τ i ( j ) ) 0 , n n 3 , or (32) x ( n ) ( j = n n i * p i ( j ) ) x ( τ i ( n i * ) ) α i - ɛ 2 x ( τ i ( n i * ) ) , n n 3 . Summing up (20) from τ i ( n i * ) to n - 1 , we find (33) x ( n ) - x ( τ i ( n i * ) ) + j = τ i ( n i * ) n - 1 p i ( j ) x ( τ i ( j ) ) 0 , n n 3 , or (34) x ( τ i ( n i * ) ) ( j = τ i ( n i * ) n - 1 p i ( j ) ) x ( τ i ( n - 1 ) ) α i - ɛ 2 x ( τ i ( n ) ) , n n 3 . Combining (32) and (34), we obtain (35) x ( n ) ( α i - ɛ 2 ) 2 x ( τ i ( n ) ) , n n 3 , or (36) x ( τ i ( n ) ) x ( n ) ( 2 α i - ɛ ) 2 , n n 3 , which means that ( z i ( n ) ) is bounded. This contradicts our assumption that φ i = . Therefore φ i < for every i = 1,2 , , m .

Dividing both sides of ( E R ) by x ( n ) , for n n 3 , we obtain (37) Δ x ( n ) x ( n ) + i = 1 m p i ( n ) x ( τ i ( n ) ) x ( n ) = 0 , or (38) Δ x ( n ) x ( n ) + i = 1 m p i ( n ) z i ( n ) = 0 . Summing up (38) from τ ρ ( n ) to n - 1 for ρ = 1,2 , , m , we find (39) j = τ ρ ( n ) n - 1 Δ x ( j ) x ( j ) + i = 1 m j = τ ρ ( n ) n - 1 p i ( j ) z i ( j ) = 0 . But (40) j = τ ρ ( n ) n - 1 Δ x ( j ) x ( j ) = j = τ ρ ( n ) n - 1 ( x ( j + 1 ) x ( j ) - 1 ) j = τ ρ ( n ) n - 1 ( 1 + ln x ( j + 1 ) x ( j ) - 1 ) = ln x ( n ) x ( τ ρ ( n ) ) , or (41) j = τ ρ ( n ) n - 1 Δ x ( j ) x ( j ) - ln z ρ ( n ) , ρ = 1,2 , , m . Combining (39) and (41), we obtain (42) - ln z ρ ( n ) + i = 1 m j = τ ρ ( n ) n - 1 p i ( j ) z i ( j ) 0 , ρ = 1,2 , , m , or (43) ln z ρ ( n ) i = 1 m j = τ ρ ( n ) n - 1 p i ( j ) z i ( j ) , ρ = 1,2 , , m . Taking limit inferiors on both sides of the above inequalities (43), we obtain (44) ln φ ρ i = 1 m φ i ( liminf n k = τ ρ ( n ) n - 1 p i ( k ) ) , ρ = 1,2 , , m and by adding we find (45) i = 1 m ln φ i i = 1 m φ i ( j = 1 m liminf n k = τ j ( n ) n - 1 p i ( k ) ) . Set (46) f ( φ 1 , φ 2 , , φ m ) i = 1 m ln φ i - i = 1 m φ i ( j = 1 m liminf n k = τ j ( n ) n - 1 p i ( k ) ) . Clearly (47) f ( φ 1 , φ 2 , , φ m ) 0 , φ 1 , φ 2 , , φ m 1 . Since (48) f φ i = 1 φ i - j = 1 m liminf n k = τ j ( n ) n - 1 p i ( k ) = 0 , for (49) φ i = 1 j = 1 m liminf n k = τ j ( n ) n - 1 p i ( k ) , i = 1,2 , , m , the function f has a maximum at the critical point (50) ( 1 j = 1 m liminf n k = τ j ( n ) n - 1 p 1 ( k ) , m , 1 j = 1 m liminf n k = τ j ( n ) n - 1 p m ( k ) ) since the quadratic form (51) i , j = 1 m 2 f φ i φ j b i b j = - i = 1 m b i 2 φ i 2 < 0 . Since f ( φ 1 , φ 2 , , φ m ) 0 , the maximum of f at the critical point should be nonnegative. Thus, (52) i = 1 m [ - ln ( j = 1 m liminf n k = τ j ( n ) n - 1 p i ( k ) ) ] - m 0 , that is, (53) max φ i 1 f ( φ 1 , φ 2 , , φ m ) = - ln i = 1 m ( j = 1 m liminf n k = τ j ( n ) n - 1 p i ( k ) ) - m . Hence (54) i = 1 m ( j = 1 m liminf n k = τ j ( n ) n - 1 p i ( k ) ) 1 e m , or (55) [ i = 1 m ( α i + j = 1 , j i m liminf n k = τ j ( n ) n - 1 p i ( k ) ) ] 1 / m 1 e , which contradicts (14).

The proof of the theorem is complete.

Theorem 4.

Assume that ( τ i ( n ) ) , 1 i m , are increasing sequences of integers such that (1) holds and ( p i ( n ) ) , 1 i m , are sequences of positive real numbers and define α i , 1 i m , by (11). If α i > 0 , 1 i m , and (56) 1 m i = 1 m α i + 2 m i < j i , j = 1 m ( liminf n k = τ j ( n ) n - 1 p i ( k ) mmmmmmmmmm × liminf n k = τ i ( n ) n - 1 p j ( k ) ) 1 / 2 > 1 e , then all solutions of ( E R ) oscillate.

Proof.

Assume, for the sake of contradiction, that ( x ( n ) ) n - w is a nonoscillatory solution of ( E R ) . Then it is either eventually positive or eventually negative. As ( - x ( n ) ) n - w is also a solution of ( E R ) , we may restrict ourselves only to the case where x ( n ) > 0 for all large n . Let n 1 - w be an integer such that x ( n ) > 0 for all n n 1 . Then, there exists n 2 n 1 such that (57) x ( τ i ( n ) ) > 0 , n n 2 , 1 i m . In view of this, ( E R ) becomes (58) Δ x ( n ) = - i = 1 m p i ( n ) x ( τ i ( n ) ) 0 , n n 2 , which means that the sequence ( x ( n ) ) is eventually decreasing.

Taking into account the fact that φ i < for i = 1,2 , , m (see proof of Theorem 3), by using (44) and the fact that (59) 1 e > ln φ ρ φ ρ , ρ = 1,2 , , m , we obtain (60) 1 e > i = 1 m ( liminf n k = τ j ( n ) n - 1 p i ( k ) ) φ i φ j , j = 1,2 , , m . Adding these inequalities we have (61) m e i = 1 m ( liminf n k = τ i ( n ) n - 1 p i ( k ) ) + i < j i , j = 1 m [ ( liminf n k = τ j ( n ) n - 1 p i ( k ) ) φ i φ j mmmmkm + ( liminf n k = τ i ( n ) n - 1 p j ( k ) ) φ j φ i ] i = 1 m ( liminf n k = τ i ( n ) n - 1 p i ( k ) ) + 2 i < j i , j = 1 m [ ( liminf n k = τ j ( n ) n - 1 p i ( k ) ) mmmmmm × ( liminf n k = τ i ( n ) n - 1 p j ( k ) ) ] 1 / 2 , or (62) 1 m i = 1 m α i + 2 m i < j i , j = 1 m ( liminf n k = τ j ( n ) n - 1 p i ( k ) mmmmmmmmm × liminf n k = τ i ( n ) n - 1 p j ( k ) ) 1 / 2 1 e , which contradicts (56).

The proof of the theorem is complete.

Similar oscillation theorems for the (dual) advanced difference equation ( E A ) can be derived easily. The proofs of these theorems are omitted, since they follow a similar procedure as in Section 2.1.

Theorem 5.

Assume that ( σ i ( n ) ) , 1 i m , are increasing sequences of integers such that (2) holds and ( p i ( n ) ) , 1 i m , are sequences of positive real numbers and define α i , 1 i m , by (11). If α i > 0 , 1 i m , and (63) [ i = 1 m ( α i + j = 1 , j i m liminf n k = n + 1 σ j ( n ) p i ( k ) ) ] 1 / m > 1 e , then all solutions of ( E A ) oscillate.

Theorem 6.

Assume that ( σ i ( n ) ) , 1 i m , are increasing sequences of integers such that (2) holds and ( p i ( n ) ) , 1 i m , are sequences of positive real numbers and define α i , 1 i m , by (11). If α i > 0 , 1 i m , and (64) 1 m i = 1 m α i + 2 m i < j i , j = 1 m ( liminf n k = n + 1 σ j ( n ) p i ( k ) m m m m m m m m m m × liminf n k = n + 1 σ i ( n ) p j ( k ) ) 1 / 2 > 1 e , then all solutions of ( E A ) oscillate.

2.3. Special Cases

In the case where p i , i = 1,2 , , m , are positive real constants and τ i are constant retarded arguments of the form τ i ( n ) = n - k i , [ σ i are constant advanced arguments of the form σ i ( n ) = n + k i ], k i N , i = 1,2 , , m , equation ( E R ) [ ( E A ) ] takes the form ( E ) Δ x ( n ) + i = 1 m p i x ( n - k i ) = 0 , n N 0 , [ x ( n ) - i = 1 m p i x ( n + k i ) = 0 , n N ] . For this equation, as a consequence of Theorems 3  and 4 , we have the following corollary.

Corollary 7.

Assume that (65) [ i = 1 m p i ] 1 / m ( i = 1 m k i ) > 1 e , or (66) 1 m ( i = 1 m p i k i ) 2 > 1 e . Then all solutions of ( E ) oscillate.

Remark 8.

A research question that arises is whether Theorems 36 are valid, even in the case where the coefficients p ( n ) oscillate (see [15, 16]). Then our results would be comparable to those in [15, 16]. This is a question that we currently study and expect to have some results soon.

3. Examples

The following two examples illustrate that the conditions for oscillations (65) and (66) are independent. They are chosen in such a way that only one of them is satisfied.

Example 1.

Consider the retarded difference equation (67) Δ x ( n ) + 1 5 e x ( n - 1 ) + 1 6 e x ( n - 2 ) + 1000 7198 e x ( n - 3 ) = 0 , n N 0 . Here m = 3 , τ 1 ( n ) = n - 1 , τ 2 ( n ) = n - 2 , τ 3 ( n ) = n - 3 , and (68) τ ( n ) = max 1 i 3 τ i ( n ) = n - 1 . It is easy to see that (69) [ i = 1 3 p i ] 1 / 3 ( i = 1 3 k i ) = 1 5 e · 1 6 e · 1000 7198 e 3 · ( 1 + 2 + 3 ) 0.3679135 > 1 e . That is, condition (65) of Corollary 7 is satisfied and therefore all solutions of equation (67) oscillate.

However, (70) 1 3 ( i = 1 3 p i k i ) 2 = 1 3 ( 1 5 e · 1 + 1 6 e · 2 + 1000 7198 e · 3 ) 2 0.3420547 < 1 e . That is, condition (66) of Corollary 7 is not satisfied.

Observe that (71) α 1 = liminf n j = n - 1 n - 1 p 1 = 1 5 e 0.0735758 , α 2 = liminf n j = n - 2 n - 1 p 2 = 2 6 e 0.1226264 , α 3 = liminf n j = n - 3 n - 1 p 3 = 3000 7198 e 0.1533256 . Thus (72) α = min { α i : 1 i 3 } 0.0735758 < 1 e . Also, (73) limsup n i = 1 3 j = n - 1 n p i = limsup n [ 2 · 1 5 e + 2 · 1 6 e + 2 · 1000 7198 e ] 0.3719953 < 1 , 0.3719953 < 1 - ( 1 - 1 - α ) 2 0.9985944 , 0.3719953 < 1 - 1 - α - 1 - 2 α - α 2 2 0.997069 , liminf n i = 1 3 j = n - 1 n - 1 p i = liminf n [ 1 5 e + 1 6 e + 1000 7198 e ] mmmmmmmiim 0.1859976 < 1 e , liminf n i = 1 3 j = τ i ( n ) n - 1 p i ( j ) = liminf n [ 1 5 e + 2 6 e + 3000 7198 e ] mmmmmmmmmm 0.349528 < 1 e and therefore none of the conditions (9), (12), (13), (8), and (10) are satisfied.

Example 2.

Consider the advanced difference equation (74) x ( n ) - 5 100 x ( n + 1 ) - 22 100 x ( n + 2 ) = 0 , n N . Here m = 2 , σ 1 ( n ) = n + 1 , σ 2 ( n ) = n + 2 , and (75) σ ( n ) = min 1 i 2 σ i ( n ) = n + 1 . It is easy to see that (76) 1 2 ( i = 1 2 p i k i ) 2 = 1 2 ( 5 100 · 1 + 22 100 · 2 ) 2 mmmmmimm 0.393323969 > 1 e . That is, condition (66) of Corollary 7 is satisfied and therefore all solutions of (74) oscillate.

However, (77) [ i = 1 2 p i ] 1 / 2 ( i = 1 2 k i ) = 5 100 · 22 100 · ( 1 + 2 ) mmmmmmmkmm 0.314642654 < 1 e . That is, condition (65) of Corollary 7 is not satisfied.

Observe that (78) α 1 = liminf n j = n + 1 n + 1 p 1 = 5 100 = 0.05 , α 2 = liminf n j = n + 1 n + 2 p 2 = 2 · 22 100 = 0.44 . Thus (79) α = min { α i : 1 i 2 } = 0.05 < 1 e . Also, (80) limsup n i = 1 2 j = n n + 1 p i = limsup n [ 2 · 5 100 + 2 · 22 100 ] = 0.54 < 1 , 0.54 < 1 - ( 1 - 1 - α ) 2 0.999358869 , 0.54 < 1 - 1 - α - 1 - 2 α - α 2 2 0.998682383 and therefore none of the conditions (9), (12), and (13) are satisfied.

At this point, we give an example with general retarded arguments illustrating the main result of Theorem 3. Similarly, one can construct examples to illustrate Theorems 46.

Example 3.

Consider the delay difference equation (81) Δ x ( n ) + c 3 n x ( [ 0.5 n ] ) + 2 c 3 n x ( [ 0.2 n ] ) = 0 , n 10 , with c = 17 / 50 .

Here τ 1 ( n ) = [ 0.5 n ] and τ 2 ( n ) = [ 0.2 n ] denote the integer parts of 0.5 n and 0.2 n . Observe that the sequences τ 1 ( n ) and τ 2 ( n ) are increasing, lim n τ 1 ( n ) = + , lim n τ 2 ( n ) = + , and (82) τ 1 ( n ) = [ 0.5 n ] 0.5 n < n or τ 1 ( n ) n - 1 , τ 2 ( n ) = [ 0.2 n ] 0.2 n < n or τ 2 ( n ) n - 1 , k k k k i mmmmmmmmmmmmmmm n 10 . Observe that, for a positive decreasing function f ( x ) , the following inequality holds: (83) b - 1 b f ( x ) d x f ( b ) b b + 1 f ( x ) d x . Based on the above inequality, we will show that (84) liminf n k = [ ( a / b ) n ] n - 1 Q k = Q · ln b a for any a , b N , a < b , and any real number Q . Indeed, (85) k = [ ( a / b ) n ] n - 1 Q k Q k = [ ( a / b ) n ] n - 1 k k + 1 d s s = Q [ ( a / b ) n ] n d s s = Q · ln n [ ( a / b ) n ] , k = [ ( a / b ) n ] n - 1 Q k Q k = [ ( a / b ) n ] n - 1 k - 1 k d s s = Q [ ( a / b ) n ] - 1 n - 1 d s s = Q · ln n - 1 [ ( a / b ) n ] - 1 . It is easy to see that (86) lim n Q · ln n [ ( a / b ) n ] = lim n Q · ln n - 1 [ ( a / b ) n ] - 1 = Q · ln b a . From the above, it follows that (87) liminf n k = [ ( a / b ) n ] n - 1 Q k = Q · ln b a . Therefore (88) α 1 = liminf n k = [ 0.5 n ] n - 1 c 3 k = c 3 ln 2 > 0 , α 2 = liminf n k = [ 0.2 n ] n - 1 2 c 3 k = 2 c 3 ln 5 > 0 , liminf n k = [ 0.2 n ] n - 1 c 3 k = c 3 ln 5 , liminf n k = [ 0.5 n ] n - 1 2 c 3 k = 2 c 3 ln 2 . Hence (89) [ i = 1 2 ( α i + j = 1 , j i 2 liminf n k = τ j ( n ) n - 1 p i ( k ) ) ] 1 / 2 = [ c 3 ln 2 + c 3 ln 5 ] · [ 2 c 3 ln 2 + 2 c 3 ln 5 ] = c 2 ln 10 3 0.369052667 > 1 e . That is, condition (14) of Theorem 3 is satisfied and therefore all solutions of (81) oscillate.

Observe, however, that (90) i = 1 2 j = τ ( n ) n - 1 p i ( j ) = j = [ 0.5 n ] n - 1 c 3 j + j = [ 0.5 n ] n - 1 2 c 3 j = j = [ 0.5 n ] n - 1 c j c ln 2 0.235670041 m m m m m m m i m m m m m as n , i = 1 2 j = τ ( n ) n p i ( j ) = j = [ 0.5 n ] n c 3 j + j = [ 0.5 n ] n 2 c 3 j = j = [ 0.5 n ] n c j c ln 2 0.235670041 mmmmmmimmmmm as n . Thus (91) liminf n i = 1 2 j = τ ( n ) n - 1 p i ( j ) = limsup n i = 1 m j = τ ( n ) n p i ( j ) 0.235670041 < 1 e . Also, (92) α = min { α i : 1 i 2 } = α 1 = c 3 ln 2 0.07855668 , 0.235670041 < 1 - ( 1 - 1 - α ) 2 0.998393464 , 0.235670041 < 1 - 1 - α - 1 - 2 α - α 2 2 0.996639107 and therefore none of the conditions (8), (9), (12), and (13) are satisfied.

Conflict of Interests

The authors declare that there is no conflict of interests regarding the publication of this paper.

Acknowledgment

The second author was supported by the Serbian Ministry of Science, Technology and Development for Scientific Research Grant no. III44006.

Fukagai N. Kusano T. Oscillation theory of first order functional-differential equations with deviating arguments Annali di Matematica Pura ed Applicata. Serie Quarta 1984 136 95 117 10.1007/BF01773379 MR765918 2-s2.0-0001816332 Ladas G. Stavroulakis I. P. Oscillations caused by several retarded and advanced arguments Journal of Differential Equations 1982 44 1 134 152 10.1016/0022-0396(82)90029-8 MR651691 2-s2.0-58149404420 Li B. Oscillation of first order delay differential equations Proceedings of the American Mathematical Society 1996 124 12 3729 3737 10.1090/S0002-9939-96-03674-X MR1363175 ZBL0865.34057 2-s2.0-21444448060 Arino O. Györi I. Jawhari A. Oscillation criteria in delay equations Journal of Differential Equations 1984 53 1 115 123 10.1016/0022-0396(84)90028-7 MR747409 2-s2.0-0000981694 Berezansky L. Braverman E. On existence of positive solutions for linear difference equations with several delays Advances in Dynamical Systems and Applications 2006 1 1 29 47 MR2287633 Berezansky L. Braverman E. Positive solutions for a scalar differential equation with several delays Applied Mathematics Letters 2008 21 6 636 640 10.1016/j.aml.2007.07.017 MR2412391 2-s2.0-41649120525 Chatzarakis G. E. Kusano T. Stavroulakis I. P. Oscillation conditions for difference equations with several variable to appear in Mathematica Bohemica Chatzarakis G. E. Manojlovic J. Pinelas S. Stavroulakis I. P. Oscillation criteria of difference equations with several deviating arguments to appear in Yokohama Mathematical Journal Chatzarakis G. E. Pinelas S. Stavroulakis I. P. Oscillations of difference equations with several deviated arguments to appear in Aequationes Mathematicae 10.1007/s00010-013-0238-2 Grammatikopoulos M. K. Koplatadze R. Stavroulakis I. P. On the oscillation of solutions of first order differential equations with retarded arguments Georgian Mathematical Journal 2003 10 1 63 76 MR1990688 Hunt B. R. Yorke J. A. When all solutions of x ' = j = 1 n q j t x t - T j t oscillate Journal of Differential Equations 1984 53 2 139 145 10.1016/0022-0396(84)90036-6 MR748236 2-s2.0-0000959553 Jaroš J. Stavroulakis I. P. Necessary and sufficient conditions for oscillations of difference equations with several delays Utilitas Mathematica 1994 45 187 195 MR1284029 Luo X. N. Zhou Y. Li C. F. Oscillation of a nonlinear difference equation with several delays Mathematica Bohemica 2003 128 3 309 317 MR2012607 Tang X. H. Zhang R. Y. New oscillation criteria for delay difference equations Computers & Mathematics with Applications 2001 42 10-11 1319 1330 10.1016/S0898-1221(01)00243-7 MR1861531 2-s2.0-0035500920 Berezansky L. Chatzarakis G. E. Domoshnitsky A. Stavroulakis I. P. Oscillations of difference equations with several oscillating coefficients Abstract and Applied Analysis 2014 2014 9 392097 10.1155/2014/392097 MR3219368 Bohner M. Chatzarakis G. E. Stavroulakis I. P. Qualitative behavior of solutions of difference equations with several oscillating coefficients Arabian Journal of Mathematics 2014 3 1 1 13 10.1007/s40065-013-0087-9 MR3177427