数学物理学报, 2023, 43(4): 1221-1243

一类具有泊松跳的脉冲中立型随机泛函微分方程的存在性及稳定性研究

何旭阳,, 毛明志,*, 张腾飞

中国地质大学(武汉)数学与物理学院 武汉 430070

Existence and Stability of a Class of Impulsive Neutral Stochastic Functional Differential Equations with Poisson Jump

He Xuyang,, Mao Mingzhi,*, Zhang Tengfei

School of Mathematics and Physics, China University of Geosciences $($Wuhan$),$ Wuhan 430070

通讯作者: *毛明志, E-mail: mingzhi-mao@163.com

收稿日期: 2022-06-22   修回日期: 2023-02-14  

基金资助: 中国地质大学(武汉)基础研究中心基金(CUGSX01)

Received: 2022-06-22   Revised: 2023-02-14  

Fund supported: FRFC(CUGSX01)

作者简介 About authors

何旭阳,E-mail:1202010933@cug.edu.cn

摘要

该文考虑一类具有泊松跳的脉冲中立型随机泛函微分方程温和解的存在唯一性以及指数稳定性. 利用逐次逼近和Picard迭代方法, 证明了在Hilbert空间中温和解的存在性; 其次, 通过Banach不动点原理, 给出了均方指数稳定和几乎必然指数稳定的充分条件.

关键词: 存在唯一性; 中立型随机泛函微分方程; 温和解; 泊松跳跃; 指数稳定性

Abstract

In this paper, we consider the existence, uniqueness and exponential stability of mild solutions for a class of impulsive neutral stochastic functional differential equations with Poisson jumps.By using successive approximation and Picard iteration method, the existence of mild solutions in Hilbert space is proved. Secondly, the sufficient conditions for the mean square exponential stability and almost certain exponential stability are given by Banach's fixed point principle.

Keywords: Existence and uniqueness; Netral stochastic functional differential equations; Mild solution; Poisson jumps; Exponential stability

PDF (516KB) 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

何旭阳, 毛明志, 张腾飞. 一类具有泊松跳的脉冲中立型随机泛函微分方程的存在性及稳定性研究[J]. 数学物理学报, 2023, 43(4): 1221-1243

He Xuyang, Mao Mingzhi, Zhang Tengfei. Existence and Stability of a Class of Impulsive Neutral Stochastic Functional Differential Equations with Poisson Jump[J]. Acta Mathematica Scientia, 2023, 43(4): 1221-1243

1 引言

中立型泛函微分方程理论因其在化学工程系统、气动弹性和自动控制等领域的潜在应用而引起了众多研究者的关注. 例如 Hale等[1]$研究了确定性中立型泛函微分方程的基本理论, Liu[2]研究了一类中立型泛函微分方程的最优控制问题. 对于随机系统, 高斯白噪声通常被用作描述随机连续稳定现象的唯一干扰源. 然而, 在实际应用中, 系统可能会受到一些突然干扰的影响. 例如, 全球金融危机引发的股市剧烈震荡, 或由于气候变暖、海啸和地震等因素而导致的物种灭绝. 从这些现象可以看出, 仅用一个平滑的干扰噪声项所描述的系统不能满足实际需要. 为了建立更真实的模型, 泊松跳被引入到随机系统中, 描述了一种不连续的随机脉冲现象.

脉冲微分系统作为近年来一个非常活跃的研究课题, 吸引了不少学者的关注, 为了更好地描述在某些时间点状态发生突变的系统, Wu等[3]首先提出了一类具有脉冲效应的非线性微分系统模型, 在Lipschitz条件下利用Cauchy-Schwarz不等式研究解的存在唯一性, 并用李雅普诺夫直接法研究了$p$阶矩的有界性.

具有泊松跳的中立型随机泛函微分方程是一类重要的随机系统, 其稳定性分析近年来受到了密切关注. 由于大多数随机变量都是显式求解的, 因此随机分析的研究是基于数值解的, Wu 等[4]首次利用半鞅收敛定理获得了Euler-Maruyama (EM)方法的几乎必然指数稳定性. 随后, 沈轶等[5]研究了一般中立型随机泛函微分方程解的渐近性质, 利用李雅普诺夫函数和半鞅收敛定理, 得到了该方程解的渐近稳定性以及指数稳定性.

众所周知,稳定性理论在中立型泛函微分方程的研究中具有很重要的作用. 其经典而强大的一个技术是基于随机形式的Lyapunov直接法, 然而, 用李雅普诺夫直接法研究稳定性时常常会遇到困难, 为解决其中的困难, 崔静等[6]用不动点方法研究了由分数阶布朗运动驱动的脉冲中立型随机泛函微分方程温和解的$p$阶矩的渐近稳定性. Razumikhin 技术在研究各种时滞微分方程的稳定性方面也是非常有效的, 参见文献[7]. 自Chang[8]首次建立了有限时滞随机泛函微分方程的Razumikhin型一致渐近稳定性判据之后, 沈轶等[9]建立了这种方程的$p$阶均值指数稳定性和几乎必然指数稳定性的Razumikhin型定理,并将这些新结果应用到具有可变时滞的随机中立型微分方程中; Yang等[11]通过建立脉冲时滞微分不等式分析了脉冲时滞系统的指数全局稳定性和指数收敛速度的估计; Guo等[12]利用Razumikhin技术和李雅普诺夫函数研究了脉冲中立型随机泛函微分方程的Razumikhin型渐近稳定性, 其结果可以应用于脉冲随机方程和具有有界或无界脉冲随机方程和随机扰动方程中. 此外, Hyers-Ulam稳定性理论的发展为稳定性分析开辟了一个新的研究方向, 例如, 赵[13]讨论了受布朗运动干扰的的均方随机泛函微分方程的Hyers-Ulam稳定性; 随后Li等[14]在有界闭合区间上研究了Lipschitz条件下的Hyers-Ulam稳定性结果.

近些年来,带有时滞和泊松跳的随机微分方程在工程、物理和电子科学等领域广泛应用[15-16]. 为将该模型更好地应用于实际生产中, 许多学者开始研究该类方程解的存在唯一性, 其中, 较为常用的方法是Picard逼近技术[17-18]. 此外, Chen等[19]利用逐次逼近方法研究了一类具有时滞和泊松跳的脉冲中立型随机偏微分方程弱解的存在唯一性; $\!\!{\rm Ren}\!\!$[10]利用Bihari不等式在非Lipschtiz条件下研究了弱解的均方存在唯一性. 作为布朗运动的一种推广, 分数阶布朗运动具有自相似和非平稳的特点, 近年来受到了广泛的关注[22-23]. 例如, Deng等[20]利用非紧性的Hausdorff测度和Mönch不动点定理, 脉冲积分不等式, 研究了Hilbert空间中由非紧半群fBm驱动的一类脉冲中立型随机泛函微分方程温和解的存在性和指数稳定性; 由于几乎周期性比周期现象在物理学、生物学中有更广泛的应用, 因此在基于算子半群法和Mönch不动点法, 以及Hyers-Ulam稳定性的基本理论上, Guo等[21]研究了非局部条件下含脉冲和含分数阶布朗运动的微分方程的几乎周期解的存在性和Hyers-Ulam稳定性.

据作者所知,关于脉冲中立型随机泛函微分方程的工作甚少, 因而, 本文建立了一类具有泊松随机测度的中立型随机泛函微分方程的模型, 旨在研究其温和解的存在唯一性及其$p$阶指数稳定性, 其贡献在于在该系统中引入泛函项, 使所研究的方程对象更具体, 为随机泛函微分方程在工程中的后续应用提供某些理论支撑. 本文的难点在于如何求解泛函项的$p$阶期望值, 如何将其$p$阶期望压缩为上确界形式; 其次, 该模型包含一个强连续半群上的无穷小生成元$A$, 在处理过程中涉及到半群理论和线性算子的有界性, 在利用线性算子的有界性进行Picard迭代过程也存在一些困难; 最后, 该模型是一个具有泊松跳的脉冲系统, 由于随机干扰项的存在, 使得这类随机泛函微分方程温和解的存在性、唯一性和稳定性的处理方法往往受到限制.

本文的组织结构如下: 第1部分为引言; 第2部分引入了一些符号、假设、定义和相关引理; 第3部分是本文的研究方法, 包括利用概率不等式、Lipschitz条件、线性算子在强连续半群上的有界性理论来证明了系统温和解的存在唯一性; 第4部分利用Banach不动点理论证明本文所研究的系统在$p$时刻处是指数稳定的; 第5部分总结本文所用到的方法以及该模型的实际应用.

2 预备知识

$\big\{ {\Omega,{\cal F},{{\left\{ {{{\cal F}_t}} \right\}}_{t \ge 0}},P} \big\}$是完备概率空间, $\sigma $域流${\rm F}= {{\left\{ {{{\cal F}_t}} \right\}}_{t \ge 0}} $满足通常条件: 即${\rm F} $是右连续的且 ${\cal F}_{0}$包含所有的零测集.

$X$,$Y$是两个可分的实Hilbert空间, 记 ${\mathfrak L} \left ( X,Y \right )$ 表示从$X$$Y$上的所有有界线性算子的集合. 让 $\left \langle \cdot \right \rangle _{X}$ 表示$X$空间上的内积, $\left | \cdot \right | _{X}$ 相应地表示$X$空间上的向量范数. 假设 $ p\left (t \right ) _{t\geq 0} $ 取值于一个可测空间$\left ( U,{\cal B}\left (U\right ) \right )$ 上的一个$\sigma$ -有限平稳${{\cal F}_{t}}$ -适应的泊松过程, 定义$N_{p}\left ( \left ( 0,t \right ] \times {\Lambda} \right ):=\sum\limits_{s\in (0,t]}1_{\Lambda } \left (p\left ( s \right ) \right )$, $\Lambda \in{\cal B}\left (U\right)$, 易知$N_{p}$是一个随机测度, 此测度通常被称为是由 $p\left ( \cdot \right )$ 生成的泊松随机测度. 再定义测度$\widetilde{N}({\rm d}t,{\rm d}y)=N_{p}\left ( {\rm d}t,{\rm d}y \right )-\nu \left ({\rm d}y\right ){\rm d}t$, 这里$\upsilon$通常被称为 $N_{p}$ 的特征测度. 令 $\omega\left ( t \right ) $$\big \{ \Omega,{\cal F},\left \{ {\cal F}_{t}\right \}_{t\geq 0},P \big\}$空间上的一个一维的布朗运动, 且有协方差算子 ${\cal Q}$, 即

$E\left \langle \omega \left ( t \right ),x\right \rangle_{X}\left \langle \omega \left ( s \right ),y \right \rangle_{X}=\left (t\wedge s \right )\left \langle {\cal Q} x,y \right \rangle_{X}, \; x,y\in X,$

其中 ${\cal Q}$$X$上的一个正的自伴迹类算子. ${\mathfrak L} _{2}^{0}\left ( X,Y \right )$ 表示从$Q^{1/2}X$$Y$ 的所有 $Q$-Hilbert-Schmide 算子的空间, 其范数定义为 $\left\| \xi \right\|_{{\mathfrak L}_{2}^{0}}^{2}:={\rm tr}\left ( \xi Q\xi ^{\ast } \right )<\infty$.

下面将介绍无穷小生成元和线性算子的连续半群的概念,在Banach空间 $B$ 中, 族 $S=\left \{ S\left ( t \right ):t\ge 0 \right \} $ 上的有界线性算子被称为是一个${\cal C}_{0} $ 类半群, 满足以下四个条件

(a) $S\left ( 0 \right )=I$, uad ($I$$B$ 上的单位算子);

(b) 对每个 $ t,s\ge 0 $, 有$S\left ( t+s \right )=S\left ( t \right )S\left ( s \right )$;

(c) 对每个 $x\in B$, $S\left ( t \right )x$$t \in\left [ 0,\infty \right ) $$B$上是连续成立的, ($C_{0} $ 的性质);

此外,

(d) 对每个 $t\ge 0$, 有$\left \|S\left ( t \right ) \right \|\le 1 $.

假设$\left \{ S\left ( t \right );t\ge 0 \right \}$ 是定义在 $B$ 上的一个线性的压缩半群, 记$D\subset B$, $x\in D $, 且$S\left (t \right )x$$t=0$ 处是右可微的, 即

$ D= \Big\{ x\in X \mid\lim\limits_{h\to +0 }\frac{S\left ( h \right )x-x }{h} \Big\}. $

$ -Ax=\lim\limits_{h \to + 0} \frac{S\left ( h \right )x-x }{h}.$

通常称$ -A$$S\left ( t \right )$上的无穷小生成元. 设$\alpha \in (\frac{1}{2},1)$, 则$\left ( -A \right )^{\alpha }$ 是预解集$D\big(\left ( -A \right ) ^{\alpha }\big) $上一个闭线性算子, 其子空间在Hilbert空间$Y$ 上是稠密的.

对于博雷尔集$ {\cal B}\left ( U-\left \{ 0 \right \} \right )$, 考虑带有泊松跳的中立型随机泛函微分方程

$\begin{matrix}\left\{\begin{array}{ll}d\big[ x\left ( t \right )+u\left ( t,x\left ( t \right ),x_{t}\right )\big] = Ax\left ( t \right ){\rm d}t+f\left ( t,x\left ( t \right ),x_{t} \right ){\rm d}t +g\left ( t,x\left ( t \right ),x_{t} \right ){\rm d}\omega \left ( t \right ) \\ +\int _{Z}h\left ( t,x\left ( t \right ),x_{t},y\right )\widetilde{N}({\rm d}t,{\rm d}y),\; t\ge 0,\\ \Delta x\left ( t_{k} \right )=x\left ( t_{k}^{+}\right)- x\left ( t_{k}^{-}\right)=I_{k}\left ( x\left ( t_{k}^{-} \right ) \right ),\;k=1,2,\cdots, \\ x\left ( \vartheta \right )=\phi \in {\cal PC},\;\vartheta \in \left [ -\tau,0 \right ],\; {\rm a.s. },\\ x_{0}\left ( \cdot \right )=\varphi \in C_{{\cal F}_{0} }^{b}\left ( \left [ -\tau,0 \right ] ;X\right ),\;{\rm a.s.} \end{array}\right.\end{matrix} $

这里, $x_{t}=\left \{ x\left ( t+\theta \right ): -\tau \leqslant \theta \leqslant 0 \right \}$ 是取值于 $C\left ( \left [ -\tau,0 \right ];X \right )$ 的随机过程, 其范数定义为 $\left|\left \| x_{t} \right \|\right|=\sup\limits_{t -\tau \le s\le t}\left \| x\left ( s \right ) \right \|$. 此外, 定义 $I_{k}\in C\left ( X,X \right )$ 是一个脉冲映射函数, 这里

$\begin{eqnarray*}& &u:R_{+}\times X\times C\left ( \left [ -\tau,0 \right ];X \right )\rightarrow X,\\ & & f:R_{+}\times X \times C\left ( \left [ -\tau,0 \right ];X \right )\rightarrow X,\\ & & g:R_{+}\times X \times C\left ( \left [ -\tau,0 \right ];X \right )\rightarrow {\mathfrak L} _{2}^{0}\left ( X,Y \right ),\\ & & h:R_{+}\times X\times C\left ( \left [ -\tau,0 \right ];X \right )\times {\Bbb Z} \to X \end{eqnarray*} $

都是博雷尔可测的. 令${\cal PC} \equiv {\cal PC} \left ( \left [-\tau,0\right ];X\right )$ 是所有几乎处处有界的 $ {\cal F}_0$ -可测函数空间, 其范数定义为 $\left \| \phi \right \|_{0}=ess\sup\limits_{\omega \in \Omega }\sup\limits_{t\in \left [-\tau,0\right ]}\left \| \phi \left ( t\right )\left ( \omega \right )\right \|$. 此外, $t_{k}$ 指的是脉冲跳跃时间, 且 $x\left (t_{k}^{+}\right )=\lim\limits_{h\rightarrow 0^{+}}x\left (t_{k}+h\right )$, $x\left ( t_{k}^{-}\right )=\lim\limits_{h\rightarrow 0^{+}}x\left ( t_{k}-h\right )$, 以及脉冲跳跃幅度为$\Delta x\left ( t_{k}\right )=x\left (t_{k}^{+}\right )-x\left ( t_{k}^{-}\right )$.

模型(2.1)是时滞随机递归神经网络领域中一个较为常用的模型, 在神经网络领域中全局指数稳定性在当前学术领域是非常感兴趣的, 通常是构造Lyapunov-Krasovskii泛函来讨论指数收敛速度估计, 从而得到时滞相关的指数稳定性条件. 对该系统稳定性的研究可以成功地将神经网络应用于模式识别、图像处理、联想记忆、优化计算和安全通信等领域, 尤其是在电路设计和超大规模电路实现的正确性方面有许多应用背景, 详情参考文献[24].

定义 2.1$X$ 值随机过程 $\left \{ x\left ( t \right ),t\in \left [ 0,T \right ] \right \}$ 满足如下两个条件

(a) $x(t)$${\cal F}_t$适应的, 且 $\int_{0}^{T}\left \langle x\left ( t \right ),x\left ( t \right ) \right \rangle_{H}{\rm d}t< \infty $ 是几乎处处成立的;

(b) $x(t)$$t\in \left [ 0,T \right ]$上几乎处处有 càdlàg 路径, 满足积分方程

$\begin{eqnarray*} \left\{\begin{array}{ll} x\left ( t \right )=S\left ( t \right )\left ( \phi \left ( 0 \right )+u\left ( 0,\phi,\varphi \right )-u\left ( t,x\left ( t \right ),x_{t} \right ) \right )-\int_{0}^{t}AS\left ( t-s \right )u\left ( s,x\left ( s \right ),x_{s} \right ){\rm d}s\\[3mm] \ uad+\int_{0}^{t}S\left ( t-s \right )f\left ( s,x\left ( s \right ),x_{s} \right ){\rm d}s+\int_{0}^{t}S\left ( t-s \right )g\left ( s,x\left ( s \right ),x_{s} \right ){\rm d}\omega \left ( s \right )\\[3mm] \ uad+\int_{0}^{t}\int _{Z}S\left ( t-s \right )h\left ( s,x\left ( s \right ),x_{s},y\right ) \widetilde{N}\left ( {\rm d}s, {\rm d}y \right )+\sum _{0<t_{k}<t}S\left ( t-t_{k} \right )I_{k}\left ( x\left ( t_{k}^{-} \right ) \right ),\\ x\left (\vartheta \right )=\phi \in PC,uad \vartheta \in \left [ -\tau,0\right ],\; {\rm a.s.},\\ x_{0}\left (\cdot \right )=\varphi \in C_{{\cal F}_{0}}^{b}\left ( \left [ -\tau,0 \right ];X \right ),\; {\rm a.s.}. \end{array}\right. \end{eqnarray*}$

则称$x(t)$是系统(2.1)的温和解

本文的主要目的是证明系统(2.1)的存在唯一性, 为此给出条件

(A1) $A$$B$上的有界线性算子 $\left \{ S\left ( t \right ),t\ge 0 \right \}$上的解析半群上的无穷小生成元, 其中 预解集$\rho \left ( -A \right )$包含0, 由于 $S(t)$ 是一致有界的, 则存在两个正数 $\gamma$$M$, 满足

$ \begin{matrix} \left \| S\left ( t \right ) \right \| \le Me^{-\gamma t}. \end{matrix}$

事实上, 存在某一个$\eta>0 $ 使得 $\left \| S\left ( t \right ) \right \|$$0\le t\le \eta $上有界, 如若不然, 则存在一个正序列$\left \{ t_{n} \right \} $, 满足$\lim\limits_{n \to \infty}t_{n}=0 $$ \left \| S\left ( t_{n} \right ) \right \|\ge n$. 由一致有界定理可得出, 对某个$x\in B$, $ \left \| S\left ( t_{n} \right )x \right \|$不是有界的, 从而矛盾. 因此, $\left \| S\left ( t \right ) \right\|\le M$ 是成立的. 由于 $\left \| T\left ( 0 \right ) \right \|=1$$M\ge 1$.$ -\gamma =\eta ^{-1}lnM$, 有 $t=n\eta +\delta$, 其中$0\le \delta < \eta $, 再由半群的性质可知

$\left \| S\left ( t \right ) \right \|=\left \| S\left ( n\eta +\delta \right ) \right \|=\left \| S\left ( \delta \right )S\left ( \eta \right )^{n} \right \|\le M^{n+1}\le MM^{\frac{t}{\eta } }=Me^{-\gamma t}.$

(A2) 映射 $f\left ( t,\cdot,\cdot \right ), g\left ( t,\cdot,\cdot \right ),h\left ( t,\cdot,\cdot\right )$ 满足如下局部Lipschitz条件和线性增长条件: 即存在一个正常数 $L_{1}, L_{2}, L_{3}$,对 $\forall \;x,y\in X$满足

$ \begin{eqnarray*}&&\left \| f\left ( t,x,x_{t}\right )-f\left ( t,y,y_{t}\right )\right \|_{X} \leqslant L_{1}\left ( \left \| x-y\right \|_{X} +\left |\left \| x_{t}-y_{t}\right \|\right | _{X} \right ),\\ &&\left \| g\left ( t,x,x_{t}\right )-g\left ( t,y,y_{t}\right )\right \|_{X} \leqslant L_{2}\left ( \left \| x-y\right \|_{X} +\left |\left \| x_{t}-y_{t}\right \|\right |_{X} \right ), \\ &&\int _{Z}\left \| h\left ( t,x,x_{t},z\right )-h\left ( t,y,y_{t},z\right )\right \|_{X} ^{2}\upsilon \left ({\rm d}z\right )\leqslant L_{3}\big ( \left \| x-y\right \|_{X} ^{2}+\left |\left \| x_{t}-y_{t}\right \|\right |_{X} ^{2}\big ),\end{eqnarray*}$

(A3) 映射 $\left ( -A\right )^{\alpha }u\left ( t,\cdot,\cdot \right )$ 满足一致Lipschitz条件: 即存在一个正常数 $L$, 使得对于任意的 $x,y\in X$,

$\left \| \left ( -A\right )^{\alpha }u\left ( t,x,x_{t}\right )-\left ( -A\right )^{\alpha }u\left ( t,y,y_{t}\right )\right \|_{X} \leq L\left ( \left \| x-y\right \|_{X} +\left |\left \| x_{t}-y_{t}\right \|\right |_{X} \right ),$

其中 $\alpha \in (\frac{1}{p},1]\left ( p\geqslant 2 \right )$$ u\left ( t,\cdot,\cdot \right )\in D\left ( \left ( -A \right )^{\alpha } \right )$.此外, 记 $\iota =\left \| \left ( -A\right )^{\alpha }\right \|L<1$.

(A4) 对 $\forall \;k\ge 1$, 有 $I_{k}\left ( 0 \right )=0$, 且对每个 $x,y\in X$, 存在一个正常数 $n_{k}$ 使得

$\left \| I_{k}\left ( x \right )-I_{k}\left ( y \right ) \right \|_{X} \leqslant n_{k}\left \| x-y \right \|_{X},uad \sum\limits_{k=1}^{+\infty }n_{k}<+\infty.$

在证明该系统的温和解的存在唯一性中, 以下三个引理很关键.

引理 2.1[25] 假设 (A1) 成立, 那么对于$\forall \;\alpha\in (0,1]$, 以下两个条件成立

(i) 对每个 $x\in {\cal D}\left ( \left ( -A\right )^{\alpha }\right ),$

$S\left ( t\right )\left ( -A\right )^{\alpha}x=\left ( -A\right )^{\alpha }S\left ( t\right )x,$

(ii) 存在一个正常数 $M_{\alpha }> 0$ 使得

$\left \| \left ( -A\right )^{\alpha }S\left ( t\right )\right \|\leqslant M_{\alpha }t^{-\alpha }e^{-\gamma t}, uad t>0.$

引理 2.2[26] 对于任意的$p\geqslant 2$ 和一个任意的 ${\cal L}_{2}^{0}$ -值可预测过程 $\phi \left ( \cdot \right )$, 有

$\sup\limits_{s\in \left [ 0,t\right ]}E\bigg \| \int_{0}^{s}\phi \left ( u\right ){\rm d}\omega \left ( u\right )\bigg \|^{p}_{X} \leqslant c_{p}\bigg ( \int_{0}^{t}\big ( E\left \| \phi \left (s\right )\right \|^{p}_{{\cal L}_{2}^{0}}\big )^{\frac{2}{p}}{\rm d}s\bigg )^{\frac{p}{2}},$

其中 $c_{p}=\left ( p\left ( p-1\right )/2\right )^{p/2}.$

引理 2.3[27]$ p\ge 1$,且 $\nu \in \left ( 0,1\right )$. 对于任意的两个实数$a,b>0,$

$\left ( a+b\right )^{p}\leq \nu ^{1-p}a^{p}+\left ( 1-\nu \right )^{1-p}b^{p}.$

3 存在唯一性

本节考虑用逐次逼近法证明系统(2.1)温和解的存在唯一性, 为此引入引理

引理 3.1 对于给定的6个实数$a_{i} $, $1\le i\le 6$, 且$ p\ge 1$, 取 $\varepsilon > 0 $, 有

$\begin{equation} \bigg(\sum\limits_{i=1}^{6}a_{i}\bigg)^{p} \le 3^{p-1}\big ( 1+\frac{1}{\varepsilon } \big )^{p-1} a_{1} ^{p}+9^{p-1}\left ( 1+\varepsilon \right )^{p-1}\sum\limits_{i=2}^{4}a_{i}^{p} +3^{p-1}\sum\limits_{i=5}^{6}a_{i}^{p}. \end{equation}$

由初等不等式可知: $ \left ( a+b \right )^{p} \le \upsilon ^{1-p} +\left ( 1-\upsilon \right )^{1-p} b^{p} $, 其中$\upsilon \in \left ( 0,1 \right ) $, 那么

$\begin{eqnarray*} \bigg(\sum\limits_{i=1}^{6}a_{i}\bigg)^{p}&\le& 3^{p-1} \bigg(\sum\limits_{i=1}^{4}a_{i}\bigg)^{p}+3^{p-1} \bigg (\sum\limits_{i=5}^{6}a_{i}\bigg)^{p}+ 3^{p-1}\sum\limits_{i=5}^{6}a_{i}^{p}\\ &\le &3^{p-1}\bigg [ \big( 1+\frac{1}{\varepsilon } \big ) ^{p-1}a_{1} ^{p}+\left ( 1+\varepsilon \right )^{p-1} \bigg(\sum\limits_{i=2}^{4}a_{i}\bigg)^{p} \bigg] +3^{p-1}\sum\limits_{i=5}^{6}a_{i}^{p}\\ &\le&3^{p-1}\bigg [ \big ( 1+\frac{1}{\varepsilon }\big)^{p-1} a_{1} ^{p}+\left ( 1+\varepsilon \right)^{p-1}3^{p-1}\sum\limits_{i=2}^{4}a_{i}^{p}\bigg]+3^{p-1}\sum\limits_{i=5}^{6}a_{i}^{p} \\ &\le& 3^{p-1}\left ( 1+\frac{1}{\varepsilon } \right )^{p-1} a_{1} ^{p}+9^{p-1}\left ( 1+\varepsilon \right )^{p-1}\sum\limits_{i=2}^{4}a_{i}^{p}+3^{p-1}\sum\limits_{i=5}^{6}a_{i}^{p}.\end{eqnarray*} $

证毕.

引理 3.2$y=t-s$, $ p\ge2$时, 由 $0<s<t$, 有

$\begin{equation} \int_{0}^{t}\left(t-s\right)^{\left(\alpha-1\right)\cdot\frac{p}{p-1}}e^{-\gamma\left (t-s\right)}{\rm d}s\le\gamma^{\frac{1-\alpha p}{p-1}}\Gamma\Big(1+\frac{p\left(\alpha-1\right)}{p-1}\Big). \end{equation}$

由数学分析知识易证.

引理 3.3 假设 (A3) 成立, 当$ p\ge 2$$\upsilon \in \left ( 0,1 \right )$时, 有

$ \begin{matrix}E\big\|u\left(t,x^{n}\left(t\right),x_{t}^{n}\right)\big\|_{X} ^{p} &\le& L^{p}\big\| \left ( -A \right )^{-\alpha }\big \|_{X} ^{p}\mu ^{1-p}E\big\| x^{n}\left ( t \right )\big \|_{X} ^{p}\\&&+L^{p}\big\| \left ( -A \right )^{-\alpha }\big\|_{X} ^{p}\xi ^{1-p}E\left|\left\|x_{t}^{n}\right \|\right|_{X} ^{p}+L^{p} \big\| \left ( -A \right )^{-\alpha }\big \|_{X} ^{p}\lambda ^{1-p}E\left \| \varphi \right \|_{X} ^{p}\\&&+\big\| \left ( -A \right )^{-\alpha }\big \|_{X} ^{p}\left ( 1-\upsilon \right )^{1-p}E\big \| \left ( -A \right )^{\alpha}u\left ( t,0,\varphi\right )\big \|_{X} ^{p}, \end{matrix}$

其中 $\mu =\upsilon ^{2}$, $\xi = \upsilon ^{2} \left ( 1-\upsilon \right ) $, $\lambda =\upsilon \left ( 1-\upsilon \right )^{2} $.

由引理 2.3 和假设 (A3), 可得出

$\begin{eqnarray*}E\left\|u\left(t,x^{n}\left(t\right),x_{t}^{n}\right)\right\|_{X} ^{p} &\le& \big\| \left ( -A \right )^{-\alpha }\big \|_{X} ^{p}\upsilon^{1-p}E\left \| \left ( -A \right )^{\alpha }u\left ( t,x^{n}\left ( t \right ),x_{t}^{n}\right )-\left ( -A \right )^{\alpha}u\left(t,0,\varphi\right)\right\|_{X} ^{p}\\&&+\left ( 1-\upsilon \right )^{1-p}\left \| \left ( -A \right )^{\alpha }\right \|_{X} ^{p}E\left \| \left ( -A \right )^{\alpha}u\left ( t,0,\varphi\right )\right \|_{X} ^{p} \\ &\le& L^{p}\big \| \left ( -A \right )^{-\alpha}\big\|_{X} ^{p}\upsilon^{1-p}E\left ( \left \| x^{n}\left(t\right)\right \|_{X} +\left \| x_{t}^{n}-\varphi\right \|_{X} \right)^{p}\\&&+\big\| \left ( -A \right )^{-\alpha } \big \|_{X} ^{p}\left ( 1-\upsilon \right )^{1-p}E\left \| \left ( -A \right )^{\alpha }u\left ( t,0,\varphi \right )\right \|_{X} ^{p}\\ &\le& L^{p}\big\| \left ( -A \right )^{-\alpha }\big \|_{X} ^{p}\mu ^{1-p}E \big \| x^{n}\left ( t \right )\big \|_{X} ^{p}+L^{p} \big\| \left ( -A \right )^{-\alpha }\big \|_{X} ^{p}\xi ^{1-p}E\left|\left\|x_{t}^{n}\right \|\right|_{X} ^{p}\\&&+L^{p}\big \| \left ( -A \right )^{-\alpha }\big \|_{X} ^{p}\lambda ^{1-p}E\left \| \varphi \right \|_{X} ^{p}\\&&+\big \| \left ( -A \right )^{-\alpha }\big \|_{X} ^{p}\left ( 1-\upsilon \right )^{1-p}E\left \| \left ( -A \right )^{\alpha}u\left ( t,0,\varphi\right )\right \|_{X} ^{p},\end{eqnarray*}$

证毕.

引理 3.4 假设(A1)和(A3)成立, 当 $ p\ge 2$$\upsilon \in \left ( 0,1 \right ) $时, 有

$\begin{matrix}& &E\left\|\int_{0}^{t}AS\left(t-s\right)u\left(s,x^{n}\left(s\right),x_{s}^{n}\right){\rm d}s\right\|_{X} ^{p} \\ &\le& \upsilon^{1-p}M_{1-\alpha}^{p}L^{p}\gamma^{1-\alpha p}\Big(\Gamma\big(1+\frac{p\left(\alpha-1\right)}{p-1}\big)\Big)^{p-1}\int_{0}^{t}e^{-\gamma\left(t-s\right)}\big[\upsilon^{1-p}E\left\|x^{n}\left(s\right)\right\|_{X} ^{p}\\&&+\left(\upsilon\left(1-\upsilon\right)\right)^{1-p}E\left|\left\|x_{s}^{n}\right\|\right|_{X} ^{p}+\left(1-\upsilon\right)^{2-2p}E\left\|\varphi\right\|_{X} ^{p}\big]{\rm d}s\\&&+\left (1-\upsilon\right)^{1-p}M_{1-\alpha}^{p}\gamma ^{1-\alpha p}\Big(\Gamma \big( 1+\frac{p\left ( \alpha -1 \right ) }{p-1}\big)\Big)^{p-1}\int_{0}^{t}e^{-\gamma\left(t-s\right)}E\left\|\left(-A\right)^{\alpha}u\left(s,0,\varphi\right)\right \|_{X} ^{p}{\rm d}s.\\ \end{matrix}$

由引理2.1知$\big \| \left ( -A \right )^{1-\alpha}S\left ( t-s \right )\big \| \le M_{1-\alpha}\left ( t-s \right )^{\alpha -1}e^{-\gamma \left ( t-s \right ) }$, 进一步, 由引理 2.1 和 2.3 知

$\begin{eqnarray*} & &E\left\|\int_{0}^{t}AS\left (t-s\right )u\left(s,x^{n}\left(s\right),x_{s}^{n}\right){\rm d}s\right\|_{X} ^{p}\\ &\le&\upsilon^{1-p}\bigg(\int_{0}^{t}LM_{1-\alpha}\left(t-s\right)^{\alpha-1}e^{-\gamma\left(t-s\right)}E\left(\left\|x^{n}\left(s\right)\right \|_{X} +\left|\left\|x_{s}^{n}-\varphi\right \|\right|_{X} \right){\rm d}s\bigg)^{p}\\&&+\left(1-\upsilon\right)^{1-p}\bigg(\int_{0}^{t}M_{1-\alpha}\left(t-s\right)^{\alpha-1}e^{-\gamma\left(t-s\right)}E\left\|\left(-A\right)^{\alpha}u\left ( s,0,\varphi\right)\right\|_{X}{\rm d}s\bigg)^{p}\\ &: =&\upsilon ^{1-p}I_{1}+\left ( 1-\upsilon \right )^{1-p}I_{2}.\end{eqnarray*}$

$I_{1}$用Holder 不等式以及引理3.2知

$\begin{eqnarray*} I_{1}&\le& M_{1-\alpha}^{p}L^{p}\bigg (\int_{0}^{t}\left (t-s\right )^{\left (\alpha -1 \right )\frac{p}{p-1}}\cdot e^{-\gamma\left(t-s\right)}{\rm d}s\bigg)^{p-1}\\&&\cdot \int_{0}^{t}e^{-\gamma \left(t-s\right)}E\left(\left\|x^{n}\left(s\right)\right\|_{X}+\left|\left\|x_{s}^{n}-\varphi\right \|\right|_{X}\right )^{p}{\rm d}s\\ &\le &M_{1-\alpha}^{p}L^{p}\gamma^{1-\alpha p}\Big(\Gamma\big(1+\frac{p\left(\alpha-1\right)}{p-1}\big)\Big)^{p-1}\int_{0}^{t}e^{-\gamma\left(t-s\right)} \big[\upsilon^{1-p}E\left\|x^{n}\left(s\right)\right\|_{X}^{p}\\&&+\left(\upsilon\left(1-\upsilon\right)\right)^{1-p}E\left|\left\|x_{s}^{n}\right\|\right|_{X}^{p}+\left(1-\upsilon\right)^{2-2p}E\left\|\varphi\right\|_{X}^{p}\big]{\rm d}s.\end{eqnarray*}$

同理有

$\begin{eqnarray*} I_{2}\le M_{1-\alpha}^{p}\gamma ^{1-\alpha p}\Big(\Gamma\big(1+\frac{p\left(\alpha-1\right)}{p-1}\big)\Big)^{p-1}\int_{0}^{t}e^{-\gamma\left(t-s\right)}E\left\|\left(-A\right)^{\alpha}u\left(s,0,\varphi\right)\right\|_{X}^{p}{\rm d}s.\end{eqnarray*}$

因此, 我们获得了引理3.4. 证毕.

引理 3.5 假设(A1)和(A2)成立, $ p\ge 2$时, 则

$ \begin{matrix} & &E\left \| \int_{0}^{t}S\left ( t-s \right )f\left ( s,x^{n-1}\left ( s \right ),x_{s}^{n-1}\right ){\rm d}s\right \|_{X}^{p}\\ &\le & M^{p}L_{1}^{p}\gamma ^{1-p}\cdot \int_{0}^{t}e^{-\gamma \left ( t-s \right ) }\big [ \upsilon ^{1-p}E \left \| x^{n-1}\left(s\right)\right\|_{X}^{p}+\left ( \upsilon \left ( 1-\upsilon \right ) \right ) ^{1-p}E\left|\left\|x_{s}^{n}\right\|\right|_{X}^{p}\\&&+\left ( 1-\upsilon\right )^{2\left ( 1-p \right )}E\left \| \varphi \right \|_{X}^{p}\big]{\rm d}s+M^{p}\gamma ^{1-p}\cdot\int_{0}^{t}e^{-\gamma \left ( t-s \right ) }E\left \| f\left ( s,0,\varphi \right ) \right \|_{X}^{p}{\rm d}s. \end{matrix}$

由假设 (A1)知$S(t-s)$ 是一致有界的, 因此当 $t\geqslant s$时, 有 $\left \| S\left ( t-s\right )\right \|\leq Me^{-\gamma \left ( t-s\right )}$, 又$f\left ( t,\cdot,\cdot \right )$ 满足 Lipschitz 和线性增长条件, 故

$\left \| f\left ( s,x^{n-1}\left ( s\right ),x_{s}^{n-1}\right )-f\left ( s,0,\varphi \right )\right\|_{X}\leq L_{1}\left ( \left \| x^{n-1}\left ( s\right )\right \|_{X}+\left|\left\|x_{s}^{n-1}-\varphi\right \|\right|_{X}\right ),$

从而有

$\begin{eqnarray*}&&E\left \| \int_{0}^{t}S\left ( t-s \right )f\left ( s,x^{n-1}\left ( s \right ),x_{s}^{n-1}\right ){\rm d}s\right \|_{X}^{p}\\&\le& M^{p}L_{1}^{p}E\left[\int_{0}^{t}e^{-\left ( \gamma \left ( p-1 \right )/p \right )\left ( t-s\right)}\cdot e^{-\left ( \gamma /p \right )\left ( t-s \right ) } \left ( \left \| x^{n-1}\left(s\right)\right\|_{X}+\left|\left\|x_{s}^{n-1}-\varphi\right \|\right|_{X}\right){\rm d}s\right]^{p}\\&&+M^{p}E\left [ \int_{0}^{t}e^{-\left ( \gamma \left ( p-1 \right )/p \right )\left ( t-s \right ) }\cdot e^{-\left ( \gamma /p \right )\left ( t-s \right ) } \left \| f\left ( s,0,\varphi\right ) \right \|_{X}{\rm d}s\right]^{p}\\&\le& M^{p}L_{1}^{p}\gamma ^{1-p}\cdot \int_{0}^{t}e^{-\gamma \left ( t-s \right ) }\big [ \upsilon ^{1-p}E \left \| x^{n-1}\left(s\right)\right\|_{X}^{p}+\left ( \upsilon \left ( 1-\upsilon \right ) \right ) ^{1-p}E\left|\left\|x_{s}^{n}\right\|\right|_{X}^{p}\\&&+\left ( 1-\upsilon\right )^{2\left ( 1-p \right )}E\left \| \varphi \right \|_{X}^{p}\big]{\rm d}s+M^{p}\gamma ^{1-p}\cdot\int_{0}^{t}e^{-\gamma \left ( t-s \right ) }E\left \| f\left ( s,0,\varphi \right ) \right \|_{X}^{p}{\rm d}s.\end{eqnarray*}$

证毕.

引理 3.6 假设 (A1)和(A2)成立, $ p\ge2$时, 则

$\begin{matrix} &&E\left \| \int_{0}^{t}S\left ( t-s \right )g\left ( s,x^{n-1}\left ( s \right ),x_{s}^{n-1}\right){\rm d}\omega \left(s\right)\right\|_{X}^{p}\\ &\le& c_{p} M^{p}\Big( \frac{p-2}{2\gamma \left ( p-1 \right ) } \Big )^{1-p/2 }\bigg \{L_{2}^{p} \cdot \int_{0}^{t} e^{-\gamma \left ( t-s \right )}\big [\upsilon ^{2-2p}E \left \| x^{n-1}\left ( s \right )\right \|_{X}^{p}\\ &&+\left ( \upsilon ^{2} \left ( 1-\upsilon \right ) \right ) ^{1-p} E\left|\left\|x_{s}^{n-1}\right\|\right|_{X}^{p}+ \big (\upsilon \left ( 1-\upsilon \right )^{2}\big)^{1-p} E\left \| \varphi \right \|_{X}^{p} \big ]{\rm d}s\\ &&+\left ( 1-\upsilon\right )^{1-p} \int_{0}^{t}e^{-\gamma \left ( t-s \right )} E\left \|g\left ( s,0,\varphi \right ) \right \|_{X}^{p}{\rm d}s \bigg \}.\end{matrix}$

其证明过程类似于引理 3.5的证明.

引理 3.7 假设(A1)和(A2)成立, $ p\ge 2$时, 则

$\begin{matrix} &&E\left\|\int_{0}^{t}\int _{{\Bbb Z}}S\left ( t-s \right )h\left (s,x^{n-1}\left ( s \right ),x_{s}^{n-1},y\right)\widetilde{N}\left({\rm d}s, {\rm d}y\right)\right\|_{X} ^{p}\\ &\le &c_{p}M^{p}2^{\frac{p}{2}}L_{3}^{p}\Big(\frac{p-2}{2\gamma\left(p-1\right)} \Big)^{\frac{p-2}{2}} \int_{0}^{t}e^{-\gamma\left(t-s\right)}\big(\upsilon^{1-p}E\left\|x^{n-1}\left(s\right)\right\|_{X} ^{p} \\ &&+\left ( \upsilon \left(1-\upsilon\right)\right)^{1-p}E\left|\left\|x_{s}^{n-1}\right\|\right|_{X} ^{p}+\left ( 1-\upsilon\right)^{2-2p}E\left\|\varphi\right\|_{X} ^{p}\big){\rm d}s\\&&+c_{p}M^{p}2^{\frac{p}{2}}\Big (\int_{0}^{t}e^{-2\gamma\left(t-s\right)}E\left \| h\left ( s,0,\varphi,y\right)\right\|_{X} ^{2}\nu \left ({\rm d}y\right ){\rm d}s\Big)^{\frac{p}{2}}.\end{matrix}$

假设(A1)和(A2)成立, 由引理2.3知

$\begin{matrix}&&E\left\|\int_{0}^{t}\int _{{\Bbb Z}}S\left ( t-s \right )h\left (s,x^{n-1}\left ( s \right ),x_{s}^{n-1},y\right)\widetilde{N}\left({\rm d}s, {\rm d}y\right)\right\|_{X} ^{p}\\&\le& c_{p}E\Big ( \int_{0}^{t}\int _{{\Bbb Z}}\left \| S\left ( t-s \right )h\left ( s,x^{n-1}\left(s\right),x_{s}^{n-1},y \right)\right \|_{X} ^{2}{\rm d}s\nu \left ({\rm d}y\right ) \Big )^{p/2}\\&\le &c_{p}M^{p}2^{\frac{p}{2}}L_{3}^{p}\Big( \int_{0}^{t}e^{-2\gamma\left(t-s\right)}E\left(\left\|x^{n-1}\left(s\right)\right\|_{X} ^{2}+\left|\left\|x_{s}^{n-1}-\varphi\right \|\right|_{X} ^{2}\right){\rm d}s\Big)^{\frac{p}{2}}\\&&+c_{p}M^{p}2^{\frac{p}{2}}\Big (\int_{0}^{t}e^{-2\gamma\left(t-s\right)}E\left \| h\left ( s,0,\varphi,y\right)\right\|_{X} ^{2}\nu \left ({\rm d}y\right ){\rm d}s\Big)^{\frac{p}{2}}.\end{matrix}$

$I_{3}=\int_{0}^{t}e^{-2\gamma\left(t-s\right)}E\left(\left\|x^{n-1}\left(s\right)\right\|_{X} ^{2}+\left|\left\|x_{s}^{n-1}-\varphi\right \|\right|_{X} ^{2}\right){\rm d}s$,用 Holder 不等式得

$\begin{matrix}I_{3}& \le& \Big ( \int_{0}^{t}\left ( e^{-\frac{2}{p}\cdot\gamma\left(t-s\right)}E\left(\left\|x^{n-1}\left(s\right)\right\|_{X} ^{2}+\left|\left\|x_{s}^{n-1}-\varphi\right \|\right|_{X} ^{2}\right)\right)^{\frac{p}{2}}{\rm d}s \Big )^{\frac{2}{p}}\\&&\cdot\Big ( \int_{0}^{t}\left ( e^{-\frac{2\left(p-1\right)}{p}\cdot\gamma\left(t-s\right)}\right)^{\frac{p}{p-2}}{\rm d}s\Big)^{\frac{p-2}{p}}\\&\le& \left ( \frac{p-2}{2\gamma\left(p-1\right)}\right)^{\frac{p-2}{p}}\cdot\bigg[\int_{0}^{t}e^{-\gamma\left(t-s\right)}\big(\upsilon^{1-p}E\left\|x^{n-1}\left(s\right)\right\|_{X} ^{p}\\&& +\left ( \upsilon \left(1-\upsilon\right)\right)^{1-p}E\left|\left\|x_{s}^{n}\right\|\right|_{X} ^{p}+\left ( 1-\upsilon\right)^{2-2p}E\left\|\varphi\right\|_{X} ^{p}\big){\rm d}s\bigg]^{\frac{2}{p}}.\end{matrix}$

将(3.9)代入到(3.8)式, 我们获得了引理3.7的结论. 证毕.

引理 3.3-3.7 的证明方法起源于Chen[19]的文章, 然而, 他们的技术并不完全适用于本文, 原因是我们的模型中有泛函项, 这将导致使用逐次逼近法收缩时, 系数有所变化.

定理 3.1 假设(A1)-(A4)成立, 则系统(2.1)有唯一的温和解.

我们可找到某个$\iota<1 $, $\frac{1}{2} <\delta < 1$, 且 $c_{p}=\left (p \left ( p-1 \right )/2 \right )^{p/2} $, 满足如下不等式

$\begin{matrix} & &9^{p-1}\left ( 1-\iota \right )^{-p}\bigg[ M_{p}^{1-\alpha }L^{p}\gamma ^{-p\alpha }\left ( \Gamma\left ( 1+p\left ( \alpha -1\right )/\left ( p-1\right )\right )\right )^{p-1} \\ && +c_{p}M^{p}L_{3}^{p}\gamma ^{-1}\left ( \frac{p-2}{2\left ( p-1\right )\gamma }\right )^{\left ( p-2\right )/2}\bigg]\\ &&+3^{p-1}\left ( 1-\iota \right )^{-p}\bigg[ M^{p}L_{1}^{p}\gamma ^{-p}+c_{p}M^{p}L_{2}^{p}\gamma ^{-1}\left ( \frac{2\gamma \left ( p-1\right )}{p-2}\right )^{1-p/2}\bigg]\\ &&+9^{p-1} \left ( 1-\iota\right )^{-p}M^{p}\bigg (\sum\limits_{0<t_{k}<t} n_{k}\bigg)^{p}<1.\end{matrix}$

为证系统(2.1)解的存在唯一性, 令 $x^{0}\left ( t \right )=S\left ( t \right )\left ( \phi \left ( 0 \right )+u\left ( 0,\phi,\varphi \right ) \right ),$$t\in \left [ 0,T \right ]$且对每个$n$, $x_{0}^{n}\left ( t \right )=\phi \left ( t \right ),t\in \left [ -\tau,0 \right ]$, 定义如下的逐次逼近序列

$\begin{matrix} x^{n}\left(t\right)&=&x^{0}\left ( t \right )-S\left ( t \right )u\left(t,x^{n}\left(t\right ),x_{t}^{n}\right)-\int_{0}^{t}AS\left(t-s\right)u\left(s,x^{n}\left(s\right),x_{s}^{n}\right){\rm d}s\\&&+\int_{0}^{t}S\left(t-s\right)f\left(s,x^{n-1}\left(s\right),x_{s}^{n-1}\right){\rm d}s +\int_{0}^{t}S\left(t-s\right)g\left(s,x^{n-1}\left(s\right),x_{s}^{n-1}\right){\rm d}\omega\left(s\right)\\&&+\int_{0}^{t}\int _{Z}S\left ( t-s \right )h\left ( s,x^{n-1}\left ( s \right ),x_{s}^{n-1},y\right)\widetilde{N}\left ( {\rm d}s, {\rm d}y \right )\\&&+\sum\limits_{0<t_{k}<t}S\left ( t-t_{k} \right ) I_{k}(x^{n-1}\left ( t_{k}^{-} \right)). \end{matrix}$

步骤 1 我们声称序列$\left \{ x^{n}\left ( t \right ),n\ge 0\right \}$ 对每个$t\in \left [ 0,T \right ] $是有界的.

首先, 根据引理2.3和3.1可得

$\begin{matrix} & &E\left\|x^{n}\left(t\right)\right\|_{X} ^{p}\\ &\le&\iota ^{1-p}E\left\|u\left(t,x^{n}\left(t\right),x_{t}^{n}\right)\right\|_{X} ^{p}+3^{p-1}\big ( 1-\iota \big )^{1-p}\big ( 1+\frac{1}{\varepsilon }\big)^{p-1}E\left \| S\left ( t \right ) \left ( \phi \left ( 0 \right )+u\left ( 0,\phi,\varphi \right ) \right ) \right \|_{X} ^{p}\\&&+9^{p-1}\left ( 1-\iota\right)^{1-p}\left ( 1+\varepsilon\right)^{p-1}E\bigg \| \int_{0}^{t}AS\left ( t-s \right )u\left ( s,x^{n}\left ( s \right ),x_{s}^{n}\right ){\rm d}s\bigg \|_{X} ^{p}\\&&+9^{p-1}\left ( 1-\iota\right)^{1-p}\left ( 1+\varepsilon\right)^{p-1}E\bigg \| \int_{0}^{t}\int _{Z}S\left ( t-s \right )h\left ( s,x^{n-1}\left ( s \right ),x_{s}^{n-1},y\right )\widetilde{N}\left ( {\rm d}s, {\rm d}y \right )\bigg\|_{X} ^{p}\\&&+9^{p-1}\left ( 1-\iota\right)^{1-p}\left ( 1+\varepsilon\right)^{p-1}E\bigg\| \sum\limits_{0<t_{k}<t}S\left ( t-t_{k}\right )I_{k}(x^{n-1}\left ( t_{k}^{-}\right ))\bigg \|_{X} ^{p}\\&&+3^{p-1}\left ( 1-\iota \right )^{1-p}E\bigg\| \int_{0}^{t}S\left ( t-s \right )f\left ( s,x^{n-1}\left ( s \right ),x_{s}^{n-1}\right ){\rm d}s\bigg\|_{X} ^{p}\\&&+3^{p-1}\left ( 1-\iota \right )^{1-p}E\bigg \| \int_{0}^{t}S\left ( t-s \right )g\left ( s,x^{n-1}\left ( s \right ),x_{s}^{n-1}\right ){\rm d}\omega \left ( s \right ) \bigg \|_{X} ^{p}.\end{matrix}$

进一步, 由假设(A4)可知

$\begin{eqnarray*} E\bigg\|\sum\limits_{0<t_{k}<t}S\left(t-t_{k}\right)I_{k}\left(x^{n-1}\left(t_{k}^{-}\right)\right) \bigg\|_{X} ^{p}&\le &E\bigg\| \sum\limits_{0<t_{k}<t}Me^{-\gamma\left(t-t_{k}\right)}\left(I_{k}\left(x^{n-1}\left(t_{k}^{-}\right)\right)-I_{k}\left(0\right)\right)\bigg\|_{X} ^{p} \\ &\le& M^{p}\bigg( \sum\limits_{t_{k}<t}q_{k}\bigg)^{p-1}\sum\limits_{t_{k}<t}q_{k}e^{-\gamma p\left ( t-t_{k}\right)}E\left \| x^{n-1}\left ( t_{k}^{-}\right)\right \|_{X} ^{p}.\end{eqnarray*}$

再利用假设(A1)和(A3), 可得

$\begin{eqnarray*} E\left \|S\left(t\right)\left(\phi\left(0\right )+u\left( 0,\phi,\varphi\right )\right )\right \|_{X} ^{p}&\le & M^{p} e^{-\gamma pt}\left ( 1+\iota \big \| \left ( -A \right )^{-\alpha }\big \|_{X} \right)^{p} \sup\limits_{s\in \left [ -\tau,0 \right ] }E\left \| \phi\left(s\right) \right \|_{X} ^{p}\\&&+M^{p} e^{-\gamma pt}\iota ^{p}\big \| \left ( -A \right )^{-\alpha }\big \|_{X} ^{p}\sup\limits_{s\in \left [ -\tau,0 \right ] }E\big \|\varphi\left(s\right) \big \|_{X} ^{p}. \end{eqnarray*}$

注意到

$\begin{eqnarray*}&&\sup\limits_{s\in \left [ -\tau,t \right ] }E\left \| x\left ( s \right ) \right \|_{X} ^{p}\le\sup\limits_{s\in \left [ -\tau,0\right ] }E\left \|\phi \left(s\right) \right \|_{X} ^{p}+\sup\limits_{s\in \left [ 0,t \right ] }E\left \| x\left ( s \right ) \right \|_{X} ^{p},\\&& E\left \| x_{t}\right \|_{X} ^{p} = E\Big \{ \sup\limits_{t-\tau \le s \le t}\left \| x\left ( s\right) \right \|_{X} ^{p}\Big\}\le\sup\limits_{s\in \left [ 0,t \right ] }E\left \| x\left(s\right)\right\|_{X} ^{p}+\sup\limits_{s\in \left [t-\tau,0 \right]}E\left\|\varphi\left(s\right) \right\|_{X} ^{p}.\end{eqnarray*}$

因此

$\begin{matrix} & &\sup\limits_{s\in \left [ 0,t \right ] }E\left \|x^{n}\left(s\right)\right\|_{X} ^{p}\\ &\le& \Big\{ 1-\big ( 1+\left ( 1-\upsilon\right )^{1-p}\big )\Big[ \iota \upsilon ^{1-p}+9^{p-1} \left ( 1-\iota\right )^{1-p}\left ( 1+\varepsilon \right )^{p-1}\upsilon ^{1-p}\\&&\cdot M_{1-\alpha }^{p}L^{p}\gamma ^{-\alpha p}\Big ( \Gamma \big ( 1+\frac{p\left ( \alpha -1 \right ) }{p-1}\big)\Big)^{p-1}\Big] \Big \}^{-1}\times\Big \{\Big[ \iota \upsilon ^{1-p}+9^{p-1}\left ( 1-\iota\right )^{1-p}\\&&\cdot\left (1+\varepsilon\right )^{p-1}\upsilon ^{2-2p}M_{1-\alpha }^{p}L^{p}\gamma ^{-\alpha p}\Big ( \Gamma \big( 1+\frac{p\left ( \alpha -1 \right ) }{p-1} \big) \Big )^{p-1}+3^{p-1}\left ( 1-\iota\right)^{1-p}\\&&\cdot\left ( 1+\frac{1}{\varepsilon}\right)^{p-1}M^{p}e^{-\gamma pt}\left ( 1+\iota \big \| \left ( -A \right )^{-\alpha } \big \|_{X} \right )^{p}\Big] \sup\limits_{s\in \left [\tau,0\right ] }E\left \| \phi\left(s\right) \right \| _{X} ^{p} \\ &&+\Big\{\big ( \left ( \upsilon \left ( 1-\upsilon\right )\right )^{1-p}+\left ( 1-\upsilon \right )^{2-2p}\big )\Big[\iota +9^{p-1}\left(1-\iota\right)^{1-p}\left ( 1+\varepsilon\right)^{p-1}\upsilon^{1-p}M_{1-\alpha}^{p}L^{p}\gamma ^{-\alpha p}\\&&\cdot \Big ( \Gamma \big ( 1+\frac{p\left(\alpha-1\right)}{p-1}\big)\Big)^{p-1}\Big]+3^{p-1}\left(1-\iota\right)^{1-p}\left ( 1+\frac{1}{\varepsilon}\right)^{p-1}M^{p}e^{-\gamma pt}\iota^{p}\big \| \left ( -A \right )^{-\alpha}\big\|_{X} ^{p}\Big\}\\&&\cdot\sup\limits_{s\in \left [\tau,0\right ] }E\left \|\varphi\left(s\right) \right \|_{X} ^{p}+\iota ^{1-p}\sup\limits_{s\in \left [\tau,t\right ] }E\left \| u\left ( s,0,\varphi\right)\right\|_{X} ^{p}\\&&+9^{p-1}\left ( 1-\iota\right )^{1-p}\left (1+\varepsilon\right )^{p-1}\upsilon ^{2-2p}M_{1-\alpha }^{p}L^{p}\gamma ^{-\alpha p}\Big ( \Gamma \big( 1+\frac{p\left ( \alpha -1 \right ) }{p-1} \big ) \Big )^{p-1}\\&&\cdot\sup\limits_{s\in\left[-\tau,t\right]}E\left \| \left ( -A \right )^{\alpha}u\left(s,0,\varphi\right)\right\|_{X} ^{p}+9^{p-1}\left(1-\iota\right)^{1-p}\left ( 1+\varepsilon\right)^{p-1}c_{p}M^{p}2^{\frac{p}{2}}L_{3}^{p}\\&&\cdot\Big ( \frac{p-2}{2\gamma\left(p-1\right)}\Big)^{\frac{p-2}{2}} \gamma ^{-1}E_{1}+9^{p-1}\left ( 1-\iota\right )^{1-p}\left (1+\varepsilon\right )^{p-1}c_{p}M^{p}\gamma ^{-\frac{p}{2}}\sup\limits_{s\in \left[-\tau,t\right] }E\left \| h\left ( s,0,\varphi\right)\right \|_{X} ^{p}\\&&+9^{p-1}\left ( 1-\iota\right )^{1-p} \left (1+\varepsilon\right )^{p-1}M^{p}\Big( \sum\limits_{t_{k}<t }q_{k}\Big)^{p-1} \sum\limits_{t_{k}<t }q_{k}e^{-\gamma p\left ( t-t_{k}\right)}\\&&\cdot\Big ( \sup\limits_{s\in \left[t\right]}E\left\|x^{n-1}\left(s\right)\right\|_{X} ^{p} +\sup\limits_{s\in \left[-\tau,0\right]}E\left\|\phi \left ( s \right ) \right\|_{X} ^{p} \Big)\\&&+3^{p-1}\left ( 1-\iota\right)^{1-p}M^{p}L_{1}^{p}\gamma ^{-p}E_{1}+3^{p-1}\left ( 1-\iota\right)^{1-p}c_{p}M^{p}\upsilon ^{2-2p}L_{2}^{p}\\&&\cdot\Big(\frac{p-2}{2\gamma\left(p-1\right)}\Big)^{\frac{2}{p}-1}\gamma ^{-1}\upsilon ^{1-p} E_{1}+ \left ( 1-\upsilon \right )^{1-p}\sup\limits_{s\in \left [ -\tau,t\right ] }E\left \|g\left ( s,0,\varphi\right)\right\|_{X} ^{p}\big]\Big \}.\end{matrix}$

其中

$\begin{matrix} E_{1}&=&\upsilon ^{1-p}\big ( \sup\limits_{s\in \left [t \right] }E\left \| x^{n-1}\left ( s \right )\right \|_{X} ^{p}+ \sup\limits_{s\in \left[-\tau,0\right]}E\left\|\phi \left(s\right)\right\|_{X} ^{p} \big) \\&& +\left(\upsilon\left(1-\upsilon\right)\right)^{1-p}\big ( \sup\limits_{s\in \left [t \right] }E\left \| x^{n-1}\left ( s \right )\right \|_{X} ^{p}+\sup\limits_{s\in \left[t-\tau,0\right]}E\left\|\varphi \left(s\right)\right\|_{X} ^{p} \big) \\&&+\left(1-\upsilon\right)^{2-2p}\sup\limits_{s\in \left[t-\tau,0\right]}E\left\|\varphi\left(s\right)\right\|_{X} ^{p}.\end{matrix}$

由(3.10)式, 可找到一个正数 $\varepsilon$ 足够小, 使得

$\begin{equation} N_{1}=\iota +9^{p-1}\left ( 1-\iota\right )^{1-p}\left ( 1+\varepsilon \right )^{p-1} M_{1-\alpha }^{p}L^{p}\gamma ^{-\alpha p}\big ( \Gamma \big ( 1+\frac{p\left ( \alpha -1 \right ) }{p-1} \big ) \big )^{p-1}<1.\end{equation}$

由于 $\upsilon \in \left ( 0,1 \right ) $, 因此

$\begin{eqnarray*}&&\big\{ 1-\left ( 1+\left ( 1-\upsilon\right )^{1-p}\right )\big[ \iota \upsilon ^{1-p}+9^{p-1}\left ( 1-\iota\right )^{1-p}\left ( 1+\varepsilon \right )^{p-1}\upsilon ^{1-p}M_{1-\alpha }^{p}L^{p}\gamma ^{-\alpha p}\\&&\cdot \big ( \Gamma \big ( 1+\frac{p\left ( \alpha -1 \right ) }{p-1}\big)\big)^{p-1}\big] \big \}^{-1}\end{eqnarray*}$

是一个有界小量. 类似地, 如下五个式子

$\begin{eqnarray*}&&9^{p-1}\left ( 1-\iota\right )^{1-p}\left ( 1+\varepsilon\right )^{p-1}c_{p}M^{p}L_{3}^{p}\Big(\frac{p-2}{2\gamma \left (p-1\right)}\Big)^{\frac{p-2}{2}}\gamma ^{-1}\upsilon ^{1-p};\\&&9^{p-1}\left ( 1-\iota\right )^{1-p}\left ( 1+\varepsilon\right )^{p-1}c_{p}M^{p}\gamma ^{-\frac{p}{2}};uad 9^{p-1}\left ( 1-\iota\right )^{1-p}\left ( 1+\varepsilon\right )^{p-1}M^{p}\Big ( \sum\limits_{t_{k}<t }q_{k}\Big)^{p};\\&&3^{p-1}\left ( 1-\iota\right )^{1-p}M^{p}L_{1}^{p}\gamma ^{-p};uad 3^{p-1}\left ( 1-\iota\right )^{1-p}c_{p} M^{p}L_{2}^{p}\Big ( \frac{p-2}{2\gamma \left (p-1\right)}\Big)^{\frac{p-2}{2}}\gamma ^{-1} \end{eqnarray*}$

都是有界的.对(3.13)式应用数学归纳法, 可证明对每个$t\in \left [ 0,T \right ] $, 序列$\left \{ x^{n}\left ( t \right ),n\ge 0\right \}$ 是有界的. 事实上, 由于 $E\left \| \phi \right \|_{X} ^{p}< \infty, E\left \| \varphi \right \|_{X} ^{p}< \infty$, 且函数 $ E\left \| \left ( -A \right )^{\alpha }u\left ( s,0,\varphi\right)\right \|_{X} ^{p}$, $E\left \| h\left ( s,0,\varphi\right )\right \|_{X} ^{p}$, $E\left \| f\left ( s,0,\varphi\right )\right \|_{X} ^{p}$, $E\left\| g\left ( s,0,\varphi\right )\right \|_{X} ^{p}$是一致有界的. 所以, 当$n=1$ 时, 有 $E\left \| x\left ( s \right ) \right \|_{X} ^{p}< \infty $; 当$n>1$ 时, 假设 $E\left \| x^{n-1} \left ( s \right ) \right \|_{X} ^{p}< \infty$ 是成立的, 由此, $E\left \| x^{n} \left ( s \right ) \right \|_{X} ^{p}< \infty$, 因而序列 $\left \{ x^{n}\left ( t \right ),n\ge 0\right \}$ 是有界的.

步骤 2 我们声称序列 $\left \{ x^{n}\left ( t \right ),n\ge 0\right \}$ 是一个 Cauchy 列.对于 $0\le t\le T$, 由 (3.11)式得

$\begin{eqnarray*} &&\sup\limits_{s\in \left [ 0,t \right ] }E\left \| x^{n+1}\left ( s \right )-x^{n}\left ( s \right )\right \|_{X} ^{p}\\ &\le &\iota \sup\limits_{s\in \left [ -\tau,t \right ] }E\left (\left \| x^{n+1}\left ( s \right )-x^{n}\left ( s \right )\right \|_{X} +\left|\left \| x_{s} ^{n+1}\left ( s \right )-x_{s} ^{n}\left ( s \right )\right \| \right|_{X} \right )^{p}\\&&+3^{p-1}\left ( 1-\iota \right ) ^{1-p} \left (1+\frac{1}{\varepsilon } \right )^{p-1}M_{1-\alpha }^{p}L^{p}\gamma ^{-\alpha p}\Big ( \Gamma \big( 1+\frac{p\left ( \alpha -1 \right ) }{p-1} \big) \Big )^{p-1} \\&&\cdot \sup\limits_{s\in \left [ -\tau,t \right ] }E\left ( \left \| x^{n+1}\left ( s \right )-x^{n}\left ( s \right )\right \|_{X} +\left|\left \| x_{s} ^{n+1}\left ( s \right )-x_{s} ^{n}\left ( s \right )\right \| \right|_{X} \right )^{p}\\&&+6^{p-1}\left ( 1-\iota\right)^{1-p}\left ( 1+\varepsilon \right ) ^{p-1} c_{p}M^{p}L_{3}^{p} \Big(\frac{p-2}{2\gamma\left(p-1\right)}\Big)^{\frac{2}{p}-1}\gamma ^{-1} \\ &&\cdot \sup\limits_{s\in \left [ -\tau,t \right ] }E\left ( \left \| x^{n}\left ( s \right )-x^{n-1}\left ( s \right )\right \|_{X} +\left|\left \| x_{s} ^{n}\left ( s \right )-x_{s} ^{n-1}\left ( s \right )\right \|\right|_{X} \right )^{p} \\&&+6^{p-1}\left ( 1-\iota\right)^{1-p}\left ( 1+\varepsilon \right ) ^{p-1}\Big(\sum\limits_{t_{k}<t }q_{k}\Big)^{p}\sup\limits_{s\in \left [ -\tau,t \right ] }E\left|\left \| x_{s} ^{n}\left ( s \right )-x_{s} ^{n-1}\left ( s \right )\right \|\right|_{X} ^{p} \\&&+3^{p-1}\left ( 1-\iota \right ) ^{1-p}M^{p}L_{1}^{p}\gamma ^{-p}\sup\limits_{s\in \left [ -\tau,t \right ] }E\left ( \left \| x^{n}\left ( s \right )-x^{n-1}\left ( s \right )\right \|_{X} +\left|\left \| x_{s} ^{n}\left ( s \right )-x_{s} ^{n-1}\left ( s \right )\right \|\right|_{X} \right )^{p}\\&&+ 3^{p-1}\left ( 1-\iota \right ) ^{1-p}c_{p} M^{p}L_{2}^{p}\gamma ^{-1}\Big ( \frac{p-2}{2\gamma\left(p-1\right)}\Big)^{\frac{p-2}{2}} \\&&\cdot \sup\limits_{s\in \left [ -\tau,t \right ] }E\big ( \left \| x^{n}\left ( s \right )-x^{n-1}\left ( s \right )\right \|_{X} +\left|\left \| x_{s} ^{n}\left ( s \right )-x_{s} ^{n-1}\left ( s \right )\right \|\right|_{X} \big )^{p},\end{eqnarray*}$

这意味着

$\begin{eqnarray*} &&\sup\limits_{s\in \left [ 0,t \right ] }E\left \| x^{n+1}\left ( s \right )-x^{n}\left ( s \right )\right \|_{X} ^{p}\\ &\le& \bigg \{1-\big ( \upsilon ^{1-p}+\left ( 1-\upsilon \right )^{1-p}\big ) \bigg [ \iota +3^{p-1}\left ( 1-\iota \right ) ^{1-p} \left (1+\frac{1}{\varepsilon } \right )^{p-1}M_{1-\alpha }^{p}L^{p}\gamma ^{-\alpha p}\\&&\cdot\Big ( \Gamma \big( 1+\frac{p\left ( \alpha -1 \right ) }{p-1} \big ) \Big)^{p-1} \bigg ]\bigg \}^{-1} \times \bigg\{6^{p-1}\left ( 1-\iota\right)^{1-p}\left ( 1+\varepsilon \right ) ^{p-1}\bigg [ c_{p}M^{p}L_{3}^{p} \Big(\frac{p-2}{2\gamma\left(p-1\right)}\Big)^{\frac{2}{p}-1}\\&&\cdot\gamma ^{-1}\left ( \upsilon ^{1-p}+\left ( 1-\upsilon \right )^{1-p}\right )+\Big(\sum\limits_{t_{k}<t }q_{k}\Big)^{p}M^{p} \bigg] +3^{p-1}\left ( 1-\iota \right ) ^{1-p}M^{p}\big ( \upsilon ^{1-p}+\left ( 1-\upsilon \right )^{1-p}\big ) \\&&\cdot\bigg [ L_{1}^{p}\gamma ^{-p}+c_{p}L_{2}^{p}\gamma ^{-1}\Big ( \frac{p-2}{2\gamma\left(p-1\right)}\Big)^{\frac{p-2}{2}} \bigg] \bigg\} \sup\limits_{s\in \left [ 0,t \right ] }E\left \| x^{n}\left ( s \right )-x ^{n-1} \left ( s \right )\right \|_{X} ^{p} \\ &=&\frac{\beta\sup\limits_{s\in \left [ 0,t \right ] }E\left \| x^{n}\left ( s \right )-x ^{n-1}\left ( s \right )\right \|_{X} ^{p} }{1-W}\\&&\vdots \\ &\le &\frac{\beta^{n} \sup\limits_{s\in \left [ 0,t \right ] }E\left \| x^{1} \left ( s \right )-x ^{0}\left ( s \right )\right \|_{X} ^{p} }{\left ( 1-W \right )^{n} },\end{eqnarray*}$

其中

$\begin{eqnarray*} \beta& =&6^{p-1}\left ( 1-\iota\right)^{1-p}\left ( 1+\varepsilon \right ) ^{p-1} \\&&\cdot\bigg [ c_{p}M^{p}L_{3}^{p} \Big(\frac{p-2}{2\gamma\left(p-1\right)}\Big)^{\frac{2}{p}-1} \gamma ^{-1}\big( \upsilon ^{1-p}+\left ( 1-\upsilon \right )^{1-p}\big ) +\Big(\sum\limits_{t_{k}<t }q_{k}\Big)^{p}M^{p} \bigg] \\ && +3^{p-1}\left ( 1-\iota \right ) ^{1-p}M^{p}\big( \upsilon ^{1-p}+\left ( 1-\upsilon \right )^{1-p}\big ) \bigg [ L_{1}^{p}\gamma ^{-p}+c_{p}L_{2}^{p}\gamma ^{-1}\Big ( \frac{p-2}{2\gamma\left(p-1\right)}\Big)^{\frac{p-2}{2}} \bigg ],\\W&=&\big ( \upsilon ^{1-p}+\left ( 1-\upsilon \right )^{1-p}\big )\\&&\cdot \bigg [ \iota +3^{p-1}\left ( 1-\iota \right ) ^{1-p} \left (1+\frac{1}{\varepsilon } \right )^{p-1}M_{1-\alpha }^{p}L^{p}\gamma ^{-\alpha p}\Big ( \Gamma \big( 1+\frac{p\left ( \alpha -1 \right ) }{p-1} \big ) \Big )^{p-1} \bigg ].\end{eqnarray*}$

类似地, 由(3.11)式, 令

$F=\sup\limits_{s\in \left [ 0,t \right ] }E\left \| x^{1}\left ( s \right )-x^{0}\left ( s \right )\right \|_{X} ^{p}+\sup\limits_{s\in \left [ -\tau,0 \right ] }E\left \| \phi \left ( s \right ) \right \| _{X} ^{p},$

$G=\sup\limits_{s\in \left [ 0,t \right ] }E\left \| x^{1}\left ( s \right )-x^{0}\left ( s \right )\right \|_{X} ^{p}+\sup\limits_{s\in \left[-\tau,0\right]}E\left \| \varphi \left ( s \right ) \right \|_{X} ^{p},$

$\begin{eqnarray*} &&\sup\limits_{s\in \left [ 0,t \right ] }E\left \| x^{1}\left ( s \right )-x^{0}\left ( s \right )\right \|_{X} ^{p} \\ &\le& \iota\big ( \upsilon ^{1-p}+ \left ( 1-\upsilon \right )^{1-p}\big)\cdot F \\&&+3^{p-1}\left ( 1-\iota \right ) ^{1-p} \left (1+\frac{1}{\varepsilon } \right )^{p-1}M_{1-\alpha }^{p}L^{p}\gamma ^{-\alpha p}\Big ( \Gamma \big ( 1+\frac{p\left ( \alpha -1 \right ) }{p-1} \big ) \Big )^{p-1} \upsilon ^{1-p}\cdot F\\&&+3^{p-1}\left ( 1-\iota \right ) ^{1-p} \left (1+\frac{1}{\varepsilon } \right )^{p-1}M_{1-\alpha }^{p}L^{p}\gamma ^{-\alpha p}\Big( \Gamma \big ( 1+\frac{p\left ( \alpha -1 \right ) }{p-1} \big) \Big )^{p-1} \left ( 1-\upsilon \right ) ^{1-p}\cdot G\\&&+6^{p-1}\left ( 1-\iota\right)^{1-p}\left ( 1+\varepsilon \right ) ^{p-1} c_{p}M^{p}L_{3}^{p} \left(\frac{p-2}{2\gamma\left(p-1\right)}\right)^{\frac{2}{p}-1}\gamma ^{-1}\\&&\cdot\bigg [ \upsilon ^{1-p}\Big ( \sup\limits_{s\in \left [ 0,t \right ] }E\left \|x^{0}\left ( s \right )\right \|_{X} ^{p} +\sup\limits_{s\in \left [ -\tau,0 \right ] }E\left \|\phi \left ( s \right ) \right \|_{X} ^{p} \Big )+\left ( 1-\upsilon \right )^{1-p} \Big ( \sup\limits_{s\in \left [ 0,t \right ] }E\left \|x^{0}\left ( s \right )\right \|_{X} ^{p}\\&&+\sup\limits_{s\in \left [ t-\tau,0 \right ] }E\left \| \varphi \left ( s \right ) \right \| _{X} ^{p} \Big ) \bigg] +\bigg[ 6^{p-1}\left ( 1-\iota\right)^{1-p}\left ( 1+\varepsilon \right ) ^{p-1}\Big(\sum\limits_{t_{k}<t }q_{k}\Big)^{p}\\&&+3^{p-1}\left ( 1-\iota \right ) ^{1-p}M^{p}L_{1}^{p}\gamma ^{-p}\upsilon ^{1-p}+ 3^{p-1}\left ( 1-\iota \right ) ^{1-p}c_{p} M^{p}L_{2}^{p}\gamma ^{-1}\Big ( \frac{p-2}{2\gamma\left(p-1\right)}\Big)^{\frac{p-2}{2}}\upsilon ^{1-p} \bigg]\\&&\cdot \Big ( \sup\limits_{s\in \left [ 0,t \right ] }E\left \|x^{0}\left ( s \right )\right \|_{X} ^{p}+\sup\limits_{s\in \left [ -\tau,0 \right ] }E\left \| \phi \left ( s \right ) \right \| _{X} ^{p} \Big )+\bigg [3^{p-1}\left ( 1-\iota \right ) ^{1-p}M^{p}L_{1}^{p}\gamma ^{-p}\left ( 1-\upsilon \right ) ^{1-p}\\&&+ 3^{p-1}\left ( 1-\iota \right ) ^{1-p}c_{p} M^{p}L_{2}^{p}\gamma ^{-1}\Big ( \frac{p-2}{2\gamma\left(p-1\right)}\Big)^{\frac{p-2}{2}}\left ( 1-\upsilon \right ) ^{1-p} \bigg] \\&&\cdot \Big( \sup\limits_{s\in \left [ 0,t \right ] }E\left \|x^{0}\left ( s \right )\right \|_{X} ^{p}+\sup\limits_{s\in \left [ t-\tau,0 \right ] }E\left \| \varphi \left ( s \right ) \right \|_{X} ^{p} \Big ).\end{eqnarray*}$

因此

$\begin{eqnarray*} &&\Big ( \sup\limits_{s\in \left [ 0,t \right ] }E\left \|x^{0}\left ( s \right )\right \|_{X} ^{p} +\sup\limits_{s\in \left [ t-\tau,0 \right ] }E\left \| \varphi \left ( s \right ) \right \|_{X} ^{p} \Big )\\ &\le&\left ( 1-W \right ) \times \bigg \{\big (\upsilon ^{1-p}+\left ( 1-\upsilon \right )^{1-p}\big ) \bigg [\iota +3^{p-1}\left ( 1-\iota \right ) ^{1-p} \left (1+\frac{1}{\varepsilon } \right )^{p-1}M_{1-\alpha }^{p}L^{p}\gamma ^{-\alpha p}\\&&\cdot\Big ( \Gamma \big( 1+\frac{p\left ( \alpha -1 \right ) }{p-1} \big ) \Big )^{p-1}+6^{p-1}\left ( 1-\iota\right)^{1-p}\left ( 1+\varepsilon \right ) ^{p-1} c_{p}M^{p}L_{3}^{p} \Big(\frac{p-2}{2\gamma\left(p-1\right)}\Big)^{\frac{2}{p}-1}\gamma ^{-1}\\ & &+3^{p-1}\left ( 1-\iota \right ) ^{1-p}M^{p}L_{1}^{p}\gamma ^{-p}+6^{p-1}\left ( 1-\iota\right)^{1-p}\left ( 1+\varepsilon \right ) ^{p-1}\Big(\sum\limits_{t_{k}<t }q_{k}\Big)^{p}\\ &&+ 3^{p-1}\left ( 1-\iota \right ) ^{1-p}c_{p} M^{p}L_{2}^{p}\gamma ^{-1}\Big ( \frac{p-2}{2\gamma\left(p-1\right)}\Big)^{\frac{p-2}{2}} \bigg] \Big ( \sup\limits_{s\in \left [ -\tau,0 \right ] }E\left \| \phi \left ( s \right ) \right \| _{X} ^{p}+ \sup\limits_{s\in \left[-\tau,0\right]}E\left \| \varphi \left ( s \right ) \right \| _{X} ^{p}\Big )\\ &&+\bigg [\big (\upsilon ^{1-p}+ \left ( 1-\upsilon \right ) ^{1-p} \big ) \Big ( 6^{p-1}\left ( 1-\iota\right)^{1-p}\left ( 1+\varepsilon \right ) ^{p-1} c_{p}M^{p}L_{3}^{p} \Big(\frac{p-2}{2\gamma\left(p-1\right)}\Big)^{\frac{2}{p}-1}\gamma ^{-1}3^{p-1}\\ &&\left ( 1-\iota \right ) ^{1-p}M^{p}L_{1}^{p}\gamma ^{-p}\left ( 1-\upsilon \right ) ^{1-p}\\ && + 3^{p-1}\left ( 1-\iota \right ) ^{1-p}c_{p} M^{p}L_{2}^{p}\gamma ^{-1}\Big ( \frac{p-2}{2\gamma\left(p-1\right)}\Big)^{\frac{p-2}{2}} \Big ) \bigg] \sup\limits_{s\in \left [ 0,t \right ] } E\left \|x^{0}\left ( s \right )\right \|_{X} ^{p} \bigg\},\end{eqnarray*}$

又由于 $ \sup\limits_{s\in \left [ 0,t \right ] }E\left \|x^{0}\left ( s \right )\right \|_{X} ^{p} \le M^{p}E\left \| \phi \left ( 0 \right ) \right \|^{p}$, 因此, $\sup\limits_{s\in \left [ 0,t \right ] }E\left \| x^{1}\left ( s \right )-x^{0}\left ( s \right )\right \|_{X} ^{p} $是有界的, 且对于$\forall\; m> n\ge 1$,有

$\begin{eqnarray*} \sup\limits_{s\in \left [ 0,t \right ] }E\left \| x^{m}\left ( s \right )-x^{n}\left ( s \right )\right \|_{X} ^{p}&\le &\sum\limits_{k=n}^{m-1 } \sup\limits_{s\in \left [ 0,t \right ] }E\left \| x^{k+1}\left ( s \right )-x^{k} \left ( s \right ) \right \| _{X} ^{p}\\ &\le &\sum\limits_{k=n}^{m-1 }\frac{\beta ^{k} }{\left ( 1-W \right )^{k} } E\left \| x^{1}\left ( s \right )-x^{0}\left ( s \right ) \right \|_{X} ^{p}.\end{eqnarray*}$

由于$\frac{\beta }{1-W} < 1$, 则$\sum\limits_{k=1}^{+\infty } \big ( \frac{\beta }{1-W} \big )^{k} <\infty.$ 进而

$\lim\limits_{n,m \to \infty}\sum\limits_{k=n}^{m-1 }\frac{\epsilon ^{k} }{\left ( 1-N\right )^{k} } E\left \| x^{1}\left ( s \right )-x^{0}\left ( s \right ) \right \| _{X} ^{p}= 0.$

因此, 对$\forall\; N>0$, $\forall\; \varepsilon >0$, 当$n,m>N$时, 有$\sup\limits_{s\in \left [ 0,t \right ] }E\left \| x^{m}\left ( s \right )-x^{n}\left ( s \right )\right \|_{X} ^{p}<\varepsilon $,这表明 $\left \{ x^{n}\left ( t \right ),n\ge 0 \right \} $ 是一个Cauchy 列.

步骤 3 我们将证明系统(2.1)解的存在唯一性. 由步骤2知, 存在一个解$x\left ( t \right ) \in X$使得

$\lim\limits_{n \to \infty}E\sup\limits_{s\in \left [ 0,t \right ]}\left \| x^{n}\left ( s \right )-x\left ( s \right ) \right \|_{X} ^{p}=0.$

Borel-Cantelli 引理表明, $x^{n}\left ( t \right ) $$ 0\le t\le T$上几乎处处一致收敛于$ x\left ( t \right ) $. 因此, 对(3.11)式左右两边同时取极限, 得

$\begin{eqnarray*} \lim\limits_{n \to \infty} x^{n}\left(t\right)&=&S\left(t\right)\left(\phi\left(0\right)+u\left(0,\phi,\varphi\right)\right)-\lim\limits_{n \to \infty} u\left(t,x^{n}\left(t\right ),x_{t}^{n}\right)\\ & &-\lim\limits_{n \to \infty} \int_{0}^{t}AS\left(t-s\right)u\left(s,x^{n}\left(s\right),x_{s}^{n}\right){\rm d}s\\ && +\lim\limits_{n \to \infty} \int_{0}^{t}S\left(t-s\right)f\left(s,x^{n-1}\left(s\right),x_{s}^{n-1}\right){\rm d}s\\ & &+\lim\limits_{n \to \infty}\int_{0}^{t}S\left(t-s\right)g\left(s,x^{n-1}\left(s\right),x_{s}^{n-1}\right){\rm d}\omega \left(s\right)\\ &&+\lim\limits_{n \to \infty} \sum\limits_{0<t_{k}<t}S\left ( t-t_{k} \right )I_{k}(x^{n-1}\left ( t_{k}^{-} \right))\\ &&+\lim\limits_{n \to \infty} \int_{0}^{t}\int _{Z}S\left ( t-s \right )h\left ( s,x^{n-1}\left ( s \right ),x_{s}^{n-1},y\right)\widetilde{N}\left ( {\rm d}s, {\rm d}y \right ).\end{eqnarray*}$

由此,$x\left(t\right)$是系统(2.1)的解.

下面证明解的唯一性. 假设 $x\left ( t \right ) $$v\left ( t \right )$分别都是系统(2.1)的解, 则

$\begin{eqnarray*} x^{n}\left(t\right)-v^{n}\left ( t \right ) &=&u\left(t,v^{n}\left(t\right ),v_{t}^{n}\right)-u\left(t,x^{n}\left(t\right ),v_{t}^{n}\right)\\&&+ \int_{0}^{t}AS\left(t-s\right)\left [u\left(s,v^{n}\left(s\right),v_{s}^{n}\right)-u\left(s,x^{n}\left(s\right),x_{s}^{n}\right) \right ]{\rm d}s\\&&+\int_{0}^{t}S\left(t-s\right)\left [f\left(s,x^{n-1}\left(s\right),x_{s}^{n-1}\right)-f\left(s,v^{n-1}\left(s\right),v_{s}^{n-1}\right) \right ]{\rm d}s\\&&+\int_{0}^{t}S\left(t-s\right)\left [ g\left(s,x^{n-1}\left(s\right),x_{s}^{n-1}\right)-g\left(s,v^{n-1}\left(s\right),v_{s}^{n-1}\right) \right ] {\rm d}\omega\left(s\right)\\&&+\int_{0}^{t}\int _{Z}S\big ( t\!-\!s \big )\left [h\left ( s,x^{n-1}\left ( s \right ),x_{s}^{n-1},y\right)\!-\!h\left ( s,v^{n-1}\left ( s \right ),v_{s}^{n-1},y\right) \right ] \widetilde{N}\left ( {\rm d}s, {\rm d}y \right )\\&&+\sum\limits_{0<t_{k}<t}S\left ( t-t_{k} \right ) I_{k}\left [x^{n-1}\left ( t_{k}^{-} \right)-v^{n-1}\left ( t_{k}^{-} \right) \right ].\end{eqnarray*}$

由步骤2, 可得

$\sup\limits_{s\in \left [ 0,t \right ] }E\left \| x^{n}\left ( s \right )-v^{n}\left ( s \right ) \right \|_{X} ^{p} \le \frac{\beta ^{n} }{\left ( 1-W \right )^{n}}\sup\limits_{s\in \left [ 0,t \right ] }E\left \| x^{0}\left ( s \right )-v^{0}\left ( s \right ) \right \|_{X} ^{p}, $

由于 $\sup\limits_{s\in \left [ 0,t \right ] }E\left \| x^{0}\left ( s \right )-v^{0}\left ( s \right ) \right \|_{X} ^{p}< \infty $$\frac{\beta }{1-W}<1 $, 因此, 可推出

$\lim\limits_{n\to \infty} \sup\limits_{s\in \left [ 0,t \right ] }E\left \| x^{n}\left ( s \right )-v^{n}\left ( s \right ) \right \|_{X} ^{p}=0, $

再由 Borel-Cantelli 引理知, $\lim\limits_{n\to \infty} x^{n} \left ( t \right ) =x\left ( t \right ),\; a.s. $$\lim\limits_{n\to \infty} v^{n} \left ( t \right ) =v\left ( t \right ),\;a.s. $ 因此, $x\left(t\right)=v\left (t\right),\;a.s.$, 这意味着系统 (2.1)的解是唯一的.证毕.

4 稳定性

在这一节中, 我们利用Banach不动点方法研究了脉冲中立型随机泛函微分方程(2.1)温和解的p阶矩的指数稳定性.

定义 4.1$p$是一个大于等于2的整数, 若对任意初始值$\phi $, 存在两个正常数$\xi$$C$,

$E\left \| x\left ( t \right ) \right \|_{X}^{p}\le M e^{-\lambda t},uad t\ge 0$

成立, 则系统(2.1)的温和解$x\left ( t \right ) $$p$时刻是指数稳定的.

下面我们给出如下两个条件

(A5) 对任意的$x,y\in C\left ( \left [ -\tau,T \right ];X\right ),t\ge 0 $, 映射$f\left ( t,\cdot,\cdot \right ),g\left ( t,\cdot,\cdot \right )$$h\left ( t,\cdot,\cdot, \cdot \right )$ 满足

$\begin{eqnarray*} &&\int_{0}^{t}e^{\gamma s}\left \| f\left ( t,x\left ( s \right ),x_{s}\right )-f\left ( t,y\left ( s \right ),y_{s} \right ) \right \|^{p}_{X}{\rm d}s\le C_{f}^{p}\int_{-\tau }^{t}e^{\gamma s} \left \| x\left ( s \right )-y\left ( s \right ) \right \|^{p} _{X}{\rm d}s,\\ &&\int_{0}^{t}e^{\gamma s} \left \| g\left ( t,x\left ( s \right ),x_{s}\right )-g\left ( t,y\left ( s \right ),y_{s} \right ) \right \|^{p}_{X}{\rm d}s\le C_{g}^{p}\int_{-\tau }^{t}e^{\gamma s} \left \| x\left ( s \right )-y\left ( s \right ) \right \|^{p} _{X}{\rm d}s,\\& &\int_{0}^{t}e^{\gamma s} \left \| h\left ( t,x\left ( s \right ),x_{s},z\right )-h\left ( t,y\left ( s \right ),y_{s},z \right ) \right \|^{2}_{X}\nu \left ({\rm d}z \right )\le C_{h}\int_{-\tau }^{t}e^{\gamma s} \left \| x\left ( s \right )-y\left ( s \right ) \right \|^{2} _{X}{\rm d}s,\\ & &\int_{0}^{t }e^{\gamma s}\left \| f\left ( s,0,\varphi \right ) \right \|^{p} _{X}{\rm d}s< \infty,\int_{0}^{t }e^{\gamma s}\left \| g\left ( s,0,\varphi \right ) \right \|^{p} _{X}{\rm d}s< \infty,\\ & &\int_{0}^{t}e^{-2\gamma\left(t-s\right)}E\left \| h\left ( s,0,\varphi,y\right)\right\|^{2}\nu \left ({\rm d}y\right ){\rm d}s< \infty, \end{eqnarray*}$

这里$C_{f},C_{g},C_{h}$ 都是某一个固定常数.

(A6) 映射$g:R_{+}\times X\times C\left ( \left [ -\tau,0 \right ];X \right )\to {\mathfrak L}_{2}^{0}\left ( X,Y \right )$ 满足

$ \begin{eqnarray*} \int_{0}^{\infty } e^{\gamma s} \left \| g\left ( s \right ) \right \|^{2}_{{\mathfrak L}_{2}^{0}\left ( X,Y \right )}{\rm d}s< \infty. \end{eqnarray*} $

定理 4.1 假设(A1), {(A3)}, {(A5)}和{(A6)}成立, 若满足

$ \begin{matrix} C^{'} :&=&2\cdot 12^{p-1}\bigg[\big \| \left ( -A \right )^{-\alpha } \big \|_{X}^{p}L^{p}+M_{1-\alpha }^{p}L^{p}\gamma ^{-\alpha p}\bigg( \Gamma \big ( 1+\frac{p\left ( \alpha -1 \right ) }{p-1} \big ) \bigg)^{p-1}+M^{p}L_{1}^{p}\gamma ^{-p}\\ &&+c_{p}M^{p}\bigg ( \frac{p-2}{2\gamma \left ( p-1 \right ) } \bigg)^{1-\frac{p}{2}}\gamma ^{-1}L_{2}^{p} +c_{p}M^{p}2^{\frac{p}{2}}\left(\frac{p-2}{2\gamma\left(p-1\right)} \right)^{\frac{p-2}{2}}L_{3}^{p}\gamma ^{-1} \\ &&+\frac{1}{2} M^{p}\Big ( \sum\limits_{t_{k}<t}q_{k}\Big)^{p} e^{- \gamma p \left ( t-t_{k} \right ) } \bigg ]<1, \end{matrix}$

则方程 (2.1)在$p$时刻是指数稳定的. 这里, 记号$\alpha,\gamma,L,L_{3},p,c_{p}$ 在第二节已分别给出.

$S$是所有${\cal F} $ -适应过程$\phi \left ( t,\omega \right ) :\left [-\tau,\infty \right )\times \Omega \to {\Bbb R} $组成的集合, 易证该集合是一个Banach空间, 对固定的$\omega \in \Omega $, 在$t$上是几乎处处连续的. 此外, 对$s\in \left [ -\tau,0 \right ] $$\phi \left ( s,\omega \right ) =\varphi \left ( s \right ) $, 当$0< \beta < \gamma $时, $\lim\limits_{n\to \infty}e^{\beta t}E\left \| \phi \left ( t,\omega \right ) \right \|_{X}^{p}= 0$. 对于$t\in \left [ -\tau,0 \right ] $定义$\left ( \pi x \right )\left ( t \right )=\psi \left ( t \right )$, $t\ge 0$时定义

$\begin{matrix} \left ( \pi x \right ) \left ( t \right )&=&S\left ( t \right )\left ( \phi \left ( 0 \right )+u\left ( 0,\phi,\varphi \right ) \right )-u\left ( t,x\left ( t \right ),x_{t} \right )-\int_{0}^{t}AS\left ( t-s \right )u\left ( s,x\left ( s \right ),x_{s}\right){\rm d}s\\ &&+\int_{0}^{t}S\left ( t-s \right )f\left ( s,x\left ( s \right ),x_{s} \right ){\rm d}s+\int_{0}^{t}S\left ( t-s \right )g\left ( s,x\left ( s \right ),x_{s} \right ){\rm d}\omega \left ( s \right ) \\ &&+\int_{0}^{t} \int _{{\Bbb Z} }S\left ( t-s \right )h\left ( s,x\left ( s \right ),x_{s},y\right )\widetilde{N}\left ( {\rm d}s, {\rm d}y \right ) +\sum\limits_{0< t_{k}< t }S\left ( t-t_{_{k} } \right )I_{k}\left (x\left ( t_{k}^{-} \right ) \right ) \\ &:=& I_{1}\left ( t \right )- I_{2}\left ( t \right )-I_{3}\left ( t \right )+I_{4}\left ( t \right )+I_{5}\left ( t \right )+I_{6}\left ( t \right )+I_{7}\left ( t \right ).\end{matrix}$

首先, 验证$\pi $$\left [0,\infty \right )$上的$p$阶矩的连续性, 令$x\in S,\; t_{1} > 0 $$\tau > 0$足够小, 有

$\begin{eqnarray*} E\left \| \left ( \pi x \right )\left ( t_{1}+\tau \right )-\left ( \pi x \right )\left ( t_{1} \right ) \right \|_{X}^{p}\le 7^{p-1}\sum\limits_{i=1}^{7}E\left \| I_{i}\left ( t_{_{1} }+\tau \right )-I_{i}\left ( t_{1} \right )\right \|_{X}^{p}.\end{eqnarray*}$

$i=1,4$$\tau \to 0$时, 显然有

$\begin{eqnarray*} E\left \| I_{i}\left ( t_{1} +\tau \right ) - I_{i}\left ( t_{1} \right ) \right \| _{X}^{p} \to 0.\end{eqnarray*}$

由假设(A3)知, 算子$\left ( -A \right )^{-\alpha } $是有界的, 且映射$\left ( -A \right ) ^{\alpha } u$$p$时刻是连续的, 因此

$\begin{eqnarray*} \lim\limits_{\tau \to 0} E\left \| I_{2}\left ( t_{1} +\tau \right ) - I_{2}\left ( t_{1} \right ) \right \| _{X}^{p}=0.\end{eqnarray*}$
$\begin{eqnarray*} & &E\left \| I_{3}\left ( t_{1}+\tau \right )- I_{3}\left ( t_{1} \right ) \right \| _{X}^{p}\\ &=&E\Big \|\int_{0}^{t_{1} }\left ( S\left ( \tau \right )-I \right )\left ( -A \right )^{1-\alpha}S\left ( t_{1}-s \right )\left ( -A \right)^{\alpha}u\left ( s,x\left ( s \right ),x_{s} \right ){\rm d}s\\&&+\int_{t_{1} }^{t_{1}+\tau }\left ( -A \right )^{1-\alpha } S\left ( t_{1}+\tau -s \right )\left ( -A \right )^{\alpha } u\left ( s,x\left ( s \right ),x_{s} \right ){\rm d}s \Big\| _{X}^{p}\\ &\le& 2^{p-1}E\Big \| \int_{0}^{t_{1} }\left ( S\left ( \tau \right )-I \right )\left ( -A \right )^{1-\alpha} S\left ( t_{1}-s \right )\left ( -A \right)^{\alpha}u\left ( s,x\left ( s \right ),x_{s} \right ){\rm d}s \Big \|_{X}^{p}\\&&+2^{p-1}E\Big\| \int_{t_{1} }^{t_{1}+\tau }\left ( -A \right )^{1-\alpha } S\left ( t_{1}+\tau -s \right )\left ( -A \right )^{\alpha } u\left ( s,x\left ( s \right ),x_{s} \right ){\rm d}s \Big \| _{X}^{p}\\ &: =&I_{31}\left ( \tau \right )+I_{32}\left ( \tau \right ).\end{eqnarray*}$

$S\left ( t \right )$的强连续性知, 对$\forall \;s\in \left [ 0,t_{1} \right ]$, 有

$\begin{equation} \lim\limits_{\tau \to 0}\left ( S\left ( \tau \right )-I \right )\left ( -A \right )^{1-\alpha}S\left ( t_{1}-s \right )\left ( -A \right)^{\alpha}u\left ( s,x\left ( s \right ),x_{s} \right )=0.\end{equation}$

再由引理3.3和假设(A1), 对$\forall \;\alpha\in \left (0,1 \right ] $, 有

$\begin{eqnarray*}& &\left \| \left ( S\left ( \tau \right )-I \right )\left ( -A \right )^{1-\alpha}S\left ( t_{1}-s \right )\left ( -A \right)^{\alpha}u\left ( s,x\left ( s \right ),x_{s} \right ) \right \|_{X}\\ &\le& \frac{2MM_{1-\alpha }}{\left ( t_{1}-s \right )^{1-\alpha } } \left \| \left ( -A \right )^{\alpha }u\left ( s,x\left ( s \right ),x_{s} \right ) \right \| _{X}.\end{eqnarray*}$

由Lebesgue控制收敛定理可知$\lim\limits_{\tau \to 0} I_{31}\left ( \tau \right )=0$. 同理可证,

$\begin{eqnarray*} &&\left \| \left ( -A \right )^{1-\alpha } S\left ( t_{1}+\tau -s \right )\left ( -A \right )^{\alpha } u\left ( s,x\left ( s \right ),x_{s} \right ) \right \| _{X} \\&\le& \frac{M_{1-\alpha }}{\left ( t_{1}+\tau -s \right )^{1-\alpha } } \left \| \left ( -A \right )^{\alpha }u\left ( s,x\left ( s \right ),x_{s} \right ) \right \| _{X}.\end{eqnarray*}$

因此, $\lim\limits_{\tau \to 0} I_{32}\left ( \tau \right )=0$,故$\lim\limits_{\tau \to 0}E\left \|I_{3}\left (t_{1}+ \tau \right )- I_{3}\left (t_{1} \right ) \right \|_{X}^{p} =0 $.

$\begin{eqnarray*} & &E\left \| I_{5}\left ( t_{1}+\tau\right)-I_{5}\left ( t_{1} \right )\right \|_{X}^{p}\\ &=&E\bigg \| \int_{0}^{t_{1}}\left ( S\left (t_{1}+\tau-s \right )-S\left (t_{1}-s \right ) \right )g\left ( s,x\left ( s \right ),x_{s} \right ){\rm d}\omega \left ( s \right ) \\&&+\int_{t_{1} }^{t_{1}+\tau }S\left (t_{1}+\tau-s \right )g\left ( s,x\left ( s \right ),x_{s} \right ){\rm d}\omega \left ( s \right )\bigg \|_{X}^{p}\\ &\le& 2^{p-1}c_{p} \bigg [ \int_{0}^{t_{1} }\left ( E\left \| \left (S\left (t_{1}+\tau-s \right )-S\left (t_{1}-s \right ) \right ) g\left ( s,x\left ( s \right ),x_{s} \right ) \right \|_{X}^{p} \right ) ^{2/p}{\rm d}s\bigg ]^{p/2}\\&& +2^{p-1}c_{p}\bigg [ \int_{t_{1} }^{t_{1}+ \tau }\left ( E\left \| S\left (t_{1}+\tau-s \right )g\left ( s,x\left ( s \right ),x_{s} \right ) \right \|_{X}^{p} \right ) ^{2/p}{\rm d}s\bigg ]^{p/2},\end{eqnarray*}$

其中 $c_{p}=\left ( p\left ( p-1\right )/2\right )^{p/2}$, 由此可知$\lim\limits_{\tau \to 0}E\left \|I_{5}\left (t_{1}+ \tau \right )- I_{5}\left (t_{1} \right ) \right \|_{X}^{p} =0$. 同理, 利用$S\left ( t \right )$的强连续性, 易证$\lim\limits_{\tau \to 0}E\left \|I_{6}\left (t_{1}+ \tau \right )- I_{6}\left (t_{1} \right ) \right \|_{X}^{p} =0$.

$\begin{eqnarray*} E\left \| I_{7}\left ( t_{1}+\tau\right)-I_{7}\left ( t_{1} \right )\right \|_{X}^{p} \le E\bigg\| \sum\limits_{0<t_{k}<t }\left ( S\left ( t_{1}+\tau -t_{k} \right )-S\left ( t_{1} -t_{k} \right ) \right )I_{k }\left ( x\left ( t_{k}^{-} \right ) \right ) \bigg\|_{X}^{p}.\end{eqnarray*}$

由此$\lim\limits_{\tau \to 0}E\left \|I_{7}\left (t_{1}+ \tau \right )- I_{7}\left (t_{1} \right ) \right \|_{X}^{p} =0$. 进而$\pi $$\left [ 0, \infty \right ) $上是$p$阶连续的.

下一步证明$\pi \left ( S \right )\subset S $, 由(4.2)式得

$\begin{matrix} e^{\beta t}E\left \| \left ( \pi x \right ) \left ( t \right ) \right \|_{X}^{p} &\le & 7^{p-1} e^{\beta t}E\left \| S\left ( t \right )\left ( \phi \left ( 0 \right )+u\left ( 0,\phi,\varphi \right ) \right ) \right \|_{X}^{p}+7^{p-1} e^{\beta t}E\left \| u\left ( t,x\left ( t \right ),x_{t} \right ) \right \|_{X}^{p} \\ &&+7^{p-1} e^{\beta t}E\bigg \| \int_{0}^{t}AS\left ( t-s \right )u\left ( s,x\left ( s \right ),x_{s}\right){\rm d}s \bigg \|_{X}^{p} \\& &+7^{p-1} e^{\beta t}E\bigg \| \int_{0}^{t}S\left ( t-s \right )f\left ( s,x\left ( s \right ),x_{s} \right ){\rm d}s \bigg \| _{X}^{p}\\ & &+7^{p-1} e^{\beta t}E\bigg \| \int_{0}^{t}S\left ( t-s \right )g\left ( s,x\left ( s \right ),x_{s} \right ){\rm d}\omega \left ( s \right ) \bigg \|_{X}^{p} \\ & &+7^{p-1} e^{\beta t}E\bigg \| \int_{0}^{t} \int _{{\Bbb Z} }S\left ( t-s \right )h\left ( s,x\left ( s \right ),x_{s},y\right )\widetilde{N}\left ( {\rm d}s, {\rm d}y \right ) \big \|_{X}^{p}\\ & & +7^{p-1} e^{\beta t}E\bigg \| \sum\limits_{0< t_{k}< t }S\left ( t-t_{_{k} } \right )I_{k}\left (x\left ( t_{k}^{-} \right ) \right ) \bigg \|_{X}^{p}\\ &:=&\sum\limits_{i=1}^{7} Q_{i}\left ( t \right ).\end{matrix}$

首先, 由假设(A1)可知

$\begin{eqnarray*} 7^{p-1} e^{\beta t}E\left \| S\left ( t \right )\left ( \phi \left ( 0 \right )+u\left ( 0,\phi,\varphi \right ) \right ) \right \|_{X}^{p}\le 7^{p-1} M^{p}e^{-p\gamma t} e^{\beta t}E\left \| \phi \left ( 0 \right )+u\left ( 0,\phi,\varphi \right ) \right \| _{X}^{p}.\end{eqnarray*}$

$\lim\limits_{t \to \infty}7^{p-1} e^{\beta t}E\left \| S\left ( t \right )\left ( \phi \left ( 0 \right )+u\left ( 0,\phi,\varphi \right ) \right ) \right \|_{X}^{p}=0$.$u\left ( t,0,\varphi \right )= 0$, 对$\forall \;x\left ( t \right )\in S$, 取$\varepsilon >0$ 足够小, $\exists\; T> 0 $, 使得当$t-\tau > T$时, 有$e^{\alpha t}E\left \| x\left ( t \right ) \right \|_{X}^{p}< \varepsilon$ 成立, 此外, 由假设(A3)可知

$\begin{eqnarray*} & &7^{p-1} e^{\beta t}E\big\| u\left ( t,x\left ( t \right ),x_{t} \right ) \big \|_{X}^{p}\\ &\le& 7^{p-1} e^{\beta t}\big\| \left ( -A \right )^{-\alpha } \big\|_{X}^{p}L^{p}E\left ( \left \| x\left ( t \right ) \right \| +\left \| x_{t} -\varphi \right \| \right )_{X}^{p}\\ &\le &7^{p-1} e^{\beta t}\big\| \left ( -A \right )^{-\alpha } \big \|_{X}^{p}L^{p}E\big ( \left \| x\left ( t \right ) \right \| +\sup\limits_{-\tau \le s\le 0}\big \| x\left ( t+s \right ) \big \| \big )^{p}\\ &\le &14^{p-1} \big\| \left ( -A \right )^{-\alpha } \big\|_{X}^{p}L^{p}e^{\beta t}E\left \| x\left ( t \right ) \right \| _{X}^{p}+ 14^{p-1} \big\| \left ( -A \right )^{-\alpha } \big \|_{X}^{p}L^{p}e^{\beta t}e^{-\beta \left (t+s \right ) }E\left \| x\left ( t \right ) \right \| _{X}^{p}\\ &\le &14^{p-1} \big \| \left ( -A \right )^{-\alpha }\big \|_{X}^{p}L^{p}\varepsilon+14^{p-1} \big\| \left ( -A \right )^{-\alpha } \big\|_{X}^{p}L^{p}e^{-\beta s}\varepsilon.\end{eqnarray*}$

这表明$ \lim\limits_{t \to \infty} 7^{p-1} e^{\beta t}E\left \| u\left ( t,x\left ( t \right ),x_{t} \right ) \right \|_{X}^{p}=0$.

由于证明$\lim\limits_{t \to \infty} Q_{i}\left ( t \right ) =0 \ (i=3,4,5,6)$的方法类似, 我们只证明$\lim\limits_{t\to \infty} Q_{6}\left ( t \right ) =0$. 假设(A1)和(A5)成立, 有

$\begin{eqnarray*} &&7^{p-1} e^{\beta t}E\bigg \| \int_{0}^{t} \int _{{\Bbb Z} }S\left ( t-s \right )h\left ( s,x\left ( s \right ),x_{s},y\right )\widetilde{N}\left ( {\rm d}s, {\rm d}y \right ) \bigg \|_{X}^{p}\\ &\le& 7^{p-1} e^{\beta t}c_{p}M^{p} E\bigg ( \int_{0}^{t}\int _{{\Bbb Z}}e^{-2\gamma \left ( t-s \right ) } \big \| h\left ( s,x\left(s\right),x_{s},y \right)\\&&-h\left ( s,0,\varphi,y\right )+h\left ( s,0,\varphi,y\right ) \big \|^{2}{\rm d}s\nu \left ({\rm d}y\right ) \bigg )^{p/2}\\ &\le& 14^{p-1} e^{\beta t}c_{p}M^{p}2^{\frac{p}{2}}L_{3}^{p}\bigg ( \int_{0}^{t}e^{-2\gamma\left(t-s\right)}E\left(\left\|x\left(s\right)\right\|^{2}+\left|\left\|x_{s}-\varphi\right \|\right|^{2}\right){\rm d}s\bigg)^{\frac{p}{2}}\\&&+14^{p-1} e^{\beta t}c_{p}M^{p}2^{\frac{p}{2}}\bigg(\int_{0}^{t}e^{-2\gamma\left(t-s\right)}E\left \| h\left ( s,0,\varphi,y\right)\right\|^{2}\nu ({\rm d}y ){\rm d}s\bigg)^{\frac{p}{2}}\\ &\le&14^{p-1} e^{\beta t} c_{p}M^{p}2^{\frac{p}{2}} \Big (\frac{p-2}{2\gamma\left(p-1\right)} \Big )^{\frac{p-2}{2}}\int_{0}^{t}e^{-\gamma\left(t-s\right)}L_{3}^{p}E\left ( \left \| x\left ( s \right ) \right \| +\left \| x_{s}-\varphi \right \| \right )^{p}{\rm d}s\\&&+14^{p-1} e^{\beta t}c_{p}M^{p}2^{\frac{p}{2}}\bigg (\int_{0}^{t}e^{-2\gamma\left(t-s\right)}E\left \| h\left ( s,0,\varphi,y\right)\right\|^{2}\nu \left ({\rm d}y\right ){\rm d}s\bigg )^{\frac{p}{2}}\\ &\le&28^{p-1} c_{p}M^{p}2^{\frac{p}{2}}\Big (\frac{p-2}{2\gamma\left(p-1\right)} \Big )^{\frac{p-2}{2}}L_{3}^{p}e^{-\left (\gamma- \beta \right )t }\cdot \int_{0}^{t}e^{\left ( \gamma -\beta \right )s }\cdot e^{\beta s}E\left \| x\left ( s \right ) \right \| _{X}^{p}{\rm d}s\\&&+56^{p-1} c_{p}M^{p}2^{\frac{p}{2}}\Big (\frac{p-2}{2\gamma\left(p-1\right)} \Big )^{\frac{p-2}{2}}L_{3}^{p}e^{-\left (\gamma- \beta \right )t-\beta \theta }\int_{0}^{t}e^{\left ( \gamma -\beta \right )s }e^{\beta \left ( s+\theta \right ) } \sup\limits_{-\tau \le \theta \le 0} E\left \| x\left ( s+\theta \right ) \right \| _{X}^{p}{\rm d}s\\&&+56^{p-1} c_{p}M^{p}2^{\frac{p}{2}}\Big (\frac{p-2}{2\gamma\left(p-1\right)} \Big )^{\frac{p-2}{2}}L_{3}^{p}e^{-\left (\gamma- \beta \right )t }\cdot \int_{0}^{t}e^{\left ( \gamma -\beta \right )s }\cdot e^{\beta s}E\left \| \varphi\left ( s \right ) \right \| _{X}^{p}{\rm d}s\\&&+14^{p-1} e^{\beta t}c_{p}M^{p}2^{\frac{p}{2}}\bigg (\int_{0}^{t}e^{-2\gamma\left(t-s\right)}E\left \| h\left ( s,0,\varphi,y\right)\right\|^{2}\nu \left ({\rm d}y\right ){\rm d}s\bigg)^{\frac{p}{2}}.\end{eqnarray*}$

由于$\int_{0}^{t}e^{-2\gamma\left(t-s\right)}E\left \| h\left ( s,0,\varphi,y\right)\right\|^{2}\nu \left ({\rm d}y\right ){\rm d}s< \infty$, 故

$\begin{eqnarray*} \lim\limits_{t \to 0}7^{p-1} e^{\beta t}E\bigg \|\int_{0}^{t} \int _{{\Bbb Z} }S\left ( t-s \right )h\left ( s,x\left ( s \right ),x_{s},y\right )\widetilde{N}\left ( {\rm d}s, {\rm d}y \right ) \bigg \|_{X}^{p}=0.\end{eqnarray*}$

剩下证明$\lim\limits_{t\to \infty} Q_{7} \left ( t \right ) =0.$ 假设(A4)成立, 则

$\begin{eqnarray*}&& 7^{p-1} e^{\beta t}E\bigg \| \sum\limits_{0< t_{k}< t }S\left ( t-t_{_{k} } \right ) I_{k}\left (x\left ( t_{k}^{-} \right ) \right ) \bigg \|_{X}^{p}\\ &\le&7^{p-1} e^{\beta t} E\bigg \| \sum\limits_{0<t_{k}<t}Me^{-\gamma\left(t-t_{k}\right)}\left(I_{k}\left(x\left(t_{k}\right)-I_{k}\left(0\right)\right)\right)\bigg \|_{X}^{p} \\ &\le &7^{p-1} M^{p}\bigg ( \sum\limits_{t_{k}<t}q_{k}\bigg)^{p-1}\sum\limits_{t_{k}<t}q_{k}e^{-\gamma p\left ( t-t_{k}\right)}e^{\beta t}E\left \| x\left ( t_{k}^{-}\right)\right \|_{X}^{p}\\ &\le &7^{p-1} M^{p}\bigg ( \sum\limits_{t_{k}<t}q_{k}\bigg)^{p}e^{-\left ( \gamma p-\beta \right )\left ( t-t_{k} \right ) }e^{\beta t_{k}}E\left \| x\left ( t_{k}^{-}\right)\right \|_{X}^{p}.\end{eqnarray*}$

由于$p>1$, 且 $0< \beta < \gamma $, 故$\lim\limits_{t\to 0}e^{-\left ( \gamma p-\beta \right )\left ( t-t_{k} \right ) }=0$, 又$e^{\beta t_{k}}E\left \| x\left ( t_{k}^{-}\right)\right \|_{X}^{p}< \varepsilon $, 从而

$\begin{eqnarray*} \lim\limits_{t \to \infty}7^{p-1} e^{\beta t}E\bigg \| \sum\limits_{0< t_{k}< t }S\left ( t-t_{_{k} } \right )I_{k}\left (x\left ( t_{k}^{-} \right ) \right ) \bigg \|_{X}^{p}=0.\end{eqnarray*}$

综上所述

$\lim\limits_{t\to \infty} e^{\beta t}E\left \| \left ( \pi x \right ) \left ( t \right ) \right \|_{X}^{p}=0,$

于是, $\pi \left ( S \right )\subset S $. 最后, 我们将证明$\pi$是压缩的, 对$\forall\; x,y\in S$, 根据(4.2)式, 我们有

$\begin{eqnarray*} & &E\sup\limits_{t\in \left [ 0,T \right ] }\left \| \left ( \pi x \right )\left ( t \right ) - \left ( \pi y \right )\left ( t \right ) \right \| _{X}^{p}\\ &\le &6^{p-1}E\sup\limits_{t\in \left [ 0,T \right ] }\left \| u\left ( t,y\left ( t \right ),y_{t} -u\left ( t,x\left ( t \right ),x_{t} \right ) \right ) \right \|_{X}^{p}\\ &&+ 6^{p-1}E\sup\limits_{t\in \left [ 0,T \right ] }\bigg \| \int_{0}^{t}AS\left ( t-s \right )\left ( u\left ( s,y\left ( s \right ),y_{s}\right)-u\left ( s,x\left ( s \right ),x_{s}\right) \right ){\rm d}s \bigg \|_{X}^{p}\\ &&+ 6^{p-1}E\sup\limits_{t\in \left [ 0,T \right ] }\bigg \| \int_{0}^{t}S\left ( t-s \right )\left ( f\left ( s,x\left ( s \right ),x_{s}\right)-f\left ( s,y\left ( s \right ),y_{s}\right) \right ){\rm d}s \bigg \|_{X}^{p}\\ &&+ 6^{p-1}E\sup\limits_{t\in \left [ 0,T \right ] }\bigg \| \int_{0}^{t}S\left ( t-s \right )\left ( g\left ( s,x\left ( s \right ),x_{s}\right)-g\left ( s,y\left ( s \right ),y_{s}\right) \right ) {\rm d}\omega \left ( s \right ) \bigg \|_{X}^{p}\\ &&+ 6^{p-1}E\sup\limits_{t\in \left [ 0,T \right ] }\bigg \| \int_{0}^{t} \int _{{\Bbb Z} }S\left ( t-s \right )\left ( h\left ( s,x\left ( s \right ),x_{s},y\right )-h\left ( s,y\left ( s \right ),y_{s},y\right ) \right ) \widetilde{N}\left ( {\rm d}s, {\rm d}y \right ) \bigg \| _{X}^{p}\\ &&+ 6^{p-1}E\sup\limits_{t\in \left [ 0,T \right ] }\bigg \| \sum\limits_{0< t_{k}< t }S\left ( t-t_{_{k} } \right )I_{k}\left (x\left ( t_{k}^{-} \right ) -y\left ( t_{k}^{-} \right ) \right ) \bigg \| _{X}^{p}\\ &\le&12^{p-1}\big \| \left ( -A \right )^{-\alpha } \big \|_{X}^{p}L^{p} \big( E\sup\limits_{t\in \left [ 0,T \right ] } \left \| x\left ( t \right )-y\left ( t \right ) \right \|_{X}^{p}+E\sup\limits_{t\in \left [ 0,T \right ] } \left \| x_{t}-y_{t} \right \|_{X}^{p} \big)\\ &&+12^{p-1}M_{1-\alpha }^{p}L^{p}\gamma ^{-\alpha p}\Big( \Gamma \big ( 1+\frac{p\left ( \alpha -1 \right ) }{p-1} \big ) \Big)^{p-1}\big( E\sup\limits_{t\in \left [ 0,T \right ] } \left \| x\left ( t \right )\!-\!y\left ( t \right ) \right \|_{X}^{p}\!+\!E\sup\limits_{t\in \left [ 0,T \right ] } \left \| x_{t}\!-\!y_{t} \right \|_{X}^{p} \big)\\ &&+12^{p-1}M^{p}L_{1}^{p}\gamma ^{-p} \big( E\sup\limits_{t\in \left [ 0,T \right ] } \left \| x\left ( t \right )-y\left ( t \right ) \right \|_{X}^{p}+E\sup\limits_{t\in \left [ 0,T \right ] } \left \| x_{t}-y_{t} \right \|_{X}^{p} \big)\\ &&+12^{p-1}c_{p}M^{p}\Big ( \frac{p-2}{2\gamma \left ( p-1 \right ) } \Big)^{1-\frac{p}{2}}\gamma ^{-1}L_{2}^{p} \big( E\sup\limits_{t\in \left [ 0,T \right ] } \left \| x\left ( t \right )-y\left ( t \right ) \right \|_{X}^{p}+E\sup\limits_{t\in \left [ 0,T \right ] } \left \| x_{t}-y_{t} \right \|_{X}^{p} \big)\\&&+12^{p-1}c_{p}M^{p}2^{\frac{p}{2}}\left(\frac{p-2}{2\gamma\left(p-1\right)} \right)^{\frac{p-2}{2}}L_{3}^{p}\gamma ^{-1} \big( E\sup\limits_{t\in \left [ 0,T \right ] } \left \| x\left ( t \right )-y\left ( t \right ) \right \|_{X}^{p}+E\sup\limits_{t\in \left [ 0,T \right ] } \left \| x_{t}-y_{t} \right \|_{X}^{p} \big)\\&&+12^{p-1} M^{p}\bigg ( \sum\limits_{t_{k}<t}q_{k}\bigg)^{p}e^{- \gamma p \left ( t-t_{k} \right ) }E\left \| x\left ( t\right)-y\left ( t\right)\right \|_{X}^{p}\\ &\le&2\cdot 12^{p-1}\bigg[\big \| \left ( -A \right )^{-\alpha } \big \|_{X}^{p}L^{p}+M_{1-\alpha }^{p}L^{p} \gamma ^{-\alpha p}\Big( \Gamma \big ( 1+\frac{p\left ( \alpha -1 \right ) }{p-1} \big ) \Big)^{p-1}+M^{p}L_{1}^{p}\gamma ^{-p}\\&&+c_{p}M^{p}\Big ( \frac{p-2}{2\gamma \left ( p-1 \right ) } \Big)^{1-\frac{p}{2}}\gamma ^{-1}L_{2}^{p} +c_{p}M^{p}2^{\frac{p}{2}}\left(\frac{p-2}{2\gamma\left(p-1\right)} \right)^{\frac{p-2}{2}}L_{3}^{p}\gamma ^{-1}\\ &&+\frac{1}{2} M^{p}\bigg ( \sum\limits_{t_{k}<t}q_{k}\bigg)^{p} e^{- \gamma p \left ( t-t_{k} \right ) } \bigg ]\sup\limits_{t\in \left [ 0,T \right ] } \left \| x\left ( t \right )-y\left ( t \right ) \right \|_{X}^{p}\\ &= &C^{'} E\sup\limits_{t\in \left [ 0,T \right ] } \left \| x\left ( t \right )-y\left ( t \right ) \right \|_{X}^{p}.\end{eqnarray*}$

由于$C^{'}<1$, 则$\pi$是一个压缩映射. 由压缩原理知, $\pi$$S$上有唯一的不动点$x\left ( t \right ) $, 即方程(2.1)的解, 且满足在$\left [ -\tau,0 \right ] $上, $x\left ( s \right ) =\phi \left ( s \right ) $$x_{_{s} }=\varphi \left ( s \right )$, 同时, 有$\lim\limits_{t\to \infty} e^{\beta t}E\left \| x\left ( t \right ) \right \|_{X}^{p} =0$ 成立, 这就意味着该唯一解$x\left ( t \right ) $$p$时刻处是指数稳定的.

