Вы находитесь на странице: 1из 7

Proca Action

Based partly on the Wikipedia

1. Introduction
The Maxwell-Proca field is the analogue of the electromagnetic field where the quanta have a non-zero rest
mass and the waves satisfy the Klein-Gordon equation. It’s what the electromagnetic field would be if the
photon had a non-zero rest mass. Equivalently, it may also be thought of as the dynamics governing the
electromagnetic field, itself, within an insulating medium.

Its dynamics are described by an action of the form


L  1 λ 

S = Ld 4 x , L ≡ = εc − g µρ g νσ Fµν F ρσ + g µν Aµ Aν  (F µν = ∂ µ Aν − ∂ ν A µ )
−g  4 2 
with constant coefficients ε > 0 and λg 00 > 0 , where c is light speed. The sign of λ is important. If one
assumed λg 00 < 0 , instead, the field would have tachyons as its quanta. The expression for the mass of the
quanta in terms of λ depends on which convention is adopted for the metric g µν and will be discussed
further below.

The Euler-Lagrange equations that result from the action principle


1
( )
∂ ν − g g µρ g νσ F ρσ − λg µν Aν = 0
−g
are called the Maxwell-Proca equations. Since Fµν = − Fνµ then it also follows, as a consequence, that
∂µ ( )
− g g µν Aν = 0 .

The action and field equations assume more familiar form if one uses to metric to raise indices
A µ = g µν Aν , F µν = g µρ g νσ Fνσ .
Then we may write
 1 
λ
L = εc − F µν F µν + A µ Aµ ,
 4 2 
1
( )
∂ ν − g F µν − λA µ = 0, ∂ µ − g A µ = 0 . ( )
−g
In Cartesian coordinates and in Minkowski space, the metric is constant and the latter two equations reduce
further to
( )
∂ µ ∂ ν A ν − ∂ ν ∂ ν Aµ − λAµ = 0, ∂ µ A µ = 0
µ µν
where ∂ = g ∂ ν . From this follows the Klein-Gordon equation
(∂ ∂ν
ν
)
+ λ Aµ = 0 .

Equivalently, starting from this equation, combined with the Lorenz gauge condition
∂ µ Aµ = 0 .
one may recover the Maxwell-Proca equations.

2. The Maxwell-Proca Equations


The analogy with the Maxwell field may be more closely seen in Minkowski space, by adopting the metric
1 1
( )
g 00 = 1 = g 00 , g i 0 = g 0 j = 0 = g i 0 = g 0 j , g ij = −c 2 δ ij , g ij = − 2 δ ij , g = det g µν = − 6
c c
for i, j = 1,2,3 ; defining the field components by
φ = − A0 , A = ( A1 , A2 , A3 ), E = (F10 , F20 , F30 ), B = (F23 , F31 , F12 )
and defining the conjugate fields, charge and current density by
∂L ∂L ∂L ∂L
D= , H=− , ρ=− , J= .
∂E ∂B ∂φ ∂A

Then, in addition to the homogeneous Maxwell equations and the field-potential relations
∂A ∂E
B = ∇ × A, E = −∇φ − , ∇ ⋅ B = 0, ∇ × B + =0
∂t ∂t
one obtains from the action
 E 2 − B 2c 2 λ φ2 − A2c 2  3

S = ε
 2
+ 2
c 2
dtd r


the inhomogeneous Maxwell equations and current conservation law
∂D ∂ρ
∇ ⋅ D = ρ, ∇ × H − = J, ∇ ⋅ J + =0,
∂t ∂t
and the Lorentz relations
1
D = εE, B = µH, εµ = 2
c
involving the permittivity ε and permeability µ .

However, one now also the following constitutive relations involving the potentials
λε
ρ = − 2 φ, J = − λεA .
c
The Maxwell-Proca field is, therefore, distinguished from the Maxwell field in that it produces a charge
and current density of its own proportional to its potential. In all other respects, it’s the same as the
Maxwell field.

In the presence of external sources ρ e and J e these relations are modified to


λε
ρ = ρe − φ, J = J e − λεA .
c2

3. The Klein-Gordon Equation and the Mass of the Field Quanta


In the absence of external sources, the inhomogeneous field equations reduce to
 2 1 ∂2  λ ∂  1 ∂φ   2 1 ∂ 2  λ  1 ∂φ 
 ∇ − 2 2 φ − 2 φ = −  ∇ ⋅ A + 2 ,  ∇ − 2 2  A − 2 A = ∇ ⋅  ∇ ⋅ A + 2 ,
 
 c ∂t  c ∂t  c ∂t   c ∂ t  c  c ∂t 
while the conservation law reduces to the Lorenz gauge condition
1 ∂φ
∇⋅A + 2 =0.
c ∂t
Using this, the field equations may be further reduced to the Klein-Gordon equations
 1 ∂2 λ   1 ∂2 λ 
 2 2 − ∇ 2 + 2 φ = 0,  2 2 − ∇ 2 + 2  A = 0 .
 c ∂t   c 
 c   c ∂t
The fields E, B, D, H will also satisfy the Klein-Gordon equations.

The relation to the field quanta may be best seen via the operator correspondence

E ↔ i , p = −i∇
∂t
for energy E and momentum p , where  is Dirac’s constant. Applying this to the field equation results in
the operator identity
 E2    
2
 E2    
2
 − p 2
− λ   φ = 0,  − p 2
− λ   A=0.
 c2  c    c2  c  
  
This is consistent with the mass-energy-momentum relation
E2
= p2 + m2c2 ,
c2
provided one takes, as the mass

m= 2 λ.
c

For stationary and spherically symmetric sources, the Klein-Gordon equation for the potential reduces to
 1 ∂2  mc  
2
d2  mc 
2
 − ∇ 2
+   φ = 0 → (rφ ) =   rφ .
 c 2 ∂t 2     dr 2
  
 
The most general solution that is bounded as r → ∞ is of the form
A  mcr 
φ = exp − 
r   
and is known as the Yukawa potential. The feature that distinguishes it from the 1 r potential of a Maxwell
field is that it drops off exponentially with an effective range given by the Compton wavelength

=
mc
of the field quanta.

If one adopts the following convention for the metric


1
( )
g 00 = − 2 , g 00 = −c 2 , g i 0 = g 0 j = 0 = g i 0 = g 0 j , g ij = δ ij = δ ij = g ij , g = det g µν = c ,
c
with opposite signature, then the action would become
 E 2 − B2c2 φ2 − A2c 2  3
S = ε

∫ 2
−λ
2
dtd r .


One then obtains the following expression for the mass

m= −λ .
c

4. The Hamiltonian Formulation and Quantization


The momenta π µ conjugate to the field coordinates Aµ are obtained by varying the action with respect to
the corresponding time derivatives ∂A µ ∂t ,
 ∂A  3
∫ (D ⋅ δE − H ⋅ δB + J ⋅ δA − ρδφ)dtd ∫
3
δS = r = − D ⋅ δ dtd r .
 ∂t 
From this, one can immediately pick out the components
(
A0 = −φ ↔ π 0 = 0; A = ( A1 , A2 , A3 ) ↔ π 1 , π 2 , π 3 = −D . )
The negative of the electric displacement D is conjugate to the potential A , while the momentum
conjugate to the potential φ suffers a constraint π 0 = 0 . A Hamiltonian may therefore only be written up
to a Lagrange multiplier, b
 ∂A 


H = − D⋅
∂t
− L + π 0 b d 3 r

 D2 ∇×A
2
λε  φ2  
  + π 0 b d 3 r.
=
 2ε∫ + D ⋅ ∇φ +

+  A 2 − 2
2  c 


 
Taking the Poisson brackets then yields the following dynamics
∂A   D2   D
∂t
= {A, H } = A,
 ∫  2ε + D ⋅ ∇φ d 3 r 
 
=−
ε
− ∇φ

∂φ
∂t
= {φ, H } {∫
= φ, π 0 bd 3 r } =b

∂D   ∇ × A ⋅ ∇ × A λε 2  3  ∇ × (∇ × A )
∂t
= {D, H } = D,
 ∫  2µ
+
2
A d r  =
 
 µ
+ λεA

∂π 0   λε φ 2  3  λε
{
= π0, H } = π 0 , ∫ D ⋅ ∇φ −

d r 
2 
= ∇⋅D+ 2 φ
∂t   2 c   c

In general, the time derivatives of the primary constraints yield the secondary constraints. Here, for π 0 , this
yields what is known as the Gaussian constraint
∂π 0 λε
= ∇⋅D + 2 φ.
∂t c
Taking a second time derivatives yields the Lorenz condition
∂ 2π  b 
2
= λ ∇ ⋅ A + 2  .
∂t  c 
Since λ ≠ 0 for Proca fields, this can be used to eliminate b = −c 2 ∇ ⋅ A , resulting in the modified
Hamiltonian and the modified equation for ∂φ ∂t
 D2 ∇×A
2
λε  φ2  
H=  ∫
+ D ⋅ ∇φ + +  A 2 − 2  − π 0 c 2 ∇ ⋅ A d 3 r,
 2ε 2µ 2  c  
 
∂φ
∂t
{∫
= {φ, H } = −c 2 φ, π 0 ∇ ⋅ Ad 3 r = −c 2 ∇ ⋅ A. }
The Poisson brackets with the primary constraint may also be derived
 
{ }
π 0 ( x ), π 0 ( y ) = 0, π 0 (x ),
∂π 0
( y ) = π 0 (x ), ∇ ⋅ D( y ) + λε2 φ( y ) = − λε2 δ (x − y ) .
 ∂t   c  c
This shows that the Gaussian constraint is second class.

A common choice of gauge to remove the Gaussian constraint is to take φ = 0 . In this case b = 0 and the
{ }
conjugate pair φ, π 0 drops out entirely. In effect, these become classical coordinates. This is the hallmark
feature of a second class constraint.

{ }
Equivalently, therefore, one may simply redefine the bracket by declaring φ, π 0 = 0 by fiat. This effects a
reduction of the Poisson bracket to what is known as the Dirac bracket. When the field is quantized, what
this amounts to doing is:
• treating the potential φ as a classical field not to be quantized at all
• treating π 0 = 0 as an operator identity, rather than as a constraint

There are separate reasons not to go with this formulation when quantizing the field. The most important is
that the 2-point functions
∂ ∂
W µν ( x − y ) = 0 Aµ ( x )Aν ( y ) 0 , 0 Fµν ( x )A ρ ( y ) 0 = µ W νρ (x − y ) − ν W µρ (x − y ), …
∂x ∂x
cannot be consistently defined. This is not specifically true of Proca fields, but is also a problem with
Maxwell fields.

The Stückelberg action


 1 λ   1 
∫ ( ) ( )
S = εc − g  − ∂ µ U ν ⋅ ∂ µ U ν − ∂ µ B ⋅ ∂ µ B + U µ U µ − B 2 d 4 x  Aµ = U µ +
 ∂ µ B 
 2 2   λ 
is sometimes used as an alternative, but suffers the same problems as the Proca action with 2-point
functions, such as
[ ]
0 T Aµ ( x )Aν (x ) 0 .
Also, the 0-mass limit, λ → 0 cannot be consistently defined.

A common fix, particularly one suited to resolve the problems with the 2-point functions, is to generalize
beyond the Maxwell-Proca equations by incorporating the constraints, themselves, more fully into the
dynamics and effectively treating them as fields in their own right. To this end, a convenient action to use is
the so-called B -field formulation
 1 λ  µc 2

S = L − g d 4 x, L = εc − F µν Fµν + A µ Aµ  + B∂ µ A µ + α
 4 2  2
B ,

with the field equations now becoming


( )
∂ ν F µν = λA µ − µc∂ µ B, ∂ µ A µ + αµcB = 0, ∂ µ ∂ µ + αλ B = 0 .
2
For α ≠ 0 , this essentially makes π into a field. A common choice for the extra coefficient is then α = 1 –
the Feynman gauge. An alternate choice, also in common use, is α = 0 – the Landau gauge. As a result of
that choice, the potential A0 = −φ then has, as its conjugate momentum, π 0 = B .

5. Scale Invariance and Renormalization


The Proca and B-field actions are invariant under the following scale change and reparametrization
Aµ F µν µ
Aµ → , F µν → , ε → Zε, µ → , λ → λ, B → Z B .
Z Z Z
By convention, the “bare” fields are defined by an action principle that has a coupling ε = 1 . One way to
think of this is that a factor of Z = 1 ε is absorbed into a redefinition of the field, itself. This is legitimate
as long as ε truly is a constant. However, what we find in quantum field theory there is a “scale
dependence” on the parameter. That is, the effective value observed, say, in a scattering process of a point-
like or concentrated source, depends on the scale of resolution at which the process probes the source.
Another way of saying this is that ε exhibits a radial dependence ε = ε (r ) on the distance r from the
scattering center.

A static point-like external source ρ e (r ) = qδ 3 (r ) will produce radial fields with


d  2 dφ  λεφ q
r ε = − δ (r ) .
dr  dr  c 2 4π
Because of the variability of ε near the source, there is an effective deviation from the Yukawa potential.
The situation is analogous to that in quantum electrodynamics, where the effective field arising from the
polarization of the vacuum around concentrated sources is approximated by the Euler-Heisenberg
Lagrangian which, like the Maxwell-Born Lagrangian, entails both a non-trivial permittivity ε and axion
coefficient, θ (see below).

The “renormalized” fields are derived from the ε = 1 action by effecting a scale change with the
renormalization coefficient Z = Z 3 .Thus, in the current literature, the coefficient Z 3 is used in place of the
permittivity ε and is referred to as the vertex renormalization coefficient. They are therefore synonymous.

A simple toy model that captures the phenomenon at a qualitative level may be given by
 1 ∂2  E 2 − B 2c 2
 2 2 − ∇ 2  log ε = Kε
 c ∂t  2
 
for some constant K . For a static electric field with ε and φ radially dependent, this reduces to
2
d2
2
(r log ε ) = − Krε  dφ  .
dr 2  dr 
Combining with the field equation and substituting u = 1 r , this results in the following system
2
d  dφ  λεφ d2 K  dφ 
u2 ε  = 2 , ε 2 (log ε ) = −  ε  , (u < ∞ )
du  du  c du 2  du 
For electromagnetism, λ = 0 , this does, indeed, exhibit anti-screening of the vacuum near a point-like
source and even a Landau pole. For Maxwell-Proca fields, λ > 0 , the situation is more complex.

Further insight into the matter may be obtained by stepping back and asking the question: what is the most
general Lagrangian that accords with the principle of relativity, possessing Lorentz invariance? Such a
Lagrangian may be constructed out of all the Lorentz invariant combinations of the fields. These include
the following
E 2 − B 2c 2 φ2 − A2c 2
I= , J = E ⋅ B, K = .
2 2
The Maxwell-Proca Lagrangian may then be recovered as the linear approximation of a more general
Lagrangian L = L (I, J , K ) , where
∂L ∂L c2 ∂L
ε (I, J , K ) = , θ (I, J , K ) = , λ(I, J , K ) = .
∂I ∂J ε (I, J , K ) ∂K
Applying the definitions
∂L ∂L ∂L ∂L
D= , H=− , ρ=− , J=
∂E ∂B ∂φ ∂A
this leads to the following constitutive relations of a form more general than the Lorentz relations
λε
D = εE + θB, H = εc 2 B − θE, ρ = − 2 φ, J = − λεA .
c

Since the inhomogeneous Maxwell equations are invariant under the transformation
D → D − θ 0 B, H → H + θ 0 E ,
for constant θ 0 the pseudo-scalar “axion” coefficient θ is only determined up to a plus or minus addition
by a constant. Therefore, one may always assume θ → 0 asymptotically, θ (0,0,0) = 0 . The asymptotic
values for the remaining coefficients ε 0 = ε (0,0,0) and λ0 = λ(0,0,0 ) then determine the nature of the field
away from concentrated sources.

6. Conversion of Maxwell to Proca via Coupling with a Scalar Field


A dynamics equivalent to that described by the Proca action will also result for a Maxwell field when it is
coupled to a scalar field that has a non-zero constant value in its minimum energy mode. A most important
case of this occurs with the Higgs mechanism. A Proca action may then be derived as the gauge-fixed
version of the Stückelberg action.

Suppose a scalar field χ is coupled to the potentials φ, A of a Maxwell field E, B , and suppose the scalar
field strengths given by
∂χ ˆ = ∇χ + Aχ .
φˆ = − + φχ , A
∂t
Assume the dynamics of the coupled system is governed by with an action of the form
 E 2 − B2c2   φˆ 2 − Aˆ 2 c 2 

S = Ldtd 3 r, L = ε  + β  −U (χ)
2   2 
   
with a constant coefficient β > 0 . Define the conjugate fields by
∂L ∂L ˆ .
ρˆ = − = − βφˆ , Jˆ = = − βc 2 A
∂φˆ ˆ
∂A
Because there is now explicit dependence on the potentials, the Maxwell field now has external charge and
current sources given, respectively, by
∂L ∂L ∂L ∂L
ρ=− = −χ = χρˆ , J = =χ = χJˆ .
∂φ ∂φˆ ∂A ∂Aˆ
Thus, one has the following chain of dependence
( ) ( )
( ρ, J ) → ρˆ , Jˆ → φˆ, Aˆ → (φ, A )
out of which may arise a relation linking the charge and current to the potentials.

Assume the scalar potential U ( χ ) at χ = χ 0 ≠ 0 and that the field is in a minimum energy state at the
constant value χ 0 . A typical example of such a potential is given by
(
U (χ) = k χ 2 − (χ0 ) .
2 2
)
Going through the chain, one step at a time: the scalar field strengths then directly reflect the Maxwell
potentials
φˆ = φχ , A ˆ = Aχ .
0 0
The constitutive law for the scalar field devolves into one involving the potentials
ρˆ = − βφχ 0 , Jˆ = − βc 2 Aχ 0 .
Finally, these relations are inherited by the external sources
ρ = − βχ 0 φχ 0 , J = − βc 2 χ 0 Aχ 0 ,
resulting in an effective value for the Proca constant given by
λε = β ( χ 0 ) .
2

Intuitively, what’s going on is that the evanescence of the scalar field turns the vacuum, itself, into an
insulating dielectric that impedes the Maxwell field. In particular, since β > 0 , the potential φ produces an
off-setting charge ρ = − λεφ that screens against concentrated sources, limiting their range effectively to a
distance proportional to  ∝ 1 λ ∝ 1 χ0 .

7. References
The Proca Action, Wikipedia, 2007 December 21 01:14, http://en.wikipedia.org/Proca_action

N. Nakanishi and I. Ojima, Covariant Operator Formalism of Gauge Theories and Quantum Gravity,
Lecture Notes in Physics Volume 27, World Scientific, 1990.

Вам также может понравиться