1 Lorentz transformations for classical fields

The Lorentz transformations are the linear global transformations (µ , ν = 0, 1, 2, 3)

µ
x0 = Λµν xν or x0 = Λ · x (1.1)

that leave invariant the quadratic form

2 2 2 2
ds2 = dx0 − dx1 − dx2 − dx3 = ηµν dxµ dxν ηµν = diag. (1, −1, −1, −1) (1.2)

It then follows that Λ has to satisfy

ηρσ Λρµ Λσν = ηµν or ΛT η Λ = η (1.3)

We also have that

∂ ∂ xν ∂ ν ∂ ∂ ∂
0 µ = 0 µ ν
= Λ−1 µ or 0
= Λ−1 (1.4)
∂x ∂x ∂x ∂ xν ∂x ∂x
Among the Lorentz transformations we have the spatial rotations and the Lorentz boosts.
The rotations on the planes (1 , 2), (2 , 3), and (3 , 1), are respectively (0 ≤ θ ≤ 2 π)
1 0 0 0
 
 0 cos θ sin θ 0 
Λ (R12 ) =  (1.5)
 
0 − sin θ cos θ 0

 
0 0 0 1

1 0 0 0
 
 0 1 0 0 
Λ (R23 ) =  (1.6)
 
0 0 cos θ sin θ

 
0 0 − sin θ cos θ

1 0 0 0
 
 0 cos θ 0 − sin θ 
Λ (R31 ) =  (1.7)
 
0 0 1 0

 
0 sin θ 0 cos θ
The Lorentz boosts on the planes (0 , i), i = 1, 2, 3, are (−∞ < α < ∞)
cosh α − sinh α 0 0
 
 − sinh α cosh α 0 0 
Λ (B01 ) =  (1.8)
 
0 0 1 0

 
0 0 0 1

cosh α 0 − sinh α 0
 
 0 1 0 0 
Λ (B02 ) =  (1.9)
 
− sinh α 0 cosh α 0

 
0 0 0 1

1
 cosh α 0 0 − sinh α
 
 0 1 0 0 
Λ (B03 ) =  (1.10)
 
0 0 1 0

 
− sinh α 0 0 cosh α

with α being the rapidity

v 1 v/c
tanh α = cosh α = q sinh α = q (1.11)
c 1 − v 2 /c2 1 − v 2 /c2

Note that the matrices Λ (Rij ) e Λ (B0i ) are real, and that Λ (Rij ) are orthogonal (ΛT (Rij ) =
Λ−1 (Rij )) and therefore are unitary (Λ† (Rij ) = Λ−1 (Rij )). However, the matrices Λ (B0i )
are symmetric and so are neither orthogonal or unitary. Consequently, such a vector repre-
sentation, of dimension 4, of the Lorentz group is not unitary. That is a particular case of a
more general fact: any finite dimensional representation of a non-compact (infinite volume)
Lie group is necessarily non unitary. The unitary representations of the Lorentz group (and
also of the Poincaré group) are infinite dimensional.
For an infinitesimal rotation we shall write
 
Rij = 1 + i θ εijk Jk + O θ2 (1.12)

and for an infinitesimal boost

 
B0i = 1 + i α Ki + O α2 (1.13)

We then have
0 0 0 0 0 0 0 0 0 0 0 0
     
 0 0 0 0   0 0 0 1   0 0 −1 0 
Λ (J1 ) = i  ; Λ (J2 ) = i  ; Λ (J3 ) = i 
     
0 0 0 −1 0 0 0 0 0 1 0 0

     
0 0 1 0 0 −1 0 0 0 0 0 0
(1.14)

and
0 1 0 0 0 0 1 0 0 0 0 1
     
 1 0 0 0   0 0 0 0   0 0 0 0 
Λ (K1 ) = i  ; Λ (K2 ) = i  ; Λ (K3 ) = i 
     
0 0 0 0 1 0 0 0 0 0 0 0

     
0 0 0 0 0 0 0 0 1 0 0 0
(1.15)

We then have the Lorentz algebra

[ Ji , Jj ] = i εijk Jk
[ Ji , Kj ] = i εijk Kk (1.16)
[ Ki , Kj ] = −i εijk Jk

2
We now introduce
1 1
Ni = (Ji + i Ki ) ; N̄i = (Ji − i Ki ) (1.17)
2 2
and so

[ Ni , Nj ] = i εijk Nk
h i
Ni , N̄j = 0 (1.18)
h i
N̄i , N̄j = i εijk N̄k

We also have that

0 − 12 0 0
 
 − 21 0 0 0 
Λ (N1 ) =  
(1.19)
0 0 0 − 2i
 
 
0 0 2i 0

0 0 − 21 0
 
 0 0 0 2i 
Λ (N2 ) =  (1.20)
 
1
−2 0 0 0 


0 − 2i 0 0

0 0 0 − 21
 
 0 0 − 2i 0 
Λ (N3 ) =  
(1.21)
0 2i 0 0 
 

− 21 0 0 0

1
0 0 0
 
2
1
  
2
0 0 0 
Λ N̄1 =  (1.22)
 
0 0 0 − 2i

 
0 0 2i 0

0 21 0 0
 
  0 0 2i
 0 
Λ N̄2 =  1  
(1.23)
0 0 0

 2 
0 − 2i 0 0

0 0 0 12
 
   0 0 − 2i 0 
Λ N̄3 =  (1.24)
 
0 2i 0 0 


1
2
0 0 0

3
Note that
1  2 1
Λ (Ni )2 = 1l Λ N̄i = 1l (1.25)
4 4
and so  
1l ± 2 Λ (Ni ) and 1l ± 2 Λ N̄i (1.26)
are projectors. In fact, if | vi is a given state of the representation, then the states
  
| v± i = (1l ± 2 Λ (Ni )) | vi | v̄± i = 1l ± 2 Λ N̄i | vi (1.27)
are eigenstates
1   1
Λ (Ni ) | v± i = ± | v± i Λ N̄i | v̄± i = ± | v̄± i (1.28)
2 2
Of course, we can not diagonalize (have eigenstates) of all Ni ’s and N̄i ’s at the same time.
Usualy we diagonilize N3 and N̄3 .

1.1 The vector field Aµ

As an example take a vector field Aµ transforming as the derivative (see (1.4))
ν
A0µ = Λ−1 µ Aν or A0 = Λ−1 · A (1.29)
Let us denote
A0
 
 A1 
A= 
(1.30)
A2
 
 
A3
We can construct
 the eigenstates of N3 and N̄3 by using the projectors. However, note that
Λ (N3 ) and Λ N̄3 mix the components A0 ↔ A3 , and A1 ↔ A2 , but not otherwise. Then
we can split each eigenvector into two. So, we write
A0
 
 A1  (A1 − iA2 ) (A1 + iA2 )
A= 
= | 1/2 , 1/2i + | −1/2 , −1/2i
A2
 
  2 2
A3
(A3 − A0 ) (A3 + A0 )
+ | 1/2 , −1/2i + | −1/2 , 1/2i (1.31)
2 2
with
0 −1
   
 1   0 
| 1/2 , 1/2i =  
| 1/2 , −1/2i =  
i 0
   
   
0 1
1 0
   
 0   1 
| −1/2 , 1/2i =  | −1/2 , −1/2i =  (1.32)
   
0 −i
 
   
1 0

4
One can check that
 
Λ (N3 ) | s1 , s2 i = s1 | s1 , s2 i Λ N̄3 | s1 , s2 i = s2 | s1 , s2 i
sa = ±1/2
(1.33)
Therefore, Aµ transforms under the doublet of the SU (2) generated by Ni , and also by the
doublet of the SU (2) generated by N̄i .
Note that under the SO(3) (or SU (2)) group of spatial rotations, generated by Ji =
Ni + N̄i , we have a triplet representation given by
1
| 1i = | 1/2 , 1/2i
2
1
| 0i = (| −1/2 , 1/2i+ | 1/2 , −1/2i) (1.34)
2
1
| −1i = | −1/2 , −1/2i (1.35)
2
and a singlet given by
1
| 0is = (| −1/2 , 1/2i− | 1/2 , −1/2i) (1.36)
2
Therefore, 3 degrees of freedom of Aµ correspond to a spin-one particle, and one degree of
freedom to a spin-zero particle. Indeed, we can write

A0
 
 A1 
A= 
= (A1 − iA2 ) | 1i + A3 | 0i + (A1 + iA2 ) | −1i + A0 | 0is (1.37)
A2
 
 
A3

1.2 Other fields

We give here the representations of the Lorentz group under which some other fields trans-
form.

1. The first one is the real scalar field φ that transform under the scalar representations
of the two SU (2)’s generated by Ni and N̄i , which we denote (0 , 0):

| φi ≡| 0i⊗ | 0i (1.38)

In the case of a complex scalar field φ = φ1 + iφ2 , the real and imaginary parts of the
field transform under the scalar representation.

2. The spinors fields transform under the spinor representations of the Lorentz group.
In fact, the Weyl left and right spinors ψL and ψR transform under the (1/2 , 0) and
(0 , 1/2) representations respectively, i.e.

| ψL i =| ±1/2i⊗ | 0i | ψR i =| 0i⊗ | ±1/2i (1.39)

5
 So, each one has two independent components. The Dirac spinor transforms under the
(1/2 , 0) + (0 , 1/2) representation, i.e.

| ψi =| ±1/2i⊗ | 0i+ | 0i⊗ | ±1/2i (1.40)

and so it has four independent components.

3. As we have seem above the vector field Aµ transforms under the (1/2 , 1/2) represen-
tation, i.e.
| Aµ i ≡| ±1/2i⊗ | ±1/2i (1.41)
and so it has four independent components, and as representations of the rotation
subgroup SO(3) it has a spin 1 and a spin 0 components, i.e.

| Aµ i =| 0i+ | 0, ±1i (1.42)

4. The field tensor Fµν or any other antisymmetric rank 2 tensor Bµν transform under
the (1 , 0) + (0 , 1) representation, i.e.

| Bµν i ≡| 0, ±1i⊗ | 0i+ | 0i⊗ | 0, ±1i (1.43)

and so it has 6 independent components. Its self-dual and anti-self-dual components

transform under the (1 , 0) and (0 , 1) respectively, i.e.
(+) (−)
| Bµν i ≡| 0, ±1i⊗ | 0i | Bµν i ≡| 0i⊗ | 0, ±1i (1.44)

with
1 i
 
(±)
Bµν = ηµρ ηνσ ± εµνρσ B ρσ (1.45)
2 2
where ηµν = diag. (1 , −1 , −1 , −1) is the Minkowski metric and εµνρσ is the totally
anti-symmetric tensor with ε0123 = 1.

5. The metric field described by a symmetric matrix gµν , transforms under the represen-
tation (1 , 1) + (0 , 0) of the Lorentz group, i.e.

| gµν i ≡| 0, ±1i⊗ | 0, ±1i+ | 0i⊗ | 0i (1.46)

and so it has 10 independent components. In terms of representations of the rotation

subgroup SO(3) it decomposes into a spin 2, a spin 1 and 2 spin 0 representations, i.e.

| gµν i ≡=| 0, ±1, ±2i+ | 0, ±1i+ | 0i+ | 0i (1.47)

The particle associated to the graviton itself has spin 2 and so only 5 independent
components. The extra 5 degrees of freedom can be eliminated by the 5 Lorentz
covariant conditions
g µν,ν = 0 gµµ = 0 (1.48)

6
2 The Poincaré group
The Poincaré group is an extension of the Lorentz group by the addition of the space-time
translations
µ
x0 = Λµν xν + aµ or x0 = Λ · x + a (2.49)
It can be realised in matrix notation as
0
x0 a0
    
x0

0
x1 Λ a1 x1
     
     
0
a2 x2
     

 x2 
 = 






 (2.50)
0

 x3



 a3 


 x3 

1 0 1 1

For infinitesimal translations we have

 
Tµ = 1 + ε Pµ + O ε2 (2.51)

The generators of infinitesimal translations being

       
1 0 0 0
0 0 0 1 0 0 0 0
       
       
       
P0 = 
 0 
 P1 = 
 0 
 P2 = 
 1 
 P3 = 
 0 


 0 


 0 


 0 


 1 

0 0 0 0 0 0 0 0

Adding a fifth row and column of zeros to the matrices (1.14) and (1.15) we get that

[ Ji , P0 ] = 0 [ Ji , Pj ] = i εijk Pk (2.52)

and
[ Ki , P0 ] = −i Pi [ Ki , Pj ] = −i δij P0 (2.53)
In addition,
[ Pµ , P ν ] = 0 (2.54)
So, the translations constitute an abelian invariant sub-algebra of the Poincaré Lie algebra,
and so the Poincaré algebra and the Poincaré group are not semisimple.
One can use a four dimensional notation and introduce the antisymmetric generators
Mµν as
1
Ki ≡ M0i Ji ≡ εijk Mjk (2.55)
2
The commutation relations for the Lorentz group become

[ Mµν , Mρσ ] = i ηνρ Mµσ − i ηµρ Mνσ − i ηνσ Mµρ + i ηµσ Mνρ (2.56)

The commutation relations (2.52) and (2.53) for the Poincaré group become

[ Mµν , Pρ ] = −i ηµρ Pν + i ηνρ Pµ (2.57)

7
 The Poicaré group has two Casimir operators. The first one in the square of momenta

P 2 = Pµ P µ (2.58)

One can check that h i h i

P 2 , Mµν = P 2 , Pµ = 0 (2.59)
The other Casimir is less trivial and it is constructed form the so-called Pauli-Lubansky
vector
1
Wµ ≡ εµνρσ P ν M ρσ (2.60)
2
which satisfies

[ Mµν , Wρ ] = −i ηµρ Wν + i ηνρ Wµ [ Wµ , Pν ] = 0 (2.61)

The second Casimir operator is

W 2 = Wµ W µ (2.62)
which satisfies h i h i
W 2 , Mµν = W 2 , Pµ = 0 (2.63)
The representations of the Poincaré group have studied by Eugene P. Wigner and fall
into three classes1 :

1. P 2 = m2 ≥ 0, (m2 real and positive), and W 2 = −m2 s (s + 1), where s is the spin and
s = 0 , 21 , 1 , 32 , 2 . . .. The representation is labelled by the mass m and the spin s, and
the states are labelled the third eigenvalues s3 of S3 component of the spin operator,
and s3 = −s , −s+1 . . . s−1 , s, and the continuous eigenvalues of Pi . Massive particle
have therefore 2 s + 1 degrees of freedom.

2. P 2 = 0 and W 2 = 0. Since Pµ W µ = 0, it follows that Pµ and Wµ are proportional,

with the constant of proportionality called helicity, and has values ±s, where s is the
spin of the representation, and s = 0 , 21 , 1 , 23 , 2 . . .. Therefore massless particles with
s 6= 0 have only 2 degrees of freedom. Note that

P02 − P~ 2 = 0 → P02 = P~ 2 → P0 = ε1 | P~ | ε1 = ±1
W02 − W ~2 = 0 → W2 = W ~2 → W 0 = ε2 | W~ | ε2 = ±1
0
P0 W0 − P~ · W
~ = 0 → P~ · W
~ = ε1 ε2 | P~ | | W
~ | → cos θ = ε1 ε2 = ±1

So
~ = λ P~
W → W 0 = ε1 ε2 | λ | P 0 (2.64)

and

P~ · W
~ = λ P~ 2 =| P~ | | W
~ | cos θ =| λ | P~ 2 cos θ → λ = ε1 ε2 | λ | → W0 = λP0
1
Eugene P. Wigner, Reviews of Modern Physics, vol. 29, n. 3, 255 (1957)

8
 So
Wµ = λ Pµ (2.65)
But

1 P~ · J~
W 0 = ε0ijk Pi Mjk = P~ · J~ = λ P 0 = ε1 | λ | | P~ | → λ = ε1 (2.66)
2 | P~ |

3. P 2 = 0 but the spin is continuous. The length of Wµ is minus the square of a positive
number. Such a representation describes a massless particle with an infinite number of
polarization states labeled by a continuous variable. A s far we know it is not realisable
in Nature.