1: \NeedsTeXFormat{LaTeX2e}
2: \documentclass[12pt]{article}
3: \usepackage{amscd,amsmath,amssymb,amstext,amsthm,exscale,latexsym}
4: \usepackage{graphicx,floatflt}
5: \setlength{\textwidth}{27pc}
6: \setlength{\textheight}{43pc}
7: \textwidth160mm
8: \textheight 230mm
9: \topmargin -10mm
10: %\oddsidemargin 10mm
11: %\evensidemargin -10mm
12: \hoffset -20mm
13: %*********************************************************************
14: \newcommand {\al} {\alpha} \newcommand {\bt} {\beta}
15: \newcommand {\g } {\gamma} \newcommand {\G } {\Gamma}
16: \newcommand {\dl} {\delta} \newcommand {\e } {\epsilon}
17: \newcommand {\z } {\zeta} \newcommand {\et} {\eta}
18: \newcommand {\ve} {\varepsilon} \newcommand {\vt} {\vartheta}
19: \newcommand {\lm} {\lambda} \newcommand {\m } {\mu}
20: \newcommand {\n } {\nu} \newcommand {\x } {\xi}
21: \newcommand {\s } {\sigma} \newcommand {\vr } {\varrho}
22: \newcommand {\ta} {\tau} \newcommand {\ph} {\phi}
23: \newcommand {\vf } {\varphi} \newcommand {\h } {\chi}
24: \newcommand {\pf} {\psi} \newcommand {\om} {\omega}
25: \newcommand {\Lm} {\Lambda} \newcommand {\Om} {\Omega}
26: \newcommand {\Th} {\Theta} \newcommand {\Ph} {\Phi}
27: \newcommand {\pl} {\partial} \newcommand {\nb} {\nabla}
28: %---------------------------------------------------------------------
29: \newcommand {\sign}{{\sf\,sign\,}}
30: \newcommand {\const}{{\sf\,const}}
31: \newcommand {\diag}{{\sf\,diag\,}}
32: \newcommand {\arctanh}{{\sf\,arcth\,}}
33: \newcommand {\ex}{{\sf\,e}}
34: %---------------------------------------------------------------------
35: \newcommand {\MG} {{\mathbb G}}
36: \newcommand {\MM} {{\mathbb M}}
37: \newcommand {\MS} {{\mathbb S}}
38: \newcommand {\MO} {{\mathbb O}}
39: \newcommand {\MU} {{\mathbb U}}
40: \newcommand {\MR} {{\mathbb R}}
41: \newcommand {\CM } {{\cal M}} \newcommand {\CN} {{\cal N}}
42: \newcommand {\Se} {{\textsc{e}}} \newcommand {\Sh} {{\textsc{h}}}
43: \newcommand {\Sm} {{\textsc{m}}} \newcommand {\Ss} {{\textsc{s}}}
44: %---------------------------------------------------------------------
45: \begin{document}
46: \title {Inside the BTZ black hole}
47: \author {G. de Berredo-Peixoto
48: \thanks{E-mail: guilherme@fisica.ufjf.br}\\ \\
49: \sl Departamento de Fisica, Universidade Federal de Juiz de Fora,\\
50: \sl Juiz de Fora. CEP 36036--330, MG, Brazil\\ \\
51: M. O. Katanaev
52: \thanks{E-mail: katanaev@mi.ras.ru}\\ \\
53: \sl Steklov Mathematical Institute,\\
54: \sl Gubkin St. 8, Moscow, 119991, Russia}
55: \date {22 November 2006}
56: \maketitle
57: \begin{abstract}
58: We consider static circularly symmetric solution of three-dimensional
59: Einstein's equations with negative cosmological constant (the BTZ black
60: hole). The case of zero cosmological constant corresponding to the
61: interior region of a black hole is analyzed in detail. We prove that
62: the maximally extended BTZ solution with zero cosmological constant
63: coincides with flat three-dimensional Minkowskian space-time without
64: any singularity and horizons. The Euclidean version of this solution
65: is shown to have physical interpretation in the geometric theory of
66: defects in solids describing combined wedge and screw dislocations.
67: \end{abstract}
68: \vskip10mm
69: %*******************************************************************
70: \section{Introduction}
71: %********************************************************************
72: Three-dimensional static circularly symmetric solution of three-dimensional
73: Einstein's equations (the BTZ black hole \cite{BaTeZa92}) attracted much
74: interest last years, and is believed to be a good relatively simple
75: laboratory for analyzing general aspects of black hole physics. The
76: BTZ solution is the most general black hole solution in three dimensions,
77: which is guaranteed by the three-dimensional version of Birkhoff's
78: theorem \cite{AyMaZa04}. It has very interesting classical and quantum
79: properties (for review see, i.e. \cite{Carlip98}).
80:
81: Global structure of the BTZ solution was analyzed in \cite{BaHeTeZa93,Steif96}.
82: In these articles the black hole space-time was considered as the quotient
83: space of the anti-de Sitter space by the action of the discrete transformation
84: group. This approach deserves a deeper analysis because the transformation
85: group has fixed points, and therefore the quotient space is not a manifold.
86: The situation is unclear especially in the interior region of the black hole.
87: To clarify the global structure of the interior region, we consider the BTZ
88: solution for zero cosmological constant. In this simple case, all coordinate
89: transformations can be written explicitly, and behavior of geodesics is
90: analyzed. We prove that the maximally extended (along geodesics) BTZ solution
91: for zero cosmological constant coincides with flat Minkowskian space-time
92: without any singularity and horizons for infinite range of the angle coordinate
93: $\vf$. The BTZ solution in the original coordinates covers only one half of the
94: Minkowskian space-time, and the two planes corresponding to $r=0$ are just
95: coordinate singularities. Two copies of the BTZ solution cover the whole
96: Minkowskian space-time and are smoothly glued at $r=0$.
97:
98: If we make the angle identification $\vf\sim\vf+2\pi$ than four cones with
99: the same vertex appear in the interior region of the BTZ black hole located
100: at the inner horizon $r_-$.
101:
102: Next we analyze the Euclidean version of the BTZ solution. In this case
103: the solution has physical interpretation in solid state physics.
104: In the geometric theory of defects developed in
105: \cite{KatVol92,KatVol99,Katana03,Katana04} (for review see \cite{Katana05})
106: it depends on the Poisson ratio and describes combined wedge and screw
107: dislocations.
108: %*******************************************************************
109: \section{The BTZ solution}
110: %********************************************************************
111: We consider a three-dimensional manifold $\MM$ with local coordinates
112: $x^\mu$, $\mu=0,1,2$. Suppose it is endowed with a
113: Lorentzian signature metric $g_{\mu\nu}(x)$, $\sign g_{\mu\nu}=(+--)$,
114: which satisfies Einstein's equations
115: \begin{equation}
116: R_{\mu\nu}=\Lm g_{\mu\nu}
117: \end{equation}
118: with a cosmological constant $\Lm=-2/l^2$. In three dimensions the full curvature
119: tensor is defined by its Ricci tensor
120: \begin{equation*}
121: R_{\mu\nu\rho\s}=-\ve_{\mu\nu\lm}\ve_{\rho\s\z}R^{\lm\z},~~~~
122: R_{\mu\nu}=R_{\mu\rho\nu}{}^\rho
123: \end{equation*}
124: where $\e_{\mu\nu\lm}$ is the totally antisymmetric tensor,
125: $\e_{012}=\sqrt{\det g_{\mu\nu}}$. Therefore, any smooth solution of Einstein's
126: equations on a manifold (by solution we mean a pair $(\MM,g)$) is a space-time
127: of constant curvature. The universal covering spaces for positive, zero, and
128: negative cosmological constant are respectively de Sitter, Euclidean, and
129: anti de Sitter spaces. All other smooth solutions $(\MM,g)$ are obtained
130: from these solutions by an action of an isometry transformation group which
131: acts freely and properly discontinuous (see, i.e.\ \cite{Wolf72}). Afterwards,
132: a solution will be isometric to some fundamental domain of de Sitter,
133: Euclidean, or anti de Sitter space with properly identified boundaries.
134: These solutions are considered as known ones though explicit action of a
135: transformation group may be quite complicated. All these solutions are smooth,
136: have no singularities and horizons, and therefore do not describe black holes.
137:
138: Theory becomes much richer if we admit the existence of singularities at
139: points, lines, or surfaces in $\MM$. The famous BTZ solution is \cite{BaTeZa92}
140: \begin{equation} \label{ebtzso}
141: \begin{split}
142: ds^2&=\left(-M+\frac{J^2}{4r^2}+\frac{r^2}{l^2}\right)dt^2
143: -\frac{dr^2}{\left(-M+\frac{J^2}{4r^2}+\frac{r^2}{l^2}\right)}
144: -r^2\left(d\vf-\frac J{2r^2}dt\right)^2,
145: \\
146: &=\left(-M+\frac{r^2}{l^2}\right)dt^2
147: -\frac{dr^2}{\left(-M+\frac{J^2}{4r^2}+\frac{r^2}{l^2}\right)}
148: -r^2d\vf^2+Jdtd\vf,
149: \end{split}
150: \end{equation}
151: where $M$ and $J$ are two integration constants, having physical
152: interpretation of the mass and angular momentum of the black hole.
153: We shall see later that the inner horizon $r_-$ becomes a line in $\MM$
154: with four cones at each point.
155: This solution is supposed to be written in cylindrical coordinate system
156: \begin{equation} \label{eracod}
157: t\in(-\infty,\infty),~~r\in(0,\infty),~~\vf\in(0,2\pi).
158: \end{equation}
159:
160: Metric (\ref{ebtzso}) has two commuting Killing vector fields: $K_1=\pl_t$ and
161: $K_2=\pl_\vf$.
162:
163: Outer $r_+$ and inner $r_-$ horizons of the BTZ solution
164: \begin{equation*}
165: r^2_\pm=\frac{Ml^2}2\left(1\pm\sqrt{1-\frac{J^2}{M^2l^2}}\right)
166: \end{equation*}
167: are defined by two positive zeroes of the function
168: \begin{equation*}
169: N(r)=-M+\frac{J^2}{4r^2}+\frac{r^2}{l^2}=0.
170: \end{equation*}
171: Here we assume that $|J|<Ml$.
172:
173: We make four comments on the form of this solution.
174:
175: 1) The space-time $(M,g)$ with metric (\ref{ebtzso}), (\ref{eracod}) is not
176: locally a constant curvature space, because any point lying in the singularity
177: $r=r_-$ do not have a neighborhood of constant curvature. Here we consider
178: singular points $(t,r=r_-,\vf)$ belonging to $\MM$. In fact, we shall see that
179: points $r=r_-$ are not points of a manifold, and do not have neighborhoods
180: diffeomorphic to a ball at all.
181:
182: 2) The range of $\vf$ is determined by its interpretation as an angle in the
183: region $r\to\infty$ where the space-time becomes asymptotically anti-de Sitter.
184: The mass is supposed to be positive, because otherwise there is no horizon.
185: The case $M<0$ and $\Lm>0$, corresponding to de Sitter asymptotic (which has a
186: horizon) is not considered because $\vf$ can not be interpreted as the angle
187: coordinate in this case. The sign of $J$ corresponds to left and right rotations
188: and does not contribute to the global structure of the solution. Thus, range
189: of coordinates (\ref{eracod}) and the signs of the cosmological constant and
190: the mass are uniquely determined by two requirements: (i) $\vf$ is an angle
191: in the asymptotic region $r\to\infty$ and (ii) the solution has at least
192: one horizon.
193:
194: 3) For zero cosmological constant $\Lm=0$ and angular momentum $J=0$ the metric is
195: \begin{equation*}
196: ds^2=-Mdt^2+\frac{dr^2}M-r^2d\vf^2.
197: \end{equation*}
198: For positive mass $M>0$, it has no conical singularity at $r=0$ because now $r$
199: and $\vf$ are coordinates on the two-dimensional Minkowskian space-time of
200: signature $(+-)$ but not on the Euclidean plane. This is in contrast with
201: a common belief that static point particles in three-dimensional gravity are
202: described by conical singularities, distributed on a space-like section of $\MM$
203: \cite{Starus63,Clemen76,DeJatH84}.
204:
205: 4) There are four regions of $r$ where coordinate lines $t,r$, and $\vf$ have
206: different types of tangent vectors. Coordinate lines $t$ and $r$ may be either
207: timelike or spacelike, depending on range of $r$ which has three distinguished
208: points: $r_+,r_-$, and $r_3=Ml^2$. The coordinate line $\vf$ is always spacelike.
209: We summarize different properties of coordinates in the Table, where plus
210: and minus signs denote respectively timelike and spacelike character of
211: coordinates.
212: \begin{table}[h]%-----------------------------------------------------
213: \begin{center}
214: \begin{tabular}{|c|c|c|c|} \hline
215: & $t$ & $r$ & $\vf$ \\ \hline
216: $r_3<r<\infty$ & $+$ & $-$ & $-$ \\ \hline
217: $r_+<r<r_3$ & $-$ & $-$ & $-$ \\ \hline
218: $r_-<r<r_+$ & $-$ & $+$ & $-$ \\ \hline
219: $0<r<r_-$ & $-$ & $-$ & $-$ \\ \hline
220: \end{tabular}
221: \caption{\label{tdegeo} Timelike ``$+$'' and spacelike ``$-$'' character
222: of coordinate lines in different regions of $r$.}
223: \end{center}
224: \end{table}%----------------------------------------------------------
225: We see that the coordinate $t$ is timelike only for large $r>r_3$. In two
226: regions, $r_+<r<r_3$ and $0<r<r_-$, all coordinates are spacelike.
227:
228: Global structure of solution (\ref{ebtzso}) was described in \cite{BaHeTeZa93}
229: in terms of the quotient space of the anti-de Sitter space, and
230: the Carter--Penrose diagrams were drawn for the metric
231: \begin{equation} \label{etomse}
232: dl^2=Ndt^2-N^{-1}dr^2.
233: \end{equation}
234: This description deserves a deeper analysis especially at the
235: vicinity of the singularity for two reasons. First, the transformation
236: group used to obtain the quotient of the anti-de Sitter space in
237: \cite{BaHeTeZa93} does not act freely because it has fixed points
238: \cite{Steif96}. Second, the coordinate lines have different character,
239: and the induced metric on sections of $\MM$ corresponding to constant angle
240: $\vf=\const$ differs essentially from (\ref{etomse}). Therefore we can not
241: say that the global solution is topologically the product of the
242: Carter--Penrose diagram on a circle. In the next section we give a different
243: description of the global structure in terms of the product of real line
244: $t'\in\MR$ on the corresponding Carter--Penrose diagram.
245: %*******************************************************************
246: \section{The interior region of the BTZ black hole}
247: %********************************************************************
248: By global solution we mean a pair $(\MM,g)$ where (i) the metric satisfies
249: Einstein's equations and (ii) any extremal (geodesic) either can be
250: continued in both directions to an infinite value of the canonical parameter
251: or it ends up at a singular point at a finite value. This solution is also
252: called maximally extended. To construct a maximally extended solution for
253: BTZ metric (\ref{ebtzso}) we shall use a conformal block method described
254: in \cite{Katana00A} which was developed for two-dimensional metrics
255: having one Killing vector field.
256:
257: To analyze global structure of the interior region of the BTZ black hole,
258: we assume that cosmological constant is zero, $\Lm=0$ or $l\to\infty$.
259: This assumption simplifies essentially formulae, which now may be written
260: explicitly. Moreover, in this case the solution has direct interpretation
261: in the geometric theory of defects describing a linear dislocation which
262: is a combination of screw and wedge dislocations (see section \ref{sscwed}).
263:
264: For later comparison, we draw the Carter--Penrose diagram for two-dimensional
265: metric (\ref{etomse}) with
266: \begin{equation} \label{ecoffi}
267: N(r)=-\al^2+\frac{c^2}{r^2},
268: \end{equation}
269: where for simplicity we introduced new notations $M=\al^2$ and $c=J/2$.
270: In the geometric theory of defects, $\al=1+\theta$, where $\theta$ is the
271: deficit angle of the wedge dislocation, and $c=b/2\pi$, where $b$ is the
272: Burgers vector of the screw dislocation. The function (\ref{ecoffi})
273: becomes a conformal factor in coordinates $t,r'$ where new radial
274: coordinate $r'$ is determined by the ordinary differential equation
275: \begin{equation*}
276: \frac{dr}{dr'}=N(r).
277: \end{equation*}
278: In these coordinates, the global structure is easily analyzed \cite{Katana00A},
279: and the Carter--Penrose diagram for the surface with metric (\ref{etomse}),
280: (\ref{ecoffi}) is shown in Fig.~\ref{fcpwro}
281: \begin{figure}[h,b,t]%------------------------------------------------
282: \hfill\includegraphics[height=60mm]{fcpwro.eps}
283: \hfill {}
284: \\
285: \centering \caption{The Carter--Penrose diagram corresponding to a surface
286: with metric (\ref{etomse}), (\ref{ecoffi}). Solid lines denote complete
287: null infinity. The timelike boundary $r=0$ is incomplete, and the
288: two-dimensional curvature is singular here. Filled circles denote complete
289: space and time infinities. \label{fcpwro}}
290: \end{figure}%---------------------------------------------------------
291: It has one horizon at $r_-$. In the limit $l\to\infty$ the outer horizon
292: $r_+$ moves to infinity and the boundary $r=\infty$ becomes geodesically
293: complete which is shown by solid thick lines. At the boundaries $r=0$,
294: the two dimensional curvature for metric (\ref{etomse}), (\ref{ecoffi})
295: is singular. Filled circles denote complete future, past and right, left
296: ``infinities''. A circle in the center denotes incomplete vertex of
297: conformal blocks. Unfortunately, the induced metric on sections of $\MM$
298: corresponding to $\vf=\const$
299: \begin{equation*}
300: dl^2=-\al^2dt^2-\frac{dr^2}{-\al^2+\frac{c^2}{r^2}}
301: \end{equation*}
302: is quite different from (\ref{etomse}), (\ref{ecoffi}). This metric is
303: degenerate at the horizon $r_-=c/\al$, where it changes signature.
304: Therefore, we are not able to draw at least one Carter--Penrose diagram
305: for it.
306:
307: To avoid this difficulty, we draw the Carter--Penrose diagram for
308: sections of constant ``time'' which is a spacelike coordinate inside
309: the BTZ black hole. First, we diagonalize the metric on $\MM$
310: \begin{equation} \label{ebtzma}
311: ds^2=-\al^2dt^2-\frac{dr^2}{-\al^2+\frac{c^2}{r^2}}-r^2d\vf^2+2cd\vf dt.
312: \end{equation}
313: Performing the linear nondegenerate coordinate transformation
314: \begin{equation} \label{enewti}
315: t=t'+\frac c{\al^2}\vf,
316: \end{equation}
317: and keeping coordinates $r$ and $\vf$ untouched, we obtain
318: \begin{equation} \label{emetrs}
319: ds^2=-\al^2dt^{\prime2}-\frac{dr^2}{-\al^2+\frac{c^2}{r^2}}
320: -\left(r^2-\frac{c^2}{\al^2}\right)d\vf^2.
321: \end{equation}
322: Now the three-dimensional space-time can be represented as a product
323: $\MM=\MR\times\MU$ where $t'\in\MR$ and $(r,\vf)\in\MU$. The two-dimensional
324: surface $\MU$ (sections of $\MM$ corresponding to $t'=\const$) possesses the
325: induced metric of Lorentzian signature. At this point we assume that
326: $\vf\in\MR$ making the identification $\vf\sim\vf+2\pi$ afterwards. Changing
327: the radial coordinate,
328: \begin{equation} \label{echrad}
329: r=\sqrt{2\al\s},~~~~\s>0~~\text{for}~~\al>0,
330: \end{equation}
331: the induced metric on $\MU$ becomes
332: \begin{equation} \label{etwmes}
333: dl^2=-\frac{\al^2 d\s^2}{c^2-2\al^3\s}+\frac{c^2-2\al^3\s}{\al^2}d\vf^2.
334: \end{equation}
335: The conformal factor for the induced metric $N(\s)=(c^2-2\al^3\s)/\al^2$ is
336: a linear function of $\s$ and therefore describes a surface of zero
337: two-dimensional curvature. It has one horizon at $\s_-=c^2/2\al^3$ corresponding
338: to inner horizon $r_-$. The maximally extended surfaces $(\s,\vf)\in\MU$ is
339: represented by the Carter--Penrose diagram in Fig.~\ref{fcpfou}.
340: \begin{figure}[h,b,t]%------------------------------------------------
341: \hfill\includegraphics[height=60mm]{fcpfou.eps}
342: \hfill {}
343: \\
344: \centering \caption{The Carter--Penrose diagram corresponding to the
345: surface $\MU$ with metric (\ref{etwmes}). It represents two-dimensional
346: flat Minkowskian plane in $\s,\vf$ coordinates. The boundary is complete.
347: Two copies of the BTZ solution are smoothly glued together along timelike
348: dashed lines $\s=0$. \label{fcpfou}}
349: \end{figure}%---------------------------------------------------------
350: Here the maximal extension requires the infinite range of coordinates
351: $\s\in\MR$ and $\vf\in\MR$. To consolidate this range of coordinates
352: with the original radial coordinate (\ref{echrad}) we assume that
353: $\s>0$ for $\al>0$ and $\s<0$ for $\al<0$, and make the identification
354: $(\s,\al)\sim(-\s,-\al)$. We see that after the identification we must
355: consider two-dimensional surfaces $r=0$ for $\MM$ as the boundary.
356: %*******************************************************************
357: \subsection{Geodesics}
358: %********************************************************************
359: The behavior of geodesics for the BTZ solution was considered in
360: \cite{FaGaSe93,CrMaPe94}. In the case of zero cosmological constant
361: equations for geodesics are greatly simplified, and a general solution
362: can be written in elementary functions and analyzed in detail.
363:
364: To understand the nature of the singularity at $r_-$ we analyze the
365: behavior of geodesics
366: $x^\mu(\tau)=\left(\vphantom{\frac ab}t(\tau),r(\tau),\vf(\tau)\right)$
367: for metric (\ref{ebtzma}) in this section.
368: The only nonzero Christoffel's symbols are
369: \begin{equation*}
370: \begin{aligned}
371: \G_{11}{}^1&=-\frac{c^2}{r(\al^2r^2-c^2)}, &\qquad
372: \G_{12}{}^0=\G_{21}{}^0&=\frac{cr}{\al^2r^2-c^2},
373: \\[2mm]
374: \G_{22}{}^1&=\frac{\al^2r^2-c^2}r, & \qquad
375: \G_{12}{}^2=\G_{21}{}^2&=\frac{\al^2r}{\al^2r^2-c^2}.
376: \end{aligned}
377: \end{equation*}
378: The equations for geodesics
379: \begin{equation} \label{egebtz}
380: \ddot x^\mu=-\G_{\nu\rho}{}^\mu\dot x^\nu\dot x^\rho,
381: \end{equation}
382: where dots denote differentiation with respect to the canonical parameter
383: $\tau$, become
384: \begin{equation} \label{egebtn}
385: \begin{split}
386: \ddot t&=-2\frac{cr}{\al^2 r^2-c^2}\dot r\dot \vf,
387: \\
388: \ddot r&=\frac{c^2}{r(\al^2r^2-c^2)}\dot r^2-\frac{\al^2r^2-c^2}r\dot\vf^2,
389: \\
390: \ddot \vf&=-2\frac{\al^2r}{\al^2r^2-c^2}\dot r\dot\vf.
391: \end{split}
392: \end{equation}
393: These equations can be integrated in elementary functions. First, we
394: note that any system of equations for geodesics has a conserved quantity:
395: the length of a tangent vector $C_0=g_{\mu\nu}\dot x^\mu\dot x^\nu$.
396: It corresponds to conservation of energy of a point particle which, by
397: assumption, moves along a geodesic line
398: \begin{equation} \label{econen}
399: C_0=-\al^2\dot t^2+\frac{r^2}{\al^2 r^2-c^2}\dot r^2-r^2\dot\vf^2+2c\dot t\dot\vf.
400: \end{equation}
401: Another two conservations laws $C_{1,2}=g_{\mu\nu}\dot x^\mu K_{1,2}^\nu$
402: correspond to the symmetry of the metric, generated by two Killing vector
403: fields: $K_1=\pl_t$ and $K_2=\pl_\vf$. They describe respectively
404: conservation of momenta and angular momenta
405: \begin{align} \label{econfi}
406: C_1&=-\al^2\dot t+c\dot\vf,
407: \\ \label{econse}
408: C_2&=~~c\dot t-r^2\dot\vf.
409: \end{align}
410:
411: Conservation laws (\ref{econen})--(\ref{econse}) can be easily solved
412: with respect to the first derivatives
413: \begin{equation} \label{efilod}
414: \begin{split}
415: \dot t&=-\frac{r^2C_1+cC_2}{\al^2r^2-c^2},
416: \\
417: \dot r^2&=\frac{\al^2r^2-c^2}{r^2}C_0+\frac{r^2C_1^2+2cC_1C_2+\al^2C_2^2}{r^2},
418: \\
419: \dot\vf&=-\frac{cC_1+\al^2C_2}{\al^2r^2-c^2}.
420: \end{split}
421: \end{equation}
422: These equations are compared with their Euclidean counterpart in section
423: \ref{sgeoeu}. We show that the connected Lorentzian manifold breaks into
424: disconnected pieces along horizons in the Euclidean case.
425:
426: Equations for geodesics (\ref{efilod}) can be integrated explicitly.
427: To make clear further integration we perform the transformation to
428: Cartesian coordinates.
429: %*******************************************************************
430: \subsection{Transformation to Cartesian coordinates}
431: %********************************************************************
432: In this section we perform the coordinate transformation which brings
433: the metric (\ref{emetrs}) into the usual Lorentzian form and shows that
434: the Carter--Penrose diagram in Fig.\ref{fcpfou} represents the Minkowskian
435: plane. We consider three-dimensional Minkowskian space $\MR^{1,2}$ with
436: the Lorentz metric in Cartesian coordinates
437: \begin{equation} \label{eflthm}
438: ds^2=dT^2-dX^2-dY^2.
439: \end{equation}
440: We introduce polar coordinates $0<R<\infty, -\infty<\Phi<\infty$ in each
441: of the four quadrants in the $T,X$ plane (see Fig.\ref{fpolmi})
442: \begin{figure}[h,b,t]%------------------------------------------------
443: \hfill\includegraphics[height=60mm]{fpolmi.eps}
444: \hfill {}
445: \\
446: \centering \caption{Polar coordinates in the Minkowskian plane. Solid
447: lines are hyperbolas of constant $R$. Arrows show increasing of the
448: angle $\Phi$. Shadowed regions denote the fundamental domain of the
449: transformation group $\Phi\rightarrow\Phi+2\pi\al$. \label{fpolmi}}
450: \end{figure}%---------------------------------------------------------
451: \begin{equation*}
452: \begin{split}
453: \text{I}:~~&T=~~R\sinh\Phi,
454: \\
455: &X=~~R\cosh\Phi,
456: \\
457: \text{III}:~~&T=-R\sinh\Phi,
458: \\
459: &X=-R\cosh\Phi,
460: \end{split}
461: \qquad
462: \begin{split}
463: \text{II}:~~&T=~~R\cosh\Phi,
464: \\
465: &X=~~R\sinh\Phi,
466: \\
467: \text{IV}:~~&T=-R\cosh\Phi,
468: \\
469: &X=-R\sinh\Phi.
470: \end{split}
471: \end{equation*}
472: This transformation is degenerate on two lines $R=0$, and metric on the
473: $T,X$ plane becomes
474: \begin{equation} \label{emimep}
475: \begin{split}
476: \text{I,III}:\quad dl^2&=-dR^2+R^2d\Phi^2,
477: \\
478: \text{II,IV}:\quad dl^2&=dR^2-R^2d\Phi^2.
479: \end{split}
480: \end{equation}
481:
482: Performing the further transformation
483: \begin{equation*}
484: \begin{aligned}
485: &\text{I,III}:\quad & R&=\frac{\sqrt{c^2-2\al^3\s}}{\al^2}, &\quad \s&<\frac{c^2}{2\al^3}
486: \\[2mm]
487: &\text{II,IV}:\quad & R&=\frac{\sqrt{2\al^3\s-c^2}}{\al^2}, &\quad \s&>\frac{c^2}{2\al^3}
488: \\[2mm]
489: &\text{I--IV}:\quad& \Phi&=\al\vf, &\quad -\infty&<\vf<\infty,
490: \\
491: && Y&=\al t', &-\infty&<t'<\infty,
492: \end{aligned}
493: \end{equation*}
494: we arrive precisely to metric (\ref{emetrs}), (\ref{etwmes}).
495:
496: Now integrating equations (\ref{egebtz}) for geodesics becomes trivial.
497: A general solution is
498: \begin{equation*}
499: \begin{split}
500: T&=t_0+v_0\tau,
501: \\
502: X&=x_0+v_1\tau,
503: \\
504: Y&=y_0+v_2\tau,
505: \end{split}
506: \end{equation*}
507: and depends on six arbitrary constants: a position of a point in the
508: Minkowskian space $t_0,x_0,y_0$ and a velocity $v_0,v_1,v_2$ of the
509: geodesic which goes through this point. The inverse transformation, for
510: example, in the first quadrant is
511: \begin{equation*}
512: \begin{split}
513: t&=\frac1\al\left[Y+\frac c{\al^2}\arctanh\left(\frac TX\right)\right],
514: \\
515: r&=\frac1\al\sqrt{c^2-\al^4(X^2-T^2)},
516: \\
517: \vf&=\frac1\al\arctanh\left(\frac TX\right).
518: \end{split}
519: \end{equation*}
520: Elementary analysis shows that constants of integration
521: (\ref{econen})--(\ref{econse}) are
522: \begin{equation*}
523: \begin{split}
524: C_0&=v_0^2-v_1^2-v_2^2,
525: \\
526: C_1&=-\al v_2,
527: \\
528: C_2&=\frac{cv_2}\al+\al(x_0v_0-t_0v_1).
529: \end{split}
530: \end{equation*}
531:
532: The considered coordinate transformations prove that the Carter--Penrose
533: diagram in Fig.\ref{fcpfou} represents a Minkowskian plane $\MR^{1,1}$.
534: The global BTZ solution with zero cosmological constant is thus a
535: product $\MR\times\MR^{1,1}=\MR^{1,2}$, i.e.\ a flat three-dimensional
536: Minkowskian space-time.
537:
538: Now the maximally extended BTZ solution with zero cosmological constant
539: (\ref{ebtzma}) and infinite range of $\vf$ becomes transparent: it is a
540: flat three-dimensional Minkowskian space-time $\MR^{1,2}$ without any
541: singularity and horizons. BTZ solution (\ref{ebtzma}) covers only one
542: half of it. Maximal extension means that the BTZ solution for $0<\s<\infty$
543: is prolonged to negative values $-\infty<\s<\infty$. For negative $\s$ we
544: can define new coordinate $r=\sqrt{2\al\s}$, $\s<0$ and $\al<0$ and again
545: arrive at the BTZ solution. Thus, two copies of BTZ solution cover the
546: whole Minkowskian space-time $\MR^{1,2}$ and are smoothly glued together
547: at $r=0$. The ``singularity'' at $r=0$ is a purely coordinate one which
548: is clear from the transformation (\ref{echrad}). We stress that the above
549: analysis was performed for the whole range of the angle $\vf\in\MR$. The
550: periodicity of the angle will be considered in the next section.
551: %*******************************************************************
552: \subsection{Periodicity of $\vf$}
553: %********************************************************************
554: Having in mind that the coordinate $\vf$ is interpreted as the angle
555: $0<\vf<2\pi$ in the exterior region of the BTZ solution with negative
556: cosmological constant, we assume that the identification $\vf\sim\vf+2\pi$
557: takes also place for zero cosmological constant. The transformation
558: group $\MG:~\vf\rightarrow\vf+2\pi$ acts freely and properly discontinuous
559: on a line $\vf\in\MR$ but not in the Minkowskian space-time $\MR^{1,2}$.
560: In the last case it has fixed points and the quotient
561: space $\MR^{1,2}/\MG$ itself is not a manifold.
562:
563: Periodicity in $\vf$ means periodicity in the polar angle $\Phi\sim\Phi+2\pi\al$
564: in the Minkowskian plane $T,X$ considered in the previous section.
565: The transformation group identifies points along hyperbolas $R=\const$
566: corresponding to different polar angles $\Phi\sim\Phi+2\pi\al$. Its
567: action is not defined on two lines $R=0$ where polar coordinates are
568: degenerate. Considering these lines as the limit of hyperbolas, we
569: assume that all their points are mapped into the origin under the action
570: of the transformation group $\MG$. Then the transformation group acting
571: in the $T,X$ plane has the fundamental domain consisting of four wedges with
572: identified boundaries shown by shadowed regions in Fig.~\ref{fpolmi}. We
573: denote them by I-IV according to the number of the corresponding quadrant.
574: Identifying boundaries of the wedge makes it a cone. The vertexes of cones
575: are glued in the origin of the Cartesian coordinate system. The
576: origin is a fixed point for the transformation group and not a point of a
577: manifold. Indeed, it does not have a neighborhood diffeomorphic to a disc.
578: At the same time, it can not be excluded from the space-time because it lies
579: at a finite distance. Thus, we have four cones in the Minkowskian plane
580: with the common vertex in the origin. They are not conical singularities
581: because the plane is equipped with the Lorentzian signature metric.
582:
583: The Killing vector $K_2=\pl_\vf=\al\pl_\Ph$ of the whole three-dimensional
584: metric (\ref{ebtzma}) is also the Killing vector for two-dimensional metric
585: (\ref{emimep}). Hence, the transformation group $\MG$ is the isometry for
586: two-dimensional Minkowskian plane $T,X$.
587:
588: The identification $\vf\sim\vf+2\pi$ acts also nontrivially on the third
589: ``time'' coordinate $t'$ in (\ref{enewti}), but it remains well defined.
590:
591: Now we analyze the behavior of geodesics in the $T,X$ plane. They are
592: clearly straight lines before the identification. Let us fix an arbitrary
593: fundamental domain $(\Phi_0,\Phi_0+2\pi\al)$ and a timelike geodesic $1$,
594: shown in Fig.~\ref{fangex}. Particle moving along geodesic 1 in the $T,X$
595: plane wraps around a cone. Its trajectory is represented by the sequence
596: of segments of straight lines in the fundamental domain, each segment
597: representing one rotation around the cone. The equation of geodesic 1 in
598: polar coordinates in the first quadrant is
599: \begin{equation*}
600: R=R_0\frac{\tan\g_1\cosh\Phi_0-\sinh\Phi_0}{\tan\g_1\cosh\Phi-\sinh\Phi},
601: \end{equation*}
602: where $\g_1$ is the inclination of the line $1$. Its segment from $R_0$ to
603: \begin{equation*}
604: R_1=R_0\frac{\tan\g_1\cosh\Phi_0-\sinh\Phi_0}
605: {\tan\g_1\cosh(\Phi+2\pi\al)-\sinh(\Phi_0+2\pi\al)}
606: \end{equation*}
607: \begin{figure}[h,b,t]%------------------------------------------------
608: \hfill\includegraphics[height=60mm]{fangex.eps}
609: \hfill {}
610: \\
611: \centering \caption{The image of the geodesic $1$ in the fundamental
612: domain $\Phi_0,\Phi_0+2\pi\al$. It consists of segments of straight lines
613: with inclinations $\g_1,\g_2,\g_3,\dotsc$ which cross the line $\Phi_0$
614: respectively at points $R_0,R_1,R_2,\dotsc$.
615: \label{fangex}}
616: \end{figure}%---------------------------------------------------------
617: belongs to the fundamental domain. The next segment corresponding to
618: polar angle $\Phi\in(\Phi_0+2\pi\al,\Phi_0+4\pi\al)$ is identified with
619: the segment of line $2$ lying in the fundamental domain $(\Phi_0,\Phi_0+2\pi\al)$
620: but having different inclination $\g_2$. Continuing this procedure, we
621: obtain equation for the $k$-th line
622: \begin{equation*}
623: R=R_{k-1}\frac{\tan\g_k\cosh\Phi_0-\sinh\Phi_0}{\tan\g_k\cosh\Phi-\sinh\Phi},
624: \end{equation*}
625: where
626: \begin{equation} \label{eqangf}
627: R_k=R_0\frac{\tan\g_1\cosh\Phi_0-\sinh\Phi_0}
628: {\tan\g_1\cosh(\Phi_0+\al_k)-\sinh(\Phi_0+\al_k)},
629: \end{equation}
630: and we introduced shorthand notation
631: \begin{equation*}
632: \al_k=2\pi\al k,~~~~k=0,\pm1,\pm2,\dotsc.
633: \end{equation*}
634: On the other hand, the segment of line $k$ in the fundamental domain
635: can be obtained as the projection of the segment of the previous line
636: $k-1$
637: \begin{equation} \label{eqrkmo}
638: R_k=R_{k-1}\frac{\tan\g_k\cosh\Phi_0-\sinh\Phi_0}
639: {\tan\g_k\cosh(\Phi_0+\al_1)-\sinh(\Phi_0+\al_1)}.
640: \end{equation}
641: Comparing Eqs.\ (\ref{eqangf}) and (\ref{eqrkmo}), we express the
642: inclination angle $\g_k$ of the $k$-th line through the original
643: inclination angle $\g_1$
644: \begin{equation} \label{eincan}
645: \tan\g_k=\frac{\tan\g_1-\tanh\al_{k-1}}
646: {1-\tan\g_1\tanh\al_{k-1}}.
647: \end{equation}
648: It is important that the inclination angles of segments depend only on
649: the original inclination angle $\g_1$. They do not depend on the fundamental
650: domain characterized by the angle $\Phi_0$ and the distance from the
651: origin $R_0$.
652:
653: Equation (\ref{eincan}) is also valid for lightlike and spacelike geodesics.
654:
655: As the consequence of Eq.(\ref{eincan}), we see that segments of all
656: lightlike geodesics $\tan\g_1=\pm1$ remain lightlike $\tan\g_k=\pm1$.
657: All segments for timelike and spacelike geodesics are respectively
658: timelike and spacelike having the same limit
659: \begin{equation*}
660: \lim_{k\to \pm\infty}\tan\g_k=\mp1.
661: \end{equation*}
662: We see also that there are no closed timelike geodesics.
663:
664: To get a deeper insight into the behavior of geodesics we consider
665: $\al_k$ as a continuous variable
666: \begin{equation*}
667: \tan\g_\al=\frac{\tan\g_1-\tanh\al}
668: {1-\tan\g_1\tanh\al}.
669: \end{equation*}
670: This function is plotted in Fig.\ref{ftgalp}. For timelike geodesics
671: $|\tan\g_1|>1$ it is singular when
672: \begin{equation} \label{edesig}
673: \tanh\al\tan\g_1=1
674: \end{equation}
675: and has two branches.
676: \begin{figure}[h,b,t]%------------------------------------------------
677: \hfill\includegraphics[height=60mm]{ftgalp.eps}
678: \hfill {}
679: \\
680: \centering \caption{The inclination of the trajectory segment versus continuous
681: parameter $\al_k\rightarrow\al$. For timelike geodesics $|\tan\g_1|>1$, it has
682: two branches with the singularity at $\tanh\al\tan\g_1=1$, corresponding to the
683: turning point. The dependence is smooth for spacelike geodesics $|\tan\g_1|<1$.
684: \label{ftgalp}}
685: \end{figure}%---------------------------------------------------------
686: The upper and lower branches correspond to the motion of a particle respectively
687: to the right and left in the first quadrant, the singularity (\ref{edesig})
688: being the turning point.
689:
690: All timelike geodesics start and end at the origin of the Cartesian
691: coordinate system at finite values of proper time parameter making
692: infinite number of rotations around the cone. Spacelike geodesics which
693: do not lie entirely in one fundamental region approach the origin on
694: the left making infinite number of rotations at a finite value of canonical
695: parameter and go to the right space infinity.
696:
697: To understand the behavior of geodesics in other quadrants, it is sufficient
698: to rotate the picture on ninety degrees.
699:
700: The behavior of geodesics at the origin of Cartesian coordinates is
701: not defined. It can be naturally done by unfolding the cones and going
702: back to Minkowskian plane $T,X$. Then timelike geodesics which do not
703: cross boundary of the fundamental region IV are continued to the fundamental
704: region II. Those timelike geodesics which cross the boundary of the
705: fundamental region IV are continued to the fundamental domain I or III
706: depending on the inclination. In the region I they move to the right at
707: a finite distance, then return back to the origin, and are continued to
708: the fundamental domain II. We draw trajectory of a static particle in
709: the $T,X$ plane after the identification in Fig.\ref{fpares}.
710: \begin{figure}[h,b,t]%------------------------------------------------
711: \hfill\includegraphics[height=60mm]{fpares.eps}
712: \hfill {}
713: \\
714: \centering \caption{The trajectory of a static particle in the $T,X$ plane.
715: The fundamental domain is chosen symmetrically $(-\pi,\pi)$ in each quadrant.
716: After the identification $\vf\sim\vf+2\pi$ it makes infinite number of
717: rotations around the cone in the fourth quadrant at a finite proper time.
718: Afterwards it moves to the right in the first quadrant, returns back,
719: making also an infinite number of rotations near the cone vertex, and
720: goes to the second quadrant. There, after an infinite number of rotations,
721: it goes to infinity.
722: \label{fpares}}
723: \end{figure}%---------------------------------------------------------
724:
725: Timelike curves $R=\const$ (which are not geodesics) are closed loops
726: after the identification $\Phi\sim\Phi+2\pi\alpha$. Thus, any particle
727: with constant acceleration (in terms of Cartesian coordinates) has a
728: closed timelike world line at regions I,III.
729:
730: To conclude this section, we draw the Carter--Penrose diagram for the
731: Minkowskian plane after the polar angle identification $\Phi\sim\Phi+2\pi\al$
732: in Fig.\ref{flower}. It looks like a flower.
733: \begin{figure}[h,b,t]%------------------------------------------------
734: \hfill\includegraphics[height=60mm]{flower.eps}
735: \hfill {}
736: \\
737: \centering \caption{The Carter--Penrose diagram for the Minkowskian plane
738: after the polar angle identification $\Phi\sim\Phi+2\pi\al$.
739: \label{flower}}
740: \end{figure}%---------------------------------------------------------
741: %*******************************************************************
742: \section{Euclidean version of the interior of the BTZ solution}
743: %********************************************************************
744: The Euclidean version of the BTZ solution was considered in \cite{CarTei95}
745: when analyzing its thermodynamical properties. In this section we
746: analyze global structure and geodesics of the Euclidean BTZ solution
747: for zero cosmological constant.
748:
749: Here, we use many notations from the previous section though their meaning in the
750: Euclidean case is often different. We hope that this step does not cause
751: any inconvenience and simplifies the comparison of the two cases.
752:
753: The Euclidean counterpart of the BTZ metric (\ref{ebtzso}) is given by
754: the transformation $r\rightarrow ir$, $l\rightarrow il$, and $J\rightarrow-J$.
755: Then the BTZ metric becomes
756: \begin{equation} \label{eclbtz}
757: ds^2=\left(-\al^2-\frac{c^2}{r^2}+\frac{r^2}{l^2}\right)dt^2
758: +\frac{dr^2}{-\al^2-\frac{c^2}{r^2}+\frac{r^2}{l^2}}
759: +r^2\left(d\vf-\frac c{r^2}dt\right)^2.
760: \end{equation}
761: Let $r_\pm$ denote two positive roots
762: \begin{equation*}
763: r_\pm^2=\frac{\al^2l^2}2\left(\sqrt{1+\frac{4c^2}{\al^4l^2}}\pm1\right).
764: \end{equation*}
765: Then metric (\ref{eclbtz}) is degenerate at $r=r_+$, has Euclidean signature
766: $(+++)$ for $r_+<r<\infty$, and Lorentzian signature $(+--)$ for $0<r<r_+$.
767: In fact, this metric describes two disjoint spaces: one for $r_+<r<\infty$
768: with the Euclidean signature metric, and the other for $0<r<r_+$ with
769: the Lorentzian metric. This phenomena was demonstrated in the two-dimensional
770: case where connected Lorentzian surface breaks into disconnected
771: Euclidean pieces along horizons when going to the Euclidean signature
772: metric \cite{KaKlKu99}, horizons giving rise to possible conical
773: singularities. The absence of conical singularities is precisely the
774: definition of the Hawking temperature. The same phenomenon occurs also
775: for zero cosmological constant (see Sec.~\ref{sgeoeu}).
776:
777: In the limit $l\rightarrow\infty$, metric (\ref{eclbtz}) reduces to the
778: metric having Lorentzian signature everywhere and cannot be considered as
779: the Euclidean version of (\ref{ebtzma}). The Euclidean counterpart of the
780: BTZ metric for zero cosmological constant (\ref{ebtzma}) is given by the
781: transformation $t\rightarrow iz$, $\vf\rightarrow i\vf$:
782: \begin{equation} \label{eclinm}
783: ds^2=\al^2 dz^2+\frac{dr^2}{\al^2-\frac{c^2}{r^2}}+r^2d\vf^2-2cd\vf dz.
784: \end{equation}
785: The range of coordinates is taken as
786: \begin{equation} \label{eraecb}
787: r\in(0,\infty),~~~~\vf\in(0,2\pi),~~~~z\in(-\infty,\infty).
788: \end{equation}
789: This metric is degenerate at the horizon $r_-=c/\al$, has Euclidean
790: signature $(+++)$ for outer region $r_-<r<\infty$, and Lorentzian signature
791: $(+--)$ for inner region $0<r<r_-$.
792:
793: So, our starting point is metric (\ref{eclinm}), (\ref{eraecb}) which is
794: the Euclidean version of (\ref{ebtzma}). The problem is to find the
795: space described by this metric. We show below that it describes two
796: disjoint maximally extended manifolds, each being topologically equivalent
797: to the Euclidean space $\MR^3$ with the wedge cut out or added to it.
798: %*******************************************************************
799: \subsection{Transformation to Cartesian coordinates}
800: %********************************************************************
801: First, we consider the region $c/\al<r<\infty$, where metric (\ref{eclinm})
802: has Euclidean signature. This domain can be easily shown to be diffeomorphic
803: to the whole Euclidean space $\MR^3$ with the wedge of angle $2\pi\theta$,
804: where $\theta=\al-1$, cut out or added. For negative and positive $\theta$
805: the wedge is respectively cut out or added to $\MR^3$.
806:
807: Let $X,Y,Z$ be Cartesian coordinates in $\MR^3$
808: \begin{equation} \label{euccap}
809: ds^2=dX^2+dY^2+dZ^2=dR^2+R^2d\Phi^2+dZ^2,
810: \end{equation}
811: where $R,\Phi$ are polar coordinates in the $X,Y$ plane. Then
812: the coordinate transformation:
813: \begin{equation} \label{ectreu}
814: \begin{aligned}
815: R&=\frac r\al\sqrt{1-\frac{c^2}{\al^2r^2}},&~~~~\frac c\al&<r<\infty,
816: \\
817: \Phi&=\al\vf, & 0&<\Phi<2\pi\al,
818: \\
819: Z&=\al z-\frac c\al \vf, & -\infty&<z<\infty,
820: \end{aligned}
821: \end{equation}
822: brings metric to the form (\ref{eclinm}). The wedge is cut out or added to the
823: Euclidean space because the angle $\Phi$ ranges from $0$ to $2\pi\al$.
824: This is determined by the original range of coordinates (\ref{eraecb}).
825: The sides of the wedge are identified (glued together). The axis $R=0$ is
826: mapped into the horizon $r_-$ which is now the surface of the cylinder.
827:
828: The inverse transformation to (\ref{ectreu}) is
829: \begin{equation*}
830: \begin{split}
831: z&=\frac1\al\left[Z+\frac c{\al^2}\arctan\left(\frac YX\right)\right],
832: \\
833: r&=\frac1\al\sqrt{c^2+\al^4(X^2+Y^2)},
834: \\
835: \vf&=\frac1\al\arctan\left(\frac YX\right).
836: \end{split}
837: \end{equation*}
838:
839: The inner region $0<r<r_-$ is diffeomorphic to the Minkowskian space-time
840: $\MR^{1,2}$ with the metric
841: \begin{equation} \label{elobtm}
842: ds^2=-dX^2-dY^2+dZ^2=-R^2d\Phi^2-R^2d\Phi^2+dZ^2.
843: \end{equation}
844: The coordinate transformation
845: \begin{equation} \label{ectren}
846: \begin{aligned}
847: R&=\frac r\al\sqrt{\frac{c^2}{\al^2r^2}-1},&~~~~0&<r<\frac c\al,
848: \\
849: \Phi&=\al\vf, & 0&<\Phi<2\pi\al,
850: \\
851: Z&=\al z-\frac c\al \vf, & -\infty&<z<\infty.
852: \end{aligned}
853: \end{equation}
854: transforms metric (\ref{elobtm}) also in the form (\ref{eclinm}). The range
855: of the angle $\Phi$ differs from $2\pi$. Therefore, the wedge is cut out
856: or added to the Minkowskian space-time as in the previous case. The axis
857: $R=0$ and infinity $R=\infty$ are respectively mapped into the horizon
858: $r_-$ and the axis $r=0$.
859: %*******************************************************************
860: \subsection{Geodesics \label{sgeoeu}}
861: %********************************************************************
862: To answer the question what is described by metric (\ref{eclinm}),
863: (\ref{eraecb}), we must analyze the behavior of geodesics. Equations
864: for geodesics (\ref{egebtn}) are
865: \begin{equation} \label{egebez}
866: \begin{split}
867: \ddot z&=-2\frac{cr}{\al^2 r^2-c^2}\dot r\dot \vf,
868: \\
869: \ddot r&=\frac{c^2}{r(\al^2r^2-c^2)}\dot r^2+\frac{\al^2r^2-c^2}r\dot\vf^2,
870: \\
871: \ddot \vf&=-2\frac{\al^2r}{\al^2r^2-c^2}\dot r\dot\vf.
872: \end{split}
873: \end{equation}
874: The only difference from Eqs.(\ref{egebtz}) is the substitution $t\rightarrow z$
875: and the sign before the second term in the second equation.
876:
877: There are three conservation laws:
878: \begin{equation} \label{egeucb}
879: \begin{split}
880: C_0&=\al^2\dot z^2+\frac{r^2}{\al^2r^2-c^2}\dot r^2+r^2\dot\vf^2
881: -2c\dot\vf\dot z,
882: \\
883: C_1&=~\al^2\dot z-c\dot\vf,
884: \\
885: C_2&=-c\dot z+r^2\dot\vf.
886: \end{split}
887: \end{equation}
888: The last two conservation laws correspond to two Killing vectors $K_1=\pl_z$
889: and $K_2=\pl_\vf$.
890:
891: A general solution of Eqs.(\ref{egebez})
892: \begin{equation*}
893: \begin{split}
894: X&=x_0+v_1\tau,
895: \\
896: Y&=y_0+v_2\tau,
897: \\
898: Z&=z_0+v_3\tau,
899: \end{split}
900: \end{equation*}
901: depends on six arbitrary constants $(x_0,y_0,z_0)$ and $(v_1,v_2,v_3)$
902: parameterizing the point and the direction of a geodesic at this point.
903: Elementary analysis shows that
904: \begin{equation*}
905: \begin{split}
906: C_0&=v_1^2+v_2^2+v_3^2,
907: \\
908: C_1&=\al v_3,
909: \\
910: C_2&=-\frac c\al v_3+\al(x_0v_2-y_0v_1).
911: \end{split}
912: \end{equation*}
913:
914: To make the difference between Lorentzian and Euclidean cases clear, we
915: solve Eqs. (\ref{egeucb}) with respect to the first derivatives
916: \begin{equation} \label{efieod}
917: \begin{split}
918: \dot z&=\frac{r^2C_1+cC_2}{\al^2r^2-c^2},
919: \\
920: \dot r^2&=\frac{\al^2r^2-c^2}{r^2}C_0-\frac{r^2C_1^2+2cC_1C_2+\al^2C_2^2}{r^2},
921: \\
922: \dot\vf&=\frac{cC_1+\al^2C_2}{\al^2r^2-c^2}.
923: \end{split}
924: \end{equation}
925: The essential difference from the Lorentzian case (\ref{efilod}) is the sign
926: before the second term in the expression for $\dot r^2$. At the horizon
927: $r_-=c/\al$ we have
928: \begin{equation*}
929: \dot r^2=-\frac{(cC_1+\al^2C_2)^2}{c^2}
930: \end{equation*}
931: We see that horizon is reached only by geodesics with $cC_1+\al^2C_2=0$
932: because the right hand side of this equation must be positive. For these
933: geodesics
934: \begin{equation*}
935: \dot\vf=0,~~~~\dot z=-\frac{C_2}c.
936: \end{equation*}
937: This means that in each plane $z=\const$ only radial geodesics reach the
938: surface of the cylinder $r=r_-$. Therefore, there are not enough geodesics
939: to consider the boundary of the cylinder as a two-dimensional surface.
940: Moreover, the circumference of the cylinder measured with metric
941: (\ref{eclinm}) is zero. We conclude that the boundary of the cylinder
942: $r=r_-$ is, in fact, a line. For the Lorentzian signature metric, there is
943: no restriction on geodesics which reach the horizon due to the difference
944: in signs.
945:
946: Previous analysis indicates that the connected Lorentzian manifold breaks
947: into disconnected pieces along horizons for the Euclidean case. To prove this,
948: we consider another coordinate transformation $R,\Phi,Z\rightarrow f,\psi,\z$:
949: \begin{equation} \label{ecotrj}
950: \begin{split}
951: R&=\frac f\al,
952: \\
953: \Phi&=\al\psi,
954: \\
955: Z&=\z-c\psi.
956: \end{split}
957: \end{equation}
958: In these coordinates the flat metric (\ref{euccap}) becomes
959: \begin{equation} \label{ejteme}
960: ds^2=\frac 1{\al^2}df^2+(f^2+c^2)d\psi^2+d\z^2-2cd\z d\psi,
961: \end{equation}
962: but now the angle range is $\psi\in(0,2\pi)$, and coordinates $f,\psi,\z$
963: cover the whole $\MR^3$ and nothing else. (We use the unusual notation
964: $f$ for the radial coordinate because in the next section it is
965: transformed $f=f(\rho)$.) This metric describes
966: the space which is topologically $\MR^3$ with the ``shifted'' conical
967: singularities along the $\z$ axis. In the case $c=0$, we have ordinary
968: conical singularity in each section $\z=\const$. This space is
969: geodesically complete at infinity $f\rightarrow\infty$ because all
970: geodesics in the original $X,Y,Z$ coordinate are complete. Thus, the
971: exterior region $r_-<r<\infty$ of the Euclidean BTZ metric for zero
972: cosmological constant describes Euclidean space with ``shifted''
973: conical singularities at the $\z$ axis. It is the maximally extended
974: manifold.
975:
976: Metric (\ref{ejteme}) is the Euclidean version of the metric considered
977: in \cite{DeJatH84} and has straightforward interpretation in solid state
978: physics considered in the next section.
979:
980: Similar analysis can be performed for the interior region, and we do not
981: repeat it here. In fact, it is obvious from coordinate transformation
982: (\ref{ectren}) that the axis $z$ for metric (\ref{eclinm}) represents
983: infinity and the surface of the cylinder is the ``shifted'' conical
984: singularity at the $Z$ axis of three-dimensional Minkowskian space-time
985: (\ref{elobtm}). This manifold is also maximally extended.
986:
987: Therefore metric (\ref{eclinm}), (\ref{eraecb}) describes two disjoint
988: maximally extended manifolds: the Euclidean space $\MR^3$ and the
989: Minkowskian space-time $\MR^{1,2}$ with ``shifted'' conical
990: singularities at the $Z$ axis.
991: %*******************************************************************
992: \section{Solid state physics interpretation \label{sscwed}}
993: %********************************************************************
994: In the geometric theory of defects (for review see \cite{Katana05})
995: pure elastic deformations describe diffeomorphisms of the flat Euclidean
996: space $\MR^3$. The presence of defects: dislocations (defects in elastic media)
997: and disclinations (defects in the spin structure), gives rise to nontrivial
998: Riemann--Cartan geometry in $\MR^3$, the curvature and torsion tensor being
999: interpreted as the surface density of Burgers and Frank vectors, respectively.
1000: In the absence of disclinations, curvature is equal to zero, and we have the
1001: space of absolute parallelism characterized only by nontrivial torsion.
1002: In this case, the $\MS\MO(3)$-connection is a pure gauge defined by the
1003: gauge conditions, and torsion is given by the triad field $e_\mu{}^i$ which
1004: satisfies Euclidean Einstein's equations with nontrivial sources
1005: (energy-momentum tensor). The corresponding Einstein's equations are
1006: written for the induced metric
1007: \begin{equation*}
1008: g_{\mu\nu}=e_\mu{}^ie_\nu{}^j\dl_{ij},~~~~\dl_{ij}=\diag(+++),
1009: \end{equation*}
1010: while the curvature tensor for the $\MS\MO(3)$-connection remains identically
1011: equal to zero.
1012: For continuous distribution of dislocations, the sources are described by
1013: continuous functions, and for single defects, we have $\dl$-function type
1014: sources in the right hand side of Euclidean Einstein's equations.
1015:
1016: Metric (\ref{ejteme}) satisfies free Euclidean Einstein's equations
1017: everywhere except the $\z$ axis, where we have a ``shifted'' conical
1018: singularity. We do not analyze the corresponding source here but
1019: simply give physical interpretation of this solution.
1020:
1021: The coordinate transformation from the flat Euclidean space (\ref{ecotrj})
1022: has the following interpretation in solid state physics. At the beginning,
1023: we have undeformed elastic media which is the flat Euclidean space
1024: $\MR^3$. For negative deficit angle $2\pi\theta$, the wedge with
1025: the edge along the $Z$ axis and the angle $2\pi|\theta|$ is cut out from
1026: the media (see Fig.). For positive deficit angle, the wedge of the
1027: same media is added.
1028: \begin{figure}[h,b,t]%------------------------------------------------
1029: \hfill\includegraphics[height=60mm]{fwescd.eps}
1030: \hfill {}
1031: \\
1032: \centering \caption{The combined wedge and screw dislocation. The picture
1033: is drawn for negative deficit angle $\theta$ and the Burgers vector $b$
1034: which is antiparallel to the $Z$ axis. \label{fwescd}}
1035: \end{figure}%---------------------------------------------------------
1036: Then the lower side of the cut is moved along the $Z$ axis in the opposite
1037: direction on the distance $b=2\pi c$, where $b$ is the Burgers vector, and
1038: both sides of the cut are glued together.
1039: The cut out or added wedge of media corresponds to the wedge dislocation,
1040: and the displacement of the lower side of the cut along the $Z$ axis
1041: describes the screw dislocation. Both types of dislocations are experimentally
1042: observed defects in real crystals (see i.e.\ \cite{Amelin64}). Thus,
1043: metric (\ref{ejteme}) qualitatively describes nontrivial geometry around
1044: combined wedge and screw dislocations. For this metric, we can construct
1045: triad field and the corresponding torsion tensor which is equal to the
1046: surface density of the Burgers vector.
1047:
1048: We used the word ``qualitatively'' because a solution of Einstein's
1049: equations can be written in an arbitrary coordinate system and does not
1050: depend on Lame coefficients which characterize the elastic properties
1051: of media. The coordinate transformation (\ref{ecotrj}) describes
1052: only cut and paste process of defect creation. In reality, the media
1053: comes to the equilibrium state after gluing both sides of the cut, and
1054: this process is driven by the elasticity theory equations. Therefore,
1055: the induced metric must depend on Lame coefficients. To consolidate
1056: gravity and elasticity theories we proposed to use the elastic gauge
1057: \cite{Katana03}. It is given by the following construction. First,
1058: we fix the cylindrical coordinates $\rho,\psi,\z$ in the Euclidean
1059: space related to the considered problem. The flat metric and the
1060: triad field are marked with a circle over a symbol
1061: \begin{equation*}
1062: \begin{split}
1063: ds^2&=\overset\circ g{}_{\mu\nu}dx^\mu dx^\nu=d\rho^2+\rho^2d\psi^2+d\z^2,
1064: \\
1065: \overset\circ e{}_\mu{}^i&=
1066: \begin{pmatrix} 1 & 0 & 0 \\ 0 & \rho & 0 \\ 0 & 0 & 1 \end{pmatrix}
1067: \end{split}
1068: \end{equation*}
1069: Then the elastic gauge is chosen to be
1070: \begin{equation} \label{elgbtz}
1071: \overset\circ g{}^{\mu\nu}\overset{\circ}{\nb}_\mu e_{\nu i}
1072: +\frac\s{1-2\s}\overset\circ e{}^\mu{}_i\overset\circ\nb_\mu e^T=0,
1073: \end{equation}
1074: where a circle over the covariant derivative $\overset\circ \nb{}_\mu$
1075: means that it is constructed for flat metric $\overset\circ g{}_{\mu\nu}$;
1076: $e^T=\overset\circ e{}^\mu{}_ie_\mu{}^i$, and $\s=\const$ is the Poisson
1077: ratio defined by the Lame coefficients \cite{LanLif70}.
1078:
1079: Physical meaning of this gauge condition is based on the linear
1080: approximation. Suppose defects are absent, and we have only elastic
1081: deformations described by the displacement vector field $u^i(x)$ which
1082: parameterizes diffeomorphisms of $\MR^3$. For simplicity, we assume
1083: that it is given in the Cartesian coordinate system. Than, for small
1084: relative displacements $|\pl_\mu u^i|\ll 1$, the induced triad field
1085: can be chosen in the form (there is a freedom in local $\MS\MO(3)$ rotations):
1086: \begin{equation} \label{elinar}
1087: e_{\mu i}\approx\dl_{\mu i}-\frac12(\pl_\mu u_i+\pl_i u_\mu).
1088: \end{equation}
1089: In the linear approximation in Cartesian coordinates, Latin and Greek indices
1090: may be identified. Then the elastic gauge condition (\ref{elgbtz}) reduces
1091: to the usual linear elasticity theory equations for the displacement vector
1092: field \cite{LanLif70}
1093: \begin{equation} \label{eqsepu}
1094: (1-2\s)\triangle u_i+\pl_i\pl_j u^j=0,
1095: \end{equation}
1096: and the reduction of Eq.~(\ref{elgbtz}) to Eq.~(\ref{eqsepu}) is quite
1097: obvious. Introduction of flat covariant derivatives in the elastic gauge
1098: (\ref{elgbtz}) allows us to use arbitrary coordinate systems.
1099:
1100: The triad field corresponding to metric (\ref{ejteme}) is
1101: \begin{equation} \label{etrwsc}
1102: e_\mu{}^i=\begin{pmatrix}
1103: \dfrac{\raise-.7ex\hbox{$1$}}{\raise .7ex\hbox{$\al$}} & 0 & 0 \\
1104: 0 & f & -c \\ 0 & 0 & 1 \end{pmatrix}.
1105: \end{equation}
1106: The $f,\psi$ part of the triad corresponds to the symmetrized linear
1107: approximation (\ref{elinar}), and the constant unsymmetrical $\psi$
1108: part does not alter the form of the elastic gauge condition. We change
1109: the radial coordinate $f=f(\rho)$ to rewrite the triad in the elastic
1110: gauge. Then the elastic gauge condition (\ref{elgbtz}) reduces to
1111: the Euler ordinary differential equation
1112: \begin{equation} \label{eulgae}
1113: \frac {f''}\al\left(1+\frac \s{1-2\s}\right)
1114: +\frac{f'}\rho\left(\frac1\al+\frac\s{1-2\s}\right)
1115: -\frac f{\rho^2}\left(1+\frac\s{1-2\s}\right)=0.
1116: \end{equation}
1117: This is the same equation which arises for pure wedge dislocation \cite{Katana03}
1118: (the interested reader can find more details there). To uniquely fix the
1119: solution of this equation, we suppose that the media fills only the cylinder
1120: of finite radius $R_0$ (this is needed to avoid divergences) and impose two
1121: additional gauge conditions
1122: \begin{equation*}
1123: e_\rho{}^\rho|_{\rho=R_0}=1,~~~~e_\psi{}^\psi|_{\rho=0}=0.
1124: \end{equation*}
1125: The first gauge condition means the absence of external forces on the
1126: surface of the cylinder, and the second one corresponds to the absence
1127: of the angular component of the deformation tensor at the core of dislocation.
1128: Afterwards, the solution of Eq.(\ref{eulgae}) is uniquely defined
1129: \begin{equation*}
1130: f=\frac\al{\g R_0^{\g-1}}\rho^\g,
1131: \end{equation*}
1132: where
1133: \begin{equation*}
1134: \g=-\theta B+\sqrt{\theta^2B^2+1+\theta},~~~~B=\frac\s{2(1-\s)}.
1135: \end{equation*}
1136: Thus, the Euclidean version of the BTZ solution for zero cosmological
1137: constant (\ref{ejteme}) in the elastic gauge (\ref{elgbtz}) becomes
1138: \begin{equation} \label{eubtze}
1139: ds^2=\left(\frac\rho {R_0}\right)^{2(\g-1)}d\rho^2
1140: +\left(\frac{\al^2}{\g^2}\left(\frac\rho{R_0}\right)^{2(\g-1)}\rho^2
1141: +c^2\right)d\psi^2+d\z^2-2cd\z d\psi.
1142: \end{equation}
1143: This is the exact solution of the Euclidean Einstein equations written in the
1144: elastic gauge. It describes the induced metric around combined wedge and screw
1145: dislocations and nontrivially depends on the Poisson ratio characterizing the
1146: elastic properties of media. Below, this solution is compared with the result
1147: obtained entirely within the elasticity theory without referring to Einstein
1148: equations.
1149:
1150: The wedge and screw dislocations are relatively simple linear defects in
1151: solids, and the corresponding displacement vector field can be explicitly
1152: found as the solution to the linear elasticity field equations (\ref{eqsepu}).
1153: The results are well known, and we skip their derivation. The displacement
1154: vector field for the wedge dislocation in cylindrical coordinates
1155: $R,\Phi,Z$ is \cite{Kosevi81}
1156: \begin{equation} \label{eweufi}
1157: u_R^{(\text{wedge})}=-\theta\frac{1-2\s}{2(1-\s)}\rho\ln\frac\rho{\ex R_0},~~~~
1158: u_\Psi^{(\text{wedge})}=-\theta\rho\phi,~~~~
1159: u_Z^{(\text{wedge})}=0,
1160: \end{equation}
1161: where $\ex$ is the base of the natural logarithm. The displacement vector
1162: field for the screw dislocation is \cite{LanLif70}
1163: \begin{equation} \label{esc}
1164: u_R^{(\text{screw})}=u_\Phi^{(\text{screw})}=0,~~~~
1165: u_Z^{(\text{screw})}=c\phi=\frac b{2\pi}\phi,
1166: \end{equation}
1167: where $b$ is the Burgers vector. The displacements vectors can be added
1168: in the linear elasticity theory, and the total displacement vector for
1169: combined wedge and screw dislocations becomes
1170: \begin{equation*}
1171: u=u^{(\text{wedge})}+u^{(\text{screw})}.
1172: \end{equation*}
1173: In our notations, the coordinate transformation corresponding to this
1174: displacement vector field is
1175: \begin{equation*}
1176: R=\rho-u_R,~~~~\Phi=\phi-\frac1\rho u_\Phi,~~~~Z=\z-u_Z.
1177: \end{equation*}
1178: The induced metric within the elasticity theory becomes
1179: \begin{equation} \label{elmets}
1180: \begin{split}
1181: ds^2_{\text{(elastic)}}&=dR^2+R^2d\Phi^2+dZ^2=
1182: \\
1183: &=\left(1+\theta\frac{1-2\s}{1-\s}\ln\frac\rho{R_0}\right)d\rho^2
1184: +\left[\rho^2\left(1+\theta\frac{1-2\s}{1-\s}\ln\frac\rho{R_0}+\theta\frac1{1-\s}
1185: \right)+c^2\right]d\phi^2
1186: \\
1187: &+d\z^2-2cd\z d\phi.
1188: \end{split}
1189: \end{equation}
1190: The linear elasticity theory equations are valid for small relative displacements
1191: $\pl_\mu u^i\ll1$. Therefore, the induced metric in the elasticity theory
1192: is expected to give correct answer for small deficit angle $\theta\ll1$,
1193: small Burgers vector $b/R_0\ll1$, and near the surface of the cylinder
1194: $\rho\sim R_0$.
1195:
1196: To compare metrics (\ref{eubtze}) and (\ref{elmets}), it is enough to consider
1197: small deficit angles. For $\theta\ll1$,
1198: \begin{equation*}
1199: \g\approx1+\theta\frac{1-2\s}{2(1-\s)}.
1200: \end{equation*}
1201: Expanding expression (\ref{eubtze}) in $\theta$, we obtain metric (\ref{elmets})
1202: in the linear approximation.
1203:
1204: So, the elasticity theory induced metric reproduces only the linear
1205: approximation of the exact solution of the Euclidean Einstein's equations
1206: within the geometric theory of defects. Metric (\ref{eubtze}) obtained
1207: within the geometric approach is simpler in its form, valid for the whole
1208: range of radius $0<\rho<R_0$, all deficit angles $-1<\theta<\infty$, and
1209: Burgers vectors $b$. The components of the induced metric are proportional
1210: to the stress tensor of media. Therefore, the result obtained within the
1211: geometric theory of defects can be verified experimentally.
1212: %*******************************************************************
1213: \section{Conclusion}
1214: %********************************************************************
1215: We considered the BTZ black hole solution for zero cosmological constant
1216: in detail. This case describes the interior region of the BTZ black hole
1217: and is simple enough to perform all calculation explicitly. We showed
1218: that points at $r=0$ are just coordinate singularities, and all geometric
1219: quantities are regular there, while the line $r=r_-$ corresponding to the
1220: inner horizon is singular: these points are not points of a manifold.
1221: There are four cones at each point of the line $r=r_-$. The singularity
1222: arises only after the identification of the polar angle $\vf\sim\vf+2\pi$.
1223:
1224: The singularity structure is probably the same in a general case for negative
1225: cosmological constant, though it is analyzed only for zero cosmological
1226: constant.
1227:
1228: In the Euclidean case, the BTZ solution for zero cosmological constant
1229: breaks into two disjoint manifolds along the horizon $r=r_-$ with
1230: Euclidean and Lorentzian signature metrics. The manifold with the
1231: Euclidean signature metric has straightforward physical interpretation
1232: in the geometric theory of defects describing combined wedge and screw
1233: dislocations in crystals. We showed that the induced metric obtained
1234: entirely within the ordinary elasticity theory provides only the linear
1235: approximation for the exact solution of Einstein's equations. The
1236: Euclidean metric in the elastic gauge obtained from the BTZ solution
1237: depends nontrivially on Lame coefficients and can be measured
1238: experimentally.
1239:
1240: The Euclidean version of the BTZ solution for negative cosmological
1241: constant has also straightforward interpretation in the geometric theory
1242: of defects, but in this case it describes continuous distribution of
1243: dislocations and is not so visual.
1244:
1245: {\bf Acknowledgments.} We are grateful to I.~L.~Shapiro for discussions.
1246: One of the authors (M.K.) thanks the Universidade Federal de Juiz de Fora
1247: for the hospitality, the FAPEMIG, the Russian Foundation of Basic Research
1248: (Grant No.\ 05-01-00884), and the Program for Supporting Leading
1249: Scientific Schools (Grant No.\ NSh-6705.2006.1) for financial support.
1250: %********************************************************************
1251: \begin{thebibliography}{10}
1252:
1253: \bibitem{BaTeZa92}
1254: M.~Ba\~nados, C.~Teitelboim, and J.~Zanelli.
1255: \newblock Black hole in three-dimensional spacetime.
1256: \newblock {\em Phys.\ Rev.\ Lett.}, 69(13):1849--1851, 1992.
1257:
1258: \bibitem{AyMaZa04}
1259: E.~Ayon-Beato, C.~Martinez, and J.~Zanelli.
1260: \newblock {\em Phys.\ Rev.} D70, 044027, 2004.
1261:
1262: \bibitem{Carlip98}
1263: S.~Carlip.
1264: \newblock {\em Quantum Gravity in $2+1$ Dimensions}.
1265: \newblock Cambridge University Press, Cambridge, 1998.
1266:
1267: \bibitem{BaHeTeZa93}
1268: M.~Ba\~nados, M.~Henneaux, C.~Teitelboim, and J.~Zanelli.
1269: \newblock Geometry of the $2+1$ black hole.
1270: \newblock {\em Phys.\ Rev.}, D48(4):1506--1525, 1993.
1271:
1272: \bibitem{Steif96}
1273: A.~R. Steif.
1274: \newblock Supergeometry of three-dimensional black holes.
1275: \newblock {\em Phys.\ Rev.\ D}, 53(10):5521--5526, 1996.
1276:
1277: \bibitem{KatVol92}
1278: M.~O. Katanaev and I.~V. Volovich.
1279: \newblock Theory of defects in solids and three-dimensional gravity.
1280: \newblock {\em Ann.\ Phys.}, 216(1):1--28, 1992.
1281:
1282: \bibitem{KatVol99}
1283: M.~O. Katanaev and I.~V. Volovich.
1284: \newblock Scattering on dislocations and cosmic strings in the geometric theory
1285: of defects.
1286: \newblock {\em Ann.\ Phys.}, 271:203--232, 1999.
1287:
1288: \bibitem{Katana03}
1289: M.~O. Katanaev.
1290: \newblock Wedge dislocation in the geometric theory of defects.
1291: \newblock {\em Theor.\ Math.\ Phys.}, 135(2):733--744, 2003.
1292:
1293: \bibitem{Katana04}
1294: M.~O. Katanaev.
1295: \newblock One-dimensional topologically nontrivial solutions in the {S}kyrme
1296: model.
1297: \newblock {\em Theor.\ Math.\ Phys.}, 138(2):163--176, 2004.
1298:
1299: \bibitem{Katana05}
1300: M.~O. Katanaev.
1301: \newblock Geometric theory of defects.
1302: \newblock {\em Physics -- Uspekhi}, 48(7):675--701, 2005.
1303:
1304: \bibitem{Wolf72}
1305: J.~A. Wolf.
1306: \newblock {\em Spaces of constant curvature}.
1307: \newblock University of California, Berkley, California, 1972.
1308:
1309: \bibitem{Starus63}
1310: A.~Staruszkiewicz.
1311: \newblock Gravitational theory in three-dimensional space.
1312: \newblock {\em Acta Phys.\ Polon.}, 24(6(12)):735--740, 1963.
1313:
1314: \bibitem{Clemen76}
1315: G.~Clem\'ent.
1316: \newblock Field--theoretic particles in two space dimensions.
1317: \newblock {\em Nucl.\ Phys.}, B114:437--448, 1976.
1318:
1319: \bibitem{DeJatH84}
1320: S.~Deser, R.~Jackiw, and G.~'t~Hooft.
1321: \newblock Three-dimensional {E}instein gravity: Dynamics of flat space.
1322: \newblock {\em Ann.\ Phys.}, 152(1):220--235, 1984.
1323:
1324: \bibitem{Katana00A}
1325: M.~O. Katanaev.
1326: \newblock Global solutions in gravity. {L}orentzian signature.
1327: \newblock {\em Proc.\ Steklov Inst.\ Math.}, 228:158--183, 2000.
1328:
1329: \bibitem{FaGaSe93}
1330: C.~Farina, J.~Gamboa, and A.~J. Segu\'i-Santonja.
1331: \newblock Motion and trajectories of particles around three-dimensional black
1332: holes.
1333: \newblock {\em Class. Quantum Grav.}, 10(11):L193--L200, 1993.
1334:
1335: \bibitem{CrMaPe94}
1336: N.~Cruz, C.~Mart\'inez, and L.~Pe\~na.
1337: \newblock Geodesic structure of the $2+1$ black hole.
1338: \newblock {\em Class. Quantum Grav.}, 11(11):2731--2740, 1994.
1339:
1340: \bibitem{CarTei95}
1341: S.~Carlip and C.~Teitelboim.
1342: \newblock Aspects of black hole quantum mechanics and thermodynamics in 2+1
1343: dimensions.
1344: \newblock {\em Phys.\ Rev.\ D}, 51(2):622--631, 1995.
1345:
1346: \bibitem{KaKlKu99}
1347: M.~O. Katanaev, T.~Kl\"osch, and W.~Kummer.
1348: \newblock Global properties of warped solutions in general relativity.
1349: \newblock {\em Ann.\ Phys.}, 276:191--222, 1999.
1350:
1351: \bibitem{Amelin64}
1352: S.~Amelinckx.
1353: \newblock {\em The direct observation of dislocations}.
1354: \newblock Academic Press, New York and London, 1964.
1355:
1356: \bibitem{Kosevi81}
1357: A.~M. Kosevich.
1358: \newblock {\em Physical mechanics of real crystals}.
1359: \newblock Naukova dumka, Kiev, 1981.
1360: \newblock [in Russian].
1361:
1362: \bibitem{LanLif70}
1363: L.~D. Landau and E.~M. Lifshits.
1364: \newblock {\em Theory of Elasticity}.
1365: \newblock Pergamon, Oxford, 1970.
1366:
1367: \end{thebibliography}
1368:
1369: \end{document}
1370: