\documentclass[reqno]{amsart} \usepackage{hyperref} \AtBeginDocument{{\noindent\small {\em Electronic Journal of Differential Equations}, Vol. 2005(2005), No. 103, pp. 1--8.\newline ISSN: 1072-6691. URL: http://ejde.math.txstate.edu or http://ejde.math.unt.edu \newline ftp ejde.math.txstate.edu (login: ftp)} \thanks{\copyright 2005 Texas State University - San Marcos.} \vspace{9mm}} \begin{document} \title[\hfilneg EJDE-2005/103\hfil Periodic trajectories] {Periodic trajectories for evolution equations in Banach spaces} \author[M. D. Voisei\hfil EJDE-2005/103\hfilneg] {Mircea D. Voisei} \address{Mircea D. Voisei \hfill\break Department of Mathematics, MAGC 3.734, The University of Texas - Pan American, Edinburg, TX 78539, USA} \email{mvoisei@utpa.edu} \date{} \thanks{Submitted August 3, 2005. Published September 28, 2005.} \subjclass[2000]{47J35, 34C25, 35K55} \keywords{Periodic solution, evolution equation of parabolic type} \begin{abstract} The existence of periodic solutions for the evolution equation $$ y'(t)+Ay(t)\ni F(t,y(t)) $$ is investigated under considerably simple assumptions on $A$ and $F$. Here $X$ is a Banach space, the operator $A$ is $m$-accretive, $-A$ generates a compact semigroup, and $F$ is a Carath\'{e}odory mapping. Two examples concerning nonlinear parabolic equations are presented. \end{abstract} \maketitle \numberwithin{equation}{section} \newtheorem{theorem}{Theorem}[section] \newtheorem{lemma}[theorem]{Lemma} \newtheorem{remark}[theorem]{Remark} \section{Introduction} Consider the nonlinear evolution equation \begin{equation} \label{e1.1} y'(t)+Ay(t)\ni F(t,y(t)), \quad 0\leq t\leq T, \end{equation} where $(X,\|\cdot \|)$ is a Banach space, $A:D(A)\subset X\to 2^{X}$ is $m$-accretive such that $-A$ generates a compact semigroup, $F:[0,T]\times \overline{D(A)}\to X$ is a Carath\'{e}odory mapping, i.e., $F(\cdot ,x):[0,T]\to X$ is strongly measurable for every $x\in \overline{D(A)}$ and $F(t,\cdot ):\overline{D(A)}\to X$ is continuous for almost every $t\in [0,T]$. The aim of this note is to investigate the existence of a mild periodic solution $y\in C([0,T];X)$, $y(0)=y(T)$ for \eqref{e1.1} in general Banach spaces. Our main result is the following. \begin{theorem} \label{thm1.1} Assume that $A$ is $m$-accretive, $-A$ generates a compact semigroup, $F$ is Carath\'{e}odory with \begin{equation} \label{e1.2} \int_{0}^{T} \sup_{x\in \overline{D(A)},\|x\|\leq r} \|F(t,x)\|dt<\infty, \end{equation} for every $r>0$, and there exist $R>0$, $b,c\in L^{1}(0,T)$, $c(t)\geq 0$, $t\in (0,T)$, $c\neq 0$, such that \begin{equation} \label{e1.3} [x,y-F(t,x)]_{+}\geq c(t)\|x\|+b(t), \quad t\in (0,T),\; x\in D(A), \; \|x\|\geq R,\; y\in Ax, \end{equation} where $[x,v]_{+}=\lim\limits_{h\downarrow 0}\frac 1h (\|x+hv\|-\|x\|)$, $x,v\in X$. Then \eqref{e1.1} admits at least one mild $T$-periodic solution. \end{theorem} This periodic problem has been intensively studied in the literature \cite{c1,h1,s1,s2,v1,v2}. The most common argument used is to apply a fixed point theorem for a suitable Poincar\'{e} operator. Generally, Schauder's fixed point theorem produced remarkable results for the general Banach space setting (Vrabie \cite{v2}, Shioji \cite{s1}). For example the main result in Shioji \cite{s1} has a statement similar to Theorem \ref{thm1.1}. In addition $X$ is separable, $\overline{D(A)}$ is convex, and condition \eqref{e1.3} is strengthened to $c(t)=c>0, $ $t\in [0,T]$. The earlier result of Vrabie \cite{v2} studies \eqref{e1.1} under the assumptions that $A-aI$ is $m$-accretive for some $a>0$ and the growth condition \eqref{e1.2} is replaced by $$ \limsup_{r\to \infty } \frac{1}{r}\sup \{\|F(t,x)\|: t\geq 0,x\in \overline{D(A)},\|x\|\leq r\}=m0$ such that $\{x\in X;$ $\|x\|=r\}\cap D(A)$ is nonempty and $$ \langle x,y-F(t,x)\rangle \geq 0, \quad \mbox{for every }x\in D(A),\; \|x\|=r,\; y\in Ax,\;0\leq t\leq T, $$ where $\langle \cdot ,\cdot \rangle $ stands for the inner product of $X$. Also, in \cite{v1} the case $m=a\geq 0$ is studied in general Banach spaces under the condition: there exists $r>0$ such that $$ [x,y-F(t,x)]_{+}\geq 0, \quad \mbox{for } 0\leq t\leq T,\; x\in D(A),\; y\in Ax,\; \|x\|\geq r. $$ Most of these new assumptions are needed for the invariance of a closed convex ball in the Schauder fixed point theorem argument. The method we use here relies on the Leray-Schauder topological degree applied for the Green operator and it is neither concerned with the initial value problem for \eqref{e1.1} nor involves the Poincar\'{e} operator. That is why, the convexity of $\overline{D(A)}$ or the strong accretivity of $A$ are no longer needed. Condition \eqref{e1.3} ensures some ``a priori'' estimates for the periodic solutions of \eqref{e1.1} and that the ``periodic solution'' operator $% P_{A}:L^{1}(0,T;X)\to C([0,T];X)$, which associates to $g\in L^{1}(0,T;X)$ all solutions $y\in P_{A}g$ of the periodic problem $y'(t)+Ay(t)\ni g(t)$, $0\leq t\leq T$, $y(0)=y(T)$, is compact. Next section is devoted to preliminaries and main notations. Section 3 contains the proof of our main result. This paper concludes with section 4 where two examples concerning nonlinear evolution equation of parabolic type are presented. \section{Preliminaries} For notions such as $m$-accretive operator, mild solution, or compact semigroup the reader is referred to Vrabie \cite{v3} and the references therein. We suggest Lloyd \cite{l1} for the theory and notations of the topological degree. Let $(X,\Vert \cdot \Vert )$ be a real Banach space and $A:D(A)\subset X\to 2^{X}$. If $A-\omega I$ is $m$-accretive for some $\omega \in \mathbb{R} $ then for each $(\xi ,f)\in \overline{D(A)}\times L^{1}(0,T;X)$ the initial value problem \begin{equation} \label{e2.1} y'(t)+Ay(t)\ni f(t), \quad 0\leq t\leq T, \; y(0)=\xi , \end{equation} has a unique mild solution denoted by $M(\xi ,f)$. For $y_{i}=M(\xi _{i},f_{i})$, where $(\xi _{i},f_{i})\in \overline{D(A)}\times L^{1}(0,T;X)$, $i=1,2$ we have the mild solution inequality \begin{equation} \label{e2.2} e^{\omega t}\Vert y_{1}(t)-y_{2}(t)\Vert \le e^{\omega s}\Vert y_{1}(s)-y_{2}(s)\Vert +\int_{s}^{t}e^{\omega \tau }\Vert f_{1}(\tau )-f_{2}(\tau )\Vert d\tau , \end{equation} for $0\le s\le t\le T$. The norm of a function $f$ in $L^{p}(0,T;X)$, $1\leq p<\infty $, is denoted by $$ \|f\|_{L^{p}}=(\int_{0}^{T}\|f(t)\|^{p}dt)^{1/p}. $$ We use the norm $\|y\|_{\infty }=\underset{t\in [0,T]}{\sup }\|y(t)\|$, for $y\in C([0,T];X)$. A family $\mathcal{G}\subset L^{1}(0,T;X)$ is called uniformly integrable if for each $\varepsilon >0$ there exists $\delta (\varepsilon )>0$ such that for every measurable set $E$ in $[0,T]$ whose Lebesgue measure is less than $\delta (\varepsilon )$ we have $\int_{E}\Vert f(t)\Vert dt<\varepsilon $, uniformly for $f\in \mathcal{G}$. Every bounded subset of $L^{p}(0,T;X)$, $10$, and $-A$ generates a compact semigroup then $P_{A}(\mathcal{G})$ is relatively compact in $C([0,T];X)$. \end{theorem} \begin{proof} The mild solution inequality and the periodicity offer us the estimate \begin{equation} \label{e2.3} \|y\|_{\infty }\leq \frac{e^{2\omega T}}{e^{\omega T}-1} \|f\|_{L^{1}},\quad\mbox{for every }y\in P_{A}f\,, \end{equation} where, without loss of generality, we assume that $0\in A0$. Since $\mathcal{G}$ is bounded, this shows that the set of initial-final data $B=\{y(0)$; $y\in P_{A}(\mathcal{G})\}$ is bounded in $X$. According to Theorem \ref{thmVrabie}, this implies that $B$ is relatively compact in $X$ and $P_{A}(\mathcal{G})$ is relatively compact in $C([0,T];X)$. \end{proof} \section{Proof of Theorem \ref{thm1.1}} In the sequel we assume that all the assumptions in Theorem \ref{thm1.1} hold. Without loss of generality we may assume that, in \eqref{e1.3}, $R>\|b\|_{L^{1}}/\|c\|_{L^{1}}$. Let $K>C:=\|b\|_{L^{1}}+R+2$ and $\rho \in C^{\infty }(\mathbb{R})$ such that $0\leq \rho \leq 1$, $\rho (u)=1$ for $|u|\leq K$, $\rho (u)=0$ for $|u|\geq K+1$. \begin{lemma} \label{lem3.1} Let $y\in C([0,T];X)$ be a mild $T$-periodic solution of \begin{equation} \label{e3.1} y'(t)+Ay(t)\ni \rho (\|y(t)\|)F(t,y(t)), \quad t\in [0,T]. \end{equation} Then $\|y\|_{\infty }\leq C$ or $\|y\|_{\infty }\geq K$. \end{lemma} \begin{proof} Assume by contradiction that $C<\|y\|_{\infty}0$ there exist \begin{itemize} \item[(i)] $0=t_{0}R$. For $\varepsilon <\min \{K-\|y\|_{\infty },\inf\limits_{[0,T]}\|y\|-R\}$, we have $R\leq \|y_{k}\|\leq K$, $k=1,\dots ,n$. From \eqref{e1.3} and (iii) we find \begin{equation} \label{e3.2} [y_{k},f_{k}-\frac{y_{k}-y_{k-1}}{t_{k}-t_{k-1} }-F(t,y_{k})]_{+}\geq c(t)\|y_{k}\|+b(t), \quad\mbox{a.e. }t\in [0,T]. \end{equation} Integrate on $[t_{k-1},t_{k}]$ and add from $k=1$ to $n$, to obtain \begin{equation} \label{e3.3} \begin{aligned} &\|y_{n}\|+\sum_{k=1}^n \|y_{k}\|\int_{t_{k-1}}^{t_{k}}c(t)dt\\ &\leq \|y_{0}\|+\sum_{k=1}^n \int_{t_{k-1}}^{t_{k}}\|f_{k}-F(t,y_{k})\| +\int_{0}^{t_{n}}|b(t)|dt \\ &\leq \|y_{0}\|+\|b\|_{L^{1}}+\sum_{k=1}^{n}\int_{t_{k-1}}^{t_{k}}\Vert \rho (\|y(t)\|)F(t,y(t))-f_{k}\Vert dt\\ &\quad +\int_{0}^{T}\|F(t,y(t))-F_{\varepsilon }(t)\|dt, \end{aligned} \end{equation} where $F_{\varepsilon }(t)=F(t,y_{k})$ if $t\in [t_{k-1},t_{k})$, $k=1,\dots ,n$, $F_{\varepsilon }(t)=F(t,y(t))$, $t\in [t_{n},T]$. According to (ii), this yields \begin{equation} \label{e3.4} \|y_{n}\|+R\int_{0}^{t_{n}}c(t)dt\leq \|y_{0}\|+\|b\|_{L^{1}}+\varepsilon +\int_{0}^{T}\|F(t,y(t))-F_{\varepsilon }(t)\|dt. \end{equation} Since $F$ is Carath\'{e}odory, from \eqref{e1.2} and the Lebesgue dominated convergence theorem, we have $\lim\limits_{\varepsilon \to 0} \int_{0}^{T}\|F(t,y(t))-F_{\varepsilon }(t)\|dt=0$. Let $\varepsilon \to 0$, $c\to T$ in \eqref{e3.4}. We find $R\leq \|b\|_{L^{1}}/\|c\|_{L^{1}}$ which is a contradiction. Therefore, $\inf\limits_{[0,T]}\|y\|\leq R$. Eventually shifting the time, we may assume without loss of generality, that there exists $t_{+}\in [0,T]$, such that $\|y(0)\|=R+1$, $\|y(t_{+})\|=\|y\|_{\infty }$, and $\|y(t)\|\geq R+1$, for every $t\in [0,t_{+}]$. By the same argument used above we obtain $\|y\|_{\infty }\leq R+1+\|b\|_{L^{1}}\leq C$ which is a contradiction. The proof is complete. \end{proof} For $0\leq \lambda \leq 1$, define the operators $L_{\lambda}:C([0,T];X)\to C([0,T];X)$, $$ L_{\lambda }v=P_{A+\omega I}(\lambda \rho (\|v\|)F(\cdot ,v)+\lambda\omega v),\quad v\in C([0,T];X), $$ i.e., $y=L_{\lambda }v$ is the unique $T$-periodic solution of \begin{equation} \label{e3.5} y'(t)+Ay(t)+\omega y(t)\ni \lambda \rho (\|v(t)\|)F(t,v(t))+\lambda \omega v(t),\quad 0\leq t\leq T. \end{equation} where $\omega >0$ is specified below. Note that $L_{0}=0$ and that Lemma \ref{lem3.1} contains an ``a priori'' estimate for the fixed points of $L_{1}$. Similarly, we provide an ``a priori'' estimate for the fixed points of $L_{\lambda }$, $0<\lambda <1$. \begin{lemma} \label{lem3.2} For $\omega >0$ big enough, every $T$-periodic solution $y\in C([0,T];X)$, of \begin{equation} \label{e3.6} y'(t)+Ay(t)+\omega (1-\lambda )y(t)\ni \lambda \rho (\|y(t)\|)F(t,y(t)),\quad 0\leq t\leq T,\; 0<\lambda <1, \end{equation} satisfies $\|y\|_{\infty }\leq C$ or $\|y\|_{\infty }\geq K$. \end{lemma} \begin{proof} Consider $a_{K}(t)=\sup\{ \|F(t,x)\|;\; x\in \overline{D(A)},\|x\|\leq K+1\}$, $t\in [0,T]$. \noindent According to \eqref{e1.2} $a_{K}\in L^{1}(0,T) $ and $\rho (\|x\|)\|F(t,x)\|\leq a_{K}(t)$, a.e. $t\in [0,T]$, $x\in \overline{D(A)}$. Assume by contradiction that $C<\|y\|_{\infty }R$. Reasoning as above we find \begin{equation} \label{e3.7} \omega (1-\lambda )RT+R\|c\|_{L^{1}}\leq (1-\lambda )\|a_{K}\|_{L^{1}}+\|b\|_{L^{1}}, \end{equation} which is absurd if $\omega >\|a_{K}\|_{L^{1}}/RT$, since $R\|c\|_{L^{1}}>\|b\|_{L^{1}}$. Therefore, $\inf\limits_{[0,T]}\|y\|\leq R$, and we may assume that there exist $0\leq s_{0}0$, where $\delta _{0}$ depends only on $a_{K}$. If $\omega >\|a_{K}\|_{L^{1}}/R\delta _{0}$, then \eqref{e3.9} provides us with a contradiction. The proof is complete. \end{proof} \begin{lemma} \label{lem3.3} $H(\lambda )=L_{\lambda }$, $0\leq \lambda \leq 1$, defines a homotopy of compact transformations in $C([0,T];X)$. \end{lemma} \begin{proof} Condition \eqref{e1.2} ensures the fact that for every $0\leq \lambda \leq 1$, the operator $C([0,T];X)\ni v\mapsto \lambda \rho (\|v\|)F(t,v)+\lambda \omega v$ transforms bounded subsets of $C([0,T];X)$ into locally integrable subsets of $L^{1}(0,T;X)$. According to Theorem \ref{thm2.1}, this implies that $L_{\lambda }:C([0,T];X)\to C([0,T];X)$ is compact, for every $0\leq \lambda \leq 1$. For $0\leq \lambda$, $\mu \leq 1$ and $v\in C([0,T];X)$, we have, from the mild solution inequality combined with the periodicity, that \begin{equation} \label{e3.10} \|H(\lambda )v-H(\mu )v\|_{\infty }\leq |\lambda -\mu |e^{2\omega T}/(e^{\omega T}-1)\|a_{K}\|_{L^{1}} \end{equation} which shows that $H$ is a homotopy, thereby completing the proof. \end{proof} \begin{proof}[Proof of Theorem \ref{thm1.1}] Pick $r_{0}\in (C,K)$ and let $B=\{v\in C([0,T];X)$; $\|v\|_{\infty }0$, $p>(n-2)/2$ such that $\rho'(r)\geq c|r|^{p-1}$ for every $r\neq 0$, and $f:\mathbb{R}\times \Omega \times \mathbb{R}\to \mathbb{R}$, $f=f(t,x,u)$ is $T$-periodic in $t$, $f(t,x,\cdot )$ is continuous for almost every $(t,x)\in \mathbb{R}\times \Omega $, $f(\cdot ,\cdot,u)$ is measurable for every $u\in \mathbb{R}$, and there exist $M>0$, $a,c\in L^{1}(0,T)$, $c(t)\geq 0$, $t\in (0,T)$, $c\neq 0$, $b,d\in L^{1}((0,T)\times \Omega )$ such that \begin{equation} \label{e4.2} |f(t,x,u)|\leq a(t)|u|+b(t,x), \quad (t,x,u)\in [0,T]\times \Omega \times \mathbb{R}, \end{equation} and \begin{equation} \label{e4.3} \begin{gathered} f(t,x,u)u\leq 0, \quad |f(t,x,u)|\geq c(t)|u|+d(t,x), \\ (t,x)\in [0,T]\times \Omega ,\quad |u|\geq M. \end{gathered} \end{equation} Then \eqref{e4.1} has at least one $T$-periodic solution. \end{theorem} \begin{remark} \label{rmk4.2} \rm For the problem above, Shioji \cite{s2} considers \eqref{e4.2} and the condition $$ \limsup_{|u|\to \infty } \mathop{\rm ess\,sup} _{(t,x)\in \mathbb{R}\times \Omega } \frac{f(t,x,u)}{u}<0, $$ which is equivalent to there exist $\delta ,M>0$, $f(t,x,u)/u\leq -\delta $ for $(t,x,u)\in \mathbb{R}\times \Omega \times \mathbb{R}$ with $|u|\geq M$. This clearly has the form of \eqref{e4.3} with $c(t)=\delta $, $t\in [0,T]$, $d=0$. \end{remark} \begin{proof}[Proof of Theorem \ref{thm4.1}] The operator $A:D(A)=\{u\in L^{1}(\Omega )$; $\rho (u)\in W_{0}^{1,1}(\Omega )$, $\Delta \rho (u)\in L^{1}(\Omega )\}\subset L^{1}(\Omega )\to L^{1}(\Omega )$ given by $Au:=-\Delta \rho (u)$, $u\in D(A)$, is $m$-accretive in $X=L^{1}(\Omega )$, $\overline{D(A)}=L^{1}(\Omega )$, and it generates a compact semigroup (Vrabie \cite{v3}). Define $F:\mathbb{R}\times L^{1}(\Omega )\to L^{1}(\Omega )$ by $F(t,u)(x):=f(t,x,u(x))$, $t\in \mathbb{R}$, $u\in L^{1}(\Omega )$. >From \eqref{e4.2} $F$ is well defined, Carath\'{e}odory, and satisfies \eqref{e1.2}. Since $A0=0$, for every $u\in D(A)$ we have \begin{equation} \label{e4.4} [u,\Delta \rho (u)-F(t,u)]_{+}\geq [u,\Delta \rho (u)]_{+}-[u,F(t,u)]_{+}\geq -[u,F(t,u)]_{+}. \end{equation} Denote by $\{u<0\}=\{x\in \Omega ;\; u(x)<0\}$, $\{u=0\}=\{x\in \Omega ;\; u(x)=0\}$, $\{u>0\}=\{x\in \Omega ;\; u(x)>0\}$, and $|\Omega |$ the measure of $\Omega $. >From \eqref{e4.2} and \eqref{e4.3}, for every $u\in L^{1}(\Omega )$ we obtain \begin{equation} \label{e4.5} \begin{aligned} &[u,F(t,u)]_{+}\\ &=\int_{\{u>0\}} f(t,x,u(x))dx- \int_{\{u<0\}} f(t,x,u(x))dx+\int_{\{u=0\}} f(t,x,0)dx\\ &\leq 2Ma(t)+3\int_{\Omega } b(t,x)dx +\int_{\{u>M\}} f(t,x,u(x))dx-\int_{\{u<-M\}} f(t,x,u(x))dx\\ &\leq -c(t)\|u\|_{L^{1}}+Mc(t)|\Omega |+2Ma(t) +3\int_{\Omega } b(t,x)dx+\int_{\Omega } |d(t,x)|dx. \end{aligned} \end{equation} Relations \eqref{e4.3} and \eqref{e4.4} combined prove that \eqref{e1.3} is fulfilled with $R=0$. According to Theorem \ref{thm1.1}, \eqref{e4.1} has at least one $T$-periodic solution. \end{proof} Next, we consider the periodic problem for the nonlinear heat equation \begin{equation} \label{e4.6} \begin{gathered} \frac{\partial y}{\partial t}-\Delta _{p}(y)=f(t,x,y(t,x)) \quad\mbox{a.e. }(t,x)\in \mathbb{R}_{+}\times \Omega, \\ y=0\quad\mbox{a.e. }(t,x)\in \mathbb{R}_{+}\times \partial\Omega, \\ y(t,x)=y(t+T,x)\quad \mbox{for every }t\geq 0,\mbox{ and a.e. } x\in \Omega , \end{gathered} \end{equation} where $\Omega $ is a bounded domain of $\mathbb{R}^{n}$, with smooth boundary $\partial \Omega $, $p\geq 2$, and $$ \Delta _{p}y=\sum_{i=1}^n \frac{\partial }{\partial x_{i}} (|\frac{\partial y}{\partial x_{i}}|^{p-2} \frac{\partial y}{\partial x_{i}}), $$ is the pseudo-Laplace operator. \begin{theorem} \label{thm4.2} Suppose that $f:\mathbb{R}\times \Omega \times \mathbb{R}\to \mathbb{R}$, $f=f(t,x,u)$ is $T$-periodic in $t$, $f(t,x,\cdot )$ is continuous for almost every $(t,x)\in \mathbb{R}\times \Omega $, $f(\cdot ,\cdot ,u)$ is measurable for every $u\in \mathbb{R}$, and there exist $M>0$, $a,k\in L^{1}(0,T)$, $a,k\geq 0$, $k\neq 0$, $b,d\in L^{1}(0,T;L^2(\Omega ))$ such that \begin{equation} \label{e4.7} |f(t,x,u)|\leq a(t)|u|+b(t,x), \quad (t,x,u)\in [0,T]\times \Omega \times \mathbb{R}, \end{equation} and \begin{equation} \label{e4.8} f(t,x,u)u\leq [c-k(t)]|u|^{p}+d(t,x)|u|, \quad (t,x)\in [0,T]\times \Omega ,\; |u|\geq M, \end{equation} where $c>0$ is such that \begin{equation} \label{e4.9} \int_{\Omega }|\nabla u(x)|^{p}dx\geq c\int_{\Omega }|u(x)|^{p}dx, \quad u\in W_{0}^{1,p}(\Omega ). \end{equation} Then \eqref{e4.6} has at least one $T$-periodic solution. \end{theorem} \begin{remark} \label{rmk4.4} \rm In \cite{v2} the existence of periodic solutions for the nonlinear heat equation governed by the pseudo-Laplace operator is showed under the conditions $$ |f(t,x,u)|\leq a|u|+b,\quad f(t,x,u)u\leq \alpha |u|^{p}+\beta ,\;\; (t,x,u)\in \mathbb{R}\times \Omega \times \mathbb{R}, $$ where $a,b,\alpha ,\beta >0$ and $\alpha 0$, $(t,x)\in \mathbb{R}\times \Omega $, since for $M=\beta /d$ and $|u|\geq M$, $d|u|\geq \beta $. \end{remark} \begin{proof}[Proof of Theorem \ref{thm4.2}] The operator $A:D(A)=\{u\in W_{0}^{1,2}(\Omega )$, $\Delta _{p}u\in L^2(\Omega )\}\subset L^2(\Omega )\to L^2(\Omega )$ given by $Au:=-\Delta_{p}u$, $u\in D(A)$, is maximal monotone in $X=L^2(\Omega )$, and it generates a compact semigroup (Vrabie \cite{v3}). Define $F:\mathbb{R}\times L^2(\Omega )\to L^2(\Omega )$ by $F(t,u)(x):=f(t,x,u(x))$, $t\in \mathbb{R}$, $u\in L^2(\Omega )$. >From \eqref{e4.7} $F$ is well defined, Carath\'{e}odory, and satisfies \eqref{e1.2}. For $u\in D(A)$, according to \eqref{e4.7}, \eqref{e4.8}, we have \begin{equation} \label{e4.10} \begin{aligned} &\|u\|_{L^2}[u,Au-F(t,u)]_{+}\\ &=\langle u,-\Delta _{p}u-F(t,u)\rangle _{L^2} \\ &=\int_{\Omega }|\nabla u(x)|^{p}dx-\int_{\Omega }f(t,x,u(x))u(x)dx\\ &\geq c\int_{\Omega }|u|^{p}dx-\int_{\{|u|\geq M\}} f(t,x,u(x))u(x)dx -\int_{\{|u|0$ and $\|u\|_{L^2}\geq R$, \begin{equation} \label{e4.11} [u,Au-F(t,u)]_{+}\geq c(t)\|u\|_{L^2}-\alpha (t), \end{equation} where $c(t)=\min \{c,k(t)\}R^{p-2}|\Omega |^{(2-p)/2}$, $t\in [0,T] $. Clearly, $c\geq 0$, $c\neq 0$ and all the conditions of Theorem \ref{thm1.1} are fulfilled. This provides us with a $T$-periodic solution of \eqref{e4.6}. \end{proof} \begin{thebibliography}{2} \bibitem{c1} R. Ca\c{s}caval and I. I. Vrabie; \emph{Existence of periodic solutions for a class of nonlinear evolution equations}, Rev. Mat. Univ. Complutense Madrid \textbf{7} (1994), 325-338. \bibitem{h1} N. Hirano and N. Shioji; \emph{Existence of periodic solutions under saddle point type conditions}, J. Nonlinear Convex Anal. \textbf{1} (2000), 115-128. \bibitem{l1} N. G. Lloyd; \emph{Degree theory}, Cambridge University Press (1978). \bibitem{s1} N. Shioji; \emph{Periodic solutions for nonlinear evolution equations in Banach spaces}, Funkc. Ekvacioj, Ser. Int. \textbf{42} (1999), 157-164. \bibitem{s2} N. Shioji; \emph{Existence of periodic solutions for nonlinear evolution equations with pseudo monotone operators}, Proc. Am. Math. Soc. \textbf{125} (1997), 2921-2929. \bibitem{v1} M. D. Voisei; \emph{Existence of periodic solutions for nonlinear equations of evolution in Banach spaces}, Math. Sci. Res. Hot-Line \textbf{3} (1999), 1-7. \bibitem{v2} I. I. Vrabie; \emph{Periodic solutions for nonlinear evolution equations in a Banach space}, Proc. Am. Math. Soc. \textbf{109}, (1990), 653-661. \bibitem{v3} I. I. Vrabie; \emph{Compactness methods for nonlinear evolutions. 2nd ed}, Pitman Monographs and Surveys in Pure and Applied Mathematics, Longman (1995). \end{thebibliography} \end{document}