Galilean boost symmetry1

D. E. Soper2
University of Oregon
4 April 2011

I offer here some background for Chapter 4 of J. J. Sakurai, Modern Quan-

tum Mechanics. This chapter discusses symmetries other than translations
and rotations, including parity, translations in a lattice, and time reversal.
It omits boost invariance, so I discuss boost invariance here.

1 Introduction
Non-relativistic classical mechanics is invariant under a change of reference
frame to one that is moving with a fixed velocity with respect to the reference
frame originally considered. To be more precise, suppose that we describe a
system in the lab frame using coordinates (t, x, y, z). Then we switch to a
moving frame using coordinates (t0 , x0 , y 0 , z 0 ). The transformation is t0 = t,
~x 0 = ~x + ~v t. Note that if we were doing relativistic physics, we would have
t0 6= t, but life is simpler with non-relativistic physics. For understanding
quantum mechanics, it is simpler to think of leaving the coordinates alone
but giving the system a boost by velocity ~v : we simply add ~v to the velocity
of every particle.

2 Boost operator
We are thus led to consider the unitary operator U (~v ) that boosts the sys-
tem. The operator U (~v ) does nothing to spins. For this reason, we consider
particles without spins. We consider a system made of particles 1 through N
with masses m1 through mN . We  begin with a momentum space description,
in which the basis vectors are p~1 , p~2 , . . . , p~N .
The operator U (~v ) adds ~v to the velocity of every particle. Thus for a
particle i with mass mi , U (~v ) adds mi~v to the momentum of the particle:
   
U (~v )p~1 , p~2 , . . . , p~N = p~1 + m1~v , p~2 + m2~v , . . . , p~N + mN ~v . (1)
1
Copyright, 2011, D. E. Soper
2
soper@uoregon.edu

1
 Now lets see what this does in a position space description. We have
Z !
X
U (~v )~x1 , ~x2 , . . . , ~xN = (2)3N/2 d~p1 d~pN exp i
 
p~i ~xi
i
 
U (~v )p~1 , p~2 , . . . , p~N
Z !
X
= (2)3N/2 d~p1 d~pN exp i p~i ~xi
i
 
p~1 + m1~v , p~2 + m2~v , . . . , p~N + mN ~v

Z
= (2) 3N/2
d~k1 d~kN

 (2)
!
X
(~ki mi~v ) ~xi ~k1 , ~k2 , . . . , ~kN

exp i
i
! Z
X
= exp i mi~v ~xi (2) 3N/2
d~k1 d~kN
i
!
X
~ki ~xi ~k1 , ~k2 , . . . , ~kN
 
exp i
i
!
X  
= exp i mi~v ~xi ~x1 , ~x2 , . . . , ~xN .
i

That is
 
~ ~x1 , ~x2 , . . . , ~xN
  
U (~v )~x1 , ~x2 , . . . , ~xN = exp iM~v R (3)

where M is the total mass of the system,

X
M= mi (4)
i

and R~ is the operator that gives the position of the center of mass of the
system
~ = 1
X
R mi~xi . (5)
M i
We learn two things from this. First, the boost operator is very simple in
the position representation. Second, the generator of boosts is M R. ~ That
is rather similar to the generator of space translations being P~ .

2
3 Transformation of operators
These results tell us how the operators ~xi and p~i transform under boosts.
(We could call these ~xi,op and p~i,op , but we dont do that here because we are
not in danger of confusing the operators with their eigenvalues.) We have
U (~v )1~xi U (~v ) = ~xi ,
(6)
U (~v )1 p~i U (~v ) = p~i + mi~v .
One operator of some importance in quantum mechanics is the total mo-
mentum, X
P~ = pi . (7)
i
It transforms according to
X
U (~v )1 P~ U (~v ) = U (~v )1 p~i U (~v )
i
X (8)
= (~pi + mi~v ) .
i

Thus
U (~v )1 P~ U (~v ) = P~ + M~v . (9)
Let us take the hamiltonian to be
X p~ 2
i
H= + V (~x1 , . . . , ~xN ) . (10)
i
2m i

With this form, we can see what the boosted hamiltonian is

X (~pi + mi~v ) 2
U (~v )1 HU (~v ) = + V (~x1 , . . . , ~xN )
i
2m i
X p~ 2
i
= + V (~x1 , . . . , ~xN ) (11)
i
2mi
X X1
+ p~i ~v + mi~v 2 .
i i
2
That is
1
U (~v )1 HU (~v ) = H + P~ ~v + M~v 2 . (12)
2
Thus the hamiltonian is not boost invariant. Rather, the hamiltonian and
the total momentum operator transform in a simple way that is independent
of the details of the potential.

3
4 The internal hamiltonian
Define
1 ~2
Hint. = H P . (13)
2M
Here int. denotes internal, M is the total mass of the system and P~ is
the total momentum of the system. The internal hamiltonian represents
the energy in a reference frame in which the total momentum of the system
vanishes. That is, if the system is in an eigenstate of P~ , then we can boost
it with velocity ~v = P~ /M and thus make the eigenvalue of P~ vanish. Then
Hint. is the hamiltonian for the suitably boosted system. Of course, the
system in general will not be in an eigenstate of P~ and ~v is a numerical
vector while P~ is an operator, so in general we cannot boost to make P~ = 0.
However, Eq. (13) is a relation among operators that can define what we
mean by Hint. .
How does Hint. transform under boosts? We note that
1 ~2 1 ~2 ~ 1
U (~v )1 P U (~v ) = P + P ~v + M~v 2 . (14)
2M 2M 2
Thus Hint. is boost invariant

U (~v )1 Hint. U (~v ) = Hint. . (15)

5 Translation invariant potential

For the remainder of these notes, we will take V will be a function only of the
absolute values of differences of the coordinates, |~xi ~xj |. Thus it is invari-
ant under space-translations and rotations. However, this invariance is not
indicated explicitly in the notation. With a translation invariant potential,
we have
[P~ , V ] = 0 . (16)
Then
[P~ , H] = 0 . (17)
That means that we can choose P~ and H to both be diagonal.
From the definition (13), we have
1 ~2
H = Hint. + P . (18)
2M
4
Choosing P~ to be diagonal makes the second term just the kinetic energy
for a particle with momentum P~ and mass M . We can say that Hint. gives
the internal energy of the system and then the full energy equals the internal
energy plus the kinetic energy from the motion of the center of mass. If
we want the energy levels of a system, it is really Hint. in which we are
interested. The kinetic energy from the center of mass motion is trivial and
uninteresting.
To construct Hint. , we simply need the momenta

~ki = p~i mi P~ . (19)

M

We note that the operators ~ki are boost invariant,

mi ~ mi
U (~v )1~ki U (~v ) = p~i + mi ~v P M~v = ~ki . (20)
M M

Furthermore, ~ki = p~i when P~ = 0. Thus if the states happen to be eigenstates

of P~ , then ~ki = p~iP
if we boost thePsystem to make the eigenvalue of P~ vanish.
Note that, using p~i = P~ and mi = M , we have
X
~ki = 0 . (21)
i

A simple calculation gives

X ~k 2 X p~ 2 P~ 2
i i
= . (22)
i
2m i i
2m i 2M

With this, we have

X ~k 2
i
Hint. = + V (~x1 , . . . , ~xN ) . (23)
i
2m i

6 Internal coordinates and momenta

We have already defined internal momenta ~ki ,

~ki = p~i mi P~ . (24)

M

5
with X
~ki = 0 . (25)
i

Then the complete system can be described using the ~ki and the total mo-
mentum P~ .
Similarly, we can define internal coordinates
~ .
~ri = ~xi R (26)

with X mi
~ri = 0 . (27)
i
M
Then the complete system can be described using the ~ri and the position of
~
the center of mass, R.
A simple calculation shows that R~ and P~ are conjugate operators in the
sense that their commutation relation is

[Rk , P l ] = ikl . (28)

A simple calculation also shows that ~ki commutes with R,

~

[Rk , kil ] = 0 . (29)

(Indeed, M R~ is the generator of boosts, so this says that ~ki is boost invari-
ant.) A simple calculation also shows that ~ri commutes with P~ ,

[rik , P l ] = 0 . (30)

(Indeed, P~ is the generator of translations, so this says that ~ri is translation

invariant, which is evident because it is the difference of two positions.) Thus
we have one set of operators, R~ and P~ , that tell about the overall motion of
the system and another set of operators, the ~ki and ~ri , that tell about the
internal motion. The internal motion operators commute with R ~ and P~ , so
we can separate the dynamics of the quantum system into two independent
parts, described by commuting sets of operators. Since we consider that the
potential depends only on differences of the positions ~xi , it can be written as
a function of the ~ri .

V (~x1 , . . . , ~xN ) = V (~r1 , . . . , ~rN ) . (31)

6
Then the hamiltonian Hint. depends only on the internal operators.
Unfortunately, the commutation relations between the ~ri and the ~ki are
a little complicated since the operators ~ki are not independent because of
Eq. (25) and neither are the operators ~ri , because of Eq. (27). We have
mj
[rik , kjl ] = ikl (ij ) . (32)
M
Suppose that you want to solve for the internal motion of three particles.
Then you need to reduce the three position vectors to just two, with two
conjugate momentum vectors. There are more than one ways to do this,
but we will not pursue them here. What we want to notice is that for any
number of particles the internal motion factors from motion of the center of
mass. Taking P~ to be diagonal, we can use a wave function
~ 0 R
~
e iP (~r1 , , ~rN ) , (33)

where here P~ 0 is the eigenvalue of the operator P~ . To find (~r1 , , ~rN ), we

use Hint. .

7 The internal hamiltonian for two particles

With just two particles, this construction is pretty simple. With our assump-
tion about V , the potential is a function of just |~x1 ~x2 |, so that
p~12 p~ 2
H= + 2 + V (|~x1 ~x2 |) . (34)
2m1 2m2
Then
~k 2 ~k 2
1
Hint. = + 2 + V (|~x1 ~x2 |) . (35)
2m1 2m2
Since ~k1 + ~k2 = 0, we can use just one momentum operator ~k with ~k1 = ~k
and ~k2 = ~k. Then
 
1 1 1~ 2
Hint. = + k + V (|~x1 ~x2 |) . (36)
m1 m2 2
Define the reduced mass by
1 1 1
+ = . (37)
m1 m2

7
Then
~k 2
Hint. = + V (|~x1 ~x2 |) . (38)
2
~ and ~r = ~r1 ~r2 :
For space coordinates, we can use R

~ = 1 (m1~x1 + m2~x2 ) .
R
M (39)
~r = ~x1 ~x2 .

Then
~k 2
Hint. = + V (|~r|) . (40)
2
We know already that

[Rk , P l ] = ikl ,
[Rk , k l ] = 0 , (41)
[rk , P l ] = 0 .

A simple computation gives

[rk , k l ] = ikl . (42)

Thus in a coordinate representation,

k l = i . (43)
rl
In a coordinate representation, we look for a wave function
~ 0 R
~
e iP (~r) , (44)

where here P~ 0 is the eigenvalue of the operator P~ . We can diagonalize the

operator P~ because it commutes with Hint. and with P~ 2 /(2M ). Then (~r) is
the internal wave function. The Schrodinger equation for (~r) is the equation
that we are used to, except that the reduced mass appears in the kinetic
energy.

8
8 Angular momentum
The angular momentum operator for our particles is
X
J~ = ~xi p~i . (45)
i

This transforms under translations and boosts according to

~ ~ iP ~ ~a
X
eiP ~a Je = (~xi + ~a) p~i ,
i
X (46)
1 ~ (~v ) =
U (~v ) JU ~xi (~pi + mi~v ) .
i

That is,
~ ~ iP ~ ~a
eiP ~a Je = J~ + ~a P~ ,
(47)
~ (~v ) = J~ + R
U (~v )1 JU ~ M~v .

We can make up an internal angular momentum that is invariant under

translations and boosts as
X
~=
S ~ri ~ki . (48)
i

Since both ~ri and ~ki are invariant under translations and boosts, so is S.
~
Let us see how S ~ is related to J.
~ We have

~ p~i mi P~
X  
S~= (~xi R)
i
M
X X mi X X mi
= ~xi p~i ~xi P~ ~ p~i +
R ~ P~
R
M M (49)
i i i i

= J~ R
~ P~ R
~ P~ + R
~ P~
= J~ R
~ P~ .

That is
J~ = L
~ +S
~ , (50)
where
~ =R
L ~ P~ . (51)

9
Thus the total angular momentum J~ is composed of the orbital angular
momentum L ~ associated with the motion of the center of mass plus the
internal angular momentum S.~ The orbital angular momentum transforms
under translations and boosts as an ordinary angular momentum,
~ ~ iP ~ ~a ~ + ~a P~ ,
eiP ~a Le =L
(52)
~ (~v ) = L
U (~v )1 LU ~ +R ~ M~v .

The internal angular momentum is invariant under translations and boosts,

~ ~
~ iP ~a = 0 ,
eiP ~a Se
(53)
~ (~v ) = 0 .
U (~v )1 SU

~ has the usual angular momentum commutation

It is easy to check that L
relations with itself,
[Lk , Ll ] = ikln Ln . (54)
~ commutes with S,
Also, L ~
[Lk , S l ] = 0 , (55)
~ and P~ commute with S.
since both R ~ Then

[J k , J l ] = [Lk + S k , Ll + S l ] = ikln Ln + [S k , S l ] (56)

but
[J k , J l ] = ikln J n , (57)
so
[S k , S l ] = ikln (J n Ln ) . (58)
That is, the internal angular momentum has the usual angular momentum
commutation relations with itself,

[S k , S l ] = ikln S n . (59)

To summarize, we have J~ = L~ +S~ where L

~ and S~ have the usual angular
momentum commutation relations with themselves and L ~ commutes with S.
~
These are the same relations as we had for the orbital angular momentum
and the spin angular momentum of a single particle.
We simplified our calculations by considering a bound system made of
many spin 0 particles. We could have used spin 1/2 particles. Then the

10
 ~ here would be made of the internal orbital angular
operator we are calling S
ri ~ki , plus the sum of their spin
P
momenta of the constituent
P particles, i~
angular momenta, i ~si . For example, a simple model for a proton is that
it is made of three spin 1/2 quarks combined, with some orbital angular
momentum, to make a system with total angular momentum 1/2.

9 Motion of the system as a whole

We have considered a system of particles that interact with each other via
a potential that depends only on absolute values of the separations between
pairs of particles. The most important case is that there is a contribution
ei ej
|~xi ~xj |

for each pair of particles. The analysis that we have applied works for any
state, but it is most useful for bound states. Thus, lets consider a bound
state here.
We can make the system more complicated by supposing that there is an
additional force that acts on our system. The simplest case is that we have a
bound state, an atom, say, and we place it in an external electric field. Then
we have an additional potential
X
V = ei (~xi ) . (60)
i

Assuming that our potential is slowly varying, we can approximate this by

~ ,
V Q (R) (61)

ei is the total charge of the atom. Then we have

P
where Q =
1 ~2 ~ .
H = Hint. + P + Q (R) (62)
2M
Here Hint. is the same hamiltonian for the internal motion that we had before.
We can imagine solving for the internal motion in the form
   
Hint.  = Eint.  . (63)

11
Then the full (time independent) Schrodinger equation is
 
1 ~2 ~
   
P + Q (R)  = (E Eint. ) . (64)
2M
~
Our atom moves in the potential (R).
A more exact approximation to the interaction of the atom with the
external field is X
V Q (R) ~ ~ R)
ei~ri E( ~ , (65)
i

where
~ =
E ~ . (66)
This next approximation is especially important if the lowest order approxi-
mation is zero because the atom is electrically neutral, Q = 0. Now we have
an interaction between the external field and the electric dipole moment of
the atom. This couples the internal and external motions. Perhaps we can
come back to this when we deal with approximation methods in the next
quarter.

10 Group multiplication law

There is a peculiarity with symmetry operations involving the boost operator.
Suppose that we multiply together
~ ~
U = e+iP ~a U (~v )1 eiP ~a U (~v ) . (67)

This corresponds to the following sequence of coordinate transformations:

~x ~x + ~v t a boost
~x + ~v t + ~a a translation
(68)
~x + ~a an inverse boost
~x an inverse translation .

That is, the product of these transformations leaves the coordinates alone. If
our quantum operators obeyed the group multiplication laws, we would have
U = 1. However, we know how P~ transforms under boosts, so
~ ~
U (~v )1 eiP ~a U (~v ) = ei(P +M~v)~a . (69)

12
Thus
U = eiM~v~a . (70)
We do not get the unit operator. But what we get is a number with absolute
value 1, a phase factor. This phase factor is applied to every vector in the
space of quantum

 mechanical
2 states. It then cancels whenever we calculate
a probability | | .

We see that the quantum theory provides a representation of the Galilei
group up to a phase. This phase doesnt matter. When we dealt with rota-
tions and translations by themselves, we could have had something similar,
but there we got a true representation with no phase. At least thats what
we got if you didnt have spin 1/2 particles. With a spin 1/2 particle, a
rotation through 2, which is the same as not rotating at all, gives a phase
factor 1.

13