数学物理学报  2016, Vol. 36 Issue (3): 462-472   PDF (362 KB)    
扩展功能
加入收藏夹
复制引文信息
加入引用管理器
Email Alert
RSS
本文作者相关文章
马晴霞
刘安平
脉冲时滞中立双曲型方程组的振动性
马晴霞1,2, 刘安平2    
1 中国科学院武汉物理与数学研究所 武汉 430071;
2 中国地质大学(武汉)数学与物理学院 武汉 430074
摘要: 研究一类非线性脉冲时滞双曲型方程组在Robin以及Dirichlet边界条件下的振动性质. 利用广义的Riccati变换、平均值方法及不等式技巧,获得了方程组在两类边界条件下振动的充分条件,并给出了应用实例用以检验结论的有效性.
关键词: 非线性     脉冲     时滞     偏微分方程组     振动    
Oscillation of Neutral Impulsive Hyperbolic Systems with Deviating Arguments
Ma Qingxia1,2, Liu Anping2    
1 Wuhan Institute of Physics and Mathematics, Chinese Academy of Sciences, Wuhan 430071;
2 School of Mathematics and Physics, China University of Geosciences (Wuhan), Wuhan 430074
Abstract: Oscillatory properties of systems of neutral type impulsive hyperbolic equations with several deviating arguments under the Robin boundary condition and the Dirichlet boundary condition are investigated, and several new sufficient conditions for oscillation are presented. The main results are illustrated by one example.
Key words: Nonlinear     Impulsive     Delay     Partial differential systems     Oscillation    
1 引言

许多自然科学和社会科学中的实际问题可以由偏微分方程来描述,例如固体的颤动,人口的增长,疾病的传播,电磁波的放射等.对这些现象的研究,人们总假设状态的变化和系统参数的改变是连续的.但是,我们容易想到某些突然出现的自然现象,比如电路系统中开关的闭合、机械钟摆的运动、电流中的脉冲、种群生态系统的定时捕捞或补给、药剂学中的定时给药的过程、地震、海啸及烈性传染病SARS等灾难的出现使得所述的系统产生了跳跃,也就是出现了脉冲.脉冲现象在现代科技各领域的实际问题中是普遍存在的,其数学模型往往可归结为脉冲微分系统.脉冲微分系统最突出的特点是能够充分考虑到瞬时突变现象对状态的影响,能更深刻、更精确地反映事物的变化规律.近年来,人们发现这类系统在航天技术、信息科学、控制系统、通讯、生命科学、医学,经济领域均有重要应用.大量的学者对脉冲微分方程的定性理论(如脉冲常微分方程或偏微分方程解的存在性、唯一性、稳定性及振动性等方面)产生了极大的兴趣,取得了一系列丰硕的成果[1, 2].

另一方面,振动理论是微分方程定性理论的一个重要的分支,也是近年来定性理论研究中一个比较活跃的方向.在自然界、工业领域、日常生活及社会生活中,都会碰到各种形式的振动现象,如机械的振动,无线电中的电磁震荡等.广泛的应用背景使得微分方程振动理论的研究在近三十年取得了丰硕的成果[3, 4],但具有时滞和脉冲的偏微分方程研究却进展缓慢,这主要是由于脉冲和时滞同时出现,使得系统呈现出新的特征,给研究带来了困难.因此关于脉冲时滞偏微分系统的定性理论之一的振动性问题有待进一步的探讨,需建立其特有的理论体系.

本文研究如下非线性脉冲时滞双曲型方程组

$\begin{align} & \frac{{{\partial }^{2}}}{\partial {{t}^{2}}}[{{u}_{i}}(x,t)+\sum\limits_{s=1}^{l}{{{c}_{s}}}(t){{u}_{i}}(x,t-{{\mu }_{s}})]+\sum\limits_{r=1}^{n}{{{b}_{ir}}}(x,t){{f}_{ir}}({{u}_{r}}(x,t-\sigma))\\ &={{a}_{i}}(t){{h}_{i}}({{u}_{i}}(x,t))\Delta {{u}_{i}}(x,t)+\sum\limits_{j=1}^{m}{{{a}_{ij}}}(t){{h}_{ij}}({{u}_{i}}(x,t-{{\tau }_{j}}))\Delta {{u}_{i}}(x,t-{{\tau }_{j}}),\ \\ & \ t\ne {{t}_{k}},(x,t)\in \Omega \times {{\mathbb{R}}^{+}}=G,\\ \end{align}$ (1.1)
$\begin{equation}\label{eq:a2} u_i(x,t^+_k)=p_{ki}(x,t_k,u_i(x,t_k)),\ \ k=1,2,\cdots, \end{equation}$ (1.2)
$\begin{equation}\label{eq:a3} u_{it}(x,t^+_k)=q_{ki}(x,t_k,u_{it}(x,t_k)),\ \ k=1,2,\cdots \end{equation}$ (1.3)
在两类边界条件

$\frac{\partial {{u}_{i}}(x,t)}{\partial \nu }+{{\beta }_{i}}(x,t){{u}_{i}}(x,t)=0,\;\;(x,t)\in \partial \Omega \times {{\mathbb{R}}^{+}},$ (1.4)
${{u}_{i}}(x,t)=0,\;\;(x,t)\in \partial \Omega \times {{\mathbb{R}}^{+}},\ i\in {{I}_{n}}=\{1,2,\cdots,n\}$ (1.5)
及初始条件

$\begin{equation}\label{eq:a6} u_i(x,t)=\Phi_i(x,t),\ \frac{\partial u_i(x,t)}{\partial t}=\Psi_i(x,t),(x,t)\in\Omega\times[-\delta,0] \end{equation}$ (1.6)
下的振动性.其中$\ \Omega\subset {\Bbb R}^N$表示具有逐片光滑边界$\partial\Omega$的有界区域,$\nu$表示边界$\partial\Omega$的单位外法向量,$u_{it}(x,t)=\frac{\partial u_i(x,t)}{\partial t},\ {\Bbb R}^+=[0,+\infty),$ $ \beta_i(x,t)\in PC(\partial\Omega\times {\Bbb R}^+,{\Bbb R}^+),PC$表示以$t=t_k$ $(k\in I_\infty)$为第一类间断点且在该处左连续的分段连续函数类.$\delta=\max\{\mu_s,\: \sigma,\ \tau_j\},\ \Phi_i(x,t)\in C^2(\Omega\times[-\delta,0],{\Bbb R}),\ \Psi_i(x,t)\in C^1(\Omega\times[-\delta,0],{\Bbb R}),\ s\in I_l,j\in I_m,i\in I_n$.

