数学杂志  2017, Vol. 37 Issue (4): 811-818   PDF    
扩展功能
加入收藏夹
复制引文信息
加入引用管理器
Email Alert
RSS
本文作者相关文章
刘宏亮
石峰
朱琳
Banach空间中一类二阶具阻尼项的积分-微分包含的可控性
刘宏亮, 石峰, 朱琳     
哈尔滨师范大学数学科学学院, 黑龙江 哈尔滨 150025
摘要:本文讨论了一类二阶具阻尼项的积分-微分包含问题.利用集值不动点定理将问题转化为求不动点问题,进而得到系统可控性的充分条件.
关键词微分包含    可控性    不动点定理    集值映射    
CONTROLLABILITY OF SECOND-ORDER DIFFERENTIAL INCLUSIONS IN BANACH SPACE WITH DAMPED AND INTEGRAL TERMS
LIU Hong-liang, SHI Feng, ZHU Lin     
College of Mathematics Science, Harbin Normal University, Harbin 150025, China
Abstract: In this paper, a class of difierential inclusions with the damped term and integral term in Banach space was studied. By using flxed point theorem, a su-cient condition for the controllability for such difierential inclusion was obtained.
Key words: difierentail inclusion     controllability     flxed point theorem     multivalued map    
1 引言

Balachandran等[1]讨论了一类二阶积分-微分系统的可控性, 利用Schaefer不动点定理, 给出了系统可控性的充分条件; Benchohra, Ntouyas[2]利用Martelli压缩集值映射不动点定理, 研究了在Banach空间中一类中立型微分和积分微分包含的存在性; Chang-Li[3]利用集值不动点定理, 考虑集值函数具有非凸值时二阶微分包含和积分微分包含的可控性.有关具阻尼项的微分包含也进一步发展起来, Benchohra等[4]利用不动点定理研究了集值映射具有凸和非凸值的情形的阻尼微分包含的存在性; Chalishajar[5]利用Ma不动点定理获得了二阶半线性具阻尼项的中立型微分包含可控性的充分条件.到目前为止, 有关具阻尼项的微分系统的研究成果参见文献[8, 9], 但有关具阻尼项的积分-微分包含系统的研究成果尚未见报道.

本文讨论如下具阻尼项的积分-微分包含

$ \begin{eqnarray} \label{eq:1} &&y^{\prime \prime }(t)-Ay(t) \in Dy^{\prime }(t)+\int_{0}^{t}\beta (t, s)F(t, y(t))ds+Bu(t)\ \ \ t\in J, \end{eqnarray} $ (1.1)
$ \begin{eqnarray} \label{eq:2} y(0)+f(y) =y_{0}, \text{ }y^{\prime }(0)=z_{0} \end{eqnarray} $ (1.2)

可控性的问题, 其中 $E$是Banach空间, $F\colon J\times E\rightarrow 2^{E}$是有界闭凸集值映射.

$ f\colon C(J, E)\rightarrow E, ~~y_{0}, ~~z_{0}\in E, $

$A$ $E=(E, \left \Vert \cdot \right \Vert )$中有界线性算子的强连续余弦算子半群 $\{C(t), t\in R\}$的无穷小生成元.

$ \beta :K\rightarrow R, ~~ K=\left \{ \left( t, s\right) \in J\times J:t\geqslant s\right \}. $

控制函数 $u\in L^{2}(J, U)$, 这里 $U$是一个Banach空间, $B$是从 $U$ $E$的有界线性算子, $D$ $E$上的有界线性算子.

2 预备知识

$(X, \left \Vert \cdot \right \Vert )$是Banach空间, $J=[0, m], $ $C(J, E)$是从 $J$ $X$的连续函数构成的Banach空间, 赋予范数 $\left \Vert x\right \Vert _{\infty }=\sup \{ \left \Vert x(t)\right \Vert :t\in J\}, $ $L^{1}(J, X)$表示Bochner可积函数 $x:J\rightarrow X$构成的Banach空间的全体.

