1001 4311 PDF
1001 4311 PDF
1001 4311 PDF
David Craig
Parampreet Singh
E-mail: craigda@lemoyne.edu
E-mail: psingh@perimeterinstitute.ca
1
I. INTRODUCTION
Whatever form a theory of quantum gravity may ultimately take string theory, loop
quantum gravity, causal dynamical triangulations, causal sets, or something else1 extract-
ing coherent physical predictions will require more than just the basic technical apparatus of
the theory. Because one of the explicit aims of such theories is to provide a complete quan-
tum description of the universe as a whole, a framework within which consistent predictions
can be made in the absence of external observers or an external notion of measurement
will also be required in an essential way.
In this essay we aim to outline an explicit realization of one such framework, the gen-
eralized consistent histories quantum theory of J.B. Hartle [2], in the context of the
Wheeler-DeWitt quantization of a flat homogeneous and isotropic Friedman-Robertson-
Walker (FRW) model universe populated with a free, massless, minimally coupled scalar
field. (Previous discussions of decoherent histories formulations of quantum cosmological
models include [38].) We will describe when and how, within the framework of decoherent
histories, consistent quantum predictions may be extracted from the theory namely, when
the histories describing the possible outcomes decohere emphasizing the fundamental quan-
tum fact that it is not normally the case that a probability may sensibly be assigned to every
history of a system. We illustrate with an example of a prediction that depends crucially on
decoherence of the appropriate histories: whether or not Wheeler-DeWitt quantized FRW
universes are singular. The classical solutions of this model are all singular and fall into
disjoint sets of ever-expanding or ever-contracting universes. We shall show that decoher-
ence is critical to a coherent quantum analysis of this question, concluding unambiguously
within this framework that all quantum Wheeler-DeWitt FRW universes are singular as well.
This stands in contrast to a naive analysis of the behavior of the quantum wave function
alone, not taking care of the proper description and decoherence of the corresponding his-
tories. This (erroneous) analysis of states which are superpositions of wavefunctions peaked
on classically disjoint expanding and contracting solutions seems to suggest that quantum
avoidance of the singularity is possible. However, a careful analysis of the quantum histories
shows that the probability of avoidance of the singularity is zero and the probability that a
Wheeler-DeWitt quantum universe encounters a singularity is unity.
The organization of this paper is as follows. Section II describes the overall framework of
generalized quantum theory. In section III we give details of the cosmological model and its
(canonical) quantization, including a discussion of the models observables, and in section
IV we discuss the analysis of the quantum singularity. (Complete technical details of the
model and its decoherent histories formulation will be given elsewhere [9].)
Quite generally, the job of any quantum theory is to do two things. For any given physical
system, a quantum theory must, for all observable quantities, tell both What are the possible
values of those observables? and How likely is each of these values? In Hilbert space
formulations of quantum theory, where observables are represented by self-adjoint operators
{A} on a Hilbert space H, the answer to the first question is found in the eigenvalues {a}
1
See [1] for an overview.
2
of those operators, and to the second, in amplitudes constructed from the corresponding
eigenstates {|ai} and the state |(t)i of the system at the moment in question:
In the textbook Copenhagen view, this framework is usually supplemented with the rule
that once a particular outcome {a } has been observed, the state of the system is replaced
by that of the observed outcome, |(t)i {|a i}. This is commonly referred to as the
collapse of the wave function.
But when does it make sense to assign such probabilities, and thence reassign the state
of the system in this discontinuous, uncontrollable way? Certainly such amplitudes do not
always make sense. Most commonly, this is the case when considering the likelihood of
sequences of outcomes, as is well known, for example, in the two-slit experiment [10]. Quite
generally, in experiments of this kind, probabilities may be consistently assigned only when
there are physical mechanisms in place (such as a measuring apparatus or a suitable physical
environment) which destroy the interference between the possible alternative histories.
Taking a rather conservative approach, this idea has been formalized into a scheme known
as the decoherent or consistent histories formulation of quantum theory, in which prob-
abilities may be consistently assigned whenever interference between histories vanishes, ir-
respective of the presence of measuring devices, external observers, or other such notions;
see [2, 11, 12] for broad overviews. In fact, the principal virtue of this perspective is that it
employs criteria entirely internal to the system to determine whether or not the assignment
of probabilities makes coherent sense. 2
The most general formulation of the consistent histories program, suitable for application
to quantum cosmology, is due to J.B. Hartle [2]. In this form it is often referred to as
generalized quantum theory. We summarize very briefly its chief elements for theories
with a homogeneous time here.
Particular sequences of physical outcomes (histories) are characterized by class op-
erators, products of Heinseberg projections onto ranges of eigenvalues of observables of
interest:
1 2 n
Ch = Pa k1
(t1 )Pa k2
(t2 ) Pakn
(tn ) (2.2a)
1 2 n
= U(t0 t1 )Pa k1
U(t1 t2 )Pa k2
U(tn1 tn )Pakn
U(tn t0 ), (2.2b)
|h i = Ch |i. (2.3)
One may see that |h i is exactly the state one would assign to this system in Copenhagen
quantum theory were the observed outcomes to be the ranges aki ; the projections corre-
spond to the collapses. In a histories-oriented formulation, however, |h i is not the state
of the system, |i is. The apparent discontinuous evolution of the state in Copenhagen
quantum theory is the result of insisting on assigning a meaning to |(t)i separately along
every possible branch of the system.
2
By which it is meant that the appropriate probability sum rules are satisfied.
3
The probability of the history h is given by the Luders-von Neumann formula:
p(h) = hh | h i. (2.4)
P
Such a probability only makes sense, however meaning h p(h) = 1 when the interference
between the possible histories vanishes, hh | h i = 0 for h 6= h.
One therefore defines the decoherence functional
d(h, h) = hh | h i. (2.5)
When the decoherence functional is diagonal on a given set of histories,3 then that set is
said to decohere or be consistent. In such sets
The decoherence functional of a system thus provides an internally defined, objective mea-
sure of interference between the possible histories of a system, independent of notions of
external observers or measurements though it does reproduce the predictions of ordinary
quantum theory in appropriate measurement situations. It is the decoherence functional
that determines when the quantum theory makes a prediction concerning a possible history,
and, when a probability may be sensibly assigned, what that probability is.
where dq 2 is a fixed, flat fiducial metric on the spatial slices. Since the flat FRW universe
is spatially infinite we choose a fiducial cell with unit spatial volume with respect to dq 2
and restrict all integrations to it. Choosing for matter content a free, massless, minimally
coupled scalar field, the action for this model cosmology is
3 1
Z Z
2
S= dt (aa ) + dt a3 2 . (3.2)
8G 2
3
Complete in the sense that it consists in a set of mutually exclusive, collectively exhaustive histories.
4
Canonical phase space variables for the model are {a, pa , , p }, the scale factor and scalar
field and their conjugate momenta. A canonical transformation to the volume V = a3 and
pa
its conjugate = 4G 3 a2
(so {, V } = 4G) [14] puts the Hamiltonian constraint in the
form
4G 2
2V 2 + p 0, (3.3)
3
from which one sees that 2 measures the energy density p2 /2V 2 in the classical theory.
Classically, p is a constant of the motion, and
1 V
= ln + c (3.4)
12G Vc
are expanding and contracting branches of the classical solution. (Vc and c are integration
constants.)
It is shown in [14, 15] that this model may be canonically quantized by setting [, V ] =
i4G~ and p = i~ /. Defining a rescaled volume variable
V
= , (3.5)
( 4
3
)3/2 lp3 /K
p
where K = 2 2/3 3 3, the constraint becomes the Wheeler-DeWitt equation
2 (, ) = ()(, ), (3.6)
4
This interpretation may provide a convenient way of conceptualizing mathematical results, but is by no
means essential to the analysis or its conclusions.
5
The inner product and hence the physical Hilbert space may be found by demanding that
the action of these observables be self adjoint; it turns out to be [15]
Z
d
h | i = (, )(, ) . (3.10)
0
The same inner product can be obtained from a more rigorous group averaging procedure
[16, 17]. (See Ref. [18] for details.)
Because the dynamics (and hence Dirac observables) preserve the positive and negative
frequency subspaces in the physical Hilbert space, we can restrict to either subspace. Here we
choose to work with positive frequency solutions. Solutions to the Wheeler-DeWitt equation
may then be represented as a sum of expanding (R for right-moving in a plot of vs.
) and contracting (L for left-moving) orthogonal branches,
(, ) = R ( ) + L (+ ), (3.11)
where = ln 12G . (Negative frequency solutions can be written in a similar way).
Defining Heisenberg projections with the dynamics U(), class operators for histories
pertaining to values of physical quantities at specific values of may be defined similarly
to Eq. (2.2). The decoherence functional for this quantum cosmological model may then be
defined just as in Eq. (2.5) with the inner product given by Eq. (3.10) [5].5
One can then show within the framework of generalized quantum theory outlined above
that the probability (for the history in which) the volume is in a range when =
is given by
d
Z
p ( ) = |(, )|2 . (3.12)
A comment concerning this result is in order. The operator does not commute with
the quantum constraint, and is therefore not a Dirac observable [1921]. However, the
corresponding relational observable | is. It is satisfying that the probability for the history
in which assumes values in at gives precisely the result one would expect from
considering expectation values of | . One thus may see how consideration of relational
observables arises naturally within histories formulations of quantum theory. (Alternately,
one may consider class operators constructed directly from projections onto ranges of values
of the Dirac observables.) For further details see [9].6
5
The striking similarity of this formulation of generalized quantum theory for our cosmological model with
that of ordinary non-relativistic quantum mechanics described in section II is a consequence of two factors.
First, in this quantization a natural clock (namely, the scalar field ) appears in the model through
Eq. (3.7). Second, we are here restricting consideration to histories of alternatives defined by sequences of
values of physical quantities at definite values of that clock. More general categories of alternatives, and
therefore class operators, are certainly of potential physical relevance. We do not however consider them
here. Previous discussions of decoherent histories formulations of quantum cosmological models may be
found in [38]. These prior formulations generally lack, however, a well-defined underlying Hilbert space
and observables. On the other hand, some of them may be better suited to models with no emergent
clock.
6
An alternative approach demands that the class operators themselves must commute with the constraint
[3, 4, 7, 8].
6
IV. QUANTUM SINGULARITIES
We turn now to the question of whether this model quantum universe is singular, just
as the classical model is. There are many incommensurate criteria one might apply to this
question. We focus on a simple one: does the universe ever achieve zero volume?
For a universe whose state is purely left-moving (contracting) or right-moving (ex-
panding) (cf. Eq. (3.11)), this question is easy to answer in the affirmative. From Eq. (3.12)
it is straightforward to show that if = [0, ] and|i = |L i or |R i,
lim pR
() = 1 lim pR
() = 0 (4.1b)
+
for any fixed choice of , no matter how small. This is just as one may have expected: con-
tracting universes will inevitably shrink to arbitrarily small volume, and expanding universes
will equally inevitably have grown from arbitrarily small volume.
For this prediction, the role of decoherence was essentially trivial since decoherence for
alternatives defined only at a single moment of time () is automatic.7
Far more interesting, however, is the general case in which the state of the universe is a
superposition
|i = |L i + |R i (4.2)
of expanding and contracting universes. Here, pL = hL | L i and pR = hR | R i (with
hL | R i = 0 and pL +pR = 1) measure the amount of each in the superposition. One may
ask, what is the likelihood that a universe in such a Schrodingers Cat-like superposition
of (possibly macroscopic) states could jump from the contracting to the expanding branch
i.e. remain peaked on a large classical solution at both early and late values of ?
Indeed, such a quantum bounce may at first seem possible. As before, one may show
that
p () = pL () + pR
(). (4.3)
Now, though,
lim p () = pR and lim p () = pL . (4.4)
+
Thus, there is in general a non-zero probability for the universe to be non-singular in both
the past (pL ) and the future (pR ). Is this not precisely the possibility (with probability
pL pR ) of a quantum bounce we were seeking?
The answer is emphatically no. We are being misled because Eq. (4.3) is the answer to
a specific quantum question: What is the probability that the volume of the universe is in
at a given value of ? But the question of whether a quantum universe bounces is
really about two values of : What is the probability that the universe is not in at both
and +? The class operator for this question is
Cbounce = lim P (1 )P (2 ), (4.5)
1 1 2
2 +
7
This is because the corresponding class operators are simply projections; the branch wave functions for
different histories are thereby automatically orthogonal.
7
where the intervals are the complements of the intervals specifying small volume.
The complementary history
Csing = 1 Cbounce (4.6)
is the one for which the universe is inevitably singular i.e. assumes zero volume (enters )
at some point in its history.
Now, however, decoherence is not automatic, and it is not trivial. In general, questions
about the volume of the universe at two different values of make no more quantum sense
than the question of which slit the particle passed through in two-slit interference. The
question can only be given a coherent answer when the histories decohere, such as when
there is a measuring device in place.
In this particular case, however, it is possible to show with complete generality that in
the given limit and +, the histories {bounce, sing} do decohere. Indeed, in
this limit
|bounce i = Cbounce |i = 0 and |sing i = Csing |i = |i, (4.7)
so that
pbounce = 0 and psing = 1. (4.8)
We have thus shown, within an explicit framework for computing quantum probabilities
in an objective, observer-independent way, that a quantum bounce is not possible in these
models: the universe is inevitably singular.
V. DISCUSSION
ACKNOWLEDGMENTS
D.C. would like to thank the Perimeter Institute, where much of this work was com-
pleted, for its repeated hospitality. Research at the Perimeter Institute is supported by the
8
In general, these histories do not decohere at finite values of , even though in the quantum theory the
left- and right-moving sectors are orthogonal and preserved by the theorys Dirac observables [9].
8
Government of Canada through Industry Canada and the Province of Ontario through the
Ministry of Research & Innovation.
[1] C. Kiefer, Quantum gravity, 2nd ed. (Oxford University Press, Oxford, 2007).
[2] J. B. Hartle, in Gravitation and Quantizations, Proceedings of the 1992 Les Houches Sum-
mer School, edited by B. Julia and J. Zinn-Justin (North Holland, Amsterdam, 1995)
arXiv:gr-qc/9304006.
[3] J. B. Hartle and D. Marolf, Phys. Rev. D56, 6247 (1997), arXiv:9703021 [gr-qc].
[4] J. J. Halliwell and J. Thorwart, Phys. Rev. D65, 104009 (2002), gr-qc/0201070.
[5] D. A. Craig and J. B. Hartle, Phys. Rev. D, 123525 (June 2004), arXiv:gr-qc/0309117v3.
[6] C. Anastopoulos and K. Savvidou, Class. Quant. Grav. 22, 1841 (2005), arXiv:0410131 [gr-qc].
[7] J. J. Halliwell and P. Wallden, Phys. Rev. D73, 024011 (2006).
[8] J. J. Halliwell(2009), arXiv:0909.2597 [gr-qc].
[9] D. A. Craig and P. Singh, Consistent probabilities in wheeler-dewitt quantum cosmology,
(2010), in preparation.
[10] G. Greenstein and A. G. Zajonc, The quantum challenge: modern research on the foundations
of quantum mechanics, 2nd ed. (Jones and Bartlett, Sudbury, 2005).
[11] R. Omnes, The interpretation of quantum mechanics (Princeton University Press, Princeton,
1994).
[12] R. B. Griffiths, Consistent quantum theory (Cambridge University Press, Cambridge, 2008).
[13] D. A. Craig and P. Singh, Consistent probabilities in loop quantum cosmology, (2010), in
preparation.
[14] A. Ashtekar, A. Corichi, and P. Singh, Phys. Rev. D77, 024046 (2008),
arXiv:0710.3565 [gr-qc].
[15] A. Ashtekar, T. Pawlowski, and P. Singh, Phys. Rev. D74, 084003 (2006),
arXiv:gr-qc/0607039.
[16] D. Marolf(1995), arXiv:gr-qc/9508015.
[17] A. Ashtekar, J. Lewandowski, D. Marolf, J. Mourao, and T. Thiemann, J. Math. Phys. 36,
6456 (1995), arXiv:gr-qc/9504018.
[18] A. Ashtekar, T. Pawlowski, and P. Singh, Phys. Rev. D73, 124038 (2006),
arXiv:gr-qc/0604013 [gr-qc].
[19] C. Rovelli, Phys. Rev. D65, 124013 (2002), arXiv:gr-qc/0110035.
[20] D. Marolf, Class. Quant. Grav. 12, 1199 (1995), arXiv:gr-qc/9404053.
[21] B. Dittrich, Class. Quant. Grav. 23, 6155 (2006), arXiv:gr-qc/0507106.