我们假设下列条件均成立:

${\rm \widetilde{H}}_1)$ $0<t_1<t_2<\cdots <t_k<\cdots,\lim\limits_{k\rightarrow\infty}t_k=\infty.\ \mu_s,\ \sigma,\tau_j$均为非负常数,$s\in I_l,j\in I_m.$

${\rm \widetilde{H}}_2)$ $a_i(t),a_{ij}(t)\in PC({\Bbb R}^+,{\Bbb R}^+),b_{ir}(x,t)\in PC(\overline{\Omega}\times{\Bbb R}^+,{\Bbb R}),$ $b_{ii}(x,t)>0,$ $ c_i(t)\in C^2({\Bbb R}^+,$ ${\Bbb R}^+),$ $ \sum\limits^l_{s=1}c_s(t)<1,r\in I_n,j\in I_m,i\in I_n.$

${\rm \widetilde{H}}_3)$ $f_{ir}(u_r)\in PC({\Bbb R},{\Bbb R})$,当$u_r\neq0$时,$f_{ir}(u_r)/u_r\geq C_{ir},C_{ir}$为正常数;$f_{ir}(-u_r)=-f_{ir}(u_r),$ $r\in I_n,i\in I_n.$

${\rm \widetilde{H}}_4)$ $h_i(u_i),h_{ij}(u_i)\in C^1({\Bbb R},{\Bbb R})$,当$u\neq 0$时,$ h_i(u_i)> 0,$ $h_{ij}(u_i)> 0;$ $ u_ih'_i(u_i)\geq 0,$ $ u_ih'_{ij}(u_i)\geq 0,$ $ h_i(0)=0,\ h_{ij}(0)=0,$ $ j\in I_m,i\in I_n.$

${\rm \widetilde{H}}_5)$ $p_{ki}(x,t_k,u_i(x,t_k)),q_{ki}(x,t_k,u_{it}(x,t_k))\in PC(\bar{\Omega}\times{\Bbb R}^+\times{\Bbb R},{\Bbb R}),$存在正常数$\overline{a}_{ki},$ $\underline{a}_{ki},$ $ \overline{b}_{ki},$ $ \underline{b}_{ki},$使得$\xi\neq 0,\eta\neq 0,k\in I_{\infty},$

$$ \min\limits_{i\in I_n}\{\underline{a}_{ki}\}=\underline{a}_k\leq\underline{a}_{ki}\leq\frac{p_{ki}(x,t_k,\xi)}{\xi}\leq\overline{a}_{ki}\leq \overline{a}_k=\max\limits_{i\in I_n}\{\overline{a}_{ki}\}, $$ $$ \min\limits_{i\in I_n}\{\underline{b}_{ki} \}=\underline{b}_k\leq\underline{b}_{ki}\leq\frac{q_{ki}(x,t_k,\eta)}{\eta}\leq\overline{b}_{ki}\leq \overline{b}_k=\max\limits_{i\in I_n}\{\overline{b}_{ki}\}, $$

其中$\overline{b}_k\leq\underline{a}_k$.

构造序列$\{\bar{t}_k\}=\{t_k\}\cup\{t_{ks}\}\cup\{t_{k\sigma}\}\cup\{t_{kj}\}$,其中$t_{ks}=t_k+\mu_s,\ t_{k\sigma}=t_k+\sigma,\ t_{kj}=t_k+\tau_j$,且$\bar{t}_k<\bar{t}_{k+1},\ s\in I_l,\ j\in I_m,\ k\in I_\infty.$

引入以下记号

$$ b_{ii}(t)=\min\limits_{x\in\overline\Omega}{b_{ii}(x,t)},\:\:\: \overline{b_{ir}(t)}=\max\limits_{x\in\overline\Omega}{|b_{ir}(x,t)|}, $$ $$ B(t)=\min\limits_{i\in I_n}\bigg\{b_{ii}(t)C_{ii}-\sum\limits_{r\neq i} \overline{b_{ri}(t)}C_{ri}\bigg\}>0,\ r\in I_n,\ i\in I_n. $$ $$ z(t)=U(t)+\sum\limits^l_{s=1}c_s(t)U(t-\mu_s),\ U(t)=\sum\limits_{i=1}^nU_i(t), $$ $$ U_i(t)=\int_\Omega Z_i(x,t){\rm d}x,\: Z_i(x,t)=\gamma_iu_i(x,t),\ \gamma_i={\rm sgn} u_i(x,t). $$

定义1.1 称向量值函数$u(x,t)=\{u_1(x,t),\ u_2(x,t),\ \cdots,\ u_n(x,t)\}$是问题(1.1)-(1.4)或问题(1.1)-(1.3),(1.5)的解,如果以下条件成立:

1) 当$-\delta\leq t\leq0$时,$u_i(x,t)=\Phi_i(x,t),\ \frac{\partial u_i(x,t)}{\partial t}=\Psi_i(x,t)$;

2) 当$0\leq t\leq \bar{t}_1=t_1$时,$u_i(x,t)$与问题(1.1)-(1.4)或问题(1.1)-(1.3),(1.5)的解一致;

3) 当$\bar{t}_k<t\leq\bar{t}_{k+1},\ \bar{t}_k\in\{t_k\}\setminus(\{t_{ks}\}\cup\{t_{k\sigma}\}\cup\{t_{kj}\})$时,$u_i(x,t)$与问题(1.1)-(1.4)或问题(1.1)-(1.3),(1.5)的解一致;

4) 当$\bar{t}_k<t\leq\bar{t}_{k+1},\ \bar{t}_k\in\{t_{ks}\}\cup\{t_{k\sigma}\}\cup\{t_{kj}\}$时,$u_i(x,t)$满足(1.4)或(1.5)并且与下列问题的解一致

$\begin{align} & \frac{{{\partial }^{2}}}{\partial {{t}^{2}}}[{{u}_{i}}(x,{{t}^{+}})+\sum\limits_{s=1}^{l}{{{c}_{s}}}(t){{u}_{i}}(x,{{(t-{{\mu }_{s}})}^{+}})]+\sum\limits_{r=1}^{n}{{{b}_{ir}}}(x,t){{f}_{ir}}({{u}_{r}}(x,{{(t-\sigma)}^{+}}))\\ &={{a}_{i}}(t){{h}_{i}}({{u}_{i}}(x,{{t}^{+}}))\Delta {{u}_{i}}(x,{{t}^{+}})+\sum\limits_{j=1}^{m}{{{a}_{ij}}}(t){{h}_{ij}}({{u}_{i}}(x,{{(t-{{\tau }_{j}})}^{+}}))\Delta {{u}_{i}}(x,{{(t-{{\tau }_{j}})}^{+}}),\ \\ & \ t\ne {{t}_{k}},(x,t)\in \Omega \times {{\mathbb{R}}^{+}}=G,\\ \end{align}$
$$ u_i(x,\bar{t}^+_k)=u_i(x,\bar{t}_k),\ \ u_{it}(x,\bar{t}^+_k)=u_{it}(x,\bar{t}_k), \:\bar{t}_k\in(\{t_{ks}\}\cup\{t_{k\sigma}\}\cup\{t_{kj}\})\setminus\{t_k\} $$ 或者

$$ u_i(x,\bar{t}^+_k)=p_{k_s}(x,\bar{t}_k,u_i(x,\bar{t}_k)),\ \ u_{it}(x,\bar{t}^+_k)=q_{k_s}(x,\bar{t}_k,u_{it}(x,\bar{t}_k)), $$ $$ \bar{t}_k\in(\{t_{ks}\}\cup\{t_{k\sigma}\}\cup\{t_{kj}\})\cap\{t_k\}, $$ 其中$k_s$由等式$\bar{t}_k=t_{k_s}$确定.

定义1.2 称向量值函数$u(x,t)=\{u_1(x,t),\ u_2(x,t),\ \cdots,\ u_n(x,t)\}$的非平凡分量$u_i(x,t)$ $(i\in I_n)$在区域$G$内非振动,如果它是最终正或最终负;否则称为振动的.

定义1.3 称问题(1.1)-(1.4)或问题(1.1)-(1.3),(1.5)的解向量$u(x,t)=\{u_1(x,t),$ $ u_2(x,t),$ $ \cdots,u_n(x,t)\}$在区域$G$内振动,如果至少有一个非平凡的分量是振动的.

以下将给出一个重要的引理,用来估计一阶脉冲微分不等式解的范围.我们仅给出引理的叙述,有兴趣的读者可以参考文献[1].

引理1.1 假设以下条件成立:

${\rm A}_1)$ 序列$\{t_k\}$满足$0\le {{t}_{0}}<{{t}_{1}}<{{t}_{2}}<\cdots ,\ \underset{k\to \infty }{\mathop{\lim }}\,{{t}_{k}}=\infty $;

${\rm A}_2)$ $m(t)\in PC^1({\Bbb R}^+,{\Bbb R})$在点列$\{t_k\}\:(k=1,2,\cdots)$处左连续;

${\rm A}_3)$ $k=1,2,\cdots $且$t\geq t_0$时,

$$ m'(t)\leq p(t)m(t)+q(t),\ \ t\neq t_k, $$ $$ m(t^+_k)\leq d_km(t_k)+e_k $$ 成立,其中$p(t),q(t)\in C({\Bbb R}^+,{\Bbb R}),\ d_k\geq0,\ e_k$是常数.则

$\begin{array}{l} m(t) \le m({t_0})\prod\limits_{{t_0} < {t_k} < t} {{d_k}} \exp \left( {\int_{{t_0}}^t p (s){\rm{d}}s} \right) + \int_{{t_0}}^t {\prod\limits_{s < {t_k} < t} {{d_k}} } \exp \left( {\int_s^t p (r){\rm{d}}r} \right)q(s){\rm{d}}s\\ \;\;\;\;\;\;\;\;\;\; + \;\sum\limits_{{t_0} < {t_k} < t} {\prod\limits_{{t_k} < {t_j} < t} {{d_j}} } \exp \left( {\int_{{t_k}}^t p (s){\rm{d}}s} \right){e_k}. \end{array}$
2 问题(1.1)-(1.4)的振动性

引理2.1 假设向量值函数$u(x,t)=\{u_1(x,t),\ u_2(x,t),\ \cdots,\ u_n(x,t)\}$是问题(1.1)-(1.4)在区域G上的一个非振动解,则函数$z(t)$满足以下脉冲微分不等式

$ \begin{equation}\label{eq:b1} z''(t)+\bigg(1-\sum\limits^l_{s=1}c_s(t-\sigma)\bigg)B(t)z(t-\sigma)\leq0,\ \ t\neq t_k, \end{equation} $ (2.1)
$\begin{equation}\label{eq:b2} \underline{a}_k\leq\frac{z(t_k^+)}{z(t_k)}\leq\overline{a}_k,\end{equation}$ (2.2)
$ \begin{equation}\label{eq:b3} \underline{b}_k\leq\frac{z'(t_k^+)}{z'(t_k)}\leq\overline{b}_k,\ \ k\in I_{\infty}.\end{equation} $ (2.3)

假设$u(x,t)=\{u_1(x,t),\ u_2(x,t),\ \cdots,\ u_n(x,t)\}$是问题(1.1)-(1.4)在区域$G$上的一个非振动解.不失一般性,假定存在$\overline{t_0}>0,$使得当$(x,t)\in\Omega\times[\overline{t_0},+\infty)$时,${\rm sgn}u_i(x,t)\neq 0,i \in I_n$.

因为$Z_i(x,t)=\gamma_iu_i(x,t),\gamma_i={\rm sgn}u_i(x,t),$则$ Z_i(x,t)>0.$根据条件${\rm \widetilde{H}}_1)$,知必存在$t_0\geq \overline{t_0}$,使得当$(x,t)\in \Omega\times[t_0,+\infty)$时,$Z_i(x,t-\mu_s)>0,Z_r(x,t-\sigma)>0,Z_i(x,t-\tau_j)>0,\ s\in I_l,r\in I_n,j\in I_m,\ i\in I_n$.

当$t\geq t_0,\ t\neq t_k(k\in I_{\infty})$时,(1.1)式两边同时在区域$\Omega$上关于变量$x$积分,得

$\begin{align} & \frac{{{\text{d}}^{2}}}{\text{d}{{t}^{2}}}[\int_{\Omega }{{{Z}_{i}}}(x,t)\text{d}x+\sum\limits_{s=1}^{l}{{{c}_{s}}}(t)\int_{\Omega }{{{Z}_{i}}}(x,t-{{\mu }_{s}})\text{d}x] \\ &+\sum\limits_{r=1}^{n}{\frac{{{\gamma }_{i}}}{{{\gamma }_{r}}}}\int_{\Omega }{{{b}_{ir}}}(x,t){{f}_{ir}}({{Z}_{r}}(x,t-\sigma))\text{d}x \\ &={{a}_{i}}(t)\int_{\Omega }{{{h}_{i}}}({{u}_{i}}(x,t))\Delta {{Z}_{i}}(x,t)\text{d}x \\ &+\sum\limits_{j=1}^{m}{{{a}_{ij}}}(t)\int_{\Omega }{{{h}_{ij}}}({{u}_{i}}(x,t-{{\tau }_{j}}))\Delta {{Z}_{i}}(x,t-{{\tau }_{j}})\text{d}x,\ i\in {{I}_{n}}.\\ \end{align}$
应用Green公式、边界条件(1.4)及条件${\rm \widetilde{H}}_4)$,得

$\begin{align} & \int_{\Omega }{{{h}_{i}}}({{u}_{i}}(x,t))\Delta {{Z}_{i}}(x,t)\text{d}x=\int_{\partial \Omega }{{{h}_{i}}}({{u}_{i}})\frac{\partial {{Z}_{i}}}{\partial \nu }\text{d}S-\int_{\Omega }{\nabla }{{h}_{i}}({{u}_{i}})\cdot \nabla {{Z}_{i}}\text{d}x \\ & \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ =-{{\gamma }_{i}}\int_{\partial \Omega }{{{\beta }_{i}}}(x,t){{u}_{i}}{{h}_{i}}({{u}_{i}})\text{d}S-{{\gamma }_{i}}\int_{\Omega }{{{h}_{{{i}'}}}}({{u}_{i}})|\text{grad}u{{|}^{2}}\text{d}x \\ & \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \le 0,\\ \end{align}$
$$ \int_\Omega h_{ij}(u_i(x,t-\tau_{j}))\Delta Z_i(x,t-\tau_{j}){\rm d}x\leq0,\ j\in I_m,i\in I_n, $$ 这里${\rm d}S$为边界$\partial\Omega$的面积元素.由条件${\rm \widetilde{H}}_3)$,可得

$$ \int_{\Omega}b_{ir}(x,t)f_{ir}(Z_r(x,t-\sigma)){\rm d}x\geq C_{ir}\int_{\Omega}b_{ir}(x,t)Z_r(x,t-\sigma){\rm d}x,\ r\in I_n,i\in I_n. $$

因$\:U_i(t)=\int_{\Omega}Z_i(x,t){\rm d}x$,很明显$U_i(t)>0$,且

$\begin{equation}\label{eq:b4} \bigg[U_i(t)+\sum\limits^l_{s=1}c_s(t)U_i(t-\mu_s)\bigg]''+b_{ii}(t)C_{ii}U_i(t-\sigma)-\sum\limits_{r\neq i}\overline{b_{ir}(t)}C_{ir}U_r(t-\sigma)\leq0,\ i\in I_n. \end{equation}$ (2.4)

因$\:U(t)=\sum\limits_{i=1}^n U_i(t),$有$U(t)>0,t\geq t_0.$将(2.4)式中的$n$个不等式垂直相加,可得

$\begin{align} & 0\ge {{[U(t)+\sum\limits_{s=1}^{l}{{{c}_{s}}}(t)U(t-{{\mu }_{s}})]}^{\prime \prime }}+\sum\limits_{i=1}^{n}{[}{{b}_{ii}}(t){{C}_{ii}}{{U}_{i}}(t-\sigma)-\sum\limits_{r\ne i}{\overline{{{b}_{ir}}(t)}}{{C}_{ir}}{{U}_{r}}(t-\sigma)] \\ & ={{[U(t)+\sum\limits_{s=1}^{l}{{{c}_{s}}}(t)U(t-{{\mu }_{s}})]}^{\prime \prime }}+\sum\limits_{i=1}^{n}{[}{{b}_{ii}}(t){{C}_{ii}}-\sum\limits_{r\ne i}{\overline{{{b}_{ri}}(t)}}{{C}_{ri}}]{{U}_{i}}(t-\sigma)\\ & \ge {{[U(t)+\sum\limits_{s=1}^{l}{{{c}_{s}}}(t)U(t-{{\mu }_{s}})]}^{\prime \prime }}+\underset{i\in {{I}_{n}}}{\mathop{\min }}\,[{{b}_{ii}}(t){{C}_{ii}}-\sum\limits_{r\ne i}{\overline{{{b}_{ri}}(t)}}{{C}_{ri}}]\sum\limits_{i=1}^{n}{{{U}_{i}}}(t-\sigma)\\ &={{[U(t)+\sum\limits_{s=1}^{l}{{{c}_{s}}}(t)U(t-{{\mu }_{s}})]}^{\prime \prime }}+B(t)U(t-\sigma),\\ \end{align}$

$\begin{equation}\label{eq:b5} \bigg [U(t)+\sum\limits^l_{s=1}c_s(t)U(t-\mu_s)\bigg]''+B(t)U(t-\sigma)\leq 0,\ t\geq t_0. \end{equation}$ (2.5)

因$z(t)=U(t)+\sum\limits^l_{s=1}c_s(t)U(t-\mu_s),$则

$$ z(t)>0,\quad z(t)\geq U(t). $$

由(2.5)式知$z''(t)\leq 0$且

$\begin{equation}\label{eq:b6} z'(t)>0. \end{equation}$ (2.6)

事实上,如果(2.6)式不成立,则存在$t_1\geq t_0$,满足$z'(t_1)<0.$因为$z'(t)$单调减,则

$$ z(t)-z(t_1)=\int^t_{t_1}z'(s){\rm d}s\leq \int^t_{t_1}z'(t_1){\rm d}s=z'(t_1)(t-t_1), $$

且$\lim\limits_{t\rightarrow+\infty}z(t)=-\infty,$此结论与$z(t)>0$矛盾,所以(2.6)式成立.

由(2.5)式可得

$$ z''(t)+B(t)U(t-\sigma)\leq0,\ t\geq t_0. $$

因为

$\begin{align} & U(t)=z(t)-\sum\limits_{s=1}^{l}{{{c}_{s}}}(t)U(t-{{\mu }_{s}})\\ &=z(t)-\sum\limits_{s=1}^{l}{{{c}_{s}}}(t)[z(t-{{\mu }_{s}})-\sum\limits_{s=1}^{l}{{{c}_{s}}}(t-{{\mu }_{s}})U(t-2{{\mu }_{s}})] \\ &=z(t)-\sum\limits_{s=1}^{l}{{{c}_{s}}}(t)z(t-{{\mu }_{s}})+\sum\limits_{s=1}^{l}{{{c}_{s}}}(t)[\sum\limits_{s=1}^{l}{{{c}_{s}}}(t-{{\mu }_{s}})U(t-2{{\mu }_{s}})] \\ & \ge z(t)-\sum\limits_{s=1}^{l}{{{c}_{s}}}(t)z(t-{{\mu }_{s}})\ge(1-\sum\limits_{s=1}^{l}{{{c}_{s}}}(t))z(t),\\ \end{align}$

所以

$$ U(t-\sigma)\geq \bigg(1-\sum\limits^l_{s=1}c_s(t-\sigma)\bigg)z(t-\sigma). $$

于是,我们得到

$$ z''(t)+\bigg(1-\sum\limits^l_{s=1}c_s(t-\sigma)\bigg)B(t)z(t-\sigma)\leq0,\ t\geq t_0. $$

当$t\geq t_0,t=t_k(k\in I_{\infty})$时,有

$\begin{array}{*{35}{l}} {{{\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{a}}}_{k}}\le \frac{{{Z}_{i}}(x,t_{k}^{+})}{{{Z}_{i}}(x,{{t}_{k}})}=\frac{{{\gamma }_{i}}{{u}_{i}}(x,t_{k}^{+})}{{{\gamma }_{i}}{{u}_{i}}(x,{{t}_{k}})}=\frac{{{p}_{ki}}(x,{{t}_{k}},{{u}_{i}}(x,{{t}_{k}}))}{{{u}_{i}}(x,{{t}_{k}})}\le {{{\bar{a}}}_{k}},\\ {{{\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{b}}}_{k}}\le \frac{{{Z}_{it}}(x,t_{k}^{+})}{{{Z}_{it}}(x,{{t}_{k}})}=\frac{{{\gamma }_{i}}{{u}_{it}}(x,t_{k}^{+})}{{{\gamma }_{i}}{{u}_{it}}(x,{{t}_{k}})}=\frac{{{q}_{ki}}(x,{{t}_{k}},{{u}_{it}}(x,{{t}_{k}}))}{{{u}_{it}}(x,{{t}_{k}})}\le {{{\bar{b}}}_{k}}.\\ \end{array}$ (2.7)

由(2.7)式可得

$$ \underline{a}_k\leq \frac{U_i(t_k^+)}{U_i(t_k)}\leq\overline{a}_k,\ \ \ \ \underline{a}_k\leq \frac{U(t_k^+)}{U(t_k)}\leq\overline{a}_k, $$ $$ \underline{b}_k\leq \frac{{U_i}'(t_k^+)}{{U_i}'(t_k)}\leq\overline{b}_k,\ \ \ \ \underline{b}_k\leq \frac{U'(t_k^+)}{U'(t_k)}\leq\overline{b}_k, $$

$$ \underline{a}_k\leq \frac{z(t_k^+)}{z(t_k)}\leq\overline{a}_k,\ \ \ \ \underline{b}_k\leq \frac{z'(t_k^+)}{z'(t_k)}\leq\overline{b}_k. $$

所以$z(t)$是脉冲微分不等式(1.1)-(1.4)的正解.证毕.

引理2.2 假设$z(t)$是脉冲微分不等式(2.1)-(2.3)的正(负)解,即假设存在$T\geq t_0$使得当$t\geq T$时,$z(t)>0(z(t)<0)$.如果

$\mathop {\lim }\limits_{t \to + \infty } \int_{{t_0}}^t {\prod\limits_{{t_0} < {t_k} < \eta } {\frac{{{{}_k}}}{{{{\bar b}_k}}}} } F(\eta ){\rm{d}}\eta = + \infty $ (2.8)

成立,则当$t\in[T,t_l]\bigcup\Big(\bigcup\limits^{+\infty}_{k=l}(t_k,t_{k+1}]\Big)$时,$z'(t)\geq0 (z'(t)\leq 0)$,其中$l=\min\{k:t_k\geq T\}$.

不妨假设$z(t)$是脉冲微分不等式(2.1)-(2.3)的正解,则当$t\geq T>t_0$时,$z(t)>0$.我们首先证明对任意的$k\geq l$,都有

$\begin{equation}\label{eq:b9} z'(t_k)\geq 0. \end{equation}$ (2.9)

如果(2.9)式不成立,则存在某个$j$,使得$j\geq l$且$z'(t_j)<0.$由(2.3)式,可得

$$ z'(t_j^+)\leq\underline{b}_j z'(t_j)<0. $$

设$z'(t_j^+)=-\beta<0.$由(2.1)式知,$z'(t)$在区间$(t_{j+i-1},t_{j+i}](i=1,2,\cdots)$内单调递减,所以

$$ z'(t_{j+1})\leq z'(t_j^+)=-\beta<0, $$ $$ z'(t_{j+2})\leq z'(t_{j+1}^+)\leq\underline{b}_{j+1}z'(t_{j+1})\leq-\underline{b}_{j+1}\beta<0, $$ $$ z'(t_{j+3})\leq z'(t_{j+2}^+)\leq\underline{b}_{j+2}z'(t_{j+2})\leq-\underline{b}_{j+2}\underline{b}_{j+1}\beta<0. $$

由归纳法可得,对任意的正整数$n\geq 2$,有

$$ z'(t_{j+n})\leq-\prod\limits_{i=1}^{n-1}\underline{b}_{j+i}\beta<0. $$

考虑以下脉冲微分不等式

$$ z''(t)\leq 0,t>t_j,t\neq t_k, $$ $$ z'(t_k^+)\leq \underline{b}_kz'(t_k),k=j+1,j+2,\cdots. $$

令$m(t)=z'(t)$,则

$$ m'(t)\leq 0,t>t_j,t\neq t_k, $$ $$ m(t_k^+)\leq \underline{b}_km(t_k),k=j+1,j+2,\cdots. $$

根据引理1.1,可得

$m(t)\leq m(t_j^+)\prod\limits_{t_j<t_k<t}\underline{b}_k,$

$z'(t)\leq z'(t_j^+)\prod\limits_{t_j<t_k<t}\underline{b}_k.$

根据(2.2)式,可得

$$ z(t_k^+)\leq \overline{a}_kz(t_k),k=j+1,j+2,\cdots. $$

再次运用引理1.1,可得

$\begin{align} & z(t)\le z(t_{j}^{+})\prod\limits_{{{t}_{j}}<{{t}_{k}}<t}{{{{\bar{a}}}_{k}}}+\int_{t_{j}^{+}}^{t}{\prod\limits_{s<{{t}_{k}}<t}{{{{\bar{a}}}_{k}}}}({z}'(t_{j}^{+})\prod\limits_{{{t}_{j}}<{{t}_{k}}<s}{{{{\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{b}}}_{k}}})\text{d}s \\ & \ \ \ \ \ \ =\prod\limits_{{{t}_{j}}<{{t}_{k}}<t}{{{{\bar{a}}}_{k}}}[z(t_{j}^{+})-\beta \int_{t_{j}^{+}}^{t}{\prod\limits_{{{t}_{j}}<{{t}_{k}}<s}{\frac{{{{\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{b}}}_{k}}}{{{{\bar{a}}}_{k}}}}}\text{d}s],\\ \end{align}$ (2.10)

由条件(2.8),可得$\lim\limits_{t\rightarrow+\infty}z(t)=-\infty$,此结论与$z(t)>0$矛盾,所以$z'(t_k)\geq 0\:(k\geq l)$.

另外由(2.3)式可推知$z'(t_k^+)\geq \underline{b}_kz'(t_k)\geq 0\:(k\geq l)$.因为$z'(t)$在区间$(t_k,t_{k+1}]\:(k\geq l)$上单调减,显然在此区间上$z'(t)\geq z'(t_{k+1})>0$,当$t\in [T,t_l]$上$z'(t)\geq z'(t_l)>0$.证毕.

定理2.1 如果条件(2.8)与以下条件

$\begin{equation}\label{eq:b11} \lim\limits_{t\rightarrow+\infty}\int^t_{t_0} \prod\limits_{t_0<t_k<\eta}\frac{\underline{a}_k}{\overline{b}_k}F(\eta){\rm d}\eta=+\infty \end{equation}$ (2.11)

均成立,则问题(1.1)-(1.4)的每一个解在区域$G$内振动,其中

$$ F(t)=\bigg(1-\sum\limits^l_{s=1}c_s(t-\sigma)\bigg)B(t). $$

用反证法.假设向量值函数$u(x,t)=\{u_1(x,t),u_2(x,t),\cdots,u_n(x,t)\}$是问题(1.1)-(1.4)的一个非振动解.不失一般性,假定存在$\overline{t_0}>0,$使得当$(x,t)\in\Omega\times[\overline{t_0},+\infty),$ $ {\rm sgn}u_i(x,t)\neq0,$ $ i \in I_n$.

令$Z_i(x,t)=\gamma_iu_i(x,t),\gamma_i={\rm sgn}u_i(x,t),$则$Z_i(x,t)>0.$由条件${\rm \widetilde{H}}_1)$知,存在$t_0\geq \overline{t_0}$,使得当$(x,t)\in \Omega\times[t_0,+\infty),$ $ Z_i(x,t-\mu_s)>0,$ $ Z_r(x,t-\sigma)>0,$ $Z_i(x,t-\tau_{j})>0,$ $ s\in I_l,r\in I_n,$ $ j\in I_m,i\in I_n$.由引理$2.1$可知,$z(t)$是脉冲微分不等式(2.1)-(2.3)的正解.

当$t\geq t_0,\ t\neq t_k(k\in I_{\infty})$时,定义

$\begin{equation}\label{eq:b12} w(t)=\frac{z'(t)}{z(t)},\ \ t\geq t_0. \end{equation}$ (2.12)

由引理$2.2$知,$w(t)\geq0,\ t\geq t_0$.不妨假设$z(t_0)=1$,由(2.12)式得

$\begin{equation}\label{eq:b13} z(t)=\exp\left(\int^t_{t_0}w(s){\rm d}s\right), \end{equation}$ (2.13)
$\begin{equation}\label{eq:b14} z'(t)=w(t)\exp\left(\int^t_{t_0}w(s){\rm d}s\right), \end{equation}$ (2.14)
$\begin{equation}\label{eq:b15} z''(t)=w^2(t)\exp\left(\int^t_{t_0}w(s){\rm d}s\right)+w'(t)\exp\left(\int^t_{t_0}w(s){\rm d}s\right). \end{equation}$ (2.15)

将式(2.13)-(2.15)代入(2.1)式,得

$\begin{align} & {{w}^{2}}(t)\exp \left(\int_{{{t}_{0}}}^{t}{w}(s)\text{d}s \right)+{w}'(t)\exp \left(\int_{{{t}_{0}}}^{t}{w}(s)\text{d}s \right)\\ &+(1-\sum\limits_{s=1}^{l}{{{c}_{s}}}(t-\sigma))B(t)\exp \left(\int_{{{t}_{0}}}^{t-\sigma }{w}(s)\text{d}s \right)\le 0,\\ \end{align}$

化简得

$\begin{equation}\label{eq:b16} w^2(t)+w'(t)+\bigg(1-\sum\limits^l_{s=1}c_s(t-\sigma)\bigg)B(t)\exp \bigg(-\int^t_{t-\sigma}w(s){\rm d}s\bigg)\leq 0. \end{equation}$ (2.16)

根据(2.16)式及条件$\overline{b}_k\leq\underline{a}_k$,易知$w(t)$是单调减函数,则当$t\geq t_0$时,$w(t)\leq w(t_0)$且

$\begin{equation}\label{eq:b17} w^2(t)+w'(t)+\exp(-\sigma w(t_0))\bigg(1-\sum\limits^l_{s=1}c_s(t-\sigma)\bigg)B(t)\leq0. \end{equation}$ (2.17)

由(2.17)式可得到

$$ w'(t)+\exp(-\sigma w(t_0))\bigg(1-\sum\limits^l_{s=1}c_s(t-\sigma)\bigg)B(t)\leq0. $$

由式(2.2),(2.3)和(2.12)式,可得

$$ w(t^+_k)=\frac{z'(t^+_k)}{z(t^+_k)}\leq\frac{\overline{b}_kz'(t_k)}{\underline{a}_kz(t_k)}=\frac{\overline{b}_k}{\underline{a}_k}w(t_k),\ k\in I_{\infty}. $$

于是,我们得到了如下脉冲微分不等式

$\begin{align} & {w}'(t)\le -\exp(-\sigma w({{t}_{0}}))(1-\sum\limits_{s=1}^{l}{{{c}_{s}}}(t-\sigma))B(t)\\ &=-\exp(-\sigma w({{t}_{0}}))F(t),\ \ t\ne {{t}_{k}},\\ & w(t_{k}^{+})\le \frac{{{{\bar{b}}}_{k}}}{{{{\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{a}}}_{k}}}w({{t}_{k}}).\\ \end{align}$

应用引理1.1,结合条件(2.11),可得

$\begin{align} & w(t)\le w({{t}_{0}})\prod\limits_{{{t}_{0}}<{{t}_{k}}<t}{\frac{{{{\bar{b}}}_{k}}}{{{{\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{a}}}_{k}}}}-\exp(-\sigma w({{t}_{0}}))\int_{{{t}_{0}}}^{t}{\prod\limits_{\eta <{{t}_{k}}<t}{\frac{{{{\bar{b}}}_{k}}}{{{{\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{a}}}_{k}}}}}F(\eta)\text{d}\eta \\ & \ \ \ \ \ \ \ =\prod\limits_{{{t}_{0}}<{{t}_{k}}<t}{\frac{{{{\bar{b}}}_{k}}}{{{{\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{a}}}_{k}}}}[w({{t}_{0}})-\exp(-\sigma w({{t}_{0}}))\int_{{{t}_{0}}}^{t}{\prod\limits_{{{t}_{0}}<{{t}_{k}}<\eta }{\frac{{{{\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{a}}}_{k}}}{{{{\bar{b}}}_{k}}}}}F(\eta)\text{d}\eta] \\ & \ \ \ \ \ <0,\\ \end{align}$

与$w(t)\geq 0$矛盾.证毕.

推论2.1 当边界条件换为(1.5)式时,引理2.1,引理2.2及定理2.1中的结论仍成立,故问题(1.1)-(1.3),(1.5)的每一个解都是振动的.

3 应用举例

例3.1 考虑以下脉冲时滞中立型双曲方程组

$\begin{align} & \frac{{{\partial }^{2}}}{\partial {{t}^{2}}}[{{u}_{2}}(x,t)+\frac{1}{2}{{u}_{2}}(x,t-2\pi)]-{{\text{e}}^{t}}{{u}_{1}}(x,t-\frac{\pi }{2})+2({{t}^{2}}+1){{u}_{2}}(x,t-\frac{\pi }{2})\\ &=2u_{2}^{2}(x,t)\Delta {{u}_{2}}(x,t)+tu_{2}^{2}(x,t-\frac{3\pi }{2})\Delta {{u}_{2}}(x,t-\frac{3\pi }{2}),\ t>1,\ t\ne {{8}^{k}},\\ & {{u}_{2}}(x,{{({{8}^{k}})}^{+}})=7{{u}_{2}}(x,{{8}^{k}}),\\ & {{u}_{2t}}(x,{{({{8}^{k}})}^{+}})={{u}_{2t}}(x,{{8}^{k}}),\ \ k\in {{I}_{\infty }},(x,t)\in(0,\pi)\times {{\mathbb{R}}^{+}},\\ \end{align}$ (3.1)

边界条件为

$\begin{equation}\label{eq:c2} \frac{\partial}{\partial x}u_i(0,t)=\frac{\partial}{\partial x} u_i(\pi,t)=0,\ \ t>1,\ t\neq 8^k,i=1,2, \end{equation}$ (3.2)

其中

$$ n=2,\ l=1,\ m=1,\ N=1,\mu_1=2\pi,\sigma=\frac{\pi}{2}, \tau_1=\frac{3\pi}{2}, $$ $$ c_1(t)=\frac{1}{2},\ a_1(t)=t^2,\:a_2(t)=2,\ h_1(u)=h_2(u)=u^2,\ a_{11}(t)=a_{21}(t)=t, $$ $$ h_{11}(u)=h_{21}(u)=u^2,f_{11}(u)=f_{12}(u)=f_{21}(u)=f_{22}(u)=u, $$ $$ b_{11}(x,t)=3(x^2+1)e^t, b_{12}(x,t)=-2t^2,\ b_{21}(x,t)=-{\rm e}^t,\ b_{22}(x,t)=2(t^2+1). $$

很容易验证条件${\rm \widetilde{H}_1)}-{\rm \widetilde{H}_5)}$及以下条件均成立

$$ b_{11}(t)=3{\rm e}^t,\ b_{22}(t)=2(t^2+1),\ \overline{b_{12}(t)}=2t^2,\ \overline{b_{21}(t)}={\rm e}^t,\ B(t)=2. $$

并且

$\begin{align} & \underset{t\to+\infty }{\mathop{\lim }}\,\int_{{{t}_{0}}}^{t}{\prod\limits_{{{t}_{0}}<{{t}_{k}}<s}{\frac{{{{\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{b}}}_{k}}}{{{{\bar{a}}}_{k}}}}}\text{d}s=\int_{1}^{+\infty }{(}\prod\limits_{1<{{t}_{k}}<s}{\frac{1}{7}})\text{d}s \\ & \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ =\int_{1}^{{{t}_{1}}}{(}\prod\limits_{1<{{t}_{k}}<s}{\frac{1}{7}})\text{d}s+\int_{t_{1}^{+}}^{{{t}_{2}}}{(}\prod\limits_{1<{{t}_{k}}<s}{\frac{1}{7}})\text{d}s \\ & \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ +\int_{t_{2}^{+}}^{{{t}_{3}}}{(}\prod\limits_{1<{{t}_{k}}<s}{\frac{1}{7}})\text{d}s+\cdots \\ & \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ge \int_{t_{1}^{+}}^{{{t}_{2}}}{\prod\limits_{1<{{t}_{k}}<s}{\frac{1}{7}}}\text{d}s+\int_{t_{2}^{+}}^{{{t}_{3}}}{\prod\limits_{1<{{t}_{k}}<s}{\frac{1}{7}}}\text{d}s+\cdots \\ & \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ =\frac{1}{7}\times({{8}^{2}}-8)+{{(\frac{1}{7})}^{2}}\times({{8}^{3}}-{{8}^{2}})+\cdots \\ & \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ =7\sum\limits_{n=1}^{+\infty }{{{(\frac{8}{7})}^{n}}}=\infty,\\ \end{align}$

所以(2.8)式成立.另外

$$ \lim\limits_{t\rightarrow +\infty}\int_{t_0}^t\prod\limits_{t_0<t_k<s}\frac{\underline{a}_k}{\overline{b}_k} F(\eta){\rm d}\eta =\int_{1}^{+\infty}\bigg(\prod\limits_{1<t_k<s}\frac{4}{3}\bigg){\rm d}\eta =\infty, $$

于是(2.11)式也成立.根据定理2.1知,问题(3.1),(3.2)的每一个解在区域$(0,\pi)\times{\Bbb R}^+$上是振动的.

注3.1 如果边界条件(3.2)换成如下Dirichlet边界条件

$\begin{equation}\label{eq:c3} u_i(0,t)=u_i(\pi,t)=0,\ \ t>1,\ t\neq 8^k,i=1,2, \end{equation}$ (3.3)

根据推论2.1,问题(3.1),(3.3)的每一个解在区域$(0,\pi)\times{\Bbb R}^+$内是振动的.

参考文献
[1] Lakshmikantham V, Bainov D D, Simenov P S. Theory of Impulsive Differential Equations. Singapore:World Scientific, 1989
[2] Bainov D D, Simeonov P S. Systems with Impulsive Effect:Stability Theory and Applications. Ellis Horwood:Chichester, 1989
[3] Hale J K. Theory of Functional Differential Equations. New York:Springer-Verlag, 1977
[4] Yoshida N. Oscillation Theory of Partial Differential Equations. Singapore:World Scientific, 2008
[5] Yan J R, Kou C H. Oscillation of solutions of impulsive delay differential equations. J Math Anal Appl, 2001, 254:358-370
[6] Agarwal R P, Meng F W, Li W N. Oscillation of solutions of systems of neutral type partial functional differential equations. Comput Math Appl, 2002, 44:777-786
[7] Minchev E, Yoshida N. Oscillations of vector differential equations of hyperbolic type with functional arguments. Math J Toyama Univ, 2003, 26:75-84
[8] Minchev E. Forced oscillation of solutions of systems of hyperbolic equations of neutral type. Appl Math Comput, 2004, 155:427-438
[9] Luo J W. Oscillation of hyperbolic partial differential equations with impulses. Appl Math Comput, 2002, 133:309-318
[10] Cui B T, Han M A. Oscillation theorems for nonlinear hyperbolic systems with impulses. Nonlinear Anal Real World Appl, 2008, 9:94-102
[11] He M X, Liu A P. Oscillation of hyperbolic partial differential equations. Appl Math Comput, 2003, 142:205-224
[12] Liu A P, Xiao L, Liu T. Oscillation of nonlinear impulsive hyperbolic equations with several delays. Electron J Differential Equations, 2004, 2004:1-6
[13] Liu A P, Ma Q X, He M X. Oscillation of nonlinear impulsive parabolic equations of neutral type. Rocky Mountain J Math, 2006, 36:1011-1026
[14] Shoukaku Y, Yoshida N. Oscillation of nonlinear hyperbolic equations with functional arguments via Riccati method. Appl Math Comput, 2010, 217:143-151
[15] Shoukaku Y. Oscillation of solutions for forced nonlinear neutral hyperbolic equations with functional arguments. Electron J Differential Equations, 2011, 2011:1-16
[16] Agarwal R P, Karakoc F. A survey on oscillation of impulsive delay differential equations. Comput Math Appl, 2010, 60:1648-1685
[17] Liu Z, Sun Y G. Interval criteria for oscillation of a forced impulsive differential equation with Riemann-Stieltjes integral. Comput Math Appl, 2012, 63:1577-1586
[18] Zhou X L, Guo Z H, Wang W S. Interval oscillation criteria for nonlinear impulsive delay differential equations with damping term. Appl Math Comput, 2014, 249:32-44
[19] Ma Q X, Liu A P. Oscillation criteria of neutral type impulsive hyperbolic equations. Acta Math Sci, 2014, 34A:1845-1853