数学物理学报, 2024, 44(1): 120-132

温度相关双扩散模型在半无穷柱体上的结构稳定性

李远飞,*, 李丹丹, 石金诚

广州华商学院应用数学系 广州 511300

Structural Stability of Temperature Dependent Double Diffusion Model on a Semi-infinite Cylinder

Li Yuanfei,*, Li Dandan, Shi Jincheng

Department of Apllied Mathematics, Guangzhou Huashang College, Guangzhou 511300

通讯作者: 李远飞, E-mail:liqfd@163.com

收稿日期: 2022-11-30   修回日期: 2023-06-20  

基金资助: 广州华商学院科研团队资助(2021HSKT01)
国家自然科学基金(11371175)

Received: 2022-11-30   Revised: 2023-06-20  

Fund supported: Research Team Project of Guangzhou Huashang University(2021HSKT01)
National Natural Science Foundation of China(11371175)

摘要

该文研究了定义在一个半无穷柱体上的温度相关双扩散模型的简化形式. 利用先验估计和加权能量分析法, 证明了当边界条件满足一定的约束条件时模型的解随空间变量指数式衰减. 利用解的先验界和衰减性结果, 得到了解对相互作用系数的结构稳定性.

关键词: 双扩散模型; 空间衰减; 结构稳定性; 相互作用系数

Abstract

A simplified temperature dependent double diffusion model defined on a semi-infinite cylinder is studied. By using a prior estimates and weighted energy analysis, it is proved that the solution of the model decays exponentially with the space variable when the boundary conditions satisfy certain constraints. The structural stability of the solution to the interaction coefficient is obtained by using the prior bounds and decay result of the solution.

Keywords: Double diffusion model; Spatial attenuation; Structural stability; Interaction coefficient

PDF (543KB) 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

李远飞, 李丹丹, 石金诚. 温度相关双扩散模型在半无穷柱体上的结构稳定性[J]. 数学物理学报, 2024, 44(1): 120-132

Li Yuanfei, Li Dandan, Shi Jincheng. Structural Stability of Temperature Dependent Double Diffusion Model on a Semi-infinite Cylinder[J]. Acta Mathematica Scientia, 2024, 44(1): 120-132

1 引言

利用 Nield[1,2]以及Nield和Kuznetsov[3]的一般理论,Falsaperla等[4]导出了具有单一温度的双扩散多孔介质中的热对流方程.

在不丧失对所研究模型的连续依赖性问题的一般性的情况下,Franchi等[5]把文献[4]中的双扩散模型简化成了如下形式

$\mu u_{i}+a(u_i-v_i)=-p_{,i}+g_iT,\ u_{i,i}=0$
$\nu v_{i}-a(u_i-v_i)=-q_{,i}+g_iT, \ v_{i,i}=0$
$T_{,t}+\alpha(u_i+v_i)T_{,i}=\Delta T,$

其中$u_i, v_i$表示流体速度,$T$表示流体的温度,$p, q$表示压强,$a$是相互作用系数,$g_i$是重力函数, 不失一般性, 假设$g_ig_i\leq1$,$\alpha$是流体的热膨胀系数,$\mu, \nu$是饱和流体的动态粘度. 假设模型 (1.1)-(1.3) 定义在有界区域$\Omega\subset\mathbb{R}^3$上, 并假设 (1.1)-(1.3) 式的解在区域上的边界上满足

$u_in_i=v_in_i=0,\ T=h(x, t),\ \text{在}\ \partial\Omega\times\{t>0\},$

或者

$\frac{\partial T}{\partial n}+\kappa T=F(x, t),\ \text{在}\ \partial\Omega\times\{t>0\},$

其中$n=(n_1, n_2, n_3)$是$\partial\Omega$上的单位外法向量,$h(x, t), F(x, t)$是大于零的给定函数,$\kappa>0$是牛顿冷却定系数. 利用最大值原理,Franchi 等[5]建立了溶液对与宏观孔隙度和微观孔隙度相关的速度之间的相互作用系数的连续依赖性, 分析了粘度系数的连续依赖性, 并且建立了解对牛顿冷却定系数的连续依赖关系.

本文的工作是把有界区域$\Omega$上的连续依赖性研究推广到半无穷区域$\mathcal{B}$上, 其中

$\mathcal{B}=\Big\{(x_1, x_2, x_3)|(x_1,x_2)\in D,\ x_3\geq0\Big\},$

$D\subset\mathbb{R}^2$是有界充分光滑的区域. 在过去的研究中, 定义在有界区域上的各种类型的偏微分方程的结构稳定性得到了充分关注, 出现了大量成果[5-9]. 半无穷柱体上的结构稳定性研究也开始受到关注. Li 和 Lin 研究半无穷柱体上 Brinkman-Forchheimer 方程的结构稳定性. 与文献[6]相比, 方程 (1.1) 和 (1.2) 中没有拉普拉斯项, 因此我们无法用$\nabla u$的$L^2$范数来控制$u$的$L^2$范数. 与文献[5]相比, 本文不仅需要考虑时间因素而且需要考虑空间变量因素, 因此文献[5,6]中的方法并不能直接应用到本文中来. 本文利用最大值原理和加权能量估计的方法, 研究方程 (1.1)-(1.3) 的解对互作用系数的连续依赖性.

令$D_z$表示$\mathcal{B}$在$x_3=z$处的截面, 即

$D_z=\Big\{(x_1,x_2,x_3)|(x_1, x_2)\in D,\ x_3=z\geq0\Big\}.$

令$\mathcal{B}_z$表示$\mathcal{B}$的子集, 即

$\mathcal{B}_z=\Big\{(x_1,x_2,x_3)|(x_1, x_2)\in D,\ x_3\geq z\geq0\Big\}.$

本文采取用逗号表示求导, 重复希腊字母表示从 1 至 2 求和, 重复英文字母表示从 1 至 3 求和. 例如:$u_{,\alpha}u_{,\alpha}=\sum\limits_{\alpha=1}^2\Big(\frac{\partial u}{\partial x_\alpha}\Big)^2,\ u_{i,i}=\sum\limits_{i=1}^3\frac{\partial u_i}{\partial x_i}$.在柱体的有限端和侧面上, 方程 (1.1)-(1.3) 的解满足以下初边值条件

$u_{i}=f_i, \ v_i=h_i,\ T=H,\ (x_1, x_2, t)\in D\times\{t>0\},$
$u_i=v_i=0,\ T=0, \ (x, t)\in\partial D\times\{t>0\},$
$T(x, 0)=0,\ x \in \mathcal{B},$
$u_i, v_i, |T|=O(1), |\nabla T|, |q|, |p|=o(x_3^{-1}), \ \text{当}\ x_3\rightarrow\infty,$

其中$\partial D$表示$D$的边界,$f_i(x_1, x_2, t), h_i(x_1, x_2, t)$和$H(x_1, x_2, t)$为已知函数.

2 先验估计

首先给出以下引理.

引理 2.1[10-12] 设$H\in L^\infty$, 则

$\sup_{[t]}||T||_\infty\leq\max\Big\{\sup_{[t]}H_\infty\Big\}\doteq T_m.$

引理 2.2[13,14] 如果$\phi\big|_{\partial D}=0,$则

$\lambda\int_D\phi^2{\rm d}A\leq\int_D\phi_{,\alpha}\phi_{,\alpha} {\rm d}A,$

其中$\lambda$是$\vartheta_{,\alpha\alpha}+\lambda\vartheta=0, \ \text{在}\ D,\ \vartheta=0,\ \text{在}\ \partial D$的第一特征值.

引理 2.3[13,14] 如果$\phi$是$D$上的 Dirichlet 可积函数,$\int_D\phi {\rm d}x=0$, 则存在向量函数$\psi=(\psi_1, \psi_2)$满足$\psi_{\alpha, \alpha}=\phi,\ \text{在}\ D, \ \psi_\alpha=0, \ \text{在}\ \partial D,$以及大于 0 的常数$k_1$满足

$\int_D \psi_{\alpha, \beta}\psi_{\alpha, \beta}{\rm d}A\leq k_1\int_D (\psi_{\alpha, \alpha})^2{\rm d}A.$

引理 2.4[15,16] 如果$\phi$是一个 Dirichlet 可积函数且$\phi\big|_{\partial D}=0, \phi\rightarrow\infty$(当$x_3\rightarrow\infty$时),

$\int_{\mathcal{B}_z}|\phi|^4{\rm d}x\leq k_2\Big(\int_{\mathcal{B}_z}\phi_{,j}\phi_{,j}{\rm d}x\Big)^2,\ \text{其中}\ k_2>0.$

接下来, 我们推导方程 (1.1)-(1.7) 解的先验估计.

引理 2.5 记$\sigma(x, t)=H(x_1, x_2, t){\rm e}^{-\gamma_1 x_3}, \gamma_1>0$. 设$H\in L^\infty(\mathcal{B}), f_i, h_i\in H^1(\mathcal{B}),$则

$\frac{1}{2}\int_\mathcal{B}T^2 {\rm d}x+\int_0^t\int_\mathcal{B}T_{,i}T_{,i}{\rm d}x{\rm d}\eta\leq m_2(t)+\varepsilon_1\int_0^t\int_\mathcal{B}(\mu u_iu_i+\nu v_iv_i){\rm d}x{\rm d}\eta,$

其中

$\begin{align*}m_2(t)=\ &2\int_0^t\int_\mathcal{B}\sigma_{,i}\sigma_{,i}{\rm d}x{\rm d}\eta+2\int_\mathcal{B}\sigma^2 {\rm d}x+\frac{2}{\lambda}\int_0^t\int_\mathcal{B}\sigma_{,\eta}^2{\rm d}x{\rm d}\eta\nonumber\\&+\frac{1}{\varepsilon_1}\alpha^2T_m^2\Big(\frac{1}{\mu}+\frac{1}{\nu}\Big)\int_0^t\int_\mathcal{B}\sigma_{,i}\sigma_{,i}{\rm d}x{\rm d}\eta-\alpha\int_0^t\int_D(f_3+h_3)H^2{\rm d}A{\rm d}\eta.\end{align*}$

显然$\sigma(x, t)$和$T$具有相同的初边值条件. 利用散度定理和方程 (1.3), 可得

$\begin{matrix}-\int_0^t\int_DT_{,3}T{\rm d}A{\rm d}\eta&=-\int_0^t\int_DT_{,3}\sigma {\rm d}A{\rm d}\eta=\int_0^t\int_\mathcal{B}(T_{,i}\sigma)_{,i}{\rm d}x{\rm d}\eta\nonumber\\&=\int_0^t\int_\mathcal{B}T_{,i}\sigma_{,i}{\rm d}x{\rm d}\eta+\int_0^t\int_\mathcal{B}\sigma\Big(T_{,\eta}+\alpha(u_i+v_i)T_{,i}\Big){\rm d}x{\rm d}\eta\nonumber\\&=\int_0^t\int_\mathcal{B}T_{,i}\sigma_{,i}{\rm d}x{\rm d}\eta+\int_\mathcal{B}T\sigma {\rm d}x-\int_0^t\int_\mathcal{B}T\sigma_{,\eta}{\rm d}x{\rm d}\eta\nonumber\\& -\alpha\int_0^t\int_D(f_3+h_3)H^2{\rm d}A{\rm d}\eta-\alpha\int_0^t\int_\mathcal{B}T(u_i+v_i)\sigma_{,i}{\rm d}x{\rm d}\eta.\end{matrix}$

利用 Hölder 不等式, 引理 2.1, 引理 2.2 和 Young 不等式, 可得

$\int_0^t\int_\mathcal{B}T_{,i}\sigma_{,i}{\rm d}x{\rm d}\eta \leq\frac{1}{4}\int_0^t\int_\mathcal{B}T_{,i}T_{,i}{\rm d}x{\rm d}\eta+\int_0^t\int_\mathcal{B}\sigma_{,i}\sigma_{,i}{\rm d}x{\rm d}\eta,$
$\int_\mathcal{B}T\sigma {\rm d}x\leq\frac{1}{4}\int_\mathcal{B}T^2{\rm d}x+\int_\mathcal{B}\sigma^2 {\rm d}x,$
$-\int_0^t\int_\mathcal{B}T\sigma_{,\eta}{\rm d}x{\rm d}\eta\leq\frac{1}{4}\int_0^t\int_\mathcal{B}T_{,i}T_{,i}{\rm d}x{\rm d}\eta+\frac{1}{\lambda}\int_0^t\int_\mathcal{B}\sigma_{,\eta}^2{\rm d}x{\rm d}\eta,$
$-\alpha\int_0^t\int_\mathcal{B}T(u_i+v_i)\sigma_{,i}{\rm d}x{\rm d}\eta\leq\frac{1}{2}\varepsilon_1\int_0^t\int_\mathcal{B}(\mu u_iu_i+\nu v_iv_i){\rm d}x{\rm d}\eta \nonumber\\ \ +\frac{1}{2\varepsilon_1}\alpha^2T_m^2\Big(\frac{1}{\mu}+\frac{1}{\nu}\Big)\int_0^t\int_\mathcal{B}\sigma_{,i}\sigma_{,i}{\rm d}x{\rm d}\eta,$

其中$\varepsilon_1$是大于 0 的任意常数. 把 (2.2)-(2.5) 式代入到 (2.1) 式, 可得

$\begin{matrix}-\int_0^t\int_DT_{,3}T{\rm d}A{\rm d}\eta\leq\ & m_1(t)+\frac{1}{2}\int_0^t\int_\mathcal{B}T_{,i}T_{,i}{\rm d}x{\rm d}\eta+\frac{1}{4}\int_\mathcal{B}T^2{\rm d}x {\rm d}x\nonumber\\&+\frac{1}{2}\varepsilon_1\int_0^t\int_\mathcal{B}(\mu u_iu_i+\nu v_iv_i){\rm d}x{\rm d}\eta,\end{matrix}$

其中

$\begin{align*}m_1(t)=\ &\int_0^t\int_\mathcal{B}\sigma_{,i}\sigma_{,i}{\rm d}x{\rm d}\eta+\int_\mathcal{B}\sigma^2 {\rm d}x+\frac{1}{\lambda}\int_0^t\int_\mathcal{B}\sigma_{,\eta}^2{\rm d}x{\rm d}\eta\nonumber\\&+\frac{1}{2\varepsilon_1}\alpha^2T_m^2\Big(\frac{1}{\mu}+\frac{1}{\nu}\Big)\int_0^t\int_\mathcal{B}\sigma_{,i}\sigma_{,i}{\rm d}x{\rm d}\eta-\alpha\int_0^t\int_D(f_3+h_3)H^2{\rm d}A{\rm d}\eta.\end{align*}$

在方程 (1.3) 的两边乘以$T$并在$R\times[t]$上积分, 可得

$\int_0^t\int_\mathcal{B}\Big[T_{,\eta}+\alpha(u_i+v_i)T_{,i}-\Delta T\Big]T{\rm d}x{\rm d}\eta=0.$

利用散度定理和边界条件 (1.4)-(1.7), 可得

$\begin{matrix}\frac{1}{2}\int_\mathcal{B}T^2 {\rm d}x+\int_0^t\int_\mathcal{B}T_{,i}T_{,i}{\rm d}x{\rm d}\eta=-\int_0^t\int_DT_{,3}T{\rm d}A{\rm d}\eta+\frac{1}{2}\alpha\int_0^t\int_D(f_3+h_3)H^2{\rm d}A{\rm d}\eta.\end{matrix}$

联合 (2.6) 和 (2.7) 式, 可以得到引理 2.5. 证毕.

引理 2.6 如果$\int_Df_3{\rm d}A=0$,$\int_Dh_3{\rm d}A=0$,$H\in L^\infty(\mathcal{B}), f_i, h_i\in H^1(R\mathcal{B}),$则

$\begin{align}\int_\mathcal{B}T^2 {\rm d}x+\int_0^t\int_\mathcal{B}T_{,i}T_{,i}{\rm d}x{\rm d}\eta&\leq m_3(t),\nonumber\\\int_{\mathcal{B}}\big(\mu u_iu_i+\nu v_iv_i\big){\rm d}x&\leq m_4(t),\nonumber\end{align}$

其中$m_3(t)$和$m_4(t)$是大于 0 的已知函数.

利用方程(1.1)和(1.2), 可得

$\begin{matrix}&\int_{\mathcal{B}_z}\big(\mu u_iu_i+\nu v_iv_i\big){\rm d}x+a\int_{\mathcal{B}_z}(u_i-v_i)(u_i-v_i){\rm d}x\nonumber\\=\ &\int_{D_z}pu_3{\rm d}A{\rm d}\eta+\int_{D_z}qu_3{\rm d}A+\int_{\mathcal{B}_z}g_iTu_i{\rm d}x{\rm d}\eta+\int_{\mathcal{B}_z}g_iTv_i{\rm d}x.\end{matrix}$

注意到

$\begin{align*}\int_{D_z}u_{3}{\rm d}A=\int_{D}u_{3}{\rm d}A+\int_{0}^{z}\int_{D_\xi}\frac{\partial u_3 }{\partial x_3}{\rm d}A{\rm d}\xi=\int_{D}f_{3}{\rm d}A-\int_{0}^{z}\int_{D_\xi}u_{\alpha,\alpha}{\rm d}A{\rm d}\xi=\int_{D}f_{3}{\rm d}A.\nonumber\end{align*}$

由于$\int_Df_3{\rm d}A=0$, 所以$\int_{D_z}u_{3}{\rm d}A=0$. 利用引理 2.3, 可知存在向量函数$\psi=(\psi_1, \psi_2)$满足

$\begin{align*}\psi_{\alpha, \alpha}=u_3,\ \text{在}\ D, \ \psi_\alpha=0, \ \text{在}\ \partial D.\end{align*}$

利用引理 2.2 和引理 2.3, 可得

$\int_{D_z}\psi_{\alpha}\psi_{\alpha} {\rm d}A\leq\frac{1}{\lambda}\int_{D_z}\psi_{\alpha, \beta}\psi_{\alpha, \beta} {\rm d}A\leq\frac{k_1}{\lambda}\int_{D_z}(\psi_{\alpha, \alpha})^2{\rm d}A=\frac{k_1}{\lambda}\int_{D_z}u_3^2{\rm d}A.$

所以

$\begin{matrix}\int_{D_z}pu_3{\rm d}A&=\int_{D_z}p\psi_{\alpha, \alpha}{\rm d}A=-\int_{D_z}p_{,\alpha}\psi_{\alpha}{\rm d}A=\int_{D_z}\Big(\mu u_\alpha+a(u_\alpha-v_\alpha)-g_\alpha T\Big)\psi_{\alpha}{\rm d}A\nonumber\\&\leq\mu\Big(\int_{D_z}u_\alpha u_\alpha {\rm d}A\Big)^\frac{1}{2}\Big(\int_{D_z}\psi_{\alpha}\psi_{\alpha} {\rm d}A\Big)^\frac{1}{2}+\Big(\int_{D_z}T^2{\rm d}A\Big)^\frac{1}{2}\Big(\int_{D_z}\psi_{\alpha}\psi_{\alpha} {\rm d}A\Big)^\frac{1}{2}\nonumber\\& \ +a\Big(\int_{D_z}(u_\alpha-v_\alpha)(u_\alpha-v_\alpha) {\rm d}A\Big)^\frac{1}{2}\Big(\int_{D_z}\psi_{\alpha}\psi_{\alpha} {\rm d}A\Big)^\frac{1}{2}\nonumber\\&\leq\frac{\mu\sqrt{k_1}}{2\sqrt{\lambda}}\int_{D_z}u_iu_i {\rm d}A+\frac{\sqrt{k_1}}{\sqrt{\mu\lambda}}\Big(\int_{D_z}T^2{\rm d}A{\rm d}\eta+\mu\int_{D_z}u_3^2 {\rm d}A\Big)\nonumber\\& \ +\frac{\sqrt{ak_1}}{\sqrt{\mu\lambda}}\Big[a\int_{D_z}(u_\alpha-v_\alpha)(u_\alpha-v_\alpha) {\rm d}A+\mu\int_{D_z}u_3^2 {\rm d}A\Big].\end{matrix}$

类似地, 有

$\begin{matrix}\int_{D_z}qu_3{\rm d}A\leq\ &\frac{\nu\sqrt{k_1}}{2\sqrt{\lambda}}\int_{D_z}v_iv_i {\rm d}A+\frac{\sqrt{k_1}}{\sqrt{\nu\lambda}}\Big(\int_{D_z}T^2{\rm d}A+\nu\int_{D_z}v_3^2 {\rm d}A\Big)\nonumber\\&+\frac{\sqrt{ak_1}}{\sqrt{\nu\lambda}}\Big[a\int_{D_z}(u_\alpha-v_\alpha)(u_\alpha-v_\alpha) {\rm d}A+\nu\int_{D_z}v_3^2 {\rm d}A\Big].\end{matrix}$

利用 Hölder 不等式和 Young 不等式, 可得

$\begin{matrix}\int_{\mathcal{B}_z}g_iTu_i{\rm d}x&\leq\frac{1}{2}\mu\int_{\mathcal{B}_z}u_iu_i{\rm d}x+\frac{1}{2\mu}\int_{\mathcal{B}_z}T_{,\alpha}T_{,\alpha}{\rm d}x,\end{matrix}$
$\begin{matrix}\int_{\mathcal{B}_z}g_iTv_i{\rm d}x&\leq\frac{1}{2}\nu\int_{\mathcal{B}_z}v_iv_i{\rm d}x+\frac{1}{2\nu}\int_{\mathcal{B}_z}T_{,\alpha}T_{,\alpha}{\rm d}x.\end{matrix}$

把(2.9)-(2.12)式代入到(2.8)式, 可得

$\begin{matrix}&\frac{1}{2}\int_{\mathcal{B}_z}\big(\mu u_iu_i+\nu v_iv_i\big){\rm d}x+a\int_{\mathcal{B}_z}(u_i-v_i)(u_i-v_i){\rm d}x\nonumber\\\leq\ &\Big(\frac{\sqrt{k_1}}{2\sqrt{\lambda}}+\frac{\sqrt{ak_1}}{\sqrt{\mu\lambda}}+\frac{\sqrt{k_1}}{\sqrt{\mu\lambda}}\Big)\cdot\mu\int_{D_z}u_iu_i {\rm d}A+\Big(\frac{\sqrt{k_1}}{2\sqrt{\lambda}}+\frac{\sqrt{ak_1}}{\sqrt{\nu\lambda}}+\frac{\sqrt{k_1}}{\sqrt{\nu\lambda}}\Big)\cdot\nu\int_{D_z}v_iv_i {\rm d}A\nonumber\\& +\Big(\frac{\sqrt{ak_1}}{\sqrt{\mu\lambda}}+\frac{\sqrt{ak_1}}{\sqrt{\nu\lambda}}\Big)\cdot a\int_{D_z}(u_\alpha-v_\alpha)(u_\alpha-v_\alpha) {\rm d}A\nonumber\\&+\Big(\frac{\sqrt{k_1}}{\sqrt{\mu\lambda}}+\frac{\sqrt{k_1}}{\sqrt{\nu\lambda}}\Big)\int_{D_z}T^2{\rm d}A+\Big(\frac{1}{2\mu}+\frac{1}{2\nu}\Big)\int_{\mathcal{B}_z}T^2{\rm d}x.\end{matrix}$

在 (2.13) 式中取$z=0$并利用 (1.4) 式, 可得

$\begin{matrix}\frac{1}{2}\int_{\mathcal{B}}\big(\mu u_iu_i+\nu v_iv_i\big){\rm d}x&+a\int_{\mathcal{B}}(u_i-v_i)(u_i-v_i){\rm d}x\leq n_1(t)+\Big(\frac{1}{2\mu}+\frac{1}{2\nu}\Big)\int_{\mathcal{B}}T^2{\rm d}x,\end{matrix}$

其中

$\begin{align*}n_1(t)=\ &\Big(\frac{\sqrt{k_1}}{2\sqrt{\lambda}}+\frac{\sqrt{ak_1}}{\sqrt{\mu\lambda}}+\frac{\sqrt{k_1}}{\sqrt{\mu\lambda}}\Big)\cdot\mu\int_{D}f_if_i {\rm d}A+\Big(\frac{\sqrt{k_1}}{2\sqrt{\lambda}}+\frac{\sqrt{ak_1}}{\sqrt{\nu\lambda}}+\frac{\sqrt{k_1}}{\sqrt{\nu\lambda}}\Big)\cdot\nu\int_{D}h_ih_i {\rm d}A\nonumber\\& +\Big(\frac{\sqrt{ak_1}}{\sqrt{\mu\lambda}}+\frac{\sqrt{ak_1}}{\sqrt{\nu\lambda}}\Big)\cdot a\int_{D}(f_\alpha-h_\alpha)(f_\alpha-h_\alpha) {\rm d}A+\Big(\frac{\sqrt{k_1}}{\sqrt{\mu\lambda}}+\frac{\sqrt{k_1}}{\sqrt{\nu\lambda}}\Big)\int_{D}H^2{\rm d}A.\end{align*}$

对 (2.14) 式从 0 到$t$积分并利用引理 2.2, 可得

$\begin{matrix}&\frac{1}{2}\int_0^t\int_{\mathcal{B}}\big(\mu u_iu_i+\nu v_iv_i\big){\rm d}x{\rm d}\eta+a\int_0^t\int_{\mathcal{B}}(u_i-v_i)(u_i-v_i){\rm d}x{\rm d}\eta\nonumber\\\leq & \int_0^tn_1(\eta){\rm d}\eta+\Big(\frac{1}{2\mu\lambda}+\frac{1}{2\nu\lambda}\Big)\int_0^t\int_{\mathcal{B}}T_{,\alpha}T_{,\alpha}{\rm d}x{\rm d}\eta.\end{matrix}$

结合引理 2.5 和 2.15 式并取$\varepsilon_1=\frac{\mu\nu\lambda}{2(\mu+\nu)}$, 可得

$\int_\mathcal{B}T^2 {\rm d}x+\int_0^t\int_\mathcal{B}T_{,i}T_{,i}{\rm d}x{\rm d}\eta\leq2m_2(t)+4\varepsilon_1\int_0^tn_1(\eta){\rm d}\eta,$
$\frac{1}{2}\int_0^t\int_{\mathcal{B}}\big(\mu u_iu_i+\nu v_iv_i\big){\rm d}x{\rm d}\eta+a\int_0^t\int_{\mathcal{B}}(u_i-v_i)(u_i-v_i){\rm d}x{\rm d}\eta \nonumber\\ \leq \int_0^tn_1(\eta){\rm d}\eta+\Big(\frac{1}{\mu\lambda}+\frac{1}{\nu\lambda}\Big)\Big[m_2(t)+2\varepsilon_1\int_0^tn_1(\eta){\rm d}\eta\Big].$

联合(2.14)和(2.16)式, 可得

$\begin{matrix}\int_{\mathcal{B}}\big(\mu u_iu_i+\nu v_iv_i\big){\rm d}x\leq 2n_1(t)+2\Big(\frac{1}{\mu}+\frac{1}{\nu}\Big)\Big[m_2(t)+2\varepsilon_1\int_0^tn_1(\eta){\rm d}\eta\Big].\end{matrix}$

在(2.16)和(2.18)式中分别取

$ m_3(t)=2m_2(t)+4\varepsilon_1\int_0^tn_1(\eta){\rm d}\eta, m_4(t)=2n_1(t)+2\Big(\frac{1}{\mu}+\frac{1}{\nu}\Big)\Big[m_2(t)+2\varepsilon_1\int_0^tn_1(\eta){\rm d}\eta\Big], $

可以完成定理 2.6 的证明. 证毕.

3 空间衰减结果

利用上一节获得的先验估计, 我们可得以下空间衰减性结果.

定理 3.1 设$\int_Df_3{\rm d}A=0$,$\int_Dh_3{\rm d}A=0$,$H\in L^\infty(\mathcal{B}), f_i, h_i\in H^1(\mathcal{B}),$则方程 (1.1)-(1.7) 的解随$z\rightarrow\infty$呈指数式衰减. 具体地,

$\begin{align*}&\frac{1}{2}\int_0^t\int_{\mathcal{B}_z}\big(\mu u_iu_i+\nu v_iv_i\big){\rm d}x{\rm d}\eta+a\int_0^t\int_{\mathcal{B}_z}(u_i-v_i)(u_i-v_i){\rm d}x{\rm d}\eta\nonumber\\&+\frac{1}{2}\int_{\mathcal{B}_z}(\xi-z)T^2{\rm d}x+\int_0^t\int_{\mathcal{B}_z}(\xi-z)T_{,i}T_{,i}{\rm d}x{\rm d}\eta\nonumber\leq b_2(t){\rm e}^{-b_1z},\end{align*}$

其中$b_1$是大于0的常数,$b_2(t)$是大于0的已知函数以及$\xi>z$.

用$(\xi-z)T$乘以方程 (1.3), 并在$R_z\times[t]$上积分可得

$\begin{matrix}& \frac{1}{2}\int_{\mathcal{B}_z}(\xi-z)T^2{\rm d}x+\int_0^t\int_{\mathcal{B}_z}(\xi-z)T_{,i}T_{,i}{\rm d}x{\rm d}\eta \\=\ & \frac{1}{2}\int_0^t\int_{\mathcal{B}_z}(u_3+v_3)T^2{\rm d}x{\rm d}\eta-\int_0^t\int_{\mathcal{B}_z}T_{,3}T{\rm d}x{\rm d}\eta\nonumber\\\leq\ & \frac{1}{2\sqrt{\mu\lambda}}T_m\Big(\int_0^t\int_{\mathcal{B}_z}\mu u_3^2{\rm d}x{\rm d}\eta\Big)^\frac{1}{2}\Big(\int_0^t\int_{\mathcal{B}_z}T_{,\alpha}T_{,\alpha}{\rm d}x{\rm d}\eta\Big)^\frac{1}{2}\nonumber\\& +\frac{1}{2\sqrt{\nu\lambda}}T_m\Big(\int_0^t\int_{\mathcal{B}_z}\nu v^2_3{\rm d}x{\rm d}\eta\Big)^\frac{1}{2}\Big(\int_0^t\int_{\mathcal{B}_z}T_{,\alpha}T_{,\alpha}{\rm d}x{\rm d}\eta\Big)^\frac{1}{2}\nonumber\\& +\frac{1}{\sqrt{\lambda}}\Big(\int_0^t\int_{\mathcal{B}_z} T^2_{,3}{\rm d}x{\rm d}\eta\Big)^\frac{1}{2}\Big(\int_0^t\int_{\mathcal{B}_z}T_{,\alpha}T_{,\alpha}{\rm d}x{\rm d}\eta\Big)^\frac{1}{2}\nonumber\\\leq\ & \frac{1}{4\sqrt{\lambda}}T_m\int_0^t\int_{\mathcal{B}_z}(\mu u_3^2+\nu v_3^2){\rm d}x{\rm d}\eta\nonumber\\&+\Big(\frac{1}{4\mu\sqrt{\lambda}}T_m+\frac{1}{4\nu\sqrt{\lambda}}T_m+\frac{1}{2\sqrt{\lambda}}\Big)\int_0^t\int_{\mathcal{B}_z}T_{,i}T_{,i}{\rm d}x{\rm d}\eta.\end{matrix}$

注意到

$\begin{align}\int_{D_z}T^2{\rm d}A=-2\int_{\mathcal{B}_z}T_{,3}T{\rm d}x& \leq\frac{2}{\sqrt{\lambda}}\Big(\int_{\mathcal{B}_z}T_{,3}^2{\rm d}x\Big)^\frac{1}{2}\Big(\int_{\mathcal{B}_z}T_{,\alpha}T_{,\alpha}{\rm d}x\Big)^\frac{1}{2}\leq\frac{1}{\sqrt{\lambda}}\int_{\mathcal{B}_z}T_{,i}T_{,i}{\rm d}x,\nonumber\end{align}$

则(2.13)式可以写为

$\begin{matrix}& \frac{1}{2}\int_{\mathcal{B}_z}\big(\mu u_iu_i+\nu v_iv_i\big){\rm d}x+a\int_{\mathcal{B}_z}(u_i-v_i)(u_i-v_i){\rm d}x\nonumber\\\leq\ & \Big(\frac{\sqrt{k_1}}{2\sqrt{\lambda}}+\frac{\sqrt{ak_1}}{\sqrt{\mu\lambda}}+\frac{\sqrt{k_1}}{\sqrt{\mu\lambda}}\Big)\cdot\mu\int_{D_z}u_iu_i {\rm d}A+\Big(\frac{\sqrt{k_1}}{2\sqrt{\lambda}}+\frac{\sqrt{ak_1}}{\sqrt{\nu\lambda}}+\frac{\sqrt{k_1}}{\sqrt{\nu\lambda}}\Big)\cdot\nu\int_{D_z}v_iv_i {\rm d}A\nonumber\\& +\Big(\frac{\sqrt{ak_1}}{\sqrt{\mu\lambda}}+\frac{\sqrt{ak_1}}{\sqrt{\nu\lambda}}\Big)\cdot a\int_{D_z}(u_\alpha-v_\alpha)(u_\alpha-v_\alpha) {\rm d}A\nonumber\\& +\Big(\frac{\sqrt{k_1}}{\sqrt{\mu}\lambda}+\frac{\sqrt{k_1}}{\sqrt{\nu}\lambda}\Big)\int_{\mathcal{B}_z}T_{,i}T_{,i}{\rm d}x+\Big(\frac{1}{2\lambda\mu}+\frac{1}{2\lambda\nu}\Big)\int_{\mathcal{B}_z}T_{,i}T_{,i}{\rm d}x.\end{matrix}$

建立"能量"函数

$\begin{matrix}E(z, t)=\ &\frac{1}{2}\int_0^t\int_{\mathcal{B}_z}\big(\mu u_iu_i+\nu v_iv_i\big){\rm d}x{\rm d}\eta+a\int_0^t\int_{\mathcal{B}_z}(u_i-v_i)(u_i-v_i){\rm d}x{\rm d}\eta\nonumber\\&+\frac{1}{2}\int_{\mathcal{B}_z}(\xi-z)T^2{\rm d}x+\int_0^t\int_{\mathcal{B}_z}(\xi-z)T_{,i}T_{,i}{\rm d}x{\rm d}\eta.\end{matrix}$

对(3.3)式微分, 可得

$\begin{matrix}-\frac{\partial}{\partial z}E(z, t)=\ &\frac{1}{2}\int_0^t\int_{D_z}\big(\mu u_iu_i+\nu v_iv_i\big){\rm d}A{\rm d}\eta+a\int_0^t\int_{D_z}(u_i-v_i)(u_i-v_i){\rm d}A{\rm d}\eta\nonumber\\&+\frac{1}{2}\int_{\mathcal{B}_z}T^2{\rm d}x+\int_0^t\int_{\mathcal{B}_z}T_{,i}T_{,i}{\rm d}x{\rm d}\eta.\end{matrix}$

若在(3.4)式中取$z=0$, 利用条件 (1.4) 和引理 2.6, 可得

$\begin{matrix}-\frac{\partial}{\partial z}E(0, t)\leq\frac{1}{2}\int_0^t\int_{D}\big(\mu f_if_i+\nu h_ih_i\big){\rm d}A{\rm d}\eta+a\int_0^t\int_{D_z}(f_i-h_i)(f_i-h_i){\rm d}A{\rm d}\eta+m_3(t).\end{matrix}$

结合(3.1)式和(2.13)式, 并注意到(3.4)式可得

$\begin{matrix}E(z, t)\leq \frac{1}{b_1}\Big(-\frac{\partial}{\partial z}E(z, t)\Big),\end{matrix}$

其中

$\begin{align*}\frac{1}{b_1}=\max\bigg\{\max\bigg\{&\frac{\sqrt{k_1}}{2\sqrt{\lambda}}+\frac{\sqrt{ak_1}}{\sqrt{\mu\lambda}}+\frac{\sqrt{k_1}}{\sqrt{\mu\lambda}},\frac{\sqrt{k_1}}{2\sqrt{\lambda}}+\frac{\sqrt{ak_1}}{\sqrt{\nu\lambda}}+\frac{\sqrt{k_1}}{\sqrt{\nu\lambda}}\bigg\}+\frac{1}{4\sqrt{\lambda}}T_m,\nonumber\\& \frac{\sqrt{k_1}}{\sqrt{\mu}\lambda}+\frac{\sqrt{k_1}}{\sqrt{\nu}\lambda}+\frac{1}{2\lambda\mu}+\frac{1}{2\lambda\nu}+\frac{1}{4\mu\sqrt{\lambda}}T_m+\frac{1}{4\nu\sqrt{\lambda}}T_m+\frac{1}{2\sqrt{\lambda}}\bigg\}.\nonumber\end{align*}$

对(3.6)式从 0 到$z$积分, 可得

$E(z, t)\leq E(0, t){\rm e}^{-b_1z}.$

在(3.6)式中取$z=0$, 并结合(3.5)式, 可得

$E(0, t)\leq b_2(t),$

其中

$b_2(t)=\frac{1}{2 b_1}\int_0^t\int_{D}\big(\mu f_if_i+\nu h_ih_i\big){\rm d}A{\rm d}\eta+\frac{a}{b_1}\int_0^t\int_{D_z}(f_i-h_i)(f_i-h_i){\rm d}A{\rm d}\eta+\frac{1}{b_1}m_3(t).$

联合(3.7)式,(3.8)式和(3.3)式可得定理 3.1. 证毕.

4 连续依赖性结果

假设$(u_i^*, v_i^*, T^*, p^*, q^*)$是方程 (1.1)-(1.7) 对应于$a=a^*$的一组解. 令

$w_i=u_i-u_i^*, r_i=v_i-v_i^*, \Sigma=T-T^*, \pi=p-p^*, s=q-q^*, \widetilde{a}=a-a^*,$

则$(w_i, r_i, \Sigma, \pi, s)$满足

$\mu w_{i}+\widetilde{a}(u_i-v_i)+a^*(w_i-r_i)=-\pi_{,i}+g_i\Sigma,\ w_{i,i}=0,$
$\nu r_{i}-\widetilde{a}(u_i-v_i)-a^*(w_i-r_i)=-s_{,i}+g_i\Sigma, \ r_{i,i}=0,$
$\Sigma_{,t}+\alpha(u_i+v_i)\Sigma_{,i}+\alpha(w_i+r_i)T^*_{,i}=\Delta \Sigma,$
$w_{i}=r_i=0,\ \Sigma=0,\ (x_1, x_2, t)\in D\times\{t>0\},$
$w_i=r_i=0,\ \Sigma=0, \ (x, t)\in\partial D\times\{t>0\},$
$\Sigma(x, 0)=0,\ x \in \mathcal{B},$
$w_i, r_i, |\Sigma|=O(1), |\nabla \Sigma|, |\pi|, |s|=o(x_3^{-1}), \ \text{当}\ x_3\rightarrow\infty.$

我们可得以下定理.

定理 4.1 设$\int_Df_3{\rm d}A=0$,$\int_Dh_3{\rm d}A=0$,$H\in L^\infty(\mathcal{B}), f_i, h_i\in H^1(\mathcal{B}),$则方程 (1.1)-(1.7) 的解连续依赖于系数$a$, 即$(w_i, r_i, \Sigma)\rightarrow0,\ \text{当}\ a\rightarrow a^*.$具体地, 或者不等式

$\begin{align}& 2b_7\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}(\xi-z)\big(\mu w_iw_i+\nu r_ir_i\big){\rm d}x{\rm d}\eta\nonumber\\& +2b_7a^*\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}(\xi-z)(w_i-r_i)(w_i-r_i){\rm d}x{\rm d}\eta\nonumber\\&+\frac{1}{2}{\rm e}^{-\omega t}\int_{\mathcal{B}_z}(\xi-z)\Sigma^2{\rm d}x+\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-\omega\eta}(\xi-z)\Big(\frac{1}{2}\omega\Sigma^2+\Sigma_{,i}\Sigma_{,i}\Big){\rm d}x{\rm d}\eta\nonumber\\\leq\ &\Big[\frac{2}{b_8}b_9(t)\widetilde{a}^2+4b_7b_4(t)\widetilde{a}^2\Big]{\rm e}^{-b_8z}+4b_7b_8b_4(t)\widetilde{a}^2z{\rm e}^{-b_8z}\nonumber\end{align}$

成立, 或者不等式

$\begin{align*}& 2b_7\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}(\xi-z)\big(\mu w_iw_i+\nu r_ir_i\big){\rm d}x{\rm d}\eta\\& +2b_7a^*\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}(\xi-z)(w_i-r_i)(w_i-r_i){\rm d}x{\rm d}\eta\nonumber\\& +\frac{1}{2}{\rm e}^{-\omega t}\int_{\mathcal{B}_z}(\xi-z)\Sigma^2{\rm d}x+\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-\omega\eta}(\xi-z)\Big(\frac{1}{2}\omega\Sigma^2+\Sigma_{,i}\Sigma_{,i}\Big){\rm d}x{\rm d}\eta\nonumber\\\leq\ & \Big[\frac{2}{b_8}b_9(t)\widetilde{a}^2+4b_7b_4(t)\widetilde{a}^2\Big]{\rm e}^{-b_8z}+\frac{4}{b_8-b_1}b_7b_8b_4(t)\widetilde{a}^2\big({\rm e}^{-b_1z}-{\rm e}^{-b_8z}\big)\nonumber\end{align*}$

成立, 其中$b_4(t), b_7, b_8, b_9(t)>0, \omega>8b_7[\frac{1}{\mu}+\frac{1}{\nu}]$.

首先, 我们建立以下辅助函数

$\begin{matrix}F_1(z, t)= & \int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}(\xi-z)\big(\mu w_iw_i+\nu r_ir_i\big){\rm d}x{\rm d}\eta\nonumber\\& +a^*\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}(\xi-z)(w_i-r_i)(w_i-r_i){\rm d}x{\rm d}\eta.\end{matrix}$

利用方程(4.1), (4.2), (4.4), (4.5)和(4.7), 由(4.8)式可得

$\begin{matrix}F_1(z, t)& =-\widetilde{a}\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}(\xi-z)(u_i-v_i)(w_i-r_i){\rm d}x{\rm d}\eta+\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}\pi w_3{\rm d}x{\rm d}\eta\nonumber\\& \ +\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}sr_3{\rm d}x{\rm d}\eta+\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}(\xi-z)g_i(w_i+r_i)\Sigma {\rm d}x{\rm d}\eta\nonumber\\&\doteq A_1(z, t)+A_2(z, t)+A_3(z, t)+A_4(z, t).\end{matrix}$

利用 Schwarz 不等式, 定理 3.1 和算术几何平均不等式, 可得

$\begin{matrix}A_1(z, t)& \leq\frac{1}{2}\widetilde{a}^2\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}(\xi-z)(u_i-v_i)(u_i-v_i){\rm d}x{\rm d}\eta \\& \ +\frac{1}{2}\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}(\xi-z)(w_i-r_i)(w_i-r_i){\rm d}x{\rm d}\eta\nonumber\\& \leq\frac{1}{2ab_1}\widetilde{a}^2b_2(t){\rm e}^{-b_1z}+\frac{1}{2}\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}(\xi-z)(w_i-r_i)(w_i-r_i){\rm d}x{\rm d}\eta.\end{matrix}$

由于$\int_{D_z}w_3{\rm d}A=\int_{D}w_3{\rm d}A=0$, 所以利用引理 2.3 可知存在向量函数$\widetilde{\phi}=(\widetilde{\phi}_1, \widetilde{\phi}_2)$使得

$\widetilde{\phi}_{\alpha, \alpha}=w_3,\ \text{在}\ D, \ \ \widetilde{\phi}_{\alpha}=0,\ \text{在}\ \partial D.$

所以

$\begin{align}A_2(z, t)&=\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}\pi_{,\alpha}\widetilde{\phi}_{\alpha}{\rm d}x{\rm d}\eta\nonumber\\&=\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}\Big[\mu w_{\alpha}+\widetilde{a}(u_\alpha-v_\alpha)+a^*(w_\alpha-r_\alpha)-g_\alpha\Sigma\Big]\widetilde{\phi}_{\alpha}{\rm d}x{\rm d}\eta.\nonumber\end{align}$

与(2.9)式类似, 可得

$\begin{matrix}A_2(z, t)&\leq\frac{\mu\sqrt{k_1}}{2\sqrt{\lambda}}\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}w_{i}w_{i}{\rm d}x{\rm d}\eta\nonumber\\& \ +\frac{\sqrt{k_1}}{2\sqrt{\lambda}}\widetilde{a}^2\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}(u_\alpha-v_\alpha)(u_\alpha-v_\alpha){\rm d}x{\rm d}\eta+\frac{\sqrt{k_1}}{2\sqrt{\lambda}}\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}w_{3}^2{\rm d}x{\rm d}\eta\nonumber\\& \ +\frac{\sqrt{k_1}}{2\sqrt{\lambda}}a^*\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}(w_\alpha-r_\alpha)(w_\alpha-r_\alpha){\rm d}x{\rm d}\eta+\frac{\sqrt{k_1}}{2\sqrt{\lambda}}a^*\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}w_{3}^2{\rm d}x{\rm d}\eta\nonumber\\& \ +\frac{\sqrt{k_1}}{2\sqrt{\lambda}}\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}\Sigma^2{\rm d}x{\rm d}\eta+\frac{\sqrt{k_1}}{2\sqrt{\lambda}}\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}w_{3}^2{\rm d}x{\rm d}\eta\nonumber\\&\leq\Big(\frac{\mu\sqrt{k_1}}{2\sqrt{\lambda}}+\frac{\sqrt{k_1}}{\sqrt{\lambda}}+\frac{\sqrt{k_1}}{2\sqrt{\lambda}}a^*\Big)\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}w_{i}w_{i}{\rm d}x{\rm d}\eta+\frac{\sqrt{k_1}}{2\sqrt{\lambda}}\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}\Sigma^2{\rm d}x{\rm d}\eta\nonumber\\& \ +\frac{\sqrt{k_1}}{2\sqrt{\lambda}}a^*\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}(w_\alpha-r_\alpha)(w_\alpha-r_\alpha){\rm d}x{\rm d}\eta+\frac{\sqrt{k_1}}{2a\sqrt{\lambda}}\widetilde{a}^2b_2(t){\rm e}^{-b_1z},\end{matrix}$

以及

$\begin{matrix}A_3(z, t)&\leq\Big(\frac{\nu\sqrt{k_1}}{2\sqrt{\lambda}}+\frac{\sqrt{k_1}}{\sqrt{\lambda}}+\frac{\sqrt{k_1}}{2\sqrt{\lambda}}a^*\Big)\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}r_{i}r_{i}{\rm d}x{\rm d}\eta+\frac{\sqrt{k_1}}{2\sqrt{\lambda}}\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}\Sigma^2{\rm d}x{\rm d}\eta\nonumber\\& \ +\frac{\sqrt{k_1}}{2\sqrt{\lambda}}a^*\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}(w_\alpha-r_\alpha)(w_\alpha-r_\alpha){\rm d}x{\rm d}\eta+\frac{\sqrt{k_1}}{2a\sqrt{\lambda}}\widetilde{a}^2b_2(t){\rm e}^{-b_1z},\end{matrix}$

利用Hölder不等式, 可得

$\begin{matrix}A_4(z, t)&\leq\frac{1}{2}\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}(\xi-z)(\mu w_iw_i+\nu r_ir_i){\rm d}x{\rm d}\eta\nonumber\\& \ +\Big(\frac{1}{2\mu}+\frac{1}{2\nu}\Big)\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}(\xi-z)\Sigma^2 {\rm d}x{\rm d}\eta.\end{matrix}$

把(4.10)-(4.13)式代入到(4.9)式, 可得

$\begin{matrix}F_1(z, t)\leq\ & b_3\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-\omega\eta}(\mu w_iw_i+\nu r_ir_i){\rm d}x{\rm d}\eta\\&+\frac{2\sqrt{k_1}}{\sqrt{\lambda}}a^*\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}(w_\alpha-r_\alpha)(w_\alpha-r_\alpha){\rm d}x{\rm d}\eta\nonumber\\& +\frac{2\sqrt{k_1}}{\sqrt{\lambda}}\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}\Sigma^2{\rm d}x{\rm d}\eta+\Big(\frac{1}{\mu}+\frac{1}{\nu}\Big)\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}(\xi-z)\Sigma^2 {\rm d}x{\rm d}\eta\nonumber\\&+b_4(t)\widetilde{a}^2{\rm e}^{-b_1z},\end{matrix}$

其中$b_3=2\max\Big\{\frac{\sqrt{k_1}}{2\sqrt{\lambda}}+\frac{\sqrt{k_1}}{\mu\sqrt{\lambda}}+\frac{\sqrt{k_1}}{2\mu\sqrt{\lambda}}a^*, \frac{\sqrt{k_1}}{2\sqrt{\lambda}}+\frac{\sqrt{k_1}}{\nu\sqrt{\lambda}}+\frac{\sqrt{k_1}}{2\nu\sqrt{\lambda}}a^*\Big\}, b_4(t)=\frac{1}{a}\big(\frac{\sqrt{k_1}}{\sqrt{\lambda}}+\frac{1}{b_1}\big)b_2(t). $

其次, 设

$\begin{matrix}F_2(z, t)& =\frac{1}{2}{\rm e}^{-\omega t}\int_{\mathcal{B}_z}(\xi-z)\Sigma^2{\rm d}x+\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-\omega\eta}(\xi-z)\Big(\frac{1}{2}\omega\Sigma^2+\Sigma_{,i}\Sigma_{,i}\Big){\rm d}x{\rm d}\eta.\end{matrix}$

利用散度定理和方程(4.3), 由(4.15)式可得

$\begin{matrix}F_2(z, t)& =\frac{1}{2}\alpha\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-\omega\eta}(w_3+r_3)\Sigma^2{\rm d}x{\rm d}\eta-\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-\omega\eta}\Sigma\Sigma_{,3}{\rm d}x{\rm d}\eta\nonumber\\& \ +\alpha\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-\omega\eta}(w_3+r_3)\Sigma T^*{\rm d}x{\rm d}\eta+\alpha\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-\omega\eta}(\xi-z)(w_i+r_i)\Sigma_{,i}T^*{\rm d}x{\rm d}\eta\nonumber\\&\doteq B_1(z, t)+B_2(z, t)+B_3(z, t)+B_4(z, t).\end{matrix}$

利用 Hölder 不等式, 引理 2.6, 引理2.4和引理2.1, 可得

$\begin{matrix}B_1(z, t)&\leq \frac{1}{2}\alpha\max\Big\{\frac{1}{\mu}, \frac{1}{\nu}\Big\}\int_0^t{\rm e}^{-\omega\eta}\Big[\int_{\mathcal{B}_z}(\mu w^2_3+\nu r^2_3){\rm d}x\Big]^\frac{1}{2}\Big(\int_{\mathcal{B}_z}\Sigma^4{\rm d}x\Big)^\frac{1}{2}{\rm d}\eta\nonumber\\&\leq \frac{1}{2}\alpha\sqrt{k_2m_4(t)}\max\Big\{\frac{1}{\sqrt{\mu}}, \frac{1}{\sqrt{\nu}}\Big\}\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-\omega\eta}\Sigma_{,i}\Sigma_{,i}{\rm d}x{\rm d}\eta,\end{matrix}$
$\begin{matrix}B_2(z, t)& \leq\frac{1}{2\sqrt{\lambda}}\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-\omega\eta}\Sigma_{,i}\Sigma_{,i}{\rm d}x{\rm d}\eta,\end{matrix}$
$\begin{matrix}B_3(z, t)& \leq\frac{1}{2\omega}\alpha^2 T_m^2\max\Big\{\frac{1}{\mu}, \frac{1}{\nu}\Big\}\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-\omega\eta}(\mu w^2_3+\nu r^2_3){\rm d}x{\rm d}\eta+\frac{1}{2}\omega\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-\omega\eta}\Sigma^2{\rm d}x{\rm d}\eta,\end{matrix}$
$\begin{matrix}B_4(z, t)&\leq\frac{1}{2}\alpha^2 T_m^2\max\Big\{\frac{1}{\mu}, \frac{1}{\nu}\Big\}\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-\omega\eta}(\xi-z)(\mu w_iw_i+\nu r_ir_i){\rm d}x{\rm d}\eta\nonumber\\& +\frac{1}{2}\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-\omega\eta}(\xi-z)\Sigma_{,i}\Sigma_{,i}{\rm d}x{\rm d}\eta.\end{matrix}$

把(4.17)-(4.20) 式代入到 (4.16) 式, 可得

$\begin{matrix}F_2(z, t)&\leq b_5(t)\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-\omega\eta}\Sigma_{,i}\Sigma_{,i}{\rm d}x{\rm d}\eta+b_6\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-\omega\eta}(\mu w^2_3+\nu r^2_3){\rm d}x{\rm d}\eta\nonumber\\& +\omega\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-\omega\eta}\Sigma^2{\rm d}x{\rm d}\eta+b_7\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-\omega\eta}(\xi-z)(\mu w_iw_i+\nu r_ir_i){\rm d}x{\rm d}\eta,\end{matrix}$

其中$b_5(t)=\alpha\sqrt{k_2m_4(t)}\max\{\frac{1}{\sqrt{\mu}}, \frac{1}{\sqrt{\nu}}\}+\frac{1}{\sqrt{\lambda}}, b_6=\frac{1}{\omega}\alpha^2 T_m^2\max\{\frac{1}{\mu}, \frac{1}{\nu}\},b_7=\frac{1}{2}\alpha^2 T_m^2\max\{\frac{1}{\mu}, \frac{1}{\nu}\}$.

第三, 设

$F(z, t)=2b_7F_1(z, t)+F_2(z, t),$

则结合(4.8)式和(4.15)式, 可得

$\begin{matrix}F(z, t)=\ &2b_7\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}(\xi-z)\big(\mu w_iw_i+\nu r_ir_i\big){\rm d}x{\rm d}\eta\nonumber\\&+2b_7a^*\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}(\xi-z)(w_i-r_i)(w_i-r_i){\rm d}x{\rm d}\eta\nonumber\\&+\frac{1}{2}{\rm e}^{-\omega t}\int_{\mathcal{B}_z}(\xi-z)\Sigma^2{\rm d}x+\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-\omega\eta}(\xi-z)\Big(\frac{1}{2}\omega\Sigma^2+\Sigma_{,i}\Sigma_{,i}\Big){\rm d}x{\rm d}\eta.\end{matrix}$

所以

$\begin{matrix}-\frac{\partial}{\partial z}F(z, t)=\ & 2b_7\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}\big(\mu w_iw_i+\nu r_ir_i\big){\rm d}x{\rm d}\eta\nonumber\\&+2b_7a^*\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}(w_i-r_i)(w_i-r_i){\rm d}x{\rm d}\eta\nonumber\\& +\frac{1}{2}{\rm e}^{-\omega t}\int_{\mathcal{B}_z}\Sigma^2{\rm d}x+\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-\omega\eta}\Big(\frac{1}{2}\omega\Sigma^2+\Sigma_{,i}\Sigma_{,i}\Big){\rm d}x{\rm d}\eta.\end{matrix}$

因为$\omega>8b_7[\frac{1}{\mu}+\frac{1}{\nu}]$, 结合(4.14)式和(4.21)式, 可得

$\begin{align}F(z, t)& \leq \frac{1}{2b_8}\Big(-\frac{\partial}{\partial z}F(z, t)\Big)+\frac{1}{2}F(z, t)+2b_7b_4(t)\widetilde{a}^2{\rm e}^{-b_1z}\nonumber\end{align}$

$\begin{matrix} F(z, t)& \leq \frac{1}{b_8}\Big(-\frac{\partial}{\partial z}F(z, t)\Big)+4b_7b_4(t)\widetilde{a}^2{\rm e}^{-b_1z},\end{matrix}$

其中$\frac{1}{2b_8}=\max\Big\{\frac{8\sqrt{k_1}b_7}{\omega\sqrt{\lambda}}, b_5(t), 2b_3b_7+b_6, \frac{4\sqrt{k_1}}{\sqrt{\lambda}}a^*b_7\Big\}$.对 (4.25) 式从 0 到$z$积分, 可得

$F(z, t)\leq F(0, t){\rm e}^{-b_8z}+4b_7b_8b_4(t)\widetilde{a}^2z{\rm e}^{-b_8z},\quad\quad\quad\quad\quad\quad\, \text{若 } b_1=b_8,$
$F(z, t)\leq F(0, t){\rm e}^{-b_8z}+\frac{4}{b_8-b_1}b_7b_8b_4(t)\widetilde{a}^2\big({\rm e}^{-b_1z}-{\rm e}^{-b_8z}\big), \text{若 } b_1\neq b_8.$

要使得 (4.27) 和 (4.26) 式有意义, 我们必须推导$F(0, t)$的上界. 在 (4.25) 式中取$z=0$可得

$\begin{matrix}F(0, t)&\leq \frac{1}{b_8}\Big[-\frac{\partial}{\partial z}F(0, t)\Big]+4b_7b_4(t)\widetilde{a}^2.\end{matrix}$

所以接下来我们只需推导$-\frac{\partial}{\partial z}F(0, t)$的上界. 为此, 对(4.9)式和(4.16)式微分, 并利用边界条件(4.4)可得

$\begin{align*}-\frac{\partial}{\partial z}F(0, t)=&-2b_7\widetilde{a}\int_0^t\int_{\mathcal{B}}{\rm e}^{-w\eta}(u_i-v_i)(w_i-r_i){\rm d}x{\rm d}\eta+2b_7\int_0^t\int_{\mathcal{B}}{\rm e}^{-w\eta}g_i(w_i+r_i)\Sigma {\rm d}x{\rm d}\eta\\&+\alpha\int_0^t\int_{\mathcal{B}}{\rm e}^{-\omega\eta}(w_i+r_i)\Sigma_{,i}T^*{\rm d}x{\rm d}\eta.\nonumber\end{align*}$

利用 Hölder 不等式, (2.17)式和引理2.1, 可得

$\begin{matrix}-\frac{\partial}{\partial z}F(0, t)\leq\ & b_7a^*\int_0^t\int_{\mathcal{B}}{\rm e}^{-w\eta}(w_i-r_i)(w_i-r_i){\rm d}x{\rm d}\eta+b_9(t)\widetilde{a}^2\nonumber\\&+b_7\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}(\xi-z)(\mu w_iw_i+\nu r_ir_i){\rm d}x{\rm d}\eta \\&+b_7\Big(\frac{1}{\mu}+\frac{1}{\nu}\Big)\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-w\eta}(\xi-z)\Sigma^2 {\rm d}x{\rm d}\eta\nonumber\\&+b_7\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-\omega\eta}(\mu w_iw_i+\nu r_ir_i){\rm d}x{\rm d}\eta+\frac{1}{2}\int_0^t\int_{\mathcal{B}_z}{\rm e}^{-\omega\eta}\Sigma_{,i}\Sigma_{,i}{\rm d}x{\rm d}\eta.\end{matrix}$

其中$b_9(t)=\frac{b_7}{aa^*}\int_0^tn_1(\eta){\rm d}\eta+\big(\frac{1}{\mu\lambda}+\frac{1}{\nu\lambda}\big)\big[m_2(t)+2\varepsilon_1\int_0^tn_1(\eta){\rm d}\eta\big]$.

由于$\omega>8b_7\big(\frac{1}{\mu}+\frac{1}{\nu}\big)$, 结合(4.24) 和 (4.29)式, 可得$-\frac{\partial}{\partial z}F(0, t)\leq2b_9(t)\widetilde{a}^2$. 由(4.28)式可得

$F(0, t)\leq \frac{2}{b_8}b_9(t)\widetilde{a}^2+4b_7b_4(t)\widetilde{a}^2.$

把(4.30)式代入到 (4.26) 式和 (4.27) 式, 并结合(4.23)式可以完成定理4.1的证明. 证毕.

注 4.1 定理 4.1 表明在$L^2(\mathcal{B}\times(0, t))$测度下, 方程 (1.1)-(1.7) 的解连续依赖于相互交换系数.

参考文献

Nield D A.

A note on modelling of local thermal non-equilibrium in a structured porous medium

Int J Heat Mass Transfer, 2002, 45: 4367-4368

DOI:10.1016/S0017-9310(02)00138-2      URL     [本文引用: 1]

Nield D A.

Effects of local thermal non-equilibrium in steady convection processes in saturated porous media: forced convection in a channel

J Porous Media, 1998, 1: 181-186

[本文引用: 1]

Nield D A, Kuznetsov A V.

The onset of convection in a bidisperse porous medium

Int J Heat Mass Transfer, 2006, 49: 3068-3074

DOI:10.1016/j.ijheatmasstransfer.2006.02.008      URL     [本文引用: 1]

Falsaperla P, Mulone G, Straughan B.

Bidispersive-inclined convection

Proc R Soc A, 2016, 472: 20160480

[本文引用: 2]

Franchi F, Nibbi R, Straughan B.

Continuous dependence on modelling for temperature-dependent bidispersiveflow

Proc R Soc A, 2017, 473(2208): 20170485.

DOI:10.1098/rspa.2017.0485      URL     [本文引用: 5]

\n We consider a model for flow in a porous medium which has a double porosity structure. There is the usual porosity herein called macro porosity, but in addition, we allow for a porosity due to cracks or fissures in the solid skeleton. The cracks give rise to a micro porosity. The model considered also allows for temperature effects with a single temperature\n T\n. This paper analyses three aspects of structural stability. The first establishes continuous dependence of the solution on the interaction coefficient between the velocities associated with the macro and micro porosity. The second analyses continuous dependence on the viscosity coefficients, while the third establishes continuous dependence on the radiation constant when Newton’s law of cooling is involved on the boundary.\n

Li Y F, Lin C H.

Continuous dependence for the nonhomogeneous Brinkman-Forchheimer equations in a semi-infinite pipe

Applied Mathematics and Computation, 2014, 244: 201-208

DOI:10.1016/j.amc.2014.06.082      URL     [本文引用: 3]

李远飞, 李志青.

具有非线性边界条件的瞬态热传导方程的二择一结果

数学物理学报, 2020, 40A(5): 1248-1258

[本文引用: 1]

Li Y F, Li Z Q.

Phragmén-Lindelöf Type Results for transient heat conduction equation with nonlinear boundary conditions

Acta Mathematica Scientia, 2020, 40A(5): 1248-1258

[本文引用: 1]

Scott N L.

Continuous dependence on boundary reaction terms in a porous medium of Darcy type

J Math Anal Appl, 2013, 399: 667-675

DOI:10.1016/j.jmaa.2012.10.054      URL     [本文引用: 1]

李远飞, 肖胜中, 曾鹏.

饱和蒸汽大气原始方程组的连续依赖性

华东师范大学学报 (自然科学版), 2021, 2021(3): 34-46

[本文引用: 1]

Li Y F, Xiao S Z, Zeng P.

Continuous dependence of primitive equations of the atmosphere with vapor saturation

Journal of East China Normal University (Natural Science), 2021, 2021(3): 34-46

[本文引用: 1]

Liu Y.

Continuous dependence for a thermal convection model with temperature-dependent solubility

Applied Mathematics and Computation, 2017, 3008: 18-30

[本文引用: 1]

Liu Y, Du Y, Lin C H.

Convergence and continuous dependence results for the Brinkman equations

Applied Mathematics and Computation, 2010, 215:, 4443-4455

DOI:10.1016/j.amc.2009.12.047      URL     [本文引用: 1]

Ciarletta M, Straughan B, Tibullo V.

Structural stability for a thermal convection model with temperature-dependent solubility

Nonlinear Analysis Real World Applications, 2015, 22: 34-43

DOI:10.1016/j.nonrwa.2014.07.012      URL     [本文引用: 1]

Payne L E, Song J C.

Spatial decay estimates for the Brinkman and Darcy flows in a semi-infinite cylinder

Continuum Mech Thermodyn, 1997, 9: 175-190

DOI:10.1007/s001610050064      URL     [本文引用: 2]

李远飞.

Keller-Segel 趋化模型解的全局存在性和爆破时间的下界估计

应用数学和力学, 2022, 43(7): 816-824

[本文引用: 2]

Li Y F.

Global existence of solutions and lower bound estimate of blow-up time for the Keller-Segel chemotaxis model

Applied Mathematics and Mechanics, 2022, 43(7): 816-824

[本文引用: 2]

Payne L E, Song J C.

Spatial decay bounds for double diffusive convection in Brinkman flow

Journal of Differential Equations, 2008, 244: 413-430

DOI:10.1016/j.jde.2007.10.003      URL     [本文引用: 1]

Horgan C O, Wheeler L T.

Spatial decay estimates for the Navier-Stokes equations with application to the problem of entry flow

SIAM J Appl Math, 1978, 35: 97-116

DOI:10.1137/0135008      URL     [本文引用: 1]

/