1:
2: \documentstyle[prl,aps,twocolumn,latexsym,amsfonts,epsf,graphicx]{revtex}
3:
4: \newcommand{\nn}{\nonumber\\}
5: \newcommand{\ul}{\underline}
6: \newcommand{\f}[1]{\mbox{\boldmath$#1$}}
7: \newcommand{\fk}[1]{\mbox{\boldmath$\scriptstyle#1$}}
8: \newcommand{\vau}{\mbox{\boldmath$v$}}
9: \newcommand{\na}{\mbox{\boldmath$\nabla$}}
10: \newcommand{\bea}{\begin{eqnarray}}
11: \newcommand{\ea}{\end{eqnarray}}
12: \newcommand{\eea}{\end{eqnarray}}
13: \newcommand{\ord}{{\cal O}}
14: \newcommand{\sumint}[1]
15: {\begin{array}{c}
16: \\
17: {{\textstyle\sum}\hspace{-1.1em}{\displaystyle\int}}\\
18: {\scriptstyle{#1}}
19: \end{array}}
20:
21: \begin{document}
22:
23: \wideabs{
24: \title{A relativistic toy model for back-reaction}
25: \author{G\"unter Plunien, Marcus Ruser, and Ralf Sch\"utzhold}
26: \address{Institut f\"ur Theoretische Physik,
27: Technische Universit\"at Dresden,
28: 01062 Dresden, Germany}
29: \date{\today}
30: \maketitle
31: \begin{abstract}
32: We consider a quantized massless and minimally coupled scalar field
33: on a circular closed string with a time-dependent radius $R(t)$, whose
34: undisturbed dynamics is governed by the Nambu-Goto action.
35: %
36: Within the semi-classical treatment, the back-reaction of the quantum
37: field onto the string dynamics is taken into account in terms of the
38: renormalized expectation value of the energy-momentum tensor including
39: the trace anomaly.
40: %
41: The results indicate that the back-reaction could prevent the collapse
42: of the circle $R\downarrow0$ -- however, the semi-classical picture
43: fails to describe the string dynamics at the turning point
44: (i.e., possible bounce) at finite values of $R$ and $\dot R$.
45: %
46: The fate of the closed string after that point (e.g., oscillation or
47: eternal acceleration) cannot be determined within the semi-classical
48: picture and thus probably requires the full quantum treatment.
49: \\
50: PACS:
51: 04.62.+v, % Quantum field theory in curved spacetime
52: 03.70.+k, % Theory of quantized fields
53: 11.15.Kc, % Classical and semi-classical techniques, General theory of fields
54: 04.60.-m. % Quantum gravity
55: %12.10.-g % Unified field theories and models
56: %12.60.Jv % Supersymmetric models
57: %12.20.Ds % Specific calculations, Quantum electrodynamics
58: %03.65.Ca % Formalism, Quantum mechanics, field theories, and SRT
59: \end{abstract}
60: }
61:
62: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
63: \section{Introduction}
64: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
65:
66: Without knowledge of the underlying theory including quantum gravity
67: the back-reaction of quantum fields onto the geometry is usually
68: described by the semi-classical Einstein equations
69: %
70: \bea
71: \label{einstein}
72: {\cal R}_{\mu\nu}-\frac12{g}_{\mu\nu}\,{\cal R}
73: =-\kappa\,\langle\hat T_{\mu\nu}\rangle_{\rm ren}
74: \,.
75: \ea
76: %
77: Although intrinsically incomplete, this semi-classical approach might
78: -- after proper renormalization of
79: $\langle\hat T_{\mu\nu}\rangle_{\rm ren}$ is employed --
80: shed light onto the following questions:
81: %several interesting questions, such as:
82: %
83: \begin{itemize}
84: \item
85: Could the back-reaction of quantum fields prevent the collapse
86: (i.e., singularity) of the space-time?
87: \item
88: Under which circumstances does the semi-classical picture apply and
89: when does it fail?
90: \end{itemize}
91: %
92:
93: As a model for gauge field theories (vector $A_\mu$ and spinor $\Psi$ fields)
94: in a 3+1 dimensional gravitational background we shall consider the
95: also conformally invariant theory of a massless and minimally coupled
96: scalar field $\Phi$ in 1+1 dimensions.
97: Since the Einstein tensor ${\cal R}_{\mu\nu}-{g}_{\mu\nu}\,{\cal R}/2$
98: vanishes identically in 1+1 dimensions,
99: one has to start with an alternative geometric action leading to a
100: preferably simple dynamics which still preserves major aspects such
101: as relativistic invariance as well as generic features of
102: higher-dimensional space-times.
103:
104: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
105: \section{Toy model}
106: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
107:
108: Let us consider a 1+1 dimensional string whose dynamics is governed by
109: the Nambu-Goto action \cite{nambu}
110: %
111: \bea
112: \label{string}
113: {\mathfrak A}_{\rm string}=-\frac{\sigma}{2}\int d^2x\,\sqrt{-{g}}
114: \,,
115: \ea
116: %
117: with $\sigma$ denoting the string tension.
118: For simplicity we shall assume rotational symmetry, i.e., a circular
119: closed string characterized by its time-dependent radius $R$ as the
120: only degree of freedom of the geometry.
121: Adopting the usual time-gauge (see, e.g. \cite{nambu,vilenkin}),
122: the induced metric reads ($\hbar=c=1$ throughout)
123: %
124: \bea
125: \label{time-gauge}
126: ds^2=\left[1-(\partial_tR)^2\right]\,dt^2-R^2d\varphi^2
127: \,,
128: \ea
129: %
130: where $t$ is the ''laboratory'' time and $\varphi$ the angular
131: coordinate.
132: As one can immediately see, the action in Eq.~(\ref{string}) is fully
133: relativistic (in contrast to any {\em ad hoc} kinetic terms like
134: $\dot R^2=(\partial_t R)^2$).
135:
136: Left alone -- i.e., without additional forces -- the free dynamics of
137: the circle governed by $\delta{\mathfrak A}_{\rm string}/\delta R=0$,
138: i.e.,
139: %
140: \bea
141: R\ddot{R}+(1-\dot{R}^2)=0
142: \label{free EOM}
143: \,,
144: \ea
145: %
146: with the initial conditions $R(0)=R_0$, $\dot{R}(0)=0$, for example,
147: leads to a collapse $R\downarrow0$ and $\dot R\uparrow1$
148: (i.e., a singularity)
149: after a finite time caused by the string tension $\sigma>0$.
150:
151: Let us consider this circle (time-dependent background metric) being
152: endowed with a real scalar field $\Phi$ such that the total action
153: reads (see, e.g., \cite{diplom})
154: %
155: \bea
156: \label{total-action}
157: {\mathfrak A}=\frac{1}{2}\int d^2x\,\sqrt{-{g}}
158: \left(\partial_\mu\Phi\,\partial^\mu\Phi-\sigma\right)
159: \,.
160: \ea
161: %
162: The (classical) equations of motion for $R(t)$ can be derived by
163: variation $\delta{\mathfrak A}/\delta R=0$
164: %
165: \begin{equation}
166: \label{eom}
167: \sqrt{1-\dot{R}^2}\left[\frac{\sigma}{2}+T^{1}_{1}
168: \right]+\frac{d}{dt}\left[\frac{R\dot{R}}{\sqrt{1-\dot{R}^2}}
169: \left(\frac{\sigma}{2}+T^{0}_{0}\right)\right]=0
170: \,,
171: \end{equation}
172: %
173: with the energy-momentum tensor of the scalar field
174: %
175: \bea
176: \label{Tmunu}
177: T_{\mu\nu}=\partial_\mu\Phi\,\partial_\nu\Phi-\frac12\,
178: {g}_{\mu\nu}\partial_\rho\Phi\,\partial^\rho\Phi
179: \,.
180: \ea
181: %
182: The equation of motion for scalar field
183: $\delta{\mathfrak A}/\delta\Phi=0$
184: follows as
185: %
186: \bea
187: \Box\Phi=\frac{1}{\sqrt{-g}}\,\partial_\mu\left(
188: \sqrt{-g}\,g^{\mu\nu}\,\partial_\nu\,\Phi\right)=0
189: \, ,
190: \ea
191: %
192: where $\Box$ denotes the d'Alembertian with respect to the metric in
193: Eq.~(\ref{time-gauge}).
194:
195: The first integral of the equations of motion (\ref{eom}) and
196: $\Box\Phi=0$ corresponds to the (conserved) total energy
197: %
198: \begin{eqnarray}
199: \label{energy}
200: E=\frac{2\pi\,R}{\sqrt{1-\dot{R}^2}}\left[\frac{\sigma}{2}+
201: T^{0}_{0}\right]
202: \,.
203: \end{eqnarray}
204: %
205: %Note that this conserved energy is the total Hamiltonian of the system
206: %$H(\Pi_\Phi,\Phi,P_R,R)$ -- but not just the sum of the Hamiltonians
207: %for $R$ and $\Phi$ separately (i.e., the Routhian).
208:
209: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
210: \section{Classical case}
211: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
212:
213: For calculating the dynamics of the field $\Phi$ it is convenient to
214: introduce another time coordinate via the transformation
215: %
216: \bea
217: \label{trafo}
218: \tau=\int dt\,\frac{\sqrt{1-(\partial_tR)^2}}{R(t)}
219: \,,
220: \ea
221: %
222: leading to conformally flat metric
223: %
224: \bea
225: \label{conformal}
226: ds^2=R^2(\tau)[d\tau^2-d\varphi^2]
227: \,.
228: \ea
229: %
230: Note that the insertion of this metric into the Nambu-Goto action in
231: Eq.~(\ref{string}) does not yield the same equations of motion for
232: $R(t)$ since Eq.~(\ref{trafo}) represents a non-algebraic transformation.
233:
234: Let us first consider the classical back-reaction of a spatially
235: homogeneous field $\partial\Phi/\partial\varphi=0$ as a solution of
236: the field equation $\Box\Phi=\partial^2\Phi/\partial\tau^2=0$ which
237: simply reads $\Phi={\mathfrak C}_\Phi\tau$ leading to the kinetic
238: term
239: %
240: \bea
241: \dot\Phi^2={\mathfrak C}_\Phi^2\,
242: \frac{1-\dot R^2}{R^2}
243: \,,
244: \ea
245: %
246: with respect to the time $t$.
247: Insertion into the total energy in Eq.~(\ref{energy}) yields
248: %
249: \begin{eqnarray}
250: \label{class energy}
251: E=\frac{2\pi\,R}{\sqrt{1-\dot{R}^2}}\left[\frac{\sigma}{2}+
252: \frac12\,\frac{{\mathfrak C}_\Phi^2}{R^2}
253: \right]
254: \,.
255: \end{eqnarray}
256: %
257: Since this energy becomes arbitrarily large for both, $R\downarrow0$
258: and $R\uparrow\infty$ (as well as for $\dot{R}^2\uparrow1$) the
259: presence of this (classical) scalar field solution prevents the
260: collapse $R\downarrow0$ for ${\mathfrak C}_\Phi\neq0$.
261: The dynamics is governed by
262: %
263: \bea
264: R\ddot{R}\left[R^2+R_E^2\right]+(1-\dot{R}^2)
265: \left[R^2-R_E^2 \right]=0
266: \,,
267: \ea
268: %
269: with $R_E^2={\mathfrak C}_\Phi^2/\sigma>0$ denoting the square of the
270: radius which minimizes the energy in Eq.~(\ref{class energy}) for
271: $\dot{R}=0$.
272: Initial conditions $R(t=0)\neq R_E$ or $\dot{R}(t=0)\neq0$ lead to an
273: eternal oscillation between $R_{\rm min}>0$ and $R_{\rm max}$ with
274: $R_{\rm min}<R_E<R_{\rm max}$ whereas for $R(t=0)=R_E$ and
275: $\dot{R}(t=0)=0$, the ring remains static forever.
276:
277: However, it should be emphasized here that a spatially homogeneous
278: solution such as $\partial\Phi/\partial\varphi=0$ -- which is also
279: called a zero-mode -- does usually not exist in higher-dimensional
280: situations with vector fields, for example (see also the next Section).
281: On the other hand, one would obtain the same result for a thermal
282: bath of the $\Phi$-field with $\langle E \rangle_\beta$ representing
283: the (classical) ensemble average of the energy for the inverse
284: temperature $\beta$.
285: The calculation of $\langle T^0_0 \rangle_\beta$ may proceed in
286: basically the same way as the derivation starting from
287: Eq.~(\ref{balance}) below, provided that one neglects all quantum
288: effects such as the trace anomaly.
289:
290: As we shall see in the next Section, the Casimir effect contributes to
291: the total energy in the same way, but with a negative sign.
292: Therefore, the above (classical) contribution and the induced
293: stabilization of the string can be negated by the (quantum) Casimir
294: effect, in particular since the amplitude of the field or the
295: temperature can decrease -- whereas the Casimir effect remains.
296: This observation motivates the consideration of the quantum field
297: effects.
298:
299: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
300: \section{Renormalization of $\langle\hat T_{\mu\nu}\rangle$ and trace anomaly}
301: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
302:
303: In order to investigate the back-reaction of the quantum field
304: $\hat\Phi$ in analogy to the semi-classical Einstein equations
305: (\ref{einstein}), we have to calculate the renormalized expectation
306: value of the energy-momentum tensor and insert it into the equation of
307: motion (\ref{eom}) for $R(t)$.
308: In both cases, the renormalization of $\langle\hat T_{\mu\nu}\rangle$,
309: i.e., the absorption of the divergences, requires a redefinition of
310: the involved coupling constants -- the cosmological constant $\Lambda$
311: and Newton's constant $\kappa$ in gravity, and the string tension
312: $\sigma$ in our model.
313:
314: Adopting the point-splitting renormalization procedure, we need the
315: two-point function $W$.
316: Fortunately, a massless and minimally coupled scalar field in 1+1
317: dimensions is conformally invariant and hence the two-point function
318: of the conformal vacuum (see, e.g., \cite{birrell}) in terms of the
319: conformal coordinates in Eq.~(\ref{conformal}) has the same for as in
320: flat space-time ($R=\rm const$)
321: %
322: \bea
323: \label{wightman}
324: \langle\hat\Phi(\tau,\varphi)\hat\Phi(\tau',\varphi')\rangle=
325: W(\tau-\tau',\varphi-\varphi')
326: \,.
327: \ea
328: %
329: In the coincidence limit, i.e., for $\tau\to\tau'$ and
330: $\varphi\to\varphi'$, the Wightman function $W$ behaves as
331: $\ln[(\tau-\tau')^2-(\varphi-\varphi')^2]$.
332:
333: Note that $W$ is only determined up to an additive constant --
334: in the usual 1+1 dimensional space-time ${\mathbb R}\times{\mathbb R}$,
335: this fact reflects the infra-red problem (in 1+1 dimensions); and, in
336: our case ${\mathbb R}\times{\mathbb S}_1$, this undetermined additive
337: constant corresponds to the zero mode $\partial\Phi/\partial\varphi=0$.
338:
339: However, zero modes do not have a zero-point energy
340: (the energy spectrum reaches zero) and decouple from the rest of the
341: modes (in our situation).
342: Therefore, we omit those modes and the related problems with their
343: quantization in the following.
344:
345: In any covariant regularization method in 1+1 dimensions
346: (where only one principal divergence exists),
347: for example point splitting after omitting direction dependent terms,
348: the regularized (but not yet renormalized) expectation value reads
349: (see also Appendix A)
350: %
351: \bea
352: \label{reg-T}
353: \langle\hat T_{\mu\nu}\rangle_{\rm bare}
354: =
355: \frac{1}{\epsilon}\,{g}_{\mu\nu}+
356: \langle\hat T_{\mu\nu}\rangle_{\rm ren}
357: \,,
358: \ea
359: %
360: where $\epsilon\to0$ is the regularization parameter
361: (e.g., the geodesic distance in point-splitting) and the remaining
362: $\langle\hat T_{\mu\nu}\rangle_{\rm ren}$ is finite.
363:
364: In the expressions for the energy (\ref{energy}) as well as the
365: equations of motion (\ref{eom}), the above divergence can be
366: completely absorbed by a renormalization of string tension via
367: %
368: \bea
369: \label{reg-sigma}
370: \sigma_{\rm bare}
371: =
372: -\frac{2}{\epsilon}+
373: \sigma_{\rm ren}
374: \,.
375: \ea
376: %
377: The question of whether $\sigma_{\rm ren}$ can acquire any dependence
378: on $R$ or $\dot R$ via the renormalization procedure (roughly similar
379: to the running coupling in QED, for example) will be answered below.
380:
381: Owing to the conformal invariance of the scalar field, the two-point
382: function of the conformal vacuum can be calculated easily in the
383: conformal metric, for example for a single scalar field with periodic
384: boundary conditions we obtain
385: %
386: \bea
387: \label{periodic}
388: &&
389: \langle\partial_\varphi\hat\Phi(\tau,\varphi)
390: \partial_{\varphi'}\hat\Phi(\tau,\varphi')\rangle
391: =
392: \langle\partial_\varphi\hat\Phi(t,\varphi)
393: \partial_{\varphi'}\hat\Phi(t,\varphi')\rangle
394: =
395: \nn
396: &&
397: =
398: \frac{1}{4\pi}\,\frac{1}{\cos(\varphi-\varphi')-1}
399: \,.
400: \ea
401: %
402: By means of this example, one can read off the symmetries
403: $\varphi\to\varphi+\varphi_0$ and $\varphi\to-\varphi$
404: which reflect the homogeneity and ${\mathbb Z}_2$-isotropy of the
405: underlying space-time.
406: These symmetries of the vacuum state and the geometry are inherited by
407: the (renormalized) expectation value of the energy-momentum tensor
408: (see, e.g., \cite{trace,birrell})
409: %
410: \bea
411: \label{flux}
412: \langle\hat T_0^1\rangle_{\rm ren}=\langle\hat T_1^0\rangle_{\rm ren}=0
413: \,,
414: \ea
415: %
416: i.e., no flux, as well as
417: %
418: \bea
419: \label{homo}
420: \partial_\varphi\langle\hat T^\mu_\nu\rangle_{\rm ren}=0
421: \,.
422: \ea
423: %
424:
425: Furthermore, one has to demand that the renormalization procedure of
426: $\langle\hat T^\mu_\nu\rangle_{\rm ren}$ respects the property of a
427: vanishing covariant divergence (see also Appendix A)
428: %
429: \bea
430: \label{balance}
431: \nabla_\mu\langle\hat T^\mu_\nu\rangle_{\rm ren}
432: &=&0\qquad\leadsto
433: \nn
434: \partial_\mu\left(\sqrt{-{g}}\,
435: \langle\hat T^\mu_\nu\rangle_{\rm ren}\right)
436: &=&
437: \frac{\sqrt{-{g}}}{2}\,\langle\hat T^{\alpha\beta}\rangle_{\rm ren}
438: \partial_\nu {g}_{\alpha\beta}
439: \,.
440: \ea
441: %
442: As it is well known \cite{trace}, one has to give up the classical
443: feature $T^\rho_\rho=0$ in this process, i.e.,
444: $\langle\hat T^\mu_\nu\rangle_{\rm ren}$ acquires an anomalous trace
445: during the renormalization
446: %
447: \bea
448: \label{anomaly}
449: \langle\hat T_\mu^\mu\rangle_{\rm ren}={\mathfrak C}_{\rm tr}\,{\cal R}
450: \,,
451: \ea
452: %
453: with ${\mathfrak C}_{\rm tr}$ denoting a constant related to the
454: central charge.
455: %
456: In 1+1 dimensions, the trace anomaly is completely determined by the
457: Ricci scalar (see, e.g., \cite{trace,birrell})
458: %
459: \bea
460: \label{ricci}
461: {\cal R}
462: =
463: 2\,\frac{\ddot{R}}{R(1-\dot{R}^2)^2}
464: =
465: 2\,\frac{R\,R''-(R')^2}{R^4}
466: \,,
467: \ea
468: %
469: with $\dot{R}=dR/dt$ and $R'=dR/d\tau$, respectively.
470:
471: Similar to the calculation of the Hawking radiation in 1+1 dimensions
472: \cite{trace} via the trace anomaly, we can integrate
473: Eq.~(\ref{balance}) with the proper boundary/initial conditions and
474: symmetries.
475: The case $\nu=1$ in Eq.~(\ref{balance}) is trivial and in the
476: remaining equation with $\nu=0$ -- when expressed in terms of the
477: conformal metric -- the trace anomaly acts somewhat similar to a
478: source
479: %
480: \bea
481: \label{source}
482: \partial_\tau(R^2\langle\hat T^0_0\rangle_{\rm ren})=
483: 2\,\frac{R\,R'\,R''-(R')^3}{R^3}\,{\mathfrak C}_{\rm tr}
484: \,.
485: \ea
486: %
487: The conformal invariance of the scalar field in 1+1 dimensions leading
488: to the absence of particle creation (conformal vacuum) manifests
489: itself in the fact that the r.h.s.\ of the above equation is a total
490: differential\footnote{If particles were created, the energy density
491: $\langle\hat T^0_0\rangle_{\rm ren}$ would depend on the history of
492: the space-time and not just on the values $R(t)$ and $\dot R(t)$ etc.,
493: at present.}.
494: Thus integrating Eq.~(\ref{source}) yields
495: %
496: \bea
497: \label{T00-conf}
498: \langle\hat T^0_0\rangle_{\rm ren}
499: ={\mathfrak C}_{\rm tr}\,\frac{(R')^2}{R^4}+
500: {\mathfrak C}_{\rm Cas}\,\frac{1}{R^2}
501: \,,
502: \ea
503: %
504: where ${\mathfrak C}_{\rm Cas}$ is {\em a priori} an integration constant.
505: However, if we consider a static circle $R=\rm const$, the only
506: contribution to the vacuum energy is the Casimir effect induced by the
507: compactness of the space (${\mathbb S}_1$).
508: In this case ${\mathfrak C}_{\rm Cas}/R^2$ is just the Casimir energy
509: density which allows us to determine the constant ${\mathfrak C}_{\rm Cas}$
510: and we keep this nomenclature also for the general time-dependent
511: situation $R'\neq0$.
512:
513: Transforming back to the laboratory time we obtain
514: (note that $\tau=\tau(t)\,\to\,t$ and hence $T^0_0\,\to\,T^0_0$)
515: %
516: \begin{equation}
517: \label{T00-lab}
518: \langle\hat T^0_0\rangle_{\rm ren}
519: =\frac{1}{R^2}\left[
520: \frac{\dot{R}^2}{1-\dot{R}^2}\;{\mathfrak C}_{\rm tr}+
521: {\mathfrak C}_{\rm Cas}\right]
522: \,.
523: \end{equation}
524: %
525: The remaining non-trivial component $\langle\hat T^1_1\rangle_{\rm ren}$
526: can be determined via the trace anomaly
527: %
528: \begin{equation}
529: \label{T11-lab}
530: \langle\hat T^1_1\rangle_{\rm ren}=
531: 2\;{\mathfrak C}_{\rm tr}\;\frac{\ddot{R}}{R(1-\dot{R}^2)^2}
532: -\langle\hat T^0_0\rangle_{\rm ren}
533: \,.
534: \end{equation}
535: %
536:
537: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
538: \section{Back-reaction}
539: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
540:
541: At this stage we are in the position to calculate the back-reaction in
542: analogy to
543: the semi-classical Einstein equations by inserting the renormalized
544: expectation values of the energy-momentum tensor
545: $T^\mu_\nu\,\to\,\langle\hat T^\mu_\nu\rangle_{\rm ren}$
546: in Eqs.~(\ref{T00-lab}) and (\ref{T11-lab}) into the equation of motion
547: (\ref{eom})
548: %
549: \begin{eqnarray}
550: \label{eom-ren}
551: R\ddot{R}\left(
552: \frac{\sigma_{\rm ren}}{2}R^2+{\mathfrak C}_{\rm Cas}
553: +
554: \frac{2+\dot{R}^2}{1-\dot{R}^2}\,
555: {\mathfrak C}_{\rm tr}
556: \right)
557: +
558: \nonumber\\
559: +
560: (1-\dot{R}^2)
561: \left(
562: \frac{\sigma_{\rm ren}}{2}R^2-{\mathfrak C}_{\rm Cas}
563: \right)
564: -\dot{R}^2{\mathfrak C}_{\rm tr}
565: =0
566: \,.
567: \end{eqnarray}
568: %
569: The consistency of the above equation -- assuming a constant
570: $\sigma_{\rm ren}$ -- with the conservation of the total
571: (renormalized) energy of the system
572: %
573: \begin{eqnarray}
574: \label{energy-ren}
575: E
576: &=&
577: \frac{2\pi\,R}{\sqrt{1-\dot{R}^2}}
578: \left(\frac{\sigma_{\rm ren}}{2}+
579: \langle\hat T^0_0\rangle_{\rm ren}
580: \right)
581: =
582: \frac{2\pi\,R}{\sqrt{1-\dot{R}^2}}
583: \times
584: \nn
585: &&\times
586: \left(
587: \frac{\sigma_{\rm ren}}{2}+
588: \frac{1}{R^2}\left[
589: \frac{\dot{R}^2}{1-\dot{R}^2}\;{\mathfrak C}_{\rm tr}+{\mathfrak C}_{\rm Cas}
590: \right]\right)
591: \,,
592: \end{eqnarray}
593: %
594: can be considered as a cross-check
595: that $\sigma_{\rm ren}$ does not acquire any non-trivial dependence on
596: $R(t)$ or $\dot R(t)$ during the renormalization procedure
597: (see also Appendix A).
598:
599: It turns out that the above equation of motion (\ref{eom-ren}) can be
600: derived from the following effective Lagrangian
601: %
602: \bea
603: \label{Leff}
604: L_{\rm eff}=2\pi\,\frac{\sqrt{1-\dot R^2}}{R}
605: \left(
606: -\frac{\sigma_{\rm ren}\,R^2}{2}
607: -{\mathfrak C}_{\rm Cas}+
608: \frac{\;{\mathfrak C}_{\rm tr}\dot{R}^2}{1-\dot{R}^2}
609: \right)
610: \,,
611: \ea
612: %
613: and that the total (renormalized) energy in Eq.~(\ref{energy-ren}) is
614: just the associated Hamiltonian $E=H=P_{\rm eff}\dot R-L_{\rm eff}$.
615:
616: As for any conservative system, possible trajectories $R(t)$ can be
617: inferred from the energy landscape according to
618: Eq.~(\ref{energy-ren}), see Fig.~\ref{full}.
619: To this end we have to specify the constants ${\mathfrak C}_{\rm tr}$
620: and ${\mathfrak C}_{\rm Cas}$.
621: For the example of a single scalar field with periodic boundary
622: conditions we have
623: %
624: \bea
625: \label{single}
626: {\mathfrak C}_{\rm tr}=-\frac{1}{24\pi}
627: \;,\;
628: {\mathfrak C}_{\rm Cas}=-\frac{\pi}{6}
629: \,.
630: \ea
631: %
632: For multiple scalar fields and/or possibly different boundary
633: conditions one would obtain other values.
634: Nevertheless, both contributions remain always negative ${\mathfrak
635: C}_{\rm tr}<0$ and ${\mathfrak C}_{\rm Cas}<0$.
636: In this situation one can infer from the qualitative behavior of the
637: energy landscape in Figs.~\ref{full} and \ref{no-cas} that a
638: trajectory $R(t)$ corresponding to a positive energy $E>0$ does never
639: reach $R=0$ and also cannot approach $\dot R=1$ for finite $R$.
640: I.e., we avoid the collapse to a singularity $R=0$ -- provided that
641: the semi-classical treatment does not break down completely
642: (cf.~the next Section).
643: %
644: \begin{figure}[!h]
645: \includegraphics[height=7cm]{figur1.eps}
646: \caption{Energy landscape (in arbitrary units) with $\sigma=20$ and
647: ${\mathfrak C}_{\rm tr}={\mathfrak C}_{\rm Cas}=-1$,
648: which has been cut off at negative energies for convenience.
649: As one can easily perceive, a circle with a positive (initial) energy
650: can never collapse -- i.e., reach $R=0$ -- unless it leaves the region
651: of validity of the semi-classical picture.}
652: \label{full}
653: \end{figure}
654: %
655: \begin{figure}[!h]
656: \includegraphics[height=7cm]{figur2.eps}
657: \caption{Energy landscape with the same values for $\sigma=20$ and
658: ${\mathfrak C}_{\rm tr}=-1$, but no Casimir effect
659: ${\mathfrak C}_{\rm Cas}=0$. Hence the Casimir effect is not
660: substantial for preventing the collapse to $R=0$.}
661: \label{no-cas}
662: \end{figure}
663: %
664:
665: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
666: \section{Bounce}
667: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
668:
669: It turns out to be very interesting to discuss possible trajectories
670: $R(t)$ in more detail.
671: Let us assume that we start with a circle with a large initial radius
672: $R(t=0)=R_0$ and a vanishing initial velocity $\dot R(t=0)=0$ -- where
673: the quantum effects are negligible.
674: In the energy landscape in Fig.~\ref{full}, this corresponds
675: to the region in the middle (around the axis $\dot R=0$) high up the
676: mountain.
677: Starting in normal region with $R(t=0)=R_0$, the radius begins to
678: shrink $\dot R<0$ and the negative velocity gradually increases in
679: magnitude.
680: After a while, i.e., for a sufficiently large velocity
681: (and, correspondingly, small radius), quantum effects become
682: important.
683: Finally, the trajectory reaches the ridge, where the radius $R$ cannot
684: decrease anymore.
685: Therefore, the velocity $\dot R$ should change its sign -- i.e., the
686: trajectory $R(t)$ should turn around.
687: But these two ridges correspond to finite velocities $\dot R$ -- one
688: positive and the other negative -- such that a change in the sign is only
689: possible by jumping from one ridge to the other
690: (corresponding to the same values of $R$, $|\dot R|$, and $E$).
691: Obviously this requires a diverging $\ddot R$ and seems to be a very
692: strange property of the semi-classical system.
693:
694: Nevertheless, it turns out that such a strange property is a necessary
695: consequence of the weird features of the energy landscape in
696: Fig.~\ref{full}.
697: The energy $E(R,\dot R)$ equals the Hamiltonian $H(R,P)$ derived from
698: the effective Lagrangian $L(R,\dot R)$ in Eq.~(\ref{Leff}).
699: For a given and fixed radius $R$, there are three values of $\dot R$
700: for which
701: %
702: \bea
703: \left(\frac{\partial E}{\partial\dot R}\right)_R=0
704: \,,
705: \ea
706: %
707: the usual $\dot R=0$ (in the middle, i.e., the normal region) and two
708: anomalous points (on the two ridges) where $\dot R\neq0$.
709: In view of the Hamilton equation
710: %
711: \bea
712: \dot R
713: =
714: \left(\frac{\partial H}{\partial P}\right)_R
715: =
716: \left(\frac{\partial H}{\partial\dot R}\right)_R
717: \left(\frac{\partial\dot R}{\partial P}\right)_R
718: \,,
719: \ea
720: %
721: for these two anomalous points where $\dot R\neq0$, we must have
722: %
723: \bea
724: \left(\frac{\partial P}{\partial\dot R}\right)_R
725: =
726: \left(\frac{\partial^2L}{\partial\dot R^2}\right)_R
727: =0
728: \,.
729: \ea
730: %
731: I.e., the ridges are critical points where the Euler-Lagrange equation
732: %
733: \bea
734: \left(\frac{\partial L}{\partial R}\right)_{\dot R}
735: =
736: \frac{d}{dt}\left(\frac{\partial L}{\partial\dot R}\right)_R
737: =
738: \ddot R\left(\frac{\partial^2L}{\partial\dot R^2}\right)_R
739: +
740: \dot R\,\frac{\partial^2L}{\partial R\,\partial\dot R}
741: \,,
742: \ea
743: %
744: cannot determine $\ddot R$ anymore, i.e., $\ddot R$ diverges.
745: Indeed, the pre-factor in front of the $\ddot R$-term in the equation
746: of motion (\ref{eom}) goes to zero at these critical points.
747:
748: Since $\ddot R$ approaches infinity there -- whereas $\dot R$ and $R$
749: remain finite -- the Ricci scalar diverges at the ridges, i.e., these
750: critical points still represent a curvature singularity.
751: %
752: \begin{figure}[!h]
753: \includegraphics[height=6.5cm]{figur3.eps}
754: \caption{Energy landscape for the classical case, i.e.,
755: ${\mathfrak C}_{\rm tr}={\mathfrak C}_{\rm Cas}=0$. All trajectories
756: corresponding to some (positive) energy inevitably lead to
757: $R\downarrow0$ and $\dot R\downarrow-1$, i.e., the circle collapses.}
758: \label{class}
759: \end{figure}
760: %
761: \begin{figure}[!h]
762: \includegraphics[height=6.5cm]{figur4.eps}
763: \caption{Energy landscape for ${\mathfrak C}_{\rm tr}=0$ and
764: ${\mathfrak C}_{\rm Cas}=-1$. In this rather artificial set-up the
765: trajectories reach the boundary $|\dot R|=1$ at a finite radius $R>0$.}
766: \label{no-tr}
767: \end{figure}
768: %
769:
770: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
771: \section{Conclusions}
772: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
773:
774: Let us compare three scenarios:
775: %
776: \begin{itemize}
777: \item Without any back-reaction (cf.~Fig.~\ref{class}), the string
778: collapses to $R=0$ and $\dot R=1$ after a finite laboratory time --
779: i.e., the metric, the action and the Ricci scalar become singular.
780: \item With purely classical back-reaction (e.g., a thermal bath) all
781: of these singularities are avoided and the string oscillates smoothly
782: forever. However, the Casimir effect generates the same contribution
783: -- but with the opposite sign -- and can therefore negate this effect.
784: \item If we take into account the Casimir effect only -- which is
785: rather artificial -- the string reaches $\dot R=1$ and hence develops
786: a singularity of the metric, the action and the Ricci scalar,
787: respectively, after a finite laboratory time and at a finite radius
788: $R>0$ (cf.~Fig.~\ref{no-tr}).
789: \item Calculating the full expectation value
790: $\langle\hat T^\mu_\nu\rangle_{\rm ren}$ in the conformal vacuum --
791: including the trace anomaly -- we find again that circle reaches the
792: critical point after a finite laboratory time, cf.~Fig.~\ref{full}.
793: However, such a critical point is characterized by a singular
794: acceleration $\ddot R$ and hence Ricci scalar only -- while the radius
795: $R>0$, the velocity $\dot R<1$, and hence both, the action as well as
796: the metric remain regular.
797: \end{itemize}
798: %
799: We observe that the back-reaction of the quantum field $\hat\Phi$
800: prevents the strong singularity of the first scenario, but still
801: generates a weaker singularity.
802: %
803: At this critical point (i.e., the ridge in the energy landscape) the
804: pre-factor in front of $\ddot R$ obtained via the semi-classical
805: treatment vanishes.
806: %
807: It appears natural to expect that the quantum back-reaction beyond the
808: semi-classic picture will become important in the vicinity of this point.
809:
810: It is evident that the semi-classical treatment of the back-reaction
811: problem has clear problems, cf.~\cite{example}.
812: %
813: The major point is that the semi-classical Einstein equations
814: -- and similarly Eq.~(\ref{eom-ren}) --
815: imply the neglect of all quantum fluctuations by considering the
816: expectation value $\langle\hat T_{\mu\nu}\rangle_{\rm ren}$ only.
817: %
818: This approximation is justified if the fluctuations are sufficiently
819: small, i.e., as long as conditions such as
820: %
821: \bea
822: \langle\hat T_{\mu\nu}^2\rangle_{\rm ren}
823: \stackrel{?}{\approx}
824: \langle\hat T_{\mu\nu}\rangle_{\rm ren}^2
825: \,,
826: \ea
827: %
828: hold.
829: %
830: Assuming the validity of the renormalization procedure such as
831: point-splitting -- e.g., that all allowed quantum states satisfy the
832: Hadamard condition -- the trace-anomaly is independent of the quantum
833: state and hence a c-number.
834: %
835: Ergo, for this particular quantity one would expect the
836: semi-classical treatment to work.
837: %
838: However, the equations of motion do not only contain
839: $\langle\hat T_\mu^\mu\rangle_{\rm ren}$, but also
840: $\langle\hat T_0^0\rangle_{\rm ren}$ etc.
841: %
842: As the derivation of these quantities is more involved and includes
843: other contributions (e.g., $\langle\hat T_{01}^2\rangle_{\rm ren}$)
844: as well as a time-integration, it is not clear that the entanglement
845: between $\hat T_{\mu\nu}$ and the corresponding operator $\widehat R$
846: can be omitted, for example
847: %
848: \bea
849: \langle\hat T_{\mu\nu}\widehat R\rangle_{\rm ren}
850: \stackrel{?}{\approx}
851: \langle\hat T_{\mu\nu}\rangle_{\rm ren}
852: \langle\widehat R\rangle_{\rm ren}
853: \,.
854: \ea
855: %
856:
857: In summary, the semi-classical theory suggests its own demise by
858: predicting a trajectory $R(t)$ which is not continuously
859: differentiable ($\ddot R$ diverges) and hence a curvature singularity.
860: %
861: As a result the fate of the circle after the bounce remains unclear in
862: the sense that it cannot be determined within the semi-classical
863: picture without additional arguments.
864: %
865: Assuming that the circle re-enters the region of validity of the
866: semi-classical theory after the bounce, there are two possibilities.
867: %
868: After jumping from one ridge to the other, the trajectory could turn
869: around and continue towards the normal regime in the middle again
870: ($\dot R$ decreases).
871: %
872: In this case, the circle would perform an eternal oscillation.
873: %
874: Alternatively, the trajectory could turn to the edge and enter the
875: anomalous regime.
876: %
877: In this situation, the velocity would increase forever and
878: asymptotically reach the speed of light.
879: %
880: These run-away solutions are somewhat similar to a trace anomaly
881: induced inflation in 3+1 dimensional gravity.
882: %
883: Interestingly, the trace anomaly of both, a scalar field in 1+1
884: dimensions on the one hand and that of, say, a vector field in 3+1
885: dimensional gravity on the other hand are such that they prevent the
886: collapse to a metric singularity $R=0$ and admit run-away solutions.
887: %
888: (The Casimir effect is not substantial in that respect,
889: cf.~Fig.~\ref{no-cas}.)
890:
891: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
892: \section{Outlook}
893: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
894:
895: As it became evident from the previous discussions, a more exhaustive
896: analysis would require the investigation of the full quantum
897: back-reaction.
898: %
899: At a first glance, one could expect that this is provided by quantizing the
900: effective action (\ref{Leff}) for the geometric sector,
901: cf.~Eq.~(\ref{polyakov}) below, after integrating out the field $\hat\Phi$.
902: %
903: However, it is by no means clear that such an approach is indeed equivalent
904: to a full quantization of the (geometric) string degrees of freedom
905: interacting with the quantum field.
906: %
907: In view of the non-trivial relation between the canonically conjugated
908: momentum $P$ and the velocity $\dot R$ as well as the occurrence of anomalies
909: and the associated non-linearities, a rigorous treatment is apparently
910: rather involved.
911: %
912: Furthermore, an {\em ab initio} quantization has to account for non-spherical
913: geometries, i.e., deviations from the rotational symmetry, since the monopole
914: mode inherently couples to higher multipole modes and their
915: quantum
916: fluctuations -- in contrast to the semi-classical case considered above.
917:
918: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
919: \section{Negative Energies and Stability}
920: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
921:
922: One might ask whether the breakdown of the semiclassical treatment for
923: the toy model under consideration is just an artifact caused by the
924: special features of this 1+1 dimensional model or whether it
925: represents a generic property also for 3+1 dimensional gravity
926: (plus quantum fields).
927: %
928: Even though this question cannot be addressed completely without
929: solving the full (3+1 dimensional) scenario, one can compare
930: characteristic features of both systems
931: (at the semiclassical level).
932:
933: For the classical string (without quantum fields) as described by the
934: Nambu-Goto action, the (conserved) energy is positive
935: $E=\pi\sigma R/\sqrt{1-\dot R^2}$.
936: %
937: Nevertheless, the equation of motion admits singular solutions
938: (collapse to $R=0$) in close analogy to classical gravity
939: (cf.~the singularity theorems).
940:
941: Owing to the trace anomaly and the Casimir effect, the renormalized
942: energy of the quantum field (in the background of the string) is
943: negative in the vacuum state.
944: %
945: However, that does not imply that the vacuum state of the quantum
946: field itself (in the semiclassical treatment) is unstable.
947: %
948: After a normal mode expansion
949: %
950: \bea
951: \hat\Phi(\tau,\varphi)
952: =
953: \sum\limits_{m=0}^\infty
954: \hat Q_m^{(c)}(\tau)\cos(m\varphi)+
955: \sum\limits_{m=1}^\infty
956: \hat Q_m^{(s)}(\tau)\sin(m\varphi)
957: \,,\!\!\!\!\!\!\!\!
958: \nn
959: \ea
960: %
961: the Hamiltonian decouples and is positive for each mode
962: %
963: \bea
964: \hat H_\Phi^\tau=\sum\limits_{m=0}^\infty\;\sum\limits_{\xi=(c,s)}
965: \left(
966: \hat P_{m,\xi}^2+
967: m^2\hat Q_{m,\xi}^2
968: \right)
969: \,.
970: \ea
971: %
972: Apart from the zero mode $m=0$, all field modes possess a unique
973: ground state -- but even for the zero mode, the quantum evolution is
974: not unstable.
975: %
976: In summary, the dynamics of the quantum field itself [in the
977: semiclassical treatment, i.e., for a fixed trajectory $R(\tau)$]
978: does not display any instabilities.
979: %
980: The stability of the full theory (string plus field), however,
981: lies outside the scope of the present investigations
982: (i.e., is still unclear) in view of the breakdown of the semiclassical
983: treatment.
984:
985: The negative (renormalized) energy of the quantum field cannot be
986: attributed to single modes. Instead, being similar to vacuum polarization
987: effects, it is a consequence of the renormalization procedure.
988: %
989: This is a generic feature of quantum fields in curved space-times and
990: occurs in the 3+1 dimensional scenario as well.
991:
992: In view of the above similarities, one might conjecture that the
993: results found for the 1+1 dimensional toy model -- i.e., the breakdown
994: of the semiclassical treatment and a possible bounce caused by the
995: trace anomaly --
996: are not just artifacts of this toy model but shed some light on the 3+1
997: dimensional situation as well.
998:
999: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1000: \section*{Appendix A: Renormalization}
1001: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1002:
1003: Starting form the obvious condition that the (renormalized)
1004: energy-momentum tensor of the total system (quantum field plus string)
1005: is conserved with respect to the (flat) 2+1 dimensional embedding
1006: space-time
1007: %
1008: \bea
1009: \partial_A T^{AB}_{\rm total}=0
1010: \,
1011: \ea
1012: %
1013: one can show the compatibility of the two conditions
1014: %
1015: \bea
1016: \nabla_\mu\langle\hat T^\mu_\nu\rangle_{\rm ren}=0
1017: \;\leftrightarrow\;
1018: \sigma_{\rm ren}={\rm const}
1019: \,,
1020: \ea
1021: %
1022: for the renormalized energy-momentum tensor of the quantum field and
1023: the string tension, respectively.
1024:
1025: As a consequence of Eq.~(38), the argument works in both directions:
1026: Demanding $\nabla_\mu\langle\hat T^\mu_\nu\rangle_{\rm ren}=0$
1027: in order to ensure energy conservation of the quantum field
1028: separately in the presence of a Killing vector $\xi$ of the induced
1029: metric, one deduces $\sigma_{\rm ren}={\rm const}$.
1030: Conversely, one may start with the assumption
1031: $\sigma_{\rm ren}={\rm const}$ and derive
1032: $\nabla_\mu\langle\hat T^\mu_\nu\rangle_{\rm ren}=0$ from
1033: $\partial_A T^{AB}_{\rm total}=0$.
1034:
1035: Since the decomposition of the total energy-momentum tensor
1036: $T^{AB}_{\rm total}$ into the contributions of the quantum field and the
1037: string is not unique, one can even impose alternative renormalization
1038: conditions.
1039: When renormalizing $\langle\hat T^\mu_\nu\rangle$, its divergent part
1040: has to be absorbed into $\sigma$, while the finite part of the
1041: counter-term remains undetermined.
1042: Here we have employed a ''minimal subtraction scheme'',
1043: cf.~Eqs.~(\ref{reg-T}) and (\ref{reg-sigma}).
1044: As an alternative renormalization scheme, for example, one may impose the
1045: condition that the trace still vanishes
1046: $\langle\hat T^\mu_\mu\rangle_{\rm ren}=0$ and abandon the property
1047: $\nabla_\mu\langle\hat T^\mu_\nu\rangle_{\rm ren}=0$ instead.
1048: Such a renormalization procedure could be envisaged as a
1049: ''non-minimal subtraction scheme''.
1050: Instead of employing Eq.~(\ref{reg-T}) we could split up
1051: $\langle\hat T_{\mu\nu}\rangle_{\rm bare}$ via
1052: %
1053: \bea
1054: \label{reg-T-alt}
1055: \langle\hat T_{\mu\nu}\rangle_{\rm bare}
1056: =
1057: \frac{1}{\epsilon}\,{g}_{\mu\nu}
1058: +
1059: \frac12
1060: {\mathfrak C}_{\rm tr}\,{\cal R}
1061: \,{g}_{\mu\nu}
1062: +
1063: \widetilde{\langle\hat T_{\mu\nu}\rangle}_{\rm ren}
1064: \,,
1065: \ea
1066: %
1067: with
1068: %
1069: \bea
1070: \widetilde{\langle\hat T_\mu^\mu\rangle}_{\rm ren}=0
1071: \;\leadsto\;
1072: \nabla_\mu\widetilde{\langle\hat T^\mu_\nu\rangle}_{\rm ren}\neq0
1073: \,,
1074: \ea
1075: %
1076: and, consequently,
1077: %
1078: \bea
1079: \label{reg-sigma-alt}
1080: \sigma_{\rm bare}
1081: =
1082: -\frac{2}{\epsilon}
1083: -{\mathfrak C}_{\rm tr}\,{\cal R}
1084: +\widetilde{\sigma}_{\rm ren}
1085: \,.
1086: \ea
1087: %
1088: Within this alternative renormalization scheme
1089: $\widetilde{\sigma}_{\rm ren}$ depends on the geometry
1090: %
1091: \bea
1092: \widetilde{\sigma}_{\rm ren}=
1093: \widetilde{\sigma}_{\rm ren}(R,\dot R,\ddot R)
1094: \,,
1095: \ea
1096: %
1097: while the equation of motion for $R(t)$ and the total energy remain
1098: the same within both schemes.
1099: Such a non-minimal subtraction scheme employed for renormalization of
1100: the of the quantum-field sector provides the possibility of inducing a
1101: space-time dependence of characteristic parameters
1102: (here simply the string tension) of the classical (geometric) sector.
1103:
1104: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1105: \section*{Appendix B: Effective action}
1106: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1107:
1108: Instead of using the equations of motion, one could investigate the
1109: back-reaction by means of the effective action, cf.~\cite{nagatani}
1110: and also \cite{brevik}.
1111: In 1+1 dimensions, the trace anomaly can be deduced from the
1112: Liouville-Polyakov effective action \cite{polyakov}
1113: %
1114: \bea
1115: \label{polyakov}
1116: {\mathfrak A}_{\rm eff}=\frac{1}{96\pi}\int d^2x\,\sqrt{-{g}}\,
1117: {\cal R}\,\Box^{-1}\,{\cal R}
1118: \,.
1119: \ea
1120: %
1121: In view of the inverse of the d'Alembert operator $\Box^{-1}$, this
1122: expression appears to be non-local, but inserting the value for the
1123: Ricci scalar in terms of the conformal coordinates, for example, it
1124: turns out that it is in fact, effectively local (no particle creation),
1125: and agrees with the term proportional to ${\mathfrak C}_{\rm tr}$
1126: in the effective Lagrangian in Eq.~(\ref{Leff}) after a coordinate
1127: transformation to the laboratory time.
1128:
1129: However, in this na{\"\i}ve treatment the Casimir term is missing and
1130: thus requires additional considerations.
1131: This already illustrates a potential danger of the effective action
1132: method -- the fact the total effective Lagrangian is just as a sum of
1133: the terms proportional to ${\mathfrak C}_{\rm tr}$ and the (static)
1134: Casimir effect (in addition to the classical terms) is not
1135: {\em a priori} obvious.
1136:
1137: In order to compare our calculation with the results of
1138: Ref.~\cite{nagatani}, a few remarks are in order:
1139: %
1140: \begin{itemize}
1141: \item Switching to the Euklidean signature -- as done in
1142: Ref.~\cite{nagatani} -- is a well-defined procedure for the
1143: calculation of the static Casimir energy, but not for general
1144: time-dependent systems.
1145: Hence the derivation of the main result as a sum of the two
1146: contributions is somewhat {\em ad hoc}.
1147: \item Contrary to the claims in Ref.~\cite{nagatani}, the scenario
1148: under consideration does not exhibit the dynamical Casimir effect
1149: (in the sense of particle creation) due to the conformal invariance.
1150: \item Furthermore, Ref.~\cite{nagatani} uses a non-relativistic
1151: kinetic term (introduced by hand).
1152: Therefore, the results can be compared with ours in the adiabatic
1153: regime $\dot R\ll1$ only.
1154: \end{itemize}
1155: %
1156: In addition to the problems mentioned in Ref.~\cite{dowker}, for
1157: example, a conclusive discussion of a possible non-trivial dependence
1158: of the renormalized string tension $\sigma_{\rm ren}(R,\dot R,\dots)$
1159: in dependence of the renormalization scheme (see Appendix A) by means
1160: of the effective action is not obvious.
1161:
1162: Having in mind the various potential difficulties of the effective
1163: action method indicated above, we would like to emphasize that,
1164: in contrast to this approach, the calculations based on the
1165: renormalized expectation values of the energy-momentum tensor
1166: $\langle\hat T_{\mu\nu}\rangle_{\rm ren}$ together with employing the
1167: equations of motion, are apparently less ambiguous and more evident.
1168:
1169: \newpage
1170:
1171: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1172: \section*{Acknowledgments}
1173: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1174:
1175: G.~P.~and M.~R.~acknowledge support from the DFG.
1176: R.~S.~gratefully acknowledges financial support by the Emmy-Noether
1177: Programme of the German Research Foundation (DFG) under grant
1178: No.~SCHU 1557/1-1 and by the Humboldt foundation.
1179:
1180: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1181: \addcontentsline{toc}{section}{References}
1182: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1183:
1184: \begin{thebibliography}{99}
1185:
1186: \bibitem{nambu}
1187: P.~A.~Collins and R.~W.~Tucker,
1188: %``Classical and quantum mechanics of free relativistic membranes,''
1189: Nucl.\ Phys.\ B {\bf 112} 150 (1976);
1190: A.~Sugamoto, Nucl.\ Phys.\ B {\bf 215}, 381 (1983).
1191:
1192: \bibitem{vilenkin}
1193: A.~Vilenkin,
1194: %``Cosmic strings,''
1195: Phys.\ Rev.\ D {\bf 24}, 2082 (1981);
1196: A.~Vilenkin and E.~P.~S.~Shellard,
1197: {\em Cosmic Strings and Other Topological Defects}
1198: (Cambridge University Press, Cambridge, 2000).
1199:
1200: \bibitem{diplom}
1201: M.~Ruser, Diploma Thesis, TU Dresden (unpublished);\\
1202: M.~Ruser, G.~Plunien and G.~Soff, {\em Particle creation in periodic
1203: backgrounds and on dynamical $p$-branes}, (preprint, 2004)
1204:
1205: \bibitem{birrell}
1206: N.~D.~Birrell and P.~C.~W.~Davies,
1207: {\em Quantum Fields in Curved Space},
1208: (Cambridge University Press, Cambridge, England 1982).
1209:
1210: \bibitem{trace}
1211: P.~C.~Davies, S.~A.~Fulling and W.~G.~Unruh,
1212: %``Energy Momentum Tensor Near An Evaporating Black Hole,''
1213: Phys.\ Rev.\ D {\bf 13}, 2720 (1976).
1214: %%CITATION = PHRVA,D13,2720;%%
1215:
1216: \bibitem{example}
1217: As an example, let us consider the semi-classical Einstein equations
1218: (\ref{einstein}) for a quantum state $|\Psi\rangle$ describing a
1219: macroscopic superposition
1220: $|\Psi\rangle=(|\Psi_1\rangle+|\Psi_2\rangle)/\sqrt{2}$.
1221: %
1222: The state $|\Psi_1\rangle$ corresponds to a large mass $M$ at the point
1223: $x_1$ whereas $|\Psi_2\rangle$ describes the same mass $M$ at a
1224: (macroscopically) different position $x_2$.
1225: %
1226: The semi-classical Einstein equations (\ref{einstein}) predict a
1227: gravitational field
1228: (since $\langle\Psi_1|\hat T^\mu_\nu|\Psi_2\rangle_{\rm ren}=0$)
1229: as if there was half the mass $M/2$ at $x_1$ and the other half $M/2$
1230: at $x_2$.
1231: %
1232: However, in reality this macroscopic superposition $|\Psi\rangle$ will
1233: exhibit an extremely fast decoherence with a negligible energy flux
1234: due to the interaction with the environment leading to a classical
1235: mixture $\hat\varrho=(\hat\varrho_1+\hat\varrho_2)/2$.
1236: %
1237: Hence one would naturally expect a classical mixture of gravitational
1238: fields.
1239:
1240: \bibitem{nagatani}
1241: Y.~Nagatani and K.~Shigetomi,
1242: %``Effective theoretical approach to backreaction of the dynamical
1243: %Casimir effect in 1+1 dimensions,''
1244: Phys.\ Rev.\ A {\bf 62}, 022117 (2000).
1245: %[arXiv:hep-th/9904193].
1246: %%CITATION = HEP-TH 9904193;%%
1247:
1248: \bibitem{brevik}
1249: I.~Brevik, K.~A.~Milton, S.~D.~Odintsov and K.~E.~Osetrin,
1250: %``Dynamical Casimir effect and quantum cosmology,''
1251: Phys.\ Rev.\ D {\bf 62}, 064005 (2000).
1252: %[arXiv:hep-th/0003158].
1253: %%CITATION = HEP-TH 0003158;%%
1254:
1255: \bibitem{polyakov}
1256: A.~M.~Polyakov,
1257: %``Quantum Geometry Of Bosonic Strings,''
1258: Phys.\ Lett.\ B {\bf 103}, 207 (1981).
1259: %%CITATION = PHLTA,B103,207;%%
1260:
1261: \bibitem{dowker}
1262: J.~S.~Dowker,
1263: %``A Note On Polyakov's Nonlocal Form Of The Effective Action,''
1264: Class.\ Quant.\ Grav.\ {\bf 11}, L7 (1994).
1265: %[arXiv:hep-th/9309127].
1266: %%CITATION = HEP-TH 9309127;%%
1267:
1268: \end{thebibliography}
1269:
1270: \end{document}
1271: