You are on page 1of 73

An Introduction to Supersymmetric Quantum

Mechanics and Shape Invariant Potentials


Jens Maluck
May 15, 2013
Supervisor: Sebastian de Haro Reader: Jan Pieter van der Schaar
Major: Science
Amsterdam University College
Abstract
This thesis represents an introduction to supersymmetric quantum mechan-
ics in 1-dimension. Demonstrations for the fundamental aspects of SUSY QM
will be given with help of simple examples. The relationship between SUSY
partner potentials, their eigenvalues and eigenfunctions are explicitly derived
and an analysis of specic attributes such as reectionless potentials is given.
Furthermore, I will demonstrate that given a potential with at least one bound
state one can, using SUSY, create a family of partner potentials having the
same energy eigenvalue and S-matrix. The elegant operator method for solving
the standard non-relativistic harmonic oscillator is extended to a group of po-
tentials classied as shape invariant. And lastly, derived tools are used to solve
multiple problems, such as to nd the rst few energy states of the Hydrogen
atom. Possible application to non-central potentials are illuminated.
Contents
List of Figures 2
1 Introduction 3
2 1-Dimensional Supersymmetric Quantum Mechanics 5
2.1 Factorisation of a General Hamiltonian . . . . . . . . . . . . . . . . . . 5
2.2 Simple Example of Two Partner Potentials . . . . . . . . . . . . . . . . 12
2.3 Supersymmetric Innite Square Well . . . . . . . . . . . . . . . . . . . 13
2.4 Supersymmetric Inuence on Scattering Relations . . . . . . . . . . . . 18
2.4.1 A Reectionless Potential . . . . . . . . . . . . . . . . . . . . . 20
2.5 Hierarchy of Hamiltonians . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.5.1 Hierachy of the Innite Square Well . . . . . . . . . . . . . . . . 25
2.6 Broken Supersymmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.7 Supersymmetric Harmonic oscillator and the SUSY Algebra Relations . 31
3 Shape Invariant and Solvable Potentials 37
3.1 Shifted Oscillator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.2 Example of Shape Invariance of a Given Wave Function . . . . . . . . . 49
3.3 Hulthen Potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
3.4 The Hydrogen Atom . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
3.5 Non-central Potentials . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
4 Conclusion 68
5 References 70
1
List of Figures
1 First Energy Levels of Two SUSY Partner Hamiltonians . . . . . . . . 10
2 Potentials V
1
and V
2
with Respect to Superpotential W(x) = x
10
. . . 13
3 SUSY Innite Square Well Partner Potentials . . . . . . . . . . . . . . 18
4 Energy Spectrum of a Hamiltonian Hierarchy via SUSY . . . . . . . . . 25
5 The Innite Square Well and its First Three Partner Potentials . . . . 27
6 Invariant Potentials and Their Characteristics . . . . . . . . . . . . . . 44
7 Plot of
(1)
0
for r
0
= A = 1. . . . . . . . . . . . . . . . . . . . . . . . . . 50
8 Plot of the superpotential to
(1)
0
for r
0
= 1. . . . . . . . . . . . . . . . 51
9 Plot of V
1
and V
2
as derived from
(1)
0
for r
0
= 1, where V
2
is the higher
of the two functions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
10 First 3 energy levels for
(1)
n
with r
0
= 1 . . . . . . . . . . . . . . . . . 57
11 The Hulthen potential for a choice of parameters V
0
= 1 and = 5. . . 59
12 The Radial Coulomb Potential . . . . . . . . . . . . . . . . . . . . . . . 60
13 Energy Levels of the Hydrogen Atom . . . . . . . . . . . . . . . . . . . 64
2
1 Introduction
Within eld theory, supersymmetry (SUSY) is a symmetry that relates fermions to
bosons in order to obtain a unied description of all basic interactions of nature. In the
context of particle physics, it predicts that each fermion is associated with a boson and
vice versa, called their superpartner. If supersymmetry is a good symmetry of nature,
each set of superpartners have the same quantum numbers and mass and only dier by
a half-integral unit in spin. With respect to the unied theory for the basic interactions
of nature, it predicts the existence of superpartners for all elementary particles, thereby
doubling the current extent of the Standard Model. Unlike most other symmetry
generators, the generators of supersymmetry are spinor charges based on a graded Lie
algebra involving also anticommutators, rather than just commutators. The theory
of supersymmetry, however, is so far not supported by experimental evidence, since
for example no scalar electron has been observed with a mass smaller than approx.
100 GeV. This suggests that supersymmetry needs to be a spontaneously broken
symmetry of nature. It was while trying to understand the SUSY breakdown in
eld theory that supersymmetric quantum mechanics (SUSY QM) was rst studied
[3]. Witten [21], in that context was rst to construct a simple quantum mechanical
supersymmetric system involving a spin-1/2 particle moving in one direction. He,
thereby, also dened the algebra that needed to be satised in order to express the
supersymmetric Hamiltonian in terms of the charge operators. These now have become
the dening algebraic relations of supersymmetric quantum mechanics.
As soon as the study of SUSY QM became more popular, it was quickly realised
that it oers a lot more than simply being a model for testing eld theory methods.
Over the last three decades SUSY QM has oered deeper insight into various aspects
of standard non-relativistic QM. For example, it claries why certain one-dimensional
potentials are analytically solvable and others are not. It, then, further helps in nding
new solvable potentials. It also exemplies the factorisation method, rst introduced
by Infeld and Hull [12], and oers a collection of new and more powerful approximation
3
methods, such as the supersymmetric WKB method. Together with the concept of
shape invariant potentials (SIPs), rst introduced by Gendenshtein [8] in 1983, a class
of potentials can be identied, which are all analytically solvable. Supersymmetry in
quantum mechanics, thereby, also illustrates the relationship between energy spectra
and wave functions of two potentials, connected by new operators emerging from the
existence of supercharges. These supercharges commute with the super-Hamiltonian
and are related to its invariance under translation in superspace, which leads to denite
relations of the spectra of the fermionic and bosonic sectors [3].
The purpose of this thesis is to oer an introduction to supersymmetric quantum
mechanics and address some of the afore-mentioned aspects it encompasses with use
of many examples. The main motivation is to derive a powerful set of tools based on
the principles of supersymmetry that can help us in solving considerably many QM
potential problems. We will see that the shape invariance condition and the method
of repeated refactorisation will play a major role in this. Furthermore, in the course
of this paper the following fundamental questions will be considered:
1. The innite square well is one of the simplest exactly solvable QM problems with
a simple discrete energy spectrum. Do any other potentials exist that correspond
to the same energy spectrum?
2. Only few of the considerable number of possible QM potentials are exactly solv-
able with analytical tools. What distinguishes these from other potentials?
3. What are the underlying aspects with respect to supersymmetry that make some
potentials reectionless?
4. Can the very fundamental, but yet elegant, operator method for solving the 1-d
QM harmonic oscillator be extended to other potentials?
Inspiration for these questions was given by [14]. The topics incorporated in this thesis
were inspired by the sources listed in the references.
4
2 1-Dimensional Supersymmetric Quantum Mechan-
ics
2.1 Factorisation of a General Hamiltonian
For the factorisation process, let us start with a general one-dimensional single particle
QM Hamiltonian
H

2
2m
d
2
dx
2
+ V (x). (1)
Instead of solving the problem for a given potential, we will take an approach that will
still allow us to vary the potential and thus give solutions for general Hamiltonians.
For this approach we will assume the ground state wave function
0
(x) and dene it to
be nodeless and to vanish at x = in order for it to be normalisable. Furthermore,
we can set the ground state energy equal to zero without losing any generality up to
a constant energy shift. Hence the Schr odinger equation
1
for the ground state wave
function
0
(x) is
0 =

2
2m
d
2
dx
2

0
(x) + V (x)
0
(x). (2)
Considering that the ground state wave function is nodeless, we can now solve equation
(2) for the potential, such that
V (x) =

2
2m

0
(x)

0
(x)
. (3)
This is generally the case. Knowing the ground state wave function, allows us to
calculate the potential V (x) up to a constant. This result will become of importance
later.
In order to go from standard quantum mechanics to supersymmetric quantum
1
In the course of this thesis, whenever it is referred to the Schrodinger equation we explicitly
mean the time-independent Schrodinger equation, unless otherwise stated.
5
mechanics, we have to make the simple Hamiltonian factorisation
H
1
= A

A (4)
where the operators A and A

are dened as
A =

2m
d
dx
+ W(x), A

2m
d
dx
+ W(x). (5)
The quantity W(x) is known as the superpotential in SUSY QM literature for reasons
that will become apparent later. Notice that the Hamiltonian is now labelled with
an index. This is important and will help us keep track of which Hamiltonian we
are discussing. Of course it still has the same underlying structure as given by eq.
(1).With help of equations (4) and (5), it is now possible for us to identify the quantity
V
1
(x) by simple insertion
H
1
(x) =
_

2m
d
dx
+ W(x)
__

2m
d
dx
+ W(x)
_
(x)
=

2

2m
d
2
dx
2
(x)

2m
[W

(x)(x) +

(x)W(x)]
+ W(x)

2m

(x) + W(x)
2
(x)
=
_

2m
d
2
dx
2

2m
W

(x) + W(x)
2
_
(x)
V
1
(x) = W
2

2m
W

(x). (6)
Equation (6) is famously known as the Riccati equation. It is important to realise
that eq. (3) still holds and eq. (6) represents an extra condition on the potential,
now known as V
1
(x). To obtain a solution for W(x) in terms of the known ground
state wave function one has to recognise that once we satisfy A
0
(x) = 0 then we
automatically get that H
1

0
= A

A
0
= 0. The operator A thereby represents an
annihilation operator, similar to the operator a in the treatment of the quantum
6
mechanical harmonic oscillator. Then
W(x) =

2m

0
(x)

0
(x)
. (7)
Notice that eq.(7) together with eq. (6) give back eq. (3).
Furthermore, using the operator A, one can nd a general solution to the ground
state wave function depending only on W(x). Since A annihilates the ground state
wave function, we see that
A
0
= 0

2m
d
dx

0
+ W(x)
0
= 0

0
= N exp

2m

x
W(k)dk

. (8)
Hereby we have managed to turn a second order dierential problem (solving the
Schr odinger equation) into a rst order dierential problem if the superpotential is
known. This already gives SUSY QM a distinct advantage in comparison to standard
QM. Now, to continue constructing the SUSY theory with regards to H
1
we dene a
second Hamiltonian H
2
by reversing the order of factorisation, thus H
2
= AA

. H
1
and H
2
are known as partner Hamiltonians. Following the same calculations as we
did to obtain equation (6), we realise that
H
2
(x) =
_

2m
d
dx
+ W(x)
__

2m
d
dx
+ W(x)
_
(x)
=

2

2m
d
2
dx
2
(x) +

2m
[W

(x)(x) +

(x)W(x)]
W(x)

2m

(x) + W(x)
2
(x)
=
_

2m
d
2
dx
2
+

2m
W

(x) + W(x)
2
_
(x)
7
and thus
H
2
=

2
2m
d
2
dx
2
+ V
2
(x) with V
2
(x) = W
2
+

2m
W

(x), (9)
where W(x) is still given by eq. (7), because eq. (3) also holds in the case of V
2
(x).
The two potentials V
1
(x) and V
2
(x) are known as supersymmetric partner potentials.
This connection via SUSY relates the two Hamiltonians and thereby their energy
levels, wave functions and S-matrices, as we will now discover. For this, rstly notice
that the energy spectra of both Hamiltonians are positive semi-denite (i.e. E
n
0).
This can be best seen in dirac notation, by
|A

A| = A|A = ||A||
2
. (10)
With help of the Schrodinger equation, we can now start to nd the relations
between the potentials, for n > 0 we observe
H
1

(1)
n
(x) = A

A
(1)
n
(x) = E
(1)
n

(1)
n
(x) (11)
implies
H
2
(A
(1)
n
(x)) = AA

A
(1)
n
(x) = E
(1)
n
(A
(1)
n
(x)). (12)
Similarly, taking the Schrodinger equation for H
2
H
2

(2)
m
(x) = AA

(2)
m
(x) = E
(2)
m

(2)
m
(x) (13)
reveals
H
1
(A

(2)
m
(x)) = A

AA

(2)
m
(x) = E
(2)
m
(A

(2)
m
(x)). (14)
Thereby, we already see one of the most important features of SUSY QM, namely the
relation between the eigenfunctions and eigenvalues between H
1
and H
2
. It is now
apparent that A
(1)
n
(x) is an eigenfunction of H
2
and A

(2)
m
(x) is an eigenfunction
of H
1
. One can further see that the corresponding energy state is in fact the one of
8
the partner Hamiltonian. This is very important, because it tells us that the energy
spectra of both Hamiltonians are identical up to perhaps zero energy level. Considering
that A
(1)
n
(x) is an eigenfunction of H
2
we can express
(2)
m
(x) = cA
(1)
n
(x), where c is
a normalisation constant. Assuming that
(1)
n
(x) is normalised and using the relation
we obtained in equation (11), we can now normalise in order to obtain c
1 =
_

(2)
m

(2)
m
= |c|
2
_

(1)
n
A

A
(1)
n
= |c|
2
E
(1)
n
_

(1)
n

(1)
n
(15)
and thus
c = (E
(1)
n
)

1
2
. (16)
Hence, we obtain the relation between the wave functions to be

(2)
m
= (E
(1)
n
)

1
2
A
(1)
n
. (17)
Including the fact that A
(1)
0
= 0 we see with help of eq. (17) that H
2
does not have
a corresponding zero energy ground state. Now with help of equations (10)-(16) it is
possible for us to determine that m = n1 and then nalise the relations of the wave
functions and energy spectra between the two partner Hamiltonians in the following
equations
E
(2)
n
= E
(1)
n+1
, E
(1)
0
= 0 (18)

(2)
n
=
_
E
(1)
n+1
_

1
2
A
(1)
n+1
(19)

(1)
n+1
=
_
E
(2)
n
_

1
2
A

(2)
n
. (20)
Notice some interesting properties. Identifying that m = n 1, immediately gave
us a relation between the energy spectra of the two Hamiltonians, namely that the
nth excited state of H
2
is equal to the (n 1)th excited state of H
1
. If
(1)
n+1
(
(2)
n
)
is normalised then from equations (19) and (20) it follows that
(2)
n
(
(1)
n+1
) is also
9
normalised. Moreover, the operators A and A

do not just convert eigenfunctions of


H
1
or H
2
to eigenfunctions of the respective other Hamiltonian with the same energy,
but by doing so they also create or annihilate an extra node in the eigenfunction.
However, the operator A annihilates the ground state of H
1
(i.e. A
(1)
0
= 0 ), leaving
us to conclude that there is not supersymmetric partner for that state. Therefore,
knowing all the eigenfunctions and energy states of H
1
, supersymmetry lets us create
all solutions of H
2
using operator A or the other way around, knowing all the eigenval-
ues and wave functions of H
2
we can, with help of the operator A

, create all solutions


of H
1
, except its ground state. In fact, as we will see in section 2.6, it is convention
to label the partner Hamiltonian, whose ground state wave function is annihilated by
A, as H
1
. The derived relationships are best summarised visually and can be seen in
Fig. 1. Here the degeneracy between the states becomes more evident.
Figure 1: First energy levels of two SUSY partner Hamiltonians and their relations
by A and A

This degeneracy of the energy spectra of H


1
and H
2
might seem quite surprising,
however investigating the underlying SUSY algebra one can easily see the reason for
its existence. Matrix notation oers the most clarity in this context. Let us consider
10
the matrix of a SUSY Hamiltonian of the form
H =
_
_
H
1
0
0 H
2
_
_
. (21)
As we will see, this matrix is part of a closed algebra with commutation and anti-
commutation relations. To represent our operators A and A

consider the following


matrices in conjunction with H
Q =
_
_
0 0
A 0
_
_
and Q

=
_
_
0 A

0 0
_
_
. (22)
The operators Q and Q

are also generally known as supercharges in SUSY literature


and can be interpreted as operators changing bosonic degrees of freedom into fermionic
degrees of freedom and the other way around. We will elaborate on this a bit further in
the treatment of the SUSY harmonic oscillator in section 2.7. The closed superalgebra
sl(1, 1), providing the algebraic framework for our formulations of supersymmetry in
1 dimension, is then described by the following relations [3]:
[H, Q] = [H, Q

] = 0,
{Q, Q

} = H , {Q, Q} = {Q

, Q

} = 0,
(23)
where {A, B} represents the anti commutator of A and B. Now, we can clearly see
that the degeneracy in the spectra of the two partner Hamiltonians comes from the
fact that Q and Q

commute with H. We will come back to eqs. (23) shortly when


we explore the spontaneous breaking of supersymmetry.
Summarising this section, it was shown that if given an exactly solvable potential
with at least one bound state, we can always construct a SUSY partner potential,
which is also exactly solvable. More specically, the energy spectra of the two poten-
11
tials are identical, except the rst ground state of V
1
. Furthermore, the bound state
energy eigenfunctions of the partner potential are easily obtained via eq. (18)-(20).
Lets visualise the implications of this section with help of two examples. The second
example, concerning the supersymmetric innite square well, will specically show the
degeneracy of the energy spectra of partner potentials.
2.2 Simple Example of Two Partner Potentials
Lets assume a simple superpotential of the form W(x) = x
5 2
, which represents a
shifted oscillator. Now, from eqs. (6) and (9) we get that
V
1
(x) =
2
x
10
5x
4
and V
2
(x) =
2
x
10
+ 5x
4
,
where we chose

2m
= 1.
Looking at their respective graphs in Fig. 2, one can see that V
1
represents a double
potential well, whereas V
2
is a single potential well. Intuitively, one would expect these
potentials to result in dierent energy spectra and wave functions, but as we know
from section 2.1 these potentials actually yield exactly same energy spectra and their
wave functions can also be derived from one another, with exception of the ground
state of V
1
. This is to illustrate the advantage supersymmetry gives us when dealing
with partner potentials.
2
This superpotential might seem quite restrictive and not give a good indication of the applica-
bility of SUSY QM to real problems, however in this context it is also merely used to illustrate the
dierence between possible partner potentials. Furthermore, when examining the resulting partner
potentials, one can actually see that the coecients are not as tightly adjusted as it might have
seemed due to the dierences in their powers. Additionally, rescaling of x tx gives additionally
freedom.
12
Figure 2: Potentials V
1
and V
2
with respect to superpotential W(x) = x
10
.
2.3 Supersymmetric Innite Square Well
In the previous example we have seen how to construct the partner potentials starting
with a superpotential. Now, we will illustrate that the opposite direction is also
possible. Let us start with the well-known potential of a innite square well and then
determine its SUSY partner potential. Assume the particle to be of mass m and the
well of length L. Hence,
V (x) = 0, for 0 x L,
V (x) = , for x 0 and x L.
(24)
The normalised ground state wave function and ground state energy of this problem
are well known and can be found in numerous QM textbooks, such as [9]. They are
given by

(1)
0
=
_
2
L
_
1/2
sin
_
x
L
_
, for 0 x L (25)
E
0
=

2

2
2mL
2
. (26)
13
In order to be able to factorise the Hamiltonian, it needs to be dened to have a zero
energy ground state, thus H
1
= H E
0
. Thus
E
(1)
n
=
n
2

2
2mL
2


2

2
2mL
2
(27)
= (n
2
1)

2

2
2mL
2
, for n = 1, 2, .... (28)
However, having n start at 1 in this context provides more confusion than service,
since we are speaking of ground state energy, which is generally labelled by 0. Hence
we will change the labelling accordingly to n n 1. Then the energy eigenvalues
and the normalised wave functions are given by
E
(1)
n
=
n(n + 2)
2

2
2mL
2
, for n = 0, 1, 2, ... (29)

(1)
n
=
_
2
L
_
1/2
sin
_
(n + 1)x
L
_
, for 0 x L. (30)
The rst and second derivative are then given by

(1)
n
=
_
2
L
_
1/2
cos
_
(n + 1)x
L
_
(n + 1)
L
, for 0 x L, (31)

(1)
n
=
_
2
L
_
1/2
sin
_
(n + 1)x
L
_
(n + 1)
2

2
L
2
, for 0 x L. (32)
We can now check if eq. (3) would return the innite square well potential that we
started o with. Then
V
1
(x) =

2
2m

_
2
L
_
1/2
sin
_
x
L
_

2
L
2
_
2
L
_
1/2
sin
_
x
L
_
=

2
2m

2
L
2
, for 0 x L. (33)
14
As we can see, it returns the innite square well potential up to a constant, which
as it turns out is the ground state energy that we shifted it by. Hence, thus far our
theory is coherent. If we now follow eq. (7), we nd that the superpotential is given
by
W(x) =

2m

L
cot
_
x
L
_
, for 0 x L. (34)
At rst it might seem surprising that a rather complicated superpotential, as given
here, can be related to the possibly simplest potential of them all, the innite square
well. However, note that W can be entirely derived from the ground state wave
function via eq. (7). Due to this dependence W(x) can potentially have extremely
unrelated forms than the potentials we start o with. Consequently, the partner
potentials derived via the superpotential can also be entirely unrelated to its simple
partner. This can be seen using eq. (9), then
V
2
(x) = W(x)
2
+

2m
W

(x), (35)
with
W(x)
2
=

2
2m

2
L
2
cot
2
_
x
L
_
, (36)
and
W

(x) =

2m

2
L
2
csc
2
_
x
L
_
, (37)
we thereby get that
V
2
(x) =

2
2m

2
L
2
cot
2
_
x
L
_
+

2
2m

2
L
2
csc
2
_
x
L
_
. (38)
Remember that cot
2
() = csc
2
() 1, then we nally obtain
V
2
(x) =

2
2m

2
L
2
_
2 csc
2
_
x
L
_
1
_
, for 0 x L. (39)
15
Notice, if we had changed the sign in V
2
from + to - we would have obtained the same
solution for V
1
we found in eq. (33), which in turn shows that our found superpotential
and partner potential are indeed correct. Of course the potential V
2
is part of a
Hamiltonian H
2
and as we have seen in section 2.1 we can calculate its wave functions
by applying the operator A to the known wave functions of H
1
. Remember that in
order to obtain the ground state wave function of H
2
you need to apply A to the rst
excited wave function of H
1
. The rst excited state wave function, can be obtained
from eq. (30) and is given by

(1)
1
(x) =
_
2
L
_
1/2
sin
_
2x
L
_
, for 0 x L. (40)
Now using eq. (19)

(2)
0
(x) =
_
E
(1)
1
_

1
2
A
(1)
1
(x), (41)
where
_
E
(1)
1
_

1
2
can be found via eq. (29) and A is given by eq. (5), we get that

(2)
0
(x) =
_
3
2

2
2mL
2
_

2m
d
dx

(1)
1
(x) + W(x)
(1)
1
(x)
_
=
_
3
2

2
2mL
2
_

2m
d
dx
_
2
L
_
1/2
sin
_
2x
L
_
+

2m

L
cot
_
x
L
_
_
2
L
_
1/2
sin
_
2x
L
_
_
.
(42)
In order to solve this equation, we best split it up in to two separate calculations. The
rst part then becomes

2m
d
dx
_
2
L
_
1/2
sin
_
2x
L
_
=

2m
_
2
L
2
L
cos
_
2x
L
_
=

2m
_
2
L
2
L
_
cos
2
_
x
L
_
sin
2
_
x
L
__
, (43)
16
the second term yields

2m

L
_
2
L
_
1/2
cot
_
x
L
_
sin
_
2x
L
_
=

2m
2
L
_
2
L
cos
2
_
x
L
_
, (44)
where we used the trigonometric identities cot() =
cos()
sin()
, cos(2) = cos
2
() sin
2
()
and sin(2) = 2 sin() cos(). Putting the terms back together and simplifying, we
obtain

(2)
0
(x) =
_
3
2

2
2mL
2
_

2m
_
2
L
2
L
_
cos
2
_
x
L
_
sin
2
_
x
L
__
+

2m
2
L
_
2
L
cos
2
_
x
L
_
_
= 2
_
2
3L
sin
2
_
x
L
_
, (45)
where in the nal result a constant was disregarded. The same procedure can be done
for any of the excited states via eq. (19). Hence, the normalised ground and rst
excited state wave functions of H
2
are given by

(2)
0
= 2
_
2
3L
sin
2
_
x
L
_
, for 0 x L, (46)

(2)
1
=
2

3L
sin
_
x
L
_
sin
_
2x
L
_
, for 0 x L. (47)
Similar to the example in section 2.2, using SUSY, we have shown that two rather dif-
ferent potentials in quantum mechanical Hamiltonians have exactly the same energy
spectra except that H
1
has one more bound state. Hence, SUSY has helped us nd
a potential sharing the same solutions with one of the most studied and best under-
stood potentials within quantum mechanics, the innite square well. This becomes
especially apparent in Figure 3. Here we have plotted the dierent potentials and their
corresponding rst few wave functions. For convenience we have chosen = 2m = 1.
17
Figure 3: Left: Innite Square Well potential with width and its ground state and
rst 2 excited wave functions. Right: Innite Square Well partner potential and its
rst 2 eigenfunctions.
2.4 Supersymmetric Inuence on Scattering Relations
For continuous spectra, supersymmetry also relates the reection and transmission
coecients of two partner potentials and thereby gives us valuable information about
the nature of the potentials. This can be achieved via employing eqs. (17)-(19) to
scattering state. First, in order to obtain a scattering process, it is necessary for both
potentials to be nite as x or x or both. For simplication, dene
W(x ) = W

. Then as x it follows that V


1,2
W

. Consider an
incoming plane wave e
ikx
with energy E coming from x . As we know, a
scattering at potential V
1,2
(x) will result in transmitted waves of the form T
1,2
(k)e
ik

x
and reected waves R
1,2
(k)e
ikx
. Therefore, as a little recap, we have

(1,2)
(k, x ) e
ikx
+ R
1,2
e
ikx
(48)

(1,2)
(k, x ) T
1,2
e
ik

x
, (49)
18
where the wave numbers k and k are dened as k =
_
E W
2

and k

=
_
E W
2
+
,
since we are dealing with scattering states. For convenience, 2m = = 1 was chosen.
As in the discrete case, SUSY connects the continuous wave functions of the partner
Hamiltonians by the relations stated in eqs. (18)-(20). Employing these relations, we
obtain
e
ikx
+ R
1
e
ikx
= C[(ik + W

)e
ikx
+ (ik + W
+
)e
ikx
R
2
], (50)
T
1
e
ik

x
= C[(ik

+ W
+
)e
ik

x
T
2
], (51)
where C is the normalisation constant. After equating the terms with the same expo-
nents and cancelling the normalisation constant C, we get the coecient relation
R
1
(k) =
_
W

+ ik
W

ik
_
R
2
(k), (52)
T
1
(k) =
_
W
+
ik

ik
_
T
2
(k). (53)
From these two equations, some interesting aspects can be observed. It is obvious that
|R
1
|
2
= |R
2
|
2
and |T
1
|
2
= |T
2
|
2
. This shows that all partner potentials have the same
transmission and reection probabilities. Furthermore, we see that for W

= W
+
we
have T
1
(k) = T
2
(k) and if W

= 0 then R
1
(k) = R
2
(k). This gives us valuable
information about the reection and transmission coecients of partner potentials
under certain congurations of the underlying superpotential. Now, one can observe
that if one of the potentials at hand is the potential of a free particle and thereby
reectionless, then its partner potential must also be reectionless. This is crucial in
order to understand the reectionless character of some important potentials, such as
potentials of the form V (x) = Asech
2
(x), which play a considerable role in the soliton
solution of the Korteweg-de Vries hierarchy. Since this is an important potential, we
will consider it in our next example.
19
2.4.1 A Reectionless Potential
Assume a superpotential of the form
W(x) = Btanh(x). (54)
Now using eqs. (6) and (9) and setting = 2m = 1, we obtain
V
1
(x) = B
2
B(B + 1)sech
2
(x), V
2
(x) = B
2
B(B 1)sech
2
(x). (55)
Having a close look, we can see that for a choice of B = 1, V
2
(x) will become a
constant and corresponds to a free particle with no bound states. Hence, T
2
(k) =
1 and R
2
(k) = 0. Thereby, V
1
(x) must be a reectionless potential following the
previous section. It has then exactly one bound state, namely E
(1)
0
, which is given by
V
1
(x) = 1 2sech
2
(x). Further, according to eq. (53) the transmission coecient of
V
1
is given by
T
1
(k, B = 1) =
_
W
+
ik

ik
_
T
2
(k), (56)
where W

= lim
x
W(x) = 1 and k = k

since W
2

= W
2
+
hence
T
1
(k, B = 1) =
1 ik
1 ik
. (57)
However, if instead we set B=2, we nd that V
1
(x) = 4 6sech
2
(x) and V
2
(x) =
4 2sech
2
(x). From above, one can immediately see that V
2
(x) now must be a
reectionless potential with one bound state since it is of the same form as V
1
(x) was
before. Thereby it follows that the new V
1
(x) where B = 2 must also be reectionless
with one more bound state, according to eqs. (18) - (20). Generalising to any integer
b for B would thereby yield a discrete parameter family of reectionless potentials of
the form
V
1
(x, B = b) = b
2
b(b + 1)sech
2
(x) (58)
20
with b bound states. Its energy eigenvalues and wave functions could be recursively
obtained by using eqs. (8) and (18)-(20). This, in fact, supplies an underlying struc-
ture in the Korteweg-de-Vries soliton solutions. However, for the scope of this paper,
simply note the recursivity used in order to obtain the next solutions. We will now use
a similar approach to obtain families of partner potentials yielding the same energy
spectra and thereby construct a Hierarchy of Hamiltonians.
2.5 Hierarchy of Hamiltonians
In section 2.1, we have seen that in order to factorise a Hamiltonian, we needed its
ground state energy to be zero. This we can always achieve by adding or subtracting
a constant to the energy ground state. Via this factorisation we obtained a part-
ner Hamiltonian with identical energy spectrum, expect the ground state. Thus the
new Hamiltonian had n 1 bound states, assuming the original Hamiltonian had
n. Following the idea developed in the previous example, we now want to obtain a
new partner Hamiltonian to H
2
, which of course is also partner to H
1
. This can be
achieved by, once again, shifting the zero energy ground state of H
2
by a constant to
equal it to zero. Then, applying the principles developed in section 2.1, we can derive
a SUSY partner Hamiltonian H
3
. Applying this process recursively allows us to nd
a chain of Hamiltonians having the same solution structure. Following these obser-
vations, each of these new Hamiltonians has one fewer bound state then the previous
one and hence this process can be repeated as many times as H
1
has bound states.
Therefore, starting with an exactly solvable potential problem for H
1
, one can nd the
solutions, i.e. energy eigenvalues, for the entire hierarchy of Hamiltonians obtained by
repeated refactorisation. Similarly, knowing all the ground energy states of a hierarchy
of Hamiltonians, we can reconstruct the solutions to the original potential problem.
This will become an important factor when combining it with shape invariance as we
will see in section 3. Mathematically, this chain of Hamiltonians can be obtained as
follows.
21
Let E
(1)
0
be the ground state energy of H
1
and let
(1)
0
be its ground state eigen-
function. For simplicity we, once again, set = 2m = 1. Then from section 2.1,
specically eq. (4), it follow that
H
1
(x) = A

1
A
1
+ E
(1)
0
=
d
2
dx
2
+ V
1
(x), (59)
where the Hamiltonian was shifted to obtain a zero energy ground state. Then fol-
lowing eq. (6), V
1
(x) is given by
V
1
(x) = W
2
1
(x) W

1
(x) + E
(1)
0
(60)
where as before
A
1
=
d
dx
+ W
1
(x), A

1
=
d
dx
+ W
1
(x), W
1
(x) =

(1)

0
(x)

(1)
0
(x)
=
dln
(1)
0
dx
, (61)
which follow from eqs. (5) and (7). We rewrote W
1
(x) here, because it will make the
computation easier since we can make use of logarithmic identities. Since we are now
dealing with more than one super partner, attention to the indices must be given. It
follows that
H
2
= A
1
A

1
+ E
(1)
0
=
d
2
dx
2
+ V
2
(x), (62)
where
V
2
(x) = W
2
1
(x) + W

1
(x) + E
(1)
0
= V
1
(x) + 2W

1
= V
1
(x) 2
d
2
dx
2
ln
(1)
0
(63)
yielding
E
2
n
= E
(1)
n+1
,

(2)
n
=
_
E
(1)
n+1
E
(1)
0
_

1
2
A
1

(1)
n+1
.
(64)
22
This is practically identical to what we have done in section 2.1 with some minor
changes to include the indices and ground state energy. Knowing that the ground
state energy of H
2
is E
(2)
0
= E
(1)
1
, we can construct a Hamiltonian H
3
by writing H
2
as follows
H
2
= A
1
A

1
+ E
(1)
0
= A

2
A
2
+ E
(2)
0
= A

2
A
2
+ E
(1)
1
, (65)
with
A
2
=
d
dx
+ W
2
(x), A

2
=
d
dx
+ W
2
(x), W
2
(x) =
dln
(2)
0
dx
. (66)
Then H
3
is given by again reversing the order of factorisation
H
3
= A
2
A

2
+ E
(1)
1
=
d
2
dx
2
+ V
3
(x). (67)
The corresponding potential thus takes the form
V
3
(x) = W
2
2
(x) + W

2
(x) + E
(1)
1
= V
2
(x) + 2W

2
= V
2
(x) 2
d
2
dx
2
ln
(2)
0
(68)
Now, we can insert the expression for V
2
(x) we found earlier, see eq. (63), then
V
3
(x) = V
1
(x) 2
d
2
dx
2
ln
(1)
0
2
d
2
dx
2
ln
(2)
0
= V
1
(x) 2
d
2
dx
2
ln
_

(1)
0

(2)
0
_
, (69)
where we used the logarithmic identity that ln(a) +ln(b) = ln(ab). As before, we can
now draw some relations for the third partner Hamiltonian
E
3
n
= E
(2)
n+1
= E
(1)
n+2
, (70)

(3)
n
=
_
E
(2)
n+1
E
(2)
0
_

1
2
A
2

(2)
n+1
,
=
_
E
(1)
n+2
E
(1)
1
_

1
2
_
E
(1)
n+2
E
(1)
0
_

1
2
A
2
A
1

(1)
n+2
, (71)
23
where eq. (64) was inserted for
(2)
n+1
. This implies that the solution of the third
partner Hamiltonian can be entirely expressed in terms of solutions to the original
Hamiltonian. It further indicates that this is possible for all Hamiltonians in this
chain. Notice explicitly that an original Hamiltonian H
1
, with m 1 bound states
with energies E
(1)
n
and eigenfunctions
(1)
n
for 0 n m, allows us to always create
a hierarchy of m 1 Hamiltonians, whose number of eigenvalues depend on their
respective positions in the hierarchy. This means that the k
th
Hamiltonian H
k
has
exactly the same energy spectrum as the original Hamiltonian H
1
, except the rst
(k-1) states. Hence, knowing that H
1
has m bound states, we can always write (for
k = 2, 3, ..., m)
H
k
= A

k
A
k
+ E
(1)
k1
=
d
2
dx
2
+ V
k
(x) (72)
with
A
k
=
d
dx
+ W
k
(x), A

k
=
d
dx
+ W
k
(x), W
k
(x) =
dln
(k)
0
dx
. (73)
And thereby the previously derived relations
E
k
n
= E
(k1)
n+1
= E
(1)
n+k1
, (74)

(k)
n
=
_
E
(1)
n+k1
E
(1)
k2
_

1
2
....
_
E
(1)
n+k1
E
(1)
0
_

1
2
A
k1
....A
1

(1)
n+k1
, (75)
V
k
(x) = V
1
(x) 2
d
2
dx
2
ln
_

(1)
0
...
(k1)
0
_
. (76)
This gives us the possibility to immediately write down all the energy eigenvalues
and eigenfunctions of the hierarchy of m 1 Hamiltonians. This is an astonishing
result, it enables us to know the solutions of many Hamiltonians almost instantly by
just knowing that they are related via supersymmetry to a Hamiltonian that we know
the solutions of. However, it is important to remember the energy shift undertaken
in order to get the ground state energy of H
1
equal to zero. To obtain the actual
energy spectrum, one needs to add this constant again and all other energies will
shift accordingly. Figure 4 best summarises the relationships we discovered. In the
24
following sections we will explore the SUSY Hamiltonian family of the innite square
well potential.
Figure 4: Energy spectrum of a Hamiltonian hierarchy via SUSY, where m = 4.
2.5.1 Hierachy of the Innite Square Well
With this example we will, starting from a analytically solvable problem like the
innite square well, see how to generate a whole class of new solvable potentials.
As we have already seen in section 2.3 we could on the basis of the normal innite
square well, nd its superpotential, partner potential and partner wave functions.
These were given by eqs. (34) - (46). Since we will be using these to obtain higher
order partner potentials to the innite square well Hamiltonian, we will quickly state
them again. For the entire derivation please see section 2.3.
W(x) =

2m

L
cot
_
x
L
_
, for 0 x L, (77)
V
2
(x) =

2

2
2mL
2
_
2 csc
2
_
x
L
_
1
_
, for 0 x L, (78)

(2)
0
= 2
_
2
3L
sin
2
_
x
L
_
, for 0 x L. (79)
To obtain higher order partner potentials, we will now start viewing V
2
as our initial
potential with the given ground state wave function
(2)
0
and apply the full machinery
of QM SUSY as we did before. Since we have done this in full detail in previous
25
sections and examples, we will now operate more solution driven. Thus, taking the
derivative of the new ground state wave function yields

(2)
0
= 2
_
2
3L
2
L
sin
_
x
L
_
cos
_
x
L
_
, for 0 x L. (80)
and then following eq. (7), we obtain the new superpotential
W
2
(x) =

2m
2
_
2
3L
2
L
sin
_
x
L
_
cos
_
x
L
_
2
_
2
3L
sin
2
_
x
L
_
=

2m
2
L
cot
_
x
L
_
, for 0 x L. (81)
Using eq. (9), the innite square wells second order supersymmetric partner potential
becomes
V
3
(x) =

2

2
2mL
2
_
6 csc
2
_
x
L
_
4
_
, for 0 x L, (82)
Of course this potential is part of a Hamiltonian H
3
and as we have seen in section
2.1, we can calculate its wave functions by applying the operator A to the known wave
functions of H
2
. Hence one can obtain that

(3)
0
sin
3
_
x
L
_
, for 0 x L. (83)
Following the same structure, one can now start with the computed ground state wave
function in order to get W
3
, then V
4
and its ground state as the following step. This
way, it is possible to construct an entire family of potentials. Doing the analysis one
can observe that these turn out to be structured as follows. For a discrete parameter
l = 0, 1, 2, ... we get that
V
(l+1)
(x) =

2

2
2mL
2
_
l(l + 1) csc
2
_
x
L
_
l
2
_
, for 0 x L, (84)
26
The energies are then found to be
E
(l+1)
n
n(n + 2l + 2)
3
, n = 0, 1, 2, ..., (85)
with ground state eigenfunctions of the form

(l+1)
0
sin
(l+1)
_
x
L
_
, for 0 x L. (86)
Now we have obtained a whole class of new solvable potentials by simply using the
QM SUSY tools outlined in section 2.1 on a simple QM potential. Once again, it is
important to stress that we already know their entire energy spectra, since the so-
lutions to the innite square well potential are well-known. The rst three partner
potentials can be seen in Figure 5. One can clearly see that the ground state energy
of each successive partner potential is higher than the previous. Of course, this also
works for other potentials fullling the requirement that SUSY is not broken.
Figure 5: The innite square well and its rst three partner potentials
3
Recall that the energies of E
(1)
n
n(n + 2)
27
2.6 Broken Supersymmetry
In general, a symmetry of a Hamiltonian, or Lagrangian for that matter, can be
spontaneously broken. This occurs if the lowest energy solution to the problem does
not obey the given symmetry [3]. This is identical in the treatment of supersymmetry.
If the lowest energy solution, here the ground state to the given Hamiltonian, does
not respect the conditions we have derived in the previous sections then we say that
supersymmetry is spontaneously broken or just supersymmetry is broken.
In the specic case of SUSY QM this means that the ground state wave function has
to be normalisable. That criteria in return depends on the form of the superpotential
W(x). For instance, let us assume that we have a given superpotential W(x) and
we are to nd the ground state wave function and partner wave functions. Note, in
previous chapters we have observed that given a ground state wave function we were
able to factorise the Hamiltonian and nd SUSY partners. Hence, we are now trying
to operate in converse order. In the case of a given W(x) we can use the tools available
to us from section 2.1 and see that there are two distinct possibilities
A
(1)
0
(x) = 0
(1)
0
(x) = Nexp
_

2m

_
x
W(x

)dx

_
(87)
or
A

(1)
0
(x) = 0
(2)
0
(x) = Nexp
_
+

2m

_
x
W(x

)dx

_
. (88)
Hence, one can see that there are two candidates, the ground state wave function for
H
1
or the ground state wave function for H
2
. However, convention dictates that from
the two partner Hamiltonians H
1
has the lowest ground state energy, which is, as we
know from previous sections, partnerless. Thus, to accomodate this convention, one
has to choose W(x) such that only H
1
(if at all) has a normalisable ground state wave
function. We see that for equation (87) to be normalisable, i.e. to vanish at ,
we must have the exponent to converge to as x . This is ensured when
choosing W such that W(x) is positive/negative for large positive/negative x. For
28
this consider the case, where
W(x) = kx
n
. (89)
Then for any odd n and positive k, we always have a normalisable ground state eigen-
state. This also holds when k is negative, since we can just choose our superpotential
to be W(x) = kx
n
. However, when k is arbitrary and n is even, we cannot nd a
normalised ground state wave function and therefore SUSY is broken. This can be
easily understood when we extract more detailed information from the fact that the
exponent must converge to as x for eq. (87) to hold. This translates to
the conditions that
_
0

W(x

)dx

= and
_

0
W(x

)dx

= . (90)
However, W(x

) is an even polynomial potential and thus symmetric. Thereby, the


previous stated conditions cannot be fullled at the same time and we do not have a
normalisable ground state and SUSY is broken.
In our matrix formulation of section 2.1, the condition for SUSY to be unbroken
translates to
Q|0 >= Q

|0 >= 0|0 >, (91)


where
|0 >
0
(x) =
_
_

(1)
0

(2)
0
_
_
. (92)
Hence, the operators Q and Q

must annihilate the vacuum. The anticommutation


relations dened at the end of section 2.1 now immediately tell us that the ground state
energy must be zero (as it was always the case in previous examples). Furthermore,
with the denitions of equation (22) we see that
(2)
0
must be zero. Hence
29

0
(x) =
_
_

(1)
0
0
_
_
, (93)
with
(1)
0
dened as in equation (87).
We have the immediate result that, if the ground state energy of H
1
is not zero,
then supersymmetry is broken and equations (18) - (20) no longer hold. Hence, the
operators A and A

do not inuence the number of nodes of the partner wave functions


anymore, and pairing between the partner Hamiltonians becomes one-to-one. The
relations obtained are then
E
2
n
= E
(1)
n
, with E
n
> 0 and n = 0, 1, 2..., (94)

(2)
n
=
_
E
(1)
n
_

1
2
A
(1)
n
, (95)

(1)
n
=
_
E
(2)
n
_

1
2
A

(2)
n
. (96)
Another way of determining whether SUSY is broken, was proposed by Edward
Witten [21]. This is also the commonly used way in supersymmetric eld theory. He
dened the so-called Witten index by
= Tr(1)
N
F
, (97)
where N
F
is the fermion number and the trace is taken, in our case, over all the bound
and continuum states of the super-Hamiltonian. The fermion number represents the
number of fermions in that state and can thus be either 0 or 1 considering Paulis
exclusion principle. When the fermion number has value 0 we talk about a bosonic
state and when its value is 1 it is referred to as a fermionic state. We will give a bit
more insight into this in the following example. It is convention to label H
1
to have
fermion number N
F
= 0.
30
In a case that SUSY is broken, we have seen that there are no zero energy eigen-
functions and by equations (94) - (96) all states come in pairs with the same energy.
Thus all the bosonic states (i.e. N
F
= 0) and the fermionic states (N
F
= 1) are paired
and their contributions to cancel leaving us with = 0.
Assuming supersymmetry is not broken, we have an additional bosonic state, since
H
1
has one more bound state than H
2
, and obtain that = 1. Hence it follows that
= 1 supersymmetry is unbroken, (98)
and we can utilise the tools obtained in previous sections.
2.7 Supersymmetric Harmonic oscillator and the SUSY Al-
gebra Relations
In this section, we will address the quantum mechanical harmonic oscillator in order
to extend it to supersymmetry. Therefore, it might be useful to recall the aspects
of the harmonic oscillator with special attention to the treatment via the operator
method, which can be obtained in detail from [9]. Some of the results will be restated
in this section. We will further rewrite the Hamiltonian using scaled variables. The
information and inspiration for this section was given by [3], [19] as well as [7]. For the
most detailed derivation, see [4]. In this example, we want to specically illustrate the
use of bosonic and fermionic creation and annihilation operators and the relations they
entail. For the treatment of the topic we will operate in dirac notation. The parameters
n
b
and n
f
will represent the boson and fermion occupation numbers, respectively. The
Hamiltonian for the harmonic oscillator is given by
H =
P
2
2m
+
1
2
m
2
q
2
. (99)
To rescale the Hamiltonian in terms of the dimensionless coordinates and moments,
x and p, respectively, we choose
31
H = H, q =
_

2m
x, P =

2mp. (100)
Then we can identify that
H =
x
2
4
+ p
2
, (101)
where x and p full the commutation relation
[x, p] = i. (102)
We can now further nd the rescaled ladder operators for the harmonics oscillator,
which are the mathematically identical to the bosonic creation and annihilation oper-
ators [7]. This is also represented in the commutation relations, since the commutator
of the ladder operators for the same boson state equal 1, while all other commutators
vanish [7]. The operators are given by
a =
x
2
+ ip, (103)
a

=
x
2
ip (104)
with the relation
[a, a

] = 1. (105)
Using the relation that [AB,C]=A[B,C]+[A,C]B, we observe that
[a

a, a] = [a

, a]a = a, [a

a, a

] + a

[a, a

] = a

. (106)
Let us dene N = a

a, then we can see that


H = N +
1
2
, (107)
32
as expected. Recall that the ground state is dened by the condition
a|0 = 0 (108)
and that the nth excited state is given by applying the raising operator n times, hence
|n =
a
n

n!
|0. (109)
Now, to consider the supersymmetric harmonic oscillator, we rewrite the previously
dened supercharges Q and Q

as a product of the bosonic operators a and a

with the
respective fermionic operators b and b

[3]. Explicitly, consider Q = ab

and Q

= a

b.
The matrix notation of the fermionic creation and annihilation operators are dened
as
b =
+
=
_
_
0 1
0 0
_
_
and b

=
_
_
0 0
1 0
_
_
. (110)
The operators b and b

respect the previously outlined algebra, namely


{b

, b} = 1, {b

, b

} = {b, b} = 0 (111)
as well as the commutation relation
[b, b

] =
3
=
_
_
1 0
0 1
_
_
. (112)
Now recall from section 2.1 that the super-Hamiltonian is also dened by the under-
lying algebra relation
{Q, Q

} = H, (113)
33
which is equivalent to stating that
H = QQ

+ Q

Q. (114)
Now, we can insert the previously found dened relations Q = ab

and Q

= a

b to
nd that the SUSY Hamiltonian of the harmonic oscillator can be written as
H =
_

d
2
dx
2
+
x
2
4
_
I
1
2
[b, b

], (115)
where I is the two-dimensional identity matrix and the last term gets rid of the
zero point energy. We can represent the state of the system either in normal matrix
notation by
=
_
_
n
b
n
f
_
_
(116)
or in dirac notation by |n
b
, n
f
. As stated earlier, n
b
and n
f
represent the boson and
fermion occupation number, respectively. Due to the anticommutation relations the
fermionic creation and annihilation operators obey, the fermion number n
f
can either
be 0 or 1. Again, states with n
f
= 0 are generally known as bosonic states, whereas
states with n
f
= 1 are fermionic states. As previously discussed, it is convention to
dene the ground state of H
1
to have n
f
= 0. This is the case, since you will nd
that bosons generally tend to congregate in the ground state of a system, when you
examine the Bose and Fermi statistics, because there are no barriers to oppose this
accumulation.
When we note that the fermionic operators are represented by Pauli spin matrices
34
(see eq. (110) and (112)), we can nd n
f
= b

b to be
n
f
= b

b =

+
=
1
2
(
1
i
2
)
1
2
(
1
+ i
2
)
=
1
2
(1
3
)
=
1
2
(1 [b, b

]). (117)
Therefore, we can evaluate the action of the operators a, a

, b and b

on the state
|n
b
, n
f
. They are then given by
a|n
b
, n
f
= |n
b
1, n
f
, b|n
b
, n
f
= |n
b
, n
f
1, (118)
a

|n
b
, n
f
= |n
b
+ 1, n
f
, b

|n
b
, n
f
= |n
b
, n
f
+ 1. (119)
We can now observe the properties of the supercharge Q

= ab

. By applying Q

we
can change a boson into a fermion without altering the energy level of the state as
it can be seen by eq. (118) and (119). This represents the boson-fermion coupling
or degeneracy that is characteristic for supersymmetric theories. However, since this
is the special example of the SUSY harmonic oscillator, we still have to discuss the
general SUSY case.
When concerning oneself with the general case of SUSY QM, we only have to re-
place the bosonic operators a and a

in the denition of the supercharges Q and Q

by
the operators A and A

that we dened in section 2.1.. The supercharges then relate


the eigenstates of the super-Hamiltonians H
1
and H
2
via the earlier dened relations.
The Hamiltonians then have fermion number 0 and 1, respectively.
35
For additional information about the topics treated in chapter 2 one may consult
[1], [3], [4], [6], [7], [14], [17], [19], [20] and [21].
36
3 Shape Invariant and Solvable Potentials
Within non-relativistic quantum mechanics there are relatively few fully analytically
solvable potentials for which the energy eigenvalues and corresponding wave functions
are explicitely known. It was long unknown what made these potentials solvable
whereas others were not. Some of the famous examples include the Coloumb, Morse,
Pschl-Teller, Eckart and harmonic oscillator potenials. However, in the mid-1980s
Gendensthein illustrated that all these potentials fall into a category labelled as shape
invariant [8]. In this chapter, we will assess what it means for a potential to be
categorised as shape invariant and illustrate with help of the tools developed in chapter
2 that we can use this to generalise the operator method known from the QM harmonic
oscillator to easily obtain all the bound state energy eigenvalues and eigenfunctions
of all common analytically solvable potentials.
In the following derivations we assume SUSY to be unbroken and we will set
= 2m = 1. Let us known look what exactly it means for a potential to be shape
invariant. Assume we have two partner potentials V
1
and V
2
related by supersymmetry
through eqs. (6) and (9). If they are similar in shape and only dier by parameters
then they are categorised to be shape invariant. This is reected by the following
condition
V
2
(x; a
1
) = V
1
(x; a
2
) + R(a
1
), (120)
where a
1
is a set of parameters, a
2
is a function of a
1
, i.e. a
2
= f(a
1
) and R(a
1
) is a
remainder. Also note, the remainder R(a
1
) is independent of x. Any scaling will also
not inuence the condition, since possible dierences in scaling can be included in the
relation f(a
1
) = a
2
. With help of equation (120) and the hierarchy of Hamiltonians
derived earlier, one can quickly derive the energy eigenvalues and eigenfunctions of
any shape invariant potential as long as SUSY is unbroken. It might be good to point
out that the given condition is an equation, in which V
2
is a function with parameter
set a
1
and V
1
contains the new parameter set a
2
. It is important to keep track of the
37
indices when dealing with SIPs.
As in most cases, we start with two partner Hamiltonians related by unbroken
SUSY. Then from section 2.1 we know that
E
(1)
0
(a
0
) = 0, (121)

(1)
0
(x; a
0
) = N exp
(

x
W
1
(y;a
0
)dy)
. (122)
With our knowledge about the hierarchy of Hamiltonians and eq. (120), we can quickly
obtain the rst Hamiltonian H
1
and its partner H
2
H
1
=
d
2
dx
2
+ V
1
(x; a
1
) (123)
and
H
2
=
d
2
dx
2
+ V
2
(x; a
1
)
=
d
2
dx
2
+ V
1
(x; a
2
) + R(a
1
), (124)
where eq. (120) was used. We can continue in similar fashion by shifting H
2
by the
the constant R(a
1
), in order to obtain a zero energy ground state for H
2
, which is
the condition for the refactorisation of the Hamiltonian (see section 2.5). By then
applying the shape invariance condition again to obtain H
3
H
3
=
d
2
dx
2
+ V
2
(x; a
2
)
=
d
2
dx
2
+ V
1
(x; a
3
) + R(a
2
), (125)
where eq. (120) was used again and a
3
= f(a
2
) = f(f(a
1
)). Now, since we had to
shift H
2
by R(a
1
) to obtain this result, we need to remember to shift H
3
back by the
38
same constant, giving us
H
3
=
d
2
dx
2
+ V
1
(x; a
3
) + R(a
2
) + R(a
1
). (126)
Collecting all three Hamiltonians,
H
1
=
d
2
dx
2
+ V
1
(x; a
1
),
H
2
=
d
2
dx
2
+ V
1
(x; a
2
) + R(a
1
),
H
3
=
d
2
dx
2
+ V
1
(x; a
3
) + R(a
2
) + R(a
1
), (127)
one can clearly see a trend in the ranking of shape invariant Hamiltonians. Thus, it
is easy to generalise this rule to the sth partner Hamiltonian
H
s
=
d
2
dx
2
+ V
1
(x; a
s
) +
s1

i=1
R(a
i
). (128)
We will now use the derived Hamiltonians to quickly obtain all energy eigenvalues of
H
1
and thereby of all its partner Hamiltonians. Let us take a closer look at H
2
and
simply apply the Schr odinger equation, then
H
2

(1)
0
(x; a
2
) =
_

d
2
dx
2
+ V
1
(x; a
2
)
_

(1)
0
(x; a
2
) + R(a
1
)
(1)
0
(x; a
2
),
= R(a
1
)
(1)
0
(x; a
2
), (129)
since
_

d
2
dx
2
+ V
1
(x; a
2
)
_

(1)
0
(x; a
2
) = H
1
(x; a
2
)
(1)
0
(x; a
2
) = 0, (130)
because by assumption that SUSY is unbroken, the ground state energy of H
1
is zero.
Thus eq. (129) yields
E
(2)
0
= R(a
1
). (131)
39
Hence, it was possible for us to express the ground state energy of H
2
and thereby the
rst excited state of H
1
only in terms of the remainder R(a
1
). This is a remarkable
result. It lets us know the ground energy of a potential simply by comparing its shape
with that of its partner potential. In a similar way, one can nd
H
s

(1)
0
(x; a
s
) =
_

d
2
dx
2
+ V
1
(x; a
s
)
_

(1)
0
(x; a
s
) +
s1

i=1
R(a
i
)
(1)
0
(x; a
2
),
=
s1

i=1
R(a
i
)
(1)
0
(x; a
s
), (132)
E
(s)
0
=
s1

i=1
R(a
i
). (133)
Eq. (133) yields the ground state energy to the sth Hamiltonian or in other words
the (s 1)th energy level to the rst Hamiltonian. Therefore, it follows that the
entire spectrum of energy eigenvalues of H
1
and thereby of course for all its partner
Hamiltonians, since H
k
has the same energy spectrum as H
1
with exception of the
rst k 1 energy levels, is given by
E
(1)
n
=
n

i=1
R(a
i
) with E
(1)
0
= 0. (134)
Concerning the corresponding bound state energy eigenfunctions
(1)
n
(x; a
1
) of shape
invariant potentials, it can be shown that these, similar to the QM 1-D harmonic
oscillator, can be easily obtained from its ground state wave function
(1)
0
(x; a
1
), which
in turn is found via eq. (8). This is achieved via the previously dened operators
A and A

, because they link up the eigenfunctions with the same energy for the
supersymmetric partner Hamiltonians H
1
and H
2
. Let us now recall equation (20) and
identify from equation (132) that its ground state wave function is given by
(1)
0
(x; a
s
).
Then we can reconstruct the nth state non-normalised wave function
(1)
n
(x; a
1
) for
the original Hamiltonian H
1
by backtracking from H
s
to H
s1
to H
s2
to ... to H
2
40
and nally to H
1
. Hence,

(1)
n
(x; a
1
) A

(x; a
1
)A

(x; a
2
)...A

(x; a
n
)
(1)
0
(x; a
n+1
). (135)
If one is familiar with the operator method of the 1-D harmonic oscillator, one can
clearly see that this is a generalisation of the method for all potentials fullling the
shape invariance condition stated in eq. (120). As a matter of fact, we can now
understand why in the harmonic oscillator a

and a are labelled as energy raising and


lowering operators, respectively. Namely, the oscillator potential is shape invariant
and its function f that transforms a
1
to a
2
is simply equal to 1, hence a
1
= a
2
=
a
3
= ... = a
n
= . This will be made clear in the next example when we consider the
shifted oscillator. Furthermore, the ground state eigenfunctions are all the same for
the Hamiltonian class of H
s
. Then, equation (135) is reduced to

n
(x) =
(a

)
n

n!

0
(x), (136)
because A

becomes proportional to a

. A more detailed analysis of this can be found


in [3] and [14].
Hence, when dealing with shape invariant potentials, one only needs to nd the
ground state eigenfunction of H
1
via the superpotential, the function f(a
1
) and the
remainder to be able to fully determine all eigenfunctions and eigenvalues of the Hamil-
tonian. Since all the familiar analytically solvable potentials fall into the category of
shape invariance, this provides us with an immensely powerful tool for computation
of their solutions, as we will see in the following section.
Although equation (135) can clearly illustrate one of the fascinating facets of shape
invariant potentials, in practise it is more useful to dene explicit expressions for the
consecutive wave functions. Thus, the following now normalised relation between wave
41
functions of the rst Hamiltonian H
1
might prove more useful

(1)
n
(x; a
1
) =
_
E
(1)
n1
_
(
1
2
)
A

(x; a
1
)
(1)
n1
(x; a
2
). (137)
The following table shows all known shape invariant potentials, labelled by V

(x; a
0
),
their bound state energy spectra E
()
n
as well as the corresponding superpotential
W(x), parameter values a
0
, a
1
and remainder R(a
1
). This table was obtained from [14],
but can also be found, in this or similar fashion, in [3] and [6]. Further information
about shape invariant potentials and their characteristics as well as categorisation
methods can be found in [2], [18], [3], [6], [13] and [20]. In the following examples we
will illustrate the idea of shape invariant potentials and have a look how knowledge of
shape invariance can help us nd the energy levels of the Hydrogen atom. Inspiration
for the examples was supplied by [3].
42
3.1 Shifted Oscillator
In this section, we will illustrate the idea of shape invariance with help of a simple
example. As usual by now, we will set = 2m = 1. This example concerns the shifted
harmonic oscillator and by starting with the superpotential given in the previous table
we will derive the other aspects, such as ground state wave function, partner potentials
and energy states.
From the table we know that the superpotential for the shifted harmonic oscillator
is given by
W(x) =
1
2
x b. (138)
The using eq. (6) and (9) we can nd the two partner potentials. For that, note that
W

(x) =
1
2
. (139)
Then
V
1
(x, ) = (
1
2
x b)
2

1
2
=
1
4

2
x
2
bx + b
2

1
2

=
1
4

2
(x
2b

)
2

1
2
(140)
and
V
2
(x, ) = (
1
2
x b)
2
+
1
2

=
1
4

2
(x
2b

)
2
+
1
2
. (141)
43
Figure 6: All known shape invariant potentials and their characteristics.
44
45
The corresponding Hamiltonians are then given by
H
1
=
d
2
dx
2
+ V
1
(x, ), (142)
H
2
=
d
2
dx
2
+ V
2
(x, ). (143)
We should now check whether SUSY is unbroken, so that we can make use of the
tools derived in the previous section
4
.For that we need to check that the ground state
wave function is normalisable. This we will do with help of eq. (87). Recall that

(1)
0
(x) = Nexp
__
x
W(x

)dx

_
, (144)
where N is the normalisation constant and
_
x
W(x

)dx =
_
x
1
2
x

bdx

=
_
x
2
4
bx

_
x
0
=
x
2
4
bx. (145)
Thus

(1)
0
(x) = Nexp
_

2m

_
x
W(x

)dx

_
= Ne

x
2
4
bx

. (146)
To nd N we need to normalise, hence
4
Of course, here we already know that SUSY is unbroken, since the superpotential was given in
the table of analytically solvable potentials, however it is still an important part of the exercise and
will let us nd the ground state eigenfunction of the shifted harmonic oscillator.
46
1 = |N|
2
_

x
2
2
2bx
dx
= |N|
2
_
2

8b
2

, (147)
where we made use of the Gaussian integral. Then
N =
_
2

_1
4
e

4b
2

(148)
and

(1)
0
(x) =
_
2

_1
4
e

4b
2

x
2
4
bx

. (149)
Hence, the ground state wave function is normalisable and SUSY is unbroken. There-
fore, we can use a supersymmetry approach as a treatment method for this problem.
Thus, we will now use the shape invariance condition to easily nd the energy levels.
Recall that the SIP condition was given by
V
2
(x; a
1
) = V
1
(x; a
2
) + R(a
1
). (150)
Inserting the two partner potentials that we found before (see eq. (140) and (141))
and noting that is our parameter of interest (i.e. a
1
= ), since x is our variable
and b is just the constant by which the harmonic oscillator was shifted, we obtain
1
4

2
(x
2b

)
2
+
1
2
=
1
4

2
2
(x
2b

2
)
2

1
2

2
+ R(). (151)
Now, considering that we are interested in the relation f() =
2
, it is enough,
according to the SIP condition, for us to compare the terms with an x dependence.
Then it is quite easy to see that
2
= f() = . Inserting this into eq. (151) and
47
solving for the remainder R() yields
1
4

2
(x
2b

)
2
+
1
2
=
1
4

2
(x
2b

)
2

1
2
+ R()
R() = . (152)
From section 3, we know that the remainder is equivalent to the ground state energy of
the partner potential and thereby the rst exited state energy of the original potential
E
(1)
1
= E
(2)
0
= R() = . (153)
Using our previously obtained knowledge that f() = , we can derive that all pa-
rameters in the Hamiltonian chain will be the same and are given by a
1
= a
2
= ... =
a
n
= , as we had already stated earlier. Furthermore, from the previous discussion of
shape invariance we know that all ground state energy levels of the Hamiltonian chain,
and thereby all excited energy levels of the original Hamiltonian, can be expressed as
a sum of the remainders. Hence
E
(1)
n
= E
(n)
0
=
n

i=1
R(
i
) =
n

i=1
R() = n. (154)
Thereby, we have now already derived the entire energy spectrum of the shifted har-
monic oscillator by making use of the simple SIP condition. We can now also nd
all the excited eigenfunctions. Since a
1
= a
2
= ... = a
n
= all ground state eigen-
functions of the Hamiltonian class are the same and, as discussed earlier, eq. (135) is
reduced to

n
(x) =
(a

)
n

n!

0
(x). (155)
Let us quickly summarise the results of the shifted harmonic oscillator found in this
section with help of shape invariance:
48
V
1
(x, ) =
1
4

2
(x
2b

)
2

1
2
, (156)
V
2
(x, ) =
1
4

2
(x
2b

)
2
+
1
2
, (157)
a
1
= a
2
= ... = a
n
= , (158)

(1)
0
(x) =
_
2

_1
4
e

4b
2

x
2
4
bx

, (159)
and
E
(1)
n
= n, (160)
which are exactly the values for the dierent parameters that can be found in
the previous table. Thus we have illustrated how the dierent values in the table of
analytically solvable potentials were obtained at the example of the shifted oscillator.
A similar treatment for the other potentials is of course also possible. In the following
example, we will consider the case, where only a ground state wave function is given
and we will try to nd its potential and energy solutions. Via this approach, we will
illustrate the advantage of using known analytically solvable shape invariant potentials
versus trying to full the SIP condition oneself.
3.2 Example of Shape Invariance of a Given Wave Function
In this example, we will start only with a given wave function and no more information
about the problem. We are going to, thereby, try to assess the resulting potentials
via the shape invariance approach and show the distinct advantage of building on the
foundations of known solvable potentials. For this, consider a wave function of the
form
49
Figure 7: Plot of
(1)
0
for r
0
= A = 1.

(1)
0
= Ar
3
e

r
2
r
2
0
5
for 0 r . (161)
Note that the given wave function is normalisable and thus supersymmetry holds. For
illustration, we have plotted the wave function in Fig. 7. Again, we set = 2m = 1 for
convenience. We will start by nding the wave functions corresponding superpotential,
for this we are going to make use of eq. (7) and therefore have to nd

(1)
0
, so

(1)
0
= A
_
3r
2
e

r
2
r
2
0
r
3
2r
r
2
0
e

r
2
r
2
0
_
. (162)
Then by eq. (7), the superpotential is given by
5
Some might already recognise this wave function, but for the sake of this section, let us assume
that is not the case.
50
Figure 8: Plot of the superpotential to
(1)
0
for r
0
= 1.
W(x) =
3r
2
e

r
2
r
2
0
r
3 2r
r
2
0
e

r
2
r
2
0
r
3
e

r
2
r
2
0
=
3
2r
2
r
2
0
r
=
2r
r
2
0

3
r
(163)
and plotted in Fig. 8. In order to use the shape invariance approach, we rst have to
nd the two partner potentials V
1
and V
2
. For this we will, as usual, use eq. (6) and
(9) for which are required to nd W

(x), which turns out to be


W

(x) =
2
r
2
0
+
3
r
2
. (164)
51
Thus
V
1
(x) = W
2
W

(x) =
_
2r
r
2
0

3
r
_
2

2
r
2
0

3
r
2
=
4r
2
r
4
0

12r
r
2
0
r
+
9
r
2

2
r
2
0

3
r
2
=
4r
2
r
4
0
+
6
r
2

14
r
2
0
(165)
and
V
2
(x) = W
2
+ W

(x) =
_
2r
r
2
0

3
r
_
2
+
2
r
2
0
+
3
r
2
=
4r
2
r
4
0
+
12
r
2

10
r
2
0
. (166)
Just by looking at the plot of V
1
and V
2
in Fig. 9, one can already suspect them to be
shape invariant. As expected, the ground state energy of V
2
is higher than that of V
1
.
We will now try to show that these are shape invariant, nd the coecients a
1
and
a
2
and use the remainder to nd the energy levels. For this we will equate V
1
and V
2
according to the SIP condition. This yields
4r
2
r
4
0
+
12
r
2

10
r
2
0
=
4r
2
r
4
0
+
6
r
2

14
r
2
0
+ R(a
1
),
12
r
2

10
r
2
0
=
6
r
2

14
r
2
0
+ R(a
1
). (167)
Now, in comparison to the other example treated before, we run into a major problem
here, since no obvious parameter is available to take as our a
1
. This is because, we
started with a given ground state wave function
(1)
0
(r, a
1
) where the parameter a
1
was already replaced by its numerical value. Considering that, we then also cannot
derive the relation f(a
1
) = a
2
, because there are extremely many possibilities to match
52
Figure 9: Plot of V
1
and V
2
as derived from
(1)
0
for r
0
= 1, where V
2
is the higher of
the two functions.
the terms, when only considering these two partner potentials
6
. To stress this point
we will show an example of a possible choice for a
1
and the relation f(a
1
) = a
2
.
Once again, the following part is only to illustrate the diculty in nding the correct
parametrisation on the basis of two partner potentials. As we will see later, there is
a much easier way to determine its parametrisation. Looking at the r-dependent part
of equation,
12
r
2
=
6
r
2
, (168)
a good guess seems to be that the parameters are related by f(a
1
) = a
1
+ 1. Then
one could nd the integer of multiplication by solving the following simple system
a
1
c
r
2
=
6
r
2
and (169)
(a
1
+ 1)c
r
2
=
12
r
2
. (170)
6
This would still be the case for 3 three partner potentials. To nd the denite relationship
f(a
1
) = a
2
and the value for a
1
, considerably more partner potentials would need to be found.
53
This provides the simple solution that a
1
= 1 and c = 6. With this we would now
have to make an assumption about the structure of the term that is independent of
r, namely
10
r
2
0
=
14
r
2
0
+ R(a
1
). (171)
For example we could chose it to be independent of a
1
. Then we can identify the
remainder R(a
1
) to be
R(a
1
) =
4
r
2
0
. (172)
Since in the case of our chosen structure to match the terms we obtained that a
1
= 1,
we can now say that
R(a
i
) =
4a
i
r
2
0
. (173)
In order to check whether this was the correct parametrisation, we now have to cal-
culate the next partner potential via the SIP condition with the identied parameters
and relations. Therefore, according to section 3, we have to shift V
2
(r, a
1
) by the
constant, since the constant represent its ground state energy. Thus
V
2
(r, a
1
) = V
1
(r, a
2
) + R(a
1
)
=
4r
2
r
4
0
+
6a
2
r
2

14
r
2
0
+ R(a
1
)
=
4r
2
r
4
0
+
12
r
2

14
r
2
0
+ R(a
1
), (174)
54
where we used the SIP condition and the fact that a
2
= a
1
+1 = 2. We now subtract
the remainder (i.e. the ground state energy of V
2
)
V
2
(r, a
1
) =
4r
2
r
4
0
+
12
r
2

14
r
2
0
= V
1
(r, a
2
). (175)
Then V
3
(r) is given by
V
3
(r) = V
2
(r, a
2
)
= V
1
(r, a
3
) + R(a
2
)
=
18
r
2

14
r
2
0
+
12
r
2
0
, (176)
where we again used the SIP condition and that a
3
= a
2
+1 = 3. Since we subtracted
the remainder R(a
2
) before, we now have to add it again to obtain the total third
partner potential
V
3
(r) =
18
r
2

14
r
2
0
+
16
r
2
0
. (177)
Unfortunately, that still does not tell us whether our previous assumptions are cor-
rect. For that we would have to compare V
3
with the third partner potential obtained
through the method of the hierarchy of Hamiltonians (see section 2.5). For this we
would have to rst nd
(2)
0
and its derivative. Then nd W
2
and apply eq. (9).
As you can see this is a very lengthy procedure to see whether we made the correct
assumptions when matching the rst two partner potentials via the shape invariance
condition and chances that we choose the correct parametrisation are very slim due
to the vast amount of possibilities. Another way of trying to determine the correct
parametrisation for the SIP condition would have been to rst nd more SUSY part-
55
ner potentials via the procedure outlined just before and then compare their shapes.
However, this is still very cumbersome and leaves possibility for failure when not con-
sidering enough potentials.
Instead of all these troublesome calculations, there is a much easier way of de-
termining the parametrisation by simply making use of the knowledge of analytically
solvable potentials. For this we can consult the table given in section 3. Already just
knowing our previously derived superpotential
W(x) =
2r
r
2
0

3
r
(178)
of the given wave function, we can almost immediately match it to one of the super-
potentials of the table, namely the 3-dimensional oscillator. Then with the simple
substitution that =
4
r
2
0
and recognizing that l = 2, we can instantly check that our
derived potentials given by eq. (165) and (166) are indeed correct. Furthermore, it is
possible for us to just read the energy levels for the given wave function of the table
without having to do any extra computations, namely
E
(1)
n
= 2n. (179)
In Fig. 10 we have plotted E
(1)
n
with on the one side our choice of parameters and on
the other side the actual energy levels for n = 1, 2, 3. As one can see, our choice of
parameters turns out not to yield the correct result, although surprisingly it did give
the exact energy for the second energy level, which is pure coincidence however.
The knowledge about the existing shape invariant potentials and their categorisa-
tion enabled us in this section to circumvent extremely cumbersome calculations by
making use of the previously supplied table. This proved a simple way to nd the
solutions to a, at rst, rather dicult problem. The aim of this section was to under-
line the importance of knowledge about the analytically solvable quantum mechanical
56
potentials and illustrate the use of the earlier presented table. The following example,
concerning the Hulthen potential, will further stress these points.
Figure 10: Left: The rst 3 energy levels derived with our determined parameters and
relations. Right: The actual rst 3 energy levels of the problem. Both assume r
0
= 1.
57
3.3 Hulthen Potential
In this example, we are going to discuss the case when only the potential is known. For
this we consider the Hulthen potential and how we can quickly solve it via another
shape invariant potential. The Hulthen potential is widely used in many areas of
physics, such as atomic physics, solid state physics, nuclear physics and chemical
physics. It is given by
V (r) = V
0
e

1 e

, (180)
where is commonly a screening parameter. The potential is visualised in Fig. 11.
V
0
is often replaced by
Z

, where the constant Z, in the case of atomic physics, is


identied with the atomic number. The potential is a short-range potential that is
always attractive. For small values for r it behaves like a Coulomb potential and
diverges like
V
0
r
. For large r it goes exponentially to 0. The Schr odinger equation
for problems of this form can be solved in a closed form for S-waves [11]. Multiple
methods have been found to approximate the solutions for Hulthen potentials when
the azimuthal number l = 0. Thereby, it represents a rather dicult potential to
solve. However, we can rewrite the potential to bring it into a dierent form, namely
V (r) = V
0
e

r
2
e
r
2
e

r
2
. (181)
Now we can make use of hyperbolic trigonometric identities and further rewrite it as
V (r) =
V
0
2
_
1 coth
r
2
_
. (182)
This is now recognisable as a special case of the Eckart potential in our table of
analytically solvable potentials. The energy eigenvalues are then given by
E
(1)
n
=
(V
0

2
(n + 1)
2
)
2
4
2
(n + 1)
2
, n = 0, 1, ..., n
max
. (183)
58
Figure 11: The Hulthen potential for a choice of parameters V
0
= 1 and = 5.
Note that the numerator gets smaller with increasing n, therefore n
max
is limited by
being the largest possible integer for that V
0

2
(n + 1)
2
> 0. Thereby, we can once
again see the advantage knowledge over analytically solvable potentials can give us. Of
course we can also now nd the Hulthens partner potential, superpotential and wave
functions via the shape invariance approach with help of the list of solvable potentials
from section 3. However, since we have done this in detail in previous examples and
will do so in the next, we will not do so now.
3.4 The Hydrogen Atom
The hydrogen atom is an often discussed example within quantum mechanics and
illustrates many properties of more complex atoms. Although it has been analysed
uncountably often, there are still some unsolved problems it posses. In this example,
we will only conne ourselves to the basics of the hydrogen atom and try to nd its
energy spectrum via the machinery of supersymmetry and shape invariance as derived
in the previous section. For the more standard QM approach to solve for the energy
spectrum of the hydrogen atom, please consult chapter 4.2 of [9]. In this approach,
we will only treat the radial equation with l = 0, where l is the azimuthal quantum
59
Figure 12: The radial Coulomb potential.
number.
The radial Schrodinger equation is given by

2
2m
d
2
u
dr
2
+
_
V (r) +

2
2m
l(l + 1)
r
2
_
u(r) = Eu(r) (184)
with V (r) being the Coulomb potential
V (r) =
e
2
4
0
1
r
. (185)
Now, in order to make use of supersymmetry we have to remember to shift the ground
state energy to zero. Hence the shifted eective potential can be found to be
V
eff
(r) =
e
2
4
0
1
r
+
_

2
l(l + 1)
2m
_
1
r
2
E
0
. (186)
To fully utilise the tools of supersymmetry, we rst need to nd the partner potential
of V
eff
. In order to do so we need to identify its superpotential. Recall the Riccati
60
equation from section 2.1, namely
V
eff
(r) = W(r)
2

2m
W

(r). (187)
To nd W(r) we need to solve this rst order dierential equation. Since we know
the structure of V
eff
we can make an educated guess for the Ansatz to be of the form
W(r) = B
C
r
. (188)
Inserting this into the Riccati equation, we obtain
V
eff
(r) = B
2

2BC
r
+
_
C
2

2m
C
_
1
r
2
. (189)
Straight away one can identify that E
0
= B
2
since B
2
is the only term independent
of r. When we compare the V
eff
obtained via our Ansatz with the actual V
eff
, we
can further equal the obtained coecients to nd that
2BC =
e
2
4
0
, (190)
C
2

2m
C =

2
l(l + 1)
2m
. (191)
Now it is easy to see that solving eq. (191) for C and using this result to nd B from
eq. (190), we can conclude that
C =

2m
(l + 1), (192)
B =

2m

e
2
8
0
(l + 1)
. (193)
Since we have now found B, we can straight away calculate the ground state energy
61
of the Hydrogen atom, where l = 0, via the identied relation of E
0
= B
2
. Thus
E
0
= B
2
=
_

2m

e
2
8
0
_
2
=
2m

2
e
4
64
2

2
0
= 2.18 10
18
J
13.6eV. (194)
Surprisingly, this rather short derivation has already yielded the known value for the
Hydrogen atoms energy ground state illustrating the capacity of a SUSY approach
in QM. The superpotential is now found by plugging the found coecients B and C
into our Ansatz. This yields
W(r) =

2m

e
2
8
0

1
r

2m
(l + 1), (195)
which supplies us with the partner potential via eq. (9). Hence,
V
2
(r) =
e
2
4
0
1
r
+

2
(l + 1)(l + 2)
2m
1
r
2
+
e
4
m
32
2

2
0
(l + 1)
2
(196)
Recalling eq. (120) and comparing V
2
with its partner potential V
eff
V
eff
(r) =
e
2
4
0
1
r
+
_

2
l
2
(l
2
+ 1)
2m
_
1
r
2
+
e
4
m
32
2

2
0
(l
2
+ 1)
2
. (197)
one can readily identify the relation between the parameters a
2
= f(a
1
) to be f(l) =
l + 1, i.e. l
2
= l + 1. With help of this relation and the shape invariance condition,
we can now obtain the remainder R(l) that will give us the energy dierence between
the ground states of these two potentials. We do this by equating V
eff
(r; f(l)) +R(l)
with V
2
(r; l) and solve for R(l). All terms with an r dependence automatically cancel,
62
when we use that f(l) = l + 1 and thus we are left with
e
4
m
32
2

2
0
(l + 1)
2
=
e
4
m
32
2

2
0
(l + 2)
2
+ R(l),
e
4
m
32
2

2
0
(l + 1)
2

e
4
m
32
2

2
0
(l + 2)
2
= R(l),
e
4
m((l + 2)
2
(l + 1)
2
)
32
2

2
0
(l + 2)
2
(l + 1)
2
= R(l),
R(l) =
e
4
m(2l + 3)
32
2

2
0
(l + 2)
2
(l + 1)
2
. (198)
From previous sections we know that the ground state energy of the partner potential
is equal to the rst exited state of the original potential. Furthermore, in section 3 we
have learned that R(l) represents the energy dierence between the two ground states
of the partner potentials. Hence, R(l) also represents the energy dierence between
the ground state of the V
eff
and its rst exited state. Now, since we had to shift the
ground state energy at the beginning of this example we now have to shift it back to
obtain the full solution for the rst exited state energy, namely
E
1
=
e
4
m
32
2

2
0
(l + 1)
2
+
e
4
m(2l + 3)
32
2

2
0
(l + 2)
2
(l + 1)
2
. (199)
Considering that we want to nd the entire energy spectrum, we would have to repeat
the previous steps multiple times. However, we can also try to generalise the found
result and nd a formula for E
n
. Since we know the remainder R(l) and the trans-
formation relation f(l) = l + 1, we have all the tools we need to do so. When staring
at eq. (199) for a while and looking at the derivation of how it was obtained, one can
notice that
R(l) =
e
4
m(2l + 3)
32
2

2
0
(l + 2)
2
(l + 1)
2
=
e
4
m(2(l + n 1) + 3)
32
2

2
0
(l + 1 + n)
2
(l + n)
2
with n = 1. (200)
From section 3, we know that the energy of any given state is equal to the ground
63
state energy plus the sum of the remainders. Hence,
E
n
= E
0
+

e
4
m(2(l + n 1) + 3)
32
2

2
0
(l + 1 + n)
2
(l + n)
2
. (201)
Considering that we were interested in the case for which l = 0, we can shorten eq.
(201) to
E
n
=
e
4
m
32
2

2
0
n
2
, (202)
which is the known formula for the energy levels of the Hydrogen atom. Thus we have
seen that supersymmetry applied to shape invariant potentials supplies powerful tools
for the computation of energy spectra. To obtain the corresponding wave functions,
one would now follow the procedure outlined in section 3. In Fig. 3.4 we can see the
rst few energy levels of hydrogen and their respective quantum numbers. These were
obtained using the preciously derived equation for E
n
.
Figure 13: The rst few energy levels of the hydrogen atom and their respective
quantum numbers as obtained from previous calculation
For further analysis of the SUSY Hydrogen atom and generalisation in d dimensions
can be found in [15] and [16].
64
3.5 Non-central Potentials
The previous examples illustrated the capacity of supersymmetry in combination with
shape invariant potentials. In Fig. 6 we can see an extensive list of algebraically solv-
able SIPs. However, most of these are one dimensional or central potentials, which
are essentially also just one-dimensional but on the half line. These potentials repre-
sent essentially almost all potentials discussed in standard non-relativistic quantum
mechanics text books. It follows naturally to ask oneself if the previously presented
method is also applicable to non-central potentials that are often left out of standard
textbooks or whether it has reached its limitations. As we will see in this section, it can
be shown that almost any non-central potential problem can be algebraically solvable
via the operator method so long as it is a separable potential and each of the separated
problems for the dierent coordinates can be classied as shape invariant [13]. In this
section, we will make use of spherical coordinates, but a similar treatment in other
orthogonal curvilinear coordinates would yield the same results. Here we will assume
the standard ranges r, and (0 r , 0 , 0 2). The Schrodinger
equation is seperable for potentials of the form [13]
V (r, , ) = V (r) +
V ()
r
2
+
V ()
r
2
sin
2
()
. (203)
This actually also represents the most general potential of that form. Note that V (r),
V () and V () do not have to be potentials of the same form and can be arbitrary.
The symbol V was merely used for simplicity. If we now choose V (r), V () and
V () to be well known SIPs, we can immediately nd their respective eigenvalues
and eigenfunctions. Hence, making use of supersymmetry and the shape invariance
condition successively, we can algebraically solve a vast group of non-central potentials.
From [3], [9] and [13] we know the Schr odinger equation for the wave function (r, , )
65
to be
_
(

r
2
+
2
r

r
)
1
r
2
(

2
+ cot

)
1
r
2
sin
2

2
_
= (E V ). (204)
We will now see why a potential of the form as eq. (203) is separable under spherical
coordinates. As it is common for the separation of variables, it is convenient to write
(r, , ) in a specic form. In our case this turns out to be
(r, , ) =
R(r)
r
H()

sin
K(), (205)
where R(r), H() and K() are somewhat arbitrary functions of their argument.
When inserting (r, , ) into eq. (204), we obtain

1
R
d
2
R
dr
2
+ V (r)
1
4r
2
+
1
r
2
_

1
H()
d
2
H()
d
2
+ V ()
1
4
csc
2

_
+
1
r
2
sin
2

1
K()
d
2
K()
d
2
+ V ()
_
= E.
(206)
Now, as customary for separation of variables, we will assume the independent func-
tions R(r), H() and K() to obey separate dierential equations outlined by eq.
(206). Hence, suppose K() respects

d
2
K()
d
2
+ V ()K() = m
2
K(), (207)
where m is a separation constant. Substituting this into eq. (206) we can rid ourselves
of the -dependence. Eq. (206) then yields

1
R
d
2
R
dr
2
+V (r)
1
4r
2
) +
1
r
2
_

1
H()
d
2
H()
d
2
+ V () + (m
2

1
4
) csc
2

_
= E. (208)
Now, we can identify H() to obey
66

d
2
H()
d
2
+ [V () + (m
2

1
4
) csc
2
]H() = l
2
H(), (209)
where l is another separation constant. If we now insert this into eq. (208) we have
completed the separation of variables. The equation then yields the radial equation

d
2
R
dr
2
+ [V (r) + (l
2

1
4
)
1
r
2
]R(r) = ER(r). (210)
Hence, via separation of variables we have turned the non-central potential problem
into 3 separate Schrodinger equations, namely

d
2
K()
d
2
+ V ()K() = m
2
K(), (211)

d
2
H()
d
2
+ [V () + (m
2

1
4
) csc
2
]H() = l
2
H(), (212)

d
2
R
dr
2
+ [V (r) + (l
2

1
4
)
1
r
2
]R(r) = ER(r). (213)
These can now be algebraically solved by implementing appropriate shape invariant
potentials for V (r), V () and V (), for which the eigenvalues and eigenfunctions can
be either found in Fig. 6 or derived with help of the introduced tools. Additional
information and examples including a 7 parameter non-central potential can be found
in [5] and [13]. This section was added for completeness purposes to include the most
often treated types of potentials in QM.
67
4 Conclusion
Via a general Hamiltonian we have seen the factorisation method and conditions
necessary for a quantum mechanical problem to be treated via a supersymmetric ap-
proach. As long as the ground state wave function of a potential is normalisable and
the corresponding ground state energy can be shifted to 0, supersymmetry holds. A
superpotential can either be derived from a given potential or vice versa and then the
partner potentials can be found. If the initial Hamiltonian has n bound states, we
can, with this method, obtain a total of n1 Hamiltonians all having the same energy
solutions, except the rst k1 states for each Hamiltonian H
k
. The wave functions of
the partner Hamiltonians are related via operators A and A

that represent annihila-


tion and creation operators, such as used in the treatment of the quantum mechanical
harmonic oscillator. With the help of various examples we have seen that partner po-
tentials can dier vastly in dependence on variables and shape. The supersymmetric
treatment of the innite square well strongly underlines this point. Furthermore, it
was seen that supersymmetry can shed some light on the appearance of reection-
less potentials by linking transmission and reection coecients of partner potentials.
Thereby, it was illustrated that reectionless potentials occur if their partner potential
is a constant potential.
The hierarchy of Hamiltonians has illuminated how knowledge of a analytically
solvable potential and whether it respects supersymmetry as a good symmetry of
nature can support us in nding other solvable potentials, whilst already knowing
their energyspectra. The shape invariance condition was illustrated and explained at
length. Furthermore, it was shown how the principle of the hierarchy of Hamiltonians
in combination with shape invariant potentials supply powerful tools in solving many
quantum mechanical potentials. Using shape invariance we saw that the energy levels
of a potential can be found via the remainder in the shape invariance condition. This
tool allowed us to almost immediately recognise energy levels of given potentials, a
point that was stressed through the treatment of the Shifted Harmonic Oscillator, the
68
Hulthen potential and the Hydrogen atom. For the latter, we used the previously
derived tools to determine the rst few energy states of the atom, limiting ourselves
to the case that the azimuthal number l = 0. Following we assessed the applicability
of supersymmetry to non-central potentials and concluded that treatment of these via
SUSY QM is possible for as long as they are separable potentials and their various
parts are shape invariant.
Considering that this thesis represents an introduction to supersymmetric quan-
tum mechanics there is still a lot of room for further investigation. After analysing
the solvable problems, one could for example focus on approximation methods and
how supersymmetry inuences these. Additionally, one should look further into the
multidimensional case of SUSY QM and generalise the treatment given within this
thesis. In that regard, an investigation of the supersymmetric dirac equation could
yield interesting results. Also, a study of the path integral formulation or superspace
formulation of SUSY quantum mechanics could be a possible next step for further
study as well as a deeper investigation of the underlying algebra.
69
5 References
[1] Jonathan Bougie, Asim Gangopadhyaya, Jery Mallow, and Constantin Rasi-
nariu. Supersymmetric quantum mechanics and solvable models. Symmetry,
4(4):452473, August 2012.
[2] Barnali Chakrabarti. Spectral properties of supersymmetric shape invariant po-
tentials. Pramana, 70(1):4150, February 2008.
[3] Fred Cooper, Avinash Khare, and Uday Pandurang Sukhatme. Supersymmetry
in quantum mechanics. World Scientic, Singapore; River Edge, N.J., 2001.
[4] R. de Lima Rodrigues. The Quantum mechanics SUSY algebra: An Introductory
review. 2002.
[5] Ranabir Dutt. Noncentral potentials and spherical harmonics using supersym-
metry and shape invariance. American Journal of Physics, 65(5):400, 1997.
[6] Ranabir Dutt, Avinash Khare, and Uday Pandurang Sukhatme. Supersymmetry,
shape invariance, and exactly solvable potentials. American Journal of Physics,
56(2):163168, 1988.
[7] Richard P Feynman. Statistical mechanics a set of lectures. Westview Press,
[Boulder, Colo.], 1998.
[8] L.E. Gendenshtein. Derivation of exact spectra of schrdinger equation by means
of supersymmetry. JETP Lett, 38(356), 1983.
[9] David J Griths. Introduction to quantum mechanics. Pearson Prentice Hall,
Upper Saddle River, NJ, 2005.
[10] E. Harikumar, V.Sunil Kumar, and Avinash Khare. Supersymmetric quantum
mechanics on non-commutative plane. Physics Letters B, 589(3-4):155161, June
2004.
70
[11] Michiel Hazewinkel. Encyclopaedia of mathematics. Springer-Verlag, Berlin; New
York, 2002.
[12] L. Infeld and T. Hull. The factorization method. Reviews of Modern Physics,
23(1):2168, January 1951.
[13] Avinash Khare. Exactly solvable noncentral potentials in two and three dimen-
sions. American Journal of Physics, 62(11):1008, 1994.
[14] Avinash Khare. Supersymmetry in quantum mechanics. volume 744, pages 133
165. AIP, 2004.
[15] A. Kirchberg, J.D. Lnge, P.A.G. Pisani, and A. Wipf. Algebraic solution of the
supersymmetric hydrogen atom in d dimensions. Annals of Physics, 303(2):359
388, February 2003.
[16] R. P. Martnez-y Romero. On solvable potentials, supersymmetry, and the one-
dimensional hydrogen atom. Communications and Network, 02(01):6264, 2010.
[17] Erik Nygren. Supersymmetric quantum mechanics. Institute for Theoretical
Physics, 2010.
[18] Satoru Odake and Ryu Sasaki. Shape invariant potentials in discrete quantum
mechanics. J.Nonlin.Math.Phys., 12:S507S521, 2005.
[19] Martin F. Sohnius. Introducing supersymmetry. Physics Reports, 128(2-3):39
204, November 1985.
[20] Candadi V. Sukumar. Supersymmetric quantum mechanics and its applications.
AIP Conf Proc, 744:166235, 2005.
[21] Edward Witten. Dynamical breaking of supersymmetry. Nuclear Physics B,
188(3):513554, October 1981.
71

You might also like