A REVIEW OF SPECIAL RELATIVITY

Contents: WHY THE “THEORY OF RELATIVITY”?

RELATIVISTIC KINEMATICS
RELATIVISTIC DYNAMICS
RELATIVISTIC TRANSFORMATIONS OF THE EM FIELDS
THE “CENTER OF MASS” ENERGY

Eliana GIANFELICE - Fermilab

1/45 < > ≪ ≫   ? i  P

Preamble
In the following we will use the notion of frame of reference where coordinates are specified.

• We will use orthogonal frames (3 axes at 900 ) where the

cartesian coordinates of a point in space are specified.

• Any frame may be made to coincide with any other by

translations and rotations.
– For this reason when considering frames attached to
moving observers we will just consider translational
motion along one common axis.
This simplifies the math.

2/45 < > ≪ ≫   ? i  P

Running in Batavia last April...

3/45 < > ≪ ≫   ? i  P

WHY THE “THEORY OF RELATIVITY”?
Quantitive description of physical events needs a frame of reference, where the coordinates of the
observed object are specified. Euclidean geometry specifies how coordinates of points in different
frames are related. For instance, if S 0 is translated by x0 wrt S along the common x̂-axis it is

x0 = x − x0 y0 = y z0 = z (1)

~ . Thus
Suppose S 0 is moving wrt S along the x̂-axis with velocity V
x0 =V t (assuming O and O 0 coincide at t=0)and making the first
and second derivates wrt time

ẋ0 = ẋ − V ẏ 0 = ẏ ż 0 = ż (2)
ẍ0 = ẍ ÿ 0 = ÿ z̈ 0 = z̈ (3)
Eqs. 1, 2 and 3 are the Galilean transformations for coordinates, velocity
and acceleration. We implicitly assumed that t0 =t and that the lengths
were invariant in the two frames.
• Eq.2 means that velocities add.
• Eq.3 says that the acceleration is invariant.

4/45 < > ≪ ≫   ? i  P

The basis of the classical mechanics are the three lawsa of dynamics.
The first dynamics law is the principle of inertia (Galileo) which states
“A free body remains in a state of rest or of uniform motion”
A reference system where a free body is at rest or it moves with uniform velocity is said to be inertial.
Because of the Galileian transformations, any frame in uniform motion wrt an inertial one is inertial
too.
The second law (Newton) states
“In an inertial system holds good ~ = m~
F a”
The variation of velocity with time (acceleration), ~ ~ , through
a, is proportional to the applied force, F
a constant, m (“inertial mass”).
Implicitly it is assumed that m is a characteristic of the body which doesn’t depend upon its motion.
The third Newton law states
“Whenever two bodies interact the force that body 1 exerts on body 2 is
equal and opposite to the force that body 2 exerts on body 1”
Third law combined with the second one gives the momentum conservation lawb for a closed system.

a Physics laws are not mathematical axioms but statements based on reproducible observations.
b Momentum: ~ ≡ m~
p v

5/45 < > ≪ ≫   ? i  P

The three laws of dynamics hold good in inertial frames. As they are all equivalent it is reasonable to
assume that all mechanics laws are the same for inertial observers.
This is expressed by the principle of relativity:
“Mechanics laws are the same for all inertial observers”.
Suppose that Alex (A) is studying the motion of a ball let to fall under
the earth gravitation force. A measures that the object is subject to a
constant acceleration of a ≈9.8 ms−2 . By using different balls he finds
that the acceleration is always the same, g. A concludes that there
must be a force acting on the balls which is directed towards the center
of the earth and has magnitude mg.

Assuming that Galileian transformations hold good, observer Beth (B) on a train moving with uniform
~ =x̂V wrt A will describe the ball motion as
velocity V

ẋ0 = ẋ − V = −V ẏ 0 = ẏ
ẍ0 = ẍ = 0 ÿ 0 = ÿ

and as the mass, m is a constant, will agree with A on magnitude and direction of the force.
Galileian transformations satisfy the the principle of relativity!
The relativity principle allows us to chose the most convenient frame for describing an event.

6/45 < > ≪ ≫   ? i  P

Is EM invariant under Galilean transformations?
In the second half of the XIX century Maxwell had summarized the whole EM phenomena into 4
differential equations containing the constant c, from which one finds the wave equation for fields and
potentials.

∂2 1 ∂2
 
Simplest case: − Φ=0
∂x2 c2 ∂t2
The constant c is the velocity of propagation of the wave and
is numerically equal to the speed of light in vacuum.
• Because of the addition of the velocities it is weird that
it is a constant, unless we assume it is the velocity wrt a
propagation medium. In Maxwell own words:
“We can scarcely avoid the inference that light
consists in the transverse undulations of the same
medium which is the cause of electric and magnetic
phenomena.”
The supporting medium was named “ether ”.

7/45 < > ≪ ≫   ? i  P

• This would mean also that Maxwell equations hold good only in that frame. Indeed s the wave
equation

∂2 1 ∂2
 
− Φ=0
∂x2 c2 ∂t2

becomes under Galileian transformation

∂2 1 ∂2 V 2 ∂2 ∂2
 
V
− − −2 Φ=0
∂x02 c2 ∂t02 c2 ∂x02 c2 ∂x0 ∂t0
– The equation is not invariant. EM laws are written in a frame connected to the ether!

8/45 < > ≪ ≫   ? i  P

Hypotheses
1. The relativity principle holds good only for the mechanics; for the EM exists a preferred frame of
reference where the speed of the light is c (the reference system where the ether is at rest).
• A bunch of experiments (starting with the famous Michelson-Morley experiment in 1887) aiming
to prove the existence of the ether failed. Their results suggested instead that the speed of the
light was a constant non dependent upon the status of motion of source or observer.
2. The at the time relatively young EM laws are wrong.
• Attempts at modifying the EM in such a way that it would be invariant under Galilean trans-
formations led to predictions of new phenomena which couldn’t be proven by experiments.
3. EM laws are correct, but the Galilean transformations (and mechanics laws) must be modified.
Because of experimental evidence, only the third hypotheses was left.
If the Galilean transformations which look so self-evident were wrong a rethinking of physics basics was
necessary.

9/45 < > ≪ ≫   ? i  P

Relativity of time
In the “Annus Mirabilis” 1905 Einstein published 4 fundamental papers. The third of them contained
the idea of relativity of time and the basis of the theory of special relativity.

The paper starts, on the basis of the exerimental evidence, by giving up the existence of ether and
introducing instead a “Principle of Relativity” based on two postulates
1) Physics laws are the same in all inertial reference systems, there is no preferred reference system.
2) The speed of the light in the empty space has the same finite value c in all inertial reference systems.

10/45 < > ≪ ≫   ? i  P

The paper goes on demonstrating that it is not possible to synchronize clocks attached to frames in
relative motion.
To find out whether two clocks at rest in different
locations of an inertial frame are synchronized we
proceed as follows. The observer A has a clock and
sends a light ray at time tA to observer B which
receives it at tB .
A mirror reflects the light back to A which receive
it at t0A .
Because of the second postulates, the clocks are
synchronized if

tB − tA = t0A − tB or ∆tA→B = ∆tB→A

If the clocks are identical they stay synchronized.
Any inertial observer can synchronize its clocks by the same procedure. How the time measured by
observers in relative motion are related?

11/45 < > ≪ ≫   ? i  P

Let’s look for instance to the synchronizing oper-
ation for two clocks attached to the ends, A and
B, of a rod moving along the x-axis as seen by a
stationary observer.
While the light moves to B, B moves further and
once reflected back to A, A moves toward the light.

Therefore for the resting observer the time needed to reach B, tB , is obtained by setting

ctB = L + V tB → tB = L/(c − V )

while the time needed to reach A is obtained from

ctA = L − V tA → tA = L/(c + V )

 2V L
∆tB→A − ∆tA→B = 2 2]
6= 0 consequence of c being finite!
c
 [1 − (V /c)

If the clocks in the moving frame would be synchronous with the stationary ones they wouldn’t be
synchronous in their own frame. The “stationary” frame would dictate the timing. However stationarity
is relative, the inertial frames are all equivalent: if there exist no privileged frame, we must abandon
the idea of universal time.

12/45 < > ≪ ≫   ? i  P

Lorentz transformations “abridged”
By assuming the speed of light constant in all reference systems, the Galilean transformations, implying
the velocity addition rule, must be modified. The new transformations must reduce to the Galileian
ones when the relative motion is slow (V  c). According to the first Einstein postulate, the empty
space is isotrope (all direction are equivalent) and homogeneous (all points are equivalent); it would
make no sense to postulate that the laws are invariant in a space which is not homogeneous and
isotrope. Homogeneity implies linearity:

x0 = a11 x + a12 y + a13 z + a14 t

y 0 = a21 x + a22 y + a23 z + a24 t wow:
z 0 = a31 x + a32 y + a33 z + a34 t 16 unknowns!
t0 = a41 x + a42 y + a43 z + a44 t

13/45 < > ≪ ≫   ? i  P

For the case we are considering of motion along the common x-axis the coordinates y and z do not
play a role and therefore it is reasonable to write

x0 = a11 x + a14 t
y0 = y
z0 = z
t0 = a41 x + a44 t

The origin of the S 0 frame is described in S by x0 = V t and by definition it is x00 =0 at any time.
Therefore
0 = x00 = a11 x0 + a14 t = a11 V t + a14 t
that is a14 /a11 = −V and

x0 = a11 (x + a14 t/a11 ) = a11 (x − V t)

14/45 < > ≪ ≫   ? i  P

For finding the values of the remaining 3 coefficients we resort to the fact that, according to the two
postulates, the speed of light is the same in S and S 0 and that the wave equation is invariant in form.
Suppose an EM spherical wave leaves the origin of
the frame S at t=0.
The propagation is described in S by the equation
of a sphere which radius increases with time as

R2 = x2 + y 2 + z 2 = c2 t2 (4)

In S 0 the wave propagates with the same speed c

and therefore

x02 + y 02 + z 02 = c2 t02
which expressing the primed coordinates in terms of those in S becomes

a211 x2 + a211 V 2 t2 − 2a11 xV t + y 2 + z 2 = c2 a241 x2 + c2 a244 t2 + 2a41 a44 xt (5)

Comparing Eqs. 4 and 5 we get a system of 3 equations in the 3 unknown a11 , a41 and a44 .

15/45 < > ≪ ≫   ? i  P

 8

Solution a11 = a44 = γ

7

a41 = −γβ/c 5

γ
with 3

2
1
β ≡ V /c (0÷1) and γ≡ p (1÷∞) 1

1 − β2 0
0 0.2 0.4 0.6 0.8 1
β

~ = x̂V
The coordinate transformation for a translational motion along x̂ with (constant) velocity V
are (Lorentz transformations)

0 t − xV /c2
t = p ≡ γ(t − V x/c2 )
1 − V 2 /c2
0 x−Vt
x = p ≡ γ(x − V t)
1 − V 2 /c2

y0 = y z0 = z
The inverse transformations are obtained replacing V with −V .

16/45 < > ≪ ≫   ? i  P

 0 t − xV /c2
t = p ≡ γ(t − V x/c2 )
1 − V 2 /c2
x−Vt
x0 = p ≡ γ(x − V t)
2
1 − V /c 2

y0 = y z0 = z
• V  c the Lorentz transformations reduce to the Galilean ones. Good!
• For V > c the transformations are meaningless because the argument of the square root,
1 − V 2 /c2 , becomes negative!
• The existence of a signal with V > c would yield to a violation of the causality principle, as we
will see.

17/45 < > ≪ ≫   ? i  P

Time is one of the 4 coordinates describing an event and as the spatial coordinates is subject to a
(Lorentz) transformation between moving frames.
For spatial coordinates it is always possible if for instance x2 > x1 to find a new coordinates system
such that x02 < x01 .
Is it possible to find a Lorentz transformation which inverts the temporal order of events?
Assume an event happening at the time t1 at the location x1 in S and a second event happens at t2
in x2 with t2 > t1 . Is it possible to find a Lorentz transformation such that t02 < t01 ? In S 0 it is

ct01 = γ(ct1 − βx1 ) and ct02 = γ(ct2 − βx2 )

c(t02 − t01 ) = γ[c(t2 − t1 ) − β(x2 − x1 )]

Therefore t02 < t01 if β(x2 − x1 ) > c(t2 − t1 ), that is if V (x2 − x1 )/(t2 − t1 ) > c2 . This may
be possible depending on the values of x2 − x1 and t2 − t1 . However if the first event in S drives
the second one, x2 and t2 are not arbitrary.
If w is the speed of the signal which triggers the second event from the first one it is

x2 − x1 = w(t2 − t1 )
w ≤ c and V < c
0 0
 V w ↓
c(t2 − t1 ) = γ[c(t2 − t1 ) − βw(t2 − t1 )] = γc(t2 − t1 ) 1 − 2 > 0
c
Causality is not violated.

18/45 < > ≪ ≫   ? i  P

Some consequences of Lorentz transformations:
time dilation and length contraction
Suppose a clock at rest in S measuring a time interval t2 − t1 between two events happening at the
same location, x1 =x2 . The time interval in the moving frame S 0 is measured by two different clocks
because according to Lorentz transformations, the events happen in S 0 in different locations. The time
difference in S 0 is proper time (time measured by the same clock)
.
t02 − t01 = γ(t2 − t1 ) ≥ t2 − t1 time dilation

Events happening at the same time, t1 =t2 , but in different places in S, will be no more simultaneous
in the moving frame S 0
c(t02 − t01 ) = γβ(x1 − x2 ) 6= 0
Consider a rod of length L0 along the x-axis and at rest in the moving frame S 0 .
The length in S is determined by the position of the rod ends at the same time (t1 =t2 ) and therefore
length at rest
↓
L0 = x02 − x01 = γ(x2 − x1 ) = γL → L = L0 /γ length contraction

However the length of a rod aligned with one of the two axis perpendicular to the direction of motion
is invariant. Angle are in general not invariant.

19/45 < > ≪ ≫   ? i  P

Warning!
• Speaking of time dilution: it is with respect to the frame where the time is measured by the same
clock (proper time).
– The statement “moving clock are slower” means moving with respect to the frame where a
single clock is needed for the time measurement.
• Length contraction is with respect to the frame where the object is at rest (proper length).

20/45 < > ≪ ≫   ? i  P

The transformation for the components of the velocity, u
~ , are obtained from the coordinate transfor-
mations
0 dx0 dx − V dt ux − V
ux ≡ = =
dt0 dt − V dx/c2 1 − ux β/c
0 dy 0 dy uy
uy ≡ = =
dt0 γ(dt − V dx/c2 ) γ(1 − ux β/c)
0 dz 0 dz uz
uz ≡ = =
dt0 γ(dt − V dx/c2 ) γ(1 − ux β/c)
with ux ≡ dx/dt, uy ≡ dy/dt and uz ≡ dz/dt.
Remember β (=V /c) refers to the motion of the frame.
• As time is not invariant, despite the lengths perpendicular to the motion direction being unchanged,
the time needed to cover them is different.
– Unlike classic kinematic, the velocity components perpendicular to the motion,
unless vanishing, are affected by the motion of the frame.
• For ux =c and uy =uz =0 it is
c−V c−V
u0x = =c =c and u0y = u0z = 0
1 − V /c2 c−V
For uy =c and ux =uz =0 it is u0x =−V , u0z =0 and
u02 02 2 2 2
x + uy = V + c (1 − (V /c) = c
2

21/45 < > ≪ ≫   ? i  P

In a similar way as for the velocity, it is possible to find the transformations for the acceleration ~
a
ax
a0x = 3
γ (1 − ux β/c)3

ay ax uy β/c
a0y = +
γ 2 (1 − ux β/c)2 γ 2 (1 − ux β/c)3
az ax uz β/c
a0z = +
γ 2 (1 − ux β/c)2 γ 2 (1 − ux β/c)3

• Acceleration in general is not invariant under Lorentz transformations.

22/45 < > ≪ ≫   ? i  P

 A historical curiosity
The relativistic transformations were named by Poincaré after the dutch physicist
Hendrik Lorentz who introduced them, before Einstein paper.
Lorentz had discovered that those transformations leave Maxwell equations invariant.
He had also introduced the notion of “local time” and of “contraction of bodies” for explaining the
negative results of the Michelson-Morley experiment because he was convinced, as many other leading
scientists, of the validity of the ether theory.
It seems that Einstein was not aware of Lorentz work... Anyway Einstein gave to the transformations a
deep physical meaning making them extendable also to mechanics and causing a revolution of classic
dynamics.

23/45 < > ≪ ≫   ? i  P

Experimental evidence of relativistic kinematics
Light aberration
Light aberration is the apparent motion of a light source due to the
movement of the observer. It was first discovered in astronomy.
Source emitting photons at an angle θ wrt to the x-axis in the S
frame where uy =c sin θ and ux =c cos θ.
In S 0 it is u0y =c0 sin θ 0 and ux =c0 cos θ 0 .
Using Galileian transformations for the velocity components
u0y = uy and u0x = ux − V
400
class.
0
tan θ = u0y /u0x = uy /(ux − V ) 350 rel.
β=0.2
300
sin θ
tan θ 0 = 250

Θ’[deg]
(cos θ − β) 200
β=0.9
150

Using instead Lorentz transformations 100

50

0 sin θ 0
0 50 100 150 200 250 300 350 400
tan θ = Θ [deg]
γ(cos θ − β)

High energy experiments involving emission of photons confirm the relativistic expression.

24/45 < > ≪ ≫   ? i  P

Lifetime of unstable particles

1
τ=0.1 s
− 0.9 τ=0.05 s
Beside e , p and n, in nature there are particles which are 0.8
0.7
produced by scattering process and unlike e+ , p̄ and n̄, are 0.6

N(t)
0.5

“short-living”. Their number decays in time as 0.4

0.3
0.2
−t/τ 0.1
N (t) = N0 e 0
0 0.1 0.2 0.3 0.4 0.5
t[s]

The lifetime of charged pions at rest is τ0 =26×10−9 s. Time needed for the pions at rest to decay
by half
N0
N (t) = N0 e−t/τ = → t = 18 ns
2
They are produced by bombarding a proper target by high energy protons and leave the target with
v ≈2.97×108 m/s that is β=0.99 and γ ≈7. It is observed that they are reduced to the half after
37 m from the target. If their lifetime would be as at rest they should become the half already after
about 5 m.
The experimental observation is explained if the pion lifetime in the laboratory frame is
τ = γτ0
as predicted by time dilation.
Time dilation may allow us realizing future colliders smashing muons!

25/45 < > ≪ ≫   ? i  P

Doppler effect
Doppler effect exists also classically: we experience it when we
hear the siren of a police car or an ambulance. The frequency
perceived by an observer at rest is higher when the car is
approaching because the number of the acoustic wave knots
per unit time is larger, while the frequency decreases when the
source is moving away. Classically there is no “transverse”
Doppler effect: in the moment the car is at the minimum
distance it is ∆f =0.
Relativistically for a light wave the situation source (S) or receiver (R) in motion are identical.
When the angle, θ, between wave propagation direction and motion is 0 the frequency is
s
1+β
f = f0 with β > 0 for R and S approaching, β < 0 when they move away
1−β

In addition because of the time dilation there is also a transverse (θ=900 ) Doppler effect

f = f0 /γ

This was predicted by Einstein who suggested an experiment using hydrogen ions for measuring it. The
experiment realized for the first time by Ives and Stilwell in 1938 proved the correctness of Einstein
prediction.

26/45 < > ≪ ≫   ? i  P

Relativistic Dynamics
Assuming F~ invariant and m constant, Newton law, F ~ = m~
a, is not invariant under Lorentz trans-
formations because we have seen that ~
a is not invariant.
In addition mass can’t be a constant because by applying a constant force to an object its speed would
increase indefinitely becoming larger than c.
• Classical mechanics must be modified to achieve invariance under Lorentz transformations.
• The new expressions must reduce to the classical ones for v/c 1.
In the 1905 paper, Einstein used the Lorentz force and the electro-magnetic field transformations to
achieve the generalization of the definition of momentum and energy.

27/45 < > ≪ ≫   ? i  P

In 1909 two MIT professors of chemistry, Lewis and Tolman, suggested a different more straightforward
approach involving purely mechanical arguments.

Let’s assume there are two observers,

Alex and Betty, moving towards each
other with the same speed as seen by a
third observer, Charlie.
Betty sits in S and Alex in S 0 .
Alex and Betty have identical elastic
balls.

28/45 < > ≪ ≫   ? i  P

Betty releases the red ball with uB B
x =0 and uz =uz 6= 0, while Alex
releases the green ball with speed u0A 0A
x =0 and ūz numerically equal
and opposite to the red ball
 velocity,
 that
 is 

u0A
z = −u B
z
%   -
Green ball in S 0 Red ball in S
The experiment is set so up that the two balls collide and rebound.
Now let’s consider Betty point of view. For Betty it is

∆pB
x = 0 ∆pB B
z = 2mB uz

∆pA
x = 0 ∆pA A
z = 2mA uz

We need here the inverse velocity transformation because we know the

numerical value of the z direction component in the moving frame S 0
u0z
uz =
γ(1 + u0x β/c)

In our case u0x =0 and u0z =−uB

z and therefore

0A 1
uA B
z = uz /γ = −uz /γ with γ = p
1 − (uA
x /c)
2

29/45 < > ≪ ≫   ? i  P

Momentum, classically defined as p
~=m~
v , is conserved if

∆pB
z = −∆p A
z

that is
1
m B uB A
z = −mA uz = mA uB
z → mA = γmB
γ

We may assume that uB z is small so that mB is the mass at rest, m0 , and mA =m(v).
So we have found that
m = γm0

We can keep the momentum definition from classic dynamic by giving up the invariance of mass.
Relativistically mass is not conserved.
A clear example is the annihilation of a e+ e− pair into 2 photons.
Let’s try modifying the classic Newton law
d~
p d~
v
~ =
F =m
dt dt
into
d~
p d dγ d~
v
~ =
F = (γm0~
v ) = m0 ~
v + m0 γ ~ and ~
F a are not parallel!
dt dt dt dt

30/45 < > ≪ ≫   ? i  P

By scalar multiplication by ~
v it is

d~
p
~ ·~
F v=~ v·
% dt
work/unit time = dE/dt

d~
p d~
v v2 3 dv v2 γ 2 3 dv
v·
~ v·
= m0 γ~ + m0 γ v = m0 γv(1 + ) = m0 γ v
dt dt c2 dt c2 dt
that is
dE dv
~ ·~
=F v = m0 γ 3 v
dt dt
It is easy to verify that this equation is satisfied by defining the energy as

E = mc2 = γm0 c2

For v=0 it is E0 =m0 c2 which is the energy at rest.

The (relativistic) kinetic energy is obtained by subtracting the rest energy from the total energy
1
T = mc2 − m0 c2 = m0 c2 (γ − 1) 6= γm0 v 2
2
which gives the classical kinetic energy T ' m0 v 2 /2 for v  c.

31/45 < > ≪ ≫   ? i  P

Kinetic energy measurement
Experiments confirmed the validity of the relativistic relationship between p
~ and ~
v.
Bertozzi experiment measured directly the velocity of e− accelerated in a linear accelerator.
• e− speed was measured through the time of flight.
• Kinetic energy relied on the knowledge of the accelerating field and on the measurement of the
heat deposited at the aluminum target.
The results also show clearly the presence of a limit speed, c.

me = e− rest mass

32/45 < > ≪ ≫   ? i  P

Importance of relativity for accelerators

3
cp[GeV]
frev[MHz]
2.5

Example of CERN PS Booster.

2
Circumference: L=157 m.
1.5
Particles are injected from Linac4 with
1
T =160 MeV and accelerated to T =2 GeV.
0.5
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
T[GeV]

• The dipole field must be ramped up according to momentum for keeping the particles on the design
orbit (ρ=p/eB).
c
• frf = hfrev . For large γ it is frev ≈ h L (1 − 2γ1 2 )
→ almost constant at high energy as the speed approaches c.
– Particularly true for e± which have 1836 larger γ for the same energy.
Relativity has basic relevance for accelerators!

33/45 < > ≪ ≫   ? i  P

Transformations of momentum, energy and force
The transformations for the momentum follows from the definition p
~=m~u and from the transformations
for the velocity. The result is
 1 
0
px = γV (px − E V /c2 ) with γV = p 
 1 − V 2 /c2
% 
γu m 0 c 2 -
0 0 frame speed
py = py pz = pz
E − V px
E0 = p
1 − V 2 /c2
• The transformations have the same form as for the coordinates transformations with
r→p
~ ~ and t → E/c2
• Relativistic energy and momentum are closely connected.
– The quantity (E/c)2 − (p2x + p2y + p2z ) is invariant. Indeed from the definitions
(m0 γ)2 c2 − (m0 γ)2 (u2x + u2y + u2z )=m20 γ 2 (c2 − u2 )=m20 γ 2 c2 (1 − β 2 )=m20 c2 =const.
• If energy and momentum are conserved in one inertial frame of reference they are conserved in all
inertial frames.
• If momentum in conserved in two inertial frames, energy too is conserved in both frames.

34/45 < > ≪ ≫   ? i  P

The transformations for the force are

uy V uz V
Fx0 = Fx − Fy − 2 Fz
c2 − u x V c − ux V
p
0 1 − V 2 /c2
Fy,z = Fy,z
1 − ux V /c2

~ 0 =F
For V  c it is F ~ which is the classic result.

If the force is acting on a particle which is instantaneously at rest in S (u=0), the transformations
simplify

1 1
Fx0 = Fx Fy0 = Fy Fz0 = Fz
γ γ
If the particle is subject to a force, the frame where it is at rest can’t be inertial!
However there is always one inertial frame where it is “instantaneously at rest”.

35/45 < > ≪ ≫   ? i  P

Transformations of EM fields
The transformations are found by applying the force transformations to the force experienced by a
charged particle moving with velocity u
~ in an EM field
~ = qE
F ~ + q~ ~
u×B (Lorentz force)

In the moving frame S 0 it must have the same form


~ ~
F = qE 0 + qu
0 ~0
~0 × B
%
q 0 =q in agreement with measurements

If the particle is instantaneously at rest in S 0 (u0 =0) it is

Fx = Fx0 Fy = Fy0 /γ Fz = Fz0 /γ
~ 0 because u
In S 0 there is no effect from B ~ 0 =0 → F ~ 0 = qE ~0
We choose for simplicity the frames orientation so that V =ux
and uy =uz =0
~ = x̂(uy Bz −uz By )+ŷ(−ux Bz +uz Bx )+ẑ(ux By −uy Bx ) = −ŷ V Bz +ẑ V By
~ ×B
u

→ The magnetic force has no x-component that is Fx =qEx .

36/45 < > ≪ ≫   ? i  P

 qEx = Fx = Fx0 = qEx0 → Ex0 = Ex
↑
force transformation
1 1
qEy − qV Bz = Fy = Fy0 = qEy0 → Ey0 = γ(Ey − V Bz )
γ γ
1 1
qEz + qV By = Fz = Fz0 = qEz0 → Ez0 = γ(Ez + V By )
γ γ
Finding the magnetic field transformations is more complicated because there is no frame where the
electric force vanishes. The result is
0 0
 V 
0
 V 
Bx = Bx By = γ By + 2 Ez B z = γ B z − 2 Ey
c c

Denoting by “parallel” and “normal” the fields components wrt to direction of motion the field trans-
formations can be written in the general form

Ek0 = Ek Bk0 = Bk

0 ~ +V
~ × B)
~ ⊥ 0 ~ −V
~ × E/c
~ 2 )⊥
E⊥ = γ(E B⊥ = γ(B
~ mus be replaced by −V
For the inverse transformations V ~.

37/45 < > ≪ ≫   ? i  P

Transformation of Source Distributions
Let us consider a distribution of charges at rest in S 0 . The
charge density is given by

0 0 0 0 0 qN
ρ (x , y , z , t ) =
dx0 dy 0 dz 0
In S, moving with velocity −V wrt S 0 , the volume element

is dx0 0 0
dx dy dz = dy dz
γ
 - length contraction
Charge density in S
qN
ρ= = γρ0
dx dy dz

As the charge distribution moves in S with velocity +x̂V , in S there is a current with density

jx = ρV = γρ0 V (in general: ~j = ρV

~ = γρ0 V
~)

There is an analogy with E/c (or mc) and p

~

ρ → m and ~j → p
~

The quantity (ρc, ~j) transforms according to the Lorentz transformations.

38/45 < > ≪ ≫   ? i  P

Invariance of Maxwell Equations
Knowing how fields and sources transform one can prove that Maxwell equations are invariant under
Lorentz transformation. This was demonstrated by Lorentz before Einstein formulated the special
relativity theory.
We want to show that if the Maxwell equations hold good in S, they hold with the same form also in
S0.
For example let’us prove that
∂Ex ∂E ∂Ez
+ ∂yy +
∂x ∂z
.
ρ 0 ρ0
~ =
∇·E ⇒ ∇ ·E = ~0
0 0
The partial derivatives in S 0 and in S are related by the cyclic rule
 
∂ ∂ct ∂ ∂x ∂ ∂y ∂ ∂z ∂ ∂ ∂
= + + + =γ +β
∂ct0 ∂ct0 ∂ct
∂ct0 ∂x ∂ct0 ∂y ∂ct0 ∂z ∂ct ∂x
 
∂ ∂ct ∂ ∂x ∂ ∂y ∂ ∂z ∂ ∂ ∂
0
= 0
+ 0
+ 0
+ 0
=γ β +
∂x ∂x ∂ct ∂x ∂x ∂x ∂y ∂x ∂z ∂ct ∂x
∂ ∂ ∂ ∂
= =
∂y 0 ∂y ∂z 0 ∂z

39/45 < > ≪ ≫   ? i  P

By using these expressions, the field transformations and the fact that Maxwell equation hold good in
S, we find

0 ∂Ex0 ∂Ey0 ∂Ez0

∇ ·E = ~0 + +
∂x0 ∂y 0 ∂z 0
∂Ex0 ∂Ey0 ∂Ez0 ∂Ex0
=γ + + + γβ
∂x ∂y ∂z ∂ct
∂Ex ∂Ey ∂Ez ∂Bz ∂By ∂Ex
=γ +γ +γ − γV + γV + γβ
∂x ∂y ∂z ∂y ∂z ∂ct
 
∂Bz ∂By ∂Ex
~
= γ∇ · E − γV − + γβ
∂y ∂z ∂ct
ρ

1 ∂E~
=γ − γV ∇ × B ~ −
0 c2 ∂t x
ρ jx
=γ − γV
0 0 c 2
γ
= (ρc − βjx )
0 c
ρ0
=
0

40/45 < > ≪ ≫   ? i  P

THE CM ENERGY
The center of momentum for an isolated ensemble of particles is defined as the inertial frame where it
holds
m0,i~vi
p 
X X
p
~i = =0
1 − V /c2 2
i i 
-
frame velocity
We have seen that (E/c)2 − |~ p|2 = m20 c2 .
For the total energy and momentum of the ensemble
 X  X
E= Ei and  P~ = p
~i
%
total energy i ↑ i
total momentum
the invariant is easily evaluated in the CM frame
energy in CM
X 2 X X  . 2
X
Ei /c − ~i ·
p p
~i = Ei0 /c

i i i i

Let us consider two simple cases:

a) two ultra-relativistic particles colliding “head-on”;
b) one ultra-relativistic particle hitting a particle at rest.

41/45 < > ≪ ≫   ? i  P

For the system of two particles it is
(E10 + E20 )2 (E1 + E2 )2
= − (~ ~2 ) · (~
p1 + p p1 + p
~2 )
c2 c2
(E1 + E2 )2
= − p21 − p22 − 2~
p1 · p
~2
c2
Moreover for ultra-relativistic particles it is
E
p = mv ' mc =
c

a) ~1 /p1 = −~
p p2 /p2

(E10 + E20 )2 E12 E22 E1 E2 E12 E22 E1 E2 E1 E2

= + +2 − − +2 =4
c2 c2 c2 c2 c2 c2 c2 c2
and thus
p
E10 + E20 = 2 E1 E2

LHC (p/p): E1 =E2 =6.5 TeV → energy in the center of mass E10 + E20 =2×6.5=13 TeV.
HERA (p/e± ): E1 =920 GeV and E2 =27.5 GeV → E10 + E20 =318 GeV.

42/45 < > ≪ ≫   ? i  P

b) ~2 = 0 and E2 = m0,2 c2
p

(E10 + E20 )2 (E1 + E2 )2

= − p21 − p22 − 2~
p1 · p
~2
c2 c2

(E10 + E20 )2 E12 E22 E1 E2 E12 E22 E1 E2

= + +2 − = +2
c2 c2 c2 c2 c2 c2 c2
and therefore
p q p
(E10 + E20 ) = 2
E2 (E2 + 2E1 ) = E2 (m0,2 c + 2E1 ) ' 2E1 E2

For example, with E2 = 0.938 GeV (proton rest mass) to get in the CM an energy of 318 GeV must
be E1 =54 TeV.
From this example we see the advantage of collider experiments wrt. fixed target ones
(intensity permitting).

43/45 < > ≪ ≫   ? i  P

A summary of some useful (?) relationships
s
1 v 1
γ≡ p β≡ = 1−
1 − (v/c)2 c γ2
 2
m0~
v v p2
m = γm0 p
~ = γm0~
v= p =
1− (v/c)2 c (m0 c)2 + p2
2 2 E m0 γc2
E = mc E0 = m 0 c = =γ
E0 m0 c2
T = E − E0 = m0 γc2 − m0 c2 = m0 c2 (γ − 1)

2 2 2 4 m20 c4 m20 c4
E = (T + E0 ) = m c = m20 γ 2 c4 = =
1− (v/c)2 1 − p2 /(m20 c2 + p2 )
m20 c4
= (m20 c2 + p2 ) = m20 c4 + c2 p2
m20 c2
E E
cp = cγm0 v = cm0 v = cm0 v = βE cp ' E for β→1
E0 m0 c2

44/45 < > ≪ ≫   ? i  P

References
[1] R. Resnick, “Introduction to Special Relativity”, John Wiley & Sons, 1968.
[2] R. P. Feynman, “Lectures on Physics”, vol. I, Addison-Wesley, 1963.

45/45 < > ≪ ≫   ? i  P