数学物理学报  2017, Vol. 37 Issue (4): 671-683   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
陈林
一类拟线性Kirchhoff型椭圆方程组多解的存在性
陈林     
伊犁师范学院数学与统计分院 新疆伊宁 835000
摘要:该文运用Nehari流形和纤维环映射方法研究非局部拟线性椭圆方程组 $ \begin{eqnarray} \left\{ \begin{array}{lll} -M(\|\nabla u\|_p ^p )\triangle_p u=\lambda g_1 (x)|u|^{r-2}u+\frac{\alpha}{\alpha+\beta}h(x)|u|^{\alpha -2}u|v|^\beta, &x\in \Omega, \nonumber\\ -M(\|\nabla v\|_p ^p )\triangle_p v=\mu g_2 (x)|v|^{r-2}v+\frac{\beta}{\alpha+\beta}h(x)|v|^{\beta -2}v|u|^\alpha,&x\in \Omega, \nonumber\\\ u(x)=v(x)=0,&x\in \partial \Omega \nonumber \end{array} \right. \end{eqnarray} $ 非平凡弱解的存在性,其中Ω⊂$ \mathbb{R} $N是一边界光滑的有界区域,△pu=div(|▽u|p-2u)是p-拉普拉斯算子,1 < p < Nα > 1,β > 1,α+β < p < pk+1) < r < p*$ {{p}^{*}}=\frac{pN}{N-p} $N > pp*=∞若Np),λμ > 0,hx),g1x),g2x)∈C(Ω)在Ω上可变号,Ms)=a+bskabk > 0.
关键词拟线性椭圆方程组    Nehari流形    纤维环映射    p-拉普拉斯算子    
Multiple Solutions for a Quasilinear Elliptic System of Kirchhoff Type
Chen Lin     
College of Mathematics and Statistics, Yili Normal University, Xinjiang Yining 835000
Abstract: In this paper, using Nehari manifold and fibering maps we study the existence of multiple nontrivial nonnegative solutions for the nonlocal quasilinear elliptic system $ \begin{eqnarray} \left\{ \begin{array}{lll} -M(\|\nabla u\|_p ^p )\triangle_p u=\lambda g_1 (x)|u|^{r-2}u+\frac{\alpha}{\alpha+\beta}h(x)|u|^{\alpha -2}u|v|^\beta,&x\in \Omega, \nonumber\\ -M(\|\nabla v\|_p ^p )\triangle_p v=\mu g_2 (x)|v|^{r-2}v+\frac{\beta}{\alpha+\beta}h(x)|v|^{\beta -2}v|u|^\alpha,&x\in \Omega, \nonumber\\\ u(x)=v(x)=0,&x\in \partial \Omega, \nonumber \end{array} \right. \end{eqnarray} $ where Ω is a bounded smooth domain of $ \mathbb{R} $N, △pu=div (|▽u|p-2u) is the p-Laplacian with 1 < p < N, α > 1, β > 1, α+β < p < p (k+1) < r < p*($ {{p}^{*}}=\frac{pN}{N-p} $ if N > p, p*=∞ if Np), λ, μ > 0, h (x), g1(x), g2(x)∈C (Ω) are functions which change sign in Ω and M (s)=a+bsk, a, b, k > 0.
Key words: Quasilinear elliptic system     Nehari manifold     Fibering map     p-Laplacian    
1 引言

本文运用Nehari流形和纤维环映射方法研究拟线性椭圆方程组

$ \begin{eqnarray}\label{1.1} \left\{ \begin{array}{lll} -M(\|\nabla u\|_p ^p )\triangle_p u =\lambda g_1 (x)|u|^{r-2}u+\frac{\alpha}{\alpha+\beta}h(x)|u|^{\alpha -2}u|v|^\beta, &x\in \Omega, \\[3mm] -M(\|\nabla v\|_p ^p )\triangle_p v =\mu g_2 (x)|v|^{r-2}v+\frac{\beta}{\alpha+\beta}h(x)|v|^{\beta -2}v|u|^\alpha, &x\in \Omega, \\ u(x)=v(x)=0,&x\in \partial \Omega \end{array} \right. \end{eqnarray} $ (1.1)

非平凡弱解的存在性, 其中$\Omega \subset {\Bbb R}^N$是一边界光滑的有界区域, $\Delta _p u=\mbox{div}(|\nabla u|^{p-2}\nabla u)$$p$ -拉普拉斯算子, $1<p<N, $ $\alpha >1, \beta>1, \alpha +\beta <p<p(k+1)<r<p^*$ ($p^*=\frac{pN}{N-p}$$N>p, $ $p^*=\infty$$N\leq p$), $ \lambda, \mu >0, $ $h(x), g_1(x), g_2(x)\in C(\Omega )$$\Omega $上可变号, $M(s)=a+bs^k, a, b, k>0.$

由于存在整个区域上的积分项, 从而此类问题通常被称为非局部问题.非局部项的存在增大了问题的研究难度, 同时也使得该类问题变得更加有趣.非局部问题来源于物理问题.当$p=2$时, 非局部项$M(\|u\|_p ^p)$最早出现于用来描述弦振动的Kichhoff方程中[1], 即

$ \begin{eqnarray}\label{1.2} \left\{ \begin{array}{lll} u_{tt}-M(\|u\|_2 ^2)\Delta u=h(x, u), (x, t)\in\Omega \times (0, T), \\ u(x, t)=0, (x, t)\in\partial \Omega \times (0, T), \\ u(x, 0)=u_0 (x), u_t (x, 0)=u_1 (x), \end{array} \right. \end{eqnarray} $ (1.2)

这里$u(x, t)$表示位移, $h(x, u)$表示外力.自Lions在文献[2]中所作的工作之后, 问题(1.2) 受到越来越多的人的关注, 得到了一些较为深刻的研究结果(参见文献[3-6]).此外, 非局部方程在生物学中也有重要的应用.例如, 非局部抛物方程常被用来描述某类特殊种群的增长和迁移规律, 其中非局部项常用来刻画种群的迁移情况.人们在这方面的研究取得了很多有意义的成果(参见文献[7-8]).

受文献[9-12]的启发, 本文将说明具有可变号权函数的非局部问题(1.1) 至少存在两个非平凡的弱解.

2 变分结构

$X=W_0 ^{1, p}(\Omega )\times W_0 ^{1, p}(\Omega )$表示通常的Sobolev空间, 其上的范数定义为

$ \begin{eqnarray}\label{2.1} \|(u, v)\|=\Big(\int_\Omega (|\nabla u|^p +|\nabla v|^p){\rm d}x\Big)^{1/p}. \end{eqnarray} $ (2.1)

下列不等式是一著名的Sobolev不等式.即存在正常数$S>0, $使得对于任意$u\in C_0 ^\infty(\Omega) $成立

$ \begin{eqnarray}\label{2.2} S\Big(\int_\Omega |u|^{p^*}{\rm d}x\Big)^{p/p^*}\leq \int_\Omega |\nabla u|^p {\rm d}x.\end{eqnarray} $ (2.2)

由逼近方法易得此不等式在空间$X$中也是成立的.

定义2.1  设$(u, v)\in X. $如果对于任意的$(\varphi, \psi )\in X $

$ \begin{eqnarray*} &&(a+b\| \nabla u\|_p ^{pk})\int_\Omega |\nabla u|^{p-2}\nabla u \nabla \varphi {\rm d}x +(a+b\| \nabla v\|_p ^{pk})\int_\Omega |\nabla v|^{p-2}\nabla v \nabla \psi {\rm d}x\nonumber\\ &=&\lambda \int_\Omega g_1 (x)|u|^{r-2}u \varphi {\rm d}x+\frac{\alpha}{\alpha +\beta }\int_\Omega h(x)|u|^{\alpha -2}u|v|^\beta \varphi {\rm d}x \nonumber \\ &&+\mu \int_\Omega g_2 (x)|v|^{r-2}v \psi {\rm d}x+\frac{\beta}{\alpha +\beta }\int_\Omega h(x)|v|^{\beta -2}v|u|^\alpha \psi {\rm d}x, \nonumber \end{eqnarray*} $

则称$(u, v)\in X$是问题(1.1) 的一个弱解.

显然, 问题(1.1) 具有变分结构.设$J_{\lambda, \mu }:X\rightarrow {\Bbb R}^1$是问题(1.1) 所对应的能量泛函, 其具体定义为

$ \begin{eqnarray} J_{\lambda, \mu }(u, v)&=&\frac{a}{p}\|(u, v)\|^p +\frac{b}{p(k+1)}(\|\nabla u\|_p ^{p(k+1)}+\|\nabla v\|_p ^{p(k+1)})\nonumber \\ & &-\frac{1}{\alpha +\beta }\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x-\frac{1}{r}G_{\lambda, \mu }(u, v), \end{eqnarray} $ (2.3)

其中

$ \begin{eqnarray} G_{\lambda, \mu }(u, v)=\lambda \int_\Omega g_1 (x) |u|^r {\rm d}x+\mu \int_\Omega g_2 (x)|v|^r {\rm d}x. \end{eqnarray} $ (2.4)

则泛函$J_{\lambda, \mu}\in C^1 (X, {\Bbb R}^1 )$, 且对任意$(\varphi , \psi)\in X$

$ \begin{eqnarray}\label{2.4} \langle J'_{\lambda , \mu}(u, v), (\varphi, \psi )\rangle &=&(a+b\| \nabla u\|_p ^{pk})\int_\Omega |\nabla u|^{p-2}\nabla u \nabla \varphi {\rm d}x\nonumber\\ &&+(a+b\| \nabla v\|_p ^{pk})\int_\Omega |\nabla v|^{p-2}\nabla v \nabla \psi {\rm d}x -\lambda \int_\Omega g_1 (x)|u|^{r-2}u\varphi {\rm d}x\\ &&-\mu \int_\Omega g_2 (x)|v|^{r-2}v\psi {\rm d}x -\frac{\alpha}{\alpha +\beta }\int_\Omega h(x)|u|^{\alpha -2}u|v|^\beta \varphi {\rm d}x\\ && -\frac{\beta}{\alpha +\beta }\int_\Omega h(x)|v|^{\beta -2}v|u|^\alpha \psi {\rm d}x, \end{eqnarray} $ (2.5)

其中$\langle, \rangle$表示通常的内积.特别地, 由(2.5) 式可得

$ \begin{eqnarray} \langle J' _{\lambda, \mu }(u, v), (u, v)\rangle &=& a\|(u, v)\|^p +b(\|\nabla u\|_p ^{p(k+1)}+\|\nabla v\|_p ^{p(k+1)})\nonumber\\ &&-\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x -G_{\lambda, \mu }(u, v). \end{eqnarray} $ (2.6)

由于泛函$J_{\lambda, \mu }$在空间$X$上没有下界, 我们引入Nehari流形

$ \begin{eqnarray} M_{\lambda, \mu }(\Omega)=\{ (u, v)\in X\setminus(0, 0)|\langle J' _{\lambda, \mu }(u, v), (u, v)\rangle =0\}. \end{eqnarray} $ (2.7)

易见, $(u, v)\in M_{\lambda, \mu}(\Omega)$当且仅当

$ \begin{eqnarray} a \|(u, v)\|^p +b(\|\nabla u\|_p^{p(k+1)}+\|\nabla v\|_p^{p(k+1)}) -\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x -G_{\lambda, \mu }(u, v)=0. \end{eqnarray} $ (2.8)

因此, 如果$(u, v)\in M_{\lambda, \mu }(\Omega )$

$ \begin{eqnarray} J_{\lambda, \mu }(u, v)&= &a\Big(\frac{1}{p}-\frac{1}{r}\Big)\|(u, v)\|^p +b\Big(\frac{1}{p(k+1)}-\frac{1}{r}\Big)(\|\nabla u\|_p^{p(k+1)}+\|\nabla v\|_p^{p(k+1)})\nonumber \\ &&+ \Big(\frac{1}{r}-\frac{1}{\alpha +\beta }\Big)\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x \label{2.8} \end{eqnarray} $ (2.9)
$ \begin{eqnarray} &=& a\Big(\frac{1}{p}-\frac{1}{\alpha +\beta }\Big)\|(u, v)\|^p +b\Big(\frac{1}{p(k+1)}-\frac{1}{\alpha +\beta}\Big)(\|\nabla u\|_p^{p(k+1)}+\|\nabla v\|_p^{p(k+1)})\nonumber \\ &&- \Big(\frac{1}{r}-\frac{1}{\alpha +\beta }\Big)G_{\lambda, \mu }(u, v).\label{2.9} \end{eqnarray} $ (2.10)

更进一步, 定义

$ \Phi _{\lambda, \mu }(u, v)=\langle J' _{\lambda, \mu }(u, v), (u, v)\rangle, \quad \forall (u, v)\in X .\nonumber $

则对于任意$(u, v)\in M_{\lambda, \mu }(\Omega)$成立

$ \begin{eqnarray} \langle \Phi ' _{\lambda, \mu }(u, v), (u, v)\rangle &=& ap\|(u, v)\|^p +bp(k+1)(\|\nabla u\|_p ^{p(k+1)}+\|\nabla v\|_p^{p(k+1)})\nonumber \\ &&- (\alpha +\beta )\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x-r G_{\lambda, \mu}(u, v) \end{eqnarray} $ (2.11)
$ \begin{eqnarray} &=& a(p-r)\|(u, v)\|^p +b(p(k+1)-r)(\|\nabla u\|_p^{p(k+1)}+\|\nabla v\|_p^{p(k+1)})\nonumber \\ &&+ (r-\alpha -\beta )\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x \label{2.12} \end{eqnarray} $ (2.12)
$ \begin{eqnarray} &=& a(p-\alpha -\beta )\|(u, v)\|^p +b(p(k+1)-\alpha -\beta )(\|\nabla u\|_p^{p(k+1)}+\|\nabla v\|_p^{p(k+1)})\nonumber \\ &&- (r-\alpha -\beta )G_{\lambda, \mu}(u, v).\label{2.13} \end{eqnarray} $ (2.13)

为便于讨论, 将$ M_{\lambda, \mu }(\Omega)$分成三部分:

$ \begin{eqnarray}\label{????} &&M_{\lambda, \mu}^+(\Omega)=\{ (u, v)\in M_{\lambda, \mu}(\Omega)| \langle \Phi'_{\lambda, \mu}(u, v), (u, v) \rangle >0 \}, \\ &&M_{\lambda, \mu}^-(\Omega)=\{ (u, v)\in M_{\lambda, \mu}(\Omega)| \langle \Phi'_{\lambda, \mu}(u, v), (u, v) \rangle <0 \}, \\ &&M_{\lambda, \mu}^0(\Omega)=\{ (u, v)\in M_{\lambda, \mu}(\Omega)| \langle \Phi'_{\lambda, \mu}(u, v), (u, v) \rangle =0 \}. \end{eqnarray} $ (2.14)

引理2.1  存在$\eta _0 >0$, 使得当$0<\lambda\|g_1 \|_\infty +\mu \| g_2 \|_\infty <\eta _0 $$M_{\lambda, \mu }^0 (\Omega )=\emptyset .$

  假设结论不成立.则对于$\eta _0 =\frac{a(p-\alpha -\beta)}{(r-\alpha -\beta )2^{r-1}c_1 ^r }\cdot (\frac{a(r-p)}{(r-\alpha-\beta)\cdot 2M_0})^{\frac{r-p}{p-\alpha-\beta}}$, 存在正常数$\lambda$$\mu, $使得当$0<\lambda\|g_1 \|_\infty +\mu \| g_2 \|_\infty <\eta _0 $$M_{\lambda, \mu }(\Omega)\neq \emptyset .$其中$c_1 $是空间$W^{1, p}(\Omega )$嵌入到空间$L^r (\Omega )$中的Sobolev常数, $M_0 =(\int_{\Omega}|h(x)|^\delta {\rm d}x)^{1/\delta}S^{-(\alpha +\beta)/p}$, $\delta =\frac{p^*}{p^* -(\alpha +\beta )}$.则当$(u, v)\in M_{\lambda, \mu }^0 (\Omega )$时, 成立

$ \begin{eqnarray}\label{2.15} \langle \Phi ' _{\lambda, \mu}(u, v), (u, v)\rangle =0. \end{eqnarray} $ (2.15)

因此, 由(2.12), (2.13) 和(2.15) 式可得

$ \begin{eqnarray} (r-\alpha -\beta )\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x \geq a(r-p)\|(u, v)\|^p. \end{eqnarray} $ (2.16)

由Hölder不等式和(2.2) 式可得

$ \begin{eqnarray}\label{2.17} \int_\Omega h(x)|u|^{\alpha +\beta }{\rm d}x\leq \Big(\int_\Omega |h(x)|^\delta {\rm d}x\Big)^{\frac{1}{\delta }}\Big(\int_\Omega |u|^{p^*}{\rm d}x\Big)^{\frac{\alpha +\beta }{p^*}}\leq M_0 \| (u, v)\|^{\alpha +\beta}. \end{eqnarray} $ (2.17)

同理可得

$ \int_\Omega h(x)|v|^{\alpha +\beta }{\rm d}x\leq M_0 \|(u, v)\|^{\alpha +\beta }. $

从而

$ \int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x\leq 2M_0 \|(u, v)\|^{\alpha +\beta }. $

因此

$ a(r-p)\|(u, v)\|^p \leq (r-\alpha -\beta )\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x\leq (r-\alpha -\beta )\cdot 2M_0\|(u, v)\|^{\alpha +\beta }. \nonumber $

从而

$ \begin{eqnarray}\label{2.18} \|(u, v)\|\leq \tilde{c}_0 ^{1/(p-\alpha -\beta)}, \end{eqnarray} $ (2.18)

其中$\tilde{c}_0=\frac{(r-\alpha -\beta )\cdot 2M_0}{a(r-p)}, $且有

$ \begin{eqnarray} &&\lambda(r-\alpha -\beta )\int_\Omega g_1 (x)|u|^r {\rm d}x+\mu (r-\alpha -\beta )\int_\Omega g_2 (x)|v|^r {\rm d}x\nonumber \\ &\leq& \lambda (r-\alpha -\beta )\|g_1 \|_\infty \|u\|_r ^r +\mu (r-\alpha -\beta )\|g_2 \|_\infty \|v\|_r ^r \nonumber \\ &\leq&(r-\alpha -\beta )c_1 ^r (\lambda \|g_1 \|_\infty \|\nabla u\|_p ^r +\mu \|g_2 \|_\infty \|\nabla v\|_p ^r )\nonumber \\ &\leq &(r-\alpha -\beta )2^{r-1}c_1 ^r (\lambda \|g_1 \|_\infty +\mu \|g_2 \|_\infty )\|(u, v)\|^r . \end{eqnarray} $ (2.19)

由(2.13) 式得

$ \begin{eqnarray} &&( r-\alpha -\beta )\Big(\lambda \int_\Omega g_1 (x)|u|^r {\rm d}x +\mu \int_\Omega g_2 (x)|v|^r {\rm d}x\Big)\nonumber \\ &=& a(p-\alpha -\beta )\|(u, v)\|^p +b(p(k+1)-\alpha -\beta )(\|\nabla u\|_p^{p(k+1)}+\|\nabla v\|_p^{p(k+1)})\nonumber \\ &\geq& a(p-\alpha -\beta )\|(u, v)\|^p .\nonumber \end{eqnarray} $

从而

$ a(p-\alpha -\beta )\|(u, v)\|^p \leq (r-\alpha -\beta )2^{r-1}c_1 ^r (\lambda \|g_1 \|_\infty +\mu \|g_2 \|_\infty )\|(u, v)\|^r . $

从而

$ \begin{eqnarray}\label{2.20} \|(u, v)\|\geq \tilde{c}_1 ^{1/(r-p)}\cdot (\lambda \|g_1 \|_\infty+\mu \|g_2 \|_\infty )^{1/(p-r)}, \end{eqnarray} $ (2.20)

其中$\tilde{c}_1 =\frac{a(p-\alpha -\beta )}{(r-\alpha -\beta )2^{r-1}c_1 ^r}.$由(2.18) 式和(2.20) 式得

$ \tilde{c}_1^{1/(r-p)}\cdot (\lambda \|g_1 \|_\infty +\mu \|g_2 \|_\infty )^{1/(p-r)}\leq \tilde{c}_0^{1/(p-\alpha -\beta)}.\nonumber $

由此可得

$ \lambda \|g_1 \|_\infty+\mu \|g_2 \|_\infty \geq \tilde{c}_1 \cdot \tilde{c}_0^{(p-r)/(p-\alpha-\beta)}=\eta _0, \nonumber $

矛盾!因此, 存在常数$\eta _0 >0 $, 使得当$0<\lambda\|g_1 \|_\infty+\mu \|g_2 \|_\infty <\eta _0 $时有$M_{\lambda, \mu }^0 (\Omega )=\emptyset .$

引理2.2  泛函$J_{\lambda, \mu }$$M_{\lambda, \mu }(\Omega)$上强制且有下界.

  对于任意$(u, v)\in M_{\lambda, \mu }(\Omega), $成立

$ \begin{eqnarray}\label{2.21} J_{\lambda, \mu }(u, v)&=& a\Big(\frac{1}{p}-\frac{1}{r}\Big)\|(u, v)\|^p +b\Big(\frac{1}{p(k+1)}-\frac{1}{r}\Big)(\|\nabla u\|_p^{p(k+1)}+\|\nabla v\|_p^{p(k+1)})\nonumber \\ &&- \Big(\frac{1}{\alpha +\beta }-\frac{1}{r}\Big)\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x \nonumber \\ &\geq&a\Big(\frac{1}{p}-\frac{1}{r}\Big)\|(u, v)\|^p +b\Big(\frac{1}{p(k+1)}-\frac{1}{r}\Big)(\|\nabla u\|_p^{p(k+1)}+\|\nabla v\|_p^{p(k+1)})\nonumber \\ &&- \Big(\frac{1}{\alpha +\beta }-\frac{1}{r}\Big)\cdot2M_0\|(u, v)\|^{\alpha +\beta }, \end{eqnarray} $ (2.21)

其中$M_0$与引理2.1中所引入的常数$M_0$相同.由于$p>\alpha +\beta $, 从而由不等式(2.21) 可知泛函$J_{\lambda, \mu}$$M_{\lambda, \mu}(\Omega)$上是强制的而且是有下界的.证毕.

引理2.3  假定$(u_0, v_0 )$是泛函$J_{\lambda, \mu }$$M_{\lambda, \mu }(\Omega)$上的一个极小值点.如果$(u_0, v_0 )\not\in M_{\lambda, \mu }^0 (\Omega ), $$(u_0, v_0 )$是泛函$J_{\lambda, \mu }$的一个临界点.

  设

$ F(u, v)= a\|(u, v)\|^p +b(\|\nabla u\|_p^{p(k+1)}+\|\nabla v\|_p^{p(k+1)}) - G_{\lambda, \mu }(u, v)-\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x. $

考虑最优化问题

$ \min\limits_{(u, v)\in M_{\lambda, \mu}(\Omega)} J_{\lambda, \mu}(u, v), $

其中$(u, v)\in M_{\lambda, \mu}(\Omega) $满足

$ F(u, v)=0. $

由Lagrange乘子原理, 存在常数$\eta\in {\Bbb R}^1 $使得

$ J'_{\lambda, \mu}(u_0, v_0)=\eta F'(u_0, v_0). $

由于$(u_0, v_0)\in M_{\lambda, \mu}(\Omega) $, 从而

$ \langle J'_{\lambda, \mu}(u_0, v_0), (u_0, v_0)\rangle=0. $

然而, $(u_0, v_0)\not\in M_{\lambda, \mu}^0(\Omega )$, 从而

$ \langle F'(u_0, v_0), (u_0, v_0)\rangle= \langle\Phi'_{\lambda , \mu}(u_0, v_0), (u_0, v_0) \rangle \neq 0. $

因此, $\eta =0$, $J'_{\lambda, \mu}(u_0, v_0)=0$.证毕.

由引理2.1, 当$0<\lambda\|g_1 \|_\infty +\mu \| g_2 \|_\infty <\eta _0 $时, $M_{\lambda, \mu}(\Omega)=M_{\lambda, \mu}^+(\Omega)\cup M_{\lambda, \mu}^-(\Omega)$.设

$ \begin{eqnarray} \delta _{\lambda, \mu}^+ = \inf\limits_{(u, v)\in M_{\lambda, \mu}^+(\Omega)} J_{\lambda, \mu}(u, v), \quad \delta _{\lambda, \mu}^-=\inf\limits_{(u, v)\in M_{\lambda, \mu}^-(\Omega)} J_{\lambda, \mu}(u, v). \end{eqnarray} $ (2.22)

引理2.4  若$\lambda$$\mu$满足$ 0<\lambda \|g_1 \|_\infty +\mu \|g_2 \|_\infty <\eta ^*$, 其中$\eta ^* =(\frac{\alpha +\beta }{p})^{\frac{r-p}{p-\alpha -\beta}}\eta _0, $

(ⅰ) $\delta_{\lambda, \mu}^+<0$;

(ⅱ) $\exists k_0>0$使得$ \delta _{\lambda, \mu}^->k_0$.

  (ⅰ) 设$(u, v)\in M_{\lambda, \mu}^+(\Omega)$.由(2.12) 式可得

$ \begin{eqnarray} && a (p-\alpha -\beta )\|(u, v)\|^p +b(p(k+1)-\alpha -\beta )(\|\nabla u\|_p^{p(k+1)}+\|\nabla v\|_p^{p(k+1)})\nonumber \\ &&- (r-\alpha -\beta )G_{\lambda, \mu }(u, v) > 0 .\nonumber \end{eqnarray} $

从而

$ \begin{eqnarray} a \frac{p-\alpha -\beta }{r-\alpha -\beta }\|(u, v)\|^p +\frac{b(p(k+1)-\alpha -\beta )}{r-\alpha -\beta }(\|\nabla u\|_p^{p(k+1)}+\|\nabla v\|_p^{p(k+1)})> G_{\lambda, \mu}(u, v).\nonumber \end{eqnarray} $

因此

$ \begin{eqnarray} J_{\lambda, \mu}(u, v)&= &a\Big(\frac{1}{p}-\frac{1}{\alpha +\beta }\Big)\|(u, v)\|^p +b\Big(\frac{1}{p(k+1)}-\frac{1}{\alpha +\beta}\Big)(\|\nabla u\|_p ^{p(k+1)}+\|\nabla v\|_p ^{p(k+1)})\nonumber \\ &&+ \Big(\frac{1}{\alpha +\beta }-\frac{1}{r}\Big)G_{\lambda, \mu }(u, v)\nonumber \\ &\leq&\frac{b\big(\alpha +\beta -p(k+1)\big)\big(r-p(k+1)\big)}{pr(k+1)(\alpha +\beta )}(\|\nabla u\|_p^{p(k+1)}+\|\nabla v\|_p^{p(k+1)}) \nonumber\\ &&+\frac{a(r-p)(\alpha+\beta -p)}{pr(\alpha +\beta )}\|(u, v)\|^p\nonumber \\ &<& 0. \end{eqnarray} $ (2.23)

从而$\delta _{\lambda, \mu }^+ <0.$

(ⅱ) 设$(u, v)\in M_{\lambda, \mu}^- (\Omega)$.由(2.12) 式可知

$ \begin{eqnarray} &&a(p-\alpha -\beta)\|(u, v)\|^p +b(p(k+1)-\alpha -\beta)(\|\nabla u\|_p^{p(k+1)}+\|\nabla v\|_p^{p(k+1)})\nonumber\\ &&-(r-\alpha -\beta )G_{\lambda, \mu}(u, v)< 0. \end{eqnarray} $ (2.24)

从而

$ a(p-\alpha -\beta)\|(u, v)\|^p <(r-\alpha -\beta )G_{\lambda, \mu }(u, v).\nonumber $

$ \|(u, v)\|^p <\frac{r-\alpha -\beta }{a(p-\alpha -\beta )}G_{\lambda, \mu}(u, v) \leq \tilde{c}_1^{-1} (\lambda \|g_1 \|_\infty +\mu \|g_2 \|_\infty )\|(u, v)\|^r .\nonumber $

因此, 对于任意的$(u, v)\in M_{\lambda, \mu }^-(\Omega)$成立

$ \|(u, v)\|>\big(\tilde{c}_1^{-1}\cdot (\lambda \|g_1 \|_\infty +\mu \|g_2 \|_\infty )\big)^{1/(p-r)}.\nonumber $

由引理2.2的证明可知

$ \begin{eqnarray} J_{\lambda, \mu}(u, v)&\geq& a\Big(\frac{1}{p}-\frac{1}{r}\Big)\|(u, v)\|^p -\Big(\frac{1}{\alpha +\beta }-\frac{1}{r}\Big)2M_0\|(u, v)\|^{\alpha +\beta }\nonumber \\ &=&\|(u, v)\|^{\alpha +\beta }\Big(a\Big(\frac{1}{p}-\frac{1}{r}\Big)\|(u, v)\|^{p-\alpha -\beta }-2M_0\Big(\frac{1}{\alpha +\beta }-\frac{1}{r}\Big)\Big)\nonumber \\ &> &\big(\tilde{c}_1^{-1}\cdot(\lambda\|g_1 \|_\infty +\mu \|g_2 \|_\infty)\big)^{(\alpha +\beta)/(p-r)}\cdot\tilde{g}, \nonumber \end{eqnarray} $

其中$\tilde{g}=\frac{a(r-p)}{pr}\cdot \big(\tilde{c}_1^{-1}\cdot (\lambda \|g_1 \|_\infty +\mu \|g_2 \|_\infty )\big)^{(p-\alpha -\beta )/(p-r)}- 2M_0(\frac{1}{\alpha +\beta }-\frac{1}{r})$.

从而当$0<\lambda \|g_1 \|_\infty +\mu \|g_2 \|_\infty <\eta ^* $时, 存在常数$k_0 =k_0 (\alpha, \beta, r, c_1, M_0, a, b, \lambda, \mu )>0$, 使得对于任意的$(u, v)\in M_{\lambda, \mu }^- (\Omega)$$J_{\lambda, \mu }(u, v)>k_0 $.证毕.

$ \begin{eqnarray} \phi _{u, v}(t)&=&J_{\lambda, \mu}{(tu, tv)} \nonumber \\ & =& t^p \frac{1}{p}a\|(u, v)\|^p +t^{p(k+1)}\frac{b}{p(k+1)}(\|\nabla u\|_p^{p(k+1)}+\|\nabla v\|_p^{p(k+1)})\nonumber \\ &&- t^r \frac{1}{r}G_{\lambda, \mu }(u, v) - t^{\alpha +\beta}\frac{1}{\alpha+\beta }\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x, \quad t>0. \end{eqnarray} $ (2.25)

$ \begin{eqnarray}\label{2.26} \phi' _{u, v}(t)&=& t^{p-1}a\|(u, v)\|^p +t^{p(k+1)-1}b(\|\nabla u\|_p^{p(k+1)}+\|\nabla v\|_p^{p(k+1)})\nonumber \\ &&- t^{r-1}G_{\lambda, \mu }(u, v) - t^{\alpha+\beta -1}\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x, \end{eqnarray} $ (2.26)
$ \begin{eqnarray} \phi '' _{u, v}(t)&=& t^{p-2}a(p-1)\|(u, v)\|^p +t^{p(k+1)-2}b(p(k+1)-1)(\|\nabla u\|_p^{p(k+1)-2}+\|\nabla v\|_p^{p(k+1)-2})\nonumber \\ &&- (r-1)t^{r-2}G_{\lambda, \mu }(u, v) -(\alpha +\beta -1)t^{\alpha +\beta -2}\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x. \end{eqnarray} $ (2.27)

易见, $(u, v)\in M_{\lambda, \mu }(\Omega )$当且仅当$\phi' _{u, v}(1)=0 $.更一般地, $\phi ' _{u, v}(t)=0 $当且仅当$tu\in M_{\lambda, \mu}(\Omega )$.即, $M_{\lambda, \mu }(\Omega )$中的元对应于纤维环映射的稳定点.另外, 由$u\in M_{\lambda, \mu }(\Omega )$可以推出$\phi '' _{u, v}(1)=\langle \Phi '(u, v), (u, v)\rangle .$

引理2.5  假定$r\geq p(k+1)$, $0<\lambda \|g_1 \|_\infty+\mu \|g_2 \|_\infty <\eta ^* .$则对于任一$(u, v)\in X$, $G_{\lambda, \mu }(u, v)>0 $, 存在$t_b >0 $使得

(ⅰ) 若$\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x\leq 0, $则存在唯一的$t^- >t_b $使得$(t^- u, t^- v)\in M_{\lambda, \mu}^-(\Omega )$且有$J_{\lambda, \mu }(t^- u, t^- v)=\sup\limits_{t\geq0} J_{\lambda, \mu}(tu, tv)$;

(ⅱ) 若$\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x> 0, $则存在唯一的$0<t^+ =t^+ (u, v)<t_b <t^- =t^- (u, v)$使得$(t^+ u, t^+ v)\in M_{\lambda, \mu}^+ (\Omega ), (t^- u, t^- v )\in M_{\lambda, \mu}^- (\Omega )$且有$J_{\lambda, \mu}(t^+ u, t^+ v )=\inf\limits_{0\leq t\leq t_b}J_{\lambda, \mu}(tu, tv ), $ $J_{\lambda, \mu}(t^- u, t^- v )=\sup\limits_{t\geq 0}J_{\lambda, \mu}(tu, tv ).$

  取$(u, v)\in X$使得$G_{\lambda, \mu }(u, v)>0$并设

$ \begin{eqnarray} S(t)&=& at^{p-\alpha -\beta }\|(u, v)\|^p +bt^{p(k+1)-\alpha -\beta }(\|\nabla u\|_p^{p(k+1)}+\|\nabla v\|_p^{p(k+1)})\nonumber \\ &&- t^{r-\alpha -\beta }G_{\lambda, \mu }(u, v), t>0 .\nonumber \end{eqnarray} $

由(2.26) 式可得

$ \begin{eqnarray}\label{2.29} \phi' _{u, v} (t)=t^{\alpha +\beta -1}\Big(S(t)-\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x\Big). \end{eqnarray} $ (2.28)

显然, $tu \in M_{\lambda, \mu}(\Omega )$当且仅当

$ \begin{eqnarray}\label{2.30} S(t)=\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x. \end{eqnarray} $ (2.29)

从而若$tu\in M_{\lambda, \mu}(\Omega), $

$ \begin{eqnarray}\label{2.31} \phi'' _{u, v}(t)=t^{\alpha+\beta -1}S'(t). \end{eqnarray} $ (2.30)

因此, $tu\in M_{\lambda, \mu }^+ (\Omega )$当且仅当$S' (t)>0 $; $tu\in M_{\lambda, \mu }^- (\Omega )$当且仅当$S'(t)<0$.由于$G_{\lambda, \mu }(u, v)>0, $从而当$t\rightarrow +\infty $时有$S(t)\rightarrow -\infty.$另一方面,

$ \begin{eqnarray} S' (t)&= &a(p-\alpha -\beta )\|(u, v)\|^p t^{p-\alpha -\beta -1}\nonumber \\ &&+ b(p(k+1)-\alpha -\beta )(\|\nabla u\|_p^{p(k+1)}+\|\nabla v\|_p^{p(k+1)})t^{p(k+1)-\alpha-\beta-1}\nonumber \\ &&- (r-\alpha -\beta )G_{\lambda, \mu}(u, v)t^{r-\alpha-\beta-1}.\nonumber \end{eqnarray} $

从而, 存在唯一的$t_b >0 $使得$S' (t_b )=0 $$S(t)$$t_b $处取得最大值.因而, 当$t\in (0, t_b )$时函数$S(t)$单调增加, 当$t>t_b $时函数$S(t)$单调减少.特别地, 当$b=0$时有

$ t_0 =\Big(\frac{(p-\alpha -\beta )(a\|(u, v)\|^p )}{(r-\alpha -\beta )G_{\lambda, \mu}(u, v)}\Big)^{\frac{1}{r-p}}>0, \nonumber $

且有

$ S(t_0 )=a\|(u, v)\|^p t_0 ^{p-\alpha -\beta }-G_{\lambda, \mu }(u, v)t_0 ^{r-\alpha -\beta }.\nonumber $

我们首先证明当$b\geq 0$$t_0 \leq t_b $.事实上, $t_0 $$t_b $分别满足

$ a(p-\alpha -\beta )\|(u, v)\|^p =(r-\alpha -\beta )G_{\lambda, \mu}(u, v)t_0 ^{r-p}\nonumber $

$ \begin{eqnarray} a( p-\alpha -\beta )\|(u, v)\|^p &=&(r-\alpha-\beta)G_{\lambda, \mu}(u, v)t_b ^{r-p}-b(p(k+1)\nonumber\\ &&-\alpha-\beta)(\|\nabla u\|_p^{p(k+1)}+\|\nabla v\|_p^{p(k+1)})t_b ^{pk}\nonumber \\ &\leq&(r-\alpha -\beta)G_{\lambda, \mu }(u, v)t_b ^{r-p}.\nonumber \end{eqnarray} $

由此可知$t_0 \leq t_b .$从而

$ \begin{eqnarray} S(t_b )&=& t_b ^{-\alpha -\beta }(a\|(u, v)\|^p t_b ^p + b(\|\nabla u\|_p^{p(k+1)}+\|\nabla v\|_p^{p(k+1)})t_b^{p(k+1)} - G_{\lambda, \mu}(u, v)t_b^r )\nonumber \\ &=&a\frac{r-p}{r-\alpha-\beta }\|(u, v)\|^p t_b ^{p-\alpha -\beta } + b\frac{r-p(k+1)}{r-\alpha -\beta }(\|\nabla u\|_p^{p(k+1)}+\|\nabla v\|_p^{p(k+1)})t_b^{p(k+1)-\alpha -\beta}\nonumber \\ &\geq &a\frac{r-p}{r-\alpha-\beta }\|(u, v)\|^p t_0 ^{p-\alpha -\beta }\nonumber \\ &= &S(t_0 ).\nonumber \end{eqnarray} $

接下来, 我们考虑以下两种情形:

(ⅰ) $\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x\leq 0.$此时, 存在唯一的$t^- >t_b $使得$ S(t^- )=\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x$$S' (t^- )<0.$从而由(2.29) 式和(2.30) 式可知$t^- u\in M_{\lambda, \mu }^- (\Omega).$由于当$t\in (0, t^- )$$S(t)>\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x$, 当$t>t^- $$S(t)<\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x$, 从而由(2.28) 式可得

$ \phi _{u, v}(t^- )=\sup\limits_{t\geq 0 }\phi _{u, v}(t), ~~ J_{\lambda, \mu}(t^- u, t^- v)=\sup\limits_{t\geq 0}J_{\lambda, \mu }(tu, tv). $

(ⅱ) $\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x>0. $由于$0<\lambda \|g_1\|_\infty +\mu \|g_2\|_\infty <\eta ^*$, 从而

$ S(0)=0<\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x\leq 2M_0\|(u, v)\|^{\alpha +\beta }\leq S(t_0 )\leq S(t_b ). $

因此, 存在唯一的$t^+ $$t^- $使得$0<t^+ <t_b <t^- $, $S(t^+ )=S(t^- )=\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x$且有$S' (t^+ )>0>S'(t^- ).$由此可得$t^+ u \in M_{\lambda, \mu}^+ (\Omega )$, $t^- u \in M_{\lambda, \mu}^- (\Omega )$.因为当$t\in (t^+, t^-)$时, $S(t)\geq \int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x$, 从而当$t\in(t^+, t ^- )$$\phi' _{u, v}(t)\geq 0 $; 当$t\in [t^+, t^-]$时, $\phi _{u, v}(t^+ )\leq \phi _{u, v}(t)\leq \phi _{u, v}(t^- )$.又由于当$t\in [0, t^+]$$S(t)<\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x $, 从而当$t\in [0, t^+]$$\phi ' _{u, v}(t)\leq 0 $且有$\phi _{u, v}(t^+ )\leq \phi _{u, v}(t)$.因此

$ \phi _{u, v}(t^+ )=J_{\lambda, \mu }(t^+u, t^+v )=\inf\limits_{0\leq t\leq t_b }J_{\lambda, \mu }(tu, tv). $

另一方面, 当$t\in [0, t^+]$

$ \phi _{u, v}(t)\leq \phi _{u, v}(0)<\phi _{u, v}(t^-), $

而当$t>t^-$$S(t)\leq \int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x $.从而, 当$t>t^-$$\phi '_{u, v}\leq 0 $.因此,

$ \phi _{u, v}(t^- )=\sup\limits_{t\geq 0 }\phi _{u, v}(t), ~~ J_{\lambda, \mu}(t^- u, t^-v)=\sup\limits_{t\geq 0}J_{\lambda, \mu}(tu, tv ). $

证毕.

引理2.6  对于满足$\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x >0$的元$(u, v)\in X$, 存在$t_{\max }>0 $使得

(ⅰ) 若$G(u, v)\leq 0, $则存在唯一的$0<t^+ <t_{\max }$使得$(t^+ u, t^+v)\in M_{\lambda, \mu}^+ (\Omega )$且有

$ J_{\lambda, \mu }(t^+u, t^+v)=\inf\limits_{t\geq 0 }J_{\lambda, \mu}(tu, tv); $

(ⅱ) 若$G(u, v)>0, $则存在唯一的$0<t^+ =t^+(u, v)<t_{\max }<t^- $使得$(t^+u, t^+v)\in M_{\lambda, \mu}^+ (\Omega), $ $(t^-u, t^-v)\in M_{\lambda, \mu }^- (\Omega)$且有

$ J_{\lambda, \mu}(t^+u, t^+v)=\inf\limits_{0\leq t\leq t_{\max }}J_{\lambda, \mu}(tu, tv), ~~ J_{\lambda, \mu}(t^-u, t^-v)=\sup\limits_{t\geq 0}J_{\lambda, \mu}(tu, tv).\nonumber $

  取$(u, v)\in X $使得$\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x>0.$

$ \begin{eqnarray}\label{2.32} \bar{S}(t)&=& a\|(u, v)\|^pt^{p-r } +b(\|\nabla u\|_p^{p(k+1)}+\|\nabla v\|_p^{p(k+1)})t^{p(k+1)-r}\nonumber \\ &&-t^{\alpha +\beta-r}\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x . \end{eqnarray} $ (2.31)

则由(2.26) 式可得

$ \phi' _{u, v}(t)=t^{r-1}(\bar{S}(t)-G_{\lambda, \mu }(u, v)). $

显然, 当$t\rightarrow 0^+ $$\bar{S}(t)\rightarrow -\infty $, 当$t\rightarrow +\infty $$\bar{S}(t)\rightarrow 0 $.又由于

$ \begin{eqnarray} \bar{S}' (t)&= &a(p-r)\|(u, v)\|^p t^{p-r-1}+b(p(k+1)-r)(\|\nabla u\|_p^{p(k+1)}+\|\nabla v\|_p^{p(k+1)})t^{p(k+1)-r-1}\nonumber \\ &&-(\alpha+\beta-r)t^{\alpha+\beta-r-1}\int_\Omega h(x)|u|^\alpha |v|^\beta {\rm d}x, \nonumber \end{eqnarray} $

从而存在$t_{\max }>0, $$t=t_{\max }$$\bar{S}' (t)=0 $, 当$t\in (0, t_{\max })$$\bar{S}' (t)>0$, 当$t>t_{\max }$$\bar{S}'(t)<0$.因此$\bar{S}(t)$$t_{\max }$点处达到最大值, 即当$t\in (0, t_{\max })$$\bar{S}(t)$单调递增, 当$t>t_{\max }$$\bar{S}(t)$单调递减.类似于引理2.5的证明即得引理2.6的结论.证毕.

3 非负弱解的存在性

为得到本文的主要结论, 我们首先讨论以下命题.

命题3.1  如果$0<\lambda \|g_1 \|_\infty +\mu \|g_2 \|_\infty <\eta ^*, $则泛函$J_{\lambda, \mu}$$M_{\lambda, \mu }^+ (\Omega)$中存在一个极小值点$(u^+ _0, v^+ _0 )$满足:

(ⅰ) $ J_{\lambda, \mu}(u_0^+, v_0^+)=\delta_{\lambda, \mu}^+$;

(ⅱ) $(u_0^+, v_0^+)$是问题(1.1) 的一个弱解且在区域$\Omega $$u_0 ^+ \geq 0, v_0 ^+ \geq 0$, $u_0 ^+ \not\equiv 0, v_0 ^+ \not\equiv 0.$

  由引理2.2以及$J_{\lambda, \mu}$$ M_{\lambda, \mu}^+(\Omega)$上有下界可知:存在一极小化序列$\{(u_k, v_k)\}\subseteq M_{\lambda, \mu}^+ (\Omega)$使得

$ \lim\limits_{k\rightarrow \infty}J_{\lambda, \mu}(u_k, v_k)=\inf\limits_{(u, v)\in M_{\lambda , \mu}^+(\Omega)} J_{\lambda, \mu}(u, v). $

由于$J_{\lambda, \mu}(u, v)$是强制的, 从而序列$\{(u_k, v_k)\}$$X$中是有界的.因此, 存在$\{(u_k, v_k)\}$的子列(不妨仍记为$\{(u_k, v_k)\}$)及$(u_0^+, v_0^+)\in X$, 使得在$X$

$ u_k\rightharpoonup u_0^+, v_k\rightharpoonup v_0^+. $

由引理2.4可得

$ J_{\lambda, \mu}(u_k, v_k)\rightarrow\delta_{\lambda, \mu}^+<0, G_{\lambda, \mu}(u_0^+, v_0^+)>0. $

接下来我们证明在$X$中有

$ u_k\rightarrow u_0^+, v_k\rightarrow v_0^+ . $

假设不然, 则有

$ \begin{eqnarray}\label{3.1} \|u_0^+\|_{W_0^{1, p}}< \lim\limits_{k\rightarrow\infty}\inf\|u_k\|_{W_0^{1, p}} \quad \mbox{及} \quad\|v_0^+\|_{W_0^{1, p}}< \lim\limits_{k\rightarrow\infty}\inf\|v_k\|_{W_0^{1, p}}. \end{eqnarray} $ (3.1)

$ E_{u, v}(t)=\bar{S}(t)-G_{\lambda, \mu }(u, v), $

其中$\bar{S}(t)$与(2.31) 式中的$\bar{S}(t)$表达式相同.显然, 当$t\rightarrow 0^+$$E_{u, v}(t)\rightarrow -\infty $, 当$t\rightarrow \infty $$E_{u, v}(t)\rightarrow -G_{\lambda, \mu }(u, v).$由于$E' _{u, v}(t)=\bar{S}'(t), $类似于引理2.6的证明过程可得:存在$\bar{t} _{\max }>0, $函数$E_{u, v}(t)$$\bar{t}_{\max }$点处取得最大值.从而当$t\in (0, \bar{t} _{\max })$$E_{u, v}(t)$单调递增, 当$t\in (\bar{t}_{\max }, +\infty )$$E_{u, v}(t)$单调递减.由于$\int_\Omega h(x)|u_0 ^+ |^\alpha |v_0 ^+|^\beta {\rm d}x>0, $从而由引理2.6可知存在唯一的$t_0 ^+ <\bar{t}_{\max }$使得$(t_0 ^+u_0^+, t_0 ^+ v_0^+)\in M_{\lambda, \mu }^+(\Omega)$且有

$ J_{\lambda, \mu}(t_0 ^+u_0^+, t_0 ^+v_0^+)=\inf\limits_{0\leq t\leq \bar{t}_{\max }}J_{\lambda, \mu }(tu_0 ^+, tv_0^+). $

从而

$ \begin{eqnarray} E_{u_0 ^+, v_0^+}(t_0^+)&=& a\|(u_0 ^+, v_0^+)\|^p(t_0^+)^{p-r}+b(\|\nabla u_0 ^+\|_p^{p(k+1)}+\|\nabla v_0 ^+\|_p^{p(k+1)})(t_0^+)^{p(k+1)-r}\nonumber\\ &&-(t_0^+)^{\alpha+\beta-r}\int_\Omega h(x)|u_0^+|^\alpha |v_0^+|^\beta {\rm d}x - G_{\lambda, \mu }(u_0^+, v_0^+)\nonumber\\ &=&(t_0 ^+)^{-r}(a\|(t_0^+u_0^+, t_0^+v_0^+)\|^p+b(\|\nabla (t_0^+u_0^+)\|_p^{p(k+1)}+\|\nabla (t_0^+v_0^+)\|_p^{p(k+1)})\nonumber \\ &&-\int_\Omega h(x)|t_0^+u_0^+|^\alpha |t_0^+v_0^+|^\beta {\rm d}x- G_{\lambda, \mu}(t_0^+u_0^+, t_0^+v_0^+))\nonumber \\ &=& 0.\nonumber \end{eqnarray} $

由(2.31) 式和(3.1) 式可得当$k$充分大时$E_{u_k, v_k}(t_0^+)>0$.由于$(u_k, v_k )\in M_{\lambda, \mu }^+(\Omega), $从而$\bar{t}_{\max }(u_k, v_k)>1.$另一方面,

$ \begin{eqnarray} E_{u_k, v_k }(1)&= &a\|(u_k, v_k)\|^p+b(\|\nabla u_k\|_p^{p(k+1)}+\nabla v_k\|_p^{p(k+1)})\nonumber \\ &&- \int_\Omega h(x)|u_k|^\alpha |v_k|^\beta {\rm d}x -G_{\lambda, \mu }(u_k, v_k )\nonumber \\ &=& 0, \nonumber \end{eqnarray} $

$E_{u_k, v_k }(t)$在区间$t\in (0, \bar{t}_{\max }(u_k, v_k))$单调递增.从而当$t\in(0, 1], $ $k$充分大时$E_{u_k, v_k}(t)<0.$从而$1<t_0^+\leq \bar{t}_{\max }(u_0, v_0).$由于$(t_0^+u_0^+, t_0^+v_0^+)\in M_{\lambda, \mu}^+(\Omega)$以及

$ J_{\lambda, \mu}(t_0^+u_0^+, t_0^+v_0^+)=\inf\limits_{0\leq t\leq \bar{t}_{\max }(u_0^+, v_0^+)}J_{\lambda, \mu}(t u_0^+, t v_0^+ ), \nonumber $

从而

$ J_{\lambda, \mu}(t_0^+u_0^+, t_0^+v_0^+)<J_{\lambda, \mu}(u_0^+, v_0^+)<\lim\limits_{k\rightarrow \infty}J_{\lambda, \mu}(u_k, v_k)=\delta_{\lambda, \mu}^+ .\nonumber $

矛盾!因此$\{u_k\}$在空间$W_0 ^{1, p}(\Omega)$中强收敛于元$u_0^+$, $\{v_k\}$在空间$W_0 ^{1, p}(\Omega)$中强收敛于元$v_0^+ $.从而当$k\rightarrow \infty $

$ J_{\lambda, \mu}(u_k, v_k )\rightarrow J_{\lambda, \mu}(u_0^+, v_0^+)=\delta_{\lambda, \mu}^+ .\nonumber $

因此$(u_0^+, v_0^+)$$J_{\lambda, \mu}$$M_{\lambda, \mu}^+(\Omega)$上的一个极小值点.由于$J_{\lambda, \mu }(u_0^+, v_0^+)=J_{\lambda , \mu }(|u_0^+|, |v_0^+|)$, $(|u_0^+|, |v_0^+|)\in M_{\lambda, \mu }^+(\Omega )$, 故由引理2.3得$(|u_0^+|, |v_0^+|)$是问题(1.1) 的一个非平凡的非负解.

命题3.2  若$0<\lambda \|g_1 \|_\infty+\mu \|g_2 \|_\infty<\eta ^* $, 则泛函$J_{\lambda, \mu }$$M_{\lambda, \mu }^- (\Omega)$上有一个极小值点$(u_0 ^-, v_0^- )$满足:

(ⅰ) $J_{\lambda, \mu}(u_0 ^-, v_0^-)=\delta_{\lambda, \mu}^-;$

(ⅱ) $(u_0^-, v_0^- )$是问题(1.1) 的一个弱解, 且在$\Omega $$u_0^- \geq 0, $ $v_0^-\geq 0$, $u_0^- \not\equiv 0, $ $v_0^- \not\equiv 0.$

  由引理2.2可知泛函$J_{\lambda, \mu}(u, v)$$M_{\lambda, \mu}^-(\Omega)$上强制、有下界.从而, 存在一极小化序列$\{(u_n, v_n)\}\subseteq M_{\lambda, \mu}^-(\Omega )$使得

$ \lim\limits_{n\rightarrow \infty}J_{\lambda, \mu}(u_n, v_n)=\inf\limits_{(u, v)\in M_{\lambda , \mu}^-(\Omega)} J_{\lambda, \mu}(u, v). $

$ J_{\lambda, \mu}(u, v)$的强制性可知序列$\{(u_n, v_n)\}$$X$中有界.因此, 存在$\{(u_n, v_n)\}$的一子序列(不妨仍记为$\{(u_n, v_n)\}$)及$(u_0^-, v_0^-)\in X $, 使得$\{(u_n, v_n)\}\rightharpoonup{(u_0^-, v_0^-)}$.由引理2.4可得当$(u, v)\in M_{\lambda , \mu}^-(\Omega )$$ J_{\lambda, \mu}(u, v)>0$.因此

$ \inf\limits_{(u, v)\in M_{\lambda, \mu}^-(\Omega)} J_{\lambda, \mu}(u, v)>0. $

另一方面, 由(2.9) 式可知$\int_{\Omega}h(x)|u_n|^\alpha |v_n|^\beta {\rm d}x>0$.因此由文献[13]可知

$ \begin{eqnarray} \int_{\Omega}h(x)|u_0^- |^\alpha|v_0^-|^\beta {\rm d}x>0 . \end{eqnarray} $ (3.2)

下面我们证明$\{u_n\}, \{v_n\}$在空间$W_0^{1, p}(\Omega)$中分别强收敛于$u_0^-$$v_0^-$.假设不然, 则有

$ \|u_0^-\|_{W_0^{1, p}}< \lim\limits_{n\rightarrow\infty}\inf\|u_n\|_{W_0^{1, p}} , \quad\|v_0^-\|_{W_0^{1, p}}< \lim\limits_{n\rightarrow\infty}\inf\|v_n\|_{W_0^{1, p}}. $

由于$\int_{{\Bbb R}^N}h(x)|u_0^-|^\alpha |v_0^-|^\beta {\rm d}x>0$, 从而由引理2.6可知存在唯一的$t_0^-$使得$(t_0^-u_0^-, t_0^-v_0^-)\in M_{\lambda , \mu}^-(\Omega)$.对于任意的$(u_n, v_n)\in M_{\lambda , \mu}^-(\Omega )$, 由引理2.5, 对于任意的$t\geq0$成立$J_{\lambda, \mu}(u_n, v_n)\geq J_{\lambda, \mu}(tu_n, tv_n)$.从而

$ J_{\lambda, \mu}(t_0^-u_0^-, t_0^-v_0^-)<\lim\limits_{k\rightarrow \infty}J_{\lambda, \mu}(t_0^-u_n, t_0^-v_n)\leq \lim\limits_{n\rightarrow \infty}J_{\lambda, \mu}(u_n, v_n)=\delta_{\lambda, \mu}^-. $

矛盾!因此, 在空间$W_0^{1, p}(\Omega)$$u_n\rightarrow u_0^-, v_n\rightarrow v_0^-$.从而, 当$n\rightarrow\infty$

$ J_{\lambda, \mu}(u_n, v_n)\rightarrow J_{\lambda, \mu}(u_0^-, v_0^-)=\delta_{\lambda, \mu}^-. $

由于$J_{\lambda, \mu}(u_0^-, v_0^-)=J_{\lambda , \mu}(|u_0^-|, |v_0^-|)$, $(|u_0^-|, |v_0^-|)\in M_{\lambda , \mu}^-(\Omega)$, 类似于命题3.1的证明可得$(u_0^-, v_0^-)$是问题(1.1) 的一个非负解.

接下来, 我们叙述并证明本文的主要结论.

定理3.1  如果参数$\lambda, \mu $满足$0<\lambda \|g_1 \|_\infty +\mu \|g_2 \|_\infty <\eta ^* $, 则问题(1.1) 至少存在两个解$(u_0 ^+, v_0^+)$$(u_0^-, v_0^-)$使得$u_0^\pm \geq 0, v_0^\pm \geq 0$$u_0^\pm \not\equiv 0, v_0^\pm \not\equiv 0.$

  由命题3.1和命题3.2, 问题(1.1) 至少存在两个非平凡的非负解$(u_0^+, v_0^+)\in M_{\lambda , \mu}^+(\Omega)$$(u_0^-, v_0^-)\in M_{\lambda, \mu}^-(\Omega)$.由于$M_{\lambda , \mu}^+(\Omega )\bigcap M_{\lambda, \mu}^-(\Omega)=\emptyset$, 从而解$(u_0^+, v_0^+)$和解$(u_0^-, v_0^-)$是不同的.证毕.

参考文献
[1] Kirchhoff G. Mechanik. Teubner:Leipzly, 1883 http://www.oalib.com/references/18608009
[2] Lions J L. On some questions in boundary value problems of mathematical physics. North-Holland Math Studies, 1978, 30: 284–346. DOI:10.1016/S0304-0208(08)70870-3
[3] Alves C O, Corrêa F J S A. On existence of solutions for a class of problem involving a nonlinear operator. Comm Appl Nonlinear Anal, 2001, 8: 43–56.
[4] Alves C O, Corrêa F J S A, Ma T F. Positive solutions for a quasilinear elliptic equation of Kirchhoff type. Comput Math Appl, 2005, 49: 85–93. DOI:10.1016/j.camwa.2005.01.008
[5] Corrêa F J S A, Menezes S D B. Existence of solutions to nonlocal and singular elliptic problems via Galerkin method. Electron J Differential Equations, 2004, 19: 1–10.
[6] Perera K, Zhang Z. Nontrivial solutions of Kirchhoff-type problems via the Yang index. J Differential Equations, 2006, 221: 246–255. DOI:10.1016/j.jde.2005.03.006
[7] Chipot M, Lovat B. Some remarks on nonlocal elliptic and parabolic problems. Nonlinear Anal, 1997, 30: 4619–4627. DOI:10.1016/S0362-546X(97)00169-7
[8] Corrêa F J S A. Opositive solutions of nonlocal and nonvariational elliptic problems. Nonlinear Anal, 2004, 59: 1147–1155. DOI:10.1016/j.na.2004.08.010
[9] Bozhkov Y, Mitidieri E. Existence of multiple solutions for quasilinear systems via fibering method. J Differential Equations, 2003, 190: 239–267. DOI:10.1016/S0022-0396(02)00112-2
[10] Rasouli S H, Afrouzi G A. The Nehari manifold for a class of concave-convex elliptic systems involving the p-Laplacian and nonlinear boundary condition. Nonlinear Anal, 2010, 73: 3390–3401. DOI:10.1016/j.na.2010.07.021
[11] 李琴, 杨作东. 带有非线性边界条件的非齐次拟线性椭圆型方程组的多解性. 数学物理学报, 2016, 36A(2): 307–316.
Li Q, Yang Z D. Multiple solutions for a class of quasilinear nonhomogeneous elliptic system with nonlinear boundary conditions. Acta Math Sci, 2016, 36A(2): 307–316.
[12] Wu T F. The Nehari manifold for a semilinear system involving sign-changing weight functions. Nonlinear Anal, 2008, 68: 1733–1754. DOI:10.1016/j.na.2007.01.004
[13] Xiu Z H, Chen C S, Huang J C. Existence of multiple solutions for an elliptic system with sign-changing weight functions. J Math Appl, 2012, 395: 531–541.