数学物理学报, 2020, 40(6): 1646-1669 doi:

论文

具有对数敏感度和混合边界的一维趋化模型解的整体存在性和收敛性

王娟,, 原子霞,

Global Existence and Convergence of Solutions to a Chemotactic Model with Logarithmic Sensitivity and Mixed Boundary Conditions

Wang Juan,, Yuan Zixia,

通讯作者: 原子霞, E-mail: yuanzixia@uestc.edu.cn

收稿日期: 2019-08-30  

基金资助: 电子科技大学中央高校基本科研业务费.  ZYGX2019J096

Received: 2019-08-30  

Fund supported: the Fundamental Research Funds for the Central Universities, UESTC.  ZYGX2019J096

作者简介 About authors

王娟,E-mail:Juanwangmath@126.com , E-mail:math@126.com

Abstract

This paper investigates the following chemotactic model with logarithmic sensitivity in a one-dimensional bounded domain: By using a Cole-Hopf type transformation, we transform the above singular repulsive chemotaxis model into a non-singular system of the form Then under some mixed boundary conditions, we prove the global existence and exponential convergence of solutions to the initial-boundary value problem of the above system with regular initial data.

Keywords: Chemotaxis ; Logarithmic sensitivity ; Mixed boundary conditions ; Global existence ; Exponential convergence

PDF (438KB) 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

王娟, 原子霞. 具有对数敏感度和混合边界的一维趋化模型解的整体存在性和收敛性. 数学物理学报[J], 2020, 40(6): 1646-1669 doi:

Wang Juan, Yuan Zixia. Global Existence and Convergence of Solutions to a Chemotactic Model with Logarithmic Sensitivity and Mixed Boundary Conditions. Acta Mathematica Scientia[J], 2020, 40(6): 1646-1669 doi:

1 引言

具有对数敏感度的趋化.趋化性在生物学和生态学中广泛存在, 它是指细胞对环境中某些具有刺激性的化学物质所产生的定向趋近或者远离的运动, 称之为吸引趋化或者排斥趋化.这种最基本的生理反应对细胞趋利避害和适应环境具有重要的作用.为了研究趋化性运动, 许多专家和学者对其进行了数学建模和分析.特别地, Keller和Segel[10]于20世纪70年代提出了一类抛物型偏微分方程组用来描述生物种群密度以及化学物质浓度的动态演化.一般形式的Keller-Segel模型为

$ \begin{equation} \left\{ \begin{array}{lll} \partial_{t}u = D\triangle u-\nabla\cdot\big(\chi u \nabla \phi(v)\big), \quad & x\in \Omega, \quad t>0, \\ \partial_{t}v = \varepsilon \triangle v+g(u, v), \quad & x\in \Omega, \quad t>0, \\ \end{array} \right. \end{equation} $

其中$ \Omega $$ {{\Bbb R}} ^n $中的有界区域或全空间, $ u $表示细胞密度, $ v $表示化学浓度.参数$ D $$ (D>0) $$ \varepsilon $$ (\varepsilon\geq0) $分别表示细胞和化学物质的扩散率. $ \chi $是趋化敏感度, $ \chi>0 $时表示吸引趋化, $ \chi<0 $时表示排斥趋化.未知函数$ g(u, v) $表示化学物质的增加或减少, 而$ \phi(v) $表示趋化敏感度函数.

在过去的几年里, 关于模型(1.1)的数学分析和研究主要集中于对函数$ \phi(v) $$ g(u, v) $的各种假设下, 在有界或无界区域中相应模型的解的性质的研究.特别地, 通过取$ \phi(v) = \ln v $, Adler[1]在实验中得到了带奇异的对数敏感度的趋化模型, Keller和Segel[11]首次在其开创性论文中使用这种趋化模型来描述由细菌趋化运动而形成的行波带传播.随后Othmer和Stevens在文献[23]中给出了关于对数敏感度的数学推导.

下面回顾关于具有对数敏感度函数$ \phi(v) = \ln v $的趋化模型(1.1)在数学领域的重要结果.对于$ \chi>0 $的情形, 当$ g(u, v) = u-v $时, Ni[22]研究了方程组(1.1)在Neumann边界条件下尖峰层稳态的存在性.当$ g(u, v) = -uv+\beta v $ ($ \beta \geq 0 $)时, Li和Wang[19]证明了对应的方程组的行波解的存在性、非线性稳定性和渐近衰减率.文献[31]研究了方程组的Cauchy问题, 首次发现了这个系统中隐藏的交互式耗散结构并且建立了能量耗散.最近, Hou等人[7]指出, 当规定为Dirichlet边界条件时, 这个系统在有界区间的边界上有边界层.文献[8]进一步证明了边界层在半平面中的收敛性, 而文献[6]证明了其稳定性.在多维空间中的适定性以及数值分析则在文献[27]中得到研究.

对于$ \chi<0 $, $ g(u, v) = uv-\mu v $时的情形, 研究主要集中在一维区间中方程组

$ \begin{equation} \left\{ \begin{array}{lll} \partial_{t}u = Du_{xx}-(\chi u(\ln v)_{x})_{x}, \quad & x\in \Omega, \quad t>0, \\ \partial_{t}v = \varepsilon v_{xx}+uv-\mu v, \quad & x\in \Omega, \quad t>0\\ \end{array} \right. \end{equation} $

解的性质的研究, 其中$ \Omega\subset{{\Bbb R}} , \, \, \ \varepsilon\geq0 $.文献[12, 23]提出了方程组$ (1.2) $用来模拟增强随机游动.为了方便研究方程组$ (1.2) $, 常用的方法是采用Cole-Hopf变换$ q = v_{x}/v $, 缩放$ \widetilde{t} = -\chi t/D $, $ \widetilde{x} = x\sqrt{-\chi}/D $, $ \widetilde{q} = q\sqrt{-\chi} $, 并令$ p = u $, $ -\chi = D = 1 $, 将方程组$ (1.2) $变换为如下方程组

$ \begin{equation} \left\{ \begin{array}{lll} p_t = p_{xx} + (pq)_x, \\ q_t = \varepsilon q_{xx} + \varepsilon (q^2)_x + p_x. \end{array} \right. \end{equation} $

与原方程组$ (1.2) $相比, 变换后的方程组$ (1.3) $去除了对数奇异性.当$ \varepsilon = 0 $时, Zhang和Zhu在文献[35]中对区间$ \Omega = (0, 1) $内具有齐次Dirichlet边界条件的方程组$ (1.3) $建立了解的全局存在性, 文献[4]研究了相应的Cauchy问题. Li和Wang在文献[16-17]中建立了行波解的存在性和非线性稳定性, 文献[14-15]分别建立了在大初值下经典解的大时间行为.当$ \varepsilon>0 $时, 在一维有界区间$ \Omega = (0, 1) $内, Fontelos等在文献[3]中建立了经典解的存在性, 并且证明了其在小扰动下收敛到稳态解. Li和Wang在文献[18]中建立了行波解的非线性稳定性和存在性, 文献[32]则建立了方程组$ (1.3) $初边值问题解的全局适定性和大时间行为.当$ \varepsilon\geq0 $时, 在一维有界区间中, 对具有Neumann-Dirichlet边界条件的方程组$ (1.3) $, Tao和Wang等在文献[28]中证明了大初值条件下解的全局存在性.

关于多维空间中排斥趋化模型和吸引趋化模型的Cauchy问题、Dirichlet边值问题和Neumann边值问题的研究也有很多优秀的研究成果(参见文献[2, 5, 9, 13, 20, 21, 24, 36]).在趋化系统中边界条件对全局流和可能的奇性都很重要(参见文献[25-26, 29-30, 33, 34]), 因此, 本文主要研究在混合边界条件下方程组$ (1.3) $解的全局存在性.

主要结果.本文研究一维有界区间$ \Omega = (0, 1) $内方程组$ (1.3) $在混合边界和初值条件

$ \begin{equation} \left\{ \begin{array}{lll} p|_{x = 0} = \overline{p}, \quad p_x|_{x = 1} = 0, \quad q|_{x = 0, x = 1} = 0, \quad & \, \, \, \, \varepsilon>0, \\ p|_{x = 0} = \overline{p}, \quad p_x|_{x = 1} = 0, \quad &\, \, \, \, \varepsilon = 0, \\ (p, q)(x, 0) = (p_0, q_0)(x) \end{array} \right. \end{equation} $

下解的全局存在性和收敛性, 这里$ \overline{p} $是一个非负常数.

bf定理1.1  假设初始值$ (p_{0}, q_{0}) \in (H^{2}(0, 1))^2 $, $ p_0(x) $非负且$ (p_{0}, q_{0}) $满足相容性条件:

这里$ \overline{p}\geq0 $.如果$ \varepsilon\geq0 $适当地小, 则问题(1.3)-(1.4)存在全局解$ (p, q) $满足

$ \bullet $$ \varepsilon >0 $时,

且对所有的$ t>0 $, 存在与$ t $无关的常数$ C>0 $满足

且有$ \|(p-\overline{p})(t)\|_{H^{2}}^2+ \|q(t)\|_{H^{2}}^2\leq \alpha e^{-\beta t}, $其中$ \alpha, \, \, \beta $是与$ t $无关的正常数$ ; $

$ \bullet $$ \varepsilon = 0 $时,

且当$ \overline{p}>0 $时, 对于所有的$ t>0 $, 存在与$ t $无关的常数$ C>0 $满足

且有$ \|(p-\overline{p})(t)\|_{H^{2}}^2+ \|q(t)\|_{H^{2}}^2\leq \alpha e^{-\beta t}, $其中$ \alpha $, $ \beta $是与$ t $无关的正常数.

研究动机.证明问题(1.3)-(1.4)解的全局存在性的主要困难在于:

$ \bullet $在混合边界条件下, 对于方程$ (1.3)_{1} $等号右边的二次非线性项, 通常的方法无效.例如, 在边界条件$ p_x|_{x = 0, x = 1} = q|_{x = 0, x = 1} = 0 $下, 对变换后的方程组$ (1.3) $应用Lyapunov泛函, 得到守恒方程

其中$ \eta(z) = z\ln(z)-z $ (参见文献[15, 32]), 然后借助修正的Sobolev不等式

控制相应的非线性项, 这里$ [p] $$ p $在(0, 1)中的平均值.这一方法主要是基于$ [p] $的守恒.然而在混合边界条件下, $ [p] $不是守恒的, 从而在高阶估计中会出现边界项, 因此需要对这些边界项进行更细致的分析.

$ \bullet $在标准的$ L^{2} $能量框架下, 估计得不到闭合.例如, 因为得不到$ \widetilde{p}_x $的耗散估计, 所以$ \int_{0}^{1}\widetilde{p}\widetilde{p}_xq {\rm d}x $不能根据标准的$ L^{2} $能量估计来处理.因此, 本文将构造一个新的能量框架:根据文献[14]中建立的$ L^{r}\, (r = 3, 4) $估计得到$ L^{2} $估计, 从而$ L^{r} $估计的右边项刚好抵消$ L^{2} $中不能被估计的项$ \int_{0}^{1}\widetilde{p}\widetilde{p}_xq {\rm d}x $.与直接应用$ L^{2} $估计相比, $ L^{r} $估计的推导更为复杂.

$ \bullet $由于缺少关于$ \varepsilon $的一致估计, 在证明非扩散问题($ \varepsilon = 0 $)解的全局存在性的时候, 不能通过取极限$ \varepsilon\rightarrow0 $而由扩散问题($ \varepsilon>0 $)的存在性结论直接得到.

本文其余部分组织如下.第二节在$ \varepsilon>0 $的情形下证明问题(1.3)-(1.4)解的全局存在性和指数收敛估计.第三节致力于$ \varepsilon = 0 $情形下的结果.

记号和定义.本文中$ \|\cdot\|_{L^{2}} $, $ \|\cdot\|_{L^{\infty}} $$ \|\cdot\|_{H^{s}} $分别为Lyapunov泛函下的$ L^{2} $范数, $ L^{\infty} $范数和Hilbert空间中的$ H^{s} $范数. $ L^{\infty}([0, T];H^{s}) $$ L^{2}([0, T];H^{s}) $表示含有时间的函数空间.常数$ C $的值可以根据上下文逐行变化.

2 扩散问题(ε>0)解的全局存在性

本节证明当$ \varepsilon>0 $时问题(1.3)-(1.4)解的全局存在性和指数收敛性.考虑变换后的方程组

$ \begin{equation} \left\{ \begin{array}{lll} p_t = p_{xx} + (pq)_x, \\ q_t = \varepsilon q_{xx} + \varepsilon (q^2)_x + p_x, \quad \varepsilon>0\\ \end{array} \right. \end{equation} $

和初边值条件

$ \begin{equation} \left\{ \begin{array}{lll} (p, q)(x, 0) = (p_0, q_0)(x), \quad p_0(x)\geq0, \quad x\in(0, 1), \\ p|_{x = 0} = \overline{p}, \quad p_x|_{x = 1} = 0, \quad q|_{x = 0, x = 1} = 0, \\ \end{array} \right. \end{equation} $

其中常数$ \overline{p}\geq0 $.本文利用标准的不动点方法得到问题(2.1)-(2.2)解的局部存在性, 建立一些先验估计, 进而得到解的全局存在性.关键步骤是利用一个Lyapunov泛函

证明耗散估计, 其中$ \eta(z) = z\ln z-z $.本节的证明分为两种情形:$ \overline{p}>0 $$ \overline{p} = 0 $.

2.1 $ \overline{p}>0 $的情形

引理2.1  (由Lyapunov泛函导出的能量估计)    假设$ (p, q) $是问题(2.1)-(2.2)的解, 则存在与$ \varepsilon $$ t $无关的正常数$ C $, 使得对所有的$ t>0 $

$ \begin{equation} \|{\widetilde{p}}(t)\|_{L^{2}}^2+\|{q}(t)\|_{L^{2}}^2+\int_{0}^{t}\bigg(\int_{0}^{1}{\frac{(\widetilde{p}_x)^2}{\widetilde{p}+\overline{p}}}{\rm d}x +\|{\widetilde{p}_x}(t)\|_{L^{2}}^2+\varepsilon\|{q_x}\|_{L^{2}}^2\bigg ){\rm d}\tau\leq C, \end{equation} $

其中$ \widetilde{p} = p-\overline{p} $.

  证明分为两个步骤.

步骤1  首先建立$ q $$ (0, 1) $中的$ L^{2} $估计.在方程$ (2.1)_{1} $两边同时乘以$ \ln p-\ln \overline{p} $, 并且在$ (0, 1) $上积分, 得到

$ \begin{equation} \int_{0}^{1}p_t(\ln p-\ln \overline{p}){\rm d}x = \int_{0}^{1}p_{xx}(\ln p-\ln \overline{p}){\rm d}x+ \int_{0}^{1}(pq)_x(\ln p-\ln \overline{p}){\rm d}x. \end{equation} $

$ \eta(z) = z\ln z-z $, 则上式左边

$ \begin{eqnarray} \int_{0}^{1}p_t(\ln p-\ln \overline{p}){\rm d}x& = &\int_{0}^{1}\bigg(\frac{\rm d}{{\rm d}t}(p \ln p )-p_t-p_t\ln \overline{p}\bigg){\rm d}x\\ & = &\int_{0}^{1}\bigg(\frac{\rm d}{{\rm d}t}\big(p(\ln p-\ln \overline{p})\big)-\frac{\rm d}{{\rm d}t}(p-\overline{p})\bigg){\rm d}x\\ & = &\frac{\rm d}{{\rm d}t}\int_{0}^{1}\bigg(p(\ln p-\ln\overline{p})-(p-\overline{p})\bigg){\rm d}x\\ & = &\frac{\rm d}{{\rm d}t}\bigg(\int_{0}^{1}\eta(p)-\eta(\overline{p})-\eta'(\overline{p}) (p-\overline{p}){\rm d}x\bigg). \end{eqnarray} $

根据边界条件(2.2), 对于$ (2.4) $式右边可得

$ \begin{eqnarray} &&\int_{0}^{1}p_{xx}(\ln p-\ln \overline{p}){\rm d}x+\int_{0}^{1}(pq)_x(\ln p-\ln \overline{p}){\rm d}x\\ & = &\int_{0}^{1}\big(p_{xx}+(pq)_x\big)(\ln p-\ln \overline{p}){\rm d}x\\ & = &(\ln p-\ln \overline{p})(p_x+pq)\big|^{1}_{0}-\int_{0}^{1}(p_x+pq)\frac{p_x}{p}{\rm d}x\\ & = &-\int_{0}^{1}\Big (p_x q+\frac{(p_x)^2}{p}\Big ){\rm d}x. \end{eqnarray} $

$ (2.5) $式和$ (2.6) $式代入$ (2.4) $式得

$ \begin{equation} \frac{\rm d}{{\rm d}t}\bigg(\int_{0}^{1}\eta(p)-\eta(\overline{p})-\eta'(\overline{p}) (p-\overline{p}){\rm d}x\bigg )+\int_{0}^{1}\Big (p_x q+\frac{(p_x)^2}{p}\Big ){\rm d}x = 0. \end{equation} $

类似地, 在方程$ (2.1)_2 $两边同时乘以$ q $, 并且在$ (0, 1) $上积分, 则有

$ \begin{equation} \int_{0}^{1}qq_t{\rm d}x = \varepsilon\int_{0}^{1}qq_{xx}{\rm d}x+\varepsilon \int_{0}^{1}q(q^2)_x {\rm d}x+\int_{0}^{1}p_x q{\rm d}x, \end{equation} $

其中, 根据$ q|_{x = 0, x = 1} = 0 $得到

$ \begin{equation} \varepsilon\int_{0}^{1}qq_{xx}{\rm d}x = \varepsilon qq_x\big|^{1}_{0}-\varepsilon\int_{0}^{1}(q_x)^2 {\rm d}x = -\varepsilon\int_{0}^{1}(q_x)^2 {\rm d}x \end{equation} $

$ \begin{equation} \varepsilon\int_{0}^{1}q(q^2)_x{\rm d}x = \frac{2}{3}\varepsilon q^3\big|^{1}_{0} = 0. \end{equation} $

$ (2.9) $式和$ (2.10) $式代入$ (2.8) $式, 可得$ \frac{1}{2}\frac{\rm d}{{\rm d}t}\|{q}\|_{L^{2}}^2-\int_{0}^{1}p_x q{\rm d}x +\varepsilon\|{q_x}\|_{L^{2}}^2 = 0 $.结合$ (2.7) $式, 有

$ \begin{equation} \frac{\rm d}{{\rm d}t}\bigg(\int_{0}^{1}\eta(p)-\eta(\overline{p})-\eta'(\overline{p})(p-\overline{p}){\rm d}x +\frac{1}{2}\|{q}\|_{L^{2}}^2\bigg)+\int_{0}^{1}\frac{(p_x)^2}{p}{\rm d}x+\varepsilon\|{q_x}\|_{L^{2}}^2 = 0. \end{equation} $

上式在$ [0, t] $上积分得到

$ \begin{eqnarray} &&\int_{0}^{1}\eta(p)-\eta(\overline{p})-\eta'(\overline{p})(p-\overline{p}){\rm d}x +\frac{1}{2}\|{q}\|_{L^{2}}^2+\int_{0}^{t}\bigg(\int_{0}^{1}{\frac{(p_x)^2}{p}}{\rm d}x+\varepsilon\|{q_x}\|_{L^{2}}^2\bigg){\rm d}\tau\\ & = &\bigg(\int_{0}^{1}\eta(p_0)-\eta(\overline{p})-\eta'(\overline{p})(p_0-\overline{p}){\rm d}x +\frac{1}{2}\|{q_0}\|_{L^{2}}^2\bigg). \end{eqnarray} $

由于$ \eta(\cdot) $具有凸性, 即$ \frac{\eta(p)-\eta(\overline{p})}{p-\overline{p}}-\eta'(\overline{p})\geq 0 $.又由$ p_{0}\geq0 $和最大值原理可知$ p\geq0 $, 从而有

$ \begin{equation} \int_{0}^{1}\eta(p)-\eta(\overline{p})-\eta'(\overline{p})(p-\overline{p}){\rm d}x\geq0. \end{equation} $

$ 0<\overline{p}<+\infty $和中值定理可得

$ \begin{eqnarray} & & \int_{0}^{1}\eta(p_0)-\eta(\overline{p})-\eta'(\overline{p})(p_0-\overline{p}){\rm d}x\\ & = &\int_{0}^{1}\big(\eta'(\xi_1)-\eta'(\overline{p})\big)(p_0-\overline{p}){\rm d}x = \int_{0}^{1}\eta''(\xi_2)(p_0-\overline{p})(\xi_1-\overline{p}){\rm d}x\\ &\leq &\big|\eta''(\xi_2)\big|\int_{0}^{1}|p_0-\overline{p}||\xi_1-\overline{p}|{\rm d}x \leq C_{0}|p_0-\overline{p}|^2 \leq C_{1}, \end{eqnarray} $

其中常数$ C_{0} $$ C_{1} $只与$ \overline{p} $有关, $ \xi_1 $介于$ p_{0} $$ \overline{p} $之间, $ \xi_2 $介于$ \xi_1 $$ \overline{p} $之间.联合(2.12)-(2.14)式可得

$ \begin{equation} \frac{1}{2}\|{q(t)}\|_{L^{2}}^2+\int_{0}^{t}\bigg(\int_{0}^{1}{\frac{(p_x)^2}{p}}{\rm d}x +\varepsilon\|{q_x}\|_{L^{2}}^2\bigg){\rm d}\tau \leq C_{2}, \end{equation} $

其中常数$ C_{2} $只与$ \overline{p} $有关.

步骤2  下面建立$ \widetilde{p} $$ (0, 1) $上的$ L^{2} $估计.由于$ \widetilde{p} = p-\overline{p} $, 将$ \widetilde{p} $代入方程组(2.1)中可得

$ \begin{equation} \left\{ \begin{array}{lll} \widetilde{p}_t-(\widetilde{p}q)_x-\overline{p}q_x = \widetilde{p}_{xx}, \\ q_t-\widetilde{p}_x = \varepsilon q_{xx} + \varepsilon (q^2)_x. \end{array} \right. \end{equation} $

方程$ (2.16)_{1} $$ \widetilde{p} $$ L^{2} $内积, 方程$ (2.16)_{2} $$ \overline{p}\, q $$ L^{2} $内积, 相加得到

$ \begin{eqnarray} \frac{\rm d}{{\rm d}t}\Big(\frac{1}{2}\|{\widetilde{p}}\|_{L^{2}}^2+\frac{\overline{p}}{2}\|{q}\|_{L^{2}}^2\Big) & = &\int_{0}^{1}\widetilde{p}\widetilde{p}_{xx}{\rm d}x+ \varepsilon\overline{p}\int_{0}^{1}qq_{xx}{\rm d}x+ \int_{0}^{1}\widetilde{p}(\widetilde{p}q)_x{\rm d}x\\ &&+\overline{p}\int_{0}^{1}\widetilde{p}q_x{\rm d}x+\overline{p}\int_{0}^{1} q\widetilde{p}_x{\rm d}x+\varepsilon\overline{p}\int_{0}^{1}q(q^2)_x{\rm d}x. \end{eqnarray} $

下面逐个估计$ (2.17) $式等号右边的各项积分.对于第一项积分, 由边界条件$ (1.4) $$ \widetilde{p} = p-\overline{p} $$ \widetilde{p}|_{x = 0} = 0 $, $ \widetilde{p}_x|_{x = 1} = 0 $, 从而应用分部积分可得

$ \begin{equation} \int_{0}^{1}\widetilde{p}\widetilde{p}_{xx}{\rm d}x = \widetilde{p}\widetilde{p}_{x}\big|^{1}_{0}-\|{\widetilde{p}_x}\|_{L^{2}}^2 = -\|{\widetilde{p}_x}\|_{L^{2}}^2. \end{equation} $

对于第二项和第三项, 根据$ q|_{x = 0, \, \, x = 1} = 0 $可知

$ \begin{equation} \varepsilon\overline{p}\int_{0}^{1}qq_{xx}{\rm d}x = \varepsilon\overline{p}qq_{x}\big|^{1}_{0}-\varepsilon\overline{p}\|{q_x}\|_{L^{2}}^2 = -\varepsilon\overline{p}\|{q_x}\|_{L^{2}}^2 \end{equation} $

$ \begin{equation} \int_{0}^{1}\widetilde{p}(\widetilde{p}q)_x{\rm d}x = {\widetilde{p}}^2q \big|^{1}_{0}-\int_{0}^{1}\widetilde{p}\widetilde{p}_x q {\rm d}x = -\int_{0}^{1}\widetilde{p}\widetilde{p}_x q {\rm d}x. \end{equation} $

对于第四和第五项显然有

$ \begin{equation} \overline{p}\int_{0}^{1}\widetilde{p}q_x{\rm d}x+\overline{p}\int_{0}^{1} \widetilde{p}_xq{\rm d}x = \overline{p}(\widetilde{p}q)\big|^{1}_{0} = 0. \end{equation} $

对于最后一项有

$ \begin{equation} \varepsilon\overline{p}\int_{0}^{1}q(q^2)_x{\rm d}x = \frac{2}{3}\varepsilon\overline{p}q^3\big|^{1}_{0} = 0. \end{equation} $

将(2.18)-(2.22)式代入(2.17)式可得

$ \begin{equation} \frac{\rm d}{{\rm d}t}\Big(\frac{1}{2}\|{\widetilde{p}}\|_{L^{2}}^2+\frac{\overline{p}}{2}\|{q}\|_{L^{2}}^2\Big)+ \|{\widetilde{p}_x}\|_{L^{2}}^2+\varepsilon\overline{p}\|{q_x}\|_{L^{2}}^2 = -\int_{0}^{1}\widetilde{p}\widetilde{p}_xq {\rm d}x. \end{equation} $

下面估计$ (2.23) $式等式右边的积分, 由于$ L^{2} $方法无法处理.因此, 回到方程组$ (2.16) $, 在方程$ (2.16)_{1} $两边同时乘以$ \widetilde{p}^{2} $, 在$ (0, 1) $上积分, 得到

$ \begin{equation} \int_{0}^{1}\widetilde{p}^{2}\widetilde{p}_t {\rm d}x-\int_{0}^{1}\widetilde{p}^{2}(\widetilde{p}q)_x {\rm d}x-\overline{p}\int_{0}^{1}\widetilde{p}^{2}q_x {\rm d}x = \int_{0}^{1}\widetilde{p}^{2}\widetilde{p}_{xx} {\rm d}x. \end{equation} $

利用分部积分和边界条件有

$ \begin{equation} -\int_{0}^{1}\widetilde{p}^{2}(\widetilde{p}q)_x {\rm d}x = -\widetilde{p}^{3}q\big|^{1}_{0}+2\int_{0}^{1}\widetilde{p}^{2}q \widetilde{p}_x {\rm d}x = 2\int_{0}^{1}\widetilde{p}^{2}q\widetilde{p}_x {\rm d}x \end{equation} $

$ \begin{equation} -\overline{p}\int_{0}^{1}\widetilde{p}^{2}q_x {\rm d}x = -\overline{p}\widetilde{p}^{2}q\big|^{1}_{0}+2\overline{p}\int_{0}^{1}\widetilde{p}q\widetilde{p}_x {\rm d}x = 2\overline{p}\int_{0}^{1}\widetilde{p}q\widetilde{p}_x {\rm d}x. \end{equation} $

类似地, 有

$ \begin{equation} \int_{0}^{1}\widetilde{p}^{2}\widetilde{p}_{xx} {\rm d}x = \widetilde{p}^{2}\widetilde{p}_x\big|^{1}_{0} -2\int_{0}^{1}\widetilde{p}(\widetilde{p}_x)^2 {\rm d}x = -2\int_{0}^{1}\widetilde{p}(\widetilde{p}_x)^{2}{\rm d}x. \end{equation} $

将(2.25)-(2.27)式代入$ (2.24) $式得到

$ \begin{equation} \frac{\rm d}{{\rm d}t}\bigg(\frac{1}{3}\int_{0}^{1}\widetilde{p}^{3}{\rm d}x\bigg) +2\int_{0}^{1}\widetilde{p}(\widetilde{p}_x)^{2}{\rm d}x = -2\int_{0}^{1}\widetilde{p}^{2}q\widetilde{p}_x {\rm d}x -2\overline{p}\int_{0}^{1}\widetilde{p}q\widetilde{p}_x {\rm d}x. \end{equation} $

类似地, 方程$ (2.16)_{1} $两边同时乘以$ \widetilde{p}^{3} $, 在$ (0, 1) $上积分, 计算可得

$ \begin{eqnarray} \frac{\rm d}{{\rm d}t}\bigg(\frac{1}{4}\|\widetilde{p}\|_{L^4}^4\bigg)+ 3\|{\widetilde{p}\widetilde{p}_x}\|_{L^{2}}^2 = -3\int_{0}^{1}\widetilde{p}^{3}q\widetilde{p}_x {\rm d}x-3\overline{p}\int_{0}^{1}\widetilde{p}^2q\widetilde{p}_x {\rm d}x. \end{eqnarray} $

$ (2.23) $式两边同时乘以$ 2\overline{p} $, 和(2.28)式相减得到

$ \begin{eqnarray} && \frac{\rm d}{{\rm d}t}\bigg(\overline{p}\|{\widetilde{p}}\|_{L^{2}}^2+\overline{p}^2\|{q}\|_{L^{2}}^2 -\frac{1}{3}\int_{0}^{1}\widetilde{p}^3{\rm d}x\bigg)+2\overline{p}\|{\widetilde{p}_x}\|_{L^{2}}^2 -2\int_{0}^{1}\widetilde{p}(\widetilde{p}_x)^2{\rm d}x+ 2\varepsilon\overline{p}^2\|{q_x}\|_{L^{2}}^2\\ & = &2\int_{0}^{1}\widetilde{p}^{2}q\widetilde{p}_x {\rm d}x. \end{eqnarray} $

$ (2.30) $式两边同时乘以$ \frac{3}{2}\overline{p} $, 和$ (2.29) $式相加得到

$ \begin{eqnarray} &&\frac{\rm d}{{\rm d}t}\bigg(\frac{3}{2}\overline{p}^2\|{\widetilde{p}}\|_{L^{2}}^2 +\frac{3}{2}\overline{p}^3\|q\|_{L^{2}}^2 -\frac{\overline{p}}{2}\int_{0}^{1}\widetilde{p}^3{\rm d}x +\frac{1}{4}\|\widetilde{p}\|_{L^4}^4\bigg)\\ &&+3\overline{p}^2\|{\widetilde{p}_x}\|_{L^{2}}^2 +3\|\widetilde{p}\widetilde{p}_x\|_{L^{2}}^2 -3\overline{p}\int_{0}^{1}\widetilde{p}(\widetilde{p}_x)^2{\rm d}x +3\varepsilon\overline{p}^3\|q_x\|_{L^{2}}^2\\ & = &-3\int_{0}^{1}\widetilde{p}^{3}q\widetilde{p}_x {\rm d}x. \end{eqnarray} $

$ \begin{eqnarray} G_1(t) & = &\frac{3}{2}\overline{p}^2\|{\widetilde{p}}\|_{L^{2}}^2 +\frac{3}{2}\overline{p}^3\|q\|_{L^{2}}^2 -\frac{\overline{p}}{2}\int_{0}^{1}\widetilde{p}^3{\rm d}x +\frac{1}{4}\|\widetilde{p}\|_{L^4}^4\\ & = &\overline{p}^2\|{\widetilde{p}}\|_{L^{2}}^2+ \frac{1}{8}\|2\overline{p}{\widetilde{p}-\widetilde{p}^2}\|_{L^{2}}^2 +\frac{1}{8}\|\widetilde{p}\|_{L^4}^4 +\frac{3}{2}\overline{p}^3\|q\|_{L^{2}}^2, \\ K_1(t) & = &3\overline{p}^2\|{\widetilde{p}_x}\|_{L^{2}}^2 +3\|\widetilde{p}\widetilde{p}_x\|_{L^{2}}^2 -3\overline{p}\int_{0}^{1}\widetilde{p}(\widetilde{p}_x)^2{\rm d}x +3\varepsilon\overline{p}^3\|q_x\|_{L^{2}}^2\\ & = &\frac{3}{2}\overline{p}^2\|\widetilde{p}_x\|_{L^{2}}^2 +\frac{3}{2}\|\overline{p}{\widetilde{p}_x-\widetilde{p}\widetilde{p}_x}\|_{L^{2}}^2 +\frac{3}{2}\|\widetilde{p}\widetilde{p}_x\|_{L^{2}}^2 +3\varepsilon\overline{p}^3\|q_x\|_{L^{2}}^2. \end{eqnarray} $

则由$ (2.31) $式有

$ \begin{equation} \frac{\rm d}{{\rm d}t}G_1(t)+K_1(t) = -3\int_{0}^{1}\widetilde{p}^{3}q\widetilde{p}_x {\rm d}x. \end{equation} $

下面估计上式右边的项.由Young不等式和$ (2.15) $式计算得到

$ \begin{eqnarray} \bigg |-3\int_{0}^{1}\widetilde{p}^{3}q\widetilde{p}_x {\rm d}x \bigg| &\leq&\frac{1}{2}\|\widetilde{p}\widetilde{p}_x\|_{L^{2}}^2 +\frac{9}{2}\|\widetilde{p}^2 q\|_{L^{2}}^2 \\ &\leq&\frac{1}{2}\|\widetilde{p}\widetilde{p}_x\|_{L^{2}}^2 +\frac{9}{2}\|\widetilde{p}\|_{L^{\infty}}^4\|q\|_{L^{2}}^2\\ &\leq&\frac{1}{2}\|\widetilde{p}\widetilde{p}_x\|_{L^{2}}^2 +C_{3}\|\widetilde{p}\|_{L^{\infty}}^4, \end{eqnarray} $

其中常数$ C_{3} $只与$ \overline{p} $有关.由上式和$ (2.33) $式可知

$ \begin{eqnarray} \frac{\rm d}{{\rm d}t}G_1(t)+\frac{3}{2}\overline{p}^2\|{\widetilde{p}_x}\|_{L^{2}}^2 +\frac{3}{2}\|\overline{p}{\widetilde{p}_x-\widetilde{p}\widetilde{p}_x}\|_{L^{2}}^2 +\|\widetilde{p}\widetilde{p}_x\|_{L^{2}}^2 +3\varepsilon\overline{p}^3\|q_x\|_{L^{2}}^2 \leq C_{4}\|\widetilde{p}\|_{L^{\infty}}^4. \end{eqnarray} $

下面对$ \|\widetilde{p}\|_{L^{\infty}} $进行估计.因为

应用Young不等式, 得到

从而有

$ \begin{equation} \|\widetilde{p}\|_{L^{\infty}}^4 \leq\big(8\|\widetilde{p}\|_{L^{4}}^4+(8+16\overline{p})\|\widetilde{p}\|_{L^{2}}^2\big) \int_{0}^{1}\frac{(\widetilde{p}_x)^2}{\widetilde{p}+\overline{p}}{\rm d}x. \end{equation} $

$ G_1(t) $的表达式可知, 存在常数$ C_{5}>0 $使得

$ \begin{equation} 8\|\widetilde{p}\|_{L^{4}}^4+(8+16\overline{p})\|\widetilde{p}\|_{L^{2}}^2 \leq C_{5}G_1(t). \end{equation} $

联合(2.35)-(2.37)式可得

$ \begin{equation} \frac{\rm d}{{\rm d}t}G_1(t)+\frac{3}{2}\overline{p}^2\|{\widetilde{p}_x}\|_{L^{2}}^2 +\frac{3}{2}\|\overline{p}{\widetilde{p}_x-\widetilde{p}\widetilde{p}_x}\|_{L^{2}}^2 +\|\widetilde{p}\widetilde{p}_x\|_{L^{2}}^2 +3\varepsilon\overline{p}^3\|q_x\|_{L^{2}}^2 \leq C_{6}G_1(t)\int_{0}^{1}\frac{(\widetilde{p}_x)^2}{\widetilde{p}+\overline{p}}{\rm d}x, \end{equation} $

因此有

$ \begin{equation} \frac{\rm d}{{\rm d}t}G_1(t) \leq C_{6}G_1(t)\int_{0}^{1}\frac{(\widetilde{p}_x)^2}{\widetilde{p}+\overline{p}}{\rm d}x. \end{equation} $

由Gronwall不等式和(2.15)式可得, 对任意的$ t\geq0 $

$ \begin{eqnarray} G_1(t)\leq G_1(0)\exp\Big({C_{5}\int_{0}^{t}\int_{0}^{1}{\frac{(\widetilde{p}_x)^2}{\widetilde{p}+\overline{p}}}{\rm d}x}\Big)\leq C_{7}. \end{eqnarray} $

将(2.40)式代入(2.38)式, 并在$ [0, t] $上积分, 计算可得

$ \begin{equation} G_1(t)+\int_{0}^{t}\bigg(\frac{3}{2}\overline{p}^2\|{\widetilde{p}_x(\tau)}\|_{L^{2}}^2 +\frac{3}{2}\|(\overline{p}{\widetilde{p}_x-\widetilde{p}\widetilde{p}_x})(\tau)\|_{L^{2}}^2 +\|\widetilde{p}\widetilde{p}_x(\tau)\|_{L^{2}}^2 +3\varepsilon\overline{p}^3\|q_x(\tau)\|_{L^{2}}^2\bigg){\rm d}\tau \leq C_{8}, \end{equation} $

其中$ C_{8} $是正常数.注意到$ C_{3}, \cdots , C_{8} $都是与$ t $$ \varepsilon $无关的常数.联合$ (2.15) $式, $ (2.32) $式和$ (2.41) $式可得(2.3)式.

下面根据标准的$ L^{2} $能量方法给出$ (p, q) $的二阶导数估计.

引理2.2  设$ (p, q) $是问题(2.1)-(2.2)的解, 则对任意的$ \varepsilon>0 $有估计

$ \begin{equation} \|(\widetilde{p}_x, q_x)(t)\|_{H^{1}}^2+\int_{0}^{t}\|\widetilde{p}_{xx}(\tau)\|_{H^{1}}^2 +\|q_{xx}(\tau)\|_{H^{1}}^2{\rm d}\tau\leq C, \end{equation} $

其中, 常数$ C $$ t $无关.

  在方程$ (2.16)_{1} $两边同时乘以$ \widetilde{p}_{xx} $, 在方程$ (2.16)_{2} $两边同时乘以$ q_{xx} $, 并且在(0, 1)上积分, 得到

$ \begin{eqnarray} &&\int_{0}^{1}\widetilde{p}_t\widetilde{p}_{xx}{\rm d}x+\int_{0}^{1}q_t q_{xx}{\rm d}x-\int_{0}^{1}(\widetilde{p}q)_x\widetilde{p}_{xx}{\rm d}x-\overline{p}\int_{0}^{1}\widetilde{p}_{xx}q_x{\rm d}x-\int_{0}^{1}\widetilde{p}_xq_{xx}{\rm d}x\\ & = &\int_{0}^{1}(\widetilde{p}_{xx})^2{\rm d}x+\varepsilon\int_{0}^{1}(q_{xx})^2{\rm d}x+\varepsilon\int_{0}^{1}(q^2)_xq_{xx}{\rm d}x. \end{eqnarray} $

根据Gagliardo-Nirenberg不等式和嵌入不等式有$ \|\widetilde{p}\|_{L^{\infty}}^2\leq C\|\widetilde{p}\|_{L^{2}}\|\widetilde{p}_x\|_{L^{2}}+C\|\widetilde{p}\|_{L^{2}} $$ \|q\|_{L^{\infty}}^2\leq C\|q\|_{L^{2}}\|q_x\|_{L^{2}} $.因为$ \widetilde{p}|_{x = 0} = 0 $, $ x\in(0, 1) $, 从而有

$ \begin{eqnarray} \int_{0}^{1}|\widetilde{p}(x)|^2{\rm d}x & = &\int_{0}^{1}\bigg|\int_{0}^{x}\widetilde{p}'(y){\rm d}y\bigg|^2{\rm d}x \leq\int_{0}^{1}\int_{0}^{x}|\widetilde{p}'(y)|^2{\rm d}y{\rm d}x \\ & \leq&\int_{0}^{1}\int_{0}^{1}|\widetilde{p}'(y)|^2{\rm d}y{\rm d}x \leq\int_{0}^{1}|\widetilde{p}'(y)|^2{\rm d}x \leq\int_{0}^{1}|\widetilde{p}_x|^2{\rm d}x, \end{eqnarray} $

$ \|\widetilde{p}\|_{L^{2}}^2\leq \|\widetilde{p}_x\|_{L^{2}}^2 $.应用Hölder不等式, $ (2.3) $式和$ (2.43) $式, 可得

$ \begin{eqnarray} &&\frac{1}{2}\frac{\rm d}{{\rm d}t}(\|\widetilde{p}_x\|_{L^{2}}^2+\|q_x\|_{L^{2}}^2)+\|\widetilde{p}_{xx}\|_{L^{2}}^2+ \varepsilon\|q_{xx}\|_{L^{2}}^2\\ & = &-\int_{0}^{1}\big((\widetilde{p}q)_x+\overline{p}q_x\big)\widetilde{p}_{xx}{\rm d}x -\int_{0}^{1}\big(\widetilde{p}_x+\varepsilon(q^2)_x\big)q_{xx}{\rm d}x\\ &\leq&\frac{1}{2}\|\widetilde{p}_{xx}\|_{L^{2}}^2+\frac{\varepsilon}{2}\|q_{xx}\|_{L^{2}}^2+ C_{9}(\|\widetilde{p}\|_{L^{\infty}}^2\|q_x\|_{L^{2}}^2+ \|q\|_{L^{\infty}}^2\|\widetilde{p}_x\|_{L^{2}}^2 \\ &&+\|q\|_{L^{\infty}}^2\|q_x\|_{L^{2}}^2+\|\widetilde{p}_x\|_{L^{2}}^2+ \|q_x\|_{L^{2}}^2)\\ &\leq&\frac{1}{2}\|\widetilde{p}_{xx}\|_{L^{2}}^2+\frac{\varepsilon}{2}\|q_{xx}\|_{L^{2}}^2+ C_{10}(\|\widetilde{p}\|_{L^{2}}\|\widetilde{p}_x\|_{L^{2}}\|q_x\|_{L^{2}}^2+\|q\|_{L^{2}}\|q_x\|_{L^{2}}\|\widetilde{p}_x\|_{L^{2}}^2\\ &&+\|q\|_{L^{2}}\|q_x\|_{L^{2}}\|q_x\|_{L^{2}}^2+\|\widetilde{p}_x\|_{L^{2}}^2+ \|q_x\|_{L^{2}}^2+\|\widetilde{p}\|_{L^{2}}^2\|q_x\|_{L^{2}}^2)\\ &\leq&\frac{1}{2}\|\widetilde{p}_{xx}\|_{L^{2}}^2+\frac{\varepsilon}{2}\|q_{xx}\|_{L^{2}}^2+ C_{11}(\|\widetilde{p}_x\|_{L^{2}}^2 +\|q_x\|_{L^{2}}^2)(\|q_x\|_{L^{2}}^2+1), \end{eqnarray} $

于是

$ \begin{eqnarray} &&\frac{1}{2}\frac{\rm d}{{\rm d}t}(\|\widetilde{p}_x\|_{L^{2}}^2+\|q_x\|_{L^{2}}^2+1)+\frac{1}{2}\|\widetilde{p}_{xx}\|_{L^{2}}^2+ \frac{\varepsilon}{2}\|q_{xx}\|_{L^{2}}^2\\ &\leq& C_{11}(\|\widetilde{p}_x\|_{L^{2}}^2+\|q_x\|_{L^{2}}^2) (\|\widetilde{p}_x\|_{L^{2}}^2+\|q_x\|_{L^{2}}^2+1). \end{eqnarray} $

由Gronwall不等式和$ (2.3) $式可得

$ \begin{equation} \|\widetilde{p}_x(t)\|_{L^{2}}^2+\|q_x(t)\|_{L^{2}}^2+\int_{0}^{t}(\|\widetilde{p}_{xx}(\tau)\|_{L^{2}}^2+ \|q_{xx}(\tau)\|_{L^{2}}^2){\rm d}\tau\leq C_{12}. \end{equation} $

由于在边界点解的高阶导数函数值未知, 因此需要交换时间导数和空间导数.在方程组$ (2.16) $两边关于$ t $求导, 然后分别乘以$ \widetilde{p}_{t} $$ q_{t} $, 并在区间$ (0, 1) $上积分, 可得

$ \begin{eqnarray} &&\int_{0}^{1}\partial_{t}\widetilde{p}_t\widetilde{p}_t{\rm d}x+\overline{p}\int_{0}^{1}\partial_{t}q_tq_t{\rm d}x -\int_{0}^{1}\partial_{t}(\widetilde{p}q)_x\widetilde{p}_{t}{\rm d}x -\overline{p}\int_{0}^{1}\partial_{t}q_x\widetilde{p}_t{\rm d}x-\overline{p}\int_{0}^{1}\partial_{t}\widetilde{p}_xq_t{\rm d}x\\ & = &\int_{0}^{1}\partial_{t}\widetilde{p}_{xx}\widetilde{p}_t{\rm d}x+\varepsilon\overline{p}\int_{0}^{1}\partial_{t}q_{xx}q_t{\rm d}x+\varepsilon\overline{p}\int_{0}^{1}\partial_{t}(q^2)_xq_t{\rm d}x, \end{eqnarray} $

从而有

$ \begin{eqnarray} &&\frac{1}{2}\frac{\rm d}{{\rm d}t}(\|\widetilde{p}_t\|_{L^{2}}^2+\overline{p}\|q_t\|_{L^{2}}^2)+\|\widetilde{p}_{xt}\|_{L^{2}}^2+\varepsilon\overline{p}\|q_{xt}\|_{L^{2}}^2\\ & = &\int_{0}^{1}\partial_{t}(\widetilde{p}q)_x\widetilde{p}_t{\rm d}x+ \varepsilon\overline{p}\int_{0}^{1}\partial_{t}(q^2)_xq_t{\rm d}x = -\int_{0}^{1}(\widetilde{p}q)_t\widetilde{p}_{xt}{\rm d}x- \varepsilon\overline{p}\int_{0}^{1}(q^2)_tq_{xt}{\rm d}x\\ &\leq&\frac{1}{2}\|\widetilde{p}_{xt}\|_{L^{2}}^2+ \frac{1}{2}\varepsilon\overline{p}\|q_{xt}\|_{L^{2}}^2+ C_{13}\big(\|\widetilde{p}\|_{L^{\infty}}^2\|q_t\|_{L^{2}}^2+\|q\|_{L^{\infty}}^2(\|\widetilde{p}_t\|_{L^{2}}^2+\|q_t\|_{L^{2}}^2)\big). \end{eqnarray} $

由Gagliardo-Nirenberg不等式和$ (2.44) $式可得$ \|\widetilde{p}\|_{L^{\infty}}^2\leq C\|\widetilde{p}_x\|_{{L^{2}}}^2 $.同理, 由$ q|_{x = 0, x = 1} = 0 $, Gagliardo-Nirenberg不等式和嵌入不等式可得$ \|q\|_{L^{\infty}}^2\leq C\|q_x\|_{L^{2}}^2 $.从而由$ (2.49) $式得

$ \begin{eqnarray} &&\frac{1}{2}\frac{\rm d}{{\rm d}t}(\|\widetilde{p}_t\|_{L^{2}}^2+\overline{p}\|q_t\|_{L^{2}}^2)+ \frac{1}{2}\|\widetilde{p}_{xt}\|_{L^{2}}^2+\frac{1}{2}\varepsilon\overline{p}\|q_{xt}\|_{L^{2}}^2\\ &\leq& C_{14}(\|\widetilde{p}_x\|_{L^{2}}^2+\|q_x\|_{L^{2}}^2)(\|\widetilde{p}_t\|_{L^{2}}^2+\overline{p}\|q_t\|_{L^{2}}^2). \end{eqnarray} $

由Gronwall不等式可以导出

$ \begin{equation} \|\widetilde{p}_t\|_{L^{2}}^2+\|q_t\|_{L^{2}}^2+\int_{0}^{t}(\|\widetilde{p}_{xt}(\tau)\|_{L^{2}}^2 +\|q_{xt}(\tau)\|_{L^{2}}^2){\rm d}\tau \leq C_{15}. \end{equation} $

对方程组$ (2.16) $$ L^{2} $内积, 根据$ (2.47) $式和$ (2.51) $式可得

$ \begin{eqnarray} \|\widetilde{p}_{xx}\|_{L^{2}}^2 & = &\|\widetilde{p}_t-(\widetilde{p}q)_x-\overline{p}q_x\|_{L^{2}}^2\\ &\leq &C_{16}\big(|\widetilde{p}_t\|_{L^{2}}^2+\|\widetilde{p}_x\|_{L^{2}}^2\|q\|_{L^{\infty}}^{2}+ \|\widetilde{p}\|_{L^{\infty}}^{2}\|q_x\|_{L^{2}}^2+\overline{p}^2\|q_x\|_{L^{2}}^2\big)\\ &\leq& C_{17} \end{eqnarray} $

$ \begin{equation} \|q_{xx}\|_{L^{2}}^2 \leq C_{18}\big(\frac{1}{\varepsilon^2}\|q_t\|_{L^{2}}^2+ \frac{1}{\varepsilon^2}\|\widetilde{p}_x\|_{L^{2}}^2+ \|q\|_{L^{\infty}}^{2}\|q_x\|_{L^{2}}^2\big) \leq C_{19}. \end{equation} $

联合$ (2.3) $式, $ (2.47) $式, $ (2.51) $式, $ (2.52) $式和$ (2.53) $式, 得到

$ \begin{equation} \|\widetilde{p}_{xx}(t)\|_{L^{2}}^2+\|q_{xx}(t)\|_{L^{2}}^2+\int_{0}^{t}(\|\widetilde{p}_{xxx}(\tau)\|_{L^{2}}^2 +\|q_{xxx}(\tau)\|_{L^{2}}^2){\rm d}\tau\leq C_{20}, \end{equation} $

其中$ C_{20} $是正常数.注意到$ C_{9}, \cdots , C_{20} $都是与$ t $无关但与$ \varepsilon $有关的常数.由(2.54)式知引理2.2成立.

下面证明问题(2.1)-(2.2)解的指数收敛性.

引理2.3  (指数收敛性)  设$ (p, q) $是问题(2.1)-(2.2)的解, 则对任意的$ \varepsilon>0 $, 存在与$ t $无关的常数$ \alpha $, $ \beta $满足

$ \begin{equation} \|(\widetilde{p}, q)(t)\|_{H^{2}}^2\leq\alpha e^{-\beta t}. \end{equation} $

  由(2.38)-(2.40)式和(2.46)-(2.51)式可得

$ \begin{equation} \frac{\rm d}{{\rm d}t}G_1(t)+\frac{3}{2}\overline{p}^2\|{\widetilde{p}_x}\|_{L^{2}}^2 +\frac{3}{2}\|\overline{p}{\widetilde{p}_x-\widetilde{p}\widetilde{p}_x}\|_{L^{2}}^2 +\|\widetilde{p}\widetilde{p}_x\|_{L^{2}}^2 +3\varepsilon\overline{p}^3\|q_x\|_{L^{2}}^2 \leq C_{19}\int_{0}^{1}\frac{(\widetilde{p}_x)^2}{\widetilde{p}+\overline{p}}{\rm d}x, \end{equation} $

$ \begin{equation} \frac{1}{2}\frac{\rm d}{{\rm d}t}(\|\widetilde{p}_x\|_{L^{2}}^2+\|q_x\|_{L^{2}}^2)+\frac{1}{2}\|\widetilde{p}_{xx}\|_{L^{2}}^2+ \frac{\varepsilon}{2}\|q_{xx}\|_{L^{2}}^2 \leq C_{20}(\|\widetilde{p}_x\|_{L^{2}}^2+\|q_x\|_{L^{2}}^2) \end{equation} $

$ \begin{equation} \frac{1}{2}\frac{\rm d}{{\rm d}t}(\|\widetilde{p}_t\|_{L^{2}}^2+\overline{p}\|q_t\|_{L^{2}}^2)+ \frac{1}{2}\|\widetilde{p}_{xt}\|_{L^{2}}^2+\frac{\varepsilon\overline{p}}{2}\|q_{xt}\|_{L^{2}}^2 \leq C_{21}(\|\widetilde{p}_x\|_{L^{2}}^2+\|q_x\|_{L^{2}}^2). \end{equation} $

比较上述不等式的系数可知, 存在适当大的正常数$ B_{1} $满足

$ (2.56) $式两边同时乘以$ B_{1} $, 与(2.57)式和(2.58)式相加, 得到

$ \begin{equation} \frac{\rm d}{{\rm d}t}M(t)+N(t)\leq B_{1}C_{19}\int_{0}^{1}\frac{(\widetilde{p}_x)^2}{\widetilde{p}+\overline{p}}{\rm d}x, \end{equation} $

其中

$ \begin{eqnarray} M(t)& = &B_{1}G_1(t)+\frac{1}{2}(\|\widetilde{p}_x\|_{L^{2}}^2+\|q_x\|_{L^{2}}^2+ \|\widetilde{p}_t\|_{L^{2}}^2+\overline{p}\|q_t\|_{L^{2}}^2), \\ N(t)& = &B_{1}\bigg(\frac{3}{2}\|\overline{p}{\widetilde{p}_x-\widetilde{p}\widetilde{p}_x}\|_{L^{2}}^2 +\|\widetilde{p}\widetilde{p}_x\|_{L^{2}}^2 +3\varepsilon\overline{p}^3\|q_x\|_{L^{2}}^2\bigg)+\frac{1}{2}\|\widetilde{p}_{xx}\|_{L^{2}}^2+\frac{\varepsilon}{2}\|q_{xx}\|_{L^{2}}^2\\ &&+\frac{1}{2}\|\widetilde{p}_{xt}\|_{L^{2}}^2+ \frac{\varepsilon\overline{p}}{2}\|q_{xt}\|_{L^{2}}^2-(C_{20}+C_{21}) (\|\widetilde{p}_x\|_{L^{2}}^2+\|q_x\|_{L^{2}}^2). \end{eqnarray} $

同理, 存在正常数$ B_{2} $满足$ B_{2}\geq B_{1}C_{19}+1. $$ (2.13) $式两边同时乘以$ B_{2} $, 与$ (2.59) $式相加, 得到

$ \begin{equation} \frac{\rm d}{{\rm d}t}\bigg(B_{2}E(p, \overline{p})+\frac{B_{2}}{2}\|q\|_{L^{2}}^2+M(t)\bigg)+(B_{2}-B_{1}C_{19})\int_{0}^{1}\frac{(\widetilde{p}_x)^2} {\widetilde{p}+\overline{p}}{\rm d}x+B_{2}\varepsilon\|q_x\|_{L^{2}}^2+N(t) \leq0, \end{equation} $

其中

则由$ (2.61) $式有

根据$ X(t) $$ Y(t) $的定义, 不等式

$ (2.44) $式和Gagliardo-Nirenberg不等式可知, 存在一个与$ t $无关的常数$ C_{22} $, 使得

因此有$ \frac{\rm d}{{\rm d}t}X(t)+C_{23}X(t)\leq0 $.结合$ X(t) $的表达式可知, 存在与$ t $无关的常数$ \alpha, \beta $满足

$ \begin{equation} \|\widetilde{p}(t)\|_{H^{2}}^2+\|q(t)\|_{H^{2}}^2 \leq \alpha e^{-\beta t}. \end{equation} $

引理2.3得证.

2.2 $ \overline{p} = 0 $的情形

$ \overline{p} = 0 $时, 根据$ p|_{x = 0} = 0 $, $ p_0>0 $, 局部存在性结果和最大值原理可知$ p>0 $.$ \widetilde{p} = p+R $, 应用扰动方法将问题(2.1)-(2.2)变换为如下初边值问题

$ \begin{equation} \left\{ \begin{array}{lll} \widetilde{p}_t-(\widetilde{p}q)_x+Rq_x = \widetilde{p}_{xx}, \\ q_t-\widetilde{p}_x = \varepsilon q_{xx} + \varepsilon (q^2)_x, \\ \widetilde{p}|_{x = 0} = R, \quad\widetilde{p}_x|_{x = 1} = 0, \\ q|_{x = 0, x = 1} = 0, \\ (p, q)(x, 0) = (p_0, q_0)(x), \end{array} \right. \end{equation} $

其中$ R>0 $是一个常数.

引理2.4  设$ (p, q) $是问题(2.1)-(2.2)的解, 则对任意的$ \varepsilon>0 $有估计

$ \begin{equation} \|(p, q)(t)\|_{H^{2}}^2+\int_{0}^{t}\|p(\tau)\|_{H^{3}}^2 +\|q(\tau)\|_{H^{3}}^2{\rm d}\tau\leq C, \end{equation} $

其中常数$ C $$ t $无关.

  证明分为五个步骤.

步骤1  设$ R $为待确定的正参数.在方程$ (2.63)_1 $两边同时乘以检验函数$ \ln{\widetilde{p}}-\ln{R} $, 在方程$ (2.63)_2 $两边同时乘以$ q $, 并在区间$ (0, 1) $上积分, 可得

因此有

$ \begin{equation} \frac{\rm d}{{\rm d}t}\bigg(\int_{0}^{1}\eta(\widetilde{p})-\eta(R)-\eta'(R) (\widetilde{p}-R){\rm d}x+\frac{1}{2}\|q\|_{L^{2}}^2\bigg)+\int_{0}^{1}\frac{(\widetilde{p}_x)^2}{\widetilde{p}}{\rm d}x+\varepsilon\|q_x\|_{L^{2}}^2 = R\int_{0}^{1}q\frac{\widetilde{p}_x}{\widetilde{p}}{\rm d}x, \end{equation} $

其中$ \eta(z) = z\ln(z)-z $.考虑(2.65)式的右边项, 根据$ \widetilde{p}\geq R $和不等式$ \|q\|_{L^{2}}^2\leq C\|q_x\|_{L^{2}}^2 $, 可以计算

$ R = \varepsilon $, 则由上式和$ (2.65) $式有

$ \begin{equation} \frac{\rm d}{{\rm d}t}\bigg(\int_{0}^{1}\eta(\widetilde{p})-\eta(R)-\eta'(R) (\widetilde{p}-R){\rm d}x+\frac{1}{2}\|q\|_{L^{2}}^2\bigg ) +\frac{1}{2}\int_{0}^{1}\frac{(\widetilde{p}_x)^2}{\widetilde{p}}{\rm d}x +\frac{\varepsilon}{2}\|q_x\|_{L^{2}}^2\leq0. \end{equation} $

应用Gronwall不等式可知

$ \begin{equation} \|q(t)\|_{L^{2}}^2+\int_{0}^{t}\|q_x(\tau)\|_{L^{2}}^2{\rm d}\tau\leq C_{24}, \end{equation} $

其中常数$ C_{24} $$ t $无关但与$ 1/\varepsilon $成比例.

步骤2  下面考虑$ \overline{p} = 0 $情形下的初边值问题(2.1)-(2.2):

$ \begin{equation} \left\{ \begin{array}{lll} p_t-(pq)_x = p_{xx}, \\ q_t-p_x = \varepsilon q_{xx} + \varepsilon (q^2)_x, \\ p|_{x = 0} = 0, \quad p_x|_{x = 1} = 0, \\ q|_{x = 0, x = 1} = 0, \\ (p, q)(x, 0) = (p_0, q_0)(x). \end{array} \right. \end{equation} $

在方程$ (2.68)_1 $两边同时乘以$ p $, 在方程$ (2.68)_2 $两边同时乘以$ q $, 并在$ (0, 1) $上积分, 可得

$ \begin{eqnarray} && \frac{1}{2}\frac{\rm d}{{\rm d}t}(\|p\|_{L^{2}}^2+\|q\|_{L^{2}}^2)+ \|p_x\|_{L^{2}}^2+\varepsilon\|q_x\|_{L^{2}}^2 = \int_{0}^{1}p_x q {\rm d}x-\int_{0}^{1}p q p_x {\rm d}x\\ &\leq&\frac{1}{4}\|p_x\|_{L^{2}}^2+\|q\|_{L^{2}}^2+\frac{1}{4}\|p_x\|_{L^{2}}^2+ \|p\|_{L^{2}}^2\|q\|_{L^{\infty}}^2\\ &\leq&\frac{1}{2}\|p_x\|_{L^{2}}^2+C_{25}\|q_x\|_{L^{2}}^2+C_{26}\|p\|_{L^{2}}^2\|q_x\|_{L^{2}}^2, \end{eqnarray} $

这里用到了Young不等式, Gagliardo-Nirenberg不等式和Poincaré不等式.于是

$ \begin{equation} \frac{1}{2}\frac{\rm d}{{\rm d}t}(\|p\|_{L^{2}}^2+\|q\|_{L^{2}}^2)+ \frac{1}{2}\|p_x\|_{L^{2}}^2+\varepsilon\|q_x\|_{L^{2}}^2\leq C_{25}\|q_x\|_{L^{2}}^2+C_{26}\|p\|_{L^{2}}^2\|q_x\|_{L^{2}}^2. \end{equation} $

应用Gronwall不等式和(2.67)式可得

$ \begin{equation} \|p(t)\|_{L^{2}}^2+\int_{0}^{t}\|p_x(\tau)\|_{L^{2}}^2{\rm d}\tau\leq C_{27}, \end{equation} $

其中常数$ C_{27} $$ t $无关但与$ 1/\varepsilon $成比例.

步骤3  类似于引理$ 2.2 $的证明方法, 可以建立解的高阶估计.在方程$ (2.16)_1 $两边同时乘以$ \widetilde{p}_{xx} $, 在方程$ (2.16)_2 $两边同时乘以$ q_{xx} $, 并在区间$ (0, 1) $上积分, 计算得到

这里用到了Hölder不等式, Sobolev不等式以及引理$ 2.2 $.于是

$ \begin{equation} \frac{1}{2}\frac{\rm d}{{\rm d}t}(\|p_x\|_{L^{2}}^2+\|q_x\|_{L^{2}}^2)+ \frac{1}{2}\|p_{xx}\|_{L^{2}}^2+\frac{\varepsilon}{2}\|q_{xx}\|_{L^{2}}^2 \leq C_{31}(\|p_x\|_{L^{2}}^2+\|q_x\|_{L^{2}}^2)(\|q_x\|_{L^{2}}^2+1). \end{equation} $

应用Gronwall不等式, $ (2.67) $式和$ (2.71) $式, 可得

$ \begin{equation} \|p_x(t)\|_{L^{2}}^2+\|q_x(t)\|_{L^{2}}^2+ \int_{0}^{t}(\|p_{xx}(\tau)\|_{L^{2}}^2+\|q_{xx}(\tau)\|_{L^{2}}^2){\rm d}\tau\leq C_{32}, \end{equation} $

其中常数$ C_{32} $$ t $无关但与$ 1/\varepsilon $成比例.

步骤4  在方程$ (2.68)_1 $和方程$ (2.68)_2 $两边关于$ t $求偏导, 分别乘以$ p_t $$ q_t $, 并在区间$ (0, 1) $上积分, 得到

于是

$ \begin{eqnarray} &&\frac{1}{2}\frac{\rm d}{{\rm d}t}(\|p_t\|_{L^{2}}^2+\|q_t\|_{L^{2}}^2)+ \frac{1}{2}\|p_{xt}\|_{L^{2}}^2+\frac{\varepsilon}{2}\|q_{xt}\|_{L^{2}}^2\\ &\leq& C_{34}(\|p_x\|_{L^{2}}^2+\|q_x\|_{L^{2}}^2)(\|p_t\|_{L^{2}}^2+\|q_t\|_{L^{2}}^2)+\|q_t\|_{L^{2}}^2. \end{eqnarray} $

根据方程$ (2.68)_2 $可得

$ \begin{eqnarray} \|q_t\|_{L^{2}}^2&\leq& C_{35}(\|q_{xx}\|_{L^{2}}^2+\|p_x\|_{L^{2}}^2+\|q_x\|_{L^{2}}^2+ \|q\|_{L^{\infty}}^2\|q_x\|_{L^{2}}^2)\\ &\leq&C_{36}(\|q_{xx}\|_{L^{2}}^2+\|p_x\|_{L^{2}}^2+\|q_x\|_{L^{2}}^2+ \|q_x\|_{L^{2}}^2\|q_x\|_{L^{2}}^2)\\ &\leq&C_{37}(\|q_{xx}\|_{L^{2}}^2+\|p_x\|_{L^{2}}^2+\|q_x\|_{L^{2}}^2), \end{eqnarray} $

这里用到了Gagliardo-Nirenberg不等式, $ (2.71) $式和$ (2.73) $式.因此由$ (2.74) $式知

$ \begin{eqnarray} &&\frac{1}{2}\frac{\rm d}{{\rm d}t}(\|p_t\|_{L^{2}}^2+\|q_t\|_{L^{2}}^2)+ \frac{1}{2}\|p_{xt}\|_{L^{2}}^2+\frac{\varepsilon}{2}\|q_{xt}\|_{L^{2}}^2\\ &\leq& C_{34}(\|p_x\|_{L^{2}}^2+\|q_x\|_{L^{2}}^2)(\|p_t\|_{L^{2}}^2+\|q_t\|_{L^{2}}^2)+C_{37}(\|q_{xx}\|_{L^{2}}^2+\|p_x\|_{L^{2}}^2+\|q_x\|_{L^{2}}^2), \end{eqnarray} $

类似于估计式$ (2.73) $, 可得

$ \begin{equation} \|p_t(t)\|_{L^{2}}^2+\|q_t(t)\|_{L^{2}}^2+ \int_{0}^{t}(\|p_{xt}(\tau)\|_{L^{2}}^2+\|q_{xt}(\tau)\|_{L^{2}}^2){\rm d}\tau\leq C_{38}, \end{equation} $

其中常数$ C_{38}>0 $$ t $无关.

步骤5  在方程$ (2.68)_{1} $和方程$ (2.68)_{2} $两边取$ L^2 $范数得

$ \begin{equation} \quad \quad \|p_{xx}\|_{L^{2}}^2\leq C_{39}\big(\|p_t\|_{L^{2}}^2+\|p\|_{L^{\infty}}^2\|q_x\|_{L^{2}}^2+ \|q\|_{L^{\infty}}^2\|p_x\|_{L^{2}}^2\big) \leq C_{40} \end{equation} $

$ \begin{equation} \|q_{xx}\|_{L^{2}}^2\leq C_{41}(\|q_t\|_{L^{2}}^2+\|p_x\|_{L^{2}}^2+ \|q\|_{L^{\infty}}^2\|q_x\|_{L^{2}}^2)\leq C_{42}, \end{equation} $

这里常数$ C_{39}, \cdots , C_{42} $都与$ t $无关.联合(2.77)-(2.79)式可知, 存在与$ t $无关但与$ 1/\varepsilon $有关的常数$ C_{43}>0 $使得

$ \begin{equation} \|p_{xx}(t)\|_{L^{2}}^2+\|q_{xx}(t)\|_{L^{2}}^2+\int_{0}^{t}(\|p_{xxx}(\tau)\|_{L^{2}}^2+ \|q_{xxx}(\tau)\|_{L^{2}}^2){\rm d}\tau\leq C_{43}. \end{equation} $

从而有(2.64)式成立.

引理2.5  (指数收敛性)  设$ (\widetilde{p}, q) $是问题$ (2.63) $的解, 则对任意的$ \varepsilon>0 $, 存在与$ t $无关的正常数$ \alpha_{1} $, $ \beta_{1} $满足

  设$ R $为待确定的正参数.在方程$ (2.63)_1 $两边同时乘以$ \widetilde{p} $, 在方程$ (2.63)_2 $两边同时乘以$ -Rq $, 关于$ x $$ (0, 1) $上积分, 两式相减可得

$ \begin{equation} \frac{\rm d}{{\rm d}t}\Big(\frac{1}{2}\|{\widetilde{p}}\|_{L^{2}}^2+\frac{R}{2}\|{q}\|_{L^{2}}^2\Big)+ \|{\widetilde{p}_x}\|_{L^{2}}^2+\varepsilon R\|{q_x}\|_{L^{2}}^2 = -\int_{0}^{1}\widetilde{p} q\widetilde{p}_x {\rm d}x. \end{equation} $

方程$ (2.63)_1 $两边同时乘以$ \widetilde{p}^{2} $, 关于$ x $$ (0, 1) $上积分, 利用分部积分可得

$ \begin{equation} \frac{\rm d}{{\rm d}t}\bigg(\frac{1}{3}\int_{0}^{1}\widetilde{p}^{3}{\rm d}x\bigg) +2\int_{0}^{1}\widetilde{p}(\widetilde{p}_x)^{2}{\rm d}x = -2\int_{0}^{1}\widetilde{p}^{2}q\widetilde{p}_x {\rm d}x +2R\int_{0}^{1}\widetilde{p}q\widetilde{p}_x {\rm d}x. \end{equation} $

方程$ (2.63)_1 $两边同时乘以$ \widetilde{p}^{3} $, 关于$ x $$ (0, 1) $上积分, 利用分部积分可得

$ \begin{equation} \frac{\rm d}{{\rm d}t}\bigg(\frac{1}{4}\|\widetilde{p}\|_{L^4}^4\bigg)+3\|{\widetilde{p}\widetilde{p}_x}\|_{L^{2}}^2 = -3\int_{0}^{1}\widetilde{p}^{3}q\widetilde{p}_x {\rm d}x +3R\int_{0}^{1}\widetilde{p}^2q\widetilde{p}_x {\rm d}x. \end{equation} $

$ (2.81) $式两边同时乘以$ 2R $, 与$ (2.82) $式相加得

$ \begin{eqnarray} && \frac{\rm d}{{\rm d}t}\bigg(R\|{\widetilde{p}}\|_{L^{2}}^2+R^2\|{q}\|_{L^{2}}^2 +\frac{1}{3}\int_{0}^{1}\widetilde{p}^3{\rm d}x\bigg)+2R\|{\widetilde{p}_x}\|_{L^{2}}^2 +2\int_{0}^{1}\widetilde{p}(\widetilde{p}_x)^2{\rm d}x+ 2R^2\varepsilon\|{q_x}\|_{L^{2}}^2 \\ & = &-2\int_{0}^{1}\widetilde{p}^{2}q\widetilde{p}_x {\rm d}x. \end{eqnarray} $

$ (2.84) $式两边同时乘以$ \frac{3}{2}R $, 与$ (2.83) $式相加, 得到

$ \begin{eqnarray} &&\frac{\rm d}{{\rm d}t}\bigg(\frac{1}{4}\|\widetilde{p}\|_{L^{4}}^4+\frac{3R^2}{2}\|\widetilde{p}\|_{L^{2}}^2+ \frac{3R^3}{2}\|q\|_{L^{2}}^2+\frac{R}{2}\int_{0}^{1}\widetilde{p}^{3}{\rm d}x\bigg)\\ &&+ \bigg(3\|\widetilde{p}\widetilde{p}_x\|_{L^{2}}^2+3R^2\|\widetilde{p}_x\|_{L^{2}}^2 +3R\int_{0}^{1}\widetilde{p}(\widetilde{p}_x)^2{\rm d}x +3 R^3\varepsilon\|q_x\|_{L^{2}}^2\bigg)\\ & = & -3\int_{0}^{1}\widetilde{p}^3q\widetilde{p}_x{\rm d}x. \end{eqnarray} $

因此有

$ \begin{equation} \frac{\rm d}{{\rm d}t}G_{2}(t)+K_2(t) = -3\int_{0}^{1}\widetilde{p}^{3}q\widetilde{p}_x {\rm d}x, \end{equation} $

这里

$ \begin{eqnarray} G_{2}(t) & = &\frac{3}{2}R^2\|{\widetilde{p}}\|_{L^{2}}^2 +\frac{3}{2}R^3\|q\|_{L^{2}}^2 +\frac{R}{2}\int_{0}^{1}\widetilde{p}^3{\rm d}x +\frac{1}{4}\|\widetilde{p}\|_{L^4}^4\\ & = &R^2\|{\widetilde{p}}\|_{L^{2}}^2+ \frac{1}{8}\|2R{\widetilde{p}+\widetilde{p}^2}\|_{L^{2}}^2 +\frac{1}{8}\|\widetilde{p}\|_{L^4}^4 +\frac{3}{2}R^3\|q\|_{L^{2}}^2, \\ K_2(t) & = &3R^2\|{\widetilde{p}_x}\|_{L^{2}}^2 +3R\int_{0}^{1}\widetilde{p}(\widetilde{p}_x)^2{\rm d}x +3\|\widetilde{p}\widetilde{p}_x\|_{L^{2}}^2 +3R^3\varepsilon\|q_x\|_{L^{2}}^2\\ & = &\frac{3}{2}R^2\|\widetilde{p}_x\|_{L^{2}}^2 +\frac{3}{2}\|R\widetilde{p}_x+\widetilde{p}\widetilde{p}_x\|_{L^{2}}^2 +\frac{3}{2}\|\widetilde{p}\widetilde{p}_x\|_{L^{2}}^2 +3R^3\varepsilon\|q_x\|_{L^{2}}^2. \end{eqnarray} $

与引理2.3中的证明方法类似, 可知存在与$ R $有关而与$ t $无关的正常数$ \alpha_{1} $, $ \beta_{1} $满足

$ \begin{equation} \|\widetilde{p}(t)\|_{H^{2}}^2+\|q(t)\|_{H^{2}}^2\leq \alpha_{1}e^{-\beta_{1}t}. \end{equation} $

引理2.5得证.

由上述$ \overline{p}>0 $$ \overline{p} = 0 $时的结果, 我们完成了定理1.1对于$ \varepsilon>0 $情形的结果的证明.

3 非扩散问题($ \varepsilon=0 $)解的全局存在性

本节证明$ \varepsilon = 0 $时问题(1.3)-(1.4)解的全局存在性.考虑变换后的初边值问题

$ \begin{equation} \left\{ \begin{array}{lll} \widetilde{p}_t-(\widetilde{p}q)_x-\overline{p}q_x = \widetilde{p}_{xx}, \\ q_t-\widetilde{p}_x = 0, \\ (\widetilde{p}, q)(x, 0) = (p_0-\overline{p}, q_0)(x), \\ \widetilde{p}|_{x = 0} = 0, \, \, \widetilde{p}_x|_{x = 1} = 0, \\ \end{array} \right. \end{equation} $

其中$ \widetilde{p} = p-\overline{p} $.首先根据不动点方法证明问题$ (3.1) $解的局部存在性, 然后利用能量法建立相应的先验估计.确切地说, 当$ \overline{p}>0 $时, 证明方法与第2节中类似.当$ \overline{p} = 0 $时, 对问题(3.1)应用扰动方法得到变换后的方程组, 再利用Lyapunov泛函证明解的先验估计.所以下面的讨论分为$ \overline{p}>0 $$ \overline{p} = 0 $两种情形.

3.1 $ \overline{p}>0 $的情形

引理3.1  设$ (\widetilde{p}, q) $是问题$ (3.1) $的解, 则对任意的$ t>0 $, 存在与$ t $无关的常数$ C $使得

$ \begin{equation} \|(\widetilde{p}, q)(t)\|_{H^{2}}^2+\int_{0}^{t}\|\widetilde{p}(\tau)\|_{H^{3}}^2 +\|q(\tau)\|_{H^{2}}^2{\rm d}\tau\leq C. \end{equation} $

  证明分为五个步骤.

步骤1  考虑引理$ 2.1 $中建立的能量估计.注意到, 当$ \varepsilon = 0 $时, $ (2.11) $式和$ (2.38) $式仍然成立, 即有

$ \begin{equation} \frac{\rm d}{{\rm d}t}\bigg(\int_{0}^{1}\eta(p)-\eta(\overline{p})-\eta'(\overline{p}) (p-\overline{p}){\rm d}x+\frac{1}{2}\|q\|_{L^{2}}^2\bigg )+ \int_{0}^{1}\frac{(p_x)^2}{\widetilde{p}+\overline{p}}{\rm d}x = 0 \end{equation} $

$ \begin{equation} \frac{\rm d}{{\rm d}t}G_1(t)+\frac{3}{2}\overline{p}^2\|{\widetilde{p}_x}\|_{L^{2}}^2 +\frac{3}{2}\|\overline{p}{\widetilde{p}_x-\widetilde{p}\widetilde{p}_x}\|_{L^{2}}^2 +\|\widetilde{p}\widetilde{p}_x\|_{L^{2}}^2\leq C_{6}G_1(t)\int_{0}^{1}\frac{(\widetilde{p}_x)^2}{\widetilde{p}+\overline{p}}{\rm d}x. \end{equation} $

对任意的$ t>0 $, $ (3.3) $式和$ (3.4) $式两端分别在$ [0, t] $上积分得

从而有

$ \begin{equation} \|\widetilde{p}(t)\|_{L^{2}}^2+\|q(t)\|_{L^{2}}^2+\int_{0}^{t}\bigg(\int_{0}^{1}\frac{(\widetilde{p}_x)^2} {\widetilde{p}+\overline{p}}{\rm d}x+\|\widetilde{p}_x(\tau)\|_{L^{2}}^2\bigg){\rm d}\tau\leq C_{46}, \end{equation} $

这里常数$ C_{46} $$ t $无关.

步骤2  方程$ (3.1)_2 $两边关于$ x $求偏导, 代入方程$ (3.1)_1 $

$ \begin{equation} q_{xt} = \widetilde{p}_t-\overline{p}q_x-(\widetilde{p}q)_x. \end{equation} $

上式两边同时乘以$ q_x $, 关于$ x $$ (0, 1) $上积分, 计算得到

$ (2.44) $式的推导类似, 由上式和Gagliardo-Nirenberg不等式可得

$ \begin{eqnarray} \frac{\rm d}{{\rm d}t}\bigg(\frac{1}{2}\|q_x\|_{L^{2}}^2-\int_{0}^{1}\widetilde{p}q_x{\rm d}x\bigg)+ \frac{3\overline{p}}{4}\|q_x\|_{L^{2}}^2 &\leq& C_{47}(\|\widetilde{p}\|_{L^{\infty}}^2\|q_x\|_{L^{2}}^2+ \|q\|_{L^{\infty}}^2\|\widetilde{p}_x\|_{L^{2}}^2) -\int_{0}^{1}\widetilde{p}\widetilde{p}_{xx}{\rm d}x\\ &\leq &C_{48}(\|\widetilde{p}_x\|_{L^{2}}^2\|q_x\|_{L^{2}}^2+\|\widetilde{p}_x\|_{L^{2}}^2)+ \frac{\overline{p}}{4}\|q_x\|_{L^{2}}^2, \end{eqnarray} $

这里我们用到了由方程$ (3.1)_2 $得到的$ q_{xt}-\widetilde{p}_{xx} = 0 $.于是有

$ \begin{equation} \frac{\rm d}{{\rm d}t}\bigg(\frac{1}{2}\|q_x\|_{L^{2}}^2-\int_{0}^{1}\widetilde{p}q_x{\rm d}x\bigg)+ \frac{\overline{p}}{2}\|q_x\|_{L^{2}}^2\leq C_{48}(\|\widetilde{p}_x\|_{L^{2}}^2\|q_x\|_{L^{2}}^2+ \|\widetilde{p}_x\|_{L^{2}}^2). \end{equation} $

$ (3.4) $式两边同时乘以$ \frac{2}{\overline{p}^2} $, 并与$ (3.8) $式相加可得

这里

根据$ G_1(t) $的定义可知, 存在常数$ C_{49}>0 $满足

$ \begin{equation} \frac{\rm d}{{\rm d}t}G_{3}(t)+K_3(t)\leq C_{50}G_{3}(t)\bigg(\int_{0}^{1}\frac{(\widetilde{p}_x)^2}{\widetilde{p}+ \overline{p}}{\rm d}x+\|\widetilde{p}_x\|_{L^{2}}^2\bigg)+C_{49}\|\widetilde{p}_x\|_{L^{2}}^2. \end{equation} $

应用Gronwall不等式和$ (3.5) $式可得

$ G_{3}(t) $$ K_3(t) $的定义可知

$ \begin{equation} \|q_x(t)\|_{L^{2}}^2+\int_{0}^{t}\|q_x(\tau)\|_{L^{2}}^2{\rm d}\tau \leq C_{52}, \end{equation} $

这里常数$ C_{52}>0 $.注意到常数$ C_{47}, \cdots , C_{52} $都与$ t $无关.

步骤3  方程$ (3.1)_{1} $两边同时乘以$ \widetilde{p}_{xx} $, 关于$ x $$ (0, 1) $上积分得

从而, 与$ (3.7) $式的推导类似, 可得

$ \begin{eqnarray} \frac{\rm d}{{\rm d}t}\|\widetilde{p}_x\|_{L^{2}}^2+\|\widetilde{p}_{xx}\|_{L^{2}}^2&\leq& C_{53}(\|\widetilde{p}q_x\|_{L^{2}}^2+\|\widetilde{p}_xq\|_{L^{2}}^2+\|q_x\|_{L^{2}}^2)\\ &\leq& C_{54}(\|\widetilde{p}_x\|_{L^{2}}^2\|q_x\|_{L^{2}}^2+\|q_x\|_{L^{2}}^2). \end{eqnarray} $

应用Gronwall不等式和$ (3.10) $式可知, 存在与$ t $无关的常数$ C_{55}>0 $满足

$ \begin{equation} \|\widetilde{p}_x(t)\|_{L^{2}}^2+\int_{0}^{t}\|\widetilde{p}_{xx}(\tau)\|_{L^{2}}^2{\rm d}\tau \leq C_{55}. \end{equation} $

步骤4  方程$ (3.1)_1 $和方程$ (3.1)_2 $两边关于$ t $求偏导, 分别乘$ \widetilde{p}_t $$ \overline{p}q_t $, 然后关于$ x $$ (0, 1) $上积分, 得到

由此可得

$ \begin{eqnarray} \frac{\rm d}{{\rm d}t}\big(\|\widetilde{p}_t(t)\|_{L^{2}}^2+\overline{p}\|q_t(t)\|_{L^{2}}^2\big)+ \|\widetilde{p}_{xt}\|_{L^{2}}^2 &\leq &C_{56}\big(\|q_x\|_{L^{2}}^2\|\widetilde{p}_t\|_{L^{2}}^2+ \|\widetilde{p}_x\|_{L^{2}}^2\|q_t\|_{L^{2}}^2\big)\\ &\leq &C_{57}\big(\|\widetilde{p}_t\|_{L^{2}}^2+\overline{p}\|q_t(t)\|_{L^{2}}^2\big)\big( \|\widetilde{p}_x\|_{L^{2}}^2+\|q_x\|_{L^{2}}^2\big), {\qquad} \end{eqnarray} $

这里我们用到了不等式$ \|\widetilde{p}\|_{L^{\infty}}^2\leq C\|\widetilde{p}_x\|_{L^{2}}^2 $$ \|q\|_{L^{\infty}}^2\leq C\|q_x\|_{L^{2}}^2 $.对(3.13)式应用Gronwall不等式, 联合$ (3.5) $式, $ (3.10) $式和$ (3.12) $式可得

$ \begin{equation} \|\widetilde{p}_t(t)\|_{L^{2}}^2+\|q_t(t)\|_{L^{2}}^2+ \int_{0}^{t}\|\widetilde{p}_{xt}(\tau)\|_{L^{2}}^2 {\rm d}\tau \leq C_{58}. \end{equation} $

由上式和(3.12)式可知, 存在与$ t $无关的常数$ C_{59}>0 $使得

$ \begin{equation} \|\widetilde{p}_{xx}(t)\|_{L^{2}}^2+\|q_{t}(t)\|_{L^{2}}^2+ \int_{0}^{t}\|\widetilde{p}_{xt}(\tau)\|_{L^{2}}^2 {\rm d}\tau \leq C_{59}. \end{equation} $

步骤5  方程$ (3.6) $两端关于$ x $求偏导, 然后乘以$ q_{xx} $, 并且关于$ x $$ (0, 1) $上积分得

$ \begin{eqnarray} \frac{1}{2}\frac{\rm d}{{\rm d}t}\|q_{xx}\|_{L^{2}}^2+\overline{p}\|q_{xx}\|_{L^{2}}^2 & = &\int_{0}^{1}\big(\widetilde{p}_{xt}-(\widetilde{p}q)_{xx}\big)q_{xx}{\rm d}x{}\\ &\leq&\frac{\overline{p}}{2}\|q_{xx}\|_{L^{2}}^2+C_{60}(\|\widetilde{p}_{xt}\|_{L^{2}}^2 +\|q\|_{L^{\infty}}^2\|\widetilde{p}_{xx}\|_{L^{2}}^2{}\\ &&+\|\widetilde{p}_x\|_{L^{2}}^2 \|q_x\|_{L^{\infty}}^2+\|\widetilde{p}\|_{L^{\infty}}^2\|q_{xx}\|_{L^{2}}^2). \end{eqnarray} $

由齐次边界条件和齐次椭圆正则性有$ \|q\|_{H^{2}} \leq C \|q_{xx}\|_{L^{2}} $.又由Sobolev嵌入不等式知$ \|q_x\|_{L^{\infty}} $$ \leq C \|q_x\|_{H^{1}}\leq C \|q\|_{H^{2}} $.因此有不等式$ \|q_x\|_{L^{\infty}}\leq C \|q_{xx}\|_{L^{2}} $.联合不等式$ \|\widetilde{p}\|_{L^{\infty}}\leq C\|\widetilde{p}_x\|_{L^{2}} $, $ \|q\|_{L^{\infty}}\leq C\|q_x\|_{L^{2}} $, $ (3.15) $式和(3.16)式可得

$ \begin{equation} \frac{1}{2}\frac{\rm d}{{\rm d}t}\|q_{xx}\|_{L^{2}}^2+\overline{p}\|q_{xx}\|_{L^{2}}^2 \leq\frac{\overline{p}}{2}\|q_{xx}\|_{L^{2}}^2+C_{61}(\|\widetilde{p}_{xt}\|_{L^{2}}^2+ \|q_x\|_{L^{2}}^2)+C_{62}\|\widetilde{p}_x\|_{L^{2}}^2\|q_{xx}\|_{L^{2}}^2. \end{equation} $

因此

$ \begin{equation} \|q_{xx}(t)\|_{L^{2}}^2+ \int_{0}^{t}\|q_{xx}(\tau)\|_{L^{2}}^2 {\rm d}\tau \leq C_{63}, \end{equation} $

这里用到了Gronwall不等式, $ (3.5) $式, $ (3.10) $式, $ (3.15) $式和$ (3.17) $式.联合$ (3.15) $式和$ (3.18) $式可得

其中常数$ C_{64}>0 $.注意到常数$ C_{60}, \cdots , C_{64} $都与$ t $无关.引理3.1得证.

非扩散情形解的指数衰减估计的证明和扩散情形的证明完全类似, 因此这里省略证明.由此, 我们完成了$ \overline{p}>0 $情形下定理1.1的证明.

3.2 $ \overline{p} = 0 $的情形

本小节证明$ \overline{p} = 0 $时非扩散情形解的全局存在性, 考虑问题(3.1)在扰动$ \widetilde{p} = p+1 $下得到的初边值问题

$ \begin{equation} \left\{ \begin{array}{lll} \widetilde{p}_t-(\widetilde{p}q)_x+q_x = \widetilde{p}_{xx}, \\ q_t-\widetilde{p}_x = 0, \\ \widetilde{p}|_{x = 0} = 1, \quad \widetilde{p}_x|_{x = 1} = 0, \\ (\widetilde{p}, q)(x, 0) = (p_0+1, q_0)(x).\\ \end{array} \right. \end{equation} $

引理3.2  设$ (\widetilde{p}, q) $是问题$ (3.19) $的解, 则对任意的$ t>0 $都有

$ \begin{equation} \|(\widetilde{p}, q)(t)\|_{H^{2}}^2+\int_{0}^{t}\|\widetilde{p}(\tau)\|_{H^{3}}^2 \leq C, \end{equation} $

其中常数$ C $$ t $有关.

  证明分为四个步骤.

步骤1  当$ \varepsilon = 0 $时, 在$ (2.65) $式中取$ R = 1 $

$ \begin{eqnarray} && \frac{\rm d}{{\rm d}t}\bigg(\int_{0}^{1}\eta(\widetilde{p})-\eta(1)-\eta'(1) (\widetilde{p}-1){\rm d}x+\frac{1}{2}\|q\|_{L^{2}}^2\bigg) +\int_{0}^{1}\frac{(\widetilde{p}_x)^2}{\widetilde{p}}{\rm d}x\\ & = &\int_{0}^{1}q\frac{\widetilde{p}_x}{\widetilde{p}}{\rm d}x \leq\frac{1}{2}\int_{0}^{1}\frac{(\widetilde{p}_x)^2}{\widetilde{p}}{\rm d}x+ \frac{1}{2}\int_{0}^{1}\frac{q^2}{\widetilde{p}}{\rm d}x\leq \frac{1}{2}\int_{0}^{1}\frac{(\widetilde{p}_x)^2}{\widetilde{p}}{\rm d}x+ \frac{1}{2}\|q\|_{L^{2}}^2, \end{eqnarray} $

上式用到了$ \widetilde{p}\geq1 $的条件.从而有

$ \begin{equation} \frac{\rm d}{{\rm d}t}\bigg(\int_{0}^{1}\eta(\widetilde{p})-\eta(1)-\eta'(1) (\widetilde{p}-1){\rm d}x+\frac{1}{2}\|q\|_{L^{2}}^2\bigg)+\frac{1}{2} \int_{0}^{1}\frac{(\widetilde{p}_x)^2}{\widetilde{p}}{\rm d}x \leq \frac{1}{2}\|q\|_{L^{2}}^2. \end{equation} $

根据Gronwall不等式和条件$ \eta(\widetilde{p})-\eta(1)-\eta'(1)(\widetilde{p}-1)\geq0 $可得

$ \begin{eqnarray} \|q(t)\|_{L^{2}}^2\leq 2e^{t}\bigg(\int_{0}^{1}\eta(\widetilde{p}_{0})-\eta(1)-\eta'(1) (\widetilde{p}_{0}-1){\rm d}x+\frac{1}{2}\|q_{0}\|_{L^{2}}^2\bigg) \leq C_{65}e^{t}, \end{eqnarray} $

这里常数$ C_{65}>0 $$ t $无关.

步骤2  现在建立$ p, \, \, q $$ L^{2} $估计.回到问题(3.19)在齐次混合边界条件下对应的初边值问题:

$ \begin{equation} \left\{ \begin{array}{lll} p_t-(pq)_x = p_{xx}, \\ q_t-p_x = 0, \\ p|_{x = 0} = 0, \quad p_x|_{x = 1} = 0, \\ q|_{x = 0, x = 1} = 0, \\ (p, q)(x, 0) = (p_0, q_0)(x). \end{array} \right. \end{equation} $

$ (2.69) $式的推导类似, 由Gagliardo-Nirenberg不等式, 带$ \epsilon $的Young不等式和$ (3.23) $式可得

于是有

$ \begin{equation} \frac{1}{2}\frac{\rm d}{{\rm d}t}(\|p\|_{L^{2}}^2+\|q\|_{L^{2}}^2)+\frac{1}{2}\|p_x\|_{L^{2}}^2 \leq C_{69}\big(e^{t}+e^{2t}\big)\big(\|p\|_{L^{2}}^2+\|q\|_{L^{2}}^2\big). \end{equation} $

由Gronwall不等式可得

$ \begin{equation} \|p(t)\|_{L^{2}}^2+\|q(t)\|_{L^{2}}^2+\int_{0}^{t}\|p_x(\tau)\|_{L^{2}}^2{\rm d}\tau\leq C_{70}(t), \end{equation} $

这里, $ C_{70}(t) $是关于$ t $的递增函数.

步骤3  下面建立$ q $的一阶估计.与$ (3.8) $式的证明类似, 可得

$ \begin{eqnarray} \frac{\rm d}{{\rm d}t}\bigg(\frac{1}{2}\|q_x\|_{L^{2}}^2-\int_{0}^{1}pq_x{\rm d}x \bigg) & = &\|p_x\|_{L^{2}}^2-\int_{0}^{1}(pq)_x q_x{\rm d}x\\ &\leq& \|p_x\|_{L^{2}}^2+\frac{1}{2}\|q_x\|_{L^{2}}^2+\frac{1}{2}\|p\|_{L^{\infty}}^2\|q_x\|_{L^{2}}^2 +\frac{1}{2}\|q\|_{L^{\infty}}^2\|p_x\|_{L^{2}}^2\\ &\leq &\|p_x\|_{L^{2}}^2+C_{71}(\|p_x\|_{L^{2}}^2+1)\|q_x\|_{L^{2}}^2. \end{eqnarray} $

根据Gronwall不等式, $ (3.25) $式和$ (3.26) $式可得

$ \begin{equation} \|q_x(t)\|_{L^{2}}^2\leq C_{72}. \end{equation} $

对于$ \widetilde{p} $的一阶估计, 与$ (3.11) $式的证明类似, 有

于是

$ \begin{equation} \frac{1}{2}\frac{\rm d}{{\rm d}t}\|\widetilde{p}_x\|_{L^{2}}^2+\frac{1}{2}\|\widetilde{p}_{xx}\|_{L^{2}}^2 \leq C_{73}(\|\widetilde{p}_x\|_{L^{2}}^2\|q_x\|_{L^{2}}^2+\|q_x\|_{L^{2}}^2). \end{equation} $

由Gronwall不等式和$ (3.28) $式可得

$ \begin{equation} \|\widetilde{p}_x(t)\|_{L^{2}}^2+\int_{0}^{t}\|\widetilde{p}_{xx}(\tau)\|_{L^{2}}^2 {\rm d}\tau \leq C_{74}. \end{equation} $

步骤4  最后建立解的高阶估计.在方程$ (3.19)_1 $和方程$ (3.19)_2 $两端关于$ t $求偏导, 分别乘以$ \widetilde{p}_t $$ q_t $, 然后关于$ x $$ (0, 1) $上积分, 得到

$ \begin{eqnarray} && \frac{1}{2}\frac{\rm d}{{\rm d}t}\bigg(\|\widetilde{p}_t(t)\|_{L^{2}}^2+\|q_t(t)\|_{L^{2}}^2\bigg)+ \|\widetilde{p}_{xt}\|_{L^{2}}^2{}\\ & = &-\int_{0}^{1}(\widetilde{p}q)_t \widetilde{p}_{xt}{\rm d}x \leq \frac{1}{2}\|\widetilde{p}_{xt}\|_{L^{2}}^2+\|q\|_{L^{\infty}}^2\|\widetilde{p}_t\|_{L^{2}}^2+ \|\widetilde{p}\|_{L^{\infty}}^2\|q_t\|_{L^{2}}^2\\ &\leq &\frac{1}{2}\|\widetilde{p}_{xt}\|_{L^{2}}^2+C_{75}(\|\widetilde{p}_t\|_{L^{2}}^2+\|q_t\|_{L^{2}}^2), \end{eqnarray} $

这里用到了不等式$ \|\widetilde{p}\|_{L^{\infty}}^2 \leq C(2+\|\widetilde{p}_x\|_{L^{2}}^2), $$ \|q\|_{L^{\infty}}^2 \leq C\|q_x\|_{L^{2}}^2 $, $ (3.28) $式和$ (3.30) $式.由Gronwall不等式可得

$ \begin{equation} \|\widetilde{p}_t(t)\|_{L^{2}}^2+\|q_t(t)\|_{L^{2}}^2+\int_{0}^{t}\|\widetilde{p}_{xt}(\tau)\|_{L^{2}}^2{\rm d}\tau \leq C_{76}, \end{equation} $

于是

$ \begin{equation} \|\widetilde{p}_{xx}(t)\|_{L^{2}}^2+\|q_{t}(t)\|_{L^{2}}^2+\int_{0}^{t}\|\widetilde{p}_{xt}(\tau)\|_{L^{2}}^2{\rm d}\tau \leq C_{77}. \end{equation} $

方程$ (3.19)_2 $两端关于$ x $求偏导, 与方程$ (3.19)_1 $相减得

$ \begin{equation} q_{xt} = \widetilde{p}_t+q_x-(\widetilde{p}q)_x. \end{equation} $

上式两端关于$ x $求偏导, 乘以$ q_{xx} $, 并关于$ x $$ (0, 1) $上积分, 得到

$ \begin{equation} \int_{0}^{1} q_{xx}q_{xxt}{\rm d}x = \int_{0}^{1}\widetilde{p}_{xt} q_{xx}{\rm d}x-\int_{0}^{1}(\widetilde{p}q)_{xx} q_{xx}{\rm d}x. \end{equation} $

从而有

$ \begin{eqnarray} \frac{\rm d}{{\rm d}t}\|q_{xx}\|_{L^{2}}^2 &\leq& C_{78}(\|\widetilde{p}_{xt}\|_{L^{2}}^2+ \|\widetilde{p}\|_{L^{\infty}}^2\|q_{xx}\|_{L^{2}}^2+\|q\|_{L^{\infty}}^2\|\widetilde{p}_{xx}\|_{L^{2}}^2+\|\widetilde{p}_x\|_{L^{\infty}}^2\|q_x\|_{L^{2}}^2)+\frac{1}{2}\|q_{xx}\|_{L^{2}}^2\\ &\leq &C_{79}(\|\widetilde{p}_x\|_{L^2}^2+1)\|q_{xx}\|_{L^{2}}^2+C_{80}(\|\widetilde{p}_{xt}\|_{L^{2}}^2 +\|\widetilde{p}_{xx}\|_{L^{2}}^2+\|q_x\|_{L^{2}}^2), \end{eqnarray} $

这里用到了不等式$ \|\widetilde{p}\|_{L^{\infty}}^2 \leq C(2+\|\widetilde{p}_x\|_{L^{2}})^2 $$ \|q\|_{L^{\infty}}^2 \leq C\|q_x\|_{L^{2}}^2 $.对(3.36)式应用Gronwall不等式, 并利用估计式$ (3.28) $, $ (3.30) $以及$ (3.33) $, 可得

$ \begin{equation} \|q_{xx}(t)\|_{L^{2}}^2\leq C_{81}. \end{equation} $

$ (3.33) $式和$ (3.37) $式知

$ \begin{equation} \|\widetilde{p}_{xx}(t)\|_{L^{2}}^2+\|q_{xx}(t)\|_{L^{2}}^2+ \int_{0}^{t}\|\widetilde{p}_{xxx}(\tau)\|_{L^{2}}^2\leq C_{82}. \end{equation} $

这里, 常数$ C_{71}, \cdots , C_{82} $都与$ t $有关.引理3.2得证.

参考文献

Adler J .

Chemotaxis in bacteria

Science, 1966, 153 (3737): 708- 716

[本文引用: 1]

Deng C , Li T .

Well-posedness of a 3D parabolic-hyperbolic Keller-Segel system in the Sobolev space framework

Journal of Differential Equations, 2014, 257 (5): 1311- 1332

URL     [本文引用: 1]

Fontelos M A , Friedman A , Hu B .

Mathematical analysis of a model for the initiation of angiogenesis

SIAM Journal on Mathematical Analysis, 2002, 33 (6): 1330- 1355

URL     [本文引用: 1]

Guo J , Xiao J X , Zhao H J , Zhu C J .

Global solutions to a hyperbolic-parabolic coupled system with large initial data

Acta Mathematica Scientia, 2009, 29 (3): 629- 641

DOI:10.1016/S0252-9602(09)60059-X      [本文引用: 1]

Hao C C .

Global well-posedness for a multidimensional chemotaxis model in critical Besov spaces

Zeitschrift für angewandte Mathematik und Physik, 2012, 63 (5): 825- 834

URL     [本文引用: 1]

Hou Q Q , Liu C J , Wang Y G , Wang Z A .

Stability of boundary layers for a viscous hyperbolic system arising from chemotaxis:one-dimensional case

SIAM Journal on Mathematical Analysis, 2018, 50 (3): 3058- 3091

URL     [本文引用: 1]

Hou Q Q , Wang Z A , Zhao K .

Boundary layer problem on a hyperbolic system arising from chemotaxis

Journal of Differential Equations, 2016, 261 (9): 5035- 5070

URL     [本文引用: 1]

Hou Q Q , Wang Z A .

Convergence of boundary layers for the Keller-Segel system with singular sensitivity in the half-plane

Journal de Mathematiques Pures et Appliquees, 2019, 130: 251- 287

URL     [本文引用: 1]

Jin H Y , Li J Y , Wang Z A .

Asymptotic stability of traveling waves of a chemotaxis model with singular sensitivity

Journal of Differential Equations, 2013, 255 (2): 193- 219

URL     [本文引用: 1]

Keller E F , Segel L A .

Initiation of slime mold aggregation viewed as an instability

Journal of Theoretical Biology, 1970, 26 (3): 399- 415

URL     [本文引用: 1]

Keller E F , Segel L A .

Traveling bands of chemotactic bacteria:a theoretical analysis

Journal of Theoretical Biology, 1971, 30 (2): 235- 248

URL     [本文引用: 1]

Levine H A , Sleeman B D .

A system of reaction diffusion equations arising in the theory of reinforced random walks

SIAM Journal on Applied Mathematics, 1997, 57 (3): 683- 730

URL     [本文引用: 1]

Li D , Li T , Zhao K .

On a hyperbolic-parabolic system modeling chemotaxis

Mathematical Models and Methods in Applied Sciences, 2011, 21 (8): 1631- 1650

URL     [本文引用: 1]

Li H , Zhao K .

Initial-boundary value problems for a system of hyperbolic balance laws arising from chemotaxis

Journal of Differential Equations, 2015, 258 (2): 302- 338

URL     [本文引用: 2]

Li T , Pan R H , Zhao K .

Global dynamics of a hyperbolic-parabolic model arising from chemotaxis

SIAM Journal on Applied Mathematics, 2012, 72 (1): 417- 443

URL     [本文引用: 2]

Li T , Wang Z A .

Nonlinear stability of traveling waves to a hyperbolic-parabolic system modeling chemotaxis

SIAM Journal on Applied Mathematics, 2009, 70 (5): 1522- 1541

URL     [本文引用: 1]

Li T , Wang Z A .

Nonlinear stability of large amplitude viscous shock waves of a generalized hyperbolic-parabolic system arising in chemotaxis

Mathematical Models and Methods in Applied Sciences, 2010, 20 (11): 1967- 1998

URL     [本文引用: 1]

Li T , Wang Z A .

Asymptotic nonlinear stability of traveling waves to conservation laws arising from chemotaxis

Journal of Differential Equations, 2011, 250 (3): 1310- 1333

URL     [本文引用: 1]

Li T , Wang Z A .

Steadily propagating waves of a chemotaxis model

Mathematical Biosciences, 2012, 240 (2): 161- 168

URL     [本文引用: 1]

Li J Y , Wang L N , Zhang K J .

Asymptotic stability of a composite wave of two traveling waves to a hyperbolic-parabolic system modeling chemotaxis

Mathematical Methods in the Applied Sciences, 2013, 36 (14): 1862- 1877

URL     [本文引用: 1]

Martinez V R , Wang Z A , Zhao K .

Asymptotic and viscous stability of large-amplitude solutions of a hyperbolic system arising from biology

Indiana University Mathematics Journal, 2018, 67 (4): 1383- 1424

URL     [本文引用: 1]

Ni W M .

Diffusion, cross-diffusion, and their spike-layer steady states

Notices of the Amer Math Soc, 1998, 45 (1): 9- 18

URL     [本文引用: 1]

Othmer H G , Stevens A .

Aggregation, blowup, and collapse:the ABC's of taxis in reinforced random walks

SIAM Journal on Applied Mathematics, 1997, 57 (4): 1044- 1081

URL     [本文引用: 2]

Peng H Y , Wen H Y , Zhu C J .

Global well-posedness and zero diffusion limit of classical solutions to the 3D conservation laws arising in chemotaxis

Zeitschrift für angewandte Mathematik und Physik, 2014, 65 (6): 1167- 1188

URL     [本文引用: 1]

Peng Y , Xiang Z .

Global solutions to the coupled chemotaxis-fluids system in a 3D unbounded domain with boundary

Mathematical Models and Methods in Applied Sciences, 2018, 28: 869- 920

URL     [本文引用: 1]

Peng Y , Xiang Z .

Global existence and convergence rates to a chemotaxis-fluids system with mixed boundary conditions

Journal of Differential Equations, 2019, 267: 1277- 1321

URL     [本文引用: 1]

Rebholz L G , Wang D H , Wang Z A , et al.

Initial boundary value problems for a system of parabolic conservation laws arising from chemotaxis in multi-dimensions

Discrete and Continuous Dynamical Systems-Series A, 2019, 39 (7): 3789- 3838

URL     [本文引用: 1]

Tao Y , Wang L , Wang Z A .

Large-time behavior of a parabolic-parabolic chemotaxis model with logarithmic sensitivity in one dimension

Discrete and Continuous Dynamical Systems-Series B, 2013, 18 (3): 821- 845

URL     [本文引用: 1]

Wang Y, Winkler M, Xiang Z. The fast signal diffusion limit in Keller-Segel (-fluid) systems. Calculus of Variations and Partial Differential Equations, 2019, 58, Article number: 196

[本文引用: 1]

Tian Y , Xiang Z .

Global solutions to a 3D chemotaxis-Stokes system with nonlinear cell diffusion and Robin signal boundary condition

Journal of Differential Equations, 2020, 269 (3): 2012- 2056

URL     [本文引用: 1]

Wang Z A , Xiang Z , Yu P .

Asymptotic dynamics on a singular chemotaxis system modeling onset of tumor angiogenesis

Journal of Differential Equations, 2016, 260 (3): 2225- 2258

URL     [本文引用: 1]

Wang Z A , Zhao K .

Global dynamics and diffusion limit of a one-dimensional repulsive chemotaxis model

Communications on Pure and Applied Analysis, 2013, 12 (6): 3027- 3046

URL     [本文引用: 2]

Wu C , Xiang Z .

The small-convection limit in a two-dimensional Keller-Segel-Navier-Stokes system

Journal of Differential Equations, 2019, 267: 938- 978

URL     [本文引用: 1]

Wu C , Xiang Z .

Asymptotic dynamics on a chemotaxis-Navier-Stokes system with nonlinear diffusion and inhomogeneous boundary conditions

Mathematical Models and Methods in Applied Sciences, 2020, 30 (7): 1325- 1374

URL     [本文引用: 1]

Zhang M , Zhu C J .

Global existence of solutions to a hyperbolic-parabolic system

Proceedings of the American Mathematical Society, 2007, 135 (4): 1017- 1027

URL     [本文引用: 1]

Zhu N , Liu Z R , Martinez V R , Zhao K .

Global Cauchy problem of a system of parabolic conservation laws arising from a Keller-Segel type chemotaxis model

SIAM Journal on Mathematical Analysis, 2018, 50 (5): 5380- 5425

URL     [本文引用: 1]

/