1: \documentclass[preprint,aps,floats,amssymb,showpacs]{revtex4}
2: \usepackage{epsfig}
3: % Uncomment next two lines for A4 paper size, comment for letter size
4: %\addtolength{\textheight}{17.6mm}
5:
6: %%%%% number equations by section %%%%%%%%
7: %\makeatletter
8: %\@addtoreset{equation}{section}
9: %\makeatother
10:
11: % Command Definitions
12:
13: \def\real{I\negthinspace R}
14: \def\zed{Z\hskip -3mm Z }
15: \def\half{\textstyle{1\over2}}
16: \def\quarter{\textstyle{1\over4}}
17: \def\sech{\,{\rm sech}\,}
18: \def\ie{{\it i.e.,}}
19: \newcommand{\be}{\begin{equation}}
20: \newcommand{\ee}{\end{equation}}
21: \newcommand{\bea}{\begin{eqnarray}}
22: \newcommand{\eea}{\end{eqnarray}}
23: \newcommand{\bml}{\begin{mathletters}}
24: \newcommand{\eml}{\end{mathletters}}
25: %\newcommand{\bml}{\begin{subequations}}
26: %\newcommand{\eml}{\end{subequations}}
27: %
28: \def\aprle{\buildrel < \over {_{\sim}}}
29: \def\aprge{\buildrel > \over {_{\sim}}}
30:
31: \begin{document}
32:
33: \preprint{IUB-TH-044}
34:
35: %\twocolumn[\hsize\textwidth\columnwidth\hsize\csname @twocolumnfalse\endcsname
36:
37: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
38:
39: %\wideabs{ % Uncomment this line for two-column output
40:
41: \title{Solitons on nanotubes and fullerenes as solutions of a modified
42: non-linear Schr\"odinger equation }
43: \renewcommand{\thefootnote}{\fnsymbol{footnote}}
44: \author{Yves Brihaye\footnote{Yves.Brihaye@umh.ac.be}}
45: \affiliation{Facult\'e des Sciences, Universit\'e de Mons-Hainaut,
46: 7000 Mons, Belgium}
47: \author{Betti Hartmann\footnote{b.hartmann@iu-bremen.de}}
48: \affiliation{School of Engineering and Sciences, International University
49: Bremen (IUB), 28725 Bremen, Germany}
50:
51:
52: \date{\today}
53: \setlength{\footnotesep}{0.5\footnotesep}
54:
55:
56: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
57: \begin{abstract}
58: Fullerenes and nanotubes consist of a large number of carbon atoms
59: sitting on the sites of a regular lattice. For pratical reasons it is often
60: useful to approximate the equations on this lattice in terms of
61: the continuous equation. At the moment, the best candidate for such an
62: equation is the modified non-linear Schr\"odinger equation.
63: In this paper, we study the modified non-linear Schr\"odinger equation, which
64: arises as continuous equation in a system describing an excitation
65: on a hexagonal lattice. This latter system can e.g.
66: describe electron-phonon interaction on fullerene related structures
67: such as the buckminster fullerene and nanotubes. Solutions of this modified
68: non-linear Schr\"odinger equation, which have solitonic character,
69: can be constructed using an Ansatz for the wave function, which has time
70: and azimuthal angle dependence introduced previously in the study of spinning
71: boson stars and Q-balls. We study these solutions on a sphere, a disc and on a cylinder.
72:
73: Our construction suggests that non-spinning as well as spinning solutions, which have
74: a wave function with an arbitrary number of nodes, exist.
75: We find that this property is closely related to the series of well-known
76: mathematical functions, namely the Legendre functions when studying the
77: equation on a sphere, the Bessel functions when studying
78: the equation on a disc and the trigonometric functions, respectively,
79: when studying the equation on a cylinder.
80: \end{abstract}
81:
82: \pacs{73.61.Wp, 11.27.+d}
83: \maketitle
84: \renewcommand{\thefootnote}{\arabic{footnote}}
85: \section{Introduction}
86: Fullerenes and fullerene related structures have gained a lot of interest
87: in the past 20 years. Fullerenes were discovered for the first time in
88: 1985 \cite{cks} and are carbon-cage molecules with a large number $n$ of
89: carbon (C) atoms bonded in a nearly spherically symmetric configuration.
90: The configuration is such that three of the valence electrons of the carbon atom
91: form the bounds with the neighbouring carbon atoms. This means, that one ``free''
92: valence electron can ``hop'' along the the $n$ positions in the fullerene.
93:
94: The $C_{60}$ fullerene, the so-called ``buckminster'' fullerene, has 60 carbon
95: atoms arranged on a lattice with 20 hexagons and 12 pentagons. It has a diameter of
96: $d_{60}\approx 7$ \AA. In general, fullerenes with $n$ carbon atoms
97: consists of lattices with 12 pentagons and $(\frac{n}{2}-10)$ hexagons, where $n\geq 24$ has to be even.
98:
99: Fullerides are alkali-doped fullerenes, i.e. fullerenes on which alkali metal atoms
100: such as rubidium (Rb), potassium (K), caesium (Cs) or sodium (Na) are put.
101: These alkali metals donate one electron each. It was observed that such ``doping''
102: leads to a metallic or even superconducting behaviour \cite{haddon}. The interesting
103: thing about this is that the transition temperature $T_c$ of the fullerides
104: is quite high for superconductors, e.g.
105: $T_c=33$ K for a RbCs$_{2}$C$_{60}$ \cite{tanigaki}.
106: In an attempt to explain this high transition temperature, it was found that
107: phonon-electron interactions in the fullerides can explain this phenomenon
108: quite well \cite{varma}.
109:
110: Nanotubes are fullerene related structures, which were discovered for the first time
111: in 1991 \cite{iijima}. These molecules have cylindrical shape and
112: the carbon atoms are arranged on a hexagonal grid. The ends
113: of the nanotubes are typically closed by caps of pentagonal rings. While the
114: tubes discovered in 1991 were multi-layered, single-walled tubes were synthesised
115: in 1993 \cite{iijima2}.
116: The tubes are distinguished
117: by their chirality, diameter and lengths. They have average diameters of 1.2 to 1.4 nm
118: and can be up to 2 mm long. Concerning their chirality, the tubes are typically
119: put into three different classes: armchair, zigzag or chiral tubes \cite{nanotubes}.
120: Depending on their structure, the tubes have different mechanical, thermal, optical
121: and electrical properties. The chirality determines e.g. whether the tubes
122: behave metallic or semiconducting.
123: These facts also lead to the conclusion that a distortion of the hexagonal
124: lattice would influence the energy-band gap. A distortion can be achieved
125: in two ways: either by external, mechanical forces such as bending, stretching or twisting
126: or by an interaction of the lattice with an internal excitation. Trying to
127: explain the dispersion-free energy transport in biopolymers,
128: Davydov realised \cite{davydov} that the interaction of an internal excitation with the
129: distorted lattice (whose distortion was initially caused by the internal
130: excitation) leads to a localised state, a soliton \cite{scott}.
131: For the case of biopolymers, the internal excitation is an amide I vibration.
132:
133:
134: Motivated by the research on fullerene related structures,
135: the interaction of a 2-dimensional hexagonal lattice with an excitation caused
136: by an ``excess electron'' was studied recently \cite{hz}.
137: The Hamiltonian is of Fr\"ohlich-type and describes ``electron-phonon'' interactions.
138: It was found that the existence of soliton-like structures depends crucially on the
139: value of the ``electron-phonon'' coupling.
140:
141: Interestingly, in the continuum limit, this latter system of equations reduces to
142: a modified non-linear Schr\"odinger equation, where the extra term
143: appearing due to the discreteness of the lattice and the
144: ``electron-phonon'' interaction, leads to the stabilisation of the
145: soliton. The appearance of this type of equation has been observed also
146: for a similar system on a quadratic lattice \cite{bepz,bpz}. Consequently, the
147: modified non-linear Schr\"odinger equation was studied on a 2-dimensional
148: plane in \cite{bhz} and on a sphere in \cite{bh} using an Ansatz previously introduced
149: in the study of boson stars \cite{bs} and Q-balls \cite{volkov}. Non-spinning as well
150: as spinning solutions with and without nodes were constructed.
151:
152: In this paper, we study the modified non-linear Schr\"odinger equation arising as
153: continuum limit of the equations describing the interaction of an electron-like excitation
154: with a hexagonal lattice. In Section II, we review some
155: generalities. Using the Ansatz introduced in \cite{bs,volkov}, we
156: construct solutions for three different cases, which we
157: describe in Sections III-V. In Section III, we discuss the results
158: for the solutions on the sphere obtained in \cite{bh} from another point of view, namely
159: we give the spectrum of the equation. In Section IV, we discuss the solutions
160: on a 2-dimensional disc with fixed radius and demonstrate how the solutions
161: obtained on the plane \cite{bhz} can be understood in this context.
162: In Section V, we discuss the equation on a cylinder with fixed radius and
163: fixed length.
164: We give our conclusions in Section VI.
165:
166:
167:
168:
169:
170:
171: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
172: \section{The Modified Non-linear Schr\"odinger equation: Generalities}
173: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
174: The starting point is a coupled systems of discrete equations
175: describing the interaction of an excitation (e.g. an ``excess electron'')
176: with a regular lattice in two dimensions. The excitation is given
177: in terms of a complex valued scalar field $\psi$, while the lattice displacements
178: are described by two fields, one standing for the displacement in $x$,
179: the other for that
180: in $y$ direction. These discrete equations were studied for the quadratic
181: lattice in \cite{bepz,bpz} and for the hexagonal lattice in \cite{hz}.
182: Remarkably, it was found
183: for the hexagonal lattice that in the stationary limit it is possible
184: to replace the full system of equation, in which the excitation field $\psi$
185: and the displacement fields are coupled, by a modified discrete
186: non-linear Schr\"odinger equation. This is not possible for the quadratic lattice.
187:
188: In the continuum limit, both systems of equations reduce to a modified
189: non-linear Schr\"odinger equation. The new term appearing in this
190: equation, which results from
191: both the discreteness of the lattice as well as from the interaction
192: between the lattice and the
193: complex valued scalar field $\psi$ (``lattice-excitation interaction''),
194: leads to the
195: stabilisation of the solution.
196: The difference between the quadratic and
197: the hexagonal case is just the coefficient in front of the non-linear
198: term in the equation, which, however is of the same order of magnitude
199: for the two cases.
200:
201: Since we are dealing with fullerene related structures here, we consider the
202: modified non-linear Schr\"odinger equation, which arises as continuum
203: limit in the hexagonal case \cite{hz, bhz}.
204: In dimensionless variables it reads:
205: \begin{equation}
206: \label{cnls}
207: i\frac{\partial \psi}{\partial t}+\Delta \psi +4g\psi\left(\vert\psi\vert^2+
208: \frac{1}{8}\Delta \vert\psi\vert^2\right)=0 \ .
209: \end{equation}
210: $g$ is the coupling constant of
211: the system
212: and determines the ``strength'' of the non-linear character of the equation.
213: Especially, as can be seen from the derivation of the modified non-linear
214: Schr\"odinger equation from the discrete equations \cite{hz}, the mass
215: of the carbon atom is encoded into $g$.
216:
217:
218:
219:
220: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
221: \section{Equation on the sphere}%%%%%%%%%%%%%%%%%%%%%%
222: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
223: In this section, we will discuss the equation (\ref{cnls}) above
224: on a 2-dimensional sphere of radius $R_s$ aiming to describe ``electronic''
225: excitations on fullerenes and fullerides.
226:
227: The Laplacian operator in this case reads:
228: \begin{equation}
229: \Delta_{sphere}=\frac{1}{R_s^2}\left(\frac{\partial^2}{\partial\theta^2}+
230: \cot\theta\frac{\partial}{\partial\theta}+
231: \frac{1}{\sin^2\theta}\frac{\partial^2}{\partial\varphi^2}\right) \ ,
232: \end{equation}
233: where $\theta$ $\epsilon$ $[0:\pi]$
234: and $\varphi$ $\epsilon$ $[0:2\pi]$. Note that
235: $\psi=\psi(t,\theta,\varphi)$ is a complex valued scalar field.
236:
237: Solutions of (\ref{cnls}) considered on the sphere with radius $R_s$
238: can be characterised by their norm:
239: \begin{equation}
240: \eta^2=\int\limits_0^{2\pi}\int\limits_0^{\pi} |\psi|^2 \sin\theta d\theta d\varphi
241: \end{equation}
242: as well as by their energy:
243: \begin{equation}
244: E=\int\limits_0^{2\pi}\int\limits_0^{\pi} \left( |\vec{\nabla}\psi|^2-2g |\psi|^4 +
245: \frac{1}{4}g(\vec\nabla |\psi|^2)^2\right) \sin\theta d\theta d\varphi \ .
246: \end{equation}
247: We will construct normalised solutions, i.e. $\eta^2=1$. Note that
248: in \cite{bh} a different strategy was employed. By
249: rescaling the wave-function $\psi\rightarrow\psi/\sqrt{g}$, the coupling
250: constant $g$ can be eliminated from the equation. (Of course,
251: the normalisation of the wave-function is lost in this approach, but
252: since $\eta$ was plotted, the corresponding value of $g$ could be easily
253: determined.)
254: In this paper, we put the emphasis on the evolution of the spectrum
255: (corresponding to normalised eigenfunctions) as a function of the
256: coupling constant $g$. Note that $g$ determines the ``strength''
257: of the non-linear part of the equation and that for $g=0$, the equation
258: becomes the ordinary Schr\"odinger equation on the sphere, which, of course,
259: is linear.
260:
261:
262:
263:
264: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
265: \subsection{Discussion of the solutions}
266: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
267: To construct explicit solutions, we employ the following axially symmetric
268: Ansatz \cite{bs,volkov}:
269: \begin{equation}
270: \label{ansatz}
271: \psi(t,\theta,\varphi)=e^{-i\omega t+i m \varphi } \Phi(\theta) \ .
272: \end{equation}
273: The solutions with $m=0$ are non-spinning solutions, while the spinning
274: solutions will have $m\neq 0$.
275:
276: Inserting this Ansatz into (\ref{cnls}), we find the following equation \cite{bh}:
277: \begin{equation}
278: \label{phi}
279: \Phi''+\cot \theta \Phi' +
280: \Phi'^2\frac{g\Phi}{1+g\Phi^2}+\frac{4g\Phi^3+\omega\Phi}{1+g\Phi^2}R_s^2
281: - \frac{m^2}{1+g\Phi^2} \frac{1}{\sin^2 \theta } \Phi=0 \ ,
282: \end{equation}
283: where the prime denotes the derivative with respect to $\theta$.
284:
285: This is a priori a non-linear equation with external parameters $g$ and $\omega$.
286: Here, we will consider it from a slightly different point of view.
287: Let us rewrite (\ref{phi}) as follows:
288: \begin{equation}
289: \label{nlsemod}
290: -\left(\Phi''+\cot \theta \Phi' +\Phi'^2\frac{g\Phi}{1+g\Phi^2}+
291: \frac{g\Phi^3 R_s^2 (4-\omega)}{1+g\Phi^2} - \frac{m^2}{1+g\Phi^2} \frac{1}
292: {\sin^2 \theta } \Phi\right)= \Omega \Phi
293: \end{equation}
294: where
295: \begin{equation}
296: \Omega\equiv\omega R_s^2
297: \end{equation}
298: is the effective spectral parameter.
299: The equation then appears as an eigenvalue Sch\"odinger equation
300: supplemented by a non-linear term coupled with the parameter $g$.
301: The solutions in the linear case, i.e. for $g=0$ are well-known
302: and it is very likely that - by continuity- their deformation due to the
303: non-linear term can be constructed at least for small values of $g$.
304: However, this does not exclude the occurence of supplementary
305: branches of solutions, which may appear typically due to the
306: non-linear structure of the equation.
307:
308: In order to construct regular solutions, (\ref{nlsemod}) has to be solved
309: subject to the following boundary conditions:
310: \begin{equation}
311: \label{bc1}
312: \Phi'|_{\theta=0}=\Phi'|_{\theta=\pi}=0 \ \ \ {\rm for} \ \ m=0
313: \end{equation}
314: and
315: \begin{equation}
316: \label{bc2}
317: \Phi(\theta=0)=\Phi(\theta=\pi)=0 \ \ \ {\rm for} \ \ m\neq 0 \ .
318: \end{equation}
319:
320:
321: To solve the equation (\ref{nlsemod}) numerically subject to the boundary
322: conditions (\ref{bc1}), respectively (\ref{bc2}) for generic
323: values of $g$ and $R_s$, we use a numerical routine based
324: on the Newton-Raphson method \cite{colsys}. Our strategy
325: is to use $\Phi(\theta=0)$, respectively $\Phi'|_{\theta=0}$
326: as a ``shooting'' parameter for the case $m=0$ and $m\neq 0$.
327:
328:
329:
330: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
331: \subsubsection{The case $g=0$}
332: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
333: In the case $g=0$, (\ref{nlsemod}) reduces to the well-known
334: Legendre equation
335: \begin{equation}
336: \Phi'' + \cot\theta \Phi' - m^2 \frac{1}{ \sin^2\theta} \Phi + \Omega \Phi=0 \ .
337: \end{equation}
338: The solutions of this equation are the associated Legendre functions:
339: \begin{equation}
340: \Phi(\theta)=P_l^m (\cos\theta) \ \ \ {\rm with} \ \ \Omega = l(l+1) \ ,
341: \end{equation}
342: where
343: \begin{equation}
344: l \in \mathbb{N} \ \ , \ \ m=-l, -l+1, ..., l-1, l \ .
345: \end{equation}
346: The spatial part of (\ref{ansatz}) are then the
347: spherical harmonics $Y^l_m(\theta,\varphi)=e^{im\varphi}P^l_m(\cos\theta)$.
348: Notice, however, that in the equation studied here there is no distinction
349: between $m$ and $-m$. From now on, we will assume $m$ to be positive,
350: but it is understood that the solution with corresponding $-m$
351: can be trivially obtained.
352:
353: The spherical harmonics have a well-defined parity, i.e.
354: they are either symmetric or anti-symmetric under the reflection
355: $\Phi(\theta) \rightarrow \Phi(\pi-\theta)$. In addition, the number
356: of zeros in the interval $]0:\pi[$ is equal to $ l-|m|$.
357:
358:
359: These results are, of course, very well- known, but we mention them for
360: completeness and because the spectrum of the
361: full equation will approach the spectrum of the Legendre equation
362: in the $g\rightarrow 0$ limit.
363:
364:
365:
366:
367: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
368: \subsubsection{Deformed Legendre functions}
369: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
370: All the solutions discussed in the previous section are
371: continuously deformed when the parameter $g$ is chosen to be larger than zero.
372: For every available spherical harmonic, we observe that
373: a branch of solutions in $g$ exists, on which the solutions have the
374: same symmetry than in the $g=0$ limit. Our numerical results
375: further indicate that the number of zeros $k$ of the solution is
376: preserved all along the branch and is equal to $k\equiv l-m$.
377:
378: Now, we will discuss in detail the numerical solutions corresponding to
379: the deformations of the spherical harmonics $Y^0_0$, $Y_0^1$ and
380: $Y_1^1$.
381:
382: \begin{itemize}
383: \item
384: The non-spinning ($m=0$) and
385: nodeless ($l=0$, $k=0$) solution corresponding to the case $Y^0_0$
386: is in fact
387: the constant solution of the equation, namely
388: \begin{equation}
389: \Phi=\sqrt{\frac{-\omega}{4g}} \ .
390: \end{equation}
391: The eigenvalue $\Omega$ of the corresponding
392: normalised solution has a linear dependence of the parameter $g$:
393: \begin{equation}
394: \Omega=-\frac{R_s^2 g}{\pi} \ .
395: \end{equation}
396: This dependence is shown in Fig.\ref{fig1} for $R_s=2$.
397:
398:
399: \item The non-spinning ($m=0$), one-node ($l=1$, $k=1$)
400: solution is anti-symmetric
401: under the reflection $\theta\rightarrow \pi-\theta$.
402: The function $Y_0^1 \propto \cos\theta$ is progressively deformed
403: by the non-linear term for $g > 0$. For $g\approx 10$ the profile of the
404: function $\Phi(\theta)$ possesses a plateau $\Phi=0$ surrounding the
405: region of $\theta=\pi/2$. Away from this plateau, the solutions
406: resemble two disconnected solitons located at the two poles of the
407: sphere, i.e. the maxima of $\Phi^2(\theta)$ are situated
408: around $\theta=\pi$ and $\theta=0$. Stated in other words, the probability
409: density vanishes in a region surrounding the equator, while it has
410: local maxima at the two poles. Let us
411: stress, that the physical parameters remain
412: monotonic functions of the ``shooting'' parameter $\Phi(\theta=0)$.
413:
414: The dependence of $\Omega$ on $g$ is shown in Fig.\ref{fig1} for $R_s=2$.
415: For $g=0$, we have $\Omega=2$, as expected. It is worth noticing
416: that the two eigenvalues corresponding to the quantum numbers
417: $m=l=0$ and $m=0$, $l=1$ cross at $g\approx 1.5$. This would
418: a priori mean, that the ``excited'' solution becomes the ``ground state''
419: of the equation. However, as we will see in the next section, the appearance
420: of new asymmetric solutions will turn out to have even lower values
421: of $\Omega$.
422:
423:
424: \item The spinning ($m=1$), no-node ($l=1$, $k=0$) solutions
425: arrive as deformations
426: of the spherical harmonic $Y_1^1\propto \sin\theta e^{i\varphi}$.
427: This solution is thus symmetric under the reflection $\theta\rightarrow \pi-\theta$.
428: In contrast to the non-spinning solutions, we notice here
429: that when the coupling constant $g$ is of order
430: $2.5$, the structure of the solutions becomes very sensitive
431: to the value of the ``shooting'' parameter $\Phi'|_{\theta=0}$.
432: In other words, there exists a critical value of $\Phi'|_{\theta=0}\equiv
433: \Phi'_{(cr)}|_{\theta=0}$
434: (depending on $R_s$) such that no solutions exist for
435: $\Phi'|_{\theta=0} >
436: \Phi'_{(cr)}|_{\theta=0}$. Our numerical results, however,
437: suggest that solutions of arbitrary large values of $g$
438: can be constructed when considering the limit $\Phi'|_{\theta=0} \rightarrow
439: \Phi'_{(cr)}|_{\theta=0}$.
440: The corresponding eigenvalue of $\Omega$
441: decreases monotonically with $g$ from $\Omega=2$ (for the case $g=0$)
442: as can be seen in Fig.\ref{fig1} for the case $R_s=2$. Of course, this curve
443: also represents the solution with $m=-1$, so that the eigenvalue $\Omega$
444: is three-fold degenerated in the limit $g=0$.
445:
446: Finally, let us state that for these solutions,
447: the function $\Phi$ has a maximum at the equator and vanishes at the poles.
448: This is an interesting result with view to \cite{harigaya}.
449: In this latter paper it was found that if a C$_{60}$ or a C$_{70}$ molecule
450: is doped with one or two excess electrons, the additional
451: charges accumulate nearly along an equatorial line of the molecule.
452:
453: \end{itemize}
454:
455: Of course, deformations of the Legendre functions for higher $m$ and
456: $l$ can be constructed in a systematic way.
457:
458:
459:
460:
461: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
462: \subsubsection{Asymmetric solutions}
463: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
464: All solutions of the non-linear equation (\ref{nlsemod})
465: discussed previously appear as continuous deformations
466: of the solutions available in the linear limit $g=0$.
467: However, in this section, we will exhibit
468: new branches of solutions, which are fundamentally non-linear
469: phenomena and as these have no linear counterparts.
470: Numerically, we find that these new branches of solutions
471: exist for sufficiently high values of the
472: coupling constant $g$ parametrising the non-linear effect and
473: that some of these branches bifurcate into the branches of the
474: deformed Legendre functions at critical values of $g=g^{(m,l)}_{cr}(R_s)$,
475: which depend on $l$, $m$
476: (and thus on the number of nodes $k=l-m$) as well
477: as on the radius $R_s$.
478:
479: These solutions are charaterised by the fact that
480: they are asymmetric with respect to the reflection
481: $\theta\rightarrow \pi-\theta$.
482:
483: %%(ii) For the values of $g$ that these solutions exist, they have
484: %%lower eigenvalue $\Omega$ than the corresponding
485: %%deformed Legendre function with the same values of $m$, $l$ and $g$.
486:
487:
488:
489: \begin{itemize}
490: \item Asymmetric, non-spinning ($m=0$) and nodeless ($l=0$, $k=0$) solutions exist
491: for $g > g^{(0,0)}_{cr}(R_s=2)\approx 0.838$.
492: When the parameter $g$ is chosen larger
493: than this critical value, our numerical results indicate
494: that this solution has a maximum at $\theta=0$ and a minimum at
495: $\theta=\pi$ and decreases monotonically between these two extrema, which
496: are strictly positive.
497: Thus, the solution is neither symmetric nor anti-symmetric under the
498: reflection $\theta\rightarrow \pi-\theta$. Instead, this reflection
499: applied to the solution leads to an equivalent solution
500: with a maximum (respectively minimum) at $\theta=\pi$ ($\theta=0$).
501: The solutions thus represent a physical state with probability density
502: concentrated in one of the hemispheres. This is a new feature that
503: cannot arise in the deformation of the Legendre functions.
504:
505:
506: In the limit $g\rightarrow g^{(0,0)}_{cr}$ the values of the maximum
507: and the minimum approach each other and the solution becomes constant.
508: The branch of these asymmetric solutions bifurcates
509: into the branch of constant solutions (deformed $Y_0^0$) described
510: in the previous section. The corresponding value of $\omega$ of these asymmetric
511: solutions is lower than the one of the symmetric branch.
512: This is shown
513: in Fig.\ref{fig1} (dotted line) for $R_s=2$.
514:
515: For generic values of $R_s$, the bifurcation of the asymmetric
516: branch occurs at
517: \begin{equation}
518: g^{(0,0)}_{cr}(R_s)=\frac{\pi}{R_s^2-\frac{1}{4}} \ \ ,
519: \end{equation}
520: with
521: \begin{equation}
522: \Omega_{cr}^{(0,0)}=-\frac{R_s^2}{R_s^2-1/4} \ .
523: \end{equation}
524:
525:
526: \item Asymmetric, non-spinning ($m=0$) and one-node ($l=1$, $k=1$)
527: solutions exist for $g > g^{(0,1)}_{cr}(R_s=2)\approx 25$. In this region
528: of parameter values, we find that in fact two branches of solutions exist
529: (see Fig.\ref{fig1}). For a fixed value of $g$, the solution with the lower
530: value of $\Omega$ has a bigger value of $\Phi(0)$ and its minimum
531: lies at smaller values of $\theta$. This is demonstrated for $g=37$, $R_s=2$
532: in Fig.\ref{fig2}.
533: The values of $\omega$ corresponding to these asymmetric solutions
534: are higher than the values of the corresponding
535: (i.e. with for the same values of $g$) antisymmetric solution on the branch
536: of deformed Legendre functions $Y_0^1$. In contrast to the other described cases,
537: we have not succeeded to construct another (a third)
538: branch which would eventually bifurcate from the
539: branch of deformed Legendre functions.
540:
541:
542:
543: %the numerical analysis turns out to be involved
544: %and it was difficult to produce the full asymmetric branch
545: %as well as to determine the critical value of $g$ to a better
546: %accurancy. The reason for this is that both the anti-symmetric (deformed
547: %Legendre) and the asymmetric solutions have profiles which are
548: %nearly equal on a big part of the interval of $\theta$
549: %(typically $\theta\in [0:0.8]$).
550: %Only for $\theta > 0.8$ do these solutions deviate significantly.
551: %We have not attempted to perform a more detailed analysis of the
552: %branch of asymmetric solutions.
553:
554:
555: \item Asymmetric, spinning ($m=1$) and nodeless ($l=1$, $k=0$) solutions
556: exist for $g > g^{(1,1)}_{cr}(R_s=2)\approx 3.79$. Solving the
557: equations for values of $g$ larger than this critical value,
558: we were able to construct a branch of asymmetric solutions
559: (and the corresponding mirror symmetric one). The maximum of these
560: solutions is located between $\theta=0$ and $\theta=\pi/2$ such that
561: the probability density is located in one of the hemispheres.
562: As before, the frequency of the asymmetric solution is lower
563: than the frequency of the corresponding (i.e. same $g$) solution
564: on the deformed $Y_1^1$ branch.
565: \end{itemize}
566:
567: The construction of asymmetric branches with higher values
568: of the spin $m$ and/or higher node numbers $k$ becomes very tricky.
569: This is mainly due to the occurence of many solutions existing
570: for the same values of the parameters and also due to the fact that the profiles
571: of the
572: symmetric and asymmetric (or anti-symmetric and asymmetric) solutions
573: can be very close to each other on a large part
574: of the $[0: \pi]$ interval. The construction of asymmetric branches
575: can thus not be achieved systematically.
576:
577:
578:
579:
580: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
581: \section{Equation on the disc}
582: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
583: In \cite{bhz} the equation (\ref{cnls}) was discussed on a two-dimensional
584: plane and non-spinning as well as spinning solutions with nodes were
585: constructed. Here, we will treat this problem differently. With view to
586: the discussion of the equation on a sphere, we will discuss here
587: the equation on a two-dimensional disc with radius $R_d$ parametrized by the
588: coordinates $\rho$ and $\varphi$. The case studied in \cite{bhz}
589: then arises as limit $R_d\rightarrow \infty$. The Laplacian
590: is given by:
591: \begin{equation}
592: \Delta_{disc}=\frac{\partial^2}{\partial \rho^2}+\frac{1}{\rho}\frac{\partial}{\partial \rho} + \frac{1}{\rho^2}\frac{\partial^2}{\partial \varphi^2}
593: \end{equation}
594: with $\varphi\in [0:2\pi]$. The norm reads:
595: \begin{equation}
596: \eta^2=\int\limits_0^{2\pi}
597: \int\limits_0^{R_d} |\psi|^2 \rho d\rho d\varphi \
598: \end{equation}
599: and the energy is given by:
600: \begin{equation}
601: E=\int\limits_0^{2\pi}\int\limits_0^{R_d} \left( |\vec{\nabla}\psi|^2-2g |\psi|^4 +
602: \frac{1}{4}g(\vec\nabla |\psi|^2)^2\right) \rho d\rho d\varphi \ .
603: \end{equation}
604:
605: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
606: \subsection{Discussion of the solutions}
607: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
608: As for the case on the sphere, we adopt the following Ansatz for the complex valued
609: function $\psi(t,\rho,\varphi)$:
610: \begin{equation}
611: \psi(t,\rho,\varphi)=e^{-i\omega t+im\varphi} \phi(\rho) \ .
612: \end{equation}
613: Inserting this Ansatz gives the following equation:
614: \begin{equation}
615: \phi''+\frac{\phi'}{\rho} + \frac{g \phi \phi'^2}{1+g\phi^2} + \frac{
616: 4g \phi^3+\omega \phi}{1+g\phi^2} - \frac{m^2 \phi}{\rho^2 (1+g\phi^2)}=0 \ ,
617: \end{equation}
618: where the prime now denotes the derivative with respect to $\rho$.
619:
620: We can rewrite, analogue to before, the equation in the following way:
621: \begin{equation}
622: \label{eqdisc}
623: -\left(\phi''+\frac{\phi'}{\rho} + \frac{g \phi \phi'^2}{1+g\phi^2}
624: + \frac{
625: g \phi^3(4-\omega)}{1+g\phi^2} - \frac{m^2 \phi}{\rho^2 (1+g\phi^2)}\right)=
626: \omega\phi \ .
627: \end{equation}
628: This equation looks like the one studied in \cite{bhz}, however, we impose
629: different boundary conditions here:
630: \begin{equation}
631: \label{bc3}
632: \partial_{\rho} \phi|_{\rho=0}=0 \ \ , \ \ \phi(R_d)=0 \ \ \ {\rm for} \ \ m=0
633: \end{equation}
634: and
635: \begin{equation}
636: \label{bc4}
637: \phi(\rho=0)=0 \ \ , \ \ \phi(R_d)=0 \ \ \ {\rm for} \ \ m\neq 0 \ .
638: \end{equation}
639: Again, we used the numerical routine described in \cite{colsys} to solve
640: the equation (\ref{eqdisc}) subject to the boundary conditions
641: (\ref{bc3}), respectively (\ref{bc4}).
642:
643:
644: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
645: \subsubsection{The $g=0$ limit}
646: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
647: In the limit $g=0$ (\ref{eqdisc}) reduces to the linear
648: Schr\"odinger equation
649: in polar coordinates
650: \begin{equation}
651: \label{eqdiscnew}
652: \phi''+\frac{\phi'}{\rho} - \frac{m^2 \phi}{\rho^2}+
653: \omega\phi =0 \ .
654: \end{equation}
655: Note that for static solutions, i.e. $\omega=0$, the equation reduces to the
656: Euler equation with solutions of the form $\psi(t,\rho,\varphi)
657: \equiv \psi(\rho)\propto
658: e^{im\varphi} \rho^{ m}$.
659: For $\omega\neq 0$, the equation becomes the Bessel equation if we define
660: $x\equiv \sqrt{\vert \omega \vert}\rho$ and $F(x)\equiv \phi(x)$:
661: \begin{equation}
662: \frac{d^2 F}{dx^2}+\frac{1}{x}\frac{d F}{dx}
663: + \left(1-\frac{m^2}{x^2}\right) F=0
664: \end{equation}
665: The solutions of this equation, which are regular at the origin are of course
666: the well-known Bessel functions $J_m(x)$ for $\omega > 0$ and the modified
667: Bessel functions $I_m(x)$ for $\omega < 0$. Remembering that
668: the Bessel functions $J_m(x)$ are oscillating and admit an infinite
669: number of nodes on the positive real axis, it is easy to construct
670: solutions of (\ref{eqdiscnew}) with $k$ nodes. This is possible assuming that
671: \begin{equation}
672: \phi(\rho) = J_m(\sqrt{\omega}\rho) \ \ {\rm with} \ \ \phi(R_d) = 0
673: \end{equation}
674: and choosing the value of $\omega$ in such the way that the
675: value $\rho = R_d$ coincides with the $k$-th zero
676: (we call it $x = x_{m,k}$ ) of the Bessel
677: function $J_m$, i.e. $\sqrt{\omega} = x_{m,k}/R_d$.
678:
679: This indicates that in the limit $g=0$ solutions with angular momentum
680: $m$ and with $k$ nodes exist and the spectrum of the equation is positive.
681: The corresponding functions are convex.
682:
683:
684: %%%\begin{equation}
685: %%%\label{besselj0}
686: %%%\phi(r>>1) = J_0(x)
687: %%%\sim \frac{cos(x-\pi/4)}{\sqrt x} \ \ , \ \ {\rm if} \ \omega < 0
688: %%%\end{equation}
689: %%%\begin{equation}
690: %%%\label{besselk0}
691: %%% \phi(r>>1) = K_0(x)
692: %%% \sim \frac{exp(-x)}{\sqrt{x}} \ \ , \ \ {\rm if} \ \omega > 0
693: %%% \end{equation}
694:
695: %%The function $\phi(r)$ therefore oscillates around $\phi = 0$
696: %%for $\omega < 0$ and decays exponentially for $\omega > 0$.
697:
698:
699:
700: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
701: \subsubsection{The $g\neq 0$ case}
702: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
703: If the coupling constant $g$ is positive but small, we expect that
704: the pattern of solutions will be similar to the one
705: available in the limit $g=0$ discussed above. Our numerical results indeed
706: confirm this expectation.
707: For small values of $g$, the solutions are still represented by convex functions
708: and the values of $\omega$ are positive and decrease slowly with $g$
709: as shown in Fig. \ref{fig3} for the case $m=0$, $k=0$.
710: Let is call this range of $g$-values the ``weak coupling-regime''.
711:
712: When $g$ is large enough,
713: the value of $\omega$ becomes negative (see Fig.\ref{fig3})
714: and the solutions have an
715: inflexion point at some intermediate value of $\rho$
716: (see Fig.\ref{fig4}). As a consequence they
717: become much more localized in the region around the origin. In this
718: ``strong coupling-regime'' the solutions can thus be described as being
719: soliton-like. We also note that
720: the value of $\omega$ decreases very strongly
721: with $g$.
722:
723: The difference between the strong and weak coupling is
724: shown in Fig.\ref{fig4}
725: for $g=1$, respectively $g=8$ and $R_d=5$.
726: The value $\omega$ is shown in dependence on $g$ for the non-spinning
727: ($m=0$) and nodeless ($k=0$) solution for different
728: values of the disc-radius
729: $R_d$ in Fig.\ref{fig3}.
730:
731: In Fig. \ref{fig3} we also superposed the values of
732: $\omega$ corresponding to the case of the full plane
733: (i.e. $R_d=\infty$).
734: As has been shown in \cite{hz,bhz}, in this case, solutions exists only for
735: large enough
736: values of $g$: $g \geq \ g_{cr} \approx 2.94$.
737: It is therefore natural to try to understand the
738: pattern available on the plane as the $R_d \rightarrow \infty$
739: limit of the pattern of solutions for the disc.
740: In this respect,
741: it is interesting to notice that in the ``strong coupling-regime''
742: the values of $\omega$ corresponding to a finite disc get very close
743: (especially for the case corresponding
744: to $R_d = 10$) to the
745: values of $\omega$ associated with the case of the infinite disc.
746: This is related to the fact that with increasing $g$, the ``soliton'' becomes
747: more and more peaked around the origin and doesn't ``notice'' whether the disc is
748: finite or infinite.
749: This observation gives a natural explanation for the fact that
750: solutions exist only for
751: $g \geq \ g_{cr}$ in the case of the plane.
752: Indeed in the case of the plane, it was noticed \cite{bhz}
753: that the linearized equation (valid asymptotically)
754: also is of Bessel type.
755: However, the nodeless, localized solutions
756: are exponentially decreasing and can only be associated
757: with the modified
758: Bessel function, which correspond to negative values of $\omega$. That's why,
759: on the plane, nodeless soliton-like solutions exist only
760: for $g \geq 2.94$ and have no ``weak coupling''-counterparts. For large
761: values of $g$ the numerical results strongly suggest the following
762: linear relation between $g$ and the frequency corresponding to the
763: fundamental mode~:
764: $\omega \approx -0.66 g + 2.19$.
765: We also studied the solutions with $k >0$ nodes
766: and $m>0$ and found that the pattern is similar to the
767: case $k=0$ and $m=0$ discussed above in detail.
768:
769: We do not find it useful to discuss them here, but we believe
770: that our results provide a robust evidence that regular solutions
771: of (\ref{eqdisc})
772: with an arbitrary number of nodes and arbitrary (integer) angular
773: momentum exist on the disc/plane.
774:
775: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
776: \section{Equation on the cylinder}%%%%%%%%%%%%%%%%%%%%%%
777: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
778: We also considered the equation on a cylinder of radius
779: $R_c$ and length $2L$.
780: This is a natural extension of the previous results in order to describe
781: soliton-like structures on nanotubes with radius $R_c$ and length $2L$.
782: Using the natural coordinates
783: $z$, $\varphi$, the Laplacian reads:
784: \begin{equation}
785: \Delta_{cylinder}=\frac{1}{R_c^2}
786: \frac{\partial^2}{\partial \varphi^2}+
787: \frac{\partial^2}{\partial z^2}
788: \end{equation}
789: with $\varphi\in [0:2\pi]$ and $z \in [-L,L]$. The norm reads:
790: \begin{equation}
791: \eta^2=\int\limits_0^{2\pi}
792: \int\limits_{-L}^{L} |\psi|^2 dz d\varphi \
793: \end{equation}
794: and the energy is given by:
795: \begin{equation}
796: E=\int\limits_0^{2\pi}\int
797: \limits_{-L}^{L} \left( |\vec{\nabla}\psi|^2-2g |\psi|^4 +
798: \frac{1}{4}g(\vec\nabla |\psi|^2)^2\right) dz d\varphi \ .
799: \end{equation}
800:
801: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
802: \subsection{Discussion of the solutions}
803: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
804: We use the following Ansatz:
805: \begin{equation}
806: \psi(t,z,\varphi) = e^{-i \omega t + i m \varphi} F(z) \ \ , \ \
807: m=0,1,2, \dots
808: \end{equation}
809: The equation (\ref{cnls}) then becomes:
810: \begin{equation}
811: \label{equacyl}
812: F'' + (F')^2 \frac{gF}{1 + g F^2}
813: + \left(4 g F^2 + \omega - \frac{m^2}{R_c^2}\right) \frac{F}{1 + g F^2}=0 \ ,
814: \end{equation}
815: where the prime denotes the derivative with respect to $z$.
816:
817: Again, we can rewrite this equation in terms of a ``spectral equation'':
818: \begin{equation}
819: -\left[F''+(F')^2 \frac{g F}{1+gF^2}
820: + \left( g F^2(4-\omega) - \frac{m^2}{R_c^2}\right) \frac{F}{1+g F^2}\right] =\omega F \ .
821: \end{equation}
822: This equation has to be solved subject to the following boundary conditions:
823: \begin{equation}
824: \label{bccy}
825: F(-L) = F(L) = 0 \ .
826: \end{equation}
827: The equation (\ref{equacyl}) is then solved subject to the boundary conditions
828: (\ref{bccy}) using the numerical routine described in \cite{colsys}.
829:
830: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
831: \subsubsection{The case $g=0$}
832: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
833: The case $g=0$ corresponds to an harmonic equation:
834: \begin{equation}
835: F''=-\left(\omega-\frac{m^2}{R_c^2}\right)F \ .
836: \end{equation}
837: The solutions of this equation subject to the boundary conditions (\ref{bccy})
838: are:
839: \begin{equation}
840: F(z) = \sin\left((z-L) \sqrt{\omega - m^2/R_c^2}\right) \ \ \ {\rm with} \ \ \
841: \omega = \left(\frac{m}{R_c}\right)^2 +
842: \left(\frac{\pi n}{2 L} \right)^2
843: \ \ , \ \ n = 1,2,3,\dots
844: \end{equation}
845: \subsection{The case $g \neq 0$}
846: It is possible to solve (\ref{equacyl}) by quadrature, which gives:
847: \begin{equation}
848: (F')^2 = \frac{C-2 g F^4 - \Omega F^2}{1+gF^2} \ \ , \ \
849: \Omega \equiv \omega - \frac{m^2}{R_c^2}
850: \end{equation}
851: and $C$ is an integration constant. The integration can
852: be performed, however, it leads to an involved and untractable expression
853: of $z$ in terms of $F$, $C$, $\Omega$. Assuming the cylinder to be
854: infinite ($L= \infty$) and $F(\infty)= F'(\infty) = 0$ leads to $C=0$.
855: If we now require in addition that the solution is symmetric under the
856: reflexion $z \rightarrow -z$ (i.e. $F'(0) = 0$) we obtain the
857: relation $\Omega = - 2 g F(0)^2$. This provides a useful
858: check for our numerical results.
859:
860: Performing a numerical analysis of the values of $\omega$ as a
861: function of $g$ (with $L$, $R_c$, $k$, $m$ fixed), we obtain a pattern
862: very similar to that available for the case on a disc
863: (see Fig.\ref{fig3}). The only noticable difference resides in the
864: fact that, in the limit $L = \infty$, a normalizable solution
865: can be constructed for all values of the parameter $g$.
866: This case is, of course, interesting with iew to nanotubes since
867: the lengths $2L$ of a tube is much larger than its diameter $2R_c$.
868: In the case
869: of the ``ground state'' solution, which is non-spinning ($m=0$) and
870: nodeless ($k=0$),
871: the following relations hold :
872: \begin{equation}
873: \omega \approx - 0.1 g^2 \ \ {\rm for} \ \ g \ll 1 \ \ \ , \ \ \
874: \omega \approx -0.555 g + 1.28 \ \ {\rm for} \ \ g \gg 1 \ .
875: \end{equation}
876: Unlike in the case of the sphere,
877: we did not succeed in constructing asymmetric solutions in the case
878: of the cylinder (i.e. solutions, which are neither even nor odd
879: under the reflexion $z \rightarrow -z$) and we believe that there
880: are no such solutions.
881:
882: \begin{equation}
883: \end{equation}
884:
885: \section{Conclusions}
886: In this paper we have presented an extended analysis
887: of a modified non-linear Schr\"odinger equation in 1+2 dimensions.
888: This equation was constructed as an effective continuous equation
889: describing an interaction of an excitation with a 2-dimensional
890: hexagonal lattice. In this paper, we think of the excitation
891: as an ``excess electron'' described by a wave function $\psi$ and the
892: interaction to be ``electron-phonon'' -like. As is well-known from amide I-excitations
893: in biopolymeric lattices such an interaction leads to the creation of a localised state,
894: a ``soliton''.
895:
896: The continuous modified non-linear Schr\"odinger
897: equation was obtained with the idea
898: to approximate the physical system on the
899: plane \cite{hz,bhz}, but it can be modified in order to
900: be considered on different geometrical manifolds, which are
901: justified physically. These are, namely, a cylinder
902: to mimick excitations on a carbon nanotube
903: and a sphere, respectively, in order to describe excitations
904: (e.g. extra electrons) on a fullerene
905: nanomolecule, which is particularly interesting
906: with view to the possibility of describing the high transition
907: temperature of superconducting fullerides.
908:
909: Completed with the different types of
910: boundary conditions related to the geometry and using an Ansatz previously introduced
911: in the study of boson stars \cite{bs} and spinnning Q-balls \cite{volkov},
912: the equation (\ref{cnls}) possesses a lot symmetries, some are
913: continuous (like rotations), some are discrete
914: (like reflexions).
915:
916:
917: In all cases considered the spectrum of solutions is rich, many
918: ``normal-mode''-types of solutions exist, mainly characterized by two
919: integers: the internal angular momentum (or spin) and the
920: number of nodes of the wave function. This is very reminiscent to the
921: principal quantum numbers of more familiar quantum mechanical systems.
922:
923: It might turn out to be difficult to classify the solutions of a
924: non-linear, partial differential equations. The
925: various types of solutions, however, can be
926: traced back from
927: the corresponding solutions
928: available in the weak coupling limit,
929: in which the equation becomes linear and its spectrum is known analytically.
930: In the case of the sphere, disc and cylinder, respectively,
931: we found deformed Legendre functions, deformed Bessel functions
932: and deformed harmonic functions.
933: However, in the case of the sphere, we demonstrated the existence of
934: additional solutions, which are specifically related to the non-linear
935: character of the equation.
936: The main feature distinguishing these supplementary
937: solutions is their asymmetry with respect to the reflexion at
938: the equator of the sphere. In other words, we exhibited
939: the spontaneous symmetry breaking (SSB)
940: of one of the natural symmetries of the initial problem.
941:
942: This SSB occurs when the non-linear coupling becomes strong enough.
943: Similarly it can occur for very different types of non-linear
944: equations. As an example, we point out the bifurcation
945: of bisphaleron solutions from the sphaleron solutions available
946: in the case of the classical
947: equations of the electroweak field theory \cite{kb,yaffe}. In this
948: case, the parity operator is spontaneously broken by the so called
949: bisphaleron solutions.\\
950: \\
951: \\
952: {\bf Acknowledgments} Y.B. gratefully acknowledges the Belgian F.N.R.S.
953: for financial support.
954:
955:
956: \begin{thebibliography}{99}
957: \bibitem{cks} H. W. Kroto, J. R. Heath, S. C. O'Brien,
958: R. F. Curl and R. E. Smalley, Nature {\bf 318} (1985), 162.
959: \bibitem{haddon} R. C. Haddon {\it et al.}, Nature {\bf 350} (1991), 320;
960: {\it for a review see} O. Gunnarsson, Rev. Mod. Phys. {\bf 69} (1997), 575.
961: \bibitem{tanigaki} K. Tanigaki, T. W. Ebbesen, S. Saito, J. Mizuki, J. S. Tsai,
962: Y. Kubo and S. Kuroshima, Nature {\bf 352} (1991), 222.
963: \bibitem{varma} C. M. Varma, J. Zaanen and K. Raghavaclari, Science {\bf 254} (1991), 989;
964: M. A. Schluter, M. Lannoo, M. Needles, G. A. Baraff and D. Tomanek, Phys. Rev. Lett. {\bf 68} (1992), 526;
965: J. Phys. Chem. Solids {\bf 53} (1992), 1473; I. I. Mazin, S. N. Rashkeev,
966: V. P. Antropov, O. Jepsen, A. I. Lichtenstein anbd O.K. Andersen, Phys. Rev. {\bf B45} (1992), 5114.
967: \bibitem{iijima} S. Iijima, Nature (London) {\bf 354} (1991), 56.
968: \bibitem{iijima2} S. Iijima and T. Ichihashi, Nature (London), {\bf 363} (1993), 603.
969: \bibitem{nanotubes} see e.g. M. S. Dresselhaus, G. Dresselhaus and P. Eklund, {\it The Science
970: of fullerenes and carbon nanotubes} (Academic, New York, 1996); {\it Carbon nanotubes, Preparation and Properties}, edited
971: by T. W. Ebbesen (CRC, Boca Raton, FL, 1996); R. Saito, G. Dresselhaus and M. S. Dresselhaus,
972: {\it Physical Properties of carbon nanotubes} (World Scientific, Singapore, 1998);
973: P. J. F. Harris, {\it Carbon Nanotubes and Related Structures} (Cambridge University Press, Cambridge, 1999);
974: {\it Carbon nanotubes: Synthesis, Structure, Properties and Applications}, edited
975: by M. S. Dresselhaus, G. Dresselhaus and P. Avouris (Springer Verlag, Berlin, 2000).
976: \bibitem{davydov} A. S. Davydov, {\it Solitons in Molecular systems} (Reidel, Dordrecht, 1985).
977: \bibitem{scott} A. Scott, Phys. Rep. {\bf 217} (1992), 1; {\it Nonlinear excitations
978: in Biomolecules}, edited by M. Peyrard (Springer, Berlin, 1996).
979: \bibitem{hz} B. Hartmann and W. J. Zakrzewski, Phys. Rev. {\bf B 68} (2003), 184302.
980: \bibitem{bepz}
981: L. Brizhik, A. Eremko, B. Piette and
982: W. J. Zakrzewski, Physica {\bf D 146} (2000), 275; {\bf D 159} (2001), 71.
983: \bibitem{bpz} L. Brizhik, B. Piette and
984: W. J. Zakrzewski, Ukr. Fiz. Zh. (Russ. Ed.) {\bf 46} (2001), 503.
985: \bibitem{bhz} Y. Brihaye, B. Hartmann and W. J. Zakrzewski, Phys. Rev.
986: {\bf D 69} (2004), 087701.
987: \bibitem{bh} Y. Brihaye and B. Hartmann, J. Phys. {\bf A 37} (2004), 1181.
988: \bibitem{bs} F. E. Schunck and E. W. Mielke, in {\it Relativity and
989: Scientific Computing}, Springer, Berlin (1996), 138.
990: \bibitem{volkov} M. Volkov and E. W\"ohnert, Phys. Rev. {\bf D 66} (2002), 085003.
991: \bibitem{colsys} U. Ascher, J. Christiansen and R. D. Russell, Math. Comput. {\bf 33},
992: (1979), 659; ACM Trans. Math. Softw. {\bf 7} (1981), 209.
993: \bibitem{harigaya} K. Harigaya, Phys. Rev. {\bf B45} (1992), 13676.
994: \bibitem{kb}J. Kunz and Y. Brihaye, Phys. Lett. {\bf B 216}, 353 (1989).
995: \bibitem{yaffe}L. G. Yaffe, Phys. Rev. {\bf D 40}, 2723 (1989).
996: \end{thebibliography}
997:
998: \newpage
999: \begin{figure}
1000: \centering
1001: \epsfysize=20cm
1002: \mbox{\epsffile{bh1.eps}}
1003: \caption{\label{fig1}
1004: The eigenvalue $\Omega=\omega R_s^2$ is shown as a function of $g$
1005: for $(m,l)=(0,0)$, $(0,1)$ and $(1,0)$. We have chosen $R_s=2$.
1006: The solid lines denote the deformed Legendre functions, while the dotted
1007: lines are the asymmetric solutions (see text for more details).}
1008: \end{figure}
1009:
1010: \newpage
1011: \begin{figure}
1012: \centering
1013: \epsfysize=20cm
1014: \mbox{\epsffile{bh2.eps}}
1015: \caption{\label{fig2}
1016: The profiles of the function $\Phi(\theta)$ and the derivative $\frac{d\Phi}{d\theta}$
1017: are shown for $(m,l)=(0,1)$ (non-spinning, nodeless solutions),
1018: $g=37$, sphere-radius $R_s=2$ and two different values of the spectral parameter $\Omega=\omega R_s$.}
1019: \end{figure}
1020:
1021:
1022: \newpage
1023: \begin{figure}
1024: \centering
1025: \epsfysize=20cm
1026: \mbox{\epsffile{bh3.eps}}
1027: \caption{\label{fig3}
1028: The eigenvalue $\omega$ is shown as a function of $g$
1029: for the non-spinning ($m=0$), nodeless ($k=0$) solutions on the disc.
1030: We have chosen $R_d=5$, $10$ and $\infty$. Note that $R_d=\infty$ corresponds
1031: to the case of the plane.}
1032: \end{figure}
1033:
1034:
1035: \newpage
1036: \begin{figure}
1037: \centering
1038: \epsfysize=20cm
1039: \mbox{\epsffile{bh4.eps}}
1040: \caption{\label{fig4}
1041: The profiles of the function $\phi(\rho)$ and the derivative $\frac{d\phi}
1042: {d\rho}$
1043: are shown for non-spinning ($m=0$), nodeless ($k=0$) solutions on the disc with radius
1044: $R_d=5$ for two different
1045: values of $g$.}
1046: \end{figure}
1047: \end{document}
1048:
1049:
1050:
1051:
1052:
1053:
1054:
1055:
1056: