Quantum Mechanics I

Solutions 5.
HS 2012
Prof. Ch. Anastasiou
Exercise 1. Uncertainty Relation
Compute
(S
x
)
2
S
2
x
S
x

2
,
where the expectation value is taken for the |S
z
, + state. Using your result check the generalized
uncertainty relation
(A)
2
(B)
2

1
4
|[A, B]|
2
with A S
x
, B S
y
.
Solution. The solution to this exercise is attached on page 5.
Exercise 2. Momentum-Space Wave Function
Prove the following:
i) p

| x| = i

p

(p

) ii) | x| =

dp

(p

)i

p

(p

) ,
where

(p

) = p

| and

(p

) = p

| are momentum-space wave functions.

(x) = Aexp

(x x
0
)
2
2d
2

.
(a) Knowing that the wave function extends over all space (i.e.

(x) L
2
(R)), nd the
normalization constant A such that ||| = 1.
Solution.
| =
_
+

dx|

(x)|
2
= |A|
2
_
+

dxe

(xx
0
)
2
d
2
= |A|
2
d
_
+

dy e
y
2
= |A|
2
d

(S.1)
where we made the change of variables y = (x x
0
)/d. Setting | = 1 and taking A real we get
A =
1

.
(b) Compute the expectation values x, p, x
2
and p
2
over the state | and describe their
dependence on d. Show that | is a minimum for the Heisenberg uncertainty relation, i.e.
show that it satises
(x)
2
(p)
2
=

2
4
.
(c) What is the interpretation of x, p, (x)
2
and (p)
2
if you consider |

(x)|
2
as a
probability density ?
1
Solution. We can already note that when (x x
0
) d, |

(x)|
2
0, so that the particle is localized
around x x
0
.
x =
_
+

dx

(x)

(x) =
1
d

_
+

dxxe

(xx
0
)
2
d
2
=
1
d

d
_
+

dy(dy + x
0
)e
y
2
=
1

x
0
_
+

dye
y
2
= x
0
(S.2)
where we made again the change of variables y = (x x
0
)/d, dx = ddy and x = x
0
+ dy, and the integral
_
+

dy ye
y
2
=
_
+

dy ye
y
2
vanishes for parity.
p = i
_
+

dx

(x)

(x) =
i
d

_
+

dxe

(xx
0
)
2
d
2
(2)
(x x
0
)
2d
2
=
i
d
3

_
x x
0
_
= 0
(S.3)
where we used
x

(x) =

(x)(2)
(xx
0
)
2d
2
and x = x
0
.
Using
_
+

dy
y
( ye
y
2
) = 0 (S.4)
as can be seen calculating the primitive at the extrema, we get:
_
+

dy
y
( ye
y
2
) =
_
+

dy
_
e
y
2
2y
2
e
y
2_
= 0 (S.5)
so that
_
+

dyy
2
e
y
2
=
1
2
_
+

dye
y
2
=

2
(S.6)
we can calculate:
x
2
=
1
d

_
+

dxx
2
e

(xx
0
)
2
d
2
=
1

_
+

dy(x
2
0
+ d
2
y
2
+ 2x
0
dy)e
y
2
= x
2
0
+
d
2
2
(S.7)
with x = x
0
+ dy. In the same way:
p
2
=

2
d

_
+

dxe

(xx
0
)
2
2d
2

2
x
e

(xx
0
)
2
2d
2
=

2
d

_
+

dx4
(x x
0
)
2
4d
4
e

(xx
0
)
2
d
2
=
=

2
d
5

_
+

dxe

(xx
0
)
2
d
2
_
x
2
+ x
2
0
2x
0
x
_
=

2
d
4
_
x
2
0
+
d
2
2
+ x
2
0
2x
0
x
0
) =

2
2d
2
(S.8)
where we used previous results and
_
+

dx

(x)

2
x

(x) =
_
+

dx
x

(x)

(x) (S.9)
We then see that the larger d, the larger the position uncertainty, while the smaller d, the larger the
momentum uncertainty, but the product is constant and independent of d, and reads:
(x)
2
(p)
2
=

2
4
(S.10)
(d) Consider now the wave-function
x| =

(x) = Aexp

(x x
0
)
2
2d
2
+ikx

.
Describe how

(x) changes when d is varied.

(x) e
ikx
and approaches a momentum eigenstate.
when d (x x
0
), d 0,

(x) (x x
0
) and approaches a position eigenstate.
x and x
2
remain equal as the additional term is just a phase factor. While
p = i
_
+

dx

(x)

(x) =
i
d

_
+

dxe

(xx
0
)
2
d
2
_
(2)
(x x
0
)
2d
2
+ ik
_
= k (S.11)
where we used
x

(x) =

(x)
_
(2)
(xx
0
)
2d
2
+ ik] and previous result. while
p
2
=

2
d

_
+

dxe

(xx
0
)
2
2d
2
ikx

2
x
e

(xx
0
)
2
2d
2
+ikx
=

2
d

_
+

dx
x
e

(xx
0
)
2
2d
2
ikx

x
e

(xx
0
)
2
2d
2
+ikx
=

2
d

_
+

dxe

(xx
0
)
2
d
2

(2)
(x x
0
)
2d
2
+ ik

=

2
2d
2
+
2
k
2
(S.12)
So that: (x)
2
(p)
2
doesnt change.
(f) Calculate the momentum representation of the state |, i.e.,
p| =

(p)
What is the relation between

(x) and

(p), i.e., through which mathematical operation

(p) varies when d changes.

(p) =
_
+

dx

(x)e
i
p

x
(S.13)
the relation is the Fourier Transform. We have to complete the square in the exponent to get:
(x x
0
)
2d
2
i(k
p

)x =
1
2d
2
_
(x x
0
)
2
+ i(
p

k)2d
2
(x x
0
+ x
0
)
_
=
=
1
2d
2
_
(x x
0
)
2
+ 2id
2
(
p

k)(x x
0
)
_
+ i(
p

k)x
0
=
1
2d
2
_
[(x x
0
+ id
2
(
p

k)]
2
+ d
4
(
p

k)
2
_
+ i(
p

k)x
0
=
1
2d
2
_
(x x
0
+ id
2
(
p

k)
_
2
+
d
2
2
(
p

k)
2
+ i(
p

k)x
0
(S.14)
where we initially collected a minus sign. The x integration is straight again with the usual change of
variables (since the other part id
2
(
p

k) in the square doesnt aect the variable change) and we are left
with the remaining part of the exponent:

(p) =
_
2
d

exp
_

d
2
2
(
p

k)
2
i(
p

k)x
0
_
(S.15)
and the normalization is dierent as in the starting integral we have A and not A
2
as usual. so the FT
of a gaussian centered at x
0
in real space is a gaussian in momentum space centered at k. so when d is
large the gaussian in momentum space is small, and has contributions only when (
p

k) 0, therefore is
sharp in momentum space; when when d is small the gaussian in momentum space is very extended and
momentum is not well dened, as we concluded also from (x)
2
, (p)
2
.
(g) We want now to understand how such wave packets spread in space when time ows. Let
us consider the wave packet

(x) with x
0
= 0. Its time evolution is governed by the
Schrodinger equation with appropriate boundary conditions

2
2m

2
x
2
(t, x) = i

t
(t, x) with (0, x) =

(x)|
x
0
=0
= Aexp

x
2
2d
2

.
3
For positive time t > 0 show that (t, x) can be written as a gaussian wave packet with a
time dependent width d(t) given by
d(t) = d

1 +

t
md
2

2
.
Hint. Make the following Ansatz
(t, x) =

dk C(k) exp

ikx i
k
2
2m
t

.
First show that (t, x) satises the Schrodinger equation and then solve the boundary
condition for C(k). You will also need (and prove if you want) the following identity

dxe
ikx
e

x
2

k
2
4
, () > 0 .
Also do not care about global phases in your nal result.
Solution. Plugging our Ansatz in the Schr odinger equation gives

2
2m
(ik)
2
= i
_
i
k
2
2m
_
,
what shows that is a solution. The boundary condition then reads
(0, x) =
_
dk C(k)e
ikx
= Aexp
_

x
2
2d
2
_
which can be directly solved by a simple Fourier transformation. We obtain
C(k) =
A
2
_
dxe
ikx
e

x
2
2d
2
=
Ad

2
e

k
2
d
2
2
,
using the hint. For positive time t > 0, the full solution can then be written as
(t, x) =
_
dk C(k) exp
_
ikx i
k
2
2m
t
_
=
Ad

2
_
dk e

k
2
d
2
2
e
ikxi
k
2
2m
t
=
Ad

2
_
dk e
k
2
(i
t
2m
+
d
2
2
)
e
ikx
= Ad
_
_
i
t
m
+ d
2
_
1
exp
_

x
2
2(i
t
m
+ d
2
)
_
. (S.16)
We have to take care of the non vanishing imaginary part for the factor in the exponential
(i
t
m
+ d
2
)
1
=
d
2
i
t
m

2
t
2
m
2
+ d
4
=
1
d
2
+
2
t
2
m
2
d
2

i
d
4
m
t
+
t
m
.
If we do not care about the imaginary parts (that do not contribute to the envelope of the wave function
but only to a global phase) we can rewrite the square root using

x =
_
|x|e
i arg(x)/2
as
_
_
i
t
m
+ d
2
_
1
=
e
i...
_
d
2
_
1 +
_
t
md
2
_
2
=
e
i...

d
_
d(t)
where d(t) as in the exercise. Plugging A we nally obtain
(t, x) =
1

1
4
_
d(t)
exp
_

x
2
2d(t)
2
_
exp {i . . .} .
4
Solution to Exercise 1.
5
Solution to Exercise 1.
6
Solution to Exercise 2.
7
Solution to Exercise 2.
8