1: \documentclass[aps,prl,twocolumn,showpacs,amsmath,amssymb]{revtex4}
2: %\documentclass[preprint,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
3: %\documentclass[preprint,aps]{revtex4}
4: %\documentclass[preprint,aps,draft]{revtex4}
5:
6: \usepackage{graphicx}% Include figure files
7: \usepackage{dcolumn}% Align table columns on decimal point
8: \usepackage{bm}% bold math
9:
10: \newcommand{\conjugate}[1]{\overline{#1}}
11: \renewcommand{\exp}[1]{e^{#1}}
12: \renewcommand{\vec}[1]{\mathbf{#1}}
13: \newcommand{\mat}[1]{\mathbf{#1}}
14: \newcommand{\order}{\mathcal{O}}
15: \newcommand{\degree}{^{\circ}}
16: \newcommand{\re}{\mathop{\mathrm{Re}}}
17: \newcommand{\im}{\mathop{\mathrm{Im}}}
18: \newcommand{\gcf}{\mathop{\mathrm{gcf}}}
19: \newcommand{\Dhat}{\mathop{\widehat{\mathcal{D}}}}
20: \newcommand{\sgn}{\mathop{\mathrm{sign}}}
21: \newcommand{\oneletter}[1]{#1}
22:
23:
24: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
25: \begin{document}
26:
27: \def\bs{\boldsymbol}
28:
29: \title{Faraday Wave Pattern Selection Via Multi-Frequency Forcing}
30: \author{Jeff Porter}
31: \email{jport@maths.leeds.ac.uk}
32: \affiliation{Department of Applied Mathematics, University of Leeds,
33: Leeds LS2 9JT, United Kingdom}
34: \author{Chad M. Topaz}
35: \affiliation{Department of Mathematics, U.C.L.A., Los Angeles, California 90095}
36: \author{Mary Silber}
37: \affiliation{Department of Engineering Sciences and Applied Mathematics,
38: Northwestern University, Evanston, Illinois 60208}
39:
40: \date{\today}
41:
42: \begin{abstract}
43: We use symmetry considerations to investigate how damped modes affect pattern
44: selection in multi-frequency forced Faraday waves. We classify and tabulate the
45: most important damped modes and determine how the corresponding resonant
46: triad interactions depend on the forcing parameters. The relative phase of the
47: forcing terms may be used to enhance or suppress the nonlinear interactions. We
48: compare our predictions with numerical results and discuss their implications for
49: recent experiments. Our results suggest how to design multi-frequency forcing
50: functions that favor chosen patterns in the lab.
51: \end{abstract}
52:
53: \pacs{05.45.-a, 47.35.+i, 47.54.+r, 89.75.Kd}
54:
55: \maketitle
56:
57: Faraday observed in 1831 that patterns of subharmonic standing waves form on the
58: surface of a fluid when its supporting container is shaken vertically with increasing
59: strength \cite{Far1831}. The first Faraday wave experiments (see
60: \cite{MilHen90} for a review) used sinusoidal forcing to shake the fluid,
61: and produced simple patterns (\emph{e.g.}, stripes, squares, and hexagons). Remarkably,
62: this experiment remains a rich source of intriguing new patterns, most recently as the
63: basic sinusoidal forcing has been replaced by periodic functions with two or three
64: frequency components. The two-frequency case has been investigated a great deal
65: experimentally \cite{Mul93,EdwFau94,ArbFin98,KudPieGol98,ArbFin02} and
66: theoretically \cite{BesEdwTuc96,ZhaVin97_2,SilTopSke00,PorSil02}, and has produced
67: a variety of states, including superlattice patterns and localized structures, as well as the
68: first experimental realizations of quasipatterns. Three-frequency forcing has been used
69: too \cite{Mul93,ArbFin02}, but to far less extent.
70:
71: A primary focus of the past 10 years of investigation of one- and two-frequency forced
72: Faraday waves has been on the role of resonant triad interactions -- the lowest order
73: nonlinear interactions -- in pattern selection. A typical resonant triad consists of two
74: critical modes and a third, damped mode with a distinct wavelength that determines
75: the angle $\theta_{\rm res}$ between the critical wave vectors; see Fig.~\ref{fig:restriad}.
76: It was originally thought that the damped mode should draw energy from the excited modes,
77: creating an ``anti-selection'' mechanism that suppresses such triads in favor of patterns that
78: avoid the resonant angle \cite{ZhaVin97_1}. However, it has since been shown that the
79: opposite effect may also occur. For instance, \cite{SilTopSke00,TopSil02} demonstrated
80: for two-frequency forced Faraday waves that a damped mode oscillating at the difference
81: of the two forcing frequencies is responsible for selecting the superlattice-I pattern
82: reported in~\cite{KudPieGol98}.
83:
84: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
85: \begin{figure}
86: \includegraphics[width=2.6in]{fig1.eps}
87: \caption{Fourier space diagram of resonant triads composed of two critical modes
88: ($|\vec{k}_1|=|\vec{k}_2|=k_1$) and a damped mode ($\vec{k}_3=\vec{k}_1+\vec{k}_2$,
89: $|\vec{k}_3|=k_3$) with (a) $k_3 <k_1$, (b) $k_1 < k_3 < 2k_1$. The resonant angle is
90: given by $\cos (\theta_{\rm res}/2) = k_3/(2k_1)$.}
91: \label{fig:restriad}
92: \end{figure}
93: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
94:
95: The richness of the Faraday system is due in large part to the vastness of the control
96: parameter space which, in principle, is infinite-dimensional. By expanding an arbitrary
97: periodic forcing function in a Fourier series
98: \begin{align}
99: F(t) &= f_m e^{im\omega t} + f_n e^{i n \omega t} + f_p e^{i p \omega t} + \cdots + c.c.,
100: \end{align}
101: it is apparent that one has the freedom to choose the commensurate frequencies
102: $(m\omega,n\omega,p\omega,\ldots)$, the amplitudes $(|f_m|,|f_n|,|f_p|,\ldots)$, and the phases
103: $(\phi_m,\phi_n,\phi_p,\ldots)$, where $\phi_u = \arg(f_u)$. Previous experimental and
104: theoretical investigations have largely been exploratory in nature, \emph{i.e.}, they have
105: focused on describing the patterns which form at particular locations in parameter
106: space. In this Letter, we take a prescriptive approach to pattern formation and
107: demonstrate how the huge parameter space may be exploited to control resonant triads.
108: In particular, we can either enhance or suppress triad interactions, largely at
109: our discretion, by an appropriate choice of forcing frequencies, amplitudes, and phases.
110:
111: Our analysis follows that of \cite{PorSil02} which uses the (broken) symmetries of time
112: translation, time reversal, and Hamiltonian structure, and is valid near onset in weakly-damped
113: systems. Our main result is a table summarizing, for up to three forcing frequencies, which
114: damped modes are likely to be important, the manner in which their coupling with critical
115: modes depends on the forcing parameters and, in many cases, the overall qualitative effect
116: (enhancing or suppressing) they have on associated patterns. These results are then applied
117: to two specific examples relevant to experiments. In the first, we show how the superlattice
118: pattern selection mechanism discussed in \cite{SilTopSke00,TopSil02} may be dramatically
119: enhanced by the addition of a third frequency component with appropriate phase. In the
120: second, we offer an explanation for why a distorted 8-fold quasipattern observed with two-frequency
121: forcing became much more regular and robust when a third forcing frequency component was
122: applied in the experiments of \cite{ArbFin02}.
123:
124: We consider resonant triads composed of two critical modes and a third linearly damped
125: mode (which may, nevertheless, be forced) oscillating with dominant frequency $\Omega > 0$.
126: The particular values of $\Omega$ most crucial to pattern selection will be determined in the
127: course of our calculation. Time is rescaled so that the common frequency is one, and we take
128: the $m$ component of the forcing to drive the critical modes, which therefore have dominant
129: frequency $m/2$. All other modes are linearly damped. We begin by expanding the fluid surface
130: height $h$ in terms of six traveling wave (TW) modes:
131: \begin{align}
132: h({\bs x},t) = \sum_{j=1}^3\sum_{\pm} Z_j^\pm (t)\,e^{i({\bs k_j}\cdot
133: {\bs x}\pm \varpi_j t)} + c.c. + ...\,,
134: \label{eq:h}
135: \end{align}
136: where $\varpi_1=\varpi_2=m/2$ and $\varpi_3=\Omega$. The TW amplitude equations respect
137: the spatial symmetries of the problem (translation, reflection through ${\bs k}_3$, inversion through
138: the origin) and the temporal symmetries
139: \begin{subequations}
140: \label{eq:tempsyms}
141: \begin{align}
142: T_\tau &: Z_j^{\pm} \rightarrow e^{\pm i \varpi_j \tau}Z_j^{\pm},
143: \:\: f_u \rightarrow e^{i u \tau} f_u , \\
144: \kappa &: Z_j^{\pm} \leftrightarrow Z_j^{\mp},\:\: (t,\gamma) \rightarrow -(t,\gamma),
145: \:\: f_u \rightarrow \bar{f}_u\,.
146: \end{align}
147: \end{subequations}
148: Here $u$ denotes any of the frequencies $\{m,n,p,\cdots\}$ and $\gamma = 2 \nu k^2$
149: is a dimensionless damping parameter formed from the kinematic viscosity $\nu$ and the
150: characteristic wave number $k(m/2)$, determined by the hydrodynamic dispersion
151: relation. Note that $T_\tau$ and $\kappa$ are not symmetries in the usual sense; time
152: translation $T_{\tau}$ is broken by the forcing and time reversal $\kappa$ by the damping.
153: Nonetheless, they are valid {\it parameter symmetries} of the problem.
154:
155: The equivariant TW equations take the form
156: \begin{subequations}
157: \label{eq:TW}
158: \begin{align}
159: \dot{Z}_1^+ & = \upsilon Z_1^+ - \lambda f_m Z_1^-
160: + {\cal Q}_1(Z_2^\pm,Z_3^\pm) + \cdots,\\
161: \dot{Z}_3^+ & = \varrho Z_1^+ - \mu F_{2\Omega} Z_3^-
162: + {\cal Q}_3(Z_1^\pm,Z_2^\pm) + \cdots,
163: \end{align}
164: \end{subequations}
165: with the remaining four equations related by symmetry. The parametric forcing term
166: $F_{2\Omega}$ represents products of the $f_u$ whose frequencies sum to $2\Omega$
167: ($F_{2\Omega}=f_{2\Omega}$ when $2\Omega$ forcing is present). The resonant terms
168: \begin{align*}
169: {\cal Q}_1 & = (Q_1Z_3^+ + Q_2Z_3^-)\bar{Z}_2^+ + (Q_3Z_3^+ + Q_4Z_3^-)\bar{Z}_2^- ,\\
170: {\cal Q}_3 & = Q_5 Z_1^+Z_2^+ + Q_6(Z_1^+Z_2^- + Z_1^-Z_2^+) + Q_7Z_1^-Z_2^-,
171: \end{align*}
172: couple the two modes at quadratic order. The resonant coefficients $Q_\ell$, according to $T_\tau$, must
173: transform as $(Q_1,\bar{Q}_5) \sim e^{i(m-\Omega)\tau}$, $(Q_2,Q_7) \sim e^{i(m+\Omega)\tau}$,
174: $(Q_3,\bar{Q}_4,\bar{Q}_6) \sim e^{-i\Omega \tau}$. The dependence of the $T_\tau$-invariant
175: coefficients $\upsilon$, $\lambda$, etc., on $\gamma$ and the $f_u$, is also determined by
176: symmetry. In particular, $\kappa$ forces most of them to be purely imaginary at
177: leading order. However, if $k_1$ and $k_3$ are defined by the local minima of the neutral stability
178: curves then the imaginary parts of $\upsilon$ and $\varrho$ (\emph{i.e.}, the detunings) vanish
179: at $\gamma =0$. We will need only the leading term in each coefficient so we write
180: \begin{align*}
181: &\upsilon = -\upsilon_r \gamma, \;\; \varrho = -\varrho_r \gamma, \;\;
182: \lambda = i \lambda_i, \;\; \mu= i \mu_i, \;\; Q_\ell = i q_\ell F_\ell ,
183: \end{align*}
184: where $\upsilon_r, \varrho_r> 0$ (they correspond to damping terms), $\lambda_i,\mu_i>0$
185: (see \cite{PorSil_pre}), $q_\ell \in \mathbb{R}$ and $F_\ell$ denotes an appropriate product
186: of the $f_u$ (or unity). The critical value of $|f_m|$ is ${\cal O}(\gamma)$
187: and all $|f_u|$ are assumed to be related by a finite $\gamma$-independent ratio, so that
188: $|f_u| \sim \gamma$.
189:
190: For a given damped mode, the task is to determine how the resonant coefficients $Q_\ell$
191: depend on the $f_u$, then apply a standard reduction procedure at the bifurcation to standing
192: waves (SW). The resulting two equations (truncated at cubic order) describe SW dynamics
193: near onset:
194: \begin{align}
195: \dot{A}_1 = \lambda A_1 + A_1\! \left(a|A_1|^2 + (b_0 + b_{\rm res})|A_2|^2\right).
196: \label{eq:SW}
197: \end{align}
198: The equation for $A_2$ is obtained by switching labels. The self-interaction coefficient $a$
199: and the ``nonresonant'' part $b_0$ of the cross-coupling coefficient are ${\cal O}(\gamma)$.
200:
201: Of primary interest here is the contribution $b_{\rm res}$ makes to the cross-coupling
202: coefficient in (\ref{eq:SW}) as a result of the slaved modes $Z_3^\pm$. Loosely
203: speaking, if $b_{\rm res}>0$ the stability of patterns involving critical modes separated by
204: the angle $\theta_{\rm res}$ will be enhanced, whereas $b_{\rm res}<0$ has a suppressing
205: effect \cite{SilTopSke00}. Since we are interested only in damped modes that have a
206: significant impact on (\ref{eq:SW}), we consider those cases where $b_{\rm res}$ is
207: ${\cal O}(\gamma)$ or larger \footnote{One can amplify the effect of modes with weaker
208: coupling by bringing them extremely close to onset but this is effectively a {\it bicritical}
209: situation which we avoid here.}.
210:
211: There are then two possibilities for obtaining large $b_{\rm res}$. If some of the $Q_\ell$
212: are ${\cal O}(1)$ then $b_{\rm res}$ is ${\cal O}(\gamma^{-1})$ and dominates in the limit
213: $\gamma \rightarrow 0$; this happens only for $\Omega = m$ and is in essence a single-frequency
214: phenomenon (the ``first harmonic" resonance). If $\Omega \neq m$ and some of the $Q_\ell$
215: are ${\cal O}(\gamma)$ then $b_{\rm res}$ is ${\cal O}(\gamma)$, and hence comparable to
216: $b_0$. This occurs when $\Omega \in \{2m, n, m \pm n, n-m \}$ for some frequency $n$.
217: However, the $\Omega=2m$ mode may be ignored because for Faraday waves it has a wavenumber
218: $k_3 > 2k_1$ and so it cannot satisfy the necessary {\it spatial} resonance condition. Each of the
219: remaining conditions on $\Omega$ generates a particular type of coupling. For example, with
220: $\Omega = n-m$ we have $F_2 = F_7 = f_n$. If more than one condition is satisfied, as when
221: $\Omega = n-m = m-p$, there are more coupling terms (here one would also have
222: $F_ 1 = \bar{F}_5 = f_p$). The maximum number of conditions that $\Omega$ can satisfy
223: simultaneously is three.
224:
225: In addition to the issue of coupling terms, it matters whether or not the damped mode is parametrically
226: forced at ${\cal O}(\gamma)$. This forcing is present when there is a frequency $2\Omega$ in $F(t)$
227: and magnifies the resonance effect as it brings the damped mode closer to criticality. In addition, it
228: leads to interesting phase dependence as the ``usual" phase, dictated by $f_{2\Omega}$, competes with
229: the phase of the nonlinear forcing ${\cal Q}_3$ (see Eqs.~\ref{eq:TW}).
230:
231: In Table \ref{tab:3ff} we give the leading contribution to $b_{\rm res}$ of the important damped modes
232: for different choices of $F(t)$ containing up to three frequencies $(m,n,p)$.
233: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
234: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
235: \begingroup
236: \squeezetable
237: \begin{table*}
238: \caption{\label{tab:3ff}Leading resonant contribution $b_{\rm res}$ in (\ref{eq:SW}) for the
239: most important damped modes under appropriate choice of three-frequency forcing. Here,
240: $m,n,p,\Omega > 0$ and $x\in\mathbb{Z}^+$. Each expression for $(m,n,p)$, given $\Omega$, is
241: excluded from those of entries further down the table. Dots in the first column indicate an arbitrary
242: commensurate frequency. For $\star$ the $\pm$ follows sign($m-n$).}
243: %\begin{footnotesize}
244: \begin{tabular*}{\textwidth}{@{\extracolsep{\fill}}lccl}
245: \hline \hline
246: $(m,n,p)$ & $\Omega$ & Leading resonant contribution $b_{\rm res}$ & relevant phase(s) \\
247: \hline \hline
248: $(m,\cdot, \cdot)$ &$m$ &$-\alpha_1/|\varrho|$ & \\
249: $(m,n, \cdot)$ &$n$ &$-\alpha_3|f_n|^2/|\varrho|$ & \\
250: $(m,n, \cdot)$ &$m \pm n$ &$-\alpha_1|f_n|^2/|\varrho|$ & \\
251: $(m,n, \cdot)$ &$n-m$ &$\alpha_2|f_n|^2/|\varrho|$ & \\
252: \hline \hline
253: $(m,2m, \cdot)$ &$m$ &$-\alpha_1P_n(\Phi)$ &$\Phi=\phi_n-2\phi_m$\\
254: $(m,n,2n)$ &$n$ &$-\alpha_3|f_n|^2P_p(\Phi)$ &$\Phi=2\phi_n-\phi_p$ \\
255: $(3x,2x, \cdot)$ &$x$ &$-\alpha_1|f_n|^2P_n(\Phi)$ &$\Phi=3\phi_n-2\phi_m$ \\
256: $(m,n,2m \pm 2n)$ &$m \pm n$ &$-\alpha_1|f_n|^2P_p(\Phi)$ &$\Phi=\phi_p -2\phi_m \mp 2\phi_n$ \\
257: $(m,n,2n-2m)$ &$n-m$ &$\alpha_2|f_n|^2P_p(\Phi)$ &$\Phi=\phi_p + 2\phi_m - 2\phi_n $\\
258: \hline \hline
259: $(m,n,|m-n|)$ &$n$ &$(-\alpha_1|f_p|^2- \alpha_3|f_n|^2+ \alpha_5|f_n||f_p|\sin\Phi)/|\varrho|$
260: &$\Phi=\phi_n - \phi_m \pm \phi_p \,\star$ \\
261: $(m,n,m+n)$ &$n$ &$(\alpha_2|f_p|^2- \alpha_3|f_n|^2 + \alpha_6|f_n||f_p|\sin\Phi)/|\varrho|$
262: &$\Phi=\phi_m + \phi_n - \phi_p$ \\
263: $(m,n,2m \pm n)$ &$m \pm n$ &$(\alpha_2|f_p|^2 -\alpha_1|f_n|^2 + \alpha_4|f_n||f_p|\cos\Phi)/|\varrho|$
264: &$\Phi=2 \phi_m - \phi_p \pm \phi_n$ \\
265: \hline \hline
266: $(3,1,2)$ &$1$ &$-\alpha_1|f_p|^2 P_p(\Phi_1 - \Phi_2) - \alpha_3|f_n|^2 P_p(\Phi_1 + \Phi_2)$
267: &$\Phi_1=\phi_n - \phi_m + \phi_p$ \\
268: & &$\mbox{}+ \alpha_5|f_n||f_p|R_p(\Phi_1,\Phi_2)$ &$\Phi_2=\phi_m + \phi_n - 2\phi_p$ \\
269: $(3,2,4)$ &$1$&$-\alpha_1|f_n|^2 P_n(\Phi_1 + \Phi_2) + \alpha_2|f_p|^2 P_n(\Phi_2 - \Phi_1)$
270: &$\Phi_1=\phi_n + \phi_p - 2\phi_m$ \\
271: & &$+ \alpha_4|f_n||f_p|R_n(\Phi_1 - 90\degree,\Phi_2 + 90\degree)$ &$\Phi_2=2\phi_n - \phi_p$ \\
272: \hline \hline
273: \end{tabular*}
274: %\end{footnotesize}
275: \end{table*}
276: \endgroup
277: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
278: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
279: To simplify the expressions therein, we define
280: \begin{align}
281: \label{eq:alphas}
282: &\alpha_1 = q_{1}q_{5}, \quad \alpha_2 = q_{2}q_{7}, \quad \alpha_3
283: = 2q_{6}(q_{3}-q_{4}), \nonumber \\
284: &\alpha_4 = q_{1}q_{7}-q_{2}q_{5}, \quad
285: \alpha_5 = \epsilon_\lambda (2q_{1}q_{6}+q_{5}(q_{3}-q_{4})), \nonumber \\
286: &\alpha_6 = \epsilon_\lambda ( 2q_{2}q_{6}-q_{7}(q_{3}-q_{4})),
287: \quad \epsilon_\lambda = {\rm sign}(\lambda_i),
288: \end{align}
289: and the functions
290: \begin{align}
291: &P_{2\Omega}(\Phi)= \frac{(|\varrho|+\mu_i|f_{2\Omega}|\sin\Phi)}
292: {(|\varrho|^2-|\mu_if_{2\Omega}|^2)}, \label{eq:P}\\
293: &R_{2\Omega}(\Phi_1,\Phi_2)= \frac{(|\varrho|\sin\Phi_1+\mu_i|f_{2\Omega}|\cos\Phi_2)}
294: {(|\varrho|^2-|\mu_i f_{2\Omega}|^2)}. \label{eq:R}
295: \end{align}
296:
297: There are four groupings in the table. The first shows the five important damped modes
298: and their contribution to $b_{\rm res}$ when there is only one type of coupling at
299: ${\cal O}(\gamma)$ or lower and no parametric forcing $f_{2\Omega}$. In these cases
300: there is no (leading order) dependence on the forcing phases. In the second section the
301: same damped modes have been parametrically forced. The factor $1/|\varrho|$ is then
302: replaced by $P_{2\Omega} (\Phi)$ of (\ref{eq:P}), a strictly positive oscillatory function
303: ($|\varrho| > |\mu_if_{2\Omega}|$ for damped modes) with extrema at $\Phi = \pm 90^\circ$.
304: Entries in the third section display two types of coupling, while the final two cases in the table
305: have $f_{2\Omega}$ forcing as well. Note that equivalent cases can be trivially generated
306: from those in Table~\ref{tab:3ff} by switching $n$ and $p$ and relabeling, for example
307: $(m,n,n-m)$, $\Omega=p$ is equivalent to $(m,n,m+n)$, $\Omega=n$.
308:
309: Hamiltonian structure in the undamped problem has important consequences for the
310: results in Table~\ref{tab:3ff}. If one assumes that at $\gamma=0$ Eqs.~(\ref{eq:TW})
311: derive from a Hamiltonian ${\cal H}$ through
312: $d Z_j^\pm /d t = \mp i \partial {\cal H}/\partial \bar{Z}_j^\pm$ (see, \emph{e.g.}, \cite{Mile84})
313: then $q_1=q_5$, $q_2=q_7$, and $q_3=q_4=q_6$. When we allow for simple rescalings:
314: $(Z_1^\pm,Z_2^\pm) \rightarrow \eta (Z_1^\pm,Z_2^\pm)$, $(Z_3^\pm) \rightarrow \xi (Z_3^\pm)$
315: with $\eta,\xi \in {\mathbb R}$, these conditions relax to $q_1=r q_5$, $q_2=r q_7$, and
316: $q_3=q_4=r q_6$ (for some $r >0$), implying
317: \begin{align}
318: \alpha_1 > 0, \quad \alpha_2 > 0, \quad \alpha_3 =0, \quad \alpha_4=0.
319: \label{eq:Ham}
320: \end{align}
321: Relations~(\ref{eq:Ham}) mean that for simple couplings (the first two sections
322: of Table~\ref{tab:3ff}) the sign of $b_{\rm res}$ is determined, and thus one knows if the
323: resonant triad has an enhancing or suppressing effect on patterns involving $\theta_{\rm res}$.
324: The $\Omega=m$ and $\Omega = m \pm n$ modes are suppressing, the $\Omega=n$ mode is
325: inconsequential, and the $\Omega=n-m$ mode is enhancing.
326:
327: We now pursue the implications of Table~\ref{tab:3ff} by examining two cases relevant to
328: recent experiments. We compare our theoretical predictions with coefficients calculated
329: numerically from the Zhang-Vi\~nals Faraday wave equations \cite{ZhaVin97_1}, which
330: describe weakly-damped fluids in deep containers. The calculation (see \cite{SilTopSke00}
331: for details) gives us the cross-coupling coefficient $b=b_0 + b_{\rm res}$ at the SW bifurcation
332: and is independent of the symmetry arguments used here.
333:
334: As our first example we consider the superlattice-I pattern observed with two-frequency
335: $(m,n)=(6,7)$ forcing in the experiments of \cite{KudPieGol98}. The wave vectors making
336: up the pattern lie on the vertices of two hexagons, one rotated by an angle
337: $\theta_{\rm h} < 30^\circ$ with respect to the other (see Fig.~3 in \cite{SilTopSke00}). It
338: was shown in \cite{SilTopSke00} that the experimentally observed angle of
339: $\theta_{\rm h} \simeq 22^\circ$ is related to a resonant triad at $\theta \simeq 158^\circ$
340: ($= 180^\circ - 22^\circ$) involving the $\Omega=n-m$ mode. This is the most interesting
341: of the damped modes because it gives $b_{\rm res}>0$ and thus acts as a selection
342: mechanism. Table \ref{tab:3ff} provides a means by which to enhance
343: this effect, namely by parametrically forcing the damped mode via $(m,n,2n-2m)=(6,7,2)$ forcing.
344: In Fig.~\ref{fig:672}a we show $b(\theta)$ for the phase $\Phi = 90\degree$ which optimizes the
345: stabilizing effect at $\theta_h \simeq 22\degree$. In Fig.~\ref{fig:672}b we show that
346: parametrically forcing the difference frequency mode with this phase quadruples the
347: stabilizing impact with respect to the two-frequency case used in the experiments. Note
348: that if the wrong phase ($\Phi=-90\degree$) is chosen, the effect of the resonance will
349: actually be diminished compared to the two-frequency case. Fig.~\ref{fig:672}c shows the
350: sinusoidal dependence of $b_{\rm res}$ on $\Phi$, in excellent agreement with our predictions.
351: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
352: \begin{figure}[ht]
353: \includegraphics[width=3.4in]{fig2.eps}
354: \caption{Effect of $\Phi$ on the stabilizing $\Omega=n-m$ resonance that selects
355: superlattice patterns \cite{KudPieGol98} with $(m,n,p)=(6,7,2)$ forcing. (a) $b(\theta)$
356: with the optimal phase $\Phi=90\degree$; the singular region heralding hexagons
357: at $\theta=60\degree$ is removed. The large dip near $\theta=67\degree$ degrees is due
358: to the strongly suppressing $\Omega = m$ resonance in the first section of Table~\ref{tab:3ff}.
359: (b) Close-up of $b(\theta)$ near $\theta=22\degree$ with $\Phi=90^\circ$ and
360: $\Phi=-90^\circ$; the two-frequency result (dotted line) with $f_p=0$ is also shown.
361: (c) Spike magnitude versus $\Phi$ (see Eq.~\ref{eq:P}). For these calculations
362: $|f_n|/|f_m|=0.75$, $|f_p|/|f_m|=0.1$. The fluid parameters in the Zhang-Vi\~{n}als
363: equations of \cite{ZhaVin97_1} are $\gamma = 0.08$, $G_0 = 1.5$.}
364: \label{fig:672}
365: \end{figure}
366: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
367:
368: As a second example, we consider recent experimental results on quasipatterns. It was reported
369: in \cite{ArbFin02} that 8-fold quasipatterns, which were distorted and difficult to observe with
370: $(m,n)=(3,2)$ forcing, became dramatically cleaner and more robust with $(m,n,p)=(3,2,4)$ forcing.
371: An explanation for this observation is provided by Table \ref{tab:3ff}. Specifically, we find that the
372: $\Omega=1$ mode forms a resonant triad with the critical modes with associated angle
373: $\theta_{\rm res} \approx 41\degree$. This is the angle present in the distorted (3,2)-forced
374: quasipatterns in~\cite{ArbFin02}. Table \ref{tab:3ff} indicates that with $(m,n,p)=(3,2,4)$
375: forcing there is a positive $\alpha_2|f_4|^2$ contribution to $b_{\rm res}$ which has a stabilizing
376: effect. In our numerical investigations, we find that the stabilizing spike in $b(\theta)$ becomes
377: broader with increasing $\gamma$, and thus it is reasonable to expect that the stabilization could
378: extend to the $45\degree$ angle associated with the perfect 8-fold quasipattern. Unfortunately,
379: the experimental value of $\gamma$ appears to be too large for a more quantitative comparison
380: with our theory.
381:
382: The theoretical framework developed in this Letter should provide some welcome guidance for
383: navigating the huge parameter space of multi-frequency forced Faraday waves. Table~\ref{tab:3ff}
384: provides a comprehensive list of the important damped modes and their effect on pattern formation
385: (additional possibilities with more than three frequencies will be considered elsewhere
386: \cite{TopPorSil_pre}). In general, the influence of these damped modes, when parametrically
387: forced, depends on $T_\tau$-invariant combination(s) of forcing phases. Using the ``proper''
388: phase greatly enhances resonance effects while the ``wrong'' phase can actually reduce them. We
389: hope that experimentalists will use Table~\ref{tab:3ff} as a tool to
390: design forcing functions conducive to particular patterns. For example, one might stabilize a
391: quasipattern composed of 2N, ${\rm N}>3$, equally spaced critical modes by arranging for an
392: $\Omega = n-m$ mode with wavenumber $k_3 = 2k_1\cos(\frac{{\rm N}-1}{\rm N} 90^\circ)$.
393: In the inviscid limit an appropriate choice of $m$ and $n$ can be estimated from this geometric
394: condition and the dispersion relation \cite{TopSil02}. Adding a third $2n-2m$ frequency would
395: then allow one to manipulate the effect by varying $\Phi$. Finally, we emphasize that the scheme
396: used to obtain Table~\ref{tab:3ff} is based on symmetry considerations and is therefore quite
397: general, requiring only that the damping be small.
398:
399: \begin{acknowledgements}
400: CMT was supported by NSF grants DMS-9983726 and DMS-9983320. MS was supported
401: by NASA grant NAG3-2364, NSF grant DMS-9972059, and by the NSF MRSEC Program
402: under DMR-0213745.
403: \end{acknowledgements}
404: \bibliography{3ff}
405: \end{document}
406:
407: