1: % Version del 3/04/2003 (MH)
2: \newcommand{\HR}[1]{{\bf HR}\footnote{\textbf{HR: {#1}}}}
3: \newcommand{\MH}[1]{{\bf MH}\footnote{\textbf{MH: {#1}}}}
4: \newcommand{\MS}[1]{{\bf MS}\footnote{\textbf{MS: {#1}}}}
5: \newcommand{\HRC}[2]{\footnote{HR: #1}{\it #2}}
6: %\newcommand{\HRC}[2]{{\it #2}}
7: %\newcommand{\HRC}[2]{{ #2}}
8: \newcommand{\MHC}[2]{\footnote{MH: #1}{\it #2}}
9: %\newcommand{\MHC}[2]{{\it #2}}
10: \newcommand{\MSC}[2]{{\it #2}}
11: %\newcommand{\MSC}[2]{\footnote{MS: #1}{\it #2}}
12: %\documentclass{elsart}
13: \documentclass[final]{siamltex}
14:
15: \usepackage{oldlfont} %uses LaTeX 2.09 fonts, e.g., \cal%
16: \usepackage{amsmath, amsfonts}
17: \usepackage{graphicx,color}
18: \input{epsf}
19: \input amssym.def
20: \input amssym
21:
22: \renewcommand{\topfraction}{1}
23: \renewcommand{\textfraction}{0}
24: \renewcommand{\bottomfraction}{1}
25:
26: \title{Near-Resonant, Steady Mode Interaction: Periodic,
27: Quasi-Periodic and Localized Patterns\thanks{The work of MH was
28: supported by the Spanish Ministerio de Ciencia y Tecnolog\'ia (BFM2001-2363).
29: The work of HR was supported by NASA grant NAG3-2113, the Department
30: of Energy (DE-FG02-92ER14303) and NSF (DMS-9804673). The work of MS was
31: supported by NASA grant NAG3-2364, and by NSF under DMS-9972059
32: and the MRSEC Program under DMR-0213745.}}
33:
34: \author{Mar\'ia Higuera\thanks{E. T. S. Ingenieros Aeron\'auticos,
35: Universidad Polit\'ecnica de Madrid, Plaza Cardenal Cisneros 3, 28040 Madrid, Spain.
36: ({\tt maria@fmetsia.upm.es}).}
37: \and Hermann Riecke\thanks{Department of Engineering Sciences and
38: Applied Mathematics, Northwestern University, Evanston, IL 602028, USA.}
39: \and Mary Silber\thanks{Department of Engineering Sciences and
40: Applied Mathematics, Northwestern University, Evanston, IL 602028, USA.}}
41:
42:
43: \bibliographystyle{siam}
44:
45: \begin{document}
46:
47: \maketitle
48:
49: \begin{abstract}
50:
51: Motivated by the rich variety of complex periodic and quasi-periodic
52: patterns found in systems such as two-frequency forced Faraday waves,
53: we study the interaction of two spatially periodic modes that are
54: {\it nearly resonant}. Within
55: the framework of two coupled one-dimensional Ginzburg-Landau equations
56: we investigate analytically the stability of the periodic solutions to general
57: perturbations, including perturbations that do not respect the
58: periodicity of the pattern, and which may lead to quasi-periodic solutions. We study the
59: impact of the deviation from exact resonance on the destabilizing
60: modes and on the final states. In regimes in which the mode
61: interaction leads to traveling waves our numerical simulations reveal
62: localized waves in which the wavenumbers are resonant and which
63: drift through a steady background pattern that has an off-resonant wavenumber
64: ratio.
65:
66: \end{abstract}
67:
68: % BEGIN MACRO DEFINITIONS
69: \def\ep{\epsilon}
70: \def\bs{\boldsymbol}
71: \def\O{\hbox{\cal O}}
72: \def\pref#1{(\ref{#1})}
73: %\newcommand{\cierto}[1]{\fbox{#1}\label{#1}}
74: \newcommand{\cierto}[1]{\label{#1}}
75: \newcommand\cc{\text{c.c.}}
76:
77:
78: \section{Introduction}
79:
80: Pattern-forming instabilities lead to an astonishing variety of
81: spatial and spatio-temporal structures, ranging from simple, periodic
82: stripes (rolls) to spatially localized structures and
83: spatio-temporally chaotic patterns. Even within the restricted class
84: of steady, spatially ordered patterns a wide range of
85: patterns \MSC{has}{have} been identified and investigated beyond
86: simple square or hexagonal planforms including patterns exhibiting
87: multiple length scales: superlattice patterns, in which the length
88: scales involved are rationally related rendering the pattern periodic
89: albeit on an unexpectedly large length scale, and quasipatterns, which
90: are characterized by incommensurate length scales and which are
91: therefore not periodic in space. These more complex two-dimensional
92: patterns have been observed in particular in the form of Faraday
93: waves on a fluid
94: layer that is vertically shaken with a two-frequency periodic
95: acceleration function \cite{EdFa94,ArFi98,KuPi98,ArFi02}, and to some
96: extent also in vertically vibrated Rayleigh-B\'enard
97: convection \cite{RoSc00a} and in nonlinear optical systems \cite{PaRa95}. The
98: Faraday system is especially suitable for experimental investigation
99: of pattern formation in systems with two competing spatial modes of
100: instability since such codimension-two points are easily accessible by
101: simply adjusting the frequency content of the periodic forcing
102: function~\cite{EdFa94,BeEd96}. The observation of both
103: superlattices and quasipatterns in this physical system raises an
104: intriguing question concerning the selection of these kinds of
105: patterns: what determines whether, for given physical parameters, a
106: periodic or a quasi-periodic pattern is obtained? This provides the
107: main motivation for the present paper in which we address certain
108: aspects of the selection problem within the somewhat simple framework
109: of mode competition in one spatial dimension.
110:
111: To capture the competition between commensurate/incommensurate length
112: scales we focus on the interaction between two modes with a
113: wavelength ratio that is close to, but not necessarily equal to, the
114: ratio of two small integers. We are thus led to consider a {\it
115: near-resonant} mode interaction. We focus on the case in which both
116: modes arise in a steady bifurcation. While at first sight it may seem
117: that an analysis of steady-state modes would not be applicable to
118: two-frequency forced Faraday waves, for which most of the patterns of
119: interest in the present context have been observed, it should be noted
120: that very close to onset the amplitude equations for these waves can be reduced
121: to the standing-wave subspace, in which the waves satisfy equations
122: that have the same form as those for modes arising from a steady
123: bifurcation (see, for example,~\cite{PoSi02}).
124:
125: The competition between commensurate and incommensurate steady
126: structures has been addressed previously in the context of a
127: spatially-periodic forcing of patterns
128: \cite{LoGo83,Co86,CoHu86,CuEl87,OgZi96}. Using a
129: one-dimensional, {\it external} forcing of two-dimensional
130: patterns in electroconvection of nematic liquid crystals,
131: localized domain walls in the local phase of the patterns were
132: observed if the forcing wavenumber was
133: sufficiently incommensurate with the preferred wavenumber of the
134: pattern \cite{LoGo83}. Theoretically, the domain walls were
135: described using a one-dimensional Ginzburg-Landau equation that
136: included a near-resonant forcing term, which reflects the small
137: mismatch between the forcing wavenumber and the wavenumber of
138: the spontaneously forming pattern \cite{Co86,CoEl87}. The
139: situation we have in mind in the present paper is similar to
140: these studies in so far as the competing modes we investigate
141: also provide a periodic
142: forcing for each other. The case of external forcing is
143: recovered if one of the two modes is much stronger than the
144: other and consequently the feedback from the forced mode on the
145: forcing mode can be ignored. We do not restrict ourselves to
146: this case, however, and thus both modes are active
147: degrees of freedom and the interaction between them is mutual.
148:
149: The analysis of the interaction of two exactly resonating modes near
150: a co-dimension-two point at which both modes bifurcate off the basic,
151: unpatterned state has revealed a wide variety of patterns and
152: dynamical phenomena. Rich behavior has been found in small systems
153: in which the interacting modes are determined by the symmetry and
154: shape of the physical domain (e.g. \cite{SiGo89,MePr86,CrKn90,FeSe89,
155: Um91,MoKn98,HiVe02}). More relevant for
156: our goal are the studies of mode-interaction in the presence of
157: translation symmetry, since they allow the extension to systems with
158: large aspect ratio, which are required to address the difference
159: between commensurate and incommensurate structures. Therefore our
160: investigation builds on a comprehensive analysis, performed by Dangelmayr \cite{Da86},
161: of the interaction between two resonant spatial modes in $O(2)$-symmetric systems
162: with wavenumbers in the ratio $m:n$, $m<n$ (Here the $O(2)$-symmetry is a
163: consequence of restricting to spatially periodic patterns in a translation-invariant
164: system.). For $m>1$ there are two primary bifurcations off the
165: trivial state to pure modes followed by secondary bifurcations to
166: mixed modes. These mixed modes can in turn undergo a Hopf bifurcation
167: to generate standing wave solutions and a parity-breaking bifurcation
168: that produces traveling waves. For $m=1$ similar results are found
169: except that there is only one pure-mode state, the one with the higher
170: wavenumber; the other primary bifurcation leads directly to a branch
171: of mixed modes. Further analyses of this system have revealed
172: structurally and asymptotically stable heteroclinic cycles near the
173: mode interaction point when $m:n=1:2$
174: \cite{JoPr87,PrJo88,ArGu88}. More recently, complex dynamics organized
175: around a sequence of transitions between distinct heteroclinic cycles
176: has been discovered in resonances of the form $1:n$
177: \cite{PoKn00,PoKn01}, in particular, in the cases $n=2$ and $n=3$.
178:
179: Although previous work on resonant mode interactions considered
180: only strictly periodic solutions it provided insight into
181: various phenomena that were observed experimentally in
182: large-aspect ratio systems involving large-scale modulations of
183: the patterns. For example, in steady Taylor vortex flow it was
184: found experimentally that not too far from threshold the band
185: of experimentally accessible wavenumbers is
186: substantially reduced compared to the stability limit obtained
187: by the standard analysis of side-band instabilities in the
188: weakly nonlinear regime \cite{DoCa86}. The origin of this strong
189: deviation was identified to be a saddle-node bifurcation
190: associated with the $1:2$ mode interaction \cite{RiPa86}. In
191: directional solidification \cite{SiBe88} localized drift waves
192: have been observed, which arise from the parity-breaking bifurcation
193: \cite{MaTr84,CoGo89,GoGu91} that is associated with the resonant mode interaction
194: with wavenumber ratio $1:2$ \cite{LeRa91}. Subsequently such
195: waves have also been obtained in a variety of other systems
196: including directional viscous fingering \cite{RaMi90}, Taylor
197: vortex flow \cite{WiMc92,RiPa92}, and premixed flames
198: \cite{BaMa94}. In our treatment of the near-resonant case the
199: mode amplitudes are allowed to vary slowly in space. It
200: therefore naturally incorporates phenomena like the localized
201: drift waves and the modification of side-band instabilities by
202: the resonance.
203:
204: In this paper we study the interaction of two nearly-resonant modes
205: in a spatially extended, driven, dissipative system. Near onset we
206: model the slow dynamics of such systems by two amplitude equations of
207: Ginzburg-Landau type, one for each mode. We focus on the weak
208: resonances ($m+n\ge 5$) in order to avoid some of the specific
209: features of the strong-resonance cases (for example, the
210: structurally stable heteroclinic cycles in the case $m:n = 1:2$). Our
211: primary goal is to investigate the transitions between periodic and
212: quasi-periodic states that take place as the result of side-band
213: instabilities, with an eye on how the detuning from exact spatial
214: resonance influences this process. We find that the detuning can
215: play an important role in the selection of the final wavenumbers of
216: the modes involved. For example, it can favor a periodic to
217: quasi-periodic transition that would otherwise (i.e., in the case of
218: exact resonance) result in a second periodic state. Among the various
219: quasi-periodic states that we find in numerical simulations
220: are several that consist of drifting localized structures with alternating
221: locked (periodic) and unlocked (quasi-periodic) domains.
222:
223: It should be noted that at present there is no rigorous justification
224: for the description of quasi-periodic patterns using low-order
225: amplitude expansions. In fact, the lack of straightforward
226: convergence of such an expansion has been investigated recently for
227: two-dimensional quasi-patterns \cite{RuRu03}. We will not discuss
228: these issues; instead we use the coupled Ginzburg-Landau equations as
229: model equations that are known to be the appropriate equations for
230: periodic patterns and at the same time also allow quasi-periodic
231: solutions.
232:
233: The organization of the paper is as follows. In Section 2 we set
234: up the coupled Ginzburg-Landau equations that are based on the
235: truncated normal form equations for the $m:n$-resonance. In
236: Section 3 we utilize and build upon on the detailed results of
237: Dangelmayr \cite{Da86} to describe the stability properties of
238: steady spatially periodic states (i.e, pure and mixed modes)
239: with respect to perturbations that preserve the periodicity of
240: the pattern. In Section 4 we turn to the question of stability
241: of the steady periodic solutions with respect to side-band
242: instabilities. Here we also determine how these instabilities
243: are affected by the detuning from perfect resonance. Numerical
244: simulations of the system are carried out in Section 5 to
245: investigate the nonlinear evolution of the system subsequent to
246: the side-band instability discussed in the previous Section.
247: Our concluding remarks are given in section 6. It
248: should be pointed out that a very recent preprint by Dawes {\em
249: et al.} presents a complementary investigation near the $1:2$
250: resonance \cite{DaPo03}.
251:
252:
253:
254:
255: \section{The amplitude equations}\cierto{sec:amplitudequations}
256:
257: We consider driven, dissipative systems in one
258: spatial dimension that are
259: invariant under spatial translations
260: and reflections. We further assume
261: that in the system of interest there are two distinct spatial
262: modes that destabilize the basic homogeneous state
263: nearly simultaneously. The wavenumbers of these two modes,
264: $q_1$ and $q_2$, correspond to minima of the neutral curves and
265: are assumed to be in approximate spatial resonance, i.e.,
266: \begin{equation}
267: n(q_1+\varepsilon\hat\gamma)= m q_2, \qquad |\varepsilon| \ll 1,
268: \cierto{1b}
269: \end{equation}
270: where $m$ and $n > m$ are positive
271: co-prime integers and the term $\varepsilon\hat \gamma$
272: measures the deviation from perfect resonance. Physical
273: fields, near onset, may then be expanded in terms of these two
274: spatial modes:
275: \begin{eqnarray}
276: u(x,t)=\varepsilon[A_1(X,T)e^{i(q_1+\varepsilon\hat{\gamma} )x}+
277: A_2(X,T)e^{iq_2x}+\cc]+\cdots \cierto{1a}
278: \end{eqnarray}
279: The two modes are allowed to vary on slow spatial and temporal
280: scales $X=\varepsilon x$ and $T=\varepsilon^2t$, respectively. Note
281: that in the expansion~(\ref{1a}) we do not expand about about the
282: minima $q_{1,2}$ of the neutral stability curves, which for
283: $\hat\gamma \neq 0$ are not in spatial resonance, but take
284: instead a mode $A_1$ which {\it is} in exact spatial resonance
285: with $A_2$. This choice of $A_1$ and $A_2$ simplifies the
286: equations, avoiding any explicit dependence on the spatial
287: variable in the resulting amplitude equations. Furthermore, we
288: allow for a small offset between the critical values of the forcing
289: amplitudes $F_{ic}$ of the two modes (see Fig.~\ref{fig:critic-modes}).
290:
291: %%%%%%% FIG00%%%%%%%%%%%%%
292: \begin{figure}[htb]
293: \centerline{\includegraphics[width=7cm]{figuras/critic_modes.ps}}
294: \caption{Sketch of the neutral stability curve for the two modes.}
295: \cierto{fig:critic-modes}
296: \end{figure}
297: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
298:
299: The equations
300: governing the evolution of $A_1$ and $A_2$ must be equivariant under the
301: symmetry operations generated by spatial translations ($T_\varphi$) and
302: spatial reflections ($R$), which act on $(A_1,A_2)$ as follows:
303: \begin{eqnarray}
304: T_\varphi :(A_1, A_2) \rightarrow &(e^{im\varphi} A_1, e^{in\varphi} A_2),
305: \quad \text{for} \quad \varphi \in \text{$[$}0, 2\pi),\\ \nonumber
306: R: (A_1, A_2) \rightarrow & (\bar A_1,\bar A_2),
307: \cierto{symmetries}
308: \end{eqnarray}
309: where the bar denotes the complex conjugate. Consistent with this
310: equivariance requirement the slow evolution of $A_1$ and $A_2$ can
311: be approximated, after rescaling, by the Ginzburg-Landau equations
312:
313: \begin{eqnarray}
314: A_{1 T}&=&\mu A_1+\delta A_{1XX}-i\gamma A_{1X}-(s|A_1|^2+\rho|A_2|^2)A_1
315: +\nu \bar A_1^{n-1}A_2^m, \cierto{2}\\
316: A_{2 T}&=&(\mu+\Delta\mu) A_2+\delta' A_{2XX}-(s'|A_2|^2+\rho'|A_1|^2)A_2
317: +\nu' \bar A_2^{m-1}A_1^n.\cierto{3}
318: \end{eqnarray}
319: The subscripts indicate partial derivatives with respect to $X$
320: and $T$. The main control parameter $\mu\propto
321: (F-F_{1c})/\varepsilon^2$ measures the magnitude of the overall
322: forcing. In addition, we keep track of the offset in the two
323: critical forcing amplitudes with $\Delta \mu
324: \equiv(F_{1c}-F_{2c})/\varepsilon^2$ (see
325: Fig.~\ref{fig:critic-modes}), and capture the detuning between $q_1$
326: and $m q_2/n$ with $\gamma\equiv 2\delta\hat \gamma$. The local
327: curvature of the neutral stability curves near $q_1$ and $q_2$ is measured by $\delta$
328: and $\delta'$, respectively. We further assume that the nonlinear
329: self- and cross- interaction coefficients satisfy the non-degeneracy
330: conditions $ss' \neq 0$ and $ss'-\rho\rho'\neq 0$ and perform a simple
331: rescaling such that $s=\pm 1$, $s'=\pm 1$.
332:
333: One goal of this paper is to gain insight into the difference
334: between periodic and quasi-periodic patterns in systems with two
335: unstable wavenumbers. If the two wavenumbers are not rationally
336: related and their irrational ratio is kept fixed as the onset
337: for the two modes is approached, then only terms of the form
338: $A_i|A_j|^{2l}$, $l=1,2,3\ldots$, appear in the equation for
339: $A_i$. For a rational ratio, however, additional nonlinear terms
340: arise, which couple the otherwise uncoupled phases of the two
341: modes $A_i$. For the $m:n$-resonance the leading-order resonance
342: terms are given by $\bar{A}_1^{n-1}A_2^m$ and
343: $\bar{A}_1^{m-1}A_1^n$ in the equations for $A_1$ and $A_2$,
344: respectively. In order to explore the connection between the
345: rational and the irrational case we consider a wavenumber ratio
346: which may be irrational, but its deviation from the ratio $m:n$
347: is of ${\mathcal O}(\varepsilon)$. This allows the
348: mismatch between the two wavenumbers to be captured by the slow
349: spatial variable $X$ and to describe periodic and quasi-periodic
350: patterns with the same set of equations (\ref{2},\ref{3}).
351: Equivalently, we could have expanded in the irrationally related
352: wavenumbers $q_1$ and $q_2$ associated with the minima of the neutral
353: curves in Fig.~\ref{fig:critic-modes}. Then the resonance terms would
354: introduce space-periodic coefficients with a period that is related to
355: the mismatch of the wavenumbers. Our choice of the expansion wavenumbers
356: (cf. (\ref{1b})) removes this space-dependence and introduces the first-order
357: derivative $-i\gamma \partial_X A_1$ in its place.
358:
359:
360: We focus on the weak resonances, $m+n\geq 5$, in which the
361: resonant terms are of higher order. We neglect, however, non-resonant terms
362: of the form $A_{i} |A_{j}|^p$ ($i, j=1,\, 2$ and $4\leq p\in\Bbb N$) which may
363: arise at lower order. This is motivated by the observation that such terms
364: do not contribute any qualitatively new effects for small
365: amplitudes. The resonant terms, in contrast, remove the unphysical
366: degeneracy that arises when the phases are left uncoupled and can
367: therefore influence dynamics in a significant way despite appearing
368: at higher order. Note, however, that near onset the resonant terms
369: are typically small and the phase coupling between $A_1$ and $A_2$
370: occurs on a very slow time scale. The coupling becomes stronger
371: further above onset where the weakly nonlinear analysis may no longer
372: be valid.
373:
374: It is often useful to recast Eqs.~(\ref{2},\ref{3}) in terms of
375: real amplitudes $R_j \ge 0$ and phases $\phi_j$ by writing $A_j=R_je^{i\phi_j}$.
376: This leads to
377: \begin{eqnarray}
378: R_{1T} &=&\mu R_1-(sR_1^2+\rho R_2^2)R_1+\nu R_1^{n-1}R_2^m
379: \cos(n\phi_1-m\phi_2) \nonumber\\
380: &&+\delta R_{1XX}-\delta\phi_{1X}^2R_1
381: -2\gamma\phi_{1X}R_1,\cierto{6a}\\
382: R_{2T} &=&(\mu+\Delta\mu) R_2-(s'R_2^2+\rho' R_1^2)R_2+\nu' R_2^{m-1}R_1^n
383: \cos(n\phi_1-m\phi_2)\nonumber\\
384: &&+\delta' R_{2XX}-\delta'\phi_{2X}^2R_2 ,\cierto{7a}\\
385: R_1\phi_{1T} &=&-\nu R_1^{n-1}R_2^{m} \sin(n\phi_1-m\phi_2)+
386: \delta\phi_{1XX}R_1+ 2\delta\phi_{1X}R_{1X}+2\gamma R_{1X},\cierto{8aa}\\
387: R_2\phi_{2T} &=&\nu' R_2^{m-1}R_1^{n} \sin(n\phi_1-m\phi_2)+
388: \delta'\phi_{2XX}R_2+ 2\delta'\phi_{2X}R_{2X}.\cierto{9a}
389: \end{eqnarray}
390: If the spatial dependence in Eqs.~(\ref{6a}-\ref{9a}) is ignored, the system reduces
391: to a set of ordinary differential equations (ODEs), equivalent to the one analyzed
392: by Dangelmayr \cite{Da86}. In this simplified problem, the translational symmetry
393: ($T_\varphi$) causes the overall phase to decouple and leaves only the three real
394: variables $R_1$, $R_2$, and the mixed phase \begin{equation}
395: \phi=n\phi_1-m\phi_2, \cierto{9b}
396: \end{equation}
397: with dynamically important roles. Dangelmayr's bifurcation analysis
398: produced expressions for the location of primary bifurcations to pure mode solutions,
399: secondary bifurcations to mixed mode solutions, and, in some instances, tertiary
400: bifurcations to standing-wave and traveling-wave solutions. These results apply
401: to a general $m:n$ resonance, and prove useful in what follows.
402:
403:
404: \section{Steady Spatially-Periodic Solutions}
405: \cierto{sec:basicsolutions}
406:
407: In this section we analyze steady solutions of
408: Eqs.~(\ref{2},\ref{3}) of the form
409: \begin{equation}
410: A_1=R_1e^{i(kX+\hat\phi_1)},\quad A_2=R_2e^{i((nk/m)X+\hat\phi_2)}, \cierto{4}
411: \end{equation}
412: where $R_{1,2}\geq 0$, and $\hat\phi_{1,2}$ and $k$ are real.
413: Such states represent spatially periodic solutions of the
414: original problem with wavenumbers $\tilde
415: q_1=q_1+\varepsilon\gamma +\varepsilon k$ and $\tilde
416: q_2=q_2+\varepsilon n k/m$ so that $\tilde{q}_1n=\tilde{q}_2m$.
417: These solutions break the continuous translational
418: symmetry ($T_\varphi$) but remain invariant under discrete
419: translations.
420:
421: Within this family of steady states there are generically only
422: two types of nontrivial solutions, pure modes and mixed modes,
423: which we describe below.
424:
425: \noindent
426: {\bf I. Pure modes} ($S_{1,2}$). These are single-mode states, which take one
427: of two forms:
428: \begin{align}
429: S_1 &: (A_1, A_2) =(\sqrt{\alpha/s}\,\,e^{i (kX+\hat\phi_1)}, 0),
430: \quad\text{for $m>1$},\nonumber\\
431: S_2 &: (A_1, A_2) =(0, \sqrt{\beta/s'}\,\,e^{i ((nk/m)X+\hat\phi_2)}),
432: \cierto{5}\end{align}
433: where $\hat\phi_1, \hat\phi_2\in [0, 2\pi)$ and
434: \begin{align}
435: &\alpha=\mu-\delta k^2+\gamma k,\nonumber\\
436: &\beta=\mu+\Delta\mu-\delta'(n k/m)^2.\cierto{5b}
437: \end{align}
438: Note that pure modes of type $S_1$ are not present if $m=1$ (see
439: Eqs.~(\ref{2},\ref{3})). Moreover, the pure modes $S_1$ and
440: $S_2$ are not isolated, but emerge as circles of equivalent
441: solutions (parametrized by $\hat\phi_1$ or $\hat\phi_2$).
442: Hereafter we consider resonances $m:n$ where $m\geq 2$, in which
443: case both pure modes $S_1$ and $S_2$ are present.
444:
445:
446: \noindent
447: {\bf II. Mixed modes} ($S_\pm$). There are two types of mixed modes,
448:
449: \begin{equation}
450: S_\pm: (A_1, A_2) =(R_1\,\,e^{i (kX+\hat\phi_1)},
451: R_2\,\,e^{i ((nk/m)X+\hat\phi_2)}),\cierto{mixmod}
452: \end{equation}
453: satisfying
454: \begin{align}
455: S_{\pm} :\quad &\cos(\phi)=\pm 1, \nonumber\\
456: & \begin{array}{ll}
457: &(s'\alpha-\rho\beta)=(ss'-\rho\rho')R_1^2\pm (s'
458: \nu R_2^2-\nu'\rho R_1^2)R_2^{m-2}R_1^{n-2},\\
459: &(s\beta-\rho'\alpha)=(ss'-\rho\rho')R_2^2\pm (s
460: \nu' R_1^2-\nu \rho' R_2^2)R_2^{m-2}R_1^{n-2}.\end{array}\cierto{6}
461: \end{align}
462: \noindent
463: Here $\phi$ is the mixed phase given by Eq.~(\ref{9b}), and $\alpha$ and
464: $\beta$ are defined by Eq.~(\ref{5b}). As in the case of the pure modes
465: $S_{1,2}$, translational symmetry implies that there are circles of equivalent
466: mixed-modes states (parametrized by $\hat\phi_1$, say). Like the
467: pure modes, the mixed modes are invariant under reflections $(R)$ through an appropriate origin.
468:
469: \subsection{Stability under homogeneous perturbations}
470:
471: The stability of $S_{1,2}$ and $S_{\pm}$ under homogeneous
472: perturbations can be obtained from Dangelmayr's analysis
473: \cite{Da86}, which we review here in some detail to provide the
474: background necessary for our analysis. The stability regions in
475: the $(\alpha, \beta)$-unfolding plane simply need to be mapped
476: to the $(k,\mu)$-plane with the (nonlinear) transformation
477: (\ref{5b}). Each intersection of the curves $\alpha=0$ and
478: $\beta=0$ corresponds to the codimension-two point of
479: \cite{Da86}; there are generically zero or two such
480: intersections. Since the nonlinear coefficients are identical in
481: the vicinity of both intersections, the response of the system
482: to homogeneous perturbations in corresponding neighborhoods of
483: the (two) intersections is identical. In particular, any
484: bifurcation set arising from one intersection arises from the
485: other as well. Several examples are given in
486: Fig.~\ref{fig:dgrgamma05}, which shows the various bifurcation
487: sets in the $(k, \mu)$-plane in four representative cases.
488: These bifurcations are described below. Note that for mode
489: $A_1$ the deviation of the wavenumber from the critical
490: wavenumber is given by $\varepsilon k$, whereas for mode $A_2$
491: it is given by $\varepsilon nk/m$. Additional details, valid in
492: sufficiently small neighborhoods of the intersections, are
493: available in \cite{Da86}.
494:
495:
496:
497: %%%%%%% FIG1%%%%%%%%%%%%%
498: \begin{figure}
499: \centerline{\includegraphics[width=6cm]{figuras/Fig2a.ps}
500: \quad \includegraphics [width=6cm]{figuras/Fig2b.ps}}
501: \centerline{\includegraphics [width=6cm]{figuras/Fig2c.ps}
502: \quad \includegraphics[width=6cm] {figuras/Fig2d.ps}}
503: \caption{Bifurcation sets indicating the creation of pure modes (thick solid
504: lines), mixed modes (thin solid lines $L_1^\pm$, $L_1$ and $L_2$), standing
505: waves (dashed lines $SW^\pm$) and traveling waves (dash-dotted lines
506: $TW^\pm$), as well as saddle-node bifurcations (dashed lines $SN$). The specific
507: resonance $m:n$ is indicated above the plots. In each case $\Delta\mu=0.5$,
508: $\gamma=0.5$, $\delta=\delta'=1$, $\rho=0.4$, and $\rho'=0.67$. In (a)
509: $s=-s'=1$,
510: $\nu=0.62$ and $\nu'=1.02$, but in (b), (c) and (d) $s=s'=-1$, $\nu= 0.62$ and
511: $\nu'=-1.02$.}
512: \cierto{fig:dgrgamma05}
513: \end{figure}
514: %%%%%%%%%%%%%%%%%%%%%%%
515:
516: In this paper we study (\ref{2},\ref{3}) with the coefficients $\nu$
517: and $\nu'$ of the higher-order resonance terms taken
518: to be of order 1. The resulting stability and bifurcation diagrams
519: therefore contain certain features that do not remain local to the
520: bifurcation when $\nu,\nu' \rightarrow 0$, i.e. in this limit these
521: features disappear at infinity, $\mu \rightarrow \infty$, and do not
522: represent robust aspects of the mode-interaction problem. We
523: return to this issue briefly at the end of this section, and indicate
524: which features of our sample bifurcation sets are not robust in the
525: limit $\nu,\nu'\rightarrow 0$.
526:
527: The pure modes $S_1$ and $S_2$ bifurcate from the trivial state when
528: $\alpha=0$ and $\beta=0$, respectively: the bifurcation to $S_1$
529: ($S_2$) is supercritical if $s=1$ ($s'=1$).
530: Their stability is determined by four eigenvalues, one of which is forced to be zero by
531: translation symmetry. In the case of $S_1$ the remaining three eigenvalues are
532: \begin{eqnarray}
533: &&\,\,\,\lambda_0^{(1)}=-2\alpha/s,\nonumber\\
534: &L_1:&\,\,\lambda_\pm^{(1)}=(\beta-\rho'\alpha)/s\pm\left\{
535: \nu'\left |\alpha/s\right|^{n/2}\right\}_{m=2}, \cierto{7}
536: \end{eqnarray}
537: where the bracketed term with subscript $m=2$ is present only if $m=2$.
538: For $S_2$ the eigenvalues are given by
539: \begin{eqnarray}
540: &&\lambda_0^{(2)}=\beta/s',\nonumber\\
541: &L_2:&\,\,\lambda_+^{(2)}=\lambda_-^{(2)}=(s'\alpha-\rho\beta)/s'.
542: \cierto{8}\end{eqnarray}
543:
544:
545: When one of the eigenvalues $\lambda_\pm^{(1)}$
546: ($\lambda_\pm^{(2)}$) changes sign the pure modes $S_1$ ($S_2$)
547: become unstable to mixed modes, respectively.
548: For $S_2$ the two eigenvalues coincide and the bifurcation occurs
549: along a single line denoted by $L_2$ in Fig.~\ref{fig:dgrgamma05}.
550: Similarly, for $m>2$ the transition from $S_1$ to the mixed modes $S_\pm$ occurs along
551: the single curve $L_1$ (Fig.~\ref{fig:dgrgamma05}a,b). For $m=2$
552: the eigenvalues $\lambda_\pm^{(1)}$ are not degenerate and the
553: line $L_1$ splits into two curves, $L_1^+$ and $L_1^-$; the
554: mixed mode $S_+$ bifurcates at $L_1^+$ and $S_-$ bifurcates at
555: $L_1^-$ ($L_1^\pm$ in
556: Figs.~\ref{fig:dgrgamma05}c-d). The splitting is due to the
557: presence of the resonant terms which are linear in $A_2$ if
558: $m=2$ but not otherwise. Both curves, $L_1^+$ and $L_1^-$,
559: become tangent to each other at the intersection
560: $\alpha=\beta=0$.
561:
562: The response of the mixed modes to amplitude perturbations is
563: decoupled (due to reflection symmetry) from the effect of phase
564: perturbations. The amplitude stability is determined by the
565: eigenvalues of a $2 \times 2$-matrix $M_{\pm}$, whose determinant
566: and trace can be written as (recall we consider $m>1$)
567: \begin{eqnarray}
568: \text{det}(M_\pm)
569: &=&-4(\rho\rho'-ss')R_1^2R_2^2\pm R_1^{n-2}R_2^{m-2}
570: H(R_1,R_2), \cierto{10}\\
571: \text{Tr}(M_\pm) & = &-2
572: s(R_1^2+ss'R_2^2)\pm R_1^{n-2}
573: R_2^{m-2}(\nu(n-2)R_2^2+\nu'(m-2)R_1^2).\cierto{11}
574: \end{eqnarray}
575: with
576: \begin{equation}
577: H(R_1,R_2)=-2\left[\nu's (m-2) R_1^4+
578: \nu s' (n-2)R_2^4-R_1^2R_2^2(\rho'\nu m+\rho\nu'n)\right]
579: \end{equation}
580: Here $R_1$ and $R_2$ are solutions of Eqs.~(\ref{6}).
581: Since in general $\nu$ and $\nu'$ are of ${\mathcal
582: O}(\varepsilon^{m+n-4})$, the contribution $H(R_1,R_2)$ from the
583: resonance term affects the steady bifurcation determined by
584: (\ref{10}) only for $m\leq 3$.
585: In particular, in the cases $m:n=3:n$ and $m:n=2:3$ the function $H$ can balance the
586: first term along curves through the codimension-2 point along which
587: $R_2 \ll R_1$ and $R_1 \ll R_2$, respectively, and one of the mixed states experiences
588: a saddle-node bifurcation (curves labeled $SN$ in Figs.~\ref{fig:dgrgamma05}b,d)
589: \cite{Da86}. For $2:n$ resonances with $n\ge 5$ the
590: term $R_1^4$ drops out in $H(R_1,R_2)$. Consequently, the $H$-term cannot balance
591: the first term in (\ref{10}) and no saddle-node bifurcations occur.
592:
593: For $m \ge 4$ the sign of det($M_\pm$)
594: does not depend on $R_{1,2}$. Then the $S_\pm$ solutions are always
595: unstable for $(\rho\rho'-ss')>0$, while for $(\rho\rho'-ss')<0$ their
596: stability must be deduced from the sign of Tr($M_\pm$). If
597: $ss'=1$ it follows from Eq.~(\ref {11}) that for $\varepsilon
598: \ll 1$ sign(Tr($M_\pm))=-s$; the mixed modes $S_\pm$ are then stable to
599: amplitude perturbations if $s=+1$. On the other hand, if
600: $ss'=-1$, the trace of $M_\pm$ can change sign, indicating the
601: possibility of a Hopf bifurcation. The resulting time-periodic
602: solutions inherit the reflection symmetry of $S_\pm$ (so
603: $\dot\phi_1=\dot\phi_2=0$) and therefore correspond to standing
604: waves ($SW$ curves in Fig.~\ref{fig:dgrgamma05}a).
605:
606:
607: Instabilities associated with perturbations of the mixed phase
608: (\ref{9b}) lead to bifurcations breaking the reflection
609: symmetry. The relevant eigenvalues are given by
610: \begin{equation}
611: TW:\,\, e_\pm=\mp(\nu n R_2^2+\nu' mR_1^2) R_1^{n-2}
612: R_2^{m-2}, \cierto{12}
613: \end{equation}
614: and may pass through zero only if sign$(\nu\nu')=-1$. In this
615: case the $S_\pm$ states undergo a pitchfork bifurcation,
616: reflection symmetry is broken, and traveling waves appear ($TW$
617: curves in Figs.~\ref{fig:dgrgamma05}b-d). These
618: traveling-wave solutions manifest themselves as
619: fixed points of the three-dimensional
620: ODE system involving $R_1$, $R_2$ and $\phi$, but are seen to be
621: traveling waves by the fact that the individual phase velocities
622: are nonzero: $\dot\phi_1/\dot\phi_2=n/m$. Since the phase
623: velocity of these waves goes to 0 at the bifurcation, they are
624: often called drift waves. We do not consider
625: the stability properties of the
626: traveling-wave solutions in this paper.
627:
628: The stability results for $S_{1,2}$ and $S_{\pm}$ described
629: above are illustrated in
630: Figs.~\ref{fig:dgrm2n5-gam0deltamu0}-\ref{fig:dgrm2n5Delta-s-}
631: for $m:n=2:5$ and the indicated parameter values.
632: They all satisfy
633: \begin{equation} s=s'=1,\qquad
634: ss'-\rho\rho'>0,\qquad \nu\nu'<0,
635: \cierto{paramset}
636: \end{equation}
637: so that the pure modes $S_{1,2}$ bifurcate supercritically in
638: all cases. Next to these plots we sketch the type of bifurcation
639: diagram one obtains when increasing $\mu$ at constant $k$
640: along the thin dashed vertical lines. Since in
641: all cases $ss'-\rho\rho'>0$, both mixed modes $S_{\pm}$ are
642: stable to amplitude perturbations (see the explanation following
643: Eqs.~(\ref{10},\ref{11})). The stability of $S_\pm$ with regard
644: to phase perturbations depends,
645: however, on the eigenvalue $e_\pm$ given by (\ref{12}). Because
646: $\nu\nu'<0$, this eigenvalue may change sign, causing the mixed
647: modes $S_\pm$ to undergo a symmetry-breaking bifurcation to
648: traveling waves.
649: Figs.~\ref{fig:dgrm2n5-gam0deltamu0}-\ref{fig:dgrm2n5Delta-s-}
650: present six different cases characterized by the following
651: quantities:
652: \begin{equation}
653: {\chi}=\gamma^2-4\Delta\mu\left(\delta-(\frac{n}{m})^2\delta'\right) \quad
654: \text{and} \quad {\Lambda}= \frac{\gamma^2}{4 \delta}-\Delta\mu.
655: \cierto{deltas}
656: \end{equation}
657: The parameter $\chi$ controls the intersection of the parabolas
658: $\alpha=0$ and $\beta=0$; they intersect if $\chi \geq 0$ and
659: not otherwise. The quantity $\Lambda$ determines the relative
660: position ($\mu$-value) of the minima of the curves $\alpha=0$
661: and $\beta=0$. It thus indicates which of the two modes, $S_1$
662: or $S_2$, is excited first. $S_2$ appears first when
663: $\Lambda>0$, while $S_1$ takes priority for $\Lambda<0$; if
664: $\Lambda=0$ both pure modes onset simultaneously.
665:
666:
667: %%%%%%%%%%%% FIG%%%%%%%%%%%%%%%%%%%%%%%
668: \begin{figure}
669: {\includegraphics[width=14cm]{figuras/figuras4/Fig4e.ps}}
670: {\includegraphics[width=14cm]{figuras/figuras4/Fig4f.ps}}
671: \caption{Stability regions of the pure modes and mixed modes for
672: the resonance $2:5$ for $\Delta\mu=0$ and
673: $\gamma=0$. Figures on the right are
674: sketches of bifurcation diagrams associated with the vertical paths
675: (dashed lines) in the stability regions on the left. Solid lines correspond
676: to stable states and dashed lines to unstable states. $T$ stands for trivial
677: state. (a) $\delta=\delta'=1$, $s=s'=1$, $\rho=0.4$, $\rho'=0.67$
678: $\nu=0.62$ and $\nu'=-1.02$. (b) $\delta= 5$, $\delta'=0.5$, $s=s'=1$, $\rho=1.5$,
679: $\rho'=0.5$ and $\nu=-\nu'=0.05$. In (a) the traveling-wave branch
680: that arises at $TW^+$ is not shown since we do not consider stability properties
681: of traveling solutions.} \cierto{fig:dgrm2n5-gam0deltamu0}
682: \end{figure}
683: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
684:
685: The degenerate case $\chi=\Lambda=0$, i.e $\gamma=\Delta\mu=0$,
686: is illustrated in Fig.~\ref{fig:dgrm2n5-gam0deltamu0}. In
687: this case the curves $\alpha=0$ and $\beta=0$ intersect only once
688: and their minima coincide. While in
689: Fig.\ref{fig:dgrm2n5-gam0deltamu0}a the neutral curve for mode
690: $A_1$ is wider than that for $A_2$, it is the other way around
691: in Fig.\ref{fig:dgrm2n5-gam0deltamu0}b. Note that the wavenumber
692: $nk/m$ of $A_2$ is larger than that of $A_1$; therefore to make
693: the neutral curve of $A_2$ wider than that of $A_1$ requires a
694: large ratio of $\delta/\delta'$. Depending on the nonlinear
695: coefficients, all four branches of pure modes and mixed modes
696: ($S_1$, $S_2$, $S_\pm$), or just the pure modes ($S_1$,$S_2$)
697: may arise at the intersection point of the neutral
698: curves. Because we are considering $s=s'=1$, $S_1$ and $S_2$
699: bifurcate supercritically.
700: For $k=0$ the eigenvalues $\lambda_\pm^{(1,2)}$ (cf.
701: Eqs.~(\ref{7},\ref{8})) determining the stability of $S_1$ and
702: $S_2$ take the simpler form:
703: \begin{eqnarray}
704: \mbox{L}_1:
705: \lambda_\pm^{(1)}&=&\mu(1-\rho')\pm\left\{\nu'\left|\mu\right|^{n/2}\right\}_{m=2}
706: \quad \text{ for $S_1$,}\cierto{eigPM}\\
707: \mbox{L}_2: \lambda_\pm^{(2)}&=& \mu(1-\rho)\qquad \text{ for
708: $S_2$.}\nonumber
709: \end{eqnarray}
710:
711: Near onset (i.e., $0<\mu \ll 1$) $S_1$ and $S_2$ are stable if $1-\rho'<0$ and
712: $1-\rho<0$, respectively. In the case $m=2$, the stability of $S_1$ can be modified
713: at larger values of $\mu$ by the resonant terms. Near onset the
714: amplitudes of the mixed modes are approximated by
715: \begin{align}
716: S_{\pm}: \quad &
717: R_1^2=\frac{1}{1-\rho\rho'}\left(\mu(1-\rho)\mp(\nu-\nu'\rho)o(\mu^2)\right)
718: \cierto{aprxMM}\\
719: &R_2^2=\frac{1}{1-\rho\rho'}\left(\mu(1-\rho')\mp(
720: \nu-\nu'\rho)o(\mu^2)\right),\nonumber
721: \end{align}
722: Note that
723: small-amplitude mixed modes $S_\pm$ can
724: therefore exist only if
725: $(1-\rho)(1-\rho')>0$. As discussed
726: above, when $\rho\rho'<1$, the stability of $S_\pm$ is controlled
727: by the phase eigenvalue $e_\pm$ of Eq.~(\ref{12}). Upon substituting
728: Eqs.~(\ref{aprxMM}) into Eq.~(\ref{12}) we find that
729: \begin{equation}
730: \text{sign}(e_\pm)=\text{sign}\left\{\mp\frac{\mu}{1-\rho\rho'}\left(n \nu
731: (1-\rho')+ m\nu'(1-\rho)\right)\right\}.\cierto{eigMM}
732: \end{equation}
733:
734: The results for $\Delta\mu=\gamma=0$ and $s=s'=1$ may now be summarized:
735: \begin{enumerate}
736: \item If $1-\rho>0$ and $1-\rho'>0$ both pure and mixed
737: modes bifurcate from the trivial state as the forcing $\mu$ is increased. The two pure
738: modes are both unstable, while one of the mixed modes, either $S_+$ or $S_-$, is stable.
739: This case is illustrated in the diagram of Fig.~\ref{fig:dgrm2n5-gam0deltamu0}a; in
740: this example sign$(e_\pm)=\mp 1$ (see Eq.~(\ref{eigMM})), implying that
741: $S_+$ is stable and $S_-$ unstable at onset.
742:
743: \item If $1-\rho<0$ or $1-\rho'<0$ only the
744: two pure modes are present at onset.When
745: $1-\rho\rho'>0$ $S_1$ is stable if $1-\rho' < 0$ and $S_2$ is
746: stable if $1-\rho < 0$. In particular, a bistable situation is
747: permitted. In the example shown in
748: Fig.~\ref{fig:dgrm2n5-gam0deltamu0} only $S_2$ is stable because $1-\rho'>0$.
749: \end{enumerate}
750:
751: Fig.~\ref{fig:dgrm2n5Delta-s+} presents two possible unfoldings
752: of the degenerate diagrams shown in
753: Fig.~\ref{fig:dgrm2n5-gam0deltamu0}. In both cases the curves
754: $\alpha=0$ and $\beta=0$ intersect twice because $\chi>0$. In
755: Fig.~\ref{fig:dgrm2n5Delta-s+}a the pure mode $S_2$ appears
756: first ($\Lambda>0$) and is stable. With increasing $\mu$,
757: however, perturbations in the direction of the other mode,
758: $A_1$, become increasingly important, eventually destabilizing
759: $S_2$ at $L_2$ in favor of the mixed modes $S_\pm$. Since $R_1
760: \ll R_2$ in the vicinity of $L_2$ and $\nu>0$, $S_+$ is the
761: stable mixed mode (cf.~Eq.~(\ref{12})). In
762: Fig.~\ref{fig:dgrm2n5Delta-s+}b we have $\Lambda<0$, and the
763: roles of $S_1$ and $S_2$ are switched. The mode $S_1$, which is
764: now stable at onset, is destabilized by a perturbation in the
765: direction of $A_2$ at $L_1^-$, generating the mixed state $S_-$.
766: It undergoes a second bifurcation involving a perturbation in
767: the direction of $A_2$ at $L_1^+$, leading to the mixed mode
768: $S_+$. In this case $S_-$ is stable, not
769: $S_+$. In contrast with Fig.~\ref{fig:dgrm2n5Delta-s+}a,
770: however, the pure mode $S_2$ is ultimately stabilized at large
771: $\mu$ since $\rho>s'=1$. Therefore the cross-interaction
772: term
773: $\rho|A_2|^2\equiv\rho(\mu+\Delta \mu)/s'$ in
774: Eqs.(\ref{2},\ref{3}),
775: dominates the linear growth rate $\mu$ of $A_1$
776: for large $\mu$ and suppresses the perturbations in the
777: direction of $A_1$.
778:
779:
780:
781: %%%%%%%%%%%%%FIG2%%%%%%%%%%%%%
782: \begin{figure}[htb]
783: \hskip0.2truecm{\includegraphics [width=14cm]{figuras/figuras4/Fig4a.ps}}
784: {\includegraphics[width=14.45cm]{figuras/figuras4/Fig4b.ps}}
785: \caption{Stability regions of the pure modes and mixed modes for the
786: resonance $2:5$. Figures on the right are bifurcation diagrams
787: associated with the vertical paths (dashed lines) in the
788: stability diagrams on the left. $T$ stands for trivial state
789: $A_1=A_2=0$. (a) Same parameters as in
790: Fig.~\ref{fig:dgrm2n5-gam0deltamu0}a but $\Delta\mu=0.366$ and
791: $\gamma=0.7$. (b) Same parameters as in
792: Fig.~\ref{fig:dgrm2n5-gam0deltamu0}b but $\Delta\mu=-0.5$ and
793: $\gamma=0.5$. The branch of traveling waves that arises from
794: $TW^-$ in (a), or that (possibly) connects the
795: $S_+$ with the $S_-$ solutions in (b) is not shown.}
796: \cierto{fig:dgrm2n5Delta-s+}
797: \end{figure}
798: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
799:
800: Fig.~\ref{fig:dgrm2n5Delta-s-} shows diagrams for the case
801: where the curves $\alpha=0$ and $\beta=0$ do not
802: cross ($\chi<0$). In Fig.~\ref{fig:dgrm2n5Delta-s-}a the pure
803: mode $S_2$ bifurcates first, while $S_1$ does so in
804: Fig.~\ref{fig:dgrm2n5Delta-s-}b. The subsequent bifurcations
805: that these states undergo are basically of the same type as
806: those in
807: Fig.~\ref{fig:dgrm2n5Delta-s+}. For example, in
808: Fig.~\ref{fig:dgrm2n5Delta-s-}a the mode $S_2$ becomes unstable
809: to the mixed modes at $L_2$, while $S_1$ eventually
810: gains stability at large $\mu$ because $\rho > s'$.
811: A feature of Fig.~\ref{fig:dgrm2n5Delta-s-}b
812: that is not present in the previous diagrams is the saddle-node
813: bifurcation on the $S_+$ branch. The appearance of this
814: bifurcation is not specific to the case $\chi<0$ but
815: depends on the nonlinear coefficients, in
816: particular on the resonance coefficients $\nu$
817: and $\nu'$ (see Eq.~(\ref{6})).
818:
819: A comment regarding the validity of the phase diagrams is in
820: order. For the weak resonances we are considering in this
821: paper the resonant terms proportional to $\nu$ and $\nu'$ are of
822: higher order in the amplitudes. Consequently, they have to be
823: considered as perturbation terms. Since they are responsible for the
824: splitting of the lines $L_1$ and $TW$ into $L_1^\pm$ and $TW^\pm$,
825: only those aspects of the results presented in Figures
826: \ref{fig:dgrm2n5-gam0deltamu0}-\ref{fig:dgrm2n5Delta-s-}
827: that persist as this splitting becomes small are expected to hold
828: systematically. Thus, the transition to traveling waves at $k=0$
829: in Fig.\ref{fig:dgrm2n5Delta-s+}b is robust, while that at $k=0$
830: in Fig.\ref{fig:dgrm2n5Delta-s+}a is shifted to ever larger
831: values of $\mu$ as the resonant terms become weaker. Similarly,
832: the saddle-node bifurcation of $S_+$ in
833: Fig.\ref{fig:dgrm2n5Delta-s-}b is not robust.
834: It disappears through a sequence of
835: bifurcations that involves a merging of the branch $S_+$ with
836: $S_1$. For very small $|\nu|$ and $|\nu'|$ the branch $S_+$
837: merges with $S_1$ close to $L_1^-$.
838:
839:
840: %%%%%%%%%%%%%FIG%%%%%%%%%%%%%%%%%%
841: \begin{figure}[htb]
842: \hskip-0.5truecm{\includegraphics[width=14.5cm]{figuras/figuras4/Fig4c.ps}}
843: {\includegraphics[width=13cm]{figuras/figuras4/Fig4d.ps}}
844: \caption{Stability regions of the pure modes and mixed modes for
845: the resonance $2:5$. Figures on the right are sketches of
846: bifurcation diagrams associated with the vertical paths
847: (dashed lines) in the
848: stability diagram on the left. $T$ stands for trivial state
849: $A_1=A_2=0$.
850: (a) $\Delta\mu=0.5$, $\gamma=0.5$, $\delta=5$, $\delta'=0.5$, $s=s'=1$, $\rho=0.5$,
851: $\rho'=1.5$ and
852: $\nu=-\nu'=0.085$
853: (b) $\Delta\mu=-0.1125$, $\gamma=0.5$, $\delta=\delta'=1$, $s=s'=1$, $\rho=1.5$,
854: $\rho'=0.5$ and $\nu=0.5$ and $\nu'=-1.01$. The branch of
855: traveling waves that (possibly) connects the $S_+$
856: with the $S_-$ solutions in (a), or that arises from $TW^-$ in (b) is not shown.}
857: \cierto{fig:dgrm2n5Delta-s-}
858: \end{figure}
859: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
860:
861:
862: \clearpage
863: \section{Side-Band Instabilities}
864: \cierto{sec:side-band}
865:
866: We now turn to the analytical core of this paper, the stability
867: of the spatially periodic solutions with respect to side-band
868: instabilities. Such instabilities are expected to destroy the
869: periodicity of the solutions and provide a possible connection
870: with quasi-periodic patterns. We therefore consider
871: perturbations of the periodic solutions in the form
872: \begin{equation}
873: A_1=(1+a_1(X,T))R_1 e^{i(kX+\hat\phi_1)},\quad
874: A_2=(1+a_2(X,T)) R_2e^{i((nk/m)X+\hat\phi_2)},\cierto{8a}
875: \end{equation}
876: where $a_j(X,T)= (a_j^+(T) e^{i Q X}+a_j^-(T)e^{-i QX})$ with $Q\neq
877: 0$ and $j=1, 2$. Note that the perturbation wavenumber $Q$ is measured
878: relative to the deviation wavenumbers $k$ and $nk/m$. The linear stability of
879: the pure modes $S_{1,2}$ and the mixed modes $S_{\pm}$ is calculated by
880: inserting the ansatz (\ref{8a}) into Eqs.~(\ref{2},\ref{3}) and linearizing in
881: $a_j^{\pm}$. The details of these calculations are given in Appendix A.
882:
883: \subsection{Pure modes $\mathbf{S_{1,2}}$}
884:
885: The linearized system for the perturbations associated with the
886: pure modes $S_{1,2}$ separates into two uncoupled
887: $2\times2$-blocks, which allows the stability of each pure mode
888: to be calculated analytically. These two blocks can be
889: associated with longwave and shortwave instabilities,
890: respectively. The block corresponding to the longwave
891: instability contains the eigenvalue related to spatial
892: translations (i.e. phase modulations). The other (shortwave)
893: block describes the evolution of amplitude perturbations in a
894: direction tranverse to the relevant pure-mode subspace. We
895: find that in addition to the instabilities discussed in Section
896: 3 the destabilization of the pure modes can occur by longwave
897: (Eckhaus) or shortwave instabilities.
898:
899: In the case of $S_1$, a straightforward calculation yields
900: \begin{eqnarray} E_1 & : & \mu-\delta k^2-\gamma k-(\gamma+2
901: k\delta)^2/2\delta=0,\cierto{7e1}\\[0.5cm] \Gamma_1 & : &
902: (s\beta-\rho' \alpha)/s+ \delta'(nk/m)^2
903: +\left\{\nu'^2|\alpha/s|^n/4\delta'(nk/m)^2\right\}_{m=2}=0,\cierto{7g1}
904: \end{eqnarray} where $E_1$ is the Eckhaus curve and $\Gamma_1$ is the
905: stability limit associated with shortwave perturbations. If the
906: curve $\Gamma_1$ is crossed, $S_1$ undergoes a steady-state
907: bifurcation with a perturbation wavenumber $Q$ given by
908: \begin{eqnarray}
909: &Q^2 &= \left(\frac{n}{m}\right)^2
910: k^2-\left\{{\nu'}^2\left|\frac{\alpha}{s}\right|^n
911: \left(\frac{m}{n}\frac{1}{2\delta'k}\right)^2
912: \right\}_{m=2}.\cierto{7wn1}
913: \end{eqnarray}
914: When $m>2$ the bracketed terms in Eqs.~(\ref{7g1}) and
915: (\ref{7wn1}) are absent. The destabilizing eigenfunction then takes
916: the simple form $(a_1, a_2) \propto (0,e^{-i k\frac{n}{m} X})$
917: (see Eqs.~(\ref{8a}) and (\ref{eqa2S1}) of Appendix A) and
918: allows a correspondingly simple physical interpretation: the
919: wavenumber of the destabilizing mode is the wavenumber at the
920: band center of the other mode $S_2$ (see Eqs.~(\ref{8a})). This
921: is no longer the case for $m=2$ because the
922: resonance terms (which are linear in $A_2$)
923: affect the stability of $S_1$. One consequence of this is that
924: the destabilizing mode is composed of two wavenumbers: $
925: k\frac{n}{m}\pm Q$ with $Q$ given by Eq.~(\ref{7wn1}).
926: Moreover, it is possible to have one or more non-zero wavenumbers
927: $k^*$, say, for which the
928: perturbation wavenumber $Q$ vanishes. At these points
929: $(k,\mu)=(k^*,\mu^*)$ the curve $\Gamma_1$ merges with the
930: curves $L_1^+$ (or $L_1^-$) describing stability under
931: homogeneous perturbations.
932:
933:
934: In the case of $S_2$ we find
935: \begin{eqnarray}
936: E_2 & : & \mu+\Delta\mu-3\delta'(n k/m)^2=0,\cierto{7e2} \\
937: \Gamma_2 & : & (s'\alpha-\rho\beta)/s'+(\gamma+2\delta k)^2/4 \delta=0,
938: \cierto{7g2}
939: \end{eqnarray}
940: where $E_2$ is the Eckhaus curve and $\Gamma_2$ is the stability
941: limit for shortwave instabilities. As $\Gamma_2$ is crossed the pure mode
942: $S_2$ undergoes a steady-state bifurcation with perturbation wavenumber $Q$
943: given by
944: \begin{equation}
945: Q=\pm|k+\gamma/2\delta|.\cierto{7wn2}
946: \end{equation}
947: The associated eigenfunction is of the form $(a_1,a_2) \propto
948: (e^{-i (k+\frac{\gamma}{2\delta})X},0)$ (see Eqs.~(\ref{8a}) and (\ref{eqa1S2})) and, as in the
949: case of $S_1$, points to a simple interpretation: the mode that destabilizes
950: $S_2$ lies at the band center of $S_1$, i.e., its wavenumber is
951: $\gamma/(2\delta)=\hat\gamma$ (cf.~Eqs.~(\ref{8a}), and Appendix A; see also
952: Fig.~\ref{fig:critic-modes}). This result holds for all resonances except
953: $m:n=1:2$, in which case the linear stability of the pure mode $S_2$ is affected
954: by the resonance terms (as $S_1$ was when $m=2$).
955:
956: Stability results for the pure modes
957: $S_{1,2}$ are shown in Figs.~\ref{fig:dgrm2n5withtw(a)} and
958: \ref{fig:dgrm2n5withtw(b)}. In
959: Fig.~\ref{fig:dgrm2n5withtw(a)}b,d and Fig.~\ref{fig:dgrm2n5withtw(b)}b
960: the parameters are as in
961: Fig.~\ref{fig:dgrm2n5Delta-s+}a,b and Fig.~\ref{fig:dgrm2n5Delta-s-}a,
962: respectively, while in the diagrams of Fig.~\ref{fig:dgrm2n5withtw(a)}a,c
963: and Fig.~\ref{fig:dgrm2n5withtw(b)}a we depart from
964: those parameters only in setting $\gamma=0$. The
965: inclusion of these last three cases allows us to gauge how much the
966: side-band instabilities are affected by the detuning from
967: perfect resonance.
968:
969: Note that due to the large ratio $\delta/\delta'$ the detuning
970: of $\gamma=0.5$ has only a small effect on the neutral curve of
971: $A_1$ ($\alpha=0$). Observe that in Fig.~\ref{fig:dgrm2n5withtw(a)}a,b
972: the pure mode $S_2$ is stable within the region bounded by the
973: curves $E_2$ and $\Gamma_2$ while $S_1$ is everywhere unstable
974: due to shortwave instabilities. In contrast,
975: Fig.~\ref{fig:dgrm2n5withtw(a)}c,d depicts a situation with
976: stable regions for both pure modes. Note that the pure mode
977: $S_2$ experiences only shortwave instabilities as the forcing is
978: decreased while the pure mode $S_1$ can lose stability to either
979: longwave or shortwave perturbations, provided the wavenumber $k$
980: is not within the interval defined by the merging points of
981: $\Gamma_1$ and $L_1^-$ (see Fig.~\ref{fig:dgrm2n5withtw(a)}c and
982: ~\ref{fig:dgrm2n5withtw(a)}d); over that interval the
983: instability suffered by $S_1$ is to
984: perturbations preserving the periodicity of $S_1$
985: and gives rise to (steady) mixed modes. The parameters of
986: Fig.~\ref{fig:dgrm2n5withtw(b)} do not allow for codimension-two
987: points (intersection of the curves $\alpha=0$ and $\beta=0$).
988: Despite this difference, the stability results for $S_2$ are
989: qualitatively similar to those of
990: Fig.~\ref{fig:dgrm2n5withtw(a)}. The effect of $\gamma$ on
991: $S_1$, however, is more striking than in
992: Fig.~\ref{fig:dgrm2n5withtw(a)}c-d. In particular, $S_1$ can
993: undergo a shortwave steady-state instability (along $\Gamma_1$)
994: when $\gamma = 0.5$ but not when $\gamma = 0$.
995:
996:
997:
998:
999:
1000: %%%%%%%%%%%%%FIG3%%%%%%%%%%%%%%%
1001:
1002: \begin{figure}[htb]
1003: \includegraphics[width=15cm,height=8cm]{figuras/figuras6y7/Fig6ab.ps}
1004: \hskip-1.5truecm\includegraphics [width=15cm,height=8.2cm]{figuras/figuras6y7/Fig6cd.ps}
1005: \caption{Stability regions of the pure modes $S_{1,2}$ and the mixed modes
1006: $S_\pm$ for the resonance $2:5$ and the following parameter sets: (a) $\Delta\mu=0.366$,
1007: $\delta=\delta'=1$, $s=s'=1$, $\rho=0.4$, $\rho'=0.67$, $\nu=0.62$, $\nu'=-1.02$ and
1008: $\gamma=0$; (b) as in (a) but $\gamma=0.7$ (cf.
1009: Fig.\ref{fig:dgrm2n5Delta-s+}a); (c) $\Delta\mu=-0.5$,
1010: $\delta=10 $, $\delta'=0.5$, $s=s'=1$, $\rho=1.5$, $\rho'=0.5$, $\nu=-\nu'=0.05$ and
1011: $\gamma=0$; (d) as in (c) but $\gamma=0.5$
1012: (cf. Fig.\ref{fig:dgrm2n5Delta-s+}b). The symbols in b) and d) indicate the
1013: parameters used in the numerical simulations described in Sec.~\ref{sec:numericalresults}.}
1014: \cierto{fig:dgrm2n5withtw(a)}
1015: \end{figure}
1016:
1017: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1018:
1019: %%%%%%%%%%%%%FIG4%%%%%%%%%%%%%%%
1020: \begin{figure}[htb]
1021: \centerline{\includegraphics [width=15cm]{figuras/figuras6y7/Fig7ab.ps}}
1022: \caption{Stability regions of the pure modes $S_{1,2}$ and the mixed modes
1023: $S_\pm$ for the resonance $2:5$ and the following parameter sets:
1024: (a) $\Delta\mu=0.5$, $\gamma=0$, $\delta=5$, $\delta'=0.5$, $s=s'=1$,
1025: $\rho=0.5$,
1026: $\rho'=1.5$, $\nu=-\nu'=0.085$. (b) as in (a) but $\gamma=0.5$ (cf.
1027: Fig.\ref{fig:dgrm2n5Delta-s-}a).}
1028: \cierto{fig:dgrm2n5withtw(b)}
1029: \end{figure}
1030:
1031: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1032:
1033:
1034: \subsection{Mixed modes $S_\pm$}
1035:
1036: The stability of the mixed modes $S_\pm$ is governed by four
1037: eigenvalues which must be determined numerically. Two of these
1038: are associated with amplitude modes and are always real. The
1039: remaining two may be real or complex and are related,
1040: respectively, with the translation mode and with the relative
1041: phase between the two modes $A_1$ and $A_2$. Throughout this
1042: stability analysis we focus on the influence of the detuning
1043: parameter $\gamma$ on the (side-band)
1044: instabilities of $S_\pm$. Furthermore, because
1045: of the invariance of Eqs.~(\ref{2},\ref{3}) under the
1046: transformation
1047: $$\gamma\rightarrow -\gamma,\quad X\rightarrow -X,$$ we may take
1048: $\gamma\geq 0$ without any loss of generality.
1049:
1050: The results of this stability analysis are shown in Fig.~\ref{fig:dgrm2n5withtw(a)}, which
1051: depicts a situation with two codimension-two points, and Fig.~\ref{fig:dgrm2n5withtw(b)},
1052: which presents a case with no codimension-two points. Since this difference seems to have
1053: only a minor effect on the stability of $S_\pm$ we focus our discussion
1054: on the case of Fig.~\ref{fig:dgrm2n5withtw(a)} (i.e. two codimension-two points).
1055:
1056:
1057: \subsubsection{Mixed mode $\mathbf {S_+}$}
1058:
1059: When $\gamma=0$ (Fig.~\ref{fig:dgrm2n5withtw(a)}a,c), the mixed
1060: mode $S_+$ becomes unstable to short-wavelength perturbations
1061: on the part of $\Gamma_{\!+}$ closer to $L_2$, while a
1062: longwave stability analysis captures the part of $\Gamma_{\!+}$
1063: closer to $TW^+$ . Both instabilities are of steady-state type
1064: with the one close to $TW^+$ breaking the reflection symmetry of
1065: the pattern. This is expected to lead to a drift of the pattern;
1066: additional calculations for other parameter sets (not shown)
1067: suggest that this situation is characteristic of $S_+$ and
1068: $\gamma=0$.
1069:
1070: When $\gamma\neq 0$ the mixed mode $S_+$ displays both
1071: steady-state and Hopf bifurcations. The two relevant
1072: eigenvalues are associated with the translation mode and the
1073: symmetry-breaking mode, respectively, with the latter eigenvalue
1074: going to zero at $TW^+$. The change
1075: in character of the instability along $\Gamma_{\!+}$ is
1076: illustrated in Fig.~\ref{fig:growthgamma04}.
1077: Figs.~\ref{fig:growthgamma04}b,c give the
1078: growthrates of the dominant modes as $\Gamma^+$ is crossed at
1079: specific points marked in Fig.~\ref{fig:growthgamma04}a. At points 1-5 the
1080: bifurcation is steady but as one continues in a counterclockwise
1081: direction (points 4-6) the eigenvalues merge and become complex
1082: over an interval of $Q$-values. This interval containing
1083: complex conjugate eigenvalues continues to grow, leading
1084: eventually to an oscillatory instability superceding the steady one.
1085: A codimension-two point
1086: therefore exists near $(k,\mu)=(-0.155, 0.581)$ at the transition from
1087: steady-state to Hopf bifurcation; at this point there are two unstable
1088: modes, at different $Q$, one steady and one oscillatory. As one goes
1089: still further in the counterclockwise direction, the $Q$-band over which
1090: the eigenvalues are complex
1091: reaches $Q=0$ and the oscillatory instability becomes a long-wave
1092: instability. This interaction was studied in the context of Taylor vortex flow
1093: \cite{RiPa92} and has been found in many others problems
1094: \cite{RaRi92,FoRa94}.
1095:
1096: %%%%%%%%%%%FIG%%%%%%%%%%%%%%%
1097: \begin{figure}
1098: \centerline{\includegraphics{figuras/figuras8/Fig8.ps}}
1099: \caption{Growth rates of the two most unstable perturbations of $S_+$ when
1100: $\Gamma_{\!+}$ is crossed at the points indicated in the upper figure:
1101: $Re(\lambda_1)$ (solid-thin line) and $Re(\lambda_2)$ (dashed-thick);
1102: $\lambda_1$ and $\lambda_2$ are complex where there is a single curve, i.e.,
1103: where $Re(\lambda_1)=Re(\lambda_2)$. We use $\gamma=0.4$ with the remaining
1104: parameters as in Figs.~\ref{fig:dgrm2n5withtw(a)}b.} \cierto{fig:growthgamma04}
1105: \end{figure}
1106: %%%%%%%%%%%%%%%%%%%%%
1107:
1108: Similar transitions occur if $\mu$ and $\gamma$ are varied (rather
1109: than $\mu$ and $k$). We illustrate this in
1110: Fig.~\ref{fig:growthm2n5k-01} by plotting the growth rate of the
1111: most unstable perturbations after crossing, at fixed $k=-0.1$, the
1112: upper part of $\Gamma_{\!+}$ which is close to $TW^+$
1113: (Fig.~\ref{fig:growthm2n5k-01}a), and the lower part of
1114: $\Gamma_{\!+}$ which is close to $L_2$
1115: (Fig.~\ref{fig:growthm2n5k-01}b). Along the upper part of
1116: $\Gamma_{\!+}$ the mixed mode $S_+$ typically sees a shortwave
1117: oscillatory instability; for small $\gamma$, $S_+$ can also lose
1118: stability to longwave perturbations. The lower part of
1119: $\Gamma_{\!+}$ is characterized by either steady-state or
1120: oscillatory instability of shortwave type (see
1121: Fig.~\ref{fig:growthm2n5k-01}b). There is again a codimension-two
1122: point, near $(\gamma,\mu)\simeq(1.035, 0.47)$, where Hopf and
1123: steady-state bifurcations coalescence. For $\gamma$ less (greater)
1124: than this critical value the bifurcation is steady (oscillatory).
1125: The route to the oscillatory instability is as described above: with
1126: increasing $\gamma$ the two modes associated with translation and
1127: reflection symmetry interact more and more, ultimately leading to a
1128: Hopf bifurcation.
1129:
1130: The steady-state shortwave instability of $S_+$ (see
1131: Fig.~\ref{fig:growthm2n5k-01}b) always appears in the vicinity
1132: of $L_2$. This proximity implies that the amplitude
1133: $A_1$ of the mixed mode is very small. Consequently,
1134: the behavior of $S_+$ in this region is similar to that of the
1135: pure mode $S_2$. The growing perturbations are predominantly in
1136: the direction of $A_1$ and the associated wavenumber $Q$ is
1137: given in a first approximation by Eq.~(\ref{7wn2}), i.e.,
1138: \begin{equation} Q\simeq \pm |k+\gamma/2\delta|. \cierto{13e}
1139: \end{equation}
1140:
1141:
1142: %%%%%%%%%%%%%FIG6%%%%%%%%%%%%%
1143: \begin{figure}
1144: \centerline{\includegraphics[height=11cm]{figuras/figuras9y12/Fig9.ps}}
1145: \caption{Growth rate of the most unstable perturbation of $S_+$ when
1146: $\Gamma_{\!+}$ is crossed vertically at $k=-0.1$ for the indicated values of $\gamma$,
1147: (a) on the upper part of $\Gamma_{\!+}$ (close to $TW^+$) and (b) on the lower part
1148: (close to $L_2$ ). The remaining parameters are as in Fig.~\ref{fig:dgrm2n5withtw(a)}a-b.
1149: Solid (dashed) lines correspond to real (complex) eigenvalues. }
1150: \cierto{fig:growthm2n5k-01}
1151: \end{figure}
1152: %%%%%%%%%%% %FIG%%%%%%%%%%%%%%
1153:
1154: %%%%%%%%%%%%%FIG7%%%%%%%%%%%%%
1155: \begin{figure}[htb]
1156: \centerline{\includegraphics[width=12cm] {figuras/figuras10y11/Fig10.ps}}
1157: \caption{Dependence of the perturbation wavenumber $Q$ of the steady mode
1158: destabilizing $S_+$ on the
1159: detuning parameter $\gamma$ for the resonace case $m:n=2:5$.
1160: Parameters as in Fig.\ref{fig:dgrm2n5withtw(a)}a,b
1161: with a) $k=-0.1$ and b) $k=0$.
1162: Open circles correspond to the perturbation wavenumber
1163: $Q$ calculated numerically from Eq.~(\ref{14b},\ref{14c}).
1164: Solid lines correspond to the indicated approximations.}
1165: \cierto{fig:maxinstvsgammam2n5k-01k0k01}
1166: \end{figure}
1167: %%%%%%%%%%%%%%%%%%%%%%%
1168:
1169: As demonstrated in Fig.~\ref{fig:maxinstvsgammam2n5k-01k0k01},
1170: Eq.~(\ref{13e}) provides a good approximation over the interval
1171: $0.1\lesssim\gamma\lesssim 0.7$ when $|k|\leq 0.1$. It is interesting
1172: to note that, when $\gamma$ is not too small, Eq.~(\ref{13e}) applies
1173: independently of the resonance $m:n$.
1174: The robustness of this result is not unexpected since the resonance
1175: terms do not influence at first order the stability properties of the
1176: mixed mode $S_+$ (because $|A_1|$ is small). However,
1177: Eq.~(\ref{13e}) ceases to apply when $\gamma$ becomes very small
1178: because resonance effects start to play a significant role in the
1179: wavenumber selection of the linear response when crossing the lower
1180: part of $\Gamma_{\!+}$. In particular, for $k=0$ and $\gamma\ll 1$
1181: one can show \begin{equation} Q\simeq(\gamma^n\nu)^{1/4}\left[ r_{20}
1182: (2s'/(ss'-
1183: \rho\rho'))^{3/4}\right]^{1/2}/2\delta+O(\gamma^{n+1}). \cierto{16a}
1184: \end{equation}
1185: where $r_{20}$ denotes the amplitude of $S_2$ (with $k=0$) on the
1186: curve $L_2$. The derivation of Eq.~(\ref{16a}), given in Appendix B, relies on
1187: the fact that for $\gamma=0$ the curve $\Gamma_{\!+}$ becomes tangent to the
1188: curve $L_2$ at $k=0$.
1189: Fig.~\ref{fig:maxinstvsgamma} shows that (\ref{16a}) is
1190: in excellent agreement with the numerical results obtained directly from the
1191: characteristic equation~(\ref{14b}) and applies to other weak resonances with
1192: $m:n\neq 2:5$. Note that for larger values of $n$ Eq.~(\ref{16a}) is valid over a
1193: wider range of $\gamma$ values (since the error is $O(\gamma^{n+1})$)
1194: and the cross-over to the behavior (\ref{13e}) is shifted to
1195: larger values of $\gamma$.
1196:
1197:
1198:
1199: %%%%%%%%%%%%%FIG8%%%%%%%%%%%%%
1200: \begin{figure}[htb]
1201: \centerline{\includegraphics[width=8cm]{figuras/figuras10y11/Fig11.ps}}
1202: \caption{Dependence of $Q$ on the
1203: detuning parameter $\gamma$ for the indicated resonances $m:n$ and $k=0$.
1204: Remaining parameters are $\Delta\mu=0.366$, $\delta=\delta'=1$, $s=s'=1$,
1205: $\rho=0.4$, $\rho'=0.67$, $\nu=0.62$ $\nu'=-1.02$.
1206: Open circles correspond to numerical results of
1207: Eq.~(\ref{14b},\ref{14c})
1208: and lines to the approximations (\ref{13e}) and (\ref{16a}),
1209: $Q\propto\gamma/2 \delta $ (dashed)
1210: and $Q\propto\gamma^{n/4}$ (solid), respectively.
1211: }
1212: \cierto{fig:maxinstvsgamma}
1213: \end{figure}
1214:
1215: %%%%%%%%%%%%%%%%%%%
1216:
1217:
1218: \bigskip
1219: \subsubsection{\bf Mixed mode $\bold {S_-}$}
1220:
1221: As illustrated in Fig.~\ref{fig:dgrm2n5withtw(a)}a, the mixed mode $S_-$ can be
1222: everywhere unstable when $\gamma=0$. For the parameters of
1223: Fig.~\ref{fig:dgrm2n5withtw(a)}a, such a situation persists until $\gamma \approx 0.6$, when
1224: a stability region for $S_-$ emerges. This stability region widens as $\gamma$ is increased
1225: (see Fig.~\ref{fig:dgrm2n5withtw(a)}b).
1226:
1227: As in the case of $S_+$, the two most important eigenvalues for
1228: the stability of $S_-$ are those associated with the
1229: translation and parity-breaking modes. The oscillatory
1230: instability results from an interaction between these two
1231: eigenvalues and, consequently, is found in the vicinity of the
1232: bifurcation set $TW^-$; the steady (shortwave) instability
1233: occurs near the initial bifurcation producing the mixed mode
1234: $S_-$ from the pure mode (i.e., near $L_1^-$). Thus, the
1235: mixed mode $S_-$ undergoes the same kinds of transitions
1236: described above for $S_+$. Depending on where $\Gamma_{\!-}$
1237: is crossed in the $(k,\mu)$- or the
1238: $(\gamma,\mu)$-plane, the instability may be either oscillatory or
1239: steady, and either of long-wave or short-wave type.
1240:
1241: %%%%%%%%%%%%%FIG6%%%%%%%%%%%%%
1242: \begin{figure}[htb]
1243: \centerline{\includegraphics[height=10cm]{figuras/figuras9y12/Fig12.ps}}
1244: \caption{Growth rate of the most unstable perturbation when $\Gamma_{\!-}$ is
1245: crossed vertically at $k=0.5$ for the indicated values of
1246: $\gamma$ and the remaining parametersas in
1247: Fig.~\ref{fig:dgrm2n5withtw(a)}a-b. (a) on the
1248: upper part of $\Gamma_{\!-}$ (close to $TW^-$) and (b) on the lower part (close
1249: to $L_1^-$ ). A solid line is used when the associated eigenvalue is real and
1250: a dashed line when it is complex.} \cierto{fig:growthm2n5k05}
1251: \end{figure}
1252:
1253: %%%%%%%%%%% %FIG%%%%%%%%%%%%%%
1254:
1255: Note that on the upper part of $\Gamma_{\!-}$ in
1256: Fig.\ref{fig:growthm2n5k05}a, as $\gamma$ is increased the two
1257: eigenvalues merge and become complex at ever smaller values of
1258: $Q$. In contrast, on the lower part of $\Gamma_{\!-}$
1259: raising $\gamma$ increases
1260: the damping of the parity-breaking mode, thus
1261: discouraging its interaction with the translation mode, this
1262: behavior is opposite to what occurs with $S_+$. It is also
1263: interesting to note that variations in the detuning parameter
1264: $\gamma$ do not induce a significant change in the perturbation
1265: wavenumber $Q$ associated with the bifurcation. This is
1266: because the instability of $S_-$ takes place close to $L_1^-$
1267: where the destabilizing perturbations are (preferentially) in
1268: the direction of $A_2$. The perturbation wavenumber $Q$
1269: associated with the instability is given, to first
1270: approximation, by Eq.~(\ref{7wn1}). Thus $Q$ depends mainly on
1271: the resonance $m:n$ and only implicitly on the parameter
1272: $\gamma$ (through the function $\alpha$ defined in
1273: Eq.~(\ref{5b})).
1274:
1275:
1276:
1277:
1278:
1279:
1280: \section{Numerical Simulations}
1281:
1282: \cierto{sec:numericalresults}
1283:
1284: The main objective in this section is to investigate the
1285: nonlinear evolution of the side-band instabilities of the pure
1286: and mixed modes discussed in Section~\ref{sec:side-band}. The
1287: results are obtained for a periodic domain of length $L$ using a
1288: finite-difference code with Crank-Nicholson time-stepping.
1289: Each simulation of (\ref{2},\ref{3}) uses as an initial
1290: condition an unstable periodic steady state (i.e., a pure or
1291: mixed mode of the type discussed in
1292: Section~\ref{sec:basicsolutions}) with a small random
1293: perturbation. The final solutions obtained in this way are
1294: typically `quasi-periodic' although periodic solutions are also
1295: found. Here we use the term quasi-periodic to refer to any
1296: solution for which the wavenumbers of $A_1$ and $A_2$ are {\it
1297: not} in the ratio $m/n$. Therefore the full reconstructed
1298: field (\ref{1a}) will not be composed of wavenumbers in the
1299: ratio $m/n$; in general it will be quasi-periodic.
1300:
1301: First we describe the evolution of system~(\ref{2},\ref{3}) when
1302: the pure modes become unstable. If the instability is longwave
1303: it affects only the excited amplitude and the solution remains
1304: in the pure-mode subspace. For example, in the case of $S_1$ (see
1305: Fig.~\ref{fig:spacetime1a}) the longwave instability involves
1306: $A_1$ but not $A_2$, which remains zero. As expected, the
1307: changes affect predominantly the phase of $A_1$ and evolve into
1308: a series of phase-slips. The final state, which is
1309: time-independent, is shown on the right part of
1310: Fig.~\ref{fig:spacetime1a}. In the case of $S_2$ (not shown)
1311: the longwave instability behaves analogously: it is now $A_2$
1312: that changes, while $A_1$ remains zero.
1313:
1314: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1315: \begin{figure}[htb]
1316: \centerline{\includegraphics[width=11cm]{figuras/figuras13/Fig13h.eps}}
1317: \caption{Evolution of the longwave instability of the pure mode
1318: $S_1$. Space-time diagram of the real part of $A_1$ (left) and
1319: final state (right) with thick (thin) lines denoting the real (imaginary)
1320: part of $A_1$. Parameters as in Fig.~\ref{fig:dgrm2n5withtw(a)}d with
1321: $k_1=k=0.15$, $\mu=0.37$ (indicated by a square in
1322: Fig.~\ref{fig:dgrm2n5withtw(a)}d) and $L=250$.}
1323: \cierto{fig:spacetime1a}
1324: \end{figure}
1325: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1326:
1327: Depending on the detuning parameter $\gamma$ the transition from
1328: the pure modes to the mixed modes can also involve side-band
1329: instabilities. For the parameters corresponding to the stability
1330: diagram shown in Fig.~\ref{fig:dgrm2n5withtw(b)} the pure mode
1331: $S_1$ becomes unstable to the mixed mode $S_-$ $via$ a side-band
1332: instability for $\gamma \ne 0$ and $k>0.15$ (cf.
1333: Fig.~\ref{fig:dgrm2n5withtw(b)}b), while for $\gamma=0$, i.e.
1334: when the two critical wavenumbers are resonant, the instability
1335: preserves the periodicity of the solution for all $k$ (cf.
1336: Fig.~\ref{fig:dgrm2n5withtw(b)}a). The final states that are
1337: reached after a small random perturbation has been applied to a
1338: pure-mode solution $S_1$ are depicted in
1339: Fig.\ref{fig:puremixed}. The movies {\tt fig5.2a.avi} and {\tt
1340: fig5.2b.avi} show the corresponding temporal evolution of $A_2$.
1341: The sequence of phase slips that is evidenced in the sharp peaks of the local
1342: wavenumber during the early stages of the evolution is a
1343: consequence of the random initial perturbation. After this
1344: transient the relevant perturbation mode starts
1345: to dominate. In the case $\gamma=0.5$ that mode has a strongly
1346: modulated wavenumber reflecting the fact that the wavenumbers of
1347: $A_1$ and $A_2$ do not satisfy the resonance condition
1348: $m:n=2:5$, resulting in a quasi-periodic reconstructed solution
1349: $u(x)$ (see also discussion leading to (\ref{19}) below).
1350:
1351: %%%%%%%%%%%%%FIG%%%%%%%%%%%%%
1352: \begin{figure}[htb]
1353: \centerline{\includegraphics[width=11cm]{figuras/figuras20/Fig20.ps}}
1354: \caption{Final state after a small random perturbation of the
1355: pure mode $S_1$.
1356: Parameters in (a) and (b) as in
1357: Fig.~\ref{fig:dgrm2n5withtw(b)}a,b, respectively
1358: with $L=250$ and $\gamma=0$, $k=0.1$, $\mu=1.4$
1359: in a)
1360: and $\gamma=0.5$, $k=0.2$, $\mu=1.4$ in b). The corresponding
1361: movies [movie:fig5.2a.avi] and [movie:fig5.2b.avi] show the
1362: temporal evolution of $A_2$. Red and yellow lines show the real
1363: and imaginary part of $A_2$, respectively, white its magnitude $|A|$,
1364: and green the local wavenumber.}
1365: \cierto{fig:puremixed}
1366: \end{figure}
1367: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1368:
1369:
1370: The longwave instability of the mixed modes affects $A_1$ as
1371: well as $A_2$ (see Fig.~\ref{fig:spacetime1b}). As with the
1372: pure modes, the phase perturbations evolve into phase slips,
1373: which change the wavenumber and lead eventually to a stationary
1374: state. In the case shown in Fig.~\ref{fig:spacetime1b} phase
1375: slips occur only in mode $A_2$. As in the mixed mode shown in
1376: Fig.\ref{fig:puremixed}b, the strong deviations of $A_{1,2}(x)$
1377: from purely sinusoidal behavior, which are due to the resonant
1378: terms proportional to $\nu$ and $\nu'$, indicate that the
1379: reconstructed solution $u(x)$ is not periodic.
1380:
1381: %%%%%%%%%%%%%FIG%%%%%%%%%%%%%%%%%%%%
1382:
1383: \begin{figure}[htb]
1384: \centerline{\includegraphics[height=7.5cm]{figuras/figuras13/Fig13g.eps}}
1385: \centerline{\includegraphics[height=4.5cm]{figuras/figuras13/Fig13g1.eps}}
1386: \caption{Evolution of longwave instability of the mixed mode $S_+$.
1387: Shown are space-time diagrams of the
1388: real part of the amplitudes and the final state
1389: with thick (thin) lines denoting the real (imaginary) parts of $A_1$ and $A_2$,
1390: respectively. Parameters as in
1391: Fig.~\ref{fig:growthgamma04}a with $k=0.1$, $\mu=0.423$
1392: (indicated by rhomb) and
1393: $L=300$.} \cierto{fig:spacetime1b}
1394: \end{figure}
1395: %%%%%%%%%%%%%FIG%%%%%%%%%%%%%
1396:
1397: The mixed modes can undergo a shortwave steady instability as
1398: well. As discussed in Section~\ref{sec:side-band} it occurs
1399: when the
1400: stability limit $\Gamma_+$ ($\Gamma_-$) of $S_+$ ($S_-$) is
1401: close to the transition line $L_2$ ($L_1^-$). The instability
1402: is primarily in the direction of the mode with smaller
1403: amplitude: $A_1$ for $\Gamma_+$ near $L_2$ and $A_2$ for
1404: $\Gamma_-$ near $L_1^-$. The resulting evolution of
1405: system~(\ref{2},\ref{3}) is shown in Fig.~\ref{fig:spacetime2}
1406: for parameters near $\Gamma_+$. It confirms the expectation
1407: that at least initially only $A_1$ changes, but not $A_2$. The
1408: space-time diagram for Re$(A_1(X,T))$ in
1409: Fig.~\ref{fig:spacetime2} shows how the growing perturbation
1410: determines the wavenumber of the final state; this change from
1411: the initial wavenumber takes place $via$ phase-slips.
1412:
1413:
1414: The wavenumber $k_1$ of $A_1$ in the final state is controlled by the
1415: detuning parameter $\gamma$ in the manner predicted by the linear
1416: analysis: $k_1\simeq \gamma/(2\delta)$. In other words, the number of maxima
1417: observed in $A_1$ is $N_1 \simeq L \gamma/(4\pi\delta)$ (see lower figures of
1418: Fig.~\ref{fig:spacetime2}).
1419:
1420: %%%%%%%%%%%%%FIG%%%%%%%%%%%%%
1421: \begin{figure}[htb]
1422: \centerline{\includegraphics[height=6cm, width=10cm]{figuras/figuras13/Fig13d.eps}}
1423: \centerline{\includegraphics[height=4cm, width=10cm]{figuras/figuras13/Fig13d1.eps}}
1424: \caption{Space-time diagram showing the real part of the amplitudes $A_1$ and $A_2$. The
1425: lower figures show the real (thick line) and imaginary (thin line) parts of $A_1$ and $A_2$
1426: corresponding to the final stable stationary state. The parameters are as in
1427: Fig.~\ref{fig:dgrm2n5withtw(a)}b ({\it square }) with $k=-0.05$, $\mu=0.27$ and $L=250$.}
1428: \cierto{fig:spacetime2}
1429: \end{figure}
1430: %%%%%%%%%%%%%FIG%%%%%%%%%%%%%
1431:
1432: The evolution of $A_1$ and $A_2$ after a shortwave steady
1433: instability on $\Gamma_-$ near $L_1^-$ is shown in
1434: Fig.~\ref{fig:spacetime3}. This time the amplitude $A_1$
1435: remains nearly constant while $A_2$ changes. The destabilizing
1436: mode was shown in Section~\ref{sec:side-band} to be of the form
1437: $a_2^+(T)e^{i(k\frac{n}{m}+Q)X}+ a_2^-(T)e^{i(k\frac{n}{m}-Q)X}$
1438: with $Q$ defined by Eq.~(\ref{7wn1}). For the cases illustrated
1439: in Fig.~\ref{fig:spacetime3} we have $|a_2^+(T)| \ll |a_2^-(T)|$
1440: and the wavenumber of the linearly unstable mode is
1441: approximately given by $k_2\approx k \frac{n}{m}-|Q|$. Note
1442: that this wavenumber is essentially determined by the resonance
1443: terms and would be 0 without them (see Eq.~\ref{7wn1}). In
1444: particular, for the example considered in
1445: Fig.~\ref{fig:spacetime3}a, where $\nu=0.62$ and
1446: $\nu'=-1.02$,
1447: we have $|k_2|\simeq 0.144$, while
1448: in Fig.~\ref{fig:spacetime3}b $\nu=-\nu'=0.05$ and $|k_2|\simeq 0.018$;
1449: note that with regard to this effect both cases are
1450: are analogous, except for the magnitude of the resonance coefficients.
1451:
1452: In addition to the wavenumber $k_2$ determined by the linear
1453: stability analysis, Fig.~\ref{fig:spacetime3} reveals a second
1454: prominent wavenumber $k_2'$, which is the result of the
1455: resonance term $\bar{A}_2^{m-1}A_1^n$. It acts as a driving term
1456: for $A_2$ and generates a mode with wavenumber $k_2'$ that is
1457: determined by
1458: \begin{equation} nk_1\pm(m-1)|k_2|=\pm |k_2'|,
1459: \cierto{19} \end{equation}
1460: where $k_1$ is the wavenumber associated with amplitude $A_1$.
1461: In the evolution shown in Fig.\ref{fig:spacetime3} $k_1$ remains
1462: unchanged. The same mechanism is at work for the short-wave
1463: instability of $S_+$ (cf. Fig.\ref{fig:spacetime2}), which
1464: occurs near $L_2$. There the wavenumber modulation of $A_1$ is,
1465: however, negligible. It is driven by the resonance term
1466: $\bar{A}_1^{n-1}A_2^m$, which is proportional to the fourth
1467: power of the amplitude $A_1$ ($n=5$) which is very small where $S_+$
1468: branches off the pure mode $S_2$. This is not the case for the
1469: short-wave instability of $S_-$ (Fig.\ref{fig:spacetime3}),
1470: which occurs near $L_1^-$.
1471: There it is $A_2$ that is small; but
1472: it enters the resonance term $\bar{A}_2^{m-1}A_1^{n}$ linearly
1473: ($m=2$) and therefore the resonance term
1474: provides a relatively strong modulation of $A_2$. Substituting the
1475: final values of $k_1$ (same as the initial value) and $k_2$
1476: (calculated above) into Eq.(\ref{19}) one obtains
1477: $|k_2'|=2.644$ in the case of Fig.~\ref{fig:spacetime3}a
1478: and $|k_2'|=0.63$ for Fig.\ref{fig:spacetime3}b.
1479: For the numbers ($N_1$,$N_2$,$N'_2$)
1480: of wavelengths associated with $k_1$, $k_2$, and $k_2'$,
1481: respectively this implies $(N_1,N_2,N_2')=(20,6,106)\simeq
1482: L/(2\pi)(0.5,0.144,2.644)$ for
1483: Fig.~\ref{fig:spacetime3}a (where $L=250$) and
1484: $(N_1,N_2,N_2')=(6,1,31)\simeq L/(2\pi)(0.124,0.02,0.63)$ for
1485: Fig.~\ref{fig:spacetime3}b (where $L=305$). These results
1486: compare quite well with the
1487: results in Fig.~\ref{fig:spacetime3}
1488: where $(N_1,N_2,N_2')=(20,6,106)$ in
1489: Fig.~\ref{fig:spacetime3}a and $(N_1,N_2,N_2')=(6,0,30)$ in
1490: Fig.~\ref{fig:spacetime3}b.
1491:
1492: %%%%%%%%%%%%%FIG%%%%%%%%%%%%%
1493: \begin{figure}[htb]
1494: \hskip1truecm{\includegraphics[height=5.5cm, width=11cm]{figuras/figuras13/Fig13e.eps}}
1495: \centerline{\includegraphics[height=3.5cm, width=10cm]{figuras/figuras13/Fig13e1.eps}}
1496: \vskip0.5truecm
1497: \hskip1truecm{\includegraphics[height=5.5cm, width=11cm]{figuras/figuras13/Fig13f.eps}}
1498: \centerline{\includegraphics[height=3.5cm, width=10cm]{figuras/figuras13/Fig13f1.eps}}
1499: \caption{Space-time diagram showing the real parts of $A_1$ and
1500: $A_2$ when mixed
1501: mode $S_-$ undergoes a shortwave steady instability. The parameters in (a) are those of
1502: Fig.~\ref{fig:dgrm2n5withtw(a)}b ({\it circle}) with $k=0.5$, $\mu=1.05$ and $L=250$. Parameters
1503: in (b) are as in Fig.~\ref{fig:dgrm2n5withtw(a)}d ({\it circle})
1504: with $k=0.125$, $\mu=0.95$ and $L=305$.}
1505: \cierto{fig:spacetime3}
1506: \end{figure}
1507: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1508:
1509: \clearpage
1510: %%%%%%%%%%%%%FIG%%%%%%%%%%%%%
1511: \begin{figure}[htb]
1512: \centerline{\includegraphics[height=8cm]{spacetime/Fig14b.ps}}
1513: \centerline{\includegraphics[height=8cm]{spacetime/Fig14a.ps}}
1514: \centerline{\includegraphics[height=5.5cm,width=12cm]{spacetime/Fig14c.ps}}
1515: \caption{Space-time diagram of the phase-gradient and the magnitude
1516: of the amplitudes $A_1$ and $A_2$. Lower figures:
1517: Real (thick) and imaginary (thin) part of $A_1$ and $A_2$ corresponding
1518: to the final stable state. Parameter as in
1519: Fig.~\ref{fig:dgrm2n5withtw(a)}b ({\it triangle})
1520: with $k=0.5$, $\mu=1.308$ and $L=250$. In the gray-scale plots dark
1521: (light) stands for low (high) amplitudes.}
1522: \cierto{fig:spacetime4}
1523: \end{figure}
1524: %%%%%%%%%%%%%FIG%%%%%%%%%%%%%
1525:
1526:
1527: Qualitatively different is the evolution of the oscillatory
1528: instability of the mixed mode, as illustrated in
1529: Figs.~\ref{fig:spacetime4}-\ref{fig:spacetime8}. Small
1530: perturbations initially evolve into standing-wave oscillations.
1531: They do not last for long, however, and decompose into left- and
1532: right-traveling disturbances, which then form localized
1533: low-wavenumber domains of waves drifting to the right and to the
1534: left, respectively. Such localized, propagating regions of
1535: `drift waves' are typical for parity-breaking instabilities
1536: because the extended drift waves are generically unstable at
1537: onset \cite{RiPa92,CaCa92,FaDo90} and may become stable only for
1538: larger amplitudes \cite{BaMa94}. The localized waves can be
1539: described using equations for the amplitude of the
1540: parity-breaking mode and the phase of the underlying pattern and
1541: can arise stably when the parity-breaking bifurcation is
1542: subcritical \cite{CoGo89,GoGu91} as well as when it is
1543: supercritical \cite{RiPa92,CaCa92}. They are related to the
1544: solitary modes observed in experiments in directional
1545: solidification \cite{SiBe88,FaCh89} as well as in Taylor vortex
1546: flow \cite{WiMc92}, and in simulations of premixed flames
1547: \cite{BaMa94}. In these systems the parity-breaking bifurcation
1548: arises from a 1:2 mode interaction \cite{LeRa91,RiPa92} rather
1549: than the 2:5-resonance considered here.
1550:
1551: %%%%%%%%%%%%%FIG%%%%%%%%%%%%%
1552: \begin{figure}[htb]
1553: \centerline{\includegraphics[height=9cm]{spacetime/Fig18b.ps}}
1554: \centerline{\includegraphics[width=10cm]{spacetime/Fig18c.ps}}
1555: \caption{a) Space-time diagram of the
1556: phase-gradients amplitudes $A_1$ and $A_2$, respectively.
1557: Parameters as in Fig.~\ref{fig:dgrm2n5withtw(a)}b but $\gamma=1$
1558: and with $k=0.5$, $\mu=1.42$ and $L=250$.
1559: Lower figures correspond
1560: to the final state: a) Real and imaginary part of the amplitudes $A_1$ and $A_2$
1561: b) phase gradients of $A_1$ and $A_2$. Dotted line indicates perfect $2:5$ resonance.
1562: The temporal evolution of $A_2$ is also shown in movie
1563: [movie:fig.5.7.avi]; red and yellow lines show the real
1564: and imaginary part of $A_2$, respectively, white its magnitude $|A|$,
1565: and green the local wavenumber.}
1566: \cierto{fig:spacetime5}
1567: \end{figure}
1568: %%%%%%%%%%%%%FIG%%%%%%%%%%%%%
1569:
1570: The localized drift waves do not always persist. In the
1571: simulation shown in Fig.\ref{fig:spacetime4} a sequence of
1572: phase slips occurs in mode $A_2$ at the trailing end of the
1573: localized drift wave. The phase slips
1574: substantially reduces the wavenumber in the growing domain
1575: between the location where the localized drift wave first
1576: is created (at $x \approx 0.6/L$) and its
1577: trailing edge. Since no phase slips occur in mode $A_1$ the
1578: wavenumbers of $A_1$ and $A_2$ are not resonant any more in this
1579: growing domain, which results in a strong modulation of the
1580: wavenumber. Due to the periodic boundary
1581: conditions the growing, localized drift wave collides
1582: eventually with this domain and is absorbed by it. After
1583: that the pattern becomes stationary. The stationary domains with
1584: strongly modulated wavenumber do not always absorb the localized
1585: drift waves. Fig.\ref{fig:spacetime5} and the movie {\tt
1586: fig5.7.avi} show a case in which the localized drift waves at
1587: times are also reflected by the stationary domains or pass
1588: through them. We have not investigated which
1589: factors determine the outcome of the collisions. In the
1590: simulation shown in Fig.\ref{fig:spacetime5} the system
1591: eventually evolves into a stationary state. It consists of
1592: relatively large domains in which the wavenumber ratio is very
1593: close to 2:5 and small domains in which the wavenumbers are
1594: strongly modulated and not in this rational ratio. One may have
1595: expected that the localized stationary domains
1596: attract each other and eventually merge; but this was not observed.
1597: Presumably, the strong wavenumber oscillations, which are
1598: particularly visible in $A_2$, lock the domains in place. Thus,
1599: in the final, stationary state shown in
1600: Fig.\ref{fig:spacetime5}b one has a periodic pattern that is
1601: interrupted by small domains in which the pattern is not
1602: periodic.
1603:
1604: The transients and the resulting final state depend very
1605: sensitively on the amplitude of the initial perturbations of the
1606: periodic state. The relatively large perturbations used in
1607: Fig.\ref{fig:spacetime5} lead to a large number of
1608: localized drift waves, which upon their collision generate
1609: localized stationary structures. Since the
1610: localized stationary structures often absorb the localized
1611: drift waves the final state tends to be stationary in this case.
1612: If the initial perturbations are much smaller, fewer drift waves
1613: arise. If they travel in the same direction no stationary
1614: localized structures are generated and the localized drift waves
1615: persist. Such a case is illustrated in
1616: Fig.\ref{fig:spacetime7},\ref{fig:spacetime8} and movie
1617: {\tt fig5.8.avi}. Here, uniformly distributed
1618: perturbations with maximal amplitude 0.001 were applied
1619: independently to the real and imaginary part of both amplitudes.
1620: In this case only a single domain forms. It
1621: develops a defect in its interior, which eventually disappears
1622: at its trailing end (cf. movie {\tt
1623: fig5.8.avi}). Initially, the propagation velocities of the
1624: leading and the trailing front of the domain are not equal and
1625: the domain grows. As time goes on, however, the front velocities
1626: converge to a common value and one sees a stable, localized,
1627: propagating domain of fixed size containing traveling waves
1628: (see Fig.~\ref{fig:spacetime8}). It is interesting to note that
1629: the wavenumbers selected in these traveling domains are in
1630: $2:5$ resonance, while the stationary regions have slightly
1631: shifted wavenumbers that are not in resonance (see upper panel
1632: of Fig.~\ref{fig:spacetime8}). The physical solution
1633: (cf.(\ref{1a})) can therefore by described as consisting
1634: of regions of localized periodic waves traveling through a
1635: stationary quasi-periodic pattern. This type of solution is
1636: generally found when the side-band oscillatory instability
1637: occurs very close to the parity-breaking bifurcation. To describe
1638: the parity-breaking bifurcation in more detail the appropriate
1639: amplitude-phase equations \cite{CoGo89,RiPa92,CaCa92,BaMa94} could be
1640: derived from the coupled Ginzburg-Landau equations. They would, in particular,
1641: allow a better understading of the localized drift waves.
1642: Such an undertaking is, however, beyond the scope of this paper.
1643:
1644:
1645: %%%%%%%%%%%%%FIG%%%%%%%%%%%%%
1646: \begin{figure}[htb]
1647: \centerline{\includegraphics[height=9cm]{spacetime/Fig19a.ps}}
1648: \centerline{\includegraphics[height=9cm]{spacetime/Fig19b.ps}}
1649: \caption{Space-time diagram for the magnitude and the phase-gradient
1650: of the amplitudes $A_1$ and $A_2$.
1651: Parameters: as in Fig.~\ref{fig:dgrm2n5withtw(a)}b but $\gamma=1.3$
1652: and with $k=0.5$, $\mu=1.42$ and $L=250$. In the gray-scale plots dark
1653: (light) stands for low (high) amplitudes. The temporal evolution of $A_2$ is also shown in
1654: movie [movie:fig.5.8.avi]; red and yellow lines show the real
1655: and imaginary part of $A_2$, respectively, white its magnitude $|A|$,
1656: and green the local wavenumber.}
1657: \cierto{fig:spacetime7}
1658: \end{figure}
1659: %%%%%%%%%%%%%FIG%%%%%%%%%%%%%
1660:
1661: %%%%%%%%%%%%%FIG%%%%%%%%%%%%%
1662: \begin{figure}[htb]
1663: \centerline{\includegraphics[height=18cm]{spacetime/Fig19c.ps}}
1664: \caption{Final, localized drift wave reached after
1665: the transient shown in Fig.~\ref{fig:spacetime7}. Dotted line in
1666: plot of $\phi_{1X}/phi_{2X}$ indicates perfect $2:5$ resonance.}
1667: \cierto{fig:spacetime8}
1668: \end{figure}
1669: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1670:
1671:
1672: Finally, in order to determine the character of the bifurcations
1673: we have also integrated Eqs.~(\ref{2},~\ref{3}) using the final states as
1674: initial conditions and varying $\mu$ so as to take the system
1675: back into the stable regions of mixed modes. The results
1676: obtained indicate that the instabilities discussed in this section are
1677: subcritical (except for the cases shown in Fig.\ref{fig:puremixed}).
1678: The solutions persist over a large interval of
1679: the forcing giving rise to bi-stability between periodic and
1680: quasi-periodic states. This behaviour has been found in all cases
1681: we have considered. Therefore we expect that the coexistence of
1682: periodic and quasi-periodic stable states may be regarded as a
1683: characteristic property of the system Eqs.~(\ref{2},~\ref{3}).
1684:
1685: \clearpage
1686:
1687:
1688:
1689:
1690:
1691: \section{Conclusions}
1692:
1693: In this paper we have investigated a near-resonant steady $m:n$
1694: mode interaction in one-dimensional dissipative systems with
1695: reflection and translation symmetry using coupled
1696: Ginzburg-Landau equations that describe the slow evolution of
1697: the two relevant mode envelopes. One of the goals was to shed
1698: some light on the competition between periodic and quasi-periodic
1699: patterns by addressing in detail the stability of the periodic
1700: solutions, emphasizing particular instabilities that can lead to
1701: quasi-periodic patterns.
1702:
1703: The simplest stationary solutions of this system are
1704: reflection-symmetric periodic patterns: the pure modes $S_{1,2}$ and
1705: the mixed modes $S_\pm$. Building on the results of Dangelmayr
1706: \cite{Da86} we established the stability properties of these
1707: solutions with respect to perturbations that preserve the periodicity
1708: of the solution. Among the many possible choices of parameters we
1709: restricted our attention to sets of values that produced two
1710: supercritical primary bifurcations to pure modes followed by
1711: secondary bifurcations to mixed modes - these mixed modes, in turn,
1712: suffered symmetry-breaking bifurcations to traveling waves.
1713:
1714: Side-band instabilities modify the stability regions of the pure and
1715: mixed modes. In particular, we find that these steady periodic patterns
1716: can undergo longwave as well as shortwave instabilities. For the
1717: pure modes the side-band instabilities are always of
1718: steady-state type. The destabilizing longwave perturbations
1719: correspond to the well-known Eckhaus instability and depend only on
1720: the linear coefficients of Eqs.~(\ref{2},\ref{3}). In contrast, the
1721: shortwave instabilities depend on nonlinear interactions and
1722: on the resonance considered (i.e., on $m$ and $n$). In particular, we
1723: find:
1724:
1725: 1. If $m>2$ and a pure mode, $S_1$ say, loses stability in a
1726: shortwave steady-state bifurcation, the wavenumber of the
1727: growing perturbation lies at the band center of the other mode
1728: (i.e. of $S_2$)); this
1729: instability is affected by the detuning parameter
1730: $\hat\gamma=\gamma/2\delta$, which measures the mismatch between
1731: the two critical wavenumbers.
1732:
1733: 2. For wavenumber ratios of the form $2:n$ one of the resonant terms
1734: enters into the stability calculation for the pure mode $S_1$.
1735: Away from the band center this pure mode can experience
1736: shortwave instabilities while close to the band
1737: center it can lose stability only to homogeneous or longwave
1738: perturbations. In the short-wave case, due to the resonant terms the
1739: wavenumber selected by the instability is shifted with respect to the
1740: band center of the other pure mode, $S_2$. As the band center is
1741: approached the perturbation wavenumber of the short-wave
1742: instability can go to 0.
1743:
1744: Close to their bifurcation from the pure modes, the mixed modes
1745: inherit the stability properties of the pure modes with respect
1746: to short-wave perturbations. In addition, they may
1747: undergo an oscillatory instability. It results
1748: from the interaction of the translation and the
1749: symmetry-breaking modes and can be strongly affected by the
1750: wavenumber mismatch.
1751:
1752:
1753: The nonlinear evolution ensuing from the instabilities was
1754: investigated through direct numerical simulation of
1755: Eqs.~(\ref{2},\ref{3}) with periodic boundary conditions for the
1756: slowly varying envelopes. It was found that the transition from
1757: the pure modes to the mixed mode can depend strongly on the
1758: wavenumber mismatch. We have identified cases in which without a
1759: wavenumber mismatch the instability of the pure mode leads to a
1760: periodic mixed mode, while in the non-resonant case the
1761: instability of the pure mode is determined by a side-band
1762: instability resulting in a quasi-periodic pattern.
1763:
1764: Side-band instabilities of the mixed modes generically destroy
1765: the spatial periodicity of the stationary, reflection-symmetric
1766: states and produce quasi-periodic patterns. They are typically
1767: characterized by a spatial modulation of the local wavenumber with a
1768: wavelength that is determined by the detuning
1769: parameter (in a manner predicted by the linear analysis) and the
1770: resonant terms (as indicated in Eq.(\ref{19})). The size of the
1771: modulation depends strongly on the strength of the resonance
1772: terms. For the weakly nonlinear 2:5 resonance discussed here the
1773: effect is only noticeable for $S_-$ (cf.
1774: Fig.\ref{fig:spacetime3}); for $S_+$ the resonant terms are too
1775: small (cf. Fig.\ref{fig:spacetime2}).
1776:
1777: The oscillatory instability tends to lead to the formation of
1778: propagating fronts separating reflection symmetric
1779: patterns from drift waves with broken
1780: reflection symmetry. The fronts may interact to form
1781: stable localized drift waves. These may subsequently collide
1782: with each other, destroying the initial periodic pattern
1783: completely in some cases and only partially in others. In the
1784: latter case the system relaxes to a very interesting pattern
1785: composed of alternating localized stationary domains of periodic
1786: and quasi-periodic states. We find other cases where all the
1787: fronts travel in the same direction, eventually achieving the
1788: same velocity and producing stable localized
1789: propagating domains of drift waves. The
1790: wavenumbers in these traveling domains are selected by the
1791: resonance condition (i.e., they are in the ratio $m/n$), whereas
1792: the wavenumbers associated with the surrounding steady periodic
1793: pattern are no longer resonant. Thus, the resulting patterns
1794: can be described as localized periodic traveling wave structures
1795: propagating through a stationary quasi-periodic background.
1796:
1797: In this paper we have focused on the periodic solutions and
1798: their stability. Complementary to this analysis would be a study
1799: of the quasi-periodic solutions. Of course, the set of solutions
1800: in which the two modes have non-resonant wavenumbers is
1801: considerably larger than that of the periodic solutions and such
1802: an analysis is beyond the scope of the present work. In
1803: preliminary numerical simulations we have considered the
1804: stability of certain quasi-periodic solutions and found that the
1805: range in the wavenumber mismatch over which they are stable is
1806: smaller than that of periodic solutions with nearby wavenumbers
1807: \cite{HiRiunpub}. Thus, at least in the regime considered, the
1808: resonant terms enhance the stability of the periodic
1809: solutions. Another promising line of investigation would be the
1810: reduction of the coupled Ginzburg-Landau equations (\ref{2},\ref{3})
1811: to two coupled phase equations. This should be possible for long-wave
1812: perturbations of the mixed-mode solution if the smallness of the
1813: resonant terms is exploited and it is assumed that
1814: such terms are of the same order as the gradients in the
1815: magnitude of the amplitudes (cf. (\ref{6a},\ref{7a})).
1816:
1817: We gratefully acknowledge discussions with J. Porter and P.
1818: Umbanhowar.
1819:
1820: \appendix
1821:
1822: \section{Stability of Pure and Mixed Modes}
1823:
1824: In this appendix we provide some of the details of
1825: the stability analysis of the pure and mixed modes, presented in
1826: Sec.\ref{sec:side-band}.
1827:
1828: \subsection{Pure modes}
1829:
1830: The perturbed pure mode $S_1$ is written as
1831: \begin{eqnarray}
1832: &&A_1= R_1e^{ikX}\left(1 +a_1^+(T)e^{iQX}+a_1^-(T)e^{-iQX}\right)\cierto{ansta1S1} \\
1833: &&A_2= e^{ik\frac {n} {m}X}\left(a_2^+(T)e^{iQX}+a_2^-(T)e^{-iQX}\right).\cierto{ansta2S1}
1834: \end{eqnarray}
1835: Inserting this ansatz in (\ref{2},\ref{3}) and linearizing in $a_1^\pm$ and $a_2^\pm$ we obtain
1836: \begin{eqnarray}
1837: \dot a_1^\pm& =&\left(\alpha - sR_1^2-\delta Q^2\mp (\gamma+2\delta k) Q\right)a_1^\pm-
1838: sR_1^2a_1^\mp,\cierto{eqa1S1}\\
1839: \dot a_2^\pm&=&\left(\beta -\rho' R_1^2-\delta' Q^2\mp 2\delta' Q(n k/m)\right)a_2^\pm
1840: +\left\{\nu' R_1^n a_2^\mp\right\}_{m=2},\cierto{eqa2S1}
1841: \end{eqnarray}
1842: where the bracketed term with subscript $m=2$ is present only if $m=2$, and
1843: $\alpha$ and $\beta$ are defined by Eq.~(\ref{5b}).
1844: The eigenvalues associated with Eq.~(\ref{eqa1S1}) are
1845: \begin{equation}
1846: \lambda_1^\pm =-(\alpha+\delta Q^2)\pm \sqrt{\alpha^2+Q^2(\gamma+2k\delta)^2},
1847: \cierto{eigva1S1}
1848: \end{equation}
1849: and with Eq.~(\ref{eqa2S1}) are
1850: \begin{equation}
1851: \lambda_2^\pm =(\beta-\rho'\frac {\alpha}{s}-\delta'Q^2)\pm \sqrt{4\delta'^2\left(k
1852: \frac {n}{m}\right)^2 Q^2+\left\{\nu'^2\left(\frac
1853: {\alpha}{s}\right)\right\}_{m=2}}.
1854: \cierto{eigva2S1}
1855: \end{equation}
1856: Eqs.(\ref{eigva1S1}) and (\ref{eigva2S1}) are related to longwave and shortwave
1857: instabilities, respectively. Note that for $m\geq 3$ the perturbation amplitudes
1858: $a_2^+$ and $a_2^-$ decouple and the destabilizing mode is composed of just one
1859: wave of the form $(a_2^+, 0)$ or $(0 ,a_2^-)$. In contrast, due to
1860: the resonant term when $m=2$, the eigevectors have (in principle)
1861: components in both directions and the destabilizing mode is composed of two
1862: waves with different wavenumbers.
1863:
1864: For the pure mode $S_2$ we have
1865: \begin{eqnarray}
1866: A_1&=& e^{ikX}\left(a_1^+(T)e^{iQX}+a_1^-(T)e^{-iQX}\right)\cierto{ansta1S3} \\
1867: A_2&=&R_2 \,e^{ik\frac {n} {m}X}\left(1+a_2^+(T)e^{iQX}+a_2^-(T)e^{-iQX}\right),\cierto{ansta2S2}
1868: \end{eqnarray}
1869: where $a_1^\pm$ and $a_2^\pm$ satisfy
1870: \begin{eqnarray}
1871: \dot a_1^\pm& =&\left(\alpha - \rho R_2^2-\delta Q^2\mp (\gamma+2\delta k) Q\right)a_1^\pm,\cierto{eqa1S2}\\
1872: \dot a_2^\pm&=&\left(\beta - 2 s' R_2^2-\delta' Q^2 \mp 2 \delta' (n k/m) Q\right)a_2^\pm
1873: -s'R_2^2 a_2^\mp.\cierto{eqa2S2}
1874: \end{eqnarray}
1875: The stability of $S_2$ is determined by two eigenvalues:
1876: \begin{eqnarray}
1877: \lambda_1^\pm & =&(\alpha -\rho \frac{\beta}{s'}-\delta Q^2 )\pm |Q||\gamma+2k\delta |,\nonumber \\[0.5cm]
1878: \lambda_2^\pm &=&-(\beta+\delta' Q^2)\pm \sqrt{\beta^2+4\delta'^2\left(\frac{n}{m}k\right)^2Q^2},
1879: \cierto{eigvS1}
1880: \end{eqnarray}
1881: which are associated with the shortwave and longwave instabilities, respectively.
1882:
1883: \subsection{Mixed modes}
1884:
1885: In this case we consider the system (\ref{6a})-(\ref{9a}) for the
1886: evolution of the real amplitudes $R_j>0$ and phases $\phi_j$ (for
1887: $j=1,2$). Thus (abusing notation) we write
1888:
1889: \begin{eqnarray}
1890: &&R_1(X,T)= R_1+r_1(X,T),\quad \phi_1(X,T)=kX+\varphi_1(X,T)\cierto{ansta1M }\\
1891: &&R_2(X,T)= R_2+r_2(X,T),\quad \phi_2(X,T)=\frac{n}{m}kX+\varphi_2(X,T),\cierto{ansta2M}
1892: \end{eqnarray}
1893: where
1894:
1895: \begin{eqnarray*}
1896: r_j &=& \left[r_j^+(T)-ir_j^-(T)\right]e^{iQX}+\left[r_j^+(T)+ir_j^-(T)\right]e^{-i
1897: Q X }
1898: \\[0.5cm]
1899: \varphi_j &=& \left[\varphi_j^+(T)-i\varphi_j^-(T)\right]e^{iQX}+\left[\varphi_j^+(
1900: T) + i \varphi_ j^-( T)\right]e^{-iQX},
1901: \end{eqnarray*}
1902: for $j=1,2$. The
1903: linearized problem
1904: for the perturbations $r_j^\pm$ and $\varphi_j^\pm$ is given
1905: by
1906:
1907: \begin{eqnarray}
1908: \left( \begin{array}{c}
1909: \dot {r_{1}^\pm}\\
1910: \dot{\varphi _{1}^\mp}\\
1911: \dot{r_{2}^\pm}\\
1912: \dot{\varphi_{2}^\mp}
1913: \end{array}\right) = \left( \begin{array}{cccc}
1914: a_1 & -2\delta Q(k+\gamma) & a_2 &0\\
1915: -2\delta Q (k+\gamma) & b_1 & 0 & b_2\\
1916: c_2 & 0 & c_1 & -2\delta' \frac{n k}{m}\\
1917: 0 & d_2 & -2\delta'\frac{n k} {m} & d_1
1918: \end{array}\right)
1919: \left( \begin{array}{c}
1920: {r_{1}^\pm}\\
1921: {\varphi _{1} ^\mp}\\
1922: {r_{2}^\pm}\\
1923: {\varphi_{2}^\mp}
1924: \end{array}\right),
1925: \cierto{a1}
1926: \end{eqnarray}
1927: where
1928: \begin{eqnarray*}
1929: a_1 &=& \mu-\delta k^2-\gamma k-\delta Q^2-(3s R_1^2+\rho R_2^2)+\nu (n-1) R_1^{n-2}R_2^m\cos(n\hat \phi_1-m\hat\phi_2),\\[0.5cm]
1930: a_2 &=& -2\rho R_1 R_2+\nu m R_1^{n-1} R_2^{m-1},\\[0.5cm]
1931: b_1&=&-\nu n R_1^{n-2}R_2^{m}-\delta Q^2,\quad b_2=\nu m R_1^{n-1}R_2^{m-1},\\[0.5cm]
1932: c_1&=&\mu-\Delta\mu-\delta' (kn/m)^2-\delta' Q^2-(3s' R_2^2+\rho' R_1^2)+\nu' (m-1) R_2^{m-2}R_1^n\cos(n\hat \phi_1-m\hat\phi_2),\\[0.5cm]
1933: c_2&=&-2\rho' R_1 R_2+\nu' n R_1^{n-1} R_2^{m-1},\\[0.5cm]
1934: d_1&=&-\nu'mR_1^{n}R_2^{m-2}-\delta' Q^2,\quad d_2=\nu' n R_1^{n-1}R_2^{m-1}.
1935: \end{eqnarray*}
1936:
1937: \section{Derivation of (\ref{16a})}
1938:
1939: To derive (\ref{16a}) we first consider the characteristic
1940: equation in the form
1941: \begin{align}
1942: \lambda^4+G_3 \lambda^3+G_2\lambda^2 +G_1\lambda+G_0=0,
1943: \cierto{14a}
1944: \end{align}
1945: where $G_j$ ($j=0,\ldots,3$) are polynomials in $Q^2$
1946: whose coefficients depend on the parameters directly and also through
1947: the amplitudes of the mixed modes. When the bifurcation is steady and of shortwave type
1948: the location of $\Gamma_{\!+}$
1949: and the wavenumber $Q\neq 0$
1950: are obtained as solutions of the system $G_0= \partial G_0/\partial
1951: Q^2 =0$, which yields
1952: \begin{equation}
1953: (g_1g_2)^2-4g_0g_2^3 - 4g_1^3-18g_0g_1g_2=0\cierto{14b}
1954: \end{equation}
1955: for the location of the lower part of $\Gamma_{\!+}$ and
1956: \begin{equation}
1957: Q^2=(g_1g_2-9g_0)/(6g_1-2g_2^2)\cierto{14c}
1958: \end{equation}
1959: for the perturbation wavenumber of the most unstable perturbation.
1960: Here the funtions $g_0$, $g_1$ and $g_2$ can be written as
1961: \begin{eqnarray*}
1962: g_0 &=&-(a_1c_1-a_2c_2)(\delta'b_1-\delta d_1)-\gamma ^2c_1d_1,\\
1963: g_1 &=& \delta\delta' (a_1c_1-a_2c_2+a_1d_1+c_1b_1)+\delta^2c_1d_1+
1964: {\delta'}^2a_1b_1+\gamma^2\delta'(c_1+d_1),\\
1965: g_2 &=& -\delta^2\delta'(c_1+d_1)-{\delta'}^2\delta(a_1+b_1)-\gamma^2{\delta'}^2,
1966: \end{eqnarray*}
1967: where $a_j$, $b_j$, $c_j$ and $d_j$ ($j=1,2$) are defined in Appendix A.
1968:
1969:
1970: To proceed, we consider the case $k=0$
1971: and make use of the fact that for $\gamma=0$ the
1972: stability limit $\Gamma^+$ coincides with $L_2$ and remains
1973: close for $\gamma \ll 1$. We therefore expand the value of $\mu$
1974: on the stability limit,
1975: \begin{equation}
1976: \mu=\mu_c+\eta^2,\quad\text{with}\quad\eta=c_1\gamma+c_2\gamma^2+\ldots,
1977: \quad |\gamma|\ll 1,
1978: \cierto{15app}
1979: \end{equation}
1980: where $\mu_c=\rho\Delta\mu /(s'-\rho)$ is the value of $\mu$ on
1981: the curve $L_2$ at $k=0$. The small parameter $\eta$ is a
1982: function of $\gamma$ and must vanish when $\gamma=0$. Moreover
1983: the amplitudes of the mixed mode $S_+$ on $\Gamma_{\!+}$ may be
1984: expressed as \begin{align} &R_1=\eta r_{11}+ \eta^2 r_{12}+
1985: \cdots,\qquad R_2=r_{20}+ \eta^2 r_{22}+\eta^3
1986: r_{23}\cdots.\cierto{15c} \end{align} where
1987: $r_{20}=\sqrt{\Delta\mu/(s'-\rho)}$ is the amplitude of the pure
1988: mode on the line $L_2$. The coefficients $r_{ij}$ are
1989: determined by substituting the expansions~(\ref{15c}) into
1990: (\ref{6}) and solving at successive orders. The number of
1991: coefficients $r_{ij}$ that must be kept depends on the
1992: particular resonance. In fact the expansion (\ref{15c}) must be
1993: carried out to $O(\eta^{n-2})$ for an $m:n$ resonance. In order
1994: to calculate the $c_j$ we substitute (\ref{15c}) into
1995: (\ref{14b}) and solve order by order. Finally, the wave number
1996: $Q$ is obtained by substituting these results into (\ref{14c}).
1997:
1998: We now sketch the necessary calculations for the case
1999: $m:n=2:5$ at $k=0$. Inserting (\ref{15app}) and (\ref{15c})
2000: into (\ref{6}) yields
2001:
2002: \begin{equation}
2003: \left.\begin{array}{r}
2004: sr_{11}^2+2\rho r_{20} r_{22}=1,\\
2005: 2s'r_{20} r_{22}+\rho' r_{11}^2=1,
2006: \end{array}\right\} \quad \Longrightarrow \quad
2007: \begin{array}{l}
2008: r_{11}^2=(s'-\rho)/(ss'-\rho\rho'),\\
2009: r_{22}=(s-\rho')/2r_{20}(ss'-\rho\rho')
2010: \end{array}
2011: \end{equation}
2012: at $O(\eta^2)$, while at $O(\eta^3)$ we have
2013: \begin{equation}
2014: \left.\begin{array}{r}
2015: 2sr_{11}r_{12}+2\rho r_{20} r_{23}=-\nu r_{20}^mr_{11}^3,\\
2016: s' r_{20} r_{23}+\rho' r_{11}r_{12}=0
2017: \end{array}\right\} \quad \Longrightarrow \quad
2018: \begin{array}{l}
2019: r_{12}=-\nu s' r_{11}^2 r_{20}^m/ 2(ss'-\rho\rho'),\\
2020: r_{23}=\nu\rho' r_{11}^3 r_{20}^{m-1}/ 2(ss'-\rho\rho').
2021: \end{array}
2022: \end{equation}
2023: If $n>5$ we would have found $r_{12}=r_{23}=0$ and it would be necessary to
2024: consider higher order corrections. Near the curve $L_2$
2025: (i.e., near the point $(\mu,k)=(\mu_c,0)$) (see Fig.~\ref{fig:dgrm2n5withtw(a)})
2026: the functions $g_j$, for $j=0,1,2$, may be written as
2027: \begin{align}
2028: g_0=\;&\eta^n \left(4 \delta' \nu (ss'-\rho\rho')r_{11}^n r_{20}^{m+2}\right)+
2029: O(\eta^{n+1}),\nonumber\\
2030: g_1=\;&\eta^2\left(4
2031: \delta\delta'(ss'-\rho\rho')r_{11}^2r_{20}^2\right)+\eta^{n-2}(2s'n+\Delta\mu(n-
2032: 1 ) r _ {20} ^{m})\delta\delta' \nu a_1^{n-2}\cierto{16}\\
2033: &+\gamma^2 \delta'(-2s'r_{20}^2-\eta^2 4s'r_{20}r_{22}-\eta^2
2034: 4s'r_{20}r_{23})+O(\eta^n+ \gamma^2\eta^{n-2})\nonumber\\
2035: g_2=\;&\delta^2\delta'(2 s' r_{20}^2)+ O(\eta^2),\nonumber\end{align} where only
2036: those terms necessary to compute $Q$ have been kept in (\ref{16}). Finally
2037: composing the expansions~(\ref{16}) and (\ref{15app}) and substituting the result
2038: into (\ref{14c}) we find (after some algebra) the result (\ref{16a}).
2039:
2040:
2041: %\bibliography{journal}
2042:
2043: \begin{thebibliography}{10}
2044:
2045: \bibitem{ArFi98}
2046: {\sc H.~Arbell and J.~Fineberg}, {\em The spatial and temporal dynamics of two
2047: interacting modes in parametrically driven surface waves}, Phys. Rev. Lett.,
2048: 81 (1998), p.~4384.
2049:
2050: \bibitem{ArFi02}
2051: \leavevmode\vrule height 2pt depth -1.6pt width 23pt, {\em Pattern formation in
2052: two-frequency forced parametric waves}, Phys. Rev. E, 65 (2002), p.~036224.
2053:
2054: \bibitem{ArGu88}
2055: {\sc D.~Armbruster, J.~Guckenheimer, and P.~Holmes}, {\em Heteroclinic cycles
2056: and modulated travelling waves in systems with o(2) symmetry}, Physica D, 29
2057: (1988), p.~257.
2058:
2059: \bibitem{BaMa94}
2060: {\sc A.~Bayliss, B.~Matkowsky, and H.~Riecke}, {\em Structure and dynamics of
2061: modulated traveling waves in cellular flames}, Physica D, 74 (1994),
2062: pp.~1--23.
2063:
2064: \bibitem{BeEd96}
2065: {\sc T.~Besson, W.~Edwards, and L.~Tuckerman}, {\em Two-frequency parametric
2066: excitation of surface waves}, Phys. Rev. E, 54 (1996), p.~507.
2067:
2068: \bibitem{CaCa92}
2069: {\sc B.~Caroli, C.~Caroli, and S.~Fauve}, {\em On the phenomenology of tilted
2070: domains in lamellar eutectic growth}, J. Phys. I (Paris), 2 (1992),
2071: pp.~281--290.
2072:
2073: \bibitem{Co86}
2074: {\sc P.~Coullet}, {\em Commensurate-incommensurate transition in nonequilibrium
2075: system}, Phys. Rev. Lett., 56 (1986), p.~724.
2076:
2077: \bibitem{CoEl87}
2078: {\sc P.~Coullet, C.~Elphick, and D.~Repaux}, {\em Nature of spatial chaos},
2079: Phys. Rev. Lett., 58 (1987), pp.~431--434.
2080:
2081: \bibitem{CoGo89}
2082: {\sc P.~Coullet, R.~Goldstein, and G.~Gunaratne}, {\em Parity-breaking
2083: transitions of modulated patterns in hydrodynamic systems}, Phys.~Rev.~Lett.,
2084: 63 (1989), p.~1954.
2085:
2086: \bibitem{CoHu86}
2087: {\sc P.~Coullet and P.~Huerre}, {\em Resonance and phase solitons in
2088: spatially-forced thermal convection}, Physica D, 23 (1986), p.~27.
2089:
2090: \bibitem{CrKn90}
2091: {\sc J.~D. Crawford, E.~Knobloch, and H.~Riecke}, {\em Period-doubling mode
2092: interactions with circular symmetry}, Physica D, 44 (1990), pp.~340--396.
2093:
2094: \bibitem{CuEl87}
2095: {\sc E.~M. Curado and C.~Elphick}, {\em Effects of an almost resonant spatial
2096: thermal modulation in the {Rayleigh}-{B\'enard} problem - quasi-periodic
2097: behavior}, J. Phys. A, 20 (1987), pp.~1205--1214.
2098:
2099: \bibitem{Da86}
2100: {\sc G.~Dangelmayr}, {\em Steady-state mode interactions in the presence of
2101: {O(2)}-symmetry}, Dyn. Stab. Systems, 1 (1986), p.~159.
2102:
2103: \bibitem{DaPo03}
2104: {\sc J.~H. Dawes, C.~M. Postlethwaite, and M.~R. Proctor}, {\em Instabilities
2105: induced by a weak breaking of a strong spatial resonance}, Physica D,
2106: (submitted).
2107:
2108: \bibitem{DoCa86}
2109: {\sc M.~Dominguez-Lerma, D.~Cannell, and G.~Ahlers}, {\em Eckhaus boundary and
2110: wave-number selection in rotating {Couette}-{Taylor} flow}, Phys. Rev. A, 34
2111: (1986), p.~4956.
2112:
2113: \bibitem{EdFa94}
2114: {\sc W.~S. Edwards and S.~Fauve}, {\em Patterns and quasi-patterns in the
2115: faraday experiment}, J. Fluid Mech., 278 (1994), pp.~123--148.
2116:
2117: \bibitem{FaCh89}
2118: {\sc G.~Faivre, S.~de~Cheveigne, C.~Guthmann, and P.~Kurowski}, {\em Solitary
2119: tilt waves in thin lamellar eutectics}, Europhys. Lett., 9 (1989), p.~779.
2120:
2121: \bibitem{FaDo90}
2122: {\sc S.~Fauve, S.~Douady, and O.~Thual}, {\em Comment on "{Parity}-breaking
2123: transitions of modulated patterns in hydrodynamic systems"}, Phys. Rev.
2124: Lett., 65 (1990), p.~385.
2125:
2126: \bibitem{FeSe89}
2127: {\sc Z.~C. Feng and P.~R. Sethna}, {\em Symmetry-breaking bifurcations in
2128: resonant surface waves}, J. Fluid Mech., 199 (1989), p.~495.
2129:
2130: \bibitem{FoRa94}
2131: {\sc L.~Fourtune, W.~J. Rappel, and M.~Rabaud}, {\em Phase dynamics near a
2132: parity-breaking instability}, Phys. Rev. E, 49 (1994), pp.~R3576--R3579.
2133:
2134: \bibitem{GoGu91}
2135: {\sc R.~Goldstein, G.~Gunaratne, L.~Gil, and P.~Coullet}, {\em Hydrodynamic and
2136: interfacial patterns with broken space-time symmetry}, Phys Rev. A, 43
2137: (1991), p.~6700.
2138:
2139: \bibitem{HiRiunpub}
2140: {\sc M.~Higuera, H.~Riecke, and M.~Silber}, (unpublished).
2141:
2142: \bibitem{HiVe02}
2143: {\sc M.~Higuera, J.~M. Vega, and E.~Knobloch}, {\em Coupled amplitude-streaming
2144: flow equations for nearly inviscid {Faraday} waves in small aspect raito
2145: containers}, J. Nonlinear Sci., 12 (2002), p.~505.
2146:
2147: \bibitem{JoPr87}
2148: {\sc C.~A. Jones and M.~R. Proctor}, {\em Strong spatial resonance and
2149: travelling waves in {B\'enard} convection}, Phys. Lett., 121 (1987), p.~224.
2150:
2151: \bibitem{KuPi98}
2152: {\sc A.~Kudrolli, B.~Pier, and J.~Gollub}, {\em Superlattice patterns in
2153: surface waves}, Physica D, 123 (1998), p.~99.
2154:
2155: \bibitem{LeRa91}
2156: {\sc H.~Levine, W.-J. Rappel, and H.~Riecke}, {\em Resonant interactions and
2157: traveling solidification cells}, Phys. Rev. A, 43 (1991), p.~1122.
2158:
2159: \bibitem{LoGo83}
2160: {\sc M.~Lowe, J.~P. Gollub, and T.~C. Lubensky}, {\em Commensurate and
2161: incommensurate structures in a nonequilibrium system}, Phys. Rev. Lett., 51
2162: (1983), p.~786.
2163:
2164: \bibitem{MaTr84}
2165: {\sc B.~A. Malomed and M.~I. Tribelsky}, {\em Bifurcations in distributed
2166: kinetic systems with aperiodic instability}, Physica D, 14 (1984), p.~67.
2167:
2168: \bibitem{MePr86}
2169: {\sc E.~Meron and I.~Procaccia}, {\em Low-dimensional chaos in surface waves:
2170: Theoretical analysis of an experiment}, Phys. Rev. A, 34 (1986), p.~3221.
2171:
2172: \bibitem{MoKn98}
2173: {\sc J.~Moehlis and E.~Knobloch}, {\em Forced symmetry breaking as a mechanism
2174: for bursting}, Phys. Rev. Lett., 80 (1998), pp.~5329--5332.
2175:
2176: \bibitem{OgZi96}
2177: {\sc A.~Ogawa, W.~Zimmermann, K.~Kawasaki, and T.~Kawakatsu}, {\em Forced
2178: periodic and quasi-periodic patterns in anisotropic systems.}, J. Phys. II
2179: (Paris), 6 (1996), p.~305.
2180:
2181: \bibitem{PaRa95}
2182: {\sc E.~Pampaloni, P.~L. Ramazza, S.~Residori, and F.~T. Arecchi}, {\em
2183: Two-dimensional crystals and quasi-crystals in nonlinear optics}, Phys. Rev.
2184: Lett., 74 (1995), p.~258.
2185:
2186: \bibitem{PoKn00}
2187: {\sc J.~Porter and E.~Knobloch}, {\em Complex dynamics in the 1 : 3 spatial
2188: resonance}, Physica D, 143 (2000), pp.~138--168.
2189:
2190: \bibitem{PoKn01}
2191: \leavevmode\vrule height 2pt depth -1.6pt width 23pt, {\em New type of complex
2192: dynamics in the 1 : 2 spatial resonance}, Physica D, 159 (2001),
2193: pp.~125--154.
2194:
2195: \bibitem{PoSi02}
2196: {\sc J.~Porter and M.~Silber}, {\em Broken symmetries and pattern formation in
2197: two-frequency forced {Faraday} waves}, Phys. Rev. Lett., 89 (2002),
2198: p.~084501.
2199:
2200: \bibitem{PrJo88}
2201: {\sc M.~Proctor and C.~Jones}, {\em The interaction of two spatially resonant
2202: patterns in thermal convection. part 1. exact 1:2 resonance}, J. Fluid Mech.,
2203: 188 (1988), pp.~301--335.
2204:
2205: \bibitem{RaMi90}
2206: {\sc M.~Rabaud, S.~Michalland, and Y.~Couder}, {\em Dynamical regimes of
2207: directional viscous fingering: Spatiotemporal chaos and wave propagation},
2208: Phys. Rev. Lett., 64 (1990), p.~184.
2209:
2210: \bibitem{RaRi92}
2211: {\sc W.-J. Rappel and H.~Riecke}, {\em Parity-breaking in directional
2212: solidification: numerics versus amplitude equations}, Phys. Rev. A, 45
2213: (1992), p.~846.
2214:
2215: \bibitem{RiPa86}
2216: {\sc H.~Riecke and H.-G. Paap}, {\em Stability and wave-vector restriction of
2217: axisymmetric {Taylor} vortex flow}, Phys. Rev. A, 33 (1986), p.~547.
2218:
2219: \bibitem{RiPa92}
2220: \leavevmode\vrule height 2pt depth -1.6pt width 23pt, {\em Parity-breaking and
2221: {Hopf} bifurcation in axisymmetric {Taylor} vortex flow}, Phys. Rev. A, 45
2222: (1992), p.~8605.
2223:
2224: \bibitem{RoSc00a}
2225: {\sc J.~Rogers, M.~Schatz, O.~Brausch, and W.~Pesch}, {\em Superlattice
2226: patterns in vertically oscillated {Rayleigh}-{B\'enard} convection}, Phys.
2227: Rev. Lett., 85 (2000), p.~4281.
2228:
2229: \bibitem{RuRu03}
2230: {\sc A.~M. Rucklidge and W.~J. Rucklidge}, {\em Convergence properties of the
2231: 8, 10, and 12 mode representations of quasipatterns}, Physica D, 178 (2003),
2232: p.~62.
2233:
2234: \bibitem{SiBe88}
2235: {\sc A.~Simon, J.~Bechhoefer, and A.~Libchaber}, {\em Solitary modes and the
2236: {Eckhaus} instability in directional solidification}, Phys. Rev. Lett., 61
2237: (1988), p.~2574.
2238:
2239: \bibitem{SiGo89}
2240: {\sc F.~Simonelli and J.~P. Gollub}, {\em Surface-wave mode interactions -
2241: effects of symmetry and degeneracy}, J. Fluid Mech., 199 (1989), p.~471.
2242:
2243: \bibitem{Um91}
2244: {\sc M.~Umeki}, {\em Faraday resonance in rectangular geometry}, J. Fluid
2245: Mech., 227 (1991), p.~161.
2246:
2247: \bibitem{WiMc92}
2248: {\sc R.~Wiener and D.~McAlister}, {\em Parity-breaking and solitary waves in
2249: axisymmetric {Taylor} vortex flow}, Phys. Rev. Lett., 69 (1992), p.~2915.
2250:
2251: \end{thebibliography}
2252:
2253: \end{document}
2254: