数学杂志  2019, Vol. 39 Issue (2): 305-316   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
李姗姗
王智勇
一类带组合项的分数阶边值问题两个解的存在性
李姗姗, 王智勇    
南京信息工程大学数学与统计学院 江苏 南京 210044
摘要:本文研究一类带组合项的分数阶边值问题
$ \begin{equation*} \left\{\begin{array}{ll} -\frac{d}{dt}\left(\frac{1}{2} {_{0}D_{t}^{-\beta}}(u'(t))+ \frac{1}{2}{_{t}D_{T}^{-\beta}}(u'(t))\right) = \nabla W\left(t, u(t)\right)\quad \text{a.e. }t\in[0, T], \\ u(0) = 0, \quad u(T) = 0, \end{array}\right.\ \end{equation*} $
其中W在无穷远处是一个超二次和次二次的组合.利用山路引理和Ekeland变分原理,得到上述问题至少存在两个非平凡解,推广和发展了已有文献中的结果.
关键词分数阶边值问题    山路引理    Ekeland变分原理    
TWO SOLUTIONS FOR A CLASS OF FRACTIONAL BOUNDARY VALUE PROBLEMS WITH MIXED NONLINEARITIES
LI Shan-shan, WANG Zhi-yong    
School of Mathematics and Statistics Nanjing University of Information Science and Technology Nanjing 210044 China
Abstract: In this paper we deal with the following kind of fractional boundary value problems with mixed nonlinearities $ \begin{equation*} \left\{\begin{array}{ll} -\frac{d}{dt}\left(\frac{1}{2} {_{0}D_{t}^{-\beta}}(u'(t))+ \frac{1}{2}{_{t}D_{T}^{-\beta}}(u'(t))\right) = \nabla W\left(t, u(t)\right)\quad \text{a.e. }t\in[0, T], \\ u(0) = 0, \quad u(T) = 0, \end{array}\right.\ \end{equation*} $ where W is a combination of superquadratic and subquadratic terms. By using mountain pass lemma and Ekeland's variational principle we prove that the above boundary value problem has at least two nontrivial solutions which extends and improves some existed literature.
Keywords: fractional boundary value problem     mountain pass lemma     Ekeland's variational principle    
1 引言

考虑如下分数阶边值问题

$ \begin{equation*} \left\{\begin{array}{ll} -\frac{d}{dt}\left(\frac{1}{2} {_{0}D_{t}^{-\beta}}(u'(t))+ \frac{1}{2}{_{t}D_{T}^{-\beta}}(u'(t))\right) = \nabla W\left(t, u(t)\right)\quad \text{a.e. }t\in[0, T], \\ u(0) = 0, \quad u(T) = 0, \end{array}\right.\ \end{equation*} $ (1.1)

其中, $ _{0}D_{t}^{-\beta} $$ _{t}D_{T}^{-\beta} $分别是$ 0\leq\beta<1 $阶的左Riemann-Liouville积分与右Riemann-Liouville积分, $ \nabla W(t, u(t)) $$ W(t, u(t)) $关于$ u $的梯度.

随着分数阶微积分的发展, 其已涵盖了科学与工程的很多应用领域, 特别是近30年以来, 已应用到流体力学, 粘弹性力学, 反常扩散, 分数控制系统, 各种电子回路, 生物系统的电传导, 神经的分数模型等, 并且已有了成熟的发展, 见文献[1-6].

近年来, 许多学者利用各种方法对问题(1.1)进行了研究, 并得到了很多重要的结果.2011年, Jiao和Zhou在文献[7]中首次建立了问题(1.1)的变分结构, 并利用极小化作用原理以及山路引理得到了问题(1.1)解的存在性; 在文献[1]中, 陈在位势函数$ W(t, x) $满足渐近二次的条件下, 利用山路引理研究了问题(1.1)解的存在性; 文献[2]中, 陈和唐运用喷泉定理和对偶喷泉定理分别考虑了位势函数$ W(t, x) $为超二次和次二次的情形, 得到了问题(1.1)存在无穷多个解.

$ \begin{equation*} W(t, u(t)) = \lambda G(t, u(t))+ F(t, u(t)), \ \end{equation*} $ (1.2)

其中$ \lambda>0 $, $ F(t, u(t)) $$ G(t, u(t)) $在无穷远处分别是超二次和次二次的.最近, 文献[4]首次研究了具有形如(1.2)式这样组合项的位势函数, 借助于山路引理和极小化方法得到问题(1.1)至少存在两个非平凡解.受上述结果的启发, 本文利用山路引理以及Ekeland变分原理, 进一步讨论此类带组合项的问题(1.1)两个解的存在性.我们有

定理1.1  假设$ W $满足如下条件

[F1] $ \forall x\in \mathbb{R}^N $, $ F(t, x) $关于$ t $是可测的, 对a.e.$ t\in[0, T] $, $ F(t, x) $关于$ x $是连续可微的, 且存在$ a\in C(\mathbb{R}^{+}, \mathbb{R}^{+}) $, $ b\in L^{1}([0, T], \mathbb{R}^{+}) $使得

$ \begin{equation*} |F(t, x)|\leq a(|x|)b(t), \quad |\nabla F(t, x)|\leq a(|x|)b(t), \end{equation*} $

$ \forall x\in \mathbb{R}^N $和\text{a.e.}$ t\in[0, T] $成立;

[F2] $ \liminf\limits_{|x|\rightarrow 0}F(t, x)\geq0 $对\text{a.e.}$ t\in[0, T] $一致成立;

[F3] $ \limsup\limits_{|x|\rightarrow 0}\frac{F(t, x)}{|x|^2}<|\cos(\pi\alpha)|\frac{\Gamma^2(\alpha+1)}{2T^{2\alpha}} $对\text{a.e.}$ t\in[0, T] $一致成立;

[F4] $ \liminf\limits_{|x|\rightarrow+\infty}\frac{F(t, x)}{|x|^2}>\frac{\pi^2}{|\cos(\pi\alpha)|\Gamma^2(2-\alpha)T^{2\alpha}(3-2\alpha)} $对\text{a.e.}$ t\in[0, T] $一致成立;

[F5]存在常数$ r>2 $, 使得$ \limsup\limits_{|x|\rightarrow+\infty}\frac{F(t, x)}{|x|^r}<+\infty $对\text{a.e.}$ t\in[0, T] $一致成立;

[F6]存在常数$ \mu>r-1 $, 使得$ \liminf\limits_{|x|\rightarrow+\infty}\frac{\left(\nabla F(t, x), x\right)-2F(t, x)}{|x|^\mu}>0 $对\text{a.e.}$ t\in[0, T] $一致成立;

[G1] $ G(t, 0) = 0 $, $ G(t, x)\in C^1([0, T], \mathbb{R}^N) $;

[G2]存在常数$ \bar{t}\in [0, T] $, $ 1<r_0<2 $, $ \omega_0>0 $以及$ \delta>0 $, 使得当$ |x|\leq\delta $时, 有

$ \begin{equation*} G(\bar{t}, x)>\omega_0|x|^{r_0}; \end{equation*} $

[G3]存在常数$ 1<r_1<r_2<\cdots<r_m<\mu-r+2<2 $, 函数$ \omega_i\in L^{\frac{2}{2-r_i}}(\mathbb{R}, \mathbb{R}^+) $, $ (i = 1, 2, \cdots, m) $使得$ \forall (t, x)\in[0, T]\times\mathbb{R}^N $, 有

$ |\nabla G(t, x)|\leq\sum\limits_{i = 1}^{m}r_i\omega_i(t)|x|^{r_i-1}, $

其中$ \Gamma $为通常的伽马函数, $ \alpha = 1-\beta/2 $, 则存在常数$ \Lambda_0>0 $, 使得当$ \lambda\in(0, \Lambda_0) $时, 问题(1.1)至少存在两个非平凡解.

注1.1  容易看出, 这里定理1.1所给的条件同文献[4]相比要更一般, 因此结果推广和发展了文献[4]中的结论.

本文结构安排如下:第二部分简要介绍一些分数阶微积分的基本概念并给出本文所需要的预备引理; 第三部分给出了定理1.1的证明; 最后一部分将给出一个例子来说明结果.

2 预备知识

本节将简要介绍一些分数阶微积分的基本概念, 并且给出问题(1.1)的工作空间和变分结构.

定义2.1  [8]$ \gamma>0 $, 函数$ f(t) $定义在$ [a, b] $上, 它的$ \gamma $阶左和右Riemann-Liouville分数积分分别定义为

$ \begin{equation*} _aD_t^{-\gamma}f(t) = \frac{1}{\Gamma(\gamma)}\int_a^t(t-s)^{\gamma-1}f(s)ds, \quad t>a \end{equation*} $

$ \begin{equation*} _tD_b^{-\gamma}f(t) = \frac{1}{\Gamma(\gamma)}\int_t^b(s-t)^{\gamma-1}f(s)ds, \quad t<b, \end{equation*} $

其中右式在$ [a, b] $上逐点有定义.相应的, 当$ \gamma = n\in \mathbb{N} $时, 上式分别与以下的$ n $次积分形式相对应

$ \begin{equation*} _aD_t^{-n}f(t) = \frac{1}{(n-1)!}\int_a^t(t-s)^{n-1}f(s)ds \end{equation*} $

$ \begin{equation*} _tD_b^{-n}f(t) = \frac{1}{(n-1)!}\int_t^b(s-t)^{n-1}f(s)ds. \end{equation*} $

定义2.2  [8]$ \gamma>0 $, 函数$ f(t) $定义在$ [a, b] $上, 它的$ \gamma $阶左和右Riemann-Liouville分数导数分别定义为

$ \begin{equation*} _aD_t^{\gamma}f(t) = \frac{d^n}{dt^n}{_aD_t^{\gamma-n}}f(t) = \frac{1}{\Gamma(n-\gamma)}\frac{d^n}{dt^n}\left(\int_a^t(t-s)^{n-\gamma-1}f(s)ds\right), \quad t>a \end{equation*} $

$ \begin{equation*} _tD_b^{\gamma}f(t) = (-1)^n\frac{d^n}{dt^n}{_tD_b^{\gamma-n}}f(t) = \frac{1}{\Gamma(n-\gamma)}(-1)^n\frac{d^n}{dt^n}\left(\int_t^b(s-t)^{n-\gamma-1}f(s)ds\right), \\\quad t<b, \end{equation*} $

其中$ t\in[a, b] $, $ n-1\leq \gamma<n $, $ n\in\mathbb{N} $, 其中$ n = [\gamma]+1 $, $ [\gamma] $$ \gamma $的整数部分.特别地, 若$ \gamma = n\in\mathbb{N} $, 则

$ \begin{eqnarray*} &&_aD_t^{0}f(t) = {_tD_b^{0}}f(t) = f(t), \\ &&_aD_t^{n}f(t) = f^{(n)}(t)\quad \text{和}\quad _tD_b^{n}f(t) = (-1)^nf^{(n)}(t), \end{eqnarray*} $

其中$ f^{(n)}(t) $$ f(t) $$ n $阶导数; 若$ 0<\gamma<1 $, 则

$ \begin{equation*} _aD_t^{\gamma}f(t) = \frac{1}{\Gamma(1-\gamma)}\frac{d}{dt}\left(\int_a^t(t-s)^{-\gamma}f(s)ds\right), \quad t>a \end{equation*} $

$ \begin{equation*} _tD_b^{\gamma}f(t) = \frac{1}{\Gamma(1-\gamma)}\frac{d}{dt}\left(\int_t^b(s-t)^{-\gamma}f(s)ds\right), \quad t<b. \end{equation*} $

$ AC\left([a, b], \mathbb{R}^N\right) $为绝对连续函数空间.对$ k\in \mathbb{N} $,

$ \begin{equation*} AC^k\left([a, b], \mathbb{R}^N\right) = \left\{f\in C^{k-1}\left([a, b], \mathbb{R}^N\right):f^{(k-1)}\in AC\left([a, b], \mathbb{R}^N\right)\right\}. \end{equation*} $

定义2.3  [8]$ \gamma>0 $, $ n\in\mathbb{N} $.若$ \gamma\in[n-1, n) $$ f(t)\in AC^n\left([a, b], \mathbb{R}^N\right) $, 则函数$ f(t) $$ \gamma $阶左和右Caputo分数导数于$ [a, b] $上几乎处处存在, 且分别定义为

$ \begin{equation*} _a^cD_t^\gamma f(t) = {_aD_t}^{\gamma-n}f^{(n)}(t) = \frac{1}{\Gamma(n-\gamma)}\left(\int_a^t(t-s)^{n-\gamma-1}f^{(n)}(s)ds\right) \end{equation*} $

$ \begin{equation*} _t^cD_b^\gamma f(t) = (-1)^n{_tD_b}^{\gamma-n}f^{(n)}(t) = \frac{(-1)^n}{\Gamma(n-\gamma)}\left(\int_t^b(s-t)^{n-\gamma-1}f^{(n)}(s)ds\right). \end{equation*} $

$ 0<\alpha\leq 1 $以及$ 1<p<+\infty $, 考虑空间

$ E_0^{\alpha, p} = \left\{u\in L^p\left([0, T], \mathbb{R}^N\right):{_0^cD_t^\alpha u(t)}\in L^p\left([0, T], \mathbb{R}^N\right), u(0) = u(T) = 0\right\} $, 其范数为

$ \begin{equation*} \|u\|_{\alpha, p} = \left[\int_0^T\left(|u(t)|^p+|_0^cD_t^\alpha u(t)|^p\right)dt\right]^{1/p}, \forall u\in E_0^{\alpha, p}. \end{equation*} $

命题2.4  [8]$ 0<\alpha\leq1 $, $ 1<p<+\infty $, 对$ u\in E_0^{\alpha, p} $, 有

$ \begin{equation} \|{u}\|_{p}\leq \frac{T^\alpha}{\Gamma(\alpha+1)}\|{_0^cD_t^\alpha u}\|_{p}. \end{equation} $ (2.1)

进一步, 若$ \alpha>1/p $$ 1/p+1/q = 1 $, 则有

$ \begin{equation*} \label{2.2} \|{u}\|_{\infty}\leq\frac{T^{\alpha-1/p}}{{\Gamma(\alpha)\left((\alpha-1)q+1\right)}^{1/q}}\|{_0^cD_t^\alpha u}\|_{p}. \end{equation*} $

特别的, 当$ p = 2 $时, 有以下两个不等式成立

$ \begin{eqnarray} &&\|{u}\|_{2}\leq\tau_2\|{_0^cD_t^\alpha u}\|_{2}, \end{eqnarray} $ (2.2)
$ \begin{eqnarray} &&\|{u}\|_{\infty}\leq\tau_\infty \|{_0^cD_t^\alpha u}\|_{2}, \end{eqnarray} $ (2.3)

其中$ \tau_2 = \frac{T^\alpha}{\Gamma(\alpha+1)} $, $ \tau_\infty = \frac{T^{\alpha-1/2}}{\Gamma(\alpha)\sqrt{2\alpha-1}} $.

注2.1由(2.1)式, $ E_0^{\alpha, p} $上的范数等价于如下定义的$ \|\cdot\|_{\alpha, p} $,

$ \begin{equation*} \|u\|_{\alpha, p} = \|{_0^cD_t^\alpha u}\|_{p} = \left(\int_0^T|_0^cD_t^\alpha u|^pdt\right)^{1/p}. \end{equation*} $

引理2.5  [7]$ 0<\alpha\leq 1 $, $ 1<p<+\infty $, 且满足$ \alpha>1/p $, 若序列$ \{u_n\} $$ E_0^{\alpha, p} $上弱收敛于$ u $, 即$ u_n\rightharpoonup u $, 则在$ C([0, T], \mathbb{R}^N) $上有$ u_n\rightarrow u $, 即当$ n\rightarrow+\infty $时, $ \|u_n-u\|_\infty\rightarrow 0 $.

下面将在空间$ E^\alpha: = E_0^{\alpha, 2} $中讨论问题(1.1)的多解性, 此时其范数为$ \|u\|_\alpha: = \|u\|_{\alpha, 2} $.

定义2.6  若对$ u\in E^\alpha $以及$ \forall v\in E^\alpha $

$ \begin{align*} &-\int_0^T\frac{1}{2}\left[\left(_0^cD_t^\alpha u(t), _t^cD_T^\alpha v(t)\right)+\left(_t^cD_T^\alpha u(t), _0^cD_t^\alpha v(t)\right)\right]dt\\\nonumber &-\int_0^T\lambda (\nabla G(t, u(t)), v(t))dt-\int_0^T(\nabla F(t, u(t)), v(t))dt = 0, \end{align*} $

则称$ u $为问题(1.1)的弱解.

考虑泛函

$ \begin{equation*} \varphi(u) = \int_0^T\left[-\frac{1}{2}\left(_0^cD_t^\alpha u(t), _t^cD_T^\alpha u(t)\right)-\lambda G(t, u(t))-F(t, u(t))\right]dt, \forall u\in E^\alpha.\ \end{equation*} $ (2.4)

根据条件(F1), (G1)和(G3), 由文献[7]中的定理4.1可知, $ \varphi\in C^1(E^\alpha, \mathbb{R}) $且有

$ \begin{align*} \langle\varphi'(u), v\rangle = &-\int_0^T\frac{1}{2}\left[\left(_0^cD_t^\alpha u(t), _t^cD_T^\alpha v(t)\right)+\left(_t^cD_T^\alpha u(t), _0^cD_t^\alpha v(t)\right)\right]dt\\\nonumber &-\int_0^T\lambda (\nabla G(t, u(t)), v(t))dt-\int_0^T(\nabla F(t, u(t)), v(t))dt, \\\quad\forall u, v\in E^\alpha, \end{align*} $ (2.5)

其中$ \langle\cdot, \cdot\rangle $表示通常的$ E^\alpha $$ (E^\alpha)^* $的对偶作用.进一步可知, $ u\in E^\alpha $为问题(1.1)的弱解当且仅当$ u $为泛函$ \varphi $的临界点.

引理2.7  [7]$ 1/2<\alpha\leq1 $, 则对$ u\in E^{\alpha} $, 有

$ \begin{equation*} |\cos(\pi\alpha)|\|{u}\|_{\alpha}^{2}\leq-\int_{0}^{T}\left(_0^cD_t^{\alpha}u(t), _t^cD_T^{\alpha}u(t)\right)dt\leq\frac{1}{|\cos(\pi\alpha)|}\|{u}\|_{\alpha}^{2}.\ \end{equation*} $ (2.6)

定义2.8  [6]$ X $是实Banach空间, $ \varphi\in C^1(X, \mathbb{R}) $, 如果$ \{u_n\}\subset X $, $ {\varphi(u_n)} $有界并且$ \varphi'(u_n)\rightarrow0(n\rightarrow+\infty) $, 则称序列$ \{u_n\}\subset X $$ \varphi $的(PS)序列.如果它的每一(PS)序列都有一收敛子列, 则称泛函$ \varphi $满足(PS)条件.

定义2.9  [6]$ X $是实Banach空间, $ \varphi\in C^1(X, \mathbb{R}) $, 如果$ \{u_n\}\subset X $, $ \varphi(u_n) $有界并且$ (1+\|u_n\|_\alpha)\varphi'(u_n)\rightarrow0(n\rightarrow+\infty) $, 则称序列$ \{u_n\}\subset X $$ \varphi $的(C)序列.如果它的每一(C)序列都有一收敛子列, 则称泛函$ \varphi $满足(C)条件.

引理2.10  [6] (山路引理)设$ X $是实Banach空间, $ \varphi\in C^1(X, \mathbb{R}) $满足(PS)条件, $ \varphi(0) = 0 $, 且有

(I1) 存在常数$ \rho, \beta>0 $, 使得$ \varphi|_{\partial B_\rho}\geq\beta $;

(I2) 存在$ e\in X\backslash \bar{B}_\rho $, 使得$ \varphi(e)\leq 0 $,

$ \varphi $存在一个临界值$ c\geq\beta $, 且$ c = \inf\limits_{g\in \Gamma}\max\limits_{s\in[0, 1]}\varphi(g(s)) $, 其中$ \bar{B}_\rho = \{u\in X: \|u\|\leq\rho\} $, $ \Gamma = \{g\in C([0, 1], X):g(0) = 0, g(1) = e\} $.

注2.2  由文献[9]可知, 山路引理在(C)条件下依然成立.

3 定理的证明

为了叙述方便, 令$ C_i(i = 1, 2, 3, \cdots) $表示一系列不同的正常数, $ X = E^\alpha $.

引理3.1  假设条件(F1), (F5), (F6), (G1)和(G3)成立, 则泛函$ \varphi $满足(C)条件.

  首先证明$ \{u_n\} $$ X $上有界.假设$ \{u_n\} $是泛函$ \varphi $的(C)序列, 即$ \{\varphi(u_n)\} $有界, 且当$ n\rightarrow+\infty $时, 有$ \|\varphi'(u_n)\|_{X^*}(1+\|u_n\|_\alpha)\rightarrow0 $, 其中$ X^* $$ X $的对偶空间, 则$ \forall n\in\mathbb{N} $, 有

$ \begin{equation*} \varphi(u_n)\leq C_1 , \quad \quad\|\varphi'(u_n)\|_{X^*}(1+\|u_n\|_\alpha)\leq C_1.\ \end{equation*} $ (3.1)

根据条件(F5), 存在常数$ R_1>0 $, 使得$ \forall|x|\geq R_1 $以及\text{a.e.}$ t\in [0, T] $, 有

$ \begin{equation*} F(t, x)\leq C_2|x|^r.\ \end{equation*} $ (3.2)

再结合条件(F1), $ \forall x\in \mathbb{R}^N $$ \text{a.e.}t\in [0, T] $, 可知

$ \begin{equation*} F(t, x)\leq C_2|x|^r+g_1(t), \ \end{equation*} $ (3.3)

其中$ g_1(t) = \max\limits_{s\in[0, R_1]}a(s)b(t) $.由条件(G3)可知

$ \begin{equation*} |(\nabla G(t, x), x)|\leq\sum\limits_{i = 1}^{m}r_i\omega_i(t)|x|^{r_i}, \end{equation*} $

所以利用(2.2)式以及Hölder不等式, 可得

$ \begin{align*} &\int_0^T\lambda|(\nabla G(t, x), x)|dt\leq\int_0^T\lambda\sum\limits_{i = 1}^{m}r_i\omega_i(t)|x|^{r_i}dt\\ \leq&\sum\limits_{i = 1}^{m}\lambda r_i\|\omega_i(t)\|_{\frac{2}{2-r_i}}\left(\int_0^T|x|^2dt\right)^{r_i/2}\\ \leq&\sum\limits_{i = 1}^{m}\lambda r_i\tau_2^{r_i}\|\omega_i(t)\|_{\frac{2}{2-r_i}}\|x\|_\alpha^{r_i} \leq C_3\sum\limits_{i = 1}^{m}\|x\|_\alpha^{r_i}. \end{align*} $ (3.4)

而由条件(G1)和(G3), 有

$ \begin{align*} |G(t, x)|& = |G(t, x)-G(t, 0)|\leq\int_0^1|\nabla G(t, sx)||x|ds\\ &\leq\int_0^1\sum\limits_{i = 1}^mr_i\omega_i(t)|x|^{r_i}s^{r_i-1}ds = \sum\limits_{i = 1}^{m}\omega_i(t)|x|^{r_i}, \end{align*} $ (3.5)

因此, 进一步依据(2.2), (3.4), (3.5)式和Hölder不等式, 可得

$ \begin{align*} \int_0^T\lambda|G(t, x)|dt&\leq\int_0^T\lambda\sum\limits_{i = 1}^{m}\omega_i(t)|x|^{r_i}dt \leq\sum\limits_{i = 1}^{m}\lambda \tau_2^{r_i}\|\omega_i(t)\|_{\frac{2}{2-r_i}}\|x\|_\alpha^{r_i} \leq C_4\\\sum\limits_{i = 1}^{m}\|x\|_\alpha^{r_i}. \end{align*} $ (3.6)

这样由(2.4), (2.6), (3.1), (3.3)和(3.6)式可得

$ \begin{align*} \frac{|\cos(\pi\alpha)|}{2}\|u_n\|_\alpha^2 &\leq \varphi(u_n)+\int_0^T\lambda G(t, u_n(t))dt+\int_0^TF(t, u_n(t))dt\\ &\leq C_1+\int_0^T\lambda|G(t, u_n(t))|dt+\int_0^Tg_1(t)dt+C_2\int_0^T|u_n(t)|^rdt\\ &\leq C_4\sum\limits_{i = 1}^{m}\|u_n\|_\alpha^{r_i}+C_2\int_0^T|u_n(t)|^rdt+C_5. \end{align*} $ (3.7)

另一方面, 由条件(F6), 存在常数$ R_2>0 $, 使得$ \forall|x|\geq R_2 $和a.e.$ t\in [0, T] $, 有

$ \begin{equation*} \left(\nabla F(t, x), x\right)-2F(t, x)\geq C_6|x|^\mu, \end{equation*} $

因此再由条件(F1), $ \forall x\in\mathbb{R}^N $和a.e.$ t\in [0, T] $, 可得

$ \begin{equation*} \left(\nabla F(t, x), x\right)-2F(t, x)\geq C_6|x|^\mu-g_2(t), \ \end{equation*} $ (3.8)

其中$ g_2(t) = (2+R_2)\max\limits_{s\in [0, R_2]}a(s)b(t)+C_6R_2^\mu $.

由条件(G3)可知$ \mu< r $.取$ \varepsilon_i = \frac{C_6|\cos(\pi\alpha)|}{4mC_2\tau_\infty^{r-\mu}\tau_2^{r_i}\lambda(r_i+2)} $, 利用(2.2), (3.5)式, 条件(G1), (G3)以及Hölder不等式和带$ \varepsilon_i $$ (i = 1, 2, \cdots m) $的Young不等式, 有

$ \begin{align*} &\int_0^T\lambda|(\nabla G(t, u_n(t)), u_n(t))|dt+\int_0^T2\lambda|G(t, u_n(t))|dt\\ \leq&\int_0^T\lambda(r_i+2)\sum\limits_{i = 1}^{m}\omega_i(t)|u_n(t)|^{r_i}dt\\ \leq&\lambda\sum\limits_{i = 1}^{m}(r_i+2)\tau_2^{r_i}\|\omega_i\|_{\frac{2}{2-r_i}}\|u_n\|_\alpha^{r_i}\\ \leq&\lambda\sum\limits_{i = 1}^{m}(r_i+2)\tau_2^{r_i}\varepsilon_i\|u_n\|_\alpha^{r_ip_i}+ \lambda\sum\limits_{i = 1}^{m}(r_i+2)\tau_2^{r_i}\frac{1}{\varepsilon_i}\|\omega_i\|_{\frac{2}{2-r_i}}^{q_i}, \end{align*} $ (3.9)

其中$ r_ip_i = \mu-r+2 $, $ 1/p_i+1/q_i = 1 $.考虑(2.4), (2.5), (3.1), (3.4), (3.6), (3.8)和(3.9)式, 有

$ \begin{align*} 3C_1\geq& 2\varphi(u_n)-\langle \varphi'(u_n), u_n\rangle\\ = &\int_0^T\left[(\nabla F(t, u_n(t)), u_n(t))-2F(t, u_n(t))\right]dt\\&+\int_0^T\lambda \left[\left(\nabla G(t, u_n(t)), u_n(t)\right)-2G(t, u_n(t))\right]dt\\ \geq& C_6\int_0^T|u_n(t)|^\mu dt-\int_0^Tg_2(t)dt-\int_0^T\lambda|((\nabla G(t, u_n(t))), u_n(t))|dt\\ \geq& C_6\int_0^T|u_n(t)|^\mu dt-\int_0^Tg_2(t)dt-\lambda\sum\limits_{i = 1}^{m}(r_i+2)\tau_2^{r_i}\varepsilon_i\|u_n\|_\alpha^{r_ip_i}\\& -\lambda\sum\limits_{i = 1}^{m}(r_i+2)\tau_2^{r_i}\frac{1}{\varepsilon_i}\|\omega_i\|_{\frac{2}{2-r_i}}^{q_i}. \end{align*} $

因此可以得到

$ \begin{equation*} \int_0^T|u_n(t)|^\mu dt\leq \frac{\lambda}{C_6}\sum\limits_{i = 1}^{m}(r_i+2)\tau_2^{r_i}\varepsilon_i\|u_n\|_\alpha^{r_ip_i} +\frac{\lambda}{C_6}\sum\limits_{i = 1}^{m}(r_i+2)\tau_2^{r_i}\frac{1}{\varepsilon_i}\|\omega_i\|_{\frac{2}{2-r_i}}^{q_i}+C_{7}.\ \end{equation*} $ (3.10)

由于$ \mu>r-1 $, 利用(3.7)和(3.10)式, 并注意到$ \varepsilon_i $的定义, 有

$ \begin{align*} &\frac{|\cos(\pi\alpha)|}{2}\|u_n\|_\alpha^2 \leq C_4\sum\limits_{i = 1}^{m}\|u_n\|_\alpha^{r_i}+C_2\tau_\infty^{r-\mu}\|u_n\|_\alpha^{r-\mu}\int_0^T|u_n(t)|^\mu dt+C_5 \\ \leq& C_2\tau_\infty^{r-\mu}\|u_n\|_\alpha^{r-\mu}\left[\frac{\lambda}{C_6}\sum\limits_{i = 1}^{m}(r_i+2)\tau_2^{r_i}\varepsilon_i\|u_n\|_\alpha^{r_ip_i}+\frac{\lambda}{C_6}\sum\limits_{i = 1}^{m}(r_i+2)\tau_2^{r_i}\frac{1}{\varepsilon_i}\|\omega_i\|_{\frac{2}{2-r_i}}^{q_i}\right]\\ &+C_2\tau_\infty^{r-\mu}\|u_n\|_\alpha^{r-\mu}C_{7}+C_4\sum\limits_{i = 1}^{m}\|u_n\|_\alpha^{r_i}+C_5\\ \leq& C_4\sum\limits_{i = 1}^{m}\|u_n\|_\alpha^{r_i}+\frac{|\cos(\pi\alpha)|}{4}\|u_n\|_\alpha^2+C_{8}\|u_n\|_\alpha^{r-\mu}+C_{5}. \end{align*} $

注意到这里$ \mu>r-1 $, 因此, $ \|u_n\|_\alpha $有界.

最后证明$ \{u_n\} $$ X $上有强收敛子列.因为$ \{u_n\} $$ X $上有界, 并且$ X $是自反的Banach空间(见文献[8]), 则存在子序列, 不妨仍记为$ \{u_n\} $, 使得在$ X $上, 有$ u_n\rightharpoonup u $.于是当$ n\rightarrow+\infty $时, 有

$ \begin{align*} \langle\varphi'(u_n)-\varphi'(u), u_n-u\rangle& = \langle\varphi'(u_n), u_n-u\rangle-\langle\varphi'(u), u_n-u\rangle\\ &\leq\|\varphi'(u_n)\|_{X^*}\|u_n-u\|_\alpha-\langle\varphi'(u), u_n-u\rangle\rightarrow0. \end{align*} $ (3.11)

由(2.3)式和引理2.5可知$ \{u_n\} $$ C([0, T], \mathbb{R}^N) $上有界.进一步, 当$ n\rightarrow+\infty $时, 有$ \|u_n-u\|_\infty\rightarrow0 $.因此再依据条件(F1)和(G3), 当$ n\rightarrow+\infty $时, 有

$ \begin{align*} &\int_0^T\nabla F(t, u_n(t))dt\rightarrow\int_0^T\nabla F(t, u(t))dt, \\ &\int_0^T\nabla G(t, u_n(t))dt\rightarrow\int_0^T\nabla G(t, u(t))dt. \end{align*} $ (3.12)

此外, 由(2.5), (2.6)和(3.12)式, 有

$ \begin{align*} &\langle\varphi'(u_n)-\varphi'(u), u_n-u\rangle\\ & = -\int_0^T(_0^cD_t^\alpha (u_n(t)-u(t)), _t^cD_T^\alpha(u_n(t)-u(t)))dt\\ &\quad-\int_0^T(\nabla F(t, u_n(t))-\nabla F(t, u(t)), u_n(t)-u(t))dt\\ &\quad-\int_0^T\lambda(\nabla G(t, u_n(t))-\nabla G(t, u(t)), u_n(t)-u(t))dt\\ &\geq|\cos(\pi\alpha)|\|u_n-u\|_\alpha^{2}-\|u_n-u\|_\infty\left|\int_0^T\left(\nabla F(t, u_n(t))-\nabla F(t, u(t))\right)dt\right|\\ &\quad-\|u_n-u\|_\infty\left|\int_0^T\lambda\left(\nabla G(t, u_n(t))-\nabla G(t, u(t))\right)dt\right|. \end{align*} $

所以结合(3.11)式可得, 当$ n\rightarrow+\infty $时, $ \|u_n-u\|_\alpha\rightarrow0 $, 即$ \varphi $满足(C)条件.

引理3.2  假设$ W $满足条件(F3), (F5), (G1)和(G3), 则存在常数$ \rho>0, \beta>0 $$ \Lambda_0>0 $, 使得当$ \|{u}\|_\alpha = \rho $, $ \lambda\in(0, \Lambda_0) $时, 有$ \varphi(u)\geq\beta $.

  根据条件(F3), (F5)可知, 存在常数$ \delta_1\in(0, |\cos(\pi\alpha)|) $, $ \forall x\in \mathbb{R}^N $$ \text{a.e.}t\in [0, T] $, 有

$ \begin{equation*} F(t, x)\leq \left(|\cos(\pi\alpha)|-\delta_1\right)\frac{\Gamma^2(\alpha+1)}{2T^{2\alpha}}|x|^2+C_{9}|x|^r.\quad\ \end{equation*} $ (3.13)

由(2.1), (2.4), (2.6), (3.13)式及条件(G1), (G3), 有

$ \begin{align*} \varphi(u)& = \int_0^T\left[-\frac{1}{2}(_0^cD_t^\alpha u(t), _t^cD_T^\alpha u(t))-\lambda G(t, u(t))-F(t, u(t))\right]dt\\ &\geq\frac{|\cos(\pi\alpha)|}{2}\|u\|_\alpha^2-\lambda\int_0^T\sum\limits_{i = 1}^{m}\omega_i(t)|u(t)|^{r_i}dt-C_{9}\int_0^T|u(t)|^r dt\\ &\quad-\left(|\cos(\pi\alpha)|-\delta_1\right)\frac{\Gamma^2(\alpha+1)}{2T^{2\alpha}}\int_0^T|u(t)|^2dt \end{align*} $
$ \begin{align*} &\geq\frac{|\cos(\pi\alpha)|}{2}\|u\|_\alpha^{2}-\lambda \sum\limits_{i = 1}^{m}\tau_2^{r_i}\|\omega_i\|_{\frac{2}{2-r_i}}\|u\|_\alpha^{r_i}-C_{10 }\|u\|_\alpha^r-\frac{\left(|\cos(\pi\alpha)|-\delta_1\right)}{2}\|u_n\|_\alpha^2 \\ &\geq\frac{\delta_1}{2}\|u\|_\alpha^2-\lambda C_{11}\sum\limits_{i = 1}^{m}\|u\|_\alpha^{r_i}-C_{10 }\|u\|_\alpha^r\\ & = \|u\|_\alpha^2\left(\frac{\delta_1}{2}-\lambda C_{11}\sum\limits_{i = 1}^{m}\|u\|_\alpha^{r_i-2}- C_{10}\|u\|_\alpha^{r-2}\right). \end{align*} $

$ \begin{equation*} g(z) = \lambda C_{11}\sum\limits_{i = 1}^{m}z^{r_i-2}+C_{10}z^{r-2}, z>0, \ \quad\Omega = \{z\in\mathbb{R}^+:g'(z) = 0\}. \end{equation*} $ (3.14)

因为$ 1<r_i<2<r $, 且当$ z\rightarrow 0^+ $$ z\rightarrow+\infty $时, 都有$ g(z)\rightarrow+\infty $, 故$ \Omega\neq\emptyset $.于是, $ \forall z_0\in\Omega $, 根据(3.14)式, 有

$ \begin{equation*} (r-2)C_{10}z_0^{r-2} = \lambda C_{11}\sum\limits_{i = 1}^{m}(2-r_i)z_0^{r_i-2}.\ \end{equation*} $ (3.15)

利用(3.14)和(3.15)式可得, 存在不依赖于$ \lambda $的常数$ C_{12} $, 使得

$ \begin{align*} g(z_0)& = \lambda C_{11}\sum\limits_{i = 1}^mz_0^{r_i-2}+C_{10}z_0^{r-2}\\ &\leq C_{12}\lambda C_{11}\sum\limits_{i = 1}^m(2-r_i)z_0^{r_i-2}+C_{10}z_0^{r-2}\\ & = \left(C_{12}(r-2)+1\right)C_{10}z_0^{r-2}. \end{align*} $

所以存在$ \hat{z}\in\mathbb{R}^+ $, 使得当$ z_0\in(0, \hat{z})\bigcap \Omega $时, 有$ 0<g(z_0) = \beta_1<\frac{\delta_1}{2} $.下面证明, 若$ \lambda $充分小, 则必有$ \Omega\subset(0, \hat{z}) $.事实上, 由(3.15)式可知, 存在不依赖于$ \lambda $的常数$ C_{13} $, 使得

$ \begin{equation*} C_{13}\lambda\geq\frac{z_0^{r-2}}{\sum\limits_{i = 1}^mz_0^{r_i-2}} = \frac{z_0^r}{\sum\limits_{i = 1}^mz_0^{r_i}}. \end{equation*} $

$ y(z) = \frac{z^r}{\sum\limits_{i = 1}^mz^{r_i}} $, $ z\in\mathbb{R}^+ $, 则$ y(z) $$ \mathbb{R}^+ $上严格递增且当$ z\rightarrow0 $时, $ y(z)\rightarrow0 $.因此存在$ y^{-1}(z) $使得, $ y^{-1}(y(z)) = z $.进一步可知$ y^{-1}(z) $$ \mathbb{R}^+ $上也是严格递增的且当$ z\rightarrow0 $时, 有$ y^{-1}(z)\rightarrow0 $.所以存在常数$ \Lambda_0>0 $, 当$ \lambda\in(0, \Lambda_0) $时, 有

$ \begin{equation*} \hat{z}>y^{-1}(C_{13}\lambda)\geq y^{-1}(y(z_0)) = z_0. \end{equation*} $

$ \beta = \left(\frac{\delta_1}{2}-\beta_1\right)\rho $, 故存在常数$ \rho = z_0>0 $, 当$ \|u\|_\alpha = \rho $$ \lambda\in(0, \Lambda_0) $时, 有$ \varphi(u)\geq\beta $.

引理3.3  假设$ W $满足条件(F1), (F4), (G1)和(G3), 则存在$ e\in X $, 并且$ \|e\|_\alpha>\rho $, 使得$ \varphi(e)<0 $.

  由条件(F4)知, 存在常数$ \delta_2>0 $$ R_3>0 $, 使得$ \forall|x|\geq R_3 $以及\text{a.e.}$ t\in[0, T] $, 有

$ \begin{equation*} F(t, x)\geq\left(\frac{\pi^2}{|\cos(\pi\alpha)|\Gamma^2(2-\alpha)T^{2\alpha}(3-2\alpha)}+\delta_2\right)|x|^2, \end{equation*} $

再由条件(F1), $ \forall x\in \mathbb{R}^N $$ \text{a.e.}t\in [0, T] $, 有

$ \begin{equation*} F(t, x)\geq\left(\frac{\pi^2}{|\cos(\pi\alpha)|\Gamma^2(2-\alpha)T^{2\alpha}(3-2\alpha)}+\delta_2\right) (|x|^2-R_3^2)-g_3(t), \quad\ \end{equation*} $ (3.16)

其中$ g_3(t) = \max\limits_{s\in[0, R_3]}a(s)b(t) $.

选取$ u_0 = (\frac{T}{\pi}\sin\frac{\pi t}{T}, 0, \cdots, 0)\in X $, 计算可得

$ \begin{equation*} \|u_0\|_2^2 = \frac{T^3}{2\pi^2}, \quad \|u_0\|_\alpha^2\leq\frac{T^{3-2\alpha}}{\Gamma^2(2-\alpha)(3-2\alpha)}.\ \end{equation*} $ (3.17)

根据(2.4), (2.6), (3.5), (3.16)和(3.17)式, 注意到$ 1<r_i<2 $, 当$ s\rightarrow+\infty $时, 有

$ \begin{align*} \varphi(su_0)& = \int_0^T\left[-\frac{s^2}{2}(_0^cD_t^\alpha u_0(t), _t^cD_T^\alpha u_0(t))-\lambda G(t, su_0(t))-F(t, su_0(t))\right]dt\\ &\leq\frac{s^2}{2|\cos(\pi\alpha)|}\|u_0\|_\alpha^2+\lambda\int_0^T|G(t, su_0(t))|+\int_0^Tg_3(t)dt\\ &\quad-\left(\frac{s^2\pi^2}{|\cos(\pi\alpha)|\Gamma^2(2-\alpha)T^{2\alpha}(3-2\alpha)}+s^2\delta_2\right)\int_0^T|u_0(t)|^2dt\\ &\leq\frac{s^2}{2|\cos(\pi\alpha)|}\cdot\frac{T^{3-2\alpha}}{\Gamma^2(2-\alpha)(3-2\alpha)}-\frac{s^2\delta_2T^3}{2\pi^2} +C_4\sum\limits_{i = 1}^ms^{r_i}\|u_0\|_\alpha^{r_i}+C_{14}\\ &\quad-\frac{s^2\pi^2}{|\cos(\pi\alpha)|\Gamma^2(2-\alpha)T^{2\alpha}(3-2\alpha)}\cdot\frac{T^3}{2\pi^2}\\ & = -\frac{s^2\delta_2T^3}{2\pi^2}+C_4\sum\limits_{i = 1}^ms^{r_i}\left(\frac{T^{3-2\alpha}}{\Gamma^2(2-\alpha)(3-2\alpha)}\right)^{r_i/2}+C_{14}\\ &\rightarrow-\infty. \end{align*} $

所以存在充分大的$ s_0 $使得$ \varphi(s_0u_0)<0 $, 取$ e = s_0u_0\in X $, 有$ \varphi(e)<0 $.

引理3.4  假设条件(F2)和(G2)成立, 则

$ \begin{equation*} -\infty<\inf\{\varphi(u):u\in \bar B_\rho\}<0, \end{equation*} $

其中$ \rho $由引理3.2给出.

  因为$ X $是Banach空间(见文献[8]), 设$ \Omega = \bar{B}_\rho\subset X $是弱列紧的序列式弱闭集, 泛函$ \varphi:\Omega\rightarrow\mathbb{R} $$ \Omega $上是弱下半连续的.首先证明$ \varphi $有下界且达到下确界, 即存在$ u_0\in \Omega $, 使得$ \varphi(u_0) = \inf\limits_{u\in \Omega}\varphi(u) $.

$ h: = \varphi(u_0) = \inf\limits_{u\in \Omega}\varphi(u) = -\infty $, 则存在$ \{u_n\}_{n = 1}^{+\infty}\subset \Omega $, 使得$ \varphi(u_n)<-n $.因为$ \Omega $是弱列紧的, 不妨认为, 当$ n\rightarrow+\infty $时, 有$ u_n\rightharpoonup u_0 $.又因为$ \Omega $是序列式弱闭集, 所以$ u_0\in \Omega $.根据$ \varphi $的弱下半连续性得

$ \begin{equation*} \varphi(u_0)\leq\lim\limits_{n\rightarrow+\infty}\inf\varphi(u_n) = -\infty. \end{equation*} $

该矛盾说明$ \varphi $$ \Omega $上有下界, 从而有下确界.记$ \{u_n\}_{n = 1}^{+\infty}\subset\Omega $是极小化序列, 即当$ n\rightarrow+\infty $时, 有$ \varphi(u_n)\rightarrow h $.因为$ \Omega $是弱列紧的序列式弱闭集, 不妨认为, 当$ n\rightarrow+\infty $时, 有$ u_n\rightharpoonup u_0\in\Omega $.利用$ \varphi $的弱下半连续性得$ h\leq\varphi(u_0)\leq\lim\limits_{n\rightarrow+\infty}\inf\varphi(u_n) = \lim\limits_{n\rightarrow+\infty}\varphi(u_n) = h, $$ \varphi(u_0) = h $.

下证存在$ \varphi(u_0) = h<0 $.

由条件(F2)知, 存在常数$ \delta_3>0 $, 使得$ \forall|x|\leq\delta_3 $以及$ \text{a.e.}t\in[0, T] $, 有

$ \begin{equation*} F(t, x)\geq0.\ \end{equation*} $ (3.18)

由条件(G2)得, 存在常数$ \delta_4>0 $, $ \forall|x|\leq \delta $$ \forall t\in(\bar{t}-\delta_4, \bar{t}+\delta_4) $, 有

$ \begin{equation*} G(t, x)\geq\omega_0|x|^{r_0}.\ \end{equation*} $ (3.19)

$ \phi\in C_0^\infty((\bar{t}-\delta_4, \bar{t}+\delta_4), \mathbb{R}^N) $.令$ \delta_5 = \min\{\delta, \delta_3\} $, 则对充分小的$ s\in(0, \frac{\delta_5}{\|\phi\|_\infty}] $, 由(2.4), (2.6), (3.18)和(3.19)式, 可得

$ \begin{align*} \varphi(s\phi)& = \int_{0}^{T}\left[-\frac{s^2}{2}(_0^cD_t^\alpha \phi(t), _t^cD_T^\alpha \phi(t))-\lambda G(t, s\phi(t))-F(t, s\phi(t))\right]dt\\ &\leq\frac{s^2}{2|\cos(\pi\alpha)|}\|\phi\|_\alpha^{2}-\lambda \omega_0s^{r_0}\int_{\bar{t}-\delta_4}^{\bar{t}+\delta_4}|\phi(t)|^{r_0}dt\\ &<0. \end{align*} $

因此有$ -\infty<\inf\{\varphi(u):u\in \bar B_\rho\}<0 $.

定理1.1的证明  根据引理3.1可知$ \varphi $满足(C)条件, 由条件(F3)和(G1)知$ W(t, 0) = 0 $, 所以有$ \varphi(0) = 0 $, 再利用引理3.2和引理3.3可知引理2.9(山路引理)中的条件(I1), (I2)成立.因此问题(1.1)存在一个非平凡解$ u_1\in X $, 使得$ \varphi(u_1) = c>0 $.另一方面, 由引理3.4和Ekeland变分原理(见文献[6]), 类似于文献[10]中定理2.1的证明, 可得有界的(PS)序列$ \{u_n\} $, 即$ \varphi(u_n) $有界, 且当$ n\rightarrow+\infty $时, $ \varphi'(u_n)\rightarrow0 $.又因为$ u_n\in \bar{B}_\rho $, 即$ \|u_n\|_\alpha<\rho $, 所以当$ n\rightarrow+\infty $时, 有$ (1+\|u_n\|_\alpha)\varphi'(u_n)_\alpha\rightarrow0 $, 即$ \{u_n\} $$ \varphi $中有界的(C)序列, 再结合引理3.1, 则问题(1.1)存在另一个非平凡解$ u_2\in X $, 使得$ \varphi(u_2)<0 $.综上所述, 问题(1.1)至少存在两个非平凡解$ u_1, u_2\in X $, 使得$ \varphi(u_2)<0<\varphi(u_1) $.

4 例子

在本节中, 给出一个具体的例子来说明我们的结果.考虑如下分数阶边值问题:

$ \begin{equation*} \left\{\begin{array}{ll} -\frac{d}{dt}\left(\frac{1}{2} {_{0}D_{t}^{-\beta}}(u'(t))+ \frac{1}{2}{_{t}D_{T}^{-\beta}}(u'(t))\right) = \nabla W\left(t, u(t)\right)\quad \text{a.e. }t\in[0, T], \\ u(0) = 0, \quad u(T) = 0, \end{array}\right. \end{equation*} $

其中$ W(t, x) = (1+t^2)\ln(1+2|x|^2)|x|^2+\lambda\frac{4}{5}t|x|^{\frac{5}{4}} $.令$ F(t, x) = (1+t^2)\ln(1+2|x|^2)|x|^2 $, $ G(t, x) = \frac{4}{5}t|x|^\frac{5}{4} $.选取$ r = 2.5 $, $ \mu = 2 $, 则容易验证$ W(t, x) $满足条件(F1)–(F6)和条件(G1)–(G3).由定理1.1, 则存在常数$ \Lambda_0>0 $, 使得当$ \lambda\in(0, \Lambda_0) $时, 问题(1.1)在$ E^\alpha $上至少有两个非平凡解.但显然$ W(t, x) $不满足文献[1-5, 7, 8]中的相关条件.

参考文献
[1] 陈静. 一类带渐近二次非线性项的分数阶Dirichlet边值问题解的存在性[J]. 应用数学学报, 2015, 38(1): 53–66.
[2] Chen J, Tang X H. Infinitely many solutions for a class of frational boundary value problem[J]. Bull. Malays. Math. Sci. Soc., 2013, 36(4): 1083–1097.
[3] Chen J, Tang X H. Infinitely many solutions for boundary value problems arising from the fractional advection dispersion equation[J]. Appl. Math., 2015, 60(6): 703–724. DOI:10.1007/s10492-015-0118-2
[4] Zhang Z H, Yuan R. Two solutions for a class of fractional boundary value problems with mixed nonlinearities[J]. Bull. Malays. Math. Sci. Soc., 2016: 1–15.
[5] Khaled B A, Abdeljabbar G, Khaled K. Existence of solutions for fractional differential equations with Dirichlet boundary conditions[J]. Electron. J. Diff. Equ., 2016, 2016(116): 1–11.
[6] Mawhin J, Willem M. Critical point theory and Hamiltonian systems[M]. New York: SpringerVerlag, 1989.
[7] Jiao F, Zhou Y. Existence of solutions for a class of fractional boundary value problem via critical point theory[J]. Comput. Math. Appl., 2011, 62(3): 1181–1199. DOI:10.1016/j.camwa.2011.03.086
[8] Zhou Y. Basic theory of fractional differential equations[M]. Singapore: World Scientific, 2014.
[9] Izydorek M, Janczewska J. Homoclinic solutions for a class of the second-order Hamiltonian systems[J]. J. Differ. Equ., 2005, 219(2): 375–389. DOI:10.1016/j.jde.2005.06.029
[10] Xu L P, Chen H B. Multiple solutions for the nonhomogeneous fourth order elliptic equations of Kirchhoff-type[J]. Taiwanese J. Math., 2015, 19(4): 1215–1226. DOI:10.11650/tjm.19.2015.4716