Home

Search

Collections

Journals

About

Contact us

My IOPscience

On time and the quantum-to-classical transition in Jordan-Brans-Dicke quantum gravity

This article has been downloaded from IOPscience. Please scroll down to see the full text article. 1993 Class. Quantum Grav. 10 2511 (http://iopscience.iop.org/0264-9381/10/12/010) View the table of contents for this issue, or go to the journal homepage for more

Download details: IP Address: 24.108.204.67 The article was downloaded on 01/10/2010 at 05:28

Please note that terms and conditions apply.

C a s Quanlum Grav. 10 (1993) 2511-2518. Rinted in the UK ls.

On time and the quantum-to-classical transition in Jardan-Brans-Dicke quantum gravity

Claus Kiefelts and Erik A Martfnez$ll t Institute for Theoretical Physics. Univenity of ZLlrich, Schanberggasse 9, C H 8W1 ZOrich,
Switzerland t lwtitute for Theoretical Physics, University of &m, S i & r s W 5,CH MI2 Bem, Switlerland Received 25 June 1993, in final form 23 July 1993 Abstract Any quantum Wy of gravity which treals lhe gravitationd constant as a r dynamical variable has to address the issue of superpositions of stales correspndifig to different eigenvalues. We show how the unobservability of such superpositions can be explained ihmugh the interaction with other gravitational depees of f r d o m (desoherence). Tie formnl framework i s canonically quantized Jordax-BonF-Dicke theory. We discuss the wncepts of inbinsic time and semi-classical time as well as the possibility of Nnnelling into regions corresponding to a negative gravitational wnstanl. We calculate l reduced density maeix k of the Jordan-Brans-Dicke field and show that the off-diagonal elements can k sufficiently suppressed to k consistent wilh experiments. The possible relevance of this mechanism for sUuctwe famation in extended inflation is briefly discussed.
PACS number: 0 4 0

It is of fundamental interest to understand the origin of coupling constants in physics. Some approaches are based on the idea of a dynamical origin of these constants within the framework of a more fundamental theory One prominent example is the possible effect of wormholes in driving the cosmological constant to zero [I]. More generally, wormholes may dynamically inhence any low energy coupling constant. Since the underlying framework is that of quantum theory, one has to address the issue of superpositionsof states corresponding to different values of the coupling constant Since quantum theory is linear, any a priori selection of a special state would amount to a selection by hand, and therefore prevent the possibility to understand dynamically the emergence of classical states. It is the purpose of this paper to provide a dynamical explanation in the particular example of the gravitational constant. Instead of working with wormholes and Euclidian quantum gravity, we restrict ourselves to the simpler framework of Jordan-Brans-Dicke (JBD) theories, where the &le of the gravitational constant is played by a scalar field [2]. Such theories also arise as low energy effective theories from string theory. We work in the framework of canonical quantization which focuses on wave functionals depending on the JBD field (apart from
C #

8 Present address: Institute for Theoretical Physics, University of Freiburg,

Hermann Herder Sfrasse 3, D 79104 Freiburg, Germany. I1 R e s n t address: Thwretical Physics Institute, Lkpartmem of Physics, University of A l b e m Edmonton. Canada T6G 2JI.

@2649381193/122511t08$0750 Q 1993 iOP Publishing Ltd

2511

2512

C Kiefer and E A Martinez

the three-metric and other fields). The issue is then to understand the non-occurrence of superpositions like

which would be in contradiction with the experience of a well defined gravitational constant. We will first outline the formal framework of our investigation, and then proceed to understand, within this framework, the emergence of the gravitational constant. We note that the effect of wormholes has also been studied in the context of the JBD theory 131 with the conclusion that the wave functional is peaked at infinite JBD parameter U , but we will not take into account the effect of any wormholes in the following. In its most simple version, the Lagrangian of JBD theories is given by

where o is a constant and no mass term or self-interaction for the JBD field is taken into account. It is straightforward to perform a 3t1 decomposition to cast the theory into Hamiltonian form. The central equation in the canonical theory is the Hamiltonian constraint which reads gn,Trn: 2w 3

2(2w

+ 3)

-k f i ( - & R

+ E h*@,d$,b) + %, 4

where l is the geometrodynamical momentum, and n, is the momentum canonically l conjugate to the JBD field. We note that the kinetic terms of the JBD field are suppressed in the limit of large w , which is the reason why the classical theory goes over to general relativity f o r o --t CO,provided one chooses the solution 4 = constant = (16nG,v)-t. The coefficients in front of the gravitational momenta are the components of DeWitts mebic in the space of three-geometries and the JBD field. Owing to the presence of the JBD field it is a 7 x 7 matrix at each space point. The components read

and

We have introduced here an auxiliary length 1. so that all components of DeWitts metric have the same dimension (note that the momentum canonically conjugate to 4 has the dimension of an inverse length, whereas the geometrodynamical momentum has the t me gravitational constant measured by obselvaion of dowly moving panicles or in time dilation exprinenls
is nor C N but (2w + 4)/(20 + 3)Gn [ ] Since we are intermed here only in the limit of large w, we will 4. We this difference into account.
MI

Time and the quantum-to-classical transition

2513

dimension of an inverse length cube). The results presented below are independent of I . Viewed as a 7 x 7 matrix at each space point, the metric can be diagonalized, with the result

where

In the limit of large w one finds that A+ and A- approach the values -$(I 14Q2/2m, respectively. The determinant of (6) is given by detGAB = -

- 3/20) and

8110Q5(2m 3)* +

w+3

(7)

An interesting point is the behaviour of the signature of this metric In general relativity the signature is hyperbolic, with the minus sign arising from the conformal mode. This minus sign is found from A+ in (6) in the limit w + 00. But now the signature also depends on the value of the JBD field and on w. As can be viewed from (6) and (7). the signature changes, for positive $, from hyperbolic to ultra-hyperbolic (mixed signature) if w becomes smaller than -3. In the case of negative $ the overall signature of the metric is changed. This might not seem to be disturbing, but one must keep in mind that the metric of non-gravitational fields is elliptic, so that the hyperbolic nature of the total metric is destroyed, for negative 4, for all values of w. Whilst this behaviour may be irrelevant at the classical level, it has important consequences for the quantum theory. There the constraint (3) is implemented as a condition on physically allowed wavefunctionals, i.e.,

If the signature of DeWitts metric is hyperbolic, a well defined initial value problem can be posed with respect to an intrinsic time variable, which is played by the conformal part of the three-metric [5]. This propem is lost in the case of mixed signature. A similar observation has been made before in the context of scalar fields which are coupled nonminimally to gravity [6]. We note that the value w = -1, which is motivated by string theory [7] and recently studied models of dilatonic black holes [8]. still lies in the hyperbolic region. We also note that in more general theories where m may depend on Q the demand for hyperbolicity would lead to a non-trivial restriction on the allowed range of the JBD field. Without such a demand one may have to deal with the presence of superspace Cauchy horizons, as in 161 whose interpretation is not clear. In the classical theory the particular solution for Q can be restricted to lie in the region of positive fields. At first glance, there does not seem to be a problem at the classical level if C is chosen t be negative, which would correspond to a gravitational repulsion. o It would, however, lead to a negative ADM energy since the positive energy theorem is no longer applicable. (A Schwarzschild solution with negative G, for example, would be equivalent to a solution with negative ADM mass.) This is reflected in the quantum theory in the change of sign in the DeWitt metric. There the wavefunctional may with negative fields. This is analogous, although different, to the

2514

C Kiefer and E A Martinez

in the asymptotically flat context which was discussed by Ashtekar and Horowitz 191, and which leads to the possibility of negative energy states in quantum gravity. In the following we will first discuss a mini-superspace model (closed Friedmann universe) where the gravitational degrees of freedom are the scale factor a and the homogeneous palt & of the JBD field. Models of this kind have been discussed in the context of extended inflation, e.g. in [lo]. The interaction with inhomogeneous degrees of freedom will then be considered below. After a field redefinition

which diagonalizes the kinetic term?, the Hamiltonian reads, in the limit of large w,

One can easily extend the discussion to a more realistic model which includes a massive scalar field ('inflaton field') but this is not necessary for the present purpose. Choosing the Laplac+Beltrami ordering for the kinetic term, the mini-superspace Wheeler-DeWitt equation reads (with a e")

This equation cannot be solved exactly, but it can be solved in a Bom4ppenheimer > approximation, since o > 1 and the kinetic term of the x field is suppressed (experimentally one has w > 500). Since we do not need the solution for $0 explicitly, we will not write it down here. It is, however, instructive to have a brief look at the mini-superspace equation in the simpler case where the Friedmann universe is open and the inflaton field is absent. Instead of ( I 1) we then have (neglecting the linear factor ordering term)

This is a wave equation with an effective speed *(we consider a as the time variable). The mini-superspace wavepacket therefore does not spread. Equation (12) can of course be easily solved in terms of plane waves,

General solutions, for example wavepackets, may be constructed by superposing many plane waves. If one had just one localised packet, this would model the classical behaviour of this theory and lead to a 'frozen' packet in the general relativistic limit of infinite w. As we will see below, however, the recovery of the time-dependent Schrijdinger equation for non-gravitational fields demands the presence of a single WKB state (13). Such a state ascribes equal probabilities to all values of ,. which would be in sharp contradiction to the y well defined observed value of the gravitational constant.

iThe choice of the variable x corresponds to the boundary condition that ihe wavefunction is zero in regions corresponding to a negative gravitational mnsfanL The general relativistic value 40 = ( l 6 n G ~ ) - 'corresponds
UIx=O.

Time and the quantum-to-classical transirion

2515

How can this apparent contradiction be resolved? The key idea is the observation that the mini-superspace degrees of freedom of equation (11) are actually correlated with a huge number of further degrees of freedom, both gravitational and non-gravitational ones. The relevant object to discuss is therefore not the mini-superspace wavefunction but the minisuperspace density matrix which is obtained by tracing out further, inaccessible, degrees of freedom. It had been shown before that interferences of different scale factors can become suppressed through this interaction [l 1,121. In fact, owing to the universal coupling of the metric to all degrees of freedom, the metric becomes classical to a high degree of accuracy and we will therefore assume in the following that the mini-superspace density matrix is already diagonal with respect to a. How, then, can one understand the emergence of a 'classical' @o? Consider an experiment where the JBD field is determined through the interaction of two masses. One can of course only measure-in full analogy to the measurement of electromagnetic fields [13]-the field average over an appropriate volume V i3 d 3 , where d is a typical distance between the masses. The definition of the mini-superspace density matrix should thus contain a coarse-graining in field space [ 141. Moreover, the state of the JBD field is quantum correlated with gravitational fields outside the volume V , since @ is really a cosmological field, in contrast to, say, the electric field. The gravitational fields in question are other masses, gravitational waves, and possibly also inhomogeneous degree?. of freedom of @ itself. These are the degrees of freedom which are to be traced over and which lead to de-coherence. It should be noted that de-coherence for the IBD field is ineffective as far as non-gravitational degrees of freedom are concemed, since only the metric couples directly to non-gravitational degrees of freedom. In the following discussion we will assume that the coarse-graining has been already performed, and we will only take the fluctuations in the JBD field itself into account. It is. however, known that these fluctuations also mimic the metric degrees of freedom [ 12,151. Technically, the discussion is carried out through an expansion of the full IBD field into harmonics on the three-sphere [16],

and the modes fnlm fn are taken into account only up to quadratic order. We note that the whole discussion will be carried out in the 'Jordan frame' in which the metric and the JBD field are not mixed by a field redefinition. Although it does not matter classically which frame one considers (galaxy formation in the framework of extended inflation, e.g., is studied more easily in the 'Einstein frame' [17]), the effect of de-coherence is sensitive to the choice of variables which are summed over. We consider it to be more sensible if the trace is performed over the original, not the re-scaled, variables, since the unscaled metric is the physical metric which is measured with rods and clocks. After a re-scaling of the lapse function, which multiplies the consaaint (3). by the JBD field (this will not change the results derived below), the Hamiltonian becomes, in the limit of large w,

2516

C Kiefer and E A Martinez

where n denotes the momentum canonically conjugate to fn. To solve the Wheeler-DeWitt . equation ( 8 ) we make the following ansatz for the wavefunction
U

where N denotes the number of inhomogeneous degrees of freedom. Assuming that *o is a single WKB wavefunction, $0 = eisn,and that the @n depend only adiabatically on a and & the wavefunctions Jr, obey a SchrXinger equation 1121 ,

with respect to the semi-classical time t which is defined by

(Vis the gradient in mini-superspace). As already remarked above, the Schrddinger equation can only be recovered if the mini-superspace wavefunction is a single WKB state (a possible em justification of this fact can be found in [18]). Neglecting the last t r in (15) (since o n 2 > 6) and taking n > 1, the Hamiltonians Hn are given by > >

H,*----+--fa2 4; 40af.Z

ia

aso

40 aa

a af.

$0

ah " af,,

asof -+-(-) aso 2 f 2 a 1

4@

a h

f o an

4 2 2

f,.
(19)

Basically, H, describes a time-dependent oscillator with mass O. The terms in (19) containing So describe, in the semi-classical limit, the time variation of the background variables a and $0. i.e.

(where 310 is the Hubble parameter), and

_

aso
ah

b a s= & h

~
'

In the following we neglect the terms which describe a time variation in the gravitational constant but keep in mind that these terms would have to be taken into account if this variation is significantt. We now proceed to solve (17) by a Gaussian ansatz and assume that the wave functions are in their adiabatic ground state. The result is

2wa2n
J= r( n) ,

114

exp

oa2n 3iE0a3 -f.' - -f.') (

40

$0

t The absolute value of the atio of the bird 10 the sewnd term in (19) is @yen by &/6&?iu than wj60O. since < IO-" yrr' from Nking experiments.

which is smaller

Time and the quantum-to-classical transirion

2517

We note that the consideration of the terms involving would amount to replacing 3 x 0 by 3 x 0 - 2o&/&. The mini-superspace density matrix is then obtained by

40

where we have already restricted ourselves to the diagonal in a space. Inserting (22) into (23) one finds

where po is the mini-superspace matrix without inhomogeneous degrees of freedom, and phase factors have not been written o t We emphasize that the first, dominating, term is u. independent of the value of o. As can be seen from this Gaussian, the influence of a single mode is ineffective, since there would still remain a coherence width in & which is equal to & itself. Only a large number of modes can de-cohere efficiently. The crucial issue is now the appropriate choice for the maximum number of modes N. We make the proposition of considering only modes with wavelength bigger than d , the separation of the masses considered above. The remaining wavelengths are already contained in the coarse-graining of the field over the region V (which we do not discuss in this letter). For d = 1 cm (laboratory) one has N = IO2* while for d w IOzs cm (supercluster) one has N w IO. Since (S@O)* = 2&N < @ both cases are consistent <: with the observed sharpness of the gravitational constant, since the remaining coherence length is much smaller than & itself. We have not tackled in this paper the question of the back-reaction of the inhomogeneous degrees of freedom onto the mini-superspace background. Its calculation is most conveniently done with the help of the coarse-grained effective action [15,20]. We assume here that this back-reaction tums out to be small, at least for sufficiently big values of the scale factor. We conclude with some brief remarks on the possible relevance of &-coherence for Structure formation in the framework of extended inflation [ 171. In this scenario the inflaton field plays a rather passive role, in the sense that it sits quietly in the false vacuum and thus triggers inflation, whilst the quantum fluctuations of the IBD field are the seeds of galaxy formation. The question then arises: at which stage and for which wavelengths do these fluctuations become classical? The idea is to take into account the interaction of these fluctuations with a reservoir of other gravitational degrees of freedom which may force the JBD fluctuations to become classical and enable the formation of galaxies. We hope to return to this issue in a future publication. Acknowledgment This work was supported by the Swiss National Science Foundation.

References
[I] Coleman S 1988 Nucl. Phys. B 310 643

2518

C Kiefer and E A Martinez

[21 Jordan P 1949 Nature 164 637 Brans C and Dicke R H 1961 Phys. Rev. 124 925 (31 Gamy L J and Garcla-Bellido J 1992 Preprint EM-Fi-59192 gr-qc/92C9015 [41 Weinberg S 1972 Cmviiafion and CosfnoloRy (New Yak Wiley-Inferscience) p 245 [SI Zeh H D 1988 Phys. Len. 126A 31 I ; 1992 The Physicar Basis of fhc Dircction of Timr (Bedin: Springer) ch 6 Kiefer C 1988 Phys. Rev. D 38 1761 [61 Kiefer C 1989 Phys. Left. 2250 227 PI Callan C, Friedan D,Maninec E and Perry M 1 1985 Nud. Phys. B 262 593 [SI Callan C. Giddings S B, Harvey J A and Strominger A 1992 Phys. Rev D 45 RIO05 and references therein [9] Ashtekar A and Homwiu 0 T 1982 Phys. Rev. D 26 3342 [IO] del Campo S and Wenkin A 1989 Phys. Rev. D 40 688 Sanyal A K and Modak B 1992 Int. J. Mod. Phys. A 7 4039 [Ill Zeh H D 1986 Phys. Lea. A116 9 [I21 Kiefer C 1987 C/um Quanfum Crov. 4 1369: 1992 Phys. Rev. 046 1658 1131 Bohr N and Rosenfeld L 1933 Kgl, duns. Wd, Selshb. XI1 8 (141 Gell-Mann M and HarUe J B 1993 Phys. Rev. D 47 3345 u [IS] Sinha S and H B L 1991 Phys. Rev. D 44 1028 1161 Hdliwell J J and Hawking S W 1985 Phys. Rev. D 31 1777 [I71 Kolb E Sdopek D S and Tumer M S 1990 Phys. Rev. D 42 3925 . [I81 Barbur 1 B 1993 Phys. Rev. D 4 l 5422 Kiefer C 1993 Phys. Rev. D 41 5414 [I91 Par J P and Sinha S 1991 Phys. Rev. D 4 1038 4 [201 Morikawa M 1989 Phys. Rev. D 40 4023