Classical Field Theory
Classical Field Theory
Sagar J C
St. Joseph’s College, Bengaluru
19PCM21015
3rd year Bachelor of Science
1 Classical Fields 1
1.1 Relativistic notations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Euler Lagrange equation of Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2.1 For a classical particle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2.2 For a classical Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.3 Lagrangian of classical fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Symmetries and Conservation Laws - Noether’s Theorem . . . . . . . . . . . . . . . . . 6
1.4 Complex Classical Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.4.1 Position independent Rotation of a complex field . . . . . . . . . . . . . . . . . 9
1.4.2 Position dependent Rotation of a complex field . . . . . . . . . . . . . . . . . . 11
1.5 Classical Gauge Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.6 Interaction between a classical particle and a classical field . . . . . . . . . . . . . . . . 13
1.6.1 Effect of the field on the particle . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.6.2 Effect of the particle on the field . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.7 Properties of Lagrangian and Lagrangian Density . . . . . . . . . . . . . . . . . . . . . 16
i
Chapter 1
Classical Fields
A particle in classical physics is an object which exists and is defined by its property of mass, charge, which
occupies a single point in space whose position is given by ‘n’ number of coordinates in ‘n’-dimensional
space. And also has various other mechanical properties like velocity, acceleration, momentum, force,
Energy, Lagrangian, etc. But a field is very different from a particle. A field is a quantity that has a
particular value at all space-time coordinates. Examples for classical fields are Temperature field (scalar
field), Electromagnetic field (vector field), Wind field (vector field). Our main concentration in this chapter
will be defining some classical fields and assigning some transformations to them, verifying their invariance,
building the Lagrangian, finding out whether the Lagrangian is invariant under the defined transformation,
and deriving some of its properties. Before going further let us define the relativistic notations that we use
throughout the paper and recall some useful relativistic concepts.
The metric signature used here is (1, −1, −1, −1). Thus the time component of a covariant four-vector (aµ )
is positive and the spatial components are all negative aµ = (a0 , −~a). The contravariant four-vector (aµ ) is
the dual of covariant four-vector and all the components here are positive aµ = (a0 , ~a). Also we can relate the
components of covariant and contravariant form of a four-vector as (a0 = a0 , a1 = −a1 , a2 = −a2 , a3 = −a3 ).
To change a contravariant tensor into a covariant tensor we have to multiply it by the metric tensor (ηµν ).
And to change a covariant tensor to a contravariant tensor we have to multiply it by the metric tensor η µν .
1
Inner product of four vectors
Let aµ be any contravariant four vector defined by aµ = (a0 , a1 , a2 , a3 ) = (a0 , ~a) and its dual, i.e., the
covariant form is given by, aµ = (a0 , −~a). Then the inner product of the four vector is a scalar defined by,
aµ aµ = a0 a0 − a1 a1 − a2 a2 − a3 a3 = a0 a0 − ~a · ~a (1.4)
If bµ = (b0 , b1 , b2 , b3 ) = (b0 , ~b) then the inner product is aµ bµ = a0 b0 − a1 b1 − a2 b2 − a3 b3 = a0 b0 − ~a · ~b
Note that any quantity which is of the form aµ bµ is a scalar and a Lorentz invariant and it means the the
corresponding components of the four vectors are multiplied and then are summed. So we get an identity,
aµ bµ = aµ bµ (1.5)
Space-time Interval
It is the distance between any two points in the space time given by, ds2 = c2 dt2 − dx2 − dy 2 − dz 2 (1.6)
The space-time coordinates are given by the four position vectors. The contravariant position vector is
given by dxµ = (cdt, dx, dy, dz) and the covariant position vector is dxµ = (cdt, −dx, −dy, −dz). Thus
the space-time interval is related to the position vector as,
Four Gradient
~ There are two forms covariant
Four Gradient operator (∂) is the analogue of three gradient operator (∇).
µ
four gradient (∂µ ) and contravariant four gradient (∂ ). Covariant 4 gradient components are defined by,
∂ 1 ∂ ∂ ∂ ∂ 1 ∂
~
∂µ = = (∂ t , ∂x , ∂y , ∂z ) = , , , = , ∇
∂xµ c ∂t ∂x ∂y ∂z c ∂t
∂ 1 ∂ ∂ ∂ ∂ 1 ∂ ~
Its contravariant form, ∂ µ = = (∂ t , ∂ x , ∂ y , ∂ z ) = ,− ,− ,− = , −∇
∂xµ c ∂t ∂x ∂y ∂z c ∂t
1 ∂ 1 ∂
∂µ = ~
,∇ and ∂µ = ~
, −∇ (1.10)
c ∂t c ∂t
1 ∂2
⇒ = − ∇2 (1.11)
c2 ∂t2
2
1.2 Euler Lagrange equation of Motion
1.2.1 For a classical particle
The consider a classical particle ‘P ’ in space time. Its position is defined by the four coordinates given by
xµ = (ct, x, y, z). The physics of that particle can be explained fully by the principle of least action. Let the
generalized coordinate of the particle be qi and q̇i which is the time rate of change of qi be the generalized
velocity, then the action of the particle is given by,
ˆ
Action = L(q, q̇, t)dt (1.12)
Where L is the Lagrangian of the particle which is generally the difference between the Kinetic energy and
Potential energy of the particle. Now the principle of least action extremize the path taken by the particle.
Let us apply some change in the generalized coordinate and keep the Lagrangian invariant for least action.
The transformation in coordinate is,
Then by differentiating it we get the transformed velocity and its change which is given by,
q̇i → q̇i + εf˙(qi ) ⇒ the change in velocity is, δ q̇i = εf˙(qi ) (1.14)
So according to the principle of least action, the change in the action has to be zero even if the coordinates
change. So the change in the Lagrangian has to be zero for action to remain invariant,
ˆ ˆ
⇒ δ L(qi , q̇i , t)dt = 0 ⇒ δL(qi , q̇i , t)dt = 0 (1.15)
ˆ " #
∂L ∂L
⇒ δqi + δ q̇i dt = 0 (1.16)
∂qi ∂ q̇i
ˆ " #
∂L ∂L ˙
⇒ εf (qi ) + εf (qi ) dt = 0 (1.17)
∂qi ∂ q̇i
ˆ " #
∂L ∂L ˙
⇒ ε f (qi ) + f (qi ) dt = 0 (1.18)
∂qi ∂ q̇i
| {z }
Integrate by parts
ˆ " #
∂L d ∂L
⇒ ε f (qi ) − f (qi ) dt = 0 (1.19)
∂qi dt ∂ q̇i
ˆ " #
∂L d ∂L
⇒ εf (qi ) − dt = 0 (1.20)
∂qi dt ∂ q̇i
We want the action to be zero for every possible change in the coordinate ‘εf (qi )’. So the only way to
equate the LHS of the above equation to zero is by equating the integrand to zero. Only in that case, the
action will be extremized and the change in action will be zero. So the resulting condition will govern the
path of the particle such that it holds the Principle of least action. This results in an equation of motion
that governs the motion of the particle which is ‘Euler-Lagrange Equation of Motion’,
∂L d ∂L
− =0 (1.21)
∂qi dt ∂ q̇i
dP ∂V
This is an equation of motion because it is analog to Newton’s second law F = − where F = −
dt ∂x
which is an equation of motion. Consider a particle moving in x-axis, then we can rearrange Eq.(1.21) as,
d ∂L ∂L
= (1.22)
dt ∂ ẋ ∂x
3
We know that L = KE − P E, for a particle moving in x-axis , K.E = 12 mẋ2 and P.E. = V (x). So,
L = 21 mẋ2 − V (x). Thus using this Lagrangian in the above equation, we get,
!
d ∂ 1 ∂ 1
mẋ2 − V (x) = mẋ2 − V (x) (1.23)
dt ∂ ẋ 2 ∂x 2
d ∂
⇒(mẋ) = − (V (x)) (1.24)
dt ∂x
∂V (x) dP
Since, mẋ = P and F = − ⇒ F = (Equation of motion) (1.25)
∂x dt
Thus Euler-Lagrange equation is a generalized equation. Where qi is the generalized coordinate, q̇i is the
∂L ∂L
generalized velocity, = Pi is the generalized momentum, and, = Fi is the generalized Force.
∂ q̇i ∂qi
ϕ(ct, x, y, z)
4D Hypersurface
ct
representing
space-time
x
z ds
Consider an arbitrary classical field which is denoted by ϕ, since it is a field, it is defined throughout the
space-time, so ϕ is a function of (ct, x, y, z). You can see in the above diagram how the field is represented.
At every point of space-time, there is a corresponding value of the field. So the ϕ axis is a 5th dimensional
axis for our 4D world. ϕ and its various derivatives are given by,
1 ∂ϕ ∂ϕ ∂ϕ ∂ϕ
ϕ = ϕ(ct, x, y, z) and ϕ̇ = , ϕx = , ϕy = , ϕz =
c ∂t ∂x ∂y ∂z
So if we want to study the ϕ field we have to consider the field as a degree of freedom that has its values
at different space-time coordinates and that varies at every coordinate. Let us generalize the derivatives by
assigning them the role of a four-vector. Using the four derivatives in Eq.(1.10), we can write,
∂ϕ 1 ∂ϕ ∂ϕ ∂ϕ ∂ϕ ∂ϕ µ µ
1 ∂ϕ ∂ϕ ∂ϕ ∂ϕ
= ϕ µ = ∂ µ ϕ = , , , and = ϕ = ∂ ϕ = ,− ,− ,− (1.26)
∂xµ c ∂t ∂x ∂y ∂z ∂xµ c ∂t ∂x ∂y ∂z
1 ∂ϕ 2 ∂ϕ 2 ∂ϕ 2 ∂ϕ 2
∴ ∂µ ϕ ∂ µ ϕ = ϕµ ϕµ = − − − (is a Lorentz invariant) (1.27)
c2 ∂t ∂x ∂y ∂z
Let us write the analog of the Euler Lagrange equation of a particle given in Eq.(1.21), which can be used
for fields. In this case, ϕ itself acts as a coordinate. The analog of Lagrangian (L) here is called as the
‘Lagrangian Density’ denoted by L. Lagrangian is then given to be the space integral of Lagrangian density.
This L depends on the value of ‘ϕ’ and its derivatives ‘ϕµ ’,
ˆ
⇒ L = L(ϕ, ϕ̇, ϕx , ϕy , ϕz ) ⇒ L= L(ϕ, ϕ̇, ϕx , ϕy , ϕz )d3 x (1.28)
V
!
∂L ∂ ∂L
Thus the analog of Euler-Lagrange equation for a classical field becomes, − =0
∂ϕ ∂xµ
∂ϕ
∂ ∂xµ
4
∂ϕ
We can simplify the above equation by using the notations where, ϕ(x) → ϕ, and → ϕµ , thus we get,
∂xµ
!
∂L ∂ ∂L
− µ =0 ⇒ ∂ϕ L − ∂µ (∂ϕµ L) = 0 (1.29)
∂ϕ ∂x ∂ϕµ
The µ index in the above equation runs from (0, 1, 2, 3) corresponding to (t, x, y, z). Since there are two ‘µ’
indices in the second term of Eq.(1.29), it can be expanded for different values of µ and summed to get,
! ! ! !
∂L ∂ ∂L ∂ ∂L ∂ ∂L ∂ ∂L
− − − − =0 (1.30)
∂ϕ ∂t ∂ ϕ̇ ∂x ∂ϕx ∂y ∂ϕy ∂z ∂ϕz
The analogue of generalized quantities of particle mechanics in the case of classical fields are,
∂ϕ
Canonical conjugate velocity, ϕµ = = ∂µ ϕ (1.31)
∂xµ
∂L
Canonical conjugate momentum, Πµ = (1.32)
∂ϕµ
∂L
Canonical conjugate Force, Fµ = (1.33)
∂ϕ
1
If V (ϕ) = m2 ϕ2 ⇒ V 0 (ϕ) = m2 ϕ
2
1 ∂2ϕ ∂2ϕ ∂2ϕ ∂2ϕ
∴ − − − 2 = −m2 ϕ ⇒ ( + m2 )ϕ = 0 (1.39)
c2 ∂t2 ∂x2 ∂y 2 ∂z
This is a different type of wave equation called as the Klein-Gordon Equation that has a great deal of
implication in Quantum Field Theory which describes the scalar fields whose quantum particles are particles
whose spin is zero (scalar particles) where the term ‘m2 ’ corresponds to the mass of the quanta of the field.
5
1.3 Symmetries and Conservation Laws - Noether’s Theorem
For particles
Consider a system of particles whose Lagrangian is L. If we apply coordinate transformation such that,
qi → qi + εf (qi ) ⇒ δqi = εf (qi ) (1.40)
⇒ q̇i → q̇i + εf˙(qi ) ⇒ δ q̇i = εf˙(qi ) (1.41)
where ε is some constant and f (qi ) is some arbitrary function of qi . This means that the coordinate of the
system is shifted by an amount of δqi = εf (qi ). Now if there is a symmetry present, i.e., if,
∂L ∂L
δL(qi , q̇i ) = 0 ⇒ δqi + δ q̇i = 0 (1.42)
∂qi ∂ q̇i
∂L
⇒ δqi + Pi δ q̇i = 0 (1.43)
∂qi
∂L dPi
Eq.(1.21), can be written in terms of Pi as, = = Ṗi (1.44)
∂qi dt
X
⇒ Ṗi δqi + Pi δ q̇i = (Ṗi δqi + Pi δ q̇i ) = 0 (1.45)
i
X d d X
⇒ Pi δqi = Pi δqi = 0 (1.46)
dt dt
i i
X
⇒ Pi δqi = constant (1.47)
i
This is Noether’s theorem that says under the transformation of coordinates qi → qi + δqi , if the Lagrangian
is invariant due to a symmetry, then there is a conserved quantity associated with it given by,
X
Pi δqi = constant (conserved) (1.48)
i
Example
Consider a free particle moving in x-axis, whose Lagrangian is given by, L = 21 mẋ2 . If we apply a transla-
tional transformation, i.e. x → x + ε, where f (x) = 0 implying it is a constant shift of coordinate. Then
ẋ → ẋ since ε is a constant, then L = 21 mẋ2 remains the same, hence δL = 0, so the conserved quantity is,
X X X
Pi δqi = constant ⇒ Px · ε = constant ⇒ Px = constant (1.49)
i
Consider the L given in Eq.(1.34), for simplicity let V (ϕ) = 0, then if we apply the transformation of
ϕ → ϕ + ε ⇒ δϕ = ε which is a constant shift in the field value, then since the L depends only on
derivatives of ϕ, then the transformation doesn’t change the L, thus by Noether’s theorem, the quantity,
X ∂L ϕ̇ ˆ
X ϕ̇
4
Πµ δϕ = = 2 −ϕx −ϕy −ϕz = constant ⇒ 2
−ϕx −ϕy −ϕ z d x = constant (1.51)
µ µ
∂ϕ µ c c
This conserved quantity is not the actual momentum, it is not that kind of momentum if it hits you knocks
you. Rather it is a canonical conjugate momentum that is associated with the rate of change of ϕ with
´ ϕ̇
4
respect to time. And that quantity given by − ϕ x − ϕy − ϕ z d x is a constant.
c2
6
Translation of field in space
ϕ(x)
ϕ(x − ε)
δϕ
ε x
Consider the physical translation of some field in space. For simplicity let us consider a plane wave moving
in +ve x-axis. Let the variation in x be a small constant quantity ε. Then the field after translation is,
∂ϕ ∂ϕ ∂ϕ
ϕ → ϕ + δϕ ⇒ δϕ =δx = (−ε) ⇒ ϕ → ϕ − ε (1.52)
∂x ∂x ∂x
The −ve sign is because some constant has to be subtracted from the argument of ϕ for the field to move
in +ve x-axis. The Lagrangian density as we saw in Eq.(1.34) was,
ϕ̇2 ϕ2x
L= − (1.53)
2c2 2
The Lagrangian density is invariant under the transformation because it is a constant time-independent
shift of the wave. Then the conserved quantity is given by,
ˆ ∂ϕ ˆ ˆ
∂L ∂ϕ ϕ̇ ϕx
Πi − dx = − dx = − dx (1.54)
∂x ∂ ϕ̇ ∂x c2
This quantity is conserved by virtue of the transnational space invariance of the wave. This conserved
momentum is the sum of both canonical and ordinary linear momentum (P~ ) carried by the wave which is,
"ˆ #
ϕ̇ ϕ x
P~ = − dx î (1.55)
c2
1 ∂2ϕ ∂2ϕ
(Wave equation is of the form,) 2 2
= (1.56)
c ∂t ∂x2
(For a right moving wave the solution is,) ϕ = ϕ(x − ct) (1.57)
ˆ
ϕ02
⇒ ϕ̇ = −cϕ0 ϕx = ϕ0 ⇒ |P~ | = dx (1.58)
c
ˆ 2
ϕ̇2 ϕ2 ϕ̇ ϕ2x
The total energy of the wave is given by, H= 2 + x ⇒ E= + dx (1.59)
2c 2 2c2 2
ˆ 02 ˆ ˆ 02
ϕ ϕ02 02 ϕ
⇒ E= + dx = ϕ dx = c dx = c|P~ | (1.60)
2 2 c
Thus for a wave, we get the Energy to be the speed times the magnitude of the momentum. And for a
usual sine wave in +ve x-axis, we get,
1 2πc 2 2π 2 µA2 c2 2π 2 µA2 c
ϕ = A sin(kx − ωt) E = µA2 = P = (1.61)
2 λ λ2 λ2
7
1.4 Complex Classical Field
Im
ϕ = ϕr + iϕi
ϕi
O ϕr Re
Consider a field that is represented by complex numbers. And every point in the field (ϕ) is the combination
of two components (ϕr , ϕi ). The field ϕ depends on its components (ϕr and ϕi ), which in turn depend on
the four coordinates of space-time. The field ϕ and its conjugate ϕ∗ (i → −i) are represented as,
√
ϕ = ϕr + iϕi and ϕ∗ = ϕr − iϕi (i = −1 and ϕr , ϕi ∈ R) (1.62)
⇒ ϕϕ∗ = (ϕr + iϕi )(ϕr − iϕi ) = (ϕr )2 − (iϕi )2 = ϕ2r + ϕ2i (1.63)
1 1 1 1
⇒ − m2 ϕ∗ − ∂µ ∂ µ ϕ∗ = 0 and − m2 ϕ − ∂ µ ∂µ ϕ = 0 (1.70)
2 2 2 2
Which results into (∂µ ∂ µ +m2 )ϕ∗ = 0 and (∂ µ ∂µ +m2 )ϕ = 0 → Klein-Gordon Equation (1.71)
These equations for ϕ and ϕ∗ vaguely indicate the existence of particles and anti-particles. And the fact
that we got Klein-Gordon equation implies that the Lagrangian for complex field we have chosen is correct,
which is again quadratic in nature. And thus has all the symmetries which we have discussed so far. But
this complex field has a special type of symmetry discussed further which has very high implications and
will be used in Quantum field theory and particle physics.
8
1.4.1 Position independent Rotation of a complex field
Im
ϕ0 = ϕ eiθ
ϕ = ϕr + iϕi
O Re
iω
q write any complex number as a vector ϕ = ϕr + iϕi and convert it in polar form as ϕ = ze where
We can
z = ϕ2r + ϕ2i is the magnitude of the complex vector and ω is the angle made by the complex vector and
the real axis. If we want to rotate the complex vector then we just have to change the value of ω, If ω
increases, then the vector is positively rotated in the counter-clockwise direction and if ω decreases, then
the vector is negatively rotated in the clockwise direction. So if we have to rotate (transform) the complex
field ϕ then we have to multiply it by the rotation factor eiθ which rotates the complex vector by an angle
of θ such rotational transformation is given by,
∗
ϕ0 → ϕ eiθ and ϕ0∗ → ϕ∗ e(iθ) i.e., ϕ0∗ → ϕ∗ (e−iθ ) (1.72)
∗
⇒ ∂µ ϕ0 = (∂µ ϕ)eiθ and ∂µ ϕ0∗ = (∂µ ϕ∗ ) e(iθ) i.e., ∂µ ϕ0∗ = (∂µ ϕ∗ )e−iθ (1.73)
⇒ (ϕ ϕ∗ )0 = ϕ iθ
e · ϕ∗ (
e−iθ
)=ϕϕ∗ (ϕ ϕ∗ is invariant under complex rotation) (1.74)
⇒ ∂µ ϕ0 ∂ µ ϕ0∗ = (∂µ ϕ)iθ
e · (∂µ ϕ∗ )
e−iθ
= ∂µ ϕ ∂ µ ϕ∗ (∂µ ϕ ∂ µ ϕ∗ is also invariant) (1.75)
Thus we see that the terms like ϕ ϕ∗ and ∂µ ϕ ∂ µ ϕ∗ are invariant under rotational transformations. And
since the Lagrangian density contains such terms only, we can tell that the Lagrangian density is also
invariant under complex rotation. Let the rotation occur such that the angle of complex vector vary in very
small quantities, then, in this case, we get the transformation to be
i2 θ2 i3 θ3
ϕ0 = ϕ(eiθ ) = ϕ 1 + iθ + + + ... (1.76)
2 6
For small rotations, we can neglect higher powers of θ thus the transformation becomes
9
When the same math is done for in normal three-dimensional space where there is a wave propagating in a
particular direction (z) but the particles oscillate in two perpendicular directions (x = ϕ1 and y = ϕ2 ) as
shown in the below figure. Then the conserved quantity becomes the angular momentum per unit length
of the wave given by Lz = xẏ − ẋy of the wave whose axis is along the direction of propagation where the
angular momentum travels along with the direction of propagation of wave as Lz → z-component.
In quantum field theory, the conserved quantity represents the charge of the quantum of the field where the
charged particles are represented by complex fields. Since the quantity is having space and time derivatives,
in some sense the complex charge vector rotates as time proceeds and also as the charged particle moves
in space. Thus charge is like an analog of the angular momentum of the rotating complex charge vector.
ˆ !
i ∂ϕ∗ ∗ ∂ϕ
The quantity given by, ϕ µ −ϕ d4 x = Jµ is a four-vector since it has four components.
2 ∂x ∂xµ
That is nothing but the four current (refer section(3.1.2)) whose time component is charge density and space
components are current density. Implying that the time component given by charge density is conserved.
We know that ϕ = ϕr + iϕi and ϕ∗ = ϕr − iϕi where ϕr and ϕi are real. Thus we get,
∂ϕ∗ ∂
ϕ µ
= (ϕr + iϕi ) µ (ϕr − iϕi ) = ϕr ∂µ ϕr + ϕi ∂µ ϕi + i(ϕi ∂µ ϕr − ϕr ∂µ ϕi ) (1.83)
∂x ∂x
∂ϕ ∂
ϕ∗ = (ϕr − iϕi ) µ (ϕr + iϕi ) = ϕr ∂µ ϕr + ϕi ∂µ ϕi − i(ϕi ∂µ ϕr − ϕr ∂µ ϕi ) (1.84)
∂xµ ∂x
∂ϕ∗ ∂ϕ
⇒ ϕ µ − ϕ∗ µ = i(ϕi ∂µ ϕr − ϕr ∂µ ϕi ) + i(ϕi ∂µ ϕr − ϕr ∂µ ϕi ) = 2i ϕi ∂µ ϕr − ϕr ∂µ ϕi
∂x ∂x
!
i ∂ϕ∗ ∗ ∂ϕ
∂ϕ
∗
∗
⇒ Jµ = ϕ µ −ϕ = − ϕ ∂ ϕ
i µ r − ϕ ∂ ϕ
r µ i = Im ϕ = Im ϕ ∂ µ ϕ (1.85)
2 ∂x ∂xµ ∂xµ
Thus the four current density Jµ is given by the imaginary part of ϕ∗ ∂µ ϕ . Since charge density ρ is the
ˆ
1 ∗ ∂ϕ 3
time component of Four current, the charge Q is given by the time component, Q = Im ϕ d x.
c ∂t
This Q in normal 3D space corresponds to the angular momentum. Now let us find the expression for ∂ µ Jµ .
∂ µ Jµ = Im ∂ µ ϕ∗ ∂µ ϕ = Im ∂µ ϕ ∂ µ ϕ∗ + ϕ∗ (∂ µ ∂µ ϕ) = Im ∂µ ϕ ∂ µ ϕ∗ + Im ϕ∗ (∂ µ ∂µ ϕ) (1.86)
By Eq.(1.65) we get ∂µ ϕ ∂ µ ϕ∗ to be a real quantity. So Im ∂µ ϕ ∂ µ ϕ∗ = 0. And by Eq.(1.71) we get
∂ µ ∂µ ϕ = −µ2 ϕ. So we get Im ϕ∗ (∂ µ ∂µ ϕ) = Im ϕ∗ (−µ2 ϕ) = Im − µ2 ϕ∗ ϕ . By Eq.(1.63) we get ϕ∗ ϕ
to be real quantity and also the quantity µ2 is a real constant. Thus we get Im ϕ∗ (∂ µ ∂µ ϕ) = 0. Hence
we get ∂ µ Jµ = 0 which is the equation of continuity. More about this is dealt in the section(2.3.2)
10
1.4.2 Position dependent Rotation of a complex field
In the previous section, we saw the rigid rotation of a field. Where the rotation θ is the same throughout
the field i.e., every point of the field is rotated by an equal amount so θ was constant with respect to space
and time. In that section, we invariably assumed that the coordinate reference frames at each place to
be parallel and aligned to each other. So only then we could tell that the wave was rotating rigidly. But
what if the coordinate reference frames are not parallel to each other but they rotated in space-time or if
the field rotation was different in different locations of space-time where the rotation is a function given
by θ = θ(t, x, y, z), then would the complex field still have a symmetry under such position dependent
rotational transformation? Let us try to evaluate. . . .
1 1
cancelling e−iθ(xµ ) and eiθ(xµ ) we get, L =∂µ ϕ + iϕ∂µ θ(xµ ) ∂ µ ϕ∗ − iϕ∗ ∂ µ θ(xµ ) − m2 ϕϕ∗ (1.92)
2 2
On expanding the first term of the Lagrangian,
!
1 1
L= (∂µ ϕ)(∂ ϕ )+iϕ(∂ ϕ )∂µ θ(x )−iϕ (∂µ ϕ)∂ θ(xµ )−i (ϕϕ )∂µ θ(x )∂ θ(xµ ) − m2 (ϕϕ∗ ) (1.93)
µ ∗ µ ∗ µ ∗ µ 2 ∗ µ µ
2 2
Interchanging the covariant and contravariant terms of four gradient of θ in the second term by using the
identity aµ bµ = aµ bµ and then taking ∂ µ θ(xµ ) common from the middle terms, we get,
!
1 1
L= ∂µ ϕ ∂ µ ϕ∗ + i∂ µ θ(xµ ) ϕ∂µ ϕ∗ − ϕ∗ ∂µ ϕ + (ϕϕ∗ )∂µ θ(xµ ) ∂ µ θ(xµ ) − m2 (ϕϕ∗ ) (1.94)
2 2
But this Lagrangian cannot be simplified further. And the form of the Lagrangian is changed under the
position dependent rotational transformation of complex fields. But we shall force the Lagrangian to become
invariant by doing some subtle changes which will be seen in the next section.
11
1.5 Classical Gauge Theory
Gauge theory is the mathematical formulation that makes the Lagrangian density invariant under local
transformations for example the position-dependent rotation of complex fields. Since the form of the La-
grangian given in Eq.(1.68) is not invariant under position-dependent rotation, we shall force it to become
invariant by adding new fields and then giving them some required transformations.
Let us define a new field called as gauge field denoted by A which has four components. The time
component is a scalar field and the spacial components are the three components of a vector field. So
Aµ = (A0 , A1 , A2 , A3 ) = (φ/c, Ax , Ay , Az ) and Aµ = (A0 , A1 , A2 , A3 ) = (φ/c, −Ax , −Ay , −Az ). This gauge
field A is a dynamical field that has its own Lagrangian and also has its own equation of motion. It is the
basis of Electro-magnetic theory which will be dealt in detail in Chapter(3).
For now let us define the Lagrangian which is used in gauge theory which has the form,
1 1
L = Dµ ϕ Dµ ϕ∗ − m2 ϕϕ∗ (1.95)
2 2
where we have just replaced ∂µ with Dµ and ∂ µ with Dµ . Here Dµ and Dµ are other types of derivative
that are called as the covariant derivative where,
To verify whether the Lagrangian is invariant or not let us first consider the non-transformed fields ϕ0 , ϕ0∗ ,
A0 , and the non-transformed Lagrangian L. Then expanding this Lagrangian we get,
1 1
L0 = (Dµ ϕ0 ) (Dµ ϕ0 )∗ − m2 ϕ0 ϕ0∗ (1.97)
2 2
1 1
⇒ L0 = (∂µ ϕ0 + ieA0µ ϕ0 )(∂ µ ϕ0∗ − ieA0µ ϕ0∗ ) − m2 ϕ0 ϕ0∗ (1.98)
2 2
µ
Now we shall apply the local rotational transformation ϕ0 → ϕ eiθ(x ) and ϕ0∗ → ϕ∗ e−iθ(xµ ) . We
have already seen what would happen to the term ∂µ ϕ0 and ∂ µ ϕ0∗ in Eq.(1.88). Let us go through each
part of the above Lagrangian one by one. Thus,
µ) µ µ)
∂µ ϕ0 + ieA0µ ϕ0 = ∂µ ϕeiθ(x + iϕeiθ(x ) ∂µ θ(xµ ) + ieA0µ ϕ eiθ(x (1.99)
!
0 1 µ
= ∂µ ϕ + ie Aµ + ∂µ θ(x ) ϕ eiθ(x )
µ
(1.100)
e
∂ µ ϕ0∗ − ieA0µ ϕ0∗ = ∂ µ ϕ∗ e−iθ(xµ ) − iϕ∗ e−iθ(xµ ) ∂ µ θ(xµ ) − ieA0µ ϕ∗ e−iθ(xµ ) (1.101)
!
1 µ
= ∂ ϕ − ie A + ∂ θ(xµ ) ϕ e−iθ(xµ )
µ ∗ 0µ ∗
(1.102)
e
1 2 0 0∗ 1 2 iθ(xµ ) ∗ −iθ(xµ ) 1 2
m ϕϕ = m ϕe ϕ e = m ϕ ϕ∗ (1.103)
2 2 2
Thus after the transformation, the Lagrangian becomes,
! !
1 1 1 1
L= ∂µ ϕ + ie A0µ + ∂µ θ(xµ ) ϕ ∂ µ ϕ∗ − ie A0µ + ∂ µ θ(xµ ) ϕ∗ − m2 ϕ ϕ∗ (1.104)
2 e e 2
12
Now we need the above transformed Lagrangian to have the same form of the non-transformed Lagrangian
1 1
in Eq.(1.98). So for that, we need the terms A0µ + ∂µ θ(xµ ) and A0µ + ∂ µ θ(xµ ) to be equal to the
e e
transformed gauge field Aµ and Aµ . Thus if we define a certain type of transformation to the gauge field
called as Gauge transformation defined by,
1 1
A0µ → Aµ − ∂µ θ(xµ ) and A0µ → A0µ − ∂ µ θ(xµ ) (1.105)
e e
Then we get the transformed Lagrangian to be,
1 1 1 1
L = (∂µ ϕ + ieAµ ϕ)(∂ µ ϕ∗ − ieAµ ϕ∗ ) − m2 ϕ ϕ∗ = (Dµ ϕ) (Dµ ϕ)∗ − m2 ϕϕ∗ (1.106)
2 2 2 2
Thus we get the Lagrangian to be invariant under local rotational transformation only when,
µ)
ϕ0 → ϕ eiθ(x and ϕ0∗ → ϕ∗ e−iθ(xµ )
1 1
A0µ → Aµ − ∂µ θ(xµ ) and A0µ → A0µ − ∂ µ θ(xµ )
e e
Dµ = ∂µ + ieAµ and Dµ = ∂ µ + ieAµ
This whole set of transformations is called as the Gauge Transformation. And the Lagrangian given in
Eq.(1.95) is said to be gauge invariant under gauge transformation. This is called as the Gauge symmetry.
The gauge field A known as the four potential has a theory of its own. Moreover, the conserved quantity
in gauge symmetry here is not any physical quantity rather it is an equation of motion that governs
the behavior of charges near the Electro-magnetic fields which will be covered in Chapter(3). Since the
electric and magnetic fields are the derivatives of the gauge field A, we can tell that the formulation of
electromagnetism is a gauge theory. Before that, we shall revise electromagnetic theory in Chapter(2)
13
Consider a particle P present in some pre-existed field ϕ. Let us assume that the field is constant with
respect to time so ϕ → ϕ(r) where r is the space coordinates. If the particle and field are coupled to
each other then we would like to know how the existence of the field varies the motion of a particle and
also how the existence of the particle varies the behavior of ϕ. For that we need a Lagrangian, let us
build a Lagrangian and then derive the equations of motion. We know the Lagrangian of a free particle in
space-time which is moving at a three-velocity v according to Eq.(1.8). That is got by the expression,
ˆ r
v2 ∂L −mv
Action = (−mc2 )dτ ⇒ L = −mc2 1 − 2 ⇒ P = =q (1.107)
c ∂v 1− v
2
c2
Now to get the Lagrangian for the interaction between the particle and field let us just add a term −gϕdτ
to the above equation for action where g is the coupling constant between the particle and the field which
determines the strength of the interaction. Thus we get,
ˆ ˆ ˆ ˆ r !
2 2 2 v2
Action = (−mc )dτ − gϕdτ = − (mc + gϕ)dτ = − (mc + gϕ) 1 − 2 dt (1.108)
c
r
2 v2 ∂L (mc2 + gϕ) cv2
⇒ L = −(mc + gϕ) 1 − 2 ⇒ P = = q (1.109)
c ∂v 1− v2
c2
Here we can see that the Lagrangian of particle-field interaction and the expression for the particle’s
momentum is very much similar to the Lagrangian and the momentum of the particle in the absence of the
field where only the term (gϕ) is added to the existing mass of the particle. So at a particular instance of
space-time, we could infer that the particle is free of the field but has a mass of (m + gϕ). Thus this type
of Lagrangian has just affected the mass of the particle. This is one of the simplest examples which shows
how the presence of field and its interaction with the particle shifts the mass of the particle. So if the mass
of the particle was initially zero and then coupled with the field, then the value of the field would have
added an apparent mass for the particle. This is the basic idea of the Higgs field, which is a scalar field.
Where the interaction of the field with particles will bestow its mass. Let us now analyze the Lagrangian.
r
2 v2
L = −(mc + gϕ) 1 − 2 (1.110)
c
Let us first consider the particle moving in non-relativistic speeds, so now expanding the square root using
binomial theorem and neglecting the higher powers, we get,
!
v 2 1 v2
L = −(mc2 + gϕ) 1 − 2 = −mc2 + mv 2 − gϕ + gϕ (1.111)
2c 2 2c2
Now the term −mc2 is doesn’t affect the Lagrangian as it vanishes in the equation of motion as it is a
v2
constant. Then the factor 2 is very negligible in non-relativistic speeds, so we can neglect those two
2c
terms and thus the Lagrangian will be,
1 1
L = mv 2 − gϕ ⇒ L = Kinetic Energy − Potential energy ⇒ K.E = mv 2 and P.E = gϕ(r)
2 2
(1.112)
Where ϕ(r) is represented as the function of space. So the equation of motion now becomes,
∂L d ∂L dmv dP~
− =0 ⇒ −gϕ0 (r) − =0 ⇒ ~
= −g ∇ϕ (1.113)
∂r dt ∂v dt dt
If we consider the above case to be the motion of a charged particle of charge g in the presence of an electric
potential ϕ, then the final equation turns out to be the good old equation of motion of a charged particle
~
in presence of electric potential given by Fe = −g ∇ϕ.
14
1.6.2 Effect of the particle on the field
To know the effect of the particle on the field we shall consider a particle that is at rest and present in the
origin. We shall assume that the particle affects the field only where it is located. The Lagrangian density
for the field as we know is given by L in Eq.(1.34), without considering the potential term the action of
this system is given by
ˆ ˆ " !2 !2 !2 !2 # ˆ
4 1 1 ∂ϕ ∂ϕ ∂ϕ ∂ϕ 4 1
Action1 = L d x = − − − d x= ∂µ ϕ ∂ µ ϕ d4 x (1.114)
2 c2 ∂t ∂x ∂y ∂z 2
ˆ r
2
!
v
Now the action of a particle in space-time in presence of the field ϕ is, − (mc2 + gϕ) 1 − 2 dt
c
ˆ
Since the particle is at rest we get, Action = −(mc2 + gϕ)dt ⇒ L = −mc2 − gϕ (1.115)
Since adding a constant term of −mc2 to the Lagrangian doesn’t affect the equations of motions we can
neglect it. Thus we get the action of this system to be,
ˆ
Action2 = −gϕdt (1.116)
But for the same system, the actions are dependent on the different variables of integration. When consid-
ering Action1 of a field it depends on d4 x but when considering the Action2 of particle it depends on dt. So
we need to convert the Action2 to d4 x integral. For that we can do a subtle change in the action integral,
however since the field is being affected only at the location of the particle, we can use the Dirac-delta
function (δ(x)). The value of the function is zero for all x 6= 0 and the value of the function is 1 at x = 0.
So we can multiply three Dirac delta functions for three spatial coordinates, δ(x)δ(y)δ(x) which can be
written as δ 3 (x, y, z). Thus the Action now can be written as,
ˆ
Action2 = −gϕδ 3 (x, y, z)d4 x (1.117)
ˆ
1
The total action is given by, Action = Action1 + Action2 = ∂µ ϕ ∂ µ ϕ − gϕδ 3 (x, y, z) d4 x (1.118)
2
1
Thus the Lagrangian density is, L = ∂µ ϕ ∂ µ ϕ − gϕδ 3 (x, y, z) (1.119)
2
Now let us find the equation of motion for the field. So, according to Eq.(1.29) we get,
!
∂L ∂ ∂L
− µ = 0 ⇒ −gδ 3 (x, y, z) − ∂µ ∂ µ ϕ = 0 ⇒ ∂µ ∂ µ ϕ = −gδ 3 (x, y, z) (1.120)
∂ϕ ∂x ∂(∂µ ϕ)
15
1 ∂2ϕ ∂2ϕ ∂2 ∂2ϕ
So the equation of motion for the field is, − − − = −gδ 3 (x, y, z) (1.121)
c2 ∂t2 ∂x2 ∂y 2 ∂z 2
This equation is very much similar to the normal wave equation = 0 at every point except for the location
x = y = z = 0 i.e., the place where the particle resides. Thus at the location of the particle, the equation is
slightly complex to solve implying that the particle residing there is affecting the field. Assuming that the
particle being at rest at the origin doesn’t change the field with respect to time i.e., if ϕ is constant with
time we get,
If in case the particle is at some arbitrary point in space (a, b, c) then the equation becomes,
If the particle moves in space and the location of the particle is a function of time a(t), b(t), c(t), then ϕ
will change with time and the equation of motion for the field becomes,
2. Lagrangian Density has to abide by the "Principle of Locality". It means that the state of an object is
directly influenced by its immediate surroundings. Thus L depends only on the generalized coordinate
(q) or (ϕ) and its derivatives q̇ or ϕµ only.
16
Chapter 2
2.1 Electrostatics
Electrostatics is the study of the charges and their behaviours which are at rest.
1 Q ~
~ =
E r̂ ⇒ ~ =F
E (2.3)
4πε0 r2 q
If there are more than one charge q0 , q1 , q2 , q3 . . . qn which are all at rest then the Electric field at a point
P which is at a distance ri from the charge qi , then the net electric field at P is given by the expression,
n
~ = 1 X qi
E rˆi (2.4)
4πε0 r2
i=1 i
17
Electric Field lines of a +ve charge Electric Field lines of a −ve charge
~
E
θ
Electric flux is defined as the number of field lines that pass through a given area. If ~a is the area vector
and E~ is the electric field passing through the area, then the electric flux is given by ΦE = E ~ ·~a = Eacos(θ).
For a small area da, the small electric flux through it dΦE is given by dΦE = E ~ · d~a = Ecos(θ)da, then the
total flux is given by, " "
ΦE = ~ · d~a =
E E cos(θ)da (2.5)
S S
Gauss’s Law
Gauss’s law states that the net flux through a closed surface is equal to the net charge enclosed by that
surface times the reciprocal of permittivity of free space.
"
qnet ~ · d~a = qnet
net ΦE = ⇒ E (2.6)
ε0 S ε0
18
Differential form of Gauss’s law
" ˚
~ we know,
By Divergence theorem for some vector field V ~
V · d~a = ~ ·V
(∇ ~ )dτ
S V
" ˚
Applying this theorem to the electric flux given by Eq.(2.5), we get ~
E · d~a = (∇~ · E)dτ
~ (2.7)
S V
Also we know that the volume charge density is given by,
˚
dq
ρ= ⇒ dq = ρdτ ⇒ qnet = ρdτ (2.8)
dτ V
˚ ˚
~ · E)dτ
~ ρ
Substituting Eq.(2.7) and Eq.(2.8) in Eq.(2.6) we get, (∇ = dτ (2.9)
V V ε0
Since the integrals are the same, we can equate the integrands and it becomes, ∇ ~ = ρ
~ ·E (2.10)
ε0
∇ ~ = ρ
~ ·E ⇒ ~ · (−∇φ)
∇ ~ =
ρ
⇒ ∇2 φ = −
ρ
(2.13)
ε0 ε0 ε0
2.2 Magnetostatics
Magnetostatics refers to the study of magnetic fields arising due to steady currents where the current
remains a constant with respect to time.
19
2.2.2 Biot-Savart’s Law
We saw how a moving charge is affected by the presence of a magnetic field. But will a moving charge
create a magnetic field? The moving charge has to create a force so that it affects its path. And the law
which gives a quantitative explanation for this phenomenon is called as the Biot-Savart’s Law. According
to this law, the small magnetic field dB~ due to a steady current I in a small length element d~` at a point
P which is at a distance ~r from the length element at an angle θ is given by,
~ ˛
~ = µ0 Id`sin(θ) n̂ = µ0 Id` × ~r ~ = µ0 I d~` × ~r
dB 2
⇒ B (2.16)
4π r 4π r3 4π C r3
C
d~`
I r ~
dB
P
Where µ0 is the permeability of free space given by 4π × 10−7 NA−2 . d` is the small path in the curve ‘C’.
The cyclic integral form of the magnetic field in the Eq.(2.16) implies that the magnetic field lines are
always in the form of loops and there is no particular source and sink for the magnetic field. Thus there
are no magnetic monopoles. Thus the net magnetic flux through a closed surface will be always equal to
zero. Thus we get, ˚
net ΦB = ~ · B)dτ
(∇ ~ =0 ⇒ ~ ·B
∇ ~ =0 (2.18)
V
The above equation is similar to Gauss’s law for electrostatics, except that the divergence of magnetic field
here is zero. This law doesn’t have a particular name, some call it ‘Gauss’s law for magnetism’ or ‘ The
law of absence of magnetic mono poles’.
20
C
r
~
B
I d~`
˛
~ · d~` = µ0 I
B (2.20)
C
~ ×B
Since the integrals are the same, we can equate the integrands and it becomes, ∇ ~ = µ0 J~ (2.24)
~
Vector Potential is a Gauge Field which is not unique to B
~ 0 ) given by,
~→A
Vector potential is not unique magnetic field. Consider the ‘Gauge transformation’ (A
~0 = A
A ~ + ∇f
~ (f is some scalar field) (2.26)
~ 0 is,
Then the curl of A ∇ ~0 = ∇
~ ×A ~ × (A
~ + ∇f
~ )=∇
~ ×A
~+∇
~ × (∇f
~ )=∇
~ ×A
~=B
~ (2.27)
Thus A ~ 0 can be the vector
~ is invariant under Gauge transformation and hence it is a Gauge field. So even A
~ ~
potential of B. i.e., B can many vector potentials.
~ × (∇
∇ ~ × A)
~ = µ0 J~ (2.28)
~,
In vector calculus we have a relation that for any vector field V ~ × (∇
∇ ~ ×V
~ ) = ∇(
~ ∇~ ·V
~ ) − ∇2 V
~.
⇒ ~ × (∇
∇ ~ × A)
~ = ∇(
~ ∇~ · A)
~ − ∇2 A
~ = µ0 J~ (2.29)
~ = −µ0 J
∇2 A (2.30)
21
2.3 Electrodynamics
We have seen the physics of a stationary charge or a steady current (I = constant) whereas Electrodynamics
is the study of accelerated charges and changing currents (I 6= constant).
~
~ and B,
By combining Eq.(2.32) and Eq.(2.34) we get a relation between E ~ ~ = ∂B
~ ×E
∇ (2.35)
∂t
dq
Charge density inside, ρ = dτ
J~
Consider an arbitrary closed surface ‘S’ having charges inside. The volume charge density is given by ‘ρ’
and let the charges go outside the surface. Then there is a current I and its density J~ set up on the surface.
We know that current at the surface is the time rate of change of charge at the surface. Also, the charge
can be expressed as the volume integral of charge density according to Eq.(2.8). Thus we get,
˚ ˚ ∂ρ
dq d
I=− =− ρdτ = − dτ (2.36)
dt dt V V ∂t
The −ve sign implies that the charges are moving outwards and the current decreases as time increase.
" ˚
In Eq.(2.19) we got, I = ~
J · d~a . Using Gauss’s divergence theorem we get, I = ~ · J)dτ
(∇ ~ (2.37)
S V
22
Equating the above two expressions we get the two volume integrals to be equal, meaning that the integrands
are equal. So we get the equation of continuity of charge which is,
~ · J~ = − ∂ρ
∇ ⇒
∂ρ ~ ~
+∇·J =0 ⇒ ~ · J~ = 0
ρ̇ + ∇ (2.38)
∂t ∂t
∇ ~ = ρ
~ ·E (Gauss’s Law)
ε0
~ ·B
∇ ~ =0 (No Name)
~
~ ×E
∇ ~ = ∂B (Faraday’s Law)
∂t
~
∇ ~ = µ0 J~ + µ0 ε0 ∂ E
~ ×B (Ampere-Maxwell’s Law)
∂t
Maxwell’s equations can also be expressed in the integral forms,
"
~ · d~a = qnet
E (Gauss’s Law)
ε0
"S
~ · d~a = 0
B (No Name)
S
˛ "~
~ · d~` = ∂B
E · d~a (Faraday’s Law)
C S ∂t
˛ " ~
~ ~ ∂E
B · d` = µ0 I + µ0 ε0 · d~a (Ampere-Maxwell’s Law)
C S ∂t
23
Chapter 3
~
J µ = (J 0 , J 1 , J 2 , J 3 ) = (cρ, Jx , Jy , Jz ) = (cρ, J)
Jµ = (J0 , J1 , J2 , J3 ) = (cρ, −Jx , −Jy , −Jz ) = (cρ, −J) ~ (3.6)
24
Four Dimensional Equation of Continuity
~ ~ ~
~ = − ∂ B = − ∂ (∇ ~ × − ∂A ~ = − ∂A
~ ×E
∇ ~ × A)
~ =∇ ⇒ E (3.11)
∂t ∂t ∂t ∂t
Thus the expression for both time varying and time constant electric field is their sum given by,
~ ∂Ay
~ = − ∂ A − ∇φ
E ~ ⇒ E1 = Ex =
∂Ax ∂φ
− , E2 = Ey = −
∂φ
, E3 = Ez =
∂Az
−
∂φ
(3.12)
∂t ∂t ∂x ∂t ∂y ∂t ∂z
~ in terms of A
By Eq.(2.25) we get the expression for B ~ as B ~ =∇ ~ ×A~.
! ! !
~ = ∂A z ∂A y ∂A z ∂A x ∂A y ∂A x
B − î − − ĵ + − k̂ (3.13)
∂y ∂z ∂x ∂z ∂x ∂y
Let the gauge transformation of ‘Aµ ’ be, A0µ = Aµ + ∂µ f , then the various derivatives of A0µ are given by,
∂ν A0µ = ∂ν Aµ + ∂ν ∂µ f (3.14)
Now we don’t know whether ∂ν ∂µ f is gauge invariant, but there is an easy way to eliminate that factor.
We could define, A0ν = Aν + ∂ν f and take the various derivatives, we get,
∂µ A0ν = ∂µ Aν + ∂µ ∂ν f (3.15)
Hence the quantity ∂µ Aν − ∂ν Aµ is gauge invariant and it is nothing but the components of the field tensor.
We can add all the components of the electric and magnetic field in field tensor Fµν which relates with Aµ
according to the equation,
Fµν = ∂µ Aν − ∂ν Aµ (3.17)
25
Which is a 4 × 4 tensor with 16 elements in it. Clearly if µ = ν i.e., the diagonal elements are 0 and When
the column and row indices are changed, the sign of the element changes. Hence it is Anti-symmetric.
F µν = η µα Fαβ η βν (3.20)
where η µα Fαβ raises the index µ and it becomes Fβµ and then by multiplying it by η βν we get, Fβµ η βν = F µν
Put µ = 1 and ν =2
∂Ay ∂Ax ∂A
y ∂Ax
F12 = ∂1 A2 − ∂2 A1 = − + =− − = −Bz (3.24)
∂x ∂y ∂x ∂y
Put µ = 1 and ν =3
∂Az ∂Ax ∂A
z ∂Ax
F13 = ∂1 A3 − ∂3 A1 = − + =− − = By (3.25)
∂x ∂z ∂x ∂z
Put µ = 2 and ν =3
∂Az ∂Ay ∂A
z ∂Ay
F23 = ∂2 A3 − ∂3 A2 = − + =− − = −Bx (3.26)
∂y ∂z ∂y ∂z
~ and B
Thus we got all the components of the Field tensor which comprises of the components of E ~ arranged
in an Anti-Symmetric manner given by,
0 Ex /c Ey /c Ez /c 0 −Ex /c −Ey /c −Ez /c
−E /c 0 −Bz By E /c 0 −Bz By
x x
µν
Fµν = F = (3.27)
−Ey /c Bz −Bx −Bx
0 Ey /c Bz 0
−Ez /c −By Bx 0 Ez /c −By Bx 0
~ →B
The dual of the covariant field tensor is given by Gµν and Gµν got by replacing E/c ~ and B
~ → −E/c,
~
0 Bx By Bz 0 −Bx −By −Bz
−B 0 Ez /c −Ey /c B 0 Ez /c −Ey /c
x x
Gµν
Gµν = = (3.28)
−By −Ez /c By −Ez /c
0 Ex /c 0 Ex /c
−Bz Ey /c −Ex /c 0 Bz Ey /c −Ex /c 0
26
Inner product of Field tensors
The inner product of field tensor is given by Fµν F µν
Fµν = ∂µ Aν − ∂ν Aµ (3.31)
µν µ ν ν µ
F =∂ A −∂ A (3.32)
Thus the Lagrangian density of EM field can have the combinations of Fµν and F µν . But we also should
make sure that the Lagrangian density is also Lorentz Invariant. Thus, it should contain terms where there
are both upper and lower indices. Thus the Lagrangian density can be,
1 1
L=− Fµν F µν = − (∂µ Aν − ∂ν Aµ )(∂ µ Aν − ∂ ν Aµ ) (3.33)
4µ0 4µ0
1
The factor − is just for conventional purposes and to preserve the dimensions of units. But we have
4µ0
already seen the value of Fµν F µν in Eq.(3.29), thus L can be expanded as,
! !
1 2 | ~ 2
E| 1 1 ~ 2 1 ~ 2
L=− µν
Fµν F = − ~ −
|B|2
= ~ − |B|
ε0 µ0 |E|2 ~ 2
= ε0 |E| − |B| (3.34)
4µ0 4µ0 c2 2µ0 2 2µ0
We know that L is the difference between the Kinetic energy density and Potential energy density. Thus,
27
Here we see nothing about currents or charges. The physics here is only of Electric and Magnetic fields in
the absence of Charges or currents. For a Lagrangian density L = L(Aν , ∂µ Aν ) the equations of motion
are,
!
∂L ∂ ∂L
− µ =0 (3.38)
∂Aν ∂x ∂(∂µ Aν )
So if we substitute the expression of Lagrangian density of Eq.(3.33) in the above equation we get,
!
∂ 1 µ ν ν µ
∂ ∂ 1 µ ν ν µ
− (∂ A
µ ν − ∂ A
ν µ )(∂ A − ∂ A ) − µ − (∂ A
µ ν − ∂ A
ν µ )(∂ A − ∂ A ) =0
∂Aν 4µ0 ∂x ∂(∂µ Aν ) 4µ0
!
µ st ∂ ∂ 1
µ ν ν µ
L doesn’t depend on A , So 1 term = 0. ∴ − (∂ µ A ν −∂ ν Aµ )(∂ A −∂ A ) =0
∂xµ ∂(∂µ Aν ) 4µ0
∂ 1 µ ν ν µ
∂ µν
⇒ (∂ A − ∂ A ) =0 ⇒ F =0 ⇒ ∂µ F µν = 0 (3.39)
∂xµ 4 ∂xµ
Using F µν given in Eq.(3.27), let us try finding the equations of ∂µ F µν = 0, when this is expanded we get,
∂µ F µν = 0 ⇒ ∂0 F 0ν + ∂1 F 1ν + ∂2 F 2ν + ∂3 F 3ν = 0 (3.40)
1∂ ∂ ∂ ∂
⇒ (F 0ν ) + (F 1ν ) + (F 2ν ) + (F 3ν ) = 0 (3.41)
c ∂t ∂x ∂y ∂z
Put ν = 0
1∂ ∂ ∂ ∂
(F 00 ) + (F 10 ) + (F 20 ) + (F 30 ) = 0 (3.42)
c ∂t ∂x ∂y ∂z
1∂ ∂ Ex ∂ Ey ∂ Ez
⇒ (0) + + + =0 (3.43)
c ∂t ∂x c ∂y c ∂z c
1~ ~ ~ ·E
~ =0
⇒ ∇·E =0 ⇒ ∇ (3.44)
c
Put ν = 1
1∂ ∂ ∂ ∂
(F 01 ) + (F 11 ) + (F 21 ) + (F 31 ) = 0 (3.45)
c ∂t ∂x ∂y ∂z
1 ∂ Ex ∂ ∂ ∂
⇒ − + (0) + (Bz ) + (−By ) = 0 (3.46)
c ∂t c ∂x ∂y ∂z
∂Bz ∂By 1 ∂Ex ~ x = µ0 ε0 ∂Ex
~ × B)
⇒ − = 2 ⇒ (∇ (3.47)
∂y ∂z c ∂t ∂t
Put ν = 2
1∂ ∂ ∂ ∂
(F 02 ) + (F 12 ) + (F 22 ) + (F 32 ) = 0 (3.48)
c ∂t ∂x ∂y ∂z
1 ∂ Ey ∂ ∂ ∂
⇒ − + (−Bz ) + (0) + (Bx ) = 0 (3.49)
c ∂t c ∂x ∂y ∂z
∂Bx ∂Bz 1 ∂Ey ~ y = µ0 ε0 ∂Ey
~ × B)
⇒ − = 2 ⇒ (∇ (3.50)
∂z ∂x c ∂t ∂t
28
Put ν = 3
1∂ ∂ ∂ ∂
(F 03 ) + (F 13 ) + (F 23 ) + (F 33 ) = 0 (3.51)
c ∂t ∂x ∂y ∂z
1 ∂ Ez ∂ ∂ ∂
⇒ − + (By ) + (−Bx ) + (0) = 0 (3.52)
c ∂t c ∂x ∂y ∂z
∂By ∂Bx 1 ∂Ez ~ z = µ0 ε0 ∂Ez
~ × B)
⇒ − = 2 ⇒ (∇ (3.53)
∂x ∂y c ∂t ∂t
~
~ ·E
Hence, for ν = 0, we got, ∇ ~ = 0 and adding the equations got by ν = 1, 2, 3 we get, ∇
~ ×B~ = µ 0 ε0 ∂ E
∂t
By these two equations of motion we can understand the physics of Electro-magnetic waves in the absence
of charges and currents. Which can be modified to get the wave equations,
1 ∂2E~ 1 ∂2E~
For Electric component of wave ~ =
∇2 E ⇒ ~ =0
− ∇2 E ⇒ ~ =0
E (3.54)
2 2
c ∂ t 2 2
c ∂ t
1 ∂2B~ 1 ∂2B~
~ =
For Magnetic component of wave ∇2 B ⇒ ~ =0
− ∇2 B ⇒ ~ =0
B (3.55)
2 2
c ∂ t 2 2
c ∂ t
As we know that at a far away distance from the source J = 0, so using the above technique in Eq.(3.57),
ˆ ˆ ˆ
1
J µ A0µ d4 x = J µ Aµ d4 x − f (∂µ J µ ) d4 x (3.58)
e
By the tensor notation of equation of continuity Eq.(3.10) we know that, ∂µ J µ = 0, thus the second integral
of the above equation becomes zero. Hence we get,
ˆ ˆ
µ 0 4
J Aµ d x = J µ Aµ d4 x ⇒ J µ Aµ is guage invariant (3.59)
Thus the term J µ Aµ is both gauge and Lorentz invariant which can be put in the Lagrangian density of EM
field. The term J µ Aµ is the interaction Lagrangian of the electromagnetic field and the full EM Lagrangian
is,
1
LEM = − Fµν F µν − J µ Aµ (3.60)
4µ0
29
Thus applying the Euler-Lagrange equations of motion, in Eq.(3.33) saw that L wasn’t dependent on four
potential rather it depended on the derivatives of four potential. But now we modified the Lagrangian to
LEM which depends even on the four potential. Thus we get the equation of motion to be,
!
∂LEM ∂ ∂LEM
− µ =0 (3.61)
∂Aν ∂x ∂(∂µ Aν )
!
∂ 1 ∂ ∂ 1
⇒ − Fµν F µν − J µ Aµ − µ − Fµν F µν − J µ Aµ =0
∂Aν 4µ0 ∂x ∂(∂µ Aν ) 4µ0
∂ µ ∂ 1 µν 1
⇒ − ν
(J A µ ) − µ
− F = −J ν + ∂µ F µν = 0
∂A ∂x µ0 µ0
∂µ F µν = µ0 J ν ⇒ ∂0 F 0ν + ∂1 F 1ν + ∂2 F 2ν + ∂3 F 3ν = µ0 J ν (3.63)
1∂ ∂ ∂ ∂
⇒ (F 0ν ) + (F 1ν ) + (F 2ν ) + (F 3ν ) = µ0 J ν (3.64)
c ∂t ∂x ∂y ∂z
Put ν = 0
1∂ ∂ ∂ ∂
(F 00 ) + (F 10 ) + (F 20 ) + (F 30 ) = µ0 J 0 (3.65)
c ∂t ∂x ∂y ∂z
1∂ ∂ Ex ∂ Ey ∂ Ez
⇒ (0) + + + = µ0 (cρ) (3.66)
c ∂t ∂x c ∂y c ∂z c
⇒ ~ ·E
∇ ~ = µ0 c2 ρ ⇒ ∇ ~ = ρ
~ ·E (3.67)
ε0
Put ν = 1
1∂ ∂ ∂ ∂
(F 01 ) + (F 11 ) + (F 21 ) + (F 31 ) = µ0 J 1 (3.68)
c ∂t ∂x ∂y ∂z
1 ∂ Ex ∂ ∂ ∂
⇒ − + (0) + (Bz ) + (−By ) = µ0 Jx (3.69)
c ∂t c ∂x ∂y ∂z
∂Bz ∂By 1 ∂Ex ~ x = µ0 Jx + µ0 ε0 ∂Ex
~ × B)
⇒ − = µ0 Jx + 2 ⇒ (∇ (3.70)
∂y ∂z c ∂t ∂t
Put ν = 2
1∂ ∂ ∂ ∂
(F 02 ) + (F 12 ) + (F 22 ) + (F 32 ) = µ0 J 2 (3.71)
c ∂t ∂x ∂y ∂z
1 ∂ Ey ∂ ∂ ∂
⇒ − + (−Bz ) + (0) + (Bx ) = µ0 Jy (3.72)
c ∂t c ∂x ∂y ∂z
∂Bx ∂Bz 1 ∂Ey ~ × B)
~ y = µ 0 J y + µ 0 ε0 ∂Ey
⇒ − = µ0 Jy + 2 ⇒ (∇ (3.73)
∂z ∂x c ∂t ∂t
30
Put ν = 3
1∂ ∂ ∂ ∂
(F 03 ) + (F 13 ) + (F 23 ) + (F 33 ) = µ0 J 3 (3.74)
c ∂t ∂x ∂y ∂z
1 ∂ Ez ∂ ∂ ∂
⇒ − + (By ) + (−Bx ) + (0) = µ0 Jz (3.75)
c ∂t c ∂x ∂y ∂z
∂By ∂Bx 1 ∂Ez ~ z = µ0 Jz + µ0 ε0 ∂Ez
~ × B)
⇒ − = µ0 J z + 2 ⇒ (∇ (3.76)
∂x ∂y c ∂t ∂t
Thus the tensor equation ∂µ F µν = µ0 J ν gives rise to the non homogeneous Maxwell’s equations. For ν = 0
we got Maxwell’s First equation (Gauss’s law). And by adding the equations got by ν = 1, 2, 3 we get
Maxwell’s 4th equation (Ampere-Maxwell’s equation).
~
∇ ~ = ρ
~ ·E and ~ = µ0 J~ + µ0 ε0 ∂ E
~ ×B
∇
ε0 ∂t
~
~ ·B
∇ ~ =0 and ∇ ~ + ∂B = 0
~ ×E (3.77)
∂t
Even they can be put in the tensor notations. But for that, we need to use the Dual of field tensor (Gµν ).
Using Gµν given in Eq.(3.28), let us expand the above equation and find out the Maxwell’s equations,
1∂ ∂ ∂ ∂
⇒ (G0ν ) + (G1ν ) + (G2ν ) + (G3ν ) = 0 (3.80)
c ∂t ∂x ∂y ∂z
Put ν = 0
1∂ ∂ ∂ ∂
(G00 ) + (G10 ) + (G20 ) + (G30 ) = 0 (3.81)
c ∂t ∂x ∂y ∂z
1∂ ∂ ∂ ∂
⇒ (0) + Bx + By + Bz = 0 (3.82)
c ∂t ∂x ∂y ∂z
∴ ~ ·B
∇ ~ =0 (3.83)
Put ν = 1
1∂ ∂ ∂ ∂
(G01 ) + (G11 ) + (G21 ) + (G31 ) = 0 (3.84)
c ∂t ∂x ∂y ∂z
1∂ ∂ ∂ Ez ∂ Ey
⇒ − Bx + (0) + − + =0 (3.85)
c ∂t ∂x ∂y c ∂z c
1 ∂Bx 1 ∂Ez ∂Ey ~ × E)
~ x+ ∂Bx
⇒ − − − =0 ⇒ (∇ =0 (3.86)
c ∂t c ∂y ∂z ∂t
31
Put ν = 2
1∂ ∂ ∂ ∂
(G02 ) + (G12 ) + (G22 ) + (G32 ) = 0 (3.87)
c ∂t ∂x ∂y ∂z
1∂ ∂ Ez ∂ ∂ Ex
⇒ − By + + (0) + − =0 (3.88)
c ∂t ∂x c ∂y ∂z c
1 ∂By 1 ∂Ex ∂Ez ~ × E)~ y + ∂By = 0
⇒ − − − = 0 ⇒ (∇ (3.89)
c ∂t c ∂z ∂x ∂t
Put ν = 3
1∂ ∂ ∂ ∂
(G03 ) + (G13 ) + (G23 ) + (G33 ) = 0 (3.90)
c ∂t ∂x ∂y ∂z
1∂ ∂ Ey ∂ Ex ∂
⇒ − Bz + − + + (0) = 0 (3.91)
c ∂t ∂x c ∂y c ∂z
1 ∂Bz 1 ∂Ey ∂Ex ~ × E)~ z + ∂Bz = 0
⇒ − − − = 0 ⇒ (∇ (3.92)
c ∂t c ∂x ∂y ∂t
Hence for ν = 0 we got Maxwell’s second equation (no name). And by adding the equations got by
ν = 1, 2, 3 we get Maxwell’s 3rd equation (Faraday’s Law) given in Eq.(3.77).
F~ = F~e + F~m = q E
~ + q(~v × B)
~ = q(E
~ + ~v × B)
~ (3.93)
Let us derive this equation by the action principle through covariant formulation.
For a particle with mass m moving in space-time we know that the action is given by,
ˆ
Action1 = −mc2 dτ (3.94)
Let that particle be charged with a charge q and move in a region of space-time filled with a four-vector
potential field Aµ which is a function of t, x, y, and z. So the small change of four potential the particle
experiences as it moves a small distance dxµ is given by Aµ dxµ . Hence the action because of the vector
potential is given by, ˆ
Action2 = qAµ dxµ (3.95)
32
ˆ ˆ r
2 µ 2 (ẋi )2 dxµ
The net action is given by, Action = −mc dτ − qAµ dx = − mc 1− − qAµ dt (3.96)
c2 dt
Thus the Lagrangian is given by,
r r
(ẋi )2 dxµ (ẋi )2 cdt dxi
L = −mc2 1 − 2 − qAµ = −mc2 1 − 2 − qA0 − qAi (3.97)
c dt c dt dt
r
(ẋi )2
⇒ L = −mc2 1 − 2 − q cA0 + Ai ẋi (3.98)
c
Since xi are spatial coordinates we get, xi = (x, y, z) = −xi . The Euler Lagrange equations for a generalized
coordinate xi and generalized vector ẋi are given by,
! !
∂L d ∂L ∂L d mẋi
− =0 ⇒ − − qAi = 0 (3.99)
∂xi dt ∂ ẋi ∂xi dt
q
i 2
1 − (ẋc2)
!
d mẋi dAi ∂L dP i ∂L dAi
⇒ +q = ⇒ = Fi = −q (3.100)
∂xi ∂xi
q
dt (ẋi )2 dt dt dt
1− c2
dAi ∂Ai ∂Ai dx ∂Ai dy ∂Ai dz dAi ∂Ai dxj ∂Ai ∂Ai j
= + + + = + j
= + ẋ
dt ∂t ∂x dt ∂y dt ∂z dt dt ∂x dt ∂t ∂xj
∂L ∂A
i ∂Ai j
⇒ Fi = − q + ẋ (3.101)
∂xi ∂t ∂xj
Let us find now the derivative of the Lagrangian with respect to xi ,
∂L ∂A
0 ∂Aj j
= −q c + ẋ (3.102)
∂xi ∂xi ∂xi
Thus we get the F i to be
!
∂A
0 ∂Aj j ∂A
i ∂Ai j ∂A0 ∂Ai ∂Ai j ∂Aj j
F i = −q c i + i
ẋ − q + j
ẋ = −q c i + + ẋ + ẋ (3.103)
∂x ∂x ∂t ∂x ∂x ∂t ∂xj ∂xi
We know that,
~
∂A
~ = −∇(cA
E ~ 0
)− and ~ =∇
B ~ ×A
~ (3.105)
∂t
So we get,
~ × B)
F i = q E i + (V ~ i ⇒ F~ = q(E
~ + ~v × B)
~ (3.106)
33