Processing math: 2%

数学物理学报, 2018, 38(6): 1205-1223 doi:

论文

记忆型非经典扩散方程在H1(Rn)×L2μ(R+;H1(Rn))中的全局吸引子

汪璇,, 韩英,, 高承华,

Global Attractor in H1(Rn)×L2μ(R+;H1(Rn)) for the Nonclassical Diffusion Equations with Fading Memory

Wang Xuan,, Han Ying,, Gao Chenghua,

通讯作者: 汪璇, E-mail: wangxuan@nwnu.edu.cn

收稿日期: 2017-07-17  

基金资助: 国家自然科学基金.  11761062
国家自然科学基金.  11561064
国家自然科学基金.  11661071
西北师范大学青年教师科研能力提升计划.  NWNU-LKQN-14-6
甘肃省高校科研项目.  2016A-003

Received: 2017-07-17  

Fund supported: the NSFC.  11761062
the NSFC.  11561064
the NSFC.  11661071
the Young Teachers Scientific Research Ability Promotion Plan of Northwest Normal University.  NWNU-LKQN-14-6
the Science Research Project for Colleges and Universities of Gansu Province.  2016A-003

作者简介 About authors

韩英,E-mail:469328224@qq.com , E-mail:469328224@qq.com

高承华,E-mail:gaochenghua@nwnu.edu.cn , E-mail:gaochenghua@nwnu.edu.cn

摘要

该文研究了带有衰退记忆和超临界非线性项的非经典扩散方程在无界域Rn中的动力学行为.运用半群理论和收缩函数方法,当外力项仅属于H-1Rn)时,证明了全局吸引子在H1(Rn)×L2μ(R+;H1(Rn))中的存在性.

关键词: 非经典扩散方程 ; 衰退记忆 ; 收缩函数 ; 全局吸引子

Abstract

In this paper, we are concerned with the dynamical behavior of the nonclassical diffusion equations with fading memory and supercritical nonlinearity on unbounded domain Rn. By applying semigroup theory and method of contractive function, we obtain the existence of global attractors in H1(Rn)×L2μ(R+;H1(Rn)), when the external forcing term g merely belongs to H-1(Rn).

Keywords: Nonclassical diffusion equation ; Fading memory ; Contractive function ; Global attractor

PDF (439KB) 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

汪璇, 韩英, 高承华. 记忆型非经典扩散方程在H1(Rn)×L2μ(R+;H1(Rn))中的全局吸引子. 数学物理学报[J], 2018, 38(6): 1205-1223 doi:

Wang Xuan, Han Ying, Gao Chenghua. Global Attractor in H1(Rn)×L2μ(R+;H1(Rn)) for the Nonclassical Diffusion Equations with Fading Memory. Acta Mathematica Scientia[J], 2018, 38(6): 1205-1223 doi:

1 引言

本文中,我们考虑了如下带有衰退记忆和超临界非线性项的非经典扩散方程解的渐近性态

{utνΔutΔu+λu0k(s)Δu(ts)ds+f(x,u)=g(x), (x,t)Rn×R+,u(x,t)=u0(x,t), (x,t)Rn×(,0],
(1.1)

其中λ为正常数,外力项g(x)\inH1(Rn),并且非线性项f(x,u)=f1(u)+a(x)f2(u)满足如下条件

\begin{eqnarray}\label{1.2}&\alpha_1|u|^p-\beta_1|u|^2\leqslant f_1(u)u\leqslant \gamma_1|u|^p+\delta_1|u|^2, ~p\geqslant 2, \\ &f_1(u)u\geqslant0, f_1^\prime(u)\geqslant -c_1, ~\mbox{且}~\beta_1 <\lambda;\end{eqnarray}
(1.2)

\begin{equation}\label{1.3} \alpha_2|u|^p-\beta_2\leqslant f_2(u)u \leqslant \gamma_2|u|^p+\delta_2, p\geqslant 2, ~\mbox{且}~ f_2^\prime(u)\geqslant -c_1;\end{equation}
(1.3)

\begin{equation}\label{1.4} a\in L^1(\mathbb{R} ^n)\cap L^\infty(\mathbb{R} ^n), ~a(x)>0, \end{equation}
(1.4)

其中\alpha_i, \beta_i, \gamma_i, \delta_i (i=1, 2)c_1均为正常数.

F_1(u)=\int_0^u{f_1(s){\rm d}s}, ~~ F_2(u)=\int_0^u{f_2(s){\rm d}s}.

根据(1.2)-(1.3)式,可知存在正常数\tilde{\alpha}_i, \tilde{\beta}_i, \tilde{\gamma}_i, \tilde{\delta}_i (i=1, 2),使得

\begin{equation}\label{1.5} \tilde{\alpha}_1|u|^p-\tilde{\beta}_1|u|^2\leqslant F_1(u)\leqslant \tilde{\gamma}_1|u|^p+\tilde{\delta}_1|u|^2, \end{equation}
(1.5)

\begin{equation}\label{1.6} \tilde{\alpha}_2|u|^p-\tilde{\beta}_2\leqslant F_2(u) \leqslant \tilde{\gamma}_2|u|^p+\tilde{\delta}_2.\end{equation}
(1.6)

具有衰退记忆的非经典扩散方程(1.1)源于流体力学,固体力学和传导介质为黏性材料的热传导理论,可参见文献[1-2, 16, 23-24]及相关文献.通过方程中函数\Delta u(\cdot)和记忆核k(\cdot)的线性卷积项,对应于(1.1)式的动力系统的能量耗散被衰退记忆所影响.因此,能量的传播不仅受到现时外力的影响,而且还受到历史外力的影响,并且能量耗散的速度要比通常的非经典扩散方程(不包含衰退记忆项)更快.

类似于文献[8],设k(\cdot)\in C^2(\mathbb{R} ^+), k(s)\geqslant 0, k(0)=k_0>0, k(\infty)=0, k^\prime(s)\leqslant 0, \forall s\in \mathbb{R}^+.进一步,设\mu(s)=-k^\prime(s)\mu(s)满足

\begin{eqnarray}\label{1.7}& \mu\in C^1(\mathbb{R} ^+)\cap L^1(\mathbb{R} ^+), ~~ \mu(s)\geqslant 0, ~~\mu^{\prime }(s)\leqslant 0, ~~\forall s\in \mathbb{R} ^+; \end{eqnarray}
(1.7)

\begin{eqnarray} \label{1.8} &\mu^\prime(s)+\delta\mu(s)\leqslant 0, ~~\forall s\geqslant 0, \end{eqnarray}
(1.8)

其中\delta为正常数,根据(1.8)式,可知

\begin{equation}\label{1.9} 0\leqslant \mu(s)\leqslant \mu(0){\rm e}^{-\delta s}, \end{equation}
(1.9)

这里蕴含着核\mu(s)沿指数速度衰退于零.

近二十年来,许多学者致力于研究动力系统吸引子理论,参见文献[6, 9, 11-13, 15, 22].关于通常的非经典扩散方程(不包含衰退记忆)的解在有界域中的渐近性态更是成为研究热点问题,见文献[3-4, 14, 20-21, 26-27].当非线性项满足次临界增长且外力项属于L^2(\Omega)时,作者在文献[26]中证明了全局吸引子在H_0^1(\Omega)中的存在性.当非线性项满足临界增长,对于自治情形当外力项仅属于H^{-1}(\Omega)时,作者在文献[21]中获得了全局吸引子的存在性;对于非自治情形,当外力项仅属于L_b^2(\mathbb{R} ;L^2(\Omega))时作者在文献[21]中还获得了一致吸引子和指数吸引子的存在性.基于以上这些结果,当非线性项满足临界指数增长时,我们研究了另一种模型,即带有衰退记忆的非经典扩散方程,在弱拓扑空间和强拓扑空间对应的有界域中分别讨论了解的动力学行为.对于自治情形,当外力项仅属于H^{-1}(\Omega)L^2(\Omega)时,我们在文献[23]中获得了全局吸引子的存在性和正则性;对于非自治情形,当外力项仅属于g(x, t)\in L_b^2(\mathbb{R} ;~L^2(\Omega))g(x, t)\in L_b^2(\mathbb{R} ;~H_0^1(\Omega))时,我们在文献[24-25]中得到了紧的一致吸引子及其结构和正则性结果.

值得注意的是以上结果均在非线性项满足临界增长或次临界增长时获得,当非线性项满足超临界增长时吸引子是否存在则成为一个值得探讨的新问题.正如我们所知,超临界非线性项在解半群的紧性验证中将会导致一些本质性困难.关于带有超临界非线性项的通常的非经典扩散方程(没有衰退记忆),文献[27]中,作者在无界域中当外力项属于H^{-1}(\mathbb{R} ^n)时讨论了此问题,并且得到了全局吸引子在H^1(\mathbb{R} ^n)中的存在性.受此结果启发,我们将在无界域中考虑带有衰退记忆和超临界非线性项的问题(1.1).在研究中我们发现衰退记忆,超临界非线性项和无界域将会带来一些无法回避的本质性研究困难.首先,因为Sobolov嵌入在无界域中非紧,使得紧性验证变得更加困难;其次,经典反应扩散方程不包含-\nu\Delta u_t项.这意味着经典问题具有某种光滑性,即尽管初值属于弱拓扑空间,解可以属于具有更高正则性的强拓扑空间.然而对于非经典问题,因为包含-\nu\Delta u_t项,当初值属于弱拓扑空间时,解仅属于弱拓扑空间,因而导致不能运用紧嵌入定理验证解半群(过程)的渐近紧性;其三,非紧的记忆空间L_\mu^2(\mathbb{R} ^+; H^1(\mathbb{R} ^n))也会给紧性验证带来困难,主要表现在不能使用试验函数(I-P_m)u验证解半群的紧性;最后,超临界非线性项在解半群的紧性验证中会带来本质性障碍.总之,由于衰退记忆和超临界非线性项在无界域中验证验证解半群的性质(紧性,渐近紧性,连续性)会变得更加困难.

运用半群理论和收缩函数方法,我们克服了由衰退记忆,超临界非线性项和无界域导致的本质性研究困难.最终我们证明了本文的主要结果,即H^1(\mathbb{R} ^n)\times L_\mu^2(\mathbb{R} ^+;~H^1(\mathbb{R} ^n))中全局吸引子的存在性(定理).

为了叙述方便,在下文中C (或者C_i)均表示正常数.

本文结构如下:在第二节,我们引入一些符号和函数空间,给出将要用到的预备结果;在第三节,证明了主要结果,即H^1(\mathbb{R} ^n)\times L_\mu^2(\mathbb{R} ^+;~ H^1(\mathbb{R} ^n))中全局吸引子的存在性.

2 预备结果

在本节中,我们将回顾一些符号,函数空间和预备结果.

如同文献[8],设

\begin{equation}\label{2.1}\eta^{t}(x, s)=\int_0^s{u(x, t-r){\rm d}r}, ~ s\geqslant0, \end{equation}
(2.1)

\begin{equation}\label{2.2}\partial_t\eta^{t}(x, s)=u(x, t)-\partial_s\eta^{t}(x, s), ~ s\geqslant0.\end{equation}
(2.2)

\mu (s)=-k^{\prime }(s),则根据假设k(\infty)=0,可得如下问题

\begin{equation}\label{2.3}\left\{\begin{array}{ll} u_{t}-\nu\Delta u_t-\Delta u+\lambda u-\int_0^\infty{\mu (s)\Delta\eta ^{t}(s){\rm d}s}+f_1(u)+a(x)f_2(u)=g(x), \\[2mm]\eta _t^t=-\eta _s^t+u, \end{array}\right.\end{equation}
(2.3)

并且相应的初值条件为

\begin{equation}\label{2.4}\left\{\begin{array}{ll}u(x, t)=u_0(x, t), &(x, t)\in \mathbb{R} ^n\times(-\infty, 0] , \\ \eta ^0(x, s)=\eta _0(x, s)=\int_0^s{u(x, -r){\rm d}r}, &(x, s)\in \mathbb{R} ^n\times \mathbb{R} ^+, \end{array}\right.\end{equation}
(2.4)

其中u(\cdot)满足:存在正常数\mathcal R\varrho=\min\{\frac{\delta}{2}, \frac{\lambda_1}{2}\},使得

\int_0^\infty{{\rm e}^{-\varrho s}\| \nabla u(-s)\|^2{\rm d}s}\leqslant{\cal R},

其中\lambda_1-\Delta的第一特征值, \|\cdot\|表示L^2(\mathbb{R} ^n)范数.易知问题(2.3)-(2.4)等价于问题(1.1).

我们将使用Pata和Squassina[17]中的一些符号.设

A=-\Delta, ~~\mbox{其域}~~D(A)= H^2(\mathbb{R} ^n).

对于s\in\mathbb{R} ,带着标准内积的Hilbert空间族D(A^{\frac{s}{2}})

\langle \cdot, ~\cdot\rangle_{D(A^{\frac{s}{2}})}=\langleA^{\frac{s}{2}}\cdot, A^{\frac{s}{2}}\cdot\rangle

及范数

\|\cdot\|_{D(A^{\frac{s}{2}})}=\|A^{\frac{s}{2}}\cdot\|,

这里\langle\cdot, ~\cdot\rangle\|\cdot\|分别表示L^2(\mathbb{R} ^n)的内积与范数.

对于0\leqslant s <3,设

{\cal H}_s=D(A^{\frac{s}{2}}), ~\mbox{范数}~\|\cdot\|_{{\cal H}_s}=\|\cdot\|_{D(A^{\frac{s}{2}})},

{\cal H}_0=L^2(\mathbb{R} ^n), {\cal H}_1=H^1(\mathbb{R} ^n){\cal H}_2= H^2(\mathbb{R} ^n).

L_\mu^2(\mathbb{R} ^+; {\cal H}_r)为函数\varphi:~\mathbb{R} ^+\rightarrow {\cal H}_r, 0 < r <3构成的Hilbert空间族,赋予内积

\langle \varphi_1, \varphi_2\rangle_{\mu, {\mathcalH}_r}=\int_0^\infty{\mu(s)\langle \varphi_1(s), \varphi_2(s)\rangle_{{\cal H}_r}{\rm d}s}

及范数

\|\varphi\|_{\mu, {\cal H}_r}^2=\int_0^\infty{\mu(s)\|\varphi(s)\|_{{\cal H}_r}^2{\rm d}s}.

下面我们引入Hilbert空间族

{\cal E}_r={\cal H}_{r}\times L_\mu ^2(\mathbb{R} ^+; {\mathcalH}_r),

赋予范数

\| z\| _{{\cal E}_r}^2=\| (u, \eta^t )\| _{{\cal E}_r}^2=\frac{1}{2}(\| u\|_{{\cal H}_{r}} ^2+\| \eta ^t\| _{\mu, {\cal H}_r}^2).

为了方便估计,我们需要以下预备结果(参见文献[5, 7, 18]).

引理2.1  设I=[0, T], \forall~ T>0.设记忆核函数\mu(s)满足(1.7)-(1.8)式,则对于任意的\eta^t\in C(I;L_\mu^2(\mathbb{R} ^+; {\cal H}_r)), 0 <r <3,存在常数\delta>0,使得

\begin{eqnarray}\label{2.5}\langle \eta^t, \eta_s^t\rangle_{\mu, {\mathcalH}_r}\geqslant\frac{\delta}{2}\|\eta^t\|_{\mu, {\cal H}_r}^2.\end{eqnarray}
(2.5)

我们还需要以下结果来证明解半群的渐近紧性和全局吸引子的存在性,并且此结果借助和运用了Sun, Cao和Duan[19]的思想和方法.

定义2.1[19]  设X为Banach空间, BX中的有界集.称函数\phi(\cdot, \cdot)定义于X\times XB\times B上的收缩函数,如果对于任意序列\{x_m\}_{m=1}^{\infty}\subset B,存在子列\{x_{m_k}\}_{k=1}^{\infty} \subset \{x_m\}_{m=1}^{\infty},使得

\lim\limits_{k\to \infty}\lim\limits_{l\to \infty}\phi(x_{m_k}, \, x_{m_l})=0.

并且, {\mathfrak C}(B)表示B\times B上的收缩函数集合.

引理2.2[19]  设\{S(t)\}_{t\geqslant 0}为作用于Banach空间(X, \, \|\cdot\|)上的半群,并且半群\{S(t)\}_{t\geqslant 0}拥有有界吸收集B_0.进一步,如果对于任意的\varepsilon>0,存在T=T(B_0, \varepsilon)\phi_{T}(\cdot, \cdot)\in {\mathfrak C}(B_0),使得

\begin{equation}\label{2.6}\|S(T)x-S(T)y\|\leqslant \varepsilon + \phi_{T}(x, \, y), \quad\forall x, y\in B_0, \end{equation}
(2.6)

其中\phi_{T}依赖于T.\{S(t)\}_{t\geqslant 0}X中渐近紧,即,对于任意的有界序列\{y_m\}_{m=1}^{\infty}\subset X\{t_m\},当m\to \infty, t_m \to \infty时, \{S(t_m)y_m\}_{m=1}^{\infty}X中相对紧.

引理2.3[6]  设\{S(t)\}_{t\geqslant 0}为Banach空间X上的连续半群.如果\{S(t)\}_{t\geqslant 0}X中渐近紧,则它在X中拥有全局吸引子.

3 H^1(\mathbb{R} ^n)\times L_\mu^2(\mathbb{R} ^+; H^1(\mathbb{R} ^n))中的全局吸引子

3.1 适定性

首先,我们给出具有衰退记忆的动力系统(2.3)-(2.4)解的定义.

定义3.1  设I=[0, T], \forall T>0.g\in H^{-1}(\mathbb{R} ^n)z_0\in{\cal E}_1.称二元组z=(u, \eta^t)带着初值z(0)=z_0满足

\begin{eqnarray}&&u\in C(I; {\cal H}_1)\capL^p(I; L^p(\mathbb{R} ^n)), \\&&u_t\in L^2(I; {\cal H}_1), \\&&\eta^t\in C(I; L_\mu^2(\mathbb{R} ^+; {\cal H}_1)), \\&&\eta_t^t+\eta_s^t\in L^\infty(I; L_\mu^2(\mathbb{R} ^+; {\cal H}_0))\capL^2(I; L_\mu^2(\mathbb{R} ^+; {\cal H}_1)), \end{eqnarray}

为问题(2.3)-(2.4)在区间I上的一个弱解,如果

\begin{eqnarray}\label{3.1}&&\langle u_t, \omega\rangle+\nu\langle u_t, \omega\rangle_{{\cal H}_1}+\langle u, \omega\rangle_{{\cal H}_1}+\lambda\langle u, \omega\rangle+\langle\eta^t, \omega\rangle_{\mu, {\cal H}_1}+\langle f_1(u), \omega\rangle+\langle a(x)f_2(u), \omega\rangle=\langle g, \omega\rangle, \\&&\langle \eta_t^t+\eta_s^t, \varphi\rangle_{\mu, {\cal H}_1}=\langle u, \varphi\rangle_{\mu, {\cal H}_1}, \end{eqnarray}
(3.1)

对于所有的\omega\in {\cal H}_1, \varphi\in L_\mu^2(\mathbb{R} ^+;{\cal H}_1)和几乎处处的 t\in I.

运用Galerkin逼近方法(参见文献[5, 18]),我们可以得到方程(2.3)-(2.4)的解在{\cal E}_1中的存在唯一性结果.

定理3.1 (存在唯一性)  若假设(1.2)-(1.4), (1.7)-(1.8)成立,且g\inH^{-1}(\mathbb{R} ^n),则对于任意给定的z_0\in {\cal E}_1及任意的T>0,方程(2.3)-(2.4)在{\cal E}_1中存在唯一解z=(u, \eta^t),并且满足

\begin{eqnarray}\label{3.2}&&u\in C([0, T]; {\cal H}_1)\cap L^p([0, T];L^p(\mathbb{R} ^n)), \\&&z\in C([0, T]; {\cal E}_1)\cap L^\infty([0, \infty); {\mathcalE}_1).\end{eqnarray}
(3.2)

进一步,解在{\cal E}_1中连续依赖于初值z_0=(u_0, \eta^0).

根据定理3.1,可定义解算子

S(t):~{\cal E}_1\rightarrow{\cal E}_1,

z_0=(u_0, \eta^0) \rightarrow (u(t), \eta^t)=z(t)=S(t)z_0.

显然, \{S(t)\}_{t\geqslant0}可构成半群.在下文中,用\{S(t)\}_{t\geqslant0}表示方程(2.3)-(2.4)对应的解半群.

3.2 有界吸收集

首先我们将证明方程(2.3)-(2.4)在{\cal E}_1中有界吸收集的存在性.

定理3.2  若假设(1.2)-(1.4), (1.7)-(1.8)成立,且g\in H^{-1}(\mathbb{R} ^n),则存在正常数\varrho_0,使得对于任意有界子集B\subset {\cal E}_1z_0\in B,有

\|S(t)z_0\|_{{\cal E}_1}\leqslant \varrho_0, ~~\mbox{对于所有的}\ t\geqslant t_0=t_0(\| B\|_{{\cal E}_1})

成立.

  将(2.3)式乘以u并且在\mathbb{R} ^n上积分,可得

\begin{eqnarray}\label{3.4}&&\frac{1}{2}\frac{\rm d}{{\rm d}t}\| u\| ^2+\frac{\nu}{2}\frac{\rm d}{{\rm d}t}\| \nabla u\| ^2+\|\nabla u\|^2+\lambda\| u\|^2-\int_0^\infty {\mu (s)\langle\Delta\eta ^t(s), u(t)\rangle{\rm d}s}\\ &=& -\langle f_1(u), u\rangle-\langle a(x)f_2(u), u\rangle+\langle g(x), u\rangle.\end{eqnarray}
(3.3)

根据u(x, t)=\eta_t^{t}(x, s)+\eta_s^t(x, s),可知

\begin{eqnarray}\label{3.5}-\int_0^\infty{\mu(s)\langle \Delta\eta^t(x, s), u(t)\rangle {\rm d}s}&=&-\int_0^\infty{\mu(s)\langle\Delta\eta^t(x, s), \eta_t^{t}(x, s)+\eta_s^t(x, s)\rangle {\rm d}s}\\&=&\int_0^\infty{\mu(s)\langle\nabla \eta^t(x, s), \nabla \eta_t^{t}(x, s)+\nabla \eta_s^t(x, s)\rangle {\rm d}s}.\end{eqnarray}
(3.4)

\mathbb{R} ^n上关于x分部积分,可得

\begin{eqnarray}\label{3.6}\int_0^\infty{\mu(s)\langle\nabla \eta^t(x, s), \nabla \eta_t^{t}(x, s)\rangle {\rm d}s}&=&\frac{1}{2}\int_0^\infty{\mu(s)\frac{\rm d}{{\rm d}t}\|\nabla \eta^t(x, s)\|^2{\rm d}s}\\&=&\frac{1}{2}\frac{\rm d}{{\rm d}t}\int_0^\infty{\mu(s)\|\nabla\eta^t(x, s)\|^2{\rm d}s}.\end{eqnarray}
(3.5)

运用引理2.1,得

\begin{eqnarray}\label{3.7}\int_0^\infty{\mu(s)\langle\nabla \eta^t(x, s), \nabla \eta_s^{t}(x, s)\rangle {\rm d}s}\geqslant\frac{\delta }{2}\|\eta^t\|_{\mu, {\cal H}_1}^2.\end{eqnarray}
(3.6)

同时根据(1.2)-(1.4)式,有

\begin{eqnarray}\label{3.8}&& -\langle {f_1(u), u} \rangle-\langle a(x)f_2(u), u \rangle+\langle g(x), u \rangle \\ &\leqslant&\int_{\mathbb{R} ^n}{(\beta_1|u|^2-\alpha_1|u|^p){\rm d}x}+ \int_{\mathbb{R} ^n}{a(x)(\beta_2-\alpha_2|u|^p){\rm d}x}+\int_{\mathbb{R} ^n}{|g||u|{\rm d}x} \\ &\leqslant &\beta_1\|u\|^2-\alpha_1\int_{\mathbb{R} ^n}{|u|^p{\rm d}x}+\beta_2\|a\|_{L^1(\mathbb{R} ^n)}-\alpha_2\int_{\mathbb{R} ^n}{a(x)|u|^p{\rm d}x} \\&& +\frac{1}{2}\|\nabla u\|^2 +\frac{1}{2}\|g\|_{H^{-1}(\mathbb{R} ^n)}^2.\end{eqnarray}
(3.7)

则将上述估计代入(3.3)式,得到

\begin{eqnarray}\label{3.9}&&\frac{1}{2}\frac{\rm d}{{\rm d}t}(\| u\|^2+\nu\|\nabla u\|^2+\|\eta^t\|_{\mu, { \mathcalH}_1}^2)+(\lambda-\beta_1)\|u\|^2+\frac{1}{2}\| \nabla u\|^2+\frac{\delta}{2}\|\eta^t\|_{\mu, {\cal H}_1}^2\\&&+\alpha_1\int_{\mathbb{R} ^n}{|u|^p{\rm d}x}+\alpha_2\int_{\mathbb{R} ^n}{a(x)|u|^p{\rm d}x}\\&\leqslant&\frac{1}{2}\| g\|_{H^{-1}(\mathbb{R} ^n)}^2+\beta_2\|a\|_{L^1(\mathbb{R} ^n)}.\end{eqnarray}
(3.8)

定义

\begin{eqnarray}\label{3.10}{\cal L}(t)=\frac{1}{2}(\| u\|^2+\nu\|\nabla u\|^2+\|\eta^t\|_{\mu, {\cal H}_1}^2).\end{eqnarray}
(3.9)

\alpha=\min\{2(\lambda-\beta_1), \frac{1}{\nu}, \delta\},可以得到

\frac{\rm d}{{\rm d}t}{\cal L}(t)+\alpha {\cal L}(t)\leqslant\frac{1}{2}\| g\|_{H^{-1}(\mathbb{R} ^n)}^2+\beta_2\|a\|_{L^1(\mathbb{R} ^n)}.

根据Gronwall引理,有

\begin{equation}\label{3.11}{\cal L}(t) \leqslant {\cal L}(0){\rm e}^{-\alpha t}+\frac{1}{\alpha}(\frac{1}{2}\| g\|_{H^{-1}(\mathbb{R} ^n)}^2+\beta_2\|a\|_{L^1(\mathbb{R} ^n)}).\end{equation}
(3.10)

由于(3.9)式,可知存在常数C_1, C_2,使得

\begin{equation}\label{3.12}C_1\|z(t)\|^2_{{\cal E}_1}\leqslant{\cal L}(t)\leqslant C_2\|z(t)\|^2_{{\cal E}_1}.\end{equation}
(3.11)

\|z(0)\|^2_{{\cal E}_1}\leqslant R_0,当t\geqslant t_0=\frac{1}{\alpha}\ln\frac{\alpha C_2R_0}{\frac{1}{2}\| g\|_{H^{-1}(\mathbb{R} ^n)}^2+\beta_2\|a\|_{L^1(\mathbb{R} ^n)}}时,可得

\|z(t)\|^2_{{\cal E}_1}\leqslant \frac{2}{C_1\alpha}\Big (\frac{1}{2}\| g\|_{H^{-1}(\mathbb{R} ^n)}^2+\beta_2\|a\|_{L^1(\mathbb{R} ^n)}\Big)=\varrho_0^2.

\begin{equation}\label{3.13} \|z(t)\|_{{\cal E}_1}= \Big(\frac{1}{2}(\|u\|_{{\cal H}_1}^2+\|\eta^t\|_{\mu, {\cal H}_1}^2) \Big)^{\frac{1}{2}} \leqslant \varrho_0 \end{equation}
(3.12)

成立.证毕.

3.3 全局吸引子的存在性

为了得到全局吸引子的存在性,我们还需要证明以下预备结果.

引理3.1  设B为{\cal E}_1中的任意有界子集,并且z(t)为方程(2.3)-(2.4)在{\mathcalE}_1中带着初值z_0\in B的解.若假设(1.2)-(1.4), (1.7)-(1.8)成立,且g\in H^{-1}(\mathbb{R} ^n),存则在常数N_0>0,对于任意的t\geqslant t_0,使得

\begin{equation}\label{3.14}\int_t^{t+1}{\|\nabla u(s)\|^2{\rm d}s}\leqslant N_0, \end{equation}
(3.13)

\begin{equation}\label{3.15}\int_t^{t+1}{\|u(s)\|_{L^p(\mathbb{R} ^n)}^p{\rm d}s}\leqslant N_0, \end{equation}
(3.14)

\begin{equation}\label{3.16}\int_t^{t+1}{\int_{\mathbb{R} ^n}a(x)|u(s)|^p{\rm d}x{\rm d}s}\leqslant N_0.\end{equation}
(3.15)

  关于(3.8)式在[t, t+1]上积分并且利用(3.12)式,可得

\begin{eqnarray}\label{3.17}&& (\lambda-\beta_1)\int_t^{t+1}{\|u(s)\|^2{\rm d}s}+\frac{1}{2}\int_t^{t+1}{\|\nabla u(s)\|^2{\rm d}s}+\frac{\delta}{2}\int_t^{t+1}{\|\eta^s(r)\|_{\mu, {\cal H}_1}^2{\rm d}s}\\&&+\alpha_1\int_t^{t+1}{\|u(s)\|_{L^p(\mathbb{R} ^n)}^p{\rm d}s}+\alpha_2\int_t^{t+1}{\int_{\mathbb{R} ^n}a(x)|u(s)|^p{\rm d}x{\rm d}s}\\&\leqslant&\frac{1}{2}\| g\|_{H^{-1}(\mathbb{R} ^n)}^2+\beta_2\|a\|_{L^1(\mathbb{R} ^n)}+C\varrho^2_0, ~~ t\geqslant t_0, \end{eqnarray}
(3.16)

故(3.13)-(3.15)式成立.证毕.

引理3.2  若引理3.1的假设成立,则存在常数N_1>0及时刻t_1\geqslant t_0,使得对于任意的z_0\in Bt\geqslant t_1,以下估计

\begin{eqnarray}\label{3.18}\|u(t)\|_{L^p(\mathbb{R} ^n)}^p\leqslant N_1\end{eqnarray}
(3.17)

成立.

  选取u_t作为试验函数,可得

\begin{eqnarray}\label{3.19}&&\|u_t\|^2+\nu\|\nabla u_t\|^2+\frac{\nu}{2}\frac{\rm d}{{\rm d}t}\| \nabla u\| ^2+\frac{\lambda}{2}\frac{\rm d}{{\rm d}t}\| u\|^2+\langle f_1(u), u_t\rangle+\langle a(x)f_2(u), u_t\rangle\\& =&\int_0^\infty {\mu (s)\langle\Delta\eta ^t(s), u_t\rangle{\rm d}s}+\langle g(x), u_t\rangle.\end{eqnarray}
(3.18)

易知

\begin{eqnarray}\label{3.20}\int_0^\infty {\mu (s)\langle\Delta\eta ^t(s), u_t\rangle {\rm d}s}&=&-\int_0^\infty {\mu (s)\langle\nabla\eta ^t(s), \nabla u_t\rangle{\rm d}s}\\ &\leqslant& \frac{\nu}{4}\|\nabla u_t\|^2+\frac{1}{\nu}\|\eta^t\|_{\mu, {\cal H}_1}^2\end{eqnarray}
(3.19)

\begin{equation}\label{3.21}\langle g(x), u_t\rangle\leqslant \frac{\nu}{4}\|\nabla u_t\|^2+\frac{1}{\nu}\|g\|_{H^{-1}(\mathbb{R} ^n)}^2.\end{equation}
(3.20)

根据(3.12)式并且运用Poincaré表达式,可得

\|u\|^2\leqslant \frac{1}{\lambda_1}\|\nabla u\|^2\leqslant\frac{2}{\lambda_1}\varrho_0^2.

利用假设(1.5)-(1.6),有

\begin{eqnarray}\label{3.22}&& \langle f_1(u), u_t\rangle+\langle a(x)f_2(u), u_t\rangle\\&=&\frac{\rm d}{{\rm d}t}\int_{\mathbb{R} ^n}{F_1(u){\rm d}x}+\frac{\rm d}{{\rm d}t}\int_{\mathbb{R} ^n}{a(x)F_2(u){\rm d}x}\\&=&\frac{\rm d}{{\rm d}t}\bigg(\int_{\mathbb{R} ^n}F_1(u){\rm d}x+\frac{2\tilde{\beta}_1\varrho^2_0}{\lambda_1}\bigg)+\frac{\rm d}{{\rm d}t}\int_{\mathbb{R} ^n}{a(x)(F_2(u)+\tilde{\beta}_2){\rm d}x}.\end{eqnarray}
(3.21)

根据(1.4)-(1.6)式,可以得到

\int_{\mathbb{R} ^n}F_1(u){\rm d}x+\frac{2\tilde{\beta}_1\varrho^2_0}{\lambda_1}\geqslant\int_{\mathbb{R} ^n}(F_1(u)+\tilde{\beta}_1|u|^2){\rm d}x\geqslant\int_{\mathbb{R} ^n}\tilde{\alpha}_1|u|^p {\rm d}x \geqslant 0,

\int_{\mathbb{R} ^n}a(x)(F_2(u)+\tilde{\beta}_2){\rm d}x\geqslant\int_{\mathbb{R} ^n}\tilde{\alpha}_2a(x)|u|^p {\rm d}x \geqslant 0.

{\cal M}(t)=\frac{\nu}{2}\|\nabla u\|^2+\frac{\lambda}{2}\|u\|^2+\int_{\mathbb{R} ^n} F_1(u){\rm d}x+\frac{2\tilde{\beta}_1\varrho^2_0}{\lambda_1}+\int_{\mathbb{R} ^n} a(x)(F_2(u){\rm d}x+\tilde{\beta}_2){\rm d}x.

则, {\cal M}(t)非负.将其代入(3.18)式,可得

\begin{equation}\label{3.23}\|u_t\|^2+\frac{\nu}{2}\|\nabla u_t\|^2+\frac{\rm d}{{\rm d}t}{\cal M}(t) \leqslant\frac{1}{\nu}(2\varrho^2_0+\|g\|_{H^{-1}(\mathbb{R} ^n)}^2)\leqslant C, \end{equation}
(3.22)

它蕴含着

\begin{equation}\label{3.24}\frac{\rm d}{{\rm d}t}{\cal M}(t)\leqslant C.\end{equation}
(3.23)

因此,根据(3.13)-(3.15)式和引理3.1,可得

\begin{equation}\label{3.25}\int_t^{t+1}{\cal M}(s){\rm d}s\leqslant C.\end{equation}
(3.24)

进一步,根据一致Gronwall引理及(3.22)-(3.23)式,可得

{\cal M}(t+1)\leqslant C, ~~ t\geqslant t_0,

\begin{equation}\label{3.26}{\cal M}(t)\leqslant C, ~~ t\geqslant t_1=t_0+1.\end{equation}
(3.25)

因此,我们得到

\begin{equation}\label{3.27}\int_{\mathbb{R} ^n}\tilde{\alpha}_1|u|^p {\rm d}x\leqslant\int_{\mathbb{R} ^n}F_1(u){\rm d}x+\frac{2\tilde{\beta}_1\varrho^2_0}{\lambda_1}\leqslant C.\end{equation}
(3.26)

\int_{\mathbb{R} ^n}|u|^p {\rm d}x\leqslant N_1

成立.证毕.

为了叙述简便,在下文中设B_0{\cal E}_1中的有界吸收集(由定理3.2和引理3.2)而得),即

B_0=\{z|z=(u, \eta^t), z\in{\cal E}_1~\mbox{且} ~u\in L^p(\mathbb{R} ^n), \|z\|_{{\cal E}_1}\leqslant \varrho_0~\mbox{且}~ \|u\|_{L^p(\mathbb{R} ^n)}^p\leqslant N_1 \}.

引理3.3  设B_0({\cal E}_1, L^p(\mathbb{R} ^n))中的有界吸收集,并且z(t)为方程(2.3)-(2.4)在{\cal E}_1中带着初值z_0\in B_0的解.若g\in H^{-1}(\mathbb{R} ^n), (1.7)-(1.8)式成立且非线性项f(x, u)=f_1(u)+a(x)f_2(u)满足(1.2)-(1.4)式.则存在正常数N_2,使得对于任意的z_0\in B_0及任意给定的T\geqslant t_2\geqslant t_1,以下估计

\begin{equation}\label{3.28}\int_0^T(\|u_t(s)\|^2+\|\nabla u_t(s)\|^2){\rm d}s\leqslant N_2\end{equation}
(3.27)

成立.

  对于任意给定的T\geqslant t_2,将(3.22)式关于t0T积分,得到

\begin{eqnarray}\label{3.29}&&\int_0^T(\|u_t(s)\|^2+\frac{\nu}{2}\|\nabla u_t(s)\|^2){\rm d}s\\&\leqslant& {\cal M}(0)-{\cal M}(T)+CT\\&\leqslant&\frac{\nu}{2}\|\nabla u(0)\|^2+\frac{\lambda}{2}\|u(0)\|^2+\int_{\mathbb{R} ^n} F_1(u(0)){\rm d}x+\frac{2\tilde{\beta}_1\varrho_0^2}{\lambda_1}\\&&+\int_{\mathbb{R} ^n} a(x)(F_2(u(0))+\tilde{\beta}_2){\rm d}x+CT\\&\leqslant&({\nu}+\frac{\lambda}{\lambda_1})\varrho^2_0+CT+\int_{\mathbb{R} ^n} F_1(u(0)){\rm d}x+\frac{2\tilde{\beta}_1\varrho^2_0}{\lambda_1}\\&&+\int_{\mathbb{R} ^n} a(x)(F_2(u(0))+\tilde{\beta}_2){\rm d}x.\end{eqnarray}
(3.28)

利用(1.5)-(1.6)式,可得

\begin{eqnarray}\label{3.30}&&\int_{\mathbb{R} ^n} F_1(u(0)){\rm d}x+\frac{2\tilde{\beta}_1\varrho^2_0}{\lambda_1}+\int_{\mathbb{R} ^n} a(x)(F_2(u(0))+\tilde{\beta}_2){\rm d}x\\&\leqslant&\int_{\mathbb{R} ^n}(\tilde{\gamma}_1|u(0)|^p+\tilde{\delta}_1|u(0)|^2){\rm d}x+\frac{2\tilde{\beta}_1\varrho^2_0}{\lambda_1}+\int_{\mathbb{R} ^n}a(x)(\tilde{\gamma}_2|u(0)|^p+\tilde{\delta}_2+\tilde{\beta}_2){\rm d}x\\&\leqslant&(\tilde{\gamma}_1+\|a\|_{L^{\infty}(\mathbb{R} ^n)}\tilde{\gamma}_2)N_1+\frac{2(\tilde{\delta}_1+\tilde{\beta}_1)\varrho^2_0}{\lambda_1}+(\tilde{\delta}_2+\tilde{\beta}_2)\|a\|_{L^{1}(\mathbb{R} ^n)}.\end{eqnarray}
(3.29)

显然(3.27)式成立.证毕.

引理3.4  设\beta=\min\{\lambda, \frac{1}{\nu}, \delta\}l=\frac{32c_1}{{C}_1\beta}(1+\|a\|_{L^{\infty}(\mathbb{R} ^n)}).若引理3.3的假设成立,则对于任意的z_0\in B_0\epsilon>0,存在\tilde{k}(\epsilon)\tilde{T}(\epsilon),使得当k\geqslant \tilde{k}(\epsilon)t\geqslant\tilde{T}(\epsilon)时,有

\int_{|x|\geqslant\sqrt{2}k}| u(t)|^2{\rm d}x\leqslant \frac{16}{\beta l}\epsilon

成立.

  选取光滑函数

\begin{equation}\label{3.31}\theta(s)=\left\{\begin{array}{ll}0, ~~& 0\leqslant s\leqslant 1, \\1, &s\geqslant 2.\end{array}\right.\end{equation}
(3.30)

显然

\begin{eqnarray}0\leqslant \theta(s)\leqslant 1, ~~ 1\leqslant s \leqslant 2, \end{eqnarray}
(3.31)

且存在正常数c,使得|\theta^{\prime}(s)|\leqslant c.

将(2.3)式乘以\theta^2(\frac{|x|^2}{k^2})u并且在\mathbb{R} ^n上积分,可得

\begin{eqnarray}\label{3.32}&&\frac{1}{2}\frac{\rm d}{{\rm d}t}\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})| u| ^2{\rm d}x-\nu\int_{\mathbb{R} ^n}\Delta u_t\cdot\theta^2(\frac{|x|^2}{k^2}) u{\rm d}x-\int_{\mathbb{R} ^n}\Delta u\cdot\theta^2(\frac{|x|^2}{k^2}) u{\rm d}x\\&&+\int_{\mathbb{R} ^n}\lambda u\cdot\theta^2(\frac{|x|^2}{k^2}) u{\rm d}x-\int_{\mathbb{R} ^n}\int_0^\infty\mu(s)\Delta \eta^t(x, s)\theta^2(\frac{|x|^2}{k^2}) u{\rm d}s{\rm d}x\\&=&-\int_{\mathbb{R} ^n}f_1(u)\theta^2(\frac{|x|^2}{k^2}) u{\rm d}x-\int_{\mathbb{R} ^n}a(x)f_2(u)\theta^2(\frac{|x|^2}{k^2}) u{\rm d}x+\int_{\mathbb{R} ^n}g(x)\theta^2(\frac{|x|^2}{k^2}) u{\rm d}x.\end{eqnarray}
(3.32)

下面我们将估计等式(3.32)左侧第二项至第五项.

首先

\begin{eqnarray}\label{3.33}&&-\nu\int_{\mathbb{R} ^n}\Delta u_t\cdot\theta^2(\frac{|x|^2}{k^2})u{\rm d}x\\&=&\nu\int_{\mathbb{R} ^n}\nabla u_t\cdot2\theta(\frac{|x|^2}{k^2})\theta^\prime(\frac{|x|^2}{k^2})\frac{2|x|}{k^2}u{\rm d}x+\nu\int_{\mathbb{R} ^n}\nabla u_t\cdot\theta^2(\frac{|x|^2}{k^2})\nabla u{\rm d}x\\&=&\nu\int_{\mathbb{R} ^n}\nabla u_t\cdot2\theta(\frac{|x|^2}{k^2})\theta^\prime(\frac{|x|^2}{k^2})\frac{2|x|}{k^2}u{\rm d}x+\frac{\nu}{2}\frac{\rm d}{{\rm d}t}\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})|\nabla u|^2{\rm d}x.\end{eqnarray}
(3.33)

进一步,根据定理3.2和条件|\theta^{\prime}(s)|\leqslant c,对于t\geqslant t_0,有

\begin{eqnarray}\label{3.34}&&\nu\int_{\mathbb{R} ^n}\nabla u_t\cdot2\theta(\frac{|x|^2}{k^2})\theta^\prime(\frac{|x|^2}{k^2})\frac{2|x|}{k^2}u{\rm d}x \\&\leqslant&\nu|\int_{\mathbb{R} ^n}\nabla u_t\cdot2\theta(\frac{|x|^2}{k^2})\theta^\prime(\frac{|x|^2}{k^2})\frac{2|x|}{k^2}u{\rm d}x|\\&\leqslant&\nu|\int_{k\leqslant|x|\leqslant \sqrt{2}k}\nabla u_t\cdot2\theta(\frac{|x|^2}{k^2})\theta^\prime(\frac{|x|^2}{k^2})\frac{2|x|}{k^2}u {\rm d}x|\\&\leqslant&\frac{4\sqrt{2}c\nu}{k}\int_{k\leqslant|x|\leqslant \sqrt{2}k} \theta(\frac{|x|^2}{k^2})|\nabla u_t| \cdot|u|{\rm d}x\\&\leqslant&\frac{2\sqrt{2}c\nu}{k}\bigg(\int_{\mathbb{R} ^n} \theta^2(\frac{|x|^2}{k^2})|\nabla u_t|^2{\rm d}x+\int_{\mathbb{R} ^n}|u|^2{\rm d}x\bigg)\\&\leqslant&\frac{2\sqrt{2}c\nu}{k}\int_{\mathbb{R} ^n} \theta^2(\frac{|x|^2}{k^2})|\nabla u_t|^2{\rm d}x+\frac{4\sqrt{2}c\nu}{k\lambda_1}\varrho^2_0, \end{eqnarray}
(3.34)

它蕴含着

\begin{eqnarray}\label{3.35}&&-\nu\int_{\mathbb{R} ^n}\Delta u_t\cdot\theta^2(\frac{|x|^2}{k^2})u{\rm d}x\\&\geqslant&\frac{\nu}{2}\frac{\rm d}{{\rm d}t}\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})|\nabla u|^2{\rm d}x-\frac{2\sqrt{2}c\nu}{k}\int_{\mathbb{R} ^n} \theta^2(\frac{|x|^2}{k^2})|\nabla u_t|^2{\rm d}x-\frac{4\sqrt{2}c\nu}{k\lambda_1}\varrho^2_0\\&\geqslant&\frac{\nu}{2}\frac{\rm d}{{\rm d}t}\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})|\nabla u|^2{\rm d}x-\frac{2\sqrt{2}c\nu}{k}\int_{\mathbb{R} ^n} \theta^2(\frac{|x|^2}{k^2})|\nabla u_t|^2{\rm d}x-\frac{\nu \tilde{C}}{k}, \end{eqnarray}
(3.35)

其中\tilde{C}=\frac{4\sqrt{2}c}{\lambda_1}\varrho_0^2且独立于k.

其次,我们来估计等式(3.32)左侧第三项,可得

\begin{eqnarray}\label{3.36}-\int_{\mathbb{R} ^n}\Delta u\cdot\theta^2(\frac{|x|^2}{k^2})u{\rm d}x&=&\int_{\mathbb{R} ^n}\nabla u\cdot2\theta(\frac{|x|^2}{k^2})\theta^\prime(\frac{|x|^2}{k^2})\frac{2|x|}{k^2}u{\rm d}x+\int_{\mathbb{R} ^n}\nabla u\theta^2(\frac{|x|^2}{k^2})\nabla u{\rm d}x\\&=&\int_{\mathbb{R} ^n}\nabla u\cdot2\theta(\frac{|x|^2}{k^2})\theta^\prime(\frac{|x|^2}{k^2})\frac{2|x|}{k^2}u{\rm d}x+\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})|\nabla u|^2{\rm d}x.~~\end{eqnarray}
(3.36)

由于

\begin{eqnarray}\label{3.37}\int_{\mathbb{R} ^n}\nabla u\cdot2\theta(\frac{|x|^2}{k^2})\theta^\prime(\frac{|x|^2}{k^2})\frac{2|x|}{k^2}u{\rm d}x&\leqslant&\bigg|\int_{k\leqslant|x|\leqslant \sqrt{2}k}\nabla u\cdot2\theta(\frac{|x|^2}{k^2})\theta^\prime(\frac{|x|^2}{k^2})\frac{2|x|}{k^2}u{\rm d}x\bigg|\\&\leqslant&\frac{4\sqrt{2}c}{k}\int_{k\leqslant|x|\leqslant \sqrt{2}k} \theta(\frac{|x|^2}{k^2})|\nabla u| \cdot|u|{\rm d}x\\&\leqslant&\frac{2\sqrt{2}c}{k}\bigg(\int_{\mathbb{R} ^n} \theta^2(\frac{|x|^2}{k^2})|\nabla u|^2{\rm d}x+\int_{\mathbb{R} ^n}|u|^2{\rm d}x\bigg)\\&\leqslant&\frac{2\sqrt{2}c}{k}\int_{\mathbb{R} ^n} \theta^2(\frac{|x|^2}{k^2})|\nabla u|^2{\rm d}x+\frac{4\sqrt{2}c}{k\lambda_1}\varrho^2_0, ~~\end{eqnarray}
(3.37)

可得

\begin{eqnarray}\label{3.38}-\int_{\mathbb{R} ^n}\Delta u\cdot\theta^2(\frac{|x|^2}{k^2})u{\rm d}x&\geqslant&\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})|\nabla u|^2{\rm d}x-\frac{2\sqrt{2}c}{k}\int_{\mathbb{R} ^n} \theta^2(\frac{|x|^2}{k^2})|\nabla u|^2{\rm d}x-\frac{4\sqrt{2}c}{k\lambda_1}\varrho^2_0\\&\geqslant&\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})|\nabla u|^2{\rm d}x-\frac{2\sqrt{2}c}{k}\int_{\mathbb{R} ^n} \theta^2(\frac{|x|^2}{k^2})|\nabla u|^2{\rm d}x-\frac{\tilde{C}}{k}.\end{eqnarray}
(3.38)

联合(3.35)与(3.38)式,可知对于任意的\epsilon>0,若

\tilde{k}_1(\epsilon)=\max\bigg\{\frac{\tilde{C}l}{\epsilon}, \frac{\nu \tilde{C}l}{\epsilon}, 8\sqrt{2}c\bigg\},

则对于k\geqslant\tilde{k}_1(\epsilon),可得

\begin{eqnarray}\label{3.39}&&-\nu\int_{\mathbb{R} ^n}\Delta u_t\cdot\theta^2(\frac{|x|^2}{k^2})u{\rm d}x-\int_{\mathbb{R} ^n}\Delta u\cdot\theta^2(\frac{|x|^2}{k^2})u{\rm d}x\\&\geqslant&\frac{\nu}{2}\frac{\rm d}{{\rm d}t}\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})|\nabla u|^2{\rm d}x-\frac{2\sqrt{2}c\nu}{k}\int_{\mathbb{R} ^n} \theta^2(\frac{|x|^2}{k^2})|\nabla u_t|^2{\rm d}x-\frac{\nu C}{k}\\&&+\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})|\nabla u|^2{\rm d}x-\frac{2\sqrt{2}c}{k}\int_{\mathbb{R} ^n} \theta^2(\frac{|x|^2}{k^2})|\nabla u|^2{\rm d}x-\frac{C}{k}\\&\geqslant&\frac{\nu}{2}\frac{\rm d}{{\rm d}t}\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})|\nabla u|^2{\rm d}x-\frac{2\sqrt{2}c\nu}{k}\int_{\mathbb{R} ^n} \theta^2(\frac{|x|^2}{k^2})|\nabla u_t|^2{\rm d}x\\&&+\frac{3}{4}\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})|\nabla u|^2{\rm d}x-\frac{2}{l}\epsilon.\end{eqnarray}
(3.39)

然后,易知等式(3.32)左侧第四项满足

\begin{eqnarray}\label{3.40}\int_{\mathbb{R} ^n}\lambda u\cdot\theta^2(\frac{|x|^2}{k^2})u{\rm d}x=\lambda\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})| u|^2{\rm d}x.\end{eqnarray}
(3.40)

最后,我们发现等式(3.32)左侧第五项满足

\begin{eqnarray}\label{3.41}&&-\int_{\mathbb{R} ^n}\int_0^\infty\mu(s)\Delta\eta^t(x, s) \theta^2(\frac{|x|^2}{k^2})u(x, t){\rm d}s{\rm d}x\\&=&\int_{\mathbb{R} ^n}\int_0^\infty\mu(s)\nabla\eta^t(x, s) \cdot2\theta(\frac{|x|^2}{k^2})\theta^\prime(\frac{|x|^2}{k^2})\frac{2|x|}{k^2}u(x, t){\rm d}s{\rm d}x\\&&+\int_{\mathbb{R} ^n}\int_0^\infty\mu(s)\nabla\eta^t(x, s) \theta^2(\frac{|x|^2}{k^2})\nabla u(x, t){\rm d}s{\rm d}x.\end{eqnarray}
(3.41)

进一步,可知

\begin{eqnarray}\label{3.42}&&\bigg|\int_{\mathbb{R} ^n}\int_0^\infty\mu(s)\nabla\eta^t(x, s) \cdot2\theta(\frac{|x|^2}{k^2})\theta^\prime(\frac{|x|^2}{k^2})\frac{2|x|}{k^2}u(x, t){\rm d}s{\rm d}x\bigg|\\&\leqslant&\frac{4\sqrt{2}c}{k}\int_{k\leqslant|x|\leqslant\sqrt{2}k}\int_0^\infty\mu(s)\nabla\eta^t(x, s) \theta(\frac{|x|^2}{k^2}) u(x, t){\rm d}s{\rm d}x\\&\leqslant&\frac{2\sqrt{2}ck_0}{k}\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})|u|^2{\rm d}x+\frac{2\sqrt{2}c}{k}\int_{\mathbb{R} ^n}\int_0^\infty\mu(s)|\nabla\eta^t(x, s)|^2{\rm d}s{\rm d}x\\&\leqslant&\frac{2\sqrt{2}ck_0}{k}\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})|u|^2{\rm d}x+\frac{4\sqrt{2}c\varrho^2_0}{k}\\&\leqslant&\frac{2\sqrt{2}ck_0}{k}\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})|u|^2{\rm d}x+\frac{\lambda_1\tilde{C}}{k}.\end{eqnarray}
(3.42)

同时,根据引理2.1及u(x, t)=\eta_t^{t}(x, s)+\eta_s^t(x, s),可知

\begin{eqnarray}\label{3.43}&&\int_{\mathbb{R} ^n}\int_0^\infty\mu(s)\nabla\eta^t(x, s) \theta^2(\frac{|x|^2}{k^2})\nabla u{\rm d}s{\rm d}x\\&\geqslant& \frac{1}{2}\frac{\rm d}{{\rm d}t}\int_{\mathbb{R} ^n}\int_0^\infty\theta^2(\frac{|x|^2}{k^2})\mu(s)|\nabla\eta^t(x, s)|^2{\rm d}s{\rm d}x \\&&+\frac{\delta}{2}\int_{\mathbb{R} ^n}\int_0^\infty\theta^2(\frac{|x|^2}{k^2})\mu(s)|\nabla\eta^t(x, s)|^2{\rm d}s{\rm d}x.\end{eqnarray}
(3.43)

易知(3.42)和(3.43)式蕴含着

\begin{eqnarray}\label{3.44}&&-\int_{\mathbb{R} ^n}\int_0^\infty\mu(s)\Delta\eta^t(x, s) \theta^2(\frac{|x|^2}{k^2})u(x, t){\rm d}s{\rm d}x\\&\geqslant& \frac{1}{2}\frac{\rm d}{{\rm d}t}\int_{\mathbb{R} ^n}\int_0^\infty\theta^2(\frac{|x|^2}{k^2})\mu(s)|\nabla\eta^t(x, s)|^2{\rm d}s{\rm d}x+\frac{\delta}{2}\int_{\mathbb{R} ^n}\int_0^\infty\theta^2(\frac{|x|^2}{k^2})\mu(s)|\nabla\eta^t(x, s)|^2{\rm d}s{\rm d}x\\&&-\frac{2\sqrt{2}ck_0}{k}\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})|u|^2{\rm d}x-\frac{\lambda_1\tilde{C}}{k}.\end{eqnarray}
(3.44)

故对于任意的\epsilon>0,若k\geqslant\tilde{k}_2(\epsilon)=\max\{\frac{\lambda_1\tilde{C}l}{\epsilon}, \frac{8\sqrt{2}ck_0}{\lambda}, \tilde{k}_1(\epsilon)\},则

\begin{eqnarray}\label{3.45}&&-\int_{\mathbb{R} ^n}\int_0^\infty\mu(s)\Delta\eta^t(x, s) \theta^2(\frac{|x|^2}{k^2})u{\rm d}s{\rm d}x\\&\geqslant& \frac{1}{2}\frac{\rm d}{{\rm d}t}\int_{\mathbb{R} ^n}\int_0^\infty\theta^2(\frac{|x|^2}{k^2})\mu(s)|\nabla\eta^t(x, s)|^2{\rm d}s{\rm d}x+\frac{\delta}{2}\int_{\mathbb{R} ^n}\int_0^\infty\theta^2(\frac{|x|^2}{k^2})\mu(s)|\nabla\eta^t(x, s)|^2{\rm d}s{\rm d}x\\&&-\frac{\lambda}{4}\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})|u|^2{\rm d}x-\frac{1}{l}\epsilon.\end{eqnarray}
(3.45)

下面我们将估计等式(3.32)的右侧各项.由于a\in L^1(\mathbb{R} ^n),则存在\tilde{k}_3(\epsilon)\geqslant\tilde{k}_2(\epsilon),对于任意的k\geqslant\tilde{k}_3(\epsilon),使得

\begin{eqnarray}\label{3.46}\int_{|x|\geqslant k}|a(x)|{\rm d}x\leqslant\frac{\epsilon}{l\beta_2}.\end{eqnarray}
(3.46)

利用(1.2)-(1.4)式,可得

\begin{eqnarray}\label{3.47}&&-\int_{\mathbb{R} ^n}f_1(u)\theta^2(\frac{|x|^2}{k^2})u{\rm d}x-\int_{\mathbb{R} ^N}a(x)f_2(u)\theta^2(\frac{|x|^2}{k^2})u{\rm d}x\\&\leqslant&\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})a(x)(\beta_2-\alpha_2|u|^p){\rm d}x\\&\leqslant&\beta_2\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})a(x){\rm d}x\\&\leqslant&\beta_2\int_{|x|\geqslant k}a(x){\rm d}x\\&\leqslant&\frac{1}{l}\epsilon.\end{eqnarray}
(3.47)

由于g\in H^{-1}(\mathbb{R} ^n),则存在\tilde{k}_4(\epsilon)\geqslant\tilde{k}_3(\epsilon),使得对于任意的k\geqslant\tilde{k}_4(\epsilon),有

\begin{eqnarray}\label{3.48}\int_{|x|\geqslant k}|A^{-\frac{1}{2}}g|^2{\rm d}x\leqslant\frac{1}{l}\epsilon.\end{eqnarray}
(3.48)

\tilde{k}_5(\epsilon)=\max\bigg\{\tilde{k}_4(\epsilon), \frac{\lambda}{8\sqrt{2}c}, \frac{2\sqrt{2}cl\|g\|_{H^{-1}(\mathbb{R} ^n)}^2}{\epsilon}\bigg\},

则对于任意的k\geqslant\tilde{k}_5(\epsilon)|x|\geqslant k,可以得到

\begin{eqnarray}\label{3.49}&&\int_{\mathbb{R} ^n}g(x)\theta^2(\frac{|x|^2}{k^2})u{\rm d}x\\&=&\int_{\mathbb{R} ^n}A^{-\frac{1}{2}}g\cdot\nabla(\theta^2(\frac{|x|^2}{k^2})u){\rm d}x\\&=&\int_{\mathbb{R} ^n}A^{-\frac{1}{2}}g(2\theta(\frac{|x|^2}{k^2})\theta^\prime(\frac{|x|^2}{k^2})\frac{2|x|}{k^2}u+\theta^2(\frac{|x|^2}{k^2})\nabla u){\rm d}x\\&\leqslant&\int_{k\leqslant|x|\leqslant \sqrt{2}k}|A^{-\frac{1}{2}}g|\cdot2\theta(\frac{|x|^2}{k^2})\theta^\prime(\frac{|x|^2}{k^2})\frac{2|x|}{k^2}|u|{\rm d}x+\int_{\mathbb{R} ^n}|A^{-\frac{1}{2}}g|\theta^2(\frac{|x|^2}{k^2})|\nabla u|{\rm d}x\\&\leqslant&\frac{4\sqrt{2}c}{k}\int_{k\leqslant|x|\leqslant \sqrt{2}k} \theta(\frac{|x|^2}{k^2})|A^{-\frac{1}{2}}g| \cdot|u|{\rm d}x+\int_{\mathbb{R} ^n}|A^{-\frac{1}{2}}g|\theta^2(\frac{|x|^2}{k^2})|\nabla u|{\rm d}x\\&\leqslant&\frac{2\sqrt{2}c}{k}\int_{\mathbb{R} ^n} \theta^2(\frac{|x|^2}{k^2})|u|^2{\rm d}x+\frac{2\sqrt{2}c}{k}\int_{\mathbb{R} ^n}|A^{-\frac{1}{2}}g|^2{\rm d}x+\int_{\mathbb{R} ^n}|A^{-\frac{1}{2}}g|^2\theta^2(\frac{|x|^2}{k^2}){\rm d}x\\&&+\frac{1}{4}\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})|\nabla u|^2{\rm d}x\\&\leqslant&\frac{\lambda}{4}\int_{\mathbb{R} ^n} \theta^2(\frac{|x|^2}{k^2})|u|^2{\rm d}x+\frac{2\sqrt{2}c}{k}\|g\|_{H^{-1}(\mathbb{R} ^n)}^2+\int_{|x|\geqslant k}|A^{-\frac{1}{2}}g|^2{\rm d}x+\frac{1}{4}\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})|\nabla u|^2{\rm d}x\\&\leqslant&\frac{\lambda}{4}\int_{\mathbb{R} ^n} \theta^2(\frac{|x|^2}{k^2})|u|^2{\rm d}x+\frac{1}{4}\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})|\nabla u|^2{\rm d}x+\frac{2}{l}\epsilon.\end{eqnarray}
(3.49)

因此,联合(3.32)-(3.49)式,可得

\begin{eqnarray}\label{3.50}&&\frac{1}{2}\frac{\rm d}{{\rm d}t}\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})| u| ^2{\rm d}x+\frac{\nu}{2}\frac{\rm d}{{\rm d}t}\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})| \nabla u| ^2{\rm d}x\\&&+\frac{1}{2}\frac{\rm d}{{\rm d}t}\int_{\mathbb{R} ^n}\int_0^\infty\theta^2(\frac{|x|^2}{k^2})\mu(s)|\nabla\eta^t(x, s)|^2{\rm d}s{\rm d}x+\frac{\lambda}{2}\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})| u| ^2{\rm d}x\\&&+\frac{1}{2}\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})| \nabla u| ^2{\rm d}x+\frac{\delta}{2}\int_{\mathbb{R} ^n}\int_0^\infty\theta^2(\frac{|x|^2}{k^2})\mu(s)|\nabla\eta^t(x, s)|^2{\rm d}s{\rm d}x\\&\leqslant& \frac{6}{l}\epsilon+\frac{2\sqrt{2}c\nu}{k}\int_{\mathbb{R} ^n} \theta^2(\frac{|x|^2}{k^2})|\nabla u_t|^2{\rm d}x.\end{eqnarray}
(3.50)

\begin{eqnarray*}{\cal L}_1(t)&=&\frac{1}{2}\bigg(\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})| u| ^2{\rm d}x+\frac{\nu}{2}\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})| \nabla u| ^2{\rm d}x\\ &&+\int_{\mathbb{R} ^n}\int_0^\infty\theta^2(\frac{|x|^2}{k^2})\mu(s)|\nabla\eta^t(x, s)|^2{\rm d}s{\rm d}x\bigg), \end{eqnarray*}

并且\beta=\min\{\lambda, \frac{1}{\nu}, \delta\},则

\begin{eqnarray}\frac{\rm d}{{\rm d}t}{\cal L}_1(t)+\beta {\cal L}_1(t)\leqslant\frac{6}{l}\epsilon+\frac{2\sqrt{2}c\nu}{k}\int_{\mathbb{R} ^n} \theta^2(\frac{|x|^2}{k^2})|\nabla u_t|^2{\rm d}x.\end{eqnarray}
(3.51)

运用Gronwall引理,可得

\begin{eqnarray}\label{3.52}{\cal L}_1(t)&\leqslant& {\rm e}^{-\beta(t-t_2)}{\cal L}_1(t_2)+\frac{6\epsilon}{\beta l}+\frac{2\sqrt{2}c\nu}{k}\int_{t_2}^t\int_{\mathbb{R} ^n} {\rm e}^{-\beta(t-s)}\theta^2(\frac{|x|^2}{k^2})|\nabla u_t(s)|^2{\rm d}x{\rm d}s\\&\leqslant &{\rm e}^{-\beta(t-t_2)}{\cal L}(t_2)+\frac{6\epsilon}{\beta l}+\frac{2\sqrt{2}c\nu}{k}\int_{0}^t\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})|\nabla u_t(s)|^2{\rm d}x{\rm d}s\\&\leqslant &C{\rm e}^{-\beta(t-t_2)}\varrho^2_0+\frac{6\epsilon}{\beta l}+\frac{2\sqrt{2}c\nu}{k}\int_{0}^t\int_{\mathbb{R} ^n}|\nabla u_t(s)|^2{\rm d}x{\rm d}s\\&\leqslant &C{\rm e}^{-\beta(t-t_2)}\varrho^2_0+\frac{6\epsilon}{\beta l}+\frac{2\sqrt{2}c\nu}{k}N_2\\&\leqslant& \frac{8\epsilon}{\beta l}, \end{eqnarray}
(3.52)

其中t\geqslant \tilde{T}(\epsilon)=t_2+\frac{1}{\beta }\ln\frac{\beta C\varrho^2_0 l}{\epsilon},且k\geqslant \tilde{k}(\epsilon)=\max\{\tilde{k}_5(\epsilon), \frac{2\sqrt{2}c\nu\beta N_2l}{\epsilon}\},且t_2, \varrho_0为引理3.3和定理3.3中给定常数.

这意味着

\begin{eqnarray}\label{3.53}\frac{1}{2}\int_{|x|\geqslant \sqrt{2}k}| u| ^2{\rm d}x&\leqslant&\frac{1}{2}\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})| u| ^2{\rm d}x+\frac{\nu}{2}\int_{\mathbb{R} ^n}\theta^2(\frac{|x|^2}{k^2})| \nabla u| ^2{\rm d}x\\&&+\frac{1}{2}\int_{\mathbb{R} ^n}\int_0^\infty\theta^2(\frac{|x|^2}{k^2})\mu(s)|\nabla\eta^t(x, s)|^2{\rm d}s{\rm d}x\\&\leqslant &\frac{8\epsilon}{\beta l}.\end{eqnarray}
(3.53)

进一步

\begin{equation}\label{3.54}\int_{|x|\geqslant \sqrt{2}k}|\ u| ^2{\rm d}x\leqslant \frac{16\epsilon}{\beta l}, \end{equation}
(3.54)

其中t\geqslant \tilde{T}(\epsilon)=t_2+\frac{1}{\beta}\ln\frac{\beta C_2\varrho^2_0 l}{\epsilon},且k\geqslant \tilde{k}(\epsilon).证毕.

为了利用收缩函数方法来证明(2.3)-(2.4)式对应系统全局吸引子的存在性,我们还需证明以下结果.

引理3.5  设B_0({\cal E}_1, L^p(\mathbb{R} ^n))中的有界吸收集,并且z_m(t)=(u_m(t), \eta_m^t(s))为(2.3)-(2.4)式在{\mathcalE}_1中满足初值条件z_{m}^0\in B_0 (m=1, 2, \cdots)的解.设k>0并且B_k=\{x \in \mathbb{R} ^n : |x| < \sqrt{2}k\}.若(1.7)-(1.8)式成立, g\in H^{-1}(\mathbb{R} ^n)且非线性项f(x, u)满足(1.2)-(1.4)式,对于任意给定的k>0T\geqslant t_2,则在L^2(0, T;L^2(B_k))中存在\{u_m(t)\}的收敛子列.

  对于任意序列\{z_{m}^{0}\}\subset B_0,需证明存在子列\{z_{m_k}^{0}\}_{k=1}^\infty\subset\{z_m^{0}\}_{m=1}^\infty使得

\begin{equation}\label{3.55}\lim\limits_{l\rightarrow\infty}\lim\limits_{p\rightarrow\infty}\int_0^T{\|u_{m_l}(s)-u_{m_p}(s)\|^2{\rm d}s}=0, \end{equation}
(3.55)

其中u_{m_l}(t)=\Pi_1 S(t)z^{0}_{m_l}, \Pi_1: ~H^1(\mathbb{R} ^n)\timesL_{\mu}^2({\Bbb R}^+;H^1(\mathbb{R} ^n)\to H^1(\mathbb{R} ^n)为投影算子.

下面仅需证明

\begin{equation}{\cal U}:=\{u_m(t), t\in[0, T]: u_m(t)=\Pi_1 S(t)z_m^0, z_m^0\in B_0\}~ \mbox{在} ~L^2([0, T]; L^2(B_k))~\mbox{中相对紧}.\end{equation}
(3.56)

首先,关于(3.8)式在[0, T]上积分并且利用(3.12)式,可知

\begin{eqnarray}\label{3.56}& &(\lambda-\beta_1)\int_0^{T}{\|u(s)\|^2{\rm d}s}+\frac{1}{2}\int_0^{T}{\|\nabla u(s)\|^2{\rm d}s}+\frac{\delta}{2}\int_0^{T}{\|\eta^s(r)\|_{\mu, {\cal H}_1}^2{\rm d}s}\\&&+\alpha_1\int_0^{T}{\|u(s)\|_{L^p(\mathbb{R} ^N)}^p{\rm d}s}+\alpha_2\int_0^{T}{\int_{\mathbb{R} ^n}a(x)|u(s)|^p{\rm d}x{\rm d}s}\\&\leqslant&\frac{T}{2}\| g\|_{H^{-1}(\mathbb{R} ^n)}^2+\beta_2\|a\|_{L^1(\mathbb{R} ^n)}T+C\varrho^2_0.\end{eqnarray}
(3.57)

因此{\cal U}L^2([0, T]; H^1(\mathbb{R} ^n))\cap L^p([0, T];L^p(\mathbb{R} ^n))中有界.显然, {\cal U}L^2([0, T]; H^1(B_k))\cap L^p([0, T];L^p(B_k))中亦有界.

然后,根据引理3.3,可得\partial_t {\cal U}L^2([0, T]; H^{1}(\mathbb{R} ^n))中有界.易知\partial_t {\cal U}L^2([0, T]; H^{1}(B_k))中有界,即, \partial_t {\cal U}L^2([0, T]; H^{-1}(B_k))中亦有界.

{\cal U}L^2([0, T]; L^2(B_k))中相对紧.证毕.

定理3.3  设\{S(t)\}_{t\geqslant 0}为方程(2.3)-(2.4)在能量空间{\cal E}_1中的解半群.若非线性项f(x, u)=f_1(u)+a(x)f_2(u)满足条件(1.2)-(1.4), g\in H^{-1}(\mathbb{R} ^n)且(1.7)-(1.8)式成立,则\{S(t)\}_{t\geqslant 0}{\cal E}_1中渐近紧.

  设z_1=(u_1, \eta_1^t)z_2=(u_2, \eta_2^t)为方程(2.3)-(2.4)的两个解,分别满足初值条件z_{1}^0=(u_{1}^0, \eta_{1}^0)z_{2}^0=(u_{2}^0, \eta_{2}^0),并且z_{1}^0z_{2}^0属于半群\{S(t)\}_{t\geqslant 0}{\cal E}_1中的有界吸收集B_0.w=u_1-u_2, \xi^t=\eta_1^t-\eta_2^t.则,利用(2.3)式,有

\begin{equation}\label{3.57}\left\{\begin{array}{ll} w_{t}-\nu\nabla w_t-\Delta w+\lambda w-\int_0^\infty{\mu (s)\Delta \xi ^t(s){\rm d}s}\\ ~~ +f_1(u_1)-f_1(u_2)+a(x)(f_2(u_1)-f_2(u_2))=0 , \\w(0)=u_{1}^0-u_{2}^0, \\\xi^0=\eta_1^0-\eta_2^0, \\w(t)=\xi _t^t+\xi _s^t.\end{array}\right.\end{equation}
(3.58)

将(3.58)式乘以w(t),并且在\mathbb{R} ^n上积分,可得

\begin{eqnarray}\label{3.58}&&\frac{1}{2}\frac{\rm d}{{\rm d}t}\| w\| ^2+\frac{\nu}{2}\frac{\rm d}{{\rm d}t}\| \nabla w\| ^2+\| \nabla w\|^2+\lambda\|w\|^2-\int_0^\infty {\mu (s)\int_{\mathbb{R} ^n}\Delta\xi^t(s)w {\rm d}x{\rm d}s}\\&&+\int_{\mathbb{R} ^n}(f_1(u_1)-f_1(u_2))w{\rm d}x+\int_{\mathbb{R} ^n}a(x)(f_2(u_1)-f_2(u_2))w{\rm d}x=0.\end{eqnarray}
(3.59)

E(t)=\frac{1}{2}(\|w\|^2+\nu\|\nabla w\|^2+\| \xi ^t\| _{\mu , {\cal H}_1}^2).

\begin{eqnarray}\label{3.59}\frac{\rm d}{{\rm d}t}E(t)&=&\frac{1}{2}\frac{\rm d}{{\rm d}t}\| w\| ^2+\frac{\nu}{2}\frac{\rm d}{{\rm d}t}\| \nabla w\| ^2+\frac{1}{2}\frac{\rm d}{{\rm d}t}\| \xi ^t\| _{\mu , {\cal H}_1}^2\\&=&-\| \nabla w\|^2-\lambda\|w\|^2-\frac{1}{2}\int_0^\infty {\mu (s)\frac{\rm d}{{\rm d}s}|\nabla \xi ^t(s)| ^2{\rm d}s}\\&&-\int_{\mathbb{R} ^n}(f_1(u_1)-f_1(u_2))w{\rm d}x-\int_{\mathbb{R} ^n}a(x)(f_2(u_1)-f_2(u_2))w{\rm d}x.\end{eqnarray}
(3.60)

利用(1.2)与(1.4)式,可得

\begin{equation}\label{3.60}-\int_{\mathbb{R} ^n}(f_1(u_1)-f_1(u_2))w {\rm d}x \leqslant c_1\| w\|^2, \end{equation}
(3.61)

\begin{equation}\label{3.61}-\int_{\mathbb{R} ^n}a(x)(f_2(u_1)-f_2(u_2))w {\rm d}x \leqslant c_1\|a\|_{L^{\infty}(\mathbb{R} ^n)}\| w\|^2.\end{equation}
(3.62)

根据引理2.1,可以得到

\begin{equation}\label{3.62}-\frac{1}{2}\int_0^\infty {\mu (s)\frac{\rm d}{{\rm d}s}| \nabla \xi ^t(s)|^2{\rm d}s} \leqslant -\frac{\delta}{2}\|\xi^t\|_{\mu, {\cal H}_1}^2.\end{equation}
(3.63)

因此

\begin{equation}\label{3.63}\frac{\rm d}{{\rm d}t}E(t)+C_\delta E(t)\leqslant c_1(1+\|a\|_{L^{\infty}(\mathbb{R} ^n)})\|w(t)\|^2, \end{equation}
(3.64)

其中C_\delta=\min\{\frac{2}{\nu}, \delta, 2\lambda\}.

由于(3.11)式,故存在常数{C}_1, {C}_2,使得

{C}_1\|z_1(t)-z_2(t)\|^2_{{\cal E}_1}\leqslant E(t) \leqslant {C}_2\|z_1(t)-z_2(t)\|^2_{{\cal E}_1}.

将(3.64)式乘以{\rm e}^{C_\delta t}并且从0T积分,有

\begin{equation}\label{3.64}E(T)\leqslant {\rm e}^{-C_\delta T} E(0)+c_1(1+\|a\|_{L^{\infty}(\mathbb{R} ^n)}){\rm e}^{-C_\deltaT}\int_0^T{{\rm e}^{C_\delta s} \|w(s)\|^2{\rm d}s}.\end{equation}
(3.65)

则对于任意的\epsilon>0,存在\bar{T}(\epsilon)=\max\{\tilde{T}(\epsilon), \frac{1}{C_\delta}\ln{\frac{4C_2\varrho^2_0}{C_1\epsilon}}\}\tilde{k}(\epsilon)(这里\tilde{T}(\epsilon), \tilde{k}(\epsilon)为引理3.4中的给定常数),使得对于任意的T\geqslant \bar{T}(\epsilon)k\geqslant\tilde{k}(\epsilon),有

\begin{eqnarray}\label{3.65}&&\|S(T)z_1^0-S(T)z_2^0\|^2_{{\cal E}_1}\\&=&\|z_1(T)-z_2(T)\|^2_{{\cal E}_1}\leqslant \frac{1}{{C}_1}E(T)\\&\leqslant& \frac{1}{{C}_1}{\rm e}^{-C_\delta T} E(0)+\frac{c_1}{{C}_1}(1+\|a\|_{L^{\infty}(\mathbb{R} ^n)}){\rm e}^{-C_\deltaT}\int_0^T{{\rm e}^{C_\delta s} \|w(s)\|^2{\rm d}s}\\&\leqslant &\frac{{C}_2}{{C}_1}{\rm e}^{-C_\delta T} \|z_1(0)-z_2(0)\|^2_{{\cal E}_1}+\frac{c_1}{{C}_1}(1+\|a\|_{L^{\infty}(\mathbb{R} ^n)}){\rm e}^{-C_\deltaT}\int_0^T{{\rm e}^{C_\delta s} \|w(s)\|^2{\rm d}s}\\&\leqslant &\frac{{C}_2}{{C}_1}{\rm e}^{-C_\delta T} (\|z_1(0)\|^2_{{\cal E}_1}+\|z_2(0)\|^2_{{\cal E}_1})+\frac{c_1}{{C}_1}(1+\|a\|_{L^{\infty}(\mathbb{R} ^n)}){\rm e}^{-C_\deltaT}\int_0^T{{\rm e}^{C_\delta s} \|w(s)\|^2{\rm d}s}\\&\leqslant &\frac{2{C}_2}{{C}_1}{\rm e}^{-C_\delta T} \varrho^2_0+\frac{c_1}{{C}_1}(1+\|a\|_{L^{\infty}(\mathbb{R} ^n)}){\rm e}^{-C_\deltaT}\int_0^T{{\rm e}^{C_\delta s} \|w(s)\|^2{\rm d}s}\\&\leqslant &{\frac{1}{2}\epsilon}+\frac{c_1}{{C}_1}(1+\|a\|_{L^{\infty}(\mathbb{R} ^n)}){\rm e}^{-C_\deltaT}\int_0^T{{\rm e}^{C_\delta s} \|w(s)\|^2{\rm d}s}.\end{eqnarray}
(3.66)

另一方面,根据引理3.4,可得

\begin{eqnarray}\label{3.66}&&\frac{c_1}{{C}_1}(1+\|a\|_{L^{\infty}(\mathbb{R} ^n)}){\rm e}^{-C_\deltaT}\int_0^T{{\rm e}^{C_\delta s} \|w(s)\|^2{\rm d}s}\\&\leqslant& \frac{c_1}{{C}_1}(1+\|a\|_{L^{\infty}(\mathbb{R} ^n)}){\rm e}^{-C_\deltaT}\int_0^T{{\rm e}^{C_\delta s} \int_{|x|\geqslant\sqrt{2}k}|w(x, s)|^2{\rm d}x{\rm d}s}\\&&+\frac{c_1}{{C}_1}(1+\|a\|_{L^{\infty}(\mathbb{R} ^n)}){\rm e}^{-C_\deltaT}\int_0^T{{\rm e}^{C_\delta s}\int_{|x| <\sqrt{2}k}|w(x, s)|^2{\rm d}x{\rm d}s}\\&\leqslant& \frac{2c_1}{{C}_1C_\delta}(1+\|a\|_{L^{\infty}(\mathbb{R} ^n)})\frac{16}{\beta l}\epsilon\\&&+\frac{c_1}{{C}_1}(1+\|a\|_{L^{\infty}(\mathbb{R} ^n)}){\rm e}^{-C_\deltaT}\int_0^T{{\rm e}^{C_\delta s}\int_{|x| <\sqrt{2}k}|w(x, s)|^2{\rm d}x{\rm d}s}.%&\leqslant& \frac{1}{2}\epsilon+\frac{c_1}{{C}_1}(1+\|a\|_{L^{\infty}(\mathbb{R} ^n)}){\rm e}^{-C_\delta T}\int_0^T{{\rm e}^{C_\delta s}\int_{|x| <\sqrt{2}k}|w(x, s)|^2{\rm d}x{\rm d}s}.\end{eqnarray}
(3.67)

\begin{eqnarray}\label{3.67}&&\|S(T)z_1^0-S(T)z_2^0\|^2_{{\cal E}_1}\\&\leqslant &C\varepsilon+\frac{c_1}{{C}_1}(1+\|a\|_{L^{\infty}(\mathbb{R} ^n)}){\rm e}^{-C_\deltaT}\int_0^T{{\rm e}^{C_\delta s} \int_{|x| <\sqrt{2}k}|w(x, s)|^2{\rm d}x{\rm d}s}.\end{eqnarray}
(3.68)

对应于引理2.2,设

\begin{equation}\label{3.68} \phi_T(z_1, z_2)=\frac{c_1}{{C}_1}(1+\|a\|_{L^{\infty}(\mathbb{R} ^n)}){\rm e}^{-C_\deltaT}\int_0^T{{\rm e}^{C_\delta s} \int_{|x| <\sqrt{2}k}|w(x, s)|^2{\rm d}x{\rm d}s}.\end{equation}
(3.69)

根据引理3.5,可知w(t)L^2([0, T];L^2(B_k))中相对紧.则\phi_T(z_1, z_2)B_0\times B_0中的收缩函数,其中B_0({\cal E}_1, L^p(\mathbb{R} ^n)中的有界吸收集.

根据引理2.2,易知\{S(t)\}_{t\geqslant 0}{\cal E}_1中相对紧,证毕.

根据引理2.3,定理3.2,引理3.2和定理3.3,可得

定理3.4  若引理3.8的假设成立,则解半群\{S(t)\}_{t\geqslant0}{\cal E}_1中拥有全局吸引子{\cal A}_0,并且按照{\cal E}_1 -范数吸引着{\cal E}_1中的每一个有界集.

注3.1  在文献[27]中,作者在复杂抽象的推导过程中应用渐近先验估计技术证明了吸引子在H^1(\mathbb{R} ^n)中的存在性,但是该结果仅仅是我们结果的一个特例.

参考文献

Aifantis E C .

On the problem of diffusion in solids

Acta Mech, 1980, 37: 265- 296

DOI:10.1007/BF01202949      [本文引用: 1]

Aifantis E C .

Gradient nanomechanics:applications to deformation, fracture, and diffusion in nanopolycrystals

Metall Mater Trans A, 2011, 42: 2985- 2998

DOI:10.1007/s11661-011-0725-9      [本文引用: 1]

Anh C T , Toan N D .

Nonclassical diffusion equations on {{\mathbb{R}}^{N}} with singularly oscillating external forces

Appl Math Lett, 2014, 38: 20- 26

DOI:10.1016/j.aml.2014.06.008      [本文引用: 1]

Anh C T , Toan N D .

Uniform attractors for non-autonomous nonclassical diffusion equations on {{\mathbb{R}}^{N}}

Bull Korean Math Soc, 2014, 51 (5): 1299- 1324

DOI:10.4134/BKMS.2014.51.5.1299      [本文引用: 1]

Borini S , Pata V .

Uniform attractors for a strongly damped wave equations with linear memory

Asymptot Anal, 1999, 20: 263- 277

URL     [本文引用: 2]

Chepyzhov V V, Vishik M I. Attractors for Equations of Mathematical Physics. Providence, RI: Amer Math Soc, 2002

[本文引用: 2]

Gatti C , Miranville A , Pata V , Zelik S V .

Attractors for semilinear equations of viscoelasticity with very low dissipation

R Mountain J Math, 2008, 38: 1117- 1138

DOI:10.1216/RMJ-2008-38-4-1117      [本文引用: 1]

Giorgi C , Rivera J E M , Pata V .

Global attractors for a semilinear hyperbolic equations in viscoelasticity

J Math Anal Appl, 2001, 260: 83- 99

DOI:10.1006/jmaa.2001.7437      [本文引用: 2]

Hale J K. Asymptotic Behavior of Dissipative Systems. Providence, RI: Amer Math Soc, 1988

[本文引用: 1]

KHanmamedov A Kh .

Global attractors for von Karman equations with nonlinear interior dissipation

J Math Anal Appl, 2006, 318: 92- 101

DOI:10.1016/j.jmaa.2005.05.031     

Kloeden P E , Siegmund S .

Bifurcation and continuous transition of attractors in autonomous and nonautonomous systems

Internat J Bifur Chaos, 2005, 15: 462- 743

URL     [本文引用: 1]

Kloeden P E , Marin-Rubio P .

Weak pullback attractors of nonautonomous difference inclusions

J Difference Equ Appl, 2003, 9: 489- 502

DOI:10.1080/1023619031000076515     

刘世芳, 马巧珍.

具有历史记忆的阻尼吊桥方程强全局吸引子的存在性

数学物理学报, 2017, 37A (4): 684- 697

URL     [本文引用: 1]

Liu S , Ma Q .

Existence of strong global attractors for damped suspension bridge equations with history memory

Acta Mathematica Scientia, 2017, 37A (4): 684- 697

URL     [本文引用: 1]

Ma Q, Liu Y, Zhang F. Global attractors in {{H}^{1}}({{\mathbb{R}}^{N}}) for nonclassical diffusion equations. Discrete Dyn Nat Soc, 2012, Article ID: 672762

[本文引用: 1]

马巧珍, 徐玲, 张彦军.

记忆型非经典反应扩散方程在{{\mathbb{R}}^{3}}中解的渐近性态

数学物理学报, 2016, 36A (1): 36- 48

DOI:10.3969/j.issn.1003-3998.2016.01.004      [本文引用: 1]

Ma Q , Xu L , Zhang Y .

Asymptotic behavior of the solution for the nonclassical diffusion equations with fading memory on the whole space {{\mathbb{R}}^{3}}

Acta Mathematica Scientia, 2016, 36A (1): 36- 48

DOI:10.3969/j.issn.1003-3998.2016.01.004      [本文引用: 1]

Peter J G , Gurtin M E , Pata V .

On the theory of heat condition involving two temperatures

Z Angew Math Phys, 1968, 19: 614- 627

DOI:10.1007/BF01594969      [本文引用: 1]

Pata V , Squassina M .

On the strongly damped wave equation

Comm Math Phys, 2005, 253: 511- 533

DOI:10.1007/s00220-004-1233-1      [本文引用: 1]

Pata V , Zucchi A .

Attractors for a damped hyperbolic equation with linear memory

Adv Math Sci Appl, 2001, 11 (2): 505- 529

URL     [本文引用: 2]

Sun C , Cao D , Duan J .

Non-autonomous dynamics of wave equations with nonlinear damping and critical nonlinearity

Nonlinearity, 2006, 19: 2645- 2665

DOI:10.1088/0951-7715/19/11/008      [本文引用: 3]

Sun C , Wang S , Zhong C .

Global attractors for a nonclassical diffusion equation

Acta Mathematica Sinica:English Series, 2007, 23: 1271- 1280

DOI:10.1007/s10114-005-0909-6      [本文引用: 1]

Sun C , Yang M .

Dynamics of the nonclassical diffusion equtions

Asympt Anal, 2008, 59: 51- 81

[本文引用: 3]

Temam R. Infinit Dimensional Dynamical System in Mechanichs and Physics. New York: Spring-Verlag, 1997

[本文引用: 1]

Wang X , Yang L , Zhong C .

Attractors for the nonclassical diffusion equations with fading memory

J Math Anal Appl, 2010, 362: 327- 337

DOI:10.1016/j.jmaa.2009.09.029      [本文引用: 2]

Wang X , Zhong C .

Attractors for the non-autonomous nonclassical diffusion equations with fading memory

Nonlinear Anal, 2009, 71: 5733- 5746

DOI:10.1016/j.na.2009.05.001      [本文引用: 2]

Wang X , Ju W , Zhong C .

Strong attractors for the non-autonomous nonclassical diffusion equations with fading memory

Chinese Annals of Mathematics, 2013, 34A (6): 671- 688

[本文引用: 1]

Xiao Y .

Attractors for a nonclassical diffusion equation

Acta Math Appl, 2002, 18: 273- 276

DOI:10.1007/s102550200026      [本文引用: 2]

Xie Y , Li Q , Zhu K .

Attractors for nonclassical diffusion equations with arbitrary polynomial growth nonlinearity

Nonlinear Anal RWA, 2016, 31: 23- 37

DOI:10.1016/j.nonrwa.2016.01.004      [本文引用: 3]

/