数学物理学报  2018, Vol. 38 Issue (3): 514-526   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
张申贵
带类p(x)-拉普拉斯算子的双非局部问题的无穷多解
张申贵     
西北民族大学数学与计算机科学学院 兰州 730030
摘要:该文运用变分方法研究一类带类px)-拉普拉斯算子的双非局部狄利克雷问题.利用喷泉定理和对称山路定理,得到了此类问题一列高能量和低能量解的存在性.
关键词变分方法    临界点    p(x)-拉普拉斯算子    非局部问题    弱解    
Infinitely Many Solutions for a Bi-Nonlocal Problem Involving p(x)-Laplacian-Like Operator
Zhang Shengui     
School of Mathematics and Computer Science, Northwest Minzu University, Lanzhou 730030
Abstract: In this article, we study a class of bi-nonlocal Dirichlet problem involving p(x)-Laplacian-like operator via variational methods. Using the fountain theorem and the symmetric mountain pass theorem, the existence of a sequence of high and low energy solutions for this problem are obtained, respectively.
Key words: Variational methods     Critical point     p(x)-Laplacian-like operator     Nonlocal problem     Weak solutions    
1 引言和结果

$\Delta_{p(x)}u={\rm div}( |\nabla u|^{p(x)-2}\nabla u )$$p(x)$ -拉普拉斯算子.带有$p(x)$ -拉普拉斯算子的数学模型可以描述许多物理模型, 如电流体的运动模型[1-2].另外, 利用变指数微分方程模型可以研究图像处理问题[3].近年来, $p(x)$ -拉普拉斯方程的吸引了一些学者的关注[4-13].在文献[14-18]中, 研究了类$p(x)$ -拉普拉斯方程的边值问题解的存在性.

本文中, 考虑带类$p(x)$ -拉普拉斯算子的双非局部Dirichlet问题

$ \begin{equation} \label{euler} \left\{ \begin{array}{ll} -a\left(\psi(u)\right){\rm div}\left(|\nabla u|^{p(x)-2}\nabla u +\frac{|\nabla u|^{2p(x)-2}\nabla u}{\sqrt{1+|\nabla u|^{2p(x)}}}\right) =b\left(\int_{\Omega} F(x, u) {\rm d}x\right)f(x, u), &x\in\Omega, \\[2mm] u=0, &x\in\partial\Omega, \\[2mm] \psi(u)=\int_{\Omega} \frac{|\nabla u|^{p(x)}+\sqrt{1+|\nabla u|^{2p(x)}}}{p(x)} {\rm d}x, \end{array}\right. \end{equation} $ (1.1)

其中$\Omega\subset{\mathbb{R}}^{N}$为具有光滑边界的有界区域, $a(t)\in C( {\mathbb{R}}^{+}, {\mathbb{R}}^{+} )$, $b(t)\in C( {\mathbb{R}}, {\mathbb{R}} )$, $f(x, u)\in C( \bar{\Omega}\times {\mathbb{R}}, {\mathbb{R}} )$, $F(x, u)=\int_{0}^{u} f(x, s) {\rm d}s$, $p\in C(\overline{\Omega})$使得$1 <p(x) <N$, $\forall x\in\overline{\Omega}$, 且

$ \begin{equation} 1 <p^{-}:=\inf\limits_{\Omega}p(x)\leq p^{+}:=\sup\limits_{\Omega}p(x) <+\infty. \end{equation} $ (1.2)

问题(1.1)方程左端的算子带有非局部系数$a\Big(\int_{\Omega} \frac{|\nabla u|^{p(x)}+\sqrt{1+|\nabla u|^{2p(x)}}}{p(x)}{\rm d}x \Big)$, 此系数依赖于$\Omega$中kinetic能量的平均值.问题(1.1)方程方程右端的外力项也带有非局部系数$b\left( \int_{\Omega} F(x, u) {\rm d}x \right)$, 这表明外力项依赖于位势能量在有界区域$\Omega$中的平均值.这导致问题(1.1)中的方程不再是一个逐点成立的等式, 此类问题常被称为双非局部问题.

问题(1.1)有着丰富的数学物理背景, 例如, 当$p(x)\equiv2$, $a(t)\equiv1$$b(t)=\lambda$时, 问题(1.1)退化为刻画"毛细现象"的广义Capiliarity方程.

(AR)存在$\mu>p^{+}$, $L>0$, 使得

$ 0 <\mu F(x, u)\leq f(x, u)u, $

对所有$x\in\Omega$$|u|\geq L$成立.条件(AR)可以保证$f(x, u)$关于变量$u$在无穷远处是超线性的, 及能量泛函(PS)序列的有界性.此条件广泛的用于变指数微分方程边值问题的研究.特别当$a(t)\equiv1$, $b(t)\equiv1$且(AR)成立时, 文献[16-17]中得到了问题(1.1)非平凡解和多重解的存在性定理.但是, 很多超线性函数不满足条件(AR).当条件(AR)不成立时, 一些学者讨论了超线性问题解的存在性[19-21].

本文中, 首先在不同于(AR)的超线性条件下, 我们将利用临界点理论中的喷泉定理[22]得到问题(1.1)无穷多解的存在性定理.假设下列条件成立:

(a$_{0}$)  设存在常数$a_{0}>0$, 有$a(t)\geq a_{0}$, 对所有$t\in{\mathbb{R}}^{+}$成立.

(a$_{1}$)  设存在常数$\eta\geq1$, 有$\widehat{a}(t):=\int^{t}_{0}a(s){\rm d}s\geq\frac{1}{\eta}a(t)t$, 对所有$t\in{\mathbb{R}}^{+}$成立.

(b$_{0}$)  设存在常数$b_{0}>0$$b_{1}>0$, 有$b_{0}\leq b(t)\leq b_{1}$, 对所有$t\in{\mathbb{R}}$成立.

(b$_{1}$)  设$\widehat{b}(t):=\int^{t}_{0}b(s){\rm d}s\leq b(t)t$, 对所有$t\in{\mathbb{R}}$成立.

($f_{0}$)  设$f(x, u)\in C( \bar{\Omega}\times {\mathbb{R}}, {\mathbb{R}} )$, 且存在常数$c>0$, 有

$ |f(x, u)|\leq c (1+|u|^{\alpha(x)-1}), $

对所有$u\in{\mathbb{R}}$$x\in\Omega$成立, 其中$C_{+}(\bar{\Omega})=\{h | h\in C(\bar{\Omega}), h(x)>1, \forall x\in\bar{\Omega}\}, $ $\alpha(x)\in C_{+}(\bar{\Omega})$, 及$p^{+} <\alpha^{-}:=\inf\limits_{\Omega}\alpha(x)\leq\alpha^{+}:=\sup\limits_{\Omega}\alpha(x) <p^{*}(x)$, 及$p^{*}(x)$为临界指数, 即当$N>p(x)$时, $p^{*}(x)=\frac{Np(x)}{N-p(x)}$; 当$N\leq p(x)$时, $p^{*}(x)=+\infty$.

($f_{1}$)  设$\lim\limits_{|u|\rightarrow\infty}\frac{F(x, u)}{|u|^{\eta p^{+}}}=+\infty$, 对$x\in\Omega$一致成立.

($f_{2}$)  记${\cal F}(x, u) : =\frac{1}{\eta p^{+}}f(x, u)u-F(x, u).$设存在常数$c_{1}>0$, $c_{2}>0$, $L>0$, $\theta\in(0, 1)$$\sigma>\frac{p^{*}}{(1-\theta)(p^{*}-p^{-})}$, 有

(ⅰ)   ${\cal F}(x, u)\geq c_{1} |u|^{p^{-}}$, 对$|u|\geq L$成立;

(ⅱ)   $|f(x, u)|^{\sigma}\leq c_{2} |u|^{\sigma(p^{-}-1)} {\cal F}(x, u)$, 对$|u|\geq L$成立.

($f_{3}$)   $f(x, -u)=-f(x, u)$, $f(x, 0)=0$, 对所有$x\in\Omega$$u\in{\mathbb{R}}$成立.

定理1.1  假设条件($a_{0}$), ($a_{1}$), ($b_{0}$), ($b_{1}$), ($f_{0}$)-($f_{3}$)都成立.则问题(1.1)有一列解$\{u_{k}\}_{k\in {\Bbb N}}$满足:当$k\rightarrow+\infty$时, 有

$ \widehat{a}\left(\int_{\Omega} \frac{|\nabla u_{k}|^{p(x)}+\sqrt{1+|\nabla u_{k}|^{2p(x)}}}{p(x)} {\rm d}x\right)-\widehat{b}\left(\int_{\Omega} F(x, u_{k}) {\rm d}x\right)\rightarrow+\infty. $

注1.1  令$a(t)=C+D t^{m}$, $b(t)=\arctan t+\pi, $其中$C>0$, $D\geq0$$m>0$.则条件(a$_{0}$), (a$_{1}$), (b$_{0}$), (b$_{10}$)成立.当$a(t)=b(t)=1, $ $p(x)=2$时, 条件(AR)可以推出条件($f_{1}$), ($f_{2}$)成立[19].另一方面, 令$a(t)=1+\frac{\cos t}{1+t^{2}}$, 则$a(t)$满足条件(a$_{0}$)-(a$_{1}$), 此时令

$ f(x, u)=\beta(x) |u|^{2p^{+}-2} u\ln(|u|+1), \beta(x)>0, $

$f$满足条件$({f}_{0}$)-($f_{3}$), 但不满足条件(AR).

定理1.2  假设条件($a_{0}$), ($a_{1}$), ($b_{0}$), ($b_{1}$), ($f_{0}$), ($f_{1}$), ($f_{3}$)都成立, 且下面条件成立:

($f_{4}$)  设存在常数$\eta p^{+} <\mu$, $\Theta\geq0$$1\leq\gamma <p^{-}$, 有

$ \begin{array}{l} &\mu F(x, u)\leq f(x, u)u+\Theta|u|^{\gamma}, \end{array} $

对所有$x\in\Omega$$|u|\geq R$成立, 其中$R$为正常数.

那么, 问题(1.1)有一列解$\{u_{k}\}_{k\in {\Bbb N}}$满足:当$k\rightarrow+\infty$时, 有

$ \widehat{a}\left(\int_{\Omega} \frac{|\nabla u_{k}|^{p(x)}+\sqrt{1+|\nabla u_{k}|^{2p(x)}}}{p(x)} {\rm d}x\right)-\widehat{b}\left(\int_{\Omega} F(x, u_{k}) {\rm d}x\right)\rightarrow+\infty. $

注1.2  易见, ($f_{4}$)比条件(AR)弱.令$a(t)=b(t)\equiv1$, 则$\eta=1$.取$\Theta=0$, 及

$ F(x, u)=g(x) |u|^{p^{+}+2}, $

其中$g(x)>0$, $g(x)\in L^{\infty}(\Omega)$.则条件($f_{4}$)成立.

最后, 当非线性项在零点附近次线性增长时, 我们将利用文献[23]中建立的对称山路定理得到问题(1.1)无穷多个解存在的充分条件.假设

($f_{5}$)  设$\lim\limits_{|u|\rightarrow0}\frac{F(x, u)}{|u|^{\eta p^{-}}}=+\infty$, 对$x\in\Omega$一致成立.

定理1.3  假设条件($a_{0}$), ($a_{1}$), ($b_{0}$), ($b_{1}$), ($f_{0}$), ($f_{3}$), ($f_{5}$)都成立.那么, 问题(1.1)有一列解$\{u_{k}\}_{k\in {\Bbb N}}$满足:当$k\rightarrow+\infty$时, 有

$ \widehat{a}\left(\int_{\Omega} \frac{|\nabla u_{k}|^{p(x)}+\sqrt{1+|\nabla u_{k}|^{2p(x)}}}{p(x)} {\rm d}x\right)-\widehat{b}\left(\int_{\Omega} F(x, u_{k}) {\rm d}x\right)\rightarrow0^{-}. $

注1.3  对问题(1.1), 目前还没有类似结论.令$p(x)\equiv2$, $a(t)=1+\frac{\sin t}{1+t^{2}}$, 则$\eta=1$.令

$ F(x, u)=d(x) |u|^{\frac{3}{2}} \ln\left(\frac{1+|u|^{2}}{4}\right), $

其中$d(x)>0$, $d(x)\in L^{\infty}(\Omega).$则定理1.3中条件($f_{0}$), ($f_{3}$)和($f_{5}$)成立.

2 预备知识

记变指数勒贝格空间$L^{p(x)}(\Omega)$

$ L^{p(x)}(\Omega)=\left\{u:\Omega\rightarrow{\mathbb{R}}, u\in S(\Omega) \Big| \int_{\Omega}|u(x)|^{p(x)}{\rm d}x < +\infty\right\}, $

其中$S(\Omega)$表示$\Omega$中可测的实值函数组成的集合, 范数为

$ |u|_{L^{p(x)}(\Omega)}=|u|_{p(x)}=\inf\left\{\lambda>0 \Big| \int_{\Omega}\left|\frac{u(x)}{\lambda}\right|^{p(x)}{\rm d}x\leq1\right\}. $

引理2.1[9]  空间$(L^{p(x)}(\Omega), |\cdot|_{p(x)})$是可分, 一致凸和自反的空间.令$\frac{1}{p(x)}+\frac{1}{q(x)}=1$.对于$u\in L^{p(x)}(\Omega)$, $v\in L^{q(x)}(\Omega)$, 有

$ \left|\int_{\Omega} uv {\rm d}x\right|\leq\left(\frac{1}{p^{-}}+\frac{1}{q^{-}}\right)|u|_{p(x)}|v|_{q(x)}\leq2|u|_{p(x)}|v|_{q(x)}. $

记变指数索伯列夫空间$W^{1, p(x)}(\Omega)$

$ W^{1, p(x)}(\Omega)=\{u\in L^{p(x)}(\Omega) \mid|\nabla u|\in L^{p(x)}(\Omega)\}, $

其范数为

$ \|u\|=|u|_{p(x)}+|\nabla u|_{p(x)}. $

$W_{0}^{1, p(x)}$$C_{0}^{\infty}(\Omega)$在空间$W^{1, p(x)}$中的闭包.

引理2.2[9]  (ⅰ)   $W^{1, p(x)}(\Omega)$$W_{0}^{1, p(x)}(\Omega)$都是自反可分的巴拿赫空间;

(ⅱ)  若$q(x)\in C_{+}(\bar{\Omega})$, 且$q(x) <p^{*}(x), \forall x\in\bar{\Omega}$, 则$W_{0}^{1, p(x)}(\Omega)\hookrightarrow L^{q(x)}(\Omega)$是紧嵌入;

(ⅲ)  存在$C>0$, 有$|u|_{p(x)}\leq C|\nabla u|_{p(x)}, \hspace{1.5mm}\forall u\in W_{0}^{1, p(x)}(\Omega)$.

引理2.3[9]  记

$ J(u)=\int_{\Omega}|\nabla u|^{p(x)}{\rm d}x, \forall u\in W_{0}^{1, p(x)}(\Omega). $

(ⅰ)   $\|u\|\geq1\Rightarrow\|u\|^{p^{-}}\leq J(u)\leq\|u\|^{p^{+}}$;

(ⅱ)   $\|u\|\leq1\Rightarrow\|u\|^{p^{+}}\leq J(u)\leq\|u\|^{p^{-}}$.

$X$为可分自反的巴拿赫空间, 则存在$\{e_{j}\}\subset X$, 使得$X_{j}={\rm span}\{e_{j}\}$, $Y_{k}=\bigoplus\limits_{j=1}^{k}X_{j}$, $Z_{k}=\overline{\bigoplus\limits_{j\geq k}X_{j}}$, 则$X=\bar{{\rm span}\{e_{j}| j\in{\Bbb N}\}}=Y_{k}\oplus Z_{k}$.

本文用到的临界点定理为:

引理2.4[22] (喷泉定理)  若泛函$I\in C^{1}(X, {\mathbb{R}})$, 满足: $I(0)=0$, $I(-u)=I(u)$, 且

(ⅰ)   $I$满足(C)条件, 即对任何点列$\{u_{n}\}\subset X$, 由$\{I(u_{n})\}$有界, $(1+\|u_{n}\|)\|I'(u_{n})\|\rightarrow0$ $(n\rightarrow+\infty)$, 蕴含$\{u_{n}\}$有收敛子列;

(ⅱ)  对于$k\in{\Bbb N}$, 存在$r_{k}>0$, 当$k\rightarrow+\infty$时, 有$\xi_{k}:=\inf_{u\in Z_{k}, \parallel u\parallel=r_{k}}I(u)\rightarrow+\infty$;

(ⅲ)  对于$\rho_{k}>r_{k}>0$, 有$\alpha_{k}:=\max_{u\in Y_{k}, \parallel u\parallel=\rho_{k}}I(u)\leq0$.

那么, 泛函$I$有一列趋向于$+\infty$的临界值.

为了证明定理1.3, 我们使用Kajikiya在文献[24]中建立的对称山路定理.设$X$为巴拿赫空间并且$\Gamma_{k}=\{A$: $A$$X$中关于原点对称的闭子集, $\{0\}\nsubseteq A$, $\gamma(A)\geq k\}$, 其中$\gamma(A)$表示集合$A$的亏格, 即$\gamma(A):=\inf\{m\in{\Bbb N}: \exists h\in C(A, {\mathbb{R}}^{m}\backslash\{0\}), -h(x)=h(-x)\}$.若对任意的$m\in{\Bbb N}$这样的映射$h$不存在, 则$\gamma(A)=+\infty$.

引理2.5[23] (对称山路定理)  记$X$为无穷维的巴拿赫空间, 设$\varphi\in C^{1}(X, {\mathbb{R}})$为偶泛函, $\varphi(0)=0$且满足下列条件:

(ⅰ)  泛函$\varphi$下方有界且满足(PS)条件, 即对任何点列$\{u_{n}\}\subset X$, 由$\{I(u_{n})\}$有界, $I'(u_{n})\rightarrow0 (n\rightarrow\infty)$, 蕴含$\{u_{n}\}$有收敛子列;

(ⅱ)  对任何$k\in{\Bbb N}$, 存在$A_{k}\in\Gamma_{k}$, 有$\sup_{u\in A_{k}}\varphi(u) <0.$

那么, 泛函$\varphi$有一列临界点$\{u_{k}\}$满足$\varphi(u_{k})\leq0$, $u_{k}\neq0$且当$k\rightarrow+\infty$, 有$\|u_{k}\|\rightarrow0$.

3 定理1.1的证明

$u\in W_{0}^{1, p(x)}(\Omega)$为问题(1.1)的弱解, 指对$\forall v\in W_{0}^{1, p(x)}(\Omega)$, 有

$ a\left(\psi(u)\right) \int_{\Omega}\left(|\nabla u|^{p(x)-2}+\frac{|\nabla u|^{2p(x)-2}}{\sqrt{1+|\nabla u|^{2p(x)}}}\right)\nabla u\nabla v {\rm d}x=b\left( \int_{\Omega} F(x, u) {\rm d}x \right)\int_{\Omega}f(x, u)v {\rm d}x, $

其中$\psi(u)=\int_{\Omega} \frac{|\nabla u|^{p(x)}+\sqrt{1+|\nabla u|^{2p(x)}}}{p(x)} {\rm d}x$, $F(x, u)=\int_{0}^{u} f(x, s){\rm d}s.$

$W_{0}^{1, p(x)}(\Omega)$上定义能量泛函$I$如下

$ I(u)=\widehat{a}\left(\psi(u)\right)-\widehat{b}\left(\int_{\Omega} F(x, u) {\rm d}x\right), $

其中$\widehat{a}(t)=\int^{t}_{0}a(s) {\rm d}s$, $\widehat{b}(t)=\int^{t}_{0}b(s) {\rm d}s$.则$u\in W_{0}^{1, p(x)}(\Omega)$是问题(1.1)的解等价于$u$是泛函$I$的临界点.且$I\in C^{1}(W_{0}^{1, p(x)}(\Omega), {\mathbb{R}})$, 对$\forall u, v\in W_{0}^{1, p(x)}(\Omega)$, 有

$ \begin{array}{l} &\langle I'(u), v\rangle=a\left(\psi(u)\right) \int_{\Omega} \left( |\nabla u|^{p(x)-2}+\frac{|\nabla u|^{2p(x)-2}}{\sqrt{1+|\nabla u|^{2p(x)}}} \right)\nabla u\nabla v {\rm d}x\\ &\hspace{20mm}-b\left( \int_{\Omega} F(x, u) {\rm d}x \right)\int_{\Omega} f(x, u)v {\rm d}x. \end{array} $

$c_{i} (i=1, 2, 3, \cdots)$表示不同的正常数.

引理3.1  若($a_{0}$), ($a_{1}$), ($b_{0}$), ($b_{1}$), ($f_{0}$)和($f_{2}$)都成立, 则泛函$I$满足(C)条件.

  设$\{u_{n}\}\subset W_{0}^{1, p(x)}(\Omega)$为泛函$I$的(C)序列, 则

$ \begin{equation}\mid I(u_{n})\mid\leq c_{3}, (1+\| u_{n}\|)\|I'(u_{n})\|\leq c_{3}. \end{equation} $ (3.1)

首先, 证明序列$\{u_{n}\}$$W_{0}^{1, p(x)}(\Omega)$中有界.反设$\{u_{n}\}$$W_{0}^{1, p(x)}(\Omega)$中无界, 则当$n\rightarrow\infty$时, 有

$ \begin{equation}\|u_{n}\|\rightarrow+\infty.\end{equation} $ (3.2)

由条件(a$_{0}$), (a$_{1}$), (b$_{0}$), (b$_{1}$)和(3.1)式, 有

$ \begin{array}{l} I(u) = \hat a\left( {\int_\Omega {\frac{{|\nabla u{|^{p(x)}} + \sqrt {1 + |\nabla u{|^{2p(x)}}} }}{{p(x)}}} {\rm{d}}x} \right) - \hat b\left( {\int_\Omega F (x,u){\rm{d}}x} \right)\\ \ge \frac{1}{\eta }a\left( {\int_\Omega {\frac{{|\nabla u{|^{p(x)}} + \sqrt {1 + |\nabla u{|^{2p(x)}}} }}{{p(x)}}} {\rm{d}}x} \right)\left( {\int_\Omega {\frac{{|\nabla u{|^{p(x)}} + \sqrt {1 + |\nabla u{|^{2p(x)}}} }}{{p(x)}}} {\rm{d}}x} \right)\\ - b\left( {\int_\Omega F (x,u){\rm{d}}x} \right)\int_\Omega F (x,u){\rm{d}}x\\ \ge \frac{{{a_0}}}{\eta }\int_\Omega {\frac{{|\nabla u{|^{p(x)}} + \sqrt {1 + |\nabla u{|^{2p(x)}}} }}{{p(x)}}} {\rm{d}}x - {b_1}\int_\Omega F (x,u){\rm{d}}x\\ \ge \frac{{{a_0}}}{{\eta {p^ + }}}\int_\Omega | \nabla u{|^{p(x)}}{\rm{d}}x - \frac{{{b_1}c}}{{{\alpha ^ - }}}\int_\Omega | u{|^{\alpha (x)}}{\rm{d}}x - {b_1}c\parallel u{\parallel _{{L^1}(\Omega )}}\\ \ge \frac{{{a_0}}}{{\eta {p^ + }}}\parallel u{\parallel ^{{p^ - }}} - \frac{{{b_1}c}}{{{\alpha ^ - }}}\int_\Omega | u{|^{\alpha (x)}}{\rm{d}}x - {c_7}\parallel u\parallel \\ \ge \left\{ \begin{array}{l} \frac{{{a_0}}}{{\eta {p^ + }}}\parallel u{\parallel ^{{p^ - }}} - \frac{{{b_1}c}}{{{\alpha ^ - }}} - {c_7}\parallel u\parallel ,\;\;\;\;\;|u{|_{\alpha (x)}}\\ \frac{{{a_0}}}{{\eta {p^ + }}}\parallel u{\parallel ^{{p^ - }}} - \frac{{{b_1}c}}{{{\alpha ^ - }}}{({\beta _k}\parallel u\parallel )^{{\alpha ^ + }}} - {c_7}\parallel u\parallel ,|u{|_{\alpha (x)}} > 1 \end{array} \right.\\ \ge \frac{{{a_0}}}{{\eta {p^ + }}}\parallel u{\parallel ^{{p^ - }}} - \frac{{{b_1}c}}{{{\alpha ^ - }}}{({\beta _k}\parallel u\parallel )^{{\alpha ^ + }}} - {c_7}\parallel u\parallel - \frac{{{b_1}c}}{{{\alpha ^ - }}}\\ = \parallel u{\parallel ^{{p^ - }}}\left[ {\frac{{{a_0}}}{{\eta {p^ + }}} - \frac{{{b_1}c}}{{{\alpha ^ - }}}{{({\beta _k})}^{{\alpha ^ + }}}\parallel u{\parallel ^{{\alpha ^ + } - {p^ - }}}} \right] - {c_7}\parallel u\parallel - \frac{{{b_1}c}}{{{\alpha ^ - }}}.\\ \end{array} $

$ \begin{equation}\left(1+\frac{1}{\eta p^{+}}\right)c_{3}\geq b_{0}\int_{\Omega} \left[ \frac{1}{\eta p^{+}}f(x, u_{n})u_{n}-F(x, u_{n}) \right]{\rm d}x. \end{equation} $ (3.3)

由条件($f_{0}$), 有

$ \left|\frac{1}{\eta p^{+}}f(x, u)u-F(x, u)\right|\leq\frac{c}{\eta p^{+}}\left(|L|+|L|^{\alpha^{+}}\right)+c|L|+\frac{c}{\alpha^{-}}|L|^{\alpha^{+}} $

对所有$x\in\Omega$$|u|\leq L$成立, 其中$\alpha(x)>1$, $\alpha^{-}=\inf\limits_{\Omega}\alpha(x)$$\alpha^{+}=\sup\limits_{\Omega}\alpha(x)$.再结合条件($f_{2}$)-(ⅰ), 可得

$ {\cal F}(x, u)=\frac{1}{\eta p^{+}}f(x, u)u-F(x, u)\geq c_{1}|u|^{p^{-}}-c_{4} $

对所有$x\in\Omega$$u\in {\mathbb{R}}$成立.因此

$ \begin{equation}\left(1+\frac{1}{\eta p^{+}}\right)c_{3}\geq\left[ c_{1}\int_{\Omega}|u_{n}|^{p^{-}}{\rm d}x-c_{4}|\Omega| \right]b_{0}.\end{equation} $ (3.4)

$v_{n}=\frac{u_{n}}{\|u_{n}\|}$, 则$\|v_{n}\|=1$.由(3.2)式和(3.4)式, 当$n\rightarrow\infty$时, 有

$ \begin{equation}\int_{\Omega} |v_{n}|^{p^{-}} {\rm d}x=\frac{1}{\|u_{n}\|^{p^{-}}}\int_{\Omega} |u_{n}|^{p^{-}} {\rm d}x\leq\frac{c_{5}}{\|u_{n}\|^{p^{-}}}\rightarrow0. \end{equation} $ (3.5)

$\sigma>\frac{p^{*}}{(1-\theta)(p^{*}-p^{-})}$, $\theta\in(0, 1)$$p^{*}>p^{-}$, 可推得

$ \sigma>1, \frac{\sigma-1}{\theta\sigma}>1, \frac{(1-\theta)\sigma p^{-}}{(1-\theta)\sigma-1} <p^{*}. $

利用引理2.2, $X\hookrightarrow L^{\frac{(1-\theta)\sigma p^{-}}{(1-\theta)\sigma-1}}(\Omega)$是紧嵌入, 故有

$ \begin{equation}\int_{\Omega} |v_{n}|^{\frac{(1-\theta)\sigma p^{-}}{(1-\theta)\sigma-1}} {\rm d}x\leq c_{6}\|v_{n}\|^{\frac{(1-\theta)\sigma-1}{(1-\theta)\sigma p^{-}}}=c_{6}.\end{equation} $ (3.6)

利用Hölder不等式及(3.6)式, 有

$ \begin{array}{l} \int_{\Omega} |v_{n}|^{p^{-}\sigma'} {\rm d}x &=&\int_{\Omega} |v_{n}|^{\frac{ \sigma p^{-}}{\sigma-1}} {\rm d}x =\int_{\Omega} |v_{n}|^{\frac{ \theta\sigma p^{-}}{\sigma-1}}|v_{n}|^{\frac{(1-\theta)\sigma p^{-}}{\sigma-1}} {\rm d}x\\ &\leq&\left[ \int_{\Omega}\left(|v_{n}|^{\frac{ \theta\sigma p^{-}}{\sigma-1}}\right)^{\frac{\sigma-1}{\theta\sigma}}{\rm d}x \right]^{\frac{\theta\sigma}{\sigma-1}}\left[ \int_{\Omega}\left(|v_{n}|^{\frac{(1-\theta)\sigma p^{-}}{\sigma-1}}\right)^{\frac{1}{1-\frac{\theta\sigma}{\sigma-1}}}{\rm d}x \right]^{1-\frac{\theta\sigma}{\sigma-1}}\\ &=&\left(\int_{\Omega} |v_{n}|^{p^{-}} {\rm d}x\right)^{\frac{\theta\sigma}{\sigma-1}}\left(\int_{\Omega} |v_{n}|^{\frac{ (1-\theta)\sigma p^{-}}{(1-\theta)\sigma-1}} {\rm d}x\right)^{1-\frac{\theta\sigma}{\sigma-1}}\\ &\leq&\left(\int_{\Omega} |v_{n}|^{p^{-}} {\rm d}x\right)^{\frac{\theta\sigma}{\sigma-1}}(c_{6})^{1-\frac{\theta\sigma}{\sigma-1}}, \end{array} $

其中$\frac{1}{\sigma}+\frac{1}{\sigma'}=1$.结合(3.5)式, 当$n\rightarrow\infty$时, 有

$ \begin{equation}\int_{\Omega} |v_{n}|^{p^{-}\sigma'} {\rm d}x\rightarrow0.\end{equation} $ (3.7)

利用条件($f_{2}$)的(ⅱ), (3.3)式和Hölder不等式, 可得

$ \begin{array}{l} \left| \int_{\Omega} \frac{f(x, u_{n})u_{n}}{\|u_{n}\|^{p^{-}}} {\rm d}x \right|&\leq&\int_{\Omega} \frac{|f(x, u_{n})|}{|u_{n}|^{p^{-}-1}} |v_{n}|^{p^{-}} {\rm d}x\\ &\leq&\left(\int_{\Omega}\left(\frac{|f(x, u_{n}) |}{|u_{n}|^{p^{-}-1}}\right)^{\sigma}{\rm d}x\right)^{\frac{1}{\sigma}}\left(\int_{\Omega} |v_{n}|^{p^{-}\sigma'} {\rm d}x\right)^{\frac{1}{\sigma'}}\\ &\leq&\left[c_{2}\int_{\Omega} {\cal F}(x, u_{n}) {\rm d}x\right]^{\frac{1}{\sigma}}\left(\int_{\Omega} |v_{n}|^{p^{-}\sigma'} {\rm d}x\right)^{\frac{1}{\sigma'}}\\ &=&\left\{c_{2}\int_{\Omega} \left[ \frac{1}{\eta p^{+}}f(x, u_{n})u_{n}-F(x, u_{n}) \right] {\rm d}x\right\}^{\frac{1}{\sigma}}\left(\int_{\Omega} |v_{n}|^{p^{-}\sigma'} {\rm d}x\right)^{\frac{1}{\sigma'}}\\ &\leq&\left[\frac{c_{2}}{b_{0}} \left(1+\frac{1}{\eta p^{+}}\right)c_{3}\right]^{\frac{1}{\sigma}}\left(\int_{\Omega} |v_{n}|^{p^{-}\sigma'} {\rm d}x\right)^{\frac{1}{\sigma'}}. \end{array} $

再结合(3.7)式, 当$n\rightarrow\infty$时, 可得

$ \begin{equation}\int_{\Omega} \frac{f(x, u_{n})u_{n}}{\|u_{n}\|^{p^{-}}} {\rm d}x\rightarrow0.\end{equation} $ (3.8)

由(3.2)式, 可设$\|u_{n}\|\geq1$, 利用引理2.3, (3.1)式, 条件($a_{0}$)和($b_{0}$), 有

$ \begin{array}{l} \parallel I'({u_n})\parallel (1 + \parallel {u_n}\parallel ) \ge \langle I'({u_n}),{u_n}\rangle \\ = a\left( {\psi ({u_n})} \right)\left( {\int_\Omega | \nabla {u_n}{|^{p(x)}}{\rm{d}}x + \int_\Omega {\frac{{|\nabla {u_n}{|^{2p(x)}}}}{{\sqrt {1 + |\nabla {u_n}{|^{2p(x)}}} }}} {\rm{d}}x} \right)\\ - b\left( {\int_\Omega F (x,{u_n}){\rm{d}}x} \right)\int_\Omega f (x,{u_n}){u_n}{\rm{d}}x\\ \ge {a_0}\left( {\int_\Omega | \nabla {u_n}{|^{p(x)}}{\rm{d}}x + \int_\Omega {\frac{{|\nabla {u_n}{|^{2p(x)}}}}{{\sqrt {1 + |\nabla {u_n}{|^{2p(x)}}} }}} {\rm{d}}x} \right)\\ - {b_1}\int_\Omega f (x,{u_n}){u_n}{\rm{d}}x\\ \ge {a_0}\int_\Omega | \nabla {u_n}{|^{p(x)}}{\rm{d}}x - {b_1}\int_\Omega f (x,{u_n}){u_n}{\rm{d}}x\\ \ge \parallel {u_n}{\parallel ^{{p^ - }}}\left( {{a_0} - {b_1}\int_\Omega {\frac{{f(x,{u_n}){u_n}}}{{\parallel {u_n}{\parallel ^{{p^ - }}}}}} {\rm{d}}x} \right). \end{array} $

结合(3.2)式, 令$n\rightarrow\infty$, 有$a_{0}\leq o(1), $其中当$n\rightarrow\infty$时, $o(1)\rightarrow 0$.这与$a_{0}>0$矛盾.故$\{u_{n}\}$$W_{0}^{1, p(x)}(\Omega)$中有界.

因为$W_{0}^{1, p(x)}(\Omega)$是自反的巴拿赫空间, 所有存在$u\in W_{0}^{1, p(x)}(\Omega)$, 使得$\{u_{n}\}$在空间$W_{0}^{1, p(x)}(\Omega)$中弱收敛于$u$, 且$\{u_{n}\}$$L^{\alpha(x)}(\Omega)$中强收敛于$u$.由Hölder不等式, 有

$ \begin{array}{l} \left|\int_{\Omega} f(x, u_{n})(u_{n}-u) {\rm d}x\right| &\leq&\int_{\Omega} |f(x, u_{n})||u_{n}-u| {\rm d}x\\ &\leq& c \int_{\Omega} |1+|u_{n}|^{\alpha(x)-1}| |u_{n}-u| {\rm d}x\\ &\leq&2c |1+|u_{n}|^{\alpha(x)-1}|_{(\alpha'(x))} |u_{n}-u|_{\alpha(x)}. \end{array} $

从而有

$ \begin{equation}\lim\limits_{n\rightarrow\infty}\int_{\Omega} f(x, u_{n})(u_{n}-u) {\rm d}x=0.\end{equation} $ (3.9)

$n\rightarrow\infty$, 有$\langle I'(u_{n}), u_{n}-u\rangle\rightarrow0$, 结合条件(a$_{0}$), (b$_{0}$)及(3.9)式, 可得

$ \langle J(u_{n}), u_{n}-u\rangle=\int_{\Omega} \left(|\nabla u_{n}|^{p(x)-2}\nabla u_{n}+\frac{|\nabla u_{n}|^{2p(x)-2}\nabla u_{n}}{\sqrt{1+|\nabla u_{n}|^{2p(x)}}}\right)(\nabla u_{n}-\nabla u) {\rm d}x\rightarrow0. $

根据文献[17, Proposition 3.1]可知, 连续映射$J: W_{0}^{1, p(x)}(\Omega)\rightarrow (W_{0}^{1, p(x)}(\Omega))^{*}$具有性质$(S_{+})$, 即$\{u_{n}\}$$W_{0}^{1, p(x)}(\Omega)$中弱收敛于$u$, 且$\limsup\limits_{n\rightarrow\infty}\langle J(u_{n})-J(u), u_{n}-u\rangle\leq0$, 可得序列$\{u_{n}\}$$W_{0}^{1, p(x)}(\Omega)$中强收敛于$u$.故泛函$I$满足(C)条件.

$X:=W_{0}^{1, p(x)}(\Omega)$, $X$为可分自反的巴拿赫空间, 则存在$\{e_{j}\}\subset X$, 使得$X_{j}={\rm span}\{e_{j}\}$, $Y_{k}=\bigoplus\limits_{j=1}^{k}X_{j}$, $Z_{k}=\overline{\bigoplus\limits_{j\geq k}X_{j}}$, 则

$ X=\bar{{\rm span}\{e_{j}| j\in{\Bbb N}\}}=Y_{k}\oplus Z_{k}. $

引理3.2[9]  若$\alpha\in C_{+}(\overline{\Omega})$, $\alpha(x) <p^{*}(x)$, $\forall x\in\overline{\Omega}$, 令

$ \beta_{k}=\sup \{ |u|_{\alpha(x)}: u\in Z_{k}, \|u\|=1 \}, $

那么, 当$k\rightarrow+\infty$时, 有$\beta_{k}\rightarrow0$.

定理1.1的证明  利用引理3.1和($f_{3}$), 可知$I$满足(C)条件, $I(-u)=I(u)$$I(0)=0$.要证明定理1.1成立, 只需验证喷泉定理(引理2.4)中环绕条件(ⅱ)和(ⅲ)成立.

先证(ⅱ)成立.由条件($f_{0}$), 可推出

$ \begin{equation} |F(x, u)|\leq c |u|+\frac{c}{\alpha^{-}} |u|^{\alpha(x)}, \end{equation} $ (3.10)

对所有$x\in\Omega$$u\in {\mathbb{R}}$成立, 其中$\alpha^{-}=\inf\limits_{\Omega}p(x)$.

对于$u\in Z_{k}$, $\|u\|=r_{k}>1$, 由条件(a$_{0}$), (a$_{1}$), (b$_{0}$), (b$_{1}$)和(3.10)式, 有

$ \begin{array}{l} I(u) = \hat a\left( {\int_\Omega {\frac{{|\nabla u{|^{p(x)}} + \sqrt {1 + |\nabla u{|^{2p(x)}}} }}{{p(x)}}} {\rm{d}}x} \right) - \hat b\left( {\int_\Omega F (x,u){\rm{d}}x} \right)\\ \ge \frac{1}{\eta }a\left( {\int_\Omega {\frac{{|\nabla u{|^{p(x)}} + \sqrt {1 + |\nabla u{|^{2p(x)}}} }}{{p(x)}}} {\rm{d}}x} \right)\left( {\int_\Omega {\frac{{|\nabla u{|^{p(x)}} + \sqrt {1 + |\nabla u{|^{2p(x)}}} }}{{p(x)}}} {\rm{d}}x} \right)\\ - b\left( {\int_\Omega F (x,u){\rm{d}}x} \right)\int_\Omega F (x,u){\rm{d}}x\\ \ge \frac{{{a_0}}}{\eta }\int_\Omega {\frac{{|\nabla u{|^{p(x)}} + \sqrt {1 + |\nabla u{|^{2p(x)}}} }}{{p(x)}}} {\rm{d}}x - {b_1}\int_\Omega F (x,u){\rm{d}}x\\ \ge \frac{{{a_0}}}{{\eta {p^ + }}}\int_\Omega | \nabla u{|^{p(x)}}{\rm{d}}x - \frac{{{b_1}c}}{{{\alpha ^ - }}}\int_\Omega | u{|^{\alpha (x)}}{\rm{d}}x - {b_1}cu{_{{L^1}(\Omega )}}\\ \ge \frac{{{a_0}}}{{\eta {p^ + }}}u{^{{p^ - }}} - \frac{{{b_1}c}}{{{\alpha ^ - }}}\int_\Omega | u{|^{\alpha (x)}}{\rm{d}}x - {c_7}u\\ \ge \left\{ \begin{array}{l} \frac{{{a_0}}}{{\eta {p^ + }}}u{^{{p^ - }}} - \frac{{{b_1}c}}{{{\alpha ^ - }}} - {c_7}u,\;\;\;\;\;|u{|_{\alpha (x)}}\\ \frac{{{a_0}}}{{\eta {p^ + }}}u{^{{p^ - }}} - \frac{{{b_1}c}}{{{\alpha ^ - }}}{({\beta _k}u)^{{\alpha ^ + }}} - {c_7}u,|u{|_{\alpha (x)}} > 1 \end{array} \right.\\ \ge \frac{{{a_0}}}{{\eta {p^ + }}}u{^{{p^ - }}} - \frac{{{b_1}c}}{{{\alpha ^ - }}}{({\beta _k}u)^{{\alpha ^ + }}} - {c_7}u - \frac{{{b_1}c}}{{{\alpha ^ - }}}\\ = u{^{{p^ - }}}\left[ {\frac{{{a_0}}}{{\eta {p^ + }}} - \frac{{{b_1}c}}{{{\alpha ^ - }}}{{({\beta _k})}^{{\alpha ^ + }}}u{^{{\alpha ^ + } - {p^ - }}}} \right] - {c_7}u - \frac{{{b_1}c}}{{{\alpha ^ - }}}. \end{array} $

$r_{k}:=\left[ \frac{b_{1}c}{\alpha^{-}}(\beta_{k})^{\alpha^{+}}\frac{2\eta p^{+}}{a_{0}} \right]^{1/(p^{-}-\alpha^{+})}$, 对$u\in Z_{k}$, $\|u\|=r_{k}$, 有

$ I(u)\geq r_{k}^{p^{-}}\frac{a_{0}}{2\eta p^{+}}-c_{7}r_{k}-\frac{b_{1}c}{\alpha^{-}}. $

注意到当$k\rightarrow+\infty$时, 有$\beta_{k}\rightarrow0$.则有$r_{k}\rightarrow+\infty$, 从而当$k\rightarrow+\infty$时, 有

$ \xi_{k}:=\inf\limits_{u\in Z_{k}, \parallel u\parallel=r_{k}}I(u)\rightarrow+\infty, $

这表明(ⅱ)成立.

下面验证(ⅲ)成立.取$t_{1}>0$, 利用条件(a$_{1}$), 可得$\frac{a(t)}{\widehat{a}(t)}\leq\frac{\eta}{t}, $对所有$t\in[t_{1}, +\infty)$成立.对此不等式积分, 有

$ \ln\frac{\widehat{a}(t)}{\widehat{a}(t_{1})}=\int_{t_{1}}^{t} \frac{a(s)}{\widehat{a}(s)} {\rm d}s\leq\int_{t_{1}}^{t} \frac{\eta}{s} {\rm d}s=\ln\left(\frac{t}{t_{1}}\right)^{\eta}, $

对所有$t\in[t_{1}, +\infty)$成立.上式表明$\widehat{a}(t)\leq\frac{\widehat{a}(t_{1})}{t_{1}^{\eta}}t^{\eta}, $对所有$t\in[t_{1}, +\infty)$成立.从而有

$ \begin{equation}\widehat{a}(t)\leq c_{8} t^{\eta}+c_{9}, \end{equation} $ (3.11)

对所有$t>0$成立, 其中$c_{8}:=\frac{\widehat{a}(t_{1})}{t_{1}^{\eta}}$, $c_{9}:=\max\limits_{t\in[0, t_{1}]}\widehat{a}(t)$.利用条件(b$_{1}$), 类似于(3.11)式的证明, 可得

$ \begin{equation}\widehat{b}(t)\geq c_{10} t+c_{11}, \end{equation} $ (3.12)

对所有$t>0$成立.从而有

$ \begin{array}{l} \widehat{a}(\psi(u))&=&\widehat{a}\left(\int_{\Omega} \frac{|\nabla u|^{p(x)}+\sqrt{1+|\nabla u|^{2p(x)}}}{p(x)} {\rm d}x\right)\\ &\leq& c_{8}\left(\int_{\Omega} \frac{|\nabla u|^{p(x)}+\sqrt{1+|\nabla u|^{2p(x)}}}{p(x)} {\rm d}x\right)^{\eta}+c_{9}\\ &\leq &c_{8}\left(\frac{1}{p^{-}}\int_{\Omega} (2|\nabla u|^{p(x)}+1) {\rm d}x\right)^{\eta}+c_{9}. \end{array} $

对于$u\in Y_{k}$, $\|u\|=\rho_{k}>r_{k}>1$, 利用引理2.3, 有

$ \begin{equation}\widehat{a}(\psi(u)\leq c_{8}\left(\frac{2}{p^{-}}\|u\|^{p^{+}}+\frac{1}{p^{-}}|\Omega|\right)^{\eta}+c_{9}.\end{equation} $ (3.13)

由条件($f_{0}$)和($f_{1}$), 对$\forall M>0$, 存在$c_{12}>0$, 有

$ \begin{equation}F(x, u)\geq M|u|^{\eta p^{+}}-c_{12}, \end{equation} $ (3.14)

对所有$x\in\Omega$$u\in{\mathbb{R}}$成立.由(3.12)式, (3.13)式和(3.14)式, 有

$ \begin{array}{l} I(u)&=&\widehat{a}\left(\int_{\Omega} \frac{|\nabla u|^{p(x)}+\sqrt{1+|\nabla u|^{2p(x)}}}{p(x)} {\rm d}x\right)-\widehat{b}\left(\int_{\Omega}F(x, u){\rm d}x\right)\\ &\leq &c_{8}\left(\int_{\Omega} \frac{|\nabla u|^{p(x)}+\sqrt{1+|\nabla u|^{2p(x)}}}{p(x)} {\rm d}x\right)^{\eta}-c_{10}\left(\int_{\Omega} F(x, u) {\rm d}x\right)+c_{12}\\ &\leq& c_{8}\left( \frac{2}{p^{-}}\|u\|^{p^{+}}+\frac{1}{p^{-}}|\Omega| \right)^{\eta}-Mc_{10}\int_{\Omega} |u|^{\eta p^{+}} {\rm d}x+c_{13}. \end{array} $

由于$\dim Y_{k} <+\infty$, 有限维空间中范数是等价的, 取$M$充分大, 则

$ \alpha_{k}:=\max\limits_{u\in Y_{k}, \parallel u\parallel=\rho_{k}}\varphi(u)\leq0, $

这表明(ⅲ)成立.

4 定理1.2的证明

引理4.1  若条件($a_{0}$), ($a_{1}$), ($b_{0}$), ($b_{1}$), ($f_{0}$)和($f_{4}$)都成立.则泛函$I$满足(C)条件.

  设$\{u_{n}\}\subset W_{0}^{1, p(x)}(\Omega)$为泛函$I$的(C)序列, 则

$ \begin{equation}\mid I(u_{n})\mid\leq c_{3}, (1+\| u_{n}\|)\|I'(u_{n})\|\leq c_{3}, \end{equation} $ (4.1)

由条件($f_{0}$)和($f_{4}$), 可知

$ \begin{equation}\mu F(x, u)-f(x, u)u\leq\Theta|u|^{\gamma}+c_{14}, \end{equation} $ (4.2)

对所有$x\in\Omega$$u\in {\mathbb{R}}$成立.由$1\leq\gamma <p^{-} <p^{*}(x)$和引理2.2, $W_{0}^{1, p(x)}(\Omega)\hookrightarrow L^{\gamma}(\Omega)$是紧嵌入, 则存在$c_{15}>0$, 对$u\in W_{0}^{1, p(x)}(\Omega)$, 有

$ \begin{equation}\int_{\Omega} |u(x)|^{\gamma} {\rm d}x\leq c_{15} \|u\|^{\gamma}.\end{equation} $ (4.3)

由条件($a_{0}$), ($a_{1}$), ($b_{0}$), ($b_{1}$), (4.1)式, (4.2)和(4.3)式, 引理2.3, 对$\|u\|>1$, 有

$ \begin{array}{l} &a_{0}\left(\frac{\mu}{\eta p^{+}}-1\right)\|u_{n}\|^{p^{-}} \\ &\leq a_{0}\left(\frac{\mu}{\eta p^{+}}-1\right)\int_{\Omega}|\nabla u_{n}|^{p(x)} {\rm d}x\\ &\leq a\left(\psi(u_{n})\right)\left(\frac{\mu}{\eta p^{+}}-1\right)\int_{\Omega}\left[|\nabla u_{n}|^{p(x)}+\sqrt{1+|\nabla u_{n}|^{2p(x)}}\right]{\rm d}x\\ &\leq \frac{\mu}{\eta} a\left(\psi(u_{n})\right)\psi(u_{n}) -a\left(\psi(u_{n})\right)\int_{\Omega}\left[|\nabla u_{n}|^{p(x)}+\sqrt{1+|\nabla u_{n}|^{2p(x)}}\right]{\rm d}x\\ &\leq \mu \widehat{a}\left(\psi(u_{n})\right)-a\left(\psi(u_{n})\right)\int_{\Omega}\left[|\nabla u_{n}|^{p(x)}+\frac{|\nabla u_{n}|^{2p(x)}}{\sqrt{1+|\nabla u_{n}|^{2p(x)}}}\right]{\rm d}x\\ &=\left[\mu\widehat{b}\left(\int_{\Omega}F(x, u_{n}){\rm d}x\right)-b\left(\int_{\Omega}F(x, u_{n}){\rm d}x\right)\int_{\Omega}f(x, u_{n})u_{n}{\rm d}x\right] +\mu I(u_{n})-\langle I'(u_{n}), u_{n}\rangle\\ &\leq b\left(\int_{\Omega}F(x, u_{n}){\rm d}x\right) \int_{\Omega} [ \mu F(x, u_{n})-f(x, u_{n})u_{n}] {\rm d}x +\mu I(u_{n})+\|I'(u_{n})\|(1+\|u_{n}\|)\\ &\leq b_{1} \int_{\Omega} [ \mu F(x, u_{n})-f(x, u_{n})u_{n} ] {\rm d}x+c_{16}\\ &\leq b_{1}\Theta\int_{\Omega} |u_{n}(x)|^{\gamma} {\rm d}x+c_{17}\\ &\leq b_{1}\Theta c_{15} \|u_{n}\|^{\gamma}+c_{17}, \end{array} $

其中$\psi(u)=\int_{\Omega} \frac{|\nabla u|^{p(x)}+\sqrt{1+|\nabla u|^{2p(x)}}}{p(x)} {\rm d}x$.由$a_{0}>0$, $\mu>\eta p^{+}$$1\leq\gamma <p^{-}$, 可得$\{u_{n}\}$$W_{0}^{1, p(x)}(\Omega)$中有界.类似于引理3.1的证明, 序列$\{u_{n}\}$$W_{0}^{1, p(x)}(\Omega)$中强收敛于$u$.故泛函$I$满足(C)条件.

定理1.2的证明  利用引理4.1和($f_{3}$), 泛函$I$满足(C)条件, $I(-u)=I(u)$$I(0)=0$.类似于定理1.1的证明, 可以证得喷泉定理(引理2.4)中环绕条件(ii)和(iii)成立.故定理1.2的结论成立.

5 定理1.3的证明

我们使用Kajikiya在文献[23]中建立的对称山路定理(引理2.5)证明定理1.3.

定理1.3的证明  对$\forall u\in X:=W_{0}^{1, p(x)}$, 定义阶段函数

$ \varphi(u)=\widehat{a}\left(\int_{\Omega} \frac{|\nabla u|^{p(x)}+\sqrt{1+|\nabla u|^{2p(x)}}}{p(x)} {\rm d}x\right)-g(\|u\|) \widehat{b}\left(\int_{\Omega} F(x, u) {\rm d}x\right), $

其中$g:{\mathbb{R}}^{+}\rightarrow[0, 1]$为单调递减的可微函数, 对任意$r>0$, 满足: $0\leq g(t)\leq1$, $t\in[0, +\infty).$对任意$r>0$, 当$t\geq r$时, 有$g(t)=0$; 当$0\leq t\leq\frac{r}{2}$时, 有$g(t)=1$.则$\varphi\in C^{1}(X, {\mathbb{R}})$.由条件($f_{3}$), 可得$\varphi(0)=0$, $\varphi(-u)=\varphi(u)$.易见, 当$\|u\|\leq\frac{r}{2}$时, $u\in X$是泛函$\varphi$的临界点, $u\in X$也是泛函$I$的临界点, 即问题(1.1)的解.

$\|u\|\geq r$, 则$g(\|u\|)=0$.利用条件($a_{0}$), ($a_{1}$)和引理2.3, 对$\|u\|>\max\{1, r\}$, 有

$ \begin{array}{l} \varphi(u)&=&\widehat{a}\left(\int_{\Omega} \frac{|\nabla u|^{p(x)}+\sqrt{1+|\nabla u|^{2p(x)}}}{p(x)} {\rm d}x\right)\\ &=&\int^{\int_{\Omega}\frac{|\nabla u|^{p(x)}+\sqrt{1+|\nabla u|^{2p(x)}}}{p(x)}{\rm d}x}_{0}a(s) {\rm d}s\\ &\geq& a_{0}\int_{\Omega} \frac{|\nabla u|^{p(x)}+\sqrt{1+|\nabla u|^{2p(x)}}}{p(x)} {\rm d}x\\ &\geq&\frac{a_{0}}{p^{+}} \int_{\Omega} |\nabla u|^{p(x)} {\rm d}x\\ &\geq&\frac{a_{0}}{p^{+}} \|u\|^{p^{-}}, \end{array} $

因此, 当$\|u\|\rightarrow+\infty$时, 有$\varphi(u)\rightarrow+\infty.$这表明泛函$\varphi$下方有界且满足(PS)条件.

$X:=W_{0}^{1, p(x)}(\Omega)$, $X$为可分自反的巴拿赫空间, 则存在$\{e_{j}\}\subset X$, 使得$X_{j}={\rm span}\{e_{j}\}$, $X={\rm span}\{e_{j}| j\in{\Bbb N}\}.$易见, $E_{k}=\bigoplus\limits_{j=1}^{k}X_{j}$$W_{0}^{1, p(x)}$$k$ -维子空间.利用有限维空间范数等价性, 存在$c_{18}>0$, $c_{19}>0$, 使得

$ \begin{equation}\|u\|^{\eta p^{-}}\leq c_{18}\int_{\Omega}|u|^{\eta p^{-}}{\rm d}x, \forall u\in E_{k}.\end{equation} $ (5.1)
$ \begin{equation}|u|\leq c_{19}\|u\|, \forall u\in E_{k}.\end{equation} $ (5.2)

由条件($f_{5}$), 对$\forall\vartheta>0$, 有

$ \begin{equation}F(x, u)\geq\vartheta |u|^{\eta p^{-}}.\end{equation} $ (5.3)

对所有$|u|\leq\delta$$x\in\Omega$成立.由(3.11)式, (3.14)式和(5.3)式, 引理2.3, 令$r$充分小, 对$\forall u\in E_{k}$, $\|u\|=\rho\leq\min\left\{\frac{r}{2}, \frac{\delta}{c_{16}}\right\} <1$, 有

$ \begin{array}{l} \varphi(u)&=&I(u)\\ &=&\widehat{a}\left(\int_{\Omega} \frac{|\nabla u|^{p(x)} +\sqrt{1+|\nabla u|^{2p(x)}}}{p(x)} {\rm d}x\right)-\widehat{b} \left(\int_{\Omega} F(x, u) {\rm d}x\right)\\ &\leq& c_{8}\left(\int_{\Omega} \frac{|\nabla u|^{p(x)} +\sqrt{1+|\nabla u|^{2p(x)}}}{p(x)} {\rm d}x\right)^{\eta} -c_{10} \int_{\Omega} F(x, u) {\rm d}x+c_{17}\\ &\leq& c_{8}\left(\int_{\Omega} \frac{2 |\nabla u|^{p(x)}+1}{p(x)} {\rm d}x\right)^{\eta}-c_{10} \int_{\Omega} F(x, u) {\rm d}x+c_{20}\\ &\leq &c_{8}\left( \frac{2}{p^{-}}\|u\|^{p^{-}}+\frac{1}{p^{-}}|\Omega| \right)^{\eta}-c_{10} \vartheta\int_{\Omega} |u|^{\eta p^{-}} {\rm d}x+c_{20}, \end{array} $ (5.4)

$\vartheta$充分大, 由(5.1)式和(5.4)式, 可推出$\varphi(u) <0$, 这表明

$ \{u\in X:\varphi(u) <0\}\supset\{u\in E_{k}:\|u\|=\rho\}. $

$A_{k}=\left\{u\in X:\varphi(u) <0\right\}$, 利用亏格$\gamma(A)$的单调性, 有$\gamma(A_{k})\geq\gamma( \{u\in E_{k}:\|u\|=\rho\} ).$因为$\{u\in E_{k}:\|u\|=\rho\}$$E_{k}$的($k-1$)维子空间, 所以$\gamma(\{u\in E_{k}:\|u\|=\rho\})=k$.从而$A_{k}\in\Gamma_{k}$$\sup_{u\in A_{k}}\varphi(u) <0.$

综上, 泛函$\varphi$满足引理2.5的所有条件, 则泛函$\varphi$有一列临界点$\{u_{k}\}$满足:$\varphi(u_{k})\leq0$, $u_{k}\neq0$$\lim\limits_{k\rightarrow+\infty}u_{k}=0$.对$\forall r>0$, 当$\|u_{k}\|\leq\frac{r}{2}$时, $u_{k}\in W_{0}^{1, p(x)}$是泛函$\varphi$的临界点, $u_{k}\in W_{0}^{1, p(x)}$也是泛函$I$的临界点, 即问题(1.1)的解.因此, 定理1.3的结论成立.

参考文献
[1] Ruzicka M. Electrorheologial Fluids:Modeling and Mathematial Throry. Berlin: Springer-Verlag, 2000.
[2] Zhikov V. Averaging of functionals of the calculus of variations and elasticity theory. Math USSR Izv, 1987, 9: 33–66.
[3] Chen Y, Levine S, Rao M. Variable exponent, linear growth functionals in image restoration. SIAM J Appl Math, 2006, 66: 1383–1406. DOI:10.1137/050624522
[4] Avci M, Cekic B, Mashiyev R A. Existence and multiplicity of the solutions of the p(x)-Kirchhoff type equation via genus theory. Mathematical Methods in the Applied Sciences, 2011, 34(14): 1751–1759.
[5] Allaoui M, Amrouss E, Rachid A. Existence results for a class of nonlocal problems involving p(x)-Laplacian. Mathematical Methods in the Applied Sciences, 2016, 39(4): 824–832. DOI:10.1002/mma.v39.4
[6] Cammaroto F, Vilasi L. Multiple solutions for a Kirchhoff-type problem involving the p(x)-Laplacian operator. Nonlinear Anal TMA, 2011, 74(5): 1841–1852. DOI:10.1016/j.na.2010.10.057
[7] Chung N T. Multiplicity results for a class of p(x)-Kirchhoff type equations with combined nonlinearities. Electron J Qual Theory Differential Equations, 2012, 42: 1–13.
[8] Fan X L, Zhao D. On the spaces Lp(x)(Ω) and Wm, p(x)(Ω). J Math Anal Appl, 2001, 263: 424–446. DOI:10.1006/jmaa.2000.7617
[9] Fan X L, Zhang Q H. Existence of solutions of p(x)-Laplacian Dirichlet problems. Nonlinear Anal TMA, 2003, 52: 1843–1852. DOI:10.1016/S0362-546X(02)00150-5
[10] Fan X L. On nonlocal p(x)-Laplacian Dirichlet problems. Nonlinear Anal TMA, 2010, 72: 3314–3323. DOI:10.1016/j.na.2009.12.012
[11] Li L, Tang C L. Existence and multiplicity of solutions for a class of p(x)-biharmonic equations. Acta Mathematica Scientia, 2013, 33B(1): 155–170.
[12] Massar M, Talbi M, Tsouli N. Multiple solutions for nonlocal system of (p(x), q(x))-Kirchhoff type. Applied Mathematics and Computation, 2014, 242: 216–226. DOI:10.1016/j.amc.2014.05.057
[13] Rǎdulescu V D. Nonlinear elliptic equations with variable exponent:old and new. Nonlinear Anal TMA, 2015, 121: 336–369. DOI:10.1016/j.na.2014.11.007
[14] Lapa E C, Rivera V P, Broncano J Q. No-flux boundary problems involving p(x)-Laplacian-like operators. Electronic J Differential Equations, 2015, 219: 1–10.
[15] Ge B. On superlinear p(x)-Laplacian-like problem without Ambrosetti and Rabinowitz condition. Bull Korean Math Soc, 2014, 51: 409–421. DOI:10.4134/BKMS.2014.51.2.409
[16] Avci M. Ni-Serrin type equations arising from capillarity phenomena with non-standard growth. Boundary Value Problems, 2013, 55(1): 1–13.
[17] Rodrigues M M. Multiplicity of solutions on a nonlinear eigenvalue problem for p(x)-Laplacian-like operators. Mediterranean J Mathematics, 2012, 9(1): 211–223. DOI:10.1007/s00009-011-0115-y
[18] Zhou Q M. On the superlinear problems involving p(x)-Laplacian-like operators without AR-condition. Nonlinear Anal RWA, 2015, 21: 161–169. DOI:10.1016/j.nonrwa.2014.07.003
[19] Ding Y H, Luan S X. Multiple solutions for a class of nonlinear Schrödinger equations. J Differential Equations, 2004, 207: 423–457. DOI:10.1016/j.jde.2004.07.030
[20] Zhou H S. An application of a mountain pass theorem. Acta Mathematica Sinica English Series, 2002, 18(1): 27–36. DOI:10.1007/s101140100147
[21] Tang C L, Wu X P. Periodic solutions for a class of new superquadratic second order Hamiltonian systems. Appl Math Lett, 2014, 34: 65–71. DOI:10.1016/j.aml.2014.04.001
[22] Bartsch T. Infinitely many solutions of a symmetric Dirichlet problem. Nonlinear Anal TMA, 1993, 20: 1205–1216. DOI:10.1016/0362-546X(93)90151-H
[23] Kajikiya R. A critical point theorem related to the symmetric mountain pass lemma and its applications to elliptic equations. J Funct Analysis, 2005, 225: 352–370. DOI:10.1016/j.jfa.2005.04.005