quant-ph0303181/MLD.tex
1: %%%%%%%%%%%%%%%%%%%%%%%%%% author.tex %%%%%%%%%%%%%%%%%%%%%%%%%
2: %
3: % sample root file for your contribution to a "contributed book"
4: %
5: % "contributed book"
6: %
7: % Use this file as a template for your own input.
8: %
9: %%%%%%%%%%%%%%%%%%%%%%%% Springer-Verlag %%%%%%%%%%%%%%%%%%%%%%%%%%
10: 
11: 
12: % RECOMMENDED %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
13: \documentclass[multphys,vecphys]{svmult}
14: 
15: % choose options for [] as required from the list
16: % in the Reference Guide, Sect. 2.2
17: 
18: \usepackage{makeidx}     % allows index generation
19: \usepackage{graphicx}    % standard LaTeX graphics tool
20:                          % when including figure files
21: \usepackage{multicol}    % used for the two-column index
22: % etc.
23: % see the list of further useful packages
24: % in the Reference Guide, Sects. 2.3, 3.1-3.3
25: 
26: \makeindex             % used for the subject index
27:                        % please use the style sprmidx.sty with
28:                        % your makeindex program
29: 
30: 
31: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
32: \def\half{\ensuremath{{\textstyle\frac{1}{2}}}}
33: 
34: \begin{document}
35: 
36: \title*{Hamiltonian monodromy as lattice defect}
37: % Use \titlerunning{Short Title} for an abbreviated version of
38: % your contribution title if the original one is too long
39: \author{B. Zhilinskii}
40: % Use \authorrunning{Short Title} for an abbreviated version of
41: % your contribution title if the original one is too long
42:  \authorrunning{Monodromy as lattice defect}
43: \institute{Universit\'e du Littoral, UMR du CNRS 8101, 59140 Dunkerque, France
44: \texttt{zhilin@univ-littoral.fr}}
45: %
46: % Use the package "url.sty" to avoid
47: % problems with special characters
48: % used in your e-mail or web address
49: %
50: \maketitle
51:         \begin{abstract}
52: The analogy between monodromy in dynamical (Hamiltonian) systems
53: and defect in crystal lattices is used in order to formulate some
54: general conjectures about possible types of qualitative features
55: of quantum systems which can be interpreted as a manifestation
56: of classical monodromy in quantum finite particle (molecular) problems.	
57:         \end{abstract}
58: 
59: \section{Introduction} \label{S:intro}
60:  The purpose of this paper is to demonstrate amazing similarity between
61: apparently different subjects: defects of regular periodic lattices,
62: monodromy of classical Hamiltonian integrable dynamical systems, and
63: qualitative features of joint quantum spectra of several commuting
64: observables for quantum finite-particle systems. First of all we
65: recall  why regular lattices and lattices with defects appear naturally
66: for classical integrable Hamiltonian systems and for their quantum analogs.
67: Then we describe several ``elementary dynamical'' defects using
68: tools and language developed in the theory of crystal defects. 
69: Comparison between defects arising in dynamical systems and crystal
70: defects leads to many interesting questions about possibility of
71: realization of certain defects in Hamiltonian dynamics and in crystals. 
72:  
73: \section{Integrable classical singular fibrations and monodromy} 
74: \label{S:classMon}
75: Let us start with the example of Liouville integrable classical 
76: Hamiltonian system with $N$ degrees of freedom 
77: \cite{Arnold}. This means that there exists a set
78: $F=\{F_1, \ldots, F_n\}$ of functions defined on $2n$-dimensional symplectic 
79: manifold $M$, which are functionally independent and mutually in involution.
80: The Hamiltonian $H$ can be locally represented as a function 
81: $H=f(F_1, \ldots, F_n)$. The mapping $F:M\rightarrow R^n$ defines the
82: integrable fibration. We call it a generalized energy-momentum map.
83: Each fiber is the union of connected component of inverse images
84: $F^{-1}(f)$ of points $f\in R^n$. If the differentials $\{dF_1,\ldots,dF_n\}$ of
85: functions from $F$ are linearly independent in each point the fibration is 
86: called regular. If moreover all fibers are compact, the fibration is toric.
87: We will be interested in integrable toric fibrations with singularities
88: of some simplest type.
89: 
90: Let us restrict ourselves to systems with two degrees of freedom.
91: Typical examples of images of singular energy-momentum maps are shown
92: in Figure \ref{F:EMmaps}. The isolated critical value of the map $F$ (see Figure 
93: \ref{F:EMmaps}, left), also known as focus-focus singularity \cite{Lerman,zung97}, 
94: appears, for example, for such problems as spherical pendulum 
95: \cite{cushman83,Duist80,cushman-duistermaat88}, 
96: champagne bottle \cite{Bates,cushman-bates}, 
97: coupling of two angular momenta \cite{SadZhPhysLett}, etc. 
98: The singular fiber in this case is a
99: pinched torus (Figure \ref{F:SingF}, left) with one isolated critical 
100: point of rank 0. 
101: 
102: Presence of a half-line of critical values, together with end point, is typical
103: for nonlinear $1:(-k)$ resonant oscillator \cite{NekhSadZhil}. Each point on the 
104: singular half-line corresponds to singular ``curled torus'' (Figure \ref{F:SingF},
105: center shows curled torus for the case $k=2$) \cite{NekhSadZhil,ColinSan}, 
106: which differs from an ordinary torus due to presence of one
107: circular trajectory which covers itself $k$-times. This particular 
108: circular trajectory is formed by critical points of rank 1 of the map $F$.
109: The end point  (see Figure \ref{F:EMmaps}, center)  
110: corresponds to pinched curled torus with multiple circle shrinking to 
111: a point. This fiber has one critical point of rank 0 and 
112: is topologically equivalent to pinched torus but its
113: immersion into $4D$-space is different. A pinched curled torus for $k=2$ is
114: shown in Figure \ref{F:SingF}, right.
115: 
116: \begin{figure}
117: \centering
118: \mbox{
119: \includegraphics[height=2.5cm]{11resIF.eps} \hskip0.5cm
120: \includegraphics[height=2.5cm]{12resIF.eps} \hskip0.5cm
121: \includegraphics[height=2.5cm]{23resIF.eps} 
122: }
123: \caption{Examples of images of the energy momentum maps for singular toric
124: fibrations.}
125: \label{F:EMmaps}      
126: \end{figure}
127: 
128: 
129: \begin{figure}
130: \centering
131: \mbox{
132: \includegraphics[height=2.5cm]{spT8cut.ps} \hskip0.5cm
133: \includegraphics[height=2.5cm]{badtorus.ps} \hskip0.5cm
134: \includegraphics[height=2.5cm]{curledzpTzs.ps} 
135: }
136: \caption{Singular fibers.  Pinched torus (left).  Curled torus (center).
137: Pinched curled torus (right).}
138: \label{F:SingF}      
139: \end{figure}
140: More general situation with two singular rays starting at one singular point
141: (as shown in Figure \ref{F:SingF}, right) corresponds to $k:(-l)$ 
142: resonant nonlinear
143: oscillator. Example of the integrable fibration corresponding to all 
144: shown in Figure  \ref{F:EMmaps} images of the energy-momentum maps with
145: two integrals $(F_1,F_2)$ in involution  can be
146: written as \cite{NekhSadZhil}
147: \begin{eqnarray}					\label{I1-def}
148:   F_1& = & m_1 \half
149: %{\ensuremath{{\textstyle\frac{m1}{2}}}}
150:  (p_1^2+q_1^2)  - m_2  \half(p_2^2+q_2^2), \\
151:   F_2& = & {\rm Im}\bigl[ (q_1+ip_1)^{m_2}(q_2+ip_2)^{m_1} \bigr]
152:             +  \bigl(m_1\half (p_1^2+q_1^2) + m_2\half  (p_2^2+q_2^2) \bigr)^s,
153: %  \quad {\rm with }\   s > m_1+m_2)/2 .		
154: 	\label{I2-def}
155: \end{eqnarray}
156: with  $s > (m_1+m_2)/2 $, and $m_1,m_2$ positive integers.
157: 
158: All regular fibers are two-dimensional tori. Their fundamental groups are
159: abelian groups $Z^2$ with two generators, corresponding to two basic cycles on a 
160: torus. The fundamental groups for different regular tori are 
161: isomorphic among themselves and to $Z^2$ integer lattice. We can establish 
162: the correspondence between basic cycles defined on different tori by choosing
163: a continuous path in the $4D$-space which is transversal to fibers and by
164: deforming basic cycles continuously along this path. 
165: In particular, for a closed path passing only through  regular
166: tori we get the automorphism of the
167: fundamental group of a chosen regular torus.  The corresponding map of
168: basic cycles is the monodromy map. It is the same for all homotopy 
169: equivalent closed paths. If the path crosses singular lines similar
170: to those taking place for integrable fibration of the 
171: (\ref{I1-def},\ref{I2-def}) resonance oscillators only subgroup of
172: chains can be continuously deformed along the path and the monodromy
173: map in such a case can be defined only for a subgroup of fundamental group
174: \cite{NekhSadZhilinpress}.
175: Nevertheless this map can be linearly extended to a whole group.  In this
176: case the extended monodromy map is represented by a matrix with fractional
177: entries; while in the case of isolated critical values the monodromy map
178: is given by integer matrix $\mu\in SL(2,Z)$. 
179: 
180: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
181: \section{Quantum monodromy} \label{S:quantM}
182: In order to study the manifestation of classical monodromy in associated
183: quantum problems we first need to recall the existence of
184: local action-angle variables  \cite{Arnold,NNaav} and to replace the transformation of 
185: angles by corresponding transformation of actions. In the case
186: of locally regular integrable fibrations local action-angle variables
187: $\{I_k,\phi_k\}$ exist and the linear transformation of angles
188: $\phi^\prime = M\phi$, imposed by the monodromy map, corresponds to the
189: transformation of actions $I^\prime=\left(M^{-1}\right)^\dag I$.
190: 
191: For quantum problems we are interested in the joint spectrum of commuting
192: operators, corresponding to classical integrals $\{F_1,F_2\}$
193: \cite{grondin-sadovskii-zhilinskii,Vu1,waalkens-dullin,child-weston-tennyson}. 
194: The collection  of joint
195: eigenvalues superimposed on the image of the energy-momentum map for
196: classical problem reveals locally the presence of a regular lattice
197: associated with integrality conditions imposed on local actions by
198: quantum mechanics. The lattice of quantum states for quantum problem
199: corresponding to classical oscillators with $1:(-1)$ and $1:(-2)$
200: resonances is represented in Fig. \ref{F:Mon} \cite{NekhSadZhil}.   
201: 
202: \begin{figure}
203: \centering
204: %\includegraphics[height=5cm]{LatDef/Mon.ps}
205: \includegraphics[height=2.8cm]{qcEM1z1.ps} \hskip0.5cm
206: \includegraphics[height=2.8cm]{qcEM1z2z.ps} 
207: \caption{Example of the lattice of quantum states with monodromy.
208: Resonant oscillators (\ref{I1-def},\ref{I2-def}) with $m_1=m_2=1$ (left)
209: and $m_1=1$, $m_2=2$ (right).}
210: \label{F:Mon}      
211: \end{figure}
212: 
213: Due to existence of monodromy, lattice of quantum states can not be
214: regular globally. From figure  \ref{F:Mon}  it is clearly seen that
215: the transport of elementary cell of the locally regular part of the lattice
216: around the singularity shows nontrivial monodromy for a non-contractible
217: close path in the base space (in the space of $F_1,F_2$ values). 
218: The presence of quantum monodromy can be interpreted as a presence of 
219: defects of locally regular lattice of quantum states
220: \cite{SadZhPhysLett}. In the case of
221: isolated critical values of classical problem (Figure  \ref{F:Mon}, left)
222: the choice of elementary cell is arbitrary and the monodromy map is integer.
223: In the case of the presence of singular line at the image of the classical
224: energy momentum map, the dimension of the cell should be increased 
225: (doubled in the case of $1:(-2)$ resonance) in order
226: to ensure the unambiguous crossing of the singular line \cite{NekhSadZhil}.
227: In both cases the presence of singular fibers in classical problem
228: is reflected in  the appearance of 
229: some specific defects of the lattice of quantum states for
230: corresponding quantum problem. We want now to describe these specific
231: defects arising in the quantum theory of Hamiltonian systems using
232: methods and tools from defect theory of periodic lattices 
233: \cite{Kleman,Mermin,Michel,Kroner}. 
234: 
235: 
236: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
237: \section{Elementary defects of lattices} \label{S:DefLat}
238: Let us play  with analogy between 2-D lattice of quantum numbers
239: (or of action variables in the classical limit) and the 2-D lattice
240: of regular solid with defects. More precisely the idea is  to see the correspondence
241: between defects of periodic solids and monodromy which is an
242: obstruction to the existence of global action-angle variables in Hamiltonian
243: dynamics (for integrable systems).
244: 
245: For 2-D  system each  quantum state
246: (or a site for a lattice formed by points) is characterized by two numbers, say
247: $(n_1,n_2)$. The existence of local order means that starting      
248: with some vertex (point of the lattice)
249:  one can  form two vectors, or
250: equivalently the elementary cell of the lattice by defining two vectors
251: as joining  respectively $(n_1,n_2)$ with $(n_1+1,n_2)$ and with
252: $(n_1,n_2+1)$. This corresponds to the choice of the elementary cell
253: with four vertices $\{(n_1,n_2), (n_1+1,n_2), (n_1,n_2+1), (n_1,n_2+1)\}$.
254: The choice of the elementary cell (or equivalently the choice of the
255: basis of the lattice) is not unique. It is defined only up to arbitrary
256: transformation with matrix $M\in SL(2,Z)$. But let us fix
257: some choice for a moment. The existence of local actions
258: in quantum-state lattice
259: language means that by elementary translations in two directions we can
260: label unambiguously all vertices by two numbers with difference in
261: numbers along each edge being 1 for one number and 0 for another.
262: This means that there is no defects (in the local region studied).
263: 
264:  Let us now analyze several different types of
265: defects which can be imagined  for periodic lattices in order to
266: find possible candidates to represent defects of lattices of
267: quantum numbers for quantum problems corresponding to classical
268: Hamiltonian systems with non-trivial (integer and fractional) monodromy.
269: 
270: \subsection{Vacations and linear dislocations} \label{sS:VacDisl}
271: %
272: The simplest point defect well known in solids is 
273: the absence of vertex (or the presence of additional vertex). This defect
274: does not  distort the system of edges not connected with the vacation.
275: The lattice is not deformed even slightly away from the point defect. The
276: elementary cell after a circular trip around the vacation has no modifications.
277: See Fig.  \ref{F:VacLDisl}, left. 
278: 
279: \begin{figure}
280: \centering
281: \includegraphics[height=2.5cm]{Vac.eps} \hskip1.0cm
282: \includegraphics[height=2.5cm]{Disl1.eps} \hskip0.6cm
283: \includegraphics[height=2.5cm]{Disl2.eps}
284: \caption{ Lattice with vacation (left).  Construction of linear dislocation
285: (centre). Lattice with linear dislocation
286: (right). }
287: \label{F:VacLDisl}      
288: \end{figure}
289: Linear dislocation  can be easily imagined to be formed through the 
290: following formal procedure. Let us remove all  vertices on the half-line
291: started at  a given vertex and join the vertices though the gap (see Fig.
292: \ref{F:VacLDisl}, center).
293: Equally, after making a cut along a line of vertices we can introduce
294: additional (one or even several) half-lines.
295: Now the circular path around this defect will show us the existence of the
296: defect. To observe this defect 
297: we should go around it by doing the same number of steps
298: in four directions (say, down, right, up, and left). If the final point will
299: not be the same as the initial point, there is a defect. The vector from
300: initial point to the final point (Burgers vector in solid state physics)
301: characterizes the dislocation. Observe that the elementary cell after the
302: round trip around the dislocation will return exactly to its initial place
303: (see Fig. \ref{F:VacLDisl}, right) because  
304: Burgers vector does not depend on the initial
305: point  and it is exactly the same
306: for all four vertices of the cell.   This means that  vacation and
307: linear dislocations can not be associated with monodromy type defects 
308: of regular lattices.
309: 
310: 
311: 
312: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
313: \subsection{Angular dislocations as elementary monodromy defect}
314: Another general idea to form defect starting from the
315: regular lattice is to remove or to introduce ``the solid angle'' 
316: and to establish
317: in some way the regular correspondence between two boundaries
318: everywhere except one central point.
319: 
320: \label{sS:ElMdef}
321: \begin{figure}
322: \centering
323: \includegraphics[height=2.5cm]{regL12.eps} \hskip0.4cm
324: \includegraphics[height=2.5cm]{regL3a.eps}  \hskip0.2cm
325: \includegraphics[height=2.8cm]{regL3b.eps}  \hskip0.2cm
326: \includegraphics[height=2.8cm]{regL4a.eps}  \hskip0.2cm
327: \includegraphics[height=2.8cm]{regL4b.eps} 
328: \caption{Construction of the angular dislocation by removing or adding
329: one of the solid angles shown on the left picture.  Reconstructed lattices
330: after removing or adding small or large sectors are shown together with
331: transport of elementary cell along a closed path around the defect on
332: the reconstructed lattice. The identification of boundaries after removing
333: or adding solid angle is done by the parallel shift of lattice points in
334: vertical direction.}
335: \label{F:AngDislPl}      
336: \end{figure}
337: 
338: It is important to note that correspondence between two boundaries
339: should be imposed in order to reconstruct the lattice. We will 
340: look for different possibilities but let us start with the simplest one:
341: After removing (or introducing) the solid angle, the reconstruction is d.ne
342: by the parallel shift of lattice points in one chosen direction. 
343: The requirement for reconstructed
344: lattice to be well defined everywhere except singular point can be
345: satisfied only for some special values of removed or added angles.
346: Namely we should impose that the number of removed (added) points at each
347: vertical line is integer and varies linearly with distance from the
348: vertex of the solid angle. Figure \ref{F:AngDislPl}, left   shows 
349: examples of removed or added solid angles.  
350: Two different solid angles correspond respectively  to  removing (adding) of 
351: one or two additional points   
352: from vertical line at each step in the horizontal direction.
353: We can remove or add solid angles in different ways. Figure \ref{F:AngDislMi}
354: illustrates the construction of the removed angle. We start with
355: one chosen point $O$ of the lattice and two basis vectors corresponding
356: to
357: ``horizontal'' and ``vertical'' directions. We put the first cut through the
358: vertex $A$ lying at the $k$-th vertical line counting from the vertex $O$.
359: ($k=6$ on figure \ref{F:AngDislMi}). To construct the second cut 
360: we go from $A$ in vertical direction up to $k$-th horizontal line.
361: Two rays $OA$ and $OB$ show the sector to be removed. Observe that with this 
362: construction $s$ points are removed from $s$-th vertical line.
363:    
364: Figures \ref{F:AngDislPl} show graphically 
365: what happens  with
366: lattice after removing (or adding) solid angles. It is important to note
367: that just by looking on the deformation of elementary cell after the round 
368: trip on the reconstructed lattice we can easily find how big was
369: removed (added) solid angle and what transformation (removing or adding)
370: was exactly done. The absolute value of removed (added) solid angle
371: can be read directly by comparing the form of the initial and final
372: cell. It is sufficient   to write the transformation of two vectors
373: forming elementary cell in matrix form. This matrix is nothing else but
374: the monodromy matrix for actions. For two examples shown in 
375: Figure \ref{F:AngDislPl} this  monodromy matrix has the form
376: $\left(\begin{array}{cc} 1& p\\ 0 & 1\end{array}\right) $
377: or
378: $\left(\begin{array}{cc} 1& 0\\ p & 1\end{array}\right) $ 
379: with $p=\pm 1$ or $p=\pm 2$. One or another form of matrix and the sign of
380: $p$ depends on the choice of the first and second basis vector and on the
381: direction of the circular trip (clockwise or counterclockwise). At the
382: same time the absolute value of $|p|$ is unambiguously related with the absolute
383: value of the removed (added) solid angle. 
384: 
385: \begin{figure}
386: \centering
387: \includegraphics[height=2.5cm]{ExLatCut1.eps} \hskip0.5cm
388: \includegraphics[height=2.5cm]{ExLatCut2.eps}  \hskip0.5cm
389: \includegraphics[height=2.5cm]{ExLatCut4.eps} 
390: \caption{Alternative constructions of the same defect. }
391: \label{F:AngDislMi}      
392: \end{figure}
393: More subtle arguments are
394: needed to distinguish between adding and   removing solid
395: angle  with the same $|p|$.  We will denote below defects obtained
396: by removing solid angle by $(-)$ and by adding solid angle by
397: $(+)$. 
398: 
399: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
400: \subsection{About the sign of the elementary monodromy defect}
401: \label{sS:SignDef}
402: The existence of the sign of Hamiltonian monodromy was conjectured by the
403: author on the basis of analogy between monodromy and 
404: $(+)$ and $(-)$ defects of lattices.
405: The proof was given by Cushman and Vu Ngoc \cite{CushmanSan}. 
406: We give here the characterization of the sign of defect in terms of
407: lattice transformation.
408: \begin{figure}
409: \centering
410: \includegraphics[height=2.3cm]{DownDef.eps} \hskip1.5cm
411: \includegraphics[height=2.3cm]{UpDef.eps}  
412: \caption{Comparison of initial and final cells after the circular path
413: around the singularity on the lattice reconstructed after removing
414: [$(-)$ defect, left] or adding [$(+)$ defect, right]
415: elementary solid angle. Both counterclockwise and 
416:  clockwise circular paths are shown for each type of defects.  }
417: \label{F:DownDef}      
418: \end{figure}
419: 
420: Let first compare initial and final cells for the same reconstructed
421: lattice (with $|p|=1$) obtained by removing simplest solid angle but for
422: both kinds of circular paths (clockwise and counterclockwise).
423: See Fig. \ref{F:DownDef}, left. The identification of initial cell with 
424: the final one
425: can be done only for two vertices. We choose the identified pair being
426: the back side of the cell in the final position (with respect to the 
427: direction given by the sense of rotation).
428: It is clearly seen from the figure  \ref{F:DownDef}, left
429:  that in order to deform the initial cell for $(-)$ defect
430: to the form of the final
431: cell we need to move the front side of the cell in the direction
432: inside the surrounded singularity. This result is unchanged if we
433: apply the same procedure to the clockwise or counterclockwise circular
434: path. 
435: 
436: A similar analysis can be done for lattice reconstructed after 
437: adding solid angle, i.e. for $(+)$ defect, 
438: (see Fig. \ref{F:DownDef}, right) shows that the deformation of the initial cell
439: after the round trip is now in the outside direction with respect to the
440: surrounded singularity. This result  remains again the same for
441: both clockwise and counterclockwise direction of the circular path.
