数学物理学报  2015, Vol. 35 Issue (5): 987-994   PDF (487 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
许艳丽
一类具有震动恢复率和时滞的传染病模型的指数收敛性
许艳丽    
湘南学院数学与金融学院 湖南郴州 423000
摘要: 考虑了一类具有震动恢复率和时滞的传染病模型,建立了使模型所有解指数收敛于零向量的一些充分条件,并用数值模拟验证了所得的结论,改进和推广了已有文献的相应结果.
关键词: 传染病模型     指数收敛     时滞     震动恢复率    
Exponential Convergence of An Epidemic Model with Delays and Oscillating Recovery Rates
Xu Yanli    
School of mathematics and Finance, Xiangnan University, Hunan Chenzhou 423000
Abstract: This paper is concerned with an epidemic model with delays and oscillating recovery rates. Some new sufficient conditions are established to ensure that all solutions of this model converge exponentially to zero vector. Some numerical simulations are carried out to support the theoretical findings. Our results improve and generalize those of the previous studies.
Key words: Epidemic model     Exponential convergence     Delay     Oscillating recovery rate    
1 引言

为了描述淋病和其他传染病的传播动态,文献[1, 2, 3, 4, 5] 建立和研究了一类非线性时滞微分系统

\begin{equation} x'_{i}(t) =-a_{i}(t)x_{i}(t) + (c_{i}(t)-x_{i}(t))\sum_{j=1}^{n}\beta_{ij}(t)x_{j}(t-\tau_{ij}(t)) ,\ i=1,2,\cdots,n. % \end{equation} (1.1)

文献[6, 7, 8, 9, 10]的作者们在系统(1.1)中考虑了连续分布时滞,得到如下系统

\begin{equation} x'_{i}(t) =-a_{i}(t)x_{i}(t) + (c_{i}(t)-x_{i}(t))\sum_{j=1}^{n}\beta_{ij}(t)\int^{0}_{-\infty}K_{ij}(s)x_{j}(t+s){\rm d}s ,\ i=1,2,\cdots, n. % \end{equation} (1.2)

系统(1.2)比系统(1.1)更符合实际,其中$c_{i}(t)$是第$i$个区域无病易感者数量 (如果不被传染就不会变成染病者), $x_{i}(t)$表示第$i$个区域在t时刻染病者数量,$\beta_{ij}(t)$ 是第$j$个区域的染病者对 第$i$个区域易感者在t时刻的传染率,$a_{i}(t)$表示第$i$个区域染病者的恢复率, $ \tau_{ij}(t)\geq 0$表示病毒 在体内的潜伏期,即染病者发病到感染其他人的时段.此外,此模型染病者不会死亡并且不具有免疫力.

收敛性在动力系统行为特征中扮演着关键的角色,种群动力学模型解的收敛性研究已经取 得了广泛的结果(见文献[11, 12, 13]). 特别地,恢复率函数$a_{i}(t)$为不震动时,即

\begin{equation} \inf\limits_{t\in R}a_{i } (t)> 0,\ \ i \in J=\{1,\ 2,\ \cdots,n \}, %$$ \end{equation} (1.3)

系统(1.1)和(1.2)已经建立了非震动条件下确保解收敛的若干充分性条件, 其结论可见文献[1, 2, 3, 4, 5, 14, 16] 和 [6, 7, 8, 9, 10, 15].

另一方面,文献[17]指出,随着季节波动,震动系数出现在线性种群动力学模型方程中, 在某些季节死亡率(收获率) 可能会大于出生率,此外由于缺乏药物和及时治疗,某些突发传染病会出现负恢复率. 那么,一个问题自然产生了: 寻求确立带有震动恢复率的系统(1.1)和(1.2)的全局指数稳定性的条件.基于上述讨论, 本文中我们将研究如下 带有时变时滞和连续分布时滞传染病模型的指数稳定性

\begin{eqnarray} x'_{i}(t) &=&-a_{i}(t)x_{i}(t) + (c_{i}(t)-x_{i}(t))\sum_{j=1}^{n}\beta_{ij}(t)x_{j}(t-\tau_{ij}(t)) \nonumber\\ & &+(\tilde{c}_{i}(t)-x_{i}(t))\sum_{j=1}^{n}\tilde{\beta}_{ij}(t) \int^{0}_{-\infty}K_{ij}(s)x_{j}(t+s){\rm d}s ,~ i\in J, % (1.4) \end{eqnarray} (1.4)

其中 $ a_{i}:R\rightarrow R$,$ c_{i},\ \beta_{ij},\tilde{c}_{i},\ \tilde{\beta}_{ij},\tau_{ij}:R\rightarrow [0,+\infty) $ 和 $K_{ij}:(-\infty,0]\rightarrow [0,+\infty)$是有界连续函数,且 $K_{ij}(t) {\rm e}^{\kappa t}$ 在$[0,+\infty)$可积, $\kappa$为一个确定正的常数,$ i,\ j\in J.$ 显然,方程(1.1)和(1.2)是模型(1.4) 的特殊情形.

基于模型(1.4)的生物学意义,我们只考虑如下初始条件

$$x_{i}(s)=\varphi_{i}(s),s\in (-\infty,\ 0],\varphi_{i}(0)>0,\ i\in J, \varphi=(\varphi_1,\varphi_2,\cdots,\varphi_n)\in C_+ ,$$ (1.5)

这里

$$C_{+}=\underbrace{C^{B}((-\infty,0],[0, +\infty))\times C^{B}((-\infty,0],[0,+\infty)) \times \cdots\times C^{B}((-\infty,0],[0,+\infty))}_n, $$

且 $C^{B}((-\infty,0],[0,+\infty))$ 是有界连续函数的集合.

本文主要目的是在舍弃假设(1.3)的前提下,给出初值问题$(1.4)$和$(1.5)$的全局指数稳定性的充分 条件. 据我们所知,具有震动恢复率的模型$(1.4)$的收敛性问题未见他人研究,本 文的结果是全新的. 本文第三部分数值模拟例子很好地说明了我们结果的有效性.

本文假设以下条件成立.

$ (H_{1})$ 存在一个正的常数$F^{i}$ 和一个连续函数 $a_{i}^{*} :R\rightarrow (0,\ +\infty)$ 使得

$$ {\rm e}^{ -\int_{s}^{t}a_{i}(u){\rm d}u}\leq F^{i} {\rm e}^{-\int_{s}^{t}a_{i}^{*}(u){\rm d}u} \ \mbox{ } t,s\in R,\mbox{ }t-s\geq 0, \ i\in J; $$

$(H_{2})$ 存在正的常数 $\xi_{1},\xi_{2},\cdots,\xi_{n}$ 使得

$$ \sup\limits_{t\in R}\bigg\{-a_{i} ^{*}(t) +\xi_{i}^{-1}F^{i}c_{i}(t) \sum_{j=1}^{n}\beta_{ij}(t)\xi_{j} +\xi_{i}^{-1} F^{i}\tilde{c}_{i}(t) \sum_{j=1}^{n}\xi_{j} \tilde{\beta}_{ij}(t)\int^{0}_{-\infty}K_{ij}(v) {\rm d}v\bigg\} < 0 ,\ i\in J. $$

为了方便,我们介绍一些符号.

$$ |x|=(|x_{1}|,\ |x_{2}|,\ \cdots,|x_{n}|)^{T}, \|x\|=\max\limits_{1\leq i\leq n} |x_{i}|, g^{+}=\sup\limits_{t\in R}|g(t)|, g^{-}=\inf\limits_{t\in R}|g(t)|. $$

记初值问题$(1.4)$和$(1.5)$的解为$ x(t;0,\varphi)= (x_{1}(t; 0,\varphi), x_{2}(t;0,\varphi),\cdots,x_{n}(t; 0,\varphi))^{T} $, 并记其最大右存在区间为$[0,\eta(\varphi))$.

2 主要结果

定理 2.1 假设 $ (H_{1}) $ 和$ (H_{2}) $ 成立. 则,当$ x_i(t; 0,\varphi)>0 ,t\in [0, \eta(\varphi)),\ i\in J,$ 集合 $\{x_i(t; 0,\varphi ): t\in [0,\eta(\varphi)), i\in J\}$ 有界,且 $\eta(\varphi)=+\infty$. 此外,当 $t\rightarrow+\infty$时,$ x(t;0,\varphi)= (x_{1}(t; 0,\varphi), x_{2}(t;0,\varphi),\cdots,x_{n}(t; 0,\varphi))^{T}$ 收敛于$(0,0,\cdots,0)^{T}$.

$$ (x_{1}(t), x_{2}(t),\cdots,x_{n}(t))^{T}=(x_{1}(t; t_{0},\varphi),x_{2}(t; t_{0},\varphi),\cdots,x_{n}(t; t_{0},\varphi))^{T}. $$

首先证明

\begin{equation} x_i(t)>0 ,\ t\in [0, \eta(\varphi)),\ i\in J. %$$ \end{equation} (2.1)

利用反证法,假设(2.1)式不成立, 那么必存在 $i\in J$ 和 $T\in [0,\eta(\varphi))$ 使得

\begin{equation} x _{i}(T) =0,\ x _{j}(t) >0 ,\ t\in (-\infty,T),\ j\in J.% $$ \end{equation} (2.2)

由(1.4)式,有

\begin{eqnarray*} x_{i}'(s) &=&-\bigg\{ a_{i}(s)+\sum\limits_{j=1}^{n} [\beta_{ij}(s)x_{j}(s-\tau_{ij}(s)) + \tilde{\beta}_{ij}(s)\int^{0}_{-\infty}K_{ij}(v )x_{j}(s+v){\rm d}v] \bigg\} x_{i}(s) \\ &&+ c_{i}(s) \sum_{j=1}^{n}\beta_{ij}(s)x_{j}(s-\tau_{ij}(s)) + \tilde{c}_{i}(s) \sum_{j=1}^{n}\tilde{\beta}_{ij}(s)\int^{0}_{-\infty}K_{ij}(v)x_{j}(s+v){\rm d}v ,\ s\in [0,T] \end{eqnarray*}

\begin{eqnarray*} 0 & = & x _{i}(T)\\ & = & x_{i} (0) {\rm e}^{-\int_{0}^{T}\{a_{i}(u)+\sum\limits_{j=1}^{n}[\beta_{ij}(u)x_{j}(u-\tau_{ij}(u)) + \tilde{\beta}_{ij}(u)\int^{0}_{-\infty}K_{ij}(v )x_{j}(u+v){\rm d}v]\}{\rm d}u}\\ & & + \int_{0}^{T}{\rm e}^{ -\int_{s}^{T}\{a_{i}(u)+\sum\limits_{j=1}^{n}[\beta_{ij}(u)x_{j}(u-\tau_{ij}(u)) + \tilde{\beta}_{ij}(u)\int^{0}_{-\infty}K_{ij}(v )x_{j}(u+v){\rm d}v]\}{\rm d}u}\\ & & \bigg[ c_{i}(s) \sum_{j=1}^{n}\beta_{ij}(s)x_{j}(s-\tau_{ij}(s)) + \tilde{c}_{i}(s) \sum_{j=1}^{n}\tilde{\beta}_{ij}(s)\int^{0}_{-\infty}K_{ij}(v)x_{j}(s+v){\rm d}v\bigg]{\rm d}s \\ & \geq &x_{i}(0) {\rm e}^{-\int_{0}^{T}\{a_{i}(u)+\sum\limits_{j=1}^{n}[\beta_{ij}(u)x_{j}(u-\tau_{ij}(u)) + \tilde{\beta}_{ij}(u)\int^{0}_{-\infty}K_{ij}(v )x_{j}(u+v){\rm d}v]\}{\rm d}u} \\ & > & 0, \end{eqnarray*}

这是一个矛盾,因此(2.1) 式成立.

其次证明 $\{x_i(t ): t\in [0,\eta(\varphi)), i\in J\}$ 有界且 $\eta(\varphi)=+\infty$.

令$ y (t) = (y_{1}(t),y_{2}(t),\cdots,y_{n}(t))^{T} = ( \xi_{1}^{-1} x_{1}(t) ,\xi_2^{-1} x_{2}(t) ,\cdots,\xi_n^{-1} x_{n}(t) )^{T}. $ 由 $ (1.4) $式 有

\begin{eqnarray*} y_{i}'(t)&=&-a_{i}(t)y_i (t )+\xi_{i}^{-1} (c_{i}(t)-x_{i}(t))\sum_{j=1}^{n}\beta_{ij}(t)x_{j}(t-\tau_{ij}(t)) \\& \ &+\xi_{i}^{-1}(\tilde{c}_{i}(t)-x_{i}(t))\sum_{j=1}^{n} \tilde{\beta}_{ij}(t)\int^{0}_{-\infty}K_{ij}(s)x_{j}(t+s){\rm d}s,\ i\in J. % \ \ \ (2.1) \end{eqnarray*}

由 $(H_{2})$,选择一个常数$\lambda \in (0,\min\{\kappa, \min\limits_{i\in J}a_{i} ^{*}\ ^{-}\}) $ 使得

\begin{eqnarray} && \sup\limits_{t\in R}\bigg\{\lambda-a_{i} ^{*}(t) +\xi_{i}^{-1}F^{i}c_{i}(t) \sum_{j=1}^{n}\beta_{ij}(t)\xi_{j} {\rm e}^{ \lambda \tau_{ij} ^{+}} \nonumber \\ && +\xi_{i}^{-1} F^{i}\tilde{c}_{i}(t) \sum_{j=1}^{n} \xi_{j}\tilde{\beta}_{ij}(t)\int^{0}_{-\infty}K_{ij}(v) {\rm e}^{-\lambda v }{\rm d}v \bigg\} < 0 . %(2.3) \end{eqnarray} (2.3)

\begin{equation} \|\varphi \|_{\xi}= \sup\limits_{t\leq 0 }\max\limits_{1\leq i\leq n }\xi_{i}^{-1}|\varphi_{i} (t) | . % \end{equation} (2.4)

对任意 $\varepsilon>0$,有$\left\| y(t) \right\|<({{\left\| \varphi \right\|}_{\xi }}+\varepsilon ){{\text{e}}^{-\lambda t}}<M({{\left\| \varphi \right\|}_{\xi }}+\varepsilon ){{\text{e}}^{-\lambda t}},\ t\in (-\infty ,0],$ 这里 $M=\sum\limits_{i=1}^{n}{{{F}^{i}}}+1.$.

下面证明

$\left\| y(t) \right\|<M({{\left\| \varphi \right\|}_{\xi }}+\varepsilon ){{\text{e}}^{-\lambda t}},\ t\in [0,\eta (\varphi )).%\text{ }$ (2.5)

否则,必存在 $i\in J$ 和 $ \theta \in (0,\eta(\varphi)) $ 使得

\begin{equation}\left\{ \begin{array}{rcl} |y_{i} (\theta)|&=&\|y(\theta)\| = M(\|\varphi \|_{\xi}+\varepsilon){\rm e}^{-\lambda \theta},\ \ \\ \ \|y(t)\| &<&M(\|\varphi \|_{\xi}+\varepsilon){\rm e}^{-\lambda t} ,\ t \in(-\infty ,\theta). \end{array} \right.% $$ \end{equation} (2.6)

注意到

\begin{eqnarray} &&y_{i}'(s)+a_{i}(s)y_i (s ) \nonumber\\&=& \xi_{i}^{-1} (c_{i}(s)-x_{i}(s))\sum_{j=1}^{n}\beta_{ij}(s)x_{j}(s-\tau_{ij}(s)) \nonumber\\ & &+\xi_{i}^{-1}(\tilde{c}_{i}(s)-x_{i}(s))\sum_{j=1}^{n} \tilde{\beta}_{ij}(s)\int^{0}_{-\infty}K_{ij}(v)x_{j}(s+v){\rm d}v,\ s\in [0,t],\ t\in [0,\theta]. % (2.7) \end{eqnarray} (2.7)

在(2.7)式的两边乘以 ${\rm e}^{ \int_{0}^{s}a_{i}(u){\rm d}u} $,并在$ [0,t]$上积分 ,得

\begin{eqnarray} y_{i} (t) &=&y_{i} (0) {\rm e}^{-\int_{0}^{t}a_{i}(u){\rm d}u}+ \int_{0}^{t}{\rm e}^{ -\int_{s}^{t}a_{i}(u){\rm d}u}\bigg[\xi_{i}^{-1} (c_{i}(s)-x_{i}(s))\sum_{j=1}^{n}\beta_{ij}(s)x_{j}(s-\tau_{ij}(s)) \nonumber\\ & &+\xi_{i}^{-1}(\tilde{c}_{i}(s)-x_{i}(s))\sum_{j=1}^{n} \tilde{\beta}_{ij}(s)\int^{0}_{-\infty}K_{ij}(v)x_{j}(s+v){\rm d}v\bigg]{\rm d}s,\ t\in [0, \theta]. % (2.8) \end{eqnarray} (2.8)

那么,由(2.3),(2.6)和(2.8)式,有

\begin{eqnarray*} |y_{i} (\theta)| &=&y_{i} (\theta)\\ &=&y_{i} (0) {\rm e}^{-\int_{0}^{\theta}a_{i}(u){\rm d}u}+ \int_{0}^{\theta}{\rm e}^{ -\int_{s}^{\theta}a_{i}(u){\rm d}u}\bigg[\xi_{i}^{-1} (c_{i}(s)-x_{i}(s))\sum_{j=1}^{n}\beta_{ij}(s)x_{j}(s-\tau_{ij}(s)) \\& \ &+\xi_{i}^{-1}(\tilde{c}_{i}(s)-x_{i}(s))\sum_{j=1}^{n} \tilde{\beta}_{ij}(s)\int^{0}_{-\infty}K_{ij}(v)x_{j}(s+v){\rm d}v\bigg]{\rm d}s \\ &\leq & y_{i} (0) {\rm e}^{-\int_{0}^{\theta}a_{i}(u){\rm d}u}+ \int_{0}^{\theta}{\rm e}^{ -\int_{s}^{\theta}a_{i}(u){\rm d}u}\bigg[ \xi_{i}^{-1}c_{i}(s) \sum_{j=1}^{n}\beta_{ij}(s)x_{j}(s-\tau_{ij}(s)) \\& &+\xi_{i}^{-1} \tilde{c}_{i}(s) \sum_{j=1}^{n}\tilde{\beta}_{ij}(s)\int^{0}_{-\infty}K_{ij}(v)x_{j}(s+v){\rm d}v\bigg]{\rm d}s \\ &\leq & y_{i} (0)F^{i} {\rm e}^{-\int_{0}^{\theta}a^{*}_{i}(u){\rm d}u}+ F^{i}\int_{0}^{\theta}{\rm e}^{-\int_{s}^{\theta}a^{*}_{i}(u){\rm d}u}\bigg[ \xi_{i}^{-1}c_{i}(s )\sum_{j=1}^{n}\beta_{ij}(s)\xi_{j}y_{j}(s-\tau_{ij}(s)) \\& \ &+\xi_{i}^{-1} \tilde{c}_{i}(s) \sum_{j=1}^{n}\tilde{\beta}_{ij}(s)\int^{0}_{-\infty}K_{ij}(v)\xi_{j}y(s+v){\rm d}v\bigg]{\rm d}s\\ & \leq & (\|\varphi \|_{\xi}+\varepsilon)F^{i} {\rm e}^{-\int_{0}^{\theta}a^{*}_{i}(u){\rm d}u}\\& \ &+ F^{i}\int_{0}^{\theta}{\rm e}^{-\int_{s}^{\theta}a^{*}_{i}(u){\rm d}u}\bigg[ \xi_{i}^{-1}c_{i}(s )\sum_{j=1}^{n}\beta_{ij}(s)\xi_{j}M(\|\varphi \|_{\xi}+\varepsilon){\rm e}^{-\lambda (s-\tau_{ij}(s))} \\& \ & +\xi_{i}^{-1} \tilde{c}_{i}(s) \sum_{j=1}^{n}\tilde{\beta}_{ij}(s)\int^{0}_{-\infty}K_{ij}(v)\xi_{j} M(\|\varphi \|_{\xi}+\varepsilon){\rm e}^{-\lambda (s+v)}{\rm d}v\bigg]{\rm d}s \\ & \leq & M(\|\varphi \|_{\xi}+\varepsilon){\rm e}^{-\lambda \theta}\bigg\{ \frac{F^{i}}{M} {\rm e}^{-\int_{0}^{\theta}(a^{*}_{i}(u)-\lambda){\rm d}u}\\& \ &+ \int_{0}^{\theta}{\rm e}^{ -\int_{s}^{\theta}(a^{*}_{i}(u)-\lambda){\rm d}u}F^{i}\bigg[ \xi_{i}^{-1}c_{i}(s) \sum_{j=1}^{n}\beta_{ij}(s)\xi_{j} {\rm e}^{ \lambda \tau_{ij} ^{+}} \\& \ &+\xi_{i}^{-1} \tilde{c}_{i}(s) \sum_{j=1}^{n}\xi_{j}\tilde{\beta}_{ij}(s)\int^{0}_{-\infty}K_{ij}(v) {\rm e}^{-\lambda v }{\rm d}v\bigg]{\rm d}s \bigg\}\\ & \leq & M(\|\varphi \|_{\xi}+\varepsilon){\rm e}^{-\lambda \theta} \bigg[\frac{F^{i}}{M} {\rm e}^{-\int_{0}^{\theta}(a^{*}_{i}(u)-\lambda){\rm d}u} + \int_{0}^{\theta}{\rm e}^{ -\int_{s}^{\theta}(a^{*}_{i}(u)-\lambda){\rm d}u}(a^{*}_{i}(s)-\lambda){\rm d}s\bigg]\\ & \leq & M(\|\varphi \|_{\xi}+\varepsilon){\rm e}^{-\lambda \theta} \bigg[1-(1-\frac{F^{i}}{M}) {\rm e}^{-\int_{0}^{\theta}(a^{*}_{i}(u)-\lambda){\rm d}u} \bigg] \\ &< & M(\|\varphi \|_{\xi}+\varepsilon){\rm e}^{-\lambda \theta}, \end{eqnarray*}

这与(2.6)式的第一个方程矛盾,所以(2.5) 式成立.令 $\varepsilon\longrightarrow 0^{+}$,由(2.5)式知

$$\|y(t)\| \leq M\|\varphi \|_{\xi}{\rm e}^{-\lambda t} \leq M\|\varphi \|_{\xi} , \ t\in [0,\eta(\varphi)) , $$

说明集合$\{x_i(t ): t\in [0,\eta(\varphi)),i\in J\}$ 有界. 根据文献[18]中的定理2.3和2.4,易得$\eta(\varphi)=+\infty$. 因此,

$$\|x(t)\|\leq \max\limits_{i\in J}\xi_{i}\|y(t)\| \leq \max\limits_{i\in J}\xi_{i} M\|\varphi \|_{\xi}{\rm e}^{-\lambda t},\ t\in [0,+\infty ),$$

这就完成了定理2.1的证明.

下面处理与模型(1.4)类似的如下系统

\begin{eqnarray} x'_{i}(t) &=&-a_{i}(t)x_{i}(t) + (c_{i}(t)-x_{i}(t))\sum_{j=1}^{n}\beta_{ij}(t)x_{j}(t-\tau_{ij}(t)) \nonumber\\ & &+(\tilde{c}_{i}(t)-x_{i}(t))\sum_{j=1}^{n}\tilde{\beta}_{ij}(t) \int^{0}_{-\bar{\tau}_{ij}(t)}K_{ij}(s)x_{j}(t+s){\rm d}s ,\ i\in J, % (2.9) \end{eqnarray} (2.9)

其中 $ a_{i}:R\rightarrow R$,$ c_{i},\ \beta_{ij},\tilde{c}_{i},\ \tilde{\beta}_{ij},\tau_{ij},\bar{\tau}_{ij}: R \rightarrow [0,+\infty) $ 和 $K_{ij}:[-\bar{\tau}_{ij}^{+},0]\rightarrow [0,+\infty)$ 都是有界连续函数,$ i,j\in J.$

令 $\tau=\max\limits_{ i,j\in J}\{\tau_{ij}^{+},\bar{\tau}_{ij}^{+}\} ,$ 和

$$\tilde{C}^{n}_{+}=\underbrace{C([-\tau,0],[0,+\infty))\times C([-\tau,0],[0,+\infty)) \times \cdots\times C([-\tau,0],[0,+\infty))}_n.$$

与系统(2.9)相关的初始条件形式如下

$$ x_{i}(s)=\varphi_{i}(s),s\in [-\tau,0],\ \varphi_{i}(0)>0,i\in J, \ \varphi=(\varphi_1,\varphi_2,\cdots,\varphi_n)\in \tilde{C}^{n}_+ . $$ (2.10)

与定理2.1的论证类似,我们得到

定理 2.2 假设$ (H_{1}) $ 成立, 存在正的常数 $\xi_{1},\xi_{2},\cdots,\xi_{n}$ 使得

$$ \sup\limits_{t\in R }\bigg\{-a_{i} ^{*}(t) +\xi_{i}^{-1}c_{i}(t) \sum_{j=1}^{n}\beta_{ij}(t)\xi_{j} +\xi_{i}^{-1} \tilde{c}_{i}(t) \sum_{j=1}^{n}\tilde{\beta}_{ij}(t)\int^{0}_{-\bar{\tau}_{ij}(t)}K_{ij}(v) {\rm d}v\xi_{j} \bigg\} < 0 ,\ i\in J. $$

设初值问题$(2.9)$和$(2.10)$的解为 $ x(t;0,\varphi) =(x_{1}(t), x_{2}(t),\cdots,x_{n}(t))^{T} $ ,并记其最大右存在区间为 $[0,\eta(\varphi))$. 则当 $x_i(t)\geq 0 ,\ t\in [0,\ \eta(\varphi)), i\in J,$ 集合 $\{x_i(t ): t\in [0,\eta(\varphi)), i\in J\}$ 有界,且 $\eta(\varphi)=+\infty$. 此外,当 $t\rightarrow+\infty$时,$ x(t;0,\varphi)$ 指数收敛于 $(0, 0,\cdots,0)^{T}$.

注 2.1 当 $a_{i}(t)\equiv a_{i}^{*}(t) (i\in J)$ 不震动时,在

$$-[a_{i }(t )-\lambda] + \sum_{j=1}^{n}c_{i }(t )\beta_{ij}(t ) {\rm e}^{\lambda \max\limits_{ i,j\in J}\{\tau_{ij}^{+}\} } <0, \lambda \mbox{ 为一个正的常数 ,}\ i \in J $$ (2.11)

$$ -[a_{i }(t )-\lambda] + \sum_{j=1}^{n}c_{i }(t )\beta_{ij}(t ) \int^{0}_{-\tau_{ij}^{+}}K_{ij}(s){\rm d}s < 0, \lambda \mbox{ 为一个正的常数,}\ i \in J $$ (2.12)

的假设下,系统(1.1)和(2.9)的指数收敛性分别在文献[14, 16]和[15]已经得证. 显然,(2.11)式是条件$(H_{2}) $ 在 $a(t)\equiv a^{*}(t)$和 $\tilde{\beta}_{ij}(t)\equiv 0$的特殊情形, (2.12)式是条件$ (H_{2}) $ 的当$a(t)\equiv a^{*}(t)$ 和 $ \beta _{ij}(t )\equiv 0$的特殊情形. 所以文献[14, 16] 和[15]的结果分别是本文定理2.1和2.2的推论.因此, 本文的结论改进和扩展了先前文献的结果.

3 一个例子

例 3.1 考虑如下带有时滞和震动恢复率的传染病模型

$\left\{ \begin{align} & {{x}_{{{1}'}}}(t)\text{=}-(18+20\sin 2000t){{x}_{1}}(t)+(2-{{x}_{1}}(t))[{{\sin }^{2}}\sqrt{3}t{{x}_{1}}(t-{{\cos }^{2}}t) \\ & \ \ \ \ \ \ \ \ \ \ +{{\sin }^{2}}t{{x}_{2}}(t-{{\cos }^{4}}(t)]+\frac{1}{100}(2-{{x}_{1}}(t))[{{\sin }^{2}}\sqrt{3}t)\int_{-\infty }^{0}{{{\text{e}}^{s}}}{{x}_{1}}(t+s)\text{d}s \\ & \ \ \ \ \ \ \ \ \ \ +{{\sin }^{2}}\sqrt{3}t\int_{-\infty }^{0}{{{\text{e}}^{s}}}{{x}_{2}}(t+s)\text{d}s], \\ & {{x}_{{{2}'}}}(t)\text{=}-(18+20\cos 2000t){{x}_{2}}(t)+(2-{{x}_{2}}(t))[{{\cos }^{2}}\sqrt{3}t{{x}_{1}}(t-{{\cos }^{2}}t) \\ & \ \ \ \ \ \ \ \ \ \ +{{\cos }^{2}}t{{x}_{2}}(t-{{\cos }^{4}}(t))]+\frac{1}{100}(2-{{x}_{2}}(t))[{{\cos }^{2}}\sqrt{3}t\int_{-\infty }^{0}{{{\text{e}}^{s}}}{{x}_{1}}(t+s)\text{d}s \\ & \ \ \ \ \ \ \ \ \ \ +{{\cos }^{2}}\sqrt{3}t\int_{-\infty }^{0}{{{\text{e}}^{s}}}{{x}_{2}}(t+s)\text{d}s]. \\ \end{align} \right.$ (3.1)

