Open Access

Oscillation and asymptotic properties of a class of second-order Emden–Fowler neutral differential equations

SpringerPlus20165:1956

https://doi.org/10.1186/s40064-016-3622-2

Received: 25 May 2016

Accepted: 1 November 2016

Published: 10 November 2016

Abstract

We consider a class of second-order Emden–Fowler equations with positive and nonpositve neutral coefficients. By using the Riccati transformation and inequalities, several oscillation and asymptotic results are established. Some examples are given to illustrate the main results.

Keywords

OscillationSecond-order differential equationEmden–Fowler equationNonpositive neutral coefficient

Mathematics Subject Classification

34K11

Background

In this paper, we study a second-order delay differential equation
$$\left( r(t)|z^{\prime }(t)|^{\alpha -1}z^{\prime }(t)\right) ^{\prime }+f(t,x(\sigma (t)))=0,\quad t\ge t_0$$
(1)
where \(z(t)=x(t)+ p(t)x(\tau (t))\) and \(\alpha\) is a positive constant. Throughout this paper, we assume that
  • (\(H_{1}\)) \(r,\;p\in {\mathrm {C}}([t_0, \infty ),{\mathbb {R}}),\; r(t)>0,\; 0 \le p(t)\le 1,\;\text {and}\; \int _{t_0}^{\infty }r^{-1/\alpha }(t) {\mathrm{d}}t=\infty ;\)

  • (\(H_{2}\)) \(r,\;p\in {\mathrm {C}}([t_0, \infty ),{\mathbb {R}}),\; r(t)>0,\; -1< -p_0 \le p(t)\le 0,\;\text {and}\; \int _{t_0}^{\infty }r^{-1/\alpha }(t) {\mathrm{d}}t=\infty ,\) where \(p_0\) is a positive constant;

  • (\(H_{3}\)) \(\tau \in {\mathrm {C}}([t_0, \infty ),{\mathbb {R}}), \; \tau (t)\le t, \; \text {and} \; \lim _{t\rightarrow \infty }\tau (t)=\infty ;\)

  • (\(H_{4}\))\(\sigma \in {\mathrm{C}}^{1}([t_0, \infty ),{\mathbb {R}}),\; \sigma (t)\le t,\; \sigma^{\prime}(t)>0,\;\text {and} \; \lim _{t\rightarrow \infty }\sigma (t)=\infty .\)

  • (\(H_{5}\)) \(f\in {\mathrm {C}}([t_0,\infty )\times {\mathbb {R}},{\mathbb {R}}),\;uf(u)\ge 0\; \text {for all}\; u\ne 0,\) and there exist a positive constant \(\beta \;\) and a function \(\; q(t)\in {\mathrm {C}}([t_0,\infty ),(0,\infty ))\; \text {such that}\; (f(t,u)/u^{\beta })\ge q(t),\) for all \(u\ne 0\), where \(1<\beta \le \alpha .\)

It is recognized that Emden–Fowler equations have a number of applications in physics and engineering; see, e.g., Berkovich (1997). As a result, there has been a great deal of interest in investigating the oscillation or nonoscillation of differential equations; see, e.g., Hale (1977), Džurina and Stavroulakis (2003), Li (2004), Li et al. (2011, 2013, 2015), Li and Rogovchenko (2015), Erbe et al. (2009), Manojlović (1999), Wang and Yang (2004), Wang (2001), Liu et al. (2012), Shi et al. (2016), Baculíková and Džurina (2011), Bohner and Li (2014), Yang et al. (2006), Xu and Meng (2006, 2007). As we known, many results of half-linear or nonlinear equations with positive neutral coefficients were established, see, e.g., Baculíková and Džurina (2011), Erbe et al. (2009), Li et al. (2011, 2013), Li and Rogovchenko (2015), Liu et al. (2012), Shi et al. (2016), Yang et al. (2006), Xu and Meng (2006, 2007). The equations with nonpositive neutral coefficients have been applied to practical life; see, for instance, Brayton (1966) and Kuang (1993, sec. 1.1.7) provided a model about the system with lossless transmission lines. And, there have been a few oscillation and asymptotic results of the equations with nonpositive neutral coefficients, see, e.g., Bohner and Li (2014), Erbe et al. (2009), Li et al. (2015), Yang et al. (2006). In the following, we provide some background details which motivated our research. Manojlović (1999), Wang (2001), and Wang and Yang (2004) considered the half-linear differential equation
$$\begin{aligned} \left( r(t)|x^{\prime }(t)|^{\alpha -1}x^{\prime }(t)\right) ^{\prime }+q(t)|x(t)|^{\alpha -1}x(t)=0, \end{aligned}$$
(2)
and gave some different oscillation results by using an inequality due to Hardy, Littlewood and Ploya and averaging functions. Motivated by these ideas, many scholars extended the results to delay differential equations or neutral delay differential equations. Džurina and Stavroulakis (2003) and Li (2004) expanded the Eq. (2) to the delay differential equation
$$\begin{aligned} \left( r(t)|x^{\prime }(t)|^{\alpha -1}x^{\prime }(t)\right) ^{\prime }+q(t)|x(\tau (t))|^{\alpha -1}x(\tau (t))=0. \end{aligned}$$
Xu and Meng (2006, 2007) extended (2) to the neutral delay differential equation
$$\begin{aligned} \left( r(t)\left( \left[ x(t)+ p(t)x(\tau (t))\right] ^{\prime }\right) ^{\alpha }\right) ^{\prime }+q(t)(x(\sigma (t)))^{\alpha }=0 \end{aligned}$$
(3)
provided that \(0\le p(t)\le 1\). Li et al. (2015) established some oscillation and asymptotic results to (1) in the case where \(-1<p(t)\le 0\). By using Riccati transformation, Erbe et al. (2009) proposed some oscillation and asymptotic results for (1), under the assumptions that \(0\le p(t)<1\) and \(-1< p(t)< 0\). Liu et al. (2012) considered the the following equation
$$\begin{aligned} \left( r(t)\left| \left[ x(t)+ p(t)x(\tau (t))\right] ^{\prime }\right| ^{\alpha -1}z^{\prime }(t)\right) ^{\prime }+q(t)|x(\sigma (t))|^{\beta -1}x(\sigma (t))=0 \end{aligned}$$
(4)
in the case where \(0\le p(t)\le 1\) and \(\alpha \ge \beta >0\). They established some oscillation and asymptotic criteria by employing averaging technique and Riccati transformation. Shi et al. (2016) extended the results of Liu et al. (2012) to dynamic equations on time scales provided that \(0\le p(t)\le 1\) and \(p(t)> 1\).

However, the results of Liu et al. (2012) and Shi et al. (2016) cannot be applied to Eq. (1) due to \(-1<p(t)\le 0\) in (1), but, in Liu et al. (2012), Shi et al. (2016) the assumption is \(0\le p(t)\le 1\) or \(p(t)>1\). Similarly, the results in Erbe et al. (2009) and Li et al. (2015) cannot be applied to Eq. (1) because there is another parameter \(\beta\) and the condition on function f in Li et al. (2015), Erbe et al. (2009) does not satisfy the hypothesis \((H_5)\). In this paper, we will extend the results of Liu et al. (2012), Shi et al. (2016) to the case of \(-1< p(t) \le 0\) and improve the results of Erbe et al. (2009), Li et al. (2015). By employing Riccati transformation, several new oscillation and asymptotic criteria are obtained under the assumptions that \((H_1)-(H_5)\). Throughout this paper, we suppose that all inequalities hold for sufficiently large t. Without loss of generality, we only consider the positive solutions of (1).

In what follows, let \(D=\{(t, s):t_0\le s \le t\}\; \mathrm {and}\;D_0=\{(t, s):t_0\le s < t\}.\) We say a function \(H=H(t,s)\) belongs to a function class P, if it satisfies
  1. (i)

    \(H(t, t)=0, t\ge t_0; H(t,s)>0, (t, s)\in D_0;\)

     
  2. (ii)
    H has partial derivatives \(\partial H/\partial t\) and \(\partial H{/}\partial s\) on \(D_0\), such that
    $$\begin{aligned} \frac{\partial H(t, s)}{\partial t}=h_1(t, s)\sqrt{H(t, s)} \end{aligned}$$
    and
    $$\begin{aligned} \frac{\partial H(t, s)}{\partial s}=-h_2(t, s)\sqrt{H(t, s)}, \end{aligned}$$
    where \(h_1\) and \(h_2\) are nonnegative continuous functions on \(D_0\).
     

Main results

In this section, we discuss the Eq. (1) under the assumptions that \(-1< p(t)\le 0\) and \(0\le p(t)\le 1\), respectively.

Oscillation of Eq. (1) when \(-1<p(t)\le 0\)

Theorem 1

Assume that \((H_2)-(H_5)\) hold. If there exists a function \(\rho \in {\mathrm{C}}^1([t_0, \infty ), (0,\infty ))\) such that, for any constant \(K>0\),
$$\begin{aligned} \int _{t_0}^\infty \left[ \rho (t)q(t)-\frac{\left( \rho ^{\prime }(t)\right) ^2}{4\rho ^2(t)\Phi (t)}\right] {\mathrm{d}}t =\infty , \end{aligned}$$
(5)
where \(\Phi (t)=[\beta \sigma ^{\prime }(t)(\xi (\sigma (t)))^{\beta -1}]/[K^{(1-\beta /\alpha )}\rho (t)(r(\sigma (t)))^{1/\alpha }]\), then all solutions of Eq. (1) are oscillatory or tend to zero as \(t\rightarrow \infty\).

Proof

Suppose x is a nonoscillatory solution of (1). Without loss of generality, there exists a \(t_1\ge t_0\), such that \(x(t)>0, x(\tau (t))>0,\;\text {and}\;x(\sigma (t))>0,\) for all \(t\ge t_1\). From (1) and the hypothesis \((H_5)\), we get
$$\begin{aligned} \left( r|z^{\prime }|^{\alpha -1}z^{\prime }\right) ^{\prime }\le 0. \end{aligned}$$
(6)
Therefore, \(r|z^{\prime }|^{\alpha -1}z^{\prime }\) is nonincreasing. We claim that \(z^{\prime }>0\). Otherwise, if \(z^{\prime }<0\), using the fact that \(r|z^{\prime }|^{\alpha -1}z^{\prime }\) is nonincreasing, there exists a positive \(k>0\), such that
$$\begin{aligned} -r(t)\left( -z^{\prime }(t)\right) ^{\alpha }\le -k<0. \end{aligned}$$
That is,
$$\begin{aligned} -z^{\prime }(t)\ge \frac{k^{\frac{1}{\alpha }}}{r^{\frac{1}{\alpha }}(t)}. \end{aligned}$$
Integrating the above inequality from \(t_1\) to t, we get
$$\begin{aligned} z(t_1)-z(t) \ge k^{\frac{1}{\alpha }}\int _{t_1}^t r^{-\frac{1}{\alpha }}(s){\mathrm{d}}s. \end{aligned}$$
It follows from \((H_2)\) that
$$\begin{aligned} \lim _{t\rightarrow \infty }z(t)=-\infty . \end{aligned}$$
(7)
We consider the following two cases.
Case 1 If x is unbounded, then there exists a sequence \(\{t_m\}\), such that
$$\begin{aligned} \lim _{m\rightarrow \infty } x(t_m)=\infty , \end{aligned}$$
where \(\{t_m\}\) satisfies \(\lim _{m\rightarrow \infty }t_m=\infty\) and \(x(t_m)=\max _{t_0\le s\le t_m}\{x(s)\}\). By the definition of \(x(t_m)\) and \(\tau (t)\le t\), we have
$$\begin{aligned} x(\tau (t_m))\le x(t_m). \end{aligned}$$
Then we get
$$\begin{aligned} z(t_m)=x(t_m)+p(t_m)x(\tau (t_m))\ge x(t_m)(1+p(t_k))>0, \end{aligned}$$
which contradicts (7).

Case 2 If x is bounded, from the definition of z and \(-1< p(t) \le 0\), z is also bounded, which also contradicts (7).

Hence, it is clear from the above discussion that \(z^{\prime }>0\), and then \(z>0\) or \(z<0\). We consider each of two cases separately.

Suppose first that \(z>0\). Considering the definition of z and \(-1<p(t)\le 0\), we get
$$\begin{aligned} z(t)=x(t)+p(t)x(\tau (t))\le x(t). \end{aligned}$$
(8)
From \(\sigma (t)\le t\) and the fact that \(r(z^{\prime })^{\alpha }\) is nonincreasing, we obtain
$$\begin{aligned} \left( r(\sigma (t))\right) ^{\frac{1}{\alpha }}z^{\prime }(\sigma (t))\ge \left( r(t)\right) ^{\frac{1}{\alpha }}z^{\prime }(t) \end{aligned}$$
(9)
and there exist a positive constant K and a \(t_2\ge t_1\), such that
$$\begin{aligned} r(t)\left( z^{\prime }(t)\right) ^{\alpha }\le K,\quad t\ge t_2. \end{aligned}$$
(10)
From the fact that \(r(z^{\prime })^{\alpha }\) is nonincreasing, we get
$$z(t) = {} z(t_1)+\int _{t_1}^t \frac{\left( r(s)\left( z^{\prime }(s)\right) ^{\alpha }\right) ^{\frac{1}{\alpha }}}{r^{\frac{1}{\alpha }}(s)} {\mathrm{d}} s \ge r^{\frac{1}{\alpha }}(t)z^{\prime }(t)\int _{t_1}^t r^{-\frac{1}{\alpha }}(s){\mathrm{d}}s = \xi (t)r^{\frac{1}{\alpha }}(t)z^{\prime }(t),$$
(11)
where \(\xi (t)=\int _{t_1}^t r^{-1/{\alpha }}(s) {\mathrm{d}}s\). Define a function \(\omega\) by
$$\begin{aligned} \omega (t)=\rho (t)\frac{r(t)\left( z^{\prime }(t)\right) ^{\alpha }}{z^{\beta }(\sigma (t))}, \end{aligned}$$
then \(\omega (t)>0\). Differentiating \(\omega\), we get
$$\omega ^{\prime }(t) = \frac{\rho ^{\prime }(t)}{\rho (t)}\omega (t)+\rho (t)\frac{\left( r(t)z^{\prime }(t)\right) ^{\prime }}{z^{\beta }(\sigma (t))} - \beta \sigma ^{\prime }(t)\rho (t)\frac{r(t)\left( z^{\prime }(t)\right) ^{\alpha }z^{\beta -1}(\sigma (t))z^{\prime }(\sigma (t))}{z^{2\beta }(\sigma (t))}.$$
From (1) and (8), we conclude that
$$\begin{aligned} \omega ^{\prime }(t) \le -\rho (t)q(t)+\frac{\rho ^{\prime }(t)}{\rho (t)}\omega (t)-\beta \sigma ^{\prime }(t)\rho (t)\frac{r(t)\left( z^{\prime }(t)\right) ^{\alpha }}{z^{2\beta }(\sigma (t))} z^{\prime }(\sigma (t))z^{\beta -1}(\sigma (t)). \end{aligned}$$
Taking into account (11), the last inequality implies
$$\begin{aligned}&\omega ^{\prime }(t)\le - \rho (t)q(t)+\frac{\rho ^{\prime }(t)}{\rho (t)}\omega (t)-\beta \sigma ^{\prime }(t)(\xi (\sigma (t)))^{\beta -1}\nonumber \\&\quad \qquad \rho (t) \frac{r(t)\left( z^{\prime }(t)\right) ^{\alpha }}{z^{2\beta }(\sigma (t))} \left( r(\sigma (t))\right) ^{\frac{\beta -1}{\alpha }}\left( z^{\prime }(\sigma (t))\right) ^{\beta }. \end{aligned}$$
(12)
It follows from (9), (10), and (12) that
$$\begin{aligned} \omega ^{\prime }(t)&\le - \rho (t)q(t)+\frac{\rho ^{\prime }(t)}{\rho (t)}\omega (t)- \frac{\beta \sigma ^{\prime }(t)\left( \xi (\sigma (t))\right) ^{\beta -1}}{\left( r(\sigma (t))\right) ^{1/\alpha }} \rho (t)\frac{r(t)\left( z^{\prime }(t)\right) ^{\alpha }}{z^{2\beta }(\sigma (t))} \left( r(t)\right) ^{\frac{\beta }{\alpha }}\left( z^{\prime }(t)\right) ^{\beta }\\&=-\rho (t)q(t)+\frac{\rho ^{\prime }(t)}{\rho (t)}\omega (t)- \frac{\beta \sigma ^{\prime }(t)\left( \xi (\sigma (t))\right) ^{\beta -1}\left( r(t)\right) ^{\frac{\beta }{\alpha }}}{\left( r(\sigma (t))\right) ^{1/\alpha }\rho (t)r(t)} \omega ^2(t)\frac{1}{\left( z^{\prime }(t)\right) ^{\alpha -\beta }}\\&\le -\rho (t)q(t)+\frac{\rho ^{\prime }(t)}{\rho (t)}\omega (t)- \frac{\beta \sigma ^{\prime }(t)\left( \xi (\sigma (t))\right) ^{\beta -1}}{K^{1-\frac{\beta }{\alpha }}\left( r(\sigma (t))\right) ^{1/\alpha }\rho (t)} \omega ^2(t)\\&= -\rho (t)q(t)+\frac{\rho ^{\prime }(t)}{\rho (t)}\omega (t)-\Phi (t)\omega ^2(t)\\&\le -\rho (t)q(t)+\frac{(\rho ^{\prime }(t))^2}{4\rho ^2(t)\Phi (t)},\\ \end{aligned}$$
(13)
where \(\Phi (t)=[\beta \sigma ^{\prime }(t)(\xi (\sigma (t)))^{\beta -1}]/[K^{(1-\beta /\alpha )}\rho (t)(r(\sigma (t)))^{1/\alpha }].\) Integrating (13) from \(t_2\) to t, we get
$$\begin{aligned} 0<\omega (t)\le \omega (t_2)-\int ^t_{t_2}\left[ \rho (s)q(s)-\frac{\left( \rho ^{\prime }(s)\right) ^2}{4\rho ^2(s)\Phi (s)}\right] {\mathrm{d}}s, \end{aligned}$$
which contradicts (5).
If \(z<0\), we claim that \(\lim _ {t\rightarrow \infty }{x(t)}=0\). Using \(z<0\) and \(z^{\prime }>0\), we deduce that
$$\begin{aligned} \lim _{t\rightarrow \infty }z(t)=l\le 0, \end{aligned}$$
where l is a constant. That is, for all sufficiently large t, z is bounded. We can easily prove that x is also bounded. From the fact that x is bounded, we get
$$\begin{aligned} \limsup _{t\rightarrow \infty } x(t)=a, \end{aligned}$$
where a is a constant. We claim that \(a=0\). Otherwise, if \(a>0\), there exists a sequence \(\{t_n\}\), such that \(\lim _{n\rightarrow \infty }t_n=\infty\) and \(\lim _{n\rightarrow \infty } x(t_n)=a.\) Letting
$$\begin{aligned} \varepsilon =\frac{a(1-p_0)}{2p_0}, \end{aligned}$$
for large enough n, we obtain
$$\begin{aligned} x(\tau (t_n))<a+\varepsilon . \end{aligned}$$
Then, from the definition of z(t) and \(p(t)\ge -p_0\), we have
$$\begin{aligned} \lim _{n\rightarrow \infty }z(t_n)\ge \lim _{n\rightarrow \infty }x(t_n)-p_0(a+\varepsilon )=\frac{a(1-p_0)}{2}>0, \end{aligned}$$
which contradicts \(z<0\). Thus \(\limsup _{t\rightarrow \infty } x(t)=0.\) By \(x>0\), we get
$$\begin{aligned} \lim _ {t\rightarrow \infty }{x(t)}=0. \end{aligned}$$
The proof is complete. \(\square\)

From Theorem 1, letting \(\rho =1\), we get the following corollary.

Corollary 1

Assume that \((H_2)-(H_5)\) hold. If
$$\begin{aligned} \int _{t_0}^{\infty }q(t) {\mathrm{d}}t=\infty , \end{aligned}$$
then all solutions of Eq. (1) are oscillatory or tend to zero as \(t\rightarrow \infty\).

Theorem 2

Assume that \((H_2)-(H_5)\) hold. If there exist two functions \(H\in P\) and \(\rho \in {\mathrm{C}}^1([t_0, \infty ), (0,+\infty ))\), such that
$$\begin{aligned} \limsup _{t\rightarrow \infty } \frac{1}{H(t,t_0)}\int _{t_0}^t\left[ H(t,s)\rho (s)q(s)-\frac{\left( \frac{\rho ^{\prime }(s)}{\rho (s)}H(t,s) -h_2(t,s)\sqrt{H(t,s)}\right) ^2}{4H(t,s)\Phi (s)}\right] {\mathrm{d}}s=\infty , \end{aligned}$$
(14)
where \(\Phi\) is as in Theorem 1, then the conclusion of Theorem 1 remains intact.

Proof

Suppose that x is a nonoscillatory solution of (1). Proceeding as in the proof of Theorem 1, we get \(z>0\) or \(z<0\).

Firstly, we consider \(z>0\). As the proof above (13) holds. That is
$$\begin{aligned} \omega ^{\prime }(t) \le -\rho (t)q(t)+\frac{\rho ^{\prime }(t)}{\rho (t)}\omega (t)-\Phi (t)\omega ^2(t). \end{aligned}$$
Multiplying this inequality by H(ts) and integrating it from \(t_2\) to t, we have
$$\begin{aligned} \int _{t_2}^t H(t,s)\omega ^{\prime }(s){\mathrm{d}}s \le \int _{t_2}^t H(t,s)\left[ -\rho (s)q(s)+ \frac{\rho ^{\prime }(s)}{\rho (s)}\omega (s)-\Phi (s)\omega ^2(s)\right] {\mathrm{d}}s. \end{aligned}$$
(15)
By the property of \(\partial H(t, s)/\partial s=-h_2(t, s)\sqrt{H(t, s)}<0\), we conclude that
$$\begin{aligned}&\int _{t_2}^t\left[ H(t,s)\rho (s)q(s)-\frac{\left( \frac{\rho ^{\prime }(s)}{\rho (s)}H(t,s) -h_2(t,s)\sqrt{H(t,s)}\right) ^2}{4H(t,s)\Phi (s)}\right] {\mathrm{d}}s\nonumber \\&\quad \le H(t,t_2)\omega (t_2)\le H(t,t_0)\omega (t_2) . \end{aligned}$$
Adding \(\int _{t_0}^{t_2}\left[H(t,s)\rho (s)q(s)- \frac{\left( \frac{\rho ^{\prime }(s)}{\rho (s)}H(t,s)-h_2(t,s)\sqrt{H(t,s)}\right) ^2}{4H(t,s)\Phi (s)}\right]{\mathrm{d}}s\) to the latter inequality and multiplying this inequality by \(1/H(t,t_0)\), we get
$$\begin{aligned} &\frac{1}{H(t,t_0)}\left[ \int _{t_0}^{t_2}+\int _{t_2}^t\right] \left[ H(t,s)\rho (s)q(s)-\frac{\left( \frac{\rho ^{\prime }(s)}{\rho (s)}H(t,s)-h_2(t,s)\sqrt{H(t,s)}\right) ^2}{4H(t,s)\Phi (s)}\right] {\mathrm{d}}s\\&\quad \quad \le \frac{1}{H(t,t_0)}\int _{t_0}^{t_2}\left[ H(t,s)\rho (s)q(s)- \frac{\left( \frac{\rho ^{\prime }(s)}{\rho (s)}H(t,s)-h_2(t,s)\sqrt{H(t,s)}\right) ^2}{4H(t,s)\Phi (s)}\right] {\mathrm{d}}s + \omega (t_2).\\&\quad \quad \le \omega (t_2)+\int _{t_0}^{t_2}\left[ \frac{H(t,s)}{H(t,t_0)}\rho (s)q(s)- \frac{\left( \frac{\rho ^{\prime }(s)}{\rho (s)}\sqrt{H(t,s)}-h_2(t,s)\right) ^2}{4H(t,t_0)\Phi (s)}\right] {\mathrm{d}}s\\&\quad \quad \le \omega (t_2)+\int _{t_0}^{t_2}\rho (s)q(s) {\mathrm{d}}s.\\ \end{aligned}$$
(16)
Letting \(t\rightarrow \infty\) in (16), we can get a contradict to (14).

If \(z<0\), repeating the proof of Theorem 1, we have \(\lim _ {t\rightarrow \infty }{x(t)}=0\). This completes the proof. \(\square\)

From Theorem 2, letting \(H(t,s)=(t-s)^{\lambda } (\lambda >0)\) and \(\rho (t)=1\), we may get the following corollary.

Corollary 2

Assume that \((H_2)-(H_5)\) hold. If
$$\begin{aligned} \limsup _{t\rightarrow \infty }\frac{1}{(t-t_0)^{\lambda }}\int _{t_0}^t (t-s)^{\lambda -2}\left[ (t-s)^2q(s)-\frac{\lambda ^2}{4\Phi (s)}\right] {\mathrm{d}}s=\infty , \end{aligned}$$
(17)
then the conclusion of Theorem 1 remains intact.

Theorem 3

Assume that \((H_2)-(H_5)\) hold. If there exist three functions \(H\in P\), \(\rho \in ([t_0, \infty ), (0,\infty ))\), and \(A\in {\mathrm {C}}([t_0,\infty ), {\mathbb {R}})\), such that
$$\begin{aligned} \int _{t_0}^{\infty }A_{+}^2(t)\Phi (t){\mathrm{d}} t=\infty \end{aligned}$$
(18)
and
$$\begin{aligned}&\limsup _{t\rightarrow \infty }\frac{1}{H(t,T)}\nonumber \\&\quad \int _T^t\left[ h(t,s)\rho (s)q(s)- \frac{\theta \left( \frac{\rho ^{\prime }(s)}{\rho (s)}H(t,s)- h_2(t,s)\sqrt{H(t,s)}\right) ^2}{4H(t,s)\Phi (s)}\right] {\mathrm{d}}s\ge A(T), \end{aligned}$$
(19)
for all \(T\ge t_0\) and for some constant \(\theta >1\), where \(A_{+}(t)=\max \{A(t),0\}\), \(\Phi\) is as in Theorem 1, and H satisfies
$$\begin{aligned} 0<\inf _{s\ge t_0}\left\{ \limsup _{t\rightarrow \infty }\frac{H(t,s)}{H(t,t_0}\right\} \le \infty , \end{aligned}$$
(20)
then every solution of Eq. (1) is oscillatory or tends to zero as \(t\rightarrow \infty\).

Proof

Suppose that x is a nonoscillatory solution of Eq. (1). Then, as in the proof of Theorem 1, \(z>0\) or \(z<0\).

We consider \(z>0\) firstly. By virtue of Theorem 2, (15) holds. That is, for all \(T\ge t_2\ge t_1,\)
$$\begin{aligned} &\frac{1}{H(t,T)}\int _{T}^t H(t,s)\rho (s)q(s){\mathrm{d}}s\\&\quad \le \omega (T)+\frac{1}{H(t,T)} \int _{T}^t \frac{\theta \left( \frac{\rho ^{\prime }(s)}{\rho (s)}H(t,s) -h_2(t,s)\sqrt{H(t,s)}\right) ^2}{4\Phi (s)H(t,s)}{\mathrm{d}}s\\&\quad -\frac{1}{H(t,T)}\int _T^t\frac{\theta -1}{\theta }H(t,s)\Phi (s)\omega ^2(s){\mathrm{d}}s.\\ \end{aligned}$$
Thus,
$$\begin{aligned} &\frac{1}{H(t,T)}\int _T^t \left[ H(t,s)\rho (s)q(s)-\frac{\theta \left( \frac{\rho ^{\prime }(s)}{\rho (s)}H(t,s)- h_2(t,s)\sqrt{H(t,s)}\right) ^2}{4\Phi (s)H(t,s)}\right] {\mathrm{d}}s\\&\quad \quad \le \omega (T)-\frac{1}{H(t,T)}\int _T^t \frac{\theta -1}{\theta }H(t,s)\Phi (s)\omega ^2(s){\mathrm{d}}s.\\ \end{aligned}$$
(21)
Taking into account (19) and (21), we deduce that
$$\begin{aligned} A(T)+\liminf _{t\rightarrow \infty }\frac{1}{H(t,T)} \int _T^t \frac{\theta -1}{\theta }H(t,s)\Phi (s)\omega ^2(s){\mathrm{d}}s\le \omega (T), \end{aligned}$$
then
$$A(T)\le \omega (T),\quad \text {for all}\quad T\ge t_2,$$
(22)
and
$$\begin{aligned} \liminf _{t\rightarrow \infty }\frac{1}{H(t,T)} \int _T^t H(t,s)\Phi (s)\omega ^2(s){\mathrm{d}}s\le \frac{\theta }{\theta -1}\left( \omega (T)-A(T)\right) <\infty . \end{aligned}$$
(23)
Now we will prove that
$$\begin{aligned} \lim _{t\rightarrow \infty }\int _{T}^t\Phi (s)\omega ^2(s){\mathrm{d}}s<\infty . \end{aligned}$$
On the contrary, if
$$\begin{aligned} \lim _{t\rightarrow \infty }\int _{T}^t\Phi (s)\omega ^2(s){\mathrm{d}}s=\infty . \end{aligned}$$
(24)
By the condition (20), there exists a positive constant c, such that
$$\begin{aligned} \inf _{s\ge t_0}\left\{ \liminf _{t\rightarrow \infty }\frac{H(t,s)}{H(t,t_0)}\right\}>c>0. \end{aligned}$$
(25)
On the other hand, using (24), for arbitrary positive M, there exists a \(t_3\ge T\), such that
$$\begin{aligned} \int _{T}^t \Phi (s)\omega ^2(s) {\mathrm{d}}s\ge \frac{M}{c}, \quad t\ge t_3. \end{aligned}$$
From the property (ii) of H and (25), we have
$$\begin{aligned} &\frac{1}{H(t,T)} \int _{T}^t H(t,s)\Phi (s)\omega ^2(s) {\mathrm{d}}s\\&\quad =\frac{1}{H(t,T)}\int _{T}^t\left[ -\frac{\partial H(t,s)}{\partial s}\right] \left[ \int _{T}^s \Phi (u)\omega ^2(u){\mathrm{d}}u\right] {\mathrm{d}}s\\&\quad \ge \frac{1}{H(t,T)}\int _{t_3}^t\left[ -\frac{\partial H(t,s)}{\partial s}\right] \left[ \int _{T}^s \Phi (u)\omega ^2(u){\mathrm{d}}u\right] {\mathrm{d}}s\\&\quad \ge \frac{M}{c}\frac{H(t,t_3)}{H(t,T)}\ge \frac{M}{c}\frac{H(t,t_3)}{H(t,t_0)}\ge M.\\ \end{aligned}$$
(26)
Taking account into the fact that M is arbitrary positive constant, (26) implies that
$$\begin{aligned} \lim _{t\rightarrow \infty }\frac{1}{H(t,T)} \int _{T}^t H(t,s)\Phi (s)\omega ^2(s) {\mathrm{d}}s=\infty , \end{aligned}$$
which contradicts (23). Thus,
$$\begin{aligned} \lim _{t\rightarrow \infty }\int _{T}^t\Phi (s)\omega ^2(s){\mathrm{d}}s<\infty . \end{aligned}$$
Using (22) and \(\Phi (t)>0\), we have
$$\begin{aligned} \int _{T}^{\infty }A_{+}^2(t)\Phi (t){\mathrm{d}}t \le \int _{T}^{\infty }\omega ^2(t)\Phi (t){\mathrm{d}}t <\infty , \end{aligned}$$
which contradicts (18).

If \(z<0\), repeating the proof in Theorem 1, we get \(\lim _{t\rightarrow \infty }x(t)=0\). The proof is complete. \(\square\)

Remark 1

Theorem 1–3 and Corollaries 1 and 2 are the oscillation and asymptotic results of (1) under the assumption that \(-1<p(t)\le 0\). However, the results in Liu et al. (2012), Shi et al. (2016) are established in the case where \(0\le p(t)\le 1\). In the hypothesis \((H_5)\), there is another parameter \(\beta\) and the condition on function f in Li et al. (2015), Erbe et al. (2009) does not satisfy \((H_5)\). Therefore, the results of Li et al. (2015), Erbe et al. (2009) can not apply to Eq. (1).

Oscillation of Eq. (1) when \(0\le p(t)\le 1\)

Theorem 4

Assume that \((H_1)\) and \((H_3)-(H_5)\) hold. If there exists a function \(\rho (t)\in {\mathrm{C}}^1([t_0,\infty ),(0,\infty ))\) such that for any positive number M,
$$\begin{aligned} \int _{t_0}^\infty \left[ \rho (t){\overline{p}}(t)- \frac{M^{1-\frac{\beta }{\alpha }}\left( \rho ^{\prime }(t)\right) ^2(r(\sigma (t)))^{\frac{1}{\alpha }}}{4\beta \sigma ^{\prime }(t)\rho (t)(\xi (t))^{\beta -1}}\right] {\mathrm{d}}t =\infty , \end{aligned}$$
(27)
where \({\overline{p}}(t)=q(t)(1-p(\sigma (t)))^\beta\) and \(\xi (t)=\int _{t_1}^t r^{-1/{\alpha }}(s) {\mathrm{d}}s\), then the Eq. (1) is oscillatory.

Proof

See “Appendix”. \(\square\)

Letting \(\rho (t)=1\), we can get the following result.

Corollary 3

Assume that \((H_1)\) and \((H_3)-(H_5)\) hold. If
$$\begin{aligned} \int _{t_0}^\infty q(t)(1-p(\sigma (t)))^\beta {\mathrm{d}}t =\infty , \end{aligned}$$
then the Eq. (1) is oscillatory.

Example 1

Consider the second-order nonlinear neutral delay differential equation
$$\begin{aligned} \left( \left| \left( x(t)+\left( 1-t^{-\frac{1}{2}}\right) x(\tau (t))\right) ^{\prime }\right| ^{\alpha -1}\left[ x(t)+\left( 1-t^{-\frac{1}{2}}\right) x(\tau (t))\right] ^{\prime }\right) ^{\prime }+\frac{\gamma }{t^2}|x(t)|x(t)=0, \end{aligned}$$
(28)
where \(\beta =2, \alpha \ge 2, r(t)=1, p(t)=1-t^{-\frac{1}{2}}, \sigma (t)=t,\; \text {and} \;q(t)=\frac{\gamma }{t^2}\), where \(\gamma\) is a positive constant.
Letting \(\rho (t)=t^2\), we have
$$\begin{aligned} \xi (t)=\int _{t_1}^t r^{-1/{\alpha }}(s) {\mathrm{d}}s=\int _{t_1}^t 1 {\mathrm{d}}s=t-t_1 \end{aligned}$$
and
$$\begin{aligned} &\int _T^t\left[ \rho (t){\overline{p}}(t)- \frac{\left( \rho ^{\prime }(t)\right) ^2(r(\sigma (t)))^{\frac{1}{\alpha }})}{4\beta \sigma ^{\prime }(t)M^{1-\frac{\beta }{\alpha }} \rho (t)(\xi (t))^{\beta -1}}\right] {\mathrm{d}}s= \int _T^t \left[ \frac{\gamma }{s}-\frac{1}{2K^{1-\frac{2}{\alpha }}(s-t_1)}\right] {\mathrm{d}} s\\&\quad =\gamma \ln t-\frac{1}{2M^{1-\frac{2}{\alpha }}}\ln (t-t_1)-\gamma \ln T+\frac{1}{2M^{1-\frac{2}{\alpha }}}\ln (T-t_1). \end{aligned}$$
Therefore, if \(\gamma > \frac{1}{2M^{1-(2/{\alpha })}}\), then
$$\begin{aligned} \int _T^{\infty }\left[ \rho (t){\overline{p}}(t)- \frac{\left( \rho ^{\prime }(t)\right) ^2(r(\sigma (t)))^{\frac{1}{\alpha }})}{4\beta \sigma ^{\prime }(t)M^{1-\frac{\beta }{\alpha }} \rho (t)(\xi (t))^{\beta -1}}\right] {\mathrm{d}}s=\infty . \end{aligned}$$
It follows from Theorem 4 that all solutions of (28) are oscillatory if \(\gamma > \frac{1}{2M^{1-(2/{\alpha })}}\). However, the Eq. (28) is oscillatory when \(\gamma > \frac{1}{M^{1-(2/{\alpha })}}\) from Theorem 1 in Liu et al. (2012). That is, if \(\frac{1}{2M^{1-(2/{\alpha })}}<\gamma \le \frac{1}{M^{1-(2/{\alpha })}}\), then Theorem 1 in Liu et al. (2012) can not apply to (28).

Remark 2

From Theorem 4, the Eq. (28) is oscillatory if \(\gamma > \frac{1}{2M^{1-(2/{\alpha })}}\). However, from Theorem 1 in Liu et al. (2012), when \(\gamma > \frac{1}{M^{1-(2/{\alpha })}}\), the Eq. (28) is oscillation. Thus, Theorem 4 improves Theorem 1 in Liu et al. (2012) in the case where \(0\le p(t)\le 1\).

Examples

In this section, we will present two examples to illustrate the main results.

Example 2

Consider the second-order nonlinear neutral delay differential equation
$$\begin{aligned} \left( \left| \left( x(t)-\frac{1}{2} x(t-2)\right) ^{\prime }\right| ^{\alpha -1}\left[ x(t)-\frac{1}{2} x(t-2)\right] ^{\prime }\right) ^{\prime }+2\left( \frac{e^2}{2}-1\right) ^2|x(t)|^{\beta -1}x(t)=0, \end{aligned}$$
(29)
where \(r(t)=1, p(t)=-1/2, \tau (t)=t-2, \sigma (t)=t,\; \text {and} \;q(t)=2\left( \frac{e^2}{2}-1\right) ^2\).
We see that
$$\begin{aligned} \int _T^{\infty } q(s){\mathrm{d}}s=\infty . \end{aligned}$$
It follows from Corollary 1 that all solutions of (29) are oscillatory or converge to zero. Letting \(\alpha =\beta =2\), we can certify that \(x(t)=e^{-t}\) is an asymptotic solution of (29).

Example 3

Consider the second-order nonlinear neutral delay differential equation
$$\begin{aligned} \left( \left| (x(t)-\lambda _0 x(\tau (t)))^{\prime }\right| ^{\alpha -1}[x(t)-\lambda _0 x(\tau (t))]\right) ^{\prime }+\frac{\gamma }{t^2} |x(\sigma _0t)|^{\beta -1}x(\sigma _0t)=0, \end{aligned}$$
(30)
where \(\alpha >\beta \ge 1\), \(0<\lambda _0<1, 0<\sigma _0<1\), \(r(t)=1, p(t)=\lambda _0, \tau (t)\le t, q(t)=\frac{\gamma }{t^2},\) and \(\sigma (t)=\sigma _0t\).
Letting \(\rho (t)=\frac{\gamma }{2} \frac{\beta \sigma _0\left( t-\frac{t_1}{\sigma _0}\right) ^{\beta -1}}{K^{1-\frac{\beta }{\alpha }}}\) and \(H(t,s)=(t-s)^2\), then
$$\begin{aligned} \xi (t)&= {} \int _{t_1}^t r^{-\frac{1}{\alpha }}(s) {\mathrm{d}}s=t-t_1,\\ \Phi (t)&= {} \frac{\beta \sigma ^{\prime }(t)(\xi (\sigma (t)))^{\beta -1}}{K^{1-\frac{\beta }{\alpha }}\rho (t)(r(\sigma (t)))^{\frac{1}{\alpha }}} =\frac{\beta \sigma _0(t-\frac{t_1}{\sigma _0})^{\beta -1}}{K^{1-\frac{\beta }{\alpha }}\rho (t)}=\frac{2}{\gamma } , \end{aligned}$$
and
$$\begin{aligned} \limsup _{t\rightarrow \infty }\frac{1}{(t-s)^2}\int _{t_0}^t\left[ (t-s)^2q(s)-\frac{1}{\Phi (s)}\right] {\mathrm{d}}s=\limsup _{t\rightarrow \infty }\frac{1}{(t-s)^2}\int _{t_0}^t\left[ \gamma (\frac{1}{2}-\frac{2t}{s}+\frac{t^2}{s^2})\right] {\mathrm{d}}s=\infty . \end{aligned}$$
That is, (17) holds. It follows from Corollary 2 that all solutions of (30) are oscillatory or converge to zero.

Remark 3

If \(\alpha >\beta\), the results of Erbe et al. (2009) and Li et al. (2015) can not apply to (29) and (30). The results of Liu et al. (2012) and Shi et al. (2016) also can not apply to (29) and (30) because \(-1< p(t)\le 0\) which does not satisfy the assumptions in Liu et al. (2012), Shi et al. (2016).

Conclusion

In this paper, we consider the oscillation of a class of second-order differential equations with positive and nonpositive neutral coefficients. It is difficult to study the nonpositive neutral coefficients equations because the sign of z is not explicit. Using Riccati transformation, some oscillation and asymptotic criteria are obtained under the assumptions that \((H_1)-(H_5)\). In Liu et al. (2012), Shi et al. (2016), the results were established for (1) in the case when \(0\le p(t)\le 1\) or \(p(t)>1\). This paper states some oscillation and asymptotic criteria for (1) in the case where \(-1<p(t)<0\) and \(0\le p(t)\le 1\). Erbe et al. (2009), Li et al. (2015) assume that \(\alpha =\beta\), however, in this paper \(\alpha \ne \beta\) is allowed. We give some examples to illustrate our results. There are two interesting questions for future study:
  1. (Q1)

    Could we obtain oscillation criteria for Eq. (1) when \(-\infty < p(t)\le 1\, \text{or} -\infty<p(t)\le-1\)?

     
  2. (Q2)

    Could we obtain some sufficient conditions which ensure that all solutions of (1) are oscillatory?

     

Declarations

Authors’ contributions

Both authors contributed equally to this paper. Both authors read and approved the final manuscript.

Acknowledgements

This work is supported by NSF of Shandong Province, China (Grant No: ZR2014FM036). The authors would like to thank the reviewers for the valuable comments.

Competing interests

Both authors declare that they have no competing interests.

Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Authors’ Affiliations

(1)
School of Control Science and Engineering, Shandong University

References

  1. Baculíková B, Džurina J (2011) Oscillation theorems for second order neutral differential equations. Comput Math Appl 61:94–99MathSciNetView ArticleMATHGoogle Scholar
  2. Berkovich LM (1997) The generalized Emden–Fowler equation. Symmetry Nonlinear Math Phys 1:155–163MathSciNetMATHGoogle Scholar
  3. Bohner M, Li T (2014) Oscillation of second-order \(p\)-Laplace dynamic equations with a nonpositive neutral coefficient. Appl Math Lett 37:72–76MathSciNetView ArticleMATHGoogle Scholar
  4. Brayton RK (1966) Bifurcation of periodic solutions in a nonlinear difference-differential equation of neutral type. Q Appl Math 24:215–224MathSciNetMATHGoogle Scholar
  5. Džurina J, Stavroulakis IP (2003) Oscillation criteria for second-order delay differential equations. Appl Math Comput 140:445–453MathSciNetMATHGoogle Scholar
  6. Erbe L, Hassan TS, Peterson A (2009) Oscillation criteria for nonlinear functional neutral dynamic equations on time scales. J Differ Equ Appl 15:1097–1116MathSciNetView ArticleMATHGoogle Scholar
  7. Hale JK (1977) Theory of functional differential equations. Springer, New YorkView ArticleMATHGoogle Scholar
  8. Kuang Y (1993) Delay differential equations with application in population dynamics. Academic Press, New YorkMATHGoogle Scholar
  9. Li Q, Wang R, Chen F, Li T (2015) Oscillation of second-order nonlinear delay differential equations with nonpositive neutral coefficients. Adv Differ Equ 35:7MathSciNetMATHGoogle Scholar
  10. Li T, Han Z, Zhang C, Sun S (2011) On the oscillation of second-order Emden–Fowler neutral differential equations. J Appl Math Comput 37:601–610MathSciNetView ArticleMATHGoogle Scholar
  11. Li T, Rogovchenko YuV, Zhang C (2013) Oscillation results for second-order nonlinear neutral differential equations. Adv Differ Equ 336:1–13MathSciNetGoogle Scholar
  12. Li T, Rogovchenko YuV (2015) Oscillation of second-order neutral differential equations. Math Nachr 288:1150–1162MathSciNetView ArticleMATHGoogle Scholar
  13. Li W (2004) Interval oscillation of second-order half-linear functional differential equations. Appl Math Comput 155:451–468MathSciNetMATHGoogle Scholar
  14. Liu H, Meng F, Liu P (2012) Oscillation and asymptotic analysis on a new generalized Emden–Fowler equation. Appl Math Comput 219:2739–2748MathSciNetMATHGoogle Scholar
  15. Manojlović JV (1999) Oscillation criteria for second-order half-linear differential equations. Math Comput Model 30:109–119MathSciNetView ArticleMATHGoogle Scholar
  16. Shi Y, Han Z, Sun Y (2016) Oscillation criteria for a generalized Emden–Fowler dynamic equation on time scales. Adv Differ Equ 3:1–12MathSciNetGoogle Scholar
  17. Wang Q, Yang Q (2004) Interval criteria for oscillation of second-order half-linear differential equations. J Math Anal Appl 291:224–236MathSciNetView ArticleMATHGoogle Scholar
  18. Wang Q (2001) Oscillation and asymptotics for second-order half-linear differential equations. Appl Math Comput 122:253–266MathSciNetMATHGoogle Scholar
  19. Xu R, Meng F (2006) Some new oscillation criteria for second order quasi-linear neutral delay differential equations. Appl Math Comput 182:797–803MathSciNetMATHGoogle Scholar
  20. Xu R, Meng F (2007) Oscillation criteria for second order quasi-linear neutral delay differential equations. Appl Math Comput 192:216–222MathSciNetMATHGoogle Scholar
  21. Yang Q, Yang L, Zhu S (2006) Interval criteria for oscillation of second order nonlinear neutral differential equations. Comput Math Appl 46:903–918MathSciNetView ArticleMATHGoogle Scholar

Copyright

© The Author(s) 2016