1:
2: \documentclass[runningheads]{cl2emult}
3:
4: \usepackage{graphicx}
5: \usepackage{subeqnar}
6: \usepackage{cropmark}
7:
8: %\bibliographystyle{plain}
9:
10: \setcounter{page}{235}
11:
12: \begin{document}
13:
14:
15:
16:
17:
18:
19:
20: %\ifthenelse{\equal{\Ox}{Ja}}{
21: %\chapter[Noise-Induced Order in Extended Systems]{Noise-Induced Order in Extended Systems:
22: %A Tutorial}
23: %\author{Jos\'e M. Sancho}
24: %\address{Departament d'Estructura i Constituents de la Mat\`eria,
25: %Univ. de Barcelona,\\ Av. Diagonal 647, E-08028 Barcelona, Spain}
26: %%\author*{and}
27: %\author{Jordi Garc\'{\i}a--Ojalvo}
28: %\address{Departament de F\'{\i}sica i Enginyeria Nuclear,
29: %Univ. Polit\`ecnica de Catalunya,
30: %Colom 11, E-08222 Terrassa, Spain}
31: %}
32: %{
33: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
34: \title*{Noise-Induced Order in Extended Systems:
35: \protect\newline A Tutorial\footnote{From ``Stochastic Processes in Physics,
36: Chemistry, and Biology'', J.~A. Freund and T.~P\"oschel (Eds.), Lecture Notes in Physics
37: vol. 557 (Springer, Berlin, 2000)}}
38:
39: \toctitle{Noise-Induced Order in Extended Systems: A Tutorial}
40:
41: \titlerunning{Noise-Induced Order in Extended Systems: A Tutorial}
42:
43: \author{
44: Jos\'e M. Sancho\inst{1}
45: \and Jordi Garc\'{\i}a--Ojalvo\inst{2}
46: }
47:
48: \authorrunning{J.M. Sancho and J. Garc\'{\i}a-Ojalvo}
49:
50: \institute{
51: Departament d'Estructura i Constituents de la Mat\`eria,
52: Univ. de Barcelona,\\ Av. Diagonal 647, E-08028 Barcelona, Spain
53: \and
54: Departament de F\'{\i}sica i Enginyeria Nuclear,
55: Univ. Polit\`ecnica de Catalunya,\\
56: Colom 11, E-08222 Terrassa, Spain}
57:
58: \maketitle
59: %}
60: \begin{abstract}
61: External fluctuations have a wide variety of constructive effects on the
62: dynamical behavior of spatially extended systems, as described by
63: stochastic partial differential equations. A set of paradigmatic situations
64: exhibiting such effects are briefly reviewed in this paper, in an attempt
65: to provide a concise but thorough introduction to this active field of
66: research, and at the same time an overview of its current status. This work
67: is dedicated to Lutz Schimansky--Geier
68: on the occassion of his 50th anniversary. Through the years, Prof.
69: Schimansky--Geier has made important contributions to the field of
70: spatiotemporal stochastic dynamics, including
71: seminal investigations in the early 1990's on noise effects in front
72: propagation, and studies of noise-induced phase transitions and
73: noise-sustained structures in excitable media, among others.
74: \end{abstract}
75:
76: \section{Introduction}
77:
78: It is well accepted nowadays that noise can have rather surprising and
79: counterintuitive effects. There are many physical situations in which noise
80: exhibits a {\em constructive}, rather than destructive, role in the behavior
81: of a nonlinear system. A relevant example of this fact, still under active
82: research nowadays, is the phenomenon of stochastic resonance, in which the
83: response of a nonlinear system to an external signal under the presence of
84: fluctuations can be enhanced by tuning the noise intensity to an optimal
85: (non-zero) value \cite{SG.Gamma98}. In another direction, studies of
86: stochastic zero-dimensional systems (i.e., systems with only temporal
87: dependence), reviewed in
88: \cite{SG.horsthemke84}, showed that noise is able to induce transitions
89: in such systems. Spatial degrees of freedom provide further interesting
90: scenarios where the non-trivial role of external noise arises, such as phase
91: transitions and critical phenomena \cite{SG.Ma76}, and pattern
92: formation out of equilibrium \cite{SG.cross}.
93:
94: The present review aims to be a brief introduction to the
95: influence of external noise on spatially extended, $d$-dimensional
96: systems (an extensive monograph on the topic can be found in \cite{SG.NISES}).
97: As far as we know, seminal works in this direction were already carried out in
98: the late 1970's by Mikhailov \cite{SG.mikhailov79,SG.mikhailov81}. Owing to
99: the existence of spatial degrees of freedom, the behavior of extended systems
100: is described by phases in a thermodynamic sense. Under these conditions, the
101: presence of certain types of external noise affects, in a non-obvious way,
102: the behavior of the corresponding deterministic (noiseless) system, or even
103: that of the associated equilibrium system (which has internal noise terms
104: obeying a fluctuation-dissipation relation). In particular, we
105: expect not only quantitative changes with respect the deterministic
106: results, but also qualitatively different, even new, features
107: {\em induced} by the presence of external fluctuations. In these situations,
108: na\"{\i}ve predictions based on a deterministic analysis are very far from
109: giving reliable results. Several examples can be enumerated, some of
110: which will be reviewed here:
111: noise-induced spatial patterns
112: %({\bf NISP})
113: \cite{SG.ojalvo93,SG.BK94,SG.par96,SG.zaikin},
114: noise-induced ordering phase transitions
115: %({\bf NIOT})
116: (both of second order
117: \cite{SG.luque94,SG.broeck94,SG.bro94b,SG.genovese},
118: and of first order \cite{SG.muller97,SG.zaikin99}),
119: noise-induced disordering phase transitions
120: %({\bf NIDT})
121: \cite{SG.broeck94,SG.genovese,SG.ojalvo96},
122: noise-induced phase dynamics
123: %({\bf NIPD})
124: \cite{SG.lacasta},
125: noise-induced fronts
126: %({\bf NIF})
127: \cite{SG.santos98},
128: and
129: noise-sustained structures in excitable media
130: %({\bf NSWSM})
131: \cite{SG.jung95,SG.kadar98,SG.hempel99}.
132:
133: Most of the noise-induced phenomena enumerated above act in the direction
134: of enhancing the order in the system. This surprising fact contrasts with
135: the noise-induced disorder that one could intuitively expect from statistical
136: mechanics. In the following pages, we review all these ordering effects.
137: In each of the different physical situations, the notion of ``order'' is
138: defined, the mechanisms through which noise produces order are discussed,
139: and a minimal model displaying the phenomenon is given.
140:
141: %The examples presented above on noise-induced some king of order, on
142: %the other hand, are counterintuitive phenomena:
143: %when fluctuations increase, order is surprisingly
144: %enhanced. In principle, these ordering effects seem to be related
145: %to the multiplicative character of the fluctuations, as compared
146: %to the disordering role of additive fluctuations. But recently this
147: %simple interpretation has changed; there is an interplay between additive
148: %and multiplicative noise terms in such a way that additive external noise
149: %can order the system \cite{SG.zaikin}. As a consequence, it appears that for
150: %some regimes
151: %additive noise can order the system, whereas multiplicative noise can
152: %disorder it.
153: %
154: %A review of all these striking phenomena is the
155: %main aim of this contribution.
156:
157: \section{Noise-Induced Phase Transitions}
158:
159: The most basic organizing phenomenon in spatially extended systems is the
160: transition between two macroscopic phases as a certain control parameter
161: is varied. Such {\em phase transitions} can be characterized by standard tools
162: in Statistical Mechanics, such as scaling functions, critical exponents,
163: and renormalization-group transformations \cite{SG.NISES}.
164:
165: From a dynamical perspective, the system can be described in a continuous way
166: by a coarse-grained field $\phi(\vec x,t)$, representing the local density
167: of a relevant physical variable (e.g., magnetization in a magnetic system,
168: or relative concentration in a binary alloy). From this viewpoint, a
169: disordered phase corresponds to the state $\phi(\vec x,t)=0$ (random
170: distribution of up- and down-spins in magnetic systems, or homogeneous mixture
171: in alloys), and an ordered phase is given by a non-zero field. As we will see
172: in what follows, external noise is able to induce phase transitions from
173: disorder to order. We consider models obeying a stochastic
174: differential equation of the general form:
175: \begin{equation}
176: \frac{\partial \phi(\vec x,t)}{\partial t} = f(\phi)+
177: g(\phi)\, \eta(\vec x,t) + \nabla^2 \phi + \xi(\vec x,t) \,,
178: \label{SG.eq:genmod}
179: \end{equation}
180: with $\vec x$ defined in a $d$-dimensional space. The additive and
181: multiplicative gaussian noises have zero mean and correlations:
182: \begin{subeqnarray}
183: & &\langle\xi(\vec x,t)\xi(\vec x',t')\rangle=2\varepsilon\;
184: \delta(\vec x-\vec x')\;\delta(t-t')\,,
185: \\
186: & &\langle\eta(\vec x,t)\eta(\vec x',t')\rangle=
187: 2\;c(\vec x-\vec x')\;\delta(t-t')\,,
188: \end{subeqnarray}
189: where $c(\vec x-\vec x')$ is the spatial correlation function of the
190: multiplicative noise $\eta(\vec x,t)$ [$c(0)$ is proportional to the noise
191: intensity]. The white additive noise $\xi(\vec x,t)$ is taken to represent
192: internal fluctuations.
193:
194: \subsection{A short-time dynamical instability}
195: \label{sec:stdi}
196:
197: We now review the most common mechanism of noise-induced ordering
198: transitions. To do so, we locally analyse the initial evolution of the
199: system. Averaging (\ref{SG.eq:genmod}) with respect to the probability
200: density of the two noise terms, and neglecting the diffusive term of the
201: equation:
202: \begin{equation}
203: \left\langle\partial_t \phi\right\rangle =
204: \langle f(\phi)\rangle+ \langle g(\phi)\, \eta(\vec x,t)\rangle \,.
205: \label{SG.eq:avmod1}
206: \end{equation}
207: We now interpret the multiplicative noise $\eta(\vec x,t)$ in the
208: Stratonovich sense \cite{SG.gardiner}. This choice is neither arbitrary
209: nor interested: we simply aim to describe realistic fluctuations, temporally
210: correlated but with a very small characteristic time.
211: Under this interpretation, the average of the noise
212: term appearing in (\ref{SG.eq:avmod1}) can be computed to be
213: \begin{equation}
214: \langle g(\phi)\,\eta(\vec x,t)\rangle=c(0)\langle g(\phi)g'(\phi)\rangle\,,
215: \label{eq:novav}
216: \end{equation}
217: where the prime indicates differentiation with respect to the argument
218: \cite{SG.NISES}.
219: Coming back to (\ref{SG.eq:avmod1}) one can say that, at the initial
220: instants of the evolution, fluctuations around the average value of the
221: field $\langle\phi\rangle$ can be neglected, so that $\langle h(\phi)\rangle
222: \approx h(\langle\phi\rangle)$ for any function $h$, and one can write
223: approximately
224: \begin{equation}
225: \partial_t \langle\phi\rangle\approx
226: f(\langle \phi\rangle)+ c(0) g(\langle\phi\rangle)g'(\langle\phi\rangle)
227: \equiv f_{\rm eff}(\langle\phi\rangle)\,.
228: \label{SG.eq:avmod2}
229: \end{equation}
230: In the zero-dimensional case (no spatial coupling), this approximation is
231: strictly valid only at short times. At long times, the system heads
232: towards the steady state, in which the noise effect is in fact
233: opposite to that of (\ref{SG.eq:avmod2}) (see \cite{SG.broeck96}). In the
234: spatially extended case, on the other hand, diffusive coupling between
235: neighbors is able to ``trap'' the system in the short-time dynamics given
236: by the effective force $f_{\rm eff}$ in (\ref{SG.eq:avmod2}). Hence, a simple
237: analysis of the zeroes of this function and their stability reveals, in
238: a na\"{\i}ve but very efficient way, the transition scenario of the system
239: \cite{SG.note1}.
240: We will now give some choices of $f$ and $g$, leading to different
241: noise-induced phase transitions.
242:
243: \subsection{Noise-Induced Second-Order Phase Transitions}
244:
245: Let us first consider the following representative model:
246: \begin{equation}
247: \partial_t \phi = -\phi(1+\phi^2)+
248: \phi\, \eta(\vec x,t) + \nabla^2 \phi + \xi(\vec x,t) \,,
249: \label{SG.eq:mod1}
250: \end{equation}
251: which corresponds, in the absence of multiplicative noise, to the well-known
252: time-dependent Ginzburg-Landau model, frequently used in studies of critical
253: dynamics \cite{SG.Ma76}, and whose universality class is that of the
254: Ising model. For the deterministic parameters chosen, the
255: system resides in the stable disordered state $\phi(\vec x,t)=0$. In the
256: presence of the multiplicative noise, and according to the discussion of
257: the previous paragraphs, the effective force (\ref{SG.eq:avmod2}) is equal
258: to $f_{\rm eff}(\phi)=[c(0)-1]\phi-\phi^3$, which shows that a non-zero
259: (ordered) value of the field can appear for large enough noise intensities
260: (for $c(0)>1$ in this approximate analysis). This noise-induced {\em
261: ordering} phase transition has been observed both numerically
262: \cite{SG.luque94} and theoretically, through linear stability \cite{SG.BK94}
263: and mean field \cite{SG.bro94b} analysis. The change of the spatially
264: averaged field $\frac{1}{V}\int \phi(\vec x,t)\,d\vec x$
265: at the bifurcation is continuous, which means that the
266: transition is of second order. The system also displays a subsequent phase
267: transition back to disorder, also of second order, for larger noise intensities
268: \cite{SG.ojalvo96}. In spite of the non-equilibrium character of the two
269: transitions, a finite-size scaling analysis shows that both of them belong to
270: the equilibrium universality class of the Ising model
271: \cite{SG.NISES}. The reason for this can be traced back to the existence
272: of the additive-noise term. In its absence (for which the disordered
273: phase becomes an absorbing state), a new universality class appears
274: \cite{SG.grinstein}.
275:
276: In spite of the simplicity of model (\ref{SG.eq:mod1}), most studies
277: performed so far on noise-induced phase transitions have been made in
278: a model introduced in 1994 by Van den Broeck et al. \cite{SG.broeck94},
279: which has an additional fifth-order saturating nonlinearity in the
280: deterministic force, and an external noise with both additive and
281: multiplicative contributions:
282: \begin{equation}
283: \partial_t \phi = -\phi(1+\phi^2)^2+
284: (1+\phi^2)\, \eta(\vec x,t) + \nabla^2 \phi \,.
285: \label{SG.eq:mod2}
286: \end{equation}
287: The associated effective force is $f_{\rm eff}(\phi)=[2c(0)-1]\phi+
288: 2[c(0)-1]\phi^3-\phi^5$. This system displays again two consecutive
289: noise-induced phase transitions, an ordering and a disordering one, both
290: of which belong to the Ising universality class \cite{SG.broeck96}
291: (which is not surprising, since the relevant terms of the deterministic
292: hamiltonian are the same as those of the Ginzburg-Landau model). The observed
293: phenomenology is basically equivalent to that of model (\ref{SG.eq:mod1}).
294: In particular, for both models the deterministic potential $U_{\rm det}=
295: -\int f(\phi)\,d\phi$ is monostable, so that the system is in all cases
296: disordered in the absence of external noise. Moreover, the stochastic
297: potential\footnote{The
298: stochastic potential $U_{\rm st}(\phi)$ is defined from the steady-state
299: probability density $P_{\rm st}(\phi)$ of the zero-dimensional system by means
300: of $P_{\rm st}(\phi)\sim\exp(-U_{\rm st})$ \protect\cite{SG.broeck96}.}
301: that can be defined for the local dynamics,
302: $U_{\rm st}=-\int [f(\phi)-c(0)g(\phi)g'(\phi)]/[c(0)
303: g^2(\phi)]\,d\phi$,
304: is also monostable for all $c(0)$ in the two cases, which indicates that
305: the corresponding zero-dimensional model does not have a noise-induced
306: transition towards order (towards non-zero field) for any of the two systems.
307:
308: In its original form, model (\ref{SG.eq:mod2}) cannot distinguish
309: between the additive and multiplicative contributions of the external
310: fluctuations to the noise-induced phenomena. Landa et al. \cite{SG.landa98}
311: slightly modified the model in order to examine the contribution of the
312: additive-noise term, and discovered the existence of noise-induced phase
313: transitions (ordering and disordering) controled by additive noise, provided
314: a multiplicative-noise noise exists.
315:
316: Another similarity between the noise-induced phase transitions exhibited
317: by models (\ref{SG.eq:mod1}) and (\ref{SG.eq:mod2}) is that, in both cases,
318: the mechanism through which noise destabilizes the disordered phase is
319: linear, as can be easily seen by examining the effective forces in the
320: two situations. A review of linear instability mechanisms of noise-induced
321: phase transitions is given in \cite{SG.marta3}. On the other hand, by taking
322: into account the discussion of Sect. \ref{sec:stdi}, one can
323: devise in a straightforward way models for which a noise-induced ordering
324: phase transition is driven by a {\em nonlinear} mechanism. The system
325: \begin{equation}
326: \partial_t \phi = -\phi^3(1+\phi^2)+
327: \phi^2\,\eta(\vec x,t) + \nabla^2 \phi + \xi(\vec x,t)\,,
328: \label{SG.eq:mod3}
329: \end{equation}
330: which has an effective force $f_{\rm eff}(\phi)=[2c(0)-1]\phi^3-\phi^5$,
331: is an example of this fact. In this case, the destabilization of the
332: disordered phase $\phi(\vec x,t)=0$ by noise is dynamically nonlinear.
333:
334: \subsection{Noise-Induced First-Order Phase Transitions}
335:
336: In all previous examples, the 0-d short-time effective force $f_{\rm eff}
337: (\phi)$ displays a supercritical pitchfork bifurcation controlled by the noise
338: intensity, corresponding to a continuous (second-order) phase transition
339: when spatial coupling is taken into account. Suitable choices of $f(\phi)$
340: and $g(\phi)$, on the other hand, lead to discontinuous bifurcations in
341: $f_{\rm eff}(\phi)$, associated to first-order noise-induced phase
342: transitions. A first simple example is given by
343: \begin{equation}
344: \partial_t \phi= -\phi+\phi^3-\phi^5+
345: \phi\,\eta(\vec x,t) + \nabla^2 \phi + \xi(\vec x,t)\,.
346: \label{SG.eq:mod5}
347: \end{equation}
348: In this case, the effective force can be seen to be $f_{\rm eff}(\phi)=
349: [c(0)-1]\phi+\phi^3-\phi^5$, which displays a subcritical pitchfork
350: bifurcation controlled by noise that corresponds to a first-order phase
351: transition in the spatially extended case. This transition is characterized
352: by an abrupt change in the spatially averaged field and by a region of
353: bistability between ordered and disordered states. Such features were
354: observed by M\"uller et al. \cite{SG.muller97} in a model with the
355: deterministic force of (\ref{SG.eq:mod5}) and an external noise with
356: both additive and multiplicative contributions. The separate role of
357: additive fluctuations was investigated in a modified version of model
358: (\ref{SG.eq:mod2}), and a first-order phase transition induced by
359: additive noise was found \cite{SG.zaikin99}, provided multiplicative
360: noise is present.
361:
362: The noise-driven discontinuous phase transition displayed by
363: (\ref{SG.eq:mod5}) is produced by a destabilization of the linear
364: coefficient of the 0-d effective force.
365: Similarly to the case of second-order phase transitions, models can be
366: defined that exhibit noise-induced first-order phase transitions driven
367: by a nonlinear mechanism. The simplest example is:
368: \begin{equation}
369: \partial_t \phi = -\phi(1+\phi^4)+
370: \phi^2\,\eta(\vec x,t) + \nabla^2 \phi + \xi(\vec x,t)\,,
371: \label{SG.eq:mod6}
372: \end{equation}
373: for which $f_{\rm eff}(\phi)=-\phi+2c(0)\phi^3-\phi^5$. This function
374: describes a subcritical pitchfork bifurcation at $c(0)=1$ in 0-d, and
375: hence a nonlinearly-driven discontinuous phase transition can be expected
376: in the spatially extended case.
377:
378: All examples presented in the last two Sections have been founded on
379: an analysis of the effective force that governs the short-time behavior of
380: the zero-dimensional system, which we already argued that would become
381: trapped by spatial coupling in the extended case. The conclusions obtained
382: by this approach can be confirmed by other methods, such as mean-field
383: approximations \cite{SG.marta1} and numerical simulations of the complete
384: models.
385:
386: \subsection{Noise-Induced Phase Dynamics}
387:
388: We have just seen that external noise can induce non-equilibrium phase
389: transitions, both of first and second order, between two {\em stationary}
390: phases. Now we aim to analyse the {\em dynamical} aspects related to the
391: appearance of a non-equilibrium phase in the system.
392: Let us consider a stationary phase, in equilibrium or not, with well-defined
393: properties (symmetries). If a control parameter, such as temperature, is
394: suddenly changed, the system undergoes a dynamical process in order to
395: reach a new steady state corresponding to the new value of the control
396: parameter. If the initial and final states belong to the same phase, the
397: dynamical process is simple and linear relaxation dynamics can be used to
398: model it. On the other hand, if the final state belongs to a phase different
399: from the initial one, then a pattern dynamics appears. The pattern is
400: composed of topological defects which evolve towards the new steady state.
401: This process is controlled by a small number of parameters, and exhibits
402: dynamics of different universality classes, which are well characterized by
403: a dynamical exponent and by the scaling properties of the structure function.
404: Two universality classes of particular interest are those of phase ordering
405: and phase separation.
406:
407: %In these two examples the dynamical process takes place in the presence of
408: %two coexisting phases and as a consequence one can speak of {\em dynamics
409: %of first order phase transitions}.
410:
411: In the following paragraphs, we examine whether external noise can induce
412: these dynamical processes, and what is the resulting dynamical universality
413: class. The answer to the first question is simple: since external noise can
414: induce ordering phase transitions, it is also able to induce the corresponding
415: dynamics. The second question is much more involved. We can advance that the
416: dynamical universality class is not changed, because the physical growth
417: mechanisms are the same.
418:
419: \subsubsection{Noise-induced phase ordering.}
420: This dynamics arises in ferromagnetic systems, when a sudden decrease in
421: temperature drives the system away from a homogeneous state of zero
422: magnetization towards a final state of finite magnetization. During this
423: process, the system exhibits magnetic domains, some of which grow at the
424: expense of the rest until the whole system presents a homogeneous finite
425: magnetization. The characteristic size of these domains grows according to
426: the Allen-Cahn law, $R(t) \sim t^{1/2}$.
427:
428: A representative model of this situation is that defined by (\ref{SG.eq:mod1}).
429: We have shown that external noise can induce an ordering phase transition in
430: that model. Therefore, beyond the transition point, noise can be considered
431: to drive an initially disordered phase into an ordered state, through the
432: formation of domains of the two coexisting (positive and negative) ordered
433: phases and their subsequent dynamics of competition as in equilibrium models.
434: Numerical simulations confirm this noise-induced dynamics \cite{SG.marta2}.
435: Moreover, the process has the characteristics of the Allen-Cahn dynamical
436: universality class: a clear evidence of the scaling behavior of the
437: structure function with a power law $ \sim t^{1/2}$ is observed. The fact
438: that the universality class is not changed can be understood because the
439: driving mechanism of the ordering process is the same as in the deterministic
440: case: the local curvature of the interface.
441:
442: \subsubsection{Noise-induced phase separation.}
443: This universality class arises in homogeneous alloys of two atomic
444: species\footnote{Phase
445: separation in fluids is another universality class not discussed here.}
446: initially at high temperature. Following a sudden cooling,
447: the two components of the alloy start to segregate, producing domains rich in
448: one of the species, whose characteristic size follows the Lifshitz-Slyozov
449: law, $R(t)\sim t^{1/3}$. The system evolves towards a final state consisting
450: on two large domains separated by an interphase.
451:
452: A dynamical model of phase separation can be constructed from model
453: (\ref{SG.eq:mod1}) if mass conservation is imposed. The explicit model reads:
454: \begin{equation}
455: \partial_t \phi = \nabla^2 \left( \phi +
456: \phi^3 - \nabla^2 \phi + \phi \,\eta({\bf x},t) \right) + \nabla\cdot \vec\xi
457: (\vec x,t)\;,
458: \end{equation}
459: where $\vec \xi(\vec x,t)$ is now a random {\em vector} field. This system
460: evolves with the restriction that the spatially averaged field,
461: $\frac{1}{V}\int \phi(\vec x,t)\,d\vec x=\phi_0$, is conserved. As in the
462: case of model (\ref{SG.eq:mod1}), the external multiplicative noise
463: $\eta(\vec x,t)$ induces an ordering phase transition, followed by a
464: disordering one. In the ordered region, domains of the two non-zero phases
465: coexist and try to grow. However, the dynamics is different from the case
466: of phase ordering. Owing to the conservation law, domains have to grow at
467: the expense of other domains, which may be located far away. The driving
468: mechanism is thus diffusion controlled by local curvature, as in the
469: deterministic case, and hence we expect the same dynamical universality
470: class of Lifshitz-Slyozov. Numerical simulations confirm this fact
471: \cite{SG.marta2}. A scaling behavior of the structure function following
472: the power law $ \sim t^{1/3}$ is observed.
473:
474: \section{Noise-Induced Structures}
475:
476: Besides phase transitions, a second and very important ordering mechanism
477: in spatially extended systems is that of {\em structure formation}. The
478: spontaneous appearance and sustenance of spatiotemporal patterns is common
479: to many non-equilibrium extended media, including hydrodinamical, optical,
480: chemical and biological systems \cite{SG.cross}. The system is usually
481: described by a stochastic partial differential equation (or a set of
482: them), for which a self-sustained non-homogeneous solution is considered
483: to be an ordered state. For instance, in the case of a single-field
484: system, an ordered state is represented by a solution $\phi(\vec x,t)$
485: which depends explicitly on $\vec x$, and which may or may not depend on
486: time. We will now review the effects of external noise on several different
487: pattern-forming systems.
488:
489: \subsection{Noise-Induced Stationary Patterns}
490:
491: Let us begin by examining whether noise is able to induce stationary
492: non-homogeneous (ordered) states in a pattern-forming system. The standard
493: model displaying such a phenomenology is:
494: \begin{equation}
495: \partial_t \phi = -\phi(1+\phi^2)+
496: \phi\, \eta(\vec x,t) - \left(\nabla^2+k_0^2\right)^2 \phi + \xi(\vec x,t) \,,
497: \label{SG.eq:she}
498: \end{equation}
499: which corresponds to the well-known Swift-Hohenberg model, widely used to
500: describe the formation of stationary structures in hydrodynamics and nonlinear
501: optics, among other fields. For the parameter region chosen, the stationary
502: solution of (\ref{SG.eq:she}) in the absence of multiplicative noise is the
503: disordered state $\phi(\vec x,t)=0$. When the external fluctuations are
504: considered, the homogeneous solution becomes unstable, as can be seen in
505: a simple way from a linear stability analysis of the first statistical
506: moment of $\phi$. To that end, we linearise (\ref{SG.eq:she}), transform
507: it to Fourier space and average the resulting equation with respect to
508: both noise distributions, which leads to:
509: \begin{equation}
510: \partial_t \langle\phi\rangle = \left[c(0)-1
511: -\left(k^2-k_0^2\right)^2\right]\langle\phi\rangle\,.
512: \label{SG.eq:sheav}
513: \end{equation}
514: This analysis shows that the disordered phase $\phi(\vec x,t)=0$ becomes
515: unstable at $c(0)=1$ for a non-zero wavenumber $k_0$, corresponding to
516: the appearance of a periodic pattern of wavelength $2\pi/k_0$. A similar
517: conclusion, at least at first order in noise intensity, can be observed when
518: the stability of higher-order statistical moments is analysed\footnote{This
519: fact contrasts with the case of 0-d systems, where statistical
520: moments of different order have different instability thresholds at all orders
521: \cite{SG.ojalvo96a}.}.
522:
523: The model of Van den Broeck et al. \cite{SG.broeck94} with the addition of
524: a Swift-Hohenberg-like spatial coupling term has also been examined in
525: search of noise-induced patterns. They have been observed coming from
526: multiplicative \cite{SG.par96} and additive \cite{SG.zaikin} noise.
527:
528: \subsection{Noise-Induced Propagation}
529:
530: The simplest non-stationary phenomenon in a spatially extended system is the
531: propagation with constant velocity of a structure through the system. We will
532: see in what follows that external noise is also able to induce such an ordering
533: effect.
534:
535: \subsubsection{Noise-Induced Fronts.}
536: A front is a spatiotemporal structure linking two different homogeneous
537: states (kink). Experiments have shown that noise supports front propagation
538: in a chain of bistable diode resonators \cite{SG.locher}.
539: We now examine the effect of multiplicative noise on the one-dimensional
540: propagation of a front in the following field equation:
541: \begin{equation}
542: \frac{\partial \phi}{\partial t} = \frac{\partial^2 \phi}{\partial x^2}
543: - \phi(1 + \phi^2)+\phi\,\eta(\vec x,t)\,,
544: \label{SG.eqfronts}
545: \end{equation}
546: In the absence of noise, the homogeneous phase $\phi(x,t)=0$ is the only
547: stable state of the system. Neither fronts nor any kind of spatial
548: structure can be sustained at large times; any initial condition
549: $\phi(x,0)\neq 0$ decays rapidly to the above-mentioned disordered state.
550: In the presence of noise, a systematic contribution to the deterministic
551: dynamics arises as shown in Sect. \ref{sec:stdi}. One can therefore write
552: an effective model that has the same behavior as (\ref{SG.eqfronts})
553: on average:
554: \begin{equation}
555: \frac{\partial \phi}{\partial t} = \frac{\partial^2 \phi}{\partial x^2}
556: - \phi(1 -c(0)+ \phi^2)+\xi_{\rm eff}(\vec x,t)\,,
557: \label{SG.eqfrontseff}
558: \end{equation}
559: where the effective noise $\xi_{\rm eff}(\vec x,t)$ has zero mean. According
560: to this equation, the homogeneous zero state becomes unstable for $c(0)>1$.
561: Under this condition, an initial localised perturbation produces a
562: kink-antikink pattern that propagates until a non-zero state $\overline
563: {\phi}_{st} \sim \sqrt{c(0) -1}$ invades all the system. The mean velocity
564: of the propagating front can be computed to be $\overline{v} = 2
565: \sqrt{c(0)-1}$. These analytical predictions can be confirmed by numerical
566: simulations of the exact model \cite{SG.santos98}.
567:
568: \subsubsection{Noise-Sustained Signal Propagation.}
569: Propagating kink-antikink combinations behave as a train of traveling pulses
570: that can act as information bits in a communication system. In a chain of
571: asymmetrical double-well oscillators, these pulses are intrinsically unstable
572: in the absence of noise. Multiplicative noise is able to sustain propagation
573: of these pulses, as can be understood in a simple way by analysing the
574: following model:
575: \begin{equation}
576: \label{SG.eq:pulse}
577: \frac{\partial \phi}{\partial t} = \phi(1-\phi)(\phi-a)+
578: D\frac{\partial^2\phi}{\partial x^2} +v\frac{\partial \phi}{\partial x}
579: +\phi\;\eta(x,t)\,.
580: \end{equation}
581: This equation contains a diffusive and a convective term in order to model
582: a uni-directional coupling in this one-dimensional system.
583: For $1/2<a<1$, the deterministic potential is an asymmetric double well,
584: with $\phi=0$ more stable than $\phi=1$ (which are the two fixed points of
585: the system). Therefore, in the absence of noise a kink-antikink pulse shrinks
586: and decays as it propagates. Noise is able to prevent this decay,
587: as can be seen by writing an effective model with a zero-mean noise, as
588: has been made in the previous section.
589: %\begin{equation}
590: %\label{SG.eq:pulseeff}
591: %\frac{\partial \phi}{\partial t} = \phi(1-\phi)(\phi-a)+c(0)\phi+
592: %D\frac{\partial^2\phi}{\partial x^2} +v\frac{\partial \phi}{\partial x}
593: %+\xi_{\rm eff}(x,t)\,.
594: %\end{equation}
595: It is easy to see that an
596: optimal value of noise intensity renders the effective potential of this
597: model symmetric \cite{SG.pulse}, which allows sustained propagation of
598: pulses through the system.
599:
600: \subsection{Noise-Supported Structures in Excitable Media}
601:
602: Due to their special characteristics, excitable systems are especially
603: sensitive to noise. Perturbations due to noise are able, for instance, to
604: make the system jump from the rest to the excited state. But besides
605: nucleating excitation pulses, external noise can also sustain their
606: propagation in the subexcitable regime (in which such motion would not
607: be possible under deterministic conditions). This fact has been observed
608: numerically for the case of
609: spiral waves \cite{SG.jung95}, in what provided the first example, up
610: to our knowledge, of spatio-temporal stochastic resonance. Additionally,
611: chemical subexcitable media have also provided the first experimental
612: evidence of noise-sustained pulse propagation \cite{SG.kadar98}. Such
613: constructive effects of noise can be modeled by cellular automata
614: \cite{SG.hempel99}; the obtained results lead also to predictions of
615: noise-induced synchronization and global oscillations. Another possible
616: modeling procedure is by means of continuous FitzHugh-Nagumo models of
617: excitable media. Analyses in this direction have shown that external
618: multiplicative noise is able to support spiral turbulent states in
619: simple activator-inhibitor models \cite{SG.ojalvo99}, and that additive
620: noise induces synchronization phenomena \cite{SG.neiman99}.
621:
622: \subsection{Other Constructive Effects of Noise}
623:
624: There are several other examples of noise-induced order in extended media
625: that have not been described in the previous pages. We have already briefly
626: mentioned the phenomenon of spatiotemporal stochastic resonance in excitable
627: media \cite{SG.jung95}. In that case, a certain non-zero but finite value of
628: noise intensity exists for which the propagation of excitation waves is
629: optimal. The extension to spatially extended systems of the phenomenon of
630: stochastic resonance can be understood in other ways. In its purest
631: interpretation, it corresponds to the enhanced response of a spatially
632: bistable system (i.e., whose Lyapunov functional has two minima corresponding
633: to two stable spatiotemporal states) to a harmonic signal \cite{SG.wio}.
634:
635: Another constructive effect of noise, in this case not as counterintuitive,
636: is the sustenance of drifting structures is systems with a convection term
637: \cite{SG.deissler,SG.palma2}. In this case, structures would escape through
638: the system boundaries, swept by convection, in the absence of noise.
639: Fluctuations ensure a continuous creation of structures, which are therefore
640: sustained by noise.
641:
642: Noise has also been seen to have a constructive influence in globally
643: coupled systems. As an example, a model of interacting Brownian particles
644: in a periodic potential has been recently found to exhibit a noise-induced
645: phase transition \cite{SG.reimann99}, in which the ordered phase can be
646: interpreted as a ratchet-like transport mechanism, even though the underlying
647: potential is symmetric.
648:
649: \section{Conclusions}
650:
651: We have tried to review, in a clear and pedagogical way, the ordering
652: influence that external noise exerts in the spatiotemporal dynamics of
653: extended media. Two main topics have been considered, namely noise-induced
654: phase transitions and noise-induced structure formation. In each situation,
655: a specific notion of order has been introduced. In the case of phase
656: transitions, order is understood in a coarse-graining sense, so that it
657: corresponds to a non-zero (even uniform) field. In the case of
658: pattern formation, order is defined in opposition to uniformity, so
659: that it corresponds to a non-uniform field profile, either depending or
660: not on time. In any of the two cases, external noise can be seen to induce
661: order. In each particular situation, the ordered phase can be characterised
662: by standard tools used in the corresponding deterministic (or equilibrium)
663: phenomenology. We interpret multiplicative noise in the Stratonovich sense,
664: impelled by a search of realistic modeling of the fluctuations. Phase
665: transitions induced by multiplicative noise in the Ito interpretation
666: have been recently found \cite{SG.munoz}, but they have a disordering
667: character. Finally, we should also note that all the phenomena reviewed
668: here can be explained by the short-time dynamical instability mechanism
669: described in Sect. \ref{sec:stdi}. But ordering transitions exist that
670: are driven by a different mechanism, an investigation of which is currently
671: in progress \cite{SG.marta4}.
672:
673: \section*{Acknowledgments}
674:
675: We acknowledge financial support from the Direcci\'on General de Ense\~nanza
676: Superior (Spain), under projects PB96-0241 and PB98-0935. J.G.O. is pleased
677: to thank Prof. Lutz Schimansky-Geier, to whom this work is dedicated, for
678: fruitful collaboration in recent years on this topic.
679:
680: \begin{thebibliography}{99}
681:
682: \bibitem{SG.Gamma98} L. Gammaitoni, P. H\"anggi, P. Jung, and F. Marchesoni,
683: %``Stochastic resonance,"
684: Rev. Mod. Phys. {\bf 70}, 223 (1998).
685:
686: \bibitem{SG.horsthemke84} W. Horsthemke and R. Lefever,
687: {\em Noise-induced transitions} (Springer, Berlin, 1984).
688:
689: %\bibitem{SG.HH} P.C. Hohenberg and B.I. Halperin,
690: %``Theory of dynamic critical phenomena,"
691: %Rev. Mod. Phys. {\bf 49}, 435 (1977).
692:
693: \bibitem{SG.Ma76} S.K. Ma, {\em Modern theory of critical phenomena}
694: (Benjamin, Reading, 1976).
695:
696: \bibitem{SG.cross} M.C. Cross and P.C. Hohenberg,
697: %``Pattern formation outside of equilibrium,"
698: Rev. Mod. Phys. {\bf 65}, 851 (1993).
699:
700: \bibitem{SG.NISES} J. Garc\'{\i}a-Ojalvo and J.M. Sancho, {\em Noise
701: in Spatially Extended Systems} (Springer, New York, 1999).
702:
703: \bibitem{SG.mikhailov79} A.S. Mikhailov,
704: %``Noise-induced phase transition in a biological system with diffusion,"
705: Phys. Lett. A {\bf 73}, 143 (1979).
706:
707: \bibitem{SG.mikhailov81} A.S. Mikhailov,
708: %``Effects of diffusion in fluctuating media: a noise-induced phase transition,"
709: Z. Phys. B {\bf 41}, 277 (1981).
710:
711: \bibitem{SG.ojalvo93} J. Garc\'{\i}a-Ojalvo, A. Hern\'andez-Machado,
712: and J.M. Sancho,
713: %``Effects of external noise on the Swift--Hohenberg equation,"
714: Phys. Rev. Lett. {\bf 71}, 1542 (1993).
715:
716: \bibitem{SG.BK94} A. Becker and L. Kramer,
717: %``Linear stability analysis for bifurcations in spatially extended
718: %systems with fluctuating control parameter,"
719: Phys. Rev. Lett. {\bf 73}, 955 (1994).
720:
721: \bibitem{SG.par96} J.M.R. Parrondo, C. Van den Broeck, J. Buceta,
722: and J. de la Rubia,
723: %``Noise-induced spatial patterns,"
724: Physica A {\bf 224}, 153 (1996).
725:
726: \bibitem{SG.zaikin} A.A. Zaikin and L. Schimansky-Geier,
727: %``Spatial patterns induced by additive noise,"
728: Phys. Rev. E {\bf 58}, 4355 (1998).
729:
730: \bibitem{SG.luque94} P. Luque, J. Garc\'{\i}a-Ojalvo, and J.M. Sancho,
731: %``Nonequilibrium phase transitions and external noise,"
732: in {\em Fluctuation phenomena: disorder and nonlinearity},
733: edited by A.R. Bishop, S. Jim\'enez, and L. V\'azquez, p. 75 (World
734: Scientific, Singapore, 1995).
735:
736: \bibitem{SG.broeck94} C. Van den Broeck, J.M.R. Parrondo, and R. Toral,
737: %``Noise-induced nonequilibrium phase transitions,"
738: Phys. Rev. Lett. {\bf 73}, 3395 (1994).
739:
740: \bibitem{SG.bro94b} C. Van den Broeck, J.M.R. Parrondo, J. Armero, and
741: A. Hern\'andez-Machado,
742: %``Mean field model for spatially extended systems in the presence of
743: %multiplicative noise,"
744: Phys. Rev. E {\bf 49}, 2639 (1994).
745:
746: \bibitem{SG.genovese} W. Genovese, M.A. Mu\~noz, and J.M. Sancho,
747: %``Nonequilibrium transitions induced by multiplicative noise,"
748: Phys. Rev. E {\bf 57}, 2495 (1998).
749:
750: \bibitem{SG.muller97} R. M\"uller, K. Lippert, A. K\"uhnel, and U. Behn,
751: %``First-order nonequilibrium phase transition in a spatially extended system,"
752: Phys. Rev. E {\bf 56}, 2658 (1997).
753:
754: \bibitem{SG.zaikin99} A.A. Zaikin, L. Schimansky-Geier, and
755: J. Garc\'{\i}a-Ojalvo,
756: %``Nonequilibrium first-order phase transition induced by additive noise,"
757: Phys. Rev. E {\bf 60}, 6275 (1999).
758:
759: \bibitem{SG.ojalvo96} J. Garc\'{\i}a-Ojalvo, J.M.R. Parrondo, J.M
760: Sancho, and C. Van den Broeck,
761: %``Reentrant transition induced by
762: %multiplicative noise in the time-dependent Ginzburg--Landau model,"
763: Phys. Rev. E {\bf 54}, 6918 (1996).
764:
765: \bibitem{SG.lacasta} J. Garc\'{\i}a-Ojalvo, A.M. Lacasta, J.M. Sancho,
766: and R. Toral,
767: %``Phase separation driven by external fluctuations,"
768: Europhys. Lett. {\bf 42}, 125 (1998).
769:
770: \bibitem{SG.santos98} M.A. Santos and J.M. Sancho,
771: %``Noise induced fronts,"
772: Phys. Rev. E {\bf 59}, 98 (1999).
773:
774: \bibitem{SG.jung95} P.~Jung and G.~Mayer--Kress,
775: Phys.~Rev.~Lett. {\bf 74}, 2134 (1995).
776:
777: \bibitem{SG.kadar98} S.~K\'ad\'ar, J.~Wang, and K.~Showalter,
778: Nature {\bf 391}, 770 (1998).
779:
780: \bibitem{SG.hempel99} H. Hempel, L. Schimansky-Geier, and J. Garc\'{\i}a-Ojalvo,
781: %``Noise-sustained pulsating patterns and global oscillations in subexcitable
782: %media,''
783: Phys. Rev. Lett. {\bf 82}, 3713 (1999).
784:
785: \bibitem{SG.gardiner} C.W. Gardiner, {\em Handbook of stochastic
786: methods} (Springer, Berlin, 1989).
787:
788: \bibitem{SG.broeck96} C. Van den Broeck, J.M.R. Parrondo, R. Toral, and
789: R. Kawai,
790: %``Nonequilibrium phase transitions induced by multiplicative noise,"
791: Phys. Rev. E {\bf 55}, 4084 (1997).
792:
793: \bibitem{SG.note1} It should be noted that this short-time
794: instability approach does not account for the quantitative influence
795: of the spatial coupling strength, since this quantity does not appear
796: explicitly in the analysis given in
797: (\protect\ref{SG.eq:avmod1})-(\protect\ref{SG.eq:avmod2}).
798: To take such an effect into account, a mean-field approach can be used
799: \protect\cite{SG.bro94b,SG.marta1}.
800:
801: \bibitem{SG.grinstein} Y. Tu, G. Grinstein, and M.A. Mu\~noz,
802: %``Systems with multiplicative noise: critical behavior from KPZ equation and
803: %numerics,"
804: Phys. Rev. Lett. {\bf 78}, 274 (1997).
805:
806: \bibitem{SG.landa98} P.S. Landa, A.A. Zaikin, and L. Schimansky-Geier,
807: %``Influence of additive noise on noise-induced
808: %phase transitions in nonlinear chains,"
809: Chaos, Solitons and Fractals {\bf 9}, 1367 (1998).
810:
811: \bibitem{SG.marta3} M. Iba\~nes, R. Toral, J. Garc\'{\i}a-Ojalvo, and
812: J.M. Sancho,
813: %``Linear instability mechanisms of noise-induced phase transitions,''
814: this book.
815:
816: \bibitem{SG.marta1} M. Iba\~nes, J. Garc\'{\i}a-Ojalvo, R. Toral, and
817: J.M. Sancho,
818: %``Noise-induced phase separation: mean field results,''
819: Phys. Rev. E {\bf 60}, 3597 (1999).
820:
821: \bibitem{SG.marta2} M. Iba\~nes, J.M. Sancho, R. Toral, and
822: J. Garc\'{\i}a-Ojalvo,
823: %``Dynamics and scaling properties of noise-induced phase separation,"
824: unpublished.
825:
826: \bibitem{SG.ojalvo96a} J. Garc\'{\i}a-Ojalvo and J.M. Sancho,
827: %``External fluctuations in a pattern-forming instability,"
828: Phys. Rev. E {\bf 53}, 5680 (1996).
829:
830: \bibitem{SG.locher} M. L\"ocher, D. Cigna, and E.R. Hunt, Phys. Rev. Lett.
831: {\bf 80}, 5212 (1998).
832:
833: \bibitem{SG.pulse} J. Garc\'{\i}a-Ojalvo, A.M. Lacasta, F. Sagu\'es, and
834: J.M. Sancho, preprint (2000).
835:
836: \bibitem{SG.ojalvo99} J. Garc\'{\i}a-Ojalvo and L. Schimansky-Geier,
837: Europhys. Lett. {\bf 47}, 298 (1999).
838:
839: \bibitem{SG.neiman99} A. Neiman, L. Schimansky-Geier, A. Cornell-Bell,
840: and F. Moss,
841: Phys. Rev. Lett. {\bf 83}, 4896 (1999).
842:
843: \bibitem{SG.wio} H.S. Wio,
844: %``Stochastic resonance in a spatially extended system,"
845: Phys. Rev. E {\bf 54}, R3075 (1996).
846:
847: \bibitem{SG.deissler} R.J. Deissler,
848: %``External noise and the origin and dynamics of structure in
849: %convectively unstable systems,"
850: J. Stat. Phys. {\bf 54}, 1459 (1989).
851:
852: \bibitem{SG.palma2} M. Santagiustina, P. Colet, M. San Miguel, and
853: D. Walgraef,
854: %``Noise-sustained convective structures in nonlinear optics,"
855: Phys. Rev. Lett. {\bf 79}, 3633 (1997).
856:
857: \bibitem{SG.reimann99} P. Reimann, R. Kawai, C. Van den Broeck, and
858: P. H\"anggi, Europhys. Lett. {\bf 45}, 545 (1999).
859:
860: \bibitem{SG.munoz} W. Genovese and M.A. Mu\~noz,
861: %``Recent results on multiplicative noise,"
862: Phys. Rev. E {\bf 60}, 69 (1999).
863:
864: \bibitem{SG.marta4} M. Iba\~nes, R. Toral, J.M. Sancho, and
865: J. Garc\'{\i}a-Ojalvo,
866: unpublished.
867:
868: \end{thebibliography}
869:
870: \end{document}
871: