数学杂志  2021, Vol. 41 Issue (4): 365-376   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
赵娇
非线性二阶离散周期边值问题的Ambrosetti-Prodi型结果
赵娇    
西北师范大学数学与统计学院, 甘肃 兰州 730070
摘要:本文研究了二阶离散周期边值问题
$ \left\{ \begin{array}{l} {\Delta ^2}u(t - 1) + f\left( {t, u\left( t \right), \Delta u\left( {t - 1} \right)} \right) = s, \;\;\;\;t \in {\left[ {1, T} \right]_{\mathbb{Z}}}, \\ u\left( 0 \right) - u\left( T \right) = \Delta u\left( 0 \right) - \Delta u\left( T \right) = 0 \end{array} \right. $
解的个数与参数s的关系,其中ftuv):[1,T]$ {\mathbb{Z}}$×$ {\mathbb{R}}$2$ {\mathbb{R}}$关于(uv)∈$ {\mathbb{R}}$2连续,s$ {\mathbb{R}}$.利用上下解方法和拓扑度理论,获得了Ambrosetti-Prodi型结果,推广了已有文献的相关结果.
关键词二阶周期边值问题    Ambrosetti-Prodi型结果    上下解方法    拓扑度理论    
AMBROSETTI-PRODI TYPE RESULTS FOR NONLINEAR SECOND-ORDER DISCRETE PERIODIC BOUNDARY VALUE PROBLEMS
ZHAO Jiao    
College of Mathematics and Statistics, Northwest Normal University, Gansu Lanzhou 730070, China
Abstract: In this paper, we discuss the relationship between the number of the solutions for second-order discrete periodic boundary value problem
$ \left\{ \begin{array}{l} {\Delta ^2}u(t - 1) + f\left( {t, u\left( t \right), \Delta u\left( {t - 1} \right)} \right) = s, \;\;\;\;t \in {\left[ {1, T} \right]_{\mathbb{Z}}}, \\ u\left( 0 \right) - u\left( T \right) = \Delta u\left( 0 \right) - \Delta u\left( T \right) = 0 \end{array} \right. $
and the parameter s, where f(t, u, v): [1, T]$ {\mathbb{Z}}$×$ {\mathbb{R}}$2$ {\mathbb{R}}$ is continuous with respect to (u, v)∈$ {\mathbb{R}}$2, s$ {\mathbb{R}}$. By using the method of the upper and lower solutions and topological degree techniques, Ambrosetti-Prodi type result is obtained, and some related conclusions on this topic are generalized.
Keywords: second-order periodic BVPs     Ambrosetti-Prodi type results     upper and lower solutions     topological degree techniques    
1 引言

Ambrosetti–Prodi型结果描述的是形如

$ \begin{align} F(x) = s \end{align} $ (1.1)

的方程所对应的边值问题解$ x $的个数与参数$ s $之间的关系, 当参数$ s $变化时, 解的个数相应改变. 1972年, A.Ambrosetti和G.Prodi在文献[1] 中对二阶椭圆边值问题

$ \begin{align} \left\{\begin{array}{ll} \Delta u+f(u) = g, \; \; x\in\Omega\\[2ex] u\mid_{\partial\Omega} = 0 \end{array} \right. \end{align} $ (1.2)

进行了研究, 得到如下定理:

定理A   设$ \Omega\subset\mathbb{R}^{n} $是一个边界充分光滑的有界子集, $ g\in C^{0, \alpha}(\bar{\Omega}) $, $ \alpha\in(0, 1) $. 令$ \lambda_{1}, \; \lambda_{2}, \; \lambda_{3}, \; \cdots $为线性齐次问题

$ \left\{\begin{array}{ll} \Delta u+f(u) = 0, \; \; x\in\Omega\\[2ex] u\mid_{\partial\Omega} = 0 \end{array} \right. $

的特征值. 若$ f:\mathbb{R}\rightarrow \mathbb{R} $$ C^{2} $连续的函数, 且满足以下条件:

(A1)   $ f(0) = 0 $;

(A2)   $ f''(t)>0 $;

(A3)   $ \lim\limits_{t\rightarrow {-\infty}}f'(t) = h', \; 0<h'<\lambda_{1} $;

(A4)   $ \lim\limits_{t\rightarrow {+\infty}}f'(t) = h'', \; \lambda_{1}<h''<\lambda_{2} $.

则存在$ M\subset C^{0, \alpha}(\bar{\Omega}) $, 在$ C^{0, \alpha}(\bar{\Omega})\backslash M $中存在两个连通分支$ A_{1} $, $ A_{2} $, 并且有如下结论:

(ⅰ)   若$ g\in A_{1} $, 则问题(1.2) 没有解;

(ⅱ)   若$ g\in A_{2} $, 则问题(1.2) 有两个解;

(ⅲ)   若$ g\in M $, 则问题(1.2) 有一个解.

1986年, Fabry, Mawhin和Nkashama在文献[2] 中运用上下解方法获得二阶周期边值问题

$ \begin{align} \left\{\begin{array}{ll} u''+F(t, u, u') = s, \; \; t\in[0, 2\pi]\\[2ex] u(0) = u(2\pi), \; u'(0) = u'(2\pi) \end{array} \right. \end{align} $ (1.3)

解的Ambrosetti–Prodi型结果, 其中$ F:[0, 2\pi]\times\mathbb{R}^{2}\rightarrow \mathbb{R} $连续且关于$ t $$ 2\pi $周期的, 并满足以下条件:

(B1)   存在$ R_{1}>0 $$ S_{1} $, 使得对任意$ t\in\mathbb{R} $$ x\leqslant-R_{1} $, 有$ F(t, x, 0)>s_{1}>F(t, 0, 0) $成立;

(B2)   $ F $满足Bernstein–Nagumo条件, 即对任意$ R\in(0, +\infty) $, 存在连续函数$ h_{R}:(0, +\infty)\rightarrow [a_{R}, +\infty)(h_{R}>0) $, 使得对所有的$ |x|\leqslant R $, $ t\in\mathbb{R} $, $ y\in\mathbb{R} $, 有$ |F(t, x, y)|\leqslant h_{R}(|y|) $, 及$ \int\limits_{a_{R}}^{+\infty}\frac{s}{h_{R}(s)}ds = +\infty $成立;

(B3)   $ \lim\limits_{|u|\rightarrow \infty}F(t, u, v) = +\infty $$ (t, v)\in[0, 2\pi]\times\mathbb{R} $一致成立.\\ 得到如下定理.

定理B   若函数$ F $连续, 关于$ t $$ 2\pi $周期的, 满足条件(B1)–(B3), 则存在常数$ s_{0}<s_{1} $, 使得当$ s_{0}<s_{1} $, 问题(1.3) 无$ 2\pi $–周期解; 当$ s\in(s_{0}, s_{1}] $时, 问题(1.3) 至少有一个$ 2\pi $–周期解.

近几年来, 对Ambrosetti–Prodi型结果的研究已经拓展到微分方程的多个领域, 然而据了解, 对差分方程Ambrosetti–Prodi型结果的研究较少. 在研究二阶周期边值问题的Ambrosetti–Prodi型结果时, 解的先验界估计是关键的一部分; 同样的, 对于二阶离散周期边值问题, 研究其Ambrosetti–Prodi型结果, 对解的先验界估计是我们所克服的主要困难. Fabry, Mawhin和Nkashama在文献[2] 中给非线性函数$ F $加上了Bernstein–Nagumo条件, 一个自然的问题是, 我们在对二阶离散周期边值问题的Ambrosetti–Prodi型结果的研究中不加Bernstein–Nagumo条件, 能否获得周期解的存在性以及解的个数与参数之间的关系?

基于此, 本文研究二阶离散周期边值问题

$ \begin{align} \left\{\begin{array}{ll} \Delta^{2}u(t-1)+f(t, u(t), \Delta u(t-1)) = s, \; \; \ \ \ \; t\in [1, T]_{\mathbb{Z}}, \\[2ex] u(0)-u(T) = \Delta u(0)-\Delta u(T) = 0 \end{array} \right. \end{align} $ (1.4)

解的个数与参数$ s $的关系, 其中$ T>1 $, $ [1, T]_{\mathbb{Z}} = \{1, 2, \cdots, T\} $, $ \Delta $是前向差分算子, 且满足$ \Delta u(t) = u(t+1)-u(t) $, $ \Delta^{2}u(t) = \Delta(\Delta u(t)) $, $ f(t, u, v):[1, T]_{\mathbb{Z}}\times\mathbb{R}^2\rightarrow \mathbb{R} $关于$ (u, v)\in\mathbb{R} $连续, $ s\in\mathbb{R} $.

本文总假定:

(H1)   $ f:[1, T]_{\mathbb{Z}}\times\mathbb{R}^{2}\rightarrow \mathbb{R} $为连续函数, 且$ \lim\limits_{|u|+|v|\rightarrow \infty}f(t, u, v) = +\infty $对任意$ t\in[1, T]_{\mathbb{Z}} $一致成立.

定理1.1   假定条件(H1) 成立, 则存在常数$ s_{1}\in\mathbb{R} $, 使得

(ⅰ)   当$ s<s_{1} $时, 问题(1.4) 无解;

(ⅱ)   当$ s = s_{1} $时, 问题(1.4) 至少有一个解;

(ⅲ)   当$ s>s_{1} $时, 问题(1.4) 至少有两个解.

2 预备知识

$ X = \{u\mid u:[0, T]_{\mathbb{Z}}\rightarrow \mathbb{R}, \; u(0)-u(T) = \Delta u(0)-\Delta u(T) = 0\} $在范数$ \|u\| = \max\limits_{t\in[1, T]_{\mathbb{Z}}}|u(t)| $下构成banach空间. 令

$ u^{+} = \max\{u, 0\}, \; \; u^{-} = \max\{-u, 0\}. $
$ u_{L} = \min\limits_{t\in[1, T]_{\mathbb{Z}}}u(t), \; \; u_{M} = \max\limits_{t\in[1, T]_{\mathbb{Z}}}u(t). $

定义投影算子$ P, Q:X\rightarrow X $, $ Pu(t) = u(0), \; Qu(t) = \frac{1}{T}\sum\limits_{s = 1}^{T}u(s), \; t\in[1, T]_{\mathbb{Z}}. $算子$ H:X\rightarrow X $定义为$ Hu(t) = \sum\limits_{s = 1}^{T}u(s), \; \; t\in[1, T]_{\mathbb{Z}}. $因为$ f:[1, T]_{\mathbb{Z}}\times\mathbb{R}^{2}\rightarrow \mathbb{R} $连续, 其相应的Nemytskii算子可定义为$ N_{f}:X\rightarrow X $, $ N_{f}(u)(t) = f(t, u(t), \Delta u(t-1)), \; \; t\in[1, T]_{\mathbb{Z}}. $对任意函数$ h\in X $, 定义$ \alpha: = Q_{\phi}(h) $如下

$ \sum\limits_{t = 1}^{T}\phi^{-1}(h(t)-\alpha) = 0, $

进一步, 函数$ Q_{\phi}:X\rightarrow \mathbb{R} $连续.

下面给出二阶周期边值问题

$ \begin{align} \left\{\begin{array}{ll} \Delta^{2}u(t-1) = f(t, u(t), \Delta u(t-1)), \; \; \ \ \ \; t\in [1, T]_{\mathbb{Z}}, \\[2ex] u(0)-u(T) = \Delta u(0)-\Delta u(T) = 0 \end{array}\right. \end{align} $ (2.1)

上下解的定义.

定义2.1   函数$ \alpha\in X $是问题(2.1) 的下解, 是指$ \alpha $满足

$ \left\{\begin{array}{ll} \Delta^{2}\alpha(t-1)\geqslant f(t, \alpha(t), \Delta\alpha(t-1)), \; \; \ \ \ \; t\in [1, T]_{\mathbb{Z}}, \\[2ex] \alpha(0) = \alpha(T), \; \Delta\alpha(0)\geqslant\Delta\alpha(T). \end{array} \right. $

函数$ \beta\in X $是问题(2.1) 的上解, 是指$ \beta $满足

$ \left\{\begin{array}{ll} \Delta^{2}\beta(t-1)\leqslant f(t, \beta(t), \Delta\beta(t-1)), \; \; \ \ \ \; t\in [1, T]_{\mathbb{Z}}, \\[2ex] \beta(0) = \beta(T), \; \Delta\beta(0)\leqslant\Delta\beta(T). \end{array} \right. $

引理2.2    若问题(2.1) 有一个下解$ \alpha(t) $和一个上解$ \beta(t) $, 使得$ \alpha(t)\leqslant\beta(t) $, 则问题(2.1) 有一个解$ u(t) $, 使得$ \alpha(t)\leqslant u(t)\leqslant\beta(t) $; 若$ \alpha $是严格上解, $ \beta(t) $是严格下解, 则有$ \alpha(t)<u(t)<\beta(t) $.

  构造辅助函数

$ \begin{eqnarray*} \tilde{u}(t) = \left\{\begin{array}{lr} \beta(t), \; \; \; \; \; \; u>\beta(t), & \\ u, \; \; \; \; \; \; \; \; \; \; \; \alpha(t)\leqslant u\leqslant\beta(t), & \\ \alpha(t), \; \; \; \; \; \; u<\alpha(t). \end{array} \right. \end{eqnarray*} $

考虑修正问题

$ \begin{align} \left\{\begin{array}{ll} \Delta^{2}u(t-1)-f(t, \tilde{u}(t), \Delta \tilde{u}(t-1))-(u(t)-\tilde{u}(t)) = 0, \; \; \ \ \ \; t\in (1, T)_{\mathbb{Z}}, \\[2ex] u(0)-u(T) = \Delta u(0)-\Delta u(T) = 0. \end{array}\right. \end{align} $ (2.2)

$ \alpha(t)\leqslant u(t)\leqslant\beta(t) $, $ t\in[1, T]_{\mathbb{Z}} $时, 问题(2.2) 和问题(2.1) 等价, 由Brouwer不动点定理可以得到问题(2.2) 至少有一个解, 要证明$ \alpha(t)\leqslant u(t) $, $ t\in[1, T]_{\mathbb{Z}} $. 反设存在$ l\in[0, T+1]_{\mathbb{Z}} $, 使得$ \alpha(l)-u(l)>0 $, $ \alpha(t)-u(t) = \max\limits_{\tau\in[0, T+1]_{\mathbb{Z}}}(\alpha(\tau)-u(\tau))>0 $, 根据$ \alpha(0)-u(0) = \alpha(T)-u(T) $$ \Delta(\alpha(0)-u(0)) = \Delta(\alpha(T)-u(T)) $, 可得$ t\in[0, T]_{\mathbb{Z}} $. 因为当$ \alpha(\tau)-u(\tau)<\alpha(0)-u(0) = \alpha(T)-u(T) $时, $ \tau\in[0, T]_{\mathbb{Z}} $, 有

$ 0>\alpha(1)-u(1)-(\alpha(0)-u(0))\geqslant\alpha(T+1)-u(T+1)-(\alpha(T)-u(T))>0, $

矛盾. 因此

$ \Delta^{2}(\alpha(t-1)-u(t-1)) = \alpha(t+1)-u(t+1)-2(\alpha(t)-u(t))+(\alpha(t-1)-u(t-1))\leqslant0, $

$ \begin{aligned} \Delta^{2}\alpha(t-1)\leqslant\Delta^{2}u(t-1)& = f(t, \tilde{u}(t), \Delta \tilde{u}(t-1))+(u(t)-\tilde{u}(t))\\& = f(t, \alpha(t), \Delta\alpha(t-1))+(u(t)-\alpha(t))\\ &<f(t, \alpha(t), \Delta\alpha(t-1))\leqslant\Delta^{2}\alpha(t-1) \end{aligned}. $

矛盾. 因此$ \alpha(t)\leqslant u(t) $, 类似可证$ u(t)\leqslant\beta(t) $. 即问题(2.1) 至少有一个解$ u(t) $, 使得$ \alpha(t)\leqslant u(t)\leqslant\beta(t) $, $ t\in [1, T]_{\mathbb{Z}} $.

引理2.3    若$ u\in X $, 则$ \|\Delta u(t-1)\|\leqslant T\|(\Delta^{2}u(t-1))^{\pm}\|. $

  根据差分中值定理, $ u(t) $定义在$ [0, T]_{\mathbb{Z}} $上, 存在$ c\in[1, T-1]_{\mathbb{Z}} $, 使得$ \Delta u(c-1)\leqslant\frac{u(T)-u(0)}{T-0} $, 结合边界条件, 当$ t\in(c, T) $, 可以得到

$ \begin{aligned} \Delta u(t-1) = \sum\limits_{s = c}^{t-1}\Delta^{2}u(s-1)+\Delta u(c-1)&\leqslant\sum\limits_{s = c}^{t-1}\Delta^{2}u(s-1)+\frac{u(T)-u(0)}{T-0}\\ &\leqslant\sum\limits_{s = c}^{t-1}\Delta^{2}u(s-1)\leqslant\sum\limits_{s = 1}^{T}\Delta^{2}u(s-1)\\ &\leqslant T\|(\Delta^{2}u(t-1))^{+}\|. \end{aligned} $

$ t\in(0, c) $, 再应用差分中值定理, 得到$ \Delta u(c)\geqslant\frac{u(T)-u(0)}{T-0} $, 结合边界条件, 则有

$ \begin{aligned} \Delta u(t-1) = \sum\limits_{s = c+1}^{t-1}\Delta^{2}u(s-1)+\Delta u(c)&\geqslant-\sum\limits_{s = t-1}^{c+1}\Delta^{2}u(s-1)\geqslant-\sum\limits_{s = 1}^{T}\Delta^{2}u(s-1)\\ &\geqslant-T\|(\Delta^{2}u(t-1))^{+}\|. \end{aligned} $

$ \|\Delta u(t-1)\|\leqslant T\|(\Delta^{2}u(t-1))^{+}\|, $同理可证$ \|\Delta u(t-1)\|\leqslant T\|(\Delta^{2}u(t-1))^{-}\| $.

引理2.4    对问题(1.4), 假设条件(H1) 成立, 对任意常数$ b\in\mathbb{R} $, 若$ b>\min f(t, u, v), \; (t, u, v)\in[1, T]_{\mathbb{Z}}\times\mathbb{R}^{2} $, 则存在$ \hat{\rho}>0 $, 使得对任意$ s\leqslant b $, 问题(1.4) 的所有可能解$ u $均属于开球$ B_{\hat{\rho}} $中.

  令$ \rho_{b} = \max\{b-f(t, u, v), \; (t, u, v)\in[1, T]_{\mathbb{Z}}\times\mathbb{R}^{2}\} $, 则$ \rho_{b}\in(0, \infty) $, 对任意的$ s\leqslant b $, 设$ u $为问题(1.4) 的解, 则$ (\Delta^{2}u(t-1))^{+} = (s-f(t, u(t), \Delta u(t-1)))^{+}\leqslant(b-f(t, u(t), \Delta u(t-1)))^{+}\leqslant \rho_{b}. $对问题(1.4) 方程两端从$ t = 1 $$ t = T $求和分, 再结合周期边界条件, 得到

$ s = \frac{1}{T}\sum\limits_{t = 1}^{T}f(t, u(t), \Delta u(t-1)) = QN_{f}(u). $

根据引理2.3可知, $ \|\Delta u(t-1)\|\leqslant T\|(\Delta^{2}u(t-1))^{\pm}\|\leqslant T\rho_{b}. $由条件(H1) 得对任意$ b\in R $, 存在$ R>0 $, 使得当$ |u|+|v|\geqslant R $时, $ f(t, u, v)>b. $如果$ u_{L}\geqslant R, $那么

$ |u|+|v|\geqslant R, \; s = \frac{1}{T}\sum\limits_{t = 1}^{T}f(t, u(t), \Delta u(t-1)) = QN_{f}(u)>b, $

这与$ s\leqslant b $矛盾, 因此$ u_{L}<R, $同理可得$ u_{M}>-R $.

$ c, d\in[1, T]_{\mathbb{Z}} $, 使得$ u_{M} = u(c) $, $ u_{L} = u(d) $, 由$ u_{M}-u_{L} = \sum\limits_{t = d+1}^{c}\Delta u(t-1) $

$ u_{M}\leqslant u_{L}+\sum\limits_{t = 1}^{T}|\Delta u(t-1)|, $

从而$ \|u\|\leqslant R+T^2\rho_{b}. $$ \hat{\rho}\geqslant R+2T^{2}\rho_{b} $, 则问题(1.4) 的所有可能解均属于开球$ B_{\hat{\rho}} $中.

引理2.5   [9] 问题(2.1) 的解可表示为

$ u = Pu+QN_{f}(u)+[H\circ(I-Q_{\phi})\circ H(I-Q)]\circ QN_{f}(u): = \Phi(u) $

其中$ \Phi:X\rightarrow X $全连续.

引理2.6   [9] 假设存在$ M>0 $, 对任意的$ (t, u, v)\in[0, T]\times\mathbb{R}^{2} $, $ |f(t, u, v)|<M $成立, 若问题(2.1) 存在下解$ \alpha $和上解$ \beta $, 满足$ \alpha(t)\leqslant\beta(t) $, $ t\in[0, T] $, 则问题(2.1) 存在解$ u $, 满足$ \alpha(t)\leqslant u(t)\leqslant \beta(t) $, $ t\in[0, T] $, 若$ \beta, \alpha $为问题(2.1) 的严格上下解, 且有$ \alpha(t)<u(t)<\beta(t) $, 则

$ d_{LS}[I-\Phi, \Omega_{\alpha, \beta}, 0] = 1. $

其中$ \Omega_{\alpha, \beta} = \{u\in X:\alpha(t)<u(t)<\beta(t), \; |u'(t)|<c, \; t\in[0, T]\} $.

$ c\geqslant 2M+r+1, \; r = \|\alpha\|+\|\beta\|. $
3 主要结果的证明

定理1.1的证明  令$ S_{j} = \{s\in\mathbb{R}:(1.1) $至少有$ j $个解$ \} $ $ (j\geqslant 1) $.

(a)   首先宣称$ {S_1} \ne Ø $.

$ s^{\ast}>\max\limits_{t\in[1, T]_{\mathbb{Z}}}f(t, 0, 0) $, 由条件(H1) 知, 存在$ R^{\ast}<0, $使得

$ \min\limits_{t\in[1, T]_{\mathbb{Z}}}f(t, R^{\ast}, 0)>s^{\ast}. $

$ \alpha\equiv R^{\ast}<0 $$ s = s^{\ast} $时问题(1.4) 的严格下解, $ \beta = 0 $是问题(1.4) 的严格上解. 对任意$ s^{\ast}\in\mathbb{R} $, 取$ n>\min\limits_{t\in[1, T]_{\mathbb{Z}}}f(t, R^{\ast}, 0)>s^{\ast}. $$ \rho_{n} = \max\{n-f(t, u, v), \; (t, u, v)\in[1, T]_{\mathbb{Z}}\times\mathbb{R}^2\}. $由条件(H1) 知, 对任意$ n>0 $, 存在$ R^{\ast}>0 $, 使得当$ |u|+|v|\geqslant R^{\ast} $时, $ f(t, u, v)>n. $$ u(t) $$ s = s^{\ast} $时问题(1.4) 的可能解, 根据引理2.4可知, $ u_{L}<R_{\ast} $成立, 令$ \rho\geqslant|R_{\ast}|+2T^{2}\rho_{n}. $

$ \tilde{f}(t, u, v) $$ f(t, u, v) $的截断函数.

$ \begin{eqnarray*} \tilde{f}(t, u, v) = \left\{\begin{array}{lr} f(t, \rho, \rho), \; \; \; \; \; \; (t, u, v)\in[1, T]_{\mathbb{Z}}\times[\rho, +\infty)\times[\rho, +\infty), & \\ f(t, \rho, v), \; \; \; \; \; \; (t, u, v)\in[1, T]_{\mathbb{Z}}\times[\rho, +\infty)\times[-\rho, \rho], & \\ f(t, \rho, -\rho), \; \; \; \; (t, u, v)\in[1, T]_{\mathbb{Z}}\times[\rho, +\infty)\times(-\infty, -\rho], & \\ f(t, u, -\rho), \; \; \; \; (t, u, v)\in[1, T]_{\mathbb{Z}}\times[-\rho, \rho]\times(-\infty, -\rho], &\\ f(t, u, v), \; \; \; \; \; \; (t, u, v)\in[1, T]_{\mathbb{Z}}\times[-\rho, \rho]\times[-\rho, \rho], &\\ f(t, u, \rho), \; \; \; \; \; \; (t, u, v)\in[1, T]_{\mathbb{Z}}\times[-\rho, \rho]\times[\rho, +\infty), &\\ f(t, -\rho, -\rho), \; \; (t, u, v)\in[1, T]_{\mathbb{Z}}\times(-\infty, -\rho]\times(-\infty, -\rho], &\\ f(t, -\rho, v), \; \; \; \; (t, u, v)\in[1, T]_{\mathbb{Z}}\times(-\infty, -\rho]\times[-\rho, \rho], &\\ f(t, -\rho, \rho), \; \; \; \; (t, u, v)\in[1, T]_{\mathbb{Z}}\times(-\infty, -\rho]\times[\rho, +\infty), &\\ \end{array} \right. \end{eqnarray*} $

$ \tilde{f}(t, u, v) $为有界函数, 构造辅助问题

$ \begin{align} \left\{\begin{array}{ll} \Delta^{2}u(t-1)+\tilde{f}(t, u(t), \Delta u(t-1)) = s, \; \; \ \ \ \; t\in [1, T]_{\mathbb{Z}}, \\[2ex] u(0)-u(T) = \Delta u(0)-\Delta u(T) = 0. \end{array} \right. \end{align} $ (3.1)

$ K = [1, T]_{\mathbb{Z}}\times[-\rho, \rho]\times[-\rho, \rho] $为一闭区域, 由引理2.6知, 辅助问题(3.1) 存在解$ u_{1}(t) $, 满足$ \alpha(t)\leqslant u_{1}(t)\leqslant \beta(t) $, $ t\in[1, T]_{\mathbb{Z}} $, 显然$ u_{1}(t) $也是问题(1.4) 的解, 因此根据引理2.6知, $ s^{\ast}\in S_{1} $.

(b)   如果$ \tilde{s}\in S_{1} $, 且$ \tilde{s}<s $, 则$ s\in S_{1} $.

$ \tilde{u}(t) $$ s = \tilde{s} $时问题(3.1) 的解, 对任意给定的$ n $, 若$ s\in(\tilde{s}, n) $, 类似于(a) 中的截断技巧, $ \tilde{u}(t) $为此时问题(3.1) 的严格上解, 再取$ R_{1}<\tilde{u}_{L} $, 使得

$ \min\limits_{t\in[1, T]_{\mathbb{Z}}}f(t, R_{1}, 0)>s, $

$ \alpha\equiv R_{1}<0 $为问题(3.1) 的严格下解, 由引理2.6及$ n $的任意性可知, 当$ n\rightarrow \infty $时, 若$ s>\tilde{s} $, 则$ s\in S_{1} $.

(c)   $ s_{1} = \inf S_{1}<\infty $$ S_{1}\supset(s_{1}, \infty) $.

$ s\in\mathbb{R} $, 设$ u $为辅助问题(3.1) 的一个解, 则$ \|\Delta u(t-1)\|\leqslant\rho $, $ s = QN_{f}(u) $, 且有$ s\geqslant c $, 其中$ c = \inf\limits_{[1, T]_{\mathbb{Z}}\times\mathbb{R}\times[-\rho, \rho]}f(t, u, \Delta u(t-1)) $. 若$ s = c $, 则$ s_{1} = \inf S_{1}<\infty $, 运用(b) 的结论, 即得$ S_{1}\supset(s_{1}, \infty) $.

(d)   $ S_{2}\supset(s_{1}, \infty) $.

$ s_{3}<s_{1}<s_{2}<n $, 对于任意的$ s\in(-\infty, n) $, 令$ \Phi(s, \cdot) $为问题(3.1) 在$ X $的不动点算子, 根据引理2.4, 可以找到相应的$ \hat{\rho} $, 使得对于$ s\in[s_{3}, s_{2}] $时, $ I-\Phi(s, \cdot) $的任意可能零点$ u $均满足$ u\in B_{\hat{\rho}} $. 因此Leray-schauder度$ \deg[I-\Phi(s, \cdot), B_{\hat{\rho}}, 0] $有定义, 且不依赖于参数$ s $, 利用$ c $的结论, 对于$ u\in X $, $ u-\Phi(s_{3}, \cdot)\neq0 $, 说明$ \deg[I-\Phi(s_{3}, \cdot), B_{\hat{\rho}}, 0] = 0 $, 且$ \deg[I-\Phi(s_{2}, \cdot), B_{\hat{\rho}}, 0] = 0 $. 根据Leray-schauder度的切除性可知, 如果$ \rho'>\hat{\rho} $, 则$ \deg[I-\Phi(s_{2}, \cdot), B_{\rho'}, 0] = 0 $.

$ \hat{u} $$ s\in(s_{1}, s_{2}) $时问题(3.1) 的解, 则$ \hat{u} $$ s = s_{2} $时问题(3.1) 的严格上解, 取$ R<\hat{u}_{L} $满足$ \min\limits_{t\in [1, T]_{\mathbb{Z}}}f(t, R, 0)>s_{2} $, 则$ R $$ s = s_{2} $时问题(3.1) 的严格下解, 由引理2.6知, 当$ s = s_{2} $时, 问题(3.1) 在$ \Omega_{R, \hat{u}} $中有一个解, 且满足$ \deg[I-\Phi(s_{2}, \cdot), \Omega_{R, \hat{u}}, 0] = 1 $, 取$ \rho' $充分大, 利用Leray-schauder度的可加性可得

$ \deg[I-\Phi(s_{2}, \cdot), B_{\rho'}\backslash\Omega_{R, \hat{u}}, 0] = \deg[I-\Phi(s_{2}, \cdot), B_{\rho'}, 0]-\deg[I-\Phi(s_{2}, \cdot), \Omega_{R, \hat{u}}, 0] = -1, $

$ s = s_{2} $时, 问题3.1在$ B_{\rho'}\backslash\Omega_{R, \hat{u}} $中有第二个解, 当$ n\rightarrow \infty $时, $ S_{2}\supset(s_{1}, \infty) $.

(e)   $ s_{1}\in S_{1} $.

$ \{\eta_{k}\} $$ (s_{1}, n) $中收敛到$ s_{1} $的一个序列, 若$ u_{k} $$ s = \eta_{k} $时问题(3.1) 的解, 则$ u_{k} = \Phi(\eta_{k}, u_{k}) $, 由引理2.4知, 对任意的$ k\geqslant1 $, 存在$ \rho_{k}>0 $, 使得$ \|u_{k}\|<\rho_{k} $. 由$ \Phi $的紧性, $ u_{k} $收敛到当$ s = s_{1} $时问题(3.1) 的解$ u $. 由$ n $任意性, $ s_{1}\in S_{1} $, 若$ u $是辅助问题(3.1) 的解, 则$ u $一定是问题(1.4) 的解, 因此, 当$ f $满足条件(H1) 时, 存在$ s_{1}\in\mathbb{R} $, 使得当$ s<s_{1} $时, 问题(1.4) 无解; 当$ s = s_{1} $时, 问题(1.4) 至少有一个解; 当$ s>s_{1} $时, 问题(1.4) 至少有两个解.

4 应用

例1 考虑二阶离散周期边值问题

$ \begin{align} \left\{\begin{array}{ll} \Delta^{2}u(t-1)+u^{2}(t)+(\Delta u(t-1))^{2} = s, \; \; \ \ \ \; t\in [1, T]_{\mathbb{Z}}, \\[2ex] u(0)-u(T) = \Delta u(0)-\Delta u(T) = 0 \end{array} \right. \end{align} $ (4.1)

解的存在性及解的个数与参数的关系.

  这里取$ f(t, u(t), \Delta u(t-1)) = u^{2}(t)+(\Delta u(t-1))^{2} $, 因为当$ |u(t)|+|\Delta u(t-1)|\rightarrow \infty $时, $ u^{2}(t)+(\Delta u(t-1))^{2}\rightarrow +\infty $, 所以$ \lim\limits_{|u|+|v|\rightarrow \infty}f(t, u, v) = +\infty $对任意$ t\in[1, T]_{\mathbb{Z}} $一致成立. 又$ f:[1, T]_{\mathbb{Z}}\times\mathbb{R}^{2}\rightarrow \mathbb{R} $为连续函数, 则$ f $满足条件(H1).

根据定理1.1, 存在常数$ s_{1}\in\mathbb{R} $, 使得当$ s<s_{1} $时, 问题(4.1) 无解; 当$ s = s_{1} $时, 问题(4.1) 至少有一个解; 当$ s>s_{1} $时, 问题(4.1) 至少有两个解.

参考文献
[1] Ambrosetti A, Prodi G. On the inversion of some differentiable mappings with singularities between Banach spaces[J]. Annali Di Matematica Pura Ed Applicata, 1972, 93(1): 231–246. DOI:10.1007/BF02412022
[2] Fabry C, Mawhin J, Nkashama M N. A Multiplicity Result for Periodic Solutions of Forced Nonlinear Second order Ordinary Differential Equations[J]. Bulletin of the London Mathematical Society, 1986(2): 173–180.
[3] Chiappinelli R, Mawhin J, Nugari R. Generalized Ambrosetti-Prodi conditions for nonlinear two-point boundary value problems[J]. Journal of Differential Equations, 1987, 69(3): 422–434. DOI:10.1016/0022-0396(87)90127-6
[4] Paiva F, Montenegro M. An Ambrosetti-Prodi type result for a quasilinear Neumann problem[J]. Proceedings Edinburgh Mathematical Society, 2012, 55: 771–780. DOI:10.1017/S0013091512000041
[5] Mawhin J, Rebelo C, Zanolin F. Continuation theorems for Ambrosetti-Prodi type periodic problems[J]. Communications in Contemporary Mathematics, 2000, 2(1): 0000007.
[6] Feng C, Chen P. The existence of solutions for nonlinear two-point boundary problems with parameter under Ambrosetti-Prodi condition[J]. Dynamics of Continuous, Discrete and Impulsive Systems Series A: Mathematical Analysis, 2007, 14(4): 545–555.
[7] Ma L Y. The Ambrosetti-Prodi type results of the nonlinear second-order Neumann boundary value problem[J]. Journal of Shandong University(Natural Science), 2015.
[8] Mawhin J. First order ordinary differential equations with several periodic solutions[J]. Zeitschrift F ü r Angewandte Mathematik Und Physik Zamp, 1987, 38(2): 257–265. DOI:10.1007/BF00945410
[9] Manasevich R, Mawhin J. Boundary value problems for nonlinear perturbations of vector plaplacian-like operators[J]. Journal of the Korean Mathematical Society, 2000, 47(5): 47–11.
[10] Filho D C, Morais D, Pereira F R. Critical Ambrosetti-Prodi type problems for systems of elliptic equations[J]. Nonlinear Analysis, 2008, 68(1): 194–207. DOI:10.1016/j.na.2006.10.041