Homework: Due 9-22-04

Sakurai Chap 1: 13,18,19,26. Chap 3: 9. Also, BJ 6.20 for spin

1

2
1.13
Label the three Stern-Gerlach apparati by 1,2,3, and let the spin state just after the three s.g. filters be 1), 2), 3), and
the intensity of the beam in the three cases be I
1
, I
2
, I
3
. Then we are told to use the normalization I
1
= 1. Since the
filters just find the probability for the beam to be in the appropriate state, we have that the intensity after s.g. 2 is
I
2
= (1 2)
2
and I
3
= I
2
(2 3)
2
. For the case at hand, in the S
z
basis, 1) = +), 2) = c +) +s -), 3) = -), where
c, s = cos
b

2
, sin
b

2
as in problem Sakurai 1.12, or BJ 6.20 (but for spin
1

2
, below). Then I
3
= c
2
s
2
.
To maximize I
3
, solve
dI
3

dc
=
d

dc
c
2
(1 -c
2
) = 2 c -4 c
3
= 0, which has solutions at c = 0,
1

H
2
. The second solution
gives the maximum of I
3
, and corresponds to b =
p

2
. The max value is found to be I
3
=
1

4
. Alternatively, recognize
that c s = sin b, so I
3
= sin
2
b, which is maximized by b =
p

2
.
1.18
a) Schwartz's inequality states that (a a) (b b) (a b)
2
. Since the inner product of any state with itself is real and
positive, one may define a
2
= (a a), and b
2
= (b b), with both a, b positive. Similarly, one may separate out the
magnitude and phase of (a b) = (a b) e
if
. Then, consider the linear combination c) = a) +l b), with l =
a

b
e
-if
,
and since (c c) 0,
((a +l
*
(b) (a) +l b)) = (a a) +l
2
(b b) +2 Re(l (a b))
= a
2
+(
a

b
)
2
b
2
+2
a

b
(a b)
= 2 a
2
+2
a

b
(a b) 0
or
(a b) < ab
which is equivalent to the Schwartz inequality. Another approach is to take l = -(b a) f (b b), which also works.
b) Suppose DA a) = lDB a). Then (a DA = l
*
(a DB, and ((DA)
2
) = (a DADA a) = l
2
((DB)
2
). It follows that
((DA)
2
) ((DB)
2
) = l
2
((DB)
2
)
2
Similarly,
hw9-22.nb: 10/3/04::11:19:57 1
1

4
([A, B|)
2
=
1

4
(AB - BA)
2
=
1

4
(a AB -BA a)
2
=
1

4
(a DADB -DBDA a)
2
=
1

4
(a (l
*
-l) (DB)
2
a)
2
=
1

4
l
*
-l
2
((DB)
2
)
2
The Schwartz inequality is saturated if
l
2
=
1

4
l
*
-l
2
which is true iff l is pure imaginary.
c) A gaussian wave function is given by
(x a) = y
a
(x) =
1

2 pd
2

4
e
ikx
e
-
x-a
2

4 d
2
Then
(X) =
]
x (a x) (x X a)
=
]
xy
a
*
(x) x y
a
(x)
=
1

H
2 pd
2

_
xxe
-
x-a
2

2 d
2
=
1

H
2 pd
2

_
y(y +a)e
-
y
2

2 d
2
= a
and
(P) =
]
x (a x) (x P a)
=
]
xy
a
*
(x) (-i

x
) y
a
(x)
=
1

H
2 pd
2

_
x(k -i
x-a

d
2
)e
-
x-a
2

2 d
2
= k
The problem asks for
(x DX a) = (x x -a a)
= (x -a) y
a
(x)
and
(x DP a) = (x p -k a)
= (-i

x
-k) y
a
(x)
= ((k -i
x-a

d
2
) -k) y
a
(x)
= (-i

d
2
(x -a)) y
a
(x)
= -i

d
2
(x DX a)
hw9-22.nb: 10/3/04::11:19:57 2
so the proportional relation with an imaginary coefficient is satisfied, and one may infer that Gaussian wave packets
satisfy the minimal uncertainty relation..
1.19
a) This is similar to 1.12 b. For the polarization state z
`
), b = 0, and the dispersion is ((DS
x
)
2
) =

2

4
. Note that (S
x
) = 0
for this state. This makes intuitive sense, since a state polarized in the z
`
direction shouldn't favor either +x
`
or -x
`
.
Similarly, ((DS
y
)
2
) =

2

4
, where S
y
=

2
s
y
, with s
y
=
l
'
.
.
0 -i
i 0
\
!
.
.. Again, notice that
(S
y
) =

2
( 1 0 )
l
'
.
.
0 -i
i 0
\
!
.
.
l
'
.
.
1
0
\
!
.
. = 0 for the current case. Also note that s
x
2
= s
y
2
= s
z
2
= 1, so
((S
x
)
2
) = ((S
y
)
2
) = ((S
z
)
2
) =

2

4
=
1

3
(S
2
) for any polarization state.
For the other side of the equation, we have [S
x
, S
y
| = i S
z
. For the z
`
polarization (S
z
) =

2
, so the right hand side
evaluates to
1

4
(i S
z
)
2
=

2

4
((S
z
))
2
=

4

16
, and we see that the relation is just satisfied.
b) For any eigenstate of an operator (a (DA)
2
a) = 0, and so for the +x
`
polarization state ((DS
x
)
2
) = 0. The L.H.S of
the inequality is therefore 0. On the R.H.S. we end up with (x
`
S
z
x
`
) = 0, and again the relation is just satisfied.
1.26
We are asked to find U that "connects" the S
z
diagonal basis and the S
x
diagonal basis. For this problem, let me adopt a
notation for states written in the S
z
diagonal basis n
`
, ), where n
`
refers to the direction of polarization. The transformed
states are given in the S
x
diagonal basis and denoted with a ', n
`
, )' . Then
Sz
`
, + =
l
'
.
.
1
0
\
!
.
., Sz
`
, - =
l
'
.
.
0
1
\
!
.
., Sx
`
, + =
1

H
2

l
'
.
.
1
1
\
!
.
., Sx
`
, - =
1

H
2

l
'
.
.
1
-1
\
!
.
., but in the S
x
diagonal basis
Sx
`
, + ' =
l
'
.
.
1
0
\
!
.
.,Sx
`
, - ' =
l
'
.
.
0
1
\
!
.
.. The transformation between the two bases for a ket a) is
a)' = U a)
Similarly, denote operators in the S
x
diagonal basis by a '. So, S
x
=

2

l
'
.
.
0 1
1 0
\
!
.
., but S
x
' =

2

l
'
.
.
1 0
0 -1
\
!
.
.. The
transformation relation for an operator S is
S
x
' = US
x
U

The eigenvalue equation for S

x
in the S
z
- basis is
S
x
x
`
, +) = +x
`
, +)
After transformation
(S
x
x
`
, +))' = (U S U

) (U x
`
, +)) = S
x
' x
`
, +)' = +x
`
, +)' = (+x
`
, +))' ,
so the transformed eigenstates are eigenstates of the transformed operators.
hw9-22.nb: 10/3/04::11:19:57 3
There are many ways one may proceed. My first reading of the problem was that the intent was to solve the operator
equation
S
x
' = US
x
U

This approach is given below. An alternative is to solve for U from the requirement that it transform the eigenstates
x
`
, +)' = U x
`
, +)
x
`
, -)' = U x
`
, -)
Since the eigenstates are all known, one may solve for U. This is equivalent to using the relation
U =
l
'
.
.
(e
1
e
1
') (e
2
e
1
')
(e
1
e
2
') (e
2
e
2
')
\
!
.
. =
1

H
2

l
'
.
.
1 1
1 -1
\
!
.
. where e
1I2
) = x
`
, ) and e'
1I2
) = x
`
, )' .
S
x
' = US
x
U

In the x
`
basis, S
x
is diagonal and is proportional to s
z
. On the other hand, if we write S
x
in the z-basis it is
proportional to s
x
. We are therefore looking for U such that
eq. 1: s
z
= Us
x
U

There are a lot of ways to do this. The crudest is to define the components of the matrix U
U =
l
'
.
.
a b
c d
\
!
.
. and then U

=
l
'
.
.
a
*
c
*
b
*
d
*
\
!
.
.. The values for a, b, c, d must satsify eq 1. Also U should be unitary
U

U =
l
'
.
.
a
*
c
*
b
*
d
*
\
!
.
.
l
'
.
.
a b
c d
\
!
.
. =
l
'
.
.
1 0
0 1
\
!
.
..
eq 1 is rewritten as
l
'
.
.
1 0
0 -1
\
!
.
. =
l
'
.
.
a b
c d
\
!
.
.
l
'
.
.
0 1
1 0
\
!
.
.
l
'
.
.
a
*
c
*
b
*
d
*
\
!
.
.
Evaluation of the 11 components of U

U = 1 and eq. 1 leads to

a
*
a +c
*
c = 1 (U

U = 1)
a
*
c +c
*
a = 1 (eq. 1)
rewriting a a e
if
a
, etc we get
a
2
+c
2
= 1 and 2 a c cos(f
a
-f
c
) = 1. These two equations require a = c =
1

H
2
, f
c
= f
a
f
a
is left undetermined.
Similarly, the 22 components yield
b
2
+d
2
= 1 and 2 b d cos(f
b
-f
d
) = -1. Which requires b = -d =
1

H
2
, f
d
= f
b
and again f
b
is undetermined.
We can also look at the 11 component of UU

= 1, which yields a
*
a +b
*
b = 1, which requires f
b
= f
a
.
The off diagonal components give relations such as a
*
b +c
*
d =
1

2
(e
i-f
a
+f
b

+e
i-f
c
+f
d

) = 0, This leads to a constraint

on phases (-f
a
+f
b
) = (-f
c
+f
d
), which is already satisfied, so there is no new information.
hw9-22.nb: 10/3/04::11:19:57 4
There is one overall phase of Uwhich cannot be resolved from the operator relation, since its conjugate always shows
up in U

. It can be resolved from the relationship between the two bases of kets. Since this phase is not observable, it is
conventionally just taken to be zero everywhere.
The solution is then
U =
1

H
2

l
'
.
.
1 1
1 -1
\
!
.
.
As a check, confirm that U = _
r
b
r
) (a
r
where the kets are in the S
z
basis,
a
1
W = ( 1 0 ), a
2
W = ( 0 1 ) , Sb
1
=
1

H
2

l
'
.
.
1
1
\
!
.
., Sb
2
=
1

H
2

l
'
.
.
1
-1
\
!
.
., so
U =
1

H
2

l
'
.
.
1 0
1 0
\
!
.
. +
1

H
2

l
'
.
.
0 1
0 -1
\
!
.
. =
1

H
2

l
'
.
.
1 1
1 -1
\
!
.
., as above.
3.9
a) Determine a spinor state from observed values of S
i

A pure ensemble is an ensemble that includes just one state. That state can be any state for the system. It is not
necessarily an energy eigenstate, but it could be expressed as a linear combination of energy eigenstates. For the 2
component spinor with an appropriate choice of overall phase, one can write the state as
l
'
.
.
a
b
\
!
.
. with a real and positive,
and b = b
r
+i b
i
complex. In matrix notation
a
2
+b
2
= 1
s
x

(S
x
) = ( a b
*
)
l
'
.
.
0 1
1 0
\
!
.
.
l
'
.
.
a
b
\
!
.
. = ab +ab
*
= 2 ab
r
s
y

(S
y
) = ( a b
*
)
l
'
.
.
0 -i
i 0
\
!
.
.
l
'
.
.
a
b
\
!
.
. = -i(ab -ab
*
) = 2 ab
i
s
z

(S
z
) = ( a b
*
)
l
'
.
.
1 0
0 -1
\
!
.
.
l
'
.
.
a
b
\
!
.
. = a
2
-b
2
which can be solved algebraically
a =
_ 1+s
z

2
b
r
=
s
x

2 a
The magnitude of b
i
is now determined by the normalization condition, but the sign is not. The magnitude of s
y
is
redundent (or acts as a consistancy check), but the sign determines the sign of b
i
.
b) Determine the density matrix from the ensemble averages S
i

The ensemble average for an operator O is given by Tr(r O). The density matrix is given by a Hermitian matrix with
Tr(r) = 1. Writing the matrix as r =
l
'
.
.
a c
c
*
b
\
!
.
. with a, b real and c = c
r
+i c
i
, we obtain the conditions
hw9-22.nb: 10/3/04::11:19:57 5
a +b = 1
s
x

[S
x
| =
2

Tr(r S
x
) = Tr
l
'
.
.
l
'
.
.
a c
c
*
b
\
!
.
.
l
'
.
.
0 1
1 0
\
!
.
.
\
!
.
. = Tr
l
'
.
.
c a
b c
*
\
!
.
. = 2 c
r
s
y

[S
y
| =
2

Tr(r S
y
) = Tr
l
'
.
.
l
'
.
.
a c
c
*
b
\
!
.
.
l
'
.
.
0 -i
i 0
\
!
.
.
\
!
.
. = i Tr
l
'
.
.
c -a
b -c
*
\
!
.
. = -2 c
i
s
z

[S
z
| =
2

Tr(r S
z
) = Tr
l
'
.
.
l
'
.
.
a c
c
*
b
\
!
.
.
l
'
.
.
1 0
0 -1
\
!
.
.
\
!
.
. = Tr
l
'
.
.
a -c
c
*
-b
*
\
!
.
. = a -b
which leads to
a =
1+s
z

2
b =
1-s
z

2
c =
s
x
-i s
y

2
Bransden and Joachain 6.20 with s = 1 s = 1 I 2
Obtain eigenvalues and eigenvectors of S
n
= n
`
S, where for spin
1

2
, S =

2
s. Work in the basis of polarization in the
z-direction, z
`
, +), z
`
, -). Then, taking the factor of

2
aside
n
`
s =
l
'
.
..
cos q sin q e
-if
sin q e
+if
-cos q
\
!
.
..
The eigenvalues are found by solving for n
`
s-l I = 0, or l
2
-cos
2
q -sin
2
q = l
2
-1 = 0, which has solutions
l = 1. Use a notation where cos q = c, sin q = s. Then, solving for the eigenvector for l = +1,
l
'
.
..
c s e
-if
s e
+if
-c
\
!
.
..
l
'
.
.
a
b
\
!
.
. = +
l
'
.
.
a
b
\
!
.
.
The upper line eqn is
c a +s e
-if
b = a
solving for a,
a = -b
s

c-1
e
-if
which gives an unnormalized eigenvector
l
'
.
.
.
.
s

1-c
e
-if
1
\
!
.
.
.
.
. Normalizing, the + eigenstate
Sn
`
, + =
l
'
.
.
.
.
.
.
..
_ 1+c

2
e
-if
_ 1-c

2
\
!
.
.
.
.
.
.
..
=
l
'
.
.
.
.
cos
q

2
e
-if
sin
q

2
\
!
.
.
.
.
Note, that for q = 0, one gets the z, +) eigenstate up to a phase, while for q = pf 2, and f = 0, pf 2 one appropriately
finds the x, +) and y, +) eigenstates. Similarly
hw9-22.nb: 10/3/04::11:19:57 6
Sn
`
, - =
l
'
.
.
.
.
.
.
..
-
_ 1-c

2
e
-if
_ 1+c

2
\
!
.
.
.
.
.
.
..
=
l
'
.
.
.
.
-sin
q

2
e
-if
cos
q

2
\
!
.
.
.
.

hw9-22.nb: 10/3/04::11:19:57 7