5 例子

下面我们给出一个带时滞和泊松跳的脉冲中立型随机微分方程的例子.

$\begin{matrix} x\left ( t \right )& =&x_{0} +\int_{0}^{t} f\left ( s,x\left ( s-\tau(s) \right ) \right ){\rm d}s+\int_{0}^{t}g\left ( s,x\left ( s-\rho (s) \right ) \right ){\rm d}\omega \left ( s \right )\\ &&+\int_{0}^{t}\int _{{\Bbb Z} } h\left ( s,x\left ( s-\delta (s) \right ),y \right ) \widetilde{N}\left ( {\rm d}s, {\rm d}y \right ) +\sum\limits_{0<t_{k}<t } I_{k} x\left ( t_{k}^{-} \right ), \end{matrix}$

这里, 令初始值 $x\left ( 0 \right )=x_{0},\; \xi(t),\; \rho(t),\; \delta(t) :\left [ 0,+\infty \right )\to \left [ 0,\tau \right ] $, 且映射 $f\left ( t,\cdot \right )$, $ g\left ( t,\cdot \right )$$h\left ( t,\cdot,\cdot \right ) $ 满足如下的 Lipschitz 条件和线性增长条件

$\begin{matrix} &&\left \| f\left ( t,x \right )-f\left ( t,y \right ) \right \|\le L_{1}\left \| x-y \right \|,uad L_{1}>0,\\ &&\left \| g\left ( t,x \right )-g\left ( t,y \right ) \right \|\le L_{2}\left \| x-y \right \|,uad L_{2}>0,\\& &\int _{{\Bbb Z} } \left \| h\left ( t,x,z \right )-h\left ( t,y,z \right ) \right \| ^{2}\nu \left ({\rm d}z \right )\le L_{3}\left \| x-y \right \|^{2},uad L_{3}>0.\end{matrix}$

注意到此例子的模型中是不含有线性算子项, 这就暗示着(3.10)式中的 $M$取值为0,

那么利用逐次逼近法和函数的Lipschitz性质可知, 对于任意的 $n\ge 1$, 存在一个连续函数 $ m\left ( t \right ) \in C\left [ 0,T \right ]$, 使得 $E\max_{s\in \left [ 0,t \right ]} \left \| x^{n} \left ( s \right ) \right \| ^{2}\le m\left ( t \right ) $, 因此, 我们可以得到 $ E\max_{s\in \left [t \right]} \left \| x^{n} \left ( s \right ) \right \| ^{2}\le C< \infty $. 进一步, 利用 Holder 不等式, Doob 不等式以及假设 (A3)可得

$\lim\limits_{n,m \to \infty}\sup E \Big( \max_{s\in \left [t \right]}\left \| x^{n} \left ( s \right )-x^{m}\left ( s \right ) \right \|^{2} \Big )=0.$

最后, 由 Borel-Cantelli 引理表明 $x^{n} \left ( t \right )\to x\left ( t \right ), v^{n} \left ( t \right )\to v\left ( t \right ) $是几乎一致成立的, 因此, $x\left(t\right)=v\left (t\right) $, 这就表明系统(5.1)的温和解是唯一存在的.

$\begin{matrix} e^{\beta t}E\left \| \left ( \pi x \right ) \left ( t \right ) \right \|_{X}^{p} &\le & 5^{p-1} e^{\beta t}x_{0}^{p} +5^{p-1} e^{\beta t}E\bigg \| \int_{0}^{t}f(s,x\left ( s-\tau(s) \right ) ){\rm d}s \bigg \| _{X}^{p}\\ & &+5^{p-1} e^{\beta t}E\bigg \| \int_{0}^{t}g\left ( s,x\left ( s-\delta (s) \right ) \right ){\rm d}\omega \left ( s \right ) \bigg \|_{X}^{p} \\ & &+5^{p-1} e^{\beta t}E\bigg \| \int_{0}^{t} \int _{{\Bbb Z} }h\left ( s,x\left ( s-\rho (s) \right ),y\right )\widetilde{N}\left ( {\rm d}s, {\rm d}y \right ) \bigg \|_{X}^{p}\\ && +5^{p-1} e^{\beta t}E\bigg \| \sum\limits_{0< t_{k}< t }I_{k}\left (x\left ( t_{k}^{-} \right ) \right ) \bigg \|_{X}^{p}\\ &:=&\sum\limits_{i=1}^{5} Q_{i}(t).\end{matrix}$

根据前面的证明过程易得

$\lim\limits_{t\to \infty} e^{\beta t}E\left \| \left ( \pi x \right ) \left ( t \right ) \right \|_{X}^{p}=0,$

说明 $\pi \left ( S \right )\subset S $

$\begin{matrix} & &E\sup\limits_{t\in \left [ 0,T \right ] }\left \| \left ( \pi x \right )\left ( t \right ) - \left ( \pi y \right )\left ( t \right ) \right \| _{X}^{p}\\ &&+ 4^{p-1}E\sup\limits_{t\in \left [ 0,T \right ] }\bigg \| \int_{0}^{t}\left ( f\left ( s,x\left ( s-\tau\left ( s \right ) \right )\right)-f\left ( s,y\left ( s-\tau\left ( s \right ) \right )\right) \right ){\rm d}s \bigg \|_{X}^{p}\\ &&+ 4^{p-1}E\sup\limits_{t\in \left [ 0,T \right ] }\bigg \| \int_{0}^{t}\left ( g\left ( s,x\left ( s-\delta\left ( s \right ) \right )\right)-g\left ( s,y\left ( s-\delta \right )\right) \right ) {\rm d}\omega \left ( s \right ) \bigg \|_{X}^{p}\\ &&+ 4^{p-1}E\sup\limits_{t\in \left [ 0,T \right ] }\bigg \| \int_{0}^{t} \int _{{\Bbb Z} }\left ( h\left ( s,x\left ( s -\rho \right ),x_{s} \right)-h\left ( s,y\left ( s-\rho \right ),y\right ) \right ) \widetilde{N}\left ( {\rm d}s, {\rm d}y \right ) \bigg \| _{X}^{p}\\ &&+ 4^{p-1}E\sup\limits_{t\in \left [ 0,T \right ] }\bigg \| \sum\limits_{0< t_{k}< t }S\left ( t-t_{_{k} } \right )I_{k}\left (x\left ( t_{k}^{-} \right ) -y\left ( t_{k}^{-} \right ) \right ) \bigg \| _{X}^{p}\\ &\le&\bigg[4^{p-1}L_{1}^{p}\gamma ^{-p} +4^{p-1}c_{p} L_{2}^{p}\gamma ^{-p}+ 4^{p-1}c_{p} L_{3}^{p}\gamma ^{-1}+4^{p-1}\Big ( \sum\limits_{t_{k}<t}q_{k}\Big)^{p} e^{- \gamma p \left ( t-t_{k} \right ) } \bigg]\\&&\cdot E\sup\limits_{t\in \left [ 0,T \right ] } \left \| x\left ( t \right )-y\left ( t \right ) \right \|_{X}^{p}\\ &= &C^{''} E\sup\limits_{t\in \left [ 0,T \right ] } \left \| x\left ( t \right )-y\left ( t \right ) \right \|_{X}^{p}.\end{matrix}$

由(4.1)式知$C^{''}<1$, 则$\pi$是一个压缩映射. 从而验证了该唯一解$x\left ( t \right ) $$p$时刻处是指数稳定的.

6 总结

脉冲中立型随机泛函微分方程在许多实际过程的建模中起着重要的作用, 因此研究这类方程的性质非常重要. 尤其是研究系统的稳定性激发了广大学者对神经网络的广泛应用产生了极大的兴趣, 例如, Li等[28]研究了随机时滞Hopfield神经网络数值解的指数稳定性; Rathinasamy[29]用分步法研究了随机时滞Hopfield神经网络的均方稳定性; Jiang等[30]讨论了随机延迟Hopfield神经网络数值模拟的稳定性; Liu等[31]研究了随机时滞Hopfield神经网络数值解的几乎确定指数稳定性.

本文用逐次逼近方法研究了具有泊松跳的脉冲中立型随机泛函微分方程温和解的存在唯一性, 并用Holder不等式和Borel-Cantelli引理完成了证明. 其次, 利用Banach不动点法证明了该唯一解在$p$时刻处是指数稳定的. 由于随机泛函微分方程的大偏差理论的研究是一个相对较新的研究方向, 也是一个非常有趣的问题. 在随后的研究工作中, 我们将考虑建立基于一般非负泛函变分表示的随机泛函微分方程适定性所需的估计, 以获得随机泛函微分方程的大偏差结果.

参考文献

Hale J K, Lunel S M V. Introduction to Functional-Differential Equations. Applied Mathematical Sciences, New York: Springer-Verlag, 1993

[本文引用: 1]

Liu K.

The fundamental solution and its role in the optimal control of infinite dimensional neutral systems

Appl Math Optim, 2009, 60(1): 1-38

DOI:10.1007/s00245-009-9065-1      URL     [本文引用: 1]

Wu S J, Meng X Z.

Boundedness of nonlinear differential systems with impulsive effect on random moments

Acta Math Appl Sin Engl Ser, 2004, 20(1): 147-154

DOI:10.1007/s10255-004-0157-z      URL     [本文引用: 1]

Wu F K, Mao X R, Szpruch L.

Almost sure exponential stability of numerical solutions for stochastic delay differential equations

Numer Math, 2010, 115(4): 681-697

DOI:10.1007/s00211-010-0294-7      URL     [本文引用: 1]

沈轶, 张玉民, 廖晓昕.

中立型随机泛函微分方程的稳定性

数学物理学报, 2005, 25(3): 323-330

[本文引用: 1]

Shen Y, Zhang Y M, Liao X X.

Stability for neutral stochastic functional differential equations

Acta Math Sci, 2005, 25(3): 323-330

[本文引用: 1]

崔静, 梁秋菊, 毕娜娜.

分数布朗运动驱动的带脉冲的中立性随机泛函微分方程的渐近稳定

数学物理学报, 2019, 39(3): 570-581

[本文引用: 1]

Cui J, Liang Q J, Bi N N.

Asymptotic stability of impulsive neutral stochastic functional differential equation driven by fractional Brownian motion

Acta Math Sci, 2019, 39(3): 570-581

[本文引用: 1]

Song Y, Zeng Z.

Razumikhin-type theorems on th moment boundedness of neutral stochastic functional differential equations with Makovian switching

J Franklin Inst B, 2018, 355(5): 8296-8312

DOI:10.1016/j.jfranklin.2018.09.019      URL     [本文引用: 1]

Chang M.

On Razumikhin-type stability conditions for stochastic functional differential equations

Math Modelling, 1984, 5(5): 299-307

DOI:10.1016/0270-0255(84)90007-1      URL     [本文引用: 1]

沈轶, 廖晓昕.

随机中立型泛函微分方程指数稳定的Razumikhin型定理

科学通报, 1998, 21: 2272-2275

[本文引用: 1]

Shen Y, Liao X X.

The Razumikhin type theorem for exponential stability of stochastic neutral type functional differential equations

Chinese Science Bulletin, 1998, 21: 2272-2275

[本文引用: 1]

Ren Y, Xia N.

Existence, uniqueness and stability of the solutions to neutral stochastic functional differential equations with infinite delay

Appl Math Comput, 2008, 210(1): 72-79

[本文引用: 1]

Yang Z, Xu D.

Stability analysis and design of impulsive control systems with time delay

IEEE Trans Automat Control, 2007, 52(8): 48-54

[本文引用: 1]

Guo Y, Zhu Q, Wang F.

Stability analysis of impulsive stochastic functional differential equations

Commun Nonlinear Sci Numer Simul, 2020, 82(C): 5-13

[本文引用: 1]

Zhao X.

Mean square Hyers-Ulam stability of stochastic differential equations driven by Brownian motion

Adv Difference Equ, 2016, 2016(1): 271

DOI:10.1186/s13662-016-1002-4      URL     [本文引用: 1]

Li S, Shu L X, Shu X B, Xu F.

Existence and Hyers-Ulam stability of random impulsive stochastic functional differential equations with finite delays

Stochastic, 2019, 91(6): 857-872

DOI:10.1080/17442508.2018.1551400      URL     [本文引用: 1]

Ferhat M, Blouhi T.

Existence and uniqueness results for systems of impulsive functional stochastic differential equations driven by fractional Brownian motion with multiple delay

Topol Methods Nonlinear Anal, 2018, 52: 449-476

[本文引用: 1]

Wu F, Yin G, Mei H.

Stochastic functional differential equations with infinite delay: existence and uniqueness of solutions, solution maps, Markov properties, and ergodicity

J Diff Equ, 2017, 62(3): 1226-1252

[本文引用: 1]

Banupriya K, Anguraj A.

Successive approximation of neutral functional impulsive stochastic differential equations with Poisson jumps

Dyn Contin Discrete Impuls Syst Ser B Appl and Algorithms, 2021, 28: 215-229

[本文引用: 1]

Boufoussi B, Hajji S.

Successive approximation of neutral functional stochastic differential equations in Hilbert spaces

Ann Math Blaise Pascal, 2010, 17(1): 183-197

DOI:10.5802/ambp.282      URL     [本文引用: 1]

Chen G, van Gaans O, Lunel S V.

Existence and exponential stability of a class of impulsive neutral stochastic partial differential equations with delays and Poisson jumps

Statist Probab Lett, 2018, 141: 7-18

DOI:10.1016/j.spl.2018.05.017      URL     [本文引用: 2]

Deng S F, Shu X B, Mao J Z.

Existence and exponential stability for impulsive neutral stochastic functional differential equations driven by fBm with noncompact semigroup via Mönch fixed point

J Math Anal Appl, 2018, 467(1): 398-420

DOI:10.1016/j.jmaa.2018.07.002      URL     [本文引用: 1]

Guo Y C, Chen M Q, Shu X B, Xu F.

The existence and Hyers-Ulam stability of solution for almost periodical fractional stochastic differential equation with fBm

Stoch Anal Appl, 2021, 39(4): 643-666

DOI:10.1080/07362994.2020.1824677      URL     [本文引用: 1]

Faizullah F, Zhu Q, Ullah R.

The existence-uniqueness and exponential estimate of solutions for stochastic functional differential equations driven by G-Brownian motion

Math Methods Appl Sci, 2021, 44: 1639-1650

DOI:10.1002/mma.v44.2      URL     [本文引用: 1]

Xiao G, Wang J, O'Regan D.

Existence and stability of solutions to neutral conformable stochastic functional differential equations

Qual Theory Dyn Syst, 2022, 21(1): 7-22

DOI:10.1007/s12346-021-00538-x      [本文引用: 1]

Tan J G, Tan Y H, Guo Y F, Feng J F.

Almost sure exponential stability of numerical solutions for stochastic delay Hopfield neural networks with jumps

Phys A: Statist Mech Appl, 2020, 545: 37-82

[本文引用: 1]

Pazy A. Semigroups of Linear Operators and Applications to Partial Differential Equations. Appl Math Sci vol. New York: Springer Verlag, 1983

[本文引用: 1]

Da Prato G, Zabczyk J. Stochastic Equations in Infinite Dimensions. Cambridge: Cambridge University Press, 1992, 45

[本文引用: 1]

Luo Q, Mao X, Shen Y.

New criteria on exponential stability of neutral stochastic differential delay equations

Syst Control Lett, 2006, 55(10): 826-834

DOI:10.1016/j.sysconle.2006.04.005      URL     [本文引用: 1]

Li R, Pang W, Leung P.

Exponential stability of numerical solutions to stochastic delay Hopfield neural networks

Appl Math Comput, 2009, 73(4): 920-926

[本文引用: 1]

Rathinasamy A.

The split-step $\theta $-methods for stochastic delay Hopfield neural networks

Appl Math Model, 2021, 36: 3477-3485

DOI:10.1016/j.apm.2011.10.020      URL     [本文引用: 1]

Jiang F, Shen Y.

Stability in the numerical simulation of stochastic delayed Hopfield neural networks

Neural Comput Appl, 2013, 22(7/8): 1493-1498

DOI:10.1007/s00521-012-0935-0      URL     [本文引用: 1]

Liu L, Zhu Q.

Almost sure exponential of numerical solutions to stochastic delay Hopfield neural networks

Appl Math Comput, 2015, 266: 698-712

[本文引用: 1]

/