数学物理学报, 2023, 43(3): 970-984

一类具有 Dirichlet 边界条件的年龄-空间结构HIV/AIDS传染病模型的动力学分析

吴鹏,1,2,*, 王秀男3, 何泽荣4

1杭州电子科技大学理学院 杭州 310018

2浙江财经大学数据科学学院 杭州 310018

3美国田纳西大学查塔努加分校数学系 查塔努加田纳西州 37403

4杭州电子科技大学运筹与控制研究所 杭州 310018

Dynamical Analysis of an Age-Space Structured HIV/AIDS Model with Homogeneous Dirichlet Boundary Condition

Wu Peng,1,2,*, Wang Xiunan3, He Zerong4

1School of Sciences, Hangzhou Dianzi University, Hangzhou 310018

2School of Data Sciences, Zhejiang University of Finance & Economics, Hangzhou 310018

3Department of Mathematics, University of Tennessee at Chattanooga, Chattanooga, TN 37403, USA

4Institute of Operational Research and Cybernetics, Hangzhou Dianzi University, Hangzhou 310018

通讯作者: *吴鹏,E-mail: hzpengwu@163.com

收稿日期: 2022-08-3   修回日期: 2023-02-12  

基金资助: 国家自然科学基金(12201557)
国家自然科学基金(11871185)
浙江省教育厅一般项目(Y202249921)

Received: 2022-08-3   Revised: 2023-02-12  

Fund supported: NSFC(12201557)
NSFC(11871185)
Foundation of Zhejiang Provincial Education Department(Y202249921)

摘要

为了探讨个体扩散、感染年龄和 Dirichlet 边界环境对 HIV/AIDS 时空传播动力学的影响, 该文构建了一类具有齐次 Dirichlet 边界条件的年龄空间结构 HIV/AIDS 传染病动力学模型. 首先, 应用特征线方法, 作者将模型转化为一个积分反应扩散方程模型. 其次, 作者给出模型基本再生数 $R_0$ 的泛函表达式, 并研究了以$R_0$ 为阈值的模型解的动力学行为. 具体地, 当$R_0<1$时, HIV/AIDS 在人群中可以被消除; 而当 $R_0>1$时, HIV 感染在人群中会持续存在. 最后, 在二维空间区域中作者通过数值模拟验证了文中理论结果.

关键词: HIV/AIDS 模型; Dirichlet 边界条件; 年龄-空间结构; 基本再生数; 阈值动力学; 一致持久性

Abstract

In order to explore the impact of human movement, infection age, and a hostile boundary environment on the HIV/AIDS spatiotemporal transmission dynamics, we construct an age-space structure model with homogeneous Dirichlet boundary condition. Applying the method of characteristics, we transform the model into a system of a reaction-diffusion equation and an integral equation. We derive the basic reproduction ratio $R_0$ and investigate the threshold dynamics in terms of $R_0$. Out theoretical results show that, under appropriate conditions, the disease can be eliminated when $R_0<1$ and the infection is uniformly persistent among the population when $R_0>1$. We verify the theoretical result by numerical simulations in a two-dimensional spatial domain.

Keywords: HIV/AIDS model; Dirichlet boundary condition; Age-space structured; Basic reproduction ratio; Threshold dynamics; Uniform persistence

PDF (705KB) 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

吴鹏, 王秀男, 何泽荣. 一类具有 Dirichlet 边界条件的年龄-空间结构HIV/AIDS传染病模型的动力学分析[J]. 数学物理学报, 2023, 43(3): 970-984

Wu Peng, Wang Xiunan, He Zerong. Dynamical Analysis of an Age-Space Structured HIV/AIDS Model with Homogeneous Dirichlet Boundary Condition[J]. Acta Mathematica Scientia, 2023, 43(3): 970-984

1 引言与问题

尽管目前 HIV/AIDS预防和治疗取得了巨大进展, 但仍有一些低收入和中等收入国家受到 HIV 流行的影响. 据估计, 截止到2020 年底全球有 3760 万 HIV 感染者, 大约 680,000 人死于与AIDS相关的疾病, 其中大多数在东部和南部非洲. 在影响 HIV/AIDS 传播和分布的所有因素中, 人群的扩散被认为是一个重要因素. 这促使关于HIV/AIDS 空间传播动力学的研究工作广泛展开. 例如, Cuadros 等[2] 利用一种基于空间变量的模型在几个非洲国家创建高分辨率的艾滋病毒感染率估计图. Ayalew 等[3] 比较了三种空间 HIV 模型, 以估计南非不同地区的艾滋病毒感染率. Wu 和 Zhao[4] 研究了一类带有抗逆转录病毒治疗的年龄结构空间艾滋病毒流行病模型的动力学行为. Zhao 等[5] 建立了一类具有三年龄段的 HIV/AIDS 流行病模型, 并进行了动力学和最优控制分析. 其他一些时空结构 HIV 感染模型也已被用来研究 HIV 在宿主体内的传播动力学(参见文献[6-10]).

事实上, 随着时间的推移, HIV 感染者可能会发展成艾滋病或保持在无症状阶段, 这在很大程度上取决于感染者的感染年龄, 即个体感染 HIV 后的时间. 因此有必要将感染年龄纳入空间 HIV/AIDS 模型, 以揭示该疾病传播的动力学机制, 这就是所谓的“年龄-空间结构”. 除了文献[4-5]所做的研究工作之外, 年龄-空间结构也被一些学者在建立其他传染病模型时所考虑. 最近, Chekroun和Kuniya[11]研究了具有 Neumann 边界条件的年龄空间结构SIR 流行病模型的全局动力学. Yang等[12]研究了年龄结构空间布鲁氏菌病模型的阈值动力学行为. Liu 等[13]研究了具有年龄结构和空间扩散的多群体SEIR流行病模型的全局演化行为. Wang等[14]建立了一类具有年龄-空间结构的 HIV 感染模型并研究了模型解的全局动力学行为. 值得提出的是, 上述这些模型都是考虑了齐次 Neumann 边界条件. 然而, 当空间区域的边界具有对个体生存不利的环境 (例如, 沼泽、沙漠或高山) 时, 模型更适合考虑齐次狄利克雷边界条件 (参见文献[15-16]). 有鉴于此, 具有狄利克雷边界条件的空间模型已被用于研究相关传染病的动力学传播. 例如, Chekroun 和 Kuniya[17]分析了 Dirichlet 边界条件下具有年龄-空间结构 SIR 反应扩散传染病模型的阈值动力学. Wang等[18]在齐次 Dirichlet 边界条件下分析了年龄-空间结构口蹄疫模型的时空动力学.

受到上述研究工作的启发, 为了研究人群扩散、感染年龄和 Dirichlet 边界条件环境对人群中 HIV/AIDS 传播的综合影响, 本文建立如下模型

$\begin{equation}\label{1} \left\{ \begin{array}{ll} \frac{\partial U_1(x,t)}{\partial t}=d_1\Delta U_1(x,t)+\Pi-U_1(x,t)\sum_{i=2}^3\int_0^{\infty}\gamma_i(b)u_i(x,b,t){\rm d}b-\sigma U_1(x,t),\\ \hskip 6cm x\in\Gamma,\ t>0,\\ \left(\frac{\partial}{\partial t}+\frac{\partial}{\partial b}\right)u_i(x,b,t)=d_i\Delta u_i(x,b,t)-(\sigma+\eta_i(b)+\delta_i(b))u_i(x,b,t),\\ \hskip 6cm x\in\Gamma,\ t>0,\ b>0,\ i=2,3,\\ \frac{\partial U_4(x,t)}{\partial t}=d_4\Delta U_4(x,t)+\int_0^{\infty}\eta_3(b)u_3(x,b,t){\rm d}b-(\sigma+\delta_4)U_4(x,t),\\ \hskip 6cm x\in\Gamma,\ t>0,\\ u_2(x,0,t)=U_1(x,t)\sum_{i=2}^3\int_0^{\infty}\gamma_i(b)u_i(x,b,t){\rm d}b,\\ \hskip 6cm x\in\Gamma,\ t>0,\\ u_3(x,0,t)=\int_0^{\infty}\eta_2(b)u_2(x,b,t){\rm d}b, x\in\Gamma,\ t>0. \end{array} \right. \end{equation}$

模型的初值条件和齐次 Dirichlet 边界条件如下所示

$\begin{equation}\label{2} \begin{array}{ll} &U_1(x,0)=\psi_1(x),\ u_i(x,b,0)=\psi_i(x,b),\ U_4(x,0)=\psi_4(x),\ x\in\overline{\Gamma}, b\ge 0,\ i=2,3,\\ &U_1(x,t)=u_2(x,b,t)=u_3(x,b,t)=U_4(x,t)=0,\ x\in\partial \Gamma,\ t>0,\ b\ge 0, \end{array} \end{equation}$

其中, $\overline{\Gamma}$$\partial \Gamma$ 分别代表有界连通开集合 $\Gamma\subset {\Bbb R} ^n$ 的闭包和光滑边界, 其中 $n\in{\mathbb N}_+$. 变量 $U_1(x,t),u_2(x,b,t),u_3(x,b,t),U_4(x,t)$ 分别表示在时间 $t$ 和位置$x$ 处感染年龄为 $b$ 的易感个体密度, 初期 HIV 感染个体密度, 慢性无症状感染个体密度和患有 AIDS 病症且性活动减少的个体密度(参见文献[19]). 参数 $d_j(j=1,2,3,4)$ 分别代表这四类人群的个体扩散率. 参数 $\Pi$ 表示易感个体的输入率, 易感个体的自然死亡率用 $\sigma$ 表示. 变量 $u_2$$u_3$ 因疾病导致的依赖年龄的死亡率分别为 $\delta_2(b)$$\delta_3(b)$. 变量 $U_4$ 的疾病导致的死亡率为 $\delta_4$. 参数 $\eta_i(b)$$\gamma_i(b)$ 分别代表变量 $u_i$, $i=2,3$ 中感染个体的转移率和感染率.

