\documentclass[reqno]{amsart} \usepackage{hyperref} \AtBeginDocument{{\noindent\small \emph{Electronic Journal of Differential Equations}, Vol. 2009(2009), No. 17, pp. 1--12.\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 2009 Texas State University - San Marcos.} \vspace{9mm}} \begin{document} \title[\hfilneg EJDE-2009/17\hfil Periodic solutions] {Periodic solutions of non-autonomous second order systems with $p$-Laplacian} \author[Z. Wang, J. Zhang \hfil EJDE-2009/17\hfilneg] {Zhiyong Wang, Jihui Zhang} % in alphabetical order \address{Zhiyong Wang \newline Department of Mathematics, Nanjing University of Information Science and Technology, Nanjing 210044, Jiangsu, China} \email{mathswzhy1979@gmail.com} \address{Jihui Zhang \newline Institute of Mathematics, School of Mathematics and Computer Sciences, Nanjing Normal University, Nanjing 210097, Jiangsu, China} \email{jihuiz@jlonline.com} \thanks{Submitted August 26, 2008. Published January 20, 2009.} \subjclass[2000]{34B15, 47J30, 58E05} \keywords{Periodic solutions; Hamiltonian system; $p$-Laplacian} \begin{abstract} We prove the existence of periodic solutions for non-autonomous second order systems with $p$-Laplacian. Our main tools are the minimax methods in critical point theory. Our results are new, even when $p=2$. \end{abstract} \maketitle \numberwithin{equation}{section} \newtheorem{theorem}{Theorem}[section] \newtheorem{lemma}[theorem]{Lemma} \newtheorem{remark}[theorem]{Remark} \newtheorem{definition}[theorem]{Definition} \section{Introduction} In this article concerns the existence of periodic solutions for the problem $$\label{e1.1} \begin{gathered} \frac{d}{dt}(|\dot{u}(t)|^{p-2}\dot{u}(t))=\nabla F(t,u(t))\quad \text{a.e. }t\in[0,T],\\ u(0)-u(T)=\dot{u}(0)-\dot{u}(T)=0, \end{gathered}$$ where $p>1$, $T>0$, $F:[0,T]\times \mathbb{R^N}\to \mathbb{R}$ satisfies the following assumption: \begin{itemize} \item[(A)] $F(t,x)$ is measurable in $t$ for every $x\in \mathbb{R^N}$ and continuously differentiable in $x$ for a.e. $t\in[0,T]$, and there exist $a\in C(\mathbb{R^+},\mathbb{R^+}),b\in L^1(0,T;\mathbb{R^+})$, such that \begin{equation*} |F(t,x)|\leq a(|x|)b(t),\quad |\nabla F(t,x)|\leq a(|x|)b(t) \end{equation*} for all $x\in \mathbb{R^N}$ and a.e. $t\in[0,T]$. \end{itemize} Considerable attention has been paid to the periodic solutions of problem \eqref{e1.1} for $p=2$, in recent years. The firsts to consider this problem when $p=2$ were Berger and Schechter \cite{Berger} in 1977, they proved the existence of solutions to problem \eqref{e1.1} for $p=2$ under the condition that $F(t,x)\to\pm\infty$ as $|x|\to\infty$ uniformly for a.e. $t\in [0,T]$. Subsequently, using the variational methods, many existence results are obtained, we refer the readers to \cite{Mawhin89, Tang98, Tang02, Tang01, Tang03, Zhao04, Mawhin87, Ben Naoum94, Ekeland02, Schechter06BVP, Schechter06, Wang08} and the references therein. However, there are few papers discussing periodic solutions for second order systems with $p$-Laplacian. In \cite{Tian07}, Tian has established the existence results for problem \eqref{e1.1} by the dual least action principle. Moreover, if problem \eqref{e1.1} with nonlinear boundary conditions, the existence of periodic solutions has also been proved in \cite{Jebelean07} by means of the least action principle and the mountain pass lemma. For $p=2$, under the assumptions that there exists $h(t)\in L^1(0,T;\mathbb{R^+})$ such that \begin{equation*} |\nabla F(t,x)|\leq h(t) \end{equation*} for all $x\in \mathbb{R^N}$ and a.e. $t\in[0,T]$, and that \begin{equation*} \int_0^TF(t,x)dt\to\pm\infty\quad\text{as } |x|\to\infty, \end{equation*} Mawhin and Willem in \cite{Mawhin89} have shown that problem \eqref{e1.1} admitted a periodic solution. After that, Tang in \cite{Tang98} generalized these results to the sublinear case. Concertely speaking, it is assumed that the nonlinearity satisfied the following restrictions $$\label{e1.2} |\nabla F(t,x)|\leq k(t)|x|^\alpha+p(t)\quad\text{for all } x\in \mathbb{R^N}\text{ and a.e. } t\in[0,T],$$ $$\label{e1.3} \frac{1}{|x|^{2\alpha}}\int_0^TF(t,x)dt\to\pm\infty\quad\text{as }|x|\to\infty,$$ here, $k(t),p(t)\in L^1(0,T;\mathbb{R^+})$ and $\alpha\in [0,1)$. Under these conditions, periodic solutions of problem \eqref{e1.1} with $p=2$ have been obtained. In addition, Tang in \cite{Tang02} first introduced the local $\alpha$-coercive conditions, that is, there exists $q(t)\in L^1(0,T;\mathbb{R^+})$ and a subset $E$ of $[0,T]$ with meas$(E)>0$ such that $$\label{e1.4} \frac{F(t,x)}{|x|^{2\alpha}}\leq q(t)$$ for all $x\in \mathbb{R^N}$ and a.e. $t\in[0,T]$, and that $$\label{e1.5} \frac{F(t,x)}{|x|^{2\alpha}}\to-\infty\quad\text{as }|x|\to\infty$$ for a.e. $t\in E$, to deal with the generalized Josephson-type systems. An interesting question naturally arises: In all the results \cite{Mawhin89, Tang98, Tang02} discussed above, the nonlinearity is required to grow at infinity at most like $|x|^\alpha$, is it possible to handle nonlinearity with faster increase at infinity and get the similar results of \cite{Mawhin89, Tang98, Tang02}. In the present paper, we will focus on this problem and give a positive answer. Here, we emphasize that our results are new even for $p=2$. We now state our main theorems. \begin{theorem} \label{thm1.1} Suppose that $F$ satisfies assumption {\rm (A)} and the following conditions: \begin{itemize} \item[(H1)] There exists a bounded nonincreasing positive function $\omega\in C((0,\infty);\mathbb{R^+})$ with the properties: \begin{itemize} \item[(i)] $\liminf_{t\to\infty}\frac{\omega(t)}{\omega(t^\gamma)}>0$\quad for some $\gamma\in(0,1)$, \item[(ii)] $\omega(t)\to0,\quad\omega(t)t\to\infty$\quad as $t\to\infty$. \end{itemize} Moreover, there exist $f\in L^2(0,T;\mathbb{R^+})$ and $g\in L^1(0,T;\mathbb{R^+})$ such that \begin{equation*} |\nabla F(t,x)|\leq f(t)[\omega(|x|)|x|]^{p-1}+g(t) \end{equation*} for all $x\in \mathbb{R^N}$ and a.e. $t\in[0,T]$; \item[(H2)] There exists a bounded non-increasing positive function $\omega\in C((0,\infty);\mathbb{R^+})$ which satisfies the conditions {\rm(i), (ii)} and \begin{equation*} \frac{1}{[\omega(|x|)|x|]^p}\int_0^TF(t,x)dx\to-\infty\quad \text{as }|x|\to\infty. \end{equation*} \end{itemize} Then problem \eqref{e1.1} has at least one solution in the set \begin{align*} W_T^{1,p} :=\big\{&u:[0,T]\to \mathbb{R^N},\; u\text{ is absolutely continuous},\\ & u(0)=u(T),\dot{u}\in L^p(0,T;\mathbb{R^N})\big\}\,. \end{align*} This set is a Banach space with the norm \begin{equation*} \|u\|:=\Big(\int_0^T|u(t)|^pdt+\int_0^T|\dot{u}(t)|^pdt\Big)^{1/p}\quad \text{for }u\in W_T^{1,p}. \end{equation*} \end{theorem} \begin{remark} \label{remark1.2}\rm Obviously, Theorem \ref{thm1.1} does not satisfy the corresponding conditions in \cite{Tang01, Tang03, Zhao04, Mawhin87, Ben Naoum94, Ekeland02, Berger, Schechter06BVP, Schechter06, Wang08}. Furthermore, there are functions $F(t,x)$ that satisfy our Theorem \ref{thm1.1} and do not satisfy the corresponding results in \cite{Mawhin89, Tang98, Tang02} even for $p=2$. For example, let \begin{equation*} \nabla F(t,x)=-\frac{|x|^{p-2}x}{[\ln(2+|x|^2)]^{p-1}}+d(t),\quad\forall t\in[0,T],x\in \mathbb{R^N}, \end{equation*} where $d(t)\in L^1(0,T;\mathbb{R})$. Let $\omega(|x|)=1/\ln(2+|x|^2),\gamma=1/2$, a straightforward computation shows that $F(t,x)$ satisfies all the conditions of our Theorem \ref{thm1.1}. However, it is clear that $F(t,x)$ neither satisfies \eqref{e1.2}, \eqref{e1.3} nor \eqref{e1.4}, \eqref{e1.5} for any $\alpha\in[0,1)$ even for $p=2$. \end{remark} \begin{theorem}\label{thm1.3} Suppose that $F$ satisfies assumption {\rm (A)} and {\rm (H1)}. Moreover assume $F$ satisfies the following conditions: \begin{itemize} \item[(H2*)] \begin{itemize} \item[(1)] There exists a bounded nonincreasing and positive function $\omega$ in \\ $C((0,\infty);\mathbb{R^+})$ and constant $C>0$ such that \begin{itemize} \item[(iii)] $\limsup_{t\to\infty}\frac{\omega(t)}{\omega(t^\gamma)}<+\infty$ for some $\gamma\in(0,1)$; \item [(iv)] $t\leq C\omega^2(t)t^2$ as $t\to\infty$; \end{itemize} \item[(2)] There exists $r(t)\in L^1(0,T;\mathbb{R^+})$ such that \begin{equation*} \frac{F(t,x)}{[\omega(|x|)|x|]^p}\leq r(t)\quad\text{for all }x\in \mathbb{R^N}\text{ and a.e. }t\in[0,T]; \end{equation*} \item[(3)] There exists a subset $E$ of $[0,T]$ with meas{\rm($E)>0$} such that for a.e. $t\in E$ \begin{equation*} \frac{F(t,x)}{[\omega(|x|)|x|]^p}\to-\infty\quad\text{as }|x|\to\infty. \end{equation*} \end{itemize} \end{itemize} Then problem \eqref{e1.1} has at least one solution in $W_T^{1,p}$. \end{theorem} \begin{remark} \label{remark1.4}\rm There are functions $F(t,x)$ that satisfy our Theorem \ref{thm1.3} and not do not satisfy the corresponding results in \cite{Mawhin89, Tang98, Tang02, Tang01, Tang03, Zhao04, Mawhin87, Ben Naoum94, Ekeland02, Berger, Schechter06BVP, Schechter06, Wang08}. For example, let \begin{equation*} F(t,x)=-\theta(t)\frac{|x|^p}{[\ln(2+|x|^2)]^{p-1}},\quad\forall t\in[0,T],x\in \mathbb{R^N}, \end{equation*} where \begin{equation*} \theta(t)=\begin{cases} \sin\frac{2\pi t}{T}, &t\in[0,T/2],\\ 0, &t\in[T/2,T]. \end{cases} \end{equation*} Take $\omega(|x|)=1/\ln(2+|x|^2)$, $\gamma=1/2$, $E=[T/6,T/4]$, by simple computation, $F(t,x)$ satisfies our Theorem \ref{thm1.3}. However, here $F(t,x)$ also neither satisfies \eqref{e1.2}, \eqref{e1.3} nor \eqref{e1.4}, \eqref{e1.5} for any $\alpha\in[0,1)$ even for $p=2$. \end{remark} \section{Preliminaries} For convenience, we will denote various positive constants as $C_i$, $i=0,1,2,\dots$. Define functional $\varphi$ on $W_T^{1,p}$ by $$\label{e2.1} \varphi(u)=\frac{1}{p}\int_0^T|\dot{u}(t)|^pdt+\int_0^TF(t,u(t))dt$$ for $u\in W_T^{1,p}$. It follows from assumption (A) that functional $\varphi$ is continuously differentiable on $W_T^{1,p}$, moreover, one has \begin{equation*} (\varphi'(u),v)=\int_0^T [(|\dot{u}(t)|^{p-2}\dot{u}(t),\dot{v}(t)) +(\nabla F(t,u(t)),v(t))]dt \end{equation*} for all $u,v\in W_T^{1,p}$. It is well known that the solutions to problem \eqref{e1.1} correspond to the critical points of the functional $\varphi$. For $u\in W_T^{1,p}$, let $\bar{u}:=\frac{1}{T}\int_0^Tu(t)dt$ and $\tilde{u}(t):=u(t)-\bar{u}$, then we have \begin{align*} & \|\tilde{u}\|_\infty\leq C_0\|\dot{u}\|_{L^p}\quad\text{(Sobolev's inequality)},\\ & \|\tilde{u}\|_{L^p}\leq C_0\|\dot{u}\|_{L^p}\quad\text{(Wirtinger's inequality)}, \end{align*} where $\|\tilde{u}\|_\infty:=\max_{0\leq t\leq T}|\tilde{u}(t)|$. To proof of our main theorems, we need the following auxiliary results. \begin{lemma}[\cite{Tang01}] \label{lemma2.1} Suppose that $G$ satisfies assumption {\rm(A)} and $E$ is a measurable subset of $[0,T]$. Assume that \begin{equation*} G(t,x)\to+\infty\quad\text{as }|x|\to\infty \end{equation*} for a.e. $t\in E$. Then for every $\delta>0$, there exists subset $E_\delta$ of $E$ with {\rm meas(}$E\backslash E_\delta{\rm )}<\delta$ such that \begin{equation*} G(t,x)\to+\infty\quad\text{as }|x|\to\infty \end{equation*} uniformly for all $t\in E_\delta$. \end{lemma} \begin{lemma}[\cite{Tang02}] \label{lemma2.2} For every constant $\beta>0$, there exists a constant $m_\beta>0$ such that \begin{equation*} \text{meas}\{t\in[0,T]||\bar{u}|0}\omega(s))^{2(p-1)}\int_0^T(T^{-1/p}\|u\|)^{p-1}dt\Big]^{1/2}+\|g\|_{L^1}\\ &\leq \|f\|_{L^2}\left[C_1\left(\omega((T^{-1/p}\|u\|)^{1/2})\|u\|\right)^{2(p-1)} +C_2\|u\|^{p-1}\right]^{1/2}+\|g\|_{L^1}\\ &\leq \|f\|_{L^2}\left[C_3\left(\omega^2((T^{-1/p}\|u\|)^{1/2})\|u\|^2 +\|u\|\right)\right]^{\frac{p-1}{2}}+\|g\|_{L^1}. \end{align*} Take $\tilde{\omega}(t):=\big[C_3\big(\omega^2((T^{-1/p}t)^{1/2})+\frac{1}{t}\big) \big]^{1/2},t>0$, then \begin{equation*} \|\nabla F(t,u(t))\|_{L^1}\leq\|f\|_{L^2}[\tilde{\omega}(\|u\|)\|u\|]^{p-1}+\|g\|_{L^1}. \end{equation*} Obviously, $\tilde{\omega}(t)$ satisfies (a) due to the properties of $\omega(t)$. Next, we come to check condition (c). Note that $\liminf_{t\to\infty} \omega(t)/\omega(t^\gamma)>0$ for $\gamma\in(0,1)$, we define $\xi:=\liminf_{t\to\infty}\frac{\omega^2(T^{-1/p}t)}{\omega^2((T^{-1/p}t) ^{1/2})}>0,$ if (H2) holds, for any $M>0$, we get $$\label{e2.2} \int_0^TF(t,u(t))dt\leq-M[\omega(|u(t)|)|u(t)|]^p+C_4,$$ which implies that for $\bar{u}\neq0$ \label{e2.3} \begin{aligned} \frac{\int_0^TF(t,\bar{u})dt}{\left[\tilde{\omega}(\|\bar{u}\|) \|\bar{u}\|\right]^p} &\leq \frac{-M\big[\omega(|\bar{u}|)|\bar{u}|\big]^p+C_4} {\big[C_3\left(\omega^2\left((T^{-1/p}\|\bar{u}\|)^{1/2}\right) \|\bar{u}\|^2+\|\bar{u}\|\right)\big]^{p/2}}\\ &= \frac{-M\left[\omega(T^{-1/p}\|\bar{u}\|)T^{-1/p}\|\bar{u}\| \right]^p+C_4}{\left[C_3\left(\omega^2 \left((T^{-1/p}\|\bar{u}\|)^{1/2}\right)\|\bar{u}\|^2+\|\bar{u}\| \right)\right]^{p/2}}. \end{aligned} By the definition of $\xi$, there exists $R>0$ such that $$\label{e2.4} \frac{\omega^2(T^{-1/p}t)t^2}{\omega^2\left((T^{-1/p}t)^{1/2}\right)t^2+t} \geq\frac{\xi}{2}\quad\text{as }t\geq R.$$ Therefore, \label{e2.5} \begin{aligned} & \frac{1}{[\tilde{\omega}(\|\bar{u}\|)\|\bar{u}\|]^p}\int_0^TF(t,\bar{u})dt \\ &\leq \frac{-MT^{-1}\big[\frac{\xi}{2}\left(\omega^2((T^{-1/p}\|\bar{u}\|)^{1/2})\|\bar{u}\|^2 +\|\bar{u}\|\right)\big]^{p/2}+C_4} {\left[C_3\left(\omega^2((T^{-1/p}\|\bar{u}\|)^{1/2})\|\bar{u}\|^2+\|\bar{u}\|\right)\right]^{p/2}}\\ &\leq -C_5M\quad\text{as }\|\bar{u}\|\to\infty, \end{aligned} condition (c) holds. Finally, we show that (d) is true. From Lemma \ref{lemma2.2}, for all $\bar{u}\neq0$, we have \[ \mathop{\rm meas}\big\{t\in[0,T]|\omega(|\bar{u}|)|\bar{u}| 0 for $\delta$ and $\beta$ small enough. By \eqref{e2.6}, for every $\eta>0$, there exists $L>0$ such that \begin{equation*} \frac{F(t,x)}{[\omega(|x|)|x|]^p}\leq-\eta \end{equation*} for all $|x|\geq L$ and a.e. $t\in E_\delta$. Furthermore, applying assumption ($H_2^*$)(2) yields \begin{equation*} F(t,x)\leq-\eta[\omega(|x|)|x|]^p+r^*(t) \end{equation*} for all $x\in \mathbb{R^N}$ and a.e. $t\in E_\delta$, where $r^*(t):=(\sup_{s>0}\omega(s))^pL^pr(t)$. Noting the definition of $\tilde{\omega}(t)$, $0<\liminf_{t\to\infty}\frac{\omega(t)}{\omega(t^\gamma)} \leq\limsup_{t\to\infty}\frac{\omega(t)}{\omega(t^\gamma)}<+\infty$ and assumption ($H_2^*$)(1)(iv), we deduce that there exist $C_6,C_7>0$ such that \begin{align*} C_6\omega(T^{-1/p}\|\bar{u}\|)\|\bar{u}\| &\leq \tilde{\omega}(\|\bar{u}\|)\|\bar{u}\|\\ &=[C_3(\omega^2((T^{-1/p}\|\bar{u}\|)^{1/2})\|\bar{u}\|^2+\|\bar{u}\|)]^{1/2}\\ &\leq C_7\omega(T^{-1/p}\|\bar{u}\|)\|\bar{u}\|\quad\text{as }\|\bar{u}\|\to\infty, \end{align*} which implies \label{e2.8} \begin{aligned} \int_{B\cap E_\delta}F(t,\bar{u})dt&\leq -\eta\int_{B\cap E_\delta}[\omega(|\bar{u}|)|\bar{u}|]^pdt+\int_{B\cap E_\delta}r^*(t)dt\\ &\leq -\eta\int_{B\cap E_\delta}[\omega(T^{-1/p}\|\bar{u}\|)m_\beta\|\bar{u}\|]^pdt+\int_{B\cap E_\delta}r^*(t)dt\\ &\leq -\eta\int_{B\cap E_\delta}\frac{{m_\beta}^p}{C_7}(\tilde{\omega}(\|\bar{u}\|)\|\bar{u}\|)^pdt+\int_{B\cap E_\delta}r^*(t)dt\\ &\leq -\eta C_8[\tilde{\omega}(\|\bar{u}\|)\|\bar{u}\|]^p+\int_{B\cap E_\delta}\hskip-.3cm r^*(t)dt\quad\text{as }\|\bar{u}\|\to\infty, \end{aligned} and \label{e2.9} \begin{aligned} \int_{[0,T]\backslash(B\cap E_\delta)}F(t,\bar{u})dt &\leq \int_{[0,T]\backslash(B\cap E_\delta)}r(t)[\omega(|\bar{u}|)|\bar{u}|]^pdt\\ &\leq \int_0^Tr(t)[\omega(T^{-1/p}\|\bar{u}\|)T^{-1/p}\|\bar{u}\|]^pdt\\ &\leq \int_0^T\frac{T^{-1}}{C_6}(\tilde{\omega}(\|\bar{u}\|)\|\bar{u}\|)^pr(t)dt\\ &\leq C_9[\tilde{\omega}(\|\bar{u}\|)\|\bar{u}\|]^p\int_0^Tr(t)dt\quad\text{as }\|\bar{u}\|\to\infty. \end{aligned} So, it follows from \eqref{e2.8} and \eqref{e2.9} that \begin{equation*} \limsup_{\|\bar{u}\|\to\infty} \frac{1}{[\tilde{\omega}(\|\bar{u}\|)\|\bar{u}\|]^p}\int_0^TF(t,\bar{u})dt\leq-\eta C_8+C_9\int_0^Tr(t)dt, \end{equation*} which implies that $$\label{e2.10} \frac{1}{[\tilde{\omega}(\|\bar{u}\|)\|\bar{u}\|]^p} \int_0^TF(t,\bar{u})dt\to-\infty \quad\text{as }\|\bar{u}\|\to\infty.$$ On the other hand, from (b) and the assumptions of (d), we get \label{e2.11} \begin{aligned} \big|\int_0^TF(t,u(t))dt-\int_0^TF(t,\bar{u})dt\big| &\leq \int_0^T\int_0^1|\nabla F(t,\bar{u}+s\tilde{u})||\tilde{u}|\,ds\,dt\\ &\leq \|\tilde{u}\|_\infty[\|f\|_{L^2}(\tilde{\omega}(\|u\|)\|u\|)^{p-1} +\|g\|_{L^1}]\\ &\leq C_{10}[\tilde{\omega}(\|u\|)\|u\|]^{p-1}\|\tilde{u}\| +C_{11}\|\tilde{u}\|. \end{aligned} As a consequence, note $\limsup_{\|u\|\to\infty}\frac{\|\tilde{u}\|}{\tilde{\omega}(\|u\|)\|u\|} <+\infty$, we then have $$\label{e2.12} C_{12}:=\limsup_{\|u\|\to\infty}\Big|\frac{1}{[\tilde{\omega}(\|u\|)\|u\|]^p} \Big[\int_0^TF(t,u(t))dt-\int_0^TF(t,\bar{u})dt\Big]\Big|<+\infty.$$ In addition, for $\|u\|\to\infty$, one knows $$\label{e2.13} 1=\frac{\|u\|}{\|u\|}=\frac{\|\bar{u}\|+\|\tilde{u}\|}{\|u\|} =\frac{\|\bar{u}\|}{\|u\|}+\frac{\|\tilde{u}\|}{\tilde{\omega}(\|u\|)\|u\|} \cdot\tilde{\omega}(\|u\|) =\frac{\|\bar{u}\|}{\|u\|}.$$ This, in conjunction with \eqref{e2.10}-\eqref{e2.12}, gives \label{e2.14} \begin{aligned} & \limsup_{\|u\|\to\infty}\frac{1}{[\tilde{\omega}(\|u\|)\|u\|]^p}\int_0^T F(t,u(t))dt\\ &\leq\limsup_{\|u\|\to\infty}\frac{1}{[\tilde{\omega}(\|u\|)\|u\|]^p}\int_0^TF(t,\bar{u})dt+C_{12}\\ &=\limsup_{\|\bar{u}\|\to\infty}\frac{1}{[\tilde{\omega}(\|\bar{u}\|)\|\bar{u}\|]^p}\int_0^TF(t,\bar{u})dt+C_{12} \to-\infty, \end{aligned} which completes the proof. \end{proof} \section{Proofs of theorems} Now, we give the proofs of the main results. \begin{proof}[Proof of Theorem \ref{thm1.1}] First, we prove that $\varphi$ satisfies the (PS) condition. Suppose that $\{u_n\}$ is a (PS) sequence for $\varphi$, that is, $\varphi'(u_n)\to0$ as $n\to\infty$ and $\{\varphi(u_n)\}$ is bounded. It follows from Wirtinger's inequality that $$\label{e3.1} \|\dot{u}_n\|_{L^p}\leq\|\tilde{u}_n\|\leq(C_0+1)^{1/p}\|\dot{u}_n\|_{L^p}$$ for all $n$. By virtue of the properties of $\tilde{\omega}(t)$, one has \label{e3.2} \begin{aligned} \tilde{\omega}(\|\bar{u}+\tilde{u}\|)\leq\min\{\tilde{\omega}(\|\bar{u}\|),\tilde{\omega}(\|\tilde{u}\|)\}, \end{aligned} which implies \label{e3.3} \begin{aligned} & \big|\int_0^T(\nabla F(t,u_n(t)),\tilde{u}_n(t))dt\big|\\ &\leq\|\tilde{u}_n\|_\infty\left[\|f\|_{L^2} (\tilde{\omega}(\|u_n\|)\|u_n\|)^{p-1}+\|g\|_{L^1}\right]\\ &\leq \|\tilde{u}_n\|_\infty\left[\|f\|_{L^2} (\tilde{\omega}(\|\bar{u}_n+\tilde{u}_n\|)\|\bar{u}_n+\tilde{u}_n\|)^{p-1}+\|g\|_{L^1}\right]\\ &\leq \|\tilde{u}_n\|_\infty\left[\|f\|_{L^2} (\tilde{\omega}(\|\bar{u}_n\|)\|\bar{u}_n\| +\tilde{\omega}(\|\tilde{u}_n\|)\|\tilde{u}_n\|)^{p-1}+\|g\|_{L^1}\right]\\ &\leq C_{13}\|\tilde{u}_n\|\left[\tilde{\omega}(\|\bar{u}_n\|)\|\bar{u}_n\|\right]^{p-1} +C_{13}\|\tilde{u}_n\|\left[\tilde{\omega}(\|\tilde{u}_n\|)\|\tilde{u}_n\|\right]^{p-1}\hskip-.2cm+C_{14}\|\tilde{u}_n\| \end{aligned} for all $n$. Thus, by \eqref{e3.1} and \eqref{e3.3}, we get \label{e3.4} \begin{aligned} \|\tilde{u}_n\| &\geq (\varphi'(u_n),\tilde{u}_n)\\ &=\int_0^T|\dot{u}_n(t)|^pdt+\int_0^T(\nabla F(t,u_n(t)),\tilde{u}_n(t))dt\\ &\geq C_{15}\|\tilde{u}_n\|^p-C_{13}\|\tilde{u}_n\|\left[\tilde{\omega}(\|\bar{u}_n\|)\|\bar{u}_n\|\right]^{p-1}\\ &\quad -C_{13}\|\tilde{u}_n\|\left[\tilde{\omega}(\|\tilde{u}_n\|) \|\tilde{u}_n\|\right]^{p-1}-C_{14}\|\tilde{u}_n\|. \end{aligned} Assume that $\{\|\tilde{u}_n\|\}$ is unbounded; that is, $\|\tilde{u}_n\|\to\infty$ as $n\to\infty$. Since $\tilde{\omega}(t)\to0$ as $t\to\infty$, it follows from \eqref{e3.4} that we can find a constant $C_{16}>0$ such that $$\label{e3.5} C_{16}\tilde{\omega}(\|\bar{u}_n\|)\|\bar{u}_n\|\geq\|\tilde{u}_n\|,$$ which implies \begin{equation*} [C_3(\omega^2((T^{-1/p}\|\bar{u}_n\|)^{1/2})\|\bar{u}_n\|^2+\|\bar{u}_n\|)]^{1/2} =\tilde{\omega}(\|\bar{u}_n\|)\|\bar{u}_n\|\to\infty \quad\text{as }n\to\infty. \end{equation*} Since $\omega$ is bounded, this leads to $$\label{e3.6} \|\bar{u}_n\|\to\infty\quad\text{as }n\to\infty.$$ On the other hand, $\|\bar{u}_n+s\tilde{u}_n\|\geq\|\bar{u}_n\|,s\in[0,1]$, by Lemma \ref{lemma2.3} (b) and \eqref{e3.5}, we see that \label{e3.7} \begin{aligned} & \int_0^T[F(t,u_n(t))-F(t,\bar{u}_n)]dt\\ &\leq \big|\int_0^T\int_0^1(\nabla F(t,\bar{u}_n+s\tilde{u}_n(t)),\tilde{u}_n(t))\,ds\,dt\big|\\ &\leq \|\tilde{u}_n\|_\infty\int_0^T\int_0^1|\nabla F(t,\bar{u}_n+s\tilde{u}_n(t))|\,ds\,dt\\ &\leq \|f\|_{L^2}\left[\tilde{\omega}(\|\bar{u}_n+s\tilde{u}_n\|)\|\bar{u}_n+s\tilde{u}_n\|\right]^{p-1} \|\tilde{u}_n\|_\infty+\|g\|_{L^1}\|\tilde{u}_n\|_\infty\\ &\leq \|f\|_{L^2}[\tilde{\omega}(\|\bar{u}_n\|)\|\bar{u}_n\|+\tilde{\omega}(\|\bar{u}_n\|)\|\tilde{u}_n\|]^{p-1} \|\tilde{u}_n\|_\infty+\|g\|_{L^1}\|\tilde{u}_n\|_\infty\\ &\leq C_{17}\|\tilde{u}_n\|[\tilde{\omega}(\|\bar{u}_n\|)\|\bar{u}_n\|]^{p-1} +C_{17}\|\tilde{u}_n\|[\tilde{\omega}(\|\bar{u}_n\|)\|\tilde{u}_n\|]^{p-1}+C_{18}\|\tilde{u}_n\|\\ &\leq C_{19}[\tilde{\omega}(\|\bar{u}_n\|)\|\bar{u}_n\|]^p +C_{20}[\tilde{\omega}(\|\bar{u}_n\|)\|\bar{u}_n\|]^p[\tilde{\omega}(\|\bar{u}_n\|]^{p-1} +C_{21}\tilde{\omega}(\|\bar{u}_n\|)\|\bar{u}_n\|, \end{aligned} which implies that \label{e3.8} \begin{aligned} \varphi(u_n) &= \frac{1}{p}\|\dot{u}_n\|_{L^p}^p+\int_0^T[F(t,u_n(t))-F(t,\bar{u}_n)]dt+\int_0^TF(t,\bar{u}_n)dt\\ &\leq C_{22}[\tilde{\omega}(\|\bar{u}_n\|)\|\bar{u}_n\|]^p +C_{20}[\tilde{\omega}(\|\bar{u}_n\|)\|\bar{u}_n\|]^p[\tilde{\omega}(\|\bar{u}_n\|]^{p-1}\\ &\quad +C_{21}\tilde{\omega}(\|\bar{u}_n\|)\|\bar{u}_n\|+\int_0^TF(t,\bar{u}_n)dt\\ &\leq [\tilde{\omega}(\|\bar{u}_n\|)\|\bar{u}_n\|]^p\Big[C_{22}+C_{20}[\tilde{\omega}(\|\bar{u}_n\|]^{p-1} +\frac{C_{21}}{[\tilde{\omega}(\|\bar{u}_n\|)\|\bar{u}_n\|]^{p-1}}\\ &\quad +\frac{1}{[\tilde{\omega}(\|\bar{u}_n\|)\|\bar{u}_n\|]^p} \int_0^TF(t,\bar{u}_n)dt\Big] \to -\infty\quad\text{as }\|\bar{u}_n\|\to\infty. \end{aligned} This contradicts the boundedness of $\varphi(u_n)$. So $$\label{e3.9} \{\|\tilde{u}_n\|\}\text{ is bounded}.$$ Suppose that $\{\|\bar{u}_n\|\}$ is unbounded and $\{\|\tilde{u}_n\|\}$ is bounded. With the similar manner above, we deduce that \label{e3.10} \begin{aligned} \varphi(u_n) &=\frac{1}{p}\|\dot{u}_n\|_{L^p}^p+\int_0^T[F(t,u_n(t))-F(t,\bar{u}_n)]dt +\int_0^TF(t,\bar{u}_n)dt\\ &\leq C_{23}\|\tilde{u}_n\|^p+C_{17}\|\tilde{u}_n\|[\tilde{\omega}(\|\bar{u}_n\|)\|\tilde{u}_n\|]^{p-1} +C_{18}\|\tilde{u}_n\|+\int_0^TF(t,\bar{u}_n)dt\\ &\leq C_{24}+C_{25}[\tilde{\omega}(\|\bar{u}_n\|)\|\tilde{u}_n\|]^{p-1}+C_{26}[\tilde{\omega}(\|\bar{u}_n\|)]^{p-1} +\int_0^TF(t,\bar{u}_n)dt\\ &\leq [\tilde{\omega}(\|\bar{u}_n\|)\|\bar{u}_n\|]^p\Big[\frac{C_{24}}{[\tilde{\omega}(\|\bar{u}_n\|)\|\bar{u}_n\|]^p} +\frac{C_{25}}{\tilde{\omega}(\|\bar{u}_n\|)\|\bar{u}_n\|} +\frac{C_{26}}{\tilde{\omega}(\|\bar{u}_n\|)\|\bar{u}_n\|^p}\\ &\quad +\frac{1}{[\tilde{\omega}(\|\bar{u}_n\|)\|\bar{u}_n\|]^p} \int_0^TF(t,\bar{u}_n)dt\Big] \to -\infty\quad\text{as }\|\bar{u}_n\|\to\infty, \end{aligned} which also contradicts the boundedness of $\varphi(u_n)$. Then $$\label{e3.11} \{\|\bar{u}_n\|\}\text{ is also bounded}.$$ From \eqref{e3.9} and \eqref{e3.11}, we have $\{\|u_n\|\}$ is bounded, thus $\varphi$ satisfies the (PS) condition. Since $W_T^{1,p}=\mathbb{R^N}\oplus \tilde{W}_T^{1,p}$, where $\tilde{W}_T^{1,p}:=\{u\in W_T^{1,p}|\int_0^Tu(t)dt=0\}$. Next, we shall prove that $$\label{e3.12} \varphi(u)\to+\infty\quad\text{as }\|u\|\to\infty\text{ in }\tilde{W}_T^{1,p}.$$ In fact, since $\tilde{\omega}(t)\to0$ as $t\to\infty$, then there exists $A>0$ such that $\tilde{\omega}(A)\leq\frac{1}{2pC_{29}}$. In a similar way to \eqref{e3.7}, we have \begin{align*} &\big|\int_0^T[F(t,u(t))-F(t,A)]dt\big|\\ &\leq C_{17}\|\tilde{u}\|[\tilde{\omega}(A)A]^{p-1} +C_{17}\|\tilde{u}\|[\tilde{\omega}(A)\|\tilde{u}\|]^{p-1}+C_{18}\|\tilde{u}\|\\ &\leq C_{27}\|\tilde{u}\|+C_{17}\tilde{\omega}(A)\|\tilde{u}\|^p\\ &\leq C_{28}\|\dot{u}\|_{L^p}+C_{29}\tilde{\omega}(A)\|\dot{u}\|_{L^p}^p\\ &\leq C_{28}\|\dot{u}\|_{L^p}+\frac{1}{2p}\|\dot{u}\|_{L^p}^p, \end{align*} which implies \label{e3.13} \begin{aligned} \varphi(u) &= \frac{1}{p}\|\dot{u}\|_{L^p}^p+\int_0^T[F(t,u(t))-F(t,A)]dt -\int_0^TF(t,A)dt\\ &\geq \frac{1}{p}\|\dot{u}\|_{L^p}^p-\frac{1}{2p}\|\dot{u}\|_{L^p}^p -C_{28}\|\dot{u}\|_{L^p}-\int_0^TF(t,A)dt \end{aligned} for all $u\in \tilde{W}_T^{1,p}$. By Wirtinger's inequality, one has \begin{equation*} \|u\|\to\infty\Leftrightarrow\|\dot{u}\|_{L^p}\to\infty\quad \text{on }\tilde{W}_T^{1,p}. \end{equation*} Hence, \eqref{e3.12} follows from \eqref{e3.13}. On the other hand, by (H2), $$\label{e3.14} \varphi(u)\to-\infty\quad\text{as }|u|\to\infty\quad\text{in }\mathbb{R^N}.$$ Combine \eqref{e3.12} and \eqref{e3.14}, applying saddle point theorem, then problem \eqref{e1.1} has at least one solution in $W_T^{1,p}$. \end{proof} \begin{proof}[Proof of Theorem \ref{thm1.3}] We commence by showing that $\varphi$ satisfies (PS) condition. Let $\{u_n\}$ be a sequence in $W_T^{1,p}$ such that $\{\varphi(u_n)\}$ is bounded and $\varphi'(u_n)\to0$ as $n\to\infty$. If $\{u_n\}$ is unbounded, without loss of generality, we may assume that $\|u_n\|\to\infty$ as $n\to\infty$. It follows from Lemma \ref{lemma2.3} (b) that \label{e3.15} \begin{aligned} \big|\int_0^T(\nabla F(t,u_n(t)),\tilde{u}_n(t))dt\big| &\leq \|\tilde{u}_n\|_\infty[\|f\|_{L^2}[\tilde{\omega}(\|u_n\|)\|u_n\| ]^{p-1}+\|G\|_{L^1}]\\ &\leq C_{30}[\tilde{\omega}(\|u_n\|)\|u_n\|]^{p-1}\|\tilde{u}_n\| +C_{31}\|\tilde{u}_n\|. \end{aligned} Hence, we have \begin{align*} \|\tilde{u}_n\| &\geq (\varphi'(u_n),\tilde{u}_n)\\ &= \int_0^T|\dot{u}_n(t)|^pdt+\int_0^T(\nabla F(t,u_n(t)),\tilde{u}_n(t))dt\\ &\geq C_{32}\|\tilde{u}_n\|^p-C_{30}[\tilde{\omega}(\|u_n\|)\|u_n\|]^{p-1} \|\tilde{u}_n\|-C_{31}\|\tilde{u}_n\|, \end{align*} which implies $$\label{e3.16} \limsup_{\|u_n\|\to\infty}\frac{\|\tilde{u}_n\|}{\tilde{\omega}(\|u_n\|)\|u_n\|}<+\infty.$$ Therefore, by Lemma \ref{lemma2.3} (d), one has $$\label{e3.17} \frac{1}{[\tilde{\omega}(\|u_n\|)\|u_n\|]^p}\int_0^T F(t,u_n(t))dt\to-\infty\quad\text{as }\|u_n\|\to\infty.$$ However, by the boundedness of $\varphi(u_n)$ and \eqref{e3.16}, we get \label{e3.18} \begin{aligned} \big|\frac{1}{[\tilde{\omega}(\|u_n\|)\|u_n\|]^p}\int_0^TF(t,u_n(t))dt \big| &=\big|\frac{\varphi(u_n)}{[\tilde{\omega}(\|u_n\|)\|u_n\|]^p} -\frac{\frac{1}{p}\int_0^T|\dot{u}_n(t)|^pdt}{[\tilde{\omega}(\|u_n\|) \|u_n\|]^p}\big|\\ &\leq \frac{\frac{1}{p}\int_0^T|\dot{u}_n(t)|^pdt}{[\tilde{\omega}(\|u_n\|) \|u_n\|]^p}\\ &\leq\frac{C_{33}\|\tilde{u}_n\|^p}{[\tilde{\omega}(\|u_n\|)\|u_n\|]^p} <+\infty, \end{aligned} which contradicts \eqref{e3.17}. So, $\varphi$ satisfies the (PS) condition. As in the proof of Theorem \ref{thm1.1}, we can obtain \begin{equation*} \varphi(u)\to+\infty\quad\text{as }\|u\|\to\infty\text{ in }\tilde{W}_T^{1,p}. \end{equation*} Furthermore, from \eqref{e2.10}, it is easy to see that \begin{equation*} \varphi(u)\to-\infty\quad\text{as } |u|\to\infty\quad\text{in }\mathbb{R^N}. \end{equation*} Thus, using the saddle point theorem, problem \eqref{e1.1} has at least one solution in $W_T^{1,p}$. \end{proof} \subsection*{Acknowledgments} This Project was Supported by Foundation of Major Project of Science and Technology of Chinese Education Ministry, SRFDP of Higher Education and NSF of Education Committee of Jiangsu Province and Foundation of Nanjing University of Information Science $\&$ Technology. The authors would like to thank the anonymous referee for his/her valuable suggestions. \begin{thebibliography}{00} \bibitem{Mawhin89} J. Mawhin, M. Willem; \emph{Critical Point Theory and Hamiltonian Systems}, Springer-Verlag, New York, 1989. \bibitem{Tian07} Y. Tian, W. Ge; \emph{Periodic solutions of non-autonomous second order systems with $p$-Laplacian}, Nonlinear Anal. 66 (2007) 192-203. \bibitem{Jebelean07} P. Jebelean, G. Morosanu; \emph{Ordinary $p$-Laplacian systems with nonlinear boundary conditions}, J. Math. Anal. Appl. 325 (2007) 90-100. \bibitem{Tang98} C. L. Tang; \emph{Periodic solutions for nonautonomous second order systems with sublinear nonlinearity}, Proc. Amer. Math. Soc. 126 (1998) 3263-3270. \bibitem{Tang02} C. L. Tang, X. P. Wu; \emph{Periodic solutions for a class of nonautonomous subquadratic second order Hamiltonian systems}, J. Math. Anal. Appl. 275 (2002) 870-882. \bibitem{Tang01} C. L. Tang, X. P. Wu; \emph{Periodic solutions for second order systems with not uniformly coercive potential}, J. Math. Anal. Appl. 259 (2001) 386-304. \bibitem{Tang03} C. L. Tang, X. P. Wu; \emph{Note on periodic solutions of subquadratic second order systems}, J. Math. Anal. Appl. 285 (2003) 8-16. \bibitem{Zhao04} F. Zhao, X. Wu; \emph{Saddle point reduction method for some non-autonomous second order systems}, J. Math. Anal. Appl. 291 (2004) 653-665. \bibitem{Mawhin87} J. Mawhin; \emph{Semi-coercive monotone variational problems}, Acad. Roy. Belg. Bull. Cl. Sci. 73 (1987) 118-130. \bibitem{Ben Naoum94} A. K. Ben Naoum, C. Troestler and M. Willem; \emph{Existence and multiplicity results for homogeneous second order differential equations}, J. Differential Equations 112 (1994) 239-249. \bibitem{Ekeland02} I. Ekeland and N. Ghoussoub; \emph{Certain new aspects of the calculus of variations in the large}, Bull. Amer. Math. Soc. 39 (2002) 207-265. \bibitem{Berger} M. Berger and M. Schechter; \emph{On the solvability of semilinear gradient operator equations}, Advance in Math. 25 (1977) 97-132. \bibitem{Schechter06BVP} M. Schechter; \emph{Periodic solutions of second-order nonautonomous dynamical systems}, Bound. Value Probl. (2006) Art. ID 25104, 1-9. \bibitem{Schechter06} M. Schechter; \emph{Periodic non-autonomous second-order dynamical systems}, J. Differential Equations 223 (2006) 290-302. \bibitem{Wang08} Z. Wang, J. Zhang and Z. Zhang; \emph{Periodic solutions of second order non-autonomous Hamiltonian systems with local superquadratic potential}, Nonlinear Anal. (2008) doi:10.1016/j.na. 2008.07.023. \end{thebibliography} \end{document}