数学物理学报, 2018, 38(6): 1135-1143 doi:

论文

一类二阶中立随机偏微分方程的吸引集和拟不变集

范英飞, 章国鹏, 江欣国,, 马剑

Attracting and Quasi-Invariant Sets of Second-Order Neutral Stochastic Partial Differential Equations

Fan Yingfei, Zhang Guopeng, Jiang Xinguo,, Ma jian

通讯作者: 江欣国, E-mail: fanyf@my.swjtu.edu.cn

收稿日期: 2017-08-9  

基金资助: 国家自然科学基金面上项目.  71771191
国家自然科学基金面上项目.  71473207
成都市科技项目.  2015-RK00-00171-ZF
西南交通大学博士研究生创新

Received: 2017-08-9  

Fund supported: the NSFC.  71771191
the NSFC.  71473207
CSTP.  2015-RK00-00171-ZF
the Doctoral Innovation Fund Program of Southwest Jiaotong University

摘要

研究了一类二阶中立随机偏微分方程.运用随机分析与不等式技巧,获得了这类方程存在吸引集和拟不变集的充分条件,推广了一些已有的相关结果.

关键词: 吸引 ; 拟不变 ; 二阶 ; 随机 ; 偏微分

Abstract

This paper is concerned with the second-order neutral stochastic partial differential equations. By using stochastic analysis and inequality techniques, sufficient conditions for the existence of attracting and quasi invariant sets are obtained. Some existing related results are generalized.

Keywords: Attracting ; Quasi-invariant ; Second-order ; Stochastic ; Partial differential

PDF (706KB) 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

范英飞, 章国鹏, 江欣国, 马剑. 一类二阶中立随机偏微分方程的吸引集和拟不变集. 数学物理学报[J], 2018, 38(6): 1135-1143 doi:

Fan Yingfei, Zhang Guopeng, Jiang Xinguo, Ma jian. Attracting and Quasi-Invariant Sets of Second-Order Neutral Stochastic Partial Differential Equations. Acta Mathematica Scientia[J], 2018, 38(6): 1135-1143 doi:

1 引言

本文研究了如下二阶中立随机偏微分方程

$\begin{equation}\label{e2.1} \left\{ {\begin{array}{ll}d[x'(t) - f\left( {t, x(t - \tau (t))} \right)]\\=[Ax(t) + g\left( {t, x(t - \tau (t))} \right)]{\rm d}t{+ \sigma \left( {t, x(t - \tau (t))} \right){\rm d}W(t), ~t \ge 0, }\\ {{x_0}(s) = \phi (s) \in BC_{{{\cal F}_0}}^b([ - \tau , 0], H), ~- \tau \le s \le 0, ~x'(0) = {\phi _1}, } \end{array}} \right. \end{equation}$

其中$H$为Hilbert空间, $A:D(A)\subset H\to H$是在$H$上的强连续cosine集族的无穷小生成子; $f, g:[0, \infty) \times {H} \to H$, $\sigma :[0, \infty) \times {H} \to {\cal L}_2^0$均为连续函数; $0 \le \tau (t) \le \tau $${\phi _1}$${H}$上的${{\cal F} _0}$ -可测,且独立于$W(t)$的随机变量.

近来,对二阶随机微分方程的研究越来越多.与一阶系统相比,二阶系统反映了更复杂的情况,可以揭示更多的现象.研究人员已经在二阶随机微分方程中得到了许多有价值的结论[1-4].分析发现,以往的研究通过一些较强的条件来研究二阶随机微分方程的稳定性.然而,这种条件往往过于严苛.

为了找到较弱的约束来估计稳定状态, Xu[5]讨论了Volterra微分方程组的不变集和吸引集.随后,吸引集和动力系统的不变组合被得到了广泛的研究.对于离散系统,参见文献[6-7];对于带有时滞项的确定性差分系统,参见文献[8-9];对于偏微分系统,参见文献[10];对于随机系统,参加文献[11-12].更进一步的, Li和Xu [13]估计了一阶随机中立偏微分方程的吸引集和拟不变集,稳定条件得以进一步减弱,估计出了更好的结果.

因此,通过使用不等式技术,本文研究了二阶中立随机偏微分系统(1.1),估计了系统的吸引集和拟不变集,拓展了文献14中的有关结论.

2 预备知识

为了对方程(1.1)的吸引集和拟不变集进行估计,本文首先定义了一些必要的空间及范数.

定义两个Hilbert空间$H$$K$的内积为${\left\langle { \cdot, \cdot } \right\rangle _H}$${\left\langle { \cdot, \cdot } \right\rangle _K}$,范数为${\left\| \cdot \right\|_H}$${\left\| \cdot \right\|_K}$; ${\cal L}(K, H)$是从$K$映到$H$的所有线性有界算子的集合; $E[f]$为函数$f$的数学期望; $\left({\Omega, {\cal F}, {{\left\{ {{{\cal F}_t}} \right\}}_{t \geqslant 0}}, P} \right)$是一个完备的概率空间,其中${{\left\{ {{{\cal F}_t}} \right\}}_{t \geqslant 0}}$满足通常的性质(右连续,且${{{\cal F}_0}}$包含了所有的$P$ -集合).

定义$\left\{ {W(t), t \geqslant 0} \right\}$是一个在空间$\left({\Omega, {\cal F}, {{\left\{ {{{\cal F}_t}} \right\}}_{t \geqslant 0}}, P} \right)$上的$K-{{\left\{ {{{\cal F}_t}} \right\}}_{t \geqslant 0}}$-Wiener过程,其中$x, y$$\in K$, $Q$是正定的、自伴随的协方差算子.因此可得到如下性质

为了定义关于$Q$-Wiener过程$W(t)$上的随机积分,本文引出了子空间${K_0} = {Q^{\frac{1}{2}}}(K)$.子空间${K_0}$是一个希尔伯特空间,空间的内积为${\left\langle {x, y} \right\rangle _{{K_0}}} = \left\langle {} \right.{Q^{ - \frac{1}{2}}}x, {Q^{ - \frac{1}{2}}}y{\left. {} \right\rangle _K}$.定义${\cal L}_2^0$是一个Hilbert-Schmidt空间,空间上的的算子是从$K_0$映到$H$上的.因此Hilbert-Schmidt空间上的范数为$\left\| \psi \right\|_{{\cal L}_2^0}^2 ={\rm tr}((\psi {Q^{\frac{1}{2}}}){(\psi {Q^{\frac{1}{2}}})^ * }), ~\psi \in {\cal L}_2^0.$且当$\psi \in {\cal L}(K, H)$,范数定义为$\left\| \psi \right\|_{{\cal L}_2^0}^2 = {\rm tr}(\psi Q{\varphi ^ * }).$

空间$C(X, Y)$由拓扑空间上$X$$Y$的连续映射组成, ${\mathbb{R}^{+} } = [0, + \infty)$.$\tau>0$,定义$C\triangleq C([-\tau, 0], \mathbb{R})$, ${C_H} = C([- \tau, 0], H)$.如果$\varphi \in C$,则${\left| {\varphi (t)} \right|_\tau } = \mathop {\sup }\limits_{ - \tau \le s \le 0} \left| {\varphi (t + s)} \right|, $${\left\| \varphi \right\|_{{C_H}}} = \mathop {\sup } \limits_{ - \tau \le s \le 0} {\left\| \varphi \right\|_H}$.定义$ BC_{{{\cal F}_0}}^b([- \tau, 0], H)$是所有有界${{\cal F}_0}$ -可测, ${C_H}$ -上的随机变量$\varphi $的集合,且有$\left\| \varphi \right\|_{{L^p}}^p = \mathop {\sup }\limits_{ - \tau \le s \le 0} E\left\| {\varphi (s)} \right\|_H^p < \infty.$

参数族${(T(t))_{t \ge 0}}$是一个强连续cosine集族[15],如果以下条件成立

(1) $T(0)=I; $

(2) $T(t)x$$\mathbb{R}$上关于$t$连续,对于所有的$x \in H; $

(3) $T(t + s) + T(t - s) = 2T(t)T(s)$,对于所有的$t, s \in \mathbb{R}$.

$A$是一个在$H$上的稠密算子,则无穷小生成子$A:D(A)\subset H\to H$是关于算子${(T(t))_{t \ge 0}}$的cosine集族,且满足$Ax= {\left. {\frac{{{{\rm d}^2}}}{{{\rm d}{t^2}}}T(t)x} \right|_{t = 0}}$.

定义强连续sine集族${(S(t))_{t \ge 0}}$满足$S(t)x = \int_0^t {T(s)x}{\rm d}s, \ t \in \mathbb{R}, \ x \in H.$

定义2.1[4] 随机过程$\left\{ {x(t), - \tau \le t \le b} \right\} (0 < b < \infty)$是方程(1.1)的柔和解,如果以下条件成立:

(1) $x(t)$是一个可测的${{\cal F}_t} $ -适应过程,且$E\int_0^b {\left\| {x(t)} \right\|} _H^2{\rm d}t < \infty; $

(2) $x(t)$满足如下积分方程

$ \begin{eqnarray} \label{e2.2} x(t)&=& T(t){\phi(0)} + S(t)({\phi _1} - f(0, {x_0})) + \int_0^t {T(t - s)f(s, {x(s-\tau(s) )})}{\rm d}s\\ &&+ \int_0^t {S(t - s)g(s, {x(s-\tau(s) )})}{\rm d}s + \int_0^t {S(t - s)\sigma (s, {x(s-\tau(s) )})} {\rm d}W(s). \end{eqnarray} $

对于$\phi \subset BC_{{{\cal F}_0}}^b([- \tau, 0], H)$,方程(1.1)的温和解常记为$x(t) = x(t, 0, {\phi})$或者${x_t}(0, {\phi})$.

定义2.2 当系统(1.1)有零解时,系统的温和解被称为$p$ -指数$(p \ge 2)$稳定,若存在两个取值为正的常数$\lambda > 0$${M_1} > 1$,对于初值为$\phi \in BC_{{{\cal F}_0}}^b([- \tau, 0], H)$的解$x(t)$满足

定义2.3 当$\phi \subset BC_{{{\cal F}_0}}^b([- \tau, 0], H)$时,集合$X \subset BC_{{{\cal F}_0}}^b([- \tau, 0], H)$被称为方程(1.1)的不变集,如果

定义2.4[13] 当$\phi \subset BC_{{{\cal F}_0}}^b([- \tau, 0], H)$时,集合$S \subset BC_{{{\cal F}_0}}^b([- \tau, 0], H)$被称为方程(1.1)的吸引集,如果

其中$p(\cdot, \cdot)$$S \subset BC_{{{\cal F}_0}}^b([- \tau, 0], H)$上的任意距离.

定义2.5[13] 集合$S \subset BC_{{{\cal F}_0}}^b([- \tau, 0], H)$被称为方程(1.1)的拟不变集,若有正数$k$$b$,对于初值$\phi \in S$,满足$[k{x_t}(0, \phi) + b] \in S, \ t \ge 0.$

引理2.1[16] 对于任意的$x \in \mathbb{R}_ + ^n$以及$p>0$,有

引理2.2[17] 对于任意的$r \ge 1$${\cal L}_2^0 $ -可预测过程$\varphi (\cdot)$,

其中${C_r} = {(r(2r - 1))^r}.$

引理2.3 设$z(t) \in C(\mathbb{R}, {\mathbb{R}_ + })$是如下时滞积分不等式的解

$\begin{equation}z(t) \le {\lambda _3}\int_0^t {{{\rm e}^{ - {c_1}(t - s)}}{{\left| {z(t)} \right|}_\tau }{\rm d}s} + {\lambda _4}\int_0^t {{{\rm e}^{ - {c_2}(t - s)}}{{\left| {z(t)} \right|}_\tau }{\rm d}s} + {\lambda _5}, ~t \ge 0, \label{e2.3}\end{equation}$

其中${\lambda _3}$, ${\lambda _4}$, ${\lambda _5}$是非负常数, ${c_1}, \ {c_2}>0$.如果

$\begin{equation}\Upsilon = \frac{{{\lambda _3}}}{{{c_1}}} + \frac{{{\lambda _4}}}{{{c_2}}} < 1, \label{e2.4}\end{equation}$

那么

$\begin{equation}z(t) \le {(1 - \Upsilon )^{ - 1}}{\lambda _5}, ~t \ge 0, \label{e2.5}\end{equation}$

$\begin{equation}z(t) \le {(1 - \Upsilon )^{ - 1}}{\lambda _5}, ~t \in [ - \tau , 0].\label{e2.6}\end{equation}$

  依据文献[18,引理8.2],引理2.3容易得证.

引理2.4 设$z(t) \in C(\mathbb{R}, {\mathbb{R}_ + })$是如下时滞积分不等式的解

$\begin{eqnarray} \label{e2.10}z(t) &\le &{\lambda _1}{{\rm e}^{ - {c_1}t}} + {\lambda _2}{{\rm e}^{ - {c_2}t}} + {\lambda _3}\int_0^t {{{\rm e}^{ - {c_1}(t - s)}}{{\left| {z(t)} \right|}_\tau }{\rm d}s} \\&&{ + {\lambda _4}\int_0^t {{{\rm e}^{ - {c_2}(t - s)}}{{\left| {z(t)} \right|}_\tau }{\rm d}s} + {\lambda _5}, \;\;\;\;t \ge 0}, \end{eqnarray}$

且满足

$\begin{equation}z(t) \le \phi (t), ~t \in [ - \tau , 0], \label{e2.11}\end{equation}$

其中$\lambda _1, \cdots, \lambda _5$是非负常数; ${c_1}, \ {c_2}>0$; $\phi (s) \in C([- \tau, 0], {R_ + })$, $s \in [- \tau, 0].$

如果(2.3)式成立,那么一定存在常数$\alpha \in (0, {c_1} \wedge {c_2})$$N>0$满足

$\begin{equation}z(t) \le N{{\rm e}^{ - \alpha t}} + {(1 - \Upsilon )^{ - 1}}{\lambda _5}, ~t \ge 0, \label{e2.12}\end{equation}$

其中

$\begin{equation}{[{\phi(0)}]_\tau } < N, ~\frac{{{\lambda _1} + {\lambda _2}}}{N} + {{\rm e}^{\alpha \tau }}\frac{{{\lambda _3}}}{{{c_1} - \alpha }} + {{\rm e}^{\alpha \tau }}\frac{{{\lambda _4}}}{{{c_2} - \alpha }} < 1.\label{e2.13}\end{equation}$

  依据文献[18,引理8.2],引理2.4容易得证.

在(2.4)式中令${\lambda _5}=0$,可得到如下引理

推论2.1 若$(2.4)$式中的所有条件都成立,那么所有关于系统(2.6)和(2.7)的解都指数收敛于零.

3 主要结果

首先给出两个重要的条件.

(H1) cosine集族${(T(t))_{t \ge 0}}$和sine集族${(S(t))_{t \ge 0}}$满足如下条件

(H2)存在非负常数${L_f}, \ {L_g}, \ {L_\sigma }, \ {b_f}, \ {b_g}$${b_\sigma }$,对于任意的$x, ~y \in {H}$, $t \ge 0$满足

定理3.1 设(H1)-(H2)满足,那么${S_1} = \left\{ {\left. {\varphi \in BC_{{{\cal F}_0}}^b([- \tau, 0], H)} \right|\left\| \varphi \right\|_{{L^p}}^p \le {{(1 - \Upsilon)}^{ - 1}}{ I}} \right\}$是系统(1.1)温和解的全局吸引集, ${S_2} = \left\{ {\left. {\varphi \in BC_{{{\cal F}_0}}^b([- \tau, 0], H)} \right|} \right.$$\left. {\left\| \varphi \right\|_{{L^p}}^p \le l, \; l > 0} \right\}$是方程(1.1)温和解的拟不变集,若如下不等式成立

$\begin{align} &\Upsilon ={{10}^{p-1}}{{M}^{p}}{{\gamma }^{1-p}}L_{f}^{p}/\gamma +{{10}^{p-1}}{{M}^{p}}({{\beta }^{1-p}}L_{g}^{p}+\frac{p(p-1)}{2}{{(\frac{2\beta (p-1)}{p-2})}^{\frac{2-p}{2}}}L_{\sigma }^{p})/\beta \\ &\ \ \ <1, \\ \end{align}$

$\begin{equation}I= {10^{p - 1}}{M^p}{\gamma ^{ - p}}b_f^p +{10^{p - 1}}{M^p}{\beta ^{ - p}}b_g^p+ {10^{p - 1}}\frac{{p(p - 1)}}{2}{(\frac{{2\beta (p - 1)}}{{p - 2}})^{\frac{{2 - p}}{2}}}\frac{{{M^p}b_\sigma ^p}}{\beta }.\label{e3.2}\end{equation}$

  由(2.1)式可得

$\begin{eqnarray}\label{e3.3}E\left\| {x(t)} \right\|_H^p &=& E\bigg\| {T(t){\phi (0)} + S(t)({\phi _1} - f(0, {x_0} ))} + \int_0^t {T(t - s)f(s, {x(s-\tau(s) )})}{\rm d}s\\&&+ \int_0^t {S(t - s)g(s, {x(s-\tau(s) )})}{\rm d}s + \int_0^t {S(t - s)\sigma (s, {x(s-\tau(s) )})} {\rm d}W(s)\bigg\|_H^p\\&\le& {5^{p - 1}}E\left\| {T(t){\phi(0)}} \right\|_H^p + {5^{p - 1}}E\left\| {S(t)({\phi _1} - f(0, {x_0}))} \right\|_H^p \\&&+ {5^{p - 1}}E\left\| {\int_0^t {T(t - s)f(s, {x(s-\tau(s) )})}{\rm d}s} \right\|_H^p\\&&+ {5^{p - 1}}E\left\| {\int_0^t {S(t - s)g(s, {x(s-\tau(s) )})}{\rm d}s} \right\|_H^p\\&&+{5^{p - 1}}E\;\left\| {\int_0^t {S(t - s)\sigma (s, {x(s-\tau(s))})} {\rm d}W(s)} \right\|_H^p\\&:=&{5^{p - 1}}\sum\limits_{i = 1}^5 {{\omega _i}(t)}.\end{eqnarray} $

由定义(H1), (3.3)式中等式右边的第一项被放缩为

$\begin{eqnarray}\label{e3.4}{\omega _1}(t) = E\left\| {T(t){\phi (0)}} \right\|_H^p \le E\left\| {{\phi(0)}} \right\|_H^p{M^p}{{\rm e}^{ - p\gamma t}}\le E\left\| {{\phi (0)}} \right\|_H^p{M^p}{{\rm e}^{ - \gamma t}}.\end{eqnarray} $

由定义(H1)-(H2), (3.3)式中等式右边的第二项被放缩为

$\begin{eqnarray}\label{e3.5}{\omega _2}(t) &=& E\left\| {S(t)({\phi _1} - f(0, {x_0 }))} \right\|_H^p \\&\le& E\left\| {({\phi _1} - f(0, {x_0 } ))} \right\|_H^p{M^p}{{\rm e}^{ - p\beta t}}\\&\le &E\left\| {({\phi _1} - f(0, {x_0 }))} \right\|_H^p{M^p}{{\rm e}^{ - \beta t}}\\&\le &E{({\left\| {{\phi _1}} \right\|_H} + {L_f}{\left| {{{\left\| {{\phi (0)}} \right\|}_H}} \right|_\tau } + {b_f})^p}{M^p}{{\rm e}^{ - \beta t}}.\end{eqnarray} $

由定义(H1)-(H2),引理2.1和Hölder不等式, (3.3)式中等式右边的第三项被放缩为

$\begin{eqnarray}\label{e3.6}{\omega _3}(t)&=& E\left\| {\int_0^t {T(t - s)f(s, {x(s-\tau(s) )})}{\rm d}s} \right\|_H^p\\&\le& E{\bigg[\int_0^t {M{{\rm e}^{ - \gamma (t - s)}}({L_f}{{\left| {{{\left\| {x(s)} \right\|}_H}} \right|}_\tau } + {b_f}){\rm d}s}\bigg ]^p} \\&\le &{2^{p - 1}}{M^p}{\gamma ^{1 - p}}L_f^p\bigg(\int_0^t {{{\rm e}^{ - \gamma (t - s)}}{{\left| {E\left\| {x(s)} \right\|_H^p} \right|}_\tau }{\rm d}s}\bigg ) + {2^{p - 1}}{M^p}{\gamma ^{ - p}}b_f^p.\end{eqnarray} $

类似于$\omega _3(t)$, $\omega _4(t)$被放缩为

$\begin{eqnarray} \label{e3.7}{\omega _4}(t)&= &E\left\| {\int_0^t {S(t - s)g(s, {x(s-\tau(s) )})}{\rm d}s} \right\|_H^p\\&\leqslant& {2^{p - 1}}{M^p}{\beta ^{1 - p}}L_g^p\bigg(\int_0^t {{{\rm e}^{ - \beta (t - s)}}{{\left| {E\left\| {x(s)} \right\|_H^p} \right|}_\tau }{\rm d}s} \bigg) + {2^{p - 1}}{M^p}{\beta ^{ - p}}b_g^p.\end{eqnarray}$

对于(3.3)式中等式右边的第五项,由定义(H1)-(H2), Hölder不等式和引理2.2可得

$\begin{eqnarray}\label{e3.8}{\omega _5}(t) &=& E\left\| {\int_0^t {S(t - s)\sigma (s, {x(s-\tau(s) )})} {\rm d}W(s)} \right\|_H^p\\&\leqslant &{M^p}{\bigg(\frac{{p(p - 1)}}{2}\bigg)^{\frac{p}{2}}}{\bigg(\int_0^t {({{\rm e}^{ - \beta p(t - s)}}E\left\| {\sigma (s, {x(s-\tau(s) )})} \right\|_{{\cal L}_2^0}^p} )^{\frac{p}{2}}}{\rm d}s{\bigg)^{\frac{p}{2}}}\\&=& {M^p}\frac{{p(p - 1)}}{2}{\bigg(\int_0^t {({{\rm e}^{ - 2\beta (t - s)}}E\left\| {\sigma (s, {x(s-\tau(s) )})} \right\|_{{\cal L}_2^0}^p})^{\frac{p}{2}}}{\rm d}s{\bigg)^{\frac{p}{2}}}\\&\quad&+{M^p}\frac{{p(p - 1)}}{2}{\bigg(\int_0^t {{{\rm e}^{ - \frac{{2\beta (p - 1)(t - s)}}{{p - 2}}}}{\rm d}s}\bigg )^{\frac{p}{2} - 1}}\bigg(\int_0^t {{{\rm e}^{ - \beta (t - s)}}E\left\| {\sigma (s, {x(s-\tau(s) )})} \right\|_{{\cal L}_2^0}^p{\rm d}s}\bigg )\\&\quad&+{2^{p - 1}}\frac{{p(p - 1)}}{2}{\bigg(\frac{{2\beta (p - 1)}}{{p - 2}}\bigg)^{\frac{{2 - p}}{2}}}{M^p}L_\sigma ^p\bigg(\int_0^t {{{\rm e}^{ - \beta (t - s)}}{{\left| {E\left\| {x(s)} \right\|_H^p} \right|}_\tau }{\rm d}s}\bigg )\\&\quad&+ {2^{p - 1}}\frac{{p(p - 1)}}{2}{\bigg(\frac{{2\beta (p - 1)}}{{p - 2}}\bigg)^{\frac{{2 - p}}{2}}}\frac{{{M^p}b_\sigma ^p}}{\beta }.\end{eqnarray} $

