数学物理学报, 2023, 43(2): 604-624

考虑病毒DNA核衣壳和细胞间传播的一般HBV扩散模型

刘利利1, 王洪刚1, 李雅芝,2,*

1山西大学复杂系统研究所 太原 030006

2黔南民族师范学院数学与统计学院 贵州都匀 558000

A Generalized HBV Diffusive Model with DNA-Containing Capsids and Cell-Cell Infection

Liu Lili1, Wang Honggang1, Li Yazhi,2,*

1Complex System Research Center, Shanxi University, Taiyuan 030006

2School of Mathematics and Statistics, Qiannan Normal University for Nationalities, Guizhou Duyun 558000

通讯作者: *李雅芝,E-mail: lyz900101@126.com

收稿日期: 2022-01-29   修回日期: 2022-10-17  

基金资助: 国家自然科学基金项目(12126349)
国家自然科学基金项目(11901326)
贵州省科学技术基金(黔科合基础-ZK[2021]一般002)
山西省卫健委资助项目(2020XM18)

Received: 2022-01-29   Revised: 2022-10-17  

Fund supported: NSFC(12126349)
NSFC(11901326)
Science and Technology Foundation of Guizhou Province, China(Qian Ke He Jichu-ZK[2021]002)
National Health Commission Foundation of Shanxi Province(2020XM18)

摘要

该文综合考虑了诱发HBV感染的两类途径、一般发生率函数和HBV病毒及其核衣壳的扩散效应, 建立了一般性的HBV 扩散模型. 证明了模型解的适定性, 两类平衡态的存在唯一性和模型的一致持久性, 然后通过构造Lyapunov 函数, 得出模型的阈值动力学行为. 最后结合数值模拟验证理论结果的同时, 揭示了扩散对各状态变量的影响, 结果显示: 扩散影响HBV感染, 且扩散系数越大, HBV感染的空间区域越大.

关键词: 一般的HBV扩散模型; 病毒DNA核衣壳; 细胞-细胞感染; 全局稳定性

Abstract

This paper investigates a generalized HBV diffusive model, where two HBV infection ways, general incidence functions and DNA-containing capsids are considered. This paper gives the well-posedness of model and then obtains the threshold dynamical behaviors of the proposed model, including the unique existence of two equalibria, the uniformly persistence of the model and global stability by using Lyapunov functionals. The numerical simulations not only verify the theoretical results, but also explore the influence of diffusion on the state variables. The result shows that diffusion affects the HBV infection, and the bigger of diffusion is, the larger of HBV infection region will become.

Keywords: Generalized HBV diffusive model; DNA-containing capsids; Cell-cell infection; Global stability

PDF (1625KB) 元数据 多维度评价 相关文章 导出 EndNote| Ris| Bibtex  收藏本文

本文引用格式

刘利利, 王洪刚, 李雅芝. 考虑病毒DNA核衣壳和细胞间传播的一般HBV扩散模型[J]. 数学物理学报, 2023, 43(2): 604-624

Liu Lili, Wang Honggang, Li Yazhi. A Generalized HBV Diffusive Model with DNA-Containing Capsids and Cell-Cell Infection[J]. Acta Mathematica Scientia, 2023, 43(2): 604-624

1 引言

乙型肝炎病毒(HBV)可导致健康肝细胞感染进而引发肝脏疾病, 如肝炎, 肝硬化和肝癌等. 据世界卫生组织报告, 目前全世界慢性乙肝病毒感染者约有3.5 亿人, 其中我国有近1.3 亿人, 乙型肝炎是我国发病率最高且危害最大的传染病之一[1]. 基于HBV感染机制, Nowak等[2]首次提出了三维常微分方程的HBV 感染模型, 此后更多学者开始关注HBV 感染模型, 并考虑不同因素使得所建立的模型更加符合实际背景. Min等[3]指出用非线性发生率函数刻画病毒感染过程更加符合实际, 李喜玲等[4]建立了一类具有免疫时滞和非线性发生率函数的分数阶HBV感染模型,利用泛函微分方程、Caputo分数阶导数和Lyapunov稳定性理论对模型进行了稳定性分析.

事实上, HBV主要活动区域是肝脏, 且病毒颗粒在肝脏内是可以自由移动的.假设病毒的扩散服从Fickian扩散原理, 王开发等[5]建立了病毒颗粒具有空间依赖的HBV感染模型, 并研究了模型行波解的存在性, 且在此基础上建立了一类具有胞内时滞的HBV扩散模型[6], 并研究了模型的稳定性; 甘庆涛等[7]利用交叉迭代和Schauder 不动点理论, 深入分析文献[6]中所建模型行波解的存在性. 进一步考虑不同的非线性发生率函数, 如饱和反应函数[8], Beddington-DeAngelis反应函数[9,10]和一般发生率函数[11]等, 学者们建立并分析HBV扩散模型的动力学性态. 需要注意上述工作均假设病毒与健康肝细胞接触, 然后引起细胞感染. 但研究表明[12,13]: 感染肝细胞与健康肝细胞接触后, 通过病毒性突触、奈米管等向健康肝细胞内释放大量病毒, 进而引发健康肝细胞被感染, 这种感染方式被称为“细胞-细胞”感染. 杨翠兰等[14]建立并分析了一类具有细胞-细胞感染和时滞的病毒感染模型的全局稳定性. Sun等[15]分析了一类具有一般非线性发生率反应函数、细胞-细胞感染和时滞的HBV扩散模型的全局稳定性, 并结合Matlab数值模拟验证了理论结果.

众所周知, HBV病毒属DNA病毒科, 且DNA遗传物质被包含在核衣壳中, 其浓度对病毒颗粒的浓度水平有着重要的影响. 鉴于此, Manna等[16]在经典HBV 病毒模型[2]的基础上增加病毒DNA核衣壳仓室, 从而建立了一个四维常微分方程模型, 并分析了模型的稳定性. 考虑染病肝细胞的感染年龄, Liu等[17]建立并分析了一个具有感染年龄和核衣壳的HBV 模型的稳定性. 进一步考虑时滞[18]、一般发生率函数[19]及免疫反应[20], 学者们建立并分析了具有核衣壳的HBV扩散模型的动力学行为.

本文将考虑病毒DNA核衣壳和细胞-细胞感染, 并考虑病毒DNA 核衣壳和病毒颗粒的自由扩散, 建立具有非线性发生率函数的HBV 扩散模型并对其进行研究.

2 模型建立

本文综合考虑病毒DNA核衣壳和细胞-细胞传播的影响, 建立了一类一般性的HBV扩散模型

$\begin{matrix}\left\{\begin{array}{ll}\displaystyle \frac{{\rm \partial} T(x,t)}{{\rm \partial} t}=s-\mu T(x,t)-h_1(T(x,t),V(x,t))-h_2(T(x,t),I(x,t)),\\ \displaystyle \frac{{\rm \partial} I(x,t)}{{\rm \partial} t}=h_1(T(x,t),V(x,t))+h_2(T(x,t),I(x,t))-\delta I(x,t),\\ \displaystyle \frac{{\rm \partial} D(x,t)}{{\rm \partial} t}=d_1\Delta D(x,t)+\theta I(x,t)-(\alpha+\delta)D(x,t),\\ \displaystyle \frac{{\rm \partial} V(x,t)}{{\rm \partial} t}=d_2\Delta V(x,t)+\alpha D(x,t)-c V(x,t).\end{array}\right.\end{matrix}$

上述反应-扩散系统(2.1)具有以下的初始条件

$\begin{matrix}T(x,0)=T_0(x)\geq0,\, I(x,0)=I_0(x)\geq0, \, D(x,0)=D_0(x)\geq0, \, V(x,0)=V_0(x)\geq0,\,x\in\Omega,\end{matrix}$

其中$\Omega\subset {\Bbb R}^n$是有界区域且具有光滑的边界${\rm \partial}\Omega$. 系统(2.1)具有以下齐次的Neumann边界条件

$\begin{matrix}\frac{{\rm \partial} D}{{\rm \partial} n}=0,\quad \frac{{\rm \partial} V}{{\rm \partial} n}=0,\quad x\in{\rm \partial}\Omega,\quad t\in(0,+\infty).\end{matrix}$

系统(2.1)中$T(x,t)$, $I(x,t)$, $D(x,t)$$V(x,t)$分别表示在时刻$t$ 位置$x$处的健康肝细胞、感染肝细胞、病毒DNA核衣壳和病毒颗粒的浓度. 其他参数的生物学意义参见表1.

表1   系统(2.1)中各参数的生物学意义

新窗口打开| 下载CSV


非线性一般发生率函数$h_1(T,V)$$h_2(T,I)$满足下面的假设.

假设2.1 (i) 对于所有的$T, I, V\geq 0$, $h_1(T,0) = h_1(0,V) = h_2(T,0) = h_2(0,I) = 0$; 并且对于所有$T, I, V> 0$, 存在$\eta_1,\eta_2>0,$ 都有$0<h_1(T,V)\leq \eta_1T$$0<h_2(T,I)\leq \eta_2T$;

(ii) 对于所有的$T, I, V\geq 0$, ${{\rm \partial} h_1(T,V)}/{{\rm \partial} V}\geq0$${{\rm \partial} h_2(T,I)}/{{\rm \partial} I}\geq0$; 并且对于所有的$T\geq0$$I, V>0$, 都有${{\rm \partial} h_1(T,V)}/{{\rm \partial} T}>0$${{\rm \partial} h_2(T,I)}/{{\rm \partial} T}>0$;

(iii) 对于所有$T, I, V\geq 0$, 都满足$V \frac{{\rm \partial}h_1(T,V)}{{\rm \partial} V}-h_1(T,V)\leq0$$I \frac{{\rm \partial} h_2(T,I)}{{\rm \partial} I}-h_2(T,I)\leq0$.

3 解的适定性

定义系统 (2.1)满足的向量空间为${\Bbb X}:=C(\overline\Omega,{\Bbb R}^4),$ 且该空间的范数定义为

$\|\phi\|_{{\Bbb X}}:=\sup_{x\in \overline\Omega}=\sup_{x\in \overline\Omega}\sqrt[]{|\phi_1(x)|^2+|\phi_2(x)|^2+|\phi_3(x)|^2+|\phi_4(x)|^2}, \quad \phi=(\phi_1,\phi_2,\phi_3,\phi_4)^T\in {\Bbb X}.$

定义线性算子$A_i:C^2(\overline\Omega,{\Bbb R})\rightarrow C^2(\overline\Omega,{\Bbb R})$$( i=1,2)$, 满足以下条件

$A_i\varphi:=d\Delta \varphi(x), \quad D(A):=\Big\{\varphi\in C^2(\overline\Omega,{\Bbb R}):\frac{{\rm \partial} \varphi}{{\rm \partial} n}=0\in{\rm {\rm \partial}}\Omega \Big\}.$

由文献[21]得, 算子$A_i$是强连续半群$\{e^{A_i t}\}_{t\geq0}(i=1,2)$$C(\overline\Omega,{\Bbb R})$ 上的一个无穷小生成元. 故定义算子${\cal A}:{\Bbb X}\rightarrow {\Bbb X}$

$\begin{matrix}{\cal A}\phi(x):=\left( \begin{array}{ccc} 0\\ 0\\ A_1\phi_3(x)\\ A_2\phi_4(x)\\ \end{array}\right),\quad \phi(x):=\left( \begin{array}{ccc} \phi_1(x)\\ \phi_2(x)\\ \phi_3(x)\\ \phi_4(x)\\ \end{array}\right)\in D({\cal A}):=C(\overline\Omega,{\Bbb R}^2)\times D(A)\subset {\Bbb X}.\end{matrix}$

显然, ${\cal A}$也是强连续半群$\{e^{{\cal A} t}\}_{t\geq0}$${\Bbb X}$上的一个无穷小生成元. 接下来定义一个非线性算子${\cal F}:{\Bbb X}^+\rightarrow {\Bbb X}^+,$

$\begin{matrix}{\cal F}(\phi)(x):=\left( \begin{array}{ccc} s-\mu\phi_1(x)-h_1(\phi_1(x),\phi_4(x))-h_2(\phi_1(x),\phi_2(x))\\ h_1(\phi_1(x),\phi_4(x))+h_2(\phi_1(x),\phi_2(x))-\delta\phi_2(x)\\ \theta\phi_2(x)-(\alpha+\delta)\phi_3(x)\\ \alpha\phi_3(x)-c\phi_4(x)\\ \end{array}\right),\end{matrix}$

其中$\phi=(\phi_1,\phi_2,\phi_3,\phi_4)^T\in {\Bbb X}^+.$ 因此, 我们将系统(2.1)转化为具有向量形式的柯西问题

$\begin{matrix}\left\{\begin{array}{ll} \frac{{\rm d} u(t)}{{\rm d}t}={\cal A}u(t)+{\cal F}(u(t)),\\[2mm]u(0)=u_0,\end{array}\right.\end{matrix}$

其中$u(t)=(T(t),I(t),D(t),V(t))^{T}$$u_0=(T_0,I_0,D_0,V_0)^{T}$. 关于上述柯西问题(3.3)解的存在性, 我们给出如下结论.

引理 3.1 对于任意$u_0\in D({\cal A}),$ 存在常数$T_{\max}>0,$ 使得柯西问题(3.3)存在唯一的局部解

$u(t)=e^{At}+\int_{0}^{t}e^{{\cal A}(t-r)}{\cal F}(u(r)){\rm d}r,\quad t\in[0,T_{\max}),$

其中$T_{\max}<+\infty.$ 此外, 若$T_{\max}=\infty,$$\lim\limits_{t\rightarrow T_{\max}}\|u(x,t;u_0)\|=\infty.$

显然, 算子${\cal F}$${\Bbb X}$上是Fréchet可微, 其导数形式

${\cal F}'[\psi]:{\Bbb X}\rightarrow {\Bbb X}, \psi=(\psi_1,\psi_2,\psi_3,\psi_4)^T\in {\Bbb X}.$
${\cal F}'[\psi]\phi(\cdot):=\left( \begin{array}{ccc} F_1, F_2, F_3, F_4 \end{array}\right)^{T},$

其中

$F_1= -\bigg[\frac{{\rm \partial} h_1(\psi_1,\psi_4)}{{\rm \partial} \psi_1}+\frac{{\rm \partial} h_2(\psi_1,\psi_2)}{{\rm \partial} \psi_1}+\mu\bigg]\phi_1(\cdot)-\frac{{\rm \partial} h_1(\psi_1,\psi_4)}{{\rm \partial} \psi_4}\phi_4(\cdot)-\frac{{\rm \partial} h_2(\psi_1,\psi_2)}{{\rm \partial} \psi_2}\phi_2(\cdot),$
$F_2=\bigg[\frac{{\rm \partial} h_1(\psi_1,\psi_4)}{{\rm \partial} \psi_1}+\frac{{\rm \partial} h_2(\psi_1,\psi_2)}{{\rm \partial} \psi_1}+\mu\bigg]\phi_1(\cdot)+\bigg[\frac{{\rm \partial} h_2(\psi_1,\psi_2)}{{\rm \partial} \psi_2}-\delta\bigg]\phi_2(\cdot)+\frac{{\rm \partial} h_1(\psi_1,\psi_4)}{{\rm \partial} \psi_4}\phi_4(\cdot),$
$F_3=\theta\phi_2(\cdot)-(\alpha+\delta)\phi_3(\cdot),$
$F_4=\alpha\phi_3(\cdot)-c\phi_4(\cdot).$

结合文献[引理2.1]. 证毕.

定理 3.1 若初始条件$u_0=(T_0,I_0,D_0,V_0)$非负, 系统(2.1)的解$u(x,t;u_0)$$[0,T_{\max})\times {\Bbb X}^+$上也是非负的.

$0=(0,0,0,0)^T,$ 对于任意非负的初始条件$u_0$和非负数$\rho$, 可以得到

$\begin{matrix}u_0(x)+\rho{\cal F}(u_0)(x)=\left( \begin{array}{ccc} T_0(x)+\rho[s-\mu T_0(x)-h_1(T_0(x),V_0(x))-h_2(T_0(x),I_0(x))]\\ I_0(x)+\rho[h_1(T_0(x),V_0(x))+h_2(T_0(x),I_0(x))-\delta I_0(x)]\\ D_0(x)+\rho[\theta I_0(x)-(\alpha+\delta)D_0(x)]\\ V_0(x)+\rho[\alpha D_0(x)-cV_0(x)]\\ \end{array}\right).\end{matrix}$

根据假设2.1, 可以得到对于所有的$T,I,V\geq0,$$0\leq h_1(T,V)\leq\eta_1T$$0\leq h_2(T,I)\leq\eta_2T$.$0\leq\rho\leq\min\{1/(\mu+\eta_1+\eta_2),1/\delta,1/(\alpha+\delta),1/c\},$ 可以得到

$\begin{matrix}u_0(x)+\rho{\cal F}(u_0)(x)\geq\left( \begin{array}{ccc} (1-\rho(\mu+\eta_1+\eta_2))T_0(x)\\ (1-\rho\delta)I_0(x)\\ (1-(\alpha+\delta))D_0(x)\\ (1-\rho c)V_0(x)\\ \end{array}\right)\geq\left( \begin{array}{ccc} 0\\ 0\\ 0\\ 0\\ \end{array}\right)=\bf{0}.\end{matrix}$

选择充分小的$\rho,$ 则有$u_0+\rho{\cal F}(u_0)\in {\Bbb X}^+.$ 特别地,

$\lim\limits_{\rho\rightarrow0^+}\frac{1}{\rho}{\rm dist}(u_0+\rho{\cal F}(u_0),{\Bbb X}^+)=0.$

证毕.

为了证明系统(2.1)解的存在性, 现定义如下空间

$\begin{matrix}\Gamma=\bigg\{u=(T,I,D,V)^T\in {\Bbb X}^+|0\leq T(x,t)+I(x,t)\leq\frac{s}{\mu}, 0\leq I(x,t)\leq\frac{s}{\mu},\\ 0 \leq D(x,t)\leq M_1, 0\leq V(x,t)\leq M_2\bigg\},\end{matrix}$

其中$M_1$$M_2$分别在式(3.5)和(3.7)给出定义.

引理 3.2$\Gamma$是系统(2.1)的正不变集.

假设$u_0=(T_0,I_0,D_0,V_0)^T\in\Gamma.$ 定义一个新变量$H(x,t)=T(x,t)+I(x,t),$ 显然$H(x,0)=H_0(x)=T_0(x)+I_0(x)\in\Gamma.$$H(x,t)$两边同时对$t$求偏导, 得到

$\begin{matrix}\frac{\partial H(x,t)}{\partial t}=s-\mu T(x,t)-\delta I(x,t)\leq s-\mu(T(x,t)+I(x,t))=s-\mu H(x,t).\end{matrix}$

解上述微分不等式, 可以得到

$H(x,t)\leq H_0(x)e^{-\mu t}+s\int_{0}^{t}e^{-\int_{r}^{t}\mu {\rm d}\rho}{\rm d}r =\bigg(H_0(x)-\frac{s}{\mu}\bigg)e^{-\mu t}+\frac{s}{\mu}\leq\frac{s}{\mu}.$

进一步地, 可以得到$I(x,t)\leq s/\mu.$$I(x,t)\leq s/\mu$带入到系统(2.1)的第三个方程, 得到下述系统

$\begin{matrix}\left\{\begin{array}{ll} \frac{{\rm \partial} D(x,t)}{{\rm \partial} t}-d_1\Delta D(x,t)\leq\theta\frac{s}{\mu}-(\alpha+\delta)D(x,t), & x\in\overline{\Omega},\quad t\in(0,+\infty),\\[3mm] \frac{\partial D}{\partial n}=0,& x\in{\rm \partial}\Omega,\quad t\in(0,+\infty),\\[2mm]D_0(x)=\phi_3(x)\geq0, & x\in\overline{\Omega}.\end{array}\right.\end{matrix}$

通过系统(3.4)可以构造下述系统

$\begin{matrix}\left\{\begin{array}{ll} \frac{{\rm d}\widetilde{D}(t)}{{\rm d}t}=\theta\frac{s}{\mu}-(\alpha+\delta)\widetilde{D}(t),\\[2mm] \widetilde{D}(0)=\mathop{\max}\limits_{x\in\overline\Omega}\phi_3(x)\geq 0.\end{array}\right.\end{matrix}$

通过解微分方程, 对于任意的$t\in[0,T_{\max}),$ 可以得到

$\begin{matrix}\label{m1}\widetilde{D}(t)\leq M_1\triangleq\max\bigg\{ \frac{\theta s}{(\alpha+\delta)\mu}, \, \mathop{\max}\limits_{x\in\overline\Omega}\phi_3(x)\bigg\}.\end{matrix}$

由比较原理[23], 显然$D(x,t)\leq\widetilde{D}(t).$ 因此, 对任意$(x,t)\in\overline\Omega\times[0,T_{\max}),$ 都有$D(x,t)\leq M_1.$ 将其代入系统(2.1)的第四个方程, 得到以下系统

$\begin{matrix}\left\{\begin{array}{ll} \frac{{\rm \partial} V(x,t)}{v t}-d_2\Delta V(x,t)\leq\alpha M_1-cV(x,t), \quad & x\in\overline{\Omega}, t\in(0,+\infty), \\[3mm] \frac{{\rm \partial} V}{{\rm \partial} n}=0,& x\in{\rm \partial}\Omega, t\in(0,+\infty),\\[2mm]V_0(x)=\phi_4(x)\geq0.\end{array}\right.\end{matrix}$

同样地, 依据系统(3.6)构造下述系统

$\begin{matrix}\left\{\begin{array}{ll} \frac{{\rm d}\widetilde{V}(t)}{{\rm d}t}=\alpha M_1-c\widetilde{V}(t),\\[2mm] \widetilde{V}(0)=\max\limits_{x\in\overline\Omega}\phi_4(x)\geq 0.\end{array}\right.\end{matrix}$

解得

$\begin{matrix}\widetilde{V}(t)\leq M_2\triangleq\max\bigg\{ \frac{\alpha M_1}{c}, \, \mathop{\max}\limits_{x\in\overline\Omega}\phi_4(x)\bigg\}.\end{matrix}$

应用比较原理[23], $V(x,t)\leq\widetilde{V}(t)\leq M_2.$ 因此, 对于任意$(x,t)\in\overline\Omega\times[0,T_{\max}),$$V(x,t)\leq M_2.$ 得证.

由引理3.1和定理3.1, 对于任意的初始条件$u_0\in\Gamma,$ 系统(2.1)存在唯一的全局解$u\in\Gamma.$ 故定义系统由(2.1) 生成的连续半流$\{\Phi_t\}_{t\geq0}: {\Bbb X}^+\rightarrow {\Bbb X}^+,$$\Phi_t(u_0):=u(\cdot,t;u_0), t\geq0.$ 引理3.2表明对于任意$t\geq0$$u_0\in\Gamma,$ 都有$\Phi_t(u_0)\in\Gamma.$

对任意的$t\geq0,$ 定义$\Phi(t):\Gamma\rightarrow\Gamma$$\Phi(t)\phi=u(\cdot,t;u_0),$ 其中

$\begin{matrix}u(\cdot,t;u_0)=(T(\cdot,t;T_0),I(\cdot,t;I_0),D(\cdot,t;D_0),V(\cdot,t;V_0))\end{matrix}$

是系统(2.1)过初值$\phi\in\Gamma$上的解. 令$\Phi:=\Phi(\omega)$是系统(2.1)的Poincaré映射. 显然, $\Phi(t)$不紧, 因此本文需要引入Kuratowski 非紧性测度的概念. 具体地, 对$\Gamma$上的任意有界子集$B,$ 定义$B$的Kuratowski非紧性测度为

$\kappa(B):=\inf\bigg\{\delta>0|B\ \mbox{可以表示为有限个开集的并}:B=\bigcap^{m}_{i=1}B_i, \mbox{且B_i直径}\ d(B_i)\leq\delta\bigg\}.$

引理 3.3[24] 对于任意$t>0,$$\Phi(t)$$\Gamma$上是$\kappa$ -压缩的, 且具有压缩函数$e^{-mt}.$ 另外, $\Phi$$\Gamma$上有一个连通的全局吸引子.

系统(2.1)的无感染平衡点为$E_0=(s/\mu,0,0,0),$ 则可在$E_0$处线性化为

$\begin{matrix}\left\{\begin{array}{ll}\displaystyle \frac{{\rm \partial} I(x,t)}{{\rm \partial} t}=h_1(s/\mu,V(x,t))+h_2(s/\mu,I(x,t))-\delta I(x,t), \quad\, x\in\Omega, t>0, \\ \displaystyle \frac{{\rm \partial} D(x,t)}{{\rm \partial} t}=d_1\Delta D(x,t)+\theta I(x,t)-(\alpha+\delta)D(x,t), \quad\quad\quad x\in\Omega, t>0, \\ \displaystyle \frac{{\rm \partial} V(x,t)}{{\rm \partial} t}=d_2\Delta V(x,t)+\alpha D(x,t)-c V(x,t), \quad\quad\quad\quad\quad\,\, x\in\Omega, t>0, \\ \displaystyle \frac{{\rm \partial} D(x,t)}{{\rm \partial} n}=\frac{{\rm \partial} V(x,t)}{{\rm \partial} n}=0, \quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\;\;\; x\in\partial\Omega, t>0.\end{array}\right.\end{matrix}$

根据系统(3.8), 可将$\Phi(t)$分解为$\Phi(t)=\Phi_1(t)+\Phi_2(t), \, t\geq0,$ 其中

$\Phi_1(t)u_0=\bigg\{e^{-\big[\delta-\frac{\partial h_2(s/\mu,I(\cdot,t;I_0))}{\partial I}\big]t}I_0(x),0,0\bigg\}, \quad t\geq0,$
$\Phi_2(t)u_0=\bigg\{\int_{0}^{t}e^{-\big[\delta-\frac{\partial h_2(s/\mu,I(\cdot,t;I_0))}{\partial I}\big](t-s)}h_1 \left(\frac{s}{\mu},V(\cdot,t;V_0)\right){\rm d}s,D(\cdot,t;D_0),V(\cdot,t;V_0)\bigg\}, \, t\geq0.$

由假设, 可以得到

$\frac{\partial h_2(T,I)}{\partial I}<\frac{h_2(T,I)}{I}<\frac{\eta_2}{I}, \quad I>0.$

因为$I>0,$ 故总存在一个$\varepsilon>0,$ 使得$I>\varepsilon>0.$ 由于${\partial h_2(T,I)}/{\partial I}<{\eta_2}/{\varepsilon}\triangleq k,$$\delta-{\partial h_2(T,I)}/{\partial I}\geq\delta-k\triangleq m.$ 因此, 可以得到

$\begin{matrix}\|\Phi_1(t)\|=\sup_{\phi\in {\Bbb X}}\frac{\|\Phi_1(t)\phi\|_{\Bbb X}}{\|\phi\|_{\Bbb X}}\leq e^{-mt}\sup_{\phi\in {\Bbb X}}\frac{\|\phi\|_{\Bbb X}}{\|\phi\|_{\Bbb X}}=e^{-mt}.\end{matrix}$

另一方面, 由$\Phi(t)$的有界性和$C_0$ -半群的紧性, 显然得到对于任意的$t>0,$ 算子$\Phi_2(t)$是压缩的. 因此, 对$\Gamma$中的任意有界集$B,$ 都有

$\begin{matrix}\kappa(\Phi(t)B)\leq\kappa(\Phi_1(t)B)+\kappa(\Phi_2(t)B)\leq\|\Phi_1(T)\|\kappa(B)+0\leq e^{-mt}\kappa(B).\end{matrix}$

综上所述, $\{\Phi(t)\}_{t\geq0}$$\Gamma$上是$\kappa$ -压缩的, 且其压缩函数为$e^{-mt}.$ 进而得到, $\Phi$$\Gamma$中有一个连通的全局吸引子[25].证毕.

4 平衡点的存在性

由下一代矩阵法可以定义系统(2.1)的基本再生数$\mathfrak{R}e_0$如下

$\mathfrak{R}e_0=\frac{\alpha\theta}{\delta c(\alpha+\delta)}\frac{{\rm \partial} h_1 (\frac{s}{\mu},0)}{{\rm \partial} V}+\frac{1}{\delta}\frac{{\rm \partial} h_2(\frac{s}{\mu},0)}{{\rm \partial} I}.$

显然, 系统(2.1)总有一个无感染平衡态$E_0=(s/\mu,0,0,0).$ 接下来, 我们将通过数学推导验证感染平衡态$E^*=(T^*,I^*,D^*,V^*)$ 的存在唯一性.

定理 4.1$\mathfrak{R}e_0>1$时, 系统(2.1)存在唯一的感染平衡态$E^*.$

已知系统(2.1)的平衡态满足方程组

$\begin{matrix}\left\{\begin{array}{ll}s-h_1(T^*,V^*)-h_2(T^*,I^*)-\mu T^*=0,\\h_1(T^*,V^*)+h_2(T^*,I^*)-\delta I^*=0,\\\theta I^*-(\alpha+\delta)D^*=0,\\\alpha D^*-c V^*=0.\end{array}\right.\end{matrix}$

通过简要计算, 可以得到

$\begin{matrix}T^*=\frac{s-\delta I^*}{\mu},\quad D^*=\frac{\theta}{\alpha+\delta}I^*,\quad V^*=\frac{\alpha\theta}{c(\alpha+\delta)}I^*.\end{matrix}$

$T^*=(s-\delta I^*)/\mu\geq0,$ 解不等式得到$I^*\leq s/\delta.$ 现将式(4.2)代入到方程组(4.1)的第二个方程中, 得到

$\begin{matrix}h_1\bigg(\frac{s-\delta I^*}{\mu},\frac{\alpha\theta}{c(\alpha+\delta)}I^*\bigg)+h_2\bigg(\frac{s-\delta I^*}{\mu},I^*\bigg)-\delta I^*=0.\end{matrix}$

在区间$[s/\delta]$上构造如下函数

$\begin{matrix}f(I)=h_1\bigg(\frac{s-\delta I}{\mu},\frac{\alpha\theta}{c(\alpha+\delta)}I\bigg)+h_2\bigg(\frac{s-\delta I}{\mu},I\bigg)-\delta I.\end{matrix}$

根据假设2.1, 显然有$f(0)=0,$$f(s/\delta)=-s<0.$ 下面将通过$f'(0)$的正负判断$f(I)$$I=0$附近的变化趋势. 又因为$h_1(s/\mu,0)=h_2(s/\mu,0)=0,$ 则显然有

${\rm \partial} h_1(s/\mu,0)/{\rm \partial} T={\rm \partial} h_2(s/\mu,0)/{\rm \partial} T=0.$

$\begin{matrix}f'(0)&=&\frac{\alpha\theta}{c(\alpha+\delta)}\frac{{\rm \partial} h_1(\frac{s}{\mu},0)}{{\rm \partial} V}+\frac{{\rm \partial} h_2(\frac{s}{\mu},0)}{{\rm \partial} I}-\delta=\delta\bigg(\frac{\alpha\theta}{c\delta(\alpha+\delta)}\frac{{\rm \partial} h_1(\frac{s}{\mu},0)}{{\rm \partial} V}+\frac{1}{\delta}\frac{{\rm \partial} h_2(\frac{s}{\mu},0)}{{\rm \partial} I}-1 \bigg )\\&=&\delta(\mathfrak{R}e_0-1).\end{matrix}$

$\mathfrak{R}e_0>1$时, 得到$f_+'(0)>0.$ 应用极限的局部保号性, 存在一个充分小的$\varepsilon>0,$ 使得对任意$I\in(0,\varepsilon)$都有$[f(I)-f(0)]/(I-0)>0.$$I_1\in(0,\varepsilon),$ 化简得到$f(I_1)>0.$ 由连续函数的零点定理, 至少存在一个$I^*\in[I_1,s/\delta]\subset[s/\mu],$ 使得$f(I^*)=0.$ 根据(4.2) 式, 可以得到$T^*,D^*,V^*$也存在, 即系统(2.1)的感染平衡态$E^*$存在.

下面证明$E^*$的唯一性. 根据假设2.1, 我们推导出对于所有的感染平衡态$E^*$都有下式

$\begin{matrix}f'(I^*)&=&-\frac{\delta}{\mu}\frac{{\rm \partial} h_1^*}{{\rm \partial} T}+\frac{\alpha\theta}{c(\alpha+\delta)}\frac{{\rm \partial} h_1^*}{{\rm \partial} V}-\frac{\delta}{\mu}\frac{{\rm \partial} h_2^*}{{\rm \partial} T}+\frac{{\rm \partial} h_2^*}{{\rm \partial} I}-\delta\\&=&-\frac{\delta}{\mu}\frac{{\rm \partial} h_1^*}{{\rm \partial} T}+\frac{\alpha\theta}{c(\alpha+\delta)}\frac{{\rm \partial} h_1^*}{{\rm \partial} V}-\frac{\delta}{\mu}\frac{{\rm \partial} h_2^*}{{\rm \partial} T}+\frac{{\rm \partial} h_2^*}{{\rm \partial} I}-\bigg(\frac{\alpha\theta h_1^*}{c(\alpha+\delta)V^*}+\frac{h_2^*}{I^*}\bigg)\\&=&-\frac{\delta}{\mu}\bigg(\frac{{\rm \partial} h_1^*}{{\rm \partial} T}+\frac{{\rm \partial} h_2^*}{{\rm \partial} T}\bigg)+\frac{\alpha\theta}{c(\alpha+\delta)V^*}\bigg(V^*\frac{{\rm \partial} h_1^*}{{\rm \partial} V}-h_1^*\bigg)+\frac{1}{I^*}\bigg(I^*\frac{{\rm \partial} h_2^*}{{\rm \partial} I}-h_2^*\bigg)<0.\end{matrix}$

假设$f(I)=0$在区间$(0,s/\mu)$上至少有两个根, 由于$f(I)$连续可微, 则必然存在另一个正零点$\tilde{I}^*,$ 使得$f'(\tilde{I}^*)\geq0.$ 显然这是矛盾的, 故当$\mathfrak{R}e_0>1$时, 系统 (2.1)的感染平衡态$E^*$存在且唯一.证毕.

5 主要结论

在本章节, 我们将讨论无感染平衡态的局部以及全局稳定性, 并且证明系统 (2.1)的一致持续性和感染平衡态的全局稳定性.

$0=\mu_0<\mu_1<\cdots<\mu_n<\cdots$是拉普拉斯算子$ -\Delta$$\overline\Omega$上满足齐次Neumann边界条件的一组特征值, 定义$E(\mu_i)$是对应于特征值$\mu_i$的特征空间, 其中$\mu_i\in C^1(\Omega)(i=0,1,2\cdots ).$ 定义$\{\phi_{ij}: j=1,2,\cdots,\dim E(\mu_i)\}$表示特征空间$E(\mu_i)$下的一组标准正交基, ${\Bbb Y}=[C^1(\overline\Omega)]^4,$ 并且${\Bbb Y}_{ij}=\{a\phi_{ij}: a\in {\Bbb R}\}.$ 那么

${\Bbb Y}=\mathop{\bigoplus}\limits_{i=1}^{\infty}{\Bbb Y}_i, \quad {\Bbb Y}_i=\mathop{\bigoplus}\limits_{j=1}^{\dim E(\mu_i)}{\Bbb Y}_{ij}.$

5.1 $E_0$的稳定性

定理5.1$\mathfrak{R}e_0<1,$ 系统(2.1)的无感染平衡态$E_0$是局部渐近稳定的. 若$\mathfrak{R}e_0>1,$$E_0$是不稳定的.

系统(2.1)在$E_0$处的线性化形式为

$\begin{matrix}\frac{\partial u}{\partial t}=\Lambda\Delta u+{\cal J}_0u,\end{matrix}$

其中$u=(T,I,D,V)^T,$$\Lambda={\rm diag}(0,0,d_1,d_2)^T,$ 并且

${\cal J}_0=\left(\begin{array}{ccccc}-\mu & -\frac{{\rm \partial} h_2(\frac{s}{\mu},0)}{{\rm \partial} I} & 0 & -\frac{{\rm \partial} h_1(\frac{s}{\mu},0)}{{\rm \partial} V}\\[3mm]0 & \frac{{\rm \partial} h_2(\frac{s}{\mu},0)}{{\rm \partial} I}-\delta & 0 & \frac{{\rm \partial} h_1(\frac{s}{\mu},0)}{{\rm \partial} V}\\[2mm]0 & \theta & -\alpha-\delta & 0\\0 & 0 & \alpha & -c\end{array}\right).$

显然, $X_i(i=0,1,2,\cdots )$在线性化下都是不变的. 我们使用指数形式的解$u(x,t)=e^{\lambda t}\Psi(x),$ 其中$\Psi\in X_i,$ 并且满足$\Delta\Psi=-\mu_i\Psi.$ 对指数形式的解两边关于$t$求偏导, 得到

${{\rm \partial} u}/{{\rm \partial} t}=\lambda e^{\lambda t}\Psi(x)=\lambda u.$

将其代入(5.1)式得到$\lambda u=-\mu_i\Lambda u+{\cal J}_0u.$简要整理, 我们得到

$\begin{matrix}(\lambda E-{\cal J}_0+\mu_i\Lambda)u=0.\end{matrix}$

考虑到方程组(5.2)有非平凡解, 根据克拉默法则, (5.2)的系数行列式$\det(\lambda E-{\cal J}_0+\mu_i\Lambda)=0$. 通过计算行列式得到系统(2.1)在$E_0$处的特征方程为

$\begin{matrix}(\lambda+\mu)(\lambda^3+A_2\lambda^2+A_1\lambda+A_0)=0,\end{matrix}$

其中

$A_2=\bigg(\delta-\frac{{\rm \partial} h_2(\frac{s}{\mu},0)}{{\rm \partial} I}\bigg)+(\alpha+\delta+c)+\mu_i(d_1+d_2),$
$A_1=\bigg(\delta-\frac{{\rm \partial} h_2(\frac{s}{\mu},0)}{{\rm \partial} I}\bigg)\bigg[(\alpha+\delta+\mu_id_1)+(c+\mu_id_2)\bigg]+(\alpha+\delta+\mu_id_1)(c+\mu_id_2),$
$A_0=\bigg(\delta-\frac{{\rm \partial} h_2(\frac{s}{\mu},0)}{{\rm \partial} I}\bigg)\bigg(\alpha+\delta+\mu_id_1\bigg)\bigg(c+\mu_id_2\bigg)-\frac{{\rm \partial} h_1(\frac{s}{\mu},0)}{{\rm \partial} V}\alpha\theta.$

显然, 特征方程(5.3)有一个负实特征根. 由$\mathfrak{R}e_0<1,$ 得到$\frac{\alpha\theta}{\delta c(\alpha+\delta)}\frac{{\rm \partial} h_1(\frac{s}{\mu},0)}{{\rm \partial} V}<1$$\frac{1}{\delta}\frac{{\rm \partial} h_2(\frac{s}{\mu},0)}{{\rm \partial} I}<1.$ 可将其分别变形为$\alpha\theta\frac{{\rm \partial} h_1(\frac{s}{\mu},0)}{{\rm \partial} V}<\delta c(\alpha+\delta)$$\delta-\frac{{\rm \partial} h_2(\frac{s}{\mu},0)}{{\rm \partial} I}>0.$ 显然得到$A_1,A_2>0.$

$\begin{aligned}A_0 & =\delta\left(\alpha+\delta+\mu_i d_1\right)\left(c+\mu_i d_2\right)-\frac{\partial h_2\left(\frac{s}{\mu}, 0\right)}{\partial I}\left(\alpha+\delta+\mu_i d_1\right)\left(c+\mu_i d_2\right)-\frac{\partial h_1\left(\frac{s}{\mu}, 0\right)}{\partial V} \theta \alpha \\& =\delta\left(\alpha+\delta+\mu_i d_1\right)\left(c+\mu_i d_2\right)\left[1-\frac{1}{\delta} \frac{\partial h_2\left(\frac{s}{\mu}, 0\right)}{\partial I}-\frac{\alpha \theta}{\delta\left(\alpha+\delta+\mu_i d_1\right)\left(c+\mu_i d_2\right)} \frac{\partial h_1\left(\frac{s}{\mu}, 0\right)}{\partial V}\right] \\& >\delta\left(\alpha+\delta+\mu_i d_1\right)\left(c+\mu_i d_2\right)\left[1-\frac{1}{\delta} \frac{\partial h_2\left(\frac{s}{\mu}, 0\right)}{\partial I}-\frac{\alpha \theta}{\delta c(\alpha+\delta)} \frac{\partial h_1\left(\frac{s}{\mu}, 0\right)}{\partial V}\right] \\& =\delta\left(\alpha+\delta+\mu_i d_1\right)\left(c+\mu_i d_2\right)\left(1-\Re_0\right)>0,\end{aligned}$
$\begin{matrix}A_2A_1-A_0&=&\bigg(\delta-\frac{{\rm \partial} h_2(\frac{s}{\mu},0)}{{\rm \partial} I}\bigg)\bigg[(\alpha+\delta+\mu_id_1)+(c+\mu_id_2)\bigg]A_2\\&&+\bigg(\delta-\frac{{\rm \partial} h_2(\frac{s}{\mu},0)}{{\rm \partial} I}\bigg)\bigg(\alpha+\delta+\mu_id_1\bigg)\bigg(c+\mu_id_2\bigg)\\&&+[(\alpha+\delta+c)+\mu_i(d_1+d_2)](\alpha+\delta+\mu_id_1)(c+\mu_id_2)+\frac{{\rm \partial} h_1(\frac{s}{\mu},0)}{{\rm \partial} V}\alpha\theta>0.\end{matrix}$

根据Routh-Hurwitz判据, 当$\mathfrak{R}e_0<1$时, 特征方程(5.3)的所有根都有负实部, 故无感染平衡态$E_0$是局部渐近稳定的. 当$\mathfrak{R}e_0>1$时, 考虑到$i$确定了扩散特征值$\mu_i,$ 故定义

$g(\lambda,i)=\lambda^3+A_2\lambda^2+A_1\lambda+A_0.$

显然, 我们有$g(0,0)=A_0<0$$\mathop{\lim}\limits_{\lambda\rightarrow+\infty}g(\lambda,0)=+\infty.$ 根据广义的零点存在性定理, $g(\lambda,i)=0$至少存在一个$\lambda_0>0,$ 使得$g(\lambda_0,0)=0.$ 因此,当$\mathfrak{R}e_0>1$时, $E_0$是不稳定的.证毕.

定理5.2$\mathfrak{R}e_0<1$时, 系统(2.1)的无感染平衡态$E_0$是全局渐近稳定的.

定义一个Lyapunov泛函

$\begin{matrix}{\cal L}(t)=\int_{\Omega}^{}\bigg(c(\alpha+\delta)I(x,t)+\alpha\frac{{\rm \partial} h_1(\frac{s}{\mu},0)}{{\rm \partial} V}D(x,t)+(\alpha+\delta)\frac{{\rm \partial} h_1(\frac{s}{\mu},0)}{{\rm \partial} V}V(x,t)\bigg){\rm d}x.\end{matrix}$

${\cal L}(t)$沿着系统(2.1)的轨线求导数, 得到

$\begin{matrix}\frac{{\rm d}{\cal L}(t)}{{\rm d}t}&=&\int_{\Omega}^{}\bigg(c(\alpha+\delta)\frac{{\rm \partial} I}{{\rm \partial} t}+\alpha\frac{{\rm \partial} h_1(\frac{s}{\mu},0)}{{\rm \partial} V}\frac{{\rm \partial} D}{{\rm \partial} t}+(\alpha+\delta)\frac{{\rm \partial} h_1(\frac{s}{\mu},0)}{{\rm \partial} V}\frac{{\rm \partial} V}{{\rm \partial} t}\bigg){\rm d}x\\&=&\delta c(\alpha+\delta)\int_{\Omega}^{}\bigg(\frac{1}{\delta}\frac{h_2(T,I)}{I}+\frac{\alpha\theta}{\delta c(\alpha+\delta)}\frac{{\rm \partial} h_1(\frac{s}{\mu},0)}{{\rm \partial} V}-1\bigg)I{\rm d}x\\&&+c(\alpha+\delta)\int_{\Omega}^{}\bigg(\frac{h_1(T,V)}{V}-\frac{{\rm \partial} h_1(\frac{s}{\mu},0)}{{\rm \partial} V}\bigg)V{\rm d}x\\&&+\alpha d_1\frac{{\rm \partial} h_1(\frac{s}{\mu},0)}{{\rm \partial} V}\int_{\Omega}^{}\Delta D{\rm d}x+(\alpha+\delta)d_2\frac{{\rm \partial} h_1(\frac{s}{\mu},0)}{{\rm \partial} V}\int_{\Omega}^{}\Delta V{\rm d}x.\end{matrix}$

根据高斯散度定理和齐次Neumann边界条件(2.3), 可以得到

$\int_{\Omega}^{}\Delta D {\rm d}x=\int_{{\rm \partial}\Omega}^{}\frac{{\rm \partial} D}{{\rm \partial} n}=0, \quad \int_{\Omega}^{}\Delta V {\rm d}x=\int_{{\rm \partial}\Omega}^{}\frac{{\rm \partial} V}{{\rm \partial} n}=0.$

由假设2.1, 得到${h_1(s/\mu,V)}/{V}$是关于$V$的减函数, 同理${h_2(s/\mu,I)}/{I}$是关于$I$的减函数. 因此

$\begin{matrix}\frac{{\rm d}{\cal L}(t)}{{\rm d}t}&\leq &\delta c(\alpha+\delta)\int_{\Omega}^{}\bigg(\frac{1}{\delta}\frac{h_2(\frac{s}{\mu},I)}{I}+\frac{\alpha\theta}{\delta c(\alpha+\delta)}\frac{{\rm \partial} h_1(\frac{s}{\mu},0)}{{\rm \partial} V}-1\bigg)I{\rm d}x\\&&+c(\alpha+\delta)\int_{\Omega}^{}\bigg(\frac{h_1(\frac{s}{\mu},V)}{V}-\frac{{\rm \partial} h_1(\frac{s}{\mu},0)}{{\rm \partial} V}\bigg)V{\rm d}x\\&\leq& \delta c(\alpha+\delta)\int_{\Omega}^{}\bigg(\frac{1}{\delta}\lim\limits_{I\rightarrow 0^+}\frac{h_2(\frac{s}{\mu},I)}{I}+\frac{\alpha\theta}{\delta c(\alpha+\delta)}\frac{{\rm \partial} h_1(\frac{s}{\mu},0)}{{\rm \partial} V}-1\bigg)I{\rm d}x\\&&+c(\alpha+\delta)\int_{\Omega}^{}\bigg(\lim\limits_{V\rightarrow 0^+}\frac{h_1(\frac{s}{\mu},V)}{V}-\frac{{\rm \partial} h_1(\frac{s}{\mu},0)}{{\rm \partial} V}\bigg)V{\rm d}x\\&=&\delta c(\alpha+\delta)\int_{\Omega}^{}\bigg(\frac{1}{\delta}\frac{{\rm \partial} h_2(\frac{s}{\mu},0)}{{\rm \partial} I}+\frac{\alpha\theta}{\delta c(\alpha+\delta)}\frac{{\rm \partial} h_1(\frac{s}{\mu},0)}{{\rm \partial} V}-1\bigg)I{\rm d}x\\&&+c(\alpha+\delta)\int_{\Omega}^{}\bigg(\frac{{\rm \partial} h_1(\frac{s}{\mu},V)}{{\rm \partial} V}-\frac{{\rm \partial} h_1(\frac{s}{\mu},0)}{{\rm \partial} V}\bigg)V{\rm d}x\\&=&\delta c(\alpha+\delta)(R_0-1)\int_{\Omega}^{}I{\rm d}x.\end{matrix}$

因此, 当$\mathfrak{R}e_0<1$时, 我们有${\rm d}{\cal L}(t)/{\rm d}t\leq 0.$显然$\{E_0\}$$\Gamma$中的最大不变集. 根据Lyapunov-LaSalle不变集原理[26], 无感染平衡态$E_0$ 是全局渐近稳定的. 证毕.

5.2 一致持续性

定义5.1 若具有初始条件$u_0\in {\Bbb X}^+$的解$u(x,t;u_0)$是有界的并且远离零点, 即对任意$u_0\in {\Bbb X}^+$都有

$\lim\limits_{t\rightarrow +\infty}\inf T(x,t;T_0)>0,\lim\limits_{t\rightarrow +\infty}\inf I(x,t;I_0)>0,$
$\lim\limits_{t\rightarrow +\infty}\inf D(x,t;D_0)>0,\lim\limits_{t\rightarrow +\infty}\inf V(x,t;V_0)>0,$

则称系统(2.1)在${\Bbb X}^+$上是一致持续的.

定义

$\Gamma_0:=\{\phi=(\phi_1,\phi_2,\phi_3,\phi_4)^T\in X^+ |\phi_2\not\equiv0 \, \mbox{或}\, \phi_3\not\equiv0 \,\mbox{或} \,\phi_4\not\equiv0 \},$
$\partial\Gamma_0:=\Gamma\backslash\Gamma_0=\{\phi=(\phi_1,\phi_2,\phi_3,\phi_4)^T\in X^+| \phi_2\equiv0 \, \mbox{且}\, \phi_3\equiv0 \,\mbox{且} \,\phi_4\equiv0 \},$
$M_\partial:=\{\phi=(\phi_1,\phi_2,\phi_3,\phi_4)^T\in{\rm \partial}\Gamma_{0} |\Phi_t(\phi)\in{\rm \partial}\Gamma_{0},\forall t\geq0\},$

其中$\omega(\phi)$表示轨$O^+(\phi):=\{\Phi_t(\phi):t>0\}$的极限集, 其中$\{\Phi_t\}_{t\geq0}$定义为系统(2.1)的解半流.

引理5.1[22]$\sigma$$\alpha$都是$\Omega$内的非负常数, 则下述系统

$\begin{matrix}\frac{{\rm \partial} \Psi(x,t)}{{\rm \partial} t}=\alpha-\sigma\Psi(x,t),\quad x\in\Omega,\quad t>0\end{matrix}$

$C(\overline\Omega,{\Bbb R})$上有一个恒为正的平衡态$\alpha/\sigma,$ 同时该平衡态在$C(\overline\Omega,R)$上是全局渐近稳定的.

定理5.3 对于系统(2.1), 有下述的三条结论成立.

(I) 对于任意$u_0 \in M_\partial,$ 都有$\omega(u_0)=\{E_0\}.$

(II) 当$x\in \overline\Omega,t>0$时, 对于任意$u_0 \in\Gamma_0,$ 都有以下四个不等式成立

$T(x,t;T_0)>0,\quad I(x,t;I_0)>0,\quad D(x,t;D_0)>0,\quad V(x,t;V_0)>0.$

(III) 若$\mathfrak{R}e_0<1,$ 则无感染平衡态$E_0=(s/\mu,0,0,0)$$\Gamma_0$上是全局吸引的. 若$\mathfrak{R}e_0>1,$ 则无感染平衡态$E_0$$\Gamma_0$的一致弱排斥子. 即存在一个任意小的$\varepsilon>0,$ 使得

$\lim\limits_{t\rightarrow +\infty}\sup\|\Phi_t(u_0)-(s/\mu,0,0,0)\|_X\geq\varepsilon,\quad \forall u_0\in\Gamma_0.$

(I) 假设$u_0\in M_{\rm \partial},$ 则系统(2.1)退化为

$\begin{matrix}\frac{{\rm \partial} T(x,t)}{{\rm \partial} t}=s-\mu T(x,t),\quad I(x,t)=D(x,t)=V(x,t)=0.\end{matrix}$

对(5.5)式积分, 则

$\begin{matrix}T(x,t)=T_{0}e^{-\mu t}+s\int_{0}^{t}e^{-\mu(t-r)}{\rm d}r=\frac{s}{\mu}+\left(T_0(x)-\frac{s}{\mu}\right)e^{-\mu t}.\end{matrix}$

因此, 当$t\rightarrow +\infty,$ 对于任意$x\in \overline\Omega,$ 都有$T(x,t)\rightarrow s/\mu.$$\omega(u_0)=\{(s/\mu,0,0,0)\}=\{E_0\}.$

(II) 首先证明对于任意$t>0,x\in\overline\Omega,$ 存在常数$\xi>0,$ 使得$\mathop{\lim}\limits_{t\rightarrow+\infty}\inf T(x,t;T_0)\geq\xi.$ 根据假设2.1和系统(2.1)的第一个方程, 我们有

$\begin{matrix}\frac{{\rm \partial} T}{{\rm \partial} t}\geq s-\mu T-(\eta_1+\eta_2)T, \quad t>0,x\in\overline\Omega.\end{matrix}$

由引理5.1, 令$\Psi(x,t)=\overline{T}(x,t),$$\alpha=s, \sigma=\mu+\eta_1+\eta_2,$ 得到了新的系统

$\begin{matrix}\frac{{\rm \partial} \overline{T}(x,t)}{{\rm \partial} t}= s-(\mu+\eta_1+\eta_2)\overline{T}(x,t), \quad t>0,x\in\overline\Omega.\end{matrix}$

则系统(5.7)存在唯一正平衡态$s/(\mu+\eta_1+\eta_2),$ 并且在空间$C(\overline\Omega,\mathbb{R})$上是全局渐近稳定的. 根据比较原理和式(5.6), 存在常数$\xi>0,$ 使得$\mathop{\lim}\limits_{t\rightarrow+\infty}\inf T(x,t;T_0)\geq\xi.$

接下来分三种情形证明不等式成立. 若$I_0(x)\not\equiv0.$ 由系统(2.1) 的第二个方程得到

$\begin{matrix}I(x,t)=e^{-\delta t}I_0(x)+\int_{0}^{t}e^{-\delta(t-s)}[h_1(T(x,s),V(x,s))+h_2(T(x,s),I(x,s))]{\rm d}s,\end{matrix}$

显然对于任意$t>0,x\in\overline\Omega,$$I(x,t)>0.$ 由系统(2.1)的第三个方程, $D(x,t)$可写成

$\begin{matrix}D(x,t)=T_3(t)D_0(x)+\theta\int_{0}^{t}T_3(t)I(x,s){\rm d}s,\end{matrix}$

其中$\{T_3(t)\}_{t\geq0}$表示由$d_1\Delta-(\alpha+\delta)$生成的$C_0$ -半群. 根据拉普拉斯算子生成的$C_0$ -半群的性质得到, $T_3(t)$是非负的, 故对所有的$t>0,x\in\overline\Omega,$ 都有$D(x,t)>0.$ 同样地, 由系统(2.1)的第四个方程, $D(x,t)$可写成如下形式

$\begin{matrix}V(x,t)=T_4(t)V_0(x)+c\int_{0}^{t}T_4(t)D(x,s){\rm d}s,\end{matrix}$

其中$\{T_4(t)\}_{t\geq0}$表示由$d_2\Delta-\alpha$生成的$C_0$ -半群, 故$T_4(t)$ 是非负的, 即对所有的$t>0,x\in\overline\Omega,$ 都有$V(x,t)>0.$

$D_0(x)\not\equiv0.$ 根据系统(2.1)的非负性与第三个方程, ${{\rm \partial} D}/{{\rm \partial} t}\geq d_1\Delta D-(\alpha+\delta)D,$$D(x,t)$为下述系统的上解

$\begin{matrix}\left\{\begin{array}{ll}\frac{{\rm \partial} \widehat{D}(x,t)}{{\rm \partial} t}=d_1\Delta \widehat{D}(x,t)-(\alpha+\delta)\widehat{D}(x,t), \quad& t>0,x\in\overline\Omega,\\[3mm]\frac{{\rm \partial} \widehat{D}(x,t)}{{\rm \partial} n}=0, & t>0,x\in{\rm \partial}\Omega,\\[2mm]\widehat{D}(x,0)=D_0(x)\not\equiv0,& x\in\overline\Omega.\end{array}\right.\end{matrix}$

由最大值原理和比较原理[23], 对于所有的$t>0,x\in\overline\Omega,$$D(x,t)\geq\widehat{D}(x,t)>0.$ 同理, 对任意$t>0,x\in\overline\Omega,$ 可得到$V(x,t)>0.$ 对于系统(2.1)的第二个方程, ${{\rm \partial} I}/{{\rm \partial} t}\geq h_1(T,V)-\delta I,$$h_1(T,V)$是严格正的, 故对所有的$t>0,x\in\overline\Omega,$ 都有$I(x,t)>0.$

$V_0(x)\not\equiv0.$ 由等式(5.9),有

$V(x,t)=T_4(t)V_0(x)+c\int_{0}^{t}T_4(t)D(x,s){\rm d}s\geq T_4(t)V_0(x)>0.$

结合$T(x,t)$$V(x,t)$的正性, 再根据等式(5.8), 显然得到对任意的$t>0,x\in\overline\Omega,$$I(x,t)>0.$ 将结果代入系统(2.1)的第三个方程, 得到

$\begin{matrix}\left\{\begin{array}{ll}\frac{{\rm \partial} D(x,t)}{{\rm \partial} t}\geq d_1\Delta D(x,t)-(\alpha+\delta)D(x,t), \quad& t>0,x\in\overline\Omega,\\[3mm]\frac{{\rm \partial} D(x,t)}{{\rm \partial} n}=0, & t>0,x\in{\rm \partial}\Omega.\end{array}\right.\end{matrix}$

由最大值原理和Hopf边界引理[23], 对所有的$t>0,x\in\overline\Omega,$ 得到$D(x,t)>0.$

(III) 根据文献[推论4.3.2], $\lambda(s/\mu)=\max\{{\rm Re}\lambda|\lambda\in\sigma({\cal J}_0-\mu_i\Lambda)\}$

是矩阵${\cal J}-\mu_i\Lambda$的主特征值, 其中$\sigma({\cal J}_0-\mu_i\Lambda)$是矩阵所有特征值的集合.

由定理5.1, 当$\mathfrak{R}e_0<1$时, 主特征值$\lambda(s/\mu)<0,$ 则存在一个充分小的$\varepsilon>0,$ 使得$\lambda(s/\mu+\varepsilon)<0.$ 进一步地, $T(x,t)$ 满足

${{\rm \partial} T(x,t)}/{{\rm \partial} t}\leq s-\mu T(x,t), t>0,x\in\overline\Omega.$

由比较原理, 存在$t_0>0,$ 使得

$T(x,t)\leq \frac{s}{\mu}+\varepsilon, \quad t\geq t_0, x\in\overline\Omega.$

故, 对所有的$t\geq t_0$都有

$\begin{matrix}\left\{\begin{array}{ll}\frac{{\rm \partial} I(x,t)}{{\rm \partial} t}\leq h_1(\frac{s}{\mu}+\varepsilon,V)+h_2(\frac{s}{\mu}+\varepsilon,I)-\delta I, \quad& t\geq t_0,x\in\overline\Omega,\\[3mm]\frac{{\rm \partial} D(x,t)}{{\rm \partial} t}=d_1\Delta D+\theta I-(\alpha+\delta)D, & t\geq t_0,x\in\overline\Omega,\\[3mm]\frac{{\rm \partial} V(x,t)}{{\rm \partial} t}=d_2\Delta V+\alpha D-cV, & t\geq t_0,x\in\overline\Omega,\\[3mm]\frac{{\rm \partial} D(x,t)}{{\rm \partial} n}=\frac{{\rm \partial} V(x,t)}{{\rm \partial} n}=0, & t\geq t_0,x\in{\rm \partial}\Omega.\end{array}\right.\end{matrix}$

可以得到下述的比较系统

$\begin{matrix}\left\{\begin{array}{ll}\frac{{\rm \partial} \widehat{I}(x,t)}{{\rm \partial} t}= h_1(\frac{s}{\mu}+\varepsilon,\widehat{V})+h_2(\frac{s}{\mu}+\varepsilon,\widehat{I})-\delta \widehat{I}, \quad & t\geq t_0,x\in\overline\Omega,\\[3mm]\frac{{\rm \partial} \widehat{D}(x,t)}{{\rm \partial} t}=d_1\Delta \widehat{D}+\theta \widehat{I}-(\alpha+\delta)\widehat{D},& t\geq t_0,x\in\overline\Omega,\\[3mm]\frac{{\rm \partial} \widehat{V}(x,t)}{{\rm \partial} t}=d_2\Delta \widehat{V}+\alpha\widehat{ D}-c\widehat{V}, & t\geq t_0,x\in\overline\Omega,\\[3mm]\frac{{\rm \partial} \widehat{D}(x,t)}{{\rm \partial} n}=\frac{{\rm \partial} \widehat{V}(x,t)}{{\rm \partial} n}=0, & t\geq t_0,x\in{\rm \partial}\Omega.\end{array}\right.\end{matrix}$

$\widehat{\xi}:=(\widehat{\xi}_2,\widehat{\xi}_3,\widehat{\xi}_4)$ 是系统(5.10)对应于主特征值$\lambda(s/\mu+\varepsilon)$的正特征向量, 则对于所有的$t\geq t_0,x\in\overline\Omega,$ 系统(5.10)的解

$(\widehat{I}(x,t),\widehat{D}(x,t),\widehat{V}(x,t))=e^{\lambda(s/\mu+\varepsilon)t}(\widehat{\xi}_2(x),\widehat{\xi}_3(x),\widehat{\xi}_4(x)).$

由于对于任意给定的$u_0\in \Gamma_0,$ 存在$\varsigma>0,$ 使得

$(I(\cdot,t_0),D(\cdot,t_0),V(\cdot,t_0))\leq \varsigma (\widehat{I}(\cdot,t_0),\widehat{D}(\cdot,t_0),\widehat{V}(\cdot,t_0)).$

根据比较原理, 可以得到

$\begin{matrix}(I(\cdot,t),D(\cdot,t),V(\cdot,t))\leq \varsigma e^{\lambda(s/\mu+\varepsilon)t}(\widehat{\xi}_2(x),\widehat{\xi}_3(x),\widehat{\xi}_4(x)), \quad \forall t\geq t_0.\end{matrix}$

$\lambda(s/\mu+\varepsilon)<0,$ 得到

$\mathop{\lim}\limits_{t\rightarrow +\infty}(I(x,t),D(x,t),V(x,t))=0, \, x\in\overline\Omega.$

并且系统(2.1)的第一个方程渐近于系统

$\begin{matrix}\frac{{\rm \partial} \widehat{T}(x,t)}{{\rm \partial} t}=s-\mu \widehat{T}(x,t), \quad t>0,x\in\overline\Omega.\end{matrix}$

由渐近自治半流理论(见文献[推论4.3]), 得到

$\mathop{\lim}\limits_{t\rightarrow +\infty}T(x,t)=s/\mu, \, x\in\overline\Omega.$

下证当$\mathfrak{R}e_0>1$时, $E_0$$\Gamma_0$上是弱排斥的. 反证法. 对$\forall \, u_0\in\Gamma_0$和充分小的$\epsilon>0,$ 假设 $\mathop{\lim}\limits_{t\rightarrow +\infty}\|\Phi_t(u_0)-E_0\|<\epsilon.$ 这意味着存在$t_1>0,$ 使得

$T(x,t)>\frac{s}{\mu}-\epsilon,\, I(x,t)<\epsilon, \, D(x,t)<\epsilon, \, V(x,t)<\epsilon, \quad\forall \, t\geq t_1,x\in\overline\Omega.$

接下来, 构造一个新的系统

$\begin{matrix}\left\{\begin{array}{ll}\frac{{\rm \partial} \widetilde{I}(x,t)}{{\rm \partial} t}= h_1(\frac{s}{\mu}-\epsilon,\widetilde{V})+h_2(\frac{s}{\mu}-\epsilon,\widetilde{I})-\delta \widetilde{I}, \quad &t\geq t_1,x\in\overline\Omega,\\[3mm]\frac{{\rm \partial} \widetilde{D}(x,t)}{{\rm \partial} t}=d_1\Delta \widetilde{D}+\theta \widetilde{I}-(\alpha+\delta)\widetilde{D}, & t\geq t_1,x\in\overline\Omega,\\[3mm]\frac{{\rm \partial} \widetilde{V}(x,t)}{{\rm \partial} t}=d_2\Delta \widetilde{V}+\alpha \widetilde{D}-c\widetilde{V}, & t\geq t_1,x\in\overline\Omega,\\[3mm]\frac{{\rm \partial} \widetilde{D}(x,t)}{{\rm \partial} n}=\frac{{\rm \partial} \widetilde{V}(x,t)}{{\rm \partial} n}=0, & t\geq t_1,x\in{\rm \partial}\Omega.\end{array}\right.\end{matrix}$

$\widetilde{\xi}:=(\widetilde{\xi}_2,\widetilde{\xi}_3,\widetilde{\xi}_4)$ 是系统(5.11)对应于主特征值$\lambda(s/\mu-\epsilon)$的正特征向量, 则系统(5.11)的解为$(\widetilde{I}(x,t),\widetilde{D}(x,t),\widetilde{V}(x,t))=e^{\lambda(s/\mu-\epsilon)t}(\widetilde{\xi}_2,\widetilde{\xi}_3,\widetilde{\xi}_4).$ 根据定理5.1, 当$\mathfrak{R}e_0>1$时, 主特征值$\lambda(s/\mu)>0,$ 由主特征值的连续性以及充分小的数$\epsilon$, 可得到$\lambda(s/\mu-\epsilon)>0.$ 对于任意给定的$u_0\in \Gamma_0,$ 存在$\zeta>0,$ 使得$(I(\cdot,t_1),D(\cdot,t_1),V(\cdot,t_1))\geq\zeta\widehat{u}(\cdot,t_1).$ 再根据比较原理[23], 得到

$\begin{matrix}(I(x,t),D(x,t),V(x,t))\geq\zeta e^{\lambda(s/\mu-\epsilon)t}(\widetilde{\xi}_2(x),\widetilde{\xi}_3(x),\widetilde{\xi}_4(x)), \quad \forall t\geq t_1, x\in\overline\Omega.\end{matrix}$

显然, $I(x,t),D(x,t),V(x,t)$都是无界的, 与假设矛盾. 故原命题成立.

定理5.4$\mathfrak{R}e_0>1$时, 系统(2.1)在$\Gamma_0$上是一致持续的.

$M=\{(s/\mu,0,0,0)\},$ 根据引理5.1, 我们得到$M_{\rm \partial}$${\rm \partial} \Gamma_0$的最大紧集, 由定理5.3中(III)得到$E_0=(s/\mu,0,0,0)$$\Gamma$上是孤立的, 并且$\Gamma_0\bigcap W^s(E_0)=\emptyset,$ 其中$W^s(E_0)$表示$E_0$的稳定流形. 故在$M_{\rm \partial}$中不存在$E_0$到自身的循环. 根据文献[定理3], 存在一个常数$\eta>0,$ 使得

$\min_{\phi\in\omega(u_0)}\Big[\min_{i=1,2,3,4}\inf_{x\in\overline\Omega}\phi_i(x)\Big]>\eta.$

即对所有的$u_0\in\Gamma_0,$ 都有

$\lim\limits_{t\rightarrow +\infty}\inf T(x,t;T_0)>\eta,\lim\limits_{t\rightarrow +\infty}\inf I(x,t;I_0)>\eta,$
$\lim\limits_{t\rightarrow +\infty}\inf D(x,t;D_0)>\eta,\lim\limits_{t\rightarrow +\infty}\inf V(x,t;V_0)>\eta.$

因此, 系统(2.1)是一致持续的.证毕.

5.3 $E^*$的稳定性

本小节我们将通过构造Lyapunov函数证明当$\mathfrak{R}e_0>1$时, 系统(2.1)的感染平衡态$E^*$的全局渐近稳定的. 为此需要引入函数$G(x)=x-1-\ln x.$ 显然$x=1$$G(x)$唯一的极值点, 并且对于任意$x>0,$ 都有$G(x)\geq0.$ 进一步地, 我们做如下假设.

假设5.1 非线性一般发生率函数$h_1(T,V)$$h_2(T,I)$分别满足下述不等式

$\begin{matrix}\left(1-\frac{T^*h_1}{Th_1^*}\right)\left(\frac{V}{V^*}-\frac{T^*h_1}{Th_1^*}\right)\leq0, \quad \forall \, T, V>0;\end{matrix}$
$\begin{matrix}\left(1-\frac{T^*h_2}{Th_2^*}\right)\left(\frac{I}{I^*}-\frac{T^*h_2}{Th_2^*}\right)\leq0, \quad \forall \, T, I>0.\end{matrix}$

定理5.5$\mathfrak{R}e_0>1$时, 系统(2.1)的感染平衡态$E^*$是全局渐近稳定的.

定义一个Lyapunov泛函

$\begin{matrix}\widetilde{{\cal L}}(t)=\int_{\Omega}^{}\bigg(T^*G\bigg(\frac{T}{T^*}\bigg)+I^*G\bigg(\frac{I}{I^*}\bigg)+\frac{h_1^*}{\theta I^*}D^*G\bigg(\frac{D}{D^*}\bigg)+\frac{h_1^*}{\alpha D^*}V^*G\bigg(\frac{V}{V^*}\bigg)\bigg){\rm d}x,\end{matrix}$

且平衡态满足以下等式

$\begin{matrix}s=h_1^*+h_2^*+\mu T^*, \quad\delta I^*=h_1^*+h_2^*, \quad\theta I^*=(\alpha+\delta)D^*,\quad\alpha D^*=c V^*.\end{matrix}$

$\widetilde{{\cal L}}(t)$沿着系统(2.1)的轨线求导数, 得到

$\begin{matrix}\frac{{\rm d} \widetilde{{\cal L}}(t)}{{\rm d}t}&=&\int_{\Omega}^{}\bigg[\bigg(1-\frac{T^*}{T}\bigg)\frac{{\rm \partial} T}{{\rm \partial} t}+\bigg(1-\frac{I^*}{I}\bigg)\frac{{\rm \partial} I}{{\rm \partial} t}+\frac{h_1^*}{\theta I^*}\bigg(1-\frac{D^*}{D}\bigg)\frac{{\rm \partial} D}{{\rm \partial} t}+\frac{h_1^*}{\alpha D^*}\bigg(1-\frac{V^*}{V}\bigg)\frac{{\rm \partial} V}{{\rm \partial} t}\bigg]{\rm d}x\\&=&\int_{\Omega}^{}\bigg[\bigg(1-\frac{T^*}{T}\bigg)\bigg(\mu T^*-\mu T+h_1^*-h_1+h_2^*-h_2\bigg)\\&&+\bigg(1-\frac{I^*}{I}\bigg)\bigg(h_1+h_2-\frac{(h_1^*+h_2^*)I}{I^*}\bigg)+h_1^*\bigg(1-\frac{D^*}{D}\bigg)\bigg(\frac{I}{I^*}-\frac{D}{D^*}\bigg)\\&&+h_1^*\bigg(1-\frac{V^*}{V}\bigg)\bigg(\frac{D}{D^*}-\frac{V}{V^*}\bigg) \bigg]{\rm d}x\\&&+\frac{h_1^*d_1}{\theta I^*}\int_{\Omega}^{}\bigg(1-\frac{D^*}{D}\bigg)\Delta D{\rm d}x+\frac{h_1^*d_2}{\alpha D^*}\int_{\Omega}^{}\bigg(1-\frac{V^*}{V}\bigg)\Delta V{\rm d}x\\&=&-\int_{\Omega}^{}\bigg(\frac{\mu(T-T^*)^2}{T}\bigg){\rm d}x+\int_{\Omega}^{}(h_1^*C_1+h_2^*C_2){\rm d}x+\frac{h_1^*d_1}{\theta I^*}\int_{\Omega}^{}\bigg(1-\frac{D^*}{D}\bigg)\Delta D{\rm d}x\\&&+\frac{h_1^*d_2}{\alpha D^*}\int_{\Omega}^{}\bigg(1-\frac{V^*}{V}\bigg)\Delta V{\rm d}x,\end{matrix}$

其中

$\begin{matrix}C_1&=&\bigg[\left(1-\frac{T^*}{T}\right)\left(1-\frac{h_1}{h_1^*}\right)+\left(1-\frac{I^*}{I}\right)\left(\frac{h_1}{h_1^*}-\frac{I}{I^*}\right)+\left(1-\frac{D^*}{D}\right)\left(\frac{I}{I^*}-\frac{D}{D^*}\right)\\&&+\left(1-\frac{V^*}{V}\right)\left(\frac{D}{D^*}-\frac{V}{V^*}\right)\bigg]\\&=&4+\frac{T^*h_1}{Th_1^*}-\frac{T^*}{T}-\frac{V}{V^*}-\frac{V^*D}{VD^*}-\frac{I^*h_1}{Ih_1^*}-\frac{D^*I}{DI^*}\\&=&G\bigg(\frac{T^*h_1}{Th_1^*}\bigg)-G\bigg(\frac{T^*}{T}\bigg)-G\bigg(\frac{V}{V^*}\bigg)-G\bigg(\frac{V^*D}{VD^*}\bigg)-G\bigg(\frac{ I^*h_1}{Ih_1^*}\bigg)-G\bigg(\frac{D^*I}{DI^*}\bigg),\end{matrix}$
$\begin{matrix}C_2&=&\bigg(1-\frac{T^*}{T}\bigg)\bigg(1-\frac{h_2}{h_2^*}\bigg)+\bigg(1-\frac{I^*}{I}\bigg)\bigg(\frac{h_2}{h_2^*}-\frac{I}{I^*}\bigg)\\&=&G\bigg(\frac{T^*h_2}{Th_2^*}\bigg)-G\bigg(\frac{T^*}{T}\bigg)-G\bigg(\frac{I^*h_2}{Ih_2^*}\bigg)-G\bigg(\frac{I}{I^*}\bigg).\end{matrix}$

使用齐次Neumann边界条件(2.3)和散度定理, 可以推出

$\begin{matrix}\int_{\Omega}^{}\left(1-\frac{D^*}{D}\right)\Delta D{\rm d}x&=&\int_{\Omega}^{}\Delta D{\rm d}x-\int_{\Omega}^{}\frac{D^*}{D}\Delta D{\rm d}x\\&=&\int_{{\rm \partial}\Omega}^{}\frac{{\rm \partial} D}{{\rm \partial} n}{\rm d}x-D^*\int_{\Omega}^{}\frac{\|\nabla D\|^2}{D^2}{\rm d}x\\&=&-D^*\int_{\Omega}^{}\frac{\|\nabla D\|^2}{D^2}{\rm d}x<0.\end{matrix}$

同样地, 我们得到

$\int_{\Omega}^{}\left(1-\frac{V^*}{V}\right)\Delta V{\rm d}x=-V^*\int_{\Omega}^{}\frac{\|\nabla V\|^2}{V^2}{\rm d}x<0.$

$\frac{{\rm d} \widetilde{{\cal L}}(t)}{{\rm d}t}\leq-\int_{\Omega}^{}\bigg(\frac{\mu(T-T^*)^2}{T}\bigg){\rm d}x+\int_{\Omega}^{}(h_1^*C_1+h_2^*C_2){\rm d}x.$

最后根据假设2.1以及$G(x)$$x=1$两侧的单调性, 可以得到

$\begin{matrix}G\left(\frac{T^*h_1}{Th_1^*}\right)\leq G\left(\frac{V}{V^*}\right), \quad G\left(\frac{T^*h_2}{Th_2^*}\right)\leq G\left(\frac{I}{I^*}\right).\end{matrix}$

因此

$\begin{matrix}\frac{{\rm d} \widetilde{{\cal L}}(t)}{{\rm d}t}&\leq&-\int_{\Omega}^{}\frac{\mu(T-T^*)^2}{T}{\rm d}x\\&&-h_1^*\int_{\Omega}^{}\bigg(G\bigg(\frac{T^*}{T}\bigg)+G\bigg(\frac{V^*D}{VD^*}\bigg)+G\bigg(\frac{ I^*h_1}{Ih_1^*}\bigg)+G\bigg(\frac{D^*I}{DI^*}\bigg)\bigg){\rm d}x\\&&-h_2^*\int_{\Omega}^{}\bigg(G\bigg(\frac{T^*}{T}\bigg)+G\bigg(\frac{I^*h_2}{Ih_2^*}\bigg)\bigg){\rm d}x\leq0.\end{matrix}$

故, 当$\mathfrak{R}e_0>1$时, 我们有${\rm d}\widetilde{{\cal L}}(t)/{\rm d}t\leq 0.$ 显然$\{E^*\}$$\Gamma$中的最大不变集. 根据Lypunov-LaSalle不变集原理[26], 感染平衡态$E^*$是全局渐近稳定的.证毕.

6 数值模拟

下面, 我们将通过数值模拟验证定理5.2和5.5 的正确性, 并探索扩散系数对HBV感染的影响. 取抽象函数$h_1={\beta_1TV}/{1+aV},$$h_2=\beta_2TI.$ 将函数代入基本再生数$\mathfrak{R}e_0$中, 可得

$\mathfrak{R}e_0=\frac{\alpha\theta\beta_1}{\delta c(\alpha+\delta)}\frac{s}{\mu}+\frac{\beta_2}{\delta}\frac{s}{\mu}.$

结合文献[1], 我们假设病毒的活动范围为有界区域$\Omega=(0,5)\times(0,5),$ 取初值 $(T_0,I_0,D_0,V_0)=(10,0,0.1,0.1).$ 取扩散系数$d_1=d_2=1.0\times10^{-4},$ 其他参数$\mu=0.01,$$\beta_1=0.01,$$a=0.05,$$\beta_2=0.01,$$\delta=1,$$\alpha=0.1,$$\delta=1,$$\theta=1,$$c=0.02.$

$s=0.14$时, 通过计算可得$\mathfrak{R}e_0=0.7764<1.$图1展示了各状态变量的空间分布情况, 图2分别取三组不同的位置研究了各状态变量的时间序列图, 图1图2 分别从空间和时间角度验证了无感染平衡态是全局渐近稳定的. 当取$s=0.75,$ 计算得到$\mathfrak{R}e_0=4.1591>1.$图3展示了各状态变量的空间分布情况, 图4 中我们取三组不同的位置, 得到各状态变量的三组时间序列图, 结合图3图4显然可以观察到感染平衡态$E^*$ 是全局渐近稳定的.

图1

图1   $\mathfrak{R}e_0=0.7764<1,$ 无感染平衡态$E_0=(13.9958,0.0002,0.0008,0.0012)$


图2

图2   $T,I,D,V$空间坐标分别取$(2,2),(2.5,2.5),(3,3)$ 的时间序列图


图3

图3   $\mathfrak{R}e_0=4.1591>1,$ 感染平衡态$E^*=(17.3326,0.5263,0.5748,2.8729)$


图4

图4   $T,I,D,V$空间坐标分别取$(2,2),(2.5,2.5),(3,3)$的时间序列图


为了进一步探究扩散系数对各个状态变量的影响, 我们设置了对照组进行研究. 假设$d_1=d_2=d,$ 并取三组扩散系数分别为: $0,$$1.0\times10^{-6},$$1.0\times10^{-4},$ 其他参数取值和图3相同. 图5 -图8显示了各状态变量取三组扩散系数下的空间分布图, 显然可以发现, 随着扩散系数的增大, 扩散区域也在逐渐增大, 感染的最终状态受空间的影响也更为明显. 这表明扩散系数越大, HBV感染的最终状态越严重.

图5

图5   取不同扩散系数时对状态变量$T$的空间演化图.

(a) $d=0$, (b) $d=1.0\times{ 10^{-6}}$, (c) $d=1.0\times{ 10^{-6}}$


图6

图6   取不同扩散系数时对状态变量$I$的空间演化图.

(a) $d=0$, (b) $d=1.0\times{ 10^{-6}}$, (c) $d=1.0\times{ 10^{-6}}$


图7

图7   取不同扩散系数时对状态变量$D$的空间演化图.

(a) $d=0$, (b) $d=1.0\times{ 10^{-6}}$, (c) $d=1.0\times{ 10^{-6}}$


图8

图8   取不同扩散系数时对状态变量$V$的空间演化图.

(a) $d=0$, (b) $d=1.0\times{ 10^{-6}}$, (c) $d=1.0\times{ 10^{-6}}$


7 结束语

本文建立了一类具有病毒DNA核衣壳和细胞间传播的一般HBV扩散模型,依据下一代矩阵法定义了模型的基本再生数, 证明了模型的一致持续性, 且证明了当$\mathfrak{R}e_0<1$时,无感染平衡态是全局渐近稳定的; 当$\mathfrak{R}e_0>1$时, 感染平衡态是全局渐近稳定的. 模型的全局稳定性排除了震荡或分支等复杂动力学行为的存在, 这表明空间上的局部扩散并不会影响模型的稳定性. 数值模拟的结果显示扩散影响着HBV感染, 扩散系数越大, HBV感染的空间区域越大.

参考文献

Li M L, Zu J.

The review of differential matrix models of HBV infection dynamics

J Virol Methods, 2019, 266: 103-113

DOI:10.1016/j.jviromet.2019.01.014      URL     [本文引用: 2]

Nowak M A, Bonhoeffer S, Hill A M, et al.

Viral dynamics in Hepatitis B virus infection

P Natl Acad Sci USA, 1996, 93(9): 4398-4402

DOI:10.1073/pnas.93.9.4398      URL     [本文引用: 2]

Treatment of chronic hepatitis B virus (HBV) infections with the reverse transcriptase inhibitor lamivudine leads to a rapid decline in plasma viremia and provides estimates for crucial kinetic constants of HBV replication. We find that in persistently infected patients, HBV particles are cleared from the plasma with a half-life of approximately 1.0 day, which implies a 50% daily turnover of the free virus population. Total viral release into the periphery is approximately 10(11) virus particles per day. Although we have no direct measurement of the infected cell mass, we can estimate the turnover rate of these cells in two ways: (i) by comparing the rate of viral production before and after therapy or (ii) from the decline of hepatitis B antigen during treatment. These two independent methods give equivalent results: we find a wide distribution of half-lives for virus-producing cells, ranging from 10 to 100 days in different patients, which may reflect differences in rates of lysis of infected cells by immune responses. Our analysis provides a quantitative understanding of HBV replication dynamics in vivo and has implications for the optimal timing of drug treatment and immunotherapy in chronic HBV infection. This study also represents a comparison for recent findings on the dynamics of human immunodeficiency virus (HIV) infection. The total daily production of plasma virus is, on average, higher in chronic HBV carriers than in HIV-infected patients, but the half-life of virus-producing cells is much shorter in HIV. Most strikingly, there is no indication of drug resistance in HBV-infected patients treated for up to 24 weeks.

Min L, Su Y M, Kuang Y.

Mathematical analysis of a basic virus infection model with application to HBV infection

Rocky Mountain J Math, 2008, 38(5): 1573-1585

[本文引用: 1]

李喜玲, 高飞, 李文琴.

具有免疫时滞的分数阶 HBV 感染模型稳定性分析

数学物理学报, 2021, 41A(2): 562-576

[本文引用: 1]

Li X L, Gao F, Li W Q.

Stability analysis of fractional-order hepatitis B virus infection model with immune delay

Acta Math Sci, 2021, 41A(2): 562-576

[本文引用: 1]

Wang K F, Wang W D.

Propagation of HBV with spatial dependence

Math Biosci, 2007, 210(1): 78-95

PMID:17592736      [本文引用: 1]

A mathematical model is proposed to simulate the hepatitis B virus (HBV) infection with spatial dependence. The existence of traveling waves is established via the geometric singular perturbation method. Numerical simulations show that the model admits non-monotone traveling profiles. Influences of various parameters on the minimum wave speed are also discussed.

Wang K F, Wang W D, Song S P.

Dynamics of an HBV model with diffusion and delay

J Theor Biol, 2008, 253(1): 36-44

PMID:18155252      [本文引用: 2]

In this paper we model and analyze the hepatitis B virus (HBV) infection in a diffusion model confined to a finite domain, induced by intracellular time delay between infection of a cell and production of new virus particles. The equilibrium solutions are obtained and the stability is analyzed if the space is assumed as homogeneous. When the space is inhomogeneous, the effects of diffusion and intracellular time delay are obtained by computer simulations.

Gan Q T, Xu R, Yang P H, et al.

Travelling waves of a hepatitis B virus infection model with spatial diffusion and time delay

IMA J Appl Math, 2010, 75(3): 392-417

DOI:10.1093/imamat/hxq009      URL     [本文引用: 1]

Xu R, Ma Z N.

An HBV model with diffusion and time delay

J Theor Biol, 2009, 257(3): 499-509

DOI:10.1016/j.jtbi.2009.01.001      PMID:19174169      [本文引用: 1]

In this paper, a hepatitis B virus (HBV) model with spatial diffusion and saturation response of the infection rate is investigated, in which the intracellular incubation period is modelled by a discrete time delay. By analyzing the corresponding characteristic equations, the local stability of an infected steady state and an uninfected steady state is discussed. By comparison arguments, it is proved that if the basic reproductive number is less than unity, the uninfected steady state is globally asymptotically stable. If the basic reproductive number is greater than unity, by successively modifying the coupled lower-upper solution pairs, sufficient conditions are obtained for the global stability of the infected steady state. Numerical simulations are carried out to illustrate the main results.

Zhang Y Y, Xu Z T.

Dynamics of a diffusive HBV model with delayed Beddington-Deangelis response

Nonlinear Anal RWA, 2014, 15: 118-139

DOI:10.1016/j.nonrwa.2013.06.005      URL     [本文引用: 1]

杨瑜, 周金玲.

一类具有扩散和Beddington-DeAngelis反应函数的病毒模型的全局稳定性

高校应用数学学报, 2016, 31(2): 161-166

[本文引用: 1]

Yang Y, Zhou J L.

Global stability of a diffusive virus dynamics model with Beddington-DeAngelis incidence function

Appl Math Journal of Chinese Universities, 2016, 31(2): 161-166

[本文引用: 1]

Hattaf K, Yousfi N.

A generalized HBV model with diffusion and two delays

Comput Math Appl, 2015, 69(1): 31-40

[本文引用: 1]

Guidotti L G, Rochford R, Chung J, et al.

Viral clearance without destruction of infected cells during acute HBV infection

Science, 1999, 284: 825-829

DOI:10.1126/science.284.5415.825      PMID:10221919      [本文引用: 1]

Viral clearance during hepatitis B virus (HBV) infection has been thought to reflect the destruction of infected hepatocytes by CD8(+) T lymphocytes. However, in this study, HBV DNA was shown to largely disappear from the liver and the blood of acutely infected chimpanzees long before the peak of T cell infiltration and most of the liver disease. These results demonstrate that noncytopathic antiviral mechanisms contribute to viral clearance during acute viral hepatitis by purging HBV replicative intermediates from the cytoplasm and covalently closed circular viral DNA from the nucleus of infected cells.

Goyal A, Murray J M.

Modelling the impact of cell-to-cell transmission in hepatitis B virus

Plos ONE, 2016, 11( 8): e0161978

[本文引用: 1]

杨翠兰, 刘贤宁.

一个具有细胞-细胞传播和时滞的病毒动力学模型

西南大学学报(自然科学版), 2015, 37(5): 97-101

[本文引用: 1]

Yang C L, Liu X N.

A virus dynamical model with cell-to-cell viral transmission and delay

Journal of Southwest University of China (Natural Science Edition), 2015, 37(5): 97-101

[本文引用: 1]

Sun H Q, Wang J L.

Dynamics of a diffusive virus model with general incidence function, cell-to-cell transmission and time delay

Comput Math Appl, 2019, 77(1): 284-301

[本文引用: 1]

Manna K, Chakrabarty S. P.

Chronic Hepatitis B infection and HBV DNA-containing capsids: modeling and analysis

Commun Nonlinear Sci Numer Simulat, 2015, 22(1-3): 383-395

DOI:10.1016/j.cnsns.2014.08.036      URL     [本文引用: 1]

Liu S H, Zhang R.

On an age-structured Hepatitis B virus infection model with HBV DNA-containing capsids

Bull Malays Math Sci Soc, 2021, 44(3): 1345-1370

DOI:10.1007/s40840-020-01014-6      [本文引用: 1]

Manna K.

Dynamics of a diffusion-driven HBV infection model with capsids and time delay

Int J Biomath, 2017, 10( 5): 1750062

[本文引用: 1]

Guo T, Liu H H, Xu C L, et al.

Global stability of a diffusive and delayed HBV infection model with HBV DNA-containing capsids and general incidence rate

Discret Contin Dyn Syst Ser B, 2018, 23(10): 4223-4242

[本文引用: 1]

Manna K, Hattaf H.

Spatiotemporal dynamics of a generalized HBV infection model with capsids and adaptive immunity

Int J Appl Comput Math, 2019, 5: 65

DOI:10.1007/s40819-019-0651-x      [本文引用: 1]

Webb G F.

A reaction-diffusion model for a deterministic diffusive epidemic

J Math Anal Appl, 1981, 84(1): 150-161

DOI:10.1016/0022-247X(81)90156-6      URL     [本文引用: 1]

Webb G F. Theory of Nonlinear Age-Dependent Population Dynamics. New York: Marcell Dekker Inc, 1985

[本文引用: 1]

Protter M H, Weinberger H F. Maximum Principles in Differential Equations. New York: Springer Science & Business Media, 2012

[本文引用: 5]

Wang J L, Wang J.

Analysis of a reaction-diffusion cholera model with distinct dispersal rates in the human population

J Dyn Differ Equ, 2021, 33: 549-575

DOI:10.1007/s10884-019-09820-8      [本文引用: 1]

Hale J K.

Asymptotic Behavior of Dissipative Systems

Providence: Amer Math Soc, 1988

[本文引用: 1]

Hale J K, Lunel S M V.

Introduction to Functional Differential Equations

Vol 99. New York:Springer Science & Business Media, 2013

[本文引用: 2]

Smith H L, Zhao X Q.

Robust persistence for semidynamical systems

Nonlinear Anal: TMA, 2001, 47(9): 6169-6179

DOI:10.1016/S0362-546X(01)00678-2      URL    

/