1: %%International Journal of Modern Physics C

2: %%15/8/2001

3: 

4: \documentstyle[twoside,ijmpc1]{article}

5: \textwidth=5truein

6: \textheight=7.26truein

7: 

8: %--------------------------------------------------------------------------

9: %DEFINITIONS

10: \def\qed{\hbox{${\vcenter{\vbox{            %HOLLOW SQUARE

11:    \hrule height 0.4pt\hbox{\vrule width 0.4pt height 6pt

12:    \kern5pt\vrule width 0.4pt}\hrule height 0.4pt}}}$}}

13: 

14: \renewcommand{\thefootnote}{\fnsymbol{footnote}}    %USE SYMBOLIC FOOTNOTE

15: 

16: \def\bsc{{\sc a\kern-6.4pt\sc a\kern-6.4pt\sc a}}   %LATEX LOGO

17: \def\bflatex{\bf L\kern-.30em\raise.3ex\hbox{\bsc}\kern-.14em

18: T\kern-.1667em\lower.7ex\hbox{E}\kern-.125em X}

19: 

20: %-------------------------------------------------------------------------

21: \pagestyle{myheadings}

22: \begin{document}

23: \setlength{\textheight}{7.7truein}  %for 2nd page onwards

24: 

25: \thispagestyle{empty}

26: 

27: \markboth{\protect{\footnotesize\it Integrability of a

28: Non-autonomous

29:  Coupled KdV System}}

30: {\protect{\footnotesize\it Ay\c{s}e Karasu (Kalkanl{\i})

31:   and Tuba K{\i}l{\i}\c{c}}}

32: 

33: \normalsize\textlineskip

34: 

35: \setcounter{page}{1}

36: 

37: \copyrightheading{}         %{Vol. 0, No. 0 (1993) 000--000}

38: 

39: \vspace*{0.88truein}

40: 

41: %\fpage{1}

42: \centerline{\bf INTEGRABILITY OF A NON-AUTONOMOUS}

43: \vspace*{0.035truein}

44: \centerline{\bf COUPLED KdV SYSTEM}

45: 

46: \vspace*{0.37truein}

47: \centerline{\footnotesize AY\c{S}E KARASU(KALKANLI)\footnote{

48: E-mail: akarasu@metu.edu.tr}

49: and TUBA KILI\c{C}}

50: \baselineskip=12pt

51: \centerline{\footnotesize\it Department of Physics, Middle East

52: Technical University}

53: \baselineskip=10pt

54: \centerline{\footnotesize\it  06531 Ankara,

55: Turkey}

56: %\centerline{\footnotesize\it E-mail: akarasu@metu.edu.tr}

57: 

58: %\vspace*{10pt}     %actual spacing

59: \vspace*{15pt}          %when needed

60: \vspace*{0.225truein}

61: \publisher{(received date)}{(revised date)}

62: 

63: \vspace*{0.25truein}

64: \abstracts{The

65: Painlev\'{e} property of  coupled, non-autonomous Korteweg-de

66: Vries (KdV) type of systems is studied. The conditions under which

67: the systems pass the Painlev\'{e} test for integrability are

68: obtained. For some of the integrable cases, exact solutions are

69: given.

70: }{}{}

71: 

72: \vspace*{5pt}

73: \keywords{non-autonomous KdV systems; Painlev\'{e} analysis;

74: exact solutions}

75: 

76: %\textlineskip          %) USE THIS MEASUREMENT WHEN THERE IS

77: %\vspace*{12pt}         %) NO SECTION HEADING

78: 

79: \vspace*{1pt}\textlineskip  %) USE THIS MEASUREMENT WHEN THERE IS

80: \section{Introduction}     %) A SECTION HEADING

81: \vspace*{-0.5pt}

82: \noindent

83: For a better

84: understanding of complicated physical phenomena scientists have

85: experienced that it is necessary to introduce mathematical models

86: whose time evolutions might show some features very similar to

87: those of the original phenomena. These models are usually systems

88: of nonlinear differential equations. These equations can be solved

89: by the use of approximation techniques. But the range of

90: applicability and usefulness of these solutions increase the

91: interest on the exact solutions and on the solution generating

92: methods for nonlinear equations. Before attempting to solve an

93: equation one usually needs to know whether the equation is

94: integrable or not. The difficulty of obtaining solutions by use of

95: inverse scattering transform technique, makes this information

96: valuable. One of the powerful tools to obtain this information is

97: the Painlev\'{e} test for integrability. There are strong

98: evidences that all integrable equations have Painlev\'{e}

99: property, that is, all solutions are single valued around movable

100: singularities.$^{1}$

101: \eject

102: In this work, we consider the non-autonomous, coupled KdV type systems

103: \begin{eqnarray}

104: u_t&=&u_{xxx}+ a(t) uu_x,\nonumber\\ v_t&=&v_{xxx}+

105: b(t) (uv)_x,

106: \end{eqnarray}

107: where $a(t)$ and $b(t)$ are some arbitrary

108: functions. We apply the Painlev\'{e} test for integrability to

109: system (1), following the Weiss-Kruskal algorithm of singularity

110: analysis $^{2,3}$ and obtain the conditions on the functions $a(t)$

111: and $b(t)$. We find the subclasses of these equations that possess

112: the Painlev\'{e} property. By using the truncated expansions we

113: obtain the exact solutions for some of these subclasses,

114: explicitly.

115: 

116: %\eject

117: 

118: \setcounter{footnote}{0}

119: \renewcommand{\thefootnote}{\alph{footnote}}

120: 

121: \section{The Non-autonomous Coupled KdV type Systems}

122: \noindent

123: Following

124: the approach of Weiss {\it et al} $^2$, we assume that the

125: solutions of (1) can be represented by the expansions, \begin{eqnarray}

126: u(x,t)&=& \sum_{r=0}^{\infty}u_r(x,t)\phi^{r+\alpha},\nonumber\\

127: v(x,t)&=& \sum_{r=0}^{\infty}v_r(x,t)\phi^{r+\beta}, \end{eqnarray} where

128: $\alpha$ and $\beta$ are integers, $u_r(x,t)$ and $v_r(x,t)$ are

129: analytic functions in a neighborhood of the singularity manifold

130: $\phi(x,t)$. A hypersurface $\phi(x,t)=0$ is noncharacteristic of

131: the system (1) if $\phi_t \phi_x \neq 0$. We choose$^3$

132: $\phi(x,t)=x+\psi(t)$, without loss of generality, hence the

133: coefficients $(u_r,v_r)$ are independent of $x$. This is the

134: simplest choice for the test, but it cannot be used to obtain the

135: particular solutions. The substitution of $u(x,t)=

136: \sum_{r=0}^{n}u_r(t)\phi^{r+\alpha},\;\;\; v(x,t)=

137: \sum_{r=0}^{n}v_r(t)\phi^{r+\beta}$ into (1) determines the

138: branches, i.e. the admissible dominant behavior of solutions, and

139: the corresponding positions $r$ of the resonances where the

140: arbitrary functions can appear in the expansions (2). The leading

141: order analysis gives that \begin{equation} \alpha=-2 ,\;\;\;

142: u_0=-\frac{12}{a(t)} ,\;\;\;

143: (\beta-2)\left[\beta(\beta-1)-12\frac{b(t)}{a(t)}\right]v_0(t)=0.

144: \end{equation} The branches satisfying (3) are \begin{equation}

145: \beta_1 = 2 ,\;\; \beta_2

146: = -m ,\;\; \beta_3 = m+1 \end{equation}

147: where $m$ is a non-negative integer

148: and \begin{equation}

149: b(t) = \frac{a(t)}{12}(m^2+m). \end{equation} \\ \\ For each branch,

150: the corresponding positions $r$ of resonances are \begin{eqnarray} \beta_1& =

151: &2, \;\; r = 0, m-1, -(m+2),\nonumber\\ \beta_2& = &-m, \;\; r =

152: 0, m+2, 2m+1,\nonumber\\ \beta_3& = &m+1, \;\; r = 0, 1-m,

153: -(2m+1). \end{eqnarray} For every branch there exists the common resonance

154: $r=0$. For $\beta_1$ and $\beta_3$, at least one of the resonances

155: always stands in a negative position. The second branch, $\beta_2

156: = -m$, has two positive resonances for every value of $m$. Hence,

157: the second branch is  generic: the expansions (2) with (3)

158: represent the general solutions near  singularity. Next, we find

159: from (1) the recursion relations for the coefficients $u_r(t) \;

160: (r=0,1,2,\ldots)$ of the expansions (2). We see that the

161: resonances occur at $r=-1,4,6$. The resonance at $r=-1$

162: corresponds to the arbitrariness of $\phi(x,t)$. For the other

163: resonances $(r=4,6)$ the recursion relations turn out to be

164: consistent if \begin{equation} a_{tt}(t)a(t)-3a_t^2(t)=0. \end{equation}

165: On the other

166: hand, the recursion relations for $v_r(t)$ are depend on $m$. For

167: each value of $m$, there exist different recursion relations with

168: different resonances, but $r=0$ is common for all those cases. We

169: check every case up to $m=15$ and find the following cases: \begin{eqnarray}

170: m&=&0 , \;\; r=-1,0,1,2,4,6;\nonumber\\ m&=&2 , \;\;

171: r=-1,0,4,4,5,6;\nonumber\\ m&=&3 , \;\; r=-1,0,4,5,6,7; \end{eqnarray}

172: for which, the compatibility conditions are automatically satisfied

173: for each resonances and the system (1) passes the Painlev\'{e}

174: test if $a(t)$ satisfies (7). We find that equation (7) has the

175: solution \begin{equation}

176: a(t)=\pm[-2c_1t+2c_2]^{-1/2}, \end{equation} where $c_1$ and

177: $c_2$ are integration constants. It follows that the system (1)

178: possesses the Painlev\'{e} property if $a(t)= k$ and

179: $a(t)=\frac{k}{\sqrt{t}}$, where $k$ is any non-zero constant.

180: 

181: For $m=3$, it is obvious from (5) that $a(t)=b(t)$ in the system

182: (1). The case $a(t)=b(t)=\frac{k}{\sqrt{t}}$ corresponds to the

183: perturbation system of the cylindrical KdV (cKdV) equation.$^4$

184: For the other case,

185: $a(t)=b(t)=k$, the system (1) is the perturbative KdV system and

186: is studied in Ref. 5. In the following sections we study two of the

187: systems (1), corresponding to $m=2$, in detail.

188: 

189: \subsection{Jordan KdV Systems}

190: \noindent

191: We consider the system of equations

192: \begin{eqnarray}

193: u_t&=&u_{xxx}+ 2kuu_x,\nonumber\\

194: v_t&=&v_{xxx}+ k (uv)_x,

195: \end{eqnarray}

196: which corresponds to $m=2$ and passes the Painlev\'{e} test.

197: Actually, this system of equations is known as a Jordan KdV system

198: and was studied in Refs. 6 and 7.

199: To gain more information on (10), we define the transformations

200: by truncating the series expansions (2) on constant level as follows:

201: \begin{equation}

202: u=\frac{u_0}{\phi^2}+\frac{u_1}{\phi}+u_2, \;\;\;\;\;

203: v=\frac{v_0}{\phi^2}+\frac{v_1}{\phi}+v_2,

204: \end{equation}

205: where $u_2(x,t)$ and $v_2(x,t)$ satisfy equations (10) and can be

206: choosen as  $u_2(x,t)=0$ and $v_2(x,t)=0$. Inserting  above expressions

207: for $u$ and $v$ into system (10) and setting the coefficients of

208: each power of $\phi$ to zero, we have

209: \begin{equation}

210: u_0=-\frac{6}{k}\phi_x^2, \;\;\;\;\;\;\; u_1=\frac{6}{k}\phi_{xx},

211: \end{equation}

212: \begin{eqnarray}

213: \phi_t-4\phi_{xxx}+3\frac{\phi_{xx}^2}{\phi_x}&=&0,\\

214: \phi_{xxxx}-2\frac{\phi_{xx}\phi_{xxx}}{\phi_x}+\frac{\phi_{xx}^3}

215: {\phi_x^2}&=&0, \end{eqnarray}

216: \begin{eqnarray}

217: v_1=-\frac{v_{0,x}}{\phi_x}+\frac{\phi_{xx}}{\phi_x^2}v_0,\\

218: v_{0,xx}-3\frac{\phi_{xx}}{\phi_x}v_{0,x}-2(\frac{\phi_{xxx}}

219: {\phi_x}-2\frac{\phi_{xx}^2}{\phi_x^2})v_0&=&0,\\

220: v_{0,t}-(2\frac{\phi_{xxx}}{\phi_x}-\frac{\phi_{xx}^2}{\phi_x^2})

221: v_{0,x}&=&0. \end{eqnarray}

222: By introducing $\psi(x,t)$, such that

223: $\phi_x=\psi^2$, equations (13) and (14) can be written as \begin{eqnarray}

224: \psi_t-4\psi_{xxx}&=&0,\\

225: \left(\frac{\psi_{xx}}{\psi}\right)_x&=&0. \end{eqnarray}

226: These equations

227: have solutions of the form

228: \begin{equation}

229: \psi(x,t)= c_1 e^{[\beta(t)+ \alpha(t)(x-1)]}+c_2 e^{-[\beta(t)+

230: \alpha(t)(x-1)]}, \end{equation} and

231: \begin{equation}

232: \psi(x,t)=c_1 x+c_2 \end{equation} where $\beta(t)= \int \alpha(t) dt$, $c_1$

233: and $c_2$ are constants.

234: 

235: As a special case, we can choose $\alpha(t)=\alpha=constant$ and

236: $c_1=c_2=1$, so that the solution (20) can be written as

237: \begin{equation}

238: \psi(x,t)=2\cosh \alpha \theta, \end{equation} where $\theta(x,t)=x+4\alpha^2

239: t$. The corresponding solution for $\phi(x,t)$ is,

240: \begin{equation}

241: \phi(x,t)=\frac{1}{\alpha}\sinh 2\alpha \theta +

242: 2\theta + 16\alpha^2 t.

243: \end{equation}

244: Using this solution we see that equations (16) and (17) are satisfied

245: if

246: \begin{equation}

247: v_{0,t}-4\alpha^2 v_{0,x}=0.

248: \end{equation}

249: Then,

250: \begin{equation}

251: v_0(x,t)=f(x+4\alpha^2t)=f(\theta), \end{equation} where $f$ is an arbitrary

252: function of its argument. Now, using (12), (15), (23) and (25) in

253: (11) we obtain the exact solutions of Jordan KdV system (10),

254: \begin{eqnarray}

255: u(x,t)&=&-\frac{24}{k}\left[\left(\frac{1+\cosh 2\alpha \theta}

256:        {\frac{1}{\alpha}\sinh 2\alpha \theta +

257:        2\theta + 16\alpha^2 t }\right)^2

258:        -\frac{\alpha \sinh 2\alpha \theta}

259:        {\frac{1}{\alpha}\sinh 2\alpha \theta +

260:        2\theta + 16\alpha^2 t}\right],\nonumber \\ \nonumber\\

261: v(x,t)&=&\frac{f(\theta)}{(\frac{1}{\alpha}\sinh 2\alpha \theta +

262:        2\theta + 16\alpha^2 t)^2}

263:        -\frac{\left[\frac{f^{\prime}

264:        (\theta)}{2(1+\cosh 2\alpha \theta)}-\frac{\alpha f(\theta)

265:        \sinh 2\alpha \theta}{(1+\cosh2 \alpha \theta)^2}\right]}

266:        {(\frac{1}{\alpha}\sinh 2\alpha \theta +

267:        2\theta + 16\alpha^2 t)}.

268: \end{eqnarray}

269: These functions can be plotted by using $Mathematica$.$^8$

270: Some results are given in $Figures(1)$ and $(2)$.

271: 

272: Next, we consider the solution in (21) and find the expression for

273: $\phi(x,t)$ as

274: \begin{equation}

275: \phi(x,t)= c_1^2 \left(\frac{x^3}{3}-4t\right)+ c_1 c_2 x^2 +

276: c_2^2 x \end{equation} that leads to the rational solutions \begin{eqnarray}

277: u(x,t)&=&-\frac{6}{k}\left(\frac{\phi_x^2}{\phi^2}-

278: \frac{\phi_{xx}}{\phi}\right)^2,\nonumber\\ v(x,t)&=& d_1

279: \left(\frac{\phi_x^2}{\phi^2}- \frac{\phi_{xx}}{\phi}\right)+ d_2

280: \left(

281: \frac{\phi_x^{3/2}}{\phi^2}-\frac{\phi_{xx}\phi_x^{-1/2}}{2\phi}\right)

282: \end{eqnarray} of Jordan KdV system (10) where $d_1$ and $d_2$ are arbitrary

283: constants.

284: 

285: \subsection{Non-autonomous Jordan KdV Systems}

286: \noindent

287: As a

288: second example, we consider the system of equations \begin{eqnarray}

289: u_t&=&u_{xxx}+ \frac{2}{\sqrt {t}} uu_x,\nonumber\\ v_t&=&v_{xxx}+

290: \frac{1}{\sqrt {t}} (uv)_x, \end{eqnarray} which corresponds to the case

291: $m=2$ in (8) and passes the Painlev\'{e} test. This system of

292: equations is known as non-autonomous Jordan KdV system and is

293: given in Ref. 9. Inserting the expansions

294: \begin{equation}

295: u=\frac{u_0}{\phi^2}+\frac{u_1}{\phi}, \;\;\;\;\;

296: v=\frac{v_0}{\phi^2}+\frac{v_1}{\phi}, \end{equation} into (29) and setting

297: the coefficients of each power of $\phi$ to zero, we obtain

298: \begin{equation}

299: u_0=-6\sqrt{t}\phi_x^2, \;\;\;\;\;\;\;  u_1= 6\sqrt{t}\phi_{xx},

300: \end{equation}

301: \begin{eqnarray}

302: \phi_t-4\phi_{xxx}+3\frac{\phi_{xx}^2}{\phi_x}&=&0,\\

303: \phi_{xxxx}-2\frac{\phi_{xx}\phi_{xxx}}{\phi_x}+\frac{\phi_{xx}^3}

304: {\phi_x^2}+\frac{\phi_x}{6t}&=&0, \end{eqnarray}

305: \begin{eqnarray}

306: v_1=-\frac{v_{0,x}}{\phi_x}+\frac{\phi_{xx}}{\phi_x^2}v_0,\\

307: v_{0,xx}-3\frac{\phi_{xx}}{\phi_x}v_{0,x}-2(\frac{\phi_{xxx}}

308: {\phi_x}-2\frac{\phi_{xx}^2}{\phi_x^2})v_0&=&0,\\

309: v_{0,t}-(2\frac{\phi_{xxx}}{\phi_x}-\frac{\phi_{xx}^2}{\phi_x^2})

310: v_{0,x}+\frac{5}{6t}v_0&=&0. \end{eqnarray}

311: The equations (32) and (33) are

312: compatible, i.e.$(\phi_{xxxx})_t = (\phi_t)_{xxxx}$, and can be

313: solved by the substitution $\phi_x=\psi^2$, where \begin{eqnarray}

314: \psi_t-4\psi_{xxx}&=&0,\nonumber\\ \psi_{xx}+\left[\frac{x}{12

315: t}+\alpha(t)\right]\psi&=&0. \end{eqnarray}

316: However, the last equation can

317: only be solved  in terms of Airy functions.$^{10}$ The result is

318: \begin{equation}

319: \psi(x,t)=(1/t)^{1/3}\left[c_1 Ai(z)+ c_2 Bi(z)\right] \end{equation}

320: where

321: \begin{equation}

322: z(x,t)=\frac{[(-1/t)^{1/3}(x+12c_0)]}{2^{2/3}3^{1/3}}, \;\;\;\;

323: \alpha(t)= \frac{c_0}{t} \end{equation} and $c_0$, $c_1$, $c_2$ are constants.

324: The corresponding solution for $\phi(x,t)$ is \begin{eqnarray}

325: \phi(x,t)&=&(1/t)^{2/3}(12c_0+x)[c_1 Ai(z)+ c_2 Bi(z)]^2

326: \nonumber\\ &+&

327: 2^{2/3} 3^{1/3}(1/t)^{1/3}[c_1

328: Ai^{\prime}(z)+ c_2 Bi^{\prime}(z)]^2. \end{eqnarray}

329: Then,

330: \begin{eqnarray}

331: u&=& 6\sqrt{t} (\ln

332: {\phi})_{xx} \nonumber\\ v&=&

333: \frac{v_0}{\phi^2}-\frac{1}{\phi}\left(\frac{v_0}{\phi_x}\right)_x,

334: \end{eqnarray} are the exact solutions of the system of equations (29) if

335: (35) and (36) are satisfied. Note that, these equations  are

336: linear in $v_0$ and a particular solution is $v_0=C\sqrt{t}

337: \phi_x^2$ with  $C=constant$. In this particular case, $v$ is

338: proportional to $u$, i.e. $v=-(C/6)u$. This implies that the

339: system of equations in (29) reduces to a cKdV equation after the

340: transformation $u \rightarrow \sqrt{t}$ $u$.$^{11-13}$ In Ref.13,

341: a hierarchy of solutions for the cKdV equation is derived in terms

342: of Airy functions. These solutions can be obtained from (41)

343: together with (40).

344: 

345: 

346: \nonumsection{Acknowledgements}

347: \noindent

348: We are grateful to Dr. Sergei Sakovich for his guidence

349: and useful comments. We also acknowledge the referee for his

350: careful reading and suggestions on the manuscript.

351: This work is supported in part by

352: the Scientific and Technical Research Council of Turkey (TUBITAK).

353: 

354: 

355: \nonumsection{References}

356: \noindent

357: 

358: \begin{thebibliography}{000}

359: 

360: \bibitem{1}

361: R. Conte (Ed.), {\it The Painlev\'{e} Property. One century

362: later} {(CRM Series in Mathematical Physics, Springer-Verlag,

363: New York, 1999)}.

364: 

365: \bibitem{2}

366: J. Weiss, M. Tabor, G. Carnevale,

367: {\bibit Journal of Mathematical Physics} {\bibbf 24}, 522 (1983).

368: 

369: \bibitem{3}

370: M. Jimbo, M. D. Kruskal, T. Miwa, {\bibit Physics Letters} {\bibbf

371: A92}, 59 (1982).

372: 

373: \bibitem{4}

374: W. X. Ma, B. Fuchssteiner,

375: {\bibit Chaos Solitons Fractals} {\bibbf 7}, 1227 (1996).

376: 

377: \bibitem{5}

378: S. Yu. Sakovich,

379: {\bibit Journal of Nonlinear Mathematical Physics} {\bibbf 5}, 230 (1998).

380: 

381: \bibitem{6}

382: S. I. Svinolupov, {\bibit Functional Analysis and Its Applications}

383: {\bibbf 27}, 257 (1994).

384: 

385: \bibitem{7}

386: A. (Kalkanl{\i}) Karasu, {\bibit International Journal of Theoretical

387:  Physics} {\bibbf 36}, 705 (1997).

388: 

389: \bibitem{8}

390: S. Wolfram, {\it The Mathematica Book (4th ed.)}

391: {(Cambridge University Press, Cambridge, 1999)}.

392: 

393: \bibitem{9}

394: M. G\"{u}rses, A. Karasu, R. Turhan, {\bibit Journal of Physics}

395: {\bibbf A34}, 5705 (2001).

396: 

397: \bibitem{10}

398: M. Abramowitz, I. A. Stegun,

399: {\it Handbook of Mathematical Functions}

400: {(Dover Publ., 1965)}, p.446.

401: 

402: \bibitem{11}

403: W. H. Steeb, M. Kloke, B. M. Spieker, W. Oevel, {\bibit Journal of Physics}

404: {\bibbf A16}, l447 (1983).

405: 

406: \bibitem{12}

407: H. Dai, {\bibit Journal of the Physical Society of Japan}

408: {\bibbf 68}, 1854 (1999).

409: 

410: \bibitem{13}

411: N. C. Freeman, G. Horrocks, P. Wilkinson,

412: {\bibit Physics Letters} {\bibbf A81}, 305 (1981).

413: 

414: \end{thebibliography}

415: 

416: \end{document}

417: