Exploring the effect of Lorentz invariance violation with ...
Photon Quantization with Lorentz Violation
Transcript of Photon Quantization with Lorentz Violation
Photon Quantization with Lorentz Violation
Don Colladay
New College of Florida
Talk presented at Miami 2012
(work done in collaboration with Patrick McDonald)
Overview of Talk
• Review of Gupta-Blueler Method of Photon Quantization
• Application to SME Photon Sector
• Issues with momentum-space expansions
Review of Gupta-Bleuler Quantization
Basic idea of Gupta-Bleuler is to add a gauge-fixing term to theLagrangian that allows for all four components of the vectorpotential to be quantized in a covariant manner
• Has advantage of maintaining Lorentz covariance explicitlyin the quantization procedure, a great advantage for calcu-lations
• Procedure introduces negative-norm states that must be re-moved using some condition on the physical Hilbert-spacestates
– S. Gupta, 1950; K. Bleuler, 1950
Formulation in the conventional case starts with Lagrangian
L = −1
4F2 −
λ
2(∂ ·A)2
where λ is a ”gauge-fixing” term
Calculation of the conjugate momenta yield
πj = F j0
and
π0 = −λ∂ ·A
(Note that π0 = 0 when λ = 0 destroying covariance)
Quantization is imposed using equal-time commutators
[Aρ(t, ~x), πν(t, ~y)] = iη νρ δ
3(~x− ~y)
and
[Aρ(t, ~x), Aν(t, ~y)] = [πρ(t, ~x), πν(t, ~y)] = 0
in analogy with Poisson-Bracket approach
Choosing λ = 1 (Feynman gauge) decouples the commutators
[Aρ(t, ~x), Aν(t, ~y)] = iηρνδ3(~x− ~y)
and
[Aρ(t, ~x), Aν(t, ~y)] = [Aρ(t, ~x), Aν(t, ~y)] = 0
(other choices for λ are not as simple...)
Expansion of the field in terms of mode operators yields
Aµ(x) =∫
d3~p
(2π)32p0
∑α
(aα(~p)εµα(~p)e−ip·x + a†α(~p)ε∗µα (~p)eip·x
)
• p0 = |~p| are positive frequency solutions to the dispersion
relation (p2 = 0)
• α = 0,1,2,3 runs through all four polarization vectors
• εµα vectors can be freely chosen as orthonormal basis
If εµα are chosen so that εµ0 = {1,0,0,0}, commutation relations
imply that mode operators satisfy
[aα(~p), a†β(~p′)] = −ηαβ2p0(2π)3δ3(~p− ~p′)
Due to the presence of the metric, a†0(~p) operators are uncon-
ventional and produce negative-norm states when they act on
vacuum
To eliminate them, impose a requirement on the physical states
〈ψ|(∂ ·A)|ψ〉 = 0, for |ψ〉 ∈ Hphysor, equivalently,
(∂ ·A+)|ψ〉 = 0
where (+) represents the positive-frequency part of the field
In terms of mode operators, the condition is
(a0(~p)− a3(~p))|ψ〉 = 0
provided the 3 direction is taken to point along the momentum
The hamiltonian takes the form
H =∫
d3~p
(2π)32p0
3∑α=1
a†α(~p)aα(~p)− a†0(~p)a0(~p)
the scalar polarization states create negative-energy states, but
the physical condition on Hphys protects the physical states from
any negative energy problems
SME Photon Sector (minimal extension, CPT-even terms)
SME photon lagrangian is given by
L = −1
4F2 + (kF )µναβF
µνFαβ − λ(∂ ·A)2
Calculation of πµ yields
πj = F j0 + kj0αβFαβ, π0 = −λ∂µAµ
Setting λ = 1 and imposing the quantization condition
[Aρ(t, ~x), πν(t, ~y)] = iη νρ δ
3(~x− ~y)
is commensurate with the Gupta-Blueler approach
Conversion to commutation relations involving A gives
[Ai(t, ~x), Aj(t, ~y)] = −iRijδ3(~x− ~y)
where Rij is the inverse matrix of δij − 2(kF )oioj, and the other
relations are conventional
It is convenient to put this into more covariant notation by set-
ting R00 = −1 and R0i = 0
[Aµ(t, ~x), Aν(t, ~y)] = −iRµνδ3(~x− ~y)
Calculation of the Hamiltonian H = πµAµ − L yields (after ap-propriate partial integrations in H =
∫d3~xH)
H =1
2
((∂jA
j)2 − (A0)2 +A0∂0(∂ ·A) + ~E2 + ~B2)
−k0i0jF0iF0j +
1
4kijklF
ijF kl
The commutation relations produce the expected action of thehamiltonian on the fields as the generator of time translations
i[H,Aµ] = ∂0Aµ
Similarly, the three-momentum operator
P i =∫d3~x
(πj∂iAj − (∂ ·A)∂iA0
)satisfies
i[P i, Aµ] = ∂iAµ
indicating that the cannonical quantization works
Gauge condition can be implemented by restricting physical state
space Hphys so that
〈ψ|(∂ ·A)|ψ〉 = 0, for |ψ〉 ∈ HphysGauge-terms then drop out of the expectation value of the hamil-
tonian and momentum as in the conventional case
Useful classification of kF terms uses type of ”dual”
(kF )µναβ =1
4εµνρσεαβγδ(kF )ρσγδ
Can split into
kF = kSDF ⊕ kASDF
• Self-dual components → no birefringence and Tr(kF ) 6= 0
• Anti-self-dual components → birefringence and Tr(kF ) = 0
(Also Weyl-tensor part of kF )
Experimental bounds on these
• birefringent terms can be bounded using cosmological tests
at 10−32 level
• non-birefringent terms bounded using laboratory scale exper-
iments at 10−14 - 10−17 level
See Data Tables for Lorentz and CPT violation for details
– A. Kostelecky and N. Russell, arXiv:0801.0287
The self-dual kF components can be handled using a coordinate
redefinition (at lowest-order in the free photon theory) and are
therefore not particularly interesting from a theoretical point of
view
Much more interesting are the anti-self-dual terms that lead to
birefringence effects, these terms cause fundamental issues in
the usual Fourier expansion technique (next part of talk...)
Issues with momentum-space expansion
The conventional expansion of the field takes form
Aµ(x) =∫
d3~p
(2π)3
∑α
1
2p0
(aα(~p)εµα(~p)e−ip·x + a†α(~p)ε∗µα (~p)eip·x
)where the p0-factor has a dependence on α, due to the birefrin-
gence
Plugging this into the commutator yields terms of the form
∑α,β
(1
p0′
)εµα(~p)ενβ(~p′)[aα(~p), a†β(~p′)]e−i(p·x−p
′·y)
To eliminate time dependence and generate the delta function,
let the commutator take the usual form
[aα(~p), a†β(~p′)] = −ηαβ2p0(2π)3δ3(~p− ~p′)
implies condition on ε vectors∑α,β
ηαβεµα(~p)ενβ(~p) = −Rµν
Analogy: Vierbein formalism in general relativity looks like this...
Special case to see if this condition feasible:
Let k0103 be anti-self-dual (birefringent) and related symmetric
components be only nonzero terms.
Energies are determined by a polynomial of form
p4(f(p20)) = 0
where f is a second-degree polynomial in p0 yielding two solutions
Plot of energy surfaces are perturbed spheres
Other solution looks like
Can get overlapping energy surfaces near cusp points
If stay away from this area, polarization sum seems to work fine
For example, if let ~p point in the 2-direction it is possible to
choose the normalization of the polarization vectors such that∑α,β
ηαβεµα(~p)ενβ(~p) = −Rµν
holds in agreement with the commutation relations
However, there can be problems where the degeneracies occur,
for example, if let ~p point in the 3-direction, one of the physical
modes becomes degenerate with the gauge mode
In this case, there are only three polarization vectors, an insuffi-
cient number to satisfy the commutation rules
It may be the case that this type of singular behavior only occurs
on a ”set of measure zero” in momentum space, but we are still
working on proving this...
Conclusions
• Gupta-Blueler quantization of the SME photon sector ap-
pears to work well when coordinate-space functions are con-
sidered
• It is unclear how to expand the momentum-space functions
to impose the quantization condition consistently for all di-
rections of the momenta
• We conjecture that ”problem points” may be a set of mea-
sure zero so that the quantization may be imposed consis-
tently on almost all of momentum space