1: %\documentclass[eqsecnum,preprint]{revtex4}
2: \documentclass[pre,twocolumn,showpacs]{revtex4}
3:
4: \usepackage{graphicx}
5: \usepackage{amsmath}
6:
7: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
8:
9: \begin{document}
10:
11: \title{Hierarchical structure description of spatiotemporal chaos}
12:
13: \author{Jian Liu$^1$, Zhen-Su She$^{1,2}\footnote{Correspondance: she@pku.edu.cn}$}
14: \affiliation{$^1$ State Key Laboratory for Turbulence and Complex
15: Systems \\
16: and Department of Mechanics and Engineering Science, Peking University, Beijing 100871, P.R. China\\
17: $^2$ Department of Mathematics, UCLA, Los Angeles, CA 90095, USA}
18:
19: \author{Hongyu Guo, Liang Li, Qi Ouyang}
20: \affiliation{Department of Physics, Peking University, Beijing,
21: 100871, P.R. China}
22:
23: \date{\today}
24:
25: \begin{abstract}
26: We develop a hierarchical structure (HS) analysis for quantitative
27: description of statistical states of spatially extended systems.
28: Examples discussed here include an experimental reaction-diffusion
29: system with Belousov-Zhabotinsky kinetics, the two-dimensional
30: complex Ginzburg-Landau equation and the modified FitzHugh-Nagumon
31: equation, which all show complex dynamics of spirals and defects.
32: We demonstrate that the spatial-temporal fluctuation fields in the
33: above mentioned systems all display the HS similarity property
34: originally proposed for the study of fully developed turbulence
35: [Z.-S. She and E. Leveque, Phys. Rev. Lett. {\bf 72}, 336 (1994)].
36: The derived values of a HS parameter $\beta$ from experimental and
37: numerical data in various physical regimes exhibit consistent
38: trends and characterize the degree of turbulence in the systems
39: near the transition, and the degree of heterogeneity of multiple
40: disorders far from the transition. It is suggested that the HS
41: analysis offers a useful quantitative description for the complex
42: dynamics of two-dimensional spatiotemporal patterns.
43: \end{abstract}
44:
45: \pacs{05.45.-a; 82.40.Ck; 47.54.+r; 47.27.-i}
46:
47: \vspace{5pt}
48:
49: \maketitle
50:
51: \section{Introduction}
52:
53: Spatiotemporal chaos have been studied extensively in recent years
54: with different level of quantitative characterization
55: \cite{rev93}. It was discovered that for some spatially extended
56: systems such as the Rayleigh-B\'{e}nard (RB) convection
57: \cite{morris93,assenheimer94}, an ordered state of straight or
58: weakly curved rolls breaks down to a spatiotemporally chaotic
59: state consisting of elementary spiral structures which appear and
60: disappear in an irregular fashion. Spiral patterns driven far from
61: equilibrium have also been studied in chemical reaction-diffusion
62: systems such as Belousov-Zhabotinsky (BZ) reaction \cite{ouyang01}
63: and in biological systems such as cardiac muscle tissue
64: \cite{cardiac}. While an individual spiral is considered as a
65: self-organized topological defect, instabilities can lead to
66: so-called defect-mediated turbulence \cite{coullet89} that is
67: characterized by an exponential decay of correlations with length
68: and time. There have been very fruitful efforts in the
69: theoretical, experimental and numerical studies on the onset phase
70: of the instability which results in the breaking down of spiral
71: waves to the early transition to the disorder states (for a review
72: on the complex Ginzburg-Landau (CGL) system, see
73: Ref.~\cite{cgl02}). The study of the transition via instabilities
74: is important for problems such as the transition from spiral to
75: spiral turbulence in heart tissue which plays an essential role in
76: cardiac arrhythmia and fibrillation \cite{cardiac}.
77:
78: Another question often raised is what quantitative measures are
79: useful when spatiotemporal patterns evolve, especially after
80: spatiotemporal chaos are developed \cite{cgl02,ch-science94}.
81: Attempt to identify defects and conduct statistical study of
82: defects in defect-mediated turbulence has yielded some interesting
83: results. In the discussion on the CGL system, it was argued
84: \cite{coullet89} that the number of defects can be a convenient
85: quantity to characterize spatiotemporal chaos. However, the
86: relation between the finite-time dimension and the number of
87: defects has not been fully established for systems of all sizes
88: and of all duration intervals, and thus the method of separation
89: has not appeared fully convincing \cite{cgl02}. In addition, when
90: spatiotemporal pattern become chaotic, the identification of the
91: basic structure (e.g. the defect) is not always possible, and its
92: statistics become dubious. In general, spirals act more like waves
93: than particles, but sometimes they may display spatially
94: intermittent features. Note that, although the RB convection, the
95: BZ reaction, and the CGL system all exhibit spiral wave behaviors
96: and defect structures, the instabilities leading to complex
97: spatiotemporal behaviors are of different nature. Moreover, local
98: pattern structures may also appear in various forms of spiral,
99: stripe, hexagon, or square. Nevertheless, transitions from ordered
100: to disordered states exist in all these systems, and a common
101: feature of these transitions is the generation of multiple scale
102: fluctuations. Therefore, simple phenomenology independent of the
103: details of the particular system is needed to reveal the essential
104: feature of spatiotemporal chaos ``buried in the wealth of
105: available data" \cite{ch-science94}. The present work propose such
106: a simple phenomenology but with a quantitative measure.
107:
108: There are several other quantitative measures proposed for
109: characterizing disorder in pattern dynamics. Correlation length
110: was directly estimated in experiments \cite{ouyang91}, and was
111: also extracted from correlation functions \cite{cross95} or from
112: the width of so-called structure factors \cite{sf, note1}. Hu
113: \textit{et al.} \cite{hu95} proposed to use local wave numbers as
114: an order parameter to characterize experimental RB patterns.
115: Gunaratne \textit{et al.} \cite{gunaratne} suggested a disorder
116: function to describe disordered patterns close to locally striped
117: structures \cite{note2}. Egolf \textit{et al.} \cite{egolf98}
118: presented a simple algorithm for real-time quantitative analysis
119: of local structure by computing the local wave number in
120: disordered striped patterns. Newell and co-workers \cite{newell98}
121: used a wave-vector field determined by the wavelet transform to
122: study the behavior of phase-diffusion equations in the presence of
123: defects. All the methods mentioned above are effective to describe
124: irregular patterns with ``visually'' distinct local structures
125: such as local striped structure. When spatiotemporal chaos are
126: overwhelmingly developed, however, local striped structures can
127: not be defined and these measures can not be applied. This is the
128: case for spiral turbulence in highly turbulent regimes of BZ
129: experiments where either exact defect structures or characteristic
130: wave numbers are difficult to obtain. The present study gives a
131: different approach which may be applicable to much more disordered
132: patterns.
133:
134: The BZ reaction system is a particularly interesting system for
135: the study of the spiral dynamics, where spirals can be supported
136: in excitable or oscillatory media \cite{rev93, ouyang01}. A recent
137: experimental study \cite{liu03a} has systematically identified
138: several regimes of spatial patterns resulted from two
139: instabilities, a Doppler instability and a long wavelength
140: instability. At various values of the chemical concentration in
141: reservoir A of malonic acid $[MA]^A$ and in reservoir B of
142: sulfuric acid $[H_2SO_4]^B$, the BZ reaction system displays a
143: variety of patterns including (see Fig.~1 which is reproduced from
144: Fig.~1 of Ref.~\cite{liu03a})
145: \begin{itemize}
146: \item{simply rotating spiral ($S$);}
147: \item{meandering spiral ($M$);}
148: \item{chemical turbulence due to the Doppler instability ($D$);}
149: \item{abnormal chemical turbulence ($A$);}
150: \item{renascent stable spiral ($R$);}
151: \item{convectively unstable spiral ($C_1$ and $C_2$);}
152: \item{chemical turbulence due to the long wavelength instability ($T$)}
153: \end{itemize}
154: This rich set of spiral patterns is an ideal system for testing
155: theoretical methods for characterizing spatiotemporal chaos.
156:
157: \begin{figure}[tbp]
158: \includegraphics[width=8cm]{fig1.eps}
159: \caption{The phase diagram of the BZ experiment reproduced from
160: \cite{liu03a}. The concentrations $[MA]^A$ and $[H_2SO_4]^B$ are
161: the control parameters. The solid lines indicate the onsets of
162: different instabilities. The dashed lines are the extrapolation of
163: the solid lines. The heavy line with arrows at $[MA]^A=0.4M$
164: indicate the case studied here.}
165: \label{Fig:phase-diagram}
166: \end{figure}
167:
168:
169: We have proposed a hierarchical structure (HS) description of
170: spatiotemporal chaos \cite{liu03a, liu03b} based on an earlier HS
171: model of She and Leveque for hydrodynamic turbulence \cite{she94,
172: she98}. The starting point of the analysis is to construct a
173: multi-scale probability density function (PDF) description for the
174: fluctuation field of our interest, and then to develop a
175: phenomenological model for the description of PDFs over the entire
176: range of scales. This set of PDFs carries the probability
177: information for fluctuations at different scales (large versus
178: small) and at different intensities (tail events versus core
179: events). The scale enters as a parameter when the fluctuation
180: field is filtered for defining random variables, and the intensity
181: is parameterized by the order of the moments calculated with the
182: PDFs. Therefore, the dependence of the moments on the scale $\ell$
183: and on the order $p$ constitute a set of quantitative measures
184: about the properties of the field.
185:
186: The HS model gives a special compact description of this set of
187: multi-scale and multi-intensity properties. The basic idea in the
188: HS model is that many statistically stationary multi-scale
189: fluctuation fields generated from nonlinear interactions between
190: many degrees of freedom have a self-organized property expressible
191: in terms of multifractal scaling whose multifractal dimension
192: function can be understood in terms of a similarity relation
193: between structures of increasing intensities of successive
194: moment-order $p$ \cite{she94}. This new similarity relation was
195: proposed for turbulence as a generalization of the Kolmogorov's
196: complete-scale-similarity \cite{she98}. Later, it was discovered
197: that the HS similarity is satisfied in a variety of nonlinear
198: systems and other complex systems, such as the RB convection
199: \cite{ching}, the Couette-Taylor flow \cite{she01}, flows in
200: rapidly rotating annulus \cite{she03c}, the climate variations
201: \cite{she02_climate}, the variation of nucleotide density along
202: DNA sequences \cite{she01_DNA}, the diffusion-limited aggregates
203: \cite{dla}, a fluctuating luminosity field of natural images
204: \cite{image} and several others \cite{passive}. More
205: interestingly, the previous studies show that the derived HS
206: parameters are closely related to global property of the turbulent
207: system. For example, in Couette-Taylor flow, one HS parameter
208: gives a signature of the breaking down of the Taylor vortex
209: \cite{she01}. It is thus hoped that the HS analysis would shed
210: light on the properties of a fluctuation field displaying
211: spatiotemporal chaos.
212:
213: In \cite{liu03a,liu03b}, the HS analysis is briefly applied to
214: describe the spiral turbulence in the BZ experiment and in the
215: numerical solutions of the two-dimensional CGL equation. It was
216: shown that the HS similarity is accurately verified for various
217: experimental and numerical two-dimensional fluctuation fields
218: which include both the ordered spirals and spatiotemporal chaos.
219: In the study of the BZ experiment, it is discovered that the
220: measured HS parameter $\beta$ shows a transition when the system
221: undergoes a transition from one kind of spiral turbulence state to
222: another (e.g. from the state $D$ to the state $T$ in
223: Fig.~\ref{Fig:phase-diagram}). Those preliminary findings have
224: inspired the present detailed study of the nature of $\beta$ while
225: a more complex sequence of transitions are analyzed and more
226: numerical models are considered.
227:
228: Specifically, we will examine more experimental data with
229: transitions between complex states. In
230: Fig.~\ref{Fig:phase-diagram}, at $[MA]^A=0.4$, the patterns vary
231: through $M$, $S$, $C_1$, $A$, $C_2$, $R$, and $T$. During the
232: transition, the amount of order (e.g. spiral structures), the
233: degree of heterogeneity (with the mixing of ordered and disordered
234: structures) and intermittency vary with parameters. We will show
235: that $\beta$ characterizes such transitions near and far from the
236: threshold in the experimental BZ system. We supplement this with a
237: further detailed study of the spiral dynamics in the CGL equation
238: \cite{cgl02} and in a two-component model equation called modified
239: FitzHugh-Nagumo (mFN) model \cite{barkley,bar} which has a number
240: of applications in biological systems. In all cases, we focus on
241: the validity of the HS description and the explanation of the
242: dynamics in terms of the degree of heterogeneity and of
243: intermittency that the value of $\beta$ indicates. We believe that
244: the relationship between the values of $\beta$ and the dynamics of
245: spatiotemporal patterns is valuable.
246:
247: The paper is organized as follows. In Section II, we present
248: briefly the HS model with special emphasis on the method of the HS
249: similarity test ($\beta$-test) and the explanation of the meaning
250: of $\beta$. Section III is devoted to the analysis of a sequence
251: of complex transitions in the experimental BZ system. In Section
252: IV, we carry out a detailed study of the transition of the spiral
253: dynamics in the CGL equation and the mFN equation. Section V
254: offers a summary and some additional discussion.
255:
256: \section{The HS description}
257:
258: The HS model has been originally proposed by She and Leveque to
259: describe inertial-range multi-scale fluctuations in a turbulent
260: fluid. The key concept in the model is a hierarchy of moment
261: ratios. Let $\varepsilon_\ell$ denotes a certain characteristic
262: variable characterizing fluctuations of the physical system at the
263: length scale $\ell$. Define the $p$th-order moments
264: $S_p(\ell)=\langle|\varepsilon_\ell|^{p}\rangle$. The HS model
265: introduce a hierarchy of functions:
266: \begin{equation}
267: F_p(\ell)=\frac{S_{p+1}(\ell)}{S_p(\ell)},
268: \end{equation}
269: each of which has the physical dimension of $\varepsilon_\ell$ and
270: hence describes a certain amplitude of the fluctuations. The
271: function $F_p(\ell)$ depends on the probability density functions
272: (PDFs) of $\varepsilon_\ell$. For a typical set of PDFs of a
273: turbulent field and of a spatiotemporal field as we study in this
274: work, the function $F_p(\ell)$ increases with $p$. In other words,
275: $F_p(\ell)$ at higher p is associated with higher intensity
276: fluctuation structures. We refer $F_p(\ell)$ to as the $p$th-order
277: HS function.
278:
279: An intuitive idea in the HS model is that in a self-organized
280: dynamical steady state, there is a similarity in the dependence of
281: $F_p(\ell)$ on $\ell$ for different $p$'s ($0\le p\le\infty$). In
282: the case of scaling dependence on $\ell$, the simplest similarity
283: is that all $F_p(\ell)$ have the same scaling. This corresponds to
284: the Kolmogorov 1941 (K41) theory of turbulence in which the large
285: and small scale statistics are completely self-similar. For
286: example, when $\varepsilon_\ell$ represents the locally averaged
287: dissipation, then the K41 theory predicts that $F_p(\ell)\sim
288: F_0(\ell)=\langle\varepsilon_\ell\rangle \sim \ell^0$ (the mean
289: dissipation is equal to the average energy flux that is constant
290: independent of the scale). This complete self-similarity law has
291: been demonstrated invalid in experiments and numerical simulations
292: during the past decade, and the dissipation fluctuations have
293: anomalous scalings in $\ell$. This is so-called intermittency
294: effect \cite{sreenivasan91}.
295:
296: She and Leveque \cite{she94} postulate that the most intense
297: structure $F_{\infty}(\ell)=\lim_{p\rightarrow \infty}F_{p}(\ell)$
298: plays a special role, and all other fluctuation structures of
299: finite $p$ obey a hierarchical similarity law, namely:
300: \begin{equation}
301: \frac{F_{p+1}(\ell)}{F_{\infty}(\ell)}=A_{p}\left(\frac{F_{p}(\ell)}{F_{\infty}
302: (\ell)}\right)^{\beta},
303: \label{hierarchical similarity}
304: \end{equation}
305: where $0 \le\beta\le 1$, $A_{p}$ are independent of $\ell$. The
306: term $F_{\infty}(\ell)$ can be eliminated by considering the ratio
307: \begin{equation}
308: \frac{F_{p+1}(\ell)}{F_{p}(\ell)}=\frac{A_{p}}{
309: A_{p-1}}\left(\frac{F_{p}(\ell)}{F_{p-1}(\ell)}\right)^\beta.
310: \label{beta-test}
311: \end{equation}
312: Both sides of Eq.~({\ref{beta-test}) can be computed from the
313: empirical PDFs (or histograms) of $\varepsilon_\ell$ calculated
314: from an experimental or numerical fluctuation field. The linearity
315: in the log-log plot of Eq.~(\ref{beta-test}) can be a direct test
316: of the validity of Eq.~(\ref{hierarchical similarity}). This will
317: be refer to as the HS similarity test, or the $\beta$-test
318: \cite{she01,she03a}. Figure~\ref{Fig:expt.beta-test} shows the
319: result of applying the $\beta$-test to the data of experimental BZ
320: reaction (for details, see below). The linear relation is
321: obviously verified, and the slope $\beta$ can be accurately
322: estimated.
323:
324: \begin{figure}[tbp]
325: \includegraphics[width=8cm]{fig2.eps}
326: \caption{The HS similarity test for three patterns from the
327: experimental BZ system: $[H_2SO_4]^B=0.367M$, $1.0M$, and $1.6M$.
328: The other control parameter is fixed at $[MA]^A=0.4M$. A straight
329: line indicates the validity of the HS similarity. The slope
330: $\beta$ is estimated by a least square fitting. For clarity, the
331: second and third set of data points are displaced vertically up by
332: a suitable amount.}
333: \label{Fig:expt.beta-test}
334: \end{figure}
335:
336:
337: The meaning of the parameter $\beta$ can be readily obtained from
338: the definition Eq.~(\ref{hierarchical similarity})
339: \cite{she98,she01}. It describes a kind of discrepancy between
340: $F_{p+1}$ and $F_p$. When $\beta$ approaches 1, high $p$ (intense)
341: structures and low $p$ (weak) structures are alike. This may be
342: realized at either extremely ordered states or completely
343: homogeneous disordered states. In the former, the structures of
344: various intensities are strongly correlated. Below, we will report
345: observations of this $\beta$ for ordered spiral states. The latter
346: situation corresponds to the hypothetical field postulated by the
347: K41 theory with complete self-similarity; unfortunately, it has
348: not yet been observed in real physical system. Another extreme
349: case is the limit $\beta \to 0$. This is the case in which
350: $F_\infty(\ell)$ stands out as the only kind of singular structure
351: which is responsible for the physical process. A mathematical
352: model displaying the dynamics of one-dimensional shocks, the
353: so-called Burgers equation, exhibits behavior close to this
354: description, where only shocks at isolated points are responsible
355: for the energy dissipation and all statistical moments are
356: determined by the discontinuities across the shocks.
357:
358: The study of hydrodynamic turbulence reveals that in many
359: realistic turbulent systems, $0<\beta<1$, namely the most intense
360: fluctuations do not completely dominate and the fluctuation
361: structures of various intensities are not completely alike. The
362: fluctuation structures of various intensities are all responsible
363: for the physical process, but they are related by the hierarchical
364: similarity law Eq.~(\ref{hierarchical similarity}) which we
365: believe is a form of the self-organization of the system.
366:
367: The results reported below on the experimental BZ system and
368: numerical spiral dynamics demonstrate that when the hierarchical
369: similarity holds, the parameter $\beta$ gives a quantitative
370: description of the degree of order/disorder and of
371: homogeneity/heterogeneity. For $\beta$ close to one, the system
372: appears to be orderly homogeneous and orderly self-organized; for
373: moderately smaller $\beta$ the system contains a mixture of
374: order/disorder and appears to be heterogeneous. The smallest
375: $\beta$ indicates the appearance of overwhelmingly disordered and
376: intermittent state with fully developed spatiotemporal chaos. The
377: details of these findings are reported below.
378:
379: \section{Analysis of experimental BZ system}
380:
381: The experimental BZ system analyzed here has been described in
382: detail in Ref.~\cite{liu03a}. An extensive phase diagram (see
383: Fig.~\ref{Fig:phase-diagram}) is established in the two
384: dimensional plane formed by the concentration $[H_2SO_4]^B$ versus
385: $[MA]^A$, in which various pattern dynamical states are identified
386: based (essentially) on visual inspection of experimental images or
387: luminosity fields. The HS analysis is performed on the total
388: variation field, $G(x,y)$, rather than directly on the luminosity
389: (image) field $f(x,y)$: which is proportional to the chemical
390: concentration:
391: \begin{equation}
392: G(x,y)=\left[(\frac{\partial f}{\partial x})^2+(\frac{\partial f}{\partial y}%
393: )^2\right]^{1/2}.
394: \label{Eq:g-define}
395: \end{equation}
396: Then, we construct a set of multi-scale variables to be the
397: coarse-grained local variation field, $\varepsilon_\ell$, similar
398: to the locally averaged energy dissipation in turbulence, which is
399: defined as below:
400: \begin{equation}
401: \varepsilon_\ell={\frac 1{{\ell ^2}}}\int\limits_x^{x+\ell
402: }\int\limits_y^{y+\ell }G(f(x',y'))dx'dy'.
403: \label{epsilon-eq}
404: \end{equation}
405: We examine the statistical properties by calculating the moments
406: of $\varepsilon_\ell$ as $\ell$ varying for increasing control
407: parameter $[H_2SO_4]^B$. The moments
408: $S_p(\ell)=\langle|\varepsilon_\ell|^{p}\rangle$ are calculated by
409: a space-average of $256\times 256$ pixels and by a time-average
410: over a number of images collected at different times. Furthermore,
411: the HS functions $F_p(\ell)$ are calculated and the ratio across
412: successive orders are evaluated and plotted, as in
413: Fig.~\ref{Fig:expt.beta-test}. It is readily seen that the
414: hierarchical similarity is satisfied and the HS parameter $\beta$
415: can be obtained (by a least square fitting) with a good accuracy.
416: Technically speaking, this completes the HS analysis of a
417: fluctuation field at a given control parameter.
418:
419: \begin{figure}[tbp]
420: \includegraphics[width=8cm]{fig3.eps}
421: \caption{The HS similarity parameter $\beta$ as a function of the
422: control parameter $[H_2SO_4]^B$ with fixed $[MA]^A=0.4M$. Symbols
423: indicate the pattern regimes as found experimentally (see the text
424: for detailed explanation).}
425: \label{Fig:expt.all-beta}
426: \end{figure}
427:
428: \begin{figure}[tbp]
429: \includegraphics[width=8cm]{fig4.eps}
430: \caption{Optical images of typical experimental states obtained
431: with different $[H_2SO_4]^B$: (a) an ordered state with $0.367M$, $\beta=1.0$;
432: (b) a heterogeneous disordered state (coexistence of spiral waves
433: and spatiotemporal chaos) with $0.833M$, $\beta=0.7$;
434: (c) an abnormal turbulent state with $0.867M$, $\beta=0.94$;
435: (d) a homogeneous intermittent disordered state with $1.4M$,
436: $\beta=0.62$. The parameter $[MA]^A$ are fixed at $0.4M$.
437: The region shown is $8.5 \times 8.5 mm^{2}$ out of a disk with $20 mm^{2}$ in diameter.}
438: \label{Fig:expt.pattern}
439: \end{figure}
440:
441: Detailed examinations of the values of $\beta$ at various control
442: parameters constitute a further step in the analysis. The
443: variation of $\beta$ for all patterns observed in the experiment
444: at each parameter value of $[H_2SO_4]^B$ is recorded in
445: Fig.~\ref{Fig:expt.all-beta}. The values of $\beta$ seem to
446: exhibit strong fluctuations, and it seems that there is no simple
447: correspondence between the value of $\beta$ and the regime. More
448: careful inspection reveals, however, that $\beta$ describes the
449: property of the pattern as expected from the theoretical
450: consideration above. We attempt to itemize our observations as
451: follows:
452:
453: (a) For the ordered patterns, the values of $\beta$ measured are
454: close to one. In Fig.~\ref{Fig:expt.pattern}(a), we plot a typical
455: experimental image of the ordered spiral state ($M$) with
456: $\beta=1$. Theoretically speaking, $\beta=1$ implies that low and
457: high intensity structures are alike, which happens with an ordered
458: pattern. Note that some values of measured $\beta$ are not close
459: to one for ordered spiral states (e.g. the two $S$ states for
460: which $\beta\approx 0.75$ in Fig.~\ref{Fig:expt.all-beta}). One
461: possible explanation for the exception of the two $S$ states with
462: low $\beta$ is the existence of experimental noise (e.g., dark
463: spots in the experimental images) that introduces artificially
464: {\it sharp} structures in the field. Our studies show that such
465: ``intrusion" of sharp objects lead to a decrease of $\beta$.
466:
467: (b) Near the transition between one regime and another, we observe
468: a depletion of the value of $\beta$ compared to that measured for
469: the states far from the transition (see
470: Fig.~\ref{Fig:expt.all-beta}). We plot a typical experimental
471: image near the transition between $C_1$ and $A$ in
472: Fig.~\ref{Fig:expt.pattern}(b) with $\beta=0.7$. Theoretically
473: speaking, a lower value of $\beta$ indicates a higher
474: intermittency of the spatial patterns.
475: Fig.~\ref{Fig:expt.pattern}(b) indeed shows a mixture of ordered
476: spirals and disordered irregular spiral elements, which displays
477: clearly a strong heterogeneity. We conclude that one kind of
478: intermittency in spatially extended systems is related to
479: heterogeneous composition of patterns of different origins. In
480: addition, we conclude that a sharp variation of $\beta$ usually
481: indicates a transition of regimes. This finding is interesting for
482: defining statistical regimes. We believe that this is the most
483: interesting finding of this work.
484:
485: (c) In fully developed spatiotemporally chaotic states, the
486: spatial patterns develop small-scale intermittent structures and
487: yield even smaller values of $\beta$ ($<0.65$). In
488: Fig.~\ref{Fig:expt.pattern}(d), we show a experimental image for
489: the chemical turbulent state $T$ with $\beta=0.62$. It is clear
490: that very small characteristic scales are developed in the
491: patterns which signify large fluctuations in the total variation
492: field (because of the spatial derivatives). This enhanced
493: intermittency appears to be stronger than noted in item (b).
494:
495: (d) There are also other disordered states (e.g. $A$ and $C_2$)
496: for which $\beta$ is close to one. Close inspection shows that
497: these disordered states all encompass large-scale irregular spiral
498: elements, as shown in Fig~\ref{Fig:expt.pattern}(c) where
499: $\beta=0.94$. These large-scale patterns are quite homogeneous and
500: spatially ordered, which may yield a large value of $\beta$. Note
501: that the HS analysis conducted in this paper is a spatially
502: multiscaling analysis which describes only the property of
503: instantaneous spatial patterns. Our conclusion is that the states
504: in the regimes $A$ and $C_2$ involve only spatially ordered
505: patterns with actively temporal dynamics. They are distinguished
506: from the other chaotic states like $T$ and $D$.
507:
508: The above analysis helps to understand the nature of the parameter
509: $\beta$ in the HS analysis of spatiotemporal chaos, and confirms
510: our previous finding that $\beta$'s variation reveals a transition
511: between two different spatiotemporally chaotic states. The spiral
512: turbulence state due to the Doppler instability has a
513: characteristic scale larger than that in the spiral turbulence
514: state due to the long wavelength instability. Both states are
515: homogeneously chaotic, but the later contains more intermittent
516: small-scale structures. The HS analysis is suitable to capture
517: such difference.
518:
519: \section{Analysis of numerical spiral turbulence}
520:
521: \begin{figure}[tbp]
522: \includegraphics[width=8cm]{fig5.eps}
523: \caption{Snapshots of the field $f(x,y)$ from the CGL equation at
524: four different parameter values of $c_3$. (a) stable spiral,
525: $c_3=0.6$; (b) coexistence of spiral and spatiotemporal chaos,
526: $c_3=0.74$; (c) coexistence of spiral and spatiotemporal chaos,
527: $c_3=0.75$; (d) fully spatiotemporal chaos, $c_3=0.8$.}
528: \label{Fig:CGL-c3-images}
529: \end{figure}
530:
531: \begin{figure}[tbp]
532: \includegraphics[width=8cm]{fig6.eps}
533: \caption{The HS similarity test for the four patterns in
534: Fig.~\ref{Fig:CGL-c3-images} of the CGL equation. A straight line
535: indicates the validity of the HS similarity. The slope $\beta$ is
536: estimated by a least square fitting. For clarity, the second, the
537: third and the fourth set of data points are displaced vertically
538: up by a suitable amount.}
539: \label{Fig:CGL-c3-beta-test}
540: \end{figure}
541:
542: \begin{figure}[tbp]
543: \includegraphics[width=8cm]{fig7.eps}
544: \caption{The HS similarity parameter $\beta$ as a function of the
545: control parameter $c_3$ in the CGL equation. Note a clear
546: transition from an ordered state to a spatiotemporally chaotic state.}
547: \label{Fig:CGL-c3-all-beta}
548: \end{figure}
549:
550: \begin{figure}[tbp]
551: \includegraphics[width=8cm]{fig8.eps}
552: \caption{A sequence of evolving fields $f(x,y)$ from the CGL
553: equation at different times. (a) $t=390$; (b) $t=500$; (c)
554: $t=550$; (d) $t=770$.}
555: \label{Fig:CGL-ev-images}
556: \end{figure}
557:
558: \begin{figure}[tbp]
559: \includegraphics[width=8cm]{fig9.eps}
560: \caption{The HS similarity test for the four patterns in
561: Fig.~\ref{Fig:CGL-ev-images} of the CGL equation. A straight line
562: indicates the validity of the HS similarity. The slope $\beta$ is
563: estimated by a least square fitting. For clarity, the second, the
564: third and the fourth set of data points are displaced vertically
565: up by a suitable amount.}
566: \label{Fig:CGL-ev-beta-test}
567: \end{figure}
568:
569: \begin{figure}[tbp]
570: \includegraphics[width=8cm]{fig10.eps}
571: \caption{The HS similarity parameter $\beta$ as a function of time
572: for a sequence of evolving patterns in the CGL equation. Note that
573: $\beta$ acts as an order parameter revealing the transition from
574: an ordered state to a spatiotemporally chaotic state.}
575: \label{Fig:CGL-ev-all-beta}
576: \end{figure}
577:
578:
579: We now turn to the analysis of numerically simulated
580: two-dimensional spatiotemporal dynamics. First, in numerical
581: calculations, the interaction mechanism responsible for the
582: generation of spatial patterns is under good control, so that
583: spatial patterns are free of noises and perturbations that are
584: inevitable in experiments. For example, optical artifacts in the
585: experimental images pose great challenge to the HS analysis (in
586: the calculation of the total variation field) and thus compromise
587: the clarity of the finding. Secondly, the experimental states
588: analyzed above are all asymptotical states. We are also interested
589: in the characterization of transient states. In other words, we
590: are interested in testing whether the HS description can
591: characterize the property of a spatial pattern which undergoes
592: actively dynamical evolution. Numerical simulations provide a
593: detailed sequence of patterns under dynamic evolution. Finally,
594: the present analysis of numerically simulated field may inspire
595: further studies of a variety of two-dimensional partial
596: differential equations of pattern dynamics, e.g. the Navier-Stokes
597: equation for the vortex dynamics. Such empirical studies are
598: important to the finding of right order parameters of chaotic
599: pattern dynamics.
600:
601: We will study two model equations for the pattern dynamics leading
602: to spatiotemporal chaos. The first equation is the two-dimensional
603: CGL equation:
604: \begin{equation}
605: \frac{\partial A}{\partial t}=A+(1+ic_1)\triangle A
606: -(1+ic_3)\left| A\right| ^2A,
607: \label{CGL-eq}
608: \end{equation}
609: where $A(\mathbf{r},t)$ is a complex function of time ($t$) and
610: space ($\mathbf{r}$), the real parameters $c_1$ and $c_3$ are
611: coefficients characterizing linear and nonlinear dispersion
612: respectively. Although the CGL equation is a normal form relevant
613: only near the threshold of a supercritical Hopf bifurcation, it
614: has become a popular model to study spatiotemporal chaos where
615: oscillations and waves are present \cite{rev93,cgl02}. Indeed, our
616: simulation gives rise to both simple spiral waves and chaotic
617: states which encompass many disordered spatiotemporal
618: oscillations.
619:
620: The numerical study of Eq.~(\ref{CGL-eq}) is performed using an
621: Euler algorithm with no-flux boundary condition on a $256 \times
622: 256$ square lattice and a time step of $\triangle t = 0.03$. The
623: initial conditions are chosen to be a simple spiral wave which is
624: the stable solution when $c_1=-1.40$ and $c_3=0.50$. We then keep
625: $c_1$ fixed and increase $c_3$ to explore possibly alternative
626: patterns. We choose to analyze the fluctuation properties of the
627: real part $f(x,y)$ of the complex field $A(x,y)= f(x,y)+ i
628: g(x,y)$. The details of the computation and the transition leading
629: to spatiotemporal chaos can be found in Ref.~\cite{liu03b}, where
630: we have established the fact that the $\beta$-test is satisfied in
631: both ordered spiral states and spatiotemporal chaos states. In the
632: present work, we report the result of a systematic HS analysis of
633: a sequence of patterns with varying parameters.
634:
635: Four snapshots of the field $f(x,y)$ are shown in
636: Fig.~\ref{Fig:CGL-c3-images}, which contribute to a so-called
637: far-field breaking-up process. As $c_3$ increases, the convective
638: instability occurs and the spiral wave breaks up far from the tip.
639: In a range of intermediate values of $c_3$, the spiral wave
640: coexists with spatiotemporal chaos surrounding it. Finally, at
641: high values of $C_3$, the spiral wave totally breaks up and
642: spatiotemporal chaos dominate the whole box ($c_3>0.76$).
643:
644: The HS analysis of the solutions of the CGL equation begins with
645: the construction of the total variation field. Let $G(x,y)$ be
646: defined as in Eq.~(\ref{Eq:g-define}) with the field $f(x,y)$. The
647: coarse-grained local variation field, $\varepsilon_\ell$, is
648: defined as in Eq.~(\ref{epsilon-eq}). The same procedure of the
649: $\beta$-test as explained in the previous section is now applied
650: to the data $\varepsilon_\ell$. As shown in
651: Fig.~\ref{Fig:CGL-c3-beta-test}, good linearity and hence the HS
652: similarity are obtained for all the states of both ordered spirals
653: and irregular spiral turbulence, consistent with the result
654: reported previously \cite{liu03b} and similar to that in
655: Fig.~\ref{Fig:expt.beta-test}. Note again that the values of
656: $\beta$ are close to one for ordered spiral states, but
657: significantly below one ($< 0.9$) when spatiotemporal chaos are
658: developed.
659:
660: The variation of $\beta$ as a function of $c_3$ is shown in
661: Fig.~\ref{Fig:CGL-c3-all-beta}. All ordered spiral states have
662: $\beta$ very close to one, and the appearance of the irregular
663: patterns surrounding the central spiral leads to a smaller
664: $\beta$. It is interesting to note that the final chaotic states
665: at high $c_3$ have a $\beta\approx 0.9$ which is bigger than that
666: of the transitional states with a mixture of ordered spirals and
667: chaos. This fact is consistent with the previous finding in the
668: analysis of the experimental BZ system that the mixed states near
669: the transition are more heterogeneous and tend to have smaller
670: $\beta$.
671:
672: Another interesting aspect of the numerical CGL dynamics is the
673: possibility to follow closely a spontaneous generation of spatial
674: disorder from an ordered state. A sequence of dynamically evolving
675: patterns are shown in Fig~\ref{Fig:CGL-ev-images}, where the
676: spiral core breaks up without meandering and then subsequently
677: develops to a complete spatiotemporal chaotic pattern. This fast
678: breaking-up process is phenomenologically similar to that of the
679: Doppler instability \cite{ouyang01,ouyang00}, but with the spiral
680: center fixed. The details of the explanation for this process can
681: be found in Ref.~\cite{liu03b}. Here, we show four typical
682: snapshots of the spatial pattern at successive times with fixed
683: control parameter $c_1=-1.40$ and $c_3=0.8$.
684:
685: \begin{figure}[tbp]
686: \includegraphics[width=8cm]{fig11.eps}
687: \caption{Snapshots of the field of the fast activator $u$ in the
688: mFN equation. (a): simple spiral, $\sigma=0.0525$; (b):
689: meandering spiral, $\sigma=0.07$; (c): outbreak of
690: breaking-up spiral, $\sigma=0.0725$; (d): spatiotemporal
691: chaos, $\sigma=0.075$. Other parameters are $a=0.84,
692: b=0.07$.}
693: \label{Fig:FN-images}
694: \end{figure}
695:
696: Figure~\ref{Fig:CGL-ev-beta-test} reports the result of the
697: $\beta$-test for the corresponding four pattern states in
698: Fig.~\ref{Fig:CGL-ev-images}. Good HS similarity property is found
699: for all pattern states, including the single spiral
700: ($\beta\approx1$), the intermediate state with generated defects
701: ($\beta<0.9$) and the full developed turbulent state
702: ($\beta\approx0.9$). Figure~\ref{Fig:CGL-ev-all-beta} shows the
703: variation of $\beta$ as a function of time during the core
704: breaking-up. The parameter $\beta$ begins with a value close to
705: one and undergoes a transition when disorder is developed. At
706: $t\approx 550$, $\beta$ reaches the minimum value when disorder is
707: mixed with order. Later, $\beta$ increases again because of the
708: homogeneity of the disorder. These facts are consistent with the
709: observations reported earlier. Once more, the HS parameter $\beta$
710: acts as an order parameter to the description of the evolution of
711: the CGL system from the ordered spiral to the disordered
712: spatiotemporal chaos.
713:
714: The second model equation we have studied is a simple
715: activator-inhibitor model: the two-dimensional mFN model
716: \cite{barkley,bar}. While the CGL equation describes an
717: oscillatory dynamics (with two components representing the real
718: and imaginary parts of a complex variable field), the mFN model is
719: a two-component model of an excitable system in which a fast
720: activator ($u$) and a slow inhibitor ($v$) are in interaction:
721: \begin{subequations}
722: \begin{equation}
723: u_{t} = -\frac{1}{\sigma}u(u-1)\left[u-\frac{v+b}{a}
724: \right]+\triangle u,
725: \end{equation}
726: \begin{equation}
727: v_{t} = f(u)-v,
728: \end{equation}
729: \begin{eqnarray}
730: f(u) &=& \left\{
731: \begin{array}{l@{\quad \quad}r}
732: 0 , & 0 \leq u<\frac{1}{3} \\
733: 1-6.75u(u-1)^{2}, & \frac{1}{3}\leq u \leq 1 \\
734: 1 , & u>1
735: \end{array} \right.
736: \end{eqnarray}
737: \label{Eq:FN}
738: \end{subequations}
739: where, the form of $f(u)$ describes an inhibitor production only
740: above a threshold value of $u$, and the value of $\sigma$ related
741: to the time scales of both activator ($u$) and inhibitor ($v$)
742: measures the degree of excitability. A nice feature of the mFN
743: model is that the degree of the excitability of the medium can be
744: continuously modified by varying the parameters. It is known
745: \cite{bar} that fast activator dynamics ($0 < \sigma \ll 1$) and
746: suitable choice for other parameters ($a<1$, $b>0$) lead to
747: excitability. When one fixes the parameters $a$ and $b$ and
748: gradually increases $\sigma$, the excited spiral waves will loss
749: its stability, begin to meander, then break up at the tip, and
750: spread chaotic spiral elements outward. Finally, the system
751: develops a state of spatiotemporal chaos. This scenario is
752: illustrated in Fig.~\ref{Fig:FN-images}, which contains a sequence
753: of images of gray-scale picture of the activator $u$ obtained at
754: various times of a dynamic evolution.
755:
756: \begin{figure}[tbp]
757: \includegraphics[width=8cm]{fig12.eps}
758: \caption{The HS similarity test for the four pattern states in
759: Fig.~\ref{Fig:FN-images} of the mFN equation. A straight line
760: indicates the validity of the HS similarity. The slope $\beta$ is
761: estimated by a least square fitting. For clarity, the second, the
762: third and the fourth set of data points are displaced vertically
763: up by a suitable amount.}
764: \label{Fig:FN-beta-test}
765: \end{figure}
766:
767: \begin{figure}[tbp]
768: \includegraphics[width=8cm]{fig13.eps}
769: \caption{The HS similarity parameter $\beta$ as a function of the
770: control parameter $\sigma$ in the mFN equation. Note a clear
771: transition from an ordered state to a spatiotemporal chaotic state.}
772: \label{Fig:FN-all-beta}
773: \end{figure}
774:
775: The numerical calculation of Eq.~(\ref{Eq:FN}) reported above uses
776: an Euler algorithm with no-flux boundary condition on a $256
777: \times 256$ square lattice. The data presented in
778: Fig.~\ref{Fig:FN-images} are obtained with fixed parameters
779: $a=0.84$ and $b=0.07$, but with several different values of
780: $\sigma$. The simulation starts with a simple spiral initial
781: condition that is the stable solution of the system at smaller
782: values of $\sigma$ (e.g. $\sigma=0.50$). At $\sigma=0.725$, a
783: transition from the simple spiral to spatiotemporal chaos as
784: described earlier is fully realized.
785:
786: To carry out the HS analysis, the total variation field $G(x,y)$
787: for the mFN model is calculated from the fast activator field
788: $u(x,y)$ as in Eq.~(\ref{Eq:g-define}). The coarse-grained local
789: variation field, $\varepsilon_\ell$, is defined as in
790: Eq.~(\ref{epsilon-eq}). We have applied the $\beta$-test for the
791: moments of $\varepsilon_\ell$, and the result is shown in
792: Fig.~\ref{Fig:FN-beta-test}. Clearly, the hierarchical similarity
793: is verified for all values of $\sigma$ and the values of $\beta$
794: are derived accurately in all cases. The variation of $\beta$ as a
795: function of $\sigma$ is presented in Fig.~\ref{Fig:FN-all-beta}.
796: The correspondence between the degrees of orders/disorder and the
797: values of $\beta$ is quite clear, as shown in the study of the CGL
798: equation. As the parameter $\sigma$ increases, the pattern
799: gradually loses the correlation and $\beta$ decreases. Detailed
800: examination shows that the tip of the spiral wave begins to
801: meander at $\sigma=0.625$, and completely breaks up at
802: $\sigma=0.725$. The break-up of the tip of the spiral marks the
803: transition from $\beta\approx 1$ to $\beta\approx 0.8$. The later
804: state corresponds to a fully developed spatiotemporal chaos.
805:
806: The above numerical studies reveal that the CGL and mFN model
807: equations have less complicated patterns than those in the
808: experimental BZ system. This significantly reduces the complexity
809: of the matter, and renders the HS description of the dynamics easy
810: to interpret. It is worth noting that the numerical data are free
811: from complicated disturbance so that the values of $\beta$ show a
812: consistent trend of variation with less fluctuations, compared to
813: the results of Fig.~\ref{Fig:expt.all-beta}. We therefore find the
814: convincing evidence that the HS parameter $\beta$ is like an order
815: parameter that describes the transition from an ordered spiral
816: state to a homogeneous disordered spiral turbulence.
817:
818: \section{Summary and discussion}
819:
820: One of the most interesting problems in the study of the pattern
821: dynamics is to find a quantitative measure that describes the
822: global property of the pattern in both ordered states and
823: turbulent states. When spatiotemporal chaos prevail and the
824: effective number of degrees of freedom is large, many familiar
825: quantities in the study of dynamical systems and chaos become hard
826: to evaluate. On the other hands, traditional statistical measures
827: such as correlation length also become ineffective when the
828: pattern presents intermittent features. A new sound phenomenology
829: is needed to capture subtle correlations across multiple scales
830: and multiple fluctuation intensities.
831:
832: We present here a sound phenomenology, the HS model, that gives a
833: concise description of a spatiotemporal field by its multifractal
834: scaling properties in terms of the HS parameter $\beta$. In this
835: work, we present detailed evidence, based on analyzing
836: experimental BZ system and numerical CGL and mFN model, that the
837: HS similarity holds for all the fields (ordered or disordered) and
838: the derived parameter $\beta$ describes the strength of the
839: correlation across scales and across intensities. For $\beta$
840: close to one, the pattern is strongly correlated or ordered; a
841: smaller $\beta$ implies a higher degree of heterogeneity and a
842: mixture of order and disorder; the smallest $\beta$ is related to
843: the intermittency associated with fully developed spatiotemporal
844: chaos. In particular, our analysis has identified two different
845: disordered states in the BZ experiment: the heterogeneous
846: disordered state with a mixture of spirals and irregular spiral
847: elements and the disordered and intermittent state of fully
848: developed spatiotemporal chaos. In those cases, $\beta$ provides a
849: useful characterization of the patterns in transition.
850:
851: Based on the above results, we speculate that the HS similarity is
852: an emergent property of self-organized complex nonlinear
853: multi-scale systems in general. This property, like the maximum
854: entropy property of the thermal equilibrium state, may be
855: difficult to be derived directly from the first principle (such as
856: the Navier-Stokes equation for fluids or the Newtonian mechanical
857: law for gases). It is thus important to conduct more empirical
858: studies for proving/disproving the above conjecture, or for
859: assessing its limit of validity.
860:
861: For the quantitative analysis of spatiotemporal chaos, we believe
862: that it is important to go beyond the simple statistics for
863: defects, wave numbers, or other local pattern properties, since
864: often they are hard to be determined when spatial patterns become
865: overwhelmingly irregular. The HS analysis is a global method that
866: takes into account the properties of the whole field, but
867: emphasizes at the same time subtle correlations and similarities
868: among structures at multiple scales and multiple intensities.
869: While the previous work \cite{liu03a, liu03b} establishes the
870: validity of the HS model for describing both experimental and
871: numerical spiral patterns, the present work gives detailed
872: evidence that the HS order parameter ($\beta$) characterizes the
873: degree of heterogeneity and intermittency of spatial patterns and
874: is particularly suitable for describing the evolution of patterns
875: from ordered states to fully developed spatiotemporally chaotic
876: states. This is demonstrated in the BZ, CGL and mFN systems. Our
877: approach may be easily extended to describe other complex
878: pattern-formation systems with different unit ``cells" than
879: spirals, such as strips, hexagons, squares, triangles, etc., and
880: this is one of the advantages of the HS analysis.
881:
882: To make a concise description for all these statistical
883: properties, a phenomenology is required. The HS analysis is one of
884: the plausible phenomenological proposals that rest on empirical
885: observations ($\beta$-test). The fact that the HS similarity is
886: found to be valid for a wide class of nonlinear processes
887: indicates the possible universal mechanism/principle behind these
888: processes. It is not yet clear how this universality should be
889: expressed (e.g. in terms of a variational principle like the
890: maximum entropy and minimum entropy production), but we believe
891: that it deserves special attention in future studies.
892:
893:
894: \acknowledgments
895:
896: We thank both anonymous referees for their helpful comments. We
897: have benefited from the research environment at the LTCS of Peking
898: University. The computation work is helped by the Turbulence
899: Simulation Center of the LTCS. We acknowledge the support by the
900: National Natural Science Foundation of China, No. 10225210, and by
901: the State Key Project for Basic Research No. 2000077305.
902:
903:
904: \begin{thebibliography}{99}
905:
906: \bibitem{rev93}
907: M. C. Cross and P. C. Hohenberg, Rev. Mod. Phys. {\bf 65}, 851
908: (1993).
909:
910: \bibitem{morris93}
911: S.W. Morris, E. Bodenschatz, D.S. Cannell, and G. Ahlers Phys.
912: Rev. Lett. {\bf 71}, 2026 (1993);
913:
914: \bibitem{assenheimer94}
915: M. Assenheimer and V. Steinberg, Nature {\bf 367}, 345 (1994).
916:
917: \bibitem{ouyang01}
918: L. Q. Zhou and Q. Ouyang, J. Phys. Chem. A {\bf 105}, 112 (2001)
919:
920: \bibitem{cardiac}
921: A. T. Winfree, Science {\bf 266}, 1003 (1994); F. X. Witkowski, L.
922: J. Leon, P. A. Penkoske, W. R. Giles, M. L. Spanol, W. L. Ditto,
923: and A. T. Winfree, Nature {\bf 392}, 78 (1998).
924:
925: \bibitem{coullet89}
926: P. Coullet, L. Gil, and J. Lega, Phys. Rev. Lett. {\bf 62}, 1619
927: (1989).
928:
929: \bibitem{cgl02}
930: I. S. Aranson and L. Kramer, Rev. Mod. Phys. {\bf 74}, 99 (2002).
931:
932: \bibitem{ch-science94}
933: M. C. Cross and P. C. Hohenberg, Science {\bf 263}, 1569 (1994).
934:
935: \bibitem{ouyang91}
936: Q. Ouyang and H. L. Swinney, Chaos {\bf 1}, 411 (1991).
937:
938: \bibitem{cross95}
939: M. C. Cross and D. I. Meiron, Phys. Rev. Lett. {\bf 75}, 2152
940: (1995).
941:
942: \bibitem{sf}
943: K. R. Elder, J. D. Gunton, and N. Goldenfeld, Phys. Rev. E {\bf
944: 56}, 1631 (1997); T. Galla and E. Moro, Phys. Rev. E {\bf 67},
945: 035101 (2003).
946:
947: \bibitem{note1}
948: Structure factors were first used as a measure of the structure of
949: local RB patterns in Ref.~\cite{morris93}. Detail comments on the
950: usage of structure factor can be found in Ref.~\cite{hu95}. This
951: measure has also been used to measure the growth of pattern
952: domains, as in Ref.~\cite{cross95} and L. Purvis and M. Dennin,
953: Phys. Rev. Lett. {\bf 86}, 5898 (2001).
954:
955: \bibitem{hu95}
956: Y. Hu, R. Ecke, and G. Ahlers, Phys. Rev. E {\bf 51}, 3263 (1995);
957: Y. Hu, R. E. Ecke, and G. Ahlers, Phys. Rev. Lett. {\bf 74}, 391
958: (1995); Phys. Rev. E {\bf 55}, 6928 (1997).
959:
960: \bibitem{gunaratne}
961: G. H. Gunaratne, R. E. Jones, Q. Ouyang, and H. L. Swinney, Phys.
962: Rev. Lett. {\bf 75}, 3281 (1995); G. H. Gunaratne, D. K. Hoffman,
963: and D. J. Kouri, Phys. Rev. E {\bf 57}, 5146 (1998); G. H.
964: Gunaratne, A. Ratnaweera, and K. Tennekone, Phys. Rev. E {\bf 59},
965: 5058 (1999);
966:
967: \bibitem{note2}
968: The disorder function calculates the local curvature of contour
969: lines and local characteristic wavenumber for patterns of local
970: stripped structures. For more complicated patterns, the
971: calculation of the disorder function is difficult. Detail comments
972: can be seen in Ref.~\cite{gunaratne}.
973:
974: \bibitem{egolf98}
975: D. A. Egolf, I. V. Melnikov, and E. Bodenschatz, Phys. Rev. Lett.
976: {\bf 80}, 3228 (1998).
977:
978: \bibitem{newell98}
979: C. Bowman and A. C. Newell, Rev. Mod. Phys. {\bf 70}, 289 (1998).
980:
981: \bibitem{liu03a}
982: H. Y. Guo, L. Li, Q. Ouyang, J. Liu, and Z.-S. She, J. Chem. Phys.
983: {\bf 118}, 5038 (2003).
984:
985: \bibitem{liu03b}
986: J. Liu, Z.-S. She, Q. Ouyang, and X. T. He, Inter. J. Mod. Phys. B
987: {\bf 17}, 4139 (2003).
988:
989: \bibitem{she94}
990: Z.-S. She and E. Leveque, Phys. Rev. Lett. {\bf 72}, 336 (1994);
991: Z.-S. She and E. C. Waymire, Phys. Rev. Lett. {\bf 74}, 262
992: (1995).
993:
994: \bibitem {she98}
995: Z.-S. She, Prog. Theor. Phys. Supp. {\bf 130}, 87 (1998).
996:
997: \bibitem {ching}
998: E. S. C. Ching, Acta Mech. Sinica {\bf 19}, 385 (2003); E. S. C.
999: Ching, C. K. Leung, X. L. Qiu, and P. Tong, Phys. Rev. E {\bf 68},
1000: 026307, (2003); E. S. C. Ching and C. Y. Kwok, Phys. Rev. E {\bf
1001: 62}, R7587 (2000).
1002:
1003: \bibitem{she01}
1004: Z.-S. She, K. Ren, G. S. Lewis, and H. L. Swinney, {\it Phys. Rev.
1005: E.} {\bf 64}, 016308 (2001).
1006:
1007: \bibitem{she03c}
1008: C. Baroud, B. Plapp, H. Swinney, and Z.-S. She, Phys. Fluids, {\bf
1009: 15}, 2091 (2003).
1010:
1011: \bibitem{she02_climate}
1012: Z.-S. She, Z. T. Fu, J. Chen, S. Liang, and S. D. Liu, Prog. in
1013: Natural Sci. {\bf 12}(10), 747-752 (2002)
1014:
1015: \bibitem{she01_DNA}
1016: J. Wang, Q. D. Zhang, K. Ren, and Z.-S. She, Chinese Sci. Bull.
1017: {\bf 46}(23), 1988-1992 (2001).
1018:
1019: \bibitem{dla}
1020: D. Queiros-Conde, Phys. Rev. Lett. {\bf 78}, 4426 (1997).
1021:
1022: \bibitem{image}
1023: A. Turiel, G. Mato, N. Parga, and J.-P. Nadal, Phys. Rev. Lett.
1024: {\bf 80}, 1098 (1998).
1025:
1026: \bibitem{passive}
1027: G. Ruiz Chavarria, C. Baudet, and S. Ciliberto, Physica D {\bf
1028: 99}, 369 (1996); E. Leveque, G. Ruiz-Chavarria, C. Baudet, and S.
1029: Ciliberto, Phys. Fluids {\bf 11}, 1869 (1999).
1030:
1031: \bibitem{bar}
1032: M. B\"ar and M. Eiswirth, Phys. Rev. E {\bf 48}, 1635 (1993);
1033:
1034: \bibitem{barkley}
1035: D. Barkley, M. Kness and L. S. Tuckerman, Phys. Rev. A {\bf 42},
1036: 2489 (1990); D. Barkley, Physica D {\bf 49}, 61 (1991); Phys. Rev.
1037: Lett. {\bf 68}, 2090 (1992); Phys. Rev. Lett. {\bf 72}, 164
1038: (1994).
1039:
1040: \bibitem{ouyang00}
1041: Q. Ouyang, H. L. Swinney, and G. Li, Phys. Rev. Lett. {\bf 84},
1042: 1047 (2000).
1043:
1044: \bibitem{sreenivasan91}
1045: K. R. Sreenivasan, Ann. Rev. Fluid Mech. {\bf 23}, 539 (1991).
1046:
1047: \bibitem{she03a}
1048: Z.-S. She and L. Liu, Acta Mech. Sinica, {\bf 19}, 453 (2003); L.
1049: Liu and Z.-S. She, Fluid Dyn. Res., {\bf 33}, 261 (2003).
1050:
1051: \end{thebibliography}
1052: \end{document}
1053: