1: \documentclass[aps,pre,twocolumn,showpacs]{revtex4}

2: 

3: \usepackage{epsfig}

4: \usepackage{graphicx}

5: \usepackage{amsmath}

6: 

7: \begin{document}

8: 

9: \title{Solutions of Fisher and Burgers' equations with finite

10: transport memory}

11: 

12: \author{Sandip Kar} 

13: 

14: \author{Suman~Kumar~Banik}

15:  

16: \author{Deb~Shankar~Ray}

17: 

18: \email{pcdsr@mahendra.iacs.res.in}

19: 

20: \affiliation{Indian Association for the Cultivation of Science, Jadavpur,

21: Calcutta 700 032, India}

22: 

23: \date{\today}

24: 

25: \begin{abstract}

26: The Fisher and Burgers' equations with finite memory transport, describing 

27: reaction-diffusion and convection-diffusion processes, respectively have 

28: recently attracted a lot of attention in the context of chemical kinetics, 

29: mathematical biology and turbulence. We show here that they admit exact 

30: solutions. While the speed of the traveling wavefront is dependent on the 

31: relaxation time in Fisher equation, memory effects significantly smoothen 

32: out the shock wave nature of Burgers' solution, without making any influence 

33: on the corresponding wave speed. We numerically analyze the ansatz for the 

34: exact solution and show that for the reaction-diffusion system the strength

35: of the reaction term 

36: must be moderate enough not to exceed a critical limit to allow

37: travelling wave solution to exist for appreciable finite memory effect.

38: \end{abstract}

39: 

40: \pacs{87.10.+e, 87.15.Vv, 87.23.Cc, 05.45.-a}

41: 

42: \maketitle

43: 

44: 

45: %\section{introduction}

46: 

47: \noindent

48: {\it Introduction}:

49: A number of nonlinear phenomena in physical \cite{ld}, chemical \cite{ire}

50: and biological processes \cite{jdm,nfb}

51: are described by the interplay of reaction and diffusion or by the interaction

52: between convection and diffusion. The well-known partial differential 

53: equations which govern a wide variety of them are Fisher \cite{raf}

54: and Burgers' \cite{jmb}

55: equations, respectively. While the Fisher equation describes the dynamics of 

56: a field variable subject to spatial diffusion and logistic growth, Burgers'

57: equation provides the simplest nonlinear model for turbulence. Since spatial

58: diffusion is common to all these processes, Fick's law forms the key element

59: in the description of transport. This description however, gets significantly

60: modified when the memory effects are taken into account, i.e., when the 

61: dispersal of the particles are not mutually independent. This implies that the

62: correlation between the successive movement of the diffusing particles may be

63: understood as a delay in the flux for a given concentration gradient. Over the

64: last several years the analysis of memory effects in diffusive processes have

65: attracted a lot of attention 

66: \cite{fort,gr1,gr2,eeh,kph1,kph2,th,vm1,wh1,wh2,mhk,ga,sf,jms,vm4,jf,kbr}

67: in chemical kinetics, mathematical biology and 

68: allied areas. The focal theme lies in the interesting traveling wave front

69: solutions which have been studied extensively by several authors under 

70: various approximations. The object of the present paper is to show that the

71: Fisher equation and the Burgers' equation with finite memory 

72: transport admit exact solutions. We numerically clarify the nature of the 

73: ansatz wherever necessary and analyze the physical implications of 

74: the solutions modified by relaxation effects and the related issues.

75: 

76: %\section{Fisher and Burgers' equations with finite memory transport}

77: 

78: \noindent

79: {\it Fisher and Burgers' equations with finite memory transport}:

80: The starting point of our analysis is the Cattaneo's modification \cite{cc}

81: of Fick's law in the form: 

82: 

83: \begin{equation}

84: \label{eq1}

85: J(x,t+ \tau)= -D \frac{\partial u(x,t)}{\partial x}

86: \end{equation}

87: 

88: \noindent

89: which takes care of adjustment of a concentration gradient at time $t$ with 

90: a flux $J(x,t+ \tau)$ at a later time $(t+ \tau)$, $\tau$ being the delay time

91: of the particles in adopting one definite direction of propagation. Here 

92: $u(x,t)$ denotes the field variable and $D$ is the diffusion 

93: coefficient of the particles.

94: 

95: The population balance equation for the particles on the other hand takes 

96: into

97: account of the conservation equation supplemented by a source function 

98: $k f(u)$ for the particles in the form 

99: 

100: \begin{equation}

101: \label{eq2}

102: \frac{\partial u(x,t)}{\partial t}= -\frac{\partial J}{\partial x} + k f(u)

103: \end{equation}

104: 

105: \noindent

106: The Fisher source function $f(u) = u (1 - u/K)$ has been the subject wide 

107: interest in various context. Here the first term in $f(u)$ signifies the

108: linear growth followed by a nonlinear decay due to the second one; $k$ and

109: $K$ being the growth rate constant of the population and carrying capacity

110: of the environment, respectively. In what follows we shall be considered with

111: two specific cases of the flux-gradient relation(1) for

112: Fisher and Burgers' problem.

113: 

114: 

115: %\subsection{The Fisher equation with nonlinear damping and finite

116: %transport memory}

117: 

118: \noindent

119: {\it A.The Fisher equation with nonlinear damping and finite

120: transport memory}:

121: We start with an expansion of $J$ in Eq.(1) \cite{mkot} upto first order in 

122: $\tau$ to obtain 

123: 

124: \begin{equation}

125: \label{eq3}

126: \tau \frac{\partial J(x,t)}{\partial t} + J(x,t) = 

127: - D \frac{\partial u}{\partial x}

128: \end{equation}

129: 

130: \noindent

131: Here $u(x,t)$ represents the density function. Differentiating (3) with 

132: respect to $x$ and differentiating (2) with respect to $t$ and eliminating 

133: $J$ from the resulting equations one has

134: 

135: \begin{equation}

136: \label{eq4}

137: \frac{{\partial}^2 u}{\partial t^2} + [ \beta - k f'(u)] 

138: \frac{\partial u}{\partial t} = \beta k f(u) 

139:  + w^2 \frac{{\partial}^2 u}{\partial x^2}

140: \end{equation}

141: 

142: \noindent

143: where we have used the following abbreviations

144: 

145: \begin{equation}

146: \label{eq5}

147:  \beta = 1/\tau  \:\:\:\: {\rm and} \:\:\:\:

148:  w^2 = \beta D

149: \end{equation}

150: 

151: \noindent

152: Eq.(4), a hyperbolic reaction - diffusion equation is a generalization of

153: Fisher equation for finite memory transport and nonlinear damping. 

154: It reduces to standard Fisher equation for

155: $\tau$=0. Over the

156: years the equation has drawn wide interest in the context of 

157: traveling wave solutions in various problems

158: \cite{fort,gr1,gr2,eeh,kph1,kph2,th,vm1,wh1,wh2,mhk,ga,sf,jms,vm4,jf,kbr}.

159: For example, Gallay and Raugel \cite{gr1,gr2}

160: have studied the propagation of front solution without the nonlinear

161: term $k f'(u)$. Horsthemke has discussed some

162: related issues in the problem of transport-driven instabilities \cite{wh2}.

163: 

164: We now look for the traveling wave solutions of Eq.(4) of the form

165: $u(x,t)= K U(x-ct) \equiv K U(z)$ with $z= x-ct$, where $c>0$ is the speed of 

166: the nonlinear wave (which, in general, is different from the linear wave $w$

167: dictated by the medium subject to the boundary condition:

168: 

169: \begin{equation}

170: \label{eq6}

171: U(-\infty)=1  \:\:\:\:\: {\rm and} \:\:\:\:\: U(+\infty)=0 \; \; .

172: \end{equation}

173: 

174: \noindent

175: Eq.(4) therefore after some algebra assumes the following form

176: 

177: \begin{eqnarray}

178: & & \frac{{\partial}^2 U}{\partial z^2} + [c(n - A)]

179: \frac{\partial U}{\partial z} - 2 A c U \frac{\partial U}{\partial z}

180: \nonumber \\

181: & & + n A m U(1 - U) = 0 = L(U)   \:\:\:\: (\rm say)

182: \label{eq7}

183: \end{eqnarray}

184: 

185: \noindent

186: where

187: 

188: \begin{equation}

189: \label{eq8}

190: m= w^2 - c^2 \:\:\:\: {\rm and} \:\:\:\: n= \beta / m \; \; .

191: {\rm and} \:\:\:\: A= k/m 

192: \end{equation}

193: 

194: \noindent

195: Following Murray \cite{jdm} we now introduce the ansatz,

196: \begin{equation}

197: \label{eq9}

198: U(z)= \frac{1}{\left [1 + a \exp(b z) \right]^s }

199: \end{equation}

200: 

201: \noindent

202: as a solution to Eq.(7), where $a$, $b$ and $s$ are positive constants to be

203: determined. Using (9) in (7) we obtain after some algebra 

204: 

205: \begin{eqnarray}

206: & &[s(s + 1) a^2 b^2 + n A m a^2 - s[c(n - A)]a^2 b 

207: \nonumber \\

208: & & - s a^2 b^2 ] e^{2 b z} + \left [ 2 a A m n - s a b^2 

209: - s[c(n - A)]a b \right ] e^{b z} 

210: \nonumber \\

211: & &+ n A m - 2 A c s a b e^{b z} (1 + a e^{b z})^{-s+1} 

212: \nonumber \\

213: & & -n A m (1 + a e^{b z})^{-s+2} = 0 = L(U) \; \ .

214: \label{eq10}

215: \end{eqnarray}

216: 

217: \noindent

218: Now for $L(U)=0$ for all $z$, the coefficients of $e^0$, $e^{b z}$, $e^{2 b z}

219: $ and $e^{3 b z}$ within the curly brackets must vanish identically. This

220: implies that $s$=0, 1 or 2. $s$=0 is not a possible solution since $s$ is a

221: positive constant by our starting assumption. For $s$=1 the coefficients of 

222: $e^{b z}$ and $e^{2 b z}$ of Eq.(10) yield the following relations,

223: 

224: \begin{equation}

225: \label{eq11}

226: s(s + 1)b^2 + n A m - s[c (n - A)] b - s b^2 = 0

227: \end{equation}

228: 

229: \begin{equation}

230: \label{eq12}

231: n A m - s b^2 - s[c (n - A)]b - 2 A c s b = 0

232: \end{equation}

233: 

234: \noindent

235: which can be solved to give $b$=0 and $b=-2 A c s/(s + 1)$.

236: 

237: Again since by initial assumption $b$ is a positive constant, both the values

238: of $b$ are unacceptable and $s$=1 is not a correct choice.

239: 

240: For $s$=2 Eq.(10) reduces to a form in which the coefficient of $e^{b z}$, $e^

241: {2 b z}$ and $e^{3 b z}$ must satisfy the following relations

242: 

243: \begin{equation}

244: \label{eq13}

245: s(s + 1) b^2 + 3 n A m -2 s[c(n - A)] b - 2 s b^2 = 0 \; ,

246: \end{equation}

247: 

248: \begin{equation}

249: \label{eq14}

250: s(s + 1) b^2 + n A m - s[c(n - A)] b - s b^2 = 0 \; {\rm and}

251: \end{equation}

252: 

253: \begin{equation}

254: \label{eq15}

255: 2 n A m - s b^2 - s[c(n - A)] b - 2 A c s b = 0 \; .

256: \end{equation}

257: 

258: \noindent

259: From Eq.(13)-(15) we obtain

260: 

261: \begin{equation}

262: \label{eq16}

263: %b^2 = \frac{n A m}{ s(s + 1)}

264: b^2 = n A m/[s(s + 1)]

265: \end{equation}

266: 

267: \noindent

268: and putting $n= \beta/m$, $A= k/m$ from (8) and $s$=2 in (16) we have

269: 

270: \begin{equation}

271: \label{eq17}

272: %b^2 = \frac{\beta k}{6 m} \; \; .

273: b^2 = \beta k/(6 m) \; \; .

274: \end{equation}

275: 

276: \noindent

277: Making use of (17) in (14) we obtain $b$ in terms of $c$ as follows.

278: 

279: \begin{equation}

280: \label{eq18}

281: %c = \frac{5 k \beta }{6 b (\beta - k)} 

282: c = 5 k \beta / [6 b (\beta - k)] 

283: \end{equation}

284: 

285: \begin{equation}

286: \label{eq19}

287: %b = \frac{5}{6 c (\frac{1}{k} - \frac{1}{\beta})}

288: b = 5/\left [6 c (\frac{1}{k} - \frac{1}{\beta}) \right ]

289: \end{equation}

290: 

291: \noindent

292: The exact speed $c$ of the traveling wave can be calculated from (17) and 

293: (19) using $m= w^2 - c^2$ as

294: 

295: \begin{equation}

296: \label{eq20}

297: c = \frac{\sqrt{\beta D}}{\left [ 1 + \frac{6}{25} (y-1/y)^2 \right ]^{1/2} }

298: \end{equation}

299: 

300: \noindent

301: with $y = \sqrt{\frac{\beta}{k}}$.

302: It may be noted that the exact value of $c$ thus derived is always greater

303: than $c_{min}$ where

304: 

305: \begin{equation}

306: \label{eq21}

307: c_{min} = \frac{w}{\left [ 1 + \frac{1}{4} (y-1/y)^2 \right ]^{1/2}} \; \; .

308: \end{equation}

309: 

310: \noindent

311: Again in the diffusive limit, i.e., $1/\beta \rightarrow 0$ or

312: $1/y \rightarrow 0$ the expression (20) results in exact Fisher value of 

313: $c$ as $c = 5 \sqrt{kD}/\sqrt{6}$. It is necessary to stress that this is 

314: not the speed selected by the front ($c = 2 \sqrt{kD}$), but it yields

315: $2.04 \sqrt{kD}$ which is very close to the selected value \cite{jdm}.

316: 

317: \begin{figure}[h]

318: \vspace*{-2.5cm}

319: \includegraphics[width=0.7\linewidth]{Fig1a.ps}

320: \vspace{-2.5cm}

321: \vspace*{-2.5cm}

322: \includegraphics[width=0.7\linewidth]{Fig1b.ps}

323: \vspace{-2.5cm}

324: \vspace*{-2.5cm}

325: \includegraphics[width=0.7\linewidth]{Fig1c.ps}

326: \vspace{-2.5cm}

327: \vspace*{-2.5cm}

328: \includegraphics[width=0.7\linewidth]{Fig1d.ps}

329: \vspace{-2.3cm}

330: \caption{A plot of trvelling wave solutions for different values of  

331: relaxation time $\tau (= \frac{1}{\beta})$ for

332: $k = 0.6$ and $D = 1.0$. The solid lines are due to numerical

333: simulation of Eq.(4) and the dotted lines are the analytic results

334: (22). (a) $\tau = 0.2$, (b) $\tau = 0.4$ (c) $\tau = 0.6$ and (d) $\tau =0.0$ 

335: (Units are arbitrary). }

336: \label{fig1}

337: \end{figure}

338: 

339: Having determined $b$ and $s$ one can write down the exact form of the

340: traveling wave solution (9) for the problem

341: 

342: \begin{equation}

343: \label{eq22}

344: U(z) = 1/

345: \left \{ 1 + a \exp \left [ \left (

346: \frac{5}{c \sqrt{6} (\frac{1}{k} - \frac{1}{\beta})} \right )

347: \frac{z}{\sqrt{6}} \right ] \right \}^2 \; .

348: \end{equation}

349: 

350: \noindent

351: Furthermore $a$ can be determined from the usual condition 

352: $U(z) = 1/2$ for $z=0$. This results in $a= (\sqrt{2} - 1)$.

353: The exact solution of the Fisher equation can be recovered from (22) in the

354: limit $1/\beta \rightarrow 0$ (i.e., $1/y \rightarrow 0$

355: ) using Fisher value of $c = 5 \sqrt{k D}/ \sqrt{6}$. This is given by 

356: 

357: \begin{equation}

358: \label{eq23}

359: U(z) = \frac{1}{ \left [ 1 + (\sqrt{2} - 1) \exp \left ( 

360: \sqrt{\frac{k}{D}} \frac{z}{ \sqrt{6}}\right ) \right ]^2} \; \; .

361: \end{equation}

362: 

363: We thus observe that the effect of memory or finite relaxation time enters 

364: into the dynamics of the reaction-diffusion system through its influence on 

365: the speed of the traveling wave front $c$. We emphasize here that for

366: $\frac{1}{\beta} = 0$ Eq.(22) does not give the solution selected by the

367: front but is much steeper although the speed is very close to the selected      

368: one. 

369: 

370: It is pertinent to point out that although exact the travelling wave solution

371: (22) does not exhaust the possibility of other solutions. This was noted

372: earlier by Murray \cite{jdm} in the context of fisher equation without 

373: memory effect which is a parabolic differential equation. For an understanding

374: of the nature of the travelling wave solution where $\beta = (1/ \tau)$ is a

375: new element of the present theory, we carry out a numerical investigation

376: of Eq.(4) using finite difference method to solve the boundary value problem.

377: The initial condition to integrate numerically is that the front is at rest

378: at $t = 0$. We fix the value of diffusion coefficient $D = 1.0$ for the 

379: entire treatment. In order to allow the variation of $\tau$ for a fixed 

380: value of $k$, we have kept $k$ at 0.6. For a higher value of $k$, i.e, 

381: where the reaction term dominates $\tau$ must be chosen appropriately over

382: a range to generate numerically stable travelling wave front solution. The

383: interplay of $\beta$ and $k$ will be considered in more detail in the later

384: part of this section.

385: 

386: In Fig.1 we compare the analytical (dotted) and the numerical (solid)

387: solutions corresponding to (22) and (4), respectively for different values

388: of $\tau$. From our analysis it is apparent that they agree fairly well for

389: $\tau$ roughly in the range between 0.1 and 0.5. In Fig.1(d) we present the

390: result for $\tau=0$, which corresponds to the typical Fisher case. The 

391: analytical curve is marginally steeper than numerical one. In Fig.2 we 

392: compare the speed of the travelling wave front computed numerically from (4)

393: with that obtained analytically following (20) for several values of $\tau$.

394: It follows that they agree reasonably well when $\tau \leq k$, i.e, in the 

395: range 0.1-0.5. As $\tau$ approaches zero the analytical value of $c$ becomes

396: lower than the numerical one. This implies that the analytical wave front

397: although moves slower is steeper than the numercal one since steepness goes

398: as $ \sim \frac{k}{c}$ as noted by Murray in his earlier analysis. For 

399: higher values of $\tau$ the disagreement between analytical and numerical 

400: values of $c$ grows rapidly.

401: 

402: \begin{figure}

403: \vspace*{-2.5cm}

404: \includegraphics[width=0.7\linewidth]{Fig2.ps}

405: \vspace{-2.3cm}

406: \caption{A plot of the speed $c$ of the travelling wave front solution vs

407: relaxation time $\tau$ (analytical, dotted line; numerical, solid line)

408: for $D = 1.0$, $k = 0.6$ (Units are arbitrary). }

409: \label{fig2}

410: \end{figure}

411: 

412: The above analysis suggests that there is a strong interplay of

413: $k$ and $\tau$ (or $\beta$) in the dynamics so far as the form and stability

414: of the travelling wave front solution is concerned. To explore this aspect

415: more clearly we now carry out an asymptotic analysis of the problem. To this

416: end we return to Eq.(7) subject to boundary condition (6). Following Murray

417: we choose the perturbation parameter $\epsilon = 1/c^2$ and look for the 

418: asymptotic solution for $0 < \epsilon << 1$ by introducing a change of 

419: variable $\xi = \frac{z}{c} = {\epsilon}^{1/2} z$ and $U(z) = g(\xi)$. With

420: these transformations Eq.(7) and (6) therefore reduces to

421: 

422: \begin{equation}

423: \label{eq24}

424: \epsilon \frac{d^2 g}{d {\xi}^2} + (n - A + 2 A g) \frac{dg}{d \xi}

425: + m n A g (1 - g) = 0

426: \end{equation}

427: 

428: \noindent

429: and

430: 

431: \begin{equation}

432: \label{eq25}

433: g(- \infty) = 1 \:\:\:\:\:\: {\rm ;}   \:\:\:\:\:\:\:  g(+ \infty) = 0 \:\: .

434: \end{equation}

435: 

436: \noindent

437: respectively. $\epsilon$ in the highest derivative in Eq.(24) identifies it as

438: a singular perturbation problem.

439: 

440: Making use of a regular perturbation series in $\epsilon$

441: 

442: \begin{equation}

443: \label{eq26}

444: g(\xi ; \epsilon) = g_0(\xi) + \epsilon g_1(\xi) + ....

445: \end{equation}

446: 

447: \noindent

448: in (24) we obtain after equating the appropriate powers of $\epsilon$

449: 

450: \begin{equation}

451: \label{eq27}

452: (n - A + 2 A g_0) \frac{dg_0}{d \xi} = 

453: - m n A g_0 (1 - g_0) \:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\: {\rm ;O(1)} \:\: .

454: \end{equation}

455: 

456: \noindent

457: and

458: 

459: \begin{equation}

460: \label{eq28}

461: (n - A + 2 A g_0) \frac{dg_1}{d \xi} + \frac{d^2 g_0}{d {\xi}^2} +

462: 2 A g_1 \frac{dg_0}{d \xi} 

463: + m n A g_1 (1 - 2 g_0) = 0 \:\:\:\:\:\:\:\:\:\:\:\:\: {\rm ;O(\epsilon)}\:\:

464: \end{equation}

465: 

466: \noindent

467: The lowest order equation (27) when integrated yields 

468: 

469: \begin{equation}

470: \label{eq29}

471: \ln \left \{\frac{(g_0)^{\beta-k}}{(1 - g_0)^{\beta+k}} \right \} =

472: - \beta k \xi + \beta k l

473: \end{equation}

474: 

475: \noindent

476: where $l$ is a constant of integration. Since we are interested in the 

477: solution in the vicinity of $z = 0$, i.e, $\xi = 0$ for which we put 

478: $g_0(\xi) = 1/2$, we obtain

479: 

480: \begin{equation}

481: \label{eq30}

482: l = \frac{1}{\beta k} \ln \left \{\frac{

483: (\frac{1}{2})^{\beta-k}}{(\frac{1}{2})^{\beta+k}} \right \}

484: \end{equation}

485: 

486: \noindent

487: Eq.(29) precludes the possibility of an explicit solution for $g_0(\xi)$.

488: Depending on $\beta$ and $k$ we therefore consider three different cases;

489: 

490: 

491: (i) $\beta>>k$ (or $\tau<<k$):

492: We have from (30) $l = 0$ and (29) reduces to 

493: 

494: \begin{eqnarray}

495: & & g_0(\xi) = (1 + \exp(k \xi))^{-1}

496: \nonumber \\ 

497: or, \:\:\:\:\:\:  & & U(z) = (1 + \exp(k z/c))^{-1} + O(\epsilon)

498: \label{eq31}

499: \end{eqnarray}

500: 

501: \noindent

502: This is the standard assymptotic solution for $U(z)$ for which the effect

503: of memory is negligible.

504: 

505: 

506: (ii) $\beta \approx k$ (i,e $\tau \approx k$):

507: We obtain similarly  from (29) and (30)

508: 

509: \begin{eqnarray}

510: & & g_0(\xi) = \left (1 - \frac{\exp(k \xi/2)}{2}

511: \right )

512: \nonumber \\ 

513: or, \:\:\:\:\:\:  & & U(z) = \left (1 - \frac{\exp((k z)/(2 c))}{2}

514:  \right ) + O(\epsilon)

515: \label{eq32}

516: \end{eqnarray}

517: 

518: \noindent

519: When both $\beta$ and $k$ are small compared to 1 and the exponential term

520: in (32) is small it is easy to put approximately the $O(1)$ term in the

521: form of (31) as 

522: 

523: \begin{equation}

524: \label{eq33}

525: U(z) \approx \left (1 + \frac{\exp((k z)/(2 c))}{2}

526: \right )^{-1}         

527: \end{equation}

528: 

529: 

530: (iii) $\beta<<k$ (i,e $\tau>>k$):

531: We obtain 

532: 

533: \begin{equation}

534: \label{eq34}

535: g_0(\xi) = \frac{1 \pm \sqrt{1 - 

536: \exp(\beta \xi)}}{2} + O(\epsilon)

537: \end{equation}

538: 

539: \noindent

540: The form of this solution is generically different from those of (32) and

541: (31) since it is independent of $k$.

542: 

543: 

544: The three cases discussed above clearly shows that monotonic solutions 

545: satisfying $U(- \infty) = 1$ and $U(\infty) = 0$ for finite wave speed 

546: $(c \ge c_{min})$ exist for the cases (i) and (ii), i,e , when $\tau$ is

547: short but finite; $\tau \le k$. The assertion of this asymptotic analysis is 

548: in clear agreement with our numerical simulation and our choice of a smaller

549: value of $k$ as discussed earlier.

550: 

551: 

552: The aforesaid analysis clearly demonstrates that

553: although the nature of the partial differential equation changes from 

554: parabolic to hyperbolic type due to the inclusion of relaxation time, the 

555: Fisher equation can be solved by Murray's ansatz \cite{jdm} to derive the 

556: exact wave speed and the traveling wave front solution for a suitable range 

557: of relaxation time $\tau$ allowed by the strength of the reaction term. 

558: A compromise between the exact and the numerical solution can be obtained

559: for relatively small reaction terms. The method can be extended further to

560: study other density dependent diffusive processes.

561: 

562: %\subsection{Burgers' equation with finite memory transport}

563: 

564: \noindent

565: {\it Burgers' equation with finite memory transport}:

566: The Burgers' equation \cite{jmb}

567: is a simple model of turbulence which illustrates an

568: interaction between convection and diffusion. The convection incorporates

569: nonlinearity in the dynamics. To include finite memory effect we proceed as 

570: follows:

571: 

572: We start with the following functional relation between flux $J(x,t+ \tau)$

573: at a time $t + \tau$ and the field variable $u(x,t)$ and its gradient term 

574: at an earlier time $t$;

575: 

576: \begin{equation}

577: \label{eq35}

578: J(x,t+ \tau) = \frac{1}{2} {u^2(x,t)} - \gamma 

579: \frac{\partial u(x,t)}{\partial x}

580: \end{equation}

581: 

582: \noindent

583: where $\gamma$ is a constant. Expanding $J$ again upto first order in $\tau$ 

584: and differentiating the resulting equation with respect to $x$ followed by 

585: differentiation of Eq.(2) for $k=0$ (i.e., in the absence of any source term)

586: with respect to time $t$ and elimination of $J$ as done in the last section 

587: we obtain 

588: 

589: \begin{equation}

590: \label{eq36}

591: \frac{\partial u}{\partial t} + u \frac{\partial u}{\partial x}

592: - \tau \frac{{\partial}^2 u}{\partial t^2}

593: = \gamma \frac{{\partial}^2 u}{\partial x^2} \; \; .

594: \end{equation}

595: 

596: \noindent

597: For $\tau$ = 0 Eq.(36) assumes the form of classical Burgers' equation 

598: \cite{ld,jmb} when

599: $u(x,t)$ and $\gamma$ are identified as the velocity field and kinematic 

600: viscosity, respectively.

601: 

602: We now seek a traveling wave solution of the Burgers' equation with memory

603: (36) in the form, $U(z) = u(x-ct)$, $z = x-ct$, where $c$ is again the wave

604: speed to be determined. This results in the following equation:

605: 

606: \begin{equation}

607: \label{eq37}

608: - \left ( \frac{c^2}{\beta} + \gamma \right ) 

609: \frac{{\partial}^2 U}{\partial z^2}

610: + U \frac{\partial U}{\partial z}

611: - c \frac{\partial U}{\partial z} = 0

612: \end{equation}

613: 

614: \noindent

615: where $\beta = 1/ \tau$.

616: 

617: We now impose the bound condition on $U(z)$ that it asymptotically tends to

618: constant values $u_1$ as $z \rightarrow -\infty$ and $u_2$ as 

619: $z \rightarrow +\infty$ and $u_1 > u_2$.

620: 

621: A direct integration of (37) yields 

622:  

623: \begin{equation}

624: \label{eq38}

625: \frac{\partial U}{\partial z} = \frac{1}{2 (\frac{c^2}{\beta} + \gamma)}

626: \left (U^2 - 2 c U - 2 A \right )

627: \end{equation}

628: 

629: \noindent

630: where $A$ is the integration constant. If $u_1$ and $u_2$ are the roots of 

631: the quadratic equation $U^2 - 2 c U - 2 A = 0$, then the wave speed $c$ and

632: the constant $A$ can be obtained as

633: 

634: \begin{equation}

635: \label{eq39}

636: c = \frac{u_1 + u_2}{2} \:\:\:\: {\rm and} \:\:\:\:\: A = - \frac{1}{2}

637: u_1 u_2 \; \; .

638: \end{equation}

639: 

640: \noindent

641: Eq.(38) can then be rewritten in the form 

642: 

643: \begin{equation}

644: \label{eq40}

645: 2 \left  (\frac{c^2}{\beta} + \gamma \right ) \frac{\partial U}{\partial z}

646: = (U-u_1) (U-u_2)

647: \end{equation}

648: 

649: \noindent

650: to integrate to obtain finally 

651: 

652: \begin{equation}

653: \label{eq41}

654: U(z) = \frac{1}{2} (u_1 + u_2) - \frac{1}{2} (u_1 - u_2) 

655: \tanh \left [\frac{z}{4 \delta} \right ]

656: \end{equation}

657: 

658: \noindent

659: where $\delta$ is given by 

660: 

661: \begin{equation}

662: \label{eq42}

663: \delta = \left ( \frac{\frac{c^2}{\beta} + \gamma}{u_1 - u_2} \right )

664: \; \; .

665: \end{equation}

666: 

667: The above analysis shows that the shape of the wave form is 

668: not only affected by kinematic viscosity $\gamma$ by also by an additional

669: contribution $c^2/ \beta$ due to finite relaxation time $\tau (= 1/ \beta)$

670: such that $(c^2/ \beta) + \gamma$ behaves as effective kinematic viscosity. 

671: It is thus apparent that the balance between the steepening effect of the

672: convection as well as smoothening effect due to kinematic viscosity 

673: is enhanced by the presence of the wave speed dependent term

674: $c^2/ \beta$. Thus although wave speed $c \left [ (u_1 + u_2)/2 \right ]$

675: itself remain unaffected by the finite memory effect in contrast to our earlier

676: case of Fisher equation, transmission layer thickness '$\delta$' - which is a

677: measure of shock thickness, increases for higher speed $c$ and relaxation time

678: $\tau$. This implies that as the wave moves faster the shock smoothens out more

679: and more so that the speed dependence of thickness $\delta$ makes the dynamics

680: self-regulating in the problem of interaction between convection and

681: diffusion.

682: 

683: %\section{Conclusions}

684: 

685: \noindent

686: {\it Conclusions}:

687: The existence of relaxation or delay time is an important feature in

688: reaction-diffusion and convection-diffusion systems. In this paper we have

689: shown that two prototypical representatives of these systems a generalized

690: Fisher equation and Burgers' equation can be solved exactly for finite

691: arbitrary delay time using conventional methods. While the wave speed is

692: significantly modified in the Fisher problem for finite memory transport,

693: speed of the traveling wave in the corresponding Burgers' problem remains

694: unaffected, delay time being effective in smoothening out the shock-wave

695: nature of the traveling wave. We also establish numerically that for the 

696: reaction-diffusion system the strength of the reaction term must not exceed

697: a critical limit to allow travelling wave front solutions to exist for

698: appreciable memory or relaxation effect.

699: In view of the fact that the studies on

700: reaction-diffusion and convection-diffusion with finite memory transport have

701: been applied to forest fire \cite{vm4} and population growth models 

702: \cite{vm1}, Neolithic transitions \cite{jf}

703: and in several other areas under various approximate schemes

704: \cite{eeh,kph1,kph2,th,wh1,wh2,mhk,ga,sf,jms}, we

705: believe that the present exact solutions for the generalized Fisher and

706: Burgers' problem are very much pertinent in this context.

707: 

708: \begin{acknowledgments}

709: This work was supported by the Council of Scientific and Industrial 

710: Research (C.S.I.R.), Government of India.

711: \end{acknowledgments}

712: 

713: \begin{thebibliography}{99}

714: 

715: \bibitem{ld} L. Debnath, {\it Nonlinear Partial Differential Equations

716: for Scientists and Engineers} (Birkh\"auser, Boston, 1997).

717: 

718: \bibitem{ire} I.R. Epstein and J.A. Pojman, {\it An Introduction to

719: Nonlinear Chemical Dynamics: Oscillations, Waves, Patterns and Chaos}

720: (Oxford, New York, 1998).

721: 

722: \bibitem{nfb} N.F. Britton, {\it Reaction-Diffusion Equations and their

723: applications to Biology} (Academic, New York, 1986).

724: 

725: \bibitem{jdm} J.D. Murray, {\it Mathematical Biology}, 

726: Second, Corrected Edition (Springer, Berlin, 1993).

727: 

728: \bibitem{raf} R.A. Fisher, Ann. Eugenics {\bf 7}, 353 (1937).

729: 

730: \bibitem{jmb} J.M. Burgers, Adv. Appl. Mech. {\bf 1}, 171 (1948).

731: 

732: \bibitem{fort} J. Fort and V. M\'endez, Rep. Prog. Phys. {\bf 65}, 895 (2002).

733: 

734: \bibitem{gr1} Th. Gallay and G. Raugel, ZAMP {\bf 48}, 451 (1997).

735:  

736: \bibitem{gr2} Th. Gallay and G. Raugel, Preprint patt-sol/9809007;

737: Preprint patt-sol/9812007 .

738: 

739: \bibitem{eeh} E.E. Holmes, Am. Nat. {\bf 142}, 779 (1993).

740: 

741: \bibitem{kph1} K.P. Hadeler, 

742: Can. Appl. Math. Quart. {\bf 2}, 27 (1994).

743: 

744: \bibitem{kph2} K.P. Hadeler, in {\it Reaction Transport Systems in 

745: Mathematics Inspired by Biology}, edited by V. Capasso and O. Diekmann,

746: CIME Lectures, Florence (Springer-Verlag, Berlin, 1998).

747: 

748: \bibitem{th} T. Hillen,

749: Math. Models Methods Appl. Sci. {\bf 8}, 507 (1998).

750: 

751: \bibitem{vm1} V. M\'endez and J. Camacho,

752: Phys. Rev. E {\bf 55}, 6476 (1997).

753: 

754: \bibitem{wh1} W. Horsthemke, Phys. Lett. A {\bf 263}, 285 (1999).

755: 

756: \bibitem{wh2} W. Horsthemke, Phys. Rev. E {\bf 60}, 2651 (1999).

757: 

758: \bibitem{mhk} K.K. Manne, A.J. Hurd and V.M. Kenkre,

759: Phys. Rev. E {\bf 61}, 4177 (2000).

760: 

761: \bibitem{ga} G. Abramson, A.R. Bishop and V.M. Kenkre,

762: Phys. Rev. E {\bf 64}, 066615 (2001).

763: 

764: \bibitem{sf} S. Fedotov, Phys. Rev. Lett. {\bf 86}, 926 (2001).

765: 

766: \bibitem{jms} J.M. Sancho and A. S\'anchez, Phys. Rev. E

767: {\bf 63}, 056608 (2001).

768: 

769: \bibitem{vm4} V. M\'endez and J.E. Llebot,

770: Phys. Rev. E {\bf 56}, 6557 (1997).

771: 

772: \bibitem{jf} J. Fort and V. M\'endez,

773: Phys. Rev. Lett. {\bf 82}, 867 (1999).

774: 

775: \bibitem{kbr} S. Kar, S.K. Banik and D.S. Ray, 

776: Phys. Rev. E {\bf 65}, 061909 (2002).

777: 

778: \bibitem{cc} C. Cattaneo, C. R. Acad. Sci. {\bf 247}, 431 (1958).

779: 

780: \bibitem{mkot} M. Kot, {\it Elements of Mathematical Ecology}

781: (Cambridge University Press, Cambridge, 2001).

782: 

783: \end{thebibliography}

784: 

785: \end{document}

786: