THE SUPERSYMMETRIC

DIRAC EQUATION
The Application to Hydrogenic Atoms
This page intentionally left blank
 THE SUPERSYMMETRIC
DIRAC EQUATION
The Application to Hydrogenic Atoms

Allen Hirshfeld
Technical University of Dortmund, Germany

Imperial College Press

ICP
 Published by
Imperial College Press
57 Shelton Street
Covent Garden
London WC2H 9HE

Distributed by
World Scientific Publishing Co. Pte. Ltd.
5 Toh Tuck Link, Singapore 596224
USA office: 27 Warren Street, Suite 401-402, Hackensack, NJ 07601
UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE

British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library.

Paul Dirac photograph by A. Börtzells Tryckeri, courtesy of AIP Emilio Segre Visual Archives,
E. Scott Bar and Weber Collections.

THE SUPERSYMMETRIC DIRAC EQUATION

The Application to Hydrogenic Atoms
Copyright © 2012 by Imperial College Press
All rights reserved. This book, or parts thereof, may not be reproduced in any form or by any means,
electronic or mechanical, including photocopying, recording or any information storage and retrieval
system now known or to be invented, without written permission from the Publisher.

For photocopying of material in this volume, please pay a copying fee through the Copyright
Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to
photocopy is not required from the publisher.

ISBN-13 978-1-84816-797-1
ISBN-10 1-84816-797-0

Typeset by Stallion Press

Email: enquiries@stallionpress.com

Printed in Singapore.

Catherine - The Supersymmetric Dirac.pmd 1 9/30/2011, 3:24 PM

August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-fm

I wish to dedicate this book to the memory of my father, Dr. Martin A. Hirshfeld,
who taught me many things, among them the meaning of proof.

v
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-fm

This page intentionally left blank

August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-fm

Preface

One morning, when I was an undergraduate, over a cup of coffee in the cafeteria
of the Weizman Institute, where he was working on his PhD in physics, Charles
Robinson opened my eyes to the wonders of the Dirac equation, and sowed the
seed for this book.
I met Paul Dirac in the early 1960s, on the occasion of his visit to the Physics
Department of the Tel-Aviv University. We talked about my physics project at that
time (I was working on my Master’s thesis), and I attended his talk, where he
explained the difficulties in the foundations of quantum field theory. Afterwards
he autographed his book for me.
I am very grateful to Jens Peder Dahl, Professor Emeritus of the Chemistry
Department of the Technical University of Denmark, for the fruitful and
encouraging correspondence on his work.
I wish to thank Dirk Fischer for his assistance with the technical aspects of
the preparation of the manuscript, and especially for his help in the preparation of
the figures.

vii
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-fm

This page intentionally left blank

August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-fm

Contents

Preface vii

List of Figures xiii

1. Introduction 1

2. The Classical Kepler Problem 5

2.1 Central Forces . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.2 The Laplace Vector . . . . . . . . . . . . . . . . . . . . . . . 8

3. Symmetry of the Classical Problem 15

3.1 Lie Groups and Lie Algebras . . . . . . . . . . . . . . . . . 15
3.2 Some Special Lie Algebras . . . . . . . . . . . . . . . . . . 18
3.3 Poisson Brackets . . . . . . . . . . . . . . . . . . . . . . . . 22
3.4 The Inverse Square Law . . . . . . . . . . . . . . . . . . . . 25

4. From Solar Systems to Atoms 31

4.1 Rutherford Scattering . . . . . . . . . . . . . . . . . . . . . 31
4.2 Conservation of the Laplace Vector . . . . . . . . . . . . . . 32
4.3 The Differential Cross Section . . . . . . . . . . . . . . . . . 33

5. The Bohr Model 35

5.1 Spectroscopic Series . . . . . . . . . . . . . . . . . . . . . . 35
5.2 The Postulates of the Model . . . . . . . . . . . . . . . . . . 35
5.3 The Predictions of the Model . . . . . . . . . . . . . . . . . 37
5.4 Correction for Finite Nuclear Mass . . . . . . . . . . . . . . 42

6. Interpretation of the Quantum Rules 47

6.1 The Sommerfeld–Wilson Quantization Conditions . . . . . . 47
6.2 de Broglie’s Wave Interpretation . . . . . . . . . . . . . . . . 49

ix
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-fm

x The Supersymmetric Dirac Equation

7. Sommerfeld’s Model for Non-Relativistic Electrons 53

7.1 Assumptions of the Model . . . . . . . . . . . . . . . . . . . 53
7.2 Results of the Model for Non-Relativistic
Hydrogen Atoms . . . . . . . . . . . . . . . . . . . . . . . . 55
7.3 The Eccentricity . . . . . . . . . . . . . . . . . . . . . . . . 57

8. Quantum Mechanics of Hydrogenic Atoms 59

8.1 Quantization . . . . . . . . . . . . . . . . . . . . . . . . . . 59
8.2 Quantum Mechanical Relation Between |A|2 and L2 . . . . . 61
8.3 Pauli’s Hydrogenic Realization of so(4) . . . . . . . . . . . . 62
8.4 so(4) and the Spectrum of Hydrogenic Atoms . . . . . . . . . 66

9. The Schrödinger Equation and the Confluent

Hypergeometric Functions 69
9.1 The Schrödinger Equation and Its Solutions . . . . . . . . . . 69
9.2 Laguerre Polynomials and Associated
Laguerre Functions . . . . . . . . . . . . . . . . . . . . . . . 73

10. Non-Relativistic Hydrogenic Atoms with Spin 77

10.1 Spin Variables, the Pauli Hamiltonian
and Factorization . . . . . . . . . . . . . . . . . . . . . . . . 77
10.2 A Theorem Concerning the Anticommutation of K . . . . . . 81
10.3 Pauli Spinors . . . . . . . . . . . . . . . . . . . . . . . . . . 82
10.4 Concerning the Operator (σ·r̂) . . . . . . . . . . . . . . . . 83
10.5 The Key Equation: Concerning the Operator (σ·A) . . . . . 84
10.6 The Factorization Method . . . . . . . . . . . . . . . . . . . 86
10.7 The Definition of Eccentricity . . . . . . . . . . . . . . . . . 88

11. Elements of Supersymmetric Quantum Mechanics 91

11.1 General Considerations . . . . . . . . . . . . . . . . . . . . 91
11.2 Supersymmetry of Non-Relativistic
Hydrogenic Atoms . . . . . . . . . . . . . . . . . . . . . . . 94

12. Sommerfeld’s Derivation of the Relativistic Energy

Level Formula 99
12.1 Assumptions of the Model . . . . . . . . . . . . . . . . . . . 100
12.2 The Energies of the Bound States . . . . . . . . . . . . . . . 102
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-fm

Contents xi

13. The Dirac Equation 105

13.1 The Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . . 105
13.2 Total Angular Momentum . . . . . . . . . . . . . . . . . . . 107
13.3 The Dirac Operator . . . . . . . . . . . . . . . . . . . . . . . 108
13.4 A Complete Set of Mutually Commuting Operators . . . . . . 110
13.5 The Dirac Spinors . . . . . . . . . . . . . . . . . . . . . . . 111
13.6 The Radial Equations in Polar Coordinates . . . . . . . . . . 111

14. The Primary Supersymmetry of the Dirac Equation 115

14.1 A Derivation of the Johnson–Lippmann Operator . . . . . . . 115
14.2 Commutation and Anticommutation Relations
of the Johnson–Lippmann Operator . . . . . . . . . . . . . . 119
14.3 Eccentricity . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
14.4 The Johnson–Lippmann Operator as the Generator
of Supersymmetry . . . . . . . . . . . . . . . . . . . . . . . 121

15. Extending the Solution Space 127

15.1 The Γ-Induced Radial Supersymmetry . . . . . . . . . . . . 129
15.2 The Supersymmetric Ground State
in the Γ Representation . . . . . . . . . . . . . . . . . . . . 133
15.3 The General Solutions in the Γ Representation . . . . . . . . 134

16. A Different Extension of the Solution Space 139

16.1 The ∆-Induced Radial Supersymmetry . . . . . . . . . . . . 141
16.2 The Supersymmetric Ground State in the ∆ Representation . 144
16.3 The General Solutions in the ∆ Representation . . . . . . . . 145

17. The Relation of the Solutions to Kramer’s Equation 149

17.1 The Eigenvalue Problem for Traceless
2 × 2 Matrices . . . . . . . . . . . . . . . . . . . . . . . . . 149
17.2 Eigenfunctions of the Operators Γ and ∆ . . . . . . . . . . . 151
17.3 Kramer’s Equation . . . . . . . . . . . . . . . . . . . . . . . 154

18. Non-Relativistic Approximation 161

18.1 The Γ Representation . . . . . . . . . . . . . . . . . . . . . 163
18.2 The ∆ Representation . . . . . . . . . . . . . . . . . . . . . 164

19. Conclusions 167

August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-fm

xii The Supersymmetric Dirac Equation

Appendices 171
Appendix A. The Confluent Hypergeometric Function . . . . . . . . 171
Appendix B. Orthogonality Relations of Hypergeometric
Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
Appendix C. More Integrals Involving Hypergeometric
Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
Appendix D. Normalization for the Γ-Induced Scheme . . . . . . . 182
Appendix E. Normalization for the ∆-Induced Scheme . . . . . . . 185

Solutions to the Exercises 189

Bibliography 195

Index 199
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-fm

List of Figures

2.1 The solar system as we understand it . . . . . . . . . . . . . . . . 6

2.2 An ellipse . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
4.1 The incoming and outgoing directions
and the scattering angle . . . . . . . . . . . . . . . . . . . . . . . 32
4.2 Scattering of an incident beam of particles by a center
of force . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
5.1 The spectral series of hydrogen . . . . . . . . . . . . . . . . . . . 36
5.2 The Bohr model of the hydrogen atom . . . . . . . . . . . . . . . 37
5.3 The electron in the ground state of the Bohr atom.
(a) Classical picture; (b) Quantum Mechanical picture . . . . . . . 38
5.4 The probability of finding the electron at a given distance from the
nucleus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
5.5 Energy-level diagram of the Bohr model . . . . . . . . . . . . . . 40
6.1 de Broglie waves associated with an electron . . . . . . . . . . . . 52
7.1 The Sommerfeld model of the hydrogen atom . . . . . . . . . . . 57
10.1 Low-lying orbits of the hydogen atom . . . . . . . . . . . . . . . 89
11.1 The supersymmetry of the non-relativistic
hydrogen spectrum . . . . . . . . . . . . . . . . . . . . . . . . . 97
14.1 The supersymmetry of the relativistic
hydrogen spectrum . . . . . . . . . . . . . . . . . . . . . . . . . 116
19.1 The development of our understanding
of the hydrogen spectrum . . . . . . . . . . . . . . . . . . . . . . 169

xiii
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-fm

This page intentionally left blank

August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch01

Chapter 1
Introduction

“He succeeded in combining quantum mechanics and the theory of relativity, . . . and
introduced the idea of an antiparticle . . . Unfortunately, only the expert can appreciate
Dirac’s awesome work. To this day, that work has not lost any of its splendor.”

Martinus Veltman, Facts and Mysteries in Elementary Particle Physics, World Scientific
Publishing, Singapore, 2003.

The sentiment expressed in this quotation is certainly true. But let us pause to
consider the phrase “only the expert can appreciate Dirac’s awesome work . . . ”. We
expect a student beginning to study quantum field theory to understand the working
of relativistic quantum theory as it applies to free particles. The next interesting
case, that of a particle in a Coulomb field, is given at best a somewhat cursory
treatment in most texts. But it was precisely this problem that was the test of the
new theory — could it explain the fine structure in the spectrum of hydrogen? The
fact of this success and the detailed structure that it predicts were the hallmark of
the theory for many years. It was the work devoted to testing this structure, with the
highest possible precision, which revealed the need for the radiative corrections of
quantum electrodynamics and provided us with the most accurate evaluation of the
fine structure, which we possess to the present day, and one of the most impressive
successes in all of physics.
It is thus highly desirable that the modern student of relativistic quantum
mechanics should have an appreciation of this decisive episode in the physics
of the twentieth century. The advances of the theory in recent years have led to
certain simplifications with respect to Dirac’s time. In particular, the discovery that
supersymmetry plays a central role in the structure of the theory of a particle in
a Coulomb field means that while gaining a mastery of this topic, students are
at the same time gaining experience that will be of importance in appreciating
the questions with which their future work will be concerned, namely, does
supersymmetry play a role in the structure of the universe? Even if this question is,
contrary to expectation, answered in the negative, the knowledge of supersymmetry

1
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch01

2 The Supersymmetric Dirac Equation

is valuable because it allows a better understanding of many solvable problems in

physics, and because of the role it plays in mathematics.
Thus, the purpose of the present book is to impart to the reader the necessary
“expertise” for appreciating the application of the Dirac theory to the hydrogen
atom. It will become clear that most of the concepts can be introduced in the more
familiar, non-relativistic context in an elementary fashion. We will also become
acquainted with a previous attempt to derive the formula for the fine structure of
hydrogen, due to Sommerfeld. In this work, the identical formula for the energy
levels of the hydrogen atom was obtained, 12 years before Dirac. This was before
the Schrödinger wave equation for the non-relativistic hydrogen atom, and before
the discovery of the spin of the electron! It is built on the Bohr model of the atom,
but takes into account the relativistic motion of the electron. We shall see that it
characterizes the orbits by their eccentricity, in a way that agrees with the Dirac
theory.
We also take advantage, in our exposition, of the opportunity provided to
relive the highlights of the history of physics. Kepler’s observation that Mars, in its
motion around the Sun, follows an elliptical orbit, marks the beginning of physics
as a quantitative science. In subsequent years, this system of two bodies interacting
through the influence of a central force has served as a model of such disparate
motions as the collision of galaxies and the orbits of electrons about the nuclei of
atoms. It is the latter development that concerns us here.
For a particle moving in a central force field the angular momentum vector L
is conserved. If the force is inversely proportional to the distance squared there is
another conserved vector, the Laplace vector A. The Laplace vector lies along the
fixed axis of the orbital motion, and its magnitude is a measure of the eccentricity
e = |A|, which characterizes the deviation of the orbit from a circle. The Laplace
vector turns out to be one of the central concepts of the book. The three components
of the angular momentum, complemented by the three components of the Laplace
vector, form, properly normalized, the generators of an SO(4) symmetry group.
The Rutherford [ Rut11] picture of the atom as a miniature solar system
suggested an approach to understanding the spectral lines of light emitted in atomic
transitions. With Bohr’s [Boh13] “quantum rules”, based on the postulate of a
smallest unit of orbital angular momentum, the era of the “old quantum theory”
was initiated.
A more refined version of the Bohr model is a non-relativistic form of
Sommerfeld’s model [Som16], based on the Sommerfeld–Wilson quantization
rules. The set of elliptical orbits predicted in the classical theory is reproduced,
along with the formula for their eccentricity.
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch01

Introduction 3

The beginning of the “new quantum theory” was marked, as a matter of fact, by
Pauli’s [Pau26] group theoretical treatment of the hydrogen atom. Pauli’s analysis
is based on the existence of the Laplace vector A as a conserved quantity of the
motion of a point particle moving under the influence of an inverse-quadratic force.
Its three components form, together with the three components of the orbital angular
momentum, the generators of the Lie algebra so(4). The quadratic Casimir operator
of this algebra then yields a formula for the spectrum of the bound-state energy
levels of hydrogenic atoms. The advantage of this approach over Schrödinger’s
[Sch26] method of solving a differential equation is that it automatically yields
an understanding of the degeneracy of the spectral lines, which is treated in the
Schrödinger approach as an “accidental” symmetry.
A full analysis of the problem of an electron with spin in a Coulomb field
involves the two-component Pauli formalism. It is only necessary to replace p by
(σ · p) to get the correct gyromagnetic ratio. The inclusion of the spin variables
provides a formalism in which quadratic operators of the theory factorize. The
effectiveness of such factorizations was already noted by Hull and Infeld [HI51],
whose work was in a sense a precursor of the theory of supersymmetry, which arose
originally in the quantum field theory of elementary particles. The Schrödinger
equation for the radial component of the wave function is seen to be the product of
two linear differential operators, whose eigenvalues are related to the magnitude of
the operator Anr = (σ ·A), in a way that is the quantum version of the eccentricity
formula e = |A|. Couching the results of the treatment of spin in a supersymmetric
language leads to an understanding of the full degeneracy of the hydrogen spectrum
as a consequence of the SO(4) × S(2) supersymmetry group.
When we ask for an understanding of the fine structure of the spectrum we
move on to the treatment of the relativistic spectrum. Today we undertake a study of
the solutions of the Dirac equation describing an electron in a Coulomb potential.
In this book we first follow in the footsteps of Sommerfeld, who undertook a
treatment of the relativistic problem, although he had at his disposal neither the
modern quantum theory nor the notion of electron spin. It is all the more astonishing
that he obtained a formula for the spectrum of hydrogen that agrees with the results
of the Dirac theory. The Sommerfeld theory not only gives the correct formula for
the energy of the hydrogen orbits, it also describes their relativistic eccentricity.
One of the purposes of this book is to explain this success, based on the work of
Biedenharn [Bie83].
When we go over to the relativistic problem the SO(4) symmetry breaks down,
but a remnant of the Laplace vector survives, namely the Johnson–Lippmann
operator A [JL50]. The Johnson–Lippmann operator is, in the non-relativistic
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch01

4 The Supersymmetric Dirac Equation

approximation, the operator Anr . In this way the degeneracy of the spectrum
reduces to that of SO(3) × S(2), with SO(3) the symmetry generated by the total
angular momentum J = L + S, where S is the spin vector that generates the SO(3)
rotation group, and S(2) is the supersymmetry group.
By describing the Dirac theory in a supersymmetric framework we are able to
derive the solutions directly, without resorting to the heuristic methods employed
formerly. Here we follow the work of Dahl and Jørgensen [DJ95]. The solutions
of the Dirac equation are obtained by investigating two different extensions of
the solution space. We describe the relation of the solutions to the second-order
Kramer’s equation, which was used formerly to construct solutions of the Dirac
equation.
Finally, the non-relativistic approximations of the solutions found for the Dirac
equation are shown to reduce to the solutions of the Pauli equation. This justifies
the choice of sign for the solutions, and allows a check on the various normalization
factors used in this book.
The Appendices deal with some of the features of the confluent hypergeometric
functions, and their relation to the wave functions of the Pauli and Dirac equations.
The SO(3) × S(2) symmetry of the Dirac theory is broken by the radiative
corrections of quantum field theory, notably in the Lamb shift [LR47], which
describes the energy difference between the 2s 12 and 2p 12 levels of hydrogen.
Because of this, the Dirac theory of bound states belongs in a course of study after
non-relativistic quantum mechanics and before the introduction to the quantum
theory of interacting fields. It is for such students that the present book is intended.

A note on units
This book uses the system of units in which
and the velocity of light are set to
unity:
= c = 1, where
= h/(2π), and h is Planck’s constant. The Einstein
convention is used throughout, according to which an index that appears in an
expression twice is summed over. The fine structure constant is α = e2 /4π, where
e is the charge of the electron.
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch02

Chapter 2
The Classical Kepler Problem

“It is not eighteen months since I first caught a glimpse of the light, three months since
the dawn, very few days since the unveiled Sun, most admirable to gaze upon, burst upon
me. Nothing restrains me; I shall indulge my sacred fury; I shall triumph over mankind by
the honest confession that I have stolen the golden vases of the Egyptians to build up a
tabernacle for my God far from the confines of Egypt. If you forgive me, I rejoice; if you are
angry, I can bear it; the die is cast, the book is written, to be read either now or by posterity,
I care not which; it may well wait a century for a reader, as God himself has waited six
thousand years for someone to behold his work.”

Johannes Kepler, Gesammelte Werke, 18 volumes, C. H. Beck, Munich, 1937–1969.

We are all familiar with the modern understanding of our solar system. The
Sun sits at the center of a ring of essentially concentric orbits, each planet following
an elliptical path around the Sun, which sits not at the center but at one focus of
the ellipse. It is the picture shown in Figure 2.1. No one has ever seen this picture.
Even today there are no satellites sitting at that vantage point tracking the orbits
and plotting their paths, much less in the seventeenth century when this picture was
first proposed by Johannes Kepler. Kepler and the astronomers before and since
him have constructed their understanding of the solar system by watching the Sun
and planets, observable as points of light, move across the sky. We only see this
two-dimensional projection of their paths.
Kepler’s “unveiled Sun” was the realization that the observational data of
Tycho Brahe, his mentor, could be explained by the following three laws:

• A planetary orbit is an ellipse with the Sun at one focus.

• A planetary orbit sweeps out equal areas in equal times.
• The square of the period of the orbit is directly proportional to the cube of its
mean distance.

Kepler’s laws of motion were explained by Newton. Newton was able to show
that they could be deduced from his universal law of gravitation, involving a central

5
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch02

6 The Supersymmetric Dirac Equation

Figure 2.1. The solar system as we understand it.

force that falls off as the square of the distance. Taking the mass of the central
attractor as infinite and the planet as a point particle leads to the problem of a point
mass in a central potential field. The motion takes place in a fixed plane because of
the conservation of angular momentum. The fact that the major axis of the ellipse
remains stationary for a 1/r potential was known, by the time of Laplace, to be due
to the conservation of the Laplace (Runge–Lenz) vector [Gol75], whose magnitude
is equal to the eccentricity of the ellipse. With the model of an atom as a miniature
solar system these concepts became relevant to the motion of electrons about the
nucleus.

2.1 Central Forces

The velocity of the point P in space is

dr dr dr̂ dr dθ
v= = r̂ + r = r̂ + r θ̂, (2.1)
dt dt dt dt dt

where r = rr̂ is the position vector and r̂ is a unit vector on the direction of r.
θ̂ is a unit vector perpendicular to r̂:

r̂ = (cos θ(t), sin θ(t)), θ̂ = (− sin θ(t), cos θ(t)). (2.2)

August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch02

The Classical Kepler Problem 7

The acceleration is


dv d2 r dr dθ d2 θ dθ dθ
a= = 2 r̂ + 2 θ̂ + r 2 θ̂ + r − r̂
dt dt dt dt dt dt dt

2 

d2 r dθ 1 d 2 dθ
= − r r̂ + r θ̂. (2.3)
dt2 dt r dt dt

Equation (2.3) tells us that the acceleration is entirely radial, that is, parallel
to r, if and only if


d dθ
r2 = 0, (2.4)
dt dt
which is equivalent to saying that r(t)2 dθ/dt is a constant independent of t.
Combining this with the next lemma gives a proof of Kepler’s second law:
Lemma. If the position of a particle over time is described by the vector function
r(t), then the rate at which the radial vector sweeps out area is given by
dA r(t)2 dθ
= . (2.5)
dt 2 dt

Proof. Given a circle with center at the origin and radius r, the area swept out by
the radius as it moves through an angle of ∆θ is given by (r 2 /2)∆θ. It follows
that if ∆A is the area swept out by the radial vector from time s to time t, and if
the distance from the origin stays constant during this time interval, then
r2
∆A = ∆θ, (2.6)
2
where ∆θ = θ(t) − θ(s). If r does not stay constant, then we can find two points in
the interval [s, t], call them t1 and t2 , where r takes on its minimum and maximum
values, respectively, over this interval: r(t1 ) ≤ r ≤ r(t2 ). It follows that
r(t1 )2 r(t2 )2
∆θ ≤ ∆A ≤ ∆θ. (2.7)
2 2
We now divide by ∆t = t − s:
r(t1 )2 ∆θ ∆A r(t2 )2 ∆θ
≤ ≤ , (2.8)
2 ∆t ∆t 2 ∆t
and take the limit as s approaches t. This forces t1 and t2 to also approach t and
yields
r(t)2 dθ dA r(t)2 dθ
≤ ≤ . (2.9)
2 dt dt 2 dt
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch02

8 The Supersymmetric Dirac Equation

Lemma. Let r(t) be the position of a particle at time t, and v(t) its velocity. The
momentum of the particle is p = mv, where m is its mass. If the particle is moving
under the influence of a central force:
dp
F = = f (r)r̂, (2.10)
dt
where f is an arbitrary function of r, then the orbital angular momentum

L=r×p (2.11)

is a constant vector of magnitude

dθ
L = |L| = mr(t)2 . (2.12)
dt

Proof. The fact that L is a constant vector is immediate:

dL d
= (r × p) = mv × v + f (r)r × r̂ = 0. (2.13)
dt dt
To prove Eq. (2.12) we write


dr dθ dθ
L = mrr̂ × r̂ + r θ̂ = mr2 (r̂ × θ̂), (2.14)
dt dt dt
so that
dθ
L = mr2 . (2.15)
dt
Since

L · r = (r × p) · r = (r × r) · p = 0, (2.16)

the motion is in a plane perpendicular to L.

2.2 The Laplace Vector

We now consider the case of a force inversely proportional to the square of the
distance from the origin:


dp k
F = = − 2 r̂. (2.17)
dt r
Kepler’s first law is a consequence of the following theorem:
Theorem. Let r(t) denote the position of a particle moving under the influence of
a force inversely proportional to the square of the distance from the origin. Then
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch02

The Classical Kepler Problem 9

there exists a constant vector A, the Laplace vector, such that

L2
|r| + r · A = . (2.18)
mk
Equivalently, if (r, θ) is the position in polar coordinates, then
L2
r(1 + e cos θ) = , (2.19)
mk
where the eccentricity e is equal to the magnitude of A: e = |A|. We recognize
Eq. (2.19) as the equation of a conic section: an ellipse, parabola, or hyperbola. In
particular, if e < 1, then it is the equation of an ellipse with one focus at the origin.
Proof. We note

L2 = L · (r × p) = (p × L) · r. (2.20)

Now



d dp k 2 dθ
(p × L) = ×L= − r̂ × mr r̂ × θ̂
dt dt r2 dt
dθ dθ d
= −mk [r̂ × (r̂ × θ̂)] = mk θ̂ = (mk r̂). (2.21)
dt dt dt
This means that the time derivative of p × L − mk r̂ is zero, so
1
A= (p × L) − r̂ (2.22)
mk
is a constant vector, and

p × L = mk(r̂ + A). (2.23)

Combining this result with Eq. (2.20), we see that

L2 = (p × L) · r = mk(|r| + r · A), (2.24)

or
L2
|r| + r · A = . (2.25)
mk
Equation (2.19) follows from the equalities |r| = r and r · A = re cos θ.
An ellipse with foci F1 and F2 can be defined as the locus of points P for
which |F1 P | + |F2 P | = 2a, with a the semimajor axis.
 The point P is described
by the vector function r(t) = r(t)r̂(t), with r(t) = x(t)2 + y(t)2 and r̂(t) =
(cos θ(t), sin θ(t)). See Figure 2.2.
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch02

10 The Supersymmetric Dirac Equation

a
F1 F2
b

Figure 2.2. An ellipse.

Lemma. The general equation of an ellipse with one focus at the origin and
semimajor axis along the x-axis is given in polar coordinates by

r(1 + e cos θ) = c, (2.26)

where e and c are constants, 0 < e < 1, c > 0. The semimajor axis is
c
a= , (2.27)
1 − e2
and the semiminor axis is
c
b= √ . (2.28)
1 − e2
The eccentricity of the ellipse is

e= 1 − (b/a)2 . (2.29)

Proof. Equation (2.26) can be rewritten as

r = c − er cos θ. (2.30)

We convert this to Cartesian coordinates (x = r cos θ, y = r sin θ) and square

each side to obtain

x2 + y 2 = c2 − 2ecx + e2 x2 , (2.31)

or

(1 − e2 )x2 + 2ecx + y 2 = c2 . (2.32)

Now add c2 e2 /(1 − e2 ) to each side:

c2 e2 c2 e2
(1 − e2 )x2 + 2ecx + + y 2 = c2 + , (2.33)
1−e 2 1 − e2
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch02

The Classical Kepler Problem 11

or

c2
(1 − e2 )(x + ea)2 + y 2 = , (2.34)
1 − e2
which may be rewritten as

(x + ea)2 y2
+ = 1. (2.35)
a2 b2
This is the equation of an ellipse in Cartesian coordinates. The center of the ellipse
is at (−ae, 0), which is a distance ae from the origin. The foci are bisected by a
line that
intersects the ellipse at two points. At these points the distance from the
foci is (ae)2 + b2 = a, and the sum of the distances from the foci, which is the
same for all points of the ellipse, is 2a.
The perigee, which is the point of closest approach, corresponds to θ = 0, so

L2
rmin = a(1 − e) = . (2.36)
mk(1 + e)

The apogee, which is the farthest distance, corresponds to θ = π, so

L2
rmax = a(1 + e) = . (2.37)
mk(1 − e)

The mean distance is 12 (rmin + rmax ) = a.

The Laplace vector A is perpendicular to the angular momentum,

A · L = 0, (2.38)

so A lies in the plane of motion of r. As a matter of fact, A lies along the line
joining the two foci.
We express the eccentricity in terms of the constants of the motion, the energy
E, and the angular momentum L. We use the identities

(a) (p × L) · (p × L) = p2 L2 ,
(b) (p × L) · r = L · (r × p) = L2 . (2.39)

We then have


p2 L2 2L2 2L2 p2 k 2L2 E
|A| = 2 2 −
2
+1 = − +1 = +1 = e2 , (2.40)
m k mkr mk2 2m r mk2
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch02

12 The Supersymmetric Dirac Equation

where
p2 k
E= − (2.41)
2m r
is the constant energy of a particle in a fixed orbit.
From the equation for the perigee we have

L2
rmin = a(1 − e) = , (2.42)
mk(1 + e)
or
L2
1 − e2 = . (2.43)
mka
Comparing this to Eq. (2.40):

2L2 E
1 − e2 = − , (2.44)
mk 2
we get

k
E=− . (2.45)
2a
The energy of a particle in orbit depends only on the semimajor axis of the ellipse.
We can now derive Kepler’s third law. By Kepler’s first law a particle
√ in orbit
moves on an ellipse. The semiminor axis of the ellipse is b = a 1 − e2 . By
Kepler’s second law the area of the ellipse is
 τ
dA L
A= dt = τ, (2.46)
0 dt 2m

where τ is the period of a complete revolution. Hence


 −2E
L 2 L L
A= τ = πab = πa 1 − e = πa
2 2 = πa3/2 √ , (2.47)
2m k m km
or

3/2 m
τ = 2πa . (2.48)
k

The square of the period of the orbit is therefore proportional to the cube of a,
which is the mean distance.
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch02

The Classical Kepler Problem 13

Notes on Chapter 2
The conservation of the Laplace vector and the interpretation of its length as
eccentricity are first mentioned by Jacob Hermann, in a letter to Bernoulli. It
was then used by Laplace in his treatise on mechanics. It was rediscovered by
Hamilton in the nineteenth century, and included in textbooks by Gibbs and Runge.
Pauli learned about it from his teacher Lenz, and after Pauli’s quantum mechanical
treatment of the hydrogen atom (Chapter 8) it is commonly referred to as the
Runge–Lenz vector. The early history of the Laplace vector was traced by Goldstein
[Gol75] [Gol76]. See also the Wikipedia page at http://en.wikipedia.org/wiki/
Laplace–Runge–Lenz_vector. The presentation here follows Bressoud [Bre91].
The origin of the Laplace vector as the boost-generator of the relativistic extension
of the theory has been proposed by Dahl [Dah95].
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch02

This page intentionally left blank

August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch03

Chapter 3
Symmetry of the Classical Problem

“Symmetry, as wide or as narrow as you may wish to define it, is one idea by which man
through the ages has tried to comprehend and create order, beauty, and perfection.”

Hermann Weyl, in Symmetry, Princeton University Press, Princeton, 1952.

In this chapter we make an excursion into mathematics. We first review the

definitions of Lie groups and Lie algebras, because these are the mathematical
basis of symmetry considerations in physics. We show how these concepts fit into
the Hamiltonian formulation of classical mechanics. We then return to the problem
of a central force which is inversely proportional to the square of the distance, and
show that the conservation of the Laplace vector, together with the conservation
of angular momentum, is equivalent to the existence of an SO(4) symmetry group
that acts on functions of the phase space.

3.1 Lie Groups and Lie Algebras

Let A be a linear operator on a finite dimensional vector space V . If V = Rn then
A is an n × n matrix. The operator valued function t
→ etA is defined by
1 1
etA = I + tA + t2 A2 + t3 A3 + · · · (3.1)
2 3!
which converges for all t. For a fixed A, the function etA behaves like the ordinary
exponential function:
d tA
etA |t=0 = I, e(t+s)A = etA esA , e = AetA . (3.2)
dt
etA is a curve of linear operators, and the LHS of the last equation gives the tangent
vector to this curve at time t. At t = 0 we have
d tA
e |t=0 = A. (3.3)
dt

15
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch03

16 The Supersymmetric Dirac Equation

If R is some other invertible operator then

(RAR −1 )n = RAn R−1 (3.4)

for all n, so
−1
et(RAR )
= Re tA R−1 . (3.5)

In particular, the tangent vector to RetA R−1 , which goes through the identity I at
t = 0, is RAR −1 .
Suppose A is fixed and R(s) is a curve of matrices that passes through I at
s = 0 and is differentiable in s. Then R(s) is invertible for s close to the identity,
and differentiating

R(s)R(s)−1 = I (3.6)

at s = 0 gives
d −1 d
R (s)|s=0 = − R(s)|s=0 = −B, (3.7)
ds ds
and
d
[R(s)AR(s)−1 ]s=0 = BA − AB , (3.8)
ds
by Leibniz’s rule. Define the Lie bracket [ , ] by

[B, A] = BA − AB . (3.9)

The Lie bracket is bilinear, antisymmetric, and satisfies Jacobi’s identity:

[A, [B, C]] = [A, BC − CB ] = [A, BC ] − [A, CB ]

= [A, B]C + B[A, C] − [A, C]B − C[A, B]
= (AB − BA)C + B(AC − CA)
− (AC − CA)B − C(AB − BA)
= [[A, B], C] + [B, [A, C]]. (3.10)

A subgroup G of the group of all invertible matrices is said to be a linear Lie

group if it is also a closed manifold.
For example, let O(n) be the n × n matrices that satisfy OOT = I, where
T
O denotes the transposed matrix. This defines O(n) as a subgroup of the n2
dimensional space of (n × n)-matrices. Let R(s) be a curve of orthogonal matrices
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch03

Symmetry of the Classical Problem 17

with R(0) =I and R
(0) = B. Differentiate R(s)R(s)T = I at s = 0, and

obtain

B + B T = 0, (3.11)

so B is antisymmetric. The dimension of the tangent space, and hence the manifold
O(n), is thus n(n − 1)/2. O(3) is a three-dimensional subgroup of the nine-
dimensional group of all invertible 3 × 3 matrices. O(4) is six-dimensional.
Conversely, let A be an antisymmetric matrix.

T T
etA = etA , (3.12)

since (AT )n = (An )T .

d  tA tA T  d T T
e (e ) = [etA etA ] = etA (A + AT )etA . (3.13)
dt dt
But A + AT = 0. Hence
d tA tA T
[e (e ) ] = 0. (3.14)
dt
At t = 0, etA (etA )T |t=0 = I, hence for all t

etA (etA )T = I. (3.15)

If A is antisymmetric, etA is orthogonal.

Let G denote the Lie group O(n) and G the space of antisymmetric matrices.
Then G is the tangent space of G at the identity. So if R(s) is a differentiable curve
of matrices in G and R(0) = I, then R
(0) ∈ G.
If A ∈ G then the curve etA is a one-parameter subgroup of G.
The Lie bracket of two antisymmetric matrices is antisymmetric:

[A, B]T = (AB − BA)T = B T AT − AT B T

= [B T , AT ] = −[AT , B T ] = −[A, B]. (3.16)

But it is more instructive to derive this from the statements above. Indeed, let B ∈ G.
Then R(s) = esB is a curve of matrices in G. Hence for each fixed s, conjugation
by R(s) carries G into itself, and carries I into itself, and hence must carry the
tangent space G into itself. Hence if A ∈ G, the curve C(s) = R(s)AR(s)−1 lies
in G; C(s) ∈ G for all s. But then C(s + h) − C(s) ∈ G for any s and h. Dividing
by h and passing to the limit implies C
(s) ∈ G. At s = 0 this tells us that for
B ∈ G and A ∈ G also [B, A] ∈ G.
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch03

18 The Supersymmetric Dirac Equation

The corresponding facts are true in general: If G is a linear Lie group then
its tangent space G at I is the space spanned by A ∈ G, and the curve etA is a
one-parameter subgroup of G. We say that G is the Lie subalgebra of the space of
all matrices.
More generally, a linear space G with a bracket mapping G ×G
→ G is called a
Lie algebra if the bracket is bilinear, antisymmetric, and satisfies the Jacobi identity.
A group G that is a finite-dimensional manifold, where the group multiplication
is consistent with the manifold structure in an obvious way, is called a Lie group.
The tangent space to G at I (in the sense of differentiable manifolds) then inherits
the structure of a Lie algebra. Given any finite-dimensional Lie algebra G over
the real numbers, there always exists a Lie group G having G as its Lie algebra.
All such G will be “locally isomorphic”, but may differ in their global structure.
There is a unique G that is connected and simply connected having G as its Lie
algebra.

3.2 Some Special Lie Algebras

We now discuss in more detail the Lie algebras with which we shall be concerned.

o(3)
Let δi denote infinitesimal rotations about the xi -axis, i = 1, 2, 3, so that
     
0 0 0 0 0 1 0 −1 0
δ1 = 0 0 −1, δ2 =  0 0 0, δ3 = 1 0 0. (3.17)
0 1 0 −1 0 0 0 0 0

Then

[δ1 , δ2 ] = δ3 , [δ2 , δ3 ] = δ1 , [δ3 , δ1 ] = δ2 . (3.18)

This can be expressed more succinctly by [δi , δj ] = ijk δk , where


 1 for (i, j, k) an even permutation of (1, 2, 3)
ijk = −1 for (i, j, k) an odd permutation of (1, 2, 3) (3.19)

0 otherwise.

The δi are clearly linearly independent and so form a basis of o(3). So if we write
the most general element of o(3) as a = a1 δi + a2 δ2 + a3 δ3 then the preceding
equations for the brackets of the basis elements show that the Lie product in o(3)
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch03

Symmetry of the Classical Problem 19

can be identified with the usual vector product,

[a, b] = a × b. (3.20)

Notice that this operation can also be described by thinking of a as a matrix and b
as a vector and multiplying the vector b by the matrix a. For example, the fact that
[δ1 , δ2 ] = δ3 can be expressed as
    
0 0 0 0 0
0 0 −1 1 = 0. (3.21)
0 1 0 0 1

e(3)
Let E(3) denote the group of all Euclidean motions in three-dimensional space. So
E(3) consists of all transformations of R3 of the form

r
→ Rr + v, (3.22)

where R ∈ O(3) and v ∈ R3 , in other words an orthogonal transformation followed

by a translation. We can realize this transformation as matrix multiplication in four
dimensions:
  
r R v r

→ , (3.23)
1 0 1 1
so E(3) can be thought of as the group of all 4 × 4 matrices of the form

R v
. (3.24)
0 1
Therefore the Lie algebra e(3) can be identified with the algebra of all 4×4 matrices
of the form

a b
, (3.25)
0 0

with a ∈ o(3) and b ∈ R3 . But recall from Eq. (3.20) that we have an identification
of o(3) with R3 . Therefore we can identify the six-dimensional algebra e(3) as a
vector space with

o(3) ⊕ o(3). (3.26)

If [ , ]o(3) denotes the bracket in the Lie algebra o(3) then the Lie bracket in e(3),
which is just the commutator of 4 × 4 matrices restricted to matrices in e(3) as
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch03

20 The Supersymmetric Dirac Equation

above, is

[(a1 , b1 ), (a2 , b2 )] = ([a1 , a2 ]o(3) , [a1 , b2 ]o(3) − [a2 , b1 ]o(3) ). (3.27)

The bracket of two elements of the form (a, 0) is just the o(3) bracket in the a
position, that is

[(a, 0), (b, 0)] = ([a, b]o(3) , 0). (3.28)

The bracket of an a and a b is the o(3) bracket in the b position, that is

[(a, 0), (0, b)] = (0, [a, b]o(3) ). (3.29)

Finally, the bracket of two bs is zero:

[(0, b1 ), (0, b2 )] = 0. (3.30)

o(4)
This is the algebra of all antisymmetric 4 × 4 matrices, so a basis is given by
     
0 0 0 0 0 0 1 0 0 −1 0 0
0 0 −1 0    0 0 0
δ1 =  , δ2 =  0 0 0 0, δ3 = 1 ,
0 1 0 0   −1 0 0 0   0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0
     
0 0 0 1 0 0 0 0 0 0 0 0
 0 0 0 0 0 0 0 1  0 0
ω1 =    , ω3 = 0 0 ,
 0 0 0 0, ω2 = 0 0 0 0   0 0 0 1
−1 0 0 0 0 −1 0 0 0 0 −1 0
(3.31)

and we write a general element of o(4) as ai δi + bi ωi , i = 1, 2, 3, or as


a b
, (3.32)
−bT 0

with a ∈ o(3) and b ∈ R3 . Once again, this six-dimensional Lie algebra can be
identified as a vector space with o(3)⊕ o(3). The bracket of two as is the same as
before, as is the bracket of an a and a b. What is new is the bracket of two bs. In
the algebra o(4) we have

[(0, b1 ), (0, b2 )] = ([b1 , b2 ]o(3) , 0). (3.33)

August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch03

Symmetry of the Classical Problem 21

o(3,1)
This is the algebra of 4 × 4 matrices of the form

a b
, (3.34)
bT 0

with a ∈ o(3) and b ∈ R3 . It can be shown that this is the Lie algebra of the Lorentz
group — the group of all linear transformations of R4 that preserve the quadratic
form

x21 + x22 + x23 − x24 . (3.35)

Once again the bracket between two as and an a and a b is as before. But now

[(0, b1 ), (0, b2 )] = (−[b1 , b2 ]o(3) , 0). (3.36)

In all three cases, e(3), o(4), o(3, 1), the Lie algebra decomposes as a vector
space direct sum

G = k + p, (3.37)

with

[k, k] ⊂ k, [k, p] ⊂ p, and [p, p] ⊂ k. (3.38)

In these three cases k = p = o(3) as vector spaces, and the [k, k] and [k, p] brackets
are the o(3) brackets. The bracket of p back into k is given by λ[ , ]o(3) , where
λ = 1 for o(4), λ = 0 for e(3), and λ = −1 for o(3, 1).
It is notable that in the o(4) case we have k = p as vector spaces, and when we
make this vector space identification all three brackets are the same: The subspaces

∆ = {(a, a)} the diagonal (3.39)

and

∆T = {(b, −b)} the antidiagonal (3.40)

are both subalgebras isomorphic to k and we have the Lie algebra direct sum
decomposition

G = ∆ ⊕ ∆T . (3.41)

Thus o(4) is Lie algebra isomorphic to o(3)⊕ o(3). This identification in Lie theory
has remarkable physical consequences as will become apparent in Chapter 8.
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch03

22 The Supersymmetric Dirac Equation

3.3 Poisson Brackets

The implementation of symmetries is seen most clearly in the Hamiltonian
formulation of classical mechanics. In this formulation an observable is a function
on phase space. In addition to the usual addition and multiplication, the space
of functions on phase space has another binary operation known as the Poisson
bracket: Given two functions, f and g, their Poisson bracket {f, g} is a third
function on phase space, and this operation satisfies the following rules:

(a) {f, g} is bilinear in f and g;

(b) { , } is antisymmetric, that is {g, f } = −{f, g};
(c) Jacobi’s identity: {f, {g, h}} = {{f, g}, h} + {g, {f, h}}; and
(d) f acting on gh satisfies the Leibniz rule: {f, gh} = {f, g}h + g{f, h}.

The first three axioms are just the axioms for a Lie algebra: The space of functions
on phase space form a Lie algebra under the Poisson bracket. Axiom (d) ties the
Poisson bracket in with usual multiplication. It says that the Poisson bracket acts
as a derivation. A commutative algebra A having a bracket satisfying the above
axioms is called a Poisson algebra.
For example, the phase space X for a single classical particle moving in
ordinary three-dimensional space is the six-dimensional space consisting of all
possible positions and momenta. So X is six-dimensional, with coordinates
(x1 , x2 , x3 , p1 , p2 , p3 ). The Poisson brackets of these coordinate functions are

{xi , xj } = {pi , pj } = 0 and {xi , pj } = δij , with i, j = 1, 2, 3. (3.42)

When these bracket relations are realized we say that the coordinates p1 , p2 , p3 are
the canonically conjugate variables to x1 , x2 , x3 . Using the axioms we find

{xi , p2j } = pj {xi , pj } + {xi , pj }pj = 2pj δij , (3.43)

and, by iteration,

{xi , pnj } = npn−1

j δij . (3.44)

Therefore
∂g
{xi , g} = , (3.45)
∂pi
and similarly
∂g
{pi , g} = − , (3.46)
∂xi
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch03

Symmetry of the Classical Problem 23

for any polynomial g = g(x, p). We then have

∂f ∂g ∂f ∂g
{f, g} = − , (3.47)
∂xi ∂pi ∂pi ∂xi
for any pair of polynomials f and g. This is, in fact, the formula for the Poisson
bracket of any pair of functions.
A transformation is called φ symplectic if the induced transformation on
functions
f
→ f ◦ φ−1 , (3.48)

preserves the Poisson bracket, that is, if

{f, g} ◦ φ−1 = {f ◦ φ−1 , g ◦ φ−1 } (3.49)
for all functions f and g. It turns out that every function f on phase space generates
a (local) one-parameter group, φt , of symplectic transformations such that
 
d
{f, g} = g ◦ φ−t . (3.50)
dt t=0
So forming the Poisson bracket with f (on the left) is the infinitesimal symmetry
associated to f . A function on phase space serves a double role: It is an observable,
and it also determines an infinitesimal symmetry of the space of observables via
the Poisson bracket.
Let G be a Lie algebra. We say that we have a Hamiltonian action of G on X if
we are given a Lie algebra homomorphism σ : G
→ A, where A denotes the space
of functions on X. Thus σ assigns to each element ξ ∈ G a function σ(ξ) on X
and

{σ(ξ), σ(η)} = σ([ξ, η]) (3.51)

for every η ∈ G.
For example, consider the case of the phase space of a single free particle
as above. Let G be the Lie algebra consisting of all 3 × 3 matrices under the
commutator bracket, and define
σ(M ) = −Mij xi pj . (3.52)

Equation (3.51) holds, since


∂(xi pj ) ∂(xm pn ) ∂(xi pj ) ∂(xm pn )
{σ(ξ), σ(η)} = Mij Nmn −
∂xk ∂pk ∂pk ∂xk
= Mij Nmn (δik δnk pj xm − δjk δmk xi pn )
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch03

24 The Supersymmetric Dirac Equation

= Mij Nmi pj xm − Mim Nmn xi pn

= (N M )ij xi pj − (M N )ij xi pj = −[M, N ]ij xi pj = σ([ξ, η]).

(3.53)

Notice also that

{σ(M ), xk } = Mik xi and {σ(M ), pk } = −Mkj pj . (3.54)

So if we identify the function xk with the kth basis element in R3 then the first
equation in (3.54) identifies the bracket with σ(M ) with the matrix Mij and so
σ defines an isomorphism between the Lie algebra gl(3, R) and the subalgebra of
A consisting of all homogeneous polynomials of degree two, which are of degree
one in x and degree one in p. For example, if L3 = x1 p2 − x2 p1 then

{L3 , x1 } = x2 , {L3 , x2 } = −x1 , {L3 , x3 } = 0, (3.55)

and

{L3 , p1 } = p2 , {L3 , p2 } = −p1 , {L3 , p3 } = 0. (3.56)

The first of these equations says that L3 acts on r as an infinitesimal rotation about
the x3 -axis. Let o(3)∈ gl(3, R) be the Lie algebra of infinitesimal rotations and let
L denote the restriction of σ to o(3). Recall that δ3 denotes infinitesimal rotation
about the x3 -axis (with similar notation for δ1 and δ2 ) and then L3 = L(δ3 ).
Let us use × to denote the vector product on R3 . We have made the
identification of o(3) with R3 so that the Lie bracket is given by ×. That is, the
elements of o(3) are regarded as vectors and the Lie bracket becomes the vector
product. The elements of δi then become the standard basis vectors. If we form the
vector-valued function

L=r×p (3.57)

then under the identification of o(3) with R3 we have, for any ξ ∈ R3 ,

L(ξ) = L · ξ, (3.58)

and (3.51) becomes

L([ξ, η]) = {L(ξ), L(η)}. (3.59)

The function L is called the angular momentum.

The time evolution of a mechanical system is the one-parameter group
generated by the Hamiltonian H. The infinitesimal change in any observable f
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch03

Symmetry of the Classical Problem 25

under this time evolution is thus given by {H, f }. We say that H is invariant under
the action σ of the Lie algebra G if

{σ(ξ), H} = 0 (3.60)

for all ξ ∈ G. This, by the antisymmetry of { , }, implies that {H, σ(ξ)} = 0. In

other words, all the functions σ(ξ) are conserved quantities under the flow generated
by H. Thus if H is rotationally invariant, this means that all the functions L(ξ),
or, what amounts to the same thing, each of the functions L1 , L2 , L3 are constant
along the trajectories of H. In other words, the vector-valued function L is constant.
This is the principle of conservation of angular momentum.
Infinitesimal translation in the x1 direction acts on phase space by the Poisson
bracket with p1 . Indeed,

{p1 , x1 } = −1, {p1 , x2 } = {p1 , x3 } = {p1 , pi } = 0, i = 1, 2, 3. (3.61)

Thus invariance of the Hamiltonian under the full translation group implies
conservation of the vector-valued function p. This is the principle of the
conservation of linear momentum.
Thus invariance of the Hamiltonian under the full group of Euclidean motions
containing all translations and rotations implies conservation of both linear and
angular momentum. One of Galileo’s principal revolutionary achievements was to
replace the geocentric Aristotelian theory, which has an O(3) symmetry centered
at the origin of the Earth, with the larger symmetry group E(3). Indeed, it is easy
to check that the map

(a, b)
→ a · L + b · p (3.62)

gives a Hamiltonian action of e(3) on phase space.

3.4 The Inverse Square Law

In the problem of a particle moving under the influence of a central force we no
longer have translational symmetry, but we retain rotational symmetry about the
origin. We do not have {H, p} = 0, but if the force is inversely proportional to the
square of the distance we have the three conserved quantities A of Eq. (2.22) in
addition to the quantities L. In the present formalism a quantity is conserved if its
Poisson bracket with the Hamiltonian vanishes. We check that this is indeed the
case. With
p2 k
H= − , (3.63)
2m r
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch03

26 The Supersymmetric Dirac Equation

we have
1 1
{H, Ai } = − ijk {1/r, pj }Lk − pj {pj , xi /r}
m m
1
= (ijk xj Lk /r 3 + pi /r − xi pj xj /r3 ), (3.64)
m

Using Eq. (3.46), since {Li , f (r)} = 0 for f an arbitrary function of r. The first
term is
1 1
ijk ksn xj xs pn /r3 = (δis δjn − δin δjs )xj xs pn /r 3 , (3.65)
m m
which cancels the second and third terms. We are using

Li = ijk xj pk . (3.66)

Exercise 3.1 Prove ijk imn = δjm δkn − δjn δkm .

Exercise 3.2 Prove ijk ijm = 2δkm .

We want to see if these six quantities close to a Lie algebra. We still have

{L(ξ), L(η)} = L([ξ, η]) (3.67)

as before. We also have

{L(ξ), A(η)} = A([ξ, η]), (3.68)

which just says that A transforms as a vector. Indeed,

1
{Li , Aj } = jsn {Li , ps Ln } − {Li , xj /r}
mk
1 1
= jsn ({Li , ps }Ln + ps {Li , Ln }) − {Li , xj }
mk r

(p × L)k xk
= ijk − = ijk Ak , (3.69)
mk r

We now wish to compute {Ai , Aj }. It is a somewhat lengthy calculation,

but we go through it in detail because of the importance of the result, which will
be needed in Chapter 8. We first calculate some Poisson brackets, which are used
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch03

Symmetry of the Classical Problem 27

below:

(mk){pi , Aj } = jsn {pi , ps Ln } − mk{pi , xj /r}

= jsn ps {pi , Ln } − mk{pi , xj }/r − mkxj {pi , 1/r}
= −jsn niq ps pq + mkδij /r − mkxj xi /r 3
= (δjq δsi − δji δsq )ps pq + mkδij /r − mkxi xj /r 3
= pi pj − p2 δij + mkδij /r − mkxi xj /r 3 , (3.70)

and

(mk){xi /r, Aj } = jsn {xi /r, ps Ln } = jsn ({xi , ps Ln }/r + {1/r, ps Ln }xi )
= jsn (ps {xi , Ln }/r + {xi , ps }Ln /r + {1/r, ps }Ln xi )
= jsn (−nik ps xk /r + δis Ln /r − xs xi Ln /r 3 )
= −(δji δsk − δjk δsi )ps xk /r + jin Ln /r − jsn xs xi Ln /r 3
= pi xj /r − δij xm pm /r − ijm Lm /r − jmn xi xm Ln /r3 .
(3.71)

The latter yields, remembering that r · L = 0, and using Exercises 3.1 and 3.2,

(mk)2 ijk {xi /r, Aj } = −2mkLk /r. (3.72)

We are now ready to do the main calculation:

(mk)2 ijk {Ai , Aj } = (mk)ijk (isn {ps Ln , Aj } − (mk){xi /r, Aj })

= (mk)ijk (isn (ps {Ln , Aj } + {ps , Aj }Ln )
− (mk){xi /r, Aj }). (3.73)

We evaluate the separate terms:

ijk isn ps {Ln , Aj } = ijk isn njq ps Aq

= ijk (δij δsq − δiq δsj )ps Aq = ijk pi Aj , (3.74)

and

(mk)ijk pi Aj = ijk jsq pi ps Lq − (mk)ijk pi xj /r

= (δiq δks − δis δkq )pi ps Lq + (mk)Lk /r
= (−p2 + mk/r)Lk . (3.75)
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch03

28 The Supersymmetric Dirac Equation

Also

(mk)ijk isn {ps , Aj }Ln = ijk isn (ps pj − p2 δsj + mkδsj /r

− mkxs xj /r3 )Ln
= (δjs δkn − δjn δks )(ps pj − p2 δsj + mk/rδsj
− mkxs xj /r3 )Ln
= (−2p2 + 2mk/r)Lk − (−p2 + mk/r)Lk
= (−p2 + mk/r)Lk . (3.76)

Combining these results, we get

(mk)2 ijk {Ai , Aj } = [(−p2 + mk/r) + (−p2 + mk/r) + 2mk/r]Lk

= −2(p2 − 2mk/r)Lk = −4mHLk . (3.77)

Multiply by ijk to finally get

(mk)2 ijk mnk {Am , An } = 2(mk)2 {Ai , Aj } = −4mijk HLk , (3.78)

and

−2H
{Ai , Aj } = ijk Lk . (3.79)
mk2
The occurrence of H(r, p) in the last equation prevents the quantities Li and
Ai from closing to a Lie algebra on the whole phase space. But when the motion
is in a fixed orbit H(r, p) = E, with E a constant energy. We can then divide the
phase space into three sets, the two open regions where H > 0 and H < 0, and
their common boundary where H = 0. In the region where H > 0 we can rescale
A by 2H/mk 2 to get an o(3, 1) algebra. In the region where H < 0 we can rescale
by −2H/mk2 to get an o(4) algebra.
That is, we define

A = N A, (3.80)

where
2H
N −2 = − , (3.81)
mk 2
to get

{Li , Lj } = ijk Lk , {Li , Aj } = ijk Ak , {Ai , Aj } = ijk Lk . (3.82)

This shows that the quantities Li and Ai close to form an o(4) Lie algebra.
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch03

Symmetry of the Classical Problem 29

We can use the Lie algebra isomorphism of o(4) with o(3)⊕ o(3) to form the
direct sum of two o(3) algebras. Define
Mi = Li + Ai , Ni = Li − Ai . (3.83)

Then the Poisson structure becomes

{Mi , Mj } = ijk Mk , {Ni , Nj } = ijk Nk , {Mi , Nj } = 0. (3.84)

We have shown that a force which is inversely proportional to the square

of the distance between the particles leads, in the case of bound states, to six
conserved quantities which form, when properly normalized, the generators of an
so(4) algebra that acts on the functions of phase space. This result, which appears
here in a somewhat abstract form, gives rise to physical results when applied to the
spectrum of hydrogen in Chapter 8.

Notes on Chapter 3
This review of Hamiltonian mechanics is standard, and can be found in any
good text on classical mechanics. It is included here to make the treatment self-
contained. The same is true for the relation between Lie groups and Lie algebras.
More information can be found in Weaver and Sattinger [WS10], or Marsden and
Ratiu [MR99]. We follow here the treatment of Guillemin and Sternberg [GS90].
The calculation of the Poisson bracket between the components of the Laplace
vector is the classical version of the calculation of the corresponding commutator
by Sudbery [Sud86].
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch03

This page intentionally left blank

August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch04

Chapter 4
From Solar Systems to Atoms

4.1 Rutherford Scattering

“It was quite the most incredible event that has ever happened to me in my life. It was almost
as incredible as if you fired a 15-inch shell at a piece of tissue paper and it came back and
hit you.”

Lord Rutherford, quoted in Rutherford and the Nature of the Atom, E. N. da C. Andrade,
Heinemann, London, 1964.

In 1897 J. J. Thompson measured the mass-to-charge ratio of the electron. In

1902 Lord Kelvin proposed that the atom consisted of a sphere of positive charge
in which the electrons were embedded. In 1906 W. Thompson reasoned that the
number of electrons in an atom was approximately equal to the atom’s atomic
weight. On this basis, there would be only one electron in the hydrogen atom.
In 1911 Rutherford [Rut11] established, from the data of scattering
experiments with alpha particles impinging on gold folie, that the positive charge
in the atom is concentrated in a nucleus 10,000 times smaller than the atom itself.
This nucleus also carries most of the mass of the atom. This was the background
on which Bohr [Boh13] developed his model of the hydrogen atom in 1913. The
electron and its nucleus form a two-body system, with an electron mass that is
negligible in comparison to the mass of the nucleus. Hence one has a point electron
moving in the Coulomb potential of a nucleus of charge one.
To understand Rutherford’s reasoning we shall derive his formula for
scattering. In this derivation we shall again make use of the Laplace vector of
the system. We do this in order to get some practice with the application of the
Laplace vector, and also because the formula is of paramount importance on the
road to understanding atomic physics.

31
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch04

32 The Supersymmetric Dirac Equation

4.2 Conservation of the Laplace Vector

We consider impinging particles of charge e, and as target a nucleus of charge Ze.
The central force is repulsive, and k in Eq. (2.17) is negative, k = −|k| = −Zα.
Then the Laplace vector is

(p × L)
A=− − r̂. (4.1)
Zαm
The scattering process is described as in Figure 4.1. Long before the collision
takes place the initial impulse of a particle in the beam is pi = mv0 ei , where v0
is the velocity of the incoming particle and ei is a unit vector in the direction of
the incoming velocity. The final velocity is pf = mv0 ef , with the same velocity
v0 (the process is elastic) and ef is a unit vector in the direction of the outgoing
velocity. Li = Lf = mv0 bk̂, where L is the conserved angular momentum, and
k̂ is a unit vector pointing up through the page, perpendicular to the scattering
plane. b is the impact parameter, the distance that the incoming particle would
pass by the point target in the absence of interaction. The initial incoming direction
is r̂i = −ei , and the final outgoing direction r̂f = ef .
The initial Laplace vector is


mv02
Ai = − bni + ei , (4.2)
Zα
e
f
n
f

b
v
f

ni

ei
m

r
b
v Θ
c

Figure 4.1. The incoming and outgoing directions and the scattering angle.
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch04

From Solar Systems to Atoms 33

where ni is the unit vector in Figure 4.1 perpendicular to ei and in the scattering
plane. The final Laplace vector is


mv02
Af = − bnf − ef , (4.3)
Zα

where nf is the unit vector perpendicular to ef and in the scattering plane.

Conservation of the Laplace vector yields

Ai ·ei = Af ·ei . (4.4)

With (ef ·ei ) = cos θ, where θ is the scattering angle, and (nf ·ei ) = cos (π/2 + θ)
= − sin θ, this leads to

Zα = mv02 b sin θ − Zα cos θ. (4.5)

Rearranging this gives a relation between b and θ,

Zα
tan (θ/2) = , (4.6)
mv02 b

and this enables us to calculate the scattering angle for any impact parameter.

4.3 The Differential Cross Section

The differential scattering cross section is defined in terms of the number of particles
scattered into the shaded band 2πdθ in Figure 4.2. From the figure this is the same
as the number of particles incident on the area 2πb db, and the differential cross

b Θ

db dΘ

Figure 4.2. Scattering of an incident beam of particles by a center of force.

August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch04

34 The Supersymmetric Dirac Equation

section is
 
dσ b  db 
=  . (4.7)
dΩ sin θ  dθ 

We calculate, from Eq. (4.6),

 
 db  Zα
 = (4.8)
 dθ  2mv 2 sin2 (θ/2) .
0

Therefore we obtain Rutherford’s scattering formula:

dσ (Zα)2
= . (4.9)
dΩ 4(mv0 )2 sin4 (θ/2)
2

The result for the Thompson atom, which involves multiple scattering,
decreases much more rapidly then the characteristic sin−4 (θ/2) of Rutherford’s
formula. This is the reason for Rutherford’s surprise, noted earlier. Together with his
associates, Geiger and Marsden, he verified the formula in great detail, including
its proportionality to Z 2 , and the inverse squared proportionality to the kinetic
energy of the incoming particles. The formula was also found to be accurate down
to small distances, and the finite size of the nucleus, at which deviations from the
formula were observed, was around 10−14 m = 10 Fermi for an aluminium foil.
The success of the formula proved the correctness of the assumptions on which it
is based, and established the solar system model for atoms.

Notes on Chapter 4
Rutherford scattering is analyzed using the Laplace vector by Basano and Bianchi
[BB80], and in textbooks on classical mechanics; see, for example, Poole, Goldstein
and Safko [PGS01].
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch05

Chapter 5
The Bohr Model

“The spectrum of the hydrogen atom has proved to be the Rosetta stone of modern physics.”

Theodor W. Hänsch, Arthur L. Schawlow, and George W. Series, in Rev. Mod. Phys. 54,
p. 697 (1982).

5.1 Spectroscopic Series

Balmer published in 1885 an empirical formula for the frequencies of the spectral
lines of the hydrogen spectrum in the visible region,


1 1
hν = R∞ − 2 , (5.1)
4 N
with N = 3, 4, 5, 6. We have used here a modern notation for the constants, with
h the constant later discovered by Planck [Pla00], ν the frequency of the light
emitted, and R∞ the Rydberg constant.
This provided the key for the Bohr model of the hydrogen atom. Subsequently
the Balmer formula was generalized to further series, these were found after Bohr
had proposed his model; see Table 5.1 and Figure 5.1.

5.2 The Postulates of the Model

In 1913 Niels Bohr presented a model of the hydrogen atom [Boh13], which
quantitatively explained the spectroscopic measurements to the accuracy which
was obtainable at that time. He based his model on the following postulates:
1. An electron in a hydrogen atom moves under the influence of an attractive
Coulomb potential between the electron and the nucleus according to the laws
of classical mechanics (i.e., in particular in circular orbits).
2. Quantization condition: Instead of the infinity of orbits possible in classical
mechanics, it is only possible for an electron to move in an orbit for which its
orbital angular momentum L is an integral number.
35
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch05

36 The Supersymmetric Dirac Equation

Table 5.1. The hydrogen series

Names Wavelength ranges Formulas


Lyman Ultraviolet ν = R∞ 1
2 − N12 N =2, 3, 4,. . .

11 
Balmer Near Ultraviolet and Visible ν = R∞ 2 − N12 N = 3, 4, 5,. . .

2 
Paschen Infrared ν = R∞ 312 − N12 N = 4, 5, 6,. . .
 
Brakett Infrared ν = R∞ 412 − N12 N = 5, 6, 7,. . .


Pfund Infrared ν = R∞ 512 − N12 N = 6, 7, 8,. . .

Figure 5.1. The spectral series of hydrogen.

3. Despite the fact that it is constantly accelerating, an electron moving in such an

allowed orbit does not radiate electromagnetic energy through bremsstrahlung.
Thus, its total energy E remains constant.
4. Electromagnetic radiation is emitted if an electron, initially moving in an orbit
of total energy Ei , discontinuously changes its motion so that it moves in an
orbit of total energy Ef . The frequency of the emitted radiation ν is equal to
the quantity (Ei − Ef ) divided by Planck’s constant 2π(= 2π
).

The first postulate bases the model on the picture introduced by Rutherford.
The second postulate introduces quantization of orbital angular momentum of an
atomic electron moving under the influence of an inverse square (Coulomb) force:

L = N, N = 1, 2, 3, . . . (5.2)

Planck’s quantization condition for the energy of a particle, such as an electron,

executing simple harmonic motion under the influence of a harmonic restoring
force, was E = 2πN ν, N = 0, 1, 2, 3, . . . , which is at first sight different from
Bohr’s condition. The third postulate removes the problem of the instability of an
electron moving in a circular orbit, due to the emission of the electromagnetic
radiation required of the electron by classical electromagnetic theory, by
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch05

The Bohr Model 37

N=1

N=2
N=3

Figure 5.2. The Bohr model of the hydrogen atom.

postulating that this particular feature of the classical theory is not valid for the case
of an atomic electron. The postulate was based on the fact that atoms are observed
by experiment to be stable — even though this is not predicted by classical theory.
The fourth postulate, 2πν = Ei − Ef , is really just Einstein’s postulate that the
frequency of a photon of electromagnetic radiation is equal to the energy carried
by the photon divided by Planck’s constant; see Figure 5.2.

5.3 The Predictions of the Model

Consider an atom consisting of a nucleus of charge +Ze and mass taken to
be infinite, compared to the mass of the electron. For a neutral hydrogen atom
Z = 1, for a singly ionized helium atom Z = 2, for a doubly ionized lithium
atom Z = 3, etc. We assume that the electron revolves in a circular orbit about
the nucleus, which, being infinitely heavy, remains fixed in space. The condition
of mechanical stability of the electron is

Zα mv 2
2
= , (5.3)
r r
where α = e2 /4π is the fine structure constant, v is the speed of the electron in
its orbit, and r is the radius of the orbit. The left-hand side of this equation is the
Coulomb force acting on the electron, and the right-hand side is ma, where a is
the centripetal acceleration of the electron in its circular orbit.
Now, the orbital angular momentum of the electron, L = mvr, is a constant,
because the force acting on the electron is entirely in the radial direction. Applying
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch05

38 The Supersymmetric Dirac Equation

Figure 5.3. The electron in the ground state of the Bohr atom. (a) Classical picture; (b) Quantum
Mechanical picture.

the quantization condition, Eq. (5.2), to L, we get

mvr = N, N = 1, 2, 3, . . . (5.4)

Solving for v and substituting into Eq. (5.3), we obtain

2
N N2
Zα = mv 2 r = mr = , (5.5)
mr mr
so
N2
r= , N = 1, 2, 3, . . . (5.6)
Zαm
and
N Zα
v= = , N = 1, 2, 3, . . . (5.7)
mr N
The application of the angular momentum quantization condition has restricted
the possible circular orbits to those of radii given by Eq. (5.6). Note that these radii
are proportional to the square of the quantum number N . If we evaluate the radius
of the smallest orbit (N = 1) for a hydrogen atom (Z = 1) by inserting the known
values of m and α, we obtain r = a0 = 1/αme ≈ 0.5 Å, the Bohr radius. We
shall show later that the electron has its minimal total energy when in the orbit
corresponding to N = 1. Consequently we may interpret the radius of this orbit
as a measure of the radius of the hydrogen atom in its ground state. It is in good
agreement with the estimate that the order of magnitude of an atomic radius is 1 Å.
Hence, Bohr’s postulate predicts a reasonable size for the atom; see Figures 5.3
and 5.4.
Evaluating the orbital velocity of an electron in the ground state of hydrogen,
we find v = 2.2×106 m/sec. It is apparent from the equation that this is the largest
velocity possible for a hydrogen atom electron. The fact that this velocity is less than
1% of the velocity of light is the justification for using non-relativistic mechanics
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch05

The Bohr Model 39

Figure 5.4. The probability of finding the electron at a given distance from the nucleus.

in the Bohr model. On the other hand, Eq. (5.7) shows that for large values of Z
the electron velocity becomes relativistic, and the model cannot be applied.
Next we calculate the total energy of an atomic electron moving in one of
the allowed orbits. Let us define the potential energy to be zero when the electron
is at rest infinitely distant from the nucleus. Then the potential energy V at any
finite distance r can be obtained by integrating the work that would be done by the
Coulomb force acting from r to ∞. Thus
 ∞
Zα Zα
V = 2
dr = − . (5.8)
r r r
The potential energy is negative because the Coulomb force is attractive; it takes
work to move the electron from r to ∞ against this force. The kinetic energy of
the electron, Ekin , can be evaluated, with the aid of Eq. (5.3), to be
1 Zα
Ekin = mv 2 = . (5.9)
2 2r
The total energy of the electron, E, is then
Zα
E = Ekin + V = − = −Ekin , (5.10)
2r
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch05

40 The Supersymmetric Dirac Equation

E [eV]

0 N=∞
-0.85 N=4
-1.51 N=3

-3.39 N=2

-13.6 N=1

Figure 5.5. Energy-level diagram of the Bohr model.

in agreement with the virial theorem. Using Eq. (5.6) for r in the preceding equation,
we have
 m  (Zα)2
E=− , N = 1, 2, 3, . . . (5.11)
2 N2
We see that the quantization of the orbital angular momentum of the electron leads
to a quantization of its total energy.
The information contained in Eq. (5.11) is presented as an energy-level
diagram in Figure 5.5. The energy of each level, as evaluated from Eq. (5.11) is
shown on the left, in terms of electron volts, and the quantum number of the level is
shown on the right. The diagram is so constructed that the distance from any level
to the level of zero energy is proportional to the energy of that level. Note that the
lowest allowed total energy occurs for the smallest quantum number N = 1. As
N increases, the total energy of the quantum states becomes less negative, with E
approaching zero as N approaches infinity. Since the state of lowest total energy
is the ground state, the normal state of the electron in a one-electron atom is the
state for which N = 1.
The energy binding the electron to the nucleus may be calculated from
Eq. (5.11). The binding energy is numerically equal to the energy of the ground
state, corresponding to N = 1. This yields, with Z = 1, E = −(m/2)α2 =
−13.6 eV , which agrees very well with the experimentally observed binding
energy for hydrogen. Next we calculate the frequency ν of the electromagnetic
radiation emitted when the electron makes a transition from the quantum state Ni
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch05

The Bohr Model 41

to the quantum state Nf , that is, when an electron initially moving in an orbit
characterized by the quantum number Ni discontinuously changes its motion so
that it moves in an orbit characterized by quantum number Nf . Using Bohr’s fourth
postulate, Eq. (5.2), and Eq. (5.11), we have
   
Ei − E f Z 2 α2 m 1 1 1 1
ν= = − 2 = R∞ Z 2 − 2 , (5.12)
2π 4π Nf2 Ni Nf2 Ni

with
α2 m
R∞ = , (5.13)
4π
and where Ni and Nf are integers.
The essential predictions of the Bohr model are contained in Eqs. (5.11) and
(5.12). We discuss the emission of electromagnetic radiation by a one-electron
Bohr atom in terms of these equations:

1. The atom is normally in its ground state N = 1.

2. In an electric discharge, or in some other process, the atom receives energy
due to collisions, etc. This means that the electron must make a transition to an
excited state, in which N > 1.
3. Obeying the common tendency of all physical systems, the atom will emit its
excess energy and return to the ground state. This is accomplished by a series
of transitions in which the electron drops to excited states of successively lower
energy, finally reaching the ground state. In each transition electromagnetic
radiation is emitted with a wavelength that depends on the energy lost by the
electron, i.e., on the initial and final quantum numbers.
4. In the very large number of excitation and de-excitation processes that take place
during a measurement of an atomic spectrum, all possible transitions occur and
the complete spectrum is emitted. The reciprocal wavelengths of the set of lines
that constitute the spectrum are given by Eq. (5.12), where we allow Ni and
Nf to take on all possible integral values subject only to the restriction that
Ni > Nf .

According to the Bohr model, each of the five known series of the hydrogen
spectrum arises from a subset of transitions in which the electron goes to a certain
final quantum state Nf . For the Lyman series Nf = 1, for the Barmer series
Nf = 2, for the Paschen series Nf = 3, for the Brakett series Nf = 4, and for the
Pfund series Nf = 5. The wavelength of the lines of all these series are fitted very
accurately by Eq. (5.12) by using the appropriate value of Nf . This was a great
triumph of the Bohr model. The success of the model was particularly impressive
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch05

42 The Supersymmetric Dirac Equation

because the Lyman, Brakett and Pfund series had not been discovered at the time the
model was developed by Bohr. The existence of these series was predicted, and the
series were soon found experimentally by the persons after whom they were named.
The model worked equally well when applied to the case of one-electron atoms
with Z = 2, i.e., singly ionized helium atoms He+. Such atoms can be produced
by passing a spark through normal helium gas. They make their presence apparent
by emitting a simpler spectrum than that emitted by normal helium atoms. In fact,
the spectrum of He+ is exactly the same as the hydrogen spectrum except that the
reciprocal wavelengths of all the lines are almost exactly four times as great. This is
explained very easily, in terms of the Bohr model, by setting Z 2 = 4 in Eq. (5.11).
The properties of the absorption spectrum of one-electron atoms are also easy
to understand in terms of the Bohr model. Since the atomic electron must have a
total energy exactly equal to the energy of one of the allowed energy states, the
atom can only absorb discrete amounts of energy from the incident electromagnetic
radiation. This fact leads to the idea that we consider the incident radiation to be a
beam of photons, and that only those photons can be absorbed whose frequency is
given by E = 2πν, where E is one of the discrete amounts of energy which can
be absorbed by the atom. The process of absorbing electromagnetic radiation is
then just the inverse of the normal emission process, and the lines of the absorption
spectrum will have exactly the same wavelengths as the lines of the emission
spectrum. Normally the atom is always initially in the ground state N = 1, so
only absorbtion processes from N = 1 to N > 1 can occur. Thus, the only
absorption lines which correspond (for hydrogen) to the Lyman series will normally
be observed. However, if the gas containing the absorbing atoms is at a very high
temperature, then, owing to collisions, some of the atoms will initially be in the
first excited state N = 2, and absorption lines corresponding to the Balmer series
will be observed. Balmer absorbtion lines are actually observed in the hydrogen
gas of some stellar atmospheres. This gives us a way of estimating the temperature
of the surface of a star.

5.4 Correction for Finite Nuclear Mass

In the previous section (5.3) we assumed the mass of the atomic nucleus to be
infinitely large compared to the mass of the atomic electron, so that the nucleus
remains fixed in space. This is a good approximation even for hydrogen, which
contains the lightest nucleus, since the mass of the nucleus is about 2000 times
larger than the electron mass. However, the spectroscopic data are so very accurate
that before we make a comparison of these data with the Bohr model we must
take into account the fact that the nuclear mass is actually finite. In such a case the
electron and the nucleus move about their common center of mass. However, it is not
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch05

The Bohr Model 43

difficult to show that in such a system the electron moves relative to the nucleus as
though the nucleus were fixed and the mass of the electron were slightly reduced to
the value µ, the reduced mass of the system. The equations of motion of the system
are the same as those we have considered if we simply substitute µ for m, where
mM
µ= , (5.14)
m+M
is less than m by a factor 1/(1 + m/M ). Here M is the mass of the nucleus.
To handle this situation Bohr modified his second postulate, Eq. (5.4), to
require
µvr = N, N = 1, 2, 3, . . . (5.15)
Using µ instead of m in this equation takes into account the angular momentum
of the nucleus as well as that of the electron. All the equations are identical with
those derived before, except that the electron mass m is replaced by the reduced
mass µ. In particular, the formula for the frequencies of the spectral lines becomes
 
1 1
ν = RM Z 2 − 2 , (5.16)
Nf2 Ni

where
µ
RM = R∞ . (5.17)
m
The quantity RM is the Rydberg constant for a nucleus of mass M . For the most
extreme case of hydrogen RM is less than R∞ by about one part in 2000. If we
evaluate RH from Eq. (5.17) we find that the Bohr model, corrected for the finite
nuclear mass, agrees with the experimental data to within three parts in 100,000.
Atoms that follow the same pattern as hydrogen, but with different values
of the parameters, are called hydrogenic atoms. They include systems in which
an electron or the proton are replaced by an exotic particle (see below). They also
include Rydberg atoms, an ordinary atom in which one electron has been elevated to
a very high quantum state, and highly ionized atoms; that is, atoms whose electrons
have been stripped away, leaving only one electron in an orbit around the nucleus.
In Rydberg atoms the methods developed here serve only as approximations, in
which the interactions between the lower-lying electrons are neglected.
An example of an exotic atom is positronium, which consists of a positron
and an electron, revolving about their common center of mass, which lies halfway
between them. The reduced mass is µ = m/2. The corresponding Rydberg constant
is RM = R∞ /2. The frequencies of the emitted spectral lines of positronium are
given by
 
R∞ 1 1
ν= − 2 . (5.18)
2 Nf2 Ni
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch05

44 The Supersymmetric Dirac Equation

The frequencies of the emitted lines are then half those of a hydrogen atom
(with infinitely heavy nucleus), Z being equal to one for positronium and for
hydrogen. The radius of the ground state of positronium is
1
rpos = = 2a0 . (5.19)
αµ
Hence, for any quantum state N the radius of the electron is twice as great in the
positronium atom as in the hydrogen atom (with infinitely heavy nucleus).
A muonic atom contains a nucleus of charge Zα and a negative muon, µ− ,
moving about it. The µ− is an elementary particle with charge −e and a mass that
is about 207 times as large as an electron mass. Such an atom is formed when a
proton, or some other nucleus, captures a µ− . The reduced mass of the system is
µ = 186m. Then, from Eq. (5.6), with N = Z = 1, and µ = 186m, we obtain
1
r1 = = 2.8 × 10−3 Å. (5.20)
186mα
Therefore the µ− is much closer to the nuclear (proton) surface than is an electron in
a hydrogen atom. It is this feature that makes muonic atoms interesting, information
about nuclear properties being revealed from their study. The binding energy of a
muonic atom with Z = 1 is calculated from Eq. (5.11), with N = 1 and µ = 186 m;
the result is E = −2530 eV . The wavelength of the first line in the Lyman series is


1
ν = RM 1 − , (5.21)
4

with RM = (µ/m)R∞ , so λ = 1/ν = 6.5 Å, so that the Lyman lines lie in the
x-ray part of the spectrum. X-ray techniques are therefore necessary to study the
spectrum of muonic atoms.
Ordinary hydrogen contains about one part in 6000 of deuterium, or heavy
hydrogen. This is a hydrogen atom whose nucleus contains a proton and a neutron.
The spectrum would be identical to that of hydrogen were it not for the correction
of finite nuclear mass. For a normal hydrogen atom

109737 cm−1
RH =  1
= 109678 cm−1 . (5.22)
1 + 1836

For an atom containing deuterium

109737cm−1
RD =   = 109707 cm−1 . (5.23)
1
1 + 2×1836
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch05

The Bohr Model 45

Hence, RD is a bit larger than RH , so that the spectral lines of the deuterium atom
are shifted to slightly lesser wavelengths compared to hydrogen. Indeed, deuterium
was discovered in 1932 by H. C. Urey following the observation of these shifted
spectral lines. By increasing the concentration of the heavy isotope above its normal
value in a hydrogen discharge tube, one can enhance the intensity of the deuterium
lines which, ordinarily, are difficult to detect. One then readily observes pairs of
hydrogen lines; the shorter wavelength members of the pair correspond exactly to
those predicted from RD . The resolution is easily obtained, the Hα -line pair being
separated by about 1.8 Å, for example, several thousand times greater than the
minimal resolvable separation.

Notes on Chapter 5
Information on Niels Bohr and the Bohr model can be found in Hydrogen: The
Essential Element, by John Rigden [Rig03]. Also in that book are further details
on hydrogenic atoms. Further recommended reading is The Infancy of Atomic
Physics: Hercules in His Cradle, by Alex Keller [Kel83]. The presentation here
follows Eisberg and Resnick [ER74].
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch05

This page intentionally left blank

August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch06

Chapter 6
Interpretation of the Quantum Rules

6.1 The Sommerfeld–Wilson Quantization Conditions

The success of the Bohr model, as measured by its agreement with experiment, was
at the time certainly very striking; but it only accentuated the mysterious nature of
the postulates on which the model was based. One of the biggest mysteries was the
question of the relation between Bohr’s quantization of the angular momentum of
an electron moving in a circular orbit and Planck’s quantization of the total energy
of an entity, such as an electron, executing simple harmonic motion. In 1916 some
light was shed upon this question by Wilson [Wil15], and by Sommerfeld [Som16],
who enunciated a set of rules for the quantization of any physical system for which
the coordinates are periodic functions of time. These rules included both the Planck
and the Bohr quantization as special cases. They were also of considerable use in
broadening the range of applicability of the quantum theory. These rules can be
stated as follows:
For any physical system in which the coordinates are periodic functions of time,
there exists a quantum condition for each coordinate. The quantum conditions are


pq dq = 2πNq , (6.1)

where q is one of the coordinates, pq is the conjugate momentum associated with 

that coordinate, Nq is a quantum number that takes on integral values, and
means that the integration is taken over one complete period of the coordinate q.
The momentum conjugate to a coordinate is in the sense of Hamiltonian mechanics,
namely {q, p} = 1, where {A, B} is the Poisson bracket of the two quantities.
The meaning of these rules can best be illustrated in terms of specific examples.
Consider a one-dimensional simple harmonic oscillator. Its total energy can be
written, in terms of position and momentum, as
kx2
E = Ekin + , (6.2)
2

47
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch06

48 The Supersymmetric Dirac Equation

or
p2x x2
+ = 1. (6.3)
2mE 2E/k

The quantization integral px dx is most easily evaluated, for the relation between
px and x that is imposed by this equation, if we consider a geometric interpretation.
The relation between px and x is the equation of an ellipse. Any instantaneous state
of motion of the oscillator is represented by some point in a plot of this equation on a
two-dimensional space having coordinates px and x. Such a space (the p − q plane)
is the phase space, and the plot is a phase diagram of the linear oscillator. During
one cycle of oscillation the point representing the position and the momentum of
the particle travels once around the ellipse. The semiaxes a and b of the ellipse
p2x /b2 + x2 /a2 = 1 are seen, by comparison with Eq. (6.3), to be
√ 
b = 2mE and a = 2E/k. (6.4)

Now the area of an ellipse is πab. The value of the integral px dx is just equal to
that area. Thus we obtain


px dx = πab. (6.5)

In our case


2πE
px dx =  , (6.6)
k/m
but

k/m = 2πν, (6.7)

where ν is the frequency of the oscillation, so that

E
px dx = . (6.8)
ν
If we now use Eq. (6.1), the Sommerfeld–Wilson quantization rule, we have


E
px dx = = 2πNx = 2πN, (6.9)
ν
or

E = 2πN ν, (6.10)

which is identical with Planck’s quantization law.

August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch06

Interpretation of the Quantum Rules 49

Note that the allowed states of oscillation are represented by a series of ellipses
in phase space, the area enclosed between successive ellipses always being 2π =
2π
= h. We find that the classical situation corresponds to 2π
→ 0, allvalues
of E and hence all ellipses being allowed if that were true. The quantity px dx
is called a phase integral; in classical physics it is the integral of the dynamical
quantity called the action over one oscillation of the motion. Hence, the Planck
energy quantization condition is equivalent to the quantization of the action.
The full quantum mechanical treatment of course gives E = 2πν(N + 12 ), so
that the Sommerfeld–Wilson quantization condition does not give the zero-point
energy correctly.
We can also deduce the Bohr quantization of angular momentum from the
Sommerfeld–Wilson rule, Eq. (6.1). An electron moving in a circular orbit of
radius r has an angular momentum, mvr = L, which is constant. The angular
coordinate is θ, which is a periodic function of time. That is, θ versus t is a saw-
tooth function, increasing linearly from zero to 2π rad in one period and repeating
this pattern in each succeeding period. The quantization rule


pq dq = 2πNq , (6.11)

becomes, in this case,

Ldθ = 2πN, (6.12)

and

 2π
Ldθ = L dθ = 2πL, (6.13)
0

so that L = N , which is identical with Bohr’s quantization law (5.2).

6.2 de Broglie’s Wave Interpretation

In his doctoral thesis, presented in 1924 to the Faculty of Science at the University of
Paris, de Broglie [Bro24] proposed the existence of matter waves. The thoroughness
and originality of his thesis was recognized at once, but because of the apparent
lack of experimental evidence, de Broglie’s ideas were not considered to have
any physical reality. It was Albert Einstein who recognized their importance and
validity and in turn called them to the attention of other physicists. Five years later
de Broglie won the Nobel Prize in physics, his ideas having been dramatically
confirmed by experiment.
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch06

50 The Supersymmetric Dirac Equation

The hypothesis of de Broglie [Bro24] was that the dual, that is the wave-
particle, behavior of radiation applies equally well to matter. Just as a photon has a
light wave associated with it that governs its motion, so a material particle (e.g., an
electron) has an associated matter wave that governs its motion. Since the universe
is composed entirely of matter and radiation, de Broglie’s suggestion is essentially
a statement about a grand symmetry of nature. Indeed, he proposed that the wave
aspects of matter are related to its particle aspects in exactly the same quantitative
way that is the case for radiation. According to de Broglie, for matter and radiation
alike the total energy of an entity is related to the frequency ν of the wave associated
with its motion by the equation

E = 2πν, (6.14)

and the momentum p of the entity is related to the wavelength λ of the associated
wave by the equation

p = 2π/λ. (6.15)

Here the particle concepts, energy E and momentum p, are connected through
Planck’s constant h = 2π
= 2π, to the wave concepts, frequency ν and
wavelength λ. Equation (6.15), in the following form, is called de Broglie’s relation:
2π
λ= . (6.16)
p
It predicts the de Broglie wavelength λ of a matter wave associated with the motion
of a material particle having a momentum p.
The wave nature of light propagation is not revealed in experiments in
geometrical optics, for the important dimensions of the apparatus used there are
very large compared to the wavelength of light. Geometrical optics involves ray
propagation, which is similar to the trajectory motion of classical particles. When
the characteristic dimension of an optical apparatus becomes comparable to, or
smaller than, the wavelength of the light going through it, we are in the domain
of physical optics, where diffraction effects are observed, and the wave nature of
light propagation becomes apparent.
To observe wavelike aspects in the motion of matter, therefore, we need
systems with apertures or obstacles of suitable small dimensions. Using apparatus
with characteristic dimensions of 1 Å, wavelike aspects of the motion of electrons
should be apparent.
Elsasser pointed out, in 1926, that the wave nature of matter might be tested
in the same way that the wave nature of x-rays was first tested, namely by allowing
a beam of electrons of appropriate energy to fall on a crystalline solid. The atoms
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch06

Interpretation of the Quantum Rules 51

of the crystal serve as a three-dimensional array of diffracting centers for the

electron wave, and so they should strongly scatter electrons in certain characteristic
directions, just as for x-ray diffraction. This idea was confirmed in experiments by
Davisson and Germer in the United States and by G. P. Thomson in Scotland.
Going back to 1924, de Broglie applied his concepts to an electron orbiting
a nucleus. The Bohr quantization of angular momentum can be written as in
Eq. (5.4) as
mvr = pr = N, N = 1, 2, 3, . . . (6.17)

where p is the momentum of an electron in an allowed orbit of radius r. If we

substitute into this equation the expression for p in terms of the corresponding de
Broglie wavelength,
2π
p= , (6.18)
λ
the Bohr equation becomes
2πr
= N, (6.19)
λ
or

2πr = N λ, N = 1, 2, 3, . . . (6.20)

Thus the allowed orbits are those in which the circumference of the orbit can
contain exactly an integral number of de Broglie wavelengths.
Imagine the electron moving in a circular orbit at constant speed. It has a wave
associated with its motion. The wave, of wavelength λ, is wrapped repeatedly
around the circular orbit. The resultant wave that is produced will have zero
intensity at any point unless the wave at each traversal is exactly in phase at that
point with the wave in other traversals. If the wave in each traversal is exactly
in phase, they join on perfectly in orbits that accommodate integral numbers of
de Broglie wavelengths, as illustrated in Figure 6.1. But the condition that this
happens is just the condition that Eq. (6.20) be satisfied.
This wave picture gives no suggestion of progressive motion. Rather, it
suggests standing waves, as in a stretched string of a given length. In a stretched
string only certain wavelengths, or frequencies of vibration, are permitted. Once
such modes are excited, the vibration goes on indefinitely if there is no damping.
To get standing waves, however, we need oppositely directed traveling waves
of equal amplitude. For the atom this requirement is satisfied by the fact that
the electron can traverse an orbit in either direction and still have the same
magnitude of angular momentum required by Bohr. The de Broglie standing wave
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch06

52 The Supersymmetric Dirac Equation

Figure 6.1. de Broglie waves associated with an electron.

interpretation, illustrated in Figure 6.1, therefore provides a satisfying basis for

Bohr’s quantization rule, and, for this case, of the more general Sommerfeld–
Wilson rule. The time-independent features of the standing wave associated with
the electron in a stationary orbit of an atom make it possible to understand in a
simple way why the electron does not emit electromagnetic radiation and spiral
into the nucleus.

Notes on Chapter 6
The presentation here follows Eisberg and Resnick [ER74].
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch07

Chapter 7
Sommerfeld’s Model for Non-Relativistic Electrons

One of the most important applications of the Sommerfeld–Wilson quantization

rules is to the case of a hydrogen atom in which the electron can move, not only in
circular orbits, but in elliptical orbits. As we have seen, this is the motion allowed
in general in a 1/r-potential by classical mechanics. In this chapter we deal with
the application of this idea to slowly moving electrons, leaving its extension to
relativistic particles to Chapter 12.

7.1 Assumptions of the Model

In non-relativistic physics the Hamiltonian for an attractive Coulomb potential is
p2 Zα
H= − , (7.1)
2m r
where Z is the nuclear charge and α the fine-structure constant α ≈ 1/137. This
equation can be written as
Zα
H+ = Ekin . (7.2)
r
From the identity
L2 = (r × p)2 = r2 p2 − (r·p)2 (7.3)
we infer, with pr = p· r̂,
L2 p2θ
p2 = p2r += p2
r + , (7.4)
r2 r2
where pr is the momentum conjugate to r and pθ = |L| is the momentum conjugate
to θ.
We calculate the radial momentum as



dr pθ dr
pr = mṙ = mθ̇ = 2 , (7.5)
dθ r dθ
using pθ = L = mθ̇r 2 .

53
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch07

54 The Supersymmetric Dirac Equation

For convenience, define the new variable

1
s= , (7.6)
r
and note that


pr ds 1 dr 1 ṙ
=− = 2 = 2 . (7.7)
pθ dθ r dθ r θ̇
Introducing these variables into Eq. (7.1), we find that

 
2
p2θ ds 2
H + Zαs = +s . (7.8)
2m dθ

One differentiates Eq. (7.8) with respect to θ and obtains the linear differential
equation:

d2 s
+ s − C = 0, (7.9)
dθ2
where we have used the abbreviation
Zαm
C= . (7.10)
p2θ

The general solution for Kepler orbits is

1
s= = A cos θ + B sin θ + C, (7.11)
r
where A and B are the integration constants and C is given in Eq. (7.10). Defining
the distance of closest approach (perihelion) to occur at θ = 0 makes B = 0,
yielding the simpler equation:
1
= A cos θ + C. (7.12)
r
This result shows that the Kepler orbits have the form of conic sections.
Let us now express the elliptic orbits in standard geometrical terms, using the
eccentricity e. If a is the semimajor axis of the ellipse, a = c/(1 − e2 ), then the
orbit equation for the ellipse reads


1 1 + e cos θ 1 1 + e cos θ
= = . (7.13)
r c a 1 − e2
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch07

Sommerfeld’s Model for Non-Relativistic Electrons 55

Quantization is achieved by applying the Sommerfeld–Wilson quantization

rules to the phase integrals for pθ and pr :
  θ=2π
pθ dθ = L dθ = 2πnθ , (7.14)
θ=0

which yields pθ = nθ , and


pr dr = 2πnr . (7.15)

To evaluate the integral in Eq. (7.15) we use the expression Eq. (7.5) for pr
and obtain

2
pr dr e2 sin2 θ
pr dr = 2 dθ = pθ dθ. (7.16)
r dθ (1 + e cos θ)2

Thus, the radial quantization condition, Eq. (7.15), takes the form
 θ=2π
1 e2 sin2 θ dθ nr
= . (7.17)
2π θ=0 (1 + e cos θ)2 nθ

The integral in this equation was evaluated by Sommerfeld [Som16]

 θ=2π
1 e2 sin2 θ dθ
= (1 − e2 )− 2 − 1,
1
(7.18)
2π θ=0 (1 + e cos θ)2
so
nr
(1 − e2 )− 2 = 1 +
1
. (7.19)
nθ

7.2 Results of the Model for Non-Relativistic Hydrogen Atoms

It is now straightforward to obtain Bohr’s formula for the energy levels from
these results. The two parameters of the orbit equation (a and e in Eq. (7.13))
are determined from the quantization conditions:

1. From Eq. (7.19) we know the eccentricity of the orbit:

2
1 nr
= 1+ . (7.20)
1 − e2 nθ
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch07

56 The Supersymmetric Dirac Equation

2. From Eq. (7.10) we get the semimajor axis:

Zαm 1
C= = , (7.21)
2
nθ a(1 − e2 )
and
e
A = eC = . (7.22)
a(1 − e2 )

We then determine H from the energy equation, Eq. (7.8), and knowledge of
the orbit s = s(θ). Eq. (7.8) becomes
n2θ 2 2
H + Zα(A cos θ + C) = (A sin θ + A2 cos2 θ + C 2 + 2AC cos θ)
2m
n2θ 2
= (A + C 2 + 2AC cos θ). (7.23)
2m
Comparing the coefficients of cos θ yields
n2θ
Zα = C, (7.24)
m
which holds because of Eq. (7.10). The constant term yields
n2θ 2 n2
H + ZαC = (A + C 2 ) = θ (e2 + 1)C 2 . (7.25)
2m 2m
This leads to
n2θ   m
(Zα)2
H= 2 − (1 − e2 ) C 2 − ZαC = − . (7.26)
2m 2 (nθ + nr )2
For a given orbit the energy is constant
m
(Zα)2
H=E=− . (7.27)
2 (nθ + nr )2
For the circular orbits e = 0, and from Eq. (7.20) it follows that nr = 0. Thus, for
Bohr’s circular orbits
m
(Zα)2
E=− . (7.28)
2 n2θ
Defining N = nθ + nr we find that there are multiple orbits for N > 1, all of the
same energy:
m (Zα)2
EN = − . (7.29)
2 N2
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch07

Sommerfeld’s Model for Non-Relativistic Electrons 57

Since nθ = 1, 2, 3, . . . and nr = 0, 1, 2, 3, . . . , N can take on the values

N = 1, 2, 3, 4, . . . (7.30)

For a given value of N, nθ can assume only the values

nθ = 1, 2, 3, . . . , N. (7.31)

The integer N is called the principal quantum number, and nθ is called the
azimuthal quantum number. nθ = 0 is not allowed; for nr = 0 it would correspond
to an orbit going through the force center.

7.3 The Eccentricity

We can write Eq. (7.19) in the form
 n
2
θ
e= 1− . (7.32)
N
This shows that the shape of the orbit (the ratio of the semiminor to the semimajor
axes) is determined by the ratio of nθ to N . For nθ = N the orbits are circular.
Figure 7.1 shows possible orbits corresponding to the first three values of the
principle quantum number. Corresponding to each value of the principal quantum
number N there are N different allowed orbits. One of these, the circular orbit, is
just the orbit described by the original Bohr model. The others are elliptical. But
despite the very different paths followed by an electron moving in the different
possible orbits for a given N , Eq. (7.29) tells us that the total energy of the electron
is the same. The total energy of an electron depends only on N . The several orbits
characterized by a common value of N are said to be degenerate. The energies of
different states of motion “degenerate” to the same total energy.

Figure 7.1. The Sommerfeld model of the hydrogen atom.

August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch07

58 The Supersymmetric Dirac Equation

This degeneracy in the total energy of an electron, due to orbits of very different
shape but common N , is the result of a very delicate balance between potential and
kinetic energy, which is characteristic of treating the inverse square Coulomb force
by the methods of classical mechanics. Exactly the same phenomena is found in
planetary or satellite motion, which is governed by the inverse square gravitational
force. For instance, a satellite may be launched into any one of a whole family of
elliptical orbits, all of which correspond to the same state of the same total energy
and have the same semimajor axis. Of course, there is no quantization of the orbit
parameters in these macroscopic cases, but as far as degeneracy is concerned they
are completely analogous to the case of the hydrogen atom.
To get the complete degeneracy of the electron, neglecting the spin, we turn
to a group-theoretical analysis in the next chapter.

Notes on Chapter 7
The history of atomic models is told in Keller [Kel83]. The importance of hydrogen
in this context is emphasized by Rigden [Rig03]. This chapter develops a non-
relativistic version of Sommerfeld’s work [Som16], following Biedenharn [Bie83].
See also Sommerfeld’s book [Som24].
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch08

Chapter 8
Quantum Mechanics of Hydrogenic Atoms

Heisenberg [ Hei25], on the basis of the spectral data of hydrogen, wrote a paper
in which he formulated for the first time the canonical commutation relations
between the basic variables of position and momenta, [xi , pj ] = iδij . He gave his
manuscript to Born, who submitted it to the Zeitschrift für Physik. Born recognized
that Heisenberg’s equations could be expressed in matrix form. The formulation of
the matrix formalism of quantum mechanics was published in two papers in 1925,
by Born and Jordan [BJ25], and by Born, Heisenberg and Jordan [MBJ26].
The immediate question was whether the new theory was capable of explaining
the spectrum of hydrogen, as the Bohr model had done. The answer was given by
Pauli [Pau26], and independently by Dirac [Dir26], towards the end of 1925. Pauli
obtained the same formula that Bohr had obtained in 1913, only this time the route
to the formula was a coherent theory — the new theory of quantum mechanics.
In the following we shall show, following Pauli, that the basic commutation
relations, and a general knowledge of the symmetry group SO(4), lead to the Balmer
formula, and a first estimate of the degeneracy of the energy levels.
The classical Kepler problem is characterized by six conserved quantities: the
three components of the orbital angular momentum L = (L1 , L2 , L3 ) and the
three components of the Laplace vector, A = (A1 , A2 , A3 ), satisfying the two
constraints A·L = 0 and |A|2 = 2EL2 /((Zα)2 m) + 1. We shall see that these
relations all have their counterparts in the quantum mechanical case.

8.1 Quantization
In general, we go from a classical to a quantum system by following the Dirac
quantization procedure [Dir53], i.e., we replace observable quantities by Hermitian
operators, and Poisson brackets of observables by commutators, according to the
rule {, } → i[, ].

59
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch08

60 The Supersymmetric Dirac Equation

For the basic observables, position xi and momentum pj , this means that
{xi , pj } = δij is replaced by [xi , pj ] = iδij . More complex commutators are
reduced by analogs of the rules (a)–(d) of Section 3.3, e.g.,

[A, BC] = [A, B]C + B[A, C], (8.1)

which follows from

[A, BC] = ABC−BCA = ABC−BAC+BAC−BCA = [A, B]C+B[A, C].

(8.2)
For an electron in a Coulomb field the Hamiltonian is
p2 Zα
H= − . (8.3)
2m r
The components of the angular momentum are

Li = ijk xj pk . (8.4)

These are Hermitian:

Li = ijk xj pk , L†i = ijk pk xj = ijk xj pk = Li . i = 1, 2, 3. (8.5)

The commutator of the Hamiltonian with a component of the angular momentum is

1 2 1
[H, Li ] = [p , Li ] = ijk [p2 , xj pk ]
2m 2m
1
= ijk (pl [pl , xj ] pk + [pl , xj ] pk pl )
2m
−i
= ijk (pj pk + pk pj ) = 0. (8.6)
2m
The angular momentum commutes of course with arbitrary functions of r. This
tells us that L is a constant vector.
If we try to take the Laplace vector A = (1/2m)(p × L) − r̂ over to quantum
mechanics it won’t work, because (ijk pj Lk )† = ijk Lk pj = −ikj Lk pj , or

(p × L)† = −L × p, (8.7)

is not Hermitian. This may be easily corrected,

1
A= (p × L − L × p) − r̂ (8.8)
2Zαm
is Hermitian, and reduces to the Laplace vector in the classical limit. More formally,
we may say that quantizing the classical vector A according to the Weyl–Moyal
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch08

Quantum Mechanics of Hydrogenic Atoms 61

quantization prescription [Tak08]

1
Ai = ijk {pj , Lk } − xi /r (8.9)
2Zαm
leads to Eq. (8.8). Now and in the following the brackets {, } are taken to mean the
anticommutator: {A, B} = AB + BA.
To show that A is a constant of the motion we wish to prove [H, A] = 0. We
do this by repeating the calculation of Section 3.4, paying attention this time to
the ordering of factors, and using the basic commutation relation [xi , pj ] = iδij
instead of the Poisson bracket {xi , pj } = δij . Equation (3.64) is now
ijk Zα
(Zαm)[H, Ai ] = {[H, pj ], Lk } − {pj , [pj , xi /r]}
2 2
iZαijk iZα
= {xj /r3 , Lk } + {pj , δij /r − xi xj /r3 }, (8.10)
2 2
using the rule [pi , g] = −i∂g/∂xi , for g a function of xi , and the first term is
iZαijk kls
xj xj  iZα  x x 
j l
x l ps + p s x l = (δ il δjs − δ is δ jl ) ps , 3 ,
2 r3 r3 2 r
(8.11)

which cancels the second term.

The quantum mechanical generalization of Eq. (2.38) is

A·L = L·A = 0. (8.12)

We prove this as follows: Use L × p = p × L − 2ip to write

1
A= (p × L − ip) − r̂. (8.13)
Zαm
Then, since p·L = r·L = 0,
1
(Zαm)A·L = ijk pj Lk Li = ijk pj [Lk , Li ] = ip·L = 0. (8.14)
2
Similarly, L·A = 0.

8.2 Quantum Mechanical Relation Between |A|2 and L2

Finally, we wish to prove the generalization of Eq. (2.40):
2H
|A|2 = (L2 + 1) + 1. (8.15)
(Zα)2 m
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch08

62 The Supersymmetric Dirac Equation

We shall use the quantum identities

(a) (p × L)·(p × L) = p2 L2 ,
(b) (p × L)·p = 2ip2 ,
(c) p·(p × L) = 0,
(d) (p × L)·r = L2 + 2i(p·r),
(e) r·(p × L) = L2 . (8.16)

Exercise 8.1 Prove these identities.

We calculate

(Zαm)2 (|A|2 − 1) = (p × L − ip)2 − (Zαm)(p × L − ip)· r̂

− (Zαm)r̂·(p × L − ip). (8.17)

Use the identities (8.16) to get

(p × L − ip)2 = p2 (L2 + 1),

L2 1
(p × L − ip)· r̂ = + ip·r , (8.18)
r r
L2 i
r̂·(p × L − ip) = − r·p.
r r
We finally obtain
 
−2 2Zαm 2 2Zαm
|A| − 1 = (Zαm)
2
p (L + 1) − 2
L −
2
r r
 2 
2 p Zα 2H
= − (L2 + 1) = (L2 + 1). (8.19)
(Zα)2 m 2m r (Zα)2 m
Thus we see that in the quantum mechanical case we have, as in the classical
case, six constants of the motion, L = (L1 , L2 , L3 ) and A = (A1 , A2 , A3 ), which
satisfy the two constraints, Eq. (8.12) and Eq. (8.15).

8.3 Pauli’s Hydrogenic Realization of so(4)

In the quantum mechanical case the commutators between the components of L
and the components of A are of interest. We shall see that, properly normalized,
these commutators are those of an so(4)-algebra.
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch08

Quantum Mechanics of Hydrogenic Atoms 63

We can calculate [Li , Lj ] from

[Li , Lj ] = ijmn (xm [Li , pn ] + [Li , xm ]pn ). (8.20)

The individual commutators follow from the definition

Li = imn xm pn , (8.21)

[Li , xm ] = irs [xr ps , xm ] = irs xr [ps , xm ] = −iirm xr = iimr xr , (8.22)

and similarly

[Li , pm ] = iqim pq , (8.23)

so

[Li , Lj ] = ijnm (inq xq pm +imq xn pq ) = i(xi pj −xj pi ) = iijk Lk . (8.24)

In general, we say that a vector operator v, whose components satisfy with the
components of the angular momentum

[Li , vj ] = iijk vk , (8.25)

transforms as a vector operator.

The commutators [Li , Aj ] = iijk Ak are the statement that A transforms as
a vector operator with respect to L; this is obvious, since A is made up of such
vectors.
We now have to calculate [Ai , Aj ]. The calculation consists of a repetition of
that with Poisson brackets, this time taking into account the ordering of factors.
We also have to take into account the additional terms in the quantum mechanical
expression for A, Eq. (8.8), compared to the classical expression, Eq. (2.22). Now
we get

1 1
(Zαm)[pi , Aj ] = jsn [pi , ps Ln ] − [pi , Ls pn ] − Zαm[pi , xj /r]
2 2
= i(pi pj − p2 δij ) + i(Zαm)(δij /r − xi xj /r 3 ). (8.26)
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch08

64 The Supersymmetric Dirac Equation

To calculate [pi , 1/r] we use the analog of the rule (3.46), generalized to the
quantum context, [pi , g] = −i∂g/∂xi , for g(x) a function of x.

(Zαm)[xi /r, Aj ] = jsn ([xi /r, ps Ln ] − [xi /r, Ls pn ])


Lk 1
= i (ijk − 3 jsn (xi xs + Ln + Ln xi xs )
r 2r
1
xj x j  1
xs xs 
+ pi + pi − δij ps + ps . (8.27)
2 r r 2 r r
The latter yields

(Zαm)2 ijk [xi /r, Aj ] = −2imLk /r 3 , (8.28)

which is the analog of Eq. (3.72).

The main calculation is

(Zαm)2 ijk [Ai , Aj ]

 
1
= (Zαm)ijk isn ([ps Ln , Aj ] + [Ln ps , Aj ]) − (Zαm)[xi /r, Aj ]
2

1
= (Zαm)ijk isn (ps [Ln , Aj ] + [ps , Aj ]Ln
2

+ Ln [ps , Aj ] + [Ln , Aj ]ps ) −(Zαm)[xi /r, Aj ] . (8.29)

We evaluate the terms separately. The first two expressions are identical to those
in the classical evaluation:

(Zαm)ijk isn ps [Ln , Aj ] = i(−p2 + (Zαm)/r)Lk . (8.30)

(Zαm)ijk isn [ps , Aj ]Ln = i(−p2 + (Zαm)/r)Lk . (8.31)

The next two expressions are new:

(Zαm)ijk isn Ln [ps , Aj ]

= iijk isn Ln (ps pj − p2 δsj + (Zαm)δsj /r − (Zαm)xs xj /r 3 )

= i(δjs δkn − δjn δks )Ln (ps pj − p2 δsj + (Zαm)δsj /r − (Zαm)xs xj /r3 )

= iLk (−p2 + (Zαm)/r). (8.32)

August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch08

Quantum Mechanics of Hydrogenic Atoms 65

(Zαm)ijk isn [Ln , Aj ]ps = i(Zαm)ijk isn njq Aq ps

= i(Zαm)(δjs δkn − δjn δks )njq Aq ps = i(Zαm)kjq Aq pj

 
1
= ikjq qmn (pm Ln + Ln pm ) − (Zαm)xq /r pj
2
 
1
= (δkm δjn − δkn δjm )(pm Ln + Ln pm ) − (Zαm)kjq xq /r pj
2

= iLk (−p2 + (Zαm)). (8.33)

Combining these results, we get, as in the classical calculation involving Poisson

brackets:
 
−2H
[Ai , Aj ] = iijk Lk = iijk N −2 Lk , (8.34)
(Zα)2 m
with
−2H
N −2 = . (8.35)
(Zα)2 m
The appearance of the hydrogen Hamiltonian in (8.34) shows that A and L do not
close under commutation to form a Lie algebra.
For the bound state problem H has nonzero negative eigenvalues, so that we
may define (N 2 )− 2 and normalize the operator A:
1

1
A = (N 2 ) 2 A. (8.36)

Now the commutation relations become

[Li , Lj ] = iijk Lk , [Li , AJ ] = iijk Ak , [Ai , Aj ] = iijk Lk . (8.37)

The L and A constitute the generators of an so(4) algebra, and provide

a realization of the so(4) generators, called Pauli’s hydrogenic realization.
Alternatively we could consider continuum states and replace H with a positive
energy and use (8.36) without the minus sign to obtain the so(3,1) commutation
relations of the Lorentz group. Indeed, Dahl [ Dah95] argues that the form of the
Laplace vector is determined by the fact that it is simply related to the generator
of Lorentz boosts.
The so(4) commutation relations can be transformed to a simpler canonical
form. If we define the two vector operators
1 1
M= (L + A), N= (L − A), (8.38)
2 2
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch08

66 The Supersymmetric Dirac Equation

then the components of M and N satisfy the commutation relations

[Mi , Mj ] = iijk Mk , [Ni , Nj ] = iijk Nk , [Mi , Nj ] = 0, (8.39)

which are equivalent to the original ones since (8.38) represents a real linear
transformation of the original generators. The new commutation relations are
simpler since M and N are two commuting angular momentum vectors. Therefore
so(4) is the direct sum of so(3) ⊕ so(3) subalgebras. The first is generated by M
and the second is generated by N.
The so(4) algebra has two Casimir operators; these are quadratic operators
commuting with all the generators:
1
C1 = L2 + A2 , C2 = (L·A + A·L). (8.40)
2

Exercise 8.2 Show that C1 and C2 are Casimir operators:

[C1 , L] = [C1 , A] = [C2 , L] = [C2 , A] = 0. (8.41)

8.4 so(4) and the Spectrum of Hydrogenic Atoms

Substituting (8.36) into (8.12) and (8.15) we obtain the important equations
A·L = L·A = 0, |A|2 + L2 + 1 = N 2 . (8.42)
These give the values of the Casimir operators of so(4). The first tells us that
C2 = 0. The second equation contains the quadratic invariant A·A + L·L and tells
us that C1 = N 2 − 1. The irreducible representations of so(4) are given by the
eigenvalues (C1 , C2 ).
The equations for C1 and C2 may also be used to obtain the Bohr formula for
the energy levels as follows. From C2 = 0 it is clear that
1 2
M2 = N2 = (L + |A|2 ). (8.43)
4
Since the eigenvalues of M 2 and N 2 are j1 (j1 + 1) and j2 (j2 + 1), respectively,
it follows that j1 = j2 . Therefore, our realization of A using the modified Laplace
vector does not give all the irreducible representations of so(4); only the so-called
diagonal ones with j1 = j2 .
This is analogous to the use of L as a realization of the so(3) generators. Only
integral values of the angular momentum are realized. The values j = 1/2, 3/2, . . .
are not possible with the realization L = r×p in terms of coordinates and momenta.
The equation C1 = N 2 − 1 may be rewritten as
L2 + |A|2 + 1 = N 2 , (8.44)
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch08

Quantum Mechanics of Hydrogenic Atoms 67

or substituting (8.43)

4M 2 + 1 = N 2 . (8.45)

Replacing M 2 by its eigenvalue j1 (j1 + 1) we find

4j1 (j1 + 1) + 1 = 4j12 + 4j1 + 1 = (2j1 + 1)2 = N 2 . (8.46)

Since j1 takes the values j1 = 0, 1/2, 1, 3/2, . . . , N 2 takes the values 1, 4, 9, . . .

That is, the operator N 2 takes the eigenvalues

N 2 → N 2. (8.47)

Finally, using Eq. (8.35), we obtain the Bohr formula

m (Zα)2
EN = − (8.48)
2 N2
for the energy levels of the hydrogenic atom. We see that N is the principal quantum
number. Given that C2 = 0, the energy levels are classified by the single number
C1 , or N .
We see that there is nothing “accidental” about the so-called “accidental
degeneracy” of the energy levels. This term arose because the degeneracy of the
2l + 1 hydrogenic states |N lm > for fixed N and l, corresponding to the energy
EN , could be explained since [L, H] = 0, but the full degeneracy of the energy
level,

N −1
N (N − 1)
(2l + 1) = 2 + N = N 2, (8.49)
2
l=0

cannot be explained on the basis of SO(3) alone, since SO(3) does not guarantee that
the different values of the angular momentum l all have the same energy. However,
it is SO(4) that is the geometrical symmetry group, not SO(3). In fact it follows
from j1 = j2 that the degeneracy of EN is N 2 , since there are N 2 = (2j1 + 1)2
basis vectors in the direct product space corresponding to the energy EN .
It must be emphasized that N 2 does not give the full degeneracy of the levels,
which is experimentally 2N 2 . The further factor of two comes from the spin of
the electron, which has so far not entered our considerations. Another indication
of this fact is discussed below.
Classically the eccentricity e is the length of the vector A. We might attempt to
take this relation over to quantum mechanics. From Eqs. (8.36), (8.42), and (8.47)
we find that

|A|2 = A·A = e2 = 1 − N −2 (L2 + 1). (8.50)

August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch08

68 The Supersymmetric Dirac Equation

Thus for L2 → l(l + 1), N 2 → N 2 , we find that the eccentricity becomes

 1
l2 + l + 1 2
e= 1− . (8.51)
N2
For circular orbits, the eccentricity vanishes classically. Quantum mechanically, we
shall see that the circular orbits (nodeless radial probability density) correspond to
l = N − 1; e is then small (≈ N − 2 for N large), but does not vanish (except for
1

N = 1).
This result for the eccentricity is strictly applicable to the orbits of spinless
particles. For the electron, which has spin- 12 , it is not applicable. In Chapter 10 we
shall find the value of the eccentricity for the orbits of the spinning electron.

Notes on Chapter 8
This chapter is a review of Pauli’s paper [Pau26]. If you need to review the
derivation of quantum mechanical commutation relations, or the application of
group-theoretical symmetry considerations, an excellent place to look is the very
lucid book by Adams [Ada94].
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch09

Chapter 9
The Schrödinger Equation
and the Confluent Hypergeometric Functions

In late 1925 Schrödinger read de Broglie’s paper in which it was proposed that
particles have an associated wavelength. Schrödinger went to work attempting to
generalize de Broglie’s concept of matter waves. The first sentence of Schrödinger’s
classic paper [Sch26] reads as follows: “In this paper I wish to consider, first,
the simplest case of the hydrogen atom, and show that the customary quantum
conditions can be replaced by another postulate, in which the concept of ‘whole
numbers’, merely as such, is not introduced.” In modern language Schrödinger is
saying that he can dispense with the use of quantum conditions; they are replaced
by the requirement that the argument of the hypergeometric function describing
the solution of the wave equation must be a nonzero negative integer if the wave
function is to fall off at infinity (see below).
The mathematics employed in the paper was familiar to physicists — it
involved the solutions of differential equations. Schrödinger set up a wave equation
in a form appropriate for the hydrogen atom — a negatively charged electron
orbiting a positively charged nucleus. The solutions are normalizable only for
discrete values of the energy. In 1926 Born suggested the interpretation of the
solution of the wave equation: The magnitude of the quantity |ψ|2 , where ψ is
the solution of the wave equation, is a measure of the probability density of the
electron. Schrödinger’s wave equation gives the maximum magnitude of |ψ|2 in
the ground state of the atom at a distance of 0.529 Å from the nucleus, in agreement
with the Bohr radius, and the known size of the atom, whose diameter is about 1 Å.

9.1 The Schrödinger Equation and Its Solutions

The Schrödinger equation is obtained by replacing the classical Hamilton function
by an operator, through the prescription p = −i∇, and applying it to a

69
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch09

70 The Supersymmetric Dirac Equation

wave function:


1 2 Zα
Hψ(r) = − ∇ − ψ(r) = Eψ(r). (9.1)
2m r
This equation is most naturally solved by transforming from Cartesian to spherical
polar coordinates, in which system it is completely separable. We use the identity

L2 = r2 p2 − r(r · p) · p. (9.2)

The component of ∇ in the direction of r is ∂/∂r, hence

∂
r · p = −ir , (9.3)
∂r
and
∂2 ∂
L 2 = r 2 p2 + r 2 2
+ 2r , (9.4)
∂r ∂r
or, more compactly
 
∂ ∂
L2 = r2 p2 + r2 . (9.5)
∂r ∂r
We can formulate this in terms of the momentem conjugate to r, which satisfies
[r, pr ] = i. Then
 
i ∂ ∂ 1
pr = − r = −i + . (9.6)
r ∂r ∂r r

Exercise 9.1 Show that pr , as defined above, is canonically conjugate to r.

Exercise 9.2 Show that the Moyal–Weyl quantization prescription, applied to p
and r̂, yields pr .
Exercise 9.3 Show that pr is Hermitian.
This yields

∂2 2 ∂ 1 ∂ 2 ∂
p2r = − 2
− =− 2 r , (9.7)
∂r r ∂r r ∂r ∂r
and
L2
p2 = p2r + , (9.8)
r2
as the quantum-mechanical counterpart to Eq. (7.4).
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch09

The Schrödinger Equation and the Confluent Hypergeometric Functions 71

With this, solutions to Eq. (9.1) are found to be of the form

ψ(r, θ, φ) = R(r)Y (θ, φ), (9.9)

where the Y (θ, φ) are spherical harmonics and R(r), the radial part of the wave
function, satisfies


2Zαm l(l + 1)
−2m(H − E) = −pr + 2mE −
2
− R
r r2


1 d 2 dR 2Zαm l(l + 1)
= 2 r + 2mE + − R = 0,
r dr dr r r2
(9.10)

subject to the conditions that limr→0 rR(r) = 0 and rR(r) be square integrable.
We introduce the dimensionless variable
 
Zαm
x=2 r, (9.11)
λ

and obtain
d2 R dR
x2 + 2x + [−µx2 + λx − l(l + 1)]R = 0. (9.12)
dx2 dx
We have defined the parameter µ by

−E λ2
µ= . (9.13)
2m (Zα)2

The parameter λ > 0 is yet to be determined. For bound states (E < 0) we also
have µ > 0.
Next we factor out the behavior near x = 0 and x = ∞, and look for solutions
of the type
√
R(x) = xl e− µx
u(x). (9.14)

For u(x) we get the differential equation

√ √
xu (x) + 2[(l + 1) − µx]u (x) − [2(l + 1) µ − λ]u(x) = 0. (9.15)
√
With µ = 1/2 this becomes

xu (x) + [2(l + 1) − x]u (x) − [l + 1 − λ]u(x) = 0. (9.16)

August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch09

72 The Supersymmetric Dirac Equation

This is of the form of the confluent hypergeometric equation,

 2 
d d
x 2 + (c − x) − a 1 F1 (a; c; x) = 0, (9.17)
dx dx
with a = l+1−λ and c = 2l+2. The general solution of Eq. (9.17) is proportional
to a confluent hypergeometric function,

u(x) = A1 F1 (l + 1 − λ; 2l + 2; x), (9.18)

if we demand that it converges at x = 0.

Since in general

1 F1 (l + 1 − λ; 2l + 2; x) −−−−→ ex , (9.19)
x→∞

the function u(x) diverges for large x unless l + 1 − λ is a negative integer or zero,
in which case the confluent hypergeometric function is a polynomial. This means
that λ must be a positive integer, which we denote by N ,

λ = N, N = 1, 2, 3, . . . (9.20)
√
Hence, with µ = 1/2, Eq. (9.13) becomes

1 E N2
µ= =− (9.21)
4 2m (Zα)2
and we find again that the energy spectrum is discrete, with

m (Zα)2
EN = − , N = 1, 2, 3, . . . (9.22)
2 N2
The variable x is now
 
Zαm
x=2 r. (9.23)
N
The spherical harmonics are normalized to unity, so the normalization of the wave
function (9.9) is
 
|R(r)|2 r2 dr = |u(r)|2 dr = 1, (9.24)

where u(r) = rR(r). Introduce the function of the dimensionless variable x


N
u(x) = u(r), (9.25)
2Zαm
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch09

The Schrödinger Equation and the Confluent Hypergeometric Functions 73

with

|u(x)|2 dx = 1. (9.26)

The radial solution, which is regular at the origin and falls off exponentially at
infinity, is
 3
u(r) 2Zαm 2 uN l (x)
RN l (r) = = (9.27)
r N x
where
uN l (x) = AN l xl+1 e− 2 x 1 F1 (−N + l + 1; 2l + 2; x),
1
(9.28)
and AN l is the normalization constant. The normalization constant is calculated in
Appendix B:

1 (N + l)!
AN l = . (9.29)
(2l + 1)! (N − l − 1)!(2N )
To obtain a polynomial solution we must have N ≥ l + 1.

9.2 Laguerre Polynomials and Associated

Laguerre Functions
We note that Eq. (9.16) has the form
xu (x) + (p + 1 − x)u (x) + (q − p)u(x) = 0, (9.30)
where p and q are integers. Since we are looking for polynomial solutions, we can
define a new function v(x) by
dp v(x)
u(x) = . (9.31)
dxp
Clearly, if u(x) is a polynomial, then v(x) is also a polynomial.
On substituting this expression into Eq. (9.30) we find, after some
rearrangement,
dp
[xv  (x) + (1 − x)v  (x) + qv(x)] = 0. (9.32)
dxp
We see that if v(x) satisfies Laguerre’s equation (also a confluent hypergeo-
metric equation)

xv  (x) + (1 − x)v  (x) + qv(x) = 0, (9.33)

August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch09

74 The Supersymmetric Dirac Equation

then u(x) is given by Eq. (9.31). The solution that is regular at the origin is

v(x) = Aq 1 F1 (−q, 1; x). (9.34)

We choose Aq = q! in Eq. (9.34) and obtain for the polynomial solution of

Laguerre’s equation

dq q −x
Lq (x) = q!1 F1 (−q, 1; x) = ex (x e ), (9.35)
dxq
see Appendix A. This function is the Laguerre polynomial of order q.
From Eq. (9.31) we see that the polynomial solution to Eq. (9.30) is obtained
by differentiating Lq (x) p times. We denote these solutions by

dp Lq (x)
Lpq (x) = . (9.36)
dxp
These functions are known as the associated Laguerre polynomials. It is a
polynomial of degree q − p. By carrying out the differentiation explicitly, we
find
∞
q!(−q)p  (−q + p)k k
Lpq (x) = x , (9.37)
p! k!(p + 1)k
k=0

where the Pochhammer symbols (a)n are defined in Appendix A.

This series is seen to describe the confluent hypergeometric function. We
generalize this result to include noninteger q and p and define the associated
Laguerre function by

[Γ(q + 1)]2
Lpq (x) = (−1)p 1 F1 (−q + p, p + 1; x). (9.38)
Γ(p + 1)Γ(q − p + 1)

From Eq. (9.28) we see that the radial functions for bound states in a Coulomb
potential are
  32
2Zαm
xl e− 2 x L2l+1
1
RN l (r) = NN l N +1 (x), (9.39)
N

where NN l is a normalization constant. In Appendix B we calculate


(N − l − 1)!
NN l = − . (9.40)
(N + l)!3 (2N )
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch09

The Schrödinger Equation and the Confluent Hypergeometric Functions 75

Notes on Chapter 9
The analysis of the Schrödinger equation is conventional, e.g., Mertzbacher
[Mer61]. The hypergeometric functions and the special functions which arise in
different physical contexts are reviewed, for example, in the book by Seaborn
[Sea80].
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch09

This page intentionally left blank

August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch10

Chapter 10
Non-Relativistic Hydrogenic Atoms with Spin

In 1922 Stern and Gerlach passed a beam of silver atoms through an inhomogeneous
magnetic field, and found that it was split into two components. Goudsmit and
Uhlenbeck proposed in 1925, on the basis of atomic spectra, that the electron was
characterized by an additional quantum number with value 1/2, which would also
explain the results of the Stern–Gerlach experiment. Pauli [Pau26] developed a
two-component formalism, which took the spin of the electron into account. The
two-component states transform as spinors, see [BJ53] or [FK05].
Our treatment of spin below has the benefit of hindsight. We present in this
chapter a complete theory of the non-relativistic hydrogenic atom with a spinning
electron. We thereby correct the long-held assumption that this theory is somehow
incomplete, needing the insight provided by the Dirac theory to understand it.
This is because the spin is a completely non-relativistic phenomena, as has been
emphasized by Lévy-Leblond [LL74]. The Dirac theory is presented in this book
as an extension of the successful two-component Pauli formalism.
We use the operator Knr to characterize the states, which was introduced by
Biedenharn and Louck [BL81] in 1981 as the non-relativistic form of an operator
that was used by Dirac [Dir53]. We then show how we are led to a factorized form
of the Pauli Hamiltonian, which allows a solution of the problem analogously to
the harmonic oscillator. This method, which was developed for the treatment of a
great variety of problems by Infeld [Inf41], and Hull and Infeld [HI51] in 1951, then
suggests a supersymmetric interpretation, which is developed in the next chapter.

10.1 Spin Variables, the Pauli Hamiltonian

and Factorization
We consider a spin- 12 Pauli particle moving in a Coulomb potential. The spin is
dynamically independent, and does not affect the energy levels, which are given
as before by the Bohr formula. The degeneracy, however, is doubled to the correct

77
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch10

78 The Supersymmetric Dirac Equation

value 2N 2 , instead of N 2 . The degeneracy becomes observable when we switch

on an electric or magnetic field, which leads to a splitting of the spectral lines.
We describe a free spin- 12 particle with the help of a two-component formalism.
If the spin is independent of the space-time variables, the Hamiltonian is
HP ψ = Eψ, (10.1)

where

ψ+
ψ= . (10.2)
ψ−
ψ+ is the spin-up component, ψ− the spin-down component. If the degeneracy of
the spin-up levels and the degeneracy of the spin-down levels is N 2 , the degeneracy
of ψ is 2N 2 . The two-component Hamiltonian HP is
p2
HP = I2 , (10.3)
2m
where I2 is the unit 2 × 2 matrix, and HP is called the Pauli Hamiltonian. This just
says that spin-up particles and spin-down particles have exactly the same dynamics.
If p is considered as a Hermitian operator in the two-dimensional complex
space then it may be expanded in terms of a basis of this space,
p = pi σi . (10.4)
If we are to have the usual relation between the square of this vector and the scalar
vector norm we must have
(σ·p)(σ·p) = (σ1 )2 p21 + (σ2 )2 p22 + (σ3 )2 p33
+{σ1 , σ2 }p1 p2 + {σ1 , σ3 }p1 p3 + {σ2 , σ3 }p2 p3
= p21 + p22 + p23 = p2 . (10.5)
This holds if
{σi , σj } = σi σj + σj σi = 2δij , (10.6)
where juxtaposition signifies matrix multiplication. Matrices which satisfy such a
relation are said to form a Clifford algebra. Since space is three-dimensional, there
are exactly three such matrices.
Let
σ1 σ2 σ3 = X. (10.7)

Then
X 2 = σ1 σ2 σ3 σ1 σ2 σ3 = σ2 σ3 σ2 σ3 = −I2 , (10.8)
and
X = σ1 σ2 σ3 = iI2 . (10.9)
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch10

Non-Relativistic Hydrogenic Atoms with Spin 79

From this we obtain

σ1 σ2 = iσ3 , (10.10)

and in general
σi σj = iijk σk . (10.11)

Then
[σi , σj ] = σi σj − σj σi = 2iijk σk . (10.12)

In the approach to theoretical physics called geometrical algebra the Clifford

algebra structure plays a central role; see references in the notes to this chapter.
The commonly used representation for the Pauli matrices σi is




01 0 −i 1 0
σ1 = , σ2 = , σ3 = . (10.13)
10 i 0 0 −1
The Pauli matrices obviously satisfy the identity
1 1
σi σj = {σi , σj } + [σi , σj ] = δij + iijk σk . (10.14)
2 2
This may also be written as

(σ·A)(σ·B) = A·B + iσ·(A × B), (10.15)

where A and B are any two vectors which commute with σ. With the help of this
identity the Hamiltonian can be written as
(σ·p)(σ·p)
HP = . (10.16)
2m
The advantage of writing the Hamiltonian as in Eq. (10.16) can be seen when
we consider an electron in a magnetic field. The presence of the magnetic field can
be accounted for by the rule of minimal substitution:

p → π = p − eA, (10.17)

where A is the vector potential (in this paragraph only), which satisfies ∇ ×A = B,
and B is the magnetic field. The Hamiltonian becomes
1 π2 i
HP → Hmag = (σ·π)(σ·π) = + σ·π × π. (10.18)
2m 2m 2m
We now have

[πi , πj ] = [pi − eAi , pj − eAj ] = ie(∇i Aj − ∇j Ai ) = (ie)ijk Bk , (10.19)

August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch10

80 The Supersymmetric Dirac Equation

and
1 1
σ·π × π = mij σm [πi , πj ] = mij ijk σm (ie)Bk = (ie)σk Bk . (10.20)
2 2
Therefore
π2 eσ π2 e π2
Hmag = − ·B = −g S ·B = − µ·B, (10.21)
2m m 2 2m m 2m
where Si denote the spin matrices, Si = 12 σi , and the magnetic moment is µ.
µB = e/(2m) is called the Bohr magneton, and g is the gyromagnetic ratio, here
g = 2, with

µ = gµB S. (10.22)

The spin matrices satisfy

[Si , Sj ] = iijk Sk , (10.23)

so that the Si are the generators of an su(2)-algebra, and S2 = s(s + 1), where
s = 1/2 is the spin of the electron.
This is in agreement with the experimental results for the interaction of the
spin of the electron with a magnetic field (up to corrections due to quantum
electrodynamics). The “classical” value of g is g = 1, where
 e 
µ=g L, (10.24)
2m
and L is the orbital angular momentum. This is nowadays recognized as the value
appropriate for spin-one photons. In fact, it is conjectured that g = 1/s in general.
The conjecture has been proved for s ≤ 2, and for any half-integral s ([LL74] and
references therein).
This method of introducing a two-component description for particles with
spin one-half, and the fact that this automatically leads to a value g = 2, was
apparently first noticed by Feynman [Sak67]. It copies Dirac’s method for the
relativistic case, but works just as well for non-relativistic fermions, and could
have been introduced in this case. It brings out the fact that spin is inherently a
non-relativistic phenomena [LL74].
The Clifford algebra structure of the Pauli matrices allows us to factorize
quadratic operators. Consider the quadratic operator L2 = L21 + L22 + L23 . We can
factorize this with the help of the spin operator σ. One finds

L2 = σ·L(σ·L + 1), (10.25)

August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch10

Non-Relativistic Hydrogenic Atoms with Spin 81

which is easily verified from Eq. (10.15) and the equation L × L = iL. The total
angular momentum is J = L + 12 σ. It factorizes as



2 3 1
J = σ·L + σ·L + . (10.26)
2 2
We now introduce, following Biedenharn and Louck [ BL81], the non-
relativistic analog of Dirac’s operator [Dir53]:

Knr = −(σ·L + 1) = L2 − J2 − 1/4. (10.27)

We can then express these factorizations as

L2 = Knr (Knr + 1), (10.28)

and



1 1
2
J = Knr + Knr − 2
= Knr − 1/4. (10.29)
2 2
Since Knr commutes with J2 and L2 this yields, when applied to a simul-
taneous eigenvector, Knr → κ, L2 → l(κ)(l(κ) + 1), J2 → j(κ)(j(κ) + 1) with

κ, κ positive
l(κ) = (10.30)
|κ| − 1, κ negative
and

j(κ) = |κ| − 1/2. (10.31)

Since j = 1/2, 3/2, . . . we find

κ = ±1, ±2, . . .(0 excluded). (10.32)

10.2 A Theorem Concerning the Anticommutation of K

The operator Knr possesses the important property of anticommuting with the
operator σ·v, where v is any vector that is perpendicular to L. In fact, a theorem
of Biedenharn and Louck [Bie83] states:
Theorem. Suppose v is a vector with respect to the angular momentum L, i.e.,

[Li , vj ] = iijk vk , (10.33)

or, in vector notation,

v × L + L × v = 2iv. (10.34)
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch10

82 The Supersymmetric Dirac Equation

Suppose also that this vector is perpendicular to L,

L·v = v·L = 0. (10.35)
Then Knr anticommutes with the operator (σ·v), which is a scalar with respect to
the total angular momentum J, i.e., it commutes with J = L + 12 σ.
Proof.
{Knr , (σ·v)} = Knr (σ·v) + (σ·v)Knr

= −(σ·L + 1)(σ·v) − (σ·v)(σ·L + 1)

= −(σ·L)(σ·v) − (σ·v)(σ·L) − 2(σ·v)

= −iσ·(L × v + v × L) − 2(σ·v) = 0. (10.36)

In the framework of the conditions of the theorem we have also
i
Knr (σ·v) = σ·(v × L − L × v). (10.37)
2
Important special cases result when we take for the vector v either r̂ (unit
radial vector), p (linear momentum) or A (Laplace vector).

10.3 Pauli Spinors

A complete set of mutually commuting operators is {H, L2 , L3 , σ3 }. This set has
energy eigenstates of the form
|ψ >= |N lm > ⊗| 12 µ > . (10.38)
An alternative complete set is {H, J2 , J3 , 12 σ3 }, with J = L + 12 σ, and J3 =
L3 + 12 µ. In the position representation the states are
1
ψ(r, θ, φ) = RN l (r)Y (l 2 )jm (θ, ϕ), (10.39)
(l 12 )jm
because the Hamiltonian is rotationally invariant, where Y (θ, ϕ) are the
Pauli central field spinors: With the notation

 

1 1 1 1 1 0
1 
ξ =  2 , 2 >= −1 
, ξ =  2 , − 2 >= , (10.40)
0 1
1
the two-component spinor Y (l 2 )jm (θ, ϕ) is given by combining the angular
momenta:
1
  l
Y (l 2 )jm (θ, φ) = C l 12 j; m − µ, µ Yµ−m (θ, φ)ξ µ . (10.41)
µ
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch10

Non-Relativistic Hydrogenic Atoms with Spin 83

Here the Yml (θ, φ) are the spherical harmonics

l+m
2l + 1 (l − m)! (−eiφ sin θ)m d
l
Ym (θ, ϕ) = (cos2 θ − 1)l .
4π (l + m)! 2l l! d cos θ
(10.42)

The coefficients C(l 12 j; m − µ, µ) are the Clebsch–Gordan coefficients.

The choice of the sign in the following results is a consequence of the phase
conventions we make for the Clebsch–Gordan coefficients. The Clebsch–Gordan
table with the Condon–Shortley phase conventions is

j/m m= 1
2
m = − 12
 
1 l+m+ 12 l−m+ 12
j =l+ 2 2l+1 2l+1
 
l−m+ 12 l+m+ 12
j =l− 1
2 − 2l+1 2l+1

We now take as a complete set of mutually commuting operators

{H, Knr , J2 , J3 }. (10.43)
The Pauli central field spinors may now be fully characterized by a notation that
employs only κ and m:
1
χκm = Y [l(κ) 2 )]j(κ)m , (10.44)
where for each prescribed integer κ from the set (10.32) the quantum number m
takes on the values
1
m = |κ| − 12 , |κ| − 13 , . . ., −|κ| + . (10.45)
2

10.4 Concerning the Operator (σ·r̂)

Consider the effect of the operator (σ·r̂) on RN l (r)χκm . This operator has no effect
on RN l (r), it suffices to consider (σ·r̂)χκm . We want to establish
(σ·r̂)χκm = −χ−κ
m . (10.46)

We note that (σ·r̂) is a scalar operator with respect to rotations, so that (σ·r̂)χκm
has the same j and m as χκm . But since (σ·r̂) anticommutes with Knr (see theorem
of Section 10.2), the result of (σ·r̂)χκm must be proportional to χ−κm . That is,

(σ·r̂)χκm = cχ−κ
m . (10.47)
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch10

84 The Supersymmetric Dirac Equation

Since (σ·r̂) is an Hermitian operator, c is a real number. Using (10.15),

(σ·r̂)(σ·r̂) = 1, (10.48)

we have |c|2 = 1 and c = ±1. The sign is fixed by choosing r̂ = eˆ3 = (0, 0, 1),
which is equivalent to taking θ = 0. From Eq. (10.42)

2l + 1
Yml (ê3 ) = δm0 , (10.49)
4π
so we have

2l(κ) + 1
χκm (r̂) = C(l(κ) 12 j(κ); 0m) ξ m . (10.50)
4π
Equation (10.46) becomes
 
2m( 2lA + 1) C lA 12 jA ; 0m = c 2lB + 1 C lB 12 jB ; 0m , (10.51)

where lA = l(κ), jA = j(κ) and lB = l(−κ), jB = j(−κ). From this it follows

that
 
2lA + 1 C lA 12 jA ; 0m
c = 2m  . (10.52)
2lB + 1 C lB 21 jB ; 0m
1
When m = 2 we have 2m = 1, and

2lA + 1 C(lA 21 jA ; 0m)
c= = −1. (10.53)
2lB + 1 C(lB 21 jB ; 0m)

When m = − 12 we have 2m = −1, and


2lA + 1 C(lA 12 jA ; 0m)
c=− = −1. (10.54)
2lB + 1 C(lB 21 jB ; 0m)
In both cases we have c = −1, which proves (10.46). From (10.46) we get, by
applying (σ·r̂), also

(σ·r̂)χ−κ
m = −χm .
κ
(10.55)

10.5 The Key Equation: Concerning the Operator (σ·A)

Applying similar considerations to the operator Anr = (σ · A) leads to a key
equation of the whole book. We get an equation relating solutions of positive κ to
solutions of negative κ, which is the essence of the application of supersymmetry,
both in the relativistic and the non-relativistic cases. This equation also involves
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch10

Non-Relativistic Hydrogenic Atoms with Spin 85

an expression for the eccentricity, which is a key concept of the book, again for
the relativistic as well as the non-relativistic cases.
Using the definition

ψEκm = RE,l(κ) χκm , (10.56)

and the fact that (σ·A) anticommutes with Knr , one finds the result that

(σ· A)ψEκm = −α(E, κ)ψE,−κ,m , (10.57)

where α(E, κ) is a constant.

Now remember Eq. (8.42):

|A|2 + L2 + 1 = N 2 . (10.58)

We use also the factorization of L2 : L2 = Knr (Knr + 1), and

(σ · A) = |A|2 + iσ·A × A = |A|2 − σ·L = |A|2 + 1 + Knr . (10.59)

Here we have used

1 i
σ · (A × A) = ijk σk [Aj , Ak ] = ijk jkm σk Lm = iσ · L. (10.60)
2 2
Inserting this to Eq. (10.58) gives us

(σ·A)2 = N 2 − Knr
2
. (10.61)

Hence

(σ·A)2 ψEκm = α(E, κ)α(E, −κ)ψE,κm = (N 2 − κ2 )ψEκm , (10.62)

and

α(E, κ) = N 2 − κ2 . (10.63)
1
For A = (N 2 ) 2 A we get

(σ·A)ψN κm = − 1 − κ2 /N 2 ψN,−κ,m . (10.64)

We shall see, in Section 10.7, that this is the expression for the eccentricity.
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch10

86 The Supersymmetric Dirac Equation

10.6 The Factorization Method

In this section we find a factorization property of the operator Anr , which will
turn out to be equivalent to the results of the supersymmetric analysis in the next
chapter.
We use Eq. (10.37) to rewrite Anr = (σ·A) as
−i i i
(σ·A) = σ·(p × L − L × p) − σ·r̂ = (σ·p)Knr − σ·r̂.
Zαm 2 Zαm
(10.65)

We now express Anr in spherical polar coordinates:

σ·p = (σ·r̂)(σ·r̂)σ·p = σ·r̂ [r̂·p + iσ·(r̂ × p)]

   
= σ·r̂ r̂·p + ir −1 σ·L = σ·r̂ r̂·p − ir−1 (Knr + 1)
   
∂ Knr + 1 Knr
= −iσ·r̂ + = −iσ·r̂ ipr + , (10.66)
∂r r r
where, according to Eq. (9.6),
i ∂
pr = − r. (10.67)
r ∂r
Inserting this into Eq. (10.65) we obtain:

 
Knr Knr
Anr = (σ·A) = (σ·r̂) ipr + −1 . (10.68)
Zαm r
This result, when combined with Eqs. (10.64) and remembering the sign change
due to Eq. (10.46), leads directly to the radial differential equation in the form

 κ  κ  κ2
ipr + − 1 RN,l(κ) (r) = 1 − 2 RN,l(−κ) (r). (10.69)
Zαm r N
When κ is positive, this defines a lowering operator on l: l(κ) = κ, l(−κ) =
κ − 1 = l − 1,

  
l d l+1 l2
+ − 1 RN l (r) = 1 − 2 RN,l−1 (r). (10.70)
Zαm dr r N
When κ is negative it defines a raising operator on l: l(κ) = |κ| − 1, κ = −(l + 1),
l(−κ) = |κ| = l + 1,

  
(l + 1) d l (l + 1)2
− − − 1 RN l (r) = 1 − RN,l+1 (r). (10.71)
Zαm dr r N2
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch10

Non-Relativistic Hydrogenic Atoms with Spin 87

These equations may be written as

A−
l+1 RN,l+1 (r) = gl+1 RN l (r), A+
l+1 RN l = gl+1 RN,l+1 (r), (10.72)

with

 12
1 1 l Zαm
gl = Zαm − 2 , A∓
l = ±ipr + − , (10.73)
l2 N r l
and A+ = (A− )† .
Applying the raising operator to RN,l (r), and then the lowering operator to
the resulting RN,l+1 (r), yields

A− + 2
l+1 Al+1 RN l = g(l+1) RN l , (10.74)

or
  
d (l + 2) Zαm d l Zαm
+ − − + − RN l (r)
dr r l+1 dr r l+1
 
1 1
= (Zαm)2 − RN l (r). (10.75)
(l + 1)2 N2
This can be simplified to
 
1 d 2 d l(l + 1) 2Zαm
r − + − 2mEN RN l (r) = 0, (10.76)
r2 dr dr r2 r
with EN given by Eq. (8.48). We have recovered the Schrödinger equation for the
radial part of the Pauli wave function. But since it has now been factorized into the
product of two factors, which raise and lower l without changing the energy, we
can now explain the degeneracy in l, and the degeneracy 2N 2 of an N -multiplet.
The tower of states terminates when A+ l+1 RN l = 0, that is, when gl+1 = 0.
From Eq. (10.73) we see that this implies N = l+1. From the equation A+ N RN l = 0
we further have
 
d N −1 1
− + − RN,N −1 (r) = 0, (10.77)
dr r N
and hence
 N −1
2(Zαm)r
e−
Zαm
RN,N −1 (r) = NN,N −1 N r . (10.78)
N
The radial wave function is nodeless. These orbits are called “circular”. For the
ground state (N = 1) we have

R10 (r) = (constant)e−(Zαm)r . (10.79)

August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch10

88 The Supersymmetric Dirac Equation

All other eigenfunctions in the N-multiplet can be generated from the explicit
result for the eigenfunction with l = N − 1. Taking κ = N − 1 (since l(N − 1) =
N − 1) we can use this same eigenfunction in Eq. (10.70), which now functions
as a lowering operator, generating the eigenfunction for l = N − 2. Iterating this
procedure yields all the eigenfunctions in an N-multiplet.
To summarize: The properties of the operator Anr are equivalent to a definition
of the radial eigenfunctions; moreover, the eigenvalue conditions deducible from
Anr → 0 are equivalent to the eigenvalue condition (obtained from analysis) for the
confluent hypergeometric function 1 F1 (a, b; z). The eigenvalue condition referred
to above is the condition that the confluent hypergeometric function reduces to a
polynomial for the parameter a being zero or a negative integer.

10.7 The Definition of Eccentricity

In the spin- 12 case the vector p is replaced by the 2 × 2 matrix (σ·p). In the same
way, the appropriate expression for the Laplace vector A becomes Anr = (σ·A).
The eccentricity is, in the classical case, the magnitude of the Laplace vector A.
In the case of a spin- 12 particle the vector A is replaced by the quantity (σ·A). So
the corresponding statement, for spin- 12 quantum particles, is that the eccentricity
is the magnitude of the quantity (σ·A).
Since

e ψN κm = (σ·A)2 ψN κm = 1 − κ2 /N 2 ψN κ (10.80)

we have

e= 1 − κ2 /N 2 , (10.81)

with κ = ±1, ±2, . . ., ±(N − 1), −N . This corresponds to the operator equation
1
eop = [1 − Knr
2
/N 2 ] 2 . (10.82)

The circular orbits have eccentricity zero, or κ = −N , and l(κ) = N − 1. Then we

have the statement, both classically and in quantum mechanics, that the circular
orbits are those with vanishing eccentricity.
Figure 10.1 is a picture of the orbits of hydrogen according to quantum
mechanics. The circular orbits for N = 1, 2, 3 are shown. The orbits with non-
vanishing eccentricity are seen to deviate from symmetry, to a degree proportional
to their eccentricity.
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch10

Non-Relativistic Hydrogenic Atoms with Spin 89

Figure 10.1. Low-lying orbits of the hydogen atom.

August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch10

90 The Supersymmetric Dirac Equation

Notes on Chapter 10
This chapter follows essentially the work of Biedenharn and Louck [Bie62].
The importance of Clifford algebras for the description of particles with spin is
especially emphasized in geometrical algebra; see Doran and Lasenby [DL03].
Spinors are discussed, at an elementary level, by Bade and Jehle [BJ53], and in the
framework of geometric algebra, by Doran and Lasenby [DL03] and Francis and
Kosowsky [FK05]. The factorization method was developed by Infeld [Inf41] and
Hull and Infeld [HI51].
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch11

Chapter 11
Elements of Supersymmetric Quantum Mechanics

The theory of supersymmetry has its origins in quantum field theories [SDB04]
and elementary particle physics [WZ74], but it is also a valuable tool for describing
the solutions of many ordinary quantum mechanical systems [Wit90], [RLB90]. It
is used, for example, to provide a simple proof of the positive-mass conjecture in
general relativity [Wit81], and in the proof of various index theorems in differential
geometry [Tak08]. The following is a summary of the most pertinent definitions
that we need in our discussion of supersymmetry in non-relativistic hydrogenic
atoms and, later, in the Dirac–Kepler problem.

11.1 General Considerations

Assume that we have a vector space V that is the direct sum of two vector spaces
V1 and V2 :

V = V1 ⊕ V2 , (11.1)

such that the vectors of V1 are even with respect to some parity P, while those of
V2 are odd. A general vector in V may then be written as the sum of an even and
an odd vector:




|ν1  |ν1  0
|ν = = + . (11.2)
|ν2  0 |ν2 

The vector space V is called Z2 -graded.

Similarly, a linear operator in the space V may be written as the sum of an
even and an odd operator:




Ω11 Ω12 Ω11 0 0 Ω12
Ω= = + . (11.3)
Ω21 Ω22 0 Ω22 Ω21 0

91
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch11

92 The Supersymmetric Dirac Equation

The element Ω11 effects an even-even transition. Similarly, the element Ω22 is
odd-odd. The element Ω12 describes the transformation of an odd state into an
even one, the element Ω21 the transformation of an even state into an odd one. The
linear operators form a graded algebra. The even and the odd operators are called
homogeneous.
The set of linear operators define a superalgebra if it is closed under a
generalized commutator, which, for two homogeneous operators Ω1 and Ω2 , is
the ordinary commutator between Ω1 and Ω2 unless both elements are odd. In
the latter case, it is the anticommutator between Ω1 and Ω2 . The definition of the
generalized commutator is extended to arbitrary operators by requiring it to be a
bilinear operation.
Consider the so-called s(2) superalgebra defined by the supercharges



1 0 A+ i 0 A+
Q1 = √ − , Q2 = √ − , (11.4)
2 A 0 2 −A 0

and the supersymmetric Hamiltonian

+ − 

A A 0 H+ 0
H= = , (11.5)
0 A− A+ 0 H−

where the operators A+ and A− are the Hermitian adjoints of each other. The
operators


1 0 0
Q = √ (Q1 + iQ2 ) = (11.6)
2 A− 0

and


1 0 A+
Q† = √ (Q1 − iQ2 ) = (11.7)
2 0 0

are the supersymmetry generators. The parity operator that is responsible for the
grading is represented by the operator


1 0
P= . (11.8)
0 −1

By explicit matrix multiplication we have



i −A+ A− 0 i A + A− 0
Q1 Q2 = , Q2 Q 1 = .
2 0 A− A+ 2 0 −A− A+
(11.9)
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch11

Elements of Supersymmetric Quantum Mechanics 93

From this {Q1 , Q2 } = 0 and similarly Q21 = Q22 = 12 H. We also have

{Q, Q} = {Q† , Q† } = 0, and {Q, Q† } = Q21 + Q22 = H. (11.10)

Therefore

[Q1 , H] = [Q1 , Q21 + Q22 ] = [Q1 , Q22 ] = −Q2 {Q1 , Q2 } + {Q1 , Q2 }Q2 = 0.
(11.11)

Similarly [Q2 , H] = 0.
Thus, the supersymmetric Hamiltonian is the sum of the squares of the
supersymmetric charges, or the anticommutator of the supersymmetry generators.
These are relations that characterize supersymmetric quantum mechanics, with the
proviso that H+ and H− be closely related to the Hamiltonians of actual physical
systems. We say that a system which satisfies these relations is an s(2) superalgebra,
and we speak of an S(2) supersymmetry.
Assume now that the vector |ν defined by Eq. (11.2) is the eigenvector of the
supersymmetric Hamiltonian with eigenvalue E:





|ν1  H+ 0 |ν1  |ν1 
H = =E . (11.12)
|ν2  0 H− |ν2  |ν2 

|ν1  is then an eigenstate of the Hamiltonian H+ with eigenvalue E, and |ν2  is an

eigenstate of H− with the same eigenvalue.
Assuming that |ν1  and |ν2  are normalizable vectors, we have

|A− |ν1 |2 = ν1 |A+ A− |ν1  = A− ν1 |A− ν1  = Eν1 |ν1 ,
|A+ |ν2 |2 = ν2 |A− A+ |ν2  = A+ ν2 |A+ ν2  = Eν2 |ν2 , (11.13)

where we have used that A+ and A− are the Hermitian adjoints of each other.
These relations show that E is positive or zero. They also show that A− |ν1  and
A+ |ν2  vanish if and only if E is zero.
Therefore the vectors




+ 
|ν1  0 † 0 A |ν2 
Q = , Q = (11.14)
0 A− |ν1  |ν2  0

vanish if E is zero. Otherwise, they are also eigenvectors of H with eigenvalue E,

because Q and Q† commute with H. Thus A− |ν1  is an eigenvector of H− with
energy E and A+ |ν2  is an eigenvector of H+ , also with energy E.
Any eigenstate of H+ with positive energy has, accordingly, a partner that is
an eigenstate of H− with the same energy and vice versa. The eigenstates of H+
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch11

94 The Supersymmetric Dirac Equation

and H− with positive energy are paired. They are transformed into each other by
the operators A− and A+ . Thus the levels of the super-Hamiltonian are doubly
degenerate. If, however, either H+ or H− has an eigenstate with energy zero, then
there is no partner state and, hence, no degeneracy.
We have seen that an eigenstate of H with energy zero is annihilated by Q and
Q† . If, conversely, an eigenstate of H is annihilated by Q and Q† , then its energy
is zero. This follows from Eq. (11.10), according to which {Q, Q† } = H.
For supersymmetry to be an exact symmetry, one requires that H have a
supersymmetrically invariant ground state, i.e., a state that is annihilated by Q and
Q† . Such a ground state has zero energy, and either |ν1  or |ν2  is zero.
We have thus seen that if supersymmetry is an exact symmetry, then the two
Hamiltonians H+ and H− have the same set of eigenvalues, except that one of
them has a zero eigenvalue that is not shared by the other Hamiltonian.
This completes our summary of elementary supersymmetric quantum
mechanics. We shall now turn to its application to non-relativistic hydrogenic atoms
with spin.

11.2 Supersymmetry of Non-Relativistic Hydrogenic Atoms

One way of applying supersymmetry to non-relativistic hydrogen atoms is to
consider the Z(2) grading to be just spin-up or spin-down. In other words an
electron is written as a two-component entity with



1 − 0
|ψ  =
+
, |ψ  = . (11.15)
0 1

This corresponds to a grading with P3 = σ3 .

Consider the Pauli Hamiltonian for an atom placed in a magnetic field along
the z-axis: B = Be3 . This field is generated by a magnetic potential A(x, y) =
(Ax , Ay , 0). The Pauli Hamiltonian restricted to the (x, y)-plane is

1
Hmag = (π 2 + πy2 ) − gµB BS3 , (11.16)
2m x

where πx , πy stand for the abbreviations

πx = px − eAx , πy = py − eAy , (11.17)

g is the gyromagnetic factor and µB is the Bohr magneton.

August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch11

Elements of Supersymmetric Quantum Mechanics 95

Introduce the supercharges

1
Q1 = √ (−πy σ1 + πx σ2 ),
2 m
1
Q2 = √ (πx σ1 + πy σ2 ). (11.18)
2 m
We may now calculate, using the properties of σi , Eq. (10.6),
1
Q 1 Q2 = (−πy πx + πx2 σ2 σ1 − πy2 σ1 σ2 + πx πy ),
4m
1
Q 2 Q1 = (−πx πy − πy2 σ2 σ1 + πx2 σ1 σ2 + πy πx ). (11.19)
4m
These two equations may be expressed more compactly as
1
{Q1 , Q2 } = (π 2 − πy2 ){σ1 , σ2 } = 0. (11.20)
4m x
We also have
1 1
Q21 = (π 2 + πy2 ) − [πx , πy ]σ1 σ2 ,
4m x 4m
1 1
Q22 = (π 2 + πy2 ) + [πx , πy ]σ2 σ1 . (11.21)
4m x 4m
Since σ1 σ2 = −σ2 σ1 we have Q21 = Q22 . For the commutators we have,
according to Eq. (10.19):

[πx , πy ] = ieB. (11.22)

Inserting this in the foregoing equations, then the comparison with the formula for
H, Eq. (11.16) — and the demand that g = 2 — leads to the equation Q21 = Q22 =
2 H. It follows that H commutes both with Q1 and Q2 . Together with Eq. (11.20)
1

we can summarize this as

{Qi , Qj } = Hδij , and [H, Qi ] = 0. (11.23)

In terms of Q, Q† we have

{Q, Q} = {Q† , Q† } = [H, Q] = [H, Q† ] = 0, {Q, Q† } = H. (11.24)

Thus we have the supersymmetry algebra s(2). We may say that the requirement
g = 2 is equivalent to demanding that the Pauli Hamiltonian, together with
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch11

96 The Supersymmetric Dirac Equation

the supercharges, form a supersymmetric algebra. We have attempted to make

clear that it is in any case a very reasonable requirement. It does not require
the Dirac equation for its justification, contrary to what is often claimed in the
literature.
Another application is a Z2 grading introduced in the Hilbert space of states
by classifying the states to be even or odd with respect to the parity operator
Pκ = Knr /|κ|, i.e., having eigenvalues ±Sign(κ), respectively. Equally, linear
operators can be assigned a grade. An operator is even if it commutes with Pκ ,
whereas operators anticommuting with Pκ are called odd. As an example of an even
operator we mention H, following from the fact that Knr is built from symmetry
operators of H. Of course, Knr itself is even as well. According to the Theorem
10.2, for any vector v that is perpendicular to L, the J scalar (σ·v) anticommutes
with Knr , and is hence an odd operator. This theorem supplies us with odd operators
(σ·v), with v equal to, for example, p, r or A.
The supercharges communicate between the subspaces with Sign(κ) = ±1,
and can be taken to be (σ·A). Since (σ·A) is Hermitian A+ = A− = (σ·A).
Then we have



0 0 † 0 (σ·A)
Q= , Q = . (11.25)
(σ·A) 0 0 0

When Q acts on the vector |ν then we have (σ · A) acting on the positive κ
component, so that (σ·A) lowers l(κ) by one unit, whereas Q† acts on the negative
κ component, and raises l(κ) by one unit.
We also have





|ν1  † |ν1  2 |ν1  |ν1 
H = {Q, Q } = (σ·A) =E , (11.26)
|ν2  |ν2  ν2 |ν2 

with
 
κ2
E= 1− 2 , (11.27)
N

see Eq. (10.81).

In general an eigenstate of (σ·A)2 with positive energy has a partner of the
same energy, but with opposite value of κ. Thus, the levels of the super-Hamiltonian
are doubly degenerate. If, however, there is an eigenstate with E = 0 then there
is no partner state and no degeneracy. Supersymmetry is exact when H has a
supersymmetrically invariant ground state, that is, a state that is annihilated by
Anr = (σ·A).
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch11

Elements of Supersymmetric Quantum Mechanics 97

0.0
E [eV]

-13.6
l=0 l=1 l=1 l=2
j=1/2 j=3/2

Figure 11.1. The supersymmetry of the non-relativistic hydrogen spectrum.

The constructed S(2) supersymmetry has implications for the spectrum. The
consequences of a (super-) symmetry group of the Hamiltonian is that the energy
eigenspaces consist of irreducible representations of the group. The irreducible
representations of S(2) are either one- or two-dimensional. States within the
subspace Hκ , with fixed j and m, transform irreducibly under the superalgebra
s(2). The multiplets have dimensionality two, unless E = 0, in which case they are
one-dimensional.
We can now interpret the spectrum of a non-relativistic spin- 12 particle in a
Coulomb field (see Figure 11.1):

1. The degeneracies of levels with fixed j and m but with l(κ) = j ± 12 are a
consequence of the s(2) supersymmetry algebra, constructed above. The level
of degeneracy is two, except for states with l = j − 12 and

m (Zα)2
E=− , (11.28)
2 (l + 1)2

which are nondegenerate.

2. The degeneracies of fixed l and m but j = l ± 12 are due to kinematic
independence of the electron spin in the non-relativistic regime.
3. The (2j + 1)-fold degeneracy of states with fixed j and l is due to rotational
invariance of the interaction.
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch11

98 The Supersymmetric Dirac Equation

Notes on Chapter 11
The relation of the factorization method to supersymmetry has been emphasized
by Stahlhofen and Bleuler [SB89]. The application of supersymmetry to the non-
relativistic hydrogen spectrum was performed by Tangerman and Tjon [TT93]. The
various applications of supersymmetry to quantum mechanics are reviewed in Roy,
Lahiri and Bagchi [RLB90] and Khare, Cooper and Sukhatma [KCS95].
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch12

Chapter 12
Sommerfeld’s Derivation of the Relativistic
Energy Level Formula

Sommerfeld [Som16] developed in 1916 a generalization of the Bohr model,

which allowed for elliptical orbits and was consistent with the theory of relativity.
In the classical relativistic extension of the problem of the motion of a particle
in a Coulomb field the orbits are precessing ellipses. Applying the quantization
conditions to these orbits leads to the Sommerfeld formula for the energy
levels.
This was done by Sommerfeld in an attempt to explain the fine structure of
the hydrogen atom. The fine structure is a splitting of the spectral lines into several
distinct components, which is found in all atomic spectra. It can be observed only by
using equipment of very high resolution since the separation, in terms of reciprocal
wavelength, between adjacent components of a single spectral line is of the order
of 10−4 times the separation between adjacent lines. This must mean that what we
had thought was a single energy state of the hydrogen atom actually consists of
several states that are very close together in energy.
The surprising feature of this formula is its agreement with the result of the
full Dirac theory. This despite the fact that Sommerfeld used the methods of
the “Old Quantum Theory”, and that his treatment completely neglects the fact
that the electron is a particle with spin. The agreement of these two formulas,
despite their different origins, is referred to as the “Sommerfeld puzzle”. In the
subsequent chapters we shall look at the Dirac description of the spinning electron,
and discuss the resolution of the puzzle, following the treatment of Biedenharn
[Bie83].
In the present chapter we present Sommerfeld’s model. It is a direct
generalization of the treatment of the non-relativistic case treated in Chapter 7.

99
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch12

100 The Supersymmetric Dirac Equation

12.1 Assumptions of the Model

The relativistic Hamiltonian for an attractive Coulomb potential −(Zα)/r, has the
quadratic form

2
Zα
H+ = p2 + m2 . (12.1)
r
2
This can be interpreted as the relativistic energy-momentum relation Ekin − p2
2
=m .
The momentum p can be resolved classically into two components:

p2θ
p2 = p2r + , (12.2)
r2
where the radial momentum is
mṙ
pr = √ , (12.3)
1 − v2
and the angular momentum is

mr2 θ̇
pθ = √ . (12.4)
1 − v2
From angular momentum conservation, the component pθ is a constant of the
motion. We define as before the variable s = 1/r, and note that


pr ds
=− . (12.5)
pθ dθ
Introducing these variables into Eq. (12.1), we find that


H + Zαs
2  p 2
ds 2 
θ
=1+ + s2 . (12.6)
m m dθ

As before, one differentiates Eq. (12.6) with respect to θ and obtains the linear
differential equation

d2 s
+ γ 2 (s − C) = 0, (12.7)
dθ 2
where we have used the abbreviations
(Zα)2
γ2 = 1 − (12.8)
p2θ
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch12

Sommerfeld’s Derivation of the Relativistic Energy Level Formula 101

and


ZαH
C= . (12.9)
γ 2 p2θ

From Eq. (12.7) we obtain the general solution for the relativistic orbits:
1
s= = A cos (γθ) + B sin (γθ) + C, (12.10)
r
where A, B are the integration constants and C is given in Eq. (12.9). Defining the
distance of closest approach (perihelion) to occur at θ = 0 makes B = 0, yielding
the simpler equation
1
= A cos (γθ) + C. (12.11)
r
This result shows that the classical Kepler orbits have the form of conic
sections — just as in the non-relativistic case — but that the correct angle variable
(in terms of which the orbit is actually a conic section) is not θ but ψ = γθ. Thus,
for elliptic orbits (bound states) to move from one perihelion (ψ = 0) to the next
(ψ = 2π) requires θ = 2π/γ. Since the parameter γ is less than unity, classical
bound state relativistic Kepler motion takes place in an ellipse whose perihelion is
advancing (that is, moving in the same sense as the orbit is transversed).
We now express the elliptic orbits in geometrical terms, using the eccentricity e.
For ψ = 0 one has the perihelion distance r = a(1−e) and, for ψ = π, the aphelion
distance r = a(1 + e). The orbit equation, Eq. (12.11), is then


1 1 1 + e cos ψ
= . (12.12)
r a 1 − e2

Except for the occurrence of ψ = γθ in place of θ, one sees that the orbits are
exactly of non-relativistic form. In other words, in a moving reference frame, such
that the perihelion appears fixed, the orbit is non-relativistic in form.
It is now easy to derive Sommerfeld’s formula by applying the Sommerfeld–
Wilson quantization rules to the phase integrals for pθ and pr :
 θ=2π
pθ dθ = 2πnθ , (12.13)
θ=0

which yields immediately pθ = nθ , and


pr dr = 2πnr . (12.14)
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch12

102 The Supersymmetric Dirac Equation

To evaluate the integral in Eq. (12.14) rewrite the radial momentum,



mṙ m dr pθ dr
pr = √ = √ θ̇ = 2 , (12.15)
1 − v2 1 − v 2 dθ r dθ

so that we obtain

2
1 dr e2 γ sin2 ψ
pr dr = pθ dθ = pθ dψ, (12.16)
r dθ (1 + e cos ψ)2

upon using the orbit Eq. (12.12). Thus, the radial quantization condition,
Eq. (12.14), takes the form

  ψ=2π 2 2
1 e sin ψ dψ nr
2
= . (12.17)
2π ψ=0 (1 + e cos ψ) γn θ

Notice that the left-hand side of Eq. (12.17) is identical to that for non-
relativistic motion, since the integral involves only the geometry of the ellipse.
It has the value
 ψ=2π 2 2
1 e sin ψ dψ
= (1 − e2 )− 2 − 1.
1
(12.18)
2π ψ=0 (1 + e cos ψ)2

By contrast, the right-hand side of Eq. (12.17) differs from the non-relativistic case
solely by the appearance of the factor γ. We may phrase this more suggestively
by saying that the sole effect of relativity in the radial quantization condition
consists in replacing the orbital angular momentum, nθ , by the “irrational angular
momentum” parameter, γnθ .

12.2 The Energies of the Bound States

It is now straightforward to obtain the formula for the energy levels. The two
parameters of the orbit equation (a and e in Eq. (12.12)) are determined from the
quantization conditions:

1. From Eq. (12.18) we know the eccentricity, e:

2
1 nr
= 1+ . (12.19)
1 − e2 γnθ

2. From Eq. (12.12) we have

A = eC. (12.20)
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch12

Sommerfeld’s Derivation of the Relativistic Energy Level Formula 103

We can then determine H from Eqs. (12.6) and (12.11). We have

H 2 + (Zα)2 C 2 + 2H(Zα)C
+ [2(Zα)2 AC + 2H(Zα)A] cos (γθ) + (Zα)2 A2 cos2 (γθ)


(Zα)2
= m + nθ A 1 −
2 2 2
+ C 2 + 2AC cos (γθ)
n2θ

A2 (Zα)2 2
+ cos (γθ) . (12.21)
n2θ
The coefficients of the terms quadratic in cos (γθ) agree, they are (Zα)2 A2 . The
coefficients of the terms linear in cos (γθ) agree, because


(Zα)2
HZα = n2θ 1 − C = n2θ γ 2 C. (12.22)
n2θ
The constant terms yield
 
2 (Zα)4 2(Zα)2 (Zα)2

H 1+ 4 4 + 2 2 = m2 + H 2 2 4 γ 2 e2 + 1 , (12.23)
nθ γ nθ γ nθ γ
or
(Zα)2  m 2
1+ [(1 − γ 2
) + 2γ 2
− γ 2 2
e − 1] = . (12.24)
n2θ γ 4 H
This simplifies to
(Zα)2 2  m 2
1+ 2 γ (1 − e 2
) = , (12.25)
nθ γ 4 H
and, inserting (1 − e2 ) from Eq. (12.18):
(Zα)2  m 2
1+ = . (12.26)
(nθ γ + nr )2 H
Equivalently, for orbits for which H = E = constant,

− 12
(Zα)2
E =m 1+ . (12.27)
(nr + γnθ )2
This is the Sommerfeld formula for the energy levels of the bound states in the
relativistic regime.
For the ground state, nθ = 1, the Lorentz transformation factor γ would appear
in a transformation from the rest frame to a frame moving, at a given instant, in a
direction tangent to the orbit with a velocity v = Zα. Such a Lorentz transformation
gives rise to a time dilation between the fixed and transformed systems by the factor
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch12

104 The Supersymmetric Dirac Equation

[1 − v 2 ]− 2 = [1 − (Zα)2 ]− 2 . This is the factor denoted by 1/γ in Eq. (12.8),

1 1

with pθ = nθ = 1. The deviation of γ from one is a measure of the magnitude of

relativistic effects, with 1/γ ≥ 1. Thus the angular velocity θ̇ appears to be smaller
(from the time dilation), or, equivalently, the angle ψ in the moving system appears
at each instant to be smaller, that is ψ = γθ.
It turns out that the quantum version of this Lorentz transformation, when
applied to the operator Γ (see Chapter 15), yields the diagonalized form with the
eigenvalues ±γ|κ|. This is the basis of Biedenharn’s analysis of the Dirac equation
[Bie62], and is another hint as to why the Sommerfeld derivation yields the correct
spectrum.
Not all possible transitions between the energy levels in the spectrum actually
occur, they have to obey certain selection rules, e.g.,

nθi − nθf = ±1, (12.28)

where nθi is the azimuthal quantum number in the initial state, and nθf in the final
state. The “Old Quantum Theory” cannot explain these selection rules, because it
makes no predictions about the probability of transitions, or about the intensities
of spectral lines. Neither can it explain the spectra of more complex atoms: It fails
even for the next more complicated atom, namely helium with two electrons.
To get these aspects it is necessary to use the complete quantum theory. We
thus turn to the Dirac theory in the next chapter.
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch13

Chapter 13
The Dirac Equation

We are now ready to tackle the relativistic problem of an electron moving rapidly
in a Coulomb potential, using the full power of modern quantum mechanics. The
appropriate formalism was developed by Dirac [Dir28]. In the present chapter we
review the standard approach to this problem: The separation of variables in
spherical polar coordinates and the derivation of differential equations for the
radial components of the wave function. We emphasize that the treatment is a
direct generalization of the treatment of a non-relativistic spin- 12 particle, with
the three-dimensional Clifford algebra replaced by the four-dimensional Clifford
algebra, corresponding to a replacement of the two-component Pauli spinors by
four-component Dirac spinors ([BJ53], [FK05]).

13.1 The Hamiltonian

In the same way as the momentum of a free non-relativistic spin- 12 particle is
described by the expression (σ · p), the momentum of a free relativistic spin- 12
particle is described by γ µ pµ , µ = 0, 1, 2, 3. The γ µ are 4 × 4 matrices which
satisfy

{γ µ , γ ν } = γ µ γ ν + γ ν γ µ = 2g µν , (13.1)

with g µν the Minkowsky metric: gµν = diag(1, −1, −1, −1). The γ µ are the
generators of a four-dimensional Clifford algebra in Minkowski space, in the same
way as the σi are the generators of a three-dimensional Clifford algebra in Euclidian
space, {σi , σj } = 2δij .
The equation of motion for a free particle with mass m is given by

γ µ pµ ψ(t, r) = ±mψ(t, r), (13.2)

105
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch13

106 The Supersymmetric Dirac Equation

as is readily seen by applying γµ pµ twice, and using

1 µ ν
γ µ γ ν pµ pν = {γ , γ }pµ pν = p2 = m2 , (13.3)
2
which is the relativistic energy-momentum relation. With pµ = i∂µ , and choosing
the plus sign, yields the Dirac equation:

(iγ µ ∂µ − m) ψ(t, r) = 0, (13.4)

where we identify the parameter m with the mass. Choosing the minus sign would
yield

(iγ µ ∂µ + m) ψ(t, r) = 0, (13.5)

which is the Dirac equation for a particle with negative mass −m. We shall consider
the Eq. (13.5) to be unphysical. It will nevertheless be used in connection with the
unphysical Hamiltonian H̄ of Eq. (15.1) below.
The gamma matrices are given (in the Dirac representation) by



I2 0 0 σ
γ0 = , γ= , (13.6)
0 −I2 −σ 0

with I2 the 2 × 2 unity matrix, and σ the Pauli matrices. We shall also need the
matrix


0 I2
γ5 = iγ 0 γ 1 γ 2 γ 3 = , (13.7)
I2 0

with γ52 = I4 . This matrix anticommutes with the others: {γ5 , γ µ } = 0, µ =

0, . . . , 3.
The Dirac equation is cast into a Hamiltonian form by multiplying by γ 0 and
introducing the new matrices


0 σ
β = γ0, α = γ0γ = . (13.8)
σ 0

Note that the new matrices are Hermitian. They satisfy

{αi , αj } = αi αj + αj αi = 2δij , {αi , β} = 0, (13.9)

with i, j = 1, 2, 3, and

(αi )2 = β 2 = I4 . (13.10)
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch13

The Dirac Equation 107

The Dirac equation becomes, upon multiplying from the left by β:

∂ψ
i = [−iα·∇ + βm] ψ(t, r). (13.11)
∂t
Hence we get the Hamiltonian equation
∂ψ
i = Hψ, (13.12)
∂t
with

H = −iα· ∇ + βm. (13.13)

For the stationary states ψ(t, r) = ψ(r)e−iEt , where E is the energy of the state,
and Eq. (13.12) reduces to

Hψ(r) = Eψ(r). (13.14)

Including a spherically symmetric potential the Hamiltonian becomes

H = α·p + βm + V (r), (13.15)

with

p = −i∇. (13.16)

13.2 Total Angular Momentum

The spin matrices are


σ 0
Σ= . (13.17)
0 σ

We have the relation between the spin matrices and the quantities α:

α = γ5 Σ = Σγ5 . (13.18)

Neither the orbital angular momentum L = r × p,

[H, Li ] = [α·p, Li ] = αj [pj , Li ] = −iijk αj pk , (13.19)

nor the spin vector are conserved:

[H, Σi ] = [αj pj , Σi ] = γ5 [Σj , Σi ]pj = 2ijik γ5 Σk pj = 2ijik αk pj ,

(13.20)
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch13

108 The Supersymmetric Dirac Equation

but the total angular momentum J = L + 12 Σ is conserved. We have used here

[Li , pj ] = iijk pk , (13.21)

which expresses the vector transformation property of p with respect to spatial

rotations. Further, relating α to Σ by Eq. (13.18), we can use the commutator

[Σi , Σj ] = 2iijk Σk , (13.22)

which expresses the so(3) structure of the Pauli matrices.

13.3 The Dirac Operator

We consider next the Dirac operator, introduced by Dirac [Dir28]


−(σ·L + 1) 0
K = −β(Σ·L + 1) = . (13.23)
0 σ·L + 1

It is Hermitian. In the following we shall use the identity

(Σ·A)(Σ·B) = A·B + iΣ·(A × B), (13.24)

which holds for any two vectors A and B which commute with Σ. This is a
generalization of the identity (10.15).
The Dirac operator is a constant of the motion. Indeed, it commutes with the
Hamiltonian for an arbitrary central potential V(r).

[H, K] = −[K, H] = [β(Σ·L + 1), α·p + βm + V (r)]

= [β(Σ·L), α·p] + [β, α·p]
= β(Σ·L)(α·p) − (α·p)β(Σ·L) + 2β(α·p)
= β(Σ·L)(α·p) + β(α·p)(Σ·L) + 2β(α·p)
= βγ5 [(Σ·L)(Σ·p) + (Σ·p)(Σ·L) + 2(Σ·p)]
= βγ5 [iΣ·(L × p + p × L) + 2Σ·p], (13.25)

using the identity (13.24). Now use L × p + p × L = 2ip to get

[H, K] = βγ5 [−2Σ·p + 2Σ·p] = 0. (13.26)

The total angular momentum commutes with the Dirac operator:

[J, K] = −[J, β(Σ·L + 1)] = −β[J, Σ·L] = 0. (13.27)

August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch13

The Dirac Equation 109

There is a relation between the orbital angular momentum squared and the
Dirac operator squared:

L2 = (Σ·L + 1)(Σ·L) = K(K + β). (13.28)

This is the generalization of the non-relativistic relation of Eq. (10.28). There is

also a relation between the total angular momentum squared and the Dirac operator
squared:

2
1 3
J2 = L + Σ = L2 + Σ·L +
2 4





3 1 1 1
= Σ·L + Σ·L + = βK + βK − , (13.29)
2 2 2 2
so
1
J2 = K 2 − . (13.30)
4
This is the relativistic generalization of Eq. (10.29).
Let the eigenvalues of K be κ. Applied to a function for which K is a good
quantum number, the relation (13.30) yields

2
1 1
κ2 = j (j + 1) + = j + , (13.31)
4 2
and κ takes on two values for definite j,
 
 1 
κ = ± j + , (13.32)
2
which generalizes Eq. (10.31).
The parity in the Dirac theory is

P = βI, (13.33)

where I is the inversion operator which inverts the position and momenta,

I(r) = −r, I(p) = −p. (13.34)

It is Hermitian, and it commutes with the Hamiltonian:

[H, P ] = [H, βI] = H(p)βI − βIH(p) = βH(−p)I − βH(−p)I = 0.

(13.35)

The Dirac operator commutes with the parity:

[K, P ] = −[β(Σ·L + 1), βI]

= −β[Σ·L, βI] + [β, βI](Σ·L + 1) = −β 2 [Σ·L, I] = 0. (13.36)
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch13

110 The Supersymmetric Dirac Equation

The total angular momentum also commutes with the parity operation:

[J, P ] = [J, βI] = β[J, I] = 0. (13.37)

13.4 A Complete Set of Mutually Commuting Operators

We now have a complete set of mutually commuting operators:

{H, J2 , J3 , K}. (13.38)

To specify the parity of the states is superfluous, since its eigenvalues are known
when those of {H, J2 , J3 , K} are given. We denote the eigenvectors, which we
assume are normalized to unity, by

|, j, mj , κ
, (13.39)

where  is the reduced energy

E
=
, (13.40)
m
and the quantum numbers j, mj , κ are defined by

J 2 |, j, mj , κ
= j(j + 1)|, j, mj , κ
,

J3 |, j, mj , κ
= mj |, j, mj , κ
, (13.41)
K|, j, mj , κ
= κ|, j, mj , κ
.
The wave function is the representation of these states in the position representation:
ψ(r) = r|, j, mj , κ
.
We now write the four-component wave function in terms of the two upper
components and the two lower ones:


ψA (r)
ψ(r) = . (13.42)
ψB (r)
Applying the Dirac operator yields for the upper components

κ = −σ·L − 1, (13.43)

and using J = L + 12 σ we find

3
J2 = L2 + σ·L + , (13.44)
4
or
3
σ·L = J2 − L2 − , (13.45)
4
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch13

The Dirac Equation 111

and Eq. (13.43) becomes

1
κ = L2 − J2 − . (13.46)
4
If σ·L is a good quantum number so is L2 , from Eq. (13.45). So
1
κ = lA (lA + 1) − j(j + 1) − . (13.47)
4
For lA = j + 12 we find κ = lA , for lA = j − 12 we find κ = −lA − 1. By
vector addition of angular momentum these are the only two possibilities. The
result is in agreement with Eq. (13.32). We see that lA = j + 12 Sign(κ), with
Sign(κ) = κ/|κ|.
For the lower components κ = σ ·L + 1, and κ = lB + 1 for lB = j − 12 ,
κ = −lB for lB = j + 12 , and lB = j − 12 Sign(κ). We have lA − lB = Sign(κ).
The cases where j = lA,B + 1/2 are referred to as “parallel coupling”, the cases
where j = lA,B − 1/2 as “antiparallel coupling”.
The spatial parity of ψA (r) is (−1)lA , the spatial parity of ψB (x) is (−1)lB =
−(−1)lA . But the conserved quantity is the total parity P = βI. The total parity
1
of ψ(r) is (−1)lA = −(−1)lB = (−1)j+ 2 Sign(κ) .

13.5 The Dirac Spinors

We can now write the Dirac spinors in the form

  m

ψA (r) RA (r)YjlAj (r̂)
ψ(, j, mj , κ; r) = = m , (13.48)
ψB (r) iRB (r)YjlBj (r̂)

j3
where Yjl (r̂) is a normalized spin-angular function formed by the combination
of the Pauli spinors with the spherical harmonics of order l. The radial functions
RA and RB depend on κ. The factor i is inserted to make RA and RB real for
bound-state solutions.

13.6 The Radial Equations in Polar Coordinates

We rewrite the Hamiltonian in spherical polar coordinates. We begin with the term
Σ·p:


i
Σ·p = (Σ·r̂)(Σ·r̂)Σ·p = (Σ·r̂)(r̂·p + iΣ·(r̂ × p)) = (Σ·r̂) r̂·p + Σ·L ,
r
(13.49)
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch13

112 The Supersymmetric Dirac Equation

so


∂ Σ·L
Σ·p = i(Σ·r̂) − + . (13.50)
∂r r
Now rewrite Σ·L in terms of the Dirac operator K,


∂ βK + 1
Σ·p = −i(Σ·r̂) + . (13.51)
∂r r
This is the relativistic generalization of Eq. (10.66). Multiply this equation from
the left by γ5 and get, by Eq. (13.18):


∂ βK + 1
α·p = −i(α· r̂) + . (13.52)
∂r r
We obtain for the Hamiltonian the expression


∂ βK + 1
H = −i(α· r̂) + + βm + V (r). (13.53)
∂r r
With this Hamiltonian the equation for the bound states becomes





 
0 σ·r̂ ∂ 1 0 −σ·r̂ κ m 0
i − − −i + + V (r)
σ·r̂ 0 ∂r r σ·r̂ 0 r 0 −m



RA YA RA Y A
× =E . (13.54)
iRB YB iRB YB

In terms of the two-component spinors the upper part of Eq. (13.54) is


dRB (1 − κ)RB
+ (σ·r̂)YB = (E − m − V (r)) RA YA . (13.55)
dr r
Now use Eq. (10.46) and get


dRB (κ − 1)RB
− + YA = (E − m − V (r)) RA YA , (13.56)
dr r
or
dRB (κ − 1)
− + RB = (E − m − V (r)) RA . (13.57)
dr r
Similarly, the lower part of Eq. (13.54) is
dRA (κ + 1)
+ RA = (E + m − V (r)) RB . (13.58)
dr r
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch13

The Dirac Equation 113

We introduce the functions

uA (r) = rRA (r), uB (r) = rRB (r), (13.59)

and the dimensionless variable x defined by



2Zαm
x = 2m 1− 2 r= r, (13.60)
na

where
Zα
na = √ . (13.61)
1 − 2
Since in the non-relativistic limit

1 (Zα)2
→1− + O((Zα)4 ) (13.62)
2 N2
we have

na → N (13.63)

in this limit. For this reason na is called the “apparent principal quantum number”.
In terms of the functions uA (x), uB (x) we find the coupled differential
equations

   
d κ 1 1 −  Zα
− uB (x) = − uA (x),
dx x 2 1+ x

    (13.64)
d κ 1 1 +  Zα
+ uA (x) = + uB (x).
dx x 2 1− x

Equation (13.64) may be written in the following matrix form:



duA duB 1 −κ −Zα
, = (ua , uB )
dx dx x Zα κ


1 0 1−
+ (uA , uB ) √ . (13.65)
2 1 − 2 1+ 0

The solution of these coupled differential equations in the literature is somewhat

ad hoc. The recognition of the supersymmetric structure of the equations allows a
systematic development.
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch13

114 The Supersymmetric Dirac Equation

Notes on Chapter 13
This is the work of Dirac [Dir28], referred to in Chapter 1. The considerations here
are by now completely standard; see the texts by Bjorken and Drell [BD64], Sakurai
[Sak67], or Rose [Ros61]. The Dirac equation is here motivated as a generalization
of the Pauli equation in Chapter 10.
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch14

Chapter 14
The Primary Supersymmetry of the Dirac Equation

An important conserved quantity of the Dirac equation was introduced by Johnson

and Lippmann [JL50], designated here as A. As we shall see at the end of this
chapter, it is the relativistic generalization of Anr = (σ·A), where A is the Laplace
vector. It is only announced in the original paper; the first derivation was given
by Biedenharn [Bie62]. The present derivation is by Khachidze and Khelashvili
[KK06].
We show that, acting on an eigenstate of the Dirac operator K, A has the effect
of interchanging the eigenvalues κ and −κ. The equation

A|, j, mj , κ >= −e|, j, mj , −κ > (14.1)

thus replaces the key equation of our non-relativistic analysis of the spectrum of the
hydrogen atom. This equation expresses the primary supersymmetry of the Dirac
equation. The supersymmetry of the Dirac equation in a Coulomb field may be seen
just by looking at the pattern formed by the lines of the spectrum in Figure 14.1.

14.1 A Derivation of the Johnson–Lippmann Operator

We first generalize the theorem of Section 10.2 to the relativistic context.

Theorem. For any vector v with respect to L, which is perpendicular to L,

v·L = L·v = 0, (14.2)

the Dirac operator K anticommutes with (Σ·v).

Proof. The proof is a straightforward generalization of the previous case. The

corollary is

K(Σ·v) = (i/2)βΣ·(v × L − L × v). (14.3)

115
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch14

116 The Supersymmetric Dirac Equation

E [eV]

κ = -1 κ=1 κ = -2 κ=2

j=1/2 j=3/2

Figure 14.1. The supersymmetry of the relativistic hydrogen spectrum.

We again have as examples v = r̂, v = p, v = A, where, by Eq. (8.8),

1
A= (p × L − L × p) − r̂. (14.4)
2Zαm
There is a relation between these three K-odd operators:
1 −i
Σ·A = Σ·(p × L − L × p) − Σ· r̂ = βK(Σ·p) − Σ· r̂,
2Zαm Zαm
(14.5)

where we have used Eq. (14.3).

The class of operators that anticommute with K, the K-odd operators, is not
restricted to the operators just discussed. Any operator of the form O(Σ·v), where
O is any operator which commutes with K, is also K-odd.
In the non-relativistic case, we found the key equation, Eq. (10.57), by utilizing
a constant of the motion which is K-odd. We try the same strategy here: We look
for a K-odd constant of the motion. Let us try a combination of (Σ·r̂) and K(Σ·p),
where the second is of the form O(Σ·p). This choice is motivated by Eq. (14.5),
which relates these operators to the Laplace vector. Both of the above operators are
represented by diagonal matrices. After commuting them with the non-diagonal
operator H, to check if they are constants of the motion, we will get anti-diagonal
terms. For example,
−2i
[Σ· r̂, H] = βKγ5 . (14.6)
r
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch14

The Primary Supersymmetry of the Dirac Equation 117

Exercise 14.1 Prove Eq. (14.6).

This new operator is anti-diagonal. It is only available in the relativistic case.

For this reason we consider the following combination of diagonal and anti-
diagonal operators:

A
= x1 (Σ· r̂) + ix2 K(Σ·p) + ix3 Kγ5 f (r), (14.7)

where f (r) is an arbitrary scalar function to be determined later. The coefficients

are chosen in such a way that A
is Hermitian with x1 , x2 , x3 arbitrary real numbers.
Indeed,

A
† = x1 (Σ· r̂) − ix2 (Σ·p)K − ix3 γ5 Kf (r) = A
. (14.8)

In order that the K-odd operator A
be a constant of the motion we must have


[A , H] = 0. To calculate this commutator we need the following terms:

Exercise 14.2 Prove [K(Σ·p), H] = −iK(Σ· r̂)V
(r).

Exercise 14.3 Prove [Kγ5 f (r), H] = iK(Σ· r̂)f
(r) + 2mKγ5 βf (r).

Together with Eq. (14.6) these yield

2i
[A
, H] = − x1 βKγ5 + x2 V
(r)K(Σ· r̂)
r
− x3 f
(r)K(Σ· r̂) − 2ix3 mβKγ5 f (r). (14.9)

We group diagonal and anti-diagonal operators separately and set Eq. (14.9) to
zero. This yields

x 
+ mx3 f (r) + K(Σ· r̂)[x2 V
(r) − x3 f
(r)] = 0.
1
−2iβKγ5 (14.10)
r
This is fulfilled if the diagonal and the anti-diagonal terms vanish separately:
x1
= −x3 mf (r), x2 V
(r) = x3 f
(r). (14.11)
r

We integrate the second equation over the interval (r, ∞), with the requirement
that f (r) and V (r) tend to zero when r → ∞, and find

x2 V (r) = x3 f (r). (14.12)

August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch14

118 The Supersymmetric Dirac Equation

From the first equation

x1 1
f (r) = − . (14.13)
x3 mr
So
x1 1
V (r) = − . (14.14)
x2 mr
Hence we have shown, within a very general framework, that the only potential for
which the Dirac Hamiltonian has an additional symmetry is a Coulomb potential.
Since the relative signs of x1 and x2 are arbitrary, we have a symmetry both for
attraction and repulsion. For attraction
Zα
V (r) = − , (14.15)
r
and it follows that
x1
x2 = . (14.16)
Zαm
The operator (14.7) becomes
 

i i
A = x1 Σ· r̂ + K(Σ·p) − Kγ5 . (14.17)
Zαm mr
This K-odd constant of the motion is actually a constant multiple of the operator
discovered by Johnson and Lippman [JL50], conventionally written in the form
−i
A= Kγ5 (H − βm) − Σ· r̂. (14.18)
Zαm
Substituting the expressions for H, Eq. (13.15), and K, Eq. (13.23), into the
expression for A brings the operator into the following form:
i iKγ5
A= β(Σ·L + 1)(Σ·p) − − Σ·r̂. (14.19)
Zαm mr
From Eq. (14.3)
i
−(Σ·L + 1)(Σ·p) = Σ·(p × L − L × p), (14.20)
2
and this enables us to write A as
 
1 iKγ5
A = Σ· β(p × L − L × p) − r̂ − . (14.21)
2Zαm mr
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch14

The Primary Supersymmetry of the Dirac Equation 119

The non-relativistic limit of the Johnson–Lippmann operator is obtained, in the

Dirac representation of the γ–matrices, by taking the “large” components of the
Dirac spinors, that is, restricting oneself to the upper two components. This amounts
to replacing β → 1 and γ5 → 0. This is because the effect of γ5 on the large
components is to add an admixture of the small components to them, which is a
negligible effect in the non-relativistic limit. We then see that

A → Anr = (σ·A). (14.22)

In this sense the Johnson–Lippmann operator A is the relativistic generalization of

(σ·A), with A the Laplace vector.

14.2 Commutation and Anticommutation Relations

of the Johnson–Lippmann Operator
The Johnson–Lippmann operator commutes with the total angular momentum J:
 
i Σ·r
[J, A] = − [J, Kγ5 (H − βm)] − J, = 0. (14.23)
Zαm r
This is because each of the commutators vanish separately; the first because J
commutes with each of the factors, the second because any scalar product and any
function of r is invariant with respect to rotations.
The Johnson–Lippmann operator anticommutes with the Dirac operator K.
K anticommutes with the first term of A: Since K commutes with the factor
(H − βm) it suffices to show that K anticommutes with Kγ5 :

K(Kγ5 ) = −K(γ5 K) = −(Kγ5 )K. (14.24)

Hence K anticommutes with the complete expression for the Johnson–Lippmann

operator A,

AK + KA = 0. (14.25)

The Johnson–Lippmann operator A also anticommutes with the total parity

operator P = βI. Write A in the form (14.18). K commutes with (H − βm). So
the first term in A anticommutes with P :

Kγ5 βI = −β(Σ·L + 1)γ5 βI = β(Σ·L + 1)βγ5 I

= βI(Σ·L + 1)βγ5 = βIβ(Σ·L + 1)γ5
= −βIKγ5 . (14.26)
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch14

120 The Supersymmetric Dirac Equation

Also the second term in A anticommutes with P :

Σ·rβI = βΣ·rI = −βIΣ·r. (14.27)

So the parity operator P anticommutes with the complete expression for the
Johnson–Lippmann operator A,

AP + P A = 0. (14.28)

14.3 Eccentricity
We now derive an expression for the magnitude of the Johnson–Lippmann operator
in terms of the eigenvalues of the Dirac operator. We evaluate the operator A2 as
A† A, and derive the following expression for A2 :
  
i −i
A2 = (H − βm)γ5 K − Σ·r̂ Kγ5 (H − βm) − Σ·r̂
Zαm Zαm
1
= K 2 (H − βm)2 + 1
(Zαm)2
i
− [(Σ·r̂)Kγ5 (H − βm) − (H − βm)γ5 K(Σ·r̂)]. (14.29)
Zαm
 
Exercise 14.4 Prove K 2 (H − βm)2 = K 2 H 2 − m2 + 2mβ Zα
r .

Exercise 14.5 Prove (Σ · r̂)Kγ5 (H −βm)−(H −βm)γ5 K(Σ · r̂) = −2iβK 2 /r.

Putting the two pieces together, and adding 1:

 
K2 Zα i βK 2
A2 = 1 + 2
H 2
− m 2
+ 2mβ − (−2i)
(Zαm) r Zαm r
1
= 1+ K 2 (H 2 − m2 ). (14.30)
(Zαm)2

Replacing the operators in this relation with their eigenvalues gives

κ2
e2 = 1 − (1 − 2 ), (14.31)
(Zα)2

where e is the magnitude of A. It is convenient to rewrite this result in the form

 2
κ
e =1−
2
, (14.32)
na
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch14

The Primary Supersymmetry of the Dirac Equation 121

where na is defined by the Eq. (13.61). We have defined e to be non-negative;

hence we have
 2
κ
e= 1− . (14.33)
na
e will be interpreted as the eccentricity. The similarity to the expression for the non-
relativistic eccentricity, Eq. (10.80), and to the classical eccentricity, Eq. (2.29), is
remarkable.

14.4 The Johnson–Lippmann Operator as the Generator

of Supersymmetry
The Johnson–Lippmann operator is a conserved quantity, which, when operating
on an eigenstate of κ, interchanges κ and −κ. If the eigenvalue κ > 0 characterizes
one sort of particle, and κ < 0 characterizes the other sort, then A is the
supersymmetric charge which communicates between the two sorts. It thus
generates a supersymmetric algebra. This is the primary supersymmetry of the
Dirac equation. This was first noted by Nieto et al. [NKT85]. Applied analogously
to the non-relativistic case it yields solutions for the radial wave functions.
The fact that A commutes with H and J but anticommutes with K implies
first that A|, j, mj , κ is zero or else an eigenstate of H, J2 , and J3 with the same
eigenvalues as |, j, mj , κ, and next that it is an eigenstate of K with an eigenvalue
that is the opposite of that associated with |, j, mj , κ. Thus we may write

A|, j, mj , κ = −e|, j, mj , −κ >, (14.34)

where e is a so-far undetermined constant. By a proper choice of the relative phase

of the vector sets |, j, mj , κ and |, j, mj , −κ >, we may ensure that e is real and
positive (or zero). The fact that A is a Hermitian operator implies then that we also
have

A|, j, mj , −κ >= −e|, j, mj , κ, (14.35)

where e is the same as before.

We see that

A (|, j, mj , κ ± |, j, mj , −κ) = ∓e (|, j, mj , κ ± |, j, mj , −κ) . (14.36)

The state vectors (|, j, mj , κ ± |, j, mj , −κ) are, accordingly, eigenstates of

the alternative set of mutually commuting operators {H, J2 , J3 , A}.
We have two types of states, those with positive values of κ, and those with
negative values of κ, with the operator A mediating between them, Eqs. (14.34)
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch14

122 The Supersymmetric Dirac Equation

and (14.35), so it is natural to identify

   
0 0 0 A
Q= , Q† = (14.37)
A 0 0 0

with the supersymmetry generators, i.e., we have A+ = A− = A, since A is

Hermitian. The supersymmetric Hamiltonian becomes
   2 
H+ 0 A 0
H= = . (14.38)
0 H− 0 A2

We recall that A commutes with J2 and J3 , which implies that the

supersymmetric generators, and the supersymmetric Hamiltonian also commute
with these operators. Thus, the S(2) supersymmetry is separate from the SO(3)
symmetry induced by the central field, and the total symmetry of the Dirac–Kepler
problem is SO(3) × SO(2).
In accordance with this, we have a pair of bosonic and fermionic stacks for each
level of j. In fact we have 2j + 1 identical pair of stacks for each j, corresponding
to the 2j + 1 values of mj . In the following, we shall assume that all levels of a pair
of stacks have been assigned the same fixed values of the SU(2) quantum numbers
j and mj .
We identify the vectors |ν1 and |ν2 with the Dirac spinors, but suppress the
quantum numbers j and mj in accordance with the above remarks. Thus, we have
 
ψ(r; , |κ|)
ψ(r; ) = . (14.39)
ψ(r; , −|κ|)

The parity operator that is responsible for the grading is

Pκ = K/|κ|, (14.40)

where K is the Dirac operator. It is represented by the matrix specified in Eq. (11.8).
The functions ψ(r; , |κ|) and ψ(r; , −|κ|) are transformed into each other
by the operator A, and they are both eigenfunctions of A2 with eigenvalue e2 .
Thus, ψ(r; ) is an eigenvector of the supersymmetric Hamiltonian. The relation
(14.33) shows that there is a one-to-one correspondence between the values of 
and e in a given pair of stacks, i.e., for a fixed value of j and, hence, of |κ|. The
supersymmetric Hamiltonian H is thus closely tied to the Hamiltonian H, as it
should be.
Having completely identified the exact supersymmetry of the Dirac–Kepler
problem, we must determine the ψ(r; ) corresponding to the bottom level, for
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch14

The Primary Supersymmetry of the Dirac Equation 123

which the eigenvalue of H, and, hence, e is zero. We ask, accordingly, for a Dirac
spinor ψ0 (r; , |κ|) such that

Aψ0 (r; , |κ|) = 0. (14.41)

In the notation of Eq. (13.48), we must determine radial functions RA (r) and
RB (r) satisfying the equation
   
RA (r)YA 0
A = . (14.42)
iRB (r)YB 0

From Eq. (13.53)

 
∂ βK 1 Zα
H − βm = −i(α· r̂) + + − . (14.43)
∂r r r r

We also use the relations (10.46) and (10.55) to get

   
RA (r)YA RA (r)YB
(Σ·r̂) =− , (14.44)
iRB (r)YB iRB (r)YA

which holds for any pair of radial functions. We use the matrix form of γ5 ,
Eq. (13.1), and the fact that ψ0 (r; , κ) is an eigenfunction of K with eigenvalue
κ. Then, using the explicit form of A, Eq. (14.18), in Eq. (14.42):
  
−i RA (r))YA
Aψ0 (r; , κ) = Kγ5 (H − βm) − Σ·r̂)
Zαm iRB (r)YB
  
κ ∂ βκ 1 RA (r)YB
=− + +
Zαm ∂r r r iRB (r)YA
     
κ −RB (r)YB RA (r)YB 0
− + = . (14.45)
mr iRA (r)YA iRB (r)YA 0

Hence, the condition (14.41) becomes

 
Zαm d
− r + r + κ + 1 RA (r) = ZαRB (r),
κ dr
 
Zαm d
− r + r − κ + 1 RB (r) = −ZαRA (r). (14.46)
κ dr
Using the variable
 
2Zαm
x= r, (14.47)
|κ|
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch14

124 The Supersymmetric Dirac Equation

which, as we shall see below, is consistent in this case with (13.60) (because
|κ| = na ), we get instead
 
1 d
− Sign(κ)x + x + κ + 1 RA (x) = ZαRB (x),
2 dx
 
1 d
− Sign(κ)x + x − κ + 1 RB (x) = −ZαRA (x). (14.48)
2 dx

Inserting one of these equations into the other shows that both RA (x) and RB (x)
satisfy the equation

 2 
d 1
x − Sign(κ)x − γ 2 |κ|2 R(x) = 0, (14.49)
dx 2

with

(Zα)2
γ2 = 1 − . (14.50)
|κ|2

This equation factorizes as follows:

  
d 1 d 1
x − Sign(κ)x − γ|κ| x − Sign(κ)x + γ|κ| R(x) = 0.
dx 2 dx 2
(14.51)
The order of the two factors is arbitrary. Hence, we end up with the two equations
 
d 1
x − Sign(κ)x − γ|κ| R(x) = 0,
dx 2
  (14.52)
d 1
x − Sign(κ)x + γ|κ| R(x) = 0,
dx 2

with the respective solutions

1
R(x) = (const)xγ|κ|−1 e 2 xSign(κ)
R(x) = (const)x−γ|κ|−1 e 2 xSign(κ) .
1
(14.53)

The second solution is physically unacceptable because of its behavior at the origin.
The first solution behaves properly at the origin, but it is only normalizable for
negative values of κ. Hence, we conclude that Eq. (14.41) only has a proper solution
for negative values of κ. Identifying the solution R(x) with RA (x) and inserting
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch14

The Primary Supersymmetry of the Dirac Equation 125

it into Eq. (14.48) gives

κ + γ|κ| Zα
RB (x) = RA (x) = RA (x). (14.54)
Zα κ − γ|κ|

The ψ0 (r; ) corresponding to the lowest energy of a double stack has now
been found to have the form
 
0
ψ0 (r; ) = , (14.55)
ψ0 (r; , κ)

with

YA
ψ0 (r; , κ) = R0 (x) κ+γ|κ| , (14.56)
i Zα
YB

and

R0 (x) = (const)xγ|κ|−1 e− 2 .
x
(14.57)

κ is negative and the functions YA and YB must be chosen according to the rules for
parallel coupling, described in Section 13.4. Thus, we have identified the ground
state of the supersymmetric Hamiltonian. The corresponding eigenvector satisfies
the equations

Qψ0 (r; ) = 0, Q† ψ0 (r; ) = 0, (14.58)

as it should for the supersymmetry to be exact.

The energy of the ground state level with a fixed value of κ is easily obtained
from the results of Section 14.3 with e = 0. The relation (14.33) shows first that
na = |κ|, and when this is inserted in Eq. (14.31), we get
2

Zα
= 1− = γ, (14.59)
κ

with γ as in Eq. (14.50). We also note that the relation na = |κ| implies that the
definitions (13.60) and (14.47) of the variable x coincide when e = 0.
Such states correspond to the “circular” orbits of the Bohr model. They are non-
degenerate under supersymmetry and hence correspond to the eigenvalue e = 0
for the operator A, which is the relativistic generalization of Anr = (σ·A). Thus
we have established the result that the states corresponding to the circular orbits
are those with vanishing eccentricity.
August 27, 2011 9:21 9in x 6in The Supersymmetric Dirac Equation b1217-ch14

126 The Supersymmetric Dirac Equation

Having determined the ground-level wave function for a double stack

corresponding to a given level of |κ|, we should focus on the energies and wave
functions of the higher levels. This could be done by solving the equation
     
ψ(r; , |κ|) ψ(r; , |κ|) ψ(r; , |κ|)
H =A 2
=e2
, (14.60)
ψ(r; , −|κ|) ψ(r; , −|κ|) ψ(r; , −|κ|)
i.e., by solving the equation

A2 ψ(r; , κ) = e2 ψ(r; , κ) (14.61)

for +κ and −κ separately. In fact, it suffices to solve the equations for one sign.
The solution corresponding to the other sign may then be obtained by Eq. (14.34).
Instead of solving Eq. (14.61) for a given value of κ, we may, of course,
focus on the radial equations for the same κ-value, and this is what we shall
do in the following chapter. In either case it is advantageous to exploit the way
supersymmetry reflects itself in an extended function space.

Notes on Chapter 14
The Johnson–Lippmann operator was announced as a conserved quantity of the
Dirac equation by Johnson and Lippmann [JL50]. The realization that it reduces to
the Laplace vector in the non-relativistic limit is due to Biedenharn [Bie83]. That
the Johnson–Lippmann operator generates a supersymmetry group was realized
by Nieto, Kostelecky and Traux [NKT85]. The parallelism with the non-relativistic
case was pointed out by Dahl and Jørgensen [DJ95].
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch15

Chapter 15
Extending the Solution Space

We may obtain a representation of the Dirac wave functions by the transformation

ψ → ψ̄ = γ5 ψ. It turns out that the supersymmetry of the Dirac equation manifests
itself in the new representation as a relation of the radial wave functions. We can
solve this relation for the radial wave functions, and then transform back to the
original representation to find the Dirac wave functions. A version of this procedure
has been known for some time [Bie62] — it was found by manipulation of the
solutions of Kramer’s equation (see Chapter 17 below). Now that we recognize the
role played by supersymmetry we can obtain the Dirac wave functions directly,
without going through the detour of Kramer’s equation.
In the same way that the Dirac equation leads to the Hamiltonian H of Eq.
(13.13), the unphysical equation Eq. (13.5) leads to the Hamiltonian
Zα
H̄ = α·p − βm − , (15.1)
r
which results from the ordinary Dirac Hamiltonian (13.15) by changing the sign
in front of m. It is referred to as the negative-mass Hamiltonian.
It is readily seen that

H̄ = γ5 Hγ5 . (15.2)

If, therefore, ψ is a solution of the positive-mass Eq. (13.14), then

ψ̄ = γ5 ψ (15.3)

is a solution of the negative-mass equation

H̄ ψ̄ = E ψ̄, (15.4)

with the same energy E.

γ5 commutes with J, but it anticommutes with K. This implies that if ψ is
characterized by the quantum numbers j, m and κ then ψ̄ has the quantum numbers

127
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch15

128 The Supersymmetric Dirac Equation

j, m and −κ. Thus, the sign of κ is reversed. The ψ̄-functions differing merely in
the sign of κ are transformed into each other by the operator Ā = γ5 Aγ5 in a
similar way as in Eq. (14.34).
It follows that the stationary bound states of the negative-mass Hamiltonian
define a supersymmetric structure that is, in all details, the same as the
supersymmetric structure defined by the positive-mass Hamiltonian, except that
the roles of κ and −κ are reversed.
If we now group the positive-mass states with a given κ-value together with the
negative-mass states with the same κ-value, then we obviously encounter a double
stack of exactly the same appearance as before. The question is then: Is it possible
to introduce a parity operator and generators such that this double stack can be
properly described by supersymmetry? It turns out that this is indeed the case.
When we group the positive- and negative-mass solutions together then the
states with the same κ-value form a new supersymmetric structure. The appearance
of this structure is a direct consequence of the supersymmetric structure that
combines states with opposite values of κ in the original Dirac–Kepler problem.
The latter structure was based on the presence of the Johnson–Lippmann operator.
We expect, accordingly, that this operator also plays a central role for the description
of the new structure.
There is a rather direct way of carrying a positive-mass state with a given
κ-value into a negative-mass state with the same κ-value, namely to apply the
operators A and γ5 in succession. Diagonalizing the operator γ5 A will therefore
separate the double stack into two columns with different eigenvalues of γ5 A. These
columns are not affected if the operator is multiplied by an arbitrary constant and
a diagonal operator added so as to obtain an operator whose eigenvalues merely
differ in sign. Such an operator defines a grading of the function space and, hence,
a center for the description of the supersymmetry.
Using the explicit form of the operator A, Eq. (14.18), we find its action on
Dirac spinors:
  

 RA (r) −
κ
(1 − )RB (r) YB
RA (r)YA  Zα 
A = 
κ  . (15.5)
iRB (r)YB i RB (r) − (1 + )RA (r) YA
Zα

Applying the matrix form of γ5 , Eq. (13.7), yields



RA (r)YA iRB (r)YB
γ5 = . (15.6)
iRB (r)YB RA (r)YA
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch15

Extending the Solution Space 129

Hence
   

 i −RB (r) +
κ
(1 − )RA (r) YA
RA (r)YA  Zα 
−iZαγ5 A = iZα    .
iRB (r)YB κ
−RA (r) + (1 + )RB (r) YB
Zα
(15.7)

Since the spinors before and after the transformation have the same value of κ,
they also have the same spin and angular variable dependence. The transformation
can then be expressed as a transformation of the radial functions, i.e., as a linear
transformation in the space of functions VR spanned by RA (r) and RB (r). That is
    
RA κ( − 1) Zα RA
−iZαγ5 A = . (15.8)
RB −Zα κ( + 1) RB
The eigenvalues of this matrix are, by the Cayley characteristic equation, the roots of
 
κ( − 1) − λ Zα
det = 0, (15.9)
−Zα κ( + 1) − λ
i.e., λ = κ ± γ|κ|. The grading may therefore be defined by
KH
Γ=− − iZαγ5 A, (15.10)
m
which has the eigenvalues ±γ|κ|.

15.1 The Γ-Induced Radial Supersymmetry

The operator Γ = −KH/m−iZαγ5 A induces the following linear transformation
in the space VR :
 
−κ −Zα
Γ(uA , uB ) = (uA , uB ) . (15.11)
Zα κ
We recognize this as the matrix in the first term of the Dirac equation for the radial
functions RA and RB , Eq. (13.65). This matrix is diagonalized by


(Zα)c1 (Zα)c2
D= , (15.12)
−(γ|κ| + κ)c1 (γ|κ| − κ)c2
with c1 , c2 arbitrary constants. Then


−1 1 (γ|κ| − κ)c2 −(Zα)c2
D = , (15.13)
det (γ|κ| − κ)c1 (Zα)c1
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch15

130 The Supersymmetric Dirac Equation

with

det = 2c1 c2 (Zα)γ|κ|, (15.14)

and


−1
γ|κ| 0
D ΓD = . (15.15)
0 −γ|κ|

The eigenvalues of Γ are thus

γ1 = γ|κ|, γ2 = −γ|κ|, (15.16)

where γ is as in Eq. (14.50). The grading operator is Pκ = Γ/γ|κ|. The new basis
vectors are


(Zα)c1 (Zα)c2
(w1 , w2 ) = D(uA , uB ) = (uA , uB ) .
−(γ|κ| + κ)c1 (γ|κ| − κ)c2
(15.17)
The vector (w1 , 0) is an eigenvector of Γ, referred to the new basis, corresponding
to the eigenvalue γ|κ|, the vector (0, w2 ) is an eigenvector of Γ corresponding to
the eigenvalue −γ|κ|. The inverse of Eq. (15.17) is


−1 1 (γ|κ| − κ)c2 −(Zα)c2
(uA , uB ) = D (w1 , w2 ) = (w1 , w2 ) .
det (γ|κ| + κ)c1 (Zα)c2
(15.18)

We shall now perform a similarity transformation to the basis (15.17) in the

Eqs. (13.65) for the radial Dirac functions duA /dx, duB /dx. We have still to
calculate the matrix in the second term of (13.65):
 
1 −1 0 1−
√ D D
2 1 − 2 1− 0
   
−(Zα) (κ−γ|κ|)2 (Zα)
√ c2 1−
− 1+
 2γ|κ| 1−2 c1 4γ|κ|(Zα) 1+ 4γ|κ| 1−

= 
    
(κ+γ|κ|)2 (Zα) (Zα)
− cc12 4γ|κ|(Zα)
1−
1+ − 4γ|κ|
1+
1−
√
2γ|κ| 1−2
(15.19)
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch15

Extending the Solution Space 131

The result of the similarity transformation on the differential equations for

duA /dx, duB /dx is the sum of the results for the first and second terms:


dw1 dw2
,
dx dx
    
−Zα (κ−γ|κ|)2
√ +
γ|κ| c2 1−
−
(Zα) 1+
 2γ|κ| 1−2 x c1 4γ|κ|(Zα) 1+ 4γ|κ| 1− 
 
= (w1 , w2 ) 
     
(κ+γ|κ|)2 (Zα) √
(Zα)
−
γ|κ|
− cc1 4γ|κ|(Zα)
1−
1+
− 4γ|κ|
1+
1− x
2 2γ|κ| 1−2

(15.20)
From this we see that if we put

c1 (κ + γ|κ|)(κ − γ|κ|)
= , (15.21)
c2 (κ − γ|κ|)(κ + γ|κ|)

then we obtain the more symmetrical form:

 
na γ|κ| ena
   − + − 
dw1 dw2  2γ|κ| x 2γ|κ| 
, = (w1 , w2 )  . (15.22)
dx dx  ena na γ|κ| 
−
2γ|κ| 2γ|κ| x
Here we have used
   
c1 1 (κ + γ|κ|)2 (1 − ) − (Zα)2 (1 + )
√
c2 4(Zα)γ|κ| 1 − 2
  
c1 1 (κ + γ|κ|)(κ − γ|κ|) −1 (κ)2 − γ 2 |κ|2
=− √ = ,
c2 2(Zα)γ|κ| 1 − 2 2γ|κ| 1 − 2
(15.23)

and
 
c2 1 (κ − γ|κ|)2 (1 − ) − (Zα)2 (1 + ))
√
c1 4(Zα)γ|κ| 1 − 2
  
c2 1 (κ − γ|κ|)(κ + γ|κ|) −1 (κ)2 − γ 2 |κ|2
=− √ = ,
c1 2(Zα)γ|κ| 1 − 2 2γ|κ| 1 − 2
(15.24)

as well as

(κ)2 − γ 2 |κ|2
= ena , (15.25)
1 − 2
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch15

132 The Supersymmetric Dirac Equation

which is a consequence of Eqs. (14.33) and (13.61). The coupled differential

equations are written out as
 
dw1 na γ|κ| ena
− − − w1 = − w2 ,
dx 2γ|κ| x 2γ|κ|
  (15.26)
dw2 na γ|κ| ena
− − w2 = − w1 .
dx 2γ|κ| x 2γ|κ|
Before proceeding to solve the differential equations we can conclude some
general results concerning the normalizations of the solutions, as well as certain
expectation values. We shall make use of these results in Section 16.1.
We multiply the first of the equations in (15.26) by w2 , the second by w1 , and
integrate over x. This yields
    
dw1 na γ|κ| ena
− w2 dx − w2 − w1 dx = − w22 dx,
dx 2γ|κ| x 2γ|κ|
    
dw2 na γ|κ| ena
w1 dx − w1 − w2 = − w12 dx. (15.27)
dx 2γ|κ| x 2γ|κ|
Now
 
dw1 dw2
− w2 dx = w1 dx (15.28)
dx dx
by integration-by-parts, which is allowed for functions that fall off suitably at
infinity. The other integrals on the left-hand side of the equations are also equal,
so those on the right-hand side must be equal as well. Thus
 
w1 dx = w22 dx,
2
(15.29)

or

w1 |w1  = w2 |w2 . (15.30)

We see that w1 and w2 are normalized in the same way.

We find a second relation for the functions w1 and w2 by multiplying the first
equation by w1 , the second by w2 , and integrating. We get
   
dw1 na γ|κ| ena
− w1 dx − w1 − w1 dx = − w1 w2 dx,
dx 2γ|κ| x 2γ|κ|
   
dw2 na γ|κ| ena
− w2 dx − w2 − w2 dx = − w2 w1 dx. (15.31)
dx 2γ|κ| x 2γ|κ|
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch15

Extending the Solution Space 133

The integral of a function times its derivative vanishes for acceptable functions.
Since the remaining integrals are equal, we find
 1    1  
   
w1  w1 = w2  w2 . (15.32)
x x

We now proceed to solve the coupled differential equations of (15.26)

explicitly. To display the supersymmetric structure of these equations, we define
the operators
   
− d γ|κ| na d γ|κ| na
b =− + − , +
b = + − . (15.33)
dx x 2γ|κ| dx x 2γ|κ|

It is clear that these operators are the Hermitian conjugates of each other. The
equations may then be written

ena ena
b− w 1 = − w2 , b+ w2 = − w1 . (15.34)
2γ|κ| 2γ|κ|

The operator b+ is the analog of A+ : it transforms an even state into an odd state.
The operator b− is the analog of A− , it transforms an odd state into an even state.
The functions w1 and w2 take the place of ν1 and ν2 , respectively. The equations
of (15.34) imply that
 2  2
ena ena
b+ b− w1 = w1 , b− b+ w2 = w2 . (15.35)
2γ|κ| 2γ|κ|

Hence the analog of Eq. (11.12) becomes

 + −    2  
b b 0 w1 ena w1
= . (15.36)
0 b− b+ w2 2γ|κ| w2

The eigenvalue of the supersymmetric Hamiltonian is zero when e = 0. Otherwise,

it is positive.

15.2 The Supersymmetric Ground State

in the Γ Representation
We are interested in the solution of the equations in (15.35) when e equals zero.
With e = 0, Eq. (14.33) tells us that na = |κ|. The corresponding reduced energy
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch15

134 The Supersymmetric Dirac Equation

 becomes γ, as shown in Eq. (14.59). Hence na = γ|κ|. Inserting this result in
(15.26) yields
   
dw1 1 γ|κ| dw2 1 γ|κ|
− − − w1 = 0, − − w2 = 0. (15.37)
dx 2 x dx 2 x
The solutions of these equations are

w1 (x) = (const)xγ|κ| e−x/2 , w2 (x) = (const)x−γ|κ| ex/2 . (15.38)

Of these, w2 (x) is not a physically acceptable solution, because of its behavior at

x = ∞. Only w1 (x) is an acceptable solution, the supersymmetric ground state
being
   γ|κ| −x/2 
w1 x e
= (const) . (15.39)
w2 0
w1 (x) corresponds to the eigenvalue γ|κ| for the operator Γ. It is equal to x times
the function R0 (x) of Eq. (14.57). We determine the structure of the Dirac spinors
by means of Eq. (15.18), we read off
Zα κ + γ|κ|
uB = uA = uA , (15.40)
κ − γ|κ| Zα
in agreement with Eq. (14.56).

15.3 The General Solutions in the Γ Representation

To determine w1 and w2 for the higher levels, we first find that
 2
d2 γ|κ|(γ|κ| − 1) na na
b+ b− = − 2 + − + ,
dx x2 x 2γ|κ|
 2 (15.41)
d2 γ|κ|(γ|κ| + 1) na na
b− b + = − 2 + − + .
dx x2 x 2γ|κ|
We then insert these expressions into (15.35) and use relations (14.33) and (14.49)
to get

(na )2 − (ena )2 = −γ 2 |κ|2 . (15.42)

This gives
 
d2 w1 γ|κ|(γ|κ| − 1) na 1
+ − + − w1 = 0,
dx2 x2 x 4
  (15.43)
d2 w2 γ|κ|(γ|κ| + 1) na 1
+ − + − w2 = 0.
dx2 x2 x 4
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch15

Extending the Solution Space 135

Next, we introduce new radial functions, G1 (x) and G2 (x), by writing

w1 (x) = xγ|κ| e−x/2 G1 (x), w2 (x) = xγ|κ|+1 e−x/2 G2 (x). (15.44)

These functions satisfy the equations

2
x ddxG21 + (2γ|κ| − x) dG
dx + (na − γ|κ|)G1 = 0,
1

(15.45)
2
x ddxG22 + (2γ|κ| + 2 − x) dG
dx + (na − γ|κ| − 1)G2 = 0.
2

The regular solutions of these equations are confluent hypergeometric functions.

Thus, our final result is
w1 (x) = N1 xγ|κ| e−x/2 1 F1 (−nr , 2γ|κ|; x),
(15.46)
w2 (x) = N2 xγ|κ|+1 e−x/2 1 F1 (−nr + 1, 2γ|κ| + 2; x),
with nr the radial quantum number

nr = na − γ|κ|. (15.47)

For w1 (x) to be a normalizable function, nr must be a positive number or zero.

For w2 (x), we get the same condition, except that nr = 0 is excluded. The number
nr counts the number of nodes of the radial function [Ros61].
We have
Zα
nr + γ|κ| = na = √ , (15.48)
1 − 2
so that
2 (nr + γ|κ|)2
= (15.49)
1 − 2 (Zα)2
and the formula for the energy levels is
 − 12
(Zα)2
= 1+ . (15.50)
(nr + γ|κ|)2
In terms of the eigenvalues of the Hamiltonian
 − 12
(Zα)2
E =m 1+ , (15.51)
(nr + γ|κ|)2
which is the same as the Sommerfeld formula (12.27), with the identification of
|κ| with nϕ . This identification follows from a comparison of (12.8) and (12.13)
with (14.50).
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch15

136 The Supersymmetric Dirac Equation

For nr = 0, we have reproduced the result (15.39). For the other solutions,
we must determine the relative values of N1 and N2 . This is done in Appendix D.
The result is
N1 2γ|κ|(2γ|κ| + 1)
=− . (15.52)
N2 ena
Our final task is to express the original radial functions uA (x) and uB (x) in
terms of w1 (x) and w2 (x). Let us put


1 1
(c1 , c2 ) =  , , (15.53)
(κ + γ|κ|)(κ − γ|κ|) (κ − γ|κ|)(κ + γ|κ|)
in accordance with Eq. (15.21). We get then, by Eq. (15.18),
 
uA (x) = Sign(κ)
2γ|κ| − (κ − γ|κ|)(κ − γ|κ|)w1 (x)
 
+ (κ + γ|κ|)(κ + γ|κ|)w2 (x) ,
  (15.54)
uB (x) = 2γ|κ|1
− (κ + γ|κ|)(κ − γ|κ|)w1 (x)
 
+ (κ − γ|κ|)(κ + γ|κ|)w2 (x) ,

where Sign(κ) is an overall sign factor commented on in Chapter 18.

Thus, we arrive at the desired expressions:

uA (x) = N Sign(κ)xγ|κ| e−x/2 (κ − γ|κ|)(κ − γ|κ|)1

× F1 (−nr , 2γ|κ|; x) + ena x

2γ|κ|(2γ|κ|+1)

 
× (κ + γ|κ|)(κ + γ|κ|)1 F1 (−nr + 1, 2γ|κ| + 2; x) ,
 (15.55)
uB (x) = N xγ|κ| e−x/2 (κ + γ|κ|)(κ − γ|κ|)1 F1 (−nr , 2γ|κ|; x)

+ ena x
2γ|κ|(2γ|κ|+1) (κ − γ|κ|)(κ + γ|κ|)1

× F1 (−nr + 1, 2γ|κ| + 2; x) .
The overall normalization constant corresponding to the normalization
condition

[u2A (x) + u2B (x)]dx = 1, (15.56)

is calculated in Appendix D, Eq. (D.16).

August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch15

Extending the Solution Space 137

Notes on Chapter 15
This type of extension has been widely discussed in the literature. Sukumar
[Suk85a; Suk85b] found the supersymmetric structure of the equations in this
decomposition, and utilized it to solve them. It was also used by Jarvis and
Stedman [JS86], who solved the second-order Kramer’s equation, rather than
the first-order Dirac equation (see Chapter 17). The same equation was solved
by Swainson and Drake [SD91], and by Goodman and Ignjatovi¢ [GI97], in an
article that prompted an avalanche of papers tracing earlier work on this subject
([Sta98], [Wal79], [Kol66], [Won86], [WY82], [AB78], [Bie83]). This method
leads to a different relative normalization of the solutions than that used here. The
uncoupled differential equations had been previously obtained by Infeld [Inf41].
The uncoupled differential equations of (15.43) may be said to arise from the
coupled equations of (15.26) by factorization, according to the general method of
Hull and Infeld [HI51]. The advantage of the method used here, which utilizes
the primary supersymmetry discussed in Chapter 14, is that it leads directly to the
solutions of the Dirac equation, and makes clearer the nature of the coefficients
appearing.
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch15

This page intentionally left blank

August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch16

Chapter 16
A Different Extension of the Solution Space

Here we investigate another representation of the solutions, more closely related

to the standard method of solution, which was actually derived in a somewhat
ad hoc manner [ Gor28]. In this representation the solutions are decomposed
into normal modes, and can accordingly be decoupled. The presentation here,
which again follows that of Dahl and Jørgensen [DJ95], profits from utilizing
the supersymmetry, which is recognized already in its primary form, from the
beginning.
Having extended our analysis to include the stationary bound states of the
Hamiltonian H̄, we shall now make a further, less obvious, extension. This is to
consider also the stationary bound states of the Hamiltonian
2
H̃ = (Σ·r̂)H(Σ·r̂) = H + βγ5 K(Σ·r̂). (16.1)
r
These bound states are obviously described by the functions

ψ̃ = (Σ·r̂)ψ, (16.2)

where ψ describes a stationary bound state of the original Hamiltonian. Like βγ5 ,
the operator Σ·r̂ commutes with J and anticommutes with K. The function ψ̃ bears,
therefore, a similar relation to ψ as does the function ψ̄. Hence, the stationary bound
states of the Hamiltonian H̃ also define a supersymmetric structure that is, in all
details, the same as the supersymmetric structure defined by the positive-mass
Hamiltonian, except that the roles of κ and −κ are reversed.
In a similar way to what we did for the negative-mass states, we may now
group the positive-mass states with a given κ-value together with the new ψ̃ states
with the same κ-value and obtain a double stack of exactly the same appearance
as previously. Again, we may ask if it is possible to introduce a parity operator and
generators such that this double stack can be properly described by supersymmetry.
We now explain how this comes about.

139
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch16

140 The Supersymmetric Dirac Equation

The function ψ̃, which corresponds to ψ but has opposite κ-value, is ψ̃ =

(Σ·r̂)ψ. The function ψ̃ with the same κ-value as ψ is ψ̃ = Σ·r̂Aψ, since the
functions ψ̃ that differ from each other in the sign of κ are related by the operator
à = (Σ·r̂)A(Σ·r̂). We recall Eq. (15.5):
 κ  

 RA (r) − (1 + )RB (r) YB
RA (r)YA  Zα 
A =   , (16.3)
iRB (r)YB κ
i RB (r) − (1 − )RA (r) YA
Zα
and Eq. (14.44):



RA (r)YA RA (r)YB
Σ·r̂ =− . (16.4)
iRB (r)YB iRB (r)YA

We see that both these operators interchange YA and YB . By applying the

two operators in succession we preserve the dependence on the angular variables
and spin. Since both Σ·r̂ and A commute with J but anticommute with K, when
we apply (Σ·r̂)A the eigenvalues of H and J are preserved, and the eigenvalue
of K is reversed in sign. Applying (Σ·r̂)A to ψ, we only encounter linear
combinations of the original radial functions RA (r) and RB (r), i.e., functions
belonging to the function space VR spanned by RA (r), RB (r). Thus the functions
ψ̄(r; , κ), ψ̄(r; , −κ) and ψ̃(r; , κ), ψ̃(r; , −κ) belong, together with ψ(r; , κ)
and ψ(r; , −κ), to the space VR . The supersymmetry which is expressed by these
operators may again be described as a symmetry in the space VR .
Explicitly,
 κ  

 −RA (r) + (1 + )RB (r) YA
RA (r)YA  Zα 
(Σ·r̂)A =    . (16.5)
iRB (r)YB κ
i −RB (r) + (1 − )RA (r) YB
Zα
The representation of (Σ·r̂)A in the space of functions VR is
 κ 
−1 (1 + )
(Σ·r̂)A =  Zα . (16.6)
κ
Zα
(1 − ) −1

The characteristic equation of this matrix is

 κ 
−1 − λ (1 + )
det  Zα  = 0. (16.7)
κ
Zα
(1 − ) −1 − λ
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch16

A Different Extension of the Solution Space 141

Hence its eigenvalues are

κ
λ = −1 ± . (16.8)
na
Thus the operator

∆ = 1 + (Σ·r̂)A (16.9)

has eigenvalues ±κ/na , and defines a parity operator

na ∆
Pκ/na = , (16.10)
κ
which determines a grading of the supersymmetry.

16.1 The ∆-Induced Radial Supersymmetry

The operator ∆ = 1 + (Σ·r̂)A induces the following linear transformation in the
space VR :


κ 0 1−
∆(uA , uB ) = (uA , uB ) . (16.11)
Zα 1+ 0
We recognize the second matrix in the Dirac radial equations, Eq. (13.65).
This matrix is diagonalized by
 d
1 d2 
√ −√
 1+ 1 + 
D= , (16.12)
 d d2 
1
√ √
1− 1−
with d1 , d2 arbitrary constants. Then
 d2 d2 
√ √
1  1− 1 + 
,
D−1 = (16.13)
det  d1 d1

−√ √
1− 1+
with
2d1 d2
det = √ , (16.14)
1 − 2
and


κ/na 0
D−1 ∆D = . (16.15)
0 −κ/na
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch16

142 The Supersymmetric Dirac Equation

So the eigenvalues of ∆ are

κ κ
δ1 = , δ2 = − . (16.16)
na na
The new basis vectors are

d1 d2 
√ −√
 1+ 1 + 
(u1 , u2 ) = D(uA , uB ) = (uA , uB ) 
 d
.
 (16.17)
1 d
2
√ √
1− 1−
The vector (u1 , 0) is an eigenvector of ∆, referred to the new basis, corresponding
to the eigenvalue κ/na . The vector (0, u2 ) is an eigenvector of ∆ corresponding
to the eigenvalue −κ/na . The inverse to Eq. (16.17) is
 d d 
√ 2 √ 2
1  1− 1 + 
.
(uA , uB ) = D−1 (u1 , u2 ) =  (16.18)
det d1 d1 
−√ √
1− 1+
The functions u1 and u2 are also the “normal modes” that decouple the differential
equations that in turn ensure that uA ; uB are solutions of the radial Dirac equations.
Indeed, these differential equations become, after a similarity transformation is
performed,
 1 n d2 (na + κ) 
a

 − −
du1 du2  2 x d1 x 
, = (u1 , u2 ) 
 d (n + κ)
.
 (16.19)
dx dx 1 a 1 a
−
d2 x 2 x
Choosing

d1 na − κ
= (16.20)
d2 na + κ
leads to the more symmetrical equations
 
1 na na

 − −
du1 du2  2 x 
x
, 
= (u1 , u2 )  , (16.21)
dx dx na 1 na 
− +
x 2 x
where we have used

e2 n2a = n2a − κ2 . (16.22)

August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch16

A Different Extension of the Solution Space 143

These equations can be written out as


du1 1 na ena
− + − u1 = − u2 ,
dx 2 x x

 (16.23)
du2 1 na ena
+ − u2 = − u1 .
dx 2 x x
Let us multiply the first of these equations from the left by u2 , the second by u1 ,
and integrate over x. The left-hand sides of the equations will then be the same
for acceptable radial functions, since −(d/dx) is the Hermitian adjoint of (d/dx).
The right-hand sides must therefore also be the same. Hence, we get
   
1 1
u1 u1 = u2 u2 . (16.24)
x x
Next, let us multiply the first of the equations of (16.23) from the left by u1 ,
the second by u2 , and integrate over x. We then compare the resulting equations,
noting that the integral of a function times its derivative vanishes for an acceptable
function. Thus, we get

u1 |u1  = u2 |u2 . (16.25)

Now replace the equations of (16.23) with the equations obtained from them
by multiplying them from the left by x. This gives
du1  x 
−x + − na u1 = −ena u2 ,
dx 2
  (16.26)
du2 x
x + − na u2 = −ena u1 .
dx 2
Let us multiply the first of these equations from the left by u2 , the second by u1 ,
and integrate over x. A comparison of the resulting equations gives
   
d d
u2 x u1 = u1 −x u2 . (16.27)
dx dx

The Hermitian adjoint of x(d/dx) is −(d/dx)x = −x(d/dx) − 1. Hence, we

conclude that u1 and u2 are orthogonal:

u1 |u2  = 0. (16.28)

This relation justifies the designation normal modes for the functions u1 (x) and
u2 (x).
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch16

144 The Supersymmetric Dirac Equation

The equations in (16.26) display the supersymmetric structure of our problem.

For if we introduce the operators a− and a+ by the definitions
d x 
a− = −x + − na ,
dx 2
  (16.29)
d x
+
a =x + − na ,
dx 2
then Eq. (16.27) allows us to consider these operators to be the Hermitian adjoints
of each other. Hence, they may take the place of the operators A− and A+ in Section
11.1, while u1 and u2 take the place of ν1 and ν2 , respectively. A supersymmetric
Hamiltonian may thus be defined as in Eq. (11.5). The equations of (16.23) may
thus be written as
a− u1 = −ena u2 ,
(16.30)
a+ u2 = −ena u1 .
They show that

a+ a− u1 = (ena )2 u1 ,
(16.31)
a− a+ u2 = (ena )2 u2 .
Hence, the analog of Eq. (16.30) becomes
    
a+ a− 0 u1 2 u1
= (ena ) . (16.32)
0 a− a+ u2 u2

Thus, the quantity (ena )2 replaces the quantity E as the eigenvalue of the
supersymmetric Hamiltonian. It is positive or zero, as it should be, and it is a
well-defined function of the physical energy. It becomes zero when e = 0.

16.2 The Supersymmetric Ground State in the ∆ Representation

The supersymmetric ground state is the solution of the equations in (16.30) when
e equals zero. With e = 0, Eq. (14.33) tells us that na = |κ|. The corresponding
reduced energy  becomes γ, as shown in Eq. (14.59). Hence, na = γ|κ|. Inserting
this result in the expressions for a− and a+ gives the equations


du1 1
−x + x − γ|κ| u1 = 0,
dx 2

 (16.33)
d2 1
x + x − γ|κ| u2 = 0.
dx 2
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch16

A Different Extension of the Solution Space 145

The solutions of these equations are

u1 (x) = (const)x−γ|κ| ex/2 ,

(16.34)
u2 (x) = (const)xγ|κ| e−x/2 .

Of these, u1 (x) is not a physically acceptable solution. The supersymmetric ground

state is accordingly
   
u1 0
= (const) γ|κ| −x/2 . (16.35)
u2 x e

The function u2 (x) equals x times the R0 (x) of Eq. (14.57), as we would expect.
From Eq. (16.18) we find
 
uB 1− 1−γ
=− =−
uA 1+ 1+γ

1 − γ2 Zα
=− =−
1+γ |κ|(1 + γ)
−|κ| + γ|κ| κ + γ|κ|
= = , (16.36)
Zα Zα
since κ is negative, in accordance with Eq. (14.56).

16.3 The General Solutions in the ∆ Representation

To determine u1 (x) and u2 (x) for the higher levels, we must solve the uncoupled
differential equations in (16.31). It is easily seen that


d2 d 1 2 1
a+ a− = −x2 − x + x − n a − x + (na )2 ,
dx2 dx 4 2
(16.37)
− + d2 d 1  
a a = −x 2
2
−x + x2 − na + 12 x + (na )2 .
dx dx 4
Hence, the differential equations in (16.30), with the help of Eq. (15.42), become

 
d2 u1 du1 1 2 1
x2 + x + − x + na − x − γ 2 |κ|2 u1 = 0,
dx2 dx 4 2
2

  (16.38)
2 d u2 du2 1 2 1
x +x + − x + na + x − γ |κ|
2 2
u2 = 0.
dx2 dx 4 2
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch16

146 The Supersymmetric Dirac Equation

The substitutions
u1 (x) = xγ|κ| e−x/2 F1 (x),
(16.39)
u2 (x) = xγ|κ| e−x/2 F2 (x),
introduce new radial functions, proportional to F1 (x) and F2 (x), that must satisfy
the equations
d2 F1 dF1
x + (2γ|κ| + 1 − x) + (na − γ|κ| − 1)F1 = 0,
dx2 dx
(16.40)
d2 F2 dF2
x 2 + (2γ|κ| + 1 − x) + (na − γ|κ|)F2 = 0.
dx dx
The regular solutions of these equations are confluent hypergeometric functions;
hence, we get
u1 (x) = N1 xγ|κ| e−x/2 1 F1 (−nr + 1, 2γ|κ| + 1; x),
(16.41)
u2 (x) = N2 xγ|κ| e−x/2 1 F1 (−nr , 2γ|κ| + 1; x),
where nr is the radial quantum number:

nr = na − γ|κ|. (16.42)

For u2 (x) to be a normalizable function, nr must be a positive integer or zero. For

u1 (x), we get the same condition, except that nr = 0 is excluded. Thus, we obtain
the energy levels described in Chapter 15.
For nr = 0, we have now reproduced the result (16.35). For the other solutions,
we must determine the relative values of N1 and N2 . We already know, from Eq.
(16.25), that u1 (x) and u2 (x) must be normalized in the same way, but we do not
know the relative signs of the functions. By explicit calculation (see Appendix E)
we get
N1 nr
= . (16.43)
N2 (na + κ)(na − κ)
We have now determined the complete solutions of the supersymmetric
eigenvalue problem (16.32). We now use the fact that u1 (x) and u2 (x) are
connected to the functions ua (x) and uB (x) through the relations (16.17) and
(16.20). We can therefore obtain the analytical expression for uA (x) and uB (x)
from Eq. (16.17). We put
  
na + γ|κ|
(d1 , d2 ) = 1, , (16.44)
na − γ|κ|
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch16

A Different Extension of the Solution Space 147

in accordance with Eq. (16.20), and get

  
1√ na − γ|κ|
uA (x) = 1 +  u1 (x) − u2 (x) ,
2 na + γ|κ|
   (16.45)
1√ na − γ|κ|
uB (x) = 1 −  u1 (x) + u2 (x) .
2 na + γ|κ|

The final expressions become

N
√
uA (x) = 1 + xγ|κ| e−x/2 [nr 1 F1 (−nr + 1, 2γ|κ| + 1; x)
2
− (na − γ|κ|)1 F1 (−nr , 2γ|κ| + 1; x)],
(16.46)
N
√
uB (x) = 1 − xγ|κ| e−x/2 [nr 1 F1 (−nr + 1, 2γ|κ| + 1; x)
2
+ (na − γ|κ|)1 F1 (−nr , 2γ|κ| + 1; x)].

The overall normalization constant is

 1/2

1 Γ(nr + 2γ|κ| + 1)
N = , (16.47)
Γ(2γ|κ| + 1) na (na − κ)nr !

see Eq. (E.12).

Notes on Chapter 16
The standard method of solving the radial equations (13.64) is by introducing a
normal-mode representation. This method was introduced by Gordon [Gor28].
It is also used by Bethe and Salpeter [BS57]. We now understand that in these
works one is actually diagonalizing the operator ∆, without paying attention to
the underlying supersymmetry. With proper attention paid to this symmetry the
procedure is simplified and the algebra becomes more transparent. The decoupled
equations (16.38) may be said to arise from the coupled equations (16.23) by
factorization.
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch16

This page intentionally left blank

August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch17

Chapter 17
The Relation of the Solutions to Kramer’s Equation

The solutions of the Dirac equation have been widely investigated in the framework
of a second-order differential equation called Kramer’s equation. The solutions of
Kramer’s equation are superpositions of the solutions of Dirac’s equation and the
solutions of an unphysical equation. When we have found the solutions of Kramer’s
equation we still have to project out the solutions of Dirac’s equation [MG58]. This
used to involve additional work. But in the present approach it turns out that this
work is already done.
In this chapter we work out the relations between the solutions given here and
those obtained by the solution of Kramer’s equation. We find that, when Kramer’s
equation is expressed in terms of the quantity Γ, it reduces to the differential
equations for w1 , w2 . When it is expressed in terms of ∆, it reduces to the
equations for u1 , u2 . So we can also find the solutions to the Dirac equation,
uA , uB , by solving Kramer’s equation for w1 , w2 , or u1 , u2 , and then using the
known relations between w1 , w2 and uA , uB , or between u1 , u2 and uA , uB . In
the following sections we shall see how this works out in detail.

17.1 The Eigenvalue Problem for Traceless 2 × 2 Matrices

Let us consider the linear space spanned by two functions, F1 and F2 . We introduce
a linear operator Ω defined by the relation


Ω11 Ω12
Ω(F1 , F2 ) = (F1 , F2 ) , (17.1)
Ω21 Ω22

and assume that

S = c1 F1 + c2 F2 (17.2)

149
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch17

150 The Supersymmetric Dirac Equation

is an eigenfunction of this operator with eigenvalue λ. We have then

 

Ω11 Ω12 c1 c1
=λ . (17.3)
Ω21 Ω22 c2 c2

We want to determine coefficients (a1 , a2 ) and (b1 , b2 ), which enable us to

write S as

S = a1 F1 + a2 ΩF1
S = b1 F2 + b2 ΩF2 . (17.4)

By using Eq. (17.1) we get

S = a1 F1 + a2 (Ω11 F1 + Ω12 F2 ) = (a1 + a2 Ω11 )F1 + a2 Ω12 F2 . (17.5)

Thus, a1 and a2 may be determined from the equations

c1 = a1 + a2 Ω11 ,
c2 = a2 Ω12 . (17.6)

This gives

c2 Ω11
a1 = c1 − ,
Ω12
c2
a2 = , (17.7)
Ω12

and, hence,
  
c1
S = a2 Ω12 − Ω11 F1 + ΩF1 . (17.8)
c2

In a similar way, we get

S = b1 F2 + b2 (Ω21 F1 + Ω22 F2 ), (17.9)

so that

c1 = b2 Ω21 ,
c2 = b1 + b2 Ω22 , (17.10)
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch17

The Relation of the Solutions to Kramer’s Equation 151

or
c1 Ω22
b1 = c2 − ,
Ω21
c1
b2 = , (17.11)
Ω21
and
  
c2
S = b2 Ω21 − Ω22 F2 + ΩF2 . (17.12)
c1
Now, Eq. (17.3) allows us to write
c2 c1
λ = Ω11 + Ω21 = Ω22 + Ω12 . (17.13)
c1 c2
We conclude, therefore, that if the matrix is traceless,

Ω11 = −Ω22 , (17.14)

then we have the following simple result:

S = a2 (λF1 + ΩF1 ) and

S = b2 (λF2 + ΩF2 ). (17.15)

By taking the difference between the two expressions for S, we finally get

Ω(a2 F1 − b2 F2 ) = −λ(a2 F1 − b2 F2 ). (17.16)

Hence, (a2 , −b2 ) is an eigenvector of the matrix in Eq. (17.3) with eigenvalue −λ.
That −λ is an eigenvalue when λ is, is a consequence of the assumption that the
matrix is traceless, Eq. (17.14).

17.2 Eigenfunctions of the Operators Γ and ∆

In the previous sections, we diagonalized the operators Γ and ∆ in the space
of functions VR spanned by the radial functions RA (r) and RB (r) that belong
to the Dirac spinor ψ characterized by the quantum number κ and the reduced
energy . Let us now, with reference to the discussion in Section 17.1, determine
the eigenfunctions of Γ in the space spanned by ψ and the spinor ψ̄ defined by
Eq. (15.3). Let us, similarly, determine the eigenfunctions of ∆ in the space spanned
by ψ and the spinor ψ̃ defined by Eq. (16.2). With reference to Eqs. (15.16) and
(16.16), we denote the eigenfunctions in question by ψγ1 , ψγ2 , ψδ1 , and ψδ2 . To
determine them, we need some special relations that hold for solutions of eigenvalue
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch17

152 The Supersymmetric Dirac Equation

problems of the kind discussed in the previous sections, the salient feature of these
problems being that the two-dimensional matrix to be diagonalized has zero trace.
The radial eigenfunctions of the operator Γ may be read off from the equations
of (15.18):
   
 
w1 uA c1 Zα −c1 (κ + γ|κ|) uA
=D = . (17.17)
w2 uB c2 (Zα) c2 (−κ + γ|κ|) uB

Each of them parallels the function S in Eq. (17.2). That function may also be
expressed in the form (17.4), with factors a2 and b2 as described in the comments to
Eq. (17.16); after identifying the vectors F1 and F2 of the previous section with the
functions uA and uB we find that we may write the radial function corresponding
to the eigenvalue γ2 = −γ|κ| as

w2 = a2 ua − b2 uB = c2 ((Zα)uA + (−κ + γ|κ|)uB ). (17.18)

Therefore

a2 = c2 (Zα), b2 = c2 (κ − γ|κ|). (17.19)

Hence, we get that the eigenfunction corresponding to the eigenvalue γ1 = γ|κ|

is, according to Eq. (17.15):

w1 = a2 (λuA + ΓuA ) = c2 (Zα)(γ|κ|uA + ΓuA ) or

w1 = b2 (λuB + ΓuB ) = c2 (κ − γ|κ|)(γ|κ|uB + ΓuB ). (17.20)

Accordingly, we get that

1
γ|κ|uA (x) + ΓuA (x) = w1 (x),
Zαc2
1
γ|κ|uB (x) + ΓuB (x) = w1 (x). (17.21)
(κ − γ|κ|)c2

The spinor ψγ1 is accordingly given by


(κ − γ|κ|) YA
xψγ1 (x) = N (1) w1 (x) , (17.22)
iZα YB

where N (1) is an arbitrary constant.

August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch17

The Relation of the Solutions to Kramer’s Equation 153

In a similar way, we find that the radial function corresponding to the

eigenvalue γ2 = −γ|κ| may be written as

w2 = c1 (Zα)(−γ|κ|uA + ΓuA ), or
w2 = c1 (κ + γ|κ|)(−γ|κ|uB + ΓuB ); (17.23)

hence, we get that


(2)
(κ + γ|κ|) YA
xψγ2 (x) = N w2 (x) . (17.24)
iZα YB

To determine the spinors ψδ1 and ψδ2 , we exploit the expression (16.12),
which allows us to write for the radial function corresponding to the eigenvalue
δ1 = κ/na the expressions
 
d2 κ
u1 = − √ uA + ∆uA , or
1 +  na
 
d2 κ
u1 = − √ uB + ∆uB . (17.25)
1 −  na
This gives that

√ 
1 +  YA
(3)
xψδ1 (x) = N u1 (x) √ . (17.26)
i 1 −  YB

Finally, we get for the radial function corresponding to the eigenvalue λ2 = −κ/na
the expressions
 
d1 κ
u2 = √ − uA + ∆uA , or
1+ na
 
d1 κ
u2 = − − uB + ∆uB , (17.27)
1− na

and, hence,

√ 
1 +  YA
(4)
xψδ2 (x) = N u2 (x) √ . (17.28)
i 1 −  YB

The eigenspinors above are in general a superposition of states of

the different supersymmetric parities. This is not the case for the ground
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch17

154 The Supersymmetric Dirac Equation

state, which has no supersymmetric partner. In this case, and in this case
alone,

 

(1) (κ − γ|κ|) YA (1
) YA
xψγ1 (x) = N w1 (x) =N w1 (x) ,
iZα YB i κ+γ|κ|
Zα
YB
(17.29)

where w1 (x) = xR0 (x), and we have used Eq. (14.54). Similarly,

√ 

1 +  YA
YA
xψδ2 (x) = N u2 (x) √
(4)
=N (4 )
u2 (x) κ+γ|κ| ,
i 1 −  YB i Zα YB
(17.30)
where u2 (x) = xR0 (x), and we have used (16.36).
The eigenspinors that we have found above are, with respect to supersymmetric
quantum mechanics, the parallels of the eigenfunctions of the Johnson–
Lippmann operator A. Thus, A turns the supersymmetric partners |, j, mj , κ
and |, j, mj , −κ > into each other (Eq. (14.34)). It has the eigenvalues ±e and
the eigenfunctions (14.36). Similarly, the operator Γ turns the spinors ψ and ψ̄
into each other. It has the eigenvalues ±γ|κ| and the eigenfunctions (17.29) and
(17.22). Finally, the operator ∆ turns the spinors ψ and ψ̃ into each other. It has
the eigenvalues ±γ|κ| and the eigenfunctions (17.22) and (17.24).

17.3 Kramer’s Equation

We now consider Kramer’s equation, which is a second-order equation satisfied
by positive-mass and negative-mass spinors alike. To obtain it, we note that the
equations (H −E)ψ = 0 (which follows from the Dirac equation) and (H̄ −E)ψ̄ =
0 (which follows from Eq. (13.5)) may be replaced by the equations O+ ψ = 0 and
O− ψ̄ = 0, where

 
Zα
O+ = β(H − E) = m + βγ5 Σ·p − β E + ,
r
 
Zα
O− = β(H̄ − E) = −m + βγ5 Σ·p − β E + . (17.31)
r
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch17

The Relation of the Solutions to Kramer’s Equation 155

The operators O+ and O− commute. Both ψ and ψ̄ will therefore satisfy the
equation
  2
Zα
O− O+ ψ = −m2 + E + + βγ5 (Σ·p)βγ5 (Σ·p)
r
    
Zα Zα
−βγ5 (Σ·p)β E + −β E+ βγ5 (Σ·p) ψ = 0.
r r
(17.32)

Now

βγ5 (Σ·p)βγ5 (Σ·p) = −(Σ·p)(Σ·p) = −p2 , (17.33)

and
   
Zα Zα
−βγ5 (Σ·p)β E + −β E+ βγ5 (Σ·p)
r r
   
Zα Zα
= γ5 (Σ·p) E + − γ5 (Σ·p) E +
r r
 
1 i
= Zαγ5 , (Σ·p) = 2 Zαγ5 (Σ·r̂). (17.34)
r r

Inserting these expressions in Eq. (17.32) yields

  2
Zα i
−m − p + E +
2 2
+ 2 Zαγ5 Σ·r̂ ψ = 0. (17.35)
r r

This is Kramer’s equation. We note that O− is not the Hermitian adjoint of O+ .

The operator O− O+ is therefore not Hermitian. Although any solution of Dirac’s
equation is automatically a solution of Kramer’s equation the converse is not true:
A solution of Kramer’s equation is a mixture of solutions of the Dirac equation
for positive mass and the Dirac equation for negative mass. To get a solution of
the Dirac equation for positive mass (the physical Dirac equation) we must, in the
conventional procedure, project it out from the solution of Kramer’s equation. This
step, which is somewhat unwieldy in the usual formulation, is here superfluous,
because the linear combination of solutions of the second-order equation, the
confluent hypergeometric functions w1 and w2 , or, alternatively, u1 and u2 , which
are also solutions of the Dirac equation, are already known.
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch17

156 The Supersymmetric Dirac Equation

In spherical polar coordinates

L2 1 ∂ 2 ∂
p2 = − 2 r , (17.36)
r2 r ∂r ∂r
and noting that L2 = K(K + β), Eq. (13.28), the equation takes the
form:

1 ∂ 2 ∂ K 2 − Z 2 α2 βK 2ZαE
−(m2 − E 2 ) + 2 r − − 2 +
r ∂r ∂r r2 r r


i 1 ∂ 2 ∂
+ 2 Zαγ5 Σ·r̂ ψ = −(m2 − E 2 ) + 2 r
r r ∂r ∂r

K 2 − Z 2 α2 Γ 2ZαE
− + 2+ ψ = 0, (17.37)
r2 r r
where

Γ = −βK + iZαγ5 (Σ·r̂). (17.38)

The operator Γ agrees with the expression we have been using, Eq. (15.10):
KH
Γ=− − iZαγ5 A, (17.39)
m
because
 
KH −i
Γ=− − iZαγ5 Kγ5 (H − βm) − Σ·r̂
m Zαm
 
KH H
=− − γ5 Kγ5 − β + iZαγ5 (Σ·r̂)
m m
 
KH H
=− +K − β + iZαγ5 (Σ·r̂)
m m

= −βK + iZαγ5 (Σ·r̂). (17.40)

The operator βK anticommutes with γ5 (Σ·r̂), because

{βK, γ5 (Σ·r̂)} = βKγ5 (Σ·r̂) + γ5 (Σ·r̂)βK = βKγ5 (Σ·r̂) − βγ5 (Σ·r̂)K

= βKγ5 (Σ·r̂) + βγ5 K(Σ·r̂) = βKγ5 (Σ·r̂) − βKγ5 (Σ·r̂)
= 0, (17.41)

and hence, from Eq. (17.40),

Γ2 = K 2 − (Zα)2 . (17.42)
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch17

The Relation of the Solutions to Kramer’s Equation 157

Inserting Eq. (17.42) into Eq. (17.37) yields Kramer’s equation in the
form


1 ∂ 2 ∂ Γ(Γ − 1) 2ZαE
r − + + (E − m ) ψ = 0.
2 2
(17.43)
r2 ∂r ∂r r2 r
We can rewrite the kinetic energy term with the identity
1 ∂ 2 ∂ ∂2 ∂ 1 ∂2
2
r = 2 + 2r = r. (17.44)
r ∂r ∂r ∂r ∂r r ∂r 2
Going over to the variable x, defined by Eq. (13.60), Eq. (17.37) then becomes


∂2 na 1
x−1 2 x − x−2 Γ(Γ − 1) + − ψ = 0. (17.45)
∂x x 4
Assume now that ψ is an eigenfuntion of Γ with eigenvalue +γ|κ|. We may then
replace the operator Γ in Eq. (17.46) by its eigenvalue γ|κ|, by which the whole
operator in front of xψ becomes independent of spin and angular coordinates.
Multiplying on the left by x we get:
 2 
∂ γ|κ|(γ|κ| + 1) na 1
+ + − xψ = 0. (17.46)
∂x2 x2 x 4
The equation may now be factorized, in accordance with Eq. (17.22), and we obtain
a radial function that is seen to satisfy the upper equation in (15.43):


d2 w1 γ|κ|(γ|κ| − 1) na 1
+ − + − w1 = 0. (17.47)
dx2 x2 x 4
Similarly, the eigenvalue −γ|κ| gives a radial function that satisfies the lower
equation in (15.43):


d2 w2 γ|κ|(γ|κ| + 1) na 1
+ − + − w2 = 0. (17.48)
dx2 x2 x 4
Thus, solutions of Kramer’s equation are solutions of the differential equations for
w1 (x) and w2 (x). These are standard differential equations, which have explicit
solutions in terms of confluent hypergeometric functions. These are connected to
the radial solutions of the Dirac equation by the known matrix of Eq. (15.17).
The fact that the differential equations (15.43) may be derived from Kramer’s
equation seems to give the functions w1 (x) and w2 (x) an elevated status with
respect to the Dirac equation. The functions u1 (x) and u2 (x) that arise from
diagonalizing the operator ∆ lead, however, to simpler expressions for uA (x) and
uB (x) and, unlike w1 (x) and w2 (x), the functions u1 (x) and u2 (x) are mutually
orthogonal. Accordingly, it is of interest to note that the second-order differential
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch17

158 The Supersymmetric Dirac Equation

equations for u1 (x) and u2 (x) may also be derived from Kramer’s equation. To
show this, we insert the form of the Hamiltonian into expression (15.10) for Γ. We
make the assumption that κ is a good quantum number and therefore replace K by
κ in expression (17.38) for Γ. We may then show that Γ can be written
∂ Zαm
Γ= r− r[1 + (Σ·r̂)A]. (17.49)
∂r κ
Indeed
−i
(Σ·r̂)A = (Σ·r̂)Kγ5 (H − βm) − 1, (17.50)
Zαm
and therefore
−i i
1 + (Σ·r̂)A = (Σ·r̂)Kγ5 (H − βm) = (Σ·r̂)γ5 K(H − βm)
Zαm Zαm
i iκ
= (Σ·r̂)γ5 (H − βm)K = (Σ·r̂)γ5 (H − βm).
Zαm Zαm
(17.51)

Now, by Eq. (13.53),

 
∂ βK + 1 Zα
(H − βm) = −iγ5 (Σ·r̂) + −
∂r r r
 
1 ∂
= −iγ5 (Σ·r̂) r + βK . (17.52)
r ∂r
Then
 
i ∂ Zα
(Σ·r̂)γ5 (H − βm) = − r + βK − (Σ·r̂)γ5 , (17.53)
r ∂r r
and
   
Zαm i ∂
r[1 + (Σ·r̂)A] = ir − r + βK − (Σ·r̂)γ5 Zα
κ r ∂r
∂
= r + βK − iZαγ5 r(Σ·r̂). (17.54)
∂r
With the definition from Eq. (17.38) this yields
∂ Zαm
Γ= r− r[1 + (Σ·r̂)A]. (17.55)
∂r κ
Now introduce ∆ from Eq. (16.9):
∂ Zαm
Γ= r− r∆. (17.56)
∂r κ
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch17

The Relation of the Solutions to Kramer’s Equation 159

In terms of the variable x from (13.60),

∂ na
Γ= x− x∆. (17.57)
∂x 2κ
Insert this expression for Γ into Eq. (17.45), and, using Eq. (17.42), get


∂2 ∂ na −1 na 1
x−1 2 x − x2 (κ2 − (Zα)2 ) + x−2 x − x ∆+ − ψ = 0.
∂x ∂x 2κ x 4
(17.58)

Note that K commutes with ∆ = 1 + (Σ · r̂)A, because K anticommutes with

Σ ·r̂ and A. Therefore we may choose ψ to be a simultaneous eigenvector of K and
∆. We may replace the operator ∆ with one of its eigenvalues, κ/na or −κ/na ,
and multiply from the left with x3 , whereby the operator in front of xψ becomes
independent of spin and angular coordinates:

2
 
2 ∂ ∂ 1 2 1
x +x − x − γ |κ| + x na ∓
2 2
(xψ) = 0. (17.59)
∂x2 ∂x 4 2
Hence, the equation may be factorized, in accordance with Eqs. (17.26) and (17.28),
and we obtain a radial function that satisfies, for the eigenvalue κ/na ,
   
d 2 u1 du1 1 1
x2 2 + x + − x2 + na − x − γ 2 |κ|2 u1 = 0. (17.60)
dx dx 4 2
For the eigenvalue −κ/Na we get
2
   
2 d u2 du2 1 2 1
x +x + − x + na + x − γ |κ| u2 = 0.
2 2
(17.61)
dx2 dx 4 2
Comparing to Eq. (16.38) we see that the ∆ and Γ representations have comparable
relations to Kramer’s equation. In the ∆ representation the solutions of Kramer’s
equation corresponding to ±κ/na are u1 (x) and u2 (x), which satisfy the
differential equations (16.38). Solving these equations gives explicit expressions
for u1 (x) and u2 (x). Finally, the radial components of the Dirac equation, uA (x)
and uB (x) are obtained from u1 (x) and u2 (x) by using the known matrix of
Eq. (16.18).

Notes on Chapter 17
Kramer’s equation was used by Martin and Glauber [MG58] to obtain solutions
of the Dirac equation, but also a similar idea was behind many of the methods
discussed in the Notes on Chapter 15.
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch17

This page intentionally left blank

August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch18

Chapter 18
Non-Relativistic Approximation

In this chapter we look at the non-relativistic limit of the theory developed in the
previous chapters. For the energy levels we have of course the Bohr formula, as
well as higher order effects. We are interested here in the reduction of the wave
functions. It is of course well known that in the non-relativistic limit the Dirac wave
functions reduce to the Pauli wave functions. Here we gain further insight into the
separation of the Dirac radial functions into the two terms w1 and w2 . It turns out
that for negative values of κ the wave function uA reduces to the two-component
Pauli wave function uN l , and for positive values to the Pauli wave function uN,l+1 .
Since the Dirac wave function is degenerate with respect to positive and negative
values of κ, due to the supersymmetry, it is clear that both components uN l and
uN,l+1 will be present in the relativistic case. In the ∆-scheme, with its separation
of the Dirac of the wave function into u1 and u2 , we find the same result after the
application of recursion relations of the hypergeometric functions.
The results of the computations in this chapter are not surprising. The explicit
demonstration is nevertheless instructive, and verifies many of the calculations
performed in the book. In particular we verify the choice of the overall phase,
Sign(κ), in Eq. (15.55), which ensures the equivalence of the expressions for the
solutions of the radial Dirac equation, which were derived independently. Hence,
our choice of phase is in coincidence with [DJ95].
Relativistic effects become important with increasing velocity. Let v be a
typical velocity. For hydrogenic atoms the Bohr formula for the lowest state,
N = 1, is
m
EBohr = EN − m = − (Zα)2
2
= kinetic energy + potential energy = −kinetic energy, (18.1)

in accordance with the virial theorem [HGS01]: potential energy =−2 × kinetic
energy. So the velocity v can be replaced by the more convenient parameter
ζ = Zα.
161
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch18

162 The Supersymmetric Dirac Equation

With the formula (A.9):

∞


− 12 (1/2)k 1 3
(1 − z) = z k = 1 + z + z 2 + O(z 3 ) (18.2)
k! 2 8
k=0

we have for the Dirac energy

  2 − 12
ζ
E = m 1+
nr + γ|κ|
 
ζ2 ζ4 3 ζ4
= m 1− + + + O(ζ 6 ), (18.3)
2N 2 N 3 (2j + 1) 8 N 4

with N = nr + |κ| = 1, 2, 3 . . . , (N − 1)/2. To order ζ 2 we have, as expected,

ζ2
 − 1 ≈ Bohr = − . (18.4)
2N 2

The terms of order O(ζ 4 ) may be calculated in first-order perturbation theory from
the Pauli formalism, for the relativistic mass-correction, the Thomas precession
[ Jac98], and the so-called Darwin term [ Dar28], which is a result of the
zitterbewegung [Ros61].
The apparent principal quantum number na has the non-relativistic
approximation

ζ
na = √ ≈ N. (18.5)
1 − 2

We then have
 
mζ
x≈2 r. (18.6)
N

Note that γ ≈ 1 + O(ζ 2 ). For positive κ we have

κ − γ|κ| ≈ κ − |κ| ≈ O(ξ 2 ) (18.7)

and for negative κ, we have

κ + γ|κ| ≈ κ + |κ| ≈ O(ξ 2 ). (18.8)

From Eq. (15.47) the radial quantum number is nr = na − γ|κ| ≈ N − |κ|.
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch18

Non-Relativistic Approximation 163

18.1 The Γ Representation

The exact expressions for the wave functions are, for the Γ-scheme:

γ|κ| −x/2

uA (x) = Sign(κ)N x e (κ) − γ|κ|)(κ − γ|κ)1 F1 (−nr , 2γ|κ|; x)

ena x
+ (κ + γ|κ|)(κ − γ|κ|)1 F1 (−nr + 1, 2γ|κ|+2; x) ,
2γ|κ|(2γ|κ| + 1)

γ|κ| −x/2

uB (x) = N x e (κ + γ|κ|)(κ − γ|κ|)1 F1 (−nr , 2γ|κ|; x)

ena x
+ (κ − γ|κ|)(κ + γ|κ|)1 F1 (−nr + 1, 2γ|κ|+2; x) .
2γ|κ|(2γ|κ| + 1)
(18.9)

In the non-relativistic limit we have, from Eq. (D.16), for |κ| = l,

 
1 Γ(2γ|κ| + nr ) 1 (N + l − 1)!
N = ≈ . (18.10)
Γ(2γ|κ| + 1) nr !(2na ) (2l)! (N − l)!(2N )

For positive κ, the orbital angular momentum is ¯l = j + 1

2
= |κ| = l + 1,
n̄r = nr − 1, and N = l + nr = ¯l + n̄r . We note that

ena ≈ N 2 − |κ|2 . (18.11)
The approximate expressions for the wave functions are

|κ|+1 −x/2 N 2 − |κ|2
uA (x) ≈ N x e 1 F1 (−nr + 1, 2|κ| + 2; x)
(2|κ| + 1)

= AN l̄ xl̄+1 e−x/2 1 F1 (−n̄r , 2¯l + 2; x) = uN,l+1 (x), (18.12)

since

1 (N + ¯l − 1)!(N + ¯l)(N − ¯l)
(2l)!(2¯l + 1)
¯ (N − ¯l)!(2N )

1 (N + ¯l)!
= ¯ = AN l̄ , (18.13)
(2l + 1)! (N − ¯l − 1)!(2N )
from Eq. (9.29). The small component is uB (x) = O(ζ).
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch18

164 The Supersymmetric Dirac Equation

For negative κ, the orbital angular momentum is l = j − 12 = |κ| − 1.

The approximate expressions for the wave functions of Eq. (18.9) are
then

uA (x) ≈ −2|κ|N x|κ| e−x/2 1 F1 (−nr , 2|κ|; x)

= −AN l xl+1 e−x/2 1 F1 (−nr , 2l + 2; x) = −uN l (x). (18.14)

where N = nr + |κ| = nr + l + 1. Here we have used


1 (N + l)!
(2l + 2)N ≈ = AN l . (18.15)
(2l + 1)! 2N (N − l − 1)!

We have, again, uB (x) ≈ O(ζ).

The physical state is, for positive κ,
 
(1/r)uN,l+1 (r)χκ (θ, φ)
ψ≈ , (18.16)
O(ζ)

and for negative κ,

 
−(1/r)uN,l (r)χκ (θ, φ)
ψ≈ . (18.17)
O(ζ)

In both cases the relativistic state reduces in the non-relativistic limit to the Pauli
state.

18.2 The ∆ Representation

For the ∆-scheme, the exact expressions for the wave functions are:

N√
uA (x) = 1 +  xγ|κ| e−x/2
2
× [nr 1 F1 (−nr + 1, 2γ|κ| + 1; x) − (na − κ)1 F1 (−nr , 2γ|κ| + 1; x)],
N√
uB (x) = 1 −  xγ|κ| e−x/2
2
× [nr 1 F1 (−nr + 1, 2γ|κ| + 1; x) + (na − κ)1 F1 (−nr , 2γ|κ| + 1; x)].
(18.18)
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch18

Non-Relativistic Approximation 165

In the non-relativistic limit, for positive κ,

N
uA (x) ≈ √ x|κ| e−x/2 nr [1 F1 (−nr + 1, 2|κ| + 1; x) − 1 F1 (−nr , 2|κ| + 1; x)],
2
uB (x) ≈ 0. (18.19)

With the identities (E.4) and (D.4) we get

1 F1 (−nr + 1, 2|κ| + 1; x) − 1 F1 (−nr , 2|κ| + 1; x)

 
x
= 1 F1 (−nr + 1, 2|κ| + 2; x). (18.20)
2|κ| + 1
Inserting this yields
 
N nr
uA (x) ≈ √ x|κ|+1 e−x/2 1 F1 (−nr + 1, 2|κ| + 2; x). (18.21)
2 (2|κ| + 1)

Now use |κ| = ¯l and n̄r = nr − 1. From Eq. (E.12),

 
1 Γ(n + 2|κ| + 1) 1 (N + |κ|)!
N =
r
≈ (18.22)
Γ(2|κ| + 1) na (na − κ)nr ! (2|κ|)! N n r nr !

and
   
nr 1 (N + ¯l)! 1 (N + ¯l)!
= ¯ = AN l̄ . (18.23)
2¯l + 1 (2¯l)! (2N ) nr nr ! (2l + 1)! (2N ) n̄r !

Hence

uA (x) ≈ AN l̄ xl̄+1 e−x/2 1 F1 (−n̄r , 2¯l + 1; x) = uN,l+1 (x). (18.24)

For negative −κ = |κ| = l + 1 we have, in the non-relativistic limit,

N
uA (x) ≈ √ x|κ| e−x/2
2
× [nr 1 F1 (−nr + 1, 2|κ| + 1; x) − (nr + 2|κ|)1 F1 (−nr , 2|κ| + 1; x)].
(18.25)

Use of the identity [GR65]

nr 1 F1 (−nr + 1, 2|κ| + 1; x) − (nr + 2|κ|)1 F1 (−nr , 2|κ| + 1; x)

= −2|κ|1 F1 (−nr , 2|κ|; x) (18.26)
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch18

166 The Supersymmetric Dirac Equation

yields
N
uA (x) ≈ − √ x|κ| e−x/2 2|κ|1 F1 (−nr , 2|κ|; x). (18.27)
2
Now use |κ| = l + 1.
N
uA (x) ≈ − √ xl+1 e−x/2 (2l + 2)1 F1 (−N + l + 1, 2l + 2; x)
2
= −AN l xl+1 e−x/2 1 F1 (−N + l + 1, 2l + 2; x) = −uN l (x), (18.28)

since

N 1 (N + l + 1)!
(2l + 2) √ ≈
2 (2l + 1)! 2N (N + l + 1) (N − l − 1)!

1 (N + l)!
= = AN l , (18.29)
2l + 1 2N (N − l + 1)!
from

1 (N + l + 1)!
N ≈ , (18.30)
(2l + 2)! N (N + l + 1)(N − l − 1)!
see Eq. (E.12).
We have regained the wave functions of Eqs. (18.16) and (18.17).
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch19

Chapter 19
Conclusions

“There’s a reason physicists are so successful at what they do, and that is they study the
hydrogen atom and the helium ion and then they stop.”
Attributed to Richard Feynman by J. Rigden, in Hydrogen: The Essential Element, Harvard
University Press, Cambridge, MA, 2002.

One of the central concepts in this book is the Laplace vector, which gives
us the connection between an attractive central force, which falls off inversely
proportional to the distance squared, and the elliptical orbits that a particle moving
under the influence of this force follows. It governs the motion of a planet about the
Sun, and of an electron about the nucleus. It is valid in the classical domain, and, in a
suitably modified form, also in the quantum domain. With a further generalization, a
remnant of the Laplace vector even appears as a conserved quantity in the relativistic
version of this problem at the quantum level.
Neglecting for a moment the spin of the electron, we see that the three
conserved components of the Laplace vector A, together with the three conserved
components of the angular momentum vector L, suitably normalised, form a
representation of the closed Lie algebra so(4). This gives a method for computing
the spectrum of a non-relativistic hydrogen-like atom, including the degeneracy of
the spectral lines.
In the classical problem the magnitude of the Laplace vector gives the
eccentricity of the orbits. Sommerfeld constructed a semi-classical model of the
hydrogen atom, which gives the correct positions of the spectral lines, and a
characterization of the orbits according to their eccentricities. The model works for
the non-relativistic hydrogen atom, as well as for the relativistic hydrogen atom.
Even though it neglects the spin of the electron it gives the fine structure of the
spectrum as it results from Dirac’s relativistic theory, and the correct expression
for the eccentricity of the orbits.
When we want to take the spin of the electron into account we are led naturally
to the Pauli formalism, where the wave function of the electron has two components.

167
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch19

168 The Supersymmetric Dirac Equation

The electron momentum is replaced by the quantity (σ·p), with σi , i = 1, 2, 3 the

Pauli matrices. In a similar way the Laplace vector A is replaced by the quantity
Anr
= (σ·A), and the magnitude of this quantity is identified with the eccentricity:
e = A2nr . An important role is played by the eigenvalues of the operator Knr =
−(σ · L + 1), the non-relativistic counterpart of a constant operator introduced
by Dirac. It turns out that the operator Anr acts on eigenstates of Knr in such a
way as to change the eigenvalue κ of Knr into −κ. In this way it connects two
states of the system with identical energies, and establishes a supersymmetry of the
system. It allows us to construct Hermitian conjugate operators A− and A+ which
act as ladder operators to construct all states of the system from a supersymmetric
vacuum. The complete degeneracy of the spectrum is accounted for.
The Dirac equation is introduced as a straightforward generalization of the
Pauli equation to four-component spinors, with the γµ matrices, µ = 1 to 4,
replacing the Pauli matrices. The primary supersymmetry of the states of an
electron in a Coulomb field is generated by the Johnson–Lippmann operator A, the
relativistic version of the operator Anr , to which it reduces in the non-relativistic
limit. The SO(4) symmetry is lost, as is the SO(3) symmetry associated to the
orbital angular momentum. But there is an SO(3) symmetry associated to the total
angular momentum, including the spin. Thus the (super)symmetry of the relativistic
hydrogen atom is SO(3) × S(2).
Recognizing the supersymmetry of the Dirac equation allows us to construct
in a systematic way its solutions. It turns out that extending the space of solutions
to include the solutions of the Dirac equation for negative mass leads to a
pair of confluent hypergeometric functions w1 (x) and w2 (x), of which definite
linear combinations form solutions of the Dirac equation. These hypergeometric
functions are solutions of second-order differential equations that factorize under
supersymmetry to yield first-order equations involving conjugate operators a− and
a+ that are relativistic generalizations of the operators A− and A+ used in the Pauli
formalism.
Another extension of the space of solutions leads to an alternative pair of
confluent hypergeometric functions u1 (x) and u2 (x), which are known forms of
the solutions of the Dirac equation.
An alternative method of solving the Dirac equation consists of solving a
second-order differential equation of Schrödinger form, the Kramer’s equation. Its
solutions w1 (x) and w2 (x) live in the extended solution space described above.
It is then necessary to project out from these solutions the appropriate linear
combinations that are solutions of the Dirac equation. An analysis of this method
reveals that it again involves a factorization of the second-order differential operator
identical to that discussed above.
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch19

Conclusions 169

Finally, performing the non-relativistic reduction of the solutions of the Dirac

equation is shown to lead to the well-known solutions of the Pauli equation. This
serves as a consistency check of the normalizations and the phase choices made in
the course of developing the formalism.
Feynman’s comment on the successes achieved through the methods employed
here is a measure of the extent to which relativistic quantum mechanics is amenable
to exact solution. Further investigations of electromagnetic processes are all limited
to studying perturbative expansions based on the exact solutions.
Figure 19.1 attempts to show how our understanding of the spectrum of
hydrogen has developed over the years, from the Bohr model for the non-relativistic
case, through the relativistic Sommerfeld model and the Dirac theory, up to
the Lamb shift, which demonstrates the need for a full quantum field-theoretic
treatment.

2s 29

2p 2p 2p
3/2 3/2

2p
1/2

2s 2s 2p 2s
1/2 1/2 1/2

Spin and Quantum

Relativity Theory Electrodynamics

Bohr 1913 Sommerfeld Dirac 1928 Lamb and

1916 Retherford 1947

Figure 19.1. The development of our understanding of the hydrogen spectrum.

August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-ch19

This page intentionally left blank

August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-App

Appendices

Appendix A. The Confluent Hypergeometric Function

The hypergeometric functions are a wide class that include all the special functions
of mathematical physics, such as the Hermite, Bessel and Legendre polynomials,
and the associated Legendre polynomials. The Laguerre polynomials and the
associated Laguerre functions are described by the confluent hypergeometric
function.
The confluent hypergeometric function is a solution of the differential equation

2 
d d
x 2 + (c − x) − a 1 F1 (a, c; x) = 0. (A.1)
dx dx
It has solutions
∞
 (a)k k
1 F1 (a, c; x) = a0 x , (A.2)
k!(c)k
k=0

where (a)k are the Pochhammer symbols defined by

(a)0 = 1,
(a)n = a(a + 1)(a + 2) · · · (a + n − 1), n = 1, 2, 3, . . . . (A.3)

They satisfy identities such as

n! = (n − m)!(n − m + 1)m ,
(c − m + 1)m = (−1)m (−c)m , (A.4)
(n + m)! = n!(n + 1)m .

A function that is analytic in the interval a < x < b may be expanded in a

Taylor series
∞
 f (n) (x0 )
f (x) = (x − x0 )n , (A.5)
n!
k=0

for any x0 in the interval.

171
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-App

172 The Supersymmetric Dirac Equation

For example, the function

f (z) = (1 − z)s (A.6)

is analytic for any real s with z in the interval −1 < z < 1. Its derivatives are

f  (z) = −s(1 − z)s−1

f  (z) = (−1)2 s(s − 1)(1 − z)s−2 (A.7)
f (k)
(z) = (−1) s(s − 1) · · · (s − k + 1)(1 − z)
k s−k
,

from which we can construct the Taylor series,

∞
 ∞

[f (k) (z)]z=0 (−1)k s(s − 1) · · · (s − k + 1)
(1 − z)s = zk = zk.
k! k!
k=0 k=0
(A.8)

This can be written

∞
 (−s)k
(1 − z)s = zk , (A.9)
k!
k=0

where we have used the identity (−s)k = (−1)k (s − k + 1)k . With s = −1, this
reduces to the geometric series

 ∞
1
= zk , (A.10)
1−z
k=0

which converges for |z| < 1.

An important identity of the Pochhammer symbols is given by Vandermonde’s
theorem:
n
(a)m (b)n−m (a + b)n
= , (A.11)
m=0
m! (n − m)! n!

where a and b are numbers independent of the summation index m. To prove it we

express s in Eq. (A.9) as the sum of two numbers a and b and set z = −x to obtain

∞
(a + b − n + 1)n n
(1 + x)a+b = x . (A.12)
n=0
n!
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-App

Appendices 173

This function can also be written as a product

∞
 ∞

a+b a b (a − k + 1)k k (b − j + 1)j
(1 + x) = (1 + x) (1 + x) = x xj .
k! j!
k=0 j=0
(A.13)

This is the Cauchy product of two series

∞
 ∞
 ∞

Sk Tj = Un , (A.14)
k=0 j=0 n=0

where

n
Un = Sm Tn−m . (A.15)
m=0

So the Cauchy product of the two series in Eq. (A.13) is

∞  n
(a − m + 1)m (b − n + m + 1)n−m n
(1 + x)a+b = x . (A.16)
n=0 m=0
m! (n − m)!

Subtraction of Eq. (A.16) from Eq. (A.12) yields

∞
 
 (a + b − n + 1)n n
(a − m + 1)m (b − n + m + 1)n−m n
− x = 0.
n=0
n! m=0
m! (n − m)!
(A.17)

Since this holds for arbitrary x in the interval −1 < x < 1, the terms in different
powers of x are linearly independent. Therefore, the coefficient of each power of
x must vanish separately, that is,

n
(a − m + 1)m (b − n + m + 1)n−m (a + b − n + 1)n
= . (A.18)
m=0
m! (n − m)! n!

Now use Eq. (A.4) to get

∞
(−a)m (−b)n−m (−a − b)n
= . (A.19)
m=0
m! (n − m)! n!

On replacing −a by a and −b by b, Eq. (A.11) follows.

August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-App

174 The Supersymmetric Dirac Equation

Laguerre polynomials
Consider the function 1 F1 (−q, 1; x). To have a polynomial q must be a positive
integer. Then, in the series
 (−q)k
1 F1 (−q, 1; x) = xk (A.20)
k!k!
k=0

all terms for k > q vanish. Now,

k
(−q)k k (q − k + 1)k k (q + 1)m (−k)k−m
= (−1) = (−1) , (A.21)
k! k! m=0
m! (k − m)!

where we have invoked Vandermonde’s theorem.

With this result we can write,
∞
 ∞ 
 (−1)k  (q + 1)m (−k)k−m
1 F1 (−q, 1; x) = xk . (A.22)
k! m=0
m! (k − m)!
k=0

Now we use the identities

(m + q)! = q!(q + 1)m (A.23)

and

k! = (−1)k−m m!(−k)k−m , m ≤ k, (A.24)

to obtain

1 F1 (−q, 1; x)

∞
 
1  
k
1 (−1)m (m + 1) · · · (m + q) m
= xk−m x
q! m=0
(k − m)! m!
k=0

∞
 
1  
k
xk−m (−1)m dq m+q
= x . (A.25)
q! m=0
(k − m)! m! dxq
k=0

We recognize this as the Cauchy product of two series. Thus,

∞ ∞
1  xr dq  (−1)s s+q 1 x dq q x
1 F1 (−q, 1; x) = x = e (x e ), (A.26)
q! r=0 r! dxq s=0 s! q! dxq
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-App

Appendices 175

which is a polynomial of degree q. It is, up to a normalization constant, the Laguerre

polynmial of Eq. (9.35).
The generating function for the Laguerre polynomials is
∞
 Lk (x)
e−xt/(1−t)
= tk . (A.27)
1−t k!
k=0

Let us differentiate both sides of Eq. (A.27) p times with respect to x,

∞
  ∞

1 dp k Lpk (x) k
L k (x) t = t . (A.28)
k! dxp k!
k=0 k=0

If this differentiation is carried out in Eq. (A.27), the generating function for the
associated Laguerre polynomials is found to be

 Lpq (x) ∞
tp
(−1)p e−xt/(1−t) = tq . (A.29)
(1 − t)p+1
q=0
q!

Notes on Appendix A
A good source of information on hypergeometric functions is the book by Seaborn
[Sea80]. The recurrence relations used in the text are given there in terms of the
associated Laguerre polynomials.

Appendix B. Orthogonality Relations

of Hypergeometric Functions
We wish to calculate the normalization of the radial eigenfunctions of the bound
states of a particle in a Coulomb potential, Eq. (9.39):

 32
2Zαm
NN l xl e− 2 x L2l+1
1
RN l (r) = N +l (x), (B.1)
N

with x = 2(Zαm/N )r. The complete eigenfunctions RN l (r)Ylm (θ, φ) are

considered to be orthonormal; therefore

∞
RN l (r)RN
l (r)r2 dr = δN N
. (B.2)
0

This leads to the appropriate orthogonality relation for the Laguerre polynomials.
By substituting from Eq. (B.1) into the integral in Eq. (B.2) and changing the
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-App

176 The Supersymmetric Dirac Equation

variable to x we find that

∞
∞
[RN l (r)]2 r 2 dr = NN
2
l x2l+2 e−x [L2l+1 2
N +1 (x)] dx. (B.3)
0 0

Clearly, we are interested in the integral

∞
xp+1 e−x [Lpq (x)]2 dx. (B.4)
0

The form of the integrand in Eq. (B.4) suggests using the generating function
in Eq. (A.29) to write
 
p+1 −x p up −xu/(1−u) p vp −xv/(1−v)
x e (−1) e (−1) e
(1 − u)p+1 (1 − v)p+1
∞ 
 ∞
uj v k
= xp+1 e−x Lpj (x)Lpk (x). (B.5)
j=0 k=0
j!k!

We integrate both sides over the full range of x to obtain

∞
  uj v k p∞
Γ(p + 2)(uv)p
(1 − u)(1 − v) = L (x)Lpk (x). (B.6)
(1 − uv)p+2 j!k! j
j=0 k=0

In carrying out this integration we use

∞
Γ(z + 1)
xz e−ax dx = (B.7)
0 az+1

to show that

∞
xp+1 e−x[1+u/(1−u)+v/(1−v)] dx
0
 −p−2
u v
= Γ(p + 2) 1 + +
(1 − u) (1 − v)
 p+2
(1 − u)(1 − v)
= Γ(p + 2) . (B.8)
(1 − uv)

We now use Eqs. (A.5) and (A.10) in

∞
 (p + 2)k
(1 − uv)−(p+2) = (uv)k (B.9)
k!
k=0
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-App

Appendices 177

to obtain
∞
 (p + 2)k
(1 − u)(1 − v)(p + 1)! (uv)k+p
k!
k=0
∞
 (p + k + 1)!
= (1 − u)(1 − v) (uv)k+p
k!
k=0

∞
= Lpj (x)Lpk (x)dx. (B.10)
0

Since u and v are arbitrary and independent, the coefficients of uj v k for any given
j and k must be the same on both sides of this equation. In particular if j = k = q,

∞
(q + 1)! q! 1
+ = xp+1 e−x [Lpq (x)]2 dx. (B.11)
(q − p)! (q − p − 1)! (q!)2 0
From this result the value of the integral is seen to be

∞
[Γ(q + 1)]3
xp+1 e−x [Lpq (x)]2 dx = (2q + 1 − p). (B.12)
0 (q − p)!
Another integral we will need for normalizing the Dirac radial
functions is

∞
xp e−x [Lpq (x)]2 dx. (B.13)
0

Following the same procedure as before we get

 
up − (1−u)
xu vp − (1−v)
xv
xp e−x (−1)p e (−1) p
e
(1 − u)p+1 (1 − v)p+1
∞ 
 ∞
uj v k
= xp e−x Lpj (x)Lpk (x). (B.14)
j=0 k=0
j!k!

We integrate both sides over the full range of x to obtain

∞  ∞

Γ(p + 1)(uv)p uj v k ∞ p −x p
= x e Lj (x)Lpk (x)dx. (B.15)
(1 − uv)p+1 j!k! 0
j=0 k=0

In carrying out this integration we this time get

∞  −p−1
u v
xp e−x[1+u/(1−u)+v/(1−v)] dx = Γ(p + 1) 1 + + .
0 (1 − u) (1 − v)
(B.16)
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-App

178 The Supersymmetric Dirac Equation

If the denominator on the left-hand side of Eq. (B.15) is expanded as in Eq. (A.9),
then Eq. (B.15) becomes
∞ ∞  ∞

(p + 1)s uj v k ∞ p −x p
Γ(p + 1) (uv)s+p = x e Lj (x)Lpk (x)dx.
s=0
s! j=0
j!k! 0
k=0
(B.17)
Now, Lpj (x)
= 0 for j < p, therefore, we can change the lower limits on the sums
on the right-hand side of Eq. (B.17) to p without affecting the sums. Next, redefine
the summation index on the left-hand side to be j = s + p. This gives
∞ ∞  ∞

(p + 1)j−p uj v k ∞ p −x p
Γ(p + 1) j
(uv) = x e Lj (x)Lpk (x)dx.
j=p
(j − p)! j=p
j!k! 0
k=p
(B.18)
Now use the definition of the Kronecker delta to rewrite the left-hand side of this
equation. After some rearrangement of terms the equation then reads
∞  ∞ 
∞
1
j k
u v xp e−x Lpj (x)Lpk (x)dx
j=p k=p
j!k! 0

Γ(p + 1)(p + 1)k−p
− δjk = 0. (B.19)
(k − p)!
The coefficients of uj v k for different combinations of j and k must vanish
separately. Thus,

∞
(k!)2 p!(p + 1)k−p
xp e−x Lpj (x)Lpk (x)dx = δjk . (B.20)
0 (k − p)!
With k − p equal to an integer Γ(p + 1)(p + 1)k−p = Γ(k + 1), from Eq. (A.4).
Therefore, for the more general case where q − p is an integer, but q and p are not
necessarily integers,

∞
[Γ(q + 1)]3
xp e−x [Lpq (x)]2 dx = . (B.21)
0 (q − p)!
Expressed in terms of 1 F1 (−n; 2a; x) the orthogonality relation for the Lpn (x)
becomes

∞

[Γ(2a + 1)]2 n!2 ∞ −x 2a 2a

e−x x2a 1 F12 (−n, 2a; x)dx = 4
e x [L2a+n (x)]2 dx
0 [Γ(2a + n)] 0

[Γ(2a)]2
= 2(a + n) n!, (B.22)
Γ(2a + n)
where we have used Eq. (9.38).
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-App

Appendices 179

From (B.12) we see that the normalized eigenfunctions are


3
(N − l − 1)! 2Zαm 2 −Zαmr/N
RN l (r) = − e
(N + l)!3 (2N ) N

l
2Zαmr
× L2l+1
N +1 (2Zαr/N )
N

3
1 (N + l)! 2Zαm 2 −(Zαmr/N )
= e
(2l + 1)! (N − l − 1)!(2N ) N

l
2Zαmr
× 1 F1 (−N + l + 1, 2l + 2; 2Zαr/N ). (B.23)
N

Notes on Appendix B
Be careful of different conventions concerning the associated Laguerre functions,
for example Mertzbacher [Mer61] vs Landau and Lipshitz [LL77].

Appendix C. More Integrals Involving Hypergeometric

Functions
For the sake of completeness, we give here the application of the formulas of
Landau and Lipshitz [LL77], Appendix F, dealing with the calculation of integrals
involving confluent hypergeometric functions.
To calculate the normalization we need the following integrals:

∞
I11 = z 2a+2 e−z [1 F1 (1 − n, 2a + 2; z)]2 dz,
0

∞
I22 = z 2a e−z [1 F1 (−n, 2a; z)]2 dz, (C.1)
0

∞
I12 = z 2a+1 e−z 1 F1 (−n, 2a; z)1 F1 (1 − n, 2a + 2; z) dz.
0

These integrals are special cases of the integral (F.16) discussed by Landau
and Lipshitz, namely

∞


Jµsp (α, α ) = z µ−1+s e−(k+k )z/2 1 F1 (α, µ; kz)1 F1 (α ; µ − p; k  z)dz.
0
(C.2)
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-App

180 The Supersymmetric Dirac Equation

To calculate these integrals we make use of the hypergeometric function

∞
(a)n (b)n n
2 F1 (a, b, c; z) = z . (C.3)
n=0
n!(c)n

The definition of the generalized hypergeometric function is

∞
(a1 )n (a2 )n · · · (ap )n n
p Fq (a1 , a2 , . . . , ap , b1 , b2 , . . . , bq ; z) = z , (C.4)
n=0
n!(b 1 )n (b2 )n · · · (bq )n

where p is the number of numerator parameters ai and q is the number of

denominator parameters bj .
To specialize (C.2) to the integrals above, we must set k = k  , which we do
by choosing k = 1 + , k = 1 −  and then taking the limit  → 0. The need
for this procedure is made clear by looking at a particular case of (C.2), namely,
s = p = 0, which, according to (F.13), is


Jµ00 (α, α ) = (−1)α Γ(µ)−(α+α ) 2 F1 (α, α , µ; −−2 ), (C.5)

where we have used the hypergeometric function

 αα z α(α + 1)α (α + 1) z 2

2 F1 (α, α , µ; z) = 1 + + + .... (C.6)
µ 1! µ(µ + 1) 2!

The indices α and α are negative integers, so that Eq. (C.6) is a polynomial which,
in the limit  → 0, is determined by the last term only. This is summarized in the
relation


0 m > 2n2 ,


n1 ! Γ(µ)
m 2 F1 (α, α , µ; −−2 ) −−−→ (−1)n2 m = 2n2
→0 
 (n1 − n2 )! Γ(µ + n2 )


undefined m < 2n2
(C.7)
where n1 = max(−α, −α ), and n2 = min(−α, −α ).
The integrals in (C.2) are

10
I11 = J2a+2 (1 − n, 1 − n),
10
I22 = J2a (−n, −n), (C.8)
02
I12 = J2a+2 (1 − n, n).
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-App

Appendices 181

The relationship (F.16), for p = 0,


α − α
Jµ10 (α, α ) = + µ − α − α Jµ00 (α, α ), (C.9)


enables us to evaluate the first two integrals of (C.9):

00
I11 = (2a + 2n)J2a+2 (1 − n, 1 − n)

[Γ(2a + 2)]2
= 2(a + n) (n − 1)!
Γ(2a + n + 1)

[Γ(2a)]2
00
I22 = (2a + 2n)J2a (−n, −n) = 2(a + n) n! (C.10)
Γ(2a + n)

The value of I22 calculated here agrees with Eq. (B.22), calculated by more
elementary methods.
A little more work is required for I12 . The recursion formula, Eq. (F.15),
namely

Jµsp (α, α ) = (µ − 1)[Jµ−1

s,p−1
(α, α ) − Jµ−1
s,p−1
(α − 1, α )], (C.11)

applied twice, yields

02
J2a+2 (1 − n, −n) = 2a(2a + 1)[J2a
00
(1 − n, −n))
− 2J2a
00
(−n, −n) + J2a
00
(−1 − n, −n)]. (C.12)

According to (C.7), only the middle term survives, and

[Γ(2a)]2
I12 = −2(2a + 1) n! (C.13)
Γ(2a + n)

Notes on Appendix C
Appendix F of Landau and Lipshitz [LL77] is an invaluable source of information
on integrals of hypergeometric functions, which often seems to be insufficiently
appreciated. The limit processes necessary for the application of the results of this
appendix to the cases considered follow [GI97].
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-App

182 The Supersymmetric Dirac Equation

Appendix D. Normalization for the Γ-Induced Scheme

We have the solutions

w1 (x) = N1 xγ|κ| e−x/2 1 F1 (−nr , 2γ|κ|; x)

(D.1)
w2 (x) = N2 xγ|κ|+1 e−x/2 1 F1 (−nr + 1, 2γ|κ| + 2; x).

To determine N1 /N2 insert this into the first of the equations in (15.34), where
−
b is defined in Eq. (15.33):


− d γ|κ| na ena
b w1 = − + − w1 = − w2 . (D.2)
dx x 2γ|κ| 2γ|κ|

First calculate


dw1 γ|κ| 1
= N1 xγ|κ| e−x/2 − 1 F1 (−nr , 2γ|κ|; x)
dx x 2

d1 F1 (−nr , 2γ|κ|; x)
+ . (D.3)
dx

Use the relation

d1 F1 (−nr , 2γ|κ|; x) −nr

= 1 F1 (−nr + 1, 2γ|κ| + 1; x). (D.4)
dx 2γ|κ|

Eq. (D.2) then becomes



na 1
N1 − 1 F1 (−nr , 2γ|κ|; x)
2γ|κ| 2

 
nr
− 1 F1 (−nr + 1, 2γ|κ| + 1; x)
2γ|κ|


ena
= N2 x1 F1 (−nr + 1, 2γ|κ| + 2; x). (D.5)
2γ|κ|

We remember that nr = na − γ|κ|. Then


nr
N1 {1 F1 (−nr , 2γ; x) − 1 F1 (−nr + 1, 2γ|κ| + 1; x)}
2γ|κ|


ena
= N2 x1 F1 (−nr + 1, 2γ|κ| + 2; x). (D.6)
2γ|κ|
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-App

Appendices 183

Use the relation [GR65]

(nr + 2γ|κ|)
− x1 F1 (−nr + 1, 2γ|κ| + 2; x)
2γ|κ|(2γ|κ| + 1)

= 1 F1 (−nr , 2γ|κ|; x) − 1 F1 (−nr + 1, 2γ|κ| + 1; x). (D.7)

This yields


nr (nr + 2γ|κ|)
N1 x1 F1 (−nr + 1, 2γ|κ| + 2; x)
2γ|κ|
2γ|κ|(2γ|κ| + 1)


ena
= −N2 x1 F1 (−nr + 1, 2γ|κ| + 2; x). (D.8)
2γ|κ|

The result is



N1 ena 2γ|κ|(2γ|κ| + 1) 1
=− =− 2γ|κ|(2γ|κ| + 1), (D.9)
N2 nr (nr + 2γ|κ|) ena

with (nr + 2γ|κ|)nr = (na + γ|κ|)(na − γ|κ|) = 2 n2a − γ 2 |κ|2 = n2a − κ2 =

(ena )2 , and we have used Eqs. (14.33), (13.61) and (14.50).
We now wish to establish the overall normalization constant. We have the
normalization condition

 2 
uA (x) + u2B (x) dx = 1. (D.10)

Inserting uA (x) and uB (x) yields



N 2
2κ(κ − γ|κ|) dx x2γ|κ| e−x [1 F1 (−nr , 2γ|κ|; x)]2

4ena 
+ (κ2 − γ 2 |κ|2 )(κ2 2 − γ 2 |κ|2 )
2γ|κ|(2γ|κ| + 1)

× dx x2γ|κ|+1 e−x 1 F1 (−nr , 2γ|κ|; x)1 F1 (−nr + 1, 2γ|κ| + 2; x)

(ena )2
+ 2κ(κ + γ|κ|)
(2γ|κ|)2 (2γ|κ| + 1)2


2γ|κ|+2 −x
× dx x 2
e [1 F1 (−nr + 1, 2γ|κ| + 2 : x)] = 1. (D.11)
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-App

184 The Supersymmetric Dirac Equation

Evaluating the integrals according to Appendix C:

[ Γ(2γ|κ|) ]2
x2γ|κ| e−x [1 F1 (−nr , 2γ|κ|; x)]2 dx = 2(2γ|κ| + nr ) nr !
Γ(2γ|κ| + nr )

x2γ|κ|+2 e−x [1 F1 (−nr + 1, 2γ|κ| + 2; x)]2 dx

[Γ(2γ|κ| + 2)]2
= 2(2γ|κ| + nr ) (nr − 1)!
Γ(2γ|κ| + nr + 1)

x2γ|κ|+1 e−x 1 F1 (−nr , 2γ|κ|; x)1 F1 (−nr + 1, 2γ|κ| + 2; x)dx

[Γ(2γ|κ|)]2
= −2(2γ|κ|)(2γ|κ| + 1) (nr )! (D.12)
Γ(2γ|κ| + nr )

so that

1 [ Γ(2γ|κ|) ]2
=4 nr !
N 2 Γ(2γ|κ| + nr )
 
κ(κ − γ|κ|)na − 2(ena )Zα|κ| 2 − γ 2


κ(κ + γ|κ|)(ena )2 na
+ . (D.13)
nr (nr + 2γ|κ|)

Now use

|κ|2 (2 − γ 2 )
nr (nr + 2γ|κ|) = (ena )2 = (D.14)
1 − 2

to reduce this to

1 8na [ Γ(2γ|κ|) ]2 2 2
= γ |κ| nr !, (D.15)
N 2 Γ(2γ|κ| + nr )

which is equivalent to

1 Γ(2γ|κ| + nr )
N = . (D.16)
Γ(2γ|κ| + 1) (2na )nr !
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-App

Appendices 185

Appendix E. Normalization for the ∆-Induced Scheme

We have the solutions

u1 (x) = N1 xγ|κ| e−x/2 1 F1 (−nr + 1, 2γ|κ| + 1; x)

(E.1)
u2 (x) = N2 xγ|κ| e−x/2 1 F1 (−nr , 2γ|κ| + 1; x),

where nr = na − γ|κ| is, for u2 (x), a positive number or zero, and, for u1 a
positive number (excluding zero).
To determine N1 /N2 insert these solutions into the second of the equations of
(16.23), where a+ is defined in Eq. (16.29):

d x 
+
a u2 = x + − na u2 = −ena u1 . (E.2)
dx 2

We calculate
du2
x = N2 xγ|κ| e−x/2
dx
 
x d1 F1 (−nr , 2γ|κ| + 1; x)
γ|κ| − F
1 1 (−n r , 2γ|κ| + 1; x) + x .
2 dx
(E.3)

We now use the recursion relation

d1 F1 (−nr , 2γ|κ| + 1; x)
x
dx
= −nr [1 F1 (−nr + 1, 2γ|κ| + 1; x) − 1 F1 (−nr , 2γ|κ| + 1; x)]. (E.4)

Inserting this in Eq. (E.2), and using Eq. (E.3), yields:

 x
N2 xγ|κ| e−x/2 γ|κ| − 1 F1 (−nr , 2γ|κ| + 1; x)
2
−nr [1 F1 (−nr + 1, 2γ|κ| + 1; x) − 1 F1 (−nr , 2γ|κ| + 1; x)]
x  
+ − na 1 F1 (−nr , 2γ|κ| + 1; x)
2
= −(ena )N1 xγ|κ| e−x/2 1 F1 (−nr + 1, 2γ|κ| + 1; x). (E.5)

The (x/2) terms cancel, and the other 1 F1 (−nr , 2γ|κ| + 1; x) terms have a
coefficient γ|κ| + nr − na , which vanishes because of (15.47). We are left with
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-App

186 The Supersymmetric Dirac Equation

N2 nr = N1 (ena ), or

N1 nr nr nr
= = = . (E.6)
N2 ena na − κ
2 2 (na + κ)(na − κ)

The normalization condition is

 2 
uA (x) + u2B (x) dx = 1. (E.7)

Substituting uA (x) and uB (x) from Eq. (16.46) leads to



N 2
(1 + ) n2r x2γ|κ| e−x [1 F1 (−nr + 1, 2γ|κ| + 1; x)]2 dx
4


2γ|κ| −x
+(na − κ) 2
x 2
e [1 F1 (−nr , 2γ|κ| + 1; x)] dx

N 2
+ (1 − ) n2r x2γ|κ| e−x [1 F1 (−nr + 1, 2γ|κ| + 1; x)]2 dx
4


2γ|κ| −x
+(na − κ) 2
x 2
e [1 F1 (−nr , 2γ|κ| + 1; x)] dx . (E.8)

The integrals are calculated similarly to the previous cases:

[ Γ(2γ|κ| + 1) ]2
x2γ|κ| e−x [1 F1 (−nr , 2γ|κ| + 1; x)]2 dx = (nr − 1)!
Γ(nr + 2γ|κ|)

[ Γ(2γ|κ| + 1) ]2
x2γ|κ| e−x [1 F1 (−nr + 1, 2γ|κ| + 1; x)]2 dx = nr !
Γ(nr + 2γ|κ| + 1)
(E.9)

so

N 2 2 [ Γ(2γ|κ| + 1) ]2 [ Γ(2γ|κ| + 1) ]2
nr (nr − 1)! + (na − κ)2 nr !
2 Γ(nr + 2γ|κ|) Γ(nr + 2γ|κ| + 1)

N 2 [ Γ(2γ|κ| + 1) ]2 nr !  
= nr (nr + 2γ|κ|) + (na − κ)2 = 1. (E.10)
2Γ(nr + 2γ|κ| + 1)

Now use

nr (nr + 2γ|κ|) + (na − κ)2 = 2na (na − κ). (E.11)

August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-App

Appendices 187

This yields
N 2 [ Γ(2γ|κ| + 1) ]2 nr !
1= 2na (na − κ),
2 Γ(nr + 2γ|κ| + 1)

1 Γ(nr + 2γ|κ| + 1)
N 2 = , (E.12)
[ Γ(2γ|κ| + 1) ]2 na (na − κ)nr !
 1/2
 1 Γ(nr + 2γ|κ| + 1)
N = .
Γ(2γ|κ| + 1) na (na − κ)nr !
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-App

This page intentionally left blank

August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-solution

Solutions to the Exercises

Exercise 3.1 Prove
ijk
imn = δjm δkn − δjn δkm .

Solution: We must have i
= j
= k if we are to have a non-vanishing contribution.

Suppose i = 1, then we can have j = 2, k = 3 or j = 3, k = 2. Also m = 2, n = 3
or m = 3, n = 2. If k = n

123
123 = δ22 δ33 − δ23 δ32 = 1.

If k
= n

123
132 = δ23 δ32 − δ22 δ33 = −1.

We proceed similarly for other permutations of the indices.

Exercise 3.2 Prove
mij
nij = 2δmn .

Solution: If m
= n we get 0. If m = n = 1 then

1ij
1ij =
123
123 +
132
132 = 2.

We proceed similarly for other permutations of the indices.

Exercise 8.1 Prove the identities

(a) (p × L)·(p × L) = p2 L2 ,
(b) (p × L)·p = 2ip2 ,
(c) p·(p × L) = 0,
(d) (p × L)·r = L2 + 2i(p·r),
(e) r·(p × L) = L2 .

189
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-solution

190 The Supersymmetric Dirac Equation

Solution:

(a)
ijk
imn pj Lk pm Ln =
ijk
imn (pj pm Lk Ln + pj [Lk , pm ]Ln )

= (δjm δkn − δjn δkm )(pj pm Lk Ln + i
kmq pj pq Ln )
= p2 L2 − pj (pm Lm )Lj + i
njq pj pq Ln = p2 L2 .
(b)
ijk pj Lk pi =
ijk (pj pi Lk + pj [Lk , pi ] = i
ijk
kiq pj pq = 2ip2 .
(c)
ijk pi pj Lk = 0.
(d)
ijk pj Lk xi =
ijk pj (xi Lk + i
kiq xq ) = L2 + 2ipi xi .
(e)
ijk xi pj Lk = L2 .

Exercise 8.2 Show that C1 and C2 are Casimir operators:

[C1 , L] = [C1 , A] = [C2 , L] = [C2 , A] = 0.

Solution:

[C1 , Li ] = [L2 , Li ] + [A2 , Ai ] = Aj [Aj , Li ] + [Aj , Li ]Aj

= −i
ijk (Aj Ak + Ak Aj ) = 0.
[C1 , Ai ] = [L2 , Ai ] + [A2 , Ai ]
= Lj [Lj , Ai ] + [Lj , Ai ]Lj + Aj [Aj , Ai ] + [Aj , Ai ]Aj
= −i
ijk (Lj Ak + Ak Lj + Aj Lk + Lk Aj ) = 0.
1
[C2 , Li ] = (Lj [Aj , Li ] + [Lj , Li ]Aj + Aj [Lj , Li ] + [Aj , Li ]Lj )
2
= −(i/2)
ijk (Lj Ak + Lk Aj + Aj Lk + Ak Lj ) = 0.
1
[C2 , Ai ] = ([L·A, Ai ] + [A·L, Ai ])
2
1
= (Lj [Aj , Ai ] + [Lj , Ai ]Aj + Aj [Lj , Ai ] + [Ai , Aj ]Lj )
2
= −(i/2)
ijk (Lj Lk + Ak Aj + Aj Ak + Ak Lj ) = 0.

Exercise 9.1 Prove that

i ∂
pr = − r
r ∂r
is canonically conjugate to r.
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-solution

Solutions to the Exercises 191

Solution:


∂ 1∂ 2
[r, pr ] = rpr − pr r = (−i) r− r = −i(1 − 2) = i.
∂r r∂

Exercise 9.2 Prove that the quantized version of pr follows from the Weyl–Moyal
quantization prescription applied to p and r̂.
Solution:



1 r r  −i ∂ ∇·r r·r ∂ 1
·p + p· = 2 + − 3 = −i + = pr .
2 r r 2 ∂r r r ∂r r

Exercise 9.3 Prove that pr is Hermitian.

Solution: We must prove

[ψ∗ (pr ψ) − (pr ψ)∗ ψ]d3 r = 0.

We have
1 ∂
pr ψ = (−i) (rψ),
r ∂r
and
1 ∂
(pr ψ)∗ = (i) (rψ)∗ .
r ∂r
Therefore

[ψ ∗ (pr ψ) − (pr ψ)∗ ψ]d3 r
  
∗ ∂ ∂ ∗
= −i d cos θ dφ (rψ) (rψ) + (rψ) (rψ)
∂r ∂r
 
∂ ∂
= −i d cos θ dφ [(rψ)(rψ)∗ ] dr = −i d cos θ dφ |rψ|2 dr.
∂r ∂r
For limr→∞ (rψ) = 0 and limr→0 (rψ) = 0 the condition for Hermiticity is
satisfied.
Exercise 14.1 Prove [Σ·r̂, H] = − 2i
r βKγ5 .

Solution:
r  ri
i
[Σ·r̂, H] = [Σ·r̂, α·p] = Σi , αj pj + [Σi , αj pj ] .
r r
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-solution

192 The Supersymmetric Dirac Equation

Calculate the individual terms:

r   
i 1 1
Σi , αj pj = Σi [ri , αj pj ] + Σi , αj pj ri
r r r
 
Σi αj 1 i α·r
= [ri , pj ] + Σi αj , pj rj = Σ·α − iΣ·r 3
r r r r
3i i 2i
= γ5 − γ5 = γ5 .
r r r
Also



ri 1 1
[Σi , αj pj ] = 2
ijk αk pj ri = 2i
ijk αk (ri pj − iδij )
r r r


1 2i
= 2iαk Lk = γ5 (Σ·L).
r r

We group these terms together and find

2i 2i 2i 2i
[Σ·r̂, H] = γ5 + γ5 (Σ·L) = γ5 (Σ·L + 1) = − γ5 βK.
r r r r

Exercise 14.2 Prove [KΣ · p, H] = −iK(Σ · r̂)V  (r).

Solution:

[K(Σ·p), H] = K[Σ·p, H] + [K, H](Σ·p) = K[Σ·p, α·p + βm + V (r)]

= K[Σ·p, α·p] + mK[Σ·p, β] + K[Σ·p, V (r)].

The individual commutators are

[Σ·p, α·p] = γ5 [Σi , Σj ]pi pj = 2i
ijk γ5 Σk pi pj = 0,

[Σ·p, β] = 0,
ri 
[Σ·p, V (r)] = Σi [pi , V (r)] = −iΣi V (r) = −i(Σ· r̂)V  (r).
r

Exercise 14.3 Prove

[Kγ5 f (r), H] = iK(Σ· r̂)f  (r) + 2mKγ5 βf (r).

August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-solution

Solutions to the Exercises 193

Solution:

[Kγ5 f (r), H]
= K[γ5 f (r), H] + [K, H]γ5 f (r) = Kγ5 [f (r), H] + K[γ5 , H]f (r)
= Kγ5 [f (r), α·p] + 2mKγ5 βf (r) = iKγ5 α· r̂f  (r) + 2mKγ5 βf (r)
= iK(Σ· r̂)f  (r) + 2mKγ5 βf (r).

Exercise 14.4 Prove K 2 (H − βm)2 = K 2 (H 2 − m2 + 2mβ Zα

r ).

Solution:
K 2 (H − βm)2 = K 2 [H 2 + m2 − m(Hβ + βH)].

We have
Zα
Hβ = m + γ5 βΣ·p − β ,
r
Zα
βH = m + βγ5 Σ·p − β ,
r
Zα
Hβ + βH = 2m − 2β .
r
Adding these together yields the desired relationship.
Exercise 14.5 Prove
−2iβK 2
(Σ·r̂)Kγ5 (H − βm) − (H − βm)γ5 K(Σ·r̂) = .
r

Solution:
(Σ·r̂)Kγ5 (H − βm) − (H − βm)γ5 K(Σ·r̂)
= Kγ5 Σ·r̂(H − βm) − γ5 K(H − βm)Σ·r̂
K −2iβK 2
= Kγ5 [Σ·r̂, H] = Kγ5 (−2iγ5 )β = .
r r
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-solution

This page intentionally left blank

August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-Bib

Bibliography

Books
[Ada94] B. G. Adams. Algebraic Approach to Simple Quantum Systems. Berlin: Springer-
Verlag, 1994.
[BD64] J. D. Bjorken and S. D. Drell. Relativistic Quantum Mechanics. New York:
McGraw-Hill, 1964.
[BL81] L. C. Biedenham and J. D. Louck. Angular Momentum in Quantum Physics:
Theory and Applications. Reading, MA: Encyclopedia of Mathematics and its
Applications 8, Addison-Wesley, 1981.
[Bre91] D. Bressoud. Second Year Calculus. New York: Springer, 1991.
[Bro24] L. de Broglie. Licht und Materie. Hamburg: H. Goverts Verlag, 1930.
[BS57] H. A. Bethe and E. E. Salpeter. Quantum Mechanics of One- and Two-Electron
Atoms. Berlin: Springer-Verlag, 1957.
[Dir53] Paul A. M. Dirac. The Principles of Quantum Mechanics. 3rd ed. Oxford:
Clarendon Press, 1953.
[DL03] C. Doran and A. Lasenby. Geometric Algebra for Physicists. Cambridge:
Cambridge University Press, 2003.
[ER74] R. Eisberg and R. Resnick. Quantum Physics of Atoms, Molecules, Solids, Nuclei,
and Particles. New York: J. Wiley, 1974.
[GR65] I. S. Gradshteyn and I. M. Ryzhik. Tables of Integrals, Series, and Products. New
York: Academic Press, 1965.
[GS90] V. Guillemin and S. Sternberg. Variations on a Theme by Kepler. Providence, R.I.:
The American Mathematical Society, 1990.
[Jac98] J. D. Jackson. Classical Electrodynamics. 3rd ed. New York: J. Wiley, 1998.
[Kel83] A. Keller. The Infancy of Atomic Physics: Hercules in his Cradle. Oxford:
Clarendon Press, 1983.
[LL77] L. D. Landau and E. M. Lipshitz. Quantum Mechanics. 3rd ed. Oxford: Pergamon
Press, 1977.
[Mer61] E. Mertzbacher. Quantum Mechanics. New York: J. Wiley, 1961.
[MR99] J. E. Marsden and T. S. Ratiu. Introduction to Mechanics and Symmetry. 2nd ed.
New York: Springer-Verlag, 1999.
[PGS01] C. Poole, H. Goldstein and J. Safko. Classical Mechanics. New York: J. Wiley,
1998.
[Rig03] J. S. Rigden. Hydrogen: The Essential Element. Cambridge, MA: Harvard
University Press, 2003.
[Ros61] E. M. Rose. Relativistic Electron Theory. New York: J. Wiley, 1961.
[Sak67] J. J. Sakurai. Advanced Quantum Mechanics. Reading, MA: Addison-Wesley,
1967.

195
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-Bib

196 The Supersymmetric Dirac Equation

[SDB04] W. Siegel, S. Duplin and J. Bagger. Concise Encyclopedia of Supersymmetry.

Dordrecht: Kluver Academic Publishing, 2004.
[Sea80] J. B. Seaborn. Hypergeometric Functions and their Applications. Berlin: Springer-
Verlag, 1980.
[Som24] A. Sommerfeld. Atombau und Spektrallinien. Braunschweig: Friedrich Vieweg
Verlag, 1924.
[Sud86] A. Sudbery. Quantum Mechanics and the Particles of Nature. Cambridge:
Cambridge University Press, 1986.
[Tak08] L. A. Takhtayan. Quantum Mechanics for Mathematicians. Providence, R.I.:
American Mathematical Society, 2008.
[WS10] O. L. Weaver and D. H. Sattinger. Lie Groups and Algebras with Applications to
Physics, Geometry, and Mechanics. New York: Springer-Verlag, 2010.

Journals
[AB78] P. R. Auvil and L. M. Brown. “The Relativistic Hydrogen Atom: A Simple
Solution”. In: Am. J. Phys. 46 (1978), p. 679.
[BB80] L. Basano and A. Bianchi. “Rutherford’s scattering formula via the Runge–Lenz
vector”. In: Am. J. Phys. 48 (1980), p. 400.
[BHJ26] M. Born, W. Heisenberg and P. Jordan. “Zur Quantenmechanik II”. In: Z. Phys.
35 (1926), p. 557.
[Bie62] L. C. Biedenharn. “Remarks on the Relativistic Kepler Problem”. In: Phys. Rev.
126 (1962), p. 845.
[Bie83] L. C. Biedenharn. “The ‘Sommerfeld Puzzle’ Revisited and Resolved”. In:
Foundations of Physics. 13 (1983), p. 13.
[BJ25] M. Born and P. Jordan. “Zur Quantenmechanik”. In: Z. Phys. 34 (1925), p. 858.
[BJ53] W. L. Bade and H. Jehle. “An Introduction to Spinors”. In: Rev. Mod. Phys. 25
(1953), p. 714.
[Boh13] N. Bohr. “On the Quantum Theory of Line Spectra”. In: Phil. Mag. 26 (1913),
p. 476.
[Dah95] J. P. Dahl. “On the Origin of the Runge–Lenz Vector”. In: Int. J. of Quant. Chem.
53 (1995), p. 161.
[Dar28] C. G. Darwin. “The Wave Equations of the Electron”. In: Proc. R. Soc. London.
Ser. A118 (1928), p. 654.
[Dir26] P. A. M. Dirac. “Quantum Mechanics and a Preliminary Investigation of the
Hydrogen Atom”. In: Proc. Roy. Soc. A110 (1926), p. 561.
[Dir28] P. A. M. Dirac. “The Quantum Theory of the Electron”. In: Proc. Roy. Soc. A118
(1928), p. 610.
[DJ95] J. P. Dahl and T. Jørgensen. “On the Dirac–Kepler Problem: The
Johnson–Lippmann Operator, Supersymmetry, and Normal-Mode Repre-
sentations”. In: Int. J. of Quant. Chem. 53 (1995), p. 161.
[FK05] M. R. Francis and A. Kosowsky. “The Construction of Spinors in Geometric
Algebra”. In: Ann. Phys. 317 (2005), p. 383.
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-Bib

Bibliography 197

[GI97] B. Goodman and S. R. Ignjatovi¢. “A Simpler Solution of the Dirac Euation in a

Coulomb Potential”. In: Am. J. Phys. 65 (1997), p. 214.
[Gol75] H. Goldstein. “The Prehistory of the Runge–Lenz Vector”. In: Am. J. Phys. 43
(1975), p. 735.
[Gol76] H. Goldstein. “More on the Prehistory of the Runge–Lenz Vector”. In: Am. J.
Phys. 44 (1976), p. 1123.
[Gor28] W. Gordon. “Energy Levels of the Hydrogen Atom in the Dirac Quantum Theory
of Electrons”. In: Z. Phys. 48 (1928), p. 11.
[Hei25] W. Heisenberg. “Ueber quantentheoretische Umdeutung kinematischer und
mechanischer Beziehungen”. In: Z. Phys. 33 (1925), p. 879.
[HI51] T. E. Hull and L. Infeld. “The Factorization Method”. In: Rev. Mod. Phys. 23
(1951), p. 21.
[Inf41] L. Infeld. “On a New Treatment of Some Eigenvalue Problems”. In: Phys. Rev.
59 (1941), p. 737.
[JL50] M. H. Johnson and B. A. Lippmann. “Relativistic Kepler Problem”. In: Phys. Rev.
78 (1950), p. 329.
[JS86] P. D. Jarvis and G. E. Stedman. “Supersymmetry in Second-Order Relativistic
Equations for the Hydrogen Atom”. In: J. Phys. A: Math. Gen. 19 (1986), p. 1373.
[KCS95] A. Khare, F. Cooper and U. P. Sukhatma. “Supersymmetry and Quantum
Mechanics”. In: Phys. Rep. 251 (1995), p. 267.
[KK06] T. T. Khachidze and A. A Khelashvili. “The Hidden Symmetry of the Coulomb
Problem in Relativistic Quantum Mechanics: From Pauli to Dirac”. In: Am. J. Phys.
74 (2006), p. 628.
[Kol66] M. Kolsrad. “On the Solution of Dirac’s Equation with Coulomb Potential”. In:
Phys. Noregica. 2 (1966), p. 43.
[LL74] J.-M. Lévy-Leblond. “The Pedagogical Role and Epistomological Significance of
Group Theory in Quantum Mechanics”. In: Rivista del Nuovo Cimento 4 (1974), p. 99.
[LR47] W. E. Lamb and R. C. Retherford. “Fine Structure of the Hydrogen Atom by a
Microwave Method”. In: Phys. Rev. 72 (1947), p. 241.
[MG58] P. C. Martin and R. F. Glauber. “Relativistic Thory of Radiative Orbital Electron
Capture”. In: Phys. Rev. 109 (1958), p. 1307.
[NKT85] M. Nieto, V. A. Kostelecky and D. Truax. “Supersymmetry and the Relationship
between the Coulomb and the Oscillator Problems in Arbitrary Dimensions”. In: Phys.
Rev. D32 (1985), p. 2627.
[Pau26] W. Pauli. “Über das Wasserstoffatom vom Standpunkt der Neuen Quantenmecha-
nik”. In: Z. Phys. 36 (1926), p. 336.
[Pla00] M. Planck. “Zur Theorie des Gesetzes der Energieverteilung im Normalspektrum”.
In: Verhandlungen der Deutschen Physikalischen Gesellschaft. (1900), p. 237.
[RLB90] P. K. Roy, A. Lahiri and B. Bagchi. “Supersymmetry in Quantum Mechanics”.
In: Int. J. Mod. Phys. A5 (1990), pp. 1383–1456.
[Rut11] E. Rutherford. “The Scattering of Alpha and Beta Particles on Matter and the
Structure of the Atom”. In: Phil. Mag. 21 (1911), p. 669.
[SB89] A. Stahlhofen and K. Bleuler. “An Algebraic Form of the Factorization Method”.
In: Nuovo Cim. 104B (1989), p. 447.
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-Bib

198 The Supersymmetric Dirac Equation

[Sch26] E. Schrödinger. “Über das Wasserstoffatom vom Standpunkt der Neuen

Quantenmechanik”. In: Ann. d. Phys. 75 (1926), p. 361.
[SD91] R. A. Swainson and G. W. F. Drake. “A Unified Treatment of the Non-Relativistic
and Relativistic Hydrogen Atom: I. The Wavefunctions”. In: J. Phys. A: Math. Gen.
24 (1991), p. 79.
[Som16] A. Sommerfeld. “Zur Quantentheorie der Spektrallinien”. In: Ann. d. Phys. 51
(1916), p. 1 and p. 125.
[Sta98] A. A. Stahlhofen. “Comment on ‘A simpler solution of the Dirac equation in a
Coulomb potential’”. In: Am. J. Phys. 66 (1998), p. 636.
[Suk85a] C. V. Sukumar. “Supersymmetry, factorisation of the Schrödinger equation and
a Hamiltonian hierarchy”. In: J. Phys. A: Math. Gen. 18 (1985), p. L57.
[Suk85b] C. V. Sukumar. “Supersymmetry and the Dirac equation for a central Coulomb
field”. J. Phys. A: Math. Gen. 18 (1985), p. L697.
[TT93] D. Tangerman and J. A. Tjon. “Exact Supersymmetry of the Non-Relativistic
Hydrogen Problem”. In: Phys. Rev. A48 (1993), p. 1089.
[Wal79] S. Waldenstrøm. “On the Dirac Equation or the Hydrogen Atom”. In: Am. J. Phys.
47 (1979), p. 1098.
[Wil15] W. Wilson. “The Quantum-theory of radiation and Line-spectra”. In: Phil. Mag.
29 (1915), p. 795.
[Wit81] E. Witten. “A new proof of the positive energy Theorem”. In: Commun. Math.
Phys. 80 (1981), p. 381.
[Wit90] E. Witten. “Spontaneous Breaking of Supersymmetry in Quantum Mechanics”.
In: Int. J. Mod. Phys. A5 (1990), pp. 1383–1456.
[Won86] M. F. K. Wong. “Infeld–Hull Factorization and the Simplified Form of the Dirac–
Coulomb Equation”. In: Phys. Rev. 34 (1986), p. 1559.
[WY82] M. F. K. Wong and H.-Y. Yeh. “Simplified Solution of the Dirac Equation with
a Coulomb Potential”. In: Phys. Rev. D25 (1982), p. 3396.
[WZ74] J. Wess and B. Zumino. “Supergauge Transformations in Four Dimensions”. In:
Nuclear Phys. B70 (1974), p. 39.
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-Index

Index

accidental symmetry, 3 154, 155, 157, 159, 161, 162, 167–169,

apparent principal quantum number, 113, 177
162 Dirac equation, 3, 106, 127, 149
associated Laguerre functions, 74 Dirac operator, 77, 108
Dirac spinors, 111, 122, 123, 128, 151
Balmer, 35, 36, 42, 59 Dirac theory, 105
Balmer series, 42
Biedenharn, 3, 58, 77, 81, 90, 99, 104, eccentricity, 2, 3, 6, 9–11, 13, 54, 55, 57,
115, 126, 127 67, 68, 85, 88, 101, 102, 121, 125, 167,
Bohr, 2, 31, 35–43, 45, 47, 49, 51, 52, 168
55–57, 59, 66, 67, 69, 77, 80, 94, 99, elementary particle, 1, 3, 44, 91
125, 161, 169 exact supersymmetry, 94, 96, 125
Bohr formula, 77
Bohr Model, 35, 37, 39–43, 45, 47, 57, 59, fine structure, 1–4, 37, 99, 167
99, 125, 169
bremsstrahlung radiation, 36 Geiger, 34
Gordan, 83
canonical commutation relations, 59 Gordon, 139
Casimir operator, 3, 66, 190 Goudsmith and Uhlenbeck, 77
circular orbits, 35, 38, 56, 68, 88, 125 graded algebra, 92
Clifford algebra, 78–80, 90, 105 graded vector space, 91
confluent hypergeometric function, 4, 69, gyromagnetic ratio, 3, 80
72, 88, 135, 146, 155, 157, 168, 171,
179 harmonic oscillator, 47, 77
Coulomb, 1, 3, 31, 35–37, 39, 53, 58, 60, Heisenberg, 59
74, 77, 97, 99, 100, 105, 115, 118, 168, helium, 37, 42, 104, 167
175 Hull and Infeld, 3, 77, 90, 137
Coulomb potential, 3, 31 hydrogenic atoms, 3, 43, 59, 66, 77, 91,
94, 161
Dahl and Jørgensen, 4, 126, 139
Darwin term, 162 Interacting fields, 4
de Broglie, 49–52, 69
degeneracy, 3, 4, 58, 59, 67, 77, 78, 87, 94, Johnson–Lippmann operator, 3, 115,
96, 97, 167, 168 119–121, 126, 128, 154, 168
Dirac, 1–4, 59, 77, 80, 81, 91, 96, 99,
104–112, 114, 115, 118, 119, 121–123, Khachidze and Khelashvili, 115
126–130, 134, 137, 141, 142, 149, 151, Kostelecky, 121

199
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-Index

200 The Supersymmetric Dirac Equation

Kramer, 4, 127, 137, 149, 154, 155, 157, quantization condition, 35, 36, 38, 47, 55,
159, 168 99, 102
Kramer’s equation, 4, 127, 137, 149, 154, quantum electrodynamics, 1, 80
155, 157, 159, 168 quantum Laplace vector, 60
quantum rules, 2, 47
Laguerre polynomials, 73, 74, 171, 174, quantum Runge–Lenz vector, 13
175
Lamb-shift, 4, 169 radial Dirac functions, 129, 130, 142
Laplace, 2, 3, 6, 8, 9, 11, 13, 15, 29, radial functions, 74, 111, 123, 129, 135,
31–34, 59, 60, 65, 66, 82, 88, 115, 116, 136, 140, 143, 146, 151, 161
119, 126, 167, 168 Rutherford, 2, 31, 34, 36
Laplace vector, 3, 6, 59, 115 Rydberg constant, 35, 43
Laplace–Lenz vector, 65
Lévy-Leblond, 77 Schrödinger, 2, 3, 69, 75, 87, 168
Lie Algebra, 3, 15, 18–26, 28, 29, Sommerfeld, 2, 3, 47–49, 52, 53, 55, 57,
65, 167 58, 99, 101, 103, 104, 167, 169
Sommerfeld formula, 99, 103
matrix formalism, 59 Sommerfeld puzzle, 99
minimal substitution, 79 spherical harmonics, 71, 72, 83, 111
spherical polar coordinates, 70, 86, 105,
Nieto, 126 111, 156
normalization, 4, 72–74, 132, 136, 137, spin, 2–4, 58, 67, 68, 77, 78, 80, 88, 90,
147, 169, 175, 179, 182, 183, 185, 186 94, 97, 99, 105, 107, 111, 129, 140,
nucleus, 6, 31, 32, 34, 35, 37, 39, 40, 157, 159, 167, 168
42–44, 51, 52, 69, 167 spinors, 77
stationary orbit, 52
Old Quantum Theory, 2, 99, 104 Stern–Gerlach experiment, 77
SU(2) supersymmetry, 122
parity, 91, 92, 96, 109–111, 119, 120, 122, superalgebra, 92, 93, 97
128, 139, 141 superalgebra s(2), 92
parity operator, 92, 96, 119, 120, 122, 128, supercharges, 92, 95, 96
139, 141 supersymmetric charge, 93, 121
Pauli, 3, 4, 13, 59, 62, 65, 68, 77–80, 82, supersymmetric eigenvalue problem, 146
83, 87, 94, 95, 105, 106, 108, 111, 114, supersymmetric generators, 122
161, 164, 167–169 supersymmetric Hamiltonian, 92, 93, 122,
Pauli formalism, 3, 77, 167 125, 133, 144
Pauli hydrogenic realization, 62, 65 supersymmetric quantum, 93, 94, 154
Pauli matrices, 79, 80, 106, 108, 168 supersymmetric quantum mechanics, 154
Pauli Spinors, 82, 105, 111 supersymmetric structure, 128, 133, 137,
Planck, 4, 35–37, 47–50 139, 144
Planck’s constant, 4, 36, 37, 50 supersymmetry, 1, 3, 4, 77, 84, 91–95, 97,
Pochhammer symbols, 74, 171, 172 98, 115, 116, 121, 122, 125–128, 137,
probability interpretation, 69 139–141, 147, 161, 168
August 27, 2011 9:22 9in x 6in The Supersymmetric Dirac Equation b1217-Index

Index 201

supersymmetry algebra, 121, 122 Thomson, 51

supersymmetry generators, 92 total angular momentum, 4, 81, 82,
symmetry algebra so(4), 62 107–110, 119, 168

theorem on odd operators, 81 Urey, 45

theory of relativity, 1, 99
Thomas precession, 162 virial theorem, 40, 161
Thompson, 31, 34
Thompson atom, 34 zitterbewegung, 162