The Lorentz-violating extension

of the Standard Model

Ralf Lehnert

CENTRA, Department of Physics, University of the Algarve, 8000-117 Faro, Portugal

Abstract. Quantum-gravity effects are expected to be suppressed by the Planck mass.

For experimental progress it is therefore important to identify potential signatures from
Planck-scale physics that are amenable to ultrahigh-precision tests. It is argued that
minuscule violations of Lorentz and CPT symmetry are candidate signals. In addition,
arXiv:hep-ph/0401124 v1 16 Jan 2004

theoretical and experimental aspects of the Standard-Model Extension, which describes

the emergent low-energy effects, are discussed.

1 Motivation
An important open question in our understanding of nature at its fundamental
level concerns a unified quantum description of all fundamental interactions in-
cluding gravity. Such a theory is likely to become dominant only as the Planck
scale is approached, so that quantum-gravity effects are expected to be minus-
cule at presently attainable energies. Moreover, the absence of a fully realistic
and viable candidate underlying theory provides a major obstacle for the iden-
tification of concrete quantum-gravity signatures that can be searched for in
present-day or near-future experiments.
A possible approach to overcome this phenomenological issue is to deter-
mine exact relations in the currently accepted laws of physics that may be vio-
lated in prospective fundamental theories and that can be tested with ultrahigh-
precision. Symmetries typically satisfy these criteria. For example, Lorentz and
CPT invariance are cornerstones of our present understanding of nature at the
fundamental level, and a variety of Lorentz and CPT tests belong to the most
sensitive null experiments available. Lorentz and CPT violation is also a key
feature of certain approaches to underlying physics.
For example, couplings varying on cosmological scales are one possible source
of Lorentz and CPT violation [1]. This fact does not come as a surprise: param-
eters dependent on spacetime break translational invariance, and translations,
rotations, and boosts are linked in the Poincaré group. Thus, violations of trans-
lation symmetry can also affect Lorentz invariance. This can be understood
intuitively as follows. The equations of motion typically contain the gradient of
the coupling, which selects a preferred direction in spacetime leading to apparent
Lorentz violation.
In this talk, we discuss some theoretical and experimental aspects of the
Standard-Model Extension (SME) [2–7], which is the low-energy framework for
Lorentz-breaking effects. The SME is a dynamical model constructed to con-
tain all Lorentz- and CPT-violating lagrangian terms consistent with coordinate
2 Ralf Lehnert

independence, which is a fundamental requirement to be discussed below. To

date, numerous Lorentz- and CPT-violation tests involving hadrons [8–21], pro-
tons and neutrons [22–31], electrons [31–41], photons [42–47], muons [48–50],
and neutrinos [2,51–54] have been analysed or identified within the framework
of the SME.
The outline of this talk is as follows. In Sec. 2, we analyse the requirement
of coordinate independence and its implementation. Section 3 contains a dis-
cussion of the SME. In particular, its construction is reviewed, its generality is
addressed, and its advantages are summarized. In Sec. 4, varying couplings are
investigated from the perspective of providing a potential source of Lorentz and
CPT violation. Section 5 comments on kinematical Lorentz tests with modified
dispersion relations. The conclusions are contained in Sec. 6.

2 Coordinate independence

One of the most fundamental principles in physics is coordinate independence.

The need for this principle in the presence of Lorentz breaking is well established
[2,3,55], and it has served as the basis for the construction of the SME. How-
ever, in some investigations of Lorentz and CPT violation coordinate-dependent
physics emerges, and occasionally Lorentz-symmetry breakdown is identified
with the loss of coordinate independence. For these reasons, it is appropriate to
review this fundamental principle and its implementation. It then also becomes
clear how coordinate independence provides a rough classification of different
approaches to Lorentz and CPT breaking.
A certain choice of labeling events in space and time is called a coordinate
system. Such a labeling scheme is typically observer dependent and thus arbi-
trary to a large extent. Coordinate systems belong to the most common and
important tools for the description of processes occurring in nature, but they fail
to possess physical reality: the choice of coordinates must leave the physics un-
affected. This principle of coordinate independence is fundamental in science. It
assures that the physics remains independent of the observer, and it is therefore
also called observer invariance. Coordinate independence is guaranteed when
spacetime is given a manifold structure and physical quantities are represented
by geometric objects, such as tensors or spinors.
Coordinate-dependent physics does break Lorentz symmetry. However, the
converse (i.e., Lorentz violation is associated with the loss of coordinate in-
dependence) is a common misconception. The principle of coordinate invari-
ance is, in fact, independent from Lorentz symmetry. For instance, Newton’s
law of gravitation and nonrelativistic classical mechanics are non-invariant un-
der Lorentz transformations but can be formulated in the coordinate-free lan-
guage of 3-vectors. The Lorentz transformations acquire a significant role only
on lorentzian spacetime manifolds where they implement changes between local
inertial frames.
Even on a lorentzian manifold, Lorentz symmetry may be broken. This fact
can be illustrated in the conventional context of a classical point particle of mass
 The Lorentz-violating Standard-Model Extension 3

m and charge q subjected to an external electromagnetic field F µν . The equation

of motion for such a particle reads
dv µ
m = qF µν vν , (1)
dτ
where τ is the particle’s proper time and v µ is its 4-velocity. Equation (1) re-
mains valid in all coordinate systems because it is a tensor equation. Thus,
observer Lorentz symmetry is maintained. However, the external F µν back-
ground violates, for example, symmetry under arbitrary rotations of the charge’s
trajectory. Among the consequences of this noninvariance is the violation of
angular-momentum conservation for the particle. Note the difference to coor-
dinate changes, which leave unaffected the physics: here, only the trajectory
is rotated, so that its orientation with respect to F µν can change. One then
says that particle Lorentz symmetry is violated, despite the presence of observer
invariance [2,55]. It is important to point out that in the above example, the
background F µν is a local electromagnetic field caused by other 4-currents that
can in principle be controlled. Our external-field illustration therefore fails to
contain Lorentz violation as a fundamental property of an effective vacuum.
The above discussion suggests the possibility of classifying different approaches
to Lorentz violation by their behavior under coordinate changes. In the remain-
ing part of this section, we discuss such a classification.
Models with coordinate-dependent physics. Although it appears to be
impossible to perform meaningful scientific investigations involving coordinate-
dependent physics, such approaches to Lorentz breaking have been considered
in the literature. More specifically, there have been two suggestions in the con-
text of neutrino phenomenology: the first one forces the masses of particle and
antiparticle to be different [56], while the second one attempts to build a model
from positive-energy eigenspinors only [57]. Both approaches are known to in-
volve coordinate-dependent off-shell physics [58,59]. In what follows, we do not
consider these approaches further.
Coordinate-covariant models involving non-lorentzian manifolds.
Another possibility to speculate how Lorentz invariance might be lost is the fol-
lowing. Local inertial frames have a structure different from the usual minkow-
skian one, so that Lorentz transformations no longer implement changes between
inertial coordinates, i.e., observer Lorentz invariance is replaced by observer in-
variance under some other symmetry transformation. But nevertheless, coordi-
nate independence is maintained. This point of view is taken in the so called
“doubly special relativities” [60,61]. We mention that both the viability and the
physical interpretation of this approach appear to be controversial at the present
time [62–66]. We leave such Lorentz-symmetry deformations unaddressed in the
present work.
Coordinate-independent models involving nontrivial vacua. In this
approach, a fully Lorentz-covariant underlying model generates a tensorial back-
ground resulting in apparent Lorentz violation. The basic idea parallels that of
our above external-field example. However, the background is outside of experi-
mental control and must be viewed as a property of the effective vacuum. Because
4 Ralf Lehnert

of the lorentzian structure of the underlying manifold and the usual Lorentz-
covariant dynamics at the fundamental level, this approach appears closest to
established theories. The physical effects in such models are perhaps compa-
rable to those inside certain crystals: the physics remains independent of the
chosen coordinates, but particle propagation, for example, can be direction de-
pendent. As an immediate consequence, one can locally still work with the metric
η µν = diag(1, −1, −1, −1), particle 4-momenta are still additive and still trans-
form in the usual way under coordinate changes, and the conventional tensors
and spinors still represent physical quantities.
Such nontrivial effective vacua can arise in various theories beyond the Stan-
dard Model. For instance, the possibility of spontaneous Lorentz and CPT break-
ing in the framework of string field theory was discovered more than a decade
ago [67–74]. Subsequent studies have considered other mechanisms for Lorentz-
violating vacua including spacetime foam [75,76], nontrivial spacetime topology
[77], loop quantum gravity [78,79], realistic noncommutative field theories [80–
83], and spacetime-varying couplings [1,84].

3 The Standard-Model Extension

The next step after determining general low-energy features is the identifica-
tion of specific experimental signatures for Lorentz breaking and the theoretical
analysis of Lorentz-violation searches. This task is most conveniently performed
within a suitable test model. Many Lorentz tests are motivated and analysed in
purely kinematical frameworks allowing for small violations of Lorentz symme-
try. Examples are Robertson’s framework, its Mansouri-Sexl extension, the c2
model, and phenomenologically constructed modified dispersion relations. But
is also clear that the implementation of general dynamical features significantly
increases the scope of Lorentz tests. For this reason, the SME mentioned in
the introduction has been developed. However, the use of dynamics in Lorentz-
violation searches has recently been questioned on the grounds of framework
dependence. We disagree with this claim and begin with a few arguments in
favor of a dynamical test model.
Such a model is constrained by the requirement that known physics must
be recovered in certain limits, despite some freedom in introducing dynamical
features compatible with a given set of kinematics rules. In addition, it seems
difficult and may even be impossible to construct an effective theory contain-
ing the Standard Model with dynamics significantly different from that of the
SME. We also mention that kinematical analyses are limited to only a subset
of potential Lorentz-violating signatures from fundamental physics. From this
viewpoint, it is desirable to explicitly implement dynamical features of sufficient
generality into test models for Lorentz and CPT symmetry.
The generality of the SME. In order to understand the generality of the
SME, we review the main elements of its construction [2,3]. Starting from the
usual Standard-Model lagrangian LSM one adds Lorentz-violating modifications
 The Lorentz-violating Standard-Model Extension 5

δL:
LSME = LSM + δL . (2)
Here, the SME lagrangian is denoted by LSME . The correction δL is formed by
contracting Standard-Model field operators of any dimensionality with Lorentz-
breaking tensorial coefficients that describe the nontrivial vacuum discussed in
the previous section. To ensure coordinate independence, this contraction must
yield observer Lorentz scalars. It becomes thus apparent that all possible con-
tributions to δL determine the most general effective dynamical description of
Lorentz violation at the level of observer Lorentz-invariant quantum field theory.
Potential Planck-scale features, such as non-pointlike elementary particles or
a discretized spacetime, are unlikely to invalidate the above effective-field-theory
approach at presently attainable energies. On the contrary, the phenomenolog-
ically successful Standard Model is normally understood as an effective-field-
theory approximation for more fundamental physics. If fundamental physics in-
deed exhibits minuscule Lorentz-breaking effects, it would seem contrived to
consider low-energy effective models outside the framework of quantum field
theory. We finally mention that the need for a low-energy description beyond
effective field theory is also unlikely to arise in the context of candidate under-
lying models with novel Lorentz-invariant aspects, such as additional particles,
new symmetries, or large extra dimensions. Lorentz-symmetric modifications can
therefore be implemented into the SME, if necessary [85–87].
Advantages of the SME. The SME permits the identification and direct
comparison of essentially all currently feasible experiments searching for Lorentz
and CPT violation. In addition, certain limits of the SME correspond to classi-
cal kinematics test models of relativity (such as the aforementioned Robertson’s
framework, its Mansouri-Sexl extension, or the c2 model) [44]. Another advan-
tage of the SME is the possibility of implementing additional desirable condi-
tions besides coordinate independence. For instance, one can choose to require
translational invariance, SU(3)×SU(2)×U(1) gauge symmetry, power-counting
renormalizability, hermiticity, and pointlike interactions. These conditions fur-
ther restrict the parameter space for Lorentz breaking. One can also adopt sim-
plifying choices, such as rotational invariance in certain coordinate systems. This
latter assumption together with additional simplifications of the SME has been
discussed in Ref. [51].

4 Varying couplings and the SME

In this section, we construct a classical cosmological solution in the framework of

the pure N = 4 supergravity in a four-dimensional spacetime. We show that this
solution leads to a spacetime variation of both the fine-structure parameter α
and the electromagnetic θ angle. Such a model fails to be fully realistic in detail,
but it is a limit of the N = 1 supergravity in 11 dimensions, which is contained
in M-theory. It could therefore give insight into generic features expected to
arise in a promising candidate underlying theory. We illustrate the associated
6 Ralf Lehnert

Lorentz-violating effects by looking at the θ-angle variation, which gives rise to

µ
the (kAF ) term contained in the SME.
Basics of the model. In Planck units, the bosonic lagrangian for our N = 4
supergravity in four dimensions takes the following form [88]:
 
√ 1 1 µν 1 µν ∂µ A∂ µ A + ∂µ B∂ µ B
L = g − 2 R − 4 M Fµν F − 4 N Fµν F̃ + . (3)
4B 2
Here, gµν represents the metric, and we have assumed that only one graviphoton,
Fµν , is excited. The dual F̃ µν = εµνρσ Fρσ /2 is defined as usual. The model
also contains two scalars A and B that can be identified with an axion and a
dilaton. The dependence of the couplings M and N on the scalars is fixed by
the supergravity framework [1]:

B(A2 + B 2 + 1) A(A2 + B 2 − 1)
M= , N= . (4)
(1 + A2 + B 2 )2 − 4A2 (1 + A2 + B 2 )2 − 4A2
In the pure N = 4 supergravity in four dimensions, the graviphoton couples
nonminimally to matter. Although the internal SO(4) symmetry can be gauged
[89], we adopt a phenomenological approach: in a realistic situation, the vector-
matter interaction must be minimal. In what follows, we therefore can identify
the graviphoton Fµν with the electromagnetic field.
Supergravity cosmology. Next, we consider the above model in a cosmo-
logical context. We begin with the standard assumption of a homogeneous and
isotropic universe. This implies that Fµν ≃ 0 on cosmological scales. We further
take the universe to be flat, i.e., k = 0. This is justified in light of recent mea-
surements [90–92]. The Friedmann-Robertson-Walker (FRW) line element has
the conventional form:

ds2 = dt2 − a2 (t)(dx2 + dy 2 + dz 2 ). (5)

Here, the usual comoving coordinates have been adopted and a(t) denotes the
cosmological scale factor. As a consequence of the above assumptions, not only
the scale factor, but also A and B depend on t only.
For phenomenological reasons it is necessary to model the known matter
content of the universe. We employ a standard approach and incorporate the
energy-momentum tensor of dust, Tµν , into our framework. If uµ is the unit
timelike vector orthogonal to the spatial hypersurfaces and ρ(t) is the energy
density of the dust, the usual arguments imply that Tµν = ρuµ uν . In the present
model, this type of matter arises from the fermionic sector of our supergravity
framework. At tree level, the scalars A and B do not couple to the fermions [88],
so that we can take Tµν as conserved separately.
It turns out that the equations of motion for our supergravity cosmology can
be integrated analytically. For example, the time dependences of A and B are
given by [1]
   
1 1 1 1
A = ±λ tanh − + A0 , B = λ sech − , (6)
τ τ0 τ τ0
 The Lorentz-violating Standard-Model Extension 7

where λ, 1/τ
√ 0 , and A0 are
p integration constants. The parameter time τ is defined
by τ = 3/4 arcoth( 3cn /4c1 t + 1), which contains two more integration
constants cn and c1 . The solution (6) implies that both A and B approach
constant values at late times t → ∞. Thus, the values of the axion A and the
dilaton B become fixed in our supergravity cosmology, despite the absence of a
dilaton potential. This is essentially a consequence of energy conservation.
Varying couplings. Next, we consider excitations of Fµν in the axion-
dilaton background determined by Eq. (6). From a phenomenological viewpoint,
we can take these excitations to be localized in spacetime regions small on cos-
mological scales. It is therefore appropriate to work in local inertial frames.
The conventional electrodynamics lagrangian in inertial coordinates can be
taken as
1 θ
Lem = − 2 Fµν F µν − Fµν F̃ µν , (7)
4e 16π 2
where we have allowed for a nontrivial θ-angle. The electromagnetic coupling
is denoted by e. Comparison with our supergravity lagrangian (3) shows that
we can identify e2 ≡ 1/M and θ ≡ 4π 2 N . It is important to note that M
and N are determined by the axion-dilaton background (6), so that e and θ
become functions of the comoving time t. In arbitrary local inertial frames, the
electromagnetic coupling and the θ angle therefore exhibit related spacetime
dependences.
We remark in passing that the functional dependence of the fine-structure
parameter α = e2 /4π on the comoving time can vary qualitatively with the
choice of model parameters. Figure (1) displays relative variations ∆α/α of the
fine-structure parameter versus fractional look-back time 1−t/tn to the big bang.
Here, tn denotes the present age of the universe and ∆α = α(t) − α(tn ). We have
set the parameter 1/τ0 to zero. The solid line corresponds
p to a constant α. Each
broken line represents a set of nontrivial choices for λ, 3cn /4c1 tn , and A0 . Input
parameters leading to a variation consistent with the Oklo constraints [95–97]
are labeled with an asterisk. The qualitative differences in the various plots, the
nonlinear features, and the sign change for α̇ in the two cases with positive A0 are
apparent. Figure (1) also depicts the recent experimental results [94] obtained
from measurements of high-redshift spectra over periods of approximately 0.6tn
to 0.8tn assuming H0 = 65 km/s/Mpc and (Ωm , ΩΛ ) = (0.3, 0.7).
Lorentz violation. The Lorentz violation associated with varying couplings
becomes perhaps most transparent in the equations of motion. Incorporating
charged matter described by a 4-current j ν , lagrangian (3) yields in a local
inertial frame:
1 2 1
∂µ F µν − 3 (∂µ e)F µν + 2 (∂µ θ)F̃ µν = j ν . (8)
e2 e 4π
Note that in the limit of spacetime-constant e and θ, the conventional inho-
mogeneous Maxwell equations are recovered. In the axion-dilaton background
(6), however, the terms containing the gradients of e and θ are associated with
apparent Lorentz violation: since the gradients can be treated as effectively non-
dynamical and constant on small cosmological scales, they select a preferred
8 Ralf Lehnert

0.5

Da/a ( x10 -5 ) - 0.5

-1

- 1.5

0 0.2 0.4 0.6 0.8 1

fractional look-back time

l 3 cn / 4 c1 tn A0

*) 2p/137 3 x 10 5 1 _ (2p/137) 2
5 _
*) 2p/137 3 x 10 1 _ (2p/137) 2
_4 3 _
(1 + 5.245 x 10 ) 2p/137 2 x 10 1 _ (2p/137) 2 + 9.3 x 10 4

Fig. 1. Relative change in the fine-structure parameter versus fractional look-back time
to the big bang for various choices of integration constants [93]. Note the qualitative
differences and nonlinear features. Also shown are the Webb data [94].

direction in the local inertial frame. As a result, symmetry under boosts and
rotations of electromagnetic fields is broken. Note that this type of Lorentz vio-
lation is not a feature of the particular coordinate system chosen. If a gradient is
is nonzero in one local inertial frame associated with a small spacetime region,
it is nonzero in all local inertial frames associated with the region in question.
By contrast, such a Lorentz-symmetry breakdown is absent in conventional
FRW cosmologies that fail to contain spacetime-dependent scalars. Although
global Lorentz invariance is usually violated, local Lorentz-symmetric inertial
frames always exist. It is also important to note that the above source for
Lorentz-violating effects is not a unique feature of our supergravity cosmology.
Equation (8) illustrates that any smooth spacetime dependence of the couplings
e and θ on cosmological scales leads to such effects. It is theoretically attractive
to associate varying couplings with quantum scalar fields acquiring spacetime-
dependent expectation values. However, from the perspective of Lorentz viola-
tion, classical scalars can be employed equally well. In fact, the variation of the
coupling need not necessarily be driven by dynamical fields at all. This suggests
that the above type of Lorentz breaking is a common feature of any model with
spacetime-varying couplings.
Next, we study how the effects of this mechanism fit into the framework of
the SME and how our supergravity model helps to resolve conceptual issues in
 The Lorentz-violating Standard-Model Extension 9

quantum field theories incorporating Lorentz breaking. This is best illustrated

by considering the θ-angle term. An integration by parts yields an equivalent
form of the electrodynamics lagrangian (7) in a local inertial frame:
1 1
L′em = − Fµν F µν + 2 (∂µ θ)Aν F̃ µν . (9)
4e2 8π
The the last term on right-hand side of Eq. (9) gives a Chern-Simons-type con-
tribution to the action. Such a term is contained in the minimal SME, and we
can identify (kAF )µ ≡ e2 ∂µ θ/8π 2 . The presence of a nonzero (kAF )µ in Eq. (9)
demonstrates explicitly Lorentz and CPT breaking at the lagrangian level.
The case of constant e and (kAF )µ has been discussed extensively in the
literature [42,3,4,77,98]. Then, lagrangian (9) becomes translationally invariant
resulting in an associated conserved energy that fails to be positive definite.
This usually leads to instabilities in the theory, and the question arises how this
problem is avoided in the present context of our positive-definite supergravity
model.1
Although in most models a Chern-Simons-type term is assumed to arise in a
fundamental theory, (kAF )µ is typically treated as constant and nondynamical
at low energies. In our supergravity cosmology, however, (kAF )µ is associated
with the dynamical scalars A and B. Excitations of Fµν therefore result in per-
turbations δA and δB in the axion-dilaton background (6). As a consequence,
the energy-momentum tensor (T b )µν of the background acquires an additional
contribution, (T b )µν → (T b )µν + δ(T b )µν . One can show [1] that the contribu-
tion δ(T b )µν does indeed compensate for the negative energies associated with
a nonzero (kAF )µ .

5 Threshold analyses
Within the SME, it is straightforward to verify that Lorentz violation typi-
cally modifies one-particle dispersion relations [42,2,3,5]. This feature permits
the prediction of possible experimental signatures for Lorentz-symmetry break-
down based purely on kinematical arguments. For instance, primary ultrahigh-
energy cosmic rays (UHECR) at energies up to eight orders of magnitude below
the Planck scale have been observed. At such energies, Lorentz-violating effects
might be pronounced relative to the ones in low-energy experiments. This leads
to potentially observable threshold modifications for particle reactions, an idea
that has been adopted in many recent studies of Lorentz-symmetry breakdown
[51,99–103]. However, it is known [55] that some threshold analyses employ phe-
nomenologically constructed dispersion relations that violate physics principles
more fundamental than Lorentz symmetry.
In this section, we investigate how some of the arbitrariness in the con-
struction of dispersion-relation modifications can be removed. Our study relies
1
The conserved symmetric energy-momentum tensor for the lagrangian (3) acquires
no contribution from the N term because the latter is independent of g µν . The other
terms contributing to the energy density are positive definite.
10 Ralf Lehnert

on the principle of coordinate invariance and on the condition of compatibility

with an effective dynamical framework like the SME. These two features appear
fundamental enough for being physical requirements, while maintaining relative
independence of the details of the Planck-scale theory. We also discuss causality
and positivity, features that further add to the viability of threshold analyses.
Throughout we assume exact conservation of energy and momentum.
Coordinate-independent dispersion relations. As argued in Sec. 2, co-
ordinate independence is essential in physics, despite the presence of Lorentz
violation. In the published literature, the usual ansatz for modified dispersion
relations is of the following form:

E 2 − p 2 = m2 + δf (E, p) , (10)

where m is the usual mass parameter and pµ = (E, p) the 4-momentum. The
function δf (E, p) controls the extent of the Lorentz violation. Coordinate inde-
pendence requires δf to be a scalar, so that

n indices
X z }| {
αβ ···
δf (E, p) = T(n) pα pβ · · · . (11)
n≥1
| {z }
n factors

αβ ···
Here, T(n) denotes a constant tensor of rank n representing the Lorentz-
breaking background. The tensor indices α, β, . . . are distinct but each one is
contracted with a 4-momentum factor. This ensures that all terms in the sum
are observer Lorentz invariant. Under mild assumptions, Eq. (11) determines the
most general Lorentz-violating dispersion relation compatible with coordinate
independence.
The implications of the general coordinate-independent ansatz (11) can be
illustrated when the common assumption of rotation invariance in certain frames
is made. In this case, the form of the Lorentz-violating tensor parameters T(n)
is constrained by the imposed rotational symmetry. As a consequence, the cor-
rection δf fails to contain odd powers of the 3-momentum magnitude |p| [55].
This result is to be contrasted with the common occurrence of |p|3 corrections in
modified dispersion relations constructed by hand without reference to principles
essential in physics.
Note that a correction ∼ E|p|2 is consistent with coordinate independence.
Then, the question arises as to whether the usual ultrarelativistic relation E ≃
|p| can introduce an effective |p|3 modification. Although such a replacement may
yield excellent approximations for the eigenenergies, it gives incorrect results in
threshold analyses. This is intuitively reasonable because this replacement rein-
troduces the conventional degeneracy of the eigenenergies. An explicit example
for the failure of the ultrarelativistic approximation is provided by photon decay
into an electron-positron pair:

γ → e+ + e− . (12)
 The Lorentz-violating Standard-Model Extension 11

This process is kinematically forbidden in conventional physics. In the present

Lorentz-violating context, the decay (12) is allowed when the coordinate-indepen-
dent correction ∼ E|p|2 for photon, electron, and positron is used. However,
when the correction term is approximated by |p|3 , the process ceases to be kine-
matically permitted.
Underlying dynamical framework. The need for underlying dynamics
can be illustrated with the following example. Consider the rotationally sym-
metric modified dispersion relation
2
(pµ pµ − m2 ) = |p|6 . (13)

Note that odd powers of |p| are absent compatible with coordinate independence.
After the usual reinterpretation of the negative-energy solutions, the particle and
antiparticle energies are given by
r
(α) |p| 3
E± (p) = (−1)α + p 2 + m2 , (14)
M
where α = 1, 2 labels the two possible particle (antiparticle) energies, which per-
haps correspond to different spin-type states. An analysis of the photon decay
(12) employing these eigenenergies reveals that six kinematically distinct decays
have to be considered. Note, however, that angular-momentum conservation as-
sociated with the enforced rotational invariance may preclude some of the six
reactions. A proper study of this case therefore requires dynamical concepts.
Causality and positivity. Causality and energy positivity are fundamental
requirements in physics. However, one or both of these requirements can be
violated in the presence of Lorentz violation [5]. From a conservative viewpoint, it
is therefore natural to ask whether reaction-threshold kinematics is significantly
affected when positivity and causality are imposed. Let M and m denote the
scales of the fundamental theory and current low-energy physics, respectively.
Then, the scale ps-c for the occurrence of spacelike momenta (and thus negative
energies in certain frames) or causality problems can be as low as [5]
√
ps-c ∼ O( mM ) . (15)

For example, if m is the proton mass and M is taken to be the Planck scale,
then ps-c ∼ 3 × 1018 eV. UHECRs with a spectrum extending beyond 1020 eV
are often employed to bound Lorentz breaking or to suggest evidence for Lorentz
violation. Thus, imposing causality and positivity could require modifications in
threshold analyses.
As a specific example, consider the decay

γ → π 0 + γ, (16)

which is kinematically forbidden in conventional physics. The usual dispersion-

relation modifications in the literature permitting this process are associated
with causality or positivity violations: take the photon energy E to be given
12 Ralf Lehnert

by E = (p 2 + δf (p))1/2 , where δf (p) excludes a mass term 2 and depends

only on the photon 3-momentum p. Then, the point (E, p) = (0, 0) on the
momentum-space lightcone must satisfy the modified dispersion relation. For
some 3-momenta p 6= 0, a nontrivial correction δf (p) forces E(p) to curve to the
outside or inside of the lightcone leading to spacelike 4-momenta or superluminal
group velocities, respectively. If, however, the photon dispersion relation is taken
to be the conventional one, spacelike pions are required for the decay to occur.3

6 Conclusion

Although Lorentz and CPT invariance are deeply ingrained in the currently
accepted laws of physics, there are a variety of candidate underlying theories
admitting the violation of these symmetries. The ultrahigh sensitivity of many
Lorentz and CPT tests therefore permits the experimental search for Planck-
scale physics.
Spacetime-dependent couplings are one potential source for apparent Lorentz
and CPT violation: the gradient of such couplings in the equations of motion
selects a preferred spacetime direction in the effective vacuum. We have argued
that variations of couplings are natural in a cosmological context of candidate
fundamental theories.
The first-order Lorentz-violating effects resulting from varying couplings and
other mechanisms for Lorentz-symmetry breakdown are described by the SME.
At the level of effective quantum field theory, the SME is the most general
dynamical framework for Lorentz and CPT violation that is compatible with
the fundamental principle of coordinate independence.
Threshold analyses with modified dispersion relations are conceptually clean
Lorentz tests and are best performed within the SME. Many purely kinematical
threshold considerations in the literature are insufficient for bounding Lorentz
breaking because they violate coordinate independence or other fundamental
principles.

Acknowledgments

The author wishes to thank H.V. Klapdor-Kleingrothaus for the invitation to

the Beyond ’03 Meeting. This work was supported in part by the Centro Mul-
tidisciplinar de Astrofı́sica (CENTRA) and by the Fundação para a Ciência e a
Tecnologia (Portugal) under grant POCTI/FNU/49529/2002.
2
A mass term would yield a Lorentz-invariant (but gauge-symmetry violating) con-
tribution, which is not of interest in the present context.
3
The allowed phase space for the decay products in the case of a lightlike pion 4-
momentum is a set of measure zero leading at best to a suppressed rate for the
reaction.
 The Lorentz-violating Standard-Model Extension 13

References
1. V.A. Kostelecký, R. Lehnert, M.J. Perry: Phys. Rev. D 68, 123511 (2003)
2. D. Colladay, V.A. Kostelecký: Phys. Rev. D 55, 6760 (1997)
3. D. Colladay, V.A. Kostelecký: Phys. Rev. D 58, 116002 (1998)
4. R. Jackiw, V.A. Kostelecký: Phys. Rev. Lett. 82, 3572 (1999)
5. V.A. Kostelecký, R. Lehnert: Phys. Rev. D 63, 065008 (2001)
6. V.A. Kostelecký, C.D. Lane, A.G.M. Pickering: Phys. Rev. D 65, 056006 (2002)
7. V.A. Kostelecký: hep-th/0312310
8. KTeV Collaboration, H. Nguyen: ‘CPT Results from KTeV’. In: CPT and Lorentz
Symmetry II, ed. by V.A. Kostelecký (World Scientific, Singapore, 2002)
9. OPAL Collaboration, R. Ackerstaff et al.: Z. Phys. C 76, 401 (1997)
10. DELPHI Collaboration, M. Feindt et al.: preprint DELPHI 97-98 CONF 80 (1997)
11. BELLE Collaboration, K. Abe et al.: Phys. Rev. Lett. 86, 3228 (2001)
12. FOCUS Collaboration, J.M. Link et al.: Phys. Lett. B 556, 7 (2003)
13. D. Colladay, V.A. Kostelecký: Phys. Lett. B 344, 259 (1995)
14. D. Colladay, V.A. Kostelecký: Phys. Rev. D 52, 6224 (1995)
15. D. Colladay, V.A. Kostelecký: Phys. Lett. B 511, 209 (2001)
16. V.A. Kostelecký, R. Van Kooten: Phys. Rev. D 54, 5585 (1996)
17. O. Bertolami, D. Colladay, V.A. Kostelecký, R. Potting: Phys. Lett. B 395, 178
(1997)
18. V.A. Kostelecký: Phys. Rev. Lett. 80, 1818 (1998)
19. V.A. Kostelecký: Phys. Rev. D 61, 016002 (2000)
20. V.A. Kostelecký: Phys. Rev. D 64, 076001 (2001)
21. N. Isgur et al.: Phys. Lett. B 515, 333 (2001)
22. L.R. Hunter et al.: ‘Limits on Local Lorentz Invariance from Hg and Cs Magne-
tometers’. In: CPT and Lorentz Symmetry, ed. by V.A. Kostelecký (World Scien-
tific, Singapore, 1999)
23. D. Bear et al.: Phys. Rev. Lett. 85, 5038 (2000)
24. M.A. Humphrey et al.: Phys. Rev. A 62, 063405 (2000)
25. V.A. Kostelecký, C.D. Lane: Phys. Rev. D 60, 116010 (1999)
26. V.A. Kostelecký, C.D. Lane: J. Math. Phys. 40, 6245 (1999)
27. R. Bluhm et al.: Phys. Rev. Lett. 88, 090801 (2002)
28. R. Bluhm et al.: hep-ph/0306190
29. D. Sudarsky, L. Urrutia, H. Vucetich: Phys. Rev. Lett. 89, 231301 (2002)
30. D.F. Phillips et al.: Phys. Rev. D 63, 111101 (2001)
31. R. Bluhm et al.: Phys. Rev. Lett. 82, 2254 (1999)
32. H. Dehmelt et al.: Phys. Rev. Lett. 83, 4694 (1999)
33. R. Mittleman et al.: Phys. Rev. Lett. 83, 2116 (1999)
34. G. Gabrielse et al.: Phys. Rev. Lett. 82, 3198 (1999)
35. R. Bluhm et al.: Phys. Rev. Lett. 79, 1432 (1997)
36. R. Bluhm et al.: Phys. Rev. D 57, 3932 (1998)
37. R. Bluhm, V.A. Kostelecký: Phys. Rev. Lett. 84, 1381 (2000)
38. B. Heckel: ‘Testing CPT and Lorentz Symmetry with a Spin-Polarized Torsion
Pendulum’. In: CPT and Lorentz Symmetry II, ed. by V.A. Kostelecký (World
Scientific, Singapore, 2002)
39. E.O. Iltan: hep-ph/0308151
40. E.O. Iltan: hep-ph/0309154
41. R. Lehnert: hep-ph/0401084
42. S. Carroll, G. Field, R. Jackiw: Phys. Rev. D 41, 1231 (1990)
14 Ralf Lehnert

43. V.A. Kostelecký, M. Mewes: Phys. Rev. Lett. 87, 251304 (2001)
44. V.A. Kostelecký, M. Mewes: Phys. Rev. D 66, 056005 (2002)
45. H. Müller et al.: Phys. Rev. D 67, 056006 (2003)
46. J.A. Lipa et al.: Phys. Rev. Lett. 90, 060403 (2003)
47. G.M. Shore: Contemp. Phys. 44, 503 2003
48. V.W. Hughes et al.: Phys. Rev. Lett. 87, 111804 (2001)
49. R. Bluhm et al.: Phys. Rev. Lett. 84, 1098 (2000)
50. E.O. Iltan: JHEP 0306, 016 (2003).
51. S. Coleman, S.L. Glashow: Phys. Rev. D 59, 116008 (1999)
52. V. Barger, S. Pakvasa, T. Weiler, K. Whisnant: Phys. Rev. Lett. 85, 5055 (2000)
53. V.A. Kostelecký, M. Mewes: hep-ph/0308300
54. V.A. Kostelecký, M. Mewes: Phys. Rev. D, in press [hep-ph/0309025]
55. R. Lehnert: Phys. Rev. D 68, 085003 (2003)
56. G. Barenboim et al.: JHEP 0210, 001 (2002)
57. G. Barenboim, J. Lykken: Phys. Lett. B 554, 73 (2003)
58. O.W. Greenberg: Phys. Rev. Lett. 89, 231602 (2002)
59. O.W. Greenberg: Phys. Lett. B 567, 179 (2003)
60. G. Amelino-Camelia: Phys. Lett. B 510, 255 (2001)
61. J. Magueijo, L. Smolin: Phys. Rev. Lett. 88, 190403 (2002)
62. J. Lukierski, A. Nowicki: Int. J. Mod. Phys. A 18, 7 (2003)
63. J. Rembielinski, K.A. Smolinski: Bull. Soc. Sci. Lett. Lódź Sér. Rech. Déform. 53,
57 (2003)
64. D.V. Ahluwalia: gr-qc/0212128
65. D. Grumiller, W. Kummer, D.V. Vassilevich: Ukr. J. Phys. 48, 329 (2003)
66. R. Schützhold, W.G. Unruh: gr-qc/0308049
67. V.A. Kostelecký, S. Samuel: Phys. Rev. D 39, 683 (1989)
68. V.A. Kostelecký, S. Samuel: Phys. Rev. D 40, 1886 (1989)
69. V.A. Kostelecký, S. Samuel: Phys. Rev. Lett. 63, 224 (1989)
70. V.A. Kostelecký, S. Samuel: Phys. Rev. Lett. 66, 1811 (1991)
71. V.A. Kostelecký, R. Potting: Nucl. Phys. B 359, 545 (1991)
72. V.A. Kostelecký, R. Potting: Phys. Lett. B 381, 89 (1996)
73. V.A. Kostelecký, M. Perry, R. Potting: Phys. Rev. Lett. 84, 4541 (2000)
74. V.A. Kostelecký, R. Potting: Phys. Rev. D 63, 046007 (2001)
75. N.E. Mavromatos: these proceedings
76. D. Sudarsky, L. Urrutia, H. Vucetich: Phys. Rev. D 68, 024010 (2003)
77. F.R. Klinkhamer: Nucl. Phys. B 578, 277 (2000)
78. J. Alfaro, H.A. Morales-Técotl, L.F. Urrutia: Phys. Rev. Lett. 84, 2318 (2000)
79. J. Alfaro, H.A. Morales-Técotl, L.F. Urrutia: Phys. Rev. D 65, 103509 (2002)
80. S.M. Carroll et al.: Phys. Rev. Lett. 87, 141601 (2001)
81. Z. Guralnik et al.: Phys. Lett. B 517, 450 (2001)
82. A. Anisimov et al.: Phys. Rev. D 68, 085003 (2003) hep-ph/0106356
83. C.E. Carlson et al.: Phys. Lett. B 518, 201 (2001)
84. O. Bertolami et al.: Phys. Rev. D, in press [astro-ph/0310344]
85. M.S. Berger and V.A. Kostelecký: Phys. Rev. D 65, 091701(R) (2002)
86. H. Belich et al.: Phys. Rev. D 68, 065030 (2003)
87. M.S. Berger: Phys. Rev. D 68, 115005 (2003)
88. E. Cremmer, B. Julia: Nucl. Phys. B 159, 141 (1979)
89. A. Das, M. Fischler, M. Roček: Phys. Rev. D 16, 3427 (1977)
90. S.J. Perlmutter et al., [Supernova Cosmology Project Collaboration]: Nature 391,
51 (1998)
 The Lorentz-violating Standard-Model Extension 15

91. A.G. Riess et al., [Supernova Search Team Collaboration]: Astron. J. 116, 1009
(1998)
92. P.M. Garnavich et al.: Astrophys. J. 509, 74 (1998)
93. R. Lehnert: hep-ph/0308208
94. J.K. Webb et al.: Phys. Rev. Lett. 87, 091301 (2001) and references therein
95. T. Damour, F. Dyson: Nucl. Phys. B 480, 37 (1996)
96. Y. Fujii et al.: Nucl. Phys. B 573, 377 (2000)
97. K. Olive et al.: hep-ph/0205269
98. C. Adam, F.R. Klinkhamer: Nucl. Phys. B 607, 247 (2001)
99. O. Bertolami, C.S. Carvalho: Phys. Rev. D 61, 103002 (2000)
100. T. Jacobson, S. Liberati, D. Mattingly: Phys. Rev. D 66, 081302 (2002)
101. T.J. Konopka, S.A. Major: New J. Phys. 4, 57 (2002)
102. F.W. Stecker: astro-ph/0308214
103. M. Jankiewicz et al.: hep-ph/0312221