\documentclass[reqno]{amsart} \usepackage{hyperref} \AtBeginDocument{{\noindent\small \emph{Electronic Journal of Differential Equations}, Vol. 2013 (2013), No. 240, pp. 1--14.\newline ISSN: 1072-6691. URL: http://ejde.math.txstate.edu or http://ejde.math.unt.edu \newline ftp ejde.math.txstate.edu} \thanks{\copyright 2013 Texas State University - San Marcos.} \vspace{9mm}} \begin{document} \title[\hfilneg EJDE-2013/240\hfil Solutions to nonlocal DEs] {Solutions to nonlocal fractional differential equations using a noncompact semigroup} \author[S. Ji, G. Li\hfil EJDE-2013/240\hfilneg] {Shaochun Ji, Gang Li} % in alphabetical order \address{Shaochun Ji (Corresponding author) \newline Faculty of Mathematics and Physics, Huaiyin Institute of Technology, Huaian 223003, China} \email{jiscmath@gmail.com} \address{Gang Li \newline School of Mathematical Science, Yangzhou University, Yangzhou 225002, China} \email{gli@yzu.edu.cn} \thanks{Submitted August 22, 2013. Published October 29, 2013.} \subjclass[2000]{26A33, 34G20} \keywords{Fractional differential equations; nonlocal conditions; \hfill\break\indent measure of noncompactness} \begin{abstract} This article concerns the existence of solutions to nonlocal fractional differential equations in Banach spaces. By using a type of newly-defined measure of noncompactness, we discuss this problem in general Banach spaces without any compactness assumptions to the operator semigroup. Some existence results are obtained when the nonlocal term is compact and when is Lipschitz continuous. \end{abstract} \maketitle \numberwithin{equation}{section} \newtheorem{theorem}{Theorem}[section] \newtheorem{lemma}[theorem]{Lemma} \newtheorem{definition}[theorem]{Definition} \newtheorem{remark}[theorem]{Remark} \allowdisplaybreaks \section{Introduction} In this article, we study the following fractional differential equations with nonlocal conditions \begin{equation} \label{e1.1} \begin{gathered} {}^CD^\alpha u(t) =Au(t)+ f(t,u(t)), \quad t\in J=[0,b], \\ u(0)=g(u), \end{gathered} \end{equation} where $A:D(A)\subseteq X\to X $ is the infinitesimal generator of a strongly continuous semigroup of bounded linear operators $\{T(t)\}_{t\geq 0}$ in a Banach space $X$; ${}^CD^\alpha$ is the Caputo fractional derivative operator of order $\alpha$ with $0<\alpha\leq 1$; $f$ and $g$ are appropriate continuous functions to be specified later. Fractional differential equations arise in many engineering and scientific problems, such as diffusion process, control theory, signal and image processing. Compared with the classical integer-order models, the fractional-order models are more realistic and practical to describe many phenomena in nature (see \cite{bon}). For some recent development on this topic, we refer to the monographs of Kilbas et al. \cite{kil}, Podlubny \cite{pod}, Lakshmikantham et al. \cite{lak1}, and \cite{aga,chang,kum,lak,peng1,yang}. By using some probability density functions, El-Borai \cite{Bor} introduced fundamental solutions of fractional evolution equations in a Banach space. Wang et al. \cite{wang} obtained the existence and uniqueness of $\alpha$-mild solutions by means of fractional calculus and Leray-Schauder fixed point theorem with a compact analytic semigroup. Ren et al. \cite{ren1,ren2} established the existence of mild solutions for a class of semilinear integro-differential equations of fractional order with delays. The study of nonlocal semilinear differential equation in Banach spaces was initiated by Byszewski \cite{bys} and the importance of the problem consists in the fact that it is more general and has better effect than the classical initial conditions $u(0)=u_0$ alone. For example, Deng \cite{den} defined the function $g$ by $$ g(u)=\sum_{j=1}^q c_ju(s_j), $$ where $c_j$ are given constants and $00$ with the lower limit zero for a function $f$ can be defined as $$ I^\alpha f(t)=\frac{1}{\Gamma(\alpha)}\int_0^t(t-s)^{\alpha-1}f(s)\,\mathrm{d}s, \quad t>0, $$ provided the right-hand side is pointwise defined on $[0,\infty)$, where $\Gamma(\cdot)$ is the gamma function. \end{definition} \begin{definition} \label{def2.2} \rm The Riemann-Liouville derivative of order $\alpha$ with the lower limit zero for a function $f$ can be written as $$ {}^{R-L}\!D^\alpha f(t)= \frac{1}{\Gamma(n-\alpha)}\frac{\mathrm{d}^n}{\mathrm{d}t^n} \int_0^t(t-s)^{n-\alpha-1}f(s)\,\mathrm{d}s,\quad t>0, $$ where $\alpha\in (n-1,n)$, $n\in \mathbb{N}$. \end{definition} \begin{definition} \label{def2.3} \rm The Caputo derivative of order $\alpha$ with the lower limit zero for a function $f$ can be written as $$ {}^C D^\alpha f(t)= \frac{1}{\Gamma(n-\alpha)}\int_0^t(t-s)^{n-\alpha-1}f^{(n)}(s)\,\mathrm{d}s , \quad t>0, $$ where $\alpha\in (n-1,n)$, $n\in \mathbb{N}$. \end{definition} If $f$ takes values in a Banach space $X$, the integrals which appear in the above three definitions are taken in Bochner's sense. Especially, when $0<\alpha< 1$, we have $$ {}^C D^\alpha f(t)=\frac{1}{\Gamma(1-\alpha)}\int_0^t \frac{f'(s)}{(t-s)^\alpha}\,\mathrm{d}s. $$ We firstly recall the concept of mild solutions to equation \eqref{e1.1} developed in \cite{Bor,zho}. \begin{definition}\label{def2.4} \rm A function $u\in C(J;X)$ is said to be a mild solution of \eqref{e1.1} if $u$ satisfies $$ u(t)=S_\alpha(t)g(u)+\int_0^t(t-s)^{\alpha-1}T_\alpha(t-s)f(s,u(s))\,\mathrm{d}s, $$ for $t\in J$, where \begin{gather*} S_\alpha(t) = \int_0^\infty \xi_\alpha(\theta)T(t^\alpha\theta)\,\mathrm{d}\theta,\quad T_\alpha(t)=\alpha\int_0^\infty \theta\xi_\alpha(\theta)T(t^\alpha\theta)\,\mathrm{d}\theta, \\ \xi_\alpha(\theta)= \frac{1}{\alpha}\theta^{-1-\frac{1}{\alpha}} \varpi_\alpha(\theta^{-\frac{1}{\alpha}})\geq 0, \\ \varpi_\alpha(\theta)= \frac{1}{\pi}\sum_{n=1}^\infty (-1)^{n-1} \theta^{-\alpha n-1}\frac{\Gamma(n\alpha+1)}{n!}\sin(n\pi \alpha),\quad \theta\in (0,\infty). \end{gather*} Here $\xi_\alpha(\theta)$ is a probability density function defined on $(0,\infty)$ satisfying $$ \int_0^\infty \xi_\alpha(\theta)\,\mathrm{d}\theta=1,\quad \int_0^\infty \theta^v \xi_\alpha(\theta)\,\mathrm{d}\theta =\frac{\Gamma(1+v)}{\Gamma(1+\alpha v)},\quad v\in [0,1]. $$ \end{definition} \begin{lemma}[\cite{wang}] \label{lem2.5} For any fixed $t\geq 0$, the operators $S_\alpha(t)$ and $T_\alpha(t)$ are linear and bounded operators; i.e., for any $x\in X$, $\|S_\alpha(t)x\|\leq M\|x\|$ and $\|T_\alpha(t)x\|\leq \frac{M\alpha}{\Gamma(1+\alpha) }\|x\|$, where $M$ is the constant such that $\|T(t)\|\leq M$ for all $t\in [0,b]$. \end{lemma} Now we give some facts on measure of noncompactness, see Banas and Goebel\cite{ban}. \begin{definition} \label{def2.6} \rm Let $E^+$ be the positive cone of an ordered Banach space $(E,\leq)$. A function $\Phi$ defined on the set of all bounded subsets of the Banach space $X$ with values in $E^+$ is called a measure of noncompactness (in short MNC) on $X$ if $\Phi(\overline{{\rm co}}\Omega)=\Phi(\Omega)$ for all bounded subsets $\Omega \subset X$, where $\overline{{\rm co}}\Omega$ stands for the closed convex hull of $\Omega$. A measure of noncompactness $\Phi$ is said to be: \begin{itemize} \item[(1)] \emph{monotone} if for all bounded subsets $\Omega_1,\Omega_2$ of $X$ we have: $(\Omega_1 \subseteq \Omega_2) \Rightarrow (\Phi(\Omega_1)\leq \Phi(\Omega_2))$; \item[(2)] \emph{nonsingular} if $\Phi(\{a\}\cup \Omega)=\Phi(\Omega)$ for every $a\in X$, $\Omega \subset X$; \item[(3)] \emph{regular} if $\Phi(\Omega)=0$ if and only if $\Omega $ is relatively compact in $X$. \end{itemize} \end{definition} One of the most important examples of MNC is the Hausdorff measure of noncompactness $\beta(\cdot)$ defined by $$ \beta(B)=\inf\{\varepsilon>0: B \text{has a finite $\varepsilon$-net in } X\}, $$ for each bounded subset $B$ in a Banach space $X$. It is well known that the Hausdorff measure of noncompactness $\beta$ enjoys the above properties. \begin{lemma}[\cite{ban}] \label{lem2.7} Let $X$ be a real Banach space and $B,C\subseteq X$ be bounded. Then the following properties are satisfied: \begin{enumerate} \item $B $ is relatively compact if and only if $\beta(B)=0$; \item $\beta(B)=\beta(\overline{B})=\beta(\operatorname{conv} B)$, where $\overline{B}$ and $\operatorname{conv}B$ mean the closure and convex hull of $B$, respectively; \item $\beta(B)\leq \beta(C)$ when $B\subseteq C$; \item $\beta(B+C)\leq \beta(B) + \beta(C)$, where $B + C=\{x+y: x\in B, y\in C\}$; \item $\beta(B\cup C)\leq \max\{\beta(B), \beta(C)\}$; \item $\beta(\lambda B )\leq |\lambda|\beta(B)$ for any $\lambda\in R$; \item If the map $Q:D(Q)\subseteq X \to Z$ is Lipschitz continuous with constant $k$, then $\beta_{Z}(QB)\leq k\beta(B)$ for any bounded subset $B \subseteq D(Q)$, where $Z$ is a Banach space. \item If $\{ W_n\}_{n=1}^\infty$ is a decreasing sequence of bounded closed nonempty subsets of $X$ and $\lim_{n\to \infty}\beta(W_n)=0$, then $\cap_{n=1}^\infty{W_n}$ is nonempty and compact in $X$. \end{enumerate} \end{lemma} We will also use the sequential MNC $\beta_0$ generated by $\beta$, that is, for any bounded subset $B\subset X$, we define $$ \beta_0(B)=\sup\big\{\beta(\{x_n :n\geq 1\}):\{x_n \}_{n=1}^\infty \text{ is a sequence in }B \big\}. $$ It follows that \begin{equation}\label{e2.1} \beta_0(B)\leq \beta(B)\leq 2\beta_0(B). \end{equation} If $X$ is a separable space, we have $\beta_0(B)=\beta(B)$. \begin{lemma}[\cite{kam}]\label{lem2.8} If $\{u_n\}_{n=1}^\infty\subset L^1(J;X)$ satisfies $\|u_n(t)\|\leq \varphi (t)$ a.e. on $[0,b]$ for all $n\geq 1$ with some $\varphi\in L^1(J;\mathbb{R_+})$, then for $t\in [0,b]$, we have $$ \beta\Big(\big\{\int_0^t u_n(s)\,\mathrm{d}s\big\}_{n=1}^\infty \Big) \leq 2\int_0^t \beta(\{u_n(s)\}_{n=1}^\infty ) \,\mathrm{d}s. $$ \end{lemma} \begin{lemma}[\cite{hen}]\label{lem2.9} Suppose $b\geq 0$, $\sigma>0$ and $a(t)$ is nonnegative function locally integrable on $0\leq t0$ and sequences $\{u_n\}\subseteq B$, $\{t_n\},~\{\overline{t}_n\}\subseteq [0,b]$, such that $t_n\to t_0$, $\overline{t}_n\to t_0$ as $n\to \infty$ and $$ \|u_n(t_n)-u_n(\overline{t}_n) \|\geq \varepsilon_0, $$ for all $n\geq 1$. Note that $$ \|u_n(t_n)-u_n(\overline{t}_n) \| \leq \sup\{ \|u(t_n)-u(\overline{t}_n)\|:u\in B\}. $$ We take the upper limit for $n$ and get that $$ \overline{\lim_n} \|u_n(t_n)-u_n(\overline{t}_n) \| \leq \operatorname{mod}_C(B(t_0))\leq \chi_2(B)=0, $$ which gives the contradiction $0<\varepsilon_0\leq 0$. Thus $B\subseteq C(J;X)$ is equicontinuous on $J$. This completes the proof. \end{proof} We will use the following hypotheses: \begin{itemize} \item[(HA)] The operator $A$ generates a strongly continuous semigroup $\{T(t)\}_{t\geq 0}$ in $X$. Moreover, there exists a positive constant $M > 0$ such that $M=\sup_{0\leq t\leq b}\|T(t)\|$ (see Pazy\cite{paz}). \item[(HG1)] $g:C(J;X)\to X$ is continuous and compact. There exists a positive constant $N$ such that $\|g(u)\|\leq N$ for all $u\in C(J;X)$. \item[(HF1)] $f:[0,b]\times X\to X$ is continuous. \item[(HF2)] there exists a constant $L>0$, such that for any bounded set $D\subset X$, $\beta(f(t,D))\leq L\beta(D)$, for a.e. $t\in J$. \end{itemize} The following lemma is useful for our proofs. \begin{lemma}\label{lem3.2} Suppose that the semigroup $\{T(t)\}_{t\geq 0}$ is strongly continuous and hypotheses {\rm (HF1), (HF2)} are satisfied. Then for any bounded set $B\subset C(J;X)$, we have $$ \beta(G_2B(t))\leq \frac{4\alpha ML}{\Gamma(1+\alpha)}\int_0^t (t-s)^{\alpha-1}\beta(B(s))\,\mathrm{d}s , $$ for $t\in [0,b]$. \end{lemma} \begin{proof} For $t\in [0,b]$, due to the inequality \eqref{e2.1}, we obtain that for arbitrary $\varepsilon >0$, there exists a sequence $\{v_k\}_{k=1}^\infty \subset B$ such that \begin{equation}\label{e3.1} \beta(G_2B(t))\leq 2\beta(\{G_2v_k(t)\}_{k=1}^\infty)+\varepsilon. \end{equation} It follows from Lemma \ref{lem2.8} and hypotheses $(Hf_1),~(Hf_2)$ that \begin{align*} % to remove numbering (before each equation) \beta(\{G_2v_k(t)\}_{k=1}^\infty )&\leq 2\int_0^t (t-s)^{\alpha-1}\beta(\{T_\alpha (t-s)f(s,v_k(s))\}_{k=1}^\infty)\,\mathrm{d}s \\ &\leq 2\int_0^t (t-s)^{\alpha-1}\frac{\alpha M}{\Gamma(1+\alpha)}L \beta(\{v_k(s)\}_{k=1}^\infty)\,\mathrm{d}s\\ &\leq \frac{2\alpha ML}{\Gamma(1+\alpha)}\int_0^t (t-s)^{\alpha-1}\beta(B(s))\,\mathrm{d}s. \end{align*} According to {e3.1}, we can derive that $$ \beta(G_2B(t))\leq \frac{4\alpha ML}{\Gamma(1+\alpha)}\int_0^t (t-s)^{\alpha-1}\beta(B(s))\,\mathrm{d}s+\varepsilon. $$ Since the above inequality holds for arbitrary $\varepsilon >0$, it follows that $$ \beta(G_2B(t))\leq \frac{4\alpha ML}{\Gamma(1+\alpha)}\int_0^t (t-s)^{\alpha-1}\beta(B(s))\,\mathrm{d}s. $$ This completes the proof. \end{proof} Now, we give the main existence result of this section. \begin{theorem} \label{thm3.3} Assume that the hypotheses {\rm (HA), (HG1), (HF1), (HF2)} are satisfied. Then the nonlocal fractional differential system \eqref{e1.1} has at least one mild solution on $[0,b]$, provided that there exists a constant $r>0$ such that \begin{equation}\label{e3.2} MN+\frac{M b^\alpha}{\Gamma(1+\alpha)} \sup_{s\in [0,b],u\in W_r}\|f(s,u(s))\| \leq r. \end{equation} \end{theorem} \begin{proof} We shall prove this result by using the Schauder's fixed point theorem. Step 1. We shall prove that $G$ is continuous on $C(J;X)$. Let $\{u_m\}_{m=1}^\infty$ be a sequence in $C(J;X)$ with $\lim_{m\to \infty}u_m=u $ in $C(J;X)$. By the continuity of $f$, we deduce that for each $s\in [0,b]$, $f(s,u_m(s))$ converges to $f(s,u(s))$ in $X$ uniformly for $s\in [0,b]$. And we have \begin{align*} \|Gu_m-Gu\| &\leq \|S_\alpha(t)g(u_m)-g(u)\|\\ &\quad +\int_0^t(t-s)^{\alpha-1}\|T_\alpha(t-s)[f(s,u_m(s))-f(s,u(s))] \|\,\ \mathrm{d}s\\ &\leq M\|g(u_m)-g(u)\|+\frac{M b^\alpha}{\Gamma(1+\alpha)} \sup_{s\in [0,b]}\|f(s,u_m(s))-f(s,u(s))\|. \end{align*} Then by the continuity of $g$, we get $\lim_{m\to \infty}G u_m=Gu$ in $C(J;X)$, which implies that $G$ is continuous on $C(J;X)$. Step 2. We construct a bounded convex and closed set $W\subset C(J;X)$ such that $G$ maps $W$ into itself. Let $W_0=\{u\in C(J;X):\|u(t)\|\leq r, t\in J \}$, where $r$ satisfies the condition \eqref{e3.2}. For any $u\in W_0$, by hypotheses (HG1), (HF1) and \eqref{e3.2}, we have \begin{align*} \|Gu(t)\| &\leq \|S_\alpha(t)g(u)\|+\| \int_0^t(t-s)^{\alpha-1}T_\alpha(t-s)f(s,u(s))\, \mathrm{d}s\\ &\leq MN+\frac{M b^\alpha}{\Gamma(1+\alpha)} \sup_{s\in [0,b],u\in W_0}\|f(s,u(s))\| \leq r, \end{align*} for $t\in [0,b]$, which implies that $GW_0\subseteq W_0$. Define $W_1=\overline{\mathrm{conv}}\{G(W_0),u_0\}$, where $\overline{\mathrm{conv}}$ means the closure of convex hull, $u_0\in W_0 $. Then $W_1\subset W_0$ is nonempty bounded closed and convex. We define $W_n=\overline{\mathrm{conv}}\{G(W_{n-1}),u_0\}$ for $n\geq 1$. It is easy to know that $\{W_n\}_{n=0}^\infty$ is a decreasing sequence of $C(J;X)$. Moreover, set $$ W=\cap_{n=0}^\infty W_n, $$ then $W$ is a nonempty, convex, closed and bounded subset of $C(J;X)$ and $GW\subseteq W$. Step 3. We claim that $W$ is compact in $C(J;X)$ by using the newly-defined MNC $\chi$. As $\{W_n\}$ is a decreasing sequence of $C(J;X)$, then $\{\beta(W_n(t))\}_{n=0}^\infty$ is nonnegative decreasing sequence for any $t\in [0,b]$. From the compactness of $g$ and Lemma \ref{lem3.2}, we get \begin{equation}\label{e3.3} \begin{aligned} \beta(W_{n+1}(t)) &\leq \beta(\{S_\alpha(t)g(u):u\in W_n \}) \\ &\quad +\beta(\{\int_0^t (t-s)^{\alpha-1}T_\alpha(t-s)f(s,u(s))\,\mathrm{d}s: u\in W_n \} ) \\ &\leq \frac{4\alpha ML}{\Gamma(1+\alpha)}\int_0^t (t-s)^{\alpha-1}\beta(W_n(s)) \mathrm{d}s. \end{aligned} \end{equation} Taking $n\to \infty$ to both sides of \eqref{e3.3}, we have \[ \beta(W(t))\leq \frac{4\alpha ML}{\Gamma(1+\alpha)}\int_0^t (t-s)^{\alpha-1}\beta(W(s)) \mathrm{d}s. \] By Lemma \ref{lem2.9} we obtain that $\beta (W(t))=0$ for any $t\in [0,b]$. Then according to the definition of $\chi_1$, we have \begin{equation}\label{e3.4} \chi_1(W_n)\to 0,\quad \text{as }n\to \infty. \end{equation} Next, we will estimate $\chi_2(W_n)$. Fix $t_0\in (0,b)$, $00$ such that $\|g(x)-g(y)\|\leq l_g\|x-y\| $, $x,y\in C([0,b];X)$. \item[(HF3)] $f:[0,b]\times X\to X$ is continuous and compact. \end{itemize} \begin{theorem} \label{thm4.2} Assume that {\rm (HA), (HG2), (HF3)} are satisfied. Then the nonlocal fractional differential system \eqref{e1.1} has at least one mild solution on $[0,b]$, provided that \eqref{e3.2} and \begin{equation}\label{e4.1} Ml_g<1. \end{equation} \end{theorem} \begin{proof} Define the solution operator $G:C(J;X)\to C(J;X)$ by $$ Gu(t)=S_\alpha(t)g(u)+\int_0^t(t-s)^{\alpha-1}T_\alpha(t-s)f(s,u(s))\,\mathrm{d}s $$ with \begin{gather*} G_1u(t)=S_\alpha(t)g(u), \\ G_2u(t)=\int_0^t(t-s)^{\alpha-1}T_\alpha(t-s)f(s,u(s))\,\mathrm{d}s, \end{gather*} for all $t\in J$. From the proof of Theorem \ref{thm3.3}, we have got that the solution operator $G$ is continuous and maps $W_r$ into itself. It remains to show that $G$ is $\beta$-condensing. For $u,v\in W_r$, we have \[ \|G_1u-G_1v\|=\|S_\alpha(t)g(u)-S_\alpha(t)g(v) \|\leq Ml_g\|u-v\|, \] which implies that $G_1$ is Lipschitz continuous with the constant $Ml_g$. From Lemma \ref{lem2.7} (7), we have \begin{equation}\label{e4.2} \beta(G_1W_r)\leq Ml_g\beta(W_r). \end{equation} Next, we shall show that $G_2$ is a compact operator. From the Ascoli-Arzela theorem, we need prove that $G_2W_r$ is equicontinuous and $G_2W_r(t)$ is precompact in $X$ for $t\in [0,b]$. For $u\in W_r$ and $0\leq t_10$, we have \begin{equation}\label{e4.5} \begin{aligned} I_2&\leq \frac{\alpha M}{\Gamma(1+\alpha)}\sup_{t\in J,u\in W_r}f(s,u(s))\cdot \int_0^{t_1} (t_2-s)^{\alpha-1}-(t_1-s)^{\alpha-1}\,\mathrm{d}s \\ &\leq \frac{M}{\Gamma(1+\alpha)}\sup_{t\in J,u\in W_r}f(s,u(s))\cdot(t_2^\alpha-t_1^\alpha). \end{aligned} \end{equation} Since $f$ is compact, then $\|[T_\alpha(t_2-s)-T_\alpha(t_1-s)]f(s,u(s))\|\to 0 $, as $t_1\to t_2$, uniformly for $s\in J$ and $u\in W_r$. This implies that, for any $\varepsilon>0$, there exists $\delta>0$, such that $$ \|[T_\alpha(t_2-s)-T_\alpha(t_1-s)]f(s,u(s)) \|<\varepsilon, $$ for $00$, there exists a sequence $\{v_k\}_{k=1}^\infty\subset W_r$, such that \begin{align*} \beta(G_2W_r(t)) &\leq 2\beta(\{ G_2v_k(t):k\geq 1\})+\varepsilon \\ &\leq 4\int_0^t (t-s)^{\alpha-1}\beta( \{T_\alpha(t-s)f(s,v_k(s)):k\geq 1\})\,\mathrm{d}s+\varepsilon. \end{align*} Noticing that $$ \beta( \{T_\alpha(t-s)f(s,v_k(s)):k\geq 1\})\leq \beta( \{T_\alpha(t-s)f(s,u(s)):t,s\in J, u\in W_r\})=0, $$ we have $\beta(G_2W_r(t))=0 $ for $t\in J$. By the Ascoli-Arzela theorem, we have that $G_2$ is a compact operator, which implies \begin{equation}\label{e4.7} \beta(G_2W_r)=0. \end{equation} So, according to \eqref{e4.2} and \eqref{e4.7}, we can conclude that \begin{equation*} \beta(GW_r)\leq \beta(G_1W_r)+\beta(G_2W_r)\leq Ml_g\beta(W_r). \end{equation*} From the condition $Ml_g<1$, $G$ is $\beta-$condensing in $W_r$. By the Darbo-Sadovskii's fixed point theorem, $G$ has at least a fixed point $u$ in $W_r$, which is just a mild solution of the problem \eqref{e1.1}. The proof is complete. \end{proof} By using Banach contraction principle, we also give the existence theorem when $f$, $g$ are uniformly Lipschitz continuous. We give the following hypothesis on $f$. \begin{itemize} \item[(HF4)] $f:C(J;X)\to X$ is continuous and there exists a constant $l_f>0$, such that $$ \|f(t,x_1)-f(t,x_2)\|\leq l_f\|x_1-x_2\|,\quad x_1,x_2\in X. $$ \end{itemize} \begin{theorem} \label{thm4.3} Assume that the hypotheses {\rm (HA), (HG2), (HF4)} are satisfied. Then the nonlocal fractional system \eqref{e1.1} has a unique mild solution on $[0,b]$, provided that \begin{equation}\label{e4.8} Ml_g+\frac{Mb^\alpha}{\Gamma(1+\alpha)}l_f<1. \end{equation} \end{theorem} \begin{proof} For $u,v\in C(J;x)$, $t\in J$, we have that \begin{align*} \|Gu(t)-Gv(t)\| &\leq \|S_\alpha(t)[g(u)-g(v)]\| \\ &\quad +\int_0^{t}(t-s)^{\alpha-1}\|T_\alpha(t-s)[f(s,u(s))-f(s,v(s))] \| \,\mathrm{d}s\\ &\leq Ml_g\|u-v\|_C +\frac{\alpha M}{\Gamma(1+\alpha)} \int_0^{t}(t-s)^{\alpha-1}\,\mathrm{d}s\cdot l_f\|u-v\|_C\\ &\leq \big(Ml_g+\frac{Mb^\alpha}{\Gamma(1+\alpha)}l_f\big)\|u-v\|_C. \end{align*} Then $$ \|Gu-Gv\|_C \leq \big(Ml_g+\frac{Mb^\alpha}{\Gamma(1+\alpha)}l_f\big) \|u-v\|_C. $$ According to \eqref{e4.8}, we find that $G$ is a contraction operator in $C(J;X)$. Thus $G$ has a unique fixed point $u$, which is the unique mild solution to the problem \eqref{e1.1}. The proof is complete. \end{proof} \subsection*{Conclusions} This article is motivated by some recent papers \cite{mop,zho,peng2}, where some fractional nonlocal differential equations are discussed when $T(t)$ is compact or equicontinuous. Since it is difficult to determine whether an operator semigroup is compact (see Pazy\cite{paz}), we do not assume that $A$ generates a compact semigroup. It allow us to discuss some differential equations which contain a linear operator that generates a noncompact semigroup. We give a simple example. Let $X=L^2(-\infty,+\infty)$. The ordinary differential operator $A=\mathrm{d}/\mathrm{d}x$ with $D(A)=H^1(-\infty,+\infty)$, generates a semigroup $T(t)$ defined by $T(t)u(s)=u(t+s)$, for every $u\in X$. The $C_0$-semigroup $T(t)$ is not compact on $X$. Another motivation of this paper is the control problem of fractional differential system. Exact controllability for fractional order systems have been discussed by many authors. Some controllability results are obtained with the assumptions that the associated semigroup $T(t)$ is compact and the inverse of control operator is bounded. However, Hern\'{a}ndez and O'Regan \cite{her} have pointed out that in this case the application of controllability result is restricted to the finite dimensional space. Here we can also deal with the fractional control problem in the similar way and give a way to remove the compactness assumptions for the nonlocal control problems. \subsection*{Acknowledgments} Research is partially supported by the National Natural Science Foundation of China (11271316, 11101353), the Foundation of Huaiyin Institute of Technology (HGC1229). \begin{thebibliography}{10} \bibitem{aga} R. P. Agarwal, V. Lakshmikantham, J. J. Nieto; \emph{On the concept of solution for fractional differential equations with uncertainty,} Nonlinear Anal. \textbf{72} (2010), 2859-2862. \bibitem{bal1} K. Balachandran, J. Y. Park; \emph{Existence of mild solution of a functional integrodifferential equation with nonlocal condition,} Bull. Korean Math. Soc. \textbf{38} (2001), 175-182. \bibitem{bal2} K. Balachandran, J. Y. Park; \emph{Nonlocal Cauchy problem for abstract fractional semilinear evolution equations,} Nonlinear Anal. \textbf{71} (2009), 4471-4475. \bibitem{ban} J. Banas, K. Goebel; \emph{Measure of Noncompactness in Banach Spaces,} Lect. Notes Pure Appl. Math., vol.60, Marcel Dekker, New York. 1980. \bibitem{bon} B. Bonilla, M. Rivero, L. Rodriguez-Germa, J. J. Trujillo; \emph{Fractional differential equations as alternative models to nonlinear differential equations,} Appl. Math. Comput. \textbf{187} (2007), 79-88. \bibitem{bys} L. Byszewski, V. Lakshmikantham; \emph{Theorem about the existence and uniqueness of solutions of a nonlocal Cauchy problem in a Banach space,} Appl. Anal. \textbf{40} (1990), 11-19. \bibitem{chang} Y. K. Chang, J. J. Nieto; \emph{Some new existence results for fractional differential inclusions with boundary conditions,} Math. Comput. Model. \textbf{49} (2009), 605-609. \bibitem{den} K. Deng; \emph{Exponential decay of solutions of semilinear parabolic equations with nonlocal initial conditions,} J. Math. Anal. Appl. \textbf{179} (1993), 630-637. \bibitem{Bor} M. M. El-Bora; \emph{Some probability densities and fundamental solutions of fractional evolution equations,} Chaos, Solitons and Fractals \textbf{14} (2002), 433-440. \bibitem{fan} Z. Fan, G. Li; \emph{Existence results for semilinear differential equations with nonlocal and impulsive conditions,} J. Funct. Anal. 258 (2010) 1709-1727. \bibitem{fan2} Z. Fan, G. Li; \emph{Existence results for semilinear differential inclusions,} Bull. Austral. Math. Soc, \textbf{76} (2007), 227-241. \bibitem{hen} D. Henry; \emph{Geometric Theory of Semilinear Parabolic Equations,} Lecture Notes in Math., vol. 840, Springer-Verlag, New York, Berlin, 1981. \bibitem{her} E. Hern\'{a}ndez, D. O'Regan; \emph{Controllability of Volterra-Fredholm type systems in Banach space,} J. Franklin Inst. \textbf{346} (2009), 95-101. \bibitem{ji} S. Ji, G. Li; \emph{A unified approach to nonlocal impulsive differential equations with the measure of noncompactness,} Adv. Differ. Equ. \textbf{Vol.2012} (2012), No. 182, 1-14. \bibitem{kam} M. Kamenskii, V. Obukhovskii, P. Zecca; \emph{Condensing Multivalued Maps and Semilinear Differential Inclusions in Banach Spaces,} De Gruyter, Berlin. 2001. \bibitem{kil} A. A. Kilbas, H. M. Srivastava, J.J . Trujillo; \emph{Theory and Applications of Fractional Differential Equations,} Elsevier, Amsterdam, 2006. \bibitem{kum} S. Kumar, N. Sukavanam; \emph{Approximate controllability of fractional order semilinear systems with bounded delay,} J. Diff. Equ. \textbf{252} (2012), 6163-6174. \bibitem{lak1} V. Lakshmikantham, S. Leela, J. Vasundhara Devi; \emph{Theory of Fractional Dynamic Systems,} Cambridge Academic Publishers, Cambridge, 2009. \bibitem{lak} V. Lakshmikantham, A. S. Vatsala; \emph{Basic theory of fractional differential equations,} Nonlinear Anal. \textbf{69} (2008), 2677-2682. \bibitem{peng1} K. Li, J. Peng, J. Jia; \emph{Cauchy problems for fractional differential equations with Riemann-Liouville fractional derivatives,} J. Funct. Anal. \textbf{263} (2012), 476-510. \bibitem{peng2} K. Li, J. Peng, J. Gao; \emph{Nonlocal fractional semilinear differential equations in separable Banach spaces,} Electron. J. Diff. Eqns. Vol. \textbf{2013} (2013), No.7, 1-7. \bibitem{mop} G. M. Mophou, G. M. N'Gu\'{e}r\'{e}kata; \emph{Existence of mild solution for some fractional differential equations with nonlocal conditions,} Semigroup Forum \textbf{79} (2009), 322-335. \bibitem{nto} S. Ntouyas, P. Tsamotas; \emph{Global existence for semilinear evolution equations with nonlocal conditions,} J. Math. Anal. Appl. \textbf{210} (1997), 679-687. \bibitem{paz} A. Pazy; \emph{Semigroups of Linear Operators and Applications to Partial Differential Equations,} Springer-Verlag, New York, 1983. \bibitem{pod} I. Podlubny; \emph{Fractional Differential Equations,} Academic Press, San Diego, 1999. \bibitem{ren1} L. Hu, Y. Ren, R. Sakthivel; \emph{Existence and uniqueness of mild solutions for semilinear integro-differential equations of fractional order with nonlocal conditions,} Semigroup Forum \textbf{79} (2009), 507-514. \bibitem{ren2} Y. Ren, Y. Qin, R. Sakthivel; \emph{Existence results for fractional order semilinear integro-differential evolution equations with infinite delay,} Integr. Equ. Oper. Theory \textbf{67} (2010), 33-49. \bibitem{wang} J. Wang, Y. Zhou; \emph{A class of fractional evolution equations and optimal controls,} Nonlinear Anal. RWA \textbf{12} (2011), 262-272. \bibitem{xue} X. Xue; \emph{Nonlocal nonlinear differential equations with a measure of noncompactness in Banach spaces,} Nonlinear Anal. \textbf{70} (2009), 2593-2601. \bibitem{yang} H. Yang; \emph{Existence of mild solutions for a class of fractional evolution equations with compact analytic semigroup,} Abstr. Appl. Anal. \textbf{Vol. 2012} (2012), Article ID 903518, 15 pages. \bibitem{zho} Y. Zhou, F. Jiao; \emph{Nonlocal Cauchy problem for fractional evolution equations,} Nonlinear Anal. RWA \textbf{11} (2010), 4465-4475. \bibitem{zhu} L. Zhu, G. Li; \emph{On a nonlocal problem for semilinear differential equations with upper semicontinuous nonlinearities in general Banach spaces,} J. Math. Anal. Appl. \textbf{341} (2008), 660-675. \end{thebibliography} \end{document}