$P(X)$表示所有非空子集的全体, $P_{k}(X)$表示所有紧子集的全体, $P_{b}(X)$表示所有有界子集的全体, $P_{f}(X)$表示所有闭子集的全体, $P_{c}(X)$表示所有凸子集的全体, $P_{bfc}(X)$表示所有有界、闭、凸子集的全体.集值映射 $G:X\rightarrow P(X), $若对于任意的 $x\in X$, $G(x)$是凸(闭、有界、紧)集, 则称 $G(x)$是凸值(闭值、有界值、紧值)映射.对于任意的 $x_{0}\in X, $若对包含 $G(x_{0})$的任意开集 $U, $存在 $x_{0}$的邻域 $V$使得 $G(V)\subset U, $则称 $G$是上半连续的.若集值函数 $G$是全连续的且具有非空紧值, 则 $G$是上半连续的等价于 $G$有闭图, 即

$ \begin{equation*} x_{n}\rightarrow x_{\ast }, y_{n}\rightarrow y_{\ast }, y_{n}\in Gx_{n}\Longrightarrow y_{\ast }\in Gx_{\ast }. \end{equation*} $

若存在 $x\in X$使得 $x\in G(x), $则称 $G$有不动点.设集值函数 $G:J\rightarrow P_{f}(X)$, 若对任意的 $x\in X$, $d(w, G(x))=\inf \{ \left \Vert w-z\right \Vert :z\in G(x)\}$ $J$上的可测函数, 则称 $G$是可测的.对于任意的 $x\in X$, $S_{G, x}$表示 $G(x(\cdot) )$的所有的Lebesgue-Bochner可积全体, 即

$ \begin{equation*} S_{G, x}=\{v\in L^{1}(J, X):v(t)\in G(x(t))\text{ a.e. }t\in J\}. \end{equation*} $

称Banach空间有界线性算子半群 $\{C(t)\colon t\in R\}$是强连续余弦算子半群当且仅当

(1) $C(0)=I$, 其中 $I$是在 $B(X)$中的单位算子;

(2) 对于任意的 $\ s$, $t\in R $, $C(t+s)+C(t-s)=2C(t)C(s)$;

(3) 对于任意 $x\in X$, 映射 $t\rightarrow C(t)x$是强连续.

结合给定的强连续余弦算半群 $\{C(t)\colon t\in R\}$, 定义

$ \begin{equation*} S(t)x=\int_{0}^{t}C(s)xds, \text{ }x\in X, t\in R. \end{equation*} $

则称 $\{S(t)\colon t\in R\}$是强连续正弦算半群.

$A\colon X\rightarrow X$, 定义为

$ \begin{equation*} Ax=\frac{d^{2}}{dt^{2}}C(t)x|_{t=0}, \text{ }x\in K(A), \end{equation*} $

其中 $K(A)=\{x\in X\colon C(t)x$是关于 $t$的二次连续可微函数 $\}$, 则称 $A$为强连续余弦族 $\{C(t)\colon t\in R\}$的无穷小生成元.定义 $E=\{x\in X\colon C(t)x$是关于 $t$的一次连续可微函数 $\}.$

引理2.1[6]  设 $X$是一个Banach空间, 设集值函数 $G:J\times X\rightarrow P_{bfc}(X)$满足

(1) 对于任意的 $x\in X, $ $(t, x)\rightarrow G(t, x)$是关于 $t$可测的;

(2) 对于几乎处处的 $t\in J, $ $(t, x)\rightarrow G(t, x)$关于 $x$是上半连续的;

(3) 对每个固定的 $x\in C(J, E)$, 集合

$ \begin{equation*} S_{G, x}=\{v\in L^{1}(J, X):v(t)\in G(t, x(t))\text{ a.e. }t\in J\} \end{equation*} $

非空的.并且设 $\Gamma :L^{1}(J, X)\rightarrow C(J, X)$是线性连续映射, 则

$ \begin{equation*} \Gamma \circ S_{G}\colon C(J, X)\rightarrow P_{bfc}(C(J, X)), \text{ }% y\rightarrow (\Gamma \circ S_{G})(y)=\Gamma (S_{G, y}) \end{equation*} $

$C(J, X)\times C(J, X)$上的闭图算子.

引理2.2[7]  (Leray-Schauder替换定理)设 $X$是Banach空间, $C\subseteq X$ $X$的非空凸子集, $0\in C$, 集值映射 $G:C\rightarrow P_{k, c}(C)$是上半连续的, 并且映 $C$的有界子集为相对紧集, 则下述论断必有一个成立

(1) 集合 $\Gamma =\{x\in C:x\in \lambda G, \lambda \in (0, 1)\}$是无界的;

(2) 集值映射 $G(\cdot )$ $C$中存在不动点.

定义2.1   函数 $y(\cdot )\in C(J, E)$满足

$ \begin{align*} y(t)\in &~(C(t)-S(t)D)\left( y_{0}-f(y)\right) +S(t)z_{0}+\int_{0}^{t}C(t-s)Dy(s)ds \\ &+\int_{0}^{t}S(t-s)\int_{0}^{s}\beta (s, \tau )F(\tau, y(\tau ))d\tau ds+\int_{0}^{t}S(t-s)Bu(s)ds, \end{align*} $

并且 $y(0)+f(y)=y_{0}$, $y^{\prime }(0)=z_{0}$, 则 $y(\cdot )$称为系统(1.1)-(1.2) 的mild解.

定义2.2   系统(1.1)-(1.2) 在 $J$上可控, 是指对任意的 $y_{0}\in K(A)$ $z_{0}\in E_{1}$, 存在控制 $u\in L^{2}(J, U)$, 使得系统(1.1)-(1.2) 的mild解满足 $ y(m)+f(y)=y_{0}$.

我们需要如下假设

(A1) $A$是Banach空间 $E$上有界线性算子的强连续余弦算子半群 $\{C(t):t\in J\}$的无穷小生成元, 且存在常数 $M>0$, 使得 $M=\sup \{ \left \Vert C(t)\right \Vert \colon t\in J\}$.

(A2) $f$是全连续的, 且存在常数 $L>0$, 使得对任意的 $y\in E$, $||f(y)||\leqslant L.$

(A3) 线性算子 $W\colon L^{2}(J, U)\rightarrow E$, 定义为 $Wu=\displaystyle\int_{0}^{m}S(t-s)Bu(s), $存在取值于 $L^{2}(J, U)\backslash \ker W$上的有界逆算子 $W^{-1}$, 且存在 $M_{1}, $ $M_{2}>0$, 使得 $ \left \Vert B\right \Vert \leqslant M_{1}$, $\left \Vert W^{-1}\right \Vert \leqslant M_{2}$.

(A4) $F\colon J\times E\rightarrow P_{bfc}(E)$, 且 $F$满足对任意的 $y\in E$, $t\rightarrow F(t, y) $是可测的, 对于几乎处处 $\mathit{t} \in \mathit{J}$, $y\rightarrow F(t, y)$是上半连续的; 对每个固定的 $y\in C(J, E)$, 集合

$ \begin{equation*} S_{F, y}=\{v\in L^{1}(J, E)\colon v(t)\in F(t, y(t))\text{ a.e.}~~t\in J\} \end{equation*} $

为非空的;

(A5) 存在一个连续非降函数 $\psi \colon R_{+}\rightarrow (0, \infty )$, 及 $p(t)\in L^{1}(J, R_{+})$对于几乎处处的 $ t\in J$, $y\in E$

$ \begin{equation*} \left \Vert F(t, y)\right \Vert =\sup \{ \left \Vert v\right \Vert \in F(t, y)\} \leqslant p(t)\psi (\left \Vert y\right \Vert ); \end{equation*} $ (11)

(A6) 对于任意的 $t\in J$, $\beta (t, s)$ $\left[0, t\right] $上可测, 并且

$ \begin{equation*} \beta (t)={\rm ess}\text{ }\sup \left \{ \left \vert \beta (t, s)\right \vert :0\leqslant s\leqslant t\right \} \end{equation*} $

$J$上有界;

(A7) 从 $J$ $L^{\infty }(J, R)$, 映射 $ t\rightarrow \beta _{t}$是连续的, 其中 $\beta _{t}(s)=\beta (t, s).$

3 主要结果

定理3.1   若假设条件(A1)-(A7) 成立, 并且存在正数 $ \theta >0$, 使得

$ \begin{equation}\label{eq:3} \frac{\theta }{N_{1}+N_{2}\left[\left \Vert D\right \Vert \theta +mM% \underset{t\in J}{\sup }\beta (t)\displaystyle\int_{0}^{m}p(s)\psi (\theta )ds\right] }>1, \end{equation} $ (3.1)

其中

$ \begin{eqnarray*} &&N_{1}=mMM_{1}M_{2}(\left \Vert y_{1}\right \Vert +L)+(1+mMM_{1}M_{2})N_{3}, \\&& N_{2}=mM(1+mMM_{1}M_{2}), \\ &&N_{3}=(M+mM\left \Vert D\right \Vert )(\left \Vert y_{0}\right \Vert +L)+mM\left \Vert z_{0}\right \Vert, \end{eqnarray*} $

则系统(1.1)-(1.2) 是可控的.

  由(A3), 对任意的 $y(\cdot )\in C(J, E)$, 定义控制函数

$ \begin{aligned} u_{y}(t)=&~W^{-1}\Big[y_{1}-f(y)-(C(m)-S(m)D)(y_{0}-f(y))-S(m)z_{0}\\ &-\displaystyle\int_{0}^{m}C(m-s)Dy(s)ds-\displaystyle\int_{0}^{m}S(m-s)\displaystyle\int_{0}^{s}\beta (s, \tau )v(\tau )d\tau ds\Big](t), \end{aligned} $

其中

$ \begin{equation*} v\in S_{F, y}=\{v\in L^{1}(J, E)\colon v(t)\in F(t, y(t))\text{ a.e.}~~t\in J\}. \end{equation*} $

利用控制函数定义集值映射 $\Phi (\cdot )\colon C(J, E)\rightarrow 2^{C(J, E)}$, 具有如下形式

$ \begin{aligned} \Phi (y) =&\Big\{ h(t)\in C(J, E)\colon h(t)=(C(t)-S(t)D)(y_{0}-f(y))+S(t)z_{0}\\ &+\displaystyle\int_{0}^{t}C(t-s)Dy(s)ds+\displaystyle\int_{0}^{t}S(t-s)\displaystyle\int_{0}^{s}\beta (s, \tau )v(\tau )d\tau ds\\ &+\displaystyle\int_{0}^{t}S(t-s)Bu_{y}(s)ds, \text{ }v\in S_{F, y}\Big\} . \end{aligned} $

我们将证明 $\Phi (\cdot )$有不动点, 这一不动点是系统(1.1)-(1.2) 的mild解.显然, $ y_{1}-f(y)\in (\Phi y)(m).$我们将证明分为以下几个步骤.

(1) 对于任意的 $y\in C(J, E)$, $\Phi y$是凸的.

$h_{1}$, $h_{2}\in \Phi y$, 存在 $v_{1}$, $v_{2}\in S_{F, y}$, 使得对任意的 $t\in J$, 有

$ \begin{aligned} h_{i}(t) =&~(C(t)-S(t)D)(y_{0}-f(y))+S(t)z_{0}+\displaystyle\int_{0}^{t}C(t-s)Dy(s)ds \\ &+\displaystyle\int_{0}^{t}S(t-s)\displaystyle\int_{0}^{s}\beta (s, \tau )v_{i}(\tau )d\tau ds+\displaystyle\int_{0}^{t}S(t-s)Bu_{y}(s)ds, \text{ }i=1, 2. \end{aligned} $

$0\leqslant \alpha \leqslant 1$, 有

$ \begin{aligned} &\left[\alpha h_{1}+\left( 1-\alpha \right) h_{2}\right] (t) \\ =&~(C(t)-S(t)D)(y_{0}-f(y))+S(t)z_{0}+\displaystyle\int_{0}^{t}C(t-s)Dy(s)ds \\ &+\displaystyle\int_{0}^{t}S(t-s)\displaystyle\int_{0}^{s}\beta (s, \tau )\left[\alpha v_{1}(s)+\left( 1-\alpha \right) v_{2}\left( s\right) \right] d\tau ds +\displaystyle\int_{0}^{t}S(t-s)Bu_{y}(s)ds, \end{aligned} $

因为 $F$有凸值, $S_{F, y}$是凸的, 则 $\alpha h_{1}+\left( 1-\alpha \right) h_{2}\in \Phi y.$

(2) $\Phi $ $C(J, E)$中的有界集映成有界集.

$V_{q}=\left \{ y\in C(J, E)\text{, }\left \Vert y\right \Vert \leqslant q\right \} $, 要证存在一个正数 $\tilde{l}$, 使得对任意的 $y\in V_{q}$, $h\in \Phi y$, 有 $\left \Vert h\right \Vert _{\infty }\leqslant \tilde{l}$.若 $h\in \Phi y$, 则存在 $v\in S_{F, y}$, 使得

$ \begin{aligned} h(t) =&\left( C(t)-S(t)D\right) (y_{0}-f(y))+S(t)z_{0}+\displaystyle\int_{0}^{t}C(t-s)Dy(s)ds \\ &+\displaystyle\int_{0}^{t}S(t-s)\displaystyle\int_{0}^{s}\beta (s, \tau )v(\tau )d\tau ds+\displaystyle\int_{0}^{t}S(t-s)Bu_{y}(s)ds, \text{ }t\in J, \end{aligned} $ (3.2)

对任意的 $t\in J$, 由假设(A1)-(A7) 知

$ \begin{aligned} &\left \Vert h(t)\right \Vert \\ \leqslant &\left \Vert C(t)-S(t)D\right \Vert (\left \Vert y_{0}\right \Vert +\left \Vert f(y)\right \Vert )+\left \Vert S(t)\right \Vert \left \Vert z_{0}\right \Vert \\ &+\left \Vert \displaystyle\int_{0}^{t}C(t-s)Dy(s)ds\right \Vert +\left \Vert \displaystyle\int_{0}^{t}S(t-s)\displaystyle\int_{0}^{s}\beta (s, \tau )v(\tau )d\tau ds\right \Vert \\ &+\left \Vert \displaystyle\int_{0}^{t}S(t-s)Bu_{y}(s)ds\right \Vert \\ \leqslant &\left( M+mM\left \Vert D\right \Vert \right) \left \Vert y_{0}+L\right \Vert +mM\left \Vert z_{0}\right \Vert +M\left \Vert D\right \Vert \displaystyle\int_{0}^{t}\left \Vert y\left( s\right) \right \Vert ds \\ &+m^{2}M^{2}\underset{t\in J}{\sup }\beta (t)\left \Vert p\right \Vert _{L^{1}}\underset{t\in J}{\sup }\psi \left( \left \Vert y(s)\right \Vert \right) +mMM_{1}M_{2}\Sigma _{2} \\ \triangleq &~\tilde{l}, \end{aligned} $

其中

$ \begin{aligned} \Sigma _{2} =&\left \Vert y_{1}\right \Vert +L+\left( M+mM\left \Vert D\right \Vert \right) (\left \Vert y_{0}\right \Vert +L)+mM\left \Vert z_{0}\right \Vert \\ &+M\left \Vert D\right \Vert \displaystyle\int_{0}^{m}\left \Vert y(s)\right \Vert ds+mM\left \Vert p\right \Vert _{L^{1}}\sup \psi \left( \left \Vert y(s)\right \Vert \right). \end{aligned} $

(3) $\Phi $ $C(J, E)$中的有界集映射为等度连续的集合.设 $t_{1}, $ $t_{2}\in J$, $0<t_{1}<t_{2}$, 对任意的 $y\in V_{q}$, $h\in \Phi y$, 使得(3.2) 式成立, 于是有

$ \begin{aligned} &\left \Vert h(t_{2})-h(t_{1})\right \Vert \\ \leqslant &\left \Vert C\left( t_{2}\right) -C\left( t_{1}\right) \right \Vert \left( \left \Vert y_{0}\right \Vert +L\right) +\left \Vert S(t_{2})D-S(t_{1})D\right \Vert (\left \Vert y_{0}\right \Vert +L) \\ &+\left \Vert S(t_{2})-S(t_{1})\right \Vert \left \Vert z_{0}\right \Vert +\displaystyle\int_{0}^{t_{1}}\left \Vert \left[ C(t_{2}-s)-C(t_{1}-s)\right] Dy(s)\right \Vert ds \\ &+M(t_{2}-t_{1})\displaystyle\int_{0}^{m}\left \Vert D\right \Vert \left \Vert y(s)\right \Vert ds \\ &+\displaystyle\int_{0}^{t_{1}}\displaystyle\int_{0}^{s_{1}}\left \Vert \left[S(t_{2}-s)-S(t_{1}-s)% \right] [\beta (s_{2}, \tau )-\beta (s_{1}, \tau )]v(s)\right \Vert d\tau ds \\ &+mM(t_{2}-t_{1})(s_{2}-s_{1})\displaystyle\int_{0}^{m}\left \Vert v(s)\right \Vert ds \\ &+\displaystyle\int_{0}^{t_{1}}\left \Vert \left[ S(t_{2}-s)-S(t_{1}-s)\right] \right \Vert M_{1}M_{2}\Sigma _{2}(s)ds \\ &+\displaystyle\int_{t_{1}}^{t_{2}}\left \Vert S(t_{2}-s)\right \Vert M_{1}M_{2}\Sigma _{2}(s)ds, \end{aligned} $

$t_{2}\rightarrow t_{1}$时, 上述不等式右端趋于零, 因而 $\Phi (V_{q})$是等度连续的.

(4) $\Phi $把有界集映射为 $E$中的相对紧集.固定 $0<t\leqslant m$, 并且实数 $ \varepsilon $满足 $0<\varepsilon <t$, 对于任意的 $y\in V_{q}$, 定义

$ \begin{aligned} h_{\varepsilon }(t) =&~(C(t)-S(t)D)(y_{0}-f(y))+S(t)z_{0}+\displaystyle\int_{0}^{t-\varepsilon }C(t-s)Dy(s)ds \\ &+\displaystyle\int_{0}^{t-\varepsilon }S(t-s)\displaystyle\int_{0}^{s-\varepsilon }\beta (s, \tau )v(\tau )d\tau ds+\displaystyle\int_{0}^{t-\varepsilon }S(t-s)Bu_{y}(s)ds. \end{aligned} $

因为 $C(t)$ $S(t)$是紧算子, 集合 $Y_{\varepsilon }(t)=\left \{ h_{\varepsilon }(t):h_{\varepsilon }(t)\in \Phi y\right \} $是相对紧集.对于每一个 $h\in \Phi y$, 当 $\varepsilon \rightarrow 0^{+}$时, $\left \Vert h(t)-h_{\varepsilon }(t)\right \Vert \rightarrow 0, $因此存在相对紧集任意逼近集合 $\left \{ h(t):h\in \Phi y\right \} .$从而, $\left \{ h(t):h\in \Phi y\right \} $是相对紧的.因此算子 $ \Phi $是全连续的集值映射.

(5) $\Phi (\cdot )$有闭图象.令 $y_{n}\rightarrow y_{\ast }$, $h_{n}\in \Phi y_{n}$, 且 $h_{n}\rightarrow h_{\ast }$, 要证 $h_{\ast }\in \Phi y_{\ast }.$事实上, $h_{n}\in \Phi y_{n}$, 则存在 $v_{n}\in S_{F, y_{n}}$, 使得

$ \begin{aligned} h_{n}(t) =&~(C(t)-S(t)D)(y_{0}-f(y_{n}))+S(t)z_{0}+\displaystyle\int_{0}^{t}C(t-s)Dy_{n}(s)ds \\ &+\displaystyle\int_{0}^{t}S(t-s)\displaystyle\int_{0}^{s}\beta (s, \tau )v_{n}(\tau )d\tau ds+\displaystyle\int_{0}^{t}S(t-s)Bu_{y_{n}}(s)ds. \end{aligned} $

需要证明存在 $v_{\ast }\in S_{F, y_{\ast }}$使得

$ \begin{aligned} h_{\ast }(t) =&~(C(t)-S(t)D)(y_{0}-f(y_{\ast }))+S(t)z_{0}+\displaystyle\int_{0}^{t}C(t-s)Dy_{\ast }(s)ds \\ &+\displaystyle\int_{0}^{t}S(t-s)\displaystyle\int_{0}^{s}\beta (s, \tau )v_{\ast }(\tau )d\tau ds+\displaystyle\int_{0}^{t}S(t-s)Bu_{y\ast }(s)ds. \end{aligned} $

$ \begin{aligned} \bar{u}_{y}(t)=&~W^{-1}\Big[y_{1}-f(y)-(C(m)-S(m)D)\left( y_{0}-f(y)\right)\\ &-S(m)z_{0}-\displaystyle\int_{0}^{m}C(m-s)Dy(s)ds\Big] \left( t\right). \end{aligned} $

$f, W^{-1}$ $D$是连续的, 则 $\overline{u}_{y_{n}}\left( t\right) \rightarrow \overline{u}_{y_{\ast }}\left( t\right), ~~\forall \text{ }t\in J.$从而当 $n\rightarrow \infty $时,

$ \begin{aligned} &\Big \Vert \Big[h_{n}(t)-S(t)z_{0}-(C(t)-S(t)D)(y_{0}-f(y_{n})) \\ &-\displaystyle\int_{0}^{t}C(t-s)Dy_{n}(s)ds-\displaystyle\int_{0}^{t}S(t-s)B\bar{u}_{y_{n}}(s)ds\Big]\\ &-\Big[h_{\ast }(t)-S(t)z_{0}-(C(t)-S(t)D)(y_{0}-f(y_{\ast }))\\ &-\displaystyle\int_{0}^{t}C(t-s)Dy_{\ast }(s)ds-\displaystyle\int_{0}^{t}S(t-s)B\bar{u}% _{y\ast }(s)ds\Big] \Big \Vert \rightarrow ~0. \end{aligned} $

定义线性连续算子 $\Gamma \colon L^{1}(J, E)\rightarrow C(J, E), $

$ \begin{aligned} v \rightarrow ~\Gamma (v)(t)=&\displaystyle\int_{0}^{t}S(t-s)\left[ \displaystyle\int_{0}^{s}\beta (s, \tau )v(\tau )d\tau \right. \\ &\left.-BW^{-1}\left( \displaystyle\int_{0}^{m}S(m-s)\displaystyle\int_{0}^{s}\beta (s, \tau )v(\tau )d\tau \right) (s)\right] ds, \end{aligned} $

并且

$ \begin{equation*} \left \Vert \Gamma \left( v\right) \right \Vert _{\infty }\leqslant mM% \underset{t\in J}{\sup }\beta (t)\left( 1+mMM_{1}M_{2}\right) \left \Vert v\right \Vert _{L^{1}}. \end{equation*} $

由引理2.1知 $\Gamma \circ S_{F, y}$是闭图算子, 并且有

$ \begin{aligned} &h_{n}(t)-(C(t)-S(t)D)(y_{0}-f(y_{n}))-S(t)z_{0} \\ &-\displaystyle\int_{0}^{t}C(t-s)Dy_{n}(s)ds-\displaystyle\int_{0}^{t}S(t-s)B\bar{u}_{y_{n}}(s)ds \in ~\Gamma \left( S_{F, y_{n}}\right) . \end{aligned} $

因为 $y_{n}\rightarrow y_{\ast }$, 所以

$ \begin{aligned} &h_{\ast }(t)-(C(t)-S(t)D)(y_{0}-f(y_{\ast }))-S(t)z_{0} \\ &-\displaystyle\int_{0}^{t}C(t-s)Dy_{\ast }(s)ds-\displaystyle\int_{0}^{t}S(t-s)B\bar{u}_{y_{\ast }}(s)ds \\ =&\displaystyle\int_{0}^{t}S(t-s)\left[ \displaystyle\int_{0}^{s}\beta (s, \tau )v_{\ast }(\tau )d\tau -BW^{-1}\left( \displaystyle\int_{0}^{m}S(m-s)\displaystyle\int_{0}^{s}\beta (s, \tau )v(\tau )d\tau \right) \left( s\right) \right] ds, \end{aligned} $

因此 $\Phi $是具有闭凸值且上半连续的集值映射.

由引理2.2知, 为了证明 $\Phi $有不动点, 还需要下面一个步骤.

令对于任意的 $\lambda \in (0, 1)$, $ y\in \lambda \Phi (y)$ $t\in J$, 存在 $v\in S_{F, y}$使得

$ \begin{aligned} y(t) =&~\lambda \left( C(t)-S(t)D\right) (y_{0}-f(y))+\lambda S(t)z_{0}+\lambda \displaystyle\int_{0}^{t}C(t-s)Dy(s)ds \\ &+\lambda \displaystyle\int_{0}^{t}S(t-s)\displaystyle\int_{0}^{s}\beta (s, \tau )v(\tau )d\tau ds+\lambda \displaystyle\int_{0}^{t}S(t-s)Bu_{y}(s)ds, \end{aligned} $

对任意的 $t\in J$, 有

$ \begin{aligned} \left \Vert y(t)\right \Vert \leqslant &\left( M+mM\left \Vert D\right \Vert \right) (\left \Vert y_{0}\right \Vert +L)+mM\left \Vert z_{0}\right \Vert \\ &+M\left \Vert D\right \Vert \displaystyle\int_{0}^{t}\left \Vert y(s)\right \Vert ds+mM\left \Vert \displaystyle\int_{0}^{t}\displaystyle\int_{0}^{s}\beta (s, \tau )v(\tau )d\tau ds\right \Vert \\ &+mMM_{1}M_{2}\Big[\left \Vert y_{1}\right \Vert +L+\left( M+mM\left \Vert D\right \Vert \right) (\left \Vert y_{0}\right \Vert +L)+mM\left \Vert z_{0}\right \Vert \\ &+M\left \Vert D\right \Vert \displaystyle\int_{0}^{m}\left \Vert y(s)\right \Vert ds+mM\displaystyle\int_{0}^{m}\displaystyle\int_{0}^{s}\beta (s, \tau )v(\tau )d\tau ds\Big] \\ \leqslant &~N_{1}+N_{2}\Big[\left \Vert D\right \Vert \left \Vert y(s)\right \Vert +mM\underset{t\in J}{\sup }\beta (t)\displaystyle\int_{0}^{m}p(s)\psi (\left \Vert y(s)\right \Vert )ds\Big]. \end{aligned} $

因此

$ \begin{equation*} \frac{\left \Vert y\right \Vert }{N_{1}+N_{2}(\left \Vert D\right \Vert \left \Vert y\right \Vert +mM\underset{t\in J}{\sup }\beta (t)\displaystyle\int_{0}^{m}p(s)\psi (||y||)ds)}\leqslant 1. \end{equation*} $

由(3.1) 式知存在 $\Phi $, 使得 $\left \Vert y\right \Vert \neq \theta $.设集合 $U=\left \{ y\in C(J, E), \left \Vert y\right \Vert <\theta \right \}.$ $U$中, 当 $ \lambda \in (0, 1)$时, 不存在 $y\in \partial U$满足 $y\in \lambda \Phi (y).$因此由引理2.2知 $\Phi $有不动点, 则系统(1.1)-(1.2) 是可控的.

参考文献
[1] Balachandran K, Park D.G, Manimegalai P. Controllability of second-order integrodifierential evolution systems in Banach spaces[J]. Comp. Math. Appl., 2005, 49: 1623–1642. DOI:10.1016/j.camwa.2005.03.001
[2] Benchohra M, Ntouyas S K. Existence results of neutral functional difierential and integrodifierentail inclusions in Banach spaces[J]. Elec. J. Difi. Equ., 2000, 20: 1–15.
[3] Chang Y K, Li W T. Controllability of second-order difierential and inegrodifierential inclusions in Banach spaces[J]. J. Optim. The. Appl., 2006, 129(1): 77–87. DOI:10.1007/s10957-006-9044-5
[4] Benchohra M, Gatsori E P, Ntouyas S K. Nonlocal quasilinear damped difierential inclusions[J]. Elec. J. Difi. Equ., 2002, 7: 1–14.
[5] Chalishajar D N. Controllability of damped second order semi-linear neutral functional difierential inclusions in Banach spaces[J]. Auto. Contr. Syst. J., 2009, 2: 1–7.
[6] Lasota A, Opial Z. An application of the Kakutani-Ky-Fan theorem in the theory of ordinary difierential equations[J]. Bull. Acad. Polon. Sci., 1965, 13: 781–786.
[7] Dugundij J, Grans A. Fixed point theory[M]. Warsaw: Mongrafle Mat. PWN, 1982.
[8] Lin Q, Yu Y. Interval oscillation criteria for emden-fowler equation with a damping term[J]. J. Math., 2012, 32(4): 716–722.
[9] Jiang L, Ding W. Periodic solutions of damped impulse systems[J]. J. Math., 2016, 36(5): 920–928.