442: 
443: Thus the simple geometrical analysis of the transformation of elementary
444: cell enables one to associate with elementary monodromy the specific
445: defect of the regular lattice. The defect obtained by
446: removing solid angle with $|p|=1$  will be called the elementary monodromy
447: defect. Exactly this defect appears in lattices of quantum states for
448: Hamiltonian systems corresponding to classical Hamiltonian systems with
449: focus-focus singularities. Observe that defects with $|p|>1$ appear naturally
450: in Hamiltonian systems with symmetries. One of the most interesting and
451: physically important systems of this kind is the integrable approximation
452: for hydrogen atom in crossed electric and magnetic fields  
453: \cite{cushman-sadovskii00}.  In classical systems 
454: monodromy with $|p|>1$ corresponds to presence of isolated singular fiber
455: which is $|p|$-times pinched torus \cite{matveev}. Cushman and
456: Vu Ngoc \cite{CushmanSan} have proved that only 
457: focus-focus singularities with the same sign
458: of monodromy can appear in a connected component of 
459: of the image of the generalised energy momentum map of an integrable Hamiltonian
460: system. In non-Hamiltonian systems monodromy of both sign can appear
461: simultaneously \cite{cushman-duistermaatNH}. 
462: 
463: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
464: \subsection{Rational cuts and rational line defects} \label{sS:RatCut}
465: We have seen in previous section that only very special cuts together with
466: matching rules enables us to construct the point defects. Now we will
467: generalize the admissible cuts but keep the matching rule. Let us 
468: start with the example of $1:2$ rational cut which is defined as follows
469: (see Figure \ref{F:1_2cut}).
470: 
471: \begin{figure}
472: \centering
473: \includegraphics[height=2.5cm]{1z2cut.eps} \hskip1.5cm
474: \includegraphics[height=2.5cm]{1z2cutC.eps}  
475: \caption{Construction of the $1:2$ rational lattice defect 
476: starting from the regular
477: square lattice. (Left) - Ambiguity in the transfer of $1\times 1$ cell through 
478: the cut.  (Right) -  
479: Unambiguous transfer of double cell through the cut. }
480: \label{F:1_2cut}      
481: \end{figure}
482: We cut out  half of the solid angle removed in the case of the
483: elementary monodromy defect. After removing this solid angle the two
484: boundaries are different. At one (lower boundary on Figure  \ref{F:1_2cut})
485: points are situated at each vertical line of the lattice. At the upper
486: boundary of the cut points are situated only at each second vertical line.
487: We keep the matching rules, i.e. identify the
488: boundaries by sliding points along the vertical lines. Naturally, the
489: reconstructed lattice is not homogeneous along the identified boundary. 
490: It is seen from the fact that the number of removed points from vertical
491: lines varies like $0,0,1,1,2,2,3,3, \ldots$ along the horizontal direction.
492: (Remark. The number of removed points can be represented in 
493: the form of the  sum of linear and oscillatory functions.) This means that the
494: reconstructed lattice has a line defect.   
495: 
496:  If we try to pass the 
497: elementary cell of the lattice through the cut the result depends on the
498: place where the cell goes through the boundary line. From Figure
499: \ref{F:1_2cut}, left,  it is clear that when the right side of the cell 
500: goes through the cut at even vertical
501: line (supposing the  vertical line going through the
502: vertex of the removed sector to be even) the form of the elementary cell 
503: remains unchanged. In contrast, when the right side of the cell 
504: goes through the cut at odd  vertical line, the form of the cell changes.
505: This ambiguity 
506: can be avoided if instead of elementary $1\times1$ cell we will use
507: larger cell. Namely, we double the dimension of the cell in the horizontal
508: direction. The double cell passes through the cut at any place in a similar
509: way. But the internal structure of the cell changes after crossing
510: the line defect. Cell transforms from 
511: ``face centered'' to ``body centered'' in the crystallographic terminology.
512: But this modification is uniform along the cut. In some way, by
513: increasing the dimension of the cell we neglect the effects comparable 
514: with the dimension of the cell. This enables us to  define the transformation
515: of lattice vectors after traversing a closed path 
516: around the origin of the removed
517: sector. Putting $e_h$ and $e_v$ as horizontal and vertical basis vectors
518: of the square lattice shown in Figure \ref{F:1_2cut}
519: and $\{e_v,e_h^{double}=2e_h\}$ as vectors forming the double cell, 
520: the transformation of  vectors forming the double cell 
521: after a close path around the origin of the removed sector in
522: the counterclockwise direction is
523: \begin{equation}
524: \left(\begin{array}{c} e_v^\prime \\ (e_h^{double})^\prime\end{array} \right) 
525: = \left( \begin{array}{cc} 1&\ 0\\  1&\ 1 \end{array}\right) 
526: \left(\begin{array}{c} e_v \\ e_h^{double}\end{array} \right). 
527: \end{equation}
528: If we extend linearly this transformation to lattice vectors themselves
529: the transformation matrix takes the form 
530: \begin{equation}
531: \left(\begin{array}{c} e_v^\prime \\ e_h^\prime\end{array} \right) 
532: = \left( \begin{array}{cc} 1&\ 0\\ 1/2&\ 1 \end{array}\right) 
533: \left(\begin{array}{c} e_v \\ e_h\end{array} \right) .
534: \end{equation}
535: The so obtained matrix  with fractional entry coincides with the
536: fractional monodromy matrix for actions in the case of $1:(-2)$ resonant
537: classical oscillator and with quantum fractional monodromy for corresponding
538: quantum problem. 
539: 
540: \begin{figure}
541: \centering
542: \includegraphics[height=2.5cm]{13Cut.eps} \hskip1.5cm
543: \includegraphics[height=2.5cm]{23Cut.eps}  
544: \caption{Construction of the $1:3$ and $2:3$ rational lattice defects 
545: starting from the regular
546: square lattice.}
547: \label{F:13Cuts}      
548: \end{figure}
549: 
550: Using the same principle we can construct, for example, line defects by
551: reconstructing lattice after $1:3$ or $2:3$ rational cuts shown in Figure
552: \ref{F:13Cuts}.   This notation means that we remove the solid angle 
553: $\varphi=\tan^{-1}(1/3)$ or $\tan^{-1}(2/3)$ respectively. 
554: We need to triple the dimension of cell in the horizontal direction in order
555: to get unambiguous transformation rules for the cell after crossing the
556: line defect on the 
557: reconstructed lattice. It is clear from the Figure \ref{F:13Cuts} that
558: the monodromy matrices for $1:3$ and $2:3$ rational defects have
559: respectively the form 
560: $\left( \begin{array}{cc} 1&0\\ 1/3&1 \end{array}\right)$
561: and  $\left( \begin{array}{cc} 1&0\\ 2/3&1 \end{array}\right)$.
562: Removing $2:3$ rational solid angle is equivalent to removing twice the
563: $1:3$ rational solid angle. 
564: Generalization to arbitrary rational cut with the same type of matching
565: rules for reconstruction of the lattice is quite obvious and leads to
566: half-line defect with fractional monodromy matrix.
567: 
568: We can also suggest alternative matching rules after rational cuts with the
569: idea to obtain reconstructed lattice with only point rather than the line defect.
570:  Let us consider again as example the $1:3$ rational cut but with different
571: matching rules for two boundaries (see Figure \ref{F:1_3Arn}). 
572: 
573: \begin{figure}
574: \centering
575: \includegraphics[height=2.5cm]{1z3cut.eps}  
576: \caption{Matching rules for reconstruction of lattice after rational
577: cut. Example of $1:3$ cut. The monodromy of the resulting defect is the
578: Arnol'd cat map. }
579: \label{F:1_3Arn}      
580: \end{figure}
581:  It is clear that
582: if we want to have on the reconstructed lattice only point defect
583: all vertices on two boundaries should be consecutively identified. This
584: identification imposes matching rule for one of the basis vectors of the
585: lattice. Another should be chosen in such a way that two new basis
586: vectors form elementary cell of the same volume, i.e. they should be 
587: related one to another with $SL(2,Z)$ transformation. 
588: %One can verify that for
589: %each rational cut it is possible to find the $SL(2,Z)$ matrix, which
590: %corresponds to matching rule for going over the cut. Otherwise for each
591: %$SL(2,Z)$ matrix one can construct a rational cut such that matching rules
592: %for this cut are expressed in terms of a chosen matrix. 
593: In fact this matrix
594: is precisely the monodromy matrix of the point defect just by the
595: construction. In the case of $1:3$ rational cut shown in
596:  Fig. \ref{F:1_3Arn} the resulting monodromy matrix has the form
597: $\left( \begin{array}{cc}0& -1\\1&\ \ 3 \end{array} \right )$.
598: This matrix is known as Arnol'd cat map
599: \cite{ArnAv}. A lot of different examples of 
600: point defects can be constructed in a similar way. But we will take 
601: as elementary point defect only $(-)$ defects corresponding to
602: elementary monodromy matrix. We will demonstrate now that all other defects
603: can be considered as more complicated objects composed in some way from
604: several elementary ones.  
605: 
606: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
607: \section{Defects with arbitrary  monodromy} \label{sS:MultD}
608: We now turn to the description of defects which can be characterized by
609: arbitrary monodromy matrices. As  soon as
610: the choice of the basis of the lattice is ambiguous, the
611: matrix representation of the monodromy transformation is basis dependent.
612: For example the monodromy matrix
613: $M_a=\left(\begin{array}{cc} 1& 1\\ 0 & 1 \end{array}\right) $
614: after the transformation to another basis through the similarity
615: $M_a^\prime = A M_a A^{-1}$ with $A$ being arbitrary $SL(2,Z)$ matrix
616: takes the form
617: $$\left(\begin{array}{cc} a& b\\ c & d\end{array}\right)
618:    \left(\begin{array}{cc} 1& 1\\ 0 & 1\end{array}\right)
619:    \left(\begin{array}{cc} d& -b\\ -c & a\end{array}\right) =
620:    \left(\begin{array}{cc} 1-ac& a^2\\ -c^2 & 1+ac\end{array}\right)$$
621: From this family of equivalent matrices it is immediately clear that
622: matrices $\left(\begin{array}{cc} 1& 1\\ 0 & 1\end{array}\right)$
623: and $\left(\begin{array}{cc} 1& 0\\ -1 & 1\end{array}\right)$ are
624: equivalent but they are written in different frames. In contrast, matrix
625: $\left(\begin{array}{cc} 1& -1\\ 0 & 1\end{array}\right)$ 
626: is equivalent to
627: $\left(\begin{array}{cc} 1& 0\\ 1 & 1\end{array}\right)$ 
628: but is not equivalent to 
629: $\left(\begin{array}{cc} 1& 1\\ 0 & 1\end{array}\right)$ in spite of the fact
630: that these two matrices are mutually inversed.
631: 
632: In order to formulate precise statement about equivalence or in-equivalence
633: of different defects we should first establish equivalence of $SL(2,Z)$
634: matrices with respect to conjugation by elements of $SL(2,Z )$, i.e. to describe 
635: classes of conjugated elements of $SL(2,Z )$ group.
636: 
637: It is well known that the trace and the determinant of the matrix are
638: invariant with respect to similarity transformation. But these invariants
639: are not sufficient to completely characterize classes of conjugated elements.
640: Before looking for $SL(2,Z)$ matrices let us start with $SL(2,R)$ ones.
641: 
642: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
643: \subsection{Topological description of unimodular matrices}
644: Let consider the subspace of $SL(2,R)$ matrices 
645: $M= \left(\begin{array}{cc} \alpha& \beta\\ \gamma & \delta\end{array}\right)$
646: with ${\rm Tr}\ M = K$. This means that four matrix elements
647: $\alpha, \beta, \gamma, \delta$ are related by two equations
648: \begin{equation}
649: \alpha\delta-\beta\gamma=1, \ \ \ \  \alpha+\delta = K.
650: \end{equation} 
651: Eliminating one parameter (say $\alpha$) we get the following relation
652: between three parameters $ \beta, \gamma, \delta$
653: \begin{equation}
654: -1+K\delta -\delta^2-\gamma\beta = 0.
655: \end{equation}
656: We can interpret this relation as the geometrical description of all 
657: $SL(2,R)$ matrices with given trace in the three dimensional space of
658: parameters $ \beta, \gamma, \delta$. The geometrical form of so obtained
659: surface depends on the value of $K$. Topologically there are three different
660: situations. 
661: 
662: If $K=\pm2$ we have double cone with vertex corresponding to
663: $\pm \left(\begin{array}{cc} 1& 0\\ 0 & 1\end{array}\right) $ matrix.
664: If $|K|>2$ we have a  hyperboloid of one-sheet
665: and if $-2<K< 2$ we have two-sheeted hyperboloid. 
666: We can schematically represent the whole family of $SL(2,R)$ matrices
667: by filling the solid torus in three-dimensional space of parameters by
668: surfaces corresponding to all possible values of traces. This 
669: representation is given in Figure \ref{torL2R}.
670: 
671: %%%%%%%%%%%%%%%%%%%%%%%%%%
672: \begin{figure}
673: \centering
674: \includegraphics[height=5cm]{torL2R.eps}
675: \caption{The topological structure of the space of 2 by 2 matrices with
676: determinant 1. The solid tore in 3-D space is foliated by levels
677: corresponding to a given value of the trace of matrix $M$. 
678: \label{torL2R} }
679: \end{figure}
680: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
681: The existence of two disjoint connected components for matrices with $-2\leq 
682: {\rm Tr} M \leq 2$ implies the existence of additional invariant which 
683: classifies matrices with the same trace into smaller subclasses of conjugate
684: elements. We need such description but only for matrices in $SL(2,Z)$.
685: The important difference between $SL(2,R)$ and $SL(2,Z)$ cases 
686:  is due to the fact that
687: there is only finite number of possible values of the trace 
688: in $SL(2,Z)$ which correspond to matrices with complex eigenvalues 
689: in $SL(2,R)$. In physical language this is the
690: consequence of the fact that only axes of second, third, fourth and sixth
691: orders are compatible with the existence of the lattice.
692:  
693: Formal proof: Characteristic polynomial for the $SL(2,Z)$ matrix $M$ has the form
694: $\lambda^2 - ({\rm Tr}\ M) \lambda +1=0$. It has complex eigenvalues only if
695: the discriminant $({\rm Tr}\ M)^2 -4 <0$. As soon as the trace is integer,
696: it is only possible that  $({\rm Tr}\ M)=0, \pm 1$.
697: 
698: Now we can return to the study of $SL(2,Z)$ case.
699: 
700: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
701: \subsection{Classes of conjugated elements and ''normal form'' of 
702: $SL(2,Z)$ matrices}
703: The matrices  $M\in SL(2,Z)$ will be named parabolic, elliptic, or
704: hyperbolic, depending on their trace. Parabolic matrices have trace
705: equal $\pm2$ and their eigenvalues are $\{+1, +1\}$ or  $\{-1, -1\}$.
706: Elliptic matrices have trace $\pm1$ or $0$. Their eigenvalues are complex
707: numbers. Hyperbolic matrices have $|{\rm Tr}\ M|>2$. Their eigenvalues
708: are real irrational numbers. 
709: Identity matrix 
710: $\left(\begin{array}{cc}  1 & 0\\  0 & 1 \end{array}\right)$ and minus identity
711: $\left(\begin{array}{cc} -1 & 0\\  0 &-1 \end{array}\right)$ commute with
712: all elements from $SL(2,Z)$ and each form a proper class of conjugate
713: elements consisting of one element. We will consider these classes
714: separately. In Fig, \ref{torL2R} these matrices correspond to vertices of
715: double cones of matrices with trace $\pm2$. 
716: 
717: Within each class of conjugate elements we can choose one matrix to be
718: the ``normal form''. All classes of conjugate elements together with
719: normal forms are listed in Table \ref{T:sl2zCE}.
720: 
721: \begin{table}
722: \centering
723: \caption{Classes of conjugated elements of $SL(2,Z)$ group together with
724: normal forms of matrices for each class.}
725: \label{T:sl2zCE}      
726: %
727: \begin{tabular}{l|ccccc}
728: \hline\noalign{\smallskip}
729: Trace    & $K, (|K|>2)$ & $\pm2$ & $\pm1$ & $0$  \\
730: \noalign{\smallskip}\hline\noalign{\smallskip}
731: Module    & $-$ & $p=0,\pm1,\ldots$ & $\varepsilon=\pm1$ & $\varepsilon=\pm1$  \\
732: \noalign{\smallskip}
733: Normal form & $\left(\begin{array}{cc} K& 1\\ -1 & 0\end{array}\right)$ 
734: & $\left(\begin{array}{cc} 1&  p \\  0 & 1\end{array}\right)$ &
735: $\left(\begin{array}{cc} \pm(1+\varepsilon)/2 & \varepsilon \\
736:            -\varepsilon & \pm(1-\varepsilon)/2 \end{array}\right)$ &
737: $\left(\begin{array}{cc} 0& \varepsilon \\ -\varepsilon & 0 
738:  \end{array}\right)$  \\
739: \noalign{\smallskip}\hline
740: \end{tabular}
741: \end{table}
742: 
743: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
744: \subsection{Several elementary monodromy defects}\label{sS:MultElMD}
745: We have defined  the construction of elementary defects using one chosen
746: lattice basis. Both cuts and matching rules were precisely defined in that
747: basis. But the choice of the lattice basis is not unique. 
748: If there are two defects, characterized both by elementary monodromy matrix
749: but the choice of basis and the orientation of cuts for these two defects
750: are different, the global monodromy, corresponding to transformation of 
751: elementary cell after a circular path around both defects, depends on
752: relative orientation of two removed solid angles. Let us again start with some
753: particular examples of systems with several elementary defects. 
754: 
755: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
756: \subsubsection{Disclinations as a composition of elementary 
757: monodromy defects}\label{ssS:Discl}
758: Figure \ref{F:90discl} shows regular square lattice with three defects
759: corresponding to elementary monodromy. For two removed angles (sectors
760: around horizontal lines) the reconstruction of lattice is done 
761: through sliding points in vertical directions. For the third removed angle
762: the identification of boundaries is done through the horizontal sliding
763: of lattice points.  
764: 
765: \begin{figure}
766: \centering
767: \includegraphics[height=4.5cm]{regLT1.eps} 
768: \hskip0.5cm
769: \includegraphics[height=4.5cm]{2defTr.eps}  
770: \caption{(Left) - Regular square lattice with three  
771:  elementary monodromy defects.  (Right) - Regular triangular
772: lattice with two elementary monodromy defects.  }
773: \label{F:90discl}      
774: \end{figure}
775:  The cumulative effect of three elementary cuts is the rotation
776: of the elementary cell by $\pi/2$. 
777: The direction of the rotation of the
778: elementary cell is defined by the direction of the circular path around    
779:  singularities. Such defect is known in solid
780: state physics as $\pi/2$ rotational disclination. 
781: It is easy to see that the same effect takes place if the
782: three cuts are distributed in another way between vertical and horizontal
783: directions (two vertical and one horizontal). To see this it is just
784: sufficient to look at the same figure after rotating it  by $\pi/2$.
785: 
786: Naturally, similar construction can be done with three singular points
787: corresponding to  adding the
788: solid angle and reconnecting new boundaries through horizontal or
789: vertical shift.
790: The resulting effect on the elementary cell is again the $\pi/2$ rotation,
791: but now the rotation of the elementary cell is in opposite direction
792: (as compared to the direction of the circular loop around the
793: singularity). 
794: 
795: The global effect in both cases can be reproduced  by removing 
796: (or adding) the
797: solid angle $\pi/2$ and by reconstructing lattice through identification
798: of two boundaries by rotating them  as it is shown in Figure
799: \ref{F:90Rotdiscl} where  $\pi/2$ solid angle is removed.  
800: Naturally one can also remove $\pi$ or $3\pi/2$ solid angle  or to
801: add $\pi/2$ or $k\pi/2$ solid angle as it is shown in 
802: Figure \ref{F:addRotdiscl}. This gives negative or  positive  rotational 
803: disclinations.
804:  
805: \begin{figure}
806: \centering
807: \includegraphics[height=4.5cm]{regL5.eps}  \hskip1cm
808: \includegraphics[height=4.5cm]{regL5a.eps}  
809: \caption{Construction of the rotational disclination
810: by removing solid angle $\pi/2$ shown on the left picture. }
811: \label{F:90Rotdiscl}      
812: \end{figure}
813: 
814: \begin{figure}
815: \centering
816: \includegraphics[height=3.0cm]{regL6.eps}  \hskip0.3cm
817: \includegraphics[height=3.0cm]{regL6a.eps}  \hskip0.3cm
818: \includegraphics[height=3.0cm]{regL7.eps}  
819: \caption{Construction of the rotational disclination
820: by removing solid angle $\pi$ (Left) and $3\pi/2$ (Right). 
821: The reconstructed lattice after removing $\pi$ solid angle (center). }
822: \label{F:180Rotdiscl}      
823: \end{figure}
824: 
825: \begin{figure}
826: \centering
827: \includegraphics[height=4.5cm]{regL8.eps}  \hskip1cm
828: \includegraphics[height=4.5cm]{regL8a.eps}  
829: \caption{Construction of the rotational dislocation (disclination)
830: by introducing solid angle $k\pi/2$. $k=1$ on the left
831: and $k=4$ on the right picture. }
832: \label{F:addRotdiscl}      
833: \end{figure}
834: 
835: The same construction made with triangular lattice and with two elementary
836: monodromy defects rotated one with respect to another over $2\pi/3$ gives
837: the cumulative effect consisting in rotation of elementary cell over
838: $2\pi/6$ after a close path surrounding two elementary defects
839: (see figure \ref{F:90discl}, right). The cumulative
840: effects of such two elementary monodromy defects is the $\pi/3$ rotational
841: disclination. Its multiple, positive or negative analogs can be
842: immediately constructed. 
843: 
844: Rotational disclinations are well known defects in the solid state physics.
845: From the point of view of defects of quantum state lattices and classical
846: Hamiltonian monodromy, the elementary monodromy defects seems to be 
847: more fundamental.
848: Any rotational disclination can be constructed as a global effect
849: in systems with several elementary monodromy defects. 
850: 
851: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
852: \subsubsection{Multiple defects with trivial global monodromy}
853: \label{ssS:Mult Triv}
854: Let us now consider in more general way the correspondence between
855: local and global monodromy for a system of elementary defects.
856:  Figure \ref{F:prodM} illustrates this relation.
857: 
858: Suppose we have two defects $c_1$ and $c_2$
859: characterized by monodromy matrices $M_1$ and $M_2$. These monodromy 
860: matrices are obtained by going around the defect $c_i$ starting from point
861: $b_i$ and using local basis associated with point $b_i$. If we are interested
862: now in global monodromy which corresponds to a close loop going around two
863: defects and starting at initial point $b_0$ with its own local basis
864: we can calculate the global monodromy by going first from $b_0$ to $b_1$,
865: making close loop around $c_1$, returning back by the same way, and repeating
866: the same for the second defect. The global monodromy calculated in this way
867: should be the same by homotopy arguments. If the modification of the
868: basis between $b_0$ and $b_i$ is described by matrices $A_i$ the global
869: monodromy $M$ can be expressed in terms of $M_1$ and $M_2$ as
870: $M= A_1 M_1 A_1^{-1} A_2 M_2 A_2^{-1}$. 
871: 
872: 
873: \begin{figure}
874: \centering
875: \includegraphics[height=4cm]{prodM.eps} 
876: %\includegraphics[height=2.5cm]{prodMa.eps}  
877: \caption{Relation between local monodromy matrices for isolated defects and
878: global monodromy for the circular path around two defects }
879: \label{F:prodM}      
880: \end{figure}
881: Naturally for an arbitrary system of elementary defects the global monodromy
882: matrix can always be represented in the form
883: $M=\prod_i A_iM_iA_i^{-1}$. As it was already noted the monodromy
884: matrix is defined up to conjugation with $SL(2,Z)$ matrices, i.e. 
885: defects with different monodromy are in one to one correspondence with classes
886: of conjugate elements of $SL(2,Z)$ matrices. 
887: 
888: One can easily verify that arbitrary $SL(2,Z)$ matrix can be represented
889: in the form of product of matrices, conjugate to elementary monodromy
890: matrices with one chosen sign \cite{CushZhil}. In particular, 
891: the identity matrix can also be
892: represented in the form of product of matrices conjugate to
893: elementary monodromy matrix. It is obvious that four $\pi/2$ rotational
894: disclinations (or six $\pi/6$ rotational disclinations)  give trivial 
895: global monodromy. In fact the elementary cell make a $2\pi$ rotation
896: when going along close path surrounding these defects and in spite of 
897: the fact that the monodromy is trivial the close path is not contractible and
898: the defect exists. An easy consequence of this statement: The monodromy
899: matrix (defined up to conjugation with $SL(2,Z)$ matrices)
900: is not sufficient to distinguish defects. Two defects with the same
901: monodromy matrix can be further labeled by the  number $k$ of $2\pi$
902: rotations of the elementary cell after a close path around a defect.
903: This additional number $k$ can be arbitrary integer $k=0, \pm1, \pm2, \ldots$.
904: Note that an elemenraty $(-)$ monodromy defect can be constructed as a 
905: cumulative effect of eleven elementary $(+)$ monodromy defects 
906: \cite{CushZhil}.
907: 
908: One can easily obtain trivial global monodromy for a close path around two
909: defects with different signs. Figure \ref{F:pm11} shows construction of
910: two elementary monodromy defects with different signs. Signs plus and minus
911: indicate respectively adding and  removing of the same solid angle.
912: 
913: \begin{figure}
914: \centering
915: \includegraphics[height=2.5cm]{pm11.eps} \hskip0.5cm
916: \includegraphics[height=2.5cm]{pm11A.eps} \hskip0.5cm
917: \includegraphics[height=2.5cm]{pm11B.eps}  
918: \caption{Lattice with two elementary monodromy defects of different sign. }
919: \label{F:pm11}      
920: \end{figure}
921:   
922: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
923: \subsection{Several rational line defects} \label{sS:MultRatL}
924: Let us now discuss examples of lattices with multiple rational line
925: defects. We assume below that all defects are of the same sign, i.e.
926: obtained by removing solid angle.
927: We start with example of two defects $1:2$ and $1:3$ which have
928: similar orientation (see Figure \ref{F:2_3cut}, left). These two defects
929: model the singularities of integrable toric fibration for $2:-3$
930: resonant oscillator (\ref{I1-def}). Elementary $1\times 1$ cell can not
931: cross unambiguously both defect half-lines. The cell should be doubled in
932: horizontal direction in order to cross unambiguously the $1:2$ defect. In a
933: similar way the cell should be tripled in the same direction in order to
934: cross the $1:3$ defect. This means that only $1\times6$ 
935: cell which is six times larger in the horizontal direction can cross
936: both defects. Using such cell we can go along a close path surrounding
937: the singular vertex. After linear extension to elementary lattice vectors
938: we get the fractional monodromy matrix
939: $\left(\begin{array}{cc} 1& \ 0\\ 5/6 &\  1\end{array}\right)$. 
940: 
941: \begin{figure}
942: \centering
943: \includegraphics[height=3.cm]{2z3cut.eps} \hskip0.5cm
944: \includegraphics[height=3.cm]{2z3cutP6.eps}  
945: \caption{Construction of lattice with two rational defects, $1:2$ and $1:3$
946: (Left)- Parallel defects which correspond to 
947: singular one-dimensional strata for $2:(-3)$
948: resonance oscillator. (Right) - Two orthogonal defects.  }
949: \label{F:2_3cut}      
950: \end{figure}
951: 
952: If  two rational defects, $1:2$ and $1:3$, have different orientations
953: the situation becomes quite different. Figure  \ref{F:2_3cut}, right shows
954: these two defects with orthogonal orientation. In order to pass through
955: horizontal $1:2$ defect the cell should be doubled in horizontal direction.
956: In order to pass through $1:3$ vertical defect the cell should be tripled in
957: vertical direction. Moreover, one should note that all vertices of the cell
958: should lie on even vertical lines and on horizontal lines having the same number
959: modulo 3. This means that we need to take at least $6\times6$ cell in order
960: to cross unambiguously both rational cuts. The resulting monodromy matrix
961: for a counterclockwise path around two singular points has the form
962: $\left(\begin{array}{cc} 5/6& \ -1/3\\ 1/2 &\  1\end{array}\right)$. 
963: This is an elliptic $SL(2,Q)$ matrix.
964: 
965: \subsubsection{Rational defect line with ends and singular points} 
966: \label{ssS:LineEnds}
967: It is quite easy to construct rational defect with two ends. It is sufficient
968: to start to cut solid angle as it was done for rational defect but at
969: some another point to change the slope and to continue with another
970: slope related to integer monodromy defect. This situation is shown in 
971: Figure \ref{F:multsing} on three different examples. Fig. \ref{F:multsing},
972: left shows $1:2$ cut which starts at point $A$ but at point $B$ the
973: angle of the cut changes. It becomes equal angle characteristic to
974: elementary monodromy defect. This means that the line $1:2$ defect on
975: reconstructed lattice has two ends and the monodromy around each end 
976: is $\left(\begin{array}{cc} 1& \ 0\\ 1/2 &\  1\end{array}\right)$. At the same
977: time the global monodromy for close path surrounding $1:2$ defect 
978: is $\left(\begin{array}{cc} 1& \ 0\\ 1 &\  1\end{array}\right)$.
979: In a similar way (see  Fig. \ref{F:multsing}, center) we can start 
980: at point $A$ with $1:3$
981: cut and change at point $B$ the angle in order to get again on the
982: reconstructed lattice  elementary monodromy for global close path. 
983: This means that surrounding point $A$ we get the monodromy
984:  $\left(\begin{array}{cc} 1& \ 0\\ 1/3 &\  1\end{array}\right)$, while
985: surrounding point $B$ the monodromy becomes
986: $\left(\begin{array}{cc} 1& \ 0\\ 2/3 &\  1\end{array}\right)$. Two ends are
987: not equivalent. Naturally, we can change angle several times. This gives
988: the line defect with singular points on it. Each singular point corresponds
989: to modification of value of solid angle removed from the lattice. 
990:  Fig. \ref{F:multsing}, right shows example with three singular points.
991: Generalization to more complicated examples is straightforward.
992: 
993: \begin{figure}
994: \centering
995: \includegraphics[height=1.8cm]{FrLendsB.eps} \hskip0.5cm
996: \includegraphics[height=1.8cm]{FrLendsC.eps} \hskip0.5cm
997: \includegraphics[height=1.8cm]{FrLendsA.eps} \hskip0.5cm  
998: \vskip10pt 
999: \includegraphics[height=1.6cm]{FrLendsD.eps} \hskip0.6cm
1000: \includegraphics[height=1.6cm]{FrLendsF.eps} \hskip0.6cm
1001: \includegraphics[height=1.6cm]{FrLendsE.eps}  \hskip0.3cm
1002:  \caption{Construction of the line defects with ends and singular points. 
1003: (Left) - Defect with equivalent ends. (Center) - Defect with inequivalent
1004: ends. (Right) - Defect with inequivalent ends and additional singular point
1005: splitting defect into two fragments.}
1006: \label{F:multsing}      
1007: \end{figure}
1008: 
1009: Figure \ref{F:HalfIntcut} demonstrates geometrically modifications
1010: which occur with elementary cell after 
1011: traversing different closed paths on the lattice with
1012: $1:2$ defect with two ends. To find the global monodromy one can use
1013: elementary $1\times1$ cell (Fig. \ref{F:HalfIntcut}, left), whereas it is
1014: not possible to cross the line with such a cell because of ambiguity of
1015: cell modifications. 
1016:  
1017: \begin{figure}
1018: \centering
1019: \includegraphics[height=2.3cm]{2endCutA.eps} \hskip0.3cm
1020: \includegraphics[height=2.3cm]{2endCutB.eps}  \hskip0.3cm
1021: \includegraphics[height=2.3cm]{2endCutC.eps}  
1022: \caption{Lattice with two half-integer defects. Defect is an interval with
1023: two end points.
1024: (Left)-  Elementary cell cannot cross unambiguously the defect line.
1025: (Center) - Both ends have the same $(1/2)$ monodromy. (Right) - Figure eight 
1026: close path is not contractible but the monodromy is trivial. }
1027: \label{F:HalfIntcut}      
1028: \end{figure}
1029: Taking $1\times2$ cell we can easily go around each singular point
1030: and see  (Fig. \ref{F:HalfIntcut}, center)
1031: that the result is exactly the same for both points, namely the half of the
1032: global modifications. 
1033: 
1034: The last Figure \ref{F:HalfIntcut}, (right) shows the evolution of the
1035: double cell along the figure eight close path which goes around two
1036: centers but in opposite directions. This close path results in trivial
1037: monodromy, the cell has no modifications after returning to the
1038: initial point.
1039: 
1040: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1041: \section{Is there mutual interest in defect - monodromy correspondence?}
1042: Stimulated by analogy between classical and quantum monodromy for
1043: Hamiltonian integrable systems from one side and defects of regular
1044: periodic lattices from another side we have suggested 
1045: construction of ``elementary integer and fractional lattice defects'' 
1046: associated with 
1047: elementary integer and fractional monodromy.  Then we have proposed
1048: how to generate more complicated defects of lattices
1049: by combination of elementary ones. 
1050: Among these more complicated defects there are defects like disclinations
1051: which are well known in solid state physics. At the same time the author
1052: does not know simple examples of dynamical Hamiltonian system with 
1053: similar defects. Reciprocally, many examples of dynamical Hamiltonian systems
1054: with elementary monodromy are known but the ``elementary monodromy defect''
1055: seems not to be individually detected experimentally in periodic
1056: solids.  These observations enables us to formulate below a number of 
1057: problems concerning Hamiltonian dynamical systems with singularities
1058: and periodic lattices with defects. 
1059: Some of these problems have more or less intuitively
1060: evident answers but the strict mathematical proofs are still absent. 
1061: In other cases even the formulation of the problem
1062: is not precise and should be critically
1063: analyzed and corrected before looking for the answer.
1064: 
1065: \begin{itemize}
1066: \item{}  {\sl About the sign of elementary monodromy defect}. How to
1067: characterize the class of dynamical systems (classical and quantum) 
1068: possessing only elementary monodromy defects of one sign? For Hamiltonian
1069: systems focus-focus singularities correspond to elementary $(-)$ defects. 
1070: Tentative answer is to say that elementary $(+)$ monodromy defects
1071: are generic for $PT$-invariant dynamical systems with non-hermitian
1072: Hamiltonians and real spectra \cite{Bender,G24Zhil}.
1073: 
1074: \item{} {\sl Correspondence between topology of singular fibers of integrable
1075: toric fibrations and integer and fractional defects of lattices.}
1076: Some simple examples of such correspondence were given. Is it possible
1077: to establish more general correspondence?
1078: In particular it seems natural that elementary $(+)$ and $(-)$ monodromy defects
1079: correspond to pinched tori with different index of transversal self-crossings
1080: \cite{matsumoto}.
1081:  
1082: \item{} {\sl Constructive methods to design Hamiltonian classical and quantum
1083: systems with prescribed
1084: type of monodromy}. Less ambitious task is to propose a list of concrete
1085: examples of classical and quantum systems which show the manifestation of
1086: different elementary and non-elementary defects.
1087: 
1088: \item{}  {\sl Existence of a topological invariant separating different
1089: singularities (defects) with the same monodromy but with different
1090: numbers of $2\pi$-rotations of elementary cell.}  The analogy
1091: between this problem and the Riemann surfaces description \cite{Cartan}
1092: was pointed out to author on several ocassions.
1093: 
1094: \item{} {\sl Extension of the correspondence between singularities and defects
1095: from 2D-systems to higher dimensional systems.} This is surely a very wide 
1096: subject and author believes that first steps in 
1097: mathematical generalization should be guided by natural physical examples.
1098: 
1099: \item{} {\sl Global restrictions on the system of defects and on the
1100: system of
1101: singularities of toric fibrations in the case of compact base space (lattices on
1102: compact spaces).} For example, singular toric fibration over $S^2$ base space
1103: should have 24 elementary focus-focus singularities \cite{zungII} or equivalently
1104: 24 elementary $(-)$ monodromy defects or 12 $(\pi/3$)-rotational disclinations
1105: known as pentagonal defects \cite{Nelson83}. This problem has obvious relation
1106: with fullerene-like materials.
1107: 
1108: \item{} {\sl The relation between number of removed vertices for a defect
1109: and the Duistermaat-Heckman measure for the reduced  Hamiltonian system.}
1110: Hint:   The slope of the function giving the number of removed vertices from
1111: vertical line as a
1112: function of the number of a vertical line coincides with the Chern class of
1113: the integrable fibration used in the Duistermaat-Heckman theorem
1114: \cite{DuistHeck,Guill}.
1115: 
1116: \item{} {\sl To find macroscopic/mesoscopic physical systems which manifest
1117: presence of elementary $(+/-)$ monodromy defects.} Possible candidates
1118: besides periodic solids or liquid crystals may be membranes 
1119: \cite{Bowick01}, fullerenes and curved carbon surfaces \cite{ConeNature},
1120: viruses \cite{SPID}, colloidal structures \cite{Macromol}, %optical lattices, 
1121: etc.
1122: 
1123: \item{} {\sl Relation between internal structure of elementary cells
1124: and possible existence of isolated elementary integer and fractional
1125: monodromy defects in real physical systems.} In what kind of systems
1126: (materials) the topological properties are more important than geometric
1127: and steric effects and enable one to see the manifestation of 
1128: elementary monodromy defects?
1129: 
1130: \item{} {\sl Physical consequences of sign conjecture.} If one accept the
1131: formulated above sign conjecture, i.e. presence of only $(-)$ defects in
1132: generic families of Hamiltonian systems depending on a small number of 
1133: parameters, there is fundamental difference between $(+)$ and $(-)$ 
1134: (or in other terms between ``right'' and ``left'')
1135: in both classical and quantum mechanics.  How to formulate this conjecture in more
1136: precise terms and what kind of physical consequences can be rigorously
1137: deduced? 
1138: 
1139: \end{itemize}
1140: 
1141: The author hopes that  Hamiltonian dynamics and  periodic
1142: solids gives  complementary points of view which are useful for both 
1143: fields of scientific interest.
1144: The present article is supposed to stimulate mutual
1145: interest, better understanding and further  cooperation between specialists
1146: working in these fields.
1147: 
1148: {\bf Acknowledgements.}
1149: {\small The author thanks Drs. R. Cushman, N. Nekhoroshev, D. Sadovskii
1150:  for many stimulating discussions.
1151: This work was essentially done during authors stay in  
1152: IHES, Bures-sur-Yvette, France,
1153: Mathematical Institute, Universitty of Warwick, UK, and 
1154: Max-Planck Institut f\"ur Physik komplexer Systeme, Dresden, Germany.
1155: The support of these instotutions is acknowledged.
1156: This work is a part of the European project MASIE,
1157: contract HPRN-CT-2000-00113.}
1158: 
1159: \input{MLDref} %%%% references
1160: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%  }
1161: 
1162: 
1163: \printindex
1164: \end{document}
1165: