数学物理学报  2016, Vol. 36 Issue (3): 500-506   PDF (274 KB)    
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
张靖
带有Hardy和Sobolev-Hardy临界指标项的扰动椭圆方程的解
张靖    
内蒙古师范大学数学科学学院 呼和浩特 010022
摘要: 考虑如下带有Hardy和Sobolev-Hardy临界指标项的扰动椭圆方程
$\begin{equation}\label{eq0} \left\{\begin{array}{ll} -\Delta u-\mu\frac{u}{|x|^{2}}+\lambda a(x)u^{q}=\frac{|u|^{2^{*}(s)-2}}{|x|^{s}}u, & x\in {\Bbb R}^{N}, \\ u>0,& u\in D^{1,\,2}({\Bbb R}^{N}), \end{array} \right. \end{equation}$ (0.1)
这里$2^{*}(s)=\frac{2(N-s)}{N-2}$ 是Sobolev-Hardy临界指标, $N\geq3$, $\lambda\in{\Bbb R}$, $0\leq s<2$, $1<q<2^{*}-1$, $0\leq \mu<\overline{\mu}=\frac{(N-2)^2}{4}$, $a(x)\in C({\Bbb R}^N)$. 在$|\lambda|$足够小的情况下, 应用临界点理论中的扰动方法来得到方程(0.1)正解的存在性. 接下来考虑anisotropic椭圆方程

$\begin{equation}\label{eq0*} \left\{\begin{array}{ll} -{\rm div} [(1+\lambda b(x))\nabla u]+\lambda a(x)u^q= \mu\frac{u}{|x|^{2}}+\frac{|u|^{2^{*}(s)-2}}{|x|^{s}}u, ~~& x\in {\Bbb R}^N, \\ u>0, & u\in{\cal D}^{1,2}({\Bbb R}^N), \end{array} \right. \end{equation}$ (0.2)
$b(x)\in C({\Bbb R}^N)$. 在$|\lambda|$足够小的情况下, 应用临界点理论中的扰动方法来得到方程(0.2)正解的存在性.

关键词: 扰动方程     Sobolev-Hardy临界指标     正解    
Positive Solutions of Perturbation Elliptic Equation Involving Hardy Potential and Critical Sobolev-Hardy Exponent
Zhang Jing    
College of Mathematics Science, Inner Mongolia Normal University, Hohhot 010022
Foundation Item: Supported by the NSFC (11571187)
Abstract: In this paper, we are concerned with the following elliptic equation involving critical Sobolev-Hardy exponent $$\begin{eqnarray*} \left\{\begin{array}{ll} -\Delta u-\mu\frac{u}{|x|^{2}}+\lambda a(x)u^{q}=\frac{|u|^{2^{*}(s)-2}}{|x|^{s}}u, ~~& x\in{\Bbb R}^{N}, \\ u>0,& u\in D^{1,\,2}({\Bbb R}^{N}), \end{array} \right. \end{eqnarray*}$$ where $2^{*}(s)=\frac{2(N-s)}{N-2}$ is the critical Sobolev-Hardy exponent, $N\geq3$, $\lambda\in{\Bbb R}$, $0\leq s<2$, $1<q<2^{*}-1$, $0\leq \mu<\overline{\mu}=\frac{(N-2)^2}{4}$, $a(x)\in C({\Bbb R}^N)$. We firstly use an abstract perturbation method in critical point theory to obtain the existence results of positive solutions of the equation for small value of|λ|. Secondly, we focus on an anisotropic elliptic equation of the form $$\begin{eqnarray*} \left\{\begin{array}{ll} -{\rm div} [(1+\lambda b(x))\nabla u]+\lambda a(x)u^q=\mu\frac{u}{|x|^{2}}+\frac{|u|^{2^{*}(s)-2}}{|x|^{s}}u, ~~&x\in{\Bbb R}^N, \\ u>0, &u\in D^{1,2}({\Bbb R}^N). \end{array} \right. \end{eqnarray*}$$ The same abstract method is used to yield existence result of positive solutions of the equation for small value of |λ|.
Key words: Perturbed     Critical Sobolev-Hardy Exponent     Positive Solution    
1 引言和主要结果

本文考虑如下带有Hardy和Sobolev-Hardy临界指标项的扰动椭圆方程

$\begin{equation}\label{eq1}\left\{\begin{array}{ll} -\Delta u-\mu\frac{u}{|x|^{2}}+\lambda a(x)u^{q}=\frac{|u|^{2^{*}(s)-2}}{|x|^{s}}u,~~& x\in{\Bbb R}^{N},\\ u>0,& u\in D^{1,2}({\Bbb R}^{N}),\end{array} \right.\end{equation}$ (1.1)

这里$2^{*}(s)=\frac{2(N-s)}{N-2}$是Sobolev-Hardy临界指标,$N\geq3$,$\lambda\in{\Bbb R}$,$0\leq s<2$,$1<q<2^{*}-1$,$0\leq \mu<\overline{\mu}=\frac{(N-2)^2}{4}$,$a(x)\in C({\Bbb R}^N)$.

当$s=0$时,方程(1.1)已经被广泛的研究,并且得到很多有意义的结果.那么一个很自然的问题就提出来了,已经得到的这些结果在$0<s<2$时是否成立?本文的想法由文献[5]和[6]引出,并且应用临界点理论中的扰动方法来得到方程(1.1)的正解存在性.

假设$a(x)$满足条件:

(a$_1)$ $a(x)\in L^\infty({\Bbb R}^N)\cap L^r({\Bbb R}^N)$,这里$r\leq\frac{2^*}{2^*-(q+1)}$;

(a$_2)$ $a(x)\geq0$,并且在${\Bbb R}^N$的一个正测度子集$\Omega$里,有$a(x)\geq\nu>0$.

类似参考文献[5],我们应用文献[1, 2, 3]中提出的扰动方法.对于足够小的$\lambda$,方程(1.1)可以看做是临界问题

$\begin{equation}\label{eq2}\left\{\begin{array}{ll} -\Delta u-\mu\frac{u}{|x|^{2}}=\frac{|u|^{2^{*}(s)-2}}{|x|^{s}}u,~~& x\in{\Bbb R}^N,\\ u>0,& u\in{\cal D}^{1,2}({\Bbb R}^N)\end{array} \right.\end{equation}$ (1.2)

的一个扰动.令$\beta=\sqrt{\overline{\mu}-\mu}$,文献[6]证明了对于$0<\mu<\overline{\mu}$,$\varepsilon>0$,方程(1.2)的所有解的形式为$u_\varepsilon(x)=\varepsilon^{\frac{2-N}{2}} u(\frac{x}{\varepsilon})$,这里

$\begin{equation}\label{eq1^*}u(x)=C_{N,s}\frac{1}{|x|^{\frac{N-2}{2}-\beta}(1+|x|^{\frac{2\beta(2-s)}{N-2}})^\frac{N-2}{2-s}},\end{equation}$ (1.3)

对于任意$\varepsilon>0$都成立,$C_{N,s}>0$是一个关于$N,s$的常数.这些解使得$A_{s}$可达,

$$A_{s}=\inf_{u\in{\cal D}^{1,2}({\Bbb R}^N),u\neq0}\frac{\int_{{\Bbb R}^N}(|\nabla u|^2-\frac{\mu u^2}{|x|^2})}{(\int_{{\Bbb R}^N}\frac{u^{2^*(s)}}{|x|^{s}})^{\frac{2}{2^*(s)}}}.$$

定理1.1 假设$a(x)$满足(a$_1)$-(a$_2)$,并且

$\begin{equation}\label{eq3}N\ge 5,\ \ 1<q<2^*-1,\ \ \beta>1.\end{equation}$ (1.4)

则存在$\overline{\lambda}>0$,$\overline{\varepsilon}\in(0,+\infty)$使得对所有$\lambda\in{\Bbb R},|\lambda|\leq\overline{\lambda}$,方程(1.1)有一个正解$u_\lambda$,而且当$\lambda\rightarrow0$时,在${\cal D}^{1,2}({\Bbb R}^N)$中有

$$u_\lambda\rightarrow u_{\overline{\varepsilon}}.$$
2 扰动方法

在这一节,我们介绍一下临界点理论中的扰动方法,参考文献[2, 3, 5].考虑Hilbert空间$E$上的$C^2$泛函$J_{\lambda}$,定义为$J_{\lambda}(u)=J(u)+\lambda \Gamma(u)$,这里$J,\Gamma\in C^2(E,{\Bbb R})$,$\lambda>0$.对于非扰动的泛函$J$有下列假设:

(1) 存在一个由$J$的临界点组成$d$维$C^2$流形$U$,$d\geq1$.

这样定义的$U$叫做$J$的临界流形.

令$T_uU$表示$U$的切空间.假设

(2) $\forall u\in U$,$D^2J(u)$是一个指标为0的Fredholm算子;

(3) $\forall u\in U$,$T_uU=\ker D^2J(u)$.

下面定理可见参考文献[2].

定理2.1 假设条件(1)-(3)成立,并且存在一个$\Gamma$的临界点$\overline{u}\in U$使得以下条件中任意一个成立:

(i) $\overline{u}$是非退化的;

(ii) $\overline{u}$是一个局部的极大或者极小值点;

(iii) $\overline{u}$是孤立的,并且$\Gamma'$在$\overline{u}$的局部拓扑度$\rm{deg}_{\rm{loc}}(\Gamma',0)\neq0$.\\则当$|\lambda|$足够小时,存在能量泛函$J_{\lambda}$的一个临界点$u_{\lambda}$使得在$\lambda\rightarrow0$时,有$u_{\lambda}\rightarrow\overline{u}$.

方程(1.1)的解是能量泛函$J_{\lambda}$的临界点,这里

$$J_{\lambda}(u)=\frac{1}{2}\int_{{\Bbb R}^N}\bigg(|\nabla u|^2-\mu\frac{u^2}{|x|^{2}}\bigg){\rm d}x-\frac{1}{2^*(s)}\int_{{\Bbb R}^N}\frac{|u|^{2^*(s)}}{|x|^{s}}{\rm d}x+\frac{\lambda}{q+1}\int_{{\Bbb R}^N}a(x)|u|^{q+1}{\rm d}x.$$ $$J(u)=\frac{1}{2}\int_{{\Bbb R}^N}\bigg(|\nabla u|^2-\mu\frac{u^2}{|x|^{2}}\bigg){\rm d}x-\frac{1}{2^*(s)}\int_{{\Bbb R}^N}\frac{|u|^{2^*(s)}}{|x|^{s}}{\rm d}x.$$

令$U=\{u_\varepsilon:\varepsilon>0\}\subset{\cal D}^{1,2}({\Bbb R}^N)$,则$U$是一个由$J$的临界点组成的一维流形,微分同胚于$(0,+\infty)$.值得一提的是当(1.4)式成立时,有$U\subset W^{1,q+1}({\Bbb R}^N)$.为了对能量泛函$J_\lambda$应用扰动方法,由参考文献[4]可知$D^2J(u_\varepsilon)=I-K$,$K$是紧算子,因此$D^2J(u_\varepsilon)$是一个指标为0的Fredholm算子.进而由参考文献[7]可得下述引理:

引理2.1 $T_{u_\varepsilon}U=\ker D^2J(u_\varepsilon)\subset{\cal D}^{1,2}({\Bbb R}^N).$

接下来,需要引入辅助函数$\Gamma:U\rightarrow{\Bbb R}$定义为

$$\Gamma(\varepsilon)=\frac{1}{q+1}\int_{{\Bbb R}^N}a(x)u_{\varepsilon}^{q+1}(x){\rm d}x.$$
3 证明定理1.1

从现在起,用$C$来表示不同的正常数.要证明定理1.1,则需证明$\Gamma$在$\overline{\varepsilon}\in(0,+\infty)$可以取得局部极大值.以下所提到的$u$皆为方程(1.3)中的$u(x)$,为使用方便简记为$u$.

引理3.1 $u^{q+1}\in L^{1}({\Bbb R}^N).$

$\int_{{\Bbb R}^N}u^{q+1}{\rm d}x=\int_{B_{1}(0)}u^{q+1}{\rm d}x+\int_{B_{1}^{c}(0)}u^{q+1}{\rm d}x.$首先考虑等式右边的第一项,

$$\begin{eqnarray*}\int_{B_{1}(0)}u^{q+1}{\rm d}x&=&C\int_{B_{1}(0)}\frac{1}{|x|^{(\frac{N-2}{2}-\beta)(q+1)}(1+|x|^{\frac{2\beta(2-s)}{N-2}})^\frac{(N-2)(q+1)}{2-s}}{\rm d}x\\&\leq& C\int_{B_{1}(0)}\frac{1}{|x|^{(\frac{N-2}{2}-\beta)(q+1)}}{\rm d}x\\&=&C\int_{0}^{1}\frac{\rho^{N-1}}{\rho^{(\frac{N-2}{2}-\beta)(q+1)}}{\rm d}\rho\\&=&C\rho^{N-(\frac{N-2}{2}-\beta)(q+1)}|_{0}^{1},\end{eqnarray*}$$

由(1.4)式知$\beta>1$,则$N-(\frac{N-2}{2}-\beta)(q+1)>0$,所以$\int_{B_{1}(0)}u^{q+1}{\rm d}x<+\infty$.

其次考虑等式右边的第二项,

$$\begin{eqnarray*}\int_{B_{1}^{c}(0)}u^{q+1}{\rm d}x&=&C\int_{B_{1}^{c}(0)}\frac{1}{|x|^{(\frac{N-2}{2}-\beta)(q+1)}(1+|x|^{\frac{2\beta(2-s)}{N-2}})^\frac{(N-2)(q+1)}{2-s}}{\rm d}x\\&\leq& C\int_{B_{1}^{c}(0)}\frac{1}{|x|^{(\frac{N-2}{2}-\beta)(q+1)}(|x|^{\frac{2\beta(2-s)}{N-2}})^\frac{(N-2)(q+1)}{2-s}}{\rm d}x\\&=&C\int_{B_{1}^{c}(0)}\frac{1}{|x|^{(\frac{N-2}{2}+\beta)(q+1)}}{\rm d}x\\&=&C\int_{1}^{+\infty}\frac{\rho^{N-1}}{\rho^{(\frac{N-2}{2}+\beta)(q+1)}}{\rm d}\rho\\&=&C\rho^{N-(\frac{N-2}{2}+\beta)(q+1)}|_{1}^{+\infty},\end{eqnarray*}$$

由(1.4)式知$\beta>1$,则$N-(\frac{N-2}{2}+\beta)(q+1)<0$,所以$\int_{B_{1}^{c}(0)}u^{q+1}{\rm d}x<+\infty$.

最终可得$u^{q+1}\in L^{1}({\Bbb R}^N)$.

引理3.2 当$\varepsilon\rightarrow0$时,有$\Gamma(\varepsilon)\rightarrow0$.

根据变量替换

$$\Gamma(\varepsilon)=\frac{1}{q+1}\int_{{\Bbb R}^N}a(x)u_{\varepsilon}^{q+1}(x){\rm d}x=\frac{\varepsilon^{N-\alpha}}{q+1}\int_{{\Bbb R}^N}a(\varepsilon x)u^{q+1}(x){\rm d}x,$$

这里$\alpha=\frac{(q+1)(N-2)}{2}$.根据引理3.1知$u^{q+1}\in L^1({\Bbb R}^N)$.而且由$a(x)$满足的条件知$a(x)$有界,则$\int_{{\Bbb R}^N}a(\varepsilon x)u^{q+1}(x){\rm d}x\leq{\parallel a\parallel}_\infty\int_{{\Bbb R}^N}u^{q+1}(x){\rm d}x$.再由$N>\alpha$,可得当$\varepsilon\rightarrow0$时,$\Gamma(\varepsilon)\rightarrow0$.

引理3.3 当$\varepsilon\rightarrow+\infty$时,$\Gamma(\varepsilon)\rightarrow0$.

由条件(a$_1)$,固定$1<r<\frac{2^*}{2^*-(q+1)}$使得$a(x)\in L^r({\Bbb R}^N)$.并且令$t=\frac{r}{r-1}$.很容易可得$(q+1)t>2^*$,且$u^{(q+1)t}\in L^1({\Bbb R}^N)$(类似证明可参考引理3.1).由H\"{o}lder不等式可知

$$\begin{eqnarray*}\int_{{\Bbb R}^N}a(\varepsilon x)u^{q+1}(x){\rm d}x&\leq &\bigg(\int_{{\Bbb R}^N}|a(\varepsilon x)|^r{\rm d}x\bigg)^\frac{1}{r}\bigg(\int_{{\Bbb R}^N}u^{(q+1)t}(x){\rm d}x\bigg)^\frac{1}{t}\\&\leq& \varepsilon^\frac{-N}{r}\bigg(\int_{{\Bbb R}^N}|a(x)|^r{\rm d}x\bigg)^\frac{1}{r}\bigg(\int_{{\Bbb R}^N}u^{(q+1)t}(x){\rm d}x\bigg)^\frac{1}{t}.\end{eqnarray*}$$

由上述不等式可知

$$\Gamma(\varepsilon)\leq\frac{\varepsilon^{N-\alpha-\frac{N}{r}}}{q+1}\bigg(\int_{{\Bbb R}^N}|a(x)|^r{\rm d}x\bigg)^\frac{1}{r}\bigg(\int_{{\Bbb R}^N}u^{(q+1)t}(x){\rm d}x\bigg)^\frac{1}{t}.$$

由于$a(x)\in L^{r}({\Bbb R}^N)$,$u^{(q+1)t}\in L^1({\Bbb R}^N)$,且$r<\frac{2^*}{2^*-(q+1)}$说明$N-\alpha-\frac{N}{r}<0$,那么由上述不等式可以得到结论当$\varepsilon\rightarrow+\infty$时,$\Gamma(\varepsilon)$收敛到$0$.

定理1.1的证明 由于$\Gamma$在$U$上是连续的,$\Gamma(0)=0$,且当$\varepsilon\rightarrow+\infty$时,$\Gamma(\varepsilon)\rightarrow0$,则$\Gamma$在某个大于0的$\overline{\varepsilon}$取得极大值.再由定理2.1,可以得到$J_\lambda$有一个临界点$u_\lambda$,当$\lambda\rightarrow0$,在${\cal D}^{1,2}({\Bbb R}^N)$中有$u_\lambda\rightarrow u_{\overline{\varepsilon}}$.并且$u_\lambda\geq0$.进而,应用Harnack不等式[8]可知$u_\lambda>0$.

4 Anisotropic Schrödinger方程

本节考虑方程

$\begin{equation}\label{eq4}\left\{\begin{array}{ll} -{\rm div} [(1+\lambda b(x))\nabla u]+\lambda a(x)u^q=\mu\frac{u}{|x|^{2}}+\frac{|u|^{2^{*}(s)-2}}{|x|^{s}}u,~~& x\in{\Bbb R}^N,\\ u>0,& u\in D^{1,2}({\Bbb R}^N),\end{array} \right.\end{equation}$ (4.1)

这里$N\geq3,\lambda\in{\Bbb R},0\leq \mu<\overline{\mu}=\frac{(N-2)^2}{4}$,$a(x)$满足条件(a$_1$)-(a$_2)$.假设

$(\rm{b_1})$ $b\in C({\Bbb R}^N),b(x)\geq0$,$b(x)\not\equiv0$,并且$\displaystyle{\lim_{|x|\rightarrow+\infty}} b(x)=0$;

$(\rm{b_2})$ $\displaystyle{\lim_{x\rightarrow0}}|x|^{\alpha-N}|b(x)-b(0)|=0$,这里$\alpha=\frac{(N-2)(q+1)}{2}.$

定理4.1 假设$a(x),b(x)$满足(a$_1)$-(a$_2)$,(b$_1)$-(b$_2)$,而且(1.4)式成立.则存在$\overline{\lambda}>0$,$\overline{\varepsilon}\in(0,+\infty)$使得当$\lambda$满足$|\lambda|\leq\overline{\lambda}$,方程(4.1)有正解$u_\lambda$,且当$\lambda\rightarrow0$时,在${\cal D}^{1,2}({\Bbb R}^N)$中有

$$u_\lambda\rightarrow u_{\overline{\varepsilon}}.$$

方程(4.1)的所有解是能量泛函

$$ \begin{eqnarray*}I_{\lambda}(u)&=&\frac{1}{2}\int_{{\Bbb R}^N}\bigg(|\nabla u|^2-\mu\frac{u^2}{|x|^{2}}\bigg){\rm d}x-\frac{1}{2^*(s)}\int_{{\Bbb R}^N}\frac{|u|^{2^*(s)}}{|x|^{s}}{\rm d}x\\&&+\frac{\lambda}{2}\int_{{\Bbb R}^N}\langle b(x)\nabla u,\nabla u\rangle {\rm d}x+\frac{\lambda}{q+1}\int_{{\Bbb R}^N}a(x)|u|^{q+1}{\rm d}x\end{eqnarray*}$$

的临界点.非扰动能量泛函

$$I(u)=\frac{1}{2}\int_{{\Bbb R}^N}\bigg(|\nabla u|^2-\mu\frac{u^2}{|x|^{2}}\bigg){\rm d}x-\frac{1}{2^*(s)}\int_{{\Bbb R}^N}\frac{|u|^{2^*(s)}}{|x|^{s}}{\rm d}x.$$

辅助函数$\Gamma : U\rightarrow{\Bbb R}$定义为

$$\Gamma(u_{\varepsilon})=\Gamma(\varepsilon)=\Gamma_{1}(\varepsilon)+\Gamma_{2}(\varepsilon),$$

这里

$$\Gamma_{1}(\varepsilon)=\frac{1}{q+1}\int_{{\Bbb R}^N}a(x)|u_{\varepsilon}|^{q+1}{\rm d}x=\frac{\varepsilon^{N-\alpha}}{q+1}\int_{{\Bbb R}^N}a(\varepsilon x)|u|^{q+1}{\rm d}x,$$ $$\Gamma_{2}(\varepsilon)=\frac{1}{2}\int_{{\Bbb R}^N}\langle b(x)\nabla u_{\varepsilon},\nabla u_{\varepsilon}\rangle {\rm d}x=\frac{1}{2}\int_{{\Bbb R}^N}\langle b(\varepsilon x)\nabla u,\nabla u\rangle {\rm d}x.$$

引理4.1 如果$\Gamma(0)=\frac{1}{2}\int_{{\Bbb R}^N} \langle b(0)\nabla u,\nabla u\rangle {\rm d}x$,则$\Gamma$在$\varepsilon=0$处连续.

由条件(b$_1)$,则有

$$|\langle b(\varepsilon x)\nabla u,\nabla u\rangle|\leq C|b(\varepsilon x)||\nabla u|^{2}\leq C|\nabla u|^{2}.$$

由$b(x)$的连续性和控制收敛定理可得

$$\lim_{\varepsilon\rightarrow0}\Gamma_2(\varepsilon)=\lim_{\varepsilon\rightarrow0}\frac{1}{2}\int_{{\Bbb R}^N}\langle b(\varepsilon x)\nabla u,\nabla u\rangle {\rm d}x=\frac{1}{2}\int_{{\Bbb R}^N}\langle b(0)\nabla u,\nabla u\rangle {\rm d}x.$$

再由引理3.2,则有

$$\lim_{\varepsilon\rightarrow0}\Gamma(\varepsilon)=\frac{1}{2}\int_{{\Bbb R}^N}\langle b(0)\nabla u,\nabla u\rangle {\rm d}x.$$

证毕.

引理4.2 当$\varepsilon\rightarrow+\infty$时,$\Gamma(\varepsilon)\rightarrow0$.

通过引理3.3,可知当$\varepsilon\rightarrow+\infty$时,$\Gamma_{1}(\varepsilon)\rightarrow0$.由条件(b$_1)$,知当$|x|\rightarrow+\infty$时,$b(x)\rightarrow0$.容易得到当$\varepsilon\rightarrow+\infty$时,$\langle b(\varepsilon x)\nabla u,\nabla u\rangle\rightarrow0$.应用控制收敛定理,可以推断

$$\lim_{\varepsilon\rightarrow+\infty}\Gamma_2(\varepsilon)=\lim_{\varepsilon\rightarrow+\infty}\frac{1}{2}\int_{{\Bbb R}^N}\langle b(\varepsilon x)\nabla u,\nabla u\rangle=0.$$

所以最终可以得到当$\varepsilon\rightarrow+\infty$时,有$\Gamma(\varepsilon)\rightarrow0$.

引理4.3 $\Gamma(\varepsilon)$在某个大于$0$的点$\overline{\varepsilon}$取得极大值.

由引理4.1知$\Gamma(0)=\frac{1}{2}\int_{{\Bbb R}^N} \langle b(0)\nabla u,\nabla u\rangle {\rm d}x$,所以只要能证明当$\varepsilon>0$足够小时,有$\Gamma(\varepsilon)\geq\Gamma(0)$成立,即可得到引理.

首先,我们知道

$$\begin{eqnarray*}\Gamma_1(\varepsilon)&=&\frac{\varepsilon^{N-\alpha}}{q+1}\int_{{\Bbb R}^N}a(\varepsilon x)u^{q+1}(x){\rm d}x\\&\geq&\frac{\varepsilon^{N-\alpha}}{q+1}\int_{\frac{\Omega}{\varepsilon}}a(\varepsilon x)u^{q+1}(x){\rm d}x\\&\geq&\frac{\varepsilon^{N-\alpha}}{q+1}\nu\int_{\frac{\Omega}{\varepsilon}}u^{q+1}(x){\rm d}x.\end{eqnarray*}$$

由于$u^{q+1}\in L^1({\Bbb R}^N)$,则有当$\varepsilon\rightarrow0$时,$\int_{\frac{\Omega}{\varepsilon}}u^{q+1}(x)\rightarrow\int_{{\Bbb R}^N}u^{q+1}(x)$.这时可以推断当$\varepsilon$足够小时,对于$C>0$有

$$\Gamma_1(\varepsilon)\geq C\varepsilon^{N-\alpha}.$$

另一方面

$\begin{equation}\label{eq5}\frac{|\Gamma_2(\varepsilon)-\Gamma_2(0)|}{\varepsilon^{N-\alpha}}\leq\frac{C}{2}\int_{{\Bbb R}^N}\frac{|b(\varepsilon x)-b(0)|}{\varepsilon^{N-\alpha}}|\nabla u(x)|^2{\rm d}x.\end{equation}$ (4.2)

由条件(b$_2)$,注意到当$\varepsilon$足够小时

$$\frac{|b(\varepsilon x)-b(0)|}{\varepsilon^{N-\alpha}}|\nabla u(x)|^2\leq|x|^{N-\alpha}|\nabla u(x)|^2.$$

那么可以推断

$$|\nabla u(x)|^2=\frac{\left[(\frac{N-2}{2}-\beta)\left(1+|x|^{\frac{2\beta(2-s)}{N-2}}\right)+2\beta|x|^{\frac{2\beta(2-s)}{N-2}}\right]^{2}}{|x|^{N-2\beta}\left(1+|x|^{\frac{2\beta(2-s)}{N-2}}\right)^{2(\frac{N-2}{2-s}+1)}},$$

由于$\beta>1$,可得$|x|^{N-\alpha}|\nabla u(x)|^2\in L^1({\Bbb R}^N)$.综上所述,由方程(4.2)和控制收敛定理,可以得到

$$\lim_{\varepsilon\rightarrow0}\frac{|\Gamma_2(\varepsilon)-\Gamma_2(0)|}{\varepsilon^{N-\alpha}}=0.$$

因此当$\varepsilon\rightarrow0$时,

$$|\Gamma_2(\varepsilon)-\Gamma_2(0)|=o(\varepsilon^{N-\alpha}).$$

所以最终可以得到$\Gamma(\varepsilon)$在某个大于0的点$\overline{\varepsilon}$取得极大值.

定理4.1的证明 由引理4.3和定理2.1,我们可以推断能量泛函$I_\lambda$有一个临界点$u_\lambda$.当$|\lambda|$足够小时,在${\cal D}^{1,2}({\Bbb R}^N)$中有$u_\lambda\rightarrow u_{\overline{\varepsilon}}$.首先清楚地知道$u_\lambda\geq0$.此外,应用Harnack不等式[8]可得$u_\lambda>0$.

参考文献
[1] Ambrosetti A, Malchiodi Andrea. Perturbation Methods and Semilinear Elliptic Problems on RN. Boston:Birkhäuser-Verlag, 2006
[2] Ambrosetti A, Badiale M. Variational perturbative methods and bifurcation of bounds states from the the essential spectrum. Proc Roy Soc, Edinburgh A, 1998, 128:1131-1161
[3] Ambrosetti A, Badiale M, Cingolani S. Semiclassical states of nonlinear Schrödinger equations. Arch Rational Mech Anal, 1997, 140:285-300
[4] Brown K J, Stavrakakis N. Global bifurcation results for a semilinear elliptic equation on all RN. Duke Math J, 1996, 85:77-94
[5] Cingolani Silvia. Positive solutions to perturbed elliptic problems in RN involving critical Sobolev exponent. Nonlinear Analysis, 2002, 48:1165-1178
[6] Kang D S, Peng S J. Positive solutions for singular critical elliptic problems. Appl Math Lett, 2004, 17:411-416
[7] Rey O. The role of the Green's function in a non-linear elliptic equation involving the critical Sobolev exponent. J Funct Anal, 1990, 89:1-52
[8] Trudinger N S. On Harnack type inequalities and their application to quasilinear elliptic equations. Comm Pure Appl Math, 1967, 20:721-747