1: \documentclass[10pt]{article}
2:
3: \usepackage{graphicx}
4: %\usepackage{showkeys}
5: \usepackage{amsmath}
6: \usepackage{amssymb}
7:
8: \numberwithin{equation}{section}
9: \setlength\arraycolsep{2pt}
10:
11: \newtheorem{thm}{Theorem}[section]
12: \newenvironment{prf}{\noindent {\it Proof} \ }{\hfill $\Box$}
13: \newenvironment{rmk}{\noindent {\it Remark} \ }{}
14: \newtheorem{prop}[thm]{Proposition}
15: \newtheorem{lemma}[thm]{Lemma}
16: \newtheorem{defn}[thm]{Definition}
17: \newtheorem{cor}[thm]{Corollary}
18: \newcommand{\eqa}{\begin{eqnarray}}
19: \newcommand{\eeqa}{\end{eqnarray}}
20: \newcommand{\beq}{\begin{equation}}
21: \newcommand{\eeq}{\end{equation}}
22: \newcommand{\nn}{\nonumber}
23: \newcommand{\pp}{\partial}
24: \newcommand{\VE}{\varepsilon}
25: \newcommand{\res}{{\rm res}}
26: \newcommand{\lm}{\lambda}
27: \newcommand{\al}{{\alpha}}
28: \newcommand{\pf}{\noindent{\it Proof \ }}
29: \newcommand{\epf}{$\quad$\hfill
30: \raisebox{0.11truecm}{\fbox{}}\par\vskip0.4truecm}
31: \setlength{\headheight}{0.0in}
32: \setlength{\headsep}{0.0in}
33: \newcommand\hb{\newline}
34: \newcommand\ve{\epsilon}
35: \newcommand\pal{\partial}
36: \textheight 22.5 truecm
37: \textwidth 15.5truecm
38: \begin{document}
39: %\baselineskip16.2pt
40: \parskip 6pt
41: \hoffset -1.8cm
42:
43: \title{A 2-Component Generalization of the Camassa-Holm Equation and Its Solutions }
44: \author{Ming Chen, Si-Qi Liu, Youjin Zhang\\
45: {\small Department of Mathematical Sciences, Tsinghua
46: University}\\
47: {\small Beijing 100084, P.R. China}}
48: \date{}
49: \maketitle
50:
51: \begin{abstract}
52: An explicit reciprocal transformation between a 2-component
53: generalization of the Camassa-Holm equation, called the 2-CH
54: system, and the first negative flow of the AKNS hierarchy is
55: established, this transformation enables one to obtain solutions
56: of the 2-CH system from those of the first negative flow of the
57: AKNS hierarchy. Interesting examples of peakon and multi-kink
58: solutions of the 2-CH system are presented.
59: \vskip 0.5 cm
60: \noindent {\bf Mathematics Subject Classifications(2000)}. 35Q53, 37K35\newline
61: {\bf Key words}: Camassa-Holm equation, AKNS hierarchy, reciprocal transformation
62: \end{abstract}
63:
64: \section{Introduction}
65:
66: The Camassa-Holm equation, which was derived physically as a shallow water wave equation
67: by Camassa and Holm in \cite{CH, CHH}, takes the form
68: \beq\label{CH}
69: u_t+\kappa\, u_x-u_{xxt}+3 u u_x=2 u_x u_{xx}+u u_{xxx}
70: \eeq
71: where $u=u(x,t)$ is the fluid velocity in the $x$ direction and the constant $\kappa$ is related to the critical
72: shallow water wave speed. The subscripts $x, t$ of $u$ denote the partial derivatives of the function $u$ w.r.t. $x, t$,
73: for example, $u_t=\pal_t u,\ u_{xxt}=\pal_t\pal_x\pal_x u$, similar notations will be
74: used frequently later in this paper.
75: This equation
76: first appeared in the work of Fuchssteiner and Fokas \cite{FF} on their theory of hereditary symmetries
77: of soliton equations.
78: As it was shown by Camassa and Holm, equation (\ref{CH}) shares most of the important properties of an
79: integrable system of KdV type, for example, the existence of Lax pair formalism, the
80: bihamiltonian structure, the multi-soliton solutions and the applicability of
81: the inverse scattering method to its initial value problem. When $\kappa=0$, the Camassa-Holm equation
82: (\ref{CH}) has a peculiar property that its soliton
83: solutions become piecewise smooth and have corners at their crests, such solutions are
84: weak solutions of (\ref{CH}) and are called ``peakons''. Since the works of Camassa and Holm,
85: this equation has become a well known example of integrable systems and has been studied
86: from various point views in, for example, \cite{AG, Alb, Cons-1, hone-1, Fo, Fu, hone,johnson,mckean-1, mckean-2, schiff}
87: and references therein.
88:
89: We consider in this paper the following 2-component generalization of the Camassa-Holm equation
90: \begin{eqnarray}
91: &&m_t+u\,m_x+2\,m\,u_x-\rho\,\rho_x=0,\label{ch-2}\\
92: &&\rho_t+(\rho\,u)_x=0.\label{ch-3}
93: \end{eqnarray}
94: Here $m=u-u_{xx}+\frac12\kappa$. Under the constraint $\rho=0$ this system is reduced to the Camassa-Holm
95: equation (\ref{CH}). Such a generalization is based on the following consideration. We note the fact
96: that both bihamiltonian structures of the Camassa-Holm hierarchy and that of the KdV hierarchy
97: are deformations of the following bihamiltonian structure
98: of hydrodynamic type
99: \begin{eqnarray}
100: &&\{u(x),u(y)\}_1=\delta'(x-y),\nn\\
101: &&\{u(x),u(y)\}_2=u(x)\delta'(x-y)+\frac12\,u(x)'\delta(x-y).\label{z8}
102: \end{eqnarray}
103: This fact also implies that the dispersionless limit of the Camassa-Holm hierarchy coincides with that of the
104: KdV hierarchy. It was shown in \cite{DLZ, LZ1} that deformations of the bihamiltonian structure (\ref{z8})
105: that depend polynomially on the variables $u_x, u_{xx},\dots$ are uniquely characterized,
106: up to Miura type transformations,
107: by a function
108: $c(u)$, this function is called the central invariant of the deformed bihamiltonian structure. For
109: the KdV hierarchy the central invariant is a nonzero constant while for the
110: Camassa-Holm hierarchy the central invariant is given by a nonzero constant multiplied by $u$, so their
111: bihamiltonian structures
112: are representatives of two different classes of deformations of (\ref{z8}). One of the main features
113: of the integrable hierarchies that correspond to bihamiltonian structures with constant central invariants
114: is the existence of tau functions \cite{DZ}, this property no longer holds true for integrable hierarchies
115: that correspond to bihamiltonian structures with nonconstant central invariants.
116: In fact, many of the well known integrable hierarchies of evolutionary PDEs
117: with one spatial variable possess bihamiltonian structures that are deformations of
118: bihamiltonian structure of hydrodynamic type with constant central invariants, and the existence
119: of tau functions plays an important role in the study of these integrable systems. The Camassa-Holm
120: hierarchy is an exceptional example of integrable systems which does not possess tau functions.
121: Now let us consider the following bihamiltonian structure of hydrodynamic type
122: \begin{eqnarray}
123: &&\{w_1(x), w_1(y)\}_1=\{w_2(x), w_2(y)\}_1=0,\nn\\
124: &&\{w_1(x), w_2(y)\}_1=\delta'(x-y).\label{z10-a}\\
125: &&\{w_1(x), w_1(y)\}_2=2 \delta'(x-y),\nn\\
126: &&\{w_1(x), w_2(y))\}_2=w_1(x) \delta'(x-y)+w_1'(x) \delta(x-y),\nn\\
127: &&\{w_2(x), w_2(y)\}_2=\left[w_2(x)\pal_x+\pal_x w_2(x))\right] \delta(x-y).\label{z10}
128: \end{eqnarray}
129: It was shown in \cite{LZ1} that the deformation with constant central invariants
130: $c_1=c_2=\frac1{24}$ leads to the nonlinear Schr\"odinger hierarchy which can be converted to the AKNS
131: hierarchy \cite{akns} after an appropriate transformation, and the one with central invariants
132: $c_1=\frac1{24}\,{(w_1+2\sqrt{w_2})^2}, c_2=\frac1{24}\,{(w_1-2\sqrt{w_2})^2}$ leads to a
133: bihamiltonian integrable hierarchy with (\ref{ch-2}),
134: (\ref{ch-3}) as the first nontrivial flow under the change of coordinates
135: \beq
136: w_1=2(u-u_x),\quad w_2=-\rho^2+(u-u_x)^2,
137: \eeq
138: the rescaling $t\mapsto -t$ and the Galilean transformation $x\mapsto x-\frac12\kappa\,t,\ t\mapsto t,\
139: u\mapsto u-\frac12 \kappa,\
140: \rho\mapsto \rho$.
141: So from the point of view of deformations of bihamiltonian structures of hydrodynamic type the systems (\ref{CH})
142: and
143: (\ref{ch-2}), (\ref{ch-3}) have the same property, i.e., both of their bihamiltonian structures have nonconstant central invariants.
144: Note that the bihamiltonian structure of the system (\ref{ch-2}), (\ref{ch-3}), as it was shown in \cite{LZ1},
145: is obtained from (\ref{z10-a}), (\ref{z10})
146: by the addition of the deformation
147: term $-\delta''(x-y)$ to the bracket $\{w_1(x), w_2(y)\}_1$. We will call the system (\ref{ch-2}), (\ref{ch-3}) the
148: 2-component Camassa-Holm (2-CH) system henceforth. This system was also derived independently by Falqui \cite{falqui}
149: by using the bihamiltonian approach.
150:
151: The main result of the present paper is the establishment of a reciprocal transformation between the 2-CH system and
152: the first negative flow of the AKNS hierarchy. Recall that
153: the Camassa-Holm equation (\ref{CH}) has a similar relation with the first negative flow of the KdV hierarchy,
154: the corresponding reciprocal transformation (also called a hodograph transformation) was found by
155: Fuchssteiner in \cite{Fu}. Several attempts have been made to obtain solutions
156: of the Camassa-Holm equation (\ref{CH}) from that of the first negative flow of the KdV hierarchy by using
157: this reciprocal transformation in, for example, \cite{Cons-1, hone, johnson,
158: LZJ1, LZJ2, schiff}. However, since the
159: inverse of this reciprocal transformation involves the solving of a nonlinear ODE of second order,
160: only particular solutions like the multi-soliton
161: solutions were obtained in explicit forms by using this approach.
162: The advantage of the reciprocal transformation between the 2-CH system and
163: the first negative flow of the AKNS hierarchy is that it gives an explicit correspondence between solutions of
164: these two systems, this correspondence is presented in Theorems \ref{fuv}, \ref{fx} and Theorems \ref{qrf},
165: \ref{fqr}. In Sec.\,\ref{sec-2} and Sec.\,\ref{sec-3}
166: we first give the construction of the
167: reciprocal transformation, then in Sec.\,\ref{sec-4}
168: we show some interesting examples of solutions of the 2-CH system that are obtained from solutions of the
169: first negative flow of the AKNS hierarchy by using the reciprocal transformation, they
170: include peakon and multi-kink solutions.
171:
172:
173: \section{A reciprocal transformation for the 2-CH system }\label{sec-2}
174:
175: The 2-CH system is equivalent to the compatibility conditions of the following
176: linear systems:
177: \eqa
178: &&\phi_{xx}+\left(-\frac14+m\,\lambda-\rho^2\,\lambda^2\right)\phi=0,\label{lax-a}\\
179: &&\phi_t=-(\frac{1}{2\lm}+u)\,\phi_x+\frac{u_x}2\phi.\label{lax-b}
180: \eeqa Here $m=u-u_{xx}+\frac12\kappa$. Since the term
181: $\frac12\kappa$ can be canceled by a Galileo transformation, we
182: assume $\kappa=0$ in this and the next section. The linear
183: equation (\ref{lax-a}) is known as the Schr\"odinger spectral
184: problem with energy dependent potential. Antonowicz and Fordy
185: considered the more general spectral problem \beq\label{fordy}
186: \phi_{xx}+(c+u_1\,\lm+\dots+u_n\,\lm^n)\phi=0 \eeq and associated
187: to it $n+1$ compatible local Hamiltonian structures in \cite{fordy1}.
188: Here $c$ is a given constant. In \cite{fordy1} it was also shown
189: that (\ref{fordy}) can be transformed to the spectral problem \beq
190: \psi_{xx}+(v_0+v_1\,\lm+\dots+v_{n-1}\,\lm^{n-1})\psi=\lm^n\psi.\label{alonso}
191: \eeq We will use similar transformations below in order to relate
192: the 2-CH system with the first negative flow of the AKNS
193: hierarchy. The relations of the spectral problem (\ref{alonso})
194: and its generalizations to multi-Hamiltonian structures and
195: integrable systems were studied in \cite{alonso, fordy2, fordy3},
196: and in the particular case of $n=2$ in \cite{jaulent}. The
197: bihamiltonian structure of the 2-CH system (\ref{ch-2}),
198: (\ref{ch-3}) was also given
199: in \cite{fordy1} from the spectral problem (\ref{fordy}) with $n=2$. In \cite{DLZ} the bihamiltonian structures
200: related to the generalized spectral problems of \cite{fordy3} were considered from the point of view
201: of deformations of bihamiltonian structures of hydrodynamic type, their central invariants are in general non-constants.
202:
203: Since when $\rho$ vanishes, the 2-CH system (\ref{ch-2}), (\ref{ch-3}) degenerates to the
204: Camassa-Holm equation (\ref{CH}), we assume hereafter $\rho \ne 0$.
205: Equation (\ref{ch-3}) shows that the $1$-form
206: \beq
207: \omega=\rho\,dx-\rho\,u\,dt
208: \eeq
209: is closed, so it define a reciprocal transformation $(x,t)\mapsto (y,s)$ by the relation
210: \begin{equation}\label{rt-1}
211: dy=\rho\,dx-\rho\,u\,dt,\quad ds=dt,
212: \end{equation}
213: and we have
214: \begin{equation}\label{zh9}
215: \frac{\pp}{\pp x}=\rho\,\frac{\pp}{\pp y},\quad \frac{\pp}{\pp t}=\frac{\pp}{\pp s}-\rho\,u\,\frac{\pp}{\pp y}.
216: \end{equation}
217: Denote $\varphi=\sqrt{\rho}\,\phi$, then the spectral problem (\ref{lax-a}), (\ref{lax-b}) is
218: converted to the following one
219: \eqa
220: &&\varphi_{yy}+\left(-\lm^2+P\,\lm+Q\right)\varphi=0,\label{nlax-a}\\
221: &&\varphi_s+\frac{\rho}{2\lm}\varphi_y-\frac{\rho_y}{4\lm}\varphi=0,\label{nlax-b}
222: \eeqa
223: where
224: \beq
225: P=\frac{m}{\rho^2},\quad
226: Q=-\frac1{4\rho^2}-\frac{\rho_{yy}}{2\rho}+\frac{\rho_y^2}{4\rho^2}.\label{tr-2}
227: \eeq
228:
229: Now let's consider the isospectral problem (\ref{nlax-a}) (\ref{nlax-b}). The compatibility conditions read
230: \begin{eqnarray}
231: &&P_s=\rho_y, \label{ed-1} \\
232: &&Q_s+\frac12\rho\,P_y+P\,\rho_y=0, \label{ed-2} \\
233: &&\frac12\rho\,Q_y+Q \rho_y+\frac14\rho_{yyy}=0. \label{ed-3}
234: \end{eqnarray}
235: By integrating the third equation (\ref{ed-3}) and comparing the resulting equation with (\ref{tr-2}) we obtain
236: \begin{equation}\label{ed-4}
237: \rho^2\,Q+\frac12\rho\,\rho_{yy}-\frac14\rho_y^2=C=-\frac14.
238: \end{equation}
239: From the equation (\ref{ed-1}) we know that
240: there exists a function $f(y,s)$ such that
241: \begin{equation} \label{tr-3}
242: P=\frac{\pp f(y,s)}{\pp y},\ \rho=\frac{\pp f(y,s)}{\pp s}.
243: \end{equation}
244: Substituting the expressions of $P,Q,\rho$ that are given by (\ref{tr-3}) and the second formula of
245: (\ref{tr-2}) into the equation (\ref{ed-2}) we arrive at the following equation for $f$:
246: \begin{equation}\label{ef}
247: \frac{f_{ss}}{2f_s^3}+f_yf_{ys}-\frac{f_{ss}f_{ys}^2}{2f_s^3}+\frac{f_{ys}f_{yss}}{2f_s^2}
248: +\frac12f_sf_{yy}+\frac{f_{ss}f_{yys}}{2f_s^2}-\frac{f_{yyss}}{2f_s}=0
249: \end{equation}
250:
251: \begin{thm}\label{fuv}
252: Let $f$ be a solution of the equation (\ref{ef}), and
253: \begin{equation}
254: u=f_yf_s^2+\frac{f_{ss}f_{ys}}{f_s}-f_{yss},\quad \rho=f_s. \label{zh6}
255: \end{equation}
256: If $x(y,s)$ is a solution of the following system of ODEs:
257: \begin{equation}
258: \frac{\pp x}{\pp y}=\frac1{\rho},\quad \frac{\pp x}{\pp s}=u, \label{zh7}
259: \end{equation}
260: then $(u(y,t),\rho(y,t),x(y,t))$ is a parametric solution of the 2-CH system (\ref{ch-2}), (\ref{ch-3}).
261: \end{thm}
262: \begin{rmk}
263: We say that the triple $(u(y,t),\rho(y,t),x(y,t))$ is a parametric solution of the 2-CH system if
264: the functions ${\bar{u}}(x,t)=u(y(x,t),t),\, {\bar{\rho}}(x,t)=\rho(y(x,t),t)$ satisfy the system (\ref{ch-2}),
265: (\ref{ch-3}), here $y=y(x,t)$ is the inverse function of $x=x(y,t)$. For simplicity, we will use the
266: same symbol $u, \rho$ to denote the functions $u(y(x,t),t),\, \rho(y(x,t),t)$ as functions of $x$ and $t$.
267: \end{rmk}
268:
269: \begin{prf}
270: Due to the definition of the reciprocal transformation, we only need to verify the validity of the equation
271: \beq
272: u-u_{xx}=m=\rho^2 P=f_s^2 f_y.\label{zh5}
273: \eeq
274: Denote by $E$ the l.h.s of equation (\ref{ef}). By using the definition (\ref{zh6}) of the function $u$
275: we obtain through a straightforward computation
276: $$u-\rho\,(\rho\,u_y)_y-f_s^2 f_y+2f_s^3 E_y+4 f_s^2 f_{ys}E=0$$
277: which yields (\ref{zh5}).
278: The theorem is proved.
279: \end{prf}
280:
281: \begin{defn} A function $f=f(y,s)$ is called a primary solution of
282: the 2-CH system (\ref{ch-2}), (\ref{ch-3}) if it satisfies the equation (\ref{ef}).
283: \end{defn}
284:
285: Given a solution $(u(x,t), \rho(x,t))$ of the 2-CH system (\ref{ch-2}), (\ref{ch-3}), the formulae
286: (\ref{tr-2}) and (\ref{tr-3}) determines a primary solution $f(y,s)$, we call it the primary solution
287: that is associated to $(u(x,t), \rho(x,t))$. On the other hand, any primary solution $f(y,s)$
288: yields a solution of the 2-CH system in a parametric form through the formulae (\ref{zh6}), (\ref{zh7}).
289: In the next theorem it will be shown that from a primary solution $f(y,s)$ one can construct another
290: solution of the 2-CH system. This solution is still in parametric form, however,
291: in this case the function $x=x(y,s)$ is given explicitly in terms of $f(y,s)$ without the need of integration.
292:
293: Due to (\ref{zh6}), we can rewrite the equations of (\ref{zh7}) in the form
294: \beq
295: \frac{\pp x}{\pp y}=\frac1{f_s}, \quad
296: \frac{\pp x}{\pp s}=f_yf_s^2+\frac{f_{ss}f_{ys}}{f_s}-f_{yss}.\label{xsfy}
297: \eeq
298: By using the equations (\ref{tr-3}), (\ref{zh7}) and the first equation of (\ref{tr-2}) we also have
299: \beq
300: \frac{\pp f}{\pp s}=\frac1{x_y}, \quad
301: \frac{\pp f}{\pp y}=\frac{m}{\rho^2}=\frac{u-u_{xx}}{\rho^2}=x_sx_y^2+\frac{x_{yy}x_{sy}}{x_y}-x_{syy}.\label{fyxs}
302: \eeq
303: The similarity of these equations suggests the existence of
304: a duality between solutions of the equation (\ref{ef}). Indeed,
305: we have the following theorem.
306:
307: \begin{thm}\label{fx}
308: Let $f(y,s)$ be a solution of the equation (\ref{ef}). Define the functions $x=x(y,s), u=u(y,s), \rho=\rho(y,s)$ by
309: \begin{equation}\label{zh8}
310: x=f(s,y),\quad u=\frac{\pp x}{\pp s},\quad \frac1{\rho}=\frac{\pp x}{\pp y}.
311: \end{equation}
312: Then $(u(y,t),\rho(y,t),x(y,t))$ is a parametric solution of the 2-CH system (\ref{ch-2}), (\ref{ch-3}).
313: \end{thm}
314: \begin{prf}
315: Substiting
316: $
317: u=\frac{\pp x}{\pp s},\ \frac1{\rho}=\frac{\pp x}{\pp y}
318: $
319: into the equations (\ref{ch-2}), (\ref{ch-3}) we know, by using the relation (\ref{zh9}), that
320: $(u(y,t),\rho(y,t),x(y,t))$ gives a solution to the 2-CH system if and only if the function $x(y,s)$ satisfies
321: \begin{equation}\label{exx}
322: x_{ss}+\frac{2x_sx_{ys}}{x_y}+\frac{x_{yy}}{x_y^4}-\frac{x_{ys}^2x_{yy}}{x_y^4}+
323: \frac{x_{yss}x_{yy}}{x_y^3}+\frac{x_{ys}x_{yys}}{x_y^3}-\frac{x_{yyss}}{x_y^2}=0.
324: \end{equation}
325: This equation follows immediately from the fact that the function $f(y,s)$ satisfies (\ref{ef}).
326: The theorem is proved.
327: \end{prf}
328:
329: Let us note that for the parametric solution (\ref{zh8}), the associated primary solution ${\bar f}(y,s)$ that is
330: determined by the formulae (\ref{tr-2}), (\ref{tr-3}) is in general different from the original primary
331: solution $f(y,s)$. This procedure yields a B\"acklund transformation $f(y,s)\mapsto {\bar f}(y,s)$ for the
332: equation (\ref{ef}). We will consider in detail such a class of B\"acklund transformation for the equation
333: (\ref{ef}) and the 2-CH system in another publication. In the next section, we will show how to construct
334: primary solutions of the 2-CH system from solutions of the first negative flow of the AKNS hierarchy.
335:
336:
337: \section{Relations to the first negative flow of the AKNS hierarchy}\label{sec-3}
338:
339: The AKNS spectral problem is given by
340: \begin{equation}\label{lax-3}
341: \left(\begin{array}{c} \phi_1 \\ \phi_2 \end{array}\right)_y=
342: \left(\begin{array}{rr} \lm & -q \\ r & -\lm \end{array}\right)
343: \left(\begin{array}{c} \phi_1 \\ \phi_2 \end{array}\right).
344: \eeq
345: The first negative flow of the ANKS hierarchy is equivalent to the compatibilty conditions of
346: (\ref{lax-3}) with the linear system
347: \beq\label{lax-4}
348: \left(\begin{array}{c} \phi_1 \\ \phi_2 \end{array}\right)_s=
349: \frac1{4\lm}\left(\begin{array}{rr} a & b \\ c & -a \end{array}\right)
350: \left(\begin{array}{c} \phi_1 \\ \phi_2 \end{array}\right).
351: \end{equation}
352: This flow can be represented in the form
353: \begin{eqnarray}
354: &&q_s=\frac12\, b,\quad r_s=\frac12\, c, \label{ak-1}\\
355: &&b_y=2\,a\,q,\quad c_y=2\,a\,r, \label{ak-2}\\
356: &&a_y+b\,r+c\,q=0.\label{ak-3}
357: \end{eqnarray}
358: Substitute the equation (\ref{ak-2}) into (\ref{ak-3}), we obtain
359: \begin{equation}\label{kk}
360: a^2+b\,c=\VE^2,
361: \end{equation}
362: where $\VE$ is a constant. We assume that $\VE \ne 0$.
363:
364: \begin{thm}\label{qrf}
365: Let $(a,b,c,q,r)$ be a solution of the equations (\ref{ak-1})--(\ref{ak-3}) with $\VE^2=1$,
366: then any function $f(y,s)$ satisfying
367: \begin{equation}\label{ff-1}
368: 2a=b\,e^{-f}-c\,e^f
369: \end{equation}
370: gives a primary solution of the 2-CH system.
371: \end{thm}
372: \begin{prf}
373: Assume that the function $f=f(y,s)$ satisfies (\ref{ff-1}). Let us first prove the following formula:
374: \beq
375: f_y=q\,e^{-f}+r\,e^f \label{ff-3}
376: \eeq
377: Due to (\ref{ff-1}) and (\ref{ak-2}), (\ref{ak-3}) we have
378: \begin{eqnarray}
379: 0&=&(2a-b\,e^{-f}+c\,e^f)_y=2\,a_y-b_y\,e^{-f}+c_y\,e^f+(b\,e^{-f}+c\,e^f)f_y \nn \\
380: &=&-2(b\,r+c\,q)-2\,a\,q\,e^{-f}+2\,a\,r\,e^f+(b\,e^{-f}+c\,e^f)f_y \nn \\
381: &=&-2(b\,r+c\,q)-(b\,e^{-f}-c\,e^f)(q\,e^{-f}-r\,e^f)+(b\,e^{-f}+c\,e^f)f_y \nn \\
382: &=&(b\,e^{-f}+c\,e^f)(f_y-q\,e^{-f}-r\,e^f)\label{zh1}
383: \end{eqnarray}
384: The formula (\ref{ff-3}) then follows from (\ref{zh1}) and the fact that
385: \begin{equation}\label{ff-2}
386: b\,e^{-f}+c\,e^f=\sqrt{(b\,e^{-f}-c\,e^f)^2+4\,b\,c}=\sqrt{4(a^2+b\,c)}=\pm 2\VE \ne 0
387: \end{equation}
388: Differentiating both sides of (\ref{ff-3}) w.r.t. $y$ and $s$ we obtain respectively
389: \begin{eqnarray}
390: f_{yy}&=&(q\,e^{-f}+r\,e^f)_y=q_y\,e^{-f}+r_y\,e^f-(q\,e^{-f}-r\,e^f)f_y \nn \\
391: &=&q_y\,e^{-f}+r_y\,e^f-(q\,e^{-f}-r\,e^f)(q\,e^{-f}+r\,e^f) \nn \\
392: &=&q_y\,e^{-f}+r_y\,e^f-q\,e^{-2f}+r\,e^{2f} \label{ff-4}
393: \end{eqnarray}
394: \begin{eqnarray}
395: f_{ys}&=&(q\,e^{-f}+r\,e^f)_s=q_s\,e^{-f}+r_s\,e^f-(q\,e^{-f}-r\,e^f)f_s \nn \\
396: &=&\frac12b\,e^{-f}+\frac12c\,e^f-(q\,e^{-f}-r\,e^f)f_s \label{ff-5}
397: \end{eqnarray}
398: For any solution $(\phi_1=\phi_1(y,s;\lm), \phi_2=\phi_2(y,s;\lm))$ of the systems (\ref{lax-3}), (\ref{lax-4}) define
399: \begin{eqnarray}
400: &&\varphi=e^{-\frac{f}2}\phi_1+e^{\frac{f}2}\phi_2, \label{zh10} \\
401: && P=f_y,\ \rho=f_s,\label{zh2}\\
402: &&Q=-\frac34q^2\,e^{-2f}-\frac12q\,r-\frac34r^2\,e^{2f}+\frac12q_y\,e^{-f}-\frac12r_y\,e^f.
403: \end{eqnarray}
404: By using equations (\ref{lax-3})-(\ref{ff-5}) and the fact $\varepsilon^2=1$, we can show through a straightforward and lengthy computation
405: that the functions $\varphi, P, Q, \rho$ satisfy the equations (\ref{nlax-a}), (\ref{nlax-b}) and (\ref{ed-4}).
406: To illustrate our computation, we give here the proof of the validity of equation (\ref{nlax-b}).
407: From (\ref{lax-3}), (\ref{lax-4}) and (\ref{ff-3}) it follows that
408: \begin{eqnarray}
409: \varphi_s&=&(e^{-\frac{f}2}\phi_1+e^{\frac{f}2}\phi_2)_s
410: =e^{-\frac{f}2}(\phi_1)_s+e^{\frac{f}2}(\phi_2)_s-(e^{-\frac{f}2}\phi_1-e^{\frac{f}2}\phi_2)\frac{f_s}2 \nn \\
411: &=&e^{-\frac{f}2}\frac{a\,\phi_1+b\,\phi_2}{4\lm}+e^{\frac{f}2}\frac{c\,\phi_1-a\,\phi_2}{4\lm}
412: -(e^{-\frac{f}2}\phi_1-e^{\frac{f}2}\phi_2)\frac{f_s}2 \nn \\
413: &=&\left(\frac{a+c\,e^f}{4\lm}-\frac{f_s}2\right)e^{-\frac{f}2}\phi_1+
414: \left(\frac{b\,e^{-f}-a}{4\lm}+\frac{f_s}2\right)e^{\frac{f}2}\phi_2
415: \end{eqnarray}
416: and
417: \begin{eqnarray}
418: \varphi_y&=&(e^{-\frac{f}2}\phi_1+e^{\frac{f}2}\phi_2)_y
419: =e^{-\frac{f}2}(\phi_1)_y+e^{\frac{f}2}(\phi_2)_y-(e^{-\frac{f}2}\phi_1-e^{\frac{f}2}\phi_2)\frac{f_y}2 \nn \\
420: &=&e^{-\frac{f}2}(\lm\,\phi_1-q\,\phi_2)+e^{\frac{f}2}(r\,\phi_1-\lm\,\phi_2)
421: -(e^{-\frac{f}2}\phi_1-e^{\frac{f}2}\phi_2)\frac{q\,e^{-f}+r\,e^f}2 \nn \\
422: &=&\left(\lm-\frac{q\,e^{-f}-r\,e^f}2\right)e^{-\frac{f}2}\phi_1+
423: \left(-\lm-\frac{q\,e^{-f}-r\,e^f}2\right)e^{\frac{f}2}\phi_2
424: \end{eqnarray}
425: By using these two formulae and (\ref{ff-5}), (\ref{zh2}) we obtain
426: \begin{eqnarray}
427: &&\varphi_s+\frac{\rho}{2\lm}\varphi_y-\frac{\rho_y}{4\lm}\varphi
428: =\varphi_s+\frac{f_s}{2\lm}\varphi_y-\frac{f_{ys}}{4\lm}\varphi \nn \\
429: &&=\frac{e^{-\frac{f}2}\phi_1-e^{\frac{f}2}\phi_2}{4\lm}\left(a-\frac{b}2\,e^{-f}+\frac{c}2\,e^f\right)=0.
430: \end{eqnarray}
431: The theorem then follows from the derivation of equation (\ref{ef}) that is given in the last section.
432: \end{prf}
433:
434: \begin{rmk}
435: The main point in the proof of the above theorem is the establishment of the gauge transformation
436: (\ref{zh10}) between the isospectral problem (\ref{nlax-a}), (\ref{nlax-b}) and (\ref{lax-3}), (\ref{lax-4}).
437: For the positive flows of AKNS hierarchy, there also exist gauge transformations of form
438: $\varphi=e^{-\frac{f}2}\phi_1+e^{\frac{f}2}\phi_2$, where $f$ satisfies $f_y=q\,e^{-f}+r\,e^f$,
439: they establish correspondences between the positive flows of the AKNS hierarchy and the ones that are related to the
440: isospectral problems (\ref{nlax-a}).
441: \end{rmk}
442:
443: The following theorem gives an explicit way of constructing a solution of the first negative flow of the AKNS hierarchy
444: from a primary solution of the 2-CH system.
445: \begin{thm}\label{fqr}
446: If $f$ is a primary solution of the 2-CH system (\ref{ch-2}), (\ref{ch-3}),
447: then we can construct a solution of the first negative flow of the AKNS hierarchy by the following formulae
448: \begin{equation}
449: q=\frac{e^f}2\left(f_y+\frac{\VE-f_{ys}}{f_s}\right),\ r=\frac{e^{-f}}2\left(f_y-\frac{\VE-f_{ys}}{f_s}\right),\
450: b=2\,q_s,\ c=2\,r_s,\ a=\frac{b\,e^{-f}-c\,e^f}2.
451: \end{equation}
452: where $\VE=1$ or $\VE=-1$.
453: \end{thm}
454: \begin{prf}
455: Since $f$ is a primary solution of the 2-CH system, we have $E=0$ where $E$ is defined as in the proof of
456: Theorem \ref{fuv}. Then by a straightforward computation, we obatin
457: \begin{equation}
458: b_y-2\,a\,q=2e^f\,E=0,\ c_y-2\,a\,r=2e^{-f}\,E=0,\ a^2+b\,c=\VE^2=1.
459: \end{equation}
460: The theorem is proved.
461: \end{prf}
462:
463: By using the freedom in the choice of signs of the parameter $\VE$ we obtain the following corollary:
464: \begin{cor}\label{bkl}
465: We have the following two B\"acklund transformations for the equation (\ref{ef})
466: \begin{equation}
467: f \mapsto B_\VE f=f+\log\left(
468: \frac{f_{ss}f_{ys}+f_s^3f_y-f_sf_{yss}+\VE(f_{ss}-f_s^2)}{f_{ss}f_{ys}+f_s^3f_y-f_sf_{yss}+\VE(f_{ss}+f_s^2)}\right),\
469: \VE=1 \ {\rm or}\ -1.
470: \end{equation}
471: \end{cor}
472: \begin{prf}
473: Since $f$ is a primary solution of the 2-CH system, by using Theorem \ref{fqr} we obtain two solutions of the first
474: negative flow of the AKNS hierarchy, we denote them by
475: $(a_\VE,b_\VE,c_\VE,q_\VE,r_\VE), \VE=1, -1$. Due to Theorem \ref{qrf}, each of these two
476: solutions yields two primary solutions of the 2-CH system
477: \begin{equation}
478: f_{\VE,\gamma}=\log\frac{\gamma-a_\VE}{c_\VE}=\log\frac{b_\VE}{\gamma+a_\VE},\quad \VE, \gamma=\pm1.
479: \end{equation}
480: It is easy to see that $f_{\VE,\gamma}=f$ when $\gamma=\VE$. The corollary is then proved if we identify
481: $B_\VE f$ with $f_{\VE,-\VE}$.
482: \end{prf}
483:
484: The B\"acklund transformations given by the above corollary can also be represented in terms of the
485: dependent variables $q, r$ of the first negative flow of the AKNS hierarchy, due to their long expressions
486: we do not present the formulae here. Instead, let us illustrate the procedure of obtaining solutions of the
487: system (\ref{ak-1})--(\ref{ak-3}) starting from the following trivial solution:
488: \begin{equation}\label{tsol1}
489: q=\sum_{i=1}^n e^{\zeta_i y+\frac{\gamma}{\zeta_i} s+\xi_i},\ r=0,\ b=2q_s,\ c=0,\ a=\gamma,
490: \end{equation}
491: where $n\in\mathbb{N}$,\ $\zeta_i, \xi_i, \gamma$ are arbitrary
492: constants with $\zeta_i$ non-vanishing and pairwise distinct. According
493: to Theorem \ref{qrf} we know that
494: $$f_0=\log\frac{b}{\gamma+a}=\log\frac{q_s}\gamma$$
495: is a primary solution of the 2-CH system. By using the above corollary, we obtain two solutions of the equation
496: (\ref{ef}), however, only one of them makes sense. By repeating this procedure, we obtain a sequence of primary
497: solutions $f_0,f_1,f_2,\dots$ of the 2-CH system. Then by using Theorem \ref{fuv} or Theorem \ref{fx} and Theorem
498: \ref{fqr}, we obtain a sequence of solutions of the 2-CH system and the first negative flow of the AKNS hierarchy.
