数学物理学报  2017, Vol. 37 Issue (5): 902-916   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
沈文国
一类半线性周期问题单侧全局区间分歧和定号解
沈文国     
兰州工业学院 基础学科部 兰州 730050
摘要:首先建立一类含不可微非线性项周期问题的单侧全局区间分歧定理.应用上述定理,可以证明一类半线性周期问题主半特征值的存在性.进而,可研究下列半线性周期问题定号解的存在性 $ \begin{array}{ll} &-x''+q(t)x=\alpha x^{+}+\beta x^{-}+ra(t)f(x),\,\,0<t<T,\\& x(0)=x(T),x'(0)=x'(T),\nonumber \end{array} $ 其中r≠0是一个参数,qaC([0,T],(0,∞)),αβC[0,T],x+=max{x,0},x-=-min{x,0};fC $\mathbb{R},\mathbb{R}$ ),当s≠0时,sfs)>0成立,并且f0∈[0,∞)且f∈(0,∞)或f0∈[0,∞]且f=0,其中 $f_{0}= \lim\limits_{|s| \rightarrow0} f(s)/s, f_{\infty}=\lim\limits_{|s| \rightarrow+\infty} f(s)/s.$ .
关键词单侧全局区间分歧    半线性问题    定号解    周期问题    
Unilateral Global Interval Bifurcation and One-Sign Solutions for the Half-Linear Periodic Problems
Shen Wenguo     
Department of Basic Courses, Lanzhou Institute of Technology, Lanzhou 730050
Abstract: In this paper, we establish a unilateral global bifurcation result from interval for a class of periodic problems with nondifferentiable nonlinearity. By applying the above result, we shall prove the existence of the principal half-eigenvalues for a class of half-linear periodic boundary problems. Moreover, we also investigate the existence of one-sign solutions for the following half-linear periodic problems. $ \begin{array}{ll} &-x''+q(t)x=\alpha x^{+}+\beta x^{-}+ra(t)f(x),\,\,0<t<T,\\& x(0)=x(T),x'(0)=x'(T),\nonumber \end{array} $ where r≠0 is a parameter, q, aC([0, T], (0, ∞)), α, βC[0, T], x+=max{x, 0}, x-=-min{x, 0};fC( $\mathbb{R},\mathbb{R}$ ), sf(s)>0 for s≠0, and f0∈[0, ∞) and f∈(0, ∞) or f0∈[0, ∞] and f=0, where $f_{0}= \lim\limits_{|s| \rightarrow0} f(s)/s, f_{\infty}=\lim\limits_{|s| \rightarrow+\infty} f(s)/s.$ .
Key words: Unilateral interval bifurcation     Half-linear problems     One-sign solutions     Periodic problems    
1 引言

近年来, 一些学者研究了周期问题[1-10].另有一些学者应用锥上不动点理论研究了周期问题[11-15].同时, 应用分歧技巧, Ma等[16-18]研究了二阶周期边值问题解的存在性问题. 2012年, Dai等[19]建立了一类周期边值问题单侧全局分歧结果.然而, 以上文献中, 非线性项在原点均是可微的.

1977年, Berestycki[20]研究了一类含不可微非线性项的二阶问题并且建立了一类重要的区间分歧定理. Schmitt和Smith[21]提升了文献[20]中的结果.最近, Ma和Dai[22]建立了一类含不可微非线性项的二阶边值问题单侧全局分歧结果.随后, Dai等[23-24]考虑了相似于文献[22]的区间分歧问题, Dai和Ma[25]亦考虑了相似的高维问题.

受上述文献启发, 本文将研究下列周期问题

$\begin{array}{ll}&- x''+q(t)x=\lambda a(t)x+F(t, x, \lambda), \, \, 0<t<T, \\& x(0)=x(T), x'(0)=x'(T), \end{array}$ (1.1)

其中$\lambda$是一个参数且$a$$q$满足

(H1) $a\in C([0, T], (0, \infty))$;

(H2) $q\in C([0, T], (0, \infty))$.

同时, $F=f+g$, 其中$f$$g$$[0, T]\times {\Bbb R}^{2}$上连续且满足

(H3) 对$t\in[0, T]$, $\lambda\in {\Bbb R}$, $0<\vert x\vert\leq1$, 都有$\left|\frac{f(t, x, \lambda)}{x}\right|\leq M_{1}$成立, 其中$M_{1}$是一个正常数;

(H4) 对$t\in[0, T]$, $\lambda$在有界集中, 在$x=0$附近, 都有$g(t, x, \lambda)=o(\vert x\vert)$一致成立.

$p=2$时, 由文献[19, 定理2.3, 定理3.1]可得

引理1.1  假设(H1)和(H2)成立.可得如下结果.

(ⅰ) 线性特征值问题

$\begin{array}{ll}&- x''+q(t)x=\lambda a(t)x, \, \, 0<t<T, \\& x(0)=x(T), x'(0)=x'(T) \end{array}$ (1.2)

存在一列正特征值

$0<\lambda_0<\lambda_{1}\leq\lambda_{2}<\lambda_{3}\leq\cdot\cdot\cdot; $

(ⅱ) 假如$0<\lambda<\lambda_{0}$则问题(1.2)没有非平凡解;

(ⅲ) $\lambda_{0}$是一个简单的特征值且它对应的特征函数在[0, T]上, 要么是正的要么是负的;

(ⅳ) 对应于$\lambda\neq\lambda_{0}$的任何特征函数$x$改变符号.

$Y=C[0, T]$装备范数$\|x\|_{\infty}=\max\limits_{t\in[0, T]}|x|.$$E=\{x\in C^{1}[0, T]|x(0)=x(T), x'(0)=x'(T)\} $装备范数$\|x\|=\max\limits_{t\in[0, T]}|x(t)|+\max\limits_{t\in[0, T]}|x'(t)|$.

定义线性算子$L:D(L)\subset E\rightarrow Y$

$Lx = - \hat Lx = - x'' + q(t)x,x \in D(L),$

其中$ D(L)=\{x\in C^{2}[0, T]|x(0)=x(T), x'(0)=x'(T)\}. $$L$是一个闭算子且$L^{-1}:Y\rightarrow E$是全连续的.

$P^{+}=\{x\in E|x>0, t\in(0, T)\}$, $P^{-}=-P^{+}, $ $P=P^{+}\cup P^{-}.$ $P^{+}$$P^{-}$$E$中不相交的开集.令${\cal S}^{\nu}$记为(1.1)式的非平凡解集在${\Bbb R}\times E$中的闭包.

在满足条件(H1)--(H4)时, 可建立问题(1.1)的区间分歧定理.在上述定理基础上, 研究下列半线性特征值问题

$\begin{array}{ll}&- x''+q(t)x=\lambda a(t)x+\alpha x^{+}+\beta x^{-}, \, \, 0<t<T, \\& x(0)=x(T), x'(0)=x'(T), \end{array}$ (1.3)

其中$x^{+} = \max\{x, 0\}, x^{-}= -\min\{x, 0\}$, $\alpha , \beta\in C[0, T]$.我们将得出问题(1.3)存在两个简单的主半特征值$\lambda^{+}$$\lambda^{-}$, 相应的解分别在$\{\lambda^{+}\}\times P^{+}$$\{\lambda^{-}\}\times P^{-}$中.

进而, 可研究下列周期问题定号解的存在性问题

$\begin{array}{ll}&- x''+q(t)x=\alpha x^{+}+\beta x^{-}+ra(t)f(x), \, \, 0<t<T, \\& x(0)=x(T), x'(0)=x'(T). \end{array}$ (1.4)

注1.1   当$\alpha =\beta=0$时, Ma等[16]研究了问题(1.4), Dai等[19]研究了问题(1.4)正解的存在性, 推广了文献[16]的结果.

注1.2   对于$a(t)=1, t\in[0, T]$, 应用Prüfer方法, Binding和Rynne [26]研究了问题(1.1)的谱理论和度理论特性.接着, 他们研究了半线性问题解的存在性与不存在性问题.

注1.3   由于$\alpha x^{+}+\beta x^{-}$, 问题(1.4)的非线性项在原点和无穷远处是线性的.显然, 文献[16-19, 27-31]的分歧结果不能直接用来获得我们的结果.

注1.4  应用单侧区间分歧定理, 可得定号解的存在性, 推广了文献[19] ($p=2$)的结果.

2 单侧全局分歧

$E$是一个实Banach空间且装备范数$\|\cdot\|$.考虑算子方程

$u=\lambda Bu+H(\lambda, u),$ (2.1)

其中$B$是一个紧线性算子, $H:{\Bbb R}\times E\rightarrow E$是紧的且在$u=0$$\lambda$在有界区间中, $H=o(\|u\|)$一致成立.假如$B$的特征值$\mu$重数为$1$${\cal S}$表示问题(2.1)的非平凡解集.由Dancer[29]可知, 存在两个不同的无界连续统${\cal C}^{+}_{\mu}$${\cal C}^{-}_{\mu}$组成了从$(\mu, 0)$发出的${\cal S}$的分歧${\cal C}_{\mu}$, 且${\cal C}^{+}_{\mu}\cap {\cal C}^{-}_{\mu}\neq\{(\mu, 0)\}$.

考虑问题

$\begin{array}{ll}&- x''+q(t)x=\mu a(t)x+g(t, x, \mu), \, \, 0<t<T, \\& x(0)=x(T), x'(0)=x'(T), \end{array}$ (2.2)

其中$\mu$是正常数.令(H1)和(H2)成立.假如$g : (0, 1)\times {\Bbb R}^{2} \rightarrow {\Bbb R}$满足Carathéodory条件, 且

$\lim\limits_{x\rightarrow0}\frac{g(t, x, \lambda)}{|x|}=0$ (2.3)

对于a.e. $t\in(0, T)$和有界集中的$\mu$一致成立.令$\lambda_{0}$是线性问题(1.2)的一个主特征值.由文献[19]可知, 当$p=2$时, 问题(2.2)存在两个不同的无界连续统${\cal C}^{+}_{\mu}$${\cal C}^{-}_{\mu}$组成了从$(\lambda_{0}, 0)$发出的问题(2.2)的分歧${\cal C}_{\mu}$, 且满足

引理2.1   ${\cal C}^{+}$${\cal C}^{-}$都是无界的且

${\cal C}^{\nu}\subset({\Bbb R}\times P^{\nu})\cup\{(\lambda_{0}, 0)\}, $

其中$\nu\in\{+, -\}$.

由文献[19], 可得

引理2.2  假设(H1), (H2), (H3)和(H4)成立, $ (\lambda, x)$是问题$(1.1)$的一个非平凡解且$x$有一个重零点, 则$x\equiv0$.

本节主要定理如下.

定理2.1  假设(H1), (H2), (H3)和(H4)成立.令$d_{1}=M_{1}/a_{0}$, 其中$a_{0}=\min\limits_{t\in[0, 1]}a(t)$, 且令$I=[\lambda_{0}-d_{1}, \lambda_{0}+d_{1}]$, 则${\cal S}^{\nu}\cup(I\times\{0\})$的包含$I\times \{0\}$的连通分支${\cal D}^{\nu}$是无界的且属于$({\Bbb R}\times P^{\nu})\cup(I\times\{0\})$, 其中$\nu=+$, $\nu=-$.

考察辅助问题

$\begin{array}{ll}&- x''+q(t)x=\lambda a(t)x+f(t, x|x|^{\epsilon}, \lambda)+g(t, x, \lambda), \, \, 0<t<T, \\& x(0)=x(T), x'(0)=x'(T). \end{array}$ (2.4)

先证如下关键引理成立.

引理2.3  令$\epsilon_{n}\rightarrow 0$, $0<\epsilon_{n}<1$.假如存在序列$(\lambda_{n}, x_{n})\in {\Bbb R}\times P^{\nu}$使得$(\lambda_{n}, x_{n})$是问题(2.4)对应于$\epsilon=\epsilon_{n}$的一个非平凡解, 且在${\Bbb R}\times E$$(\lambda_{n}, x_{n})\rightarrow (\lambda, 0)$, 则$\lambda\in I$.

  不失一般性, 假设$\|x_{n}\|\leq 1$.令$y_{n}=x_{n}/\|x_{n}\|$, 则$y_{n}$满足

$\begin{array}{ll} &{ - {y_n}^{\prime \prime }}+q(t)y_{n}=\lambda a(t)y_{n}+f_{n}(t)+g_{n}(t), \, \, 0<t<T, \\& y_{n}(0)=y_{n}(T), y_{n}'(0)=y_{n}'(T), \end{array}$ (2.5)

其中

$f_{n}(t)=\frac{f(t, x_{n}|x_{n}|^{\epsilon_{n}}, \lambda_{n})}{\|x_{n}\|}, g_{n}(t)=\frac{g(t, x_{n}, \lambda_{n})}{\|x_{n}\|}.$

$\overline{g}(t, x, \lambda)=\max\limits_{0\leq|s|\leq x}|g(t, s, \lambda)|$, $t\in (0, T)$, 且$\lambda$在有界集中, 则$\overline{g}$对于$x$是非减的且

$\lim\limits_{x\rightarrow0^{+}}\frac{\overline{g}(t, x, \lambda)}{x}=0$ (2.6)

对于$t\in(0, T)$和有界集中的$\lambda$一致成立.进而从(2.6)式可得

$ \frac{|g(t, x, \lambda)|}{\|x\|}\leq\frac{\overline{g}(t, |x|, \lambda)}{\|x\|}\leq\frac{\overline{g}(t, \|x\|_{\infty}, \lambda)}{\|x\|}\leq\frac{\overline{g}(t, \|x\|, \lambda)}{\|x\|}\rightarrow0, \|x\|\rightarrow0 $

对于$t\in(0, T)$和有界集中的$\lambda$一致成立.进而

$\lim\limits_{n\rightarrow\infty}g_{n}(t)=0.$ (2.7)

显然, 对$t\in[0, T]$, 由(H2)可得

$|f_{n}(t)|=\frac{|f(t, x_{n}|x_{n}|^{\epsilon_{n}}, \lambda_{n})|}{\|x_{n}\|} \leq\frac{|x_{n}|^{\epsilon_{n}}M_{1} |x_{n}|}{\|x_{n}\|} =|x_{n}|^{\epsilon_{n}}M_{1} |y_{n}| \leq M_{1} |y_{n}|.$ (2.8)

由(2.5), (2.7)和(2.8)式可得$y_{n}$$C^{2}$中有界.由Arzelà-Ascoli定理, 在$E$中可假设$y_{n}\rightarrow y$$\|y\|=1$.显然, $y\in \overline{P^{\nu}}$.假设$y\in P^{\nu}$, 由引理2.2, 可得$y\equiv0$, 与$\|y\|=1$产生矛盾.故$y\in P^{\nu}$.

下证$\lambda$有界.为此, 将$y_{n}$$\varphi _{\nu}$用于Picone恒等式[32] (或文献[33]中的(0.4)式), 其中$\varphi _{\nu}\in P^{\nu}$是(1.2)式对应于$\lambda_{0}$的特征函数.由于$y_{n}$满足

$ \left[{{y_n}^{\prime \prime }}-q(t)y_{n}\right]+\left[\lambda a(t)y_{n}+\frac{f(t, x_{n}|x_{n}|^{\epsilon}, \lambda_{n})}{\|x_{n}\|}+\frac{g(t, x_{n}, \lambda_{n})}{\|x_{n}\|}\right]=0 $

$\varphi _{\nu}$满足

$ ({\varphi _v}^{\prime \prime }-q(t)\varphi _{\nu})+\lambda_{0} a(t)\varphi _{\nu}=0. $

不失一般性仅考虑$\nu=+$.由Picone[32] (或文献[33]中的(0.4)式)可得

$\int_{0}^{T}\left[\frac{y_{n}}{\varphi _{+}}\left(\varphi _{+}y_{n}'-y_{n} \varphi _{+}'\right)\right]'{\rm d}t=A_{1}+B_{1},$ (2.9)
$ A_{1}=\int_{0}^{T}\frac{y_{n}}{\varphi _{+}}\left(\varphi _{+} \widehat{L}y_{n}-y_{n}\widehat{L}\varphi _{+}\right){\rm d}t, B_{1}=\int_{0}^{T}\left(y_{n}'-\frac{y_{n}}{\varphi _{+}}\varphi _{+}'\right)^{2}{\rm d}t. $

现证$\varphi _{+}(0)\neq0$$\varphi _{+}(T)\neq0$.假设$\varphi _{+}(0)=0$$\varphi _{+}(T)=0$, 则$\varphi _{+}(T)=\varphi _{+}(0)=0$.结合$\varphi _{+}(t)\geq0, t\in[0, T]$, 可得$\varphi _{+}'(0)\geq0, \varphi _{+}'(T)\leq0$.进而, 可得$\varphi _{+}'(0)=\varphi _{+}'(T)=0$.由引理2.2可得$\varphi _{+}(t)\equiv0, \forall t\in[0, T]$, 产生矛盾.

进而, 可得(2.9)式的左边等于零.显然$B_{1}\geq0$.进而

$0\geq A_{1}=\int_{0}^{T}\frac{y_{n}}{\varphi _{+}}(\varphi _{+}\widehat{L}y_{n}-y_{n}\widehat{L}\varphi _{+}){\rm d}t \\ \quad \quad \quad =\int_{0}^{T}(\lambda_{0}-\lambda_{n})a(t)y_{n}^{2}{\rm d}t -\int_{0}^{T}y_{n}[f_{n}(t)+g_{n}(t)]{\rm d}t.$ (2.10)

相似的

$0\geq A_{2}=\int_{0}^{T}\frac{\varphi _{+}}{y_{n}}[\varphi _{+}\cdot \widehat{L}y_{n}-y_{n}\cdot \widehat{L}\varphi _{+}]{\rm d}t \\ \quad \quad \quad =\int_{0}^{T}(\lambda_{n}-\lambda_{0})a(t)\varphi _{+}^{2}{\rm d}t +\int_{0}^{T}\frac{\varphi _{+}^{2}}{y_{n}}[f_{n}(t)+g_{n}(t)]{\rm d}t.$ (2.11)

由(2.7), (2.8)和(2.10)式, 令$n\rightarrow+\infty$, 可得

$\int_{0}^{T}(\lambda_{0}-\lambda)a(t)y^{2}{\rm d}t\leq\lim\limits_{n \rightarrow\infty}\int_{0}^{T}f_{n}(t)y {\rm d}t\leq M_{1}\int_{0}^{T}y^{2}{\rm d}t.$ (2.12)

由(2.7), (2.8)和(2.11)式, 同理可得

$\int_{0}^{T}(\lambda-\lambda_{0})a(t)\varphi _{+}^{2}{\rm d}t\leq\lim\limits_{n \rightarrow\infty}\int_{0}^{T}\frac{\varphi _{+}^{2}}{y_{n}}f_{n}(t){\rm d}t\leq M_{1}\int_{0}^{T}\varphi _{+}^{2}{\rm d}t.$ (2.13)

进而, 假若$\lambda\leq\lambda_{0}, $ $\int_{0}^{T}\left[(\lambda_{0}-\lambda)a_{0}-M_{1}\right]y^{2}{\rm d}t\leq0, $$\lambda\geq\lambda_{0}-d_{1}.$同理, 假若$\lambda\geq\lambda_{0}, $ $\int_{0}^{T}\left[(\lambda-\lambda_{0}) a_{0}{\rm d}t-M_{1}\right]\varphi _{+}^{2}{\rm d}t \leq0$, 则$\lambda\leq\lambda_{0}+d_{1}$.因此, 可得$\lambda\in I$.

定理2.1的证明  仅证${\cal D}^{+}$的情况, 同理可证${\cal D}^{-}$.令${\cal D}^{+}$${\cal S}^{+}\cup(I\times\{0\})$的包含$I\times \{0\}$的紧连通分支.分两步.

(1) 先证${\cal D}^{+}\subset({\Bbb R}\times P^{+})\cup(I\times\{0\}).$

对任何$(\lambda, x)\in {\cal D}^{+}$, 存在两种情况.

(ⅰ) $x\in P^{+}$.

(ⅱ) $x \in \partial {P^ + }$.

当满足(ⅰ)时, 显然$(\lambda, x)\in {\Bbb R}\times P^{+}$.由(ⅱ)可得$x$在[0, T]中至少有一对零点, 由引理2.2可得$x\equiv0$.因此, 存在序列$(\lambda_{n}, x_{n})\in {\Bbb R}\times P^{+}$使得$(\lambda_{n}, x_{n})$是(2.4)式对应于$\epsilon= 0$的一个解, 且在${\Bbb R}\times P^{+}$$(\lambda_{n}, x_{n})\rightarrow(\lambda, 0)$.由引理2.3, 可得$\lambda\in I$, i.e., $(\lambda, x)\in ( I\times\{0\})$.因此, ${\cal D}^{+}\subset ({\Bbb R}\times P^{+})\cup(I\times\{0\})$.

(2) 再证${\cal D}^{+}$无界.

反设${\cal D}^{+}$有界.用类似于文献[20]中定理1的论证, 可以找到${\cal D}^{+}$的一个领域$\mathcal{O}$使得$\partial \mathcal{O}\cap{\cal S}^{+}=\emptyset$.

考虑问题$(2.4)$.对于$\epsilon>0$, 易知$f(t, x|x|^{\epsilon}, \lambda)+g(t, x, \lambda)$满足条件(2.3).令${\cal S}_{\epsilon}$是问题(2.4)的非平凡解.由引理2.1, 存在两个从$(\lambda_{0}, 0)$分歧出的无界连通分支${\cal D}_{\epsilon}^{+}\subseteq {\cal S}_{\epsilon}^{+}$${\cal D}_{\epsilon}^{-}\subseteq{\cal S}_{\epsilon}^{-}$, 使得${\cal D}^{\nu}_{\epsilon}\subset({\Bbb R}\times P^{\nu})\cup\{(\lambda_{0}, 0)\}.$因此, 对所有$\epsilon>0$存在$(\lambda_{\epsilon}, x_{\epsilon})\in {\cal D}^{+}_{\epsilon}\cap\partial \mathcal{O}$.因为在${\Bbb R}\times P^{+}$$\mathcal{O}$是无界的, 由方程(2.4)可知$(\lambda_{\epsilon}, x_{\epsilon})$${\Bbb R}\times C^{4}$中是有界且不依赖与$\epsilon$.由于$L^{-1}$是紧的, 可找到序列$\epsilon_{n}\rightarrow0$使得$(\lambda_{\epsilon}, x_{\epsilon})$收敛于(1.1)式的解$(\lambda, x)$.因此$x\in \overline{P^{+}}$.假如$x\in P^{+}$, 则由引理2.2可得$x\equiv0$.由于$ {\cal D}^{+}_{\epsilon}\cap\partial \mathcal{O}$${\Bbb R}\times P^{+}$的闭子集, 故记$(\lambda, x)\in {\cal D}^{+}_ {\epsilon}\cap\partial \mathcal{O}$.由引理2.3, $\lambda\in I, $$\mathcal{O}$的定义矛盾.另外, 假如$x\in P^{+}$, 则$(\lambda, x)\in ( {\cal S}^{+}\cap\partial \mathcal{O})$, 与${\cal S}^{+}\cap\partial \mathcal{O}=\emptyset$相矛盾.

由定理2.1, 易得

推论2.1  在${\Bbb R}\times E$中, 存在从$I\times\{0\}$分歧出的问题(1.1)解的两个无界子连续统${\cal C}^{+}$, ${\cal C}^{-}$, 使得${\cal C}^{\nu}\subset({\Bbb R}\times P^{\nu})\cup(I\times\{0\})$$\nu=+$$\nu=-$.

注2.1  定理2.1和推论2.1的结果是单侧全局的, 相比文献[16]中主要结果, 它们给出了更多分歧信息.

3 半线性问题的谱

这部分, 将考虑半线性问题(1.3).由于在锥$u > 0$$u<0$中, 问题(1.3)是正齐次的且是线性的, 故称之为半线性的.相似于文献[20], 假若问题(1.3)存在非平凡解$(\lambda, x_{\lambda})$, 则称$\lambda$是问题(1.3)的半特征值.假若对问题(1.3)的所有解$(\lambda, y)$都有$y=cx_{\lambda}, c > 0$, 则称$\lambda$是简单的.假若半特征值对应的特征函数是正的或负的, 则称之为主半特征值.

为了证明定理3.1, 需要得出下列比较结果.

引理3.1  对于$t\in(0, T)$$b_{i}(t)\in C(0, T), i = 1, 2$, 令$b_{2}(t)\geq \max\{b_{1}(t), b_{1}(t)+\alpha +\beta, b_{1}(t)-\alpha -\beta\}$.令$u_{1}, u_{2}$分别是下列方程的解

$\begin{array}{l} u'' - q(t)u + {b_i}(t)u + \alpha {u^ + } + \beta {u^ - } = 0,i = 1,2,t \in (0,T),\\ u(0) = u(T),u'(0) = u'(T). \end{array} $

假如$(c, d)\subseteq(0, T)$, $u_{1}(c) = u_{1}(d) = 0$, $u_{1}(t)\neq0$, 则要么存在$\tau\in(c, d)$使得$u_{2}(\tau ) = 0$成立, 要么$b_{2} = b_{1}$$b_{2} = b_{1}+\alpha +\beta$$b_{2} = b_{1}-\alpha -\beta$并且对于常数$\mu\neq0$, $u_{2}(t) = \mu u_{1}(t)$成立.

  分四种情况.

(ⅰ)当$t\in(c, d)$时, $u_{1}(t)>0$, $u_{2}(t)>0$.由Picone恒等式[32] (或文献[33]中的(0.4)式), 可得

$ \int_{c}^{d}\left[\frac{u_{1}}{u_{2}}(u_{2}u_{1}'-u_{1}u_{2}')\right]'{\rm d}t =\int_{c}^{d}\frac{u_{1}}{u_{2}}\left(u_{2}\widehat{L}u_{1}-u_{1}\widehat{L}u_{2}\right){\rm d}t+\int_{c}^{d}\left[u_{1}'-\frac{u_{1}}{u_{2}}u_{2}'\right]^{2}{\rm d}t. $

$\int_{c}^{d}\left[\frac{u_{1}}{u_{2}}(u_{2}u_{1}'-u_{1}u_{2}')\right]'{\rm d}t =\int_{c}^{d}\left\{(b_{2}-b_{1})u_{1}^{2}+\left[u_{1}'-\frac{u_{1}}{u_{2}}u_{2}'\right]^{2}\right\}{\rm d}t.$ (3.1)

(3.1)式左边等于

$ -\lim\limits_{t\rightarrow c^{+}}\frac{u_{1}[u_{2}u_{1}'-u_{1}u_{2}']}{u_{2}}+\lim\limits_{t\rightarrow d^{-}}\frac{u_{1}[u_{2}u_{1}'-u_{1}u_{2}']}{u_{2}}=-H_{c}+H_{d}. $

现证$H_{c}=0$.假若$u_{2}(c)\neq0$, 则$H_{c}=0$.假若$u_{2}(c)=0$, 由引理2.2, 可得$u_{2}'(c)>0$.由罗尔定理可得

$ H_{c}=\lim\limits_{t\rightarrow c^{+}}\frac{u_{1}[u_{2}u_{1}'-u_{1}u_{2}']}{u_{2}}=\lim\limits_{t\rightarrow c^{+}}\frac{u_{1}'[u_{2}u_{1}'-u_{1}u_{2}']+u_{1}[u_{2}u_{1}'-u_{1}u_{2}']'}{u_{2}'}=0. $

相似地, 可证$H_{d}=0$.因此, (3.1)式左边等于零.故(3.1)式右边亦等于零.则存在常数$\mu\neq0$使得$u_{2} = \mu u_{1}$$b_{2} = b_{1}$.

相似与(ⅰ), 可证(ⅱ) $u_{1}(t)>0$, $u_{2}(t)<0$. (ⅲ) $u_{1}(t)<0$, $u_{2}(t)<0$. (ⅳ) $u_{1}(t)<0$, $u_{2}(t)>0$.证毕.

由引理3.1, 可得

引理3.2   假设(H1)和(H2)成立.令$g_{n}\in C([0, T], (0, +\infty))$, 则

$ \lim\limits_{n\rightarrow+\infty}g_{n}(t)=+\infty, \, t\in(0, T). $

$y_{n}$是下列方程的一个解

$ \left\{ \begin{array}{ll} {{y_n}^{\prime \prime }}(t)-q(t)y_{n}+g_{n}(t)y_{n}+\alpha y_{n}^{+}+\beta y_{n}^{-}=0, 0<t<T,&\\ y_{n}(0)=y_{n}(T), y_{n}'(0)=y_{n}'(T)=0.& \\ \end{array} \right. $

$n\rightarrow+\infty$时, 则$y_{n}$$(0, T)$上一定变号.

$\alpha ^{0} := \max\limits_{t \in[0, T]}|\alpha (t)|$, $\beta^{0} := \max\limits_{t \in[0, T]}|\beta(t)|$.通过简单计算可得

$g_{n}(t)+\alpha \frac{y_{n}^{+}}{y_{n}}+\beta \frac{y_{n}^{-}}{y_{n}}y_{n}^{-}\geq g_{n}(t)-\alpha ^{0}-\beta^{0}, t \in(0, T).$

$j\rightarrow+\infty$, 假设

$ g_{n}(t)-\alpha ^{0}-\beta^{0}\geq \lambda_{j}, t \in(0, T), $

其中$\lambda_{j}$是下列问题第$j$个特征值

$ \left\{ \begin{array}{ll} {u''}(t)-q(t)u+\lambda u=0, 0<t<T,&\\ u(0)=u(T), u'(0)=u'(T).& \\ \end{array} \right. $

$\psi_{j}$是对应与$\lambda_{j}$的特征函数.当$j\rightarrow\infty$, 由于$\lambda_{j}\rightarrow\infty$, 由引理1.1可知, $\psi_{j}$$(0, T)$上一定变号.假若$\alpha =\beta= 0$, 引理3.1亦成立.由引理3.1, 当$n\rightarrow+\infty$时, 可得$y_{n}$$(0, T)$上变号.

定理3.1  问题(1.3)存在两个半特征值$\lambda^{+}$$\lambda^{-}$, 对应的半线性解分别在$\{\lambda^{+}\}\times P^{+}$$\{\lambda^{-}\}\times P^{-}$中, 且除了$\lambda^{+}$$\lambda^{-}$, 没有别的半线性特征值对应于正的或负的函数.

  由定理2.1, 问题(1.3)至少存在一个解$(\lambda^{\nu}, x^{\nu})\in {\Bbb R}\times P^{\nu}$, 其中$\nu = +$$ \nu = -$.由正齐次问题(1.3)可知$\{(\lambda^{\nu}, cx^{\nu}), c>0\}$是在$\{\lambda^{\nu}\}\times P^{\nu}$中的半线性解.由引理2.2可知, 问题(1.3)的任何非平凡解$u$属于$ P^{\nu}$.

以下分两步仅证$\nu=+$. (ⅰ)对任何问题(1.3)的解$(\lambda, x)$$x\in P^{\nu}$, 可得$\lambda=\lambda^{\nu}$.

对于$x$$x^{+}$, 在$[0, T]$上应用Picone恒等式[32] (或文献[33]中的(0.4)式), 可得

$\int_{0}^{T}\left\{\frac{x^{+}}{x}[x(x^{+})'-x^{+}x']\right\}'{\rm d}t=\int_{0}^{T}\frac{x^{+}}{x}(x\widehat{L}x^{+}-x^{+}\widehat{L}x){\rm d}t+\int_{0}^{T}\left[(x^{+})'-\frac{x^{+}}{x}x'\right]^{2}{\rm d}t.$ (3.2)

相似与引理2.3, 可得$\lambda^{+}\leq\lambda$, 同理可得$\lambda^{+}\geq\lambda$, 即$\lambda^{+}=\lambda$.

(ⅱ) $x=c \lambda^{\nu}$, 其中$c$是常数.

因为$\lambda=\lambda^{+}$, 由(3.2)式可得$\int_{0}^{T}\left[(x^{+})'-\frac{x^{+}}{x}x'\right]^{2}{\rm d}t=0.$$x=cx^{+}$, 其中$c$是常数.

现考虑问题

$\begin{array}{ll} &{ - x''}+q(t)x=\lambda a(t)x+\alpha x^{+}+\beta x^{-}+g(t, x, \lambda), \, \, 0<t<T, \\& x(0)=x(T), x'(0)=x'(T), \end{array}$ (3.3)

其中$g$满足(H4).可得如下结论.

定理3.2   对于$\nu=+, -$, $(\lambda^{\nu}, 0)$是问题(3.3)的一个分歧点.进而, 存在问题(3.3)的一个解的无界连通分支${\cal D}^{\nu}$, 使得${\cal D}^{\nu}\subset(({\Bbb R}\times P^{\nu})\cup\{(\lambda^{\nu}, 0)\})$.

  令$\alpha ^{0} := \max\limits_{t \in[0, T]}|\alpha (t)|$$\beta^{0} := \max\limits_{t \in[0, T]}|\beta(t)|$, 且$I_{0}=\left[\lambda_{0}-\frac{\alpha ^{0}+\beta^{0}}{a_{0}}, \lambda_{0}+\frac{\alpha ^{0}+\beta^{0}}{a_{0}}\right]$.由推论2.1可知, 在${\Bbb R}\times E$中存在从$I_{0}\times\{0\}$分歧出的问题(3.3)的两个无界子连续统${\cal D}^{+}$${\cal D}^{-}$, 满足${\cal D}^{\nu}\subset({\Bbb R}\times P^{\nu})\cup(I_{0}\times\{0\})$, 其中$\nu=+$$\nu=-$.下证$(\lambda^{\nu}, 0)$是问题(3.3)的一个分歧点.事实上, 假若存在问题(3.3)的收敛于$(\lambda, 0)$的一列解$(\lambda_{n}, x_{n})$.令$y_{n}=\frac{x_{n}}{\|x_{n}\|}$, 则$y_{n}$是下列问题的一个解

$ y_{n}=L^{-1}\left(\lambda_{n} ay_{n}+\alpha y_{n}^{+}+\beta y_{n}^{-}+\frac{g(t, x_{n}, \lambda)}{\|x_{n}\|}\right). $

$n\rightarrow+\infty$时, 由(2.7)式和$L^{-1}$的紧性可得, $y_{n}\rightarrow y_{0}$.进而

$ Ly_{0}=\lambda a(t)y_{0}+\alpha y_{0}^{+}+\beta y_{0}^{-} $

$\|y_0\|=1.$$\lambda=\lambda^{\nu}$, $\nu\in\{+, -\}$.

4 半线性特征值问题的定号解

本节研究周期问题(1.4)定号解的存在性问题.其中$a$$q$分别满足(H1)和(H2).另外, 假设$f$满足

(H5) $sf(s)>0$, $s\neq0;$

(H6) $f_{0}, f_{\infty}\in (0, +\infty);$

(H7) $f_{0}=0$$f_{\infty}\in (0, \infty);$

(H8) $f_{0}\in (0, \infty)$$f_{\infty}=0;$

(H9) $f_{0}=\infty$$f_{\infty}=0;$

(H10) $f_{0}=0$$f_{\infty}=0$.

其中

${f_0} = \mathop {\lim }\limits_{|s| \to 0} \frac{{f(s)}}{s},\qquad {f_\infty } = \mathop {\lim }\limits_{|s| \to + \infty } \frac{{f(s)}}{s}.$

首先, 引述下列引理.

引理4.1 [34]   令$X$是一个Banach空间, 令$\{C_{n}|n=1, 2, \cdots \}$$X$的一个闭连通子序列集.假设

(ⅰ)存在$z_{n}\in C_{n}, \, \, n=1, 2, \cdots $$z^{\ast}\in X$, 使得$z_{n}\rightarrow z^{\ast}$;

(ⅱ) $r_{n}=\sup\{\|x\||x\in C_{n}\}=\infty;$

(ⅲ)对所有$R>0$, $\Big(\bigcup\limits_{n=1}^{\infty}C_{n}\Big)\cap B_{R}$$X$中的相对紧集, 其中

$ B_{R}=\{x\in X|\|x\|\leq R\}. $

则在${\Bbb D}$中, 存在一个无界连通集$C$$z^{\ast}\in C, $其中${\Bbb D}:=\limsup\limits_{n\rightarrow\infty}C_{n}=\{x\in X|\exists\{n_{i}\}\subset{\Bbb N}$, $x_{n_{i}}\in C_{n_{i}}$, 使得$x_{n_{i}}\rightarrow x\}$ (可参考文献[35]).

先考虑

$\begin{array}{ll} &{ - x''}+q(t)x=\alpha x^{+}+\beta x^{-}+\lambda ra(t)f(x), \, \, 0<t<T, \\& x(0)=x(T), x'(0)=x'(T), \end{array}$ (4.1)

其中$\lambda>0$是一个参数.令$\zeta\in C({\Bbb R}, {\Bbb R})$使得

$ f(x)=f _{0}x+ \zeta(x) $

$\lim\limits_{|x| \rightarrow0}\frac{\zeta(x)}{x}=0. $考虑

$\begin{array}{ll} &{ - x''}+q(t)x=\lambda ra(t)f_{0}x+\alpha x^{+}+\beta x^{-}+\lambda ra(t)\zeta(x), \, \, 0<t<T, \\& x(0)=x(T), x'(0)=x'(T) \end{array}$ (4.2)

作为从$x\equiv0$分歧出的一个分歧问题.

将定理3.2应用到问题(4.2), 可得

引理4.2  对于$\nu= +, -$, $(\frac{\lambda^{\nu}}{rf_{0}}, 0)$是问题(4.2)的一个分歧点.进而, 存在问题(4.2)的一个无界连续统${\cal D}^{\nu}$, 使得${\cal D}^{\nu}\subset(({\Bbb R}\times P^{\nu})\cup\{(\frac{\lambda^{\nu}}{rf_{0}}, 0)\})$.

下列可得本文主要结论.

定理4.1  假设(H1), (H2), (H5)和(H6)成立.对于$\nu=+, -$, 要么$ \frac{\lambda^{\nu}}{f_{\infty}}<r<\frac{\lambda^{\nu}}{f_{0}}$成立, 要么$ \frac{\lambda^{\nu}}{f_{0}}<r<\frac{\lambda^{\nu}}{f_{\infty}}$成立.则问题$(1.4)$存在两个解$x^{+}$, $x^{-}$, 使得$\nu x^{\nu}>0$.

注4.1  由(H5)和(H6)可知:存在一个正数$Q$, 使得$\frac{f(s)}{s}\geq Q$, $s\neq0$.

注4.2  假若$\alpha =\beta\equiv0$, 定理4.1的结果等价于文献[19] ($p=2$)中定理5.1.可见, 定理4.1推广了文献[19] ($p=2$)中的结果.

定理4.1的证明  显然问题$(4.1)$形如$(1, x)$的解一定产生问题$(1.4)$的一个解$x$.下证${\cal D}^{\nu}$穿过超平面$\{1\}\times E$.为此, 仅证${\cal D}^{\nu}$连接$(\frac{\lambda^{\nu}}{rf_{0}}, 0)$$(\frac{\lambda^{\nu}}{rf_{\infty}}, +\infty)$.令$(\mu_{n}, x_{n})\in {\cal D}^{\nu}\backslash\{(\frac{\lambda^{\nu}}{rf_{0}}, 0)\}$满足

$ |\mu_{n}|+\|x_{n}\|\rightarrow+\infty. $

(ⅰ) $\frac{\lambda^{\nu}}{f_{\infty}}<r<\frac{\lambda^{\nu}}{f_{0}}$.

分两步完成.

(a) 假若对于足够大的$n\in{\Bbb N}$, 存在常数$M>0$, 使得$ \mu_{n}\in(0, M]$, 进而可得$\|x_{n}\|\rightarrow+\infty$.令$\xi\in C({\Bbb R}, {\Bbb R})$使得$f(x)=f_{\infty}x+\xi(x)$$\lim\limits_{|x|\rightarrow+\infty}\frac{\xi(x)}{x}=0$.令$ \overline{\xi}(x)=\max\left\{|\xi(s)|:0\leq|s|\leq x\right\}$, 则$\overline{\xi}$是非减的且

$\lim\limits_{x\rightarrow+\infty}\frac{\overline{\xi}(x)}{x}=0.$ (4.3)

分解

$\begin{array}{ll}&- {x_n}^{\prime \prime }+q(t)x_{n}=\mu_{n}ra(t)f_{\infty}x_{n}+\alpha x_{n}^{+}+\beta x_{n}^{-}+\mu_{n}ra(t)\xi(x_{n}), \, \, t\in(0, T), \\&x_{n}(0)=x_{n}(T), x_{n}'(0)=x_{n}'(T). \end{array}$

上式两端同除$\|x_{n}\|$且令$y_{n}=\frac{x_{n}}{\|x_{n}\|}$.因为$y_{n}$$E$中有界, 可以选择它的一个收敛子序列, 不妨仍记为$y_{n}$, 使得在$C[0, T]$中存在$y\in E$, 满足$y_{n}\rightarrow y$.进而, 由$(4.3)$式和$\overline{\xi}$是减函数, 可得

$ \lim\limits_{n\rightarrow\infty}\frac{|\xi\left(x_{n}(t)\right)|}{\|x_{n}\|}=0, \, \, \forall t\in[0, T]. $

因为

$ \frac{|\xi(x_{n}(t))|}{\|x_{n}\|}\leq\frac{\overline{\xi}(|x_{n}(t)|)}{\|x_{n}\|} \leq\frac{\overline{\xi}(\|x_{n}(t)\|_{\infty})}{\|x_{n}\|}\leq\frac{\overline{\xi}( \|x_{n}(t)\|)}{\|x_{n}\|}\rightarrow0, n\rightarrow\infty\, \, \forall t\in[0, T]. $

$L^{-1}$的紧性, 可得

$\begin{array}{ll} &{-y''}+q(t)y=\mu ra(t)f_{\infty}y+\alpha y^{+}+\beta y^{-}, \, \, t\in(0, T), \\&y(0)=y(T), y'(0)=y'(T), \end{array}$

其中$\mu:=\lim\limits_{n\rightarrow\infty}\mu_{n}$.

由于${\cal D}^{\nu}$${\Bbb R}\times E$中是闭的, 故$\|y\|=1$$y\in\overline{{\cal D}^{\nu}}\subseteq{\cal D}^{\nu}$.进而, 由定理3.1, 可得$\mu rf_{\infty}=\lambda^{\nu}$, 因此

$ \mu =\frac{\lambda^{\nu}}{rf_{\infty}}. $

进而${\cal D}^{\nu}$连接$\left(\frac{\lambda^{\nu}}{rf_{0}}, 0\right)$$\left(\frac{\lambda^{\nu}}{rf_{\infty}}, +\infty\right)$.

(b) 以下证明:对充分大的$n\in{\Bbb N}$, 存在$M>0$, 使得$|\mu_{n}|\in(0, M]$.

反设$\lim\limits_{n\rightarrow+\infty}|\mu_{n}|=+\infty$.由$L^{-1}$的紧性可得

$\begin{array}{ll}&- {x_n}^{\prime \prime }+q(t)x_{n}=\mu_{n}ra(t)\tilde{f_{n}}(t)x_{n}+\alpha x_{n}^{+}+\beta x_{n}^{-}, \, \, t\in(0, T), \\&x_{n}(0)=x_{n}(T), x_{n}'(0)=x_{n}'(T), \end{array}$

其中

$ \widetilde {{f_n}}(t) = \left\{ {\begin{array}{*{20}{l}} {\frac{{f({x_n})}}{{{x_n}}},}&{{x_n} \ne 0,}\\ {{f_0},}&{{x_n} = 0.} \end{array}} \right.$

由注4.1, 可得$\lim\limits_{n\rightarrow+\infty}\mu_{n}r\tilde{f_{n}}(t)=\pm\infty$.

$\psi^{\nu}$是对应于$\lambda^{\nu}$的特征函数.假如$\lim\limits_{n\rightarrow+\infty}\mu_{n}r\tilde{f_{n}}(t)=-\infty$, 将$y_{n}$$\psi^{\nu}$应用到引理3.1, 则对于充分大的$n$, $\psi^{\nu}$一定变号, 这是不可能的.因此$\lim\limits_{n\rightarrow+\infty}\mu_{n}r\tilde{f_{n}}(t)=+\infty$.由引理3.2, 对于充分大的$n$, $y_{n}$一定变号, 于$y_{n}\in P^{\nu}$矛盾.

(ⅱ) $\frac{\lambda^{\nu}}{f_{0}}<r<\frac{\lambda^{\nu}}{f_{\infty}}$.

此时, 假如$(\mu_{n}, x_{n})\in {\cal D}^{\nu}$, 使得$\lim\limits_{n\rightarrow\infty}(|\mu_{n}|+\|x_{n}\|)=+\infty.$

由(ⅰ)的(b)可知:存在$M>0$使得对于充分大的$n\in N$, 使得$|\mu_{n}|\in(0, M]$成立.应用(ⅰ)的(a)可得: $ (\mu_{n}, x_{n})\rightarrow\left(\frac{\lambda^{\nu}}{rf_{\infty}}, \infty\right), \, \, n\rightarrow\infty.$即得${\cal D}^{\nu}$连接$\left(\frac{\lambda^{\nu}}{rf_{0}}, 0\right)$$\left(\frac{\lambda^{\nu}}{rf_{\infty}}, +\infty\right)$, 结论得证.

定理4.2  假设(H1), (H2), (H5)和(H7)成立.对于$\nu\in\{+, -\}$, 假设下列条件之一成立.

(ⅰ) 对于$\lambda^{\nu}>0$, 使得$r\in(\frac{\lambda^{\nu}}{f_{\infty}}, +\infty)$成立;

(ⅱ) 对于$\nu\lambda^{\nu}>0, $使得$ r\in(\frac{\lambda^{+}}{f_{\infty}}, +\infty)\cup(-\infty, \frac{\lambda^{-}}{f_{\infty}})$成立;

(ⅲ) 对于$\nu\lambda^{\nu}<0$, 使得$r\in(\frac{\lambda^{-}}{f_{\infty}}, +\infty)\cup(-\infty, \frac{\lambda^{+}}{f_{\infty}})$成立;

(ⅳ) 对于$\lambda^{\nu}<0, $使得$r\in(-\infty, \frac{\lambda^{\nu}}{f_{\infty}})$成立.

则问题$(1.4)$存在两个解$x^{+}$, $x^{-}$, 使得$\nu x^{\nu}>0$.

  仅证(ⅰ), 同理可证(ⅱ), (ⅲ)和(ⅳ).

由文献[36], 定义截断函数$f$如下

$ f^{[n]}(s):=\left\{ \begin{array}{ll} \frac{1}{n}s, &s\in[-\frac{1}{n}, \frac{1}{n}], \\[3mm] \left[f(\frac{2}{n})-\frac{1}{n^{2}}\right](ns-2)+f(\frac{2}{n}), & s\in(\frac{1}{n}, \frac{2}{n}), \\[3mm] -\left[f(-\frac{2}{n})+\frac{1}{n^{2}}\right](ns+2)+f(-\frac{2}{n}), ~ & s\in(-\frac{2}{n}, -\frac{1}{n}), \\[3mm] f(s), &s\in(-\infty, -\frac{2}{n}]\cup[\frac{2}{n}, +\infty). \end{array} \right. $

考虑

$\begin{array}{ll} &{ - x''}+q(t)x=\alpha x^{+}+\beta x^{-}+\lambda ra(t)f^{[n]}(x), \, \, 0<t<T, \\& x(0)=x(T), x'(0)=x'(T). \end{array}$ (4.4)

显然, 可得$\lim\limits_{n\rightarrow+\infty}f^{[n]}(s)= f(s)$, $(f^{[n]})_{0}=\frac{1}{n}$$(f^{[n]})_{\infty}=f_{\infty}$.

相似与定理4.1的证明, 由引理4.2可知, 存在从$(\frac{n\lambda^{\nu}}{r}, 0)$发出的问题(4.4)解的无界连续统${\cal D}^{\nu[n]}$, 使得${\cal D}^{\nu[n]}\subset(({\Bbb R}\times P^{\nu})\cup\{(\frac{n\lambda^{\nu}}{r}, 0)\})$, 且${\cal D}^{\nu[n]}$连接$(\frac{n\lambda^{\nu}}{r}, 0)$$(\frac{\lambda^{\nu}}{rf_{\infty}}, \infty)$.

$z_{n}=(\frac{n\lambda^{\nu}}{r}, 0)$, $z^{\ast}=(\infty, 0)$, 则$z_{n}\rightarrow z^{\ast}$.

因此, 引理4.1中(ⅰ)满足且$z^{\ast}=(\infty, 0)$.

显然

$ r_{n}=\sup\left\{\lambda+\|x\||(\lambda, x)\in {\cal D}^{\nu[n]}\right\}=\infty, $

即引理4.1中(ⅱ)成立.由Arezela-Ascoli定理和$f^{[n]}$的定义可推得引理4.1中(ⅲ)成立.

因此, 由引理4.1可知, $\limsup\limits_{n\rightarrow\infty}{\cal D}^{\nu[n]}$包含一个无界连通分支${\cal D}^{\nu}$$(\infty, 0)\in {\cal D}^{\nu}$.

$\lim\limits_{n\rightarrow+\infty}f^{[n]}(s)= f(s)$, (4.4)式等价于(4.1)式.因为${\cal D}^{\nu[n]}\subset P^{\nu}$, 可得${\cal D}^{\nu}\subset P^{\nu}$.进而由(1.4)式, 可得${\cal D}^{\nu}\subset {\cal S}^{\nu}$.

相似于定理4.1中(ⅱ)的证明, 可证$(\frac{\lambda^{\nu}}{rf_{\infty}}, \infty)\in {\cal D}^{\nu}$.

定理4.3  假设(H1), (H2), (H5)和(H8)成立.对于$\nu\in\{+, -\}$, 假设下列条件之一成立.

(ⅰ) 对于$\lambda^{\nu}>0$, 使得$r\in(\frac{\lambda^{\nu}}{f_{0}}, +\infty)$成立;

(ⅱ) 对于$\nu\lambda^{\nu}>0$, 使得$ r\in(-\infty, \frac{\lambda^{-}}{f_{0}})\cup(\frac{\lambda^{+}}{f_{0}}, +\infty)$成立;

(ⅲ) 对于$\nu\lambda^{\nu}<0$, 使得$r\in(-\infty, \frac{\lambda^{+}}{f_{0}})\cup(\frac{\lambda^{-}}{f_{0}}, +\infty)$成立;

(ⅳ) 对于$\lambda^{\nu}<0, $使得$r\in(-\infty, \frac{\lambda^{\nu}}{f_{0}})$成立.

则问题$(1.4)$存在两个解$x^{+}$, $x^{-}$, 使得$\nu x^{\nu}>0.$

  仅证(ⅰ), 同理可证(ⅱ), (ⅲ)和(ⅳ).

假若$(\lambda, x)$是问题(1.4)的任意一个非平凡解, 在问题(1.4)两端同除以$\|x\|^{2}$, 令$y =\frac{x}{\|x\|^{2}}$, 可得

$\begin{array}{ll} {-y''}+q(t)y=\alpha y^{+}+\beta y^{-}+\lambda ra(t)\frac{f(x)}{\|x\|^{2}}, \, \, 0<t<1, \\[2mm] y(0)=y(1)=0. \end{array}$ (4.5)

定义

$ {\tilde{f}}(y):=\left\{ \begin{array}{ll} \|y\|^{2}f(\frac{y}{\|y\|^{2}}), ~ &\mbox{若} y\neq0, \\[2mm] 0, & \mbox{若} y=0. \end{array} \right. $

进而, 问题(4.5)等价于

$\begin{array}{ll} &{-y''}+q(t)y=\alpha y^{+}+\beta y^{-}+\lambda ra(t){\tilde{f}}(y), \, \, 0<t<1, \\& y(0)=y(1)=0. \end{array}$ (4.6)

显然$(\lambda, 0)$是问题(4.6)的一个解.简单计算可得${\tilde{f}}_{0} = f_{\infty}$, ${\tilde{f}}_{\infty} = f_{0}$.由定理4.2可知, 存在从$(\infty, 0)$发出的问题(4.6)解的一个无界连续统${\cal C}^{\nu}$, 使得${\cal C}^{\nu}\subset(({\Bbb R}\times P^{\nu})\cup\{(\infty, 0)\})$, 且${\cal C}^{\nu}$连接$(\infty, 0)$$(\frac{\lambda^{\nu}}{r{\tilde{f}}_{\infty}}, \infty)$.

由逆映射$y\rightarrow\frac{y}{\|y\|^{2}}=x$, 可得${\cal C}^{\nu}\rightarrow{\cal D}^{\nu}$是问题(1.4)解的一个无界连续统, 并且${\cal D}^{\nu}$连接$(\frac{\lambda^{\nu}}{rf_{0}}, 0)$$(\infty, \infty)$.

定理4.4   假设(H1), (H2), (H5)和(H9)成立.对于$\nu\in\{+, -\}$, 假设下列条件之一成立.

(ⅰ) 对于$\lambda^{\nu}>0$, 使得$r\in(0, +\infty)$成立;

(ⅱ) 对于$\nu\lambda^{\nu}<0$, 使得$r\in(0, +\infty)\cup(-\infty, 0)$成立;

(ⅲ) 对于$\lambda^{\nu}<0, $使得$r\in(-\infty, 0)$成立.

则问题$(1.4)$存在两个解$x^{+}$, $x^{-}$, 使得$\nu x^{\nu}>0.$

  仅证(ⅰ), 同理可证(ⅱ), (ⅲ)和(ⅳ).定义

$ f^{[n]}(s):=\left\{ \begin{array}{ll} \frac{1}{n} s, &s\in(-\infty, -2n]\cup[2n, +\infty), \\[3mm] \frac{2+f(-n)}{n}(s+n)+f(-n), &s\in(-2n, -n), \\[3mm] \frac{2-f(n)}{n}(s-n)+f(n), & s\in(n, 2n), \\ f(s), & s\in[-n, -\frac{2}{n}]\cup[\frac{2}{n}, n], \\[3mm] -\left[f(-\frac{2}{n})+1\right](ns+2)+f(-\frac{2}{n}), ~ &s\in(-\frac{2}{n}, -\frac{1}{n}), \\[3mm] \left[f(\frac{2}{n})-1\right](ns-2)+f(\frac{2}{n}), & s\in(\frac{1}{n}, \frac{2}{n}), \\[3mm] ns, & s\in[-\frac{1}{n}, \frac{1}{n}]. \end{array} \right. $

考虑

$\begin{array}{ll} &{ - x''}+q(t)x=\alpha x^{+}+\beta x^{-}+\lambda ra(t)f^{[n]}(x), \, \, 0<t<T, \\& x(0)=x(T), x'(0)=x'(T). \end{array}$ (4.7)

显然, 可得$\lim\limits_{n\rightarrow+\infty}f^{[n]}(s)= f(s)$, $(f^{[n]})_{0}=n$$(f^{[n]})_{\infty}=\frac{1}{n}$.

由引理4.2, 存在从$(\frac{\lambda^{\nu}}{rn}, 0)$发出的问题(4.7)一个解的无界连续统${\cal D}^{\nu[n]}$, 使得${\cal D}^{\nu[n]}\subset(P^{\nu}\cup\{(\frac{\lambda^{\nu}}{rn}, 0)\})$, 且${\cal D}^{\nu[n]}$连接$(\frac{\lambda^{\nu}}{rn}, 0)$$(\frac{n\lambda^{\nu}}{r}, \infty)$.

相似与定理4.1的证明, 可知$\limsup\limits_{n\rightarrow\infty}{\cal D}^{\nu[n]}$包含一个无界连通分支${\cal D}^{\nu}$$(0, 0)\in {\cal D}^{\nu}$.

相似于定理4.3证明, 可得$(\infty, \infty)\in {\cal D}^{\nu}$.定理得证.

定理4.5  令(H1), (H2), (H5)和(H10)成立.对于$\nu\in\{+, -\}$, 假设下列条件之一成立.

(ⅰ) 对于$\lambda^{\nu}>0$, 存在一个$\lambda_{\nu}^{\nu}>0$, 使得$r\in(\lambda_{\nu}^{\nu}, +\infty);$

(ⅱ) 对于$\nu\lambda^{\nu}>0$, 存在一个$\nu\lambda_{\nu}^{\nu}>0$, 使得$r\in(-\infty, \lambda_{-}^{-})\cup(\lambda_{+}^{+}, +\infty)$成立;

(ⅲ) 对于$\nu\lambda^{\nu}<0$, 存在一个$\nu\lambda_{\nu}^{\nu}<0$, 使得, $r\in(-\infty, \lambda_{+}^{+})\cup(\lambda_{-}^{-}, +\infty)$成立;

(ⅳ) 对于$\lambda^{\nu}<0$T, 存在一个$\lambda_{\nu}^{\nu}<0$, 使得$r\in(-\infty, \lambda_{\nu}^{\nu})$成立.

则问题$(1.4)$存在两个解$x^{+}$, $x^{-}$, 使得$\nu x^{\nu}>0.$

  仅证(ⅰ), 同理可证(ⅱ), (ⅲ)和(ⅳ).

定义

$ f^{[n]}(s):=\left\{ \begin{array}{ll} \frac{1}{n}s, &s\in(-\infty, -2n]\cup[2n, +\infty), \\[3mm] \frac{2+f(-n)}{n}(s+n)+f(-n), & s\in(-2n, -n), \\[3mm] \frac{2-f(n)}{n}(s-n)+f(n), & s\in(n, 2n), \\[3mm] f(s), &s\in[-n, -\frac{2}{n}]\cup[\frac{2}{n}, n], \\[3mm] -\left[f(-\frac{2}{n})+\frac{1}{n^{2}}\right](ns+2)+f(-\frac{2}{n}), ~ & s\in(-\frac{2}{n}, -\frac{1}{n}), \\[3mm] \left[f(\frac{2}{n})-\frac{1}{n^{2}}\right](ns-2)+f(\frac{2}{n}), & s\in(\frac{1}{n}, \frac{2}{n}), \\[3mm] \frac{1}{n}s, & s\in[-\frac{1}{n}, \frac{1}{n}]. \end{array} \right. $

考虑

$\begin{array}{ll} &{ - x''}+q(t)x=\alpha x^{+}+\beta x^{-}+\lambda ra(t)f^{[n]}(x), \, \, 0<t<T, \\& x(0)=x(T), x'(0)=x'(T). \end{array}$ (4.8)

显然, 可得$\lim\limits_{n\rightarrow+\infty}f^{[n]}(s)= f(s)$, $(f^{[n]})_{0}=\frac{1}{n}$$(f^{[n]})_{\infty}=\frac{1}{n}$.

相似与定理4.2的证明, 可知存在一个无界连通分支${\cal D}^{\nu}\subset{\cal S}^{\nu}$ with $(\infty, 0)\in {\cal D}^{\nu}$.

相似与定理4.3的证明, 可得$(\infty, \infty)\in {\cal D}^{\nu}$.

参考文献
[1] Camassa R, Holm D D. An integrable shallow water equation with peaked solitons. Phys Rev Lett, 1993, 71: 1661–1664. DOI:10.1103/PhysRevLett.71.1661
[2] Camassa R, Holm D D, Hyman J M. A new integrable shallow water equation. Adv Appl Mech, 1994, 1: 1–33.
[3] 冯育强, 王蔚敏, 李寿贵. 带有奇异非线性项的分数微分方程周期解的存在性与唯一性. 数学物理学报, 2015, 35A(3): 1059–1070.
Feng Y Q, Wang W M, Li S G. Existence and uniqueness results for the periodic boundary value problems of fractional differential equations with singular nonlinearities. Acta Mathematica Scientia, 2015, 35A(3): 1059–1070.
[4] Constantin A, McKean H P. A shallow water equation on the circle. Commun Pure Appl Math, 1999, 52: 949–982. DOI:10.1002/(ISSN)1097-0312
[5] 魏含玉, 夏铁成. 广义Broer-Kaup-Kupershmidt孤子方程的拟周期解. 数学物理学报, 2016, 36A(2): 317–327.
Wei H Y, Xia T C. Quasi-periodic solution of the generalized Broer-Kaup-Kupershmidt soliton equation. Acta Mathematica Scientia, 2016, 36A(2): 317–327.
[6] 陈少伟, 肖利琴. 一类带大参数的周期Thomas-Fermi-Dirac-von Weizsäcker方程非零解的存在性. 数学物理学报, 2016, 36(5): 965–977.
Chen S W, Xiao L Q. Existence of nontrivial solution of a periodic Thomas-Fermi-Dirac-von Weizsäcker equation with a large. Acta Mathematica Scientia, 2016, 36(5): 965–977.
[7] Johnson R S. Camassa-Holm, Korteweg-de Vries and related models for water waves. J Fluid Mech, 2002, 455: 63–82.
[8] 匡杰, 王泽军. 非齐次Burgers方程周期解的大时间行为. 数学物理学报, 2015, 35(1): 1–14.
Kuang J, Wang Z J. The large time behavior of inhomogeneous Burgers equation with periodic initial data. Acta Mathematica Scientia, 2015, 35(1): 1–14.
[9] 张兴永. 一阶带线性部分Hamilton系统的周期解. 数学物理学报, 2013, 33(5): 894–905.
Zhang X Y. Periodic solutions for the first order Hamiltonian system with linear part. Acta Mathematica Scientia, 2013, 33(5): 894–905.
[10] Constantin A, Lannes D. The hydrodynamical relevance of the Camassa-Holm and Degasperis-Procesi equations. Arch Ration Mech Anal, 2009, 192: 165–186. DOI:10.1007/s00205-008-0128-2
[11] Atici F M, Guseinov G S. On the existence of positive solutions for nonlinear differential equations with periodic boundary conditions. J Comput Appl Math, 2001, 132: 341–356. DOI:10.1016/S0377-0427(00)00438-6
[12] Jiang D Q, Chu J F, O'Regan D, Agarwal R P. Multiple positive solutions to superlinear periodic boundary value problems with repulsive singular forces. J Math Anal Appl, 2003, 286: 563–576. DOI:10.1016/S0022-247X(03)00493-1
[13] Torres P J. Existence of one-signed periodic solutions of some second-order differential equations via a Krasnosel'skii fixed point theorem. J Differential Equations, 2003, 190: 643–662. DOI:10.1016/S0022-0396(02)00152-3
[14] O'Regan D, Wang H Y. Positive periodic solutions of systems of second order ordinary differential equations. Positivity, 2006, 10: 285–298. DOI:10.1007/s11117-005-0021-2
[15] Graef J R, Kong L J, Wang H Y. Existence, multiplicity, and dependence on a parameter for a periodic boundary value problem. J Differential Equations, 2008, 245: 1185–1197. DOI:10.1016/j.jde.2008.06.012
[16] Ma R Y, Xu J, Han X L. Global bifurcation of positive solutions of a second-order periodic boundary value problem with indefinite weight. Nonlinear Anal, 2011, 74: 3379–3385. DOI:10.1016/j.na.2011.02.013
[17] Ma R Y, Xu J, Han X L. Global structure of positive solutions for superlinear second-order periodic boundary value problems. Appl Math Comput, 2012, 218: 5982–5988. DOI:10.1016/j.amc.2011.11.079
[18] 马如云, 高承华. 二阶常微分方程周期解的全局分歧. 数学物理学报, 2009, 29A(5): 1223–1232.
Ma R Y, Gao C H. Global bifurcation of periodic solutions to second order ordinary differential equations. Acta Mathematica Scientia, 2009, 29A(5): 1223–1232.
[19] Dai G W, Ma R Y, Wang H Y. Eigenvalues, bifurcation and one-sign solutions for the periodic p-Laplacian. Commun Pure Appl Anal, 2013, 12(6): 2839–2872. DOI:10.3934/cpaa
[20] Berestycki H. On some nonlinear Sturm-Liouville problems. J Differential Equations, 1977, 26: 375–390. DOI:10.1016/0022-0396(77)90086-9
[21] Schmitt K, Smith H L. On eigenvalue problems for nondifferentiable mappings. J Differential Equations, 1979, 33: 294–319. DOI:10.1016/0022-0396(79)90067-6
[22] Ma R Y, Dai G W. Global bifurcation and nodal solutions for a Sturm-Liouville problem with a nonsmooth nonlinearity. J Funct Anal, 2013, 265: 1443–1459. DOI:10.1016/j.jfa.2013.06.017
[23] Dai G W. Global bifurcation from intervals for Sturm-Liouville problems which are not linearizable. Electr J Qual Theory Differ Equ, 2013, 65: 1–7.
[24] Dai G W, Ma R Y. Unilateral global bifurcation for p-Laplacian with non-p-1-lineariza-tion nonlinearity. Discrete Contin Dyn Syst, 2015, 35(1): 99–116.
[25] Dai G W, Ma R Y. Global bifurcation, Berestycki's conjecture and one-sign solutions for p-Laplacian. Nonlinear Anal, 2013, 91: 51–59. DOI:10.1016/j.na.2013.06.003
[26] Binding P A, Rynne B P. Half-eigenvalues of periodic Sturm-Liouville problems. J Differential Equations, 2004, 206: 280–305. DOI:10.1016/j.jde.2004.05.014
[27] Kielhöfer H. Bifurcation Theory:An Introduction with Applications to PDEs. New York:Springer-Verlag, 2004
[28] Rabinowitz P H. Some global results for nonlinear eigenvalue problems. J Funct Anal, 1971, 7: 487–513. DOI:10.1016/0022-1236(71)90030-9
[29] Dancer E N. On the structure of solutions of non-linear eigenvalue problems. Indiana Univ Math J, 1974, 23: 1069–1076. DOI:10.1512/iumj.1974.23.23087
[30] Dancer E N. Bifurcation from simple eigenvaluses and eigenvalues of geometric multiplicity one. Bull Lond Math Soc, 2002, 34: 533–538. DOI:10.1112/S002460930200108X
[31] López-Gómez J. Spectral Theory and Nonlinear Functional Analysis. Boca Raton:CRC, 2001.
[32] Picone M. Sui valori eccezionali di un parametro da cui dipende un'equazione differenziale lineare ordinaria del second ordine. Ann Scuola Norm Pisa, 1910, 11: 1–141.
[33] Kusano T, Jaros T, Yoshida N. A Picone-type identity and Sturmian comparison and oscillation theorems for a class of half-linear partial differential equations of second order. Nonlinear Anal, 2000, 40: 381–395. DOI:10.1016/S0362-546X(00)85023-3
[34] Ma R Y, An Y L. Global structure of positive solutions for nonlocal boundary value problems involving integral conditions. Nonlinear Anal, 2009, 71: 4364–4376. DOI:10.1016/j.na.2009.02.113
[35] Whyburn G T. Topological Analysis. Princeton: Princeton University Press, 1958.
[36] Ambrosetti A, Calahorrano R M, Dobarro F R. Global branching for discontinuous problems. Comment Math Univ Carolin, 1990, 31: 213–222.