1: %\documentstyle[aps,psfig]{revtex}
2: %\documentstyle[preprint,aps,psfig]{revtex}
3: \documentstyle[aps,prl,psfig,multicol]{revtex}
4: %\baselineskip=12pt
5: \begin{document}
6: \draft
7:
8: \title{Harmonic forcing of an extended oscillatory system: \\
9: Homogeneous and periodic solutions}
10:
11: \author{Jeenu Kim,$^1$ Jysoo Lee,$^{2,3}$ and Byungnam Kahng$^1$}
12: \address{$^{1}$School of Physics and Center for Theoretical Physics,
13: Seoul National University, Seoul 151-742, Korea\\
14: $^{2}$National Creative Research Initiative Center for Neurodynamics
15: and Department of Physics,\\
16: Korea University, Seoul 136-701, Korea\\
17: $^{3}$Supercomputing Research Department,
18: Korea Institute of Science and Technology Information,\\
19: P.O.Box 122, Yusong, Daejon 305-806, Korea}
20: \date{\today}
21: \maketitle
22:
23: \begin{abstract}
24: In this paper we study the effect of external harmonic forcing on a
25: one-dimensional oscillatory system described by the complex Ginzburg-Landau
26: equation (CGLE). For a sufficiently large forcing amplitude, a homogeneous
27: state with no spatial structure is observed. The state becomes unstable
28: to a spatially periodic ``stripe'' state via a supercritical bifurcation
29: as the forcing amplitude decreases. An approximate phase equation is derived,
30: and an analytic solution for the stripe state is obtained, through which
31: the asymmetric behavior of the stability border of the state is explained.
32: The phase equation, in particular the analytic solution, is found to be
33: very useful in understanding the stability borders of the homogeneous
34: and stripe states of the forced CGLE.
35: \end{abstract}
36:
37: \pacs{PACS numbers: 05.45.Xt, 89.75.Kd, 47.20.Ky}
38:
39: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
40: \begin{multicols}{2}
41:
42: \section{Introduction}
43:
44: % motivation
45:
46: Nonequilibrium pattern formation is widely observed in many physical,
47: chemical and biological systems. Significant progresses have been
48: made in this field during the last few decades. For example, it has
49: been found that nonequilibrium patterns can be grouped into a few
50: universality classes \cite{Cross,m90,Walgraef}. In many cases, such a
51: system is in constant interaction with its environment, and
52: understanding the effect of extrinsic perturbation is of great
53: theoretical and practical importance. In particular, it is
54: interesting to study the deformation of an existing pattern or the
55: formation of a new pattern under an external forcing. Our
56: understanding in such a direction is far from complete.
57:
58: % previous experimental results
59:
60: Petrov {\em et al.}, and later Lin {\em et al.}, studied the light
61: sensitive Belousov-Zhabotinsky (BZ) reaction in an oscillatory regime
62: in the presence of a periodic modulation of the intensity of
63: illumination \cite{Petrov,Lin}. They observed ``entrainment bands''
64: in which the system is frequency locked. Different spatial
65: patterns---stationary fronts, standing waves of labyrinth and more
66: complex shapes---are observed within the bands. In a similar BZ
67: reaction setup, Vanag {\em et al.} studied spatial patterns and
68: transitions among them in detail, and observed localized
69: irregular/standing clusters as well as the above patterns
70: \cite{Vanag,Vanag2}.
71:
72: % previous theoretical results
73:
74: Continuum models of forced pattern forming systems can be grouped into
75: ones based on a kinetic model or on an amplitude equation. In the
76: first group, an unforced system is modeled by a coupled kinetic model,
77: such as the Brusselator, Oregonator or FitzHugh-Nagumo model, and
78: parameters in the model are modulated to simulate the effect of
79: external forcing (e.g. \cite{Lin,Steinbock}). On the other hand, near
80: a bifurcation onset of a pattern, small differences among systems
81: become irrelevant, and they are all described by one of a few
82: universal equations. If the bifurcation is supercritical and
83: oscillatory, and if the most unstable wave number is zero, the complex
84: Ginzburg-Landau equation (CGLE) is the equation for the class of
85: systems. In the presence of an external periodic modulation, it is
86: shown that the CGLE with an additional forcing term becomes the
87: appropriate equation \cite{Walgraef,Coullet}. There exist a few
88: studies on forced CGLE, and diverse behaviors are observed depending
89: on several factors, such as the spatial dimension, the mode of the
90: frequency locking, and the behavior of the corresponding unforced
91: system
92: \cite{Walgraef,Coullet,Lega,Elphick,Elphick2,Chate3,Lin3,Park,Gallego}.
93: However, we do not even know what behaviors are possible, let alone
94: understand them.
95:
96: Even the simplest case of the 1:1 locking in one dimension displays a
97: wide variety of behaviors. At large amplitude of the forcing, a
98: homogeneous state with no spatial structure is stable. Chat{\'e} {\em
99: et al.} found that the homogeneous state becomes unstable to a
100: periodic ``stripe'' or ``kink-breeding'' state as the forcing
101: amplitude decreases, and a ``turbulent synchronized'' state---chaotic
102: with its average phase is locked to that of the forcing---can appear,
103: as the amplitude decreases further \cite{Chate3}.
104:
105: % short summary of our results
106:
107: In this paper, we study in detail the homogeneous and stripe states of
108: one dimensional forced CGLE around the 1:1 locking. There are two
109: borders regarding the homogeneous state (1) the stability border,
110: below which the state loses its stability, and (2) the existence
111: border, below which a homogeneous solution does not exist. In
112: general, the existence and stability borders do not coincide. It is
113: known that the stability border of the homogeneous state lacks a
114: reflection symmetry around the $\nu = \alpha$ line. Here, $\nu$ is
115: the difference between the natural and external frequencies, and
116: $\alpha$ is a nonlinear detuning parameter. We find the asymmetry
117: can be explained by the linear stability of the state. Also, the
118: condition under which the existence and stability borders of the
119: homogeneous state coincide is found. The stability border of the
120: stripe state also lacks a reflection symmetry. An approximate phase
121: equation is derived from the forced CGLE, and it is found that its
122: qualitative behavior is identical to that of the original equation, at
123: least in the region of present interest. An analytic expression of
124: the stripe state for the phase equation is obtained, which is used to
125: explain the asymmetry of the border of the stripe state.
126:
127: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
128:
129: \section{Forced Complex Ginzburg-Landau Equation}
130:
131: \subsection{Complex Ginzburg-Landau equation}
132:
133: Near the stability border of a homogeneous state of an extended
134: pattern forming system, the time evolution in a large spatial and
135: temporal scale is given by one of a few universal equations
136: \cite{Cross,m90,Walgraef}. If the instability is oscillatory and
137: supercritical, and the wave number of the most unstable mode is zero,
138: %the complex Ginzburg-Landau equation (CGLE),
139: the CGLE,
140: \begin{equation}
141: \label{Eq:CGLE}
142: \partial_{t} A = A - (1 + i \alpha) |A|^2 A + (1 + i \beta)
143: \nabla^2 A
144: \end{equation}
145: is the governing equation. Here, $A$ is complex amplitude, and
146: $\alpha, \beta$ are real constants. The behavior of the CGLE is
147: relatively well understood, especially in one and two dimensions
148: \cite{Shraiman,Chate1,Chate2}. It has plane wave solutions, which are
149: stable only if $1 + \alpha \beta > 0$. Otherwise, the Benjamin-Feir
150: instability sets in, making the solutions unstable. Near the unstable
151: side of the stability border ($1 + \alpha \beta = 0$ line), ``phase
152: turbulence'' is observed, which is characterized by disordered
153: cellular structure and the absence of a defect ($|A| = 0$). ``Defect
154: turbulence'' is observed further in the unstable region, where
155: constant creation and annihilation of defects is observed
156: \cite{cl88,cgl89}. In this paper, the value of $\alpha = -3/4, \beta
157: = 2$ is mainly used, which is in the phase turbulence region.
158:
159: \subsection{Homogeneous state}
160:
161: Consider the case wherein a sinusoidal forcing is applied to the system
162: of Eq.~(\ref{Eq:CGLE}). It was shown that an additional forcing term
163: should be included, and its form can be determined from the conditions
164: of the spatial and temporal translation invariance \cite{Coullet}.
165: For a harmonic forcing (near the 1:1 tongue), the resulting equation
166: is
167: \begin{equation}
168: \label{Eq:FCGLE}
169: \partial_{t} A = (1 + i \nu)A - (1 + i \alpha) \left|A\right|^2 A
170: +(1+ i \beta) \nabla^2 A + B,
171: \end{equation}
172: where $\nu$ is the difference between the natural and forcing
173: frequencies, and $B$ is related to the amplitude of the forcing.
174:
175: % homogeneous solution
176:
177: We first seek for the homogeneous solution of Eq.~(\ref{Eq:FCGLE}).
178: In polar coordinates [$A = R \exp(i \Phi)$], the equation becomes
179: \begin{eqnarray}
180: \label{Eq:RP}
181: \partial_{t} R & = & R - R^3 + B \cos{\Phi} \nonumber\\ && + R_{xx} - \beta R
182: {\Phi}_{xx} - 2 \beta R_{x} {\Phi}_{x} - R{\Phi}_{
183: x}^2, \nonumber \\
184: R \partial_{t} \Phi & = & \nu R - \alpha R^3 -
185: B \sin{\Phi} \nonumber\\ && + \beta R_{xx} + R {\Phi}_{xx} + 2 R_{x}
186: {\Phi}_{x} - \beta R {\Phi}_{x}^2.
187: \end{eqnarray}
188: For a sufficiently large $B$, the system is expected to lock to the
189: forcing. Its homogeneous solution is
190: \begin{eqnarray}
191: \label{Eq:R0}
192: B \cos{\Phi}_0 & = & - R_0 (1 - R_0^2), \nonumber \\
193: B \sin{\Phi}_0 & = & R_0 (\nu - \alpha R_0^2),
194: \end{eqnarray}
195: which can have one or three roots depending on the parameters. For the
196: three root case, only the one corresponding to the largest $R_{0}$ is
197: stable. The region of the parameter space in which a locked
198: homogeneous solution exists is shown in Fig.~\ref{fig:phase-d}(a).
199:
200: % linear stability of homogeneous solution
201:
202: We apply the linear stability analysis to the homogeneous solution
203: \cite{Chate3}, where the behavior of a small deviation from the
204: solution $r = R - R_{\rm 0}$ and $\phi = \Phi - \Phi_{\rm 0}$ is
205: studied. The growth rate of the mode with wave number $k$ is found to
206: be
207: \begin{eqnarray}
208: \label{Eq:Rate}
209: \lambda(k) &=& 1 -2 R_0^2 -k^2 \nonumber\\
210: &&+ \sqrt {(1+\alpha^2)R_0^4 - (\nu -2\alpha R_0^2 -\beta k^2)^2},
211: \end{eqnarray}
212: which has the maximum value of
213: \begin{equation}
214: \label{Eq:Ratemax}
215: \lambda_{\rm max} = 1 -2 R_0^2 -{1 \over \beta} [\nu -\{ 2
216: \alpha +\sqrt{(1+\alpha^2)(1+\beta^2)} \} R_0^2 ]
217: \end{equation}
218: at $k = k_{\rm max}$, corresponding to the most unstable mode, which
219: is given by
220: \begin{equation}
221: \label{Eq:kmax}
222: \beta k_{\rm max}^2 = \nu -\left( 2\alpha +
223: \sqrt{\frac{1+\alpha^2}{1+\beta^2}} \right) R_0^2.
224: \end{equation}
225: The stability border $B_{\rm s}$ of the homogeneous solution is
226: obtained by solving numerically $\lambda_{\rm max} = 0$, which is also
227: shown in Fig.~\ref{fig:phase-d}(a).
228:
229: % shape of phase diagram
230:
231:
232:
233: A distinct feature of the stability border is that it is not symmetric
234: to the $\nu = \alpha$ line. As shown in the figure, the difference
235: between the existence and the stability border is smaller at the $\nu <
236: \alpha$ side. Moreover, the difference vanishes for $\nu \le \nu_{\rm
237: c}$ with $\nu_{\rm c} \simeq -1.067$. This feature can be understood
238: from the $\nu$ dependence of $k_{\rm max}$, which is given by
239: Eq.~(\ref{Eq:kmax}). It is found that $k_{\rm max}$ is an increasing
240: function of $\nu$: it is zero for $\nu \le \nu_{\rm c} =
241: [2\alpha+\sqrt{(1+\alpha^2)/(1+\beta)}]R_0^2$, and proportional to
242: $\sqrt{\nu - \nu_{\rm c}}$ slightly above $\nu_{\rm c}$. Since the
243: wave number of the most unstable mode is zero for $\nu \le \nu_{\rm
244: c}$, and since the solution Eq.~(\ref{Eq:R0}) with the largest $R_0$
245: is stable to a zero wave number perturbation, the existence of the
246: homogeneous solution guarantees its stability.
247:
248: As will be discussed later, an approximate phase equation is derived
249: from Eq.~(\ref{Eq:FCGLE}), which gives an additional insight into the
250: stability border. The origin of the instability of the homogeneous
251: state of the phase equation can be traced to a Laplacian term, whose
252: coefficient is a decreasing function of $\nu$, and becomes negative at
253: ${\nu}_{\rm c}^{\phi}$. Thus the homogeneous state is stable for $\nu
254: \le \nu_{\rm c}^{\phi}$, and it becomes more unstable as $\nu$
255: increases.
256:
257: \subsection{Stripe state}
258:
259: The behavior below the stability border is investigated numerically.
260: The forced CGLE in one dimension is integrated using a pseudospectral
261: method for various $\nu$ and $B$ \cite{f91}. The spatial resolution
262: $\Delta x$ and time step $\Delta t$ used are $0.1$ and $0.01$,
263: respectively. Also, a periodic boundary condition is used. For most
264: cases, the linear size of the system is chosen to be $4096$, and the
265: time interval of $2 \times 10^{4}$ is used. Larger systems for longer
266: intervals are also studied, and no change in the behavior is observed.
267:
268: The numerical integrations confirm the prediction that the homogeneous
269: state is stable above the stability border. It is found that the
270: state undergoes a supercritical bifurcation to a spatially periodic
271: static ``stripe'' state as $B$ decreases below the border, and the
272: modulation amplitude of the state, defined as $\delta \phi = \sqrt
273: {\langle (\phi - \langle \phi \rangle_{x})^2 \rangle_{x}}$,
274: behaves as $\sqrt{B_{\rm s} - B}$ close to the border
275: [Fig.~\ref{fig:stripe}(a)]. The wave number of the stripe state is
276: found to agree very well with $k_{\rm max}$ of Eq.~(\ref{Eq:kmax}),
277: especially near the border. In order to check how the nature of
278: the transition depends on the unforced dynamics, the transition from
279: a homogeneous state is examined for four different values of
280: $(\alpha,\beta)$: $(-2,2)$, $(-1.11, 1)$, $(-2,0)$, and
281: $(-0.75,0.5)$. It is found that a supercritical transition to the
282: stripe state is observed for the first two cases belonging to the
283: Benjamin-Feir (BF) unstable region, while a transition to a disordered
284: structure is observed for the other two cases belonging
285: to the BF stable region.
286:
287: As $B$ decreases further, the stripe state becomes unstable to a
288: fluctuating stripe or ``kink-breeding'' state, depending on $\nu$
289: \cite{Chate3}. The stability border of the stripe state is
290: determined, and is plotted in Fig.~\ref{fig:phase-d}(a). Again, the
291: border is not symmetric to the $\nu = \alpha$ line. The region of the
292: stripe state is much broader on the $\nu > \alpha$ side. Moreover, it
293: extends to the region where a locked homogeneous state does not exist.
294: The origin of the asymmetry will be discussed using a phase equation,
295: and is found to be very different from the case of the homogeneous
296: state.
297:
298: \section{Phase Equation}
299:
300: \subsection{Derivation}
301:
302: An approximate phase equation can be derived from Eq.~(\ref{Eq:FCGLE})
303: as follows. Define small variables $r = R - R_{0}$ and $\phi = \Phi -
304: \Phi_{0}$, and assume that the time scale for $\phi$ is much larger
305: than that for $r$. The variable $r$ is then slaved to $\phi$.
306: Starting from Eq.~(\ref{Eq:RP}), it can be shown that
307: \begin{equation}
308: \label{Eq:rs}
309: r = {R_0 \over 3R_0^2-1} \left[(1-R_0^2) + {B \over R_0} \cos(\Phi_0 +
310: \phi) - \beta \phi_{xx} - \phi_{x}^{2} \right],
311: \end{equation}
312: where additional terms higher than the second order in $\phi$ are
313: ignored, and $B$ is assumed to be small. Substituting this to the
314: phase part of Eq.~(\ref{Eq:RP}),
315: \begin{eqnarray}
316: \label{Eq:Phase}
317: R_0 \partial_{t} \phi &=& \nu R_0 - \alpha R_0^3 - B \sqrt{1+a^2}
318: \sin(\Phi_0 +\phi+\delta) \nonumber\\ && + b \phi_{x}^2 + c \phi_{xx} + d
319: \phi_{xxxx} + e,
320: \end{eqnarray}
321: where $a, b, c, d, e, \delta$ are constants depending on $\alpha,
322: \beta, \nu$, and $R_0$. Since $R_0$ is $1$ at $\nu = \alpha$, and is a
323: slowly varying function of $\nu$, it is expected that setting $R_0 =
324: 1$ does not change the qualitative behavior of the equation. On the
325: other hand, the constants are simplified to
326: \begin{eqnarray}
327: \label{const}
328: a &=& \alpha - (\nu - \alpha) / 2, \nonumber \\
329: b &=& \alpha - \beta - (\nu - \alpha)/2, \nonumber \\
330: c &=& 1 + \alpha \beta - \beta(\nu-\alpha)/2, \nonumber \\
331: d &=& -\beta^2 / 2, \nonumber \\
332: e &=& 0, \nonumber \\
333: \delta &=& \tan^{-1} (a).
334: \end{eqnarray}
335: For the remainder of the paper, $R_0$ will be set to $1$ in the
336: equation. Note that Eq.~(\ref{Eq:Phase}) is a generalized version of
337: the phase equation obtained by Coullet and Emilsson, which is derived
338: for the special case of $\nu \simeq \alpha$ \cite{Coullet}. Also,
339: $\phi_{xxxx}$ term is added for the stability of the solution in
340: the phase and defect turbulence regions.
341:
342: \subsection{Homogeneous state}
343:
344: The phase equation is studied in a way parallel to the analysis of the
345: forced CGLE. Homogeneous states, given by
346: \begin{equation}
347: \label{Eq:P0}
348: \Phi_0 = \sin^{-1} ({\nu - \alpha \over B \sqrt{1+a^2}}) - \delta
349: \end{equation}
350: exist for $B \ge (\nu - \alpha) / \sqrt{1+a^2}$. There exist two
351: solutions $\Phi_0^{\rm s}$ and $\Phi_0^{\rm u}$, in the $[0, 2\pi]$
352: interval satisfying Eq.~(\ref{Eq:P0}) as shown in
353: Fig.~\ref{fig:flat-sol}. The $\Phi_0^{\rm s}$ solution is stable
354: under homogeneous perturbation, while the $\Phi_0^{\rm u}$ solution is
355: unstable. A linear stability analysis of the stable homogeneous state
356: shows that the maximum growth rate is
357: \begin{equation}
358: \label{Eq:rate}
359: \lambda_{\rm max}^{\phi} = - B \sqrt{1+a^2} \cos(\Phi_0+\delta) -
360: c^2/4d
361: \end{equation}
362: for the mode with wave number $k_{\rm max}^{\phi} = \sqrt{c/2d}$ (if $c
363: \le 0$). This state is found to be linearly stable above the stability
364: border $B_{\rm s}^{\phi}$, which is given as
365: \begin{equation}
366: \label{Eq:Bs}
367: B_{\rm s}^{\phi} = \sqrt{
368: {(c^2/4d)^2 + (\nu-\alpha)^2 \over 1+a^2}}.
369: \end{equation}
370: The existence and stability borders are plotted in
371: Fig.~\ref{fig:phase-d}(b) for $\alpha = -3/4, \beta = 2$. Note that
372: the shapes of the borders are qualitatively the same as those of the
373: forced CGLE: the stability border is asymmetric to the $\nu = \alpha$
374: line, and the two borders meet for $\nu \le \nu_{\rm c}^{\phi}$.
375: Since the wave number of the most unstable mode should be zero at $\nu
376: = \nu_{\rm c}^{\phi}$, one arrives at
377: \begin{equation}
378: \nu_{\rm c}^{\phi} = {2 + 3 \alpha \beta \over \beta}.
379: \end{equation}
380: For the above values of $\alpha$ and $\beta$, $\nu_{\rm c}^{\phi} =
381: -5/4$, which is comparable to the value of $\nu_{\rm c}$ for the
382: forced CGLE.
383:
384: The simple structure of the phase equation makes its interpretation
385: simple. The reason for the instability is that $c$ can be negative,
386: while the $\phi_{xxxx}$ term always tries to suppress such an
387: instability. The value of $c$ remains positive for $\nu < \nu_{\rm
388: c}^{\phi}$, and the homogeneous state is stable. As $\nu$ increases
389: further, $c$ becomes negative. Since $c$ is a decreasing function of
390: $\nu$, the instability becomes stronger with increasing $\nu$, which
391: explains the fact that the difference between the existence and
392: stability border increases with $\nu$.
393:
394: \subsection{Stripe state}
395:
396: The behavior of the phase equation below the stability border is
397: studied numerically. As one crosses the border, the homogeneous state
398: goes through a supercritical bifurcation to a stripe state, and the
399: modulation amplitude $\delta \phi = \sqrt {\langle (\phi - \langle
400: \phi \rangle_{x})^2 \rangle_{x}}$ behaves as $\sqrt{B_{\rm
401: s}^{\phi}-B}$ close to the border. A typical dependence of $\delta
402: \phi$ on $B$ is shown in Fig.~\ref{fig:stripe}(b), where $\nu =
403: -0.75$. As $B$ decreases further, the stripe state becomes unstable.
404: The stability border of the stripe state determined numerically is
405: plotted in Fig.~\ref{fig:phase-d}(b). Again, the border is not
406: symmetric to the $\nu = \alpha$ line, and even extends below the
407: existence border of the homogeneous state. Although the phase
408: equation is simpler than the forced CGLE, their qualitative behaviors
409: are essentially the same, at least for the homogeneous and stripe
410: states.
411:
412: % A1, B2
413:
414: The simple structure of the phase equation allows an analytic
415: expression for the stripe solution. The solution may be expanded in
416: terms of harmonic functions
417: \begin{eqnarray}
418: \label{Eq:series}
419: \phi (x) = \phi_1 && + S_1 \sin{(k_0x)} + C_1 \cos{(k_0x)}
420: \nonumber\\ && + S_2 \sin{(2k_0x)} + C_2 \cos{(2k_0x)},
421: \end{eqnarray}
422: where higher harmonics are ignored. $k_0$ is the wave number of the
423: most unstable mode, and the coefficient $C_1$ can always be set to $0$
424: by choosing an appropriate origin. Substituting it in
425: Eq.~(\ref{Eq:Phase}), we find
426: \begin{eqnarray}
427: \label{Eq:A1B2}
428: && S_2 = 0, \nonumber \\
429: \sin(\Phi_0\!+\!\phi_1\!+\!\delta)~&& =
430: {\nu - \alpha + (S_1^2+4C_2^2) ~bc/4d \over B \sqrt{1+a^2} ~
431: (1-(S_1^2+C_2^2)/4)}, \nonumber\\
432: S_1^2 = 4 C_2 &&~{-B \sqrt{1+a^2} \cos(\Phi_0\!+\!\phi_1\!+\!\delta) + 2c^2/d \over
433: B \sqrt{1+a^2} \sin(\Phi_0\!+\!\phi_1\!+\!\delta) - bc/d}, \nonumber\\
434: C_2 = 2 &&~{-B\sqrt{1+a^2}\cos(\Phi_0\!+\!\phi_1\!+\!\delta) - c^2/4d \over
435: B \sqrt{1+a^2} \sin(\Phi_0\!+\!\phi_1\!+\!\delta) + 2bc/d},
436: \end{eqnarray}
437: which can be solved numerically for $S_1, C_2$ and $\phi_1$. Near the
438: stability border $B_{\rm s}^{\phi}$, an approximate analytic solution
439: can be obtained, which is
440: \begin{eqnarray}
441: \label{Eq:stsol}
442: S_1 & \propto & \sqrt{B_{\rm s}^{\phi} - B}, \nonumber \\ C_2 &
443: \propto & S_1^2, \nonumber \\ \phi_1 & \simeq & {1 \over 4 B
444: \sqrt{1+a^2} \cos (\Phi_0 + \delta)} ~(\nu - \alpha + {bc \over d})
445: ~S_1^2,
446: \end{eqnarray}
447: where the proportionality constants are rather complex except for the
448: case of $\phi_1$. The analytic solution agrees well with the results
449: using numerical integration: as shown in Fig.~\ref{fig:stripe}(b), the
450: modulation amplitude $\delta \phi$ vs $B$ curve obtained from the
451: above expression is in good agreement with the corresponding numerical
452: values.
453:
454: % region of stationary stripes
455:
456: The analytic solution confirms not only the square root dependence of
457: $A_1$, but it also provides an explanation for the asymmetry of the
458: stability border of the stripe solution. Shown in
459: Fig.~\ref{fig:shift}(a) are two homogeneous solutions---stable
460: $\Phi_0^{\rm s}$ and unstable $\Phi_0^{\rm u}$---of the phase equation
461: with $B$ a little below the stability border. As $B$ decreases from
462: the homogeneous toward the stripe region, the modulation amplitude
463: around $\Phi_{0}^{\rm s}$ increases with decreasing $B$. For
464: sufficiently small $B$, $\phi(x)$ at certain $x$ approaches the
465: unstable fixed point $\Phi_0^{\rm u}$, which then makes the stripe
466: solution unstable. Note that $\phi_1$ in Eq.~(\ref{Eq:stsol}) is
467: nonzero---it is negative when $\nu$ is not very different from
468: $\alpha$. Thus, the average phase of a stripe state is shifted toward
469: a value smaller than $\Phi_0^{\rm s}$. The average phase of the
470: stripe solution measured using numerical integration is also shown,
471: which confirms the shift. As shown in Fig.~\ref{fig:flat-sol}, the
472: shift moves the stripe solution toward (away from) $\Phi_0^{\rm u}$
473: for $\nu < \alpha$ ($\nu > \alpha$), resulting in the asymmetry (a
474: related argument is given in \cite{Park}). The situation is entirely
475: similar for the forced CGLE. The phase of the stable and unstable
476: solution is plotted against $\nu$ in Fig.~\ref{fig:shift}(b)
477: \cite{note}. Also plotted is the average phase of the stripe
478: solution. Here again, the average phase is shifted toward (away from)
479: the unstable solution for $\nu < \alpha$ ($\nu > \alpha$).
480:
481: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
482:
483: \section{Conclusion}
484:
485: Despite its simplicity, the forced CGLE displays a large variety of
486: phenomena. The homogeneous and stripe states are mainly discussed
487: here, and the phase equation is found to be very useful in
488: understanding the stability borders of the forced CGLE. For a
489: sufficiently large forcing amplitude, a homogeneous state with no
490: spatial structure is observed. The state becomes unstable to a
491: spatially periodic stripe state via a supercritical bifurcation as the
492: forcing amplitude decreases. We obtained an analytic solution for the
493: stripe state of the phase equation, through which an argument for the
494: asymmetry of the stability border of the state is formulated. As $B$
495: decreases further, more complex behaviors such as, the kink breeding,
496: ``depinning,'' and ``roughening,'' are expected, which are currently
497: under investigation.
498:
499: % Acknowledgement
500:
501: This work was supported in part by grant No. 2000-2-11200-002-3 from
502: the BRP program of the KOSEF. J. Lee was supported by Creative
503: Research Initiatives of the Korean Ministry of Science and Technology.
504: The computation was done on a PC cluster in the Supercomputing Center
505: of Korea Institute of Science and Technology Information. We acknowledge
506: for the generous computer time, and thank Mr. Jeong-Woo Hong for
507: helpful technical assistance.
508:
509: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
510:
511: \begin{thebibliography}{99}
512:
513: \bibitem{Cross} M.C. Cross and P.C. Hohenberg, Rev. Mod. Phys. {\bf
514: 65}, 851 (1993).
515:
516: \bibitem{m90} P. Manneville, {\em Dissipative Structures and Weak
517: Turbulence} (Academic Press, New York, 1990).
518:
519: \bibitem{Walgraef} D. Walgraef, {\em Spatio-Temporal Pattern
520: Formation} (Springer-Verlag, New York, 1997).
521:
522: \bibitem{Petrov} V. Petrov, Q. Ouyang and H.L. Swinney, Nature
523: (London) {\bf 388}, 655 (1997).
524:
525: \bibitem{Lin} A.L. Lin, M. Bertram, K. Martinez, H.L. Swinney,
526: A. Ardelea, and G.F. Carey, Phys. Rev. Lett. {\bf 84}, 4240 (2000).
527:
528: \bibitem{Vanag} V.K. Vanag, L. Yang, M. Dolnik, A.M. Zhabotinsky, and
529: I.R. Epstein, Nature (London) {\bf 406}, 389 (2000).
530:
531: \bibitem{Vanag2} V.K. Vanag, A.M. Zhabotinsky, and I.R. Epstein,
532: Phys. Rev. Lett. {\bf 86}, 552 (2001).
533:
534: \bibitem{Steinbock} O. Steinbock, V. Zykov, and S.C. M{\"u}ller,
535: Nature (London) {\bf 366}, 322 (1993).
536:
537: \bibitem{Coullet} P. Coullet and K. Emilsson, Physica D {\bf 61}, 119
538: (1992).
539:
540: \bibitem{Lega} P. Coullet, J. Lega, B. Houchmanzadeh, and
541: J. Lajzerowicz, Phys. Rev. Lett. {\bf 65}, 1352 (1990).
542:
543: \bibitem{Elphick} C. Elphick, A. Hagberg, and E. Meron,
544: Phys. Rev. Lett. {\bf 80}, 5007 (1998).
545:
546: \bibitem{Elphick2} C. Elphick, A. Hagberg, and E. Meron, Phys. Rev. E
547: {\bf 59}, 5285 (1999).
548:
549: \bibitem{Chate3} H. Chat{\'e}, A, Pikovsky, and O. Rudzick, Physica D
550: {\bf 131}, 17 (1999).
551:
552: \bibitem{Lin3} A.L. Lin, A. Hagberg, A. Ardelea, M. Bertram,
553: H.L. Swinney, and E. Meron, Phys. Rev. E {\bf 62}, 3790 (2000).
554:
555: \bibitem{Park} H.-K. Park, Phys. Rev. Lett. {\bf 86}, 1130 (2001).
556:
557: \bibitem{Gallego} R. Gallego, D. Walgraef, M. San Miguel, and R. Toral,
558: %nlin/0104059 (preprint).
559: e-print nlin-sys/0104059.
560:
561: \bibitem{Shraiman} B.I. Shraiman, A. Pumir, W. van Saarloss,
562: P.C. Hohenberg, H. Chat{\'e}, and M. Holen, Physica D {\bf 57}, 241
563: (1992).
564:
565: \bibitem{Chate1} H. Chat{\'e}, Nonlinearity {\bf 7}, 185 (1994).
566:
567: \bibitem{Chate2} H. Chat{\'e} and P. Manneville, Physica A {\bf 224},
568: 348 (1996).
569:
570: \bibitem{cl88} P. Coullet and J. Lega, Europhys. Lett. {\bf 7}, 511
571: (1988).
572:
573: \bibitem{cgl89} P. Coullet, L. Gil, and J. Lega, Phys. Rev. Lett. {\bf
574: 62}, 1619 (1989).
575:
576: \bibitem{f91} C.A.J. Fletcher, {\it Computational Techniques for
577: Fluid Dynamics} (Springer-Verlag, New York, 1991).
578:
579: \bibitem{note} For the forced CGLE, there exist three homogeneous
580: solutions, instead of two for the phase equation. The third solution,
581: which belongs to the small $|A|$ branch, is irrelevant in the
582: discussion of the stripe solution.
583:
584: \end{thebibliography}
585:
586: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
587:
588: % Figures
589:
590: \begin{figure}[ht]
591: \centerline{\psfig{figure=fig1a.ps,angle=270,width=0.45\textwidth}}
592: \centerline{\psfig{figure=fig1b.ps,angle=270,width=0.45\textwidth}}
593: \caption{(a) The existence and stability borders of the homogeneous
594: state for the harmonically forced CGLE Eq.~(\ref{Eq:FCGLE}) with
595: $\alpha = -3/4$ and $\beta = 2$. The stability border of the stripe
596: state is also shown. Note that the stability borders are not
597: symmetric to the $\nu = \alpha$ line. (b) Corresponding borders for
598: the phase equation Eq.~(\ref{Eq:Phase}) with $\alpha = -3/4$, $\beta =
599: 2$, and $R_0 = 1$.}
600: \label{fig:phase-d}
601: \end{figure}
602:
603: \begin{figure}[ht]
604: \centerline{\psfig{figure=fig2a.ps,angle=270,width=0.45\textwidth}}
605: \centerline{\psfig{figure=fig2b.ps,angle=270,width=0.46\textwidth}}
606: \caption{(a) Modulation amplitude $\delta \phi$ of the stripe
607: solution of the forced CGLE Eq.~(\ref{Eq:FCGLE}) is shown against $B$
608: for $\nu = -0.75$. Also shown is a square root fit $D \sqrt{B_{\rm s}
609: - B}$ with $D = 1.25$ and $B_{\rm s} = 0.02101$. (b) $\delta \phi$ of
610: the stripe solution of the phase equation Eq.~(\ref{Eq:Phase}) is
611: shown for $\nu = -0.75$. Analytic expression Eq.~(\ref{Eq:series})
612: with the coefficients given by Eq.~(\ref{Eq:stsol}) is found to be a
613: good approximation.}
614: \label{fig:stripe}
615: \end{figure}
616:
617: \begin{figure}[ht]
618: \centerline{\psfig{figure=fig3.ps,angle=270,width=0.45\textwidth}}
619: \caption{A schematic view of the determination of the homogeneous
620: solution of the phase equation Eq.~(\ref{Eq:Phase}): There exist one
621: stable $\Phi_0^{\rm s}$ and one unstable $\Phi_0^{\rm u}$ solutions.
622: Here, the one with positive $\cos(\Phi_0+\delta)$ is stable. For $\nu
623: > \alpha, \Phi_0^{\rm u} > \Phi_0^{\rm s}$, while $\Phi_0^{\rm u} <
624: \Phi_0^{\rm s}$ for $\nu < \alpha$.}
625: \label{fig:flat-sol}
626: \end{figure}
627:
628: \begin{figure}[ht]
629: \centerline{\psfig{figure=fig4a.ps,angle=270,width=0.45\textwidth}}
630: \centerline{\psfig{figure=fig4b.ps,angle=270,width=0.45\textwidth}}
631: \caption{(a) The stable $\Phi_0^{\rm s}$ and unstable $\Phi_0^{\rm u}$
632: homogeneous solutions of the phase equation Eq.~(\ref{Eq:Phase}) just
633: below the stability border ($B - B_s^{\phi} = 5 \times 10^{-3}$) are
634: plotted against $\nu$. The average phase of the stripe solution of
635: the equation for the same $B$ is also plotted. The average phase is
636: smaller than $\Phi_0^{\rm s}$, which moves away from (toward)
637: $\Phi_0^{\rm s}$ for $\nu > \alpha$ ($\nu < \alpha$). (b) The same
638: plot for the forced CGLE. The qualitative behavior is identical.}
639: \label{fig:shift}
640: \end{figure}
641:
642: \end{multicols}{2}
643: \end{document}
644:
645: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
646: