1: %----------------- Settings ----------------------------
2:
3: %\documentclass[preprint, aps, pre, eqsecnum, amsmath,amssymb,showpacs]{revtex4}
4: \documentclass[preprint, aps, pre, preprintnumbers,amsmath,amssymb,showpacs]{revtex4}
5: %\documentclass[draft,twocolumn, aps, pre, eqsecnum,amsmath, amssymb,showpacs]{revtex4}
6: %\documentclass[draft,twocolumn, aps, pre, preprintnumbers,amsmath, amssymb,showpacs]{revtex4}
7:
8:
9:
10: \usepackage[final]{graphicx} % to include figures
11: %\usepackage[draft]{graphicx} % includes only space for figures
12: %\usepackage{psfig}
13:
14: \renewcommand{\arraystretch}{1.5} %faktor 1.5 fuer tabellenzeilenabstand
15: \newcommand{\Nabla}{\mbox{\bf\boldmath $\nabla$}}
16: \newcommand{\ve}[1]{\stackrel{\rightharpoonup}{#1}}
17: \renewcommand{\vec}[1]{{\bf #1}}
18: \newcommand{\Vxi}{\mbox{\bf\boldmath $\xi$}}
19: \newcommand{\Vphi}{\mbox{\boldmath $\varphi$}}
20: \newcommand{\Vzeta}{\mbox{\boldmath $\zeta$}}
21: \newcommand{\VPsi}{\mbox{\boldmath $\Psi$}}
22: \newcommand{\Overrightarrow}[1]{\stackrel{\textstyle\rightarrow}{#1}}
23: \newcounter{saveeqn}
24: \newcommand{\alpheqn}{\setcounter{saveeqn}{\value{equation}}
25: \stepcounter{saveeqn}
26: \setcounter{equation}{0}
27: \renewcommand{\theequation}{\mbox{\arabic{section}.\arabic{saveeqn}\alph{equation}}}
28: }
29:
30:
31: \newcommand{\reseteqn}{\setcounter{equation}{\value{saveeqn}}
32: \renewcommand{\theequation}{\mbox{\arabic{section}.\arabic{equation}}}\\[-3mm]
33: }
34:
35: % info und seite in kopfzeile
36: \pagestyle{myheadings}
37: % datum in kopfzeile. falls 'twoside': {leftside}{rightside}
38: \markboth{\today}{\today}
39:
40: \bibliographystyle{prsty}
41:
42: \begin{document}
43:
44: \title{Competition between Traveling Fluid Waves of Left and Right Spiral Vortices and
45: Their Different Amplitude Combinations}
46:
47: \author{A.~Pinter, M.~L\"ucke, and Ch.~Hoffmann}
48: \affiliation{Institut f\"ur Theoretische Physik, Universit\"at des Saarlandes,
49: Postfach 151150, \\ D-66041 Saarbr\"ucken, Germany}
50:
51: \date{\today}
52:
53: %-------------- Abstract --------------------------------
54: \begin{abstract}
55: Stability, bifurcation properties, and the spatiotemporal behavior
56: of different nonlinear combination structures of spiral vortices in
57: the counter rotating Taylor-Couette system are investigated by full
58: numerical simulations and by coupled amplitude equation
59: approximations. Stable cross-spiral structures with continuously
60: varying content of left and right spiral modes are found. They
61: provide a stability transferring connection between the initially
62: stable, axially counter propagating wave states of pure spirals and
63: the axially standing waves of so-called ribbons that become stable
64: slightly further away from onset of vortex flow.
65: \end{abstract}
66:
67: %--------------- PACS -----------------------------------
68: \pacs{47.20.-k, 47.54.+r, 47.32.-y, 47.10.+g}
69:
70: %****************************************************************
71: % 47.20.-k : Hydrodynamic stability
72: % 47.54.+r Pattern selection; pattern formation
73: % 47.32.-y Rotational flow and vorticity
74: % 47.10.+g : General theory (of fluid dynamics)
75: % 47.60.+i Flows in ducts, channels, nozzles, and conduits
76: %****************************************************************
77: \maketitle
78: %---!---10----!---20----!---30----!---40----!---50----!---60----!---70----!---80
79:
80: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
81: %\section{introduction}
82: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
83: The combination of two counter propagating plane linear traveling
84: waves (TWs) of equal amplitude creates a standing wave (SW). This
85: property is also the reason for the possibility of SW structures in
86: a variety of nonlinear pattern forming systems that show an
87: oscillatory instability and inversion symmetry in one or more
88: spatial directions \cite{CH93}. Then two symmetry degenerate counter
89: propagating TW structures as well as a nonlinear SW solution appear
90: at the common oscillatory bifurcation threshold. However, in many systems like,
91: e.g., binary mixture convection TW solutions bifurcate subcritically being
92: initially amplitude unstable and then become strongly nonlinear on the stable
93: upper solution branch. Moreover, the SW solution is often amplitude unstable
94: and/or unstable against phase propagation and TW formation.
95:
96: Thus, we consider here nonlinear combination states of left-spiral
97: vortex (L-SPI) and right-spiral vortex (R-SPI) structures that ---
98: being mirror images of each other --- travel either upwards or
99: downwards in the annulus between two counter rotating cylinders of a
100: Taylor-Couette setup \cite{CI94, T94}. This system offers an easy
101: experimental and numerical access ({\em i}) to forwards bifurcating
102: initially stable TWs (i.e., L-SPI and R-SPI), ({\em ii}) to forwards
103: bifurcating SWs [called ribbons (RIBs) in the Taylor-Couette
104: literature \cite{CI94, T94}] that become stable slightly above
105: threshold, and in particular also ({\em iii}) to stable
106: cross-spirals (CR-SPI), i.e., combinations of L-SPI and R-SPI with
107: different amplitudes which provide a stability transferring
108: connection between TW and SW solution branches. This bifurcation
109: scenario does not seem to have been reported so far \cite{RIBs}.
110:
111: Here, we describe stability, bifurcation properties, and the spatiotemporal
112: behavior of the three oscillatory structures ({\em i})-({\em iii}) by two
113: amplitude equations with a coupling that contains necessarily also a
114: quintic order term to enable CR-SPI solutions. Their results are compared with
115: numerical solutions of the full Navier-Stokes equations that we obtained for a
116: system with radius ratio $\eta=1/2$ by methods described in \cite{HLP04}.
117:
118: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
119: %\section{structures}
120: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
121: %---!---10----!---20----!---30----!---40----!---50----!---60----!---70----!---80
122: The vortex structures ({\em i})-({\em iii}) are axially and
123: azimuthally periodic with wave numbers $k=2\pi /\lambda$ and $M$, respectively,
124: with $\lambda=1.3$ and $M=2$ throughout this paper.
125: They rotate with characteristic constant angular velocities into the same
126: direction as the inner cylinder \cite{HLP04} and thereby they are forced to
127: propagate axially except for RIB vortices which only rotate but do not propagate.
128:
129: In order to characterize the spatiotemporal properties of the
130: vortices we used also Fourier decompositions in azimuthal and axial
131: direction
132: \begin{eqnarray} \label{modenansatz}
133: f(r,\varphi,z,t) = \sum_{m,n} f_{m,n}(r,t)\,e^{i(m\varphi + nkz)}\,.
134: \end{eqnarray}
135: For example, L-SPI (R-SPI) flow contains only modes with $m=nM$
136: ($m=-nM$) since it does not depend on $\varphi,z,t$ separately but
137: only on the phase combination $\phi_A= M\varphi + kz-\omega_A t$
138: ($\phi_B= M\varphi - kz-\omega_B t$). Here
139: we use $A$ and $B$ to identify L-SPI and R-SPI properties,
140: respectively. For SPI that are axial mirror images of each other the
141: modes in (\ref{modenansatz}) oscillate according to $f_{m,n}(r,t)
142: =\delta_{m,\pm nM} e^{-i(m/M)\omega t} \tilde{f}_{\pm n}(r)$ with a
143: common frequency $\omega_A= \omega_B=\omega(k,M)$. Thus, the
144: color-coded plots in Fig.~\ref{v-w-Plots}(a,b) of the SPI in the
145: $\varphi-z$ plane of an 'unrolled' cylindrical surface are patterns
146: of straight stripes. They both move in positive $\varphi$-direction,
147: i.e., into the direction of rotation of the inner cylinder
148: \cite{HLP04} with angular phase velocity $\dot{\varphi}_{SPI} =
149: \omega/M$. The L-SPI propagates upwards with axial phase velocity
150: $w_{ph}^A=\omega/k$ and and its mirror image the R-SPI propagates
151: downwards, $w_{ph}^B=-\omega/k$ .
152:
153: On the other hand, a CR-SPI is a kind of combination of a L-SPI and
154: a R-SPI with different flow amplitudes and different frequencies.
155: Thus, the CR-SPI spectrum contains two frequencies, say, $\omega_S$
156: and $\omega_D$, that are in general incommensurate. Within the
157: amplitude equation approximation presented below $\omega_S=(\omega_A
158: + \omega_B)/2$ and $\omega_D=(\omega_A - \omega_B)/2$. For the
159: CR-SPI in Fig.~\ref{v-w-Plots}(c) the amplitude of the L-SPI content
160: is larger than the R-SPI content while $\omega_A < \omega_B$.
161: Then the periodic deformation of the L-SPI-dominated stripe pattern in
162: Fig.~\ref{v-w-Plots}(c) that is caused by the minority R-admixture
163: propagates axially downwards with velocity $\omega_D/k$ while
164: rotating with angular velocity $\omega_S/M$.
165: The CR-SPI in Fig.~\ref{v-w-Plots}(d) is the mirror
166: image of Fig.~\ref{v-w-Plots}(c) with reverse amplitude and
167: frequency relations. For all CR-SPI investigated here
168: $\omega_S\gg|\omega_D|$. RIB states are degenerate CR-SPI states
169: with equal content of R-SPI and L-SPI contributions and $\omega_D=
170: 0$ which do not propagate axially.
171:
172: Fig.~\ref{bifurcationdiagram} shows for two different $R_1$ bifurcation diagrams
173: of SPI, CR-SPI, and RIB solutions versus $R_2$ and its reduced distance
174: $\mu=(R_2 -R_2^0)/|R_2^0|$
175: from the onset, $R_2^0$, of SPI and RIB flow. Here $R_1$ and $R_2$ are the
176: Reynolds numbers defined by the rotational velocities of the inner and outer
177: cylinder, respectively. Order
178: parameters are the squared amplitudes ($|A|^2,|B|^2$) and frequencies
179: ($\omega_A,\omega_B$) of the dominant modes
180: \begin{equation} \label{EQ:u2pm1}
181: u_{2,1}(t)=|A|e^{-i\omega_A t} \quad , \quad u_{2,-1}(t)=|B|e^{-i\omega_B t}\,
182: \end{equation}
183: in the decomposition (\ref{modenansatz}) of the radial velocity $u$
184: at mid-gap. For all vortex structures investigated here the moduli
185: and frequencies of $u_{2,\pm1}$ in Eq.~(\ref{EQ:u2pm1}) are
186: constant.
187:
188: The L-SPI ($A\neq 0, B=0$) shown in Fig.~\ref{bifurcationdiagram} by circles and
189: the RIB solution ($A =B$) marked by diamonds bifurcate at $\mu = 0$ with common
190: linear frequency $\omega^0$ out of the
191: unstructured CCF. However, with increasing $\mu$ their frequencies
192: vary differently. The squared amplitudes of these two states grow
193: basically linearly with $\mu$, albeit with different slopes.
194:
195: Initially, the SPI is stable and the RIB is unstable. But then there appears
196: at larger $\mu$ a stable CR-SPI solution (triangles) which
197: transfers stability from the SPI to the RIB. The CR-SPI ($A \neq B$) bifurcates
198: with $B=0$ and finite $\omega_B$ out of the L-SPI as shown in
199: Fig.~\ref{bifurcationdiagram} (bifurcation out of
200: the symmetry degenerate R-SPI is analogous --- just exchange $A$ and $B$). Then
201: $|B|$ grows and $|A|$ decreases while $\omega_A$ increases and $\omega_B$ becomes
202: smaller until the CR-SPI solution branches end with
203: $A=B$ and $\omega_A = \omega_B$ in the RIB state. The latter loses stability
204: outside the plot range of Fig.~\ref{bifurcationdiagram} to amplitude-modulated
205: CR-SPI that are not discussed here.
206:
207: Spatiotemporal properties, bifurcation, and stability behavior of
208: the SPI, RIB, and CR-SPI states close to onset can reasonably well
209: be explained and described within an amplitude-equation approach.
210: Therein the fields are represented by the superposition
211: --- here written down, e.g., for the radial velocity $u(r,\varphi,z,t)$ ---
212: \begin{equation} \label{EQ:u-amp-repres}
213: u=\left[ A(t)\hat{u}_A(r)e^{ikz} + B(t)\hat{u}_B(r)e^{-ikz} \right]
214: e^{i(M\varphi-\omega^0t)} + c.c.
215: \end{equation}
216: of just the linear critical modes
217: $\hat{u}_{A,B}(r)e^{\pm ikz}e^{i(M\varphi-\omega^0t)}$ of L- and R-SPI with
218: slowly varying
219: amplitudes $A$ and $B$, respectively. Here we normalize $|\hat{u}_A|$ and
220: $|\hat{u}_B|$ to 1 at mid-gap. There the representation (\ref{EQ:u-amp-repres})
221: is quite sufficient. But closer to the
222: inner cylinder the vortex fields contain further axial and azimuthal modes that
223: are generated via nonlinear interactions of the critical modes. However,
224: it seems that close to onset they do not influence decisively the bifurcation
225: and stability behavior that is governed by the modes retained in the
226: approximation (\ref{EQ:u-amp-repres}).
227:
228: In order to reproduce the bifurcation and stability behavior of
229: the aforementioned vortex states including the CR-SPI one needs
230: coupled equations for $A$ and $B$ of at least quintic order \cite{FDT91}.
231: Higher-order terms that are suggested in \cite{CI94} are not necessary to
232: ensure existence of CR-SPI solutions. Symmetry
233: arguments \cite{CI94} and simplicity considerations discussed below restrict the
234: form of the equations to
235: \begin{subequations} \label{EQ:amp-eq}
236: \begin{eqnarray}
237: \dot{A}=\left[ a\mu +b|A|^2 +c|B|^2 + e\left(|A|^2 - |B|^2\right)|B|^2 \right] A \\
238: \dot{B}=\left[ a\mu +b|B|^2 +c|A|^2 + e\left(|B|^2 - |A|^2\right)|A|^2 \right] B \, .
239: \end{eqnarray}
240: \end{subequations}
241: The coefficients $a-e$ are complex but only the moduli $|A|$ and $|B|$
242: enter into the square brackets. We have determined all coefficients
243: by comparing with bifurcation diagrams obtained for the flow
244: at mid-gap position.
245:
246: We are seeking solutions of (\ref{EQ:amp-eq}) of the form
247: $A(t)=|A|e^{-i\Omega_A t} \, , \, B(t)= |B|e^{-i\Omega_B t}$
248: with constant moduli and frequencies. Here
249: $\Omega_{A,B}= \omega_{A,B} -\omega^0$ are the deviations of the
250: frequencies from the critical ones according to
251: (\ref{EQ:u2pm1}, \ref{EQ:u-amp-repres}).
252:
253: In view of
254: Fig.~\ref{bifurcationdiagram} we have neglected here for simplicity reasons
255: quintic
256: contributions to SPI amplitudes and frequencies by discarding terms of the form
257: $d|A|^4A$ and $d|B|^4B$ in (\ref{EQ:amp-eq}a) and (\ref{EQ:amp-eq}b),
258: respectively. Taking
259: $a= 1 + a_i$ one then has $|A|^2=\mu/(-b_r)$ and $\Omega_A=-a_i\mu -b_i |A|^2 $
260: for, say, a L-SPI (for a R-SPI replace $A$ by $B$). The indices $r$ and $i$
261: denote real and imaginary parts,
262: respectively. As an aside we mention that a linear variation of $\Omega_A$
263: with $|A|^2$ holds for SPI also significantly further away from onset than in
264: Fig.~\ref{bifurcationdiagram}.
265:
266: Motivated by Fig.~\ref{bifurcationdiagram} we also discard quintic contributions
267: to the modulus and to the frequency of the RIB state. This is enforced by
268: making in the terms $f|B|^4A$ and $f|A|^4B$ that appear on general grounds in
269: (\ref{EQ:amp-eq}a) and (\ref{EQ:amp-eq}b), respectively, the special choice
270: $f=-e$. Then the RIB solution is characterized by
271: $|A|^2=|B|^2=\mu/(-b_r-c_r)$ and $\Omega_A= \Omega_B= -a_i\mu -(b_i+c_i)|A|^2$.
272:
273: The CR-SPI solution with finite $A \neq B$ exists only for nonzero
274: coupling $e$. It is most conveniently found and expressed in terms
275: of the combined order parameters
276: \begin{subequations} \label{EQ:S-D-def}
277: \begin{eqnarray}
278: S = (|A|^2 + |B|^2)/2 \quad &,& \quad D = (|A|^2 - |B|^2)/2 \\
279: \Omega_S= (\Omega_A +\Omega_B)/2 \quad &,& \quad \Omega_D= (\Omega_A -\Omega_B)/2 \, .
280: \end{eqnarray}
281: \end{subequations}
282: Then the radial velocity field (\ref{EQ:u-amp-repres}) of the CR-SPI at mid-gap
283: can be written into the form
284: $ u=\left[ |A|e^{i(kz-\Omega_Dt)} + |B|e^{-i(kz-\Omega_Dt)} \right]
285: e^{i(M\varphi-\omega^0t-\Omega_St)} + c.c.$. It
286: shows the aforementioned rotation frequency $\omega_S=\omega^0+\Omega_S$
287: and the axial propagation of the deformation that is governed by the
288: frequency $\omega_D=\Omega_D$.
289:
290: The CR-SPI solution of (\ref{EQ:amp-eq}) reads
291: \begin{subequations} \label{EQ:CR-SPI-solution}
292: \begin{eqnarray}
293: S=(c_r-b_r)/(2e_r)\, &,& \, D^2 = [\mu + (b_r+c_r)S]/(2e_r)\\
294: -\Omega_S=a_i\mu + (b_i+c_i)S -2e_i D^2 \, &,& \,
295: -\Omega_D=(b_i-c_i +2e_i S)D \, .
296: \end{eqnarray}
297: \end{subequations}
298: Note that $S$ is constant in this CR-SPI. $D$ grows via a pitchfork bifurcation
299: at $\mu_2=-(b_r+c_r)S$ out of the RIB ($D=0$) and connects at
300: $\mu_1= -2 b_r S$ with $D=|A|^2/2=S$ with the L-SPI and with $D=-|B|^2/2=-S$
301: with the R-SPI.
302:
303: In Figs.~\ref{bifurcationdiagram} - \ref{FIG-R1=200} we compare
304: bifurcation diagrams of SPI, RIB, and CR-SPI solutions resulting from the
305: amplitude equations \cite{coefficients} (thin lines) with those from the full
306: numerical simulations (symbols). For the sake of presentational clarity the
307: stability of the former is not indicated in these figures. But
308: stability analyses of the three vortex states in question confirms that the
309: CR-SPI solution (\ref{EQ:CR-SPI-solution}) transfers stability between the RIB
310: and the SPI states as indicated in Fig.~\ref{3d-Plot}.
311:
312: We should like to mention that keeping the discarded quintic terms by setting
313: $d\neq 0$ and $f\neq -e$ slightly
314: improves the agreement between the solutions of the Navier-Stokes equations and
315: the amplitude equation approximation without changing the bifurcation topology
316: and the stability behavior of the latter. However, then the CR-SPI solution
317: would still have $S=$ const. To
318: obtain deviations from this behavior as displayed, e.g., by the triangles
319: of Fig.~\ref{FIG-R1=200}(a) one needs higher orders in eqs.~(\ref{EQ:amp-eq}).
320:
321: So, the model (\ref{EQ:amp-eq}) is a minimal one in
322: the sense that the quintic terms that are necessary to ensure the existence of
323: $A \neq B$ solutions enter only into the coupling between R- and L-SPI
324: modes but not into the SPI and RIB solutions.
325: Note furthermore that the coupled amplitude equations (\ref{EQ:amp-eq}) apply
326: equally well to other systems with two distinct $A$ and $B$ solutions --- be
327: they stationary or oscillatory --- that
328: are related to each other by analogous (symmetry) requirements with similar
329: nonlinear couplings. Thus, bifurcation diagrams in which the pure
330: $A$, the pure $B$, and the $A=B$ solutions are connected with a stability
331: transfer by a $A \neq B$ solution branch could also arise in these systems with
332: a form that would be (topologically) similar to that of Fig.~\ref{3d-Plot}.
333:
334: This work was supported by the Deutsche Forschungsgemeinschaft.
335:
336: \newpage
337: %----------------- Bibliography ---------------------
338: \begin{thebibliography}{999}
339: %----------------------------------------------------------------
340: \bibitem{CH93}
341: M.~C.~Cross and P.~C.~Hohenberg,
342: %{\it Pattern formation outside of equilibrium},
343: Rev.~Mod.~Phys. {\bf 65}, 851 (1993).
344: %----------------------------------------------------------------
345: \bibitem{T94}
346: R.~Tagg,
347: %{\it The Couette-Taylor Problem},
348: Nonlinear Science Today {\bf 4}, 1 (1994).
349: %----------------------------------------------------
350: \bibitem{CI94}
351: P.~Chossat and G.~Iooss, {\it The Couette-Taylor Problem},
352: (Springer, Berlin, 1994).
353: %----------------------------------------------------
354: \bibitem{RIBs}
355: Stable RIBs were found \cite{TESM89} in a long system of
356: aspect ratio $\Gamma=36$ with radius ratio $\eta=0.727$ where theory
357: \cite{DI84,CI94} and numerical simulations \cite{TESM89} predicted a subcritical
358: transition. However, no CR-SPI were obseved. RIBs in end-plate-dominated short
359: systems ($\Gamma < 10, \eta=1/2$) were recently reported \cite{LPA03-04} to come
360: with two different symmetries \cite{KP92}.
361: %-------------------------------------------------------------------------------
362: \bibitem{TESM89}
363: R.~Tagg, W.~S.~Edwards, H.~L.~Swinney, and P.~S.~Marcus,
364: %{\it Nonlinear standing waves in Couette-Taylor flow},
365: Phys.~Rev.~A {\bf 39}, 3734 (1989).
366: %-------------------------------------------------------------------------------
367: \bibitem{DI84}
368: Y.~Demay and G.~Iooss,
369: %{\it Calcul des solutions bifurquées pour le problème de Couette-Taylor avec les deux
370: %cylindres en rotation},
371: J. Mec. Theor. Appl., Spec. Suppl., 193 (1984).
372: %-------------------------------------------------------------------------------
373: \bibitem{LPA03-04}
374: J.~Langenberg, G.~Pfister, and J.~Abshagen,
375: %{\it Standing waves in flow between finite counterrotating cylinders},
376: Phys.~Rev.~E {\bf 68}, 056308 (2003);
377: %{\it The effect of physical boundaries on oscillatory bifurcation in
378: %counterrotating Taylor-Couette flow},
379: Phys. of Fluids, {\bf 16}, 2757 (2004).
380: %-------------------------------------------------------------------------------
381: \bibitem{KP92}
382: E.~Knobloch and R.~Pierce,
383: %{\it Spiral vortices in finite cylinders},
384: in {\it Ordered and turbulent patterns in Taylor-Couette Flow},
385: ed. C.~D.~Andereck and F.~Hayot, (Plenum Press, NY, 1992), p. 83.
386: %-------------------------------------------------------------------------------
387: \bibitem{HLP04}
388: Ch.~Hoffmann, M. L\"ucke, and A. Pinter,
389: %{\it Spiral vortices and Taylor vortices in the annulus between rotating cylinders and the effect
390: %of an axial flux},
391: Phys.~Rev.~E {\bf 69}, 056309 (2004).
392: %----------------------------------------------------
393: %\bibitem{notation}
394: %In \cite{HLP04} the axial wave number is
395: %positive and $M=\pm|M|$ while we use here $k=\pm|k|$ but positive $M$.
396: %-------------------------------------------------------------------------------
397: \bibitem{FDT91}
398: S.~Fauve, S.~Douady, and O.~Thual,
399: %{\it Drift instabilities of cellular patterns},
400: J.~Phys.~II {\bf 1}, 311 (1991) show that a non-potential linear coupling
401: which is forbidden here can cause a pitchfork drift bifurcation of
402: parametrically excited surface SWs into reflection symmetry breaking traveling
403: states with $A \neq B$.
404: %-------------------------------------------------------------------------------
405: \bibitem{coefficients}
406: We have determined the coefficients as follows: a linear
407: analysis of the flow dynamics gave $\omega^0= 150.9 \,\,(117.0)$ and
408: $a_i= 45.13\,\,(15.97)$ for $R_1=240$ ($R_1=200$).
409: Then $b_r$ and $b_i$ were obtained by fitting $|A|^2$ and
410: $\Omega_A$, respectively, of the nonlinear numerical SPI solution. Thereafter
411: $c_r$ and $c_i$ were fixed analogously with the numerical RIB
412: solutions. Finally, $e$ was determined with the CR-SPI.
413: The results are
414: \mbox{$100 b=-0.245+i\,83.54\,\,(-0.659+i\,132.8)$},
415: \mbox{$100 c=-1.06-i\,554\,\,(-2.517-i\,383.9)$},
416: \mbox{$100 e=-3.251+i\,118.4\,\,(-1.976+i\,89.2)$}.
417: %----------------------------------------------------------------
418: \bibitem{control}
419: Unstable SPI solutions were obtained by suppressing the $m=0-$mode of $u$. This
420: stabilizes the SPI against the transition to CR-SPI and
421: RIB states which have finite $m=0-$ content. The unstable RIB solution was
422: stabilized by imposing the $z \rightarrow -z$ symmetry of this state thereby
423: suppressing SPI and CR-SPI solutions.
424: %----------------------------------------------------------------
425:
426:
427: \end{thebibliography}
428: \clearpage
429: %----------------- Figures-------------------------------------------------
430: %----------------- FIG 1 --------------------------------------------------
431: \begin{figure}
432: \includegraphics[clip=true,width=7.5cm, angle=0]{./fig1.eps}
433: \caption{(Color online) Snapshots of the radial flow $u$ in the
434: $\varphi-z$-plane of the 'unrolled' mid-gap cylindrical surface.
435: White identifies the nodes. Light (dark) grey [red (blue) in the electronic
436: version] marks radial out-flow (in-flow). Arrows indicates how the phase moves. For the CR-SPI in
437: (c) [(d)] the undulated pattern moves upwards [downwards] (full
438: arrows) while the undulation itself propagates slowly downwards
439: [upwards] (dashed arrows). Parameters are $R_2=-605.2 (a, b),
440: -604.725 (c,d), -604.5 (e)$ and $R_1=240$. \label{v-w-Plots}}
441: \end{figure}
442: %----------------- FIG 2 --------------------------------------------------
443: \begin{figure}
444: \includegraphics[clip=true,width=8cm, angle=0]{./fig2.ps}
445: \caption{(Color online) Bifurcation diagrams of SPI (red circles),
446: RIB (blue diamonds), and CR-SPI (purple triangles) obtained from numerical
447: solutions of the full Navier-Stokes equations for $R_1=240$ (a,b) and
448: $R_1=200$ (c,d) as functions of $R_2$ and $\mu$. Shown are the
449: squared mode amplitudes $|A|^2,|B|^2$ and frequencies $\omega_A,\omega_B$ of the
450: dominant modes $u_{2,1}(t),u_{2,-1}(t)$ (\ref{EQ:u2pm1}), respectively, in the
451: expansion (\ref{modenansatz}) of the radial velocity field $u$ at mid-gap.
452: Full (open) symbols denote stable (unstable) solutions \cite{control}. Lines
453: show amplitude-equation approximations. Here a CR-SPI with
454: $|A|>|B|, \omega_A<\omega_B$ transfers stability from an L-SPI ($A\neq 0, B=0$)
455: to the RIB state ($A=B$).
456: The analogous diagram with an R-SPI is obtained by interchanging $A$ and $B$.
457: \label{bifurcationdiagram}}
458: \end{figure}
459: %\clearpage
460: %----------------- FIG 3 --------------------------------------------------
461: \begin{figure}
462: \includegraphics[clip=true,width=8cm, angle=0]{./fig3.ps}
463: \caption{(Color online) Bifurcation diagrams of SPI (red circles),
464: RIB (blue diamonds), and CR-SPI (purple triangles) obtained from numerical
465: solutions of the full Navier-Stokes equations for $R_1=240$ using the order
466: parameters (\ref{EQ:S-D-def}): Relation between $\mu$ and
467: $S$ (a) and $D$ (b), respectively; CR-SPI frequencies
468: $\Omega_S=(\Omega_A+\Omega_B)/2=(\omega_A+\omega_B)/2-\omega^0$ (c) and
469: $\Omega_D=(\Omega_A-\Omega_B)/2=(\omega_A-\omega_B)/2=\omega_D$ (d),
470: respectively, versus $D$. Full (open) symbols denote stable (unstable)
471: solutions \cite{control}. Lines show amplitude-equation approximations. Here a
472: CR-SPI ($|A|>|B|,D>0$ and $\Omega_A<\Omega_B,\Omega_D<0$)
473: transfers stability from an L-SPI ($B=0$, i.e., $D=S$) to the RIB state
474: ($A=B$, i.e., $D=0$).
475: \label{FIG-R1=240}}
476: \end{figure}
477: %\clearpage
478: %----------------- FIG 4 --------------------------------------------------
479: \begin{figure}
480: \includegraphics[clip=true,width=8cm, angle=0]{./fig4.ps}
481: \caption{(Color online) Bifurcation diagrams for $R_1=200$ as in
482: Fig.~\ref{FIG-R1=240}.
483: \label{FIG-R1=200}}
484: \end{figure}
485: %----------------- FIG 5 --------------------------------------------------
486: \begin{figure}
487: \includegraphics[clip=true,width=7cm, angle=0]{./fig5.eps}
488: \caption{(Color online)
489: Bifurcation diagrams of SPI, RIB, and CR-SPI resulting from the
490: amplitude-equation approximation. (a) 3D plot in $|A|^2-|B|^2-\mu$-space;
491: (b) projection onto the $|A|^2-|B|^2$-plane or the $D-S$-plane, respectively.
492: Full (dashed) lines denote stable (unstable) solutions. The CR-SPI
493: connects with constant $S$ the two SPI solution branches to the RIB thereby
494: transferring stability from the former to the latter.
495: \label{3d-Plot}}
496: \end{figure}
497: %---------------------------------------------------------------------------
498: \end{document}
499: