1: \documentstyle[aps,prl,multicol]{revtex}
2: \input{epsf}
3: \begin{document}
4: \draft \title{Drifting Pattern Domains in a Reaction-Diffusion System
5: with Nonlocal Coupling}
6:
7:
8: \author{ Ernesto M. Nicola,
9: Michal Or-Guil, Wilfried Wolf and Markus B\"{a}r}
10:
11: \address{ Max-Planck-Institut f\"ur Physik komplexer Systeme, \\
12: N\"othnitzer Stra{\ss}e 38, D-01187 Dresden, Germany }
13: \date{May 19, 2000}
14: \maketitle
15:
16:
17: \begin{abstract}
18:
19: Drifting pattern domains (DPDs), {\it i.e.} moving
20: localized patches of traveling
21: waves embedded in a stationary (Turing) pattern
22: background and {\it vice versa}, are observed in simulations of a
23: reaction-diffusion model with nonlocal coupling.
24: %
25: Within this model, a region of bistability between Turing patterns and
26: traveling waves arises from a codimension-2 Turing-wave bifurcation
27: (TWB).
28: %
29: DPDs are found within that region in a substantial distance from the TWB.
30: %
31: We investigated the dynamics of
32: single interfaces between Turing and wave patterns.
33: %
34: It is found that DPDs exist due to a locking of the interface velocities,
35: which is imposed by the absence of space-time defects near these interfaces.
36: %
37: \end{abstract}
38:
39:
40:
41:
42: \pacs{PACS numbers:
43: 03.40.Kf, %Waves and wave propagation: general mathematical aspects
44: 47.54.+r, %Pattern selection
45: 82.40.Ck %Pattern formation in vortices diffusion systems}]
46: }
47:
48: \begin{multicols}{2}
49:
50:
51:
52: %\newpage
53:
54: %\subsection{Introduction}
55:
56: {\em - Introduction - }
57: Pattern forming processes in nonequilibrium systems
58: can be classified according to the primary instability of the
59: spatially homogeneous state.
60: %
61: Ref. \cite{Cross-Hohenberg} distinguishes three basic types of
62: instabilities in unbounded systems:
63: {\it (i)} spatially periodic and stationary in time,
64: {\it (ii)} spatially periodic and oscillatory in time and {\it (iii)}
65: spatially homogeneous and oscillatory in time.
66: %
67: Within the reaction-diffusion literature, these instabilities are known
68: as Turing, wave and Hopf bifurcation, respectively.
69: %
70:
71: Many chemical and biological patterns are well captured by so called
72: activator-inhibitor models \cite{Mikhailov} describing the
73: dynamics of two reacting and diffusing substances with two coupled
74: partial differential equations.
75: %
76: In such two component reaction-diffusion models only Turing and Hopf
77: instabilities are possible.
78: %
79: Recently, numerical investigations of chemical
80: reaction-diffusion systems with three components \cite{Zhabotinsky}
81: and nonlocal coupling \cite{MexPRL} have yielded the occurrence of wave
82: instabilities and the corresponding patterns.
83: %
84: A universal description of patterns near these instabilities
85: is achieved within the framework of amplitude equations
86: \cite{Cross-Hohenberg,MvHecke}.
87: %
88: %
89:
90: Here, we study a simple FitzHugh-Nagumo model with
91: inhibitory nonlocal coupling that is obtained as a limiting case
92: of a three component reaction-diffusion system.
93: %
94: It describes the interaction of an activator species with
95: an inhibitor.
96: %
97: For slow inhibitor diffusion (compared to the activator diffusion),
98: the model exhibits wave instabilities,
99: while, for fast inhibitor diffusion, Turing instabilities are found.
100: %
101: The two instabilities occur simultaneously at a codimension-2 Turing-wave
102: bifurcation (TWB).
103: %
104: Such a situation has been found earlier within a model for
105: binary convection \cite{Schoepf} and is a generalization of the well
106: investigated Turing-Hopf instability in reaction-diffusion systems
107: \cite{TuHo}.
108: %
109: Basic properties of a TWB have been studied
110: theoretically in amplitude equations \cite{Walgraef} as well as
111: experimentally in a one-dimensional gas-discharge system \cite{Purwins}.
