数学杂志  2019, Vol. 39 Issue (3): 431-439   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
宋海风
彭临平
一个三次等时中心在非光滑扰动下的极限环分支
宋海风1, 彭临平1,2    
1. 北京航空航天大学数学与系统科学学院, 北京 100191;
2. 北京航空航天大学数学、信息与行为教育部重点实验室, 北京 100191
摘要:本文研究了一个三次等时中心在非光滑扰动下的极限环分支问题.利用非光滑系统的一阶平均方法,获得了在任意小的分段三次多项式扰动下,从未扰动系统的周期环域中至多分支出7个极限环,而且此上界可以达到,推广了光滑扰动下的结果.
关键词三次等时中心    非光滑扰动    极限环    平均方法    
LIMIT CYCLES BIFURCATION FROM A CUBIC ISOCHRONOUS CENTER UNDER NON-SMOOTH PERTURBATIONS
SONG Hai-feng1, PENG Lin-ping1,2    
1. School of Mathematics and System Sciences, Beihang University, Beijing 100191, China;
2. Key Laboratory of Mathematics, Information and Behavior of the Ministry of Education, Beihang University, Beijing 100191, China
Abstract: This paper is devoted to study the bifurcation of limit cycles from a cubic isochronous center under any small non-smooth perturbations. By using the averaging theory for discontinuous differential systems, it proves that under any small piecewise cubic polynomial perturbations, at most seven limit cycles bifurcate from the period annulus sounding the center of the unperturbed system, and this upper bound can be reached, which extends the resultant under smooth perturbations.
Keywords: cubic isochronous center     non-smooth perturbations     limit cycles     averaging method    
1 引言

众所周知, 研究多项式微分系统的极限环个数是实平面微分系统定性理论的重要部分之一, 这类问题的主要研究方法有逆积分因子法, Abelian积分法及平均方法等.对于光滑系统的极限环个数, 已经得到了很多重要结论[1-10].

由于非光滑现象广泛存在于自然科学和工程技术等领域, 近年来非光滑系统越来越被重视, 特别是关于非光滑系统极限环分支的研究工作更是层出不穷, 其中有两个重要的工作要特别指出: Liu和Han研究了平面分段光滑的Hamilton系统的闭轨族的极限环分支问题, 推导证明了分段光滑系统的一阶Melnikov函数计算公式[11]; Llibre等介绍了不连续系统的平均理论, 给出了平均函数的表达式[12].又如Liang和Han利用Melnikov方法, 研究了擦边闭轨内外两侧闭轨族分支出的极限环的最大个数, 给出了广义同宿分支的条件及二重广义分支的条件[13]. Li和Liu研究了分段光滑二次微分系统的闭轨族的极限环分支问题, 借助于不连续微分系统的平均理论和复方法, 估计了所考虑的系统在$ n $次扰动下分支产生的极限环的最大个数[14]. Cen等利用一阶平均方法及Chebyshev准则, 研究了一类二次等时中心在分段光滑的二次多项式扰动下的极限环分支现象, 给出了分支产生的极限环的最大个数[15].

本文选取了如下三次可积非Hamilton系统

$ \begin{equation} \begin{split} \left\{ \begin{aligned} \dot{x}& = -y+x^{2}y, \\ \dot{y}& = x+xy^{2}. \end{aligned} \right. \end{split} \end{equation} $ (1.1)

显然, 系统(1.1)有首次积分

$ H(x, y) = \frac{x^{2}+y^{2}}{1-x^{2}} = h, $

相应的积分因子为$ \mu = \frac{2}{(1-x^{2})^{2}} $, $ (0, 0) $是系统(1.1)的等时中心, 以此中心及两条过赤道上退化奇点的不变线$ x = \pm1 $为边界构成该系统的周期环域, 即

$ \{(x, y)|H(x, y) = h, \quad h\in(0, +\infty)\}. $

系统(1.1)的相图如图 1所示.

图 1 系统(1.1)的相图

利用非光滑系统的一阶平均方法, 研究系统(1.1)的任意分段三次多项式小扰动即

$ \begin{equation} \begin{split} \left(\begin{aligned}&\dot{x}\\&\dot{y}\end{aligned}\right)& = \left\{\begin{aligned} \left(\begin{aligned} -y+x^{2}y+\varepsilon P_{1}(x, y)\\ x+xy^{2}+\varepsilon Q_{1}(x, y)\end{aligned}\right), \quad x>0, \\ \left(\begin{aligned} -y+x^{2}y+\varepsilon P_{2}(x, y)\\ x+xy^{2}+\varepsilon Q_{2}(x, y)\end{aligned}\right), \quad x<0 \end{aligned} \right. \end{split} \end{equation} $ (1.2)

的极限环分支现象, 其中$ 0<|\varepsilon|\ll1 $, $ P_{i}(x, y), Q_{i}(x, y)\; (i = 1, 2) $是关于变量$ x $$ y $的三次多项式, 表达式如下

$ \begin{equation} \begin{split} &P_{1}(x, y) = \sum\limits_{i+j = 0}^{3}a_{ij}x^{i}y^{j}, \quad Q_{1}(x, y) = \sum\limits_{i+j = 0}^{3}b_{ij}x^{i}y^{j}, \\ &P_{2}(x, y) = \sum\limits_{i+j = 0}^{3}c_{ij}x^{i}y^{j}, \quad Q_{2}(x, y) = \sum\limits_{i+j = 0}^{3}d_{ij}x^{i}y^{j}.\nonumber \end{split} \end{equation} $

根据系统特点, 借助于一些数学技巧, 证明了下述定理.

定理1.1   对于三次等时中心(1.1), 在任意小的分段三次多项式扰动下, 从未扰系统的周期环域中至多分支出7个极限环, 而且此上界是可以达到的.

  Li和Zhao已在文献[1]中研究了系统(1.1)在光滑扰动情况下的极限环分支问题, 得到的结论是最多可以从未扰动系统的周期环域中分支出3个极限环.与得到的结论相比可知, 非光滑扰动比光滑扰动能分支出更多的极限环.

本文的结构安排如下:第二部分简要介绍非光滑系统的一阶平均理论;第三部分计算系统(1.2)的一阶平均函数;第四部分证明定理$ 1.1 $.

2 非光滑系统的平均理论

在定理1.1的证明过程中将会用到一阶平均方法, 所谓的平均方法即是将极限环的个数问题转化为研究一阶平均函数的简单零点个数.下面首先对平均理论做一个简单介绍, 详细论述见文献[12, 16, 17].

引理2.1 [12]  考虑如下非光滑微分系统

$ \begin{equation} \frac{dr}{d\theta} = \varepsilon F(\theta, r)+\varepsilon^{2}R(\theta, r, \varepsilon), \end{equation} $ (2.1)

其中

$ \begin{equation} \begin{split} &F(\theta, r) = F_{1}(\theta, r)+{\rm sign}(h(\theta, r))F_{2}(\theta, r), \\ &R(\theta, r, \varepsilon) = R_{1}(\theta, r, \varepsilon)+{\rm sign}(h(\theta, r))R_{2}(\theta, r, \varepsilon), \nonumber \end{split} \end{equation} $

$ F_{1}, F_{2}:\mathbb{R}\times D \rightarrow \mathbb{R}, R_{1}, R_{2}:\mathbb{R}\times D\times (-\varepsilon_{0}, \varepsilon_{0}) \rightarrow \mathbb{R}, h:\mathbb{R}\times D \rightarrow \mathbb{R} $都是连续函数, 且这些函数均是关于变量$ \theta $以T为周期的周期函数, $ D $$ \mathbb{R} $中的开子集, $ {\rm sign}(h) $是如下定义的符号函数

$ \begin{equation} \begin{split} {\rm sign}(h) = \left\{ \begin{aligned} &1, \quad h>0, \\ &0, \quad h = 0, \\ &-1, \quad h<0. \end{aligned} \right. \end{split} \nonumber \end{equation} $

假定$ h(\theta, r) $$ C^{1} $函数且$ 0 $为其正则值, 记$ \mathcal{M} = h^{-1}(0) $, $ \Sigma = \{0\} \times D\nsubseteq \mathcal{M} $, $ \Sigma_{0} = \Sigma\setminus\mathcal{M}\neq\varnothing $, 和$ z\equiv(0, z)\notin \mathcal{M} $.定义系统(2.1)的一阶平均函数$ f:D\rightarrow \mathbb{R} $如下

$ \begin{equation} f(r) = \int_{0}^{T}F(\theta, r)d\theta. \end{equation} $ (2.2)

假定系统(2.1)满足下面(ⅰ), (ⅱ), (ⅲ)三个条件.

(ⅰ) $ F_{1}, F_{2}, R_{1}, R_{2} $$ h $关于$ r $满足局部Lipschitz条件;

(ⅱ) 对于$ a\in \Sigma_{0} $, 存在$ a $的一个邻域$ V $, 使得对任意的$ z\in \bar{V} \backslash \{a\} $, 有$ f(z)\neq0 $及Brouwer度函数$ d_{B}(f, V, a)\neq0 $;

(ⅲ) 如果$ \frac{\partial h}{\partial \theta}\neq0 $, 那么对所有的$ (\theta, r)\in \mathcal{M} $, 有$ \frac{\partial h}{\partial \theta}(\theta, r)\neq0 $.如果$ \frac{\partial h}{\partial \theta}\equiv0 $, 则对所有的$ (\theta, z) \in [0, T]\times \mathcal{M} $, 有$ \langle\nabla_{r}h, F_{1}\rangle^{2}-\langle\nabla_{r}h, F_{2}\rangle^{2}>0 $, 其中$ \nabla_{r}h $是函数$ h $关于变量$ r $的梯度.则当$ |\varepsilon|>0 $充分小时, 系统(2.1)存在一个周期为$ T $的解$ r(\theta, \varepsilon) $, 使得当$ \varepsilon\rightarrow0 $时, $ r(0, \varepsilon)\rightarrow a $ (在Hausdorff度量意义下).

为了更方便地验证引理$ 2.1 $中的条件(ⅱ), 本文中我们用文献 [16]中一个充分条件$ \tilde{\rm (ii)} $来代替条件(ⅱ).

注2.1   $ \tilde{\rm (ii)} $假定$ f:D \rightarrow \mathbb{R} $$ C^{1} $函数且$ f(a) = 0 $, 其中$ D $$ \mathbb{R} $中的开子集且$ a\in D $.如果雅克比行列式$ J_{f}(a)\neq0 $时, 存在$ a $的一个邻域$ V $使得对所有的$ r\in \bar{V}\setminus\{a\} $, 有$ f(r)\neq0 $.则$ d_{B}(f, V, 0)\neq0 $.

考虑如下形式的平面微分系统

$ \begin{equation} \begin{split} \left\{ \begin{aligned} \dot{x}& = P(x, y)+\varepsilon p(x, y), \\ \dot{y}& = Q(x, y)+\varepsilon q(x, y), \end{aligned} \right. \end{split} \end{equation} $ (2.3)

其中$ P(x, y), Q(x, y), p(x, y), q(x, y):\mathbb{R}^{2}\rightarrow \mathbb{R} $是连续函数, $ \varepsilon $是小参数.假设系统$ (2.3)|_{\varepsilon = 0} $围绕中心$ (0, 0) $的周期环域

$ \{\gamma_{h}\}\subset\{(x, y)| H(x, y) = h, \quad h\in(h_{c}, h_{s})\}, $

其中$ H(x, y) = h $是系统$ (2.3)|_{\varepsilon = 0} $的一个首次积分, $ h_{c} $$ h_{s} $分别对应中心和分界线处的Hamilton函数值.

引理2.2 [16]  考虑系统$ (2.3)|_{\varepsilon = 0} $及其首次积分$ H(x, y) = h $.假设对于周期环域中的所有$ (x, y) $, 都有$ xQ(x, y)-yP(x, y)\neq0 $成立.设$ \rho:(\sqrt{h_{c}}, \sqrt{h_{s}})\times[0, 2\pi)\rightarrow[0, +\infty) $是一个连续函数, 并且对所有的$ r\in(\sqrt{h_{c}}, \sqrt{h_{s}}) $$ \theta\in [0, 2\pi) $都有

$ \begin{equation} H\left(\rho(r, \theta)\cos\theta, \rho(r, \theta)\sin\theta\right) = r^{2}, \nonumber \end{equation} $

则对于系统(2.3), 描述$ h $的平方根$ r = \sqrt{h} $和角度$ \theta $之间相关性的微分方程为

$ \begin{equation} \frac{dr}{d\theta} = \varepsilon\frac{\mu(x^{2}+y^{2})(Qp-Pq)}{2r(Qx-Py)+2r\varepsilon(qx-py)}, \nonumber \end{equation} $

其中$ \mu = \mu(x, y) $是系统$ (2.3)|_{\varepsilon = 0} $相应于首次积分$ H(x, y) = h $的积分因子, $ x = \rho(r, \theta)\cos\theta $, $ y = \rho(r, \theta)\sin\theta $. $ P, Q, p $$ q $同前.

3 平均函数的计算

对于系统(1.1)的Hamilton函数

$ H(x, y) = \frac{x^{2}+y^{2}}{1-x^{2}}, $

选择如下的函数

$ \begin{equation} \rho(r, \theta) = \frac{r}{\sqrt{1+r^{2}\cos^{2}{\theta}}}, \nonumber \end{equation} $

使得

$ H(\rho(r, \theta)\cos\theta, \rho(r, \theta)\sin\theta) = r^{2}, \quad r\in(0, +\infty). $

对系统(1.2)做变换

$ \begin{equation} x = \frac{r\cos\theta}{\sqrt{1+r^{2}\cos^{2}{\theta}}}, \quad y = \frac{r\sin\theta}{\sqrt{1+r^{2}\cos^{2}{\theta}}}, \end{equation} $ (3.1)

则有

$ \begin{equation} \begin{split} \frac{dr}{d\theta} = &\left\{\begin{aligned} \varepsilon X_{1}(\theta, r)+\varepsilon^{2}Y_{1}(\theta, r, \varepsilon), \quad \cos\theta>0, \\ \varepsilon X_{2}(\theta, r)+\varepsilon^{2}Y_{2}(\theta, r, \varepsilon), \quad \cos\theta<0, \end{aligned} \right. \end{split} \end{equation} $ (3.2)

其中

$ \begin{eqnarray*} &&X_{i}(\theta, r) = \sqrt{1+r^{2}\cos^{2}\theta} \cos\theta(1+r^{2})P_{i}(\theta, r)+\sqrt{1+r^{2}\cos^{2}\theta} \sin\theta Q_{i}(\theta, r), \\ &&Y_{i}(\theta, r, \varepsilon) = -X_{i}(\theta, r)\frac{\sqrt{1+r^{2}\cos^{2}\theta}(\cos\theta Q_{i}(\theta, r) -\sin\theta P_{i}(\theta, r))}{r+\varepsilon\sqrt{1+r^{2}\cos^{2}\theta}(\cos\theta Q_{i}(\theta, r)-\sin\theta P_{i}(\theta, r))}, \end{eqnarray*} $

这里$ P_{i}(\theta, r) $$ Q_{i}(\theta, r) $是(1.2)式中的$ P_{i}(x, y), Q_{i}(x, y)\; (i = 1, 2) $通过(3.1)式变换得到的.

$ \begin{equation} \begin{split} F_{i}(\theta, r)& = \frac{1}{2}\Big(X_{1}(\theta, r)-(-1)^{i}X_{2}(\theta, r)\Big), \\ R_{i}(\theta, r, \varepsilon)& = \frac{1}{2}\Big(Y_{1}(\theta, r, \varepsilon)-(-1)^{i}Y_{2}(\theta, r, \varepsilon)\Big), \quad i = 1, 2. \end{split} \nonumber \end{equation} $

则系统(3.2)可以写成以下标准形式

$ \begin{equation} \frac{dr}{d\theta} = \varepsilon F(\theta, r)+\varepsilon^{2}R(\theta, r, \varepsilon), \end{equation} $ (3.3)

其中

$ \begin{equation} \begin{split} F(\theta, r)& = F_{1}(\theta, r)+{\rm sign}(\cos\theta)F_{2}(\theta, r), \\ R(\theta, r, \varepsilon)& = R_{1}(\theta, r, \varepsilon)+{\rm sign}(\cos\theta)R_{2}(\theta, r, \varepsilon). \end{split} \nonumber \end{equation} $

易验证方程(3.3)满足引理$ 2.1 $中的三个条件, 其一阶平均函数为

$ \begin{equation} \begin{split} f(r) = \int_{0}^{2\pi}F(\theta, r)d\theta = \int_{-\frac{\pi}{2}}^{\frac{\pi}{2}}X_{1}(\theta, r)d\theta+\int_{\frac{\pi}{2}}^{\frac{3\pi}{2}}X_{2}(\theta, r)d\theta, \end{split} \end{equation} $ (3.4)

对上式中第二个积分做变量替换$ \theta\rightarrow\pi-\theta $, 则(3.4)式转化为

$ \begin{equation} f(r) = \int_{-\frac{\pi}{2}}^{\frac{\pi}{2}}\left[\sqrt{1+r^{2}\cos^{2}\theta}\cos\theta(1+r^{2})P(\theta, r)+\sqrt{1+r^{2} \cos^{2}\theta}\sin\theta Q(\theta, r)\right]d\theta, \end{equation} $ (3.5)

其中

$ \begin{eqnarray*} &&P(\theta, r) = \sum\limits_{i+j = 0}^{3}(a_{ij}-(-1)^{i}c_{ij})\left(\frac{r\cos\theta}{\sqrt{1+r^{2}\cos^{2}\theta}}\right)^{i} \left(\frac{r\sin\theta}{\sqrt{1+r^{2}\cos^{2}\theta}}\right)^{j}, \\ &&Q(\theta, r) = \sum\limits_{i+j = 0}^{3}(b_{ij}+(-1)^{i}d_{ij})\left(\frac{r\cos\theta}{\sqrt{1+r^{2}\cos^{2}\theta}}\right)^{i} \left(\frac{r\sin\theta}{\sqrt{1+r^{2}\cos^{2}\theta}}\right)^{j}. \end{eqnarray*} $

直接计算可得下列结论.

引理3.1   以下积分等式成立

$ \begin{equation} \begin{split} &\int_{-\frac{\pi}{2}}^{\frac{\pi}{2}}\cos\theta\sqrt{1+r^{2}\cos^{2}\theta}d\theta = 1+\frac{1+r^{2}}{r}\arctan(r), \\ &\int_{-\frac{\pi}{2}}^{\frac{\pi}{2}}\frac{\cos\theta}{\sqrt{1+r^{2}\cos^{2}\theta}}d\theta = \frac{2}{r}\arctan(r), \\ &\int_{-\frac{\pi}{2}}^{\frac{\pi}{2}}\frac{\cos^{3}\theta}{\sqrt{1+r^{2}\cos^{2}\theta}}d\theta = \frac{1}{r^{2}}+\frac{r^{2}-1}{r^{3}}\arctan(r), \\ &\int_{-\frac{\pi}{2}}^{\frac{\pi}{2}}\frac{1}{1+r^{2}\cos^{2}\theta}d\theta = \frac{\pi}{\sqrt{1+r^{2}}}, \\ &\int_{-\frac{\pi}{2}}^{\frac{\pi}{2}}\frac{\cos^{2}\theta}{1+r^{2}\cos^{2}\theta}d\theta = \frac{\pi}{r^{2}}-\frac{\pi}{r^{2}\sqrt{1+r^{2}}}, \\ &\int_{-\frac{\pi}{2}}^{\frac{\pi}{2}}\frac{\cos^{4}\theta}{1+r^{2}\cos^{2}\theta}d\theta = \frac{\pi}{2r^{2}}-\frac{\pi}{r^{4}}+\frac{\pi}{r^{4} \sqrt{1+r^{2}}}.\nonumber \end{split} \end{equation} $

将引理$ 3.1 $代入(3.5)式整理得

$ \begin{equation} f(r) = k_{1}f_{1}(r)+k_{2}f_{2}(r)+k_{3}f_{3}(r)+k_{4}f_{4}(r)+k_{5}f_{5}(r)+k_{6}f_{6}(r)+k_{7}f_{7}(r)+k_{8}f_{8}(r), \end{equation} $ (3.6)

其中

$ \begin{equation} \begin{split} &f_{1}(r) = 1, \; \; \; f_{2}(r) = r, \; \; \; f_{3}(r) = r^{2}\; \; \; f_{4}(r) = r^{3}, \; \; \; f_{5}(r) = r\sqrt{1+r^{2}}, \\ &f_{6}(r) = \frac{\sqrt{1+r^{2}}-1}{r}, \; \; f_{7}(r) = \frac{r^{2}+1}{r}\arctan(r), \; \; f_{8}(r) = \frac{r^{4}-1}{r}\arctan(r), \nonumber \end{split} \end{equation} $

$ \begin{equation} \begin{split} &k_{1} = (a_{00}-c_{00})+(a_{20}-c_{20})-(a_{02}-c_{02})-(b_{11}-d_{11}), \\ &\begin{split}k_{2} = &\frac{\pi}{2}[(a_{10}+c_{10})-(a_{30}+c_{30})+3(a_{12}+c_{12})+(b_{01}+d_{01})\\ &-3(b_{03}+d_{03})+(b_{21}+d_{21})], \end{split}\\ &k_{3} = (a_{00}-c_{00})+(a_{20}-c_{20})-(a_{02}-c_{02}), \\ &k_{4} = \frac{\pi}{2}[(a_{10}+c_{10})+(a_{30}+c_{30})+(a_{12}+c_{12})], \\ &k_{5} = \pi[-(a_{12}+c_{12})+(b_{03}+d_{03})], \\ &k_{6} = \pi[(a_{30}+c_{30})-(a_{12}+c_{12})+(b_{03}+d_{03})-(b_{21}+d_{21})], \\ &k_{7} = 2(a_{00}-c_{00})+2(a_{02}-c_{02})+(b_{11}-d_{11}), \\ &k_{8} = (a_{00}-c_{00})+(a_{20}-c_{20})+(a_{02}-c_{02}).\nonumber \end{split} \end{equation} $

命题3.1   式(3.6)中的$ 8 $个函数$ f_{i}:(0, +\infty)\rightarrow \mathbb{R}\; (i = 1, 2, 3, 4, 5, 6, 7, 8) $是线性无关的且系数$ k_{1}, k_{2}, \cdots, k_{8} $关于扰动参数是相互独立的.

  易知函数$ f_{i}:(0, +\infty)\rightarrow \mathbb{R} (i = 1, 2, 3, 4, 5, 6, 7, 8) $$ r = 0 $处的泰勒展开式分别为

$ \begin{equation} \begin{split} &f_{1}(r) = 1, \\ &f_{2}(r) = r, \\ &f_{3}(r) = r^{2}, \\ &f_{4}(r) = r^{3}, \\ &f_{5}(r) = r+\frac{1}{2}r^{3}-\frac{1}{8}r^{5}+\frac{1}{16}r^{7}+\mathcal{O}(r^{8}), \\ &f_{6}(r) = \frac{1}{2}r-\frac{1}{8}r^{3}+\frac{1}{16}r^{5}-\frac{5}{128}r^{7}+\mathcal{O}(r^{8}), \\ &f_{7}(r) = 1+\frac{2}{3}r^{2}-\frac{2}{15}r^{4}+\frac{2}{35}r^{6}+\mathcal{O}(r^{8}), \\ &f_{8}(r) = -1+\frac{1}{3}r^{2}+\frac{4}{5}r^{4}-\frac{4}{21}r^{6}+\mathcal{O}(r^{8}).\nonumber \end{split} \end{equation} $

通过计算可以得到

$ \frac{\partial (f_{1}(r), f_{2}(r), f_{3}(r), f_{4}(r), f_{5}(r), f_{6}(r), f_{7}(r), f_{8}(r))}{\partial (1, r, r^{2}, r^{3}, r^{4}, r^{5}, r^{6}, r^{7})}\bigg|_{r = 0} = \frac{1}{50400}\neq 0, $

所以这些函数是线性无关的.进一步有

$ \begin{equation} \frac{\partial(k_{1}, k_{2}, k_{3}, k_{4}, k_{5}, k_{6}, k_{7}, k_{8})}{\partial(a_{00}, a_{10}, a_{20}, a_{02}, b_{01}, b_{11}, b_{03}, b_{21})} = \pi^{4}\neq0, \nonumber \end{equation} $

由此可知系数$ k_{1}, k_{2}, \cdots, k_{8} $关于扰动参数是相互独立的.

4 主要结果的证明

这一部分主要研究$ f(r) $的零点个数, 为此介绍一个重要的引理.

引理4.1 [2]  考虑$ n $个线性无关的解析函数$ h_{i}(x):D\rightarrow\mathbb{R}, i = 1, 2, \cdots, n $, 其中$ D\subset\mathbb{R} $是一个区间.假设存在$ k\in \{1, 2, \cdots, n\} $使得$ h_{k}(x) $$ D $上不变号, 则一定存在$ n $个常数$ c_{i}\; (i = 1, 2, \cdots, n) $使得$ \sum\limits_{i = 1}^{n}c_{i}h_{i}(x) $$ D $上至少有$ n-1 $个简单零点.

由此引理及命题3.1得

命题4.1   函数$ f(r) $在区间$ (0, +\infty) $上零点个数的最大值至少为$ 7 $.

定理1.1的证明   为了估计$ f(r) $$ r\in(0, +\infty) $上零点个数的最小上界, 令$ F(r) = rf(r). $显然$ F(r) $$ f(r) $$ r\in(0, +\infty) $上有完全相同的零点个数.进一步有

$ \begin{eqnarray*} F(r)& = &-k_{6}+k_{1}r+k_{2}r^{2}+k_{3}r^{3}+k_{4}r^{4}\\ &&+(k_{5}r^{2}+k_{6})\sqrt{1+r^{2}}+(k_{7}-k_{8}+k_{7}r^{2}+k_{8}r^{4})\arctan(r). \end{eqnarray*} $

$ F(r) $求五阶导得

$ \begin{equation} F^{(5)}(r) = -\frac{\sqrt{1+r^{2}}A(r)+(60k_{6}-15k_{5})r^{3}+(90k_{5}-45k_{6})r}{(1+r^{2})^{\frac{9}{2}}}, \end{equation} $ (4.1)

其中$ A(r) = (96k_{8}-80k_{7})r^{2}-96k_{8}+16k_{7}. $因为$ \sqrt{1+r^{2}}>1 $, 所以令

$ \sqrt{1+r^{2}} = \frac{1+t^{2}}{1-t^{2}}, \qquad t\in(0, 1), $

那么

$ r = \frac{2t}{1-t^{2}}. $

则函数(4.1)变为

$ \begin{equation} F^{(5)}(r) = g(t) = \frac{2(-1+t^{2})^{6}}{(1+t^{2})^{9}}B(t), \nonumber \end{equation} $

其中

$ \begin{eqnarray*} B(t)& = &(48k_{8}-8k_{7})t^{6}+(45k_{6}-90k_{5})t^{5}+(168k_{7}-240k_{8})t^{4}\\ &&+(240k_{5}-330k_{6})t^{3} +(168k_{7}-240k_{8})t^{2}\\ &&+(45k_{6}-90k_{5})t+(48k_{8}-8k_{7}). \end{eqnarray*} $

由于多项式$ B(t) $的系数是对称的, 若$ t_{0}\neq 0 $是其零点的话, 则$ 1/t_{0} $也为其零点, 所以$ g(t) $$ (0, 1) $上至多有3个零点, 即$ F^{(5)}(r) $$ (0, +\infty) $上至多有3个零点, 从而$ F(r) $$ [0, +\infty) $上至多有8个零点.由于$ F(0)\equiv0 $, 所以$ F(r) $$ (0, +\infty) $上至多有7个零点, 由其定义知函数$ f(r) $$ (0, +\infty) $至多有7个零点.再根据命题$ 4.1 $$ f(r) $零点个数的最大值为7.

根据引理2.1, 系统(1.2)至多有$ 7 $个极限环从未扰动系统的周期环域中分支出来, 而且最大个数$ 7 $是可以达到的.定理$ 1.1 $得证.

参考文献
[1] Li S M, Zhao Y L. Limit cycles of perturbed cubic isochronous center via the second order averaging method[J]. Intern. J. Bifur. Chaos, 2014, 24(3): 1450035. DOI:10.1142/S0218127414500357
[2] Coll B, Gasull A, Prohens R. Bifurcation of limit cycles from two family of centers[J]. Dyn. Contin. Discrete Impuls. Syst., 2005, 12(2): 275–287.
[3] Han M A, Li J B. Lower bounds for the Hilbert number of polynomial systems[J]. J. Diff. Equ., 2012, 252(4): 3278–3304. DOI:10.1016/j.jde.2011.11.024
[4] Han M A, Yu P. Normal forms, Melnikov functions and bifurcations of limit cycles[M]. New York: Springer, 2012.
[5] Iliev I D. Perturbations of quadratic centers[J]. Bull. Sci. Math., 1998, 122(2): 107–161. DOI:10.1016/S0007-4497(98)80080-8
[6] Chen F D, Li C Z, Llibre J, Zhang Z H. A unified proof on the weak Hilbert 16th problem for n=2[J]. J. Diff. Equ., 2006, 221(2): 309–342. DOI:10.1016/j.jde.2005.01.009
[7] Wang J H, Xiao D M. On the number of limit cycles in small perturbations of a class of hyper-elliptic Hamiltonian systems with one nilpotent saddle[J]. J. Diff. Equ., 2011, 250(4): 2227–2243. DOI:10.1016/j.jde.2010.11.004
[8] Zoladek H. Quadratic systems with center and their perturbations[J]. J. Diff. Equ., 1994, 109(2): 223–273. DOI:10.1006/jdeq.1994.1049
[9] Zhang P G. On the distribution and number of limit cycles for quadratic systems with two foci[J]. Qual. The. Dyn. Syst., 2002, 3(2): 437–463. DOI:10.1007/BF02969414
[10] Zhang Z F, Li C Z. On the number of limit cycles of quadratic Hamiltonian systems under quadratic perturbations[J]. Adv. Math., 1997, 26(5): 445–460.
[11] Liu X, Han M A. Bifurcation of limit cycles by perturbing piecewise Hamiltonian systems[J]. Intern. J. Bifur. Chaos, 2010, 20(5): 1–12.
[12] Llibre J, Novaes D D, Teixeira M A. On the birth of limit cycles for non-smooth dynamical systems[J]. Bull. Sci. Math., 2015, 139(3): 229–244. DOI:10.1016/j.bulsci.2014.08.011
[13] Liang F, Han M A. Limit cycles near generalized homoclinic and double homoclinic loops in piecewise smooth systems[J]. Chaos, Soli. Frac., 2012, 45(4): 454–464. DOI:10.1016/j.chaos.2011.09.013
[14] Li S M, Liu C J. A linear estimate of the number of limit cycles for some planar piecewise smooth quadratic differential systems[J]. J. Math. Anal. Appl., 2015, 428(2): 1354–1367. DOI:10.1016/j.jmaa.2015.03.074
[15] Cen X L, Li S M, Zhao Y L. On the number of limit cycles for a class of discontinuous quadratic differential systems[J]. Math. Anal. Appl., 2017, 449(1): 314–342. DOI:10.1016/j.jmaa.2016.11.033
[16] Buiča A, Libre J. Averaging methods for finding periodic orbits via brouwer degree[J]. Bull. Sci. Math., 2004, 128(1): 7–22. DOI:10.1016/j.bulsci.2003.09.002
[17] Sanders J A, Verbulst F. Averaging methods in nonlinear dynamic systems[M]. New York: SpringVerlag, 1985.