Math. Nachr. 288, No. 10, 1150–1162 (2015) / DOI 10.1002/mana.

201300029

Oscillation of second-order neutral differential equations

Tongxing Li∗1,2 and Yuriy V. Rogovchenko∗∗3
1
LinDa Institute of Shandong Provincial Key Laboratory of Network Based Intelligent Computing, Linyi
University, Linyi, Shandong, 276005, P. R. China
2
School of Informatics, Linyi University, Linyi, Shandong, 276005, P. R. China
3
Department of Mathematical Sciences, University of Agder, PO Box 422, N-4604, Kristiansand, Norway

Received 2 February 2013, revised 2 October 2014, accepted 21 October 2014

Published online 20 February 2015

Key words Oscillation, neutral differential equations, delayed argument, advanced argument
MSC (2010) 34K11
We study oscillatory behavior of a class of second-order neutral differential equations under the assumptions
that allow applications to differential equations with both delayed and advanced arguments, and not only. New
theorems complement and improve a number of results reported in the literature. Illustrative examples are
provided.


C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1 Introduction
This paper is concerned with the oscillatory behavior of solutions to a nonlinear second-order neutral differential
equation
 γ 
a(t) [z  (t)] + q(t)x β (σ (t)) = 0, (1.1)
where t ∈ I := [t0 , ∞), t0 ∈ R, z(t) := x(t) + p(t)x(τ (t)), the functions a, p, q, τ , σ ∈ C(I) and a, p, q take on
positive values. We also assume that β, γ ∈ R, where R is the set containing all ratios of odd natural numbers.
The analysis of qualitative properties of Eq. (1.1) is important for applications since, for instance, its particular
case, an Emden-Fowler type equation
 γ 
a(t) [x  (t)] + q(t)x β (σ (t)) = 0,
has numerous applications in mathematical and theoretical physics; see [21].
By a solution of Eq. (1.1) we mean a function x ∈ C([Tx , ∞)), Tx ≥ t0 , such that a (z  )γ ∈ C1 ([Tx , ∞))
and x(t) satisfies (1.1) on [Tx , ∞). We consider only those solutions x(t) of (1.1) that satisfy the condition
sup{|x(t)| : t ≥ T } > 0, for all T ≥ Tx , and we tacitly assume that (1.1) possesses such solutions. As usual, a
solution x(t) of (1.1) is called oscillatory if it has arbitrarily large zeros on [Tx , ∞); otherwise, it is termed
non-oscillatory. Eq. (1.1) is called oscillatory if all its solutions are oscillatory.
During the last three decades, conditions for oscillatory or non-oscillatory behavior of various classes of
differential equations have been frequently discussed in the literature; see, e.g., the papers [1]–[25] and the
references cited therein. In what follows, we briefly comment on related results that motivated our study. In one of
the first important contributions to the field, Grammatikopoulos et al. [8] proved that a linear second-order neutral
differential equation
(x(t) + p(t)x(t − τ )) + q(t)x(t − σ ) = 0 (1.2)
is oscillatory provided that 0 ≤ p(t) ≤ 1 and
 ∞
q(s)(1 − p(s − σ )) ds = ∞.
t0

∗ e-mail: litongx2007@163.com, Phone: +86 13869959692, Fax: +86 539 5797055

∗∗ Corresponding author: e-mail: yuriy.rogovchenko@uia.no, Phone: +47 3814 2245, Fax: +47 3814 1001


C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Math. Nachr. 288, No. 10 (2015) / www.mn-journal.com 1151

Oscillation criteria for Eq. (1.2) which require that q(t) ≥ q0 > 0, p1 ≤ p(t) ≤ p2 , and p(t) is not eventually
negative can be found in the monograph by Erbe et al. [7]. Later on, Xu and Xia [24] proved that the conditions
0 ≤ p(t) < ∞ and q(t) ≥ M > 0 ensure oscillation of (1.2). Recently, Baculı́ková and Džurina [3], Han et al.
[10], and Li et al. [17] studied asymptotic behavior of solutions to a neutral differential equation
(r (t)[x(t) + p(t)x(τ (t))] ) + q(t)x(σ (t)) = 0
under the assumptions that 0 ≤ p(t) ≤ p0 < ∞ and τ ◦ σ = σ ◦ τ ; further extensions of these results can be
found in another paper by Baculı́ková and Džurina [4].
For γ ≥ 1 and 0 ≤ p(t) ≤ p0 < ∞, Baculı́ková et al. [5] established oscillation results for a general class of
even-order neutral differential equations which also includes a second-order differential equation
 γ 
r (t) ([x(t) + p(t)x(τ (t))] ) + q(t)x γ (σ (t)) = 0 (1.3)
assuming that σ (t) ≤ τ (t) ≤ t, whereas Li et al. [16] deduced oscillation of Eq. (1.3) under the assumptions
that τ ◦ σ = σ ◦ τ, τ (t) ≥ t, and σ (t) ≥ t. In the case where 0 ≤ p(t) < 1, Eq. (1.3), its particular cases and
modifications have been studied by Dong [6], Hasanbulli and Rogovchenko [11], [12], Liu and Bai [18], and Xu
and Meng [22], [23]. For p(t) > 1, Li et al. [15] explored oscillatory properties of Eq. (1.3) assuming that either
τ (t) > t or τ (t) < t. One should also note that results reported by Baculı́ková et al. [5] and Li et al. [15], [16] do
not apply to Eq. (1.1) when γ
= β.
Very recently, Zhong et al. [25] established oscillation results for Eq. (1.1) assuming that a  (t) ≥ 0, p(t) =
p0
= 1, τ (t) = t − τ0 , τ0 ≥ 0, and σ (t) ≤ t − τ0 , although their results fail to apply to Eq. (1.1) for σ (t) ≥ t.
The problem of oscillation of Eq. (1.1) in the case where σ (t) ≤ t has been also discussed by Baculı́ková and
Džurina [4]. However, the latter study left without answers questions regarding oscillation of (1.1) in the following
two cases: (i) β ≥ γ ; (ii) τ (t) ≤ σ (t) ≤ t.
This study was strongly motivated by the research of Baculı́ková and Džurina [4] and Zhong et al. [25]. Its
purpose is not only to analyze oscillation of Eq. (1.1) in two cases mentioned above, but to derive also new
oscillation criteria for (1.1) without imposing the assumption σ (t) ≤ t. Main results in this paper are organized
into three parts in accordance with different assumptions on the coefficient p(t). In Section 2, new oscillation
results for Eq. (1.1) are established in the case where p(t) is bounded, 0 ≤ p(t) ≤ p0 < ∞. In Section 3,
different methods are used to derive new oscillation criteria for (1.1) in a very important particular case when
p(t) is constant, 0 ≤ p(t) = p0
= 1. Finally, in Section 4, we present oscillation theorems for (1.1) for p(t) ≥ 1
assuming that p(t) may be even unbounded.
As usual, all functional inequalities considered in this paper are supposed to be satisfied for all t large enough.
Without loss of generality, we deal only with positive solutions of Eq. (1.1) since under our assumptions −x(t) is
obviously a solution of this equation provided that x(t) is a solution.

2 Oscillation results for bounded p(t)

In this section, τ (t) can be both a delayed or an advanced argument. Throughout this section, in addition to the
main assumptions formulated in the introduction, it is also supposed that
(H1 ) 0 ≤ p(t) ≤ p0 < ∞, for some constant p0 ;
(H2 ) τ ∈ C1 (I), τ  (t) ≥ τ∗ > 0, τ ◦ σ = σ ◦ τ , and there exists a function η ∈ C(I) such that η(t) ≤ σ (t)
and
 ∞ lim t→∞ η(t) = ∞;
−1/γ
(H3 ) t0 a (s) ds = ∞.

In the sequel, we use the notation τ −1 for the function which is inverse to τ ; Q(t) := min{q(t), q(τ (t))},
 ∞  t
A(t) := a −1/γ (s) ds, B(t) := a −1/γ (s) ds,
t t1
α
Q α (t) := Q(t)[B(η(t))] , and Q β (t) := Q(t)[B(η(t))]β ,
where t1 is large enough and the choice of α is to be specified later.
We need the following auxiliary result due to Baculı́ková and Džurina [4, Lemma 3].

www.mn-journal.com 
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1152 T. Li and Yu. V. Rogovchenko: Oscillation of second-order neutral differential equations


Lemma 2.1 Assume that x(t) is a positive solution of (1.1). Then, z(t) > 0, z  (t) > 0, and (a(t) [z  (t)]γ ) < 0,
for all t large enough.
Now we state and prove four theorems that ensure oscillation of Eq. (1.1).
Theorem 2.2 Let 0 < β ≤ 1 and η(t) < t ≤ τ (t). Suppose that conditions (H1 )−(H3 ) are satisfied, and there
exists a number α ∈ R such that α ≤ β and α < γ . If
 ∞
Q α (t) dt = ∞, (2.1)

Eq. (1.1) is oscillatory.

P r o o f . Assume that Eq. (1.1) is not oscillatory. In this case, it has solutions that eventually do not change
sign. Without loss of generality, we can suppose that x(t) is a positive solution of (1.1) on I. As in the proof of [4,
Theorem 1], one arrives at the inequality
 
γ p0 β γ
a(t) [z  (t)] + a(τ (t)) [z  (τ (t))] + Q(t)z β (σ (t)) ≤ 0, (2.2)
τ∗
which is valid for all t ≥ t∗ and for some t∗ ≥ t0 . In view of Lemma 2.1, we know that z  (t) > 0. Hence, by (2.2)
and the condition η(t) ≤ σ (t), one has
 
γ p0 β γ
a(t) [z  (t)] + a(τ (t)) [z  (τ (t))] + Q(t)z β (η(t)) ≤ 0. (2.3)
τ∗
Furthermore, due to monotonicity of z(t), there exists a constant M > 0 such that
z β (η(t)) = z β−α (η(t))z α (η(t)) ≥ M β−α z α (η(t)). (2.4)
Combining (2.3) and (2.4), one obtains
 
 γ p0 β  γ
a(t) [z (t)] + a(τ (t)) [z (τ (t))] + M β−α Q(t)z α (η(t)) ≤ 0. (2.5)
τ∗
An application of Lemma 2.1 yields that the function
γ
ω(t) := a(t) [z  (t)] (2.6)
is positive and decreasing. Therefore,
z(t) ≥ a 1/γ (t)z  (t)B(t) = ω1/γ (t)B(t). (2.7)
Using (2.7) in (2.5), we conclude that ω(t) is a positive solution of a delay differential inequality
 
p0 β
ω(t) + ω(τ (t)) + M β−α Q α (t)ωα/γ (η(t)) ≤ 0. (2.8)
τ∗
Define now a new function y(t) by
p0 β
y(t) := ω(t) + ω(τ (t)). (2.9)
τ∗
Then, by monotonicity of ω(t),
 
p0 β
y(t) ≤ ω(t) 1 + . (2.10)
τ∗
Substituting (2.10) into (2.8), we observe that y(t) is a positive solution of a delay differential inequality
 α/γ
 β−α τ∗
y (t) + M Q α (t)y α/γ (η(t)) ≤ 0.
τ∗ + p0 β


C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mn-journal.com
Math. Nachr. 288, No. 10 (2015) / www.mn-journal.com 1153

Then, by virtue of [19, Theorem 1], the associated delay differential equation
 α/γ
τ∗
y  (t) + M β−α Q α (t)y α/γ (η(t)) = 0 (2.11)
τ∗ + p0 β
also has a positive solution. However, [13, Theorem 2] implies that under the assumption (2.1) Eq. (2.11) is
oscillatory. Therefore, Eq. (1.1) cannot have positive solutions. This contradiction with our initial assumption
completes the proof. 
Theorem 2.3 Suppose that 0 < β = γ ≤ 1 and η(t) ≤ t ≤ τ (t). If conditions (H1 )−(H3 ) are satisfied and
 t
τ∗ 1
lim inf Q β (s) ds > , (2.12)
τ∗ + p0 β t→∞ η(t) e
Eq. (1.1) is oscillatory.
P r o o f . Assume that Eq. (1.1) is non-oscillatory, and let x(t) be a positive solution defined on I. Proceeding
as in the proof of Theorem 2.2, one comes to the conclusion that a delay differential equation
τ∗
y  (t) + Q β (t)y(η(t)) = 0 (2.13)
τ∗ + p0 β
has a positive solution y(t). On the other hand, application of condition (2.12) along with [4, Lemma 4] implies
that Eq. (2.13) is oscillatory, a contradiction. The proof is complete. 
In order to illustrate efficiency of our theorems, we recall a recent result due to Baculı́ková and Džurina [4,
Corollary 1] that reads as follows.
Corollary 2.4 Suppose that 0 < β ≤ 1, β ≤ γ , and σ (t) ≤ t ≤ τ (t). Assume also that conditions (H1 )−(H3 )
are satisfied and
 β/γ  t  β
σ (s)
τ∗ −1/γ 1
lim inf Q(s) a (v) dv ds > .
τ∗ + p0 β t→∞ σ (t) t1 e
Then Eq. (1.1) is oscillatory.
Remark 2.5 Note that for β = γ , Theorem 2.3 extends [4, Corollary 1] because in our result the assumption
σ (t) ≤ t is no longer required.
Example 2.6 Consider a second-order nonlinear neutral differential equation
b
(x(t) + 2x(θ t)) + x 1/3 (λt) = 0, (2.14)
t 4/3
where θ ≥ 1, b > 0, and λ ∈ (0, ∞). An application of Corollary 2.4 yields oscillation of Eq. (2.14) for any
λ ∈ (0, 1) provided that
θ −1 1 1
bλ1/3 ln > .
(θ + 21/3 )1/3 λ e
However, choosing in Theorem 2.2 γ = 1, α = β = 1/3, λ1 ∈ (0, 1), η(t) = λ1 t ≤ λt, one concludes that
Eq. (2.14) is oscillatory for all λ ∈ (0, ∞) and b > 0, which demonstrates superiority of Theorem 2.2 over
Corollary 2.4.
Theorem 2.7 Let 0 < β ≤ 1 and η(t) < τ (t) ≤ t. Assume also that there exists a number α ∈ R such that
α ≤ β and α < γ . If conditions (H1 )−(H3 ) and (2.1) are satisfied, Eq. (1.1) is oscillatory.
P r o o f . As above, let x(t) be a positive solution of Eq. (1.1). It has been established in the proof of Theorem
2.2 that the function ω(t) defined by (2.6) is positive, decreasing, and satisfies the inequality (2.8). Introducing
again y(t) by (2.9) and using the monotonicity of ω(t), we conclude that
 
p0 β
y(t) ≤ ω(τ (t)) 1 + . (2.15)
τ∗

www.mn-journal.com 
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1154 T. Li and Yu. V. Rogovchenko: Oscillation of second-order neutral differential equations

Substitution of (2.15) into (2.8) implies that, for sufficiently large t, y(t) is a positive solution of a delay differential
inequality
 α/γ
 β−α τ∗  
y (t) + M Q α (t)y α/γ τ −1 (η(t)) ≤ 0. (2.16)
τ∗ + p0 β
By virtue of [19, Theorem 1], a delay differential equation
 α/γ
 β−α τ∗  
y (t) + M Q α (t)y α/γ τ −1 (η(t)) = 0 (2.17)
τ∗ + p0 β
associated with (2.16) also has a positive solution. Using condition (2.1) in [13, Theorem 2], one concludes that
Eq. (2.17) is oscillatory, which contradicts the fact that y(t) is a positive solution of (2.17). The proof is complete
now. 
Theorem 2.8 Suppose that 0 < β = γ ≤ 1 and η(t) ≤ τ (t) ≤ t. If conditions (H1 )−(H3 ) are satisfied and
 t
τ∗ 1
lim inf Q β (s) ds > , (2.18)
τ∗ + p0 β t→∞ τ −1 (η(t)) e

Eq. (1.1) is oscillatory.

P r o o f . Following the same lines as in the proof of Theorem 2.7, we observe that a delay differential equation

τ∗  
y  (t) + Q β (t)y τ −1 (η(t)) = 0 (2.19)
τ∗ + p0 β

has an eventually positive solution y(t). On the other hand, using condition (2.18) in [4, Lemma 4], one deduces
that Eq. (2.19) is oscillatory. This contradiction completes the proof. 
Remark 2.9 In this section, assuming as Baculı́ková and Džurina [4] that τ ◦ σ = σ ◦ τ and using a modified
technique, we established new oscillation criteria that provide answers to two open problems formulated in the
cited paper. We also improved some of results reported there, cf. Example 2.6. Note that oscillation theorems can
be also obtained for β ≥ 1; in this case, one simply has to replace Q(t) := min{q(t), q(τ (t))} in [4, Lemma 1]
with a function Q(t) := 21−β min{q(t), q(τ (t))} and proceed as above.

3 Oscillation results for constant p(t)

Throughout this section, in addition to condition (H3 ), we also assume that

(H4 ) a  (t) ≥ 0 for t ≥ t0 , p(t) = p0

= 1;
(H5 ) τ (t) = t − τ0 , τ0 ∈ R, and there exist a function η ∈ C(I) and a constant δ0 ≥ 0 such that η(t) ≤
t − δ0 ≤ σ (t) and limt→∞ η(t) = ∞.

In what follows, we use the notation R(t) := min{q(t), q(t − τ0 )}.

Theorem 3.1 Let 0 < β ≤ 1, η(t) < t, and τ0 ≤ 0. Assume that there exists a number α ∈ R such that α ≤ β
and α < γ . If conditions (H3 )–(H5 ) are satisfied and, for some sufficiently large t1 ≥ t0 ,
 ∞
R(t)[B(η(t))]α dt = ∞, (3.1)

Eq. (1.1) is oscillatory.

P r o o f . Assume the opposite. Let x(t) be a positive solution of (1.1). Using Lemma 2.1 and monotonicity
of a(t), we conclude that z  (t) < 0 eventually. Therefore, there exists an l ≥ 0 such that limt→∞ z  (t) = l ≥ 0,


C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mn-journal.com
Math. Nachr. 288, No. 10 (2015) / www.mn-journal.com 1155

in which case, by virtue of [9], one has

l
lim x  (t) = ≥ 0.
t→∞ 1 + p0
Consider two cases.
Case I. First, assume that l > 0. Then x  (t) > 0 eventually, and, using the assumption (H5 ), we deduce that
x(σ (t)) ≥ x(t − δ0 ). Hence, Eq. (1.1) yields that
 γ 
a(t) [z  (t)] + q(t)x β (t − δ0 ) ≤ 0. (3.2)
In view of (3.2), one has
 γ 
p0 β a(t − τ0 ) [z  (t − τ0 )] + p0 β q(t − τ0 )x β (t − τ0 − δ0 ) ≤ 0. (3.3)
Combining (3.2) and (3.3), we arrive at
 γ γ 

a(t) [z  (t)] + p0 β a(t − τ0 ) [z  (t − τ0 )] + R(t) x β (t − δ0 ) + p0 β x β (t − τ0 − δ0 ) ≤ 0.
Using the inequality (X + Y )l ≤ X l + Y l which holds for all l ∈ [0, 1] and for any X, Y ≥ 0, cf. [4, Lemma 2],
one deduces that
 γ γ 
a(t) [z  (t)] + p0 β a(t − τ0 ) [z  (t − τ0 )] + R(t)z β (t − δ0 ) ≤ 0. (3.4)
Since Lemma 2.1 yields that z(t) is strictly increasing, using (2.4), (3.4) and assumption (H5 ), we arrive at the
inequality
 γ γ 
a(t) [z  (t)] + p0 β a(t − τ0 ) [z  (t − τ0 )] + M β−α R(t)z α (η(t)) ≤ 0. (3.5)
Taking into account that z(t) > 0 and the function ω(t) defined by (2.6) is positive and decreasing, as in Theorem
2.2 we conclude that (2.7) is satisfied. Substitution of (2.7) into (3.5) yields that ω(t) is a solution of a delay
differential inequality
 
ω(t) + p0 β ω(t − τ0 ) + M β−α R(t)[B(η(t))]α ωα/γ (η(t)) ≤ 0. (3.6)
Define now
u(t) := ω(t) + p0 β ω(t − τ0 ). (3.7)
Then
 
u(t) ≤ ω(t) 1 + p0 β . (3.8)
Substituting (3.8) into (3.6), we conclude that u(t) is a positive solution of a delay differential inequality
M β−α
u  (t) + R(t)[B(η(t))]α u α/γ (η(t)) ≤ 0.
(1 + p0 β )α/γ
It follows now from [19, Theorem 1] that the associated delay differential equation
M β−α
u  (t) + R(t)[B(η(t))]α u α/γ (η(t)) = 0 (3.9)
(1 + p0 β )α/γ
also has a positive solution. However, under the condition (3.1), [13, Theorem 2] yields that Eq. (3.9) is oscillatory,
which contradicts the latter statement.
Case II. Assume now that l = 0. Since z(t) > 0 and z  (t) > 0, there exists a positive l1 (possibly +∞) such
that
lim z(t) = l1 > 0.
t→∞

Consequently,
l1
lim x(t) = >0
t→∞ 1 + p0

www.mn-journal.com 
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1156 T. Li and Yu. V. Rogovchenko: Oscillation of second-order neutral differential equations

and
x(σ (t) − τ 0 )
lim = 1.
t→∞ x(σ (t))
The latter condition means that, for any  ∈ (0, 1) and for t large enough,
x(σ (t) − τ0 )
1− < < 1 + .
x(σ (t))
Then,
x(σ (t)) x(σ (t)) 1 1
= = ≥ . (3.10)
z(σ (t)) x(σ (t)) + p0 x(σ (t) − τ0 ) x(σ (t)−τ0 )
1 + p0 x(σ (t)) 1 + p 0 (1 + )

Combining (1.1) with (3.10) and using the fact that η(t) ≤ σ (t), we observe that z(t) is a positive solution of a
delay differential inequality
 γ  q(t)
a(t) [z  (t)] + z β (η(t)) ≤ 0.
(1 + p0 (1 + ))β
Repetition of an argument similar to the one used above leads to the conclusion that the associated delay differential
equation
M β−α
u  (t) + q(t)[B(η(t))]α u α/γ (η(t)) = 0 (3.11)
(1 + p0 β )α/γ
also has a positive solution. Using condition (3.1) in [13, Theorem 2], we conclude that (3.11) is oscillatory. This
contradiction completes the proof. 
Theorem 3.2 Let 0 < β = γ ≤ 1 and τ0 ≤ 0. Suppose that assumptions (H3 )−(H5 ) are satisfied. If, for some
sufficiently large t1 ≥ t0 ,
 t
1 1
lim inf R(s)[B(η(s))]β ds > , (3.12)
1 + p0 β t→∞ η(t) e
Eq. (1.1) is oscillatory.
P r o o f . As in the proof of Theorem 3.1, one can show that the differential equation
1
y  (t) + R(t)[B(η(t))]β y(η(t)) = 0 (3.13)
1 + p0 β
has a positive solution y(t) and thus, equation
q(t)
u  (t) + [B(η(t))]β u(η(t)) = 0 (3.14)
(1 + p0 (1 + ))β
also has a positive solution u(t). On the other hand, using (3.12) in [4, Lemma 4], we conclude that Eq. (3.13) is
oscillatory, and so is (3.14). This contradiction completes the proof. 
Example 3.3 Consider a second-order nonlinear neutral differential equation

1/3  b
t 1/4 (x(t) + 2x(t − τ0 )) + 13/12 x 1/3 (λt) = 0, (3.15)
t
where τ0 < 0, b > 0, and λ ∈ [1, ∞). Let η(t) = λ1 t, for some λ1 ∈ (0, 1). Using Theorem 3.2, we conclude that
Eq. (3.15) is oscillatory provided that
41/3 1 1
bλ1 1/12 ln > .
1 + 21/3 λ1 e
Observe that neither results reported in [4] nor those in Section 2 can be applied to Eq. (3.15) since, for λ
= 1,
one has λ(t − τ0 )
= λt − τ0 . Furthermore, oscillation criteria reported by Zhong et al. [25] also fail to apply to
Eq. (3.15) since λ ≥ 1.


C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mn-journal.com
Math. Nachr. 288, No. 10 (2015) / www.mn-journal.com 1157

Theorem 3.4 Let 0 < β ≤ 1, τ0 ≥ 0, and η(t) < t − τ0 . Assume also that there exists an α ∈ R such that
α ≤ β and α < γ . If, for some sufficiently large t1 ≥ t0 , conditions (H3 )−(H5 ) are satisfied and (3.1) holds,
Eq. (1.1) is oscillatory.
P r o o f . Let x(t) be a positive solution of (1.1). It has been established in Theorem 3.1 that the function ω(t)
defined by (2.6) is decreasing and satisfies (3.6). Introducing a new function u(t) by (3.7), one observes that, by
virtue of (3.7),
 
u(t) ≤ ω(t − τ0 ) 1 + p0 β . (3.16)
Substituting (3.16) into (3.6), we conclude that u(t) is a positive solution of a delay differential inequality
M β−α
u  (t) + R(t)[B(η(t))]α u α/γ (η(t) + τ0 ) ≤ 0.
(1 + p0 β )α/γ
It follows then from [19, Theorem 1] that the associated delay differential equation
M β−α
u  (t) + R(t)[B(η(t))]α u α/γ (η(t) + τ0 ) = 0 (3.17)
(1 + p0 β )α/γ
also has a positive solution. On the other hand, using condition (3.1) in [13, Theorem 2], we conclude that Eq.
(3.17) is oscillatory. This contradiction completes the proof. 
Theorem 3.5 Let 0 < β = γ ≤ 1, τ0 ≥ 0, and η(t) ≤ t − τ0 . If, for some sufficiently large t1 ≥ t0 , assumptions
(H3 )−(H5 ) are satisfied and
 t
1 1
lim inf R(s)[B(η(s))]β ds > , (3.18)
1 + p0 β t→∞ η(t)+τ0 e
Eq. (1.1) is oscillatory.
P r o o f . As in the proof of Theorem 3.4, application of [19, Theorem 1] yields that the differential equation

1
y  (t) + R(t)[B(η(t))]β y(η(t) + τ0 ) = 0 (3.19)
1 + p0 β
has a positive solution. However, by virtue of condition (3.18), [4, Lemma 4] implies that Eq. (3.19) is oscillatory.
This contradiction completes the proof. 
Remark 3.6 Results presented in this section complement and improve several oscillation criteria reported
in [3], [4], [5], [16], [17], [25] because our theorems apply even if σ (t) − τ0
≡ σ (t − τ0 ) and σ (t) ≥ t − τ0 .
Oscillation criteria for Eq. (1.1) for β ≥ 1 can be deduced in a similar manner. To this end, while using [4, Lemma
1], one simply has to define R(t) := 21−β min{q(t), q(t − τ0 )}. Note that as in the paper by Zhong et al. [25],
methods used in this section require that τ (t) = t − τ0 and p(t) = p0 .

4 Oscillation results for unbounded p(t)

Throughout this section, we assume that
(H6 ) p(t) ≥ 1 and p(t)
≡ 1 eventually;
(H7 ) τ is strictly increasing and limt→∞ τ (t) = limt→∞ σ (t) = ∞;
∞ −1/γ
(H8 ) t0 a (s) ds < ∞.
The following lemma whose validity can be easily established is required in this section.
Lemma 4.1 Let x(t) be a positive solution of Eq. (1.1). Then, for t large enough, z(t) > 0, (a(t)[z  (t)]γ ) < 0,
and either
z  (t) > 0 or z  (t) < 0.

www.mn-journal.com 
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1158 T. Li and Yu. V. Rogovchenko: Oscillation of second-order neutral differential equations

Theorem 4.2 Assume that there exist α, λ ∈ R such that α ≤ β ≤ λ and α < γ < λ. Suppose further that
conditions (H6 )−(H8 ) hold, and there exist four functions δ∗ , δ ∗ , η∗ , η∗ ∈ C(I) such that δ∗ (t) ≤ t ≤ δ ∗ (t),
η∗ (t) < t < η∗ (t),
τ (δ∗ (t)) ≤ t ≤ τ (δ ∗ (t)), τ (η∗ (t)) ≤ σ (t) ≤ τ (η∗ (t)), (4.1)
and
lim η∗ (t) = lim δ∗ (t) = ∞.
t→∞ t→∞

Finally, assume that, for t ≥ t1 large enough,

 ∞
q(t)( p ∗ (σ (t)))β dt = ∞ (4.2)

and
 ∞
q(t)( p∗ (σ (t)))β Aλ (η∗ (t)) dt = ∞, (4.3)

where
  
∗ 1B τ −1 (δ ∗ (t)) 1
p (t) := 1−
p(τ −1 (t)) B(τ −1 (t)) p(τ −1 (τ −1 (t)))
and
  
1 A τ −1 (δ∗ (t)) 1
p∗ (t) := 1− .
p (τ −1 (t)) A(τ −1 (t)) p(τ −1 (τ −1 (t)))
Then (1.1) is oscillatory.
P r o o f . Let x(t) be a positive solution of (1.1). By Lemma 4.1, z  (t) does not change sign eventually.
Case I. Assume first that z  (t) > 0. Using the definition of z(t) and (4.1), we obtain
1   −1   
x(t) = z τ (t) − x τ −1 (t)
p(τ −1 (t))
      
1  −1  z τ −1 τ −1 (t) − x τ −1 τ −1 (t)
= z τ (t) −
p(τ −1 (t)) p (τ −1 (τ −1 (t)))
   
1  −1  z τ −1 τ −1 (t)
≥ z τ (t) −
p(τ −1 (t)) p (τ −1 (τ −1 (t)))
  
1  −1  z τ −1 (δ ∗ (t))
≥ z τ (t) − . (4.4)
p(τ −1 (t)) p (τ −1 (τ −1 (t)))

By Lemma 4.1, (a(t) [z  (t)]γ ) < 0. Therefore,
 t  t
 γ 1/γ −1/γ
z(t) ≥ a(s) [z  (s)] a (s) ds ≥ a 1/γ (t)z  (t) a −1/γ (s) ds. (4.5)
t1 t1

Consequently,
 
z(t)
≤ 0. (4.6)
B(t)
Using (4.6) and the condition δ ∗ (t) ≥ t, we conclude that
 
 −1 ∗  B τ −1 (δ ∗ (t))  −1 
z τ (δ (t)) ≤ z τ (t) .
B(τ −1 (t))


C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mn-journal.com
Math. Nachr. 288, No. 10 (2015) / www.mn-journal.com 1159

The inequality (4.4) yields

   
z τ −1 (t) B τ −1 (δ ∗ (t)) 1  
z(t) ≥ −1
1− −1 −1 −1
= p ∗ (t)z τ −1 (t) . (4.7)
p(τ (t)) B(τ (t)) p(τ (τ (t)))
Combining (1.1) and (4.7), one arrives at
 γ   
a(t) [z  (t)] + q(t) ( p ∗ (σ (t)))β z β τ −1 (σ (t)) ≤ 0. (4.8)
Hence, by (4.1) and (4.8), we deduce that
 γ 
a(t) [z  (t)] + q(t) ( p ∗ (σ (t)))β z β (η∗ (t)) ≤ 0. (4.9)

Since z (t) > 0, there exists a constant M > 0 such that
z β (η∗ (t)) = z β−α (η∗ (t))z α (η∗ (t)) ≥ M β−α z α (η∗ (t)). (4.10)
Combining (4.9) and (4.10), we get
 γ 
a(t) [z  (t)] + M β−α q(t)( p ∗ (σ (t)))β z α (η∗ (t)) ≤ 0. (4.11)
Using (4.5) in (4.11), we conclude that the function ω(t) defined by (2.6) is a positive solution of a delay differential
inequality
ω (t) + M β−α q(t)( p ∗ (σ (t)))β ωα/γ (η∗ (t)) ≤ 0.
It follows from [19, Theorem 1] that the associated delay differential equation
ω (t) + M β−α q(t)( p ∗ (σ (t)))β ωα/γ (η∗ (t)) = 0 (4.12)
also has a positive solution. On the other hand, if (4.2) holds, Eq. (4.12) cannot have positive solutions by virtue
of [13, Theorem 2], a contradiction.
Case II. Assume now that z  (t) < 0. By Lemma 4.1, one has a(s) [z  (s)]γ ≤ a(t) [z  (t)]γ , for all s ≥ t.
Consequently,
z  (t)
z  (s) ≤ a 1/γ (t) .
a 1/γ (s)
Integrating the latter inequality from t to ∞, we deduce that
z(t) ≥ −a 1/γ (t)A(t)z  (t), (4.13)
which, in turn, implies that
 
z(t) 
≥ 0. (4.14)
A(t)
Using the condition δ∗ (t) ≤ t, (4.1), (4.14), and proceeding as in the proof of Case I, we obtain
  
1  −1  z τ −1 (τ −1 (t))
x(t) ≥ z τ (t) −
p(τ −1 (t)) p(τ −1 (τ −1 (t)))
  
1  −1  z τ −1 (δ∗ (t))
≥ z τ (t) −
p(τ −1 (t)) p(τ −1 (τ −1 (t)))
  
z(τ −1 (t)) A τ −1 (δ∗ (t)) 1
≥ 1−
p(τ −1 (t)) A(τ −1 (t)) p(τ −1 (τ −1 (t)))
 
= p∗ (t)z τ −1 (t) . (4.15)
It follows from (1.1) and (4.15) that
 γ   
a(t) [z  (t)] + q(t)( p∗ (σ (t)))β z β τ −1 (σ (t)) ≤ 0.

www.mn-journal.com 
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1160 T. Li and Yu. V. Rogovchenko: Oscillation of second-order neutral differential equations

The latter inequality and (4.1) yield

 γ 
a(t) [z  (t)] + q(t)( p∗ (σ (t)))β z β (η∗ (t)) ≤ 0. (4.16)
Taking into account that z(t) is strictly decreasing, one can find a constant M∗ > 0 such that
z β (η∗ (t)) = z β−λ (η∗ (t))z λ (η∗ (t)) ≥ M∗ β−λ z λ (η∗ (t)). (4.17)
It follows then from (4.13), (4.16), and (4.17) that
 γ 
λ
a(t) [z  (t)] − M∗ β−λ q(t)( p∗ (σ (t)))β Aλ (η∗ (t)) a 1/γ (η∗ (t))z  (η∗ (t)) ≤ 0.
Let ω(t) be defined by (2.6). Obviously, ω(t) < 0, and we have
−ω (t) − M∗ β−λ q(t)( p∗ (σ (t)))β Aλ (η∗ (t))(−ω)λ/γ (η∗ (t)) ≥ 0.
Hence, −ω(t) is a positive solution of a differential inequality
y  (t) − M∗ β−λ q(t)( p∗ (σ (t)))β Aλ (η∗ (t))y λ/γ (η∗ (t)) ≥ 0.
By virtue of [2, Lemma 2.3], the associated differential equation
y  (t) − M∗ β−λ q(t)( p∗ (σ (t)))β Aλ (η∗ (t))y λ/γ (η∗ (t)) = 0 (4.18)
also has a positive solution. On the other hand, if (4.3) is satisfied, Eq. (4.18) has no positive solutions by virtue
of [13, Theorem 1]. This contradiction completes the proof. 
Corollary 4.3 Assume that conditions (H6 )−(H8 ) hold, β = γ , and there exist four functions δ∗ , δ ∗ , η∗ ,
∗
η ∈ C(I) satisfying the assumptions of Theorem 4.2. If
 t
1
lim inf q(s)( p ∗ (σ (s)))β ds > (4.19)
t→∞ η∗ (t) e
and
 η∗ (t)
1
lim inf q(s)( p∗ (σ (s)))β Aβ (η∗ (s))ds > , (4.20)
t→∞ t e
Eq. (1.1) is oscillatory.
P r o o f . Let α = β = γ = λ. Then Eqs. (4.12) and (4.18) in Theorem 4.2 assume, respectively, the form
ω (t) + q(t)( p ∗ (σ (t)))β ω(η∗ (t)) = 0 (4.21)
and
y  (t) − q(t)( p∗ (σ (t)))β Aβ (η∗ (t))y(η∗ (t)) = 0. (4.22)
Observe that both (4.21) and (4.22) have positive solutions. On the other hand, applying condition (4.19) to (4.21),
one concludes that (4.21) cannot have positive solutions by [14, Theorem 2.1.1]. Similarly, condition (4.20)
yields, by virtue of [14, Theorem 2.4.1], that Eq. (4.22) has no positive solutions. This contradiction completes
the proof. 
Example 4.4 For t ≥ 1, consider a second-order neutral differential equation
 t  √    π
e (x(t) + 2e2π x(t + 2π )) + 2 1 + 2e2π et x t − = 0. (4.23)
4
Choosing δ∗ (t) = t − 2π , δ ∗ (t) = t, η∗ (t) = t − 9π/4, and η∗ (t) = t + π , we conclude that Eq. (4.23) is oscil-
latory by Corollary 4.3. In fact, x(t) = sin t is one oscillatory solution of this equation.
Remark 4.5 Oscillation theorems in [1], [10], [15], [16] are derived under the assumptions τ (t) ≥ t, σ (t) ≥ t,
σ (t) ≤ τ (t) ≤ t, or τ (t) ≤ σ (t) ≤ t. Using different methods, we improve cited results by providing sufficient
conditions which insure that all solutions of (1.1) are oscillatory without requiring similar conditions on the
functions τ and σ . Oscillation criteria established
 ∞for Eq. (1.1) in this section complement
∞ theorems reported by
Agarwal et al. [1], because we use assumption t0 a −1/γ (s) ds < ∞ rather than t0 a −1/γ (s) ds = ∞. On the


C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mn-journal.com
Math. Nachr. 288, No. 10 (2015) / www.mn-journal.com 1161

other hand, our results complement those by Li et al. [10], [16] since our conditions allow unbounded functions
p(t).

Remark 4.6 We point out versatility of the obtained results with respect to the behavior of the functions τ and
σ . Our theorems apply in the cases where τ (t) ≤ t, τ (t) ≥ t, as well as when τ (t) − t oscillates. Similarly, σ (t)
can be a delayed argument, an advanced argument, and σ (t) − t can also oscillate.

Remark 4.7 As in the paper by Agarwal et al. [1], in the case where p(t) ≥ 1, we require τ to be strictly
increasing. The question regarding oscillatory behavior of Eq. (1.1) without monotonicity assumption on τ remains
open at the moment.

Acknowledgements The authors thank three anonymous referees for careful reading of the manuscript and useful comments
that helped to correct several inaccuracies in the original submission. The research of the first author is supported by the AMEP
of Linyi University, P. R. China.

References
[1] R. P. Agarwal, M. Bohner, and W. T. Li, Nonoscillation and Oscillation: Theory for Functional Differential Equations
(Marcel Dekker Inc., New York, 2004).
[2] B. Baculı́ková, Properties of third-order nonlinear functional differential equations with mixed arguments, Abstr. Appl.
Anal. 2011, 1–15 (2011).
[3] B. Baculı́ková and J. Džurina, Oscillation theorems for second order neutral differential equations, Comput. Math. Appl.
61, 94–99 (2011).
[4] B. Baculı́ková and J. Džurina, Oscillation theorems for second-order nonlinear neutral differential equations, Comput.
Math. Appl. 62, 4472–4478 (2011).
[5] B. Baculı́ková, J. Džurina, and T. Li, Oscillation results for even-order quasilinear neutral functional differential equa-
tions, Electron. J. Diff. Equ. 2011, 1–9 (2011).
[6] J. G. Dong, Oscillation behavior of second order nonlinear neutral differential equations with deviating arguments,
Comput. Math. Appl. 59, 3710–3717 (2010).
[7] L. H. Erbe, Q. Kong, and B. G. Zhang, Oscillation Theory for Functional Differential Equations (Marcel Dekker Inc.,
New York, 1995).
[8] M. K. Grammatikopoulos, G. Ladas, and A. Meimaridou, Oscillation of second order neutral delay differential equation,
Rad. Math. 1, 267–274 (1985).
[9] I. Győri and G. Ladas, Oscillation Theory of Delay Differential Equations with Applications (Clarendon Press, Oxford,
1991).
[10] Z. Han, T. Li, S. Sun, and Y. Sun, Remarks on the paper [Appl. Math. Comput. 207 (2009) 388–396], Appl. Math.
Comput. 215, 3998–4007 (2010).
[11] M. Hasanbulli and Yu. V. Rogovchenko, Asymptotic behavior of nonoscillatory solutions of second order nonlinear
neutral differential equations, Math. Ineq. Appl. 10, 607–618 (2007).
[12] M. Hasanbulli and Yu. V. Rogovchenko, Oscillation criteria for second order nonlinear neutral differential equations,
Appl. Math. Comput. 215, 4392–4399 (2010).
[13] Y. Kitamura and T. Kusano, Oscillation of first-order nonlinear differential equations with deviating arguments, Proc.
Amer. Math. Soc. 78, 64–68 (1980).
[14] G. S. Ladde, V. Lakshmikantham, and B. G. Zhang, Oscillation Theory of Differential Equations with Deviating
Arguments (Marcel Dekker Inc., New York, 1987).
[15] T. Li, R. P. Agarwal, and M. Bohner, Some oscillation results for second-order neutral differential equations, J. Indian
Math. Soc. 79, 97–106 (2012).
[16] T. Li, Z. Han, C. Zhang, and H. Li, Oscillation criteria for second-order superlinear neutral differential equations, Abstr.
Appl. Anal. 2011, 1–17 (2011).
[17] T. Li, Z. Han, P. Zhao, and S. Sun, Oscillation of even-order neutral delay differential equations, Adv. Difference Equ.
2010, 1–9 (2010).
[18] L. H. Liu and Y. Z. Bai, New oscillation criteria for second-order nonlinear neutral delay differential equations, J.
Comput. Appl. Math. 231, 657–663 (2009).
[19] Ch. G. Philos, On the existence of nonoscillatory solutions tending to zero at ∞ for differential equations with positive
delays, Arch. Math. 36, 168–178 (1981).
[20] X. Tang, Oscillation for first order superlinear delay differential equations, J. London Math. Soc. 65, 115–122 (2002).
[21] J. S. W. Wong, On the generalized Emden-Fowler equation, SIAM Rev. 17, 339–360 (1975).

www.mn-journal.com 
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1162 T. Li and Yu. V. Rogovchenko: Oscillation of second-order neutral differential equations

[22] R. Xu and F. Meng, Some new oscillation criteria for second order quasi-linear neutral delay differential equations,
Appl. Math. Comput. 182, 797–803 (2006).
[23] R. Xu and F. Meng, Oscillation criteria for second order quasi-linear neutral delay differential equations, Appl. Math.
Comput. 192, 216–222 (2007).
[24] R. Xu and Y. Xia, A note on the oscillation of second-order nonlinear neutral functional differential equations, Int. J.
Contemp. Math. Sci. 3, 1441–1450 (2008).
[25] J. Zhong, Z. Ouyang, and S. Zou, An oscillation theorem for a class of second-order forced neutral delay differential
equations with mixed nonlinearities, Appl. Math. Lett. 24, 1449–1454 (2011).


C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mn-journal.com