1 Lorentz Invariance and QM

One of the major misconceptions in the discussion of QFT, is that it is necessary extension of QM to be
compatible with Lorentz invariance. This is not entirely correct, as QM is inherently Lorentz invariant.
Take for example the path integral formulation of QM, where the transition amplitude is given by
Z Z tf  2
i m d~x
h~xi , ti |~xf , tf i = d~x(t) e h S , S = dt (1)
ti 2 dt
where ~x is the three component spatial vector, and we integrate over all path that begins at (~xi , ti ) and
ends at (~xf , tf ). Lets set m = 1. The usual statement is that time is single out and treated in a fashion
different than space. This is not entirely true as we will see.
Lets now consider another action
1 1 dX dX
Z  
S = d (2)
v( ) 2 d d
where X is now a four vector with:
X 0 = t, (X 1 , X 2 , X 3 ) = ~x, (3)
You will notice that now this new action is Lorentz invariant since it is constructed from the Lorentz
invariant inner product X X = (X 0 )2 (X 1 )2 (X 2 )2 (X 3 )2 . Now this new action has a remarkable
symmetry, it is unchanged under a change of coordinates!. To see this first note that the new function
v( ) is defined such that under arbitrary change of coordinates, we have
d 0 v 0 ( 0 ) = d v( ) (4)
where 0 is a new coordinate and can be an arbitrary new function of the old , i.e. 0 ( ). Note that the
above equation then tells us that
d
v 0 ( 0 ) = 0 v( ) (5)
d
Finally, the function X( ) is a scalar, which means that under a change of coordinates
X 0 ( 0 ) = X( ) . (6)
Now you should be able to convince yourself that the action is indeed invariant under coordinate change,
i.e.
1 1 dX 0 dX0 1 1 dX dX
Z   Z  
d 0 0 0 = d (7)
v ( ) 2 d 0 d 0 v( ) 2 d d
To let you have a better feeling, you can study the change of the various characters. For example consider
a coordinate change where
0 = ( ) (8)
where ( ) is an infinitesimal function. For the scalar we then have:1
dX( )
X 0 ( 0 ) = X( ) X 0 ( ) ( ) = X( ) (9)
d
1 You dX( ) dX 0 ( )
might wonder why in the second equality why it is ( ) d instead of ( ) d . The reason is that we
are considering tiny changes, and ( ) is already infinitesimal. So at this point it is sufficient to consider X( ) as an
approximation to X 0 ( ).

1
Thus under a change of coordinates, we find that the scalar transforms as:

dX( )
X( ) X 0 ( ) X( ) = ( ) . (10)
d
HW1 Use eq.(5), show that the infinitesimal change for the function v( ) is given by:

d( )v( )
v( ) v 0 ( ) v( ) = (11)
d
HW2 Even though it is redundant, show that the action is also invariant under an infinitesimal change
of coordinates.
Cool! with all this, we are ready to do some business. Note that since the action is invariant under a
change of coordinates, we can choose any coordinate system we want. Lets then choose

X 0 ( ) = . (12)

if you recall that X 0 is our original time coordinate, we are essentially choosing our coordinate to be
exactly of t. For those of you who are bait more well informed, this is called a gauge choice. Since
we are choosing a special coordinate to make X 0 special, lets look at its Equation of Motion. Using the
variation principle we find
L d L v 1
0
= 2 X 0 X 0 = 0 (13)
X d X 0 v v
where L is the Lagrangian. Now since weve already chosen the coordinate system where we have eq.(12),
one then has X 0 = 0. Thus for the equation of motion to be satisfied, we can conclude,

v = 0 v = constant . (14)

Putting everything together, we find that the action in eq.(2) indeed becomes eq.(1)! (Theres an extra
constant terms, but constants in the Lagrangian are irrelevant, at least for now.)
So now youve learned three lessons: 1. QM is in fact Lorentz invariant! 2. Often time in physics,
formalism is just formalism. Many important properties of the system may not be manifest. In the
present case, Lorentz invariance was not manifest, but its there. 3. Often times, manifest symmetry can
be achieved by introducing redundancy in the system. In this case the redundancy was the invariance
under change of coordinates. The last point will be a recurrent theme in the setting up the necessary
formalism for QFT beyond scalars.
Ok so if QM is Lorentz invariant, then whats wrong with it? Well, consider the description of a
charged particle coupling to an external electric or magnetic field:

E(x), B(x)

2
Now from your EM class you shouldve learned that to show Lorentz symmetry, we should expressed the
~ (more on this in the
electric and magnetic fields in terms for potentials A , where E A0 , B A
next class). Now we can use our Lorentz invariant action in eq.(2). The must natural extension would be
1 1 dX dX dX
Z  
S[A] = d +e A (X) (15)
v( ) 2 d d d
where e is the coupling charge.
HW3
Is S[A] still invariant under a change of coordinate? If so, how should A (X) transform ?
HW4
From classical electrodynamics, you know that any thing that is described by A must be invariant under
the change A = (X)X where (X) is some random whatever function. Show that S[A] is indeed
invariant.
Great now again plug in our special choice of coordinates in eq.(12), we find that the action becomes:
Z  
1 d~x d~x ~ t)
S[A] = dt + eA0 (x, t) ex A(x, (16)
2 dt dt
This is what Lorentz invariance tells us about how a charged particle can couple. If you recall that the
momenta of a system is given by L
~x
, in this case we find,
L d~x ~ t)
p~ = = eA(x, (17)

~x dt
This is precisely why in QM class, your professor tells you that in the presence of a electric magnetic
field, the canonical momentum is shifted as d~ x d~
x ~
dt dt eA(x, t).
Great, so now youve seen that the QM not only is Lorentz invariance compatible with what ever we
are doing, it actually tells us how interactions can be described. So what is wrong? Well, youll notice
that the field A(X) is just a classical field. Theres nothing quantum about it! In other words, we are
just describing a quantum particle traveling through a classical background. If you want to describe
interactions that would be described by the following diagram

you are out of luck, because we dont know how to describe a path integral, where the path itself (the
) actually branches out! To be more mathematically rigorous, the correct statement is that the above
diagram is not a true one-dimensional manifold. For those of you who have learned GR, youll notice that
the invariance of our action under change of coordinates is nothing by diffeomorphism invariance, and
what we were describing before is essentially one-dimensional gravity theory. So QM in its very essence
is simply a one-dimensional quantum gravity theory (sounds catchy doesnt it). Its inability to describe
interactions is due to the fact that one-dimensional manifolds are simple, you can only have a straight
line, or a circle.
Thus in summary, QM fails to describe systems where all degrees of freedom including the interaction
itself, are described by quantum particles. For such system we need something else, and that is the reason
you are in the QFT class today (I hope).

3
2 The physical states
Before moving on to working out the basics of QFT, it is instructive to stop and think about what types
of degrees of freedom should we expect from a world with Lorentz invariance such as ours. After all,
we should be familiar with what we want to calculate or describe first, before studying a 700 page book
learning about some tools to do the task.
Lorentz invariance means that there are only two meaning full distinction of particles, those which are
massless, p2 = 0, and those that are massive p2 = m2 . Now we certainly see a lot more particle then just
two types, so there better be a more refined version of categorization then just the above. Lets begin
with massless particle first.
Imagine a massless particle traveling, and we some how tag along. We can then go to the particles
frame where the four-momenta takes the form

P = E(1, 0, 0, 1) (18)

Now let us try to think what are the different ways a particle might look as we maintain traveling
parallel to it. Since we must be parallel to the particle, the only way we can move around is to rotate in
the plain that is perpendicular to the particle, i.e. a rotation in the plane that has (0, 0) in eq.(18). This
is illustrated as

This is a two dimensional rotation, also known as SO(2). Since we can do this rotation, we can classify
the particle by how much it changes. It can not change at all, but if it does, we know that it must rotate
back to itself after rotating 2. This implies that the particle are allowed to change as:

|0 i = ei |i (19)

where = 0, 1, 2 . Actually, in reality it is sufficient for the particle to come back to itself up to
a sign (we will learn more about this soon), so half integers are also allowed. Thus in summary we find
that we there is a whole zoo of states that are classified by how they transform under this SO(2) little
group:
1 3
= 0, , 1, , 2, . (20)
2 2
The 0 is what we called a scalar state. The 1 is a spin-1 state, and note that it has two degrees of
freedom. Infact all massless states besides the scalar has two degree of freedom. This will be something
that has a very interesting explanation from QFT, but here we see that this was already fixed physically

4
a priori. For fundamental particles, nature has already given us states with 0, 12 , 1, 2 (states are
approximately massless at high enough energies). Anything beyond 2 will have inconsistent probablity
amplitudes which we will explain at some point. So that leaves us with 32 , is there something waiting
to be discovered?
We can now turn to massive particles, which things are more familiar. The momentum is now:

P = E(1, 0, 0, 0) (21)

where we are essentially in the particles rest frame. The little group, whose formal definition is simply
the symmetry group that leaves the momentum invariant, is SO(3), the three-dimensional rotation group,
and we for sure (I hope) know the classification of distinct states, these are the spherical harmonics, and
they have spin
` = 0, 1, 2, 3, (22)
for ` = 1, we have `z = 1, 0, 1 and similarly for other spins. If we consider the covering group SU(2),
then we also have spin-1/2 representations. All of this you should have learned in your QM class.
The important thing to notice here is that for massive particles, the spin-1 state has 3 degrees of
freedom, where as in massless case it has only 2. Since the degrees of freedom are different, this tells you
that a massless spin-1 state likes to stay massless. This is unlike spin-0 or spin- 21 , where it has the
same degrees of freedom either massless or massive, so it can be bi-curious. In this course you will learn
from a QFT standpoint why massless vectors stay massless. As for the bi-curious nature of scalars, this
is the source of the huge headache for us trying to understand the Higgs boson. In other words, youve
already seen a glimpse of the frontier of current research!