利用特征线法, 将系统 (1.1)-(1.2) 改写为积分反应扩散方程. 为此, 令 $b-t=a$, $u_i(x,b,t)=u_i(x,t+a,t)$, $\eta_i(t)=\eta_i(t+a),\delta_i(t)=\delta_i(t+a)$, $i=2,3$. 类似文献[11-12,20-21] 中的推导过程, 我们可以得到 $u_i(x,b,t)$ 表达式如下

$\begin{equation}\label{3} u_i(x,b,t)=\left\{ \begin{array}{ll} \Omega_i(b)\int_{\Gamma}G_i(x,y,b)u_i(y,0,t-b){\rm d}y,\ & t>b,\\ \frac{\Omega_i(b)}{\Omega_i(b-t)}\int_{\Gamma}G_i(x,y,t)\psi_i(y,b-t){\rm d}y,\ &b\ge t, \end{array} \right. \end{equation} $

其中 $\Omega_i(b)=\exp\{-\int_0^b(\sigma+\eta_i(s)+\delta_i(s))ds\}$, $G_i(x,y,t)$ 是关于拉普拉斯算子 $d_i\Delta u_i$ 的格林函数, $i=2,3$, $G_i(x,y,t)$ 函数的性质参加文献[22-23].

$U_i(x,t)=u_i(x,0,t)$ 并将 (1.3) 带入到系统 (1.1) 第四个方程中可得

$\begin{eqnarray*}\label{4} U_2(x,t) &=&U_1(x,t)\int_0^t\gamma_2(b)\Omega_2(b)\int_{\Gamma}G_2(x,y,b)U_2(y,t-b){\rm d}y{\rm d}b\\ &&+U_1(x,t) \int_0^t\gamma_3(b)\Omega_3(b)\int_{\Gamma}G_3(x,y,b) \int_0^{\infty}\eta_2(c)u_2(y,c,t-b){\rm d}c{\rm d}y{\rm d}b\\ &&+U_1(x,t)\sum_{i=2}^3\int_t^{\infty}\gamma_i(b) \frac{\Omega_i(b)}{\Omega_i(b-t)}\int_{\Gamma}G_i(x,y,t)\psi_i(y,b-t){\rm d}y{\rm d}b. \end{eqnarray*}$

由于 $U_3(x,t)$$U_4(x,t)$$U_2(x,t)$ 所确定, 因此我们接下来研究如下积分反应扩散方程模型, 其中 $x\in\Gamma, t>0$,有

$\begin{equation}\label{5} \left\{ \begin{array}{ll} \frac{\partial U_1(x,t)}{\partial t}=d_1\Delta U_1(x,t)+\Pi-\sigma U_1(x,t)-U_2(x,t),\ x\in\Gamma, t>0,\\[3mm] U_2(x,t)=U_1(x,t)\int_0^t\gamma_2(b)\Omega_2(b)\int_{\Gamma}G_2(x,y,b)U_2(y,t-b){\rm d}y{\rm d}b\\[3mm] +U_1(x,t) \int_0^t\gamma_3(b)\Omega_3(b)\int_{\Gamma}G_3(x,y,b) \int_0^{t-b}\eta_2(c)\Omega_2(c)\int_{\Gamma}G_2(y,z,c)\\[3mm] \times U_2(z,t-b-c){\rm d}z{\rm d}c{\rm d}y{\rm d}b+ ({\cal F}_1(x,t)+{\cal F}_2(x,t)),\\[3mm] U_1(x,0)=\psi_1(x),U_2(x,0)=\psi_1(x)\sum_{i=2}^3\int_0^{\infty}\gamma_i(b)\psi_i(x,b){\rm d}b,\ x\in\overline{\Gamma}, \\[3mm] U_1(x,t)=U_2(x,t)=0,\ x\in\partial \Gamma, t>0. \end{array} \right. \end{equation}$

其中

$\begin{eqnarray*} {\cal F}_1(x,t)&=&U_1(x,t)\sum_{i=2}^3\int_t^{\infty}\gamma_i(b)\frac{\Omega_i(b)}{\Omega_i(b-t)}\int_{\Gamma}G_i(x,y,t)\psi_i(y,b-t){\rm d}y{\rm d}b,\\ {\cal F}_2(x,t)&=&U_1(x,t)\int_0^{t}\gamma_3(b)\Omega_3(b)\int_{\Gamma}G_3(x,y,b)\int_{t-b}^{\infty}\eta_2(c)\frac{\Omega_2(c)}{\Omega_2(b+c-t)}\\ &&\times\int_{\Gamma}G_2(y,z,t-b)\psi_2(z,b+c-t){\rm d}z{\rm d}c{\rm d}y{\rm d}b. \end{eqnarray*}$

接下来, 我们致力于研究系统 (1.5) 的动力学行为.

2 系统(1.5)的适定性

为了研究系统 (1.5) 的适定性,我们首先做出如下假设.

${\bf定义 2.1}$ 我们假设

(1) 模型参数 $\Pi,\sigma,\delta_4$, $d_j\ (j=1,2,3,4)$ 都是正的;

(2) $\gamma_i(\cdot)\in L^1_+({\Bbb R} _+)\cap L^{\infty}_+({\Bbb R} _+)$. 此外, 存在 $0<b_1<b_2$ 使得 $\gamma_i(b)>0, b\in (b_1,b_2)$;

(3) $\eta_i(\cdot),\delta_i(\cdot)\in L^{\infty}_+({\Bbb R} _+)$, $m^{\infty}=ess\cdot\sup\limits_{b\in{\Bbb R} _+}m(b)<\infty$, 其中 $m(\cdot)=\gamma_i(\cdot),\eta_i(\cdot),\delta_i(\cdot),$$ i=2,3$.

我们令空间 ${\mathbb M}\triangleq C({\Bbb R},\overline{\Gamma})$ 具有范数 $\| \psi\| _{{\mathbb M}}\triangleq\sup\limits_{x\in\Gamma}|\psi(x)|,\psi\in{\mathbb M}$. 定义空间 ${\mathbb Q}\triangleq BC({\Bbb R} _+,{\mathbb M})$, 其范数为 $\| \phi\| _{{\mathbb Q}}\triangleq\sup\limits_{b\ge 0}\| \phi(b)\| _{{\mathbb M}}=\sup\limits_{(x,b)\in\Gamma\times{\Bbb R} _+}|\phi(x,b)|$, $\phi\in{\mathbb Q}$. 定义 ${\mathbb M}^0\triangleq\{\psi\in{\mathbb M}|\psi(x)=0,$$x\in\partial\Gamma\}$, ${\mathbb Q}^0\triangleq\{\phi\in{\mathbb Q}|\phi(x,\cdot)=0,$$x\in\partial \Gamma\}$ 分别是空间 ${\mathbb M}$${\mathbb Q}$ 的正锥. 在空间 ${\mathbb Q}^0$ 中定义线性算子 $\Phi$

$\begin{eqnarray*} (\Phi\phi)(x,t)&\triangleq&\int_0^{t}\gamma_2(b)\Omega_2(b)\int_{\Gamma}G_2(x,y,b)\phi(y,t-b){\rm d}y{\rm d}b\\ &&+\int_0^t\gamma_3(b)\Omega_3(b)\int_{\Gamma}G_3(x,y,b) \int_0^{t-b}\eta_2(c)\Omega_2(c)\\ &&\times \int_{\Gamma}G_2(y,z,c)\phi(z,t-b-c){\rm d}z{\rm d}c{\rm d}y{\rm d}b, \phi\in{\mathbb Q}^0,x\in\overline{\Gamma}. \end{eqnarray*}$

根据格林函数 $G_i(x,y,t)$ 性质以及假设2.1, 易知 $\Phi({\mathbb Q}^0)\subset {\mathbb Q}^0_+$, 其中 ${\mathbb Q}^0_+$ 是空间 ${\mathbb Q}^0$ 的正锥. 此外, 由 $\sigma+(\Phi U_2)(x,t)+{\cal F}_1+{\cal F}_2$ 的有解性和连续性, 再根据强极大值原理[24] 和假设 2.1, 我们可得如下结论.

${\bf引理 2.1}$ 如果 $(U_1,U_2)$ 是系统 (1.5) 具有 $(\psi_1,\psi_2)\in{\mathbb M}^0\times{\mathbb Q}^0$ 初值条件的解, 那么存在 $t_0>0$ 使得 $U_1(x,t)>0,U_2(x,t)\ge 0$, $(x,t)\in\Gamma\times(0,t_0]$.

进一步地, 我们可得系统(1.5)解的全局存在性和一致有界性的结论.

${\bf引理 2.2}$ 如果假设 2.1 成立且 $(\psi_1,\psi_2,\psi_3)\in{\mathbb M}^0\times{\mathbb Q}^0\times{\mathbb Q}^0$, 那么系统 (1.5)有正解 $(U_1,U_2)$, $(x,t)\in\Gamma\times[0,\infty)$. 此外, $U_1$$U_2$ 都是一致有界的, 即存在两个正数 ${\cal B}_1={\cal B}_1(\psi_1,\psi_2,\psi_3), $${\cal B}_2={\cal B}_2(\psi_1,\psi_2,\psi_3)$, $(\psi_1,\psi_2,\psi_3)\in{\mathbb M}^0_+\times{\mathbb Q}^0_+\times{\mathbb Q}^0_+$, $(x,t)\in\Gamma\times(0,+\infty)$, 有 $0<U_1(x,t)\le{\cal B}_1,0<U_2(x,t)\le{\cal B}_2$ 成立.

${\bf证}$ 从系统 (1.5) 第一个方程中可得

$U_1(x,t)=\int_0^te^{-\sigma b}\int_{\Gamma}G_1(x,y,b)(\Pi-U_2(y,t-b)){\rm d}y{\rm d}b+{\cal F}_3(x,t), (x,t)\in\overline{\Gamma}\times(0,+\infty), $

其中 ${\cal F}_3(x,t)=e^{-\sigma t}\int_{\Gamma}G_1(x,y,t)\psi_1(y){\rm d}y$. 于是, $U_2(x,t)$ 可以被表示为

$\begin{eqnarray*} \label{7} U_2(x,t)&=&\left(\int_0^te^{-\sigma b}\int_{\Gamma}G_1(x,y,b)(\Pi-U_2(y,t-b)){\rm d}y{\rm d}b+{\cal F}_3(x,t)\right)\\ &&\times((\Phi U_2)(x,t)+{\cal F}_1(x,t)+{\cal F}_2(x,t)). \end{eqnarray*} $

由此, 我们可知当(2.1)式存在解 $U_2(x,t)$, 那么 $U_1(x,t)$ 就可以被确定.

${\cal F}^{\infty}_k:=\sup\limits_{(x,t)\in\Omega\times{\Bbb R} _+}{\cal F}_k(x,t),k=1,2,3$. 由于 $(\psi_1,\psi_2,\psi_3)\in{\mathbb M}^0_+\times{\mathbb Q}^0_+\times{\mathbb Q}^0_+$ 以及格林函数 $G_i(x,y,t)\ (i=2,3)$ 的性质可知 ${\cal F}^{\infty}_k$ 是有界的. 设 ${\cal B}>{\cal F}^{\infty}_3({\cal F}^{\infty}_1+{\cal F}^{\infty}_2)$ 是足够大的数以及

$\begin{eqnarray*} &&f_1(x,t)=\frac{\Pi+{\cal B}}{\sigma}(1-e^{-\sigma t})+{\cal F}_3(x,t), \\ &&f_2(x,t)=\frac{\gamma^{\infty}_2{\cal B}}{\sigma}(1-e^{-\sigma t})+\frac{\gamma^{\infty}_3\eta^{\infty}_2{\cal B}}{\sigma^2}(1-e^{-\sigma t})^2+{\cal F}_1(x,t)+{\cal F}_2(x,t). \end{eqnarray*} $

注意到 $f_1(x,0)={\cal F}_3(x,0),f_2(x,0)={\cal F}_1(x,0)+{\cal F}_2(x,0)$, 我们可选取足够小的 $T>0$ 使得

$ \sup\limits_{(x,t)\in\Gamma\times(0,T)}[f_1f_2]<{\cal B}, $
$ \tilde{f}:=\sup\limits_{(x,t)\in\Gamma\times(0,T)}\frac{1}{\sigma}\bigg[(\gamma^{\infty}_2 +\frac{\gamma^{\infty}_3\eta^{\infty}_2}{\sigma}(1-e^{-\sigma t})f_1(x,t)+f_2(x,t)\bigg][1-e^{-\sigma t}]<1. $

对任意固定的 $T$, 我们令 ${\mathbb M}_T:=C({\mathbb M},[T])$ 具有范数

$\| \psi\| _{{\mathbb M}_T}=\sup\limits_{t\in(0,T)}\| \psi(t)\| _{{\mathbb M}}=\sup\limits_{(x,t)\in\Gamma\times(0,T)}|\psi(x,t)|, \psi\in{\mathbb M}_T. $

定义

$ {\mathbb M}^0_T:=\{\psi\in{\mathbb M}_T:\psi(x,\cdot)=0,x\in\partial\Gamma\}, {\mathbb M}^0_{T,{\cal B}}:=\{\psi\in{\mathbb M}^0_T:\| \psi\| _{{\mathbb M}_T}<{\cal B}\}, $

以及 ${\mathbb M}^0_{T,{\cal B}}$ 中的线性算子 ${\cal H}$, $(x,t)\in\bar{\Gamma}\times(0,T]$,有

$\begin{eqnarray*} ({\cal H}\phi)(x,t)&=&\left[\int_0^te^{-\sigma b}\int_{\Gamma}G_1(x,y,b)[\Pi-\psi(y,t-b)]{\rm d}y{\rm d}b+{\cal F}_3(x,t)\right]\\ &&\times\left[\int_0^t\gamma_2(b)\Omega_2(b)\int_{\Gamma}G_2(x,y,b)\psi(y,t-b){\rm d}y{\rm d}b+\int_0^t\gamma_3(b)\Omega_3(b)\int_{\Gamma}G_3(x,y,b)\right.\\ &&\times \left.\int_0^{t-b}\eta_2(c)\Omega_2(c)\int_{\Gamma}G_2(y,z,c)\psi(z,t-b-c){\rm d}z{\rm d}c{\rm d}y{\rm d}b+{\cal F}_1(x,t)+{\cal F}_2(x,t)\right] \end{eqnarray*}$

$\psi\in{\mathbb M}^0_{T,{\cal B}}$, 我们可得 $({\cal H}\psi)(x,\cdot)=0,\ x\in\partial \Gamma$,有

$\begin{eqnarray*} \| {\cal H}\psi\| _{{\mathbb M}_T}&\le& \sup\limits_{(x,t)\in\Gamma\times(0,T)}\left(\frac{\Pi+{\cal B}}{\sigma}(1-e^{-\sigma t})+{\cal H}_3(x,t)\right)\\ &&\times\left(\frac{\gamma^{\infty}_2{\cal B}}{\sigma}(1-e^{-\sigma t})+\frac{\gamma^{\infty}_3\eta^{\infty}_2{\cal B}}{\sigma^2}(1-e^{-\sigma t})^2+{\cal F}_1(x,t)+{\cal F}_2(x,t)\right)\\ & \le&\sup\limits_{(x,t)\in\Gamma\times(0,T)}[f_1(x,t)f_2(x,t)]<{\cal B}, \end{eqnarray*}$

这意味着 ${\cal H}\psi\in{\mathbb M}^0_{T,{\cal B}}$. 由于 ${\cal H}({\mathbb M}^0_{T,{\cal B}})\subset{\mathbb M}^0_{T,{\cal B}}$ 我们可知 ${\cal H}$${\mathbb M}^0_{T,{\cal B}}$ 中是严格收缩的. 为了延拓解的存在区域, 我们还需证明解 $(U_1,U_2)$ 永远不会爆破. 从系统 (1.5) 第一个方程中, 可知 $U_1(x,t)\le \Pi/\sigma(1-e^{-\sigma t})+e^{-\sigma t}\| \psi_1\| _{{\mathbb M}}<+\infty$. 于是, 存在 ${\cal B}_1:=\Pi/\sigma+\| \psi_1\| _{{\mathbb M}}$ 使得 $0<U_1(x,t)\le {\cal B}_1$.

假设 $U_2(x,t)$ 会爆破, 则存在 $\tilde{t}>0, \tilde{x}\in\Gamma$ 使得 $\lim\limits_{t\to\tilde{t}^{-}}\partial_tU_1(\tilde{x},t)=-\infty$, 这意味着 $U_1(\tilde{x},t)$$\tilde{t}$ 邻域中会变为负的. 这显然与 $U_1(x,t)$ 的正性相矛盾. 因此, $U_2(x,t)$ 不会爆破. 于是, 存在 ${\cal B}_2$ 使得 $0<U_2(x,t)\le{\cal B}_2$. 引理证毕.

${\bf定理 2.1}$ 如果假设条件 2.1 成立, 那么系统 (1.5) 存在唯一的具有初始条件 $(\psi_1,\psi_2,\psi_3)\in{\mathbb M}^0_+\times{\mathbb Q}^0_+\times{\mathbb Q}^0_+$ 的有界解 $(U_1,U_2)$, $(x,t)\in\Gamma\times[0,+\infty)$.

${\bf证}$ 我们已经在引理 2.2 中证明了系统 (1.5) 经典解的全局存在性. 现在我们利用反证法证明解的唯一性. 假设系统 (1.5) 有两个解 $(\widetilde{U}_1,\widetilde{U}_2)$$(\widehat{U}_1,\widehat{U}_2)$. 从引理 2.2 中我们可知存在 ${\cal B}^+>0$ 使得 $0\le\widetilde{U}_2,\widehat{U}_2\le {\cal B}_+$, $(x,t)\in\Gamma\times(0,+\infty)$. 注意到 $\widetilde{U}_2={\cal H}\widetilde{U}_2, \widehat{U}_2={\cal H}\widehat{U}_2$.$U^*_2=\widetilde{U}_2-\widehat{U}_2$, 可得

$\begin{eqnarray*} |\widetilde{U}_2-\widehat{U}_2|&=&|{\cal H}U^*_2(x,t)| \\ &\le&\left(\frac{\Pi+{\cal B}^+}{\sigma}+{\cal H}^{\infty}_3\right)\int_0^t\| U^*_2(b)\| _{{\mathbb M}}{\rm d}b \\ &&+\left(\frac{\gamma^{\infty}_2{\cal B}^+}{\sigma}+\frac{\gamma^{\infty}_3\eta^{\infty}_2{\cal B}^+}{\sigma^2} +{\cal H}^{\infty}_1+{\cal H}^{\infty}_2\right)\int_0^t\| U^*_2(b)\| _{{\mathbb M}}{\rm d}b\\ &:=&{\bf A}^+\int_0^t\| U^*_2(b)\| _{{\mathbb M}}{\rm d}b. \end{eqnarray*}$

于是利用 Gronwall 不等式, 可得 $\| U^*_2(b)\| _{{\mathbb M}}=0,t>0$. 这意味着 $\widetilde{U}_2=\widehat{U}_2$. 类似地, 可得 $U_1(x,t)$ 的唯一性. 定理证毕.

3 模型的基本再生数

此节中, 我们致力于推导出系统 (1.5) 的基本再生数泛函表达式. 根据命题 4.1 和引理 4.1[17], 可知方程

$\begin{eqnarray*} \left\{ \begin{array}{ll} 0=d_1\Delta U_{10}(x)+\Pi-\sigma U_{10}(x),\ &x\in\Gamma,\\ U_{10}(x)=0,\ &x\in\partial \Gamma, \end{array} \right. \end{eqnarray*}$

有唯一的非负解 $U_{10}\in{\mathbb M}^0_+\setminus\{0\}$, 其在 $\Gamma$ 中是严格正的且满足 $\limsup\limits_{t\to+\infty}U_1(x,t)\le U_{10}(x)$, $x\in\overline{\Gamma}$. 系统 (1.5) 中 $U_2(x,t)$ 在无病平衡态 $E_0(U_{10}(x),0)$ 处的线性化方程

$\begin{eqnarray*} U_2(x,t)&=&U_{10}(x)\bigg[\int_0^t\gamma_2(b)\Omega_2(b)\int_{\Gamma}G_2(x,y,b)U_2(y,t-b){\rm d}y{\rm d}b +\int_0^t\gamma_3(b)\Omega_3(b)\\ &&\times \int_{\Gamma}G_3(x,y,b)\int_0^{t-b}\eta_2(c)\Omega_2(c) \int_{\Gamma}G_2(y,z,c)U_2(z,t-b-c){\rm d}z{\rm d}c{\rm d}y{\rm d}b\bigg]. \end{eqnarray*}$

根据文献[25], 我们定义再生算子 ${\cal R}:{\mathbb M}^0\to{\mathbb M}^0$

$\begin{eqnarray*}\label{8} {\cal R}(\psi)(x)&=&U_{10}(x)\bigg[\int_0^{\infty}\gamma_2(b)\Omega_2(b)\int_{\Gamma}G_2(x,y,b)\psi(y){\rm d}y{\rm d}b \\ &&+\int_0^{\infty}\gamma_3(b)\Omega_3(b)\int_{\Gamma}G_3(x,y,b)\int_0^{\infty}\eta_2(c)\Omega_2(c) \int_{\Gamma}G_2(y,z,c)\psi(z){\rm d}z{\rm d}c{\rm d}y{\rm d}b\bigg], \end{eqnarray*}$

其中 $\psi\in{\mathbb M}^0, x\in\overline{\Gamma}$. 由文献[11-12]中的结论可知系统 (1.5) 的基本再生数 $R_0$ 定义为 $R_0=r({\cal R})$, 其中 $r({\cal R})$ 是算子 ${\cal R}$ 的谱半径. 此外, 根据文献[17,引理 4.2,命题 4.2], 我们有如下结论.

${\bf定理 3.1}$ 若算子 ${\cal R}$ 如(3.1)式所定义, 则 ${\cal R}$ 是强正的, 有界的和紧的. 此外, $R_0$${\cal R}$ 唯一简单的特征值, 其具有正特征函数 $\zeta_0\in{\mathbb M}^0_+\setminus\{0\}$.

4 模型无病平衡态的全局吸引性

此节中, 我们讨论系统 (1.5) 无病平衡态 $E_0(U_{10}(x),0)$ 的全局吸引性. 为此, 我们首先给出如下引理.

${\bf引理 4.1}$ 如果 $(\psi_1,\psi_2,\psi_3)\in{\cal D}$, $R_0<1$, 那么有

$0\le U_1(x,t)\le U_{10}(x), 0\le U_2(x,t)\le \epsilon\zeta_0 {\rm for}\ (x,t)\in\overline{\Gamma}\times(0,+\infty), $

其中 $\epsilon>0$ 是足够大的常数, ${\cal D}$如下所示

$\begin{eqnarray*} {\cal D}:&=&\bigg\{(\psi_1,\psi_2,\psi_3)\in{\mathbb M}^0_+\times{\mathbb Q}^0_+\times{\mathbb Q}^0_+ :\psi_1\le U_{10}(x),\psi_2(x,b)\\ && \le \epsilon\bigg[\Omega_2(b)\int_{\Gamma}G_2(x,y,b)\zeta_0(y){\rm d}y\\ && +\Omega_3(b)\int_{\Gamma}G_3(x,y,b)\int_0^{\infty}\Omega_2(c) \int_{\Gamma}G_2(y,z,c)\zeta_0(z){\rm d}z{\rm d}c{\rm d}y{\rm d}b\bigg]\bigg\}\ \mbox{for}\ b\ge 0,x\in\overline{\Gamma}. \end{eqnarray*}$

${\bf证}$ 根据引理 2.2 易知 $0\le U_1(x,t)\le U_{10}(x)$.$U_2(x,0)$ 的方程中我们可得

$\begin{eqnarray*} U_2(x,0)&=&\psi_1(x)\bigg[\int_0^{\infty}\gamma_2(b)\Omega_2(b)\epsilon\int_{\Gamma}G_2(x,y,b)\zeta_0(y){\rm d}y{\rm d}b\\ &&+\int_0^{\infty}\gamma_3(b)\epsilon\Omega_3(b)\int_{\Gamma}G_3(x,y,b)\int_0^{\infty}\eta_2(c)\Omega_2(c)\int_{\Omega}G_2(y,z,c)\zeta_0(z){\rm d}z{\rm d}c{\rm d}y{\rm d}b\bigg]\\ &\le &\epsilon({\cal R}\zeta_0)(x)\le R_0\epsilon\zeta_0(x)<\epsilon \zeta_0(x). \end{eqnarray*}$

假设存在 $t_2>0$$x_2\in\Gamma$ 使得 $U_2(x,t)<\epsilon\zeta_0(x)$, $t\in(0,t_2)$ 以及 $U_2(x,t_2+\rho)>\epsilon\zeta_0(x)$$\rho>0$ 足够小. 从而可得

$\begin{eqnarray*} U_2(x,t_2+\rho)&\le&U_{10}(x)\int_0^{t_2+\rho}\gamma_2(b)\Omega_2(b)\int_{\Gamma}G_2(x,y,b)\epsilon\zeta_0(xy){\rm d}y{\rm d}b\\ &&+U_{10}(x)\int_{t+\rho}^{\infty}\gamma_2(b)\Omega_2(b)\int_{\Gamma}G_2(x,y,t_2+\rho)\int_{\Gamma}G_2(y,z,b-t_2-\rho)\epsilon\zeta_0(z){\rm d}z{\rm d}y{\rm d}b\\ &&+U_{10}(x)\int_0^{t_2+\rho}\gamma_3(b)\Omega_3(b)\int_{\Gamma}G_3(x,y,b)\epsilon\int_0^{t_2+\rho-b}\gamma_2(c)\Omega_2(c)\\ &&\times \int_{\Gamma}G_2(y,z,c){\rm d}z{\rm d}c{\rm d}y{\rm d}b\\ &\le& U_{10}(x)\bigg[\int_0^{\infty}\gamma_2(b)\Omega_2(b)\int_{\Gamma}G_2(x,y,b)\epsilon\zeta_0{\rm d}y{\rm d}b\\ &&+\int_0^{\infty}\gamma_3(b)\Omega_3(b)\int_{\Gamma}G_3(x,y,b)\int_0^{\infty}\eta_2(c)\Omega_2(c)\int_{\Gamma}G_2(y,z,c)\epsilon\zeta_0dzdc{\rm d}y{\rm d}b\bigg]\\ &\le&\epsilon R_0\zeta_0(x)\le \epsilon \zeta_0(x). \end{eqnarray*}$

引理证毕.

${\bf定理 4.1}$ 假设条件 2.1 成立, $\epsilon$ 如引理 4.1 中所示. 如果 $R_0<1$$(\psi_1,\psi_2,\psi_3)\in{\cal D}$, 那么 $E_0$ 是全局吸引的, 即 $\lim\limits_{t\to\infty}\| U_1(\cdot,t)-U_{10}(x)\| _{{\mathbb M}}=0, \lim\limits_{t\to\infty}\| U_2(\cdot,t)\| _{{\mathbb M}}=0$.

${\bf证}$$U^{\infty}_2(x)=\limsup\limits_{t\to\infty}U_2(x,t),x\in\overline{\Gamma}$. 显然地, $U^{\infty}_2(x)\le\epsilon\zeta_0(x)$. 于是对任意的 $x\in\overline{\Gamma}$, 我们有

$\begin{eqnarray*} U^{\infty}_2(x)&\le&U_{10}(x)\bigg[\int_0^{\infty}\gamma_2(b)\Omega_2(b)\int_{\Gamma}G_2(x,y,b)(\limsup\limits_{t\to\infty}U_2(y,t)){\rm d}y{\rm d}b\\ &&+\int_0^{\infty}\gamma_3(b)\Omega_3(b)\int_{\Gamma}G_3(x,y,b)\int_0^{\infty}\eta_2(c)\Omega_2(c)\\ &&\times \int_{\Gamma}G_2(y,z,c)(\limsup\limits_{t\to\infty}U_2(z,c)){\rm d}z{\rm d}c{\rm d}y{\rm d}b\bigg]\\ &\le&\epsilon({\cal R}\zeta_0)(x)\le R_0\epsilon\zeta_0(x), \end{eqnarray*}$

这意味着 $\lim\limits_{t\to\infty}\| U_2(\cdot,t)\| _{{\mathbb M}}=0$. 于是, 对任意的 $\varrho>0$, 存在 $T>0$, $\| U_2(\cdot,t)\| _{{\mathbb M}}\le \varrho$, $t\ge T$, 使得

$\begin{eqnarray*} U_1(x,t)&=&\int_T^te^{-\sigma(b-T)}\int_{\Gamma}G_1(x,y,b-T)[\Pi-U_2(y,t-b+T)]{\rm d}y{\rm d}b\\ &&+e^{-\sigma(t-T)}\int_{\Gamma}G_1(x,y,t-T)U_1(y,T){\rm d}y\\ &\ge&(\Pi-\varrho)\int^{t-T}_0e^{-\sigma b}\int_{\Gamma}(x,y,b){\rm d}y{\rm d}b, \end{eqnarray*}$

继而可得

$ U^{\varrho}_{10}(x)\le\liminf\limits_{t\to\infty}U_1(x,t)\le \limsup\limits_{t\to\infty}U_1(x,t)\le U_{10}(x), x\in\overline{\Gamma}, $

其中 $U^{\varrho}_{10}(x)=(\Pi-\varrho)\int_0^{\infty}e^{-\sigma b} \int_{\Omega}G_1(x,y,b){\rm d}y{\rm d}b$. 由于 $\varrho>0$, 可知 $\lim\limits_{t\to\infty}\| U_1(\cdot,t)-U_{10}(\cdot)\| _{{\mathbb M}}=0$. 定理证毕.

5 疾病的一致持久性

此节中, 我们证明当 $R_0>1$ 时, HIV 感染在人群中会持久存在的. 我们首先给出如下结论.

${\bf引理 5.1}$ 如果假设 2.1 成立, $(\psi_1,\psi_2,\psi_3)\in{\mathbb M}^0_+\times{\mathbb Q}^0_+\times{\mathbb Q}^0_+$, 那么存在一个紧的连续解半流 $\Sigma(t,\psi_1,\psi_2,\psi_3)=(U_1(\cdot,t),U_2(\cdot,t)) \in{\mathbb M}^0_+\times{\mathbb Q}^0_+,t>0$.

${\bf证}$ 根据文献[11,引理 5.1], 可得 $\Sigma(t,\psi_1,\psi_2,\psi_3)$的存在性. 接下来, 我们证明半流的紧性. 设 ${\cal B}\subset{\mathbb M}^0_+\times{\mathbb Q}^0_+\times{\mathbb Q}^0_+$ 是一个有界集, 定义

$\widetilde{B}:=\sup\limits_{(\psi_1,\psi_2,\psi_3)\in{\cal B}}B_k (\psi_1,\psi_2,\psi_3), k=1,2. $

选取 ${\psi^n=(\psi^n_1,\psi^n_2,\psi^n_3)}^{\infty}_{n=1} \in{\cal B}$, 则存在 $C>0$ 使得

$\max{\| \psi^n_1\| {{\mathbb M}},\| \psi^n_2\| {{\mathbb Q}}, \| \psi^n_3\| {{\mathbb Q}}}<C, n\in{\mathbb N}+. $

我们用 $(U^n_1,U^n_2)$ 表示具有初始条件 $(\psi^n_1,\psi^n_2,\psi^n_3)$ 的解. 接下来, 我们证明对固定的 $t>0,$${(U^n_1(\cdot,t)U^n_2 (\cdot,t)}^{\infty}_{n=1}$ 存在一个收敛子序列. 根据文献[11,引理 5.1], 对于 $(x,t)\in\overline{\Gamma}\times(0,+\infty),$$ k=1,2$, 我们有 $0\le U^n_k(x,t)\le \widetilde{B}k$. 因此, ${U^n_k(\cdot,t)}^{\infty}_{n=1}$是一致有界的. 基于 $G_1(x,y,t)$ 的性质, 可知存在 $\varsigma>0$, 使得对于任意 $\xi>0$, 当 $|x_1-x_2|<\varsigma$$x_1,x_2\in\overline{\Gamma}$ 时, 有

$\int_{\Gamma}|G_1(x_1,y,t)-G_1(x_2,y,t)|{\rm d}y<\xi. $

因此, 对于 $|x_1-x_2|<\varsigma$, 有

$|U^n_1(x_1,t)-U^n_1(x_2,t)|\le\xi\Big(\frac{\Pi+\widetilde{B}1}{\sigma}+\frac{\widetilde{B}2} {\sigma^2}+{\cal B}\Big). $

继而可知 ${U^n_1(\cdot,t)}^{\infty}_{n=1}$ 具有等度连续性. 同样地, ${U^n_2(\cdot,t)}^{\infty}_{n=1}$ 也具有等度连续性. 综上, ${U^n_k(\cdot,t)}^{\infty}_{n=1}, k=1,2$ 有收敛子序列, 从而半流的紧性得证. 引理证毕.

${\bf引理 5.2}$ 如果假设 2.1 成立以及 $R_0>1$, 那么存在 $\xi_1>0$ 使得

$ \limsup\limits_{t\to\infty}\| U_2(\cdot,t)\| _{{\mathbb M}}>\xi_1\ \mbox{for}\ (\psi_1,\psi_2,\psi_3)\in{\cal D}_0, $

其中

$\begin{eqnarray*}{\cal D}_0:&=&\bigg\{(\psi_1,\psi_2,\psi_3)\in{\mathbb M}^0_+\times{\mathbb Q}^0_+\times{\mathbb Q}^0_+, \\ && \psi_1(x)\bigg[\int_0^{\infty}\gamma_2(b)\psi_2(x,b){\rm d}b+\int_0^{\infty} \gamma_3(b)\psi_3(x,b){\rm d}b\bigg]>0\ \mbox{对些}\ x\in\Gamma\bigg\}. \end{eqnarray*}$

${\bf证}$ 对足够小的 $\xi_1>0$ 和足够大的 $g>0$, 根据 Kerin-Rutman 定理可知 $r(\widetilde{{\cal R}})>1$, 其中

$\begin{eqnarray*} (\widetilde{{\cal R}}\psi)(x):&=&\widetilde{U}_{10}(x) \bigg[\int_0^{\infty}\gamma_2(b)\Omega_2(\lambda)\int_{\Gamma}G_2(x,y,b)\psi(y){\rm d}y{\rm d}b\\ &&+\int_0^{\infty}\gamma_3(b)\Omega_3(\lambda)\int_{\Gamma}G_3(x,y,b)\int_0^{\infty}\eta_2(c)\Omega_2(\lambda) \\ &&\times \int_{\Gamma}G_2(y,z,c)\psi(c){\rm d}z{\rm d}c{\rm d}y{\rm d}b\bigg], \psi\in{\mathbb M}^0, \end{eqnarray*}$

这里 $\widetilde{U}_{10}(x)=(\Pi-\xi_1)\int_0^ge^{-\sigma b}\int_{\Gamma}G_1(x,y,b){\rm d}y{\rm d}b$, $\Omega_i(\lambda)=\exp\{-\int_0^b(\sigma+\eta_i(s)+\delta_i(s)+\lambda){\rm d}s\},$$i=2,3$. Set $\widetilde{R}^+_0=r(\widetilde{{\cal R}}^+)$, 其中

$\begin{eqnarray*} (\widetilde{{\cal R}}^+\psi)(x):&=&\int_0^{\infty}\gamma_2(b)\Omega_2(\lambda)\int_{\Gamma}G_2(x,y,b)\widetilde{U}_{10}(y)\psi(y){\rm d}y{\rm d}b\\ &&+\int_0^{\infty}\gamma_3(b)\Omega_3(\lambda)\int_{\Gamma}G_3(x,y,b)\int_0^{\infty}\eta_2(c)\Omega_2(\lambda)\int_{\Gamma}G_2(y,z,c)\widetilde{U}_{10}(c)\psi(c){\rm d}z{\rm d}c{\rm d}y{\rm d}b. \end{eqnarray*}$

从而可得 $R^+_0=r(\widetilde{{\cal R}}^+)=r(\widetilde{{\cal R}})>1$. 进而可知 $R^+_0\zeta^+_0(\cdot)=\widetilde{{\cal R}}^+\zeta^+_0$, 其中 $\zeta^+_0\in{\mathbb M}^0_+\setminus\{0\}$. 假设引理 5.2 结论不成立, 那么存在足够大的 $T^*>0$ 使得 $U_2(x,t)\le\xi_1$, $t\ge T^*$. 于是, 可得 $U_1(x,t)=(\Pi-\xi_1)\int_0^{t-T^*}e^{-\sigma b}\int_{\Gamma}G_1(x,y,b){\rm d}y{\rm d}b$.$(x,t)\in\overline{\Gamma}\times(0,+\infty)$, 我们有

$\begin{matrix}\label{11} U_2(x,t)&\ge& \widetilde{U}_{10}(x)\bigg[\int_0^t\gamma_2(b)\Omega_2(b)\int_{\Gamma}G_2(x,y,b)U_2(y,t-b){\rm d}y{\rm d}b\\ &&+\int_0^{t}\gamma_3(b)\Omega_3(b)\int_{\Gamma}G_3(x,y,b)\int_0^{t-b}\eta_2(c)\Omega_2(c) \\ &&\times \int_{\Gamma}G_2(y,z,c)U_2(z,t-b-c){\rm d}z{\rm d}c{\rm d}y{\rm d}b\bigg]. \end{matrix}$

${\cal L}(U_2)(\lambda)=\int_0^{\infty}e^{-\lambda t}U_2(x,t){\rm d}t$ 并对式 (5.1) 两边同乘以 $\zeta^+_0(x)$ 可得

$\begin{eqnarray*} \int_{\Gamma}\zeta^+_0(x){\cal L}(U_2(x,\cdot))(\lambda){\rm d}x &\ge& \int_{\Gamma}\bigg[\int_0^{\infty}\gamma_2(b)\Omega_2(\lambda)\int_{\Gamma}G_2(x,y,b)\widetilde{U}_{10}(x)\zeta^+_0(x){\rm d}x{\rm d}b \\ &&+\int_0^{\infty}\gamma_3(b)\Omega_3(\lambda)\int_{\gamma}G_3(x,y,b)\int_0^{\infty}\eta_2(c)\Omega_2(\lambda) \\ &&\times\int_{\Gamma}G_2(y,z,c)\widetilde{U}_{10}(x)\zeta^+_0(x)\bigg]{\cal L}[U_2(y,\cdot)](\lambda){\rm d}y\\ &=&\int_{\Gamma}(\widetilde{{\cal R}}^+\zeta^+_0)(y){\cal L}[U_2(y,\cdot)](\lambda){\rm d}y\\ &=&R^+_0\int_{\Gamma}\zeta^+_0{\cal L}[U_2(x,\cdot)](\lambda){\rm d}x, \end{eqnarray*}$

这与 $R^+_0>1$ 矛盾. 引理证毕.

${\bf定理 5.1}$ 如果假设 2.1 成立以及 $R_0>1$, 那么系统 (1.5) 是一致持久的, 即存在 $\xi=\xi(x,b)>0$ 使得 $ \liminf\limits_{t\to\infty}u_2(x,b,t)>\xi$, $(\psi_1,\psi_2,\psi_3)\in {\cal D}_0$. 此外, 系统 (1.5) 至少存在一个空间依赖的正平衡态 $E^+=(U^+_1(x),U^+_2(x))$.

${\bf证}$ 类似文献[11,命题5.3] 中的证明过程, 我们可以断言存在 $\xi_0>0$ 使得

$\lim\limits_{t\to\infty}\| U_2(\cdot,t)\| _{{\mathbb M}}>\xi_0 {\rm for } (\psi_1,\psi_2,\psi_3)\in{\cal D}_0. $

$\kappa(t)=(U_1(\cdot,t),u_2(\cdot,\cdot,t))$ 是算子 $\Sigma$ 使得 $\Sigma(\kappa(s),t)=\kappa(t_s)$, $(s,t)\in{\Bbb R} \times(0,+\infty)$ 的总轨迹. 那么有

$\begin{equation}\label{12} u_2(x,b,t)=\Omega_2(b)\int_{\Gamma}G_2(x,y,b)U_2(y,t-b){\rm d}y,\ (x,b,t)\in\overline{\Gamma}\times[0,+\infty)\times{\Bbb R}. \end{equation}$

$b>0,x\in\Gamma$, 定义$\rho_k:{\mathbb M}^0_+\times{\mathbb Q}^0_+\times{\mathbb Q}^0_+\to{\Bbb R} ^+, k=1,2$的算子如下

$\rho_1(\kappa(t)):=\| U_2(\cdot,t)\| {{\mathbb M}}, \rho_2(\kappa(t)):=\Omega_2(b)\int{\Gamma}G_2(x,y,b)U_2(y,t-b){\rm d}y. $

由于 $\lim\limits_{t\to\infty}\| U_2(\cdot,t)\| {{\mathbb M}}>\xi_0$, 因此$(\psi_1,\psi_2,\psi_3)\in{\cal D}0$的一致 $\rho_1-$持久性(见文献[26,定义3.1]) 成立. 根据文献[26,推论4.22], 可假设$\kappa(t)$ 是具有预紧缩值域的总轨迹, 且$\inf{t\in{\Bbb R} }\rho_1(\kappa(t))=\inf{t\to{\Bbb R} }\| U_2(\cdot,t)\| {{\mathbb M}}>0$, 因而可得 $\rho_2(\kappa(0))>0$. 根据文献[26,推论4.22],可知存在$\xi=\xi(x,b)>0$ 使得 $\inf{t\to+\infty}\rho_2(\kappa(t))>\xi$. 再根据 (5.2)式, 可知, 对于 $b>0$, 有 $\liminf\limits_{t\to+\infty}u_2(x,b,t)>\xi$.

接下来, 我们证明系统 (1.5) 存在空间依赖的正稳态 $E^+$. 为此, 我们首先定义 ${\mathbb M}^0_+$ 上的线性算子

$\begin{eqnarray*} (\Phi\phi)(x):&=&\int_0^{\infty}\gamma_2(b)\Omega_2(b)\int_{\gamma}G_2(x,y,b)\phi(y){\rm d}y{\rm d}b\\ &&+\int_0^{\infty}\gamma_3(b)\Omega_3(b)\int_{\Gamma}G_3(x,y,b)\int_0^{\infty}\eta_2(c)\Omega_2(c)\int_{\Gamma}G_2(y,z,c)\phi(c){\rm d}z{\rm d}c{\rm d}y{\rm d}b,\ x\in\overline{\Gamma}. \end{eqnarray*}$

然后,正稳态满足方程

$\begin{eqnarray*} \left\{ \begin{array}{ll} &0=d_1\Delta U^+_1(x)+\Pi-(\sigma+(\Phi U^+_2)(x))U^+_1(x),\\ &U^+_2(x)=U^+_1(x)(\Phi U^+_2)(x). \end{array} \right. \end{eqnarray*}$

从而可得

$\begin{eqnarray*} &&U^+_1(x)=P^x\left[\Pi\int_{0}^{s\Gamma}\exp\bigg\{-\int_0^{\tau}(\sigma+(\Phi U^+_2))\chi_{U_2}{\rm d}U_2\bigg\}{\rm d}\tau\right],\\ &&U^+_2(x)=P^x\left[\Pi\int_0^{s\Gamma}\exp\bigg\{-\int_0^{\tau}(\sigma+\Phi U^+_2)(\chi_{U_2}){\rm d}U_2\bigg\}{\rm d}\tau\right](\Phi U^+_2)(x), \end{eqnarray*} $

其中 Itô是关于 $d_1\Delta$ is $\{\chi_t\}$ 的扩散, $P^x$ 表示 $\chi_t$ 的概率分布的期望, $s\Gamma:=\inf\{t>0:\chi_t\notin\Gamma\}$. 定义非线性算子 $L^1_+(\Gamma)$

$\begin{eqnarray*} {\cal A}(\phi)(x):=P^x\left[\Pi\int_0^{s\Gamma}\exp\left\{-\int_0^{\tau}(\sigma+(\Phi\phi)(\chi_{U_2})){\rm d}U_2\right\}{\rm d}\tau\right](\Phi\phi)(x),\ \phi\in L^1_+(\Gamma),\ x\in\overline{\Gamma}, \end{eqnarray*}$

以及从 $L^1_+(\Gamma_{\varpi})$$L^1_+(\Gamma)$ 的算子

$\begin{eqnarray*} (\Phi_{\varpi}\phi)(x):&=&\int_0^{\infty}\gamma_2(b)\Omega_2(b)\int_{\Gamma}G_2(x,y,b)\phi(y){\rm d}y{\rm d}b\\ &&+\int_0^{\infty}\gamma_3(b)\Omega_3(b)\int_{\Gamma}G_3(x,y,b)\int_{0}^{\infty}\eta_2(c)\Omega_2(c)\int_{\Gamma}G_2(y,z,c)\phi(c){\rm d}z{\rm d}c{\rm d}y{\rm d}b, \end{eqnarray*} $

其中 $\phi\in L^1_+(\Gamma_{\varpi}),\ \bar\omega>0$. 定义如下从 $L^1_+(\Gamma_{\varpi})$$L^1_+(\Gamma_{\varpi})$ 的算子

$\begin{eqnarray*} &&{\cal A}_{\varpi}(\phi)(x):=P^x\left[\Pi\int_0^{s\Gamma}\exp\left\{-\int_0^{\tau}(\sigma+(\Phi_{\varpi}\phi)(\chi_{U_2})){\rm d}U_2\right\}{\rm d}\tau\right](\Phi_{\varpi}\phi)(x)\Big|_{\Gamma_{\varpi}},\\ &&({\cal R}_{\varpi}\phi)(x):=P^x\left[\Pi\int_0^{s\Gamma}e^{-\sigma\tau}{\rm d}\tau\right](\Phi_{\varpi}\phi)(x)\Big|_{\Gamma_{\varpi}}=U^0_1(x)(\Phi_{\varpi}\phi)(x)\Big|_{\Gamma_{\varpi}}, \end{eqnarray*} $

其中 ${\Gamma_{\varpi}}$$\Gamma$ 的一组适当的子域, 满足 $d(\Gamma_{\varpi},\partial \Gamma)>0$. 对于 $\varpi_1<\varpi_2$, 有 $\Gamma_{\varpi_1}\subset\Gamma_{\varpi_2}$$\lim\limits_{\varpi\to 0}\Big|\Gamma\setminus\Gamma_{\varpi}\Big|{\Gamma{\varpi}}=0$, 其中 $\Big|{\Gamma{\varpi}}$ 表示域在 $\Gamma_{\varpi}$ 上的约束. 注意到, 当 ${\cal A}{\varpi}$$L^1+(\Gamma_{\varpi})\setminus{0}$ 中具有正定点 $\phi^+{\varpi}$ 时, 则 ${\cal A}$ 有一个正定点.

一方面, 由于 $R_0=r({\cal R})>1$, 可知显然存在一个足够小的常数 $\varpi>0$ 使得 $r({\cal R}{\varpi})>1$. 对于固定的 $\varpi$, 可知 ${\cal A}{\varpi}(L^1+(\gamma_{\varpi})\subset L^1_+(\Gamma_{\varpi})$${\cal A}{\varpi}(0)=0$. 另一方面, 根据文献[17,引理 7.1], 可知 ${\cal A}'{\varpi}(0)={\cal R}{\varpi}$, ${\cal A}'{\varpi}(\infty)=0$ 相对于 $L^1_+(\Gamma_{\varpi})$. 此外, ${\cal A}'{\varpi}(\infty)$ 的谱是 0, 并位于以 $0$ 为中心的半径小于 $1$ 的圆内. 事实上, ${\cal A}'{\varpi}(0)={\cal R}{\varpi}$ 有一个正特征函数 $\vartheta_{\varpi}\in L^1_+(\Gamma_{\varpi})\setminus{0}$ 使得 ${\cal A}'(0)\vartheta_{\varpi}={\cal R}{\varpi}\vartheta{\varpi}=r({\cal R}{\varpi})\vartheta{\varpi}$. 根据Krein-Rutman定理, 由于 $r({\cal R}{\varpi})>1$, ${\cal A}'(0)$ 没有任何特征函数对应特征值1. 此外, ${\cal A}{\varpi}$ 的紧性得证 (详见文献[17,引理7.1]). 应用 Krasnoselskii 不动点定理, 可得 ${\cal A}{\varpi}$ 至少有一个非平凡不动点 $\phi^+{\varpi}\in L^1_+(\Gamma_{\varpi})\setminus{0}$, 这意味着系统 (1.5) 至少有一个空间依赖的正稳定态解 $E^+=(U^+_1,U^+_2)$. 证毕.

6 数值模拟

在本节中, 我们根据中国50岁及以上感染艾滋病人群的报告病例, 对系统 (1.5) 进行数值模拟. 与文献[17]类似, 我们考虑矩形域 $\Gamma=(0,w_1)\times(0,w_2)\subset {\Bbb R} ^2, w_1=m,w_2=1/m, m>0$. 因此, 格林函数 $G_i(x,y,t)$ 形式为

$\begin{eqnarray*} G_i(x_1,x_2,y_1,y_2,t):=\frac{4}{w_1w_2}\sum_{p,q=1}^{\infty}\sin\frac{p\pi x_1}{w_1}\sin\frac{q\pi x_2}{w_2}\sin\frac{p\pi y_1}{w_1}\sin\frac{q\pi y_2}{w_2}e^{-d_i\left(\frac{p^2}{w^2_1}+\frac{q^2}{w^2_2}\right)\pi^2t}, \end{eqnarray*}$

其中 $x_1,y_1\in[w_1]$, $x_2,y_2\in[w_2], i=2,3$. 根据文献[5-6], 我们假设模型 (1.5) 参数值与初值如下

$\begin{equation}\label{100} \begin{array}{ll} \Pi=3960000,\ \sigma=0.00207,\ \delta_1=0.0456,\ \delta_2=0.066,\ \eta_2=3\times 10^{-3},\ \eta_3=6\times 10^{-3},\\ \gamma_i(b)=\gamma_i\times 10^{-9},\ i=2,3,\ d_1=0.05,\ d_2=0.05, d_3=0.04, \delta_3=0.05,\ b=10. \end{array} \end{equation} $
$\begin{equation} \begin{array}{ll} \psi_1(x)=U^0_1(x),x\in\overline{\Gamma},\\ \psi_i(x,b)=\left\{ \begin{array}{ll} U_i(0) e^{-(\sigma+\eta_i+\delta_i)b}\prod_{k=1}^{2}(x_k-0.2w_k)(0.8w_k-x_k), (x_1,x_2)\in\Gamma_0,\\ 0,\ \mbox{其它}, \end{array} \right. \end{array} \end{equation}$

其中 $i=2,3$, $U_2(0)=2318,U_3(0)=1128$, $\Gamma_0=(0.2w_1,0.8w_1)\times(0.2w_2,0.8w_2)\subset \Gamma$.

利用文献[27,3.1.2 节]中的 Fredholm 离散化方法, 我们可以得到 $R_0$ 随迭代次数 $m\in(0,40]$ 的变化图.当$\gamma_1=5.7, \gamma_2=6.8$$\gamma_1=7.6, \gamma_2=9$$R_0$ 的值变化分别在图1(a)(b)中给出. 这些图表明: 基本再生数 $R_0$ 随着 $m$ 的增加先增加后减少, 并且 $R_0$的值在图 1(a)(b)中分别在 $m=10$ 时达到最大值 $0.927$$1.492$.图 2 中, 我们可以看到 当 $R_0<1$$U_2(x,t)$ 密度分布曲面随着时间的推移逐渐收敛于零平面, 这与定理 4.2 的结果一致. 图 3 表明在 $R_0>1$$U_2(x,t)$ 密度分布曲面随着时间的推移远离零平面, 验证了定理 5.1 的结果. 图 4 比较了扩散率 $d_1$, $d_2$$d_3$$R_0$ 的影响. 从中我们可以看到, 当这些扩散率变小时 $R_0$ 变大. 此外,$R_0$$d_1$ 最敏感, 对 $d_3$ 最不敏感.

图 1

图 1   (a) $R_0$ 的值, 其中参数值同式 (6.1), $\gamma_1=5.7, \gamma_2=6.8$; (b) $R_0$ 的值, 其中参数值同式 (6.1) 以及 $\gamma_1=7.6, \gamma_2=9$}


图 2

图 2   $R_0\approxeq 0.927<1$$U_2(x,t)$ 的演化行为


图 3

图 3   $R_0\approxeq 1.492\gg 1$$U_2(x,t)$ 的演化行为


图 4

图 4   扩散系数 $d_n (n=1,2,3)$ 对基本再生数 $R_0$ 的影响. 其他参数值同式 (6.1)


7 结论

在本文中, 我们建立了一类具有齐次 Dirichlet 边界条件的年龄空间结构 HIV/AIDS 传染病模型. 这种边界条件给动力学分析带来了挑战. 首先, 由于系统的无病稳态是非常数的, 即使模型参数是空间无关的, 这与齐次Neumann 边界条件下的空间同质模型的结论不同. 其次, 由于反应扩散方程和积分方程的耦合系统的耗散性不能直接得到, 我们证明了系统的一致持久性之后, 不能直接获得依赖于空间的正稳态解的存在性. 为了克服这些困难, 我们首先研究了系统的适定性, 并导出了基本再生数 $R_0$ 的泛函表达式. 然后我们得到了无病稳态的全局吸引性, 并证明了系统的一致持久性. 最后, 通过使用强 Fréchet 导数、紧致性理论和 Krasnoselskii 不动点定理证明了空间依赖的正稳态的存在性. 数值模拟部分, 我们在二维空间区域中通过数值模拟验证了文中理论结果. 值得指出的是, 本文所用的方法也可推广到研究其他具有年龄空间结构和齐次 Dirichlet 边界条件的传染病模型中.

参考文献

UNAIDS: AIDS by the numbers in 2020. https://www.unaids.org/en

URL    

Cuadros D F, Li J, Branscum A J, et al.

Mapping the spatial variability of HIV infection in Sub-Saharan Africa: Efective information for localized HIV prevention and control

Sci Rep, 2017, 7: 9093

DOI:10.1038/s41598-017-09464-y      PMID:28831171      [本文引用: 1]

Under the premise that in a resource-constrained environment such as Sub-Saharan Africa it is not possible to do everything, to everyone, everywhere, detailed geographical knowledge about the HIV epidemic becomes essential to tailor programmatic responses to specific local needs. However, the design and evaluation of national HIV programs often rely on aggregated national level data. Against this background, here we proposed a model to produce high-resolution maps of intranational estimates of HIV prevalence in Kenya, Malawi, Mozambique and Tanzania based on spatial variables. The HIV prevalence maps generated highlight the stark spatial disparities in the epidemic within a country, and localize areas where both the burden and drivers of the HIV epidemic are concentrated. Under an era focused on optimal allocation of evidence-based interventions for populations at greatest risk in areas of greatest HIV burden, as proposed by the Joint United Nations Programme on HIV/AIDS (UNAIDS) and the United States President's Emergency Plan for AIDS Relief (PEPFAR), such maps provide essential information that strategically targets geographic areas and populations where resources can achieve the greatest impact.

Ayalew K A, Manda S, Cai B.

A comparison of Bayesian spatial models for HIV mapping in South Africa

Int J Environ. Res Public Health, 2021, 18(21): 11215

DOI:10.3390/ijerph182111215      URL     [本文引用: 1]

Despite making significant progress in tackling its HIV epidemic, South Africa, with 7.7 million people living with HIV, still has the biggest HIV epidemic in the world. The Government, in collaboration with developmental partners and agencies, has been strengthening its responses to the HIV epidemic to better target the delivery of HIV care, treatment strategies and prevention services. Population-based household HIV surveys have, over time, contributed to the country’s efforts in monitoring and understanding the magnitude and heterogeneity of the HIV epidemic. Local-level monitoring of progress made against HIV and AIDS is increasingly needed for decision making. Previous studies have provided evidence of substantial subnational variation in the HIV epidemic. Using HIV prevalence data from the 2016 South African Demographic and Health Survey, we compare three spatial smoothing models, namely, the intrinsically conditionally autoregressive normal, Laplace and skew-t (ICAR-normal, ICAR-Laplace and ICAR-skew-t) in the estimation of the HIV prevalence across 52 districts in South Africa. The parameters of the resulting models are estimated using Bayesian approaches. The skewness parameter for the ICAR-skew-t model was not statistically significant, suggesting the absence of skewness in the HIV prevalence data. Based on the deviance information criterion (DIC) model selection, the ICAR-normal and ICAR-Laplace had DIC values of 291.3 and 315, respectively, which were lower than that of the ICAR-skewed t (348.1). However, based on the model adequacy criterion using the conditional predictive ordinates (CPO), the ICAR-skew-t distribution had the lowest CPO value. Thus, the ICAR-skew-t was the best spatial smoothing model for the estimation of HIV prevalence in our study.

Wu P, Zhao H.

Mathematical analysis of an age-structured HIV/AIDS epidemic model with HAART and spatial diffusion

Nonlinear Analysis: RWA, 2021, 60: 103289

DOI:10.1016/j.nonrwa.2021.103289      URL     [本文引用: 2]

Zhao H, Wu P, Ruan S.

Dynamic analysis and optimal control three-age-class HIV/AIDS epidemic model in China

Disc Conti Dyn Syst Seri B, 2020, 25(9): 3491-3521

[本文引用: 3]

Wu P, Zhao H.

Dynamical analysis of a nonlocal delayed and diffusive HIV latent infection model with spatial heterogeneity

J Franklin Institute, 2021, 358(10): 5552-5587

DOI:10.1016/j.jfranklin.2021.05.014      URL     [本文引用: 2]

Gao Y, Wang J.

Threshold dynamics of a delayed nonlocal reaction-diffusion HIV infection model with both cell-free and cell-to-cell transmissions

J Math Anal Appl, 2020, 488(1): 124047

DOI:10.1016/j.jmaa.2020.124047      URL     [本文引用: 1]

Ge Q, Wang X, Rong L.

A delayed reaction-diffusion viral infection model with nonlinear incidences and cell-to-cell transmission

International Journal of Biomathematics, 2021, 14(8): 2150100

DOI:10.1142/S179352452150100X      URL     [本文引用: 1]

In this paper, we propose a reaction–diffusion viral infection model with nonlinear incidences, cell-to-cell transmission, and a time delay. We impose the homogeneous Neumann boundary condition. For the case where the domain is bounded, we first study the well-posedness. Then we analyze the local stability of homogeneous steady states. We establish a threshold dynamics which is completely characterized by the basic reproduction number. For the case where the domain is the whole Euclidean space, we consider the existence of traveling wave solutions by using the cross-iteration method and Schauder’s fixed point theorem. Finally, we study how the speed of spread in space affects the spread of cells and viruses. We obtain the existence of the wave speed, which is dependent on the diffusion coefficient.

Wang W, Wang X, Feng Z.

Time periodic reaction-diffusion equations for modeling 2-LTR dynamics in HIV-infected patients

Nonlinear Analysis: Real World Applications, 2021, 57: 103184

DOI:10.1016/j.nonrwa.2020.103184      URL     [本文引用: 1]

王晶囡, 杨德中.

具有时滞扩散效应的病原体-免疫模型的稳定性及分支

数学物理学报, 2021, 41A(4): 1204-1217

[本文引用: 1]

Wang J N, Yang D Z.

Stability and Bifurcation analysis of a delayed diffusive pathogen-immune model

Acta Math Sci, 2021, 41A(4): 1204-1217

[本文引用: 1]

Chekroun A, Kuniya T.

An infection age-space structured SIR epidemic model with Neumann boundary condition

Appl Anal, 2020, 99: 1972-1985

DOI:10.1080/00036811.2018.1551997      URL     [本文引用: 6]

Yang J, Xu R, Li J.

Threshold dynamics of an age-space structured brucellosis disease model with Neumann boundary condition

Nonlinear Analysis: RWA, 2019, 50: 192-217

DOI:10.1016/j.nonrwa.2019.04.013      URL     [本文引用: 3]

liu P, Li H.

Global behavior of a multi-group SEIR epidemic model with age structure and spatial diffusion

Math Biosci Eng, 2020, 17: 7248-7273

DOI:10.3934/mbe.2020372      PMID:33378896      [本文引用: 1]

Different epidemic models with one or two characteristics of multi-group, age structure and spatial diffusion have been proposed, but few models take all three into consideration. In this paper, a novel multi-group SEIR epidemic model with both age structure and spatial diffusion is constructed for the first time ever to study the transmission dynamics of infectious diseases. We first analytically study the positivity, boundedness, existence and uniqueness of solution and the existence of compact global attractor of the associated solution semiflow. Based on some assumptions for parameters, we then show that the disease-free steady state is globally asymptotically stable by utilizing appropriate Lyapunov functionals and the LaSalle's invariance principle. By means of Perron-Frobenius theorem and graph-theoretical results, the existence and global stability of endemic steady state are ensured under appropriate conditions. Finally, feasibility of main theoretical results is showed with the aid of numerical examples for model with two groups which is important from the viewpoint of applications.

Wang J, Zhang R, Gao Y.

Global threshold dynamics of an infection age-space structured HIV infection model with Neumann boundary condition

J Dyn Diff Equat, 2021, https://doi.org/10.1007/s10884-021-10086-2

URL     [本文引用: 1]

An Q, Wang C, Wang H.

Analysis of a spatial memory model with nonlocal maturation delay and hostile boundary condition

Discrete Contin Dyn Syst Ser A, 2020, 40(10): 5845-5868

DOI:10.3934/dcds.2020249      URL     [本文引用: 1]

Yi T, Zou X.

On Dirichlet problem for a class of delayed reaction-diffusion equations with spatial non-locality

J Dyn Diff Equat, 2013 25(4): 959-979

[本文引用: 1]

Chekroun A, Kuniya T.

Global threshold dynamics of an infection age-structured SIR epidemic model with diffusion under the Dirichlet boundary condition

J Differential Equations, 2020, 269: 117-148

DOI:10.1016/j.jde.2020.04.046      URL     [本文引用: 6]

Wang X, Sun H, Yang J.

Temporal-spatial analysis of an age-space structured foot-and-mouth disease model with Dirichlet boundary condition

Chaos, 2021, 31: 053120

DOI:10.1063/5.0048282      URL     [本文引用: 1]

Shen M, Xiao Y, Rong L.

Global stability of an infection-age structured HIV-1 model linking within-host and between-host dynamics

Math Biosci, 2015, 263: 37-50

DOI:10.1016/j.mbs.2015.02.003      PMID:25686694      [本文引用: 1]

Although much evidence shows the inseparable interaction between the within-host progression of HIV-1 infection and the transmission of the disease at the population level, few models coupling the within-host and between-host dynamics have been developed. In this paper, we adopt the nested approach, viewing the transmission rate at each stage (primary, chronic, and AIDS stage) of HIV-1 infection as a saturated function of the viral load, to formulate an infection-age structured epidemic model. We explicitly link the individual and the host population scale, and derive the basic reproduction number R0 for the coupled system. To analyze the model and perform a detailed global dynamics analysis, two Lyapunov functionals are constructed to prove the global asymptotical stability of the disease-free and endemic equilibria. Theoretical results indicate that R0 provides a threshold value determining whether or not the disease dies out. Numerical simulations are presented to quantitatively investigate the influence of the within-host viral dynamics on between-host transmission dynamics. The results suggest that increasing the effectiveness of inhibitors can decrease the basic reproduction number, but can also increase the overall infected population because of a lower disease-induced mortality rate and a longer lifespan of HIV infected individuals. Copyright © 2015 Elsevier Inc. All rights reserved.

Gao Y, Wang J.

Threshold dynamics of a delayed nonlocal reaction-diffusion HIV infection model with both cell-free and cell-to-cell transmissions

J Math Anal Appl, 2020, 488: 124047

DOI:10.1016/j.jmaa.2020.124047      URL     [本文引用: 1]

Yang J, Jin Z, Xu F.

Threshold dynamics of an age-space structured SIR model on heterogeneous environment

Appl Math Lett, 2019, 96: 69-74

DOI:10.1016/j.aml.2019.03.009      URL     [本文引用: 1]

Ito S.

Diffusion Equations

Providence, RI: American Mathematical Society, 1992

[本文引用: 1]

Mckenna P J, Walter W.

On the Dirichlet problem for elliptic systems

Appl Anal, 1986, 21: 207-224

DOI:10.1080/00036818608839592      URL     [本文引用: 1]

Protter M H, Weinberger H F. Maximum Principle in Differential Equations. New York: Springer-Verlag, 1984

[本文引用: 1]

Wang W, Zhao X Q.

Spatial invasion threshold of Lyme disease

SIAM J Appl Math, 2015 73: 1142-1170

[本文引用: 1]

Smith H L.

Monotone Dynamical Systems: An Introduction to the Theory of Competitive and Cooperative Systems

Providence, RI: American Mathematical Society, 1995

[本文引用: 3]

Chatelin F.

The spectral approximation of linear operators with applications to the computation of eigenelements of differential and integral operators

SIAM Rev, 1981, 23: 495-522

DOI:10.1137/1023099      URL     [本文引用: 1]

/