Physics 701 - Solutions 3

Department of Physics
University of New Hampshire

Due Feb 17, 2010

Reading: Please read 2.1, 2.2 and 2.3 in Griffiths.

General statements about solutions to the Schr

odinger
Equation

[10] Problem 2.3 in Griffiths, reworded see also problems 2.1 and 2.2 in the
extra section.
By explicitly solving the Schrodinger equation for the infinite square well potential,
show that there are no acceptable solutions to the time-independent Schrodinger
equation for E = 0 and E < 0. Since V = 0, the implication is that there are no
solutions when E V everywhere.
Solution: When E = 0 the time-independent Schrodinger equation becomes:
d2
(x) = 0
dx2
which can be integrated twice to result in a solution:
(x) = A + Bx

(1)

(2)

where A and B are constants. The boundary conditions require (0) = 0, so we

must have A = 0 and (a) = 0, so Ba = 0 and thus B = 0. This results in a wave
function psi(x) = 0, which cannot be normalized, hence E = 0 is not an allowed
energy.
For E < 0, we solve the time-independent Schrodinger equation to get:

h2 d2
(x) = E(x)
2m dx2

d2
2mE
2
(x)
=
k
(x),
with
k
=
, real
dx2
h

The solution for this differential equation is then:

(x) = Ae+kx + Bekx
1

(3)

(4)

+0
0
The boundary conditions result

in: (x) = 0 Ae + Be = 0 B = A and
+ka
ka
(a) = 0 A e
e
= 0. We thus get either A = 0, but then (x) = 0
everywhere, or we get e+ka eka = 0 +ka = ka k = 0, but then again
(x) = 0, which cannot be normalized.

We conclude that E 0 is not allowed for this system.

Infinite Square Well Problems

(1) [10] Problem 2.4

Calculate hxi, hx2 i, hpi, hp2 i, x and p , for the nth stationary state (energy
eigenstate) of the infinite square well. Check that the uncertainty principle
is satisfied for all n. Which state comes the closest to the uncertainty limit?
For each of your answers, check to make sure you have the correct dimension
(units).
Solution: We use the general solution for a stationary state:
r
 n 
2
sin
x
(5)
n (x) =
a
a
We now compute:
Z
Z a


2 a
2 n
x sin
x dx
(6)
hxi = hn |x| n i =
n xn dx =
a 0
a
0
The function we integrate over is odd, but the limits are not symmetric, so we
cannot immediate make any immediate conclusions. Use an integral table,
or the trig identity: sin2 = 21 12 cos(2) and then integrate by parts on the
a
x) term using u = x and dv = cos( 2n
x), so v = 2n
sin( 2n
x). We
x cos( 2n
a
a
a
get:

Z a
2n
x x
hxi =
cos(
x) dx
a a
a
0

a

a
Z
11 2
1 a a
2n
x
2n
=
x

sin(
x)dx +
sin(
x)
a2
a 0 2n
a
2n
a
0
0

a
a
a
2n
=
+
cos(
x)
2
4n2 2
a
0
a
a
=
+
(1 cos(2n))
2 4n2 2
a
=
(7)
2
2

The result has units of length and is between 0 and a.The integral for hx2 i
is quite similar, except that you have to integrate by parts twice:
Z a
Z a



 2 
2
2
2 n
2


x sin
x dx
n x n dx =
x
= n x n =
a 0
a
0


Z a
x2 x2
2n
=

cos(
x) dx
a
a
a
0
 3 a Z a
 2
a
2x a
x
2n
x
2n
=
+
sin(
x)dx
sin(
x)
3a 0
a 2n
a
n
a
0

a 0
Z a
1 a
2n
ax
2n
a2

cos(
x) +
cos(
x)
=
3
a
2n2 2
a
0 n 2n
0
a

2
2
a
ax
2n
a
a
=
+ 2 2
sin(
x) 2 2 cos(2n)
3
2n 2n
a
2n
0


2
2
a
a
1
1
=
2 2 = a2
2 2
3
2n
3 2n
(8)
This result has units of length squared, as expected, and is larger than hxi2
for all n. Now we know we can do hpi the fast way:
hpi = m

hxi = 0
t

(9)

The units here are not important since the result is zero.
What about hp2 i? Since we know the energy E of the wave functions we can
take a shortcut:
|i
E |i = H
 
  2

p

1
2

hEi =

+ V  =
p
+ hV i
2m
2m

2
p

= 2m (hEi hV i)
(10)
In our case the potential is zero, so

2
n2 2 h
2
p
= 2m hEi =
a2
3

(11)

The long, but reliable, method for finding the same is:
Z a

2
n d2
n
22
=
h
sin( x) 2 sin( x)dx
p

a 0
a dx
a
2 Z a
2 2
2h
n n
n
=
sin( x)
sin( x)dx
2
a
a
a
a
0
2 2 2 Z a
nh

(x)(x)dx
=
2
a
0
2 2 2
n h

=
2
a

(12)

To check the units we reduce everything to dimensions of combinations of

mass, length and time. Since h
has units of Joules times seconds, h
/a has
units J s/m = N m s/m = kg m/s2 s = kg m/s, which are indeed the
units for momentum (remember classically p = mv). Note that the integral
that needed to be evaluated is just the normalization integral.
We then have:
r
r
1
1
1
a 1
2
1
1
2 2 =a
2 2 =
2 2
= a
3 2n
4
12 2n
2 3 n
nh
=
a r
r
nh a 1
2
h
n2 2
=

=
2
(13)
a 2 3 n2 2
2
3
r

x
p
x p

The uncertainty principle holds for all n and the ground state comes closest
to the uncertainty limit.
(2) Problem 2.5
A particle in the infinite square well of width a has as its initial wave function
an even mixture of the first two energy eigenstates:
(x, 0) = A [1 (x) + 2 (x)]

(14)

where n (x) are the properly normalized solutions of the time-independent

Schrodinger equation for the infinite square well potential.
a. [5] Normalize (x, 0) without performing any actual integral.
Solution: We know the n wave functions are orthonormal (since they
4

are solutions to an eigen equation) so:

Z
n (x)m (x)dx = nm
hn |m i =

(15)

then:
h |i = A A (h1 | + h2 |) (|1 i + |2 i)
= |A|2 (h1 |1 i + h1 |2 i + h2 |1 i + h2 |2 i )
1
= |A|2 2 = 1 A =
(16)
2
In integral notation:
Z
(x, 0)(x, 0)dx
h |i =

= AA
(1 (x) + 2 (x)) (1 (x) + 2 (x)) dx
Z

(1 (x)1 (x) + 2 (x)1 (x) + 1 (x)2 (x) + 2 (x)2 (x)) dx

= |A|2

1
= |A|2 2 = 1 A =
2

(17)

b. [5] Find (x, t) and |(x, t)|2 . Express the latter as a sinusoidal function of time. To simplify your result, let = 2 h
/2ma2 .
Solution: Normally (see problem 2.7) you need to do an integral to get
the coefficients of the wave function expansion:
X
cn |n i eiEn t/h
|(t)i =
n

(x, t) =

cn n (x)eiEn t/h

(18)

where the coefficients are given by:

Z
cn = hn |(0)i =

n (x)(x, 0)dx

(19)

But in this case, you can just match the cn to equation 14 and imme2
h
diately write: c1 = c2 = A and all other cn are zero. Setting = 2ma
2,
5

and using the energy levels En = n2 h

we can thus write:
1
1
(x, t) = 1 (x)eiE1 t/h + 1 (x)eiE1 t/h
2
2
r
r





2
1
1
it
=
sin
x e
+
sin
x ei4t
a
a
a
a

(20)

When we compute ||2 we need to be careful about the cross terms,

which do not drop out (until you integrate over them). We get:

  


1

2
2
it
i4t

|(x, t)| =
sin
x e + sin
x e
a
a
a
  




2
it
i4t
sin
x e
+ sin
x e
a
a



 
1
2
2
2
x + sin
x
=
sin
a
a
a



 


2
i3t
i3t
+ sin
x sin
x e
+e
a
a






 
 
2
1
2
2
2
=
x + sin
x + 2 sin
x sin
x cos(3t)
sin
a
a
a
a
a
(21)
c. [5] Compute hxi and find the angular frequency and amplitude of the
oscillation.
Solution:
Z a
hxi =
x |(x, t)|2 dx





Z0 a 
 
 
1
2
2
2
2
=
x sin
x + x sin
x + 2x sin
x sin
x cos(3t)
a
a
a
a
0 a
(22)
The first two integrals we have done already, together they equate to a2 .
The last integral requires:


Z a
 
2
x sin
x sin
x dx =
a
a
0


Z a   

3
16a2
x cos
x cos
x
= 2
a
a
9
0
(23)
6

So:
hxi =

a 16a

cos(3t)
2 9 2

The angular frequency is 3 =

so less than a2 .

3 2
h
,
2ma2

the amplitude is

(24)
16a
9 2

a2 0.3603...

d. [5] Compute hpi. (No integration needed! If you want to, do the integral
and very this.)
Solution: We can just take the derivative with respect to time of the
previous answer:

hxi
x
16ma
sin(3t)
=
3 2
8h
=
sin(3t)
3a

hpi = m

(25)

e. [5] If you measured the energy of this particle, what values might you
get and what is the probability of getting these values? Compute the
expectation value hHi and comment on the result. (How does it compare with E1 and E2 ).
2
h2
Solution: A measurement of the energy would results in either E1 = 2ma
2
2 2
or in E2 = 2mah2 with equal probability: P1 = P2 = 12 . The expectation
value for E is then:
hEi =

X
n

1
2h
2
5 2 h
2
|cn | En = (1 + 4)
=
2
2ma2
4ma2

(26)

This result is right in between E1 and E2 , as you would expect for an

equal weighted average.
(3) [10] Problem 2.6
Although the overall phase (embedded in the constant A) of the wave function is of no physical significance, the relative phase of the coefficients in
equation 2.17, does matter. For example, suppose we change the relative
phase in the previous problem to:


(x, 0) = A 1 (x) + ei 2 (x)
(27)

where is a constant. Find (x, t), |(x, t)|2 and hxi and compare your
answer with what you got before. Study the special cases of = 0, = /2
and = .
Solution: The exponential
term drops out of the normalization integral, so
q
we still have A =

1
.
2

Now:

r
(x, t) =
|(x, t)|2 =
+
=

r


 
2
1
1 i
it
sin
x e
+
e sin
x ei4t
a
a
a
a



 
2
1
2
2
sin
x + sin
x
a
a
a



 


2
i3t+i
i3ti
x sin
x e
+e
sin
a
a






 
 
1
2
2
2
2
sin
x + sin
x + 2 sin
x sin
x cos(3t )
a
a
a
a
a
(28)

So then:
hxi =

a 16a

cos(3t )
2 9 2
(29)

For = 0 we get our previous result. For = /2 we get hxi = a2

16a
16a
sin(3t), so hxi starts at a2 instead of a2 9
2.
9 2
a
16a
16a
For = we get hxi = 2 + 92 cos(3t), so hxi starts at a2 + 9
2 instead of
16a
a
92 .
2

Extra Problems

This extra problem section is for those of you who want to dig a little more deeply
into the material. These problems are a bit more difficult. Sometimes they may
be a bit too advanced to expect everyone to do them. Thus, they are not required,
but I encourage you to think about them, and then check the solutions when they
become available.
(1) Problem 2.1 in Griffiths.
8

a. Suppose E is not real, then the solution to the time dependence of

(x, t) has an energy term E = Er + iEi where now Er and Ei are both
real. We then get:
|(t)i = |i ei(Er +iEi )t/h = |i eEi t/h eiEr t/hbar

(30)

The normalization then leads to:

h |i = e2Ei t/hbar h |i = 1

(31)

Since h |i = 1 for all time t, we must have e2Ei t/hbar = 1 Ei = 0.

2

+
b. If satisfies the time independent Schrodinger equation: 2m
x2
V = E, then, noting that V , E and m are real, the complex conjuh2 2

+ V = E so that both and satisfy the

gate gives: 2m
x2
Schrodinger equation for the same energy E. Then any linear combination of and also satisfy the Schrodinger equation (with the same
energy E) Check for yourself by inserting A = c1 + c2 . Thus the
combinations ( + ) and i( ), both of which are real, also satisfy
the Schrodinger equation. So from any complex solution we can always
construct two real solutions.

c. If (x) satisfies the time independent Schrodinger equation, then changing variables x x and noting that 2 /(x)2 = 2 /x2 , we get:

h
2 2
(x) + V (x)(x) = E(x)
2m x2

(32)

Now if V (x) = V (x), in other words, the potential is an even function,

then (x) is a solution to the same Schrodinger equation. Thus the
combinations = (x) (x) are also solutions, where + is even
and is odd. For any particular energy level, in one dimension, one
or the other will be zero.
(2) Problem 2.2 in Griffiths. Writing the Schrodinger equation as:
d2
2m
= 2 [V (x) E]
2
dx
h

(33)

you can see that if E < Vmin then and 00 always have the same sign. If
is positive (negative) then 00 is also positive (negative). This means that
always curves away from the x-axis. But we must have a normalizable
9

solution, which is not possible if continues to curve away from the x-axis.
To get to curve back so that it approaches zero as x goes to infinity, it needs
to turn over, which requires a negative second derivative for a positive
and visa versa. We conclude that for E < Vmin we cannot get a normalizable
solution.
Total number of points: 55

10