1: %\pagestyle{myheadings}
2: %\markboth{\textit{My Paper}\qquad \qquad \qquad \rm Vol. 23,
3: %No.1}{{\footnotesize \textit{Li Lian-gang, et al.}} \footnotesize
4: %\rm The Analysis of Vector Field for the pendulum Equation}
5:
6:
7: \documentclass[10pt,twoside]{article}
8: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
9: \usepackage{amsfonts}
10: \usepackage{fancyhdr}
11: \usepackage{titlesec}
12: \usepackage{cite}
13: \usepackage{ifthen}
14: \usepackage{graphicx}
15:
16: %TCIDATA{OutputFilter=Latex.dll}
17: %TCIDATA{Version=5.50.0.2890}
18: %TCIDATA{<META NAME="SaveForMode" CONTENT="1">}
19: %TCIDATA{BibliographyScheme=Manual}
20: %TCIDATA{LastRevised=Monday, July 21, 2008 22:15:30}
21: %TCIDATA{<META NAME="GraphicsSave" CONTENT="32">}
22:
23: \DeclareGraphicsRule{.jpg}{eps}{.bb}{}
24: \newcommand\upcite[1]{\textsuperscript{\cite{#1}}}
25: \titleformat{\section}{\centering\large\bfseries}{\S\arabic{section}}{1em}{}
26: \newboolean{first}
27: \setboolean{first}{true}
28: \renewcommand{\headrulewidth}{0pt}
29: \textheight 205mm \textwidth 145mm \setlength{\oddsidemargin}{5.6mm}
30: \setlength{\evensidemargin}{5.6mm}
31: \renewcommand\baselinestretch{1.2}
32: \newcommand{\hs}{\hskip 0.3cm}
33: \newcommand{\sign}{\mbox{sign}}
34: \newcommand{\mod}{\mbox{mod}}
35: \newcommand{\Discr}{\mbox{Discr}}
36: \newcommand{\transrow}[1]{\stackrel{#1}{\longrightarrow}}
37: \newtheorem{theorem}{Theorem}
38: \newtheorem{lemma}{Lemma}
39: \def\thefootnote{}
40:
41:
42: \begin{document}
43:
44: \title{\textbf{\Large The analysis of rotated vector field on the pendulum
45: equation}}
46: \author{Lian-Gang Li \\
47: %EndAName
48: {\small \emph{Department of Physics, School of Sciences, Tianjin
49: University, Tianjin, 300072, China}}\date{} } \maketitle
50:
51: \setlength \abovedisplayskip{2pt}
52: \setlength \abovedisplayshortskip{0pt}
53:
54: \footnote{%
55: corresponding author.\newline
56: Ph. Doctor, \emph{e-mail}: liliank@tju.edu.cn}
57:
58: \begin{center}
59: \begin{minipage}{135mm}
60: {\bf \small Abstract}.\hskip 2mm {\small The driven, damped pendulum
61: equation is a mathematical model of pendulum. It is a nonlinear
62: differential equation which is non-integrable. By the method of
63: rotated vector field, this paper obtains the relation between the
64: external drive $\beta$ and the periodic solution. An other
65: conclusion is that the critical value of $\beta$ remains fixed in
66: the over damping situation. These results is very useful in the
67: study of charge-density wave in physics. \vskip 5mm \bf \small
68: Keyword}:\hskip 2mm {\small pendulum, rotated vector fields,
69: limit-cycle, charge-density wave (CDW)}
70: \end{minipage}
71: \end{center}
72:
73: %\thispagestyle{fancyplain} \fancyhead{}
74: %\fancyhead[L]{\textit{My paper}\\
75: %2008, 23(1): ***-***} \fancyfoot{}
76: \vskip 6mm
77:
78: \setcounter{section}{1} \setcounter{equation}{0}
79: \renewcommand{\theequation}
80: {\arabic{equation}} \setcounter{figure}{0}\renewcommand{\thefigure}{%
81: \arabic{figure}} %%
82:
83: \begin{flushleft}
84: {\large \textbf{1.\hspace*{2mm}Introduction\newline
85: }}
86: \end{flushleft}
87:
88: The driven,damped pendulum model is an essential model of nonlinear
89: sciences, which displays many manifestations including chaos and appears in
90: many physical subjects such as the Jesophson junctions and the
91: charge-density waves\cite{Ott,Bohr,Gruner}. This paper is involves the
92: damped pendulum subject to an external drive, for which the equation of
93: motion is%
94: \begin{equation}
95: \frac{\mathrm{d}^{2}\phi }{\mathrm{d}t^{2}}+\gamma \frac{\mathrm{d}\phi }{%
96: \mathrm{d}t}+\sin \phi =\beta , \label{eq:f1}
97: \end{equation}%
98: where $\beta $ is the external drive, $\gamma $ is the friction coefficient.
99: Equation (\ref{eq:f1}) is also used in the theory of charge-density waves
100: (CDW)\cite{Gruner}. The relation between $\beta $ and its solution is
101: important to describe the properties of conductivity associated with CDW\cite%
102: {Li}. This differential equation is nonlinear and non-integrable. It is the
103: purpose of this paper to obtain the relation between $\beta $ and its
104: solutions by the method of rotated vector fields.
105:
106: In 1881-1886, Poincar\'{e} originated a theory named "vector field" in his
107: papers\cite{Poinc}. With this approach, the solutions of differential
108: equation can be regarded as integral curves in the phase space. The
109: qualitative properties of solutions can be obtained geometrically. After
110: that, many scientists have studied in this area of theory. In 1953, G. F.
111: Duff proposed the rotated vector fields\cite{Duff}, then G. Seifert \emph{et
112: al}. developed it to the general rotated vector fields\cite{Seifert,Chen,Ma}.
113:
114: By the method of rotated vector field, this paper provides the proofs about
115: the relation between $\beta$ and the periodic solutions of the equation. By
116: analysis, it is found that if the $\gamma$ is bigger enough (over damping),
117: the critical value of $\beta$, with which this equation will have a critical
118: periodic solution, remains fixed all the time, namely, $\beta_{0}\equiv1$.
119: This explicit conclusion is very useful for the paper\cite{Li}.
120:
121: %\setcounter{section}{1}
122: %\setcounter{equation}{0}
123: %\renewcommand{\theequation}
124: %{\arabic{equation}}
125:
126: \begin{flushleft}
127: {\large \textbf{2.\hspace*{2mm}Preliminary definitions and lemmas\newline
128: }}
129: \end{flushleft}
130:
131: First of all, some statements are introduced in preparation for the proofs.
132: On the $\phi -z$ phase plane, Equation (\ref{eq:f1}) takes the form of
133: vector field:%
134: \begin{equation}
135: \left\{
136: \begin{array}{ll}
137: \frac{\mathrm{d}\phi }{\mathrm{d}t}=z & \\
138: \frac{\mathrm{d}z}{\mathrm{d}t}=\beta -\sin \phi -\gamma z &
139: \end{array}%
140: \right. , \label{eq:f2}
141: \end{equation}%
142: where $\gamma >0$, $\beta \geq 0$, whose solutions correspond to-
143: trajectories on the plane. Each trajectory, has a direction running as time
144: goes forward, to which the tangents become vectors which constitute the
145: vector field. The points satisfying $\frac{d\phi }{dt}=0$ and $\frac{dz}{dt}%
146: =0$ is the singularities of Equation (\ref{eq:f2}). When the vectors are
147: rotating owing to the change of some parameter of equation, they constitute
148: the rotated vector field, and accompanying with the moving of singularities
149: as well, the field is named as the general rotated vector field\cite%
150: {Duff,Seifert,Chen,Ma}. According to $\frac{d\phi }{dt}=z$ indicates that,
151: the trajectories direct from left to right while $z>0$, and they direct from
152: right to left while $z<0$. In the theory of differential equation, owing to
153: the uniqueness of the solution, the trajectories don't intersect each other
154: except at singularities\cite{Perko}. This is very useful property in the
155: following analysis.
156:
157: When $\beta >1$, there exist no singularities. When $0\leq \beta \leq 1$,
158: there exist singularities on the $\Phi $-axis, denoted by $(\phi _{n},0)$,
159: where
160: \begin{equation}
161: \left\{
162: \begin{array}{ll}
163: \phi _{n}=n\pi +(-1)^{n}\phi _{0} & \\
164: \phi _{0}=\arcsin \beta &
165: \end{array}%
166: \right. ,\quad \quad n\in \mathbb{Z}. \label{eq:f3}
167: \end{equation}%
168: For convenience, let $W(\phi ,z)=z$, $Q(\phi ,z)=\beta -sin\phi -\gamma z$.
169: Because $W(\phi ,z)=W(\phi +2\pi ,z)$, $Q(\phi ,z)=Q(\phi +2\pi ,z)$, both
170: with a period of $2\pi $ along the $\Phi $-axis, the distribution of the
171: vector field, determined by $W(\phi ,z)$ and $Q(\phi ,z)$, is also with a
172: period of $2\pi $ along the $\Phi $-axis, the plane can be roll up into a
173: cylindrical surface along the $\Phi $-axis, that is a cylindrical surface
174: system. Owing to the properties of cylindrical surface system, we just
175: discuss the interval $[-\pi -\phi _{0},\pi -\phi _{0}]$ without declaiming
176: and name it main interval. The singularities, denoted by the following
177: symbols as default, have the properties as following.
178:
179: \vskip2mm
180: \begin{minipage} [32mm]{138mm}
181: 1)\hskip 2mm When $0\leq \beta<1$, $A_{k}(\phi_{2k-1},0)$ is the
182: saddle of the equation. The slopes of two separatrices are
183: respectively
184: \begin{equation} \label{eq:f4}
185: \lambda_{1}=-\frac{\gamma}{2}+\sqrt{\left(\frac{\gamma}{2}\right)^{2}+\cos
186: \phi_{0}} \hskip 1mm ,
187: \end{equation}
188: \begin{equation} \label{eq:f5}
189: \lambda_{2}=-\frac{\gamma}{2}-\sqrt{\left(\frac{\gamma}{2}\right)^{2}+\cos
190: \phi_{0}} \hskip 1mm .
191: \end{equation}
192: When $\gamma>0$, $B_{k}(\phi_{2k},0)$ is the stable focus of the
193: equation.
194: \par
195: When $\gamma=0$, $B_{k}(\phi_{2k},0)$ is the center of the equation.
196: But it is not the subject of the current analysis.
197: \par
198: 2)\hskip 1mm When $\beta=1$, $\phi_{0}=\frac{\pi}{2}$,
199: $\phi_{2k+1}=\phi_{2k}$, namely, $A_{k+1}$ and $B_{k}$ coincide in
200: position. They become combining singularities.
