1:
2: % Spatial period-multiplying instabilities of hexagonal Faraday waves
3: % by Tse, Rucklidge, Hoyle & Silber
4: % submitted to Physica D 14 July 1999
5: % revised 23 May 2000
6:
7: % former title:
8: % Secondary instabilities of standing hexagons in a Faraday wave experiment
9:
10: % Upper-case A B C D E F G H I J K L M N O P Q R S T U V W X Y Z
11: % Lower-case a b c d e f g h i j k l m n o p q r s t u v w x y z
12: % Digits 0 1 2 3 4 5 6 7 8 9
13: % Exclamation ! Double quote " Hash (number) #
14: % Dollar $ Percent % Ampersand &
15: % Acute accent ' Left paren ( Right paren )
16: % Asterisk * Plus + Comma ,
17: % Minus - Point . Solidus /
18: % Colon : Semicolon ; Less than <
19: % Equals = Greater than > Question mark ?
20: % At @ Left bracket [ Backslash \
21: % Right bracket ] Circumflex ^ Underscore _
22: % Grave accent ` Left brace { Vertical bar |
23: % Right brace } Tilde ~
24:
25: \documentclass[12pt]{elsart}
26:
27: \setlength{\textheight}{9.0in}
28: \setlength{\textwidth}{6.2in}
29: \setlength{\oddsidemargin}{-0.2in} % real is (8.2-6)/2 + 1.2
30: %\renewcommand{\baselinestretch}{1.2}
31: \parindent 0pt
32:
33: %%%%%%%%%%%%%%%%%%%%%%%%%% use packages %%%%%%%%%%%%%%%%%%%%%%%
34:
35: \usepackage[thinlines,thiklines]{easytable}
36: \usepackage[thinlines,thiklines]{easymat}
37: %\usepackage{fancyhdr}
38: %\usepackage{lastpage}
39: \usepackage{array}
40: \usepackage{amsmath,amsthm}
41: \usepackage{amsfonts}
42: \usepackage[dvips]{graphicx}
43: \usepackage{epsfig}
44: \usepackage{amssymb}
45: \usepackage{latexsym}
46: \usepackage{multirow}
47: %\usepackage[dark,none,bottom]{draftcopy}
48: %\usepackage{eepic}
49: %\usepackage{amstex}
50: %\usepackage{bibsty}
51: %\usepackage{tabularx}
52: %\usepackage{smallcapt}
53:
54:
55: %%%%%%%%%%%%%%%%%%%%%%%%%% new definitions %%%%%%%%%%%%%%%%%%%%%%%
56:
57: \def\kerL{\mbox{ker\/$L$}}
58: \def\fixsp{\mbox{Fix$(\Sigma)\/$}}
59: \def\fixspgam{\mbox{Fix$(\Gamma)\/$}}
60:
61: \def\etal{et al.}
62:
63: \include{macros}
64:
65:
66:
67: \begin{document}
68: \begin{frontmatter}
69:
70:
71: \title{Spatial period-multiplying instabilities of hexagonal Faraday waves}
72:
73:
74: \author{D.P.~Tse\thanksref{DAWNADDRESS}},
75: \author{A.M.~Rucklidge\thanksref{ALASTAIREMAIL}},
76: \author{R.B.~Hoyle}
77: \address{Department of Applied Mathematics and Theoretical Physics,\\
78: University of Cambridge, Silver Street, Cambridge CB3 9EW, UK}
79: \and
80: \author{M.~Silber\thanksref{MARYEMAIL}}
81: \address{Department of Engineering Sciences and Applied Mathematics,\\
82: Northwestern University, Evanston, IL 60208 USA}
83:
84: \thanks[DAWNADDRESS]{Current address: Warburg Dillon Read, 100 Liverpool
85: Street, London EC2M 2RH, UK}
86: \thanks[ALASTAIREMAIL]{E-mail: A.M.Rucklidge@damtp.cam.ac.uk}
87: \thanks[MARYEMAIL]{E-mail: m-silber@northwestern.edu}
88:
89:
90: \begin{abstract}
91: A recent Faraday wave experiment with two-frequency forcing reports two types
92: of `superlattice' patterns that display periodic spatial structures having two
93: separate scales~\cite{KudrPier98}. These patterns both arise as secondary
94: states once the primary hexagonal pattern becomes unstable. In one of these
95: patterns (so-called `superlattice-two') the original hexagonal symmetry is
96: broken in a subharmonic instability to form a striped pattern with a spatial
97: scale increased by a factor of $2\sqrt{3}$ from the original scale of the
98: hexagons. In contrast, the time-averaged pattern is periodic on a hexagonal
99: lattice with an intermediate spatial scale ($\sqrt{3}$~larger than the original
100: scale) and apparently has $60^{\circ}$~rotation symmetry. We present a
101: symmetry-based approach to the analysis of this bifurcation. Taking as our
102: starting point only the observed instantaneous symmetry of the superlattice-two
103: pattern presented in~\cite{KudrPier98} and the subharmonic nature of the
104: secondary instability, we show (a)~that a pattern with the same instantaneous
105: symmetries as the superlattice-two pattern can bifurcate stably from standing
106: hexagons; (b)~that the pattern has a spatio-temporal symmetry not reported
107: in~\cite{KudrPier98}; and (c)~that this spatio-temporal symmetry accounts for
108: the intermediate spatial scale and hexagonal periodicity of the time-averaged
109: pattern, but not for the apparent $60^{\circ}$~rotation symmetry. The approach
110: is based on general techniques that are readily applied to other secondary
111: instabilities of symmetric patterns, and does not rely on the primary pattern
112: having small amplitude.
113: \end{abstract}
114:
115: \begin{keyword}
116: 02.20.Df 47.20.Ky 47.20.Lz 47.54.+r\newline
117: Faraday waves; secondary instabilities; spatial period-multiplying;
118: superlattice patterns; averaged symmetries of attractors
119: \end{keyword}
120:
121: \end{frontmatter}
122:
123: \section{Introduction}\label{sec:intro}
124:
125: The classical hydrodynamic problem of parametrically driven surface waves -- or
126: Faraday waves -- concerns the spontaneous generation of standing waves at the
127: free surface of a horizontal layer of fluid when subjected to vertical
128: oscillations whose amplitude exceeds a critical value. Its usefulness as a tool
129: to study nonlinear pattern-forming dynamics in non-equilibrium systems is
130: reflected in the considerable amount of interest shown in the subject by
131: experimentalists and theoreticians alike. A review of earlier works, mostly
132: conducted with low-viscosity fluids in small vessels and a single forcing
133: frequency, can be found in~\cite{MilesH90}. More recently, Edwards \&
134: Fauve~\cite{EdwaFauv94} have performed experiments in the small-depth,
135: high-viscosity and large-aspect ratio regime using a forcing function with two
136: commensurate frequency components that modulates gravity periodically. In this
137: regime, where it can be shown that the wavenumber of the selected pattern is
138: less sensitive to the size and shape of the container, and that long-wavelength
139: modes are heavily damped, observations of spatially periodic patterns (stripes,
140: squares, hexagons), circular patterns (targets and spirals) and quasi-patterns
141: have been reported~\cite{EdwaFauv94,KudrolliG96}. A survey of more recent
142: results has been carried out by M\"{u}ller \etal~\cite{MullerFP98}. Over the
143: past two years, a new class of `superlattice patterns' has been independently
144: observed by Kudrolli \etal~\cite{KudrPier98} and Arbell \&
145: Fineberg~\cite{ArbellF98} in experiments employing two-frequency forcing
146: functions, and by Wagner \etal~\cite{WagnerMK98} in experiments using
147: non-Newtonian fluids. These superlattices are so termed because of their
148: distinctive feature of having spatial structures on two different length scales
149: when viewed at any instant in time~\cite{KudrPier98}. Steady patterns that
150: display similar characteristics have also been observed in convection
151: experiments on fluids with temperature-dependent viscosity~\cite{White88} and
152: have been investigated in a model of long wave convection~\cite{SkeldonS98} and
153: in reaction-diffusion systems near a Turing bifurcation~\cite{JuddS98}.
154:
155: Two types of superlattice patterns have been reported in~\cite{KudrPier98} for
156: different parameter values. Both of them, despite their different spatial and
157: spatio-temporal symmetry properties, are found to be possible transitions from
158: harmonic standing hexagons as the forcing amplitude is increased. (In this
159: context, {\em harmonic} indicates an oscillation with the same period as that
160: of the external forcing, denoted by~$T$, while {\em subharmonic} indicates an
161: oscillation with twice that period.) The first of these patterns (called
162: `superlattice-one' by Kudrolli \etal~\cite{KudrPier98}) is a harmonic response
163: with triangular symmetry on a small scale and hexagonal lattice periodicity on
164: a larger scale. This pattern has been studied by Silber \&
165: Proctor~\cite{SilberP98}, who showed that it (along with standing hexagons) can
166: arise in a bifurcation from the flat, undisturbed state when a hexagonal
167: lattice with spatial periodicity larger than that dictated by the critical
168: wavelength is considered~\cite{DionneG92,DionSkel97}. Silber \&
169: Proctor~\cite{SilberP98} also suggested that stability might be transfered from
170: standing hexagons to superlattice-one through an intermediate branch.
171:
172: \begin{center}
173: \begin{figure}[htbp]
174: \hfil (a) \hfil (b) \hfil (c) \\
175: \smallskip
176: {\resizebox{!}{6.26truecm}{\includegraphics{fig1.eps}}}\\
177: \caption{\sl \small (a) Time-averaged image of the superlattice-two pattern
178: displays a well-defined hexagonal symmetry on two spatial scales. (b) \& (c)
179: Instantaneous snapshots of the same pattern separated by $3/20$th of the
180: external driving period $T$ reveals a time-dependent stripe-like modulation.
181: The pattern in (a) is a different realization of the experiment from those in
182: (b) \&~(c). Courtesy of Kudrolli \etal~\cite{KudrPier98}, reproduced with
183: permission.}
184: \label{fig:original}
185: \end{figure}
186: \end{center}
187:
188: The second type of superlattice pattern (`superlattice-two'~\cite{KudrPier98}),
189: in contrast to the first, arises in a period-doubling (or subharmonic)
190: instability of the standing hexagons. If we let $u(\xv,t)$ measure the
191: deformation of the free fluid surface at time~$t$, it satisfies
192: %
193: \begin{equation}\label{pdouble}
194: u(\xv,t+2T) = u(\xv,t), \quad
195: u(\xv,t+T) \neq u(\xv,t).
196: \end{equation}
197: %
198: Further, this pattern exhibits a complicated mixture of spatial symmetry and
199: time-dependent behaviour. When averaged over two periods of the driving
200: function, its image displays hexagonal symmetry with two well-defined spatial
201: scales in the ratio $1:\sqrt{3}$ (\mbox{figure~\ref{fig:original}(a)}).
202: Remarkably, at any instant, a wavy, stripe-like spatial modulation destroys the
203: average hexagonal symmetry, resulting in a pattern that appears vastly
204: different from its time-averaged image (see figures~\ref{fig:original}(b)
205: and~\ref{fig:original}(c)). Arbell \& Fineberg (unpublished) have also found
206: the superlattice-two pattern for similar experimental parameters.
207:
208: The superlattice-two pattern presents a number of theoretical challenges that
209: motivate this paper: the disappearance of the stripes from the time-averaged
210: pattern; the reduced spatial period in the time-average; and the apparent
211: $60^{\circ}$~rotation symmetry of the time-average. We present a symmetry-based
212: approach to the study of this pattern by taking the view that it arises as a
213: symmetry-breaking instability from the underlying standing hexagons in a
214: spatial period-multiplying bifurcation. Our aim is to classify qualitatively
215: the range of possible bifurcating solutions and to understand how their
216: symmetry properties can be related to the experimental observations described
217: above. We emphasize that we are examining instabilities of fully nonlinear
218: states, so our approach differs from weakly nonlinear studies of the primary
219: Faraday instability.
220:
221: There are three stages in our approach. First, by using the experimentally
222: observed instantaneous spatial symmetry information of the superlattice-two
223: instability and by making the assumption that all solutions are periodic in the
224: plane, we can restrict all patterns to a suitably chosen spatially periodic
225: lattice. This lattice in turn defines a compact symmetry group, which we denote
226: by~$\Gamma^s$, with the key properties that its action leaves standing hexagons
227: invariant and that it has a subgroup, which we denote by~$\Sigma^s$, that
228: describes the {\em instantaneous} spatial symmetry of the bifurcating
229: superlattice pattern. Due to the compactness and special structure
230: of~$\Gamma^s$, we can compute explicitly all its irreducible representations.
231: Second, we observe that since $\Sigma^s$ is by definition the isotropy subgroup
232: of the bifurcating solution under the action of~$\Gamma^s$, it must have a
233: non-trivial fixed-point set. This restriction allows us to identify the one
234: relevant irreducible representation of $\Gamma^s$ that describes the spatial
235: symmetry properties of the marginal eigenfunctions at the superlattice
236: bifurcation point. Finally, by considering the action of the time-shift
237: symmetry $\tau_t: t \rightarrow t + T$ on the period-doubling marginal modes,
238: we obtain the irreducible representation of the full symmetry group (denoted
239: by~$\Gamma$) and hence the normal form of the bifurcation problem. We can then
240: invoke the equivariant branching lemma~\cite{GolubitskySS88} to show that there
241: are at least six primary branches of solutions bifurcating from standing
242: hexagons.
243:
244: With one proviso, the superlattice-two pattern observed by
245: Kudrolli~\etal~\cite{KudrPier98} can be identified as one of these branches,
246: which we show can bifurcate as a stable branch from standing hexagons. By
247: applying techniques for studying the averaged symmetry of periodic orbits
248: (cf~\cite{BaranyDG93}), we show that the time-average of this branch of
249: solutions has the hexagonal lattice periodicity observed in the experiment (as
250: in figure~\ref{fig:original}(a)); this change in the spatial length scale on
251: time-averaging is a consequence of the branch of solutions possessing a
252: spatio-temporal symmetry. This symmetry was not reported in~\cite{KudrPier98}.
253: The proviso mentioned above is that our time-average pattern does not possess
254: $60^{\circ}$~rotation symmetry -- we will return to this discrepancy in the
255: final section.
256:
257: A further stability analysis predicts that other patterns, displaying different
258: spatial and spatio-temporal symmetry properties, can bifurcate as stable
259: branches of solutions from standing hexagons in different regions of parameter
260: space. More generally, our analysis indicates that patterns that display
261: superlattice structures can arise in two-dimensional spatial period-multiplying
262: bifurcations from an underlying non-trivial solution, and our approach could
263: also be used to investigate other superlattice patterns of Arbell \&
264: Fineberg~\cite{ArbellF98} and Wagner \etal~\cite{WagnerMK98}. In particular, it
265: may be possible to analyse some of those experimental results in terms of other
266: irreducible representations of the same group~$\Gamma^s$.
267:
268: The issue of spatial period-multiplying instabilities is an interesting one
269: that has arisen in a variety of experimental and theoretical contexts.
270: Period-multiplying bifurcations in one lateral direction have arisen in
271: convection problems~\cite{McKenzie88}, magnetoconvection~\cite{ProctorWeiss93},
272: Taylor--Couette experiments~\cite{Iooss86} and in numerical solutions of the
273: Kuramoto--Sivashinsky equations~\cite{AstonSW92,AmdjadiAP97}. Much less is
274: known about spatial period-multiplying bifurcations in two directions. There
275: are now several experimental observations of this phenomenon in the Faraday
276: wave problem~\cite{KudrPier98,ArbellF98,WagnerMK98} as well as in convection
277: experiments~\cite{White88,McKenzie88,ProcMatt96} and magnetoconvection
278: calculations~\cite{Dawes2000,RWBMP2000}.
279:
280: In the next section, we introduce some fundamental definitions and results from
281: equivariant bifurcation theory~\cite{GolubitskySS88} to help us describe how
282: this problem can be cast into a theoretical framework. In
283: section~\ref{sec:findgamma}, we fully describe the symmetry group of the
284: bifurcation problem that will give rise to the observed symmetry-breaking
285: behaviour. We also show that, under suitable phenomenological assumptions, we
286: can identify and hence explicitly compute the irreducible representation that
287: is relevant to the action of the symmetry group on the observed bifurcating
288: modes. The normal form of the bifurcation problem and a stability analysis are
289: presented in section~\ref{sec:bifprob}. Discussions of our approach and a
290: comparison with the experiments follow in section~\ref{sec:discuss}.
291:
292: \section{Group theoretic ideas}\label{sec:groupideas}
293:
294: In order to study the superlattice-two pattern as a symmetry-breaking
295: instability from standing hexagons, it is necessary to identify all the
296: symmetries that are initially present. Due to the apparent absence of side-wall
297: effects in the observed patterns, we consider the mathematical idealisation
298: that all physical fields are defined in a laterally unbounded domain. Standing
299: hexagons are then easily seen to be invariant under the action generated by a
300: reflection, a $60^{\circ}$~rotation, and two linearly independent translations.
301: The group generated by these symmetry actions is isomorphic to $\Zset^2
302: \dotplus \deesix$, which is non-compact. (Here $\Zset$ denotes the group of
303: integers under addition, and $\deesix$ is the twelve-element symmetry group of
304: a regular hexagon.) Consequently a bifurcation problem that is equivariant
305: under the action of this group can have an infinite number of modes related by
306: symmetry becoming marginally stable simultaneously. This difficulty can be
307: resolved if we restrict possible solutions to doubly-periodic functions defined
308: on a suitably chosen lattice, an assumption justified by the distinct spatial
309: periodicity of the observed patterns. A suitable lattice, which can be viewed
310: as a finite cell with periodic boundary conditions, is one that captures the
311: spatial periodicity of both the bifurcating modes as well as the standing
312: hexagons. With respect to such a periodic cell, the symmetries that leave
313: standing hexagons invariant now form a finite, and hence compact group that can
314: be studied via representation theory. Our task therefore, is to make use of the
315: available symmetry information taken from experimental observations to choose a
316: lattice on which we can define a suitable spatial symmetry group $\Gamma^s$
317: with the properties outlined in section~\ref{sec:intro}. The idea of
318: suitability can be made precise after we have introduced some basic group
319: theoretic results.
320:
321: Since we are considering bifurcations from a time-periodic solution, we
322: formulate the bifurcation problem of the superlattice-two pattern, a
323: period-doubling instability, by expanding about standing hexagons using a
324: stroboscopic map $\mathcal{G}$ in the manner described by Crawford \&
325: Knobloch~\cite{CrawKnob91} and Silber \& Proctor~\cite{SilberP98}.
326: Specifically, we are assuming that standing hexagons, a fixed point of the
327: map~$\mathcal{G}$, lose stability to subharmonic waves with period $2T$ as a
328: bifurcation parameter $\mu$ is varied past zero. This implies that the
329: linearised map $D\mathcal{G}$ evaluated at the fixed point has a real
330: eigenvalue passing through the value $-1$ as the bifurcating waves become
331: unstable. With the assumption that all fields are defined in a periodic cell
332: such that symmetries of the standing hexagons are described by a compact
333: group~$\Gamma^s$, the linearised map $D\mathcal{G}$ has a finite number ($p$)
334: of marginal eigenfunctions associated with the eigenvalue $-1$ as $\mu$ crosses
335: zero. We denote the amplitudes of these $p$ marginal modes at time $t=qT, \; q
336: \in \Zset$ by $\zv_q = \left[z_1(qT), \ldots , z_p(qT)\right]\in\Rset^p$. In
337: addition, the pattern has two neutrally stable modes (eigenvalues equal to~$1$)
338: associated with translations of the standing hexagons
339: (see~\cite{Iooss86,RuckSilb98}); the amplitudes of these two modes, which
340: correspond to the translation of the pattern in the plane, are denoted
341: by~$\dv_q$. Close to the onset of the period-doubling instability,
342: $\mathcal{G}$~can be reduced to a finite-dimensional map $\gv$ defined on the
343: centre manifold spanned by these amplitudes:
344: %
345: \begin{equation}\label{normalform}
346: \zv_{q+1} = \gv(\zv_q; \mu), \quad
347: \gv: \Rset^p \times \Rset \rightarrow \Rset^p,
348: \end{equation}
349: %
350: coupled with a map $\hv: \Rset^p \times \Rset \rightarrow \Rset^2$ describing
351: how the perturbation drives translations of the pattern:
352: %
353: \begin{equation}\label{drifteqn}
354: \dv_{q+1} = \dv_q + \hv(\zv_q;\mu).
355: \end{equation}
356: %
357: The map $\gv$ is forced by symmetry to be $\Gamma^s$-equivariant:
358: %
359: \begin{equation}\label{gammaequiv}
360: \gamma \gv(\zv_q;\mu) = \gv(\gamma \zv_q;\mu) \quad \mbox{for all}\;
361: \gamma \in \Gamma^s,
362: \end{equation}
363: %
364: while the map $\hv$ obeys
365: %
366: \begin{equation}\label{hequivariant}
367: N_{\gamma}\hv\left(\zv_q;\mu\right) =
368: \hv\left(\gamma \zv_q; \mu\right) \quad \mbox{for all}\;
369: \gamma \in \Gamma^s,
370: \end{equation}
371: %
372: where $N_\gamma$ is the $2\times 2$ matrix that represents how the symmetry
373: $\gamma$ acts on a horizontal displacement vector~\cite{RuckSilb98}. In terms
374: of the marginal modes, standing hexagons correspond to the trivial state,
375: \mbox{$\zv = \mathbf{0}$.} When considered as a symmetry-breaking bifurcation
376: from the underlying standing hexagons, the superlattice pattern corresponds to
377: a non-trivial, period-two solution to the map~$\gv$, denoted
378: by~$\zv^{\ast}_{q}$, whose instantaneous spatial symmetry is specified by its
379: isotropy subgroup $\Sigma_{\zv^{\ast}_{q}}^s$:
380: %
381: \begin{equation}\label{isotropy}
382: \Sigma_{\zv^{\ast}_{q}}^s = \left\{\sigma \in \Gamma^s: \sigma
383: \zv^{\ast}_{q} = \zv^{\ast}_q \right \}
384: \subset \Gamma^s.
385: \end{equation}
386: %
387: In fact, this solution must lie in the fixed-point subspace of
388: $\Sigma_{\zv^{\ast}_q}^s$:
389: %
390: \begin{equation}\label{fixsigma}
391: {\mbox{Fix}}(\Sigma_{\zv^{\ast}_q}^s) = \left\{\zv \in \Rset^p :
392: \sigma \zv = \zv, \;\mbox{for all}\;
393: \sigma \in \Sigma_{\zv^{\ast}_q}^s\right\},
394: \end{equation}
395: %
396: which is a linear subspace of $\Rset^p$ and invariant
397: under~$\gv$~\cite{GolubitskySS88}.
398:
399: Since standing hexagons do not possess spatio-temporal
400: symmetries~\cite{RobertsSW86}, $\Gamma^s$-equivariance is sufficient to
401: determine the normal form of the map $\gv$ in the case of (temporal)
402: period-preserving bifurcations. However, for period-doubling bifurcations there
403: is an extra symmetry pertaining to the normal form, related to the time-shift
404: action $\tau_t: t \rightarrow t + T$ on the bifurcating modes. In this case,
405: the normal form of the map $\gv$ is $\left(\Gamma^s \times \mrom{Z}_2
406: \right)$-equivariant~\cite{ChossatG88,LambM99}. Once the full symmetry group of
407: the normal form of $\gv$ is determined, we can apply the equivariant branching
408: lemma, which, with suitable interpretation, states that if certain
409: non-degeneracy conditions are satisfied, there is a unique branch of
410: bifurcating solutions for each isotropy subgroup of $\Gamma \equiv \Gamma^s
411: \times \mrom{Z}_2$ with a one-dimensional fixed-point subspace. So instead of
412: solving for solutions of the nonlinear vector field~$\gv$, we can simply look
413: for isotropy subgroups of $\Gamma$ with this property. To apply the equivariant
414: branching lemma, we need to know explicitly how symmetry acts on all the
415: marginal modes, but experimental observations only provide information about
416: the instantaneous spatial symmetry of one of these modes. We cannot infer
417: directly from the observations the total number of marginal modes that are
418: related by symmetry at the bifurcation point, nor the set of matrices that
419: represent the action of the symmetry group $\Gamma$ on the marginal modes and
420: the map~$\gv$. However, this difficulty can be resolved if we make the
421: (generic) assumption that the bifurcation is associated with an irreducible
422: representation of the group~$\Gamma$, and this is where the need to invoke
423: representation theory arises.
424:
425: In order to introduce the key properties of irreducible representations
426: (irreps) and describe how they can be computed for a finite group~$\Gamma$, we
427: recall the following definitions~\cite{Cornwell}.
428: %
429: \begin{enumerate}
430: \renewcommand{\labelenumi}{(\roman{enumi})}
431: %
432: \item A {\em representation} of the group $\Gamma$ is a homomorphism $\psi$
433: that maps $\Gamma$ into a set of invertible $n\times n$ matrices
434: $\rombf{M}_{\Gamma}$ acting on $\Rset^n$ or $\Cset^n$, in other words
435: \begin{displaymath}
436: \psi\left(\gamma\right) = M_{\gamma}, \quad \quad
437: \gamma \in \Gamma,\; M_{\gamma} \in \rombf{M}_{\Gamma}
438: \end{displaymath}
439: such that $\psi\left(\gamma_1 \gamma_2\right) = \psi(\gamma_1)\psi(\gamma_2)\:
440: \mbox{for all}\, \gamma_1, \gamma_2 \in \Gamma$. The integer~$n$ is the {\em
441: dimension} of the representation.
442: \item Two $n$-dimensional representations $\rombf{M}_{\Gamma}$ and
443: $\rombf{N}_{\Gamma}$ of $\Gamma$ are called {\em equivalent} if there is an
444: invertible $n \times n$ matrix $Q$ such that for each $\gamma \in \Gamma$,
445: \begin{displaymath}
446: N_\gamma = Q^{-1}M_\gamma Q, \quad M_\gamma \in \rombf{M}_{\Gamma},\,
447: N_\gamma \in \rombf{N}_{\Gamma}.
448: \end{displaymath}
449: \item A {\em conjugacy class} of $\Gamma$ is a subset $C$ of $\Gamma$ such
450: that $\gamma^{-1}c\gamma \in C, \mbox{for all}\, c \in C \;\mbox{and}\;
451: \gamma\in\Gamma$.
452: \item The {\em character} of an element $\gamma \in \Gamma$ in a
453: representation $\rombf{M}_{\Gamma}$ is defined to be the trace of the matrix
454: $M_{\gamma}$, and we denote this value by~$\chi_{M_{\gamma}}$.
455: \item A representation of $\Gamma$ on $\Rset^n$ ($\Cset^n$) is said to be {\em
456: irreducible} if it does not leave invariant any proper subspace of $\Rset^n$
457: ($\Cset^n$).
458: \end{enumerate}
459: Simplistically we can consider a representation as a set of $n\times n$
460: nonsingular matrices that specifies the action of $\Gamma$ on the vector space
461: $\Rset^n$ or $\Cset^n$ and at the same time preserves the group structure. It
462: is possible to show that every representation of a finite group is equivalent
463: to a unitary representation -- one in which all matrices are
464: unitary~\cite{Cornwell}. A simple result of definition (iv) is that the
465: character of the identity element in a representation is always equal to the
466: dimension~$n$ of that representation, and definitions (i), (iii) and (iv) imply
467: that elements in the same conjugacy class have the same character. The
468: characters of the irreps of $\Gamma$ obey a set of rules inherited from the
469: orthogonality theorem governing the underlying irreps~\cite{RileyHB98} and for
470: simple groups such as $\mrom{Z}_2$, $\mrom{Z}_6$ and $\deesix$, the character
471: tables can easily be constructed by appealing to those rules. The orthogonality
472: theorem also implies that the number of irreps of a group is equal to the
473: number of conjugacy classes. For finite groups with a semi-direct product
474: structure of the form $\Gamma = \mathcal{A} \dotplus \mathcal{B}$ such that
475: $\mathcal{A}$ is a normal (or invariant) subgroup of~$\Gamma$ (that is,
476: $\gamma^{-1}a\gamma\in\mathcal{A}$ for every $a\in\mathcal{A}$ and
477: $\gamma\in\Gamma$), the characters of the irreps of $\mathcal{A}$ and
478: $\mathcal{B}$ form the building blocks in determining all the characters and
479: constructing unitary irreps of the group $\Gamma$ via a special
480: algorithm~\cite{Cornwell}.
481:
482: In summary, analysis of the superlattice pattern using these group theoretic
483: tools depends on our being able to find a spatial lattice or periodic cell on
484: which standing hexagons and the marginal modes exhibiting the observed
485: symmetries fit. The arrangement of the standing hexagons in the periodic cell
486: then gives us a suitable symmetry group~$\Gamma^s$, which has a subgroup
487: $\Sigma_{\zv^{\ast}_q}^s$, defined in~\eqref{isotropy}, whose elements are
488: determined from experimental observations. Once we have calculated all the
489: characters of~$\Gamma^s$, the restriction provided by the requirement that
490: $\Sigma_{\zv^{\ast}_q}^s$ be the isotropy subgroup of the observed pattern
491: enables us to isolate the one irrep that describes the action of $\Gamma^s$ on
492: all the marginal modes related to the observed bifurcating mode by symmetry.
493: Indeterminacy in the choice of irreps can be avoided if we choose a unit cell
494: that captures exactly one spatial period of the observed pattern. The details
495: of this procedure are the subject of the next section.
496:
497: \begin{center}
498: \begin{figure}[htbp]
499: \begin{tabular}{ccc}
500: (a) & (b) & (c) \\
501: \resizebox{!}{6.2cm}{\includegraphics{newhexbox0.eps}} &
502: \resizebox{!}{6.2cm}{\includegraphics{litebox2.eps}} &
503: \resizebox{!}{6.2cm}{\includegraphics{litebox3.eps}} \\
504: \end{tabular}
505: \caption{\sl \small (a) Schematic representation of standing hexagons, where
506: $\tilde{\ev}_1$ and $\tilde{\ev}_2$ denote the vectors of translations defined
507: in~(\ref{tauonetwo}). (b)~Diagram depicting the instantaneous spatial symmetry
508: of the superlattice-two pattern (see figure~\ref{fig:original}(c)), whose
509: spatial periodicity can be captured by hexagonal cells mapped to one another by
510: the translations $\ev_1$ and $\ev_2$ as shown in~(c). The superlattice-two
511: pattern is left unchanged by $\ttc$, $\kappa_x$ and $\toc\rosixty^3$. The small
512: circles, dotted lines and shading in (b) and (c) serve to identify equivalent
513: hexagons related by translations, rotations and reflections of the pattern.}
514: \label{fig:sl2box}
515: \end{figure}
516: \end{center}
517:
518: \section{Finding the symmetry group~$\Gamma$ and the irrep for the
519: superlattice-two bifurcation problem}\label{sec:findgamma}
520:
521: A closer examination of images obtained from the experiment reveals that it is
522: possible to impose a hexagonal lattice on the observed patterns, whose
523: instantaneous spatial symmetries are depicted in
524: {\mbox{figure~\ref{fig:sl2box}}}. The choice of lattice is not unique as it can
525: be shown that there are many possible candidates (for example a $\sqrt{3}:1$
526: rectangular lattice), but a hexagonal lattice is a natural choice due to the
527: symmetry of the standing hexagons. Let us denote the two generating vectors of
528: the hexagonal lattice $\mathcal{L}$ by $\ev_1, \ev_2 \in \Rset^2$ such that
529: %
530: \begin{equation}\label{genvect}
531: \nom{\ev_1} = \nom{\ev_2} = c,
532: \end{equation}
533: %
534: where $c$ is a scaling factor (figure~\ref{fig:sl2box}(c)). Functions in the
535: plane that are doubly-periodic with respect to $\mathcal{L}$ satisfy
536: %
537: \begin{equation}\label{doublepd}
538: u(\xv,t) = u(\xv + \lv,t), \quad \xv = (x,y) \in \Rset^2,
539: \; \lv \in \mathcal{L},
540: \end{equation}
541: %
542: where the lattice is defined as
543: %
544: \begin{displaymath}
545: \mathcal{L} = \left\{n_1\ev_1 + n_2\ev_2: \;
546: (n_1, n_2) \in \Zset^2\right\}.
547: \end{displaymath}
548: %
549:
550: First let us consider the spatial symmetries of the standing hexagons shown
551: schematically in figure~\ref{fig:sl2box}(a). They are invariant under the
552: action of $\deesix$ as well as two translations, which we define as follows:
553: %
554: \begin{equation}\label{tauonetwo}
555: \ton: \xv \rightarrow \xv + \tilde{\ev}_1, \quad
556: \ttw: \xv \rightarrow \xv + \tilde{\ev}_2,
557: \end{equation}
558: %
559: and let $\nom{\tilde{\ev}_1} = \nom{\tilde{\ev}_2} = c_0$ be the observed size
560: of the periodic cell in which the basic standing hexagons fit. Our aim is to
561: pick a value for the scaling factor $c$ in \eqref{genvect} in terms of~$c_0$.
562: As indicated at the end of section~\ref{sec:groupideas}, a suitable choice of
563: the value~$c$ is one that gives a hexagonal cell whose size captures precisely
564: one spatial period of the bifurcating modes, as shown schematically in
565: figure~\ref{fig:sl2box}(c). In fact, the observed ratio of the two lengths
566: $\nom{\ev_i}$ and $\nom{{\tilde{\ev}}_i}$ is $c/c_0 = 2\sqrt{3}$, and for this
567: value of~$c$, the symmetry group~$\Gamma^s$ of the standing hexagons includes
568: non-trivial translations generated by $\ton$ and~$\ttw$. The structure of the
569: group $\langle \ton, \ttw \rangle$ can be determined if we express each of the
570: translations in \eqref{tauonetwo} in terms of $\ev_1$ and~$\ev_2$. We can then
571: look for the lowest powers $n_1, n_2, n_3, n_4 \in \Zset^+$ such that
572: $\ton^{n_1}$, $\ttw^{n_2}$, $\ton^{n_3}\ttw^{n_4}$ map the lattice
573: $\mathcal{L}$ to itself, and thus determine the order of the group
574: $\langle\ton,\ttw\rangle$.
575:
576: Guided by the experimental observations, we choose $\tilde{\ev}_1 =
577: c_0\left(\frac{\sqrt{3}}{2}, \haf\right)$, $\tilde{\ev}_2 = c_0\left(0, 1
578: \right)$ such that $\ev_1=4\tilde{\ev}_1-2\tilde{\ev}_2=c_0(2\sqrt{3},0)$,
579: $\ev_2=2\tilde{\ev}_1+2\tilde{\ev}_2=c_0(\sqrt{3},3)$ (see
580: figure~\ref{fig:sl2box}). The translations can now be written as
581: %
582: \begin{equation}\label{newtauonetwo}
583: \ton : \xv \rightarrow \xv + \frac{1}{6}\ev_1 + \frac{1}{6}\ev_2, \quad
584: \ttw : \xv \rightarrow \xv - \frac{1}{6}\ev_1 + \frac{1}{3}\ev_2,
585: \end{equation}
586: %
587: and we can easily show that they satisfy $\ton^6 = \ttw^6 = \ton^2\ttw^2 =
588: \hbox{identity}$, as vectors of the form $\xv + m_1\ev_1 + m_2\ev_2$ for any
589: integers $m_1$ and $m_2$ lie in $\mathcal{L}$ and are therefore identified.
590: Since $\ton$ and $\ttw$ commute we can also see that every element generated by
591: $\ton$ and $\ttw$ can be written as $\ton^n\ttw$ or $\ton^n$ for $n =
592: 0,\ldots\,,5$. In total there are twelve different translations, forming a
593: group that is isomorphic to $\mrom{Z}_6 \times \mrom{Z}_2$. The order of this
594: group being twelve corresponds to the fact that each of the large hexagonal
595: cells in figure~\ref{fig:sl2box}(c) contains exactly twelve of the smaller
596: hexagons.
597:
598: \begin{table}[htb]
599: \begin{center}
600: \small
601: \begin{tabular}{|c|r|r|r|r|r|r|r|r|r|r|r|r|r|r|r|} \hline
602: & \multicolumn{15}{c|}{Conjugacy classes of $\Gamma^{s}$} \\
603: \hline
604: \multirow{3}{6mm}{Irrep} & $id$ & $\kappa_x$
605: & $\rho$ & $\rho\kappa_x$
606: & $\rho^2$ & $\rho^3$
607: & $\toc\rho^3\kappa_x$
608: & $\ttw\rho^2$ & $\toc\rho^3$
609: & $\ttw$
610: & $\ton\kappa_x$ & $\ttw\kappa_x$ & $\ttc$
611: & $\toc\kappa_x$ & $\ton^2$ \\
612: & & $\ttc\kappa_x$ & & & & &
613: $\ton\ttw\rho^3\kappa_x$ & &
614: $\ton\ttw\rho^3$ & & & & & & \\
615: & $(1)\ast$ & $(6)\ast$ & $(24)$ & $(18)$ & $(8)$
616: & $(3)$ & $(18)\ast$ & $(16)$ & $(9)\ast$
617: & $(6)$ & $(12)$ & $(12)$ & $(3)\ast$ & $(6)$ & $(2)$ \\ \hline
618: $\mrom{M}_{\Gamma^s}^{1}$ & $1$ & $1$ & $1$ & $1$ & $1$
619: & $1$ & $1$ & $1$ & $1$ & $1$
620: & $1$ & $1$ & $1$ & $1$ & $1$ \\
621: $\mrom{M}_{\Gamma^s}^{2}$ & $1$ & $-1$ & $1$ & $-1$ & $1$
622: & $1$ & $-1$ & $1$ & $1$ & $1$
623: & $-1$ & $-1$ & $1$ & $-1$ & $1$ \\
624: $\mrom{M}_{\Gamma^s}^{3}$ & $1$ & $1$ & $-1$ & $-1$ & $1$
625: & $-1$ & $-1$ & $1$ & $-1$ & $-1$
626: & $-1$ & $-1$ & $1$ & $-1$ & $1$ \\
627: $\mrom{M}_{\Gamma^s}^{4}$ & $1$ & $-1$ & $-1$ & $1$ & $1$
628: & $-1$ & $1$ & $1$ & $-1$ & $1$
629: & $-1$ & $-1$ & $1$ & $-1$ & $1$ \\
630: $\mrom{M}_{\Gamma^s}^{5}$ & $2$ & $0$ & $1$ & $0$ & $-1$
631: & $-2$ & $0$ & $-1$ & $-2$ & $2$
632: & $0$ & $0$ & $2$ & $0$ & $2$ \\
633: $\mrom{M}_{\Gamma^s}^{6}$ & $2$ & $0$ & $-1$ & $0$ & $-1$
634: & $2$ & $0$ & $-1$ & $2$ & $2$
635: & $0$ & $0$ & $2$ & $0$ & $2$ \\
636: $\mrom{M}_{\Gamma^s}^{7}$ & $2$ & $2$ & $0$ & $0$ & $2$
637: & $0$ & $0$ & $-1$ & $0$ & $-1$
638: & $-1$ & $-1$ & $2$ & $2$ & $-1$ \\
639: $\mrom{M}_{\Gamma^s}^{8}$ & $2$ & $-2$ & $0$ & $0$ & $2$
640: & $0$ & $0$ & $-1$ & $0$ & $-1$
641: & $1$ & $1$ & $2$ & $-2$ & $-1$ \\
642: $\mrom{M}_{\Gamma^s}^{9}$ & $3$ & $1$ & $0$ & $1$ & $0$
643: & $3$ & $-1$ & $0$ & $-1$ & $-1$
644: & $-1$ & $1$ & $-1$ & $-1$ & $3$ \\
645: $\mrom{M}_{\Gamma^s}^{10}$ & $3$ & $-1$ & $0$ & $1$ & $0$
646: & $-3$ & $-1$ & $0$ & $1$ & $-1$
647: & $1$ & $-1$ & $-1$ & $1$ & $3$ \\
648: $\mrom{M}_{\Gamma^s}^{11}$ & $3$ & $-1$ & $0$ & $-1$ & $0$
649: & $3$ & $1$ & $0$ & $-1$ & $-1$
650: & $1$ & $-1$ & $-1$ & $1$ & $3$ \\
651: $\mrom{M}_{\Gamma^s}^{12}$ & $3$ & $1$ & $0$ & $-1$ & $0$
652: & $-3$ & $1$ & $0$ & $1$ & $-1$
653: & $-1$ & $1$ & $-1$ & $-1$ & $3$ \\
654: $\mrom{M}_{\Gamma^s}^{13}$ & $4$ & $0$ & $0$ & $0$ & $-2$
655: & $0$ & $0$ & $1$ & $0$ & $-2$
656: & $0$ & $0$ & $4$ & $0$ & $-2$ \\
657: $\mrom{M}_{\Gamma^s}^{14}$ & $6$ & $-2$ & $0$ & $0$ & $0$
658: & $0$ & $0$ & $0$ & $0$ & $1$
659: & $-1$ & $1$ & $-2$ & $2$ & $-3$ \\
660: $\mrom{M}_{\Gamma^s}^{15}$ & $6$ & $2$ & $0$ & $0$ & $0$
661: & $0$ & $0$ & $0$ & $0$ & $1$
662: & $1$ & $-1$ & $-2$ & $-2$ & $-3$ \\
663: \hline
664: \end{tabular}
665: \vspace{4mm}
666: \caption{\sl \small Character table of the group of spatial symmetries
667: $\Gamma^s$ constructed via the algorithm taken from~\cite{Cornwell}. A
668: representative element is shown for each conjugacy class, and the number of
669: elements in the class is given in brackets. Classes marked by $\ast$ contain
670: elements (specified at the top of the table) of the eight-element group
671: $\Sigma^{s}_{\zv^{\ast}_q}=\langle\;\ttc,\;\kappa_x,\;\toc\rosixty^3\,\rangle$.
672: \vspace{8mm}}
673: \label{tab:chartable}
674: \end{center}
675: \end{table}
676:
677: So in terms of the lattice~$\mathcal{L}$, the full spatial symmetry of the
678: standing hexagons is given by the group
679: $\Gamma^s=\left(\mrom{Z}_6\times\mrom{Z}_2\right)\dotplus\deesix$, where
680: $\deesix$ is generated by a reflection $\kappa_x$ and a $60^{\circ}$~rotation
681: $\rho$ and its standard action on $\Rset^2$ is given by
682: %
683: \begin{equation}
684: \kappa_x : (x,y) \rightarrow (-x,y), \quad
685: \rho: (x,y)\rightarrow \haf\left(x - \sqrt{3}y, \sqrt{3}x+y\right),
686: \end{equation}
687: %
688: and $\mrom{Z}_6 \times \mrom{Z}_2$, an invariant subgroup of~$\Gamma^s$, is
689: generated by the two translations
690: %
691: \begin{equation}
692: \ton : (x,y) \rightarrow
693: \left(x+ \frac{\sqrt{3}}{2}c_0, y + \frac{1}{2}c_0\right),
694: \quad
695: \ttw : (x,y) \rightarrow
696: \left(x, y + c_0\right).
697: \end{equation}
698: %
699: The group~$\Gamma^s$ has the semi-direct product structure mentioned in
700: {\mbox{section~\ref{sec:groupideas}}}. As a result we can apply the algorithm
701: taken from~\cite{Cornwell} to calculate all its characters and irreps, and we
702: present the characters of its irreps in {\mbox{table~\ref{tab:chartable}}}.
703:
704: Any elements that have the same character as the identity in a unitary irrep
705: of~$\Gamma^s$ must also act like the identity~\cite{Collins90}. Using this
706: simple idea and the information taken from experimental observations about the
707: spatial symmetries of the unstable mode, we can single out the irrep of
708: $\Gamma^s$ that describes the instantaneous symmetry-breaking behaviour.
709: Careful study of snapshots of the superlattice-two pattern shows that it is
710: invariant under the action of $\ttc$, $\kappa_x$ and $\toc\rosixty^3$ (see
711: figures~\ref{fig:original}(b) and~\ref{fig:original}(c), where the pattern is
712: shown at a slightly tilted angle, and figure~\ref{fig:sl2box}(b)). The group
713: generated by these elements is by definition the isotropy subgroup of the
714: bifurcating mode under the action of $\Gamma^s$ (cf~\eqref{isotropy}),
715: therefore
716: %
717: \begin{equation}\label{isosub}
718: \Sigma^{s}_{\zv^{\ast}_q} \equiv
719: \langle \; \ttc, \;\kappa_x ,\;\toc \rosixty^3\, \rangle
720: \subset \Gamma^s.
721: \end{equation}
722: %
723: We can now go through the list of characters of $\Gamma^s$ given in
724: {\mbox{table~\ref{tab:chartable}}} and determine which irrep satisfies the
725: criteria of permitting $\Sigma^{s}_{\zv^{\ast}_q}$ defined in~\eqref{isosub} to
726: be an isotropy subgroup of the bifurcating solution. First, any irreps that
727: satisfy
728: %
729: \begin{equation}\label{elimcond}
730: \chi_{M_{\gamma_s}} = \chi_{M_{id}} \;
731: \mbox{for some}\; \gamma_s \in \Gamma^s \;\mbox{and} \;
732: \gamma_s \not\in \Sigma^{s}_{\zv^{\ast}_q}
733: \end{equation}
734: %
735: must be rejected, because in these cases the isotropy subgroup of
736: $\zv^{\ast}_q$ must contain spatial symmetry elements apart from those that are
737: observed. This eliminates representations 1--12. In representation~13, the
738: class containing~$\ttc$ is represented by the identity, but this class also
739: contains $\toc$ and $\ton\ttw$, which are not in $\Sigma^{s}_{\zv^{\ast}_q}$,
740: so eliminating this irrep and leaving only 14 and~15. Then we can use the trace
741: formula~\cite{GolubitskySS88} to calculate the dimension of the fixed-point
742: subspace of $\Sigma^{s}_{\zv^{\ast}_q}$:
743: %
744: \begin{equation*}
745: \mbox{dim}\,\mbox{Fix}\left(\Sigma\right)
746: = \frac{1}{\nom{\Sigma}}\sum_{\sigma \in \Sigma}
747: \chi_{M_{\sigma}},
748: \end{equation*}
749: %
750: which gives 0 for representation~14 and 1 for representation~15. Clearly we
751: require $\mbox{dim}\,\mbox{Fix}\left(\Sigma^{s}_{\zv^{\ast}_q}\right)\neq0$,
752: since $\mbox{Fix}\left(\Sigma^{s}_{\zv^{\ast}_q}\right)$ is non-trivial. Thus
753: the six-dimensional irrep $\mrom{M}_{\Gamma^s}^{15}$ is the only one in which
754: $\Sigma^{s}_{\zv^{\ast}_q}$ satisfies the conditions of being an isotropy
755: subgroup of the observed mode.
756:
757: In addition to being equivariant under the action of spatial symmetries as
758: specified by this irrep, the normal form of the period-doubling bifurcation
759: problem has an extra symmetry corresponding to a translation in time by one
760: period of the external forcing:
761: %
762: \begin{equation}\label{timeshift}
763: \tau_t: t \rightarrow t + T.
764: \end{equation}
765: %
766: This element can be viewed as a spatio-temporal symmetry with a trivial spatial
767: action, and it acts independently from elements in $\Gamma^s$ with respect to
768: the standing hexagons. So the full symmetry group~$\Gamma$ of the normal form
769: for the superlattice bifurcation problem is a direct product between $\Gamma^s$
770: and the group $\langle \tau_t \rangle$, which, as can be seen
771: from~\eqref{pdouble} and~\eqref{timeshift}, is isomorphic to~$\mrom{Z}_2$,
772: hence $\Gamma = \Gamma^s \times \mrom{Z}_2$ as we pointed out in
773: section~\ref{sec:groupideas}. We can write each element $\gamma \in \Gamma$ as
774: %
775: \begin{equation}\label{directprod}
776: \gamma = \left(\gamma_s, {\sigma}_t\right), \;\gamma_s \in
777: \Gamma^s,\;{\sigma}_t \in \langle\tau_t \rangle\, ,
778: \end{equation}
779: %
780: such that for $\gamma_1 = \left(\gamma_{s_1}, {\sigma}_{t_1}\right)$, $\gamma_2
781: = \left(\gamma_{s_2}, {\sigma}_{t_2}\right)$, $\gamma_1\gamma_2 =
782: \left(\gamma_{s_1}\gamma_{s_2}, {\sigma}_{t_1}{\sigma}_{t_2}\right)$. Because
783: of the direct product structure of $\Gamma$ and the period-doubling nature of
784: the bifurcating solution, $\tau_t$ must act like $-1$ on the amplitudes of the
785: marginal modes. Therefore the irrep of $\Gamma$ that specifies the action of
786: spatial and spatio-temporal symmetry elements on the marginal modes and the
787: normal form of $\gv$ can be constructed from the set of matrices
788: $M_{\gamma_s}\! \in \mrom{M}^{15}_{\Gamma^s}$ as follows:
789: %
790: \begin{displaymath}
791: M_{\gamma} = \left\{ \begin{array}{ll}
792: M_{\gamma_s} & \mbox{if} \quad {\sigma}_t = \hbox{identity} \\
793: -M_{\gamma_s} & \mbox{if} \quad {\sigma}_t = \tau_t
794: \end{array}
795: \right.
796: \end{displaymath}
797: %
798: for all $\gamma = \left(\gamma_s, {\sigma}_t\right) \in \Gamma$. This irrep,
799: which we denote by~$\mrom{M}_{\Gamma}$, is of the same dimension as
800: $\mrom{M}^{15}_{\Gamma^s}$, which implies that we have a six-dimensional centre
801: manifold at the bifurcation point. So all bifurcating solutions can be written
802: as $u(\xv, t) = u_0(\xv, t) + \zeta(\xv,t)$ such that
803: %
804: \begin{equation}\label{planform1}
805: \zeta(\xv, qT) = A_qf_1(\xv) + B_qf_2(\xv) + C_qf_3(\xv) +
806: \mbox{c.c.} + \mbox{h.o.t.}, \quad q \in \Zset
807: \end{equation}
808: %
809: where $u_0(\xv,t)$ represents standing hexagons, \hbox{c.c.} denotes complex
810: conjugate, \hbox{h.o.t.} denotes the higher-order terms, and $A_q$, $B_q$,
811: $C_q\in\Cset$ are the small amplitudes of $f_1$, $f_2$ and $f_3$, the three
812: complex marginal eigenfunctions that form a basis for the neutral eigenspace
813: (excluding the two zero eigenvalues corresponding to translating the underlying
814: pattern). Note that by including the higher-order terms, $\zeta$~represents the
815: nonlinear perturbation from the standing hexagons.
816:
817: Applying the method described in~\cite{Cornwell}, we can construct all the
818: $6\times6$ matrices $M_{\gamma}$ that specify the action of $\Gamma$ on
819: $\Rset^6$ (or $\Cset^3$) for the irrep~$\mrom{M}_{\Gamma}$. Rather than
820: describe this procedure, we find it convenient to specify the group action by
821: choosing a small number of Fourier modes to represent the marginal
822: eigenfunctions, and working out how the amplitudes of these modes $A_q$, $B_q$
823: and $C_q$ transform under the generating elements of~$\Gamma$. Since
824: representations are defined only up to a similarity transformation, the choice
825: of Fourier modes we make will not matter, as long as we are careful not to
826: introduce any accidental symmetries (which would become apparent on checking
827: the characters).
828:
829: Any function $u(\xv,t)$ defined on the lattice~$\mathcal{L}$ can be written as
830: a double Fourier series of the form
831: %
832: \begin{equation}
833: u(\xv,t) = \sum_{j_1\in \Zset} \sum_{j_2\in \Zset}
834: u_{j_1,j_2}(t)\, e^{2\pi i \left(j_1\kv_1 + j_2\kv_2\right)
835: \cdot \xv} ,
836: \end{equation}
837: %
838: where $\kv_1$ and $\kv_2$ are the generating wavevectors of the dual lattice
839: $\mathcal{L}^{\ast}$ related to $\ev_1$ and $\ev_2$ by $\kv_i \cdot \ev_j =
840: \delta_{ij}$ such that \eqref{doublepd} holds. Our choice of the vectors
841: $\ev_1$ and $\ev_2$ requires $\kv_1 = k\left(\frac{\sqrt{3}}{2},-\haf\right)$,
842: and $\kv_2 = k(0,1)$, where $k = \frac{1}{3c_0}$.
843:
844: We use the observed instantaneous symmetry of the pattern (see
845: figure~\ref{fig:sl2box}(c)) to select a representative function from the full
846: set of Fourier modes, starting with a single Fourier mode $e^{2\pi
847: i(j_1\kv_1+j_2\kv_2)\cdot\xv}$ (and its complex conjugate) for some choice of
848: integers $j_1$ and~$j_2$. If the pattern is to be invariant under $\ttc$,
849: $j_1$~must be even, so set $j_1=2m$, and, for later convenience, set $j_2=m+n$,
850: where $m$ and $n$ are integers. With this choice, the Fourier mode is $e^{2\pi
851: ik(\sqrt{3}mx+ny)}$. The pattern is also invariant under $\toc\rosixty^3$. Now
852: $\rosixty^3$ replaces the chosen mode by its complex conjugate, and $\toc$
853: multiplies the mode by a complex number with unit modulus. Since $\toc$ is of
854: order two but not equal to the identity, it must act by multiplying the mode
855: by~$-1$. This forces $m+n$ to be odd (and so for the observed pattern, the
856: amplitude of the Fourier mode must be pure imaginary). The translation $\tau_1$
857: must act with order~6 (otherwise the pattern would be invariant under a lesser
858: translation in that direction), so $3m+n\equiv1\mod 6$ or $3m+n\equiv5\mod 6$.
859: The second of these is essentially the complex conjugate of the first, so we
860: choose $3m+n\equiv1\mod 6$; $(m,n)$ could be $(0,1)$, $(2,1)$ or $(1,4)$, for
861: example. Finally, the reflection $\kappa_x$ generates a new function $e^{2\pi
862: ik(-\sqrt{3}mx+ny)}$, so the superlattice-two pattern can be exemplified by a
863: mode of the form $f_1=e^{2\pi ik(\sqrt{3}mx+ny)}+e^{2\pi ik(-\sqrt{3}mx+ny)}$.
864: Sixty degree rotations of this function generate $f_2$ and $f_3$, so we have:
865: %
866: \begin{equation}\label{marginalmodes}
867: f_1 = e^{2\pi i\Kv_1\cdot \xv} + e^{2\pi i\Kv_2\cdot \xv}, \;
868: f_2 = e^{2\pi i\Kv_3\cdot \xv} + e^{2\pi i\Kv_4\cdot \xv}, \;
869: f_3 = e^{2\pi i\Kv_5\cdot \xv} + e^{2\pi i\Kv_6\cdot \xv}
870: \end{equation}
871: %
872: (the true eigenfunctions will be made up of linear combinations of such
873: functions), where
874: %
875: \begin{center}
876: \begin{tabular}{ll}
877: $\Kv_1 = k(\sqrt{3}m, n)$, &
878: $\Kv_2 = k(-\sqrt{3}m, n)$, \\
879: $\Kv_3 = \frac{k}{2}\left(\sqrt{3}(m+n), (-3m+n)\right)$, &
880: $\Kv_4 = \frac{k}{2}\left(\sqrt{3}(-m+n), (3m+n)\right)$, \\
881: $\Kv_5 = \frac{k}{2}\left(\sqrt{3}(m+n), (3m-n)\right)$, &
882: $\Kv_6 = \frac{k}{2}\left(\sqrt{3}(-m+n), -(3m+n)\right)$. \\
883: \end{tabular}
884: \end{center}
885: %
886: These wavevectors have the same wavenumber $K(m,n) = k\sqrt{3m^2+n^2}$, with
887: $m$ and $n$ satisfying $3m+n \equiv 1\mod 6$. With this choice of basis
888: functions, the relevant irrep of $\Gamma$ can be specified by the action of the
889: generating elements of $\Gamma$ on the amplitudes $(A_q, B_q, C_q)$:
890: %
891: \begin{eqnarray}\label{amplitudeirrep}
892: \kappa_x : (A_q, B_q, C_q) &\rightarrow& (A_q, \bar{C_q}, \bar{B_q}),
893: \label{amplitudeirrepfirst} \\
894: \rosixty: (A_q, B_q, C_q) &\rightarrow& (B_q, C_q, \bar{A_q}), \\
895: \ton: (A_q, B_q, C_q) &\rightarrow&
896: (e^{\frac{i\pi}{3}}A_q, e^{\frac{i2\pi}{3}}B_q, e^{\frac{i\pi}{3}}C_q), \\
897: \ttw: (A_q, B_q, C_q) &\rightarrow&
898: (e^{\frac{i2\pi}{3}}A_q, e^{\frac{i\pi}{3}}B_q, e^{-\frac{i\pi}{3}}C_q), \\
899: \tau_{t}: (A_q, B_q, C_q) &\rightarrow& (-A_q, -B_q, -C_q).
900: \label{amplitudeirreplast}
901: \end{eqnarray}
902: %
903: We include the subharmonic action of~$\tau_{t}$ here for completeness. The same
904: representation could be constructed using the method described
905: in~\cite{Cornwell}.
906:
907: \section{Normal form of the bifurcation problem}\label{sec:bifprob}
908:
909: We now have sufficient information to invoke the equivariant branching
910: lemma~\cite{GolubitskySS88} and describe the different patterns that must be
911: formed in the instability that created the superlattice-two from standing
912: hexagons. Before doing this, we will compute the normal form for the
913: bifurcation since we need it to work out the stability of the various patterns.
914: The irrep~(\ref{amplitudeirrepfirst}--\ref{amplitudeirreplast}) we identified
915: in section~\ref{sec:findgamma} implies that the reduced map $\gv$ introduced in
916: \eqref{normalform} is six-dimensional, and we let ${\mathbf{z}}_q = (A_q, B_q,
917: C_q)$, $q \in \Zset$, $A_q,B_q,C_q\in\Cset$. As indicated earlier, the action
918: of $\tau_t$ defined in \eqref{timeshift} is due to the subharmonic nature of
919: the bifurcating modes with respect to the overall driving period $T$ given in
920: \eqref{pdouble}. If each iteration in $\zv_q$ corresponds to advancing in time
921: by $T$, then $\zv_{q+2} = -\zv_{q+1} = \zv_{q}$. Consequently, $\gv(\zv_{q}) =
922: \zv_{q+1} = -\zv_{q+2} = -\gv(\zv_{q+1})= -\gv(-\zv_{q})$~\cite{ElphickTBCI87}.
923: So the map $\gv$ will be an odd function of the amplitudes $A_q$, $B_q$ and
924: $C_q$, as well as being $\Gamma$-equivariant. This information enables us to
925: write down the form of $\gv$ including up to fifth order terms:
926: %
927: \begin{eqnarray}
928: A_{q+1} &=& -\left(1+\mu\right) A_q + \alpha_1\nomsq{A_q}A_q +
929: \alpha_2\left(\nomsq{B_q} + \nomsq{C_q}\right)A_q
930: + \beta_1\nom{A_q}^4 A_q \nonumber \\
931: & & + \mbox{ }
932: \beta_2\left(\nom{B_q}^4 + \nom{C_q}^4\right)A_q
933: + \beta_3\nomsq{A_q}\left(\nomsq{B_q} + \nomsq{C_q}\right)A_q
934: + \beta_4 \nomsq{B_q}\nomsq{C_q}A_q \nonumber \\
935: & & + \mbox{ }
936: \beta_5 B_{q}^2 \bar{C}_{q}^{2}\bar{A_q}
937: + \nu \bar{A}_{q}^5 ,
938: \label{Aqdot} \\
939: B_{q+1} &=& -\left(1+\mu\right) B_q + \alpha_1\nomsq{B_q}B_q +
940: \alpha_2\left(\nomsq{A_q} + \nomsq{C_q}\right)B_q
941: + \beta_1\nom{B_q}^4 B_q \nonumber \\
942: & & + \mbox{ }
943: \beta_2\left(\nom{A_q}^4 + \nom{C_q}^4\right)B_q
944: + \beta_3\nomsq{B_q}\left(\nomsq{A_q} + \nomsq{C_q}\right)B_q
945: + \beta_4 \nomsq{A_q}\nomsq{C_q}B_q \nonumber \\
946: & & + \mbox{ }
947: \beta_5 A_{q}^2 C_{q}^{2}\bar{B_q}
948: + \nu \bar{B}_{q}^5 ,
949: \label{Bqdot}\\
950: C_{q+1} &=& -\left(1+\mu\right) C_q + \alpha_1\nomsq{C_q}C_q +
951: \alpha_2\left(\nomsq{A_q} + \nomsq{B_q}\right)C_q
952: + \beta_1\nom{C_q}^4 C_q \nonumber \\
953: & & + \mbox{ }
954: \beta_2\left(\nom{A_q}^4 + \nom{B_q}^4\right)C_q
955: + \beta_3\nomsq{C_q}\left(\nomsq{A_q} + \nomsq{B_q}\right)C_q
956: + \beta_4 \nomsq{A_q}\nomsq{B_q}C_q \nonumber \\
957: & & + \mbox{ }
958: \beta_5 \bar{A}_{q}^2 B_{q}^{2}\bar{C_q}
959: + \nu \bar{C}_{q}^5,
960: \label{Cqdot}
961: \end{eqnarray}
962: %
963: where all coefficients are forced by symmetry to be real.
964:
965: Apart from the $\nu$~terms, the equations above are equivalent to the
966: $\mrom{T}^2\dotplus\deesix\times\mrom{Z}_2$-equivariant amplitude equations
967: (truncated to the same order) that arise in the context of Boussinesq
968: convection on a hexagonal lattice~\cite{GolubitskySS88,GoluSwif84}, once they
969: are re-interpreted as amplitude equations rather than a map. The $\nu$~terms
970: have the effect of breaking the full $\mrom{T}^2$ (two-torus) symmetry group of
971: translations in a periodic domain to the discrete translations allowed by the
972: underlying pattern. A natural question to ask is why we needed to work out the
973: details of the representation before writing down these amplitude equations.
974: The main reason is that we did not know in advance how many linearly
975: independent marginal eigenfunctions are involved in the instability. Even if we
976: had assumed that there were six, it has turned out that there are two
977: six-dimensional irreps, only one of which is involved in the bifurcation. The
978: other six-dimensional irrep is generated by taking $f_1=e^{2\pi
979: ik(\sqrt{3}mx+ny)}-e^{2\pi ik(-\sqrt{3}mx+ny)}$ and (following a similar
980: analysis) results in the same amplitude equation. Without realising this, one
981: might conclude incorrectly that patterns that are odd under $\kappa_x$
982: reflection might also be found in this instability. All the other irreps in
983: table~\ref{tab:chartable} have dimension less than six (that is, there are
984: fewer than six independent marginal eigenfunctions), so the order of the
985: relevant normal forms would be correspondingly less.
986:
987: We also use~\eqref{hequivariant} to write down the dynamics of the position
988: $\dv_q$ of the underlying standing hexagons, truncated to quartic order:
989: %
990: \begin{equation}\label{dqdot}
991: \dv_{q+1} = \dv_q + \xi\, \mbox{Im} \left[
992: \begin{array}{c}
993: A_q^2(\bar{C}_q^2 - B_q^2) - 2B_q^2C_q^2 \\
994: -\sqrt{3} A_q^2(B_q^2 + \bar{C_q}^2)
995: \end{array} \right],
996: \end{equation}
997: where $\xi$ is a constant.
998:
999: \begin{table}[htb]
1000: \begin{center}
1001: \setlength{\extrarowheight}{2pt}
1002: \begin{tabular}{|l|lll|c|}\hline
1003: \multicolumn{3}{|l}{representative solution branch} &
1004: \multicolumn{2}{l|}{isotropy subgroup \mbox{\hspace{3mm}}
1005: averaged symmetry}\\\hline
1006: \multirow{2}{6mm}{I}
1007: & $1$.
1008: & $A\in \Rset$, $B=C=0$
1009: & $\langle\ttc,\;\kappa_x,\;\rosixty^3,\;{\widetilde{\tau}_1}^3\rangle$
1010: & \multirow{2}{28mm}{$\langle\toc,\;\ttc,\;\kappa_x,\;\rosixty^3\rangle$} \\
1011: & $2$.
1012: & $A\in i\Rset$, $B=C=0$
1013: & $\langle\ttc,\;\kappa_x,\;\toc\rosixty^3,\;{\widetilde{\tau}_1}^3\rangle$
1014: & \\
1015: \hline
1016: \multirow{2}{6mm}{II}
1017: & $3$.
1018: & $A=B=C\in \Rset$
1019: & $\deesix=\langle\kappa_x,\;\rosixty\rangle$
1020: & \multirow{2}{28mm}{$\deesix$} \\
1021: & $4$.
1022: & $A=-B=C\in i\Rset$
1023: & $\langle\kappa_x,\;\widetilde{\rosixty}\rangle$
1024: & \\
1025: \hline
1026: \multirow{2}{6mm}{III}
1027: & $5$.
1028: & $A=0$, $B=C\in \Rset$
1029: & $\langle\kappa_x,\;\rosixty^3,\;{\widetilde{\tau}_2}^3\rangle$
1030: & \multirow{2}{28mm}{$\langle\ttc,\;\kappa_x,\;\rosixty^3\rangle$} \\
1031: & $6$.
1032: & $A=0$, $B=-C\in i\Rset$
1033: & $\langle\kappa_x,\;\ttc\rosixty^3,\;{\widetilde{\tau}_2}^3\rangle$
1034: & \\
1035: \hline
1036: \end{tabular}
1037: \vspace{4mm}
1038: \caption{\sl \small Primary solution branches of the normal form
1039: (\ref{Aqdot}--\ref{Cqdot}) and their isotropy subgroups, grouped into three
1040: types, I--\hbox{III}. Using the analysis presented in
1041: section~\ref{subsec:timeavg} we can show that the two branches within each type
1042: of solution share the same time-averaged spatial symmetries. Branch~2 of type~I
1043: corresponds to the superlattice-two pattern. \vspace{8mm}}
1044: \label{tab:primarysolns}
1045: \end{center}
1046: \end{table}
1047:
1048: We can show that there are six isotropy subgroups whose fixed-point subspaces
1049: are one-dimensional, so the equivariant branching lemma tells us that there are
1050: at least six primary bifurcating branches of solutions from standing hexagons,
1051: and we summarise these solutions in table~\ref{tab:primarysolns}. Elements
1052: accented by a tilde represent spatio-temporal symmetries, which, using the
1053: notation introduced in~\eqref{directprod}, can be written as
1054: ${\widetilde{\tau}_1}^3 = \left(\toc, \tau_t\right)$, and similarly for
1055: ${\widetilde{\tau}_2}^3$ and~$\widetilde{\rosixty}$. The superlattice-two
1056: pattern corresponds to branch~2 of type~\hbox{I}.
1057:
1058: \begin{center}
1059: \begin{figure}[htbp]
1060: \begin{tabular}{cccc}
1061: (a) standing hexagons & (b) branch $3$ &
1062: (c) branch $4$, $t = 0$ & (d) branch $4$, $t = T$\\
1063: \multicolumn{4}{c}{\resizebox{!}{40mm}{\includegraphics{fig3abcd.eps}}}\\
1064: (e) branch $2$, $t=0$ & (f) branch $2$, $t=T$&
1065: (g) branch $1$, $t=0$ & (h) branch $1$, $t=T$ \\
1066: \multicolumn{4}{c}{\resizebox{!}{40mm}{\includegraphics{fig3efgh.eps}}}\\
1067: (i) branch $5$, $t=0$& (j) branch $5$, $t=T$&
1068: (k) branch $6$, $t=0$& (l) branch $6$, $t=T$\\
1069: \multicolumn{4}{c}{\resizebox{!}{40mm}{\includegraphics{fig3ijkl.eps}}}
1070: \end{tabular}
1071: \caption{\sl \small Instantaneous planforms of the different solution branches
1072: summarised in table~\ref{tab:primarysolns} and illustrated here in frames
1073: (b)--(l) as small-amplitude perturbations to standing hexagons. Solid squares
1074: represent lattice points of~$\mathcal{L}$.
1075: (a) Standing hexagons, which have the full $\Gamma$ symmetry.
1076: (b) Solution branch 3 with $\deesix$ symmetry, referred to
1077: in~\cite{DionSkel97} as `superhexagons'. The periodic hexagonal boxes are
1078: delineated by light borders surrounding each cell.
1079: (c) \& (d) Solution branch 4 at $t = 0$ and $T$ showing $\mrom{D}_3$ symmetry
1080: as well as the spatio-temporal symmetry~$\widetilde{\rho}$.
1081: (e) \& (f) The superlattice-two pattern corresponds to solution branch 2 as
1082: they share the same instantaneous spatial symmetries. This pattern is shown
1083: here at $t = 0$ and $T$ with spatio-temporal symmetry
1084: $\widetilde{\tau}_{1}^{3}$ evident.
1085: (g) \& (h) Similar spatio-temporal symmetry is displayed by solution branch~1.
1086: (i)--(l) Branches $5$ \& $6$ have very similar symmetry properties: both are
1087: invariant under the action of $\widetilde{\tau}_{2}^{3}$ and instantaneously
1088: they differ only by a shift of the reflection symmetry~$\kappa_y$.}
1089: \label{fig:solutions}
1090: \end{figure}
1091: \end{center}
1092:
1093: For the choice of wave integer pair $(m,n)=(2,1)$, the instantaneous planforms
1094: of these six solution branches are illustrated schematically in
1095: {\mbox{figure~\ref{fig:solutions}}. We can compare
1096: figures~\ref{fig:solutions}(e) and~\ref{fig:solutions}(f)
1097: with~\ref{fig:original}(b) and~\ref{fig:original}(c) and notice that the
1098: appearance of stripes at regular intervals in the grey-scale plots of solution
1099: branch 2 closely resembles the essential features of the experimentally
1100: observed superlattice-two pattern.
1101:
1102: None of these primary branches leads to a net drift of the underlying hexagonal
1103: pattern; this can be seen in two ways: first, because the rate of drift (from
1104: \eqref{dqdot}, truncated to quartic order) is zero on all six primary branches;
1105: second (and more convincing) since $\rho^3$ is in the symmetry group of all the
1106: time averaged patterns (see below). In other words, the patterns are all pinned
1107: by the $180^{\circ}$ rotation symmetry on average.
1108:
1109: \subsection{Stability results}
1110:
1111: We summarise in {\mbox{table~\ref{tab:branchevals}}} the branching equations
1112: and the Floquet multipliers of the period~$2T$ patterns, for each of the six
1113: primary solutions guaranteed to exist by the equivariant branching lemma.
1114: Floquet multipliers greater than one in magnitude indicate instability. We
1115: group the six branches into three types and denote them by I, II and III as
1116: shown in tables~\ref{tab:primarysolns} and~\ref{tab:branchevals}. It is evident
1117: that branches within each type are degenerate up to third-order terms, thus
1118: necessitating the inclusion of quintic terms. In particular, only one solution
1119: branch within each of types I and II can be stable depending on the signs
1120: of~$\nu$ and $\beta_5+\nu$, and both branches in type III are always unstable.
1121: Only one branch can bifurcate stably, and all branches must be supercritical
1122: for one of them to be stable. One of the requirements for the observed
1123: superlattice-two pattern (\ie branch 2 of type~I) to be stable is that the
1124: quintic coefficient $\nu>0$. If we also assume the non-degeneracy conditions
1125: $\alpha_1\neq0$, $\alpha_1+2\alpha_2\neq0$, $\alpha_1\pm\alpha_2\neq0$,
1126: $\nu\neq0$ and $\beta_5+\nu\neq0$, close to the bifurcation point the relative
1127: stability of branches of the three types is illustrated by the bifurcation
1128: diagrams shown in figure~\ref{fig:bifndgm}.
1129:
1130: \begin{table}[htb]
1131: \begin{center}
1132: \setlength{\extrarowheight}{3pt}
1133: \begin{tabular}{|c|l|l|}\hline
1134: \multicolumn{2}{|c|}
1135: {Primary solutions and branching equations} &
1136: Floquet multipliers (multiplicity) \\ \hline \hline
1137: \multirow{4}{4mm}{I}
1138: &1. $A_q=R_1$, $B_q=C_q=0$,
1139: & $1-4\alpha_1 R_1^2$,
1140: $1-2\left(\alpha_2-\alpha_1\right)R_1^2$ (4 times), \\
1141: & $0 = -\mu + \alpha_1 R_1^2 + \left(\beta_1 + \nu\right)R_1^4$
1142: & $1+12\nu R_1^4$ \\
1143: \cline{2-3}
1144: &2. $A_q=iR_2$, $B_q=C_q=0$,
1145: & $1-4\alpha_1 R_2^2$,
1146: $1-2\left(\alpha_2-\alpha_1\right)R_2^2$ (4 times), \\
1147: & $0 = -\mu + \alpha_1 R_2^2 + \left(\beta_1 - \nu\right)R_2^4$
1148: &$1-12\nu R_2^4$ \\
1149: \hline
1150: \multirow{6}{4mm}{II}
1151: &3. $A_q=B_q=C_q=R_3$,
1152: & $1-4\left(\alpha_1+2\alpha_2\right)R_3^2$, \\
1153: & $0 = -\mu + \left(\alpha_1 + 2\alpha_2\right)R_3^2$
1154: & $1+4\left(\alpha_2-\alpha_1\right)R_3^2$ (2 times), \\
1155: & $\quad{}+\left(\beta_1+2\beta_2+2\beta_3+\beta_4+\beta_5+\nu\right)R_3^4$
1156: & $1+12\nu R_3^4$ (2 times),
1157: $1+12\left(\beta_5+\nu\right)R_3^4$ \\
1158: \cline{2-3}
1159: &4. $A_q=-B_q=C_q=iR_4$,
1160: & $1-4\left(\alpha_1+2\alpha_2\right)R_4^2$, \\
1161: & $0 = -\mu + \left(\alpha_1 + 2\alpha_2\right)R_4^2$
1162: & $1+4\left(\alpha_2-\alpha_1\right)R_4^2$ (2 times), \\
1163: & $\quad{}+\left(\beta_1+2\beta_2+2\beta_3+\beta_4-\beta_5-\nu\right)R_4^4$
1164: & $1-12\nu R_4^4$ (2 times),
1165: $1-12\left(\beta_5+\nu\right)R_4^4$ \\
1166: \hline
1167: \multirow{6}{4mm}{III} &
1168: 5. $A_q=0$, $B_q=C_q=R_5$,
1169: & $1-4\left(\alpha_1+\alpha_2\right)R_5^2$,
1170: $1+4\left(\alpha_2-\alpha_1\right)R_5^2$, \\
1171: & $0 = -\mu + \left(\alpha_1 + \alpha_2\right)R_5^2$
1172: & $1-2\left(\alpha_2-\alpha_1\right)R_5^2$ (2 times), \\
1173: & $\quad{}+\left(\beta_1 + \beta_2 + \beta_3 + \nu\right)R_5^4$
1174: & $1+12\nu R_5^4$ (2 times) \\
1175: \cline{2-3}
1176: &6. $A_q= 0$, $B_q=-C_q=iR_6$,
1177: & $1-4\left(\alpha_1+\alpha_2\right)R_6^2$,
1178: $1+4\left(\alpha_2-\alpha_1\right)R_6^2$, \\
1179: & $0 = -\mu + \left(\alpha_1 + \alpha_2\right)R_6^2$
1180: & $1-2\left(\alpha_2-\alpha_1\right)R_6^2$ (2 times), \\
1181: & $\quad{}+
1182: \left(\beta_1 + \beta_2 + \beta_3 - \nu\right)R_6^4$
1183: & $1-12\nu R_6^4$ (2 times) \\
1184: \hline
1185: \end{tabular}
1186: \vspace{4mm}
1187: \caption{\sl \small Branching equations and Floquet multipliers for the six
1188: primary period-two solutions of the normal form~(\ref{Aqdot}--\ref{Cqdot})
1189: listed in table~\ref{tab:primarysolns}. $A_q$, $B_q$ and $C_q$ are complex
1190: amplitudes of the marginal modes defined in~(\ref{planform1}). Only leading
1191: order terms in $R_i$ are shown, and the multiplicities of the Floquet
1192: multipliers (computed from the second iterate of the map) are indicated.
1193: \vspace{8mm}}
1194: \label{tab:branchevals}
1195: \end{center}
1196: \end{table}
1197:
1198: \begin{figure}[htb]
1199: \begin{center}
1200: \resizebox{!}{4.0truein}{\includegraphics{unfsltwo.eps}}
1201: \caption{\sl \small Bifurcation diagrams for the $\Gamma$-equivariant normal
1202: form (\ref{Aqdot}--\ref{Cqdot}). The sign of the cubic coefficient $\alpha_1$
1203: determines whether solution type~I bifurcates sub- or supercritically. If we
1204: assume that the quintic coefficient $\nu>0$, then the superlattice-two pattern
1205: (branch~2 of type~I) can occur as a stable branch in the region of the
1206: $(\alpha_1,\alpha_2)$ space given by $\left\{(\alpha_1,\alpha_2): \alpha_1>0,\,
1207: \alpha_2>\alpha_1\right\}$.}
1208: \label{fig:bifndgm}
1209: \end{center}
1210: \end{figure}
1211:
1212: The experimental results~\cite{KudrPier98} suggest that the bifurcation may
1213: have subcritical branches as there is a parameter regime in which standing
1214: hexagons and the superlattice-two pattern may coexist. On the other hand, the
1215: experimentalists report no hysteresis between standing hexagons and
1216: superlattice-two, while they do report hysteresis between hexagons and other
1217: patterns at other parameter values, so it is not clear whether or not there is
1218: a direct bifurcation from standing hexagons to the superlattice-two pattern in
1219: the experiments. With our parameters, we require $\alpha_1>0$,
1220: $\alpha_2>\alpha_1$ and $\nu>0$ for the superlattice-two pattern (branch~2 of
1221: type~I) to bifurcate stably, but the branch could also be stable in the region
1222: $\alpha_1<0$, $\alpha_2>\alpha_1$ and $\nu>0$ if there were a saddle-node
1223: bifurcation on branch~\hbox{I}.
1224:
1225: \subsection{Time-averaged behaviour}\label{subsec:timeavg}
1226:
1227: We can study the symmetry properties of the time-averaged image of the observed
1228: solution by integrating over a full period of the newly created periodic orbit
1229: (cf~\cite{BaranyDG93,RuckSilb98,DellnitzGN94}). Specifically, we let
1230: $u_0(\xv,t)$ be the standing hexagons solution and $\zeta(\xv,t)$ the nonlinear
1231: perturbation to this solution such that $u(\xv,t)=u_0(\xv,t)+\zeta(\xv,t)$
1232: represents the observed pattern. We know that $u_0(\xv,t)$ is
1233: $\Gamma$-invariant, \ie $\gamma u_0(\xv,t)=u_0(\xv,t)$ for all
1234: $\gamma\in\Gamma$, and have also found that the spatial and spatio-temporal
1235: symmetry of $\zeta(\xv,t)$ is given by its isotropy subgroup
1236: $\langle\ttc,\;\kappa_x,\;\toc\rosixty^3,\;{\widetilde{\tau}_1}^3\rangle \equiv
1237: \Sigma_{\zeta}\subset\Gamma$. Let $\gamma_s$ and $\gamma_t$ denote respectively
1238: the purely spatial symmetry elements and the spatial part of spatio-temporal
1239: symmetry elements in $\Sigma_{\zeta}$ that act on $\zeta(\xv,t)$ as follows:
1240: %
1241: \begin{equation}\nonumber
1242: \gamma_s \zeta(\xv,t) = \zeta(\xv,t), \quad
1243: \gamma_t \zeta(\xv,t) = \zeta(\xv,t+T).
1244: \end{equation}
1245: %
1246: The time-averaged value of $u(\xv,t)$ can be obtained by integrating over the
1247: full period of the bifurcating solution:
1248: %
1249: \begin{eqnarray}
1250: \bar{u}(\xv) &=& \frac{1}{2T}\int_{0}^{2T} u_0(\xv,t) +
1251: \zeta(\xv,t) \,\ud t
1252: \nonumber \\
1253: &=& \bar{u}_0(\xv)
1254: + \frac{1}{2T}\int_{0}^{T} \zeta(\xv,t) +
1255: \zeta(\xv,t+T) \,\ud t \label{baru},
1256: \end{eqnarray}
1257: %
1258: where we have used the fact that $u_0(\xv,t) = u_0(\xv,t+T)$. Clearly,
1259: $\bar{u}$ shares the same spatial symmetry with $\zeta$ because both
1260: $\bar{u}_0$ and the individual entries of the integrand in \eqref{baru} are
1261: invariant under the action of $\gamma_s \in \Sigma_{\zeta}$. It is also
1262: invariant under~$\gamma_t$ because the integrand in \eqref{baru} as a whole is
1263: invariant under~$\gamma_t$:
1264: %
1265: \begin{eqnarray}
1266: \gamma_t \bar{u}(\xv) &=& \gamma_t\bar{u}_0(\xv)
1267: + \frac{1}{2T}\int_0^{T} \gamma_t \zeta(\xv,t)
1268: + \gamma_t \zeta(\xv,t+T) \, \ud t \nonumber \\
1269: &=& \bar{u}_0(\xv)
1270: + \frac{1}{2T}\int_0^{T} \zeta(\xv,t+T)
1271: + \zeta(\xv,t) \, \ud t \nonumber \\
1272: &=& \bar{u}(\xv). \nonumber
1273: \end{eqnarray}
1274: %
1275: This result in fact follows readily from more general results on the symmetries
1276: of chaotic attractors~\cite{BaranyDG93,DellnitzGN94}.
1277:
1278: In the case of the observed superlattice-two pattern with isotropy
1279: subgroup~$\Sigma_{\zeta}$, the spatial component of the spatio-temporal
1280: symmetry element, namely~$\toc$, will show up alongside $\ttc$, $\kappa_x $ and
1281: $\toc\rho^3$ in the time-averaged image to generate an augmented spatial
1282: symmetry group $\Sigma_{\bar{u}} = \langle \toc, \ttc, \kappa_x, \rho^3
1283: \rangle$. This prediction is in agreement with experimental results
1284: (figure~\ref{fig:original}(a)) and can be understood in the following way. The
1285: action of the translations $\ton$ and $\ttw$ on $\bar{u}(\xv)$ is of order
1286: three since $\toc \bar{u}(\xv) = \ttc \bar{u}(\xv) = \bar{u}(\xv)$, whereas the
1287: order of the same action on $u(\xv)$ is six. So the averaged pattern will
1288: appear to be periodic on a lattice $\mathcal{L}_{\mrom{av}}$ spanned by basis
1289: vectors $\ev_{\mathrm{av}}$ such that $\nom{\ev_{\mathrm{av}}} =
1290: \haf\nom{\ev_i} = \haf c$. We have shown in section~\ref{sec:findgamma} that
1291: $c_0 = \frac{c}{2\sqrt{3}}$, it follows that $\nom{\ev_{\mathrm{av}}} =
1292: \sqrt{3}\nom{\tilde{\ev}_i}$. Therefore the ratio of spatial period of the
1293: averaged pattern to that of the basic standing hexagons is $1:\sqrt{3}$, which
1294: is consistent with the observation reported by~\cite{KudrPier98} as shown in
1295: {\mbox{figure \ref{fig:original}}}(a). Using the same reasoning and information
1296: from the isotropy subgroups of the primary solutions given in
1297: table~\ref{tab:primarysolns}, we therefore predict both branches in each type
1298: of solutions to have the same time-averaged symmetries.
1299:
1300: \section{Discussion}\label{sec:discuss}
1301:
1302: Starting from the observed instantaneous symmetry of the superlattice-two
1303: pattern reported in~\cite{KudrPier98}, we have been able to show (a)~that a
1304: pattern with the same instantaneous spatial symmetry as the superlattice-two
1305: pattern can bifurcate stably from standing hexagons in a spatial
1306: period-multiplying instability; (b)~that the pattern has the spatio-temporal
1307: symmetry (not reported in~\cite{KudrPier98}) of advancing one driving period in
1308: time combined with a translation by three units in space
1309: (figure~\ref{fig:solutions}(e) and~\ref{fig:solutions}(f)); and (c)~that this
1310: spatio-temporal symmetry accounts for the intermediate spatial scale and
1311: periodicity on a hexagonal lattice of the time-averaged pattern
1312: (figure~\ref{fig:original}(a)). We should emphasise that the intermediate
1313: spatial periodicity of the time-averaged pattern is not the spatial periodicity
1314: of the larger hexagonal lattice that we have assumed.
1315:
1316: Arbell \& Fineberg (unpublished) have found the superlattice-two state in their
1317: experiments and have confirmed that it does have the spatio-temporal symmetry
1318: that we predict. Our results also suggest that $60^{\circ}$~rotations are not
1319: in fact symmetries of the time-averaged pattern, but should be weakly broken.
1320: The breaking of $60^{\circ}$~rotational symmetry, if it is present, is
1321: evidently a small effect since the hexagons in figure~\ref{fig:original}(a) do
1322: appear to be invariant under $60^{\circ}$~rotations~\cite{KudrPier98} (this has
1323: been confirmed by Gollub, private communication). For other parameter values,
1324: the symmetry breaking effect may be more pronounced: Fineberg (private
1325: communication) reports that his experimental time-averaged pattern is not
1326: invariant under $60^{\circ}$~rotations. Clearly this would be an interesting
1327: issue to investigate in more detail, but the measurements are delicate and are
1328: liable to be prone to systematic errors or imperfections, so confirming our
1329: prediction could be difficult.
1330:
1331: The spatio-temporal symmetry of superlattice-two arises because the instability
1332: of standing hexagons is subharmonic. Other patterns, with different
1333: combinations of spatial and spatio-temporal symmetries, are possible stable
1334: branches in the same bifurcation problem. Not all branches of solutions have
1335: spatio-temporal symmetries, and some of the patterns share the same
1336: time-averaged symmetries even though they have different instantaneous
1337: planforms. The method we have presented is based entirely on symmetry arguments
1338: and is able to deal with instabilities of a fully nonlinear time-periodic
1339: solution.
1340:
1341: Spatial period-multiplying instabilities have arisen in a variety of contexts,
1342: in both one~\cite{ProctorWeiss93,Iooss86,AstonSW92,AmdjadiAP97} and two lateral
1343: directions
1344: \cite{KudrPier98,ArbellF98,WagnerMK98,White88,McKenzie88,ProcMatt96,Dawes2000,RWBMP2000}.
1345: Most of these situations involved relatively simple groups; part of the
1346: difficulty and interest here has been the size of the symmetry group, enlarged
1347: because of the number of translations broken by the new pattern. Only one of
1348: the 15 representations is involved in the superlattice-two bifurcation; other
1349: representations may be relevant to other experiments
1350: (particularly~\cite{ArbellF98,WagnerMK98}) in which standing hexagons lose
1351: stability to patterns that fit into the larger hexagonal cells we have used
1352: here.
1353:
1354: As can be seen in section~\ref{sec:findgamma}, a heuristic step in our method
1355: involves the choice of a suitable periodic cell that accommodates the observed
1356: patterns and whose size coincides with exactly one spatial period of the
1357: bifurcating modes. The arrangement of the underlying basic state in this cell
1358: then defines a spatial symmetry group~$\Gamma^s$ of the bifurcation problem and
1359: the instantaneous symmetries of the superlattice instability form its isotropy
1360: subgroup~$\Sigma^{s}$. If a larger hexagonal periodic cell that captures more
1361: than one spatial period of the bifurcating modes had been chosen, the
1362: translations $\ton$ and $\ttw$ given in~\eqref{tauonetwo} that leave standing
1363: hexagons invariant would have had higher order, resulting in a larger spatial
1364: symmetry group. In this case there would have been more than one irrep of
1365: $\Gamma^s$ in which $\Sigma^{s}$ satisfied the conditions of being an isotropy
1366: subgroup. By choosing the smallest possible periodic cell, we have found that
1367: such indeterminacy can be avoided.
1368:
1369: The method we have described in this paper for analysing certain types of
1370: symmetry-breaking instabilities bifurcating from a non-trivial basic state is
1371: based entirely on the observed spatial symmetries of these patterns. However,
1372: information on spatial symmetries of the new pattern alone may not be
1373: sufficient for our approach to be applicable in some problems. For example,
1374: consider a bifurcation problem defined on a spherical domain. Suppose a basic
1375: state with $\mrom{O}(3)$ symmetry loses stability and the observed bifurcating
1376: solutions are axisymmetric, then the isotropy subgroup of the bifurcating modes
1377: is given by~$\mrom{O}(2)$. If the eigenfunctions are expanded in spherical
1378: harmonics, it is known that $\mrom{O}(2)$ is a maximal isotropy subgroup of
1379: $\mrom{O}(3)$ for {\em all} even values of the spherical harmonic
1380: index~$l$~\cite{GolubitskySS88} and so an infinite number of irreps is relevant
1381: to the observed bifurcation. This example illustrates the fact that our method
1382: breaks down if the observed symmetries of the bifurcating modes form an
1383: isotropy subgroup for more than one irrep of the symmetry group of the basic
1384: state.
1385:
1386: We are currently involved in applying a similar method to the study of the
1387: `superlattice-one' pattern reported in~\cite{KudrPier98} as a bifurcating
1388: instability from standing hexagons. Preliminary analysis of the experimental
1389: data reveals that a suitable periodic box in this case will give rise to an
1390: arrangement of standing hexagons with a `hidden' reflection
1391: symmetry~\cite{DionSkel97}, which leads to extra complications in determining
1392: the spatial symmetry group. It is an interesting problem that deserves further
1393: investigation.
1394:
1395: Unlike some time-periodic solutions (for example, standing rolls), which can
1396: also be defined on a hexagonal lattice, standing hexagons possess only trivial
1397: spatio-temporal symmetries~\cite{RobertsSW86}. So our treatment of the
1398: superlattice patterns as symmetry-breaking instabilities from standing hexagons
1399: is relatively simple because only instantaneous spatial symmetries are needed
1400: to define the isotropy subgroup of these solutions. In general, our approach
1401: can be applied to the study of spatial period-multiplying bifurcations from
1402: solutions with spatio-temporal symmetries and used to investigate some of the
1403: possible symmetry-breaking behaviour, if techniques discussed by Rucklidge \&
1404: Silber~\cite{RuckSilb98} and Lamb~\&~Melbourne~\cite{LambM99} are also
1405: included.
1406:
1407: \begin{ack}
1408: We would like to thank Jerry Gollub for inspiring this work, Jonathan Dawes,
1409: Marty Golubitsky, Paul Matthews and Michael Proctor for sharing their insights
1410: with us, and Jay Fineberg for showing us his unpublished experimental results.
1411: DPT is grateful to the Croucher Foundation for financial support. The research
1412: of AMR is supported by EPSRC. The research of RBH was supported by King's
1413: College, Cambridge. The research of MS is supported by NSF grant DMS-9972059,
1414: NSF CAREER award DMS-9502266, and by NASA grant NAG3-2364.
1415:
1416: \end{ack}
1417:
1418: \bibliographystyle{PRB_normal}
1419: \bibliography{trhs}
1420:
1421: \end{document}
1422:
1423:
1424: