p.

1 Eamples of physical 4-vectors

1 Some four-vectors, old and new

We will now introduce a few four-vectors. A few of these will be related
to a moving object or particle (4-analogs of velocity, acceleration, etc). To
prepare for defining these, we will first review our understanding proper
time (Subsection 1.1), with particular reference to a moving object.
Once we have identified a 4-vector, we can also identify a corresponding
Lorentz scalar or invariant: its norm. These invariants sometimes have
easy interpretations, but not always.

1.1 Preparation: proper time and time derivatives

We will construct several 4-vectors describing the kinematics or dynamics
of a moving body, such as the 4-vector analogs of ordinary displacement
~r, velocity ~u, acceleration ~a, momentum ~p, etc.
In ordinary mechanics, derivatives with respect to time (rates of change)
are important: the velocity is the temporal derivative of displacement,
the acceleration is the temporal derivative of velocity, etc. In nonrela-
tivistic mechanics, time is invariant (frame-independent). However, when
we consider corresponding derivatives in relativistic mechanics, time will
have to be specified relative to a particular frame.
It is useful to consider the time as measured from an inertial frame that
is, at the moment of consideration, attached to the object, i.e., relative to
which the object is at rest. (If the object has nonzero acceleration, the iner-
tial frame cannot remain attached to it, i.e., the object cannot remain at rest
relative to the frame. However, at any point of the history of the object,
one can identify an inertial frame relative to which the object is at rest.)
This is known as the rest frame of the object.
If two nearby events occur at the location of the object, then the time inter-
val between them as measured in the rest frame of the object is the proper
time interval.
If the object moves with velocity ~u relative to the laboratory frame, then
the proper time interval between two events on the object is
q
cdτ = c2 (dt)2 − (dx )2 − (dy)2 − (dz)2 .

This is the time interval measured from a frame relative to which the object
pframe, (dx, dy, dy) is the displacement of the
is at rest. Relative to the lab
object in time dt, thus u = (dx )2 + (dy)2 + (dz)2 /dt. Therefore

dt
q p
cdτ = c2 (dt)2 − u2 (dt)2 = cdt 1 − u2 /c2 =⇒ dτ =
γu
p. 2 Eamples of physical 4-vectors

If the motion of the object is uniform, one could write this in terms of finite
differences:
∆t
∆τ =
γu

1.2 Space-time or 4-displacement

Space-time is the 4-vector we have been dealing with from early on. A
space-time 4-vector is the coordinate of an event

(ct, x, y, z)

or an ‘interval’ between events:

(c∆t, ∆x, ∆y, ∆z) or (cdt, dx, dy, dz).

Each event is a point in Minkowski space, thus a spacetime 4-vector is a

‘displacement’ in Minkowski space. It is the 4-vector analog of the dis-
placement between spatial points in ordinary 3-space. So we could call it
a 4-displacement.
The norm of every 4-vector is a scalar, i.e., a Lorentz invariant. Consider
⇒
the 4-displacement ∆ x = (c∆t, ∆x, ∆y, ∆z) representing the interval be-
tween two events. Its norm is
 ⇒  ⇒
∆ x ? ∆ x = c2 (∆t)2 − (∆x )2 − (∆y)2 − (∆z)2

which is the Lorentz-invariant interval between the events, as we estab-

lished very early in the semester.
⇒
If the interval ∆ x is time-like, i.e., if the norm is positive, then the norm
could be written as c2 (∆τ )2 , where τ is the proper time interval between
the events. If the norm is negative (the interval is space-like), then there
is no frame in which the two events are equilocal, so it would not make
sense to talk about “proper time”.

1.2.1 Notation blues

⇒
Above, I have used the notation x = (ct, x, y, z) = (ct,~r ) for the space-
time coordinate 4-vector. Remember the warning that the double-vector
notation is not standard, and only invented for these notes.
More conventional notation would be to simply use x for the spacetime
coordinate. Then you might write

x = (ct, x, y, z).
p. 3 Eamples of physical 4-vectors

This has the unpleasant feature that the same symbol x means different
things on left and right sides of the equation.
It is also common to refer to the spacetime coordinate as x µ . You might
write
x µ = (ct, x, y, z).
You might find this also a bit strange: does x µ refer to a particular compo-
nent of the 4-vector depending on the value of µ, or does it represent the
complete 4-vector? You have to get used to x µ sometimes representing the
µ-th element of the 4-vector, and sometimes the 4-vector itself.
There is no standard notation that is completely satisfactory. You will en-
counter both of these mildly inconsistent manners of describing 4-vectors.
We will gradually wean ourselves off the double-vector notation and move
to these more standard notations, and hope that from the context you can
always figure out which symbol refers to what.

1.3 4-velocity
In non-relativistic mechanics, the velocity of an object is obtained by taking
d~r
the derivative of displacement: ~u = .
dt
Since the displacement (position) vector now appears as components of a
4-vector (spacetime), we could think of defining a velocity 4-vector as the
temporal derivative of the spacetime coordinate:
⇒ ? d
U= (ct,~r ) = (c, ~u).
dt
One can show (through a somewhat painful calculation) that this combi-
nation does not transform like a 4-vector. More easily, one observes that
c2 − |~u|2 is not Lorentz invariant and hence is not the norm of a 4-vector.
To see what went wrong, we write this is

d 1
(ct,~r ) = lim (c∆t, ∆~r )
dt ∆t→0 ∆t

Here ∆~r is the displacement of the particle in time interval ∆t, as seen from
some frame which is not the rest frame of the particle. We know that ∆t is
not a Lorentz scalar. Since we are dividing a 4-vector by a number which is
not a scalar, we cannot expect to find a 4-vector. The solution is to divide
by a scalar (Lorentz invariant) quantity, i.e., to replace ∆t by the proper
time interval ∆τ = ∆t/γu . Thus we define the 4-velocity as
⇒ d d
U= (ct,~r ) = γu (ct,~r ) = γu (c, ~u)
dτ dt
p. 4 Eamples of physical 4-vectors

You can show (going through a cumbersome calculation) that this object
trasforms under Lorentz transformations as a 4-vector should. We also
immediately see that the norm is γu 2c2 − γu 2v2 = c2 , which we know to
be Lorentz invariant. Thus we have successfully constructed a 4-vector.
Definition: 4-velocity

If an object has 3-velocity ~u = (u x , uy , uz ) and speed u = |~u|, then its

4-velocity is


γu (c, ~u) = (γu c, γu~u) = γu c, γu u x , γu uy , γu uz

• The 4-velocity of any particle with nonzero mass is time-like, i.e., has
positive norm in the (+, −, −, −) metric. Exercise: Show!

• For a photon (massless particle), the speed is u = c so that γu is infinite.

The definition of 4-velocity unfortunately doesn’t make much sense in
this case.
Why does the definition of the 4-velocity not work for a photon? The
definition relied on the notion of proper time: the time measured in a
frame relative to which the object is at rest. But proper time is not de-
fined for a photon, because there is no frame relative to which a photon
is at rest. So, no surprise that the definition only applies to massive
objects.

• Using a capital U for 4-velocities seems to be common notation.

Our definition of the 4-vector could be written as


U µ = γu (c, ~u) = γu c, γu u x , γu uy , γu uz

with the understanding that U µ on the left represents the whole 4-vector
and not one component of the 4-velocity. We will be gradually shift-
ing toward this type of notation and stop using the non-standard thick-
vector notation.
Some texts also use U or U ~ to represent the 4-vector. I would advise
against this. No need to encourage confusion between 4-vectors and
3-vectors!

• For a particle/object at rest, the 4-velocity is simple: (c, 0, 0, 0). Exercise: Show!
In other words, the 4-velocity of any massive object in the frame moving
with the object is (c, 0, 0, 0)

• The fact that a 4-velocity is a 4-vector (transforms according to Lorentz

transformations) can be used to derive the equations of relativistic 3-
vector addition. Consider a particle having 4-velocity U = γu (c, u x , uy , uz )
p. 5 Eamples of physical 4-vectors

relative to Σ and U 0 = γu0 (c, u0x , u0y , u0z ) relative to Σ0 . If Σ and Σ0 are re-
lated to the standard boost, then
 00     0
U γv −γv (v/c) 0 0 U
U 01  −γv (v/c) 1
γ v 0 0  U 
 
  = 
U 02   0 0 1 0  U 2 
U 03 0 0 0 1 U3

These equations (and their inverted form) can then be used to express
~ 0 , and vice versa.
the components of ~u in terms of the components of u

Exercise! This is hopefully irresistible.

• Adding two 4-velocities does not physically mean much, in fact, the
result might not even be a valid 4-velocity.

Exercise: Take two relatively simple 4-velocities and add them.

(E.g., the 4-velocities of a particle at rest and a particle moving
along the x direction.) Find out if the resulting 4-vector has the
form of a 4-vector.

1.4 4-momentum
We have already encountered this 4-vector:
Definition: 4-momentum
An object with energy E and 3-momentum ~p = ( p x , py , pz ) has 4-
momentum ( E/c, ~p).

• For a massive particle of mass m moving with 3-velocity ~u, we know

that E = γu mc2 and ~p = γu m~u. The 4-momentum is

Pµ = (γu mc, γu m~u) = γu m(c, ~u) = γu m(c, u x , uy , uz ).

We notice that the right side is m times the 4-velocity! Thus

(
⇒ ⇒ for objects with
Pµ = mU µ or P = mU
nonzero mass.

This is similar to the familiar 3-vector relation ~p = m~v, which is very

pleasant. Of course, you cannot in general explect 4-vectors to obey the
same relations among themselves as the corresponding 3-vectors do.
p. 6 Eamples of physical 4-vectors

• We have previously seen that the combination (γu mc, γu m~u) transforms
like a 4-vector under Lorentz transformations. In fact, this is how we
motivated the notion of 4-vectors.

Exercise: Actually, we worked out the transformation properties

only for the case where the object moves in the same direction as
the boost. (~u was in the common x, x 0 direction, for our standard
boost.) Can you work out the transformations when the object
3-velocity is in an arbitrary direction?

• For a massless particle, i.e., a photon, we know how to express the mo-
mentum and energy in terms of its frequency f or its wavelength λ:
hc hf h
E = hf = and ~p = n̂ = n̂
λ c λ
where n̂ is a unit vector in the direction of propagation of the photon.
Hence the 4-momentum is
   
µ hf hf h h
P = , n̂ = , n̂ .
c c λ λ
For example, if the photon were traveling in the z direction, its 4-momentum
would be (h/λ, 0, 0, h/λ).
• The 4-momentum of a photon also transforms under LT’s as a 4-vector
should, although we did not show this explicitly. To show this, we need
to use our knowledge of the relativistic Doppler shift.

Exercise: Consider inertial frame Σ̃ related to Σ by a standard

boost. A photon moving in the common x, x 0 direction has fre-
quency f˜ as measured from Σ̃. Find the frequency measured from
Σ. Hence show that the 4-momentum of the photon, as measured
from Σ̃ and Σ, are related by the standard Lorentz transformation
for this boost.

• The expression Pµ = mass×U µ does not make sense for photons. This
is because 4-velocity is not defined for photons, and the mass is zero.
• For a massive particle, the norm of the 4-momentum is Pµ Pµ = m2 c2
(if you didn’t previously: Show! ). This is Lorentz invariant, of course.
Also, since this is positive, the 4-momentum of particles with nonzero
mass is time-like.
For a mass-less particle, the norm is Pµ Pµ = 0, also an invariant. The
4-momentum in this case is null or light-like.
p. 7 Eamples of physical 4-vectors

1.5 4-acceleration
We constructed the 4-velocity from the 4-displacement by differentiating
with respect to the proper time. Similarly, the 4-acceleration is obtained by
differentiating the 4-velocity with respect to the proper time:

dU µ
Definition: 4-acceleration Aµ =
dτ

• If the three-velocity of the particle/object is ~u, then

d d
Aµ = (γu c, γu~u) = γu (γu c, γu~u) = γu (γ˙u c, γ˙u~u + γu~a).
dτ dt
Here the dot represents a derivative with respect to the coordinate time
t (and not with respect to the proper time τ). Also, ~a = ddt~u is the 3-
acceleration.
• Early in the semester, we derived the expressions
1 3 1 3
γ˙u = γ u u
(~ ·~
a ) = γ uu̇
c2 c2 u
for the time derivative of the Lorentz factor. Here, u = |~u| is the speed
and u̇ is the time derivative of the speed, which is not the acceleration.
So the 4-acceleration can be written as
 
1 4 1 4 2
µ
A = γ ~u ·~a, 2 γu (~u ·~a) ~u + γu~a
c u c
or  
µ 1 4 1 4 2
A = γ uu̇, 2 γu uu̇~u + γu~a .
c u c
Admittedly, neither of these expressions are very pretty, and not worth
memorizing.
• Consider the frame of the object, i.e., the inertial frame in which the
object is (at this instant) at rest. This frame has to be re-defined at every
instant — if we kept a frame attached to an accelerating objject, that
would not be an intertial frame.
In its instantaneous frame, the object has zero velocity and zero speed.
The 4-acceleration becomes Aµ = (0,~a0 ) ( Show! ), where ~a0 is the ac-
celeration measured in the rest frame of the object, i.e., the acceleration
experienced by the object itself. This is called the proper acceleration.
• Like other 4-vectors, the norm Aµ Aµ is an invariant. There is a sim-
ple expression only in the frame where the object is (instantaneously)
at rest: Aµ Aµ = −~a0 · ~a0 . The norm of the 4-acceleration is thus the
negative square of the proper acceleration.
p. 8 Eamples of physical 4-vectors

• The 4-acceleration of a particle happens to be always ‘orthogonal’ to its

4-velocity.

⇒ ⇒
Exercise: Show that Uµ Aµ = 0 , or U ? A = 0 , using the ex-
pressions derived for them, for a particle having 3-velocity ~v and
3-acceleration ~a.

Since inner products are Lorentz invariant, if Uµ Aµ is zero in one inertial

frame, then it vanishes in any inertial frame.

Exercise: Write down U µ and Aµ in the rest frame of the particle.

Use these to show that Uµ Aµ = 0.

1.6 4-force
˙ in the by-now familiar
We generalize the defintion of the 3-force, ~F = ~p,
way. The 4-force on a particle/object is
dPµ
Definition: 4-force Kµ =
dτ
where Pµ of is the 4-momentum of the particle/object.
⇒
• Notation: Why K µ instead of, say, F µ or F ?
Because we will encounter an object called F µν (with two indices) in
electromagnetism, and it would be confusing to have two quantities
called F. So the notation K µ is common. It probably stems from the
German word for force, Kraft.
If you are not dealing with electromagnetism, using F µ is fine of course.
d
• Since Pµ = mU µ and Aµ = µ
dτ U , we get for objects which are not
losing/gaining mass
K µ = mAµ
The 4-force is mass times the 4-acceleration. Nice! Even though the rel-
ativistic 3-force and 3-acceleration refuse to obey the familiar equation,
the 4-force and 4-acceleration do obey a relativistic version.
d µ
• Noting dτ P = γu dtd ( E/c, ~p), we get
 
µ 1 dE ~
K = γu ,F
c dt
The 3-vector part of the 4-force is γu times the 3-force.
p. 9 Eamples of physical 4-vectors

• Using similar calculations as previously done for the 4-acceleration (or

just using K µ = mAµ ), one obtains the not-pretty expressions
m m 
Kµ = γu4 ~u ·~a, 2 γu4 (~u ·~a) ~u + mγu2~a
c c
and m m 4 
µ
K = γu4 uu̇, ~ 2
~
γ uu̇u + mγu a .
c c2 u
• If the velocity at some instant is in the x direction (v = v x and vy = vz =
0), then m 
Kµ = γu4 u x a x , γu4 ma x , γu2 may , γu2 maz .
c
Exercise: Show. This corresponds to the 3-force being
 
~F = γu3 ma x , γu may , γu maz .
p. 10 Eamples of physical 4-vectors

1.7 4-potential in electromagnetism

If you have taken a semester on electromagnetism or electrodynamics, you
might know that the electric field ~E and the magnetic field ~B can be ex-
pressed in terms of two potentials: the scalar potential φ and the vector
potential A:~
~
~E = −∇ ~ φ − ∂ A , ~B = ∇~ ×A
~
∂t
(In an introductory electromagnetism module, it might be more common
to use V (~r, t) instead of φ(~r, t). We will use φ.)
The scalar and vector potential combine together to form a 4-vector, which
we could call the 4-potential:
 
µ φ ~
A = ,A .
c

You might also remember (or learn soon) that the physical fields ~E and ~B
remain unchanged under gauge transformations:

∂f ~ →A ~ f,
~ −∇
φ → φ+ , A
∂t
where f (~r, t) is a scalar function. In 4-vector language, this has the concise
form
Aµ → Aµ + ∂µ f .
It’s almost special relativity was custom-built for electromagnetism! (Re-
member: it was indeed.) You might have guessed that ∂µ , the 4-gradient,
is the 4-vector operator
 
1∂ ∂ ∂ ∂
µ
∂ = ,− ,− ,− .
c ∂t ∂x ∂y ∂z

Notice the minus sign on the spatial derivatives. Where did those come
from? This will hopefully become clearer when we introduce index nota-
tion.

What about the electric and magnetic fields themselves? It seems likely that ~E and
~B should also be bundled into objects that transform nicely under Lorentz
transformations? They are, but these objects are not 4-vectors, rather, they are
4-tensors which are more complicated. We will postpone discussion of the
electromagnetic field tensor (and their connection to the 4-potential) until after
we’ve mastered index notation. At that point we will formulate all the basic laws of
electrodynamics in terms of 4-vectors and 4-tensors.
p. 11 Eamples of physical 4-vectors

The invariant associated with the 4-potential is the quantity

φ2 ~ |2
− |A
c2
I have no idea what this means or why this particular quantity is invariant
under Lorentz transformations.
~ |2 is Lorentz invariant. Is it also
Exercise: The quantity φ2 /c2 − | A
gauge-invariant?

1.8 Density-current, or 4-current-density

In electromagnetism, you would have met the charge density ρ (the charge
per unit volume) and the current density ~J, a 3-vector with direction point-
ing towards the flow of charge and magnitude giving the current through
a unit perpendicular area. These two quantities together form a 4-vector

Definition: 4-current J µ = (cρ, ~J ) or J µ = µ0 (cρ, ~J )

The first definition is more common. The second definition (including the
magnetic permeability of SI units) arguably makes future formulae easier
if we are using SI units.
As you know, electromagnetism has the annoyance of “choice of units.”
In special relativity, choices of metric and other conventions are a plague.
Taken together, when one studies electromagnetism in relativistic nota-
tion, a nightmare of differing notation can be expected: comparing differ-
ent texts can be very difficult. Using c = 1 and another set of electromag-
netism units (the Heaviside-Lorentz units instead of SI units) might make
things easier. We will get through unit issues somehow.

• Transformation under standard boost:

We will now show that the combination (cρ, ~J ) indeed transforms like
a Lorentz 4-vector. We focus on the case of the standard boost with the
charge having velocity component only in the common x, x 0 direction.
Generalization to arbitrary directions of ~J and arbitrary Lorentz trans-
formations is of course possible, but cumbersome.
Consider a blob of charge moving with velocity ~u, u ~ 0 relative to Σ, Σ0
0
frames. Mercifully, these velocities only have x, x components. In the
rest frame of the charge (call it Σ0 ), the charge density is ρ0 . Observed
from Σ or Σ0 , the charge density will be different (ρ or ρ0 ). Then the
p. 12 Eamples of physical 4-vectors

current densities observed from the Σ and Σ0 are respectively ~J = ρ~u

and ~J 0 = ρ0 u
~ 0 . We need only consider the x components:

Jx = ρu = ρu x , Jx0 = ρ0 u0 = ρ0 u0x .

As observed from the Σ frame, the size (length) of the charge-carrying

region in the x direction is contracted with respect to the rest frame by
a factor γu . The contraction factor is γu because u is the relative speed
between Σ and the rest frame Σ0 . Since the volume of the charged region
charge
is observed from Σ to be smaller, the density = is observed to
volume
be larger: ρ = ρ0 γu . Similarly, the charge density is observed from Σ0 to
be ρ0 = ρo γu0 .
We can now work out how (ρ, Jx ) are related to (ρ0 , Jx0 ), by using the
velocity transformation equation:

u0 + v  
u= , γu = γu0 γv 1 + u0 v/c2 . (1)
1 + u0 v/c2
We obtain
 
ρ = γu ρ0 = γu0 γv 1 + u0 v/c2 ρ0
0 v v 0
  
0
= γv ρ 0 γu 0 + ρ 0 γu 0 u 2 = γv ρ + 2 J x
c c

Similarly, you can show that Jx = γv ( Jx0 + vρ0 ). Exercise: Show.

Thus, we have shown that (cρ, Jx ) transforms like (ct, x ) for a standard
Lorentz boost.
This result can be extended, of course, to show that (cρ, ~J ) transforms
like (ct,~r ) for an arbitrary Lorentz transformation.

y y'

v
x x'

Figure 1: A charge distribution (cloud of charge) moves in the common

x, x 0 direction. It has speed u0 relative to Σ0 and speed u relative to Σ. Its
charge density is ρo in its rest frame, ρ when measured from the Σ frame,
and ρ0 when measured from Σ0 .
p. 13 Eamples of physical 4-vectors

• In electrodynamics, the conservation of charge is expressed through the

∂ρ ~ ~
beautiful continuity equation: + ∇ · J = 0. In index notation, this is
∂t
expressed in the (even more beautiful) form:

continuity equation ∂µ J µ = 0 or ∂µ Jµ = 0.

We will revisit this after we’ve learned index notation.

• If we consider mass density (also just known as density) and mass cur-
rent density, than exactly the same derivation holds. Mass density and
mass current density also form a 4-vector.

• At the end of an electrodynamics class, you study the conservation of

electromagnetic energy: the energy density and energy current den-
sity (Poynting vector) together obey a continuity equation. One won-
ders: Should the energy density and energy current density also form a
Minkowski 4-vector?
The answer is unfortunately no. This is because the energy itself is not
invariant under Lorentz transformations, in contrast to mass or charge.
So, the derivation above breaks down. If we try a calculation analogous
the one above, the energy density does not get just a factor of γu (unlike
charge density or mass density) when transforming from Σ0 to Σ; its
transformation is more complicated. As a result we cannot construct a
valid 4-vector out of energy-density and energy-current.
p. 14 Eamples of physical 4-vectors

1.9 Frequency-wavevector, or 4-wavevector

1.10 4-gradient