1: \documentclass[10pt]{article}
2: \usepackage{graphicx}
3: \usepackage{capt-of}
4: \graphicspath{{c/}}
5: \DeclareGraphicsExtensions{.eps,.ps}
6: \begin{document}
7: \vspace*{2.5cm}
8: {\Large\bf The complete quantum collapse scenario of a 2 + 1 dust shell:
9: Preliminary calculations}
10:
11: \vspace*{0.8cm}
12: {\bf L. Ort{\'\i}z and M. P. Ryan, Jr.
13: \footnote{Instituto de Ciencias Nucleares, Universidad Nacional Aut\'onoma
14: de M\'exico, A. Postal 70-543, M\'exico 05249 D. F., Mexico}}
15:
16: \vspace*{0.5cm}
17: \hspace*{1.5cm}\begin{minipage}[t]{9.5cm}\footnotesize{\it Abstract}
18:
19: \hrulefill
20:
21: If we consider the gravitational collapse of a material object to a black
22: hole, we would expect, for ranges of mass where a black hole would form, the
23: following scenario. A large enough object would collapse classically until
24: an event horizon forms, and to an external observer the object would be lost
25: from view. However, once the horizon has formed, the black hole will begin to emit
26: Hawking radiation and the hole will lose mass and the horizon will shrink.
27: The final state of this process could be either a zero-mass ``black hole'' with consequent
28: information loss, or some sort of ``quantum remnant.''
29:
30: A complete investigation of this process would require: 1) A complete and
31: consistent theory of quantum gravity coupled to some kind of field that
32: would provide the Hawking radiation \cite{hawk} (which could be the gravitational field itself ---
33: gravitons); 2) Some kind of definition of a ``horizon'' in this quantum gravity, and;
34: 3) The calculational tools to achieve a description of the scenario. Lacking these, one may
35: resort to toy models to try to give some sort of a preliminary answer.
36:
37: In this paper we will consider the collapse of an infinitesimally thin
38: dust shell in 2 + 1 gravity, where an exact minisuperspace quantum solution
39: exists, and try to make rough estimates of the collapse-Hawking radiation-remnant
40: formation process.
41:
42: \hrulefill
43: \end{minipage}
44: \vspace{1cm}
45: \section{Introduction}
46:
47: There has been some
48: interest over the years in the minisuperspace quantization of thin shells as
49: models of the full quantum collapse of more complicated objects. One
50: important question that could be asked is whether enough information
51: could be squeezed out of the grossly oversimplified model to mock up
52: a first look at the complete collapse scenario. In Ref. \cite{meleo} we used previous
53: work by Peleg and Steif \cite{stipel}, Israel \cite{Isra}, and Cris\' ostomo and Olea
54: \cite{crisol} on the classical equations of motion of a dust shell to construct a
55: quantum formulation of the collapse of such a shell.
56:
57: This quantum formulation was used to study the possibility of the formation
58: of a horizon, that is, whether some sort of ``quantum black hole,'' will form
59: or whether a shell of non-interacting particles will simply collapse to a point
60: where the uncertainty principle will provide a ``repulsive force'' and the
61: shell will reexpand into the same universe.
62: Simple considerations can answer this question up to a point. It seems
63: obvious that for large enough masses one might expect a reasonably
64: peaked wave function centered on a radius near to but outside the
65: classical Schwarzschild radius of the shell would, as it moves toward
66: zero radius, maintain enough coherence to pass beyond the Schwarzschild
67: radius almost in its entirety and a ``black hole'' would form with high
68: probability. For small masses one might expect that rapid spreading
69: would overwhelm the coherence of the wave function and the shell would
70: reexpand into our own universe with probability essentially (or exactly) one.
71:
72: In full relativistic quantum gravity these simple ideas are fraught with
73: difficulties. The most serious of these are:\\
74:
75: 1) The quantum shell problem becomes unphysical for masses much above the Planck
76: mass. While the minisuperspace approach has frozen out all radiative modes
77: of the gravitational field, and one is insisting on a single-particle
78: interpretation of the shell problem, the wave function in the Schr\" odinger
79: picture can still become pathological at a point where one might expect
80: graviton production to begin. H\'aj{\'\i}\v cek \cite{haj1}
81: has given a sufficient
82: but not necessary condition that shows that we might expect problems
83: above a couple of Planck masses. In \cite{cruz}, a qualitative
84: argument was presented that used Compton wavelength considerations to give a
85: similar bound.
86:
87: 2) There are very serious technical problems in the formulation of the
88: problem. Many authors have used the full ADM method to construct a
89: Hamiltonian for the system,
90: where they have chosen an internal time for the system in order to have a
91: true Hamiltonian
92: and a Schr\" odinger equation that allows the study of the time evolution of
93: the quantum system. The choice of an internal time leads to the problem
94: of quantum formulations that are not unitarily equivalent. Another problem
95: with these Hamiltonian formulations is that they often require an {\it ad
96: hoc\/} choice of a Hamiltonian in terms of variables defined on the shell
97: itself. Several of these Hamiltonians have been given by various authors
98: \cite{haj1}, \cite{hkk}. A Hamiltonian due to
99: H\'aj{\'\i}\v cek and Kucha\v r \cite{hyk}
100: has the advantage of being defined by a coherent procedure with no {\it ad
101: hoc\/} choices, and is formulated in terms of foliations of spacetime
102: by timelike surfaces. We will discuss this Hamiltonian in more detail
103: below. All of these Hamiltonians have limits on the mass of the shell
104: of a few
105: Planck masses. One formulation that does not seem to
106: have a mass limit is based
107: on the Wheeler-DeWitt equation corresponding to one of the Hamiltonians
108: in \cite{hkk}.
109:
110: 3) Many of the Hamiltonians are quite complicated and there is no real
111: chance of finding analytic solutions to the Schr\" odinger equations of
112: these models. Numerical solutions have been presented by a group consisting
113: of A. Corichi, G. Cruz, A. Minzoni. M. Rosenbaum and M. Ryan of the UNAM, N.
114: Smyth of the University of Edinburgh, and T. Vukasinac of the University
115: of Michoacan \cite{cruz}.\\
116:
117: The idea of Ref. \cite{meleo} was to study the quantum collapse
118: problem in 2 + 1 gravity, where one can address some of the difficulties
119: mentioned above in a context where
120: some of the problems mentioned above do not exist. The quantum problem is
121: unambiguous for all masses, so there is no problem of wave function
122: pathologies. As will be shown below, one possible Hamiltonian for this
123: problem has the form of that of a harmonic oscillator. This will allow us to find
124: simple analytic solutions that can be used to illustrate the development of the
125: quantum collapse of the shell, and allowed us to investigate the problem of
126: horizon formation.
127:
128: The usual view of black hole evaporation and remnant formation has a vast literature.
129: A recent review article \cite{koch} cites a large number of authors. This usual
130: view has three phases \cite{stromtom}\\
131:
132: 1) The {\it balding} phase, where the black hole radiates away all its multipole
133: moments, losing mass due to classical gravitational radiation.\\
134:
135: 2) The {\it evaporation} phase, where Hawking radiation of thermally distributed
136: quanta carries away mass until the Planck mass is reached.\\
137:
138: 3) The {\it Planck} phase, where quantum gravity is important and a remnant is
139: formed or all the mass is carried away.\\
140:
141: We will not be interested in the balding phase, only in the circularly symmetric
142: evaporation and Planck phases. There are a number of arguments for remnant
143: formation (see \cite{koch} and references therein), one of which is similar
144: to what we will consider here, arguments based on the uncertainty principle.
145:
146: Our idea is to investigate Hawking radiation in the quantum collapse of a
147: dust shell. This is a minisuperspace model of quantum gravity associated
148: with collapse. Our scenario is slight different from the usual view given
149: above, where the existence of a stable black hole created some time in
150: the past evaporates toward a stable remnant. The back reaction of the
151: Hawking radiation causes the mass of the collapsing shell to change,
152: affecting the shell evolution. We will attempt to show that this effect
153: causes a shell that has collapsed below its (classical) horizon will
154: reappear with a reduced mass.
155:
156: One possible way to calculate this scenario would be to define a
157: ``shell cloud'' from the probability density of shell radius, in the
158: same way that we define an ``electron cloud'' from the electron
159: probability density around a nucleus. For the electron cloud we
160: can calculate its classical electromagnetic field, while for the
161: shell cloud we can calculate its classical metric. The study of
162: Hawking radiation in this background could be illuminating.
163:
164: Since this idea is difficult to carry out, we will use a rough
165: approximation consisting of a classical shell that follows
166: the path of the expectation value of the quantum shell, and a model
167: Hawking radiation. Since in Ref. \cite{meleo} we saw that a shell
168: wave packet collapses to a minimum radius and the reexpands due to
169: uncertainty principle effects,
170: this scenario can have the
171: shell falling below its classical horizon, and the resulting black hole
172: begins to evaporate, and the shell, reappearing with mass loss as some sort
173: of ''dynamic remnant'' in the form of an expanding shell.
174:
175: The plan of the rest of the article is as follows. Section 2 will be
176: a brief resum\'{e} of the literature on the general relativistic minisuperspace
177: problem. Section 3 will consider the classical and quantum 2 + 1 problems and
178: discuss a model of horizon formation,
179: Section 4 will model the complete collapse-Hawking radiation-final state scenario, while
180: Section 5 will be conclusions and suggestions for future study.
181:
182: \section{The collapse of thin shells in relativistic quantum gravity}
183:
184: The study of the quantum collapse of dust shells is about a decade old \cite
185: {strom}. The classical collapse problem is fairly straightforward, and
186: can in principle be solved exactly. One assumes a $\delta$-function massive
187: (or null) shell where Birkhoff's theorem tells us that outside the shell
188: the metric is Schwarzschild and the metric inside the shell is Minkowski.
189: The Israel junction conditions can be used to derive the equation for the
190: evolution of the shell in terms of intrinsic variables on the shell itself,
191: the proper time, $\tau$, of an observer riding on the shell and the curvature
192: radius, $R(\tau)$, of the shell that he would measure. The equation for the
193: motion of the shell becomes \cite{Isra}
194: \begin{equation}
195: M = m\left \{ 1 + \left (\frac{dR}{d\tau}\right )^2 \right \}^{1/2} -
196: \frac{m^2}{2R}, \label{iseq}
197: \end{equation}
198: where $m$ is the rest mass of the shell (a constant of motion) and
199: $M$ is the Schwarzschild mass of the exterior metric. It is straightforward
200: to define $x = R/m$, $\tau \rightarrow \tau/m$, $V \equiv dx/d\tau$ and $M/m$ as our
201: Hamiltonian. Assuming $V = V(P)$ and solving
202: $\partial H/\partial P = V(P)$ for the ``momentum" $P$, we find that $V =
203: \sinh^{-1} (P)$ and our Hamiltonian becomes
204: \begin{equation}
205: H = \cosh P - \frac{m}{2x}, \label{hajham}
206: \end{equation}
207: a Hamiltonian
208: given by H\'aj{\'\i}\v cek \cite {haj1}. Unfortunately, this is not the only
209: Hamiltonian that gives the equation of motion (\ref{iseq}), and we are left
210: with the problem of defining an ``appropriate'' Hamiltonian for the
211: problem. A number of Hamiltonians for different choices time, including
212: the time of an observer at the center of the shell where space is flat, and
213: a Wheeler-DeWitt equation identical to that for a relativistic charged
214: particle radially falling in a Coulomb potential are given
215: by H\'aj{\'\i}\v cek, Kay and Kucha\v r \cite{hkk}.
216:
217: Kucha\v r and H\'aj{\'\i}\v cek \cite{hyk}, dissatisfied with such
218: {\it ad hoc\/} Hamiltonians,
219: have managed to construct a Hamiltonian for collapsing dust shells that
220: comes directly from an ADM reduction of the Hilbert-plus-matter action.
221: The problem with this approach is that, as spacetime quantities, the matter
222: variables are proportional to $\delta [R - R_0(\tau)]$, where $R_0 (\tau)$ is
223: the position of the shell as a function of shell proper time. It is
224: virtually impossible to reduce the time derivatives of these delta functions
225: to reasonable variables in the matter Lagrangian that can give a shell
226: Hamiltonian that describes the motion purely in terms of canonical variables
227: on the shell. Kucha\v r and H\'aj{\'\i}\v cek used an
228: ingenious method (following \cite{hajki}) based on the fact that
229: the ADM reduction by a $3 + 1$ foliation is not restricted to foliation
230: by spacelike surfaces, but works just as well for foliations by timelike
231: surfaces. Unfortunately, this approach usually leads to an ill-posed problem, but
232: in the case of the shell it does not. Using this approach and a
233: formulation of the dust fluid
234: velocity in terms of velocity potentials, they define a new Hamiltonian.
235: The cost of this consistent formulation is a very complicated Hamiltonian,
236: \begin{equation}
237: H= -\sqrt{2}R\left (1 - \frac{M}{R} - \sqrt{1 - \frac{2M}{R}}\cosh\frac{P}
238: {R}\right )^{1/2} \qquad R \geq 2M, \label{karham1}
239: \end{equation}
240: \begin{equation}
241: H = -\sqrt{2}R\left (1 - \frac{M}{R} - \sqrt{\frac{2M}{R} - 1} \sinh
242: \frac{P}{R} \right )^{1/2} \qquad 0 \leq R \leq 2M. \label{karham2}
243: \end{equation}
244:
245: Almost all of the Hamiltonians that have been given
246: (except for the Wheeler-DeWitt equation of
247: \cite{hkk}) seem to have mass limits beyond which the wave functions
248: become pathological. These Hamiltonians are all so complicated that it
249: seems impossible to find analytic solutions to assist in the interpretation.
250: In \cite{cruz}, Corichi et
251: al. have given a series of numerical solutions that give the evolution of
252: wave functions that are initially sharply peaked over a radius near the classical
253: horizon, $R = 2M$, for the $\cosh P$ Hamiltonian (\ref{hajham}) and an
254: approximation to the
255: Hamiltonian given by Eqs. (\ref{karham1}-\ref{karham2}). All of these solutions
256: show evolution of the peak toward $R = 0$ with a bounce caused by the
257: boundary conditions at $R = 0$, with the appearance of interference fringes as well as
258: a rapid spread of the wave packet. The
259: Kucha\v r-H\'aj{\'\i}\v cek Hamiltonian has wave functions similar to those of
260: (\ref{hajham}), but with many rapid oscillations superimposed.
261:
262: Since these Hamiltonians are self-adjoint, unitary evolution implies that
263: a peak formed from scattering states will always rebound to $R = \infty$.
264: One can ask whether this behavior means that all quantum collapse of this
265: sort implies a rebound into our own universe. Since $\tau$ is proper time
266: on the shell and $R$ is also a shell variable, such questions can only
267: be answered by knowing the global quantum spacetime surrounding the shell.
268: In any case, the scenario of the shell observer is that he sees (begging
269: questions of quantum measurement and the reduction of
270: the wave packet) the shell
271: collapse to some point near $R = 0$, where uncertainty principle effects
272: change the classical equations of motion and the shell rebounds (actually,
273: a shell where the particles do not interact directly with one another
274: ``passes through itself'' and reexpands, that is, each radially infalling
275: particle passes through $R = 0$ and the azimuthal angle $\theta$ jumps from
276: the initial $\theta_0$ to $\theta_0 + \pi$). Even if the shell has
277: collapsed below its classical horizon, in finite proper time it will again
278: be above the horizon and traveling toward $R = \infty$. This
279: quasi-classical scenario is not surprising. In proper time, a classical shell
280: that manages to avoid forming a curvature singularity at $R = 0$ would behave
281: in this way, but as $R$ becomes greater than $2M$ the shell would be in
282: a universe beyond our temporal infinity ($i_{+}$), or in ``another universe.''
283:
284: This quasi-classical scenario is what one might expect to see for a large
285: mass where the quantum fluctuations would be small compared to the mass,
286: so the evolution of the wave packet would be coherent long enough for
287: the shell to collapse past its horizon and the shell would emerge from
288: the horizon into a new universe. However, if the wave packet spreads
289: sufficiently so that the width is greater than the classical horizon radius,
290: we can see that a horizon might never form, and the shell would reexpand
291: into our own universe.
292:
293: The problem of horizon formation in the quantum system is very difficult.
294: Event horizons are global features and one has to try to define a global
295: feature in a fluctuating manifold. Of course, this quantum manifold must
296: be constructed in terms of the full minisuperspace canonical
297: quantum gravity of the shell-metric system.
298: In the shell case we tend to use some kind of approximation to construct
299: the spacetime metric. Kucha\v r \cite{ku2} argues that for the simple
300: shell minisuperspace we may just replace the the shell mass (Schwarzschild
301: mass) and the shell radius in the metric outside the shell by the
302: corresponding operators to make a ``metric operator.'' The problem with
303: this metric operator is that it is a function of shell proper time, and
304: studies of the metric close to the shell \cite{tatj} cannot tell us
305: whether a true event horizon (tied to observer time at infinity) forms.
306:
307: Other approximations are under study \cite{us}. The simplest calculation
308: is to calculate $<\hat R(\tau)>$ and the uncertainty $\Delta R = \sqrt{<(\hat R - <\hat R>)^2>}$
309: and check whether $\Delta R$ becomes very large as $<\hat R>$ becomes small so that
310: $<\hat R>$ does not fall below the classical horizon and $\Delta R$ is larger than
311: the classical horizon, which
312: can be taken as an indication of the non-formation of a horizon. In
313: \cite{us}, numerical evaluations of these two quantities will be presented
314: for the Hamiltonians (\ref{hajham}) and (\ref{karham1}-\ref{karham2}), and
315: they
316: suggest that no horizon forms for small masses. Another possibility (to
317: be considered in \cite{us}) would be to take M$|\psi (R, \tau)|^2$ to be a
318: classical density $\rho (R, \tau)$ and calculate the classical metric due
319: to a classical fluid with this mass distribution and see whether a horizon
320: forms. Note that $M$ should be either the rest mass or the Schwarzschild
321: mass. It is not yet clear which. There are
322: technical problems with this calculation. We have to
323: calculate the metric from a density that is given in terms of a solution
324: of the Schr\" odinger equation for our Hamiltonian, and there is no
325: guarantee that this density can be made to obey the equation
326: $T^{\mu \nu}_{\, \, \, \, \, ;\nu} = 0$ for our fluid. Another possibility
327: considered in
328: \cite{us} is that of the ``metric operator'' mentioned above, which was
329: extended to the whole
330: manifold outside the shell and used to define a ``quantum'' stress-energy
331: tensor.
332:
333: The rotationless 2 + 1 problem has some advantages over the 3 + 1 problem.
334: The Hamiltonian can be constructed fairly easily, and, as will be shown,
335: has the form of a harmonic oscillator. The Schr\" odinger
336: equation for this Hamiltonian has well-known analytic solutions. The
337: expectation value of $R$ and its uncertainty can, in principle, be
338: calculated analytically. The analogue of the other calculation using $\rho
339: (R, \tau)$ is much simpler than in the 3 + 1 case. The classical horizon is
340: easily
341: found. In the next section these ideas will be considered.
342:
343: \section{The classical and quantum collapse of shells in 2 + 1 gravity}
344:
345: The first element we need for this problem is an equation for the radius of
346: the shell. This problem has been studied in detail by Peleg and Steif
347: \cite{stipel}, using the 2 + 1 version of the original formulation of
348: Israel \cite{Isra}, and Cris\' ostomo and Olea \cite{crisol}, using canonical methods.
349:
350: Israel
351: studied the collapse of a shell in 3 + 1 gravity, represented by a
352: delta-function sphere of dust of
353: radius $R(\tau)$. In 2 + 1 gravity the shell is a
354: circle, i.e.
355: a ring of matter, also of radius $R(\tau)$. The metric of spacetime will be
356: written in circular coordinates, where flat space is represented by the metric
357: \begin{equation}
358: ds^2 = -dt^2 + dr^2 + r^2 d\theta^2.
359: \end{equation}
360:
361: The equation for a circle $R(\tau)$ is
362: \begin{equation}
363: ^{(3)}r = R(\tau), \qquad ^{(3)}\theta = \theta, \qquad ^{(3)}t = t(\tau).
364: \end{equation}
365: We will use the notation $i, j = 1, 2, 3$ and $A, B = 1, 2$. We will now need
366: a set of coordinates $\xi_A$ on the circle, which we will take to be $\xi_A =
367: (\tau, \theta)$.
368:
369: Off the ring of matter, the 2 + 1 version of Birkhoff's theorem says
370: that the three-dimensional metric is a static or stationary solution to
371: \begin{equation}
372: R_{ij} - \frac{1}{2}g_{ij} R = -\Lambda g_{ij},
373: \end{equation}
374: (where we have to have a cosmological constant $\Lambda$ to avoid locally completely
375: flat solutions) both
376: inside and outside the circle. Here, as in Ref. \cite{meleo}, we will study
377: the static case,
378: where the matter has no angular momentum. In this case the metric
379: has the form
380: \begin{equation}
381: ds^2 = -f(r)dt^2 + \frac{1}{f(r)}dr^2 + r^2 d\theta^2,
382: \end{equation}
383: and the well-known solution is
384: \begin{equation}
385: f = B - \Lambda r^2,
386: \end{equation}
387: $B$ a constant. Since $\Lambda$ has units of inverse length, we will
388: write, as is common, $\Lambda = \pm 1/\ell^2$.
389: The final form of the static, circularly-symmetric metric is
390: \begin{equation}
391: ds^2 = - \left( -M \mp \frac{r^2}{\ell^2}\right ) dt^2 + \frac{1}{\left ( -M \mp \frac{r^2}{\ell^2}\right )}
392: dr^2 + r^2 d\theta^2, \label{metrc}
393: \end{equation}
394: where we have taken $B$, following Ba\~nados, Teitelboim and Zanelli (BTZ) \cite{BTZ}, to be $-M$,
395: M a Schwarzschild mass. We will take
396: the metric inside the ring to be the 2 + 1 AdS metric where $M = -1$
397: \cite{BTZ2}, \cite{carl}. As in the 3 + 1 case, we
398: expect that outside the ring we will have a black hole metric with some
399: ``Schwarzschild mass" $M$, that is,
400: \begin{equation}
401: ds^2 = - \left(-M + \frac{r^2}{\ell^2}\right ) dt^2 + \frac{1}{\left (-M + \frac{r^2}{\ell^2}\right )}
402: dr^2 + r^2 d\theta^2, \label{metfin}
403: \end{equation}
404: and inside the ring,
405: \begin{equation}
406: ds^2 = - \left( 1 + \frac{r^2}{\ell^2}\right ) dt^2 + \frac{1}{\left ( 1 + \frac{r^2}{\ell^2}\right )}
407: dr^2 + r^2 d\theta^2.
408: \end{equation}
409:
410: With these preliminaries, in Ref. \cite{meleo}
411: the equation of motion of the shell was calculated using the
412: 2 + 1 analogue of the Israel formulation of the equation of
413: motion of a shell in 3 + 1 gravity, which was similar to that
414: used by Peleg and Steif \cite{stipel}.
415:
416: In \cite{meleo} we used the jump in the Einstein equations between the interior and
417: exterior of the shell, and
418: the Lanczos relation,
419: \begin{equation}
420: \gamma_{AB} - g_{AB}\gamma = 8\pi S_{AB}, \label{lanc}
421: \end{equation}
422: where $\gamma_{AB}$ is the jump of $K_{AB}$, the extrinsic curvature of the shell as
423: seen from the inside and the outside of the shell, the $g_{AB}$ are the components of
424: the induced metric on the surface in
425: terms of $\tau$ and $\theta$,
426: $\gamma = g^{AB}\gamma_{AB}$, and $S_{AB}$ is the surface stress-energy
427: tensor, (In our case, we will
428: be interested in a dust shell, so we will take $S_{AB} = \sigma u_A u_B$,
429: $\sigma$ the rest mass density
430: of the ring, where $\sigma = m/2\pi R$, $m$ the total rest mass of the shell.)
431: to find the classical equation of motion for the shell,
432: \begin{equation}
433: \ddot R = -R/\ell^2.\label {eq2}
434: \end{equation}
435:
436: Equation (\ref{eq2}) has a first
437: integral,
438: \begin{equation}
439: E = \frac{1}{2} \dot R^2 + \frac{R^2}{2\ell^2},\label{eeq}
440: \end{equation}
441: and, using the Lanczos equation directly,
442: we find
443: \begin{equation}
444: \sqrt{1 + 2E} - \sqrt{-M + 2E} = 4m,
445: \end{equation}
446: which can be solved for E as
447: \begin{equation}
448: E = \left (\frac{M}{32m} + \frac{1}{32m} + \frac{m}{2}\right )^2 - \frac{1}{2}.
449: \label{eeq2}
450: \end{equation}
451:
452: Equation (\ref{eeq}) was used to construct a Hamiltonian
453: formulation of the problem.
454: This equation is simply the energy equation for a harmonic oscillator,
455: so it is obvious that
456: if we take $E$ as our Hamiltonian we can define a
457: momentum $P_R$ as $\dot R$, and we have
458: \begin{equation}
459: H = \frac{1}{2}P_R^2 + \frac{R^2}{2\ell^2}.
460: \end{equation}
461: Hamilton's equations for this Hamiltonian are equivalent to the equation
462: of motion $\ddot R = -R/\ell^2$.
463:
464: Once we have a Hamiltonian in terms of the ring variables $R$ and $\tau$,
465: it is possible to construct a Schr\" odinger equation for the problem.
466: If we write the classical relation
467: \begin{equation}
468: \left (\frac{M}{8m} + \frac{1}{8m} + 2m\right )^2 - 1 = \dot R^2 + \frac{R^2}{\ell^2},
469: \label{hameq1}
470: \end{equation}
471: on the right-hand-side $M$ and $m$ are dimensionless, so $M \equiv M/M_{\rm Pl}$ and $m
472: \equiv m/M_{\rm Pl}$, where $M_{\rm Pl}$ is the Planck mass, $M_{\rm Pl} = c^2/G^{(2)}$
473: (we will later need the Planck length, $L_{\rm Pl} = \hbar G^{(2)}/c^3$), where $G^{(2)}$
474: is the two-dimensional Newton's constant with units $[L^2]/[M][T^2]$.
475:
476: In conventional units, our equation for $R$ becomes
477: \begin{equation}
478: \left (\frac{M}{8m} + \frac{M_{\rm Pl}}{8m} + \frac{2m}{M_{\rm Pl}}\right )^2 - 1 = \frac{1}{c^2}
479: \left (\frac{dR}{d\tau} \right )^2 +
480: \frac{R^2}{\ell^2}.
481: \end{equation}
482: Our ``Hamiltonian'' should have the units of energy for a conventional
483: quantum Hamiltonian, so
484: we should multiply Eq. (\ref{hameq1}) above by $M_{\rm Pl} c^2$, our
485: $R$-momentum $P_R$ becomes
486: $M_{\rm Pl} \dot R$ , and with
487: \begin{equation}
488: H = \frac{M_{\rm Pl}c^2}{2} \left (\frac{M}{8m} ^2 + \frac{M_{\rm Pl}}{8m} +
489: \frac{2m}{M_{\rm Pl}}\right)^2 - \frac{M_{\rm Pl}c^2}{2}, \label{hamval}
490: \end{equation}
491: and
492: \begin{equation}
493: H = \frac{P^2_R}{2M_{\rm Pl}} + \frac{M_{\rm Pl}}{2}\omega_0^2 R^2,
494: \end{equation}
495: where $\omega_0 = c/\ell$.
496:
497: In \cite{meleo} we quantized this system with $H$ replaced by an operator $\hat H$.
498: If we look at (\ref{hamval}), we see that the right-hand-side must become an
499: operator, so the left-hand-side must also be an operator. We also took
500: $M$ to be a $q$-number, and $m$ and $M_{\rm Pl}$ $c$-numbers.
501: Of course, there is no special reason to make this choice, but we do so to
502: make contact with previous work (see, for example, \cite{ku2}). If we
503: consider Eq. (\ref{hajham}), this choice is motivated by the fact that
504: the Schr\"odinger equation for (\ref{hajham}) is similar to the hydrogen atom
505: Schr\"odinger equation, with $m$ playing the role of $e^2$, and it is usual in the
506: hydrogen atom to take $e^2$ as a $c$-number rather than a $q$-number.
507:
508: We can now take the wave function of the system to be $\tilde \psi =
509: \psi_M \psi (R, \tau)$, with $\psi_M (M)$ an approximate eigenstate of
510: $\hat M$ with eigenvalue $M_0$. An exact eigenstate of $\hat M$ of this
511: type would be $\delta (M - M_0)$, but to avoid problems with the integral of
512: the square of a delta function we will assume that $\psi_M$ is an
513: extremely sharply peaked wave function centered on $M = M_0$. In this case,
514: our Schr\" odinger equation becomes ($\hat M^2 \psi_M \approx M_0^2 \psi_M$,
515: and realizing $\hat P_R$ as $-i\hbar \partial/\partial R$)
516: \begin{eqnarray}
517: \left [ \frac{M_{\rm Pl}c^2}{2} \left (\frac{M_0}{8m}+
518: \frac{M_{\rm Pl}}{8m} +
519: \frac{2m}{M_{\rm Pl}}\right)^2 -
520: \frac{M_{\rm Pl}c^2}{2}\right ] = \nonumber \\
521: = i\hbar \frac{\partial \psi(R, \tau)}{\partial \tau} = -\frac{\hbar^2}
522: {2M_{\rm Pl}} \frac{\partial^2\psi (R, \tau)}{\partial R^2} +
523: \frac{M_{\rm Pl}}{2}\omega_0^2 R^2 \psi (R, \tau).
524: \label{schreq}
525: \end{eqnarray}
526: One curious fact about this equation is that, if we define $\psi =
527: \exp(-iE\tau/\hbar)\psi(R)$, then the energy eigenvalues are
528: \begin{equation}
529: E_n = (n + \frac{1}{2})\hbar \omega_0 = (n + \frac{1}{2})
530: \frac{\hbar c}{\ell},
531: \end{equation}
532: and since $E$ is given by the first line of (\ref{schreq}), there is a
533: discrete relation between the Schwarzschild mass, $M_0$, and the rest mass,
534: $m$,
535: \begin{equation}
536: \left (\frac{M_0}{8m} +
537: \frac{M_{\rm Pl}}{8m} +
538: \frac{2m}{M_{\rm Pl}}\right)^2 - 1 =
539: (2n + 1)\frac{\hbar G}{c^3 \ell} = (2n + 1)\frac{L_{\rm Pl}}{\ell}.
540: \end{equation}
541: We will now return to units where $G = c = \hbar = 1$, $M_0$ now meaning
542: $M_0/M_{\rm Pl}$, $m$ now meaning $m/M_{\rm Pl}$, and $\ell$ meaning
543: $\ell/L_{\rm Pl}$. If we solve for $M_0$ in terms of $m$ and $\ell$, we find
544: \begin{equation}
545: M_0 = 16m^2\left [\sqrt{\frac{1}{4m^2} + \frac{1}{2m^2}\left[
546: \frac{1}{\ell}\left (n + \frac{1}{2}\right )\right ]} - 1\right ] - 1.\label{meq}
547: \end{equation}
548:
549: For a moment we will return to conventional units, and define
550: the following dimensionless variables. We define $y = R/\sqrt{\ell
551: L_{\rm Pl}}$ and
552: $T = c\tau/\ell$. The Schr\" odinger equation now becomes
553: \begin{equation}
554: i\frac{\partial \psi (y, T)}{\partial T} = -\frac{\partial^2 \psi (y, T)}
555: {2\partial y^2} + \frac{y^2}{2} \psi(y, T)
556: \end{equation}
557: which has eigensolutions
558: \begin{equation}
559: \psi = e^{-i(n + \frac{1}{2})T}\frac{1}{\sqrt{2^n n!}(\pi)^{1/4}}
560: e^{-y^2/2}H_n (y),
561: \end{equation}
562: $H_n$ Hermite polynomials.
563:
564: In Ref. \cite{meleo} we studied
565: the evolution of a wave packet sharply peaked around a value of $y = y_0$,
566: a point some distance outside the point where a classical horizon would form if the
567: radius of the shell were to fall below $R_H = \ell \sqrt{M_0}$ and
568: followed its movement as the packet fell toward $R = 0$.
569: In the 2 + 1 case it is possible to give an exact analytic
570: expression for the wave packet as a coherent harmonic-oscillator state.
571: However, even though the eigensolutions are nothing but harmonic oscillator wave
572: functions, the radial variable $R$ cannot be negative. In order to keep
573: this from happening we will took the potential to be that of a half
574: oscillator with an infinitely hard wall at $R = 0$ This potential is
575: shown in Figure 1
576:
577: \vspace{1cm}
578: \centerline{\scalebox{.4}{\includegraphics{fig1.eps}}}
579: \captionof{figure}{\small The harmonic oscillator potential for our problem with $M_{\rm Pl} \omega_0^2/2 = 1$. The dashed line shows a typical position of the classical horizon radius, $r_H$.}
580: \vspace{0.5cm}
581:
582: \noindent This meant that $\psi (0, \tau)$ was always
583: zero. This boundary condition can be enforced by only expanding in odd-$n$
584: harmonic oscillator eigenfunctions.
585:
586: In \cite{meleo} we began with a difference
587: of two Gaussian states, one peaked around $y = +y_0$ and the other around
588: $y = -y_0$, so that their sum at $y = 0$ was zero.
589: This state is
590: \begin{equation}
591: \psi (y, 0) = \alpha [e^{-\frac{1}{2}(y - y_0)^2} - e^{-\frac{1}{2}(y +
592: y_0)^2}],\label{init}
593: \end{equation}
594: only valid for $y > 0$, with $\alpha$ a normalization constant. Since this
595: is a sum of two Gaussian states, we can use standard techniques to construct
596: the difference between two coherent Gaussian states with (\ref{init}) as an
597: initial condition. The result is
598: \begin{eqnarray}
599: \psi(y,\tau) = \alpha e^{-i\omega_0 \tau/2} e^{i y_0^2 \sin 2\omega_0
600: \tau/4}\times \\ \nonumber \label{cohstat}
601: [\exp(-\frac{1}{2}\{ (y - y_0\cos \omega_0 \tau )^2\})
602: \exp (-iy y_0 \sin \omega_0 \tau )\\ \nonumber
603: - \exp(-\frac{1}{2}\{ (y + y_0\cos \omega_0 \tau)^2 \} ) \exp (iy y_0
604: \sin \omega_0 \tau) ],
605: \end{eqnarray}
606: which is zero at $y = 0$ for all $\tau$. The normalization $\alpha$ is
607: easily found to be
608: $\alpha^2 = [\sqrt{\ell}\sqrt{\pi}(1 - e^{-y_0^2})]^{-1}$.
609:
610: We connect the variables that describe the wave function with
611: the radius of the
612: classical horizon, $r_H$, shown in Fig. 1.
613: The
614: unnormalized probability density, $\rho \equiv \psi^{*} \psi/\alpha^2$ is
615: \begin{equation}
616: \rho = e^{-(w - \lambda \cos T)^2} + e^{-(w + \lambda \cos T)^2}
617: -2e^{-w^2} e^{-\lambda^2 \cos^2 T} \cos(2\lambda w \sin T), \label{wpsi}
618: \end{equation}
619: (where $w = y/\sqrt{\ell}\sqrt{M}$).
620: Figures showing the evolution of this probability density are given in \cite{meleo}.
621: This evolution begins with a Gaussian packet at $T = 0$ that collapses toward $y = 0$,
622: developing interference fringes as it reaches a minimum for some
623: $y < y_0$, then rebounds toward $y = y_0$ again.
624: This pattern then repeats forever. We will only be interested in one cycle of this pattern.
625:
626: \subsection{The formation of a horizon}
627:
628: In previous articles the quantization of the shell collapse was used to
629: study the possibility of the formation of
630: a horizon in the quantum collapse. As mentioned in Sec. 2, this concept has
631: many difficulties. An event horizon is a global construct and it has no
632: local definition. This means that in quantum gravity one would have to
633: return to the starting point and try to define what a ``quantum horizon''
634: might be. Once this definition has been decided upon, one must try to find
635: out if some collapse process will result in the formation of such a horizon,
636: with the result being a probability of
637: horizon formation. Theories of quantum gravity in their present state are
638: far from being able to give us this result,
639: so, in shell collapse, some articles \cite{cruz}, \cite{us}, \cite{vinart}
640: have tried to give an estimate of horizon formation by finding out if a
641: sharply peaked wave packet, during its collapse toward $R = 0$, will fall,
642: in some sense, below the classical horizon radius, $r_H$. In some sense,
643: because the packet will usually spread and basically will never lie entirely
644: below $R = r_H$. One could use the integral of $\psi^{*} \psi$ from $R = 0$
645: to $R = r_H$, which is a number less than one, as the probability of horizon
646: formation. Another possibility would be to use the operator $\hat M$ in the
647: expression $\hat R_H = \ell\sqrt{2\hat M - 1}$ to define a
648: ``horizon operator'' that could be used to define the probability of horizon
649: formation. In previous work it was decided to use $<\hat R>$ as a function
650: of $\tau$, a quantity that falls from $R_0$ (the $R$ associated with $y_0$)
651: to a minimum and then begin to
652: increase again. If this minimum is below the classical horizon, one
653: can say that a horizon forms and if not, not. This is a yes-no answer
654: instead of a probability, but it is
655: a quick estimate. We will use this concept below.
656:
657: It is not difficult to calculate $<\hat R>(\tau)$ from the wave packet given
658: in (\ref{cohstat}), but the result is a complicated function that contains
659: the error function and Dawson's function, so trying to find the minimum of
660: $<\hat R>(\tau)$ by finding the point where $d<\hat R>/d\tau = 0$ requires
661: the solution of a transcendental algebraic equation, making it difficult to
662: give an analytic expression for the point where a horizon would form.
663: To avoid this problem, in Ref. \cite{meleo} we used
664: the fact that at $T = \pi/2$, the point
665: where the packet begins to rebound, the peak nearest $R = 0$ is high and
666: narrow. We used the position of this peak as a parameter to tell us
667: whether a horizon would form or not. If the position of the peak is below $r_H$,
668: a horizon forms, and if not, not. In previous work it was found that for
669: large masses the shell collapse was so rapid that the wave packet fell below
670: $r_H$ so quickly that quantum mechanics did not allow it to rebound before
671: that point, while for small masses the rebound occurred for $R > r_H$. We
672: were able to give a range of masses where no horizon would form.
673:
674: In the present article, we will study the behavior of $<\hat R>(\tau)$, but
675: we will use the threshold mass where a horizon will form from \cite{meleo}
676: in order to estimate the point where a horizon forms. We will not give any details
677: of the calculation in \cite{meleo}, only the results.
678:
679: We can define several purely numerical quantities that will be used to
680: explain the mass limits. We will key off $r_H = \ell \sqrt{M}$, the value of $R$ of the
681: shell as it passes its classical horizon. We will also make use of the
682: variable $y$. The peak of the initial wave function is at $y_0$, and
683: using the fact that the value of $y$ corresponding to the classical horizon
684: is $y_H = \sqrt{\ell} \sqrt{M}$, we can define $y_0 = \lambda y_H$, $\lambda >
685: 1$. The point where the largest peak reaches its minimum at $y_c = \gamma y_H$.
686:
687: Studying the motion of the wave packet, it was possible to find an analytic
688: relation between $\lambda$ and $\gamma$ in terms of $M_0$, where, for
689: a given $\lambda$, $M_0$ is a monotonically decreasing function of $\gamma$ (and $\ell$).
690: Since $\gamma$ must be less than $\lambda$ for collapse to make sense, we have
691: a minimum of $M_0$ as a function of $\ell$. Another minimum can be found by
692: studying $H$ as a function of $M_0$, $m$ and $\ell$. If we take $<2\hat H>$
693: equal to $\left (\frac{M_0}{8m} +
694: \frac{M_{\rm Pl}}{8m} + \frac{2m}{M_{\rm Pl}}\right)^2 - 1$, and calculate
695: $<2\hat H>$ as a function of $M_0$, $m$ and $\ell$, we find a real
696: solution for $M_0$ only for $m$ greater than a minimum value. Taking $m$
697: to be this minimum, we find another minimum for $M_0$. Equating these two
698: minima, we find a numerical value for $\ell$, $\ell \approx 0.18$. For this
699: value of $\ell$, we find that for $M_0 < 1.48$ there is no real solution for
700: $M_0$, and for
701: \begin{equation}
702: 1.48 < M_0 < 3.32,
703: \end{equation}
704: no horizon forms, while for $M_0 > 3.32$ one does.
705:
706: Of course, as has been mentioned above, the rebound of the wave packet,
707: once it has passed the horizon does not mean that the shell returns through
708: the horizon into the same spacetime where it began. In spite of the fact
709: that in terms of proper time the shell exits the horizon in a finite time,
710: in terms of the time, $t$, of an observer at infinity this exit occurs at
711: time {\it after} $t = \infty$, or into ``another universe.''
712:
713: \section{The complete collapse--Hawking radiation--final state scenario}
714:
715: As we mentioned in Sec. 1, we want to consider the complete evolution of a
716: collapsing quantum shell. Without a complete, consistent quantum theory of
717: gravity, we have to appeal to some sort of approximation. One possible
718: approximation that might be the closest we can come to a real theory would
719: be to define a ``shell cloud'' in the same way we define an ``electron cloud''
720: when we consider such concepts as electron shielding of the charge of a
721: nucleus. If we consider a hydrogen atom with the electron in some quantum
722: state, $\psi({\bf x}, t)$, we can define an electric charge density by
723: $\rho = e\psi^{*} \psi$, and then calculate the electromagnetic field
724: due to this ``electron cloud.'' We can do the same with the shell, that is,
725: define a classical mass density as something like $\rho = M_0 |\psi (R, \tau)|^2_{R
726: \rightarrow r}$, and try to calculate a classical metric $g_{ij} (r, \tau)$ from
727: this mass density.
728:
729: Of course, $\rho(r,\tau)$ is not enough to solve the 2 + 1 Einstein equations,
730: we need a real stress-energy tensor, $T^{ij}$. If we can define such a $T^{ij}$, we
731: can write the spherically symmetric metric as
732: \begin{equation}
733: ds^2 = -e^{\nu(r, t)}dt^2 + e^{\lambda(r, t)}dr^2 + r^2 d\theta^2,
734: \end{equation}
735: which, if $T^{ij} = T^{ij}(r,t)$, we have(see Synge \cite{sing})
736: \begin{equation}
737: e^{-\lambda} = -\Lambda r^2 + 16\pi \int^r_0 r T^0_0 dr,
738: \end{equation}
739: \begin{equation}
740: \nu = -\lambda - 16\pi \int^r_0 r e^{\lambda} (T^0_0 - T^r_r)dr.
741: \end{equation}
742: It is easy to relate $T^0_0$ to the mass density $M\sigma \psi^* \psi$,
743: where $\sigma$ is a numerical factor to take into account the fact that
744: we want $e^{-\lambda}$ to be $-\Lambda r^2 - M$ for large $r$, but
745: $T^r_r$ is undefined. We also have to study the rest of the components
746: of $T^i_j$ for possible singular behavior. We can show, as in the 3 + 1 case,
747: that $T^{ij}$ should diagonal except for $T^{0r}$. This means that we need
748: only find $T^{rr}$ and $T^{\theta \theta}$. The Bianchi identities give
749: \begin{equation}
750: T^r_0 = -\frac{1}{r} \int^r_0 r\frac{\partial (T^0_0)}{\partial t} dr,
751: \end{equation}
752: \begin{equation}
753: T^{\theta}_{\theta} = rT^0_0 + r\frac{\partial (T^0_r)}{\partial t},
754: \end{equation}
755: so, in reality, we only need to define $T^r_r$ as long as $T^0_0$ can be calculated from
756: from $M\sigma \psi^* \psi$. We can consider a fluid stress energy of the type
757: \begin{equation}
758: T^i_j = (\rho + p)u^i u_j + p\delta^i_j, \qquad u^{\theta} = 0.
759: \end{equation}
760: From $u^i u_i = -1 = -e^{\nu}(u^0)^2 + e^{\lambda} (u^r)^2$, and
761: defining $e^{\lambda} (u^r)^2 \equiv U^2$, we have
762: \begin{equation}
763: T^0_0 = -(1 + U^2)\rho - U^2 p,
764: \end{equation}
765: \begin{equation}
766: T^r_r = U^2 \rho + (1 + U^2) p,
767: \end{equation}
768: and
769: \begin{equation}
770: T^0_0 - T^r_r = -(1 + 2U^2)(\rho + p),
771: \end{equation}
772: and
773: \begin{equation}
774: T^{\theta}_{\theta} = p.
775: \end{equation}
776: We can have $T^0_0 - T^r_r = 0$ if $p = -\rho$ and
777: $T^0_0 = -\rho$. If we have $\rho = M\sigma \psi^* \psi$, and since we have taken
778: the Schr\" odinger equation to have the form of Eq. (\ref{schreq}), what we call $\psi$ is actually
779: $\sqrt{r} \psi$, we will have (taking $\sigma = 1/16\pi$ in order to make $e^{-\lambda}$
780: go to $-\Lambda r^2 - M $as $r \rightarrow \infty$)
781: \begin{equation}
782: e^{-\lambda} = -\Lambda r^2 - M\int^r_0 \psi^* \psi (r, \tau)dr,
783: \end{equation}
784: and $\nu = -\lambda$.
785:
786: We can calculate $T^r_0$ and $T^{\theta}_{\theta}$, which in this particular case
787: ($T^0_0 - T^r_r = 0$), have singular points. A better choice of $U$ would make $T^i_j$
788: well behaved, but we have an almost infinite choice of $U$, although tying $U$ to
789: the quantum mechanical current associated with our Schr\" odinger equation (\ref{schreq}) would
790: be best.
791:
792: Once we have a reasonable metric (taking into account the caveats we have mentioned)
793: we could, in principle calculate the Hawking radiation in this background, as well as the
794: effect of its back reaction on the quantum evolution of the shell. This is still a massive
795: undertaking, involving difficult numerical analysis.
796:
797: Our plan in this article is to attempt a simplified analysis, where we can use purely
798: analytic constructs (with the exception of a numerical solution to a simple first-order
799: ODE) to model the complete collapse scenario.
800:
801: The first element we need is some quantity to represent the shell evolution. We will
802: consider the expectation value of $\hat R$, $<\hat R> (\tau)$, to represent a classical
803: shell evolving with this $R(\tau)$. We have
804: \begin{eqnarray}
805: <\hat R>(\tau) = \frac{\lambda \ell \sqrt{M}}{1 - e^{-\lambda^2 \ell M}}
806: [\cos T {\rm erf}\, (\lambda \sqrt{\ell} \sqrt{M} \cos T) \nonumber \\
807: - \sin T e^{-\lambda^2
808: \ell M} i{\rm erf}\, (i\lambda \sqrt{\ell} \sqrt{M} \sin T)].
809: \end{eqnarray}
810: Figure 2 shows $<\hat R>(\tau)$ for $\lambda = 3$, $\ell = 0.18$, and $M = 5$.
811:
812: This expectation value begins at $<\hat R> \approx 1.20$, falls to a minimum and rises again
813: to the original value. In Fig. 2, the value of the classical horizon $r_H \approx 0.4$, is shown as a
814: dashed line, and $<\hat R>(\tau)$ dips below this value, and later (after a relatively
815: short proper time interval) rises above $r_H$ again. Of course, the amount of time
816: $\Delta t$ measured by an observer far from the black hole between the moment when
817: $<\hat R>(\tau)$ dips below the classical horizon and the moment when it reappears
818: is greater than infinity.\\
819:
820: The next element needed for the complete scenario is Hawking radiation.
821: While Hawking radiation has been calculated for 2 + 1 gravity \cite{lifort}
822: \cite{Hlee}, in the spirit of
823: our simplified calculations, we will use a quick estimate of the average
824: energy of particles emitted in Hawking radiation and of $T_H$, the Hawking
825: temperature. In the Appendix we calculate this energy in 3 + 1 gravity by
826: assuming that pair production occurs in regions of the size of the Compton
827: wavelength of the particle produced, $\lambda_c$, and that if a pair is
828: produced in a small region between $r_H$ and $\lambda_c$, one of the pair
829: may fall into the black hole, and the other appears as a real particle
830: with velocity $c$.
831:
832: \vspace{1cm}
833: \centerline{\scalebox{.4}{\includegraphics{fig2.eps}}}
834: \captionof{figure}{\small The quantity $<\hat R>(\tau)$ as a function of $\tau$ from $3r_H$ until it
835: returns to $3r_H$ again for $M = 5$. The dashed line shows the position of the
836: horizon $r_H = \ell \sqrt{M}$.}
837: \vspace{0.5cm}
838:
839: \noindent A simple Newtonian calculation of the energy at infinity
840: of this particle, assuming that it is created at $r_H + \lambda_c/2$ and
841: moves radially toward $r = \infty$, gives an energy that is almost equal
842: to the average particle energy in Hawking radiation.
843:
844: In 2 + 1 gravity there are several difficulties. One is that there is
845: no Newtonian limit to 2 + 1 gravity, and another is that there is a
846: cosmological constant. Since, in the Appendix, we use Newtonian
847: calculations, we can do the same here (using two-dimensional Newtonian
848: gravity). For the cosmological constant we can use the Newtonian
849: cosmological constant introduced by Seeliger and Neumann \cite{seenue} at the
850: end of the nineteenth century.
851:
852: The two-dimensional Newtonian equation of motion for the radial
853: motion of a particle of mass $m$ with a cosmological constant in
854: the spherical field of a point mass $M$ is
855: \begin{equation}
856: m\frac{d^2 r}{dt^2} = - \frac{G^{(2)} Mm}{r} - \frac{mc^2}{\ell^2}r,
857: \end{equation}
858: where $c^2/\ell^2$ is the Newtonian cosmological constant, which has
859: a first integral
860: \begin{equation}
861: E = \frac{m}{2} \left (\frac{dr}{dt}\right )^2 + G^{(2)} Mm \ln r +
862: \frac{mc^2}{2\ell^2}r^2.
863: \end{equation}
864: From Figure 3 we have $r_H = \ell \sqrt{G^{(2)} M/c^2}$, and $\lambda_c =
865: \hbar/mc$,
866:
867: \vspace{1cm}
868: \centerline{\scalebox{.3}{\includegraphics{fig3.eps}}}
869: \captionof{figure}{\small The circles showing the horizon $r_H$ and the thin region between $r_H$ and $r_H + \lambda_c$,
870: $\lambda_c$ the Compton wavelength of a particle of mass $m$.}
871: \vspace{0.5cm}
872:
873: \noindent and the energy of a particle moving outward at velocity
874: $c$ at a position $r_H + \lambda_c/2$ is
875: \begin{equation}
876: E = \frac{mc^2}{2} + G^{(2)}Mm \ln \left (\ell \sqrt{\frac {G^{(2)}M}{c^2}}
877: + \frac {\lambda_c}{2}\right ) + \frac{1}{\ell^2} \left (\frac{\ell \sqrt{G^{(2)}
878: M}}{c^2} + \frac{\lambda_c}{2}\right )^2 mc^2.
879: \end{equation}
880: For $\lambda_c << r_H$,
881: \begin{equation}
882: E \approx \frac{mc^2}{2} + G^{(2)} Mm \left [1 + \ln \left
883: (\ell \sqrt{\frac{G^{(2}) M}{c^2}} \right ) \right ] + \frac{3}{4}
884: \hbar \frac{\sqrt{G^{(2)} M}}{\ell}.
885: \end{equation}
886:
887: The first term is not zero as in the 3 + 1 problem, but the last term
888: can be taken to be the average energy of Hawking radiation particles.
889: This quantity is close to the average energy calculated in \cite{lifort}
890: \cite{Hlee}, $E =
891: \hbar \sqrt{G^{(2)} M}/2\pi \ell$, and for our approximation we will
892: take it as the average energy. We can now define the Hawking temperature
893: ($k_B$ the Boltzmann constant),
894: \begin{equation}
895: T_H = \frac{3\hbar}{4k_B} \frac{\sqrt{G^{(2)} M}}{\ell}.
896: \end{equation}
897:
898: The effect of back reaction can be calculated (using the
899: two-dimensional Stefan-Boltzmann equation) as in the Appendix, and
900: leads to a rate of mass evaporation of the black hole. The black hole can be treated
901: as a black body of radius $r_H$ with a surface {\it length}, $2\pi r_H$,
902: where
903: \begin{equation}
904: \frac{dE}{dt} = - \zeta (3) \left (\frac{27}{32}\right ) \frac{\hbar^2 c^2}{\ell^2}
905: \left (\frac{G^{(2)} M}{c^2} \right ),
906: \end{equation}
907: and the Schwarzschild mass decreases $dM/dt = (dE/dt)/c^2$, and, since
908: $\zeta (3) (27/32)$ $\approx 1$, we have
909: \begin{equation}
910: \frac{dM}{dt} \approx -\frac{\hbar}{\ell^2} \left (\frac{G^{(2)} M}{c^2} \right )^2 =
911: -\frac{M_{\rm Pl} c}{L_{\rm Pl}}\left (\frac{L_{\rm Pl}}{\ell} \right )^2
912: \left (\frac{M}{M_{\rm Pl}} \right )^2,
913: \end{equation}
914: or, in our units,
915: \begin{equation}
916: \frac{dM}{dt} = -\frac{M^2}{\ell^2}.\label{dmdt}
917: \end{equation}
918: this can be solved for $M(t)$ as
919: \begin{equation}
920: M(t) = \frac{M(0)}{\frac{M(0)}{\ell^2} t + 1},\label{m(t)}
921: \end{equation}
922: assuming the $M(t)$ at $t = 0$ is $M(0)$.
923:
924: With these elements we can consider the complete collapse scenario. We
925: have to ask whether the classical shell represented by $<\hat R>(\tau)$ can fall
926: below the classical horizon radius $r_H = \ell \sqrt{M(0)}$ and return through
927: the smaller horizon at $r_H = \ell \sqrt{M(t)}$ at some later, finite, $t$-time.
928: Naively, one might think that the falling horizon radius will meet the rising
929: shell that has passed through its minimum radius and is rebounding, ``sooner''
930: than it would have in the static case. If we were to consider that we have
931: been using some sort of ``Newtonian'' time, which is unique, this would
932: always happen. However, we are trying to model relativity, where $t$
933: should be observer time at large $r$ which is not the same as $\tau$,
934: the proper time on the shell.
935:
936: Classically, we can find a relation between $t$ and $\tau$ by using the
937: metric (\ref{metrc}) with $\theta = \theta_0 = $ constant., and $r = R(\tau)$,
938: \begin{equation}
939: -d\tau^2 = -\left (-M + \frac{R^2}{\ell^2} \right ) dt^2 + \frac{\dot R^2}
940: {-M + \frac{R^2}{\ell^2}} d\tau^2,
941: \end{equation}
942: or
943: \begin{equation}
944: \frac{dt}{d\tau} = \frac{\sqrt{\dot R^2 + R^2/\ell^2 - M}}{\left |
945: -M + \frac{R^2}{\ell^2} \right |}.\label{dtdtau}
946: \end{equation}
947: This expression has the problem that the denominator is zero whenever
948: $R = \ell \sqrt{M}$, so when the shell crosses the horizon (either descending
949: or ascending), $dt/d\tau$
950: becomes singular, the origin of the fact that $t \rightarrow +\infty$ before
951: the rebounding shell can exit the horizon.
952:
953: We want to use Eq. (\ref{dtdtau}) modified by quantum considerations with $M = M(t)$,
954: $M(t)$ given by (\ref{m(t)}). Since $\dot R^2 + R^2/\ell^2 = 2E$, we will take this $2E$
955: to be $(2n + 1)/\ell$. We now want to use $R(\tau) \rightarrow <\hat R>(\tau)$.
956: Unfortunately, the denominator of (\ref{dtdtau}) still becomes zero at $<\hat R>(\tau) =
957: \ell \sqrt{M(t)}$, and the relation between $t$ and $\tau$ is still singular at
958: this point. Arguing that the quantum uncertainty in the shell position
959: makes the exact position $<\hat R>(\tau)$ unacceptable in our calculation
960: we can try to model this uncertainty by using $<\hat R^2> [M(t), \tau]$ in
961: our expression for $dt/t\tau$. Finally, we will try to solve the simple
962: differential equation
963: \begin{equation}
964: \frac{dt}{d\tau} = \frac{\sqrt{\frac{2n + 1}{\ell} - M(t)}}{\left |-M(t) +
965: \frac{<\hat R^2 >[M(t), \tau]}{\ell^2}\right |},\label{dt2}
966: \end{equation}
967: with $M(t)$ given by (\ref{m(t)}). This equation is a nonlinear, first-order differential
968: equation that must be solved numerically.
969:
970: Our complete collapse scenario will now be:\\
971:
972: 1) Collapse from $R = \lambda \ell \sqrt{M(0)}$ to the horizon formation
973: point $R = \ell \sqrt{M(0)}$, with essentially no Hawking radiation, since
974: the horizon is necessary for our calculation of $M(t)$.\\
975:
976: 2) From horizon formation to the point where
977: $<\hat R>(\tau)$ is again above the horizon, we solve (\ref{dt2}) for $t(\tau)$ and
978: $M[t(\tau)]$, taking $t = 0$ at the point where the ring crosses the
979: horizon.\\
980:
981: 3) Study $<\hat R>(\tau)$ to see if, on the rebound, it crosses the
982: horizon in finite $t$-time.\\
983:
984: \noindent If this scenario occurs, we will have to discuss the meaning of
985: the ``quantum remnant'' we have.
986:
987: \vspace{0.5cm}
988: \centerline{\scalebox{.35}{\includegraphics{fig4.eps}}}
989: \captionof{figure}{\small For $M(0) = 5$, the time $t$ at large $r$ as a function of $\tau$, the proper time on the shell.}
990:
991: We will now take $M(0) = 5$ and $n = 3$ ($n$ must be odd), a somewhat
992: arbitrary choice. Note that for $n = 3$ the maximum $M(0)$ for (\ref{dtdtau})
993: to make sense is $\approx 38.9$. For $\lambda = 3$, $\ell = 0.18$, we
994: can calculate $t(\tau)$ numerically. This was done using Maple, from
995: $\tau_0$ where the shell falls below the
996: horizon, to $\tau_1$ where it would
997: reemerge classically with $M(0) =$ constant $= 5$. The answer is given in Figure 4.
998: The use of $<\hat R^2>$ makes $t$ a regular function of $\tau$ in this entire
999: region. Figure 5 shows the horizon size, $\ell \sqrt{M(t)}$.
1000:
1001:
1002: \vspace{1cm}
1003: \centerline{\scalebox{.45}{\includegraphics{fig5.eps}}}
1004: \captionof{figure}{\small For $M(F0) = 5$, the horizon radius, $r_H = \ell \sqrt{M(\tau)}$ as a function of $\tau$.}
1005: \vspace{0.5cm}
1006: \vskip 15 pt
1007:
1008: \noindent Figure 6
1009: shows $<\hat R>(\tau)$ and the horizon size $\ell\sqrt{M[t(\tau)]}$ in this
1010: region. We can see that $<\hat R>(\tau)$ would never cross the horizon,
1011: and we can interpret this to mean that Hawking radiation would change $M(t)$
1012: so quickly that no horizon would form.
1013:
1014: If we now take $M(0) = 10$, we can see from Figure 7 that the form of $t(\tau)$
1015: does not change very much. Figure 8 shows $\ell \sqrt{M[t(\tau)]}$, and
1016: Figure 9 shows both $<\hat R>(\tau)$ and $\ell \sqrt{M[t(\tau)]}$.
1017: In this case, $<\hat R>(\tau)$ still does not fall below the horizon, but
1018: a more detailed study of $<\hat R>(\tau)$ near the initial value of $\tau$
1019: shows that the difference between $<\hat R>(\tau)$ and $r_H (\tau)$ is only
1020: of the order of $10^{-5}$, so the shell grazes $r_H(\tau)$.
1021:
1022: \vspace{1cm}
1023: \centerline{\scalebox{.4}{\includegraphics{fig6.eps}}}
1024: \captionof{figure}{\small For $M(0) = 5$, the horizon radius $r_H$ (solid line), and the expectation value of the shell
1025: radius $<\hat R>[M\{t(\tau)\}]$ (dashed line) as functions of $\tau$.}
1026: \vspace{1cm}
1027: \centerline{\scalebox{.4}{\includegraphics{fig7.eps}}}
1028: \captionof{figure}{\small For $M(0) = 10$, the time $t$ at large $r$ as a function of $\tau$, the proper time on the shell.}
1029: \vspace{0.5cm}
1030: \centerline{\scalebox{.4}{\includegraphics{fig8.eps}}}
1031: \captionof{figure}{\small For $M(0) = 5$, the horizon radius $r_H$ (solid line), and the expectation value of the shell
1032: radius $<\hat R>[M\{t(\tau)\}]$ (dashed line) as functions of $\tau$.}
1033: \vspace{0.5cm}
1034: \centerline{\scalebox{.4}{\includegraphics{fig9.eps}}}
1035: \captionof{figure}{\small For $M(0) = 10$, the horizon radius $r_H$ (solid line), and the expectation value of the shell
1036: radius $<\hat R>[M\{t(\tau)\}]$ (dashed line) as functions of $\tau$.}
1037: \vspace{0.5cm}
1038:
1039: If we now take $M(0) = 20$, we can see from Fig. 10 that the form of $t(\tau)$
1040: still does not change very much. Figure 11 shows $\ell \sqrt{M[t(\tau)]}$, and
1041: Figure 12 shows both $<\hat R>(\tau)$ and $\ell \sqrt{M[t(\tau)]}$ from
1042: $\tau \approx 0.22139$.
1043:
1044: \vspace{0.25cm}
1045: \centerline{\scalebox{.4}{\includegraphics{fig10.eps}}}
1046: \captionof{figure}{\small For $M(0) = 20$, the time $t$ at large $r$ as a function of $\tau$, the proper time on the shell.}
1047:
1048: \centerline{\scalebox{.4}{\includegraphics{fig11.eps}}}
1049: \captionof{figure}{\small For $M(0) = 20$, the horizon radius, $r_H = \ell \sqrt{M(\tau)}$ as a function of $\tau$.}
1050:
1051: \centerline{\scalebox{.4}{\includegraphics{fig12.eps}}}
1052: \captionof{figure}{\small For $M(0) = 20$, the horizon radius $r_H$ (solid line), and the expectation value of the shell
1053: radius $<\hat R>[M\{t(\tau)\}]$ (dashed line) as functions of $\tau$.}
1054: \vspace{20pt}
1055:
1056: \noindent Figure 13 shows ten times the difference between $<\hat R>[M\{t(\tau)\}]$ and $r_H(\tau)$,
1057: making the dip below $r_H$ more obvious.
1058:
1059: \vspace{0.5cm}
1060: \centerline{\scalebox{.35}{\includegraphics{fig13.eps}}}
1061: \captionof{figure}{\small For $M(0) = 20$, ten times $<\hat R>(\tau) - r_H(\tau)$, showing the dip of the expectation value
1062: below $r_H$ and its return above $r_H$.}
1063: \vspace{0.5cm}
1064:
1065: We can see that the shell falls below the horizon at $\tau \approx \tau_0$, $\approx 0.22139$ and
1066: reemerges at $\tau \approx 0.22241$, at which point Hawking radiation ceases.
1067: Since $t$ is a regular function of $\tau$ in this region, $t$ at reemergence
1068: is finite. We can interpret this scenario as loosely defining a ``dynamic remnant.''
1069: Many calculations in 3 + 1 gravity of Hawking radiation back reaction assume
1070: an eternal, static or stationary black hole, and that any remnant would be some sort
1071: of static or stationary construct. In our case, the fact that our quantum shell
1072: ``rebounds,'' seems to mean that our ``remnant'' would be an expanding quantum shell.
1073:
1074: The important point here is that there is mass loss during the Hawking radiation
1075: epoch is $\approx 11$, or 55\%. For $M(0) = 30$, the behavior is
1076: similar, with mass losses of $\approx 20$, or 67\% but the loss is not 100\%.
1077: As $M(0) \rightarrow \infty$, the mass loss grows, but it should never be 100\%.
1078: In the 3 + 1 case, since $dM/dt$ decreases with growing $M$, we might expect the mass
1079: loss to decrease.
1080:
1081: Our scenario is the collapse below a horizon, mass loss due to Hawking radiation, and
1082: the reappearance of the shell in expansion, but with a reduced mass. We can consider
1083: this a ``dynamic remnant.'' There are a number of problems with this calculation beyond
1084: its rough nature. One is that we have assumed in our Hawking radiation calculation that
1085: the Compton wavelength, $\lambda_c$, is much smaller that $r_H$. However, our units show
1086: that $M$ is of the order of a few Planck masses, so $\lambda_c$ for any reasonable
1087: particle is very much larger than $r_H$. We would have to drastically modify our
1088: Hawking radiation approximation in this case. This problem reflects the well-known
1089: difficulty of calculating Hawking radiation for particle masses above the horizon
1090: radius (see \cite{koch} and references therein).
1091:
1092: Another problem is that we can still have information loss. The existence of a remnant
1093: does not necessarily solve all problems. There is always the necessity for an
1094: infinite number of states which allows for the unbounded information content inherited
1095: from the original state.
1096:
1097: \section{Conclusions and suggestions for further research}
1098:
1099: While the above results are suggestive, there are several caveats, and
1100: one should be careful in interpreting the results. One problem is that
1101: Eq. (\ref{dmdt}) has the rate of mass loss {\it increasing} as $M$ increases, while
1102: in 3 + 1 gravity it {\it decreases}, so even if we accept the results, they
1103: will be numerically quite different from those in 3 + 1 gravity. Of course,
1104: our results are, at best, crude, due to the drastic simplifications we have
1105: made. Perhaps the most sensitive choices we have made are in the equation
1106: for $dt/d\tau$. If we were to use the approximation of the classical $T^{ij}$
1107: constructed from $\psi^* \psi$, some of these difficulties could be sidestepped,
1108: but there will still be a difficult problem in that the proper time $\tau$ is
1109: valid only for {\it one} shell, and we would somehow have to change the
1110: problem to account for different proper times on different shells. One advantage
1111: we have in our simplified calculation is that we can argue that our proper time
1112: is valid for the {\it one} shell we have.
1113:
1114: Another time problem is that we are using (essentially) Schwarzschild time for
1115: $t$, while a coordinate system such as Kruskal might be a better
1116: choice. However, the coordinate change from Schwarzschild contains $M$,
1117: and a changing $M$ makes the coordinate change difficult. In the classical
1118: $T^{ij}$ problem, we would have to {\it invent} the equivalent of Kruskal
1119: coordinates.
1120:
1121: The one advantage of 2 + 1 gravity is that the quantum collapse problem is
1122: exactly solvable, but numerical work in 3 + 1 would be more believable.
1123:
1124: However, we can make several suggestions for further research:
1125: \vskip 20 pt
1126:
1127: 1) Do the $T^{ij}$ problem, with due attention to the difficulties with
1128: proper time.\\
1129:
1130: 2) Return to 3 + 1 gravity and try to do a similar simplified calculation
1131: to the one attempted here.\\
1132:
1133: 3)Redo the calculation with due care given to the fact that, in general
1134: $\lambda_c >> r_H$.
1135: \vskip 20 pt
1136:
1137: \noindent Each of these would require numerical solutions.
1138:
1139: \section*{Appendix}
1140:
1141: We want to do the same simplified calculation of Hawking radiation
1142: energy that we did in Sec. 4 for the 2 + 1 black hole. If we use
1143: Fig. 3, taking the circles to be spheres, we have a thin shell between
1144: $r_H$ and $r_H + \lambda_c$ where we can have the production of virtual
1145: particles with mass $m$, with one of the pair falling into the hole,
1146: and the other becoming real with outward radial velocity $c$. The
1147: (constant) Newtonian energy of the particle, if produced at $r =
1148: R_H + \lambda_c/2$, is
1149: \begin{equation}
1150: E = \frac{mc^2}{2} - \frac{GMm}{r_H + \lambda_c/2},
1151: \end{equation}
1152: and if $\lambda_c << r_H$,
1153: \begin{equation}
1154: E \approx \frac{mc^2}{2} - \frac{GMm}{r_H}\left (1 - \frac{\lambda_c}{2r_H}\right ),
1155: \end{equation}
1156: \begin{equation}
1157: = \frac{mc^2}{2} - \frac{mc^2}{2} + \frac{\hbar c^3}{8GM} = \frac{\hbar c^3}{8GM}.
1158: \end{equation}
1159: The exact average energy of Hawking radiation particles at infinity is
1160: \begin{equation}
1161: \left( \frac{\Gamma (4) \zeta (4)}{\pi \Gamma(3) \zeta (3)}\right ) \frac{\hbar c^3}{8GM},
1162: \end{equation}
1163: which is within 14\% of our simplified result.
1164:
1165: We can now define a Hawking temperature
1166: \begin{equation}
1167: T_H = \frac{\hbar c^3}{8GM k_B},
1168: \end{equation}
1169: and treating the black hole as a black body with radius $r_H$, we have an
1170: energy loss due to Hawking radiation as ($\sigma$ the Stefan-Boltzmann constant)
1171: \begin{equation}
1172: \frac{dE}{dt} = -\sigma (4\pi r_H^2) T_H^4,
1173: \end{equation}
1174: \begin{equation}
1175: = -\frac{\pi^3}{15360}\left (\frac{\hbar c^6}{G^2}\right ) \left(\frac{1}{M^2}\right).
1176: \end{equation}
1177: The mass loss equation is
1178: \begin{equation}
1179: \frac{dM}{dt} = -\frac{\pi^3}{15360}\left (\frac{\hbar c^4}{G^2}\right ) \left(\frac{1}{M^2}\right).
1180: \end{equation}
1181: Note that $dM/dt$ {\it decreases} with increasing $M$, rather than growing
1182: as in the 2 + 1 case. Also, $T_H$ is {\it independent} of $m$, so, formally, we
1183: can take $m \rightarrow 0$ to assume the production of massless particles.
1184:
1185: Notice that we have assumed $r_H << \lambda_c$, and if we wanted to use
1186: our simplified calculation for Planck-sized black holes, we would have to
1187: change our approximation.
1188:
1189:
1190:
1191: \begin{thebibliography}{99}
1192:
1193: \bibitem{hawk} S. W. Hawking, Comm. Math. Phys. {\bf 43}, 199 (1975);
1194: Phys. Rev. {\bf 14}, 2460 (1976).
1195:
1196: \bibitem{meleo} L. Ort{\'\i}z and M. Ryan, in press, General Relativity and
1197: Gravitation.
1198:
1199: \bibitem{stipel} Y. Peleg and A. Steif, Phys. Rev. D {\bf 51}, 3992 (1995).
1200:
1201: \bibitem{Isra} W. Israel, Nuovo Cimento B {\bf 44}, 1 (1966).
1202:
1203: \bibitem{crisol} J. Cris\' ostomo and R. Olea, Phys. Rev. D {\bf 69}, 104023 (2004).
1204:
1205: \bibitem{haj1} P. H\'aj{\'\i}\v cek, Commun. Math. Phys. {\bf 150}, 545
1206: (1992).
1207:
1208: \bibitem{cruz} A. Corichi. G. Cruz, A. Minzoni, P. Padilla, M. Rosenbaum,
1209: M. Ryan, N. Smyth and T Vukasinac, Phys. Rev. D {\bf 65}, 064006 (2002).
1210:
1211: \bibitem{hkk} P. H\'aj{\'\i}\v cek, B. Kay and K. Kucha\v r, Phys. Rev. D
1212: {\bf 46} (1992).
1213:
1214: \bibitem{hyk} P. H\'aj{\'\i}\v cek and K. Kucha\v r, in preparation.
1215:
1216: \bibitem{koch} B. Koch, M. Bleicher and S. Hossenfelder, J. High. En. Phys. 10-053 (2005).
1217:
1218: \bibitem{stromtom} S. Giddings and S. Thomas, Phys. Rev D {\bf 65}, 056010 (2002).
1219:
1220: \bibitem{hajki} P. H\'aj{\'\i}\v cek and J. Kijowsky, Phys. Rev. D {\bf 62}, 044025 (2000).
1221:
1222: \bibitem{vinart} M. Ryan, Class. Quant. Grav. {\bf 21}, S323 (2004).
1223:
1224: \bibitem{us} G. Cruz, K. Kucha\v r, A. Minzoni, M. Rosenbaum, M. Ryan and
1225: N. Smyth, in preparation.
1226:
1227: \bibitem{strom} The literature is vast. See the list of references in
1228: K. Kucha\v r, Int. J. Theor. Phys. {\bf 38}, 1033 (1999).
1229:
1230: \bibitem{ku2} K. Kucha\v r, Int. J. Theor. Phys. {\bf 38}, 1033 (1992).
1231:
1232: \bibitem{tatj} G. Cruz, A. Minzoni, M. Rosenbaum, M. Ryan, N. Smyth and
1233: T. Vukasinac, Rev. Mex. Fis. {\bf 49} (Suppl. 2), 122 (2003).
1234:
1235: \bibitem{BTZ} M. Ba\~ nados, C. Teitelboim, and J. Zanelli,
1236: Phys. Rev. Lett. {\bf 69}, 1849 (1992).
1237:
1238: \bibitem{BTZ2} M. Ba\~nados, M. Henneaux, C. Teitelboim, and J. Zanelli,
1239: Phys. Rev. D {\bf 48}, 1506 (1993).
1240:
1241: \bibitem{carl} S. Carlip {\it Quantum Gravity in 2 + 1 Dimensions} (Cambridge, Cambridge, 1998).
1242:
1243: \bibitem{sing} J. L. Synge, {\it Relativity: The general theory} (North-Holland, Amsterdam, 1960).
1244:
1245: \bibitem{lifort} G. Lifschytz and M. Ortiz, Phys. Rev. D {\bf 49}, 1929 (1994).
1246:
1247: \bibitem{Hlee} S. Hyun, G. H. Lee, and J. H. Yee, Phys. Lett. B {\bf 322}, 182 (1994).
1248:
1249: \bibitem{seenue} H. Seeliger, Astronomische Nachrichten {\bf 137}, 129 (1895); C. G.
1250: Neumann, {\it \" Uber das Newtonische Prinzip der Fernwirkung} (Leipzig, 1895).
1251:
1252: \end{thebibliography}
1253: \end{document}
1254:
1255:
1256:
1257:
1258: