Corichi1998, Ambiguities in Loop Quantization Area Vs Electric Charge

Modern Physics Letters A, Vol. 13, No.
17 (1998) 1339–1346

c World Scientific Publishing Company
AMBIGUITIES IN LOOP QUANTIZATION:

AREA VS. ELECTRIC CHARGE
ALEJANDRO CORICHI∗
Instituto de Ciencias Nucleares, Universidad Nacional Autónoma de México,
A. Postal 70-543, México D.F. 04510, México
KIRILL V. KRASNOV
Center for Gravitational Physics and Geometry, Department of Physics,
Penn State University, University Park, PA 16802, USA
Received 26 February 1998

Mod. Phys. Lett. A 1998.13:1339-1346. Downloaded from www.worldscientific.com
by UNIVERSITY OF QUEENSLAND on 05/13/13. For personal use only.
In this letter we compare the ambiguity that, as pointed out by Immirzi, arises in the
loop quantization of general relativity with a somewhat similar ambiguity in the quan-
tization of Maxwell theory. The “loop” quantization leads to a quantum theory in
which the fundamental excitations are loop-like rather than particle-like. Each such loop
plays the role of a quantized Faraday’s flux line. For the case of Maxwell theory, we show
that the quantization depends on an arbitrary choice of a parameter ε that carries the
dimension of electric charge. For each value of ε the electric charge that can be con-
tained inside a bounded spatial region is automatically quantized in units of ~/ε. The
requirement of consistency with the quantization of electric charge observed in our Uni-
verse fixes a value of the, so far arbitrary, parameter ε of the theory. We compare the
ambiguity in the choice of parameter ε with the β-ambiguity of quantum gravity, and
comment on the possible way this ambiguity can be fixed.
It is a fairly well-known result of the “loop” approach to quantum gravity1 that the
area of surfaces is quantized. The fundamental excitations of this quantum theory
are one-dimensional, loop-like, and the loops play the role of quantized
p flux lines:
each loop, labeled with spin j, contributes an area proportional to j(j + 1) times
the Planck area to the area of a surface it transversely intersects.2,3 It is, however,
considerably less well-known that the proportionality coefficient that fixes the area
spectrum of this theory is not predicted by the theory itself: it plays the role of a
free parameter. The reason for this is twofold. First, as it was initially pointed out
by Immirzi in Ref. 4, there is an ambiguity in the definition of an SU(2) connection
field A that is used to construct the quantum theory (see also Ref. 5). Possible
choices of the connection field A are labeled by a real parameter β. Starting from
different A, one gets non-equivalent quantum theories. In particular, the eigenvalues
∗ E-mail: corichi@nuclecu.unam.mx
1339
1340 A. Corichi & K. V. Krasnov
of operators measuring areas of surfaces happen to depend on the choice of 6 β

Xp
AS = 8πβG~ jv (jv + 1) . (1)
v
Here the sum is taken over all edges intersecting S transversally. Second, even if one
manages to fix β from some independent considerations, it is not granted that the
dimensionfull constant called G in (1) should be set equal to the physical Newton
constant.7 Indeed, G is a parameter of the quantum theory on the same footing
as, say, β. It can well happen that the predicted value of the constant governing
gravitational interactions between macroscopic bodies is a complicated function of
parameters G, β, etc. Then one would have to adjust the values of the parameters
of the theory in such a way that the predicted value of this “macroscopic” constant
is just equal to the physical Newton’s constant. Thus, a priori there is no reason
to expect G to be equal to Newton’s constant, but on dimensional grounds one can
expect G to be proportional to GN . Summarizing, one can say that the quantum

theory predicts that the area is quantized, the area eigenvalues are proportional to
GN ~ = lp2 , but the spectrum
Xp
AS = γGN ~ jv (jv + 1) (2)
v
depends on an additional physical parameter γ. Here γ is a dimensionless parameter

that takes care of both ambiguities in the area spectrum mentioned above.
In this letter we try to comprehend the physical significance of the ambiguity
represented by γ by comparing it with a somewhat similar ambiguity that arises in
the loop quantization of Maxwell theory for the charge spectrum. At the outset,
we wish to point out that none of the results we obtain on the loop quantization of
Maxwell field are quite new. The main result at which we will arrive, namely the
existence of a one-parameter family of loop quantizations of the free Maxwell theory,
seems to have been known already to, for example, Gambini and Trias,8 although,
to the best of our knowledge, it was never stated in this form. The purpose of this
letter is not to discover new facts about Maxwell theory, but to try to find a new
point of view on the old facts, a point of view that will also shed some light on the
physical significance of the γ-ambiguity in quantum gravity.
The quantization of the free Maxwell theory that considers the Wilson loop
functionals of the electromagnetic potential was first studied by Gambini and Trias
in Ref. 8. Since then, canonical quantization of Maxwell theory using loop variables
has been the subject of attention of many authors (see, for example, Ref. 9 and
references therein, and Ref. 10). Most of this research was concentrated, however,
on the use of loop variables and their Fock space representation (see, however,
Refs. 11 and 12). In this letter we shall be concerned with a different quantization
procedure. Namely, we consider a natural quantization coming from a choice of a
non-canonical algebra of observables in Maxwell theory. We shall use the recently
Ambiguities in Loop Quantization: Area vs. Electric Charge 1341
introduced spin network techniques; to the best of our knowledge such a treatment
for Maxwell theory does not exist in the literature.
Let us concentrate on the kinematics of the free Maxwell theory in the Hamilto-
nian framework. Let Σ denote the spatial hypersurface (which for technical reasons
is supposed to be a smooth manifold). Note that we do not assume any background
structure on Σ, and our entire discussion is, therefore, diffeomorphism invariant.
Let Aa be the electromagnetic potential. Throughout this letter we use the Gauss
system of units (in which there is a factor of 1/16π in front of the action of Maxwell
theory)p and assume the speed of light c = 1. In these units Aapcarries the dimen-
sion of M/L. Let Ẽ a be the electric field (of the dimension M/L3 ) that plays
the role of the canonically conjugate variable. Here “tilde” over the symbol of the
electric field indicates that it is a densitized vector field.
The phase space of the theory consists of pairs (Aa , Ẽ a ) satisfying appropriate
fall-off conditions at the spatial infinity in the case of a non-compact Σ. The Poisson
bracket between the canonically conjugate variables is given by

{Ab (x), Ẽ a (y)} = 4πδba δ 3 (x, y) , (3)

R
where δ 3 (x, y) stands for the δ-function defined so that Σ δ 3 (x, y)f (y)d 3 y = f (x).
The electromagnetic potential has a dimension different from the standard di-
mension 1/L of a connection field. In order to convert the electromagnetic potential
into a U(1) connection field, one has to introduce a dimensionfull parameter into
the theory. A possible way to do this √ is to introduce a parameter ε that carries
the dimension of electric charge [ε] = M L. Then Aa /ε has the dimension of a
connection. Thus, the set of fields εi Aa becomes the set of U(1) connection fields
on some U(1) bundle over Σ. One can use εi Aa to construct a holonomy hγ along
a path γ ∈ Σ Z
i
hγ := exp A . (4)
ε γ
Since the connection εi Aa is Abelian, we use the ordinary exponential instead of a
path ordered one. Note that so far ε is an arbitrary parameter with the dimension
of electric charge.
Let us now turn to the quantum theory. The algebra of observables we want
to regard as fundamental in the quantization procedure is the one generated by
holonomies hγ of the U(1) connection as configuration observables, and fluxes of
electric field E[S] as momenta. Recall that since Ẽ a is a vector density of weight
one, there is a naturally defined two-form Eab associated to it: Eab := ηabc Ẽ c . In
the same way that connection one-forms are objects that one can naturally integrate
along loops (in order to define a holonomy hγ ), oneR can naturally integrate two-forms
over surfaces. Therefore, the functions E[S] := S Eab dσab are the corresponding
momenta observables. They satisfy the following Poisson bracket relations,
i
{hγ , E[S]} = 4π hγ I(γ, S) , (5)
ε
where I(γ, S) denotes the oriented intersection number between the loop γ and the
surface S. This corresponds to the “loop-surface algebra” of Ref. 13. It is important
to note that there is a one-parameter family of such algebras, labeled precisely by
the parameter ε.
The so-called loop quantization of a gauge theory is constructed by taking the
traced holonomies of the connection as main configuration observables that de-
termine the quantum representation.14 The kinematics of the resultant quantum
theory was described in details in Ref. 15. For our purposes it is sufficient to recall
that there is a basis of quantum states given by spin networks, i.e. graphs embedded
in Σ with edges labeled by irreducible representations of the gauge group and the
vertices labeled by intertwining operators. In our case the gauge group is U(1) and
irreducible representations are in one-to-one correspondence with integers q (called
“charges”). Given a vertex of a spin network and a set of incoming and outgoing
edges, the intertwining operator exists if the sum of charges labeling the incoming
edges is equal to the sum of charges labeling the outgoing edges. In this case the
intertwining operator is unique up to a multiplicative constant factor.
We shall now see that the edges of spin networks play the role of quantized
flux lines of electric field. The smeared electric field observable E[S] becomes an
operator in the quantum theory. First, let us consider the canonical commutation
relations between the fundamental variables,
[Âb (x), Ê a (y)] = 4πi~δba δ 3 (x, y) . (6)
The standard way to satisfy the commutation relations is to represent Ê a (x), heuris-
tically, as a functional derivative with respect to Aa (x)
~ δ
Ê a (x) := 4π . (7)
i δAa (x)
The operator Ê[S] can then be promoted into a well-defined operator in the
Hilbert space using the regularization technique developed in Ref. 3. The resulting
operator is diagonal in the basis formed by spin network states
~ X 1 (u)
Ê[S] · |Ψi = 4π (q − qv(d) )|Ψi , (8)
ε v 2 v
where the sum on the right-hand side is taken over all vertices v of the spin network
(u) (d)
Ψ lying on the surface S, and qv , qv are the sum of all charges labeling edges
lying up and down the surface S respectively.
In the case when all vertices of Ψ lying on S are bi-valent (i.e. those coming
from simple intersections of edges of Ψ with S), the formula (8) simplifies
~X
Ê[S]|Ψi = 4π qv |Ψi , (9)
ε v
where qv are the charges labeling edges of Ψ intersecting S. It is straightforward to

check that the operators thus defined satisfy the quantum algebra coming from (5).
Thus, in our quantum theory the edges of spin networks indeed play the role of
the quantized flux lines of electric field: flux of electric field through S acquires value
via intersections with these edges, each edge labeled with charge q contributing a
flux equal to q times 4π~/ε for each transverse intersection.
Let us now consider the operator of electric charge. According to the Gauss’s
law, the total electric charge contained inside a closed surface S in Σ is determined
by the flux of electric field through S
1
QS = E[S] , (10)
4π
where S is a closed surface. The corresponding quantum operator Q̂S is given by
1
( 4π times) Eq. (8).
Let us analyze the spectrum of Q̂S . In the free Maxwell theory, which we were
considering so far, a charge contained inside a closed surface S in Σ is zero when
the topology of Σ is trivial. It is interesting to note, however, that even in the free
Maxwell theory one can have a nonzero charge inside a closed surface in the case
of a nontrivial topology of the spatial manifold (the condition is that H 2 (Σ) 6= 0).
This possibility, first pointed out by Misner and Wheeler,16 is fully realized in our
theory.
To have the possibility of a nonzero charge inside a closed surface in the case of
trivial topology for Σ, one has to include charged matter into the theory.
The most natural possibility would be to couple the theory to fermionic matter.
The corresponding quantum theory has been constructed in Refs. 12 and 17. How-
ever, we will not need the details of that construction here. Only the following
two points are important to us here. First, to couple the free Maxwell theory to
fermionic matter, one has to introduce an independent coupling constant e0 that
has the meaning of the (unrenormalized) fermion “electric charge”. The coupling
of matter to the Maxwell field is, of course, the standard minimal coupling in which
the “momentum” operator i~∂a is replaced by the covariant derivative operator
i~∂a − e0 Aa . The requirement that the “fermionic loop variable”
R
(i A)
ψfin e ε γ ψ̄in (11)
is invariant under gauge transformations
Aa → Aa − ~∂a χ ,
(12)
ψ → ψeie0 χ ,
renders
~
= e0 . (13)
ε
Thus, there is only one independent physical parameter in Maxwell theory coupled
to fermionic matter: the unrenormalized electric charge e0 of the fermions. The
parameter ε of the free Maxwell theory gets related to e0 via (13). Note that
a priori there is no reason to expect that the unrenormalized electric charge e0

coincides with the observed electric charge of the fermions. Indeed, we are used
to the fact that in perturbative QED, e0 undergoes an (infinite) renormalization.
Second, in the case when a fermionic charged matter is present in the theory, the
eigenvalues of the flux operator Ê[S], S being closed, are all eigenvalues one finds
in Eqs. (8). Thus, when charged matter is present in the theory, the electric charge
inside a closed surface S is not necessarily zero.
Therefore, as it can be seen from (9), the electric charge that can be contained
inside a closed surface S is quantized in our theory in integer multiples of the
“elementary” charge ē
~
ē = . (14)
ε
This holds both for the case of trivial and nontrivial topologies of Σ. In the case of
the trivial topology, and a charged matter present, the charge is quantized in units
of ē, which in this case is equal to e0 . A nontrivial topology of Σ charge contained

inside a closed surface can even be nonzero in the free Maxwell theory. In this case
charge is again quantized in units of ē. Thus, first, the electric charge is quantized
not in units of charge ε, but in units of some different charge ē. Second, in the
case fermionic matter is present, the quantum of electric charge ē coincides with
the parameter of the theory e0 . Thus, no renormalization occurs.
As we have shown, the quantization of the theory of the type adopted here
automatically guarantees that the electric charge is quantized. Note, however, that
in nature the electric charge is known to be quantized in the units of the charge e of
the electron (or, as in the standard model of elementary particles, in units of e/3).
Thus, the requirement that the theory is consistent with observations must fix ē = e
(or, possibly, ē = e/3). Two remarks are in order. Note that Eq. (14) is very similar
in form to Dirac’s quantization condition18 : qg = 2πm, m ∈ Z, where in the analogy
ε is related to the magnetic charge g. However, the obtained charge quantization
is a natural consequence of loop quantization without the necessity to introduce
a magnetic charge nor a test field as is needed in Dirac’s derivation (for details
see Ref. 19 and references therein). Secondly, an independent derivation of charge
quantization, when working in the strong coupling regime on a lattice, was given by
Polyakov (see for instance Ref. 20). Although similar in form, loop quantization is
a nonperturbative formalism defined on the continuum, without taking any limit.
To summarize, let us say that the loop quantization of Maxwell theory depends
on an arbitrary choice of the parameter ε having the dimension of electric charge.
The quantum theory, however, predicts that the charge is quantized in units of ~/ε,
and the requirement of consistency with the quantum of charge that is observed in
nature determines ε unambiguously.
We can now return to the main interest of our letter: the ambiguity in the area
spectrum of loop quantum gravity, and compare it with the ambiguity in Maxwell
theory previously mentioned. First, note that the two ambiguities have technically
very different origins. Indeed, the ambiguity in the case of Maxwell theory arises
because the “scale” of the connection used to construct the loop variables is free.
On the other hand, as it was emphasized by Rovelli and Thiemann,5 the SU(2)
connection in the case of gravity is not free to re-scale, and the origin of the β
ambiguity lies in the fact that there are two different connections available (see
Ref. 5). However, conceptually, the two ambiguities are similar, because in both
cases the spectrum of a physical important operator is not fixed by the theory itself
but depends on an additional free physical parameter. It is interesting to note that
in both cases of area and charge, the ambiguity can be reduced to the arbitrariness
of the value of the smallest nonzero quantum of the quantity of interest as expressed
in terms of the fundamental constants of the theory. Indeed, in quantum gravity the
ambiguity in the area spectrum would be fixed if one knew the smallest quantum of
area in the theory in terms of the constants GN , ~: the only fundamental constants
in the theory. In the case of Maxwell theory the ambiguity is fixed by relating the
quantum of charge to ~, the only fundamental constant.
The comparison with the Maxwell case also suggest a way the ambiguity (2)
should be fixed. In the case of Maxwell theory the problem was resolved by a
requirement of consistency with the observed quantization of electric charge. Thus,
ideally, to fix the area spectrum ambiguity of quantum gravity one would simply
have to measure the smallest quantum of area in nature in terms of GN ~. Of course,
this does not seem to be a very practical way to fix the problem because of the tiny
value of lp2 as compared with our ordinary scale. Thus, the only practical way to
fix the area ambiguity has to be indirect.
Let us conclude this letter by pointing out a possible indirect way the area
spectrum ambiguity in quantum gravity can be fixed. Since a direct experimental
check of the spectrum (1) does not seem to be possible nowadays, one would have
to find a macro-scale consequence of the theory that depends on the detailed form
of the area spectrum. Then, comparing such a prediction of the quantum theory
with experiment, or with results predicted by other well-established theories, one
could fix the value of β from the requirement of consistency. As such a result
of the quantum theory one can take, for example, the recent statistical mechanical
calculations of black hole entropy based on the formalism of loop quantum gravity.21
These calculations yield for the statistical mechanical entropy of Schwarzschild black
hole S = cA/γlp2 , where c is the dimensionless constant determined by calculations.
A comparison of this result with the Bekenstein–Hawking entropy S = A/4lp2 can
be used to determine a value of parameter γ = 4c. This provides one with a possible
way the parameter γ can be fixed in quantum gravity.
Acknowledgments
The authors would like to thank all the participants of the Penn State Mini-
workshop on Quantum Gravity, February 1997, for stimulating discussions that clar-
ified the analogy between the ε-ambiguity in Maxwell theory and the β-ambiguity
in quantum gravity. A.C. was supported by Universidad Nacional Autónoma de
México (DGAPA, UNAM). K.K. was supported, in part, by the Braddock fellowship
of Penn State University. This work was in part supported by NSF-grant PHYS
95-14240 and the Eberly research fund of Penn State.
References
1. For a recent review see: C. Rovelli, “Loop quantum gravity”, gr-qc/9710008.
2. C. Rovelli and L. Smolin, Nucl. Phys. B442, 593 (1995); B456, 734(E) (1995).
3. A. Ashtekar and J. Lewandowski, Class. Quantum Grav. 14, A55 (1997).
4. G. Immirzi, Nucl. Phys. Proc. Suppl. 57, 65 (1997); G. Immirzi, Class.Quantum Grav.
14, L177 (1997).
5. C. Rovelli and T. Thiemann, Phys. Rev. D57, 1009 (1998).
6. K. Krasnov, Class. Quantum Grav. 15, L1 (1998).
7. T. Jacobson, seminar given at Vienna Workshop on Mathematical Problems of Quan-
tum Gravity, July 1996.
8. R. Gambini and A. Trias, Phys. Rev. D22, 1380 (1980); C. Di Bartolo, F. Nori,
R. Gambini and A. Trias, Lett. Nuovo Cimento 38, 497 (1983).
9. R. Gambini and J. Pullin, Loops, Knots, Gauge Theory and Quantum Gravity
(Cambridge Univ. Press, 1996).

10. A. Ashtekar and A. Corichi, Class. Quantum. Grav. 14, A45 (1997).
11. R. Gambini and J. Pullin, Phys. Rev. D47, R5214 (1993).
12. K. Krasnov, Phys. Rev. D53, 1874 (1996).
13. A. Ashtekar and C. Isham, Phys. Lett. B274, 393 (1992).
14. A. Ashtekar and C. Isham, Class. Quantum Grav. 9, 1433 (1992).
15. A. Ashtekar, J. Lewandowski, D. Marolf, J. Mourao and T. Thiemann, J. Math. Phys.
36, 6456 (1995).
16. C. W. Misner and J. A. Wheeler, Ann. Phys. (N. Y.) 2, 525 (1957).
17. J. C. Baez and K. V. Krasnov, J. Math. Phys. 39, 1251 (1998)
18. P. A. M. Dirac, Proc. R. Soc. London A133, 60 (1931).
19. P. Goddard and D. I. Olive, Rep. Prog. Phys. 41, 91 (1978).
20. A. M. Polyakov, Gauge Fields and Strings (Harwood, 1987).
21. C. Rovelli, Phys. Rev. Lett. 77, 3288 (1996); K. V. Krasnov, Phys. Rev. D55, 3505
(1997); K. V. Krasnov, Gen. Rel. Grav. 30, 53 (1998).
This article has been cited by:
1. O. P. Dimitriev. 2011. Spin motion versus monopole in the charge quantization problem. Physics Essays 24:4, 512-517. [CrossRef]
2. Hanno Sahlmann. 2011. Black hole horizons from within loop quantum gravity. Physical Review D 84:4. . [CrossRef]
3. E Minguzzi, C Tejero Prieto, A López Almorox. 2006. Weak gauge principle and electric charge quantization. Journal of Physics
A: Mathematical and General 39:30, 9591-9610. [CrossRef]
4. P. Arias, E. Fuenmayor, Lorenzo Leal. 2004. Interacting particles and strings in path and surface representations. Physical Review
D 69:12. . [CrossRef]
5. Alejandro Corichi. 2003. Quasinormal modes, black hole entropy, and quantum geometry. Physical Review D 67:8. . [CrossRef]
6. Madhavan Varadarajan. 2002. Gravitons from a loop representation of linearized gravity. Physical Review D 66:2. . [CrossRef]
7. Ernesto Fuenmayor, Lorenzo Leal, Ryan Revoredo. 2002. Loop representation of charged particles interacting with Maxwell and
Chern-Simons fields. Physical Review D 65:6. . [CrossRef]
8. Martin Bojowald. 2001. The semiclassical limit of loop quantum cosmology. Classical and Quantum Gravity 18:18, L109-L116.
[CrossRef]
9. Madhavan Varadarajan. 2001. Photons from quantized electric flux representations. Physical Review D 64:10. . [CrossRef]
10. Joseph Samuel. 2001. Comment on “Immirzi parameter in quantum general relativity”. Physical Review D 64:4. . [CrossRef]
11. M Bojowald, H A Kastrup. 2000. Symmetry reduction for quantized diffeomorphism-invariant theories of connections. Classical
and Quantum Gravity 17:15, 3009-3043. [CrossRef]
12. Martin Bojowald. 2000. Abelian BF-theory and spherically symmetric electromagnetism. Journal of Mathematical Physics 41:7,
4313. [CrossRef]

Corichi1998, Ambiguities in Loop Quantization Area Vs Electric Charge

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Corichi1998, Ambiguities in Loop Quantization Area Vs Electric Charge

Uploaded by

Copyright:

Available Formats

Modern Physics Letters A, Vol. 13, No.

AMBIGUITIES IN LOOP QUANTIZATION:

Received 26 February 1998

of operators measuring areas of surfaces happen to depend on the choice of 6 β

expect G to be proportional to GN . Summarizing, one can say that the quantum

depends on an additional physical parameter γ. Here γ is a dimensionless parameter

bracket between the canonically conjugate variables is given by

{Ab (x), Ẽ a (y)} = 4πδba δ 3 (x, y) , (3)

[Âb (x), Ê a (y)] = 4πi~δba δ 3 (x, y) . (6)

where qv are the charges labeling edges of Ψ intersecting S. It is straightforward to

is invariant under gauge transformations

a priori there is no reason to expect that the unrenormalized electric charge e0

of ē, which in this case is equal to e0 . A nontrivial topology of Σ charge contained

(Cambridge Univ. Press, 1996).

You might also like