physics0003077/2.tex
1: %       CNLS Conference, May 16-21, 1997 Los Alamos Natl.Lab., for Physica D.
2: %       
3: \documentstyle[12pt]{article}
4: \textwidth15.6cm
5: \oddsidemargin0cm
6: \evensidemargin0cm
7: \topmargin0.5cm
8: \headheight0cm
9: \headsep0cm
10: \textheight22cm
11: \parindent1.0cm
12: \baselineskip15pt
13: \begin{document}
14: \title{Patterns on  liquid surfaces: cnoidal waves, compactons and scaling}
15: \author{A. Ludu and J. P. Draayer \\
16: \normalsize{Department of Physics and Astronomy, Louisiana State
17: University,} \\
18: \normalsize{Baton Rouge, LA 70803-4001, U.S.A.}}
19: \date{\today}
20: \maketitle
21: \begin{abstract}
22: 
23: Localized patterns and nonlinear oscillation formation on the bounded free surface of
24: an ideal incompressible liquid are analytically investigated . Cnoidal modes, solitons
25: and compactons, as traveling  non-axially symmetric shapes are discused. A
26: finite-difference differential generalized Korteweg-de Vries equation is shown to
27: describe the three-dimensional motion of the fluid surface and the limit of long and
28: shallow channels one reobtains the well known KdV equation. A tentative expansion
29: formula for the  representation of the general solution of a nonlinear equation, for
30: given initial condition is introduced on a graphical-algebraic basis. The model is
31: useful in multilayer fluid dynamics, cluster formation, and nuclear physics since, up
32: to an overall scale, these systems display liquid free surface behavior.
33: 
34: 
35: 
36: \end{abstract}
37: 
38: PACS : 47.55.Dz, 68.10.Cr, 47.20.Ky, 47.20.Dr, 97.60.j, 83.70, 36.40.-c
39: 
40: \setlength{\baselineskip} {4ex.}
41: 
42: \vfill
43: \eject
44: 
45: \section{Introduction}
46: 
47: Liquid oscillations on bounded surfaces have been studied intensively, both theoretically [1-3] and
48: experimentally [4-6]. The small-amplitude oscillations of incompressible drops maintained by
49: surface tension are usually characterized by their fundamental linear modes of motion in
50: terms of spherical harmonics [1-3]. Nonlinear oscillations of a liquid drop introduce new
51: phenomena and more complicated patterns (higher resonances, solitons, compactons, breakup and
52: fragmentation, fractal structures, superdeformed shapes) than can be described
53: by a linear theory.  Nonlinearities in the description of an ideal drop demonstrating
54: irrotational flow  arise from Bernoulli's equation for the pressure field and from the
55: kinematic surface boundary conditions [7]. Computer simulations have been carried for non-linear
56: axial oscillations and they are in very good agreement with experiments [4-6].
57: 
58: The majority of experiments show a rich variety of complicated shapes, many related to the
59: spinning, breaking, fission and fusion of liquid drops. There are experiments [6] and numerical
60: simulations [2] where special rotational patterns of circulation emerge:  a running wave originates on
61: the surface of the drop and then propagates inward.  Recent results (superconductors [8], catalytic
62: patterns [9], quasi-molecular spectra [10], numerical tests on higher order non-linear equations [11] and
63: analytical calculations on the non-compact real axis [12-13])  show shape-stable traveling waves for
64: nonlinear systems with compact geometry.  Recent studies showed that a
65: similar one-dimensional analysis for the process of cluster emission from heavy nuclei and
66: quasi-molecular spectra of nuclear molecules yields good agreement with experiment [10]. Such
67: solutions are stable and express to a good extent the formation and stability of patterns, clusters,
68: droplets, etc. However, even localised, they have nor compact support neither periodicity (excepting
69: some intermediate steps of the cnoidal solutions, [10,13]), creating thus difficulties when analysing on
70: compact surfaces.
71: 
72: In the present paper we comment on the cnoidal-towards-solitons solution investigated in [9,12],
73: especially from the energy point of view. We introduce here a new nonlinear 3-dimensional dynamical
74: model of the surface, in compact geometry (pools, droplets, bubbles, shells), inspired by [12], and we
75: investigate the possibilities to obtain compacton-like solutions for this model. We also study the scale
76: symmetries of such solutions.
77: 
78: The model in [9,12] consider the nolinear hydrodynamic equations of the surface of a liquid
79: drop and show their direct connection to KdV or MKdV systems. Traveling solutions that
80: are cnoidal waves are obtained [10,13] and they generate multiscale patterns ranging from small
81: harmonic oscillations (linearized model), to nonlinear oscillations, up to solitary waves. These
82: non-axis-symmetric localized shapes are  described by a KdV Hamiltonian system, too,
83: which
84: results as the second order approximation of the general Hamiltonian, next corrextion from the linear
85: harmonic shape oscillations. Such rotons were observed experimentally when the shape
86: oscillations of a droplet became nonlinear [4,6,8,13].
87: 
88: 
89: \section{Liquid drop cnoidal and soliton solutions from Hamiltonian approach}
90: 
91: The dynamics governing one-dimensional surface oscillations of a perfect ($\rho =$const.), irrotational
92: fluid drop (or bubble, shell) can be described by the velocity field $\Phi$ and a corresponding
93: Hamiltonian [1-3,7,10,13].  By expanding the Hamiltonian and dynamical equations in terms of a small
94: parameter, i.e. the amplitude of the perturbation $\eta $ over the
95: radius of drop $R_0$, the usual linear theory is recovered in the first order. Higher order non-linear
96: terms introduce deviations and produce large surface oscillations like cnoidal waves [7].
97: These oscillations, under conditions of a rigid core of radius $R_0 -h$ and non-zero angular momentum,
98: transform into solitary waves. In the following, by using the calculation developed in [10],
99: we present the Hamiltonian approach for the liquid drops nonlinear oscillations. However, this approach
100: is different from the nuclear liquid drop model point of view in [10], since we do not use here
101: the nuclear interaction (shell corrections) responsible for the formation of different potential valleys.
102:  
103: 
104: The total hydrodynamic energy $E$ consists of the sum of the kinetic $T$ and potential $U$ energies of
105: the liquid drop. The shape function is assumed to factorize, $r(\theta ,\phi , t)=R_0 (1+g(\theta
106: )\eta (\phi ,t))$. All terms that depend on $\theta $ are absorbed in the coefficients of some
107: integrals and the energy reduces to a
108: functional of $\eta $ only.  The potential energy is given by the surface energy $U_S =\sigma
109: ({\cal A}_{\eta }-{\cal A}_{0})|_{V_{0}}$, where $\sigma $ is the surface pressure coefficient,
110: ${\cal A}_{\eta }$ is the area of the deformed drop, and ${\cal A}_0$ the area of the spherical
111: drop, of constant volume  $V_0$. 
112: The kinetic energy 
113: $T=\rho \oint_{\Sigma }\Phi \nabla \Phi \cdot d{\vec S}/2$, [1-3,10,13],
114: the kinematic free surface boundary condition $\Phi _r=\partial _t r +(\partial _{\theta }r)
115: \Phi _{\theta
116: }/r^2 +(\partial _{\phi }r)\Phi _{\phi }/r^2 \sin \theta $, and the  boundary condition
117: for the radial velocity  on the inner surface
118: $\partial _r  \Phi | _{r=R_0 -h}=0$, [7], result in the expression [2,3,10]
119: \begin{eqnarray}
120: T={{R_{0}^{2}\rho } \over 2}\int_{0}^{\pi}\int_{0}^{2\pi}
121: {{
122: R_0 \Phi \eta _t \sin \theta  +{1 \over {R_0}}
123: g\eta _{\phi } \Phi \Phi _{\phi }(1-\sin \theta )} \over 
124: {\sqrt{1+g_{\theta }^{2}\eta ^2 +g^2 \eta _{\phi }^{2}}}} d\theta d\phi .
125: \end{eqnarray}
126: If the total energy, written in the second order in $\eta $, is taken to be a
127: Hamiltonian $ H[\eta ]$, the time derivative of any quantity
128: $F[\eta ]$ is given by $F_t =[F,H]$. Defining $F=\int_{0}^{2\pi} 
129: \eta (\phi -Vt)d\phi $  it results ([10], last reference) 
130: \begin{eqnarray}
131: {{dF}\over {dt}}=\int_{0}^{2\pi }\eta _t d\phi =\int_{0}^{2\pi}
132: (2 C_2 \eta_ {\phi }+6 C_3 \eta \eta _{\phi }-2 C_4 
133: \eta _{\phi \phi \phi }) d\phi =0,
134: \end{eqnarray}
135: which leads to the KdV  equation.
136: Here  
137: ${\cal C}_{2}=\sigma R_{0}^{2}(S_{1,0}^{1,0}+S_{0,1}^{1,0}/2) + 
138: R_{0}^{6}\rho V^2
139: C_{2,-1}^{3,-1}/2$, ${\cal C}_{3}=\sigma R_{0}^{2}S_{1,2}^{1,0}/2+
140: R_{0}^{6}\rho V^2 (2S_{-1,2}^{3,-1}R_0+S_{-2,3}^{5,-2}+ R_0 
141: S_{-2,3}^{6,-2})/2$,
142: ${\cal C}_{4}=\sigma R_{0}^{2}S_{2,0}^{-1,0}/2$, with
143: $S_{i,j}^{k,l}=R_{0}^{-l}\int_{0}^{\pi}h^{l}g^{i}g_{\theta }^{j}{\sin 
144: }^{k}\theta d\theta $.
145: Terms proportional to $\eta \eta _{\phi }^{2}$ can be neglected
146: since they introduce a factor $\eta _{0}^{3}/L^2 $ which is small compared
147: to $\eta _{0}^{3}$, i.e. it is in the third order. In order to verify the correctness of the
148: above approximations, we present, for a typical soliton
149: solution  $\eta (\phi ,t)$, some terms occuring in the expresion of $E$, Fig. 1. 
150: All details of calculation are given in [10,13].
151: Therefore, the energy of the non-linear liquid drop model can be interpreted as the Hamiltonian of the
152: one-dimensional KdV equation.
153: The coefficients in eq.(2) depend on two stationary functions
154: of  $\theta $ (the depth $h(\theta )$ and the transversal profile $g(\theta )$), hence, under the
155: integration, they involve only  a parametric dependence.
156: 
157: The KdV equation has the following cnoidal wave (Jacobi elliptic function) as exact solution
158: \begin{eqnarray}
159: \eta =\alpha _3 +(\alpha _2 -\alpha _3 )sn^2 \biggl ( \sqrt{{{C_3 (\alpha _3
160: -\alpha _2 )}\over {12C_4 }}}(\phi -Vt)\bigg | m \biggr ) ,
161: \end{eqnarray}
162: where $\alpha _1 ,\alpha _2 ,\alpha _3 $ are constants of integration, $m^2 =(\alpha _3 -\alpha _2
163: )/(\alpha _3 -\alpha _1 )$. 
164: This solution  oscillates between $\alpha _2$ and $\alpha _3$,
165: with a period
166: $T=2K(m)\sqrt{{(\alpha _3 -\alpha _2 )C_3 }\over {3C_4 }}$, where $K(m)$ is the period of a Jacobi
167: elliptic function $sn(x | m)$. 
168: The parameter $V$ is the velocity of the cnoidal waves and  $\alpha _1 +\alpha _2 +\alpha _3
169: ={{3(V-C_2 ) }\over {2C}}$. In the limit
170: $\alpha _1 =\alpha _2 =0$  the solution eq.(3)
171: approaches
172: \begin{eqnarray}
173: \eta =\eta _0 sech ^2 \biggl [
174: \sqrt{{{\eta _0 C_3 }\over {12 C_4}}}(\phi -Vt)
175: \biggr ],
176: \end{eqnarray}
177: which is the soliton solution of amplitude $\eta _0 $. Small oscillation occur when
178: $\alpha _3 \rightarrow \alpha _2 $ and $m \rightarrow 0 , T\rightarrow \pi
179: /2$.
180: Consequently, the system has two limiting solutions, a periodic and 
181: a localized traveling profile, which deform one into the other,
182: by  the initial conditions and the velocity parameter $V$. A figure showing the
183: deformation from the
184: $l=5$ cnoidal mode towards a soliton is shown in Figs. 2.
185: 
186: The cnoidal solution eq.(3) depends on the  parameters $\alpha _i$
187: subjected to the
188: volume conservation and the  periodicity condition of the solution (for the final soliton state this
189: condition should be taken as a  quasi-periodicity realised by the rapidly decreasing profile. This a
190: problem of the basic model, [10]). The periodicity restriction reads
191: \begin{eqnarray}
192: K\biggl ( \sqrt{{\alpha _3 -\alpha _2 }\over {\alpha _3 -\alpha _1 }}\biggr ) =
193: {{\pi}\over {n}}\sqrt{\alpha _3 -\alpha _1 }, \ \ \ n=1,2,\dots,  2\sqrt{\alpha _3 -\alpha _1 } .
194: \end{eqnarray}
195: Hence, a single free parameter remains, which
196: can be taken either one out of the three $\alpha
197: $'s, $V$ or $\eta _0 $.
198: Equatorial  cross-sections of the drop are shown in Fig. 2b for the
199: cnoidal solution at several values of the parameter $\eta _0$. 
200: All explicite calculations are presented in detail in [10].
201: 
202: In Fig. 3 we present the total energy plotted versus the parameters $\alpha _1 , \alpha _2$ for constant
203: volume. From the small oscillation limit ($\alpha _2 \simeq 3$ in the figure) towards the solitary wave
204: limit ($\alpha _2 =1$ in the figure) the energy increases and has a valley for $\alpha _1 \simeq 0.1$
205: and  $\alpha_2 \in (1.2, 1.75)$ (close to the $l=2$ mode). 
206: In order to introduce more realistic results, the total hydrodynamic energy is plotted versus 
207: $\alpha _1 ,\alpha _2$ for constant volume, too  but we marked those special solutions 
208: fulfilling the periodicity condition. In Fig. 4 we present the total energy valley, from the small
209: oscillations limit  towards the solitary wave limit. We notice that 
210: the energy constantly increases but around  $\alpha _2 \in (1.2, 1.75)$ (close to the
211: linear $l=2$ mode) it has a valley providing some stability for solitary
212: solution (also called roton [13]).
213: 
214: 
215: \section{The three-dimensional nonlinear  model}
216: 
217: In the following we introduce a sort of generalized KdV equation for fluids.
218: We consider the three-dimensional irrotational flow of an ideal incompressible fluid layer
219: in a semi-finte rectangular channel subjected to uniform vertical gravitation ($g$ in $z$ direction)
220: and to surface pressure [12]. The
221: depth of the layer, when the fluid is at rest is $z=h$. Boundary conditions at the finite spaced walls
222: consist in annilation of the  normal velocity component, i.e. on the bottom of the layer ($z=0$) and on 
223: the walls $x=x_{0}\pm L/2$ of the channel of width $L$. The following results remain valid if the
224: walls expand arbitrary, e.g. $L\rightarrow \infty$, and the flow is free. We choose for the
225: potential of the velocities the form
226: \begin{eqnarray}
227: \Phi =\sum_{k\ge 0}\alpha _k (t)
228: \cos {{k\pi (x-x_0 )} \over L}\cosh {{\sqrt{2}k\pi (y-y_0 )} \over L}\cos {{k\pi z} \over L},
229: \end{eqnarray}
230: where $\alpha _k (t)$ are arbitrary functions of time and $L$ is a free parameter.
231: Eq.(7) fulfils $\triangle \Phi =0$ and the above boundary conditions at the walls. However there is
232: another boundary condition at the free surface of the fluid [7]
233: \begin{eqnarray}
234: ( \Phi _z -\eta _t -\eta _x \Phi _x )_{z=h+\eta }=0,
235: \end{eqnarray}
236: where $\eta (x,t)$ describes the shape of the free surface. By introducing the
237: function
238: \begin{eqnarray}
239: f(x,t)=\sum_{k=0}^{\infty }{{\alpha _k (t)k\pi } \over L}
240: \biggl ( \sin {{k\pi (x-x_0 )} \over L}\cosh {{\sqrt{2}k\pi (y-y_0 )} \over L}\biggr ) , 
241: \end{eqnarray}
242: the velocity field on the free surface can be written
243: \begin{eqnarray}
244: \Phi _x |_{z=h+\eta } =- \cosh(z\partial _{x})f , \nonumber \\ 
245: -\Phi _z |_{z=h+\eta }=-\sinh(z\partial _{x})f .
246: \end{eqnarray}
247: Eqs (10) do not depend on $L$ and the case $L \rightarrow \infty $ of unbounded channels and free
248: travelling profiles remains equaly valid. Since the unique force field in the problem is
249: potential, the dynamics is described by the Bernoulli equation, which, at the free surface,
250: reads
251: \begin{eqnarray}
252: \Phi _{xt}+\Phi _{x}\Phi _{xx}+\Phi _{z}\Phi _{xz}+g{\eta }_{x}+{1 \over {\rho }}P_{x}=0.
253: \end{eqnarray}
254: Here $P$ is the surface pressure obtained by equating
255: $P$'s first variation with the local mean curvature of the surface, under the restriction of the
256: volume conservation
257: \begin{eqnarray}
258: P\bigg |_{z=h+\eta }={{\sigma {\eta }_{xx}} \over {{(1+{\eta }_{x}^{2})}^{3/2}}},
259: \end{eqnarray}
260: and $\sigma $ is the surface pressure coefficient. The pressure in  eq.(12) approaches $-\sigma {\eta
261: }_{xx}$, for small enough relative amplitude of the deformation
262: $\eta /h$.  In order to solve the system of the two partial differential equations (8,11) with respect
263: to the unknown functions $f(x,t)$  and $\eta (x,t)$, we consider the approximation of small
264: perturbations of the surface compared to  the depth, $a=max|{\eta }^{(k)}(x,t)|<<h$,  where $k=0,...,3$
265: are orders of differentiation. Inspired by [12] and using a sort of perturbation  technique in $a/h$, we
266: obtain from eqs.(6-11) the  generalised KdV equation
267: \begin{eqnarray}
268: {{\eta }_{t}+{{c_{0}}\over h}\sin(h\partial ){\eta } +  {{c_{0}}\over h}({\eta }_{x}\cosh(h\partial )
269: {\eta }} +\eta \cosh (h\partial ){\eta }_{x})0. 
270: \end{eqnarray}
271: If we approximate $\sin (h\partial )\simeq h\partial -{1 \over  6}(h\partial )^{3}$, $\cosh (h\partial
272: )\simeq 1-{1 \over 2} (h\partial )^{2}$, we obtain,  from eq.(9), the polynomial differential equation:
273: \begin{eqnarray}
274: a{\tilde {\eta }}_{t}+2c_{0}{\epsilon }^{2}h{\tilde {\eta }}
275: {\tilde {\eta }}_{x}+c_{0}\epsilon h{\tilde {\eta }}_{x}
276: -c_{0} \epsilon {{h^{3}} \over 6}{\tilde {\eta }}_{xxx}-
277: {{c_{0}{\epsilon }^{2}h^{3}} \over 2}\biggl (
278: {\tilde {\eta }}_{x}{\tilde {\eta }}_{xx}+{\tilde {\eta }}{\tilde 
279: {\eta }}_{xxx}\biggr ) & = & 0, \hfill
280: \end{eqnarray}
281: where $\epsilon ={a \over h}$. The first four terms in eq.(20)  correspond  to the zero order
282: approximation terms in $f$, obtained  from the boundary  condition at the free surface,  i.e. the
283: traditional way of obtaining the KdV equation in shallow channels.
284: 
285: In order to find an exact solution for  eq.(12) we can  write it in the form:
286: $$
287: Ah{u }_{X}(X)+{{u (X+h)-u (X-h)} \over {2i}}
288: +{u }_{X}(X){{u (X+h)+u (X-h)} \over 2} 
289: $$
290: \begin{eqnarray}
291: +u (X)
292: {{{u }_{X}(X+h)+{u }_{X}(X-h)
293: } \over {2}}=0, 
294: \end{eqnarray}
295: where $X=x+Ac_0 t$ and $A$ is an arbitrary real constant.
296: We want to stress here that eq.(14) is a finite-difference differential equation, which is rather the
297: exception than the rule fir such systems. Hence, it may contain among its symmetries, the scaling
298: symmetry. Actualy, the first derivative of $u(X)$ is shown to be alinear combination of translated
299: versions of the original function. In this way, the theory of such equations can be related with the
300: wavelet, or other self-similarity systems, theory, [13]. In the following we study the solutions with a
301: rapid decreasing  at infinity and make a change of variable:
302: $v=e^{BX}$ for
303: $x \in (-\infty ,0)$ and $ v=e^{-BX}$ for $x \in (0, \infty )$, with  $B$ an arbitrary constant.
304: Writing $u(X)=-hA+f(v)$, and choosing for the solution the form
305: of a power series in $v$:
306: \begin{eqnarray}
307: f(v)=\sum_{n=0}^{\infty }a_{n}v^{n},
308: \end{eqnarray}
309: we obtain a nonlinear recurrsion relation for the
310: coefficients $a_{n}$:
311: $$
312: \biggl ( Ahk+{{\sin(Bhk)} \over {B}}\biggr ) a_{k} 
313: $$
314: \begin{eqnarray}
315: =-\sum_{n=1}^{k-1}n\biggl (\cosh\left( Bh(k-n)\right )+\cosh (Bh(k-1))\biggr )a_{n}a_{k-n}. 
316: \end{eqnarray}
317: With the coefficients given in eq.(16) the general solution $\eta $ can be written analyticaly.
318: In order to verify the consistency of this solution
319: we study a limiting case of the relation, by replacing 
320: $\sin$  and $\cosh$ expressions with their lowest nonvanishing terms in their power 
321: expansions
322: Thus, eq.(16) reduces to
323: \begin{eqnarray}
324: {\alpha }_{k}={6 \over {B^{2}h^{3}k(k^{2}-1)}}
325: \sum_{n=1}^{k-1}n{\alpha }_{n}{\alpha }_{k-n},
326: \end{eqnarray}
327: and
328: \begin{eqnarray}
329: {\alpha }_{k}=
330: \biggl ( {1 \over {2B^{2}h^{3}}}\biggr )^{k-1}k  \hfill
331: \end{eqnarray}
332: is the solution of the above recurrence relation.
333: In this approximation, the solution of eq.(12) reads
334: \begin{eqnarray}
335: \eta (X) & = & 2B^{2}h^{3}\sum_{k=1}^{\infty }k
336: \biggl ( -e^{-B|X|}
337: \biggr )^{k}={{B^{2}h^{3}} \over 2}{1 \over {\left(
338: \cosh(BX/2)\right)^{2}}},
339: \end{eqnarray}
340: which is just the single-soliton solution of the
341: KdV equation and it was indeed obtained by assuming $h$ small 
342: in the recurrence relation (16).
343: Hence, we have shown that the KdV equation describing the shallow liquids can be generalised for any
344: depths and lengths. This result may be the starting point to search for more interesting symmetries. 
345: It would be interesting to interpret the generalized-KdV eq.(12) as the Casimir element of a certain
346: algebra. 
347: 
348: 
349: \section{Compacton and self-similar solutions}
350: 
351: Eq.(12) has a special character, namely contains both infinitesimal and finite difference operators.
352: This particularity relates it to another field of nonlinear systems, that is scaling functions and
353: wavelet basis, functions or distributions with compact support and self-similarity properties. In the
354: following we investigate a particular case of eq.(12), that is when
355: $h\ll
356: \eta$,
357: $h\ll \delta$, where $\delta $ is the half-width of the solution, if this has bounded or compact
358: support. In this approximations, from eq.(12) we keep only the terms
359: \begin{eqnarray}
360: {1 \over {c_0 }}\eta _t +\eta _x +{1 \over h} \eta \eta _x -{{h}\over {2}}\eta _x \eta _{xx}
361: +{1 \over h} \eta \eta _x -{h \over 2} \eta \eta _{xxx} +{\cal O}_{3}\simeq 0.
362: \end{eqnarray}
363: This equation is related to another intergable system, namely the K(2,2) equation, investigated in
364: [11]
365: \begin{eqnarray}
366: \eta _t +(\eta ^2 )_x +(\eta ^2)_{xxx}=0.
367: \end{eqnarray}
368: The main property of the K(2,2) equation is the equal occurence of non-linearity,
369: dispersion and the existence of a Lagrangian and Hamiltonian system associated with it. The
370: special solutions of this equation are the compactons
371: \begin{eqnarray}
372: \eta _{c}={{4 \eta _{0}}\over {3}}\cos ^2 \biggl ( {{x-\eta _{0}t}\over {4}} \biggr )  , \ \ \
373: |x-\eta _{0} t|\geq 2\pi ,
374: \end{eqnarray}
375: and $\eta _c =0$ otherwise.
376: This special solutions have compact support and special
377: properties concerning the scattering between different such solutions.
378: As the authors comment in [11], the robustness of these solutions makes it clear that a new mechanism is
379: underlying this system. In this respect, we would like to add that, taking into account eq.(12),
380: this new mechanism might be related to selfsimilarity and multiscale properties of nonlinear systems.
381: 
382: 
383: 
384: 
385: \vskip 1cm
386: \section{Conclusions}
387: \vskip 1cm
388: 
389: In the present paper we introduced a non-linear hydrodynamic model describing new modes of  motion of
390: the free surface of a liquid. The total energy of this nonlinear liquid drop model, subject to
391: non-linear boundary conditions at the free surface and the inner surface of the fluid layer,
392: gives the Hamiltonian of the Korteweg de Vries equation. We have studied the stability of the cnoidal
393: wave and solitary wave solutions, from the point of view of minima of this Hamiltonian.
394: 
395: The non-linear terms yield rotating steady-state solutions that are cnoidal waves on the surface
396: of the drop, covering continuously the range from small harmonic oscillations, to anharmonic 
397: oscillations, and up to solitary waves.
398: The initial one-dimensional model [10] was extend to a three-dimensional model. A kind of new
399: generalized KdV equation, together with some of its analytical solutions have been presented.
400: We also found a connection between the obtained generalized KdV equation, and another one (i.e. K(2,2)),
401: in a certain approximation. In this case, compacton solutions have been found and new symmetries (e.g.
402: self-similarity) were put into evidence.
403: 
404: The analytic solutions of the  non-linear
405: model presented in this paper,  make  possible the study of clusterization
406: as well as to explain or predict the existence of new strongly deformed shapes, or new patterns having
407: compact support or finite wavelength. The model applies not only in fluid and rheology theories, but may
408: provide insight into similar processes occurring in other fields and at other scales, such as the
409: behavior of superdeformed nuclei, supernova, preformation of cluster in hydrodynamic models (metallic,
410: molecular, nuclear), the fission of liquid drops (nuclear physics), inertial fusion, etc. 
411: 
412:  
413: 
414: 
415: \vskip 1cm
416: Supported  by the U.S. National Science Foundation through a regular grant, No. 9603006, and a 
417: Cooperative Agreement, No. EPS-9550481, that includes a matching component from the Louisiana
418: Board of Regents Support Fund.  One of the authors (A.L.) would like to thank Peter Herczeg from the
419: T5 Division at Los Alamos National Laboratory, and at the Center for Nonlinear Studies at Los
420: Alamos for hospitality. 
421: 
422: 
423: 
424: \begin{thebibliography}{99}
425: 
426: %theory
427: \bibitem{1}
428: H-L. Lu and R. E. Apfel, {\it {J. Fluid Mech.}} {\bf {222}} 351 (1991);
429: T. Shi and R. E. Apfel, {\it {Phys. Fluids}} {\bf {7}} 1545 (1995);
430: Y. Tian, R. G. Holt and R. E. Apfel, {\it {Phys. Fluids}} {\bf {7}} 2938 (1995);
431: W. T. Shi, R. E. Apfel and R. G. Holt, {\it {Phys. Fluids}} {\bf {7}} 2601 (1995).
432: 
433: %Lagrangean,roton modes,MIT
434: \bibitem{2}
435: R. Natarajan, R. A. Brown, {\it {J. Fluid Mech.}} {\bf {183}} 95 (1987); {\it {Phys. Fluids}}
436: {\bf {29}} 2788 (1986).
437: 
438: % cel mai complet sist de eq+Mit
439: \bibitem{3}
440: J. A. Tsamopoulos and R. A. Brown, {\it {J. Fluid Mech.}} {\bf {127}} 519 (1983).
441: 
442: %experiment
443: \bibitem{4}
444: E. H. Trinh, R. G. Holt and D. B. Thiessen, {\it {Phys. Fluids}} {\bf {8}} 43 (1995);
445: P. L. Marston and S. G. Goosby, {\it {Phys. Fluids}} {\bf {28}} 1233 (1985);
446: E. Trinh and T. G. Wang, {\it {J. Fluid Mech.}} {\bf {122}} 315 (1982).
447: 
448: %exp-columbia
449: \bibitem{5}
450: R. E. Apfel {\it {et al}}, {\it {Phys. Rev. Lett.}} {\bf {78}} 1912 (1997)
451: 
452: %experiment ROTONI!!
453: \bibitem{6}
454: E. Trinh and T. G. Wang, {\it {J. Fluid Mech.}} {\bf {122}} (1982) 315
455: 
456: \bibitem{7} 
457: G. L. Lamb, {\it Elements of Soliton Theory} (John Wiley \& Sons, New York, 1980);  C. Rebbi
458: and G. Soliani, {\it Solitons and Particles} (World Scientific, Singapore, 1984);  S.
459: Novikov, S. V. Manakov, {\it Theory of Solitons: The Inverse Scattering Method} (Consultants
460: Bureau, New York, 1984);  R. K. Bullough and P. J. Caudrey, Eds., {\it {Solitons}} (Topics in
461: Current Physics, Springer-Verlag, Berlin, 1980).
462: 
463: \bibitem{8}
464: A. Ustinov, {\it {Solitons in Josephson Junctions}}, 
465: {\it {non-linear Waves and Solitons in Physical Systems}}
466: (May 12-16, 1997, CNLS, Los Alamos, NM) to appear in {\it {Physica D}}.
467: 
468: \bibitem{9}
469: Y. G. Kevrekidis, {\it{Catalytic Pattern on Microdesigned Domains}},
470: {\it {non-linear Waves and Solitons in Physical Systems}}
471: (May 12-16, 1997, CNLS, Los Alamos, NM) to appear in {\it {Physica D}}.
472: 
473: \bibitem{10}
474: A. Ludu, A. Sandulescu and W. Greiner, {\it {Int. J. Modern Phys. E}} {\bf 1} 169 (1992) ; 
475: {\bf 2} 4 (1993) 855;  {\it J. Phys. G: Nucl. Part. Phys.} {\bf 21} 1715 (1995); {\it J. Phys. G: Nucl.
476: Part. Phys.} {\bf 23} 343 (1997).
477: 
478: \bibitem{11}
479: P. Rosenau and J. M. Hyman, {\it {Phys. Rev. Let.}} {\bf {70}} 564 (1993);
480: F. Cooper, J. M. Hyman and A. Khane, to be published.
481: 
482: \bibitem{12}
483: A. Ludu and W. Greiner, {\it {Found. Phys.}} {\bf {26}} 665 (1996).
484: 
485: \bibitem{13}
486: A. Ludu and J. P. Draayer, in preparation; Internal seminar CNLS and T8, Los Alamos, May, 1996.
487: 
488: \bibitem{14}
489: R. Abraham and J. E. Marsden, {\it Foundations of Mechanics}
490: (The Benjamin/Cummings Publishing Company, Inc., Reading, Massachusetts,
491: 1978).
492: 
493: \end{thebibliography}
494: 
495: \vfill
496: \eject
497: 
498: \centerline{FIGURE CAPTIONS}
499: \vskip 1cm
500: Fig. 1
501: 
502: The order of smallness of four typical terms depending on $\phi $ and occuring
503: in the Hamiltonian, eqs.(1,2).  Order zero holds for $\eta ^2$, order 1 for $\eta
504: _{\phi }^{2}$, order 2 for $\eta ^3, $ and order 3 for $\eta \eta _{\phi }^{2}$.
505: 
506: \vskip 1cm
507: Figs. 2
508: 
509: 2a.
510:  
511: The transition of the cnoidal solution, from a $l=5$ mode to the soliton limit:  shape of the
512: cross-section for $\theta =\pi /2$ function as a function of $\alpha _2 $ with $\alpha _{1,3}$ fixed by
513: the volume conservation and periodicity conditions.
514: 
515: 2b.
516: 
517: Cnoidal solutions (cross-sections of $\Sigma 1$ for $\theta =\pi /2$) subject to the
518: volume conservation constraint. Results for the $l=6$ mode to the $l=2$ mode and a soliton are
519: shown. The corresponding linear modes, i.e. spherical harmonics, are superimposed on the
520: non-linear solutions.
521: 
522: 2c.
523: 
524: Pictorial view of a soliton deformation of a drop, on the top of the original undeformed sphere.
525: The supporting sphere for the soliton has smaller radius  because of the volume conservation.
526: 
527: \vskip 1cm
528: 
529: 
530: \vskip 1cm
531: Fig. 3
532: 
533: The energy plotted versus $\alpha _1 , \alpha _2$ for constant volume. From the
534: small oscillation limit ($\alpha _2 \simeq 3$) towards the solitary wave limit ($\alpha _2
535: =1$) the energy increases and has a valley for $\alpha _1 \simeq 0.1$ and 
536: $\alpha_2 \in (1.2, 1.75)$ (close to the $l=2$ mode). 
537: 
538: \vskip 1cm
539: Fig. 4
540: 
541: The total energy plotted versus $\alpha _1 ,\alpha _2$ for
542: constant volume (small circles). Larger circles indicate the patterns
543: fulfilling the periodicity condition. From the small oscillations limit
544: ($\alpha _2 \simeq 3$) towards the solitary wave limit ($\alpha _2 =1$)
545: the energy increases but for $\alpha _2 \in (1.2, 1.75)$ (close to
546: $l =2$ mode) it has a valley..
547: 
548: 
549: 
550: \end{document}
551: 
552: