数学物理学报  2018, Vol. 38 Issue (1): 71-82   PDF    
扩展功能
加入收藏夹
复制引文信息
加入引用管理器
Email Alert
RSS
本文作者相关文章
张明书
朱泽奇
赵才地
Ladyzhenskaya流体力学方程组的确定模与确定结点个数估计
张明书1, 朱泽奇2, 赵才地1     
1. 温州大学数学与信息科学数学学院 浙江温州 325035;
2. 中国科学院武汉岩土力学研究所 武汉 430071
摘要:论文给出了二维有界区域上Ladyzhenskaya流体力学方程组的确定模与确定结点的个数估计.结果表明若该方程组的任意两个弱解的前有限个傅立叶模有相同的渐近行为,则这两个解就具有相同的渐近行为;若该方程组的任意两个强解在有限个空间中的点上有相同的渐近行为,则这两个解几乎在整个空间上具有相同的渐近行为.
关键词Ladyzhenskaya流体力学方程组    确定模    确定结点    渐近行为    
Determining Modes and Determining Nodes to the Fluid Flow of Ladyzhenskaya Model
Zhang Mingshu1, Zhu Zheqi2, Zhao Caidi1     
1. Department of Mathematics and Information Science, Wenzhou University, Zhejiang Wenzhou, 325035;
2. State Key Laboratory of Geomechanics and Geotechnical Engineering, Institute of Rock and Soil Mechanics, Chinese Academy of Sciences, Wuhan 430071
Abstract: This article estimates the finite number of determining modes and determining nodes for the fluid flow of Ladyzhenskaya model on two-dimensional bounded smooth domains. The finite number of determining modes implies that the solutions of the addressed fluids are determined completely by their first finite number of Fourier modes. The determining nodes reveals that whenever two different solutions of the fluid have the same asymptotic behavior at finite number of points in the physical space, then they also possess the same asymptotic behavior at almost everywhere points of the physical space.
Key words: Ladyzhenskaya model     Determining modes     Determining nodes     Asymptotic behavior    
1 引言

本文研究下面非自治Ladyzhenskaya流体力学方程组解的渐近行为

$ \begin{equation} {\bf u}_{t}-\triangle {\bf u}+({\bf u}\cdot\nabla){\bf u} - \nabla\cdot T({\bf e}({\bf u})) + \nabla p={\bf f}({\bf x}, t), ~~~ {\bf x}\in \Omega, \label{1} \end{equation} $ (1.1)
$ \begin{equation} \nabla\cdot {\bf u}=0, \label{2} \end{equation} $ (1.2)
$ \begin{equation} {\bf u}|_{\partial \Omega}=0, ~~~ T_{ij}n_jn_i=0, ~~~ i, j=1, 2, ~~~ {\bf x}\in \partial \Omega, \label{3} \end{equation} $ (1.3)
$ \begin{equation} {\bf u}({\bf x}, 0)={\bf u}_0({\bf x}), \label{4} \end{equation} $ (1.4)

其中$\Omega\subseteq {\mathbb{R}}^2$是适当光滑的有界区域, 未知函数${\bf u}$表示流体的速度场, ${\bf u}_0$表示流体的初始速度场, $p$代表压力, ${\bf f}$表示外力函数, $\nabla\cdot {\bf u}=0$描述流体的不可压缩性, $\nabla=({\partial}/{\partial x_1}, {\partial }/{\partial x_2})$是梯度算子, $\overrightarrow{{\bf n}}=(n_1, n_2)$$\partial \Omega$的单位外法向量, $T({\bf e}({\bf u}))$称为流体的本构关系.不同的流体对应不同的本构关系.本文中$ T({\bf e}({\bf u}))= (T_{ij}({\bf e}({\bf u})))_{2\times 2} $是二阶方阵, 其分量为

$ \begin{equation} T_{ij}({\bf e}({\bf u})) = \mu_0(\epsilon+|{\bf e}({\bf u})|^{2})^{-\gamma/2} {\bf e}_{ij}({\bf u}), ~~~ i, j=1, 2, \end{equation} $ (1.5)

这里$|e({\bf u})|^2=\sum\limits_{i, j=1}^2|e_{ij}({\bf u})|^2$, 而$e_{ij}({\bf u})=\frac{1}{2}(\frac{\partial {\bf u}_i}{\partial x_j}+\frac{\partial {\bf u}_j}{\partial x_i})$表示应变速率张量.此外$\epsilon$, $\mu_0$, $\gamma$是与流体相关的参数, 本文将假设$\epsilon>0$, $\mu_0>0$, $\gamma\in (0, 1)$.

方程组(1.1)-(1.5)由前苏联数学家Ladyzhenskaya在文献[30]中提出, 被称为Ladyzhenskaya流体模型.方程(1.3)中的第一个条件表示流体在边界上没有滑溜, 第二个条件表示流体的牵引力在边界上消失.当参数$\gamma < 0$时, 该流体被称为增稠流[6]; 当$\gamma>0$时, 该流体被称为稀化流; 当$\gamma=0$时, 方程组(1.1)-(1.2)就变成了著名的Navier-Stokes方程组.关于方程组(1.1)-(1.4)的详细物理背景, 可以参考文献[4-5, 29, 31].目前已有大量文献研究了Ladyzhenskaya流体模型(或者其相关流体模型).例如, 文献[31]首先讨论了解的存在唯一性; 文献[5]研究了Young测度值解的存在性; 文献[2-3]证明了解的存在性、唯一性和正则性; 文献[36]研究了Cauchy问题; 文献[13, 32]讨论了时间衰减率; 文献[38-40]研究了吸引子及其相关性质.

本文首先讨论方程组(1.1)-(1.5)的弱解的确定模问题.记${\mathbb{P}}_m$(定义见(3.5)式)是与Stokes算子$A$ (定义见(2.1)式)相关的投影算子.

定义1.1  设${\bf u}({\bf x}, t)$${\bf v}({\bf x}, t)$是方程组(1.1)-(1.5)的两个弱解.如果

$ \begin{eqnarray} \lim\limits_{t\rightarrow +\infty}\int_\Omega|{\mathbb{P}}_m {\bf u}({\bf x}, t)-{\mathbb{P}}_m {\bf v}({\bf x}, t)|^2{\rm d}{\bf x} =0 \end{eqnarray} $ (1.6)

意味着

$ \begin{eqnarray} \lim\limits_{t\rightarrow +\infty} \int_\Omega| {\bf u}({\bf x}, t)- {\bf v}({\bf x}, t)|^2{\rm d}{\bf x} =0, \end{eqnarray} $ (1.7)

则称与${\mathbb{P}}_m$相关的前$m$个傅立叶模为方程组(1.1)-(1.5)的确定模.

本文的第一个结果是证明方程组(1.1)-(1.5)存在有限个确定模.这表明方程组(1.1)-(1.5)的弱解的渐近行为完全由它们的前有限个傅立叶模的渐近行为所确定.关于偏微分方程组的确定模问题, 已有一些文献可参考.例如Navier-Stokes方程组(参见文献[8, 10-11, 15-16, 18, 20-21, 24-25, 27, 33-34]), 非线性耗散系统[9], 双曲MHD湍流模型[12], 耗散随机动力系统[17], 非线性Schrödinger方程[26], 三维Navier-Stokes-Voight方程组[28], 二阶梯度流方程[35], 二维微极流方程[37].

本文讨论的第二个问题是方程组(1.1)-(1.5)的强解的确定结点个数.众所周知, 在许多实际情形中, 用到的实验数据都是从物理空间中有限个测量点采集得到.因此关于确定结点的研究有重要意义.

定义1.2  考虑空间$\Omega$中的$N$个测量点, 记为$\Lambda=\{{\bf x}^1, {{\bf x}}^2, \cdots, {{\bf x}}^N\}.$${{\bf u}}({{\bf x}}, t)$${{\bf v}}({\bf x}, t)$是方程组(1.1)-(1.5)的两个强解.如果

$ \begin{equation} \lim\limits_{t\rightarrow +\infty}\max\limits_{j=1, \cdots, N} |{\textbf u}({\textbf x}, t)-{\textbf v}({\textbf x}, t)|_{{\textbf x} ={\textbf x}^j} =0 \end{equation} $ (1.8)

意味着

$ \begin{equation} \lim\limits_{t\rightarrow +\infty} \int_\Omega |{\textbf u}({\textbf x}, t)-{\textbf v}({\textbf x}, t)|^2{\rm d}{\bf x} =0, \end{equation} $ (1.9)

则称$\Lambda$为方程组(1.1)-(1.5)的确定结点集, $\Lambda$中的点称为方程组(1.1)-(1.5)的确定结点.

本文的第二个结果是证明方程组(1.1)-(1.5)的强解存在有限个确定结点.这一结果表明方程组(1.1)-(1.5)的强解的渐近行为完全可以被它们物理空间中有限个点的渐近行为所确定.这将为人们应用计算机对Ladyzhenskaya流体方程组进行数值模拟提供理论支撑.关于非线性偏微分方程组的确定结点问题研究, 可以参考Navier-Stokes方程组[14, 22-23], Kuramoto-Sivashinsky方程[19]等.

2 预备知识

本文中$\mathbb{R}$表示实数集, $\mathbb{N}$表示自然数集, $L^p(\Omega)$表示Lebesgue空间, 范数为$|\cdot|_{p}$; $W^{m, p}(\Omega)$表示Sobolev空间, 范数为$|\cdot|_{m, p}$, 其中

$ |\varphi|_{p}= \bigg (\int_\Omega |\varphi|^p{\rm d}{\bf x} \bigg)^{1/p}, ~~ |\varphi|_{m, p}= \bigg(\sum\limits_{|\beta|\leqslant m} \int_\Omega |D^\beta \varphi|^p{\rm d}{\bf x} \bigg)^{1/p}. $

${\mathbb{L}}^p(\Omega)=L^p(\Omega)\times L^p(\Omega)$表示二维Lebesgue空间, 范数为$\|\cdot\|_{{\mathbb{L}}^p(\Omega)}$, 特别地$\|\cdot\|=\|\cdot\|_{{\mathbb{L}}^2(\Omega)}$. ${\mathbb{W}}^{m, p}(\Omega)={W}^{m, p}(\Omega)\times {W}^{m, p}(\Omega)$表示二维Sobolev空间[1], 范数为$\|\cdot\|_{m, p}$, 这里

$ \|\varphi\|_{{\mathbb{L}}^p(\Omega)} =(|\varphi_1|_p^p+|\varphi_2|_p^p)^{1/p}, ~~ \varphi=(\varphi_1, \varphi_2)\in {\mathbb{L}}^p(\Omega), $
$ \|\varphi\|_{m, p} = (|\varphi_1|_{m, p}^p+|\varphi_2|_{m, p}^p)^{1/p}, ~~ \varphi=(\varphi_1, \varphi_2)\in {\mathbb{W}}^{m, p}(\Omega). $

另外, 记${\mathbb{H}}_0^1(\Omega)=\{ \varphi\in{{\cal C}}_{0}^\infty(\Omega)\times{{\cal C}}_{0}^\infty(\Omega)\}$${\mathbb{H}}^1(\Omega)$范数下的闭包,

$ {{\cal V}} =\left\{ \varphi \in {{\cal C}}_0^\infty(\Omega) \times {{\cal C}}_0^\infty(\Omega)| \varphi=(\varphi_1, \varphi_2), \nabla \cdot \varphi=0 \right\}, $
$ H = {\cal V} \mbox{在}{\mathbb{L}}^2(\Omega){空间中的闭包, 范数为} \|\cdot\|_H=\|\cdot\|, {对偶空间为}H^*, $
$ \cal V = V \mbox{在}{\mathbb{H}}^1(\Omega) {空间中的闭包, 范数为} \|\cdot\|_V =\|\cdot\|_{1, 2}, {对偶空间为} V^*. $

同时, 我们记$(\cdot, \cdot)$${\mathbb{L}}^2(\Omega)$或者$H$中的内积, $\langle\cdot, \cdot \rangle$$V$$V^*$的对偶积, 分别定义为

$ ({\bf u}, {\bf v}) = \int_{\Omega} {\bf u(x)}\cdot {\bf v(x)}{\rm d}{\bf x}, \qquad \langle {\bf u}, {\bf v} \rangle = \sum\limits_{k=1}^{2} \int_\Omega\frac{\partial{\bf u}}{\partial x_k} \frac{\partial {\bf v}}{\partial x_k}{\rm d}{\bf x}. $

下面介绍三个算子.首先定义算子$A$:

$ \begin{equation} \langle A{\bf u}, {\bf v} \rangle = \sum\limits_{k=1}^{2} \int_\Omega\frac{\partial{\bf u}}{\partial x_k} \frac{\partial {\bf v}}{\partial x_k}{\rm d}{\bf x}, ~~~ \forall {\bf u}, \, {\bf v}\, \in V. \end{equation} $ (2.1)

易知$A:V\mapsto V^*$是连续线性算子, 且$A: D(A)= \{{\bf u}\in V\big| A{\bf u}\in H\}\mapsto H$是线性算子.事实上, $ A{\bf u}=-P_{L}\triangle {\bf u}, \forall {\bf u}\, \in D(A), $其中$P_{L}$: ${\mathbb{L}}^2(\Omega)\longrightarrow H$是Leray投影算子.其次在${\mathbb{H}}^1(\Omega)$上定义如下的三线性型

$ \begin{equation} b({\bf u}, {\bf v}, {\bf w}) = \sum\limits_{i, j=1}^{2} \int_\Omega {\bf u}_i\frac{\partial {\bf v}_j}{\partial x_i}{\bf w}_j\, {\rm d}x, ~~~ \forall {\bf u}, \, {\bf v}, \, {\bf w}\, \in {\mathbb{H}}^1(\Omega). \end{equation} $ (2.2)

对任意${\bf u}, \, {\bf v}\in{\mathbb{H}}^1(\Omega)$, 记$\langle B({\bf u}, {\bf v}), {\bf w}\rangle= b({\bf u}, {\bf v}, {\bf w}), ~~~\forall {\bf w} \in {\mathbb{H}}^1(\Omega)$.特别地记$B({\bf u})=B({\bf u}, {\bf u})$.由于$V\subset {\mathbb{H}}^1(\Omega)$, 所以$B(\cdot, \cdot)$$V\times V$$V^*$的连续泛函.关于$b(\cdot, \cdot, \cdot)$, 我们有

引理2.1[15]  存在仅依赖于$\Omega$的正常数$c_i$ $(i=1, 2)$使得

$ \begin{equation} b({\bf u}, {\bf v}, {\bf v}) =0, ~~~~ b({\bf u}, {\bf v}, {\bf w}) =-b({\bf u}, {\bf w}, {\bf v}), ~~~\forall {\bf u}, ~~~{\bf v}, ~~~{\bf w} \in V, \end{equation} $ (2.3)
$ \begin{equation} |b({\bf u}, {\bf v}, {\bf w})| \leqslant c_1 \|{\bf u}\|^{\frac{1}{2}} \|{\bf u}\|^{\frac{1}{2}}_{V} \|{\bf v}\|^{\frac{1}{2}}\|{\bf v}\|^{\frac{1}{2}}_{V}\|{{\bf w}} \|_{V}, ~~~\forall {\bf u}, ~~~{\bf v}, ~~~{\bf w} \in V, \end{equation} $ (2.4)
$ \begin{equation} |b({\bf u}, {\bf v}, {\bf w})| \leqslant c_1 \|{\bf u}\|^{\frac{1}{2}} \|{\bf u}\|^{\frac{1}{2}}_{V} \|{{\bf v}}\|_{V}{\bf w}\|^{\frac{1}{2}}\|{\bf w}\|^{\frac{1}{2}}_{V}, ~~~\forall {\bf u}, ~~~{\bf v}, ~~~{\bf w} \in V, \end{equation} $ (2.5)
$ \begin{equation} |b({\bf u}, {\bf v}, {\bf w})| \leqslant c_2\|{\bf u}\|_{V}\| {\bf v}\|_{V}\|A {\bf w}\|, ~~~\forall\, {\bf u}, ~~~{\bf v}\in V, ~~~{\bf w}\in D(A). \end{equation} $ (2.6)

最后, 对任意${\bf u}\in V$, 记$ \mu({\bf u})= \mu_0(\epsilon+|{\bf e}({\bf u})|^{2})^{-\gamma/2} $并定义

$ \begin{eqnarray} \langle N({\bf u}), {\bf v}\rangle = \sum\limits_{i, j=1}^{2}\int_\Omega \mu({\bf u})e_{ij}({\bf u})e_{ij}({\bf v})\, {\rm d}{\bf x}, ~~~\forall \, {\bf u}, \, {\bf v}\in V, \end{eqnarray} $ (2.7)

$N({\bf u})$是从$V$$V^*$的连续泛函.

应用上面的记号和算子, 方程组(1.1)-(1.5)在零散度场中的弱的形式可以写成

$ \begin{equation} \frac{\partial{{\bf u}}}{\partial{t}} +A{\bf u} +B({\bf u}) +N({\bf u})={\bf f}(t), ~~ ~~~ t>0, \end{equation} $ (2.8)
$ \begin{equation} {\bf u}({\bf x}, 0)={\bf u}_0, \end{equation} $ (2.9)

其中(2.8)式是在${{\cal D}}'(0, +\infty; V^*)$分布意义下成立.下面给出关于方程组(2.8)-(2.9)解的定义.

定义2.2  如果函数${\bf {u}}\in L^2(0, +\infty; H )\cap L^2(0, +\infty; V)\cap L^\infty(0, +\infty;H), ~~~ {\bf {u}}({\bf {x}}, 0)={\bf {u}}_0$, 且${\bf {u}}$以及它的导数$\partial_t {\bf {u}}$${{\cal D}}'(0, +\infty; V^*)$分布意义下满足(2.8)式, 则称${\bf {u}}$是方程组(2.8)-(2.9)的弱解.如果${\bf {u}}$是弱解且${\bf {u}}\in L^2(0, +\infty; V) \cap L^2(0, +\infty; D(A))\cap L^\infty(0, +\infty;V)$, 则称${\bf {u}}$是方程组(2.8)-(2.9)的强解.

定理2.3  假设$\epsilon>0, \, \mu_0>0, \, \gamma\in(0, 1)$${\bf f}\in L^2(0, +\infty; H)$.则对任意${\bf u}_0\in H$, 方程组(2.8)-(2.9)存在唯一弱解; 且对任意${\bf u}_0\in V$, 方程组(2.8)-(2.9)存在唯一强解.

我们可以应用Galerkin方法证明定理2.3的结论, 过程与文献[4]相似, 在此略去.

下面的引理在我们估计Ladyzhenskaya流体方程组的确定模与确定结点时起重要作用.

引理2.4[15]  设$\phi(t)$$\psi(t)$$[0, +\infty)$上的实值函数, 且存在$T>0$, 使得

$ \begin{equation} \liminf\limits_{t\rightarrow +\infty}\frac{1}{T} \int_t^{t+T}\phi(\tau){\rm d}\tau>0, \end{equation} $ (2.10)
$ \begin{equation} \limsup\limits_{t\rightarrow +\infty}\frac{1}{T} \int_t^{t+T}\phi^-(\tau){\rm d}\tau < \infty, \end{equation} $ (2.11)
$ \begin{equation} \lim\limits_{t\rightarrow +\infty}\frac{1}{T} \int_t^{t+T}\psi^+(\tau){\rm d}\tau=0, \end{equation} $ (2.12)

其中$\phi^-(t) =\max\{-\phi(t), 0\}$, $\psi^+(t)=\max\{\psi(t), 0\}$.设$\xi(t)$$[0, +\infty)$上的非负绝对连续函数.若

$ \begin{eqnarray} \frac{{\rm d}\xi(t)}{{\rm d}t}+\phi(t) \xi(t)\leqslant\psi (t), \quad \mbox{a.e.} {于}\ ~~~ [0, +\infty), \end{eqnarray} $ (2.13)

则当$t\longrightarrow +\infty$时有$\xi(t)\longrightarrow 0$.

3 确定模个数估计

这一节的主要目标是估计方程组(2.8)-(2.9)的弱解的确定模个数.为此, 设${\bf u}$${\bf v}$是方程(2.8)对应于压力${\bf f}={\bf f}({\bf x}, t)$${\bf g}={\bf g}({\bf x}, t)$的两个弱解, 且${\bf f}({\bf x}, t)$${\bf g}({\bf x}, t)$具有相同的渐近行为, 即当$t\rightarrow+\infty$时, 有

$ \begin{eqnarray} \int_\Omega|{\bf f}({\bf x}, t)-{\bf g}({\bf x}, t)|^2{\rm d}{\bf x} \longrightarrow 0. \end{eqnarray} $ (3.1)

外力项${\bf f}({\bf x}, t)$${\bf g}({\bf x}, t)$的渐近强度用它们的$L^2$范数刻画为

$ \begin{eqnarray} \limsup\limits_{t\rightarrow +\infty}\|{\bf f}({\bf x}, t)\| = \limsup\limits_{t\rightarrow +\infty}\|{\bf g}({\bf x}, t)\| \triangleq {\bf F}. \end{eqnarray} $ (3.2)

根据算子$A$的定义, 易知算子$A$是自伴椭圆算子.由椭圆算子的经典理论[7, 15], 知存在一列特征值$\{\lambda_n\}_{n=1}^\infty$满足

$ \begin{eqnarray} 0 < \lambda_1\leqslant\lambda_2\leqslant \cdots\leqslant\lambda_n\leqslant\cdots, \end{eqnarray} $ (3.3)

且当$n\rightarrow +\infty$$\lambda_n\rightarrow+\infty $, 同时在$D(A)$中存在一列特征向量$\{{\bf w}_n({\bf x})\}_{n=1}^\infty$, 该列特征向量是空间$V$的标准正交基, 在$H$中正交, 使得

$ \begin{eqnarray} A{\bf w}_n=\lambda_n {\bf w}_n, ~~~\forall\, n\in\, {\mathbb{N}}. \end{eqnarray} $ (3.4)

根据这列特征向量, 我们可以将方程(2.8)的每个解展开成下面形式

$ {\bf u}({\bf x}, t) = \sum\limits_{n=1}^\infty \hat{u}_n(t){\bf w}_n({\bf x}), \qquad {\bf v}({\bf x}, t) =\sum\limits_{n=1}^\infty \hat{v}_n(t){\bf w}_n({\bf x}). $

这样, 我们定义与前$m$个傅立叶模相关的Galerkin投影算子${\mathbb{P}}_m$ (参见文献[13, p123])如下

$ \begin{eqnarray} {\mathbb{P}}_m{\bf u}({\bf x}, t) =\sum\limits_{n=1}^m \hat{u}_n(t){\bf w}_n({\bf x}), \qquad {\mathbb{P}}_m{\bf v}({\bf x}, t) =\sum\limits_{n=1}^m \hat{v}_n(t){\bf w}_n({\bf x}). \end{eqnarray} $ (3.5)

本节主要结果如下:

定理3.1  假定$\epsilon>0$, $\mu_0>0$, $\gamma\in(0, 1)$

$ \begin{eqnarray} \frac{\lambda_1}{\lambda_1+1}> 2\mu_0\epsilon^{-\gamma/2}\sqrt{4+\frac{12}{\epsilon^2}}. \end{eqnarray} $ (3.6)

$m\in{\mathbb{N}}$使得

$ \begin{eqnarray} \lambda_{m+1} >\frac{16(\lambda_1+1){c_1}^2|{\bf {f}}|_{{L^\infty}(0, +\infty, H)}^2}{\lambda_1^2 (\frac{\lambda_1}{\lambda_1+1}-2\mu_0\epsilon^{-\gamma/2}\sqrt{4+\frac{12}{\epsilon^2}})}, \end{eqnarray} $ (3.7)

则与Galerkin投影算子${\mathbb{P}}_m$相关的前$m$个傅立叶模是方程组(1.1)-(1.5)的确定模.

  设${\bf u}$${\bf v}$是方程(2.8)对应于压力${\bf f}={\bf f}({\bf x}, t)$${\bf g}={\bf g}({\bf x}, t)$的两个弱解, 且${\bf f}$${\bf g}$满足(3.1)式.根据定义1.1, 我们需证明当$m$足够大时, 若(1.6)式成立则(1.7)式成立.为此, 记${\bf w}={\bf u}-{\bf v}$, $Q_m =I-{\mathbb{P}}_m $, 其中$I$是单位算子.我们只需证明$\lim\limits_{t\rightarrow +\infty}\|Q_m{\bf w}(t) \|^2=0$.

因为${\bf u}$${\bf v}$是方程(2.8)对应于压力${\bf f}={\bf f}({\bf x}, t)$${\bf g}={\bf g}({\bf x}, t)$的两个弱解, 所以

$ \begin{equation} \frac{{\rm d}{\textbf w}}{{\rm d} t} +A{\bf w} +B({\bf w}, {\bf u})+B({\bf v}, {\bf w})+N({\bf u})-N({\bf v}) ={\bf f}(t)-{\bf g}(t). \end{equation} $ (3.8)

$Q_m {\bf w}$与方程(3.8)在$H$中作内积, 得

$ \begin{eqnarray} && \frac{1}{2}\frac{{\rm d}}{{\rm d}t}\|Q_m {\bf w}\|^2 + \|\nabla Q_m {\bf w}\|^2+ b({\bf w}, {\bf u}, Q_m {\bf w}) +b({\bf v}, {\bf w}, Q_m{\bf w}) \nonumber\\ && +(N({\bf u}), Q_m {\bf w}) - (N({\bf v}), Q_m{\bf w}) = ({\bf f}(t)-{\bf g}(t), Q_m{\bf w}), ~~~~\mbox{a.e.} 于\ ~~~ [0, +\infty). \end{eqnarray} $ (3.9)

应用(2.3)式, 我们把$ b({\bf w}, {\bf u}, Q_m{\bf w})$写成

$ \begin{eqnarray} b({\bf w}, {\bf u}, Q_m{\bf w}) = b({\mathbb{P}}_m{\bf w}, {\bf u}, Q_m {\bf w}) + b(Q_m{\bf w}, {\bf u}, Q_m {\bf w}). \end{eqnarray} $ (3.10)

由(2.4)和(2.5)式得

$ \begin{equation} |b({\mathbb{P}}_m {\bf w}, {\bf u}, Q_m {\bf w})| \leqslant c_1\|{\mathbb{P}}_m {\bf w}\|^{\frac{1}{2}} \|{\mathbb{P}}_m {\bf w}\|^{\frac{1}{2}}_{V} \|{\bf u}\|^{\frac{1}{2}} \|{\bf u}\|^{\frac{1}{2}}_{V} \|Q_m{\bf w}\|_{V}, \end{equation} $ (3.11)
$ \begin{eqnarray} |b(Q_m {\bf w}, {\bf u}, Q_m {\bf w})| &\leqslant& c_1\|Q_m{\bf w}\| \| Q_m {\bf w}\|_{V} \|{\bf u}\|_{V} \nonumber\\ &\leqslant& \frac{4{c_1}^2(\lambda_1+1)}{\lambda_1}\|{\bf u}\|^2_{V}\| Q_m{\bf w}\|^2+\frac{\lambda_1}{4(\lambda_1+1)}\| Q_m {\bf w}\|^2_{V}. \end{eqnarray} $ (3.12)

类似地, 有

$ \begin{eqnarray} |b( {\bf v}, {\bf w}, Q_m {\bf w})| = |b( {\bf v}, {\mathbb{P}}_m{\bf w}, Q_m {\bf w})| &\leqslant c_2 \|{\mathbb{P}}_m {\bf w}\|^{\frac{1}{2}} \|{\mathbb{P}}_m {\bf w}\|^{\frac{1}{2}}_{V} \|{\bf v}\|^{\frac{1}{2}} \|{\bf v}\|^{\frac{1}{2}}_{V}\|Q_m{\bf w}\|_{V} . \end{eqnarray} $ (3.13)

为了估计非线性项$(N({\bf u})-N({\bf v}), Q_m{\bf w})$, 我们记

$ \begin{eqnarray} {{\cal F}}({\mathbb{S}})=\mu_0(\epsilon+|{\mathbb{S}}|^2)^{-\gamma/2}{\mathbb{S}}, \end{eqnarray} $ (3.14)

其中${\mathbb{S}}= \left( \begin{array}{cc} s_{11}& s_{12} \\ s_{21}& s_{22} \\ \end{array} \right)\in {\mathbb{R}}^{2\times 2}_{\rm sym}$(二阶对称方阵), $|{\mathbb{S}}|^2=\sum\limits _{j, k=1}^2s_{jk}^2$, $s_{jk} \in {\mathbb{R}} ~~~ j, k=1, 2$.通过计算(参见文献[38, (3.10)式]), 可知${{\cal F}}({\mathbb{S}})$的一阶Frechét导数满足

$ \begin{eqnarray} \|D{{\cal F}}({\mathbb{S}})\| \leqslant c_3\triangleq \mu_0\epsilon^{-\gamma/2}\sqrt{4+\frac{12}{\epsilon^2}}, ~~~ \forall\, {\mathbb{S}}\in {\mathbb{R}}^{2\times 2}_{\rm sym}. \end{eqnarray} $ (3.15)

因此对任意${\mathbb{S}}_1, {\mathbb{S}}_2\in {\mathbb{R}}^{2\times 2}_{\rm sym}$, 有

$ \begin{eqnarray} {{\cal F}}({\mathbb{S}}_2)-{{\cal F}}({\mathbb{S}}_1) =\int_0^1 D{{\cal F}}({\mathbb{S}}_1+\tau ({\mathbb{S}}_2-{\mathbb{S}}_1)) ({\mathbb{S}}_2-{\mathbb{S}}_1)\, {\rm d}\tau. \end{eqnarray} $ (3.16)

注意到对任意${\bf u}\in V$$\|{\bf e}({\bf u})\| \leqslant \|{\bf u}\|_{V}$.结合(3.14)-(3.16)式可推得

$ \begin{eqnarray} |(N({\bf u})-N({\bf v}), Q_m{\bf w})| &=& \Big|\int_\Omega \big\{\nabla\cdot\big[{{\cal F}}({\bf e}({\bf u})) -{{\cal F}}({\bf e}({\bf v}))\big]\big\} \cdot Q_m{\bf w}{\rm d}{\bf x}\Big|\nonumber\\ &=& \Big| \int_\Omega [{{\cal F}}({\bf e}({\bf u})) -{{\cal F}}({\bf e}({\bf v}))] \cdot {\bf e}(Q_m{\bf w}){\rm d}{\bf x}\Big| \nonumber\\ &=& \Big|\int_\Omega\int_0^1 D{{\cal F}}({\bf e}({\bf u})+\tau {\bf e}({\bf w})) {\bf e}({\bf w}){\rm d}\tau \cdot{\bf e}(Q_m{\bf w}){\rm d}{\bf x}\Big|\nonumber\\ &\leqslant& c_3\|{\bf e}({\bf w})\|\|{\bf e}(Q_m{\bf w})\| \leqslant c_3\|{\bf w}\|_{V}\|Q_m{\bf w}\|_{V} \nonumber\\ &\leqslant& c_3(\|Q_m{\bf w}\|_{V}\|{\mathbb{P}}_m{\bf w}\|_{V} +\|Q_m{\bf w}\|^2_{V}). \end{eqnarray} $ (3.17)

现由Poincaré不等式得

$ \begin{eqnarray} \frac{\lambda_1}{\lambda_1+1}\|{\bf u}\|^2_{V} \leqslant \langle A{\bf u}, {\bf u}\rangle \leqslant \|{\bf w}\|^2_{V}, ~~~ \forall {\bf u}\in V. \end{eqnarray} $ (3.18)

结合(3.3), (3.4)和(3.18)式, 可得

$ \begin{eqnarray} \|{\mathbb{P}}_m{\bf w}\|^2_{V} \leqslant \frac{\lambda_1+1}{\lambda_1}\langle A{\mathbb{P}}_m{\bf w}, {\mathbb{P}}_m{\bf w}\rangle \leqslant \frac{(\lambda_1+1)\lambda_{m+1}}{\lambda_1}\|{\mathbb{P}}_m{\bf w}\|^2. \end{eqnarray} $ (3.19)

把估计式(3.19)代入(3.17)式得

$ \begin{eqnarray} |(N({\bf u})-N({\bf v}), Q_m{\bf w})| \leqslant c_3\sqrt{\frac{(\lambda_1+1)\lambda_{m+1}}{\lambda_1}} \|{\mathbb{P}}_m{\bf w}\|\|Q_m{\bf w}\|_{V} +c_3\|Q_m{\bf w}\|^2_{V}. \end{eqnarray} $ (3.20)

最后显然有

$ \begin{equation} |({\bf f}(t)-{\bf g}(t), Q_m {\bf w})| \leqslant \|{\bf f}(t)-{\bf g}(t)\| \| Q_m {\bf w}\|, \end{equation} $ (3.21)
$ \begin{equation} \|\nabla Q_m {\bf w}\|^2 \geqslant \frac{\lambda_1}{\lambda_1+1}\|Q_m{\bf w}\|^2_{V}\label{44}. \end{equation} $ (3.22)

从(3.9)-(3.13)式及(3.20)-(3.22)式推得

$ \begin{eqnarray} &&\frac{{\rm d}}{{\rm d} {t}}\|Q_m{{\bf w}}\|^2 + (\frac{\lambda_1}{\lambda_1+1}-2c_3)\|Q_m{\bf w}\|^2_{V} - \frac{8{c_1}^2(\lambda_1+1)}{\lambda_1}\|{\bf u}\|^2_{V} \|Q_m{\bf w}\|^2\nonumber\\ &\leqslant& 2c_1\|{\mathbb{P}}_m {\bf w}\|^{\frac{1}{2}} \|{\mathbb{P}}_m {\bf w}\|^{\frac{1}{2}}_{V} \|{\bf w}\|^{\frac{1}{2}} \|{\bf w}\|^{\frac{1}{2}}_{V}\|Q_m{\bf w}\|_{V}\nonumber\\ &&+ 2c_1 \|{\mathbb{P}}_m {\bf w}\|^{\frac{1}{2}} \|{\mathbb{P}}_m{\bf w}\|^{\frac{1}{2}}_{V}\| {\bf v}\|^{\frac{1}{2}} \|{\bf v}\|^{\frac{1}{2}}_{V}\|Q_m{\bf w}\|_{V} \nonumber\\ &&+ 2\|{\bf f}(t)-{\bf g}(t)\| \| Q_m {\bf w}\| +2c_3\sqrt{\frac{(\lambda_1+1)\lambda_{m+1}}{\lambda_1}} \|{\mathbb{P}}_m{\bf w}\| \|Q_m{\bf w}\|_{V}, \nonumber\\ && \mbox{a.e.} 于\ ~~~ [0, +\infty). \end{eqnarray} $ (3.23)

应用不等式$\lambda_{m+1}\|Q_m{\bf w}\|^2\leqslant \|Q_m{\bf w}\|^2_{V}$且记

$ \begin{equation} \xi_1(t) \triangleq\|Q_m{{\bf w}}(t)\|^2, \end{equation} $ (3.24)
$ \begin{equation} \phi_1(t) \triangleq(\frac{\lambda_1}{\lambda_1+1}-2c_3)\lambda_{m+1} -\frac{8{c_1}^2(\lambda_1+1)}{\lambda_1} \|{\bf u}(t)\|^2_{V}, \end{equation} $ (3.25)
$ \begin{eqnarray} \psi_1(t) &\triangleq& 2c_1\|{\mathbb{P}}_m {\bf w}(t)\|^{\frac{1}{2}} \| {\mathbb{P}}_m{\bf w}(t)\|^{\frac{1}{2}}_{V} \|{\bf w}(t)\|^{\frac{1}{2}}\|{\bf w}(t)\|^{\frac{1}{2}}_{V} \|Q_m{\bf w}(t)\|_{V}\nonumber\\ && + 2c_1\|{\mathbb{P}}_m {\bf w}(t)\|^{\frac{1}{2}} \| {\mathbb{P}}_m{\bf w}(t)\|^{\frac{1}{2}}_{V} \|{\bf v}(t)\|^{\frac{1}{2}}\|{\bf v}(t)\|^{\frac{1}{2}}_{V} \|Q_m{\bf w}(t)\|_{V}\nonumber\\ && + c_3\sqrt{\frac{(\lambda_1+1)\lambda_{m+1}}{\lambda_1}} \|{\mathbb{P}}_m{\bf w}(t)\| \|Q_m{\bf w}(t)\|_{V} + 2\|{\bf f}(t)-{\bf g}(t)\| \| Q_m {\bf w}(t)\|. \end{eqnarray} $ (3.26)

则微分不等式(3.23)可写成

$ \begin{eqnarray} \frac{{\rm d}\xi_1(t)}{{\rm d}t}+\phi_1(t)\xi_1(t) \leqslant \psi_1(t), \quad \mbox{a.e.}~ 于\ ~~~ [0, +\infty). \end{eqnarray} $ (3.27)

接下来将验证(3.24)-(3.27)式中的$\phi_1(t)$$\psi_1(t)$满足引理$2.4$的条件.事实上, 因为${\bf u}$是(2.8)-(2.9)式的弱解, 先用${\bf u}$与(2.8)式在$H$中作内积, 然后在$[t, t+T]$上积分($t$$T$均为正数), 再应用(2.3)式, (3.18)式, 不等式$\langle N({\bf u}), {\bf u}\rangle\geqslant 0$

$ \int_t^{t+T} ({\bf f}(s)\cdot {\bf u}(s) ){\rm d}s \leqslant \frac{\lambda_1}{2(\lambda_1+1)}\int_t^{t+T} \|{\bf u}(s)\|^2_{V}{\rm d}s + \frac{\lambda_1+1}{\lambda_1}\int_t^{t+T} \|{\bf f}(s)\|^2{\rm d}s, $

可以得到

$ \begin{eqnarray} \|{\bf u}(t+T)\|^2 + \frac{\lambda_1}{\lambda_1+1}\int_t^{t+T} \|{\bf u}(s)\|^2_{V}{\rm d}s \leqslant \|{\bf u}(t)\|^2 + \frac{\lambda_1+1}{\lambda_1}\int_t^{t+T} \|{\bf f}(s)\|^2{\rm d}s. \end{eqnarray} $ (3.28)

注意到${\bf u}\in L^\infty(0, +\infty;H)$${\bf f}\in L^\infty(0, +\infty;H)$.从(3.28)式可推知当$T$足够大时有

$ \begin{eqnarray} \frac{1}{T}\int_t^{t+T}\|{\bf u}(s)\|^2_{V}{\rm d}s \leqslant \frac{4(1+\lambda_1)^2|{\bf f}|_{{L^\infty}(0, +\infty;H)}^2}{\lambda_1^2}. \end{eqnarray} $ (3.29)

$\phi_1(t)$满足条件(2.11).同时, 应用(3.25)和(3.29)式可得

$ \begin{eqnarray} \lim\limits_{t\rightarrow +\infty}\frac{1}{T}\int_t^{t+T}\phi_1(\tau) {\rm d}\tau \geqslant (\frac{\lambda_1}{\lambda_1+1}-2c_3)\lambda_{m+1}- \frac{32(\lambda_1+1)^3{c_1}^2|{\bf f}|_{{L^\infty}(t, t+T, H)}^2}{\lambda_1^3}. \end{eqnarray} $ (3.30)

因此, 如果(3.6)式成立且$m$充分大使得(3.7)式成立, 则$\phi_1(t)$也满足条件(2.10).最后, 因为方程(2.8)的弱解都属于$L^2(0, +\infty;V)\cap L^{\infty}(0, +\infty;H)$, 根据(3.26)式中$\psi_1(t)$的表达式, 条件(1.6)及假设(3.1)式可推知$\psi_1(t)$满足条件(2.12).至此, 我们由引理$2.4$和不等式(3.27)推得

$ \lim\limits_{t\rightarrow +\infty}\xi_1(t)=\lim\limits_{t\rightarrow +\infty}\|Q_m{\bf w}(t) \|^2=0. $

证毕.

4 确定结点个数估计

本节的目标是估计方程组(2.8)-(2.9)的强解的确定结点个数.我们先介绍下面的引理.

引理4.1[15]  设$\Omega$被取$N$个恒等的正方形覆盖, 集合$\Lambda=\{{\textbf x}^1, {\textbf x}^2, \cdots, {\textbf x}^N\}\subset \Omega$, 且每个点${\textbf x}^i$属于且只属于其中的一个正方形.则对任意${\bf w}\in D(A)$, 有

$ \begin{eqnarray} \|{\bf w}\|^2_{V} \leqslant c_4N\eta({\bf w})^2+\frac{c_4}{\lambda_1N}\|A{\bf w}\|^2, \end{eqnarray} $ (4.1)

其中$ \eta({\textbf w})=\max\limits_{1\leqslant j\leqslant N}\big|{\bf w}({\bf x}^j)\big|$, $c_4$是仅依赖于$\Omega$的常数.

我们仍将应用(3.14)式定义的函数${{\cal F}}({\mathbb{S}})$来处理方程(2.8)中的非线性项$N(\cdot)$.事实上, 通过计算(参见文献[39, (3.10)式])可知${{\cal F}}({\mathbb{S}})$的一阶和二阶Fréchet导数满足

$ \begin{equation} \|D{{\cal F}}({\mathbb{S}})\|+\|D^2{{\cal F}}({\mathbb{S}})\| \leqslant c_5\triangleq c(\epsilon, \gamma, \mu_0), ~~~ \forall\, {\mathbb{S}}\in {\mathbb{R}}^{2\times 2}_{\rm sym}. \end{equation} $ (4.2)

为了估计方程组(1.1)-(1.5)的强解的确定结点个数, 我们需假定

$ \begin{equation} c_5 < 1/4. \end{equation} $ (4.3)

本节的主要结果如下:

定理4.2  假设$\epsilon>0$, $\mu_0>0$, $\gamma\in(0, 1)$且(4.3)式成立; 设压力${\textbf f}\in L^{\infty}(0, +\infty;H)$, ${\textbf g}\in L^{\infty}(0, +\infty;H)$且满足(3.1)-(3.2)式.若$\Omega$$N$个恒等的正方形覆盖, 而集合${\mathcal E}=\{{\textbf x}^1, {\textbf x}^2, \cdots, {\textbf x}^N\}\subset \Omega$, 且每个点${\textbf x}^i$属于且只属于其中的一个正方形.则存在常数

$ c_6\triangleq \frac{48(1+\lambda_1)^2c^2_2c_4{{\bf F}}^2}{\lambda^3_1(1-4c_5)}, $

如果

$ \begin{equation} N \geqslant c_6, \end{equation} $ (4.4)

${\mathcal E}$是方程组(1.1)-(1.5)的强解的确定结点集.

  设${\bf u}$${\bf v}$是方程(2.8)对应于压力${\bf f}={\bf f}({\bf x}, t)$${\bf g}={\bf g}({\bf x}, t)$的两个强解, 且${\bf f}$${\bf g}$满足(3.1)-(3.2)式.设$\Omega$$N$个恒等的正方形覆盖, 而集合${\mathcal E}=\{{\textbf x}^1, {\textbf x}^2, \cdots, {\textbf x}^N\}\subset \Omega$, 且每个点${\textbf x}^i$属于且只属于其中的一个正方形.记${\bf w}(t)={\bf u}(t)-{\bf v}(t)$.我们需要证明:在定理$4.2$的条件下, 若(1.8)式成立, 则有$\lim\limits_{t\rightarrow+\infty}\|{\bf w}(t)\|= 0$.事实上, 我们可以证明$\lim\limits_{t\rightarrow+\infty}\|{\bf w}(t)\|_V=0$.

首先, 由

$ \begin{eqnarray} \frac{{\rm d}{\bf w}}{{\rm d}t}+A{\bf w}+ B({\bf w}, {\bf u})+B({\bf v}, {\bf w})+N({\bf u})-N({\bf v}) ={\bf f}(t)-{\bf g}(t), \end{eqnarray} $ (4.5)

$A {\bf w}$与方程(4.5)在$H$中作内积, 得

$ \begin{eqnarray} && \frac{1}{2}\frac{{\rm d}}{{\rm d}t}\|\nabla {\bf w}\|^2 + \|A{\bf w}\|^2+ b({\bf w}, {\bf u}, A{\bf w}) + b({\bf v}, {\bf w}, A{\bf w}) \nonumber\\ &&+ (N({\bf u}), A{\bf w}) - (N({\bf v}), A{\bf w}) = ({\bf f}(t)-{\bf g}(t), A{\bf w}), \quad \mbox{a.e. 于} ~~~ [0, +\infty). \end{eqnarray} $ (4.6)

应用(2.6)式和Cauchy不等式得

$ \begin{equation} |b({\bf w}, {\bf u}, A{\bf w})| \leqslant c_2 \|{\bf w}\|_{V} \|{\bf u}\|_{V}\|A{\bf w}\| \leqslant 3c_2^2\|{\bf w}\|^2_{V}\| {\bf u}\|^2_{V} + \frac{1}{6}\|A{\bf w}\|, \end{equation} $ (4.7)
$ \begin{equation} |b({\bf v}, {\bf w}, {\bf Aw})| \leqslant 3c_2^2\|{\bf w}\|^2_{V}\| {\bf v}\|^2_{V} + \frac{1}{6}\|A{\bf w}\|^2. \end{equation} $ (4.8)

为了估计$(N({\bf u})-N({\bf v}), A{\bf w})$, 我们再次应用(3.14)式中定义的函数${{\cal F}}({\mathbb{S}})$.事实上, 由(4.2)-(4.3)式可得

$ \begin{eqnarray} |(N({\bf u})-N({\bf v}), Q_m{\bf w})| &=& \Big|\int_\Omega \big\{\nabla\cdot\big[{{\cal F}}({\bf e}({\bf u})) -{{\cal F}}({\bf e}({\bf v}))\big]\big\} \cdot A{\bf w}{\rm d}{\bf x}\Big| \nonumber\\ &=& \Big|\int_\Omega\big\{\nabla\cdot\int_0^1 D{{\cal F}}\big({\bf e}({\bf u})+\tau {\bf e}({\bf w})\big) {\bf e}({\bf w}){\rm d}\tau\big\} \cdot {\rm A}{\bf w}{\rm d}{\bf x}\Big|\nonumber\\ &\leqslant& \int_\Omega \int_0^1 \|D^2{{\cal F}}\big({\bf e}({\bf u})+\tau {\bf e}({\bf w})\big)\|{\rm d}\tau ~~~ |{\bf e}({\bf w})| | {\rm A}{\bf w}|{\rm d} {\bf x}\nonumber\\ && +\int_\Omega \int_0^1 \|D{{\cal F}}\big({\bf e}({\bf u})+\tau {\bf e}({\bf w})\big)\|{\rm d}\tau~~~ |\nabla{\bf e}({\bf w})| | {\rm A}{\bf w}|{\rm d} {\bf x}\nonumber\\ &\leqslant& c_5(\|\nabla{\bf w}\|+\|\triangle{\bf w}\|)\|A{\bf w}\| \leqslant 2c_5\|A{\bf w}\|^2. \end{eqnarray} $ (4.9)

另外, 显然有

$ \begin{eqnarray} ({\bf f}(t)-{\bf g}(t), A{\bf w}) \leqslant \frac{1}{6}\|A{\bf w}\|^2+3\|{\bf f}(t)-{\bf g}(t)\|^2. \end{eqnarray} $ (4.10)

从(4.6)-(4.10)式可推得

$ \begin{eqnarray} \frac{{\rm d}}{{\rm d}t}\|\nabla {\bf w}(t)\|^2 + (1-4c_5)\|A{\bf w}\|^2 \leqslant 6c^2_2\|{\bf w}\|^2_{V} (\|{\bf u}\|^2_{V}+\|{\bf v}\|^2_{V}) + 6\|{\bf f}(t)-{\bf g}(t)\|^2. \end{eqnarray} $ (4.11)

结合上式和引理$4.1$中的(4.1)式得到

$ \begin{eqnarray} && \frac{{\rm d}}{{\rm d}t}\|{\bf w}(t)\|^2_{V} + \|{\bf w}(t)\|^2_{V} \Big[\frac{\lambda_1(1-4c_5)N}{c_4}-6c^2_2 (\|{\bf u}(t)\|^2_{V}+\|{\bf v}(t)\|^2_{V})\Big]\nonumber\\ &\leqslant& (1-4c_5)N^2(\eta({\bf w}(t)))^2 + 6\|{\bf f}(t)-{\bf g}(t)\|^2, \quad \mbox{a.e. 于} ~~~ [0, +\infty). \end{eqnarray} $ (4.12)

$ \begin{equation} \xi_2(t) \triangleq \|{{\bf w}}(t)\|^2_{V}, \end{equation} $ (4.13)
$ \begin{equation} \phi_2(t) \triangleq \frac{\lambda_1(1-4c_5)N}{c_4} - 6c^2_2(\|{\bf u}(t)\|^2_{V}+\|{\bf v}(t)\|^2_{V}), \end{equation} $ (4.14)
$ \begin{equation} \psi_2(t) \triangleq (1-4c_5)N^2(\eta({\bf w}(t)))^2 +6\|{\bf f}(t)-{\bf g}(t)\|^2. \end{equation} $ (4.15)

则微分不等式(4.12)可写成

$ \begin{eqnarray} \frac{{\rm d}\xi_2(t)}{{\rm d}t}+\phi_2(t)\xi_2(t) \leqslant \psi_2(t), \quad \mbox{a.e. 于}\ ~~~ [0, +\infty). \end{eqnarray} $ (4.16)

下面验证(4.13)-(4.16)式中的$\phi_2(t)$$\psi_2(t)$满足引理$2.4$的条件.事实上由于${\bf u}$${\bf v}$是方程(2.8)的强解, 故都属于$L^2(0, +\infty;V)\cap L^{\infty}(0, +\infty;V)$, 所以$\phi_2(t)$满足条件(2.11).此外, 如果条件(4.3)-(4.4)成立, 则对于充分大的$T$, 我们可由(3.2)式和(3.28)-(3.29)式推得

$ \begin{eqnarray} \lim\limits_{t\rightarrow+\infty}\frac{1}{T}\int_t^{t+T} (\|{\bf u}(\tau)\|^2_{V}+\|{\bf v}(\tau)\|^2_{V}) {\rm d}\tau \leqslant \frac{8(\lambda_1+1)^2{\bf F}^2}{\lambda_1^2}. \end{eqnarray} $ (4.17)

从而

$ \begin{eqnarray} \liminf\limits_{t\rightarrow+\infty}\frac{1}{T}\int_t^{t+T} \psi_2(\tau){\rm d}\tau &=& \frac{\lambda_1(1-4c_5)N}{c_4} -\frac{6c^2_2}{T}\int_t^{t+T} (\|{\bf u}(\tau)\|^2_{V}+\|{\bf v}(\tau)\|^2_{V}) {\rm d}\tau\nonumber\\ & \geqslant& \frac{\lambda_1(1-4c_5)N}{c_4} -\frac{24(\lambda_1+1)^2{\bf F}^2}{\lambda_1^2} >0. \end{eqnarray} $ (4.18)

$\phi_2(t)$也满足条件(2.10).最后, 条件(1.8)表明

$ \begin{equation} \lim\limits_{ t\longrightarrow +\infty}\eta({\bf w}(t))=0. \end{equation} $ (4.19)

注意到${\bf f}\in L^{\infty}(0, +\infty;H)$${\bf g}\in L^{\infty}(0, +\infty;H)$.从(3.1)-(3.2), (4.3), (4.15)及(4.19)式可以推得

$ \begin{eqnarray*} \lim\limits_{t\rightarrow+\infty}\frac{1}{T}\int_t^{t+T} \psi^+_2(\tau) {\rm d}\tau = \lim\limits_{t\rightarrow+\infty}\frac{1}{T}\int_t^{t+T} \Big(6\|{\bf f}(t)-{\bf g}(t)\|^2 + (1-4c_5)N^2(\eta({\bf w}))^2\Big) {\rm d}t = 0. \end{eqnarray*} $

因此, 函数$\psi_2(t)$满足条件(2.12).至此, 我们由引理$2.4$和不等式(4.16)推得

$ \lim\limits_{t\rightarrow +\infty}\xi_2(t)=\lim\limits_{t\rightarrow +\infty}\|{\bf w}(t)\|^2_V=0. $

证毕.

参考文献
[1] Adams R A. Sobolev Spaces. New York: Academic Press, 1975.
[2] Bae H. Existence regularity and decay rate of solutions of non-Newtonian flow. J Math Anal Appl, 1999, 231: 467–491. DOI:10.1006/jmaa.1998.6242
[3] Bae H, Choe H. Existence and regularity of solutions of non-Newtonian flow. Quart Appl Math, 2000, 58: 101–110.
[4] Bloom F, Hao W. Regularization of a non-Newtonian system in an unbounded channel:Existence and uniqueness of solutions. Nonlinear Anal, 2001, 44: 281–309. DOI:10.1016/S0362-546X(99)00264-3
[5] Bellout H, Bloom F, Nečas J. Young measure value solutions for non-Newtonian incompressible fluids. Comm Partial Differential Equations, 1994, 19: 1768–1803.
[6] Bohme G. Non-Newtonian Fluid Mechanics. Amsterdam: North-Holland, 1987.
[7] Boyer F, Fabrie P. Mathematical Tools for the Study of the Incompressible Navier-Stokes Equations and Related Models. New York: Springer, 2013.
[8] Cao C, Rammaha M A, Titi E S. The Navier-Stokes equations on the rotating 2-D sphere:Gevrey regularity and asymptotic degrees of freedom. Z Angew Math Phys, 1999, 50: 341–360. DOI:10.1007/PL00001493
[9] Cockburn B, Jones D A, Titi E S. Eestimating the number of asymptotic degrees of freedom for nonlinear dissipative systems. Math Comp, 1997, 219: 1073–1087.
[10] Constantin P, Foias C, Manley O, Temam R. Determining modes and fractal dimension of turbulent flows. J Fluid Mech, 1985, 150: 427–440. DOI:10.1017/S0022112085000209
[11] Constantin P, Foias C, Manley O, Temam R. On the dimension of the attractor in two-dimensional turbulence. Phys D, 1988, 30: 427–40.
[12] Catania D. Global attractor and determining modes for a hyperbolic MHD turbulence model. J Turbulence, 2011, 40: 1–20.
[13] Dong B, Li Y. Large time be havior to system of incompressible. J Math Anal Appl, 2004, 209: 667–676.
[14] Foias C, Temam R. Determination of the solutions of the Navier-Stokes equations by a set of nodal values. Math Comp, 1984, 43: 117–133. DOI:10.1090/S0025-5718-1984-0744927-9
[15] Foias C, Manley O, Rosa R, Temam R. Navier-Stokes Equations and Turbulence. Cambridge: Cambridge University Press, 2004.
[16] Foias C, Jolly M, Kravchenko R, Titi E S. A determining form for the two-dimensional Navier-Stokes equations:The Fourier modes case. J Math Phys, 2012, 53: 1–30.
[17] Flandolf F, Langa J A. Determining modes for dissipative random dynamical systems. Stochastics:An Inter J Prob Stoch Processes, 1999, 66: 1–25.
[18] Foias C, Manley O, Temam R, Tréve Y M. Asymptotic analysis of the of the Navier-Stokes equations. Phys D, 1983, 9: 157–188. DOI:10.1016/0167-2789(83)90297-X
[19] Foias C, Kukavica I. Determining nodes for the Kuramoto-Sivashinsky equation. Dynamics Differential Equations, 1995, 7: 365–373. DOI:10.1007/BF02219361
[20] Galdi G P. Determining modes, nodes and volume elements for stationary solutions of the Navier-Stokes problem past a three-dimensional body. Arch Rational Mech Anal, 2006, 180: 97–126. DOI:10.1007/s00205-005-0395-0
[21] Hale J K, Raugel G. Regularity, determining modes and Galerkin methods. J Math Pures Appl, 2003, 82: 1075–1136. DOI:10.1016/S0021-7824(03)00045-X
[22] Hoff D, Ziane M. The global attractor and finite determining nodes for the Navier-Stokes equations of compressible flow with singular initial data. Indiana University Math J, 2003, 49: 843–890.
[23] Jones D A, Titi E S. On the number of determining nodes for the 2D Navier-Stokes equations. J Math Anal Appl, 1992, 168: 72–88. DOI:10.1016/0022-247X(92)90190-O
[24] Jones D A, Titi E S. Determining finite volume elements for the 2D Navier-Stokes equations. Physica D, 1992, 60: 165–174. DOI:10.1016/0167-2789(92)90233-D
[25] Jones D A, Titi E S. Upper bounds on the number of determining modes, and volum elements for the Navier-Stokes equations. Indiana Univ Math J, 1993, 42: 875–887. DOI:10.1512/iumj.1993.42.42039
[26] Jollya M S, Sadigov T, Titi E S. A determining form for the damped driven nonlinear Schrödinger equation-Fourier modes case. J Differential Equations, 2015, 258: 2711–2744. DOI:10.1016/j.jde.2014.12.023
[27] Ju N. Estimates of asymptotic degrees of freedom for solutions to the Navier-Stokes equations. Nonlinearity, 2000, 13: 777–789. DOI:10.1088/0951-7715/13/3/314
[28] Kalantarov V K, Titi E S. Global attractors and determining modes for the 3D Navier-Stokes-Voight equations. Chinese Annals Math-B, 2009, 30: 697–714. DOI:10.1007/s11401-009-0205-3
[29] 郭柏灵, 林国广, 尚亚东. 非牛顿流动力系统. 北京: 国防工业出版社, 2006.
Guo B L, Lin G G, Shang Y D. Non-Newtonian Fluid Dynamic System. Beijing: National Defence Industry Press, 2006.
[30] Ladyzhenskaya O. The Mathematical Theory of Viscous Incompressible Fluids. New York: Gordon and Breach, 1969.
[31] Ladyzhenskaya O. New equations for the description of the viscous incompressible fluids and solvability in the large of the boundary value problem for them. Trudy Steklov's Math Inst, 1967, 102: 85–104.
[32] Nečasová Š, Penel P. L2 decay for weak solution to equation of non-Newtonian incompressible fluids in the whole space. Nonlinear Anal, 2001, 47: 4181–4191. DOI:10.1016/S0362-546X(01)00535-1
[33] Olson E, Titi E S. Determining modes for continuous data assimilation in 2D turbulence. J Stat Phys, 2003, 113: 799–840. DOI:10.1023/A:1027312703252
[34] Olson E, Titi E S. Determining modes and Grashof number in 2D turbulence:a numerical case study. Theoretical Comp Fluid Dyna, 2008, 22: 327–339. DOI:10.1007/s00162-008-0086-1
[35] Paicu M, Raugel G, Rekalo A. Regularity and finite-dimensional behaviour of the global attractor of the second grade fluids equations. J Differential Equations, 2012, 252: 3695–3751. DOI:10.1016/j.jde.2011.10.015
[36] Pokorný M. Cauchy problem for non-Newtonian incompressible fluids. Appl Math, 1996, 41: 169–201.
[37] Szopa P. Determining modes for 2-D micropolar fluid flows. Math Comp Modelling, 2005, 42: 1079–1088. DOI:10.1016/j.mcm.2004.09.007
[38] Zhao C, Zhou S. Pullback attractors for nonautonomous incompressible non-Newtonian fuild. J Differential Equations, 2007, 238: 394–425. DOI:10.1016/j.jde.2007.04.001
[39] Zhao C, Li Y, Zhou S. Regularity of trajectory attractor and upper semicontinuity of global attractor for a 2D non-Newtonian fuild. J Differential Equations, 2009, 247: 2331–2363. DOI:10.1016/j.jde.2009.07.031
[40] Zhao C, Jia X, Yang X. Uniform attractor non-autonomous incompressible non-Newtonian fluid with a new class of external forces. Acta Math Sci, 2011, 31: 1803–1812. DOI:10.1016/S0252-9602(11)60362-7