499: Such class of solutions of the 2-CH system have very nice properties, we will study them in more detail
500: in the next section.
501:
502: \section{Particular solutions of the 2-CH system}\label{sec-4}
503: We are now ready to present in this section some examples of solutions of the 2-CH system, they include the peakon and
504: multi-kink solutions.
505:
506: {\bf Example 1.} Let us first try, without referring to the reciprocal transformation
507: constructed above, to find travelling wave solutions of the 2-CH system.
508: Assume $u=h(x+v\,t),\rho=g(x+v\,t)$, where $v$ is a constant. Then the equation (\ref{ch-2}) and (\ref{ch-3}) become
509: \eqa\label{fff}
510: &&v(h'-h''')+3h\,h''-2h'\,h''-h\,h'''+\kappa\,h'-g\,g'=0,\\
511: &&v\,g'+g\,h'+h\,g'=0.
512: \eeqa
513: We can solve the second equation directly
514: \begin{equation}\label{gf} g=\frac{A}{v+h}\end{equation}
515: where A is a constant. Then the equation (\ref{fff}) becomes an ODE for $h$. This ODE can be solved by
516: a standard method. By carefully choosing the integration constant, we obtain the following solution
517: \begin{equation}\label{pk-1}
518: u=\chi-\sqrt{\chi^2-v^2},\ \rho=\sqrt{v\,K}\left(1+\sqrt{\frac{\chi-v}{\chi+v}}\right),\ {\rm where}
519: \ \chi=(v-K)\cosh(x+v\,t)+K,
520: \end{equation}
521: where $K=-4\kappa$. If $v>0,K>0$, this is a travelling peakon solution, see Figure 1.
522:
523: {\bf Example 2.} The first negative flow of the AKNS hierarchy has an important reduction.
524: Under the assumption
525: \begin{equation}
526: q=-\frac{w_y}2,\ r=\frac{w_y}2,\ b=-\sinh w,\ c=\sinh w,\ a=\cosh w,
527: \end{equation}
528: it is reduced to the sinh-Gordon equation
529: \begin{equation}
530: w_{ys}=\sinh w.
531: \end{equation}
532:
533: Now let us employ the results of the previous sections to obtain a stationary peakon solution of the 2-CH
534: system (\ref{ch-2}) and (\ref{ch-3})
535: with $\kappa=0$. Due to Theorem \ref{qrf}, a solution of the sinh-Gordon equation leads to a primary solution
536: of the 2-CH system
537: \begin{equation}
538: f(y,s)=\varepsilon \log\left(-\tanh\frac{w}2\right), {\rm where}\ \varepsilon=\pm1.
539: \end{equation}
540: Then the equations in (\ref{tr-3}) become
541: \begin{equation}
542: P=\varepsilon \frac{w_y}{w_{ys}},\ \rho=\varepsilon \frac{w_s}{w_{ys}}.
543: \end{equation}
544: By using Theorem \ref{fuv}, we find a parametric solution of the 2-CH system with $\kappa=0$
545: \begin{equation}\label{sg}
546: x(y,s)=\varepsilon \log w_s+x_0,\quad u(y,s)=\varepsilon \frac{w_{ss}}{w_s},\quad \rho=\varepsilon \frac{w_s}{w_{ys}}.
547: \end{equation}
548: where $x_0$ is an arbitrary constant.
549:
550: Now let us choose a kink solution of the sinh-Gordon equation
551: \begin{equation}
552: w(y,s)=4 \tanh^{-1}\left(e^{p_1\, y+\frac{s}{p_1}+q_1}\right),
553: \end{equation}
554: with some constants $p_1\ne 0, q_1$ and substitute it into (\ref{sg}), we obtain a stationary solution of the 2-CH system
555: \begin{equation}
556: u=\frac{\sqrt{1+e^{2\varepsilon (x-x_0)}}}{p_1},\ \rho=\frac1{p_1\,\sqrt{1+e^{2\varepsilon (x-x_0)}}}
557: \end{equation}
558: Then it is easy to see that the following $u,\rho$
559: \begin{equation}\label{pk-2}
560: u=\frac{\sqrt{1+e^{-2|x-x_0|}}}{p_1},\ \rho=\frac1{p_1\sqrt{1+e^{-2|x-x_0|}}}
561: \end{equation}
562: is a stationary peakon solution of the 2-CH system, see Figure 1. Note that a peakon solution with constant
563: speed $-\frac12\kappa$ for the 2-CH system without the assumption of $\kappa=0$ can be obtained by the
564: Galilean transformation $x\mapsto\tilde x= x-\frac12\,\kappa\,t,\ t\mapsto \tilde t=t, \ u\mapsto \tilde u=u-\frac12\,\kappa,
565: \ \rho\mapsto \tilde \rho=\rho$.
566: If we choose $w$ as a multi-kink solution of the
567: sinh-Gordon equation, we will arrive at some very strange
568: solutions of the 2-CH system whose peaked points are also
569: stationary.
570:
571: \begin{figure}
572: \begin{center} \small
573: $$
574: \begin{array}{cc}
575: \includegraphics[width=0.45\textwidth]{example1.eps} &
576: \includegraphics[width=0.45\textwidth]{example2.eps} \\
577: \mbox{(a) (\ref{pk-1}) with } v=2,K=1,t=1. &
578: \mbox{(b) (\ref{pk-2}) with } p_1=1,x_0=2.
579: \end{array}
580: $$
581: \end{center}
582: \caption{Travelling peakon and stationary peakon}
583: \end{figure}
584:
585: {\bf Example 3.} In this example we give some explicit expressions of kink and 2-kink interaction solutions of
586: the 2-CH system with $\kappa=0$. These solutions
587: are derived from the particular trivial solutions of the first negative flow of the AKNS hierarchy
588: given in the end of the last section and by using the B\"acklund transformations of Corollary \ref{bkl}. It's easy to
589: see that there are only constant
590: solutions when $n=1$. We consider here the cases when $n=2$ and $n=3$.
591:
592: Let us first assume $n=2$. Denote $\xi_i=p_i\,y+\frac{s}{p_i} +q_i$, then we have the following solution
593: for the first negative flow of the AKNS hierarchy
594: \begin{equation}
595: q=p_1 e^{\xi_1}+p_2 e^{\xi_2},\ r=0,\ b=2(e^{\xi_1}+e^{\xi_2}),\ c=0,\ a=1,
596: \end{equation}
597: where $p_1 \ne p_2$. By applying Theorem \ref{qrf} and the B\"acklund transformations of Corollary \ref{bkl} we arrive
598: at the following two primary solutions of the 2-CH system
599: \begin{equation}
600: f_0=\log\left(e^{\xi_1}+e^{\xi_2}\right),\
601: f_1=\log\left(\frac{(p_1-p_2)^2 e^{\xi_1+\xi_1}}{p_1^2 e^{\xi_1}+p_2^2 e^{\xi}_2}\right).
602: \end{equation}
603: Note that a further application of the B\"acklund transformations of Corollary \ref{bkl} leads to
604: $f_2=\log(0)$. So in this case we can only obtain two primary solutions. By using Theorem \ref{fx} we obtain
605: two solutions of the 2-CH system which have the form (\ref{zh8}) with the function $x(y,s)$ given respectively by
606: \begin{equation}\label{kink}
607: x_0=\log\left(e^{\tilde{\xi}_1}+e^{\tilde{\xi}_2}\right),\
608: x_1=\log\left(\frac{(p_1-p_2)^2 e^{\tilde{\xi}_1+\tilde{\xi}_1}}{p_1^2 e^{\tilde{\xi}_1}+p_2^2 e^{\tilde{\xi}_2}}\right).
609: \end{equation}
610: where $\tilde{\xi}_i=p_i\,s+\frac{y}{p_i}+q_i$. The solution obtained from $x_0$ (respectively, $x_1$) is an antikink
611: (respectively, kink), see Figure 2 where the profiles of $\rho$ are represented by the dashed curves.
612:
613: \begin{figure}
614: \begin{center}\small
615: $$
616: \begin{array}{cc}
617: \includegraphics[width=0.45\textwidth]{example3-1.eps} &
618: \includegraphics[width=0.45\textwidth]{example3-2.eps} \\
619: \mbox{(a) } u,\rho \mbox{ obtained from } x_0. &
620: \mbox{(b) } u,\rho \mbox{ obtained from } x_1.
621: \end{array}
622: $$
623: \end{center}
624: \caption{Kink and antikink. ((\ref{kink}) with $p_1=1,p_2=2,q_1=0,q_2=0.$)}
625: \end{figure}
626:
627: Now let us consider the case when $n=3$, in a similar way to the previous case we obtain three
628: solutions of the 2-CH system which have the form (\ref{zh8}) with the function $x(y,s)$ given respectively by
629: \begin{eqnarray}
630: x_0&=&\log\left(e^{\tilde{\xi}_1}+e^{\tilde{\xi}_2}+e^{\tilde{\xi}_3}\right), \label{kk-x0}\\
631: x_1&=&\log\left(\frac{p_1^2(p_2-p_3)^2e^{\tilde{\xi}_2+\tilde{\xi}_3}+p_2^2(p_3-p_1)^2e^{\tilde{\xi}_3+\tilde{\xi}_1}
632: +p_3^2(p_1-p_2)^2e^{\tilde{\xi}_1+\tilde{\xi}_2}}{p_2^2p_3^2e^{\tilde{\xi}_1}+p_3^2p_1^2e^{\tilde{\xi}_2}+
633: p_1^2p_2^2e^{\tilde{\xi}_3}}\right), \label{kk-x1} \\
634: x_2&=&\log\left(\frac{(p_1-p_2)^2(p_2-p_3)^2(p_3-p_1)^2e^{\tilde{\xi}_1+\tilde{\xi}_2+\tilde{\xi}_3}}{
635: p_1^4(p_2-p_3)^2e^{\tilde{\xi}_2+\tilde{\xi}_3}+p_2^4(p_3-p_1)^2e^{\tilde{\xi}_3+\tilde{\xi}_1}
636: +p_3^4(p_1-p_2)^2e^{\tilde{\xi}_1+\tilde{\xi}_2}}\right), \label{kk-x2}
637: \end{eqnarray}
638: where $p_1, p_2, p_3$ are pairwise distinct.
639: These solutions describe the antikink-antikink, antikink-kink and kink-kink interactions, see Figure 3,4,5.
640: These figures are drawn with parameters
641: \begin{equation}
642: p_1=1,\ p_2=2,\ p_3=3,\ q_1=0,\ q_2=0,\ q_3=0
643: \end{equation}
644: and only for $u$, since the figures of $\rho$ corresponding to the above three solutions of
645: the 2-CH system are
646: very similar to that of $u$, like the figures of the single kink solutions drawn in Figure 2.
647:
648: In general, for any $n\in\mathbb N$ we expect to arrive at in this way $n$ solutions of the 2-CH system, each of which
649: is a $(n-1)$-kink solution if we choose the parameters $p_i, q_i$ in an appropriate way. They correspond
650: to the interactions of $k$ antikinks and $n-1-k$ kinks with $k=0,\dots, n-1$.
651: Except for the cases of $n=2, 3$, we can also check
652: this assertion for the case when $n=4$. We will leave the analysis of the general case to a subsequent
653: publication.
654:
655:
656: \begin{figure}
657: \begin{center}
658: \includegraphics[width=0.45\textwidth]{example4-01.eps}
659: \includegraphics[width=0.45\textwidth]{example4-02.eps}
660: \includegraphics[width=0.45\textwidth]{example4-03.eps}
661: \includegraphics[width=0.45\textwidth]{example4-04.eps}
662: \end{center}
663: \caption{The interaction of two antikinks (\ref{kk-x0}).}
664: \end{figure}
665:
666: \begin{figure}
667: \begin{center}
668: \includegraphics[width=0.45\textwidth]{example4-11.eps}
669: \includegraphics[width=0.45\textwidth]{example4-12.eps}
670: \includegraphics[width=0.45\textwidth]{example4-13.eps}
671: \includegraphics[width=0.45\textwidth]{example4-14.eps}
672: \end{center}
673: \caption{The interaction of an antikink and a kink (\ref{kk-x1}).}
674: \end{figure}
675:
676: \begin{figure}
677: \begin{center}
678: \includegraphics[width=0.45\textwidth]{example4-21.eps}
679: \includegraphics[width=0.45\textwidth]{example4-22.eps}
680: \includegraphics[width=0.45\textwidth]{example4-23.eps}
681: \includegraphics[width=0.45\textwidth]{example4-24.eps}
682: \end{center}
683: \caption{The interaction of two kinks (\ref{kk-x2}).}
684: \end{figure}
685:
686: \section{Conclusion}
687: We have constructed an explicit reciprocal transformation between the 2-CH system (\ref{ch-2}), (\ref{ch-3}) and the first negative
688: flow of the AKNS hierarchy (\ref{ak-1})--(\ref{ak-3}) with $\VE^2=1$. The primary solution $f(y,s)$ satisfying
689: the equation (\ref{ef})
690: plays a crucial role in this construction. This transformation comprises of two
691: step, the first step is the correspondence between solutions of the 2-CH system and the primary solutions
692: satisfying (\ref{ef}), it is given by the formulae (\ref{tr-2}), (\ref{tr-3}) and Theorems \ref{fuv}, \ref{fx}.
693: The second step is the correspondence between solutions of the first negative
694: flow of the AKNS hierarchy and the primary solutions of the 2-CH system, it is given by Theorems \ref{qrf}, \ref{fqr}.
695: These correspondences are presented in simple and explicit forms, they enable us to obtain solutions of the
696: 2-CH system from that of the first negative flow of the AKNS hierarchy, which includes in particular the well
697: known sine-Grodon and the sinh-Gordon equations.
698:
699: In terms of the primary solutions we also obtained
700: in Sec.\ref{sec-2}, \ref{sec-3} two kinds of B\"acklund transformations for the 2-CH system or
701: the first negative flow of the AKNS hierarchy, and showed in Sec.\ref{sec-4}
702: that the B\"acklund transformations given in this section
703: lead to interesting multi-kink solutions of the 2-CH system. It would be interesting to express
704: in terms of the primary solutions of the 2-CH system the
705: B\"acklund transformations of the AKNS hierarchy that are well known in the literatures, see for example \cite{gz,
706: matveev}.
707:
708: We note that the travelling peakon solution of the 2-CH system that is given by
709: the first example of Sec.\ref{sec-4} assumes that the constant $K>0$, when $K=0$
710: it degenerates to a peakon solution of the Camassa-Holm equation (\ref{CH}).
711: The existence of multi-kink solutions of the 2-CH system is also
712: an intersting phenomenon for a system of hydrodynamic type with dispersive perturbation terms.
713: We will return to analyze in a subsequent paper in more details the various properties of particular solutions of the
714: 2-CH system, including the problem of existence of multi-peakon solutions.
715: Although the
716: 2-CH system that we considered here was derived from the problem of classification of deformations of
717: bihamitionian structures of hydrodynamic type, we do expect that it would find for itself physically
718: important applications.
719:
720:
721: \vskip 0.2truecm \noindent{\bf Acknowledgments.} The authors are
722: grateful to Boris Dubrovin for helpful discussions and comments. They also
723: thank Yishen Li for drawing their attention to the interesting papers \cite{LZJ1, LZJ2}.
724: The researches of Y.Z. were partially
725: supported by the Chinese National Science Fund for Distinguished
726: Young Scholars grant No.10025101 and the Special Funds of Chinese
727: Major Basic Research Project ``Nonlinear Sciences''.
728:
729:
730:
731:
732: \begin{thebibliography}{99}
733: \bibitem{AG} Abenta S., Grava T., Modulation of Camassa-Holm
734: equation and reciprocal transforations, preprint 2004.
735: \bibitem{akns}Ablowitz M. J., Kaup D. J., Newell A. C., Segur H., The inverse scattering
736: transform-Fourier analysis for nonlinear problems. Studies in Appl. Math. {\bf 53} (1974), 249--315.
737: \bibitem{Alb} Alber M. S., Camassa R., Fedorov Yu. N., Holm D. D., Marsden J. E.,
738: The complex geometry of weak piecewise smooth solutions of integrable nonlinear
739: PDEs of shallow water and Dym type, Comm. Math. Phys. {\bf 221} (2001) 197-227.
740: \bibitem{fordy1}Antonowicz M., Fordy A. P.,
741: Coupled KdV equations with multi-Hamiltonian structures. Phys. {\bf D 28}
742: (1987) 345--357.
743: \bibitem{fordy2} Antonowicz M., Fordy A. P.,
744: Coupled Harry Dym equations with multi-Hamiltonian structures, J. Phys.
745: {\bf A 21} (1988) L269--L275.
746: \bibitem{fordy3} Antonowicz M., Fordy A. P.,
747: Factorisation of energy dependent Schr\"odinger operators:
748: Miura maps and modified systems. Comm. Math. Phys. {\bf 124} (1989)
749: 465--486.
750: \bibitem{CH} Camassa R. and Holm D. D., An integrable shallow water equation with peaked solitons,
751: Phys. Rev. Lett. {\bf 71} (1993) 1661-1664.
752: \bibitem{CHH} Camassa R., Holm D. D. and Hyman J. M., A new integrable shallow water equation,
753: Adv. Appl. Mech. {\bf 31} (1994), 1-33.
754: \bibitem{Cons-1}Constantin A., On the scattering problem for
755: the Camassa-Holm equation. R. Soc. Lond. Proc. Ser. A Math. Phys. Eng. Sci. {\bf 457} (2001),
756: 953--970.
757: \bibitem{hone-1}Degasperis A., Holm D. D., Hone A. N. W., Integrable and non-integrable equations with peakons.
758: Nonlinear physics: theory and experiment, II (Gallipoli, 2002), 37--43, World Sci. Publishing, River Edge, NJ, 2003.
759: \bibitem{DZ} Dubrovin B., Zhang Y., Normal forms of integrable
760: PDEs, Frobenius manifolds and Gromov-Witten invariants, math.DG/0108160.
761: \bibitem{DLZ} Dubrovin B., Liu S. Q., Zhang Y., On Hamiltonian perturbations of hyperbolic systems of conservation
762: laws, math.DG/0410027.
763: \bibitem{falqui}Falqui G., On a two-component generalization of the CH equation, talk given at the conference
764: ``Analytic and geometric theory of the Camassa-Holm equation and Integrable systems'', Bologna, September 22- 25, 2004.
765: \bibitem{Fo} Fokas A. S., On a class of physically important integrable
766: equations, Physica {\bf D 87}(1995) 145-150.
767: \bibitem{Fu} Fuchssteiner B., Some tricks from the symmetry-toolbox for nonlinear equations:
768: Generalizations of the Camassa-Holm equation, Physica {\bf D 95}
769: (1996) 229-243.
770: \bibitem{FF} Fuchssteiner B., Fokas A. S., Symplectic structures, their B\"acklund transformations and
771: hereditary symmetries, Physica {\bf D 4} (1981) 47-66.
772: \bibitem{gz}Gu C. H., Zhou Z. X., On the Darboux matrices of
773: B\"acklund transformations for AKNS systems. Lett. Math. Phys. {\bf 13} (1987), 179--187.
774: \bibitem{hone} Hone A. N. W., The associated Camassa-Holm equation and the KdV equation. J. Phys. A {\bf 32}
775: (1999), L307--L314.
776: \bibitem{jaulent}Jaulent M., Jean C., The inverse problem for the one-dimensional Schr\"odinger equation with
777: an energy-dependent potential. I. Ann. Inst. H. Poincar¨¦ Sect. A (N.S.) {\bf 25} (1976), 105--118.
778: \bibitem{johnson} Johnson R. S., On solutions of the Camassa-Holm equation. R. Soc. Lond. Proc. Ser. A Math. Phys. Eng.
779: Sci. {\bf 459} (2003), 1687--1708.
780: \bibitem{LZJ1}Li Y. S., Zhang J. E., The multiple-soliton solution of the Camassa-Holm equation, R. Soc. Lond. Proc. Ser. A Math. Phys. Eng.
781: Sci. {\bf 460} (2004),
782: \bibitem{LZJ2} Li Y. S., Zhang J. E., Analytical multiple-soliton solution of the Camassa-Holm equation, preprint 2004, to
783: appear in J. Nonlinear Math. Phys.
784: \bibitem{LZ1} Liu S. Q., Zhang Y., Deformations of semisimple bihamiltonian structures of
785: hydrodynamic type, math-DG/0405146, to appear in J. Geom. Phys.
786: \bibitem{alonso} Mart\'{\i}nez Alonso L., Schr\"odinger spectral problems with energy-dependent potentials as
787: sources of nonlinear Hamiltonian evolution equations, J. Math. Phys. {\bf 21}
788: (1980) 2342--2349.
789: \bibitem{matveev} Matveev V. B., Salle M. A., Darboux transformations and solitons.
790: Springer Series in Nonlinear Dynamics. Springer-Verlag, Berlin, 1991.
791: \bibitem{mckean-1}McKean H., The Liouville correspondence between the Korteweg-de Vries and the
792: Camassa-Holm hierarchies. Dedicated to the memory of J¨¹rgen K. Moser. Comm. Pure Appl. Math. {\bf 56} (2003), 998--1015.
793: \bibitem{mckean-2}McKean H., Breakdown of the Camassa-Holm equation. Comm. Pure Appl. Math. {\bf 57} (2004), 416--418.
794: \bibitem{schiff}Schiff J., The Camassa-Holm equation: a loop group approach. Phys. D {\bf 121} (1998), no. 1-2, 24--43.
795: \end{thebibliography}
796:
797: \eqa
798: {\rm email\ addresses:} && {\rm {\mbox{chen-02}}@mails.tsinghua.edu.cn,\ lsq99@mails.tsinghua.edu.cn} \nn\\
799: && {\rm yzhang@math.tsinghua.edu.cn}\nn
800: \eeqa
801: \end{document}
802: