nlin0108005/RSF.tex
1:  
2: \documentclass[twoside,draft,reqno]{birkart} 
3:  
4: % this is for the mathbb to get Rset below
5: \usepackage{amsfonts}
6: 
7: % for included graphics
8: \input epsf
9: 
10: % locally defined commands
11:  
12: \newcommand{\etal}{\hbox{\it et al.}}
13: \newcommand{\Rset}{\mathbb{R}}
14: 
15: \begin{document} 
16: \setcounter{page}{1} 
17:  
18: \title[Secondary instabilities of hexagonal Faraday waves] 
19: {Secondary instabilities of hexagons: \\
20: a bifurcation analysis of experimentally \\
21: observed Faraday wave patterns}
22: \author[A.M.~Rucklidge]{A.M.~Rucklidge}
23:  
24: \address{%% 
25: Department of Applied Mathematics,\br 
26: University of Leeds, Leeds LS2 9JT UK}
27: 
28: \email{A.M.Rucklidge@leeds.ac.uk} 
29:  
30: \author[M.~Silber]{M.~Silber}
31: 
32: \address{%%
33: Department of Engineering Sciences and Applied Mathematics,\br
34: Northwestern University, Evanston IL 60208 USA}
35: 
36: \email{m-silber@northwestern.edu}
37: 
38: \author[J.~Fineberg]{J.~Fineberg}
39: 
40: \address{%%
41: The Racah Institute of Physics,\br
42: The Hebrew University of Jerusalem,\br
43: Givat Ram, Jerusalem 91904 Israel}
44: 
45: \email{jay@vms.huji.ac.il}
46: 
47: \begin{abstract} 
48: We examine three experimental observations of Faraday waves generated by
49: two-frequency forcing, in which a primary hexagonal pattern becomes
50: unstable to three different superlattice patterns. We use the
51: symmetry-based approach developed by Tse \etal~\cite{refT53} to
52: analyse the bifurcations involved in creating the three new
53: patterns. Each of the three examples reveals a different situation
54: that can arise in the theoretical analysis.
55: \end{abstract} 
56:  
57: \maketitle 
58:  
59: \section{Introduction} 
60: The classic Faraday wave experiment consists of a horizontal layer of fluid
61: that spontaneously develops a pattern of standing waves on its surface as it is
62: driven by vertical oscillation with amplitude exceeding a critical value. 
63: Recent experiments have revealed a wide variety of complex patterns, 
64: particularly in the large aspect ratio regime and with a forcing function 
65: containing two commensurate frequencies~\cite{refE10,refK84,refM118}. 
66: Transitions from the flat surface to a primary, spatially
67: periodic, pattern can be studied using equivariant bifurcation
68: theory~\cite{refG59}. These group theoretic techniques may also be
69: applied to secondary spatial period-multiplying transitions to
70: patterns with two distinct spatial scales (so called {\em
71: superlattice} patterns) as demonstrated by Tse \etal~\cite{refT53}.
72: 
73: We apply the method of Tse \etal~\cite{refT53} to the analysis of
74: three superlattice patterns observed when secondary subharmonic
75: instabilities destroy the basic hexagonal standing wave pattern in
76: two-frequency Faraday wave experiments.  We can make use not only of
77: the general symmetry-based approach from~\cite{refT53} but also of
78: many of the detailed results. The reason for this is that in
79: their paper, Tse \etal\ considered instabilities of hexagonal patterns
80: that broke the translation symmetry of the hexagons, but that remained
81: periodic in a larger hexagonal domain comprising twelve of the
82: original hexagons. The instabilities under consideration here satisfy
83: exactly the same conditions (though in fact they remain periodic in
84: smaller domains as well).
85: \looseness=-1
86: 
87: We begin by specifying the coordinate system and symmetries
88: we will use in section~\ref{sec:coordinates}, then describe the symmetries of
89: the three experimental patterns in section~\ref{sec:experimental}. In
90: section~\ref{sec:method}, we apply Tse \etal's method of analysis to these
91: three patterns, and present normal forms and stability calculations in
92: section~\ref{sec:normalforms}. We conclude in section~\ref{sec:discussion}.
93: 
94:  \begin{figure}
95:  \begin{center}
96:  \centerline{\epsfxsize3.0truein\epsffile{RSF_fig_1.eps}}
97: \caption{The coordinate system and certain elements of the symmetry
98: group~$\Gamma$. The origin of the coordinate system is at the centre
99: of the diagram, and the point $(1,0)$ is indicated. The small hexagons
100: represent the primary pattern, which is invariant under reflections
101: ($\kappa_x$ and~$\kappa_y$), $60^\circ$~rotations ($\rho$) and
102: translations ($\tau_1$ and~$\tau_2$). The secondary patterns are all
103: periodic in the larger hexagonal box. The three corner points labelled
104: with solid circles are identified through the assumed periodicity.}
105:  \label{fig:coordinates}
106:  \end{center}
107:  \end{figure}
108:  
109: \section{Coordinates and symmetries}\label{sec:coordinates}
110: The primary pattern is made up of regular hexagons, which are invariant under
111: the group~$D_6$ (made up of $60^\circ$~rotations and reflections) combined with
112: translation from one hexagon to the next (see figure~\ref{fig:coordinates}).
113: Tse \etal~\cite{refT53} studied experimental patterns reported
114: in~\cite{refK67}, which had the feature that after the secondary instability,
115: the pattern remained periodic in the larger hexagonal box in
116: figure~\ref{fig:coordinates}. The 144-element spatial symmetry group of the
117: primary hexagonal pattern within this box is~$\Gamma$, generated by the
118: following reflection~$\kappa_x$, rotation~$\rho$ and translations $\tau_1$
119: and~$\tau_2$:
120:  %
121:  \begin{alignat}{2}\label{eqn:symmetries}
122:  \kappa_x &: (x,y) \rightarrow (-x,y)                                 
123:           & \qquad
124:  \tau_1   &: (x,y) \rightarrow 
125:              (x,y) + \left(\frac{3}{2},\frac{\sqrt{3}}{2}\right)      \\
126:  \rho     &: (x,y) \rightarrow 
127:              \left(\frac{1}{2}x-\frac{\sqrt{3}}{2}y,
128:                    \frac{\sqrt{3}}{2}x+\frac{1}{2}y\right)              
129:           & \qquad
130:  \tau_2   &: (x,y) \rightarrow (x,y)+\left(0,\sqrt{3}\right)
131:  \end{alignat}
132:  %
133: We also define $\kappa_y=\kappa_x\rho^3$, and note the following identities:
134:  %
135:  \begin{gather}
136:  \kappa_x^2=\kappa_y^2=\rho^6=\tau_1^6=\tau_2^6=\tau_1^2\tau_2^2
137:            =\hbox{identity},                                          \\
138:  \rho\kappa_x = \kappa_x\rho^5,\qquad 
139:  \tau_1\kappa_x = \kappa_x\tau_1^5\tau_2,\qquad
140:  \tau_2\kappa_x = \kappa_x\tau_2,                                     \\
141:  \tau_1\rho = \rho\tau_1^3\tau_2,\qquad
142:  \tau_2\rho = \rho\tau_1,\qquad
143:  \tau_1\tau_2 = \tau_2\tau_1.
144:  \end{gather}
145:  %
146: The time translation $\tau_T$ advances time by one period~$T$ of the forcing
147: function, which is the same as the temporal period of the hexagonal pattern.
148: This time translation is combined with the spatial symmetries above to give
149: spatio-temporal symmetries.
150: 
151:  \begin{figure}
152:  \begin{center}
153:  \centerline{%
154:  \hbox to0.30\hsize{\hfil(a)\hfil}\hfil
155:  \hbox to0.30\hsize{\hfil(b)\hfil}\hfil
156:  \hbox to0.30\hsize{\hfil(c)\hfil}}
157:  \centerline{% 
158:  \epsfxsize0.30\hsize\epsffile{RSF_fig_2a.eps}\hfil
159:  \epsfxsize0.30\hsize\epsffile{RSF_fig_2b.eps}\hfil
160:  \epsfxsize0.30\hsize\epsffile{RSF_fig_2c.eps}}
161:  \centerline{%
162:  \hbox to0.30\hsize{\hfil(d)\hfil}\hfil
163:  \hbox to0.30\hsize{\hfil(e)\hfil}\hfil
164:  \hbox to0.30\hsize{\hfil(f)\hfil}}
165:  \centerline{% 
166:  \epsfxsize0.30\hsize\epsffile{RSF_fig_2d.eps}\hfil
167:  \epsfxsize0.30\hsize\epsffile{RSF_fig_2e.eps}\hfil
168:  \epsfxsize0.30\hsize\epsffile{RSF_fig_2f.eps}}
169:  \centerline{%
170:  \hbox to0.30\hsize{\hfil  \hfil}\hfil
171:  \hbox to0.30\hsize{\hfil(g)\hfil}\hfil
172:  \hbox to0.30\hsize{\hfil(h)\hfil}}
173:  \centerline{% 
174:  \hbox to 0.30\hsize{\hfil}\hfil
175:  \epsfxsize0.30\hsize\epsffile{RSF_fig_2g.eps}\hfil
176:  \epsfxsize0.30\hsize\epsffile{RSF_fig_2h.eps}}
177:  \caption{Experimental and idealised secondary patterns. (a-c)~Experimental 
178:  patterns, visualised from above. (d-f)~Idealised versions of~(a-c). 
179:  (g-h)~patterns (e-f) but seen one forcing period~$T$ later. The idealisations 
180:  are all rotated by about~$30^\circ$ compared with the experimental pictures.}
181:  \label{fig:idealised}
182:  \end{center}
183:  \end{figure}
184: 
185: \section{Experimental patterns}\label{sec:experimental}
186: The three experimentally observed patterns are shown in
187: figure~\ref{fig:idealised}(a-c), visualised using the techniques described
188: in~\cite{refA57}. Patterns (a) and (b) are both obtained using Dow-Corning
189: silicone oil with viscosity $47\,\hbox{cSt}$ and layer depth $0.35\,\hbox{cm}$,
190: while pattern (c) was found using a $23\,\hbox{cSt}$ oil layer of depth
191: $0.155\,\hbox{cm}$.  All three patterns are obtained with forcing function
192: containing two frequencies in the ratio $2:3$; pattern~(a) is found with
193: frequencies $50$ and $75\,\hbox{Hz}$, pattern~(b) with frequencies $70$ and
194: $105\,\hbox{Hz}$, and pattern~(c) with $40$ and $60\,\hbox{Hz}$ driving
195: frequencies. Pattern (c) was reported previously in~\cite{refA57}. Typically,
196: the secondary bifurcations occur at forcing amplitudes between 10 and 50\%
197: larger than the critical acceleration for the primary hexagonal state. Further
198: experimental details can be found in~\cite{refA57,refL61}.
199: 
200: For the purposes of the analysis, we consider the idealised versions of these
201: experimental patterns, shown in figure~\ref{fig:idealised}(d-f). The first
202: pattern in figure~\ref{fig:idealised}(a,d) retains the $D_6$~symmetry of the
203: original hexagons but breaks certain translation symmetries.  It is periodic in
204: the medium-sized dashed hexagon in figure~\ref{fig:idealised}(d), which implies
205: that the pattern is invariant under the translations~$\tau_1^3$
206: and~$\tau_1\tau_2$. It has no spatio-temporal symmetries. The second pattern is
207: similar, although it possesses only triangular ($D_3$) symmetry
208: instantaneously. Moreover, it has the spatio-temporal symmetry given by a
209: $60^\circ$ rotation combined with advance in time by one period~$T$ of the
210: forcing, as in figure~\ref{fig:idealised}(e,g). In fact, this spatio-temporal
211: symmetry was first suggested by the analysis below, and found to be consistent
212: with the experimental observations. The third pattern in
213: figure~\ref{fig:idealised}(c,f) is quite different: the dark lozenges in
214: figure~\ref{fig:idealised}(f) represent the enlarged gaps between the hexagons
215: in figure~\ref{fig:idealised}(c). The pattern is periodic in the medium-sized
216: dashed hexagon in figure~\ref{fig:idealised}(f), so is invariant under
217: translations $\tau_1^2$ and~$\tau_2^2=\tau_1^4$. It is also invariant under the
218: group of symmetries of a rectangle $D_2$, and possesses the spatio-temporal
219: symmetry of the translation~$\tau_2$ combined with advance in time by one
220: period~$T$ of the forcing, as in figure~\ref{fig:idealised}(f,h).
221: 
222: Using the information above, we write down the instantaneous (spatial) symmetry
223: groups of the three patterns from figure~\ref{fig:idealised}(a-c) in terms of
224: their generators:
225:  %
226:  \begin{equation}
227:  \Sigma_a=\langle\kappa_x,\rho,\tau_1^3,\tau_1\tau_2\rangle,\quad
228:  \Sigma_b=\langle\kappa_x,\rho^2,\tau_1^3,\tau_1\tau_2\rangle,\quad
229:  \Sigma_c=\langle\kappa_x,\kappa_y\tau_2,\tau_1^2\rangle.
230:  \end{equation}
231:  %
232: These groups are of order 48, 24 and 12 respectively. For the full
233: spatio-temporal symmetry groups, we would also include $\rho\tau_T$ in the
234: generators of~$\Sigma_b$, and $\tau_2\tau_T$ in the generators of~$\Sigma_c$,
235: but initially we will work with the spatial symmetry groups. The reason for
236: this is that the instantaneous (spatial) symmetries can be determined reliably
237: from a single experimental image, while extracting spatio-temporal symmetries
238: from the experimental data is more involved.
239: 
240: % needed to adjust the text size (small) and intercolumn spacing (8pt)
241: % to fit into the specified style.
242: 
243: \begin{table}
244: \begin{center}
245: \small
246: \begin{tabular}{r|r@{\hspace{8pt}}r@{\hspace{8pt}}r@{\hspace{8pt}}
247:                   r@{\hspace{8pt}}r@{\hspace{8pt}}r@{\hspace{8pt}}
248:                   r@{\hspace{8pt}}r@{\hspace{8pt}}r@{\hspace{8pt}}
249:                   r@{\hspace{8pt}}r@{\hspace{8pt}}r@{\hspace{8pt}}
250:                   r@{\hspace{8pt}}r@{\hspace{8pt}}r} 
251: \hline 
252:              & a & b & c & d & e & f & g & h & i & j & k & l & m & n & o \\
253:                     &
254: id                  &
255: $\kappa_x$          &
256: $\kappa_y$          &
257: $\tau_1$            &
258: $\tau_1^2$          &
259: $\tau_1^3$          &
260: $\kappa_x\tau_1$    &
261: $\kappa_x\tau_2$    &
262: $\kappa_x\tau_1^3$  &
263: $\kappa_y\tau_1^3$  &
264: $\rho$              &
265: $\rho^2$            &
266: $\rho^3$            &
267: $\rho^2\tau_1$      &
268: $\rho^3\tau_1^3$    \\
269:    & 1 & 6    & 18   & 6    & 2    & 3    & 12   & 12   & 6    & 18   & 24   & 8    & 3    & 16   & 9 \\
270: \hline
271: %    a     b      c      d      e      f      g      h      i      j      k      l      m      n      o
272:  1 & 1 &  $1$ &  $1$ &  $1$ &  $1$ &  $1$ &  $1$ &  $1$ &  $1$ &  $1$ &  $1$ &  $1$ &  $1$ &  $1$ &  $1$ \\ 
273:  2 & 1 & $-1$ & $-1$ &  $1$ &  $1$ &  $1$ & $-1$ & $-1$ & $-1$ & $-1$ &  $1$ &  $1$ &  $1$ &  $1$ &  $1$ \\
274:  3 & 1 &  $1$ & $-1$ &  $1$ &  $1$ &  $1$ &  $1$ &  $1$ &  $1$ & $-1$ & $-1$ &  $1$ & $-1$ &  $1$ & $-1$ \\
275:  4 & 1 & $-1$ &  $1$ &  $1$ &  $1$ &  $1$ & $-1$ & $-1$ & $-1$ &  $1$ & $-1$ &  $1$ & $-1$ &  $1$ & $-1$ \\
276:  5 & 2 &  $0$ &  $0$ &  $2$ &  $2$ &  $2$ &  $0$ &  $0$ &  $0$ &  $0$ &  $1$ & $-1$ & $-2$ & $-1$ & $-2$ \\
277:  6 & 2 &  $0$ &  $0$ &  $2$ &  $2$ &  $2$ &  $0$ &  $0$ &  $0$ &  $0$ & $-1$ & $-1$ &  $2$ & $-1$ &  $2$ \\
278:  7 & 2 &  $2$ &  $0$ & $-1$ & $-1$ &  $2$ & $-1$ & $-1$ &  $2$ &  $0$ &  $0$ &  $2$ &  $0$ & $-1$ &  $0$ \\
279:  8 & 2 & $-2$ &  $0$ & $-1$ & $-1$ &  $2$ &  $1$ &  $1$ & $-2$ &  $0$ &  $0$ &  $2$ &  $0$ & $-1$ &  $0$ \\
280:  9 & 3 &  $1$ &  $1$ & $-1$ &  $3$ & $-1$ & $-1$ &  $1$ & $-1$ & $-1$ &  $0$ &  $0$ &  $3$ &  $0$ & $-1$ \\
281: 10 & 3 & $-1$ &  $1$ & $-1$ &  $3$ & $-1$ &  $1$ & $-1$ &  $1$ & $-1$ &  $0$ &  $0$ & $-3$ &  $0$ &  $1$ \\
282: 11 & 3 & $-1$ & $-1$ & $-1$ &  $3$ & $-1$ &  $1$ & $-1$ &  $1$ &  $1$ &  $0$ &  $0$ &  $3$ &  $0$ & $-1$ \\
283: 12 & 3 &  $1$ & $-1$ & $-1$ &  $3$ & $-1$ & $-1$ &  $1$ & $-1$ &  $1$ &  $0$ &  $0$ & $-3$ &  $0$ &  $1$ \\
284: 13 & 4 &  $0$ &  $0$ & $-2$ & $-2$ &  $4$ &  $0$ &  $0$ &  $0$ &  $0$ &  $0$ & $-2$ &  $0$ &  $1$ &  $0$ \\
285: 14 & 6 & $-2$ &  $0$ &  $1$ & $-3$ & $-2$ & $-1$ &  $1$ &  $2$ &  $0$ &  $0$ &  $0$ &  $0$ &  $0$ &  $0$ \\
286: 15 & 6 &  $2$ &  $0$ &  $1$ & $-3$ & $-2$ &  $1$ & $-1$ & $-2$ &  $0$ &  $0$ &  $0$ &  $0$ &  $0$ &  $0$ \\
287: \hline
288: \end{tabular}
289: \vspace{2mm}
290:  \caption{Character table of the group~$\Gamma$, taken from Tse 
291:  \etal, with corrections. A representative element is shown on the second line
292:  for each conjugacy class (see also figure~\ref{fig:classes}),
293:  and the number of elements in the class is on the 
294:  third row. The next fifteen rows give the characters associated with 
295:  each conjugacy class for each of the fifteen representations.}
296:  \label{tab:chartable}
297:  \end{center}
298:  \end{table}
299: 
300: Each of the three instabilities that generates the three different patterns
301: will be associated with a set of marginally stable eigenfunctions; the new
302: pattern, at least near onset, can be thought of as (approximately) a linear
303: combination of these marginal eigenfunctions and the original hexagonal
304: pattern. Which linear superpositions are consistent with the nonlinearity
305: inherent in the pattern formation process is determined by our bifurcation
306: analysis. The symmetries in~$\Gamma$ all leave the primary hexagonal pattern
307: unchanged, so they must send marginal eigenfunctions onto linear combinations
308: of marginal eigenfunctions, which induces an action on the amplitudes of these
309: functions. In other words, if there are $n$~marginal eigenfunctions $f_1$,
310: \dots,~$f_n$, with $n$~amplitudes $\mathbf{a}=(a_1,\dots,a_n)\in\Rset^n$, each
311: element $\gamma\in\Gamma$ sends $\mathbf{a}$ to $R_\gamma\mathbf{a}$, where the
312: set of $n\times n$ orthogonal matrices $R_\gamma$ forms a
313: representation~$R_\Gamma$ of the group~$\Gamma$. For
314: subharmonic instabilities of the type of interest here, this will generically
315: be an irreducible representation (irrep)~\cite{refG59}. Tse \etal~\cite{refT53}
316: have computed all the irreps of the group~$\Gamma$; the character table of
317: these representations is reproduced in table~\ref{tab:chartable}. Recall that
318: the character of a group element~$\gamma$ in a representation is the trace of
319: the matrix~$R_\gamma$, and that conjugate elements (which form a conjugacy
320: class) have the same characters.
321: 
322: Once the representation associated with each of the three transitions is
323: identified, we can write down the normal form, work out what other patterns are
324: created in the same bifurcation, and compute stability of the patterns in terms
325: of the normal form coefficients.
326: 
327:  \begin{figure}
328:  \begin{center}
329:  \centerline{%
330:  \hbox to0.191\hsize{\hfil(a) identity (1)\hfil}\hfil
331:  \hbox to0.191\hsize{\hfil(b) $\kappa_x$ (6)\hfil}\hfil
332:  \hbox to0.191\hsize{\hfil(c) $\kappa_y$ (18)\hfil}}
333:  \centerline{% 
334:  \epsfxsize0.191\hsize\epsffile{RSF_fig_3a.eps}\hfil
335:  \epsfxsize0.191\hsize\epsffile{RSF_fig_3b.eps}\hfil
336:  \epsfxsize0.191\hsize\epsffile{RSF_fig_3c.eps}}
337:  \centerline{%
338:  \hbox to0.191\hsize{\hfil(d) $\tau_1$ (6)\hfil}\hfil
339:  \hbox to0.191\hsize{\hfil(e) $\tau_1^2$ (2)\hfil}\hfil
340:  \hbox to0.191\hsize{\hfil(f) $\tau_1^3$ (3)\hfil}}
341:  \centerline{% 
342:  \epsfxsize0.191\hsize\epsffile{RSF_fig_3d.eps}\hfil
343:  \epsfxsize0.191\hsize\epsffile{RSF_fig_3e.eps}\hfil
344:  \epsfxsize0.191\hsize\epsffile{RSF_fig_3f.eps}}
345:  \centerline{%
346:  \hbox to0.191\hsize{\hfil(g) $\kappa_x\tau_1$ (12)\hfil}\hfil
347:  \hbox to0.191\hsize{\hfil(h) $\kappa_x\tau_2$ (12)\hfil}\hfil
348:  \hbox to0.191\hsize{\hfil(i) $\kappa_x\tau_1^3$ (6)\hfil}}
349:  \centerline{% 
350:  \epsfxsize0.191\hsize\epsffile{RSF_fig_3g.eps}\hfil
351:  \epsfxsize0.191\hsize\epsffile{RSF_fig_3h.eps}\hfil
352:  \epsfxsize0.191\hsize\epsffile{RSF_fig_3i.eps}}
353:  \centerline{%
354:  \hbox to0.191\hsize{\hfil(j) $\kappa_y\tau_1^3$ (18)\hfil}\hfil
355:  \hbox to0.191\hsize{\hfil(k) $\rho$ (24)\hfil}\hfil
356:  \hbox to0.191\hsize{\hfil(l) $\rho^2$ (8)\hfil}}
357:  \centerline{% 
358:  \epsfxsize0.191\hsize\epsffile{RSF_fig_3j.eps}\hfil
359:  \epsfxsize0.191\hsize\epsffile{RSF_fig_3k.eps}\hfil
360:  \epsfxsize0.191\hsize\epsffile{RSF_fig_3l.eps}}
361:  \centerline{%
362:  \hbox to0.191\hsize{\hfil(m) $\rho^3$ (3)\hfil}\hfil
363:  \hbox to0.191\hsize{\hfil(n) $\rho^2\tau_1$ (16)\hfil}\hfil
364:  \hbox to0.191\hsize{\hfil(o) $\rho^3\tau_1^3$ (9)\hfil}}
365:  \centerline{% 
366:  \epsfxsize0.191\hsize\epsffile{RSF_fig_3m.eps}\hfil
367:  \epsfxsize0.191\hsize\epsffile{RSF_fig_3n.eps}\hfil
368:  \epsfxsize0.191\hsize\epsffile{RSF_fig_3o.eps}}
369:  \caption{The 15 conjugacy classes of~$\Gamma$. One 
370:  element from and the number of elements in each class are indicated. The 
371: letters (a)--(o) correspond to the columns of table~\ref{tab:chartable}.}
372:  \label{fig:classes}
373:  \end{center}
374:  \end{figure}
375: 
376: \section{Method}\label{sec:method}
377: The first task is to identify which representation is relevant for each
378: bifurcation. Tse \etal~\cite{refT53} outlined a two-stage method to accomplish
379: this. First, any symmetry element that is represented by the identity matrix in
380: a particular representation must appear in the symmetry group of every branch
381: of solutions created in a  bifurcation with that representation. This can be
382: used to eliminate from consideration any representation that has an element
383: with character equal to the character of the identity that does not appear in
384: the symmetry group of the observed pattern. Second, we make use of the trace
385: formula from~\cite{refG59}, which gives the dimension of the subspace of
386: $\Rset^n$ that is fixed by a particular isotropy subgroup~$\Sigma$ of~$\Gamma$
387: with representation given by the matrices~$R_\Gamma$:
388:  %
389:  \begin{equation}\label{eq:traceformula}
390:  \hbox{dim}\,\hbox{fix}(\Sigma) = 
391:    \frac{1}{|\Sigma|}\sum_{\sigma\in\Sigma}\hbox{Tr}\,R_\sigma,
392:  \end{equation}
393:  %
394: where $|\Sigma|$ is the number of elements in~$\Sigma$. Specifically, we use
395: the trace formula to eliminate those representations for which the spatial
396: symmetry group of the pattern fixes a zero-dimensional subspace (implying that
397: the subgroup is not an isotropy subgroup); only the remaining representations
398: need be examined in more detail.
399: \looseness=-1
400: 
401: We proceed by first counting the number of elements in each conjugacy class for
402: each of the symmetry groups $\Sigma_a$, $\Sigma_b$ and~$\Sigma_c$.
403: Figure~\ref{fig:classes} shows representative elements from each class and is
404: helpful for this categorization. The result of this is:
405:  $\Sigma_a$ contains:
406:  %
407:  \begin{equation}
408:  a:1,\, b:6,\, c:6,\, f:3,\, i:6,\, j:6,\, k:8,\, l:8,\, m:1,\, o:3
409:  \end{equation}
410:  %
411:  (that is, one element from class~$a$, six from class~$b$ etc.); $\Sigma_b$
412:  contains:
413:  %
414:  \begin{equation}
415:  a:1,\, b:6,\, f:3,\, i:6,\, l:8;
416:  \end{equation}
417:  %
418:  and $\Sigma_c$ contains:
419:  %
420:  \begin{equation}
421:  a:1,\, b:1,\, c:3,\, e:2,\, h:2,\, o:3.
422:  \end{equation}
423:  %
424: The element $\tau_1^2$ does not appear in the symmetry groups of patterns (a)
425: and (b), which eliminates representations 1--6 and 9--12 (since $\tau_1^2$ is
426: represented by the identity matrix in all these: see
427: table~\ref{tab:chartable}). Similarly, $\tau_1\tau_2$ in class~$f$ and $\rho^3$
428: do not appear in $\Sigma_c$, which eliminates representations 1--9, 11 and 13
429: from consideration for that bifurcation problem.
430: 
431: Next, by applying~(\ref{eq:traceformula}), we find that pattern~(a) has a
432: non-zero dimensional fixed point subspace only in representation~7, as does
433: pattern~(b). The spatial symmetry group of pattern~(a) fixes a one-dimensional
434: subspace, and that of pattern~(b) fixes a two-dimensional subspace. Pattern~(c)
435: has a one-dimensional fixed point subspace in representations 10 and~12, and
436: zero in other representations.
437: 
438: We are therefore faced with three different situations: the spatial symmetry
439: group~$\Sigma_a$ fixes a one-dimensional subspace in representation~7, so we
440: expect by the Equivariant Branching Lemma (see~\cite{refG59}) that such a
441: pattern will generically be found in a bifurcation problem with that
442: representation.
443: 
444: Pattern~(b), on the other hand, has a spatial symmetry group that fixes a
445: two-dimensional subspace. However, we must take into account that the pattern
446: arises in a subharmonic (period-doubling) instability, and extend the
447: groups~$\Gamma$ and~$\Sigma_b$ to the spatio-temporal symmetry groups that
448: arise by including time translations. We may then show that the spatio-temporal
449: symmetry group of pattern~(b) fixes a one-dimensional subspace, and so also
450: arises generically in a subharmonic bifurcation with representation~7. This is
451: the same representation as with pattern~(a), obtained for similar experimental
452: parameter values. Extending to include the subharmonic nature of the
453: instability does not affect the branching of pattern~(a).
454: 
455: The third situation arises with pattern~(c), which on symmetry arguments alone
456: could be associated with either representation~10 or representation~12.
457: Including information about the spatio-temporal symmetry of the pattern does
458: not distinguish between these two representations. However, information on the
459: Fourier transform of the pattern does allow a choice to be made between the two
460: possibilities; in order to show this, we first need to work out which
461: combinations of Fourier modes are associated with each pattern.
462: 
463: It is useful to have sample Fourier modes for the basic hexagonal 
464: pattern:
465:  %
466:  \begin{equation}\label{eq:basichexgons}
467:  f_0(x,y)=    \cos2\pi\left(\frac{2x}{3}\right) +
468:               \cos2\pi\left(-\frac{x}{3}+\frac{y}{\sqrt{3}}\right) +
469:               \cos2\pi\left(-\frac{x}{3}-\frac{y}{\sqrt{3}}\right),
470:  \end{equation}
471:  %
472: with wavevector of length $\frac{4\pi}{3}$, as well as sample Fourier modes
473: for representations~7, 10 and~12. The method described by Tse
474: \etal~\cite{refT53} yields Fourier functions that would be included in the
475: eigenfunctions associated with representation~7; representative functions with
476: the shortest wavevectors include:
477:  %
478:  \begin{align}\label{eq:repn7fns1}
479:  f_1(x,y)&=    \cos2\pi\left(\frac{x}{3}+\frac{y}{3\sqrt{3}}\right) +
480:                \cos2\pi\left(\frac{x}{3}-\frac{y}{3\sqrt{3}}\right) +
481:                \cos2\pi\left(\frac{2y}{3\sqrt{3}}\right)   \\
482:  f_2(x,y)&=    \sin2\pi\left(\frac{x}{3}+\frac{y}{3\sqrt{3}}\right) +
483:                \sin2\pi\left(-\frac{x}{3}+\frac{y}{3\sqrt{3}}\right) +
484:                \sin2\pi\left(-\frac{2y}{3\sqrt{3}}\right),
485:  \label{eq:repn7fns2}
486:  \end{align}
487:  %
488: which is made up of wavevectors of length equal to $\frac{1}{\sqrt{3}}$ of that
489: of the basic hexagonal pattern. Eigenfunctions for representation~10 are made
490: up of Fourier functions that include:
491:  %
492:  \begin{equation}\label{eq:repn10fns}
493:  f_1=\sin2\pi\left(\frac{x}{6}+\frac{y}{2\sqrt{3}}\right)\quad
494:  f_2=\sin2\pi\left(\frac{-x}{6}+\frac{y}{2\sqrt{3}}\right)\quad
495:  f_3=\sin2\pi\left(\frac{-x}{3}\right),
496:  \end{equation}
497:  %
498: with wavevector of length $\frac{1}{2}$ the fundamental; and representation~12
499: has:
500:  %
501:  \begin{equation}\label{eq:repn12fns}
502:  f_1=\sin2\pi\left(\frac{x}{2}+\frac{-y}{2\sqrt{3}}\right)\quad
503:  f_2=\sin2\pi\left(\frac{x}{2}+\frac{y}{2\sqrt{3}}\right)\quad
504:  f_3=\sin2\pi\left(\frac{y}{\sqrt{3}}\right),
505:  \end{equation}
506:  %
507: with wavevector of length $\frac{\sqrt{3}}{2}$ the fundamental. In each case,
508: we have chosen the Fourier modes with the shortest wavevectors, as these are
509: easiest to identify in an experimental Fourier transform. 
510: 
511: The images of the Fourier transform of pattern~(c) in~\cite{refA57} show that
512: the mode created in the instability contains wavevectors that are a factor of~2
513: shorter than the shortest in the basic hexagonal pattern, which is consistent
514: with representation~10 but not~12. In this way, information about the power
515: spectrum of the pattern is necessary to supplement the arguments based
516: entirely on symmetries and to distinguish between the two choices.
517: 
518: \section{Normal forms}\label{sec:normalforms}
519: Using the functions specified above as a basis for representations~7 and~10,
520: the matrices that generate the two relevant representations are, for
521: representation~7:
522:  %
523:  \begin{equation}
524:  R_{\kappa_x}= I_2,    \quad
525:  R_\rho=       \begin{bmatrix}1&0\\0&-1\end{bmatrix},   \quad
526:  R_{\tau_1}=   \begin{bmatrix}-\frac{1}{2}        &  \frac{\sqrt{3}}{2}\\
527:                               -\frac{\sqrt{3}}{2} & -\frac{1}{2} \end{bmatrix},   \quad
528:  R_{\tau_2}=   R_{\tau_1}^2,  \quad
529:  R_{\tau_T}=   -I_2, 
530:  \end{equation}
531:  %
532: where $I_n$ is the $n\times n$ identity matrix; 
533: and for representation~10:
534:  %
535:  \def\AMRpone{\phantom{-}1}
536:  \def\AMRmone{         - 1}
537:  \def\AMRzero{\phantom{-}0}
538:  \begin{align}
539:  R_{\kappa_x}&=  \begin{bmatrix}\AMRzero&\AMRpone&\AMRzero\\\AMRpone&\AMRzero&\AMRzero\\\AMRzero&\AMRzero&\AMRmone\end{bmatrix},  \qquad
540:  R_\rho=         \begin{bmatrix}\AMRzero&\AMRzero&\AMRmone\\\AMRpone&\AMRzero&\AMRzero\\\AMRzero&\AMRpone&\AMRzero\end{bmatrix},  \\
541:  R_{\tau_1}&=    \begin{bmatrix}\AMRmone&\AMRzero&\AMRzero\\\AMRzero&\AMRpone&\AMRzero\\\AMRzero&\AMRzero&\AMRmone\end{bmatrix},  \qquad
542:  R_{\tau_2}=     \begin{bmatrix}\AMRmone&\AMRzero&\AMRzero\\\AMRzero&\AMRmone&\AMRzero\\\AMRzero&\AMRzero&\AMRpone\end{bmatrix},  \qquad
543:  R_{\tau_T}=     -I_3.
544:  \end{align}
545:  %
546: The perturbation amplitude at time $j+1$ times the forcing period, given the
547: perturbation at time~$j$, is given by
548: $\mathbf{a}_{j+1}=\mathbf{f}(\mathbf{a}_j)$, where the equivariance condition
549: amounts to $R_\gamma\mathbf{f}(\mathbf{a})=\mathbf{f}(R_\gamma\mathbf{a})$ for
550: all $\gamma\in\Gamma$. Using this, we can determine the relevant normal form
551: associated with these two representations:
552:  %
553:  \begin{equation}\label{eq:repn7nf}
554:  z_{j+1} = -(1+\mu) z_j + P|z_j|^2z_j + Q|z_j|^4z_j + R{\bar z}^5
555:  \end{equation}
556:  %
557: for representation~7 (truncated at quintic order), where the two amplitudes
558: of~$f_1$ and $f_2$ in (\ref{eq:repn7fns1}--\ref{eq:repn7fns2}) are the real and
559: imaginary parts of~$z$, and $P$, $Q$ and~$R$ are real constants. For
560: representation~10 we truncate at cubic order and obtain:
561:  %
562:  \begin{align}\label{eq:repn10nf1}
563:  a_{j+1} &= -(1+\mu) a_j + Pa_j^3 + Q(a_j^2+b_j^2+c_j^2)a_j, \\
564:  b_{j+1} &= -(1+\mu) b_j + Pb_j^3 + Q(a_j^2+b_j^2+c_j^2)b_j, \\
565:  c_{j+1} &= -(1+\mu) c_j + Pc_j^3 + Q(a_j^2+b_j^2+c_j^2)c_j, 
566:  \label{eq:repn10nf3}
567:  \end{align}
568:  %
569: where $P$ and $Q$ are (different) real constants. In these two sets of
570: equations, $\mu$~represents the bifurcation parameter. The $-1$ Floquet
571: multipliers at $\mu=0$ arise because these are subharmonic bifurcations. In
572: representation~7, equivariance with respect to $R_{\tau_T}=-I_2$ is a normal
573: form symmetry, so even terms up to any order can be removed
574: from~(\ref{eq:repn7nf}) by coordinate transformations~\cite{refE2}. With
575: representation~10, the matrix~$-I_3=R_\rho^3$ appears as a spatial symmetry, so
576: the normal form symmetry is in fact exact, and every solution branch has the
577: spatio-temporal symmetry $\tau_T\rho^3$, a rotation by~$180^\circ$
578: followed by time-translation by one period.
579: 
580: The patterns are neutrally stable with respect to translations in the two
581: horizontal directions, and so also have two Floquet multipliers equal to~$1$
582: associated with translation modes. We have neglected these as all the patterns
583: we find are pinned by reflection symmetries that prohibit drifting.
584: 
585: The final stages are to determine the solutions that are created in each of
586: these bifurcations, their symmetry and stability properties, and to compare 
587: these with experimental observations. 
588: 
589: The first normal form (\ref{eq:repn7nf}) generically has two types of
590: period-two points, found by solving~$f(z)=-z$:
591:  %
592:  \begin{equation}
593:  z_a=          \sqrt{\frac{\mu}{P} - 2\mu^2\frac{Q+R}{P^3}}, \qquad
594:  z_b=\hbox{i}\,\sqrt{\frac{\mu}{P} - 2\mu^2\frac{Q-R}{P^3}}.
595:  \end{equation}
596:  %
597: The first of these has exactly the symmetry group~$\Sigma_a$ of pattern~(a),
598: with no spatio-temporal symmetries, while the second has exactly the spatial
599: symmetry group~$\Sigma_b$ of pattern~(b), as well as spatio-temporal symmetries
600: generated by~$\rho\tau_T$. Reconstructions of these two are shown in
601: figure~\ref{fig:representation7}(a) for pattern~(a) and
602: figure~\ref{fig:representation7}(b,c) for pattern~(b), using the Fourier
603: functions from above. Linearising the normal form about these two period-two
604: points readily yields stability information: if $P>0$, then both patterns are
605: supercritical but only one is stable, while if $P<0$, both are subcritical and
606: neither is stable.
607: 
608:  \begin{figure}
609:  \begin{center}
610:  \centerline{%
611:  \hbox to0.30\hsize{\hfil(a)\hfil}\hfil
612:  \hbox to0.30\hsize{\hfil(b)\hfil}\hfil
613:  \hbox to0.30\hsize{\hfil(c)\hfil}}
614:  \centerline{% 
615:  \epsfxsize0.30\hsize\epsffile{RSF_fig_4a.eps}\hfil
616:  \epsfxsize0.30\hsize\epsffile{RSF_fig_4b.eps}\hfil
617:  \epsfxsize0.30\hsize\epsffile{RSF_fig_4c.eps}}
618:  \caption{Reconstructed patterns from the two solutions that arise in
619: representation~7, using the Fourier functions
620: (\ref{eq:repn7fns1}--\ref{eq:repn7fns2}) added to a function of the form
621: of~(\ref{eq:basichexgons}). (a) has the spatial symmetries of pattern~(a) and
622: no spatio-temporal symmetries (cf.~\ref{fig:idealised}a,d); (b) has the
623: symmetry properties of pattern~(b) (c is one period~$T$ later;
624: cf.~figure~\ref{fig:idealised}b,e,g)}
625:  \label{fig:representation7}
626:  \end{center}
627:  \end{figure}
628: 
629:  \begin{figure}
630:  \begin{center}
631:  \centerline{%
632:  \hbox to0.22\hsize{\hfil(a)\hfil}\hfil
633:  \hbox to0.22\hsize{\hfil(b)\hfil}\hfil
634:  \hbox to0.22\hsize{\hfil(c)\hfil}\hfil
635:  \hbox to0.22\hsize{\hfil(d)\hfil}}
636:  \centerline{% 
637:  \epsfxsize0.22\hsize\epsffile{RSF_fig_5a.eps}\hfil
638:  \epsfxsize0.22\hsize\epsffile{RSF_fig_5b.eps}\hfil
639:  \epsfxsize0.22\hsize\epsffile{RSF_fig_5c.eps}\hfil
640:  \epsfxsize0.22\hsize\epsffile{RSF_fig_5d.eps}}
641:  \centerline{%
642:  \hbox to0.22\hsize{\hfil(e)\hfil}\hfil
643:  \hbox to0.22\hsize{\hfil(f)\hfil}\hfil
644:  \hbox to0.22\hsize{\hfil(g)\hfil}\hfil
645:  \hbox to0.22\hsize{\hfil(h)\hfil}}
646:  \centerline{% 
647:  \epsfxsize0.22\hsize\epsffile{RSF_fig_5e.eps}\hfil
648:  \epsfxsize0.22\hsize\epsffile{RSF_fig_5f.eps}\hfil
649:  \epsfxsize0.22\hsize\epsffile{RSF_fig_5g.eps}\hfil
650:  \epsfxsize0.22\hsize\epsffile{RSF_fig_5h.eps}}
651:  \caption{Reconstructed patterns from irreps~10 and~12: 
652:  (a,b)~10: $(a,b,c)=(1,1,0)$ (cf.~figure~\ref{fig:idealised}c,f,h);
653:  (c,d)~12: same amplitudes and same symmetries as (a,b);
654:  (e,f)~10: $(a,b,c)=(1,0,0)$;
655:  (g,h)~10: $(a,b,c)=(1,1,1)$.}
656:  \label{fig:representation1012}
657:  \end{center}
658:  \end{figure}
659: 
660: The second normal form (\ref{eq:repn10nf1}--\ref{eq:repn10nf3}) generically has
661: three types of period-two points~$(a,b,c)$:
662:  %
663:  \begin{equation}\label{eq:repn10branches}\quad
664:  \sqrt{\frac{\mu}{P+Q}}
665:    \begin{pmatrix}1\\0\\0\end{pmatrix},\quad
666:  \sqrt{\frac{\mu}{P+2Q}}
667:    \begin{pmatrix}1\\1\\0\end{pmatrix},\quad
668:  \sqrt{\frac{\mu}{P+3Q}}
669:    \begin{pmatrix}1\\1\\1\end{pmatrix}.\quad
670:  \end{equation}
671:  %
672: The middle branch has the spatio-temporal symmetries of pattern~(c), with
673: 12~elements in the spatial part of the symmetry
674: group~($\Sigma_c=\langle\kappa_x,\kappa_y\tau_2,\tau_1^2\rangle$).
675: Figure~\ref{fig:representation1012}(a,b) illustrates this pattern
676: (cf.~figure~\ref{fig:idealised}c,f,h). For comparison, the pattern that would
677: have been obtained with modes from representation~12 is in
678: figure~\ref{fig:representation1012}(c,d): the symmetry group is the same, but
679: the appearance of the pattern does not match the experimental observation. The
680: first branch has a 24~element spatial symmetry group
681: $\langle\rho^3\tau_1,\kappa_x\rho\tau_1^5\tau_2,\tau_1^2\rangle$
682: (figure~\ref{fig:representation1012}e,f), and the third branch has an
683: 18~element group $\langle\kappa_y\tau_2,\kappa_x\rho^5,\tau_1^2\rangle$
684: (figure~\ref{fig:representation1012}g,h). The three patterns also have the
685: spatio-temporal symmetry~$\rho^3\tau_T$ (since $R_\rho^3=-I_3$), so $\rho^3$
686: will appear in the symmetry group of the time-average of each of the
687: patterns, as discussed in~\cite{refT53}.
688: 
689: The first branch has Floquet multipliers $-1+2\mu$ and $-1-\frac{P}{P+Q}\mu$
690: (twice); the second branch $-1+2\mu$, $-1-\frac{P}{P+2Q}\mu$ and
691: $-1+\frac{2P}{P+2Q}\mu$; and the third branch $-1+2\mu$ and
692: $-1+\frac{2P}{P+3Q}\mu$ (twice). As a result, if $P+Q>0$ and $P+3Q>0$, then all
693: branches bifurcate supercritically, and either the first branch will be stable
694: (when~$P<0$) or the last will be stable (when~$P>0$). If any branch bifurcates
695: subcritically, none are stable. The middle branch, which is the one
696: corresponding to the experimentally observed pattern~(c), is always unstable at
697: onset. 
698: %This is discussed in more detail below.
699: 
700: \section{Discussion}\label{sec:discussion}
701: Using the symmetry-based approach of Tse \etal~\cite{refT53}, we have
702: analysed three experimentally observed spatial period-multiplying transitions
703: from an initial hexagonal pattern. The three patterns illustrate three
704: situations that can arise in this kind of analysis. Pattern~(a) was
705: straight-forward, in that a single representation of~$\Gamma$ had a
706: one-dimensional space fixed by the spatial symmetry group of the pattern. The
707: existence of a solution branch of the form of pattern~(b) could also be
708: inferred using the Equivariant Branching Lemma, though in this case it was
709: necessary to include the temporal symmetry associated with period-doubling
710: bifurcation. Specifically, the spatial symmetries selected a two-dimensional
711: fixed point space which was further reduced to a one-dimensional fixed point
712: space when spatio-temporal symmetries were taken into account. Experimentally,
713: these two patterns were found for the same fluid parameters and same
714: $2\omega:3\omega$ forcing function but for different frequencies~$\omega$:
715: $\omega=25\,\hbox{Hz}$ for~(a) and $\omega=35\,\hbox{Hz}$ for~(b). This
716: suggests that the transition between these patterns, which arise for
717: instabilities associated with the same representation, might be observed by
718: tuning the frequency~$\omega$.
719: 
720: Pattern~(c), on the other hand, had a spatial symmetry group that fixed
721: one-dimensional subspaces in two different representations, and we appealed to
722: the measured power spectrum of the pattern to choose between the two
723: possibilities. In this situation, symmetry considerations alone were not
724: enough. Similar situations arise in other bifurcation problems, for example,
725: knowing that a stable axisymmetric pattern is found in a spherically symmetric
726: bifurcation problem does not provide enough information to determine which is
727: the relevant representation. 
728: 
729: The experimentally observed transition between hexagons and pattern~(c) occurs
730: by means of a propagating front that separates domains of hexagons and the
731: secondary pattern. The front is initiated at the lateral boundaries of the
732: system and emanates radially inward. There is little if any hysteresis, and the
733: reverse transition also occurs via the same scenario. The occurrence of a front
734: in this transition suggests bistability of the hexagonal pattern and
735: pattern~(c).  This is certainly consistent with the theoretical prediction that
736: pattern~(c) is unstable at small amplitude, that is, at onset. However, we have
737: not explored the possible stabilization mechanisms for pattern~(c).
738: 
739: It is worth emphasizing that an understanding of group representation theory is
740: useful in classifying and analysing secondary instabilities of patterns,
741: not only in the Faraday wave experiment as described here, but also in
742: convection and other pattern formation problems (see~\cite{refR59}). It
743: is also worth mentioning that the examples studied here indicate that
744: spatio-temporal symmetries readily arise in secondary subharmonic
745: instabilities, and that careful experimental characterization of these, either
746: by still images taken one forcing period apart or by time-averaging over two
747: forcing periods, can be helpful. Subsequent instabilities of patterns that have
748: spatio-temporal symmetries can be analysed using methods described
749: in~\cite{refR40,refL51}.
750: 
751: The approach outlined in~\cite{refT53} and here is useful for taking an
752: experimental observation of a secondary transition and casting it into its
753: equivariant bifurcation theory context, but it does not predict which
754: transitions should be expected in an experiment. However, in these
755: two-frequency Faraday wave experiments, three-wave interactions of the type
756: described in~\cite{refS105} may select a third wavevector that could appear in
757: the secondary transition. Each of the representations in the problem under
758: consideration is associated with a set of wavevectors, providing a possible
759: mechanism for selecting between possibilities.
760: 
761: \noindent{\bf Acknowledgements.} This paper builds on earlier published results
762: obtained with Dawn Tse, Rebecca Hoyle and Hagai Arbell. AMR is grateful for
763: support from the EPSRC. The research of MS is supported in part by NSF grant
764: DMS-9972059 and NASA grant NAG3-2364. JF is grateful for support from the
765: Israel Academy of Science (grant 203/99).
766:  
767: \begin{thebibliography}{99} 
768: 
769: \itemsep=\smallskipamount 
770: 
771: \def\author#1#2{#2, #1}%
772: \def\and{\&\ }
773: 
774: \def\PT{Phil. Trans. R.~Soc. Lond.~A}
775: \def\PRS{Proc. R.~Soc. Lond.~{\rm A}}
776: \def\JFM{J.~Fluid Mech.}
777: \def\GAFD{Geophys. Astrophys. Fluid Dynamics}
778: \def\PCPS{Proc. Camb. Phil. Soc.}
779: \def\CUP{Cambridge University Press}
780: \def\PRL{Phys. Rev. Lett.}
781: \def\PRE{Phys. Rev. E}
782: 
783: \def\refarticle#1#2#3#4#5#6#7{%
784: \bibitem{#1} #2, {\em #4,} #5, {\bf #6} (#3), {#7}.}
785: 
786: %\def\refarticleshort#1#2#3#4{%
787: %\bibitem{#1} #2 (#3) #4}
788: 
789: \def\refbook#1#2#3#4#5#6{%
790: \bibitem{#1} #2 (#3) {\it #4}. #5: #6.}
791: 
792: \def\refartbook#1#2#3#4#5#6#7#8#9{%
793: \bibitem{#1} #2 (#3) #4. In {\it #5}
794: \begingroup\def\author##1##2{##1 ##2}(ed. #6),\endgroup\ pp.~#9. #7: #8}
795: 
796: \refarticle{refT53}                                                 
797: {\author{D.P.}{Tse}, \author{A.M.}{Rucklidge}, \author{R.B.}{Hoyle} 
798:  \and \author{M.}{Silber}}{2000}                                            
799:             {Spatial period-multiplying instabilities of  
800:              hexagonal Faraday waves}
801:             {Physica}{146D}{367--387}                        
802: 
803: \refarticle{refE10}
804: {\author{W.S.}{Edwards} \and \author{S.}{Fauve}}{1994}
805:             {Patterns and quasi-patterns in the Faraday experiment}
806:             {\JFM}{278}{123--148}
807: 
808: \refarticle{refK84}
809: {\author{A.}{Kudrolli} \and \author{J.P.}{Gollub}}{1996}
810:             {Patterns and spatiotemporal chaos in parametrically forced
811:               surface waves: a systematic survey at large aspect ratio}
812:             {Physica}{97D}{133--154}
813: 
814: \refartbook{refM118}
815: {\author{H.W.}{M\"{u}ller}, \author{R.}{Friedrich} 
816:  \and \author{D.}{Papathanassiou}}{1998}
817:            {Theoretical and experimental investigations of the Faraday 
818:             instability}
819:            {Evolution of Spontaneous Structures in Dissipative Continuous 
820:             Systems}
821:            {\author{F.H.}{Busse} \and \author{S.C.}{M\"uller}}
822:            {Springer}{Berlin}{230--265}
823: 
824: \refbook{refG59}
825: {\author{M.}{Golubitsky}, \author{I.}{Stewart} \and \author{D.G.}{Schaeffer}}
826:     {1988}
827:     {Singularities and Groups in Bifurcation Theory. Volume~II}
828:     {Springer}{New York}
829: 
830: \refarticle{refK67}
831: {\author{A.}{Kudrolli}, \author{B.}{Pier} \and \author{J.P.}{Gollub}}{1998}
832:             {Superlattice patterns in surface waves}
833:             {Physica}{123D}{99--111}
834: 
835: \refarticle{refA57}                   
836: {\author{H.}{Arbell} \and \author{J.}{Fineberg}}{1998}              
837:             {Spatial and temporal dynamics of two interacting modes in
838:              parametrically driven surface waves}                           
839:             {\PRL}{81}{4384--4387}                        
840: 
841: \refarticle{refL61}
842: {\author{O.}{Lioubashevski}, \author{H.}{Arbell} \and 
843:  \author{J.}{Fineberg}}{1996}
844:             {Dissipative solitary states in driven surface waves}
845:             {\PRL}{76}{3959--3962}      
846: 
847: \refarticle{refE2}                                                            
848: {\author{C.}{Elphick},                                          
849:  \author{E.}{Tirapegui},                                             
850:  \author{M.E.}{Brachet},                                   
851:  \author{P.}{Coullet} \and                                      
852:  \author{G.}{Iooss}}{1987}                                     
853:             {A simple global characterization for normal forms of
854:              singular vector fields}                  
855:             {Physica}{29D}{95--127}      
856: 
857: \refarticle{refR59}
858: {\author{A.M.}{Rucklidge}, \author{N.O.}{Weiss},
859:  \author{D.P.}{Brownjohn}, \author{P.C.}{Matthews} 
860:  \and \author{M.R.E.}{Proctor}}{2000}
861:             {Compressible magnetoconvection in three dimensions:  
862:              pattern formation in a strongly stratified layer}
863:             {\JFM}{419}{283--323}
864: 
865: \refarticle{refR40}                                             
866: {\author{A.M.}{Rucklidge} \and \author{M.}{Silber}}{1998}                   
867:             {Bifurcations of periodic orbits with spatio-temporal symmetries}
868:             {Nonlinearity}{11}{1435--1455}                        
869: 
870: \refartbook{refL51}
871: {\author{J.S.W.}{Lamb} \and \author{I.}{Melbourne}}{1999}
872:             {Bifurcation from periodic solutions with spatiotemporal symmetry}
873:             {Pattern Formation in Continuous and Coupled Systems}
874:             {\author{M.}{Golubitsky}, \author{D.}{Luss} \and 
875:              \author{S.H.}{Strogatz}}
876:             {Springer}{New York}{175--191}
877: 
878: \refarticle{refS105}                                                            
879: {\author{M.}{Silber}, \author{C.M.}{Topaz} \and \author{A.C.}{Skeldon}}{2000}
880:             {Two-frequency forced Faraday waves: 
881:              weakly damped modes and patterns selection}
882:             {Physica}{143D}{205--225}
883: 
884: 
885: \end{thebibliography} 
886:  
887: \end{document} 
888: 
889: 
890: 
891: