New Oscillatory Theorems for Third-Order Nonlinear Delay Dynamic Equations on Time Scales

This paper is concern ed with the oscillatory properties of the third-order nonlinear delay dynamic equations of the form on time scales , where is a quotient of odd positive integers. Applying the inequality technique we present two new sufficient conditions which ensure that every solution of equations is oscillatory or converges to zero. The results obtained improve and complement some known results in the literature.

• Oscillatory
• Third Order
• Delay Dynamic Equation
• Generalized Riccati Transformation
• Time Scale

1. Introduction

Beginning with the landmark contribution work of Hilger [1] , the time scales theory, which in order to unify the continuous and discrete analysis, has received significant attention. In the recent years, there has been increasing interest in obtaining sufficient conditions for the oscillation and nonoscillation of solutions of various equations on time scales; we refer the reader to the papers [2] - [18] . Following this trend, we shall consider oscillation for the third-order nonlinear delay dynamic equation

( r 2 ( t ) [ ( r 1 ( t ) x Δ ( t ) ) Δ ] α ) Δ + q ( t ) f ( x [ τ ( t ) ] ) = 0,     t ∈ T ,     t ≥ t 0 , (1)

where α ≥ 1 is a quotient of odd positive integers.

Throughout this paper, assume that

(H₁) T is a time scale (i.e., a nonempty closed subset of the real numbers ℝ ) which is unbounded above, and t 0 ∈ T with t 0 > 0 . We define the time scale interval of the form [ t 0 , ∞ ) T by [ t 0 , ∞ ) T = [ t 0 , ∞ ) ∩ T .

(H₂) r 1 ( t ) , r 2 ( t ) , q ( t ) are positive, real-valued rd-continuous functions defined on T , and r 1 ( t ) , r 2 ( t ) satisfy

∫ t 0 ∞ 1 r 1 ( s ) Δ s = ∞ , ∫ t 0 ∞ ( 1 r 2 ( s ) ) 1 α Δ s = ∞ .

(H₃) τ : T → T is a strictly increasing and differentiable function such that

τ ( t ) ≤ t , lim t → ∞ τ ( t ) = ∞   and   τ ( T ) = T .

(H₄) f : ℝ → ℝ is a continuous function, and there exists a positive number

K such that f ( x ) x α ≥ K > 0 for x ≠ 0 .

By a solution of (1) , we mean a nontrivial function x ( t ) satisfying (1) which has the properties x ( t ) ∈ C r d 1 ( [ T x , ∞ ) T , ℝ ) for T x ≥ t 0 , and

r 2 ( t ) [ ( r 1 ( t ) x Δ ( t ) ) Δ ] α ∈ C r d 1 ( [ T x , ∞ ) T , ℝ ) . Our attention is restricted to those

solutions of (1) which satisfy s u p { | x ( t ) | : t ≥ T } > 0 for all T ≥ T x . A solution x of Equation (1) is said to be oscillatory on [ T x , ∞ ) T if it is neither eventually positive nor eventually negative. Otherwise it is called nonoscillatory. The equation itself is called oscillatory if all its solutions are oscillatory.

If α = 1 , τ ( t ) = t , then (1) simplifies to the third-order nonlinear dynamic equation

( r 2 ( t ) [ ( r 1 ( t ) x Δ ( t ) ) Δ ] ) Δ + q ( t ) f ( x ( t ) ) = 0,     t ∈ T ,     t ≥ t 0 . (2)

If, furthermore, r 1 ( t ) = r 2 ( t ) = 1 , f ( x ) = x , τ ( t ) = t , then (1) reduces to the third-order linear dynamic equation

x Δ Δ Δ ( t ) + q ( t ) x ( t ) = 0 , t ∈ T ,   t ≥ t 0 . (3)

If, in addition, α = 1 , then (1) reduces to the nonlinear delay dynamic equation

( r 2 ( t ) [ ( r 1 ( t ) x Δ ( t ) ) Δ ] ) Δ + q ( t ) f ( x [ τ ( t ) ] ) = 0,     t ∈ T ,     t ≥ t 0 . (4)

In [11] , Erbe et al. established some sufficient conditions which ensure that every solution of Equation (2) is oscillatory or converges to zero. In [12] , Erbe et al. studied the third-order linear dynamic Equation (3), and they obtained Hille and Nehari type oscillation criteria for the Equation (3). In [16] , Han et al. extended and improved the results of [11] [12] , meanwhile obtained some oscillatory criteria for the Equation (4). In [18] , Gao et al. considered the third-order nonlinear dynamic Equation (1). By employing the generalized Riccati transformation and the integral averaging technique, they established three sufficient conditions which ensure that every solution of Equation (1) is oscillatory or converges to zero. On this basis, we continue to study Equation (1). If (4.11) in ( [18] , Theorem 4.3) is not hold, then we obtain two new sufficient conditions which guarantee that every solution of Equation (1) is oscillatory or converges to zero. Our results will improve some previous results. The usual notation and concepts of the time scales calculus, which will be used throughout the paper, can be found in [19] [20] .

2. Several Lemmas

Lemma 1 Assume that x ( t ) is an eventually positive solution of (1). Then there exists T ∈ [ t 0 , ∞ ) T such that either

(I) x ( t ) > 0 , x Δ ( t ) > 0 , ( r 1 ( t ) x Δ ( t ) ) Δ > 0 , t ∈ [ T , ∞ ) T ;

or

(II) x ( t ) > 0 , x Δ ( t ) < 0 , ( r 1 ( t ) x Δ ( t ) ) Δ > 0 , t ∈ [ T , ∞ ) T .

The proof is similar to that of ( [11] , Lemma 1).

Lemma 2 (see [19] , Theorem 1.90]) If x is differentiable, then

( x γ ) Δ = γ x Δ ∫ 0 1 [ h x σ + ( 1 − h ) x ] γ − 1 d h . (5)

Lemma 3 (see [21] , Theorem 41]) Assume that X and Y are nonnegative real numbers. Then

λ X Y λ − 1 − X λ ≤ ( λ − 1 ) Y λ   for   all   λ > 1 , (6)

where the equality holds if and only if X = Y .

Throughout this paper, for sufficiently large T, we denote

R ( t , T ) = ∫ T t ( 1 r 2 ( s ) ) 1 α Δ s .

Lemma 4 Assume that x ( t ) is an eventually positive solution of (1) which satisfies case (I) in Lemma 1. Then there exists T ∈ [ t 0 , ∞ ) T , such that

x Δ ( t ) ≥ R ( t , T ) r 1 ( t ) r 2 1 α ( t ) ( r 1 ( t ) x Δ ( t ) ) Δ , t ∈ [ T , ∞ ) T . (7)

The proof is similar to that of (18], Lemma 3.4).

Lemma 5 Assume that x ( t ) is an eventually positive solution of (1) which satisfies case (I) in Lemma 1. Furthermore, assume that r 1 Δ ( t ) ≤ 0 and

∫ t 0 ∞     q ( s ) τ α ( s ) Δ s = ∞ . (8)

Then there exists T ∈ [ t 0 , ∞ ) T such that x ( t ) > t x Δ ( t ) , and x ( t ) t is strictly decreasing on [ T , ∞ ) T .

The proof is similar to that of ( [16] , Lemma 2.3).

Lemma 6 Assume that x ( t ) is an eventually positive solution of (1) which satisfies case (II) in Lemma 1. Furthermore,

∫ t 0 ∞ 1 r 1 ( t ) ∫ t ∞ [ 1 r 2 ( s ) ∫ s ∞ q ( u ) Δ u ] 1 α Δ s Δ t = ∞ . (9)

Then lim t → ∞ x ( t ) = 0 .

Proof Assume that x ( t ) is an eventually positive solution of (1) which satisfies the case (II) in Lemma 1. Then x ( t ) is decreasing and lim t → ∞ x ( t ) = l ≥ 0 . We assert that l = 0 . If not, then x [ τ ( t ) ] ≥ x ( t ) ≥ l > 0 for t ∈ [ t 0 , ∞ ) T . Integrating (1) from t to ∞ , we get

− r 2 ( t ) [ ( r 1 ( t ) x Δ ( t ) ) Δ ] α ≤ − K ∫ t ∞ q ( s ) x α [ τ ( s ) ] Δ s ≤ − K l α ∫ t ∞ q ( s ) Δ s ,     t ∈ [ t 0 , ∞ ) T .

Hence, we have

− ( r 1 ( t ) x Δ ( t ) ) Δ ≤ − l [ 1 r 2 ( t ) ∫ t ∞ K q ( s ) Δ s ] 1 α .

Integrating the above inequality from t to ∞ , we obtain

r 1 ( t ) x Δ ( t ) ≤ − l K 1 α ∫ t ∞ [ 1 r 2 ( s ) ∫ s ∞ q ( u ) Δ u ] 1 α Δ s .

Integrating the last inequality again from T to t, we have

x ( t ) − x ( T ) ≤ − l K 1 α ∫ T t 1 r 1 ( s ) ∫ s ∞ [ 1 r 2 ( u ) ∫ u ∞ q ( v ) Δ v ] 1 α Δ u Δ s .

Since condition (9) holds, we obtain lim t → ∞ x ( t ) = − ∞ , which contradicts x ( t ) > 0 . Hence l = 0 . This completes the proof.

Lemma 7 (see [22] , Theorem 3]) Let a , b ∈ T and a < b , for positive rd-continuous functions f , g : [ a , b ] → ℝ , we have

∫ a b | f ( s ) g ( s ) | Δ s ≤ ( ∫ a b | f ( s ) | p Δ s ) 1 p ( ∫ a b | g ( s ) | q Δ s ) 1 q , (10)

where p > 1 and 1 p + 1 q = 1 .

3. Main Results

Theorem 1 Assume that (8) and (9) hold, r 1 Δ ( t ) ≤ 0 . Furthermore, assume that there exist functions H , h ∈ C r d ( D , ℝ ) , where D ≡ { ( t , s ) : t ≥ s ≥ T } such that

H ( t , t ) = 0 , t ≥ T ; H ( t , s ) > 0 , t > s ≥ T , (11)

and H has a nonpositive continuous D-partial derivative H Δ s ( t , s ) with respect to the second variable and satisfies

H Δ s ( σ ( t ) , s ) + ( δ Δ ( s ) ) + δ ( s ) H ( σ ( t ) , s ) = − h ( t , s ) δ ( s ) H α α + 1 ( σ ( t ) , s ) , (12)

and, for all sufficiently large T, that there exists T 0 ≥ T ,

0 < inf s ≥ T 0 [ lim inf t → ∞ H ( σ ( t ) , s ) H ( σ ( t ) , T 0 ) ] ≤ ∞ , (13)

lim sup t → ∞ 1 H ( σ ( t ) , T 0 ) ∫ T 0 σ ( t ) h − α + 1 ( t , s ) [ σ α ( s ) r 1 ( s ) ] α [ s α δ ( σ ( s ) ) R ( s , T ) ] α Δ s < ∞ , (14)

and a real rd-continuous function Ψ : [ t 0 , ∞ ) T → ℝ such that

∫ T ∞ s α δ ( σ ( s ) ) R ( s , T ) δ 1 + 1 α ( s ) σ α ( s ) r 1 ( s ) Ψ + 1 + 1 α ( s ) Δ s = ∞ , (15)

lim sup t → ∞ 1 H ( σ ( t ) , T 0 ) ∫ T 0 σ ( t ) [ H ( σ ( t ) , s ) Q ( s ) − h − α + 1 ( t , s ) [ σ α ( s ) r 1 ( s ) ] α ( α + 1 ) α + 1 [ s α δ ( σ ( s ) ) R ( s , T ) ] α ] Δ s ≥ Ψ ( T 0 ) , (16)

where δ ( t ) is a positive D-differentiable function, Q ( t ) = K q ( t ) δ ( σ ( t ) ) ( τ ( t ) σ ( t ) ) α .

h − ( t , s ) = max { 0 , − h ( t , s ) } , h + ( t , s ) = max { 0 , h ( t , s ) } , Ψ + ( t ) = max { 0 , Ψ ( t ) } .

Then every solution x ( t ) of Equation (1) is either oscillatory or converges to zero.

Proof Assume that (1) has a nonoscillatory solution x ( t ) on [ t 0 , ∞ ) T . Without loss generality we may assume that there exists sufficiently large T ≥ t 0 such that x ( t ) > 0 and x [ τ ( t ) ] > 0 for all t ∈ [ T , ∞ ) T . By Lemma 1, we see that x ( t ) satisfies either case (I) or case (II).

If case (I) holds, then x Δ ( t ) > 0, t ∈ [ T , ∞ ) T . Define the function ω ( t ) by

ω ( t ) = δ ( t ) r 2 ( t ) ( ( r 1 ( t ) x Δ ( t ) ) Δ x ( t ) ) α , t ∈ [ T , ∞ ) T .

Obviously ω ( t ) > 0 . Using the product and quotient rule of D-differential, we obtain

ω Δ ( t ) = δ Δ ( t ) δ ( t ) ω ( t ) − δ ( σ ( t ) ) q ( t ) f ( x [ τ ( t ) ] ) x α ( σ ( t ) )     − δ ( σ ( t ) ) r 2 ( t ) [ ( r 1 ( t ) x Δ ( t ) ) Δ x ( t ) ] α ( x α ( t ) ) Δ x α ( σ ( t ) ) .

By Lemma 2, we get

ω Δ ( t ) ≤ ( δ Δ ( t ) ) + δ ( t ) ω ( t ) − δ ( σ ( t ) ) q ( t ) f ( x [ τ ( t ) ] ) x α ( σ ( t ) )     − α δ ( σ ( t ) ) r 2 ( t ) [ ( r 1 ( t ) x Δ ( t ) ) Δ x ( t ) ] α x α ( t ) x Δ ( t ) x α ( σ ( t ) ) x ( t ) ,

where ( δ Δ ( t ) ) + = m a x { 0, δ Δ ( t ) } , and get

ω Δ ( t ) ≤ ( δ Δ ( t ) ) + δ ( t ) ω ( t ) − δ ( σ ( t ) ) q ( t ) f ( x [ τ ( t ) ] ) x α ( σ ( t ) )     − α δ ( σ ( t ) ) r 2 ( t ) [ ( r 1 ( t ) x Δ ( t ) ) Δ x ( t ) ] α x α ( t ) x Δ ( t ) x α ( σ ( t ) ) x ( t ) .

From Lemma 5, we obtain x ( τ ( t ) ) x ( σ ( t ) ) ≥ τ ( t ) σ ( t ) , x ( t ) x ( σ ( t ) ) ≥ t σ ( t ) , and using (7), so we obtain

ω Δ ( t ) ≤ − K δ ( σ ( t ) ) q ( t ) ( τ ( t ) σ ( t ) ) α + ( δ Δ ( t ) ) + δ ( t ) ω ( t )     − α δ ( σ ( t ) ) R ( t , T ) r 1 ( t ) ( t σ ( t ) ) α r 2 1 + 1 α ( t ) [ ( r 1 ( t ) x Δ ( t ) ) Δ x ( t ) ] α + 1 .

Hence, by the definition of ω ( t ) , Q ( t ) , we obtain

ω Δ ( t ) ≤ − Q ( t ) + ( δ Δ ( t ) ) + δ ( t ) ω ( t ) − α δ ( σ ( t ) ) R ( t , T ) δ 1 + 1 α ( t ) r 1 ( t ) ( t σ ( t ) ) α ω 1 + 1 α ( t ) . (17)

Multiplying both sides of (17), with t replaced by s, by H ( σ ( t ) , s ) , integrating with respect to s from T 0 to σ ( t ) , t ≥ T 0 ≥ T , we get

∫ T 0 σ ( t ) H ( σ ( t ) , s ) Q ( s ) Δ s ≤ − ∫ T 0 σ ( t ) H ( σ ( t ) , s ) ω Δ ( s ) Δ s + ∫ T 0 σ ( t ) H ( σ ( t ) , s ) ( δ Δ ( s ) ) + δ ( s ) ω ( s ) Δ s       − ∫ T 0 σ ( t ) α s α H ( σ ( t ) , s ) δ ( σ ( s ) ) R ( s , T ) δ 1 + 1 α ( s ) σ α ( s ) r 1 ( s ) ω 1 + 1 α ( s ) Δ s .

Integrating by parts and using (11), we obtain

∫ T 0 σ ( t ) H ( σ ( t ) , s ) Q ( s ) Δ s ≤ H ( σ ( t ) , T 0 ) ω ( T 0 )     + ∫ T 0 σ ( t ) H Δ s ( σ ( t ) , s ) ω ( s ) Δ s + ∫ T 0 σ ( t ) H ( σ ( t ) , s ) ( δ Δ ( s ) ) + δ ( s ) ω ( s ) Δ s     − ∫ T 0 σ ( t ) α s α H ( σ ( t ) , s ) δ ( σ ( s ) ) R ( s , T ) δ 1 + 1 α ( s ) σ α ( s ) r 1 ( s ) ω 1 + 1 α ( s ) Δ s ≤ H ( σ ( t ) , T 0 ) ω ( T 0 ) + ∫ T 0 σ ( t ) [ − h ( t , s ) H α 1 + α ( σ ( t ) , s ) δ ( s ) ω ( s )     − α s α H ( σ ( t ) , s ) δ ( σ ( s ) ) R ( s , T ) δ 1 + 1 α ( s ) σ α ( s ) r 1 ( s ) ω 1 + 1 α ( s ) ] Δ s ,

and so

∫ T 0 σ ( t ) H ( σ ( t ) , s ) Q ( s ) Δ s ≤ H ( σ ( t ) , T 0 ) ω ( T 0 ) + ∫ T 0 σ ( t ) [ − h ( t , s ) H α 1 + α ( σ ( t ) , s ) δ ( s ) ω ( s )     − α s α H ( σ ( t ) , s ) δ ( σ ( s ) ) R ( s , T ) δ 1 + 1 α ( s ) σ α ( s ) r 1 ( s ) ω 1 + 1 α ( s ) ] Δ s , (18)

Now set

X λ = α s α H ( σ ( t ) , s ) δ ( σ ( s ) ) R ( s , T ) δ 1 + 1 α ( s ) σ α ( s ) r 1 ( s ) ω λ ( s ) ,

Y λ − 1 = h − ( t , s ) [ σ α ( s ) r 1 ( s ) ] 1 λ λ [ α s α δ ( s ) R ( s , T ) ] 1 λ ,

where λ = α + 1 α > 1 , X ≥ 0 and Y ≥ 0 . Using the inequality (6), we obtain

h − ( t , s ) H 1 λ ( σ ( t ) , s ) δ ( s ) ω ( s ) − α s α H ( σ ( t ) , s ) δ ( σ ( s ) ) R ( s , T ) δ λ − 1 ( s ) σ α ( s ) r 1 ( s ) ω λ ( s ) ≤ h − α + 1 ( t , s ) [ σ α ( s ) r 1 ( s ) ] α ( α + 1 ) α + 1 [ s α δ ( σ ( s ) ) R ( s , T ) ] α . (19)

Combining (18) and (19), we get

1 H ( σ ( t ) , T 0 ) ∫ T 0 σ ( t ) [ H ( σ ( t ) , s ) Q ( s ) − h − α + 1 ( t , s ) [ σ α ( s ) r 1 ( s ) ] α ( α + 1 ) α + 1 [ s α δ ( σ ( s ) ) R ( s , T ) ] α ] Δ s ≤ ω ( T 0 ) .

From (16), we obtain

Ψ ( T 0 ) ≤ ω ( T 0 ) , T 0 ∈ [ T , ∞ ) T ,

lim sup t → ∞ 1 H ( σ ( t ) , T 0 ) ∫ T 0 σ ( t ) H ( σ ( t ) , s ) Q ( s ) Δ s ≥ Ψ ( T 0 ) . (20)

By (18), we get

1 H ( σ ( t ) , T 0 ) ∫ T 0 σ ( t ) H ( σ ( t ) , s ) Q ( s ) Δ s ≤ ω ( T 0 ) + 1 H ( σ ( t ) , T 0 ) ∫ T 0 σ ( t ) h − ( t , s ) H α 1 + α ( σ ( t ) , s ) δ ( s ) ω ( s ) Δ s     − 1 H ( σ ( t ) , T 0 ) ∫ T 0 σ ( t ) α s α H ( σ ( t ) , s ) δ ( σ ( s ) ) R ( s , T ) δ 1 + 1 α ( s ) σ α ( s ) r 1 ( s ) ω 1 + 1 α ( s ) Δ s . (21)

We denote

u ( t ) = 1 H ( σ ( t ) , T 0 ) ∫ T 0 σ ( t ) h − ( t , s ) H α 1 + α ( σ ( t ) , s ) δ ( s ) ω ( s ) Δ s ,

v ( t ) = 1 H ( σ ( t ) , T 0 ) ∫ T 0 σ ( t ) α s α H ( σ ( t ) , s ) δ ( σ ( s ) ) R ( s , T ) δ 1 + 1 α ( s ) σ α ( s ) r 1 ( s ) ω 1 + 1 α ( s ) Δ s ,

meanwhile noting that (16), we obtain

lim inf t → ∞ [ v ( t ) − u ( t ) ] ≤ ω ( T 0 ) − Ψ ( T 0 ) < ∞ .

Now we assert that

∫ T ∞ s α δ ( σ ( s ) ) R ( s , T ) δ 1 + 1 α ( s ) σ α ( s ) r 1 ( s ) ω 1 + 1 α ( s ) Δ s < ∞ (22)

holds. Suppose to the contrary that

∫ T ∞ s α δ ( σ ( s ) ) R ( s , T ) δ 1 + 1 α ( s ) σ α ( s ) r 1 ( s ) ω 1 + 1 α ( s ) Δ s = ∞ , (23)

by (13), there exists a constant ε > 0 such that

inf s ≥ T 0 [ lim inf t → ∞ H ( σ ( t ) , s ) H ( σ ( t ) , T 0 ) ] > ε > 0 , (24)

from (23), there exists T 1 ∈ [ T 0 , ∞ ) T for arbitrary real number M > 0 such that

∫ T 1 σ ( t ) s α δ ( σ ( s ) ) R ( s , T ) δ 1 + 1 α ( s ) σ α ( s ) r 1 ( s ) ω 1 + 1 α ( s ) Δ s ≥ M α ε , for   t ∈ [ T 1 , ∞ ) T .

Using the integration by parts formula of D-differential, we obtain

v ( t ) = 1 H ( σ ( t ) , T 0 ) ∫ T 0 σ ( t ) { α H ( σ ( t ) , s ) ( ∫ T 0 s u α δ ( σ ( u ) ) R ( u , T ) δ 1 + 1 α ( u ) σ α ( u ) r 1 ( u ) W 1 + 1 α ( u ) Δ u ) Δ s } Δ s = 1 H ( σ ( t ) , T 0 ) ∫ T 0 σ ( t ) { − α H Δ s ( σ ( t ) , s ) ∫ T 0 σ ( s ) u α δ ( σ ( u ) ) R ( u , T ) δ 1 + 1 α ( u ) σ α ( u ) r 1 ( u ) ω 1 + 1 α ( u ) Δ u } Δ s ≥ 1 H ( σ ( t ) , T 0 ) ∫ T 1 σ ( t ) { − α H Δ s ( σ ( t ) , s ) ∫ T 0 σ ( s ) u α δ ( σ ( u ) ) R ( u , T ) δ 1 + 1 α ( u ) σ α ( u ) r 1 ( u ) ω 1 + 1 α ( u ) Δ u } Δ s ≥ 1 H ( σ ( t ) , T 0 ) ∫ T 1 σ ( t ) − α H Δ s ( σ ( t ) , s ) M α ε Δ s = M ε H ( σ ( t ) , T 1 ) H ( σ ( t ) , T 0 ) .

From (24), there exists T 2 ∈ [ T 1 , ∞ ) T , we get H ( σ ( t ) , T 1 ) H ( σ ( t ) , T 0 ) ≥ ε for t ∈ [ T 2 , ∞ ) T , so that v ( t ) ≥ M . Since M is arbitrary, we obtain

lim t → ∞ v ( t ) = ∞ . (25)

Selecting a sequence { t n } n = 1 ∞ : t n ∈ [ T 0 , ∞ ) T with lim n → ∞ t n = ∞ satisfying

lim n → ∞ [ v ( t n ) − u ( t n ) ] = lim inf t → ∞ [ v ( t ) − u ( t ) ] < ∞ ,

then there exists a constant M 0 > 0 such that

v ( t n ) − u ( t n ) ≤ M 0 (26)

for sufficiently large positive integer n. From (22), we can easily see

lim n → ∞ v ( t n ) = ∞ , (27)

(26) implies that

lim n → ∞ u ( t n ) = ∞ . (28)

From (26) and (27), we obtain

u ( t n ) v ( t n ) − 1 ≥ − M 0 v ( t n ) > − M 0 2 M 0 = − 1 2 ,

i.e.,

u ( t n ) v ( t n ) > 1 2

for sufficiently large positive integer n, which together with (28) implies

lim n → ∞ [ u ( t n ) ] α + 1 [ v ( t n ) ] α = lim n → ∞ [ u ( t n ) v ( t n ) ] α u ( t n ) = ∞ . (29)

On the other hand, by Lemma 7, we obtain

u ( t n ) = 1 H ( σ ( t n ) , T 0 ) ∫ T 0 σ ( t n ) h − ( t n , s ) H α α + 1 ( σ ( t n ) , s ) δ ( s ) ω ( s ) Δ s = ∫ T 0 σ ( t n ) h − ( t n , s ) H α α + 1 ( σ ( t n ) , s ) H ( σ ( t n ) , T 0 ) δ ( s ) ω ( s ) Δ s = ∫ T 0 σ ( t n ) { [ α s α H ( σ ( t n ) , s ) δ ( σ ( s ) ) R ( s , T ) H ( σ ( t n ) , T 0 ) σ α ( s ) r 1 ( s ) ] α α + 1 ω ( s ) δ ( s ) }     × { h − ( t n , s ) H α α + 1 ( σ ( t n ) , s ) H ( σ ( t n ) , T 0 ) [ α s α H ( σ ( t n ) , s ) δ ( σ ( s ) ) R ( s , T ) H ( σ ( t n ) , T 0 ) σ α ( s ) r 1 ( s ) ] − α α + 1 } Δ s

≤ { ∫ T 0 σ ( t n ) α s α H ( σ ( t n ) , s ) δ ( σ ( s ) ) R ( s , T ) H ( σ ( t n ) , T 0 ) σ α ( s ) r 1 ( s ) [ ω ( s ) δ ( s ) ] α + 1 α Δ s } α α + 1     × { ∫ T 0 σ ( t n ) h − ( t n , s ) H α α + 1 ( σ ( t n ) , s ) H α + 1 ( σ ( t n ) , T 0 ) [ α s α H ( σ ( t n ) , s ) δ ( σ ( s ) ) R ( s , T ) H ( σ ( t n ) , T 0 ) σ α ( s ) r 1 ( s ) ] − α Δ s } 1 α + 1 = [ v ( t n ) ] α α + 1 { 1 α α H ( σ ( t n ) , T 0 ) ∫ T 0 σ ( t n ) h − α + 1 ( t n , s ) σ α 2 ( s ) r 1 α ( s ) [ s α δ ( σ ( s ) ) R ( s , T ) ] α Δ s } 1 α + 1 .

The above inequality show that

[ u ( t n ) ] α + 1 [ v ( t n ) ] α ≤ 1 α α H ( σ ( t n ) , T 0 ) ∫ T 0 σ ( t n ) h − α + 1 ( t n , s ) σ α 2 ( s ) r 1 α ( s ) [ s α δ ( σ ( s ) ) R ( s , T ) ] α Δ s .

Hence, (29) implies

lim n → ∞ 1 α α H ( σ ( t n ) , T 0 ) ∫ T 0 σ ( t n ) h − α + 1 ( t n , s ) σ α 2 ( s ) r 1 α ( s ) [ s α δ ( σ ( s ) ) R ( s , T ) ] α Δ s = ∞ .

This contradicts (14). Therefore (22) holds. Noting Ψ ( T 0 ) ≤ ω ( T 0 ) for T 0 ∈ [ T , ∞ ) T , by using (22), we obtain

∫ T ∞ s α δ ( σ ( s ) ) R ( s , T ) δ 1 + 1 α ( s ) σ α ( s ) r 1 ( s ) Ψ + 1 + 1 α ( s ) Δ s ≤ ∫ T ∞ s α δ ( σ ( s ) ) R ( s , T ) δ 1 + 1 α ( s ) σ α ( s ) r 1 ( s ) ω 1 + 1 α ( s ) Δ s < ∞ .

This contradicts (15).

If case (II) holds, from (9), by Lemma 6 lim t → ∞ x ( t ) = 0 . The proof is complete.

Theorem 2 Assume that (8), (9), (12), (13), (15) and r 1 Δ ( t ) ≤ 0 hold, where H , h and δ are defined in Theorem 1. Furthermore, assume that there is a real rd-continuous function Ψ : [ t 0 , ∞ ) T → ℝ such that

lim inf t → ∞ 1 H ( σ ( t ) , T 0 ) ∫ T 0 σ ( t ) H ( σ ( t ) , s ) Q ( s ) Δ s < ∞ , (30)

lim inf t → ∞ 1 H ( σ ( t ) , T 0 ) ∫ T 0 σ ( t ) [ H ( σ ( t ) , s ) Q ( s ) − h − α + 1 ( t , s ) [ σ α ( s ) r 1 ( s ) ] α ( α + 1 ) α + 1 [ s α δ ( σ ( s ) ) R ( s , T ) ] α Δ s ≥ Ψ ( T 0 ) , (31)

for T 0 ∈ [ T , ∞ ) T , where Q ( t ) , Ψ ( t ) are defined in Theorem 1. Then every solution x ( t ) of Equation (1) is either oscillatory or converges to zero.

Proof Assume that (1) has a nonoscillatory solution x ( t ) on [ t 0 , ∞ ) T . Without loss generality we may assume that there exists sufficiently large T ≥ t 0 such that x ( t ) > 0 and x [ τ ( t ) ] > 0 for all t ∈ [ T , ∞ ) T . By Lemma 1, we see that x ( t ) satisfies either case (I) or case (II).

If case (I) holds, proceeding as in the proof of Theorem 1, we get

1 H ( σ ( t ) , T 0 ) ∫ T 0 σ ( t ) [ H ( σ ( t ) , s ) Q ( s ) − h − α + 1 ( t , s ) [ σ α ( s ) r 1 ( s ) ] α ( α + 1 ) α + 1 [ s α δ ( σ ( s ) ) R ( s , T ) ] α ] Δ s ≤ ω ( T 0 ) ,

From (31), we obtain

Ψ ( T 0 ) ≤ ω ( T 0 ) , T 0 ∈ [ T , ∞ ) T ;

lim inf t → ∞ 1 H ( σ ( t ) , T 0 ) ∫ T 0 σ ( t ) H ( σ ( t ) , s ) Q ( s ) Δ s ≥ Ψ ( T 0 ) ; (32)

and

lim inf t → ∞ 1 H ( σ ( t ) , T 0 ) ∫ T 0 σ ( t ) H ( σ ( t ) , s ) Q ( s ) Δ s − lim inf t → ∞ 1 H ( σ ( t ) , T 0 ) ∫ T 0 σ ( t ) h − α + 1 ( t , s ) [ σ α ( s ) r 1 ( s ) ] α ( α + 1 ) α + 1 [ s α δ ( σ ( s ) ) R ( s , T ) ] α Δ s ≥ lim inf t → ∞ 1 H ( σ ( t ) , T 0 ) ∫ T 0 σ ( t ) [ H ( σ ( t ) , s ) Q ( s )     − h − α + 1 ( t , s ) [ σ α ( s ) r 1 ( s ) ] α ( α + 1 ) α + 1 [ s α δ ( σ ( s ) ) R ( s , T ) ] α ] Δ s ≥ Ψ ( T 0 ) . (33)

Using (30) and (33), we get

lim inf t → ∞ 1 H ( σ ( t ) , T 0 ) ∫ T 0 σ ( t ) h − α + 1 ( t , s ) [ σ α ( s ) r 1 ( s ) ] α ( α + 1 ) α + 1 [ s α δ ( σ ( s ) ) R ( s , T ) ] α Δ s < ∞ .

Thus, there exists a sequence { t n } n = 1 ∞ : t n ∈ [ T 0 , ∞ ) T with lim n → ∞ t n = ∞ such that

lim n → ∞ 1 H ( σ ( t n ) , T 0 ) ∫ T 0 σ ( t n ) h − α + 1 ( t , s ) [ σ α ( s ) r 1 ( s ) ] α ( α + 1 ) α + 1 [ s α δ ( σ ( s ) ) R ( s , T ) ] α Δ s < ∞ .

We define u ( t ) and v ( t ) also, as in the proof of Theorem 1. From (18) and (32), we obtain

lim sup t → ∞ [ v ( t ) − u ( t ) ] ≤ ω ( T 0 ) − lim inf t → ∞ 1 H ( σ ( t ) , T 0 ) ∫ T 0 σ ( t ) H ( σ ( t ) , s ) Q ( s ) Δ s < ∞ .

For the above sequence { t n } n = 1 ∞ , we get

lim n → ∞ [ v ( t n ) − u ( t n ) ] ≤ lim sup t → ∞ [ v ( t ) − u ( t ) ] < ∞ .

Similar to the proof of Theorem 1, we get (22). The rest of the proof is similar to that of Theorem 1, and hence is omitted. The proof is complete.

Remark 1 From Theorems 1 and 2, we can obtain different sufficient conditions for the oscillation of Equation (1) with different choices of the functions

δ and H . For example, H ( t , s ) = ( t − s ) m or H ( t , s ) = ( ln t + 1 s + 1 ) m .

Remark 2 The theorems in this paper are new even for the cases of T = ℝ and T = ℤ .

Example 1 Consider the third-order nonlinear delay dynamic equation

( t ( ( 1 t 2 x Δ ( t ) ) Δ ) 5 3 ) Δ + 1 t 8 3 ( x ( t 2 ) ) 5 3 ( 1 + ln ( 1 + x 2 ( t 2 ) ) ) = 0, t ∈ 2 Z ¯ , t ≥ t 0 : = 2. (34)

Here α = 5 3 , r 1 ( t ) = 1 t 2 , r 2 ( t ) = t , q ( t ) = 1 t 8 3 , f ( x ) = x 5 3 ( 1 + ln ( 1 + x 2 ) ) and τ ( t ) = t 2 < t .

The conditions (H₁)-(H₃) are clearly satisfied, (H₄) holds with K = 1 . f ( x ) x α = 1 + ln ( 1 + x 2 ) ≥ K > 0 . r 1 Δ ( t ) = − 3 4 t 3 < 0 , and

∫ t 0 ∞ q ( s ) τ α ( s ) Δ s = ∫ 2 ∞ 1 s 8 3 ( s 2 ) 5 3 Δ s = 2 − 5 3 ∫ 2 ∞ s − 1 Δ s = ∞ ,

∫ t 0 ∞ 1 r 1 ( t ) ∫ t ∞ [ 1 r 2 ( s ) ∫ s ∞ q ( u ) Δ u ] 1 α Δ s Δ t = ∫ 2 ∞ t 2 ∫ t ∞ [ 1 s ∫ s ∞ 1 u 8 3 Δ u ] 3 5 Δ s Δ t = ( 1 1 − 2 − 5 3 ) 3 5 ∫ 2 ∞ t 2 ∫ t ∞ 1 s 8 5 Δ s Δ t = ( 1 1 − 2 − 5 3 ) 3 5 1 1 − 2 − 3 5 ∫ 2 ∞ t 7 5 Δ t = ∞ ,

so (8), (9) hold. For larger enough t > T , we have

R ( t , T ) = ∫ T t ( 1 r 2 ( s ) ) 1 α Δ s = ∫ T t     s − 3 5 Δ s = t 2 5 − T 2 5 2 2 5 − 1 > 1.

Let δ ( t ) = t , since σ ( t ) = 2 t , we have

Q ( t ) = K q ( t ) δ ( σ ( t ) ) ( τ ( t ) σ ( t ) ) α = 1 2 7 3 t 5 3 .

Let H ( t , s ) = ( t − s ) 2 , that there exists a function h ( t , s ) = − 4 ( t − s ) 2 ( 2 t − s ) 5 4 such that

H Δ s ( σ ( t ) , s ) + ( δ Δ ( s ) ) + δ ( s ) H ( σ ( t ) , s ) = − h ( t , s ) δ ( s ) H α α + 1 ( σ ( t ) , s ) .

It follows that

0 < inf s ≥ T 0 [ lim inf t → ∞ H ( σ ( t ) , s ) H ( σ ( t ) , T 0 ) ] = inf s ≥ T 0 [ lim inf t → ∞ ( 2 t − s ) 2 ( 2 t − T 0 ) 2 ] = 1 < ∞ ,

lim sup t → ∞ 1 H ( σ ( t ) , T 0 ) ∫ T 0 σ ( t ) h − α + 1 ( t , s ) [ σ α ( s ) r 1 ( s ) ] α [ s α δ ( σ ( s ) ) R ( s , T ) ] α Δ s ≤ 2 10 9 ⋅ T 0 − 4 1 − 2 − 4 < 2 6 T 0 4 < ∞ ,

so (12), (13) and (14) hold. Let Ψ ( t ) = 1 2 t , we have

∫ T ∞ s α δ ( σ ( s ) ) R ( s , T ) δ 1 + 1 α ( s ) σ α ( s ) r 1 ( s ) Ψ + 1 + 1 α ( s ) Δ s = 1 2 49 15 ⋅ ( 2 2 5 − 1 ) ⋅ ∫ T ∞ ( s 1 5 − s − 1 5 ⋅ T 2 5 ) Δ s = ∞ ,

and

lim sup t → ∞ 1 H ( σ ( t ) , T 0 ) ∫ T 0 σ ( t ) [ H ( σ ( t ) , s ) Q ( s ) − h − α + 1 ( t , s ) [ σ α ( s ) r 1 ( s ) ] α ( α + 1 ) α + 1 [ s α δ ( σ ( s ) ) R ( s , T ) ] α ] Δ s ≥ 2 − 7 3 ( 1 − 2 − 2 3 ) T 0 2 3 − 2 10 9 ( 8 3 ) 8 3 ( 1 − 2 − 4 ) T 0 4 > 1 2 T 0 2 3 > 1 2 T 0 = Ψ ( T 0 ) .

Then, by Theorem 1, every solution x ( t ) of Equation (34) is either oscillatory or converges to zero. But the results in [18] cannot be applied in (34).

Acknowledgements

The authors are grateful to the reviewers for their comments and suggestions. This research is supported by Shandong Provincial Natural Science Foundation (ZR2013AM003), Technology Planning Program of Shandong Province (J14LI54).

References

1. Hilger, S. (1990) Analysis on Measure Chains—A Unified Approach to Continuous and Discrete Calculus. Results in Mathematics, 18, 18-56. https://doi.org/10.1007/BF03323153

2. Bohner, M. and Saker, S.H. (2004) Oscillation of Second Order Nonlinear Dynamic Equations on Time Scales. Rocky Mountain Journal of Mathematics, 34, 1239-1254. https://doi.org/10.1216/rmjm/1181069797

3. Erbe L. ,et al. (2001)Oscillation Criteria for Second Order Linear Equations on a Time Scale Canadian Applied Mathematics Quarterly 9, 345-375.

4. Agarwal, R.P., Bohner, M. and Saker, S.H. (2005) Oscillation of Second Order Delay Dynamic Equations. Canadian Applied Mathematics Quarterly, 13, 1-18.

5. Erbe, L., Peterson, A. and Saker, S.H. (2007) Oscillation Criteria for Second Order Nonlinear Delay Dynamic Equations. Journal of Mathematical Analysis and Applications, 333, 505-522. https://doi.org/10.1016/j.jmaa.2006.10.055

6. Crace, S.R., Bohner, M. and Agarwal, R.P. (2009) On the Oscillation of Second-Order Half-Linear Dynamic Equations. Journal of Difference Equations and Applications, 15, 451-460. https://doi.org/10.1080/10236190802125371

7. Crace, S.R., Agarwal, R.P., Kaymakcalan, B. and Sae-jie, W. (2010) Oscillation Theorems for Second Order Nonlinear Dynamic Equations. Journal of Applied Mathematics and Computing, 32, 205-218. https://doi.org/10.1007/s12190-009-0244-7

8. Sun, S., Han, Z. and Zhang, C. (2009) Oscillation of Second Order Delay Dynamic Equations on Time Scales. Journal of Applied Mathematics and Computing, 30, 459-468. https://doi.org/10.1007/s12190-008-0185-6

9. Zhang, Q. and Gao, L. (2010) Oscillation Criteria for Second Order Half-Liner Delay Dynamic Equations with Damping on Time Scales. Scientia Sinica Mathematica, 40, 673-682. (In Chinese)

10. Zhang, Q., Gao, L. and Liu, S. (2011) Oscillation Criteria for Second Order Half-Liner Delay Dynamic Equations with Damping on Time Scales . Scientia Sinica Mathematica, 41, 885-896. (In Chinese)

11. Erbe, L., Peterson, A. and Saker, S.H. (2005) Asymptotic Behavior of Solution of a Third-Order Nonlinear Dynamic Equations. Journal of Computational and Applied Mathematics, 181, 92-102. https://doi.org/10.1016/j.cam.2004.11.021

12. Erbe, L., Peterson, A. and Saker, S.H. (2007) Hille and Nehari Type Criteria for Third Order Nonlinear Dynamic Equations. Journal of Mathematical Analysis and Applications, 329, 112-131. https://doi.org/10.1016/j.jmaa.2006.06.033

13. Hassan, T.S. (2009) Oscillation of Third Order Nonlinear Delay Dynamic Equations on Time Scales. Mathematical and Computer Modelling, 49, 1573-1586. https://doi.org/10.1016/j.mcm.2008.12.011

14. Li, T., Han, Z., Sun, S. and Zhao, Y. (2011) Oscillation Results for Third Order Nonlinear Delay Dynamic Equations on Time Scales. Bulletin of the Malaysian Mathematical Sciences Society, 34, 639-648.

15. Agarwal, R.P., Bohner, M., Tang, S., Li, T. and Zhang, C. (2012) Oscillation and Asymptotic Behavior of Third-Order Nonlinear Retarded Dynamic Equations. Applied Mathematics and Computation, 219, 3600-3609. https://doi.org/10.1016/j.amc.2012.09.054

16. Han, Z., Li, T., Sun, S. and Zhang, M. (2011) Oscillation Behavior of Solution of Third-Order Nonlinear Delay Dynamic Equations on Time Scales. Communications of the Korean Mathematical Society, 26, 499-533. https://doi.org/10.4134/CKMS.2011.26.3.499

17. Li, T., Han, Z., Sun, S. and Zhao, Y. (2011) Asymptotic Behavior of Solutions for Third-Order Half-Linear Delay Dynamic Equations on Time Scales. Journal of Applied Mathematics and Computing, 36, 333-346. https://doi.org/10.1007/s12190-010-0406-7

18. Gao, L., Zhang, Q.X. and Liu, S.H. (2014) Oscillatory Behavior of Third-Order Nonlinear Delay Dynamic Equations on Time Scales. Journal of Computational and Applied Mathematics, 256, 104-113. https://doi.org/10.1016/j.cam.2013.07.024

19. Bohner, M. and Peterson, A. (2001) Dynamic Equations on Time Scales: An Introduction with Applications. Birkhauser, Boston. https://doi.org/10.1007/978-1-4612-0201-1

20. Agarwal, R.P., Bohner, M., Crace, S.R. and O’Regan, D. (2005) Discrete Oscillation Theory. Hindawi Publishing Corporation, New York.

21. Hardy, G.H., Littlewood, J.E. and Polya, G. (1988) Inequalities. 2nd Edition, Cambridge University Press, Cambridge.

22. Tuna, A. and Kutukcu, S. (2008) Some Integral Inequalities on Time Scales. Applied Mathematics and Mechanics, 29, 23-29. https://doi.org/10.1007/s10483-008-0104-y