显然,

$$a_{1}(t)=18+20\sin 2000 t,\ a_{2}(t)=18+20\cos 2000 t,\ a_{1}^{*}(t)=a_{2}^{*}(t)=18,\xi_{i} =1. $$

$$ {\rm e}^{-\int_{s}^{t}a_{i}(u){\rm d}u}\leq {\rm e}^{\frac{1}{100}} {\rm e}^{-18(t-s)},\ t,s\in R ,\ t-s\geq 0, i=1,2. $$

\begin{eqnarray*} &&\sup\limits_{t\in R }\bigg\{-a_{i} ^{*}(t) +\xi_{i}^{-1}F^{i}c_{i}(t) \sum_{j=1}^{n}\beta_{ij}(t)\xi_{j} +\xi_{i}^{-1} F^{i}\tilde{c}_{i}(t) \sum_{j=1}^{n} \tilde{\beta}_{ij}(t)\int^{0}_{-\infty}K_{ij}(v) {\rm d}v\xi_{j} \bigg\} \\ &\leq &-18+ {\rm e}^{\frac{1}{100}}(2+2+2+2)<-1 ,\ \xi_{i} =1,\ i=1,2. \end{eqnarray*}

即系统(3.1)满足了定理2.1的所有条件.故当$t\rightarrow+\infty$时, 系统(3.1)具有初值$\varphi \in C_{+}$条件下的所有解指数收敛于$(0, 0)$.数值模拟验证了此结果(见图 1).

图 1 系统(3.1)在初值$ \varphi(s)\equiv (7,8)^T, (1,1)^T,(6,3)^T$ 下的数值解

注 3.1 系统(3.1)是一个带有震动恢复率$ a_{1}(t)=18+20\sin 2000 t $ 和 $ a_{2}(t)=18+20\cos 2000 t$的简单传染病模型,很显然,文献[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 14, 15, 16]和其中的参考文献的所有结果不能 直接获得系统(3.1)的所有解的指数收敛性, 这说明本文结论是全新的.

参考文献
[1] Zhang X, Chen L. The periodic solution of a class of epidemic models. Comput Math Appl, 1999, 38(3/4):61-71
[2] Lajmanovich A, Yorke T A. Adeterministic model for gonorrhea in a nonhomogeneous populaton. Math Biosci, 1976, 28:221-236
[3] Aronsson G, Mellander I. Adeterministic model in biomathematics, asymptotic behavior and threshold conditions. Math Biosci, 1980, 49:207-222
[4] Busenberg S, Cooke K L. Periodic solution of a periodic nonlinear delay differential equation. SIAM J Appl Math, 1978, 35:704-721
[5] Ruan S, Wang W, Levin S A. The effect of global travel on the spread of SARS. Math Biosci, 2006, 3:205-218
[6] Chen L, Sun J. The positive periodic solution for an epidemic model. J Sys Sci Math Sci, 2006, 26(4):456-466
[7] Chen L. The positive periodic solution for a class of nonlinear system with continuous time delay. Math in Prac Theory, 2006, 36(4):151-157
[8] Mukandavire Z, Garira W, Chiyaka C. Asymptotic properties of an HIV/AIDS model with a time delay. J Math Anal Appl, 2007, 330(2):834-852
[9] Zhang F, Zhao X. A periodic epidemic model in a patchy environment. J Math Anal Appl, 2007, 325(1):496-516
[10] d'Onofrio A. Stability properties of pulse vaccination strategy in SEIR epidemic model. Math Biosc, 2002, 26(4):57-72
[11] Smith H L. An Introduction to Delay Differential Equations with Applications to the Life Sciences. New York:Springer, 2011
[12] Wang W. Exponential extinction of Nicholson's blowflies system with nonlinear density-dependent mortality terms. Abstr Appl Anal, 2012, Article ID 302065, 15 pages
[13] Chen Z. A shunting inhibitory cellular neural network with leakage delays and continuously distributed delays of neutral type. Neural Comput & Applic, 2013, 23:2429-2434
[14] Yi X, Gong S, Wang L. Global exponential convergence of an epidemic model with time-varying delays. Nonlinear Anal Real World Appl, 2011, 12(1):450-454
[15] Xiao B, Liu B. Exponential convergence of an epidemic model with continuously distributed delays. Math Comput Model, 2008, 48(3/4):541-547
[16] Zhao C, Yang M. Peng G. Exponential convergence of an epidemic model with time-varying delays. Nonlinear Anal Real World Appl, 2010, 11(1):117-121
[17] Berezansky L, Braverman E. On exponential stability of a linear delay differential equation with an oscillating coefficient. Applied Mathematics Letters, 2009, 22:1833-1837
[18] Hale J K, Kato J. Phase space for retarded equations with infinite delay. Funkcialaj Ekvaci, 1978, 21:11-41