1: \documentstyle[prl,aps,amsmath,graphicx,epsfig,psfig,
2: pstricks]{revtex}
3:
4: \newpsobject{showgrid}{psgrid}{subgriddiv=1,griddots=10,gridlabels=7pt}
5:
6:
7: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
8: %% A few handy macros
9: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
10: \DeclareMathOperator{\I} {i}
11: \newcommand{\diff}[2]{\partial_#2 #1}
12: \newcommand{\abs} [1]{\lvert#1\rvert}
13: \newcommand{\nabn}{\nabla^2}
14: \newcommand{\eno} [1]{\mbox{\!(\ref{#1})}}
15:
16: \begin{document}
17: \draft
18: \twocolumn
19: \title{Self-organized stable pacemakers near the onset of birhythmicity}
20: \author{Michael Stich, Mads Ipsen, and Alexander S. Mikhailov}
21: \address{Fritz-Haber-Institut der Max-Planck-Gesellschaft, Faradayweg
22: 4-6, D-14195 Berlin, Germany}
23: \date{\today}
24: \maketitle
25:
26: \begin{abstract}
27: General amplitude equations for reaction-diffusion systems
28: near to the soft onset of birhythmicity described by a
29: supercritical pitchfork-Hopf bifurcation are derived. Using
30: these equations and applying singular perturbation theory, we
31: show that stable autonomous pacemakers represent a generic
32: kind of spatio\-temporal patterns in such systems. This is
33: verified by numerical simulations, which also show the
34: existence of breathing and swinging pacemaker solutions. The
35: drift of self-organized pacemakers in media with spatial
36: parameter gradients is analytically and numerically
37: investigated.
38: \end{abstract}
39:
40: \pacs{82.40.Bj, 82.40.Ck, 82.20.Wt}
41:
42: %\twocolumn
43:
44: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
45: %% Introduction.
46: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
47:
48: Oscillatory reaction-diffusion systems, such as the
49: Belousov-Zhabotinsky (BZ) chemical reaction, exhibit a rich
50: variety of nonlinear wave patterns. The first complex pattern
51: discovered in this reaction was the target pattern, where
52: concentric waves were emitted by a pacemaker representing a
53: periodic wave source~\cite{Zaikin70}. Subsequently, similar
54: target structures have been observed in many other chemical,
55: physical, and biological
56: systems~\cite{JakRo90,AssStein93,NaSaSa89,Lee97}. A simple
57: theoretical explanation of target patterns in oscillatory
58: chemical systems is that their pacemakers are created by
59: impurities which increase the local oscillation frequency in
60: the medium~\cite{TyFi80}. Most of the target patterns seen in
61: the BZ~reaction are indeed caused by small local
62: inhomogeneities, such as dust particles. A general question is
63: whether self-organized target patterns, representing an
64: intrinsic dynamical property, are also possible in
65: reaction-diffusion systems. Examples of stable autonomous
66: pacemakers with localized or extended wave patterns are indeed
67: known for several reaction-diffusion
68: models~\cite{Vasiliev86,Mik92,KawCom92,Saka92,Kokubo94,DeiBra94,ZhDoEp95,OhHaKo96,RoZhEp97,NekShe98}.
69:
70: The aim of the present Letter is to show that stable autonomous
71: pacemakers with extended wave patterns represent a {\em generic\/}
72: pattern-forming object in oscillatory reaction-diffusion systems near
73: the onset of {\em bi\-rhyth\-micity}. Birhythmicity, the coexistence
74: of two stable limit cycles corresponding to uniform oscillations with
75: different frequencies, is possible in various
76: systems~\cite{DeGold82,AlEp83,LaHu85,PerezI98}, including glycolytic
77: oscillations~\cite{Gold96} and the photosensitive BZ
78: reaction~\cite{KrKu90}. Here, we derive two coupled amplitude
79: equations yielding the normal form of such a dynamical system near a
80: supercritical pitchfork-Hopf bifurcation which leads to birhythmicity.
81: Using singular perturbation theory, an analytical solution for
82: autonomous pacemakers is then constructed and its stability is
83: numerically confirmed. In addition, target patterns with breathing or
84: swinging pacemakers are observed. Finally, we show that autonomous
85: pacemakers can drift under the influence of a parameter gradient and
86: determine the drift velocity.
87:
88: The derivation of normal forms for various kinds of reaction-diffusion
89: systems has recently been discussed by one of the
90: authors~\cite{Trans99,IpsKraSo00}. Here, we focus our attention on
91: systems near to a supercritical pitchfork-Hopf bifurcation, where a
92: real uniform eigenmode and a pair of complex conjugate uniform
93: eigenmodes start to grow simultaneously. This implies that either a
94: stable small-amplitude limit cycle becomes unstable and gives rise to
95: two stable limit cycles or a pair of stable and unstable limit cycles
96: of small amplitude emerges near to a stable limit cycle. At least
97: three species are needed to realize this bifurcation.
98:
99: Using the approach described in~\cite{Trans99}, we have derived the
100: normal form of this bifurcation for a general reaction-diffusion
101: system~\cite{IpsStiMik01}. The normal form is given by
102: \begin{subequations}
103: \begin{align}
104: \label{eq:ScaledEqn-A}
105: \diff{A}{t} &= A - (1 + \I\alpha )\abs{A}^{2} A +
106: (1+\I\beta )\nabn A + (1 - \I \epsilon)A z, \\
107: \label{eq:ScaledEqn-B}
108: \tau \diff{z}{t} &= \sigma - \gamma \abs{A}^2 + z -\nu z^3
109: + \lambda^{2}\nabn z,
110: \end{align}
111: \label{eq:ScaledEqn}
112: \end{subequations}
113: The system~\eno{eq:ScaledEqn} represents a complex
114: Ginzburg-Landau equation (CGLE) for a supercritical Hopf
115: bifurcation~\eno{eq:ScaledEqn-A} which is coupled to an
116: equation describing an imperfect pitchfork
117: bifurcation~\eno{eq:ScaledEqn-B}. Here $A$ is the complex
118: oscillation amplitude and $z$ is the amplitude of the slow
119: real mode. The coefficients $\tau$ and $\lambda$,
120: respectively, are the ratios of the characteristic time and
121: length scales of the real and the oscillatory mode. The
122: parameter $\epsilon$ specifies the frequency shift of the
123: oscillatory mode due to coupling to the real mode, $\gamma $
124: characterizes the strength of the feedback from the
125: oscillatory to the real mode, and $\nu$ determines the
126: nonlinear saturation of the real mode. When $\sigma =0$,
127: Eq.~\eno{eq:ScaledEqn-B} describes a supercritical pitchfork
128: bifurcation, whereas $\sigma \neq 0$ corresponds to an
129: imperfect pitchfork (or ``cusp''~\cite{Kutz95})
130: bifurcation. Only positive parameters $\gamma ,\nu$, and
131: $\sigma $ will be considered in this Letter. We assume that uniform
132: oscillations are modulationally stable in this system, i.e.\
133: the Benjamin-Feir-Newell condition $1+\alpha \beta >0$ is
134: satisfied, and that the waves have positive dispersion, i.e.\
135: $\beta -\alpha >0$.
136:
137: It can be seen that Eqs.~\eno{eq:ScaledEqn} have solutions
138: corresponding to two uniform stable limit-cycle oscillations
139: with frequencies $\Omega _{1}$ and $\Omega _{3}$ and to an
140: unstable limit cycle with frequency $\Omega _{2}$. The
141: frequencies are given by $\Omega _{1,2,3}=\alpha +\left(
142: \alpha +\epsilon \right) z_{1,2,3}$ where $z_{1,2,3}$ are the
143: real roots of the equation $\nu z^{3}-(1-\gamma )z+\gamma
144: =\sigma $. When $\alpha +\epsilon <0$ (which is the case that
145: we chose in the simulations), the smallest root $z_{1}$
146: corresponds to the most rapid oscillations, i.e.\ $\Omega
147: _{3}<\Omega _{2}<\Omega _{1}$.
148:
149: In addition to these uniform oscillations, the system
150: described by Eqs.~\eno{eq:ScaledEqn} may have stable
151: nonuniform solutions representing self-organized
152: pacemakers. To create a pacemaker, a sufficiently strong local
153: perturbation should be applied to the state comprised of
154: uniform oscillations with the lower frequency $\Omega_{3}$, so
155: that a small core region is formed where oscillations have a
156: higher frequency. Inside this region, the variable $z$ is
157: close to $z_{1}$, whereas outside, it is near to $z_{3}$. The
158: core starts to send out waves and hence a pacemaker is
159: created. The core expands ($\gamma >\sigma$ is a necessary
160: condition for this, otherwise it will contract) and the
161: frequency and the wavenumber of the emitted waves slowly
162: increase at the same time. In turn, this leads to a decrease
163: of the oscillation amplitude of emitted waves. This amplitude
164: controls the propagation velocity of the front, representing
165: the boundary of the expanding core. When a critical wavenumber
166: is reached, the front velocity becomes zero and a stationary
167: pacemaker is formed.
168:
169: We have constructed an analytical solution for stationary
170: pacemakers in the one-dimensional
171: system~(\ref{eq:ScaledEqn}). The phase $\phi$ and amplitude
172: $\rho $ are introduced by \mbox{$A=\rho\exp[-\mbox{i}(\Omega_3
173: t+\phi)]$}. Then $\rho$ is adiabatically eliminated using
174: \mbox{$\rho ^{2}\approx 1+z-(\nabla \phi )^{2}+\beta \nabla
175: ^{2}\phi $}. The phase equation approximation~\cite{Kuramoto}
176: is valid for smooth phase perturbations and $\nu \gg
177: 1-\gamma$. Assuming that the characteristic length scale of
178: the real mode, determining the front width, is much shorter
179: than the characteristic length scale of the oscillatory
180: subsystem (i.e.\ that $\lambda \ll 1$), we apply singular
181: perturbation theory to this problem. The derivation will be
182: published separately~\cite{StIpMi2} and only selected results
183: are reported in this Letter.
184:
185: The velocity of the front $V$ (of the expanding core) depends
186: on the wavenumber $k$ and is given by
187: \begin{equation}
188: V(k)=3\frac{\lambda }{\tau }\sqrt{\frac{\nu }{2}}\widetilde{z}_{2}(k),
189: \label{V}
190: \end{equation}
191: where $\widetilde{z}_{2}(k)$ is the middle root of the cubic
192: equation \mbox{$\nu z^{3}-\left( 1-\gamma +\gamma a\right)
193: z+\gamma (1+az_{3}-k^{2})=\sigma$}, with \mbox{$a=\beta
194: (\alpha +\epsilon )/(1+\alpha \beta )$.} On the other hand, if
195: the core radius $R$ is known, the wavenumber $k$ of emitted
196: waves can be analytically found if the condition $\lambda
197: k\ll1$ is satisfied (cf.~\cite{Mik92}). The inverted
198: dependence $R(k)$ and the velocity $V(k)$ are displayed in
199: Fig.~\ref{fig1}(a).
200:
201: For stationary pacemakers, the front velocity $V$
202: vanishes. This determines the wavenumber $k_{0}$ of a
203: stationary pacemaker and thus allows us to find its core
204: radius $R_{0}$. Analytical solutions for these key properties
205: have been constructed. We find that
206: \begin{eqnarray}
207: k_{0}&=&\sqrt{1-\sigma /\gamma +az_{3}}\,, \label{k0} \\
208: R_{0}&=&\frac{1+\alpha \beta }{(\beta -\alpha )\sqrt{k_{{\rm {max}}
209: }^{2}-k_{0}^{2}}}\,{\rm tan}^{-1}\!\left( \frac{k_{0}}{\sqrt{k_{{\rm {max}}
210: }^{2}-k_{0}^{2}}}\right),
211: \end{eqnarray}
212: where $k^2_{{\rm {max}}} = (\alpha+\epsilon)
213: (z_{1}-z_{3})/(\beta-\alpha)$. The frequency $\Omega _{0}$ of
214: a stationary pacemaker is $\Omega _{0}=\Omega _{3}+(\beta
215: -\alpha )k_{0}^{2}$. The wavenumber $k_{0}$ and the radius
216: $R_{0}$ of a stationary pacemaker are shown as functions of
217: the coupling coefficient $\gamma$ in Fig.~\ref{fig1}(b).
218:
219: Examining the constructed solutions, we note that generally
220: $\Omega_{3}<\Omega _{0}<\Omega _{1}$. The frequency $\Omega
221: _{0}$ of a stationary pacemaker approaches the frequency
222: $\Omega _{1}$ of rapid uniform oscillations, when the core
223: radius $R_{0}\rightarrow \infty$ (and $k_{0}\rightarrow
224: k_{\max }$). On the other hand, when the core is small,
225: $k_{0}$ is small and the frequency $\Omega _{0}$ is close to
226: $\Omega _{3}$. Stationary pacemakers exist inside an interval
227: of the coupling intensity $\gamma$ [see
228: Fig.~\ref{fig1}(b)]. Our approximate analysis based on
229: singular perturbation theory is only valid when the core is
230: not too small, i.e.\ $R_{0}\gg \lambda$.
231:
232: Some conclusions about the stability of stationary pacemakers
233: can already be drawn from Fig.~\ref{fig1}(a). Suppose the
234: radius $R$ has increased above the stationary radius
235: $R_{0}$. This leads to an increase of the wavenumber $k$ of
236: emitted waves which, in turn, will make the front velocity $V$
237: negative. Therefore the front will retreat, decreasing the
238: radius $R$ back to its stationary value. This argument is,
239: however, only applicable when the characteristic time scale of
240: the core evolution is much longer than the time needed for the
241: wave pattern to adjust to its changes, i.e.\ when $\tau \gg
242: 1$. Generally, the stability of stationary pacemakers should
243: be numerically investigated.
244:
245: The system described by Eqs.~\eno{eq:ScaledEqn} was integrated
246: with an explicit Euler scheme where the Laplacian operator was
247: discretized with a nearest-neighbor approximation. No-flux
248: boundary conditions were used. Figure~\ref{stable1d1}(a)
249: displays the evolution of a stationary pacemaker from a small
250: initial perturbation of the real mode $z$. In the first stage,
251: the core grows with approximately constant speed. Later, the
252: growth is terminated and a stationary object is
253: formed. Figure~\ref{stable1d1}(b) displays the creation and
254: emission of waves in the oscillatory subsystem. The profile of
255: the asymptotic stable stationary pacemaker is shown in
256: Fig.~\ref{stable1d1}(c).
257:
258: Pacemakers are stable for sufficiently large $\tau$. When
259: $\tau$ is decreased, numerical integrations show that
260: stationary pacemakers become unstable. Close to the
261: instability boundary, stable breathing and swinging pacemakers
262: were found [Figs.~\ref{comb1d1}(a)--\ref{comb1d1}(b)]. For a
263: breathing pacemaker, the center remains stationary whereas the
264: radius oscillates. For swinging pacemakers, the radius stays
265: approximately constant while the position of the pacemaker
266: oscillates. Further lowering of $\tau$ leads to the
267: disappearance of any stable pacemaker solutions.
268:
269: Similar to plane waves in the CGLE~\cite{JanPum92}, the waves
270: emitted by a pacemaker may become unstable when their
271: wavenumber $k_0$ exceeds the Eckhaus wavenumber given by
272: $k_{{\rm {EH}}}\approx \sqrt{(1+z_{3})(1+\alpha \beta
273: )(3+\alpha \beta +2\alpha ^{2})^{-1}}$. Among other effects,
274: this may lead to the destabilization of a stationary
275: pacemaker, as illustrated by
276: Figs.~\ref{comb1d1}(c)--\ref{comb1d1}(d). Phase singularities
277: and thus oscillation amplitude defects are periodically
278: generated at the core boundary, giving rise to a
279: short-wavelength regime in the core and to a long-wavelength
280: regime in the periphery. The core gradually grows and
281: eventually the whole medium is occupied by rapid uniform
282: oscillations.
283:
284: In contrast to pacemakers which are created by local
285: heterogeneities, autonomous pacemakers are not pinned and
286: their location is determined only by the initial
287: conditions. Moreover, such self-organized structures are able
288: to move through the medium when spatial parameter gradients
289: are present. Suppose, for example, that the parameter $\gamma
290: $ varies with a constant gradient $\kappa$, i.e.\
291: $\gamma(x)=\gamma _{0}+\kappa (x-x_{0})$, where $\gamma _{0}$
292: is the value of $\gamma$ in the center of the pacemaker. For
293: sufficiently small gradients ($\kappa R_{0}\ll \gamma _{0}$),
294: linear perturbation theory can be used. Its application
295: (see~\cite{StIpMi2}) allows us to determine analytically the
296: drift velocity $V_{{\rm {D}}}$ as \mbox{$V_{{\rm {D}}}=\kappa
297: R_{0}\, \partial_{\gamma} V(\gamma_0,k_0)$} where $V(\gamma
298: ,k)$ is the front velocity given by Eq.~\eno{V}.
299:
300: The simulation displayed in Fig.~\ref{drift1d1}(a) was
301: initiated with a stable stationary pacemaker. After a constant
302: gradient in the parameter $\gamma$ was introduced, the
303: pacemaker drifted through the medium in the direction of
304: increasing $\gamma$. When the gradient was removed, the drift
305: of the pacemaker terminated and a spatially shifted stationary
306: pacemaker was recovered. The emission and propagation of waves
307: persisted during the drift [Fig.~\ref{drift1d1}(b)]. In
308: addition, the Doppler effect led to a small increase of $k$ in
309: the direction of motion.
310:
311: In this Letter, we have analytically constructed
312: self-organized pacemaker solutions in the vicinity of a
313: pitch\-fork-Hopf bifurcation. Our numerical investigations
314: have shown that such self-organized patterns are stable for a
315: wide range of parameters. To create autonomous pacemakers, a
316: sufficiently strong local perturbation should be applied to
317: the state corresponding to stable uniform oscillations. This
318: is in contrast to~\cite{RoZhEp97}, where autonomous target
319: patterns were found near a Hopf bifurcation with a finite
320: wavenumber and thus uniform oscillations of the medium were
321: absolutely unstable. Our approach is also different from the
322: model~\cite{Saka92,Kokubo94} that was constructed to explain
323: target pattern formation in electrohydrodynamic convection and
324: which is based on a Hopf bifurcation of a cellular spatial
325: structure. On the other hand, Ohta {\em et
326: al.}~\cite{OhHaKo96} have investigated a two-component
327: activator-inhibitor model with coexistence of excitable
328: kinetics and stable uniform oscillations, and reported several
329: different kinds of autonomous wave sources. The subsequent
330: numerical studies~\cite{OhtaPC} have, however, shown that
331: while localized target patterns are stable, target patterns
332: which extend over the whole medium are unstable in the model
333: and slowly evolve into uniform oscillations. Stable localized
334: target patterns are also found in the quintic
335: CGLE~\cite{DeiBra94}.
336:
337: Since our analysis is based on general amplitude equations,
338: the results presented here are valid for any
339: reaction-diffusion system near a soft onset of birhythmicity
340: with small-amplitude limit cycles. In a separate
341: publication~\cite{IpsStiMik01}, this analysis will be applied
342: to a particular chemical model system. As in the case of a
343: Turing-Hopf bifurcation, the results of our analysis based on
344: the amplitude equations may remain (qualitatively) applicable
345: even at significant separation from the bifurcation point.
346: Finally, we note that the physical mechanism responsible for
347: the stabilization of pacemakers in the considered system
348: involves a long-range negative feedback, similar to the one
349: necessary for the formation of stable localized spots in
350: reaction-diffusion models with fast inhibitor diffusion. Here,
351: however, an infinite-range inhibition is caused not by
352: diffusion, but by non-damped propagation of waves emitted from
353: the core region. The effect of pacemaker drift in systems with
354: spatial parameter gradients provides a convenient experimental
355: method to identify self-organized pacemakers and distinguish
356: them from other target patterns caused by local
357: heterogeneities in the medium.
358:
359: The authors thank T. Ohta for an interesting discussion and
360: H. Engel for bringing Ref.~\cite{KrKu90} to our
361: attention. Financial support of the Humboldt Foundation
362: (Germany) is gratefully acknowledged.
363:
364:
365:
366: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
367: %% Bibliography.
368: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
369: \bibliographystyle{prsty}
370: \bibliography{main}
371:
372:
373:
374: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
375: %% Figures.
376: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
377: \begin{figure}[htbp]
378: \noindent
379: \begin{pspicture}(0,0)(8.5,3.3)
380: \rput[bl](0.0,0.0){%
381: \includegraphics[width=8.65cm]{fig1.eps}
382: }
383: \end{pspicture}
384: \caption{%
385: (a) Dependence of the front velocity $V$ on the wavenumber
386: $k$ of emitted waves (solid line) and dependence of $k$ on
387: the core radius $R$ (dashed line, plotted as $R$ {\em vs.}
388: $k$). (b) The wavenumber $k_{0}$ of the waves emitted by a
389: stationary pacemaker (solid line) and the corresponding
390: radius $R_{0}$ (dashed line) as functions of the coupling
391: coefficient $\gamma$. The parameters are $\alpha =1.4$,
392: $\beta=2.3$, $\epsilon =-2.1$, $\lambda=1$, $\tau =5$,
393: $\nu =20$, $\gamma=0.13$, $\sigma=0.1$.}
394: \label{fig1}
395: \end{figure}
396:
397:
398:
399: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
400: % FIG 2
401: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
402: \begin{figure}[htbp]
403: \noindent
404: \begin{pspicture}(0.0,0)(8.5,5.4)
405: \rput[bl](0.05,3.05){%
406: \includegraphics[width=8.5cm]{fig2ab.eps}%
407: }
408: \rput[bl](0.05,0.25){%
409: \includegraphics[width=9.1cm]{fig2c.eps}
410: }
411: \end{pspicture}
412: \caption{%
413: Development of a stable stationary pacemaker (a,b) and its
414: asymptotic profile (c). Frame (a) shows the evolution of
415: the real mode amplitude $z$ after an initial
416: perturbation. Frame (b) displays the evolution of
417: Re$A$. In frame (c) the spatial distribution of the
418: variables $z$ (solid line), Re$A$ (dotted line) and $|A|$
419: (dashed line) are presented. The system size is $L=100$
420: and the time interval is $0<t<500$; the same parameters as
421: in Fig.~\protect\ref{fig1}. In our gray-scale plots the
422: black and white levels always correspond to the minimum
423: and the maximum values of the plotted variable,
424: respectively. In the space-time plots, time runs along the
425: horizontal axis.}
426: \label{stable1d1}
427: \end{figure}
428:
429:
430: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
431: %% FIG 3
432: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
433: \begin{figure}[htbp]
434: \noindent
435: \begin{pspicture}(0.0,0)(8.5,4)
436: \rput[bl](0.0,0.1){
437: \includegraphics[width=8.5cm]{fig3.eps}
438: }
439: \end{pspicture}
440: \caption{%
441: Breathing (a), swinging (b), and Eckhaus-unstable (c-d)
442: pacemakers. The displayed coordinate and time ranges are
443: $\Delta L=50$, $\Delta T=125$ (a-b) and $\Delta L=100$,
444: $\Delta T=625$ (c-d). The parameters are $L=100$,
445: $\alpha=1.38$, $\epsilon =-3.18$, $\lambda=0.8$, $\nu=83$,
446: $\gamma=5.59\cdot 10^{-4}$, $\sigma=3.4\cdot 10^{-4}$, and
447: for (a): $\beta=3.0, \tau=0.001$, (b):
448: $\beta=2.65,\tau=0.001$, (c-d): $\beta=2.1, \tau=0.025$.}
449: \label{comb1d1}
450: \end{figure}
451:
452:
453:
454:
455: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
456: %% FIG 4
457: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
458: \begin{figure}[htbp]
459: \noindent
460: \begin{pspicture}(0.0,0)(8.5,2)
461: \rput[bl](0.0,0.1){
462: \includegraphics[width=8.5cm]{fig4.eps}
463: }
464: \end{pspicture}
465: \caption{%
466: Drift of a pacemaker. The spatial gradient of the
467: parameter $\gamma $ with $\kappa /\gamma _{0}=0.003$ is
468: applied inside the time interval indicated by vertical
469: dashed lines in frame (a), which shows the evolution of
470: $z$ in the time interval $0<t<2\cdot 10^{5}$. The
471: pacemaker is drifting in the direction of increased
472: $\gamma$. The parameters are $\gamma_0=5.59\cdot 10^{-4}$,
473: $\beta=2.3$, $\tau=2$. The rest are the same as in
474: Fig.~\protect\ref{comb1d1}. Frame (b) displays the
475: drifting wave pattern within a narrow time interval
476: $\Delta T=500$ during the drift, marked by the dotted
477: vertical line in frame (a).}
478: \label{drift1d1}
479: \end{figure}
480:
481:
482: \end{document}
483: