1

Relativistic Quantum Field Theory. Fys4170 2021

The following are lecture notes for Fys 4170 from the fall 2021. Things are presented
close to how they were lectured, but not all details are presented. Some misprints might
still be present
Jan Olav Eeg

I. THE DIRAC EQUATION. A HISTORICAL DESCRIPTION.

The Schrödinger equation may be written in the form:

Hψ = Eψ . (1)

For a non-relativistic free particle the Hamiltonian is

(~p)2
H = HS = , (2)
2m

where
~~ ∂
~p → ∇ , E → i ~∂t ; ∂t ≡ . (3)
i ∂t
The Schrödinger equation is based on the non-relativistic relation between energy E and
spatial momentum p~ = m ~v :
1 (~p)2
E = m (~v )2 = . (4)
2 2m
The corresponding relativistic relation is

E 2 = (c ~p)2 + (m c2 )2 , (5)

where c is the speed of light.

The relation (5) is the starting point for the Klein-Gordon equation for the field Φ (at
first erronously intepreted as a wave function):
h i
(i~ ∂t )2 Φ = ~ 2 + (m c2 )2 Φ ,
(−ci~ ∇) (6)

which in covariant form with four-momentum

E ∂ i~ ~~
pµ = ( , ~p) → i ~ ∂ µ = i ~ = ( ∂t , ∇) , (7)
c ∂ xµ c i
 2

can be written :
pµ pµ Φ = (m c)2 Φ . (8)
~ with ∂~ = − ∇.
Note that ∂ µ = (∂0 , ∂) ~
The adjoint field Φ† satisfy the same equation:

pµ pµ Φ† = (m c)2 Φ† . (9)

If we multiply the equation (8) with Φ† from left and (9) with (-Φ) from right and sum we
obtain the continuity equation
∂ ρKG

µ
∂µ jKG = ~ · ~jKG = 0 ; j µ ≡ i Φ† ∂ µ Φ − (∂ µ Φ† ) Φ .
+ ∇ (10)
KG
∂t
The Klein-Gordon describe a spinless particle (s = 0). But because ρKG might be negative
it cannot be intepreted as a probabillity density such as in non-relativistic (Schrödinger)
quantum mechanics with probabillity density ρS = ψS† ψS . The Klein-Gordon equation was
therefore rejected as basis for a (possible ) relativistic quantum mechanics. However, in
quantum field theory (QFT) it is fine that ρKG might be both positive and negative because
eρKG will be intepreted as an operator for charge density. (Here e is the electric charge).
The KG equation has solutions with both positive and negative energy. Thus, for a free
p
particle (and for ~ = c = 1), these energies are E = ±Ep , with Ep = (~p)2 + m2 . The
plane wave solutions of the KG equation for positive and negative energy have the form

Φ(+)
p = N e−ip·x , Φp(−) = N e+ip·x , (11)
p
where N = 1/ 2Ep is a normalisation factor and p · x = Ep t − p~ · ~x . Note that the spatial
momentum is quantised to −~p when the energy is negative. The plane wave solutions Φ(±)
then satisfy the ortogonality relation
Z  
(ε) (ε′ ) (ε′ ) † (ε′ )
(Φp , Φp′ ) = i d3 x (Φ(ε)
p (x)) †
(∂ Φ
t p′ (x)) − (∂ Φ(ε)
t p (x)) Φp′ (x)

= ε δεε′ (2π)3 δ (3) (~p − p~′ ) . (12)

In QFT one uses units such that ~ = c = 1. The motivation is simplification of formulae
(which often contains powers of ~ and c). Using these units causes no problem because the
physical units may be retained in the end.(Still ~ and c are kept in the wave eqs below)

Dirac required an equation of Schrödinger form with a Hamiltonian

~ · p~ + βmc2 ,
HD = c α (13)
 3

where α
~ and β are operators which do not depend on space-time coordinates x = (ct, ~x) or
on ∂ µ . Because HD must be hermitean, α
~ and β must also be hermitean:

~† = α
α ~ , β† = β . (14)

In order that (13) should be the equation for a relativistic particle, HD must satisfy the
equation:
(HD )2 = (c ~p)2 + (mc2 )2 . (15)

Squaring HD in (13), we see that in order to satisfy (15), β and αi (where i = 1, 2, 3) must
satisfy anti-commutation rules, and their square must be equal to the unity operator 1 in
the space which αi and β act :

β2 = 1 , { α
~ , β } = 0 , {αi , αj } = 2 δ ij 1 . (16)

Operators with this kind of algebra can be represented by matrices. The matrices must be
quadratic according to (14).
One might try to represent β and αi with 2 × 2 matrices by choosing αi = σ i og β = 1
but then β og αi would not anti-commute. (However, for mass m = 0 one may use αi = σ i
because β is irrelevant in this case. Then parity symmetry is broken, which is also the case
for massless neutrinos. See later).
Because the matrices αi and β have a square equal to 1 they must have eigenvalues equal
to ±1. Further, because the four matrices anti-commutes one may easily show that they
have a trace equal to zero:

T r(β) = 0 , T r(αi ) = 0 . (17)

If one for example choose β diagonal there must be equally many +1 as −1 on the diagonal.
We have already excluded 2 × 2 matrices. 3 × 3 matriser with eigenvalues ±1 cannot be
traceless, and therefore the next possibillity is to try 4 × 4 matriser. There are infinitely
many sets of β and αi which satisfy (16). If one finds one set which fits, one may find
another one by using a similarity-transformation. The representation which is mostly used
is the Pauli-representation which has β diagonal and can be written in block form
   
1 0 0 ~σ
βP =   , α ~P =   , (18)
0 −1 ~σ 0
 4

where    
1 0 0 0
1 ≡ 12×2 =   , 0 ≡ 02×2 =   . (19)
0 1 0 0
The Pauli-representation is for instance useful when studying the non-relativistic limit for
the Dirac equation. Another representation is the Kramer-Weyl (KW) representation:
   
0 1 −~σ 0
βKW =   , α ~ KW =  , (20)
1 0 0 ~σ

which is useful when studying the Lorentz transformations of the Dirac equation. Note,
however, that the spin operator
 
k
~ k σ 0
Sk = Σ ; Σk =   = γ5 α k . (21)
2 0 σ k

is the same in all representations. For the pseudoscalar matrix γ5 we have

i
γ5 = εµνσρ γ µ γ ν γ σ γ ρ ; (γ5 )2 = 14×4 , (22)
4!

where
γ µ = (γ 0 , ~γ ) ; γ 0 ≡ β ; ~γ ≡ β α
~ (23)

Note that α ~ describing spin is an axial vector operator. For

~ is a vector operator, while Σ
the two representations, Pauli- og Kramer-Weyl- , we have
   
0 1 −1 0
(γ5 )P =   ; (γ5 )KW =   (24)
1 0 0 1

The relation between the Kramer-Weyl (KW)-representation and the Pauli (P) represen-
tation (which has diagonal γ0 ) is given by
 
1 1 1 
(~ α)P S −1
α)KW = S (~ ; S = √ . (25)
2 1 −1

and similar for β and γ5 .

Multiplying the Dirac equation (13) from left β and using space-time derivatives, we find
the covariant version of the Dirac equation:

i~ γ µ ∂µ ψ = m c ψ (26)
 5

In the KW representation the Dirac matrices have the nice form:

 
µ
0 σ
(γ µ )KW = (γ 0 , ~γ ) =   , (27)
eµ 0
σ

where
σ µ ≡ (1, ~σ) , and similarly eµ ≡ (1, −~σ ) .
σ (28)

Taking the hermittean conjugate of equation (26) we find the adjoint Dirac equation

−i~ (∂µ ψ)γ µ = m c ψ ; ψ ≡ ψ† β . (29)

It is found that the Dirac equation has solutions with both positive and negative energy.
Thus, for a free particle (and for ~ = c = 1), these energies are E = ±Ep , with Ep =
p
(~p)2 + m2 . The plane wave solutions of the Dirac equation for positive and negative
energy have the form

(+)
ψpσ = N uσ (~p)e−ip·x (−)
, ψpσ = N vσ (~p)e+ip·x , (30)
p
where N = 1/ 2Ep is a normalisation factor and p · x = Ep t − p~ · ~x . Note that the spatial
momentum is quantised to −~p when the energy is negative. Within the Pauli representation,
the u- and v-spinors are given by:
   
~
σ ·~
p
χσ χ
Ep +m σ
(uσ (~p)) = K 
P
 , (vσ (~p))P = K   . (31)
~
σ ·~
p
χ
Ep +m σ
χσ
p
where K = Ep + m is some other normalization factor, and χσ are Pauli-spinors satisfying
χ†σ χσ′ = δσσ′ , where σ, σ ′ are spin quantum numbers.
In the KW representation, the u- and v spinors are
   
√ √
p · σχσ p · σ χσ
(uσ (~p))KW =  p  ; (vσ (~p))KW =  p  . (32)
p·σ eχσ − p·σ eχσ

as given in the book of Peskin and Schroeder. Note that with cosh(η) = Ep /m, one has

p · σ = m (cosh(η) − sinh(η)σp ) = m exp(−ησp ) , (33)

where σp = p~ · ~σ /|~p|. Then one obtains

√ √ 1 √  η η 
p·σ = m exp(− ησp ) = m cosh( ) − sinh( )σp , (34)
2 2 2
 6

where r r
η Ep + m η Ep − m
cosh( ) = ; sinh( ) = . (35)
2 2m 2 2m
p
One obtains similar relations for p · σ
e when σp → −σp .
The plane wave solutions satisfy the ortogonality relations
Z
(ε′ )
† (ε′ )
(ε)
(ψpσ , ψp′ σ′ ) = (ε)
d3 x ψpσ (x) ψp′ σ′ (x) = δεε′ δσσ′ (2π)3 δ (3) (~p − p~′ ) . (36)

The Dirac spinors have four degrees of freedom, corresponding to two energy signs and two
spin states for each energy sign.
The Dirac equation gives the correct answer for “relativistic hydrogen”, when the
Coulomb potential V = −e2 /|x| is added to the Dirac-Hamiltonian in (13). But still it
was considered problematic with the negative energy states which had to be there because
of mathematical reasons.
In order to cure the “negative energy problem”, one tried to use the Dirac “sea -theory”
where vacuum should correspond to a state where all the negative energy states were filled
. Abscence of a negativ energy state (a “hole in the sea”) should then be intepreted as
an anti-particle. But by doing so, one is leaving the intepretation of the Dirac equation
as a description of solely one particle. Even if one can find correct answers by using the
“sea-theory”, the basic problem can only be solved within QFT. A consistent relativistic
quantum mechanics does not exist, because the negative energy solutions cannot be ignored
and because the number of particles are not conserved in a relativistic theory, only the
energy ( cfr. E = mc2 ).

Multiplying the equation (26) with ψ from left and (29) with (-ψ) from right and sum,
we find the continuity equation

µ ∂ ρD ~ · ~jD = 0 ; j µ ≡ ψ † γ µ ψ
∂µ jD = +∇ D
∂t
0
jD ≡ ρD = ψ † ψ ; j~D = ψ † α~ψ (37)

The Dirac equation describes a particle with spin s = 1/2. It was thought that ρD should
not be negativ, and this was good news for a possible relativistic quantum mechanical
description. But within QFT this is not a point. There the charge density eρD could also
be negative because particles and anti-particles have opposite charges. Mathematically, this
is obtained because the operators ψ og ψ contain anti-commuting components.
 7

II. GROUP THEORY

: A group G = {gi } is a set (collection of elements) with the following properties:

* ∃ multiplication such that gi gj ∈ G ;
* ∃ a unity element g0 = 1 such that gi g0 = g0 gi = gi ;
* (g1 g2 ) · g3 = g1 · (g2 g3 ) (Associative law)
* ∃ an inverse (gi )−1 for all gi ∈ G

A. Lie groups

In field theory and particle physics we will mainly use continuous groups, called Lie
groups. A Lie group is a group where the elements are functions of continuous parameters:

g(α1, α2 , ...., αn ) , (38)

where all αi might be chosen real.

Within field theory and particle physics the most important groups are
 iα
U(1) = e ; α∈R , (39)

i.e. the collection of all phase factors, and


SU(n) = U = n × n matrix (n = 2, 3, ..) ; Ukl ∈ C , U † U = 1 , detU = 1 (40)

Here R is the set of all real numbers, C is the set of all complex numbers, and k and l run
from 1 to n. SU(n) has n2 − 1 independent parameters (which can easily be shown)

B. Lie Algebra

The group elements in SU(n) may be written

( )
X
U = exp −i αk Lk , (41)
k

where αj ; j = 1, 2, ..., (n2 − 1) is a set real parameters. and the generators

 
∂U
Lk = i , (42)
∂αk {αj =0}
 8

satisfies a Lie algebra (sum over m is understood):

[Lk , Ll ] = ifklm Lm , (43)

where fklm is a set coefficients (structure constants of the group) completely antisymmetric
in the indices klm.
If {Lk } is a set of generators for a group, we may obtain a new set of generators {L′k }
(-which also satisfies the Lie algebra in (43)) by means of a similarity transformation:

Lk → L′k = S Lk S −1 , (44)

where S is an n × n matrix (with existing inverse). Because

U = eA ⇒ detU = eT rA , (45)

we obtain from detU = 1 that

T rLk = 0 , (46)

for all k = 1, 2, ..., (n2 − 1).

C. Irreducible representations

For all Lie groups, a representation which can be transformed to block diagonal form by
means of a similarity transformation, i.e.
 
U (A) 0
S U S −1 =   , (47)
(B)
0 U

will be called reducible. The fundamental representations for SU(n) are all irreducible,
i.e. they can not be transformed to block diagonal form. For SU(2) the representations
corresponding to spin s = 1/2, 1, 3/2, 2, ... (also called 2, 3, 4, 5, ...) are irreducible. For
SU(3) the representations 3, 6, 8, 10, ... are irreducible. (This subsection has not been
mentioned in the lectures, only available in the separate “Group theory note”.)

D. Contragredient representations

Let {Mk } be a reprentation of {Lk }, such that [Mk , Ml ] = ifklm Mm . For all SU(n) groups
it can be shown that the structure constants fklm are real, and therefore all M fk = − M ∗
k
 9

will also satisfy the Lie algebra. The representations {U} and {U c } are given by

∗ f
U = e−iαk Mk , U ∗ = eiαk Mk = e−iαk Mk = U c , (48)

will be called mutually contragredient.

It is an axiom within particle physics that particles and the corresponding anti-particles
transform after mutually contragredient representations. For SU(2) the mutually contra-
gredient representations are equivalent. (They are in worst case related by a similarity
transformation. Physically, there is, for instance, no “anti-spin”). For SU(3) there are
contragredient representations 3, 6, 8, 10. (Here 8 is equivalent to 8). Under colour SU(3),
quarks transform under the 3 representation and anti-quarks under the 3 representation.
(This subsection has not been mentioned in the lectures, only available in the separate
“Group theory note”.)

E. The group SU(2)

: In the fundamental representation of SU(2) the generators are given by the Pauli
matrices (k=1,2,3):
1 (1/2)
Lk = σk ≡ Sk , (49)
2
which corresponds to spin 1/2. Higher representations (s = 1, 3/2, ...) have the same Lie
algebra
h i
(s) (s) (s)
Sk , Sl = i ǫklm Sm , (50)

where ǫklm is the Levi-Civita tensor in three dimensions. But other matrix relations are
different for different spin. For example, σj 2 = 1 is only valid for spin 1/2. For spin
(1) (1) (1)
1, (Sj )2 6= 1, but (Sj )3 = Sj . The spin 1 representation of SU(2) corresponds to the
fundamental representation of the rotation group, often given in the cartesian representation,
(1)
namely : (Sk )ln = −iǫkln .

F. Connection between SU(2) and spatial rotations

Let us consider an arbitrary three vector ~x. We define the 2x2 matrix

S ≡ ~x · ~σ ⇒ detS = −(~x)2 . (51)

 10

The determinant is easily found by means of the standard expressions for the Pauli matrices.
Note also that T r(S) = 0 because the Pauli matrices are traceless. Now we let S transform
after Uǫ SU(2) :
S → S ′ = USU † . (52)

Here S ′ will have no contribution from the unit matrix because

T r(S ′) = T r(USU † ) = T r(U † US) = T r(S) = 0 . (53)

Then S ′ can be written in terms of a transformed quantity (vector) ~x ′ :

S ′ = ~x ′ · ~σ ⇒ detS ′ = −(~x ′ )2 . (54)

On the other hand,

detS ′ = detU · detS · detU † = |detU|2 · detS = detS , (55)

such that the transformation (52) generates a rotation of ~x, conserving its lenght:

(~x ′ )2 = (~x)2 . (56)

A rotation may in general be expressed as

xi → x′i = Rij xj . (57)

From (51) and (54) and T r(σi σj ) = 2δij we find the coefficients

1

Rij = T r σi Uσj U † . (58)
2

Also, using (52) and (57) one finds

U † σi U = Rij σj . (59)

If we write
i
U = exp(− θ ~n · ~σ ) = cos(θ/2) − i~n · ~σ sin(θ/2) , (60)
2
this represents a rotation an angle θ around the direction ~n (we have |~n| = 1). In this case
we find
Rij = cosθ δij + (1 − cosθ) ni nj + sinθ ǫijk nk , (61)
 11

which is the rotation matrix in three spatial dimensions. Further, ~v ≡ χ†~σ χ will transform
as a vector, i.e.
vi → vi′ = Rij vj when χ → χ′ = Uχ , (62)

and χ is a two component spinor.

Note that the connection between SU(2) and the rotation group is not one to one. For
example, it is seen from (52) that both U and (−U) correspond to the same rotation.

G. Lorentz transformations

A Lorentz transformation of a four vector xµ = (x0 , ~x) is in analogy with (57) given by

xµ → x′µ = Lµ ν xν , (63)

where
ν
(x′ )2 ≡ x′µ x′µ = xµ xµ ≡ x2 ⇒ Lα µ Lα ν = g µ ν ⇒ (L−1 )µ = Lν µ . (64)

The Lie algebra for the Lorentz group is given by

[Jk , Jl ] = iǫklm Jm , [Jk , Kl ] = iǫklm Km , [Kk , Kl ] = −iǫklm Jm , (65)

where Jk are the rotation generators and Kl are the boost generators. In the Minkowski
representation the rotation generators are in principle 4x4 matrices, but with only zeros for
all time components (components where at least one index is 0) and the usual 3x3 rotation
generator matrices in the lower right corner. For Kx ≡ K1 we have (Kx )01 = (Kx )10 = −i,
and the rest of the elements equal to zero. A pure boost along the unit vector e may be
written
~ ~ − (coshψ − 1)(~e · K)
~ 2,
L = e−iψ ~e·K = 1 − i sinhψ ~e · K (66)

where we have used

~ 3 = −~e · K
(~e · K) ~ and coshψ = (1 − v 2 /c2 )−1/2 . (67)

The 6 generators may also be collected in an antisymmetric tensor J µ ν with :

J k l = ǫkln Jn ; Jl 0 = Kl (68)

with the algebra

[J µ ν , J σ ρ ] = i(g νρJ µ σ − g µρ J ν σ − g νσ J µ ρ + g µσ J ν ρ ) . (69)

 12

H. Connection between SL(2C) and Lorentz transformations

The group SL(2C) is defined by

SL(2C) = {A = 2 × 2 matrix ; Akl ∈ C , detA = 1} . (70)

The group SL(2C) is related to the Lorentz group in the same way as SU(2) is related to
the rotation group. And SU(2) is a subgroup of SL(2C). In the SL(2C) representation,
the generators are given by
1 ~ = i ~σ ,
J~ = ~σ and K (71)
2 2
which is easily seen to satisfy the algebra in (65). For A = U,i.e a unitary matrix, we
therefore have A = (A−1 )† A generalisation of subsection F might be obtained by first
defining the matrices

σ µ ≡ (1, ~σ ) and eµ ≡ (1, −~σ ) .

σ (72)

Then we define the two by two matrices

T ≡ xµ σ µ and Te ≡ xµ σ
eµ , (73)

and we find that

detT = detTe = xµ xµ (74)

We now perform transformations with (A−1 )† and A , which gives

T → T ′ = (A−1 )† σ · xA−1 = σ · x′ = σ µ Lµ ν (A) xν ,

Te → Te′ = Ae
σ · xA† = σ eµ Lµ ν (A) xν .
e · x′ = σ (75)

Using Lµ ν (A) = Lν µ (A−1 ) and letting A−1 → A throughout the equations in (75), one
obtains
A−1 σ
eµ (A−1 )† = Lµ ν (A)e
σν , A† σ µ A = Lµ ν (A)σ ν (76)

We have
T ′ = x′µ σ µ ⇒ detT ′ = (x′ )2 . (77)

On the other hand,

detT ′ = detA · detT · detA† = |detA|2 · detT = detT , (78)

 13

such that the norm (“length”) of the four vector is conserved:

(x′ )2 = x2 . (79)

The same is valid for Te′ . Thus an element of SL(2C) generates Lorentz transformations.
We find in analogy with (58) :

1

Lµ ν (A) = eµ Aσν A† .
Tr σ (80)
2

Further, the quantity v µ ≡ χ† σ µ χ will transform as a four vector, i.e..

vµ → vµ′ = Lµ ν vν when χ → χ′ = Aχ , (81)

and χ is a two component spinor. One can show that an arbitrary element of SL(2C) can

be written as
A = H ·U where H † = H and U † = U −1 , (82)

which means that any Lorentz transformation might be written as a boost times a rotation.
Furthermore, H can be written in analogy with U in (60):

~ = exp( 1 ψ ~e · ~σ ) = cosh(ψ/2) + ~e · ~σ sinh(ψ/2) .

H = exp(−iψ ~e · K) (83)
2

It is an important point that the 2×2 matrices in SL(2C) are simpler to deal with than the
4×4 matrices L in the Minkowski representations.

I. The Dirac representation of the Lorentz group

Studying Lorentz transformations within Dirac theory it is an advantage to use the Weyl
(or Kramer-Weyl) representation (KW) of the Dirac matrices:
 
µ
0 σ
γ µ = (γ 0 , ~γ ) =   ≡ (γ µ )
KW . (84)
µ
σ
e 0

The relation between the KW-representation and the Pauli (P) representation (which has
diagonal γ0 ) is given by
 
1 1 1
(γ µ )KW = S (γ µ )P S −1 ; S = √   , (85)
2 1 −1
 14

where 1 means the 2 by 2 unit matrix.

The transformation for a Dirac-spinor Ψ under a Lorentz transformation is (Λ is here
given in the KW representation with A ∈ SL(2C))
 
(A−1 )† 0
Ψ(x) → Ψ′ (x′ ) = Λ Ψ(x) ; Ψ′ = Ψ Λ−1 ; Λ =   . (86)
0 A

The Dirac representation of the rotation and boost generators are

1~ ~ = iα
J~ = Σ , K ~ , (87)
2 2
where (in all Dirac representations, like in KW and P):
 
~σ 0
~ = 
Σ  , and α ~ .
~ = γ 0 ~γ = γ5 Σ (88)
0 ~σ

In the KW-representation
 
−1 0
γ5 =   , (89)
0 1

The analogue of (66) is

~ = exp( 1 ψ ~e · α
Λ = exp(−i ψ ~e · K)
ψ ψ
~ ) = cosh( ) 1 + sinh( ) ~e · α
~. (90)
2 2 2
Using the KW representation for the Dirac matrices, Λ given as in (86) and the eqs. in
(76), we obtain the transformations for the Dirac-matrices:

Λ−1 γ µ Λ = Lµ ν γ ν , (91)

The Dirac equation has the same form in the transformed and the untransformed frame:

(γ · ∂ − m ) Ψ(x) = 0 ; (γ · ∂ ′ − m ) Ψ(x′ ) = 0 , (92)

where ∂µ′ = Lµ ν ∂ν denotes derivative with respect to the Lorentz transformed coordinates
x′µ = Lµ ν xν . We define

S(x) ≡ Ψ(x) Ψ(x) ; jµ (x) ≡ Ψ(x) γµ Ψ(x) . (93)

Using (86) and (91), we find that the quantities S and jµ transform as a Lorentz scalar and
a Lorentz vector, respectively:

S ′ (x′ ) = S(x) ; jµ′ (x′ ) = Lµ ν jν (x) . (94)

 15

The KW representation is sometimes also called “chiral basis” because the right- and left-
handed projections given by
1 1
ΨR ≡ R Ψ , ΨL ≡ L Ψ ; R ≡ (1 + γ5 ) , L ≡ (1 − γ5 ) , (95)
2 2
have zero upper and lower two-components.

III. LAGRANGE FORMALISM FOR FIELDS.

We are familier with the Lagrange formalism for mechanical systems. Here one defines
an action Z
W = dtL (96)

where the lagrange function L is a function of the coordinates qi of the system and their
time derivatives dqi /dt. The canonical momenta here is the time derivative of L with respect
to dqi /dt. Normally the kinetic energy minus the potential energy L = T − V . Then
the dynamical equations for a system is obtained by a variation principle leading to Euler
Lagrange equations
In field theory,the canonical coordinates are some fields Φi and their space-time derivatives
∂µ Φi . The fields might be Klein-Gordon, Dirac, Maxwell fields. The action (contains one
field per particle type):
Z Z
W = dtL = d4 x L(Φi , ∂µ Φi ) . (97)
Ω

where L is the Lagrangian density for the system of fields, and where Ω is some space-time
volume,- which might be the complete space-time.
Performing a small variation of the fields small variation Φi → Φi + δΦi ⇒ one obtains
the first order variation
Z  
∂L 4 ∂L
δW = dx δΦi + δ(∂µ Φi ) (98)
Ω ∂Φi ∂(∂µ Φi )
This can be reformulated to
Z    Z  
4 ∂L ∂L 4 ∂L
δW = dx − ∂µ + d x∂µ δΦi ; (99)
Ω ∂Φi ∂(∂µ Φi ) Ω ∂(∂µ Φi )
Now the last term can be rewritten as a surface term which vanish. Moreover, according
to the variational principle: One requires δW = 0 for arbitrary (small)first order variations
 16

inside the space-time voluume Ω. Then we obtain theEuler-Lagrange’s equations:

 
∂L ∂L
− ∂µ = 0 (100)
∂Φi ∂(∂µ Φi )

for all fields i = 1, 2, 3...

One defines canonical momenta:

∂L
πi ≡ , (101)
∂(∂t Φi )

where ∂t = ∂/∂t . Then one defines a Hamiltonian as an integral over a Hamiltonian density:
Z
H = d3 x H where H = πi (∂t Φi ) − L ; (102)

Now, the field equations are known from before. Thus we need to find a suitable la-
grangian densities L which via the Euler-Lagrange equsations reproduces the known field
equations.
For the Klein-Gordon case we find the suitable Lagrangian density

LKG = ∂µ φ† ∂ µ φ − m2 φ† φ . (103)

We find

∂L ∂L ∂L ∂L
†
= −m2 φ , †
= −m2 φ , = ∂ µ φ† , = ∂µφ , (104)
∂φ ∂φ ∂(∂µ φ) ∂(∂µ φ† )

leading, via the Euler Lagrange equations to the known Klein-Gordon equation for φ and φ†

(∂µ ∂ µ + m2 )φ = 0 , (∂µ ∂ µ + m2 )φ† = 0 (105)

The corresponding cononical momenta are

πφ = ∂t φ† ; πφ† = ∂t φ ; (106)

Thus the Hamiltonian density will be

 †  
HKG † ~
= (∂t φ) · (∂t φ) + ∇φ ~
· ∇φ + m2 φ† φ (107)

Here the term

 †    
~
∇φ ~
· ∇φ ~ · φ† ∇π
= ∇ ~ − φ † ∇2 φ . (108)
 17

which- using the KG equation may be reformulated such that the Hamiltonian gets a very
simple form, namely, we find
 
~ · φ† ∇π
HKG = (∂t φ† ) · (∂t φ) − φ† (∂t2 φ) + ∇ ~ (109)

Integrating over space to get the Hamiltonian, the last term will be a surface term which
can be neglexted, and we obtain HKG as a scalar product
Z
HKG = d3 x HKG = i(φ, (∂t φ)) (110)

where the scalar product for KG fields is defined as

Z


(φA , φB ) = i d3 x φ† (∂t φB ) − (∂t φ† )φB (111)

For the Dirac equation one finds the suitable Lagrangian density

LD = ψ (γµ i∂ µ − m)ψ , (112)

As shown before, the γ matrices satisfy the anticommutator relations

{γ µ , γ ν } = 2g µν 1 ; (113)

Now we find
∂L ∂L ∂L ∂L
= −mψ , = (iγµ ∂ − m) ψ , = iψγ 0 = iψ † , = 0. (114)
∂ψ ∂ψ ∂(∂µ ψ) ∂(∂µ ψ)

Then we obtain the Dirac equation for ψ and the adjoint equation for ψ read :

(γµ i∂ µ − m)ψ = 0 , −i∂ µ ψ γµ = m ψ . (115)

The canonical momenta can be read off equation (114)

πψ = iψ † ; πψ† = 0 (116)

The Dirac equation has a positive definite scalar product, as mentioned in the note “The
Dirac equation...” available from the web-page of the course. The Hamiltonian form is :

HD = πψ (∂t ψ) − LD = ψ hD ψ , hD ≡ α ~ + mβ
~ · (−i∇) (117)

For the Maxwell case the Lagrangian density is

1 1 ~2 
~ 2) ,
LM = − F µν Fµν = E −B (118)
4 2
 18

where
F µν ≡ ∂ µ Aν − ∂ ν Aµ ; (119)

Using this definition one obtains

1 1
LM = − (∂ µ Aν )(∂µ Aν ) + (∂ µ Aν )(∂ν Aµ ) . (120)
2 2

Taking derivatives by using

∂(∂µ Aν )
= gµα gνβ , (121)
∂(∂ α Aβ )
we obtain the equation of motion
∂µ F µν = 0 (122)

which give us half of the Maxwell equations without sources. The other half of the equations
is given by
∂µ ǫµνσρ Fσρ = 0 (123)

which is automatically valid by symmetry

The canonical momenta are

∂L
πAµ = , giving π k = E k and π 0 = 0 (124)
∂(∂t Aµ )

Here
~ = −∇A
E ~ 0 − ∂t A
~ ; B
~ = ∇
~ ×A
~ (125)

Then the Hamiltonian densitie for Maxwell case is

1~ 2 
~ 2 + ∆HM
HM = − E k (∂t Ak ) − LM = E⊥ + B (126)
2

Normally
1 ~2 ~ 2)
HM = (E + B (127)
2
but now
1 ~ 2 ~ 2 ) + ∆HM
HM = (E⊥ + B (128)
2
where ∆HM contains the Coulomb interaction.
 19

IV. NOETHER’S THEOREM

Noethers theorem, due to Emily Noether states that for each symmetry transformation
under which the Lagrangian is symmetric,- there is a correnponding conserved quantity.
Consider a (infinitesimal) transformation of the fields

Φi → Φ′i = Φi + δΦi (129)

Then we have a priori the same mathematics as in equation (98)-(99)

    
∂L ∂L ∂L
δL = − ∂µ δΦi + ∂µ δΦi ; (130)
∂Φi ∂(∂µ Φi ) ∂(∂µ Φi )
(with sum over i as usual). However, the intepretation of this equation is different from in
(98)-(99). Namely, now the transformation is known, and the fields are assumed to satisfy
the Euler Lagrange equations such that the first term is zero. Thus, if the Lagrangian
density is symmetric under the transformation (129), then
∂L
∂µ f µ = 0 , where f µ = δΦi (131)
∂(∂µ Φi )
Then the integral Z
d3 xf0 = constant(conserved) (132)

Example: The Dirac Lagrangian is invariant under the global phase factor transformation
where α is real and constant

ψ → ψ ′ = eiα ψ → (1 + iδα)ψ + O((δα)2 ) (133)

Then
∂L ∂L µ
fµ = δψ + δψ = (iψγ µ ) · (iδα ψ) = − δα jD (134)
∂(∂µ ψ) ∂(∂µ ψ)
µ
where jD is the well known Dirac current:

µ µ
∂µ jD = 0 ; where jD = ψ γµ ψ (135)
R
Then the electric charge Q = d3 x jD
0
(x) is conserved in time. Similarly for the KG case:

µ

jKG = i φ† (∂ µ φ) − (∂ µ φ† ) φ (136)

We may also consider a transformation involving γ5

ψ → ψ ′ = eiαγ5 ψ → (1 + iδαγ5 + O((δα)2)ψ (137)

 20

where α is still a constant. Here, γ5 anticommutes with γ µ such that the kinetic term in LD
in invariant, but the mass term in not:

mψψ → mψ ′ ψ ′ = mψe2iαγ5 ψ (138)

The current corresponding to this transformation is the axial current

jAµ = ψγ µ γ5 ψ . (139)

But here ∂µ jAµ = 0 only in the limit m → 0

WE will now consider invariance under space and time translations. We consider now a
total variance, due to both a change in Φi and in the coordinate. The total variation of a
function F (x) is then

∂F
δT F = F ′ (x′ ) − F (x) = (F ′ (x′ ) − F (x′ )) + (F (x′ ) − F (x)) = δF (x′ ) + δxµ (140)
∂xµ

Here δF (x′ ) = δF (x) to first order in the variations. Now

δT L = δL + (∂µ L)δxµ (141)

Here  
∂L
δL = ∂µ δΦi ; (142)
∂(∂µ Φi )
as before . Further
(∂µ L)δxµ = ∂µ (Lδxµ ) − L(∂µ δxµ ) (143)

giving  
∂L
δT L = ∂µ δΦi + Lδxµ − L(∂µ δxµ ) (144)
∂(∂µ Φi )
Now the total variation of the field is in general:

δT Φi = δ Φi + (∂µ Φi ) δxµ (145)

Now combining our equations we obtain:

 
∂L µα
δT L = ∂µ δT Φi − T δxα (146)
∂(∂µ Φi )

where
∂L
T µα = (∂ α Φi ) − L g µα (147)
(∂µ Φi )
 21

is the energy-momentum tensor. Note that for translations δT Φi = 0 and δxα is a constant.
Then Z
α
P = d3 xT 0α (148)

is a constant, and is intepretes as the four momentum.

For a Lorentz transformation xµ = Lνµ xν → xµ + δxµ where δxµ = ωµν xν where ω is
small in this case there might also be a transformation of the fields such that

δT Φi = ωµν Sijµν Φj (149)

Then one obtains

 
∂L
∂α S µν Φs (x) + xµ T αν − xν T αµ =0. (150)
(∂α Φr ) rs

which gives conserved boost and rotation (angular momentum) generators.

V. QUANTIZATION OF FREE FIELDS

A. The Klein Gordon Field

In quantum mechanics the well known commutator

[qi , qj ] = i~δij (151)

is replaced by a commutator among a field and a canonical momenta at equal times

[Φj (x) , πl (y)]x0 =y0 = iδjl δ (3) (x − y) (152)

The fields and canonical momenta commute amomng themselves at equal times

[Φj (x) , Φl (y)]x0 =y0 = [πj (x) , πl (y)]x0 =y0 = 0 (153)

We expand of the complex KG field in terms of plane waves::

Z
d3 k  (+) † (−)

φ(x) = a ϕ
k k (x) + b ϕ
k k (x) (154)
(2π)3

Thus the adjoint field is

Z
† d3 k  † (−) (+)

φ(x) = a ϕ (x) + b ϕ
k k (x) (155)
(2π)3 k k
 22

which means that ap and b†p , which before quantization were fourier coefficients will from
now be operators. The plane wave solutions are:

1  ∗
(±) (∓)
ϕk (x) = √ e∓ik·x = ϕk (x) (156)
2ωk
p
where ωk = ~k 2 + m2 is the energy for mass m and space momentum ~k. These plane wave
solutions have a orthonormal scalar product
 
(ǫ)
= ǫ δǫ,ǫ′ (2π)3 δ (3) (~k − p~)
′
ϕk (x) , ϕp(ǫ ) (x) (157)

where both ǫ and ǫ′ might be ±1, and where

Z
(φA , φB ) = i d3 x((φ)† (∂t φB ) − (∂t φ† )φB ) (158)

Note that for equal momenta this it the integgral over the zeroth comonent of the KG current
µ
jKG . Using this scalar product we find
Z
d3 x ip·x
ap = (ϕ(+)
p , φ(x)) = i p e ((∂t φ(x)) − iωp φ(x)) (159)
2ωp

and similarly
Z
(−) d3 y −ik·y

a†k = −(ϕk , φ(x)) = −i √ e (∂t φ(y)†) − iωk φ(y)† (160)
2ωk
Here the time derivaties are canonical momenta

∂t φ = πφ† and ∂t φ† = πφ (161)

Then we obtain the commutator

Z Z
† d3 x d3 y ip·x −ik·x
[ap , ak ] = p √ e e [(ωp φ(x) + iπφ† (x)) , ([(ωk φ(y)† + iπφ (y)))]
2ωp 2ωk

As the ap and a†k are time independet , we put x0 = y0 within the integral. Now the
commutators between a field and its canonical momentum is a delta function, and the
commutator between two fields and two momenta are zero, we ombtain in end

[ap , a†k ] = (2π)3 δ (3) (~p − (~k) (162)

Similarly we obtain
[bp , b†k ] = (2π)3 δ (3) (~p − (~k) (163)
 23

The rest of the commutators between the a’s and the bs are zero.
A one particle state for a-type and b-type particles are given by
√ √
|a; ~ki = 2ωk a†k |0i , and |b; ~ki = 2ωk b†k |0i (164)

The scalar product of two states are

h~p|~qi = 2ωp (2π)3 δ (3) (~p − ~q) (165)

which is Lorentz invariant (see PS page 23). Now the Hamiltonian is given by the scalar
product
HKG = i(φ(x) , (∂t φ(x))) (166)

The field φ(x) is given above, and

Z
d3 k  (+) † (−)

∂t φ(x) = ak ϕk (x)(−iωk ) + bk ϕk (x)(+iωk ) (167)
(2π)3
(−)
Using the scalar product between the various ϕk (x) we obtain the Hamiltonian
Then we obtain the Hamiltonian operator
Z
d3 k  † †

HKG = a ak + bk bk ωk (168)
(2π)3 k
However, we have to put the operators in normal order, such that
Z
d3 k a b


HKG = N k + N k ωk (169)
(2π)3
where the number operators for particles and anti-particles:

Nka = a†k ak ; Nkb = b†k bk (170)

The charge operator is (with normal order found to be :

Z
d3 k a b


Q = N k − N k (171)
(2π)3

B. The Dirac field

The Dirac field can be expanded in terms of plane wave solutions

X Z d3 p

(+) † (−)
ψ(x) = c pσ ψpσ (x) + d pσ ψpσ (x) (172)
σ
(2π)3
 24

where the plane wave solutions

(+) e−ip·x (−) e+ip·x

ψpσ = p uσ (~p) ; ψpσ = p vσ (~p) (173)
2Ep 2Ep
are given in Peskin and Schroeder and in the note “The Dirac equation. Fys4170 H21. A
historical descriptions”.
For a two-particle state, according to the Pauli principle we have

|e− (~pσ) e− (~qτ )i = − |e− (~qτ )e− (~pσ)i (174)

This cannot be fulfilled if the c’s and d’s have commutation rules, but anticommutators will
be consistent with the Pauli principle. Therefore we assume:

{ckσ , c†k′ σ′ } = {dkσ , d†k′σ′ } = (2π)3 δ (3) (~k − k~′ ) δσ σ′ (175)

all other anti-commutators are = 0. Now we obtain the Hamiltonian

Z
d3 p

HD = 3
Ep c†pσ cpσ − dpσ d†pσ (176)
(2π)
Note the relative minus due to negative enrgy solutions. The scalar product is positive
definite, and therefor we obtain
Z
d3 p

QD = c†p cp + dp d†p (177)
(2π)3
with a relative plus sign between the two terms. But HD and QD must also be in normal
order, in order to obtain

h0|HD |0i = 0 , and h0|QD |0i = 0 . (178)

To obtain this we do the replacent

dpσ d†pσ → − d†pσ dpσ (179)

Then Z
d3 p  
† †
HD = Ep cp~ cp~ + dp~dp~ (180)
(2π)3
and Z
d3 p  † †

QD = c c
~ p
p ~ − d d
~ p
p ~ (181)
(2π)3
NB! Normal order removes “Dirac sea” to give correct H and Q. The energy will always
be positive, and the charge may take both signs.
 25

C. The Maxwell field

The expansion of the Maxwell field reads:

Z
d3 k  (+) ~ † (−) ~

Aµ (x) = akλ Aµ (x; k, λ) + akλ Aµ (x; k, λ) (182)
(2π)3
Plane waves (with frequences ±)
1
A(±)
µ (x; k, λ) = √ e∓ik·x ǫµ (~kλ)(∗) (183)
2ωk

where ωk = |~k|, and ǫµ (~kλ) is the photon polarization vector. For physical photons (helicity
λ = ±1) the time component is zero and the space part is ǫ~k = (1, iλ, 0). The sign (∗) means
a complex conjugate for the negative frequency (corresponding to an outgoing photon).
Now we should remember that within Lorentz gauge each component µ = 0, 1, 2, 3 is
satisfying the Klein-Gordon equation However for the radiation part of the field only two
helicities are corresponing to real photons. Non-zero commutator (others zero):

[akλ , a†k′ λ′ ] = (2π)3 δ (3) (~k − k~′ ) δλλ′ (184)

The transverse part of the Hamiltonian is

XZ d3 k  † 
T
HM = a akλ (185)
λ
(2π)3 kλ

Note: There are some technical subtelties due to gauge invariance. A four vector field
has 4 digrees of freedom, while the physical photon has two. We come back to this.

D. generalities....

A quantized field may be intepreted as an infinite sum over harmonic oscillators, one for
each momentum ~k (and spin).
The space of states formed by action on vacuum state (ground state) |0i with creation
operators
Example: An electron-photon state might be expressed as
p √
|e− (p, σ) , γ(k, λ)i = 2Ep 2ωk c†pσ a†kλ |0i (186)

At this point we should talk about dimensions of fields and states:

 26

The action W has the dimension energy times time, which is the same dimension as
Plancks constant. There fore the action is dimensionless. For c = ~ = 1, energy, momentum
and mass has the same dimension. We call this dimension one. As plane waves has the form
exp(−ip · x), length (and time) has dimension minus one. Thus the Lagrangian density has
dimension four, and the dervative ∂µ has the dimension of momentum which is one. Then
we find that the KG field and the Maxwell field has dimension one, while the Dirac field has
dimentsion 3/2. The creation and annihilation operators have dimension minus 3/2, and a
one particle state has dimension minus one.

VI. QUANTUM ELECTRODYNAMICS (QED)

The Lagrangian density for QED is

LQED = L0 + Lint ; L0 = LD + LM ; (187)

µ
Lint = −ejD Aµ = − Hint (188)

where the interaction Lagrangian Lint has been obtained from LD by the “minimal substi-
tution”:
i∂µ → iDµ = i∂µ − e Aµ (189)

In Quantum Field Theory one uses the Interaction picture (I), where the time development
of the fields are as for free fields. A state |A, tiI in the interaction (I) picture is defines from
the Schrødinger (S) picture as

|A, tiI = exp (−iH0 (t − t0 ) |A, t0 iS (190)

If the free Hamiltonian H0 had been replaced by the total Hamilton here, we had gotten
the Heisenberg picture. The time development of the states are driven by the interaction
Hamiltonian:
d I
i~ |A, tiI = Hint (t) |A, tiI . (191)
dt
This can be integrated to
Z t
1 I
|A, tiI = |A, t0 iI + dt1 Hint (t1 )|A, t1 iI . (192)
i~ t0
 27

Then we can do an iteration by writing down this for time t1

Z
1 t1 I
|A, t1 iI = |A, t0 iI + dt2 Hint (t2 )|A, t2 iI , (193)
i~ t0

and insert into the first equation to obtain

Z Z t Z t1
1 t I 1 I I
|A, tiI = |A, t0 iI + dt1 Hint (t1 )|A, t0 iI + 2
dt1 dt2 Hint (t1 )Hint (t2 )|A, t2 iI
i~ t0 (i~) t0 t0
(194)
Here, as explained in PS page 85, one can take both integral from t0 to t and divided by 2.
A formal solution in terms of time development operator UI (in interaction picture I) is
obtained by iteration:
|A, tiI = U(t, t0 )I |A, t0 iI

In scattering or decay processes we intepret t0 = −∞ and t = +∞ ⇒ Then the initial

state is
|ii = |A, t0 = −∞iI (195)

The state of system after scattering which contains all possibilities within QED:

|fei i ≡ |A, t = +∞)iI = S|ii (196)

Thus the scattering operator S, found perturbatively (by iteration) is

Z
S ≡ U(+∞, −∞)I = T exp(−i d4 x HInt (x)) . (197)

Here T is the time ordering operator defined by:

T (A(t1 )B(t2 )) = θ(t1 − t2 )A(t1 )B(t2 ) ± θ(t2 − t1 )B(t2 )A(t1 ) , (198)

wherethe minus sign is valid if both the operators are fermionic fields.
The probability amplitude to measure a state |f i -chosen by experiment - after scattering
(eventually decay in weak processes, to be desribed later) is:

hf |fei i = hf |S|ii ≡ Sf i (199)

We consider the perturbative expansion

X Z
(n) (0) (1)
S = S ; where S = 1 ; and S = −i d4 x HInt (x) (200)
n
 28

These terms are not interesting. The zeroth term is where nothing happens. The second
term in some sense illustrates the Feynman rules (se later). But it does not contribute to
pysicaal processes. Note: HInt (x) is in normal order N [...] ! The second term, which term
contributes to important QED processes, is
Z Z   
(2) (−i)2
S = T d4 xd4 y ψ(x)eγµ Aµ (x)ψ(x) ψ(y)eγν Aν (y)ψ(y) (201)
2!
and in general
Z Z Z
(n) (−i)n
S = T ... d4 x1 d4 x2 ...d4 xn × HInt (x1 ) HInt (x2 )....HInt (xn ) (202)
n!
To handle such operators one uses Wick’s theorem which states that a time ordered product
of operators are equal to the normal ordered product of the operators plus all possible
contractions:
X
T (A1 A2 A3 .... An ) = N (A1 A2 A3 .... An ) (203)
C

We sum over all possible contractions within the normal ordered product. The zeroth
order means no contractions, Then there are single contractions, double etc. By some
complicated arguments one may show that there should be no contractions in same space-
time point!
In general the contraction between two operators is

C (A(x)B(y)) = h0|T (A(x)B(y)) |0i . (204)

(The standard notation is to use a line between the contracted operators, which is too
complicated for me to do in latex). Only a few contractions are non-zero. Because the
components of the Dirac field and the components of the Maxwell field commute completely,
it is easily shown that a contractions that contractions involving a Dirac field and a Maxwell
 
field are zero. Also the contraction C (ψ(x)α ψ(y)β ) and C ψ(x)α ψ(y)β are zero. An
important contraction which is non-zero is the Feynman Dirac propagator:
   
C ψ(x)α ψ(y)β ≡ SF (x − y)αβ = h0|T ψ(x)α ψ(y)β |0i

. This contraction can bewritten

XZ d3 p  (+)
(+)
SF (x − y)αβ = 3
θ(x0 − y0 ) ψpσ (x)α ψpσ (x)β (205)
σ
(2π)

(−) (−)
+ θ(y0 − x0 ) ψpσ (x)α ψpσ (x)β .
 29

This can also, via spin sums for u and v-spinors be written as an integral over d4 p over the
contour shown in page 31 in PS.(See also page 63).
Z
d4 p
SF (x − y) = SF (p) e−ip·(x−y) (206)
(2π)4
where
(γ · p + m) i
SF (p) = i 2 2
≡ . (207)
(p − m ) (γ · p − m)
This means that the Feynman Dirac propagator can be written as a sort of “inverse” Dirac
Lagrangian:
i
SF (x − y) = δ (4) (x − y) (208)
(γ · i∂ − m)
where
Z
(4) d4 p −ip·(x−y)
δ (x − y) = e (209)
(2π)4
The Lagrangian for the Maxwell field has no such inverse due to gauge invariance. On
general grounds one writes
Z
d4 k
DF (x − y)µν = h0|T (A(x)µ A(y)ν ) |0i = 4
DF (k)µν e−ik·(x−y) ,
(2π)
where
i(−gµν + rµν )
DF (k)µν = . (210)
k 2 + iǫ
Further, this rµν is, due to Lorentz invariance, expected to depend solely on kµ , and it is
therefore assumed that rµν = α kµ kν /k 2 , where α is a real number. (see the Path inte-
gral section in the end). In diagrams without loops, rµν is irrelevant because of “current
conservation”, in momentum space

kµ uσ′ (p′ )γ µ uσ (p) = 0 ; where k = (p − p′ ) . (211)

The second order term is a sum of 6 terms,

S (2) = S (2eγ) + S (2ee) + ....... (212)

where those with only one contractions are:

Z Z h i
(2eγ) 4 4 µ ν
S = − d x d y N ψ(x) eγ Aµ (x) i SF (x − y) eγ Aν (y) ψ(y) (213)
 30

which contribute to e± γ scattering, e+ e− → γγ, γγ → e+ e− and the second which con-

tribute to e− e− , e+ e− , e+ e+ scattering is
Z Z
(2ee) 1
S = − d4 x d4 y N [e j µ (x) i DF (x − y)µν e j ν (y)] (214)
2

where jµ (x) = ψ(x)γµ ψ(x) is the Dirac current density operator as usual. The first term
S (2eγ) is actually the sum of two equal terms, and therefore the factor 1/2 is lacking here. The
dots in (212) contain two (-and even three)contractions which correspond to loop diagrams,
and will shortly be mentioned later.
For Compton scattering, the initial state is
p √
|e− (p, σ) , γ(k, λ)i = 2Ep 2ωk c†p~σ a~†kλ |0i . (215)

For field operators acting on states:

ψ (+) (x) A(+) −

µ (x) |e (p, σ)γ(k, λ)i
p √ (+)
= ( 2Ep 2ωk ) ψpσ (x) A(+)
µ (x; kλ) |0i

= uσ (~p) e−ip·x ǫµ (~kλ) e−ik·x |0i , (216)

(+)
where ψ(x)(+) is the positive frequency part of the Dirac field and ψpσ (x) is the plave
(+)
wave solution for momentum p~ and spin σ = ±1/2. Similarly Aµ (x; kλ) is the plane
wave solution of the Maxwell field with momentum ~k and helicity λ = ±1. For Compton
scattering, we find the amplitude expressed in the coordinate space:
!Z
Y p
(S − 1)f i = 2En d4 xd4 y
n∈f i
(+)
ψpf σf (x) ieγ µ A(−) ν (+) (+)
µ (x; kf λf ) SF (x − y) ieγ Aν (y; ki λi ) ψpi σi (y)
 
+ γi ↔ γf
(2)
= (2π)4 δ (4) (Pi − Pi ) Mf i . (217)

The Scattering amplitude will contain this exponential e−ip·x , and similar exponentials from
point x and also point y from the plane wave solutions and the Feynman propagator. Then
R 4 R
d x and d4 y will give conservation of four momentum at point x and y . In momentum
space, we get the scattering amplitude for Compton scattering: e− (pi )γ(ki ) → e− (pf )γ(kf )
 31

as the sum of two terms (here, say, ui is short hand for uσi (pi )):
 
(2)
Mf i e− γ → e− γ =
−e2 uf γ µ ǫµ (kf )iSF (pi + ki ) γ ν ǫν (ki ) ui
 
+ ǫµ (kf ) ↔ ǫµ (kf ) ; ki → −kf (218)

where the ǫ’s are the polarisation vectors introduced in (183). This illustrates the Feynman
rules in momentum space, as for example seen in PS, page 118 and 123. Note that the
momenta in SF are now off-shell because (pi + ki )2 6= m2 and (pi − kf )2 6= m2 . The result for
e− (p)e+ (q) → γ(k1 )γ(k2 ) can be obtained from the result (218) above by the substtitution
(called “crossing”):

pi → p , pf → −q , ki → −k1 , kf → k2 . (219)

Here u(−q) wll be intepreted as v(q).

Doing all the anticommuting operator business, we obtain for e− e− elastic scattering
(ab → cd):
   
(2)
Mf i (ab → cd) = (−e2 ) uc γ µ ua iDF (pa − pc )µν ud γ ν ub
 
− c↔d (220)

The last term with minus-sign comes from c ↔ d and is in accordance with the Pauli
principle. Mathematically, this is a consequence of the anticommutation rules of fermion
fields. Note also that for e+ e− scattering we can also obtain this by crossing from the
e− e− -amplitude, namely one scatterin diagram (“t-channel”) and one annihilation diagram
(“s-channel”).
N B!:
One should remember that Feynman diagrams are llustrations of our perturbative calcu-
lations ,- with certain mathematical rules. They give our intepretation of our perturbative
formalism. Remember that what is measureable is the initial and final states, in other
words, the external particles. Later, for electroweak interactions and QCD,- we take ad-
vantage of QED, and (See textbook for explicite expressions): Then Feynman rules may,
to large extent, be read off the interaction Lagrangian ! Propagators follow from the free
particle Lagrangians.
 32

ψ

Aµ


ψ

e e

e e


γ e

γ e

R
FIG. 1: Graphical pictures (Feynman diagrams) for S (1) = i d4 x LInt (x) where LInt = −e jµ Aµ
is the “building block”(top figure) Next S (2ee) , corresponding to Møller and Bhabha scattering,
and last S (2eγ) , conrresponding to Compton scattering and the process ee → γγ.

VII. DECAY RATES AND CROSS SECTIONS

The S-matrix element Sf i has the general structure:

f
(S − 1)f i = i(2π)4 δ (4) (Ptot i
− Ptot ) Mf i (221)

where Mf i depend on spesific process. NB! This is valid for all processes/interactions
The (differential) transition probability for a process i → f is

|h f |(S − 1)f i )| ii|2 nf

d nf P = d Π (222)
hf |f i hi|ii

where dnf Π is the differential phase space (sometimes called “the number of states”) for the
 33

nf particles in the final space, borrowed fram statistical mechanics

Y V 
nf nf
d Π = d pf , (223)
n∈f
(2π)3

where V may be intepreted as a “reaction volume”.

From the orthogonality of states
 
h~pσ|p~′ σ ′ i = 2Ep (2π)3 δ (3) (~p − p~′ ) δσσ′ , (224)

such that


h~pσ|~pσi = 2Ep (2π)3 δ (3) (0) = 2Ep V . (225)

Then we obtain
Y Y
hf |f i hi|ii = (2Ef V ) (2Ei V ) . (226)
n∈f n∈i

Further, we have

f f
|(2π)4 δ (4) (Ptot i
− Ptot )|2 = V T × (2π)4 δ (4) (Ptot i
− Ptot ), (227)

where T is intepreted as the interaction time. Now we obtain

TV
d nf P = Q |Mf i |2 d3nf Φf , (228)
n∈i (2Ei V )
where
Y d3 p~n  f
d3nf Φf ≡ (2π)4 δ (4) (Ptot i
− Ptot ) (229)
n∈f
2En (2π)3
is a Lorentz invariant phase space factor, often called dLips.
For a decay process, with one particle (with mass M )in the initial state ni = 1 and we
obtain the differential transition probabillity per unit time in the rest frame of the decaying
particle).
1 3nf
d3nf ΓD
i→f = d
3nf
Pi→f /T = d Φf |Mf i |2 (230)
2M
For a differntial cross section with two particles in the final state we have also to divide by
the flux F = vrel /V to obtain
d3nf Pi→f 1
d3nf σi→f = = d3nf Φf |Mf i |2 (231)
TF 4 f (p1, p2 )
where p1 and p2 = momenta of colliding particles
p
f (p1 , p2 ) ≡ E1 E2 vrel = (p1 · p2 )2 − (m1 m2 )2 (232)
 34

For physical processes, the auxilary quantities T and V drops out.

In general, unmeasured momenta might be integrated out.
A comment about traces:
The γ-matrices are not four vectors. However they are connected to Lorentz transforma-
tions through the eq (91). Therefore the traces must be numbers which are the same time
Lorentz tensors. Thus the results must be products of gµν ’s , or eventually also the Levi
Civita tensor ǫµνσρ if γ5 is involved. One may write down the most general possible result
with arbitrary couefficients. Then the coefficients may be obtained by using the symmetry
of the trace (usic the cyclic property, say), and some simple examples for the four indices.
Cross sections for the most common processe in QED are in PS, chapter 5.

VIII. TRYING TO BUILD A THEORY FOR STRONG INTERACTIONS

At the time when pions and nucleons were beleived to be elementary one tried to construct
a theory for interactions between pions and nucleons as a field theory where photons were
replaced by pions and electrons by nucleons in an isispin symmetric way. Then one should
have Feynman diagrams like in Fig. 2. An interaction isospin symmetric Lagrangian for
strong interactions was written down::
 
p
LπN = GπN N̄ (τi Φi ) γ5 N , N =  
n

where N is the nucleon doublet containing the proton field p and the neutron field n, Φ =
the pion triplet (π ± , π 0 ) and τi the Pauli matrices which here represents isospin.
From the pure mathematical point of view this theory seemed to be OK. BUT: In contrast
to QED, the pi-nucleon coupling had to be much to big for a perturbative expansion:

(GπN )2
≃ 15 (233)
4π

Therefore this theory failed to describe pi-nucleon physics.

 35

π N N
N π π
π∗ N∗

N N
N N N

FIG. 2: Some π-Nucleon Feynman diagrams

IX. QUARKS AND PARTONS

A. Multiplets for baryons and mesons

It was shown by Gell-Mann in the beginning of the 1960’s that the pions and kaons
make an octet of pseudo-scalar particles (0− ) within SU(3)-symmetry, as shown in Fig.3.
There were also a similar octet containining the proton, neutron and the strange Σ, Λ and
Ξ-particles. More particles were found, for instance the were spin 1-octets containing the
ρ, K ∗ , ω, φ-particles, and a SU(3) decuplet containing the spin 3/2 particles ∆, Σ∗ , Ξ∗ , Ω− .
All these particles were explaines as built up of the fundamental SU(3)- triplet ()u, d, s.
As more and more baryons and mesons were discovered , some physicists were staring to
think that all these were not elemntary particles, but were composed from “smaller”, more
fundamental particles. In Fig. 3 we see the meson and baryon SU(3) octets. Maybe the
elementary triplet could correspond to fundamental constituents of the baryons and mesons?
Sunch fundamental particles were called quarks or simply partons.. This hypothesis were
from the beginning controversial, and those who “believed” in quarks were sometimes called
“frogs sitting around a pond saying quark, quark,quark “.
The octet baryon symmetry might be obtained by combining three quarks from the SU(3)
triplet in Fig. (3). These might then be the u-quark with charge +2/3 and isospin + 1/2,
the d-quark with charge -1/3 and isospin -1/2, and the s-(strange) quark withc zero charge
-1/3 ad isospin zero. Then the proton should contain two u-quarks and one d-quark, and
the neutron should contain one u-quark and two d-quarks. One important point is that we
have baryons with spin 3/2, Assuming the SU(3) triplet in Fig. 3 build up all the baryons,
there are three cases where baryons are built up from three u-quarks (the ∆++ particle),
three d-quarks(the ∆− particle), or three s-quarks(the Ω− particle), with all their three
quarks having spin plus 1/2 along the same axis. BUT: Three identical particles in the same
 36

K0 K+ p n

u d

π− (π 0 , η8) π+ Σ− (Σ0, Λ) Σ+

s
K− K0 Ξ− Ξ0

−
FIG. 3: Pseudoscalars (0− , left figure) and baryons( 12 )(figure in middle put in octet (8-plets)
representations of SU (3). This was making order i hadronic particles. BUT the elementary triplet
(3-plet) representation (“quarks”) was apparently not realized - or...? NB: I discovered in last
moment that in the central figure, p and n was interchanged. And in the figure to the right u and
d was interchanged.

state should be forbidden according to the Pauli exclusion principle;- ....unless..... there is
another quantum number which is different for the three quarks. This new quantum number
is for some peculiar reason called color. And in order to save the Pauli principle, there
must be at least three different values of this new quantum number. Further if we say that
there are just three different colors, it is natural to choose SU(3) as the color group. This
group is called SU(3) color and written SU(3)c . Then the matematicians can tell us that by
combining three color triplet quarks we may obtain a color singlet baryon state, with other
words, the color singlet baryon has no open color. Mesons can be built from a triplet and
an anti-triplet.

B. An elektron-proton collision and the 1974 revolution

So, is the proton elementary, or does it consist of smaller pieces? To try to find out this
there was from 1969 going on an experiment at SLAC(California) were hard electrons were
shot against nucleons:
What was seen was that new particles could occurr, for instance that an extra ppion was
coming out, i. e. the reaction ep → epπ 0 or ep → enπ + (parton) were occurring. The hard
electron could not shoot loose a single quark. However, the energy from the electron may
create a q − q̄-pair inside the proton. And a q − q̄-pair may come out as a color-less physical
 37

FIG. 4: Elektron-proton collision at high energy (at Stanford Linear Accellerator Center, iCalifor-
nia, in 1969) e p → e N π

e e

γ∗

FIG. 5: Collision ep → eX in the Parton Model. After the collision, quarks, anti-quarks and gluons
will hadronize to make physical, color singlet baryons and mesons

meson. The conclusion was that a quark cannot be free! The “color”-charge is confined !
i.e. bound to a baryon or a meson.
The intepretation of data is the following: A proton consisted of the three valence quarks
uud. At high energies, the proton in addition consists of “sea-quarks” = quarks and anti-
quarks and also gluons. Thus the picture of the proton is three quarks with different colors
combined to a color singlet. The quarks cannot be kicked out of the proton. The energy
from the colloding electron (transmitted via the virtual photon) is used to make a quark
anti-quark pair. Then for a (very) short time the proton contains four quarks and an anti
quark, which combine to one baryon and one meson. The quark anti-quark pair is thought
to be made via the emission of a gluon which again makes the quark anti-quark pair. This
 38

gluon is the analogue of the photon in elecromagnetic interactions, where a photon can make
an electron-positron pair. Thus the picture of the proton is then three valence quaks for
small energien plus some gluons plus some quark anti-quark pairs when the energy grows.
In 1974 came the “1974-revolution”: Then the experimentalists found the socalled Ψ-
particles. The decay spectrum of these particles behaved qualitatively as the decay-spectrum
of positronium e+ e− . The Ψ particles were intepreted as a system of a heavy (for that time)
quark c and its anti-particle c̄. After this, the quark picture was accepted among (almost)
all particle physicists. And, actually there were a good candidate for a field theory for the
color force,quarks and gluons, namely the Yang Mills theory.

X. YANG MILLS THEORY. GAUGE TRANSFORMATIONS.

A. Gauge theory

The Yang-Mills (YM) theory , which is used i modern field theory within the Standard
model were introduced 1954 already. This theory was a generalisation of QED:
1
LQED = ψ γ µ (iDµ − m) ψ − F µν Fµν , (234)
4
where the covariant derivative Dµ is given by

iDµ = i∂µ − e Aµ , (235)

and field tensor Fµν might be written via a commutator between covariant derivatives:

[iDµ , iDν ] = ieFµν , Fµν = ∂µ Aν − ∂ν Aµ . (236)

The gauge group of QED is U(1)em , where the subscript em means “electromagnetic”. The
QED lagrangian is invariont under the following combined transformation for fermion and
gauge field respectively: The gauge transformations are
1
ψ(x) → ψ(x)′ = eiα(x) ψ(x) , Aµ (x) → Aµ (x)′ = Aµ (x) − ∂µ α(x) (237)
e
The mass term is trivially gauge invariant, and so is the field tensor Fµν . But the term
including the covariant derivative is invariant if the object (Dµ ψ) transforms as ψ itself :

Dµ ψ(x) → Dµ′ ψ ′ = eiα(x) Dµ ψ(x) (238)

 39

Yang Mills (YM) theories have the generic form:

1
LY M = f γ µ (iDµ − mf ) f − F a,µν Fµν
a
, (239)
4
where Dµ is the covariant derivative

iDµ ≡ i∂µ − g Aµ , where Aµ = ta Aaµ (240)

Here the ta ’s are the generators of some gauge group. There are then as many gauge fields
as there are generators in the gauge group. Because the ta matrices do not commute among
a
themselves, the field tensor Fµν = Fµν has an extra term - to obtain gauge invariance, and
is given by the commutator of two covariant derivatives:

[i Dµ , i Dν ] = −ig Fµν , where Fµν = ∂µ Aν − ∂ν Aµ + ig[Aν , Aν ] . (241)

or
a
Fµν = ∂µ Aaν − ∂ν Aaµ − g f abc Abν , Acν , with [ta , tb ] = if abc tc . (242)

Now there is an extra term in the field tensor because the generators do not commue. The
fermion muliplet field f has n components (i.e there are n Dirac fields). The generators ta
are n × n-matrices. The gauge group may be for say SU(n), or some orthogonal group. It is
important that the vector bosons interact among themselves! We obtain triple and quartic
vertices when the tensor Fµν is squared in the lagrangian. The tensor may be written

a (0)a
Fµν = Fµν − gf abc Abµ Acν ; Fµν
(0)a
≡ ∂µ Aaν − ∂ν Aaµ (243)

(0)a
where Fµν is the naively expected term. Squaring the full tensor there is e crossed term
the form

g abc (0)a b,µ c,ν

Lint
G = − f Fµν A A
2
g 2 abc ade b c d,µ e,ν
− f f Aµ Aν A A (244)
4

The YM field theory came in some sense “too early”. It was apparently no use for it at
the time it was presented. There were no known massless vector bosons, except the photon.
 40

FIG. 6: Triple boson vertex ∼ g dictated by non-Abelian gauge invariance

FIG. 7: Four boson vertex ∼ g 2 dictated by non-Abelian gauge invariance

The ρ had a mass, and the YM has no mass term for the vectors, which would break gauge
invarians. But at that time it was apparently no use for it. Was Y.-M. -just a mathematical
exercise?
For Yang-Mills theory the corresponding transformation for fermion fields is
a ta
f (x) → f (x)′ = U(x) f (x) ; U(x) = eiα ∈ SU(n) (−for inst.)

where αa (x) is a set of real functions.

The mass term is again trivially gauge invariant But the term including the covariant
derivative is invariant if Dµ f transforms as f itself:

Dµ f (x) → Dµ′ f (x)′ = U(x) Dµ f (x) (245)

Gauge field transformations in the non-Abelian case will contain a rotation among fields:
i
Aµ → A′µ = U Aµ U † + (∂µ U) U † ,
g
One then obtains after a lengthy calculation
′
Fµν → Fµν = (∂µ (A′ν ) − ∂ν (A′µ )) + ig[(A′ν ) , (A′ν )] = ......
a αa (x)
...... = U Fµν U † ; U = U(x) = eit , (246)

such that
1 1 a a,µν
T r(Fµν F µν ) = Fµν F (247)
2 4
is invariant under gauge transformations.
 41

B. Quantum chromo dynamics (QCD)

Assuming that quarks exsist and carry three different colors, one may construct a quark
gluon field theory of Yang-Mills theory type Thus QCD has the form
X
1
LQCD = q γ µ i∂µ − gs ta Aaµ − mq q − F a,µν Fµν
a
(248)
q
4

Note that there is a sum over quark flavors q = u, d, s, ... The field tensor has an extra term

a
Fµν = (∂µ Aaν − ∂ν Aaµ ) + gs f abc Abν Acν (249)

All the six quark fields q (q = u, d, s, c, b, t) have 3 components (i.e 3 Dirac fields). The
ta matrices are generators (3 × 3-matrices) of the gauge group SU(3)c . It is important
that vector bosons interact among themselves! We obtain triple and quartic vertices. The
triple gluon vertex has dramatic consequences for QCD compared to QED. The important
argument for (at least) three colors is : Otherwise the paricle Ω− (sss) cannot be explained!
Remember the Pauli-principle !

gs ta γµ

FIG. 8: The quark gluon coupling(left diagram),Quark-quark scattering (second diagram), and
quark-gluon scattering (third and fourth diagram)

QCD is significantly different from QED due to the triplet coupling. The gauge copling gs
is universal, the same for quark-gluon and triple gluon couplings. Perturbative QCD breaks
down for low energies (say 1-2 GeV). See later.

XI. RENORMALIZATION. COMPARISON OF ASSYMTOTIC BEHAVIOUR

OF QED AND QCD

In quantum filed theories one might say that the particles, leptons and quarks, and
also gauge bosons and their antiparticles are exitations of the fields. In quantum field
 42

theories (QFT) the number of particles are not conserved, only the total energy and the
(electric) charge. Particles might appear, live for a very short time and disappear, as quantum
fluctuations.
In order to understand the color force of QCD better we will compare it with QED.

A. Renormalization of QED. A simplified description

. A single electron or a single photon “alone in the universe” should be stable. That is,
they should not change by the scattering operator S, i.e. that:

S|e− (p)i = |e− (p)i ; S|γ(k)i = |γ(k)i (250)

The scattering operator might be written as

∞
X
S = 1 + ∆S with ∆S = S (n) , (251)
n=1
(n) n
where S ∼ e . This means that for electrons and photons ∆S is expected to have no
effect for particles on the mass-shell, p2 = m2 :

he− (p)|∆S|e− (p)i = 0 ; hγ(k)|∆S|γ(k)i = 0 . (252)

For off shell particles, as part of higher order diagrams, or for bound fermions, the effect
may in general be non-zero. It turns out that eq. (252) is not satisfied. This means that
the theory has to be redefined. Then the quantities in the original Lagrangian will be
intepreted as bare quantities which has to be replaced by renormalized quantities intepreted
as the physical quantities. In the following this renormalization is described to order e2 in a
qualitative way.
The explicite calculations of these corrections (loop diagrams) are found in textbooks,
for instance in the textbook of Peskin and Schroeder (PS).

• Renormalization of the electron mass and electron field

In Fig. 1 the left diagram (just the electron line) corresponds to the bare electron state
or electron propoagator. The lowest order e2 correction to the electron (state or propagator)
is given in the middle. This diagram is called the electron self energy diagram. By direct
computation it gives a result proportional to

δ (4) (p − p′ )u(p′) e2 Σ(p) u(p) , (253)

 43

FIG. 9: Self-energy correction(s) to electron propagator

where u(p) is a Dirac spinor for quantised four momentum p and Σ(p) is a four by four
R
matrix in Dirac space, given as a calculable integral d4 k over photon loop momentum k
(calculated explicitly in PS):
An explicite calculation gives

e2 Λ
−i(e2 Σ(p))γ·p = m = δm ∼ m ln( ) (254)
4π 2 m

The infinite δm is handeled by parametrising the divergence by a cut-off Λ. One should

note that to obtain δm ∼ mR , the cut-off Λ must be huge...
Now there are two problems: First, an explicite calculation shows that the result (253) is
not zero in contrast to (252). Second, δm is even infinite due to a divergent loop integration
for Σ(p) (see PS). The first problem is solved by renormalizing (i.e redefining) the mass by
defining the renormalized mass mR :

mR = m + δm , (255)

where mR is intepreted as the physical mass. Then the mass part of the Lagrangian split in
two parts
Lmass = −mR ψψ + δm ψψ . (256)

The last term ∼ δm is taken as part of the interaction together with the electromagnetic
interaction −ej µ Aµ . Now the extra contribution in diagram 1c (the counterterm) to the
self energy ∼ δm u(p) u(p) substracts the value of diagram 1b to get zero at the mass shell
p2 = m2R .
But mass renormalization is not enough to stabilise the theory. One must also require
that the fermion propagator have the correct pole at p2 = m2R . From the self enery diagram
we obtain the following corrected propagator

S(p)corr = S(p) + S(p) e2 Σ(p) S(p) + O(e4 ) (257)

 44

Now we may write Σ(p) as an expansion in (p2 − m2R ):

]
−iΣ(p) = A + B(p2 − m2R ) + (p2 − m2R )2 Σ(p) (258)

] is finite.
where A and B are momentum independent and divergent, while Σ(p)
Using the expression S(p) = i/(γ · p − m) for the propagator I give for completeness three
equations (259) (260) (261) not given in the lectures. Using the general expansion

1 1 1
= H + ..... (259)
G−H G G

and one finds the corrected propagator

i
S(p)corr = (260)
]
(γ · p − m − e2 A − e2 B(p2 − m2R ) − (p2 − m2R )2 e2 Σ(p))

Here A is proportional to δm such that m + e2 A = mR , and therefore

i
S(p)corr = (261)
]
(1 − e2 B)(γ · p − mR ) − (γ · p − mR )2 e2 Σ(p)

Now, as the propagator is the contraction of ψ and ψ, the factor (1 − e2 B) in the denom-
inator can be swollowed by the field. Thus one defines the renormalized electron fields ψ, ψ
as
ψ → ψRen = (1 + e2 B)−1/2 ψ + O(e4 ) (262)

After the renormalizing procedures have been performed the bare electron propagator
will be replaced by the quantity

S(p)Rcorr = S(p)R (1 − e2R Σc (p)) + O(e4R ) (263)

] and SR contains only renormalized quantities. Also

where Σc (p) = (γ · p − mR ) Σ(p),
S(p)Rcorr has a pole ar γ · p = mR . Note also that we have antisipated renormalization og
the electric charge eR .
The remaining part Σc of diagram 1b for off-shell electrons are nonzero! It turns out that
for a bound electron (which is off-shell)in a Hydrogen atom will give the main contribution
to Lamb-shift. The degenerations of 2S1/2 and 2P1/2 in Hydrogen are now removed (lifted)
by the self-energy contribution for bound (off-shell) electrons. A detailed analysis shows that
this gives the famous Lamb-shift in Hydrogen giving the λ ≃ 21 cm line.
 45

FIG. 10: Correction(s) for photon propagator gives vacuum polarisation

• Vacuum polarisation

In Fig 10 the left diagram, just the wavy line, represents the photon state or photon
propagator. The diagram in the middle of Fig 10 gives a corrections for the photon, usually
called the vacuum polarisation diagram which means that electron-positron pairs appear
and disappears This can be seen as screening of electrix charge!
The contrubution from this diagram can be written as

hγf |∆S|γi i ∼ δ (4) (kf − ki ) (ǫµf )∗ ǫνi e2 Π(k)µν , (264)

where Πµν is the vacuum polarization tensor.

Now current conservation(gauge invariance) protects the photon from getting an extra
mass., ∂µ (ψγ µ ψ) = 0, which implies k µ Π(k)µν = 0, and thereby

e2 Π(k)µν = (kµ kν − k 2 gµν )e2 D(k 2 ) (265)

The quantity D obtained from the loop integration for vacuum polarization in QED can be
written
D(k 2 ) = D(0) + k 2 Πc (k 2 ) (266)

where D(0) is a divergent constant and Πc (k 2 ) is finite. Here a factor (1 + e2 D(0))−1/2 can
be swallowed into the electromagnetic field analogous to (263) .

Aµ → AµR = (1 + e2 D(0))−1/2 Aµ (267)

and the corrected renormalized photon propagator will be

−1 −1
( 2
) → ( 2 )(1 + e2R Πc (k 2 )) (268)
k k

• Renormalization of the vertex (coupling e)

 46

FIG. 11: Vertex correction(s) for electromagnetic interaction

In this case the diagram in the middle of 11

ieγµ → ie(γµ + e2 Λ(p, p′)) where Λ(p, p′ ) = K + Λc (p, p′) (269)

where Λ(p, p′ ) contains a divergent part K which is compensated by the counterterm to

the right in Fig. 11. Going through the renormalization procedure, the divergent part is
swallowed into the charge e such that the renormalized version reads

ieγµ → ieR (γµ + e2R Λc (p, p′ )) (270)

where Λc (p, p′ ) is finite and goes to zero as q = (p − p′ ) goes to zero.

Within non-relativistic quantum mechanics , and also for relativistic equation of Dirac,
the gyromagnetic factor of the electron is ge = 2. Now the diagram in the middle of Fig. 11
will give a finite contribution/correction to the gyromagnetic factor:

αem
ge = 2 + (271)
π

B. Running couplings in QED and QCD

It should be noted that the there are diagrams with self energy interaction on the external
legs and with a vacuum polarization graph on the external photon leg. However, there is
a relation between the vertex correction Λ(p, p′) and the self-energy Σ, called the Ward
identity:
∂Σ(p)
= Λ(p, p) (272)
∂p
Then it turns out that the net renormalization of the charge is due to vacuum polarization
alone.

• Running electromagnetic coupling

 47

In Fig. 12 we display ee scattering to lowest order (left) then the possible fermion loops
(f means a charged particle), and to the right a diagram with a W -bosons on the loop.

e e e e e
f¯ e
W+
γ γ γ γ γ

e f e e
e W− e
e

FIG. 12: Diagrams explaining αem , which grows with energy! αem ≃ 1/129 at energy = MZ .
Alternative: Modification of Coulomb potential

Now the interaction between two electrons con be obtained by the diagrams in Fig. 12

e2R 2 2

µ
∼ 1 + eR Πc (k ) j (1) jµ (2) (273)
k2

where jµ (1, 2) are the dirac currents in momentum space for the electron lines. The effective
measurable fine structure constant at arbitrary k 2 :

em 2 e2R

αef f (k ) = 1 + e2R Πc (k 2 ) + O(e6 ) (274)
4π

and the physical, renormalized fine structure constant, measured at k 2 → 0 is:

em em e2R e2   1
αR = αef f (0) = = 1 + e2 D̂(0) + O(e6 ) ≃ (275)
4π 4π 137

Using cut-off regularization, D̂(0) ∼ ln(Λ/m). The cut-off Λ is hidden in eR ! Previously

this diagram(s) were presented as a correction to the Coulomb potential. Now , in modern
field theory, one usually intepretes this as an energy dependent fine structure. This means
1 1
that αem = 137
for low energy photons and αem > 137
for larger energies - giving energy
dependent αem Adding later also other charged particles, in the loop i.e more charged leptons,
1
quarks and W-bosons make αem ≃ 128
at the energy MZ c2 .
Note that the physical quantities are not dependent on Λ! ; All dependences of Λ are
hidden/included in renormalized physical quantities! QED is said to be renormalizable ,
which means that all divergences can be hidden in a few quantities.
Similar renormalization procedures in electroweak interactions
 48

• . Running of strong coupling

In QCD there is “vacum polarization” due to quark anti-quark pairs. Due to the triplet
gluon coupling there is also“gluonic vacuum polarization”. This diagram has opposite sign
of the corresponding diagram for quark anti-quark pairs! This has dramatic consequence
s
for high energies: QED and QCD has opposite properties at high energies αR (k 2 ) → 0 for
k 2 → ∞. (“assymptotic freedom”), and is of order one for |k| ∼ 1 GeV.
Experiments gives:
s
αR (MZ2 ) ≃ 0.12 . (276)

Using the results from renormalization of QCD, one obtains for lower energies.

s
⇒ αR ((1GeV )2 ) ≃ 0.5 (±10%) and αR
s
((0.7GeV )2 ) ≃ 0.6 to 1.2 (277)

The great uncertainties for lower energies below 1 GeV shows that perturbative QCD breaks
down at small energies! Other methods are needed. The most used method nowadays is
lattice gauge theory.
In Fig. 13 we show qq-scattering to lowest order in perturbative QCD (left) with correc-
tions with a quark loop(middle) and gluonic loop (right).

q
q q q q q
G
q̄
G G G G G

q q q
q q q
q G

FIG. 13: Diagrams explaining αs , which become smaller with growing energy. The quark and the
gluon loops have different signs. The gluon loop [Link] has dramatic consequenses for the
high-energy behavious of perturbative QCD

.
 49

FIG. 14: Qualitative behaviour of effective αem and of effective αs

FIG. 15: Attractive nuclear forces

C. Nuclear forces

Two protons will repel each other electrically. But if they come very close the may attract
each other because the color forces are stronger than the electric ones at very close distance.
Two protons feels the “tail” of quark-gluon-forces, in spite of electric repulsion
Nuclear forces are much stronger than electromagnetic forces
-at short distance
The color forces are so strong that the quarks can not come out of the proton.
The proton (-and the neutron) are color neutral. Now the picture is that the true strong
force is the color force. While the nuclear force is only the tail of the color force, some times
called the deduced color force. The color force binds quarks together to baryons and mesons,
while the nuclear force bind nucleons to varioous nucleuses.

XII. ELECTROWEAK INTERACTIONS

Historically, what was known about weak interactions was the various weak decays. The
weak force was often called weak nuclear force because it was first seen in nuclear physics,
 50

as beta-decay.

A. Weak decays

The most known weak process (decay) is β-decay : (publically known from the C − 14-
test) where a neutron, possibly inside a nucleus, decay to a proton, an electron and an (anti-)
electro-neutrino:
n → p e− νe , (278)

as seen in Fig. 16. This process breaks parity-symmetry. (-as shown by the chinese-american
physicist C.S. Wu, often called Madame Wu)
The charged π’s and K’s could also decay to leptons, as seen in Fig. 17.

π + and K + → µ+ νµ or e+ νe . (279)

Here the decays to the heavier lepton µ is most probable. The pure leptonic decay µ− →
e− νe νµ is another important example. For massive fermions and anti-fermions there are in
total four degrees of freedom. BUT: For massless fermions/anti-fermions there are in total
only two degrees of freedom Then one makes the following choice:
Neutrinos are considered as lefthanded particles (νe )L and antineutrinos as righthanded
antiparticles (νe )R . Then one defines the leptonic charges for the electron, the muon and
their neutrinos:

• Lepton number Le conserved:

Le (e− ) = Le (νe ) = +1 and Le (e+ ) = Le (νe ) = −1

• Lepton number Lµ conserved:

Lµ (µ− ) = Lµ (νµ ) = +1 and Lµ (µ+ ) = Lµ (νµ ) = −1
(Similar for Lτ )

The weak processes above all conserve the lepton numbers. (-until the neutrino oscillations
was seen... many years later)
Consider the process π + → µ+ νµ in the rest frame of π + . Then µ+ and νµ are going back
to back away from the decayed π + . Assume that µ+ goes upwards and νµ goes downwards.
Now, when the neutrino is mass-less it is left-handed and have its spin(helicity) in the
 51

n p
d u

?
e−
W e−
νe
νe

FIG. 16: Fermi-diagram for beta-decay where a neutron decays, left. To the right the modern
microscopic version where a d-quark in the neutron is decaying with W - exchange.

u
µ+ µ− νµ

W
π+ , K + e−
W
νµ
d, s
νe

FIG. 17: Left: Decay of π + or K + to leptons. The decay to e+ νe is also possible but less probable.
Right: Pure leptonic decay of µ

opposite direction compared to its momentum, i.e it has its spin upwards. Let us consider
a parity transformation on the system. Then its momenta are reversed, while its spins are
unchanged because they are pseudo-vectors. Now both the neutrino momentum and its spin
are upwards, and we have apparently a right-handed neutrino which does not exist, which
also illustrates that parity is broken. Let us then go further and perform in addition a charge
conjugation transformation which leaves the momenta and spins unchanged , but replaces
particles ↔ anti − particles. Then we have gotten an anti-neutrino which is right-handed,
showing that the combined CP -symmetry is intact (-in this case. Later we will mention that
also CP -symmetry might be broken). Now CP -symmetry means that the proces π − → µ− νµ
has same probabillity as π + → µ+ νµ .
In the early 1960’s there were no theory for weak interactions like QED for the electro-
magnetic interactions. Fermi wrote down an effective interaction for four fermion fields
action in a point. Little was from the beginning known of the interaction except for its
 52

strength, the Fermi constant

10−5
GF ≃ , (280)
m2p
where mp is the proton mass. Later, in 1958 Feynman and Gell-Mann found that the
interaction could be written as a product of two left-handed currents
GF 1 V

LF = 4 √ j(W, N)µ j(W, l)µ , j(W )µ = jµ − jµA , (281)
2 2
i.e a vector current minus an axial current: In beta-decay the nucleonic (N) and an leptonic
(l) currents can be written
1 1
j(W, N)µ = ψp γ µ (1 − γ5 )ψn , and j(W, l)µ = ψe γ µ (1 − γ5 )ψνe , (282)
2 2
respectively. The ψ’s are the various particle fields in a self-explained way. The currents
can be written in terms of the left and right projectors PL,R in Dirac space:
1
j(W )µ = ψγµ PL ψ ; PL,R ≡ (1 ∓ γ5 ) . (283)
2
(Later I have also used PL = L and PR = R ). Here

(PL )2 = PL , (PR )2 = PR , PL · PR = PR · PL = 0 , PL + PR = 1 . (284)

A Dirac field can be split in two parts:

ψ = ψL + ψR , ψL = PL ψ , ψR = PR ψ , (285)

where ψL is a left spinning (left screw) particle, and ψR is a right spinning (right screw)
particle. In this way parity violation is built into the theory. Note that the free Dirac
Lagrangian with zero mass can be written

ψγµ i∂ µ ψ = ψL γµ i∂ µ ψL + ψR γµ i∂ µ ψR . (286)

Note also that if the neutrino νe is purely lefthanded, then the current for νe → e− is
automatically lefthanded:

j(W, e)µ = ψe γµ ψνe = ψe γµ (ψνe )L = ψe γµ PL ψνe . (287)

Note that ψL = (PL ψ)† γ0 = ψ PR , such that:

ψL γµ ψL = ψ γµ PL ψ , ψR γµ ψR = ψ γµ PR ψ , ψ ψ = ψL ψR + ψR ψL . (288)
 53

From the last equation we observe that mass terms mixes left- and right-handed fields. We
also observe that symbolically, the squared left-handed current has the form

(V − A)2 = (V V + A A) − (V A + A V ) , (289)

where the last term violates parity

If the two currents in β decay were mediated by a heavy weak boson W coupling with
strength gW , then one would expect that

GF g2
− 4√ → 2 W 2 . (290)
2 q − MW

If the four momentum q is small compared to MW ; i.e |q 2 | << MW

2
, then

GF g2
4 √ ≃ W2 . (291)
2 MW

There existed an hypothes already in the 1930’s that weak interactions were mediated by a
massive boson W . But at that time the W was very hypothetic, and how massive it should
be was not known. (Even its spin was also not clear). But if one assumed that gW were of
the same order of magnitude as the electric charge e, one could deduce from the measured
value of GF that the mass of the W would be of the order 40 proton masses, which turned
out to be correct up to a factor 2.

XIII. SU (2)L × U (1)Y SYMMETRY

In 1967 S. Weinberg published a paper with the title “A model of leptons”. This was a
model of weak interactions for leptons which was a gauge theory with gauge group SU(2)L ×
U(1)Y for two lepton families. At roughly the same time S. Glashhow and A. Salam had
similar ideas.
The W -bosons should be vectors (have spin one). One tried to adapt weak interactions
to a gauge theory;- although gauge boson fields were a priori massless. The simplest
gauge theory is QED where electrons interact via photons. Now, the gauge boson sector
should in addition contain W-and Z-bosons. Yang Mills theory was already known, and
could be used to build an electroweak theory, step by step. A model with gauge group
SU(2)L for two doublets , one lepton doublet EL and one quark doublet QL were considered:
 54

   
νe uL
EL =   and QL =   , (292)
−
e dL
L L

where νe ,e, u and d now means the Dirac fields of the corresponding particles. Inspired by the
concepts of isospin and hypercharge within strong interactions (remember the SU(3) flavor
3
octets), we say that the upper components νe and uL have weak isospin up (IW = +1/2)
3
and the lower components eL and dL have weak isospin down (IW = −1/2). Note: There
are many objects in physics related to the group SU(2), for instance spin. The index L just
means we are dealing with left-handed fermion doublets. The mathematics is the same for
all SU(2) groups. The Yang Mills Lagrangian for SU(2)L based on (292) is
1 k,µν k
LEW 1L = EL γ µ iDµL EL + QL γ µ iDµL QL − W Wµν , (293)
4
where the covariant derivative is given by
τk
iDµL = i∂µ − g( )Wµk . (294)
2
There are three generators 12 τ k and thereby three gauge fields for SU(2)L , namely the gauge
bosons Wµk , k = 1, 2, 3. The genarators 12 τ k = IW
k
are also called the weak isospin operators.
The gauge field tensor is given by:

k
Wµν = ∂µ Wνk − ∂µ Wµk − g ǫkln Wµl Wνn , (295)

where ǫkln is the usual Levi-Citita tensor. Now, the kinetic part of the Lagrangian LEW 1L
can be rewritten

LEW 1KL = EL γµ i∂ µ EL + QL γµ i∂ µ QL − g j kµ Wµk , (296)

where the weak isospin currents jµk coupling to Wµk are given by
1 1
jµk = EL γµ ( τ k )EL + QL γµ ( τ k )QL . (297)
2 2
Note that the weak isospin generators τ k /2 commute completely with all Dirac matrices
(they work in different spaces).
Now we form the current components

jµ(±) = (jµ1 ± i jµ2 ) = EL γµ τ (±) EL + QL γµ τ (±) QL . (298)

 55

Using explicitly the projector matrices τ ± = (τ 1 ± iτ 2 )/2, we obtain

jµ(+) = νe γµ eL + uL γµ dL , and jµ(−) = eL γµ νe + dL γµ uL , (299)

which are just the currents needed to describe beta-decay with currents mediated by W ± .
The third component of the current is
1 1

jµ3 = (νe γµ νe − eL γµ eL ) + u L γµ u L − dL γµ dL . (300)
2 2
We understand that W 3 cannot be the photon because jµ3 has wrong structure.
Now, the free kinetic Dirac Lagrangian is given in (286). Thus the right handed parts
have in some way to be added. We add for the electron the right handed SU(2)L singlet
term eR γ · i∂eR . And we extend the derivative operator i∂µ to be a covariant derivative
containing a new U(1)Y gauge field Bµ . We do this also for the righht-handed parts of the
u- and d quarks. Thus we obtain an additional Lagrangian term:
µ µ µ 1 µν
LEW 1R = eR γµ iDR eR + uR γµ iDR uR + dR γµ iDR dR − B Bµν , (301)
4
where the covariant derivatives for the fermions fR = eR , uR , dR are given by
1
iDµR = i∂µ − g ′ Y (fR ) Bµ . (302)
2
The U(1)Y field tensor is given by

Bµν = ∂µ Bν − ∂ν Bµ , (303)

similar to the electromagnetic field. Here the Y (fR )’s are just numbers called the weak
hypercharge. From the mathematical point of view, U(1)Y is as U(1)em for electric charge.
Therefore the Y ’s are not restricted, in contrast to non-abelian theories. For instance our
fermions in one generation have the charges 0, −1, +2/3, and −1/3. The Y (fR )’s are deter-
mined by the requirement that we get the correct electromagnetic currents in the end. The
weak left-handed doublets EL and QL are also assigned a weak hypercharge such that the
covariant derivative for left-handed doublets in (294) is extended to be
1 1
iDµL = i∂µ − g ( τ k )Wµk − g ′ YL Bµ , (304)
2 2
for YL = Y (EL ) and YL = Y (QL ). The weak hypercharge current coupling to the U(1)Y
field Bµ is then:



(jµY ) = Y (EL ) EL γµ EL + Y (QL ) QL γµ QL


+ Y (eR ) (eR γµ eR ) + Y (uR ) (uR γµ uR ) + Y (dR ) dR γµ dR . (305)
 56

All particles up to here are apparently massless in order to have gauge invariance ! The
answer to this “mass problem” is the Higgs mechanism, to be presented below.
The right-handed e and u and d-quark fields are SU(2)L singlets. Without a right-handed
field for the the neutrino, they are mass-less (-which is anyway a good approximation).
The masses of the W ± and Z-bosons will- after some assumptions about the Higgs field
- come from a term LHiggs to be presented below. The fermion masses are coming from
LY ukawa ,- also to be presented soon.
Later we will see that a physical intepretation can be obtained in a spesific gauge !

XIV. THE HIGGS MECHANISM

In 1964, Higgs (and others) demonstrated how to introduce masses in gauge theories
without breaking gauge invariance. One could introduce a complex scalar field or a multiplet
or more scalars (multiplets) and assume that some part of it has a nonzero value in vacuum
(a Vacuum expextation value, called VEV). This could give a mass to (some of) the gauge
bosons.

A. A simple example

Consider a complex Klein Gordon field φ with an additional φ4 interaction term which
can be written
Lφ4 = (i∂α φ)† (i∂ α φ) − V (φ) , (306)

where


2
V (φ) ≡ µ2 φ† φ + λ φ† φ . (307)

Apparently this field has mass µ. However, if we take the derivative of V (φ) with respect
to |φ| , we find
∂V

= 2|φ| µ2 + 2λ|φ|2 , (308)
∂|φ|
which means that if µ2 is assumed to be negative, the potential V has a minimum at
r
v −µ2
|φ| = √ , where v = . (309)
2 λ
 57

Then the fields can be reparametrised

1
φ = √ (v + σ + i η) . (310)
2
This is called Spontanous Symmetry Breaking (SSB). The U(1) symmetry of the origial
Lagrangian is not there when written in terms of σ and η. In terms of these fields the
Lagrangian will be
1 1
(∂α σ)(∂ α σ) + (∂α η)(∂ α η) − V (φ)SSB ;
L =
2 2
1 2 2 1 2 2 λ 4 1 λ
V (φ)SSB = mσ σ + mη η + v + aσ + λvσ 3 + λvσ 2 η + λσ 2 η 2 + (σ 4 + η 4(311)
),
2 2 4 2 4
where
m2σ = µ2 + 3λv 2 = −2µ2 , m2η = µ2 + λv 2 = 0 , a = vµ2 + λv 3 = 0 , (312)
p
which means that we have gotten one particle σ with mass mσ = −2µ2 , and one particle
η with mass mη = 0, which is a Goldstone boson. It is very important that the linear term
in σ cancels because such linear terms cannot be intepreted in QFT. Note that PS uses a
priori opposite sign for µ2 . We may now do a new reparametrisation
1 η′
φ = √ (v + σ ′ ) ei v . (313)
2
In addition we extend the derivative in (306) with a covariant derivative

LDφ4 = (iDα φ)† (iD α φ) − V (φ) , (314)

where iDα = i∂α − eAα as in QED. Without SSB, this equation would be QED for a
charged scalar particle. Assuming SSB, and making a gauge transformation with η ′ /v as
gauge parameter we have
η′ 1 1
φ → φ′ = e−i v φ = √ (v + σ ′ ) ; Aα → A′α = Aα − ∂α η . (315)
2 ev
A massless vector field has two spin degrees of freedom;- helicity ±1. Now the transformed
vector field A′α has gotten an extra spin zero part ∼ ∂α η and become massive. For the
transformed covariant dervative we have (using ∂α v = 0)
1
iDα φ → iDα′ φ′ = (ı∂α − eA′α )φ′ = = √ (i∂α σ ′ − eA′α (v + σ ′ )) . (316)
2
Then we find:
1 σ′
L′Dφ4 = (∂α σ ′ )(∂ α σ ′ ) + (ev)2 (Aα Aα )′ (1 + )2 − V (φ′ )SSB , (317)
2 v
 58

where
µ2 λ
V (φ′ )SSB = (v + σ ′ )2 + (v + σ ′ )4 . (318)
2 4
Now the Goldstone boson η is “swallowed” by the gauge transformation which gives the
transformed field A′µ a mass
mA′ = ev . (319)

B. SSB for SU (2)L × U (1)Y

For the SU(2)L × U(1)Y case one introduces a complex SU(2) doublet. The Lagrangian
for the Higgs doublet φ is :

†

LHiggs = iDα(φ) φ α
iD(φ) φ − V (φ) , (320)

where


2
V (φ) ≡ µ2 φ† φ + λ φ† φ , (321)

as before, but now we have a complex doublet, i.e. four real fields. Then the complex Higgs
doublet interact with gauge bosons through covariant derivatives:
 
 k ′

φ+ τ g
φ =   ; (iDα )φ = i∂α − g Wα − Y (φ) Bα φ .
(φ) k
(322)
φ0 2 2

For µ2 > 0 , we have a theory for spin zero particles with mass µ, interacting with massless
vector fields. And the minimum of V (φ) is at φ = 0 , or h0|φ|0i = 0 for the quantum case.
Again, if µ is assumed to be imaginary, i.e. µ2 < 0, and at the same time λ > 0, the
potential V has minimum for a value |φ| =
6 0, and SSB will occurr. This means that the
vacuum has not the full symmetry of the dynamical equations, and the vacuum value of the
field φ may be taken to be
r  
v −µ2 0
h0|φ|0i = √ χV ; v ≡ ; χV ≡   , (323)
2 λ 1

where v is the vacuum value measured in GeV.

With SSB, a triplet ξ j (x) of massless neutral (Goldstone) boson fields and one massive
p
neutral Higgs field H with mass mH = −2µ2 will appear.
 59

We now do a reparametrisation, and write

1
φ = Uξ−1 χV √ (v + H) where Uξ−1 = exp(iξ k τ k /2v) · exp(iξ 3 Y (φ)/2v) . (324)
2
Now the Goldstone triplet fields ξ i will be used as gauge parameters and transformed
into three linear combinations (W (±) , Z) of Wµk and Bµ to make them massive, while the
photon remains massless. This very special SU(2)L × U(1)Y gauge transformation contains
the Goldstone fields ξ j via Uξ :

1
φ → φ′ = Uξ φ = √ (v + H) χV . (325)
2
This is the “Physical gauge”, and H is the physical Higgs boson field. The gauge where
Goldstone bosons are transformed away, and particles have their physical masses are called
unitary gauge. After Higgs-mechanism has been applied, all transformed fields, vector and
fermion, depend on the three Goldstone fields ξ k .
(±)
The transformed gauge fields combine to the physical bosons Wµ , Zµ , Aµ in the following
way:

1
Wµ± = √ (Wµ1) ∓ i Wµ2 )′ , Zµ = (cW Wµ3 + sW Bµ )′ , Aµ = (−sW Wµ3 + cW Bµ )′ , (326)
2
where the “primes” (’) denotes the transformed fields. Here sW ≡ sinθW and cW ≡ cosθW ,
where θW is the weak mixing angle, which has to be determined by experiment. One sees
that the gauge bosons (Wµk , Bµ ) having apriori two degrees of freedom get an additional
longitudinal (spin 0) term ∼ ∂µ ξ k and thereby obtains a mass.
The transformed covariant derivative is a two by two matrix, and acting on the Higgs
doublet it may be written:
   

′ A C C (v + H)
iDα(φ) φ =   √1 (v + H) χV = √1   . (327)
C† K 2 2 K (v + H)

Here
1 g′
C = −gW Wα(−) , and K = i∂α + (g Wα3 − Y (φ) Bα)′ , (328)
2 2
√
where the gauge fields are the transformed ones, and gW = g/ 2. The quantity A also,
as K, contains the derivative, but another linear combination of (Wα3 )′ and Bα′ . But A is
projected out because of the zero in χV . Plugging in the expressions for (the transformed)
 60

Wα3 and Bα fields we obtain

1
K(v + H) = i∂α H + (gcW + g ′Y (φ)sW )) (v + H)Zα
2
1
+ (gsW − g ′ Y (φ)cW )(v + H)Aα . (329)
2
Now , the generic form of the mass term for a vector field Vµ with mass mV is (for neutral
fields there is an extra factor 1/2 at the right hand side):

LVM ass = m2V Vµ† V µ . (330)

To avoid a mass term for the photon we therefore require

gsW − g ′Y (φ) cW = 0 (331)

The coefficient for the Zα term will then be gZ (v + H)/2 , where

g
gZ ≡ . (332)
cos(θW )
The transformed covariant derivative can now be written
 
(−)

′ 1 −gW Wα (v + H)
iDα(φ) φ = √   . (333)
2 1
i∂α + gZ Zα (v + H)
2

The Uξ transformed Lagrangian is :

1 µ2 λ
L′φ = (∂α H)(∂ α H) − (v + H)2 − (v + H)4
2 2 4
 g 2  g 2
Z
.+ Wα(+) W (−)α (v + H)2 + Zα Z α (v + H)2 (334)
2 2
(φ)
N B! With the choice made in (331), the linear combination for the γ-field Aµ in (Dα )’
has disappeared , which means that the photon remains mass-less. The mass of the Higgs
particle is
p p
mH = µ2 + 3λv 2 = −2µ2 . (335)

Now , from (330) and (334) we can read off the masses of the W and Z-bosons:
1 1
MW = g v , MZ = gZ v , (336)
2 2
and furthermore, from these equations we predict an important relation between the W and
Z masses:
MZ 1
= , (337)
MW cosθW
which is confirmed by experiment (-up to higher order corrections).
 61

C. Yukawa interactions for fermion masses

Mass terms for fermions should also appear in the correct way, namely as

∼ m(ψ R ψL + ψ L ψR ) . (338)

The way to obtain this is to write down a Yukawa type Lagrangian for interactions
between the electron and the Higgs field:

(e)

LY ukawa = − Ge EL φ eR + eR φ† EL . (339)

In addition to SU(2)L this Lagrangian term should also be U(1)Y invariant. This means
that when we multiply the fermion and Higgs fields with a phase factor of the type
i
exp( g ′ Y · θ) . (340)
2
Here θ is an arbitrary real parameter. The Yukawa-term in (339) term must be invariant.
Then we must have the relation

−Y (EL ) + Y (φ) + Y (eR ) = 0. (341)

Plugging in the transformation (325) we obtain

(e) Ge
(LY ukawa )′ = − √ (v + H) (eR ′ e′L + eL ′ e′R ) . (342)
2
To obtain a mass for the d quark we introduce

(d)

LY ukawa = − Gd QL φ dR + dR φ† QL , (343)

with the relation

−Y (QL ) + Y (φ) + Y (dR ) = 0 . (344)

In order to get fermion masses for those with weak isospin +1/2, we need to use the adjoint
(charge conjugated) field φe given by:
   
†
φ0 v
φe ≡ φc = iτ 2 (φ† )T =  e 0 = √1   .
 ; hφi (345)
† 2
− φ+ 0

Then the u-quark gets its mass from the Yukawa interaction:
 
LY ukawa = − Gu QL φe uR + uR φe† QL ,
(u)
(346)
 62

with

e + Y (uR ) = 0
− Y (QL ) + Y (φ) (347)

e = − Y (φ). After transformation with Uξ one obtains the generic form

Note that here Y (φ)
Gf
L′Y uk f H = − √ (v + H) f ′ f ′ . (348)
2
Thus the result for fermion masses is
Gf
mf = √ v , (349)
2
All the Gf ’s are determined by experimental values of fermion masses. It follows that the
Higgs boson coupling to a fermion f is
mf
gf H = g. (350)
2MW
Therefore the Higgs particle couple most strongly to heavy ferions ! This is a consequance
of the gauge invariant Yukawas.
We may perfectly well add a righthanded neutrino νR in the SM in the same way as the
rest of the fermions! :
 
(ν ) (1) (1) (1)
e
LY ukawa = − Gν e EL φe eR + νR φe† EL ; EL = ((νe )L , eL ) . (351)

If the ν’s are Majorana particles, i.e. their own antiparticles, things are more complicated,
and we do not go deeper into this question in this course.
The Higgs -boson was discovered in 2012 at LHC at CERN i Geneve, with a mass ≃ 125
GeV/c2 . It could be produced as “gluon fusion” as shown at the Figure 18. A proton at
high energy contains gluons. When two protons collide, a gluon from one of the protons
might collide with a gluon from the other, i.e. two gluons might collide. Then a Higgs
particle may appear as a fusion from two gluons and make a Higgs boson. But this Higgs
boson is very unstable and may decay into two photons. In both diagrams it happens when
the Higgs couples to a top quark, which has the strongest coupling to the Higgs-boson .

D. Currents

The un-transformed Lagrangian for currents interacting with gauge bosons is

g′ Y µ
Lint
f EW = − g jµk W kµ − j B , (352)
2 µ
 63

g
t
H
t

g t

γ
t
H
t

t γ

FIG. 18: Feynman diagram for Higgs production through gluon fusion. And decay of Higgs to two
photons via a triangle loop with a top quark.

where jµk is given in (297) and jµY in (305). After the SU(2)L × U(1)Y transformation Uξ
has been applied one obtains the Lagrangian in the “physical gauge” : Using the relations
between the physical vector bosons Aµ , Zµ and the transformed (Wµ3 )′ and Bµ′ given in (326),
the interaction Lagrangian takes the form

′
Lint
f EW = − gγ jµem Aµ − gZ jµZ Z µ


− gW jµ(+) W (−)µ + jµ(−) W (+)µ . (353)

The physical weak and electromagnetic currents are known ! These will now be in-
tepreted in the transformed (by Uξ ) version, and we know that in the end, for the first
generation/family the electromagnetic current must be:
X
jµem = qf f γµ f = − (e γµ e)′
f
2 1
+ (u γµ u)′ − (d γµ d)′ . (354)
3 3
The weak currents are already known in (299), but now in the transformed version

jµ(+) = (νe γµ L e)′ + (u γµ L d)′ , (355)

jµ(−) = (e γµ L νe )′ + (d γµ L u)′ ; (356)

where L = (1 − γ5 )/2, and the fermion fields are understood to be the transformed, weak
eigenstates fW = fL′ + fR′ .
 64

Now
1 1
′
(jµ3 )′ = (νe γµ νe − eL γµ eL )′ + u L γµ u L − dL γµ dL , (357)
2 2
where the fermion fields are understood to be the Uξ transformed ones, fW = fL′ + fR′ .
The hypercharge current jµY has several terms. For the first generation it is

′
(jµY )′ = Y (EL ) (νe γµ νe + eL γµ eL )′ + Y (QL ) uL γµ uL + dL γµ dL

′
+ Y (eR ) (eR γµ eR )′ + Y (uR ) (uR γµ uR )′ + Y (dR ) dR γµ dR . (358)

Now we obtain the coefficient of Aµ in (353):

1
gγ jµem = (νe γµ νe )′ (g sW + g ′ Y (EL )cW ) +
2
1 1
(eL γµ eL )′ (−g sW + g ′Y (EL )cW ) + (eR γµ eR )′ (g ′Y (eR )cW ) +
2 2
1 1
(uL γµ uL )′ (+g sW + g ′Y (QL )cW ) + (uR γµ uR )′ (g ′ Y (uR )cW )
2 2
1 1
+ (dL γµ dL )′ (−g sW + g ′ Y (QL )cW ) + (uR γµ uR )′ (g ′Y (dR )cW ) (359)
2 2
Now, in order to get the correct electromagnetic current terms as in (354), we obtain the
requirements:
(g sW + g ′ Y (EL )cW ) = 0 , (360)
1
(−g sW + g ′ Y (EL )cW ) = − gγ , (361)
2
1 ′
(g Y (eR )cW ) = − gγ , (362)
2
1 2
(g sW + g ′Y (QL )cW ) = gγ , (363)
2 3
1 ′ 2
(g Y (uR )cW ) = gγ , (364)
2 3
1 1
(−g sW + g ′ Y (QL )cW ) = − gγ , (365)
2 3
1 ′ 1
(g Y (dR )cW ) = − gγ . (366)
2 3
Now we choose (cfr. PS) Y (φ) = 1. Then the requirements to obtain the correct electro-
magnetic current can be collected in the formula:

3 1
qf = (IW )f + Y (f ) , (367)
2
where qf is the fermion charge. (Note that PS has replaced Y /2 by just Y .) This is also
in agreement with the the relations for weak hyper-charges obtained from the Lagrangian
 65

f W W−
e e e
γ, Z γ, Z
νe

e f¯ e
e W W+

FIG. 19: Examples: e+ e− → f f¯ and two diagrams for e+ e− → W + W −

terms of the various Yukawa interactions. (Remember that the right-handed singlets have
3
weak isospin (IW )fR = 0). It is also the analogue to the formula we have for the SU(3)
flavor octets in hadronic physics. And it is consistent consistent with
 ′
em 3 1 Y
jµ = jµ + jµ , (368)
2
which will lead to (367) when integrated over space (cfr. charge conservation via Noethers
theorem)
The couplings for the physical bosons γ, Z, W ± are
g g g′
|e| ≡ gγ = g sinθW , gZ = , gW = √ , tanθW = (369)
cosθW 2 g
The neutral current jµZ is now completely determined to be:

jµZ = (jµ3 )′ − sin2 θW jµem . (370)

Experimentally sin2 θW ≃ 0.23. From αem and GF , one finds v ≃ 246 GeV. We have

obtained a partial unification of electromagnetic and weak interactions: The three couplings
gγ , gW , gZ has become two independent couplings!
The Feynman rules for fermions coupling to the electroweak bosons γ W ± and Z are

3
gγ qf γµ , gW γ µ L , gZ γµ (IW L − s2W qf ) . (371)

There is also a tripple boson vertex obtained from the Lagrange term
1

LW 3 = − gǫkln ∂µ Wνk − ∂ν Wµk W lµ W lν , (372)
2
which gives a vertex (interaction) with (W 1 W 2 W 3 )′ . Here (the transformed) W 1 and W 2
can be combined to W + and W − , and (W 3 )′ will be a linear combination of A(γ) and Z.
Thus (372) can be converted to γW W and ZW W vertices (see PS p.802, eq. (A.11) for
details).
 66

Many people think of the Higgs potential V (φ) to be analogous to a magnet which
has a rotation symmetric interaction Hamiltonian for the individual magnets. For high
temperatures, the individual magnets are randomly oriented, but for low temperatures there
is a ground state where individual magnets are aligned. So maybe the Higgs potential had
a symmetrical shape in the early(earlier) universe (?) .

E. Quark mixing

The first sign of quark mixing was the beta-decay of the strange particle Σ0 , when the
process Σ0 → p e− νe was seen. The explanation made by the italian physicist Nicola Cabibbo
was that there was a mixing between the d-quark and the s-quark, such that the partner of
the u-quark was not solely the d-quark, but the mixture

dθC = d cos(θC ) + s sin(θC ) , (373)

where θC is called the Cabibbo angle. Note that θC is not related to the weak mixing angle
θW . Experimentally sin(θC ) ≃ 0.22. Later, when more quarks had been observed, the idea
of Cabibbo was generalized by Kobayashi and Maskawa, and for three families, three lepton
doublets and three quark doublets we have (still not considering mixing of neutrinos):
           
νe νµ ντ u c t
  ,  ,    ,  ,  , (374)
− − −
e µ τ dCKM sCKM bCKM
L L L L L L

-plus all the right-handed singlets. Now all the quarks with charge -1/3 mix, and we have
the CKM (Cabibbo-Kobayashi-Maskawa)- mixing of quarks

dCKM = Vud d + Vus s + Vub b ,

sCKM = Vcd d + Vcs s + Vcb b ,
bCKM = Vtd d + Vts s + Vtb b , (375)

where (d, s, b) are the physical quarks (with definite mass), and Vij ’s are Kobayashi-Maskawa
matrix elements. How this mixing are forced to occurr in electroweak interactions we will
show below.
The point now is the following: The mass part of the quark sector has non-diagonal
terms like u¯′L c′R (charge 2/3), d¯′L b′R (charge -1/3). Namely, there is no symmetry which
 67

can prevent such mass-like terms with the same electric charge. The Lagrangian generating
fermion masses will be more general than shown in the subsection about Yukawa terms:
X

LM ix
Y ukawa = − CFLfR FL φ fR + fR φ† FL
FL ,fR
X  i
+ e f
C F L fR FL φe ff f e†
R + fR φ FL . (376)
FL ,ff
R

Note: There are two right-handed singlets (fR , ff

R ) with different (hyper-) charges per left-

handed doublet (FL = EL , QL ) The eq. (376) implies

e = Y (FL ) − Y (ff
Y (φ) = Y (FL ) − Y (fR ) ; Y (φ) R) . (377)

Note that there are summed over all 6 left-handed doublets FL and all 9 right-handed singlets
(or 12 if neutrinos are included). But there are of course no mixing between quarks and
leptons (This would have been violation of baryon number and an unstable proton, say)
Now the quark mass part can be written in a very compact notation:

Lmass (q)′ = − UW MU UW − DW MD DW ; (378)

   
u d
 W   W 
   
UW ≡  cW  ; D W ≡  sW  , (379)
   
tW bW

where the subscript W means weak eigenstate, which is the same as the primed fields.
The entrys of the non-dagonal mass matrices MU and MD are proportional to v and the
coefficients of various “non-diagonal”Yukawa terms in (376).
Physical quark states are mass eigenstates: UP , DP , and have diagonal mass matrices.
The non-diagonal mass matrices are diagonalised by a similarity transformations SU and
SD :

Mdiag
Q
†
= SQ M Q SQ ; QP = SQ QW (380)

for Q = U and Q = D, respectively. The subscript P means “physical”.

(±)
Then we obtain the weak currents jµ (q) for three generation of quarks:

jµ(+) (q) = UW γµ L DW = UP γµ L DCKM , (381)

 68

and

†
jµ(−) (q) = DW γµ L UW = DP γµ L VCKM , UP (382)

where

†
DCKM ≡ VCKM DP ; UCKM ≡ VCKM UP . (383)

Here VCKM ≡ SU SD† is the Kobayashi-Maskawa quark mixing matrix. This matrix appear
in the charged weak currents because of mismatch of diagonalization (up vs. down sector),
technically seen as SU 6= SD , which implies VCKM 6= 1 (Later we might drop subscript P ).
The quark mixing matrix is a unitary matrix with three real and one imaginary parameter;
- to be determined experimentally. It has been chosen to attach the mixing matrix to the
D-quark sector as in (374) and (375). If one considers mixing in the lepton sector, it has
been chosen to attach it to neutrinos.
†
For the neutral currents jµem and jµZ the quark mixing disappears because SQ SQ = 1
leads (for Q = U, D) to QW Γ QW = QP Γ QP , where Γ is some (product of) Dirac matrices.
The neutral currents in quark sector can then be written ( remembering eq. (370) ):

2 1
jµem (q) = U γµ U − D γµ D , (384)
3 3
1

jµ3 (q) = U γµ L U − D γµ LD . (385)
2

These expressions are valid both for QP and QW because the neutral currents are diagonal
in the quark fields. Flavor changing neutral currents (FCNC) only appear at loop level (For
example as b → sγ in the hadronic process B → γK ∗ , as shown in Fig. 22 below. )

F. Phenomenologial consequenses of quark mixing

Quark mixing is responsible for the many decays of exotic particles, mesons as well as
baryons. Without quark mixing, flavor change could only occurr within the same quark
doublet. This means only the transitions u ↔ d, c ↔ s and t ↔ b would occurr. Thus the
b-quark and the s-quark would have been stable because they have heavier partners. With
quark mixing the “charges” like “bottomness”, “charm” and “strangeness”associated with
b, c and s-quarks are not conserved in weak interactions. This is illustrated in Fig. 20.


69

b c b u

W W

u q c q

FIG. 20: Basic W-exchange diagrams generating various non-leptonic decays of B-mesons. (Here
q = d or s). Some quark lines have to be bended, and a “spectator quark” line has to be added in
order to obtain a diagram for a physical process. Topologically similar to the right-hand parts of
the diagrams in Fig. 1 and Fig.2

s (u, c, t) d s W µ− s
(u, c, t) µ−
W W (u, c, t) W
νµ
Z
(ū, c̄, t̄) s̄ d¯ µ+ d¯ µ+
d¯ W

FIG. 21: Quark diagram for K 0 → K 0 determining K 0 − K 0 mixing and CP violation in neutral
K-meson decays. (left diagram), It was found from this diagram that mc < 2 GeV . Middle and
right: Diagrams for K 0 → µ+ µ− , which goes diferent from the charged decay K + → µ+ νµ in
Fig.2. When u, c, t quarks are going in loops there are a partial cancellation between the three
contributions, which is due to the unitarity of the CKM-matrix.( called GIM-mechanism).

The complex entrys of VCKM means that there are possibillities of CP -violating effects,
which may show up in flavor changing decays triggered by W -exchange.
As the quarks are sitting inside hadrons, for instance in mesons, the K, D, and B-mesons
have short lifetimes. CP -violation effects are seen in some of such decays. BUT: CP -
violating effects in the SM are too small to explain early universe cosmology. (Unfortunately
it has not been time to discuss CP -violation in the lectures). Decays of kaons (K-mesons)
have played an important role in understanding of the SM.

A diagram to illustrate K 0 → π + π − -or 2π 0 . CP -violation is different for charged and

 70

u, c, t
b s

B q W K∗

FIG. 22: The quark decay b → sγ decay at one loop level “sitting” inside the process B → K γ.
The lower q line is called spectator quark because it doesnt take part in the electroweak interaction,
but is bound to B or K ∗ by non-perturbative QCD.

W
d
s ū, d¯

(u, c, t)
???
G

d¯ u, d

d¯

FIG. 23: Example: The “Penguin diagram”. Perturbative weak and strong interactions in play in
a triangle loop. And there is a non-perturbative QCD interaction symbolized by the ??-sign

neutral pions The non-perturbative QCD part is difficult.(To be handeled by Lattice gauge
theory).
As said previously, there will also be mixing in the leptonic sector, implying that the
lepton flavors Le , Lµ , Lτ are not separately conserved. This leads to neutrino oscillations,
which have been observed experimentally. It also means that processes like µ± → e± γ is
not forbidden. If this mixing has the same origin as for quarks or if it as another like for
istance of Majorana type is not clear.

XV. SUMMING UP ELECTROWEAK INTERACTIONS

The electroweak interactions can be formulated as a complete symmetric SU(2)L ×

U(1)Y gauge theory. Gauge-symmetric theories are chosen because: Gauge theories are
renormalizable-and have a minimum of parameters. Gauge invariance implies relations be-
tween couplings and masses.
First one writes down a completely gauge symmetric Lagrangian. The free parameters
 71

are :
e in (376) .
g , g ′ , µ2 , λ , all the hypercharges Y , all the C, C (386)

After SSB: The U(1)em symmetry is still visible. The free parameters in electroweak
interactions,- after the SSB (and Higgs-mechanism) are : gγ , θW , all 12 fermion masses
mf , the four independent parameters of quark mixing matrix VCKM . There is also a mixing
p
of neutrinos, up to now of unknown origin. The Higgs mass MH = −2µ2 and Higgs self-
interaction coupling λ were a priori free parameters (to be determined by experiment). But
p
their ratio is fixed by v = −µ2 /λ, experimentally known from value of Fermi’s coupling
constant GF . Knowing the Higgs mass experimentally to be ≃ 125 GeV/c2 , the value of λ
is fixed to lowest order.
All other couplings of bosons to fermions, triple and quartic couplings of vector bosons,
and Higgs couplings to all fermions and bosons are fixed by gauge symmetry, and obtained
from the parameters above! Note that the Higgs self-interactions ∼ λ are also fixed (to
lowest order)when the Higgs mass is experimentally known.

XVI. PATH INTEGRALS

Here I have followed the book (PS), as far as I went. I started to define the generating
functional Z[J] for the KG equation, and defined the functional derivatives which gave the
propagator after two such derivatives. Then I wrote down the corresponding expression
R
DAeiS[A] , and went through the gauge fixing leading to the photon propagator

F −i  2

Dµν = g µν − (1 − ξ)k k
µ ν /k . (387)
k 2 + iǫ

Then I defined the Grassman variables and how to integrate over them. I defines the
generating functional with sources for the Dirac equation. At the end I considered the
Fadeev-Popov ghost for QCD as an additional term ∼ ca ∂ µ Dµab cb .

NB !
For Path integrals, and for renormalization, only qualitative knowledge of the subjects
are required at the final oral exam.