依据(3.3)-(3.8)式,有

$\begin{eqnarray}\label{e3.9}E\left\| {x(t)} \right\|_H^p &\le& {5^{p - 1}}E\left\| {\phi (0)} \right\|_H^p{M^p}{{\rm e}^{ - \gamma t}}+ {5^{p - 1}}E{({\left\| {{\phi _1}} \right\|_H} + {L_f}{\left| {{{\left\| {\phi (0)} \right\|}_H}} \right|_\tau } + {b_f} )^p}{M^p}{{\rm e}^{ - \beta t}}\\&&+ {10^{p - 1}}{M^p}{\gamma ^{1 - p}}L_f^p\bigg(\int_0^t {{{\rm e}^{ - \gamma (t - s)}}{{\left| {E\left\| {x(s)} \right\|_H^p} \right|}_\tau }{\rm d}s} \bigg)\\&& + {10^{p - 1}}{M^p}{\beta ^{1 - p}}L_g^p\bigg(\int_0^t {{{\rm e}^{ - \beta (t - s)}}{{\left| {E\left\| {x(s)} \right\|_H^p} \right|}_\tau }{\rm d}s}\bigg )\\&&+ {10^{p - 1}}\frac{{p(p - 1)}}{2}{\bigg(\frac{{2\beta (p - 1)}}{{p - 2}}\bigg)^{\frac{{2 - p}}{2}}}{M^p}L_\sigma ^p\bigg(\int_0^t {{{\rm e}^{ - \beta (t - s)}}{{\left| {E\left\| {x(s)} \right\|_H^p} \right|}_\tau }{\rm d}s}\bigg )\\&&+ {10^{p - 1}}{M^p}{\gamma ^{ - p}}b_f^p + {10^{p - 1}}{M^p}{\beta ^{ - p}}b_g^p+ {10^{p - 1}}\frac{{p(p - 1)}}{2}{\bigg(\frac{{2\beta (p - 1)}}{{p - 2}}\bigg)^{\frac{{2 - p}}{2}}}\frac{{{M^p}b_\sigma ^p}}{\beta }.\\\end{eqnarray} $

接下来估计系统(1.1)的吸引集和拟不变集.

${\varphi \in BC_{{F_0}}^b([- \tau, 0], H)}$和(3.1)式,存在常数$N>0$$\alpha \in (0, \gamma \wedge \beta)$满足${\left| {E\left\| {{\phi (0)}} \right\|_H^p} \right|_\tau } \leqslant N$

$\begin{array}{l}\;\;\;\;{5^{p - 1}}{M^p}[E\left\| {\phi (0)} \right\|_H^p + E{({\left\| {{\phi _1}} \right\|_H} + {L_f}{\left| {{{\left\| {\phi (0)} \right\|}_H}} \right|_\tau } + {b_f})^p}]/N\\\;\;\;\; + {{\rm{e}}^{\alpha \tau }}{10^{p - 1}}{M^p}{\gamma ^{1 - p}}L_f^p/\gamma - \alpha \\\;\;\;\; + {{\rm{e}}^{\alpha \tau }}{10^{p - 1}}{M^p}[{\beta ^{1 - p}}L_g^p + \frac{{p(p - 1)}}{2}{(\frac{{2\beta (p - 1)}}{{p - 2}})^{\frac{{2 - p}}{2}}}L_\sigma ^p]/\beta - \alpha \\ < 1.\end{array}$

由引理2.4

因此${S_1}$是系统(1.1)解的吸引集.

(3.9)式通过放缩可得

依据(2.3)式,进一步放缩为

其中

因此, ${S_2}$是系统(1.1)的拟不变集.证毕.

接下来将讨论系统(1.1)的一些特殊情形.

推论3.1 设(H1)-(H2)成立,且${b_f} = {b_g} = {b_\sigma } = 0$, $p = 2$,那么系统(1.1)的解是全局均方指数稳定的,若

$\begin{equation}5{M^2}\bigg[\frac{{L_f^2}}{{{\gamma ^2}}} + \frac{{L_g^2}}{{{\beta ^2}}} + \frac{{L_\sigma ^2}}{\beta }\bigg] < 1.\label{e3.11}\end{equation}$

  首先建立不等式如下

$\begin{equation}{(a + b + c + d + e)^2} \le 5{a^2} + 5{b^2} + 5{c^2} + 5{d^2} + 5{e^2}, \label{e3.12}\end{equation}$

其中$a-e$是任意正的常数.

由(3.9)式, (3.12)式和Hölder不等式

$\begin{eqnarray}E\left\| {x(t)} \right\|_H^2 &\leqslant &5E\left\| {T(t){\phi (0)}} \right\|_H^2 + 5E\left\| {S(t)({\phi _1} - f(0, {x_0})} \right\|_H^2\\&\quad&+ 5E\left\| {\int_0^t {T(t - s)f(s, {x(s - \tau(s) )})}{\rm d}s} \right\|_H^2\\&\quad&+ 5E\left\| {\int_0^t {S(t - s)g(s, {x(s - \tau(s) )})}{\rm d}s} \right\|_H^2\\&\quad&+ 5E\;\left\| {\int_0^t {S(t - s)\sigma (s, {x(s - \tau(s) )})} {\rm d}Ws} \right\|_H^2\\&\le &5E\left\| {\phi (0)} \right\|_H^2{M^2}{{\rm e}^{ - \gamma t}} +5E{({\left\| {{\phi _1}} \right\|_H} + {L_f}{\left| {{{\left\| {\phi (0)} \right\|}_H}} \right|_\tau } + {b_f})^2}{M^2}{{\rm e}^{ - \beta t}}\\&\quad& + 5{M^2}{\gamma ^{ - 1}}L_f^2\bigg(\int_0^t {{{\rm e}^{ - \gamma (t - s)}}{{\left| {E\left\| {x(s)} \right\|_H^2} \right|}_\tau }{\rm d}s} \bigg)\\ &\quad &+ 5{M^2}({\beta ^{ - 1}}L_g^2 + L_\sigma ^2)\bigg(\int_0^t {{{\rm e}^{ - \beta (t - s)}}{{\left| {E\left\| {x(s)} \right\|_H^2} \right|}_\tau }{\rm d}s}\bigg ). \label{e3.13}\end{eqnarray}$

依据(3.11)式, (3.13)式和推论2.1可知,系统(1.1)的温和解全局均方指数稳定.证毕.

注3.1 当去掉脉冲项时,因为本文的时滞项是与时间$t$有关的函数,因此推论$3.1$可看为文献[14]中推论$3.3$的推广.比较文献[14]中的证明过程,本文的方法更加简洁.

$f(t, {x(t - \tau(t))}) \equiv 0, $那么系统(1.1)变为了如下的二阶随机偏微分方程

$ \begin{equation} \left\{ \begin{array}{l} dx'(t) = [Ax(t) + g(t, {x(t - \tau(t) )})]{\rm d}t + \sigma (t, {x(t - \tau(t) )}){\rm d}W(t), \;\;\;\;t \ge 0, \\ x_0(s) = \phi (s), \;\;\;\; - \tau \le s \le 0, \ \ \ \ \;x'(0) = {\phi _1}. \end{array} \right.\label{e3.14} \end{equation} $

推论3.2 设(H1)-(H2)和${b_g} = {b_\sigma } = 0$成立,那么系统(3.14)的温和解是全局$p$ -指数稳定的,若

  由(2.1), (2.2)和(3.9)式和引理2.1,命题得证.

注3.2   Arthi等在文献[14]中的引理$3.5$研究了与(3.14)类似的系统.如果不考虑脉冲项的影响,本文在时滞项上的研究可以看做是原来结论的推广.

4 实例分析

考虑如下系统

$\begin{eqnarray} \label{e3.15} \partial\bigg [\frac{\partial }{{\partial t}}v(t, x) - {\omega _1}v(t, x(t-1))\bigg]{\rm{ }} &=&\bigg[\frac{{{\partial ^2}}}{{\partial {x^2}}}v(t, x)+{\omega _2}v(t, x(t-1))+\sin t\bigg]{\rm d}t\\&&+ {\omega _3}v(t, x(t-1)){\rm d}W(t), \ \ t\ge 0, \;0 \le x \le \pi\end{eqnarray} $

$\begin{eqnarray} \label{e3.16} \left\{ \begin{array}{ll} \frac{\partial }{{\partial t}}v(0, x) = {v_1}(x), \ \ x \in [0, \pi ], \\v(s, x) = \varphi (s, x), \;\;\; s \in [ - 1 , 0], \ \ x \in [0, \pi ], \end{array} \right. \end{eqnarray} $

其中$ \varphi (s, \cdot) \in H, $$ \varphi (\cdot, x) \in BC_{{{\cal F}_0}}^b([-1, 0], H), $$H = {L^2}[0, \pi]$, ${\omega_i}(i = 1, 2, 3)$是非负的常数, $W(t)$是一个一维的Wiener过程.

$K = \mathbb{R}$,无穷小生成子为$A= \frac{{{\partial ^2}}}{{\partial {x^2}}}:H \to H$, $D(A) = \{ v \in H:\ v (\pi) = 0, \ v'' \in H\}$,那么

其中${\eta _k}(x) = \sqrt {\frac{2}{\pi }} \sin kx, \ k = 1, 2, \cdot \cdot \cdot, n, $${\left\{ {{\eta _k}} \right\}_{n \in {\Bbb N}}}$是关于特征向量$A$的完备正交集.

定义$f, \ g$$\sigma$满足

若(H1)-(H2)成立,且$M=1$, $\gamma = \beta = \pi$,由定理3.1,系统(4.1)和(4.2)温和解的吸引集为${S_1} = \left\{ {\varphi \in BC_{{{\cal F}_0}}^b} \right.$$\left. {\left. {([-1, 0], H)} \right|\left\| \varphi \right\|_{{L^p}}^p \le {{(1 - \Upsilon)}^{ - 1}}{I}} \right\}$,系统(4.1)和(4.2)温和解的拟不变集为${S_2} = \left\{ {\left. {\varphi \in BC_{{{\cal F}_0}}^b([-1, 0], H)} \right|\left\| \varphi \right\|_{{L^p}}^p \le R, \ R > 0} \right\}$,若

参考文献

Liu K . Stability of Infinite Dimensional Stochatic Differential Equations with Applications. London: Chapman Hall, 2006

[本文引用: 1]

Caraballo T , Real J , Taniguchi T .

The exponential stability of neutral stochastic delay partial differrential equations

Discrete Contin Dyn Syst, 2007, 18 (2/3): 295- 313

DOI:10.3934/dcdsa     

Bao J H , Hou Z T .

Existence of mild solutions to stochastic neutral partial differrential equations with non-Lipschitz coefficients

Comput Math Appl, 2010, 59 (1): 207- 214

DOI:10.1016/j.camwa.2009.08.035     

Chen H .

The asymptotic behaviour for second-order neutral stochastic partial differential equations with infinite delay

Discrete Dynamics in Nature and society, 2011, 2011: 531- 536

[本文引用: 2]

Xu D Y .

Invariant and attracting sets of Volterra difference equations with delays

Comput Math Appl, 2003, 45 (6): 1311- 1317

URL     [本文引用: 1]

Cheban D , Mammana C .

Invariant manifolds, global attractors and almost periodic solutions of nonautonomous difference equations

Nonlinear Anal, 2004, 56 (4): 465- 484

DOI:10.1016/j.na.2003.09.009      [本文引用: 1]

Xu D Y .

Asymptotic behavior of nonlinear difference equations with delays

Comput Math Appl, 2001, 42 (3): 393- 398

URL     [本文引用: 1]

Bernfeld S R , Corduneanu C , Ignatyev A O .

On the stability of invariant sets of functional differential equations

Nonlinear Anal, 2003, 55 (6): 641- 656

DOI:10.1016/j.na.2003.08.002      [本文引用: 1]

Liao X X , Luo Q , Zeng Z G .

Positive invariant and global exponential attractive sets of neural networks with time-varying delays

Neurocomputing, 2008, 71 (4): 513- 518

URL     [本文引用: 1]

Wang L S , Xu D Y .

Asymptotic behavior of a class of reaction-diffusion equations with delays

J Math Anal Appl, 2003, 281 (2): 439- 453

DOI:10.1016/S0022-247X(03)00112-4      [本文引用: 1]

Duan J Q , Lu K N , Schmalfu B B .

Invariant manifolds for stochastic partial differential equations

Ann Probab, 2003, 31 (4): 2109- 2135

DOI:10.1214/aop/1068646380      [本文引用: 1]

Lu K N , Schmalfu B B .

Invariant manifolds for stochastic wave equations

J Differential Equations, 2007, 236 (2): 460- 492

DOI:10.1016/j.jde.2006.09.024      [本文引用: 1]

Li Dingshi , Xu D Y .

Atttacting and quasi-invariant sets of stochastic neutral partial functional differential equations

Acta Mathematica Scientia, 2013, 33 (2): 578- 588

DOI:10.1016/S0252-9602(13)60021-1      [本文引用: 3]

Arthi G , Park Ju H , Jung H Y .

Exponential stability for second-order neutral stochastic differential equations with impulses

International Journal of Control, 2015, 88 (6): 1300- 1309

DOI:10.1080/00207179.2015.1006683      [本文引用: 3]

Fattorini H O . Second-Order Partial Differential Equations in Banach Spaces. North Holland: Elsevier, 2011

[本文引用: 1]

Wu F , Hu S , Liu Y .

Positive solution and its asymptotic behaviour of stochastic functional Kolmogorov-type system

J Math Anal Appl, 2010, 364 (1): 104- 118

DOI:10.1016/j.jmaa.2009.10.072      [本文引用: 1]

Ichikawa A .

Stability semigroups of stochastic evolution equations

J Math Anal Appl, 1982, 90 (1): 12- 44

DOI:10.1016/0022-247X(82)90041-5      [本文引用: 1]

Iko K , Nisio M .

On stationary solutions of a stochastic differential equation

J Math Kyoto Univ, 1964, 1964 (4): 1- 75

URL     [本文引用: 2]

/