Physics 70007, Fall 2009

Solutions to Homework #3

October 30, 2009

1. (Sakurai 1.28)

(a) Let x and px be the coordinate and linear momentum in one dimension. Evaluate the classical
Poisson bracket
[x, F (px )]classical

The classical Poisson bracket is dened in Sakurai (1.6.48):

X  ∂A ∂B ∂A ∂B

[A (q, p) , B (q, p)]classical ≡ −
s
∂qs ∂ps ∂ps ∂qs

so
∂x ∂F (px ) ∂x ∂F (px )
[x, F (px )]classical = −
∂x ∂px ∂px ∂x
∂F
= 1· −0·0
∂px
∂F
=
∂px

(b) Let x and px be the corresponding quantum mechanical operators this time. Evaluate the com-
mutator   
ipx a
x, exp
~

First we expand the exponential in a power series, and use the fact that the commutator is linear
in its arguments:
   X ∞  n
ipx a 1 ia
x, exp = [x, pnx ]
~ n=0
n! ~

Now we show by induction that [x, pnx ] = i~ · npn−1

x for n ≥ 1:

• Firstly for n = 1, the statement becomes [x, px ] = i~, which we accept as true (this is one of
our basic axioms);

• we then assume that the statement holds for n and prove that it holds for n + 1:

x, pn+1 = xpn+1 − pn+1

 
x x x x
= xpnx px − pnx px x
= ([x, pnx ] + pnx x) px − pnx px x
= i~npn−1 + pnx x px − pnx px x


x
= i~npnx + pnx [x, px ]
= i~ (n + 1) pnx as desired.

1
Using this result, we have
   ∞  n
ipx a X 1 ia
x, exp = i~npn−1
x (the n=0 term gives zero)
~ n=1
n! ~
  ∞  n−1
ia X 1 ia
= (i~) pn−1
x
~ n=1
(n − 1)! ~
 
ipx a
= −a exp (shifting the n index)
~
From part (a), we nd that the corresponding classical Poisson bracket would be ∂p∂x exp ip~x a =


−a ipx a
. So we see that the correspondence rule between classical Poisson brackets and


i~ exp ~
quantum-mechanical commutators (Sakurai 1.6.47), wherein we take the classical result, promote
x and p to operators, and multiply by i~ to obtain the quantum commutator, is satised.

(c) Using the result obtained in (b), prove that

 
ipx a
exp |x0 i , (x |x0 i = x0 |x0 i)
~
is an eigenstate of the coordinate operator x. What is the corresponding eigenvalue?

Let |ψi ≡ exp ipx a

|x0 i; then


~
 
ipx a
x |ψi = x̂ exp |x0 i
~
     
ipx a ipx a
= exp x + x, exp |x0 i
~ ~
    
ipx a 0 ipx a
= exp x − a exp |x0 i
~ ~
= (x0 − a) |ψi
So |ψi is an eigenstate of x with eigenvalue x0 − a. Apparently, the exponential operator has
performed a spatial translation of the position eigenstate.

2. (Sakurai 1.29)

(a) On page 247, Gottfried (1966) states that

∂G ∂F
[xi , G (~
p)] = i~ , [pi , F (~x)] = −i~
∂pi ∂xi
can be easily derived from the fundamental commutation relations, for all functions F and G that
can be expressed as power series in their arguments. Verify this statement.

First we note that although G is a function of all the components of p~, the only component which
does not commute with xi is pi ; so for this computation, we can consider G to be a function of pi
only (the other components of p~ will act like constants as far as commutation with xi is concerned.)
We then expand G as a power series in pi , apply the linearity of the commutator, and use the result
we derived in part (b) of the previous problem:
∞
X
[xi , G (~
p)] = gn [xi , pni ]
n=0
X∞
= gn · i~npn−1
i (the n=0 term is zero)
n=0
∞
∂ X
= i~ gn pni
∂pi n=0
∂G
= i~
∂pi

2
The second relation can be derived in a very similar way (the needed lemma is [pi , xni ] = −i~nxn−1 i
whose proof is very similar to the one already performed.)
 2 2  2 2
(b) Evaluate x , p . Compare your result with the classical Poisson bracket x , p .
classical

Our strategy to evaluate x2 , p2 = xxpp − ppxx is to move the x operators in the rst term one
 

by one, through the p operators:

 2 2
x , p = xxpp − ppxx
= x p2 x + x, p2 − p2 x2
 

= xp2 x + 2i~xp − p2 x2 (from part a)

= p x + x, p x + 2i~xp − p2 x2
2 2
 

= p2 x2 + 2i~px + 2i~xp − p2 x2
= 2i~ (xp + px)
In contrast, the classical Poisson bracket is
 2 2 ∂x2 ∂p2 ∂x2 ∂p2
x , p classical = −
∂x ∂p ∂p ∂x
= (2x) (2p) − 0 · 0
= 4xp
At rst glance, it might seem as though the classical-quantum correspondence rule is failing us here.
In fact this is not the case  there is an ambiguity in the rule, due to the fact that the classical x
and p commute (as they are just ordinary c-numbers) while the corresponding quantum operators
do not. So when we are presented with a classical quantity like xp, we can quantize it as is, giving
us the operator x̂p̂, or we could commute the classical quantities rst to get px and then quantize,
giving us a dierent operator p̂x̂! Both choices appear equally valid, yet they cannot possibly both
be correct. The prescription that appears to work is the symmetric prescription, which respects
both choices equally by writing xp as 1/2 (xp + px) before quantizing; at any rate, it makes classical-
quantum correspondence work for this problem. The bottom line, however, is that it is often far
from clear how to quantize a classical system once we move beyond the simplest cases.

3. (Sakurai 1.30)

The translation operator for a nite (spatial) displacement is given by

!
p · ~l
−i~
T ~l = exp


~
where p~ is the momentum operator.
h
i
(a) Evaluate xi , T ~l .

First note that we can write the translation operator as

 
~

−i (px lx + py ly + pz lz )
T l = exp
~
     
−ipx lx −ipy ly −ipz lz
= exp exp exp
~ ~ ~
since the various components of momentum commute with each other. Then, since only pi out of
all the components of momentum does not commute with xi , we can treat the exponentials not
involving pi as constants and factor them out of the commutator:
 
 
h
i Y −ip l
j j 
xi , T ~l = xi , exp
j
~
 
    
Y −ip l
j j  −ipi li
=  exp xi , exp
~ ~
j6=i

3
The remaining commutator is the same one we evaluated in problem 1, part b:
 
    
h
i Y −ip l
j j  −ipi li
xi , T ~l =  exp li exp
~ ~
j6=i

= li T ~l .

(b) Using (a) (or otherwise), demonstrate how the expectation value h~xi changes under translation.

We consider some state |ψi versus its translated counterpart |ψ 0 i ≡ T ~l |ψi, and look at the

expectation value of some component of position:

†
hψ 0 | xi |ψ 0 i = hψ| T ~l xi T ~l |ψi



†
† h
i
= hψ| T ~l T ~l xi |ψi + hψ| T ~l xi , T ~l |ψi

†
†
= hψ| T ~l T ~l xi |ψi + li hψ| T ~l T ~l |ψi

= hψ| xi |ψi + li

where we have used the fact that T ~l is unitary, since the components of momentum are hermitian.

This result holds for each component i, so we can combine our results into the vector equation
0
h~xi = h~xi + ~l

which shows that the expectation value of position is translated, as we might have suspected.

4. (Sakurai 2.5)

Consider a particle in one dimension whose Hamiltonian is given by

p2
H= + V (x).
2m
By calculating [[H, x] , x], prove

X 2 ~2
|ha00 | x |a0 i| (Ea0 − Ea00 ) =
2m
a0

where |a0 i is a [normalized] energy eigenket with eigenvalue Ea0 [and the sum goes over the complete
set of energy eigenstates].

First we note that

∗
ha00 | x |a0 i = ha0 | x† |a00 i = ha0 | x |a00 i
since x is hermitian. Then calling the desired sum S , we have
2
X
S = |ha00 | x |a0 i| (Ea0 − Ea00 )
a0
X
= ha00 | x |a0 i ha0 | x |a00 i (Ea0 − Ea00 )
a0
X X
= Ea0 ha00 | x |a0 i ha0 | x |a00 i − Ea00 ha00 | x |a0 i ha0 | x |a00 i
a0 a0
X X
= ha00 | xEa0 |a0 i ha0 | x |a00 i − Ea00 ha00 | x |a0 i ha0 | x |a00 i
a0 a0
X X
00 0 0 00
= ha | xH |a i ha | x |a i − Ea00 ha00 | x |a0 i ha0 | x |a00 i
a0 a0
X X
00
= ha | xH (|a i ha |) x |a i − Ea00 ha00 | x
0 0 00
(|a0 i ha0 |) x |a00 i
a0 a0
00 00 00 2 00
= ha | xHx |a i − E a00 ha | x |a i

4
Now, looking at the second term, we can proceed in two dierent ways: we can move Ea00 into the
matrix element, either to the left or to the right of the x2 . Either way, it ends up next to an a00
eigenstate and so can be turned into the Hamiltonian operator H ; so in the former case, we obtain

S = ha00 | xHx − Hxx |a00 i = ha00 | [x, H] x |a00 i

= − ha00 | [H, x] x |a00 i

while in the latter case we get

S = ha00 | xHx − xxH |a00 i = ha00 | x [H, x] |a00 i

Adding these two expressions, we nd

2S = ha00 | [x, [H, x]] |a00 i = − ha00 | [[H, x] , x] |a00 i

or
1
S = − ha00 | [[H, x] , x] |a00 i
2
The nested commutator is easy to evaluate:
 2 
p 1  2  −1 −i~
[H, x] = + V (x) , x = p ,x = (2i~p) = p
2m 2m 2m m

and so
2
−i~ (−i~) ~2
[[H, x] , x] = [p, x] = =−
m m m
So at last,  2
~2

1 ~
S= − − ha00 |a00 i = .
2 m 2m