112: %
113: In our model, we find a pattern previously unknown in
114: %
115: \begin{figure}[t]
116: \label{FIG 1}
117: \epsfxsize=80mm
118: \centerline{\epsffile{fig1.eps}}
119: \vspace{2mm}
120: \caption[]{Space-time plots of the field $u$ in grey scale for three examples of
121: DPDs found in numerical simulations of Eqs.\ (1).
122: In all three cases $a=6.0$, and only a part of the system of length $L=409.6$
123: is shown. (a) Large DPD with $\delta=0.84$. (b),(c) DPDs consisting of a
124: single cell of Turing and wave respectively. In (b) $\delta=0.80$ and in (c)
125: $\delta=0.91$. Other parameters, see \cite{codim2}.
126: }
127: \end{figure}
128: %
129: \noindent reaction-diffusion systems: drifting pattern domains (DPDs),
130: {\it i.e.}
131: localized patches of traveling waves embedded in a
132: Turing background and {\it vice versa} (see Fig. 1).
133: %
134: These patches have
135: constant width and move (drift) with constant speed.
136: %
137: As they drift along, maxima of
138: the concentrations of activator and inhibitor are conserved
139: beyond both boundaries of the DPD,
140: where Turing and wave patterns are joined together.
141: %
142: Consequently, formation of space-time defects by coalescence of maxima and
143: minima is prevented.
144: %
145: Similar patterns have been reported in a variety of hydrodynamical
146: experimental systems, see {\it e.g.} \cite{Fless,LiqCol} and have been related
147: to secondary instabilities (parity breaking)
148: of stationary patterns \cite{Fless,GoldPB}.
149: %
150: DPDs exist in a broad region of the para-
151: %
152: \begin{figure}
153: \label{FIG 2}
154: \epsfxsize=80mm
155: \centerline{\epsffile{fig2.eps}}
156: %
157: \caption[]{
158: (a) Critical wavenumber $k_W^c$ against inverse nonlocal coupling range
159: $\sigma$. (b) Real ($\chi$) and imaginary part ($\omega$) of $\lambda(k)$ at the
160: TWB, for parameters see \cite{codim2}.}
161: \end{figure}
162: %
163: \noindent meter space, but appear only in a substantial distance to the onset of pattern formation.
164: %
165: Their existence region is characterized by
166: bistability between waves and Turing patterns.
167: %
168: Near the boundary of DPD existence small DPDs containing only a
169: single wave or Turing ,,cell'' inside a background of the respective other state are encountered
170: (see Figs. 1b,c).
171: %
172: Outside the DPD existence region,
173: the width of pattern domains shrinks or expands;
174: these transient domains exhibit defects at the interfaces.
175: %
176: Large DPDs are composed of two interfaces
177: separating wave and Turing patterns.
178: %
179: In this Letter we will study the dynamics of such interfaces.
180: %
181: These interfaces typically select the wavenumber of the invading domain and
182: form one parameter families characterized by the wavenumber
183: of the invaded domain.
184: %
185: Far away from the TWB, interfaces can exhibit a locking mechanism
186: of their velocities due to the absence of defects.
187: %
188: This locking implies that the interface velocity is fixed by
189: the wavenumbers and frequency of the patterns that they separate.
190: %
191: If both interfaces in a DPD are locked,
192: they have to travel with equal speed.
193: %
194: This mechanism is responsible for the existence of DPDs with constant
195: width.
196: %
197:
198:
199: {\em - Model equations and linear stability - }
200: %
201: We study a variant of the FitzHugh-Nagumo equation supplemented by an
202: inhibitory nonlocal coupling in the dynamics of the activator $u$
203: %
204: \begin{eqnarray}
205: \partial_{t}u & = & a u +\beta u^2 - \alpha u^3 -bv
206: + \partial_{x}^2 u\nonumber \\
207: & &- \mu \int_{-\infty}^{+\infty} e^{-\sigma |x-x^{'}|}
208: u(x^{'},t)dx^{'}
209: \nonumber \\
210: \partial_{t}v &=& c u - d v + \delta \partial_{x}^2 v. \label{eq_nlc}
211: \end{eqnarray}
212: %
213: Eqs. (1) represent a limiting case of a three variable model involving the
214: activator $u$, the inhibitor $v$ and an additional fast inhibitor \cite{3var}.
215: %
216: Related three variable models have been introduced previously to describe
217: pattern formation on sea shells and in cell biology \cite{Meinhardt} as well
218: as spot dynamics in gas discharges \cite{Discharge} and concentration patterns
219: in heterogeneous catalysis \cite{Moshe}.
220: %
221: Here, the emphasis is on the onset of pattern formation resulting
222: from destabilization of a single homogeneous steady state.
223: %
224: Eqs. (1) possess the trivial homogeneous fixed point
225: ${\bf u}_0 {\stackrel{{\scriptscriptstyle \rm def}}{=}} (u_0,v_0)^T = (0,0)^T$
226: for all parameter values.
227: %
228: Here, we consider the regime where this fixed point is the only one present,
229: {\it i.e.} $a < bc/d + 2 \mu/\sigma$ and consider perturbations
230: $\propto e^{ikx - \lambda(k) t}$, where $\lambda(k) = \chi(k) + i \omega(k)$.
231: %
232: The growth rates $\lambda(k)$ are given by the eigenvalues of the Jacobian.
233: %
234: Linear stability analysis reveals that Eqs. (1) exhibit wave
235: instabilities if the nonlocal coupling is of sufficiently
236: long range $\sigma < \sigma_c = (2 \mu / (1 + \delta))^{1/3}$.
237: %
238:
239: In the following we vary the control parameters $a$ and $\delta$; the
240: ,,driving force'' $a$ represents the kinetics, whereas the ratio of diffusion
241: coefficients $\delta$ describes the spatial coupling in the medium.
242: %
243: All other parameters of Eqs. (1) have been fixed \cite{codim2}.
244: %
245: For the wave bifurcation,
246: the critical wavenumber $k_W^c$ and parameters $a_W, \delta_W$ are
247: obtained from the condition $\lambda (k_W^c) = \pm i \omega_0$ where the perturbation with $k_W^c$ is the fastest growing mode with
248: $ (k_W^{c})^2 = \sqrt{\frac{2\mu\sigma}{1+\delta}} -{\sigma^2} $.
249: %
250: Note, that for both $\sigma = \sigma_C$ and $\sigma = 0$
251: (global coupling limit) the critical wavenumber is $k_W^c = 0$, see Fig. 2a.
252: %
253: Similarly, a competing Turing instability appears for a critical
254: parameter $a_T$ with a wavenumber $k_T^c$, where the leading eigenvalue
255: $\lambda(k_T^c) = 0$.
256: %
257: For large enough driving $a$,
258: the wave instability appears for small $\delta$, while for
259: large $\delta$ the Turing instability destabilizes the homogeneous
260: state.
261: %
262: For the chosen parameter values, the system exhibits a TWB point (see Fig. 3a and \cite{codim2}).
263: %
264: For the corresponding $\lambda(k)$, see Fig. 2b.
265:
266:
267: {\em - Weakly nonlinear analysis - }
268: Near the TWB, we can write ${\bf{u}}{\stackrel{{\scriptscriptstyle \rm def}}{=}}(u,v)^T $ as a perturbative
269: expansion around $\bf{u}_0$ using a small parameter $\varepsilon$, indicating the distance to
270: the instability threshold: ${\bf{u}}={\bf u}_0 + \varepsilon
271: {\bf{u}}_1+ \varepsilon^2 {\bf{u}}_2+\varepsilon^3{\bf{u}}_3 +\cdots$
272: and use the following multiple scale ansatz:
273: \cite{TwoTimes}
274: %
275: \begin{eqnarray*}
276: {\bf u}_1\!&=&\! [A(\!X\!,\!T_{_1}\!,\!T_{_2}\!){\bf U}_{\!\!A}
277: e^{i(\omega_0 t+k_W^c x)}+ B(\!X\!,\!T_{_1}\!,\!T_{_2}\!)
278: {\bf U}_{\!\!B}e^{i(\omega_0 t-k_W^c x)} \nonumber \\
279: & &+{\cal R}(\!X\!,\!T_{_1}\!,\!T_{_2}\!){{\bf U}_{\!\!{\cal R}}}e^{ik_T^c x} +
280: c.c.]/2. \nonumber
281: \end{eqnarray*}
282: %
283: This leads to a set of coupled equations for the amplitudes $A$, $B$ and
284: $\cal{R}$ for left-, right-going waves and Turing pattern that
285: depend on slow time and space variables.
286: %
287: After reestablishing the original time
288: and space variables and performing further $\varepsilon$-independent
289: scaling, one obtains:
290: %
291: \begin{eqnarray}
292: \partial_{t}{\cal R}&=&\eta {\cal R}
293: - |{\cal R}|^{2}{\cal
294: R} + \xi \partial_{x}^2{\cal R} - \zeta (|A|^{2}+ |B|^{2}){\cal
295: R}
296: \nonumber \\
297: \partial_{t}A+ c_g \partial_{x} A&=& \rho A+ (1+i c_1) \partial_{x}^2 A -
298: (1-ic_3)|A|^{2}A \nonumber \\
299: & &- g (1-ic_2)|B|^{2}A-\nu(1-i\kappa)|{\cal R}|^{2}A
300: \nonumber \\
301: \partial_{t}B- c_g \partial_{x} B&=& \rho B +(1+i c_1) \partial_{x}^2 B-
302: (1-ic_3)|B|^{2}B \nonumber \\
303: & &- g (1-ic_2)|A|^{2}B-\nu(1-i\kappa)|{\cal R}|^{2}B.
304: \end{eqnarray}
305: %
306: For the detailed values of all coefficients, see \cite{coeff}.
307: %
308: Note, that the nonlocal term of Eqs. (1) only enters into the
309: diffusion coefficients of Eqs. (2) and does
310: not give rise to a nonlocal term in Eqs. (2).
311: %
312: Knowledge of the coefficients of Eqs. (2) allows analytical
313: predictions of the pattern dynamics.
314: %
315: Here, traveling waves are always preferred over standing waves ($ g >
316: 1$, see \cite{Cross-Hohenberg}) and bistability between wave and
317: Turing patterns is found ($\nu \zeta > 1$).
318: %
319: In this bistability region in parameter space (see Fig.\ 3a),
320: a family of stable Turing patterns and two
321: families of stable left- and right-traveling waves parametrised by their
322: corresponding wavenumbers coexist.
323: %
324: To get further insight, we take a closer look at the dynamics of
325: single interfaces separating domains of Turing and wave patterns.
326: %
327:
328:
329: {\em - Interface dynamics. -}
330: %
331: With suitable initial conditions,
332: a moving interface between Turing and wave patterns will be formed in
333: simulations of Eqs. (1).
334: %
335: We can distinguish two types of interfaces
336: depending on whether the phase velocity of the waves points towards the interface or
337: away from it.
338: %
339: This classification is independent from the direction in which the interface is moving.
340: %
341: In the following, we will call the first type {\it inward-interfaces} and
342: the second {\it outward-interfaces}.
343: %
344: Fig.\ 3b and 3c show examples of the latter type.
345: %
346:
347: Near the TWB, we have studied general properties of such interfaces
348: in amplitude equations (3) by counting arguments \cite{MvHecke}
349: as well as by direct numerical simulations.
350: %
351: Counting arguments are applied to ordinary differential equations
352: obtained from a coherent structure ansatz in a comoving
353: frame.
354: %
355: We observe that, typically, the wavenumber of the invaded domain remains
356: constant, while it adapts in the invading domain.
357: %
358: In other words, the interface selects a particular wavenumber for
359: the invading state, while the initial wavenumber of the invaded
360: state is a free parameter.
361: %
362: The velocity of the interface is a function of this parameter.
363: %
364: For Turing patterns the selected wavenumber is always the critical one,
365: {\it i. e.} $k_T^{sel} = k_T^c$, while for waves typically
366: $k_W^{sel} \ne k_W^c$ and therefore $\omega^{sel} \ne \omega_0$.
367: %
368: This is valid for both inward- and outward-interfaces.
369: %
370: Thus, we typically have two one-parameter families of interfaces for a
371: given point in parameter space.
372: %
373: %
374: These results are confirmed by numerical simulations of the nonlocal model (1)
375: near the TWB.
376:
377: %
378: Simulations of interfaces in
379: Eqs. (1) far away from the TWB, show qualitatively similar
380: behavior with respect to the selected wavenumbers.
381: %
382: In addition,
383: interfaces far away from the TWB may
384: exhibit a locking mechanism of their velocities, which are
385: fixed by the wavenumbers and frequencies of the Turing and wave domains.
386: %
387: More specifically, the selected velocity is determined by the
388: absence of defects at the interface.
389: %
390: For geometrical reasons an interface without defects, that
391: connects a wave state with wavenumber $k_W$
392: and frequency $\omega$ and a Turing state with $k_T$, has a speed
393: $|v_{lock}|= \omega / (k_T - k_W)$.
394: %
395: This velocity locking mechanism is found for both types of interfaces.
396: %
397: Two examples of areas where locking occurs ({\it locking tongues})
398: in parameter space for inward- and outward-interfaces are given in Fig. 3a.
399: %
400: A locked outward-interface is displayed in Fig. 3b.
401: %
402: Outside the tongue the
403: outward-interfaces display phase slips (see Fig. 3c).
404: %
405: The area of the locking tongues depends only weakly on the front
406: parameter $k_W$ for inward-interfaces and $k_T$ for outward-interfaces.
407: %
408: Note, that the locking tongues open at a sub-
409: %
410: \begin{figure}[t]
411: \label{FIG3}
412: \epsfxsize=70mm
413: \centerline{\epsffile{fig3a.eps}}
414: \vspace{-4mm}
415: \epsfxsize=85mm
416: \centerline{\epsffile{fig3bc.eps}}
417: \vspace{1mm}
418: \caption{(a) Parameter space $a$-$\delta$ near the TWB point (black circle).
419: %
420: The light grey region indicates bistability between the Turing and
421: wave pattern with $k_T^c, k_W^c$ as predicted from Eqs.\ (2).
422: %
423: Dark grey regions correspond to two examples of locking tongues for an
424: outward-interface with $k_T=k_T^c$ (region {\bf OT}) and an
425: inward-interface with $k_W= k_W^c$ (region {\bf IT}).
426: %
427: These two tongues are shown only up to $a\approx 5.6$.
428: %
429: The dashed lines show where the selected velocities of interfaces in
430: Eqs. (2) coincide with $v_{lock}$.
431: %
432: White circles indicate parameter values of simulations shown in (b) and (c).
433: %
434: See Fig. 4 for a description of the dashed regions.
435: %
436: (b) and (c) show space-time plots of $u$ in grey scale
437: from simulations of Eqs.\ (1) showing outward-interfaces for $a=5.2$.
438: %
439: In (b) an example inside the locking tongue is shown for $\delta=0.80$ and in (c) an
440: interface outside the tongue exhibiting defects (inside the white circles)
441: for $\delta=0.86$ is shown.}
442: \end{figure}
443: %
444: \noindent
445: stantial distance from the
446: TWB.
447: %
448: Since the rapidly
449: varying space and time scales have been factored out in the amplitude equations (2),
450: locking tongues cannot be found therein. However on a line in the parameter space (see Fig.\ 3a), the velocity of
451: interfaces in Eqs. (2) coincides with the velocity prescribed by the locking
452: mechanism.
453: %
454: The locking mechanism arises when
455: the characteristic width of the interfaces is of the same order than the
456: characteristic length scale of the patterns.
457: %
458:
459: {\em - DPDs and their Phase Diagram. - }
460: Consider that a large DPD is composed of an inward-interface and an
461: outward-interface, that practically do not interact (see {\it e.g.}
462: Fig.\ 1a).
463: %
464: If both interfaces exhibit no defects and are locked,
465: their velocities have equal magnitude $|v_{lock}|$ but opposite signs.
466: %
467: This ensures constant width and allows construction of DPDs of arbitrary size.
468: %
469: Indeed, the region of existence of large DPDs starts to open
470: where the locking tongues for both interface types begin to
471: overlap (see Fig. 3a).
472: %
473: This is the case for $a \gtrsim 5.7$.
474: Above that
475: %
476: \begin{figure}
477: \label{FIG4}
478: \epsfxsize=70mm
479: \centerline{\epsffile{fig4.eps}}
480: \caption{
481: The region of existence of DPDs obtained in simulations of Eq.\ (1)
482: in the $a$-$\delta$ parameter space shown dashed. The light grey area
483: corresponds to the bistable region between the critical wave and Turing patterns
484: as calculated from the amplitude equations (3). In region {\bf B} DPDs of any
485: size exist; the size being determined only by the initial condition. In
486: region {\bf C} small domains of wave patches traveling
487: in a Turing background are found. In region {\bf A} only
488: Turing-droplets are stable. The three circles correspond to the location
489: of simulations shown in Fig.\ 1.
490: }
491: \end{figure}
492: %
493: \noindent
494: value, DPDs spontaneously form from a variety of initial conditions.
495: %
496: We have determined the parameter region,
497: where they propagate with constant
498: width and drift speed, from extensive simulations in systems with
499: sizes $L > 400$ and periodic boundary conditions. The results are shown in the phase diagram of Fig. 4.
500: %
501:
502: %
503: We can distinguish three different subregions.
504: %
505: In region B, DPDs of any size, with two
506: locked interfaces traveling at the same speed, are found (see Fig. 1a).
507: %
508: In region A, the inward-interface is no longer locked and its speed is
509: smaller than $|v_{lock}|$.
510: %
511: Therefore large domains of Turing (wave) patterns contract (expand)
512: in size until only a stable DPDs containing a single Turing cell is
513: left (see Fig. 1b).
514: %
515: In region C, the outward-interface selects a $k_W^{sel}$ which would be
516: unstable against Turing patterns in an infinite domain.
517: %
518: Therefore, the wave domain forming the DPD is mostly replaced
519: by a Turing pattern.
520: %
521: However, small DPDs with a few wavelength of wave pattern
522: are still encountered.
523: %
524: At the outer boundary of region C, only DPDs with a single wave cell
525: are found to be stable (see Fig. 1c).
526:
527: {\em - Conclusion - }
528: We found a large variety of drifting pattern domains in a
529: reaction-diffusion model with nonlocal coupling.
530: %
531: Their ingredients include a bistability between wave and Turing
532: patterns near a codimension-2 point as well as absence of defects
533: at the interface.
534: %
535: They exist as robust patterns only in a finite distance to
536: the onset of pattern formation.
537: %
538: Our results are not limited to the reaction-diffusion
539: model studied here and should carry over to other physical systems
540: with similar pattern forming instabilities.
541: %
542: Altogether, DPDs and their constituting interfaces
543: represent a generalization of simpler structures such as fronts and
544: pulses in bistable reaction-diffusion systems, which do not simply
545: combine two homogeneous states, but, instead, select their
546: constituents from whole families of possible traveling or
547: stationary periodic patterns.
548: %
549: %
550: %
551:
552: \begin{thebibliography}{99}
553:
554: \vspace{-10mm}
555:
556: % GENERAL
557:
558: \bibitem{Cross-Hohenberg}
559: M. C. Cross and P. C. Hohenberg, Rev. Mod. Phys.
560: {\bf 65},851 (1993).
561:
562: % WAVE BIFURCATION
563:
564: \bibitem{Mikhailov}
565: A. S. Mikhailov, {\em Foundations of Synergetics I},
566: (Springer Verlag, New York, 1990).
567:
568: \bibitem{Zhabotinsky}
569: A. M. Zhabotinsky, M. Dolnik and I. R. Epstein, J. Chem. Phys.
570: {\bf 103}, 10306 (1995).
571:
572: \bibitem{MexPRL}
573: M. Hildebrand, A. S. Mikhailov and G. Ertl, Phys. Rev. Lett. {\bf 81},
574: 2602 (1998).
575:
576: % AMPLITUDE EQUATION AND COUNTING
577:
578: \bibitem{MvHecke}
579: M. van Hecke, C. Storm and W. van Saarloos,
580: Physica D, {\bf 134}, 1 (1999).
581:
582: % CODIMENSION 2 BIFURCATION: THEORY
583:
584: \bibitem{Schoepf}
585: W. Sch\"opf and W. Zimmermann, Phys. Rev. E {\bf 47}, 1739 (1993).
586:
587:
588: \bibitem{TuHo}
589: J. J. Perraud {\it et. al.}, Phys. Rev. Lett. {\bf 71}, 1272 (1993);
590: A. De Wit {\it et. al.}, Phys. Rev. E {\bf 54},
591: 261 (1996); M. Meixner {\it et. al.},
592: Phys. Rev. E {\bf 55}, 6690 (1997);
593: M. Or-Guil and M. Bode, Physica A {\bf 249}, 174 (1998).
594:
595: \bibitem{Walgraef}
596: D. Walgraef, Phys. Rev. E {\bf 55}, 6887 (1997).
597:
598: % CODIMENSION 2 BIFURCATION: EXPERIMENT
599:
600: \bibitem{Purwins}
601: H. Willebrand {\it et. al.}, Contrib. Plasma Phys. {\bf 31}, 57
602: (1991).
603:
604: % PARITY BREAKING INSTABILITY: EXPERIMENT
605: \bibitem{Fless}
606: J. M. Fleselles, A. J. Simon and A. J. Libchaber,
607: Adv. Phys. {\bf 1}, 1 (1991).
608:
609: \bibitem{LiqCol}
610: C. Counillon {\it et. al.},
611: Phys. Rev. Lett. {\bf 80}, 2117 (1998).
612:
613:
614: % PARITY BREAKING INSTABILITY: THEORY
615:
616:
617: \bibitem{GoldPB}
618: R. E. Goldstein {\it et. al.}, Phys. Rev. A
619: {\bf 43}, 6700 (1991).
620:
621: \bibitem{3var}
622: Eqs. (1) are obtained from the three variable system
623: $\partial_{t}u = a u + \beta u^2 - \alpha u^3 - bv - g w
624: + \partial_{x}^2 u;
625: \partial_{t}v = c u - d v + \delta \partial_{x}^2 v;
626: \tau_w \partial_{t}w = e u - f w + \gamma \partial_{x}^2 w$
627: in the limit $\tau_w = 0$.
628: The parameters characterizing the nonlocal coupling in Eqs. (1)
629: are then found as $\sigma = \sqrt{f/\gamma}$ and $\mu = g \sqrt{e^2/\gamma f}$.
630:
631: % MOTIVATION OF THE MODEL
632:
633: \bibitem{Meinhardt}
634: H. Meinhardt and M. Klingler, J. Theor. Biol. {\bf 126}, 63 (1987);
635: H. Meinhardt, J. Cell. Sci. {\bf 112}, 2867 (1999).
636:
637: \bibitem{Discharge}
638: C. P. Schenk {\it et. al.}, Phys. Rev. Lett. {\bf 78}, 3781 (1997),
639: M. Or-Guil {\it et. al.}, Phys. Rev. E {\bf 57}, 6432 (1998).
640:
641: \bibitem{Moshe}
642: M. Sheintuch and O. Nekhamkina, J. Chem. Phys. {\bf 107}, 8165 (1997).
643:
644: \bibitem{codim2}
645: We chose the parameters $b = 4, c = d = 1, \sigma = 1$ and $\mu = 2$,
646: $\alpha=4/3$ and $\beta = 0$.
647: %
648: All results have been checked for $\beta \ne 0$ and do not
649: change qualitatively as long as $|\beta|$ is not too large.
650: The TWB point is found at $a_{TW}= 4 \sqrt{2} -
651: 1$ and $\delta_{TW} = 1$
652: with $k_T^c = \sqrt{2^{3/2}-1} \simeq 1.352 $,
653: $k_W^c =\sqrt{2^{1/2}-1} \simeq 0.644$ and $\omega_0=\sqrt{2}$.
654: %
655: %
656:
657:
658: %
659: % GL comments
660:
661: \bibitem{TwoTimes}
662: Note that {\it two} new time scales ($T_{_1} \propto \varepsilon t$ and
663: $T_{_2} \propto \varepsilon^2 t$) are needed.
664: For a detailed discussion
665: %in the context of amplitude equations for wave instability;
666: see E. Knobloch, in {\em Pattern
667: Formation in Complex Dissipative Systems}, eds. A. Doelman and A. van
668: Harten, (Longman, Burnt Mill, 1995).
669: %
670: %For more details about this topic see \cite{Ernesto}.
671:
672:
673: % GL coefficients
674:
675: \bibitem{coeff}
676: At the TWB point
677: for the parameters given in \cite{codim2}
678: the coefficients of Eqs. (2) are $c_g = 1.1892, c_1 = 0.8536, c_3 = 1,
679: g = 2, c_2 = 1, \nu = 1/2, \zeta = 8, \kappa = 1$ and $\xi
680: = 4.4142$.
681: %
682: The small parameters in Eqs. (2) are defined by
683: $\rho = [((a-a_{TW}) - (k_W^c)^2
684: (\delta - \delta_{TW})]/2$ and $\eta = 2 [(a-a_{TW})+ (k_T^c)^2
685: (\delta - \delta_{TW})/2]$.
686:
687: %\bibitem{Ernesto}
688: %E. Nicola, M. Or-Guil and M. B\"ar, to be published.
689:
690:
691: \end{thebibliography}
692:
693:
694:
695:
696: \end{multicols}
697: \end{document}
698:
699: