N ATIONAL I NSTITUTE OF S CIENCE E DUCATION

AND R ESEARCH , B HUBANESWAR

S IXTH S EMESTER (P398) P ROJECT R EPORT

Classical Field Theory & Relativistic

Quantum Mechanics

Author: Supervisor:
Joydeep N ASKAR Dr. Yogesh Kumar
S RIVASTAVA

A report submitted in fulfillment of the requirements

for the the course P398 Physics Project
in the

School of Physical Sciences

April 20, 2019

ii

“The Mathematician plays a game in which he himself invents the rules while the Physicist
plays a game in which the rules are provided by Nature. But as time goes on it becomes
increasingly evident that the rules which the Mathematician finds interesting are the same
as those which Nature has chosen.”

Paul A.M. Dirac

 iii

NATIONAL INSTITUTE OF SCIENCE EDUCATION AND RESEARCH,

BHUBANESWAR

Abstract
Sixth Semester(P398) Physics Project

Classical Field Theory & Relativistic Quantum Mechanics

by Joydeep N ASKAR

This is the Sixth Semester P398 Physics Project report which was a reading project.
In this report, I have discussed Relativistic Quantum Mechanics in the first chapter.
It covers the Klein-Gordon equation and then moves on to Dirac Equation and its
derivation through the Lorentz Group. The algebra of spinors is covered and an
example of electron magnetic moment is provided. In the second chapter, concepts
of Classical Field Theory is discussed. I started with Real Scalar Fields and traversed
through Complex Scalar and Electromagnetic Fields. Yang-Mills Field is studied and
the chapter is concluded with some geometric insights of Gauge Fields.
 v

Acknowledgements
I would like to thank my project supervisor Dr. Yogesh Kumar Srivastava, whose
expertise, understanding, generous guidance and support made it possible to work
on the topic of my interest. It was a pleasure working with him.
 vii

Contents

Abstract iii

Acknowledgements v

1 Relativistic Quantum Mechanics 1

1.1 Klein-Gordon Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Dirac Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2.1 Rotation Group . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
SO(3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
SU(2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2.2 SL(2,C) & Lorentz Group . . . . . . . . . . . . . . . . . . . . . . 5
1.2.3 Derivation of Dirac Equation . . . . . . . . . . . . . . . . . . . . 6
1.3 Tackling the Problems of Klein-Gordon Equation . . . . . . . . . . . . . 7
1.4 Dirac Spinors: Construction & Example . . . . . . . . . . . . . . . . . . 9
1.4.1 Standard Representation of Spinors . . . . . . . . . . . . . . . . 9
1.4.2 Example: Electron Magnetic Moment . . . . . . . . . . . . . . . 11

2 Introduction to Continuous Systems and Fields 13

2.1 Transition from a Discrete to Continuous System . . . . . . . . . . . . . 13
2.2 Lagrangian Forumlation for Continuous Systems . . . . . . . . . . . . . 14
2.2.1 Real Scalar Field . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Hamiltonian Forumlation . . . . . . . . . . . . . . . . . . . . . . 15
2.2.2 Complex Scalar Fields . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2.3 Complex Scalar Fields & Electromagnetic Field . . . . . . . . . . 16
2.2.4 Yang-Mills Field . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.3 The Geometry of Gauge Fields . . . . . . . . . . . . . . . . . . . . . . . 19

A Derivations 23
A.1 A0µ = SAS−1 − gi (∂µ S)S−1 . . . . . . . . . . . . . . . . . . . . . . . . . . 23
A.2 Relation between φR (p) and φL (p) . . . . . . . . . . . . . . . . . . . . . 23

B Derivation of Euler-Lagrange Equation from Variational Principle 25

C Spinor Algebra: Bilinear Quantities 27

 1

Chapter 1

Relativistic Quantum Mechanics

This chapter introduces relativistic quantum mechanics. It starts with the Klein-
Gordon equation and the associated problems in brief. It then moves to Dirac Equa-
tion and describes it in little detail.

1.1 Klein-Gordon Equation

From special relativity the 4-momentum of a particle is given by pµ = ( E/c, p). It
gives an invariant
E2
pµ pµ = 2 − p.p = m2 c2
c
Let’s put E = ih̄ ∂t∂ , p = −ih̄∇. For a particle with one component φ,

1 ∂2 2 m2 c2
( − ∇ ) φ + φ=0
c2 ∂t2 h̄2
In units of h̄ = 1, c = 1, we have

(2 + m2 ) φ = 0

This is the Klein-Gordon equation.

p2
In the non-relativistic limit, E = 2m where E is kinetic energy, it yields the free-
particle Schrodinger equation

h̄2 2 ∂φ
∇ φ = −ih̄
2m ∂t
In one way, we can say that Schrodinger’s equation is a non-relativistic approxima-
tion of Klein-Gordon equation. The probability density of Schrondinger equation
is
ρ = φ∗ φ
and the probability current is

ih̄ ∗
j=− (φ ∇φ − φ∇φ∗ )
2m
They obey continuity equation

∂ρ
+∇·j = 0
∂t
In case of Klein-Gordon equation, in 4-vector form we have
2 Chapter 1. Relativistic Quantum Mechanics

ih̄ ∗ ←→ ← → ih̄ ←→
jµ = (ρ, j) = φ ( ∂0 , ∇ ) φ = φ ∗ ∂ µ φ
m m
←→
where A ∂µ B = 21 [ A(∂µ B) − (∂µ ) B]. So the continuity equation becomes

ih̄ ∗
∂µ jµ = (φ 2φ − φ2φ∗ ) = 0
2m
Now if we look closely at
ih ∗ ∂φ ∂φ∗
ρ= (φ −φ )
2m ∂t ∂t
The density is not positive definite. It may assume negative values. So its interpre-
tation as probability density has to be abandoned, which means, the interpretation
of Klein-Gordon equation as single-particle equation, with wave-function φ also has
to be abandoned.
Another problem is that solutions to Klein-Gordon equation contain both positive
energy and negative energy (E = ±(m2 c4 + p2 c2 )1/2 ). For an interacting particle,
there exists a possibility of cascading down to negatively infinite energies.

1.2 Dirac Equation

We discussed the Klein- Gordon equation in the previous section and the problems
associated with it.
The Dirac equation attempts to solve those problems, which we will discuss here.
The Dirac equation can be derived from the transformation properties of spinors
under the Lorentz group.

1.2.1 Rotation Group

SO(3)
A general rotation is of the form:
 0  
x x
 y0  = R  y 
z0 z

or r’=Rr
Rotations preserve distance from origin. So

r0T r0 = r T r
r T R T Rr = r T r
RT R = 1
R is an orthogonal 3X3 matrix. These matrices form a continuous, non-abelian
group of unit determinant referred as SO(3). It is a Lie group.
For an active clockwise rotation by an angle θ in the x-y plane(about z-axis), the
rotation matrix is denoted by
 
cos θ sin θ 0
Rz (θ ) = − sin θ cos θ 0
0 0 1
1.2. Dirac Equation 3

Similarly, we can define rotation matrices for rotations about x & y axes. We
can define the generators. For example, the generator matrix of the rotation matrix
defined above is
 
0 −i 0
1 dRz (θ ) 

Jz = =  i 0 0
i dθ θ =0

0 0 0
The generators are Hermitian and infinitesimal rotations are given by

Rz (δθ ) = 1 + iJz δθ

The generators follow the commutation relations

Jx Jy − Jy Jx = [ Jx , Jy ] = iJz

and cyclic permutations. Upto second order,

Rz (δθ ) R x (δφ) R− 1 −1
z ( δθ ) R x ( δφ ) = 1 − iJy ( δθδφ )

→ ∞) in the x-y
The matrix for a finite rotation through an angle θ = Nδθ ( N −
plane (about z-axis) is

θ N
Rz (θ ) = [ Rz (δθ )] N = (1 + iJz δθ ) N = (1 + iJz
) = eiJz θ
N
A finite rotation about an axis n through an angle θ is denoted

Rn (θ ) = eiJ.θ = eiJ.nθ

SU(2)
Consider the group SU(2), consisting of 2X2 unitary matrices with unit determinant,
i.e, UU † = 1 , det U = 1
 
a b
U=
−b∗ a∗
,
| a |2 + | b |2 = 1
 
ξ1
Consider a spnior ξ = in 2-D complex space and the transformation
ξ2

ξ −→ Uξ, ξ † −→ ξ † U †

If we look closely at the transformation ξ 0 = Uξ

ξ 10 = aξ 1 + bξ 2

ξ 20 = −b∗ ξ 1 + a∗ ξ 2
And the transformation ξ † 0 = ξ † U †

ξ ∗01 = a∗ ξ 1∗ + b∗ ξ 2∗

ξ ∗02 = −b(ξ 1∗ ) + aξ 2∗
4 Chapter 1. Relativistic Quantum Mechanics

 
0 −1
Let us define a matrix τ = . While ξ and ξ † do not transform in the same
1 0
way. However, (Let ’∼’ means transforms like.), it is observed that
 ∗
∗ −ξ 2
, ξ † ∼ (τξ )T = −ξ 2 ξ 1


ξ ∼ τξ = ∗
ξ1

Hence
−ξ 1 ξ 2 ξ 12
   
† ξ1

ξξ ∼ −ξ 2 ξ 1 =
ξ2 ξ 22 ξ1ξ2
Let’s call this matrix − H. Under SU(2) transformation H −→ UHU †
We can construct a matrix h from the position vector r and the Pauli matrices σ.
 
z x − iy
h = σ·r =
x + iy −z

h is Hermitian and the transformation

h −→ UhU † = h’

preserves Hermiticity and tracelessness of h if U is unitary. If U belongs to SU(2),

then det h’=det h, i.e,
x 02 + y 02 + z 02 = x 2 + y 2 + z 2
. Hence the transformation on h induces a rotation of the position vector r. Identify-
ing h and H, we can conclucde.  
  x
ξ1
An SU(2) transformation on ≡ SO(3) transformation on y 
ξ2
z
1 2 2 1 2 2
with x = 2 (ξ 1 − ξ 2 ), x = 2i (ξ 1 + ξ 2 ), z = ξ 1 ξ 2 .
Now we put a = eiα/2 , b = 0, we get
 
 iα/2  cos α sin α 0
e 0
U= − ←→ − sin α cos α 0
0 e iα/2
0 0 1

In general, the correspondence between an SU(2)

 transformation in spinor space
  x
ξ1
and an SO(3) transformation in the space  y  is
ξ2
z

θ θ
U = eiσ·θ/2 = cos + iσ.n sin ←→ eiJ.θ
2 2
σy
The Pauli matrices obey the relation [ σ2x , 2 ] = i σ2z and cyclic permutations, simi-
lar to the algebra of J’s. So we identify
σ
J=
2
This homomorphism between SU(2) and SO(3) implies that they have a similar
structure and thier generators must follow the same algebra. It is easily verified that
the Pauli matrices follow the same algebra.
Note that: There is a two-to-one mapping of the elements of SU(2) onto those of SO(3)
1.2. Dirac Equation 5

1.2.2 SL(2,C) & Lorentz Group

Analogous to the correspondence between SU(2) and the rotation group, there is
correspondence between SL(2,C) and the Lorentz group.
Pure boost Lorentz transformations connect two inertial frames, moving with a
relative speed v. If the relative motion is along the common x axis, then x 0 = x+2vt1/2 ,
(1− v2 )
c
t+vx/c2
y0 = y, z0 = z, t0 = 2 1/2 . Put γ =
1
2 1/2 , β = v/c, x0 = ct, x1 = x and so
(1− v2 ) (1− v2 )
c c
on. Define cosh φ = γ and sinh φ = βγ where tanh φ = β The pure boost Lorentz
transformation is given by boost matrix:
 
cosh φ sinh φ 0 0
 sinh φ cosh φ 0 0
Bx ( θ ) = 
 0

0 1 0
0 0 0 1
The generator K of the boost this boost is:
 
0 1 0 0
1 dBx (φ)  1 0 0 0
Kz = = −i 
0

i dφ φ=0 0 0 0

0 0 0 0
In this 4 X 4 matrix notation, the generator of rotation about z-axis looks like
 
0 0 0 0
0 0 1 0
Jz = −i 
0 −1

0 0
0 0 0 0

The most general Lorentz transformation is a boost along three directions and
rotations in three planes(about three axes) and the six generators follow the commu-
tations relations (along with their cyclic permutations)

[Kx , Ky ] = −iJz
[ Jx , Jy ] = iJz
[ Jx , Ky ] = ik z
[ Jx , K x ] = 0
An interesting observation is that pure boost transformations do not form a group.
To relate boost transfromations to Pauli matrices, we identify
σ
K = ±i
2
There exists two types of spinor, for the two different values of K. Let us now define:

1
A= ( J + iK )
2
1
B= ( J − iK )
2
6 Chapter 1. Relativistic Quantum Mechanics

The commutation relations become (with cyclic permutations)

[ A x , Ay ] = iAz

[ Bx , By ] = iBz
[ Ai , Bj ] = 0 (i, j = x, y, z)
A and B each generate a SU(2) group and they commute. The Lorents group is
SU(2) ⊗ SU(2) transforming in a well-defined way labelled by two angular momenta
0
(j,j ), the first one corresponding to A and the second to B. As special cases, one or
the other will correspond to spin zero.

( j, 0) −→ J ( j) = iK ( j) ( B = 0)

(0, j) −→ J ( j) = −iK ( j) ( A = 0)
Thus we may define two types of spinor:
Type I: ( 12 , 0) : J (1/2) = σ/2, K (1/2) = −iσ/2. Let us denote the spinor ξ.
Let (θ, φ) be the parameters of rotation and boost, then ξ transforms as
σ
ξ −→ exp[i .(θ − iφ)]ξ = Mξ
2

Type II: (0, 12 ) : J (1/2) = σ/2, K (1/2) = iσ/2.

Let us denote the spinor η. η transforms as
σ
η −→ exp[i .(θ + iφ)]ξ = Nξ
2
Note that they are inequivalent representations. But if we include the parity
operation, then velocity in boost changes from v to -v. So K changes to -K, like a
polar vector. But J does not change sign, behaving like axial vector. So under parity,
the two representations (j,0) and (0,j) are just interchanged. They can no longer be
treated separately as one can
 be transformed into other by operation of parity. Thus
ξ
we define a 4-spinor ψ = . The 4-spinor ψ is an irreducible representation of
η
Lorentz group extended by parity.

1.2.3 Derivation of Dirac Equation

Let’s relabel the previously defined 2-spinors ξ & η as φR & φL respectively. For a
pure boost, we have

φR −→ eσ.φ/2 φR = cosh φ/2 + σ.n sinh φ/2

where n is a unit vector in the direction of Lorentz boost.

From now onwards, we will be working in c=1 units, unless explicitly stated.
Let’s say the original spinor refers to a particle at rest, φR (0), and the transformed
one to with a particle with momentum p, φR ( p). We have cosh φ/2 = [(γ + 1)/2]1/2
and sinh φ/2 = [(γ − 1)/2]1/2 (Note that this comes using from hyperbolic trigono-
metric identities and the definition of cosh φ.) So,

γ + 1 1/2 γ − 1 1/2
φR ( p) = [( ) + σ. p̂( ) ] φR (0)
2 2
1.3. Tackling the Problems of Klein-Gordon Equation 7

For a particle with energy E, mass m and momentum p, γ = E/m,

E + m + σ.p
φR ( p ) = φR (0)
[2m( E + m)]1/2
Similarly,
E + m − σ.p
φL ( p ) = φ L (0)
[2m( E + m)]1/2
Since originally the particle was at rest, φR (0) = φL (0). It can be shown that ( I
have done the proof in Appendix A.2):

E + σ.p
φR ( p ) = φL ( p )
m
E − σ.p
φL ( p ) = φR ( p )
m
These equations can be written in matrix form as:
  
−m p0 + σ.p φR ( p )
=0
p0 − σ.p −m φL ( p )
   
φR ( p ) 0 1
Define 4-spinor ψ( p) = and 4 × 4 matrices γ0 = and γi =
φL ( p ) 1 0
0 −σi
 
. So we get
σi 0
( γ0 p0 + γ i p i − m ) ψ ( p ) = 0
Since pµ = ( E, −p),
(γµ pµ − m)ψ( p) = 0
This is the Dirac equation for massive spin 1/2 particles.
For massless particles, we have:

( p0 + σ.p)φL ( p) = 0

( p0 − σ.p)φR ( p) = 0
These are called Weyl equations. Since for massless particles, p0 = |p|, so

σ. p̂φL = −φL , σ. p̂φR = φR

The operator σ. p̂ measures the hellicity (component of spin in the direction of
momentum). Thus φR and φL are eigenstates of hellicity.

1.3 Tackling the Problems of Klein-Gordon Equation

The Klein-Gordon equation suffered from two faults: first with the interpretation of
negative probability density and the second was that of negative energy.
Dirac equations removes one of them trivially and the other predicts the existence of
antiparticles. Before we proceed to tackle those problems using Dirac equation, in
co-ordinate space, Dirac equation can be written as:

(iγµ ∂µ − m)ψ = 0
8 Chapter 1. Relativistic Quantum Mechanics

This is a first order equation. Applying the operator iγµ ∂µ again, we get

[−(γµ ∂µ )(γν ∂ν ) − i (γµ ∂µ )m]ψ = 0

( γ µ γ ν ∂ µ ∂ ν + m2 ) ψ = 0
Now ∂µ ∂ν = ∂ν ∂µ , so
1 µ ν
{ γ , γ } ∂ µ ∂ ν ψ + m2 ψ = 0
2
where curly braces mean anti-commutator.
To satisfy energy-momentum-mass relation, each component of ψ satisfies the Klein-
Gordon equation. Therefore comparing the coefficients of ∂µ ∂ν , we arrive at

{γµ , γν } = 2gµν

.
Now, let us construct a probablity current jµ to check if the probability density is
positive. So we have
(iγµ ∂µ − m)ψ = 0
Taking Hermititan conjugate of this equation [Note: γ0† = γ0 but γi† = −γi ],

ψ† (−iγ0 ∂0 + iγi ∂i − m) = 0

operating on the left, becomes

ψ(γµ ∂µ + m) = 0

where ψ = ψ† γ0 is called the adjoint spinor. We define the current jµ = ψγµ ψ.

We see that current is conserved.

∂µ jµ = (∂µ ψ)γµ ψ + ψγµ (∂µ ψ) = (imψ)ψ + ψ(−imψ) = 0

The density j0 is
j0 = ψγ0 ψ = ψ† ψ > 0
j0 is thus fit to serve as the probability density of the particle in question. So it solves
the problem of negative probability. [However note that j0 is not invariant.]
Now the second problem of negative energy persists. Let’s take the example of
Dirac particle at rest.

γ0 p0 ψ = mψ
p0 ψ = mγ0 ψ
The eigenvalues of γ0 are +1(twice) and -1(twice). So there are 2 positive energy
solutions(+m) and 2 negative energy solutions(-m). From the Dirac equation, the
eigenvalues of E are:
E = +(m2 + p2 )1/2 twice
E = −(m2 + p2 )1/2 twice
There are two positive energy solutions and there two negative energy solutions.
Dirac’s solution to this problem stands on the fact that the particle concerned is a
half-spin particle and obeys Pauli’s exclusion principle. Dirac’s assumption is that
1.4. Dirac Spinors: Construction & Example 9

all the negative energy states are completely filled, and the exclusion principle pre-
vents any more half-spin particle from entering the "sea" of negative energy states.
It is important to note here that having assumed the above logic, the applicability
of Dirac’s equation is limited to Fermions only. The logic of exclusion principle fails
with Bosons.

1.4 Dirac Spinors: Construction & Example

We shall study
 the transformation properties of ψ and other bilinear expressions. We
φR
had ψ = . A general Lorentz transformation would give:
φL

i i
φR −→ exp[ σ.(θ − iφ)]φR , φL −→ exp[ σ.(θ + iφ)]φL
2 2
Similarly,

−i −i
φR† −→ φR† exp[ σ.(θ + iφ)], φL† −→ φL† exp[ σ.(θ − iφ)]
2 2
It is clear that
ψ † ψ = φR † φR + φ L † φ L
is not invariant under Lorentz transformation. However, it is invariant under pure
rotation.  †
φL
The adjoint spinor ψ = . Before we proceed, define a matrix γ5 = iγ0 γ1 γ2 γ3 .
φR †
 
5 1 0
γ =
0 −1

Now we will construct a few quantities and state how they behave upon trans-
formation (Refer to Appendix C for more details):
ψψ = φL † φR + φR † φL −→ scalar
ψγ5 ψ = φL † φR − φR † φL −→ pesudoscalar
ψγµ ψ = (φL † φL + φR † φR , −φL † σφL + φR † σφR ) −→ vector
ψγµ γ0 ψ = (φR † φR − φL † φL , φL † σφL + φR † σφR ) −→ axial vector
ψ(γµ γν − γν γµ )ψ −→ anti-symmetric tensor

1.4.1 Standard Representation of Spinors

Lets now play with the spinors. Consider an arbitrary state of motion of a Dirac
particle. But here will no longer with the chiral representation where γ5 is diagonal
but in a representation with γ0 diagonal. This is called the standard representation.
For a particle at rest, the plane wave solutions are:
ψ( x ) = u(0)e−imt −→ positive energy
ψ( x ) = v(0)eimt −→ negative energy
10 Chapter 1. Relativistic Quantum Mechanics

with two positive and negative energy spinors

       
1 0 0 0
0  u (2) ( 0 ) =   v (1) ( 0 ) =   v (2) ( 0 ) =  0 
1 0
u (1) ( 0 ) = 
       
0 0 1 0
0 0 0 1

with γ0 in the standard representation

 
1 0 0 0
0 1 0 0 
γ0 = 


0 0 −1 0 
0 0 0 −1

In the condensed form written as

 
0 1 0
γ =
0 −1
Note that this is related to the chiral representation
  via a transformation matrix S
0 = Sγ0 S−1 where S = 1 1 1
such that γSR CR 2 1 −1 . In the standard representation,


  
φR −1 1 φR + φ L
ψ=S S = √
φL 2 φR − φ L
Consider now a Lorentz boost to a moving frame. We had previously
 0 1
!   
φR 1 e 2 σ.φ 0 φR φ
0e 2 σ.φ
= =M R
φL 0 0
−1
e2 σ.φ φ L φL

In the standard representation, the boost matrix is

 
−1 cosh φ/2 σ.n sinh φ/2
MSR = SMCR S =
σ.n sinh φ/2 cosh φ/2
−m 1/2
+m 1/2
We remember that cosh(φ/2) = ( E2m ) ,sinh(φ/2) = ( E2m ) tanh(φ/2) =
p
E+m . Hence,

p x −ipy
 
pz
1 0 E+m E+m 
p x +ipy − pz 
E + m 1/2 

0 1 E+m E+m 
MSR =( )  pz p x −ipy
2m 
 E+m E+m 1 0 

p x +ipy − pz
E+m E+m 0 1

The corresponding plane wave spinors are

ψ(α) = u(α) ( p)e−ip.x positive energy
1.4. Dirac Spinors: Construction & Example 11

ψ(α) = v(α) ( p)eip.x negative energy

where α = 1, 2 and the transformed spinors are given below. Let p± = p x ± ipy .

0
   
1
E + m 1/2  0  E + m 1/2  1 
u (1) =( )   z ,
p  u (2) =( )  p 
 E+−m 
2m E+m 2m
p+ − pz
E+m E+m

pz 
  p− 
E+m E+m
p+  − pz 
E + m 1/2  E + m 1/2 
v (1) = ( )  E+m  ,
 1  v (2) = ( )  E+m 
2m 2m  0 
0 1
Normalisation and some other properties of the spinors are:
0
u(α) ( p)u(α ) ( p) = δαα0
0
v(α) ( p)v(α ) ( p) = δαα0
0
u(α) ( p ) v(α ) ( p ) = 0
0 0 E
u(α)+ ( p)u(α ) ( p) = v(α)+ ( p)v(α ) ( p) = m δαα
0

1.4.2 Example: Electron Magnetic Moment

Particles with spin carry an intrinsic magnetic moment. Taking charge of electron as
-e, the electron magnetic moment arising due to its spin is given by1

e e eh̄
µ=2 S= S= σ
2m m 2m
In the presence of an electromagnetic potential Aµ = (φ, A), the Dirac equation is
given by
γ0 ( E − eφ)ψ − γ.(p − eA)ψ = mψ
      
u 0 σ u u
( E − eφ) − (p − eA). =m
−v −σ 0 v v
Explicitly writing the two equations:

( E − eφ)u − .(p − eA)v = mu

−( E − eφ)v + .(p − eA)u = mv

From the second equation we get,

v = ( E + m − eφ)−1 .(p − eA)u

. In the non-relativistic and weak field limit, E + m − eφ ≈ 2m and p = mv p where

v
v p is veloctiy of particle. We have β = cp . So

1
v= σ.(p − eA)u ∼ o ( β)u
2
.
In the non-relativistic limit, β << 1. So we see that the top two components of ψ is
1 This material is covered in P305
12 Chapter 1. Relativistic Quantum Mechanics

much larger than the bottom two.

Coming back to problem in question, we get

σ.(p − eA)σ.(p − eA)

Eu = + mu + eφu
2m
Separating the mass term from expression of energy, let E = m + W. We have

σ.(p − eA)σ.(p − eA)

Wu = + eφu
2m
From the algebra of σ matrices, we get

σ.(p − eA)σ.(p − eA) = (p − eA).(p − eA) + iσ.(p − eA) × (p − eA)

Now we have
(p − eA) × (p − eA) = p × A + A × p
Using operator equation2 , we have [ pi , A j ] = −ih̄i A j . Interchanging i and j and
taking a difference

( pi A j − p j Ai ) + ( Ai p j − A j pi ) = −ih̄(∂i A j − ∂ j Ai )

Multiplying by eijk and summing over all components, we get3

p × A + A × p = −ih̄∇ × A = −ih̄B

Finally, in the non-relativistic limit, substituting Hu=Wu, we get the Hamiltonian

H

1 eh̄
(p − eA)2 + eφ −
H= σ.B
2m 2m
The first two terms correspond to the Classical Hamiltonian of an electron in a
EM field. The last term is the interaction energy of magnetic moment with the mag-
netic field. Hence the Dirac equation predicts the correct electron magnetic moment,
with gs = 2.

References
The text has been mostly referred from Lewis H. Ryder, Quantum Field Theory, Sec-
ond Edition. Some portion on Dirac Equation has been borrowed from J.J. Sakurai,
Advanced Quantum Mechanics, Pearson. Rotation groups and the two-to-one homo-
morphism between SU(2) and SO(3) are referred from J.J. Sakurai and Jim J. Napoli-
tano, Modern Quantum Mechanics, Second Edition, Pearson. The electron spin magnetic
moment is referred from B.H. Bransden, C.J. Joachain, Physics of Atoms and Molecules,
Second Edition, Pearson. The Hamiltonian of a charged particle in an electromagnetic
field and corresponding changes in momentum and energy is referred from Gold-
stein, Poole, Safko, Classical Mechanics, Third Edition, Pearson.

2 This material is studied in P303

3 This material has been covered in P301
 13

Chapter 2

Introduction to Continuous
Systems and Fields

2.1 Transition from a Discrete to Continuous System

Having learnt the Lagrangian and Hamiltonian formalism for Discrete particles in
P201, we move on to describe the physics of continuous systems under the same
formalism.
Consider an infinitely long elastic rod composed of discrete particles of equal
mass m capable of undergoing small oscillations longitudinally through an imagi-
nary massless spring. If the displacement of the ith particle is ηi , then the Lagrangian
is given by:
1
L = T − V = ∑[mη̇i 2 − k (ηi+1 − ηi )2 ]
2 i
where T & V are kinetic and potential energies, respectively. The equation can ve
re-written as
1 m ( ηi + 1 − ηi ) 2
L = ∑ a[ η̇i 2 − ka ] = ∑ Li
2 i a a2 i

The resulting equation of motion becomes

m ( ηi + 1 − ηi ) (η − η )
η̈i − ka 2
+ ka i 2 i−1 = 0
a a a
In going from discrete to continuous case, the integer index i identifying the partic-
ular mass point becomes the continuous position coordinate x, instead of discrete
variable ηi we have a continuous field variable ηx . Furthermore,

( ηi + 1 − ηi ) (ηx+a − ηx )
=
a a
dη
approaches the limit (as a approaches zero) dx . The Lagrangian appears as

1 dη 2
Z
L= [µη̇ 2 − Y ( ) ]dx
2 dx
where µ = ma is the mass per unit length and Y = ka is identified as the Young’s
modulus of the rod.
The Lagrangian density L becomes

1 dη dη
L = [ µ ( )2 − Y ( )2 ]
2 dt dx
14 Chapter 2. Introduction to Continuous Systems and Fields

In case of 3-D, Z Z Z
L= Ldxdydx

2.2 Lagrangian Forumlation for Continuous Systems

In this section we shall study how the Lagrangian formalism can be applied to var-
ious types of fields, the corresponding Euler-Lagrange equations and the conserved
quantities.

2.2.1 Real Scalar Field

Consider a field φ( x µ ) and Lagrangian density

1 m2 2
L(φ, ∂µ φ) = (∂µ φ)(∂µ φ) − φ
2 2
The Euler Lagrange equation for φ is given by (proved in Appendix B)

∂L ∂ ∂L
− µ[ ]=0
∂φ ∂x ∂(∂µ φ)

So we have,
∂L ∂L
= −m2 φ, = ∂µ φ
∂φ ∂(∂µ φ)
The Euler-Lagrange equation gives

∂ µ ∂ µ φ + m2 φ ≡ (2 + m2 ) φ = 0

which is the Klein-Gordon equation. We define action as

Z
S= L(φ, ∂µ φ)d4 x

We define energy-momentum tensor as

µ ∂L µ
θν = ∂ν φ − δν L
∂(∂µ φ)
Now suppose the action S is invariant under a certain type of transformations on
x µ and φ, whose infinitesimal forms are:

∆x µ = Xν δω ν , ∆φ = Φµ δω µ
µ

characterised by parameter δω ν , where Xν is a matrix and Φ is a set of numbers. On

µ

demanding δS = 0, we get

∂L
Z
Φν − θκ Xνκ ]δω ν dσµ = 0
µ
[
∂R ∂(∂µ φ)

over some boundary ∂R of a region in R.

Since δω µ is arbitrary, Z
µ
Jν dσµ = 0
∂R
2.2. Lagrangian Forumlation for Continuous Systems 15

where
∂L
Φν − θκ Xνκ
µ µ
Jν =
∂(∂µ φ)
µ
∂µ Jν d4 x = 0 and since R is arbitrary,
R
From Gauss’s Theorem we have R
µ
∂µ Jν = 0
µ
Thus we have a conserved (divergenceless) current Jν which gives rise to a con-
served charge (time independent) Qν defined as
Z
µ
Qν = Jν dσµ
σ

where the integral is taken over a spacelike hypersurface σµ . At a frozen time t,

Z
Qν = Jν0 d3 x
ν

where the integral is taken over a 3-volume V. By conscervation of Qν , integrating

over V: Z Z
∂0 Jν0 d3 x + ∂i Jνi d3 x = 0
V V
By Gauss’s theorem, taking suitable far-away surface, the second term vanishes leav-
ing behind
d dQν
Z
Jν0 d3 x = =0
dt dt
This is Noether’s theorem.
Conservation of energy and momentum, holds for any system whose Lagrangian
does not depend on x µ .
Putting the scalar field Lagrangian in the energy-momentum tensor, we arrive at

θ µν = (∂µ φ)(∂ν φ) − gµν L

This equation is symmetric in µ and ν. Hence for a scalar field, energy-momentum

tensor is symmetirc. But it is not in general true. Since it is not unqiue either. We
may add term ∂λ f λµν where f λµν = − f µλν , so that ∂µ ∂λ f λµν = 0. Hence, we define
the canonical energy-momentum tensor

T µν = θ µν + ∂λ f λµν

We have
∂µ T µν = ∂µ θ µν = 0
We can always choose f λµν such that T µν is symmetric.

Hamiltonian Forumlation
One can obtain the Hamiltonian formulation for fields. Conjugate to each field com-
ponent ηi canonical momentum density πi = ∂∂η̇Li . Construct the Hamiltonian density
H as
H = π η̇ − L
The Hamiltonian formulation suppresses the parallel treatment of time and space
variables.
16 Chapter 2. Introduction to Continuous Systems and Fields

2.2.2 Complex Scalar Fields

Any complex field is described by two independent components, the complex field
(φ) itself and its complex conjugate (φ∗ ). Another choice is the real part and imagi-
nary part but here we stick with the former. Consider the Lagrangian density

1 m2 ∗
L(φ, ∂µ φ, φ∗ , ∂µ φ∗ ) = (∂µ φ)(∂µ φ∗ ) − φ φ
2 2
The Euler- Lagrangian equations are given by

(2 + m2 )φ = 0, (2 + m2 ) φ ∗ = 0

For a constant Λ the Lagrangian is invariant under the Gauge transformation of

first kind, i.e,
φ −→ e−iΛ φ, φ∗ −→ eiΛ φ∗
It’s infinitesimal form is
δφ = −iφ, δφ∗ = iφ∗
And,
δ(∂µ φ) = −iΛ∂µ φ, δ(∂µ φ∗ ) = iΛ∂µ φ∗
The conserved current is
∂L ∂L
Jµ = (−iφ) + (iφ∗ ) = i (φ∗ ∂µ φ − φ∂µ φ∗ )
∂(∂µ φ) ∂(∂µ φ∗ )

The 4-divergence of vanishes on boundary, i.e, ∂µ J µ = 0. The conserved quantity is

∂φ∗
Z Z
∂φ
Q= 0
J dV = i (φ∗ −φ )dV
∂t ∂t
If we identify this quantity with electric charge, we need to note the following
observations on Q:

1. It is conserved, i.e, dQ/dt = 0.

2. It is continuous and not quantized in steps of e, and is a classical quantity.

3. When φ is real, Q = 0.

Here Q emerged as a conserved quantity from the invariance of the Lagrangian

under the gauge transformation of first kind. But such a gauge transformation is
a global gauge transformation i.e, performing the transformation at one point, the
same transformation happens at all other points at the same time. In spirit, it violates
special relativity.

2.2.3 Complex Scalar Fields & Electromagnetic Field

We will now look at a local gauge transformation, also called gauge transformation
of second kind, which differs from point to point. Let Λ( x µ ) be an arbitrary function
of space-time. For infinitesimal Λ,

φ −→ φ − iΛφ =⇒ δφ = −iΛφ
2.2. Lagrangian Forumlation for Continuous Systems 17

And,

∂µ φ −→ ∂µ φ − i (∂µ Λ)φ − iΛ(∂µ φ) =⇒ δ(∂µ φ) = −i (∂µ Λ)φ − iΛ(∂µ φ)

Similarly,
δφ∗ = iΛφ∗ & δ(∂µ φ∗ ) = i (∂µ Λ)φ∗ + iΛ(∂µ φ∗ )
These extra terms renders the Lagrangian to be no longer invariant.

δL = δ[(∂µ φ)(∂µ φ∗ )] − m2 δ(φ∗ φ)

= [δ(∂µ φ]∂µ φ∗ + (∂µ φ[δ∂µ φ∗ ] − m2 · 0
= [−iΛ(∂µ φ) − i (∂Λ)φ]∂µ φ∗ + (∂µ φ)[iΛ(∂µ φ∗ ) + i (∂µ Λ)φ∗ ] (2.1)
= (∂µ Λ)[−iφ∂µ φ∗ + iφ∗ ∂µ φ]
= (∂µ Λ) J µ

To restore gauge invariance we add an extra term, introducing a 4-vector Aµ

which couples directly to J µ :

L1 = −eJ µ Aµ
(2.2)
= −ie(φ∗ ∂µ φ − φ∂µ φ∗ ) Aµ

Under gauge transformation of the second kind,

1
Aµ −→ Aµ + ∂µ Λ
e
So,
δL1 = −e(δJ µ ) Aµ − eJ µ (δAµ ) = e(δJ µ ) Aµ − J µ ∂µ Λ
Since J µ = 2φ∗ φ∂µ Λ, so

δL + δL1 = −2eAµ (∂µ Λ)φ∗ φ

We thus add another term L2 = e2 Aµ Aµ φ∗ φ.

δL2 = 2e2 Aµ δAµ φ∗ φ = 2eAµ (∂µ Λ∗φ φ

We see that
δ L + δ L1 + δ L2 = 0
Even though the total Lagrangian L + L1 + L2 is now invariant, by logical consider-
ations, the introduced field Aµ which couples to the current J µ of the complex field
φ, must contribute by itself to the Lagrangian. So we need a L3 which is gauge in-
variant constructed from Aµ .
Let us define
Fµν = ∂µ Aν − ∂ν Aµ
. The new term in Lagrangian is

1
L3 = − F µν Fµν
4
Hence the total Lagrangian is

1
Ltot = L + L1 + L2 + L3 = (∂µ φ + ieAµ φ)(∂µ φ∗ − ieAµ φ∗ ) − m2 φ∗ φ − F µν Fµν
4
18 Chapter 2. Introduction to Continuous Systems and Fields

Here Fµν is the electromagnetic field tensor. It us worthwhile to do all these

to recognise the fact the electromagnetic field arises naturally by demanding gauge
invariance of Lagrangian under Gauge Transformation of the second kind (i.e, local
transformations).
Let us define covariant derivative Dµ φ = (∂µ + ieAµ )φ. It transforms covariantly
as can be seen:

δ( Dµ φ) = δ(∂µ φ) + ie(µ )φ + ieAµ δφ = −iΛ( Dµ φ)

Similarly, Dµ φ∗ = (∂µ − ieAµ )φ∗ It can be thought of such that φ describes a field
with charge e and φ∗ describes a field with charge -e. The electro-magnetic field can
be thought of as the gauge filed which has to be introduced to guarantee invariance
under local U(1) gauge transformations.
The Euler-Lagrange equations for Aµ gives:

∂L ∂L
− ∂ν [ ]=0
∂Aµ ∂(∂ν Aµ )
It gives

∂ν F µν = −ie(φ∗ ∂µ φ − φ∂µ φ∗ ) + 2e2 Aµ |φ|2 = −ie(φ∗ D µ φ −µ φ∗ ) = −eJ µ

where J µ = i (φ∗ D µ φ −µ φ∗ ) is the covariant current, and conserved when the elec-
tromagnetic field is present.
It is to be noted that the electromagnetic field is massless. A mass terms in the
Lagrangian would be of the form

L M = M2 Aµ Aµ

which is not invariant under the gauge transformation. Hence gauge invariance
requires that the gauge field is massless.
One last point to be noted is that here the charge e plays a double role. It is a
conserved quantity. It also appears as a coupling constant and measures the strength
with which a particle interacts with electric and magnetic fields.

2.2.4 Yang-Mills Field

The Yang-Mills field is an example of non-Abelian gauge field. Let us assume that
the field φ has three components, φ = (φ1 , φ2 , φ3 ), in the internal space, and the
gauge transformations are in this space. This gives a vector quantity which is con-
served (unlike charge which is scalar). Consider the gauge transformation (rotations
in internal space):

φ10 = cos Λ3 φ1 + sin Λ3 φ2 , φ20 = cos Λ3 φ2 − sin Λ3 φ1 , φ30 = φ3

If Λ3 is infinitesimal,

φ10 = φ1 + Λ3 φ2 , φ20 = φ2 − Λ3 φ1 , φ30 = φ3

which is essentially,

φ −→ φ0 = φ − Λ × φ =⇒ δφ = −Λ × φ
2.3. The Geometry of Gauge Fields 19

We also have

∂µ φ −→ ∂µ φ0 = ∂µ φ − ∂µ Λ × φ − Λ × ∂µ φ =⇒ δ(∂µ φ) = −∂µ Λ × φ − Λ × ∂µ φ

Introducing a gauge potential Wµ , we can write the covariant derivative as

Dµ φ = ∂µ φ + gWµ × φ

We require δ( Dµ φ) = −Λ × ( Dµ φ). Now Wµ transforms like

1 1
Wµ −→ Wµ − Λ × Wµ + ∂µ Λ =⇒ δWµ = −Λ × Wµ + ∂µ Λ
g g
And,

δ( Dµ φ) = δ(∂µ φ) + g(δWµ ) × φ + gWµ × (δφ)

= −Λ × ∂µ φ − g[(Λ × Wµ ) × φ + Wµ × (Λ × φ)]
(2.3)
= −Λ × (∂µ φ + gWµ × φ)
= − Λ × Dµ φ

It is to be noticed that Wµ is the analogue of of Aµ . Let’s call the analogue of

field strength tensor Fµν as Wµν. The requirement is δ(Wµν) = −Λ × Wµν. This
requirement is fulfilled by choosing

Wµν = ∂µ Wν − ∂ν Wµ + gWµ × Wν

Hence the Lagrangian is given by:

1 m2 1
L = ( Dµ φ ) · ( D µ φ ) − φ · φ − Wµν · Wµν
2 2 4
The Euler Lagrange equations are given by

∂L ∂L
− ∂ν [ ]=0
∂ (W i µ ) ∂(∂ν W i µ )
where i is an internal index. We finally get

∂ν Wµν + gW ν × Wµν = g[(∂µ φ) × φ + g(Wµ × φ) × φ]

And,
D ν Wµν = g( Dµ φ) × φ ≡ gJµ
In the absence of source term,

D ν Wµν = 0 −→ ∂ν Wµν = − gW ν × Wµν

which means that the field Wµν acts as a source for itself.

2.3 The Geometry of Gauge Fields

In this section we will try to talk the gauge transformations in the language of ge-
ometry. Rewriting the Yang-Mills gauge transformation of φ in its iso-space,

φ −→ φ0 = φ − Λ × φ
20 Chapter 2. Introduction to Continuous Systems and Fields

This is the infinitesimal form of

φ −→ φ0 = exp(iI.λ)φ

where the matrix generators I are given by

     
0 0 0 0 0 i 0 −i 0
I1 = 0 0 −i  I2 =  0 0 0 I3 =  i 0 0
0 i 0 −i 0 0 0 0 0

It can be seen ( Ii )mn = −ieimn .

0
φmn = (1 + iIi Λi )mn φn
= (δmn + eimn Λi )φn
(2.4)
= φm − emin Λi φn
= (φ − Λ × φ)m

Now let Λ depend on x µ .

φ −→ φ0 = exp(iI.λ( x ))φ = S( x )φ

For a general n component vector,

ψ( x ) −→ ψ0 ( x ) = exp[iM a Λ a ( x )]ψ( x ) = S( x )ψ( x )

where a is summed from 1 to 3, ψ is a n-component vector and M a are three genera-

tors.
ψ does not transform covriantly

∂µ ψ0 = S(∂µ ψ) + (∂µ S)ψ

The problem is that we are performing different iso-rotation at different points. ψ( x )

and ψ( x + dx ) are measured in different coordinate systems as discussed in Figure
2.1. To form a properly covariant derivative, ψ( x + dx ) shouldn’t be compared with
ψ( x ) but ψ( x ) + δψ where δψ is defined by parralel transport.
We assume that the form of δψ is δψ = igM a Aµa dx µ ψ where Aµa is an additional
potential- which tells us how axes in isospace vary at different points. Now

Dψ = (ψ + dψ) − (ψ + δψ) = dψ − δψ = dψ − igM a Aµa dx µ ψ

So
Dψ
= Dµ ψ = ∂µ ψ − igM a Aµa dx µ ψ
dx µ
This equation defines the covariant derivative for an arbitrary field ψ transforming
under a group whose generators are M a . Let’s check with some examples we have
already studied:

1. For the group U(1), put M=-1 and g=e, it gives

Dµ = ∂µ + ieAµ
2.3. The Geometry of Gauge Fields 21

F IGURE 2.1: (a) dψ carries information about variation in ψ as well as

change in coordinates. (b) δψ is defined by parallel transport

2. For the group SU(2), put ( M a )mn = −ieamn , it gives

Dµ φm = ∂ Dµ φm − ig( M a )mn Aµa φn

= ∂µ φm − geamn Aµa φn (2.5)
= (∂µ φ + gAµ × φ)m

A generic vector ψ transforms in isospace as ψ −→ Sψ. The covariant derivate

also transforms as
Dµ ψ −→µ ψ0 = SDµ ψ
Let us define Aµ = M a Aµa . Then,

Dµ ψ = (∂µ − igAµ )ψ

Transforming to a new iso-frame

(∂µ − igA0µ )ψ0 = S(∂µ − igAµ )ψ

Putting ψ0 = Sψ gives (worked out in Appendix A.1)

i
A0µ = SAµ S−1 − (∂µ S)S−1
g

This is the rule for gauge transformation of potential. For example: For the group
U(1), put S = e−iΛ , ∂S = −i (∂µ Λ)e−iΛ , with g=e and M=-1 gives

1
A0µ = Aµ + ∂µ Λ
e
22 Chapter 2. Introduction to Continuous Systems and Fields

For the case of SU(2), we have S = exp( 2i τ · Λ) and ∂µ S = i

2 · ∂µ Λ · S. We arrive at

1
A0µ = Aµ − Λ × Aµ + ∂µ Λ
g

References
The text has been mostly referred from Lewis H. Ryder, Quantum Field Theory, Sec-
ond Edition. Some portion has been borrowed from Goldstein, Poole, Safko, Classical
Mechanics, Third Edition, Pearson. For Electromagnetic field, Laundau Lifshitz, The
Classical Theory of Fields, 4th Edition was referred. For non-Abelian Gauge Fields,
Steven Weinberg, The Quantum Theory of Fields II: Modern Applications, Cambridge was
referred.
 23

Appendix A

Derivations

A.1 A0µ = SAS−1 − gi (∂µ S)S−1

(∂µ − igA0µ )ψ0 = S(∂µ − igAµ )ψ

Putting ψ0 = Sψ gives

(∂µ − igA0µ )Sψ = S(∂µ − igAµ )ψ

=⇒ ∂µ Sψ − igA0µ Sψ = S∂µ ψ − igS( Aµ ψ)

=⇒ S∂µ + (∂µ )ψ = S∂µ ψ − igSAµ ψ + igA0µ ψ
=⇒ (−igA0 S)ψ = (−igSAµ − ∂µ S)ψ
We take igA0µ S ≡ igSAµ + ∂µ S. Then,

1
=⇒ A0µ SS−1 = SAS−1 + ∂µ S
ig

i
: .A0µ = SAS−1 − (∂S)S−1
g

A.2 Relation between φR (p) and φL (p)

For a particle with energy E, mass m and momentum p, γ = E/m,

E + m + σ.p
φR ( p ) = φR (0)
[2m( E + m)]1/2
Similarly,
E + m − σ.p
φL ( p ) = φ L (0)
[2m( E + m)]1/2
Since originally the particle was at rest, φR (0) = φL (0). Acting by ( E + σ.p) on
φL (p) gives

( E + σ.p)( E + m − σ.p)
( E + σ.p)φL ( p) = φ L (0)
[2m( E + m)]1/2
24 Appendix A. Derivations

Now,

( E + σ.p)( E + m − σ.p)φL (0) = [ E2 + Em + Eσ · p − Eσ · p + mσ · p − (σ · p)(σ · p)]φR (0)

= ( E2 − p2 + Em + mσ · p)φR (0)
= (m2 + Em + mσ · p)φR (0)
= m ( m + E + σ · p ) φR (0)
= mφR ( p)

So
E + σ.p
φR ( p ) = φL ( p )
m
In a similar way, we can prove

E + m − σ.p
φL ( p ) = φ L (0)
[2m( E + m)]1/2
 25

Appendix B

Derivation of Euler-Lagrange
Equation from Variational
Principle

For a given Lagrangian density L(φ, ∂µ φ), the action S is

Z
S= L(φ, ∂µ φ)d4 x

The field φ traces out a 4-dimensional region R. The initial and final spacelike hy-
persurfaces may be taken to be time-slices t = t1 and t = t2 , which form part of the
boundary ∂R of region R. On subjecting the field and co-ordinates to a variation that
vanishes at the boundary ∂R:

x µ −→ x 0µ = x µ + δx µ

φ( x ) −→ φ0( x ) = φ( x ) + δφ( x )
Let’s consider the case where L depends on x µ : L(φ, ∂µ φ, x µ ).
Introducing the total variation in φ

φ( x 0 ) = φ( x ) + ∆φ( x )

To the first order in δx,

∆φ = φ0 ( x 0 ) − φ( x 0 ) + φ( x 0 ) − φ( x ) = δφ + (δµ φ)δx µ

The variation in the action is

Z Z
δS = L(φ0 , ∂µ φ0 , x 0µ )d4 x 0 − L(φ, ∂µ φ, x µ )

where d4 x 0 = J ( x 0 /x )d4 x, where J(x’/x) is the Jacobian of the transformation. We

have
∂x 0µ µ
= δλ + δλ δx µ
∂x λ
So
x0 ∂x 0µ
J ( ) = det( λ = 1 + δµ (δx µ )
x ∂x
Hence Z
δS = (δL + L∂µ δx µ d4 x
26 Appendix B. Derivation of Euler-Lagrange Equation from Variational Principle

where
∂L ∂L ∂L
δL = δφ + δ(∂µ φ) + µ δx µ
∂φ ∂(∂µ φ) ∂x
We see that δ(∂µ φ) = ∂µ δφ. So

∂L ∂L
Z
δS = [ δφ + ∂µ (δφ) + ∂µ (Lδx µ )]d4 x
R ∂φ ∂(∂µ φ)

Now

∂L ∂L ∂L
∂µ (δφ) = ∂µ [ δφ] − ∂µ [ ]δφ
∂(∂µ φ) ∂(∂µ φ) ∂(∂µ φ)
Using Gauss’s Theorem,

∂L ∂L ∂L
Z Z
δS = { − δµ [ ]}δφd4 x + [ δφ + Lδx µ ]dσµ
R ∂φ ∂(∂µ φ) ∂R ∂(∂µ φ)

Since the variation in φ and x µ vanishes on the boundary of R, so the second term
vanishes. The condition for stationary action is that the first term also vanishes over
all of R, regardless of R. Hence we get the Euler-Lagrange equation

∂L ∂ ∂L
− µ[ ]=0
∂φ ∂x ∂(∂µ φ)
 27

Appendix C

Spinor Algebra: Bilinear

Quantities

The spinors transform under general Lorentz transformation (boost and rotation)
like:
i i
φR −→ exp[ σ.(θ − iφ)]φR , φL −→ exp[ σ.(θ + iφ)]φL
2 2
Similarly,

−i −i
φR† −→ φR† exp[ σ.(θ + iφ)], φL† −→ φL† exp[ σ.(θ − iφ)]
2 2
Under parity, they transform as

φL ⇐⇒ φR

. We constructed a few quantities and stated how they behave upon transformation:
ψψ = φL † φR + φR † φL −→ scalar
ψγ5 ψ = φL † φR − φR † φL −→ pesudoscalar
ψγµ ψ = (φL † φL + φR † φR , −φL † σφL + φR † σφR ) −→ vector
ψγµ γ0 ψ = (φR † φR − φL † φL , φL † σφL + φR † σφR ) −→ axial vector
ψ(γµ γν − γν γµ )ψ −→ anti-symmetric tensor
Let us study some of them in detail.

1. ψψ = φL † φR + φR † φL −→ scalar
Under Lorentz transformation,
† †
ψ0 ψ0 = φL0 φR0 + φR0 φL0 = φL † φR + φR † φL = ψψ

Under parity,
† †
ψ0 ψ0 = φL0 φR0 + φR0 φL0 = φR † φL + φL † φR = ψψ

2. ψγ5 ψ = φL † φR − φR † φL −→ pesudoscalar
Under Lorentz transformation,
† †
ψ0 γ5 ψ0 = φL0 φR0 − φR0 φL0 = φL † φR − φR † φL = ψγ5 ψ

Under parity,
† †
ψ0 γ5 ψ0 = φL0 φR0 − φR0 φL0 = −(φL † φR − φR † φL )
 28 Appendix C. Spinor Algebra: Bilinear Quantities

3. ψγµ ψ = (φL † φL + φR † φR , −φL † σφL + φR † σφR ) −→ vector

0th component:
Under spatial rotations,
† †
ψ0 γ0 ψ0 = φL0 φL0 + φR0 φR0 = φL † φL + φR † φR = ψγµ ψ

It is clearly evident that ψγ0 ψ = φL † φL + φR † φR is invariant under parity transfor-

mation.
ith component:
Under spatial rotations,

−i i −i i
ψ0 γψ0 = −φL† e 2 σ·θ σφL e 2 σ·θ + φR† e 2 σ·θ σφR e 2 σ·θ
= −φL† (1 − 2i σ · θ )σφL (1 + 2i σ · θ ) + φR† (1 − 2i σ · θ )σ · θσφR (1 + 2i σ · θ )σ ·
= −φL† (σ − θ × σ)φL + −φR† (σ − θ × σ)φR
= ψγψ − θ × (ψγψ)

Under parity transformation, ψγi ψ = −φL † σφL + φR † σφR changes sign. It is clearly
visible.
4. ψγµ γ0 ψ = (φR † φR − φL † φL , φL † σφL + φR † σφR ) −→ axial vector
It behaves under rotation in the same way as the above one. However, it exhibits
opposite behaviours under parity transformation.
5. Wµν = ψ(γµ γν − γν γµ )ψ −→ anti-symmetric tensor
After pursuing the algebra, the elements of Wµν can be summarized as

Wi0 = 2[φR† σi φL − φL† σi φR ]

Wij = −2eijk [φR† σk φL + φL† σk φR ]