201: \par
202: 3)\hskip 1mm When $\beta>1$, there exist no singularities.
203: \par Notice $k\in \mathbb{Z}$ above mentioned.
204: \end{minipage}
205: \vskip2mm
206:
207: While $\gamma >0$, $0\leq \beta <1$, in the main interval, there are four
208: special trajectories departing from or going into the two saddles. Let $R$, $%
209: V$ denote the trajectory departing from $A_{0}$ and $A_{1}$ respectively;
210: Let $S$, $U$ denote the trajectory going into $A_{0}$ and $A_{1}$
211: respectively; as shown in Figure \ref{fig:p1}. As default symbols, $R$, $V$,
212: $S$, $U$ would represent the four special trajectories given above. In order
213: to demarcate different regions, we draw a curve $G$ as shown in Figure \ref%
214: {fig:p1}, which expression is%
215: \begin{equation}
216: z=\frac{\beta -\sin \phi }{\gamma }. \label{eq:f6}
217: \end{equation}
218:
219: \begin{figure}[h]
220: \centering
221: \includegraphics[width=2.80in, height=1.80in]{myfig1.png}
222: \caption{Special trajectories in the main interval.}
223: \label{fig:p1}
224: \end{figure}
225:
226: \begin{minipage} [32mm]{138mm}
227: Let $\Lambda_{1}$ denotes the domain closed by $G$ and the $\Phi$-axis for
228: $z>0$ ;\\
229: Let $\Lambda_{2}$ denotes the domain closed by $G$ and the $\Phi$-axis for $z<0$;\\
230: Let $\Lambda_{3}$ denotes the domain above $G$ for
231: $z>0$;\\
232: Let $\Lambda_{4}$ denotes the domain below $G$ for $z<0$.\\
233: \end{minipage}
234:
235: From Equation (\ref{eq:f2}), it is easy to obtain:%
236: \begin{equation}
237: \frac{\mathrm{d}z}{\mathrm{d}\phi }=\frac{\beta -\sin \phi }{z}-\gamma .
238: \label{eq:f7}
239: \end{equation}
240:
241: From Equation (\ref{eq:f7}), it is easy to derive the following properties:
242:
243: \begin{minipage} [32mm]{138mm}
244: a) $\frac{\mathrm{d}z}{d\phi}>0$, while $(\phi,z)\in \Lambda_{1}.or.\Lambda_{2}$;\\
245: b) $\frac{\mathrm{d}z}{d\phi}<0$, while $(\phi,z)\in \Lambda_{3}.or.\Lambda_{4}$;\\
246: c) $\frac{\mathrm{d}z}{d\phi}=0$, while $(\phi,z)\in G$;\\
247: d) $\frac{\mathrm{d}z}{d\phi}=\infty$, while z=0 and
248: $\phi \neq \phi_{n}$.
249: \end{minipage}
250:
251: On the base of the properties above, it is easy to give the Lemma 1 about
252: the trajectory $R$.
253:
254: \noindent \textbf{Lemma 1.} \textit{Let $K_{0}:\phi =\phi _{1}$ expresses a
255: line perpendicular to the $\Phi $-axis through point $A_{1}$, as shown in
256: Figure 1. Then the trajectory $R$ departing from $A_{0}$ must passe over the
257: upper half plane and is bound to reach at point $C(\phi _{c},0)$ of the $%
258: \Phi $-axis and $\phi _{c}\in \lbrack \phi _{0},\phi _{1}]$ as well, or at
259: point $C(\phi _{1},z_{c})$ of line $K_{1}$ and $z_{c}\in (0,\frac{\beta +1}{%
260: \gamma })$ as well.}
261:
262: we can also give Lemma 2.
263:
264: \noindent \textbf{Lemma 2.} \textit{If a periodic solution $z=z(\phi )$
265: exists for Equation (\ref{eq:f2})}, i.e., \textit{$z(\phi )=z(\phi +2\pi )$,
266: $\phi \in (-\infty ,+\infty )$, then it must be satisfied}
267: \begin{equation}
268: \int_{\phi }^{\phi +2\pi }z(\phi )\,\mathrm{d}\phi =\frac{2\pi \beta }{%
269: \gamma }. \label{eq:f8}
270: \end{equation}%
271: The proof of the Lemma 2 is very simple. According to Equality (\ref{eq:f7})
272: and noticing the periodicity of $z(\phi )$, it is easy to obtain Equality (%
273: \ref{eq:f8}).
274:
275: For the proofs in the next section, two definitions are provided as
276: following.
277:
278: \noindent \textbf{Definition 1.} \textit{Here $\beta$ is fixed ($\beta \geq0$%
279: ), take $\gamma$ $(\gamma \geq0)$ as a parameter, it constructs a rotated
280: vector field. $\theta$ is the angle between the vector and the $\Phi$-axis,
281: the changing ratio of $\theta$ with respect to $\gamma$ is
282: \begin{equation} \label{eq:f9}
283: \frac{\mathrm{d}\theta}{\mathrm{d}\gamma}=-\frac{z^{2}}{z^{2}+(\beta-\sin
284: \phi-\gamma \cdot z)^{2}}\leq0 .
285: \end{equation}
286: }
287:
288: \noindent \textbf{Definition 2.} \textit{Here $\gamma$ is fixed ($\gamma
289: \geq0 $), take $\beta$ $(\beta \geq0)$ as a parameter, it constructs a
290: general rotated vector field. $\theta$ is the angle between the vector and
291: the $\Phi $-axis, the changing ratio of $\theta$ with respect to $\beta$ is}
292: \begin{equation} \label{eq:f10}
293: \frac{\mathrm{d}\theta}{\mathrm{d}\beta}=\frac{z}{z^{2}+(\beta-\sin
294: \phi-\gamma \cdot z)^{2}},
295: \end{equation}
296:
297: \textit{While $\gamma \in \lbrack 0,+\infty )$, $\beta \in \lbrack 0,+\infty
298: )$ and $z>0$, we have $-\frac{\pi }{2}<\theta <\frac{\pi }{2}$. }
299:
300: Meanwhile, we give the third lemma.
301:
302: \noindent \textbf{Lemma 3.} \textit{Here is $\beta =0$ , for arbitrary $%
303: \gamma _{1}>0$, while $\gamma =\gamma _{1}$, then the trajectory $R$
304: departing from $A_{0}$ must passe over upper half of plane and is bound to
305: reach at point $C(\phi _{c},0)$ of the $\Phi $-axis, and the point $C$
306: locates between focus $B_{0}$ and saddle $A_{1}$, i.e. $\phi _{c}\in \lbrack
307: \phi _{0},\phi _{1})$. }
308:
309: \noindent \textbf{Proof.} Let $\gamma =0$, $\beta =0$, thus Equation (\ref%
310: {eq:f2}) degenerate to be a integrable system. It is a sine-Gordon equation,
311: Namely, ${\displaystyle\frac{\mathrm{d}^{2}\phi }{\mathrm{d}t^{2}}+\sin \phi
312: =0}$. It is easy to give a solution,
313: \begin{equation}
314: z=\sqrt{2(\cos \phi +1)}. \label{eq:f11}
315: \end{equation}%
316: As shown in Figure \ref{fig:p2}, its corresponding trajectory $L_{0}$
317: connects the saddle $A_{0}$ with $A_{1}$.
318:
319: \begin{figure}[h]
320: \centering
321: \includegraphics[width=2.80in, height=1.80in]{myfig2-1.png}
322: \caption{The Trajectories as $\protect\beta=0$.} \label{fig:p2}
323: \end{figure}
324:
325: \noindent The slope of its tangent line at saddle $A_{0}$, $A_{1}$ is $\pm 1$
326: respectively. While $\gamma =\gamma _{1}$ $(\gamma _{1}\geq 0)$ and $\beta
327: =0 $, according to Equality (\ref{eq:f4}) gives that the slope of the
328: tangent of trajectory $R$ at point $A_{0}$ is $\lambda _{1}$, where $%
329: 0<\lambda _{1}<1 $. Therefore, $R$ is bound to locate at the inside of curve
330: $L_{0}$ at the beginning. According to Definition 1 gives that, $\frac{%
331: \mathrm{d}\theta }{\mathrm{d}\gamma }<0$ for the vectors at each point on
332: the curve $L_{0}$ except point $A_{0}$ and $A_{1}$, so the vectors direct
333: toward the inside of curve $L_{0}$ while $\gamma =\gamma _{1}$. According to
334: the uniqueness of the solution, it can be obtained that, the trajectory $R$
335: runs at inside of curve $L_{0}$ all the time, or intersect at the
336: singularity $A_{1}$. If we assume that trajectory $R$ intersects at point $%
337: A_{1}$, thus, the slope of $R $ at point $A_{1}$ is $\geq -1$ because it
338: reaches point $A_{1}$ from inside of $L_{0}$. According to Equality (\ref%
339: {eq:f5}) gives its slope is $\lambda _{2}<-1$. Therefore, they contradict
340: each other. So $R$ can not reach point $A_{1}$. According to of Lemma 1
341: indicates that $R$ must passe over the upper half plane and reaches point $%
342: C(\phi _{c},0)$ on the $\Phi $-axis, and point $C(\phi _{c},0)$ locates
343: between point $B_{0}$ and $A_{1}$, i.e. $\phi _{0}\leq \phi _{c}<\phi _{1}$.
344:
345: This completes the proof.
346:
347: \setcounter{section}{1} %\setcounter{equation}{0}
348: \renewcommand{\theequation}
349: {\arabic{equation}}
350:
351: \begin{flushleft}
352: {\large \textbf{3.\hspace*{2mm} The proofs of theorem\newline
353: }}
354: \end{flushleft}
355:
356: Now we propose the theorem as following and prove it on the basis of the
357: lemmas above.
358:
359: \noindent \textbf{Theorem.} \textit{Suppose $\gamma=\gamma_{1}$, where $%
360: \gamma _{1}>0$. $R(\beta)$ denotes the trajectory departing from point $%
361: A_{0} $, it must have the propositions as following.}
362:
363: \vskip 2mm
364: \begin{minipage} [32mm]{140mm}
365: \textit{ a)\hspace*{2mm} Equation (\ref{eq:f2}) must have a value of
366: $\beta_{0}(\gamma_{1})$, $0<\beta_{0} \leq1$, while
367: $\beta=\beta_{0}(\gamma_{1})$, $R(\beta_{0})$ is bound to intersect
368: the $\Phi$-axis at point $A_{1}$. Here this trajectory is named as a
369: critical periodic solution or pseudo-periodic solution. $\beta_{0}$
370: is called the critical value, with which the equation will have a
371: critical periodic solution or pseudo-periodic solution.}
372: \par
373: \textit{b)\hspace*{2mm} $\beta_{0}(\gamma_{1})$ is unique.}
374: \par
375: \textit{c)\hspace*{2mm} While $\beta>\beta_{0}$, a periodic solution
376: $z=z_{T}(\phi)$ exists for equation (\ref{eq:f2}), that is
377: $z_{T}(\phi)=z_{T}(\phi+2\pi)$, moreover $z_{T}(\phi)>0$, $\phi \in
378: (-\infty,+\infty)$; while $\beta<\beta_{0}$, no periodic solution
379: exists for equation (\ref{eq:f2}).}
380: \par
381: \textit{d)\hspace*{2mm} The periodic solution of equation
382: (\ref{eq:f2}) is unique and stable.}
383: \par
384: \textit{e)\hspace*{2mm} Suppose $\beta_{0}(\gamma_{1})=1$, and if
385: $\gamma_{2}>\gamma_{1}$, then $\beta_{0}(\gamma_{2})=1$.}
386: \end{minipage}
387:
388: \noindent \textbf{Proof.}
389:
390: a)\hspace*{2mm} First, Let $\beta =0$, $\gamma =\gamma _{1}$, thus, $\phi
391: _{0}=0$. Three singularities on the main interval are denoted by $%
392: A_{0}^{\ast }(\phi _{-1}^{\ast },0)$, $B_{0}^{\ast }(\phi _{0}^{\ast },0)$, $%
393: A_{1}^{\ast }(\phi _{1}^{\ast },0)$ respectively. According to Lemma 3
394: derives that the trajectory $R(0)$ departing from point $A_{0}^{\ast }$ must
395: reach the point $C^{\ast }(\phi _{c}^{\ast },0)$ on the $\Phi $-axis,
396: moreover $0\leq \phi _{c}^{\ast }<\pi $, as shown in Figure \ref{fig:p3}.
397: \begin{figure}[h]
398: \centering
399: \includegraphics[width=2.80in, height=1.79in]{myfig3.png}
400: \caption{The trajectories $R$ as $\protect\beta =0$ and $\protect\beta >0$.}
401: \label{fig:p3}
402: \end{figure}
403:
404: Now let $\beta >0$, but $\beta \leq 1$, thus, $\phi _{0}=\arcsin \beta >0$,
405: moreover $\phi _{0}\leq \frac{\pi }{2}$. The new singularities on the main
406: interval are written as $A_{0}(\phi _{-1},0)$, $B_{0}(\phi _{0},0)$, $%
407: A_{1}(\phi _{1},0)$ respectively. $R(\beta )$ denotes the trajectory
408: departing from $A_{0}$. Let $K_{1}$ is a line perpendicular to the $\Phi $%
409: -axis through the point $A_{1}$. Obviously, point $A_{0}$ is on the left of
410: point $A_{0}^{\ast }$, the trajectory $R(\beta )$ run at outside of the
411: trajectory $R(0)$ at the beginning. According to Definition 2 gives that $%
412: \frac{\mathrm{d}\theta }{\mathrm{d}\beta }>0$ for each point on the curve $%
413: R(0)$ except point $A_{0}^{\ast }$ and $C^{\ast }$. Therefore, while $\beta
414: >0$, the vector of field defined by the equation rotates anticlockwise. The
415: direction of the vector at each point of curve $R(0)$ is towards outside.
416: Hence, according to the uniqueness of solution and Lemma 1, it can be proved
417: that the trajectory $R(\beta )$ must run at the outside of $R(0)$,
418: furthermore, it reaches point $C(\phi _{c},0)$ on the $\Phi $-axis, where $%
419: \phi _{max}\leq \phi _{c}\leq \phi _{1}$, $\phi _{max}=max\{\phi _{c}^{\ast
420: },\phi _{0}\}$; or it reaches point $C(\phi _{1},z_{c})$ on line $K_{1}$,
421: where $0<z_{c}<\frac{\beta +1}{\gamma _{1}}$, as shown in Figure \ref{fig:p3}%
422: .
423:
424: Assume that $R(\beta )$ intersect line $K_{1}$ at point $C(\phi _{1},z_{c})$%
425: , and $0<z_{c}<\frac{\beta +1}{\gamma _{1}}$. According to the continuing
426: dependence of solution on the parameter\cite{Perko}, there must exist a
427: trajectory $R(\beta _{0})$, where $0<\beta _{0}<\beta $, which intersects
428: point $A_{1}$. Proposition a) is correct.
429:
430: Assume that $R(\beta)$ intersect point $C(\phi_{c},0)$ on the $\Phi$-axis,
431: and $\phi_{max}\leq \phi_{c}\leq \phi_{1}$, where $\phi_{max}=max\{
432: \phi_{c}^{\ast },\phi_{0}\}$. We notice that point $C(\phi_{c},0)$ must not
433: locate on the left of point $C^{\ast}(\phi_{c}^{\ast},0)$ at least by
434: increasing of $\beta$, point $A_{1}$ moving leftward monotonously. And while
435: $\beta=1$, $\phi _{0}=\frac{\pi}{2}$, $B_{0}$ coincides with $A_{1}$, $C$
436: must be between the two point. According to the these situations, it can be
437: declared that there must exist a trajectory $R(\beta_{0})$, where $%
438: 0<\beta_{0}\leq1$, which intersects point $A_{1}$. Proposition a) is also
439: correct. Thus the proof of proposition a) is complete.
440:
441: b)\hspace*{2mm} Proof by contradiction. Assume that $\beta_{0}(\gamma_{1})$
442: is not unique, here is $\beta_{0}^{\prime}(\gamma_{1})\neq
443: \beta_{0}(\gamma_{1}) $, we may set $\beta_{0}^{\prime}(\gamma_{1})>%
444: \beta_{0}(\gamma_{1}) $.
445:
446: While $\beta =\beta _{0}$, two saddles are denoted by $A_{0}^{(1)}$ and $%
447: A_{1}^{(1)}$ respectively. The trajectory connecting them is denoted by $%
448: R(\beta _{0})$, as shown in Figure \ref{fig:p4}.
449:
450: While $\beta =\beta _{0}^{\prime }$, two saddles are denoted by $A_{0}^{(2)}$
451: and $A_{1}^{(2)}$ respectively. The trajectory connecting them is denoted by
452: $R(\beta _{0}^{\prime })$, as shown in Figure \ref{fig:p4}.
453: \begin{figure}[h]
454: \centering
455: \includegraphics[width=2.80in, height=1.79in]{myfig4.png}
456: \caption{Two critical trajectories which intersect each other.}
457: \label{fig:p4}
458: \end{figure}
459:
460: Because $\beta _{0}^{\prime }>\beta _{0}$, obviously, point $A_{0}^{(2)}$ is
461: on the left of point $A_{0}^{(1)}$, point $A_{1}^{(2)}$ on the left of point
462: $A_{1}^{(1)}$. Therefore $R(\beta _{0}^{\prime })$ must intersect with $%
463: R(\beta _{0})$, the intersection is not on the $\Phi $-axis, as shown in
464: Figure \ref{fig:p4}. Referring to the preceding proofs will prove that $%
465: R(\beta _{0}^{\prime })$ can not intersect $R(\beta _{0})$. The result
466: contradicts with the preceding assumption, hence $\beta _{0}(\gamma _{1})$
467: must be unique. The proof of proposition b) is complete.
468:
469: c)\hspace*{2mm} First, we prove for the case of $\beta>\beta_{0}$.
470:
471: Suppose that $\beta=\beta_{0}$, the equation has a critical periodic
472: trajectory $R(\beta_{0})$, the saddles connecting by it are denoted by $%
473: A_{0}^{\ast}$ and $A_{1}^{\ast}$ respectively.
474:
475: While $\beta >\beta _{0}$, the trajectory departing from the new singularity
476: $A_{0}$ is denoted by $R(\beta )$. Referring to the preceding proofs derives
477: that $R(\beta )$ must run above $R(\beta _{0})$, as shown in Figure \ref%
478: {fig:p5}. The function for $R(\beta )$ is expressed as following:
479: \begin{equation}
480: z=z_{\beta }(\phi ),\quad \quad \phi \in (-\infty ,+\infty ). \label{eq:f12}
481: \end{equation}
482:
483: \noindent Obviously, here is $z_{\beta}(-\pi-\phi_{0})=0$, $z_{\beta}(\pi
484: -\phi_{0})>0$, thus $z_{\beta}(-\pi-\phi_{0})<z_{\beta}(\pi-\phi_{0})$.
485: \begin{figure}[h]
486: \centering
487: \includegraphics[width=2.80in, height=1.80in]{myfig5-1.png}
488: \caption{The trajectories as $\protect\beta>\protect\beta_{0}$.}
489: \label{fig:p5}
490: \end{figure}
491:
492: According to Equality (\ref{eq:f7}) gives that $\displaystyle\frac{\mathrm{d}%
493: z}{\mathrm{d}\phi }\leq 0$ for each point on the line $L$:$z=\frac{\beta +1}{%
494: \gamma _{1}}$. The vectors of field direct from left to right, therefore the
495: vectors of field on line $L$ point downwards of $L$. Moreover, we notice
496: that the vectors of field above line $L$ direct downwards absolutely. So we
497: take a trajectory passing through point $H_{0}(-\pi -\phi _{0},\frac{\beta +1%
498: }{\gamma _{1}})$, the expression of function for it is:
499: \begin{equation}
500: z=z_{m}(\phi ),\quad \quad \phi \in (-\infty ,+\infty ). \label{eq:f13}
501: \end{equation}%
502: Thus there are $z_{m}(-\pi -\phi _{0})=\frac{\beta +1}{\gamma _{1}}$, $%
503: z_{m}(\pi -\phi _{0})\leq \frac{\beta +1}{\gamma _{1}}$, and $z_{m}(-\pi
504: -\phi _{0})\geq z_{m}(\pi -\phi _{0})$. According to the continuing
505: dependence of solutions on the initial values\cite{Perko}, there must exists
506: a point $P$ on the line segment $A_{0}H_{0}$, through which the trajectory
507: satisfies that $z_{T}(-\pi -\phi _{0})=z_{T}(\pi -\phi _{0})$, obviously, $%
508: z_{T}(\phi )>0$. According to the properties of cylindrical surface system
509: derives,
510: \begin{equation}
511: z_{T}(\phi )=z_{T}(\phi +2\pi ),\quad \quad \phi \in (-\infty ,+\infty ),
512: \label{eq:f14}
513: \end{equation}
514:
515: \noindent i.e. Equation (\ref{eq:f2}) has a periodic solution, moreover $%
516: z_{T}(\phi )>0$. The proof is complete for the proposition of $\beta >\beta
517: _{0}$.
518:
519: Second, we prove for the case of $\beta<\beta_{0}$.
520:
521: Suppose that $\beta =\beta _{0}$, two saddles are denoted by $A_{0}^{\ast }$
522: and $A_{1}^{\ast }$ respectively. Let $R(\beta _{0})$ denotes the trajectory
523: which connects two saddles, for which the expression of function is $%
524: z=z_{0}(\phi )$, hence $z_{0}(\phi )=z_{0}(\phi +2\pi )$, $\phi \in (-\infty
525: ,+\infty )$.
526:
527: While $\beta <\beta _{0}$, two saddles are denoted by $A_{0}$ and $A_{1}$
528: respectively. Let $R(\beta )$ denotes the trajectory departing from point $%
529: A_{0}$, Let $U(\beta )$ denotes the one going into point $A_{1}$, as shown
530: in Figure \ref{fig:p6}. We will prove by contradiction. Assuming there is a
531: periodic solution, the expression of function for it is $z=z_{\beta }(\phi )$%
532: . Obviously, $z_{\beta }(\phi )=z_{\beta }(\phi +2\pi )$, $\phi \in (-\infty
533: ,+\infty )$.
534: \begin{figure}[h]
535: \centering
536: \includegraphics[width=2.80in, height=1.79in]{myfig6.png}
537: \caption{The trajectories as $\protect\beta <\protect\beta _{0}$.}
538: \label{fig:p6}
539: \end{figure}
540:
541: As analogue of the preceding proofs, it is easy to derive the relation of
542: position of trajectory $R(\beta _{0})$ with $R(\beta )$ and $U(\beta )$
543: geometrically. $R(\beta )$, $U(\beta )$ locates by each side of $R(\beta
544: _{0})$ respectively, as shown in Figure \ref{fig:p6}. Obviously, if $%
545: z=z_{\beta }(\phi )$ passes through the line segment $A_{0}H_{0}^{\prime }$,
546: it must be through axis $\Phi $. Observing the directions of the vectors of
547: field at each side of axis $\Phi $, we can declare that the trajectory
548: passing through axis $\Phi $ isn't a periodic trajectory. Therefore, $%
549: z=z_{\beta }(\phi )$ is impossible the one passing through the line segment $%
550: A_{0}H_{0}^{\prime }$. So it is declared that there are only two relations
551: between $z=z_{\beta }(\phi )$ and $z=z_{0}(\phi )$ as following.
552:
553: In the first case, $z_{\beta }(\phi _{-1})<z_{0}(\phi _{-1})$, because the
554: trajectory of $z=z_{\beta }(\phi )$ doesn't crossover the $\Phi $-axis, we
555: have $z_{\beta }(\phi )\leq 0$, $\phi \in (-\infty ,+\infty )$. While $\beta
556: >0$, $\gamma >0$, according to Lemma 2 indicates that this kind of periodic
557: solution is impossible.
558:
559: In the second case, $z_{\beta }(\phi )>z_{0}(\phi )$, $\phi \in \lbrack \phi
560: _{-1}^{\ast },\phi _{1}^{\ast }]$. While $\beta <\beta _{0}$, according to
561: Lemma 2 indicates that this situation is also impossible.
562:
563: Summarizing the conclusions above, we conclude that the equation does not
564: exists the periodic solution while $\beta<\beta_{0}$. The proof is complete
565: for the proposition of $\beta<\beta_{0}$.
566: \begin{figure}[h]
567: \centering
568: \includegraphics[width=1.40in, height=1.60in]{cylinder1.png}
569: \caption{The cycle of the second kind on the cylindrical surface.}
570: \label{fig:p7}
571: \end{figure}
572:
573: d)\hspace*{2mm} According to Lemma 2, it is easy to prove that the periodic
574: solution is unique. In the proof, it just need to notice the property that
575: trajectories don't intersect each others.
576:
577: The periodic solution is a cycle of the second kind on the cylindrical
578: surface, as shown in Figure \ref{fig:p7}. The characteristic exponent of the
579: limit-cycle is $-\gamma <0$, therefore it is a stable limit-cycle\cite{Perko}%
580: . The periodic solution of the equation is a stable periodic solution.
581:
582: e)\hspace*{2mm} Assume $\beta _{0}(\gamma _{1})=1$. Let $\gamma =\gamma
583: _{1}, $ $\beta =1$, $R(\gamma _{1})$ denotes the critical periodic
584: trajectory connecting singularity $A_{0}$ and $A_{1}$, as shown in Figure %
585: \ref{fig:p8}. The singularity $B_{0}$ coincides with $A_{1}$. Let $\gamma
586: =\gamma _{2}>\gamma _{1}$, $\beta =1$, $R(\gamma _{2})$ denotes the
587: trajectory departing from $A_{0}$. Referring to the preceding proof derives
588: that, trajectory $R(\gamma _{2})$ must run at the inside of $R(\gamma _{1})$%
589: . According to Lemma 1 derives that $R(\gamma _{2})$ must intersect with
590: axis $\Phi $ at point $A_{1}$ $(i.e.$ $B_{0})$. it is the critical periodic
591: trajectory yet. Notice that $\beta =1$ still. According to the uniqueness of
592: $\beta _{0}$ gives that $\beta _{0}(\gamma _{2})=1$, $\gamma _{2}>\gamma
593: _{1} $. In this case, the slope by which $R(\gamma _{2})$ going into point $%
594: A_{1}$ is $\lambda _{1}=0$.
595:
596: This completes the proof.
597: \begin{figure}[h]
598: \centering
599: \includegraphics[width=2.80in, height=1.79in]{myfig7.png}
600: \caption{The critical periodic trajectories as $\protect\gamma _{2}>\protect%
601: \gamma _{1}$.}
602: \label{fig:p8}
603: \end{figure}
604:
605: \begin{flushleft}
606: {\large \textbf{4.\hspace*{2mm} Discussion of the conclusion }}
607: \end{flushleft}
608:
609: The theorem above gives a conclusion that, for arbitrary $\gamma >0$,
610: Equation (\ref{eq:f2}) exists a critical value $\beta _{0}$, where $0<$ $%
611: \beta _{0}\leq 1$. While $\beta \geq \beta _{0}$, it exists a unique,
612: positive and periodic solution that is a stable limit-cycle of the second
613: kind on the cylindrical surface. According to the Bendixson's Criteria\cite%
614: {Perko}, it is easy to prove that Equation (\ref{eq:f2}) exists no cycle of
615: the first kind on the cylindrical surface for $\gamma >0$. It indicates
616: that, whatever the initial state it is, the final state will be in a
617: periodic motion toward positive. While $\beta <\beta _{0}$, Equation (\ref%
618: {eq:f2}) exists no any periodic solution. In fact, all trajectories on the
619: phase plane will go into the saddles or focuses in this situation. The
620: focuses are stable. The saddles are unstable. If there are some
621: disturbances, it will deviate from the saddles. In the final state, all the
622: trajectories will go into the focuses. In physics, whatever the initial
623: state it is, the system will stay at equilibrium points (focuses) when it
624: reaches the stable state.
625:
626: Proposition e) of the theorem indicates that, supposing a minimum value $%
627: \gamma _{min}$ such that $\beta _{0}(\gamma _{min})=1$, while $\gamma \geq
628: \gamma _{min}$, the critical value $\beta _{0}$ will be 1 all the time. By
629: numerical calculations, M. Urabe has given the definite value, that is $%
630: \gamma _{min}=1.193$. These conclusions above are very useful in the
631: research of CDW. The derivations in detail was presented in another paper
632: \cite{Li}. It is unnecessary to go into details in this paper.
633:
634: \vspace{0.4cm} \noindent{\large \textbf{Acknowledgments}}
635:
636: The author thanks Professor L. Zhao for his help on LaTex software.
637:
638: %\vspace{1cm} \noindent \textbf{References}
639:
640: \begin{thebibliography}{99}
641: \bibitem{Ott} {\small E. Ott, Chaos in Dynamical Systems, World Publishing
642: Corporation, Beijing, 2005.}
643:
644: \bibitem{Bohr} {\small T. Bohr, P. Bak, M.H. Jensen, Transition to chaos by
645: interaction of resonances in dissipative systems. II. Josephson junctions,
646: charge-density waves, and standard maps, Phys. Rev. A 30(1984) 1970-1980. }
647:
648: \bibitem{Gruner} {\small G. Gr\"{u}ner, A. Zawadowski \& P.M. Chaikin,
649: Nonlinear conductivity and noise due to charge-density wave depining in
650: NbSe3, Phys. Rev. Lett. 46(7)(1981) 511-515. }
651:
652: \bibitem{Li} {\small Li L.G., Ruan Y.F., The analysis on the classic model
653: of the charge density wave, Acta Phys. Sin. 55(1)(2006) 441-445 (in
654: Chinese). }
655:
656: %\bibitem{Fleming} {\small R.M. Fleming, Electric-field depining of
657: %charge-density wave in NbSe3, Phys. Rev. B 22(1980) 5606-5612. }
658:
659: \bibitem{Poinc} {\small H. Poincar\'{e}, M\'{e}moire sur les courbes d\'{e}%
660: finies par une \'{e}quation diff\'{e}rentielle, J. Math. Pures Appl.
661: 7(3)(1881) 375-422; 8(1882) 251-296; 1(4)(1885) 167-244; 2(1886) 151-217. }
662:
663: \bibitem{Duff} {\small G.F.D. Duff, Limit-cycles and rotated vector fields,
664: Ann. Math. 57(1953) 15-31. }
665:
666: \bibitem{Seifert} {\small G. Seifert, Contributions to the theory of
667: nonlinear oscillations, IV(1958) 125-140. }
668:
669: \bibitem{Chen} {\small Chen Xiang-yan, General rotated vector fields, Acta
670: Nanjing Uni. 1(1975) 100-108 (in Chinese). }
671:
672: \bibitem{Ma} {\small Ma Zhi-en, The motion of singular closed trajectory in
673: the rotated vector field, J. Xi'an Jiaotong Uni. 4(1978) 49-65 (in Chinese).
674: }
675:
676: \bibitem{Perko} {\small L. Perko, Differential Equations and Dynamical
677: Systems, Springer-Verlag, NewYork, 1991. }
678:
679: \bibitem{Urabe} {\small M Urabe, The least upper bound of a damping
680: coefficient insuring the existence of a periodic motion of a pendulum under
681: constant torque, J. Sci. Hiroshima Uni. A 18(1954) 379-389.}
682:
683: %\bibitem {Portis}{\small A.M. Portis, Class model of charge density wave
684: %transport, Mol. Cryst. Liq. Cryst. 81(1982) 59-72.}
685: \end{thebibliography}
686:
687: {\small \vskip 10mm %\noindent
688: %Physics Doctor, School of Science, Tianjin University, Tianjin,
689: %300072, China.
690: }
691:
692: \end{document}
693: