ntica I

Fsica Cua

(2016-17)

Exercise sheet 3
Lecturer: Kostas Glampedakis ([email protected])

Key:

= extension of theory
F = more challenging than most problems
 = straightforward

1. Use the parabolic approximation to calculate the first few energy levels of the potential
2

V (x) = V0 ex

What can you say about the higher energy states of this potential as compared to those
of the harmonic oscillator?
Solution
We need to Taylor-expand the potential assuming a small x (x2  1):
1
V (x) = V0 [ 1 + x2 + ...] V0 + V0 x2 = V0 + (2V0 )x2
2
That is, the approximate potential is that of a harmonic oscillator with spring constant
k = 2V0 , displaced by a constant value V0 . This system is described by the Schrodinger
equation:


2m
1 2
00
+ 2 E V0 kx = 0
~
2
We therefore ought to get displaced harmonic oscillator energy levels:


p
1

~,
= k/m,
n = 0, 1, 2, ...
En = En V0 = n +
2
That is,

 
1/2
1
2V0
En = V0 + n +
~
2
m
Since the full potential V (x) grows faster than the parabolic one, we expect that the
spacing between successive energy levels should increase as we move up in the spectrum.

2. The Stark effect: a charged harmonic oscillator is immersed in a static and homogenous
electric field E. The resulting potential is
1
V = kx2 qEx
2
where q is the particles charge. Calculate the shift in the systems energy spectrum
caused by the electric field.
Solution
Solving this exercise may seem difficult but it is actually rather easy. We notice that we
can write the potential as a perfect square plus a constant term:

2
1
qE
q2E 2
1 2
V = k x

k
x + V0
2
k
2k
2
This new potential describes a harmonic oscillator displaced in space x x and with
its energy shifted by a constant amount E E V0 . The former spatial displacement
cannot change the energy spectrum (otherwise my quantum oscillator here would have
different energy levels than your oscillator there). We thus expect that the electric field
causes a uniform displacement of the energy levels:




1
q2E 2
q2E 2
1
= n+
~
~
En = n +
2
2k
2
2m 2
This is the so-called Stark effect.
If you are not entirely convinced by the above argument, heres another one. Since
x = x qE/k, we can write the Schrodinger equation as


d2 2m 1 2
+ 2 E k
x = 0,
E = E V0
d
x2
~
2
This clearly leads to the energy spectrum quoted above.
3. A harmonic oscillator is found, at t = 0, in the state
(x) = N (2x + 1)ex

2 /2

Find the time evolution of the mean value hxi(t) using two different methods:
i) The Ehrenfest theorem.
ii) The expansion of (x) in energy eigenfunctions n (x).
Solution

We first determine the constant N by normalising (x):

Z +
Z +
2
2
2
dx|| = 1 N
dx(1 + 2x)2 ex = 1

The integral is Gaussian and we find

1
1
N2 = N =
3
3 1/4
i) According to Ehrenfests theorem, quantum mechanical mean values obey the classical
equations of motion. As we have seen, for the harmonic oscillator we have
d2 hxi
d2 hxi
2
+

hxi
=
0

+ hxi = 0
dt2
dt2
where the last equation assumes natural units (~ = m = = 1). The solution is
hxi(t) = A cos t + B sin t,

A, B = const.

As usual, A and B are fixed by imposing initial conditions:

hxi(0) = hxi0 ,

1
1
dhxi
(0) = hpi(0) = hpi0
dt
m
m

Then,
hpi0
m
Since we know the state of the system at t = 0 we can readily calculate the initial mean
values. For the momentum
Z +
dx 0 = 0
hpi0 = i
A = hxi0 ,

B=

This result follows trivially from the fact that is real-valued.

For the position we find
hxi0 = N

2
2
dxx(1 + 2x)2 ex = 2 N 2 =
3

Therefore, the Ehrenfest theorem led us to:

hxi(t) =

2
cos t
3

ii) Given that the oscillators first two energy eigenstates are

1 x2 /2
2
2
0 (x) = 1/4 e
,
1 (x) = 1/4 xex /2

and the rest are built by polynomials of higher order, it is obvious that
(x) = c0 0 (x) + c1 1 (x)
From this we get,

2
1
1
x + = c0 + c1 2x c0 = ,
3
3
3

r
c1 =

2
3

That is,
r
1
2
(x) = 0 (x) +
1 (x)
3
3
The time-evolution of this wavefunction is then,
r
r
1
2
1
2
iE0 t
iE1 t
it/2
(x, t) = 0 e
+
1 e
= 0 e
+
1 e3it/2
3
3
3
3
This can be simplified:
"
(x, t) = eit/2

1
0 +
3

2
1 eit
3

We know that a time-dependent overall phase is of no physical importance, hence

r
1
2
1 eit
(x, t) = 0 +
3
3
In exercise sheet 1 we have derived a general hxi(t) formula for a dual-state such as the
present one. After adapting the notation we have
hxi(t) = + cos(t + )
where
= |c0 |2 x00 + |c1 |2 x11 ,

= 2,

ei = c0 c1 x01 ,

xmn = hm |x|n i

As well known, x00 = x11 = 0 = 0, and = 0 since c0 , c1 and x12 are all real
quantities. Finally,

2 2
= 2c0 c1 x12 =
x01
3

For the remaining matrix element we find x01 = 1/ 2. Thus,

hxi(t) =

2
cos t
3

in perfect agreement with the calculation of (i).

4

4.  Obtain the energy eigenstates for the truncated parabolic potential:


,
x<0
V (x) =
1
2
kx , x > 0
2

Solution
This exercise is almost trivial: it is obvious that the wavefunctions consistent with the
presence of a potential wall at x = 0 are the odd (=anti-symmetric) eigenstates of the
harmonic oscillator (since for these states we have n (x) = n (x) n (0) = 0):
n (x) Hn (x)ex

2 /2

1
En = n + ,
2

n = 1, 3, 5, ...

5. The 3-D anisotropic oscillator: a particle of mass m moves under the influence of the
potential:
1
1
1
V (x, y, z) = k1 x2 + k2 y 2 + k3 z 2
2
2
2
where k1 , k2 , k3 are constants.
(i) Find the systems energy levels and compare the resulting spectrum against that an
isotropic (k1 = k2 = k3 = k) and a weakly anisotropic (k1 = k2 = k, k3 = 4k) oscillator.
What is the general conclusion you can draw from this comparison?
(ii) Derive the explicit form of the ground state wavefunction for the isotropic and weakly
anisotropic systems.
Solution
(i) After writing the 3-D wavefunction as (x, y, z) = x (x)y (y)z (z), the standard
separation of variables method leads to three 1-D harmonic oscillator equations:




1
1
2m
2m
2
2
00
00
x + 2 Ex k1 x x = 0, y + 2 Ey k2 y y = 0,
~
2
~
2


2m
1
z00 + 2 Ez k1 z 2 z = 0
~
2
where a prime stands for a derivative with respect to the functions argument and
E = Ex + Ey + Ez
Each of the above equations can be solved in the usual way and independently of the
other two. We thus have the energy spectrum:






1
1
1
E = nx +
~1 + ny +
~2 + nz +
~3 ,
nx , ny , nz = 0, 1, 2, ...
2
2
2
5

with i =

p
ki /m, i = {1, 2, 3}. The wavefunction is given by
nx ,ny ,nz (x, y, z) = nx (x)ny (y)nz (z)

In general, this 3-D oscillator does not display any degeneracy unless the i are related
in a special way (i.e. when we have resonances such as 1 /2 = integer etc.).
The simplest example is provided by
and therefore 1 = 2 = 3 = . In
becomes:
3
En = n + ,
2
In terms of (nx , ny , nz ) the spectrum

the isotropic oscillator, which has k1 = k2 = k3 = k

natural units ~ = m = = 1 the energy spectrum
n = nx + ny + nz = 0, 1, 2, ...
looks like:

2nd excited state:

E2 =

7
2

(1, 1, 0) (1, 0, 1) (0, 1, 1) (2, 0, 0) (0, 2, 0) (0, 0, 2)

1st excited state:

E1 =

5
2

(1, 0, 0) (0, 1, 0) (0, 0, 1)

= {100 , 010 , 001 }

ground state:

E0 =

3
2

(0, 0, 0)

= 000

The isotropic oscillator obviously exhibits some degeneracy in its energy states. This is
not surprising given that the x, y, z directions are completely equivalent here.
A second example of an oscillator with resonant frequencies is that of a weakly anisotropic
system with k1 = k2 = k, k3 = 4k 1 = 2 = , 3 = 2.
Adopting again natural units, we have the energy spectrum:


1
1
1
En = n + 2,
E = nx + + ny + + 2 nz +
2
2
2

n = 0, 1, 2, ...

The corresponding picture for the spectrum is:

2nd excited state:

E2 = 4 (1, 1, 0) (0, 0, 1)

1st excited state:

E1 = 3 (1, 0, 0) (0, 1, 0)

= {100 , 010 }

E0 = 2 (0, 0, 0)

= 000

ground state:

This system too exhibits degeneracy in its energy states. Note, however, that the degree
of degeneracy is lower than that of the previous isotropic oscillator. This is an example
(another one is the 3-D box potential discussed in a previous exercise) of the general rule:

higher symmetry

higher degree of degeneracy

(ii) After having obtained the energy spectra we crack on with the calculation of the
ground state wavefunctions.
For the isotropic oscillator we have (natural units)
0 (x) =

ex
1/4

2 /2

0 (y) =

ey
1/4

2 /2

0 (z) =

ez
1/4

2 /2

The full wavefunction is spherically symmetric:

000 (x, y, z) =

1
3/4

e(x

2 +y 2 +z 2 )/2

000 (r) =

1
3/4

er

2 /2

The symmetry of the ground state is no coincidence. The isotropic parabolic potential is
itself spherical:
1
1
V = k(x2 + y 2 + z 2 ) = kr2
2
2
and as a general rule the ground state always shares the symmetry as the potential.
The normal human units are easily restored using the characteristic length-scale
r
~
a0 =
m
Then
(x, y, z)

1
3/2

a0

(x/a0 , y/a0 , z/a0 )

which for the ground state leads to

1

000 (r) =

3/2
a0 3/4

er

2 /2a2
0

The case of the weakly anisotropic oscillator is slightly less straightforward. We first
restore units in the partial wavefunctions nx (x), ny (y) (which are identical to that of
the isotropic oscillator):
(x)

1
1/2
a0

(x/a0 ),

(y)

1
1/2
a0

(y/a0 )

that is,
0 (x) =

1
1/2
a0 1/4

ex

2 /2a2
0

0 (y) =

1
1/2
a0 1/4

ey

2 /2a2
0

The z-partial oscillator has a different frequency 3 = 2 and therefore a different characteristic length:
r
~
1
az =
= a0
m3
2
This means that in order to restore units in nz (z) we need to rescale the usual dimensionless result as
 1/4
1
2
2
2
1/2
(z) 1/2 (z/az ) 0 (z) =
a0 ez /a0

az
The full ground state wavefunction is:
 1/4
2
2
2
2
2
3/2
000 (x, y, z) =
a0 e(x +y +2z )/2a0
3

or in terms of the cylindrical radius $2 = x2 + y 2 ,

 1/4
2
2
2
2
3/2
000 ($, z) =
a0 e($ +2z )/2a0
3

This time, both the potential and the ground state wavefunction are axisymmetric about
the z-axis.
6. Use the algebraic formalism for the harmonic oscillator to show that in the energy state
|ni the following expressions are true:
1
hx2 i = hp2 i = n + ,
2

3
hx4 i = (2n2 + 2n + 1),
4

x = p =

p
n + 1/2

Solution
These relations have been worked out elsewhere (theory lecture) using the algebraic formalism. Here we present a different derivation. We shall only discuss hx2 i; the other
mean values can treated in a very similar way.
Previously, we calculated this mean value in the following manner:
1
1
hx2 i = hn|x2 |ni = hn|(a + a )(a + a )|ni = hn|a2 + a a + aa + (a )2 |ni
2
2
1
1
1
1
= hn|a a + aa |ni = hn|1 + 2a a|ni = hn|1 + 2N |ni = + n
2
2
2
2
where N = a a is the number occupation operator.
Now we show an alternative, and more geometric derivation. We note that hn|x2 |ni is
also the norm (squared) of the Hilbert space vector x|ni:
hn|x2 |ni = (x|ni) x|ni = ||x|ni||2
8

For this vector we have,

1
x|ni = (a + a )|ni =
2

n
|n 1i +
2

n+1
|n + 1i
2

This is the decomposition of x|ni in the orthonormal base |ni. Hence,

||x|ni||2 =

1
n n+1
+
=n+
2
2
2

hn|x2 |ni = n +

1
2

7. Coherent states:
(i) Calculate the mean value hEi and uncertainty E of a harmonic oscillators energy
in a coherent state.
(ii) Show that the coherent states lack the property of orthogonality (which is expected,
given that the annihilation operator a is not hermitian). In particular, verify that for
, R
2
h|i = e() /2
(iii) Show that an arbitrary coherent state with eigenvalue = 1 + i2 is also a state
of minimum uncertainty,
that is, it obeys xp = 1/2 (~ = 1). In addition, show that

x = p = 1/ 2.
Solution
(i) This part has been solved in the classroom.
(ii) As we have seen in the theory, a coherent state can be expanded in energy eigenstates
as

X
n
||2 /2|
|ni
|i = e
n!
n=0
For the another coherent state |i we similarly have,
|i = e||

2 /2|

X
m
|mi
m!
m=0

Then, the Dirac bracket of this pair of states with real eigenvalues , is
X X n m
X X n m
2
2
hn|mi = e( + )/2
nm
n! m!
n! m!
n
m
n
m
X ()n
2
2
2
2
h|i = e( + )/2
= e( + )/2+
n!
n
h|i = e(

2 + 2 )/2

h|i = e(

2 2 )/2

This is the result we were seeking.

9

(iii) We first need to compute the mean values hxi, hx2 i, hpi, hp2 i. We have:

1
1 
hxi = h|x|i = h|a + a |i = + h|a |i
2
2
Using h|a = h| in the last term,

1
hxi = ( + ) = 2Re[] = 21
2
We carry on with the computation of hx2 i:
1
1
hx2 i = h|x2 |i = h|(a + a )(a + a )|i = h|a2 + aa + a a + (a )2 |i
2
2
1 2
= [ + h|aa |i + h|a |i + h|(a )2 |i ]
2
1
= [ 2 + h|1 + a a|i + h|a |i + h|(a )2 |i ]
2
1
1
= [ 2 + 1 + 2h|a |i + h|(a )2 |i ] = [ 2 + 1 + 2 + h|a |i ]
2
2
1 2
= [ + 1 + 2 + ( )2 ]
2
1 1
1
1
hx2 i = + ( + )2 = + 2(Re[])2 = + 212
(1)
2 2
2
2
Then, the uncertainty in the particles position is:
(x)2 = hx2 i hxi2 =

1
2

1
x =
2

(~ = 1)

Next, we work out the momentum mean value:

1
1
hpi = h|a a |i = ( ) = 2Im[]
i 2
i 2
that is,
hpi = h|p|i =

22

In a similar way we obtain (exercise for the dedicated reader!)

hp2 i =

1
+ 222
2

Then,
1
1
p =
(~ = 1)
2
2
We have proven an important result: a coherent state minimises the uncertainty principle
with position and momentum participating on an equal footing (in natural units):
(p)2 = hp2 i hpi2 =

1
x = p = ,
2
10

x p =

1
2

A couple of comments:
i) We have opted for the elegance of a fully algebraic derivation, but we could have
arrived to the same result by direct use of the position wavefunction (x) = hx|i and
the calculation of integrals (of the generalised Gaussian type).
ii) We have found that in a coherent state hpi 6= 0. This result deserves some discussion
since we know that an energy eigenstate of a bound particle always leads to hpi = 0. Our
result can be understood better with the help of the quantum Newtons law (Ehrenfests
theorem!),
dhpi
= hF (x)i = hxi
(m = = ~ = 1)
dt
The fact that a coherent state has hxi =
6 0 allows a time-evolving hpi; this invalidates the
argument that led us to hpi = 0 in an energy eigenstate.
8. F Squeezed states:
The previous exercise naturally leads to the following question: can we find eigenstates
of minimum uncertainty, xp = 1/2, that do not have the extra property x = p (in
~ = 1 units)?
In order to explore such a possibility we define the generalised ladder operators:
1
A = ( x + ip ),
2

1
A = ( x ip ),
2

We also define the generalised coherent states as the solutions of the eigenvalue problem:
A|i = |i
i) Develop the algebraic formalism of A, A and show that
[A, A ] =
ii) Use the formalism to show that
1
xp = ,
2

r
x =

,
2

1
p =
2

iii) The above states are called squeezed (you may want to google the term squeezed
light). The reason behind this name can be understood if you calculate the wavefunction
(x) = hx|i and see how it behaves with respect to the parameter .
9. F Show that the energy levels of the potential,
1
g
V (x) = kx2 + 2
2
x
11

are given by (in natural units ~ = m = k = 1)

1
En = 2n + s + ,
2

n = 0, 1, 2, ... ,

s=


p
1
1 + 1 + 8g
2

As a sanity check of the quantum result verify that the ground state energy has the correct
classical limit.
Solution
Comment: this is a rather long exercise as it contains several new bits of ODE theory
applied to the solution of the Schrodinger equation.
The potential in this exercise represents a harmonic oscillator with a centrifugal potential
barrier (the g/x2 term). We will solve it following the general method we used in the
treatment of the harmonic oscillator.
Step 1: introduce natural units ~ = m = k = 1. The Schrodinger equation becomes


2
g
00
2
+ 2E x 2 = 0
x
where g is the (dimensionless) value of g in natural units. This equation has to be solved
in the interval 0 < x < + (since the potential goes to infinity as x 0).
A quick inspection
g x2 ) = 0
00 +( 2E x2 2
|{z} |{z} |{z} | {z }
2

+2

reveals that our equation is of third grade (remember that the grade of a term xm dn y/dxn
is defined as m n). In order to produce a second-grade equation (and therefore one
admitting a closed-form solution) we need to subtract the asymptotic solution.
Step 2: Obtain the asymptotic solution for x +.
For large x the Schrodinger equation becomes
00

x2 = 0

(x) = ex

2 /2

(we keep the decaying exponential and throw away the growing one).
Then, we write
(x) = (x)F (x) = ex

2 /2

F (x)

After inserting this back into the Schrodinger equation we obtain a new ODE for F (x):


2
g
00
0
F 2xF + 2E 1 2 F = 0
x
This new equation is of second grade (with terms of grade 2 and 0).
12

Step 3: Demand that the F -equation has a truncated power-series solution. That is, if
the solution takes the general form of a Frobenious series
F (x) = x

+
X

ak x k

k=0

we want to terminate the series at some power k = N . In such case

F (x) = xs PN (x),

PN (x) = polynomial of order N

We first need to obtain the starting power s. This is done by studying the x 0 limit
of the F -equation:
[s(s 1) 2
g ]xs2 = 0

s =

1 1p
1 + 8
g

2 2

(general rule: the dominant terms in the x 0 limit are the ones with the lowest grade).
Since (0) = 0 we should only keep the positive s-solution, i.e.
s = s+ =

1 1p
+
1 + 8
g
2 2

At x + we should have
F (x) x

with = s + N

Inserting this in the F -equation (general rule: the dominant terms at x + are the
ones with the highest grade). We are happy to find that the equation does admit a
power-law solution at infinity provided the energy takes certain discrete values:
(2 + 2E 1)x = 0

E=+

1
1
=s+N +
2
2

Thus we have obtained the energy E (the same way we did for the oscillator).
However, we are not done yet. According to ODE theory, the Frobenius series solution of
a second-grade ODE (of grades m and n) only contains powers which are integer multiples
of the equations step ` = |m n|. That is,


F (x) = xs a0 + a` x` + a2` x2` + ...
(this is because second-grade ODEs lead to recurrence relations of the form ak+` = f (k)ak
for the coefficients of the Frobenious series).
Our F -equation has step ` = |0 (2)| = 2 and therefore
N = 2n,

n = 0, 1, 2, ...

13

Therefore, the energy spectrum becomes

En = s + 2n +

1
1p
= 2n + 1 +
1 + 8
g
2
2

which is the result we were asked to prove (and we achieved that without actually solving
the Schrodinger equation!).
The eigenfunctions are of the form:
n (x) = e

x2 /2

s x2 /2

Fn (x) = x e

Pn (x),

Pn =

+
X

a2n x2n

n=0

and obviously satisfy the required boundary conditions at x = 0 and x = +.

Step 4: Restore human units.
The units of energy and length are the same as those of the simple harmonic oscillator,
i.e.
r
~
 = ~, a0 =
m
p
where of course = k/m. The parameter g has dimensions of (energy) (length)2 and
therefore its unit is
~2
g0 = a20 =
m
Then,
En En ,

g = g/g0

En =

1
2n + 1 +
2

8mg
1+ 2
~

!
(n = 0, 1, 2, ...)

Finally, we study the classical limit of the quantum ground state energy and compare it
against the classical result. The classical limit amounts to taking ~ = 0 or m . For
the n = 0 level we get

p
2mg
E0
~ = 2kg
~
The ground state of the classical system is the one of minimum energy, that is, the
particle will have zero kinetic energy and minimised potential energy. The latter is:
g
1
(E0 )cl = kx2 + 2
2
x
Then

d(E0 )cl
=0
dx

p
1
g
Emin = kx2min + 2 = 2kg
2
xmin

in agreement with the classical limit of the quantum result.

14

x2min =

2g/k

Although we were not asked to, we can also calculate the first few eigenfunctions of the
system. For the ground state, n = 0, we know that P0 = const. and therefore
2 /2

0 (x) = Cxs ex

where the constant C can be fixed by normalisation. As expected, the ground state
eigenfunctions has no nodes.
For the first excited state, n = 1, we should have
F1 (x) = xs (1 + ax2 )
The constant a is easily found if we insert F1 in the F -equation. The result is
a=
Then
s

1 (x) = Cx

2
2s + 1


2
2
2
x ex /2
1
2s + 1

and this has a single node (again, as expected).

Naturally, we can obtain a closed form solution for the general eigenfunction n (x) by
directly solving the F -equation. The analysis is similar to that of the oscillator and is
based on the method of asymptotic comparison (for more details see the mathematical
appendix of the oscillator lecture). Given that the x + local solutions of the F 2
equation are a power-law ( x ) and an exponential ex we know that the equation
can be transformed into a confluent hypergeometric equation (i.e. one of the very well
studied ODEs of mathematical physics).
The general solution of the F -equation is thus expressed in terms of the hypergeometric
function F (a; c; x) (also denoted as 1 F1 (a; c; x) in the literature) and it turns out to be:


F (x) = xs c1 F (a; c; x2 ) + c2 x2(1c) F (a + 1 c; 2 c; x2 )
where c1 , c2 are free constants and
1
1
1p
1
1 + 8
g=
a = (s+ ) = (1 E) +
2
2
4
2



1
E + s +
2

1
1p
1
c = 1 + (s+ s ) = 1 +
1 + 8
g = +s
2
2
2

(2)

For a solution regular at x = 0 one must set c2 = 0. Then, the exponential divergence
2
F ex is avoided provided we set
a = n

F (n; c; x2 ) = polynomial of order 2n

15

(n = 0, 1, 2, ...)

The condition a = n is precisely the energy quantisation formula we derived earlier.

The final result for the n-level eigenfunction is,


1 2 x2 /2
s
n (x) = Cx F n; s + ; x e
2
This concludes our analysis of this lengthy exercise.

16