Electromagnetism

Luis Fernando Alday

Chapters

1.- Introduction to Electrostatics

2.- Boundary-value problems in electrostatics
3.- Magnetostatics
4.- Time dependent electromagnetism
5.- Electromagnetic waves
6.- Epilogue: Electromagnetism and special relativity

Reading list
R. Feynman, Lectures in Physics, Vol.2. Electromagnetism, Addison Wesley.
J.D. Jackson, Classical Electrodynamics, John Wiley.
Landau and Lifshitz, The Classical Theory of Fields (to read if you are brave enough!)
1 Introduction to Electrostatics
This is the subject of static electricity, which we’ve all played with: recall rubbing a pen on
your cat and picking up bits of paper. We interpret this as charging up or adding charge to
the pen. As greeks didn’t have pens, they used amber. The greek word for amber is electron
and this is where the subject gets its name from.

1.1 Coulomb’s law

We will assume the following facts about charge:

• there exist charges in nature;

• charge can be positive or negative;
• charge cannot be created or destroyed;

For a more sophisticated standpoint, we may know that charge is discrete, and that electrons
have charge -1 and protons charge +1 in suitable units, but for our purposes we want to
think of charge as continuous. As a mathematical idealization, we represent charge by a
real function, the charge density ρ(x, y, z, t), with the property that the total charge Q in a
region V of space is
Z ZZZ
Q= ρdV = ρ dxdydz. (1)
V V
Other idealizations are possible, and we shall sometimes use them. For example, it may be
convenient to think of charges residing on a surface, something two-dimensional, and then
charge surface density σ(x, y, t) is charge per unit area, so that the total charge on a piece
S of surface is
Z
Q= σdS; (2)
S
similarly, for a one-dimensional distribution of charge, charge line density is charge per unit
length on some curve. Finally, a point charge q corresponds to a finite, nonzero charge
localized at a point.

After introducing the concept of charge, the natural question is: how do charges interact
with each other? An important experimental fact about point charges is the
Coulomb’s Law: Given two point particles P, P 0 with charges q, q 0 at positions ~r, ~r 0 , the
electric force on P due to P 0 is

1 ~r − ~r 0
F~ = qq 0 (3)
4π0 |~r − ~r 0 |3
where 0 is a constant and accounts for the correct units, see figure.

2
 P’

r−r’
r

P
r
0

We treat this law as an experimental fact. To see how one might have discovered it, note
that:

• it is an inverse-square law, acting along the line joining the particles;

• it is proportional to the product of the charges, so that like charges repel and unlike
charges attract.

1.2 Electric field and scalar potential

It is convenient to think of Coulomb’s law as follows: P 0 generates an electric field to which
P responds. Put P 0 at the origin and define the electric field at ~r to be

~ r) = 1 0 ~r
E(~ q (4)
4π0 r3
Then P , which has charge q, when placed in the field generated by P 0 is subject to (or ”feels”)
the force

F~ = q E
~ (5)
Replace P 0 by various particles P1 , P2 , ..., Pn , at points ~r1 , ~r2 , ..., ~rn with charges q1 , q2 , ..., qn .
The forces will add as vectors so if we defined the electric field now as

n
~ r) = 1 X qi (~r − ~ri )
E(~ , (6)
4π0 i=1 |~r − ~ri |3
then, as before, P placed at ~r is subject to the force

F~ (~r) = q E(~
~ r) (7)
The electric field (6) generalizes easily to the electric field for a volume distribution of charge
density

(~r − ~r 0 ) 0 0 0
Z
~ r) = 1
E(~ ρ(~r 0 ) dx dy dz , (8)
4π0 |~r − ~r 0 |3

3
It is instructive to ask how (8) reduces to (6). In order to answer that question we need
to understand which charge density ρ(x, y, z) corresponds to a point charge of magnitude q
located at (x0 , y0 , z0 ). This function should be such that it vanishes identically outside the
point (x0 , y0 , z0 ) but at the same time
Z
ρ(x, y, z)dV = q (9)
V

for any volume including the point (x0 , y0 , z0 ). We introduce the one-dimensional Dirac delta
function δ(x), defined by the following properties:

1. δ(x) = 0, for x 6= 0
R
2. δ(x)dx = 1, if the region of integration includes x = 0, and is zero otherwise.

Useful properties, which can be derived from the two above are the following:

Z
f (x)δ(x − a)dx = f (a), if the region of integration includes x = a, (10)
X 1
δ(f (x)) = δ(x − xi ), with f (xi ) = 0. (11)
i
|f 0 (x i )|

In higher dimensions the Dirac delta function is just the product of the cartesian delta
functions. We see that the volume charge density with the correct properties to correspond
to a point charge of magnitude q located at (x0 , y0 , z0 ) is given by

ρ(x, y, z) = qδ(~r − ~r0 ) ≡ qδ(x − x0 )δ(y − y0 )δ(z − z0 ) (12)

P
By using the properties of the delta function you can check that for ρ(~r) = qi δ(~r − ~ri ) (8)
reduces to the potential (6) for a collection of particles.

We want to explore (6). Note first that it is the gradient of another function:
n
~ = −∇Φ, 1 X qi
E Φ= . (13)
4π0 i=1 |~r − ~ri |

This scalar function Φ is called the electric or scalar potential. The scalar potential has a
very nice physical interpretation. Let us think about Newton’s equations of motion for a
charged particle of mass m and charge q subject to the force (7)

m~a = F~ = q E
~ = −q∇Φ (14)

Then

4
 3
∂Φ dxi
 
d 1 X dΦ
m~v · ~v = m~v · ~a = −q~v · ∇Φ = −q i
= −q (15)
dt 2 i=1
∂x dt dt
so that
1
E = m|~v |2 + qΦ = constant (16)
2
In the motion of the charged particle, the kinetic energy changes with time, but the sum
of the kinetic energy and qΦ is constant; thus qΦ is the potential energy of the particle of
charge q in the electric field with potential Φ. 1

1.3 Gauss law and Poisson equation

In the previous section we have seen that the electric field for a collection of particles is the
gradient of a scalar function. In particular, therefore

~ = 0,
∇∧E (17)
From (6) we can also explicitly check

~ = ∇2 Φ = 0,
∇·E except at ~r = ~ri (18)

Consider the diagram

large sphere S

S1
S3
P1
P3

S2
V
P2

i.e. S is a large sphere containing all Pi , Si is a small sphere containing only Pi and V̂ is the
region in between. Then ∇ · E ~ = 0 on V̂ , so
1
Note that this is completely analogue to the usual (gravitational) potential energy, the mass m is replaced
by the charge q and the potential gh is replaced by the potential Φ.

5
 Z n Z
X Z
~ · dS −
E ~ · dS =
E ~
∇ · EdV =0 (19)
S i=1 Si V̂

the first two are surface integrals (with dS pointing away from the center of the corresponding
sphere), while the last one is a volume integral. Make sure you understand the signs in this
equation. Let us focus in the integral over S1 and consider only the charge P1 . Taking ~r1 = 0
for simplicity we obtain:

Z Z
~ · dS = 1 ~r
E q1 3 · ndS
S1 S1 4π0 r
Z
1 1 2
= q1 r sin θdθdφ
4π0 r2
1
= q1 ,
0

in the second step we used n = ~rr . Now, the divergence of the electric field generated by
the other charges vanishes identically inside S1 , so we can use the divergence theorem to
conclude that they will not contribute to the integral above. Each term in the sum (19) can
be calculated like this, so we obtain:
Z n Z
~ · dS = 1 (q1 + q2 + ... + qn ) = 1 × total charge inside S
X
~ · dS =
E E (20)
S i=1 Si 0 0
This is the Gauss’s Law in its integral form:
Gauss’s Law: The flux of E ~ out of V = 1 × total charge in V .
0
If we have a smoothed out charge density instead of point charges, Gauss’s law would read
Z Z
~ · dS = 1
E ρdV, (21)
S 0 V
where V is the volume inside the closed surface S. We will assume this to be true.
Now, by Divergence theorem we have
Z  
~ 1
∇ · E − ρ dV = 0 (22)
V 0
but if this is to hold for all possible regions V , then

∇·E ~ = 1 ρ, (23)
0
which is the differential version of Gauss’s Law.
This together with E ~ = −∇Φ implies Poisson’s equation

6
 1
∇2 Φ = − ρ (24)
0
What is the solution for this equation? remember that for point particles we had (see (13))

n
1 X qi
Φ= (25)
4π0 i=1 |~r − ~ri |
so we might guess that the solution of (24) is

ρ(~r 0 )
ZZZ
1
Φ(~r) = 0
dx0 dy 0 dz 0 (26)
4π0 V |~r − ~r |
where V is the region in which ρ 6= 0. In order check that this is the correct guess, we need
to apply the Laplacian to both sides. In doing so we are led to consider

1 d2
   
21 1
∇ = 2
r. = 0, for r 6= 0 (27)
r r dr r
At r = 0 we need to be a little bit careful. Indeed, integrating around a little sphere around
the origin we find
Z   Z   Z  
2 1 1 1
∇ dV = ∇·∇ dV = n·∇ ds = −4π (28)
V r V r S r
hence we arrive to the following beautiful equation
 
1
∇ 2
= −4πδ(~r − ~r 0 ) (29)
|~r − ~r 0 |
using this it is immediate to check that (26) is a solution to the Poisson equation. Incidentally,
also note that (29) implies the Poisson equation also works for point charges, where the charge
density has the form (12).

An example
~ given the charge distribution. We have
The main problem of Electrostatics is to obtain E
solved that problem with (26): Given the density ρ, integrate this to find Φ and then
~ = −∇Φ. However, in simple cases we can go straight to E
E ~ by using (4). Here is an
example:
Total charge Q is spread out uniformly round a plane circular wire of radius a. Find the
electric field at a point P on the axis of the circle, at a distance b from the centre.

7
 a e
unit vectors
P α
k
L

~ =
We can do this one directly from the inverse square law: E q ~ r
.
4π0 r3
Cut the circular wire into elements, each of length adθ; then each contains charge equal to
Q
dθ and so contributes 2πQ 1 ~e
dθ 4π 2 , (with ~
~ at P . Adding these up
e as in the figure) to E
2π 0 L

around the circle leads, by symmetry, to a vector along ~k. Note, from the diagram, that
~e · ~k = cos α, so that E(p)
~ = E~k, with E a scalar and given by

Z
Q 1 cos α
E = · dθ · ·
2π 4π0 L2
Q cos α
=
4π0 L2
Q b
=
4π0 (a + b2 )3/2
2

which is the answer.

~ for a point P 0 not on the axis.
It is trickier to find E

Another example: Discontinuity as you cross a surface density

Suppose we want to compute the discontinuity in the electric field as we cross a surface layer
of charge, with surface density σ(x, y), see figure.

8
In order to answer this question we can use the Gauss law in its integral form (21) for a
volume similar to a coin (as shown in the figure), with one face on each side of the surface
and a negligible thickness. We obtain
Z Z Z Z Z
~ · dS =
E ~ 2 · ~ndS −
E ~ 1 · ~ndS = 1
E ρdV =
1
σ(x, y)dS (30)
S S S 0 V 0 S

where ~n is the normal to the surface. Hence we have a discontinuity in the normal component
of the electric field as we cross the layer
 
0 E~2 − E
~ 1 · ~n = σ(x, y) (31)

~ = 0 together with Stokes theorem to show that the tangential

You can now try to use ∇ ∧ E
components are continuous.

1.4 Boundary problems, Green’s functions

If electrostatic problems always involved localized distributions of charge with no boundary
surfaces, the solution would be simply given by (26) or (8) and this would be an eight hours
course :-). Unfortunately, many problems involve finite regions of space with prescribed
boundary conditions on the bounding surfaces. There are two natural boundary conditions:

• Dirichlet boundary conditions: The potential Φ is specified on a boundary surface.

∂Φ
• Neumann boundary conditions: The normal derivative of the potential ∂n
= ~n · ∇Φ is
specified on a boundary surface.

To handle such problems we will make use of the Green’s theorem:

Z Z  
2

2 ∂ψ ∂φ
φ∇ ψ − ψ∇ φ dV = φ −ψ dS (32)
V S ∂n ∂n
∂
where φ and ψ are two arbitrary scalar fields, S is the boundary of V and ∂n
is the normal
derivative at the surface S, directed outwards from inside the volume V .
In order to exploit this theorem in the context of electrostatics we introduce the concept of
Green’s function. A Green’s function is a function G(~r, ~r 0 ) which satisfies

∇02 G(~r, ~r 0 ) = −4πδ(~r − ~r 0 ) (33)

Of course, the solution to that equation is not unique, but we know the general solution is
of the form:
1
G(~r, ~r 0 ) = + F (~r, ~r 0 ) (34)
|~r − ~r 0 |

where F (~r, ~r 0 ) satisfies the Laplace’s equation inside the volume V

9
 ∇02 F (~r, ~r 0 ) = 0 (35)
we will use the freedom to choose F (~r, ~r 0 ) momentarily. Next, let us consider Green’s theorem
with φ = Φ(~r 0 ) the scalar potential and ψ = G(~r, ~r 0 ) the Green’s function. We obtain

r, ~r 0 ) r 0)
Z Z  
0 ∂G(~ 0 ∂Φ(~
Φ(~r )∇ G(~r, ~r ) − G(~r, ~r )∇ Φ(~r ) dV 0 =
0 02 0 0 02 0
dS 0(36)


Φ(~r ) 0
− G(~r, ~r ) 0
V S ∂n ∂n

Using (33) and the properties of the delta function we obtain

Z
Φ(~r 0 )∇02 G(~r, ~r 0 )dV 0 = −4πΦ(~r) (37)
V
where we have assumed that the observation point ~r is inside the volume V . In this case we
obtain

r 0) r, ~r 0 )
Z Z  
1 0 0 1 0 0 ∂Φ(~ 0 ∂G(~
Φ(~r) = ρ(~r )G(~r, ~r )dV + G(~r, ~r ) 0
− Φ(~r ) 0
dS 0 (38)
4π0 V 4π S ∂n ∂n

Now we can use the freedom in the definition of the Green’s function. If we are solving a
problem with Dirichlet boundary conditions we demand

GD (~r, ~r 0 ) = 0, for ~r 0 in S. (39)

Once we have found a Green’s function with these boundary conditions, the solution for the
potential is simply given by

r, ~r 0 )
Z Z  
1 0 1 0 0 0 ∂GD (~
Φ(~r) = ρ(~r )GD (~r, ~r )dV − Φ(~r ) 0
dS 0 (40)
4π0 V 4π S ∂n
For Neumann boundary conditions we must be more careful. We cannot require the normal
derivative of G to vanish: application of the Gauss theorem to the definition of Green’s
function gives

∂G(~r, ~r 0 ) 0
Z
ds = −4π (41)
S ∂n
The easiest condition we can require is

∂GN (~r, ~r 0 ) 4π
=− (42)
∂n A
where A is the total surface area of S. Then the solution is

∂Φ(~r 0 ) 0
Z Z
1 0 1 0 0
Φ(~r) = hΦiS + ρ(~r )GN (~r, ~r )dV + GN (~r, ~r 0 ) dS (43)
4π0 V 4π S ∂n0

10
where hΦiS is the average value of the potential over the whole surface. In the case of an
”exterior problem” the volume is bounded by two surfaces, one finite and closed and the
other one at infinity. In this case the surface area diverges and hΦiS vanishes.
Let us add that for both boundary conditions we can always choose the Green’s function to
be symmetric under the interchange of ~r and ~r 0 .
In practice it can be quite hard to find the Green’s function with the appropriate boundary
conditions. In the next section we will see some examples, and introduce different methods
to solve boundary problems in electrostatics.

11
2 Boundary-value problems in electrostatics
Many problems in electrostatics involve boundary surfaces on which either the potential or
the normal derivative of the potential is specified. We have already obtained the formal solu-
tion to such problems by the method of Green’s functions. To find such functions, however,
can be quite complicated, and several methods have been developed to solve boundary-value
problems in electrostatics. In this chapter we will see some of them.

2.1 Method of images

Charge in the presence of a conductor plane

Imagine we have a conductor plane at z = 0, which forces the potential to vanish there, .i.e.
Φ(x, y, 0) = 0, and we place a point charge of magnitude q at a distance d from the plane
(see figure below on the left). We want to find the potential in the region z > 0.

Φ=0

q -q q

z<0 z>0
Let us locate the charge at (0, 0, d) for simplicity. The method of images consist on adding
image charges outside the region of interest in such a way that the boundary conditions are
satisfied. The potential is then the sum of the potentials for the point charge (or charges)
and their images. In the present case, it is enough to add a single image charge located at
(0, 0, −d) and of magnitude −q. The total potential is

q 1 q 1
Φ(x, y, z) = 1/2
− (44)
4π0 (x2 + y 2 + (z − d)2 ) 4π0 (x2 + y 2 + (z + d)2 )1/2
the first term is simply the potential due to the point charge at (0, 0, d) while the second term
is the potential of the image charge, which we added in order for the boundary conditions
at z = 0 to be satisfied. Indeed, you can check that the above potential satisfies both the
Poisson equation and the correct boundary conditions

q
∇2 Φ(x, y, z) = − δ(x)δ(y)δ(z − d), for z > 0, (45)
0
Φ(x, y, 0) = 0 (46)

12
Note that it is important that the image charge is located outside the region of interest (see
figure above on the right), hence the total potential will still satisfy the Poisson equation in
the region of interest. More precisely, we could have added the delta function corresponding
to the image charge, but this vanishes identically for z > 0.
The method of images is intimately connected to the method of Green’s functions. In order to
make this relation clear, let us solve this problem by using the method of Green’s functions.
The Green’s functions for Dirichlet boundary conditions on the plane z = 0 take the form

1
G(~r, ~r 0 ) = + F (~r, ~r 0 ), ∇02 F (~r, ~r 0 ) = 0, z, z 0 > 0 (47)
|~r − ~r 0 |
where F (~r, ~r 0 ) has to be chosen such as to satisfy

G(~r, ~r 0 ) = 0, for z 0 = 0. (48)

we can choose

1
F (~r, ~r 0 ) = − (49)
|~r − ~r 0R |
where ~r 0R is the reflection of ~r 0 on the plane, namely ~r 0R = (x0 , y 0 , −z 0 ). Note that in the
region of interest the Laplacian of F (~r, ~r 0 ) vanishes. The relation with the method of images
is now clear, the Green’s function for Dirichlet boundary conditions on z = 0
1 1
GD (~r, ~r 0 ) = − (50)
|~r − ~r | |~r − ~r 0R |
0

is simply the potential for a point charge of unit magnitude (in units of 4π0 ) in the presence
of a conductor plane at z = 0, where ~r 0 is the location of the point charge. This has to be
the case, indeed, by definition the Green’s function satisfies the Poisson equation with a unit
charge as a source and vanishes at z = 0. 2
Having constructed the appropriate Green’s function we can now plug the desired charge
distribution in (40). For example, for a point charge at d~ we obtain
Z  
1 1 q q
Φ(~r) = q δ(~r − d~ )
0
− dV 0 = − (51)
V |~r − ~r | |~r − ~r 0R |
0 ~
|~r − d| |~r − d~R |
exactly as expected from the method of images.
Returning to the potential (44) there are many questions we can ask. For instance, what is
the force acting on the charge q? the simplest way to calculate this is by computing the force
done on the charge q by its image, an attractive force along the z direction with magnitude
1 q2
|Fz | = (52)
4π0 4d2
2
Note that the Green’s function is symmetric under interchange of ~r and ~r 0 , so it also vanishes for z 0 = 0,
which was the original requirement for Green’s functions in section one.

13
There is another, more instructive, way to compute the force. Though the potential vanishes
at z = 0 its normal derivative does not:

∂ 1 qd
Φ|z=0 = (53)
∂z 2π0 (x + y 2 + d2 )3/2
2

Through (31) we can interpret this result as saying that the charge is inducing a surface
density:

1 qd
σ(x, y) = − (54)
2π (x + y 2 + d2 )3/2
2

An infinitesimal piece of surface dxdy possesses total charge σ(x, y)dxdy, hence the total
force over the charge is

1 q2
Z
q d
Fz = σ(x, y) dxdy = − (55)
4π0 (x2 + y 2 + d2 )3/2 4π0 4d2
In perfect agreement with our previous result.
In the following we will discuss another geometry where it is possible to guess the location
of the image charges.

Charge in the presence of a conductor sphere

Imagine we place a point charge of magnitude q at a distance b from the center of a sphere of
radius a < b. The sphere is a grounded conductor, so we require the potential to vanish on
its surface, namely Φ(~r) = 0 for |~r| = a. We want to find the potential for this configuration
in the region |~r| > a (see figure).

Φ=0
q

q'

14
Let us denote the vector position of the point charge as ~rq = nq b and let us assume that only
one image charge of magnitude q 0 is enough. By symmetry, the image charge should lie on
the ray from the origin to the original point charge. The scalar potential would then be

q0
 
1 q
Φ(~r) = + (56)
4π0 |~r − nq b| |~r − nq b0 |
Now we need to adjust q 0 and b0 such that

q0
 
1 q
Φ(na) = + =0 (57)
4π0 |na − nq b| |na − nq b0 |
for any unit vector n. You can explicitly check that this can be achieved by choosing

a a2
q 0 = − q, b0 = (58)
b b
Which gives the full potential for the problem at hand
 
1 q a q
Φ(~r) = − (59)
4π0 |~r − nq b| b |~r − nq a2 /b|

Green’s function for the sphere

As for the case of the conductor plane, the Dirichlet Green’s function for the sphere is simply
the potential for a unit charge (including its image) in units of 4π0 , where ~r 0 refers to the
location of the unit charge. We obtain

1 a 1 1 1
G(~r, ~r 0 ) = 0
− 0 2 = −
(r + r − 2rr0 cos γ)1/2
|~r − ~r | r |~r − r02 ~r |
a 0 2 02 r2 r02

1/2
a2
+ a2 − 2rr0 cos γ
(60)
0
where γ is the angle between ~r and ~r . With the Green’s function at our disposal, we can
write down the general solution to the Dirichlet problem on the sphere. For instance, suppose
for simplicity that there is no charge distribution and the prescribed boundary conditions
at the sphere fix Φ(a, θ, φ) = V (θ, φ), where we are using spherical coordinates. Then, from
(40), the solution to the potential takes the form

a(r2 − a2 )
Z
1
Φ(~r) = V (θ0 , φ0 ) sin θ0 dθ0 dφ0 (61)
4π (r2 + a2 − 2ar cos γ)3/2

2.2 Method of orthonormal functions

A very important feature of the equations of electrostatics, (17) and (23) or equivalently (24),
is that they are linear equations (in the fields or the potential). This leads to the principle
of superposition: the sum of solutions is also a solution. Conversely any solution can also be

15
expressed in a convenient basis of solutions (you have already met this phenomenon, when
solving the equations of motion for a vibrating string).
The representation of solutions of potential problems by expansions in orthonormal functions
is a powerful technique that can be used in a large class of problems. The particular set
chosen depends on the symmetries of the problem.

Generalities

We start by recalling some general properties of orthonormal functions. Consider an inter-

val (a, b) and a set of real or complex functions in one variable Un (x), n = 1, 2, .... The
orthonormality condition takes the form
Z b
∗
Um (x)Un (x)dx = δmn (62)
a

If the above set is complete then an arbitrary function f (x) (square integrable in the interval
(a, b)) can be written as a linear combination of the orthonormal functions

∞
X
f (x) = an Un (x) (63)
n=1

where the coefficients are given by

Z b
an = Un∗ (x)f (x)dx (64)
a

Plugging this expression into (63) we obtain

∞
!
Z b X
f (x) = Un∗ (x0 )Un (x) f (x0 )dx0 (65)
a n=1

recalling the properties of the Dirac delta function, it should be the case that

∞
X
Un∗ (x0 )Un (x) = δ(x0 − x) (66)
n=1

This is called completeness or closure relation. The most famous example of orthogonal
functions are the sines and cosines that enter in the Fourier expansion. For the interval
(−a/2, a/2) they are
r   r  
2 2πnx 2 2πnx
sin , cos (67)
a a a a
The extension to higher dimensions is obvious. For instance, a function of two variables
admits an expansion in terms of the sets Un (x) and Vn (y).

16
 X Z Z
∗
f (x, y) = amn Um (x)Vn (y), where amn = dx dy Um (x)Vn∗ (y)f (x, y) (68)

If the interval becomes infinite the set of orthogonal functions Un (x) may become an uncount-
able set, namely Un (x) → Uk (x) with k a real number. In this case sums are replaced by
integrals and the Kronecker delta is replaced by the Dirac delta. The most famous example
is the exponential Fourier series expansion in terms of the set

1
Uk (x) = √ eikx (69)
2π
An arbitrary function can be expanded in terms of those
Z ∞
1
f (x) = √ A(k)eikx dk (70)
2π −∞

This expansion is usually called exponential Fourier expansion. The ”coefficients” A(k)
(which now depend on a continuous function) are given by
Z ∞
1
A(k) = √ e−ikx f (x)dx (71)
2π −∞
The orthonormality condition is
Z ∞
1 0
ei(k−k )x dx = δ(k − k 0 ) (72)
2π −∞

while the completeness relation takes the form

Z ∞
1 0
eik(x−x ) dk = δ(x − x0 ) (73)
2π −∞

These last two integrals give useful representations of the delta function, and will be used
below.

Cartesian coordinates

For electrostatic problems with rectangular symmetries it is convenient to use cartesian

coordinates. The Laplace equation in cartesian coordinates is

∂ 2Φ ∂ 2Φ ∂ 2Φ
+ + 2 =0 (74)
∂x2 ∂y 2 ∂z
Looking for separable solutions of the form Φ(x, y, z) = X(x)Y (y)Z(z) we find

1 d2 X 1 d2 Y 1 d2 Z
+ + =0 (75)
X(x) dx2 Y (y) dy 2 Z(x) dz 2

17
which implies

1 d2 X 1 d2 Y 1 d2 Z
2
= −α2 , 2
= −β 2 , 2
= α2 + β 2 (76)
X(x) dx Y (y) dy Z(x) dz

Hence we find the following basis of solutions

√
2 2
Φ = e±iαx e±iβy e± α +β z (77)
where at this point α and β are arbitrary, but we have chosen them to be real. Any linear
combination of the above solutions is a solution. When specific boundary conditions are
imposed on the potential, we will have a restriction on the values that α and β can take.
For example, let us consider the problem of finding the potential inside a rectangular box
with dimensions (a, b, c) in the (x, y, z) directions, such that the potential vanishes at all
faces except at z = c, where it takes some prescribed value Φ(x, y, c) = V (x, y), see figure.

Φ = V(x,y)
c

a
b

Requiring the potential to vanish at x = 0, y = 0 and z = 0 we see the potential should be

the sum of terms of the form
p
Φ = sin αx sin βy sinh α2 + β 2 z (78)
Requiring the potential to vanish at x = a and y = b further fixes the form of α and β

mπ nπ
α= , β= (79)
a b
for m, n integer numbers. Hence, requiring the boundary conditions on five of the six faces
fixes the form of the solution to be

18
 ∞  mπ   nπ  r
X mπ 2  nπ 2
Φ(x, y, z) = Am,n sin x sin y sinh + z (80)
m,n=1
a b a b
The remaining boundary condition requires

∞  mπ  r
X nπ mπ 2  nπ 2
Am,n sin x sin y sinh + c = V (x, y) (81)
m,n=1
a b a b

Which is basically the sine Fourier expansion of V (x, y). Hence the coefficients Am,n are
given by
Z a Z b
4 mπ nπ
Am,n = q dx dy sin x sin yV (x, y) (82)
mπ 2

nπ 2


0 0 a b
ab sinh a
+ b
c
where we have considered an odd extension of V 3 and used the orthonormality of the func-
tions entering the solution:
Z a  mπ   0 
mπ
dx sin x sin x = aδmm0 (83)
−a a a
for m 6= 0.

Another problem which can be easily solved by the method of rectangular orthogonal func-
tions is that of a point charge q inside a grounded box. Imagine the potential is required
to vanish at the six faces x, y, z = 0 and x = a, y = b, z = c, and we locate a charge q at
(x0 , y0 , z0 ). The potential should satisfy the corresponding Poisson equation
q
∇2 Φ(x, y, z) = − δ(x − x0 )δ(y − y0 )δ(z − z0 ) (84)
0
The boundary conditions suggest we expand the solution in terms of the set of orthogonal
functions

∞ r  
X 8  mπx   nπy  `πz
Φ(x, y, z) = Am,n,` sin sin sin (85)
m,n,`=1
abc a b c
Note that the boundary conditions are automatically satisfied. Plugging this expression into
the right hand side of the Poisson equation (84) we find

∞ r  2 !  
X 8  mπ 2  nπ 2 `π  mπx   nπy  `πz
− Am,n,` + + sin sin sin
m,n,`=1
abc a b c a b c
(86)
3
Remember from Fourier series, in order to do a sine expansion often you have to extend your functions
to the region (−a, 0) in a fashion consistent with the odd symmetry of sin x.

19
On the other hand, using the completeness relation (66) for the orthonormal functions at
hand gives the following representation of the delta function

∞
X 2  mπx   mπx 
0
δ(x − x0 ) = sin sin (87)
m=1
a a a
and similar for δ(y − y0 ) and δ(z − z0 ). We see the Poisson equation is satisfied if
r  2 !−1  
q 8  mπ 2  nπ 2 `π  mπx 
0
 nπy 
0 `πz0
Am,n,` = + + sin sin sin
0 abc a b c a b c

which gives the final solution

∞ mπx0 nπy0
sin `πz



 
q X 8 sin a
sin b c
0  mπx   nπy  `πz
Φ(x, y, z) =
2 sin sin sin
0 m,n,`=1 abc mπ 2 nπ 2 a b c
+ `πc



a
+ b

It would have been extremely hard to solve this problem by other methods!

Finally, in order to solve problems with boundary surfaces it is sometimes convenient to

expand the corresponding Green’s function in the particular set of orthonormal functions
under consideration. For instance, lets consider Dirichlet boundary conditions at the plane
z = 0. We have already found the Green’s function for this problem and have obtained
1 1
G(~r, ~r 0 ) = − (88)
|~r − ~r | |~r − ~r 0R |
0

In order to expand this in rectangular orthonormal functions, recall the integral representa-
tion (73) for the Dirac delta function
Z ∞ Z ∞ Z ∞
1
dα dβ dγeiαx eiβy eiγz = δ(x)δ(y)δ(z) (89)
(2π)3 −∞ −∞ −∞

This suggests

∞ ∞ ∞
−4π eiαx eiβy eiγz
Z Z Z
1
dα dβ dγ 2 = 2 (90)
(2π)3 −∞ −∞ −∞ α +β +γ 2 2 (x + y + z 2 )1/2
2

Since acting with the Laplacian would produce −4πδ(~r) on both sides. Integrating over γ
we obtain
√
Z ∞ Z ∞ −|z| α2 +β 2
1 1 e
= − dα dβeiαx eiβy p (91)
(x2 + y 2 + z 2 )1/2 2π −∞ −∞ α2 + β 2
Hence we obtain for the Green’s function

eiα(x−x ) eiβ(y−y )  −|z−z0 |√α2 +β 2 √

Z ∞ Z ∞0 0
1 1 1 −|z+z 0 | α2 +β 2

0
− 0
=− dα dβ p e −e (92)
|~r − ~r | |~r − ~r R | 2π −∞ −∞ α2 + β 2

20
As expected, note that the r.h.s. vanishes if either z = 0 or z 0 = 0.

Spherical coordinates

The Laplace equation in spherical coordinates (r, θ, φ) takes the form

1 ∂2 ∂ 2Φ
 
1 ∂ ∂Φ 1
2
(rΦ) + 2 sin θ + 2 2 =0 (93)
r ∂r r sin θ ∂θ ∂θ r sin θ ∂φ2
Note that the Laplacian splits into a ”radial” part and an ”angular” part

1 ∂2 1
∇2 Φ = 2
(rΦ) + 2 ∇2θ,φ Φ (94)
r ∂r r
1 ∂ 2Φ
 
1 ∂ ∂Φ
∇2θ,φ = sin θ + (95)
sin θ ∂θ ∂θ sin2 θ ∂φ2
It is convenient to look for separable solutions of the form

Φ(r, θ, φ) = R(r)Y (θ, φ) (96)

Proceeding as before, we find the expansion in terms of orthonormal functions takes the form

∞ X
X `
A`,m r` + B`,m r−(`+1) Y`m (θ, φ)


Φ= (97)
`=0 m=−`

The orthonormal functions Y`m (θ, φ) depend only on the angular variables and are called
spherical harmonics. Here ` = 0, 1, 2, ... and for a fixed ` , m takes the integer values from
−m to m. They satisfy

∇2θ,φ Y`m (θ, φ) = −`(` + 1)Y`m (θ, φ) (98)

This is a complete set of functions, in the sense that any (single valued!) function of the
angular variables g(θ, φ) can be written in terms of those. They satisfy
∗
Y`,−m (θ, φ) = (−1)m Y`,m (θ, φ) (99)

The orthonormality condition takes the form

Z 2π Z π
dφ sin θdθY`∗0 ,m0 (θ, φ)Y`,m (θ, φ) = δ`,`0 δm,m0 (100)
0 0

while the completeness relation takes the form

∞ X
X `
∗
Y`,m (θ0 , φ0 )Y`,m (θ, φ) = δ(φ − φ0 )δ(cos θ − cos θ0 ) (101)
`=0 m=−`

21
The first few spherical harmonics are

1
Y00 = √ (102)
4π
r
3
Y11 = − sin θeiφ (103)
8π
r
3
Y10 = cos θ (104)
4π
r
1 15
Y22 = sin2 θe2iφ (105)
4 2π
r
15
Y21 = =− sin θ cos θeiφ (106)
8π
r  
5 3 2 1
Y20 = cos θ − (107)
4π 2 2
The prototypical problem with spherical symmetry is that of finding the potential inside a
sphere of radius a with prescribed potential at the sphere. Namely the potential satisfies the
Laplace equation and the boundary conditions

Φ(a, θ, φ) = V (θ, φ) (108)

Φ(r, θ, φ) is bounded as r → 0 (109)
Where V (θ, φ) is some prescribed function. The general solution to this problem takes the
form

∞ X
X `
Φ(r, θ, φ) = A`,m r` Y`m (θ, φ) (110)
`=0 m=−`
where we have used the fact that the solution is bounded as r → 0, so as to keep only
non-negative powers or r. Note that if we were interested in solving the problem outside the
sphere, we would require that the potential is bounded at infinity, and we would keep only
the negative powers. The coefficients A`,m can be fixed by requiring the correct boundary
conditions at r = a:

∞ X
X `
Φ(a, θ, φ) = A`,m a` Y`m (θ, φ) = V (θ, φ) (111)
`=0 m=−`

Using the orthonormality condition (100) we obtain

Z 2π Z π
∗
A`,m = dφ sin θdθY`,m (θ, φ)V (θ, φ) (112)
0 0

22
As in the case of rectangular coordinates, for problems including boundaries, it will be
useful to expand the Dirichlet Green’s function for a sphere of radius a in terms of spherical
orthonormal functions
X
G(~r, ~r 0 ) = A`,m Y`,m (θ, φ) (113)
`,m

where (θ, φ) are the angular coordinates of the point r and the ”coefficients” A`,m depend on
(r, r0 , θ0 , φ0 ). Since the l.h.s is symmetric under interchange of ~r and ~r 0 we can furthermore
write
X
G(~r, ~r 0 ) = ∗
Y`,m (θ0 , φ0 )A`,m (r, r0 )Y`,m (θ, φ) (114)
`,m

In order to compute the coefficients A`,m (r, r0 ) we can proceed as follows. The Dirac delta
function in spherical coordinates takes the form 4
1
δ(~r − ~r 0 ) = δ(r − r0 )δ(φ − φ0 )δ(cos θ − cos θ0 ) (115)
r2
Using the completeness relation (101) we can write

∞ X `
0
X 1
δ(~r − ~r ) = 2
δ(r − r0 )Y`,m
∗
(θ0 , φ0 )Y`,m (θ, φ) (116)
`=0 m=−`
r
hence
∞
X
∗
X 1
∇ 2
Y`,m (θ0 , φ0 )A`,m (r, r0 )Y`,m (θ, φ) = −4π 2
δ(r − r0 )Y`,m
∗
(θ0 , φ0 )Y`,m (θ, φ) (117)
`,m `,m
r

which implies
1 d2 `(` + 1) 4π
2
(rA`,m ) − 2
A`,m = − 2 δ(r − r0 ) (118)
r dr r r
0 0
For the regions r > r and r < r the delta function vanishes and we obtain

 Ar` + Br−(`+1) , r < r0
A`,m (r, r0 ) = (119)
0 ` 0 −(`+1) 0
Ar +Br , r>r


Now we need to fix the constants of integration A, B, A0 , B 0 (careful: since the equation is
on r, these constants could actually depend on r0 ). Remember that we interpret ~r as the
observation point and ~r 0 as the location of the unit point charge. Since the Green’s function
4
To express δ(x − x0 ) = δ(x1 − x01 )δ(x2 − x02 )δ(x3 − x03 ) in terms of coordinates another set of coordinates
(y1 , y2 , y3 ) we need to divide by the Jacobian, so that δ(x − x0 )d3 x is invariant.

23
is bounded as we take the observation point to infinity, we need A0 = 0. Furthermore, the
Green’s function vanishes as r = a, which gives a relation between A and B and continuity
at r = r0 leave us only with an overall constant
  
a2`+1

 A r` − r(`+1)
, r < r0
A`,m (r, r0 ) = (120)
 02`+1 2`+1

−(`+1) 0
A r −a r , r>r


The overall constant can be fixed as follows. Multiplying (118) by r and integrating both
sides over the interval r = r0 −  to r = r0 +  we obtain
   
d 0 d 0 4π
(rA`,m (r, r )) − (rA`,m (r, r )) =− 0 (121)
dr r0 + dr r0 − r
from where it follows

r0−(`+1)
A = 4π (122)
2` + 1
Putting all together we can write the Green’s function as

a2`+1
 
0
X 1 −(`+1) ∗
G(~r, ~r ) = 4π `
r< − `+1 r> Y`,m (θ0 , φ0 )Y`,m (θ, φ) (123)
`,m
2` + 1 r<

where r> and r< denote the bigger and smaller between r and r0 . Note that the Green’s
function vanishes for both r = a and r0 = a (when one of the the two conditions is satisfied,
the other radius is bigger, since we are looking at the sphere from outside). In very much
the same way one can work out The Green’s function for the interior of the sphere. In this
case we obtain

`
 
0
X 1 −(`+1) r> ∗
G(~r, ~r ) = 4π `
r r − 2`+1 Y`,m (θ0 , φ0 )Y`,m (θ, φ) (124)
`,m
2` + 1 < > a

24
3 Magnetostatics
Frequently, people find magnetism a little more mysterious than electricity. Rather than
thinking of fridge magnets, you should think of electromagnets: charges in motion give rise
to electric currents and these produce magnetic fields; many of you will have visualized
these fields in experiments by sprinkling iron filings on sheets of cardboard transverse to a
current-carrying wire. (The fridge magnet, or any permanent magnet, derives its magnetism
from microscopic currents.) The subject corresponding to electrostatics, which is produced
by time-independent charges, is magnetostatics, which is produced by time-independent, or
”steady”, currents.

+ v + v
+ v
+ v
+ v

Given a collection of point charges Qi in motion with velocities ~vi , we define the corresponding
electric currents as: X
~j = qi~vi

This is a vector field made by adding up elementary contributions. By analogy with charge
density ρ as a smoothed-out distribution of point charges, we introduce a vector field, the
~
Rcurrent density J(x, y, z, t), so that the total electric current in a region V is the integral
~ .
JdV
V
In this section we shall usually be thinking of steady currents, but there is one thing to deal
with first, namely the mathematical expression for the physical observation than charge is
conserved.

3.1 Conservation of charge

Consider a region V with surface S with charges moving through.

S
V

25
 R
The total charge inside V is Q = V
ρdV , so that

Z
dQ ∂ρ ∂ρ
= dV allowing ∂t
6= 0 for the moment
dt V ∂t
= rate of increase of Q
= rate charge goes in − rate charge goes out
Z
= − J~ · dS
S
Z
= − ~
∇ · JdV by divergence theorem
V

an so
Z  
∂ρ
+ ∇ · J~ dV = 0 (125)
V ∂t
this is to be true for all regions V , then
∂ρ
+ ∇ · J~ = 0 (126)
∂t
which is the charge conservation equation.
For the rest of this section, we suppose none of the quantities of interest depends on time.
For this time independent situation (126) reduces to ∇ · J~ = 0.

3.2 Magnetic Field Strength

~ the magnetic field strength B
Like the electric field E, ~ can be defined from an experimentally
stablished force law. Analogously to (7), a charged particle with charge q, moving with
~ is subject to a force
velocity ~v in a magnetic field B

F~ = q~v ∧ B
~ (127)

F
B

26
 ~ but note a peculiar feature of this force law, that the force is
This serves to define B,
orthogonal to the velocity. In particular, therefore ~v · F~ = 0 and the force does no work.

Force between two parallel current-carrying wires

lines of force

k
j a
i

I1= I 1k I2= I 2k

To see (127) in action, let us consider the force between two parallel current-carrying wires.
What can we say about the magnetic field B ~ generated by a single wire? let us choose coordi-
nates such that the wire is along the z direction and let us use cylindrical-polar coordinates.
It can be experimentally verified that if we take a charge q and move it around the wire, the
charge is subject to a force. This force vanishes, i.e. there is no force, if the charge moves
along the θ direction, while there is a force if the charge moves in any other direction. This
means that B ~ has a component only in the θ direction5 ; using the symmetry in θ and z we
can write B~ = B(R)~eθ , where R is the distance to the wire.
Now suppose we have parallel wires, wire 1 carrying current I1 and wire 2 carrying current
I2 . Each wire gives rise to a B-field and the field from one exerts a force on the current in
the other. Using (127), and thinking of the currents as charged particles in motion, we have:
Force per unit length on wire 2 due to wire 1 is

F~12 = I~2 ∧ B
~ 1.

where B~ 1 means B ~ at wire 2 due to current in wire 1, and from the geometry I~2 = I2~k.
Similarly, the force on wire 1 due to wire 2 is

F~21 = I~1 ∧ B
~ 2.
~ at wire 1 due to wire 2, and I1 = I1~k. These forces must be equal in magnitude
~ 2 is B
where B
and opposite, due to Newton’s third law:
5
This is consistent with the following fact: if you sprinkle iron filings on a cardboard held at right angle
to a current-carrying wire, you can see that the magnetic lines of force are concentric circles, centered on the
wire.

27
 I~2 ∧ B
~ 1 = −I~1 ∧ B
~ 2.
From what was said about the direction of B, ~ we have that at wire 2, B
~ 1 = ~jB1 in terms of
some magnitude B1 , and in wire 1, B~ 2 = −~jB2 in terms of some magnitude B2 . Therefore

|F~12 | = I2 B1 = |F21 | = I1 B2 .

It is an experimental fact that this magnitude is

µ0
|F~ | =
I1 I2
2πa
in terms of a constant µ0 (which, as with 0 is needed to get the right dimensions).
From this experimental fact we deduce that, at wire 2,

~ = µ0 I1 ~j
B
2πa
At a general point, in terms of cylindrical polar coordinates (R, θ, z) with the wire along the
z-axis this is

µ0 I 1
= ~eθ
2π R
which we can write in cartesian coordinates as

~ = µ0 I − y , x , 0 ,
 
B R 2 = x2 + y 2 (128)
2π R2 R2
This is the magnetic field due to an infinite straight wire with constant current, which can
be though of as the elementary magnetic field, much as (4), E~ = Q ~r3 gives the elementary
4π0 r
electric field.

3.3 Differential equations for the magnetic field

Our aim for the next few pages is to calculate the B ~ due to current in an arbitrary wire,
much as the integral (26) together with the relation E ~ = −∇Φ gives E ~ for an arbitrary
charge distribution. We achieve this aim with (137) below. As a step towards this aim, we
note that, when Γ is a horizontal circle of radius R centered on a straight wire

I I
~ · d~` = µ0 I
B
1
~eθ · ~eθ Rdθ (129)
Γ 2π R
= µ0 I (130)
Even though we have derived this for a straight wire, the result is much more general:

~ · d~` = µ0 × total current through Γ.

H
Ampere’s Law: Γ
B

28
 I

Compare ampere’s law to Gauss Law in the first form we had.

For a current density instead of a wire, we obtain an integral version of Ampere’s Law:

I Z
~ · d~` = µ0
B J~ · dS Σ spans Γ
Γ Σ
Z
= ~ · dS
∇∧B by Stoke’s theorem
Σ

and for this to be true for all Σ we obtain the differential version of Ampere’s law:

~ = µ0 J~
∇∧B (131)

(recall that ∇ · J~ = 0 from charge conservation and time independence, and we need this for
(131) to make sense). This equation is supplemented by

~ =0
∇·B (132)

Which can be explicitly checked for the straight wire, from (128). Equation (132) is to be
~ = 1 ρ. Thus we interpret (132) as saying ”there are no magnetic
contrasted with (23): ∇ · E 0
charges”.

The magnetic potential

It follows from (17) that, in a simply-connected region, there exists a function Φ such that
~ = −∇Φ. It is a less familiar fact that
E
Claim: if ∇ · B ~ = 0 in a suitable region, then there exists in that region a vector field A
~
such that
~ =∇∧A
B ~

29
~ is the magnetic potential (also called vector potential, in which case Φ is called the scalar
A
potential).
Note that a change A ~ →A ~ + ∇ζ, for any scalar function ζ, leaves B ~ unchanged. We may
exploit this freedom to impose another condition, namely

~=0
∇·A
~ 6= 0 and change A
for suppose F = ∇ · A ~→A ~ + ∇ · ζ, then this would change F → F + ∇2 ζ;
2
now, choose ζ such that ∇ ζ = −F , so that now ∇ · A ~ = 0.
Equation (131) can be turned into an equation for A~ as follows:

~ = ∇ ∧ (∇ ∧ A)
∇∧B ~ = ∇(∇ · A) − ∇2 A
~

= µ0 J~ (133)

~ = 0, then we have
with the choice ∇ · A

~ = −µ0 J~
∇2 A (134)
~ = 0. This has the vector form of
Remember that this is true for the particular choice ∇ · A
the Poisson’s equation, and we’ve solved Poisson’s equation before: recall (26) and compare
with (24). So we solve (134) by
Z ~ r 0)
~ ~r ) = µ0
A(
J(~
dV 0
4π 0
|~r − ~r |
This is for a volume distribution of current. If we have just a line current J~ = I~t, where I is
constant and ~t is the tangent (of unit length) to a curve L (where the curve L is the ”wire”),
then instead

~td`
Z
~ = µ0 I
A
4π L |~r − ~r 0 (`)|

30
where ` parametrizes where we are along the wire.
~ we can calculate B.
From the vector potential A ~ For a volume distribution

Z ~ 0
~ ~ µ0 J(~r )
B(~r) = ∇ ∧ A = ∇∧ dV 0 (135)
4π |~r − ~r 0 |
Z ~ 0
µ0 J(~r ) ∧ (~r − ~r 0 ) 0
= dV (136)
4π |~r − ~r 0 |3

while for a line distribution/wire

~t ∧ (~r − ~r 0 )
Z
B(~ ~ = µ0 I
~ r) = ∇ ∧ A d`
4π L |~ r − ~r 0 |3
or
d~` ∧ (~r − ~r 0 )
Z
µ0 I
B(~r) = (137)
4π L |~r − ~r 0 |3
Where d~` = d`~t. This is now B ~ at all points of space due to an arbitrary wire, and so
generalizes (128). We can use it to obtain an expression for the force between two arbitrary
wires, by chopping the second wire into elementary pieces.

P1 r
P2

I1 I2

At P2 on wire 2

dF~ = I2 d~`2 ∧ B
~1
so
Z
F~ = I2 d~`2 ∧ B
~1

d~`2 ∧ (d~`1 ∧ (~r2 − ~r1 ))

ZZ
µ0 I1 I2
= (138)
4π |~r2 − ~r1 |3

where ~r1 = ~r1 (`1 ) belongs to wire 1 and ~r2 = r2 (`2 ) belongs to wire 2. This is the Biot-Savart
Law for the force between two arbitrary current-carrying wires (more precisely, the force on
wire 2 done by wire 1). Its interest is largely theoretical, as it is rather hard to use.

31
3.4 The story so far
We’ve studied time-independent electricity (or electrostatics) and time-independent mag-
netism (or magnetostatics), and found these to be governed by the following system of
equations

~ = 1ρ
∇·E ~ =0
∇·B
0
~ =0
∇∧E ~ = µ0 J~
∇∧B (139)

with the understanding that

~
∂E ~
∂B ∂ J~ ∂ρ
=0= = = .
∂t ∂t ∂t ∂t
The electric and magnetic field strengths themselves can be defined from the force laws (7)
and (127), which we combine as follows: a particle with charge Q moving with velocity ~v in
~ and a magnetic field B
a combination of an electric field E ~ is subject to the force given by

F~ = Q(E
~ + ~v ∧ B).
~

In addition we have one equation which we expect to be valid also in the time-dependent
case, namely the charge conservation equation
∂ρ
+ ∇ · J~ = 0. (140)
∂t
The problem now is how to extend all these equations to the time-dependent case, i.e. how
to put back the terms that are zero in the time-independent case. This will be the subject
of the next part of this course.

32
4 Time dependent electromagnetism

4.1 Maxwell equations

Experimental evidence suggests two qualitative facts:

~ produces an E.
• A time-varying B ~
~ produces a B.
• A time-varying E ~

This experimental evidence can be summarized quantitatively in the following two integral
laws:

I Z
~ ~ d ~ · dS
E · d` = − B Faraday’s law of induction (141)
Γ dt Σ
I Z Z ~
~ · d~` =
B ~ + 0 µ0 ∂ E · dS
µ0 J~ · dS ~ (142)
Γ Σ Σ ∂t

Where Σ spans Γ. Apply Stoke’s theorem to both to obtain

!
Z ~
~ + ∂B
∇∧E ~=0
· dS (143)
Σ ∂t
!
Z ~
~ − µ0 J~ − 0 µ0 ∂ E
∇∧B ~=0
· dS (144)
Σ ∂t

and if these hold for all Σ we obtain the

Maxwell equations:

~
∂B 1 ∂E~
~+
∇∧E = 0; ~ = µ0 J~ +
∇∧B
∂t c2 ∂t
~ ~ 1
∇·B = 0, ∇·E = ρ (145)
0
1
with c2 = 0 µ0
.
These are the Maxwell’s equations, the basic equations of the new theory of electromagnetism,
in which electricity and magnetism are merged.
Note that

• Two of the equations always have zero on the right; the other two contain the sources
~
ρ and J.

33
 • If all time-derivatives are set to zero, we recover the equations of the previous section.

• The system obtained by setting ρ and J~ to zero is called source-free Maxwell equations.

• Regarding ρ and J~ as given, there are eight equations for six unknowns, so there should
be a consistency condition.

To see what the consistency condition is, calculate:

!
∂ρ ∂ ~
∂E
~ = 0 ∇ ·
= 0 (∇ · E)
∂t ∂t ∂t
1 
~ − µ0 J~

= ∇· ∇∧B
µ0
= −∇ · J~

which we recognize as the charge conservation equation. From this point of view the charge
conservation equation is a consistency condition for the Maxwell’s equations. As an exercise,
show that there is a second consistency condition but that it is automatically satisfied.

Energy of the Electromagnetic Field

There is a deductive way to approach this question, but we shall rely on intuition and
guesswork. We want an energy density (energy-per-unit-volume) which is something like
”1/2m~v · ~v ”.
After playing around with Maxwell equations for a while, we are lead to consider

1 ~ 2 1 ~ 2
E = 0 |E| + |B| (146)
2 2µ0
Then, using the Maxwell equations, we calculate

∂E ∂E~ 1 ~ ∂B ~
~·
= 0 E + B ·
∂t ∂t µ0 ∂t
1 ~  ~ − µ0 J~ − 1 B

~ ·∇∧E
~
= E· ∇∧B
µ0 µ0
 
1 ~ ~ ~ · J~
= −∇ · E∧B −E
µ0

For the source-free case, we set J~ = 0 so that the last term vanishes. Then, this has the
form of a conservation equations

∂E
+ ∇ · P~ = 0 (147)
∂t

34
Remember that E is the energy density. P~ ≡ µ10 E ~ ∧B
~ is called the Poynting vector and has
the interpretation of the rate of energy flow, or momentum density.
If we now reinstate the sources, this equation has a name:
Poynting’s theorem

∂E
= −∇ · P~ − E
~ · J~
∂t
We integrate Poynting’s theorem over a volume V with surface S to obtain an energy balance
equation:

Z Z Z
d
EdV = − P~ · dS
~− ~ · JdV
E ~
dt V S V

(1) (2) (3)

where each term has the following interpretation:

1. Rate of increase of electromagnetic energy.

2. Rate of energy flow into V .

3. Rate of work done by the field on sources.

To justify the interpretation of (3), consider a single charge. Remember that work is the
product of force and displacement W = F~ · d~`, so the rate of work is dW
dt
= F~ · ~v . For a single
charge F~ = q E
~ and q~v is by definition the current ~j, so that dW
dt
=E ~ · ~j. This satisfactory
interpretation of the third term reinforces the interpretation of E as energy density.

4.2 Electromagnetic potentials

The introduction of potentials in the time-dependent case is similar to that in the time-
independent case. Assume we are interested in a suitable (simply-connected, etc) region of
space, then the Maxwell equations imply, first:

~ = 0 ⇒ ∃A
∇·B ~ such that B
~ = ∇ ∧ A;
~
~
∂B ~
∂A
next ∂t
=∇∧ ∂t
, so that
!
~ ~
~ + ∂B = ∇ ∧
∇∧E ~ + ∂A
E = 0,
∂t ∂t

whence

35
 ~
~ + ∂ A = −∇Φ
∃Φ such that E
∂t
i.e.

~ = ∇∧A
B ~ (148)
~
~ = −∇Φ − ∂ A
E (149)
∂t

As before, we have the freedom to modify A ~ and Φ slightly without changing the electric
and magnetic fields. This freedom has the name of gauge transformation:

~→A
A ~ + ∇ζ (150)
∂ζ
Φ→Φ− (151)
∂t
This gauge transformation and can be exploited, as before, to simplify potentials.
Plugging the expression for the fields in terms of their potentials into the Maxwell equations
we obtain

∂ ~ =−1ρ
∇·A ∇2 Φ + (152)
∂t 0
~
1 ∂ 2A

1 ∂Φ

2~ ~
∇ A− 2 2 −∇ ∇·A+ 2 = −µ0 J~ (153)
c ∂t c ∂t

This is a nice set of equations for the potentials, but they are coupled. You can show that
the gauge transformations can be used in order to choose potentials such that

~ + 1 ∂Φ = 0
∇·A (154)
c2 ∂t
For this particular choice we obtain

1 ∂ 2Φ 1
∇2 Φ − 2 2
=− ρ (155)
c ∂t 0
~
1 ∂ 2A
~−
∇2 A = −µ0 J~ (156)
c2 ∂t2
These equations, supplemented with our gauge choice (154) are completely equivalent to
Maxwell equations. In order to solve these equations, and give the time-dependent of (26)
and (137), we need to introduce a new concept.

36
4.3 Time-dependent Green’s functions
The inhomogeneous wave equations found in the previous section have the structure

1 ∂ 2ψ
∇2 ψ − = −4πf (~r, t) (157)
c2 ∂t2
where c is the velocity and f (~r, t) is a known source term. In order to solve this equation,
it is convenient to find the corresponding Green’s function, as we did for electrostatics. By
definition

1 ∂2
∇2 G(~r, t; ~r 0 , t0 ) −
2 2
G(~r, t; ~r 0 , t0 ) = −4πδ(~r − ~r 0 )δ(t − t0 ) (158)
c ∂t
We will focus on situations without boundaries and choose our Green’s function to be a
function of the differences ~r − ~r 0 and t − t0 . As we have seen, the delta functions have the
following integral representation
Z ∞ Z ∞
0 0 1 ~ 0 0
δ(~r − ~r )δ(t − t ) = 3
dk dωeik·(~r−~r ) e−iω(t−t ) (159)
(2π)4 −∞ −∞

Hence it is natural to represent the Green’s function as

Z ∞ Z ∞
~ 0 0
0 0
G(~r, t; ~r , t ) = 3
dk dωg(~k, ω)eik·(~r−~r ) e−iω(t−t ) (160)
−∞ −∞

Acting on both sides of (160) with the wave operator

1 ∂2
∇2 − (161)
c2 ∂t2
and comparing with the delta function representation (159) we obtain

1 1
g(~k, ω) = 3 ω2
(162)
4π k 2 − c2

The manipulations we did here are exactly the same we did to arrive to (90), however, there
is a crucial difference: as we plug back (162) into (160) and try to do the integral over ω, we
find two poles along the real line, at

ω = ck, ω = −ck (163)

Note that this is due to the minus sign in front of the second time derivative in the wave
equation. In order to compute (160) we need to give a prescription on how to treat these
poles. In similar situations in physics one adds a small imaginary part and moves the poles
a little above or below the real axis. In order to know which one of the choices is the correct
one, we turn into physical considerations.

37
 What is the meaning of G(~r, t; ~r 0 , t0 )? it represents the wave caused by a disturbance at
~r = ~r 0 , t = t0 . Hence, for t < t0 we demand G(~r, t; ~r 0 , t0 ) to vanish identically (since the
disturbance didn’t happen yet!)

G(~r, t; ~r 0 , t0 ) = 0, for t < t0 (164)

Looking at (160) we see that for t < t0 we can compute the integral over ω by closing the
contour at infinity from above. Indeed, for t − t0 < 0 the contribution from the circle above
vanishes exponentially. The correct prescription for the poles is then to push them slightly
below the real axis:

ω = ck − i, ω = −ck − i (165)

so that the integral over ω vanishes for the case t < t0 . See figure

t<t'

ω-plane

-c k - i ε ck-iε

t>t' C

Now, lets turn out to the more interesting situation t > t0 . In this case we can compute the
integral over ω by closing the contour from below. By residue theorem we obtain

sin (ck(t − t0 ))
Z
1 −iω(t−t0 )
dω (ω+i)2
e = 2πc (166)
C k2 − k
c2
Hence
∞
sin (ck(t − t0 ))
Z
0 0 c ~ 0)
G(~r, t; ~r , t ) = d3 keik·(~r−~r (167)
2π 2 −∞ k
Z ∞
2c
= dk sin(k|~r − ~r 0 |) sin (ck(t − t0 )) (168)
π|~r − ~r 0 | 0
Z ∞
c
= dk sin(k|~r − ~r 0 |) sin (ck(t − t0 )) (169)
π|~r − ~r 0 | −∞

where in the second line we have used spherical coordinates for ~k and integrated over the
angular variables. In the third line we have simply extended the integration region to the
whole real line. Then, using the exponential form of the trigonometric functions we end up

38
with the integral representation for the delta function (73). One delta is identically zero as
t − t0 > 0, while the other one gives

 
0 |~
r−~r 0|
δ t + c
−t
G(~r, t; ~r 0 , t0 ) = (170)
|~r − ~r 0 |

We see that the Green’s function vanishes identically outside the sphere |~r − ~r 0 | = c(t − t0 ).
This result fits perfectly with the interpretation of G(~r, t; ~r 0 , t0 ) as a wave that originates
from the perturbation at ~r = ~r 0 and t = t0 and then expands spherically at velocity c. This
Green’s function is also called the retarded Green’s function: the effect observed at0 the point
~r at time t is due to the perturbation originated at a retarded time t0 = t − |~r−~ c
r |
. Hence,
the final solution to the inhomogeneous wave equation (157) in the absence of boundaries is
 
Z δ t0 + |~
r−~r 0| 0
c
−t Z
f (~r 0 , t − |~r−~r |
) 0
0 0 0 0 c
ψ(~r, t) = f (~r , t )dV dt = dV (171)
|~r − ~r 0 | |~r − ~r |0

This allows to write the solution for the scalar and vector potential wave equations (155) in
the absence of boundaries for arbitrary charge and current distributions.

0
ρ(~r 0 , t − |~r−~r |
) 0
Z
1 c
φ(~r, t) = dV (172)
4π0 |~r − ~r 0 |
r 0|
J(~r , t − |~r−~
Z ~ 0
~ r, t) = µ0 c
) 0
A(~ dV (173)
4π |~r − ~r 0 |

We leave it as an exercise to show that this solution indeed satisfies the gauge (154).

4.4 Maxwell equations in macroscopic/dielectric media

Consider the basic equations of electrostatics

~ = 0,
∇∧E ∇·E ~ = 1ρ (174)
0
writing them down for a real situation would require a precise knowledge of the charge
density at all points of space. Except some ideal situations (e.g. a collection of charges in
the vacuum) such a knowledge is impossible. For instance, if we are studying electrostatics in
water, on a cubic centimeter there would be billions of atoms with their electrons and protons
(which are furthermore vibrating!) and they would have to be accounted for. Luckily one
can write an effective set of equations which describes electrostatics in macroscopic media.
For the simplest case of media they amount to simply replacing the value of 0 valid for he
vacuum by , which characterizes the medium:

39
 ~ = 0,
∇∧E ~ =ρ
∇ · ( E) (175)
where now the medium itself does not contribute to the charge density. The ratio /0 is
called the dielectric constant of the medium, such that the dielectric constant of the vacuum
is one. For other materials the dielectric constant is always larger than one. For instance,
the dielectric constant of air is very close to 1 (about 1.0005), while the dielectric constant
of water is about 2.
These effective equations can be understood as follows. Usually a macroscopic medium
contains dipoles 6 . In the absence of an electric field these dipoles are randomly aligned. As
we turn an electric field, however, these dipoles align, producing themselves a net electric
field. This alignment (and the field the dipoles produce) is proportional to the applied electric
field, effectively changing the value of  and giving (175). For most examples studied in this
lectures  will be just a constant, for which the effective equations exactly coincide with the
equations on the vacuum, upon replacing 0 → . An exception will be problems where a
plane separates two regions with different values of . In this case, by applying Gauss law,
we can see that as we cross the boundary
 
~ 1 − 2 E
1 E ~ 2 · ~n = 0 (176)
On the other hand, by applying Stokes theorem we obtain
 
~1 − E
E ~ 2 ∧ ~n = 0 (177)
where ~n is the normal to the boundary. Hence, as we cross the boundary, we must impose
~ and the tangential components of E.
continuity for the normal component of E ~
A similar description can be given for the equations of magnetostatics, and the effective
equations take the form

1~ ~ ~ =0
∇∧ B = J, ∇·B (178)
µ
The ratio µ/µ0 is called permeability of the medium. It is one for the vacuum, extremely
close to one for air and slightly less than one for water. As before, as we cross the boundary
between two regions with different magnetic permeabilities we obtain
 
1 ~ 1 ~ 
~ ~

B1 − B2 ∧ ~n = 0, B1 − B2 · ~n = 0 (179)
µ1 µ2
Hence, as we cross the boundary the tangential components of µ1 B ~ and the normal com-
~ are continuous. Finally, the effective Maxwell equations in the presence of a
ponents of B
medium with constant , µ take the form
6
You can think of a dipole as two opposite charges very close to each other. The dipole is said to be
aligned in the direction of separation of the charges.

40
 ~ ~
~ + ∂ B = 0;
∇∧E ~ = µJ~ + µ ∂ E
∇∧B
∂t ∂t
~
∇·B = 0, ~ = 1ρ
∇·E (180)


41
5 Electromagnetic waves
In this section we will study the source-free Maxwell equations and show that they admit
wave solutions. We will then study many properties of these solutions.

5.1 Plane waves

We start from the Source-free Maxwell equations:

~ = 0,
∇·B ~ =0
∇·E (181)
~+ ~
∂B ~− 1 ∂E~
∇∧E ∂t
= 0, ∇∧B c2 ∂t
=0

~ and B.
we would like to decouple the equations for E ~ Calculate:

~ = ∇(∇ · E)
∇ ∧ (∇ ∧ E) ~ − ∇2 E
~ = −∇2 E
~
~
∂B ∂ ~
1 ∂ 2E
= −∇ ∧ ~
= − (∇ ∧ B) = − 2 2
∂t ∂t c ∂t
so that
~
1 ∂ 2E
~−
∇2 E =0 (182)
c2 ∂t2
which is the wave-equation with c = (0 µ0 )−1/2 as the wave speed. Similarly (exercise)

~
1 ∂ 2B
2~
∇ B− 2 2 =0 (183)
c ∂t
We want to solve these wave equations, so think first about a scalar version

1 ∂ 2F ∂ 2F ∂ 2F ∂ 2F 1 ∂ 2F
∇2 F − = + + − =0 (184)
c2 ∂t2 ∂x2 ∂y 2 ∂z 2 c2 ∂t2
and try F = f (~k · ~r − ωt) with constant ω and ~r. Then

∂F
= −ωf 0 , ∇F = ~kf 0
∂t
∂ 2F
= ω 2 f 00 , ∇2 F = k 2 f 00
∂t2
ω2
Hence, F = f (~k · ~r − ωt) is a solution for any (twice-differentiable) f provided k 2 = c2
. Note
that F is constant on surfaces

42
 ~k · ~r − ωt = const
at a fixed t this is the equation of a plane, so these solutions are called plane waves. As time
goes forward, the wavefront propagates in the direction of ~k. If f is of the form

~
f (~k · ~r − ωt) ∼ ei(k·~r−ωt) (185)
then they are called harmonic waves or or monochromatic plane waves with a single frequency
ω and wave number k. Note that both the real part and imaginary part of the exponential
above will be solutions of the wave equation.
After this discussion, we try

~ = ~e E0 ei(~k·~r−ωt)
E (186)
~ = ~b B0 ei(~k·~r−ωt) ,
B
With the understanding that the electric and magnetic fields are the real part of the solutions.
Here ~e and ~b are unit vectors characterizing the direction of the fields and E0 and B0 are
constants. Plugging these expressions into ∇ · B~ = 0, ∇ · E~ = 0 we obtain

~e · ~k = 0, ~b · ~k = 0 (187)
This means that both E ~ and B
~ are perpendicular to the direction of propagation of the wave.
Such a wave is called transverse wave. The rest of the Maxwell equations imply

(~k ∧ ~e)E0 − ω~bB0 = 0 (188)

ω
(~k ∧ ~b)B0 + 2 ~eE0 = 0 (189)
c
2
This system implies the expected relation k 2 = ωc2 , shows that ~e and ~b are perpendicular to
each other and fixes the relative magnitudes:
1 ~b = 1 ~k ∧ ~e → ~e · ~b = 0
B0 = E0 , (190)
c k

These are transverse waves (unlike e.g. sound). The waves travel in the direction of ~k, while
~ and B
E ~ lie in wave fronts, transverse to the direction of propagation, and are orthogonal
to each other, see figure below.

B, b i

direction of wave
K

E, e i

43
5.2 Polarization
Let us discuss the phenomenon of polarization. To simplify the expressions, take ~k in the
z-direction, ~k · ~r = kz. Look in the plane z = 0 so that the argument of the wave solutions
is simply ωt. Taking linear combinations of the solutions above we can write

~ = ~e1 cos ωt + ~e2 sin ωt

E (191)
1~ 1
~ =
B k ∧ ~e1 cos ωt + ~k ∧ ~e2 sin ωt (192)
ω ω
where ~e1 and ~e2 are two a priori independent vectors. To discuss the phenomenon of polari-
sation, we distinguish some particular cases:
(i) ~e1 , ~e2 proportional, say to ~i, then

~ = E~i cos(ωt + δ)
E (193)

~ = 1 E~j cos(ωt + δ)
B (194)
c
for some constant δ;

cB
E
x

a wave like this is said to be lineary polarised; E ~ and B ~ oscillate parallel to two fixed
orthogonal directions.
(ii) ~e1 , ~e2 orthogonal, of equal length, e.g. ~e1 = E~i, ~e2 = E~j, then

~ = E(~i cos ωt + ~j sin ωt)

E (195)

~ = E (−~i sin ωt + ~j cos ωt);

B (196)
c
a wave like this is said to be circularly polarised; E~ and B ~ rotate at a constant rate about
the direction of propagation.

44
 y

cB E

ωt
x

The direction is anti-clockwise and the wave is left circularly polarized if ~e1 ∧ ~e2 · ~k > 0, and
the rotation is clockwise and the wave is right circularly polarized if ~e1 ∧ ~e2 · ~k < 0.
(iii) the general case can be regarded as a combination of waves with different polarizations.

5.3 Reflection and refraction of electromagnetic waves

If you have ever been to a swimming pool you know about the funny properties of light as
it crosses the boundary between air and water. The basic laws of optics regarding reflection
and refraction of light can be summarized as follows:

1.- The angle of reflection equals the angle of incidence.

2.- Each material has an index of refraction n and the angles of incidence α and
refraction α0 satisfy

n sin α = n0 sin α0

For air the index of refraction is almost exactly 1, while for water is about 1.3.
Let us discuss how the laws of reflection and refraction of light at a plane surface between
two media of different dielectric properties (for instance air and water) can be deduced from
the Maxwell equations. Consider a plane surface located at z = 0 separating two media with
µ and 0 µ0 , and consider an incident wave along ~k which ”splits” into a refracted wave along
~k 0 and a reflected wave along ~k 00 , see figure:

45
 z

k k''
α α''

εμ

ε'μ'

α'
k'

For z > 0 the solution corresponds to the incident plus the reflected waves

~ 0 ei(~k.~r−ωt) + E
~ = E
E ~ 00 ei(~k00 .~r−ωt) (197)
0

~ 1 ~ ~ i(~k.~r−ωt) ~ 00 ~ 00 i(~k00 .~r−ωt) 


B = k ∧ E0 e + k ∧ E0 e (198)
ω
while for z < 0 the solution corresponds to the refracted wave

E ~ 00 ei(~k0 .~r−ωt)
~ = E (199)

~ = 1 ~ 0 ~ i(~k0 .~r−ωt)
B k ∧ E0 e (200)
ω

Here E~ 0 and ~k are arbitrary (since we can send in any wave we want) and we need to fix
~ ,E
E 0 ~ and ~k 0 , ~k 00 in terms of these, by requiring the correct boundary conditions at z = 0.
00
0 0
The wave numbers satisfy

ω ω
k = k 00 =, k0 = 0 (201)
v v
where v and v 0 are the velocities of the waves in the two media, related to the properties of
the medium by
r r
1 0 µ0 0 1 0 µ0
v=√ = c, v =√0 0 = c (202)
µ µ µ 0 µ0

Whatever conditions we have at z = 0 the spacial variation of the three waves, as we move
along the z = 0 plane, should be the same (as they should work for all x, y) hence

~k.~r |z=0 = ~k 0 .~r |z=0 = ~k 00 .~r |z=0 (203)

46
independent of the nature of the boundary conditions. In particular the three wave vectors
should lie on a plane, let’s say (x, z), and their component along the x direction should
coincide, namely

k sin α = k 0 sin α0 = k 00 sin α00 (204)

This implies
1 1
α = α00 , sin α = 0 sin α0 (205)
v v q
These are the two basic laws of optics! the refraction index is usually defined as n = 0µµ0 .
We haven’t yet imposed the precise boundary conditions at z = 0. As we cross the boundary,
remember that the normal components of E ~ and B~ are continuous and the tangential
~ −1 ~
components of E and µ B:

 
(E ~0 + E ~ 000 ) − 0 E
~ 00 · ~n = 0 (206)
 
~k ∧ E
~ 0 + ~k 00 ∧ E~ 000 − ~k 0 ∧ E ~ 00 · ~n = 0 (207)
 
~0 + E
E ~ 000 − E
~ 00 ∧ ~n = 0 (208)
  
1 ~ ~  1
k ∧ E0 + ~k ∧ E
00 ~ 0 − ~k ∧ E
00 0 ~ 0 ∧ ~n = 0
0
(209)
µ µ0
These equations can be used to compute the magnitude of the reflected and refracted fields
in terms of the incident one. For instance, assume for simplicity that the electric field is
linearly polarized along the direction perpendicular to the incidence plane (the plane formed
by ~k and ~n), namely E
~ 0, E
~0,E
0
~ 00 are all along the y direction (going out from the picture). In
0
this case, (206) imply

E0 + E000 − E00 = 0 (210)

s
0 0
r

(E0 − E000 ) cos α − E cos α0 = 0 (211)
µ µ0 0
Assuming µ = µ0 (which is almost true for instance for light from air reflected on water) and
using (205) we obtain

2 cos α sin α0
E00 = E0 (212)
sin(α + α0 )
sin(α − α0 )
E000 = − E0 (213)
sin(α + α0 )
The correct description of light and the laws of optics was one of the biggest triumphs of
electromagnetism!

47
6 Epilogue: Electromagnetism and Special Relativity
In order to apply the second law of Newton F~ = m~a, compute the path of particles under
central forces, etc, you need to choose a system of reference. However, provided the system
is inertial you can choose any system you want. In particular, your description and the
description of your classmate should be equivalent if your two systems are moving with
constant velocity with respect to each other (all inertial frames move with constant velocity
with respect to each other). This is also true for other laws of Physics, and so we can make
the following postulate:

1.- The laws of Physics are the same in all inertial frames of reference.

Should electromagnetism be any different? Imagine it is not. More precisely, if I, in my nice

stationary lab, measure forces between wires and charges, I will find out they are described
by the electromagnetism equations with some specific 0 and µ0 . Now, if your lab is moving
(at a constant velocity) and you repeat the same experiments, you should find the same
results, with the same 0 and µ0 . In particular, if we both produce electromagnetic waves,
both of us will see that these waves move at speed c = √10 µ0 . This leads to the second
postulate:

2.- The speed of light c has the same value in all inertial frames.

Let us see which consequences can have these two innocent and quite reasonable postulates.
Imagine Stella and Leonardo are using two coordinate systems O and O0 , with space time
coordinates (x, y, z, t) and (x0 , y 0 , z 0 , t0 ), which are moving at relative speed v along the z
component. See figure:

O' v
O z z'

Imagine furthermore that at the common origin of times t = t0 = 0 the two systems overlap.
Now, at t = t0 = 0 a flash of light is originated at the origin (which is common) and expands
in a spherical shape. According to postulate 2 the velocity of the wave in both systems is c,
so in you ask Stella what is the shape of this sphere she will tell you:

x2 + y 2 + z 2 − c2 t2 = 0 (214)

48
If instead you ask Leonardo what is the shape of the sphere in his system he will tell you:

x02 + y 02 + z 02 − c2 t02 = 0 (215)

There should be some way now to translate between the two systems. Galileo will tell you
that this is easy, indeed, looking at the pictures:

x0 = x (216)
y0 = y (217)
z 0 = z − vt (218)
t0 = t (219)

This is called a Galilean transformation, is the transformation you would use, and has been
used for centuries, but it does not agree with (214) and (215)!! If you assume that the
relations between the set of variables is linear and you insist on consistency with (214) and
(215) you find:

x0 = x, y0 = y (220)
z − vt
z0 = q (221)
2
1 − vc2

0 t − cv2 z
t = q (222)
2
1 − vc2

These are called the Lorentz transformations and you can see that they reduce to Galilean
transformations for velocities much smaller than c. One of the most surprising aspects of
these transformations is that the time intervals between the two systems are not the same!
and Stella (t) will look at the watch of Leonardo (t’) and think that Leonardo’s watch is
going slower!
Lorentz actually found these transformations by studying the Maxwell equations, and realiz-
ing that they were not invariant under Galilean transformations. Einstein proved that these
transformations actually followed from his postulates 1 and 2 (which however have wider
validity). In this way special relativity was born.

49