1: \documentclass[12pt]{iopart}
2: %\usepackage{amsmath}
3: \usepackage{graphicx}
4: \usepackage{setstack,cite}
5: %\bibliographystyle{plain}
6: \begin{document}
7: \date{}
8:
9: \title[General solution for the modified Emden type equation]
10: {On the general solution for the modified
11: Emden type equation $\ddot{x}+\alpha x\dot{x}+\beta x^3=0$}
12:
13: \author{V~K~Chandrasekar, M~Senthilvelan and M~Lakshmanan}
14:
15: \address{Centre for Nonlinear Dynamics, Department of Physics, Bharathidasan
16: University, Tiruchirapalli - 620 024, India.}
17:
18: \begin{abstract}
19: In this paper, we demonstrate
20: that the modified Emden type equation (MEE),
21: $\ddot{x}+\alpha x\dot{x}+\beta x^3=0$, is integrable either explicitly or by
22: quadrature for any value of $\alpha$ and $\beta$. We also prove that the MEE
23: possesses appropriate time-independent Hamiltonian function for the full range
24: of parameters $\alpha$ and $\beta$. In addition, we show that the MEE is
25: intimately connected with two well known nonlinear models, namely the
26: force-free Duffing type oscillator equation and the two dimensional
27: Lotka-Volterra (LV) equation and thus the complete integrability of the latter
28: two models can also be understood in terms of the MEE.
29: \end{abstract}
30: \pacs{02.30.Hq, 02.30.Ik, 05.45.-a}
31: %Integrable systems, Nonlinear dynamics, Ordinary differential equation
32: \section{Introduction}
33: \label{sec1}
34: One of the well discussed models in nonlinear dynamics is the modified Emden
35: type equation (MEE), also called the modified Painlev\'e-Ince equation,
36: \begin{eqnarray}
37: \ddot{x}+\alpha x\dot{x}+\beta x^3=0,
38: \label {eq01}
39: \end{eqnarray}
40: where over dot denotes differentiation with respect to time and
41: $\alpha$ and $\beta$ are arbitrary parameters. This equation has
42: received attention from both mathematicians and physicists for more than a
43: century \cite{Painleve,Ince,Davis,
44: Kamke,Murphi}. For example, in the nineteenth
45: century, Painlev\'e had
46: studied this equation and identified general solution for two parametric
47: choices, namely, $(i)\;\beta=\frac{\alpha^2}{9}$ and
48: $(ii)\;\beta=-\alpha^2$ \cite{Painleve,Ince}. The above
49: differential equation (\ref{eq01}) arises in a variety of mathematical
50: problems such as univalued
51: functions defined by second order differential equations \cite{Gobulev} and
52: the Riccati equation \cite{Chisholm}. On the other hand
53: physicists have shown that this equation arises in different contexts:
54: for example, it occurs in the study of
55: equilibrium configurations of a spherical gas cloud acting under the mutual
56: attraction of its molecules and subject to the laws of thermodynamics
57: \cite{Dix:1990,Moreira} and in the modelling of the fusion of
58: pellets \cite{Erwin}. It also governs spherically symmetric expansion or
59: collapse of a relativistically gravitating mass \cite{McVittie}. This equation
60: can also be thought of as a one-dimensional analogue \cite{Chisholm,Yang} of
61: the boson `gauge-theory'
62: equations introduced by Yang and Mills.
63: Apart from the above, for the past two decades or so the invariance and integrability
64: properties of this equation alone have been studied in detail by a number of
65: authors,
66: see for example Ref. \cite{mahomed:1985,Duarte2,Bouquet,Sarleti,leach:1988a,
67: Steeb,feix:1997,Ibragimov1,Leach1,Chand,Chand1}. In a nutshell, the MEE
68: (\ref{eq01})
69: has been found to possess explicit general solution only for the following
70: parametric choices, that is,
71: $(i)\; \alpha=0,\;(ii)\;\beta=0,\;(iii)\;\beta=\frac{\alpha^2}{9}$ and
72: $(iv)\;\beta=-\alpha^2$. While the cases (i) and (ii) can be integrated
73: trivially, in the third case the equation is linearizable to a free particle
74: equation and in the fourth case the general solution can be expressed in terms
75: of Weierstrass elliptic function \cite{Painleve,Ince,Davis,
76: Kamke,Murphi,mahomed:1985,Duarte2,Bouquet,Sarleti,leach:1988a,Steeb,feix:1997,
77: Ibragimov1,Leach1,Chand,Chand1,Euler}.
78: Equation (\ref{eq01}) has also been noted to possess Painlev\'e property only
79: for certain values of $r=\frac{\alpha}{4\beta}(\alpha\pm\sqrt{\alpha^2-8\beta})$
80: \cite{leach:1988a,feix:1997}. Finally, we mention that equation
81: (\ref{eq01}) admits a two parameter Lie point symmetry group for arbitrary
82: values of $\alpha$ and $\beta$, while for the chioce $\beta=\frac{\alpha^2}{9}$
83: the equation (\ref{eq01}) possesses eight-parameter Lie point symmetries
84: \cite{mahomed:1985}. However, the
85: general solution for the equation (\ref{eq01}) with $\alpha$ and $\beta$ are
86: arbitrary is yet to be explored.
87:
88: Very recently \cite{Chand1}, the present authors have studied a generalized
89: version of this equation from a different perspective and shown that the equation
90: (\ref{eq01}), for $\alpha^2 \geq 8\beta$, possesses time independent integrals
91: and admits a Hamiltonian formalism which in turn ensures its complete
92: integrability \cite{Chand1}. However, due to the complicated form of
93: the first integral the general solution was not obtained in the previous work.
94: Keeping in mind the historical importance and popularity of this model we
95: kept exploring the general solution of equation (\ref{eq01}) in phase-space.
96: Based on our investigation, in this paper, we construct the time
97: independent integrals for the equation (\ref{eq01}) for arbitrary values of
98: $\alpha$ and $\beta$ (including the case $\alpha^2 < 8\beta$). Since the first
99: integrals are not in simple polynomial forms it
100: is difficult to obtain the general solution by just directly integrating them.
101: In order to overcome this difficulty, first we identify time
102: independent Hamiltonians from these time independent integrals
103: and making use of suitable canonical transformations we convert the
104: Hamiltonians into standard forms. We then integrate the new Hamiltonians and
105: obtain the general solutions or reduce to quadratures. In this way
106: we report the general solution for the equation (\ref{eq01}) for
107: arbitrary values of $\alpha$ and $\beta$ for the first time. Our motivation
108: to explore this
109: solution is, as we see below, also due to the fact that the MEE (\ref{eq01})
110: is not a stand-alone
111: model. It is intimately connected to force-free Duffing oscillator type equation
112: (vide equation (\ref{eq08}) below) and two dimensional Lotka-Volterra (LV) equation
113: (vide equation
114: (\ref{eq10}) below). The popularity and importance of these two models need no
115: emphasis \cite{Chand1,Bluman,Laks1,Murray,Cairo1,Cairo2}. Thus
116: exploring the general solution for the
117: equation (\ref{eq01}) also serves to establish the complete integrability
118: of these two models for appropriate parameters besides understanding the
119: dynamics of the other models mentioned in the introduction.
120:
121: The plan of the paper is as follows. In Section \ref{sec2}, we give the
122: time independent integrals and corresponting Hamiltonians for the MEE
123: (\ref{eq01}). Using suitable canonical transformations, we obtain the
124: general solutions for the parameter ranges $\alpha^2=8\beta,
125: \;\alpha^2 >8\beta$ and $\alpha^2 <8\beta$ separately in Section \ref{sec3}. In Section
126: \ref{sec4},
127: we show that the MEE (\ref{eq01}) is intimately connected with two other
128: well known nonlinear models, namely, the force-free Duffing oscillator type
129: equation and two dimensional Lotka-Volterra equation.
130: Finally, in Section \ref{sec5} we summarize our
131: results.
132:
133: \section{Time independent integrals and Hamiltonian description}
134: \label{sec2}
135: As pointed out in the introduction, the MEE (\ref{eq01}) cannot be
136: straightforwardly integrated. Making use of the modified Prelle-Singer method
137: developed by us recently \cite{Chand,Chand1} following the earlier work of
138: Duarte et al
139: \cite{Duarte}, in this section we first identify the first integrals
140: separately for each of
141: the three ranges (i) $\alpha^2 =8\beta$,
142: (ii) $\alpha^2 >8\beta$ and (iii) $\alpha^2 <8\beta$. Then we identify suitable
143: canonical Hamiltonian description for each of these cases.
144: \subsection{Time independent integrals}
145: \label{sec20}
146: In Ref. \cite{Chand1}, the time independent integrals for the equation
147: (\ref{eq01}) with $\alpha^2 \geq 8\beta$ has been reported using the modified
148: Prelle-Singer procedure. However, improving the ansatz
149: given in Ref. \cite{Chand1} one can obtain the time independent integrals
150: for all values of $\alpha$ and $\beta$ and the method of deriving the
151: integrals for the equation (\ref{eq01}) with $\alpha$ and $\beta$ arbitrary
152: is given in Appendix A. Using this method we identify the following
153: time independent first integrals for the
154: equation (\ref{eq01}) with $\alpha$ and $\beta$ arbitrary, that is,
155: \begin{eqnarray}
156: \fl\mbox{Case 1:}\;\alpha^2=8\beta &\;\;(r=2)\nonumber\\
157: &I=\log(\alpha^2x^2+4\alpha \dot{x})
158: -\frac{4\alpha\dot{x}}{\alpha ^2x^2+4\alpha\dot{x}}&
159: \label {intm03}\\
160: \fl\mbox{Case 2:}\; \alpha^2 >8\beta &\;\;(r\neq0,1,2) \nonumber\\
161: &I=\frac{(r-1)}{(r-2)}
162: \bigg(\dot{x}+\frac{(r-1)}{2r}\alpha x^2\bigg)^{1-r}
163: \bigg(\dot{x}+\frac{\alpha }{2r}x^2\bigg)
164: & \label {intm02}\\
165: \fl\mbox{Case 3:}\; \alpha^2 <8\beta & \nonumber\\
166: &I=\frac{1}{2}\log(2\dot{x}^2+\alpha x^2\dot{x}+\beta x^4)
167: +\frac{\alpha}{\omega}tan^{-1}\bigg[\frac{\alpha \dot{x}+2\beta x^2}
168: {\omega \dot{x}}\bigg],& \label {intm01}
169: \end{eqnarray}
170: %\label{com02}
171: %\end{subequations}
172: where $\omega=\sqrt{8\beta-\alpha^2}$ and
173: $r=\frac{\alpha}{4\beta}(\alpha\pm\sqrt{\alpha^2-8\beta})$.
174: Note that $r=0$ and $1$ correspond to the trivial cases $\alpha=0$ and
175: $\beta=0$, respectively, and so they are not considered here separately.
176:
177: As it is very difficult to integrate equations (\ref{intm03})-(\ref{intm01}) and
178: obtain the general solutions by direct integration, we
179: correlate these integrals with appropriate Hamiltonians. In the following we
180: briefly give the method of obtain the Hamiltonian from the known time
181: independent integral.
182: \subsection{Hamiltonian description}
183: \label{sec21}
184: To explore the Hamiltonian description of (\ref{intm03}), let us assume the
185: existence of a Hamiltonian
186: \begin{eqnarray}
187: I(x,\dot{x}) = H(x,p) = p \dot{x}-L(x,\dot{x}),
188: \label{mlin13}
189: \end{eqnarray}
190: where $L$ is the Lagrangian and $p$ is the canonically conjugate
191: momentum. Then
192: \begin{eqnarray}
193: &&\frac{\partial{I}}{\partial\dot{x}} = \frac{\partial{H}}{\partial\dot{x}} =
194: \frac{\partial{p}}{\partial\dot{x}}\dot{x} + p - \frac{\partial{L}}{\partial\dot{x}} =
195: \frac{\partial{p}}{\partial\dot{x}} \dot{x}.
196: \label{mlin13a}
197: \end{eqnarray}
198: From (\ref{mlin13a}) we identify
199: \begin{eqnarray}
200: p =\int \frac{I_{\dot{x}}}{\dot{x}} d\dot{x}+f(x),
201: \label{mlin13b}
202: \end{eqnarray}
203: where $f(x)$ is an arbitrary function of $x$ and the Lagrangian $L=p\dot{x}
204: -I(x,\dot{x})$.
205: Here, without loss of generality, we take $f(x)=0$. Substituting the integrals
206: (\ref{intm03})-(\ref{intm01}) into (\ref{mlin13b}) and integrating
207: the resultant integrals we can obtain the expression for the canonical
208: momentum $p$. Substituting
209: back the latter into the equation~(\ref{mlin13}) and simplifying the resultant
210: equation we arrive at the following Lagrangian,
211: \begin{eqnarray}
212: L=\left\{
213: \begin{array}{ll}
214: -\log(\dot{x}+\frac{\alpha}{4} x^2), & \;\;\; \alpha^2=8\beta\\
215: \frac{1}{(2-r)}(\dot{x}+\frac{(r-1)}{2r}\alpha x^2)^{(2-r)},
216: &\;\;\;\alpha^2>8\beta\\
217: \frac{1}{\omega}
218: \bigg(tan^{-1}\bigg[\frac{4\dot{x}+\alpha x^2}{\omega x^2}\bigg]
219: (\frac{4\dot{x}}{x})-\alpha tan^{-1}\bigg[\frac{\alpha \dot{x}+2\beta x^2}
220: {\omega \dot{x}}\bigg]\bigg)\\
221: \qquad \qquad -\frac{1}{2}\log(2\dot{x}^2+\alpha x^2\dot{x}+\beta x^4),
222: &\;\;\;\alpha^2 <8\beta
223: \end{array}\right.\label{mlin14}
224: \end{eqnarray}
225: and Hamiltonian
226: \begin{eqnarray}
227: H=\left\{
228: \begin{array}{ll}
229: \log(-\frac{4\alpha}{p})-\frac{\alpha}{4}px^2,
230: &\qquad\qquad\qquad\alpha^2=8\beta \\
231: \displaystyle{\frac{(r-1)}{(r-2)}(p)^{\frac{r-2}{r-1}}
232: -\frac{(r-1)}{2r}\alpha x^2p},&\qquad\qquad\qquad
233: \alpha^2>8\beta \\
234: \displaystyle{\frac{1}{2}\log[x^4 sec^2(\frac{\omega}{4} x^2p)]
235: -\frac{\alpha }{4}x^2p},
236: &\qquad\qquad\qquad\alpha^2 <4\beta
237: \end{array}\right.\label{mlin16}
238: \end{eqnarray}
239: where the canonically conjugate momentum
240: \begin{eqnarray}
241: p=\left\{
242: \begin{array}{ll}
243: -\frac{1}{(\dot{x}+\frac{\alpha}{4}x^2)}
244: &\qquad\qquad\qquad\qquad\quad\;\; \alpha^2=8\beta\\
245: \displaystyle{\bigg(\dot{x}+\frac{(r-1)}{2r}\alpha x^2
246: \bigg)^{1-r}}&\qquad\qquad\qquad\qquad\quad\;\;\alpha^2>8\beta\\
247: \displaystyle{\frac{4}{\omega x^2}
248: tan^{-1}\bigg[\frac{4\dot{x}+\alpha x^2}{2\omega x^2}\bigg]},
249: &\qquad\qquad\qquad\qquad\quad\;\; \alpha^2 <8\beta
250: \end{array}\right.\label{mlin15}
251: \end{eqnarray}
252: respectively for the MEE (\ref{eq01}).
253:
254: One can easily check that the canonical equations of motion for the above
255: Hamiltonians are nothing but the equation of motion (\ref{eq01}) of the
256: MEE in the appropriate parametric regimes.
257:
258: \section{General solutions}
259: \label{sec3}
260: In this section, we consider each of the above three cases separately,
261: namely, (i) $\alpha^2 =8\beta$,
262: (ii) $\alpha^2 >8\beta$ and (iii) $\alpha^2 <8\beta$,
263: and obtain their respective general solutions using suitable canonical
264: transformations.
265: \subsection{Case 1: $\alpha^2=8\beta$}
266: \label{sec22}
267: To derive the general solution for this parametric choice first we consider the
268: Hamiltonian given in (\ref{mlin16}) for $\alpha^2=8\beta$ as
269: \begin{eqnarray}
270: H=\log(-\frac{4\alpha}{p})-\frac{\alpha}{4}px^2.\label{eq04}
271: \end{eqnarray}
272: Introducing the canonical transformation
273: \begin{eqnarray}
274: x=\frac{4P}{\alpha U},\qquad\;\;p=-\frac{\alpha U^2}{8},
275: \label {eq05}
276: \end{eqnarray}
277: the Hamiltonian (\ref{eq04}) can be recast into the standard form
278: \begin{eqnarray}
279: H=\frac{1}{2}P^2+\log(\frac{32}{U^2})\equiv \hat{E},\label{eq06a}
280: \end{eqnarray}
281: where $\hat{E}$ is a constant. From the corresponding canonical equations
282: $\dot{U}=P,\;\dot{P}=\frac{2}{U}$,
283: equation (\ref{eq06a}) can be rewritten as
284: \begin{eqnarray}
285: E=\frac{1}{2}\dot{U}^2-2\log(U),\;\;\; E=\hat{E}-\log(32).\label{eq06}
286: \end{eqnarray}
287: Rewriting the above equation we get
288: \begin{eqnarray}
289: \frac{dU}{\sqrt{2E+4\log(U)}}=dt.\label{eq06b}
290: \end{eqnarray}
291: Integrating the equation (\ref{eq06b}) we obtain
292: \begin{eqnarray}
293: U =\displaystyle{e^{-\frac{1}{2}\bigg(E+2\mbox{erf}^{-1}(z)^2\bigg)}},
294: \label{eq07a}
295: \end{eqnarray}
296: where $\displaystyle{z=\frac{2e^{\frac{E}{2}}(t_0+it)}{\sqrt{\pi}}}$ and
297: $t_0$ is the second arbitrary integration constant and {\it erf} is the error
298: function
299: \cite{Abramowitz}.
300:
301: Substituting the equation (\ref{eq07a}) into equation (\ref{eq05})
302: we get the general solution for the equation (\ref{eq01}), with the
303: parametric choice $\beta=\frac{\alpha^2}{8}$, in the form (after some
304: modifications)
305: \begin{eqnarray}
306: x(t) =\displaystyle{\frac{8}{\alpha}\mbox{erf}^{-1}(\bar{z})
307: e^{\frac{1}{2}\bigg(E+2\mbox{erf}^{-1}(\bar{z})^2\bigg)}}, \;\;
308: \bar{z}=\frac{2e^{\frac{E}{2}}(t_0-t)}{\sqrt{\pi}}.
309: \label{eq07}
310: \end{eqnarray}
311: In Fig. 1, we have plotted the solution of MEE (\ref{eq01}) given by the
312: expression (\ref{eq07}) for the parametric choice $\alpha^2=8\beta$ for
313: different initial conditions.
314: \begin{figure}[!ht]
315: \begin{center}
316: \includegraphics[width=.5\linewidth]{figerf}
317: \caption{Solution plot for case $\alpha^2=8\beta$ for different
318: initial conditions.}
319: \end{center}
320: \end{figure}
321:
322:
323: \subsection{Case 2: $\alpha^2>8\beta$}
324: \label{sec23}
325: Now let us consider the Hamiltonian for the case $\alpha^2>8\beta$ from
326: (\ref{mlin16}), that is,
327: \begin{eqnarray}
328: H=\displaystyle{\frac{(r-1)}{(r-2)}(p)^{\frac{r-2}{r-1}}
329: -\frac{(r-1)}{2r}\alpha x^2p}.\label{com03}
330: \end{eqnarray}
331:
332: Interestingly, here we identify a canonical transformation for the
333: Hamiltonian (\ref{com03}) in the form
334: \begin{eqnarray}
335: &x=-a\frac{P}{U},\qquad p=\frac{ U^2}{2a},
336: \label {com05}
337: \end{eqnarray}
338: where $a=\frac{2r}{\alpha(1-r)}$. It is straightforward to check that when $U$ and
339: $P$ are canonical so do $x$
340: and $p$ (and vice versa) so that the Hamiltonian $H$ in Eq.~(\ref{com03}) can
341: be rewritten as
342: \begin{eqnarray}
343: &H=\frac{1}{2}P^2-k U^{m}\equiv E,
344: \label{com06}
345: \end{eqnarray}
346: where $\displaystyle{m=\frac{2(r-2)}{(r-1)}}$ and
347: $\displaystyle{k=\frac{(1-r)}{(2a)^{\frac{m}{2}} (r-2)}}$. From the canonical
348: equations then we have $\dot{U}=P$ and $\dot{P}=mkU^{m-1}$.
349:
350: To obtain the general solution,
351: we introduce a transformation $X=U^m$ in equation (\ref{com06})
352: so that the latter can be brought to a quadrature of the form
353: \begin{eqnarray}
354: t-t_0=\int \frac{X^{\frac{1-m}{m}}dX}{m\sqrt{2E+2kX}}, \;\; m\neq0,2,4.
355: \label{int01}
356: \end{eqnarray}
357: Now fixing $\frac{1-m}{m}=n$, the above integral leads us to the following
358: expression \cite{Gradshteyn}, that is,
359: \begin{eqnarray}
360: \fl t-t_0=\int \frac{X^{n}dX}{m\sqrt{2E+2kX}}
361: %\nonumber\\
362: =\frac{1}{m}\bigg(\frac{X^n\sqrt{2E+2kX}}{2k(n+\frac{1}{2})}
363: -\frac{2En}{2k(n+\frac{1}{2})}\int \frac{X^{n-1}dX}{\sqrt{2E+2kX}}\bigg).
364: \label{int02}
365: \end{eqnarray}
366: On the other hand fixing $\frac{1-m}{m}=-n$ in (\ref{int01}), we end up with the
367: following expression \cite{Gradshteyn}
368: \begin{eqnarray}
369: \fl \qquad\qquad t-t_0&=&\int \frac{dX}{mX^{n}\sqrt{2E+2kX}}
370: \nonumber\\
371: &=&\frac{1}{m}\bigg(-\frac{\sqrt{2E+2kX}}{2EX^{n-1}(n-1)}
372: -\frac{k(n-\frac{3}{2})}{E(n-1)}\int \frac{dX}{X^{n-1}\sqrt{2E+2kX}}\bigg).
373: \label{int03}
374: \end{eqnarray}
375:
376: For integer values of $n$ the integrals (\ref{int02}) and (\ref{int03}) can be
377: integrated explicitly by repeated use of these formulas. When $n$ is a
378: noninteger value the final term in the
379: integrals (\ref{int02}) and (\ref{int03}) can be integrated in terms of
380: logarithmic function or beta function \cite{Gradshteyn}
381: (since the value of $n-1$ in the final integral lies between 0 to 1). We also
382: note here that for the special choices
383: $n=0$ or $-\frac{3}{2}$ (which corresponds to the case
384: $\beta=\frac{\alpha^2}{9}$) and $n=-\frac{2}{3}$ or $-\frac{5}{6}$ (which
385: corresponds to the case $\beta=-\alpha^2$) respectively in equations
386: (\ref{int02}) and (\ref{int03}), one can get the respective solutions reported
387: in the literature \cite{Painleve,Ince,Davis,
388: Kamke,Murphi,mahomed:1985,Duarte2,Bouquet,Sarleti,leach:1988a,Steeb,feix:1997,
389: Ibragimov1,Leach1,Chand,Chand1,Euler}.
390:
391: \subsection{Case 3: $\alpha^2<8\beta$}
392: \label{sec24}
393: Finaly, let us focus our attention in the regime $\alpha^2<8\beta$. In this
394: parametric regime the Hamiltonian takes the form (vide equation (\ref{mlin16}))
395: \begin{eqnarray}
396: H=\displaystyle{\frac{1}{2}\log[x^4 sec^2(\frac{\omega}{4} x^2p)]
397: -\frac{\alpha }{4}x^2p}.\label{com03a}
398: \end{eqnarray}
399: For the present case, let us choose the canonical transformation
400: in the form
401: \begin{eqnarray}
402: &x=\frac{U}{P},\qquad\;\; p=\frac{P^2}{2},\label {com05aa}
403: \end{eqnarray}
404: and transform the Hamiltonian to
405: \begin{eqnarray}
406: &H=\displaystyle{\frac{1}{2}\log[\frac{U^4}{P^4}sec^2(\frac{\omega}{8} U^2)]
407: -\frac{\alpha }{8}U^2} \equiv E.\label{com06a}
408: \end{eqnarray}
409:
410: Now making use of the canonical equations, $\dot{U}=-\frac{2}{P}$ and
411: $\dot{P}=\frac{1}{4U}((\alpha-\omega\tan(\frac{\omega}{8} U^2))U^2-8)$,
412: equation (\ref{com06a}) can be rewritten as
413: \begin{eqnarray}
414: E=\frac{1}{2}\log[\frac{\dot{U}^4U^4}{16}sec^2(\frac{\omega}{8} U^2)]
415: -\frac{\alpha }{8}U^2.
416: \label{int04}
417: \end{eqnarray}
418: Introducing now the transformation $V=\frac{U^2}{2}$ in (\ref{int04}) we arrive
419: at
420: \begin{eqnarray}
421: E=\frac{1}{2}\log[\frac{\dot{V}^4}{16}sec^2(\frac{\omega}{4} V)]
422: -\frac{\alpha }{4}V.
423: \label{int04a}
424: \end{eqnarray}
425: Integrating the equation (\ref{int04a}) we get
426: \begin{eqnarray}
427: t-t_0=\frac{e^{-\frac{E}{2}}}{2}\int \sqrt{sec(\frac{\omega}{4} V)}
428: e^{\displaystyle{-\frac{\alpha }{8}V}} dV,
429: \label{int05}
430: \end{eqnarray}
431: where $t_0$ is the second integration constant. By substituting $W=e^V$
432: in (\ref{int05}) we obtain
433: \begin{eqnarray}
434: \fl \quad t-t_0=\frac{e^{-\frac{E}{2}}}{\sqrt{2}}
435: \int \frac{W^{q_1-1}dW}{\sqrt{1+W^{q_2}}}
436: =\frac{e^{-\frac{E}{2}}W^{q_1}}{\sqrt{2}q_1} F(\frac{1}{2},\frac{q_1}{q_2};
437: 1+\frac{q_1}{q_2};-W^{q_2})
438: \label{int05a}
439: \end{eqnarray}
440: where $q_1=\frac{-\alpha+i\omega}{8},
441: \;q_2=\frac{i\omega}{2}$ and $F(\alpha,\beta;\gamma;z)$
442: is the hypergeometric function \cite{Abramowitz,Gradshteyn}.
443:
444: \section{Connection to two other nonlinear models}
445: \label{sec4}
446: In this section, we show that the MEE (\ref{eq01}) is intimately connected
447: with two other
448: well known nonlinear models, namely, the force-free Duffing oscillator type
449: equation and two dimensional Lotka-Volterra (LV) equation.
450: \subsection{Force-free Duffing oscillator type equation}
451: \label{sec31}
452: The MEE equation (\ref{eq01}) can be transformed to the following
453: oscillator equation
454: \begin{eqnarray}
455: w''+(\alpha w+\gamma)w'+\beta w^3+\frac{\alpha \gamma}{3}w^{2}
456: +\frac{2\gamma^2}{9}w=0,\;\;\;('=\frac{d}{d\tau})
457: \label {eq08}
458: \end{eqnarray}
459: through the invertible point transformation $x = we^{\frac{\gamma}{3}\tau},
460: t = -\frac {3}{\gamma}e^{-\frac{\gamma}{3}\tau}$, where $\gamma$ is an arbitrary
461: parameter. Equation (\ref{eq08}) includes
462: force-free Duffing oscillator (in the specific case $\alpha=0$)
463: \cite{Chand1,Bluman,Laks1} and quadratic oscillator with $\beta=0$
464: \cite{Chand1}.
465:
466: One can deduce the general solution of the equation (\ref{eq08})
467: from the general solution of MEE. For example, in the parametric choice
468: $\beta=\frac{\alpha^2}{8}$ the general solution of the equation (\ref{eq08})
469: can be derived from (\ref{eq07}) in the form
470: \begin{eqnarray}
471: w(\tau) =\frac{8}{\alpha}\mbox{erf}^{-1}(\hat{z})
472: e^{\frac{1}{2}\bigg(E-\frac{2}{3}\gamma\tau+2\mbox{erf}^{-1}(\hat{z})^2\bigg)},
473: \label{eq09}
474: \end{eqnarray}
475: where now $\displaystyle{\hat{z}=\frac{6e^{\frac{E}{2}}(\tau_0+e^{-\frac{\gamma}{3}\tau})}
476: {\gamma\sqrt{\pi}}}$ and $E$ and $\tau_0$ are arbitrary integration constants.
477: Similarly, one can fix the general solution for the equation (\ref{eq08}) in the
478: other parametric regimes also, that is, $\alpha^2>8\beta$ and $\alpha^2<8\beta$.
479: \subsection{Two dimensional Lotka-Volterra equation}
480: \label{sec32}
481: Let us consider the two dimensional Lotka-Volterra equation of the form
482: %\begin{subequations}
483: \begin{eqnarray}
484: \dot {x}=x(a_1+a_2 x+a_3 y),\quad
485: %\label {eq10a}\\
486: \dot {y}=y(b_1+b_2x+b_3y),
487: %\label {eq10b}
488: \label{eq10}
489: \end{eqnarray}
490: %\end{subequations}
491: where $a_i$'s and $b_i$'s are six real parameters.
492: Equation (\ref{eq10}) models two species in competition in ecology and is
493: being analyzed for the past three decades or so in mathematical biology
494: \cite{Murray,Cairo1,Cairo2}. Interestingly, equation (\ref{eq10}) can also be
495: transformed to (\ref{eq08}) as follows.
496: Rewriting the first equation in (\ref{eq10}) for the variable $y$ and substituting
497: it into the second equation in equation (\ref{eq10}) we get the following second order
498: ODE for $x$, namely,
499: \begin{eqnarray}
500: \fl \qquad\ddot{x}-(1+\frac{b_3}{a_3})\frac{\dot{x}^{2}}{x}+
501: ((2a_2\frac{b_3}{a_3}-a_{2}-b_2)x+(2a_{1}\frac{b_3}
502: {a_3}-b_1))\dot{x}
503: \nonumber\\
504: \fl\qquad\quad+(b_2a_{2}-\frac{b_3}{a_3}a_2^{2})x^{3}
505: +(a_2 b_1+b_2a_{1}-2a_{1}a_2\frac{b_3}{a_3})x^{2}
506: +(a_1b_1-\frac{b_3}{a_3}a_{1}^{2})x=0.
507: \label {eq11}
508: \end{eqnarray}
509: Let us choose the parameters in (\ref{eq11}) in the form $b_3=-a_3$
510: and $b_1=a_1$ so that equation (\ref{eq11}) can be brought to the form
511: \begin{eqnarray}
512: \fl\qquad\ddot{x}-\left((3a_2+b_2)x+3a_1\right)\dot{x}
513: +a_2(a_2+b_2)x^{3}+a_1(3a_2+b_2)x^{2}+2a_1^2x=0. \label {eq12}
514: \end{eqnarray}
515: The associated LV equation takes the form
516: %\begin{subequations}
517: \begin{eqnarray}
518: \dot {x}=x(a_{1}+a_2x+a_3y),\quad
519: %\label {eq13a}\\
520: \dot {y}=y(a_{1}+b_2x-a_3y).
521: %\label {eq13b}
522: \label{eq13}
523: \end{eqnarray}
524: %\end{subequations}
525: Now comparing the equations (\ref{eq08}) and (\ref{eq12}) we obtain
526: $\alpha=-(3a_2+b_2)$, $\beta=a_2(a_2+b_2)$ and
527: $\gamma=-3a_1$. Choosing $a_2=b_2$ ($\beta=\frac{\alpha^2}{8}$) the
528: general
529: solution for the LV equation (\ref{eq12}) can be obtained from (\ref{eq09})
530: and using this in the first equation in (\ref{eq13}) we arrive the general
531: solution for the LV equation (\ref{eq13}) in the form
532: \begin{eqnarray}
533: \fl \qquad x(t) =-\frac{2}{a_{2}}\mbox{erf}^{-1}(z)
534: e^{\frac{1}{2}\bigg(E+2a_1t+2\mbox{erf}^{-1}(z)^2\bigg)},\quad
535: %\nonumber\\
536: y(t) = -\frac{e^{\frac{1}{2}\bigg(E+2a_1t+2\mbox{erf}^{-1}(z)^2\bigg)}}
537: {a_3\mbox{erf}^{-1}(z)},
538: \label{eq14}
539: \end{eqnarray}
540: where $\displaystyle{z=\frac{2e^{\frac{E}{2}}(t_0-e^{a_1t})}{a_1\sqrt{\pi}}}$
541: and $E$ and $t_0$ are arbitrary integration constants.
542: Finally, we mention that the general solution
543: for the equation (\ref{eq12}) for other parametric choices can be obtained
544: from equation (\ref{eq08}) in
545: a similar manner. Again (to our knowledge) the complete integrability of the LV system
546: (\ref{eq13}) in these regimes is new to the literature.
547: \section{Conclusions}
548: \label{sec5}
549: In this paper, we have shown that the MEE (\ref{eq01}) is integrable either
550: explicitly or by quadratures for any value of $\alpha$ and $\beta$.
551: We have also obtained the
552: time independent Hamiltonians for
553: the equation (\ref{eq01}). We have transformed the Hamiltonians into simpler
554: forms, with appropriate canonical transformations, and deduced the general
555: solution by direct integration so that our approach
556: helps to understand the dynamics of equation (\ref{eq01}) in phase-space
557: clearly.
558: Further, we have demonstrated that the complete integrability of
559: equation (\ref{eq01}) also helps one to understand the dynamics of two other
560: nonlinear models, namely, the generalized oscillator equation and the two
561: dimensional LV equation. As a consequence the
562: solutions which we have explored for the equation (\ref{eq01}) also provide
563: solutions for these models as well.
564:
565: \section*{Acknowledgment}
566: The work of VKC is supported by CSIR in the form of a CSIR Senior Research
567: Fellowship. The work of ML forms part of a Department of Science and
568: Technology, Government of India sponsored research project and is supported by a
569: Department of Atomic Energy Raja Ramanna Fellowship.
570:
571: \appendix
572:
573: \section{Method of deriving integrals of motions}
574: \label{sec6}
575: In the following we briefly explain the generalized extended or modified
576: Prelle-Singer (PS) procedure for second order ODEs \cite{Chand,Chand1,Duarte}
577: which is used to identify the integrals of motions (\ref{intm03})-(\ref{intm01}).
578:
579: To begin with, let us rewrite equation~(\ref{eq01}) in the form
580: \begin{eqnarray}
581: \ddot{x}=-(\alpha x\dot{x}+\beta x^3)\equiv \phi(x,\dot{x}).
582: \label{eq01a}
583: \end{eqnarray}
584: Further, we assume that the ODE (\ref{eq01a})
585: admits a first integral $I(t,x,\dot{x})=C,$ with $C$ constant on the
586: solutions, so that the total differential becomes
587: \begin{eqnarray}
588: dI={I_t}{dt}+{I_{x}}{dx}+{I_{\dot{x}}{d\dot{x}}}=0,
589: \label{met3}
590: \end{eqnarray}
591: where each subscript denotes partial differentiation with respect
592: to that variable. Rewriting equation~(\ref{eq01a}) in the form
593: $\phi dt-d\dot{x}=0$ and adding a null term
594: $S(t,x,\dot{x})\dot{x}$ $ dt - S(t,x,\dot{x})dx$ to the latter, we obtain that on
595: the solutions the 1-form
596: \begin{eqnarray}
597: (\phi +S\dot{x}) dt-Sdx-d\dot{x} = 0, \quad \phi=-(\alpha x\dot{x}+\beta x^3).
598: \label{met6}
599: \end{eqnarray}
600: Hence, on the solutions, the 1-forms (\ref{met3}) and
601: (\ref{met6}) must be proportional. Multiplying (\ref{met6}) by the
602: factor $ R(t,x,\dot{x})$ which acts as the integrating factors
603: for (\ref{met6}), we have on the solutions that
604: \begin{eqnarray}
605: dI=R(\phi+S\dot{x})dt-RSdx-Rd\dot{x}=0.
606: \label{met7}
607: \end{eqnarray}
608: Comparing Eq.~(\ref{met3})
609: with (\ref{met7}) we have, on the solutions, the relations
610: \begin{eqnarray}
611: I_{t} = R(\phi+\dot{x}S),\quad
612: I_{x} = -RS, \quad
613: I_{\dot{x}} = -R.
614: \label{met8}
615: \end{eqnarray}
616: Then the compatibility conditions,
617: $I_{tx}=I_{xt}$, $I_{t\dot{x}}=I_{{\dot{x}}t}$, $I_{x{\dot{x}}}=I_{{\dot{x}}x}$,
618: between the Eqs.~(\ref{met8}), provide us
619: \begin{eqnarray}
620: S_t+\dot{x}S_x+\phi S_{\dot{x}} &=&
621: -\phi_x+\phi_{\dot{x}}S+S^2,\label {lin02}\\
622: R_t+\dot{x}R_x+\phi R_{\dot{x}} & =&
623: -(\phi_{\dot{x}}+S)R,\label {lin03}\\
624: R_x-SR_{\dot{x}}-RS_{\dot{x}} &= &0.
625: \qquad \qquad\qquad \;\;\;\label {lin04}
626: \end{eqnarray}
627:
628: Solving equations~(\ref{lin02})-(\ref{lin04}) one can obtain expressions for $S$ and
629: $R$. It may be noted that any set of special solutions $(S,R)$ is sufficient for
630: our purpose. Once these forms are determined the integral of motion
631: $I(t,x,\dot{x})$ can be deduced from the relation
632: \begin{eqnarray}
633: I= r_1
634: -r_2 -\int \left[R+\frac{d}{d\dot{x}} \left(r_1-r_2\right)\right]d\dot{x},
635: \label{met13}
636: \end{eqnarray}
637: where
638: \begin{eqnarray}
639: r_1 = \int R(\phi+\dot{x}S)dt,\quad
640: r_2 =\int (RS+\frac{d}{dx}r_1) dx. \nonumber
641: \end{eqnarray}
642: Equation~(\ref{met13}) can be derived straightforwardly by integrating the
643: equation~(\ref{met8}).
644:
645: As our motivation is to explore time independent integral of motion for the
646: equation~(\ref{eq01a}) let us choose $I_t=0$.
647: In this case one can easily fix the null form $S$ from the first equation in
648: (\ref{met8}) as
649: \begin{equation}
650: S = \frac{-\phi}{\dot{x}}=
651: \frac{(\alpha x\dot{x}+\beta x^3)}{\dot{x}}.
652: \label{mlin06}
653: \end{equation}
654: Substituting this form of $S$ into (\ref{lin03}) we get
655: \begin{eqnarray}
656: \dot{x}R_x-(\alpha x\dot{x}+\beta x^3)R_{\dot{x}}
657: =-\frac{\beta x^3}{\dot{x}}R.
658: \label {lin03a}
659: \end{eqnarray}
660:
661: Equation~(\ref{lin03a}) is a first order linear partial differential equation
662: with variable coefficients.
663: As we noted earlier any particular solution is
664: sufficient to construct an integral of motion (along with the function $S$).
665: To seek
666: a particular solution for $R$ one can make a suitable ansatz instead of looking
667: for the general solution. We assume $R$ to be of the form,
668: \begin{equation}
669: R = \frac{\dot{x}}{(A(x)+B(x)\dot{x}+C(x)\dot{x}^2)^r},
670: \label{mlin08}
671: \end{equation}
672: where $A,\;B$ and $C$ are functions of their arguments, and $r$ is a constant
673: which are all to be
674: determined. {\it We demand the above form of ansatz (\ref{mlin08}), which is very
675: important to derive the Hamiltonian structure associated with the given
676: equation}, due to the following
677: reason. To deduce the first integral $I$ we assume a rational form for $I$,
678: that is,
679: $I=\frac{f(x,\dot{x})}{g(x,\dot{x})}$, where $f$ and $g$
680: are arbitrary functions of $x$ and $\dot{x}$ and are independent of $t$, from
681: which we get
682: $I_x=\frac{f_{x}g-fg_x}{g^2}$ and $I_{\dot{x}}=\frac{f_{\dot{x}}g-fg_{\dot{x}}}{g^2}$.
683: From (\ref{met8}) one can see that $R=I_{\dot{x}}=\frac{f_{\dot{x}}g-fg_{\dot{x}}}{g^2},
684: \;S=\frac{I_x}{I_{\dot{x}}}=\frac{f_xg-fg_x}{f_{\dot{x}}g-fg_{\dot{x}}}$ and
685: $RS=I_x$, so that the denominator of the function $S$ should be the numerator of the
686: function $R$. Since the denominater of $S$ is $\dot{x}$ (vide Eq.~(\ref{mlin06}))
687: we fixed the numerator of $R$ as $\dot{x}$. To seek a suitable function in the
688: denominator initially one can consider an
689: arbitrary form $R=\frac{\dot{x}}{h(x,\dot{x})}$. However, it is difficult to
690: proceed with this choice of $h$. So let us assume that $h(x,\dot{x})$ is a
691: function which is polynomial in
692: $\dot{x}$. To begin with let us consider the case where $h$ is quadratic in
693: $\dot{x}$,
694: that is, $h=A(x)+B(x)\dot{x}+C(x)\dot{x}^2$, which is a generalized version of
695: the form considered in Ref. \cite{Chand1}, where only the linear form in
696: $\dot{x}$ was investigated (that is $C(x)=0$). Since $R$ is in rational form while
697: taking differentiation or integration the form of the denominator remains same
698: but the power of the denominator decreases or increases by a unit order from
699: that of the initial one. So instead of considering $h$ to be of the form
700: $h=A(x)+B(x)\dot{x}+C(x)\dot{x}^2$, one may consider a more general form
701: $h=(A(x)+B(x)\dot{x}+C(x)\dot{x}^2)^r$, where $r$ is a constant to be
702: determined. The parameter $r$ plays an important role, as we see below.
703:
704: Substituting (\ref{mlin08}) into (\ref{lin03a}) and solving the resultant
705: equations, we arrive at the relation
706: \begin{eqnarray}
707: \fl \quad r\bigg[\dot{x}(A_x+B_x\dot{x}+C_x\dot{x}^2)
708: -(\alpha x\dot{x}+\beta x^3)(B+2C\dot{x})\bigg]
709: =-\alpha x(A+B\dot{x}+C\dot{x}^2).
710: \label{mlin09}
711: \end{eqnarray}
712: Solving equation~(\ref{mlin09}), we can fix the forms of $A,\;B,\;C$ and $r$ and
713: substituting them into equation~(\ref{mlin08}) we can get the integrating factor
714: $R$. Doing so, we find
715: \begin{eqnarray}
716: R=\left\{
717: \begin{array}{ll}
718: \displaystyle{\frac{\dot{x}}{(\dot{x}+\frac{(r-1)}{2r}\alpha x^2)^{r}}},
719: &\;\;\alpha^2\geq8\beta \\
720: \displaystyle{\frac{\dot{x}}{(2\dot{x}^2+\alpha x^2\dot{x}+\beta x^4)}},
721: &\;\;\alpha^2 <8\beta \\
722: \displaystyle{\dot{x}}, & \;\; \alpha=0
723: \end{array}\right.\label{lam107}
724: \end{eqnarray}
725: where $r=\frac{\alpha}{4\beta}\bigg[\alpha\pm\sqrt{\alpha^2-8\beta}\bigg]$.
726: One can easily check the functions $S$ and $R$ given in (\ref{mlin06}) and
727: (\ref{lam107}), respectively, satisfy (\ref{lin04}) also. Finally, substituting
728: $R$ and $S$ into the form (\ref{met13}) for the integral we get the integrals of
729: motion (\ref{intm03})-(\ref{intm01}).
730:
731:
732: \section*{References}
733:
734: \begin{thebibliography}{11}
735: %\Bibliography{11}
736: \bibitem{Painleve}
737: Painlev\'e P 1902 \emph{Acta Math.} {\bf 25} 1
738:
739: \bibitem{Ince}
740: Ince E L 1956 \emph{Ordinary Differential Equations} (New York: Dover)
741:
742: \bibitem{Davis}
743: Davis H T 1962 \emph{Introduction to Nonlinear Differential and Integral
744: Equations} (New York: Dover)
745:
746: \bibitem{Kamke}
747: Kamke E 1983 \emph{Differentialgleichungen Losungsmethoden und Losungen}
748: (Stuggart: Teubner)
749:
750: \bibitem{Murphi}
751: Murphy G M 1960 \emph{Ordinary Differential Equations and Their Solutions}
752: (New York: D. Van Nostrand)
753:
754: \bibitem{Gobulev}
755: Gobulev V V 1950 \emph{Lectures on Analytical Theory of Differential Equations}
756: (Moscow: Gostekhizdat)
757:
758: \bibitem{Chisholm}
759: Chisholm J S R and Common A K 1987 \emph{J. Phys. A: Math. Gen.} {\bf20}
760: 5459-5472
761:
762: \bibitem{Moreira}
763: Moreira I C 1984 \emph{Hadronic. J.} {\bf 7} 475; Leach P G L 1985
764: \emph{ J.~Math.~Phys.~} {\bf 26} 2510
765:
766: \bibitem{Dix:1990}
767: Chandrasekhar S 1957 \emph{ An Introduction to the Study of Stellar
768: Structure} (New York: Dover);
769: Dixon J M and Tuszynski J A 1990 \emph{ Phys. Rev. A} {\bf41} 4166
770:
771: \bibitem{Erwin}
772: Erwin V J, Ames W F and Adams E 1984 \emph{Wave Phenomena: Modern Theory and
773: Applications} ed C Rogers and J B Moodie (Amsterdam: North-Holland)
774:
775: \bibitem{McVittie}
776: McVittie G C 1933 \emph{Mon. Not. R. Astron. Soc.} {\bf93}
777: 325; 1967 \emph{Ann. Inst. H Poincar\'e} {\bf 6} 1; 1984
778: \emph{Ann. Inst. H Poincar\'e} {\bf 40} 3, 231
779:
780: \bibitem{Yang}
781: Yang C N and Mills R L 1954 \emph{Phys. Rev.} {\bf96} 191
782:
783: \bibitem{mahomed:1985}
784: Mahomed F M and Leach P G L 1985 \emph{Quaestiones Math. } {\bf 8} 241;
785: 1989 {\bf 12} 121
786:
787: \bibitem{Duarte2}
788: Duarte L G S, Duarte S E S and Moreira I C 1987 \emph{J. Phys. A: Math. Gen.}
789: {\bf 20} L701
790:
791: \bibitem{Bouquet}
792: Bouquet S E, Feix M R and Leach P G L 1991 \emph{J. Math. Phys.}
793: {\bf 32} 1480
794:
795: \bibitem{Sarleti}
796: Sarlet W, Mahomed F M and Leach P G L 1987
797: \emph{J. Phys. A: Math. Gen.} {\bf 20} 277
798:
799: \bibitem{leach:1988a}
800: Leach P G L, Feix M R and Bouquet S 1988 \emph{J. Phys. A: Math. Gen.}
801: {\bf 29} 2563; Lemmer R L and Leach P G L 1993 \emph{J. Phys. A: Math. Gen.}
802: {\bf 26} 5017
803:
804: \bibitem{Steeb}
805: Steeb W H 1993 \emph{Invertible Point Transformations and Nonlinear
806: Differential Equations} (London: World Scientific)
807:
808: \bibitem{feix:1997}
809: Feix M R, Geronimi C, Cairo L, Leach P G L, Lemmer R L and Bouquet S 1997
810: \emph{J. Phys. A: Math. Gen.} {\bf 30} 7437
811:
812: \bibitem{Ibragimov1}
813: Ibragimov N H 1999 \emph{Elementary Lie Group Analysis and Ordinary
814: Differential Equations} (New York: John Wiley \& Sons)
815:
816: \bibitem{Leach1}
817: Leach P G L Cotsakis S and Flessas G P 2000 \emph{J. Math. Anal. Appl.}
818: {\bf 251} 587
819:
820: \bibitem{Chand}
821: Chandrasekar V K, Senthilvelan M and Lakshmanan M 2005
822: \emph{Proc. R. Soc. London } {\bf A461} 2451; \emph{Phys. Rev.}
823: {\bf E72} 066203; \emph{Chaos, Solitons and Fractals} {\bf 26} 1399
824:
825: \bibitem{Chand1}
826: Chandrasekar V K, Pandey S N, Senthilvelan M and Lakshmanan M 2006
827: \emph{J. Math. Phys.} {\bf 47} 023508
828:
829: \bibitem{Euler}
830: Euler M, Euler N and Leach P G L 2005/2006 \emph{The Riccati and Ermakov-Pinney
831: hierarchies}, Report no. {\bf08}, Institut Mittag-Leffler, Sweden
832:
833: \bibitem{Bluman}
834: Bluman G W and Anco S C 2002 \emph{Symmetries and Integration Methods
835: for Differential Equations} (New York: Springer-Verlag)
836:
837: \bibitem{Laks1}
838: Lakshmanan M and Rajasekar S 2003 \emph{Nonlinear Dynamics: Integrability
839: Chaos and Patterns} (New York: Springer-Verlag)
840:
841: \bibitem{Murray}
842: Murray J D 1989 \emph{Mathematical Biology} (New York: Springer-Verlag)
843:
844: \bibitem{Cairo1}
845: Cairo L and Feix M R 1992 \emph{J. Math. Phys.} {\bf 33} 2440;
846: Almeida M A, Magalhaes M E and Moreira I C 1995 \emph{J. Math. Phys.}
847: {\bf 36} 1854
848:
849: \bibitem{Cairo2}
850: Hua D D, Cairo L, Feix M R, Govinder K S and Leach P G L 1996
851: \emph{Proc. R. Soc. London A} {\bf 452} 859;
852: Cairo L, Feix M R and Llibre J 1999 \emph{J. Math. Phys.}
853: {\bf 40} 2074
854:
855: \bibitem{Duarte}
856: Duarte L G S, Duarte S E S, da Mota A C P and J E F Skea 2001
857: \emph{J. Phys. A} {\bf 34} 3015
858:
859: \bibitem{Abramowitz}
860: Abramowitz M and Stegun I A, eds. 1972 \emph{Handbook of Mathematical
861: Functions with Formulas, Graphs, and Mathematical Tables} (New York:
862: Dover)
863:
864: \bibitem{Gradshteyn}
865: Gradshteyn I S and Ryzhik I M 1980 \emph{Table of Integrals, Series and
866: Products} (London: Academic press)
867:
868:
869: \end{thebibliography}
870: \end{document}
871:
872:
873:
874: