This action might not be possible to undo. Are you sure you want to continue?

BooksAudiobooksComicsSheet Music### Categories

### Categories

### Categories

### Publishers

Scribd Selects Books

Hand-picked favorites from

our editors

our editors

Scribd Selects Audiobooks

Hand-picked favorites from

our editors

our editors

Scribd Selects Comics

Hand-picked favorites from

our editors

our editors

Scribd Selects Sheet Music

Hand-picked favorites from

our editors

our editors

Top Books

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Audiobooks

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Comics

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Sheet Music

What's trending, bestsellers,

award-winners & more

award-winners & more

P. 1

K Schulten - 2000 - Notes on Quantum Mechanics|Views: 45|Likes: 0

Published by Zheng Li

See more

See less

https://www.scribd.com/doc/69624242/K-Schulten-2000-Notes-on-Quantum-Mechanics

10/20/2011

text

original

- Lagrangian Mechanics
- 1.1 Basics of Variational Calculus
- 1.2 Lagrangian Mechanics
- 1.3 Symmetry Properties in Lagrangian Mechanics
- Quantum Mechanical Path Integral
- 2.1 The Double Slit Experiment
- 2.2 Axioms for Quantum Mechanical Description of Single Parti-
- 2.3 How to Evaluate the Path Integral
- 2.4 Propagator for a Free Particle
- 2.5 Propagator for a Quadratic Lagrangian
- 2.6 Wave Packet Moving in Homogeneous Force Field
- 2.7 Stationary States of the Harmonic Oscillator
- The Schr¨odinger Equation
- 3.1 Derivation of the Schr¨odinger Equation
- 3.2 Boundary Conditions
- 3.3 Particle Flux and Schr¨odinger Equation
- 3.4 Solution of the Free Particle Schr¨odinger Equation
- 3.5 Particle in One-Dimensional Box
- 3.6 Particle in Three-Dimensional Box
- Linear Harmonic Oscillator
- 4.1 Creation and Annihilation Operators
- 4.2 Ground State of the Harmonic Oscillator
- 4.3 Excited States of the Harmonic Oscillator
- 4.4 Propagator for the Harmonic Oscillator
- 4.6 Momentum Representation for the Harmonic Oscillator
- 4.7 Quasi-Classical States of the Harmonic Oscillator
- Theory of Angular Momentum and
- 5.1 Matrix Representation of the group SO(3)
- 5.2 Function space representation of the group SO(3)
- 5.3 Angular Momentum Operators
- 5.4 Angular Momentum Eigenstates
- 5.6 Wigner Rotation Matrices
- 5.8 Generators and Rotation Matrices of SU(2)
- 5.9 Constructing Spin States with Larger Quantum Numbers
- 5.10 Algebraic Properties of Spinor Operators
- 5.12 Mapping of SU(2) onto SO(3)
- 6.1 Clebsch-Gordan Coeﬃcients
- 6.2 Construction of Clebsch-Gordan Coeﬃcients
- 6.3 Explicit Expression for the Clebsch–Gordan Coeﬃcients
- 6.4 Symmetries of the Clebsch-Gordan Coeﬃcients
- 6.5 Example: Spin–Orbital Angular Momentum States
- 6.6 The 3j–Coeﬃcients
- 7.1 Radial Schr¨odinger Equation
- 7.2 Free Particle Described in Spherical Coordinates
- 8.2 Planar Electromagnetic Waves
- 8.3 Hamilton Operator
- 8.4 Electron in a Stationary Homogeneous Magnetic Field
- 8.5 Time-Dependent Perturbation Theory
- 8.6 Perturbations due to Electromagnetic Radiation
- 8.7 One-Photon Absorption and Emission in Atoms
- 8.8 Two-Photon Processes
- Many–Particle Systems
- 9.1 Permutation Symmetry of Bosons and Fermions
- 9.2 Operators of 2nd Quantization
- 9.3 One– and Two–Particle Operators
- 9.4 Independent-Particle Models
- 9.5 Self-Consistent Field Theory
- 9.7 Properties of the SCF Ground State
- 9.8 Mean Field Theory for Macroscopic Systems
- Relativistic Quantum Mechanics
- 10.1 Natural Representation of the Lorentz Group
- 10.2 Scalars, 4–Vectors and Tensors
- 10.3 Relativistic Electrodynamics
- 10.5 Klein–Gordon Equation
- 10.6 Klein–Gordon Equation for Particles in an Electromagnetic
- 10.7 The Dirac Equation
- 10.8 Lorentz Invariance of the Dirac Equation
- 10.9 Solutions of the Free Particle Dirac Equation
- 10.10 Dirac Particles in Electromagnetic Field

K. Schulten

Department of Physics and Beckman Institute University of Illinois at Urbana–Champaign 405 N. Mathews Street, Urbana, IL 61801 USA

(April 18, 2000)

Preface Preface

i

The following notes introduce Quantum Mechanics at an advanced level addressing students of Physics, Mathematics, Chemistry and Electrical Engineering. The aim is to put mathematical concepts and techniques like the path integral, algebraic techniques, Lie algebras and representation theory at the readers disposal. For this purpose we attempt to motivate the various physical and mathematical concepts as well as provide detailed derivations and complete sample calculations. We have made every eﬀort to include in the derivations all assumptions and all mathematical steps implied, avoiding omission of supposedly ‘trivial’ information. Much of the author’s writing eﬀort went into a web of cross references accompanying the mathematical derivations such that the intelligent and diligent reader should be able to follow the text with relative ease, in particular, also when mathematically diﬃcult material is presented. In fact, the author’s driving force has been his desire to pave the reader’s way into territories unchartered previously in most introductory textbooks, since few practitioners feel obliged to ease access to their ﬁeld. Also the author embraced enthusiastically the potential of the TEX typesetting language to enhance the presentation of equations as to make the logical pattern behind the mathematics as transparent as possible. Any suggestion to improve the text in the respects mentioned are most welcome. It is obvious, that even though these notes attempt to serve the reader as much as was possible for the author, the main eﬀort to follow the text and to master the material is left to the reader. The notes start out in Section 1 with a brief review of Classical Mechanics in the Lagrange formulation and build on this to introduce in Section 2 Quantum Mechanics in the closely related path integral formulation. In Section 3 the Schr¨dinger equation is derived and used as an alternative description of continuous quantum o systems. Section 4 is devoted to a detailed presentation of the harmonic oscillator, introducing algebraic techniques and comparing their use with more conventional mathematical procedures. In Section 5 we introduce the presentation theory of the 3-dimensional rotation group and the group SU(2) presenting Lie algebra and Lie group techniques and applying the methods to the theory of angular momentum, of the spin of single particles and of angular momenta and spins of composite systems. In Section 6 we present the theory of many–boson and many–fermion systems in a formulation exploiting the algebra of the associated creation and annihilation operators. Section 7 provides an introduction to Relativistic Quantum Mechanics which builds on the representation theory of the Lorentz group and its complex relative Sl(2, C). This section makes a strong eﬀort to introduce Lorentz–invariant ﬁeld equations systematically, rather than relying mainly on a heuristic amalgam of Classical Special Relativity and Quantum Mechanics. The notes are in a stage of continuing development, various sections, e.g., on the semiclassical approximation, on the Hilbert space structure of Quantum Mechanics, on scattering theory, on perturbation theory, on Stochastic Quantum Mechanics, and on the group theory of elementary particles will be added as well as the existing sections expanded. However, at the present stage the notes, for the topics covered, should be complete enough to serve the reader. The author would like to thank Markus van Almsick and Heichi Chan for help with these notes. The author is also indebted to his department and to his University; their motivated students and their inspiring atmosphere made teaching a worthwhile eﬀort and a great pleasure. These notes were produced entirely on a Macintosh II computer using the TEX typesetting system, Textures, Mathematica and Adobe Illustrator. Klaus Schulten University of Illinois at Urbana–Champaign August 1991

ii

Preface

Contents

1 Lagrangian Mechanics 1.1 Basics of Variational Calculus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Lagrangian Mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3 Symmetry Properties in Lagrangian Mechanics . . . . . . . . . . . . . . . . . . . . . 2 Quantum Mechanical Path Integral 2.1 The Double Slit Experiment . . . . . . . . . . . . . . . 2.2 Axioms for Quantum Mechanical Description of Single 2.3 How to Evaluate the Path Integral . . . . . . . . . . . 2.4 Propagator for a Free Particle . . . . . . . . . . . . . . 2.5 Propagator for a Quadratic Lagrangian . . . . . . . . 2.6 Wave Packet Moving in Homogeneous Force Field . . 2.7 Stationary States of the Harmonic Oscillator . . . . . 3 The 3.1 3.2 3.3 3.4 3.5 3.6 Schr¨dinger Equation o Derivation of the Schr¨dinger Equation o Boundary Conditions . . . . . . . . . . . Particle Flux and Schr¨dinger Equation o Solution of the Free Particle Schr¨dinger o Particle in One-Dimensional Box . . . . Particle in Three-Dimensional Box . . . 1 1 4 7 11 11 11 14 14 22 25 34 51 51 53 55 57 62 69 73 74 77 78 81 83 88 90

. . . . . Particle . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . . . . . . . . . . . . . Equation . . . . . . . . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

4 Linear Harmonic Oscillator 4.1 Creation and Annihilation Operators . . . . . . . . . . . 4.2 Ground State of the Harmonic Oscillator . . . . . . . . . 4.3 Excited States of the Harmonic Oscillator . . . . . . . . 4.4 Propagator for the Harmonic Oscillator . . . . . . . . . 4.5 Working with Ladder Operators . . . . . . . . . . . . . . 4.6 Momentum Representation for the Harmonic Oscillator 4.7 Quasi-Classical States of the Harmonic Oscillator . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

5 Theory of Angular Momentum and Spin 97 5.1 Matrix Representation of the group SO(3) . . . . . . . . . . . . . . . . . . . . . . . . 97 5.2 Function space representation of the group SO(3) . . . . . . . . . . . . . . . . . . . . 104 5.3 Angular Momentum Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106

iii

iv 5.4 5.5 5.6 5.7 5.8 5.9 5.10 5.11 5.12

Contents Angular Momentum Eigenstates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Irreducible Representations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Wigner Rotation Matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Spin 1 and the group SU(2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Generators and Rotation Matrices of SU(2) . . . . . . . . . . . . . . . . . . . . . . . Constructing Spin States with Larger Quantum Numbers Through Spinor Operators Algebraic Properties of Spinor Operators . . . . . . . . . . . . . . . . . . . . . . . . Evaluation of the Elements dj m (β) of the Wigner Rotation Matrix . . . . . . . . . m Mapping of SU(2) onto SO(3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Spin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 120 123 125 128 129 131 138 139 141 145 147 151 160 163 172 176 179

6 Quantum Mechanical Addition of Angular Momenta and 6.1 Clebsch-Gordan Coeﬃcients . . . . . . . . . . . . . . . . . . 6.2 Construction of Clebsch-Gordan Coeﬃcients . . . . . . . . . 6.3 Explicit Expression for the Clebsch–Gordan Coeﬃcients . . 6.4 Symmetries of the Clebsch-Gordan Coeﬃcients . . . . . . . 6.5 Example: Spin–Orbital Angular Momentum States . . . . 6.6 The 3j–Coeﬃcients . . . . . . . . . . . . . . . . . . . . . . . 6.7 Tensor Operators and Wigner-Eckart Theorem . . . . . . . 6.8 Wigner-Eckart Theorem . . . . . . . . . . . . . . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

7 Motion in Spherically Symmetric Potentials 183 7.1 Radial Schr¨dinger Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184 o 7.2 Free Particle Described in Spherical Coordinates . . . . . . . . . . . . . . . . . . . . 188 8 Interaction of Charged Particles with Electromagnetic Radiation 8.1 Description of the Classical Electromagnetic Field / Separation of Longitudinal and Transverse Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2 Planar Electromagnetic Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3 Hamilton Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4 Electron in a Stationary Homogeneous Magnetic Field . . . . . . . . . . . . . . . . 8.5 Time-Dependent Perturbation Theory . . . . . . . . . . . . . . . . . . . . . . . . . 8.6 Perturbations due to Electromagnetic Radiation . . . . . . . . . . . . . . . . . . . 8.7 One-Photon Absorption and Emission in Atoms . . . . . . . . . . . . . . . . . . . . 8.8 Two-Photon Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Many–Particle Systems 9.1 Permutation Symmetry of Bosons and Fermions . 9.2 Operators of 2nd Quantization . . . . . . . . . . 9.3 One– and Two–Particle Operators . . . . . . . . 9.4 Independent-Particle Models . . . . . . . . . . . 9.5 Self-Consistent Field Theory . . . . . . . . . . . . 9.6 Self-Consistent Field Algorithm . . . . . . . . . . 9.7 Properties of the SCF Ground State . . . . . . . 9.8 Mean Field Theory for Macroscopic Systems . . 203 . . . . . . . . 203 206 208 210 215 220 225 230

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

239 . 239 . 244 . 250 . 257 . 264 . 267 . 270 . 272

Contents 10 Relativistic Quantum Mechanics 10.1 Natural Representation of the Lorentz Group . . . . . . . . . . . 10.2 Scalars, 4–Vectors and Tensors . . . . . . . . . . . . . . . . . . . 10.3 Relativistic Electrodynamics . . . . . . . . . . . . . . . . . . . . . 10.4 Function Space Representation of Lorentz Group . . . . . . . . . 10.5 Klein–Gordon Equation . . . . . . . . . . . . . . . . . . . . . . . 10.6 Klein–Gordon Equation for Particles in an Electromagnetic Field 10.7 The Dirac Equation . . . . . . . . . . . . . . . . . . . . . . . . . 10.8 Lorentz Invariance of the Dirac Equation . . . . . . . . . . . . . 10.9 Solutions of the Free Particle Dirac Equation . . . . . . . . . . . 10.10Dirac Particles in Electromagnetic Field . . . . . . . . . . . . . . 11 Spinor Formulation of Relativistic Quantum Mechanics 11.1 The Lorentz Transformation of the Dirac Bispinor . . . . . 11.2 Relationship Between the Lie Groups SL(2,C) and SO(3,1) 11.3 Spinors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.4 Spinor Tensors . . . . . . . . . . . . . . . . . . . . . . . . . 11.5 Lorentz Invariant Field Equations in Spinor Form . . . . . .

v 285 286 294 297 300 304 307 312 317 322 333 351 351 354 359 363 369

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

12 Symmetries in Physics: Isospin and the Eightfold Way 371 12.1 Symmetry and Degeneracies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371 12.2 Isospin and the SU (2) ﬂavor symmetry . . . . . . . . . . . . . . . . . . . . . . . . . 375 12.3 The Eightfold Way and the ﬂavor SU (3) symmetry . . . . . . . . . . . . . . . . . . 380

vi

Contents

there are 3N conﬁgurational coordinates.. .1) from a space F of vector-valued functions q(t) onto the real numbers. qM (t)).4) d δq(t)| < . | a functional δS[ · .2) . δq2 (t)] . For this purpose we will review the relevant concepts of Classical Mechanics. namely. Deﬁnition: A functional S[ ] is diﬀerentiable. ∀t. i. 1. δS[ · . t1 ] ⊂ R → RM . Deﬁnition: A functional S[ ] is a map S[ ] : F → R .3) (ii) δS[q(t). It is important to note at the outset that for the description of a d classical system it will be necessary to provide information q(t) as well as dt q(t).1 Basics of Variational Calculus The derivation of the Principle of Least Action requires the tools of the calculus of variation which we will provide now. δq(t)] + O( 2 ) (1. . q(t) diﬀerentiable} (1. that along a path describing classical motion the action integral assumes a minimal value (Hamiltonian Principle of Least Action). In case of N classical particles holds M = 3N . δq(t)] is linear in δq(t). . · ] is called the diﬀerential of S[ ]. often in the Cartesian coordinate system. An important concept is that the equations of motion of Classical Mechanics can be based on a variational principle. |δq(t)| < . δq(t) ∈ F. if for any q(t) ∈ F and δq(t) ∈ F where F = {δq(t).Chapter 1 Lagrangian Mechanics Our introduction to Quantum Mechanics will be based on its correspondence to Classical Mechanics. The latter is the velocity vector of the system. namely. q : [t0 . t1 ] ⊂ R} dt (1. q2 (t). α1 δq1 (t) + α2 δq2 (t)] = α1 δS[q(t). The linearity property above implies δS[q(t).e. q(t) is called the trajectory of a system of M degrees of freedom described by the conﬁgurational coordinates q(t) = (q1 (t). · ] exists with the properties (i) S[q(t) + δq(t)] = S[q(t)] + δS[q(t). 1 (1. δq1 (t)] + α2 δS[q(t). t ∈ [t0 . the position coordinates of the particles in any kind of coordianate system. F = {q(t).

For this purpose we consider M ˙ ˙ ˙ L(q(t) + δq(t). We will later often assume that only variations of a trajectory q(t) are permitted for which δq(t0 ) = 0 and δq(t1 ) = 0 holds. but rather to a diﬀerential of the function which is simply the df diﬀerentiated function multiplied by the diﬀerential increment of the variable. q(t). j=1 We will now consider a particular class of functionals S[ ] which are expressed through an integral d over the the interval [t0 . . · ] in conventional diﬀerential calculus does not correspond to a diﬀerentiated function. e. i. t) + j=1 ∂L ∂L δqj + δ qj ˙ ∂qj ∂ qj ˙ + O( 2 ) (1. t0 (1. · ) is a function diﬀerentiable in its three arguments.6) For a proof we can use conventional diﬀerential calculus since the functional (1. dt q(t). t) (1. df = dx dx or. df = M ∂xj dxj .6) immediately.7) through Taylor expansion and identiﬁcation of terms linear in δqj (t). t1 ] where the integrand is a function L(q(t).2 Lagrangian Mechanics Note: δq(t) describes small variations around the trajectory q(t). t) of the conﬁguration d vector q(t). ∂f in case of a function of M variables.g. It holds M M ∂L t1 d ∂L ∂L δS[q(t). · . q(t). δq(t)] = dt − δqj (t) + δqj (t) ∂qj dt ∂ qj ˙ ∂ qj ˙ t0 j=1 j=1 t1 . q(t) + δ q(t).8) We note then using d dt f (t)g(t) ˙ = f˙(t)g(t) + f (t)g(t) ∂L δqj ∂ qj ˙ − d ∂L dt ∂ qj ˙ δqj . q(t) + δ q(t). We focus on such functionals because they play a central role in the so-called action integrals of Classical Mechanics. at the ends of the time interval of the trajectories the variations vanish. It is also important to appreciate that δS[ · . t) = L(q(t).9) d ∂L δ qj = ˙ ∂ qj ˙ dt This yields for S[q(t) + δq(t)] t1 M S[q(t)] + t0 dt j=1 d ∂L − ∂qj dt ∂L ∂ qj ˙ t1 M δqj + t0 dt j=1 d dt ∂L δqj ∂ qj ˙ + O( 2 ) (1. i. t) (1..e.6) is expressed in terms of ‘normal’ functions. equating these terms with δS[q(t). the velocity vector dt q(t) and time t. ˙ In the following we will often use the notation for velocities and other time derivatives d q(t) = q(t) and dtj = xj . We attempt to evaluate t1 S[q(t) + δq(t)] = t0 ˙ ˙ dt L(q(t) + δq(t). ˙ Theorem: Let S[q(t)] = t0 dx dt t1 ˙ dt L(q(t). q(t) + δq(t) is a ‘slightly’ diﬀerent trajectory than q(t)..e. (1. δq(t)].10) From this follows (1.5) where L( · .

. According to the Lemma above follows then (1. δq(t)] = 0 for all δq(t) ∈ F .12) The proof of this theorem is based on the property Lemma: If for a continuous function f(t) f : [t0 . . y(x1 ) = y1 .16) This property holds for any δqj with δq(t) ∈ F . Let us assume that the two points are connected by the path y(x). For this purpose we deﬁne Deﬁnition: An extremal of a diﬀerentiable functional S[ ] is a function qe (t) with the property δS[qe (t).1.6) which reads in the present case M ∂L t1 d ∂L δS[q(t). The length of such path can be determined starting from the fact that the incremental length ds in going from point (x. y1 ) and (x2 . On the other side. (1. An Example As an application of the above rules of the variational calculus we like to prove the well-known result that a straight line in R is the shortest connection (geodesics) between two points (x1 . This condition is stated in the following theorem. if and only if it satisﬁes the conditions (j = 1. dx (1. . y(x2 ) = y2 . . it holds in case δq(t0 ) = δq(t1 ) = 0 that qe (t) is an extremal.12) for j = 1. t1 ] ⊂ R → R holds t1 (1. .1: Variational Calculus 3 We now consider the question for which functions the functionals of the type (1.14) for any continuous function h(t) ∈ F with h(t0 ) = h(t1 ) = 0. The proof of the above theorem starts from (1. y2 ). .17) .12) for j = 1. . ∂qj dt ∂ qj ˙ t0 j=1 (1. . y(x + dx)) is ds = (dx)2 + ( dy dx)2 = dx dx 1+( dy 2 ) .15) We will not provide a proof for this Lemma. from (1. t1 ]. 2.16) the property δS[qe (t). . the above theorem. · ] ≡ 0 and.5). hence. M ) d dt ∂L ∂ qj ˙ − ∂L = 0 ∂qj (1. M . M and δqj (t0 ) = δqj (t1 ) = 0 follows according to (1. 2. . 2. Theorem: Euler–Lagrange Condition For the functional deﬁned through (1.11) The extremals qe (t) can be identiﬁed through a condition which provides a suitable diﬀerential equation for this purpose.13) dt f (t)h(t) = 0 t0 (1. then f (t) ≡ 0 on [t0 . y(x)) to (x + dx. (1. δq(t)] = dt − δqj (t) .5) assume extreme values.

1: Show that the shortest path between two points on a sphere are great circles. j = 1. hence. (1.22) ˙ where L(q(t). 2. . ˙2 2 (1. dx ) = 1 + (dy/dx)2 . . . Velocity–dependent potentials which describe particles moving in electromagnetic ﬁelds will be considered below. 1. . . circles whose centers lie at the center of the sphere.5) with L(y(x). y = const. t) is the so-called Lagrangian M ˙ L(q(t).18) dy s is a functional of y(x) of the type (1.e.4 The total path length is then given by the integral x1 Lagrangian Mechanics s = x0 dx 1+( dy 2 ) . i. .19) From this follows y / 1 + (y )2 = const and. This in turn yields y(x) = ax + b. One obtains y 1 − y2 y(x) = (x − x2 ) + y2 . The shortest path is an extremal of s[y(x)] which must. obey the Euler–Lagrange dy condition. the so-called action integral.21) are extremals of the functional. q(t). (1. . ..1. t) = j=1 1 mj qj − U (q1 . dx (1. qM ) . q(t).20) x1 − x2 Exercise 1. Threorem: Hamiltonian Principle of Least Action The trajectories q(t) of systems of particles described through the Newtonian equations of motion d ∂U (mj qj ) + ˙ = 0 dt ∂qj .23) Presently we consider only velocity–independent potentials. Using y = dx the condition reads d dx ∂L ∂y = d dx y 1 + (y )2 = 0. according to the theorems above. M (1. t1 S[q(t)] = t0 ˙ dt L(q(t). t) (1.2 Lagrangian Mechanics The results of variational calculus derived above allow us now to formulate the Hamiltonian Principle of Least Action of Classical Mechanics and study its equivalence to the Newtonian equations of motion. q(t). The constants a and b are readily identiﬁed through the conditons y(x1 ) = y1 and y(x2 ) = y2 . q2 .

t) as follows B E = ×A 1 ∂A . The transformation which leaves the ﬁelds invariant is A (r.25) of a particle moving in the ﬁeld.29) where ρ(r.27) (1.31) = − V − Gauge Symmetry of the Electromagnetic Field It is well known that the relationship between ﬁelds and potentials (1. t) and B(r. respectively.32) (1.25) = 0 = 0 4π J c = 4πρ = (1.30) (1.2: Lagrangian 5 For a proof of the Hamiltonian Principle of Least Action we inspect the Euler–Lagrange conditions associated with the action integral deﬁned through (1. 1. These conditions read in the present case ∂L d ∂L ∂U d − = 0 → − − (mj qj ) = 0 ˙ (1.1.26) (1. 1. t) . t) describes the charge current density. t) c ∂t (1.24) ∂qj dt ∂ qj ˙ ∂qj dt which are obviously equivalent to the Newtonian equations of motion. x3 (t)) moving in an electromagnetic ﬁeld described through the electrical and magnetic ﬁeld components E(r. t) obey the Maxwell equations ×E + ·B ×B − ·E 1 ∂ c ∂t E 1 ∂ c ∂t B (1. t) − K(r. t) describes the charge density present in the ﬁeld and J(r.23). t) and a vector potential A(r.28) can be satisﬁed implicitly if one represents the ﬁelds through a scalar potential V (r. x2 (t). t). F (r. t) = V (r.33) . dt The ﬁelds E(r. c ∂t (1.30.31) allows one to transform the potentials without aﬀecting the ﬁelds and without aﬀecting the equations of motion (1. t) + v × B(r. t) + V (r. t) = A(r.22. The equations of motion for such a particle are d q ˙ (mr) = F (r. t) and B(r.28) (1. t) = q E(r. Equations (1. t) dt c where dr = v and where F (r.27) and (1. t) 1∂ K(r. Particle Moving in an Electromagnetic Field We will now consider the Newtonian equations of motion for a single particle of charge q with a trajectory r(t) = (x1 (t). t) is the Lorentz force.

Comparing then (1. 2 c (1.40. (1.25) follows from the Hamiltonian Principle of Least Action..36) This expression allows us to show that (1. 1.6 Lagrangian of Particle Moving in Electromagnetic Field Lagrangian Mechanics We want to show now that the equation of motion (1.35) is equivalent to the Euler–Lagrange condition d dt The second term in (1. d ∂V q (mv1 ) = F1 = −q + [v × B]1 .30).35) − ∂A1 ∂x2 [v × B]1 = x2 B3 − x3 B2 = x2 ˙ ˙ ˙ ∂A2 ∂A1 − ∂x1 ∂x2 − x3 ˙ ∂A1 ∂A3 − ∂x3 ∂x1 . 1.41) with (1.34) For this purpose we consider only one component of the equation of motion (1.37) is ∂L ∂V q = −q + ∂x1 ∂x1 c The ﬁrst term in (1. in fact.39) together yield ∂V q d (mx1 ) = −q ˙ + O dt ∂x1 c where O = ∂A1 ∂A2 ∂A3 ∂A1 ∂A1 ∂A1 x1 + ˙ x2 + ˙ x3 − ˙ x1 − ˙ x2 − ˙ x3 ˙ ∂x1 ∂x1 ∂x1 ∂x1 ∂x2 ∂x3 ∂A2 ∂A1 ∂A1 ∂A3 = x2 ˙ − − x3 ˙ − ∂x1 ∂x2 ∂x3 ∂x1 (1. t) = 1 q mv 2 − q V (r. t) + A(r. . r. equivalent.38) ∂L ∂ x1 ˙ − ∂L = 0.37) The results (1.25). namely.35) shows that the Newtonian equations of motion and the Euler–Lagrange conditions are. dt ∂x1 c We notice using (1. t) · v . (1. e.36) in the Newtonian equation of motion.40) (1.g.39) ∂A1 ∂A2 ∂A3 x1 + ˙ x2 + ˙ x3 ˙ ∂x1 ∂x1 ∂x1 . if one assumes for a particle the Lagrangian ˙ L(r. ∂x1 (1. (1.41) which is identical to the term (1. B3 = ∂A2 ∂x1 (1.38.37) is d dt ∂L ∂ x1 ˙ = d q dA1 d q (mx1 ) + ˙ = (mx1 ) + ˙ dt c dt dt c ∂A1 ∂A1 ∂A1 x1 + ˙ x2 + ˙ x3 ˙ ∂x1 ∂x2 ∂x3 .

and since these properties allow one often to simplify mathematical descriptions.45) . t) c t1 t0 t1 t0 ˙ dtL(q. a term not aﬀected by the variations allowed.44) Since the condition δq(t0 ) = δq(t1 ) = 0 holds for the variational functions of Lagrangian Mechanics. in fact. q. i.44) implies that the gauge transformation amounts to adding a constant term to the action integral. The gauge symmetry. We will consider in the following two symmetries. For this purpose we consider the transformation of the single particle Lagrangian (1. 1.32. t) = q K(q. t) + q K(q. t) + dt c (1.34) q 1 q d ˙ L (r.3 Symmetry Properties in Lagrangian Mechanics Symmetry properties play an eminent role in Quantum Mechanics since they reﬂect the properties of the elementary constituents of physical systems. t) . t) · v + K(r. dt ∂x1 ∂x2 ∂x3 ∂t ∂t (1. t) + K ·v (1. t) L (q. q. (1. appear in a diﬀerent. t) = mv 2 − q V (r. t) − + 2 c ∂t c 1 q = mv 2 − q V (r. equivalent to the gauge transformation (1.43) into (1.32. Theorem: Gauge Transformation of Lagrangian The equation of motion (Euler–Lagrange conditions) of a classical mechanical system are unaﬀected by the following transformation of its Lagrangian d q ˙ ˙ K(q. q.43) To prove this theorem we determine the action integral corresponding to the transformed Lagrangian t1 S [q(t)] = t0 ˙ dtL (q. t) c t1 t0 = S[q(t)] + (1.42) This transformation is termed gauge transformation.33) of electromagnetic potentials.. 1. t) 1 1 ∂K q mv 2 − q V (r. surprisingly simple fashion in Lagrangian Mechanics.33) of electyromagnetic potentials. Note that one adds the total time derivative of a function K(r.33) of electromagnetic potentials. They are the subject of the following theorem.46) (1. t) = L(q. t) · v . One can conclude then immediately that any extremal of S [q(t)] is also an extremal of S[q(t)].3: Symmetry Properties 7 1. q. We want to demonstrate now that the transformation (1.45) and reordering terms yields using (1. t) = x1 + ˙ x2 + ˙ x3 + ˙ = ( K) · v + . This term is d ∂K ∂K ∂K ∂K ∂K K(r. t) + A (r.32.32. t) + A(r. gauge symmetry and symmetries with respect to spatial transformations. Eq.1.e. r.33) ˙ L (r. 1. 2 c = A(r. The factor q has been introduced to make this c transformation equivalent to the gauge transformation (1. encountered above in connection with the transformations (1. 1. 2 c c dt Inserting (1.42) is. r. t) the Lagrangian.

ˆ (1. t). the reason being that our interest lies in achieving familiarity with the principles (just mentioned above ) of symmetry properties rather than in providing a general tool in the context of Classical Mechanics. for such vectors holds a · a = 1.52) x3 0 0 1 x3 . The transformations considered are speciﬁed in the following deﬁnition.42) corresponds to replacing in the Lagrangian potentials V (r. We consider now invariance properties connected with coordinate transformations. Such invariance properties are very familiar. but nevertheless instructive way considering rotations around the x3 –axis. which has an important analogy in Quantum Mechanics. i. A(r.42) and (1.33). t) by gauge transformed potentials V (r.48) In the following we will denote unit vectors as a. r ∈ R . for example.47) (1. ) = r + R(r) + O( 2 ) . A (r. R(r) = through ∂r ∂ (1. ) . Deﬁnition: Inﬁnitesimal One-Parameter Coordinate Transformations A one-parameter coordinate transformation is decribed through r = r (r.50) A non-trivial example is furnished by the inﬁnitesimal rotation around axis e ˆ r = r + e×r .. the transformation (1. The corresponding inﬁnitesimal transformation is deﬁned for small r (r. the equivalence of (1. we like to ˆ provide the transformation here anyway for later reference r = r + e. We have proven. In this case the transformation can be written in matrix form cos −sin 0 x1 x1 x2 = sin cos 0 x2 (1. and for the introduction of inﬁnitesimal transformations which will provide a crucial method for the study of symmetry in Quantum Mechanics. therefore. for the connection between invariance properties and constants of motion. In case of a translation transformation in the direction e nothing new is gained. ˆ ˆ ˆ Examples of Inﬁnitesimal Transformations The beauty of inﬁnitesimal transformations is that they can be stated in a very simple manner. in the case of central force ﬁelds which are invariant with respect to rotations of coordinates around the center. t). r. t). where the origin of is chosen such that r (r.8 Lagrangian Mechanics Obviously.51) We would like to derive this transformation in a somewhat complicated.e. The following description of spatial symmetry is important in two respects. 0) = r . The transformations we consider are the most simple kind. ˆ (1. However. 1.32.49) =0 ∈ R (1.

e. Before the embark on this theorem we will comment on what is meant by the statement that a classical mechanical system is invariant under a coordinate transformation. q. however. the property holds for any system. then M ∂L j=1 Qj ∂ xj is a constant of motion. a classical mechanical system is invariant with respect to a coordinate transformation ˙ a constant of motion exists. i. Theorem: Noether’s Theorem ˙ If L(q.3: Symmetry Properties 9 In case of small this transformation can be written neglecting terms O( 2 ) using cos = 1 + O( 2 ). ˙ ∂qk ˙ Inserting these inﬁnitesimal changes of qj and qj into the Lagrangian L(q.. expansion. Using the fact. t) + j=1 ∂L Qj + ∂qj M j. We consider here only the ‘particle version’ of the theorem.1.56) ∂Qj qk . Anytime. ˙ We have generalized in this theorem the deﬁnition of inﬁnitesimal coordinate transformation to M –dimensional vectors q.e. q. a quantity C(r. t) = L(q. .55) (1.57) ∂ d where we used dt Qj = M ( ∂qk Qj )qk . (1.k=1 ∂L ∂Qj qk ˙ ∂ qj ∂qk ˙ (1. We have used here the notation corresponding to single particle motion. that along the classical path holds d ∂L ∂L the Euler-Lagrange condition ∂qj = dt ( ∂ qj ) one can rewrite the sum of the second and third term ˙ in (1. q. In order to prove Noether’s theorem we note qj qj ˙ = qj + = qj + ˙ k=1 Qj (q) M (1.58) From this follows the statement of the theorem.. the second and third term ˙ k=1 in (1.57) M Qj j=1 d dt ∂L ∂ qj ˙ + ∂L d Qj ∂ qj dt ˙ = d dt M Qj j=1 ∂L = 0 ∂ qj ˙ (1. t) is invariant with respect to an inﬁnitesimal transformation q = q + Q(q). The property has been shown to hold in a more general context. by Noether. sin = + O( 2 ) 0 − 0 x1 x1 x1 x2 = x2 + 0 0 x2 + O( 2 ) .51) reads r = r − x2 e1 + x1 e2 ˆ ˆ (1. In the context of Lagrangian Mechanics this implies that such transformation leaves the Lagrangian of the system unchanged.53).57) must cancel each other or both vanish. namely for ﬁelds rather than only for particle motion. r) which is constant along the classical path of the system. q. t) yields after Taylor ˙ 2 ).53) x3 x3 0 0 0 x3 One can readily verify that in case e = e3 (ˆj denoting the unit vector in the direction of the ˆ ˆ e xj –axis) (1. i. Invariance implies L = L. neglecting terms of order O( M ˙ ˙ L (q.54) which is equivalent to (1.

51). 2 We ﬁrst determine the constant of motion in case of invariance with respect to translations as deﬁned in (1. ˆ ˙ ˆ j=1 (1. Noether’s theorem yields ˆ ˆ the constant of motion (qj = xj .50). 3) 3 Qj j=1 ∂L = e· ˆ ∂ xj ˙ 3 ej mxj = e · mv . 3 and.59) yields the constant of motion ˆ e (ˆ × r) · mv. It is not necessary to investigate the consequences of global. 2. v) = 1 mv 2 − U (r). In this case we have Qj = ej · e. except using now Qj = ej · (ˆ × r). ˆ The important result to be remembered for later considerations of symmetry transformations in the context of Quantum Mechanics is that it is suﬃcient to know the consequences of inﬁnitesimal transformations to predict the symmetry properties of Classical Mechanics. for which translational invariance holds. A calculation similar to that in (1. is conserved. of course. . j = 1. It was.59) We obtain the well known result that in this case the momentum in the direction. We will now investigate the consequence of rotational invariance as described according to the inﬁnitesimal transformation (1.59).10 Application of Noether’s Theorem Lagrangian Mechanics We consider brieﬂy two examples of invariances with respect to coordinate transformations for the Lagrangian L(r. to be expected that this is the constant of motion. hence. Using the cyclic property (a × b) · c = (b × c) · a = (c × a) · b allows one to rewrite the e constant of motion e · (r × mv) which can be identiﬁed as the component of the angular momentum ˆ mr × v in the e direction. i. not inﬁnitesimal transformations. In this case we will use the same notation as in (1.e. 2. j = 1.

t1 ] . t ∈ [t0 .Chapter 2 Quantum Mechanical Path Integral 2. The probability that the particle is detected at space–time point r.1) (2. t is |ψ(r. At present one may think of f (r) as a sum over δ–functions which represent a multi–slit screen. 11 . (2. We provide now a set of formal rules which state how the probability to observe such a particle at some space–time point r. t) ψ : R3 ⊗ R → C. r. The wave function ψ(r. 1.2) (2. 3. placed into the space at some particular time and with a detector behind each slit.1 The Double Slit Experiment Will be supplied at later date 2. t is described in Quantum Mechanics. The particle is described by a wave function ψ(r.2 Axioms for Quantum Mechanical Description of Single Particle ˙ Let us consider a particle which is described by a Lagrangian L(r. t) where z is the conjugate complex of z. 4. 2. t)|2 = 1 Ω ∀t. t)|2 Ω (2. t). The probability to detect the particle with a detector of sensitivity f (r) is d3 r f (r) |ψ(r.3) where Ω is the space volume in which the particle can exist. t)ψ(r.4) a condition which enforces that the probability of ﬁnding the particle somewhere in Ω at any particular time t in an interval [t0 . t) is normalized d3 r |ψ(r. t)|2 = ψ(r. t1 ] in which the particle is known to exist. is unity.

9) for N → ∞. t0 ) = Ωd 3r N −1 Ωd 3r N −2 ··· Ωd 3r 1 φ(r. The deﬁnition will actually assume an inﬁnite number of intermediate times and express the path integral through integrals of the type (2. t ) φ(r. The generalization of the completeness property to N − 1 intermediate points t > tN −1 > tN −2 > . The following so-called completeness relationship holds for the propagator (t > t [t0 . t|r0 . This is diﬀerent from the classical situation where the particle follows a discrete path and. This must hold for any ψ(r . t1 ]) d3 r φ(r. t ∈ [t0 . t) = Ω d3 r φ(r. t ) = δ(r − r ) Ω (2. t ) which requires d3 r φ(r. t|r .7) t. t| rN −1 . t0 ) = r(t0 )=r0 d[r(t)] Φ[r(t)] . t|r . t) is described by a linear map of the type ψ(r.. described by ψ(r. (2. > t1 > t0 is φ(r. tN −1 |rN −2 . t0 ) = δ(r − r0 ). Since (2. The time evolution of ψ(r. t|r .8) a knowledge of the state at all space points r ∈ Ω at some intermediate time t . This symbol will be deﬁned further below.. t ) ψ(r . . at any intermediate time the particle needs only be known at one space point. hence. t1 ] (2.e. t ) Ωd 3 r |ψ(r.11) which denotes an integral over all paths r(t) with end points r(t0 ) = r0 and r(tN ) = rN . namely the point on the classical path at time t . tN −1 ) φ(rN −1 .9) Employing a continuum of intermediate times t ∈ [t0 . t. t). the path integral r(tN )=rN d[r(t)] · · · r(t0 )=r0 (2. t = )|2 = 1 . t1 ]) Ωd Ωd 3r Ωd 3r Ωd 3r 3 r |ψ(r. t|r . t|r . ψ(r0 . .e. t ) ψ(r . i.10) We have introduced here a new symbol.4) holds for all times. . t ∈ [t0 . t ) ψ(r . the propagator is unitary. The state of a system at time t.12 Quantum Mechanical Path Integral 5. t1 ] yields an expression of the form r(tN )=rN φ(r. t0 ) = φ(r. t |r0 . t )φ(r. t|r0 . t ) φ(r . t0 ) . (2. requires then according to (2. t ∈ (2. t|r .5) 6.6) φ(r.8) 7. t0 ) Ω This relationship has the following interpretation: Assume that at time t0 a particle is generated by a source at one point r0 in space. t)|2 = (2. t ) t > t . t1 |r0 . (t > t . 8. tN −2 ) · · · φ(r1 . t. t|r0 . i.

The constant given in (2. This implies that small variations of the path near the classical path alter the value of the action integral by very little. 2 2 (2. This number is much smaller than the one for the macroscopic particle considered above and one expects that variations of the exponent of Φ[r(t)] are of the order of unity for protons.13) and = 1. The value of S[r(t)] is then Scl = 1 1 m v 2 t = erg s .11) is Φ[r(t)] = exp where S[r(t)] is the classical action integral tN 13 i S[r(t)] (2. The functional Φ[r(t)] in (2.12. the exponent is a very large imaginary number. However. t) (2.13) are built on action integrals for all possible paths.15) The exponent of (2. 2. interferences should be much less dramatic than in case of the macroscopic particle. Situations which are well described classically will be distinguished through the property that the classical path gives the dominant. One would still expect signiﬁcant descructive interference between contributions of diﬀerent paths since the value calculated is comparable to 2π. r. 2.. Scl / ≈ 8.12) is then Scl / ≈ 0.13) L(r.5 · 1027 . (2. Since this number is multiplied by ‘i’.. i.e. such that destructive interference of the contributions of such paths does not occur.14) has the same dimension as the action integral S[r(t)]. i.14) ˙ In (2.13). Any variations of Scl would then lead to strong oscillations of the contributions exp( i S) to the path integral and one can expect destructive interference betwen these contributions. for microscopic particles like the electron this is by no means the case. actually often essentially exclusive. it is comparable to action integrals as they arise for microscopic particles under typical circumstances. However. Only for paths close to the classical path is such interference ruled out. not only for the classical path. a very large number. . To show this we consider the value of the action integral for a particle of mass m = 1 g moving over a distance of 1 cm/s in a time period of 1 s. However. namely due to the property of the classical path to be an extremal of the action integral. contribution to the path integral (2.2.12) S[r(t)] = t0 ˙ dt L(r. Its value is extremely small in comparision with typical values for action integrals of macroscopic particles.12.2: Axioms 9. accordingly. in complete distinction from Classical Mechanics. The situation is very diﬀerent for microscopic particles. t) is the Lagrangian of the classical particle.e. expressions (2. r. for the electron many paths contribute and the action integrals for non-classical paths need to be known. In case of a proton with mass m = 1.0545 · 10−27 erg s .6725 · 10−24 g moving over a distance of 1 ˚ in a time period of 10−14 s the value of S[r(t)] is A Scl ≈ 10−26 erg s and. However.

The spacings between the times tj+1 and tj will all be identical.20) Here. tj ) to the next (xj+1 . namely tj+1 − tj = (tN − t0 )/N = N . respectively.14 Quantum Mechanical Path Integral 2.17) The discretization in time leads to a discretization of the paths x(t) which will be represented through the series of space–time points {(x0 .21) N 2. (xN −1 . . ˙ 2 (2. (2. (2. .16) In order to deﬁne the path integral we assume. This allows us to write the evolution operator using (2. tj+1 ) we assume that kinetic energy and potential energy are constant. > t1 > t0 letting N go to inﬁnity later. however. t0 ). tN |x0 . . These assumptions lead then to the following N 2 Riemann form for the action integral N −1 S[x(t)] = lim N →∞ N j=0 1 (xj+1 − xj )2 m − U (xj ) 2 2 N . tN )} .18) The time instances are ﬁxed. t1 ). (xN . the xj values are not. The particles have associated with them a Lagrangian L(x. For the sake of simplicity we will consider the case of particles moving in one dimension labelled by the position coordinate x. x2 .4 Propagator for a Free Particle As a ﬁrst example we will evaluate the path integral for a free particle following the algorithm introduced above. x.20) can be properly taken. at the same time. tN −1 ). (x1 . as in (2. etc. . .9). provides an avenue to evaluate path integrals. . 2. They can be anywhere in the allowed volume which we will choose to be the interval ] − ∞. t) = ˙ 1 mx2 − U (x) .12. ∞[. (2. In passing from one space–time instance (xj .12) φ(xN . namely 1 m(xj+1 − xj )2 / 2 and U (xj ). Its value is CN = m 2πi N 2 (2. −∞ dxN −1 (xj+1 −xj )2 1 − U (xj ) .19) The main idea is that one can replace the path integral now by a multiple integral over x1 .10) and (2. . a series of times tN > tN −1 > tN −2 > . . CN is a constant which depends on N (actually also on other constant in the exponent) which needs to be chosen to ascertain that the limit in (2. t0 ) = limN →∞ CN exp i N N −1 j=0 +∞ +∞ +∞ −∞ dx1 −∞ dx2 .13) and.3 How to Evaluate the Path Integral In this section we will provide an explicit algorithm which deﬁnes the path integral (2. 2 2m N (2.

using (2. tN |x0 . tN ) is to be evaluated. it is more suitable to deﬁne new integration variables yj . of (2. x(t ) = x ] for any path x(t) can then be expressed through action integral 0 0 N N an action integral over the path y(t) relative to the classical path.24) implies for the second term on the r.24) Also we use the fact that the velocity of the classical path xcl = (xN −x0 )/(tn −t0 ) is constant.25) + The condition (2. t0 ) = exp 1 im (xN − x0 )2 2 tN − t0 φ(0.24).e. which can also be written φ(xN .s.25) is. .29) This expression corresponds to the action integral in (2. 2. the origin of which coincides with the classical path of the particle. t0 ) and (xN .4: Propagator for a Free Particle 15 Rather then using the integration variables xj . x(tN ) = 0. The ˙ 1 S[x(t)|x(t ) = x .26) The ﬁrst term on the r.28) i.31) We have denoted explicitly that the action integral for a path connecting the space–time points (x0 . x(tN ) = xN ] = tN 1 ˙2 t0 dt 2 mxcl tN 1 ˙2 ˙ ˙ t0 dt 2 m(xcl + 2xcl y tN tN mxcl t0 dty + t0 dt 1 my 2 . Inserting the result into (2. x(tN ) = xN ] is S[x(t)|x(t0 ) = x0 .h. can be expressed through a path integral with endpoints x(t0 ) = 0. To see the beniﬁt of such approach we deﬁne a path y(t) as follows x(t) = xcl (t) + y(t) (2.30) d[x(t)] exp 2 tN − t0 x(t0 )=0 a result. One obtains S[x(t)|x(t0 ) = x0 . tN |0. ˙ t0 (2. tN |x0 .. due to (2. since x(t0 ) = xcl (t0 ) = x0 and x(tN ) = xcl (tN ) = xN . ˙ ˙ ˙ 2 + y2) = ˙ (2. it holds y(t0 ) = y(tN ) = 0 .10.2. t0 ) = exp S[x(t)] (2. tN dt y = y(tN ) − y(t0 ) = 0 .22) where xcl (t) is the classical path which connects the space–time points (x0 . ˙ 2 t0 (2.27) The third term can be written in the notation introduced tN 1 dt my 2 = S[x(t)|x(t0 ) = 0.13).h. (2. tN ). The resulting expression for S[x(t)|x(t0 ) = x0 . x(tN ) = 0] . xN − x0 xcl (t) = x0 + ( t − t0 ) . x(tN ) = xN ] 1 (xN − x0 )2 m + 0 + 2 tN − t 0 + S[x(t)|x(t0 ) = 0. namely. = (2. (2.s. tN dt t0 1 1 (xN − x0 )2 m x2 = m ˙ cl . 2 2 tN − t0 (2. t0 ) and (xN .23).23) tN − t0 It is essential for the following to note that. x(tN ) = 0] . t0 ) (2.12) yields x(tN )=0 im (xN − x0 )2 i φ(xN .

. (2. . . . ..37) det(Ajk ) .34) of (ajk ) deﬁning a new matrix (Ajk ) through ajk = Using det(ajk ) = m 2 N N −1 m 2 N Ajk . symmetric matrix (bjk ) and det(bjk ) = 0. .38) . 0 0 0 0 0 0 The following integral +∞ +∞ (2. ..k=1 dy1 · · · −∞ −∞ dyN −1 exp i yj bjk yk = (iπ)d det(bjk ) 1 2 . (2.. 2 −1 −1 2 (2. 2 N . real. as the quadratic form E = im 2 2 ( 2y1 − y1 y2 − y2 y1 + 2y2 − y2 y3 − y3 y2 2 N 2 2 + 2y3 − · · · − yN −2 yN −1 − yN −1 yN −2 + 2yN −1 ) N −1 = i j. In the appendix we prove +∞ +∞ d j. 0 0 . tN |0..36) which holds for a d-dimensional.31) for a free particle one needs to evaluate the following path integral φ(0. . . 0 0 .. .33) where ajk are the elements of the following symmetric (N 2 −1 0 −1 2 −1 2 m 0 −1 ajk = . .32) we split oﬀ the factor 2 mN in the deﬁnition (2. t0 ) = limN →∞ × +∞ +∞ −∞ dy1 · · · −∞ dyN −1 exp i N m 2πi N 2 N × (2.k=1 yj ajk yk (2..35) must be determined. . . In order to complete the evaluation of (2. (2.32) (yj+1 − yj )2 N −1 1 2 j=0 2 m N The exponent E can be written..16 Evaluation of the necessary path integral Quantum Mechanical Path Integral To determine the propagator (2.. . . 0 0 . noting y0 = yN = 0. .34) dy1 · · · −∞ −∞ dyN −1 exp i N −1 j.k=1 yj ajk yk − 1) × (N − 1) matrix .

D1 = 2 and obtain from (2. . . Our derivation has provided us with the value det(Ajk ) = N . 2 −1 0 0 0 −1 2 For this purpose we expand (2. deﬁning t = tN . . det(Ajk ) (2..40) . Inserting this into (2.. let say n.. . To solve this three term recursion relationship one needs two starting values. . . D3 . .42) m 2πi NN 1 2 (2. t0 ) = limN →∞ and with NN (2. .39) yields φ(0. which follows from (2.45) Expressions for Free Particle Propagator We have now collected all pieces for the ﬁnal expression of the propagator (2.. . . Using D1 = |(2)| = 2 . 0 0 Dn = . tN |0. t0 ) = m 2πi (tN − t0 ) 1 2 . . . . 0 0 −1 2 −1 . tN |0. . t|x0 .41) the same result for D2 .. t0 ) = m 2πi (t − t0 ) 1 2 exp im (x − x0 )2 2 t − t0 .41) D2 = 2 −1 −1 2 = 3 (2. .2. one can readily verify Dn = n + 1 ..18) we obtain φ(0. . One can readily verify that this procedure leads to the following recursion equation for the determinants Dn = 2 Dn−1 − Dn−2 .40) in terms of subdeterminants along the last column. 2.44) = tN − t0 .. variable.46) . .43) We like to note here for further use below that one might as well employ the ‘artiﬁcial’ starting values D0 = 1. t0 ) = limN →∞ m 2πi N 2 17 N 2πi m N N −1 2 1 . . . .. 0 0 0 . presently N − 1. 0 0 0 −1 2 .39) In order to determine det(Ajk ) we consider the dimension n of (Ajk ).31) and obtain. . .. x = xN φ(x. (2. (2.4: Propagator for a Free Particle a property which follows from det(cB) = cn detB for any n × n matrix B. . . tN |0. n = 1. (2. We seek then to evaluate the determinant of the n × n matrix 2 −1 0 . we obtain φ(0. (2.

51) as follows Eo (xo . (2. t) = 1 πδ 2 1 4 m 2πi t 1 2 +∞ dx0 exp −∞ im (x − x0 )2 x2 po − 02 + i xo 2 t 2δ Eo (xo .51) For this evaluation we adopt the strategy of combining in the exponential the terms quadratic (∼ x2 ) and linear (∼ x0 ) in the integration variable to a complete square 0 ax2 + 2bx0 = a 0 x0 + b a 2 − b2 a (2. The result (2. (2.46) can be generalized to three dimensions in a rather obvious way. and describes a single particle since 1 πδ 2 1 2 +∞ dx0 exp −∞ − x2 0 δ2 = 1. We need to evaluate for this purpose the integral ψ(x. x) + E(x) in (2. x) + E(x) (2. thereby.52) and applying (2. Below we will apply this to a particle at rest and a particle forming a so-called wave packet.53) .50) One refers to such states as wave packets.10) φ(r. according to (2. t0 ) = m 2πi (t − t0 ) 3 2 exp im (r − r0 )2 2 t − t0 . t0 )| = 2 1 πδ 2 1 2 exp − x2 0 δ2 (2. centered around x0 = 0. t0 ) = 1 πδ 2 1 4 exp − x2 po 0 +i x 2 2δ (2.46).5) to (2.48) Obviously. the associated probability distribution |ψ(x0 . t) of a free particle. t|r0 . We want to apply axiom (2.48) as the initial state using the propagator (2.5) allows us to predict the time evolution of any state function ψ(x.47) One-Dimensional Free Particle Described by Wave Packet We assume a particle at time t = to = 0 is described by the wave function ψ(x0 .49) is Gaussian of width δ.18 Quantum Mechanical Path Integral This propagator.247). We devide the contributions to the exponent Eo (xo . the wave function of the particle at later times. One obtains then for the propagator (2. We will obtain. x) = im 2 t x2 1 + i o t mδ 2 − 2xo x − po t m + f (x) (2.

57) and needs to determine the integral +∞ I = −∞ +∞ dx0 eEo (xo . We consider a transformation to a new integration variable ρ deﬁned through i 1−i t mδ 2 ρ = x0 − x− 1+ po m t t i mδ2 . (2.51) ψ(x. 2 t One chooses then f (x) to complete. We proceed as follows. t) = 1 πδ 2 1 4 2 t 1 + i mδ2 . however.52).53) E(x) = x− po mt 19 (2.55) im xo 2 t 1+i t − mδ 2 x− 1+ po m t t i mδ2 . The beginning and the end of such path are the points z1 = −∞ × i 1−i t mδ 2 .56) m 2πi t 1 2 +∞ eE(x) −∞ dx0 eEo (xo .x) im xo 2 t t − mδ 2 x− 1+ x− 1+ po m = −∞ +∞ dx0 exp 1+i t 2 2 t i mδ2 = −∞ im dx0 exp 2 t t 1+i 2 mδ xo − po m t t i mδ2 .x) (2. making certain.4: Propagator for a Free Particle im x2 − f (x) . according to (2. z2 = +∞ × i 1−i t mδ 2 (2.59) An integration path in the complex x0 –plane along the direction i 1−i t mδ 2 (2. 2 (2.2.60) is then represented by real ρ values. the square in (2.58) The integrand is an analytical function everywhere in the complex plane and we can alter the integration path. x) = One can write then (2.61) . (2. that the new path does not lead to additional contributions to the integral. (2.54) f (x) = This yields Eo (xo .

One obtains i po i po x x2 2m t E(x) = − 2 + − t t t 2δ (1 + i mδ2 ) 1 + i mδ2 1 + i mδ2 2 (2.57) reads then 1 πδ 2 1 4 2πi t .58) has a leading contribution in x0 of −x2 /δ 2 the integrand of (2. We can conclude then that (2.65) t 1 − i mδ2 t 1 + i mδ2 . Hence. (2. Quantum Mechanical Path Integral (2. 1+b 1+b ﬁnally E(x) = − (x − po m (2. the paths between z1 → z1 and z2 → z2 have a real part of x0 of ±∞.64) ψ(x.62) If one can show that an integration of (2.67) We need to determine ﬁnally (2.54) using (2. z2 = +∞ . one can show that z1 lies at −∞ − i × ∞ and z2 at +∞ + i × ∞. using a ab = a − .63) holds and.20 whereas the original path in (2.66) yields ψ(x. Since the exponent in (2.58) along the path z1 → z1 and along the path z2 → z2 gives only vanishing contributions one can replace (2. I = Equation (2. In fact. accordingly. (2. t) = t 1 − i mδ2 t 1 + i mδ2 1 4 1 πδ 2 (1 + 2 t2 m2 δ 4 1 4 ) exp [ E(x) ] .55).58) by t mδ 2 +∞ I = i 1−i dρ exp − −∞ m 2 t 1+ t mδ 2 2 ρ2 (2.58) vanishes 0 for Re x0 → ±∞. t m(1 + i mδ2 ) (2. t) = Seperating the phase factor 1 t 1 + i mδ2 1 4 1 2 exp [ E(x) ] .63) which can be readily evaluated.58) has the end points z1 = −∞ . (2.69) t)2 t 2δ 2 (1 + i mδ2 ) + i po x − i p2 o t 2m (2.70) .68) and.

t) = exp i i p2 po o x − (t − to ) m 2m . instead of (2.72) 2 t2 δ 1+ 2 4 (2.72). (2.49).71). (2.48) and for the ﬁnal state (2.71) conserves the phase factor exp[i(po / )x] of the original wave function (2.71) × exp − (x − po m t)2 ) 2δ 2 (1 + 2 t2 m2 δ 4 (1 − i .72) for the initial state (2. respectively. the case of arbitrary to is recovered iby replacing t → to in (2. Another interesting observation is that the wave function (2.49).f. The width of the distribution (2.72) moves in the direction of the positive x-axis with velocity vo = po /m which identiﬁes po as the momentum of the particle.2. (2.75) i. t0 ) = exp i po i p2 o xo − to m 2m .74) remains invariant in time.e. 2. (2. (2..67) provides the complete expression of the wave function at all times t ψ(x. a timedependent phase factor exp[− i (p2 /2m) t] arises which is related to the energy = p2 /2m of a o o particle with momentum po . t0 ) = exp i po xo m . coinciding at t = 0 with the width of the initial distribution (2.73) m δ increases with time. t) = exp i po i p2 o x − t m 2m .48) for δ → ∞. In this case holds ψ(x. the spatial dependence of the initial state (2.4: Propagator for a Free Particle which inserted in (2.76) From this we conclude that an initial wave function ψ(xo . The conservation of this factor is particularly striking for the (unnormalized) initial wave function ψ(x0 . (2.74) which corresponds to (2. (2.48)] to = 0. The center of the distribution (2. This yields.77) . t) = t 1 − i mδ2 t 1 + i mδ2 1 4 21 1 πδ 2 (1 + 2 t2 ) m2 δ 4 1 4 × po i p2 t o ) + i x − t mδ 2 2m (2.71. t)| = 2 1 πδ 2 (1 + 2 t2 m2 δ 4 1 2 ) exp − (x − δ 2 (1 + po m t)2 ) 2 t2 m2 δ 4 .48) and that the respective phase factor is related with the velocity of the classical particle and of the center of the distribution (2. However. The corresponding probability distribution is |ψ(x. We had assumed above [c.72).72) Comparision of Moving Wave Packet with Classical Motion It is revealing to compare the probability distributions (2.75) ψ(x. This ‘spreading’ of the wave function is a genuine quantum phenomenon.

81) vanishes2 .78) i. y(t). For this purpose we use d xcl y = ˙ ˙ (xcl y) − xcl y ˙ ¨ (2. the end points of y(t) are y(t0 ) = 0 Inserting (2.22 becomes at t > to ψ(x. the spatial as well as the temporal dependence of the wave function remains invariant in this case. x. In order to simplify this task we deﬁne again a new path y(t) x(t) = xcl (t) + y(t) (2. (2.84) . t0 ) = x(t0 )=x0 d[x(t)] exp i S[x(t)] (2. t) ˙ δL 1 1 mx2 − c(t)x2 − e(t)xcl ˙ cl cl 2 2 1 1 = my 2 − c(t)y 2 ˙ 2 2 = mxcl y(t) − c(t)xcl y − e(t)y . (2. ˙ ˙ = (2.. 2. xcl .83) (2. 1. t) ˙ L (y.79) for a quadratic Lagrangian L(x. y(tN ) = 0 . t) + δL ˙ ˙ ˙ ˙ where L(xcl .81) for an arbitrary path x(t) with end points x(t0 ) = x0 and x(tN ) = xN . 2.10. y(t).e.5 Propagator for a Quadratic Lagrangian x(tN )=xN We will now determine the propagator (2.85) We want to show now that the contribution of δL to the action integral (2.12. t) + L (y. Such states play a cardinal role in quantum mechanics. 2 . x. 2. tN |x0 . xcl + y(t). t) ˙ (2.80) For this purpose we need to determine the action integral S[x(t)] = t0 dt L(x. y(t)] which vanishes according to the Hamiltonian principle as discussed in Sect. Obviously.82) one obtains L(xcl + y. One refers to the respective states as stationary states. xcl .80) into (2. t) = ˙ 1 1 mx2 − c(t)x2 − e(t)x .13) φ(xN . t) = L(xcl . ˙ 2 2 tN (2.86) dt The reader may want to verify that the contribution of δL to the action integral is actually equal to the diﬀerential δS[xcl . t) = exp i Quantum Mechanical Path Integral po i p2 o x − t m 2m .82) which describes the deviation from the classical path xcl (t) with end points xcl (t0 ) = x0 and xcl (tN ) = xN .

.. 2 −1 0 0 0 −1 2 c1 0 0 .93) . 0 0 m ajk = . of (2. tj = t0 + form j. cN −2 0 0 0 0 0 cN −1 (2. x t0 (2. ..2. . . For the evaluation of φ(0... . . 0 0 0 c2 0 .79) i ˜ φ(xN . . tN |0.5: Propagator for a Quadratic Lagrangian and obtain tN t0 23 dt δL = m [xcl y]|t0 − ˙ t N tN dt [ m¨cl (t) + c(t) xcl (t) + e(t) ] y(t) .88) and. One can then express the propagator (2. 2 N . tN |0. tN |0. .91) (yj+1 − yj )2 1 2 2m N where cj = c(tj ). . t0 ) = exp S[xcl (t)] φ(0. . . . 0 0 0 .88) vanishes. (2. . 0 0 N 0 − .. vanishes. tN |x0 .32).91) through the quadratic N −1 E = i j.24) to the Lagrangian (2. t0 ) we will adopt a strategy which is similar to that used for the evaluation of (2. . . . . also the second contribution on the r.k=1 yj ajk yk (2. . One can express the exponent E in (2.h. . . . . t0 ) = y(tN )=0 d[y(t)] exp y(t0 )=0 i tN dt L (y. 0 0 0 c3 . t0 ) (2. . y. . .31) for the free particle propagator. .s. . . .. 2 . t) ˙ t0 .. .90) Evaluation of the Necessary Path Integral We have achieved for the quadratic Lagrangian a separation in terms of a classical action integral and a propagator connecting the end points y(t0 ) = 0 and y(tN ) = 0 which is analogue to the ˜ result (2. Applying the Euler–Lagrange conditions (1.92) where ajk are the elements of the following (N − 1) × (N − 1) matrix 2 −1 0 . .. hence. 0 0 0 −1 2 . tN |0.s.87) According to (2. 0 0 0 . .h. . .80) yields for the classical path m¨cl + c(t) xcl + e(t) = 0 x (2.83) the ﬁrst term on the r. .89) where ˜ φ(0. 0 0 −1 2 −1 . . t0 ) = limN →∞ × +∞ +∞ −∞ dy1 · · · −∞ dyN −1 exp N i N N −1 j=0 m 2πi N 2 N × − 1 2 cj 2 yj (2. . The discretization scheme adopted above yields in the present case ˜ φ(0. . . .

24

Quantum Mechanical Path Integral

In case det(ajk ) = 0 one can express the multiple integral in (2.91) according to (2.36) as follows ˜ φ(0, tN |0, t0 ) = limN →∞ m 2πi m 2πi

N 2

N

(iπ)N −1 det(a) 1

1 2

= limN →∞

1

2

N

2

N

N −1

m

det(a)

.

(2.94)

**˜ In order to determine φ(0, tN |0, t0 ) we need to evaluate the function f (t0 , tN ) = limN →∞ According to (2.93) holds DN −1 = = 2−
**

m c1

2 N 2 N

2

N

N

N −1

m

det(a)

.

(2.95)

def

2

N

N −1

m

det(a) 0 0 0 . . .

m cN −2

2 N

(2.96) 0

−1 −1 2 − . . . 0 0

2 N

0 ... −1 . . . ... . .. . . .

m c3

−1 2 − 0 . . . 0 0

m c2

0 ... 2 − 0

−1 2 −

m cN −1

2 N

0 0 . . . −1

In the following we will asume that the dimension n = N − 1 of the matrix in (2.97) is variable. One can derive then for Dn the recursion relationship Dn = 2−

2 N

m

cn

Dn−1 − Dn−2

(2.97)

using the well-known method of expanding a determinant in terms of the determinants of lower dimensional submatrices. Using the starting values [c.f. the comment below Eq. (2.43)] D0 = 1 ; D1 = 2 −

2 N

m

c1

(2.98)

this recursion relationship can be employed to determine DN −1 . One can express (2.97) through the 2nd order diﬀerence equation Dn+1 − 2Dn + Dn−1

2 N

= −

cn+1 Dn . m

(2.99)

Since we are interested in the solution of this equation in the limit of vanishing N we may interpret (2.99) as a 2nd order diﬀerential equation in the continuous variable t = n N + t0 d2 f (t0 , t) c(t) = − f (t0 , t) . 2 dt m (2.100)

**2.5: Propagator for a Quadratic Lagrangian The boundary conditions at t = t0 , according to (2.98), are f (t0 , t0 )
**

df (t0 ,t) dt t=t0

25

= =

N N

D0 = 0 ; D1 − D0

N

= 2−

2 N

m

c1 − 1 = 1 .

(2.101)

**We have then ﬁnally for the propagator (2.79) φ(x, t|x0 , t0 ) = m 2πi f (to , t)
**

1 2

exp

i

S[xcl (t)]

(2.102)

where f (t0 , t) is the solution of (2.100, 2.101) and where S[xcl (t)] is determined by solving ﬁrst the Euler–Lagrange equations for the Lagrangian (2.80) to obtain the classical path xcl (t) with end points xcl (t0 ) = x0 and xcl (tN ) = xN and then evaluating (2.81) for this path. Note that the required solution xcl (t) involves a solution of the Euler–Lagrange equations for boundary conditions which are diﬀerent from those conventionally encountered in Classical Mechanics where usually a solution for initial conditions xcl (t0 ) = x0 and xcl (t0 ) = v0 are determined. ˙

2.6

Wave Packet Moving in Homogeneous Force Field

We want to consider now the motion of a quantum mechanical particle, decribed at time t = to by a wave packet (2.48), in the presence of a homogeneous force due to a potential V (x) = − f x. As we have learnt from the study of the time-development of (2.48) in case of free particles the wave packet (2.48) corresponds to a classical particle with momentum po and position xo = 0. We expect then that the classical particle assumes the following position and momentum at times t > to y(t) p(t) = = po 1 f (t − to ) + (t − to )2 m 2m po + f (t − to ) (2.103) (2.104)

The Lagrangian for the present case is L(x, x, t) = ˙ 1 m x2 + f x . ˙ 2 (2.105)

This corresponds to the Lagrangian in (2.80) for c(t) ≡ 0, e(t) ≡ −f . Accordingly, we can employ the expression (2.89, 2.90) for the propagator where, in the present case, holds L (y, y, t) = 1 my 2 ˙ ˙ 2 ˜ tN |0, t0 ) is the free particle propagator (2.45). One can write then the propagator such that φ(0, for a particle moving subject to a homogeneous force φ(x, t|x0 , t0 ) = m 2πi (t − t0 )

1 2

exp

i

S[xcl (τ )] .

(2.106)

Here S[xcl (τ )] is the action integral over the classical path with end points xcl (to ) = xo , xcl (t) = x . (2.107)

26 The classical path obeys m xcl = f . ¨ The solution of (2.107, 2.108) is xcl (τ ) = xo +

Quantum Mechanical Path Integral

(2.108)

x − xo 1 f − (t − to ) t − to 2m

τ +

1 f 2 τ 2m

(2.109)

as can be readily veriﬁed. The velocity along this path is xcl (τ ) = ˙ x − xo 1 f f − (t − to ) + τ t − to 2m m (2.110)

**and the Lagrangian along the path, considered as a function of τ , is g(τ ) = = 1 mx2 (τ ) + f xcl (τ ) ˙ cl 2 2 1 x − xo 1 f x − xo 1 f m − (t − to ) + f − (t − to ) τ 2 t − to 2m t − to 2m 1 f2 2 x − xo 1 f 1 f2 2 + τ + f xo + f − (t − to ) τ + τ 2 m t − to 2m 2 m 1 m 2 + x − xo 1 f − (t − to ) t − to 2m f2 2 τ + f xo m
**

t 2

=

+ 2f

x − xo 1 f − (t − to ) t − to 2m

τ (2.111)

**One obtains for the action integral along the classical path S[xcl (τ )] =
**

to

dτ g(τ ) x − xo 1 f − (t − to ) t − to 2m 1 f x − xo − (t − to ) t − to 2m

2

=

1 m 2 +f +

(t − to ) (t − to )2

1 f2 (t − to )3 + xo f (t − to ) 3 m

=

1 1 f2 1 (x − xo )2 m + (x + xo ) f (t − to ) − (t − to )3 2 t − to 2 24 m (2.112)

**and, ﬁnally, for the propagator φ(x, t|xo , to ) = × exp m 2πi (t − to )
**

1 2

×

(2.113)

im (x − xo )2 i i f2 + (x + xo ) f (t − to ) − (t − to )3 2 t − to 2 24 m

2.5: Propagator for a Quadratic Lagrangian

27

The propagator (2.113) allows one to determine the time-evolution of the initial state (2.48) using (2.5). Since the propagator depends only on the time-diﬀerence t − to we can assume, withoult loss of generality, to = 0 and are lead to the integral

+∞ m 1 1 4 2 ψ(x, t) = dx0 πδ 2 2πi t −∞ im (x − x0 )2 x2 po i i f2 3 exp − 02 + i xo + (x + xo ) f t − t 2 t 2δ 2 24 m

1

(2.114)

Eo (xo , x) + E(x) To evaluate the integral we adopt the same computational strategy as used for (2.51) and divide the exponent in (2.114) as follows [c.f. (2.54)] Eo (xo , x) = im 2 t x2 1 + i o im 2 t t mδ 2 x2 + − 2xo x − po f t2 t− m 2m − + f (x) (2.115) (2.116)

E(x) =

f t2 x − f (x) m

1 f 2 t3 . 24 m

**One chooses then f (x) to complete, according to (2.52), the square in (2.115) 2 2 x − po t − f t m 2m f (x) = . t 1 + i mδ2 This yields Eo (xo , x) = im xo 2 t 1+i x− t− t − 2 mδ t 1 + i mδ2
**

1 4 2 t2 ) m2 δ 4

(2.117)

po m

f t2 2m

2

.

(2.118)

**Following in the footsteps of the calculation on page 18 ﬀ. one can state again ψ(x, t) =
**

t 1 − i mδ2 t 1 + i mδ2

1 4

1 πδ 2 (1 +

exp [ E(x) ]

(2.119)

**and is lead to the exponential (2.116) E(x) = − where S(x) = = + x2 1+i t mδ 2 + x
**

2

1 f 2 t3 im S(x) + t 24 m 2 t(1 + i mδ2 ) f t2 m 1+i t mδ 2 − po f t2 t− m 2m

2 2

(2.120)

1+i t mδ 2 −

− x−

x−

x− x

f t2 po t− m 2m

po f t2 t− m 2m

2

f t2 + 2x m

po f t2 t+ m 2m

f t2 po t+ m 2m

1+i

t mδ 2 (2.121)

**28 Inserting this into (2.120) yields x− i
**

po m

Quantum Mechanical Path Integral

E(x)

=

− +

t−

2δ 2 1 + i

f t2 2m t mδ 2

2

(2.122) i 2m po t + p o f t 2 + f 2 t3 f 2 t3 + 4 12

(po + f t) x −

**The last term can be written − i 2m p o t + po f t 2 + f 2 t3 3 = − i 2m
**

t

dτ (po + f τ )2 .

0

(2.123)

**Altogether, (2.119, 2.122, 2.123) provide the state of the particle at time t > 0
**

t 1 − i mδ2 t 1 + i mδ2

1 4

ψ(x, t)

=

1 πδ 2 (1 + x− 2δ 2

po m

2 t2 ) m2 δ 4

1 4

×

× exp − × exp i

t−

f t2 2m

2

1+

2 t2 m2 δ 4

1−i

t

t mδ 2

× . (2.124)

(po + f t) x −

i

0

dτ

(po + f τ )2 2m

**The corresponding probablity distribution is |ψ(x, t)|2 = 1 πδ 2 (1 +
**

2 t2 m2 δ 4 1 2

)

x− t− exp − 2 2 δ 2 1 + m2tδ4

po m

f t2 2m

2

.

(2.125)

Comparision of Moving Wave Packet with Classical Motion It is again [c.f. (2.4)] revealing to compare the probability distributions for the initial state (2.48) and for the states at time t, i.e., (2.125). Both distributions are Gaussians. Distribution (2.125) moves along the x-axis with distribution centers positioned at y(t) given by (2.103), i.e., as expected for a classical particle. The states (2.124), in analogy to the states (2.71) for free particles, exhibit a phase factor exp[ip(t)x/ ], for which p(t) agrees with the classical momentum (2.104). While these properties show a close correspondence between classical and quantum mechanical behaviour, the distribution shows also a pure quantum eﬀect, in that it increases its width . This increase, for the homogeneous force case, is identical to the increase (2.73) determined for a free particle. Such increase of the width of a distribution is not a necessity in quantum mechanics. In fact, in case of socalled bound states, i.e., states in which the classical and quantum mechanical motion is conﬁned to a ﬁnite spatial volume, states can exist which do not alter their spatial distribution in time. Such states are called stationary states. In case of a harmonic potential there exists furthermore the possibility that the center of a wave packet follows the classical behaviour and the width remains constant in time. Such states are referred to as coherent states, or Glauber states, and will be

2.5: Propagator for a Quadratic Lagrangian

29

studied below. It should be pointed out that in case of vanishing, linear and quadratic potentials quantum mechanical wave packets exhibit a particularly simple evolution; in case of other type of potential functions and, in particular, in case of higher-dimensional motion, the quantum behaviour can show features which are much more distinctive from classical behaviour, e.g., tunneling and interference eﬀects.

**Propagator of a Harmonic Oscillator
**

In order to illustrate the evaluation of (2.102) we consider the case of a harmonic oscillator. In this case holds for the coeﬃcents in the Lagrangian (2.80) c(t) = mω 2 and e(t) = 0, i.e., the Lagrangian is 1 1 L(x, x) = m x2 − m ω 2 x2 . ˙ ˙ (2.126) 2 2 .We determine ﬁrst f (t0 , t). In the present case holds ¨ f = −ω 2 f ; f (t0 , t0 ) = 0 ; f˙(to , to ) = 1 . (2.127)

The solution which obeys the stated boundary conditions is f (t0 , t) = sinω(t − t0 ) . ω (2.128)

We determine now S[xcl (τ )]. For this purpose we seek ﬁrst the path xcl (τ ) which obeys xcl (t0 ) = x0 and xcl (t) = x and satisﬁes the Euler–Lagrange equation for the harmonic oscillator m¨cl + mω 2 xcl = 0 . x This equation can be written xcl = −ω 2 xcl . ¨ the general solution of which is xcl (τ ) = A sinω(τ − t0 ) + B cosω(τ − t0 ) . The boundary conditions xcl (t0 ) = x0 and xcl (t) = x are satisﬁed for B = x0 ; and the desired path is xcl (τ ) = where we introduced c = cosω(t − to ) , s = sinω(t − to ) (2.134) We want to determine now the action integral associated with the path (2.133, 2.134)

t

(2.129) (2.130)

(2.131)

A =

x − x0 cosω(t − to ) , sinω(t − t0 )

(2.132)

x − x0 c sinω(τ − t0 ) + x0 cosω(τ − t0 ) s

(2.133)

S[xcl (τ )] =

t0

dτ

1 1 mx2 (τ ) − mω 2 x2 (τ ) ˙ cl cl 2 2

(2.135)

30

Quantum Mechanical Path Integral

For this purpose we assume presently to = 0. From (2.133) follows for the velocity along the classical path x − x0 c xcl (τ ) = ω ˙ cosωτ − ω x0 sinωτ (2.136) s and for the kinetic energy 1 (x − x0 c)2 mω 2 cos2 ωτ 2 s2 x − x0 c −mω 2 xo cosωτ sinωτ s 1 + mω 2 x2 sin2 ωτ o 2 Similarly, one obtains from (2.133) for the potential energy 1 mx2 (τ ) ˙ cl 2 = 1 mω 2 x2 (τ ) cl 2 = 1 (x − x0 c)2 mω 2 sin2 ωτ 2 s2 x − x0 c +mω 2 xo cosωτ sinωτ s 1 + mω 2 x2 cos2 ωτ o 2 =

(2.137)

(2.138)

Using 1 1 + cos2ωτ 2 2 1 1 2 sin ωτ = − cos2ωτ 2 2 1 cosωτ sinωτ = sin2ωτ 2 the Lagrangian, considered as a function of τ , reads cos2 ωτ g(τ ) = 1 1 mx2 (τ ) − mω 2 x2 (τ ) ˙ cl cl 2 2 = 1 (x − x0 c)2 mω 2 cos2ωτ 2 s2 x − x0 c −mω 2 xo sin2ωτ s 1 − mω 2 x2 cos2ωτ o 2

t 0 dτ g(τ )

(2.139) (2.140) (2.141)

(2.142)

**Evaluation of the action integral (2.135), i.e., of S[xcl (τ )] =
**

t

requires the integrals (2.143) (2.144)

dτ cos2ωτ

0 t

= =

dτ sin2ωτ

0

1 1 sin2ωt = sc 2ω ω 1 2 1 [ 1 − cos2ωt ] = s 2ω ω

where we employed the deﬁnition (2.134) Hence, (2.135) is, using s2 + c2 = 1, S[xcl (τ )] = = = 1 (x − x0 c)2 x − x0 c 2 1 mω s c − mωxo s − mω 2 x2 s c o 2 2 s s 2 mω (x2 − 2xxo c + x2 c2 ) c − 2xo xs2 + 2x2 cs2 − x2 s2 c o o o 2s mω (x2 + x2 ) c − 2xo x o 2s

(2.145)

2.5: Propagator for a Quadratic Lagrangian and, with the deﬁnitions (2.134), S[xcl (τ )] = mω (x2 + x2 ) cosω(t − t0 ) − 2x0 x . 0 2sinω(t − t0 )

31

(2.146)

**For the propagator of the harmonic oscillator holds then φ(x, t|x0 , t0 ) = × exp
**

imω 2 sinω(t − t0 ) mω 2πi sinω(t − t0 )

1 2

× . (2.147)

(x2 + x2 ) cosω(t − t0 ) − 2x0 x 0

Quantum Pendulum or Coherent States As a demonstration of the application of the propagator (2.147) we use it to describe the time development of the wave function for a particle in an initial state ψ(x0 , t0 ) = mω π

1 4

exp

−

mω(x0 − bo )2 i + po xo 2

.

(2.148)

The initial state is decribed by a Gaussian wave packet centered around the position x = bo and corresponds to a particle with initial momentum po . The latter property follows from the role of such factor for the initial state (2.48) when applied to the case of a free particle [c.f. (2.71)] or to the case of a particle moving in a homogeneous force [c.f. (2.124, 2.125)] and will be borne out of the following analysis; at present one may regard it as an assumption. If one identiﬁes the center of the wave packet with a classical particle, the following holds for the time development of the position (displacement), momentum, and energy of the particle po sin ω(t − to ) displacement mω p(t) = − mωbo sin ω(t − to ) + po cos ω(t − to ) momentum b(t) = bo cos ω(t − to ) + p2 o + 1 mω 2 b2 o 2 2m

o

(2.149)

=

energy

We want to explore, using (2.5), how the probability distribution |ψ(x, t)|2 of the quantum particle propagates in time. The wave function at times t > t0 is

∞

ψ(x, t) =

−∞

dx0 φ(x, t|x0 , t0 ) ψ(x0 , t0 ) .

(2.150)

Expressing the exponent in (2.148) imω 2 sinω(t − to ) i (xo − bo )2 sinω(t − to ) + 2po xo sin ω(t − to ) mω (2.151)

**(2.147, 2.150, 2.151) can be written ψ(x, t) = mω π
**

1 4

m 2πiω sinω(t − t0 )

1 2

∞

dx0 exp [ E0 + E ]

−∞

(2.152)

32 where E0 (xo , x) = imω 2 s imω 2 s

Quantum Mechanical Path Integral

x2 c − 2xo x + isx2 − 2isxo bo + o o

2po xo s + f (x) mω (2.153) (2.154) (2.155)

E(x)

=

x2 c + isb2 − f (x) o

.

c = cos ω(t − to ) ,

s = sin ω(t − to ) .

Here f (x) is a function which is introduced to complete the square in (2.153) for simpliﬁcation of the Gaussian integral in x0 . Since E(x) is independent of xo (2.152) becomes ψ(x, t) = mω π

1 4

m 2πiω sinω(t − t0 )

1 2

eE(x)

∞

dx0 exp [ E0 (xo , x)]

−∞

(2.156)

We want to determine now Eo (xo , x) as given in (2.153). It holds Eo = imω po x2 eiω(t−to ) − 2xo (x + isbo − s) + f (x) o 2 s mω po 2 −iω(t−to ) s) e . mω

2

(2.157)

For f (x) to complete the square we choose f (x) = (x + isbo − One obtains for (2.157) E0 (xo , x) = imω po exp [iω(t − t0 )] x0 − (x + isbo − s) exp (−iω(t − t0 ) 2 s mω

1 2

(2.158)

.

(2.159)

**To determine the integral in (2.156) we employ the integration formula (2.247) and obtain
**

+∞

dx0 e

−∞

E0 (x0 )

=

2πi sinω(t − t0 ) mω exp[iω(t − t0 )]

(2.160)

**Inserting this into (2.156) yields mω 1 E(x) 4 e π For E(x) as deﬁned in (2.154) one obtains, using exp[±iω(t − to )] = c ± is, ψ(x, t) = E(x) =
**

imω 2 s

(2.161)

**x2 c + isb2 − x2 c + isx2 − 2isxbo c − 2s2 xbo o
**

po po + s2 b2 c − is3 b2 + 2 mω xsc + 2i mω bo s2 c o o po po − 2i mω xs2 + 2 mω bo s3 − p2 o m2 ω 2

s2 c + i mpo 2 s3 2ω

2

=

− mω 2

**x2 + c2 b2 − 2xbo c + 2ixsbo − ib2 sc o o
**

p2 o m2 ω 2 po s2 − 2i mω xc

2

po po −2 mω xs + 2 mω bo sc + po −2i mω bo s2 + i mpo 2 sc 2ω

=

− mω (x − cbo − 2 − i

2

po mω

s)2 +

i

(− mωbo s + po c) x i po bo s2 (2.162)

po ( 2mω − 1 mωb2 )sc + o 2

**2.5: Propagator for a Quadratic Lagrangian We note the following identities
**

t

33

dτ

to

p2 (τ ) 2m p2 mωb2 o o − 2mω 2 sc − 1 bo po s2 2 (2.163)

1 1 o (t − to ) + 2 2 t 2 b2 (τ ) mω dτ 2 to = = 1 2

o (t

− to ) −

1 2

p2 mωb2 o o − 2mω 2

sc +

1 bo po s2 2

(2.164)

**˙ where we employed b(τ ) and p(τ ) as deﬁned in (2.149). From this follows, using p(τ ) = mb(τ ) and the Lagrangian (2.126),
**

t to

˙ dτ L[b(τ ), b(τ )] =

p2 mωb2 o o − 2mω 2

sc − bo po s

(2.165)

**such that E(x) in (2.162) can be written, using again (2.149)), E(x) = − mω i 1 i [x − b(t)]2 + p(t) x − i ω (t − to ) − 2 2
**

t to

˙ dτ L[b(τ ), b(τ )]

(2.166)

Inserting this into (2.161) yields, ψ(x, t) = mω 1 mω 4 × exp − [x − b(t)]2 × π 2 i 1 i t ˙ × exp p(t) x − i ω (t − to ) − dτ L[b(τ ), b(τ )] 2 to (2.167)

where b(t), p(t), and in (2.149).

o

are the classical displacement, momentum and energy, respectively, deﬁned

**Comparision of Moving Wave Packet with Classical Motion The probability distribution associated with (2.167) |ψ(x, t)|2 = mω π
**

1 2

exp −

mω

[x − b(t)]2

(2.168)

is a Gaussian of time-independent width, the center of which moves as described by b(t) given in (2.148) , i.e., the center follows the motion of a classical oscillator (pendulum) with initial position bo and initial momentum po . It is of interest to recall that propagating wave packets in the case of vanishing [c.f. (2.72)] or linear [c.f. (2.125)] potentials exhibit an increase of their width in time; in case of the quantum oscillator for the particular width chosen for the initial state (2.148) the width, actually, is conserved. One can explain this behaviour as arising from constructive interference due to the restoring forces of the harmonic oscillator. We will show in Chapter 4 [c.f. (4.166, 4.178) and Fig. 4.1] that an initial state of arbitrary width propagates as a Gaussian with oscillating width.

34

Quantum Mechanical Path Integral

In case of the free particle wave packet (2.48, 2.71) the factor exp(ipo x) gives rise to the translational motion of the wave packet described by po t/m, i.e., po also corresponds to initial classical momentum. In case of a homogeneous force ﬁeld the phase factor exp(ipo x) for the initial state (2.48) gives rise to a motion of the center of the propagating wave packet [c.f. (2.125)] described by 1 (po /m)t + 2 f t2 such that again po corresponds to the classical momentum. Similarly, one observes for all three cases (free particle, linear and quadratic potential) a phase factor exp[ip(t)x/ ] for the propagating wave packet where fp(t) corresponds to the initial classical momentum at time t. One can, hence, summarize that for the three cases studied (free particle, linear and quadratic potential) propagating wave packets show remarkably close analogies to classical motion. We like to consider ﬁnally the propagation of an initial state as in (2.148), but with bo = 0 and po = 0. Such state is given by the wave function ψ(x0 , t0 ) = mω π

1 4

exp

−

mωx2 iω 0 − to 2 2

.

(2.169)

where we added a phase factor exp(−iωto /2). According to (2.167) the state (2.169) reproduces itself at later times t and the probablity distribution remains at all times equal to mω π

1 2

exp

−

mωx2 0

,

(2.170)

i.e., the state (2.169) is a stationary state of the system. The question arises if the quantum oscillator posesses further stationary states. In fact, there exist an inﬁnite number of such states which will be determined now.

2.7

Stationary States of the Harmonic Oscillator

In order to ﬁnd the stationary states of the quantum oscillator we consider the function W (x, t) = exp 2 mω x e−iωt − e−2iωt − mω 2 iωt x − 2 2 . (2.171)

We want to demonstrate that w(x, t) is invariant in time, i.e., for the propagator (2.147) of the harmonic oscillator holds

+∞

W (x, t) =

−∞

dxo φ(x, t|xo , to ) W (xo , to ) .

(2.172)

We will demonstrate further below that (2.172) provides us in a nutshell with all the stationary states of the harmonic oscillator, i.e., with all the states with time-independent probability distribution. In order to prove (2.172) we express the propagator, using (2.147) and the notation T = t − to

2 mω φ(x, t|xo , to ) = e × π (1 − e−2iωT ) 1 + e−2iωT mω 2x xo e−iωT mω 2 × exp − (xo + x2 ) − 2 1 − e−2iωT ω 1 − e−2iωT 1

− 1 iωT 2

(2.173)

2.5: Propagator for a Quadratic Lagrangian One can write then the r.h.s. of (2.172) I = e where Eo (xo , x) = − mω 2 x2 o 1 + e−2iωT +1 1 − e−2iωT 2xe−iωT +2 1 − e−2iωT mω e−iωto + f (x)

− 1 iωt 2

35

mω π (1 − e−2iωT )

1 2

+∞

dxo exp[ Eo (xo , x) + E(x) ]

−∞

(2.174)

(2.175)

+ 2xo E(x) = − mω 2 x2

**1 + e−2iωT 2 + e−2iωto − f (x) −2iωT 1−e mω
**

2

(2.176)

Following the by now familiar strategy one choses f (x) to complete the square in (2.175), namely, 1 f (x) = (1 − e−2iωT ) 2 This choice of f (x) results in Eo (xo , x) = − mω 2 xo 2 1 − e−2iωT

2

2xe−iωT + 2 1 − e−2iωT

mω

e

−iωto

.

(2.177)

=

**1 − e−2iωT 2xe−iωT + +2 2 1 − e−2iωT mω i (xo + zo )2 i (e−2iωT − 1)
**

+∞

mω

e

−iωto

(2.178)

**for some constant zo ∈ C. Using (2.247) one obtains dxo e
**

−∞ Eo (xo ,x)

=

π (1 − e−2iωT mω

1 2

(2.179)

and, therefore, one obtains for (2.174) I = e− 2 iωt eE(x) . For E(x), as given in (2.176, 2.177), holds E(x) = − mω 2 −4 mω 2 x2 2 2x2 e−2iωT 1 + e−2iωT + e−2iωto − 1 − e−2iωT mω 1 − e−2iωT x e−iωT − 2 (1 − e−2iωT ) mω e−2iωto

1

(2.180)

mω

=

−

x2 − 4

mω mω

x e−iωt +

2 e−2iωt mω (2.181)

=

−

mω 2 x + 2 2

x e−iωt − e−2iωt

of (2. t − to ) = φ(x.172) I = exp 2 mω x e−iωt − e−2iωt − Quantum Mechanical Path Integral mω 2 1 x − iωt 2 2 . xo .171.188) identiﬁes the functions ψn (x. t) +∞ −∞ dto 1 ˜ exp[−iω(m + 1 ) to ] φm (xo ) 2 m! (2. . t).s.183) where the expansion coeﬃcients are functions of x. t) = n=0 1 ˜ exp[ − iω(n + 1 ) t ] φn (x) 2 n! (2. We want to inspect the consequences of the invariance property (2. t − to ) 1 ˜ φn (xo ) n! (2. n = 1. i.e.188) ˜ ˜ Equation (2. In contrast to W (x.184) 1 ˜ exp[−iω(n + 1 ) t] φn (x) 2 n! ∞ +∞ = m=0 −∞ dxo Φ(x..185) Replacing t → t + to yields ∞ n=0 1 ˜ exp[−iω(n + 1 ) (t + to )] φn (x) 2 n! ∞ +∞ = m=0 −∞ dxo Φ(x.171) can be expanded in terms of e−inωt . t|xo . but not of t. (2.172) is a function of t − to and deﬁning accordingly Φ(x.. one obtains for the r.172). t − to ) 1 ˜ exp[−iω(m + 1 ) to ] φm (xo ) 2 m! (2. .187) or 1 ˜ exp[−iω(n + 2 ) t] φn (x) +∞ = −∞ ˜ dxo φ(x. to ).h. 1 exp[iω(n + 2 ) to ] · · · . 2 (2. . Accoordingly. t|xo . . We note that the factor exp(2 mω/ xe−iωt − e−2iωt ) in (2. to ) exp[−iω(n + 1 ) to ] φn (xo ) . xo . xo . to ) we express (2.171) ∞ W (x.182) Comparision with (2. Noting that the propagator (2.36 Altogether. which also exhibits such invariance. 2. one can expand (2. 2. results in 1 ˜ exp[−iω(n + 1 ) t] φn (x) 2 n! +∞ = −∞ dxo Φ(x. t) = exp[−iω(n + 1 ) t] φn (x) as invariants under the 2 action of the propagator φ(x.186) Fourier transform.147) in (2.172). t|xo .171) concludes the proof of (2. xo .172) in the form ∞ n=0 (2.

191) (2.183).197) are called Hermite polynomials which are polynomials of degree n which will be evaluated below. t) such that +∞ −∞ 2 dx |ψ(x.2. ˜ |φ stationary wave functions ψn (x. t) = z 2 e−y where w(y. The Hermite Polynomials ˜ The function (2. In the following we will characterize the functions φn (x) and determine the normalization constants Nn . in a ‘nutshell’ all information on the Hermite polynomials.194) plays a central role for the Hermite polynomials since it contains.183). .192) w(y. t) of the quantum mechanical harmonic oscillator ˜ ψn (x. Expansion (2.183). Actually. characterizes the wave functions φn (x). z) = exp(2yz − z 2 ) . Expression (2.5: Propagator for a Quadratic Lagrangian 37 ˜ ˜ the functions ψn (x. .183) reads then w(y. (2. To obtain ˜n (x) we simplify the expansion (2.196) where Hn (y) = ey 2 /2 (2. according to (2.190) ˜ is obeyed. .198) ∂z n z=0 .193) (2. 1. (2.171).189) ˜ dx φ2 (x) = 1 n (2. z) (2.197) The expansion coeﬃcients Hn (y) in (2. through expansion (2. t)|2 = ˜n (x)|2 . t) = exp[−iω(n + 1 ) t] Nn φn (x) .195) zn Hn (y) n! ˜ φn (y) .194) = z2 n=0 ∞ 1 ∞ zn ˜ φn (y) n! (2. z) = Hn (y) (2. z) = n=0 1 2 /2 1 2 /2 = = mω e−iωt x (2. 2. through the expansion coeﬃcients φn (x) in (2. t) provide all stationary states of the quantum mechanical harmonic oscillator. 2 Here Nn are constants which normalize ψn (x.194).171) W (x. This follows from ∂n w(y. For this purpose we introduce ﬁrst closed expressions for φ the new variables y z and write (2. t) are associated with a time-independent probablity density |ψn (x. t)|2 = Nn +∞ −∞ n = 0. z) z 2 e−y or w(y. we have identiﬁed then. We will also argue that the functions ψn (x.

normalization factors for φ(y). ν+µ ν H3 (y) = 8y 3 − 12y.196). 1989). z) ∂z n = = z=0 2 ∂ n 2 y z − z2 e ∂z n e−(y−z) 2 ey z=0 2 ∂ n −(y−z)2 e ∂z n 2 z=0 2 (−1)n ey ∂n ∂y n = (−1)n ey z=0 ∂n ∂y n e−y (2. ∂y n (2. (2.38 Quantum Mechanical Path Integral which is a direct consequence of (2. For this purpose we expand the generating function (2. ν+µ (2. Reading. 2.201) ν−µ µ µ ν−µ Figure 2. Boston. 1990) is a useful introduction to this tool as is a chapter in the eminently useful Concrete Mathematics by R..Knuth. The identity (2.L. We want to derive now explicit expressions for the Hermite polynomials. m lattice (right diagram). that Hn (y) is a polynomial of degree n.Patashnik (AddisonWesley. H1 (y) = 2y. As will become evident in the present case generating functions provide an extremely elegant access to the special functions of Mathematical Physics3 .196) results in the so-called Rodrigues formula for the Hermite polynomials Hn (y) = (−1)n ey 2 ∂ n −y2 e . and O. .Wilf (Academic Press.e.196) to ˜ derive. µ lattice (left diagram) corresponds to a summation over only every other point in an n. z) the generating function for the Hermite polynomials. Massachusetts. . One calls w(y. .200) yields for the ﬁrst Hermite polynomials H0 (y) = 1.194..202) generatingfunctionology by H. and recursion equations for the eﬃcient evaluation of Hn (y). .199) Comparision with (2. ν H2 (y) = 4y 2 − 2. The diagrams illustrate that a summation over all points in a ν. among other properties.E.S.200) One can deduce from this expression the polynomial character of Hn (y).1: Schematic representation of change of summation variables ν and µ to n = ν + µ and m = ν −µ. i. We start from 2 e2yz − z ∞ ν = ν=0 µ=0 ∞ ν 1 ν! 1 ν ν µ ν µ z 2µ (−1)µ (2y)ν−µ z ν−µ (−1)µ (2y)ν−µ z ν+µ = ν=0 µ=0 3 (2. We will employ (2.196) ∂n w(y. The diagrams also identify the areas over which the summation is to be carried out.194) in a Taylor series in terms of y p z q and identify the corresponding coeﬃcient cpq with the coeﬃcient of the p–th power of y in Hq (y). Inc. closed expressions for Hn (y).198) for the Hermite polynomials can be expressed in a more convenient form employing deﬁnition (2.Graham. D.

0 ≤ k ≤ [n/2] where [x] denotes the largest integer p. for odd n. µ expressend in terms of n. (2. .e. From (2.209). One can write then using m = n − 2k e 2yz − z 2 ∞ = n=0 zn n! [n/2] k=0 n! (−1)k (2y)n−2k .2. 2 µ = n−m . m are ν = n+m . H2 (y) = 4y 2 − 2. k! (n − 2k)! = Hn (y) (2. m must be restricted such that either both n and m are even or both n and m are odd. H1 (y) = 2y.. (2.194) holds w(−y. 2 (2. Hence.207) one can deduce that Hn (y).205) Since (n − m)/2 is an integer we can introduce now the summation variable k = (n − m)/2 . With this restriction in mind one can express (2.208) which agrees with the expressions in (2. k! (n − 2k)! (2. the sum in (2. (2.207) contains only even powers. H3 (y) = 8y 3 − 12y. . −z) and. µ are integers the summation over n.210) from which one can conclude the property (2. According to (2.203) The old summation variables ν. 0 ≤ m ≤ n .206) From this expansion we can identify Hn (y) [n/2] Hn (y) = k=0 (−1)k n! (2y)n−2k . z) = w(y. 39 (2. m = ν − µ 0 ≤ n < ∞. hence.207) This expression yields for the ﬁrst four Hermite polynomials H0 (y) = 1.197) ∞ n=0 zn Hn (−y) = n! ∞ n=0 (−z)n Hn (y) = n! ∞ n=0 zn (−1)n Hn (y) n! (2.5: Propagator for a Quadratic Lagrangian and introduce now new summation variables n = ν + µ. .204) Since ν. it contains only odd powers. p ≤ x.209) This property follows also from the generating function. i.1. 2. it holds Hn (−y) = (−1)n Hn (y) .201). is a polynomial of degree n. according to 2. otherwise. in fact.202) e 2yz − z 2 ∞ = n=0 zn n! ≤n m≥0 n! (−1) n−m 2 n−m 2 ! m! (2y)m . . The lattices representing the summation terms are shown in Fig. In case of even n .

(2.214) We want to derive (??) using the generating function. (2.217) gives H1 (y) − 2y H0 (y) = 0.188. for example. 2. . as given in (2.215) ∂z which can be readily veriﬁed using (2.218) The reader should recognize the connection between the pattern of the diﬀerential equation (??) and the pattern of the recursion equation (??): a diﬀerential operator d/dz increases the order n of Hn by one. n! (2. 2. as deﬁned through (2. For this purpose one considers w(0. (2. . 1. have a node at y = 0.213) which allow one to evaluate Hn (y) from H0 (y) and H1 (y) given by (2. Starting point of the derivation is the property of w(y. 2. 2. . z) − (2y − 2z) w(y. indeed satisﬁes the latter relationship. (2. n = 1.194). z) ∂ w(y.215) yields ∞ n=1 z n−1 Hn (y) − 2y (n − 1)! ∞ n=0 zn Hn (y) + 2 n! ∞ n=0 z n+1 Hn (y) = 0 . .216) Combining the sums and collecting terms with identical powers of z ∞ n=1 zn Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) n! + H1 (y) − 2yH0 (y) = 0 (2. .40 Quantum Mechanical Path Integral The generating function allows one to determine the values of Hn (y) at y = 0.211) m! n! m=0 n=0 Comparing terms on both sides of the equation yields (2n)! . Substituting expansion (2. . H2n (0) = (−1)n Recursion Relationships A useful set of properties for special functions are the so-called recursion relationships.194).212) n! This implies that stationary states of the harmonic oscillator φ2n+1 (x). . . One can then readily state which diﬀerential equation of w(y. z) = 0 (2. dy n = 1.196) into the diﬀerential equation (2.208). . a factor z reduces the order of Hn by one and introduces also a factor n. . (2.209) since odd functions have a node at the origin. z) should be equivalent to the relationship (??). Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) = 0 . . Hn+1 (y) − 2y Hn (y) + 2n Hn−1 (y) = 0. a property which is consistent with (2. dw/dy − 2zw = 0. For Hermite polynomials holds. namely. H2n+1 (0) = 0 . . n = 1.233) below.197) above and given by (2. Another relationship is d Hn (y) = 2n Hn−1 (y). z) = exp(−z 2 ) and carries out the Taylor expansion on both sides of this expression resulting in ∞ ∞ (−1)m z 2m zn = Hn (0) . . 2. n = 0. z). The reader may verify that w(y.

5: Propagator for a Quadratic Lagrangian Integral Representation of Hermite Polynomials An integral representation of the Hermite polynomials can be derived starting from the integral +∞ 41 I(y) = −∞ dt e2iy t − t .222) Employing this expression now on the r. 1. .221) 1 = √ π +∞ −∞ dt e2iyt − t . (2. 2 (2.228) . z) w(y. through a double series over Hermite polynomials using (2. ﬁnally. of the Rodrigues formula (2. e−y 2 π e−y 2 (2.196) the orthogonality properties of the Hermite polynomials.194.194.s.h.226a).219) which can be written I(y) = e−y 2 +∞ −∞ dte−(t−iy) = e−y √ 2 2 +∞ −∞ dze−z . in the integral representation of the Hermite polynomials 2n (−i)n ey √ Hn (y) = π Orthonormality Properties 2 (2. 2.247) for a = i one obtains I(y) = and.220) Using (2.227) Comparing the terms of the expansions allows one to conclude the orthonormality conditions +∞ −∞ dy Hn (y) Hn (y) e−y 2 √ = 2n n! π δn n . acording to the deﬁnition (2. (2. dy n (2. 2 √ π e2 z z (2. z ) e−y 2 +∞ = e2 z z = √ ∞ dy e−(y−z−z ) = . 2.226) π −∞ 2n z n z n n=0 n! Expressing the l. . .224) +∞ −∞ dt tn e2iy t − t . 2 n = 0.h. 2.200) yields Hn (y) = The identity (−1)n y2 √ e π +∞ dt −∞ dn 2iyt − t2 e .225) We want to derive from the generating function (2.s. 2 (2.196) yields ∞ n.2. 2 (2.223) dn 2iy t − t2 2 e = (2 i t)n e2iy t − t n dy results. For this purpose we consider the integral +∞ −∞ dy w(y.n =0 +∞ −∞ dy Hn (y) Hn (y) e −y 2 zn z n = n! n ! ∞ n=0 √ zn z n 2n n! π n! n! (2.

197) holds 2 ˜ φn (y) = e−y /2 Hn (y) . 2. According to (2.190)] φn (y) = Nn e−y 2 /2 and for the normalization constants Nn follows from (2.229) Hn (y) . 2. 3. Normalized Stationary States The orthonormality conditions (2. 2. (2.228) +∞ 2 Nn −∞ √ 2 2 2 dy e−y Hn (y) = Nn 2n n! π = 1 1 2n n! 1 2n n! We conclude Nn = and can ﬁnally state the explicit form of the normalized stationary states φn (y) = −y √ e π 2 /2 The stationary states (2. 3.230) (2. negative for n = 2 and vanishes for n = 1. 4 and odd for n = 1. 4 in Fig. 4. 3. 2.189. Furthermore. 1. The normalized states are [c. ¡ ¢ ¡ -4 -2 0 2 4 y (2.232) Hn (y) . that the wave functions are even for n = 0.2.42 Quantum Mechanical Path Integral E 9/2 h ω ¥ ¥ ¥ ¥ ¥ 7/2 h ω 5/2 h ω ¤ 3/2 h ω £ 1/2 h ω ¦ Figure 2. the value of the wave function at y = 0 is positive for n = 0. (2. (2. in harmony with (??). One can recognize. 4.233) are presented for n = 0.228) allow us to construct normalized stationary states of the harmonic oscillator.2: Stationary states φn (y) of the harmonic oscillator for n = 0.231) √ π (2.f. 3. 2. 1. in agreement with our above discussions.233) . One can also recognize that n is equal to the number of nodes of the wave function.

???] and relativistic [see Sect. w(x) = 0 only at a discrete set of points xk ∈ Ω (2. denoted by P (x) and introduced in Sect. The various orthonogal polynomials diﬀer in the spaces Ω ⊂ R over which they are deﬁned and diﬀer in a weight function w(y) which enter in their orthonogality conditions. the ultra-spherical harmonics which o arise in n–dimensional Schr¨dinger equations and the associated Laguerre polynomials which arise o for the stationary quantum states of particles moving in a Coulomb potential.156.235) Completeness of the Hermite Polynomials The Hermite polynomials are the ﬁrst members of a large class of special functions which one encounters in the course of describing stationary quantum states for various potentials and in spaces of diﬀerent dimensions. 5. (5.239) .179] holds Ω = [−1. The latter are written for polynomials pn (x) in the general form dx pn (x) pm (x) w(x) = In δnm Ω (2. Comparision with (2. ??? and eq.236) where w(x) is a so-called weight function with the property w(x) ≥ 0. (2.237) and where In denotes some constants. denoted by Ln (x) and encountered in case of the stationary states of the nonrelativistic [see Sect. The Hermite polynomials are so-called orthonogal polynomials since they obey the conditions (2. Other examples of orthogonal polynomials are the Legendre and Jacobi polynomials which arise in solving three-dimensional stationary Schr¨dinger equations.e.e.233) and (2.. 1].f.2.237) for Ω = R. +∞[. (10.. holds Ω = [0.234) The explicit form of the stationary states of the harmonic oscillator in terms of the position variable x is then. and I = 2/(2 + 1).189) by the Jacobian dx/dy.151. i. (2. The orthogonal polynomials pn mentioned above have the important property that they form a complete basis in the space F of normalizable functions. f. 5 below [c. 10.189) φn (x) = √ 1 2n n! mω π 1 4 e− mωx2 2 Hn ( mω x) .236 . 5. In case of the associated Laguerre (α) polynomials. w(x) = xα e−x .231) of the wave functions diﬀers from that postulated in (2.228) shows that the orthonogality condition of the Hermite polynomials is in complience with (2. w(x) = √ exp(−x2 ). w(x) ≡ 1. In = Γ(n+α+1)/n! where Γ(z) is the so-called Gamma function.238) where the space is endowed with the scalar product (f |g) = Ω dx f (x) g(x) w(x) = < ∞ . 5. In case of the Legendre polynomials. by dx = dy mω 1 4 . 2. using (2. (2.228).10 and eq. Ω (2. of functions which obey dx f 2 (x) w(x) = < ∞ . and In = 2n n! π.150 .459] hydrogen atom.5: Propagator for a Quadratic Lagrangian 43 The normalization condition (2. i. g ∈ F .

i. . (2.188. t) = n=0 1 2 dn exp[−iω(n + )t] e−y /2 Hn (y) .244) For the state to be stationary |ψ(x. the states (2. If one replaces for all f ∈ F: f (x) → w(x) f (x) and. i.. t) is ∞ ψ(y. t) = n=0 cn (t) e−y 2 /2 Hn (y) . Since the Hermite polynomials form a complete basis for such states we can expand ∞ ψ(y.245) must be time-independent. ∞ n. hence.240) where cn = 1 In dx w(x) f (x) pn (x) .247) . ψ(y.197) it must hold cn (t) = dn exp[−iω(n + 1 )t] and.. t) must be identical to one of the stationary states (2. t)|2 . Appendix: Exponential Integral We want to prove +∞ +∞ I = dy1 . 2 (2. symmetric a. Therefore. n 2 (2. det(a) (2.e. 2. except for a single n = no . t) which is stationary under the action of the harmonic oscillator propagator (2.m=0 d∗ dm exp[iω(m − n)t] e−y Hn (y)Hm (y) .e. Ω (2. i.147). pn (x) → w(x) pn (x) the scalar product (2.243) To be consistent with(2.172) holds. the 2 stationary state ψ(y. −∞ dyn ei −∞ n j. (2.233)..246) for det(a) = 0 and real.242) and the existence of a normalizable state ψ(y.44 As a result holds for any f ∈ F f (x) = n Quantum Mechanical Path Integral cn pn (x) (2.242) Let us assume now the case of a function space governed by the norm (2.e. The only possibility for this to be true is dn = 0.233) exhaust all stationary states of the harmonic oscillator.e.239) becomes the conventional scalar product of quantum mechanics f |g = Ω dx f (x) g(x) .241) The latter identity follows from (2. aT = a.k yj ajk yk = (iπ)n . in particular.236). a state for which (2. In case of n = 1 this reads +∞ −∞ dx ei a x = 2 iπ . i. . a (2.

. yk = k Skj yj . .255) .254) ajj . det(S) a = (det(S))−1 det(a) det(S) = det(a) .249) The akk are the eigenvalues of a and are real. This property allows one to simplify the bilinear ˜ form n yj ajk yk by introducing new integration variables j. (2. .246) reads then in terms of yj ˜ n n j. .250) The bilinear form in (2. . according to (2. 0 0 . (2. . . i. ann ˜ where S can be chosen as an orthonormal transformation.k yj ajk yk ˜˜ ˜ (2. ST S = 1 1 or S = S−1 . . The proof of (2. . One can conclude det(a) = j=1 n (2.246) exploits that for any real.248.2.m (S T )jl alm Smk . ˜ (2. .. . ˜ (2.249) n ajk = ˜ l.248) .k n n yj = ˜ k (S −1 )jk yk .k yj ajk yk n = j.. 0 ˜ ˜ S−1 a S = a = . .252) ˜ For the determinant of a holds det(˜) = a n ajj ˜ j=1 (2.k n m n y Sj ajk Skm ym ˜ ˜ y (S T ) j ajk Skm ym ˜ ˜ j. .251) where. . 0 ˜ 0 a22 . symmetric matrix exists a similarity transformation such that a11 0 .k n m = = j.7: Appendix / Exponential Integral 45 which holds for a ∈ C as long as a = 0.253) as well as det(˜) = det(S−1 aS) = det(S−1 ) det(a) . 2. . (2. .e.

yn ) ∂(˜1 . . .250) holds J = S and. . yn ) y ˜ = ( det S )2 (2.. Accordingly. .. . .264) I = d˜1 . 2.250) allows one to express (2. . y −∞ d˜n det y −∞ ∂(y1 . . .265) k=1 −∞ which leaves us to determine integrals of the type +∞ dx eicx −∞ 2 (2. .262) (2. yn ) ∂(˜1 . hence. .46 We have assumed det(a) = 0. . . . (2.261) ∂(y1 . . . . . i.e. n (2. . .263) (2. d˜n e y −∞ = (2.258) where we introduced the Jacobian matrix J = with elements Jjs = According to (2. yn ) y ˜ ∂yj .266) .. yn ) y ˜ ei n k akk yk ˜ ˜2 . y −∞ +∞ d˜n ei y −∞ n k akk yk ˜ ˜2 +∞ i˜11 y1 a ˜2 n i˜nn yn a ˜2 +∞ a ˜ d˜k ei˜kk yk y 2 = d˜1 e y −∞ .256) such that none of the eigenvalues of a vanishes.250) +∞ +∞ I = d˜1 .257) Substitution of the integration variables (2. ∂(˜1 .249) follows 1 = det ST S such that one can conclude det S = ±1 One can right then (2.258) +∞ +∞ ∂(y1 . . . det( From (2. . ∂ ys ˜ (2. . ajj = 0 . .259) (2. yn ) ) = det(S) . . . holds n Quantum Mechanical Path Integral ajj = 0 ˜ j=1 (2. . ˜ for j = 1. .260) (2. . . .

268) vanishes. Jk = γk dz eicz 2 (2. 2 since eicz is a holomorphic function. for x.2.269) The contributions Jk can be expressed through integrals along a real coordinate axis by realizing that the paths γk can be parametrized by real coordinates x p γ1 : z = x γ2 : z = ix + p γ3 : z = √ ix J1 = −p p dx eicx 2 J2 = J3 = = i dx eic(ix+p) 0√ − 2p √ √ 2p 2 i dx eic( 2p √ ix)2 √ − i 0 √ (2. We consider ﬁrst the case c > 0 and discuss the case c < 0 further below.257) holds c = 0.267) by considering the contour integral J = γ dz eicz = 0 2 (2. One can relate integral (2.e.270) dx e−cx 2 2 − 2p √ γ4 : z = ix − p J4 = −p i dx eic(ix−p) ..268) along the path γ = γ1 + γ2 + γ3 + γ4 displayed in Figure 2.268) can be written as the sum of the following path integrals J = J1 + J2 + J3 + J4 . (2.271) . according to (2.7: Appendix / Exponential Integral 47 where. The contour integral (2.c > 0 . Substituting −x for x into integral J4 one obtains 0 J4 = p p (−i) dx eic(−ix+p) 2 = 0 i dx eic(ix−p) 2 = J2 . p ∈ IR.3.266) to the well-known Gaussian integral +∞ dx e−cx −∞ 2 = π . c (2. The contour intergral (2. i. the integrand does not exhibit any singularities anywhere in C.

This follows from the following calculation p p→+∞ lim |J2 or 4 | = p→+∞ lim | 0 p i dx eic(ix+p) | ≤ It holds |e ic(p2 −x2 ) p→+∞ 0 lim |i| dx |eic(p 2 −x2 ) | = 1 since the exponent of e is purely imaginary.48 Quantum Mechanical Path Integral Im(z) ip γ2 ¡ γ4 ¡ γ3 ¡ -i p Figure 2.268) for p = +∞. (2.272) = 0 . c ¡ -p γ1 p Re(z) 2 | |e−2cxp | . One can state accordingly ∞ J = dx e −∞ icx2 − √ ∞ i dx e−cx −∞ 2 Using 2. Hence. p p→+∞ lim |J2 or 4 | ≤ = p→+∞ 0 lim dx |e−2cxp | 1 − e−2cp 2cp = 0.3: Contour path γ in the complex plain. We will now show that the two integrals J2 and J4 vanish for p → +∞. (2.267) one has shown then ∞ dx eicx −∞ 2 = iπ . (2.275) .273) p→+∞ lim J2 and J4 do not contribute then to integral (2.274) (2.

277) We apply the above results (2.279) . It holds n I = k=1 iπ = akk ˜ (iπ)n .268). This leads to ∞ icx2 J = dx e −∞ + √ −∞ −i ∞ dx ecx = 0 2 (2. but with a path γ that is reﬂected at the real axis.277) to (2.275. c (2.7: Appendix / Exponential Integral 49 One can derive the same result for c < 0.2.255) this result can be expressed in terms of the matrix a I = which concludes our proof.276) and (c < 0) ∞ dx eicx −∞ 2 = −iπ = −|c| iπ . 2. (iπ)n det(a) (2. if one chooses the same contour integral as (2. n ˜ j=1 ajj (2.265).278) Noting (2.

50 Quantum Mechanical Path Integral .

x3 )T .s. t + |r0 . In the limit of very small it is suﬃcient to employ a single discretization interval in (2.s. t) . It holds then according to (2.4) − U (r. of this equation in terms of powers of and we will demonstrate that the terms of order require that ψ(r. t + ) = × exp i 2 +∞ +∞ +∞ −∞ dx1 −∞ dx2 −∞ dx3 m x2 + x2 + x2 1 2 3 m 2πi 3 2 × (3.1 Derivation of the Schr¨dinger Equation o We will consider now the propagation of a wave function ψ(r.20.h. Generalizing (2. t) . (3. t) ψ(r0 . t) by an inﬁnitesimal time step . t) . namely the Schr¨dinger equation. . t + ) = Ω m 2πi 3 2 exp i m (r − r0 )2 − 2 U (r. t) ψ(r0 . (3.1) can be expressed through the discretization scheme (2. x3 51 .21).2) d 3 r0 m 2πi 3 2 exp i m (r − r0 )2 − 2 U (r.1) We will expand the l. x2 . even in x1 .5) ψ(r. We will denote the components of s by (x1 .20) to evaluate the propagator. The propagator in (3. t + ) = Ω d3 r0 φ(r. and the r. Notable exceptions are non-stationary systems involving time-dependent linear and quadratic Lagrangians. 2. but by no means all.20) to R one obtains then for small φ(r. t) . t) ψ(r + s. x2 .3) In order to carry out the integration we set r0 = r + s and use s as the new integration variable.Chapter 3 The Schr¨dinger Equation o 3.h. t + |r0 . t) = From this follows ψ(r. the Schr¨dinger o o equation provides the simpler avenue towards describing quantum systems than the path ingral formulation of Section 2. For many situations. t) satisﬁes a partial diﬀerential equation. (3. This yields ψ(r.

.11) and. 1.4) one identiﬁes 2 1 = a m = O( ) (3. t) j .. t). x3 . the terms collected in (3. a (3. O( 7 2 ) . t) + . Since the kinetic energy contribution in (3. .. of the expansion of ψ(r + s. t) and the kinetic energy term..12) Here one needs to note that we are actually dealing with a three-fold integral. ∂xj ∂xk (3. n = 0.6) 3 4 ∂4 j=1 xj ∂x4 ψ(r. Since the latter contributes to (3. (3. x2 . O( 7 2 ). 1 2 3 2 ∂2 j=1 xj ∂x2 ψ(r. but over s = (x1 .8) (3. t) .10) It is now important to note that in case of integral (3. we need then to evaluate integrals of the type +∞ In (a) = −∞ dx x2n exp i a x2 . 2 (3. only terms of the expansion (3.7) According to (2. .7) shows 1 ∂ In (a).8) one can evaluate recursively all integrals In (a). e.g.9) iπ a (3. t) + ∂xj 2 3 xj xk j. a supposition which will be examined below. O( 5 2 ) .4) is even in all three coordinates x1 . (3.4) only for small x2 + x2 + x2 values 1 2 3 we expand 3 ψ(r + s. t) = ψ(r.k=1 ∂2 ψ(r. consequently. It is then suﬃcient to consider the terms ψ(r. The integration involves only the wave function ψ(r +s. t) + j=1 xj ∂ 1 ψ(r. 1 4 3 2 2 ∂4 j..k=1 xj xk ∂x2 ∂x2 ψ(r. t) j k . x3 )T . Obviously.6) make contributions of the order O( 3 2 ) . x2 . It holds In+1 (a) = I1 (a) = i 2a iπ . U (r.52 Schr¨dinger Equation o It is important to note that the integration is not over r. According to (3..36) holds I0 (a) = Inspection of (3.5) which are even separately in all three coordinates yield non-vanishing contributions.5) assuming that only the leading terms contribute.11) holds 3 m 3 iπ 2 2 × = 1 (3. a I2 (a) = − 3 4a2 iπ . t) is a constant with respect to this integration. t) j 1 12 . i ∂a Starting from (3.13) 2πi a .

a solution is thought for t ≥ t0 and the solution is speciﬁed at the intial time t0 . ψ(r. The 2nd order spatial derivatives require that one speciﬁes also properties of the solution on a closed boundary ∂Ω surrounding the volume Ω in which a solution is to be determined. t). t) + O( 2 ) . The following general remarks can be made about the solution. Obviously. one speciﬁes the so-called initial condition.14) This expansion in terms of powers of suggests that we also expand ∂ ψ(r. 1st order in time. t1 ) = f (r). t) + O( 2 ) . t) . t) (3.e.g. t) ψ(r. As we will discuss in Chapter 5 below the solutions of the Schr¨dinger equation are restricted to particular Hilbert spaces H which are o .15) ψ(r.18) This is the celebrated time-dependent Schr¨dinger equation. t) + 1 2i 4 m 2 53 ψ(r. t) . using (3. t) ψ(r. t) + O( 2 ) .3. e. This equation is often written in the o form ∂ ˆ i ψ(r. ψ(r.. In order O( ) the equation reads i ∂ ψ(r. The 1st order time derivative requires that for any solution a single temporal condition needs to be speciﬁed.19) ∂t where 2 2 ˆ H = − + U (r. t + ) = ψ(r. Due to its linear character any linear combination of solutions of the time-dependent Schr¨dinger o equation is also a solution. t) = 1 − (3. this equation is trivially satisﬁed to order O( 0 ). t)ψ(r. (3. We will derive brieﬂy the type of boundary conditions encountered. t + ) = exp − i U (r. t) = H ψ(r. t) = ψ(r.2: Boundary Conditions and one can conclude.10). 2nd o order in the spatial variables and linear in the solution ψ(r.16) Inserting this into (3. t) = ∂t 2 − 2 2m + U (r. Usually. t) − + i 2 2 i U (r.2 Boundary Conditions The time-dependent Schr¨dinger equation is a partial diﬀerential equation. t) (3.17) 2m ψ(r. (3.14) results in ψ(r. t) + and exp − i U (r. t) + ∂ ∂t ψ(r. i. (3. t) + O( 2 ) ∂t i U (r..20) 2m 3. (3.

that H is 2 .54 Schr¨dinger Equation o linear vector spaces of functions f (r) in which a scalar product between two elements f.20). The reader is advised to consult a reference text on ‘Linear Algebra’ to follow this argument. g|r) is called the ˆ concomitant of H and is P (f ∗ .23). hence. (3. 1975). Interchanging f ∗ (r) and g(r) results in g| H |f Ω = Ω ˆ d3 r g(r) Hf ∗ (r) . . for example. implies hermitian ˆ f | H |g 1 2 Ω ˆ = g| H |f Ω (3. in principle.28) See. Chapter 1..25) ˆ f | H |g Ω ˆ = g| H |f Ω + ∂Ω da · P (f ∗ .26) where ∂Ω da · A(r) denotes an integral over the surface ∂Ω of the volume Ω. diﬀer from each other.24) Using Green’s theorem1 Ωd 3r f ∗ (r) 2 g(r) − g(r) 2 f ∗ (r) = one obtains the identity ∂Ω da · ( f ∗ (r) g(r) − g(r) f ∗ (r) ) (3.22) ˆ where H is deﬁned in (3. g|r) = − 2 2m ( f ∗ (r) g(r) − g(r) f ∗ (r) ) (3. t) g(r) − Ω d3 rg(r) U (r.26) the vector–valued function P (f ∗ .23) ˆ Since H is a diﬀerential operator the expressions (3. Calssical Electrodynamics. f ∗ (r) d3 rg(r) Ω 2 = − 2m d3 rf ∗ (r) Ω 2 g(r) − f ∗ (r) (3. The hermitian property. The diﬀerence between the integrals is g| H |f = Ω Ω = − g| H |f d3 rf ∗ (r) − Ω 2 2 2m g(r) − Ω d3 rg(r) 2 − 2 2m f ∗ (r) + Ω 2 d3 rf ∗ (r) U (r. D. however. 2nd Ed. (John Wiley.21) Ω This leads one to consider the integral f | H |g Ω = Ω ˆ d3 rf ∗ (r) H g(r) (3. J. Since energy is a real ˆ ˆ quantity one needs to require that the eigenvalues of the operator H are real and.27) ˆ We will postulate below that H is an operator in H which represents energy.22) and (3. the surface elements da pointing out of the surface in a direction normal to the surface and the vector–valued function A(r) is taken at points r ∈ ∂Ω. New York. In (3. g|r) (3. g ∈ H is deﬁned as follows f |g Ω = d3 rf ∗ (r)g(r) (3. t). Jackson.

. In this case one can postulate both conditions (3. in which case ∂Ω1 is the inner sphere and ∂Ω2 is the outer sphere. t)∂t ψ(r. A most common boundary condition is encountered for the volume Ω = R3 in which case one postulates lim f (r) = 0 “natural boundary condition” .3: Particle Flux 55 and.32) the wave function must decay for |r| → ∞ rapidly enough to be square integrable. Often the closed surface of a volume ∂Ω is the union of disconnected surfaces3 ∂Ωj .30) each condition holding on an entire surface ∂Ωj . 3. t) on a single connected surface ∂Ωj only either one of the conditions (3.e.3 Particle Flux and Schr¨dinger Equation o The solution of the Schr¨dinger equation is the wave function ψ(r. It must hold then for all f ∈ H f (r) = 0 ∀r.30) can be postulated. t) which describes the state of o a particle moving in the potential U (r. In this case one can expect that the particle density is localized in the area where the energy of the particle exceeds the potential eenrgy. to avoid discontinuities in ψ(r.g. i. t)|2 .3. (3. 3. . However. like exp(−κr). r ∈ ∂Ω or da · f (r) = 0 ∀r.. t) ] .. t) = ω 3 d3 r [ ψ ∗ (r. |ψ(r. i. i. Since the total probability of ﬁnding the particle anywhere in space is d3 r|f (r)|2 = 1 (3. . namely. g|r) vanish on ∂Ω.33) The time derivative of p(ω.29) In fact..31) usually arises when a particle existing in a bound state is described.. and that the density decays rapidly when one moves away from that area. In either case does f (r) and all of its derivatives vanish asymptotically. The latter property stems from the fact that the boundary condition (3. ∂Ω = ∂Ω1 ∩ ∂Ω2 ∩ ∂Ω3 ∩ . The observable directly linked to the wave function is the probability to ﬁnd the particle at position r at time t. α > 2. in this case also all derivatives of f (r) vanish at inﬁnity. r ∈ ∂Ω (3. (3.31) |r|→∞ (3. t)∂t ψ ∗ (r. t) + ψ(r. therefore. t). e.e. t) is ∂t p(ω. t)|2 .29. 3. we can only allow functions which make the diﬀerential da · P (f ∗ .32). if f and g satisfy these conditions than also does any linear combination αf + βg.34) An example is a volume between two concentric spheres.29.30) Note that these boundary conditions are linear in f . t) = ω d3 r |ψ(r. The probability to observe the particle anywhere in the subvolume ω ⊂ Ω is p(ω. (3. κ > 0 or like r−α .e. obey (3.

t)|2 is a probability density with units 1/volume. t) = 0 (3. (3.27) this can be written i ∂t p(ω.38) holds. ψ|r. One can multiply the solution of (3.40) One can rewrite then (3. (3. t) ψ ∗ (r.18) by any complex number and accordingly one can deﬁne ψ(r. t) − ψ(r.44) ˆ Note that the Hamiltonian H involves only real terms. (3. We will assume in the remainder of this Section that the solutions discussed are normalized. t) = [ ψ(r. t) + 4 · j(r. . t) H ψ(r.37) can be expressed through a volume integral according to da · A(r) = ∂ω ω (3. The surface integral (3. t) ψ(r.41) where j(r. t) + ω · j(r. t) .37) d3 r ∂t ρ(r. t) = |ψ(r.27) one can express this j(r. t) (3. t) .26. Accordingly the probability to observe the particle anywhere in the total volume Ω is constant. Note that for a normalized wave function the quantity ρ(r. t) = ∂ω da · P (ψ ∗ (r. 3. t) = 0 .30) ∂t p(Ω. A natural choice for this constant is 1.43) (3.42) 2mi Since (3.56 Using (3. t) H ψ ∗ (r. 3. t)r. t) − ψ ∗ (r.19) and its conjugate complex4 −i yields i ∂t p(ω. t) = P (ψ ∗ . t) ] .37) If one applies this identity to ω = Ω one obtains according to (3.39) d3 r · A(r) (3.35) d3 r . t) = ω Schr¨dinger Equation o ∂ ∗ ˆ ψ (r. t). t) such that d3 r|ψ(r. One refers to such solution as normalized. t) = 0. Using (3. ψ(r.36) According to (3. (3.41) holds for any volume ω ⊂ Ω one can conclude ∂t ρ(r. t)|2 = 1 Ω (3. t) = H ψ ∗ (r. t) ∂t ˆ ˆ ψ ∗ (r.29.

t) points in the direction to the outside of volume ω. (3.148. t) has vanishing ﬂux anywhere. 2 we had demonstrated that a factor exp(ipo xo / ) induces a motion of 1-dimensional wave packets such that po /m corresponds to the initial velocity. t0 ) = [2π]− 2 Ω∞ 3 ˜ d3 k φ(k) exp i(k · r − ωt0 ) . j(r. t) = [2π]− 2 where the dispersion relationship holds ω = k2 .46) i.48. (2. 2m (3. (2.50) 3 ˜ d3 k φ(k) exp i(k · r − ωt) Ω∞ (3.49) reads at t = t0 ψ(r. the initial condition at ψ(r.4: Free Particle 57 The interpretation of j(r. One can readily show by insertion into (3.48) that the general solution is of the form ψ(r.f. (2. (3.f. t) = 0 2 ∂ 2 i ψ(r. and for particles moving in a linear [c.45) Obviously. equal to the velocity of the particle..51) . Such case arose in Sect. m (3.41) written in the form ∂t ω d3 r ρ(r.167)] potential.47) .71)]. 3.f. Note that j(r. 2. This ﬁnding is consistent with the present evaluation of the particle ﬂux: a factor exp(ipo xo / ) gives rise to a ﬂux po /m. t) gives rise to a descrease of the total probability in volume ω due to the disappearence of probability density at the surface ∂ω. The generalization to three dimenisons implies then that the factor exp(i k · r) corresponds to an intial velocit k/m and a ﬂux of the same magnitude.. for f (r) ∈ R.3. t0 ) determines φ(k). t) . This follows directly from an inspection of Eq.49) ˜ Obviously. (3.48) ∂t 2m which describes the motion of free particles. t) (3. It is of interest to note from (3. t) is that of density ﬂux. Equation (3. 2. t) = − ψ(r.4 Solution of the Free Particle Schr¨dinger Equation o We want to consider now solutions of the Schr¨dinger equation (3. One often encounters wave functions of the type φ(r) = f (r) ei k·r The corresponding ﬂux is j(r) = k 2 f (r) . In Sect. 2.105.43) that any real wave function ψ(r.e.e. (3.18) in Ω∞ = R in the case o U (r. 2 for a free particle [c. i. t) = − ∂ω da · j(r. arises solely from the complex factor exp(i k · r).125)] and in a quadratic [c.

t|r0 .52) above ψ(r.49) is square integrable at all subsequent times and.54) valid for this case.51. 3. t|r0 . evaluating the integral in (3.70) derived below for a particle in a box and the harmonic oscillator. t0 ) (3. in (2.53) where φ(r. Below we will generalize the propagator (3.54) to the case of non-vanishing potentials U (r).58 The inverse Fourier transform yields 3 ˜ φ(k) = [2π]− 2 Schr¨dinger Equation o d3 r0 exp(−ik · r0 ) ψ(r0 .54) yields (2. t0 ) .47). 3. t0 ) ψ(r0 .e.247). hence. The general form for this propagator involves an expansion in terms of a complete set of eigenfunctions as in (3. t) as given by (3.54) an alternative representation of the propagator (2. (t − t0 ) (3. (3.52) We have not speciﬁed the spatial boundary condition in case of (3.. i.e. that the “natural boundary condition” (3. respectively.49. t0 ) = 1 2π 3 d3 k exp Ω∞ ik · (r − r0 ) − i 2 k2 2m (t − t0 ) .81) .114) and (4.51) to be square integrable it follows according to the properties of the Fourier–transform that ψ(r. t) = Ω∞ d3 r0 φ(r. To show this one needs to note 1 2π = +∞ dk1 exp −∞ ik1 (x − x0 ) − 1 2 i 2 k2 1 2m .49). We have identiﬁed then with (3.56) and using (2. The ensuing solutions are called wave packets.31) applies. i. In case of the harmonoc oscillator the expansion can be stated in a closed form given in (4.47).54) This expression obviously has the same form as postulated in the path integral formulation of Quantum Mechanics introduced above.55) m 2πi (t − t0 ) exp im (x − x0 )2 2 t − t0 This follows from completion of the square ik1 (x − x0 ) − (t − t0 ) 2m (t − to ) m x − xo = −i k1 − 2m t − to i 2 k2 1 2 + i m (x − xo )2 2 t − to (3. If one chooses the initial state f (r) deﬁned in (3. Ω∞ (3. derive an expression similar to (3. The solution as stated is deﬁned in the inﬁnte space Ω∞ = R . In fact.5). Comparision with Path Integral Formulation One can write solution (3..

r ∈ ∂Ω (3. (3.f.62) We will restrict the space of allowed solutions to a volume Ω such that the functions also make the concomitant (3. t) 2 k2 (3. t|x0 . t0 ) ψ(x0 .58) Integration over x0 leads to the integral +∞ dx0 exp −ikx0 + −∞ i po x0 − x2 0 2δ 2 = √ 2πδ 2 exp − (k − po 2 2 ) δ 2 (3. (3. t) is given by (??). t0 ) = 1 2π +∞ dk exp −∞ ik (x − x0 ) − i 2m (t − t0 ) . t|x0 .20) which are of o the form ψ(r.60) =???? Combining (3.57) where φ(x. the functions obey boundary conditions of the type (3. 3.61) × exp − (x − po m t)2 2δ 2 (1 + 2 t2 ) m2 δ 4 (1 − i . The 1-dimensional version of (3. Stationary States We consider now solutions of the time-dependent Schr¨dinger equation (3.3.56)].27) vanish on the surface ∂Ω of Ω. t) = −∞ dx0 φ(x. r ∈ ∂Ω or da · φ(r) = 0 ∀r.. Accordingly. We have demonstrated then in this case that the Schr¨dinger formulation of Quantum o Mechanics is equivalent to the Feynman path integral formulation.59) which is solved through completion of the square in the exponent [c.52) to the case that the initial state of a 1-dimensional free particle ψ(x0 . a result which is identical to the expression (??) obtained by means of the path integral propagator (2.63) . (3.55. t) = f (t) φ(r) .49. The remaining integration over k leads to the integral +∞ −∞ dk exp ikx − 1 2 (k − po 2 2 ) δ − i k2 m (t − t0 ) (3. 3.64) (3.53.57–3.29.e. 3. 3.4: Free Particle Free Particle at Rest 59 We want to apply solution (3. i.19. the boundary conditions are φ(r) = 0 ∀r.54) is +∞ ψ(x.46). (3. t) = t 1 − i mδ2 t 1 + i mδ2 1 4 1 πδ 2 (1 + 2 t2 m2 δ 4 1 4 ) × t po i p2 o ) + i x − t mδ 2 2m (3.60) yields with t0 = 0 ψ(x.30).

(3.65) is constant. We must postulate therefore ∂t f (t) ˆ H φ(r) = E f (t) = E φ(r) . according to (3. n = 1.e. i.62) we insert it into (3. One calls such states stationary states. (3.45) that the total probability d3 r ρ(r. (3.62) exists.62) which consist of two factors. one factor depending only on the time variable and the other only on the space variables are called separable in space and time. 2. It follows from the observation that for the solution space considered (3.69) If these two equations can be solved simultaneously a solution of the type (3. We will encounter many such problems in the subsequent Sections. t)|2 = |f (t)|2 Ω d3 r |φ(r)|2 (3. . solutions exist only for a set of discrete E ˆ values. We want to demonstrate this property now. It is important to realize that the separable solutions (3. t) = f (0) exp − i E φE (r) where ˆ H φE (r) = E φE (r) .63.. . At this point we will accept that solutions φ(r) of the type (3. At this point we state without proof that. This can hold only if |f (t)| is time-independent. As pointed out above.71) . often only for a discrete set of values En .68) can hold only for all t and all r if g1 (t) = E g2 (t) and E h1 (r) = h2 (r) for some E ∈ C. α ∈ R . This yields an expression g1 (t) h1 (r) = g2 (t) h2 (r) (3.e. the eigenvalues of the operator H. in general. by no means all solutions are of this type.. (3. We have then shown that ψ(r. hence. We may note in passing that solutions of the type (3.19). a common case is Ω = Ω∞ and the ‘natural boundary condition’ (3. the solutions o (3.62) are special solutions of the timedependent Schr¨dinger equation.31). One can conclude that the probability density for the state (3. t) vanishes on the surface of ∂Ω.67) (3. It turns out that a solution for f (t) can be found for any E. t)|2 = |φ(r)|2 . g2 (t) = f (t). if f (t) = eiα . We will demonstrate that solutions of the type (3. In fact. is time-independent.70) The task of ﬁnding solutions φ(r) which solve (3. namely f (t) = f (0) exp − i Et .62) have the particular property that the associated probability distributions are independent of time. It follows then from (3. t) = Ω Ω d3 r |ψ(r. h1 (r) = φ(r).60 Schr¨dinger Equation o and aﬀect only the spatial wave function φ(r).42) the ﬂux j(r.62) is |ψ(r. The identity (3. . In order to further characterize the solution (3.68) ˆ where g1 (t) = i ∂t f (t). for the eigenvalue problems in the conﬁned function space.69) exist.62) do exist and we will characterize the two factors of the solution f (t) and φ(r).69) is called an eigenvalue problem. 3.66) i.27) holds and. i. however.e..64). for functions required to obey boundary conditions (3. and h2 (r) = H φ(r). We denote the corresponding solution by φE (r).

hence. should be positive. m (3.71) are. e.e.72) According to (3. We start our proof using the property (3.. − 2 2m φ(r) = E φ(r) . The solution φE (r) corresponding to the eigenvalue problem posed by (3. 3. E (3. The resulting total energy values E are positive.4: Free Particle 61 is a solution of the time-dependent Schr¨dinger equation (3.g.74) The classical free particle with constant energy E > 0 moves without bounds in the space Ω∞ .75) 2m One can ascertain this statement by inserting the expression for φk (r) into the eigenvalue problem posed in (3. The ﬂux corresponding to (3.19. ˆ d3 r φ∗ (r) H φE (r) = E Ω Ω ˆ d3 r φE (r) H φ∗ (r) . We will show in Section 5 that E can be interprerted as the total energy of a stationary state.66) E must be real. E (3.43) is j(r. We want to prove now that the eigenvalues E which arise in the eigenvalue problem (3.76) has kinetic energy 2 k 2 /2m.76) according to (3. .28) for the special case that f and g in (3.28) both represent the state φE (r). of course.74) is actually best labelled by an index k. real. E ∈ R. a property which is to be expected since the energy is purely kinetic energy which. in fact. one can interpret then k as the magnitude of the momentum of the particle. t) = exp − i 2 k2 2m t exp i k · r (3.77) Noting that k can be interpreted as the magnitude of the momentum of the particle the ﬂux is equal to the velocity of the particle v = k/m multiplied by |N |2 . o According to (3.71) this yields E Ω d3 r φ∗ (r) φE (r) = E ∗ E Ω d3 r φE (r) φ∗ (r) . i. We will wave this assumption as we always need to do later whenever we deal with unbound particles. Stationary State of a Free Particle We consider now the stationary state of a free particle described by i ψ(r. = E. Obviously. As a result we cannot postulate in the present case that wave functions are localized and normalizable. t) = exp − E t 2 φE (r) . The corresponding stationary solution ψ(r.74) using exp i k · r = ik exp i k · r .20). k ∈ R 2 k2 φk (r) = N exp i k · r .73) from which follows E = E ∗ and. (3. (3. particles scattered of a potential.3. t) = |N |2 k .

. that the solutions obey this symmetry as well. a] ⊂ R which vanish on the surface ∂Ω1 = {−a. 2m dx2 2 (± ) = } (3.83) The solutions (3. (o) 2 k2 (e) 2 k2 2m = E (3. 3.79) The stationary solutions have the form ψ(x.84) . The boundary conditions which according to (3.82) satisfy the diﬀerential equation in (3. t) = exp(−iEt/ )φE (x) where φE (x) is determined by 2 d2 − φE (x) = E φE (x) . two types of solutions.81.80) 2m dx2 We note that the box is symmetric with respect to the origin. We can expect. (3.78) (3. a] ⊂ R assuming that the potential outside of this interval is inﬁnite.. . We assume. In case of the even solutions (3.5 Particle in One-Dimensional Box As an example of a situation in which only bound states exist in a quantum system we consider the stationary states of a particle conﬁned to a one-dimensional interval [−a. a] ⊂ R → R. a}. (3. f continuous. therefore. and so-called odd solutions obeying φ(x) = −φ(−x) φE (x) = A sinkx.62 Schr¨dinger Equation o 3. 3. 5 . . φ(±a) = 0 . t) .82) One can readily verify that (3. Setting up the Space F1 of Proper Spatial Functions The presence of the inﬁnite energy wall is accounted for by restricting the spatial dependence of the solutions to functions f (x) deﬁned in the domain Ω1 = [−a.e. It turns out that this boundary condition can only be satisﬁed for a discrete set of k–values kn . Hence. f ∈ F1 = {f : [−a.o) (−a) = 0 (3.80) need to be satisﬁed are φE (e. let say the one at x = a. i. (3. We will refer to this as a particle in a one-dimensional ‘box’.o) (a) = 0 and φE (e.81) 2m = E. 3. t) = − d2 ψ(x. n ∈ N.82) have the property that either both boundary conditions are satisﬁed or none. so-called even solutions obeying φ(x) = φ(−x) φE (x) = A coskx . 2a n = 1.80).81.81) they are kn = nπ . Solutions of the Schr¨dinger Equation in F1 o The time–dependent solutions satisfy i ∂t ψ(x. hence. we have to consider only one boundary condition.

a n = 2.82) only the kn –values cos(kn a) = cos kn = nπ .81.1). .o) (a. 6 . .91) holds. .84. 6 . (3.e. 2a n = 1. . . 6 . are En = 2π2 8ma2 n2 . (3.92) and for the odd states +a |An |2 −a dx sin2 nπx = |An |2 a = 1 .87) (Note that. . i. n = 2. n = 1.89) nπx . respectively. (3. according to (3. 5 . 4. (3. It is desirable to normalize the wave functions such that +a −a dx |φ(e.. (3. = 0.88) where the energies for odd (even) n–values correspond to the even (odd) solutions given in (3. 2a n = 1. 3.90) 2a The wave functions represent stationary states of the particle in a one-dimensional box. 3. (3.5: Particle in One-Dimensional Box since for such kn 63 nπa nπ = cos 2a 2 In case of the odd solutions (3. n is assumed to be even. x) = An sin n nπx .86) satisfy the boundary condition since for such kn nπa sin(kn a) = sin 2a = sin nπ 2 = 0. The wave functions for the ﬁve lowest energies En are presented in Fig. Notice that the number of nodes of the wave functions increase by one in going from one state to the state with the next higher energy En .) The Energy Spectrum and Stationary State Wave Functions The energy values corresponding to the kn –values in (3. (3. x) = An cos n and φ(o) (a. .82). (3.85) (3. By counting the number of their nodes one can determine the energy ordering of the wave functions. φ(e) (a. 4.94) . . 4. . 5 . x)|2 = 1 n (3. .93) The normalization constants are then An = (3. 3.3.86). . 3. according to the dispersion relationships given in (3. .86). This condition implies for the even states +a |An |2 −a dx cos2 nπx = |An |2 a = 1 .81) and (3. . 3 . 2a n = 2. 2a 1 . 2.82).

64 Schr¨dinger Equation o E h2 ] [ 8 m a2 25 ¢ 16 9 4 Figure 3. 2.o) (a. 3.1: Eigenvalues En and eigenfunctions φn (e. x) for n = 1. 5 of particle in a box. 4. ¡ 1 -a a x .

.e.3. Similarly.103) We will show in Section 5 that the property of a scalar product do indeed apply.95) 1 a cos nπx 2a sin nπx 2a for n = 1. Because of the property (3. i. for n = 2. In case of n. n = 1. (3. too. the integral arises φn |φm 1 a Ω1 = 1 a +a nπx −a dx cos 2a cos mπx = 2a (3.e.99) Since in this case the integrand is a product of an even and of an odd function.} where φn (a.100) +a −a dx cos (n−m)πx + cos (n+m)πx 2a 2a The periods of the two cos-functions in the interval [−a. . a] are N.94) B1 = {φn (a. m even φn |φm 1 a Ω1 = 1 a +a nπx −a dx sin 2a sin mπx = 2a (3.97) form an orthonormal basis set. In particular. the integral vanishes.98) the elements of B1 must be linearly independent. it holds φn |φm Ω1 = δn m . Obviously. and (iii) the mixed case. m odd. the integrals vanish. g ∈ F1 (3. one obtains for n. . the integrand is odd. i. n = m. m both odd.89.. 3. 6 . f. this integral vanishes. In case of n = m we have to consider three cases.98) ∞ f |f 5 Ω1 = n=1 d2 . 4. In fact. for ∞ f (x) = n=1 dn φn (a.101) +a −a dx cos (n−m)πx − cos (n+m)πx 2a 2a and.5: Particle in One-Dimensional Box The Stationary States form a Complete Orthonormal Basis of F1 We want to demonstrate now that the set of solutions (3. 3.90. 3.96) f |g Ω1 = −a dxf (x) g(x) . . (ii) n. x) . Hence we need to consider only the ﬁrst two cases. (3. n (3. 3. N ≥ 1. . x) (3. 2. . x) = together with the scalar product5 +a 65 (3. (i) n. 5 .102) holds according to (3. . it holds: f |f Ω1 = 0 → f (x) ≡ 0. 2a 2a (3.. m both even. .98) The latter property is obviously true for n = m. The latter case leads to integrals φn |φm Ω1 = 1 a +a dx cos −a nπx mπx sin . hence.

3. 4. . n = 1.106) and generalizing to t ≥ t0 one can write ∞ +a ψ(x. and bn .95. Accordingly.e. a] yields then also an expansion for any element in F1 and B1 is a complete basis for F1 . (3.66 Schr¨dinger Equation o f (x) ≡ 0 implies f |f Ω1 = 0 which in turn implies dn = 0 since d2 ≥ 0.106) Using the orthonormality property (3. n = 0. the coeﬃcients an . .109) Insertion of (3.107) Inserting this into (3. i.e. x0 ) ψ(x0 . t) at times t > t0 . 3.78) can be expressed as a linear combination of the elements of B1 deﬁned in (3. 2. . Restricting the expansion (3.95) is an orthonormal basis. x0 ) ψ(x0 . t0 ) = n=1 dn φn (a. It follows that B1 n deﬁned in (3.95) is also complete. they can be expanded in terms of a Fourier series. . 3.105) f an cos 2a 2a n=0 n=1 ˜ The functions f corresponding to the functions in the space F1 have zeros at x = ±m a. x) . x) cn (t) −a dx0 φn (a.. . ( ) = ([ + ] − ) (3.98) one obtains +a dx0 φm (a. there exist real constants {an . t0 ) = +a ∞ i n=1 − En φn (a. We like to show ﬁnally that the basis (3. . The functions f are periodic with period 2a. 2. however. For this purpose we extend the deﬁnition of the elements of F1 to the whole real axis through ˜ f : R → R. such demonstration can be based on the theory of Fourier series.110) . Evaluating the Propagator We can now use the expansion of any initial wave function ψ(x.79) requiring the o initial condition cn (t0 ) = 1 . n = 1. Demonstration of completeness is a formidable task. 1. .108) into the Schr¨dinger equation yields o +a ∞ n=1 φn (a. x) cn (t) −a dx0 φn (a. n = 2. For this purpose we expand ∞ ψ(x. m = 1. . t0 ) (3. Hence. in (3.104) ˜ where [y]a = y mod2a. 5 . −a (3. (3. t) = n=1 φn (a. This implies that only the trigonometric functions which are elements of B1 enter into the Fourier ˜ series. . x) ∂t cn (t) −a dx0 φn (a.108) where the functions cn (t) are to be determined from the Schr¨dinger equation (3. x0 ) ψ(x0 . . t0 ) in terms of eigenfunctions φn (a.96). .} and {bn ..} such that ∞ ∞ nπx nπx ˜ = + an sin (3. i. t0 ) . . any element of the function space F1 deﬁned in (3. In the present case. t0 ) = dm . We have then shown that any f corresponding to elements of F1 can be expanded in terms of elements in B1 . . x0 ) ψ(x0 .105) to the interval [−a.105) must vanish. x) to obtain an expression for ψ(x. (3. .

i. k0 = . but the collision with the left wall leads to new interference eﬀects. t) for any initial condition ψ(x. Often the propagator φ(x.108. t0 |x0 . x0 ) .. One can recognize that the particle moves ﬁrst to the left and that near the right wall of the box interference eﬀects develop.78).3.98).5). t|x0 . In case that diﬀerent boundary conditions hold the propagator will be diﬀerent as well. t0 ) itself is a solution of the time-dependent Schr¨dinger o equation (3. t0 ) = δ(x − x0 ) as can be readily veriﬁed using (3. We have identiﬁed then with (3. This solution can be written +a ψ(x. t|x0 . cm (t0 ) = 1 . according to (3.112) determine now ψ(x. a] yields. The last frame shows the wave front reaching again the right wall and the onset of new interference. 67 (3. The respective initial condition is φ(x.114) This expression has the same form as postulated in the path integral formulation of Quantum Mechanics introduced above. It is of interest to note that φ(x. rather than by an integral (3. Example of a Non-Stationary State As an illustration of a non-stationary state we consider a particle in an initial state ψ(x0 .115) This initial state corresponds to the particle being localized initially near x0 = 0 with a velocity v0 = 15 /m a in the direction of the positive x-axis. . 4 a (3. (3. t) = −a dx0 φ(x.79) which lies in the proper function space (3. The particle moves then to the left. being reﬂected at the right wall. x) and integrating over [−a.e. t0 ) = 1 2πσ 2 1 4 exp − x2 0 + i k0 x0 2σ 2 . the elements of which satisfy the appropriate boundary conditions.2 presents the probability distribution of the particle at subsequent times.54) as in the case of the free particle system which does not exhibit any bound states. σ = a 15 . t0 ) = ∞ φn (a.112) Equations (3.5: Particle in One-Dimensional Box Multiplying both sides by φm (a. t0 ) (3.109) are cm (t) = exp i − Em (t − t0 ) .113). 3. i ∂t cm (t) = − Em cm (t) . In the present system which is composed solely of bound states the propagator is given by a sum. in (2. The interference pattern begins to ‘smear out’ ﬁrst. t|x0 . t|x0 .111) The solutions of these equations which satisfy (3. t0 ) is also referred to as a Greens function. Note that the propagator has been evaluated in terms of elements of a particular function space F1 . t0 ) ψ(x0 . Figure 3. (3.113) where φ(x. x) exp n=1 i − En (t − t0 ) φn (a.114) the representation of the propagator for a particle in a box with inﬁnite walls. t0 ).

2: Stroboscopic views of the probability distribution |ψ(x.68 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 1.00 m a h2 |Ψ|2 2 t = 0.92 m a h2 |Ψ|2 2 t = 2.16 m a h2 1 πσ 1 πσ -a a -a a Figure 3.48 m a h2 |Ψ|2 2 t = 0. .96 m a h2 |Ψ|2 2 t = 1.68 Schr¨dinger Equation o |Ψ|2 2 t = 0. t)|2 for a particle in a box starting in a Gaussian distribution with momentum 15 /a.72 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 0.24 m a h2 1 πσ k = 15 a 1 πσ -a a -a a |Ψ|2 2 t = 0.20 m a h2 1 πσ 1 πσ -a a -a a |Ψ|2 2 t = 1.44 m a h2 |Ψ|2 2 t = 1.

e.95). A complete basis of F1 is given by the inﬁnite set B1 (3. x3 ) . 3. x3 ) (3. (3.6 Particle in Three-Dimensional Box We consider now a particle moving in a three-dimensional rectangular box with side lengths 2a1 . An important property of this basis and. t) = exp − E t φE (x1 . x2 . ) = ∀( .117) 2m 2 2 2 ∂1 + ∂2 + ∂3 (3.. x3 . x1 = ±a1 } ∪ {(x1 . x2 . The description employed a space of functions F1 deﬁned in (3. a1 ] ⊗ [−a2 . j = 1.114) in terms of which the solutions for all initial conditions can be expressed. x3 . We have deﬁned in the space F1 a scalar product (3. ( . We seek then solutions of the time-dependent Schr¨dinger equation o ˆ i ∂t ψ(x. φ(j) (±aj ) = 0 . x2 . The corresponding solutions have the form i ψ(x1 . 2a3 . (3. ˆ = O(x1 ) + O(x2 ) + O(x3 ). x2 .121) where − 2 2m 2 ∂j φ(j) (xj ) = Ej φ(j) (xj ) . x3 ) = j=1 φ(j) (xj ) (3. x3 ) = E φE (x1 . can be counted.e. i. t) . 2a2 .6: Particle in Three-Dimensional Box Summary: Particle in One-Dimensional Box 69 We like to summarize our description of the particle in the one-dimensional box from a point of view which will be elaborated further in Section 5. . a2 ] ⊗ [−a3 . hence. x3 )T ∈ Ω. x3 ) is an element of the function space F3 deﬁned in (3. x3 –axes with the edges of the box yields spatial boundary conditions which are obeyed by the elements of the function space F3 = {f : Ω → R.122) .3. 3 (3. )T ∈ ∂ } . x2 .116) where Ω is the interior of the box and ∂Ω its surface Ω ∂Ω = [−a1 . 2. x3 = ±a3 } . a3 ] ⊂ R = {(x1 .120) ˆ Since the Hamiltonian H is a sum of operators O(xj ) each dependent only on a single variable. of the space F1 is that the elements of the basis set can be enumerated by integer numbers. x2 . i. x3 )T ∈ Ω.. x2 .118) which are stationary states. . Placing the origin at the center and aligning the x1 .78).97) with respect to which the eigenfunctions are orthonormal.119) where φE (x1 . f continuous. x2 . x2 = ±a2 } ∪ {(x1 . x2 . t) = H ψ(x1 . x2 . one can express H 3 φ(x1 . x2 .116) and obeys the partial diﬀerential equation ˆ H φE (x1 . ˆ H = − 2 (3. This property allowed us to evaluate the propagator (3. x3 )T ∈ Ω.

For example. (3. x2 .n2 . if all three lengths a1 .e. r0 ) . . a2 . the solutions of (3. 3. 2. 2. n1 . This additional symmetry is also reﬂected by degeneracies in the energy levels. x3 ) = 1. a2 .n3 ) (a1 .124) The same considerations as in the one-dimensional case allow one to show that B3 = {φ(n1 . .. . x3 ) . n2 .126) Symmetries The three-dimensional box conﬁning a particle allows three symmetry operations that leave the box unchanged. n3 = 1. 3. (3. x3 ) n1 . 3. Such degeneracies are always a signature of an underlying symmetry. x1 ) φn2 (a2 . x2 ) φn3 (a3 .122) are given by (3. Actually. n2 . The energies and corresponding degeneracies of the particle in the three-dimensional box with all side lengths equal to 2a are given in the following Table: n1 1 1 1 1 2 1 2 1 2 3 1 n2 1 1 2 1 2 2 2 1 3 3 1 n3 1 2 2 3 2 3 3 4 3 3 5 E/[ 2 π 2 /8ma2 ] 3 6 9 11 12 14 17 18 22 27 27 degeneracy single three-fold three-fold three-fold single six-fold three-fold three-fold three-fold single three-fold One can readily verify that the symmetry of the box leads to three-fold and six-fold degeneracies. a2 .n2 .n3 ) (a1 . a3 . t0 ) = n1 .n3 ) (a1 . . a1 = a2 = a3 = a then rotation around the x1 .120) can be written φ(n1 . x1 . x2 . x1 .n2 . n 2 .70 Schr¨dinger Equation o and E1 + E2 + E3 = E.n2 . 2. . a2 . hence. a2 .122) with (3. x3 –axes. .n2 . n 3 and E(n1 . n3 = 1.125) is a complete orthonormal basis of F3 and that the propagator for the three-dimensional box is φ(r. Comparing (3.n3 ) (a1 . .n3 ) = 2π2 = φn1 (a1 . If two or all three orthogonal sides of the box have the same length further symmetry operations leave the system unaltered. . a3 . This symmetry has been exploited in deriving the stationary states.80) shows that the solutions of (3. namely rotation by π around the x1 .n2 . a3 . a3 . x2 . n1 .96) and. (3. t|r0 .n3 =1 φ(n1 . x2 . a3 are identical.123) 8m n2 n2 n2 1 2 3 + 2 + 2 a2 a2 a3 1 .} ∞ (3. r) exp i − E(n1 n2 n3 ) (t − t0 ) φ(n1 . in the present case . . i. x3 –axes by π/2 also leaves the system unaltered.

for example. 5) as shown in the Table above. a. the identity 32 + 32 + 32 = 12 + 12 + 52 . Any linear combination of wave functions ˜ φ(r) = α φ(1. However. r) + β φ(2. in case of degenerate states one cannot necessarily expect that stationary states are orthogonal. φ(1.127) ˜ ˆ ˜ obeys the stationary Schr¨dinger equation H φ(r) = E122 φ(r).2) (a. a. However. 6 This is a result of linear algebra which the reader may ﬁnd in a respective textbook.2. to construct n orthogonal stationary states6 . due to the hermitian character of ˆ H. this linear combination o is not necessarily orthogonal to other degeneraste states. a. n2 . a. a. 3. r) + γ φ(2. Hence.2) (a. a.3.2. for (n1 . We consider the degeneracy of the energy E122 which is due to the identity E122 = E212 = E221 . n3 ) = (3.1.6: Particle in Three-Dimensional Box 71 ‘accidental’ degeneracies also occur.1) (a. .2) (a. in case of an n–fold degeneracy it is always possible. One particular aspect of the degeneracies illustrated in the Table above is worth mentioning.2.g. a. e. 1. r).. a. 3) (1. however. The origin of this degeneracy is. r) (3.

72 Schr¨dinger Equation o .

the most eminent role of this oscillator is its linkage to the boson. as we will demonstrate below. Due to the pedagodical nature of this Section we will link carefully the algebraic treatment with the diﬀerential equation methods used so far in studying the Schr¨dinger equation description of quantum systems. its propagator.2) 2 2m ∂x 2 It comprises one of the most important examples of elementary Quantum Mechanics. at least approximately. i. For example. from the path integral (propagotor) perspective and from the Schr¨dinger equao tion perspective. We have encountered the harmonic oscillator already in Sect.. The linear harmonic oscillator. providing the basis for its quantization. the phonons. The Schr¨dinger equation approach will allow us to emphasize the algebraic aspects of quantum o theory. however. t) = H ψ(x. The path integral approach gave us a direct route to study time-dependent properties. The most likely reason for this connection with fundamental properties of matter is that the harmonic oscillator Hamiltonian (4. o Both approaches. in the context of a path integral approach. In the present section we approach the harmonic oscillator in the framework of the Schr¨dinger o equation. the motion of coherent states. 2 where we determined. Many more physical systems can. However. The important role of the harmonic oscillator certainly justiﬁes an approach from two perspectives. even though it may represent rather non-elementary objects like a solid and a molecule. and its stationary states.Chapter 4 Linear Harmonic Oscillator The linear harmonic oscillator is described by the Schr¨dinger equation o ˆ i ∂t ψ(x. (4. be described in terms of linear harmonic oscillator models. The linear harmonic oscillator describes vibrations in molecules and their counterparts in solids. bosons describe the modes of the electromagnetic ﬁeld. yield the same stationary states and the same propagator. provides a window into the most elementary structure of the physical world. both operators appearing as quadratic terms. t) for the Hamiltonian ˆ H = − 2 (4.2) is symmetric in momentum and position. This Section will be the ﬁrst in which an algebraic formulation will assume center stage. o 73 . one of the conceptual building blocks of microscopic physics. In this respect the material presented provides an important introduction to later Sections using Lie algebra methods to describe more elementary physical systems. There are several reasons for its pivotal role.e. the Schr¨dinger equation approach is suited particularly for stationary state properties.1) ∂2 1 + m ω 2 x2 .

We will point out explicitly where assumptions are made which built on this restriction.235) had been determined. however.1 Creation and Annihilation Operators The Hamiltonian operator (4. only for a discrete set of E values can the boundary conditions (4.4) Equation (4. 4.8) i which holds for any position and momentum operator.5) →±∞ where C∞ denotes the set of functions which together with all of their derivatives are continuous. in which the operators used below. however. (4. beside the representation (4. The stationary states of the harmonic oscillator have been considered already in Chapter 2 where the corresponding wave functions (2. (4.6) the ﬁrst being a diﬀerential operator and the second a multiplicative operator.7) of the Hamiltonian. act.7) which is why these operators are of interest to us.74 Linear Harmonic Oscillator In the following we consider ﬁrst the stationary states of the linear harmonic oscillator and later consider the propagator which describes the time evolution of any initial state. therefore. In the following algebraic solution of (4.4) be satisﬁed. t) where ˆ H φE (x) = E φE (x) .5) ˆ H = 1 1 2 p + m ω 2 x2 ˆ ˆ 2m 2 (4.3) can be solved for any E ∈ R. is the commutation property 1 1 (4.5).3) Due to the nature of the harmonic potential which does not allow a particle with ﬁnite energy to move to arbitrary large distances. lim ( ) = } (4. The cardinal property exploited below. This property states that p and x obey an ˆ ˆ algebra in which the two do not commute. the commutator has a simple form. i dx x = x ˆ (4. ∈ C∞ . If this restriction would not apply and all functions f : R → R would be admitted. In the framework of the Schr¨dinger equation the stationary o states are solutions of (4.3) we restrict the ˆ Hamiltonian H and the operators appearing in the Hamiltonian from the outset to the space of functions N1 = {f : R → R. It is important to keep in mind this restriction of the space. The Hamiltonian H can be expressed in terms of the operators acting on the space (4.1) of the form ψ(x. The operators act ˆ on the space of functions N1 deﬁned in (4. In order [ˆ.2) can be expressed in terms of the two operators p = ˆ d . t) = exp(−iEt/ )φE (x. all stationary states of the harmonic oscillator must be bound states and. x] = p ˆ . the spectrum of ˆ H in (4.3) would be continuous and the eigenfunctions φE (x) would not be normalizable. the natural boundary conditions apply x→±∞ lim φE (x) = 0 .

Our next step is an attempt to factorize the Hamiltonian (4.11) ˆ The reader may note that we have attempted. 1 ω 2 (4.8) a ˆ ˆ ˆ ˆ ˆ implies [ˆ− . the commutator property (4.12). a ˆ 1 (4. ˆ ˆ .10) holds for a ˆ ˆ [ˆ− . x] f (x) = p ˆ d d x f (x) − x f (x) = f (x) = 1 f (x) . 2mω dx (4. x] p ˆ on such functions must be considered. hence. 1 i dx i dx i i (4. 1 ω 2 (4.4. In fact.1: Creation and Annihilation Operators 75 to derive (4.15) in a form which will be useful below ˆ H = ˆ H = ω a− a+ − ˆ ˆ ω a+ a− + ˆ ˆ ω 1 1 2 ω 1 . Before we continue we like to write (4.12) To prove this commutation property we determine using (4.11) a− a+ = ˆ ˆ (4. 2mω dx a− = ˆ mω x + ˆ 2 d .13) (4.7) and (4. ˆ 2 2m ω a− = ˆ mω i x + √ ˆ p. One obtains [ˆ.7) assuming that the factors are easier to handle than the Hamiltonian in yielding spectrum and eigenstates. to factor H/ ω.8) we recall that all operators act on functions in N1 and.14) and (4.18) It is of interest to note that the operators a+ and a− are real diﬀerential operators. a+ ] = 1 . 4. a+ ] = a− a+ − a+ a− does not necessarily vanish.15) together lead to the commutation property (4. 1 2 (4.15) 1 ˆ 1 H + 1 .10) we deﬁne a+ = ˆ mω i x − √ ˆ p.16) (4. the action of [ˆ.17) d dx We also express a+ and a− directly in terms of the coordinate x and the diﬀerential operator ˆ ˆ a+ = ˆ mω x − 2 d . Being guided by the identity for scalar numbers (b − ic)(b + ic) = b2 − i(cb − bc) + c2 = b2 + c2 (4.9) From this follows (4. x] .8).14.14) 1 i mω 2 x + ˆ p2 + ˆ [ˆ. Since a+ and a− are ˆ ˆ + and a− since operators and not scalars we cannot simply expect that the identy (4. in fact.8) yield a− a+ = ˆ ˆ Similarly one can show a+ a− = ˆ ˆ 1 ˆ 1 H − 1 . ˆ 2 2m ω (4. p ˆ 2 2m ω 2 (4.

21) ˆˆ H a+ φE (x) ω 1 ) a+ φE (x) 1 ˆ 2 ω = a+ ( ω a− a+ − ˆ ˆ ˆ 1 + ω 1 ) φE (x) 1 1 2 ˆ = a+ ( H + ω ) φE (x) ˆ = ( ω a+ a− + ˆ ˆ = (E + ω ) a+ φE (x) . it holds that for a stationary state φE (x) of energy E deﬁned through (4.22. (4. 4.24) − m This property states that the operators in the function space N1 are the hermitian conjugate of each other.19) in the form1 f |ˆ+ g a Ω∞ = g|ˆ− f a Ω∞ .19). g ∈ N1 +∞ +∞ dx f (x) a+ g(x) = −∞ −∞ dx g(x) a− f (x) .19) This property states that the operators a+ and a− are the adjoints of each other.17.23) can be generalized to m-fold application of the operators a+ and a− ˆ ˆ ˆ a m H (ˆ+ ) φE (x) ˆ a m H (ˆ− ) φE (x) 1 = ( E + m ω ) a+ ˆ = (E − m ω) a ˆ m φE (x) φE (x) .21) is available. 4.17. a+ and a− as Ladder Operators ˆ ˆ The operators a+ and a− allow one to generate all stationary states of a harmonic oscillator once ˆ ˆ one such state φE (x) ˆ H φE (x) = E φE (x) (4.16. The property ˆ ˆ follows directly from (??). ˆ (4.23) Together. (4. (4. This property of operators is investigated more systematically in Section 5. ˆ The results (4. . 4. 4. 4.22) Similarly one can show using again (4.12. In fact.16. 4.16. 4. The derivation will be based solely on the properties (4.17. Using (??) we like to state (4.20) ˆ In the following we will determine the spectrum of H and its eigenstates.21) ˆˆ H a− φE (x) ω 1 ) a− φE (x) 1 ˆ 2 ω = a− ( ω a+ a− + ˆ ˆ ˆ 1 − ω 1 ) φE (x) 1 1 2 ˆ = a− ( H − ω ) φE (x) ˆ = ( ω a− a+ − ˆ ˆ = (E − ω ) a− φE (x) . one obtains using (4. 4.76 Relationship between a+ and a− Linear Harmonic Oscillator The operators a+ and a− are related to each other by the following property which holds for all ˆ ˆ functions f.21) a− φE (x) is ˆ a stationary state of energy E − ω and a+ φE (x) is a stationary state of energy E + ω. ˆ (4.

The property (??) has an important consequence for the stationary states φE (x). however. 4. (4. φ0 (y) dy dy the solution of which is 1 lnφ0 (y) = − y 2 + c0 2 φ0 (y) = c0 exp 1 − y2 2 . m ∈ Z when m is decreased. two diﬃculties: (i) one needs to know at least one stationary state to start the construction. (4. in fact. Another name is creation (ˆ+ ) and annihilation (ˆ− ) operators since these operators a a create and annihilate vibrational quanta ω.5). Using (4.26) Of course. (ii) the construction appears to yield energy eigenvalues without lower bounds when. one expects that E = 0 should be a lower bound.4. For obvious reasons one refers to a+ and a− as ladder ˆ ˆ operators. In fact.25) needs to be normalizable in order to represent a bound state of the harmonic oscillator.28) 4.32) for some constant c0 or (4. Let φE (x) and φE (x) be two normalized stationary states corresponding to two diﬀerent energies E. (4. For f (x) = φE (x) and g(x) = φE (x) in (??) follows (Note that according to (??) the eigenvalue E is real. 2 2 (4. on ˆ the other side such a state is itself an eigenstate of H as can be shown using (4. the solution φ0 (x) of (4. i.29) d 1 d φ0 (y) = = lnφ0 (y) = − y . φ0 (x) should be an element of the function space N1 deﬁned in (4.25) φ0 (y) = 0 (4. be found.17) ˆ H φ0 (x) = ( ω a+ a− + ˆ ˆ 1 ω 1 ) φ0 (x) = 1 ω φ0 (x) .25) can. There arise.29) where y = mω x. E = E .25) on one side would lead to termination of the sequence E0 + m. E .11) one can rewrite (4.. It turns out that both diﬃculties can be resolved together.6.2: Ground State 77 One can use these relationships to construct an inﬁnite number of stationary states by stepping up and down the spectrum in steps of ± ω.27) Since E = E one can conclude φE |φE Ω∞ = 0.30) Assuming that φ0 (y) does not vanish anywhere in its domain ] − ∞. a state φ0 (x) which obeys the property a− φ0 (x) = 0 ˆ (4. in fact.31) (4. +∞[ one can write (4.2 Ground State of the Harmonic Oscillator d + y dy A suitable solution of (4. (4.e.33) .) ˆ ˆ 0 = φE |HφE Ω∞ − φE |HφE Ω∞ = ( E − E ) φE |φE Ω∞ .

. . we construct the states for n = 1.. mω (4. i.. . .26) the energy value associated with this state is 1 ω. i. . . 2 n = 0.34) dx |φ0 (x)|2 = |c0 |2 −∞ −∞ dx exp − mωx2 2 = |c0 |2 π . 1.78 Linear Harmonic Oscillator This solution is obviously normalizable. and the kinetic energy contribution.2) there are two competing contributions to the energy. . . 1. 1..37) above the ground state energy.36) n times. choose the normalization constants cn and the functions φn (x) real as well.e.. therefore.24) can then be written as follows En = ω(n + 1 ). the so-called excited states. (4. a state conﬁned to the minimum corresponding to the classical state of lowest energy. 2. n = 0. the potential energy contribution which for a state δ(x). it holds ˆ H φn (x) = ω(n + 1 ) φn (x) . we can now construct the stationary states corresponding to energies (4. .e. which for a narrowly localized state yields a large positive value.39).35) The appropriate ground state is φ0 (x) = mω π 1 4 exp − mωx2 2 . (4.3 Excited States of the Harmonic Oscillator Having obtained a suitable stationary state with lowest energy. 4. The set of allowed energies of the oscillator according to (4. (4. 2 n = 0. (4.25) admits only one solution there is only one set of states with energy E + m ω. 2. . The ground state (4. 2.36) assumes a functional form such that both terms together assume a minimum value. Such states need to be suitably normalized for which purpose we introduce a normalization constant cn φn (x) = cn a+ ˆ n φ0 (x) . .37) It is most remarkable that the energy 1 ω of the ground state is larger than the lowest classically 2 allowed energy E = 0.39) We notice that the ground state wave function φ0 (x) as well as the operators (ˆ+ )n are real.36) Since the ﬁrst order diﬀerential equation (4. would yield a vanishing contribution. 2 This state of lowest energy is called the ground state of the oscillator. m ∈ Z which properly terminate at some minimum value E + m0 ω ≥ 0. For this purpose we apply the operator a+ to the ˆ ground state (4. We recall that according to (4. .38) These states correspond to the energy eigenvalues (4. 2. i. We a can.e. . The reason is that in the Hamiltonian (4. The conventional normalization condition φ0 |φ0 reads +∞ +∞ Ω∞ = 1 (4. We will consider this point more systematically in Section 5.

according to (4. For n = 0 holds c0 = 1.47). n = 0. they obey φn |φn = δn n . . (4. βn must be real as well. ˆ Repeated application of this relationship yields φn−s (x) = (n − s)! − a ˆ n! s 2 1 = βn n φn |φn Ω∞ 2 = βn n . 2. (4. n = 0.47) for some suitable constants βn . .24) holds in analogy to (4. Since a− is a real diﬀerential operator [see (4.e. (4. To determine βn we note using (4.20) 1 = φn−1 |φn−1 Equations (4. 1. (4. We consider then the situation that we have obtained a properly normalized state φn (x)..42) ˆ Using (4. n.3: Excited States 79 We need to determine now the normalization constants cn .40) φn−1 (x) = βn a− φn (x) ˆ (4.48) (4. 2.4.41) Employing the adjoint property (4. A properly normalized state φn+1 (x) is then of the form φn+1 (x) = αn a+ φn (x) ˆ for some real constant αn which is chosen to satisfy φn+1 |φn+1 Ω∞ 2 = αn a+ φn |ˆ+ φn ˆ a Ω∞ (4. .44) One can conclude then that the stationary states of the oscillator are described by the functions 1 φn (x) = √ a+ ˆ n! n φ0 (x) . .40–4.45) We like to note that these functions according to (4. i. 1.14) together with H φn (x) = ω(n+ 1 ) φn (x) leads to the condition (note that we assumed 2 φn (x) to be normalized) 2 2 αn ( n + 1 ) φn |ˆ− a+ φn Ω∞ = αn ( n + 1 ) = 1 a ˆ √ From this follows αn = 1/ n + 1 and. . 4. . √ a+ φn (x) = n + 1 φn+1 (x) .18)] and since the ˆ φn (x) are real functions.40).15 . To determine these constants we adopt a recursion scheme.43) (4. according to (4.28) and the construction (4. (4. (4.50) φn (x) .45) form an orthonormal set.51) . √ a− φn (x) = n φn−1 (x) .40) = 1.39) yield √ From this follows βn = 1/ n and.46) According to (4. ˆ (4. Ω∞ 2 = βn a− φn |ˆ− φn ˆ a Ω∞ 2 = βn φn |ˆ+ a− φn a ˆ Ω∞ .49) (4.20) yields 2 αn φn |ˆ− a+ φn a ˆ Ω∞ = 1. .

56) can be expressed in terms of Hermite polynomials Hn (y) introduced in Sect.e.35) rather than (4. by dx = dy mω 1 4 .56) Relationship to Hermite Polynomials We want to demonstrate now that the expression (4. We will demonstrate below the identity Hn (y) = e y2 2 y − d dy n e− y2 2 (4.80 Evaluating the Stationary States Linear Harmonic Oscillator We want to derive now an analytical expression for the stationary state wave functions φn (x) deﬁned through (4. namely the square root of the Jacobian dx/dy. (4. (4.. It should be noted that √ the normalization (4. 2.7. i. for example. (4. .55) 1 y2 2 (4.57) which includes the factor 1/ n! according to Eqs.35) by a constant. In terms of y the ground state wave function is φ0 (y) = π − 4 e− and a+ is ˆ 1 a+ = √ ˆ 2 y − d dy .54) The eigenstates of the Hamiltonian are then given by φn (y) = − √ e n n! π 2 1 y2 2 e y2 2 y − d dy n e− y2 2 . however.200).57) such that one can write the stationary state wave functions of the harmonic oscillator φn (y) = − √ e 2n n! π 1 y2 2 Hn (y) .233) derived in Sect. by introducing the variable y deﬁned in (4. by the Rodrigues formula (2.30) and employing the normalization +∞ −∞ dy φ2 (y) = 1 n (4.39).46) yields a set of normalized states.53) We will later re-introduce this factor to account for the proper normalization (4.52) This normalization of the wave functions diﬀers from that postulated in (4. (4. n–independent factor.58) This result agrees with the expression (2. simplifying the calculation. 2.28) of the ground state and the deﬁnition (4. For this purpose we start from expression (4. (4.45).40–4.52).7 and given.

4.64) which implies that (4.61) and.30) and return later to the coordinate x.60) and.106–3.65) . (4. For the sake of notational simplicity we employ initially the coordinate y as deﬁned in (4. (4. i. (4.200) we need to verify y − which implies Hn (y) = (−1)n ey 2 d dy n e− y2 2 = (−1)n e y2 2 dn −y2 e .59) as follows g(y) = −e = −e y2 2 y2 2 n y2 d d − y2 2 e 2 (−1)n e 2 e−y n dy dy d − y2 d n − y2 e 2 y − e 2 .4 Propagator for the Harmonic Oscillator We consider now the solution of the time-dependent Schr¨dinger equation of the harmonic oscillator o (4.59) hold.4.59) by induction noting ﬁrst that (4. dy dy (4. t0 ).62) Denoting f (y) = (y − d/dy)exp(−y 2 /2) and employing y2 y2 d d − y2 e 2 f (y) = −y e− 2 f (y) + e− 2 f (y) dy dy (4.59) holds for n = 0.39) ∞ ψ(y0 . (3. We prove (4. Starting point of our derivation is the assumption that the initial condition can be expanded in terms of the eigenstates φn (y) (4. and showing then that the property also holds for n + 1 in case it holds for n. therefore.59) dn −y2 e dy n (4. 1..1. t0 ) = n=0 dn φn (y0 ) .e. g(y) = (−1)n+1 e y2 2 dn+1 −y2 e = dy n+1 y − d dy n+1 e− y2 2 .57) of the hermite polynomials and the deﬁnition given by (2. 4.61) One can factor g(y) and employ (4.4: Propagator 81 To verify the realtionship between the deﬁnition (4.2) for an arbitrary initial wave function ψ(x0 . Our derivation will follow closely the procedure adopted for the case of a ‘particle in a box’ [see Eqs.200). hence. dy n (4.63) one obtains g(y) = y − d dy f (y) (4. the Rodrigues formula (2.114)].

2. y0 .108–3.82 Linear Harmonic Oscillator Such expansion is possible for any f (y0 ) = ψ(y0 . y0 .70) and (2. 1965) Sect. N. In order to prove the equivalence of (4.5). 4. 1 t = e−iω(t1 −t0 ) (4.46) the expansion coeﬃcients dn are +∞ dn = −∞ dy0 ψ(y0 . We want to demonstrate now that this propagator is identical to the propagator (2.h. 2 A demonstration of this property can be found in Special Functions and their Applications by N. Employing orthogonality condition (4.N.67) for which according to (3.68) Altogether one can express then the solution +∞ ψ(y.71) Applying (??) to express Hn (y) and Hn (y0 ) yields exp(−2tuv) w(y. t0 ) which is an element of N1 deﬁned in (4.70). t1 |y0 . Inc.15.70) is the propagator of the linear harmonic oscillator.69) where ∞ φ(y. pp.Lebedev (Prentice Hall.. (4. 2n n! π 2 2 (4. t1 ) = −∞ dy0 φ(y. t1 ) = n=0 dn φn (y) cn (t) (4. a supposition which is not proven here2 . We consider for this purpose the following expression for |t| < 1 ∞ w(y..s. sec:harm.7. Englewood Cliﬀs.147) we employ the technique of generating functions as in Sect. . (4.147) for the harmonic iscillator determined in Sect.112) and (4.39) holds cn (t1 ) = exp −i ω ( n + 1 ) ( t1 − t0 ) 2 .72) × exp( −u − v + 2iyu + 2iy0 v ) . 68. of (4. t) (4. t|y0 . t0 ) φn (y0 ) .65) to times t ≥ t0 through insertion of time-dependent coeﬃcients cn (t1 ) ∞ ψ(y. t0 ) = n=0 φn (y) φn (y0 ) tn+ 2 . t0 ) ψ(y0 . t) = π −3/2 exp 2 2 y 2 y0 + 2 2 2 +∞ +∞ ∞ du −∞ −∞ dv n=0 1 ( −2tuv )n × n! (4. this is an excellent textbook from which we have borrowed heavily in this Section.225) which allows one to derive a generating function for products of Hermite polynomials which can be applied to the r.66) One can extend expansion (4. t) = n=0 Hn (y)e−y /2 Hn (y0 )e−y0 /2 n √ t . For this purpose we start from the integral representation (2.J.

The resulting propagator is G(t1 .s.147) derived by means of the path integral description. (4.4. i. t|x0 . −∞ (4...30) to replace y and y0 and that we multiply the propagator by mω/ . t) √ 1 π(1−t2 ) 2 exp − 1 (y 2 + y0 ) 1 + t2 + 2 y y0 2 1−t Hn (y)e−y /2 Hn (y0 )e−y0 /2 n ∞ √ t n=0 2n n! π 2 2 2 t 1 − t2 = . . indeed.77).4: Propagator Carrying out the sum on the r. t0 ) = and 1 2 i π F (t1 .77) The sum in (4. a Re a2 > 0 (4.e. (4. t0 ) − i y y0 2 F (t1 . (4.h. This requires that we employ (4.70) is. it holds φ(y.73) The incomplete Gaussian integral +∞ −∞ dx e−a 2 x2 √ − 2bx = π b2 /a2 e . t0 ) (4.79) 1 + e−2iω(t1 −t0 ) cosω(t1 − t0 ) = . y0 . t0 ) exp i 2 1 2 (y + y0 ) G(t1 .58) of the harmonic oscillator ∞ φn (y) φn (y0 ) tn = n=0 2 1 t 2 1 + t exp − (y 2 + y0 ) + 2 y y0 2 2) 2 1−t 1 − t2 π(1 − t 1 .75) Applying (4.74) a second time yields ﬁnally together with the deﬁnition (4. (4. identical to the generating function (4.71) of w(y. t0 ) = i φ(x. t0 ) = exp mω 2iπ sinω(t1 − t0 ) 1 2 × . t0 ) = where F (t1 . y0 . t1 |y0 . one obtains w(y.78) 1 − e−2iω(t1 −t0 ) = sinω(t1 − t0 ) 2 i e−iω(t1 −t0 ) (4.81) imω 2 sinω(t1 − t0 ) (x2 + x2 ) cosω(t1 − t0 ) − 2 x x0 0 This result agrees with the propagator (2. t) = π −1 exp − 2 y 2 y0 + 2 2 +∞ dv exp(−(1 − t2 ) v 2 − 2i (yt − y0 ) v) .80) −2iω(t1 −t0 ) sinω(t1 − t0 ) 1−e We can ﬁnally express the propagator (4.76) One can express this in terms of the stationary states (4.78) in terms of the coordinates x and x0 .74) applied once results in w(y. t) = π −3/2 exp +∞ 83 y2 y2 + 0 2 2 +∞ dv exp( −v 2 + 2iy0 v ) × −∞ × −∞ du exp( −u2 − 2u(tv − iy) ) . i. √ πexp( t2 v 2 − 2iytv − y 2 ) (4. by a factor 4 mω/ for both φn (y) and φn (y0 ). y0 .e.

i. both quantum systems are endowed with a large number of modes. phonons. a (4. a− must be equivalent. ˆ ˆ Calculus of Creation and Annihilation Operators We summarize ﬁrst the key properties of the operators a+ and a− ˆ ˆ [ˆ− .85) is convergent everywhere in R.84 Linear Harmonic Oscillator 4. e.g. each corresponding to a single harmonic oscillator of the type studied presently by us.86) f (ˆ+ ) = a a ˆ ν! ν=0 .. ˆ ˆ To put the following material in the proper modest perspective we may phrase it as an approach which rather than employing the coordinate y and the diﬀerential operator d/dy uses the operators 1 d 1 d y − . a+ ] a ˆ a− φ0 (y) ˆ a+ φn (y) ˆ a− φn (y) ˆ a− f |g Ω∞ ˆ We note that these properties imply √ φn (y) = (ˆ+ )n φ0 (y)/ n! . the use of generating functions. For the present there is actually no pressing need to apply such techniques since the techniques ˆ borrowed from the theory of diﬀerential equations serve us well in describing harmonic oscillator type quantum systems. Such functions are deﬁned for ˆ ˆ a f : R → R in case that the Taylor expansion ∞ f (y) = ν=0 f (ν) (0) ν y ν! (4.5 Working with Ladder Operators In the last section we have demonstrated the use of diﬀerential equation techniques.82) a+ = √ ˆ dy dy 2 2 √ √ Obviously.e. d/dy and a+ . We want to introduce now techniques based on the ladder operators a+ and ˆ a− . It is with quantum electrodynamics and solid state physics in mind that we cease the opportunity of the single quantum mechanical harmonic oscillator to develop a working knowledge for a+ and a− in the most simple setting. one can express y = 2(ˆ− + a+ ) and d/dy = 2(ˆ− − a+ ) and. (4. f (ˆ+ ). The reason for introducing the calculus of the operators a+ and a− is that ˆ ˆ this calculus proves to be useful for the description of vibrations in crystals. Here f (ν) (y0 ) denotes the ν-th derivative of f (y) taken at y = y0 . In this case we deﬁne ∞ f (ν) (0) + ν (4. the approaches a ˆ a ˆ using y.. and of the modes of the quantized electromagnetic ﬁeld.84) = 1 1 = 0 √ = n + 1 φn+1 (y) √ = n φn−1 (y) = f |ˆ+ g a Ω∞ . a− = √ ˆ y + .83) We will encounter below functions of a+ and a− . (4. hence.

.88) (4. The ﬁrst case is trivial. a a a 85 (4. a a a ˆ 1 a Let us assume that (4. ˆ a a a ˆ In particular. An example is the so-called shift operator exp(uˆ− ). we will only assume operator functions acting on φ0 (y). a+ ] = 1 = f1 (ˆ+ ) . the second case follows from [ˆ− . fn+1 (ˆ+ )] a a = [ˆ− .90) we show that (4. i.88) is (ˆ− )n f (ˆ+ ) φo (y) = f (n) (ˆ+ ) φ0 (y) . a+ ] + [ˆ− .91) (4.88) is a fundamental one and will be used in the following.90) holds for f (ˆ+ ). Hence. a An important operator function is the exponential function. property (4.83) and can be proven for all powers fn by induction and then inferred for all proper functions f (ˆ± ). The following important property holds a a− f (ˆ+ ) = f (1) (ˆ+ ) + f (ˆ+ ) a− .90) holds for fn .e. f1 (ˆ+ )] = [ˆ− . It holds a a euˆ f (ˆ+ ) φ0 (y) = f (ˆ+ + u) φ0 (y) .93) one can reduce all operators acting on some proper state function by operators acting on the state φ0 (y).92) f (y) = n=0 dn φn (y) = n=0 (ˆ+ )n a dn √ φ0 (y) n! (4.94) This identity follows from (4. a a a We like to state the following property of the functions f (ˆ± ) a f (ˆ− ) φ|ψ = φ|f (ˆ+ ) ψ . The convergence of the Taylor expansion ascertains then that (4.4. f (ˆ+ )] + f (ˆ+ ) a− ˆ a a a a ˆ which implies that in order to prove (4. f (ˆ+ )] = f (1) (ˆ+ ) . (ˆ+ )n ] a+ a a ˆ a a ˆ a a ˆ = (ˆ+ )n 1 + n (ˆ+ )n−1 a+ = (n + 1) (ˆ+ )n a 1 a ˆ a Since any function f (y) in the proper function space N1 can be expanded ∞ ∞ (1) (4.96) .90) holds for f0 and for f1 .89) (4. For fn+1 follows then [ˆ− . ˆ a We proceed by induction noticing ﬁrst that (4.87) (4.90) To prove (4. a a (4. As long as the operators act on φ0 (y) one can state then that a− behaves like a diﬀerential operator with respect to functions f (ˆ+ ). We note ˆ a− f (ˆ+ ) = [ˆ− .88) we need to show actually [ˆ− . a a − (4. 4. a− f (ˆ+ ) φ0 (y) = f (1) (ˆ+ ) φ0 (y) ˆ a a which follows from a− φo (y) = 0. (ˆ+ )n a+ ] = (ˆ+ )n [ˆ− .87.95) (4.5: Working with Creation and Annihilation Operators and similarly for f (ˆ− ). Note that ˆ a an immediate consequence of (4.90) holds for any function fn (ˆ+ ) = (ˆ+ )n which is a power of a a a+ .

98) for u = −α∗ and v = α yields e−α This together with (4. (4. + (4.104) (4.105) a eαˆ + − α∗ a− ˆ φ0 (y) = e αα∗ 2 e−α ∗ a− ˆ a eαˆ φ0 (y) .101) 1.102) Similarly. + ∗ + (4. z ∈ C.106) Applying (4.98) The related operators exp[vˆ+ ± v ∗ a− ] play also an important role. without explicitly stating this.106) yields a eαˆ + ∗ a− ˆ a a eαˆ φ0 (y) = e−αα eαˆ φ0 (y) .100) ˆ ˆ To solve this diﬀerential equation we deﬁne C(u) = D(u) exp(−u2 /2) which leads to d ˆ D(u) = du zˆ+ + z ∗ a− a ˆ ˆ D(u) (4.96) is applied is a a a euˆ evˆ φ0 (y) = ev (ˆ − + + + u) a φ0 (y) = euv evˆ φ0 (y) . hence.86 To prove this property we expand exp(uˆ− ) a ∞ ν=0 Linear Harmonic Oscillator uν − ν (ˆ ) f (ˆ+ ) φ0 (y) = a a ν! ∞ ν=0 uν (ν) + f (ˆ ) φ0 (y) = f (ˆ+ + u) φ0 (y) . + ∗ a− ˆ a − u euzˆ euz + ∗ a− ˆ = (4.108) . We assume in the following a ˆ derivation.101) ˆ ˆ ˆ C(u) obviously obeys C(0) = 1 This results in D(0) = 1 and. 1 + + uz ∗ a− ). one obtains a evˆ + − v ∗ a− ˆ a φ0 (y) = e− 2 vv evˆ e− v 1 ∗ + ∗ a− ˆ φ0 (y) (4. the solution of (4. a a ν! (4.97) An example in which (4. Below we will use the operator identity which follows for the choice of v = α in (4. ∗ = exp(uzˆ a ˆ and determine its derivative d du ˆ C(u) a = zˆ+ euzˆ euz a + ∗ a− ˆ a + euzˆ z ∗ a− euz ˆ + ∗ a− ˆ = + zˆ+ + z ∗ a− a ˆ Using (4.107) − α∗ a− ˆ φ0 (y) = e− αα∗ 2 a eαˆ φ0 (y) .103) (4. One can conclude then using the deﬁnition of C(u) and deﬁning ˆ ˆ is D(u) = exp(uzˆ a ˆ v = uz 1 ∗ ˆ− ∗ ∗ ˆ− a+ a+ evˆ + v a φ0 (y) = e 2 vv evˆ ev a φ0 (y) . + (4. that the operators considered act on φ0 (y). euzˆ ˆ euz ∗ a− ˆ (4. To express ˆ these operators as products of operators exp(vˆ+ ) and exp(v ∗ a− ) we consider the operator C(u) = a ˆ + ) exp(uz ∗ a− ) where u ∈ R. + (4.99) ˆ C(u) = zˆ a zˆ+ + z ∗ a− a ˆ + +z a ˆ ∗ − a euzˆ euz ˆ − u C(u) .90) and zz ∗ = 1 we can write this d du e uzˆ+ a e uz ∗ a− ˆ a − z ∗ a− .105) e e vˆ+ a + v ∗ a− ˆ φ0 (y) = e − 1 vv ∗ 2 e v ∗ a− ˆ e vˆ+ a φ0 (y) vˆ+ a − v ∗ a− ˆ φ0 (y) = e 1 vv ∗ 2 e −v ∗ a− ˆ e vˆ+ a φ0 (y) .

.58) the Hermite polynomials Hn (y) by the eigenstates φn (y) √ ∞ 1 ( 2t)n y2 /2 2yt − t2 √ e φn (y) (4.112) Expanding both sides of this equation and using (4. Similarly.58) one obtains π− 4 1 ∞ ν=0 (−1)ν u2ν = 2ν ν! ∞ ν=0 uν √ φν (0) = ν! ∞ ν=0 uν Hν (0) √ 2ν π ν! (4. We start from the generating function (??) and replace according to (4.116) = δµ ν . is the equivalent of the generating function (??).s.111) yields π − 4 exp − 1 u2 2 a = euˆ φ0 (0) .114) where we have employed (4.111) the orthonormality properties of the wave functions φn (y).95) one obtains a a a a euˆ φ0 |evˆ φ0 = φ0 |euv evˆ φ0 = euv evˆ φ0 |φ0 = euv − + + + − (4. the expected from which follows by comparision of each term on both sides φµ |φν orthonormality property. we can reproduce by means of the generating function (4.h. and ˆ l.111) This expression.s. Expanding r. Using (4.h.84) yields ∞ ν. and on the r.113) Comparision of all terms on the l. (4.e.84) and deﬁning 2t = u one can write then π − 4 exp − or π − 4 exp − 1 1 √ y2 u2 + 2uy − 2 2 ∞ = n=0 un + n (ˆ ) φ0 (y) a n! + (4. i. Setting y = 0 in (4. in the ladder operator calculus. + (4.115) and using (4. 4. of (4.95).115) a where the latter step follows after expansion of evˆ and using a− φ0 (y) = 0.s. . (??).98) and again (4.110) √ y2 u2 + 2uy − 2 2 a = euˆ φ0 (y) . We want to derive now the values Hn (0).84.5: Working with Creation and Annihilation Operators Generating Function in Terms of a+ ˆ 87 We want to demonstrate now that generating functions are as useful a tool in the calculus of the ladder operators as they are in the calculus of diﬀerential operators. For this purpose we consider a a a a euˆ φ0 |evˆ φ0 = φ0 |euˆ evˆ φ0 + + − + (4. We will derive the equivalent of a generating function and use it to rederive the values Hn (0) and the orthonormality properties of φn (y).s.h.4.µ=0 uµ v ν √ φµ |φν µ!ν! ∞ = µ=0 uµ v µ µ! (4. provide the same values for Hn (0) as provided in Eq.h.109) e = π4 n! n=0 √ Using (4.

e. (4. is given by a function ψ(x). according to (4. of (4. For this purpose we employ the Schr¨dinger equation in the momentum o representation. ˜ (4.119) 2m 2 dp2 Accordingly. ..6) and the Hamiltonian (4. for an oscillator in a stationary state φn (x) as given by (2. (−i)n ˜ φn (p) = √ 2n n! m˜ ω π 1 4 exp − m˜ p2 ω 2 Hn ( m˜ ω p) ..2).118) d2 1 2 + p dp2 2m ˜ ˜ φE (p) = E φE (p) .88 Linear Harmonic Oscillator 4.2. (4.120) Multiplying this equation by 1/m2 ω 2 yields − where ˜ E = E / m2 ω 2 . As to be expected. (4. i. The momentum representation eigenfunctions are. i.7) is 1 2 m 2 ω 2 d2 p − . 2.123) (4. are ˜ En = or.117) In this section we want to provide a representation for the harmonic oscillator which is most natural ˜ if one wishes to determine for a stationary state of the system the probability density P (p) of ﬁnding the system at momentum p. n = 0.3).125) .121) ω (n + ˜ 1 ) 2 .235). the momentum and position operators are p = p. ω = 1 / m2 ω .121).e. the momentum and position operators are as stated in (4. 1.. dp (4.7) is given by (4. . The solutions can be stated readily exploiting the earlier results. (4. En = ω (n + 1 ) 2 d2 1 + m˜ 2 p2 ω 2 2m dp 2 2 ˜ ˜˜ φE (p) = E φE (p) (4. the probability density is P (x) = |φn (x)|2 .122) For a solution of the Schr¨dinger equation in the momentum representation. is given by the ˜ function ψ(p).6 Momentum Representation for the Harmonic Oscillator The description of the harmonic oscillator provided so far allows one to determine the probability density P (x) of ﬁnding an oscillator at position x.37). using (4. In the position representation the wave function is a function of x. (4. using (2. The eigenvalues. For example.124) ). the eigenvalues are identical to those determined in the position representation. the time-independent Schr¨dinger equation for the oscillator can be written o ˆ H = − mω 2 2 2 x = i ˆ d .e. 2. ˆ and the Hamiltonian (4. n = 0. . we o note that the posed eigenvalue problem is formally identical to the Schr¨dinger equation in the o position representation as given by (4.122). . . i. 1. 4. . In the momentum representation the wave function is a function of p.235).

through direct integration. of course. hence.125.134).30) and (4.235)] absorbs the replacement dx → dy.127) exp − p2 m ω 2 Hn ( 1 p) .e. α ∈ R. ˜ We can now state the probability density Pn (p) for an oscillator with energy En to assume momentum p.129) follows ky = px/ and.126). this allowed us to introduce in (4. the Jacobian.130) dx exp(−ipx/ ) φn (x) −∞ (4.129) m ω one obtains.131) in the present case employing dimensionless units.. 1 ˜ φn (k) = √ 2π +∞ (4. Proceed as follows.4.6. φn (k) = We want to ﬁnally apply the relationship 1 ˜ φn (p) = √ 2π +∞ 2 (−i)n −k √ e 2 Hn (k) .126) in terms of dimensionless units following the procedure adopted for the position representation where we employed the substitution (4.132) dy exp(−iky) φn (y) −∞ (4. i. m ω (4.6: Momentum Representation or with (4.122) (−i)n ˜ φn (p) = √ 2n n! 1 πm ω 1 4 89 exp − p2 2m ω Hn ( 1 p) . 2n n! π (4. deﬁned only up to an arbitrary phase factor eiα . Deﬁning k = m˜ / p or ω 1 k = p (4.30) to obtain (4. m ω (4.126) a phase factor (−i)n .126) Normalized eigenfunctions are. i.133) where we note that a change of the normalization of φn (y) [c. This implies the property of Hermite polynomials (−i)n e− k2 2 +∞ Hn (k) = −∞ dy exp(−iky) e− y2 2 Hn (y) (4.1: Demonstrate the validity of (4. According to the theory of the momentum representation holds ˜ ˜ Pn (p) = |φn (p)|2 or 1 ˜ Pn (p) = n 2 n! 1 πm ω 1 2 (4.f. 4. instead of (4. From (4.58). (4. .30) and (2.. the correctness of the phase factors (−i)n .134) Exercise 4.e.128) We may also express the eigenstates (4.

137) where y(t) describes the center of the particle and where the mass of the particle remains narrowly distributed around y(t) for an arbitrary period of time. Such states are referred to as quasi-classical states. J.Dupont-Roc.7 Quasi-Classical States of the Harmonic Oscillator A classical harmonic oscillator.C.194. +∞ Linear Harmonic Oscillator dy exp(−iky) g(z. a pendulum swinging at small amplitudes. The quasi-classical states can be obtained by generalizing the construction of the ground state of the oscillator.Grynberg (John Wiley & Sons. z) = exp 2yz − z 2 − y 2 /2 (4.136) (b) Employ the generating function (2.134).138). carries out a periodic motion described by y(t) = Re y0 eiωt (4. 2.90 (a) Prove ﬁrst the property 1 f (k. for example. except that α changes periodically in time. 1989) which discusses quasi-classical states of the free electromagnetic ﬁeld in Sect. For this purpose we assume such state exists and expand ∞ ∞ φ(α) (y) = n=0 d(α) φn (y) . 4. The question arises if similar states exist also for the quantum oscillator..Chen-Tannoudji. a useful role in Quantum Electrodynamics since they allow to reproduce with a quantized ﬁeld as closely as possible the properties of a classical ﬁeld3 . z) = exp −2ikz + z 2 − k 2 /2 . coherent states. namely through the eigenvalue problem for normalized states a− φ(α) (y) = α φ(α) (y) . ˆ α ∈ R.140) An excellent textbook is Photons and Atoms: Introduction to Quantum Electrodynamics by C.135) g(k. New York. z) = √ 2π where f (k. pp. 4. φ(α) | φ(α) = . for example.138) We will show that a harmonic oscillator prepared in such a state will remain forever in such a state.196) of Hermite polynomials and equate coeﬃcients of equal powers of z to prove (4.138) yields using a− φn (y) = n φn−1 (y) ˆ ∞ n=0 √ n d(α) φn−1 (y) − α d(α) φn (y) n n = 0 (4.4.139) where we have added the condition that the states be normalized. We want to construct and characterize such states. We ﬁrst construct the solution of (4.— III. 192 3 . (4. or Glauber states of the harmonic oscillator and they play.1 showing in its top row the attributes of such state just discussed. n n=0 d(α) n 2 = 1 (4. Inc. We will ﬁnd that α can be characterized as a displacement form the minimum of the oscillator and that the state φ(α) (y) displays the same spatial probability distribution as the ground state. Inserting this expansion into √ (4. The answer is ‘yes’. y) −∞ (4. and G. The motion of the probability distribution of such state is presented in Fig.

145) Using the generating function in the form (4. Comparing (4.147) allows one to interprete T (α) as an operator √ which shifts the ground state by 2α. n! (4.142) The solution is αn d(α) = √ c n n! ∞ ∞ (4.141) Because of the linear independence of the states φn (y) all coeﬃcients multiplying φn (y) must vanish and on obtains the recursion relationship (α) dn+1 = √ α d(α) .150) for α presently real..84) φ(α) (y) = exp − α2 2 ∞ n=0 αα∗ (αˆ+ ) a a+ φ0 (y) = e− 2 eαˆ φ0 (y) . n! (4. hence. .108) allows one to write φ(α) (y) = T + (α) φ0 (y) = T + (α) [π]− 4 e−y where a T + (α) = eαˆ + 1 2 /2 (4. we will see below that α ∈ C are also admissible and will provide a corresponding interpretation.7: Quasi-Classical States or ∞ n=0 91 √ n + 1 dn+1 − α d(α) n (α) φn (y) = 0 .146) √ 1 − (y − 2 α)2 2 (4.149) with (4. This interpretation justiﬁes the choice of α ∈ R in (4. the quasi-classical states are φ (α) (y) = exp α2 − 2 ∞ n=0 αn √ φn (y) .139)] 1 = n=0 d(α) n 2 = |c|2 n=0 α2 2 = |c|2 eα . [π]− 4 e− one can write φ(α) (y) = [π]− 4 exp 1 1 y2 2 + √ 2 α y − α2 = exp − α2 2 ∞ n=0 αn √ φn (y) . n! (4. (4.148) The identity (4.144) Normalization requires c = exp(−α2 /2) and.147) √ which identiﬁes this state as a ground state of the oscillator displaced by 2α.109).149) − α∗ a− ˆ (4. n n+1 (4.145) using (4. i. n! n (4. One can also write (4.138).4.e. However.143) for a constant c which is determined through the normalization condition [see (4.

(4.h.95) to linear combinations αˆ+ + βˆ− and reads then a a f (αˆ+ + βˆ− ) φ|ψ = φ|f (α∗ a− + β ∗ a+ ) ψ . is the adjoint of the operator function on the l.152) T + (α) f |T + (α) g = f |T (α) T + (α) g = f |g .150) for complex α obeys the unitary property T + (α) T (α) = 1 1 (4. Following the procedure adopted in determining the propagator of the harmonic oscillator [see (4. it holds T + (α) φ0 |T + (α) φ0 = f |T (α) T + (α) g = φ0 |φ0 .153) This is obviously the inverse of T + (α) as deﬁned in (4. 1. becomes real. 4.145) as the initial state φ(α) (y. Relationship (4.155) (4.159) Noticing the identity which holds for |u|2 = 1 u2 + 1 = (Re u)2 − (Im u)2 + 1 + 2 i Re u Im u = 2 (Re u)2 + 2 i Re u Im u (4. .h. one can write in case of T + (α) T + (α) + = T (α) = eα ∗ a− ˆ − αˆ+ a a = e − (αˆ + − α∗ a− ) ˆ . n! u = e−iω(t1 − t0 ) .e.68)] one can write the time-dependent solution with (4.158) Comparision of this expression with (4. t1 ) = exp or φ (α) − α2 2 ∞ exp n=0 −i ω ( n + ∞ n=0 1 ) ( t1 − t0 ) 2 αn √ φn (y) n! (4. α ∈ C (4.151) which follows from a generalization of (4.s.150) and one can conclude that T + (α) is a so-called unitary operator with the property T + (α) T (α) = T (α) T + (α) = 1 1 or applying this to a scalar product like (4.160) . i.. n = 0.157) (y.156) (4.109) serves again to simplify this sum √ y2 1 1 α2 (u2 + 1) φ(α) (y. t1 ) = u exp 1 2 α2 − 2 (α u)n √ φn (y) .s.92 Linear Harmonic Oscillator The shift operator as deﬁned in (4. t1 ) = π − 4 u 2 exp − + 2yαu − 2 2 E (4.154) . . the shift operator leaves the ground state normalized.152) Since the operator function on the r. . In particular.145) is replaced by a complex number αu which at times t1 = t0 + 2πn/ω. (4. a a ˆ ˆ (4.67.145) shows that α in (4.

162) 2 with a periodic change.163) also through the shift operator. experiences a change of its width (relative to .158) allows one to write 1 φ(α) (y. t1 ) = π − 4 exp − [y − y0 (t1 )]2 + iy k0 (t1 ) − iφ0 (t1 ) − ω(t1 − t0 ) 2 2 . a complex α corresponds to a displaced oscillator which has initially a non-vanishing velocity. Arbitrary Gaussian Wave Packet Moving in a Harmonic Potential We want to demonstrate now that any initial state described by a Gaussian shows a time dependence very similar to that of the Glauber states in that such state remains Gaussian.138. This shows clearly that the Glauber state is a close analogue to the classical oscillator. except that it is not pointlike.148.159) is then 1 1 i φ(α) (y. The diagram illustrates that the wave function retains its Gaussian shape with constant width for all times. 4. (4. 4. t1 ) = u 2 T + (α u) φ0 (y) (4. moving solely its center of mass in an oscillator fashion around the minimum of the harmonic potential.150) T + (α u) = exp α ( e−iω(t1 −t0 ) a+ − eiω(t1 −t0 ) a− ) ˆ ˆ This provides a very compact description of the quasi-classical state. Our interpretation 2 explains also the meaning of a complex α in (4. It is of interest to note that the displacement y0 of the Gaussian as well as the momentum k0 and phase φ0 associated with (4.165) . In Fig.7: Quasi-Classical States the exponent E can be written E = − √ √ y2 + 2 αy Reu − α2 (Re u)2 + i 2 αy Imu − i α2 Reu Imu 2 √ √ 2 1 = − y − 2αReu + i 2yα Imu − i α2 Reu Imu 2 phase term momentum displaced Gaussian 93 (4.150) with (4. this characterization corresponds closely to that of the possible initial states of a classical oscillator.138): a real α corresponds to a displacement such that initially the oscillator is at rest. a momentum factor and a phase factor.163) This solution corresponds to the initial state given by (4.1 (top row) we present the probability distribution of the Glauber state for various instances in time.161) are time-dependent √ y0 (t1 ) = 2 α cosω(t1 − t0 ) √ k0 (t1 ) = − 2 α sinω(t1 − t0 ) 1 φ0 (t1 ) = − α2 sin2ω(t1 − t0 ) (4.4. The oscillations of the mean position y0 (t1 ) and of the mean momentum k0 (t1 ) are out of phase by π just as in the case of the classical oscillator. Comparing (4. being displaced around the center with period 2π/ω and. (4. in general. The time-dependent wave function (4. obviously. 4.164) where according to (4.147).161) and is found to describe a displaced Gaussian. One can express (4.

168) dy e−p(y−q) Hn (y) where p = 1 + σ2 . To describe such state which satisﬁes the initial condition φ(ao . t0 ) +∞ = √ 1 n n!πσ 2 e−ao q/2 = √ 2n n!πσ dy exp −∞ +∞ −∞ − (y − ao )2 y2 − 2σ 2 2 2 Hn (y) (4.1 (second and third row).167) One can derive for the expansion coeﬃcients +∞ dn = −∞ dy φn (y) φ(ao . t0 ) = n=0 dn φn (y) . This behaviour is illustrated in Fig.1: Time dependence of Gaussian wave packets in harmonic oscillator potential. σ|y. t0 ) = [πσ 2 ]− 4 exp we expand ∞ 1 − (y − ao )2 2σ 2 (4.169) . 2σ 2 q = a . 1 + σ2 (4. σ|y. σ|y. (4. that of the Glauber states) with a period π/ω. One can recognize the cyclic displacement of the wave packet and the cyclic change of its width with a period of twice that of the oscillator.94 E E E E Linear Harmonic Oscillator E 2h ω 2h ω 2h ω 2h ω 2h ω π t= 0ω σ2 = 1 ¥ ¦ £ y ¨ ¡ ¢ ¡ π t = 1/4 ω σ2 = 1 ¥ ¦ £ ¡ y ¨ ¢ ¡ π t = 1/2 ω σ2 = 1 ¥ ¦ £ ¡ y ¨ ¢ ¡ π t = 3/4 ω σ2 = 1 ¥ ¦ £ ¡ y ¨ ¢ ¡ π t= 1 ω σ2 = 1 ¥ ¦ £ ¤ ¤ ¤ ¤ ¤ y ¡ ¢ ¡ ¨ -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 0 2 4 E E E E E 2h ω π t = 0ω σ2 = 1/3 ¥ ¤ § £ ¢ 2h ω π t = 1/4 ω σ2 = 1/3 ¥ § £ 2h ω π t = 1/2 ω σ2 = 1/3 ¥ § £ 2h ω π t = 3/4 ω σ2 = 1/3 ¥ § £ 2h ω π t= 1 ω σ2 = 1/3 ¥ § £ ¤ ¤ ¤ y ¨ y ¨ ¢ y ¨ ¢ y ¨ ¢ ¤ y 0 ¢ -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 2 4 © © © © E E E E E 2h ω π t = 0ω σ2 = 2 ¥ £ 2h ω π t = 1/4 ω σ2 = 2 ¥ £ 2h ω π t = 1/2 ω σ2 = 2 ¥ £ 2h ω π t = 3/4 ω σ2 = 2 ¥ £ © 2h ω π t= 1 ω σ2 = 2 ¥ ¤ ¤ ¤ ¤ y ¨ ¡ ¢ ¡ y ¨ ¢ ¡ y ¨ ¢ ¡ y ¨ ¢ ¡ ¤ y ¡ ¢ ¡ ¨ ¡ ¡ ¡ £ ¨ -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 0 2 4 -4 -2 0 2 4 Figure 4. 4.166) φ(ao .

172) One can then conclude e−a /4 n lim dn = √ a σ→1 2n n! (4. p−1 1 − σ4 t = 1 − σ 2 −iω(t1 −to ) e 1 + σ2 (4. 1 (4.4.7: Quasi-Classical States The solution of integral (4.177) − 1 ω(t1 − to ) 2 . This expansion can be written φ(ao .P. i.157) as follows φ(ao .143).168) is4 e−aq/2 (p − 1)n/2 dn = √ Hn (q 2n n!σ p(n+1)/2 p ).174) Hn (q p p−1 ) φn (y) . For this purpose we note that according to the explicit expresssion (??) for Hn (y) the leading power of the Hermite polynomial is Hn (y) = (2 y)n + and. σ→1 p−1 2 (4.. Yu. this 3 volume integral table is likely the most complete today. σ|y. σ|y. σ|y. New York. In this limit the coeﬃcients dn should√ become identical to the coeﬃcients (4.166) in analogy to (4. t1 ) = √ × where yo = q p a = √ .143) of the Glauber state.176) 1 √ pσ π a 2 exp − 1 1+σ2 + 1 yo − 1 ω(t1 − to ) 2 2 2 2 × (4. 1 a2 2 1 + σ2 + 1 2 2 y o + 2 y yo This integral can be found in Integrals and Series.170) In order to check the resulting coeﬃcients we consider the limit σ → 1. 2 by A.e. Prudnikov.171) lim p−1 p n 2 Hn (q p ) = lim (2q)n = an . unfortunately very expensive. vol. namely $750 4 . Brychkov. 1990). Following the strategy adopted previously one can express the time-dependent state corresponding to the initial condition (4. t1 ) = e−aq/2 √ pσ ∞ n=0 e −iω(n+ 1 )(t1 −to ) √ 2 1 2n n! p−1 p n/2 (4.173) which is. t1 ) = √ 2 × exp − 1 (y 2 + yo ) 1+t2 − 2 1−t 2 1 1 √ √ pσ π 1 − t2 × t 1 − t2 (4. in fact. σ→1 O(y n−2 ) for n ≥ 2 1 for n = 0.175) ∞ tn n=0 2n n! Hn (yo ) e−yo /2 Hn (y) e−y 2 2 /2 The generating function (4. in agreement with the expected result (4. Marichev (Wiley. a worthy successor of the famous Gradshteyn. and O.76) permits us to write this φ(ao . a ground state shifted by ao = 2α.I. therefore. p−1 95 (4.

96 Linear Harmonic Oscillator .

We will also introduce the generators of this group and their algebra as well as the representation of rotations through exponential operators. argon. or the proton and the neutron made up of three quarks. The mathematical techniques presented in this section will be used throughout the remainder of these notes. The reason is that these transformations and groups are closely tied to the properties of elementary particles. the building blocks of matter. In this section we want to investigate how elementary and composite systems are described. but also to the properties of composite systems. 5. neon.. As an illustration we will consider ﬁrst rotational transformations acting on vectors r in 3-dimensional space. and ﬁnally transformations of products of wavefunctions like N ψj (rj ) which represent a system of N (spinj=1 zero) particles in R . To develop a systematic description of rotational properties of composite quantum systems the consideration of rotational transformations is the best starting point. we will then consider transformations of wavefunctions ψ(r) of single particles in R . 97 . the group SO(3) of the associated rotation matrices and the corresponding transformation matrices of spin– 1 states forming the group SU(2) occupy a very 2 important position in physics. e. helium.. i.1 Matrix Representation of the group SO(3) In the following we provide a brief introduction to the group of three-dimensional rotation matrices.e. we will ﬁnd that the algebraic properties of operators governing spatial and spin rotation are identical and that the results derived for products of angular momentum states can be applied to products of spin states or a combination of angular momentum and spin states..Chapter 5 Theory of Angular Momentum and Spin Rotational symmetry transformations. Examples of the latter with particularly simple transformation properties are closed shell atoms. all composite systems which appear spherical as far as their charge distribution is concerned.g. We will also review below the well-known fact that spin states under rotations behave essentially identical to angular momentum states. i. r ∈ R .e. the magic number nuclei like carbon.

i. (5. namely around an axis given by the direction of ϑ and by an angle |ϑ|. 0.4) With the deﬁnition of the transposed matrix RT [RT ]jk ≡ Rkj this property can be written R(ϑ) RT (ϑ) = RT (ϑ) R(ϑ) = 1 1 .e. In the latter case the determinant of R(ϑ = (0. Let us deﬁne the rotated vector as r = R(ϑ) r . (5. From (5. In the following we want to exclude inversions and consider only rotation matrices with detR = 1. ϕ)T ) = ± sinϕ cosϕ 0 (5.7) Let us consider brieﬂy an example to illustrate rotational transformations and to interpret the sign of detR(ϑ).5) This equation states the key characteristic of rotation matrices. . x3 )T ..(5. =1 [R(ϑ)]jk [R(ϑ)]j ak b = j=1 aj bj . i.6) follows immediatety for the determinant of R(ϑ) using detA B = detA detB and detAT = detA detR(ϑ) = ±1 . therefore. ϕ)T ) is negative.k. in case of a prefactor −1 a rotation and an inversion at the origin. 3 . 2.e. It holds then j=1 3 3 a ·b = j.e.6) (5. the matrix represents a proper rotation. x2 .8) 0 0 1 In case of a prefactor +1. in Cartesian coordinates r = (x1 . (5.. it follows 3 [R(ϑ)]jk [R(ϑ)]j = δk j=1 . (5. (5. i. where ϑ denotes the rotation. can be represented by 3 × 3 matrices R(ϑ). A rotation around the x3 -axis by an angle ϕ is described by the matrix cosϕ − sinϕ 0 R(ϑ = (0. if the thumb in your right hand ﬁst points in the direction of ϑ.98 Properties of Rotations in R Theory of Angular Momentum and Spin Rotational transformations of vectors r ∈ R .7) implies that a rotation is associated with an inversion. they conserve a · b = 3 aj bj . ϑ parametrizes the rotations uniquely as long as |ϑ| < 2π. (5. a minus sign in Eq.1) In Cartesian coordinates this reads 3 xk = j=1 [R(ϑ)]kj xj k = 1. 0. are linear and. then the ﬁngers of your ﬁst point in the direction of the rotation.2) Rotations conserve the scalar product between any pair of vectors a and b.3) Since this holds for any a and b. We assume the convention that rotations are right-handed.

hence. We need to demonstrate that this inverse belongs also to −1 T SO(3). b ∈ G the product c = a ◦ b is also in G. It follows that R3 is also an element of SO(3).9) This set of matrices is closed under matrix multiplication and. For this purpose we must demonstrate that the group properties (i–iv) hold. R2 ∈ SO(3) and do not write out “◦” explicitly since it represents the matrix product. (5. (ii) The group element ‘e’ is played by the 3 × 3 identity matrix 1 1. one can show R3 R3 = 1 Furthermore. Since (R1 )T = (R1 )−1 it follows −1 −1 −1 T (R1 )T R1 = (R1 )−1 R1 = −1 which implies R1 ∈ SO(3). there exists a real 3 × 3–matrix −1 R1 which is the inverse of R1 . (iv) For products of three elements holds the associative law a ◦ (b ◦ c) = (a ◦ b) ◦ c. (ii) There exists an element e in G with the property ∀a ∈ G → e ◦ a = a ◦ e = a. T R1 R1 −1 = 1 −1 = 1 1 1 (5. 3 × 3–matrix. R RT = RT R = 1 detR = 1 } 1. a ∈ G → ∃a−1 ∈ G with a ◦ a−1 = a−1 ◦ a = e. (i) Obviously. R3 = R1 R2 is a real. We would like to point out the obvious property that for all elements R of SO(3) holds R−1 = RT (5.11) (iii) From detR1 = 0 follows that R1 is non-singular and. In the following we will assume R1 . (iii) ∀a. (5.12) (iv) Since the associative law holds for multiplication of any square matrices this property holds for the elements of SO(3).5. It holds T T T T R3 R3 = ( R1 R2 )T R1 R2 = R2 R1 R1 R2 = R2 R2 = 1 . forms a group G satisfying the axioms (i) For any pair a. in fact.9) forms a group.13) . 1 T Similarly. 99 (5.1: Matrix Representation of SO(3) Deﬁnition of the Group SO(3) We will consider then the following set SO(3) = { real 3 × 3 matrices R. it holds 1. We have shown altogether that the elements of SO(3) form a group.10) detR3 = det ( R1 R2 ) = detR1 detR2 = 1 . We want to prove now that SO(3) as deﬁned in (5.

17) . −1 0 0 −1 . establish if 1-1 homomorphic mappings exist. J3 as follows i R(ϑ) = exp − ϑ · J (5. J2 . we arrive at the property exp ϑk LT k from which follows LT = −Lk . M4 } = 1 0 0 1 .14) together with matrix multiplication as the binary operation ‘◦’.1.18) (5. k = exp ( −ϑk Lk ) (5. 0 −1 1 0 (5. we want to assume that they exist. rotation by 180o . Lie Algebra We want to consider now a most convenient. will play a central role in the representation of many other symmetry transformations in Physics.20) (5. x2 –plane {rotation by Oo . −i} together with multiplication as the binary operation ‘◦’. the matrices i Lk = − Jk (5. rotation by 90o . (c) the set of four rotations in the x1 .19) (5.15) Below we will construct appropriate matrices Jk . i. (b) the set of matrices {M1 . Generators. −1.16) must be antisymmetric. This representation expresses rotation matrices in terms of three 3 × 3– matrices J1 . Geometric intuition tells us R(ϑk ek )−1 = R(−ϑk ek ) = exp ( −ϑk Lk ) ˆ ˆ Using the property which holds for matrix functions [f (R)]T = f (RT ) we can employ (5. presently. mappings between the sets which conserve the group properties: (a) the set of real and imaginary numbers {1. due to the uniqueness of the inverse. We want to show that for the property R RT = 1 to hold. rotation by 270o } and consecutive execution of rotation as the binary operation ‘◦’. identify subgroups. 0 1 −1 0 .100 Theory of Angular Momentum and Spin Exercise 5. compact representation of R(ϑ) which will be of general use and. i. M2 . To demonstrate this property consider rotations with ϑ = ϑk ek where ek ˆ ˆ is a unit vector in the direction of the Cartesian k–axis.e. in fact.1: Test if the following sets together with the stated binary operation ‘◦’ form groups. Lie Group.13) R(ϑk ek )−1 = R(ϑk ek )T = [exp ( ϑk Lk )]T = exp ϑk LT ˆ ˆ k and. M3 .

0.5. (5. A must have the form 0 a b A = −a 0 c .1: Matrix Representation of SO(3) 101 We can conclude then that if elements of SO(3) can be expressed as R = eA . ϑ2 .22). ϑ3 and three matrices L1 . L3 which are independent of the rotation angles. 0)T the rotation is the identity.23) (5. ϑ3 )T ] = lim sinϑ3 cosϑ3 0 = ϑ3 ϑ3 →0 ϑ3 →0 0 0 1 0 0 1 Insertion into (5.26) (5. ϑ3 )T such that for ϑ = (0.8) we can state cosϑ3 −sinϑ3 0 1 −ϑ3 0 1 0 . (5. 0. 0)T . the real matrix A is anti-symmetric.1. 0 0 0 0 0 0 (5. This property gives A then only three independent matrix elements. 0.21) −b −c 0 It is this the reason why one can expect that three-dimensional rotation matrices can be expressed through (5.2: Determine the generator L1 according to Eq. Exercise 5.22) for k = 3 yields L3 = ϑ−1 3 One obtains in this way 0 0 0 i − J1 = L1 = 0 0 −1 0 1 0 0 0 1 i − J2 = L2 = 0 0 0 −1 0 0 0 −1 0 i 1 0 0 − J3 = L3 = 0 0 0 The matrices Lk are called the generators of the group SO(3).(5.15) with three real parameters ϑ1 . One can determine then the matrices Lk deﬁned in (5. 0.25) 0 −ϑ3 0 0 −1 0 ϑ3 0 0 = 1 0 0 .e. i. L2 . Let us use this deﬁnition to evaluate L3 . ϑ2 .22) and similar for L2 and L3 . lim R[(0. We assume now that we parameterize rotations through a three-dimensional rotation vector ϑ = (ϑ1 . Using the deﬁnition of partial derivatives we can state L1 = lim ϑ−1 R(ϑ = (ϑ1 .15) as the derivatives of R(ϑ) with respect to the Cartesian components ϑk taken at ϑ = (0. 0)T ) − 1 1 1 ϑ1 →0 (5. Using (5.24) (5.27) ..

if the resulting products of exponential operators can be expressed in the form (5..g. The expression for Zn are actually rather complicated. The latter property implies that for any ϑ1 and ϑ2 there exists a ϑ3 such that exp i − ϑ1 · J exp i − ϑ2 · J = exp i − ϑ3 · J (5. etc.28) The construction of the generators through the limit taken in Eq. = 2. Z2 = 2 [A. J ] = i k m Jm .30) are called Lie groups. Finite rotations can be obtained by applications of many inﬁnitesimal rotations of the type R(ϑ = ( 1 .30) is called the Lie algebra. and the constants fk m are called the structure constants. the property (5.30) Groups with the property that their elements can be expressed like (5. (5. the elements of if any two indices are identical for k = 1. as we see later. Z3 = 12 ( [A. 3 )T ) = exp( 3 L3 ). 0. only the latter example and . 0. 2 . x2 -. i.. i. In case of the group SO(3) the structure constants are particularly simple. it must hold 3 [Lk .31) where k m are the elements which are 0 1 k m = 1 −1 of the totally antisymmetric 3-dimensional tensor.15) obey the group property. = 2.8) in case of ϑ = (0.g. R(ϑ = (0. e.33) We want to demonstrate now that exp(ϑ · L) yields the known rotational transformations.e. (5.15). 0)T ) = exp( 1 L1 ). We want to consider. in fact. 0)T ) = exp( 2 L2 ) and R(ϑ = (0. however.29) where Z1 = A + B and where the remaining operators Zn are commutators of A and B. 1 1 e. m = 3 . B]. An answer can be found considering the Baker-Campbell-Hausdorf expansion ∞ exp( λA ) exp( λB ) = exp( n=1 λn Zn ) (5. the matrix (5.15) will be closed under matrix multiplication only in the case that commutators of Lk can be expressed in terms of linear combinations of generators. = 2..15) and their generators obey the property (5.102 Theory of Angular Momentum and Spin Sofar it is by no means obvious that the generators Lk allow one to represent the rotation matrices for any ﬁnite rotation. The question is then. L ] = m=1 fk m Lm . [A. 0. three being a special case.32) For the matrices Jk = i Lk which. however. ϕ)T . m = 3 for any odd permutation of k = 1. the Lie algebra of SO(3) can be written [Lk . From this result one can conclude that expressions of the type (5.22) implies. B]. B] ). commutators of commutators of A and B.e. In fact. (5. Of course. L ] = k m Lm (5. Lie groups can have any number of generators. that the generators represent inﬁnitesimal rotations with 1 >> |ϑ| around the x1 -. B]] + [[A. are related to angular momentum operators. (5. that the operators (5. m = 3 for any even permutation of k = 1. holds the algebra [Jk . and x3 -axes.

1.41) cos ϕ − sin ϕ sin ϕ cos ϕ (5. using (5.5.39) Recognizing the Taylor expansions of cos ϕ and sin ϕ one obtains eA = and. 0 −1 1 0 = (−1)n 0 −1 1 0 (5. ∞ 2n = − 1 0 0 1 2n+1 (5.35) To determine expA = ∞ ν ν=1 A /ν! we split its Taylor expansion into even and odd powers 1 A2n + (2n)! 2 ∞ n=0 ∞ eA = n=0 1 A2n+1 .40) . (5. (5. that the matrix ϕL3 .36) The property 0 −1 1 0 allows one to write 0 −1 1 0 and.38) eA = n=0 (−1)n 2n ϕ (2n)! 1 0 0 1 ∞ + n=0 (−1)n ϕ2n+1 (2n + 1)! 0 −1 1 0 .1: Matrix Representation of SO(3) determine exp(ϕL3 ).34) One obtains then for the rotational transformation exp A 0 0 0 0 0 1 0 0 1 0 0 0 1 0 0 1 0 0 0 1 0 0 1 ∞ ν=1 ∞ ν=1 0 0 1 = = = + + 1 ν! 1 ν! A 0 0 0 0 0 ν Aν 0 0 0 0 0 0 0 1 ∞ 1 ν ν=0 ν! A 0 0 = eA 0 0 . eϕL3 which agrees with (5. cos ϕ − sin ϕ 0 = sin ϕ cos ϕ 0 0 0 1 (5. can be written ϕL3 = 0 103 A 0 0 0 0 . We note.37) = (−1)n 1 0 0 1 . according to (5. (2n + 1)! (5.34. accordingly.35). Exercise 5.27).8).3: Show that the generators Lk of SO(3) obey the commutation relationship (5. (5. A = ϕ 0 −1 1 0 . 5.31).

Deﬁnition We ﬁrst deﬁne how a rotational transformation acts on a wavefunction ψ(r ).e. Since this holds for any ψ(r ) one can conclude ρ R(ϑ1 ) ρ R(ϑ2 ) group property (5. Exercise 5. the transformation R(ϑ): C∞ ( ) → C∞ ( ) is related to the transformation R(ϑ): R → R .45) (5.47) = ρ R(ϑ1 ) R(ϑ2 ) . we will learn that transformation patterns and invariances are connected with the angular momentum states of quantum mechanics. For this purpose we require stringent continuity properties of the wavefunction: the wavefunctions under consideration must be elements of the set C∞ ( ). In analogy to R(ϑ) being a linear map R → R .2 Function space representation of the group SO(3) In this section we will investigate how rotational transformations act on single particle wavefunctions ψ(r). for A. (5. we deﬁne rotations R(ϑ) as linear maps C∞ ( ) → C∞ ( ). i.1: Assume the deﬁnition ρ ρ (A B) = ρ (B) ρ (A).42) Obviously.43) ρ( ) conserves the group property of SO(3).44) holds. in fact. (5. i. = ψ([ R(ϑ1 ) R(ϑ2 ) ]−1 r) = ρ R(ϑ2 ) ψ([R(ϑ1 )]−1 r) (5. one isomorphic to SO(3) and. hence. its elements can be represented through an exponential R(ϑ) = exp i − ϑ·J (5.2. Show that in this case holds Generators One can assume then that R(ϑ) should also form a Lie group.. B ∈ SO(3) holds ρ(A B) = ρ(A) ρ(B) .46) (5.104 Theory of Angular Momentum and Spin 5.e. complex-valued functions over the 3-dimensional space R which can be diﬀerentiated inﬁnitely often.44) This important property which makes ρ( SO(3) ) also into a group can be proven by considering ρ R(ϑ1 ) R(ϑ2 ) ψ(r) = ψ([R(ϑ2 )]−1 [R(ϑ1 )]−1 r ) = ρ R(ϑ1 ) ρ R(ϑ2 ) ψ(r) . In particular. namely R(ϑ)ψ(r ) = ψ(R−1 (ϑ) r ) . i.e. a relationship which we like to express as follows R(ϑ) = ρ R(ϑ) . the R(ϑ) ψ(r) = ψ(R(ϑ) r ). (5.48) .

0. i − J3 f (r) = = Using (5.2: Function Space Representation of SO(3) where the generators can be determined in analogy to Eq.2: (a) Derive the generators Jk . 0.54) (5. . (5.2. 0. (5. J3 ] and [J3 .5. (5. J2 ].49) to a function f (r). show that (5. i.56) We like to verify this property for [J1 . (5. in case of J3 . −ϑ3 ) r ) − f (r) ) . (5.57) Exercise 5. but also the Lie algebra of the generators. it holds [Jk .55) Not only is the group property of SO(3) conserved by ρ( ). J1 ]. 0)T ) − 1 1 ϑ1 →0 105 (5.49). In fact.24) one can expand ϑ3 →0 ϑ3 →0 lim ϑ−1 R(ϑ = (0.(5.56) holds for [J2 .52) ( x2 ∂1 − x1 ∂2 ) f (r) . 5.50) and Taylor expansion yields i − J3 f (r) = = ϑ3 →0 lim ϑ−1 ( f (r) + ϑ3 x2 ∂1 f (r) − ϑ3 x1 ∂2 f (r) − f (r) ) 3 (5. J ] = i k m Jm . − i J ] f (r) 1 2 =− 2× [ (x3 ∂2 − x2 ∂3 ) (x1 ∂3 f − x3 ∂1 f ) − (x1 ∂3 − x3 ∂1 ) (x3 ∂2 f − x2 ∂3 f ) ] 2 +x1 x3 ∂2 ∂3 f − x1 x2 ∂3 f − x2 ∂1 ∂2 f + x2 x3 ∂1 ∂3 f + x2 ∂1 f 3 2(x ∂ 2 1 2 −x1 ∂2 f − x1 x3 ∂2 ∂3 f + x2 ∂1 ∂2 f + x1 x2 ∂3 f − x2 x3 ∂1 ∂3 f ] 3 = − − x1 ∂2 ) f (r) = i J3 f (r) .42)) and similarly for J2 and J3 . 0. −ϑ3 ) r ≈ −ϑ3 1 0 x2 = −ϑ3 x1 + x2 .53) (5.53–5..49) (note the minus sign of −ϑ1 which originates from the inverse of the rotation in Eq.55) and f (r) = f [J1 .e. 0 0 1 x3 x3 (5. 2 by means of limits as suggested in Eq. ϑ3 )T ) f (r) − f (r) 3 (5. The generators Jk can be determined by applying (5.23.51) Inserting this into (5. One obtains using (5. J2 ] f (r) = − = − 2[ 2 [− i J .22) according to i − J1 = lim ϑ−1 R(ϑ = (−ϑ1 . k = 1.50) lim ϑ−1 ( f ( R(0. 3 1 ϑ3 0 x1 x1 + ϑ3 x2 R(0. Carrying out similar calculations for J1 and J2 one can derive i − J1 i − J2 i − J3 = = = x3 ∂2 − x2 ∂3 x1 ∂3 − x3 ∂1 x2 ∂1 − x1 ∂2 .

62) J1 [J1 .59) which. 2.53–5. φ) m (θ. according to the correspondence principle between classical and quantum mechanics.61) simultaneous eigenstates of J 2 and of one of the Jk .58) i = r ×p ˆ (5.3 Angular Momentum Operators The generators Jk can be readily recognized as the angular momentum operators of quantum mechanics. J3 ] = = = 2 2 [J1 + J2 . The commutation relationships imply that no simultaneous eigenstates of all three Jk exist. φ).2. C] = A[B.106 Theory of Angular Momentum and Spin Exercise 5. J3 ] J2 −i J1 J2 − i J2 J1 + i J2 J1 + i J1 J2 = 0 According to (5. Equation (5. identiﬁes J as the quantum mechanical angular momentum operator. Using [AB. φ) = = 2 ( + 1) Y m (θ. Jk ] = 0 .56) are the famous commutation relationships of the quantium mechanical angular momentum operators.60) commutes with all three Jk . (5. To show this we ﬁrst note that (5. J3 ] (5.56) one obtains [J 2 . C] + [A. φ).64) mY m (θ. We will show now that the properties of spherical harmonics J2Y J3 Y m (θ.61) We demonstrate this property for k = 3. usually chosen as J3 . Show that for J3 as deﬁned above holds exp( iα J3 ) ψ(φ) = ψ(φ + α). x2 = r sinθ sinφ. [J 2 .3: Consider a wave function ψ(φ) which depends only on the azimutal angle φ of the spherical coordinate system (r. However. and x3 = r cosθ. C]B and (5. k = 1. the system related to the Cartesian coordinates through x1 = r sinθ cosφ.65) . φ) (5. i. 3 . (5. J3 ] + [J2 ..e. 2. φ) as well as the eﬀect of the so-called raising and lowering operators J± = J1 ± i J2 (5. θ. J3 ] + [J1 . the operator 2 2 2 J 2 = J1 + J2 + J3 (5. using the momentum operator p = ˆ J = r × i . k = 1. i. can be found. 3.55) can be written as a vector product of r and i − J = −r × From this follows. 5. . These eigenstates are the well-known spherical harmonics Y m (θ.e. J3 ] J1 + J2 [J2 .63) (5.

φ) 107 = = 0 = = 0 ( + m + 1)( − m) Y ( + m + 1)( − m) Y m+1 (θ.e. L1 − iL2 ] = −2iL3 and L+ | = 0 one obtains L+ L− | = −2iL3 | = −2 | .56). Using [L3 . L ] = k m Lm and let L± = L1 ± i L2 . The arguments are very similar to the ones used above and we can be brief. φ) m+1 (θ. We will show that this property holds then also for m − 1. φ) are a consequence of the Lie algebra (5. (∗) (5. φ) (θ. It holds L+ L− | = (L+ L− − L− L+ + L− L+ )| = ([L+ . α −1 β −1 = 2 .3: Angular Momentum Operators on the spherical harmonics.e.67) (5. i. We note L3 L− | = ( [L3 . The deﬁnitions of the coeﬃcients α −1 and β −1 yield L+ L− | = L+ ( iβ −1 | − 1 = − α −1 β −1 | .74) (5. For this purpose we prove the following theorem.76) We now show that property (ii) also holds for m = − 1 proceeding in a similar way. 3 be operators acting on a Hilbert space H which obey the algebra [Lk . L− ] + L− L+ )| .70) (5.69) J+ Y (θ.1 Let Lk . Noting [L+ . 2. We consider L+ L− | m = ( [L+ . Let there exist states | in 1 2 3 H with the property L+ | L3 | then the states deﬁned through L− | m + 1 = −i βm | m have the following properties (i) L+ | m (ii) (iii) L3 | m L2 | m = −i αm | m + 1 = −i m | m = − ( + 1) | m (5. J+ Y J− Y J− Y m (θ. L− ] = [L1 + iL2 . L− ] = [L3 . k = 1.. For this purpose we ﬁrst demonstrate that the property holds for m = − 1 by considering (i) for the state L− | ∼ | − 1 . φ) (5. L1 − iL2 ] = L2 + iL1 = i (L1 − iL2 ) = iL− and L3 | = −i | we obtain L3 L− | = ( iL− + L− L3 )| = i(− + 1) L− | .5. φ) − m (θ. L2 = L2 + L2 + L2 . L+ raises the m-value of | − 1 from m = − 1 to m = .66) (5.68) (5. and since we have already shown | − 1 = (iβ −1 )−1 L− | we can state L3 | − 1 = −i( − 1)| − 1 . An Important Theorem Theorem 1. Continuing our proof through induction we assume now that property (i) holds for m.72) = 0 = −i | (5. i.75) (5. L− ] + L− L+ )| m = ( −2iL3 + . namely. From this follows that L− | is an eigenstate of L3 . in fact. L− ] + L− L3 ) | .71) To prove this theorem we ﬁrst show by induction that (i) holds.73) (5.

e.73.77) We will show now that if (ii) holds for m.e. (5.66. It follows then L | m = − 2 αm−1 βm−1 + 2 αm βm + m | m = 1 − 2 ( + m + 1)( − m) + 1 ( + m)( − m + 1) + m2 = − ( + 1)| m . it also holds for m − 1. We note that we can write L2 = 1 1 1 1 2 2 2 2 L+ L− + 2 L− L+ + L3 .70. We will have various opportunities to employ the Theorem just derived. given in (5. (5.83) (5. in particular.63. Using the familiar formula + m − n k=0 k = n (n + 1)/2 − (m + 1) m = ( + 1) m − (m + 1) m (5. in terms of spherical coordinates deﬁned through x1 x2 x3 = = = r sin θ cos φ r sin θ sin φ r cos θ (5.78) ( + m + 1)( − m) ( + m + 1)( − m) . 2 We note that Eqs. it holds αm βm = 2 one obtains αm βm = = ( + 1) k=m+1 k . 5. θ. We can deduce from (∗) and (∗∗) α α α −1 β −1 −2 β −2 −3 β −3 = 2 = 2( − 1 + ) = 2( − 2 + · · −1 + ) Obviously..82) (5.84) .81) i. For this purpose we need to express the angular momentum operators in terms of spherical coordinates (r. This completes the proof of the theorem.64). ﬁnally αm = βm = (5.79) read L+ | m L− | m + 1 = −i = −i ( + m + 1)( − m) | m + 1 ( + m + 1)( − m) | m . 5. We note L3 L− | m = ( [L3 . 5.108 Theory of Angular Momentum and Spin L− L+ )| m = (−2m − αm βm )| m .. 5.72. A ﬁrst application will involve the construction of the eigenstates of the angular momentum operators as deﬁned in (5. Angular Momentum Operators in Spherical Coordinates We want to express now the generators Jk . from L+ L− | m = − αm−1 βm−1 | m follows the recursion relationship αm−1 βm−1 = 2m + αm βm . One can normalize the states | m such that αm = βm . This implies that (i) holds for m − 1. Before we can ﬁnally show that property (iii) holds as well for all − ≤ m ≤ we need to determine the coeﬃcients αm and βm . 5. (∗∗) (5.69).53–5.79) We can now show that (iii) holds for all proper m–values. yield properties (5. L− ] + L− L3 )| m = ( iL− − i m L− )| m = −i (m − 1) L− | m from which we can deduce L3 | m − 1 = −i(m − 1)| m − 1 . φ. i.55). Again the argumemts are similar to the ones used above.80) (5.

We like to express now the operators J3 and J 2 .85).93) and similarly.60).88) (5. applying repeatedly the chain rule. (L1 )2 = = ∂ 2 ∂ + cot θ cos φ ∂θ ∂φ 2 ∂ ∂ 1 ∂2 sinφ cosφ sin2 φ 2 − + 2 cotθ sinφ cosφ ∂θ ∂φ ∂θ∂φ sin2 θ ∂2 ∂ + cot2 θ cos2 φ 2 + cotθ cos2 φ ∂θ ∂φ sin φ i ∂φ .82–5.87) To derive these properties we consider. According to (5. cos θ ∂ cos2 φ + sin θ cos φ ∂φ cos2 θ sin θ sin φ ∂ sin θ sin φ ∂ + + sin2 θ f (θ.91) which agrees with expression (5.92) (5.5. Using (5. (L2 )2 = − cos φ ∂ ∂ + cot θ sin φ ∂θ ∂φ 2 . the latter deﬁned in (5.85). In fact. φ) (5. involve only the angular variables θ and φ and not the radius r. φ). (5. φ) ∂x2 /x1 ∂ ∂x3 /r ∂ = x3 + x3 ∂x2 ∂ tan φ ∂x2 ∂ cos θ ∂x2 /x1 /r ∂ ∂ − x2 − x2 ∂x3 3 ∂ cos θ f (θ. in terms of spherical coordinates. using ∂/∂tanφ = cos2 φ ∂/∂φ and ∂/∂cosθ = −(1/sinθ) ∂/∂θ.86). using (5. φ) = (5. φ) sin θ ∂θ sin θ ∂θ L1 f (θ.86) (5. we will prove − − − i i i J1 J2 J3 = L1 = L2 = L3 ∂ ∂ + cot θ cos φ ∂θ ∂φ ∂ ∂ = − cos φ + cot θ sin φ ∂θ ∂φ ∂ = − .85) (5. φ) = (x3 ∂2 − x2 ∂3 )f (θ. φ) ∂x ∂x3 ∂ tan φ =0 (5. L1 f (θ.89) = ∂ x2 x2 ∂ x2 + x2 ∂ x3 − 33 − x2 1 3 2 x1 ∂ tan φ r ∂ cos θ r ∂ cos θ f (θ. actually. for example.90) This yields.84 ) one obtains. using (5. L1 f (θ.3: Angular Momentum Operators 109 We ﬁrst like to demonstrate that the generators Jk . φ) = ( x3 ∂2 − x2 ∂3 ) f (θ.87) holds J3 = To determine J 2 we note. ∂φ = sin φ (5.

5. C∞ (S ). sphere S2 .100) (5. one can conclude J2 = − 2 sin θ 2 sinθ ∂ ∂θ 2 + ∂2 ∂φ2 . 5.97).85. i. 5. The ˙ corresponding expression for the quantum mechanical kinetic energy operator is ˆ T = − which follows from the identity 2 2 2 2m = − 1 ∂2 J2 r + .64) can be constructed as follows. 1 sin2 θ sinθ ∂ ∂θ (5.96) and the relationship between Lk and Jk as given in (5. 5.110 = cos2 φ Theory of Angular Momentum and Spin ∂2 1 ∂ ∂2 + sinφ cosφ − 2 cotθ sinφ cosφ ∂θ2 ∂φ ∂θ∂φ sin2 θ 2 ∂ ∂ + cotθ sin2 φ + cot2 θ sin2 φ 2 . 5.101) 1 ∂2 r r ∂r2 and comparision with (5.87). 2m r ∂r2 2m r2 2 (5.94) It follows (L1 )2 + (L2 )2 + (L3 )2 = With cot2 θ + 1 = 1/sin2 θ.99) = = ∂2 2 ∂ 1 + + 2 2 2 ∂r r ∂r r sin θ ∂2 2 ∂ + 2 ∂r r ∂r sinθ ∂ ∂θ 2 + ∂2 ∂φ2 (5.97) Kinetic Energy Operator The kinetic energy of a classical particle can be expressed J2 p2 p2 = r + class 2m 2m 2m r2 (5.95) sinθ ∂ ∂θ = ∂2 ∂ + cotθ .4 Angular Momentum Eigenstates According to the theorem on page 107 above the eigenfunctions (5. namely on the space of complex-valued functions on the unit 3-dim. The functions in C∞ (S ) have real variables . One ﬁrst identiﬁes the Hilbert space H on which the generators Lk operate.86. 2 ∂θ ∂θ (5.e.85. These generators act on a subspace of C∞ ( ). 5.98) where Jclass = r × mv is the angular momentum and pr = mr is the radial momentum.63. ∂θ ∂φ ∂2 ∂ ∂2 + cotθ + ( cot2 θ + 1 ) 2 . ∂θ2 ∂θ ∂φ (5. (5.86.87). The operators in the present case are Lk = − i Jk where the Jk are diﬀerential operators in θ and φ given in (5.

φ) = ∆( .106) The latter property is obviously satisﬁed for the chosen φ-dependence.104. . One obtains in this way Y m (θ. . The norm in H is deﬁned through |f |2 = S2 dΩ f ∗ (θ. (5. normalizing these functions. (5.110) = 0 (5.105) explicitly. m) = 1 −m J− 1 2 Y (θ. .104) (5. .109) using (5.108) m (θ. φ. φ) = 2 + 1 ( − m)! P 4π ( + m)! m (cosθ) eimφ . . must be cyclic in φ with period 2π. } applying L− Y m+1 (θ. For this purpose we split oﬀ a suitable normalization factor as well as introduce the assumption that the dependence on φ is described by a factor exp(imφ) Y It must hold L+ Y (θ. The function space endowed with this norm and including only functions for which the integral (5. φ). φ) = −i Y (θ. − 2.63. φ) (5. m) Constructing Y → Y →→ ··· Y ( + m)! (2 )! ( − m)! −1 − We like to characterize the resulting eigenfunctions Y m (θ.85–5.4: Angular Momentum Eigenstates 111 θ. φ) in H which satisfy L+ Y (θ.87). .102) exists. . 0 ≤ φ < 2π. and to be admissable for description of quantum states. φ) = = 0 −i Y (θ. (5. φ) = −i ( + m + 1)( − m)Y m (θ. (5. φ) L3 Y (θ. 5.102) where dΩ = sinθ dθ dφ is the volume element of S2 .107) (5. .107) reads ∂ ∂ + i cotθ ∂θ ∂φ P (cosθ) ei φ i e±iφ ∂ ∂ ± i cotθ ∂θ ∂φ . and then determining the family {Y m (θ. φ) . The raising and lowering operators L± = L1 ± i L2 . 0 ≤ θ < π. − .103) iteratively for m = − 1.105) ∆( . the so-called spherical harmonics. The eigenfunctions (5. can be expressed L± = Accordingly. φ) (5. φ). (5. m = − . − + 1. by carrying out the construction according to (5.64) can be constructed then by seeking ﬁrst functions Y (θ. φ). φ) (5. φ) = 0 as well as L3 Y (θ. is indeed a Hilbert space.111) .5. 5. φ)f (θ.

5. (5. A proper solution of this equation.e. one for which π dθ sinθ |P (cosθ)|2 (5. ”Classical Electrodynamics.112) where we employed the deﬁnition of P (cosθ) in (5. (5.113) 0 is ﬁnite.118) 2 +1 1 1 sin θ ei 4π (2 )! 2 ! φ . .106.116) implies |N |2 from which we conclude N = (−1) 1 · 3 · 5 · · · (2 − 1) (5. One obtains.119) provide the following expression for Y Y (θ.120) See.119) where the factor (−1) has been included to agree with convention1 . Jackson (John Wiley. New York.114) dθ sinθ |Y (cosθ)|2 = 1 0 (5. (5.112 or Theory of Angular Momentum and Spin ∂ − ∂θ cotθ P (cosθ) = 0 (5. 5. 5. 2nd Ed. = = x (1 − x2 ) 2 x3 3 + 2 −1 +1 −1 −1 dx x2 (1 − x2 ) + 2 · (2 − 2) 1·3 −1 +1 (1 − x2 ) dx x4 (1 − x2 ) −1 −2 +1 2 · (2 − 2) · · · 2 dx x2 1 · 3 · 5 · · · (2 − 1) −1 2 · (2 − 2) · · · 2 2 (2 )! 2 = 2 2 +1 1 · 3 · 5 · · · (2 − 1) 2 + 1 [1 · 3 · 5 · · · (2 − 1)] (5.. . 1975) . using the new integration variable x = cos θ. (5.D.116) The integral appearing in the last expression can be evaluated by repeated integration by parts. for example. π +1 dθ sin2 0 +1 θ = −1 +1 −1 dx (1 − x2 ) +1 = = . φ) = (−1) 1 1 = 1 [1 · 3 · 5 · · · (2 − 1)]2 (5.107). i.” by J.106.114.115) holds. We note that (5. is P (cos θ) = N sin θ as one can readily verify.114) yield |N |2 2 +1 1 2π 4π (2 )! π dθ sin2 0 +1 θ = 1. The normalization factor N is chosen such that π (5.117) Accordingly.

123). as speciﬁed in (5.124).127) ∂ −m−1 (x2 − 1) . according to (5. are real. therefore. Actually.126) Then holds. i.103. through repeated application of (5. and can obtain now..121) states then for the functions P m (cos θ) ∂ + (m + 1) cotθ ∂θ P m+1 (cos θ) = − ( + m + 1) ( − m) P m (cos θ) .124) reads P (x) = = (2 )! (1 − x2 )− 2 (x2 − 1) 2 ! (−1) 1 · 3 · 5 · · · (2 − 1) sin θ (5.124) called the associated Legendre polynomials. − deﬁned in (5. describes the φ-dependence of Y m (θ. P m (x) = −1 ( + m)! 2 ! ( − m)! × (1 − x2 )− m+1 2 −(1 − x2 ) 2 1 ∂ x + (m + 1) 1 ∂x (1 − x2 ) 2 × (5. ∂x −m−1 . The reader should note that the associated Legendre polynomials. We want to demonstrate now that the recursion equation (5. namely Y (θ.106). φ). . .110) together with (5. − 2.110).125) which agrees with (5.122) The latter identity can be written. m = − 1. (5. 5. For m = (5.63. φ). m (cos θ) e (5. P m (x) = −1 × ( + m + 1) ( − m) × −(1 − x2 ) 2 1 (5.64). we seek to determine the functions P m (cos θ) and.114. Let us then assume that (5. To prove (5. 5.124) holds for m + 1. . φ).5. 5. indeed. every application of L− reduces the power of eiφ by one.121) This expressions shows that the factor eimφ .e. use (5.103. 5. (5.124) we proceed by induction.119). employing again the variable x = cos θ.123) P m+1 (x) ∂ x + (m + 1) 1 ∂x (1 − x2 ) 2 .106) i e−iφ ∂ ∂ − i cotθ ∂θ ∂φ 2 + 1 ( − m − 1)! P 4π ( + m + 1)! 2 + 1 ( − m)! P 4π ( + m)! i(m+1)φ m+1 (cos θ) e = −i ( + m + 1)( − m) imφ .123) leads to the expression P m (x) = −m m ∂ 1 ( + m)! (1 − x2 )− 2 (x2 − 1) 2 ! ( − m)! ∂x −m (5. the eigenfunctions Y m (θ. P m+1 (x) = −m−1 m+1 ∂ 1 ( + m + 1)! (1 − x2 )− 2 (x2 − 1) .4: Angular Momentum Eigenstates 113 We have constructed a normalized solution of (5. 2 ! ( − m − 1)! ∂x −m−1 (5.

namely.135) . It holds. We ﬁrst determine Y − (θ. However. 5.130) (5.106. appears not very interesting.114 By means of −(1 − x2 ) 2 = Theory of Angular Momentum and Spin −m−1 m+1 ∂ ∂ (1 − x2 )− 2 (x2 − 1) ∂x ∂x −m−1 −m m ∂ −(1 − x2 )− 2 (x2 − 1) ∂x −m −m−1 m+1 ∂ x (1 − x2 )− 2 − (m + 1) 1 ∂x −m−1 (1 − x2 ) 2 1 (5. P − (5. We want to test if the recursion (5. if L− Y (θ. φ) choosing the proper sign. φ) as given in (5. In that case (1 − x2 ) is a polynomial of degree 2 and. φ) (5.132). it holds then for all m. by repeated application of the operator L+ . Since (5. expression (5. φ) = (−1)m Y m −m (θ. Such construction reproduces the eigenfunctions Y m (θ. which is equivalent to recursive application of L− . φ) ..127) reproduces (5. and constructs then the eigenfunctions Y − +1 . therefore. i. etc. therefore.124). Y ∗ (θ. that expression (5.124) and. from such construction emerges an important symmetry property of Y m (θ.132) which reduces the number of spherical harmonics which need to be evaluated independently roughly by half .124) holds for m = . (5. it holds ∂2 (x2 − 1) = (2 )! ∂x2 and.124) holds for m if it holds for m + 1.133) Since the term with the highest power of (x2 − 1) is x2 . φ) = 0 (5.106.e. φ) = Y − (θ.124) and.134) (x) = 1 2 ! (1 − x2 ) 2 . we embark on this construction in order to prove (5.131) identiﬁes f (θ. Y − +2 . φ) = = determines ﬁrst a normalized solution 0 i f (θ.128) one can show that (5. φ) L3 f (θ. 5. φ) An alternative route to construct the eigenfunctions Y of L− f (θ.123) terminates for m = − . using x = cos θ. 2 ! (2 )! ∂x2 1 (5. yields a vanishing expression for m = − − 1 as long as is a non-negative integer.124). (5. In fact. φ) using (5. Constructing Y − → Y − +1 → ···Y m (θ. the derivative ∂ 2 +1 /∂x2 +1 of this expression vanishes. P − (x) = 1 ∂2 (1 − x2 ) 2 (x2 − 1) .129) holds. therefore. φ). hence. therefore.

106.80) and (5. According to (5.122.142) For this purpose we proceed by induction. (5. (5.138) Employing (5.120) for Y obey the postulated relationship (5. φ) = 2 +1 1 1 sin θ e−i 4π (2 )! 2 ! φ 115 − .139) = ∂ − m cotθ ∂θ P m (cos θ) .e..136) is normalized and has the proper sign. (5. (5.141) We want to demonstrate now that this recursion equation leads to the expression P m (x) = +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) 2 ! ∂x +m (5.4: Angular Momentum Eigenstates Due to (5.136) We note in passing that this expression and the expression (5. following closely the proof of eq. 5.137) m (θ. (5. φ) sin θ e−i φ = 0. i. We ﬁrst note that for m = − (5.106).136) to construct all other Y recursion eiφ ∂ ∂ + i cotθ ∂θ ∂φ Y m (θ. the expression (5. (5.132).5. One can readily verify that this expression provides a solution of (5.124). a sign consistent with the family of functions Y m constructed above.140) Introducing again the variable x = cos θ leads to the recursion equation [c. Obviously.123)] P m+1 (x) = −(1 − x2 ) 2 1 ∂ x − m 1 ∂x (1 − x2 ) 2 P m (x) . φ) .142) yields P − (x) = (−1) 1 (1 − x2 )− 2 (x2 − 1) = (1 − x2 ) 2 2 ! 2 ! (5.124) one arrives at Y (θ. φ). (5.143) . This follows from the identity ∂ ∂ − i cotθ ∂θ ∂φ One can use (5. 5.f. one can conclude for the associated Legendre polynomials ( − m)! ( + m)! = or P m+1 (cos θ) ∂ − m cotθ ∂θ P m (cos θ) ( + m + 1)( − m) ( − m − 1)! P ( + m + 1)! m+1 (cos θ) (5.129).110) applies the = ( + m + 1)( − m) Y m+1 (θ.

one can conclude P m (x) = (−1)m m (θ. are identical.144) reads P m+1 (x) = +m+1 m+1 ∂ (−1)m+1 (1 − x2 ) 2 (x2 − 1) .141) reads then P m+1 (x) = 1 (5.e.148) According to (5. of which agrees with the r. Hence. (5. We then assume that (5.149) (5. 1 (1 − x ) ∂x +m (1 − x2 ) 2 1 (5.h.144) ∂ x − m 1 ∂x (1 − x2 ) 2 +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) .s. Since (5.124). yields the same eigenfunctions as the previous construction beginning with Y and stepping down the series of functions Y −1 .147) the r.132). (5. 2 ! ∂x 1 (5.142) holds for m. equivalently. of (5. (5. (5.146) i. However. i.116 Theory of Angular Momentum and Spin wich agrees with (5.. Y −2 .124) yields (−1)m ( − m)! P ( + m)! −m (x) = +m m ∂ (−1)m (1 − x2 ) 2 (x2 − 1) .142) one can state P (x) = ∂ (x2 − 1) .142) and by (5. According to both (5.135).s.150) .124) and (5. etc. m → −m − 1 yields −(1 − x2 ) 2 = +m m ∂ ∂ (1 − x2 ) 2 (x2 − 1) ∂x ∂x +m +m+1 m+1 ∂ −(1 − x2 ) 2 (x2 − 1) ∂x +m+1 +m x 2 m ∂ 2 +m .142) holds for m + 1 if it holds for m. given by (5. also the associated Legendre polynomials determined this way.142). Accordingly.. P (x) = P 0 (x) are called Legendre polynomials. 2 ! ∂x +m+1 (5. 2 ! ∂x +m (5.145) Hence.128) m + 1 → −m or. Y − +2 . The construction beginning with Y − and continuing with Y − +1 .h.e. 2 ! ∂x +m −(1 − x2 ) 2 Replacing in (5.106) this implies for Y The Legendre Polynomials The functions the identity (5.142) holds for m = − we veriﬁed that it holds for all m. etc. one notes that application of (5. φ) ( − m)! P ( + m)! −m (x) .

J. (5. In case q = 1 and |r2 | < |r1 | holds the identity2 ∞ 1 r2 = P (cos γ) (5. For example. N. t) = √ 1 − 2xt + t2 ∞ 1 − 2 x t + t2 .. 1. Lebedev (Prentice-Hall. Englewood Cliﬀs.153) P (cos θ) eimφ .150) and (5. 45 of “Special Functions and their Applications” by N. Using x = cos γ. in case of x = 1 holds w(1.154) |r1 − r2 | r +1 =0 1 where γ is the angle between r1 and r2 . t = r2 /r1 .4: Angular Momentum Eigenstates The ﬁrst few polynomials are P0 (x) = 1 .156) yields P (1) = 1 . . the spherical harmonics for m ≥ 0 Y m (θ. P1 (x) = x . for example. accordingly. 2.152) and. pp. =0 (5. (5. r1 is the point where the potential is measured.157) = 0. e.132). Jackson (John Wiley. .N.” by J.158) see. φ) = 2 + 1 ( − m)! (−1)m sinm θ × 4π ( + m)! × ∂ ∂ cos θ m (5.155) P (x) t . t) is called a generating function of the Legendre polynomials3 .g. 2 3 1 1 = = 2 1−t 1 − 2t + t ∞ t .D. 92 To prove this property see. 1975).142) allows one to express the associated Legendre polynomials in terms of Legendre polynomials P m (x) = (−1)m (1 − x2 ) 2 m ∂m P (x) . pp. P3 (x) = 1 2 117 P2 (x) = − 3x) 1 2 2 (3x − 1) . . The Legendre polynomials arise in classical electrodynamics as the expansion coeﬃcients of the electrostatic potential around a point charge q/|r1 − r2 | where q is the charge. and |r1 − r2 | = r1 (5. The generating function allows one to derive useful properties of the Legendre polynomials. 2nd Ed. The spherical harmonics for m < 0 can be obtained using (5.5. 1965) which is an excellent compendium on special functions employed in physics. . and r2 denotes the location of the charge. m ≥ 0 ∂xm (5. t) = √ Comparision with (5. ”Classical Electrodynamics.154) can be written 1 = w(x. New York.156) w(x. (5..151) (5x3 Comparision of (5. =0 (5.

φ) m (θ.167) (5.159) by w(1 − 2xt + t2 ) leads to the diﬀerential equation obeyed by w(x. ∂t 1 − 2xt + t2 (5. Noting exp[im(π + φ)] = (−1)m exp(imφ) we determine Y Properties of Y m (θ. Accordingly.161) Collecting coeﬃcients with equal powers t yields 0 = P1 (x) − x P0 (x) ∞ (5. using P0 (x) = 1 [c.142) allows one to conclude P m (−x) = (−1) +m P m (x) (5.159) Using ∂lnw/∂t = (1/w) ∂w/∂t and multiplying (5. Due to cos(π − θ) = − cos θ the replacement r → −r alters P m (x) into P m (−x). under inversion Y m (θ. φ) goes over to Y m (π − θ.162) − (2 + 1)x P (x) + P −1 (x) ] t + =1 [( + 1) P +1 (x) The coeﬃcients for all powers t must vanish individually. φ) Under inversion at the origin vectors r are replaced by − r. (5. (5. Inspection of (5.168) mY m (θ. π − θ. Inversion Symmetry of Y m (θ. t) holds Theory of Angular Momentum and Spin ∂lnw(x. ∂t P (x)t −1 −1 (5.166) We want to summarize the properties of the spherical harmonics Y derived above.156) and (∂/∂t) to P (x)t = ∞ ∂w + (t − x) w = 0 . π + φ) = (−1) Y m (θ. φ) . φ) = = 2 ( + 1) Y m (θ. θ. φ) m (θ.163) (5.f. . t) x−t = . t) (1 − 2xt + t2 ) Employing (5. φ) (5. . The spherical harmonics are eigenfunctions of the angular momentum operators J2Y J3 Y m (θ. then the coordinates of − r are (r. π + φ). π + φ). Accordingly. φ) m (π − θ.118 For w(x. holds P1 (x) ( + 1) P +1 (x) = = x P0 (x) (2 + 1) x P (x) − P −1 (x) (5.164) which.165) since ∂ n /∂(−x)n = (−1)n ∂ n /∂xn . . If the spherical coordinates of r are (r. 1.160) this diﬀerential equation is equivalent ∞ (1 − 2xt + t2 ) =0 P (x) t + (t − x) =0 P (x) t = 0 (5. φ). . φ) ..151)] allows one to determine P (x) for = 1. 2.

m (5. (5. a complete basis of C∞ (S2 ).172) (5. where ( = 1. uniquely deﬁned functions over the unit sphere S2 which are inﬁnitely often diﬀerentiable π 2π sin θdθ 0 0 dφY ∗m (θ. The spherical harmonics obey the recursion relationships J+ Y J− Y m (θ. 4π (5. in fact.176) = 1 +1 [ (2 + 1) x P (x) − are the Legendre polynomials. except for the factor exp(imφ).5.175) (5.173) m+1 (θ. 5. φ) C = =0 m=− π C sin θdθ 0 m Y m (θ. φ) where J± = J1 ± iJ2 .) P0 (x) = 1 . φ) (5. P +1 (x) P1 (x) = x .4: Angular Momentum Eigenstates 119 2.169) 3. φ)f (θ. φ) = (−1)m Y ∗ (θ.178) 7. φ) = = ( + m + 1)( − m) Y ( + m + 1)( − m) Y m+1 (θ. For the Legendre polynomials holds the orthogonality property +1 dx P (x) P ( (x) = −1 2 δ 2 +1 . The spherical harmonics form. The spherical harmonics for m < 0 are given by Y −m (θ. φ) 2π 0 (5. The spherical harmonics are given by the formula Y m (θ. (5.e. φ) Y m (θ. φ) ∈ C∞ (S2 ) holds ∞ f (θ.177) Note that the spherical harmonics are real. φ) = δ δm m . m 4. φ) m (θ.170) (5. . for any f (θ. φ) . φ) .174) m ≥ 0 P (cos θ) eimφ .. φ) = 2 +1 P (cos θ) .179) . 2. 6. φ) = 2 + 1 ( − m)! (−1)m sinm θ × 4π ( + m)! 2 ! × ∂ ∂ cos θ m (5. The spherical harmonics form an orthonormal basis of the space C∞ (S2 ) of normalizable.171) m = dφ Y ∗ (θ. . i. P −1 (x) ] (5. The spherical harmonics Y 0 are Y 0 (θ. .

4.2: Derive the orthogonality property for the Legendre polynomials (5. Exercise 5.185) (5.1: Derive expressions (5.4.120 8.183) (5.179) using the generating function w(x. (5.189). .s. 2 are given by Y00 = 1 √ 4π 3 sin θ eiφ 8π 3 cos θ 4π 1 4 − 15 sin2 θ e2iφ 2π 15 sin θ cos θ eiφ 8π 5 4π 3 1 cos2 θ − 2 2 (5. φ). φ) . Exercise 5.. of (5. =0 t + −1 dxP (x) P (x) . (5.181) = 0. 1. 4π (5. For the Laplacian holds 2 = = = (5.177). For this purpose start from the identity +1 ∞ +1 dx w(x.h. φ) = δm0 2 +1 . t)2 = −1 . 11. (5.188) Exercise 5.4.86).156). The spherical harmonics obey the inversion symmetry Y 10.s.87).3: Construct all spherical harmonics Y3m (θ.h.186) (5. t) stated in (5.187) h(r) Y m (θ. φ) = 1 ∂2 r − r ∂r2 ( + 1) r2 h(r) Y m (θ. φ) = (−1) Y m (π − θ. π + φ) .184) Y22 Y21 Y20 together with (5.189) evaluate the integral on the l.180) 9. The spherical harmonics for θ = 0 are Y m (θ Theory of Angular Momentum and Spin = 0. expand the result in powers of t and equate the resulting powers to those arising on the r.182) Y11 Y10 = = − (5. (5. The spherical harmonics for m (θ.

and (ii) they form the lowest dimensional sets X obeying C∞ (S ) = X which have this invariance property. φ ) exp m One can also represent C∞ (S ) C∞ (S ) = [ {Y . (5. . assumes a particularly simple form. . = − .193) where [ { } ] denotes closure of a set by taking all possible linear combinations of the elements of the set. = . We denote the image of a function f (θ.m D(ϑ) m. (5. φ) = exp Since the spherical harmonics Y space C∞ (S ) one can expand ˜ f (θ. = 0. The transformations exp − i ϑ · J . φ) i − ϑ·J f (θ.197) The subspaces X have the important property that (i) they are invariant under rotations exp − i ϑ · J .193) shows ∞ = − .196) C∞ (S ) = = X . φ) . We like to argue that the matrix representing the rotational transformations of the function space C∞ (S ).5. φ ) (5. φ) m.195). }] . .195) i.. For this purpose we consider the subspaces of C∞ (S ) X = [ {Y . . ∞. Comparision with (5.5 Irreducible Representations We will consider now the eﬀect of the rotational transformations introduced in (5. (5.42). m = S2 i dΩ Y ∗m (θ. are characterized completely if one speciﬁes the transformation of any Y m (θ..e.e. . . φ) exp − ϑ · J . . }] (5. called the irreducible representation. φ) (5.191) i − ϑ·J f (θ . .. φ) ˜ Y m (θ. − + 1. therefore. exp − i ϑ · J (X ) = X . φ) D(ϑ) = .. φ). .5: Irreducible Representations 121 5.. . m . These coeﬃcients can be considered the elements of an inﬁnite-dimensional matrix which provides the representation of exp − i ϑ · J in the basis {Y m . 1. if one speciﬁes the functional form of the coeﬃcients [D(ϑ)] m .190) provide a complete.. φ) (5..194) Y m (θ. . φ) ∈ C∞ (S ). }. φ) ˜ under such rotational transformation by f (θ. (5. = .m m m (θ. on functions f (θ.− + . m Y m (θ. m = − . .. .. exp − i ϑ · J .− + .. . . ˜ f (θ.. .e.192) = S2 dΩ Y ∗ (θ .. ∞. i. i. orthonormal basis for the function c m Y m (θ. with elements given by (5. φ) c = .

the transformation exp − i ϑ · J leaves X invariant as well.201) From expression (5. . One = contribute can conclude then that in the expansion (5.56. one concludes for the matrix elements (5. 5. i. Similarly. n The action of J− on Y m produces Y m−1 or. φ) exp m i − ϑ·J Y m (θ.n3 n n n Accordingly. . one needs to consider the action of J+ 1 J− 2 J3 3 on the subspaces X . i. with blocks of dimension 1. .198) 2 2 where ϑ± = ϑ1 ( 1 ϑ− J+ 2 + 1 2 ϑ+ J− ϑ2 .200) leaves X invariant. (5..n3 J+ 1 J− 2 J3 3 . m = δ S2 dΩ Y ∗ (θ.. φ) . Since any additive term n n n cn1 .199) [J+ . . leaves X invariant. and J3 1 1 ϑ · J = ϑ− J+ + ϑ+ J− + ϑ3 J3 (5.e. J− ] = 2 J3 . If one orders the basis according to the partitoning (5.195) [D(ϑ)] m. one can conclude that J+ leaves X invariant... i. (5. J− .202) D(ϑ) = (2 + 1) × (2 + 1) . (5. 1×1 3×3 .e..e. 5.65). J− 2 also leaves n1 X invariant. The exponential operator exp − i ϑ · J + ϑ3 J3 )n .122 Theory of Angular Momentum and Spin The invariance of the subspaces X under rotations follows from a Taylor expansion of the exponential operator exp − i ϑ · J . can be expanded in powers One employs then the commutation properties [J3 .n2 . equivalently. J± ] = ± J± (5. .200) of the rotational transformations one can also conclude that the X are the smallest invariant subspaces of C∞ (S ). in case m = − .200) in (5.n2 .194) only spherical harmonics with or.199) allows one to express exp i − ϑ·J = n1 . One proceeds by expanding ﬁrst ϑ · J in terms of J+ . Since the Y m are n3 eigenfunctions of J3 the action of J3 reproduces the basis functions of X .197) of C∞ (S ) the matrix representation of D(ϑ) assumes a block-diagonal form. produces 0. 3.. . Application of (5. which follow readily from (5.

(d) Show that the similarity transformation T (ϕ) deﬁned in the space of non-singular. Determine the generator of ρ(ϕ) in analogy to (5. k = 1. though seemingly natural. A ∈ C. and periodic functions f (α). The representation is referred to as the irreducible representation.49). and for which detR = 1. i. does not provide the simplest mathematical description of rotations. 5. (a) Show that SO(2) with the group operation ◦ deﬁned as matrix multiplication. (b) Prove that the elements of SO(2) can be completely characterized through a single parameter.6: Wigner Rotation Matrices 123 This representation of D(ϑ) is the one wich has blocks of the lowest dimensions possible.e. f (α + 2π) = f (α). 2. (c) Show that the map R(ϕ) x y = x y = cosϕ −sinϕ sinϕ cosϕ x y = R(ϕ) x y deﬁnes a representation of SO(2).e. ∈ N} invariant. i. 3 certainly allow one to describe any rotation around the origin.1: Representations of the group SO(2) SO(2) is the set of all 2×2 matrices R which are orthogonal. other representations are termed reducible representations. this parametrization. for which holds RT R = R RT = 1 1. A more suitable parametrization had been suggested by Euler: every rotation exp − i ϑ · J can be represented also uniquely by three consecutive rotations: (i) a ﬁrst rotation around the original x3 –axis by an angle α. 5. However. real 2 × 2– matrices O through O = T (ϕ)O = R(ϕ)OR−1 (ϕ) with R(ϕ) as in (c) is also a representation of SO(2).5. .6 Wigner Rotation Matrices We will now take an important ﬁrst step to determine the functional form of the matrix elements of D(ϑ). The three components ϑk . is a group. (f) The transformation ρ(ϕ) as deﬁned in (e) leaves the following subspace of functions considered in (e) Xm = { f (α) = A e−imα . (iii) a third rotation around the new x3 –axis by an angle γ. (ii) a second rotation around the new x2 –axis by an angle β. the map ρ(ϕ) deﬁned through ρ f (α) → g(ϕ) = f (α − ϕ) = ρ(ϕ)f (α) also deﬁnes a representation of SO(2). Give the corresponding expression of ρ(ϕ). (e) In the space C∞ ( ) of inﬁnitely often diﬀerentiable.22. Exercise 5.5. This step reconsiders the parametrization of rotations by the vector ϑ assumed sofar.

β. however. and the r. is e− i γJ3 = e− i βJ2 e− i αJ3 e− i γJ3 e i αJ3 e i βJ2 (5.e.208) αJ3 βJ2 = e− αJ3 e− i βJ2 e i (5. φ) = S2 dΩ Y ∗m (θ. γ)] = . (i) and (ii). one can replace (5.. β. in terms of rotations deﬁned with respect to the original frame. γ ∈ R such that exp i − ϑ·J = e− i γJ3 − i βJ2 − i αJ3 e e (5.h. m [D(α. followed by J2 in the original frame.m m.203) For any ϑ ∈ R one can ﬁnd Euler angles α.194. the transformation from the rotated frame to the original frame.209).206) The expression (5.207) are equivalent.203).210) The third rotation in (5. Accordingly.124 Theory of Angular Momentum and Spin The angles α. i. (5. 5. i. (5.209) The ﬁrst two rotations in (5. m i Y m (θ.209) in this expression one obtains e− i γJ3 = e− i αJ3 e− i βJ2 γJ3 e i αJ3 e− i i αJ3 e− i i γJ3 e i i αJ3 γJ3 e− e i i αJ3 e i i βJ2 e i αJ3 = e− i = e− αJ3 e− βJ2 e− βJ2 e αJ3 (5. followed by transformation (i). For this purpose we notice that J2 can be expressed through the similarity transformation J2 = e− i αJ3 J2 e i αJ3 (5. φ) e− γJ3 − i βJ2 − i αJ3 e e . in analogy to (5. the l. β. For any similarity transformation involving operators A and S holds eS A S Accordingly. φ) [D(α.203) has the disadvantage that it employs rotations deﬁned with respect to three diﬀerent frames of reference. γ will be referred to as Euler angles.. We will demonstrate that (5. β.203) can be expressed.205) Y m (θ. of (5.e..211) Using (5. Obviously. The axis x2 is deﬁned in the coordinate frame which is related to the original frame by rotation (i).203). The Euler rotation replaces exp − i ϑ · J by e− i γJ3 − i βJ2 − i αJ3 e e . the axis x3 is deﬁned in the coordinate frame which is related to the original frame by the consecutive rotations (i) and (ii).204) is satisﬁed.s. can then be written e− i βJ2 − i αJ3 e = e− i αJ3 e− i βJ2 (5. we can write e− i −1 = S eA S −1 .195) by ˜ Y m (θ. the transformation from the original frame to the rotated frame.s.e.212) . i.207) which replaces J2 by the inverse transformation (i). γ)] m. φ) (5.h. i (5.

φ) e− αJ3 e− βJ2 Y m (θ.216) Using. φ) e− m i βJ2 Y m (θ.219). β. This allows one to express the rotational trasnformation (5. but many other elementary components of matter are endowed with spin-like properties.206) by ˜ Y m (θ. the muon µ and its neutrino νµ of the second .5. 5. γ)] m. (5. Examples are the other ﬁve members of the lepton family to which the electron belongs. 5. φ) e− m S2 i αJ3 f (θ. m ( ) S2 dΩ Y ∗ (θ. The eigenvalue property (5. (5. γ)] = .215) The evaluation of the matrix elements (5. γ)] m. 5. φ) . the electron e and the electron neutrino νe of the ﬁrst generation.205. (5.203) with respect to the original (unprimed) frame one simply needs to reverse the order of the rotations. β. m (5. φ) e−imγ . γ)] m.e.m m. m i Y m (θ. φ) = S2 dΩ f (θ. (5. φ) e− S2 i γJ3 Y m (θ. β. φ) e− m i βJ2 Y m (θ. φ) [D(α. φ) (5.64) yields dΩ f (θ. Best known is the spin of the electron.218) Deﬁning the so-called Wigner rotation matrix dmm (β) = one can express the rotation matrices [D(α. φ) = e−im α S2 dΩ Y ∗ (θ.214) e− γJ3 = S2 dΩ Y ∗m (θ. m = e−iαm S2 dΩ Y ∗ (θ.7 Spin 1 2 and the group SU(2) The spin describes a basic and fascinating property of matter. φ) Y m (θ. φ) e−iγm δ (5. (5.217) Accordingly. φ) . to redeﬁne the three rotations in (5. φ) f (θ.7: Spin 1 2 and the group SU(2) 125 Multiplication from the left with (5. the self-adjointness of the operator J3 one can state dΩ Y ∗ (θ.310)] an explicit expression for the Wigner rotation matrix (5. m [D(α.309. β.219) = e−iαm dm m (β) e−iγm δ ( ) . in addition.220) We will derive below [see Eqs.215) beneﬁts from the choice of Euler angles for the parametrization of rotational transformations.213) i. φ) i i (5.210) yields the simple result e− i γJ3 − i βJ2 − i αJ3 e e = e− i αJ3 e− i βJ2 e− i γJ3 . one can write [D(α..

each component Sk . So do the three generations of six quarks.126 Theory of Angular Momentum and Spin generation. the spin of the electron is likely the most important property in Chemistry.221) One can show readily that this set forms a group with the groups binary operation being matrix multiplication. have properties beyond those of single spins. There are also the mediators. there are together ﬁve conditions in terms of real quantities to be met by the matrix elements and.g. and because of detU = 1. c± ∈ C. Such systems can also assume any linear combination c+ χ+ + c− χ− . j. the gluon. then allowed + − symmetry transformations of spin states are described by 2 × 2-matrices U with complex-valued matrix elements Ujk . which might be the reason why consideration of rotational symmetry is so often fruitful in the study of matter. in principle. This speciﬁcation implies that we consider transformations save for overall factors eiφ since such factors are known not to aﬀect any observable properties. like superconductivity and magnetism.e. k = 1. We will consider ﬁrst only the so-called spin– 1 . In any case.3 . eight real numbers to specify the matrix elements. We may ﬁnally mention that the spin is at the heart of many properties of condensed matter systems. as long as |c+ |2 + |c− |2 = 1. generalizing then further below. carry other spin-like properties which together. 2. a relationship. k = 1. The presence of this property permeates matter also at larger scales than those of the elementary particles mentioned. the photon. 3 representing a 2 × 2 matrix. the degrees of freedom of the matrices U are three real quantities. two of which (in certain linear combinations) make up the vector mesons which carry spin 1. leaving its imprint on the properties of nuclei. Conservation of the scalar product under symmetry transformations requires ∗ the property U U † = U † U = 1 where U † denotes the adjoint matrix with elements U † jk = Ukj . How can the elements of SU(2) be parametrized. Because of the unitarity condition U † U = 1 which are really four equations in terms of real 1 quantities.2 . The transformation matrices are then elements of the set SU (2) = { complex 2 × 2 matrices U .1 systems 2 2 can be related to two states which we denote by χ+ and χ− . however. and three of which make up the baryons which carry spin. four real and four imaginary parts of Ujk . atoms and molecules.1 and spin. There . in which we deﬁne the scalar product between any state |1 = a+ χ+ + a− χ− and |2 = b+ χ+ + b− χ− as 1|2 = a∗ b+ + a∗ b− . the quarks. As complex 2 × 2 matrices one needs. the tau τ and its neutrino ντ of the third generation and their antiparticles carry a 1 spin. i. particles which mediate the strong and the electro-weak interactions. in fact. Spin. The particles mentioned. the two W± and the 2 2 Zo .222) where the vector S has three components. It appears to be rather impossible for a Physicist not to be enamored with the spin property. One can show that the unitarity condition requires these matrices to be hermitian. hence. [Sk ]mn = ([Sk ]nm )∗ . there is nothing more elementary to matter than the spin property. (5. and the condition detU = 1 requires the Sk to have vanishing trace. and which carry spin 1. We will ﬁnd that the spin in its transformation behaviour is closely related to angular momentum states. If we identify the states χ+ and χ− with the basis of a Hilbert space. one for each matrix element of U † U . The important feature is that all U ∈ SU(2) can be parametrized by an exponential operator U = exp − i ϑ · S (5. U U † = U † U = 1 detU = 1 } 1. 1 We will specify for the transformations considered detU = 1. e. 2.

5. (5.228) (5.g. z A = In fact. (5.230) where a. they are associated with the important fermion property of matter. hermitian 2 × 2–matrices A can be expandend. (5.k=1 j<k + σ j σ k aj bk . Cliﬀord algebras play an important role in the mathematical structure of physics.229) and avoids commuting the components aj and bk .224) provide a basis in terms of which all traceless. it holds. σ2 = 0 −i i 0 . the simplest choice being S1 = 1 σ1 . 2.227) which is essentially identical to the Lie algebra (5.227.224) Algebraic Properties of the Pauli Matrices The Pauli matrices (5. σk ]+ = 2δjk 1 . k = 1. (5.k=1 σ σ aj bk + j>k aj bk = 3 j. One can readily verify the commutation property σj σk − σk σj = [σj . σj 2 (5.. i. Any such A can be expressed in terms of three real parameters x.231) . In fact. 1 i. The proof of this relation rests on the commutation relationship and the anti-commutation relationship (5. At this point we will state a useful property. 2 S2 = 1 σ2 .229) = 1 . A = x σ 1 + y σ 2 + z σ3 . σ3 = 1 0 0 −1 . ∈ R . According to this property the Pauli matrices generate a 3dimensional Cliﬀord algebra C3 . (σ · a) (σ · b) = a · b 1 + iσ · (a × b) 1 (5.226) The Pauli matrices satisfy very special commutation and anti-commutation relationships. it might hold aj bk − bk aj = 0. 2 S3 = 1 σ3 2 (5. 3 are the Pauli matrices σ1 = 0 1 1 0 . 5. y. namely.k=1 σ σ 3 j k j.31) of the group SO(3). b are vectors commuting with σ.223) where σk . σk ]− = 2i jk z x − iy x + iy −z . We will show below that a 2-1–homomorphic mapping exists between SU(2) and SO(3) which establishes the close relationship between the two groups.225) σ (5.e.e. but their comonents must not necessarily commute with each other. y. z. e.7: Spin 1 2 and the group SU(2) 127 exist three such linear independent matrices. x. 1 σj σk = − σk σj for j = k which can also be readily veriﬁed. The Pauli matrices obey the following anti-commutation properties σj σk + σk σj = [σj . one obtains (σ · a) (σ · b) = 3 j 2 j=1 (σ ) aj bj 3 j k j.

128 Theory of Angular Momentum and Spin Using (σ j )2 = 1 the ﬁrst term on the r. i.k=1 j>k σ j σ k (aj bk − ak bj ) − 1 2 3 j k j.k=1 j>k σ j σ k (aj bk − ak bj ) = 1 2 3 j. 1 1 1 ( + 1) χ± 2 2 . (5.238) S 2 χ± = . since there is a slight diﬀerence: when you interprete the 3-component ϑ in (5. however. we like to draw in this respect also on a close analogy to angular momentum states which developes 2 2 2 if one considers the operators S 2 = S1 + S2 + S3 and S3 . The commutation property (5.230). Well almost like it.233) This result together with (5.k=1 j>k i σl (aj bk − ak bj ) = i =1 σ a×b . We like to derive this result now by evaluating the transformations of SU(2) explicitly. 1 (5. (5. hence.e.237) + 1)1 and.231) proves (5.222) as a rotation vector and you rotate once.h.s. S ] = i k m Sm (5. yields a · b.8 Generators and Rotation Matrices of SU(2) Sofar there is nothing which relates the transformations U . The algebra of the generators of a transformation is such a basic property that the ‘accident’ that the generators of spin transformations behave in this respect like the generators of 3-dimensional rotations makes spins appear in their physical behaviour like rotations. however.k=1 (σ σ j>k σ k σ j (aj bk − ak bj ) = (5.234) 5.222) by the Euler form exp(−iγS3 ) exp(−iβS2 ) exp(−iαS3 ) . The two remaining terms yield using σ j σ k = −σ k σ j for j = k and altering ‘dummy’ summation indices 3 j. only a 720o rotation leaves the spin invariant. In the special case a = b ∈ R holds (σ · a)2 = a 2 1 .235) which is identical to that of the generators of SO(3). (5. to rotations in space.236) The matrix elements of this 2 × 2 operator can be labelled by the basis states χ+ and χ− . through the algebra obeyed by the operators Sk [Sk .k=1 j>k 1 2 3 j.222).227) leads to 3 3 jk j. For this purpose we choose to replace (5. let say around the x2 –axis by 360o . (5. S3 χ± = ± 1 χ± 2 .232) − σ k σ j ) (aj bk − ak bj ) . This relationship emerges. Noticing the idempotence of the Sk ’s 2 Sk = 3 one obtains S 2 = 4 1 = 1 1 1 2(2 1 4 1 0 0 1 (5. spin changes sign.

239) in the notation (5. β. can be constructed formally from spin 2 2 states χ± if one considers the two properties χ± to be carried by two kinds of bosons. (5. (5. use the label 2 2 | 1 ± 1 for the states χ± . following Jourdan and Schwinger. (5.240) . i. . .5.236) 1 2m | exp(−iγS3 ) exp(−iβS2 ) exp(−iαS3 ) | 1 m 2 1 2m = e−iγm | exp(−iβS2 ) | 1 m e−iαm 2 (1) 2 = e−iγm dmm (β)e−iαm (5.e. One cannot consider the 1 entities carrying the spin– 2 to be particles in the ordinary sense for it can be shown that spin– 1 2 particles have fermion character.and sinfunctions and one obtains for the rotation matrices 2 dmm (β) (1) = cos β 2 sin β 2 − sin β 2 cos β 2 .219.9: Constructing Spin States with S > 1 2 129 This result implies that the states χ± behave in relation to the generators of SU(2) like an angular momentum state | 1 ± 1 in relation to the generators of SO(3).242) The expressions in brackets [.] can be identiﬁed with the Taylor expansion of the cos.220) are [D(α. therefore. 2 2 We obtain with this notation for the transformations (5. 3 . (5. .e. The idempotence property of Sk yields particularly simple expressions for these powers. .5.245) 5. 1.e. namely.241) 2n+1 Taylor expansion of the exponential operator yields then ∞ exp(−iβS2 ) = n=0 (−1)n β 2 2n ∞ 1 + 1 n=0 (−1)n β 2 0 1 −1 0 (5. − + 1.244) i. no two such particles can exist in the same state. . 1 . ( − β S2 )2n = (−1)n ( − β S2 )2n+1 = (−1)n β 2 β 2 2n 1 1 0 1 −1 0 2n+1 (5. . m = cos β e−i( 2 + 2 2 γ α sin β ei(− 2 + 2 2 α γ − sin β ei( 2 − 2 2 α γ cos β ei( 2 + 2 2 α γ . m = − .9 Constructing Spin States with Larger Quantum Numbers Through Spinor Operators In this section we like to demonstrate. In order to determine the Wigner matrix in (5. The complete matrix elements (5. identical particles any number of which can exist in the same state χ+ and χ− . that states | m for higher quantum numbers = 0. For this purpose one needs to determine the powers of −iβS2 . . .. .219). 2. We may. i.243) We note the property of this rotation matrix 2 dmm (2π) (1) = −1 1 .239 ) one expands the exponential operator exp(−iβS2 ). γ)] m.239) where we made use of the notation (5. rotation by 360o changes the sign of the spin state.

b † ζ ζ =0 (5.e. 2 + 2 − † (5. j. b− )T is expressed through the commutation relationships + − (ζ. . b† )T and b = (b+ . j = 0. 3 .251) One can show using this property and (5.249) In the following will refer to b† = (b† . −) bζ . b ζ = b ζ. . m) = b† − j−m (j + m)! (j − m)! |Ψ0 . therefore. b† )T and b = (b+ . .5.252) (5. .253) = δζζ . i. also operators of the type x b† = x+ b† + x− b† + − . b ζ = 0 b ζ. We consider. + − For example. b ζ = b† . .e. b± |Ψ0 = 0 (5. . These operators are associated with a given spatial reference system. (5. b ζ The States |Ψ(j.130 Theory of Angular Momentum and Spin Deﬁnition of Spinor Creation and Annihilation Operators We present the states χ± through creation operators b† which when applied to a formal vacuum ± state |Ψ0 generate χ± . ζ = +. 1 . b− † = − sin β † β b + cos b† .243) the creation operators in a coordinate system rotated by an angle β around the y–axis are b+ † = cos β β † b + sin b† 2 + 2 − . using (5. x∗ b = x∗ b+ + x∗ b− + − (5.250) which for x∗ x = x∗ x+ + x∗ x− = 1 represent spinor operators in an arbitrary reference system.254) We will show below that these states are orthonormal and form spin states with higher quantum numbers.247) (5. −j + 1. b† |Ψ0 = χ− − .246) The corresponding adjoint operators are denoted by b+ and b− . 2 2 . m) † and bζ obey the commutation (5. The operators b† allow one to construct a set of states which represent j + m–fold and j − m–fold ± χ+ and χ− states as follows b† + j+m |Ψ (j.248) bζ . m = −j. (5. 1. i. . The boson character of the operators b† = (b† . b† |Ψ0 = χ+ + .247. b− )T as spinor creation and annihilation + − operators.248) that bζ relationships † † b ζ. b † ζ For the vacuum state |Ψ0 holds = δζζ .

(5.255) In order to determine how this operator is modiﬁed by multiplication from the left by b+ we note b+ f (b† ) |state = + b+ .256) and (5. 5.10 Algebraic Properties of Spinor Operators We want to establish ﬁrst a few important and useful algebraic properties which result from the commutation relationships (5. in b† and b† . b† + n . (5.258) Equations (5. b† + n = n cn b+ .257) motivate us to determine the commutator (5. b† + n+1 = = b+ .249) b+ f (b† ) |Ψ0 = + b+ .248) and from (5. f b† Obviously.248) we obtain b+ .5. For this purpose we consider the following polynomial of creation operators N j+m j−m f (b† ) = + n=0 cn b† + n .254). b† + = n b† + n−1 . f (b† ) + (5. + b+ . f (b† ) |Ψ0 . An example + − is the monomial b† which generates the states (5. The Spinor Derivative Operator A most important operation is the action of the annihilation operators b+ and b− on operators which can be expressed as polynomials. b† + b+ . .247. b † + = n b† + n−1 b† + b † + + n = (n + 1) b† + .256) In the special case |state = |Ψ0 this reads using (5.259) The property is obviously true for n = 0. possibly inﬁnite power series. (5.260) . If it is true for n then using (5.249). b† + n b† + b† + b† + + n n n b+ . + + (5.257) which for (5. We will derive the result.10: Algebraic Properties of Spinor Operators 131 5. b† + − well-known for the quantum mechanical harmonic oscillator. we need to evaluate b+ .255) is b+ . f (b† ) |state + f (b† ) b+ |state . We will show that for this commutator holds n b+ . that the annihilation operators play a role similar to the diﬀerential operator in calculus.

One can ﬁnally conclude for polynomials (5. b† ) in b† and b† holds (ζ = +. f (b† ) − = f (b† ) − (5.268) .247) the more general property for polynomials f (b† ... f (x)] g(x) = (∂k f (x))g(x) or [∂k . m) . (5.262) Because of the commutation relationships (5. m) m=−j (5. b† ) |Ψ0 . (5.259) holds for all n ∈ N. m) = x− b† − j−m (j + m)! (j − m)! |Ψ0 . m) We want to prove now the property ∞ j exp xb † |Ψ0 = 1 j=0. 2 . f (x)] = ∂k f (x) .267) exp xb† |Ψ0 is called a generating function of |Ψ(j.132 Theory of Angular Momentum and Spin i. + − (5.264) This demonstrates the equivalence of the spinor operators bζ and the derivative operation..263) corresponds to the product rule of calculus ∂k f (x)g(x) = (∂k f (x))g(x) + f (x)(∂k g(x)) which can be written ( ∂k f (x) − f (x)∂k ) g(x) = [∂k .257) we can conclude in particular bζ f (b† . b† ) |Ψ0 = + − ∂ ∂b† ζ f (b† . Equation (5.266) where xb† has been deﬁned in (5. By induction we can conclude that (5. + − (5.263) According to (5. b† ) . f (b† . φjm (x) Ψ (j. one can prove b− . f (b† ) + where f (x) = df dx . the property holds also for n + 1.266) we compare the terms x+ b† + j+m φjm (x) Ψ (j.1. (5.250) and where φjm (x) represents the function of the two variables x+ and x− closely related to |Ψ(j.265) Generating Function of the States |Ψ(j.261) Similarly. b† ) + − = ∂ ∂b† ζ f (b† .255) b+ .e. = f (b† ) + (5. In order to derive (5. m) φjm (x) = j−m xj+m x− + (j + m)! (j − m)! . −) + − + − bζ .

270) s x+ b† + N −s x− b† − s |Ψ0 . m) 1 j=0. 3 .. one obtains φjm (x) Ψ (j. chosing the summation index j = 0. 1.. m=−j = j=0...e. .273) Change of the summation indices allows one to conclude j φjm (x) Ψ (j. m) = = |Ψ0 (N − s)! s! N −s 1 N! x+ b† x− b† + − N ! (N − s)!s! N −s 133 (5. (5.10: Algebraic Properties of Spinor Operators with the s–th term in the binomial expansion of (a + b)n n! an−s bs .274) Summing this expression over 2j = 0.1. (5. 1. . 2..1... 2 2 leads to ∞ j ∞ φjm (x) |Ψ (j.275) = exp xb† |Ψ0 . . s! (n − s)! Deﬁning j+m = N −s j−m = s. m) m=−j = = 1 x+ b† + x− b† + − (2j)! 2j 1 xb† |Ψ0 . 2 ∞ 1 xb† 2j! u 2j |Ψ0 = u=0. 2.5. . . 1 .1.. 1 .272) The summation over s can be written in terms of j and m using (5. (2j)! 2j |Ψ0 (5....269) (5. 1 xb† u! |Ψ0 (5.271) Summation over s from s = 0 to s = N yields 1 N! = = N s=0 N! (N − s)!s! x+ b† + x− b† − s |Ψ0 N 1 x+ b† + x− b† |Ψ0 + − N! 2j 1 x+ b† + x− b† |Ψ0 + − (2j)! (5. i. 2 .270) N 2j j → s=0 j−m=0 → m=−j . .

279) One can generalize this to † † s (bζ )s e(yb ) |Ψ0 = yζ e(yb ) |Ψ0 .276) = j.254) are orthonormal.281) and. m) |Ψ j . (5. b† ] = 0 allow one to state that for + − + any polynomial f (b+ .m.284) . i. (5. According to (5. For this purpose we consider the inner product † † e(xb ) Ψ0 |e(yb ) Ψ0 = δjj δmm (5. m) We want to show now that the states (5. we can conclude † † ∗ e(xb ) Ψ0 |e(yb ) Ψ0 = e(x y) (5. of (5. Orthonormality of the States |Ψ (j. [b† .282) where we deﬁned x∗ y = x∗ y+ + x∗ y− .134 Theory of Angular Momentum and Spin which concludes our derivation. b− ) holds † † f (b+ .249) for any non-vanishing integer s holds + − Ψ0 |bs Ψ0 = 0 and. one can write † † e(xb ) Ψ0 |e(yb ) Ψ0 † ∗ Ψ |e(x y) e(yb ) Ψ = 0 0 = e (x∗ y) † Ψ0 |e(yb ) Ψ0 (5. m holds.m φjm (x∗ ) φj m (y) Ψ (j. hence. (5.e. In order to evaluate the ∗ b) (yb† ) operator e(x e we notice that the derivative property (5.s. y− )e(yb ) |Ψ0 (5. b− )e(yb ) |Ψ0 = f (y+ .277) To evaluate this expression we ﬁrst notice † e(xb ) † = e(x ∗ b) . b† ] = 0 and [b− .283) (5.277) by Ψ0 |e(x b) e(yb ) Ψ0 .h. (5. that Ψ (j. m .j .264) implies † † bζ e(yb ) |Ψ0 = yζ e(yb ) |Ψ0 . b− ] = 0. m) |Ψ j . therefore.278) † ∗ which allows us to replace the l. The generating function allows one to derive various properties of the states |Ψ (j.280) The commutation properties [b+ . m) and will be used for this purpose below.

. etc.ζ where σk .287) ζ 2 ζ. .276). 3 . 1.m . One deﬁnes corresponding operators Jk through 1 Jk = b† < ζ |σk | ζ > bζ . Jj ] = = 1 4 1 4 < n |σi | m >< n |σj | m > b† bm .m.n . (5. 3 denote the Pauli spin matrices (5.m +b† b† .223.m. 5.m φjm (x∗ ) φjm (y) . The 2 operators are explicitly.j .282) and (5. In our present notation the matrix elements < ζ |σk | ζ > for ζ = ± correspond to the matrix elements < ζ |σk | ζ > for σ = ± 1 . J1 = J2 = J3 = 1 † b b− + b† b+ − 2 + 1 † b b− − b† b+ − 2i + 1 † b b+ − b† b− . < + |σ1 | − > = < − |σ1 | + > = 1.5. (5.275) one can show e(x ∗ y) = j. 1 New Representation of Spin 0. (5. using < + |σ1 | + > = < − |σ1 | − > = 0.289) < n |σi | m >< n |σj | m > {b† bm . . (5.n . b† bm n n n.285) Following steps similar to those in Eqs. States 2 We want to demonstrate now that the states |Ψ (j.268–5.277) φjm (x∗ ) φj m (y) Ψ (j.m. − 2 + (5. bm bm } n n = 1 4 < n |σi | m >< n |σj | m > {b† bm δmn − b† bm δm n } n n n.m 135 = e(x ∗ y) .n .286) from which follows immediately the orthonormality property (5.224).m ijk Jk . m) as given in (5.10: Algebraic Properties of Spinor Operators and comparing (5.m. . b† bm n n n. Jj ] = i This property is derived as follows [Ji . (5. k = 1.288) The three operators obey the Lie algebra of SU(2) [Ji .254) are eigenstates of operators J 2 and J3 with eigenvalues j(j + 1) and m where J 2 and J3 are representations of the spin 2 2 2 operators S 2 = S1 + S2 + S3 and S3 deﬁned in 5.224). m) |Ψ j . m j. 2 . 2.

292). m) as Eigenstates of J 2 and J3 We wish to show now that the states |Ψ (j.295) The last result together with (5.294) .m.m. J+ ] = [J1 − iJ2 .291) Here we have deﬁned J+ J− = J1 + iJ2 = J1 − iJ2 = b† b− + = b† b+ .m. (5. m) are eigenstates of J 2 and J3 . To this end we note 2 2 2 J 2 = J1 + J2 + J3 = 1 1 2 J+ J− + J− J+ + J3 2 2 .297) . σj ]| m > b† bm n n.m 1 4 < n |σj | m >< m |σi | m > b† bm n n . J1 ] + i [J1 .289) yield [J− .293) . 2 J 2 = J+ J− + J3 − J3 = J3 (J3 − 1) + J+ J− (5.k ijk ijk σk | m > b † bm n = i = i 1 2 < n |σk | m > b† bm n n. (5.m < n |[σi . (5.290) |Ψ (j. J2 ] = −2 J3 from which we can conclude the property 1 1 J− J+ = J+ J− − J3 2 2 Hence.136 = 1 4 − = = 1 4 1 4 Theory of Angular Momentum and Spin < n |σi | m >< m |σj | m > b† bm n n.288) yields ﬁrst the expression J2 = which results in J2 = 1 4 1 † ( b + b + − b † − b− ) ( b † + b+ − b† − b − − 2 ) + b † + b − b † − b + 4 .296) ( b † + b + b † + b+ + b † + b + b † − b − − 2 b † + b + − b † − b − b † + b + + b † − b − b † − b − + 2 b † − b − + 4 b † + b − b† − b + ) (5.292) The commutation relationships (5.m ijk Jk . (5.m < n |2i n. − (5. (5. J1 + iJ2 ] = −i [J2 . (5.

b † + b + + b † + b+ b † − b − 1 1 b† + b + + b † + b + b † − b− + b † − b − b † + b + . . m) .10. (5. (5. Ik ] = i jk I . . n2 .298) (5.. Show λ = n1 +n2 n1 +n2 2 ( 2 + 1) and m = 1 2 (n1 − n2 ). 1 2 k (a) Show that the eigenstates are given by 1 |n1 . [bj .e. One can then conclude J 2 |Ψ (j.301) Obviously. n2 = √ b+ n1 ! 1 n1 |0 1 1 √ b+ n2 ! 2 n2 |0 2 where the vacuum states are deﬁned through bj |0 j = 0. the number of states to the possible energy eigenvalues. . m = −λ.s. 1 2. . . . n} eigenstates of I 2 = I1 + I2 + I3 and of I3 I 2 ||λ. Construct. n1 = 0. m) One can furthermore derive readily J+ |Ψ (j. n = n1 + n2 ﬁxed. . b† .299) (5. and eigenstates of k with eigenvalues j.300) (5. m = λ||λ.e.303) Exercise 5. using operators I± = I1 ± iI2 2 2 2 and the subspace {|n1 . . m + 1) (j + m) (j − m + 1)|Ψ (j.10: Algebraic Properties of Spinor Operators The last term on the r. I3 ||λ.254) are eigenstates of + b+ and − b− with eigenvalues j + m and j − m. (b) Show that the three operators 1 + 1 1 + (b1 b2 + b+ b1 ) . can be written b† + b − b † − b + = = One can then state J2 = 1 (b† + b+ )2 + b† + b+ b† − b− + b† − b− b† + b+ 4 + (b† − b− )2 + 2 b† + b+ + 2 b† − b− Deﬁning the operator 1 ˆ k = ( b† + b+ + b† − b− ) 2 one can write ﬁnally ˆ ˆ J2 = k ( k + 1 ) b† . m . m) J− |Ψ (j. . 2 2 137 (5. m = m||λ. [Ij . . m) J3 |Ψ (j.305) = = j ( j + 1 ) |Ψ (j. λ. the states (5.h.304) (5. −λ + 1. i. I2 = (b1 b2 − b+ b1 ) . . I3 = (b+ b1 + b+ b2 ) 2 2 2 2 2i 2 1 satisfy the Lie algebra of SU(2). b+ ] = δjk .302) (5. . m − 1) . . m) m |Ψ (j. i. 2. State the corresponding eigenvalues and the degree of degeneracy. j = 1. 1.1: The system investigated in this section is formally identical to a 2-dimensional isotropic harmonic oscillator governed by the Hamiltonian H = ω(b+ b1 + b+ b2 + 1) . m) = = (j + m + 1) (j − m) |Ψ (j. 2 .5. ˆ respectively. m I1 = where λ = 0.

309) . m) .251).306) and (5. (j) (5. m ) are related to the + − states |Ψ (j. + − One may expect that these two conditions restrict both σ and σ.306. The prefactor of (b† )j+m (b† )j−m which according to (5. 1 1 √ sinβ 2 (1 + cosβ) 2 (1) 1 cosβ (dm m ) = − √2 sinβ 1 1 √ 2 (1 − cosβ) − 2 sinβ and in case j = 1 2 1 2 (1 − cosβ) 1 √ sinβ 2 1 (1 + cosβ) 2 (5. For this purpose we expand + − + − b† + cos β b† + sin β b† 2 + 2 − j+m σ =0 j−m σ=0 j+m j+m b† − j−m = j−m −sin β b† + cos β b† 2 + 2 − cos β 2 j−m −σ σ = j+m σ σ sin β 2 σ +σ j+m −σ j−m σ −sin β 2 cos β 2 b† + b† − 2j−σ −σ (5.11 Evaluation of the Elements dj m (β) of the Wigner Rotation m Matrix The spinor algorithm allows one to derive expressions for the Wigner rotation matrix elements dj m (β). m) in the original coordinate system by j |Ψ j. both conditions are satisﬁed for σ = j + m − σ. In case j = 1 this expression yields. m = m=−j dm m (β) |Ψ (j.5.254) are |Ψ j.307) shows that the elements of the rotation matrix can be obtained by binomial expansion of b † and b † in terms of b† and b† .307) can be + − (j) identiﬁed with the elements dm m (β) of the rotation matrix can then be written (j) dm m (β) = (j + m)!(j − m)! (j + m )!(j − m )! β 2 j−m × (−1)j−m −σ sin σ=0 2j−m−m −2σ j+m j+m−σ cos β 2 j−m σ m+m +2σ × (5. . The combination of σ.138 Theory of Angular Momentum and Spin 5.308) The latter sum involves terms (b† )j+m (b† )j−m for σ + σ = j + m and 2j − σ − σ = j − m. On the other side the states |Ψ (j. σ values which yields (b† )j+m (b† )j−m is + − then σ = j + m − σ. For this purpose we note that the states |Ψ(jm) in a rotated coordinate system according m to (5.306) where b † and b † are given by (5.307) Comparision of (5.243).310) it reduces to (5. for example. However. m = b† + j+m b† − j−m (j + m )! (j − m )! |Ψ0 (5.

Hence.311) (5. It is of interest to trace the mapping from SU(2) onto SO(3) as represented by (5..310) the similarity transformation ˜ A = U† A U where U is the 3 × 3 unitary matrix which establishes the transformation 1 1 i √ ( −x1 − ix2 ) − √2 − √2 0 x1 2 x3 0 1 . Evaluation of A yields √ −1 0 1 1 √ cosβ + 2 sinβ 1√ cosβ − 1 ˜ 0 A = 4 i √ i − 2 sinβ 2 cosβ 2 sinβ × √ 0 2 0 1 − cosβ − 2 sinβ 1 + cosβ √ −1 −i √ 0 −2 cosβ −2 2 sinβ 2 cosβ 2i 2i × × 0 0 2 = 1 4 √0 1 −i 0 −2 sinβ 2 2 cosβ 2 sinβ −1 −i √ 0 cosβ 0 −sinβ 1 0 (5. the orthogonal 3 × 3 matrix which describes a rotation around the x2 –axis.313) × 0 0 2 = 0 sinβ 0 cosβ 1 −i 0 which.310) establishes a mapping of SU(2) onto SO(3). e.12: Mapping of SU(2) onto SO(3) 139 5.309) does not distinguish between SU(2) transformations (5. U = 0 1 1 i √ ( x1 − ix2 ) √ x3 − √2 0 2 2 (5. = U x2 .e. Replacing the latter transformation by its negative form.312) The choice of this transformation derives from a property shown further below. This can be shown by (1) applying to the matrix A = (dm m ) in (5.g. conclude that the mapping from SU(2) onto SO(3) is a 2–1 mapping.5.. i.309) and the particular matrix (5. i.12 Mapping of SU(2) onto SO(3) The representation (5. of (5.e.314) in case of integer j–values. x2 . . namely that the component of the vector on the l. The SU(2) transformation entered in (5. One can.314) leaves (5. This factor implies. the matrix A should represent a rotation around the x2 –axis in the space of vectors ˜ (x1 .251).310). in case of j = 1. 2 2 − (5.h.309).310) to the SU(2) transformation assumed in deriving the general result (5. b+ † = −cos β † β b+ − sin b† 2 2 − . b− † = sin β † β b+ − cos b† .s.308) unaltered except for a factor (−1)2j which multiplies then also the ﬁnal result (5.308) and had the form (5. x3 ) ∈ R . therefore.251) and (5. that the the representation (5. however. in fact. is the expected element of SO(3).312) transforms like an angular momentum state ˜ |1 m .

140 Theory of Angular Momentum and Spin .

In this case a proper choice for a description of interactions would be to consider the vectors rAB = rA − rB and ρC = (mA rA + mB rB )/(mA + mB ) − rC . Further below we will consider composite systems involving spin states.. on the distances between the three particles.g.. e. give rise to good quantum numbers. but neither on the position of the center of mass of the particles nor on the overall orientation of the three particle system with respect to a laboratory–ﬁxed coordinate frame.g. φ) as eigenfunctions and/or spin. and (rAB . ρC ). commute with the Hamiltonian of the composite system and. Example: Three Particle Scattering Consider the scattering of three particles A. How should they be chosen? Actually there is no unique choice. |ρC |. three Eulerian angles α. for example. Since the Hamiltonian does not depend on the position of the center of mass R = (mA rA + mB rB + mC rC )/(mA +mB +mC ). To specify the dependency of the Hamiltonian on the particle coordinates we start from the nine numbers which specify the Cartesian components of the three position vectors rA .Chapter 6 Quantum Mechanical Addition of Angular Momenta and Spin In this section we consider composite systems made up of several particles. The overall orientation of any three particle conﬁguration can be speciﬁed by three parameters1 . rC of the particles. since related operators. and to express the Hamiltonian in terms of |rAB |. by a rotational vector ϑ. e. We like to illustrate this for an example involving particle motion. six parameters must suﬃce to describe the interaction of the system. The rotational part of the scattering motion is described then in terms of the unit vectors rAB and ˆ We remind the reader that. B. is the quantity of interest. hence. β. and particle C impinges on the compound AB coming from a large distance. sums of orbital angular momentum and of spin operators of the particles. C governed by a Hamiltonian H which depends only on the internal coordinates of the system. classically speaking the sum of all angular momenta and spins of the composite system. Often the socalled total angular momentum. γ are needed to specify a general rotational transformation 1 141 . This eleminates three further parameters from the dependency of the Hamiltonian on the three particle conﬁguration and one is left with three parameters. each carrying orbital angular momentum decribed by spherical harmonics Y m (θ. We like to consider a choice which is physically most reasonable in a situation that the scattering proceeds such that particles A and B are bound. rB .

= 0. 2} 1.m2 Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ). One can describe the rotational degrees of freedom of the three-particle scattering process through the basis B = {Y 2 1 m1 (ˆAB ) Y r 2 m2 (ˆC ). will make their appearence a natural consequence of the symmetry of the building blocks of matter.max 2. The latter often arrives at the physical properties of composite systems by adding the .max = 1 2 IA−B E .. The rotational symmetry of the interaction between the particles allows one.max d(B) = 1 =0 (2 1 + 1) 2 =0 (2 2 + 1) = ( 1.1) where 1.max = 10 one obtains d(B) = 14 641. each of which stands for two angles. rather than by Classical Mechanics. however. − 1 ≤ m1 ≤ 1. . .4) The basis set which provides a maximum degree of decoupling between rotational states is of great principle interest since the new states behave in many respects like states with the attributes of a single angular momentum state: to an observer the three particle system prepared in such states my look like a two particle system governed by a single angular momentum state.3) For rather moderate values 1. n = 1.142 Addition of Angular Momenta and Spin ρC .max = ∆ 2 mA mB mC E . by the total energy E. There is yet another important reason why the following section is of fundamental importance for the theory of the microscopic world governed by Quantum Mechanics.e.max = 2. Such large number of dynamically coupled states would constitute a serious problem in any detailed description of the scattering process. . . as we will demonstrate below. mA mB + mB mC + mA mC 2. to separate the 14 641 dimensional space of rotational states Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ) into r ρ subspaces Bk . 2.max . − 2 ≤ m1 ≤ (6. ρ 1 = 0.max and 2. 2 . mB . (6. One may consider then to describe the motion in terms ˆ of products of spherical harmonics Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ) describing rotation of the compound AB r ρ and the orbital angular momentum of C around AB. . composite systems behaving like elementary objects are common. have not even be considered. mC . In fact. . . i. 2. in fact.m1 . . and by the moment of inertia IA−B of the diatomic molecule A–B approximately as follows 1. Obviously. the values of which are determined by the size of the interaction domain ∆V . B1 ⊕ B2 ⊕ . = B such.max + 1)2 (6. 1.m1 2 . (6.m2 c (n) r ρ 1.max denote the largest orbital and rotational angular momentum values. albeit puzzling. 14 641} . . and the following mathematical description will shed light on their ubiquitous appearence in physics. in particular. by the masses mA . . . a very large number. the dimensions d(Bk ) of these subspaces does not exceed 100. vibrations and rearrangement of the particles in reactions like AB + C → A + BC.2) The dimension d(B) of B is 1. Such extremely useful transformation of the problem can be achieved through the choice of a new basis set B = { 1 .max . . 1. that only states within the subspaces Bk are coupled in the scattering process. since further important degrees of freedom. .max + 1)2 ( 2.

One can equivalently express therefore (6. In this sense. in fact. It holds then H R(ϑ) ψ = R(ϑ) H ψ. ρC ) deﬁned through R(ϑ) ψ(rAB . ∂yAB ∂zAB i (2) ∂ ∂ − J3 = ρy − ρx . i. be diﬀerent from H. presumably a facility the reader would like to acquire with great eagerness. Following our description of rotations of single particle wave functions we express (6. from which follows in turn the well-known result that H is related to H through the similarity transformation H = R(ϑ) H R−1 (ϑ). Since this is true for any ψ(rAB . 3 (6. connected with the fact that the Hamiltonian H does not depend on the overall orientation of the three interacting particles.8) Obviously.5) do not aﬀect the Hamiltonian. rotations R(ϑ) of the wave functions ψ(rAB .10) .55) holds i (1) − J1 = zAB ∂ ∂ − yAB .6) For the following it is essential to note that H is not invariant under rotations of only rAB or ρC .7) where the generators J (k) are diﬀerential operators acting on rAB (k = 1) and on ρC (k = 2). the reader will learn in the following section how to add and subtract in the microscopic world of Quantum Physics.6. Jq(2) = 0 for p. ∂ρx ∂ρy (6.0: Addition of Angular Momenta and Spin 143 corresponding physical properties of the elementary components. the existence of a basis (6. Rotational Symmetry of the Hamiltonian As pointed out already.. For ˆ ˆ example. examples are the total momentum or the total angular momentum of a composite object which are the sum of the (angular) momenta of the elementary components.48) i i R(ϑ) = exp − ϑ · J (1) exp − ϑ · J (2) (6. but solely under simultaneous and identical rotations of rAB or ρC . R−1 (ϑ) ρC ) (6. the commutation relationships (1) Jp .9) hold since the components of J (k) are diﬀerential operators with respect to diﬀerent variables.5) according to (5. Describing quantum mechanically a property of a composite object as a whole and relating this property to the properties of the elementary building blocks is then the quantum mechanical equivalent of the important operation of addittion.7) i R(ϑ) = exp − ϑ · J (6. q = 1.e. (6. ρC ) it follows H R(ϑ) = R(ϑ) H. 2.4) which decouples rotational states is connected with the rotational symmetry of the Hamiltonian of the three particle system considered. according to (5. ρC ) = ψ(R−1 (ϑ) rAB . assuming presently that H might. 5. Hence. To specify this property mathematically let us denote by H the Hamiltonian in the rotated frame. The invariance of the Hamiltonian under overall rotations of the three particle system implies then H = R(ϑ) H R−1 (ϑ) .53.

We recognize.9). i. k = 1. Jk ] = 0 .11) By means of (6.e.81)]. i.71–5.6) for rotational invariance of the Hamiltonian in the form i i H = exp − ϑ · J H exp + ϑ · J . i. 2.15) are satisﬁed. 3 cannot have simultaneous eigenstates among each other. k = 1. ˆ (6. One can show readily that the commutation relationships [Jk . (6. In the following we will use the notation r ρ Ω1 = rAB .11) we can write the condition (6. B + C] = [A.16) following the procedure stated in the theorem above [c. that the components Jk .144 where J = J (1) + J (2) Addition of Angular Momenta and Spin . J ] = i k m Jm for n = 1.e. the (n) (n) (n) properties [Jk .14) We will refer in the following to Jk . For a proof one uses (6.14) implies that the components of the total angular momentum operator (6.. componentwise. In fact. r ρ Deﬁnition of Total Angular Momentum States The commutation property (6.14) implies that the total angular momentum is conserved during the scattering process. B] + [A. a supposition which can be tested through the commutation properties of these operators. J ] = i k m Jm (6. We suspect. of course. 2. (6.12) each individually can have simultaneous eigenstates with the Hamiltonian. Before we apply the procedure (5. (6. 3 . the operators Jk . (6. 2 together with the property [A. Eqs. ˆ Ω 2 = ρC . 2.e.. that energy. To describe the scattering process of AB + C most concisely one seeks eigenstates YJM of J2 and J3 which can be expressed in terms of Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ).71–5. and the eigenvalues of J2 and J3 are good quantum numbers. To order O(|ϑ|) one obtains i ϑ · JH .f. however. 3 as the three components of the total angular momentum operator. (5. for |ϑ| H ≈ 1 − 1 i ϑ·J H 1 + 1 i ϑ·J ≈ H + i 1. According to the theorem above this property implies that one can construct eigenstates YJM of J3 and of J2 = J2 + J2 + J2 1 2 3 (6. [H.12) We consider this equation for inﬁnitesimal rotations.17) . k = 1. 3 do not commute. C]. we will ﬁnd that the states yield the basis B with the desired property of a maximal uncoupling of rotational states. The property (6.13) Hϑ · J − Since this holds for any ϑ it follows HJ − J H = 0 or. 2. the important fact that the Jk obey the Lie algebra of SO(3).81) we want to consider the relationship between YJM and Y 1 m1 (ˆAB ) Y 2 m2 (ˆC ). k = 1.

therefore. or the closely related Wigner 3j-coeﬃcients introduced further below. The two missing quantum 2 2 numbers are 1 and 2 corresponding to the eigenvalues of J (1) and J (2) . J3 . and J3 .1: Show that J2 . 2 |Ω1 .20) This equation can be satisﬁed only if the Clebsch-Gordan coeﬃcients satisfy (J M | 1 m1 2 m2 ) = 0 for m1 + m2 = M .22) We will not adopt such explicit summation since it leads to cumbersum notation. of (6.. J (2) .e. and three quarks..h. 2 |Ω1 . and J (1) · J (2) commute. the reader should always keep in mind that conditions equivalent to (6. i. restrict the sum in (6. m1 . Since YJM sofar speciﬁes solely two quantum numbers. Ω2 ) = m1 . ρC ) are normalized.m2 2 |Ω1 .1 Clebsch-Gordan Coeﬃcients In order to determine YJM we notice that the states Y 1 m1 (Ω1 ) Y 2 m2 (Ω2 ) are characterized by four 2 2 (1) (2) quantum numbers corresponding to eigenvalues of J (1) . 2 |Ω1 . Exercise 6.18) to avoid summation of vanishing terms YJM ( 1 .e. Ω2 ) (1) (2) = 1 m1 2 m2 ) Y 1 m1 (Ω1 ) Y 2 m2 (Ω2 ) (m1 + m2 ) (J M | .1. Equivalent coeﬃcients exist for other symmetry properties of multi–component systems. J3 YJM = M YJM .18) where the states YJM ( 1 .1: Clebsch-Gordan Coeﬃcients 145 6. Ω2 ) = m1 (J M | 1 m1 2 M − m1 ) Y 1 m1 (Ω1 ) Y 2 m2 (Ω2 ) . k = 1. (6. (6. J (1) .m2 (J M | 1 m1 2 m2 ) Y 1 m1 (Ω1 ) Y 2 m2 (Ω2 ) (6.21) One can.18) yields using the property (k) J3 Y k mk (Ωk ) = mk Y k mk (Ωk ) . Why can states YJM then not be speciﬁed by 5 quantum numbers? 2 2 Properties of Clebsch-Gordan Coeﬃcients A few important properties of Clebsch-Gordan coeﬃcients can be derived rather easily. i. These coeﬃcients. (6. i. (6. . We. J3 .s.6. play a cardinal role in the mathematical description of microscopic physical systems. two further quantum numbers need to be speciﬁed for a complete characterization of the total angular momentum states. mesons. The expansion coeﬃcients (J M | 1 m1 2 m2 ) ˆ are called the Clebsch-Gordan coeﬃcients which we seek to determine now. 2 M YJM ( 1 . the eigenvalue being speciﬁed by the quantum number M . baryons. assume the expansion YJM ( 1 . an important example being the symmetry groups SU(N) governing elementary particles made up of two quarks.21) hold. We ﬁrst notice that YJM in (6. hence. However.18) is an eigenfunction of J3 . J (2) .e.19) Noting J3 = J3 + J3 and applying this to the l.

1 + 2 . = {YJM ( 1 . . to a new basis B ( 1 . . The basis B ( 1 . (6. it is not immediately obvious what the J– values are.26) We will show below in an explicit construction of the Clebsch-Gordan coeﬃcients that. J. . In this case the range of |J|–values would be the interval [ |J (1) | − |J (2) | . . the basis B ( 1 .18) constitutes a change of an orthonormal basis B( 1 . m2 = − 2 . |J (1) | + |J (2) |].e. − + 1. Our derivation below will also yield real values for the Clebsch-Gordan coeﬃcients. .24) The basis B( 1 .m2 2 M and with (6. . .18) applied to Y∗ and to YJ JM (J M | m1 . .2: Prove Eq.s. Exercise 6. However. 2 ) has (2 1 + 1)(2 2 + 1) elements. Ω2 ). . 2 ). M = −J.25) i. − 2} 1 + 1. The property dΩ1 dΩ2 Y∗ ( 1 . J (1) and J (2) .23) corresponding to the r. | 1 − 2 | + 1. Since YJM represents the total angular momentum state and Y 1 m1 (Ω1 ) and Y 2 m2 (Ω2 ) the individual angular momenta one may start from one’s classical notion that these states represent angular momentum vectors J.27) together with (6. −J + 1. . . . respectively. 2 ) is also orthonormal2 and must have the same number of elements. . .h..25) We want to state now two summation conditions which follow from the orthonormality of the two basis sets B( 1 . 2) = {Y 1 m1 (Ω1 ) Y 2 2 m2 (Ω2 ).1.28) This property follows from the fact that the basis elements are eigenstates of hermitian operators with diﬀerent eigenvalues. in fact. . J} .146 Theory of Angular Momentum and Spin The expansion (6. 1 + 2. Ω2 ) = δJJ δM M (6. In fact. The orthonormality (Ω1 ) Y 2 m2 (Ω2 )Y 1 m1 Ω1 ) Y 2 m2 (Ω2 ) = δ 1 1 δm1 m1 δ 2 2 δm2 m2 . Ω2 ) YJ M ( 1. m1 = − 1 . 1.24) yields 1 m1 2 m2 ) ∗ 1 m1 2 m2 ) (J M| = δJJ δM M . B2 2) should be 2 |Ω1 . . For each quantum number J there should be 2J + 1 elements YJM with M = −J. (6. JM 2 |Ω1 .h. −J + 1. 2 ) and B ( 1 . . . 2) . . | 1 − 2| + 1. (6. . and that the states can be normalized. This obviously corresponds quantum mechanically to a range of J–values J = | 1 − 2 |. J = | 1 − 2 |. it holds 1+ 2 ( 2J + 1 ) = (2 J=| 1 − 2 | 1 + 1) (2 2 + 1) .s. (6. 2 |Ω1 . (6.. the range of values assumed for J is correct. (6. property implies dΩ1 dΩ2 Y 1 m1 corresponding to the l.

1 m1 2 m2 ) = 0 if J < | 1 − 2| .71–5. therefore.. (6.32) which complements (6. 2 |Ω1 . One can show readily using the same reasoning as applied in the derivation of (6. The (J M | 1 m1 2 m2 ) cJ M .. i. . A similar 2 construction will also be applied to composite systems governed by other symmetry groups. Ω2 )Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) .e. 1+ 2 J Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) = J =| 1 − 2 | M =−J cJ M YJ M ( 1. to composite systems involving spin. 2 ) and B ( 1 . Hence. 2 |Ω1 .. e. also apply the results. it is possible to expand Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ) in terms of YJM ( 1 .34) 6. or baryons multiplets involving three quarks. Ω2 ).28) allows one to conclude that the coeﬃcients cJ (J M | 1 m1 2 m2 )∗ . Ω2 ) .2: Construction of Clebsch-Gordan Coeﬃcients 147 The second summation condition starts from the fact that the basis sets B( 1 . The result of this construction will include all the properties previewed above. Ω2 ) . 2 |Ω1 .30) The latter property follows from multiplying (6. 2 ) span the same function space. i.e.81) stated above.28) from (6. (6. and actually also the properties of Clebsch-Gordan coeﬃcients stated above. or J > 1 + 2 (6. will be based solely on the commutation properties of the operators J and J (k) . (6.e. rather the convention (J M | has been assumed. (6. the group SU(3) in case of meson multiplets involving two quarks. We can.2 Construction of Clebsch-Gordan Coeﬃcients We will now construct the Clebsch-Gordan coeﬃcients.18) by Y∗ J orthogonality property (6.31) are identical to Comparision with (6. At this point we like to stress that the construction will be based on the theorems (5. (6.1 states. Ω2 ) and integrating.18) that the Clebsch-Gordan coeﬃcients obey the second summation condition (J M | 1 m1 2 m2 )∗ (J M | 1 m1 2 m2 ) = δm1 m1 δm2 m2 .6. Y = J =| 1 − 2 | M =−J 1 m1 M (Ω1 )Y 1+ 2 2 m2 (Ω2 ) (J M | ∗ 1 m1 2 m2 ) YJ M J ( 1.m2 M ( 1. 2 |Ω1 .27) yields δJJ δM M = m1 .18).33) JM The latter summation has not been restricted explicitly to allowed J–values.g.. i.29) where the expansion coeﬃcients are given by the respective scalar products in function space cJ M = dΩ1 dΩ2 Y∗ J M ( 1. 2 |Ω1 .

Ω2 ) = (J + M + 1)(J − M )YJM ( 1 . J3 YJJ ( 1 .66. 2 |Ω1 .64) J3 = = (1) + J3 (1) 1 (2) Y 1 1 (Ω1 ) Y 2 2 2 2 (Ω2 ) 1 1 J3 Y ( + 1 1 (Ω1 ) Y 1 1 (Ω2 ) + Y 2 2 (Ω1 ) J3 Y (2) 2 2 (Ω2 ) (6. 1+ 2. 1+ 2 ( 1.39) into (6.35) 1 − 2 |. Ω2 ) . We obtain using (5. 2 |Ω1 .148 For the construction of YJM we will need the operators J± = J1 + iJ2 . The corresponding classical total angular momentum vector Jclass (1) (2) would be obtained by aligning both Jclass and Jclass also along the z–axis and adding these two vectors. Ω2 ) (Ω1 ) dΩ2 Y 2 2 (Ω2 ) = 1. The construction assumes a particular choice of J ∈ {| such J–value seeks an expansion (6. have shown Y 1+ 2. . (6. −J + 1.68) J+ = (1) + J+ (1) (2) Y 1 1 (Ω1 ) Y 2 2 2 2 (Ω2 ) 1 1 (6. −J. 1+ 2 In fact. 2 |Ω1 . | 1 − 2| + 1. Ω2 ). 1+ 2 ( 1.38) for M = J − 1. . 2 |Ω1 . .5.37) The solution needs to be normalized. . Similarly. Quantum mechanically this corresponds to a state G 1+ 2. therefore. 1 + 2} and for (6. . we can also demonstrate using (??) that G dΩ1 = We. (6. . .37). 2 |Ω1 . .43) .41) 2) Y (Ω1 ) Y (Ω2 ) .25) using (6. We embark on the suggested construction for the choice J = 1 + 2 .11) and (5. Ω2 ) Addition of Angular Momenta and Spin (6. 2 |Ω1 . Ω2 ) is normalized dΩ2 G dΩ1 Y 1 1 1+ 2. M = −J. (6. 2 |Ω1 . Having determined such YJJ we then construct the whole family of functions XJ = {YJM ( 1 .37) and replace (1) (2) according to (6.18) which satisﬁes J+ YJJ ( 1 . 2 |Ω1 .36) = 0 = J YJJ ( 1 . Ω2 ) . we can demonstrate condition (6. For this purpose we insert (6.11) J+ by J+ + J+ . To ﬁnd GJJ we start from the observation that GJJ represents the state with the largest possible quantum number J = 1 + 2 with the largest possible component M = 1 + 2 along the z–axis. J − 2.42) ( 1. Ω2 ) = Y 1 1 (Ω1 ) Y 2 2 (Ω2 ) . 2 |Ω1 . .39) which we will try for a solution of (6. . We ﬁrst seek an unnormalized solution GJJ and later normalize. Ω2 ) = Y 1 1 (Ω1 ) Y 2 2 (Ω2 ) (6. Ω2 ) 2 |Ω1 . 1+ 2 ( 1.40) (Ω1 ) J+ Y (2) 2 2 J+ Y 1 1 (Ω1 ) Y (Ω2 ) + Y (Ω2 ) = 0 . J} by applying repeatedly J− YJM +1 ( 1 . (6.

). 1 + 2 − 3.37) is satisﬁed. .44) (Ω2 ) . We demonstrate the procedure ex(1) (1) plicitly only for M = 1 + 2 − 1. We seek for this purpose ﬁrst an unnormalized solution G 1 + 2 −1 1 + 2 −1 of (6. 2 = 1 in Table 1. ( + )}. Normalization yields furthermore Y = 1 1 + 2 −1 1 + 2 −1 We have thereby determined ( 1 2 |Ω1 Ω2 ) (6. in particular.45).46) 2 1 +c Y (Ω1 ) Y 2 2 (Ω2 ) = 0 . To satisfy this equation one needs to choose c = − 1 / 2 . Y 1 1 −1 (Ω1 )Y (Ω2 ) + Y 1 1 This construction can be continued to obtain all 2( 1 + 2 ) + 1 elements of B + and.37).21) we set G 1 + 2 −1 1 + 2 −1 ( 1 2 |Ω1 Ω2 ) = Y 1 1 −1 (Ω1 )Y 2 2 (Ω2 ) + c Y 1 1 (Ω1 )Y 2 2 −1 (Ω2 ) (6.s. 2 |Ω1 . The l. ). M = −( + − ).47) (Ω2 ) − 1 1 2 + Y 2 1 1 −1 (Ω1 )Y 2 2 + Y 2 1 1 (Ω1 )Y 2 2 −1 (Ω2 ) . and all total angular momentum functions for a given choice of 1 and 2 . 2 |Ω1 . One can obviously continue the construction outlined to determine Y 1 + 2 −2 1 + 2 −2 . To demonstrate (6. 1 + 2 − 1. The r. Ω2 ).2: Construction of Clebsch-Gordan Coeﬃcients 149 We now employ property (6. Expression (6.1: Construct following the procedure above the three functions YJM ( 1 .27). Exercise 6. G 1 + 2 −1 1 + 2 −1 in (6. We have provided in Table 1 the explicit form of YJ M ( 1 2 |Ω1 Ω2 ) for 1 = 2 and 2 = 1 to illustrate the construction. with the symmetry pattern and with the terms Y 1 m1 Y 2 m2 contributing to each YJ M . 2 |Ω1 . One obtains then Y 1 1+ 2 1 + 2 1 + 2 −1 2 2 ( 1.h. Show that the resulting functions are orthonormal.s. of (6.2.36. Having constructed this family we have determined the Clebsch-Gordan coeﬃcients ( 1 + 2 − 1M | 1 m1 2 m2 ). . 6.6.39) as required by (6. . | . | . . Ω2) for M = 1 + 2 − 2 and J = 1 + 2 .h.45) for some unknown constant c. Ω2 ) 2 1+ 2 = (Ω1 )Y 2 2 −1 (6. One can readily show that (6. ( + − )}. of (6. etc.36) we proceed as above and obtain J+ Y = (1) 1 1 −1 (Ω1 ) Y 2 2 2 2 (Ω2 ) + c Y 1 1 1 1 (Ω1 ) J+ Y (2) 2 2 −1 (Ω2 ) (6. . According to the condition (6. This expression is orthogonal to (6. The reader should familiarize himself with the entries of the Table. . . The result is illustrated for the case 1 = 2. −( 1 + 2 − 1). . all the Clebsch-Gordan coeﬃcients ( 1 + 2 M | 1 m1 2 m2 ). thereby.38) yields pression 2( 1 + 2 )Y 1 + 2 1 + 2 −1 ( 1 . M = −( + ).47) can serve now to obtain recursively the elements of the family B + − for M = 1 + 2 − 2. . 1 + 2 − 2. .38) yields with J− = J− + J− the ex√ √ 2 1 Y 1 1 −1 (Ω1 )Y 2 2 (Ω2 ) + 2 2 Y 1 1 −1 (Ω1 )Y 2 2 −1 (Ω2 ). . We like to construct now the family of total angular momentum functions B + − = {Y + − M ( . .38) to construct the family of functions B + = {Y + M ( . .

Ω2) Y3−1 (2. 1|Ω1 .57735 −0.547723 0.632456 0.774597 0 2 5 8 15 1 6 3 10 1 3 0. 1|Ω1 . 1|Ω1 .150 Addition of Angular Momenta and Spin Y33 (2. 1|Ω1 . Ω2) Y32 (2.547723 0. 1|Ω1 . 1|Ω1 .1: Some explicit analytical and numerical values of Clebsch-Gordan coeﬃcients and their relationship to the total angular momentum wave functions and single particle angular momentum wave functions.816497 Y22 (Ω1 )Y1−1 (Ω2 ) 1 15 1 3 3 5 1 5 1 2 3 10 2 5 1 2 1 10 2 3 1 3 Y20 (Ω1 )Y11 (Ω2 ) Y21 (Ω1 )Y10 (Ω2 ) 0. Ω2) Y30 (2. 1|Ω1 .632456 0.707107 0. Ω2) Y1−1 (2.316228 0. Ω2) Y2−1 (2. 1|Ω1 . Ω2) Y22 (2.816497 −0. Ω2) Y11 (2. Ω2) Y2−2 (2.774597 0.57735 0.774597 − − − − 0.707107 0. 1|Ω1 .816497 Y2−2 (Ω1 )Y11 (Ω2 ) Y2−1 (Ω1 )Y10 (Ω2 ) Y20 (Ω1 )Y1−1 (Ω2 ) Y2−2 (Ω1 )Y10 (Ω2 ) − 2 3 Y2−1 (Ω1 )Y1−1 (Ω2 ) Y2−2 (Ω1 )Y1−1 (Ω2 ) 1 Table 6. 1|Ω1 .258199 0.57735 0.447214 −0.547723 0. 1|Ω1 . .408248 −0. 1|Ω1 . Ω2) Y31 (2.816497 0. Ω2) − − − − Y22 (Ω1 )Y11 (Ω2 ) 1 Y21 (Ω1 )Y11 (Ω2 ) 2 3 1 3 2 5 1 2 1 10 1 5 1 2 3 10 1 15 1 3 3 5 Y22 (Ω1 )Y10 (Ω2 ) 1 3 2 3 8 15 1 6 0. 1|Ω1 .730297 0.57735 0. Ω2) Y3−2 (2.316228 0. 1|Ω1 .707107 0.258199 −0.632456 −0.57735 0.408248 −0.57735 3 10 3 5 Y2−1 (Ω1 )Y11 (Ω2 ) Y20 (Ω1 )Y10 (Ω2 ) Y21 (Ω1 )Y1−1 (Ω2 ) −0.707107 0.447214 0. Ω2) Y3−3 (2. Ω2) Y21 (2. Ω2) Y20 (2.547723 0. 1|Ω1 . Ω2) Y10 (2.730297 −0.

51) The operators aζ . m − 2 . For this purpsose we will employ the spinor operators introduced in Sections 5. aζ . These expressions will be derived now. + 2 − J−M +1 for = J + 1 2J+2 2 J−M for = J − 1 2J 2 = .6. a† ζ . b † ζ = δζζ = δζζ . − 1 2 2 2 J+M +1 for = J + 1 2J+2 2 The construction described provides a very cumbersome route to the analytical and numerical values of the Clebsch-Gordan coeﬃcients. b† refer to diﬀerent particles and. ζ = ± and the matrix elements < ζ |σk | ζ > are as deﬁned in Section 5.54) 2 + = √ b† ( 2 +m2 )! √− ( 2 −m2 )! .2. bζ = a† . (6. m + 1 .52) and | 2 m2 2 According to Section 5. 2 . M | . 6.ζ ζ. Accordingly. b† ζ = 0. bζ . we are dealing now with two particles carrying angular momentum or spin.48) (6. 1 .3: Explicit Expressions of CG–Coeﬃcients 151 Exercise 6.53) (6. The creation and annihilation operators are again of the boson type with commutation properties aζ . 5. M | .10 [cf. 5. hence. b† ζ ζ = aζ . (6. b† ζ ζ = 0 = 0 . commute with each other ζ ζ aζ .9. bζ = = a† .ζ a† < ζ |σk | ζ > aζ ζ b† < ζ |σk | ζ > bζ ζ (6. we extent deﬁnition (5. It is actually possible to state explicit expressions for any single coeﬃcient (JM | 1 m1 2 m2 ). J. aζ bζ . a† and bζ .254)] the angular momentum / spin eigenstates | 1 m1 of the two particles are | 1 m1 | 2 m2 1 + = √ a† 1 +m1 ( 1 +m1 )! 2 +m2 √− b† a† 1 −m1 ( 1 −m1 )! 2 −m2 |Ψ0 |Ψ0 .287) to two particles (1) J k k = = (2) J 1 2 1 2 ζ.50) (6.10.49) where ζ. (5. Deﬁnition of Spinor Operators for Two Particles In contrast to Sections 5. 1 (6.2: Use the construction for Clebsch-Gordan coeﬃcients above to prove the following formulas J+M 1 for = J − 2 2J 1 1 1 = J.9.10.3 Explicit Expression for the Clebsch–Gordan Coeﬃcients We want to establish in this Section an explicit expression for the Clebsch–Gordan coeﬃcients (JM | 1 m1 2 m2 ).10 where we studied single particle angular momentum and spin. a† ζ ζ b† .

63) according to At this point. m2 a† + ( 1 2 = a† − ( 1 1 −m1 (6. (6. .300).. 5. mj j = j | j .53. m1 1 | 2 . (6. (6. M ( 1 . 2 .55) . 3 .54).67) = j |J. hence. M ( 1 . 6.59) We seek to determine states |J.56) The states | 1 .66) (6. (1) J (2) J 3 | 1 m1 1 3 | 2 m2 2 = m1 | 1 m1 = m2 | 2 m2 1 2 (6. which describe a two particle system according to (6. M ( 2) 2) = J(J + 1) |J. M ( 1 . (j) 2 ˆ ˆ J = kj ( kj + 1 ) . 2) + 1) |J. 1. M ( 1 . m2 |J. as usual are denoted by their respective quantum numbers J.303). M ( 1 . M ( 1 . are | 1 . 2) . J3 |J. |J. 2 )) . mj j and.18) as follows = | 1 m1 . m1 1 | 2 . (2) J which.62) . M ( 1 . −J ≤ M ≤ J . 2 . 2) 2) (6.57) b† + ( (2) 2 2 +m2 1 +m1 b† − ( 2 2 −m2 + m1 )! − m1 )! (1) + m2 )! − m2 )! |Ψo .152 It holds.e.64) (6.61) (6. 2 . M ( 1 . M ( 1 .302. m2 |J. M ( 1 . k2 = b b+ + b† b− k1 = + − − 2 2 + namely. J3 .60) (6. (1) J 2 |J. j = 1. in analogy to Eqs. The operator of the total angular momentum/spin of the two particle system is J = with Cartesian components Jk = (1) J + J (6. M ( 1 . i. (2) J 2 .65) (6.m2 | 1 . ˆ For the operators kj holds ˆ kj | j . k = 1. we like to recall for future reference that the operators (5. (1) J 2 . M. m1 . 2) We will also require below the raising and lowering operators associated with the total angular momentum operator (6. can be expressed in terms of the number operators 1 † 1 ˆ ˆ a† a+ + a† a− . m2 1 2 ( 1 . M ( 1 . M ( 1 . The aim of the present Section is to determine closed expressions for the Clebsch–Gordan coeﬃcents ( 1 . (6. 2) 1. (1) J . 1 . = = 1( 1 2( 2 2) 2) (6.59) J± = J1 ± i J2 . (1) J 2 | (2) J 2 | 1 m1 1 2 m2 2 Addition of Angular Momentum and Spin = = 1( 1 2( 2 + 1) | 1 m1 + 1) | 2 m2 1 2 . 2. for such states should hold J2 |J.69) − 2| ≤ J ≤ + 2 . m1 1 | 2 . + 1) |J. = M |J. M ( (2) J 2 |J. (6.58) Jk + (2) Jk . (5. 2 .68) The states |J. 2 )). 2 ) which are simultaneous eigenstates of the operators 2 2 J2 . m1 . ˆ kj |J. m2 2 . . 2) 2) can be expressed in terms of Clebsch-Gordan coeﬃcients (6.

K† J3 . K † (j) J 2 .73) (6. K † ] = (j) J 2 . + − − + 153 (6. This operator obeys the following commutation relationships with the other pertinent angular momentum / spin operators ˆ kj . K † J± . 6.73) yields J3 . 2 2 3 ˆ = K † kj + K † . a† b† − a† a− .75) The relationships (6. b† = 0 . j = 1.65) and (6. + − + − 2 2 − + = . (6. (6. due to J2 = K† 1 2 J+ J− 1 † K . We note that. ˆ A similar calculation yields [k2 . the relationships (6. and applying the relationships (6.70) will play a crucial role in the evaluation of the Clebsch-Gordan-Coeﬃcients. 4 Using J3 = (1) J 3 + (2) J 3 .50. K † (kj + 1) 2 2 1ˆ † 1 ˆ kj K + K † (kj + 1) 2 2 1 ˆ 1 ˆ K † kj + kj . a† b† + − − + 2 + 2 − 1 1 − b† a† b+ . j = 1. K † 1 a† a+ − a† a− + b† b+ − b† b− .73) can be readily proven. + − − + 2 2 = 1 † 2K . K † 2 1ˆ ˆ 1 ˆ ˆ kj kj + 1. expressing (k) J 3 through the creation and annihilation operators according to (5. a† b† + + − − + 2 2 − 1 1 = a† b† − a† b† = K† .72) (6. For example. K † + K † 2 2 3 † †ˆ K kj + K . 2 4 = 0 = = 0.6. a† b† − a† b† − + − − + 2 + 1 1 = a† a+ . 6.288).64. b† − b† a† b− .74) + 1 J− J+ + J2 . using (6. a† b† + a† a− . Employing (6. K † 1 † a a+ + a† a− .71–6.74) imply 3 2 J2 . K † = 0.71) one can show K† = = = = = 1 ˆ ˆ kj (kj + 1).3: Explicit Expressions of CG–Coeﬃcients The Operator K † The following operator K † = a† b† − a† b† .71–6. 6. a† b† − a† b† + − + − + − − + 2 1 1 = a† a+ .71) (6.73. K † + kj .51) one obtains ˆ k1 .

We consider now the case that (K † )n acts on the simplest total angular momentum / spin state. 6.63. M ( 1 . on the state |j1 + j2 .77). 2) = = (j) 2 J . j2 + ) 2 2 n n = N (n. 2 2 + n ) 2 (6. M ( 1 . M ( 1 + .72) and (6.. K † = Addition of Angular Momentum and Spin a† a− + b† b− . j2 ) . M ( 1 . Action of K † on the states |J.2. = − a† a− . M ( 1 . The commutation properties (6.76) Here N is an unknown normalization constant. 2 (6.73.76) and state K† n |J. (6.75) ascertain that under the action of K † the states |J. j + 2 2 2) However. M ( 1 . 6.e. this result implies that K † |J. 1 2 + 2 . K † ] = 0 is demonstrated in an analoguous way. j1 1 + . 2) = N |J. 2) 2) ˆ K † kj + = K† = j 3 † K + K † j ( j + 1) |J. j1 + j2 (j1 . |j1 + j2 . 2) is a state with quantum numbers 1 1 + 1 2 and = N |J. j1 + ) K † 2 2 (6. M ( 1 . j2 ) = |j1 . One can generalize property (6.292) one can derive similarly J+ . a† b† + b† a† b− .62. K† + K † (j) J 2 |J. b† + − + + + − The property [J− .154 Starting from (5. M ( 1 . 2 ) .78) a state which has been used already in the construction of Clebsch-Gordan coeﬃcients in Section 6.67) (j) J 2 K † |J. it holds K † |J. 2) We want to demonstrate now that the action of K † on the states |J. M ( 1 + n .79) n |j1 + j2 . M ( 1. 2 2 1 + ) . To demonstrate that the resulting states are eigenstates of (1) J 2 and (2) J 2 we exploit (6. 2 ) remain eigenstates of J2 and J3 with the same quantum numbers. 6.77) where N is another normalization constant. M ( 1 . Application of (K † )n to this state yields. j1 1 |j2 . but for diﬀerent 1 and 2 quantum numbers of the operators (1) J 2 and (2) J 2 . according to (6. namely. j2 2 . i. j1 + j2 (j1 . 4 3 + j ( j + 1) |J. j1 + j2 (j1 + n n . M ( 1 . a† b− − a† b† + + + − + = 0. M ( 1 . 2 ) produces again total angular momentum eigenstates to the same J and M quantum numbers of J2 and J3 . 2 ) j + 4 1 3 + K † |J.

79). M ( 1 . Strategy for Generating the States |J. 2) = N( 1 + 2 − J.105) of the spherical harmonics.s.3: Explicit Expressions of CG–Coeﬃcients 155 where we denoted the associated normalization constant by N (n.85) ∆(J.82) for |J. J( 1 .83) The derivation of this expression will be provided further below (see page 158 ﬀ).f.82) 1 × | (J + 2 1 × | (J + 2 1 (J + 2 1 − 1 ) . 2) 2) . j2 as follows J = j1 + j2 which is equivalent to n j1 j2 Accodingly. 2) = ∆(J.79) n. 1 = j1 + n 2 . holds |J. 6. 2) Our construction of the states |J. M ) (J− )J−M |J. j1 . J( 1 . Combining (6.84) (6. 2 ) can be expressed through the r. 2 ) ∆(J.80) = = = 1 1 2 2 − J 1 n − = (J + 2 2 n 1 − = (J + 2 2 + 1 1 − − 2) 2) . (6. (J + 2 − − 2) 1 1) 2 The normalization constant appearing here is actually N( + − J. M ( 1 . M ) (J + × K† 1 − 2 )! (J + 2 − 1 )! J+ 1 − 1 + 2 −J (J− )J−M a† + × 2 (6. M ( 1 . in analogy to the construction (5. M ) . (6. M ( 1 . 1 . allow one to obtain the states |J. J( 1 . 2 = j2 + n 2 (6. K† 1 2 1 + 2 −J × × . M ( 1 . 2 ) . 1.84) with (6. j2 + n ).82. (6.86) b† + J+ 2 − 1 |Ψo . (6.74)] yields |J. = (J + M )! (2J)! (J − M )! 1 2 2) (6. 5. of (6. 2) 1 2 = ( 1 + 2 (2J + 1)! − J)! ( 1 + 2 + J + 1)! 1 2 .104. − 1. For this purpose one needs to choose in (6. 2) . The latter states. 2 ) for −J ≤ M ≤ J as follows |J. J( 1 .6.h. j1 + n .57) and exploiting the fact that J− and K † commute [c. 2 2 It is now important to notice that any state of the type |J. 2 ) exploits the expression (6.81) = N( 1 + 2 1 2 − J.

y) = λr xs y t r |Gst .87) + − + − i. in terms of | 1 . (6. b† ] = 0 [cf.. st The desired expansion of |Gr can be obtained from an alternate evaluation of I(λ.87). x. Comparision with (6.t (6.s.88). r!s!t! r. y) which is st based on the properties aζ f (a† ) |Ψo = ζ ∂ ∂a† ζ f (a† ) |Ψo ζ bζ f (b† ) |Ψo = ζ ∂ ∂b† ζ f (b† ) |Ψo ζ (6.50–6. y) = exp (λJ− ) exp x a† exp x b† |Ψo . (6.90) (6.93) = f (a† + λ a† ) g(b† + λ b† ) |Ψo .50)] − − exp ( λJ− ) f (a† ) g(b† ) |Ψo + + = exp a† a+ − = u f (a† ) exp b† b+ + u g(b† ) |Ψo + λu u! a† a+ − λv v! u f (a† ) × + v × v b† b+ − ∂ g(b† ) |Ψo + u = u λ a† − u! × v λ b† − v! ∂a† + v ∂ ∂b† + f (a† ) × + v g(b† ) |Ψo + (6. follows from the commutation properties (6. y) is a generating function for the states |Gr deﬁned in (6. + + Taylor expansion of the two exponential operators yields immediately I(λ.86) |Gr = Jr a† st − + s b† + t |Ψo (6.92) − − and noting [a+ . 2 [cf.89) i. One obtains then using J− = a† a+ + b† b+ (6.156 Addition of Angular Momentum and Spin Our strategy for the evaluation of the Clebsch-Gordan-coeﬃcients is to expand (6. m2 Gordan-coeﬃcients.57)].e. I(λ.91) which. a† ] = [b+ . (6. + − + − ..e. x.69) yields then the Clebsch- Expansion of an Intermediate State We ﬁrst consider the expansion of the following factor appearing in (6. x.88) in terms of monomials (6.264). For this purpose we introduce the generating function I(λ.86) in terms of monomials 1 +m1 1 −m1 2 +m2 2 −m2 a† a† b† b† |Ψo . in analogy to (5. x. m1 1 | 2 .52).

t xs y t s! t! xs y t s! t! a† + a† + λ a† + − s u u s b† + λ b† + − t v b† + t−v t |Ψo = s.q (−1)s (6.3: Explicit Expressions of CG–Coeﬃcients We conclude I(λ.94) yields I(λ. |J. 157 (6.50).97) yields. y) = s.t × a† + s−q s q b† + t r−q t−r+q a† − q b† − |Ψo (6. 2) = s. With K † given by (6.t = q r! × a† + s q s−q t r−q a† − q × b† + t−r+q b† − r−q |Ψo (6. m2 2 .90) allows one to infer |Gr s. y) = exp a† + λ a† + − exp b† + λ b† + − |Ψo . one can write the right factor in (6. 2 ) in terms of states | 1 . Operation of this operator on (6. M ( 1 .99) . m1 1 | 2 . x.6. Expanding the exponentials in st (6.97). x.t × λ v b† − × r−q v × = t s−u v λu a† − r! q |Ψo λr xs y t r! s! t! r. M ( 1 .95) Comparision with (6.70) holds K† 1 + 2 −J = s 1 + 2 −J b† − s 1 + 2 −J−s (−1)s 1 + 2 −J−s (6.94) One can infer from this result the desired expressions for |Gr . using the commutation property (6. to obtain the desired expansion of |J.88).96) and.98) a† − s a† + b† + s .97) + q)! |Ψo × (J − M )!(J + 1 − 2 )!(J + 2 − 1 )! q!(J + 1 − 2 − q)!(J − M − q)!(M + 2 − J+ 1 − 2 −q × Final Result a† + a† − q b† + M + 2 − 1 +q b† − J−M −q Our last step is to apply the operator (K † ) 1 + 2 −J to expression (6.86) (J− )J−M a† + = q J+ 1 − 2 b† + J+ 2 − 1 |Ψo 1 (6.s. using the deﬁnition (6.

54) between creation operator monomials and angular momentum states allow one to write this |J. j2 .100) ( 1 + 2 − J)!(J − M )!(J + 1 − 2 )!(J + 2 − 1 )! s!( 1 + 2 − J − s)!q!(J + 1 − 2 − q)!(J − M − q)!(M + 2 − 1 1 + q)! × (2 ×| 1 . j2 . m2 = M − 1 + q + s. Using (6.101). (6.102) We want to determine now the expression (6. For this purpose we introduce j1 = 1 (J + 2 1 2 − J. M − +q+s One can conclude that this expression reproduces (6. √ 2J + 1 1 2 . J( 1 . m2 in (6.101) Note that m1 + m2 = M holds.103) this can be written 1 = N 2 ψ(j1 .82). 2 ) − 2) . n)|ψ(j1 .82) and (6. The summation over q corresponds then to the summation over m1 . j2 = 1 (J + 2 2 − 1) . M ( 1 . n = 1 + 2 −J. 2 ) (6. − q − s)!(q + s)! 2 × (M + 1 − 1 + q + s)!( 1 1 + 2 − M − q − s)! 2 − q − s 1 | 2. m2 |J. J( 1 . 2 ) we consider the scalar product J. (6.q (−1)s × (6. M ) 1 = + J)!(− 1 + 2 + J + 1)! 2 2 + 2 − J)!( 1 − ( 1+ + J)! 1 2 × [( × s + m1 )!( − m1 )!( 2 + m2 )!( 2 − m2 )!(J + M )!(J − M )!] 2 (−1)s s!( 1 − m1 − s)!( 2 + m2 − s)! 1 2 1 × The Normalization 1 ( 1 + 2 − J − s)!(J − 1 − m2 + s)!(J − + m1 + s)! (6. To determine N = N ( 1 + 2 −J.53. n) 2 )|J. The Clebsch-Gordan coeﬃcents are then ﬁnally ( 1 .69) since. − 1 .158 Addition of Angular Momentum and Spin ( 1 + 2 − J)!(J − M )!(J + 1 − 2 )!(J + 2 − 1 )! s!( 1 + 2 − J − s)!q!(J + 1 − 2 − q)!(J − M − q)!(M + 2 − a† + 2 1 −q−s 1 + q)! a† − q+s b† + M + 2 − 1 +q+s b† − 1 + 2 −M −q−s |Ψo The relationships (6.83) of the normalization constant N ( 1 + deﬁned through (6. 1 .103) = 1. according to (6. 2 ) ∆(J.69) if one identiﬁes m1 = 1 −q − s . m1 . 1 . M ) (J + 2 )! (J + 2 − 1 )! s.104) . q = 1 − m1 − s and m2 = M − m1 . × 2) = N( 1 + 2 1 − J.6.

j1 1 |j2 .105) The ﬁrst step of our calculation is the expansion of ψ(j1 . we obtain |ψ(j1 . n) = K† n 159 |j1 . j2 .6.108) = λ=0 (n1 + ν)! n1 ! (6. m1 1 |j2 . n) in terms of states |j1 . (6.109) and Taylor expansion of the left hand side of (6. (6. n) = 1 (2j1 )!(2j2 )! a† + n! (2j1 )!(2j2 )! a† + 2j1 −n−s s 2j1 s n s 2j2 a† b† + − |Ψo = n−s (−1)s a† b† − + s b† + (−1)s a† − s (2j1 + n − s)!s!(2j2 + s)!(n − s)! s!(n − s)! 2j2 +s b† + b† − n−s (2j1 + n − s)!s!(2j2 + s)!(n − s)! |Ψo . applying (6. j2 .70) for the operator K † . j2 . m2 2 .108) yields the identity n1 + r − s n1 n2 + s n2 = n1 + n2 + r + 1 n1 + n2 + 1 .106) The orthonormality of the states occurring in the last expression allows one to write (6. Accordingly. One obtains then.104) 1 = N2 (n!)2 (2j1 )!(2j2 )! (2j1 + n − s)!(2j2 + s)! s!(n − s)! 2j2 + s 2j2 (6.108). We employ the expression (6. 1 1−λ n1 +1 1 1−λ n2 +1 = m1 .3: Explicit Expressions of CG–Coeﬃcients where |ψ(j1 .107) s = (n!)2 s 2j1 + n − s 2j1 The latter sum can be evaluated using 1 1−λ a property which follows from ∂ν ∂λν 1 1−λ n1 +1 n1 +1 = m1 n1 + m1 n1 λm1 (6. (6.111) Comparision with (6.112) s .110) which can be written 1 1−λ n1 +n2 +2 = r s n1 + r − s n1 n2 + s n2 λr (6.108) twice. j2 2 .m2 n1 + m1 n1 n2 + m2 n2 λm1 +m2 (6.57) for these states and the expression (6.

interchanging the operators J (1) and J (2) results in J = J (2) + J (1) . M ) = (−1) 1 + 2 −J ( 2 . −m1 . m2 ) by ( 1 . m1 .s. m1 ). M ) corresponding to (6. and .. (6. m1 . 2 . −m1 2|J. m2 |J. M | 1 . 1 . M ) ( 1 . 1 . m2 ) and. 2 . (6. 2 . (2j1 + 2j2 + 1)! (6.g. and J (2) are vectors in R . To derive relationship (6. e. one can interchange also the operator J on the l. 2 . M ) .h. of this equation J (1) = J (2) − J .114) show a simple relationship to the Clebsch Gordan coeﬃcients ( 2 . m1 .116) one expresses the Clebsch-Gordan coeﬃcient on the r. m2 . m1 |J.114) as long as J.4 Symmetries of the Clebsch-Gordan Coeﬃcients The Clebsch-Gordan coeﬃcients obey symmetry properties which reﬂect geometrical aspects of the operator relationship (6. −M ) = (−1) 1 + 2 −J ( 1 .160 Applying this to (6. 2 1+1 (6. ( 1 .116) If one takes the negatives of the operators in (6. J (1) on the r.114) For example. For the quantum mechanical addition of angular momenta the Clebsch Gordan coeﬃcients ( 1 . ( 2 . m1 ) by ( 2 .113) Using the identities (6. m1 .83). −m2 |J.s. m2 |J. M ) = (−1) 2 +m2 2J + 1 ( 2 . of (6. manner to the coeﬃcients ( 1 .118) Finally.114) one obtains − J = − J (1) − J (2) The respective Clebsch-Gordan coeﬃcients ( 1 . m2 |J.116). of (6.h. vice versa. (6.118). M ) corresponding to (6.h.120) can be readily derived from the expression (6. −m2 . (6.11) J = J (1) + J (2) . M ) .103) one obtains the desired result (6.102) by replacing ( 1 .115) This relationship is a trivial consequence of (6. m2 |J. 6.119) The corresponding symmetry property of the Clebsch-Gordan coeﬃcients is ( 1 . m1 ) . m1 |J.120) The symmetry properties (6. m2 . (6.s.116) through formula (6.115). m1 .107) yields 1 = N 2 (n!)2 Theory of Angular Momentum and Spin 2j1 + 2j2 + n + 1 2j1 + 2j2 + 1 n!(2j1 + 2j2 + n + 1)! = N2 . and (6. −m1 . −M ) . J (1) . J.117) are again related in a simple 2 . 2 .102) of the Clebsch-Gordan coeﬃcients. (6. (6. 2 .114) by. We will demonstrate this now. m2 |J. namely.

6.4: Symmetries of the Clebsch-Gordan Coeﬃcients

161

seeks then to relate the resulting expression to the original expression (6.102) to prove identity with the l.h.s. Inspecting (6.102) one recognizes that only the sum S( 1 , m1 , ×

2 , m2 |J, M )

=

s

(−1)s s!( 1 − m1 − s)!( 2 + m2 − s)!

2

1 ( 1 + 2 − J − s)!(J − 1 − m2 + s)!(J −

+ m1 + s)!

(6.121)

is aﬀected by the change of quantum numbers, the factor in front of S being symmetric in ( 1 , m1 ) and ( 2 , m2 ). Correspondingly, (6.116) implies S( 1 , m1 ,

2 , m2 |J, M )

= (−1)

1 + 2 −J

S( 2 , m2 ,

1 , m1 |J, M )

.

(6.122)

**To prove this we note that S on the r.h.s. reads, according to (6.121), S( 2 , m2 ,
**

1 , m1 |J, M )

=

s

(−1)s s!( 2 − m2 − s)!( 1 + m1 − s)! 1 − m1 + s)!(J − 2

1

×

(

1+

2 − J − s)!(J −

+ m2 + s)!

.

(6.123)

**Introducing the new summation index s = and using the equivalent relationships s =
**

1 1

+

2

−J −s

(6.124)

+

2

− J − s ,

−s = J − −

1

−

2

+ s

(6.125)

**to express s in terms of s in (6.123) one obtains S( 2 , m2 , (−1)
**

1 , m1 |J, M )

1

= (−1)−s ( 1 + 2 − J − s )!(J − 1 − m2 + s )!(J −

2 2

+

2

−J s

+ m1 + s )! (6.126)

×

s !(

1 − m1 − s )!( 1

+ m2 − s )!

.

Now it holds that 1 + 2 − J in (6.124) is an integer, irrespective of the individual quantum numbers 1 , 2 , J being integer or half-integer. This fact can best be veriﬁed by showing that the construction of the eigenstates of (J (1) + J (2) )2 and (J (1) + J (2) )3 in Sect. 6.2 does, in fact, imply this property. Since also s in (6.102) and, hence, in (6.122) is an integer, one can state that s , as deﬁned in (6.124), is an integer and, accordingly, that (−1)− s = (−1)s (6.127)

holds in (6.126). Reordering the factorials in (6.126) to agree with the ordering in (6.121) leads one to conclude the property (6.122) and, hence, one has proven (6.116).

162

Theory of Angular Momentum and Spin

To prove (6.118) we note that in the expression (6.102) for the Clebsch-Gordan coeﬃcients the prefactor of S, the latter deﬁned in (6.121), is unaltered by the change m1 , m2 , M → −m1 , −m2 , −M . Hence, (6.118) implies S( 1 , m1 ,

2 , m2 |J, M )

= (−1)

1 + 2 −J

S( 1 , −m1 ,

2 , −m1 2|J, −M )

.

(6.128)

**We note that according to (6.121) holds S( 1 , −m1 , ×
**

2 , −m1 2|J, −M ) = s

(−1)s s!( 1 + m1 − s)!( 2 − m2 − s)!

2

1 ( 1 + 2 − J − s)!(J − 1 + m2 + s)!(J −

− m1 + s)!

.

(6.129)

Introducing the new summation index s as deﬁned in (6.124) and using the relationships (6.125) to replace, in (6.129), s by s one obtains S( 1 , −m1 , (−1)

2 , −m2 |J, −M )

=

1

1 + 2 −J

s

(−1)−s ( 1 + 2 − J − s )!(J − 2 + m1 + s )!(J −

1

− m2 + s )! (6.130)

×

1 s !( 2 + m2 − s )!(

− m1 − s )!

.

For reasons stated already above, (6.127) holds and after reordering of the factorials in (6.130) to agree with those in (6.121) one can conclude (6.128) and, hence, (6.118). We want to prove ﬁnally the symmetry property (6.120). Following the strategy adopted in the proof of relationships (6.116) and (6.118) we note that in the expression (6.102) for the ClebschGordan coeﬃcients the prefactor of S, the latter deﬁned in (6.121), is√ symmetric in the pairs of quantum numbers ( 1 , m1 ), ( 2 , m2 ) and (J, M ), except for the factor 2J + 1 which singles out J. However, in the relationship (6.120) this latter factor is already properly ‘repaired’ such that (6.120) implies S( 1 , m1 , According to (6.121) holds S( 2 , −m2 , J, M | 1 , m1 ) =

s 2 , m2 |J, M )

= (−1)

2 +m2

S( 2 , −m2 , J, M | 1 , m1 ) .

(6.131)

(−1)s s!( 2 + m2 − s)!(J + M − s)!

1

×

1 ( 2 + J − 1 − s)!( 1 − 2 − M + s)!(

− J − m2 + s)!

.

(6.132)

**Introducing the new summation index s = and, using the equivalent relationships s =
**

2 2

+ m2 − s

(6.133)

+ m2 − s ,

−s = −

2

− m2 + s

(6.134)

**6.5: Spin–Orbital Angular Momentum States to replace s by s in (6.132), one obtains S( 1 , −m1 , (−1)
**

2 , −m1 2|J, −M )

163

=

2

2 +m2

s

(−1)−s ( 2 + m2 − s )!s !(J −

+ m1 + s )!

1

×

1 (J − 1 − m2 + s )!( 1 − m1 − s )!(

+

2

− J − s )!

.

(6.135)

Again for the reasons stated above, (6.127) holds and after reordering of the factorials in (6.135) to agree with those in (6.121) one can conclude (6.131) and, hence, (6.120).

6.5

Example: Spin–Orbital Angular Momentum States

Relativistic quantum mechanics states that an electron moving in the Coulomb ﬁeld of a nucleus experiences a coupling ∼ J · S between its angular momentum, described by the operator J and wave functions Y m (ˆ), and its spin- 1 , described by the operator S and wave function χ 1 ± 1 . As a r 2 2 2 result, the eigenstates of the electron are given by the eigenstates of the total angular momentumspin states Yjm ( , 1 |ˆ) = r ( , m , 1 , σ|j, m) Y m (ˆ) χ 1 σ r (6.136) 2 2

2

m ,σ

which have been deﬁned in (6.18). The states are simultaneous eigenstates of (J (tot) )2 , J3 , J 2 , and S 2 and, as we show below, also of the spin-orbit coupling term ∼ J · S. Here J (tot) is deﬁned as J (tot) = J + S . (6.137) Here we assume for S the same units as for J , namely, , i.e., we deﬁne S = rather than (5.223). Two-Dimensional Vector Representation One can consider the functions χ 1 ± 1 to be represented alternatively by the basis vectors of the

2 2

(tot)

2

σ

(6.138)

space C χ1 1 =

2 2

1 0

,

χ1−1 =

2 2

0 1

.

(6.139)

The states Yjm ( , 1 |ˆ), accordingly, can then also be expressed as two-dimensional vectors. Using r 2 m = m − σ; one obtains Yjm ( , 1 |ˆ) r 2 = ( , m − 1 , 1 , 1 |j, m) Y 2 2 2 r m− 1 (ˆ) 2 1 0 0 1 (6.141)

1 σ = ±2

(6.140)

+ ( , m + 1 , − 1 , 1 |j, m) Y 2 2 2

r m+ 1 (ˆ) 2

164 or Yjm ( , |ˆ) = r

1 2

Addition of Angular Momentum and Spin

( , m − 1 , 1 , 1 |j, m) Y 2 2 2

r m− 1 (ˆ) 2 r m+ 1 (ˆ) 2

1 ( , m + 1 , − 1 , 2 |j, m) Y 2 2

.

(6.142)

In this expression the quantum numbers ( , m ) of the angular momentum state are integers. According to (6.140), m is then half-integer and so must be j. The triangle inequalities (6.220) state in the present case | − 1 | ≤ j ≤ + 1 and, therefore, we conclude j = ± 1 or, equivalently, 2 2 2 1 = j ± 2 . The diﬀerent Clebsch-Gordon coeﬃcients in (6.141) have the values

1 (j − 2 , m − 1 , 1 , 1 |j, m) 2 2 2

=

j +m 2j j −m 2j − j − m +1 2j + 2 j + m +1 2j + 2

(6.143)

(j − 1 , m + 1 , 1 , − 1 |j, m) 2 2 2 2 (j + 1 , m − 1 , 1 , 1 |j, m) 2 2 2 2 (j + 1 , m + 1 , 1 , − 1 |j, m) 2 2 2 2

=

(6.144)

=

(6.145)

=

(6.146)

which will be derived below (see pp. 170). Accordingly, the spin-orbital angular momentum states (6.141, 6.142) are j +m r 2j Yj− 1 m− 1 (ˆ) 2 2 Yjm (j − 1 , 1 |ˆ) = (6.147) r 2 2 j −m Yj− 1 m+ 1 (ˆ) r 2j 2 2 j − m +1 − Yj+ 1 m− 1 (ˆ) r 2j + 2 2 2 . Yjm (j + 1 , 1 |ˆ) = r (6.148) 2 2 j + m +1 Yj+ 1 m+ 1 (ˆ) r 2j + 2

2 2

Eigenvalues For the states (6.147, 6.148) holds (J + S)2 Yjm (j (J + S)3 Yjm (j J Yjm (j S 2 Yjm (j Furthermore, using (J (tot) )2 = ( J + S )2 = J 2 + S 2 + 2J · S or, equivalently, 2J · S = (J (tot) )2 − J 2 − S 2 (6.154) (6.153)

2

1 2 1 2 1 2

, 1 |ˆ) r 2 , 1 |ˆ) r 2 , |ˆ) r

1 2

= = =

2

j(j + 1) Yjm (j

1 2

1 2

, 1 |ˆ) r 2

**(6.149) (6.150) (6.151)
**

1 2

m Yjm (j

2

, 1 |ˆ) r 2

1 2

(j

2

1 2

)(j

1 2

+ 1) Yjm (j

, |ˆ) r

1 2

1 2

, 1 |ˆ) r 2

=

3 4

Yjm (j

, 1 |ˆ) r 2

(6.152)

6.5: Spin–Orbital Angular Momentum States

165

one can readily show that the states Yjm ( , 1 |ˆ) are also eigenstates of J · S. Employing (6.149, r 2 6.151, 6.152) one derives 2J · S Yjm (j − 1 , 1 |ˆ) r 2 2 = = and 2J · S Yjm (j + 1 , 1 |ˆ) r 2 2 = =

2 2 2

[j(j + 1) − (j − 1 )(j + 1 ) − 3 ] Yjm (j − 1 , 1 |ˆ) r 2 2 4 2 2 (j − 1 ) Yjm (j − 1 , 1 |ˆ) r 2 2 2 (6.155)

[j(j + 1) − (j + 1 )(j + 3 ) − 3 ] Yjm (j + 1 , 1 |ˆ) r 2 2 4 2 2

3 ( −j − 2 ) Yjm (j + 1 , 1 |ˆ) . r 2 2

2

(6.156)

Orthonormality Properties The construction (6.141) in terms of Clebsch-Gordon coeﬃcients produces normalized states. Since eigenstates of hermitean operators, i.e., of (J + S)2 , (J + S)3 , J 2 with diﬀerent eigenvalues are orthogonal, one can conclude the orthonormality property

+π 2π

sin θdθ

−π 0

∗ ∗ dφ [ Yj m ( , 1 |θ, φ) ]T Yjm ( , 1 |θ, φ) = δjj δmm δ 2 2

(6.157)

where we have introduced the angular variables θ, φ to represent r and used the notation [· · ·]T ˆ ∗ 1 to denote the transpose of the two-dimensional vectors Yj m ( , 2 |θ, φ) which deﬁnes the scalar product T a∗ c c = a∗ b∗ = a ∗ c + b∗ d . (6.158) ∗ b d d The Operator σ · r ˆ Another important property of the spin-orbital angular momentum states (6.147, 6.148) concerns the eﬀect of the operator σ · r on these states. In a representation deﬁned by the states (6.139), ˆ this operator can be represented by a 2 × 2 matrix. We want to show that the operator σ · r in the basis ˆ

1 r r {( Yjm (j − 2 , 1 |ˆ), Yjm (j + 1 , 1 |ˆ) ) , 2 2 2

j = 1, 2 assumes the block-diagonal form

3 2

. . . ; m = −j, −j + 1, . . . + j }

(6.159)

0 −1 −1 0 0 −1 σ·r = ˆ −1 0

(6.160)

..

.

166

Addition of Angular Momentum and Spin

1 where the blocks operate on two-dimensional subspaces spanned by {Yjm (j − 2 , 1 |ˆ), Yjm (j + r 2 1 1 , 2 |ˆ)}. We ﬁrst demonstrate that σ · r is block-diagonal. This property follows from the commur ˆ 2 tation relationships (tot) [Jk , σ · r ] = 0 , k = 1, 2, 3 (6.161)

tot where Jk is deﬁned in (6.137) To prove this we consider the case k = 1. For the l.h.s. of (6.161) holds, using (6.137),

[ J1 + S1 , σ1 x1 + σ2 x2 + σ3 x3 ] = σ2 [ J1 , x2 ] + [ S1 , σ2 ] x2 + σ3 [ J1 , x3 ] + [ S1 , σ3 ] x3 . (6.162)

The commutation properties [cf. (5.53) for J1 and (5.228), (6.138) for σ and S1 ] [ J1 , x2 ] [ J1 , x3 ] [S1 , σ2 ] [S1 , σ3 ] = = = = −i [ x2 ∂3 − x3 ∂2 , x2 ] = i x3 −i [ x2 ∂3 − x3 ∂2 , x3 ] = − i x2 2 2 [ σ1 , σ2 ] = i σ3 [ σ1 , σ3 ] = − i σ2 (6.163) (6.164) (6.165) (6.166)

**allow one then to evaluate the commutator (6.161) for k = 1 [J1
**

(tot)

, σ · r] = i ( σ2 x3 + σ3 x2 − σ3 x2 − σ2 x3 ) = 0 .

(6.167)

One can carry out this algebra in a similar way for the k = 2, 3 and, hence, prove (6.161). (tot) Since the diﬀerential operators in Jk do not contain derivatives with respect to r, the property (6.161) applies also to σ · r, i.e., it holds ˆ [Jk From this follows J (tot) = =

2

1 σ · r Yjm (j ± 2 , 1 |ˆ) ˆ r 2

(tot)

, σ · r] = 0 , ˆ

k = 1, 2 3 .

(6.168)

σ · r J (tot) ˆ

2

2

1 Yjm (j ± 2 , 1 |ˆ) r 2

j(j + 1) σ · r Yjm (j ± 1 , 1 |ˆ) , ˆ r 2 2

(6.169)

1 i.e., σ · r Yjm (j ± 2 , 1 |ˆ) is an eigenstate of (J (tot) )2 with eigenvalue 2 j(j + 1). One can prove ˆ r 2 (tot) similarly that this state is also an eigenstate of J3 with eigenvalue m. Since in the space spanned by the basis (6.159) only two states exist with such eigenvalues, namely, Yjm (j ± 1 , 1 |ˆ), r 2 2 one can conclude 1 σ · r Yjm (j + 2 , 1 |ˆ) = α++ (jm) Yjm (j + 1 , 1 |ˆ) + α+− (jm) Yjm (j − 1 , 1 |ˆ) ˆ r r r 2 2 2 2 2

(6.170)

and, similarly, σ · r Yjm (j − 1 , 1 |ˆ) = α−+ (jm) Yjm (j + 1 , 1 |ˆ) + α−− (jm) Yjm (j − 1 , 1 |ˆ) . ˆ r r r 2 2 2 2 2 2 (6.171)

6.5: Spin–Orbital Angular Momentum States

167

We have denoted here that the expansion coeﬃcients α±± , in principle, depend on j and m. We want to demonstrate now that the coeﬃcients α±± , actually, do not depend on m. This property follows from (tot) [ J± , σ · r ] = 0 ˆ (6.172) which is a consequence of (6.168) and the deﬁnition of J± notation α±± (j)

(tot)

[c.f. (6.35)]. We will, hence, use the

Exercise 6.5.1: Show that (6.172) implies that the coeﬃcients α±± in (6.170, 6.171) are independent of m. We want to show now that the coeﬃents α++ (j) and α−− (j) in (6.170, 6.171) vanish. For this 1 purpose we consider the parity of the operator σ · r and the parity of the states Yjm (j ± 2 , 1 |ˆ), ˆ r 2 i.e., their property to change only by a factor ±1 under spatial inversion. For σ · r holds ˆ σ · r → σ · (−ˆ) = − σ · r , ˆ r ˆ (6.173)

i.e., σ · r has odd parity. Replacing the r-dependence by the corresponding (θ, φ)-dependence and ˆ ˆ noting the inversion symmetry of spherical harmonics [c.f. (5.166)] Yj+ 1 m± 1 (π − θ, π + φ) = (−1)j+ 2 Yj+ 1 m± 1 (θ, φ)

2 2 2 2 1

(6.174)

one can conclude for Yjm (j + 1 , 1 |ˆ) as given by (6.142) r 2 2 Yjm (j + 1 , 1 |θ, φ) → Yjm (j + 1 , 1 |π − θ, π + φ) = (−1)j+ 2 Yjm (j + 1 , 1 |θ, φ) . 2 2 2 2 2 2 Similarly follows for Yjm (j − 1 , 1 |ˆ) r 2 2

1 Yjm (j − 1 , 2 |θ, φ) → (−1)j− 2 Yjm (j + 1 , 1 |θ, φ) . 2 2 2 1 1

(6.175)

(6.176)

We note that Yjm (j + 1 , 1 |ˆ) and Yjm (j − 1 , 1 |ˆ) have opposite parity. Since σ · r has odd parity, r r ˆ 2 2 2 2 i.e., when applied to the states Yjm (j ± 1 , 1 |ˆ) changes their parity, we can conclude α++ (j) = r 2 2 α−− (j) = 0. The operator σ· r in the two-dimensional subspace spanned by Yjm (j ± 1 , 1 |ˆ) assumes ˆ r 2 2 then the form 0 α+− (j) σ·r = ˆ . (6.177) α−+ (j) 0

∗ Since σ · r must be a hermitean operator it must hold α−+ (j) = α+− (j). ˆ According to (5.230) one obtains ( σ · r )2 = 1 . ˆ 1

(6.178)

This implies |α+− (j)| = 1 and, therefore, one can write σ·r = ˆ 0 e−iβ(j) eiβ(j) 0 , β(j) ∈ R . (6.179)

One can demonstrate that σ · r is, in fact, a real operator. For this purpose one considers the ˆ operation of σ · r for the special case φ = 0. According to the expressions (6.147, 6.148) for ˆ

f.1 states χ 1 ± 1 2 2 2 [c. φ) = − Yj 1 (j + 1 . 1 |θ. 1 |ˆ) = −Yjm (j ˆ r 2 2 ˆ The Operator σ · p ˆ The operator σ · p plays an important role in relativistic quantum mechanics.180).168 Addition of Angular Momentum and Spin 1 Yjm (j ± 2 . 1 |ˆ) = ((σ · p f (r)))Yjm (j ± 1 . 1 |θ = 0. 1 |ˆ). 1 |ˆ) r r r 2 2 2 2 2 2 (6. (6. 1 |θ = 0.160).181.184) 0 1 space χ 1±1 2 2 one can conclude from (6.186) Here (( · · · )) denotes again conﬁnement of the diﬀusion operator ∂r to within the double bracket.147. therefore. 2 |ˆ) and Yj 1 (j + 1 . We want to demonstrate ﬁnally that the “−”-sign holds in (6.188) .224)] 1 0 σ·r = ˆ . 1 |θ = 0.174–5. According to (5. 6.180) where the sign could depend on j. φ) . have proven (6. (6.180) and. 6. 1 |ˆ) at θ = 0 are r r 2 2 2 2 Yj 1 (j − 1 . One can conclude then ˆ σ·r = ± ˆ 0 1 1 0 (6. 1 |θ = 0.187) 1 2 . φ) 2 2 2 = − j+ 1 2 4π (6.185) We have. Since f (r) is independent of θ and φ follows ˆ ((σ · p f (r))) = −i (( ∂r f (r) )) σ · r .181) 0 = j+ 1 2 4π Yj 1 (j + 1 . ˆ 2 2 2 2 2 2 for θ = 0. Noting that p = −i r is a ﬁrst order 2 2 diﬀerential operator it holds ˆ ˆ ˆ σ · p f (r) Yjm (j ± 1 .183) in case θ = 0 becomes in the standard representation with respect to the spin.177) one notes that for φ = 0 the spin-angular momentum states 2 are entirely real such that σ · r must be real as well. given by ˆ σ · r = σ1 sin θ cos φ + σ2 sin θ sin φ + σ3 cos θ . for θ = 0 (6. ˆ (6. For this purpose we consider the application of σ · r in the case of θ = 0. We want to determine ˆ its action on the wave functions f (r) Yjm (j ± 1 . φ) 2 2 2 . 1 |ˆ) r 2 (6.182) σ · r Yj 1 (j − 1 . identiﬁed the sign of (6. (5.182) 0 Since σ · r. hence. φ) and (5. The result can also be stated in the compact form σ · r Yjm (j ± 1 .148) the particular ˆ 1 1 states Yj 1 (j − 2 . ˆ (6.180) and (6. 1 |ˆ) + f (r) σ · p Yjm (j ± 1 .

hence.187) one obtains ˆ σ · p f (r) Yjm (j ± 1 . x3 x2 ˆ ˆ ( p × r )1 h = −i (∂2 − ∂3 ) h (6. 1 |ˆ) . we conclude the intuitively expected identity 1 ˆ ˆ p×r = − J . ∂3 (1/r) = −x3 /r3 and ∂2 x3 = x3 ∂2 .186) for both terms in (6. 1 |ˆ) r 2 (6.193) r r 1 1 1 = −i (∂2 x3 − ∂3 x2 ) h − i h (x3 ∂2 − x2 ∂3 ) .156) this yields ﬁnally 1 ˆ σ · p f (r) Yjm (j + 2 . 1 |ˆ) r 2 1 2 (6.53).191) (6. Corresponding results are obtained for the other components of p × r and.189) The celebrated property of the Pauli matrices (5.5: Spin–Orbital Angular Momentum States Using (6.194) ˆ ˆ where J1 is deﬁned in (5. r r r Using ∂2 (1/r) = −x2 /r3 . r 2 (6. 6. ∂3 x2 = x2 ∂3 we obtain ˆ ˆ ( p × r )1 h = −i 1 1 (x3 ∂2 − x2 ∂3 ) h = − J1 h r r (6. 1 |ˆ) = r 2 2 ˆ i ∂r f (r) − f (r) σ · p σ · r Yjm (j ˆ 1 2 169 .195) .. 1 |ˆ) r 2 (6. ˆ For the test function h(r) holds ˆ ˆ p·rh = = Using (1/r) = −r/r3 and r · ˆ −i −i · r h r = −i h r ·r − i r· h. h r (6. 1 |ˆ) = i [ ∂r + r + ] f (r) Yjm (j 2 r r Using (6. 1 |ˆ) r 2 (6. e.6.g. To demonstrate this we consider one of its cartision components. r 2 2 1 J ·S 2 1 ˆ σ · p f (r) Yjm (j ± 2 .155.190) can be related to the angular momentum operator.190) 3 h − i hr · r 1 − i r· ˆ r h = ∂r h one can conclude ˆ ˆ p · r h = −i 2 + ∂r r h (6. 1 |ˆ) = 0 and σ = 2S/ . 1 |ˆ) r 2 = i ∂r + −j r 1 g(r) Yjm (j + 2 .192) ˆ ˆ The operator p × r in (6.196) = i j+ ∂r + r 1 2 3 2 1 f (r) Yjm (j − 2 . r Altogether we obtain.198) .197) 1 ˆ σ · p g(r) Yjm (j − 2 .230) allows one to express ˆ ˆ ˆ ˆ ˆ σ · p σ · r = p · r + i σ · (p × r) . using ∂r Yjm (j ± 1 .

144) for m = j. i.206) .143–6. 1 |ˆ) r 2 2 The Clebsch-Gordan coeﬃcients are then (j − 1 . consider ﬁrst the case of the largest m-value m = j. adopting the method in Sec.198).202) Yjj (j − 1 .151).2. 1 |ˆ) r 2 2 r 2 3 i i 1 2 1 r = [ i∂r + ( − j) ] [ i∂r + (j + ) ] f (r) Yjm (j + 2 . In this case holds. 6.199). 1 |ˆ) 2 r 2 r 2 j+3 ( 1 − j)(j + 3 ) 2 2 2 2 2 = [ −∂r − ∂r + − 2 ] f (r) Yjm (j + 1 . 1 |ˆ) = Yj− 1 j− 1 (ˆ) χ 1 1 . 1 |j. 5. 6. 1 |ˆ) = r 2 2 which agrees with (6.205) which agrees with the expressions (6.197. We begin with the coeﬃcients (6. Evaluation of Relevant Clebsch-Gordan Coeﬃcients We want to determine now the Clebsch-Gordan coeﬃcients (6. (6. .2.204) (6. Yjj−1 (j − 1 . 1 |ˆ) r 2 2 i 3 ˆ = σ · p [ i∂r + (j + ) ] f (r) Yjm (j − 1 .201) 1 3 (j + )(j + ) Yjm (j + 1 .45). 1 . r 2 2 2 2 2 2 − r 2 ∂r r + J2 r2 f (r) Yjm (j + 1 . r ∂r2 r 2 (6.99) holds ˆ (σ · p)2 = − 2 2 = ∂2 J2 r + 2 .203) (6. j 2 + 2j + 2 3 4 (6. j + .146).43).143.200) = (j + 1 )(j + 3 ). according to (6. 6. 2 2 (6. 6.101).143. j) 2 2 2 2 (j − . 1 |ˆ) = r 2 2 2j − 1 Yj− 1 j− 3 χ 1 1 + 2 2 2 2 2j 1 Y 1 1 χ1 1 2j j− 2 j− 2 2 − 2 (6.170 Addition of Angular Momentum and Spin To demonstrate the validity of this key result we note that according to (5.199) We want to show that eqs. using (5. r (6.230. according to (6. 1 |ˆ) = J 2 Yjm (j + 1 . 6. − |j.e. in fact.144) and. j − 1 . j) 1 2 1 2 1 2 1 2 = = 1 0 (6. For m = j − 1 one can state. 1 |ˆ) r r 2 2 2 2 2 2 2 yields ˆ (σ · p)2 f (r) Yjm (j + 1 . as well as (6. 1 |ˆ) r 2 2 r r2 r2 j 2 + 2j + 3 2 2 2 4 = [ −∂r − ∂r + ] f (r) Yjm (j + 1 . 1 |ˆ) r 2 2 r r2 and. For this purpose we use the construction method introduced in Sec. are consistent with this identity.. We note ˆ (σ · p)2 f (r) Yjm (j + 1 .

1 |ˆ) = r 2 2 j+m 2j + (j + m − 1)(j − m + 1) Yj− 1 m− 3 χ 1 1 2 2 2 2 j+m Yj− 1 m− 1 χ 1 − 1 2 2 2 2 2j j−m 2j (j + m)(j − m) Yj− 1 m− 1 χ 1 − 1 2 2 2 2 + or Yjm−1 (j − 1 .f.h. − 1 |j. (6. 6. 1 |ˆ) = r 2 2 Applying J− (tot) j+m Yj− 1 m− 1 χ 1 1 + 2 2 2 2 2j j−m Yj− 1 m+ 1 χ 1 − 1 .s. 2 2 2 2 2j .209) (j − 1 .2.144) for m = j − 1. (6. j − 1) 2 2 2 171 = = 2j − 1 2j 1 2j (6. are obtained by iterating the application of (6.143. etc.5: Spin–Orbital Angular Momentum States The corresponding Clebsch-Gordan coeﬃcients are then 1 (j − 1 .210)] yields (j + m)(j − m + 1) Yjm−1 (j − 1 . [c.f.s.210) to (6. and J− + S− to the r.208) (6. 1 |ˆ) = r 2 2 j+m−1 Yj− 1 m− 3 χ 1 1 + 2 2 22 2j (j − m) = This implies (j − 1 . 2 |j. j − 3 . j − 1 .143. j − 1) 2 2 2 2 which again agrees with the expressions (6. as described in Sec.212) 1 + 2j(j − m + 1) 1 2j(j − m + 1) Yj− 1 m− 1 χ 1 − 1 2 2 2 2 j+m−1 Yj− 1 m− 3 χ 1 1 + 2 2 2 2 2j j−m+1 Yj− 1 m− 1 χ 1 − 1 . 6.143.137)] J− (tot) = J− + S− (6. Expression (6.207) (6..202).h.6. 1 |j. (· · · |jj − 3).144) for the Clebsch-Gordan coeﬃcients by induction. 1 .144) implies for j = m Yjm (j − 1 .211) to the l. 6. The further Clebsch-Gordan coeﬃcients (· · · |jj − 2). Let us verify then the expression (6. (6. is obtained by applying the operator [c. 1 . 6. m − 1) 2 2 2 2 = j +m−1 2j (6.210).206). 1 . m − 3 . 2 2 2 2 2j (6.

by the set j1 . m − 1) 2 2 2 2 = (6.214) yields (j. 6.172 Addition of Angular Momentum and Spin j −m+1 2j (j − 1 . − 1 |j.218) where we have replaced the quantum numbers J.143) by applying the symmetry relationships (6. 2j + 2 (6. one can obtain expression (6.144) by induction. However. 2 . m2 1 (6. the relationship (6. hence. 1 .215). The latter relationships applied together read ( 1 . 1 .143. m1 . m + 1 . Jclass and Jclass play equivalent roles.217) 2j + 2 (j + 1 . The Clebsch-Gordan coeﬃcients (6. −m2 | 1 . m1 . 1 |j + 1 .145) from (6. m) 2 2 2 2 2j + 1 (6. m − 1 .144) for j = m − 1.213) which is in agreement with (6. m2 | 3 . j2 .216) 6. 2 . proven (6.143. (1) (2) (−tot) Obviously. 1 . 1 |j + 1 . m1 ) (6. using (6. 6. m.144). In this context Jclass and Jclass play the same role. one obtains (j + 1 . − 1 |j. j3 . to reﬂect in the notation the symmetry of these quantities.120). a higher degree of symmetry is obtained if one rather (−tot) (1) (2) (−tot) considers classically to obtain a vector Jclass with the property Jclass + Jclass + Jclass = 0. They are related in a simple manner to the ClebschGordan coeﬃcients j1 j2 j3 m1 m2 m3 = (−1)j1 −j2 +m3 (2j3 + 1)− 2 (j3 − m3 |j1 m1 . 1+1 3 For 1 (j. m3 . 2j + 1 (6. j+m+1 . leading quantum mechanically to a symmetry of the Clebsch-Gordan coeﬃcients (JM | 1 m1 2 m2 ) with respect to exchange of 1 m1 and 2 m2 . m2 . m + 1 . m3 ) = (−1) × 2 + 3 − 1 + 2 +m2 × 2 . We have.215) which follows from (6.6 The 3j–Coeﬃcients The Clebsch-Gordan coeﬃcients describe the quantum mechanical equivalent of the addition of two (1) (2) classical angular momentum vectors Jclass and Jclass to obtain the total angular momentum vector (tot) (1) (2) (1) (2) Jclass = Jclass + Jclass . m. (1) (2) (−tot) The coeﬃcients which are the quantum mechanical equivalent to Jclass + Jclass + Jclass = 0 are the 3j–coeﬃcients introduced by Wigner. m + 1 ) = 2 2 2 2 and.143).116. m3 . m) = 2 2 2 2 Similarly. 6. all three vectors Jclass . m + 1 ) = 2 2 2 j+m+1 . m1 .214) 2 2 +1 ( 3 . 1 .146) can be obtained from (6. − 1 |j. M. 1 . j2 m2 ) 2 .

j2 . However. with respect to alltogether 12 symmetry operations. the so-called triangle condition.102) of the Clebsch-Gordan coeﬃcients. = (6.220) The latter condition |j1 − j2 | ≤ j3 ≤ j1 + j2 .6. 6. According to the deﬁnition of the 3j–coeﬃcients one would expect symmetry properties with respect to exchange of j1 . e. are identical to the integer arguments which enter the analytical expression (6. The two integer entries J − 1 − m2 and J − 2 + m1 in (6. In case of a non-cyclic exchange of rows and columns the 3j–coeﬃcent assumes a prefactor (−1)Σ . The Regge symbol also vanishes in case that two rows or columns are identical and Σ is an odd integer. except when all elements are non-negative integers and each row and column has the same integer value Σ = j1 + j2 + j3 .223) (Σ + 1)!n11 !n22 !n33 !n23 !n32 !)) n=0 n31 n32 n33 . i. The reader may note that the entries of the Regge-symbol.. states that j1 . m2 and j3 .102) are obtained each through the diﬀerence of two entries of the Regge-symbol. For this purpose one can use the symmetry transformations to place the smallest element into the upper left corner of the Regge symbol. √ safe for the prefactor 2j3 + 1 which is cancelled according to the deﬁnition (6.221) j1 j2 j3 −m1 −m2 −m3 where the the results of a cyclic. discovered by Regge. m3 .218) relating 3jcoeﬃcients and Clebsch-Gordan coeﬃcients. In this way the values of 12 3j-coeﬃcients are related. m1 . there exist even further symmetry properties. j3 form the sides of a triangle and the condition is symmetric in the three quantum numbers. m2 . To represent the full symmetry one expresses the 3j–coeﬃcients through a 3 × 3–matrix.g.e. the Regge-symbol.21. The Regge symbol reﬂects a remarkable degree of symmetry of the related 3j–coeﬃcients: One can exchange rows. j2 .34) imply j1 j2 j3 m1 m2 m3 j1 j2 j3 m1 m2 m3 = 0 if not m1 + m2 + m3 = 0 = 0 if not |j1 − j2 | ≤ j3 ≤ j1 + j2 . for which no known classical analogue exists. as follows −j1 + j2 + j3 j1 − j2 + j3 j1 + j2 − j3 j1 j2 j3 j1 − m1 j2 − m2 j3 − m3 .6: The 3j–Coeﬃcients We ﬁrst like to point out that conditions (6. Because of its high degree of symmetry the Regge symbol is very suited for numerical evaluations of the 3j–coeﬃcents.219) (6. m3 and with respect to sign reversals of all three values m1 . one can exchange columns and one can reﬂect at the diagonal (transposition). Assuming this placement the Regge symbol can be determined as follows (n11 is the smallest element!) n11 n11 n12 n13 n12 !n13 !n21 !n31 ! n21 n22 n23 = (−1)n23 +n32 sn (6. −j1 + j2 + j3 . anti-cyclic exchange and of a sign reversal are stated.222) m1 m2 m3 j1 + m1 j2 + m2 j3 + m3 The Regge symbol vanishes. 173 (6. These symmetries follow the equations j1 j2 j3 m1 m2 m3 = (−1)j1 +j2 +j3 = j2 j3 j1 m2 m3 m1 = (−1)j1 +j2 +j3 j2 j1 j3 m2 m1 m3 (6. These symmetry operations relate altogether 72 3j–coeﬃcients.

(6.223-6.224) (6. m1 + m2 ) (2 − m1 )! (2 + m1 )! √ 6 (2 − m)! (2 + m)! (1 − m2 )! (1 + m2 )! 1 ≥ |m2 | ∧ 2 ≥ |m| ∧ 2 ≥ |m1 | (6. m2 ) √ 2 1 ≥ |m1 |) 2 (6.231) Here is an explicit value of a Clebsch-Gordan coeﬃcient: 1 1 (70 2 −15 1 |120 −10 60 2 −5 1 ) = 2 2 √ √ √ 4793185293503147294940209340 127 142 35834261990081573635135027068718971996984731222241334046198355 0. m) δ(−m1 .1.232) .227) (2m|2m1 1m2 ) = (−1)m+m1 (m + 2m2 ) δ(m.225) sn−1 . m1 + m2 ) (3 − m)! (3 + m)! (3m|2m1 1m2 ) = √ 105 (2 − m1 )! (2 + m1 )! (1 − m2 )! (1 + m2 )! 1 ≥ |m2 |) ∧ 2 ≥ |m1 | ∧ 3 ≥ |m| (6. (6.10752786393409395427444450130056540562826159542886 We also illustrate the numerical values of a sequence of 3j-coeﬃcients in Figure 6.229) 1 1 (0m| m1 m2 ) = 2 2 i(−1)1−m2 δ(0.228) √ (−1)2m1 −2m2 7 δ(m.174 where Σ s0 sn = − Addition of Angular Momentum and Spin = n11 + n12 + n13 = j1 + j2 + j3 23 ! 32 ! = (n23n− n11 )! (n32n− n11 )! (n11 +1−n)(n22 +1−n)(n33 +1−n) n(n23 −n11 +n)(n32 −n11 +n) (6.226) We like to state ﬁnally a few explicit analytical expressions for Clebsch-Gordan coeﬃcients which were actually obtained using (6.226) by means of a symbolic manipulation package (Mathematica) (1m|2m1 1m2 ) = (−1)1+m+m1 δ(m. m1 + m2 ) (2 − m1 )! (2 + m1 )! √ 10 (1 − m)! (1 + m)! (1 − m2 )! (1 + m2 )! 1 ≥ |m| ∧ 1 ≥ |m2 | ∧ 2 ≥ |m1 | (6. m1 + m2 ) (1 − m)! (1 + m)! √ 1 1 1 1 6 2 − m1 ! 2 + m1 ! 2 − m2 ! 2 + m2 ! 1 1 ≥ |m1 | ∧ ≥ |m2 | ∧ 1 ≥ |m| 2 2 (6.230) 1 1 (1m| m1 m2 ) = 2 2 √ (−1)2m1 −2m2 3 δ(m.

i. . (2 (2 1 2 + 1) (2 + 1) × (2 1 2 + 1) + 1) . . .1: Oscillatory behavior of 3j-coeﬃcients.6. Irreducible Representation We had stated above that the spherical harmonics Y m (Ω)). 1 ..+ 2 } the matrix has the blockdiagonal form D(ϑ) = 1·1×1·1 ¤ -10 60 700 m 10-m -40 -20 0 20 40 60 m 1·3×1·3 . (6. . Similarly. . m1 = − 1 . then for a basis {Y 1 m1 (Ω1 )Y 2 m2 (Ω2 ).+ 1 . If we deﬁne the matrix representation of R(ϑ) by D(ϑ). . the 2–particle total angular momentum wave functions YJM ( 1 . 2 = 1. . .6: The 3j–Coeﬃcients 10 -3 ¥ ¢ £ 175 3j coefficient 20 120 18 16 14 12 10 8 6 4 2 0 -2 -4 -6 -8 -10 -12 -14 -16 ¡ ¢ -60 ¡ Figure 6. . 2 |Ω1 .5). . m2 = − 2 . provide the irreducible representation for D(ϑ). . the rotations in single particle function space.e. Ω2 ) provide the irreducible representation for the rotation R(ϑ) deﬁned in (6. 2.e.233) . . rotations in 2–particle function space. eigenfunctions of the single particle angular momentum operators J 2 and J3 . .. i.

. (2( 1 + 2 ) + 1) ×(2( 1 + 2 ) + 1) Partitioning in smaller blocks is not possible.2: How many overall singlet states can be constructed from four spin– 1 states 2 | 1 m1 (1) | 1 m2 (2) | 1 m3 (3) | 1 m4 (4) ? Construct these singlet states in terms of the product wave 2 2 2 2 functions above. . . 2.233. Exercise 6.1: Prove Eqs. 1). Ω2 ).3: Two triplet states |1m1 (1) |1m2 (2) are coupled to an overall singlet state Y00 (1. . each of the blocks in (6.6. . . .234) . 2 = 1. 1 . Exercise 6. .234) Exercise 6. .6.6. J} (2 (2 1 2 + 1) (2 + 1) × (2 1 2 + 1) + 1) = (2| 1 − 2 | + 1) ×(2| 1 − 2 | + 1) (2| 1 − 2 | + 3) ×(2| 1 − 2 | + 3) (6. . .176 Addition of Angular Momentum and Spin 1 For the basis {YJM ( 1 . M = −J.6. 2 |Ω1 . . . (6. .233) is further block-diagonalized as follows + 2. J = | 1 − 2 |. Show that the probability of detecting a triplet substate |1m1 (1) for arbitrary polarization (m2 – 1 value) of the other triplet is 3 .

Examples of Tensor Operators The multiplicative operators C∞ ( ) → C∞ ( ) Y ( ) = def Y ( ) (6.223). The latter implies that T transforms like an angular momentum or spin state | m . −k + 1.236) In this equation Dq q (ϑ) denotes the rotation matrix Dq q (ϑ) = kq |R(ϑ)|kq . a mul∂2 ∂2 tiplicative operator T ψ(r) = f (r) ψ(r) or a diﬀerential operator T ψ(r) = ( ∂x2 + ∂x2 + ∂2 ) ψ(r). This implies. Let T |ψ denote the state obtained after the operator T has been applied and let R(ϑ) denote a rotation in the representation of SO(3) or SU(2) which describes rotational transformations of the quantum states under consideration. Note that in the examples mentioned the operator T would be deﬁned within the same representation as R(ϑ).236. (6. S 2 = S1 + S2 + S3 or any other polynomial of Sk . q = −k. The latter can be written T |ψ where T denotes T in the rotated frame given by T = R(ϑ) T R−1 (ϑ) . e. −k + 1. that T belongs to a family of operators {Tkq . .238) .237) The operators T ∈ {Tkq . the operator (5. 6.g.g. (k) (6. e. Such operators T can be characterized through their behaviour under rotational transformations. e. . . some of the r r examples considered below involve tensor operators T of this type. that in case (i) T is an operator C∞ ( ) → C∞ ( ) acting on single particle wave functions. k} such that k Tkq = q =−k (k) Dq q (ϑ) Tkq . k} with the transformation property (6. q = −k.235) The property that T imparts onto states |ψ angular momentum or spin corresponds to T behaving as an angular momentum or spin state multiplying |ψ .7 Tensor Operators and Wigner-Eckart Theorem In this Section we want to discuss operators which have the property that they impart angular momentum and spin properties onto a quantum state. The operator T may also act on multi-particle states like Y 1 m1 (ˆ1 )Y 2 m2 (ˆ2 ). . . .g.e. ∂x2 3 1 2 2 2 2 case (ii) the operator T could be a spin operator Sk deﬁned in (5.42) in case (i) of the position representation of single particle wave functions or the operator (5.7: Tensor Operators 177 6. Rotations transform |ψ as |ψ = R(ϑ)|ψ and T |ψ as R(ϑ)T |ψ .237) are called tensor operators of rank k. i. (k) (6. for example.222) in case (ii) of single particle spin operators. In fact.6.

239) These operators can be expressed in terms of the coordinates x1 .203). x2 . Exercise 6. .6.235 . In fact. 2. The fact that these operators form tensor operators of rank 1.2: Express the transformed versions of the following operators (a) T = x2 − x2 and 1 2 (b) S2 S3 in terms of Wigner rotation matrices and untransformed operators.237) do not require that the rotation R−1 (ϑ) is expressed in terms of Euler angles according to (5. 1. 2. Exercise 6.178 are tensor operators of rank k.7. we will assume presently that the rotation is chosen as follows R(ϑ) = exp ( ϑ+ L+ + ϑ− L− + ϑ3 L3 ) (6. This choice of parametrization allows us to derive conditions which are equivalent to the property (6.240) 1 where we have deﬁned ϑ± = 2 (ϑ1 iϑ2 ) and L± = L1 ± iL2 . 3 follows from the transformation properties of the spherical harmonics derived in Section 1. Examples are Y00 Y1±1 Y10 Y2±2 Y2±1 Y20 1 = √ 4π Theory of Angular Momentum and Spin 3 3 r sinθ e±iφ = (x1 ± ix2 ) 8π 8π 3 3 = r cosθ = x3 4π 4π 1 15 2 2 ±2iφ 1 15 = r sin θ e = (x1 ± ix2 )2 4 2π 4 2π 15 2 15 = r sinθ e±iφ = (x1 ± ix2 ) x3 8π 8π 5 2 3 2 1 5 3 2 r2 = r cos θ − = x3 − 4π 2 2 4π 2 2 = (6.237). x3 .3. but rather any rotation and any parametrization can be assumed.1: Show that the following set of spin operators T00 T1±1 T10 T2±2 T2±1 T20 = 1 = 1 √ S± 2 S± 2 = S3 = = = (S3 S ± + S ± S3 ) 2 2 (3S3 − S 2 ) 3 are tensor operators of rank 0.7. For the following it is important to note that the rotation matrix elements (6. but are far easier to ascertain for any particular operator.

(6.235 . ϑ3 )T . 6.241) Using R(ϑ) = 1 + ϑ+ L+ + O(ϑ2 ) this equation can be rewritten neglecting terms of order O(ϑ2 ) 1 + + k (1 + ϑ+ L+ ) Tkq (1 − ϑ+ L+ ) = 1 1 q =−k kq |1 + ϑ+ L+ |kq Tkq . J3 yields [J+ . (6.242) from which follows by means of kq |1 1|kq equation ϑ+ [L+ . Exercise 6.4: Is the 1-particle Hamiltonian 2 H = − 2 2m + V (|r|) (6.246. (k + q + 1)(k − q) (k + q)(k − q + 1) (6. Similar equations can be derived for inﬁnitesimal rotations of the form ϑ = (0.6.246) (6.3: Derive Eqs.248) a tensor operator? Exercise 6. J− .5: Consider a system of two spin. Tkq ] [J3 . Tkq ] = = δqq and by subtracting Tkq on both sides of the k ϑ+ kq |L+ |kq Tkq .6. Expressing the results in terms of the angular momentum operators J+ . Exercise 6. 0)T . 1 (6.247) These properties often can be readily demonstrated for operators and the transformation properties (6.1 particles for which the ﬁrst spin is described by 2 the operator S (1) and the second spin by the operator S (2) . 0)T in case of small ϑ+ .80) follows kq |L+ |kq = −i (k + q + 1)(k − q)δq and.247).237) yields ﬁrst k R(ϑ) Tkq R −1 (ϑ) = q =−k kq |R(ϑ)|kq Tkq . We ﬁrst consider a rotation with ϑ = (ϑ+ . For this 2 2 purpose state ﬁrst the proper rotation operator R(ϑ) and note then that S (1) · S (2) commutes with 1 the generators of the rotation of | 2 m1 (1) | 1 m2 (2) .7: Tensor Operators 179 The conditions can be derived if we consider the property (6. Show that S (1) · S (2) is a tensor operator of rank 0 in the space of the products of the corresponding spin states | 1 m1 (1) | 1 m2 (2) .237) be assumed then. 2 Exercise 6. Tkq ] [J− .6: Form tensor operators of rank 1 in terms of the three components of acting on the space of 1-particle wave functions. 0.245) (6.6.237) for transformations characterized by inﬁnitesimal values of ϑ+ . .7.244) [L+ . 0. Tkq ] = = = Tkq+1 Tkq−1 q Tkq . ϑ− .243) From (5.7. Tkq ] = −iTk q+1 (k + q + 1)(k − q) . hence. (0. ϑ3 . q =−k q+1 (6.6. (6.235 .7.235 . ϑ− .7. The property (6.

J3 | 2 m2 γ2 = m2 | 2 m2 γ2 . m1 . 2 |γ2 ) are orthogonal to | 1 m1 in case of diﬀerent quantum numbers 1 . J3 .252) ∼δm1 q+m2 = ( 1 m1 |kq 2 m2 ) Tkq Similarly. Starting point of the derivation is the fact that the states Tkq | 2 m2 γ2 behave like angular momentum states of a composite system of two particles each carrying angular momentum or spin. 2 |γ2 ) which correspond to total angular momentum states. J 2 | 2 m2 γ2 = 2 2( 2 + 1) | 2 m2 γ2 . 2 |γ2 ) is an eigenstate of J 2 with eigenvalue 2 1 ( 1 + 1). 1 m1 |Φ 1 m1 (k. .249) which demonstrates.e. Before we proceed we like to point out that the states | 2 m2 γ2 are also eigenstates of J 2 .e. 2 |γ2 ) = ( 1 m1 |kq 2 m2 ) Tkq | 2 m2 γ2 .m2 2 2 2 We want to show now that these states are eigenstates of J 2 = J1 + J2 + J3 and of J3 where J1 . The hermitian property of J3 and J 2 implies that states Φ 1 m1 (k. To prove this we consider the transformation of Tkq | 2 m2 γ2 R(ϑ) Tkq | 2 m2 γ2 = R(ϑ) Tkq R−1 (ϑ) R(ϑ) | 2 m2 γ2 = q m2 Dq q Tkq Dm2 m2 | 2 m2 γ2 2 (k) ( ) (6.m2 ( 1 m1 |kq 2 m2 ) (J3 Tkq − Tkq J3 + Tkq J3 ) | 2 m2 γ2 = qTkq = q.8 Wigner-Eckart Theorem A second important property of tensor operators Tkq beside (6. which is characterized also by a set of other quantum numbers γ1 which are not aﬀected by the rotational transformation R(ϑ).m2 2 m2 ) (q + m2 ) Tkq | 2 m2 γ2 | 2 m2 γ2 (6. behave like |kq | 1 m1 .21) of Clebsch-Gordan coeﬃcients J3 Φ 1 m1 (k. J2 . hence.250) q.e. | 1 m1 γ1 denotes an angular momentum (spin) state. (6.253) Exercise 6. J3 are the generators of the rotation R(ϑ).235 . in fact.1: Show that Φ 1 m1 (k. possibly the total angular momentum– spin state of a compositie system. 2 |γ2 ) = q.251) The corresponding property for Φ 1 m1 (k. J3 | 2 m2 γ2 = m2 γ2 | 2 m2 and the property (6. One can.6. 2 |γ2 ) = Cδ 1 1 δm1 m1 2 1( 1 (6. i. i.8. one can show that Φ 1 m1 (k. The relationships among the matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 are stated by the Wigner–Eckart theorem which we will derive now.237) is that their matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 obey simple relationships expressed in terms of Clebsch-Gordan coeﬃcients.18) are deﬁned through Φ 1 m1 (k. construct states Φ 1 m1 (k. i. (6. the stated property.247). These states according to (6. 2 |γ2 ) is an eigenstate of J 2 with eigenvalue + 1).m2 ( 1 m1 |kq m1 q.180 Theory of Angular Momentum and Spin 6. 2 |γ2 ) can be shown readily as follows using (6.

e.32) Tkq | 2 m2 γ2 = and orthogonality property (6.253) 1 m1 γ1 |Tkq | 2 m2 γ2 1 m1 γ1 |Tkq | 2 m2 γ2 we express using the equivalent of 1 m1 ( 1 m1 |kq 2 m2 ) Φ 1 m1 (k 2 |γ2 ) (6. γ2 = 2 1 +1 1 m1 . the matrix elements 1 m1 γ1 |Tkq | 2 m2 γ2 can be reduced to an m1 –independent factor.258) We can then ﬁnally express the matrix elements of the tensor operators Tkq as follows 1 m1 .g. To prove this property we consider 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) for a diﬀerent m1 value. One can then evaluate also the corresponding Clebsch-Gordan coeﬃcient ( 1 m1 |kq 2 m2 ) and determine 1 .8: Wigner-Eckart Theorem 181 In order to evaluate the matrix elements (6. γ2 is determined by applying (6. γ1 ||Tk || 2 . γ1 |Tkq | 2 m2 . In order to express the m1 – independence explicitly we adopt the following notation 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) = (−1)k− 2+ 1 √ 1 2 1 +1 1 . γ2 (−1)k− 2 + 1 ( 1 m1 |kq 2 m2 ) (6. i.8.259) The socalled reduced matrix element 1 .6. γ1 |Tkq | 2 m2 . γ1 ||Tk || 2 . γ2 . one obtains 1 m1 + 1γ1 |Φ 1 m1 +1 (k 2 |γ2 ) = 1 √ 1 m1 γ1 |J− Φ 1 m1 +1 (k 2 |γ2 ) ( 1 +m1 +1)( 1 −m1 ) 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) = (6. γ2 ( 1 m1 |kq (6.h. (6. m2 . γ1 ||Tk || 2 . γ1 ||Tk || 2 .s.257) which establishes the m1 –independence of 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) . for which the l.260) Exercise 6.255) At this point the important property can be proven that 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) is independent of m1 . (6. can be evaluated as easily as possible. say m1 + 1. q .254) = 1 m1 γ1 |Φ 1 m1 (k 2 |γ2 ) ( 1 m1 |kq 2 m2 ) .2: Determine the matrix elements of the gradient operator d3 rF (r) G(r) of the type (6. Using | 1 m1 + 1γ1 = 1 ( 1 + m1 + 1)( 1 − m1 ) J+ | 1 m1 γ1 (6. its m1 –dependence being expressed solely through a Clebsch-Gordan coeﬃcient.256) and noting that the operator adjoint to J+ is J− . γ2 k− 2 + 1 √ 1 2 m2 ) (−1) 2 1 +1 = 1 . m1 = q = m2 = 0.e.261) .259) to a combination of magnetic quantum numbers m1 .

182 Theory of Angular Momentum and Spin to a tensor when the functions F (r) and G(r) are of the type f (r)Y m (ˆ). For this purpose relate r operator T1q . (The necessary evaluations are cumbersome. φ) +1m (θ. φ) = ( −m)( +m) (2 −1)(2 +1) Y +1m (θ. φ) sinθ Y √ ( +1) (2 +1)(2 +3) Y m (θ. evaluate the matrix for m1 = q = m2 = 0 using cosθ Y ( +1−m)( +1+m) (2 +1)(2 +3) m (θ. but a very useful exercise!) . φ) − √ ( −1) 2 −1)(2 +1) and express the remaining matrix elements using the Wigner–Eckart theorem. φ) + Y −1m (θ. φ) = Y −1m (θ.

The potentials (7.4) is by far the most relevant of the four choices. in potentials which are solely a function of r and are independent of the angles θ. 7. for example.3) describes the motion of a charge in a uniformely charged sphere and can be employed to describe the motion of protons and neutrons in atomic nuclei1 . The second potential is of the square well type −Vo 0 ≤ r ≤ R . φ. Switzerland. It leads to the stationary electronic states of the hydrogen atom (Z = 1).1) conﬁnes a freely moving particle to a spherical box of radius R.2) serve as schematic descriptions of quantum particles. in case of the so-called bag model of hadronic matter.4) governs the motion of electrons in hydrogen-type atoms. i. and crystals.1.2) V (r) = 0 r > R The third potential V (r) = 1 mω 2 r2 2 (7. Potential (7. See. for example. The fourth potential V (r) = − Z e2 r (7. The corresponding wave functions serve basis functions for multi-electron systems in atoms. Simple Models of Complex Nuclei / The Shell Model and the Interacting Boson Model by I. molecules..e.3) describes an isotropic harmonic oscillator . Poststrasse 22. Four examples will be studied. 1993) 1 183 . Talmi (Harwood Academic Publishers. The ﬁrst potential V (r) = 0 0 ≤ r ≤ R ∞ r > R (7. (7.Chapter 7 Motion in Spherically Symmetric Potentials We describe in this section the stationary bound states of quantum mechanical particles in spherically symmetric potentials V (r). The potential (7. 7000 Chur.

Employing in this plane the coordinates r. 2 . (7.14) (7. ˆ A classical particle moving in a potential V (r) is governed by the Newtonian equation of motion (7. i. ˆ H ψE. J3 ψE.1 Radial Schr¨dinger Equation o ˙ mv = − er ∂r V (r) . using (7.184 Spherically Symmetric Potentials 7. The expression of the kinetic energy 1 1 1 J2 p2 ˙ = m r2 + m r2 θ2 = m r2 + ˙ ˙ 2m 2 2 2 2mr2 and conservation of energy yield J2 1 m r2 + ˙ + V (r) = E . stationary states ψE.m (r) = = = E ψE. Jk } where H is the Hamiltonian.m (r) m ψE. (5.87) for Jk as well as the commutation property (5.f. 2 2mr2 (7. ˆ dt (7.5) In the case of an angular independent potential angular momentum J = m r × v is a constant of ˙ motion. ( + 1) ψE.5) and r = v. Jk ] = 0 .15) (7. k = 1.8) This property can also be proven readily employing the expression of the kinetic energy operator [c.e. 3.m (r) . Accordingly.11) (7.99)] 2 1 2 J2 2 2 = − ∂r r + .61) of J 2 and Jk . k = 1.m (r) (7. one can state ˙ J = m r2 θ . 3 . J ψE.12) . 2. .7) ˆ The correspondence principle dictates then for the quantum mechanical Hamiltonian operator H and angular momentum operators Jk ˆ [H.m (r) can be chosen as simultaneous eigenstates of the Hamiltonian ˆ operator H as well as of J 2 and J3 . . 2. d ˙ J = v × mv + r × mv = 0 + r × er (−∂r V (r)) = 0 .85– 5.9) − 2m 2m r 2mr2 the expressions (5. (7.. the time variation of J can be written. For this purpose one concludes ﬁrst that the conservation of angular momentum J implies a motion of the particle conﬁned to a plane.m (r) . θ for the distance from the origin and for the angular position. 2 .6) ˙ Since Jk is also equal to the poisson bracket {H. Jk } = 0 .13) .m (r) .10) (7. In fact. (7. In classical mechanics one can exploit the conservation of angular momentum to reduce the equation of motion to an equation governing solely the radial coordinate of the particle. one can conclude ˙ Jk = {H.

Multiplying (7. together with the original potential.13).m ψE. (7. (7. ∞[ governed by the eﬀective potential (7. but can also trap particles in the latter space giving rise to strong scattering resonances (see Section ??).20) by −2mr/ 2 yields the so-called radial Schr¨dinger equation o 2 ∂r − ( + 1) − U (r) − κ2 r2 r vκ. Employing the kinetic energy operator in the form (7. 2 (7.16) m 2mr2 Once.7. .m (r) 2 .9) one can write the stationary Schr¨dinger o equation (7.m vE.18) = vE.m (r) describes the radial motion as a one-dimensional motion in the interval [0.24) .20) in the form of the one-dimensional Schr¨dinger equation o 2 − where 2m 2 ∂r + Veﬀ (r) − E φE (r) = 0 . − Adopting for ψE. in the present case. ( + 1) 2mr2 . . 7.20) Since this equation is independent of the quantum number m we drop the index m on the radial wave function vE. .m (r) 1 2 J2 ∂r r + + V (r) − E 2m r 2mr2 the functional form ψE.10) − 1 2 ∂ r + 2m r r 2 2 ( + 1) + V (r) − E 2mr2 . can exclude particles from the space with small r values. equations (7.m (r) = 0. (7. It can be solved by integration of 1 2 2 J2 r = ± ˙ E − V (r) − . (7.10).m (r) = 0. φ) (7. mr2 (t) (7.e. r(t) is determined the angular motion follows from (7.22) which is the original potential 2 ( +1) V (r) with an added rotational barrier potential 2mr2 .21) Veﬀ (r) φE (r) = = V (r) + r vE. .11. using (7.11).22) (7. This barrier.m (r) and E . .17) In analogy to the classical description one can derivem.1: Radial Schr¨dinger Equation o 185 This is a diﬀerential equation which governs solely the radial coordinate. for the wave function of a quantum mechanical particle a diﬀerential equation which governs solely the r-dependence. φ) are the angular momentum eigenstates deﬁned in Section 5.19) where Y m (θ.m (r) Y m (θ.m (r) . i. (7.m .12) are obeyed and one obtains for (7.. One can write (7.23) This demonstrates that the function r vE. (r) = 0 . by integration of ˙ θ = J . . .4.

ψE. Galindo and P. see. (r) ∼ A r + B r− −1 +1 + B r− (7.” by J. (r) = 0 . For r → 0 one may assume that the term ( + 1)/r2 becomes larger than the potential U (r). This follows for > 0 from consideration of the integral which measures the total particle density.33) A detailed discussion of the proper boundary conditions. . We replaced in (7. κ assumes real values. r → 0 (7. For = 0 the contribution of Br−( function is. at r = 0 is found in the excellent monographs Quantum Mechanics I. Berlin. in particular.D.29) Only the ﬁrst term is admissable. 2 2 2m 2 ψE. (r) ∼ A r or vκ.28) (7. assumes the general form r vκ. . 4π|r| (7. for example.m (r) ∼ √ B . (r) (7.25) (7. 2nd Ed. accordingly.30) and.26) 2m 2 In case E < 0. Jackson (John Wiley.20) the index E by the equivalent index κ.20) one needs to specify proper boundary conditions2 . Boundary Conditions In order to solve (7.m (r) ∼ √ 4π B δ(x1 )δ(x2 )δ(x3 ) . The radial part of this integral is ∞ 0 2 dr r2 vκ.31) to the complete wave which. using the expression (5. for > 0 is not integrable. In this case the solution is governed my ( + 1) 2 r vκ. .32) The total kinetic energy resulting from this contribution is. 1975).27) ∂r − r2 and.182) for Y00 . the term Br−( +1) would contribute |B|2 0 dr r−2 +1) (7. Pascual (Springer. according to a well-known result in Classical Electromagnetism3 . 1990) 3 We refer here to the fact that the function Φ(r) = 1/r is the solution of the Poisson equation 2 = −4πδ(x)δ(y)δ(z). New York. ”Classical Electrodynamics. (7. hence. II by A.186 where we deﬁned U (r) κ2 = = − − 2m 2 Spherically Symmetric Potentials V (r) E (7.

7.1: Radial Schr¨dinger Equation o

187

Since there is no term in the stationary Schr¨dinger equation which could compensate this δo function contribution we need to postulate that the second term in (7.29) is not permissible. One can conclude that the solution of the radial Schr¨dinger equation must obey o r vκ, (r) → 0 for r → 0 . (7.34)

The boundary conditions for r → ∞ are governed by two terms in the radial Schr¨dinger equation, o namely, 2 ∂r − κ2 r vκ, (r) = 0 for r → ∞ . (7.35) We have assumed here limr → ∞ V (r) = 0 which is the convention for potentials. The solution of this equation is r vκ, (r) ∼ A e−κr + B e+κr for r → ∞ . (7.36) For bound states κ is real and, hence, the second contribution is not permissible. We conclude, therefore, that the asymptotic boundary condition for the solution of the radial Schr¨dinger equao tion (7.20) is r vκ, (r) ∼ e−κr for r → ∞ . (7.37) Degeneracy of Energy Eigenvalues We have noted above that the diﬀerential operator appearing on the l.h.s. of the in radial Schr¨dinger equation (7.24) is independent of the angular momentum quantum number m. This o implies that the energy eigenvalues associated with stationary bound states of radially symmetric potentials with identical , but diﬀerent m quantum number, assume the same values. This behaviour is associated with the fact that any rotational transformation of a stationary state leaves the energy of a stationary state unaltered. This property holds since (7.8) implies i ˆ [H, exp(− ϑ · J ) ] = 0 . Applying the rotational transformation exp(− i ϑ · J ) to (7.10) yields then i ˆ H exp(− ϑ · J ) ψE,

,m (r)

(7.38)

i = E exp(− ϑ · J ) ψE,

,m (r)

,

(7.39)

i.e., any rotational transformation produces energetically degenerate stationary states. One might also apply the operators J± = J1 ± iJ2 to (7.10) and obtain for − < m < ˆ H J± ψE,

,m (r)

= E J± ψE,

,m (r)

,

(7.40)

**which, together with the identities (5.172, 5.173), yields ˆ H ψE,
**

,m±1 (r)

= E ψE,

,m±1 (r)

(7.41)

where E is the same eigenvalue as in (7.40). One expect, therefore, that the stationary states for spherically symmetric potentials form groups of 2 + 1 energetically degenerate states, so-called multiplets where = 0, 1, 2, . . .. Following a convention from atomic spectroscopy, one refers to the multiplets with = 0, 1, 2, 3 as the s, p, d, f -multipltes, respectively. In the remainder of this section we will solve the radial Schr¨dinger equation (7.20) for the potentials o stated in (7.1–7.4). We seek to describe bound states for the particles, i.e., states with E < 0. States with E > 0, which play a key role in scattering processes, will be described in Section ??.

188

Spherically Symmetric Potentials

7.2

Free Particle Described in Spherical Coordinates

We consider ﬁrst the case of a particle moving in a force-free space described by the potential V (r) ≡ 0 . The stationary Schr¨dinger equation for this potential reads o

2

(7.42)

−

2

2m

− E

ψE (r) = 0 .

(7.43) The general solution of (7.43), as (7.44)

**Stationary States Expressed in Cartesian Coordinates expressed in (3.74–3.77), is ψ(k|r) = N eik·r where E =
**

2 k2

≥ 0. (7.45) 2m The possible energies can assume continuous values. N in (7.44) is some suitably chosen normalization constant; the reader should be aware that (7.44) does not represent a localized particle and that the function is not square integrable. One chooses N such that the orthonormality property d3 r ψ ∗ (k |r)ψ(k|r) = δ(k − k)

Ω∞

(7.46)

holds. The proper normalization constant is N = (2π)−3/2 . In case of a force-free motion momentum is conserved. In fact, the Hamiltonian in the present case

2

Ho = −

2

2m

(7.47)

ˆ commutes with the momentum operator p = ( /i) and, accordingly, the eigenfunctions of (7.45) can be chosen as simultaneous eigenfunction of the momentum operator. In fact, it holds ˆ p N eik·r = k N eik·r . (7.48)

as one can derive using in (7.48) Cartesian coordinates, i.e., ∂1 = ∂2 , k · r = k1 x1 + k2 x2 + k3 x3 . ∂3

(7.49)

Stationary States Expressed in Spherical Coordinates Rather than specifying energy ˆ through k = |k| and the direction of the momentum through k = k/|k| one can exploit the 2 and J given in (5.97) and in (5.92), respectively, fact that the angular momentum operators J 3 commute with Ho as deﬁned in (7.47). This latter property follows from (5.100) and (5.61). Accordingly, one can choose stationary states of the free particle which are eigenfunctions of (7.47) as well as eigenfunctions of J 2 and J3 described in Sect. 5.4.

7.2: Free Particle

189

The corresponding stationary states, i.e., solutions of (7.43) are given by wave functions of the form ψ(k, , m|r) = vk, (r) Y where the radial wave functions obeys [c.f. (7.20)] − 1 2 ∂ r + 2m r r

2 2 m (θ, φ)

(7.50)

( + 1) − E 2mr2

2

,m

vk, (r) = 0 .

(7.51)

**Using (7.45) and multiplying (7.20) by −2mr/
**

2 ∂r −

yields the radial Schr¨dinger equation o r vk, (r) = 0 . (7.52)

( + 1) + k2 r2

We want to determine now the solutions vk, (r) of this equation. We ﬁrst notice that the solution of (7.52) is actually only a function of kr, i.e., one can write vk, (r) = j (kr). In fact, one can readily show, introducing the new variable z = kr, that (7.52) is equivalent to d2 ( + 1) − + 1 z j (z) = 0 . (7.53) 2 dz z2 According to the discussion in Sect. 7.1 the regular solution of this equation, at small r, behaves like j (z) ∼ z for r → 0 . (7.54) There exists also a so-called irregular solution of (7.53), denoted by n (z) which behaves like n (z) ∼ z −

−1

for r → 0 .

(7.55)

We will discuss further below also this solution, which near r = 0 is inadmissable in a quantum mechanical wave function, but admissable for r = 0. For large z values the solution of (7.53) is governed by d2 + 1 dz 2 the general solution of which is j (z) ∼ 1 sin(z + α) z for r → ∞ (7.57) z j (z) = 0 for r → ∞ (7.56)

for some phase α. We note in passing that the functions g (z) = j (z), n (z) obey the diﬀerential equation equivalent to (7.53) d2 2 d ( + 1) + − + 1 g (z) = 0 . (7.58) 2 dz z dz z2 Noting that sin(z + α) can be written as an inﬁnite power series in z we attempt to express the solution of (7.53) for arbitary z values in the form

∞

j (z) = z f (z 2 ) ,

f (z 2 ) =

n=0

an z 2n .

(7.59)

190

Spherically Symmetric Potentials

The unknown expansion coeﬃcients can be obtained by inserting this series into (7.53). We have introduced here the assumption that the factor f in (7.59) depends on z 2 . This follows from d2 z dz 2

+1

f (z) = z

+1

d2 d f (z) + 2( + 1)z f + ( + 1)z dz 2 dz

−1

f

(7.60)

from which we can conclude d2 2 d + ( + 1) + 1 2 dz z dz Introducing the new variable v = z2 yields, using 1 d d = 2 , z dz dv the diﬀerential equation d2 2 +3 d 1 + + 2 dv 2v dv 4v f (v) = 0 (7.63) d2 d2 d = 4v 2 + 2 , 2 dz dv dv (7.62) f (z) = 0 . (7.61)

which is consistent with the functional form in (7.59). The coeﬃcients in the series expansion of f (z 2 ) can be obtained from inserting ∞ an z 2n into (7.63) (v = z 2 ) n=0 an n (n − 1) v n−2 +

n

1 1 (2 + 3) an v n−2 + an v n−1 2 4

= 0

(7.64)

**Changing the summation indices for the ﬁrst two terms in the sum yields an+1 n (n − 1) +
**

n

1 1 (2 + 3) an + an 2 4

v n−1 = 0 .

(7.65)

In this expression each term ∼ v n−1 must vanish individually and, hence, an+1 = − One can readily derive a1 = − 1 1 a0 , 2 1! (2 + 3) a2 = 1 1 a0 . 4 2! (2 + 3)(2 + 5) (7.67) 1 1 an 2 (n + 1) (2n + 2 + 3) (7.66)

**The common factor a0 is arbitrary. Choosing a0 = the ensuing functions ( = 0, 1, 2, . . .) j (z) = z 1 · 3 · 5 · · · (2 + 1) 1 −
**

1 2 2z

1 . 1 · 3 · 5 · (2 + 1)

(7.68)

1!(2 + 3)

+

( 1 z 2 )2 2 − + ··· 2!(2 + 3)(2 + 5)

(7.69)

are called regular spherical Bessel functions.

**7.2: Free Particle One can derive similarly for the solution (7.55) the series expansion ( = 0, 1, 2, . . .) n (z) = − 1 · 3 · 5 · · · (2 − 1) z +1 1 −
**

1 2 2z

191

1!(1 − 2 )

+

( 1 z 2 )2 2 − + ··· 2!(1 − 2 )(3 − 2 )

.

(7.70)

These functions are called irregular spherical Bessel functions. Exercise 7.2.1: Demonstrate that (7.70) is a solution of (7.52) obeying (7.55). The Bessel functions (7.69, 7.70) can be expressed through an inﬁnite sum which we want to specify now. For this purpose we write (7.69) j (z) = z 2 1 The factorial-type products 1 3 · ··· 2 2 + 1 2 (7.72) 1 · · 5 ···( 2 iz 2 ( 2 + 1!( + 3 ) 2

1 2 3 2

+ 1) 2 (( iz 2 + + ··· 2!( + 3 )(2 + 5 ) 2 2

4

(7.71)

**can be expressed through the so-called Gamma-function4 deﬁned through
**

∞

Γ(z) =

0

dt tz−1 e−t .

(7.73)

**This function has the following properties5 Γ(z + 1) = Γ(n + 1) = Γ( )
**

1 2

z Γ(z) n! for n ∈ N √ π π . sin πz

(7.74) (7.75) (7.76) (7.77)

= =

Γ(z) Γ(1 − z) from which one can deduce readily Γ( + 1 ) = 2 One can write then j (z) = √ √

π·

1 3 · ··· 2 2

+

1 2

.

(7.78)

π 2

z 2

∞

iz 2n 2

n! Γ(n + 1 + + 1 ) 2 n=0

.

(7.79)

4 For further details see Handbook of Mathematical Functions by M. Abramowitz and I.A. Stegun (Dover Publications, New York) 5 The proof of (7.74–7.76) is elementary; a derivation of (7.77) can be found in Special Functions of Mathematical Physics by A.F. Nikiforov and V.B. Uvarov, Birkh¨user, Boston, 1988) a

**192 Similarly, one can express n (z) as given in (7.70) n (z) = − √ Using (7.77) for z = 2 πz 1 Γ( + ) 2 1 +
**

iz 2 2 1!( 1 − 2

Spherically Symmetric Potentials

+1

)

+

iz 4 2 2!( 1 − )( 3 2 2

− )

+ ··· .

(7.80)

+ 1 , i.e., 2 Γ( + 1 ) = (−1) 2 Γ( 1 2 π − )

2

(7.81)

yields n (z) = (−1)

+1 √

π

2 z

+1

iz 1 2 + Γ( 1 − ) 1! Γ( 1 − )( 1 − ) 2 2 2 iz 4 2 )( 1 − 2 +1 ∞ n=0

+ or n (z) = (−1)

+1

2! Γ( 1 − 2 √ π 2 2 z

)( 3 − ) 2

+ ···

iz 2n 2

.

(7.82)

n! Γ(n + 1 − − 1 ) 2

.

(7.83)

Linear independence of the Regular and Irregular spherical Bessel Functions We want to demonstrate now that the solutions (7.69) and (7.69) of (7.53) are linearly independent. For this purpose we need to demonstrate that the Wronskian W (j , n ) = j (z) d d n − j (z) n dz dz ( + 1) f1,2 − f1,2 z2 (7.84)

does not vanish. Let f1 , f2 be solutions of (7.53), or equivalently, of (7.58). Using d2 2 d f = − f1,2 + 2 1,2 dz z dz one can demonstrate the identity (7.85)

d 2 W (f1 , f2 ) = − W (f1 , f2 ) dz z This equation is equivalent to d 1 d ln W = ln 2 dz dz z the solution of which is ln W =

(7.86)

(7.87)

c (7.88) z2 for some constant c. For the case of f1 = j and f2 = n this constant can be determined using the expansions (7.69, 7.70) keeping only the leading terms. One obtains c = 1 and, hence, W (j , n ) = 1 . z2 (7.89)

The Wronskian (7.89) doesn not vanish and, therefore, the regular and irregular Bessel functions are linearly independent.

7.2: Free Particle Relationship to Bessel functions

193

The diﬀerential equation (7.58) for the spherical Bessel functions g (z) can be simpliﬁed by seeking the corresponding equation for G + 1 (z) deﬁned through

2

1 g (z) = √ G z Using d g (z) dz d2 g (z) dz 2 = = 1 √ z 1 √ z

+ 1 (z). 2

(7.90)

d 1 √ G 1 (z) G + 1 (z) − 2 dz 2z z + 2 3 d2 1 d √ G G 1 (z) + G 1 (z) − √ dz 2 + 2 4z 2 z z z dz + 2

(7.91)

+ 1 (z) 2

(7.92) one is lead to Bessel’s equation d2 1 d ν2 + − 2 + 1 dz 2 z dz z Gν (z) = 0 (7.93)

where ν = + 1 . The regular solution of this equation is called the regular Bessel function. Its 2 power expansion, using the conventional normalization, is given by [c.f. (7.79)] Jν (z) = z 2

ν ∞ n=0

(iz/2)2n . n! Γ(ν + n + 1)

(7.94)

One can show that J−ν (z), deﬁned through (7.94), is also a solution of (7.93). This follows from the fact that ν appears in (7.93) only in the form ν 2 . In the present case we consider solely the case ν = + 1 . In this case J−ν (z) is linearly independent of Jν (z) since the Wronskian 2 W (J

+1 , 2

J−

−1 ) 2

= (−1)

2 πz

(7.95)

is non-vanishing. One can relate J + 1 and J− − 1 to the regular and irregular spherical Bessel 2 2 functions. Comparision with (7.79) and (7.83) shows j (z) = n (z) = π J 2z (−1)

+ 1 (z) 2

(7.96)

− 1 (z) 2

+1

π J 2z −

.

(7.97)

These relationships are employed in case that since numerical algorithms provide the Bessel functions Jν (z), but not directly the spherical Bessel functions j (z) and n (z). Exercise 7.2.2: Demonstrate that expansion (7.94) is indeed a regular solution of (7.93). Adopt the procedures employed for the function j (z). Exercise 7.2.3: Prove the identity (7.95).

and using the Rodrigues formula for Legendre polynomials (5.e.99) We want to determine the expansion coeﬃcients b . one given o by (7.104) dx eizx P (x) −1 = (−1) 2 ! (iz) = 2 ! +1 dx (x2 − 1) −1 +1 dx (1 − x2 ) eizx . of (7.99) +1 d cos θ eikr cos θ P (cos θ) = b j (kr) −1 2 .178) are given in terms of Legendre polynomials P (cos θ) one can replace the expansion in (7.98) The l. 7. This holds for −1 and one can conclude +1 consecutive integrations by part d izx e dx (7. One can show d −1 2 (x − 1) ∼ (x2 − 1) × polynomial in x dx −1 (7.102) −1 d −1 +1 −1 dx −1 (x2 − 1) .50).h.43) has two solutions.101) 2 ! ∂x −1 −1 Integration by parts yields +1 dx eizx P (x) −1 = 1 2 ! − 1 2 ! eizx d −1 2 (x − 1) dx −1 dx d izx e dx +1 (7. One can expand the former solution in terms of solutions (7. In this case the expansion on the r.s. (7.h. according to (5. hence. 2 +1 (7.103) and.m a mj (kr) Y m (θ.50). (7. φ).179) yield from (7.194 Generating Function of Spherical Bessel Functions Spherically Symmetric Potentials The stationary Schr¨dinger equation of free particles (7. z = kr. the surface term ∼ [· · ·]+1 vanishes. φ) . φ) for non-vanishing m has a non-trivial φ-dependence as described by (5. (7. in case of a free particle moving along the x3 -axis one expands eik3 x3 = .98) by ∞ eik3 x3 = =0 b j (kr) P (cos θ) . −1 .44. namely. For example. The orthogonality properties (5. the wave function does not depend on φ.150) one obtains +1 +1 1 ∂ dx eizx P (x) = dx eizx (x2 − 1) .45) and one given by (7.98) does not involve any non-vanishing m-values since Y m (θ.s.106). i.. can be written exp(ikr cos θ). Since the spherical harmonics Y 0 (θ.100) Deﬁning x = cos θ.

2: Free Particle Comparision with (7.69)] b z 2 (iz) = 1 · 3 · 5 · · · (2 + 1) 2 + 1 2 ! +1 dx (1 − x2 ) . after insertion into (7.112) We want to consider the expression Gν (z) = aν z ν fν (z) where we deﬁne fν (z) = C (7.108) where the last factor is equal to unity.h.105) must hold for all powers of z. Comparision with the l.s. using ν = Jν (z) = √ 1 πΓ(ν + 1 ) 2 z 2 ν + 1.99). (7.105) and (7.114) .117) one can write the r. Integral Representation of Bessel Functions Combining (7.109) results in the integral representation of j (z) j (z) = (z) 2 +1 ! +1 dx (1 − x2 ) eizx .111) Employing (7. −1 (7. as the generating function of the spherical Bessel functions. 1 (7. for the leading power x [c.105) This expression allows one to determine the expansion coeﬃcients b . z i or i (2 + 1) z (2 )! 2 2 2 +1 2 ! [1 · 3 · 5 · · · (2 − 1)] 1 (7.106) yields ﬁnally b = i (2 + 1) or. 2 +1 −1 dx (1 − x2 )ν− 2 eizx .109) eikr cos θ = =0 i (2 + 1) j (kr) P (cos θ) . 1 (7. ∞ (7. The identity (7.106) Employing (5. of (7.96) one can express this.s.113) dt (1 − t2 )ν− 2 eizt . (7.h.h. −1 (7.s.7.110) One refers to the l.107) 1 2 (2 )! 1 · 3 · 5 · · · (2 + 1) 2 + 1 1 · 2 · 3 · 4 · · · (2 − 1) · 2 (7. −1 (7. in particular.100) gives b j (kr) 2 (iz) = 2 +1 2 ! +1 195 dx (1 − x2 ) eizx .f.

115) Integration by part yields fν (z) = − i 2ν + 1 (1 − t2 )ν+ 2 eizt 1 t2 t1 − z 2ν + 1 dt (1 − t2 )ν+ 2 eizt .93).113) satisﬁes (7.113) obeys the Bessel equation (7.93) as long as the integration path in (7. To prove this we note fν (z) = i C dt (1 − t2 )ν− 2 t eizt .2. 4. for properly chosen endpoints t1 .121) (7.117). 1 j = 1. 3. obeys Bessel’s equation (7.122) (7. In fact. t2 .121).118) z 2ν + 1 fν (z) + fν (z) . 1 (7.117) dt (1 − t2 )ν− 2 t2 eizt 1 (7. t2 of the integration paths C.196 Spherically Symmetric Potentials Here C is an integration path in the complex plane with endpoints t1 . Gν (z). it holds for the derivatives of Gν (z) Gν (z) G (z) = = ν aν z ν fν (z) + aν z ν fν (z) z 2ν ν(ν − 1) aν z ν fν (z) + aν z ν fν (z) + aν z ν fν (z) 2 z z (7. C 1 (7. We can now demonstrate that Gν (z) deﬁned in (7. Exercise 7.119) We note that equations (7.93) for arbitrary ν. 7. z (7.123) Insertion of these identities into Bessel’s equation leads to a diﬀerential equation for fν (z) which is identical to (7. 2.114.120) such that we can conclude that Gν (z) for proper integration paths is a solution of (7.124) . (7. (7. We consider now the functions u(j) (z) = √ 1 πΓ(ν + 1 ) 2 z 2 ν Cj dx (1 − x2 )ν− 2 eizx .116) fν (z) = − or fν (z) = − From this we can conclude fν (z) + 2ν + 1 fν (z) + fν (z) = 0 .120) z 2ν + 1 dt (1 − t2 )ν− 2 eizt + C C 1 1 t2 t1 = 0 (7.120) imply also the property (2ν + 1) fν (z) + z fν+1 (z) = 0 .4: Prove (7.116) In case that the endpoints of the integration path C satisfy (1 − t2 )ν+ 2 eizt one can write (7.

2: Free Particle 197 for the four integration paths in the complex plane parametrized as follows through a real path length s C1 (t1 = −1 → t2 = +1) : C2 (t1 = 1 → t2 = 1 + i ∞) : C3 (t1 = 1 + i ∞ → t2 = −1 + i ∞) : C4 (t1 = −1 + i ∞ → t1 = −1) : t = s t = 1 + is t = 1 + is t = −1 + is −1 ≤ s ≤ 1 (7.93).135) are known as the Poisson integrals of the Bessel functions.125) 0 ≤ s < ∞ (7. . one can state Jν (z) = − u(2) (z) + u(4) (z) .128) C = C1 ∪ C2 ∪ C3 ∪ C4 is a closed path. (7. for Rez > 0 and ν ∈ R obey (7.117) and. hence. Comparision with (7. u(2) (z) and u(2) (z) are both solutions of Bessel’s equation (7. 1 (7. therefore.124) is analytical in the part of the complex plane surrounded by the path C we conclude 4 u(j) (z) ≡ 0 j=1 (7.127) 0 ≤ s < ∞ (7. (7. the integral expression (1) Hν (z) 1 1 1 1 (7. one can derive (2) Hν (z) = e −i(z−πν/2−π/4) ∞ 0 √ ds [s(1 − is/2)]ν− 2 e−zs . (7.130) holds Jν (z) = (1) 1 2 1 (1) (2) Hν (z) + Hν (z) .134) and (7. we introduce the so-called Hankel functions (1) Hν (z) = −2 u(2) (z) . Following convention.112) shows u(1) (z) = Jν (z).130) We note that the endpoints of the integration paths C2 and C4 . Since the integrand in (7. (2) Hν (z) = −2 u(4) (z) . 1 (7.7. dt = i ds.129) The integrand in (7. and (1 − t2 )ν− 2 eizt = [(1 − t)(1 + t)]ν− 2 eizt = [−is(2 + is)]ν− 2 eiz e−zs = ei(z−πν/2+π/4) 2ν− 2 [s(1 + is/2)]ν− 2 e−zs . Accordingly.134) Similarly.131) According to (7. u(3) (z) ≡ 0.124) vanishes along the whole path C3 and.132) For Hν (z) one derives.126) −1 ≤ s ≤ 1 (7.133) = e i(z−πν/2−π/4) √ 2 zν π Γ(ν + 1 ) 2 2 zν π Γ(ν + 1 ) 2 √ √ ∞ 0 ds [s(1 + is/2)]ν− 2 e−zs .135) (7. using t = 1 + is.

hence.134) and (7. πz (7.136) (1) 2 (z) = 0 ds [s(s ± is/2)] e−zs . For such solutions g(z). We employ for this purpose the Poisson integrals (7. converge fast for |z| → ∞. Hν (z) as well as H−ν (z) are solutions of this equation. h) must be a non-vanishing function. i..140).2) (z) +1 2 in terms of z −1 such that (z) = e±i[z − ( ∞ +1) π ] 2 2z z ( 1 ) f 2 (z) π ! (7. the Wronskian W (g.140) Bessel Functions with Negative Index Since ν enters the Bessel equation (7.137) (z) = r=0 r ± i 2 r 0 ∞ ds s +r e−zs .141) (1) (1) the derivation of which follows the derivation on page 192 for the Wronskian of the radial Schr¨dinger equation.e.139) (z) = 2 ±i[z − ( e πz +1) π ] 2 r=0 ( + r)! r!( − r)! i ± 2z r (7.93) only as ν 2 .143) . One obtains W (z) = − 4i . For the Wronskian connected with the Bessel equation (7. we obtain H (1) 2 1 +2 (1) 2 (z) = r=0 ( + r)! ! r!( − r)! ± i 2 r 1 z +r+1 (7. z (7.198 Asymptotic Behaviour of Bessel Functions We want to obtain now expansions of the Hankel functions H Spherically Symmetric Potentials the expansions converge fast asymptotically. The general solution of (7.142) and h(z) = H (2) (z) +1 2 one can identify the constant c by using the leading terms in the expansions (7.93) holds the identity 1 W = − W z (7.141) is o W (z) = − In case of g(z) = H (1) (z) +1 2 c . (7. (7. h(z) to be linearly independent.135) which read for ν = + 1 2 H where f The binomial formula yields f (1) 2 (1) 2 1 +2 (1.138) The formula for the Laplace transform of sn leads to f and. As a second order diﬀerential equation the Bessel equation has two linearly independent solutions.

147) this can be written n (z) = 1 2i h (1) (z) − h (2) (z) . (7. We conclude 2 H−( Simarly. 7. 2z 2 (1.153) .97.2) H + 1 (z) . From (7. one can show H−( Spherical Hankel Functions In analogy to equations (7.152) (7. For |z| → ∞ the leading terms yield π π π π π 1 1 1 √ ei(z+ 2 ) = √ A ei(z− 2 − 2 ) + √ B e−i(z− 2 − 2 ) z z z |z| → ∞ .151) Equations (7.96.146) (z) = −i (−1) H (2) 1 (z) +2 .147) (z) = π (1. 7. H (1.151) are equivalent to h h (1) (2) (z) (z) = = j (z) + i n (z) j (z) − i n (z) .97) one deﬁnes the spherical Hankel functions h (1.148) Following the arguments provided above (see page 193) the functions h (z) are solutions of the radial Schr¨dinger equation of free particles (7.149.53).140). 7.146.2: Free Particle i.144) The expansion coeﬃcients A.132.90. in fact. 7.e. 2 (1.7.148) holds for the o regular spherical Bessel function j (z) = 1 (1) (2) [ h (z) + h (z) ] .2) (z) +1 2 199 are. According to (7.149) We want to establish also the relationship between h (z) and the irregular spherical Bessel function n (z) deﬁned in (7.2) (2) +1) 2 (1) +1) 2 (z) = i (−1) H (1) (z) +1 2 . H−ν (z) = A H (1) (1) (z) +1 2 One can then expand + BH (2) (z) +1 2 . B can be obtained from the asymptotic expansion (7. linearly independent. (7.132) follows n (z) = 1 2 π (−1) 2z +1 H−( (1) 1 +2) (z) + H−( (2) +1) 2 (z) . 7. (7. (7.97). (7. (7.150) According to (7.145) This equation can hold only for B = 0 and A = exp[i( + 1 )π].2) (7.2) (7. 7. (7..

at large |z|.140. = −cos(z − π/2) results in 2 2 the alternative expressions j (z) n (z) = = sin(z − π/2) cos(z − π/2) p (z) + q (z) z 2z 2 cos(z − π/2) sin(z − π/2) q (z) − p (z) .157) i q (z) 2z = i Im r=0 i 2z i 2z r = [ −1/2] ( +2r+1)! r=0 (2r+1)!( −2r−1)! −1 r 4z 2 0 ≥ 1 = 0 (7. are complex conjugates. From (7.148) one obtains readily h (1) 2 (1.200 Asymptotic Behaviour of Spherical Bessel Functions Spherically Symmetric Potentials We want to derive now the asymptotic behaviour of the spherical Bessel functions h and n (z).148).161) (7. 2z 2 z sin(z − π/2) z cos(z − π/2) − . at large |z|.158) one can derive then from (7. 7.154) The leading term in this expansion. 2 2z z (7. j (z) (z) = ( i) z +1 e±iz r=0 ( + r)! r!( − r)! ± i 2z r (7. (1) (7.154) the identities j (z) n (z) = = sin[z − ( + 1) π ] cos[z − ( + 1) π ] 2 2 p (z) − q (z) z 2z 2 cos[z − ( + 1) π ] cos(z − π/2) 2 q (z) + p (z) .159) (7. z (7.160) Employing cos[z − ( + 1) π ] .156) ( + r)! r!( − r)! ( + r)! r!( − r)! i 2z r [ /2] = r=0 ( + 2r)! (2r)!( − 2r)! −1 4z 2 r (7.164) . ne ll(z) = Im[h (1) (z)] . = sin(z − π/2) and sin[z − ( + 1) π ] . are j (z) n (z) = = (7. Hence.2) (z).140) and (7. (7. as given by (7.155) (z) and To determine j (z) and n (z) we note that for z ∈ R the spherical Hankel functions h (2) h (z).162) The leading terms in these expansions. is h (1) 2 (z) = ( i) z +1 e±iz . it follows z ∈ R : Using p (z) = Re r=0 j (z) = Re[h (1) (z)] .163) (7.

161.171.176) .2) relationships hold for g (z) = h (z).175) Using Γ( + 1) = !.114) we can write g (z) = a z f 2 ! + 1 (z) 2 .149.169) (7.173) A (z) = z ( +1) z2 +2 z (7.171) −1 (z) (7. (1. 7.151) follows then that the relatiosnhips hold also for g (z) = j (z) and g (z) = n (z). (7.124.174) and from (7. 7.166) (7. From the linearity of the relationships (7. j (z) and n (z). 7. 7. 1.2: Free Particle Expressions for the Spherical bessel Functions j (z) and n (z) 201 The identities (7.168) (7. deﬁning a = −1 and employing fν (z) as deﬁned in (7.131. One can combine (7.167) (7.171–7. 1 (7.172) to +1 (z) +1 (z) = A (z) 1− g (z) g (z) −1 (7.170) Recursion Formulas of Spherical Bessel Functions The spherical Bessel functions obey the recursion relationships g g +1 (z) +1 (z) = = (1. For this purpose we need to demonstrate only that the (1.174) We want to prove these relationships.172) where g (z) is either of the functions h obtain the recursion relationship g g (z). One obtains for = 0.165) (7.2) g (z) − g (z) z 2 +1 g (z) − g z (7. 7.148) and express g (z) = h (1.171) holds for g (z) = h (z) we employ (7. 2 j0 (z) n0 (z) j1 (z) n1 (z) j2 (z) n2 (z) = = = = = = sin z z cos z − z sin z cos z − z2 z cos z sin z − 2 − z z 3 1 3 − sin z − 2 cos z 3 z z z 3 1 3 − 3 + cos z − 2 sin z z z z (7.162) allow one to provide explicit expressions for j (z) and n (z).2) To demonstrate that (7.4 dx (1 − x2 )ν− 2 eizx .7.2) (z) = −2 π 1 √ 2z πΓ(ν + 1 ) 2 z 2 ν C2.

In order to prove (7. is equivalent to (7. n0 (z) using (7.171).58) it holds o 2 g (z) = − g (z) + z ( + 1) g (z) − g (z) . j0 (z) using (7. z2 (7.173). 7. j2 (z) from j0 (z). Exercise 7.2.121).6: Employ the recursion relationship (7.180) Since g (z) is a solution of the radial Schr¨dinger equation (7.179) g (z) = − z2 g (z) + g (z) − g +1 (z) z .173. The second component of (7.2. Exercise 7. i.178) (7. (7.165).171).173.172). . To prove (7. f yields. in fact. (7.e.182) Replacing all ﬁrst derivatives employing (7.179) leads to (7.177) (z) = − z f 2 +2 + 3 (z) 2 . (7.172). n2 (z) from n0 (z).171).173) is equivalent to (7.5: Provide a detailed derivation of (7.182).174) we start from (7. 7.202 The derivative of this expression is g (z) = Employing (7. and (b) n1 (z). z (7.176) g (z) = z g (z) − g +1 (z) +1 2 Spherically Symmetric Potentials z g (z) + a z f 2 ! +1 2 (z) .179) from which follows (7..181) Using this identity to replace the second derivative in (7.180) yields g +1 (z) = g (z) − ( + 2) g (z) + z2 +2 g (z) . together with (7. The ﬁrst component of (7.174) to determine (a) j1 (z).172) we diﬀerentiate (7.166).

. We assume that the system considered is in vacuum in which charge and current sources described by the densities ρ(r.29). However. for the existence of discrete photons. then state the Hamiltonian which describes the resulting interaction. It turns out that the proper description of the electromagnetic ﬁeld requires a little bit of eﬀort.g.27– 1. These sources enter the two inhomogeneous 203 .g.1 Description of the Classical Electromagnetic Field / Separation of Longitudinal and Transverse Components The aim of the following derivation is to provide a description of the electromagnetic ﬁeld which is most suitable for deriving later a perturbation expansion which yields the eﬀect of electromagnetic radiation on a bound charged particle. t) are present. This is achieved in the following derivation. and the external electromagnetic ﬁeld are of electromagnetic origin and. an electron in a hydrogen atom. the attractive Coulomb force between proton and electron in case of the hydrogen atom. e. on an electron in a hydrogen atom.Chapter 8 Interaction of Charged Particles with Electromagnetic Radiation In this Section we want to describe how a quantum mechanical particle. We will later introduce a simple rule which allows one to account to some limited degree for the quantum nature of the electromagnetic ﬁeld.e. We will describe the electromagnetic ﬁeld classically.. i. For this purpose we need to establish a suitable description of this ﬁeld. is aﬀected by electromagnetic ﬁelds. t) and J(r.e. The classical electromagnetic ﬁeld is governed by the Maxwell equations stated already in (1.. or other charged particles. and are aﬀected by the external electromagnetic ﬁeld. Such description should be suﬃcient for high quantum numbers. i.g.. 8. hence. The problem is that the latter electron. e. both the Coulomb interaction due to charges contributing to binding the particle. e.. for situations in which the photons absorbed or emitted by the quantum system do not alter the energy content of the ﬁeld. are aﬀected by the Coulomb interaction V (r) which is part of the forces which produce the bound state. must be described consistently.

t) is a scalar function. t) −→ V (r. t) · B(r. t) + ∂t B(r.7) Gauge Transformations We have expressed now the electric and magnetic ﬁelds E(r. accordingly. 2nd Edition. but leave the ﬁelds unaltered: A(r. The particle.9) (8.11) V (r. t) . D.8) = 0 (8. called the scalar potential. 1975) for a discussion of these and other conventional units. t) = − V (r.204 Maxwell equations1 · E(r. t) (8. t) × B(r. contributes to the charge density ρ(r. the relationship between ﬁelds and potentials is not unique. t) + χ(r. t) = − V (r. The reader is referred to the well-known textbook ”Classical Electrodynamics”.1) the term qδ(r − ro (t)) and to the current density ˙ J(r. In the non-relativistic limit holds p ≈ mr and (8.1) (8.2) the term q r˙o δ(r − ro (t)). From this follows E(r. obeys the equation of motion d p = q dt E[ro (t). t) Interaction of Radiation with Matter = 4 π ρ(r. alter the potentials. by J. t) − ∂t E(r. Scalar and Vector Potential Setting B(r. called gauge transformations. t] (8. t) where V (r. t) + ∂t A(r. called the vector potential. t) −→ 1 A(r. t) = 0 = 0. New York. Jackson (John Wiley & Sons. t) = 4 π J(r. t) . t) and B(r.3) (8. As is well known. t) (8. t) + ∂t A(r. t). . t) which is solved by E(r. t) through the scalar and vector potentials V (r. The following substitutions. t) = × A(r. (8.6) for some vector-valued function A(r.4) Lorentz Force A classical particle with charge q moving in the electromagnetic ﬁeld experiences the so-called Lorentz force q[E(r.5) where p is the momentum of the particle and ro (t) it’s position at time t. t] + v × B[ro (t). in turn.10) (8. t) − ∂t A(r. Equation (8. t) in (8. t) − ∂t χ(r.2) In addition. t) in (8.3) reads then × E(r. t)] and.4).5) above agrees with the equation of motion as given in (1.25). t) + v × B(r. t) . (8. solves implicitly (8. t). t) and A(r. the two homogeneous Maxwell equations hold × E(r. (8. We assume so-called Gaussian units.

t). For this purpose we examine the oﬀending term ∂t V (r. t) + ∂t V (r. Using the expressions (8. t) can be obtained employing (8. t) = − 4 π J(r. 4π (8. (8. t) = ∂t = 4 π ρ(r. and a component due to the remaining currents.19) couples the vector potential A(r. be achieved. t) |r − r | (8. t) − ∂t A(r. a name which is due to the form of the resulting scalar potential V (r. t = 4 π J(r. t) Using · ∂t A(r. (8.1: Electromagnetic Field 205 This gauge freedom will be exploited now to introduce potentials which are most suitable for the purpose of separating the electromagnetic ﬁeld into a component arising from the Coulomb potential connected with the charge distribution ρ(r. t) = 0 . such that the latter describes the electromagnetic radiation.19) and deﬁne J (r. t) − ∂t A(r. t) and the current due to moving net charges.19) Unfortunately. t) (8. t) . t) = − 4 π ρ(r. t) = Ω∞ for r ∈ ∂Ω∞ (8. t) + ∂t A(r.17) A(r. (8.18) together with condition (8. t) together with (8. the gauge freedom allows us to impose on the vector potential A(r. t) = 1 ∂t V (r. One would prefer a description in which the Coulomb potential (8. t) .12) yields then the Poisson equation 2 V (r. t) and V (r. t) − ∂t V (r. t) . in fact.9) for the ﬁelds results in × The identity × × A(r. the second inhomogeneous Maxwell equation. (8. t) = 0 the solution is given by the Coulomb integral V (r. The vector potential A(r.9) into (8.2). Such description can. t) the condition · A(r.6) and (8.14) In case of the boundary condition V (r.12) The corresponding gauge is referred to as the Coulomb gauge.12) leads us to 2 2 A(r. this potential results from inserting (8. t) − 2 × A(r. t) = · A(r. (8. t) .16) This is the potential commonly employed in quantum mechanical calculations for the description of Coulomb interactions between charged particles. In fact.1) · − V (r. t) in (8.16) and the vector potential are uncoupled. t).13) · A(r. t) .15) d3 r ρ(r . and the former the Coulomb interactions in the unperturbed bound particle system.20) . equation (8.8. In fact.

t) = 0 (8. t) contributes solely an electric ﬁeld component E (r.2 Planar Electromagnetic Waves The current density Jt describes ring-type currents in the space under consideration. V (r. 8.22) 1 ∂t 4π · J (r. t) . t) − ∂t A(r. t).26) turns into the well-known wave equation 2 2 A(r. no divergence-free currents. V (r. using ∂t · J (r. t) + ∂t ρ(r. t) + ∂t ρ(r.21) and (8.. t) holds. t) = 0). t) = − V (r. t) = 0 and from this follows · Jt (r. t) = 0).26) and (8. in a ring-shaped antenna which exhibits no net charge.g. the name longitudinal electric ﬁeld. The deﬁnitions of J and Jt applied to (8. t) = − 4 π Jt (r. Presently.25) one refers to J and Jt as the longitudinal and the transverse currents.14). respectively.16) yield ﬁnally the electric and magnetic ﬁelds.12) and the Coulomb potential (8. t) as the current due to the time-dependence of the charge distribution ρ(r. (8. t) = or = 2 Interaction of Radiation with Matter (8. t) be the total current of the system under investigation and let Jt = J − J . t) = 0 . hence. yet a current. t) Bt (r. In this case (8. Let J(r.21) ∂t and the Poisson equation (8. for example. t) = × A(r. = − ∂t A(r. t) (8. A(r. t) These ﬁelds are obviously divergence -free (e. hence. exist in the space considered. t) . i. t) = 0 . the name transverse ﬁelds.206 For the curl of J holds × J (r. t) (8.19) yield 2 2 A(r.29) · Et (r.e. (8. t) contributes an electrical ﬁeld component as well as the total magnetic ﬁeld Et (r.28) (8. t) − ∂t A(r.27) which is obviously curl-free ( × E (r. The vector potential determined through (8.23) This continuity equation identiﬁes J (r.25) Because of properties (8. t) = − ∂t ρ(r. t) = 0 . (8.24) (8. such current densities exist. For the divergence of J (r. we want to assume that no ring-type currents..26) This equation does not couple anymore scalar and vector potentials. For J also holds the continuity equation · J(r.30) . (8.

39) 8π The deﬁnition incoming waves and outgoing waves is rationalized below in the discussion following Eq.31) and the resulting ﬁelds (8. t) = ±i ω A(r. (8. ˆ (8. t) = ± ω2 ˆ |Ao |2 k . Time average over one period 2π/ω yields S(r. the constant k is referred to as the wave vector. Here the “-” sign corresponds to so-called incoming waves and the “+” sign to outgoing waves2 .31.31).38) ˆ In this expression for the energy ﬂux one can interprete k as the propagation velocity (note c = 1) and. The Coulomb gauge condition (8.158).37) ˆ ˆ where k is the unit vector k = k/|k|. This ﬂux is given by 1 S(r.34) (8. A complete set of solutions is given by the so-called plane waves A(r.35) one obtains S(r. accordingly. obviously.8. 2 . t) and Bt (r. are orthogonal to each other and are both orthogonal to the wave vector k. see also the comment below Eqs. there exist two linearly ˆ u independent orientations for u corresponding to two independent planes of polarization.38. The corresponding electric and magnetic ﬁelds. The latter vector describes the direction of propagation of the energy ﬂux connected with the plane wave electromagnetic radiation. Et (r. (8. Note that in the units chosen the velocity of light is c = 1.2: Planar Electromagnetic Waves 207 which describes electromagnetic ﬁelds in vacuum. t) . t) = ± ω2 ˆ |Ao |2 k sin2 ( k · r − ωt ) 4π (8.36) 4π Using the identity a × (b × c) = b (a · c) − c (a · b) and (8.35) The vector potential in (8.32. 8.39).28.34. 8π (8. 8. ˆ We want to characterize now the radiation ﬁeld connected with the plane wave solutions (8.12) yields u·k = 0. at each point r and moment t. ω2 = |Ao |2 (8. t) × Re B(r. (8.34. (8. 8. In applying the potential and ﬁelds to physical observables and processes we will only employ the real parts.35). 8.34. t) . according to (8. t) in (8. 8.32) holds. t) = i k × A(r. is orthogonal to k.29). 8. t) = Re Et (r.31) u is a unit vector (|ˆ| = 1) which. hence. Obviously. t) = Ao u exp i(k · r ˆ where the dispersion relationship |k| = ω (8. t) Bt (r. are Et (r.35) are complex-valued quantities. 8.33) ωt) (8.

t) 2m + 1 8π Ω∞ 2 H = + qV (r) 1 16π d3 r |Et |2 + |Bt |2 Ω∞ d3 r E 2 + . t) = 8πNω u exp i(k · r − ωt) ˆ ωV . 8.42) 8. A ‘poor man’s’ quantization of the ﬁeld is possible at this point by expressing the energy density (8. These photons each carry the energy ω. A correct description of the electromagnetic ﬁeld requires that the ﬁeld be quantized. Bt (r. whereas in the case of outgoing waves the energy is transported in the direction of k. the energy of the plane wave is transported in the direction of −k. deﬁned below Eqs. (8.44) . (8. t) 2m 2 + qV (r) .39) and (8.31).31) allows one ﬁnally to state for the planar wave vector potential A(r.28.29) together with the potentials (8.38) implies that for incoming waves. t) ˆ replacing the classical momentum p by the diﬀerential operator p = of the particle is then described by the Schr¨dinger equation o ˆ i ∂t Ψ(r.16.3 Hamilton Operator The classical Hamiltonian for a particle of charge q in a scalar and vector potential V (r) and A(r. i (8.8.43).32). (8.208 Interaction of Radiation with Matter as the energy density. t) is real 1 and that Et (r.31).39) through the density of photons connected with the planar waves (8. u · k = 0 . |k| = ω . The integrals express the integration over the energy density of the ﬁelds. t) . Note that E (r. (8. 8. t). If we consider a volume V with a number of photons Nω the energy density is obviously Nω ω . a ˆ speciﬁc k and a speciﬁc u.40) = V It should be pointed out that Nω represents the number of photons for a speciﬁc frequency ω.27. The sign in (8. t) = H Ψ(r. 8.31.41) ωV Inserting this into (8. hence. t) are complex leading to the diﬀerence of a factor 2 in the energy densities of the lontitudinal and transverse components of the ﬁelds. t). The wave function Ψ(r. Comparision of (8.45) . (8. (8.43) Here the ﬁelds are deﬁned through Eqs. is p − q A(r. respectively.40) allows one to express then the ﬁeld ˆ amplitudes 8πNω Ao = . We are left with a classical Hamiltonian function which has an obvious quantum mechanical analogue ˆ H = ˆ p − q A(r. We assume that the energy content of the ﬁelds is not altered signiﬁcantly in the processes described and. we will neglect the respective terms in the Hamiltonian (8. ˆ (8.

t)/ . (8. in the Schr¨dinger equation o 2 ˆ p − qA i ∂t Ψ(r. t) inside the double brackets.10. does not change expectation values which contain the probability densities3 . hence.50) The equivalence of (8. hence.46) the ﬁelds enter. 8.t)/ Ψ(r.8.t)/ Ψ(r. (8.11) of the potentials is equivalent to multiplying the wave function Ψ(r. This leads to the question in how far the gauge transformations (8. It should be noted.t)/ p + q(( χ)) Ψ(r. by the group 3 The eﬀect on other expectation values is not discussed here. such phase factor does not change the probability density |Ψ(r.t)/ [ i ∂t − q((∂t χ)) ] Ψ(r.48) 2m For this purpose one notes i ∂t eiqχ(r.. that this principle applies only to fundamental interactions.44. an element of the group U(1). t) and V (r. 8. i. t) and A(r.47.t)/ constitutes a unitary transformation of a scalar quantity. t) = + qV Ψ(r. not to phenomenological interactions like the molecular van der Waals interaction. t) enter whereas in the classical equations of motion ˙ m¨ = q E(r. 8.11) to (8.47) is equivalent to 2 ˆ p − qA i ∂t eiqχ(r. The idea just stated can be generalized by noting that multiplication by a phase factor eiqχ(r. t) by a local and time-dependent phase factor eiqχ(r. t) = + qV eiqχ(r. An important conceptual step of modern physics has been to turn the derivation given around and to state that introduction of a local phase factor eiqχ(r. have no eﬀect on the motion of the particle described by (8. (8. t) = eiqχ(r.47) i ∂t Ψ(r. e.11) aﬀect the quantum mechanical description. Elementary constituents of matter which are governed by other symmetry groups.t)/ should not aﬀect a system and that..45) leads to the Schr¨dinger equation o 2 ˆ p − q A − q(( χ)) + qV − q((∂t χ)) Ψ(r. t) are necessary to compensate terms which arise through the phase factor.51) 2m the potentials A(r. 8. t) ˆ p eiqχ(r.46). t)|2 and. t) = 2m where ((· · ·)) denotes derivatives in (( χ)) and ((∂t χ)) which are conﬁned to the function χ(r. t) ˆ = eiqχ(r.10.e.48) implies that the gauge transformation (8. accordingly. . Obviously.10. One can show that (8. t) (8.49) (8. t) .t)/ Ψ(r. however. t) . t) r (8. 8.t)/ Ψ(r.3: Hamilton Operator 209 Gauge Transformations It is interesting to note that in the quantum mechanical description of a charged particle the potentials V (r. t) . In the classical case such question is mute since the gauge transformations do not alter the ﬁelds and. t) + q r × B(r.g. Applying the gauge transformations (8.

x2 . The vector potential (8. (8. (8. can be described by various vector potentials. the so-called Landau gauge ˆ associates the vector potential AL (r) = Bo x1 e2 ˆ (8. In this o case holds V (r. therefore. Using Cartesian coordinates this yields 2 H = − 2me 2 2 2 ∂1 + ∂2 + ∂3 + 2 eBo e2 Bo 2 x1 ∂2 + x i me 2me 1 (8.t) where σ is the vector of Pauli matrices.210 Interaction of Radiation with Matter SU(2). thereby.45) with Hamiltonian (8.44) can be expanded ˆ2 p q H = − 2m 2m For any function f (r) holds ˆ ˆ p · A − A · p f (r) = i A· f + f ·A − A· f = i f ·A.59) .. aﬀects the complexity of the mathematical derivation of the wave functions. j = 1. x3 ) = exp(ik2 x2 + ik3 x3 ) φE (x1 ) . x1 . H = 2m m 2m (8.52) This expression vanishes in the present case since since ˆ ˆ p · Af = A · pf and. k3 . consequently. 3.g. t) = Bo .55). (8. one can employ the Hamiltonian (8. The Hamiltonian (8. The stationary homogeneous magnetic ﬁeld B(r.54) ˆ2 p q ˆ q2 2 − p·A + A + qV . for Bo = Bo e3 in (8. We want to describe the stationary states corresponding to the Hamiltonian (8. holds (8.53) ˆ ˆ p·A + A·p + q2 2 A + qV 2m (8. The choice of a vector potential aﬀects the form of the wave functions describing the eigenstates and. the generators of SU(2).12)].58). Accordingly.57) with a homogeneous magnetic ﬁeld Bo .58) where ∂j = (∂/∂xj ). e.56). For this purpose we use the wave function in the form Ψ(E. Solution for Landau Gauge A particularly convenient form for the Hamiltonian results for a choice of a so-called Landau gauge for the vector potential A(r. t) ≡ 0. 2. likewise can demand the existence of ﬁelds which compensate local transformations described by eiσ·χ(r.55). k2 . The resulting ﬁelds are called Yang-Mills ﬁelds. (8. · A = 0 [cf.55) 8. t).57) satisﬁes · A = 0 and. In case of a homogeneous potential pointing in the x3 -direction.4 Electron in a Stationary Homogeneous Magnetic Field We consider now the motion of an electron with charge q = −e and mass m = me in a homogeneous magnetic ﬁeld as described by the Schr¨dinger equation (8.56) due to the gauge freedom.

57). k3 . (8.61) leads to − where 2 2me 2 ∂1 + 2 k2 1 3 me ω 2 ( x1 + x1o )2 + 2 2me φE (x1 ) = E φE (x1 ) .59). (8.62) x1o = and where ω = k2 eBo eBo me (8. Unfortunately. From this observation one can immediately conclude that the wave shifted (by e 3 function of the system. The energies corresponding to these states are E(n. 2 2me (8. We want to employ . x3 ) = exp(ik2 x2 + ik3 x3 ) × √ 1 2n n! me ω π 1 4 1 exp − me ω(x2 +x1o ) 2 Hn mω (x1 + x1o ) (8. It is important to note that the completion of the square absorbs the kinetic energy term of the motion in the x2 -direction described by the factor exp(ik2 x2 ) of wave function (8. However.56) through the vector potential A(r) = 1 Bo × r .65) which localize the electrons. The stationary Schr¨dinger equation (8. according to (8.62) is that of a displaced (by x1o ) harmonic oscillator with o 2 k 2 /2m ) energies.4: Electron in Homogenous Magnetic Field This results in a stationary Schr¨dinger equation o 2 211 − 2me 2 ∂1 + 2 k2 2 2me + 2 k2 3 2me + 2 eBo k2 e2 Bo 2 x1 + x φE (x1 ) me 2me 1 = E φE (x1 ) . the familiar harmonic oscillator quantum number.67) .65) where we replaced the parameter E by the integer n. the so-called symmetric gauge which expresses the homogeneous potential (8.and x2 -coordinates. is not symmetric in the x1 . like the corresponding gauge (8.59).63) (8. the states are degenerate in the quantum number k2 describing displacement along the x2 coordinate.63) this induces a spread of the wave function in the x1 direction. k3 ) = ω(n + 2 k2 1 3 ) + . therefore. according to (8.66) Obviously. k2 . k2 .65). x1 . 2 (8. the wave function (8. x2 .60) Completing the square 2 2 e2 Bo 2 eBo k2 e2 Bo x1 + x1 = 2me me 2me x + k2 eBo 2 − 2 k2 2 2me (8.64) is the classical Larmor frequency (c = 1).8. Without aﬀecting the energy one can form wave packets in terms of the solutions (8. Solution for Symmetric Gauge The solution obtained above has the advantage that the derivation is comparatively simple. is Ψ(n.

70) ˆ ˆ p · A f = − Bo · p × r f .71) (8. y-plane with constant velocity v on a ring of radius r.74) We consider a simple case to relate (8. In this case the current density measures −e v oriented tangentially to the ring. The latter can be rewritten. (8.52) becomes ˆ2 e e2 p + Bo · L + H = 2me 2me 8me Of particular interest is the contribution Vmag = e L · Bo 2me Bo × r 2 . 8.78) .73) where µclass is the magnetic moment connected with a current density j µclass = 1 2 r × j(r) dr (8.212 Interaction of Radiation with Matter One can readily verify that this vector potential satisﬁes the condition (8. an electron moving in the x.72) allows one to interpret µ = − e L 2me (8.67) one can write ˆ p·A = Using · (u × v) = − u · function f (r) ×v + v· 2i · Bo × r . namely. Vmag = e 2me class = r me v e3 of the electron ˆ (8.68) × u yields. (8. For the vector potential (8. namely.75) 2 The latter can be related to the angular momentum µclass = − and.12) for the Coulomb gauge. the Hamiltonian (8. the magnetic moment (8. for any (8. using × (uf ) = −u × f +f ˆ ˆ Bo · p × r f = Bo · r × p f . accordingly. (8.72) to Hamiltonian (8. Vmag = −µclass · Bo (8. ˆ Identifying r × p with the angular momentum operator L.71). Accordingly.74).76) e 2me class class · Bo .69) × u and × r = 0.73. The theory of classical electromagnetism predicts an analogue energy contribution .72) and (8. ˆ (8. (8.74) is in the present case 1 µclass = − e r v e3 . in the present case of constant Bo .77) Comparision with (8.

For a magnetic ﬁeld (8. likewise. e gives rise to an energy contribution (8. x2 ) .88) me ω xj . The functions ψ(x1 ..86) 1 2 2me 1 2 ˆ as well as of the angular momentum operator L3 .82) 1 ω (x1 ∂2 − x2 ∂1 ) i 2me + 1 me ω 2 x2 + x2 1 2 2 2 k2 3 + (8. Accordingly.83) describes two identical oscillators along the x1 -and x2 -directions which ˆ are coupled through the angular momentum operator L3 . x2 ) = E ψ(x1 .79) To obtain the stationary states. x2 ) obey then ˆ Ho ψ(x1 . 10 that the spin of the electron. A derivation of his property and the value of g. the solutions of ˆ H ΨE (x1 . for Bo = Bo e3 . To obtain these eigenstates we introduce the customary dimensionless variables of the harmonic oscillator Xj = (8. x2 .67) yields a more symmetric solution which decays to zero along both the ±x1 .83) (8. we separate the variable x1 .and the ±x2 -direction.e.e. x3 ) = exp(ik3 x3 ) ψ(x1 .84) . 2me (8. x2 ) where ˆ Ho = − 2 2 2 ∂1 + ∂2 (8.71) is ˆ H = 2 ˆ2 p e2 Bo + 2me 8me x2 + x2 + 1 2 eBo L3 . 10. x2 . the so-called gyromagnetic ratio of the electron. x3 ) = E ΨE (x1 . In this case.4: Electron in Homogenous Magnetic Field 213 as the quantum mechanical magnetic moment operator for the electron (charge −e). 2 . We will demonstrate in Sect. x2 . E = E − (8. we seek stationary states which are simultaneous eigenstates of the Hamiltonian of the two-dimensional isotropic harmonic oscillator 2 1 2 ˆ Hosc = − ∂ 2 + ∂2 + me ω 2 x2 + x2 (8. ˆ i. (8. the Hamiltonian (8.87) .56) pointing in the x3 -direction the symmetric gauge (8. i.81) (8. requires a Lorentz-invariant quantum mechanical description as provided in Sect.8.85) The Hamiltonian (8. x2 from x3 setting ΨE (x1 . j = 1.72) with an associated magnetic moment − g 2me S where g ≈ 2. x3 ) .80) (8. (8. 2me We have used here the expression for the angular momentum operator ˆ L3 = ( /i)(x1 ∂2 − x2 ∂1 ) .83) can then be expressed 1 ˆ 1 Ho = − ω 2 ∂2 ∂2 + 2 2 ∂X1 ∂X2 + 1 2 2 X1 + X2 2 + 1 i X1 ∂ ∂ − X2 ∂X2 ∂X1 . described by the operator S.

x2 ) .as well as for the x2 -oscillator. x2 ) . one obtains 1 ˆ 1 H = a† a1 + a† a2 + 1 + 1 1 2 ω i 1 i a† a2 − a† a1 1 2 Xj − ∂ ∂Xj 1 . 2 2 . (8. x2 ) = 2m Ψ (j. x1 . . 0. therefore.214 Employing the creation and annihilation operators 1 a† = √ j 2 and the identity ˆ ω L3 = which can readily be proven. m . x2 ) = a† 1 j−m (j + m)! (j − m)! Ψ(0. x1 .92) where Ψ(0. it holds a† a1 + a† a2 + 1 1 1 2 Ψ (j. x2 ) is the wave function for the state with zero vibrational quanta for the x1 . . (8. (8. 2 (8. m . the total vibrational energy being ω(2j + 1). Such eigenstates can be expressed.95) (j) ˆ We want to choose the coeﬃcients αmm such that (8. x2 ) . m.92) are not eigenstates of L3 . In order to cover all posible vibrational quantum numbers one needs to choose j. attempt to express eigenstates in terms of vibrational wave functions a† 1 j+m Ψ(j. m. x1 .96) ˆ holds. x2 ) = m=−j αmm Ψ(j. x1 . .. (8.90) a† a2 − a† a1 1 2 .94) is an eigenstate of Ho ˆ Ho Ψ (j. m . If this property is. We. . x1 .e.92) represents a state with j + m quanta in the x1 -oscillator and j − m quanta in the x2 -oscillator. x2 ) = ( 2j + 1 ) Ψ (j. (j) (8.94) is an eigenstate of the vibrational Hamiltonian. obeyed. m . x1 .97) . however.e. . 1 2 since one phonon is annihilated and one created. (8. x1 . such that 1 i a† a2 − a† a1 1 2 Ψ (j. +j . 1. (8. in fact. x1 .93) ˆ The states (8. x2 ) (8.. . (8. x1 . through a combination of states j Ψ (j. . m .94) is also an eigenstate of L3 . x1 . m . 0. 1 3 . aj = √ 2 Interaction of Radiation with Matter Xj + ∂ ∂Xj . m as follows: j = 0. m . j = 1. −j + 1. i. x1 . m = −j. . i. x2 ) (8.91) We note that the operator a† a2 − a† a1 leaves the total number of vibrational quanta invariant.94) Since this state is a linear combination of states which all have vibrational energy (2j + 1) ω.89) . (8. x2 ) = ω 2j + 2m + 1 Ψ (j.

5. an example is the Coulomb potential V (r. t) introduced above. 5.98) then the states Ψ(j. (8.5: Time-Dependent Perturbation Theory (j) 215 In order to obtain coeﬃcients αmm we can proﬁtably employ the construction of angular momentum states in terms of spin– 1 states as presented in Sects. Other radiation ﬁelds can .288)] 1 ˆ J3 = 2 a† a1 − a† a2 1 2 (8.102) We have identiﬁed.92) correspond to the eigenstates |Ψ(j.84) above. Using the explicit form of the Wigner rotation matrix as given in (5. b† .31). the eigenstates of (8.5. The required rotation must transform the x3 –axis into the x2 –axis. x1 . 5. x2 ) (8. The ﬁrst rotation contributes a factor exp(−im π ). m .55) conﬁnes the particle to stationary bound states. the second rotation a factor dmm ( π ).(5. x2 ) .5 Time-Dependent Perturbation Theory We want to consider now a quantum system involving a charged particle in a bound state perturbed by an external radiation ﬁeld described through the Hamiltonian (8. We assume that the scalar potential V in (8. thus. 0) is a rotation matrix which describes the rotation around the x3 –axis by π 2 2 2 and then around the new x2 –axis by π . a† . If we identify 2 a† . m. π . x1 . m. i. According to the derivation given there.10.e. a2 1 2 present notation ←→ b† .9. the 2 2 latter representing the Wigner rotation matrix of Sect.11 such transformation is provided through π π (j) Ψ (j. 5.8..11.100) J2 = 1 2 2i 1 ˆ i. x1 .83) and conﬁrmed the eigenvalues stated in (8. a1 . m . m.9.10. x1 .5. x1 . 5. x2 ) = j+m m+m −t 1 2j 2 j m=−j j−m t=0 (j+m)!(j−m)! (j+m )!(j−m )! (j) (j) j−m t (−1)j−m −t (−i)m Ψ(j.5. a transformation moving the x3 –axis into the x2 – 2 axis.. is identical to the operator L3 introduced in (8. the states are eigenstates of the operator [we use for the operator the notation of Sect. 5. 0) Ψ(j. x1 . is in the notation of the present section 1 ˆ a† a2 − a† a1 .10. m.11 (8. Eq.10. which corresponds there to the angular momentum in the x2 –direction. 5. b− + − notation in Sects. According to Sect. The connection with the present problem arises due to the fact that the operator J2 in Sect.55). This implies that ˆ we can obtain eigenstates of L3 by rotation of the states Ψ(j.97).11. cf.101) 2 2 where Dmm ( π . . (8. x2 ). 5. except for a factor 2 .9.309) yields ﬁnally Ψ (j.e. m) in Sect. In the simplest case the radiation ﬁeld consists of a single planar electromagnetic wave described through the potential (8. x2 ) deﬁned in (8. 8.99) with eigenvalue m. b+ . t) = 1/4πr conﬁning an electron with energy E < 0 to move in the well known orbitals of the hydrogen atom. x2 ) = Dmm ( . The external radiation ﬁeld is accounted for by the vector potential A(r.

i. t). using (8.105) for the case of radiationinduced transitions in atomic systems.104) (8.109) The reader should note that the estimates are very crude since we are establishing an order of magnitude estimate only. one can estimate that the perturbation. orthonormal basis. the term ∼ p · A(r.106) where we adopted the Dirac notation for the states of the quantum system.. is much smaller than the eigenvalue diﬀerences near ˆ typical atomic bound states. We will see below that the perturbation resulting from a ‘pure’ plane wave radiation ﬁeld will serve us to describe also the perturbation resulting from a radiation ﬁeld made up of a superposition of many planar waves.e. The Hamiltonian of the particle in the radiation ﬁeld is then described through the Hamiltonian H Ho VS = = = Ho + VS ˆ2 p + qV 2m q ˆ q2 2 p · A(r.e.216 Interaction of Radiation with Matter be expanded through Fourier analysis in terms of such plane waves. this assumption can be waved as our results below will not depend on it.. i. is much 2 (r. can be considered a weak perturbation.e.e.41) e ˆ e p2 p·A ∼ 2me me me 2me 4 1 2 8πNω . as given in (8. However. This result will allow us to neglect the larger than the second term. t).105). t) − m 2m (8.. and that the ﬁrst term in (8. i. Estimate of the Magnitude of VS We want to demonstrate now that the interaction VS (t). ωV (8.. .103) (8. (8. . i.105) as follows4 . t) is given by (8. the term ∼ A second term in (8. t) + A (r.105) in further calculations and to expand the wave function in terms of powers of VS (t) in a perturbation calculation. In fact. The states |n are thought to form a complete. we discount the possibility of a continuum of eigenstates. n = 0. we assume n|m = δnm and for the identity ∞ (8.107) 1 = 1 n=0 |n n| . For an electron charge q = −e and an electron mass m = me one can provide the estimate for the ﬁrst term of (8. Here the so-called unperturbed system is governed by the Hamiltonian Ho with stationary states deﬁned through the eigenvalue problem Ho |n = n |n . 1. in this case.108) We assume for the sake of simplicity that the eigenstates of Ho can be labeled through integers.105) where A(r. . We ﬁrst note. (8. 2 .42).

e.42) it holds VS ≈ e m k. the ﬁrst term in (8. replace the perturbation (8. m (8.115) (8.118) . We can. We want to estimate now the second term in (8. hence.41) one can state e2 e2 1 8πNω ω A2 ∼ 2me 2me ω 2 V For the same assumptions as above one obtains e2 e2 A2 ∼ · 2me 8πao ao 4 ω λ me c2 .. and me c2 ≈ A A 500 keV 2 ao ω ≈ 10−8 (8. i. i. 2me (8.112) π λ me c2 and with e2 /ao ≈ 27 eV. Using again (8. u) p · u exp i(k · r − ωt) ˆ ˆ ˆ kV . (8.5 ˚. Hence.114) Employing for the second factor the estimate as stated in (8.105) for radiation ﬁelds can be considered a small perturbation.105).8. (8.117) In case that such perturbation acts on an electron and is due to superpositions of planar waves described through the vector potential (8. one obtains for (8.111) For ω = 3 eV and a corresponding λ = 4000 ˚ one obtains. altogether. one can neglect this term as long as the ﬁrst term gives non-vanishing contributions.109) using V = λ 4π 2 c2 /ω 2 e2 e ˆ p·A ∼ me 4πao 2 ao ω π λ me c2 1 2 (8..116) This term is obviously much smaller than the ﬁrst term. a volume V = λ3 where λ is the wave length corresponding to a plane wave with frequency ω.5: Time-Dependent Perturbation Theory 217 The virial theorem for the Coulomb problem provides the estimate for the case of a hydrogen atom p2 2me ∼ 1 e2 2 ao (8. and as long as the photon densities Nω /V are small. Consequently. me (8. λ = 2πc/λ.110) where ao is the Bohr radius. Nω = 1.ˆ u 4πNk α(k.105) due to a radiation ﬁeld by VS = − q ˆ p · A(r. Assuming a single photon.112) yields e2 A2 ∼ 10 eV · 10−8 = 10−7 eV .113) This magnitude is much less than the diﬀerences of the typical eigenvalues of the lowest states of the hydrogen atom which are of the order of 1 eV. with ao ≈ 0.e. e ˆ p · A ∼ 10 eV · 10−4 = 10−3 eV . t) .

|Ψ(to ) = |0 (8..121) (8. One seeks to predict the probability to observe the particle in one of the states |n . under ordinary circumstances.120) (8.108) as the eigenstates of the unperturbed Hamiltonian Ho .127) i ∂t |ΨD (t) = VD (t) |ΨD (t) . The probability to ﬁnd the particle in the state |n at time t is then p0→n (t) = | n|ΨS (t) |2 . . n = 0 at some later time t ≥ to . A factor α(k. Using i i ∂t exp − Ho (t − to ) and (8. For this purpose one needs to determine the state |ΨS (t) of the particle. 2. (8.106–8. n = 1.119) In order to determine the wave function ΨS (t) we choose the so-called Dirac representation deﬁned through i |ΨS (t) = exp − Ho (t − to ) |ΨD (t) (8.42) through k = |k| = ω. |n are deﬁned in (8. u) has been added to describe eliptically or circularly polarized ˆ waves. The states |0 . describes the eﬀect of a radiation ﬁeld on an electron system and which will be assumed below to describe radiative transitions.126) Ho (t − to ) yields ﬁnally .218 Interaction of Radiation with Matter where we have replaced ω in (8. The sum runs over all possible k vectors and might actually be an integral. .125) (8.122) where |ΨD (to ) = |0 . the sum over u involves the two possible polarizations of planar ˆ electromagnetic waves. Perturbation Expansion The generic situation we attempt to describe entails a particle at time t = to in a state |0 and a radiation ﬁeld beginning to act at t = to on the particle promoting it into some of the other states |n .124) (8. . This state obeys the Schr¨dinger equation o i ∂t |ΨS (t) = [ Ho + VS (t) ] |ΨS (t) subject to the initial condition |ΨS (to ) = |0 . Equation (8.119) one obtains i exp − Ho (t − to ) ( Ho + i ∂t ) |ΨD (t) i = [ Ho + VS (t) ] exp − Ho (t − to ) |ΨD (t) from which follows i i exp − Ho (t − to ) i ∂t |ΨD (t) = VS (t) exp − Ho (t − to ) |ΨD (t) .118) is the form of the perturbation which. Multiplying the latter equation by the operator exp i (8.123) i = Ho exp − Ho (t − to ) (8.

133.133–8. From (8. n| (8. to (8.137) follows |ΨD (t) = |0 .128) We note that the transition probability (8.127) we assume the expansion ∞ |ΨD (t) = n=0 (n) |ΨD (t) (n) (8.137) One can readily verify that (8.132–8. (8.130) = exp − i n (t − to ) − to )]| = 1.137) can be solved recursively.121) expressed in terms of ΨD (t) is i p0→n (t) = | n|exp − Ho (t − to ) |ΨD (t) |2 . 0 VD (t) |ΨD (t) (1) VD (t) |ΨD (t) (0) (8. . Equations (8.137) are consistent with (8. We will consider here only the two leading contributions to |ΨD (t) .134) (8. .133) (8.138) 1 i t dt VD (t ) |0 . Accord(n) ingly.8.132) where |ΨD (t) accounts for the contribution due to n-fold products of VD (t) to |ΨD (t) .5: Time-Dependent Perturbation Theory where VD (t) = exp 219 i i Ho (t − to ) VS (t) exp − Ho (t − to ) . we deﬁne |ΨD (t) through the evolution equations i ∂t |ΨD (t) i ∂t |ΨD (t) i (2) ∂t |ΨD (t) (1) (0) = = = .127. Due to the Hermitean property of the Hamiltonian Ho holds n|Ho = i n|exp − Ho (t − to ) from which we conclude. (8.135) i ∂t |ΨD (t) (n) VD (t) |ΨD (n−1) (t) (8. .134. (8. consequently. 8.137) |ΨD (t) = (1) (0) (8.129) n| and. .139) . 8. = .131) p0→n (t) = | n |ΨD (t) |2 .128). using |exp[− i n (t n (8. In order to determine |ΨD (t) described through (8. . Employing this result one obtains for (8.136) together with the initial conditions |ψD (to ) = |0 0 for n = 0 for n = 1. 8. . 2. .

(8.220 This result.144) Here the factor I describes the strength of the radiation ﬁeld (for the speciﬁed planar wave) as determined through the photon density Nj /V and the factor II describes the polarization of the planar wave.135. For the sake of including the eﬀect of superpositions of plane waves we will assume.118). A2 are deﬁned through (8.144) r is the position of the electron and p = ( /i) is the momentum operator of the electron.34.143) ˆ ˆ where V1 and V2 are time-independent operators deﬁned as e ˆ Vj = m 8πNj ωj V I ˆ ˆ p · uj eik·r .142). A factor exp(λt). This factor will serve mainly the purpose of keeping in the following derivation all mathematical quantities properly behaved. 8.141) + . The resulting perturbation on an electron system. the direction of ˆ the propagation being determined by k = k/|k|. non-singular. in turn. to → −∞ (8.. i. The factor III in (8. II III (8. yields for (8. however. We will demonstrate below that the the sign of iωt determines if the energy of the planar wave is absorbed (“-” sign) or emitted (“+” sign) by ˆ the quantum system. deﬁnes the direction of the E-ﬁeld of the ˆ radiation. The coeﬃcients A1 .137) |ΨD (t) = (2) Interaction of Radiation with Matter 1 i 2 t t dt to to dt VD (t ) VD (t ) |0 . . according to (8.142) combining an incoming and an incoming or outgoing wave.144) describes the propagation of the planar wave. is VS = ˆ ˆ V1 exp(−iω1 t) + V2 exp( iω2 t) eλt . that two planar waves simulataneously interact with an electronic system. t) = + A1 u1 exp i (k1 · r − ω1 t) ˆ A2 u2 exp i (k2 · r ˆ ω2 t) incoming wave incoming or outgoing wave (8..118) above that the eﬀect of a radiation ﬁeld on an electronic system is accounted for by perturbations with a so-called harmonic time dependence ∼ exp(−iωt).140) Altogether we have provided the formal expansion for the transition amplitude n|ΨD (t) = n|0 + ∞ 1 i 1 i t dt n|VD (t ) |0 to 2 t t (8. 8.6 Perturbations due to Electromagnetic Radiation We had identiﬁed in Eq. note that uj . We want to apply now the perturbation expansion derived to such perturbations. (8.e.41). + m=0 dt to to dt n|VD (t )|m m|VD (t ) |0 8. λ → 0+ . In (8. according to (8. λ → 0+ has been introduced which describes that at time to → −∞ the perturbation is turned on gradually. such that the combined radiation ﬁeld is decribed by the vector potential A(r..

According to (8. using (8.147) ˆ ˆ n|V1 |0 e−iω1 t + n|V2 |0 e Carrying out the time integration and taking the limit limt→−∞ yields (1) n|ΨD (t) = λ→0+ lim e λt exp ˆ n|V1 |0 o − o − ω1 ) t + + ω1 − n + iλ ( n i i ˆ + n|V2 |0 2nd Order Processes exp o − o ω2 ) t ± ω2 − n + iλ ( n . (8.143) to the expansion (8. (8.150) and. (8.151) ˆ ˆ n|V1 |m m|V1 |0 exp ˆ ˆ + n|V2 |m m|V2 |0 exp i i n − − m − ω1 − i λ)t exp ω2 − i λ)t exp m o n m m o . yields n|ΨD (t) = − (2) 1 ∞ t t 2 λ→0+ t→−∞ lim lim ( ( dt m=0 to to dt i i ( ( − − − ω1 − i λ)t ω2 − i λ)t (8.128) one obtains k|VD (t) | = k|VS (t)| exp i ( k − ) (8.141).148) We consider now the 2nd order contribution to the transition amplitude.8.140. employing the perturbation (8.143).128). (8.141) this is n|ΨD (t) = − (2) 1 2 ∞ m=0 t t dt to to dt n|VD (t ) |m m|VD (t ) |0 .130) and (for m = 0) i i exp − Ho (t − to ) |m = exp − for (8. (8.146) = 1 λ→0+ t→−∞ i lim lim × t dt exp to i ( n − o − i λ) t iω2 t × . For the 1st order contribution to the transition amplitude n|ΨD (t) = (1) 1 i t dt n|VD (t ) |0 to (8. 8.6: Perturbations due to Electromagnetic Radiation 1st Order Processes 221 We employ now the perturbation (8.149) Using the deﬁnition of VD stated in (8.145) n|ΨD (t) (1) m (t − to ) |m .145) we obtain then.

the time derivative of p0→n (t). (8.152) m=0 i ˆ ˆ n|V1 |m m|V1 |0 exp m − o − ω1 − i λ + − 2 ω1 − 2i λ)t n − o − 2 ω1 − 2i λ ( n o − ˆ ˆ n|V2 |m m|V2 |0 exp ω2 − i λ m− o ˆ1 |m m|V2 |0 exp ˆ n|V + ω2 − i λ m− o ˆ2 |m m|V1 |0 exp ˆ n|V + m − o − ω1 − i λ 1st Order Radiative Transitions i 2 ω2 − 2i λ)t 2 ω2 − 2i λ n− o i ( n − o − ω1 ω2 − 2i λ)t ω2 − 2i λ n − o − ω1 ( n o i − − ω1 ω2 − 2i λ)t ω2 − 2i λ n − o − ω1 ( n o − The 1st and 2nd order transition amplitudes (8. respectively. in which case one can expect that it is larger than the 2nd (2) order contribution n|ΨD (t) which we will neglect. We also assume for the ﬁnal state n = 0 such that n|0 = 0 and p0→n (t) = | n|ΨD (t) |2 holds.e. Using (8.152).131).155) λ→0+ ( o We are actually interested in the transition rate.222 ˆ ˆ + n|V1 |m m|V2 |0 exp ˆ ˆ + n|V2 |m m|V1 |0 exp i i ( ( − − Interaction of Radiation with Matter − ω1 − i λ)t exp ω2 − i λ)t exp i i ( ( − − ω2 − i λ)t − ω1 − i λ)t n m m o n m m o Carrying out the integrations and the limit limt→−∞ provides the result (2) n|ΨD (t) = − 1 ∞ 2 λ→0+ lim (8.153) (8.148) and (8. provide now the transition probability p0→n (t) according to Eq. i.154) yields p0→n (t) = lim e2λt + + 2 Re ˆ n|V1 |0 |2 ( o + ω1 − n )2 + (λ )2 ˆ n|V2 |0 |2 ( o ± ω1 − n )2 + (λ )2 ˆ ˆ n|V1 |0 0|V2 |n exp i (± ω2 − ω1 ) t + ω1 − n + iλ ) ( o ± ω2 − n − iλ ) (8.148) and ∗ |z1 + z2 |2 = |z1 |2 + |z2 |2 + 2Re(z1 z2 ) (1) (8.. For this rate holds . We assume ﬁrst that the ﬁrst order transi(1) tion amplitude n|ΨD (t) is non-zero.

Using lim x2 + 2 δ→0+ = π δ(x) (8.e. however. the matrix elements can vary strongly for diﬀerent states |n of the system. which is observed through the so-called spectral intensities of transitions |0 → |n . The result supports our deﬁnition of incoming and outgoing waves in (8.159) leads to ﬁnal states |n with energy n = o + ω1 . describing either an incoming or an outgoing plane wave due to the vector potential A2 u2 exp i (k1 · r ˆ ω2 t) . Hence. The δ-functions appearing in this expression reﬂect energy conservation: the incoming plane wave contribution of (8. The time average will be denoted by · · · t .158) ˆ + | n|V2 |0 |2 δ( − ω2 ) Obviously. First. The result of such average is.158).6: Perturbations due to Electromagnetic Radiation d p0→n (t) dt lim e2λt + × 2 Re ˆ 2 λ n|V1 |0 |2 ( o + ω1 − n )2 + (λ )2 ˆ 2 λ n|V2 |0 |2 d + 2λ + × ( o ± ω1 − n )2 + (λ )2 dt ˆ ˆ n|V1 |0 0|V2 |n exp i (± ω2 − ω1 ) t + ω1 − n + iλ ) ( o ± ω2 − n − iλ ) 223 = λ→0+ (8.31) and (8. 8.144). these matrix elements determine the so-called selection rules for the transition: the matrix elements vanish for many states |n and |0 on the ground of symmetry and geometrical properties.160) leads to ﬁnal states |n with energy n = o ± ω2 .156) such that the 1st order contributions of the two planar waves of the perturbation simply add. For the resulting expression the limit limλ→0+ can be taken. . i.156) ( o The period of electromagnetic radiation absorbed by electronic systems in atoms is of the order 10−17 s.8. of this expression describe the individual eﬀects of the two planar wave contributions of the perturbation (8. due to the vector potential A1 u1 exp i (k1 · r − ω1 t) . The second contribution to (8. due to the uncertainty relationship would introduce a considerable perturbation to the system.142–8. is much shorter than could be resolved in any observation.157) one can conclude for the average transition rate k = d p0→n (t) dt t = 2π ˆ | n|V1 |0 |2 δ( n − n o − ω1 ) o (8. in fact. In case the matrix elements are non-zero.143. to cancel the third term in (8. ˆ (8.158) play an essential role for the transition rates of radiative transitions. one should average the rate over many periods of the radiation. a property. any attempt to do so..144). the two terms apearing on the rhs.142) ˆ ˆ The matrix elements n|V1 |0 and n|V2 |0 in (8. (8.

|z1 + z2 + z3 |2 t = |z1 |2 + |z2 |2 + |z3 |2 (8.224 2nd Order Radiative Transitions Interaction of Radiation with Matter We now consider situations where the ﬁrst order transition amplitude in (8.161) To determine the transition rate we proceed again.k=1 j=k 2 2 2 2 ∗ zj zk (8. We deﬁne ∞ z1 = m=0 ˆ ˆ n|V1 |m m|V1 |0 m − o − ω1 − i λ exp i × (8.153–8.163) × (8.. (8.. in Eqs.e.165) × ∞ z3 = m=0 2 ω2 − 2i λ)t 2 ω2 − 2i λ n− o ˆ ˆ ˆ ˆ n|V2 |m m|V1 |0 n|V2 |m m|V1 |0 + ω2 − i λ m − o − ω1 − i λ m− o ( n o i − × It holds exp − ω1 ω2 − 2i λ)t ω2 − 2i λ n − o − ω1 ( n o 3 − |z1 + z2 + z3 | = |z1 | + |z2 | + |z3 | + j.152).162) × and.164) (8.g.e.158).. (2) (8.167) Taking now the limit limλ→0+ and using (8. in analogy to (8. similarly. hence. Time average · · · t of expression (8. as we did in the the case of 1st order transitions. it holds p0→n (t) = | n|ΨD (t) |2 . exp[ i (±ω2 − ω1 )]. e.153) vanishes such that the leading contribution to the transition probability p0→n (t) arises from the 2nd order amplitude (8.166) over many periods of the radiation yields exp[ i (±ω2 − ω1 )] t = 0 and.166) In this expression the terms |zj |2 exhibit only a time dependence through a factor e2λt whereas the terms zj z ∗ k for j = k have also time-dependent phase factors. i. k = = d p0→n (t) dt 2π ∞ m=0 t 2 ˆ ˆ n|V1 |m m|V1 |0 m − o − ω1 − i λ δ( m − o − 2 ω1 ) (8.168) absorption of 2 photons ω1 . i.158). ∞ − 2 ω1 − 2i λ)t n − o − 2 ω1 − 2i λ ( n o − z2 = m=0 ˆ ˆ n|V2 |m m|V2 |0 ω2 − i λ m− o exp i × (8.157) yields.

168) the interpretation is ∞ ˆ n|V1 |m m=0 1 m − o − ω1 − i λ virtually occupied state |m ˆ m|V1 |0 pert. namely absorption/ emission ﬁrst of photon ω2 and then absorption of photon ω1 and.. according to its δ-function factor. interference between the contributions from all intermediate states |m can arise. this state is only virtually excited. a second perturbation. 8. accordingly. Similarly.158) which accounts for such transitions. i. accordingly. describes processes which lead to ﬁnal states |n with energy n = o + 2 ω1 and. ﬁrst absorption of photon ω1 and then absorption/ emission of photon ω2 . The remaining two contributions in (8.. Similarly.168) describes two intermediate histories. i. |m ← |0 (8. which is stationary and energy is conserved. the third term describes processes in which a photon of energy ω1 is absorbed and a second photon of energy ω2 is absorbed/emitted. (8.168) describe similar histories of the excitation process. each of energy ω1 .168) describe the time sequence of the two photon absorption/ emission processes. For this purpose we will employ the transition rate as given in Eq. describe the absorption of two photons.169) pert. Most remarkably. promotes the system then to the state |n . The factors | · · · |2 in (8. describe the absorption/emission of two photons.e. the third term in (8. m = o + ω1 ) and the evolution of state |m is described by a factor 1/( m − o − ω1 −i λ).7: One-Photon Emission and Absorption 2π ∞ m=0 225 2 + ˆ ˆ n|V2 |m m|V2 |0 ω2 − i λ m− o δ( m − o 2 ω2 ) absorption/emission of 2 photons ω2 + 2π ∞ m=0 ˆ ˆ n|V2 |m m|V1 |0 + m − o − ω1 − i λ + ˆ ˆ n|V2 |m m|V1 |0 ω2 − i λ m− o 2 δ( n − o − ω1 ω2 ) absorption of a photon ω1 and absorption/emission of a photon ω2 This transition rate is to be interpreted as follows.e. The expression sums over all possible virtually occupied states |m and takes the absolute value of this sum. i. the second term describes the processes leading to ﬁnal states |n with energy n = o ± 2 ω2 and. i.e. |n ← |m . each of energy ω2 ..7 One-Photon Absorption and Emission in Atoms We ﬁnally can apply the results derived to describe transition processes which involve the absorption or emission of a single photon. . vice versa. The ﬁrst term. it must hold n = o + 2 ω1 . the system is perturbed through absorption of a photon with energy ω1 from the initial state |0 into a state |m . In case of the ﬁrst term in (8. there is no energy conservation necessary (in general.8.e.. through absorption of a photon.

the vector potential actually assumed is A(r. φ) (8. For example.170) We will employ only the real part of this potential.e.174) We seek to evaluate the matrix element ˆ M = n| p eik·r |0 . φ) × × eik·r e−r/ao Y00 (θ. .144) VS = where e 2πN ˆ ˆ p · u e±ik·r . (8.172) δ( n − o − ω) (8. to → −∞ (8. the second term can be discounted in case of absorption. The absorption rate. the wave functions are (n.143. φ) a3 ao o 2p (8. ˆ (8. in this range one can expand eik·r ≈ 1 + ik · r + .176) ψn=2. in case of a radiative transition from the 1s state of hydrogen to one of its three 2p states. =0. (8. .177) √ M = 6 0 ∞ 1 2π ia4 o r2 dr −1 dcosθ 0 ∗ dφ r e−r/2ao Y1m (θ. .m (r. (8. and the integral is =1. φ) = 6 r −r/2ao e Y1m (θ. φ) = 2 − 1 2 1 −r/ao e Y00 (θ. t) = 2πN ωV u exp ˆ ı (k · r − ωt) + 2πN ωV u exp ˆ ı (−k · r + ωt) .. λ → 0+ . plane polarized wave described through the complex vector potential A(r. φ) a3 o 1s (8.171) The perturbation on an atomic electron system is then according to (8.158).143) will contribute to the absorption process. θ.178) These wave functions make signiﬁcant contributions to this integral only for r-values in the range r < 10 ao .179) . i.2 = m ωV Only the ﬁrst term of (8. 8.226 Absorption of a Plane Polarized Wave Interaction of Radiation with Matter We consider ﬁrst the case of absorption of a monochromatic.175) The matrix element involves a spatial integral over the electronic wave functions associated with states |n and |0 . t) = 8πN ωV u exp ˆ ı (k · r − ωt) . However.173) V1. θ. (8. according to (8. is then kabs = Dipole Approximation 2π e2 2πN m2 ωV e ˆ u · n| p eik·r |0 ˆ 2 ˆ ˆ V1 exp(−iωt) + V2 exp(+iωt) eλt .m=0 (r. m denote the relevant quantum numbers) ψn=1.

Ho ] i where Ho is the Hamiltonian given by (8. Using |k| = 2π/λ.180) ∆E2p−1s and ao = 0.7: One-Photon Emission and Absorption 227 One can estimate that the absolute magnitude of the second term in (8.184) and the commutation property [xk . is Ho = ˆ (p)2 e2 + V (r) . Ho ] = ˆ2 p [ r.178) holds eik·r ≈ A 1 1 + O( 50 ) such that one can approximate eik·r ≈ 1 . Transition Dipole Moment A further simpliﬁcation of the matrix element (8. pk ] + ˆ ˆ k=1 1 2me 3 [ r.182) (8. pk ] pk ˆ k=1 (8.186) δ( n − o − ω) (8.181) For the commutator in (8. in case of the hydrogen atom.181) and (8.8. V (r) ] 2me =0 = Using r = 3 ˆ j=1 xj ej 1 2me 3 pk [ r.179) and other terms are never larger than 20π ao /λ. the value of the wave length for the 1s → 2p transition 2π c λ = = 1216 ˚ A (8.175).174) yields kabs = 4π 2 e2 N ω V u · n| ˆ |0 ˆ r 2 m( o − m n| [r.185) from which follows (8. (8.183) (8.k=1 i m 3 pk ek = ˆ j.187) . 2me r (8. pj ] = i δkj one obtains ˆ [ r. Ho ] |0 = i i n) n|r |0 .175) can then be ˆ achieved and the diﬀerential operator p = i replaced by by the simpler multiplicative operator r. One refers to this approximation as the dipole approximation. We are now in a position to obtain an alternative expression for the matrix element (8.182) one ﬁnds [ r. This simpliﬁcation results from the identity m ˆ p = [ r. ] + [ r.529 ˚ one concludes that in the signiﬁcant integration range in (8. Ho ] = i m 3 pk ej δjk = ˆ j. V (r) = − .182) one obtains M ≈ Insertion into (8.182).104) and. Using (8.k=1 i ˆ p m (8.

ρ) holds then u cosα = cosϑ1 cosϑ + sinϑ1 sinϑ cos(ϕ1 − ϕ) = sinϑcosϕ and cosβ = cosϑ2 cosϑ + sinϑ2 sinϑ cos(ϕ2 − ϕ) = sinϑsinϕ .33) are orthogonal to each other. u2 and k ˆ ˆ coordinate system. there are ony two linearly independent directions of u possible. a very hot oven. For the two angles α = ∠(ˆ1 .190) (8.188) ˆ Here dk is the integral over all orientations of k. the direction of u1 is described by the ˆ ˆ ˆ ˆ angles ϑ1 = π/2. three-dimensional vector. planar waves as deﬁned in (8. will actually be replaced by a distribution function which reﬂects (1) the ﬁnite life time of the states |n . Accordingly. We have demonstrated in our derivation of the rate of one-photon processes (8. We can. i. according to (8. obtain the transition rate in the present case by adding the individual transition rates of all planar waves present in the oven. k ˆ As a result. in practical situations. The unit ˆ ˆ ˆ ˆ can be chosen to point along the x1 . The δ-function appearing in this expression. k frame is described by the angles ϑ.191) (8. Instead of adding the components of all possible k values we integrate over all k using the following rule +∞ =⇒ k u ˆ V −∞ k 2 dk (2π)3 ˆ dk u ˆ (8. Absorption of Thermal Radiation We want to assume now that the hydrogen atom is placed in an evironment which is suﬃciently hot. x3 -axes of a right-handed cartesian vectors u1 . ρ) and ˆ u β = ∠(ˆ2 . ϕ1 = 0 and of u2 by ϑ2 = π/2. The direction of this vector in the u1 . ϕ2 = π/2. and (2) the fact that strictly monochromatic radiation cannot be prepared such that any radiation source provides radiation with a frequency distribution. and all polarizations of the radiation.e.. x2 .189) where the factor 1/ arose from the integral over the δ-function.158) above that in ﬁrst order the contributions of all components of the radiation ﬁeld add. say u1 and u2 .228 Interaction of Radiation with Matter where we used the fact that due to the δ-function factor in (8. u2 . ˆ u ˆ (8.193) .35). Let us assume that the wave functions describing states |n and |0 have been chosen real such that ρ = n|r|0 is a real. ϕ. one can express | u · n| r |0 |2 = |ρ|2 ( cos2 α + cos2 β ) = sin2 θ .192) and obtain ˆ dk u ˆ u · n| ˆ |0 ˆ r 2 2π 1 = |ρ|2 0 −1 dcosϑ (1 − cos2 ϑ) = 8π 3 (8. such that the thermal radiation present supplies a continuum of frequencies. ˆ In order to carry out the integral dk we note that u describes the possible polarizations of the ˆ ˆ and u. hence. directions.187) and using k c = ω results in the total absorption rate kabs (tot) = e2 Nω ω 3 2π c3 ˆ dk u ˆ u · n| ˆ |0 ˆ r 2 (8. |0 .31–8.174) one can replace n − o by ω. Integrating and summing accordingly over all contributions as given by (8.

In case of a classical radiation ﬁeld one would expect that emission cannot occur. however.37 × 1019 1 ao | n|r|0 |2 × s λ λ2 .194) combined are typically somewhat smaller than (1 ˚/1000 ˚)3 = A A 10−9 . The most intensely absorbing molecules are long. i. . . 1. However. Nω photons before absorption.e. the resulting total rate of emission bears a diﬀerent dependence on the number of photons present in the environment. (8. i. the absorption rate is of the order of 109 s−1 or 1/nanosecond.194) for the absorption rate shows that the essential property of a molecule which determines the absorption rate is the so-called transition dipole moment | n| r |0 |.194) For absorption processes involving the electronic degrees of freedom of atoms and molecules this radiation rate is typicaly of the order of 109 s−1 . for Nω = 0.197) o (tot) (8. This diﬀerence between emission and absorption is due to the quantum nature of the radiation ﬁeld.173) contributes.172.7: One-Photon Emission and Absorption 229 This geometrical average. there is also a contribution to the emission rate which is proportional to Nω . can be inserted into (8. The radiation ﬁeld is described. The corresponding process is termed spontaneous emission. The quantum nature of radiation manifests itself in that the number of photons Nω msut be an integer.. a problem in case of emission by quantum systems in complete darkness. Accordingly.8. in particular. The transition dipole moment can vanish for many transitions between stationary states of a quantum system. linear molecules. Emission of Radiation We now consider the rate of emission of a photon. Nω = 0. 2 . by planar waves with vector potential (8.. For this purpose we use ω/c = 2π/λ and obtain 4 e2 ω 3 32π 3 e2 ao 1 ao = = 1.189) to yield the total absorption rate kabs (tot) = Nω 4 e2 ω 3 | n|r|0 |2 3 c3 .196) The last two factors in (8. a quantum mechanical treatment of the radiation ﬁeld leads to a total emission rate which is proportional to Nω + 1 where Nω is the number of photons before emission. in agreement with observations. The value of | n| r |0 | determines the strength of an optical transition. ˆ In case of emission only the second term V2 exp(+iωt) in (8. . the calculation of the emission rate proceeds as in the case of absorption.173). (8. However.37 × 1019 × 3 3 c3 3 ao λ3 s λ and kabs where λ = 2πc n − (8.. that emission occurs even if no photon is present in the environment. for atoms or symmetric molecules.171) and perturbation (8. This dependence predicts. Transition Dipole Moment The expression (8. However. Otherwise. 8. ﬁnally.e. as for the absorption process. For practical evaluations we provide an expression which eliminates the physical constants and allows one to determine numerical values readily.195) = Nω 1. This poses.

respectively.200). Examples are radiative transitions in which two photons are absorbed or emitted or scattering of radiation by matter in which a photon is aborbed and another re-emitted. For these numbers holds Nn / No = exp[−( n − o )/kB T ] (8.204) Nω . Deﬁning the rate of spontaneous emission 4 e2 ω 3 | n|r|0 |2 (8. e.198) to determine the stationary distribution of photons ω in an oven of temperature T . Nω + 1 (8.8 Two-Photon Processes In many important processes induced by interactions between radiation and matter two or more photons participate. The total rate of emission.e. the well-known Planck radiation formula. In the following we discuss several examples. it must hold Nω ksp N0 = (Nω + 1) ksp Nn (8. Under stationary conditions the number of hydrogen atoms undergoing an absorption process |0 → |n must be the same as the number of atoms undergoing an emission process |n → |0 . Hence..202) It follows.g.199) Planck’s Radiation Law The postulate of the Nω + 1 dependence of the rate of emission as given in (8. The number of atoms undergoing absorption in unit time are Nω ksp No and undergoing emission are (Nω +1)ksp Nn . 8.203) i. Let No and Nn denote the number of atoms in state |0 and |n . in lasers. To demonstrate this property we apply the transition rates (8. is (tot) kem 4 e2 ω 3 | n|r|0 |2 (spontaneous emission) 3 3c 4 e2 ω 3 | n|r|0 |2 (induced emission) + Nω 3 3c 4 e2 ω 3 = ( Nω + 1 ) | n|r|0 |2 3 3c Nω photons before emission.230 Interaction of Radiation with Matter which is termed induced emission since it can be induced through radiation provided.201) ksp = 3 c3 the rates of absorption and emission are Nω ksp and (Nω + 1)ksp . = (8. exp[ ω/kB T ] − 1 (8. exp[− ω/kB T ] = This equation yields Nω = 1 . using (8.195) and (8.198) is consistent with Planck’s radiation law which reﬂects the (boson) quantum nature of the radiation ﬁeld. We assume n − o = ω.198) (8. accordingly. . respectively.200) where kB is the Boltzmann constant..

(8. .207) does not converge rapidly.206) − o − 2 ω) . ˆ (8. then expression (8.173).208) has been prepared. In this case one needs to choose the energy of the photons to obey n = o + 2 ω.207) one is.182) yields. dropping the index 1 characterizing the radiation. The transition requires that the transition dipole moment n| r |0 does not vanish for two states |0 and |n .104) with stationary states |n deﬁned through (8. The perturbation which accounts for the coupling of the electronic system and the radiation ﬁeld is the same as in case of 1st order absorption processes and given by (8. however. The intense radiation ﬁelds of lasers allow one to increase transition rates to levels which can readily be observed in the laboratory. However. k = 2π e2 2πNω m2 ωV e × δ( m 2 ∞ m=0 n|ˆ · p eik·r |m m|ˆ · p eik·r |0 u ˆ u ˆ m − o − ω1 − i λ 2 × (8. under ordinary circumstances.172. k = Nω V 2 8π 3 e4 × δ( m− ∞ m=0 o ( n − ˆ m) u · n|ˆ |m ( m − o ) u · m|ˆ |0 r ˆ r ω( m − o − ω − i λ) 2 − 2 ω) . One obtains.8: Two-Photon Processes Two-Photon Absorption 231 The interaction of electrons with radiation. t) = Ao1 u1 cos(k1 · r − ω1 t) .106). faced with the dilemma of having to sum over all intermediate states |m of the system.8. We assume that a planar wave with wave vector k1 and polarization u1 .207) for the rate of 2-photon transitions shows that the transition |0 → |n becomes possible through intermediate states |m which become virtually excited through absorption of a single photon.187). a transition between the states |0 and |n may be possible.205) The respective radiative transition is of 2nd order as described by the transition rate (8. If the sum in (8. We wish to describe an observation in which a detector is placed at a solid angle element dΩ2 = sinθ2 dθ2 dφ2 with respect to the origin of the coordinate system in which the electron is described.207) Expression (8. but then requires the absorption of two photons. which is not necessarily the case.168) where the ﬁrst term describes the relevant contribution. Employing the dipole approximation (8. Scattering of Photons at Electrons – Kramers-Heisenberg Cross Section We consider in the following the scattering of a photon at an electron governed by the Hamiltonian Ho as given in (8. induce single photon absorption processes as described by the transition rate Eq. however.207) does not provide a suitable avenue of computing the rates of 2-photon transitions. The electron absorbs the radiation and emits immediately a second photon. ﬁnally. as described through the vector potential ˆ A(r. even if n| r |0 vanishes. In applying (8. The resulting rate of the transition depends on 2 Nω .181) and using (8. (8. in case ˆ of absorption only V1 contributes. (8. 8.

211) are given by VS (t) = e ˆ p · A+ u1 exp[i(k1 · r − ω1 t)] + A− u2 exp[−i(k2 · r − ω2 t)] o1 ˆ o2 ˆ 2me contributes in 2nd order + e2 4me A+ A− u1 · u2 exp{i[(k1 − k2 ) · r − (ω1 − ω2 ) t]} ˆ o1 o2 ˆ contributes in 1st order The eﬀect of the perturbation on the state of the electronic system is as stated in the perturbation expansion (8.105). later we will allow the quantum system to select appropriate values.209) cosθ2 where the value of k2 has been ﬁxed.210) is accounted for by the following contributions of (8. o1 ˆ o2 ˆ (8.210) A(r. the amplititude A− deﬁned in (8.212) where N1 /V is the density of photons for the wave described by (8. u2 .210). u1 . The relevant terms of the perturbation (8. the amplitude A+ is given by o1 A+ = o1 8πN1 ω1 V (8. the wave characterized through k1 . The second term in (8. for the components of the wave (8. Let us assume for the present that the emitted photon has a wave vector ˆ k2 with cartesian components sinθ2 cosφ2 k2 = k2 sinθ2 sinφ2 (8. ˆ The perturbation which arises due to the vector potential (8. hence.213) where N2 /V is the density of photons characterized through k2 . In the present case we consider only scattering processes which absorb radiation corresponding to the vector potential (8. (8.208.. 8.214) . t) = A+ u1 exp[ i (k1 · r − ω1 t) ] + A− u2 exp[ i (k2 · r − ω2 t) ] .208).208.6 above that the absorption of the radiation in (8.210). We know already from our description in Section 8.210) The vector potential which describes both incoming wave and outgoing wave is a superposition of the potentials in (8.211) accounts for the emitted wave and.211) is o2 A− = o2 8π (N2 + 1) ω1 V (8.232 Interaction of Radiation with Matter We assume that the experimental set-up also includes a polarizer which selects only radiation with a certain polarization u2 . according to the ˆ description of emission processes on page 229.208) and emit radiation corresponding to the vector potential (8.141).211) The ﬁrst term describes the absorption of a photon and. The vector potential describing the emitted plane wave is then A(r.e. t) = Ao2 u2 cos(k2 · r − ω2 t) . ˆ (8.105) using the vector potantial (8.211) is stated in Eq. in the present case. This expansion yields. i.208) and the emission of the radiation in (8. 8. however.

Only the second (1st order) and the third (2nd order) terms in (8.8. one needs to integrate the rate as given by (8.215) correspond to scattering processes in which the radiation ﬁeld ‘looses’ a photon ω1 and ‘gains’ a photon ω2 .215) and taking the limits limto →−∞ and limλ→0+ yields the transition rate k − m = 2π δ( n − o − ω1 + ω2 ) e2 A+ A− u1 · u2 n|0 ˆ ˆ 4m2 o1 o2 e 2 (8.213) for the amplitudes A+ and A− results o1 o2 in the Kramers-Heisenberg formula for the scattering rate k = N1 c 2 ω2 r (N2 + 1) dΩ2 u1 · u2 n|0 ˆ ˆ V o ω1 1 n|ˆ2 · p |m m|ˆ1 · p |0 u ˆ u ˆ n|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ − + me m m − o + ω2 m − o − ω1 (8. Following closely the procedures adopted in evaluating the rates of 1st order and 2nd order radiative transitions on page 222–225.181) in stating this result. one needs to carry out the integration 2 V(2π)−3 k2 dk2 · · ·. me c2 (8.. 8.215) t dt ei( to n− o− ω1 + ω2 +i λ)t 1 i 2 e2 4m2 e A+ A− × o1 o2 × ˆ ˆ u1 · n| p |m u2 · m| p |0 × ˆ ˆ t t × to dt to dt ei( n− m− ω1 +i λ)t i( e m− o+ ω2 +i λ)t ˆ ˆ + u2 · n| p |m u1 · m| p |0 × ˆ ˆ t t × to dt to dt ei( n− m+ ω2 +i λ)t i( e m− o− ω1 +i λ)t We have adopted the dipole approximation (8.216) over all available modes of the ﬁeld.8: Two-Photon Processes function accounting for absorption and re-emission of a photon n|ΨD (t) = + + m=0 233 n|0 + 1 e2 A+ A− u1 · u2 n|0 ˆ ˆ i 4me o1 o2 ∞ (8.217) 2 Here ro denotes the classical electron radius ro = e2 = 2.8 · 10−15 m .e.. Accordingly. i. Hence.e. i.218) .212. evaluating the time integrals in (8. only these two terms contribute to the scattering amplitude.216) e2 + − A A 4me o1 o2 n|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ n|ˆ2 · p |m m|ˆ1 · p |0 u ˆ u ˆ + m − o + ω2 m − o − ω1 We now note that the quantum system has the freedom to interact with any component of the radiation ﬁeld to produce the emitted photon ω2 . Inserting also the values (8.

224) = m 1 − o ( m ω ( ω)2 + + . 2 ... i.227) u ˆ 0|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ 0|ˆ2 · p |m m|ˆ1 · p |0 u ˆ + me ( m − o )3 me ( m − o )3 (8.221) Using |n = |0 and.217) to the scattering cross section deﬁned through dσ = It holds then dσ = 2 ro rate of photons arriving in the the solid angle element dΩ2 ﬂux of incoming photons ω2 (N2 + 1) dΩ2 u1 · u2 n|0 ˆ ˆ ω1 1 n|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ me m m (8.220) n|ˆ2 · p |m m|ˆ1 · p |0 u ˆ u ˆ + m − o − ω1 2 − − o + ω2 In the following we want to consider various applications of this formula. a process in which the electronic state remains unaltered after the scattering.228) S (0) = m S (0) = 2 m . Rayleigh scattering is deﬁned as the limit in which the wave length of the scattered radiation is so long that none of the quantum states of the electronic system can be excited.221) suggests to expand S( ω) S( ω) = S(0) + S (0) ω + Using 1 − o ± ω one can readily determine m (8. − o )2 ( m − o )3 (8.. .222) where S( ω) = 1 me 0|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ 0|ˆ2 · p |m m|ˆ1 · p |0 u ˆ u ˆ + m − o + ω m − o − ω 1 S (0)( ω)2 + .219) (8. in fact. Accordingly. for all states |m of the electronic system (8. Rayleigh Scattering We turn ﬁrst to an example of so-called elastic scattering. one assumes the even stronger condition ω1 << | o − m| .226) u ˆ 0|ˆ2 · p |m m|ˆ1 · p |0 u ˆ u ˆ 0|ˆ1 · p |m m|ˆ2 · p |0 u ˆ − me ( m − o )2 me ( m − o )2 (8.223) m Condition (8. (8.225) S(0) = m u ˆ 0|ˆ1 · p |m m|ˆ2 · p |0 u ˆ u ˆ 0|ˆ2 · p |m m|ˆ1 · p |0 u ˆ + me ( m − o ) me ( m − o ) (8. one can relate (8.234 Interaction of Radiation with Matter The factor N1 c/V can be interpreted as the ﬂux of incoming photons. ω1 = ω2 . consequently.e. it follows 2 dσ = ro (N2 + 1) dΩ2 |ˆ1 · u2 − S( ω) |2 u ˆ (8. .

222) of the cross section for Rayleigh scattering dσ = 2 ro m2 ω 4 (N2 + 1) dΩ2 × e (8.224. honey bees.222).8: Two-Photon Processes 235 ˆ These three expressions can be simpliﬁed using the expression (8.226) into S(0) = 1 i ( 0|ˆ1 ·r |m m|ˆ2 · p |0 − 0|ˆ2 · p |m m|ˆ1 ·r |0 ) u u ˆ u ˆ u m .233) m Employing again (8.182) for p and the expression (8. the blue color of the sky and the polarization pattern in the sky which serves many animals. (8.229) both to u1 · p and to u2 · p which results in ˆ ˆ ˆ ˆ me S (0) = 2 ( 0|ˆ2 ·r |m m|ˆ1 ·r |0 − 0|ˆ1 ·r |m m|ˆ2 ·r |0 ) .108) yields S (0) = me 2 0| [ˆ2 ·r.229) (8.8. i.232. Expression (8.234) where we used for the second identity the fact that u1 ·r and u2 ·r commute.182) 0|ˆ1 · p |m u ˆ 1 = 0|ˆ1 ·r |m u me ( m − o ) i This transforms (8. u1 ·r ] |0 = 0 u ˆ (8. For this purpose we replace p using (8.227) vanishes.226).235) We can now combine eqs.e. m|ˆ1 · p |0 u ˆ 1 = − m|ˆ1 ·r |0 u me ( m − o ) i (8.230) According to (8. this term cancels the u1 · u2 term in (8. .234. ˆ ˆ S (0) given in (8. u u u u (8.222). ˆ We want to simplify ﬁrst (8. It explains. pk ] = i δjk yields ﬁnally ˆ S(0) = 1 S(0) = i 3 3 (8.. for example.236) 2 × m 0|ˆ∗ ·r |m m|ˆ2 ·r |0 u1 u 0|ˆ∗ ·r |m m|ˆ1 ·r |0 u2 u + m − o m − o We have applied here a modiﬁcation which arises in case of complex polarization vectors u which ˆ describe circular and elliptical polarizaed light. 8. ˆ ˆ We want to prove now that expression (8. as a compass. pk ]|0 = u u ˆ j.232) Obviously. u ˆ ˆ ˆ ˆˆ i The commutator property [xj .231) (ˆ1 )j (ˆ2 )k 0|[xj .k=1 j.k=1 (ˆ1 )j (ˆ2 )k δjk = u1 · u2 u u ˆ ˆ (8.228) provides then the ﬁrst non-vanishing contribution to the scattering cross section (8. 8. Using again (8.108) this is 1 0|ˆ1 ·r u2 · p − u2 · p u1 ·r |0 .108) for the identity operator. For this purpose we apply (8.229) both for the u1 · p and the to u2 · p terms in (8.228) we obtain ˆ ˆ ˆ ˆ S (0) = 2me 2 m 0|ˆ2 ·r |m m|ˆ1 ·r |0 u u 0|ˆ1 ·r |m m|ˆ2 ·r |0 u u + m − o m − o (8. 8.235) and obtain the leading contribution to the expression (8.236) is of great practical importance.

240) cosθ2 We choose for the polarization of the incoming radiation the directions along the x1 . u2 = u2 ˆ ˆ ˆ ˆ cos2 φ2 cos2 θ2 sin2 φ2 for u1 = u1 .237) The resulting scattering is called Thomson scattering. the emitted u ˆ radiation is decribed by the wave vector sinθ2 cosφ2 k2 = k1 sinθ2 sinφ2 (8. u2 = u2 ˆ ˆ ˆ ˆ (1) (2) (2) (1) (1) (2) (1) (2) (2) (1) needs to be (8. however. (8. hence. we choose for the polarization of the emitted radiation two perpendicular directions u2 ˆ (2) and u2 which are also orthogonal to the direction of k2 . though. Since u2 ˆ ˆ (1) orthogonal to k2 as well as to u2 the sole choice is ˆ cosθ2 cosφ2 (1) k2 × u2 ˆ (2) u2 = ˆ = cosθ2 sinφ2 (1) |k2 × u2 | ˆ −sinθ2 The resulting scattering cross sections 2 dσ = ro (N2 + 1) dΩ2 × for the various choices of polarizations are sin2 φ2 for u1 = u1 . (8.244) . ao Bohr radius . u2 = u2 ˆ ˆ ˆ ˆ (8.243) cos2 θ2 cos2 φ2 for u1 = u1 . u1 = 1 ˆ ˆ (8.240).236 Thomson Scattering Interaction of Radiation with Matter We consider again elastic scattering. |n = |0 and ω1 = ω2 = ω in (8.. u ˆ 1 ao .241) 0 0 Similarly.e. assume now that the scattered radiation has very short wave length such that ω >> | o − m| . We want to assume.242) |k2 × k1 | 0 where the second identity follows readily from k1 = e3 and from (8.220) 2 dσ = ro (N2 + 1) dΩ2 |ˆ1 · u2 |2 . i. that the dipole approximation is still valid which restricts the applicability of the following derivation to k1 >> One obtains immediately from (8.238) (8. for all states |m of the electronic system .. u2 = u2 ˆ ˆ ˆ ˆ for u1 = u1 . To evaluate |ˆ1 · u2 |2 we assume that k1 is oriented along the x3 -axis and.239) We will show below that this expression decribes the non-relativistic limit of Compton scattering.220).and the x2 -axes 1 0 (1) (2) u1 = 0 . The ﬁrst choice is ˆ sinφ2 k2 × k1 (1) u2 = ˆ = −cosφ2 (8.

one speaks of Stokes scattering. one speaks of anti-Stokes scattering. ˆ In case that one measures the scattered radiation irrespective of its polarization. φ(vibr. the resulting scattering cross section is 2 ro (N2 + 1) ( 1 + cos2 θ2 ) dΩ2 . e.. Since the scattering is inelastic. i. but most often involves other degrees of freedom.246) 2 This expression is the non-relativistic limit of the cross section of Compton scattering.)n . in case of a diatomic molecule |n = |φ(elect. In the case that the scattering excites other than electronic degrees of freedom.247) ( 1 − cosθ2 ) (8.247.247) becomes identical with the Thomson scattering cross section (8. In the case of such scattering an electron system absorbs and re-emits radiation without ending up in the initial state.e. the ﬁrst term in (8. deﬁned in (8. molecular vibrations or crystal vibrational modes. The Compton scattering cross section which is derived from a model which treats photons and electrons as colliding relativistic particles is dσ tot = dσ tot where (rel) = 2 ro (N2 + 1) 2 ω2 ω1 2 ω1 ω2 + − sin2 θ2 ω2 ω1 dΩ2 (8.)n . Such scattering is called Raman scattering. the scattering is termed Brillouin scattering. the scattered radiation is polarized to some degree..245) 1 cos2 θ for u = u(2) ˆ ˆ 2 2 2 2 The result implies that even though the incident radiation is unpolarized. the states |n etc. for c → ∞ the Compton scattering cross section (8.e.246).8. If energy is absorbed by the system.248) me c2 One can readily show that in the non-relativistic limit.8: Two-Photon Processes 237 In case that the incident radiation is not polarized the cross section needs to be averaged over (1) (2) the two polarization directions u1 and u1 . or in case of translational motion of molecules in liquids. The excitation can be electronic. 8.. Raman Scattering and Brillouin Scattering We now consider ineleastic scattering described by the Kramers-Heisenberg formula. if energy is released.g.220) vanishes and one obtains in case of Raman scattering ω2 2 dΩ2 | u2 · R · u1 |2 ˆ ˆ (8. i. The energy deﬁcit is used to excite the system. The radiation scattered at right angles is even completely polarized (1) along the u2 -direction. One obtains then for the scattering cross section of ˆ ˆ unpolarized radiation (1) 1 for u2 = u2 ˆ ˆ 2 2 dσ = ro (N2 + 1) dΩ2 × (8. In case that the nuclear degrees of freedom excited absorb very little energy. For electronic systems in molecules or crystals the degrees of freedom excited are nuclear motions. (8.220) represent actually electronic as well as nuclear motions. as in the case of excitations of accustical modes of crystals.249) dσ = ro (N2 + 1) ω1 −1 −1 ω2 − ω1 = .

a case called resonant Raman scattering.238 where R represents a 3 × 3-matrix with elements Rjk ω2 = = 1 me Interaction of Radiation with Matter m n n| pj |m m| pk |0 ˆ ˆ n| pk |m m| pj |0 ˆ ˆ + m − o + ω2 m − o − ω1 − o )/ (8.g. Of course. the Raman scattering cross section will be much enhanced.. . Nevertheless. no singlularity developes in such case due to the ﬁnite life time of the state |m . ω1 ≈ m − o for a particular state |m . the cross section for resonant Raman scattering can be several orders of magnitude larger than that of ordinary Raman scattering. In case that the incoming photon energy ω1 is chosen to match one of the electronic excitations.k xj Rjk yk . e. a property which can be exploited to selectively probe suitable molecules of low concentration in bulk matter.251) ω1 − ( We deﬁne x · R · y = j.250) (8.

of the j-th particle. xN |Ψ . rj and σj denote position and spin. (9.. mainly those tools which are connected with eﬀective single–particle descriptions. 1 respectively. We will assume systems composed of N particles. the quantum Hall eﬀect. however. This has the consequence that one cannot distinguish between states which diﬀer only in a permutation of particles. i. ﬁnally. σj ). . In fact. Permutations The essential aspect of the systems described is that the particles are assumed to be identical. e.e. for example. . collective excitations in nuclei. The wave function in the space–spin representation is then Ψ(x1 . In the following we will introduce the rudimentary tools.1 Permutation Symmetry of Bosons and Fermions We seek to determine the stationary states of systems of many identical particles as described through the stationary Schr¨dinger equation o H |Ψ = E |Ψ . atoms and molecules. The interaction among many identical particles gives rise to a host of fascinating phenomena. .e. superconductivity. as electrons in crystals. A description of the mentioned phenomena requires a suﬃcient account of the interactions among the particles and of the associated many–particle motions. i. as protons and neutrons in nuclei and. relativistic descriptions. The spin variable for electrons. quantum theory dictates that all such states are identical. The latter systems require.2) 239 .g. the particle index j runs from 1 to N .. 9. x2 . as vibrations and combination of electrons and phonons in crystals. . as photons in the electromagmetic ﬁeld. as quarks in mesons and baryons..1) Here |Ψ represents the state of the system which in the following will be considered in a representation deﬁned by space–spin–coordinates xj = (rj . Systems to which this chapter applies appear in many disguises.Chapter 9 Many–Particle Systems In this chapter we develop the quantummechanical description of non-relativistic many–particlesystems. molecules and atoms. is σj = ± 2 . . x2 . are not counted (9. . xN ) = x1 .

the set of all permutations of N objects. . . 5) T (5. 5) in case of P2 . . Transpositions denoted by T (j. Denoting by P (j) the image of j.. xN ) = Ψ(xP (1) . the top row representing the numbers 1 to N labelling the N particles under consideration. This property will be established now.5) . i. However. even though these properties. The group SN is then composed of two disjunct classes of even and of odd permutations.e. i. (9. .3) P2 which aﬀects only two particles. . . . . xP (2) .. 5) (9. one can state P Ψ(x1 . .4) both permutations being obviously odd. . (j. One calls permutations of the ﬁrst type even permutations and permutations of the second type odd permutations.2). . k). xj .e. is called a transposition. i. a given P can either only be presented by even numbers of transpositions or by odd numbers of transpositions. Examples for an eight particle system are 1 2 3 4 5 6 7 8 1 2 4 5 6 3 8 7 1 2 3 4 5 6 7 8 1 2 3 5 4 6 7 8 P1 = . 6) T (6. in notation (9. k) of the two particles transposed. 4) T (4. x2 . The latter factorization has the essential property that the number of factors is either even or odd.e. rather only states which obey a certain transformation property under permutations are acceptable.3) the j-th index in the second row. P2 = . and explicitly the fact that any P ∈ SN can be given as a product of transpositions T (j. In the following sections we will then introduce the operators of second quantization which provide a means to ascertain that manipulation of wave functions always leave the transformation property under permutations uncompromised. number k under the entry j of the ﬁrst row indicating that particle j is switched with particle k. For the following we will require only two properties of SN . particles 4 and 5. . The product P1 P2 of even and odd permutations P1 and P2 obey the following multiplication table: P2 \ P1 even odd even even odd odd odd even We ﬁnally determine the action of permutations P on the wave functions (9. . We will not discuss here at any depth the properties of the permutation groups SN . The permutations we need to consider for a system of N particles are the elements of the group SN .. . the representations in the space of N –particle wave functions are extremely useful in dealing with N –particle systems. not any state can be chosen as to represent the many– particle system. i. For the permutations P1 and P2 given above holds P1 = T (3. The elements P ∈ SN can be represented conveniently as 2 × N matrices. 3) T (7. but in a diﬀerent order. . xP (j) . A reader interested in the quantum mechanical description of N particle systems is strongly encouraged to study these representations. (9. and the second row showing the numbers 1 to N again. 5). P2 = T (4. k) are characterized by the two indices (j. We can then state P2 = T (4.. leaving all other particles unchanged.240 Many–Particle Systems repeatedly like degenerate states. xP (N ) ) . . namely implicitly the group property. in particular. .e. k) = (4. . 8) . . .

−1.e. usually a very large number. xk ) is symmetric in its two arguments. From this property follows that H is independent of any speciﬁc assignment of indices to the particles (note that (9. k = 1.-1}.e.1: Bosons and Fermions Permutation Symmetry 241 A second consequence of the identical character of the particles under consideration is that the Hamiltonian must reﬂect the permutation symmetry. 2. The ﬁrst case applies to bosons. {1} (trivial) and {1. it is a postulate of quantum mechanics that for any description of many–particle systems the permutation operators must be diagonal. Presently. .. The boson and fermion property stated implies that for a system of N bosons holds P |Ψ = |Ψ For fermions holds P |Ψ = where P P ∀P ∈ SN .. a diagonal representation can only berealized in a simpler group. . all particles being governed by the same interactions.5) implies that a permutation eﬀectively changes the indices of the particles) and..9) = 1 for P even −1 for P odd (9. in turn.e.e. In fact. implies that the permutation operators can be chosen in a diagonal representation.e. hence.. i. i. i. the second group to fermions.6) which describes one–particle interactions and two–particle interactions of the system. The single particle states are assumed to be orthonormal. particles with integer spin values. Obviously. In fact. for the scalar product holds j|k = δj k . S where S denotes the number of available single particle states. From the latter results [H. . i. Fock-Space Our derivation will start by placing each particle in a set of S orthonormal single particle states x|k = φk (x). the Hamiltonian must be of the type N H = j=1 1 F (xj ) + 2 N G(xj .11) . G(xj . ∀P ∈ SN . Since permutations do not necessarily commute. ) are the same for all particles and for all pairs of particles. the representation is either isomorphic to the group composed of only the ‘1’.k=1 (9. (9. P ] = 0 . xk ) j. it is essential that the functions F ( ) and G( .10) We will construct now wave functions which exhibit the appropriate properties. (9. i.8) |Ψ (9. provide a representation of the group structure represented by the multiplication table of even and odd permutations given above. The terms of the Hamiltonian will be discussed further below. In fact. particles with half–integer spin values. 1 together with multiplication. In the latter case all even permutations are represented by ‘1’ and all odd permutations by ‘-1’. (9.9. or the corresponding group of two elements 1. the Hamiltonian for permuted indices P HP −1 is identical to H. both groups.7) This property.

. . . . . . T (j. . the nj specify how often a single particle state |j is occupied. xN |ΨFock (λ1 . . . This detail is connected with the fact that a two–particle state of the type ΨFock (λ. λ2 . 9. . which do not yet obey the symmetries (9. . .11. .e. e. each particle j into a speciﬁc state |λj . hence. λ2 . λN ) = j=1 φλj (xj ) .15) (nj !) N! 1 2 P ∈Λ(λ1 . λ2 . (9. The numbers nj . . . referred to as the occupation numbers.. λ2 . the so-called Fock-states. with λj = λk = . |λ2 . λN ) only if λj = λk . nS ) = S j=1 (9.. .14) Obviously. It holds N ΨFock (λ1 . . k. i. n2 . λ2 . . . .9). . . . therefore. . . . x2 . xN |ΨB (n1 . ...λN ) P x1 .13) These states form an orthonormal basis of N -particle states.15) completely and. λN ) do not exhibit the symmetries dictated by (9. |λN . nS ). k) ∈ Λ(λ1 . A wave function which obeys the boson symmetry (9. ..e. x2 . . λj ∈ {1. .λN ) Here the states (λ1 .15) with λk = j.12) where the second factor represents the scalar product between spin states. An important detail of the deﬁnition (9. .g. . . S} of the type (9. . . The integers nj are equal to the number of λk in (9. . (9. λN ) = j=1 δλj λj .12).λN ) on the rhs. are essential. appear severalfold in (9. . . Wave Functions with Boson Symmetry in the Occupation Number Representation Wave functions with the proper symmetries can be obtained as linear combinations of Fock states. λ) in which two particles occupy the same single– particle state |λ does not allow transposition of particles. Λ(λ1 . λN )|ΨFock (λ1 .15) is that the sum over permutations does not involve permutations among indices j.242 Many–Particle Systems The scalar product involves spin as well as space coordinates and is explicitly given by j|k = d3 r φ∗ (r) φk (r) · σj |σk j (9. . i.. .8) is x1 . 9. λ2 . .9). . x2 . . . . .15). . n2 . 2. is a particular choice of placing N particles consistent with the occupation numbers (n1 . xN |Ψ Fock (λ1 . . . . λN ) ⊂ SN is the set of all permutations involving only particles j with diﬀerent λj . . . 9. . In a second step we form linear combinations of Fock states with the desired symmetries. the states |ΨFock (λ1 . . . The reader is well advised to pause and grasp the deﬁnition of the nj . . .8. . N x1 . . . We ﬁrst consider orthonormal many–particle wave functions.8. . the reason is that such transposition duplicates the state which would. characterize the wave function (9. . The Fock states represent N particles which are placed into S states |λ1 .

. for two sets of occupation numbers N = (n1 . x2 .1. . . √1 N! (9. . nS ) holds S Ψ (N )|Ψ (N ) = j=1 B B δnj nj . . for example. .9). the integers nj denote the occupancy of the single–particle state |j . .. n2 . φλN (xN ) Here. x2 . Such wave function is x1 . . xN |ΨF (n1 . n2 . x2 . nS ) and N = (n1 . .17) Here the states (λ1 . . (9.e. φλ2 (x1 ) φλ2 (x2 ) . . . . for example. . . correspond to a particular choice of placing N particles consistent with the occupation numbers (n1 . . . .1: (a) Show that the states (9.20) . .16) Exercise 9. . . .15) describe. For any value nj > 1 two or more of the columns of the Slater matrix are identical and the wave function vanishes. i. the photons of the electromagnetic ﬁeld or the phonons in a crystal. for two sets of occupation numbers N = (n1 . These states form an orthonormal basis. (9. λ2 . Wave Functions with Fermion Symmetry in the Occupation Number Representation One can construct similarly a wave function which obeys the fermion symmetry property (9. nS ) holds S ΨF (N )|ΨF (N ) = j=1 δnj nj . .16). xN |ΨFock (λ1 . n2 . .1: Bosons and Fermions 243 The wave functions (9. . n2 . . . . The important property holds that nj can only assume the two values nj = 0 or nj = 1. nS ). . . . φλN (x1 ) φλN (x2 ) . . . . a molecule or a crystal. . . . nS ) = √1 N! P ∈SN P P x1 .15) holds (9. .8). . . λ2 .15) obey the boson symmetry (9. (b) Show that for the states (9.. ..18) this wave function can also be expressed through the so-called Slater determinant x1 ..9. nS ) = φλ1 (x1 ) φλ1 (x2 ) . i. . . . n2 . . These states form an orthonormal basis. . .19) φλ1 (xN ) φλ2 (xN ) . . . xN |ΨF (n1 . . . n2 . Using the identity governing the determinant of N × N –matrices A N det(A) = P ∈SN P j=1 Aj P (j) (9.λN ) (9.. . The fermion states (9. . . . .17) describe. n2 . nS ) and N = (n1 . .e.λN ) on the rhs. . electrons in an atom..

.9). .17) is commonly referred to as the occupation number representation since the wave functions are uniquely speciﬁed by the set of occupation numbers N = (n1 . .23) (9. . . Exercise 9. These operators allow one to construct and manipulate many–particle wave functions while preserving permutation symmetry. nj . nS ) = j nj + 1 ΨB (n1 . . .1. nS ) .244 Many–Particle Systems The representation of wave functions (9. To prove that (9.22) follows from (9. ak ] = 0 . n . . . n − 1.21) we consider the matrix Aj corresponding to aj in the basis of many–particle states (9.15). . [a† . . a† ] = δj k .24) [aj . .25) . are the so-called operators of 2nd quantization. . n ) are normalΨ 1 j 1 j S S ized according to (9. .16) one obtains (Aj )N = ΨB (N )| aj |ΨB (N ) = √ S N nj δnj nj −1 k=1 k=j δnk nk . . . . n2 . . . . since both the N -particle wave function B (n . . . k The operators a† and aj are referred to as the creation and annihilation operators of bosons for the j orthonormal single–particle states |j . a† ] = 0 j k (9.21) √ The factor nj appears here on the rhs. . nj − 1. . For the adjoint operator a† holds j a† ΨB (n1 .15). 9. . . nS ) nj ≥ 1 0 nj = 0 √ . .17) obey the orthonormality property (9. . . . . . nj + 1. Creation and Annihilation Operators for Bosons We consider the operator aj deﬁned through aj ΨB (n1 . . . nS ) = nj ΨB (n1 . . . (9. . .15) and (9. nj .22) The operators aj and a† obey the commutation properties j [aj . . . . . . (9. .2: (a) Show that the states (9.17) obey the fermion symmetry (9. guaranteeing the correct permutation properties of the many–particle states. . (b) Show that the states (9. nS ). . . .20).2 Operators of 2nd Quantization An important tool to describe mathematically systems of many bosons or fermions. Employing the superindices N and N introduced above and using the orthonormality property (9. . . . n ) and the N − 1-particle wave function ΨB (n . . (9.

. .29) |ΨB (n1 . nS ) = j=1 nj |ΨB (n1 . One can readily show using (9.2: Operators of 2nd Quantization Let A† be the matrix adjoint to Aj . nS = 0) as follows S (9. . . . the properties (9.24). . i. j (9. . .34) = . It holds for j = k (aj a† − a† aj )|Ψ(N ) = (nj + 1 − nj )|Ψ(N ) = |Ψ(N ) .21.31) This operator is diagonal in the occupation number representation. .e.27) Equations (9.29) yield the commutation relationship (9.22) ˆ Nj |ΨB (N ) = nj |ΨB (N ) . .22) follows from (9. .30) Of particular interest is the operator ˆ Nj = a† aj . . nk + 1 . . 9. We will prove the commutation properties (9. . .26) From this result one can immediately conclude that (9. . . obviously. . a† ] = 1 For j = k follows similarly 1. .. n2 = 0.23) follow in an analogous way. One obtains for its matrix elements j A† j N N 245 = ΨB (N )| a† |ΨB (N ) = (Aj )N N j S k=1 k=j nj δnj nj −1 δnk nk = nj + 1 δnj nj +1 S k=1 k=j δnk nk .21) and vice versa. (9.. nS ) = j=1 a† j nj nj ! |0 . j (aj a† − a† aj )|Ψ(N ) = k k ( nj (nk + 1) − (nk + 1)nj )|Ψ(n1 . (9. . nS ) . . . the eigenvalues of Nj are the occupation numbers nj . n2 .24). (9. . One refers to Nj as the occupation number operator. (9. n2 . The creation operators a† allow one to construct boson states ΨB (N ) from the vacuum state j |0 = |ΨB (n1 = 0. nS ) = |ΨB (N ) . . . nj − 1 . . . 9. nS ) N |ΨB (n1 . j j Since this holds for any |Ψ(N ) it follows [aj . nS ) = 0 . S ˆ N |ΨB (n1 . for basis states |ΨB (n1 .33) is called the particle number operator since.9. .32) ˆ ˆ i. .28) (9. n2 .27.e. (9. n2 . The related operator S ˆ N = j=1 ˆ Nj (9. . n2 . .

. .29. . . . . . one needs to apply only the commutation propertes (9. operators which can alter the occupancy of the wave function (9. 9.. . . xN |ΨF (n 1 .. from the vacuum state |0 or states |ΨB (N ) . As long as one starts from a wave function with the proper boson symmetry. . nj+1 = 1. Before proceeding in this respect we need to account for the following property of the fermion wave function which applies in the case nj . Exercise 9. x2 . nS ) .23. ..2. . . of (9. . . in general.. . . . x2 . nj . one does not need to worry ever about proper symmetries of wave functions. i. nS ) = − x1 .1: The commutation relationships (9. nS ) = .35) and where the derivative operation is deﬁned through ∂f (A) = ∂A ∞ n=1 1 f (n) (0) An−1 . ..g.2. .17. . . . xN |ΨF (n1 . f (aj )] j = ∂f (a† ) j ∂a† j ∂f (aj ) ∂aj = − where the operator function is assumed to have a convergent Taylor expansion ∞ f (A) = n=0 1 (n) f (0) An n! (9.24) and aj |0 = 0 imply that the the properties (9. φj+1 (x1 ) φj+1 (x2 ) . . . Obviously.. Prove this by induction. √1 N! .19) without aﬀecting the fermion symmetry (9. 9. . nj+1 . in case it holds for nj .. . .30). φj (xN ) φj+1 (xN ) . 9. . .21. i. since they are induced through the algebra of a† and aj . 9. nj (9. (9. j Exercise 9.. φj (x1 ) φj (x2 ) . .. 9. nj+1 . .37) (9. . . .38) . xN |ΨF (n1 . . e. nj . .36) Creation and Annihilation Operators for Fermions We want to consider now creation and annihilation operators for fermions..246 Many–Particle Systems In using the boson creation and annihilation operators. in case that the single particle states |j and |j + 1 are both occupied.23. 9.24) and the property aj |0 = 0. x1 . To prove this property we notice that the l.22. . nj+1 .e. showing that the property holds for nj = 0 and. x2 . the fermion wave function changes sign when one exchanges the order of the occupancy.h.9). .38) corresponds to x1 .e. .s. (n − 1)! (9. it also holds for nj + 1. . f (a† )] j [a† . .39) .2: Show that the boson operators a† and aj satisfy j [aj .23) hold for the state deﬁned through (9.. .

nS ) = nj (−1)qj |ΨF ( n1 .s. x2 . . must be chosen once and for all at the beginning of a calculation and these states must be occupied always in the order of increasing indices. . nS = 0) φ2 (x1 ) φ5 (x1 ) 1 1 = √2 = √2 [φ2 (x1 )φ5 (x2 ) − φ5 (x1 )φ2 (x2 )] . φ2 (x2 ) φ5 (x2 ) (9. the rhs. . .. . . nj = 0. in order to be consistent with this convention.19).e. .. . . . . i. .. n3 . . For this purpose one must adhere to a strict convention: the labelling of single–particle states by indices j = 1. to move the particle to the ﬁrst position (to be annihilated there) or to move it from the ﬁrst position to its canonical position (after it had been created at the ﬁrst position). 2. .40) .19). . . nS ) qj states occupied In case that the single particle state |j is not occupied. A proper example is the two particle fermion wave function x1 . .e. . of (9. . . it is necessary to specify for a fermion wave function the order in which the single– particle states |λj are occupied. nS ) as follows: j−1 qj = k=1 nk . .2: Operators of 2nd Quantization The r. We are now ready to deﬁne the annihilation operator cj for fermions in the single–particle state |j as follows cj |ΨF (n1 . n2 . . .. nj−1 .. qj is the number of states |k with k < j which are occupied in a fermion wave function.h. . . . φj (x1 ) φj (x2 ) .9. n2 = 1. n2 . .. particles are being created or annihilated. nj+1 . however. the expressions (9.43) . We will be adopting below the notational convention that qj corresponds to occupancies N = (n1 . . n2 .. nj . nS ) = . nS ). nj+1 . i. φj+1 (x1 ) φj+1 (x2 ) . . 247 − √1 ! N . . x2 |ΨF (n1 = 0. Obviously. . . One adopts the convention that particles are created and annihilated by these operators always at the ﬁrst position of the wave function. ..e. . . i.41) Before we consider the deﬁnition of fermion creation and annihilation operators we need to take notice of the fact that one also needs to deﬁne at which position in the wave function. vanishes. .42) i.e. Because of the property of the determinant to change sign when two columns are interchanged.. . xN |ΨF (n1 .38) reads − x1 . (9. n5 = 1. at which column of the Slater determinant (9.. nj . n3 = 0. at the ﬁrst column of the Slater determinant (9. n6 = n7 = . .. . This requires one. . . nj → 0.40) are identical. . . φj+1 (xN ) φj (xN ) . (9. that occupancies are always ordered according to a monotonous increase of the single–particle state index.39) and (9. n4 = 0. .. . (9. n2 . . This change of position is connected with a possible sign change (−1)qj where qj is deﬁned for a given N = (n1 .

.46) [cj .48) From this follows (9. . . c† ]+ = δj k k where we have used the deﬁnition of the so-called anti-commutators [A.45) (9. The anti-commutation properties (9. . [c† .45) follow in an analogous way and will not be derived here. We consider (9. nj . The elements of Cj are then ( note that nj only assumes values 0 or 1) S (Cj )N N = Ψ (N )| cj |Ψ (N ) = nj (−1) δnj nj −1 k=1 k=j F F qj δnk nk . nj+1 . nS ) = j (−1)qj (nj + 1))(−1)qj (1 − nj ) |ΨF (N ) = (1 − nj ) |ΨF (N ) (9.48) vanishes. .44). . n2 . We have used here the deﬁnition qj = k<j nk along with qj = qj . n3 .47) Let C† be the matrix adjoint to Cj . . nj−1 . . nS ) = j j (−1)qj (1 − nj + 1)(−1)qj nj |ΨF (N ) = nj ΨF (N ) (9. nS ) = j Many–Particle Systems (−1)qj (1 − nj ) |ΨF ( n1 . . c† ]+ = 0 j k (9.19). (9.43). nS ) qj states occupied The operators thus deﬁned obey the commutation properties [cj . One obtains for its matrix elements j C† j NN = ΨF (N )| c† |ΨF (N ) = (Cj )N j S N = nj (−1)qj δnj nj −1 k=1 k=j δnk nk S = (−1) (1 − nj ) δnj nj +1 k=1 k=j qj δnk nk . B]+ = AB + BA. nj → 0 . . We will prove now the anti-commutation property (9. . the last factor on the rhs.49) The derivation involves realization of the fact that a non-zero result is obtained only in case nj = 1. n2 . 9. . otherwise. . Similarly one obtains cj c† |ΨF (N ) = cj (−1)qj (1 − nj ) |ΨF (n1 . . It holds c† cj |ΨF (N ) = c† (−1)qj nj |ΨF (n1 . . nj → 1 .44) (9. nj → 1.46) ﬁrst for the case j = k. We ﬁrst show that (9.50) . (9.248 The operator c† adjoint to cj exhibits the following property j c† |ΨF (n1 .17.46). ck ]+ = 0 .44) follows from the deﬁnition (9. . . of (9. Let Cj be the matrix corresponding to the operator cj in the basis of fermion states (9.

nk = 1 . is referred to as the occupation number operator.55) is called the particle number operator. λ λ λ (9.. 2 (9. nλ1 .56) Here we have made use of cj cj = 0 and c† c† = 0 which follows from nj ∈ {0. We j will often separate the states |j .49) and (9.e. j j j j j j j (9. On the r. . k k 249 (9. . i. i. . of this equation we meant to indicate only those N occupation numbers nλj which are non-vanishing.) = c† 1 c† 2 · · · c† N |0 . . σ 1 (σ = ± 2 ) .e. coordinates x and index j into a space part and a spin part. . j j We ﬁnally note that the creation operators c† allow one to construct any fermion state |ΨF (N ) j from the vacuum state |0 deﬁned as above (see (9. Correspondingly.52) hold for any |ΨF (N ) one can conclude (9.46). nλN .51) (9. the creation operators must operate in the proper order. .54) ˆ N = j=1 ˆ Nj (9.49) shows that the operator ˆ Nj = c† cj j (9.58) .50) yield ( cj c† + c† cj ) |ΨF (N ) = |ΨF (N ) . . in the basis |ΨF (N ) . for which purpose one needs to envoke only the algebraic (anti-commutation) properties (9. .. nS ) k c† cj |ΨF (N ) = (−1)qj +qk −1 nj (1 − nk )|ΨF (n1 . ( cj c† + c† cj ) |ΨF (N ) = 0 .51.53) is diagonal in the occupation number representation. nS ) k and. . j → j.h. 1 In the following we will apply fermion operators c† and cj only to electrons which carry spin 2 . . nj = 0 . 9. .57) On the l. . nj = 0 .s..2: Operators of 2nd Quantization (9. . . Equation (9. ˆ It is an interesting exercise to demonstrate that Nj only has eigenvalues 0 or 1. an operator cj must be on the left of an operator ck for j < k.h.46).29)) |ΨF (. i. . x|j → φj (r) | 1 σ .e. hence. S (9. 1}.52) One can obtain similarly the same relationship in the case j > k.s. . . . nλ2 .9. hence. . Since (9. j j For j < k one obtains cj c† |ΨF (N ) = (−1)qj +qk nj (1 − nk )|ΨF (n1 . with eigenvalues equal to the occupation numbers nj ˆ Nj |ΨF (N ) = nj |ΨF (N ) ˆ Nj . nk = 1 . The stated property ˆ follows from the idempotence of Nj which can be derived as follows † ˆ2 ˆ Nj = c† cj c† cj = c† (1 − c† cj )cj = c† cj − c† c† cj cj = cj cj = Nj . .

Operators of the type 1 ˆ G = 2 N g (xj .F (N ) . Since the many–particle states are built-up from a basis |r .6) above. . j = 1. respectively.15.59) and (9. (9.250 Many–Particle Systems 9. r = 1.k=1 j=k (9.17) and integrating over all particle space-spin coordinates x1 . i. respectively – which are all identical. S of single–particle spinorbital states (see page 241) and the operators are speciﬁed through the associated generic operators ˆ f and g .F (N )|G|ΨB. The action of the operators boson and many–fermion states |Ψ is obviously described through the many–particle state matrix elements ˆ ΨB. σr ) f (x) ψs (r. ˆ . 2.15. . ˆ r|f |s = 1 ∗ ˆ d3 r ψr (r.. These operators had been introduced already in Eq. xN . 9. .17).e.60) involves generic operators – acting on xj and on xj . 2.59) ˆ ˆ are called single–particle operators. An operator N ˆ R = j=1 rj (xj ) ˆ (9.F (N )|F |ΨB. The operators g (x. (9.60) is that the sum over particles in (9. xk ) ˆ j. . The essential aspect of deﬁnition (9.F (N ) . . We adopt a similar expression to distinguish two–particle operators G and g . 9.63) ˆ ˆ This expression has been speciﬁed for the purpose of these notes to distinguish F and f and is not common ˆ terminology. . . x ) which constitute G are ˆ called the associated generic two–particle operators. . The operators f (x) which constitute F are called the associated generic single–particle operators1 . This procedure is most tedious and becomes essentially impossible in the general case that many-particle basis functions are linear combinations of wave functions of the type (9.59. .59. .61) is not a one particle operator as long as the N operators rj (x). N are not all identical. 9. ˆ ˆ ˆ We seek to determine now how one–particle operators F and two–particle operators G act on many– B (N ) and |ΨF (N ) .3 One– and Two–Particle Operators Deﬁnition Operators acting on N particle systems of the type N ˆ F = j=1 ˆ f (xj ) (9. x2 . 9. ˆ ΨB. (9.62) These matrix elements can be obtained by chosing the operators and many-particle states in the position representation. xk .60) ˆ correspondingly are called two–particle operators. namely. σs ) (9. one expects that the matrix elements can actually be expressed in terms of matrix elements ˆ involving solely the generic operators and the single–particle states.59) and over pairs of particles in (9.

9. ∂r 2 (9. according to (9. g g (9.70) Both operators.68) Another single–particle operator is the one–particle density operator N ρ(r ) = ˆ˜ j=1 δ(r − rj ) . (9. that the symmetry implies that one can switch simultaneously the pairs of indices r.69) ˆ In this case the generic operator is f (x) = δ(r − r).64) We like to note an important symmetry of the matrix element r.71) . The aim of the present section is to express the matrix elements of single– and two–particle operators in a basis of many–particle states (9.63.k=1 j=k q2 |rj − rk | (9.64). is t(x) = (− 2 /2m)(∂ 2 /∂r 2 ). Examples of One– and Two–Particle Operators An example for a single–particle operator is the kinetic energy operator N ˆ T = j=1 2 − 2m ∂2 . The matrix elements of the generic operator ˜ in the one–particle basis are ˆ rσ|f |sσ = δσσ d3 r φ∗ (r) δ(r − r) φs (r) = δσσ φ∗ (r) φs (r) δσσ ˜ ˜ r r ˜ (9.62) in terms of the matrix elements of single–particle states (9. σu ) . ∂r 2 ∂x2 ∂x2 ∂x2 1 2 3 (9. s|ˆ|t. are spin-independent as is evident from the factor δσσ . The matrix elements of this operator in a single–particle basis are ˆ rσ|t|sσ = δσσ d3 r φ∗ (r) r 2 − 2m ∂2 φs (r) ..3: One– and Two–Particle Operators r. s|ˆ|t. ˜ (9. u = s. ∂rj2 (9. r|ˆ|u. i. σr )ψs (r .67) ˆ In this case the generic single–particle operator.e.66) Here we adopt the notation for the Laplacian ∂2 ∂2 ∂2 ∂2 = + + . ˆ 251 (9. An example for a two–particle operator is the Coulomb repulsion operator 1 ˆ V = 2 N j. t . u = g d3 r ∗ ∗ d3 r ψr (r.69).59). one cannot switch the indices of one pair individually.65) This symmetry will be exploited repeatedly below. s and t. u. u which follows directly g from the deﬁnition (9. s|ˆ|t.64) r. 9. σs ) g (x. Notice. x ) ψt (r. We will show that the boson and fermion creation and annihilation operators serve this purpose.66) and (9. σt )ψu (r .

. i.+ . Sj. However. α|ˆ+ |s. Sj is the spin operator for particle j with components (in the spherical representation) Sj.252 Many–Particle Systems which is also spin-independent.k=1 (9. α|ˆ3 |s. u. σ .e. β s r.− + S1. The operator N S = j=1 Sj (9. Here.+ S2. In this case the generic operator is v (x.3 . σ.3 2 2 (9. (9.75) is a two–particle operator. The generic operator is s12 = S1 · S2 = ˆ 1 1 S1.− . u.+ + S1. σ = 2 q φ (r )φ (r ) δ δ d3 r2 φ∗ (r1 )φ∗ (r2 ) r s |r1 − r2 | t 1 u 2 σσ σ σ d 3 r1 (9. α s r. s = 1 σ. not in the strict sense of our deﬁnition above since the restriction 2 j = k does not apply in the summation. σ | s12 |t.76) the single–particle state matrix elements of which are r. The non-vanishing matrix elements of the spherical components of s are (α = + 1 . σ . β = ˆ 2 2 1 −2) r. one can obviously extract the term j Sj to be left with a two–particle operator in the strict sense of our deﬁnition. s.74) and (9. however.72) As a ﬁnal example of one– and two–particle operators we consider operators of total spin. s. α s r. The generic operator is given through the Pauli matrices.77) are orbital-independent as evidenced by the factors δrs and δrt δsu in the respective formulas. x ) = q 2 /|r1 − r2 |. β|ˆ− |s.− S2. .77) δσ σ ( δσσ − δ−σσ ) The operators (9. β|ˆ3 |s.3 S2.73) is a one-particle operator.74) = N j=1 Sj = 2 N Sj · Sk j. σ. The ˆ matrix elements of the generic operator in terms of single–particle states are ˆ r. σ ˆ + 1 δ− 1 σ δ 1 σ δ 1 σ δ− 1 σ 2 2 2 2 2 = δrt δsu + 1 4 δσσ 1 2 δ 1 σ δ− 1 σ 2 2 δ− 1 σ δ 1 σ 2 2 + . σ | t |t. Sj. β s The operator S 2 = δrs = δrs = = + − 1 2 1 2 δrs δrs (9.

s ΨF (N )|cs ΨF (N ) = nr (−1)qr ns (−1)qs δnr −1 nr δns ns −1 δnt nt We refer the reader to Condon and Shortley’s ‘Theory of Atomic Spectra’.63) and (9. 9. u . 9.23. To show that the resulting values of the matrix elements (9. u a† a† at au bosons g r s † † r.F (N ) have the same values as the respective matrix elements (9. These rules will be provided below only for the case of fermions.k=1 j=k S r.46). g and not ct cu . Expressions (9. r ΨB (N )|a† a† at au |ΨB (N ) r s ΨF (N )|c† c† ct cu |ΨF (N ) r s bosons fermions (9.176 where the matrix elements of fermion states can be found.81) For the latter matrix elements simple rules can be derived from the algebraic properties of the boson and fermion operators (9. xk ) → ˆ j.79) have the following implication: The operators (9.78) and (9.u=1 (9.63.t.e.79) It is crucial to notice in the expression for the fermion two–particle operator that the order of the annihilation operators in (9.59) and G in (9.u=1 S r.s. i.62) through single–particle state matrix ˆ ˆ elements (9. namely cu ct .64) one replaces the operators F in (9.s. s|ˆ|t. u as deﬁned in (9.s=1 S r.62).62) are correct one needs to compare the result derived with conventional derivations of the matrix elements2 .45.78..78) 1 2 g (xj . r S Ψ (N In case of r = s one obtains F )|c† cr |ΨF (N ) r = nr r=1 δnr nr . s|ˆ|t.t.3: One– and Two–Particle Operators Deﬁnition in Terms of Creation and Annihilation Operators 253 In order to express many–particle state matrix elements (9. 9.24) and (9.60) as follows N ˆ f (xj ) → j=1 N 1 2 1 2 S r. r ΨF (N )|c† cs |ΨF (N ) . 9. respectively.83) t=1 t=r.9.64). .82) ΨF (N )|c† cs |ΨF (N ) = r cr 2 (9. s|ˆ|t.171 and pp. and then in g a second step the matrix elements ΨB (N )|a† as |ΨB (N ) .79) is opposite to that in the matrix element r.s=1 ˆ r|f |s a† as bosons r ˆ r|f |s c† cs fermions r r.79) in the basis of many–particle states |ΨB. pp. The Matrix Elements ΨF (N )|c† cs |ΨF (N ) r We consider ﬁrst the case r = s. The matrix elements are then actually those of the number operator c† cr which is diagonal in the occupation number representation. In order to determine these matrix elements one needs to evaluate ﬁrst the matrix elements ˆ of the generic operators r|f |s r. respectively. s|ˆ|t. u cr cs cu ct g fermions (9. (9.80) (9.

where each contribution corresponds to one of the three cases mentioned. 1 2 S r s|ˆ|s r c† c† cr cs .s=1 a. c2 = 0 r (9.i. i. ˆ ˆ Nr Ns .d.’ (all indices diﬀerent) that only tuples (r..d.i.s=1 a. s. Case 2 all four indices are diﬀerent. t. (r.86) into three contributions.88) .d.53) 1 ˆ G0 = 2 S r s|ˆ|r s − r s|ˆ|s r g g r. The Matrix Elements ΨF (N )|c† c† ct cu |ΨF (N ) r s Before we determine these matrix elements we will investigate which possible combination of indices r.78) does yield the same results as the operator deﬁned in (9. t. u) are included in the summation for which all indices are diﬀerent.u r=s. s.87) Here and in the following we denote by ‘a. we split the sum in (9.. t=u 2 = 0. ˆ ˆ ˆ ˆ Accordingly.85) All other matrix elements vanish. Case 1 three of the four indices are diﬀerent. r = t etc. The anti-commutation property (9.e.254 From (9.84) we can conclude that for diagonal elements holds S Many–Particle Systems ˆ ΨF (N )|F |ΨF (N ) = r=1 ˆ nr r|f |r (9. 9. for which holds r = s.i.s=1 a.84) The oﬀ-diagonal elements with nr = nr except for r = s. g r s r. i.s.79) and using c† r we can write S 1 ˆ G = rs|ˆ|tu c† c† cu ct g r s 2 r. s).86) For the combination of indices r. t.t.45) yields together with the deﬁnition of the occupation number operator (9.i.d. u need to be considered. (9.82. u in this sum three possibilities exist Case 0 two of the four indices are diﬀerent. s. ˆ The ﬁrst contribution G0 is 1 ˆ G0 = 2 S r s|ˆ|r s c† c† cs cr + g r s r. those for which N and N diﬀer in more than two occupation numbers. t). in which case ns = ns + 1 and nt = nt − 1 holds.e. t. Comparision with the results in Condon&Shortley (pp. (r. (9. Starting from the deﬁnition (9.59) when the matrix elements are determined between Slater determinant wave functions. namely. G = G0 + G1 + G2 . s. are ˆ ˆ ΨF (N )|F |ΨF (N ) = (1 − ns )(−1)qs nt (−1)qt s|f |t (9. 171) shows that the operator deﬁned in (9.

t=1 a.t>u r s|ˆ|t u − r s|ˆ|u t g g c† c† cu ct .s r.u r>s.t. S r. N = N or nr = nr for all r. this contribution obviously behaves similarly to the oﬀ-diagonal part (9.45).s.s. S r. s|ˆ|r.t=1 a. Commutation of ct cu according to (9.d.i. A similar contribution does not arise for F . employing the symmetry property (9.t=1 a.u r>s. nt .t.u r<s.t=1 a. g r s (9. s (9. ˆ The second contribution G1 is ˆ G1 = + + + 1 2 1 2 1 2 1 2 S r.d.s.86). r s|ˆ|r t − r s|ˆ|t r g g ˆ c† ct Nr .65).t.s.89) S r. r g g .s. r s|ˆ|r t c† c† ct cr g r s r s|ˆ|t s c† c† cs ct g r s r s|ˆ|s t c† c† ct cs g r s r s|ˆ|t r c† c† cr ct . r s (9. ˆ Case 0 This case covers diagonal matrix elements. ˆ A similar series of transformations allows one to express G2 which can be written ˆ G2 = + + + as follows S 1 2 1 2 1 2 1 2 r.d.t<u S r s|ˆ|t u c† c† cu ct g r s r s|ˆ|t u c† c† cu ct g r s r s|ˆ|t u c† c† cu ct g r s r s|ˆ|t u c† c† cu ct g r s (9.s.s.t. F (N )|G|ΨF (N ) .i. One obtains 1 ˆ ΨF (N )|G|ΨF (N ) = 2 S nr ns r. (9. given in (9.t>u S r.90) ˆ Since Nr is diagonal. For this purpose we consider three ˆ We can state now the matrix elements Ψ cases which actually correspond to the three cases considered below Eq.3: One– and Two–Particle Operators 255 Obviously.9.84) ˆ of the one–particle operator F . Only G0 . this contribution to the two–particle operator is diagonal in the occupation number ˆ representation. this part accounts for all diagonal contributions to G.t<u S r. s − r.s.u r<s.88). (9.d. i.t=1 a.i. ns ..d. s|ˆ|s.93) .t. relabelling ˆ summation indices and using c† c† ct cr = c† c† cr ct = c† ct Nr yields r s s r s S ˆ G1 = r.91) ˆ G2 = r. As we will ﬁnd.92) ˆ This contribution to G has non-vanishing matrix elements only when N diﬀers from N in four ˆ occupation numbers nr .s. contributes in this case.t>u S r.i.s.u r>s. nu .e.i.

Derive the non-vanishing matrix elements (9. r=1 r=1 Exercise Exercise Exercise Exercise Exercise 9. One obtains ˆ ΨF (N )|G|ΨF (N ) = ± (−1)qs +qr +qu +qv (1 − ns )(1 − nr )nu nv r.256 Many–Particle Systems Case 1 This case covers oﬀ-diagonal elements with nr = nr except for r = s. Spin Operators ˆ ˆ In this section we will brieﬂy consider the spin operators S given in (9. u g g (9.1: 9. Only G1 . u > v. One obtains for the three components of S ˆ S3 ˆ S+ ˆ S− = 1 2 S ˆ ˆ Nrα − Nrβ = = r=1 S c† crβ rα r=1 S c† crα rβ r=1 (9.92). u. nu = nu − 1. nt = nt + 1. u > v or s > r.4: 9. s|ˆ|v.92) from (9. particle number is conserved. u < v.91). Only G2 . given in (9.. Derive (9. contributes in this case.e.3. t − r. s|ˆ|u.89).73) and S 2 given in (9. s|ˆ|r.62) for bosons.90). u < v or s > r.3.3: 9. One obtains ˆ ΨF (N )|G|ΨF (N ) = (−1)qs +qt (1 − ns )nt r r=s. t. The matrix elements of the corresponding generic operators are (9.95) from (9.95) In the latter formula the ‘+’-sign applies for s < r.2: 9. Derive (9. We like to express these operators through ˆ fermion creation and annihilation operators.94) Case 2 This case covers oﬀ-diagonal elements with nr = nr except for r = s. the ‘−’-sign applies for s < r.76). for ˆ non-vanishing matrix elements ΨF (N )|G|ΨF (N ) at most four occupation numbers nr and nr can diﬀer.90) from (9. s|ˆ|t. i. nv = nv − 1.t nr r.74) and (9. it holds S nr = S nr .92). Furthermore. ˆ ˆ Show that the matrix elements of F and of G conserve particle number. All matrix elements which are not covered by the three cases above vanish. v for which holds ˆ ns = ns + 1. respectively. In particular.5: Derive (9.3. v − r.3.75) which are. given in (9. contributes in this case.3. r g g (9.96) . t for which holds ˆ ns = ns + 1 and nt = nt − 1. one–particle and two–particle operators.

4 Independent-Particle Models In the remaining part of this chapter we will apply the technique of operators of 2nd quantization to the description of many–fermion systems as they arise. In all these systems the Hamiltonian is a sum of one– and two–particle operators S H = r=1 1 ˆ r r|t|s c† cs + r. evaluation of their stationary states and excitation energies.9. however. The fortunate side of this is.97). t. the systems investigated involve a macroscopic number of components such that statistical mechanical approaches are envoked.s. the study of problems governed by Hamiltonians of the type (9.97) is over the orbital states. the description of many–particle systems. 2 2 Exercise 9. Often.. In fact. u refer only to the orbital part of the single–particle basis.98.g.98) In many cases the Hamiltonian is spin-independent and the equivalent notation S H = r.96. e.s=1 σ 1 ˆ rσ r|t|s c† csσ + 2 S r. the concerted behaviour of interacting systems can deviate qualitatively from that of systems made up of independent components. In the latter case the indices r. v r s 2 r.3. atoms and nuclei.97) The summation in (9.96) and (9. If it were not for the two–particle contribution to the Hamiltonian. rα sβ (9.99) has preoccupied an important part of all intellectual eﬀorts in Theoretical Physics during the past ﬁfty years. Actually.99) will be used.t. Unfortunately. superconductivity. s. s|ˆ|u. The great generality of the concepts developed in the context of many–particle systems can be judged from the fact .s=1 r..s=1 c† c† crβ csα . e.g.σ (9. the spin part of the single–particle states is represented by α = 1 and β = − 1 . s|ˆ|t. it is one of the main intellectual achievements of Physics among the Sciences to have addressed systematically the study of systems composed of many strongly interacting components. but also ordinary phenomena like freezing and evaporation.t.u=1 σ. u c† c† cuσ ctσ v rσ sσ r. The outcome of these studies is that even when interactions between components (particles) are simple.u S (9.4: Independent-Particle Models ˆ For S 2 one obtains 1ˆ 1 ˆ S2 = N + 2 4 S S 257 ˆ ˆ ˆ ˆ (Nrα − Nrβ )(Nsα − Nsβ ) − r. molecules.s. in crystals.6: Derive (9. which represents the interactions between particles. v c† c† cu ct . 9. 9. 9. Examples is the laser action. would be a simple exercise in linear algebra. that the two–particle operator representing interactions between particles induces a variety of interesting phenomena. the two–particle operator makes the description of many–particle systems a very hard problem.

electrons and nucleons. Amit (Cambridge University Press.J. i. What might happen? Independent-Particle Hamiltonian A many–fermion system governed by an independent-particle Hamiltonian. would not exist. Chemistry essentially would know only one element and Life would not exist.. namely of neurons3 . We will restrict our description in the following to systems composed of an even number (2N ) of particles and to spin-independent interactions. If it were’nt for the fermion nature.s=1 σ ˆ rσ r|t|s c† csσ (9. .100) in a form in which the factors c† csσ become rσ occupation number operators.1: Imagine that in a closed.e. ranging from nuclei to the molecules of living systems . New York.103) An account of some of these eﬀorts can be found in Modelling Brain Function – The World of Attractor Neural Networks by D. The electronic properties of atoms with diﬀerent numbers of electrons would diﬀer little. i. a Hamiltonian without a two–particle operator contribution accounting for interactions among the particles.e. 1989) . (9. where S |m) = r=1 Urm |r . The Hamiltonian of such system is S Ho = r. according to which nuclei and atoms change their qualitative properties when they grow larger. Exercise 9. ultimately depend in a crucial way on the fermion character of its constituent building blocks. {|m). can be described in a rather straightforward way. The Aufbau principle.. water-ﬁlled jar all electrons of water turn their fermion nature into a boson nature.101) The unitary property of Urn U† U = U U† = 1 1 or S S (9. all material systems would condensate into states which would depend little on particle number. . another prototypical systems of many strongly interacting components. To alter the representation of Ho we apply a unitary transformation U to the single–particle states |r to obtain a new basis of states |m). 2.4. r=1 3 Urm U ∗sm = δrs m=1 (9. .102) U ∗rm Urn = δmn . m = 1. Before continuing one may ﬁnally point out that the material world as we know it and as it establishes the varieties of natural substances.100) ˆ ˆ We will denote the matrix elements of t as r|t|s = trs .258 Many–Particle Systems that today these concepts are providing insight into the functioning of biological brains. Transformation to a New Set of Creation and Annihilation Operators Our aim is to represent the Hamiltonian (9. In such a representation Ho is diagonal and the stationary states can be stated readily. S}.

112) . using (9.9. .4: Independent-Particle Models allows one to express conversely |r in terms of |m) S 259 |r = m=1 ∗ Urm |m) . expand S trs = mn ∗ ˜ tmn Urm Usn (9. csσ = n=1 Usn dnσ . of (9.f. [d† . ∗ ∗ Urm Urn .111) can be written. d† ]+ = sσ nσ r. d† ]+ = 0 (9.101) are occupied.45.110) mσ nσ [dmσ . (9. dnσ = r=1 ∗ Urn crσ (9. (9. [dmσ . the expressions drived above for the matrix elements of one.46) as c† and crσ . behave exactly like the operators c† and crσ behave for states |ΨF (N ) . the operators d† and dnσ .s.108) and (9. for nσ rσ example. (9.and two-particle operators hold in an analogous way for d† and dnσ . they are Fermion creation and annihilation operators. d† ]+ = δmn δσσ nσ (9.s r The unitarity property r ∗ Urm Urn = δmn yields immediately identity (??).h.110) can be demonstrated similarly.105) where ˜ ˆ tmn = (m|t|n) = S ∗ trs Urm Usn = r. accordingly. of N -Fermion states in which single particle states |nj ) as deﬁned in (9. The unitary property (9. The l. In a basis of states N d† j σj |0 n j=1 (9. i. nσ We demonstrate here the anti-commutation property (9.109) The operators (9.e.108) These operators describe the creation and annihilation operators of fermions in states |n) which are linear combinations (9. 9. c† ]+ = δσσ [dmσ .s=1 ˆ U † tU mn (9.108) obey the same anti-commutation properties (9.100) yields ˜ Ho = tmn d† dnσ mσ where d† nσ = r=1 S S (9.111) and.113) Urm Usn [crσ .104)] S S ∗ Urm d† mσ m=1 c† rσ = .106) which together with (9.111).101) of states |r . rσ namely. dnσ ]+ = 0 . the anti-commutation properties (9. (9..46).107) Urn c† rσ .103) allows one to express [c.104) One can. hence.

The condition (9. .114) together with (9.101) gives us the freedom to choose the matrix (9. A non-degenerate orbital state which is occupied by two fermions. 1 as well as a fermion in a spin state | 1 .116) which follows from (9. 2 2 2 2 i. the associated ( properly normalized) eigenvectors Vn are the columns of (Urn ).. . . It is. σj ) = (nk . This equation poses the eigenvalue problem for the hermitian matrix (trs ).106) and the unitary property of Urn . .115) with eigenvalues 2N E(n1 . ) = j=1 nj .115) can be stated immediately since any many–particle wave function in the occupation number representation is suitable.114) Ho = n=1 † n dnσ dnσ (9. 1 and |β = | 1 ..106) is equivalent to S trs Usn = s=1 n Urn ∀r. transformation matrices Urn and energy values n . j j (9. n2N . n = 1. σ2 . a fermion in a spin state | 1 . hence. .e.106) diagonal. . The 2N fermion state 2N d† j σj |0 n j=1 where (nj . 2.e. σ1 . σk ) for j = k (9. .. − 1 . The corresponding eigenvalues n . We apply for this purpose (9. .117) are eigenstates of (9. . a simple matter to nσ state many–particle states in the new representation.260 Diagonal Representation Many–Particle Systems The transformation (9. n2 . i. Vn = (U1n . . . r = 1.118) Ground State In case of an ordering m < n for m < n the state of lowest energy. U2n . S are T real. . i.. Before proceeding we like to address the issue how the new representation.57) to the case of 2N particles. USn ). − 1 is called a closed 2 2 2 2 .119) In this state the N lowest orbital eigenstates of trs are occupied each by an electron with spin |α = | 1 .e. σ2N . S (9. i. Eigenstates of (9. (9.e. (9. 2. the so-called ground state. is obtained.115) ˜ and involves solely occupation number operators N nσ = d† dnσ . is N |Φo = j=1 d† α d† β |0 . ˜ tmn = In this case Ho has the desired form S n δmn .

α |β. Alltogether. However. One can conclude immediately that the ˜ ˜ ˜ ˜ (N mα − N mβ )(N nα − N nβ ) does not give any contribution second term in (9. We will demonstrate now that this property endows the ground state with total spin zero.97). ˆ S 2 ||β.n=1 d† d† dmβ dnα mα nβ (9. Excited States We want to construct now excited states for the system governed by the independent-particle Hamiltonian (9. β .e. 1 S 4 m.120) is particularly simple for closed shells.100). of ∆E = and |β. The ground state has only closed shells.121) All four states have the same excitation energy.n=1 † † if either |m) or |n) are closed shells.n=N (9. (9.4: Independent-Particle Models 261 shell. is a singlet state.n=1 m..120) where the occupation number operators refer to the single-particle states |m). There are four such states. i. α N +1 − N . α . Similarly. |α.97).122) This property can be derived as follows: The two states are obviously diagonal with respect to the ˆ following contribution to S 2 1ˆ 1 ˆ Σ1 = N + 2 4 N −1 N −1 ˆ ˆ ˆ ˆ (Nmα − Nmβ )(Nnα − Nnβ ) − m. the third term − S m. namely.e. α = 2 |β. the states diﬀer in their spin character.9.123) ˆ which acts only on the N − 1 closed shells (except for N ) of the two states and has an eigenvalue ˆ N + 0 − (N − 1) = 1 in both cases. namely. ˆ In order to investigate the total spin of |Φo we apply to this state the total spin operator S 2 as given by (9. energy above the ground state..n=1 cmα cnβ cmβ cnα does not give a contribution if m = n and either |m) or |n) are closed shells. The two states |α. the states with energy closest to the ground state are those in which a particle is promoted from the highest occupied state |N ) to the lowest unoccupied state |N + 1). Obviously. i. both with eigenvalues 2 ˆ S 2 |α.124) . β = d† +1α dN α |Φo N = d† +1α dN β |Φo N = d† +1β dN α |Φo N = d† +1β dN β |Φo N (9.n=1 d† d† dmβ dnα mα nβ (9. β = 2 |α. ˆ one can conclude that |Φo is an eigenstate of S 2 with eigenvalue zero. β |β.120). The remaining contributions to S 2 which act on partially occupied orbitals |N and |N + 1 are 1 ˆ Σ2 = 4 N +1 ˆ ˆ ˆ ˆ (Nmα − Nmβ )(Nnα − Nnβ ) m.n=1 m. α |α. The spin operator in the present representation of single-particle states is 1 1ˆ ˆ S2 = N + 2 4 S S ˜ ˜ ˜ ˜ (N mα − N mβ )(N nα − N nβ ) − m. β ˆ are eigenstates of S 2 given in (9. In case of m = n it gives a contribution ‘1’ for each closed shell which cancels the contribution of the ﬁrst term. The action of the operator (9.

. β both with eigenvalues 1. r = 1. . do not need to be considered since they ˆ give vanishing contributions. . The two states |α.100) outlined above we consider a system of 2N electrons which move in a set of atomic orbitals |r . One can.σ d† +1. β . c† +1 σ = c† σ 1 2N (9.262 and ˆ Σ3 = − m.122) is correct. 2N which are located on a ring. however.127) ˆ are triplet and singlet states. β are not eigenstates of S 2 . . also eigenstates of Σ2 . . for example. for |α.n=N d† d† dmβ dnα d† +1α dN β |Φo = 0 mα nβ N (9. conclude that (9. ˆ 3 on the two states produces . 2N ) drσ = √ 1 2N 2N e−irnπ/N cnσ n=1 (9.e. hence. . β and |β. 2. . α as triplet states. one with eigenvalue zero (singlet) and one with eigenvalue ‘2’ (triplet). (9. α are.2: Construct four operators O ˆ O ˆ such that O ˆ S3 .σ dN σ N σ. .126) which follows from the occurence of squares of fermion operators which. |N + 3 . α . vanish. ˆ The remaining states |α. .n=N Many–Particle Systems N +1 d† d† dmβ dnα . apropriate eigenstates of the operators S 2 and Example: 2N Independent Electrons on a Ring In order to illustrate the procedure to obtain eigenstates of one-particle Hamiltonians (9.128) The cyclic property of the system is expressed through c2N +1 σ = c1 σ . mα nβ (9. ˆ An eigenvalue ‘2’ of the spin operator S 2 identiﬁes the states |α.130) .σ = (9.125) ˆ Contributions to S 2 acting on states |N + 2 . ˆ Exercise 9.129) The symmetry of the Hamiltonian suggests to choose a new representation deﬁned through the operators (r = 1. We identify the states |2N + 1 = |1 to avoid cumbersome summation limits. The system is described by the Hamiltonian 2N ˆ Ho = − t n=1 σ c† σ cn σ + c† σ cn+1 σ n n+1 . i. β =− m. The system is assumed to posses a 2N –fold symmetry axis and interactions connect only orbital states |r → |r ± 1 . |β. of course. etc. m |Φo m m m γσ. . α and |β. the linear combinations 1/2(|α. of course. β qualify as eigenstates. The action of Σ N +1 ˆ Σ3 |α. 2.4. α ± |β. The same result holds for |β.

131) constitute a unitary transformation U deﬁned through Urn = unitarity property follows from 2N eirsπ/N = 0 r=1 for 0 < s < 2N . of (9. 9.137) where 1 ˜ trs = e−irπ/N 2N Using (9.135) c† nσ 1 = √ 2N e−irnπ/N d† rσ r=1 (9.s=1 σ ˜ rσ trs d† dsσ (9.132) an identity which can be derived employing the well-known expression for a ﬁnite geometric series m as = a s=1 1 − am .136) ˆ and. (9.132) gives for r = 0.139) −2t cos r=1 σ rπ N d† drσ . eisπ/N 1 − e2N isπ/N 1 − eisπ/N (9.132) one can conclude 2N ei(s−r)nπ/N + eisπ/N n=1 1 2N 2N ei(r−s)nπ/N .133) Application to the l.134) which. (9. . 4N. vanishes for integer s.f. n=1 (9.4: Independent-Particle Models d† = √ rσ 1 2N 2N 263 eirnπ/N c† nσ n=1 √ 1 eirnπ/N .140) The Hamiltonian is now diagonal and the construction outlined above can be applied.h.131) The (9. 1−a (9. . indeed. 2N (9.138) ˜ trs = from which follows ˆ Ho = eirπ/N + e−irπ/N 2N δrs (9.s.109)] ¸ cnσ = √ 1 2N 2N eirnπ/N drσ r=1 2N (9. rσ (9. . .9. 2N. thereby. one obtains the new representation of Ho 2N ˆ Ho = −t r. One can then express[.130. .

Since independent-particle behaviour is so universal one may wonder why the presence of particleparticle interactions. does not spoil it.3: Construct the ground state and the lowest energy excited state(s) for a system of 2N electrons moving in a linear chain of single-electron orbitals |n . (9. in particular. and that independent-particle behaviour surfaces mainly because it applies well to the ground state.5 Self-Consistent Field Theory Observations of many-fermion systems show that properties of such systems often can be explained to a surprisingly good degree on the basis of the assumption that the fermions move like independent particles.. i.. the everpresent Coulomb repulsion between electrons.e. Many electronic properties of solids can be understood in a similar way. i. n = 1.. A similar principle accounts approximately for properties of nuclei as described by the nuclear shell model. i. if one places fermions into an independent-particle ground state of the type (9.g. 2N with interactions between neighbouring (n. . one expects that mean-ﬁeld descriptions should be rather suﬃcient. The reason is mainly connected with the fact that most many-fermion systems are found in their ground state or removed from it through low energy excitations. The slow decay of the interaction makes the motion of any one electron rather independent of the exact position of most other electrons. properties of many-fermion properties deviant from such simple description is then of as much interest as independent fermion descriptions are useful.e. n ± 1) orbitals and Hamiltonian 2N −1 ˆ Ho = − t n=1 σ c† σ cn σ + c† σ cn+1 σ n n+1 . e.e. . if one can choose the singleparticle states such that the residual perturbations are small.e. . 9.119) then perturbations due to pair interactions according to the Pauli exclusion principle can only involve independent particle states not occupied in the ground state.e. assuming that electrons move independently from each other in so-called bands. The properties of molecules can be understood largely in terms of models which place electrons into so-called molecular orbitals in which the electrons move as if they were not interacting with other electrons... These states do not altogether neglect the pair interactions. We will present in this and the following section a method which constructs such optimal single-fermion states. i. Of course.141) State the excitation energy ∆E. The regularities of the elements can be rationalized in terms of electrons moving in hydrogen atom-type orbitals.4. in states which are constructed according to the ‘Aufbauprinzip’ in a manner which neglects the mutual repulsion between electrons. the fermion nature restricts the possibility for perturbations on the system.264 Many–Particle Systems Exercise 9. valence bands or conduction bands. There are two reasons why this is so: ﬁrst. The most famous example is the periodic table of the elements. rather they assume that each particle experiences the average interaction due to the fermions frozen into the ground state. i.. A second reason why independent-particle models can be successful is that the pair interaction for electrons is the slowly decaying Coulomb interaction. those with energy above that of the (energetically) highest occupied single-particle state. Hence. . 2.

p. 2.s=1 σ r.142) is based. 2. . .t. m = 1. Mean-Field Potential The mean ﬁeld for the reference state |Φo (U) is deﬁned in a straightforward way by averaging the two-particle contribution to (9. u refer to the initial basis of single-particle states |r . This is done as follows: ˆ Vmf (|Φo (U) ) = 1 2 S r.u=1 σ. U = ( Urm ) as deﬁned in (9. ˜ (9. . S on which (9. as in in (9. The states |m are connected with the states |r through (m = 1.σ and derive a single-particle operator which approximates this Hamiltonian. N . . s|ˆ|t.108). j j (9.145) which will play a pivotal role: it is this the state in which the fermions are assumed to be moving and relative to which the mean interactions acting on individual fermions is determined. . n.143) with r. . . S) ˜ S S |m = ˜ r=1 Urm |r . t. . . 2.145). the corresponding U = ( Urm ) forms an S × N matrix and. s. r = 1. .144) deﬁned as in (9.σ c† c† rσ sσ cuσ ctσ + c† c† rσ sσ c† ctσ cuσ rσ c† cuσ rσ ctσ cuσ ctσ + c† c† cuσ rσ sσ ctσ + c† sσ − c† c† ctσ cuσ − c† rσ sσ sσ (9.5: Self-Consistent Field Theory 265 In this section we will consider systems of 2N fermions described by the spin-independent Hamiltonian S S 1 ˆ rσ H = r. . . .146) . m = 1.9. S} the elements of which will be labelled by indices ˜ m. is not unitary.s. |r = m=1 ∗ Urm |m . s|ˆ|t. naturally. We will adopt a second set of single-particle states {|m .142) over this reference state to turn the contribution into an eﬀective one-particle operator. If one restricts m = 1.142) r|t|s c† csσ + rσ sσ 2 r. In our formulas below we will adopt strictly the convention that indices r. 2.t. 2.143) These S states are associated with the creation and annihilation operators S d† mσ = r=1 S Urm c† rσ ∗ Urm crσ r=1 dmσ = (9. q. The states |m serve to deﬁne a reference state of the system ˜ N |Φo (U) = j=1 d† α d† β |0 .s. . S is a unitary matrix. . u c† c† cuσ ctσ v (9.u=1 σ. u v r. .

m|ˆ|t.150) r.σ S d† dnσ cuσ mσ − r.σ S 1 − 2 where. d† d† mσ nσ dmσ dnσ d† dnσ mσ = = = = and.s. ˆ Vmf (|Φo (U) ) = 1 2 S Φo (U)|d† d† |Φo (U) = 0 mσ nσ Φo (U)|dmσ dnσ |Φo (U) = 0 Φo (U)|d† dnσ mσ δmn δσσ 0 |Φo (U) m ≤ N m > N (9.n.σ S ctσ + m. by means of the rules (9. s|ˆ|t.σ S d† d† mσ nσ cuσ ctσ + r.s. hence.t=1 σ. n c† c† v ˜ ˜ rσ sσ r. m|ˆ|t.n=1 σ.n=1 σ.t. n c† d† dnσ ˜ v ˜ rσ mσ r. u ˜ ˜v m. n = ˜ ˆ ˜ s.266 Here ··· denotes the average O = Φo (U)| O |Φo (U) .n. s|ˆ|m.149) The remaining matrix elements appearing in (9. m. n|ˆ|t.148) one obtains.σ S dmσ dnσ + r. s| v t. u c† ˜ vn sσ m. n c† ˜ v ˜ sσ m. for example. ˆ (9.t.t=1 σ.u=1 σ. u Usm Uun .147) Since the reference state |Φo (U) is deﬁned in terms of the single-particle states |m it is preferable ˜ to switch the representation of (9.84).u=1 σ. For the averages ··· in (9.u ∗ r. u c† d† dnσ cuσ ˜ vn rσ mσ r.148) are obtained in a similar way by transforming only two of the four single-particle states into the new representation. m| v |t.σ .m.148) r. n c† ctσ δσ σ ˜ v ˜ rσ m≤N r.n. s|ˆ|˜ . m|ˆ|˜ . Since the averages aﬀect only two creationannihilation operators actually a mixed representation is most suitable ˆ Vmf (|Φo (U) ) 1 = 2 1 2 1 2 1 2 1 2 S m.146) accordingly.σ S d† dnσ mσ ctσ (9.n.m.s. Many–Particle Systems (9.u=1 σ.m.

The one-particle operator (9. u = s. s|ˆ|t. t| v |u. r| v |u.. t| v |s.m.σ r.. n c† ctσ δσ σ ˜ v ˜ rσ − r.N (9.u=1 σ. t and renaming dummy summation indices one can demonstrate ˆ ˆ that term one and term two as well as term three and four are identical and one obtains S ˆ Vmf (|Φo (U) ) = 2 r. s ) ˆ ˆ m≤N ∗ Utm Uum (9.s σ where r| vmf (|Φo (U) ) |s = ˆ t.u=1 σ. m|ˆ|t.9.u=1 σ S − m.2. . m|ˆ|˜ .154) ˆ The mean ﬁeld approximation replaces then the Hamiltonian (9.152) can then be written in the form ˆ Vmf (|Φo (U) ) = r| vmf (|Φo (U) ) |s c† csσ ˆ (9.m..σ S r. ˜ v ˜ sσ m≤N s. n c† ctσ . The mean ﬁeld approximation. We will address now a proper choice of the reference state. u c† cuσ δσσ ˜ vn sσ m≤N s.. u c† cuσ ˜ vn rσ m≤N r. leaves open the question how the reference state should be chosen.155) which is a function of the S × N –matrix U = ( Urm )r=1. as introduced here.5: Self-Consistent Field Theory 1 2 1 2 1 2 S 267 m..151) Carrying out the sum σ δσσ = 2 leads to a factor 2 for the ﬁrst two terms.153) rσ r.u ( 2 r.t=1 σ. u c† cuσ ˜ vn rσ (9. s|ˆ|˜ . At this point this state is completely arbitrary.142) by Hmf (U) deﬁned through S ˆ Hmf (U) = r. Exploiting the symmetry r.t=1 σ (9. m|ˆ|˜ ..2. u − r.149) for the matrix elements in the mixed representation one can state the right hand side explicitly in terms of Urm .S.156) which deﬁnes the reference state |Φo (U) . s| v |t. m=1.s=1 σ ˆ r|t|s + r| vmf (|Φo (U) ) |s ˆ c† csσ rσ (9.152) By means of expression (9.σ S + − r.

(9. .e. determining in a step (2) the corresponding mean ﬁeld Hamiltonian (9. the independent-particle nature stemming from the fact that the functional form of the ground state. is exact only for an independent-particle Hamiltonian. i. S . . however.142).. 9. (9. under conditions which often are realized. . . 2.s=1 σ ˆ r|hmf |s c† csσ rσ (M ) (9. r = 1. . . is based on the mean ﬁeld approach and provides an algorithm to construct a reference state |Φo (U) for a 2N fermion system described by a spin-independent Hamiltonian (9.157) and determines the associated diagonal representation deﬁned through S (1) ˆ r|t|s Usm = s=1 (1) (1) m Urm . SCF-Algorithm.145). . U(1) is the S ×N –matrix (1) of the ﬁrst N column vectors of Urm . assuming in an intial step (1) a properly chosen reference state.155). .6 Self-Consistent Field Algorithm The Self-Consistent Field (SCF) approximation.158) where the labels m are ordered to obey the condition (1) m < (1) n for m < n . can achieve this goal only iteratively.268 Many–Particle Systems 9. M = 1.e.e.s=1 σ S ˆ r|t|s + r| vmf |Φo (U(M ) ) ˆ |s c† csσ rσ = r. S . We will argue below that the self-consistent ﬁeld ground state.155) following the method outlined in Section 9.4 and deﬁning this as the new reference state. 2. . . Step 1: Choosing an Initial Reference State One deﬁnes the Hamiltonian ˆ (1) Hmf = r.. S.160) . . i.119. . Step 2M. m = 1. Let us state now the construction of the self-consistent ﬁeld ground state in more detail. This procedure. is the lowest energy independent-particle ground state one can construct.s=1 σ S ˆ rσ r|t|s c† csσ (9. . and obtaining in a step (3) the ground state of this one-particle Hamiltonian according to the construction in Section 9.: Determine Mean Field Hamiltonian In this step the mean ﬁeld Hamiltonian S ˆ (M ) Hmf = r. The procedure determines the reference state as the ground state of the independent-particle Hamiltonian (9.159) One deﬁnes the reference state |Φo (U(1) ) using the deﬁnition (9. 2. S . the reference state resulting from step 2M + 1 (within numerical error) is equal to the reference state resulting from step 2M − 1. 2. The state thus determined is referred to as the self-consistent independent-particle ground state. . SCF-Algorithm. 2. i. m = 1. (9. . often also referred to as the Hartree-Fock approximation.4. steps (2) and (3) are being repeated M times until the procedure converges.145). r = 1.

6: Self-Consistent Field Algorithm is evaluated. (9. . 2. 2. (c) Show that any spin-independent one-particle operator S ˆ F = r.s=1 (9. 2. SCF-Algorithm: Continuation and Convergence Condition When step 2M + 1 is completed.166) ˆ (d) Deﬁne a similar two-particle operator Prstu . . SCF-Algorithm. i.e.164) ˆ (a) Determine the expectation value of Prs for the state as deﬁned in (9.s=1 σ ˆ r|f |s c† csσ rσ (9. . .: Diagonalize Mean Field Hamiltonian In this step the eigenvalue problem S (M r|hmf |s Usm +1) = s=1 (M ) (M +1) (M +1) Urm m . S (9. step 2M + 2 is carried out. . r = 1. . . S .145). s = 1. Step 2M + 1. the S × S–matrix S (M ) r|hmf |s N 269 ˆ = r|t|s + t. M = 1. S do not exceed a preset threshold. i. (e) Consider the creation operators N † gqσ = m=1 Vmq d† mσ (9. r = 1. u − r. 2. t| v |u. .u=1 ( 2 r. . . t| v |s.9. indicating that the state converged. U(M +1) is the (M +1) S × N –matrix of the ﬁrst N column vectors of Urm . (9. ˆ Exercise 9.165) can be written ˆ F = S ˆ ˆ r|f |s Prs r. . ˆ (b) How can the expectation values of the operator Prr be interpreted. s ) ˆ ˆ m=1 Utm ∗(M ) (M Uum ) (9. ˆ t. etc. . σ|ˆrs |u. . m = 1. .. σ p = δtr δsu δσσ .167) .161) is calculated. The procedure is continued until it ∗(M ) (M ) ∗(M +1) (M +1) Uum − N Urm Uum | is detected that the one-particle density diﬀerences | N Urm m=1 m=1 for r. .145). i.e.163) The result yields the reference state |Φo (U(M +1) ) using the deﬁnition (9. m = 1.162) is solved adopting an ordering of the labels m which obeys the condition (M +1) m < (M +1) n for m < n .6.. S . . 2 .1: Let Prs be the one-particle operator with the generic operator prs = |r s|. 2.e. . S.

2. (SCF ) m < (SCF ) n for m < n (9.270 Many–Particle Systems which are connected with d† through a unitary N × N –matrix (Vqm ).2: Determine the SCF ground state and its energy expectation value for a system of two particles described through the Hamiltonian (S = 2) H = σ ( c† c1σ + c† c2σ ) − t ( c† c2σ + c† c1σ ) 1σ 2σ 1σ 2σ ( c† c1σ c† c2σ 1σ 2σ σ. We denote the representation which results from the SCF algorithm after its convergence as follows: S |m = ˜ r=1 (SCF Urm ) |r . (9. i. Equation (9.163). .7 Properties of the SCF Ground State We want to investigate now the properties of the SCF ground state..163). r = 1. .169) 9. (f) Can one distinguish the states (9.145).171) implies that for the mean ﬁeld Hamiltonian holds S ˆ Hmf (U(SCF ) ) = m=1 σ (SCF ) † dmσ dmσ m .σ + 2v ( c† c1α c† c1β + c† c2α c† c2β ) + v 1α 2α 1β 2β + c† c2σ c† 2σ 1σ c1σ ) . (9. S denotes the initial single-particle states and Urm is the unitary S × S– transformation matrix obtained in (9.173) . Note that this is not an mσ S × S–matrix! Show that the reference state deﬁned through N † † gqα gqβ |0 q=1 (9.e. (9.171) where the convention (9. We begin by summarizing the result of the SCF algorithm.6.162. For this representation holds N m| t |˜ + ˜ ˆn m 2 mm | v |˜ m ˜ ˜ ˆ n˜ − mm | v |m n ˜˜ ˆ ˜ ˜ = (SCF ) δmn m (9.145) and (9.9.172) is obeyed. .170) (SCF ) Here |r .168) through a physical observation? Exercise 9.168) ˆ ˆ has the same expectation values for Prs and Prstu as the reference state (9.

. .174) The ﬁrst property we consider is the total spin character of SCF . v (SCF ) Show that the resulting eigenvalue problem of the type (9.173) holds N ˆ SCF Hmf (U(SCF ) ) SCF = 2 r=1 (SCF ) m . 2 m=1 (SCF ) m − m.162) is non-linear in Urm .7: Properties of the SCF Ground State 271 i.2: Derive (9. i. Finally. mα mβ (9.. (SCF ) ˆ Exercise 9.175) However.142).1: Reformulate (9. u and Urm . one can also determine the expectation value of SCF for the Hamiltonian (9.e. the SCF ground state is N SCF = m=1 d† d† |0 . We like to determine next the energy expectation value of SCF .177).7. the mean ﬁeld potential ˆ Vmf (|Φo (U) ) deﬁned in (9. r. once as a member of the ground state.142).171) in terms of matrix elements r|t|s . s|ˆ|t. Employing expressions (9. Exercise 9. (9. Since the system under consideration has only 2N particles altogether. ˆ SCF H SCF .9.142). the mean-ﬁeld Hamiltonian is diagonal in the SCF representation. and once as a probe particle being subject to the mean pair interaction.e.146) counts a particle twice.4 that SCF is a total singlet state. exploiting the spinindependence of (9. 2 m=1 m|t|m + ˜ ˆ˜ m. Since this state is composed only of closed shells one can conclude following Section 9. This over-counting leads to the correction term in (9.177).7.177) 2 mm | v |mm ˜˜ ˆ ˜˜ − mm | v |m m ˜˜ ˆ ˜ ˜ . Within the mean ﬁeld approximation as described by (9.84) and (9.176) 2 mm | v |mm ˜˜ ˆ ˜˜ − mm | v |m m ˜˜ ˆ ˜ ˜ .93) for the expectation values (diagonal elements) of the one-particle and two-particle part of the Hamiltonian (9.m =1 The second term originates from the fact that in the mean ﬁeld approximation one assumes that each of the fermions is subject to an average pair interaction involving a 2N -particle reference state. one obtains ˆ SCF H SCF N = N (9.171) yields ˆ SCF H SCF N = N (9.m =1 Comparision with (9.

According to (9.s. approximation schemes like the self-consistent ﬁeld approximation play an important role. The operator c† (crσ ) creates (destroys) an electron with spin σ (σ = α. The case n = 1 is termed the half ﬁlled band case. . t is assumed to be a positive. of (9.178).4–9. is identical to that of the independent–particle Hamiltonian H0 (9. † ˆ Ho = −t c† .u σσ where r.h. of (9.7 for ﬁnite systems. . the second term on ther.σ rσ The potential energy term.e.σ + v 2 c† c† crσ crσ .. v (9. in the usual two–particle operator form. cannot be solved exactly. i. v (9. is characterized through the parameters t. t = v δru δst δrs .. respectively) in state |r at lattice site r.e. in the two-dimensional copper oxide lattices of high temperature superconductors.h. labeled r = −N0 + 1.178).178) and through the so-called ﬁlling factor n n= 2N N = . β for ‘up’ and ‘down’ rσ spins. We consider as an example the Hubbard model of an inﬁnite linear lattice and apply the model to describe magnetic instabilities in metals. real number.272 Many–Particle Systems 9.178).182) n can assume values 0 ≤ n ≤ 2. i. In spite of the simplicity of its Hamiltonian the Hubbard model. t describes the coupling of state |r to states |r ± 1 at the two neighboring lattice sites.181) The Hubbard model.σ crσ + crσ cr+1. except in case of one-dimensional systems.8 Mean Field Theory for Macroscopic Systems The SCF algorithm has been developed in Sections 9. . s|ˆ|t.179) r+1. The one-dimensional Hubbard model is described by the Hamiltonian ˆ H = −t rσ † c† r+1. rσ rσ r σσ (9. as stated through Hamiltonian (9.t. reads 1 ˆ r. . (9. S N0 (9. N0 which are populated by 2N electrons (we choose an even number of electrons for convenience).s. for example.. Both N0 and N are macroscopically large numbers. v represents the ‘on-site’ Coulomb repulsion. referring to the fact that in this case the band of single particle energies (9.e. u c† c† cuσ ctσ .192) derived below is ﬁlled half. In the present Section we want to adapt the algorithm to systems containing a macroscopically large number of fermions. r ∈ Z. v of the Hubbard Hamiltonian (9. The ﬁrst term on the r. The Hubbard Model The Hubbard model serves today the important role as the simplest manifestation of strongly correlated electron systems which arise in many instances in molecular and solid state physics. Due to the lack of an exact solution in dimensions higher than one.180) V = rσ sσ 2 r.s. describes the sites of a linear lattice. .178) v contributes only in case that two electrons occupy the same lattice site.128).σ crσ + crσ cr+1. s|ˆ|u. We assume that there are altogether S = 2N0 lattice sites.178) Here the index r. i.

185) ˆ According to (9. i.184) ˆ We are now ready to apply step 1 of the SCF algorithm and diagonalize term H0 .144).140). N0 (9.186) one obtains for the mean ﬁeld Hamiltonian N0 ˆ (2) Hmf = m=−N0 +1 σ m d† dmσ + v mσ N 2N0 c† crσ . Accordingly. we identify |r + N0 = |r and later let N0 go to inﬁnity. < ±(N0 −1) < N0 holds as can be readily veriﬁed. rσ rσ (9. For this purpose we apply the SCF theory presented in Section 9.9.178) within the framework of the self-consistent ﬁeld theory.181. mα mβ (9.174).155. 9.. SCF Ground State for the Hubbard Model We seek to determine the ground state of the Hubbard model stated by (9. is Urm = √ 1 exp (irmπ/N0 ) .179). rσ (9. This task has ˆ been solved already on page 262. Using (9.154). boundary eﬀects are assumed to be negligible.e.187) The energy levels are labeled such that 0 < ±1 < . 9. we assume that the SCF ground state is given by (9. the operator H0 can be re-written in the diagonal form N0 ˆ H0 = m=−N0 +1 σ m d† dmσ mσ (9. a freedom which we employ to adopt so-called periodic boundary conditions like those for the example “2N independent electrons on a ring” on page 262. Inserting (9.185) into (9.186) where m = −2t cos mπ . by N −1 ||SCF = m=0 d† d† |0 . in fact. The unitary transformation which diagonalizes H0 . The self-consistent ﬁeld Hamiltonian Hmf which corresponds to the Hubbard Hamiltonian (9.183) where the operators d† are related to the original creation operators c† through the unitary mσ rσ ˆ transformation U deﬁned in (9.6.184) and using (9. Accordingly. one obtains for the mean ﬁeld Hamiltonian N −1 ˆ ˆ Hmf (U) = H0 + v r. We can now embark on step 2 of the SCF algorithm.188) . 2N0 (9.8: Macroscopic Systems 273 Since we are dealing presently with a macroscopic system. .σ m=0 ∗ Urm Urm c† crσ . .178) can be determined by applying (9.

is already diagonal.. The ground state of the system. the self–consistency condition is exactly met and the SCF algorithm converged after two steps. mσ (9. for km = mπ/N0 holds km ∈ −π + N0 . One can see that −2t ≤ (k) ≤ 2t holds. called the Fermi energy.182)]. crσ through d† . namely. Usually. there exist several ways of formulating mean ﬁeld approximations.191) where m is given in (9. i. can be obtained. N0 ˆ (2) ˆ Hmf = Hmf = m=−N0 +1 σ m + vn † dmσ dmσ 2 (9.189) and.f.182) of the ﬁlling factor n [c.183) is not the only candidate for the mean ﬁeld ground state of the Hubbard model.187) form a quasi1 2 continuous energy band. The energy Emf of this state can be calculated readily by using (9. π]. (9. as given in (9.1). 9.178).177) and mm |ˆ|mm = mm |ˆ|m m = ˜˜ v ˜˜ ˜˜ v ˜ ˜ which holds in the present case. 4 So far we have not exploited the fact that N0 and N are macroscopically large quantities. denoted by F (see Fig. i. but may diﬀer quantitatively.187). according to the Pauli principle.e.185) one can prove rσ mσ N0 c† crσ = rσ rσ m=−N0 +1 σ d† dmσ . π and. in the limit N0 → ∞. 9. through the so-called dispersion law (k) = −2t cos k .190) We have used in the latter expression the deﬁnition (9. Indeed.190).e. by ﬁlling up this energy band with electrons from the bottom of the band (which corresponds to k = 0) to a maximum energy. To demonstrate this we consider an alternative formulation of the mean ﬁeld approximation for Hamiltonian (9. . Emf is then N −1 v 2N0 Emf = 2 m=0 vn vn −N = 2 m+ 2 2 N −1 m m=0 + vn 4 (9. the electrons in the present description behave like independent particles with energies ¯m = m + vn .274 Many–Particle Systems Expressing c† . the results of diﬀerent formulations agree qualitatively.. the state with the lowest possible energy. . The bottom of the energy band corresponds to (0) = −2t and the width of the energy band is 4t. The energy m can be expressed as a function of a continuous variable k ∈] − π. −π + N0 . . dnσ according to (9. (9.1.109) and using (9. Apparently. π]. Alternative Description of the Mean Field Approximation The state (9. In other words. holds k ≡ km ∈] − π. Indeed. the ground state of the Hubbard model in the self-consistent ﬁeld approximation can be determined exactly.192) This dispersion law is presented in Fig. ˆ The Hamiltonian Hmf . When N0 becomes macroscopically large the single particle discrete energy levels (9. −π ≤k ≤π . hence. even for one and the same Hamiltonian. .

rσ rσ rσ (9. i. (9.193) into (9.8: Macroscopic Systems 275 Figure 9.56). Inserting (9.9. we have employed (9.e.194) ˆ For a given reference state.. i.45.195) ˆ ˆ ˆ and where δ (Nrσ ) = Nrσ − Nrσ describes the ﬂuctuations of Nrσ . the deviation of Nrσ from its mean value Nrσ for the corresponding reference state.196) (9. The anti-commutation properties (9. F = (±kF ) denotes the Fermi energy. Nrσ = SCF ||Nrσ ||SCF . in the last term on the right hand side.. The mean ﬁeld approximation neglects the .178) and performing the summation over the spin indices results in ˆ ˆ H = H0 + v r ˆ ˆ Nrα Nrβ . say the one given in (9. the operator Nrσ can be written ˆ Nrσ = Nrσ + δ (Nrσ ) . (9.e. First. let us express the interaction term in (9.193) where.183).178) in terms of the occupation number operaˆ tors Nrσ = c† crσ . where Nrσ is the mean occupation number of the one particle state |rσ .1: The dispersion law (k) for independent electrons on an inﬁnite one dimensional lattice. 9.46) of the fermionic creation and rσ annihilation operators yield c† c† crσ crσ = −c† c† crσ crσ = rσ rσ rσ rσ ˆ ˆ ˆ2 c† crσ c† crσ − c† crσ δσσ = Nrσ Nrσ − Nrσ δσσ .

non-vanishing magnetization. Spin-Polarized Mean Field Ground State ˆ We want to consider now a ground state for Hmf deﬁned through Nrσ = nσ = const (9. Note that the procedure employed in (9.198) Let us assume that the ground state of the Hubbard model. In case nα = nβ the following construction will lead to a ground state which coincides with the non-magnetic ground state (9.199) which is identical to expression (9. For such state the mean occupation number nrσ is uniform at all lattice sites.m ∗ Urm Urm d† dm σ mσ = m=0 ∗ Urm Urm = n N = .195). separating the occupation number operator into a mean value plus ﬂuctuation and neglecting the ﬂuctuations in 2nd order.194) by dropping all the terms which contain the ﬂuctuations to second order. 2N0 2 holds ˆ Hmf = = mσ mv ˆ H0 + 2 vn2 2 rσ nv † vn dmσ dmσ − N . is given by (9.198) as the one obtained by using the mean ﬁeld potential (9. however.200) allowing. In this approximation one can express ˆ ˆ Nrα Nrβ = ≈ = ( Nrα + δ (Nrα )) ( Nrβ + δ (Nrβ )) Nrα Nrβ + Nrα δ (Nrβ ) + Nrβ δ (Nrα ) ˆ ˆ Nrα Nrβ + Nrβ Nrα − Nrα Nrβ (9.197) and the corresponding mean ﬁeld Hamiltonian becomes ˆ ˆ Hmf = H0 + v r ˆ ˆ Nrα Nrβ + Nrβ Nrα − v r Nrα Nrβ . that nα and nβ assume diﬀerent values. i. Therefore. consequently. yields essentially the same mean ﬁeld Hamiltonian (9.183).191) for the ground state energy.146). (9.276 Many–Particle Systems ﬂuctuations to second order and the mean ﬁeld Hamiltonian is obtained from (9. To determine a magnetic ground state we note ˆ Nrσ = r mm r ∗ Urm Urm d† dm σ = mσ mm ˆ δmm Nmσ = m ˆ Nmσ . but the mean number of particles with spin α can be diﬀerent from the mean number of particles with spin β. in the mean ﬁeld approximation.. m+ 2 2 c† crσ − 2N0 rσ (9.e. . the ground state of the system can have a non-zero local spin and.183) . Because of Nrσ = c† crσ rσ N −1 = m.

206) ˜mf is minimized where µ is the Lagrangian multiplier considered an independent variable. one can see that an uneven occupation by electrons of the two energy bands yields a relative shift of the energy bands with respect to each other.208) .. (9.205) Such minimization is realized through the method of Lagrangian multipliers. the smaller is σ (k) for a given k.207) where (k) is given by (9.σ with respect to nα and nβ . as is shown schematically in Figure 9. e. Therefore. β) are determined by minimizing the energy density of the ground state Hmf 1 = [ mσ Nmσ ] − vnα nβ (9. (9. a spin–polarized ground state.206). (9. The ground state (corresponding to the lowest possible energy) of the many electron system is obtained by ﬁlling these two energy sub–bands with electrons up to the same maximum energy value..e.206) ascertains that condition (9.202) This Hamiltonian describes a system of non-interacting electrons with a spin-dependent dispersion law (9. For this purpose the sum in (9. m (9..e.203) mσ = m + vn−σ . In fact.8: Macroscopic Systems from which we obtain nσ = The mean ﬁeld Hamiltonian is then ˆ Hmf = m 277 1 2N0 Nmσ . The additional term in (9. N0 ∼ 1023 . one can expect the system to lower its energy by assuming nα = nβ and. This last equation tells us that the electrons with spin α (β) are accommodated by an energy sub–band α (k) ( β (k)) which is obtained from the dispersion law of the non-interactive electrons (k) by an overall shift of vnβ (vnα ). if nα > nβ then β (k) > α (k). hence. E with respect to nα . the so-called Fermi energy F = µ (see below).201) ( m ˆ + vnβ ) Nmα + ( m ˆ + vnα ) Nmβ − 2N0 vnα nβ .206) needs to be evaluated.204) Emf ≡ 2N0 2N0 m. In this limit the discrete energy spectrum (9. 2π (9. At this point we exploit the fact that our system is macroscopically large. According to this method one minimizes ˜ Emf = Emf + µ (n − nα − nβ ) . The actual values of nσ (σ = α. for any function f mπ holds (compare with the deﬁnition of a deﬁnite integral N0 as the limit of the corresponding Riemann sum) 1 N0 →∞ 2N0 lim N0 f m=−N0 +1 mπ N0 π = −π dk f (k) .192).203) is provided by the continuous function σ (k) = (k) + vn−σ . such that the larger nσ . In the limit N0 → ∞ this sum can be expressed as an integral. (9. nβ and µ.2a. The values of nα and nβ are determined by minimizing the energy density (9. i. subject to the constraint n = nα + nβ . i. Since β (k) − α (k) = v(nα − nβ ).205) is met.g.9.

210) Here δ(x) is the Dirac–delta function and ρ( ). unit energy and a given spin orientation of the particle. δ(k + π − arccos( /2t)) + δ(k − π + arccos( /2t)) (2t)2 − 2 . namely k1.2: (a) Relative position of the two energy sub–bands α (k) and β (k) for ∆ = nα − nβ > 0. Inserting this last result into (9. −π . Since the equation f (k) ≡ 2t cos k + = 0 has two simple roots.211) In order to calculate the integral we recall the following property of the Dirac–delta function δ [f (x)] = i δ(x − xi ) . (b) Graphical deﬁnition of the limiting energies α and β . The shaded areas represent the ﬁlled portions of these bands by electrons.278 Many–Particle Systems (a) (b) Figure 9. If the function f depends explicitly only on (k) one can state π −π dk f ( (k)) = 2π π ∞ ρ( )d f ( ) −∞ (9. In our case (k) is given by (9. which is usually called the density of states. and noting |f (k)| = |2t sin k| = 2t one obtains δ( − (k)) = 1 − cos2 k = (2t)2 − 2 . |f (xi )| (9. For a given dispersion law (k) the density of states ρ( ) can be calculated by employing equation (9.212) where xi are the simple roots of the function f (x). 2π (9.211) and using π dkδ(k ± (π − arccos( /2t))) = θ(2t − | |) .210). gives the the number of available one particle states per site. therefore. 2π (9.2 = ±(π − arccos( /2t)).209) where ρ( ) = −π dk δ ( − (k)) . holds π ρ( ) = −π dk δ( + 2t cos k) .192) and.

(9.8: Macroscopic Systems 279 where θ(x) is the step function. F = α + vnβ = β + vnα . µ. i. ∂ α ˜ ∂ Emf =0. We can employ the results obtained to express the ground state energy density (9. + vnα − µ = 0 . (9. ∂µ (9.213) can simply be replaced by 1. π α.214) where σ denotes the value of (k) which corresponds to the top of the ﬁlled portion of the energy sub–band σ (k) [c.206)] depends only on continuous quantities. (9.. 9. namely. β (9.216) Using (9. ∂ β ˜ ∂ Emf =0.214) are given by the sum (9.219–9.217) and µ. Replacing again the sum by an integral over energy one can write σ nσ = −2t ρ( )d (9.220) µ can be readily obtained µ= 1 ( 2 α + β) + vn .e. θ(x) = 1 if x > 0 and θ(x) = 0 if x < 0. (9. on conditions for a ground state of minimum energy are ˜ ∂ Emf =0..218) The derivatives can be readily determined and the conditions (9. in principle.e.213) The presence of the θ function in the above formula guaranties that the density of states vanishes for | | > 2t. From (9. the necessary ˜ Since Emf [c. (9.f.201). v and n.218) read ˜ ∂ Emf = ∂ α ˜ ∂ Emf = ∂ β and α + vnβ − µ = 0 .f. Fig. of t. outside the energy band (k).219..e.213) and carrying out the resulting integral yields nσ = 1 arcsin( σ /2t) . to determine the unknown quantities α . i.222) .216) allow one.219) (9.221) and (9. namely. nα and nβ as a function of the parameters of the Hubbard model.204) through an integral expression.220) β ˜ ∂ Emf = n − nα − nβ = 0 .2]. 2 (9. one arrives at the following expression of the density of states ρ( ) = θ(2t − | |) π (2t)2 − 2 . β . 2t[ then the θ function in (9. 9. (9.215) nα and nβ in (9.221) ∂µ Equations (9. One obtains ˜ Emf = α β ρ( )d + −2t −2t ρ( )d + vnα nβ + µ(n − nα − nβ ) . if we assume that is restricted to the interval ] − 2t. i.9.

.223) ∆ = 0 corresponds to the ground state (9. (9. First. (9. a magnetic ground state.223) follows α α − v β .222) and (9. (9. From the deﬁnition (9. non trivial.226) ∆ = f (∆) .228)] vc ρ ( β ) = 1 .226) follows that f (∆) is a monotonically increasing function of ∆. The Stoner criterion (9. through the condition [c. The slope of f (∆) at the origin depends on the Hubbard model parameter v. For this purpose we consider the magnetization of the mean ﬁeld ground state ∆ ≡ nα − nβ . Accordingly.227) will have a second. One can express ∆ as a function of α and β in two diﬀerent ways. if. If the slope of f (∆) at the origin is less than 1. the Hubbard model has a ground state with ∆ = 0.223 one obtains ∆= and second.2b). Figure 9.229) has a wider range of validity than one can infer from the present analysis of the Hubbard model.3]. For v above a critical value vc . f (∆) converges to a maximum value as ∆ → ∆max = n.217) and (9.e. from (9.227) can be solved graphically.e. (9.f. i. (9.227). Equation (9.225) Deﬁning the function v∆ f (∆) = 0 ρ( β + )d . (9. the slope of f (∆) is positive for all ∆ ≥ 0. from (9. For this purpose one plots f (∆) versus ∆ as is shown schematically in Figure 9.f.215) reveals that µ is indeed equal to the Fermi energy F (see also Figure 9. One still needs to determine α and β .228) Since the total number of states per site is ﬁnite. i. condition (9. (9.220) and (9.227) has only the trivial solution ∆ = 0. so called ferromagnetic solution..226) one can obtain ∆ by solving (cf. (9. We assume in the following discussion that the implicit ∆ dependence of β in f (∆) can be neglected. the slope of f (∆) at the origin is larger than 1 [c.3. Indeed.227) Since f (0) = 0 one can infer that ∆ = 0 is always a solution (the so called paramagnetic solution) of condition (9. . and only if.229) This cindition is known as the Stoner criterion and gives the critical value of v which determines the onset of the ferromagnetic long range order.224) v∆ ∆= β ρ( )d = 0 ρ( β + )d . One can determine vc by equating the slope of f (∆) at the origin to 1.219–9. as reﬂected by the expression d[f (∆)] = vρ( d∆ β + ∆) = π (2t)2 v − (v∆ + 2 β) >0. condition (9.183) without magnetization.280 Many–Particle Systems Comparision of (9. (9.

9. v (9. is presented in Figure 9.233) Employing the approximation sin(y + δ) − sin y ≈ δ cos y for small δ. (9.9..3: Graphical solution of the mean ﬁeld equation ∆ = f (∆). α.231) follows y= δ π π (n − 1) − ≈ (n − 1) . the plot of vc vs. t. n). n. The corresponding 2t analytical expression provides the phase diagram of the ground state: when v approaches vc from above.220. the ground state is characterized by the quantities (v.219. hence. Accordingly. By parameterizing α and β α = 2t sin x . i. (9.230) one obtains from (9.4.234) 2t 2 The resulting phase diagram of the ground state. any .e.205) the following set of equations determining x.232) states that x − y is proportional to ∆. We like to determine ﬁnally the dependence of vc on the ﬁlling factor n. 2t As already mentioned. Equation (9.232) For a given 2t the values of x and y can be obtained numerically and other relevant quantities v mentioned above can be calculated. β x+y x−y = = π(n − 1) 2t π (sin x − sin y) = π∆ . (9.233) yield the desired expression vc π = π cos (n − 1) .8: Macroscopic Systems 281 Figure 9. From (9.231) (9. 2 2 2 (9.232) together with (9. Since ∆ vanishes near vc we set x = y + δ where δ is a small quantity. 9. the magnetization ∆ of the ground state vanishes and remains zero for values v < vc . β = 2t sin y . y and.

even in these materials. Concluding Remarks At the end of our analysis of the mean ﬁeld ground state of the one-dimensional Hubbard model. . can lead to serious modiﬁcations of the mean ﬁeld ground state. The magnetic nature of the ground state of the system depends on whether the representative point lies inside the ferromagnetic or inside the paramagnetic domain of the phase diagram. the method described above gives a systematic way of singling out the state with the lowest possible energy which might be a good candidate for the real ground state of the system. For example. The real ground state of the one-dimensional Hubbard model is quite diﬀerent in nature from the mean ﬁeld ground state discussed here. in one spatial dimension is very strong. strong electron correlation eﬀects . Figure 9. Nevertheless. the eﬀect of quantum ﬂuctuations is getting less important as the dimensionality of the system is increased.282 Many–Particle Systems mean ﬁeld ground state is represented by a point in the phase diagram. First. Well. However.202). the application of the mean ﬁeld theory for these systems provides an excellent testing opportunity of these theories by comparing their predicted results with the exact ones. the analysis presented above is not quite useless. In general. which are neglected in the mean ﬁeld theory. the mean ﬁeld theory of the one-dimensional Hubbard model can be extended in a straightforward way to higher spatial dimensions. it is natural to ask oneself if the results obtained reﬂect. Second. once we specify a class of possible states to which the real ground state might belong to. Third. The reason is that the eﬀect of quantum ﬂuctuations. in three spatial dimensions the theory presented above works ﬁne in the case of the transitional–metal oxides such as NiO and CoO.4: Ground state phase diagram of the mean ﬁeld Hamiltonian (9. since in one spatial dimension there are many exact results available. the answer is no. at least qualitatively. the properties of the real ground state of the Hubbard model.

The available mean ﬁeld theories cannot account for all the striking. In fact.9. in the case of the copper oxide high temperature superconductors.8: Macroscopic Systems 283 In case of two spatial dimensions things are more complicated. and on the other hand. the lack of exact solutions. On the one hand. which involve strongly correlated quasi two-dimensional electron systems. a reliable microscopic theory is still lacking. unusual physical properties of these materials. . the strong eﬀect of quantum ﬂuctuations of the strongly correlated electron system described by the Hubbard Hamiltonian makes all the presently existing solutions questionable.

284 Many–Particle Systems .

Obviously.e.45)] described by an electrical potential V (r. The transformations beween such frames according to the Theory of Special Relativity are described by Lorentz transformations. the components representing the spin attribute of particles and also representing together with a particle its anti-particle.. It is not possible to uncouple the equations to describe only a single type particle without aﬀecting negatively the Lorentz invariance of the equations.2) The replacement of (10. We will introduce below Lorentz-invariant diﬀerential equations which take the place of the Schr¨dinger equation of a paro ticle of mass m and charge q in an electromagnetic ﬁeld [c. these ˆ transformations are x1 = x1 − v1 t 1 − v1 2 c . t = t− 1 − v1 x c2 1 v1 2 c . x3 .. We will ﬁnd that actually several Lorentz–invariant equations which replace (10. the equations need to be interpreted as actually describing many–particle-systems: the equivalence of mass and energy in relativistic formulations of physics allows that energy converts into particles such that any particle described will have ‘companions’ which assume at least a virtual existence. x3 . spin–0 particles. x2 = x2 . some of the equations describe a particle together with its anti-particle. t) ∂ i ψ(r. t) are vectors of dimension larger one. spin– 1 particles. x2 . x2 . t) c 2 i + qV (r. etc.Chapter 10 Relativistic Quantum Mechanics In this Chapter we will address the issue that the laws of physics must be formulated in a form which is Lorentz–invariant. In case that v is oriented along the x1 –axis. Here c denotes the velocity of light. t) in one frame with space time coordinates (x1 .f. i.e.2) will result. t) (10. Furthermore. 2 As mentioned.1) which connect space time coordinates (x1 . (refeq:ham2. it will be necessary to begin this Chapter with an investigation of the group of Lorentz transformations and their representation in the space of position r and time t. 8.g..2) by Lorentz–invariant equations will have two surprising and extremely important consequences: some of the equations need to be formulated in a representation for which the wave functions ψ(r. x3 = x3 (10. t) ψ(r. t) and a vector potential A(r. any of these equations being speciﬁc for certain classes of particles. The representation 285 . i. the description should not allow one to diﬀerentiate between frames of reference which are moving relative to each other with a constant uniform velocity v. v = v1 x1 . e. t ) in another frame. t) = ∂t 1 2m q − A(r.

e. Denoting by xµ the . for the transformed space-time–cordinates x µ = (x 0 . x 2 . This will provide us with a general set of Lorentz–invariant equations which for various particles take the place of the Schr¨dinger equation. This space is called the Minkowski space. 5 to the groups SO(3) and SU(2) of rotation transformations of space coordinates and of spin.4 to cover ﬁelds. 10. t) as components. Note that the components of xµ all have the same dimension.1). x 1 . namely that of length. Minkowski Space Historically. x2 .5. x3 ) (10.e. This choice implies dim(time) = dim(length). Before introduco ing these general Lorentz–invariant ﬁeld equations we will provide in Sects. however. The reason for this choice is that the transformations (10. i.7 a heuristic derivation of the two most widely used and best known Lorentz–invariant ﬁeld equations. 10.1) leave the quantity s2 = (x0 )2 − (x1 )2 − (x2 )2 − (x3 )2 (10.1).286 Relativistic Quantum Mechanics in Sect.e. 10. assume new units for time such that the velocity of light c becomes c = 1. x3 ) = r describes the space coordinates. wave functions ψ(r. t) and vectors with functions ψ(r.1 Natural Representation of the Lorentz Group In this Section we consider the natural representation of the Lorentz group L. We will ﬁnd that this deﬁnition will lead us back to the transformation law (10. i. we will provide a more basic deﬁnition of the transformations.1). Rather than starting from (10. The Group of Lorentz Transformations L = O(3.5) and the Dirac (Sect.7) equation.4) invariant. In such a space Lorentz transformations correspond to 4-dimensional rotations. We will. x 3 ) holds (x0 )2 − (x1 )2 − (x2 )2 − (x3 )2 = (x )2 − (x )2 − (x )2 − (x )2 . We will denote these vectors as follows def xµ = (x0 .1 will be extended in Sect..5) √ One can interprete the quantity −s2 as a distance in a 4–dimensional Euclidean space if one chooses the time component purely imaginary. 10. Rather than following this avenue we will introduce Lorentz transformations within a setting which does not require real and imaginary coordinates.1) The Lorentz transformations L describe the relationship between space-time coordinates xµ of two reference frames which move relative to each other with uniform ﬁxed velocity v and which might be reoriented relative to each other by a rotation around a common origin. The elements L ∈ L act on 4–dimensional vectors of position– and time–coordinates. 0 1 2 3 (10. 10. i.3) where x0 = ct describes the time coordinate and (x1 . x1 . 10. namely the Klein–Gordon (Sect. but in a setting of representation theory methods as applied in Secti. the group of Lorentz transformations (10. the Lorentz transformations were formulated in a space in which the time component of xµ was chosen as a purely imaginary number and the space components real. x2 . 10. henceforth..

the notation in (10. the latter set constituting a group. Aµ ν xν = µ=0 new old Aµ ν xν . This will be explained further below. i.6) allows us to introduce the so-called summation conventon: any time the same index appears in an upper and a lower position. We will exploit this property below to determine the general form of the Lorentz transformations.e. 1. i.. The upper index closer to ‘L’ denotes the ﬁrst index of the matrix and the lower index ν further away from ‘L’ denotes the second index. R).11) This condition speciﬁes the key property of Lorentz transformations. Aµ ν B ν ρ = ν=0 Aµ ν B ν ρ . (10. Aµν . This condition can be written xµ gµν xν = x gµν x where µ ν (10.12) (10.9) 1 0 0 0 −1 0 0 ( gµν ) = 0 0 −1 0 = g . (10.6) x µ = Lµ ν xν .e. [ A more conventional notation would be Lµν .10) Combining condition (10.7) The reason for the notation is two-fold. the .10.8) new old old The summation convention allows us to write (10. 3: 4-vector: xµ .5) invariant. 2. R). Aµν . the latter notation will be used for diﬀerent quantities further below. First. ν=0 (10.1: Natural Representation of the Lorentz Group 287 coordinates in one reference frame and by x µ the coordinates in the other reference frame.6) Here Lµ ν are the elements of a 4 × 4–matrix representing the Lorentz transformation. 3 3 3 yµ xµ = new yµ xµ .5) into scalar products. The subset of GL(4. Aµ ν . The Lorentz transformations form the subgroup of all matrices which leave the expression (10. (10. xµ . The Lorentz transformations are non-singular 4 × 4–matrices with real coeﬃcients.. L ∈ GL(4. 0 0 0 −1 (10.9) and (10. the Lorentz transformations constitute a linear transformation which we denote by 3 x µ = ν=0 Lµ ν xν . summation over that index is assumed without explicitly noting it. ν = 0. however.6) yields Lµ ρ gµν Lν σ xρ xσ = gρσ xρ xσ . Since this holds for any xµ it must be true Lµ ρ gµν Lν σ = gρσ . 4 × 4 tensor: Aµ ν .] The following possibilities exist for the positioning of the indices µ. The second reason is that upper and lower positions allow us to accomodate the expression (10.

= g LT g = 0 1 2 −L 2 L 2 L 2 L3 2 −L0 3 L1 3 L2 3 L3 3 (10. One can also show that if L ∈ O(3. This stipulates that O(3. This set is identical with the set of all Lorentz transformations L.e.18) (10. i. the associated inverse obeys (10. (10.10) of g one can rewrite the invariance property (10. the inverse of L 1 L−1 (10. L3 0 L3 1 L3 2 L3 3 Using the deﬁnition (10. We want to show now L = O(3. 10. L−1 ∈ O(3. In fact.12) LT gL = g .16) The corresponding expression for (LT )−1 is obviously (LT )−1 = (L−1 )T = g L g .h. 1) = { L.21) i.1) is.22) (10. i. in fact. To demonstrate the group property of O(3.1).20) results in the desired property (L−1 )T g L−1 = g .19) i.20) Multiplying (10. 3 2 1 2 1 2 (10. To simplify the following proof of the key group properties we like to adopt the conventional matrix notation for Lµ ν 0 L 0 L0 1 L0 2 L0 3 L1 L1 1 L1 2 L1 3 (10. .1).17) one obtains (L−1 )T g L−1 = gLgggLT g = gLgLT g .1) since it satisﬁes (10. L2 ∈ O(3. of (10. a group. R) is a group. L3 ∈ O(3.15) one can derive LT gLgLT = LT and multiplying this from the left by by g(LT )−1 yields L g LT = g Using this result to simplify the r. hence. For L3 = L1 L2 holds LT g L3 = LT LT g L1 L2 = LT (LT gL1 ) L2 = LT g L2 = g . LT gL = g } .14) holds for the inverse of L. L ∈ GL(4.e..1).13) L = ( Lµ ν ) = 2 0 L 0 L2 1 L2 2 L3 3 .15) (10.14). of O(3.e. We consider 1 then L1 . employing expressions (10.14). (10.e..s.288 Relativistic Quantum Mechanics elements of which satisfy this condition. R).1).16.17) we note ﬁrst that the identity matrix 1 is an element of O(3. From this one can obtain using g2 = 1 1 (gLT g)L = 1 and...1). property (10. is called O(3.1).14) from the right by gLT and using (10. (10.14) L0 0 −L1 0 −L2 0 −L3 0 −L0 1 L1 1 L2 1 L3 1 .1) ⊂ GL(4.

+ We start our investigation by demonstrating that L↑ forms a group. L↓ . A statement on this value can be made on account of property (10.12).27) (10. 1). For this purpose we consider ﬁrst the value of det L. Properties (10. (10.32) (10. Using det AB = det A det B and det AT = det A yields (det L)2 = 1 or det L = ±1 . Considering in (10. From this we can conclude L0 0 ≥ 1 or L0 0 ≤ −1 . = { L.31) (10. + = − L↑ g + (10. The above shows that the set of proper Lorentz transformations L↑ allows one to generate all Lorentz transformations. det L = −1. L↓ are disjunct sets deﬁned as follows + + − − L↑ + L↓ + L↑ − L↓ − = { L.25) i. ∃M2 . Lorentz transformations in L↑ . (10.23) One can classify Lorentz transformations according to the value of the determinant into two distinct classes. L↑ .14).30) ≤ −1} .28) (10. 1).. 1). det L = −1. L ∈ O(3. + − gL↑ + = L↑ g . entirely suitable to investigate ﬁrst only 1. (10.34) j=1 j=1 . For this purpose we consider the value of C 0 0 = A0 µ B µ 0 = 3 A0 j B j 0 + A0 0 B 0 0 . Schwartz’s inequality yields j=1 2 3 3 3 j=1 A jB 0 ≤ 0 j A 0 2 j Bj 0 2 .23) and (10. L ∈ O(3. + except for the trivial factors g and −1 It is. there exist two other distinct classes.24) or since (L1 0 )2 + (L2 0 )2 + (L3 0 )2 ≥ 0 it holds (L0 0 )2 ≥ 1. (10. Obviously. L ∈ O(3.33) where we used the deﬁnition aM = {M1 . = { L. One can also 1 verify that one can write L↑ − L↓ + L↓ − = gL↑ + = = −L↑ . det L = 1. B ∈ L+ holds C = AB ∈ L+ . L0 0 ≤ −1} . = { L. L ∈ O(3. M1 = a M2 }.1: Natural Representation of the Lorentz Group Classiﬁcation of Lorentz Transformations 289 We like to classify now the elements of L = O(3. L0 0 ≥ 1} .12) the case ρ = 0.25) can be stated as follows: The set of all Lorentz transformations L is given as the union L = L↑ ∪ L↓ ∪ L↑ ∪ L↓ + + − − where L↑ .10.1). + + ↑ ↑ We can also demonstrate that for A. L0 0 ≥ 1} . 1). L↑ contains 1 1.14). L0 0 (10. σ = 0 yields L0 0 2 − L1 0 2 − L2 0 2 − L3 0 2 = 1. A second class property follows from (10.29) (10. det L = 1.26) (10. hence. M2 ∈ M.14) which we employ in the formulation (10. It holds g ∈ L and −1 ∈ L as one can readily verify testing for property (10.e.

1) ascertains CT gC = g it must hold C 0 0 ≥ 1. .12) follows (B 0 0 )2 − 3 (B j 0 )2 = 1 or 3 (B j 0 )2 = (B 0 0 )2 − 1. holds L0 0 > 0 and the value of the determinant is close to unity. obviously j=1 A0 0 B 0 0 ≥ 1.40) and. since the group property of O(3. + transformations in a small neighborhood of 1 which we parametrize by inﬁnitesimal parameters. Inﬁnitesimal Lorentz transformations The transformations in L↑ have the property that they are continously connected to the identity 1 1. 1 0 0 = ≥ 1 and.35) One can conclude. if we enforce (10..16). µ ν small . these transformations can be parametrized such that a continuous variation of the parameters connects any element of L↑ with 1 This property will be exploited now in that we consider ﬁrst 1.36) It holds.1). δ00 = 1 and It should be noted that according to our present deﬁnition holds δµν = gµρ δ δ11 = δ22 = δ33 = −1. L↑ is called the subgroup of proper.13).37) (10. | 3 A0 j B j 0 | < A0 0 B 0 0 . This relationship shows (L−1 )0 0 = L0 0 .14) actually L0 0 ≥ 1 and det L = 1 must hold. we consider transformations Lµ ν = δ µ ν + µ ν . one can j=1 j=1 3 0 )2 = (A0 )2 − 1.1).21) j=1 (A j 0 3 j=1 A0 j B j 0 ≤ 2 (A0 0 )2 − 1 (B 0 0 )2 − 1 < (A0 0 )2 (B 0 0 )2 . The next group property of L↑ to be demonstrated is the existence of the inverse. 1 ∈ L↑ . For the inverse of + ↑ any L ∈ L+ holds (10. hence. (10.14) implies 1 + T g (1 + ) = g 1 1 (10. + i.e. Using the above expression for C 0 0 one can state C 0 0 > 0. (10. Similarly. Since A0 0 ≥ 1 and B 0 0 ≥ 1. 1 We will then employ the Lie group properties to generate all transformations in L↑ . In the following we + will consider solely this subgroup of SO(3. therefore. accordingly. + Accordingly.e.34) provides then the estimate conclude from (10.. Since the associative property holds for matrix multipli1 1 + ↑ cation we have veriﬁed that L+ is indeed a subgroup of SO(3. i. ρ ν (10. (10. In fact. from which one can conclude L−1 ∈ L↑ . obviously. orthochronous Lorentz transformations.39) where we have employed the matrix form T 1 deﬁned as in (10.38) For these transformations. We also note that the identity operator 1 has elements 1 + 1 µν = δµν 1 where we deﬁned1 δµν = 1 for µ = ν 0 for µ = ν (10. To order O( 2 ) holds g + g = 0.290 Relativistic Quantum Mechanics From (10. Property (10.

K3 = 0 −1 0 0 0 0 0 0 0 0 0 0 0 −1 0 0 0 0 These commutators obey the following commutation relationships [ Jk . J2 = 0 0 0 −1 0 0 0 0 0 0 0 0 0 0 0 −1 0 1 . 2. ϑ2 .50) .47) 0 −1 0 0 0 0 −1 0 0 −1 0 0 0 0 0 0 0 0 K1 = . w3 ) = −w2 ϑ3 0 −ϑ1 −w3 −ϑ2 ϑ1 0 (10. w2 . 2. = − k m Km (10. K2 = . 3) Jk = (ϑk = 1.44) This result allows us now to deﬁne six generators for the Lorentz transformations(k = 1. 3 (10. j.41) which reads explicitly 0 0 0 0 0 1 2 3 1 1 1 1 0 1 2 3 2 2 2 2 0 1 2 3 3 3 3 3 0 1 2 3 = − 0 1 2 3 0 0 0 0 0 1 1 1 1 0 2 2 2 2 0 3 3 3 3 − − − 1 2 3 − − − 1 2 3 − − − 1 2 3 .42) This relationship implies µ 0 j µ j = 0 = j 0 . w1 . (ϑ1 . The generators are explicitly 0 0 0 0 0 0 0 0 J1 = 0 0 0 −1 0 0 1 0 (10. 2.1: Natural Representation of the Lorentz Group Using (10. J = 0 3 0 1 0 0 0 0 0 0 0 0 0 0 0 (10. ϑ3 .46) 0 0 0 0 . K ] [ Jk . (10.10. other ﬁve parameters zero) Kk = (wk = 1. J ] [ Kk .48) (10. k = 1.45) (10. 3 .15) one can conclude T 291 = −g g (10. k j j = 1. K ] The operators also obey J · K = J1 J1 + J2 J2 + J3 J3 = 0 = = k m Jm k m Jm 0 −1 0 0 0 0 0 0 (10.43) k = − Inspection shows that the matrix has 6 independent elements and can be written 0 −w1 −w2 −w3 −w1 0 −ϑ3 ϑ2 .49) . other ﬁve parameters zero) .

For the parameters ϑk of the Lorentz transformations holds obviously ϑk ∈ [0. 3 is closed.51) for w = 0 correspond to rotations of the spatial coordinates. that the transformations (10. 10.49. k = 1. (10. w) = exp ϑ · J + w · K . 2.52) where the 3 × 3–matrices Lk are the generators of SO(3) deﬁned in Chapter 5. 2. 2π[ . and using the relationship Jk = 0 0 0 Lk (10. w ∈ R3 3 k=1 wk Kk .35) it issuﬃcient to consider in the present case the 2 × 2–matrix L = exp since w1 0 −1 −1 0 ∞ = n=0 n w1 n! 0 −1 −1 0 0 0 . Kk . A boost in the x1 –direction is L = exp(w1 K1 ). 3 (10. Following the treatment of the rotation group SO(3) one can express the elements of L↑ through + the exponential operators L(ϑ. In analogy to equation (5. + The commutation relationships imply that the algebra of the generators Jk .56) = 1 0 0 1 = 1 1 (10. Exercise 7. i. 0 1 n (10.51) One can readily show..57) . (10. k = 1. Relativistic Quantum Mechanics The commutation relationships (10. however. w = 0) = 0 0 0 R(ϑ) . constitutes an overcomplete parametrization of the rotations (see Chapter 5).53) Here R(ϑ) are the 3 × 3–rotation matrices constructed in Chapter 5. L(ϑ.54) which.1: Demonstrate the commutation relationships (10.55) L 0 0 0 0 exp (w1 K1 ) = exp Using the idempotence property 0 −1 −1 0 2 0 0 1 0 (10. 3 where we have deﬁned ϑ · J = k=1 ϑk Jk and w · K = following the algebra in Chapter 5.e. We consider now the Lorentz transformations for ϑ = 0 which are referred to as ‘boosts’.49) deﬁne the Lie algebra associated with the Lie group L↑ . To determine the explicit form of this transformation we evaluate the exponential operator by Taylor expansion.50).292 as can be readily veriﬁed. ϑ.

cosh2 w1 − 1 and coshw1 = cosh2 w1 − 1 = cosh w1 sinh w1 sinh2 w1 + 1 .1: Natural Representation of the Lorentz Group one can carry out the Taylor expansion above: ∞ 293 L = n=0 2n w1 1 + 1 (2n)! ∞ n=0 2n+1 w1 (2n + 1)! 0 −1 −1 0 = cosh w1 −sinh w1 −sinh w1 cosh w1 . (10. This property of the wk -values contrasts with the property of the parameters ϑk specifying rotational angles which assume only values in a compact range.58) = cosh w1 1 + sinh w1 1 0 −1 −1 0 The conventional form (10. vk deﬁned in (10.59) v1 = These two equations yield cosh w1 = 1/ and (10. consistent with (10.3. x3 = x3 (10.59) follows.56. (10.60) sinh w1 = v1 / 2 1 − v1 .10. We expect that vk can only assume values less than the velocity of light c which in the present units is c = 1. The range of the parameters wk can now be speciﬁed.1) of the Lorentz transformations is obtained through the parameter change sinh w1 v1 = = tanh w1 (10. . in fact.63) which agrees with (10. 1[. 10. one obtains from (10.62) According to (10. ∞[ .59) cosh w1 Using cosh2 w1 − sinh2 w1 = 1 one can identify sinhw1 = sinh2 w1 + 1.64) We note that the range of wk -values is not a compact set even though the range of vk -values is compact.59) for the case k = 1 corresponds to the relative velocity of two frames of reference. (10.59) can be written 1 2 1 − v1 √−v1 2 1 − v1 2 1 − v1 .59). we can state vk ∈ ] − 1. From (10. Accordingly. Correspondingly.6. for wk wk ∈ ] − ∞. 10. (10. 10. This property is. t = t − v1 x1 2 1 − v1 . however.1).61) √ √−v1 √ exp (w1 K1 ) = 2 1 − v1 1 2 1 − v1 0 0 0 0 0 0 0 0 1 0 0 1 (10.51) the explicit transformation for space–time–coordinates is then x1 = x1 − v1 t 1 − 2 v1 . x2 = x2 .

we will encounter an impressive example of physical properties.4). Presently. ﬁnally.3) are the so-called 4–vectors aµ .65) An example is s2 deﬁned in (10. x2 .67) the transformation behaviour of which we will demonstrate further below. i. . E = √ m mv . we have not yet deﬁned representations of Lorentz transformations beyond the ‘natural’ representation acting in the 4–dimensional space of position– and time–coordinates. namely. Such quantities appear in the description of physical systems and statements about transformation properties are often extremely helpful and usually provide important physical insight. t) are the electrical and the vector potential of an electromagnetic ﬁeld. a1 . of a system of charges.e.2 Scalars. t) and J(r. p = √ 2 1−v 1 − v2 (10.69) where V (r. scalars like r = x2 + x2 + x2 . r2 ) and. J) (10. A third 4-vector is the so-called current vector J µ = (ρ. Counterexamples are the energy. t)|2 or the scalar product r1 · r2 of two particle positions. spherical harmonics Y m (ˆ). Some of these r ˆ quantities were actually deﬁned with respect to representations of the rotation group in function spaces. the charge density. 2 |ˆ1 . not any physical property f ∈ R is a scalar. A) (10. tensor operators Tkm . We have encountered examples in connection with rotational transformations.294 Relativistic Quantum Mechanics 10. These quantites always come as four components (a0 . the square of the electric ﬁeld |E(r..66) Examples of 4-vectors beside xµ are the momentum 4-vector pµ = (E. t) are the charge density and the current density. However. Nevertheless. not in the so-called natural representation associated with the 3–dimensional Euclidean space E3 . We will see below how true scalars under Lorentz transformations can be constructed. Scalars The quantities with the simplest transformation behaviour are so-called scalars f ∈ R which are invariant under transformations. respectively. Another example is the potential 4-vector Aµ = (V. another example is the rest mass m of a particle. 4-Vectors The quantities with the transformation behaviour like that of the position–time vector xµ deﬁned in (10. x3 )T . 4–Vectors and Tensors In this Section we deﬁne quantities according to their behaviour under Lorentz transformations. (10. f = f. the z–component x3 of a particle. p) . (10. a3 )T and transform according to a µ = Lµ ν aν . t) and A(r. total angular momentum r 1 2 3 states of composite systems like Y m ( 1 . The 4-vector character of J µ and of Aµ will be demonstrated further below.68) where ρ(r. a2 . our deﬁnition of quantities with special properties under Lorentz transformations presently is conﬁned to the natural representation. vectors like r = (x1 . Hence.

78) To prove the 4-vector property of ∂µ we will show that g µν ∂ν transforms like a contravariant 4vector. g µν ∂ν = Lµ ρ g ρσ ∂σ . ∂t (10.e.73) Note that according to (10. If aµ and bµ are 4-vectors aµ gµν bν (10. gσν xν = Lρ σ gρµ x µ and g µσ gσν = δ µ ν . (10. one can employ the tensors g µν to raise and lower indices of L ν establish here the consistency of the ensuing notation. (10.66) together with (10. −a3 ) (10.71) Contravariant and Covariant 4-Vectors It is convenient to deﬁne a second class of 4-vectors.66). We have duplicated the expression for the inverse of Lµ ν to obtain the correct notation in terms of covariant. one can express [(L−1 )T ]µ ν = (L−1 )ν µ and.75) (10. Covariant 4-vectors transform like a µ = gµν Lν ρ g ρσ aσ where we deﬁned g µν = gµν . The 4-Vector ∂µ (10. We like to point out that from deﬁnition (10. Multiplication (and summation) of x µ = Lµ ν xν by Lρ σ gρµ yields.16). We do not fact. using (10.10. We start from x µ = Lµ ν xν .70) is a scalar. The respective vectors aµ are associated with the 4-vectors aµ . the relationship being aµ = gµν aν = (a0 . In fact.17) can be written (L−1 )ν µ = Lµ ν . This property follows from (10. i. In any case one can express (10. xν = g νσ Lρ σ gρµ x µ .72) of the covariant 4-vector follows = In µν and g µ as well.76) An important example of a covariant 4-vector is the diﬀerential operator ∂µ = ∂ = ∂xµ ∂ .e.74) g µν aν . 4–Vectors and Tensors Scalar Product then 295 4-vectors allow one to construct scalar quantities.. i.12)..2: Scalars. lower.77) The transformed diﬀerential operator will be denoted by ∂µ = def ∂ .12) a gµν b µ ν = Lµ ρ gµν Lν σ aρ bσ = aρ gρσ bσ (10.79) This is the inverse Lorentz transformation consistent with (10. aµ (10.72) where aν is a vector with transformation behaviour as stated in (10. (10. and . accordingly. ∂x µ (10.73) aµ = Lµ σ aσ .17) Lµ σ is the transformation inverse to Lσ µ . −a2 . One calls 4-vectors aµ covariant and 4-vectors aµ contravariant. −a1 .

1 − v2 1 − v 2 dt (10.79) allows us to determine the connection between ∂µ and ∂µ .g.67) transforms like (10. dτ 1 − v 2 dt p0 = E = √ m m dt = √ . For this purpose we consider the scalar diﬀerential (dτ )2 = dxµ dxµ = (dt)2 − (dr)2 It holds dτ dt from which follows 2 (10. Using the chain rule of diﬀerential calculus we obtain 3 ∂µ = ν=0 ∂xν ∂ = g νσ Lρ σ gρµ ∂ν = Lµ ν ∂ν ∂x µ ∂xν (10. (10. 2 2 dt 1 − v 1 − v (10.82) (10.86) One can write (10.84) = 1 − (v)2 (10.e. e. d’Alembert Operator We want to construct now a scalar diﬀerential operator. For this purpose we deﬁne ﬁrst the contravariant diﬀerential operator ∂ µ = g µν ∂ν = Then the operator 2 ∂µ ∂ µ = ∂t − 2 (10.89) .− ∂t . i. (10.. indices. ∂µ does indeed transform like a covariant vector. this operator is equal to the d’Alembert operator which is known to be Lorentz-invariant.296 Relativistic Quantum Mechanics contravariant. In fact. i.. p1 = √ m v1 m dx1 = √ .80) Multiplication by g λµ (and summation over µ) together with g λµ gρµ = δ λ ρ yields g λµ ∂µ = Lλ σ g σν ∂ν . 2 dt dτ 1 − v (10.85) 1 d d = √ .88) One can express then the momentum vector pµ = √ d m dxµ = m xµ .. Proof that pµ is a 4-vector We will demonstrate now that the momentum 4-vector pµ deﬁned in (10.81) ∂ .66).e.87) The remaining components of pµ can be written. upper.83) is a scalar under Lorentz transformations.

. m2 m2 v 2 − = m2 1 − v2 1 − v2 (10. for these potentials. . namely.10. This gauge is not Lorentz-invariant and we will adopt here another gauge. and had chosen the so-called Coulomb gauge (8. of (10. namely pµ pµ = pµ gµν pν = E 2 − p 2 Evaluation of the r. the so-called Lorentz gauge.67) E2 − p 2 = or pµ pµ = m2 which.89) must be a 4-vector.77) which allow one to express (10.93) (10. t) + · A(r. hence. in fact. One obtains E = ±m ± p2 2m (p 2 )2 + O 4m3 (p 2 )3 4m5 .3 Relativistic Electrodynamics In the following we summarize the Lorentz-invariant formulation of electrodynamics and demonstrate its connection to the conventional formulation as provided in Sect. If we can show that Aµ deﬁned in (10.s. We will demonstrate now the 4-vector property of Aµ . yields according to (10. respectively.94) 1 m In the non-relativistic limit the rest energy m is the dominant contribution to E.90) (10.91) (10. The momentum 4-vector allows us to construct a scalar quantity.89) alltogether transforms like a contravariant 4-vector.h.h.96) The Lorentz-invariance of this gauge. 8. We like to rewrite the last result E 2 = p 2 + m2 or E = ± p 2 + m2 . in (10. Lorentz Gauge In our previous description of the electrodynamic ﬁeld we had introduced the scalar and vector potential V (r.h. ∂t V (r.s. kinetic energy ± 2m and relativistic corrections. is a scalar. can be demonstrated readily using the 4-vector notation (10. the r. (10. 10. and. · A = 0. equivalently. a contravariant 4-vector then the l. hence. t). (10. i.92) (10.3: Relativistic Electrodynamics 297 d The operator m dτ transforms like a scalar. in fact.s. of (10. Expansion in should then be rapidly convergent.97) We have proven already that ∂µ is a contravariant 4-vector.e.69) for the electrodynamic potential and the 4-vector derivative (10.97) and.s.96) is a scalar and.h. (10. t) = 0 .96) in the form ∂µ Aµ = 0 . in (10.95) 2 p This obviously describes the energy of a free particle with rest energy ±m. t) and A(r. Since xµ transforms like a contravariant 4-vector. pµ on the l. Lorentz-invariant.12).69) is.

e.18) and.9).82) as follows F µν = ∂ µ Aν − ∂ ν Aµ . (10. yields ∂µ ∂ µ Aν (xσ ) = 4 π J ν (xσ ) . a 4-vector. t) + · J(r.69). 2.298 Transformation Properties of J µ and Aµ Relativistic Quantum Mechanics The charge density ρ(r. t) = −∂t V (r. t) = −∂t V (r. Equation (10. F µν = x Ey B z 0 −Bx Ez −By Bx 0 Alternatively. (8.96).96).98) which reﬂects the principle of charge conservation. The l. k. t).h. t) − 2 V (r. . t) and current density J(r. using (8. t) from (8. (10.32). The Field Tensor The electric and magnetic ﬁelds can be collected into an anti-symmetric 4×4 tensor 0 −Ex −Ey −Ez E 0 −Bz By . using (10.99) Since this equation must be true in any frame of reference the right hand side must vanish in all frames. m.83) and (10. Since ∂µ ∂ µ transforms like a scalar one can conclude that Aν (xσ ) must transform like a 4-vector. one obtains for A(r. prove that Aµ is. also the l.17) using the identity (8.77) and (10. n = 1. it follows that J µ .102) In this equation the r. ∂µ J µ (xµ ) = 0 . (10. this can be stated F k0 = −F 0k = E k . Consequently. The respective equation for A0 = V can be obtained from Eq.105) . t) = 4πρ(r. i. 10. t) together with (10. · A(r.. i. t) . t) − 2 A(r. One can readily verify. t) 2 ∂t A(r.h.s. F mn = − mn (10.97) and.13). using (10.98) can be written.s. 3 (10. of (10. transforms like a 4-vector. that F µν can be expressed through the potential Aµ in (10. (10. This principle should hold in any frame of reference. in fact. t) are known to obey the continuity property ∂t ρ(r.101).h. t) = ∂t · A(r.104) where mn = mn is the totally anti-symmetric three-dimensional tensor deﬁned in (5. must be a scalar. t) = 4 π J(r. · A(r. according to (10.68). has to transform like a contravariant 4-vector. must transform likewise. one obtains 2 ∂t V (r.100.103) B .99) must be a scalar.s.6) and (8.101) Combining equations (10. in fact. t) . Using · ∂t A(r.e. Since ∂µ transforms like a covariant 4-vector.. (10. thereby. t) = 0 (10.69) and ∂ µ in (10.100) Similarly. We want to derive now the diﬀerential equations which determine the 4-potential Aµ in the Lorentz gauge (10.

4). equivalently. 10. B . Noting ∂µ F µν = ∂µ ∂ µ Aν − ∂µ ∂ ν Aµ = ∂µ ∂ µ Aν − ∂ ν ∂µ Aµ = ∂µ ∂ µ Aν .108) yields then the expressions for the transformed ﬁelds E and B . 8. the components of the ﬁelds parallel and perpendicular to the velocity v which determines the Lorentz transformation. .104) establishe the transformation behaviour of E(r. B and E⊥ .102) ∂µ F µν = 4π J ν . E⊥ = B⊥ E⊥ + v × B √ 1 − v2 B⊥ − v × E = √ 1 − v2 (10.106) In case that the Lorentz transformation Lµ ν is given by (10.103. These equations show that under Lorentz transformations electric and magnetic ﬁelds convert into one another.3: Relativistic Electrodynamics 299 The relationships (10. The results can be put into the more general form E B = = E .63).97). Maxwell Equations in Lorentz-Invariant Form One can express the Maxwell equations in terms of the tensor F µν in Lorentz-invariant form.110) where E . t).3.62) or.10.1.113) which can be shown to be equivalent to the two homogeneous Maxwell equations (8. From the deﬁnition (10. one obtains Ey −v B Ez +v B 0 −Ex − √ 1 2z − √ 1 2y 1−v1 1−v1 Bz −v1 Ey By +v1 Ez √ 2 Ex 0 − √ 2 1−v1 1−v1 F µν = E −v B (10.111) One can readily prove that this equation is equivalent to the two inhomogeneous Maxwell equations (8. respectively. 8. (10. t) and B(r. In a new frame of reference holds F µν = Lµ α Lν β F αβ (10.112) (10. where we used (10. by (10.105) and (10. one can conclude from (10.109) (10.107) y 1 z Bz −v1 Ey √ 2 √ 2 0 −Bx 1−v1 1−v1 Ez +v1 By By +v1 Ez √ 2 − √ 2 Bx 0 1−v1 1−v1 Comparision with F µν 0 −Ex −Ey −Ez E 0 −Bz By = x Ey B z 0 −Bx Ez −By Bx 0 (10. B⊥ are.2).105) of the tensor F µν one can conclude the property ∂ σ F µν + ∂ µ F νσ + ∂ ν F σµ = 0 (10.

86) and (10.120) (10. hence. hence.119) which is the x-component of the equation of motion (8. also from the equation of motion (8.e. of (10.87)] and retain the deﬁnition E for the electric ﬁeld.114) is. We have.120) is referred to as the Lorentz force. dE q = p·E dτ m or with (10. The µ = 0 component of (10.h.5).118) p· dt √ where we exploited the fact that according to p = mv/ 1 − v 2 holds p v. demonstrated that (10. 1 dE 2 1 dp 2 = = qp·E 2 dt 2 dt (10. provides an alternative description of the action of the Lorentz force.114) reads then.s. dτ m √ Employing again (10.103) dpx q = ( EEx + py Bz − pz By ) . It holds for a particle with 4-momentum pµ as given by (10.67). For the spatial components.114).5).117) (10.86). (10.114) where d/dτ is given by (10.5) taking the scalar product with p dp = qp · E (10.. E = m/ 1 − v 2 .114) reads using (10. equivalent to (8. yields dpx = q dt Ex + (v × B)x (10. . using (10.67) and charge q dpµ q = pν F µν dτ m (10.5) where p = mv/ 1 − v 2 .104).116) (10.. in fact. i. for µ = 1. The term on the r. dt E From this one can conclude.300 Lorentz Force Relativistic Quantum Mechanics One important property of the electromagnetic ﬁeld is the Lorentz force acting on charged particles moving through the ﬁeld. We want to express this force through the tensor F µν . We want to √ demonstrate now that this equation is equivalent to the equation of motion (8. however.g. employing (10. Equation (10. e. To avoid √ confusion we will employ in the following for the energy of the particle the notation E = m/ 1 − v 2 [see (10.115) This equation follows.93).86) dE q = p·E .

122) This result gives rise to the deﬁnition of the function space representation ρ(Lµ ν ) of the Lorentz group (ρ(Lµ ν )ψ)(xµ ) = ψ((L−1 )µ ν xν ) . obey a diﬀerent transformation law. K3 ) are the generators of L(ϑ. In this section we will investigate the transformation properties of scalar functions ψ(xµ ). taken at the pairs of points (x µ = Lµ ν xν .121) by (L−1 )µ ν xν and obtains ψ (xµ ) = ψ((L−1 )µ ν xν ) . w) = exp ϑ · J + w · K .4 Function Space Representation of Lorentz Group In the following it will be required to decribe the transformation of wave functions under Lorentz transformations..7. In analogy to the situation for SO(3) we seek an expression for ρ(Lµ ν ) in terms of an exponential operator and transformation parameters ϑ. i. We like to express now ψ (x µ ) in terms of the old coordinates xµ . (10. w)) = ρ eϑ·J + w·K which we present in the form L(ϑ. Functions which represent 4-vectorsor other non-scalar entities.127) . (10. xµ ).48) of SO(3). However.1) is a generalization (rotation + boosts) of the group SO(3). For this purpose one replaces xµ in (10.126) ρ ewk Kk − 1 1 .51) for the natural representation of the Lorentz group.e. ψ ∈ C∞ (4). We will denote the intended representation by L(ϑ. w.e. the charge-current density in case of Sect.. i.124) In this expression J = (J1 .42) of SO(3). def (10.3 or the bi-spinor wave function of electron-positron pairs in Sect. e.123) This deﬁnition corresponds closely to the function space representation (5. and which can be constructed following the procedure adopted for the function space representation of SO(3). K2 . in as far as SO(3.125)]. J2 .4: Function Space Representation of Lorentz Group 301 10. Accordingly. The resulting expression should be a generalization of the function space representation (5. one can evaluate Jk as follows Jk = lim and Kk Kk = lim wk →0 ϑk →0 1 ϑ1 1 w1 ρ eϑk Jk −1 1 (10. We need to emphasize that (10.. (5. 10.125) def (10.121) which states that the function values ψ (x µ ) at each point x µ in the new frame are identical to the function values ψ(xµ ) in the old frame taken at the same space–time point xµ . 10. w) = ρ(Lµ ν (ϑ.121) covers solely the transformation behaviour of scalar functions. we seek an expression which corresponds to (10.48) and (10. J3 ) and K = (K1 . (10.g. w) which correspond to the generators Jk and Kk in (10. For such functions holds in the transformed frame ψ (Lµ ν xν ) = ψ(xµ ) (10.47). in the present case we exclude the factor ‘−i’ [cf.10.

49) as the generators of the natural representation. namely [J1 . and [J1 .126) for J1 we consider ﬁrst −1 Relativistic Quantum Mechanics K1 = x0 ∂1 + x1 ∂0 K2 = x0 ∂2 + x2 ∂0 K3 = x0 ∂3 + x3 ∂0 (10. sinhw1 x0 + coshw1 x1 .128). x3 ) 2 = ψ(xµ ) + w1 (x1 ∂0 + x0 ∂1 ) ψ(xµ ) + O(w1 ) (10. (10. x3 ∂1 ] = −x1 ∂2 + x2 ∂1 = J3 . which we like to demonstrate for J1 and K1 . K ] [ Jk . K2 ] = K3 : [ J1 . J3 = x2 ∂1 − x1 ∂2 . the expression for K1 in (10. [ Jk . (10. [K1 . J2 ] = J3 . One can determine similarly K1 starting from coshw1 sinhw1 sinhw1 coshw1 = 0 0 0 0 0 0 1 0 0 0 . i. The generators J .133) We demonstrate this for three cases. x1 . In order to evaluate (10. J ] [ Kk .134) . obviously.132) and. J2 = x1 ∂3 − x3 ∂1 . x1 ∂3 − x3 ∂1 ] = [x3 ∂2 . K ] = = k m Jm k m Jm = − k m Km . obviously. 1 This result. K obey the same Lie algebra (10.130) ew1 K1 −1 = e−w1 K1 (10.302 One obtains J1 = x3 ∂2 − x2 ∂3 . x2 . reproduces the expression for J1 in (10. x1 ∂3 ] − [x2 ∂3 . −sinϑ1 x2 + cosϑ1 x3 ) = ψ(xµ ) + ϑ1 (x3 ∂2 − x2 ∂3 ) ψ(xµ ) + O(ϑ2 ) .e. 0 1 (10. J2 ] = [x3 ∂2 − x2 ∂3 .129) which yields for small ϑ1 ρ eϑ1 J1 ψ(xµ ) = ψ(x0 .131) This yields for small w1 ρ ew1 K1 ψ(xµ ) = ψ(coshw1 x0 + sinhw1 x1 . K2 ] = −J3 . cosϑ1 x2 + sinϑ1 x3 .128) eϑ1 J1 = e−ϑ1 J1 1 0 = 0 0 0 0 0 1 0 0 0 cosϑ1 sinϑ1 0 −sinϑ1 cosϑ1 (10.126).

Eqs.85-5.139.4: Function Space Representation of Lorentz Group [ K1 .142) (10.10. x2 ∂0 ] − [x2 ∂3 .137). K2 ] = [x0 ∂1 + x1 ∂0 . K2 ] = [x3 ∂2 − x2 ∂3 . (10. x3 as follows x0 = R coshΩ . 0 ∂cothΩ ∂x x = ∂R x0 = − ∂x3 R (10.138) x3 x0 . x0 ∂2 + x2 ∂0 ] = [x0 ∂1 . 3 ∂R 3 ∂Ω ∂x ∂x R ∂R x ∂Ω = ∂Ω ∂tanhΩ 1 = cosh2 Ω 0 3 ∂tanhΩ ∂x x ∂Ω ∂cothΩ 1 = = − sinh2 Ω 3 . cothΩ = 3 . 10.135) (10..128).137) where K3 is given in (10. One-Dimensional Function Space Representation The exponential operator (10. ∂Ω .125) in the case of a one-dimensional transformation of the type L(w3 ) = exp w3 K3 . [ J1 .136) (10. In both cases the radial coordinate is the quantity conserved under the transformations.140) x0 x The transformation to hyperbolic coordinates closely resembles the transformation to radial coordinates for the generators of SO(3) in the function space representation [cf. The relationships (10.143) .140) allow one to ∂ ∂ express the derivatives ∂0 . x0 ∂2 ] = −x2 ∂1 + x1 ∂2 = − J3 . 303 (10. x0 ∂2 + x2 ∂0 ] = [x3 ∂2 . a relationship which can also be stated R = and + − (x0 )2 − (x3 )2 (x0 )2 − (x3 )2 if x0 ≥ 0 if x0 < 0 (10.e.87)].141) (10. We note tanhΩ = ∂R x0 = . Ω which are connected with x0 . ∂3 in terms of ∂R . In the following we consider solely the case x0 ≥ 0. can be simpliﬁed considerably. ∂x0 R and ∂Ω ∂x3 ∂Ω ∂x0 The chain rule yields then ∂0 ∂3 ∂R ∂ ∂Ω ∂ x0 ∂ 1 ∂ + = − sinh2 Ω 3 0 ∂R 0 ∂Ω ∂x ∂x R ∂R x ∂Ω ∂R ∂ ∂Ω ∂ x3 ∂ 1 ∂ = + = − + cosh2 Ω 0 . x2 + x2 + x2 1 2 3 in the case of SO(3) and (x0 )2 − (x3 )2 in case of transformation (10. x0 ∂2 ] = x3 ∂0 + x0 ∂3 = K3 . i. x2 ∂0 ] − [x1 ∂0 . For this purpose one expresses K3 in terms of hyperbolic coordinates R. (5.139) x3 = R sinhΩ (10.

which allows one to replace classical observables by quantum mechanical operators acting on wave functions.304 Inserting these results into the deﬁnition of K3 in (10. Further below we will provide a more systematic.121. − ) = i ∂ µ . (10. complex-valued function. namely the Klein–Gordon and the Dirac equations. In the following we will employ so-called natural units = c = 1. 10. In the position representation the correspondence principle states ˆ E =⇒ E = − ∂t i ˆ p =⇒ p = i which. In natural units the Klein–Gordon equation (10.145) ∂Ω 10.148) ∂ µ ∂µ + m2 (10. therefore.5 Klein–Gordon Equation In the following Sections we will provide a heuristic derivation of the two most widely used quantum mechanical descriptions in the relativistic regime.92) one obtains the wave equation − or 2 2 µ ∂ ∂µ ψ(xν ) = m2 ψ(xν ) ψ(xν ) = 0 .147) (10.151) is called the Klein-Gordon equation. in 4-vector notation reads pµ =⇒ pµ = i (∂t . (10.144) The action of the exponential operator (10. ∂Ω Relativistic Quantum Mechanics (10. representation theoretic treatment. Free Particle Case A quantum mechanical description of a relativistic free particle results from applying the correspondence principle. We will provide a ‘derivation’ of these two equations which stem from the historical development of relativistic quantum mechanics. ) = i ∂µ .128) yields K3 = x0 ∂3 + x3 ∂0 = ∂ . and (time)−1 all have the same dimension. extremely useful.122). The latter property implies that upon change of reference frame ψ(xµ ) transforms according to (10. mass. momentum. The historic route to these two equations. In these units the quantities energy. (length)−1 . but certainly is short and.146) Applying the correspondence principle to (10. however.149) where ψ(xµ ) is a scalar.137) on a function f (Ω) ∈ C∞ (1) is then that of a shift operator ∂ L(w3 ) f (Ω) = exp w3 f (Ω) = f (Ω + w3 ) . The partial diﬀerential equation (10. ˆ pµ =⇒ pµ = i(∂t . ˆ (10.150) .151) reads ∂µ ∂ µ + m2 ψ(xµ ) = 0 (10. is not very insightful.

t) ) 2m 1 = ( ψ ∗ (r. and that (as postulated) ψ(xµ ) is a scalar. t) + where ρS (r. (10. t) jKG (r. Current 4-Vector Associated with the Klein-Gordon Equation As is well-known the Schr¨dinger equation of a free particle o i∂t ψ(r. t)∂t ψ(r. In order to obtain the conservation law connected with the Klein–Gordon equation (10. To 1 derive this conservation law one rewrites the Schr¨dinger equation in the form (i∂t − 2m 2 )ψ = 0 o and considers 1 2 Im ψ ∗ i∂t − ψ = 0 (10.151) One can notice immeadiately that (10. t) ψ(r. t) + where ρKG (r. t) = 0 (10.152) which has the same form as the Klein–Gordon equation in the original frame. t) − ψ(r.161) . t) = ψ ∗ (r. t) (10. t) ψ(r. t) = · jKG (r.10.150) one considers Im ψ ∗ ∂µ ∂ µ + m2 ψ = 0 (10. t)∂t ψ ∗ (r.150) becomes ∂µ ∂ µ + m2 ψ (x µ ) = 0 (10.154). t) ψ(r. t) ) . t) ] 2mi (10.155) describes the positive deﬁnite probability to detect a particle at position r at time t and where jS (r.153) is associated with a conservation law for particle probability ∂t ρS (r.154) describes the current density connected with motion of the particle probability distribution. t) 1 [ ψ ∗ (r. t) ψ ∗ (r.158) which yields 2 2 ψ ∗ ∂t ψ − ψ∂t ψ ∗ − ψ ∗ 2ψ +ψ ψ∗ − 2ψ∗ = ψ) = 0 (10. t) = 0 i ( ψ ∗ (r.160) ∂t (ψ ∗ ∂t ψ which corresponds to − ψ∂t ψ∗) + · (ψ ψ∗ ∂t ρKG (r.150) is invariant under Lorentz transformations.157) 2m which is equivalent to (10. This follows from the fact that ∂µ ∂ µ and m2 are scalars. t) − ψ(r. 2mi (10. t) ψ ∗ (r.5: Klein–Gordon Equation or 2 ∂t − 2 305 + m2 ψ(xµ ) = 0 . t) = − 1 2m 2 ψ(r.159) (10. t) = (10. t) − ψ(r.156) · jS (r. Under Lorentz transformations the free particle Klein–Gordon equation (10.

150) is a continuum of positive energies E ≥ m.150) are actually determined by po and by the choice of sign ±.154). λ|xµ ) = N e−iλmt . λ) associated with the corresponding wave functions ψo (p. corresponding to λ = −. λ = ± (10. The proper interpretation of the two cases is that the Klein–Gordon equation describes particles as well as anti-particles.151) yields 2 2 Eo − p0 − m2 ψ(r. The density ρKG (p. t). i. ±) = ± 2 m2 + po (10. Generating a Solution Through Lorentz Transformation A particle at rest. p0 . m 2 2 E0 = m2 + po .164) The corresponding energy is Eo (po . We denote this by summarizing the solutions as follows ∂µ ∂ µ + m2 ψo (p. λ|xµ ) = Nλ. t) = 0 (10. When the Klein-Gordon equation had been initially suggested this lack of positive deﬁniteness worried physicists to a degree that the Klein–Gordon equation was rejected and the search for a Lorentz–invariant quantum mechanical wave equation continued.93] dispersion relationship connecting E0 . Solution of the Free Particle Klein–Gordon Equation Solutions of the free particle Klein–Gordon equation are ψ(xµ ) = N e−ipµ x µ = N ei(p0 ·r − Eo t) .162) Inserting this into the Klein–Gordon equation (10.161) and (10. λ) actually describes charge density rather than probability.163) which results in the expected [see (10. (10. and the density ρKG (p. λ|xµ ) according to (10. corresponding to λ = +. according to (??) is decribed by the r–independent wave function ψo (p = 0. (10. and of negative energies E ≤ −m. (10. is actually that of a charge density. The proper interpretation of ρKG (r. it had been realized later. λ = ± . for example pions.168) .. λ|xµ )ψo (p.p ei(p·r − λEo (p)t) Eo (p) = The spectrum of the Klein–Gordon equation (10. λ|xµ ) (10.165) This result together with (10. the Klein-Gordon equation is considered as a suitable equation to describe spin–0 particles. λ) = λ ψo (p. the anti-particles carry a charge opposite to that of the associated particles. with p = 0. λ|xµ ) = 0 2 m2 + po .167) m which is positive for λ = + and negative for λ = −.166) is Eo (p) ∗ ρKG (p. not of particle probability.166) ψo (p. o namely.306 Relativistic Quantum Mechanics This conservation law diﬀers in one important aspect from that of the Schr¨dinger equation (10.162) shows that the solutions of the free particle Klein-Gordon equation (10. Today.e. in that the expression for ρKG is not positive deﬁnite.

of (10.172) as the energy [c. t). (10.. In case of λ = −.175) as the momentum of the particle and one obtains L(w3 )ψo (p = 0.173) (10.e.f.166)] of the particle. This yields.6 Klein–Gordon Equation for Particles in an Electromagnetic Field We consider now the quantum mechanical wave equation for a spin–0 particle moving in an electromagnetic ﬁeld described by the 4-vector potential Aµ (xµ ) = (V (r.169) The addition theorem of hyperbolic functions cosh(Ω + w3 ) = coshΩ coshw3 + sinhΩ sinhw3 allows us to rewrite the exponent on the r. (10.s. λ = +|xµ ) = N ei(px 3 − Eo (p)x0 (10..171) One can interpret then for λ = +.169) −iλ ( m coshw3 ) ( R coshΩ ) − iλ ( m sinhw3 ) ( R sinhΩ ) . Aµ (xµ ) = (V (r.170) The coordinate transformation (10.10. λ = −|xµ ) = N ei(px 3 + Eo (p)x0 . Ω.174) which agrees with the expression given in (10.6: Klein–Gordon Equation with Electromagnetic Field 307 We want to demonstrate now that the wave functions for p = 0 in (??) can be obtained through appropriate Lorentz transformation of (10.168). t). t)) . For this purpose we consider the wave function for a particle moving with momentum velocity v in the direction of the x3 –axis. i. for negative energy solutions.e.h. for positive energy solutions.138) and the relationships (10.177) . t)) (10.138) to replace t = x0 by hyperbolic coordinates R. one has to interprete p = mv/ 1 − v 2 (10. (10. using (10. p = − mv/ 1 − v 2 as the momentum of the particle relative to the moving frame and √ m = 1 − v2 m2 = 1 − v2 m2 + m2 v 2 = 1 − v2 m2 + p2 = Eo (p) (10. for the wave function (10. (10. 2 2 1−v 1−v (10. L(w3 )ψo (p = 0. In case of λ = + one obtains ﬁnally L(w3 )ψo (p = 0.176) 10. λ|xµ ) = = exp w3 ∂ ∂Ω N e−iλmRcoshΩ 3) N e−iλmRcosh(Ω+w . i.168). −A(r. Such wave function should be generated by applying the Lorentz transformation in the function space representation p (10. A(r.61) yield for this expression −iλ √ m mv x0 − iλ √ x3 .145) choosing m = sinhw3 .166). in fact.

e. albeit in a form.180) + m2 ψ(r. according to the rules shown in Table 10. (10.150) −g µν (i∂µ )(i∂ν ) + m2 ψ(xµ ) = 0 .. m2 ) ψ(xµ ) = 0 .150). it’s rest energy.182) and seek an equation for Ψ(r. t))2 ψ(r. To obtain the appropriate wave equation we follow the derivation of the free particle Klein–Gordon equation above and apply again the correspondence principle to (10. i∂t and −i in (10. momenta for classical particles and the respective operators for quantum mechanical particles in the manner shown. t) 2 (10. t) = −i − q A(r.308 free classical particle classical particle in ﬁeld (V. in . According to the so-called minimum coupling principle the presence of the ﬁeld is accounted for by altering energy. in keeping withnthe non-relativistic limit. and focus on the remaining part of the wave function. is much larger than all other energy terms. we deﬁne ψ(r. We will also assume. (10. For this purpose we split-oﬀ a factor exp(−imt) which describes the oscillations of the wave function due to the rest energy. which couples a particle of charge q to an electromagnetic ﬁeld described through the potential Aµ (xν ). t) .. t) . that the mass m of the particle. i.178) According to the replacements in Table 10. A) free quantum particle quantum particle in ﬁeld (V. (10. (10. For this purpose one writes the Klein-Gordon equation (10. i. In terms of space-time derivatives this reads (i∂t − qV (r.1 this becomes g µν (i∂µ − qAµ )(i∂ν − Aν ) ψ(xµ ) = m2 ψ(xµ ) which can also be written g µν πµ πν − .e. −A). According to the principle of minimal coupling [see (10.147). A) or Aµ = (V.e..1: Coupling of a particle of charge q to an electromagnetic ﬁeld described by the 4-vector potential Aµ = (V. t). i.93). t) = e−imt Ψ(r.181) Non-Relativistic Limit of Free Particle Klein–Gordon Equation In order to consider further the interpretation of the positive and negative energy solutions of the Klein–Gordon equation one can consider the non-relativistic limit.1. See also Eq.69)] one replaces the quantum mechanical operators. A) Relativistic Quantum Mechanics energy momentum 4-vector E p pµ E − qV p − qA pµ − qAµ i∂t ˆ p = −i i∂µ i∂t − qV ˆ ˆ p − qA = π i∂µ − qAµ = πµ Table 10.179) (10.

The process of π − meson capture involves the so-called Auger eﬀect. | i∂t Ψ | << m . (10.183) The term on the l. π − mesons are generated through inelastic proton–proton scattering p + p −→ p + p + π − + π + ..186) then are slowed down. t) = − Ze and A(r. larger than |i∂t Ψ/Ψ| and alrger than qV .e. i. atoms in which one or more electrons are replaced by π − mesons.188) Inserting this into (10. i. This application demonstrates that the Klein–Gordon equation describes spin zero particles. t) 2m Ψ(r. (10. Pionic Atoms To apply the Klein–Gordon equation (10.s. ﬁltered out of the beam and ﬁnally fall as slow pions onto elements for which a pionic variant is to be studied. t) (10.e.181) can then be approximated as follows: (i∂t − qV )2 e−imt Ψ = = (i∂t − qV ) (me−imt Ψ + e−imt i∂t Ψ − qV e−imt Ψ) m2 e−imt Ψ + me−imt i∂t Ψ − qV e−imt Ψ +me−imt i∂t Ψ − e−imt ∂ 2 Ψ − qV e−imt i∂t Ψ −me−imt qV Ψ − e−imt i∂t qV Ψ + q 2 V 2 e−imt Ψ ≈ m2 e−imt Ψ − 2mqV e−imt Ψ − 2me−imt i∂t Ψ (10. (10.189) . t)]2 + qV (r.185) This is. Ψ |q V | << m .183). spin-0 mesons.. of (10. however. t) = ˆ [p − q A(r.187) We want to investigate in the following a description of a stationary state of a pionic atom involving a nucleus with charge +Ze and a π − meson.181) yields (we assume now that the Klein–Gordon equation describes a 2 particle with mass mπ and charge −e) for qV (r. To ‘manufacture’ pionic atoms. A stationary state of the Klein–Gordon equation is described by a wave function ψ(xµ ) = ϕ(r ) e−i t . the binding of a negative charge (typically an electron) while at the same time a lower shell electron is being emitted π − + atom −→ (atom − e− + π − ) + e− .2) of a non-relativistic spin-0 particle o moving in an electromagnetic ﬁeld.g. identical to the Schr¨dinger equation (10.181) to a physical system we consider pionic atoms. t) ≡ 0 r + Ze2 r 2 + 2 − m2 π ϕ(r ) = 0 .181) reads then i ∂t Ψ(r.184) where we neglected all terms which did not contain factors m.10. The Klein-Gordon equation (10. (10. The approximation is justiﬁed on the ground of the inequalities (10.6: Klein–Gordon Equation with Electromagnetic Field particular.h. 309 (10. e..

φ) = ( + 1) Y m (θ.199) ( + 1) − Z 2 e4 . i.196) In this expression the number n deﬁned through n = n− −1 (10.e. i. (10. (10. n − 1 .190) With this expression and after expanding ( + one obtains 2 ˆ d2 L2 − Z 2 e4 2 Ze2 − + + dr2 r2 r − m2 π r φ(r) = 0 .e.192) ˆ are spherical harmonics.310 Relativistic Quantum Mechanics Because of the radial symmetry of the Coulomb potential we express this equation in terms of spherical coordinates r. 2n2 = 0. . .192) leads then to the ordinary diﬀerential equation d2 − dr2 ( + 1) − Z 2 e4 2 Ze2 + + r2 r 2 − m2 π R (r) = 0 . φ. φ) (10.198) .191) ˆ The operator L2 in this equation suggests to choose a solution of the type ϕ(r ) = where the functions Y in (10. (10. .194) Bound state solutions can be obtained readily noticing that this equation is essentially identical to 2 that posed by the Coulomb problem (potential − Ze ) for the Schr¨dinger equation o r d2 − dr2 ( + 1) 2mπ Ze2 + + 2mπ E r2 r R (r) = 0 (10. (10. r r r r sin θ r sin θ Ze2 2 r ) (10.191) m (θ. 1. . The Laplacian is 2 = ˆ 1 2 1 1 1 2 L2 2 ∂r r + 2 2 ∂θ sinθ∂θ + 2 2 ∂φ = ∂r r − 2 .. 2.193) (10. φ) . the eigenfunctions of the operator L2 ˆ L2 Y m (θ. The similarity of (10.194) and (10. this quantity deﬁnitely must be an integer. (10. θ. φ) R (r) Y r m (θ.195) The latter problem leads to the well-known spectrum En = − mπ (Ze2 )2 . n = 1.195) can be made complete if one determines λ such that λ( ) (λ( ) + 1) = The suitable choice is λ( ) = − 1 + 2 ( + 1 2 ) − Z 2 e4 2 (10.197) counts the number of nodes of the wave function. . . ..

. 1. . . which corresponds to a bound = 0. 2mπ 2 (n + λ( ) + 1)2 (10. 1. . results in the (10. 10.197. (10.194) d2 λ( ) ( λ( ) + 1 ) 2 Ze2 − + + dr2 r2 r 2 311 − m2 π R (r) = 0 . .205) α √ 2 2 1+ (n − β + β −α) and one obtains the series of approximations 1 1+ ≈ ≈ ≈ 1+ ≈ 1+ (n − β + α √ β 2 −α)2 1 1+ (n − α 2β α + O(α2 ))2 1 1+ n2 − α n + O(α2 ) β α 1 α n2 + α2 βn3 + O(α3 ) − α2 8n4 1 α 2n2 + α2 2βn3 + O(α3 ) .204) reads 1 (10.6: Klein–Gordon Equation with Electromagnetic Field and one can write (10..202) Solving this for state) yields (choosing the root which renders = 1 + mπ Z 2 e4 ( n + λ( )+1 )2 ≤ mπ . .199) and deﬁning EKG (n.200) The bound state solutions of this equation should correspond to from (10. Introducing α = Z 2 e4 and 1 β = + 2 (10. 1. + (10. i. − + 1. . . . 2mπ 2 −→ λ( ) . . n = 0. .10. (10.203) spectrum .196) if one makes the replacement 2 values which can be obtained E −→ One obtains − m2 π . . .e. . = 0. . EKG = Using (10. .201) mπ Z 2 e4 m2 − m2 π π = − .204) In order to compare this result with the spectrum of the non-relativistic hydrogen-like atom we expand in terms of the ﬁne structure constant e2 to order O( 8 ). . 2. . . . n − 1 m = − . m) = mπ 1 + Z 2 e4 (n− − 1 2 + ( + 1 )2 − Z 2 e4 )2 2 n = 1. 2 e2 −→ e2 mπ .

i. m) ≈ m − mZ 2 e4 mZ 4 e8 − 2n2 2n3 1 + − 1− Relativistic Quantum Mechanics α α2 α2 α2 − + + + O(α3 ) . t) (10.208). one with a positive-deﬁnite density.206) 1 2 3 + O(Z 6 e12 ) .. 4n (10. 10.209) which. 2n2 2βn3 8n4 4n4 (10.. In order to describe electron spectra one must employ the Lorentz-invariant 1 wave equation for spin. a feature which is connected with the 2nd order time derivative in this equation. (10. A more satisfactory Lorentz–invariant wave equation. however. Application of the correspondence principle (10.209).2 particles. but not an operator iγ · as suggested by the principle of relativity (equivalence of space and time). Dirac succeeded. Finally. the wave function changes sign under reﬂection. in fact.. the Dirac equation introduced below. but in the present case rather the negative of the energy. .e. is identical to the two equations (10. are unsatisfactory since expansion of the square root operator involves all powers of the Laplace operator.208. The latter spectrum shows. in turn. arises because the Klein–Gordon equation. The latter term agrees with observations of pionic atoms. t) . but for a long time to no avail. 10. for example.204) EKG (n.208) These two equation can be combined i∂t + m2 − 2 i∂t − m2 − 2 Ψ(r. however. His solution .208.146) leads to the wave equation i∂t Ψ(r. = 1 states into groups of two and four degenerate states. would have only a ﬁrst order time derivative. the Klein–Gordon equation had been rejected since it did not yield a positive-deﬁnite probability density. a splitting of the six n = 2.312 ≈ From this results for (10. because of the equivalence of space and time coordinates in the Minkowski space such equation necessarily can only have then ﬁrst order derivatives with respect to spatial coordinates.207) Here the ﬁrst term represents the rest energy. Heuristic Derivation Starting from the Klein-Gordon Equation An obvious starting point for a Lorentz-invariant wave equation with only a ﬁrst order time derivative is E = ± m2 + p 2 . This derivative. and the third term gives the leading relativistic correction.e. 10. i. i. the second term the non-relativistic energy. Many attempts were made by theoretical physicists to ‘linearize’ the square root operator in (10. However.e. It must be pointed out here that does not denote energy. t) = ± m2 − 2 Ψ(r. through the correspondence principle.209). It should feature then a diﬀerential operator of the type D = iγ µ ∂µ . Equations (10.7 The Dirac Equation Historically. the π − meson is a pseudoscalar particle. it does not agree with observations of the hydrogen spectrum. is related to the equation E 2 = m2 + p 2 of the classical theory which involves a term E 2 . Also.

213) reads (γ µ )2 = 1 µ = 0 −1 µ = 1.211) into (10. We seek to identify the coeﬃcients γ µ . this would lead to the two wave equations (P − m)Ψ = 0 and (P + m)Ψ = 0 which have a ﬁrst order time derivative and.212) where we have changed ‘dummy’ summation indices. namely. 3 . . the 4-dimensionsional representation discovered by Dirac involved new physical properties. ψ2 (xµ ). Starting point is to boldly factorize. γ 3 can only be realized by d×d–matrices requiring that the wave function Ψ(xµ ) is actually a d–dimensional vector of functions ψ1 (xµ ). the operator of the Klein–Gordon equation ∂µ ∂ µ + m2 = −( P + m ) ( P − m ) (10. Accordingly. (10.211) Obviously. j = 1.213) We want to determine now the simplest algebraic realization of γ µ .e. beautiful mathematical theory and that. γ j . Initially.7: Dirac Equation 313 to the problem involved an ingenious step.215) . (10. the realization that the linearization can be carried out only if one assumes a 4-dimensional representation of the coeﬃcients γ µ . it was assumed that the 4-dimensional space introduced by Dirac could be linked to 4vectors. but did not commute the. one can impose the condition γ0 is hermitian . Inserting (10. Properties of the Dirac Matrices Let us now trace Dirac’s steps in achieving the linearization of the ‘square root operator’ in (10. 3 has imaginary eigenvalues ±i. It turns out that no 4-vector of real or complex coeﬃcients can satisfy these conditions. 3 are anti-hermitian . (10. Comparing the left-most and the right-most side of the equations above one can conclude the following property of γµ γ µ γ ν + γ ν γ µ = [ γ µ . therefore. The discovery by Dirac. these descriptions ultimately merging with the true theory of matter. strengthens the believe of many theoretical physicists today that the properties of physical matter ultimately derive from a. exploited ∂µ ∂ν = ∂ν ∂µ . j = 1. . so far.208). 2. according to (10. 2. γ 1 .210) yields −g µν ∂µ ∂ν − m2 = (iγ µ ∂µ + m)(iγ µ ∂µ − m) 1 = −γ µ γ ν ∂µ ∂ν − m2 = − ( γ µ γ ν ∂µ ∂ν + γ ν γ µ ∂ν ∂µ ) − m2 2 1 µ ν ν µ = − ( γ γ + γ γ ) ∂µ ∂ν − m2 2 (10.214) From this follows that γ 0 has real eigenvalues ±1 and γ j . the quantities γ 0 .209). ψd (xµ ).10. In fact. Instead. However. .. For µ = ν condition (10. achieved through a beautiful mathematical theory. this was not so. yet to be discovered. 2. one route to important discoveries in physics is the creation of new mathematical descriptions of nature. unspeciﬁed algebraic objects γ µ and γ ν . therefore.1 2 and anti-particles.66). are associated with a positive-deﬁnite particle density. γ 2 .210) where P = iγ µ ∂µ . γ ν ]+ = 2 g µν (10. spin. quantities with the transformation law (10. i.

218) of the Pauli matrices one can readily prove that condition (10. A proper choice is γ0 = 1 1 0 0 −1 1 . (10.220) where Ψ(xµ ) represents a 4-dimensional wave function. in fact. the γ µ are anti-commuting.217) Obviously. holds 2 σj = 1 . is unique.224) . rather than a scalar wave function. We will argue further below that the choice f γ µ . (10. The Dirac Equation Altogether we have shown that the Klein–Gordon equation can be factorized formally ( iγ µ ∂µ + m ) ( iγ µ ∂µ − m ) Ψ(xµ ) = 0 (10.216) i. namely the Pauli matrices σ 1 . (10.223) Inserting this into (10.e. σ 3 for which. to construct four matrices γ µ for the next possible dimension d = 4. is implied by (10. 2.314 For µ = ν (10. From this one can conclude for the determinants of γ µ det(γ µ γ ν ) = det(−γ ν γ µ ) = (−1)d det(γ ν γ µ ) = (−1)d det(γ µ γ ν ) .213) reads γ µ γ ν = −γ ν γ µ . γj = 0 −σ j σj 0 . ψ3 . For d = 2 there exist only three anti-commuting matrices.221) =0 (10. except for similarity trasnformations. This property can also be written (γ µ )† = γ 0 γ µ γ 0 .215).213) is satisﬁed. The Adjoint Dirac Equation The adjoint equation is Ψ† (xµ ) i(γ µ )† ∂µ + m ← ← (10.222) ∗ ∗ ∗ ∗ where we have deﬁned Ψ† = (ψ1 . One can readily show using the hermitian property of the Pauli matrices (γ 0 )† = γ 0 and (γ j )† = −γ j for j = 1. as long as det(γ µ ) = 0 the dimension d of the square matrices γ µ must be even so that (−1)d = 1. Relativistic Quantum Mechanics (10.222) and multiplication from the right by γ 0 yields the adjoint Dirac equation Ψ† (xµ ) γ 0 iγ µ ∂µ + m ← =0. however. ψ4 ) and where ∂µ denotes the diﬀerential operator ∂ µ operating to the left side.. σ 2 . ψ2 . σ j σ k = − σ k σ j for j = k .219) Using property (10.218) The Pauli matrices allow one. 1 (10. (10. in fact. rather than to the right side. From this equation one can conclude that also the following should hold ( iγ µ ∂µ − m ) Ψ(xµ ) = 0 which is the celebrated Dirac equation. 3 which.

233) The eigenstates and eigenvalues of H correspond to the stationary states and energies of the particles described by the Dirac equation. (10.230) Schr¨dinger Form of the Dirac Equation o Another form in which the Dirac equation is used often results from multiplying (10.225) . 2.Chiral Representation 315 The Dirac equation can be subject to any similarity transformation deﬁned through a non-singular ˜ 4 × 4–matrix S.219) is the socalled chiral representation deﬁned through ˜ Ψ(xµ ) = S Ψ(xµ ) . (10. t) = 0 (10. j = 1.227) A representation often adopted beside the one given by (10. [˜ µ .227) leaves the algebra of the Dirac matrices unaﬀected and commutation property (10.221) from the left by γ 0 ˆ ˆ i ∂t + i α · − β m Ψ(r. γj = ˜ 0 σj 1 S = √ 2 −σ j 0 1 1 1 1 1 −1 1 1 (10.227). ˆ ˆ ˆ Ho = α · p + β m ..228) (10. (10. γ ˜ Exercise 10.229) The similarity transformation (10. j = 1. 3 . αj = ˆ 0 σj σj 0 .e. Deﬁning a new representation of the wave function Ψ(xµ ) ˜ Ψ(xµ ) = S Ψ(xµ ) leads to the ‘new’ Dirac equation ˜ ( i˜ µ ∂µ − m ) Ψ(xµ ) = 0 γ where γ µ = S γ µ S −1 ˜ (10. i. 2.213).222. 3 .232) This form of the Dirac equation is called the Schr¨dinger form since it can be written in analogy o to the time-dependent Schr¨dinger equation o i∂t Ψ(xµ ) = Ho Ψ(xµ ) .7. γ ν ]+ = 2 g µν .226) (10. (10. 3 and ˆ β = 1 1 0 0 −1 1 .10. (10.1: Derive (refeq:Dirac-intro20a) from (10. 2. . 10.231) ˆ whereα has the three components αj .7: Dirac Equation Similarity Transformations of the Dirac Equation . and γ0 = ˜ 0 1 1 1 0 1 .213) still holds. j = 1.

Good. the matrix elements of C = A ⊗ B are Cjk.H.Mod.e. j1 < j2 < · · · < js s ≤ 4 (10. The associative algebra generated by d1 .235) and. any product 1 of the dj ’s can be brought to the form (10. page 187. d3 .237) which are the ordered products of the operators ±1 and d1 . in case of C4 . The Cliﬀord algebra C4 entails a subgroup G4 of elements ± dj1 dj2 · · · djs . d4 is called a Cliﬀord algebra C4 . Γ±14 = ±d1 d3 d4 .235). Γ±3 = ±d2 . d4 . For example. hence. Γ±9 = ±d2 d3 Γ±10 = ±d2 d4 . Obviously. form a Cliﬀord algebra C3 .213). The three Pauli matrices also obey the property (10. The representations of Cliﬀord algebras Cm are well established. d2 .235) can diﬀer only with respect to unitary similarity transformations. Rev. to Dirac matrices. The reader may . Γ±7 = ±d1 d3 . j = 1. 3 .316 Cliﬀord Algebra and Dirac Matrices The matrices deﬁned through dj = iγ j . i. satisfy the anti-commutation property dj dk + dk dj = 2 for j = k 0 for j = k Relativistic Quantum Mechanics d4 = γ 0 (10.236) where ‘⊗’ denotes the Kronecker product between matrices. One can conclude then that also any set of 4 × 4–matrices obeying (10. d4 = 1 0 0 −1 1 0 0 −1 0 1 1 0 0 i −i 0 (10.235) as can be readily veriﬁed from (10. (1955). a representation of the dj ’s is d1 = d3 = 0 1 1 0 0 i −i 0 1 0 0 1 1 0 0 1 . To complete the proof in this section the reader may consult Miller ‘Symmetry Groups and their Application’ Chapter 9. . Γ±5 = ±d4 Γ±6 = ±d1 d2 . d2 = . Γ±13 = ±d1 d2 d4 .6 and R. . Γ±4 = ±d3 . 27. Γ±11 = ±d3 d4 Γ±12 = ±d1 d2 d3 . Γ±15 = ±d2 d3 d4 Γ±16 = ±d1 d2 d3 d4 (10.238) These elements form a group since obviously any product of two Γr ’s can be expressed in terms of a third Γr . Since the Γj are hermitian the similarity transformations are actually given in terms of unitary transformations.e. The representations of this group are given by a set of 32 4 × 4–matrices which are equivalent with respect to similarity transformations.213).Phys.. i. m = Aj Bkm . 2. Γ±8 = ±d1 d4 . There are (including the diﬀerent signs) 32 elements in G4 which we deﬁne as follows Γ±1 = ±1 1 Γ±2 = ±d1 ..234) (10.237) by means of the property (10. This property extends then to 4 × 4–matrices which obey (10.

and derivatives according to (10.. namely.241) .6).6). the form of the Dirac equation is identical in equivalent frames of reference..4: Show that from (10.8 Lorentz Invariance of the Dirac Equation We want to show now that the Dirac equation is invariant under Lorentz transformations.218) of the Pauli matrices σ j . 10. Hence.. Except for a similarity transformation. in frames connected by Lorentz transformations. Inﬁnitesimal Bispinor State Transformation We want to show now that a suitable transformation of Ψ(xµ ) does.e. x µ = Lµ ν xν (10. Exercise 7.213. exist.10.e. in a moving frame holds iγ µ ∂µ − m Ψ (x ) = 0 .3: Demonstrate the anti-commutation relationships (10. x = Lµ ν xν . we do not know yet how to transform the 4-dimensional wave function Ψ and the Dirac matrices γ µ . a result one could have also obtained by applying the chain rule to (10. Exercise 7. therefore. However. i. We can. transform coordinates and derivatives of the Dirac equation.218) of the Dirac matrices γ µ . it must hold γ µ = γ µ . i..80). in fact.e. The transformation is assumed to be linear and of the form Ψ (x ) = S(Lµ ν ) Ψ(xµ ) µ µ (10. 10.239) Form invariance implies that the matrices γ µ should have the same properties as γ µ .236) with the Dirac representation (10. j = 1.76) yields ∂ρ = Lν ρ ∂ ν .8: Lorentz Invariance of the Dirac Equation 317 also want to establish the unitary transformation which connects the Dirac matrices in the form (10. Exercise 7. i.214) follows that γ 0 has real eigenvalues ±1 and can be represented by a hermitian matrix. Lorentz Transformation of the Bispinor State Actually.e. and γ j . (10. The latter transformations imply that coordinates transform according to (10.80) by Lµ ρ and using (10. i. Multiplication and summation of (10.240) . i. these properties determine the matrices γ µ uniquely. in a moving frame should hold iγ ∂µ − m Ψ (x ) = 0 . we will approach the proof of the Lorentz invariance of the Dirac equation by testing if there exists a transformation of the bispinor wave function Ψ and of the Dirac matrices γ µ which together with the transformations of coordinates and derivatives leaves the form of the Dirac equation invariant. 3 has imaginary eigenvalues ±i and can be represented by an anti-hermitian matrix.215). µ µ (10.2: Demonstrate the anti-commutation relationships (10. 2..219).e.

245) The construction of S(Lη ξ ) will proceed using the avenue of inﬁnitesimal transformations.12) in the form Lν µ gνσ Lσ ρ = gµρ = gρµ .246) (10. (10. i. of the equation. µ g ρν = µν and.244) from which.s.241) implies a similarity transformation Sγ µ S −1 . The proper starting point for a constructiuon of S(Lη ξ ) is actually (10. Ob1 viously. state that the Dirac equation (10. in the expression of the inﬁnitesimal transformation ρ Lµν = g µν + µν (10.e. It holds then S( µν i ) = 1 − σµν 1 4 µν µν = S(− νµ i ) = 1 + σµν 1 4 µν . the coeﬃcients of which depend on the matrix Lµ ν deﬁning the Lorentz transformation in such a way that S(Lµ ν ) = 1 for Lµ ν = δ µ ν holds. We had introduced in (10.243) in a form in which the Lorentz transformation in the form Lµν is on the r. Multiplication of (10.249) from which we can conclude σµν = −σµν νµ = −σνµ i.250) σµν = −σνµ .243) We want to determine now the 4 × 4–matrix S(Lη ξ ) which satisﬁes this condition. it must hold (10.h. (10. using gρµ γ µ = γρ . . The elements of g are.38) the inﬁnitesimal Lorentz transformations in the form Lµ ν = δ µ ν + µ ν where the inﬁnitesimal operator µ ν obeyed T = − g g. For this purpose we exploit (10.. hence.247) can be expanded S( µν i ) = 1 − σµν 1 4 µν (10. follows S(Lη ξ )γρ S −1 (Lη ξ ) = Lσ ρ γσ .242) The form invariance of the Dirac equation under this transformation implies then the condition S(Lη ξ )γ µ S −1 (Lη ξ ) Lν µ = γ ν . S( µν ) should not change its value if one replaces in its argument µν by − νµ . the transformation (10.221) upon Lorentz transformation yields iS(Lη ξ )γ µ S −1 (Lη ξ ) Lν µ ∂ν − m Ψ (x ) = 0 .e. One can ﬁnally conclude multiplying both sides by g ρµ S(Lη ξ )γ µ S −1 (Lη ξ ) = Lν µ γν . νµ (10.318 Relativistic Quantum Mechanics where S(Lµ ν ) is a non-singular 4 × 4–matrix.248) Here σµν denote 4 × 4–matrices operating in the 4-dimensional space of the wave functions Ψ. Multiplication of this property by g from the right yields ( g)T = − g. µ (10. The inﬁnitesimal transformation S(Lρ σ ) which corresponds to (10.243) from the left by Lσ ρ gσν yields S(Lη ξ )γ µ S −1 (Lη ξ ) gρµ = Lσ ρ gσν γσ . (10. hence. One can.. however. g is an anti-symmetric matrix.247) the inﬁnitesimal matrix µν is anti-symmetric.

In this representation the Dirac matrices γµ = (˜ 0 . 2. β [ σαβ . 10. Algebra of Generators of Bispinor Transformation We want to construct the bispinor Lorentz transformation by exponentiating the generators σµν . β and operate in the same space as the Dirac matrices.8: Lorentz Invariance of the Dirac Equation 319 One can readily show expanding SS −1 = 1 to ﬁrst order in µν that for the inverse inﬁnitesimal 1 transformation holds i (10. the expression on the r. σjk = [˜j . For this purpose we need to verify that the algebra of the generators involving addition and multiplication is closed.e.5: Show that the σαβ deﬁned through (10. For this purpose one needs to express formally the r.228. γβ ]− 2 (10.h.252) also as a sum over both indices of αβ . Furthermore.254). γ µ ]− = 2i ( δ µ β γα − δ µ α γβ ) .h. 3 .252) results in the condition for each α.h.248.252) Since six of the coeﬃcients αβ can be chosen independently. In fact. (10.254) The proper σαβ must be anti-symmetric in the indices α. of (10.213. j = 1. must be symmetric with respect to interchange of the indices α and β. i. this condition can actually be expressed through six independent conditions. For this purpose we express νµ γν = 1 2 αµ γα + 1 2 βµ γβ = = 1 2 αβ δ µ β γα + 1 2 αβ δ µ α γβ (10.s. namely. Comparing this with the l..251) into (10. (10. Exercise 7. γk ] = ˜ γ ˜ jk σ 0 0 σ ..10. (10. γj = ˜ 0 −σ j σj 0 .251) S −1 ( µν ) = 1 + σµν µν 1 4 Inserting (10.s.h.255) which can be demonstrated using the properties (10. (10.246) results then in a condition for the generators σµν − i ( σαβ γ µ − γ µ σαβ ) 4 αβ = νµ γν . ˜ ˜ σ only six generators need to be determined) σ0j = ˜ i [˜0 .257) .s. like the expression on the l.216) of the Dirac matrices.229). γj ] = γ ˜ 2 −iσ j 0 0 iσ j . (10.256) One can readily verify that the non-vanishing generators σµν are given by (note σµν = −˜νµ . of (10. 10.254) is σαβ = i [ γα . 10.253) 1 2 αβ ( δ µ β γα − δ µ α γβ ) .255) satisfy condition (10. the chiral representation introduced above in Eqs. a solution of condition (10.s. −γ ) are ˜ γ ˜ γ0 = ˜ 0 1 1 1 0 1 . For this purpose we inspect the properties of the generators in a particular representation.

261) We note that this operator is block-diagonal.e. ϑ) = e 2 (w − iϑ)·σ 0 − 1 (w + iϑ)·σ 0 e 2 1 (10.248) allows us to write the transformation S for any. and since the algebra of the Pauli matrices is closed. any representation of the generators..260) Inserting this into (10. the representation (10. Since such operator does not change its block-diagonal form upon exponentiation the bispinor transformation (10.259) yields the desired connection between the Lorentz transformation (10.241) bispinor wave functions from one frame of reference into another frame of reference. therefore. the algebra of these generators is closed under addition and multiplication. 2 0 −(w + iϑ) · σ 13 + σ23 ˜ 23 ) (10.262) This expression allows one to transform according to (10.44) 0 −w1 −w2 −w3 w 0 ϑ3 −ϑ2 = 1 w2 −ϑ3 0 ϑ1 w3 ϑ2 −ϑ1 0 µν (10.51) and S.255) yields a closed algebra.259) becomes in the chiral representation ˜ S(w.259) We had stated before that the transformation S is actually determined through the Lorentz transformation Lµ ν . be able to state S in terms of the same parameters w and ϑ as the Lorentz transformation in (10. µν in the exponential form i S = exp − σµν 4 µν . one can express the tensor µν through w and ϑ using µν = µ ρ g ρν and the expression (10. therefore. Finite Bispinor Transformation The closedness of the algebra of the generators σµν deﬁned through (10. i.258) again into block-diagonal operators. . We can ﬁnally note that the closedness of the algebra of the generators σµν is not aﬀected by similarity transformations and that. In fact. since both operations convert block-diagonal operators A 0 0 B (10.51). In this representation holds i ˜ − 4 σµν µν i = − (˜01 01 + σ02 02 + σ03 03 + σ12 12 + σ13 σ ˜ ˜ ˜ ˜ 2 1 (w − iϑ) · σ 0 = . (10. not necessarily inﬁnitesimal. In order to construct an explicit expression of S in terms of w and ϑ we employ again the chiral representation.320 Relativistic Quantum Mechanics Obviously. One should. in particular.

268) holds for the bispinor Lorentz transformation. the property applies then in any representation of S. of (10.221) from the left by Ψ† (xµ )γ 0 . The time-like component ρ of j µ 4 iγ µ ∂µ + iγ µ ∂µ ← Ψ(xµ ) = 0 . For this purpose we prove ﬁrst the relationship S −1 = γ 0 S † γ 0 .10.263) does hold. (10. and addition yields Ψ† (xµ )γ 0 The last result can be written ∂µ Ψ† (xν )γ 0 γ µ Ψ(xν ) = 0 .s.h. in fact.267) has the desired property of being positive deﬁnite.269) One can readily show that the same operator is obtained evaluating γ S (w.263) Starting point are the Dirac equation in the form (10.268) We will prove this property in the chiral representation..263) allows one to conclude that j µ must transform like a contravariant 4-vector as the notation implies. (10.264) (10.e.8: Lorentz Invariance of the Dirac Equation Current 4-Vector Associated with Dirac Equation 321 We like to derive now an expression for the current 4-vector j µ associated with the Dirac equation which satisﬁes the conservation law ∂µ j µ = 0 .266) ρ(xµ ) = Ψ† (xµ )Ψ(xµ ) = s=1 |ψs (xµ )|2 (10. The reason is that the r. The conservation law (10. We will now determine the relationship between the ﬂux j µ = Ψ (x )γ 0 γ µ Ψ (x ) † µ (10. −ϑ) = e− 2 (w − iϑ)·σ 0 1 (w + iϑ)·σ 0 e2 1 (10. (10. ϑ) = S(−w.224). (10. ϑ)˜ ˜ γ 0 ˜† 0 = 0 1 1 1 0 1 e 2 (w + iϑ)·σ 0 − 1 (w − iϑ)·σ 0 e 2 1 0 1 1 1 0 1 .271) .221) and the adjoint Dirac equation (10. For our proof we note ﬁrst ˜ ˜ S −1 (w. for j µ (xµ ) = (ρ. j µ must transform like a contravariant 4-vector. Multiplying (10.224) from the right by Ψ(xµ ). i.263) obviously is a scalar under Lorentz transformations and that the left hand side must then also transform like a scalar.270) We conclude that (10. Obviously.265) (10. the conservation law (10. This transformation behaviour can also be deduced from the transformation properties of the bispinor wave function Ψ(xµ ). (10. Since ∂µ transforms like a covariant 4-vector. j) = Ψ† (xµ )γ 0 γ µ Ψ(xµ ) .

268) µ j = Ψ† (xµ )S † γ 0 γ µ SΨ(xµ ) = Ψ† (xµ )γ 0 S −1 γ µ SΨ(xµ ) . as just mentioned. the identiﬁcation of observables and their quantum mechanical operators. As it turns out. commute with each other.246) S −1 (Lη ξ )γ µ S ( Lη ξ ) = (L−1 )νµ γν = (L−1 )ν γ ν = Lµ ν γν . Like in non-relativistic quantum mechanics the free particle wave function plays a central role. We will start from the Dirac equation in the Schr¨dinger form (10. The operators commute also with Ho in (10. not only as the most simple demonstration of the theory.233) o i∂t Ψ(xµ ) = Ho Ψ(xµ ) . as is required for the mentioned property. only two degrees of freedom of the bispinor four degrees of freedom are independent [c. Note that we have assumed in (10.271) that the Dirac matrices are independent of the frame of reference. (10.232. In case of non-relativistic quantum mechanics the free particle wave function has a single component ψ(r. 10. The independent degrees of freedom allow one to choose the states of the free Dirac particle as eigenstates of the ˆ ˆ 4-momentum operator pµ and of the helicity operator Γ ∼ σ · p/|p| introduced below. Combining this with (10.273) where we have employed (10. i. bispinor wave functions and we need to determine corresponding components of the wave function. The solutions provide also a complete. The additional degrees of freedom described by the four components of the bispinor wave function require.9 Solutions of the Free Particle Dirac Equation We want to determine now the wave functions of free particles described by the Dirac equation.274) 10.282. the Dirac particles are described by 4-dimensional.231.322 Relativistic Quantum Mechanics in a moving frame of reference and the ﬂux j µ in a frame at rest. Like for the free particle wave functions of the non-relativistic Schr¨dinger and the Klein–Gordon o equations one expects that the space–time dependence is governed by a factor exp[i(p · r − t)]. (10.272) results in the expected transformation behaviour µ j = Lµ ν j ν .. (10.272) With S −1 (Lη ξ ) = S((L−1 )η ξ ) one can restate (10. In the following we want to provide this characterization. additional characterizations. In relativistic quantum mechanics a Dirac particle can also be characterized through a momentum. One obtains using (10. (10. µ (10. 10. t) and is determined through the momentum p ∈ R3 . of which the wave functions are eigenfunctions as well. but also as providing a basis in which the wave functions of interacting particle systems can be expanded and characterized. however. As pointed out. a property which leads to the energy– momentum (dispersion) relationship of the Dirac particle. For this purpose we consider .76). orthonormal basis and allows one to quantize the Dirac ﬁeld Ψ(xµ ) much like the classical electromagnetic ﬁeld is quantized through creation and annihilation operators representing free electromagnetic waves of ﬁxed momentum and frequency.233). 10. These operaˆ tors.f. the wave function has four components which invite further characterization of the free particle state.275) The free particle wave function is an eigenfunction of Ho .e. speciﬁc for the Dirac free particle.283)].

one ﬁnds using (5.279) the well-known relativistic dispersion relationship 2 = m2 + p 2 (10.9: Solutions of the Free Particle Dirac Equation the following form of the free Dirac particle wave function Ψ(xµ ) = φ(xµ ) χ(xµ ) = φo χo ei(p·r − t) 323 (10.93.277) To solve this problem we write (10. +m (10. Inserting (10.280) σ·p (σ · p)2 φ0 = χ0 = +m ( + m) ( + m) p2 χ = χ0 . 2 − m2 0 (10. (10.276) into (10. from (10. These two relationships are consistent with each other. later to be identiﬁed with momentum and energy.276).282) (10. hence. two-dimensional spinor state.10.284) The relationships (10.234) and (10. In fact.280) which has a positive and a negative solution = ± E(p) . conclude 1. We want to show now that these degrees of freedom correspond to a spin-like property. (10.283) imply that the bispinor part of the wave function allows only two degrees of freedom to be chosen independently. reproduce the classical relativistic energy–momentum relationships (10.234) of Pauli matrices holds (σ · p)2 = p 2 1 One can.283) where is deﬁned in (10.279) According to the property (5.231. like the Klein–Gordon equation.276) where p and together represent four real constants. (10.281) Obviously.277) explicitly ( − m) 1 φo − σ · p χo 1 − σ · p φo + ( + m) 1 χo 1 = 0 = 0.278) provides us with information about the components of the bispinor wave function (10.232) leads to the 4-dimensional eigenvalue problem m σ·p σ · p −m φo χo = φo χo . 10. 10. and φo . 10. . E(p) = m2 + p 2 .94) Equation (10.278) Multiplication of the 1st equation by ( +m)1 and of the second equation by −σ · p and subtraction 1 of the results yields the 2-dimensional equation ( 2 − m2 ) 1 − (σ · p)2 φo = 0 . χo each represent a constant.281).282. namely φo and χo are related as follows φ0 χo = = σ·p χ0 −m σ·p φ0 . the socalled helicity of the particle. the Dirac equation. 1 (10.

First. ±) γ 0 Ψ(0..s.241). Hence.s. (10.s.287) will be justiﬁed now. i.259).. (10. For this purpose we consider ﬁrst the positive energy solution. u† u = |u1 |2 + |u2 |2 = 1 .286) Here N+ (p) is a constant which will be chosen to satisfy the normalization condition Ψ† (p. through χo given by χo = u1 u2 = u ∈ C2 . i.. (10.285) The corresponding free Dirac particle is then described through the wave function Ψ(p.h.h. ±). Ψ† (p. 0) γ 0 u 0 = |N+ (0)|2 . ±). ±)γ 0 Ψ(p. u† u = |u1 |2 + |u2 |2 = 1 .324 Relativistic Quantum Mechanics For our further characterization we will deal with the positive and negative energy solutions [cf. where S is given by (10. (10.281)] separately. Ψ† (0. (10. (10. (10. = +E(p) . i. ±) Ψ† (0. (10. = −E(p) . of (10. 10. is invariant under Lorentz transformations.e.268). and let Ψ(0. The connection between the solutions. the solution for = −E(p). ±) = S Ψ(0. ±) S † γ 0 S Ψ(0. ±) = = = Ψ† (0. (10.287.290) which diﬀers from the normalization condition (10. One can see this as follows: Let Ψ(p. +) γ 0 Ψ(0. hence. ±) denote the solution of a free particle moving with momentum p in the laboratory frame. we present the negative energy solution. +) = −1 . ±) Ψ† (0.289) Here N− (p) is a constant which will be chosen to satisfy the normalization condition Ψ† (p. +) = 1 . is Ψ(p. we demonstrate that the product Ψ† (p.h. ±) γ 0 S −1 γ 0 γ 0 S Ψ(0. using γ 0 as given in (10.288) corresponding to the wave function Ψ(p.287.287) in the minus sign on the r.290).e. according to (10. +) = |N+ (0)|2 (u† .f.219) and u† u = 1 [c.286) for p = 0 yields. The form of this condition and of (10. (10.291) Note that we have used that. S † = γ 0 S −1 γ 0 . ±) γ 0 Ψ(p.e.290) should diﬀer. +) γ 0 Ψ† (p. +|xµ ) = N+ (p) u σ·p E(p) + m u ei(p·r − t) . S −1 = γ 0 S † γ 0 and. 10. −|xµ ) = N− (p) −σ·p E(p) + m u u ei(p·r − t) . ±) . we present φo through the normalized vector φo = u1 u2 = u ∈ C2 .287) the form of which will be justiﬁed further below.285)]. +) γ 0 Ψ† (p. of (10. Employing (10. the l. For the positive energy solution. Similarly.292) . ±) denote the corresponding solution of a particle in its rest frame. according to (10. the solution for = +E(p). We want to demonstrate now that the signs on the r.

286). 2m (10. ( m + E(p) )2 − p 2 (10.299) ( m + E(p) )2 . −) γ 0 Ψ(0. Condition (10. σ·p u∗ E(p) + m T γo u σ·p E(p) + m u = 1 (10. u† ) γ 0 0 u = − |N− (0)|2 . 325 (10.9: Solutions of the Free Particle Dirac Equation The same calculation for the negative energy wave function as given in (10.287) written explicitly using (10.293) Obviously.295) Evaluating the l.301) yields the normalization coeﬃcient N+ (p) = m + E(p) .298) This result completes the expression for the wave function (10.10.s.f.298) becomes N+ (p) = m + E(p) . We consider ﬁrst the positive energy solution. this requires the choice of a negative side on the r. 2 E(p) (10.294) We want to determine now N± (p) for arbitrary p.297) from which follows N+ (p) = Noting ( m + E(p) )2 − p 2 = m2 − p 2 + 2mE(p) + E 2 (p) = 2(m + E(p)) m the normalization coeﬃcient (10.286) is 2 N+ (p) (u∗ )T . Exercise 7.h. −) = |N− (0)|2 (0. using γ 0 as given in (10. (10.s.300) (10. (5.285) results in 2 N+ (p) 1 − p2 (E(p) + m)2 = 1 (10.6: Show that the normalization condition 2 N+ (p) (u∗ )T . σ·p u∗ E(p) + m T u σ·p E(p) + m u = 1 (10.219) yields 2 N+ (p) u† u − u† (σ · p)2 u (E(p) + m)2 = 1.290) to assign a positive value to |N− (0)|2 .296) Replacing (σ · p)2 by p 2 [c.h.302) . We can also conclude from our derivation N± (0) = 1 . (10. of (10.289) yields Ψ† (0.234)] and using the normalization of u in (10.

i∂µ Ψ(p.289).289) is T −σ·p −σ · p 2 E(p) + m u N− (p) u∗ ..296) for the normalization constant N+ (p) of the positive energy solution.286. thereby.233) and ˆ ˆ ˆ from the two identities 1 1 0 0 −1 1 σ 0 0 σ ·p − ˆ σ 0 0 σ ·p ˆ 1 1 0 0 −1 1 = 0 (10.309) . The commutation property [σ · p.305) and. λ|xµ ) = λ E(p) Ψ(p.300) have been constructed to satisfy the free particle Dirac equation (10. pj ] = 0 . yields 2 N− (p) u† (σ · p)2 u − u† u (E(p) + m)2 = −1 .289.e.286) into (10. Rather than considering the ˆ observable (10. conclude N− (p) = m + E(p) 2m (10. λ|xµ ) (10.307) where pµ = ( .308) 2 |p| ˆ Note that p represents here an operator. (10.307) we investigate ﬁrst the observable due to the simpler operator σ · p. (u∗ )T γ o = −1 (10. The corresponding operator which measures this observable is 1 p ˆ Λ = σ· . The wave functions (10. j = 1. Helicity The free Dirac particle wave functions (10. 3 is fairly obvious. 2. λ|xµ ) = pµ Ψ(p. λ|xµ ) . Inserting (10.326 Relativistic Quantum Mechanics We consider now the negative energy solution. the wave functions constructed represent eigenstates of Ho .286. This degree of freedom 1 describes a spin– 2 attribute. not a constant vector. hence. (10. This can be veriﬁed expressing the space–time factor of Ψ(p. 10. −p). deﬁned as the component of the particle spin along the direction of motion. We can. λ|xµ ) in 4-vector notation. The property [σ · p. This attribute is the so-called helicity.e.h.275) yields Ho Ψ(p. The wave functions are also eigenstates of the momentum operator i∂µ . i. the two components of u indicate another degree of freedom which needs to be deﬁned.304) This condition is.289) are not completely speciﬁed.. Condition (10. We ˆ want to show that this operator commutes with Ho and p to ascertain that the free particle wave ˆ function can be simultaneously an eigenvector of all three operators.275).290) written explicitly using (10. Ho ] = 0 follows from (10. i..303) E(p) + m u Evaluating the l. 10. identical to the condition (10. 10.s. however. exp[i(p · r − t)] = exp(ipµ xµ ).e. have completed the determination for the wave function (10. (10.306) i.

t) = Np e 1 0 0 p m + Ep i(px3 + E t) 1 p e Ψ(pˆ3 .313) 2 0 1 where Ep and Np are deﬁned in (10. t) = Np e 2 1 p m + Ep 0 0 i(px3 − E t) 1 p e Ψ(pˆ3 . The wave functions which are eigenfunctions of the helicity operator are in this case 1 −p m + Ep i(px3 + E t) 0 1 p e Ψ(pˆ3 . in principle.308) above. 0)T and (0. hence. 2m (10. According to the deﬁnition (5. = +E(p). ˆ The condition that the wave functions (10.. . 1 p = (0. +. 1)T are eigenstates of 2 σ 3 with eigenvalues ± 1 .312). i. + 1 |r. are 1 i(px3 − E t) 0 p e Ψ(pˆ3 .e. − 1 |r. −. t) = Np e (10. 0.. = −E(p)..312) We assume now particles with negative energy. i.10. We consider ﬁrst the simplest case that particles move along the x3 –direction. States which are simultaneously eigenvectors of these three operators are also simulteneously eigenvectors of the three operators p.224) 1 of σ3 the two u vectors (1. + 2 |r. In this case Λ = 2 σ 3 . Np = m + Ep . Ho and σ·p commute with each other and. Therefore. +. Ho and Λ deﬁned in (10.311) 2 −p 0 m + Ep 1 where e3 denotes the unit vector in the x3 -direction and where ˆ Ep = m2 + p2 . Since helicity is deﬁned relative to the direction of motion of a particle the characterization of u as an eigenvector of the helicity operator.9: Solutions of the Free Particle Dirac Equation and 0 σ σ 0 = ·p ˆ σ 0 0 σ ·p − ˆ − σ 0 0 σ ·p ˆ 0 σ σ 0 ·p ˆ = 0. 2 the wave functions which are eigenstates of the helicity operator. p3 ).e. − 1 |r. can ˆ ˆ be simultaneously diagonal. 327 0 (σ · p)2 ˆ 2 (σ · p) ˆ 0 0 (σ · p)2 ˆ 2 (σ · p) ˆ 0 (10.310) We have shown altogether that the operators p. We assume ﬁrst particles with positive energy. i.286) are eigenfunctions of Λ as well will specify now the vectors u.e. t) = Np e (10. −. is independent of the direction of motion of the particle.

+ 1 ) 2 p 1 m + Ep u(p. 0)T and (0. p) e |p| (10. t) = Np ei(p·r + Ep t) (10. 10. + 2 ) u(p. γ and are 1 0 1 ˜ Ψ(p = 0. 10.320) where Ep and N are again given by (10. Generating Solutions Through Lorentz Transformation The solutions (10. The corresponding wave functions are determined through ˜ i˜ 0 ∂t − m Ψ(t) = 0 . 1 |t) = √2 e−imt .312) can be obtained also by means of the Lorentz transformation (10. t) 2 = Np u(p. + ) 2 1 u(p. t) = N√ ei(p·r + Ep t) (10. − 1 ) 2 −p 1 m + Ep u(p. γ . These eigenstates are obtained through a rotational transformation (5.317) Ψ(p.319) 1 Ψ(p.314) (10. 1)T have to be replaced by eigenstates u± (p) of the spin operator along the direction of p.123). 5.315) e3 × p ˆ ∠(ˆ3 . + 2 ) p 1 m + Ep u(p. +. − 2 |r. [One can also express the rotation through Euler angles α.318) 1 Ψ(p. − ) 2 where ϑ(p) = = = exp exp i − ϑ(p) · σ 2 i − ϑ(p) · σ 2 1 0 0 1 . +.312). − 2 ) u(p. + 2 ) ei(p·r − Ep t) (10. +. −. −. t) 2 = Np 1 u(p.313) except that the states (1.321) .226. 2 1 0 (10.316) describes a rotation which aligns the x3 –axis with the direction of p.223) as follows 1 u(p.e.262) for the bispinor wave function and the transformation (10. (10..222. + 1 |r.229). in which case the transformation is given by (5. − 1 |r.] The corresponding free particle wave functions are then Ψ(p.311.311.220). denoted by ˜. for an r–independent wave function. − 1 ) 2 ei(p·r − Ep t) (10. a wave function which represents free particles at rest. 10.328 Relativistic Quantum Mechanics To obtain free particle wave functions for arbitrary directions of p one can employ the wave functions (10. − 2 ) −p 1 m + Ep u(p. i. For this purpose one starts from the solutions of the Dirac equation in the chiral representation (10. β. + 2 |r.

313) in the p → 0 limit.311) e 1 w 2 3 ˜ Ψ(p(w3 )ˆ3 . −1 e 0 0 1 +imt 1 ˜ Ψ(p = 0.325) One should note that φo .323) Transformation (10. This correspondence justiﬁes the characterization ±.176) imt → i ( p3 x3 and for the bispinor part according to (10. χo ∈ 1 0 .262) for a boost in the x3 –direction.322).311. 10. φo . yields for the exponential space–time dependence according to (10. 10. − 1 |t) = √2 . −.228) yields the corresponding solutions (10. i. 0.323) should yield the solutions for non-vanishing momentum p in the x3 –direction. for w = (0. χo are eigenstates of σ 3 with eigenvalues ±1. +. − 1 |r.262) φo χo → e 2 w3 σ 0 1 3 Et ) (10. t) e 2 1 = √ 2 e− 2 w3 1 1 1 0 1 0 0 1 0 1 i(px3 − E t) p e ˜ Ψ(p(w3 )ˆ3 . 2 0 1 1 0 +imt 1 1 ˜ Ψ(p = 0.324) 0 e − 1 w3 σ 3 2 φo χo = e 2 w3 σ φo 1 3 e− 2 w3 σ χo 1 3 . 2 |t) = √2 . + 1 |r. − 1 |t) = √2 e−imt .e. (10.324. For the resulting wave functions in the chiral representation one can use then a notation corresponding to that adopted in (10. +. 0 1 (10. +.10. −.. 10. 2 The solutions (10.322) The reader can readily verify that transformation of these solutions to the Dirac representationsas deﬁned in (10.174. w3 ).9: Solutions of the Free Particle Dirac Equation 0 1 1 ˜ Ψ(p = 0. t) e 2 e− 2 w3 1 = √ 1 2 e 2 w3 i(px3 − E t) p e . ± 1 of the wave functions stated in (10. Applying (10.325) to (10.322) can be written in spinor form 1 √ 2 φo χo e imt . 2 0 e −1 329 (10.

t) e 2 = i(px3 + E t) p e 1 Ψ(p(w3 )ˆ3 . −.326) where according to (10. t) e 1 = √ 2 −e− 2 w3 − 1 w3 2 1 i(px3 + E t) p e ˜ Ψ(p(w3 )ˆ3 . 2 (10.330 e 1 w 2 3 Relativistic Quantum Mechanics 1 0 1 0 0 1 0 1 1 ˜ Ψ(p(w3 )ˆ3 .327) Employing the hyperbolic function properties cosh x = 2 coshx + 1 . −.328) the relationship (10. and the expression (10. 2 sinh x = 2 coshx − 1 .311) for Ep one obtains 1 = √ 2 1 1 √ 2 + 1 = 1 − v3 2 m2 + = 2 m2 v3 2 1 − v3 cosh w3 2 1+ 2 v3 2 + 1 1 − v3 +m = 2m Ep + m 2m (10.228) yields w3 cosh 2 = sinh w3 2 1 Ψ(p(w3 )ˆ3 . t) e = i(px3 + E t) p e (10. + 2 |r. − 2 |r.61) between the parameter w3 and boost velocity v3 . t) e = cosh w3 2 − sinh w3 2 sinh w3 2 cosh w3 2 − sinh w3 2 cosh w3 2 i(px3 − E t) p e Ψ(p(w3 )ˆ3 .61) p(w3 ) = m sinhw3 . +. −. + 1 |r. Transformation to the Dirac representation by means of (10. + 2 |r. −. − 2 |r.329) . t) e 2 e 1 = √ 1 2 − e 2 w3 i(px3 + E t) p e (10. − 1 |r. +. t) e 1 0 1 0 0 1 0 1 1 0 1 0 0 1 0 1 i(px3 − E t) p e 1 Ψ(p(w3 )ˆ3 .

ρ(Lη ξ ) has been deﬁned in (10. (10. In the new derivation we consider the particle described by the wave function transformed.331) where S(Lη ξ ) denotes again the transformation acting on the bispinor character of the wave function Ψ(xµ ) and where ρ(Lη ξ ) denotes the transformation acting on the space-time character of the wave function Ψ(xµ ). thereby. The reason why we provide another derivation of (10.313) for −p. reproduces the positive energy wave functions (10. λ.332) Here we have made use of the fact that S(Lη ξ ) commutes with ∂µ and ρ(Lη ξ ) commutes with γ µ . we will derive this equation only for inﬁnitesimal transformations. This transformation.329.168–10.330) into (10.221) using (10.243) considered solely the limit of inﬁnitesimal transformations anyway. that if Ψ(xµ ) is a solution of the Dirac equation that Ψ (xµ ) as given in (10. the proper transformation S(Lη ξ ). refered to as the active transformation. is suﬃcient since (1) it must hold then for any ﬁnite transformation. expresses the system in the old coordinates. Λ|xµ ) from the solutions describing particles at rest Ψ(p = 0. Λ|t). which however. but not the observer. Making this expectation a postulate allows one to derive the condition (10.10. 10.311) as well as the negative energy solutions (10. To show this we rewrite the Dirac equation (10.331) is a solution as well. Invariance of Dirac Equation Revisited At this point we like to provide a variation of the derivation of (10. Such description will be essential for the formal description of Lorentz invariant wave equations for arbitray spin further below. = γν .243) is to familiarize ourselves with a formulation of Lorentz transformations of the bispinor wave ﬁnction Ψ(xµ ) which treats the spinor and the space-time part of the wave function on the same footing. We expect.243).331) i S(Lη ξ )γ µ S −1 (Lη ξ ) ρ(Lη ξ )∂µ ρ−1 (Lη ξ ) − m Ψ (xµ ) = 0 (10. when we generated the solutions Ψ(p. Actually.330) Inserting expressions (10. indeed. of course.9: Invariance of Dirac Equation Revisited and similarly sinh w3 = 2 Ep − m = 2m p 2m (Ep + m) 331 (10.243) and.333) . the essential property stating the Lorentz–invariance of the Dirac equation.123) above and characterized there. and since (2) the calculations following (10. The transformation is Ψ (xµ ) = S(Lη ξ ) ρ(Lη ξ ) Ψ(xµ ) (10. The fact that any such Ψ (xµ ) is a solution of the Dirac equation allows us to conclude S(Lη ξ )γ µ S −1 (Lη ξ ) ρ(Lη ξ )∂µ ρ−1 (Lη ξ ) = γ ν ∂ν which is satisﬁed in case that the following conditions are met ρ(Lη ξ )∂µ ρ−1 (Lη ξ ) S(Lη ξ )γ µ S −1 (Lη ξ ) Lν µ = Lν µ ∂ν . The change of sign for the latter solutions had to be expected as it was already noted for the negative energy solutions of the Klein–Gordon equation (10. λ. in general. Such transformation had been applied by us.327).176).334) (10.

∂µ ] = (K )ν µ ∂ν .341) [K3 . The second condition is identical to (10. As mentioned already we will show condition (10.337) [J2 .335) shows that we need to demonstrate [J . 0.332 Relativistic Quantum Mechanics We will demonstrate now that the ﬁrst condition is satisﬁed by ρ(Lη ξ ).334) for inﬁnitesimal Lorentz transformations Lη ξ . 0).e. .336) We will proceed with this task considering all six cases: [J1 .128) to express ρ(Lη ξ ) in its inﬁnitesimal form and evaluate the expression 1 + ϑ·J + w·K 1 ∂µ 1 − ϑ·J − w·K 1 (10. of course.. We will proceed by employing the generators (10. 0). 0 0 ∂3 −∂2 0 −∂3 0 ∂1 0 ∂2 −∂1 0 −∂1 −∂0 0 0 −∂2 0 −∂0 0 −∂3 0 0 −∂0 µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ µ = = = = = = = = = = = = = = = = = = = = = = = = 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 (10. w = (0. ∂µ ] = (10. ∂µ ] = (J )ν µ ∂ν . w = (0.262). ∂µ ] = [x0 ∂3 + x3 ∂0 . ∂µ ] = (10.342) One can readily convince oneself that these results are consistent with (10. . 1. ∂µ ] = (10. 0. ∂µ ] = (10.331) is again a solution of the Dirac equation. w = (0.336). [K . . ∂µ ] = [x0 ∂2 + x2 ∂0 . 0. ∂µ ] = [x3 ∂2 − x2 ∂3 .335) = ∂µ + M ν µ ∂ν + O( 2 ) The result will show that the matrix M ν µ is identical to the generators of Lν µ for the six choices ϑ = (1. We have demonstrated. 0).340) [K2 . 0.243) and. . ∂µ ] = [x1 ∂3 − x3 ∂1 . ϑ = (0. 1) . that any solution Ψ(xµ ) transformed according to (10. ϑ = (0. ∂µ ] = (10. 0). it is met by S(Lη ξ ) as given in the chiral representation by (10. therefore. ∂µ ] = (10. 0. 0). i. ∂µ ] = [x2 ∂1 − x1 ∂2 . Inspection of (10. the Dirac equation is invariant under active Lorentz transformations.338) [J3 . . ∂µ ] = [x0 ∂1 + x1 ∂0 .339) [K1 .

The Dirac equation in the Schr¨dinger form reads then o i∂t Ψ(xµ ) = ˆ ˆ where α and β are deﬁned in (10. This solution exhibits a timedependence exp[−i(m + )t] where for holds in the non-relativistic limit | | << m. accordingly.1 in Sect 10.347) (10. 10.221) reads then [ iγ µ (∂µ + iqAµ ) − m ] Ψ(xν ) = 0 (10. (10. σ3 )T one obtains then i∂t φ i∂t χ = = ˆ σ · π χ + qV φ + mφ ˆ σ · π φ + qV χ − mχ .351) X (x ) µ . we replace in the Dirac equation the momentum operator pµ = i∂µ by i∂µ − qAµ where q is the charge of the respective particles ˆ (see Table 10. we deﬁne φ(xµ ) = e−imt Φ(xµ ) χ(x ) = e µ −imt (10. σ2 . Following the respective procedure developed for the Klein-Gordon equation in Sect. (10.g. Non-Relativistic Limit We want to consider now the Dirac equation (10.233) by replacing i∂t by (see Table 10.345) in the so-called non-relativistic limit in which all energies are much smaller than m.10.350) (10. Equivalently. we replace the operator ∂µ by ∂µ + iqAµ . The Dirac equation (10.343) One may also include the electromagnetic ﬁeld in the Dirac equation given in the Schr¨dinger form o ˆ (10.1) i∂t − qV and p by ˆ ˆ π = p − qA ..6 above). Accordingly.6 we assume that the ﬁeld is described through the 4-vector potential Aµ and.349) We want to focus on the stationary positive energy solution.346) ˆ ˆ ˆ α · π + qV + β m Ψ(xµ ) (10.232).345) holds |qV | << m . For this purpose we choose the decomposition Ψ(xµ ) = φ(xµ ) χ(xµ ) .344) Using the notation σ = (σ1 . for the scalar ﬁeld V in (10.348) (10. e.10: Dirac Particles in Electromagnetic Field 333 10.345) (10.10 Dirac Particles in Electromagnetic Field We like to provide now a description for particles governed by the Dirac equation which includes the coupling to an electromagnetic ﬁeld in the minimum coupling description.

accordingly. 5. ˆ ˆ For the components of π × π holds ˆ ˆ π×π = jk (10. derived in Sect.353) (10.334 and assume that for the time-derivative of Φ and X holds ∂t Φ << m . Ak ] .353) by X ≈ to obtain a closed equation for Φ ˆ σ·π 2m 2 (10.356) i∂t Φ ≈ Φ + qV Φ .358) ( πj πk − πk πj ) = jk [πj .356). and. ∂k ] + q [Aj .352) allow one to approximate (10.348. (10. identiﬁes X as the small component of the bi-spinor wave function which. 10. ∂k ] + q [ ∂j .355) ˆ σ·π Φ 2m (10.357) Equation (10.359) We want to evaluate the latter commutator. due to the m−1 factor. Ak ] i i i i =0 =0 = q q [Aj .346.354) The properties (10.349) yields i∂t Φ i∂t X = = ∂t X X Relativistic Quantum Mechanics << m . henceforth.7. 10.361) . one can replace X in (10. ˆ Equation (10.357) for Φ can be reformulated by expansion of (σ · π)2 . ∂k ] + [∂j .230). 10.352) ˆ σ · π X + qV Φ ˆ σ · π Φ + qV X − 2mX . (10. i i (10. Ak ] ) f = ( ∂j Ak − Ak ∂j + Aj ∂k − ∂k Aj ) f . (10. π ] .354) ˆ 0 ≈ σ · π Φ − 2m X . which in the present case states ˆ ˆ2 ˆ ˆ (σ · π)2 = π + i σ · (π × π) . Φ Using (10. does not need to be considered anymore. One obtains [πj .350. Ak ] + q 2 [Aj .351) in (10. ∂k ] + [∂j . (10. πk ] = = 1 1 [ ∂j + qAj .360) For an arbitrary function f (r) holds ( [Aj . (10. For this purpose we employ the identity (5. ∂k + qAk ] i i 1 1 1 1 [ ∂j .

0.223) shows that the propagator in (10. 10.10. The respective bispinor wave function Ψ(xµ ) ∈ C4 is described as the stationary solution of the Dirac equation (10.364) which is referred to as the Pauli equation. t) 2m 2m Φ(r. In other words. Equations (10. (10. 5.363) allow us to write (10. The formal solution of this equation is Φ(t) = e−iqtB·σ Φ(0) . one obtains ( [Aj ..10: Dirac Particles in Electromagnetic Field Using ∂j Ak f = ((∂j Ak )) f + Ak ∂j f ∂k Aj f = ((∂k Aj )) f + Aj ∂k f 335 where ((∂j · · ·)) denotes conﬁnement of the diﬀerential operator to within the brackets ((· · ·)). t) (10.343) for the vector potential Aµ = (− Ze2 .366) can be interpreted as a rotation around the ﬁeld B by an angle qtB. does not leave the theory again when one takes the non-relativistic limit.362) or. Ak ] ) f = [ ((∂j Ak )) − ((∂k Aj )) ] f (10. using (10.366) (10.365) Comparision of this expression with (5. r (10. [πj .222. i. 0) .363) where we employed B(r.e.359. ∂k ] + [∂j . 10. the spin. πk ] = q q (( ∂j Ak − ∂k Aj )) = − i i ×A jk q = − B i jk (10. t) ≈ ˆ [p − q A(r.1 with the magnetic ﬁeld.2) governing a one-dimensional wave function ψ ∈ C. For 2 this purpose we disregard the spatial degrees of freedom and assume the Schr¨dinger equation o i∂t Φ(t) = q σ · B Φ(t) . t) [see (8. Dirac Particle in Coulomb Field .358. 10. −A).364) governing a two-dimensional wave function Φ ∈ C2 with the corresponding non-relativistic Schr¨dinger equation (10. o reveals a stunning feature: the Pauli equation does justice to its two-dimensional character.1 which emerged in the Lorentz-invariant theory as an algebraic 2 necessity. 0. but rather remains as a steady “guest” of non-relativistic physics with the proper interaction term.360) and Aµ = (V. Let us consider brieﬂy the consequences of the interaction of a spin.2 around the magnetic ﬁeld. t) = × A(r. t) + q V (r.344.357) in the ﬁnal form i∂t Φ(r.367) .1 particle with a magnetic 2 ﬁeld B. t)]2 q − σ · B(r.Spectrum We want to describe now the spectrum of a relativistic electron (q = −e) in the Coulomb ﬁeld of a nucleus with charge Ze. Comparision of (10. the interaction qσ · B induces a 1 precession of the spin. while agreeing in all other respects with the non-relativistic Schr¨dinger equation (10.6)].2) it introduces the o extra term qσ · B Φ which describes the well-known interaction of a spin.

(˜ 0 )2 = 1 and (˜ j )2 = γ 1 γ −1 j = 1. (5. For this purpose we ‘square’ the Dirac equation.375) Following the algebra that connected Eqs. according to (10.368) is also a solution of (10. ˜ However. we solve ˜ [ i˜ µ (∂µ + iqAµ ) − m ] Ψ(xµ ) = 0 γ (10. 1. respectively. 3. 3 µ=0 ˆ2 (˜ µ )2 πµ = π0 − π . µ = ν and altering ‘dummy’ summation ˜ ˜ ˜ ˜ . Employing πµ as deﬁned ˜ in Table 10.372) (10.1 one can write (10. but the converse is not necessarily true.374) The ﬁrst term on the r.6.368) ˜ ( γ µ πµ − m ) Ψ(xµ ) = 0 .369) For our solution we will adopt presently a strategy which follows closely that for the spectrum of pionic atoms in Sect. (5.228) and in (10. noting from (10. once a solution Ψ(xµ ) of (10.369).369). but diﬀers from it in an essential way. 10. ˜ (10.ν=1 µ =ν γ µ γ ν πµ πν .232) in Sect. This follows from [ i˜ µ (∂µ + iqAµ ) + m ] [ i˜ µ (∂µ + iqAµ ) − m ] γ γ = [ i˜ µ (∂µ + iqAµ ) − m ] [ i˜ µ (∂µ + iqAµ ) + m ] γ γ according to which follows from (10.7 one can write the second term in (10.371).370) resembles closely the Klein-Gordon equation (10.180).230) γ µ γ ν = − γ ν γ µ . This yields ˜ [ i˜ µ (∂µ + iqAµ ) + m ] [ i˜ µ (∂µ + iqAµ ) − m ] Ψ(xµ ) γ γ ˜ = (˜ µ πµ γ ν πν − m2 ) Ψ(xµ ) = 0 .370) is obtained then ˜ [ i˜ µ (∂µ + iqAµ ) + m ] Ψ(xµ ) γ is a solution of (10.e.368) ˜ where Ψ(xµ ) and γ µ are deﬁned in (10. 5. γ ˆ ˜ ˆ (10. γ ˆ2 ˆ2 (10.371) such that we can conclude that (10.370). indeed. can be rewritten using. is a solution of (10. ˜ ˜ ˆ ˆ (10. i.336 Relativistic Quantum Mechanics For the purpose of the solution we assume the chiral representation.s.h. The diﬀerence arises from the term γ µ πµ γ ν πν in (10.369) from the left by γ ν πν + m.230).373) (10. 2.370) for which holds ˜ ˆ ˜ ˆ 3 γ πµ γ πν = ˜ ˆ ˜ ˆ µ=0 µ ν (˜ µ )2 πµ + γ ˆ2 µ.229).370) ˜ [ i˜ µ (∂µ + iqAµ ) − m ] [ i˜ µ (∂µ + iqAµ ) + m ] Ψ(xµ ) = 0 γ γ (10.374). multiplying (10. Equation (10.231).370) Any solution of (10.

γ 0 ] [ˆj .379) The commutators [ˆ0 .ν=1 µ =ν 337 1 2 ( γ µ γ ν πµ πν + γ ν γ µ πν πµ ) ˜ ˜ ˆ ˆ ˜ ˜ ˆ ˆ µ. γ ˜ π ˆ j=1 (10.380) γ µ γ ν πµ πν = i ˜ ˜ ˆ ˆ µ.ν=1 µ =ν = 1 4 1 4 ( γ µ γ ν πµ πν − γ ν γ µ πµ πν + γ ν γ µ πν πµ − γ µ γ ν πν πµ ) ˜ ˜ ˆ ˆ ˜ ˜ ˆ ˆ ˜ ˜ ˆ ˆ ˜ ˜ ˆ ˆ µ. γ j ] are γ ˜ [˜ 0 ..367) holds qA0 = r ˆ Ze2 . πν ] γ ˜ π ˆ µ.367) of Aµ . πj ] + γ ˜ π ˆ 4 0 j [˜ j . π0 ] γ ˜ π ˆ j=1 = [˜ 0 . the commutators [˜ 0 . it holds 1 4 [˜ µ . t) is a suitable test function and where ((· · ·)) denotes the range to which the derivative is limited.ν=1 µ =ν 3 j=1 3 3 = 1 4 1 2 1 [˜ .229). Altogether. i. πj ] in (10.367) and the deﬁnition (10.10: Dirac Particles in Electromagnetic Field indices.e.378) can be evaluated using (10. γ ] [ˆ0 .380) where f = f (r.10.378) According to the deﬁnition (10. γ j ] γ ˜ = = 2 0 1 1 1 0 1 σj 0 0 −σ j 0 σj −σ j 0 − 0 σj −σ j 0 0 1 1 1 0 1 (10. due to A = 0. r2 (10. 3 (10. πν ] γ ˜ π ˆ µ.376) This expression can be simpliﬁed due to the special form (10.377) which follows readily from the deﬁnition (10. Since [ˆµ . 2. ν = 1. γ ν ] [ˆµ .381) According to (10. πj ] π ˆ = = (−i∂t + qA0 . one can summarize (10.ν=1 µ =ν σ 0 0 −σ · (( qA0 )) (10. γ µ γ ν πµ πν = ˜ ˜ ˆ ˆ µ. γ ν ] [ˆµ .ν=1 µ =ν = [˜ µ . πν ] = 0 π ˆ for µ. −i∂j ] = − [ (∂t + iqA0 ) ∂j − ∂j (∂t + iqA0 ) ] f i ((∂j qA0 )) f (10.344) π ˆ [ˆ0 .ν=1 µ =ν (10.376–10.344). πj ] . γ j ] [ˆ0 .382) .

r r 2 2 2 j = 1. it is this quantity that we want to determine. 6. (10. a solution of the form ∼ Y m (ˆ) can be obtained.385) 1 We split the wave function into two spin. . 6. (10. 1 |ˆ) ) .147. (10.5 which describe the coupling of orbital angular momentum and spin 1 {( Yjm (j − 2 . m values are coupled.6. . 2 3 2 . (10.384) into (10. Accordingly.368) reads − Ze2 ∂t − i r 2 + 2 + i σ·r ˆ 0 0 −σ · r ˆ Ze2 − m2 r2 Ψ(xµ ) = 0 (10. + j } (10. Combining this result with (10.148).e.388) Except for the term iσ · r this equation is identical to that posed by the one-dimensional Kleinˆ Gordon equation for pionic atoms (10. i. in r ˆ fact. . Yjm (j + 1 . . We note that these states are also eigenstates of the angular r 2 . . is genuinely two-dimensional and. 1 |ˆ).189) for the Laplacian and expansion of the term (· · ·)2 result in the twodimensional equation 2 ∂r − ˆ L2 − Z 2 e4 iσ · r Ze2 ˆ 2Ze2 + r2 r + 2 − m2 r φ± (r) = 0 .1 .388) in terms of states introduced in Sect.375) the ˆ ‘squared’ Dirac equation (10. m = −j. The term iσ · r.384) can be interpreted as the energy of the stationary state and.2 components ˜ Ψ(r) = and obtain for the separate components φ± (r) Ze2 + r 2 ˜ φ+ (r) ˜ φ− (r) (10.191) solved in Sect. In the latter case. 2 |ˆ)} as given in (6.386) + 2 ± iσ · r ˆ Ze2 − m2 r2 φ± (r) = 0 . Insertion of (10. . Yjm (j + r 2 1 1 .389) According to the results in Sect. 1 |ˆ).. only the two states {Yjm (j − 2 . −j + 1.374).383) yields the purely spatial four-dimensional diﬀerential equation + Ze2 r 2 + 2 + i σ·r ˆ 0 0 −σ · r ˆ Ze2 − m2 r2 Φ(r) = 0 . however. we express 2 the solution of (10. (10. 6. Such solutions are of the form ˜ ˜ Ψ(xµ ) = Φ(r) e−i t .338 Relativistic Quantum Mechanics where r = r/|r| is a unit vector.387) The expression (10. (10.5 the operator iσ · r is block-diagonal in this basis such that only ˆ 1 the states for identical j.381). 10. couples the orbital angular momentum of the electron to its spin.383) We seek stationary solutions of this equation. hence.

2 (j) + 1) 2 Ze2 + + r2 r 2 − m2 f1. except for the slight diﬀerence in the expression of λ1. we want to determine solely the spectrum.199).10.. . we require only the eigenvalues of the matrices B± = 1 (j − 2 )(j + 1 ) − Z 2 e4 2 ±iZe2 ±i Ze2 1 (j + 2 )(j + 3 ) − Z 2 e4 2 . m and expand φ± (r) = Using σ · r Yjm (j ± 1 .2 (j) being 2 j = 1 . conclude that the . (6. Since.395) (10.200). 1 |ˆ) = − Yjm (j ˆ r 2 2 1 2 h± (r) g± (r) Yjm (j − 1 . 1 |ˆ) r 2 (10. which states that the states Yjm (j ± 1 . 10.10: Dirac Particles in Electromagnetic Field 339 ˆ momentum operator L2 [cf. .2 (j) as given by (10.392) h± (r) g± (r) = 0.396) and (10. the missing additive term − 1 . unaltered.f. 1 |ˆ) are r 2 2 ˆ eigenfunctions of L2 . (6.394) Equation (10.391) derived in Sect.2 (j)[λ1. Any similarity transformation leaves the ﬁrst term in (10.151]. involving the 2 × 2 unit matrix.2 (j). 3 .186)]. not the wave functions. the eigenvalues are independent of m.397) This equation is identical to the Klein-Gordon equation for pionic atoms written in the form (10. property (6.392) reads then in the diagonal representation 2 ∂r − λ1. and except for the fact that 2 2 we have two sets of values for λ1. .5 [c.151). We select. − 1 r2 (j − 1 )(j + 1 ) − Z 2 e4 2 2 ±i Ze2 (j + )(j + ) − Z 2 e4 We seek to bring (10. 1 |ˆ) r 2 2 2 2 r r .390) (10.396) and λ2 (j) [λ2 (j) + 1] (10. together with the orthonormality of these two states leads to the coupled diﬀerential equation 2 ∂r + 2Ze2 + r 2 −m 1 0 0 1 ±iZe2 1 2 3 2 (10. λ1 (j) and λ2 (j). a speciﬁc pair of total spin-orbital angular momentum quantum numbers j. 6. We can. . as in the case of pionic atoms. The two eigenvalues of both matrices are identical and can be written in the form λ1 (j) [λ1 (j) + 1] where λ1 (j) λ2 (j) = = (j + 1 )2 − Z 2 e4 2 (j + 1 )2 − Z 2 e4 − 1 2 (10. Obviously.395. . rather than = 0. hence. such transformation can be chosen as to diagonalize the second term. namely. However. 1 |ˆ) + r Yjm (j + 1 . the values of the argument of λ1.392). . therefore. namely. 1.2 (r) = 0 (10. in the present treatment.392) into diagonal form. (10.393) but do not explicitly consider further the wavefunctions.

403) √ Z 2 e4 2 ED (n. 1 ms = ± 2 (10. j = 1 . . . . of these equations describe the binding energy. For a given value of n the energies 1 and 2 for identical j-values correspond to mixtures of states 1 with orbital angular momentum = j − 2 and = j + 1 . the stationary states have binding energies E = − mZ 2 e4 n = 1. 2. (10. 2. 10. .2 . 1.402) In this expression n is the so-called main quantum number. 10. . (10. . One obtains in case of (10.404) = 0.399) mZ 2 e4 + O(Z 4 e8 ) 2 (n + j + 3 )2 2 mZ 2 e4 ≈ m − + O(Z 4 e8 ) 2 2 (n + j + 1 )2 2 n = 0. .401) if one attributes to the respective 1 states the angular momentum quantum numbers = j + 2 and = j − 1 . albeit with some modiﬁcations. −j + 1. 2 2 1 ≈ m − (10. m) 2 n = 1. . j where 1 corresponds to λ1 (j) as given in (10. j = + 1 . . . = 1 + m Z 2 e4 √ ( n − − 1 + 2 − Z 2 e4 )2 . . . . 2. counts the nodes of the wave function. . . j . It is given by n = n + + 1 where is the orbital angular momentum quantum number and n = 0.398. One can equate (10. .395) and 2 corresponds to λ2 (j) as given in (10. . . . . accordingly.400) (10.396) we obtain. Expanding the energies in terms of this parameter allows one to identify is determined by Z the relationship between the energies 1 and 2 and the non-relativistic spectrum.400) and (10. .396). 1. m = − . 1 = 1 + m Z 2 e4 (n + 1 + (j+ 1 )2 − Z 2 e4 )2 2 Z 2 e4 (n + (j+ 1 )2 − Z 2 e4 )2 2 . j = − 2 . m) = 1 + m (n− + . .399) n = 0. 1. One may also state 2 this in the following way: (10. 2.401) corresponds to a non-relativistic state 2 with quantum numbers n. 3 . . . .400) corresponds to a non-relativistic state with quantum numbers n. . . the second term on the r.340 Relativistic Quantum Mechanics spectrum of (10. . The magnitude of the relativistic eﬀect 2 2 e4 .s. 2n2 = 0. j . . . −j + 1. . ( +1) − Z 2 e4 )2 (10. . . .395. (10. . . 1. n − 1 . . . (10. . n − 1 . and spin-orbital angular momentum j = + 1 . Using (10. and spin-orbital angular momentum j = − 1 . .401) These expressions can be equated with the non-relativistic spectrum. m = −j. . . 2 2 m = −j. including spin. 1. . . −j + 1.h. (10.402) with (10. m = −j. . These considerations 2 are summarized in the following equations 1 ED (n. − + 1. . . j = 1. . .398) 2 = 1 + m . In case of non-relativistic hydrogen-type 1 atoms. . 3. . . .203). Obviously.397) is again given by eq.

2: Binding energies for the hydrogen (Z = 1) atom. 10. .60583 -3.e. . the nonrelativistic degeneracy for the six 2p states of the hydrogen atom: in the present. i.402)] and the relativistic [cf. the 2p 3 states with r 2 2 2 2 1 involve four degenerate states corresponding to Y 3 m (1. the non-relativistic and relativistic energies are hardly discernible.036 by means of Eqs. .404) ﬁnally into the single formula ED (n.2 the non-relativistic [cf.e.51176 ⇑ ⇑ ⇑ ⇑ 2s 1 2 2p 1 2 2p 3 2 3s 1 2 3p 1 2 3p 3 2 3d 3 2 3d 5 2 Table 10. is in the range of 10 eV. j. . n − 1 j = 1 2 1 2 ± m = −j. in fact.0041 keV and e2 = 1/137. However. .60601 -3. relate closely to the corresponding nonrelativistic states. 1. . mom.403. relativistic.551177 ⇑ . The 2p 1 states with j = j = 1 2 3 2 2 2 2 for = 0 otherwise involve two degenerate states corresponding to Y 1 m (1. 2. r 2 2 2 .3. in case of heavier nuclei the kinetic energy of bound electrons in the ground state scales with the nuclear charge Z like Z 2 such that in case Z = 100 the kinetic energy is of the order of the rest mass and relativistic eﬀects become important. Of particular interest is the eﬀect of spin-orbit coupling which removes. = 0.402. binding energy / eV Eq.405)] spectrum of the hydrogen atom.. 10. (10.405) In order to demonstrate relativistic eﬀects in the spectrum of the hydrogen atom we compare in Table 10. (10. . 2 |ˆ) for m = ± 1 . .40146 ⇑ ⇑ -1. The table entries demonstrate that the energies as given by the expression (10. case these six states are split into energetically diﬀerent 2p 1 and 2p 3 states. j 1 2 1 2 1 2 3 2 1 2 1 2 3 2 3 2 5 2 341 rel. i. 1 |ˆ) for m = ± 1 .551176 spectr. The reason is that the mean kinetic energy of the electron in the hydrogen atom.405) in terms of the non-relativistic quantum numbers n.551178 ⇑ . much less than the rest mass of the electron (511 keV). . −j + 1.10: Dirac Particles in Electromagnetic Field main quantum number n 1 2 2 2 3 3 3 3 3 orbital angular mom. j (10.405 ) -13. One can combine the expressions (10. m) = m 1 + Z 2 e4 (n−j −1 + 2 (j+ 1 )2 − Z 2 e4 )2 2 n = 1.. (10. for example. 0 0 1 1 0 1 1 2 2 spinorbital ang. notation 1s 1 2 non-rel.402) -13. The energies were evaluated with m = 511.10. (10. (10. This is clearly demonstrated by the comparision of non-relativistic and relativistic spectra of a hydrogen-type atom with Z = 100 in Table 10. ± 3 . .40147 -1.405).1. binding energy / eV Eq.40151 ⇑ . . Degeneracies are denoted by ⇑. .1.3.

15.405).405) reads 1 (10. 0. 1 2 2 2 2 E (2p 3 ) − E (2p 1 ) ≈ − 2 2 mZ 4 e8 2 · 23 1 −1 2 = mZ 4 e8 .405 ) -161. to estimate the diﬀerence between the energies of the states 2p 3 and 2p 1 (cf. j 1 2 1 2 1 2 3 2 1 2 1 2 3 2 3 2 5 2 Relativistic Quantum Mechanics non-rel. (10.3 spectr.407) 1 2 This expression allows one. In order to investigate further the deviation between relativistic and non-relativistic spectra of hydrogen-type atoms we expand the expression (10.342 main quantum number n 1 2 2 2 3 3 3 3 3 orbital angular mom.10.3). (10. It holds for n = 2 and j = 3 . 32 (10.8 ⇑ .406) α √ 1+ (n − β + β 2 −α)2 The expansion (10. Introducing α = Z 2 e4 1 and β = j + 2 (10.402) -136. (10.0 ⇑ ⇑ -15. 0.402. (10.1 -34. . binding energy / keV Eq. 0) (10. mom. We assume for the wave function the stationary state . 10.35.405) to order O(Z 4 e8 ).3: Binding energies for the hydrogen-type (Z = 100) atom. Tables 10.409) where V (r) is spherically symmetric. notation 1s 1 2 2s 1 2 2p 1 2 2p 3 2 3s 1 2 3p 1 2 3p 3 2 3d 3 2 3d 5 2 Table 10.2 -17. The energies were evaluated with m = 511. m) ≈ m − mZ 2 e4 mZ 4 e8 − 2n2 2n3 1 j+ − 3 4n + O(Z 6 e12 ) .6 -42. j.1 ⇑ . binding energy / keV Eq.0041 keV and e2 = 1/137. for example.408) Radial Dirac Equation We want to determine now the wave functions for the stationary states of a Dirac particle in a 4-vector potential Aµ = (V (r). Degeneracies are denoted by ⇑.2.206) provides in the present case ED (n. 0 0 1 1 0 1 1 2 2 spinorbital ang.15.036 by means of Eqs.1 ⇑ ⇑ ⇑ ⇑ rel.9 ⇑ . An example for such potential is the Coulomb potential V (r) = −Ze2 /r considered further below.

2 |ˆ) + r Yjm (j − 2 . m-values. The arguments above allow one to eliminate the angular dependence by expanding Φ(r) and X (r) 1 in terms of Yjm (j + 2 .198)] not couple states with diﬀerent j.412) are four-dimensional with r-dependent wave functions. ˆ σ · p X + m Φ + V (r) Φ ˆ σ · p Φ − mX + V (r) X = = Φ X (10.232.410) 1 where Φ(r).416) The diﬀerential equations (10.5.414) These equations can be brought into a more symmetric form using ∂r + which allows one to write (10.413) ˆ σ · p g(r) Yjm (j − 1 . 1 |ˆ) introduced in r 2 2 1 Sect. 2 |ˆ) r r r . 2 |ˆ) + r Yjm (j − 2 .2 degrees of freedom. X (r) ∈ C2 describe the spatial and spin. σ · p ˆ σ · p f (r) Yjm (j + 1 .417) c(r) d(r) X (r) 1 1 1 1 Yjm (j + 2 .413. the term is block-diagonal in the space spanned by the states Yjm (j ± 1 . 10.5 [see. 10. 1 |ˆ) r 2 2 = i ∂r + j+ r 1 2 3 2 f (r) Yjm (j − 1 . 1 |ˆ) r 2 = i ∂r − 1 2 r g(r) Yjm (j + 1 . but are timeindependent. (10. i. 1 |ˆ) .. 1 |ˆ). r 2 2 (10. i. (6.f. 1 |ˆ) r 2 1 1 = ∂r r r r (10.414) ˆ σ · p rf (r) Yjm (j + 1 . 6. r 2 (10. 1 |ˆ) and Yjm (j − 1 . 168]: the term is a scalar (rank zero tensor) in the space of the spin-angular momentum states Yjm (j ± 1 .e.412) ˆ In this equation a coupling between the wave functions Φ(r) and X (r) arises due to the term σ · p. 1 |ˆ) r 2 2 = i j+ ∂r + r j+ r 1 2 1 r f (r) Yjm (j − 2 . 1 |ˆ) r 2 2 = i ∂r + −j r 1 g(r) Yjm (j + 2 . 1 |ˆ) r 2 2 (10. 2 |ˆ) and does r 2 ˆ has odd parity and it holds [c.197.345). r r 2 2 2 a(r) b(r) 1 1 1 1 Yjm (j + 2 .415) 1 ˆ σ · p rg(r) Yjm (j − 2 .411. 6. 6.e. according to (10. 2 |ˆ) r Φ(r) r . The Dirac equation reads then. pp. = r (10. 1 |ˆ) .411) (10.. This term has been discussed in detail in Sect.10: Dirac Particles in Electromagnetic Field form Ψ(xµ ) = e−i t 343 Φ(r) X (r) . in particular. 10.10.

416). the orthonormality property (6. c(r) are coupled. Inserting (10.423) .418) ∂r + j+ r 1 2 c(r) + [ m + V (r) − ] b(r) = 0 i j+ ∂r − r 1 2 b(r) + [ −m + V (r) − ] c(r) = 0.417) of the form f1 (r) i Yjm (j + 1 . 1 |ˆ) r Φ(r) 2 2 r = (10.415. and multiplying by r results in the following two independent pairs of coupled diﬀerential equations i ∂r − j+ r j+ r 1 2 d(r) + [ m + V (r) − ] a(r) = 0 i and i ∂r + 1 2 a(r) + [ −m + V (r) − ] d(r) = 0 (10.420) g1 (r) X (r) 1 1 − Yjm (j − 2 .418) holds for f1 (r). 2 |ˆ) r r f2 (r) 1 1 r Φ(r) i r Yjm (j − 2 .417) into (10. there exist two independent solutions (10. d(r) are coupled and b(r).419) Obviously.411. Presently. m values. 2 |ˆ) = (10.422) j+ r j+ r 1 2 g2 (r) + [ − m − V (r) ] f2 (r) = 0 ∂r − 1 2 f2 (r) − [ + m − V (r) ] g2 (r) = 0 (10.421) g2 (r) X (r) 1 1 − Yjm (j + 2 . 2 |ˆ) r r where the factors i and −1 have been introduced for convenience. Accordingly. (10.412). we consider the case that only states for one speciﬁc j. m pair contribute. g1 (r) ∂r − j+ r j+ r 1 2 g1 (r) + [ − m − V (r) ] f1 (r) = 0 ∂r + and for f2 (r). such expansion must include states with all possible j.157). 10.344 Relativistic Quantum Mechanics In general. According to (10. using (10. only a(r). 10. g2 (r) ∂r + 1 2 f1 (r) − [ + m − V (r) ] g1 (r) = 0 (10.

405). determining wave functions of the type (10. We assume the 4-vector potential of pure Coulomb type (10. one demonstrates ﬁrst that the wave function at r → 0 behaves as rγ for some suitable γ. . the radial wave functions for Dirac particles in the potential (10. 3d 5 .422) describes the states 2p 1 .422) near r = 0. 10. 3p 3 .425) r 0 covers states with angular momentum 1s 1 .10. Dirac Particle in Coulomb Field . except for the opposite sign of the term (j + 1 ). to states with angular momentum = j + 2 . Equation (10. 2 2 2 2 2 2 = j − 1 2 and. can be written ∂r − j+ r j+ r 1 2 g1 (r) + Ze2 f1 (r) r Ze2 g1 (r) r = 0 ∂r + 1 2 f1 (r) − = 0.409). . . . etc.Wave Functions We want to determine now the wave functions of the stationary states of hydrogen-type atoms which correspond to the energy levels (10.422. According to this procedure.422). 2 2 2 (10. . (10.423) apply for V (r) = −Ze2 /r. and obtains ﬁnally a polynomial function p(r) such that rγ exp(−µr)p(r) solves (10. Hence. Φ in (10. 3d 3 ..420).426) . (10. Similarly. the 2 equations determine..422) and (10. for small r.424) Ψ(xµ ) ≈ r 0 1 i. In the non-relativistic limit. 1.422) determines solutions of the form (10.422). = 0. correspodingly the states We consider ﬁrst the solution of (10. one demonstrates then that the wave functions for r → ∞ behaves as exp(−µr) for some suitable µ. enforcing the polynomial to be of ﬁnite order leads to discrete eigenvalues . 3p 1 . n − 1.422) follows in this case from the same procedure as that adopted for the radial wave function of the non-relativistic hydrogen-type atom. (10.405). i.422) corresponds to states f1 (r) Yjm (j + 1 . (10. . The solution of (10. 1 |ˆ) r i 2 2 . . 2s 1 .e. . We note that (10. Hence.e.405) of the relativistic hydrogen atom the corresponding states have quantum numbers n = 1. Behaviour at r → 0 We consider ﬁrst the behaviour of the solutions f1 (r) and g1 (r) of (10. the ones given in (10. 3s 1 .423) are identical.420) is the large component and X is the small component.423).422).421). in the non-relativistic limit wave functions f2 (r) 1 1 i Yjm (j − 2 . namely.10: Dirac Particles in Electromagnetic Field 345 Equations (10. According to the discussion of the spectrum (10. 2. 2p 3 . (10. together with the appropriate boundary conditions at r = 0 and r → ∞. 2 |ˆ) r Ψ(xµ ) ≈ .367) which is spherically symmetric such that equations (10. etc.

becomes imaginary. in case j + 1 )2 < Ze2 . (10. we conclude f1 (r) →∞ ∼ e−µr .430) 2 Note that the exponent in (10.427).427) = = 0 0. for small r.434) √ The solutions of these equations are f1 . r2 (10.432) = = − ( − m.429) This equation poses an eigenvalue problem (eigenvalue −γ) for proper γ values.422) becomes ∂r g1 (r) ∂r f1 (r) Iterating this equation once yields 2 ∂r g1 (r) 2 ∂r f1 (r) 1 . µ = m2 − 2 (10. We have.427) with 1 γ = (j + )2 − Z 2 e4 .346 Setting f1 (r) yields γ b rγ−1 − (j + 1 ) b rγ−1 + Ze2 a rγ−1 2 γ a rγ−1 + (j + 1 ) a rγ−1 − Ze2 b rγ−1 2 or γ + (j + 1 ) −Ze2 2 2 Ze γ − (j + 1 ) 2 a b →0 Relativistic Quantum Mechanics ∼ a rγ . 10.427) makes only the positive solution 2 possible. determined that the solutions f1 (r) and g1 (r). according to (10.435) . The assumed r-dependence in (10. Such r-dependence 2 would make the expectation value of the potential r2 dr ρ(r) 1 r (10.428) = 0. (10. One obtains γ = ± (j + 1 )2 − Z 2 e4 . assume the r-dependence in (10. (10.420). g1 ∼ exp(± m2 − 2 r). g1 (r) →∞ ∼ e−µr . for the particle density holds then ρ(r) ∼ |rγ−1 |2 = Behaviour at r → ∞ For very large r values (10. Only the exponentially decaying solution is admissable and. hence.267. hence. g1 (r) →0 ∼ b rγ (10.433) = = ( m2 − 2 ) g1 (r) ( m2 − 2 ) f1 (r) (10. 10. ) f1 (r) ( + m ) g1 (r) (10.266.431) inﬁnite since.

443) From this results after a little algebra j+1 m + Ze2 2 ] (φ1 − φ2 ) − (φ1 + φ2 ) + φ2 = 0 ρ m− ρ j+1 m − Ze2 2 [∂ρ + ] (φ1 + φ2 ) + (φ1 − φ2 ) − φ2 = 0 .441) The last two terms on the l.442) which leads to a partial cancellation of the asymptotically dominant terms.440) and (10. We also introduce the new variable ρ = 2µ r .434) √ m + a e−µ r .445) . of both (10.441) correspond to (10.10: Dirac Particles in Electromagnetic Field For bound states holds 347 < m and. (10. Let us consider then for the solution of (10.444) (10. Accordingly.434) results in √ √ (m − ) m + a − (m − ) m + a √ √ (m + ) m − a − (m + ) m − a which is obviously correct.s. In the present case the functions f1 and g1 cannot be chosen ˜ ˜ identical due to the terms in the diﬀerential equations contributing for ﬁnite r. Solution of the Radial Dirac Equation for a Coulomb Potential To solve (10. However. µ is real.438) (10.435 ). (10. Equation (10.422 ) leads to √ √ j+1 Ze2 ˜ 2 − m − [∂r − ] g1 + m + ˜ f1 + r √r √ ˜ (m − ) m + g1 − (m − ) m + f1 = 0 ˜ 1 2 √ √ j+2 ˜ Ze m + [∂r + ] f1 + m − g1 − ˜ r r √ √ ˜ (m + ) m − f1 + (m + ) m − g1 = 0 ˜ (10. hence.422) for the Coulomb potential V (r) = −Ze2 /r We assume a form for the solution which is adopted to the asymptotic solution (10.436) f1 (r) = Insertion into (10. g1 (r) = φ1 (r) − φ2 (r) ˜ (10.436).10.440) (10.h.437) where they ˜ ˜ cancelled in case f1 = g1 = a.437) (10. without loss of generality we can choose ˜ f1 (r) = φ1 (r) + φ2 (r) .439) − m− e g1 (r) ˜ where µ is given in (10. √ g1 (r) = − m − a e−µr . we set f1 (r) g1 (r) = = √ √ ˜ m + e−µr f1 (r) −µr = = 0 0 (10. ρ m+ ρ [∂ρ − (10.

430) which conform to the proper r → 0 behaviour determined above [c. (10.447) of the form n φ1 (ρ) = ργ s=0 n γ s=0 αs ρ2 βs ρ2 (10. Inserting (10.453) (s + γ)2 + Z 2 e4 − (j + 1 )2 2 2 √ Ze m2 − 2 βs .450) −√ mZe2 m2 − 2 βs ρs+γ−1 = 0 mZe2 m2 − αs (s + γ) β2 + (j + 1 ) αs + √ 2 s 2 +√ From (10.452) From (10.447) leads to (s + γ) αs + (j + 1 ) βs − √ 2 s Ze2 m2 − 2 αs (10.348 Relativistic Quantum Mechanics Addition and subtraction of these equations leads ﬁnally to the following two coupled diﬀerential equations for φ1 and φ2 ∂ρ φ1 + ∂ρ φ2 + j+ ρ 1 2 Ze2 mZe2 φ1 − √ φ2 = 0 m2 − 2 ρ m2 − 2 ρ mZe2 Ze2 φ1 + √ φ1 + √ φ2 − φ2 = 0 m2 − 2 ρ m2 − 2 ρ j+ ρ φ2 − √ 1 2 (10.446.430)].449) into (10.448) φ2 (ρ) = ρ (10.446) (10.448. (10.430) one can write this βs = s+γ − 2 √ Ze m2 − 2 s (s + 2γ) βs−1 .454) .426–10. 10.f. 10. = m− Ze2 βs s + γ − √m2 − 2 2 (10.447) We seek solutions of (10.450) follows Ze2 m2 − 2 βs − βs−1 =0 (10.451) follows βs−1 = = Ze2 s+γ + √ m2 − s+γ − βs + m2 Z 2 e4 m2 − 2 − (j + 1 )2 2 2 √ Ze m2 − 2 2 s+γ − βs (10.449) for γ given in (10. 10.446 .451) 1 √mZe − (j + 2 ) 2 2 αs . Using (10.

.448..459) − 2 √mZe m2 − 2 (10.452) follows j+ αs = 1 2 349 Ze2 m2 − 2 −γ (10. 2. hence. 3d 3 holds n = 1. 2γ + 1. x) = 1 + a a(a + 1) x2 x + + . 10..457) One can relate the polynomials φ1 (ρ) and φ2 (ρ) deﬁned through (10. ρ) . conclude 2 2 2 that the polynomials in (10. . .457) with the conﬂuent hypergeometric functions F (a.405) allows one to identify n = n − j + 2 which.457) for βs and αs imply that so must then be an integer. x) . (s − so ) β0 s! (2γ + 1)(2γ + 2) · · · (2γ + s) (10.10. n = 0.455) this conﬁnement of so implies discrete values for .449) must be of ﬁnite order.461) In order that the wave functions remain normalizable the power series (10. . 10. . . This requires that all coeﬃcients αs and βs must vanish for s ≥ n for some n ∈ N. c c(c + 1) 2! (10.449) and (10. (s − so ) β0 s! (2γ + 1)(2γ + 2) · · · (2γ + s) (10.455) s − so βs−1 s(s + 2γ) (s − 1 − so )(s − so ) βs−2 (s − 1)s (s − 1 + 2γ)(s + 2γ) (1 − so )(2 − so ) . 3p 1 .461 for values given by (10. i. for the states 2p 1 . so = n . . For example. . equivalently..458) or. namely.448. in fact. is an integer.460) (10. = From (10. Comparision with (10. 1. We can. 1.455) are ﬁnite. . .405) and the ensuing so values ( 10. 10. The expressions (10.e.456.456) − so 2 √mZe m2 − 2 (−so )(1 − so )(2 − so ) . 2. m (n ) = . c. n It holds φ1 (ρ) φ2 (ρ) = = ργ F (−so . α + 1. According to the deﬁnitions (10. ρ) so β0 ργ F (1 − so .456) and (10.430.405). β0 j+ 1 2 (10. (10. 10.462) Z 2 e4 1 + 1 (n + (j+ 2 )2 − Z 2 e4 )2 This expression agrees with the spectrum of relativistic hydrogen-type atoms derived above and 1 given by (10. with the associated Laguerre polynomials L(α) = F (−n. 2γ + 1.10: Dirac Particles in Electromagnetic Field Deﬁning so = √ one obtains βs = = .

which involve the conﬂuent hypergeometric functions in (10.421) correspond to non-relativistic states with orbital angular momentum 1 1 = j + 2 . 6. The wave functions (10. j.m (j + 1 . m|xµ ) = e−i t iF1 (r) Yj.461).421) with radial wave functions f1 (r). (10.463) where Φ and X . = j + 2 . where2 F± (κ|r) = G1 (r) = F+ (κ|r) .460.442).468) −κ n! µ γ n = = = = (m − )(m + ) (j + 1 )2 − Z 2 e4 2 n−j − m 1+ 1 2 Z 2 e4 (n +γ)2 .467) ± n F (1 − n .267).350 Relativistic Quantum Mechanics Altogether we have determined the stationary states of the type (10.439). 2µr) (2µ) 2 Γ(2γ + 1) 3 N = and m 4m )Γ(2γ + n + 1) (n +γ)m (n +γ)m (10. is time-independent. are two-dimensional vectors determined through the explicit form of the spin-orbital angular momentum states Yjm (j ± 1 . They are described through quantum numbers n. m.438. given by expression (10.m (j − 1 . (n + γ)m N (2µr)γ−1 e−µr − κ F (−n . Greiner. 1 |ˆ) in (6.461) are to be chosen to satisfy a normalization condition and to assign an overall phase.463) and yield [note the 1/r factor in (10. Due to the form (10. can follow the procedure adopted for the wave functions of the non-relativistic hydrogen atom and will not be carried out here.461). 10.464) 0 The evaluation of the integrals. as in (10.466) F1 (r) = F− (κ|r) .148). (Springer. and (10. 1990).469) This formula has been adapted from ”Relativistic Quantum Mechanics” by W. The complete wave function is given by the following set of formulas 1 Ψ(n.158) of the latter states absorb the angular integral in (10. The normalization integral is then ∞ 0 π 2π r2 dr 0 sin θdθ 0 dφ (|Φ(r)|2 + |X (r)|2 ) = 1 (10. 2γ + 1.465) (10. 10. The orthonormality 2 2 r properties (6. 2 . 1 |ˆ) 2 2 r κ = j+ 1 2 (10.460. 2γ + 1.421)] ∞ dr (|f1 (r)|2 + |g1 (r)|2 ) = 1 (10. (10.410) of the stationary state wave function the density ρ(xµ ) of the states under consideration. Sect. 6. The coeﬃcients β0 in (10. g1 (r) determined by (10.460.157. 9. 10. j. 1 |ˆ) 2 2 r G1 (r) Yj. 2µr) (10. = j + 2 .147.6. Berlin. 10.421).

423) which diﬀers from the radial Dirac 1 equation for f1 (r) and g1 (r) solely by the sign of the terms (j + 2 )/r.m (j − 1 . this wavefunction has an orbital angular momentum quantum number = j − 1 and. hence.469).10. 2p 3 .470) r 0 Obviously. . One can verify. j. 1 |ˆ) 2 2 r G2 (r) Yj. 2s 1 .10: Dirac Particles in Electromagnetic Field 351 We want to consider now the stationary states of the type (10.421) are governed by the radial Dirac equation (10.472) F2 (r) = F− (κ|r) .469). The radial wave functions f2 (r) and g2 (r) in (10. obtained closed expressions for the wave functions of all the stationary bound states of relativistic hydrogen-type atoms. in the non-relativistic limit. G2 (r) = F+ (κ|r) . 3p 3 .465–10. become f2 (r) 1 1 i Yjm (j − 2 .421) which. describes the complementary set of states 1s 1 . We have. 3s 1 .m (j + 1 .467–10. tracing all steps which lead from (10. not not 2 2 2 2 2 2 covered by the wave functions given by (10.469) that the following wave functions result Ψ(n. 2 |ˆ) r Ψ(xµ ) ≈ e−1 t . etc. = j − µ 1 2 .422) to (10. where F± (κ|r) are as given in (10. m|x ) = e −i t iF2 (r) Yj. (10.471) (10. 3d 5 . 2 accordingly. 1 |ˆ) 2 2 r κ = −j − 1 2 (10.

352 Relativistic Quantum Mechanics .

229). in particular.1 The Lorentz Transformation of the Dirac Bispinor We will provide in the following a new formulation of the Dirac equation in the chiral representation ˜ deﬁned through (10. We will characterize the space on which the transformations a(z) We will see below that the representations a(z) and b(z) are. This transformation bispinor wave function Ψ can be written ˜ S(z) a(z) b(z) z = = a(z) 0 0 b(z) exp 1 z·σ 2 1 − z ∗· σ 2 (11.e.1) (11.. a complex variable z. ϑ through 3. that the corresponding representation of the Lorentz group is more basic than the natural representation and may serve as a building block for all representations. The lower dimensionality of a(z) and b(z) implies. expressing its dependence on w. a(z) and b(z). must be two-dimensional irreducible representations of the Lorentz group. ϑ) for the ˜ in the chiral representation as given by (10. 1 353 . ϑ).Chapter 11 Spinor Formulation of Relativistic Quantum Mechanics 11.3) (11.e. in a sense. indeed. We will proceed by building as much as possible on the results obtained sofar in the chiral representation of the Dirac equation. diﬀerent L(w. z ∈ C ˜ Because of its block-diagonal form each of the diagonal components of S(z).225–10.. what will be achieved in the following. This fact is remarkable since it implies that the representations provided through a(z) and b(z) are of lower dimension then the four-dimensional natural representation1 L(w. Starting point is the Lorentz transformation S(w. isomorphic to the natural representation. ˜ We have altered here slightly our notation of S(w. in fact. This is. i.262). may be exploited to express the Lorentz-invariant equations of relativistic quantum mechanics. i.4) = exp = w − iϑ . ϑ)µ ν . ϑ)µ ν correspond to diﬀerent a(z) and b(z).2) (11.

i.. θ ∈ R3 . express 4-vectors Aµ . ϑ ∈ R3 (11. For ϑ = (0. ± 1 represents the familiar spin– 1 states. Ψ2 (xµ ) and Ψ3 (xµ ).8) We like to stress that there exists. β. Ψ2 (xµ ) transform according to a(z) whereas Ψ3 (xµ ). the so-called spinor space. Ψ4 (xµ ) in case z = w + iϑ for w = 0. (11. and since b(z) ˜ ˜ acts on the third and fourth component of Ψ one can characterize Ψ ˜ Ψ1 (xµ ) ˜ Ψ2 (xµ ) Ψ3 (xµ ) ˜ ˜ Ψ4 (xµ ) ∼ ∼ ∼ ∼ 1 |2. a space of vectors state1 for which holds state2 state 1 ∼ | 1 . and the Klein–Gordon equation in the spinor representation.243).e. will establish the map between L(w. 2 2 2 ˜ Since a(z) acts on the ﬁrst two components of the solution Ψ of the Dirac equation. + 1 2 2 state 2 ∼ | 1 .e. ﬁnally.1 states as described by Dm m (ϑ). Here | 1 . are elements of SU(2) and correspond. In this case holds a(z) = b(z) ˜ 1 (xµ ). γ (see Chapter 5). the operator ∂µ and the Pauli matrices σ in the new representation and. −1 2 2 |1. namely. +1 2 |1. neutrino equation.9) a∗ (z) = exp 2 . This characterization allows one to draw conclusions regarding the state space in which a(z) and b(z) operate. usually expressed 2 as product of rotations and of functions of Euler angles α. (11. in 2 fact. A First Characterization of the Bispinor States We note that in case w = 0 the Dirac transformations are pure rotations.7) where “∼” stands for “transforms like”. i. In this case a(z) and b(z) are identical and read a(i ϑ) = b(i ϑ) = exp 1 − ϑ·σ 2 . formulate the Dirac equation. however.5) The transformations in this case. (11. Aν . ϑ ∈ R3 . −1 2 2 z = iϑ. b(z). β. ϑ ∈ R3 .6) as given by (5. 0)T the transformations are (1) a(i β e2 ) = b(i β e2 ) = ˆ ˆ 2 dmm (β) (1) = cos β 2 sin β 2 − sin β 2 cos β 2 . for z = iϑ.. Ψ4 (xµ ) transform according to ˜ ˜ ˜ and Ψ b(z). Relationship Between a(z) and b(z) The transformation b(z) can be related to the conjugate complex of the transformation a(z). to 1 ∗ ∗ z ·σ (11. to the rotational transformations of spin. − 1 2 2 z = iϑ.354 Spinor Formulation and b(z) act. a distinct diﬀerence in the transformation behaviour of ˜ ˜ ˜ ˜ Ψ1 (xµ ). ϑ)µ ν and a(z). +1 2 2 |1.

one can show −1 −1 −1 = −σ1 = σ2 = −σ3 ∗ = −σ1 ∗ = −σ2 σ2 −1 (11. 1. The answer lies in the necessity that application of spatial inversion should transform a solution of the Dirac equation into another possible solution of the Dirac equation. −1 = ∗ −σ3 = −σ ∗ .2. (11.12).10. (11. Hence.11) one ﬁrst demonstrates that for and −1 as given in (11. in fact.224)] ∗ σ1 = σ1 . σ3 ).12) does.18) with a(z) given by (11. One can readily verify [c. The eﬀect of inversion on Lorentz transformations is. Ψ (xµ ) even though these pairs of ˜ ˜3 ˜4 featuring the components Ψ1 (x ˜ 2 components transform independently of each other. Ψ (xµ ) as well as Ψ (xµ ).17) We conclude. (11.14) . using f (a) −1 = f ( a −1 ).10) From this one can derive b(z) = where = 0 1 −1 0 . hold −1 = 1 One notices then. that they alter w into −w. (5.4) and . This demonstrates that a(z) is the ˜ transformation which characterizes both components of S(z). 11. therefore. a result to be used further below. one can express a∗ (z) −1 = exp 1 − z ∗· σ 2 = b(z) . Spatial Inversion ˜ One may question from the form of S(z) why the Dirac equation needs to be four-dimensional.15) (11. 11.11) 0 −1 1 0 = . µ ). ∗ σ 3 = σ3 .13) Explicit matrix multiplication using (5. ∗ σ 2 = − σ2 . however. 355 (11. (11.1) can be written in the form ˜ S(z) = a(z) 0 0 a∗ (z) −1 (11. −1 given by (11. but leave rotation angles ϑ unaltered. −1 a∗ (z) −1 (11.224.12) To prove (11.12) yields σ1 σ3 or in short σ Similarly.11. . a∗ (z) −1 = exp 1 ∗ z σ∗ 2 −1 .f. that the transformation (11.1: Lorentz Transformation of Dirac Bispinor ∗ ∗ ∗ where σ ∗ = (σ1 . σ2 .16) σ = −σ ∗ . 11.

. det(M ) = 1 } . non-singular matrix M holds2 det eM and from [c. (11.3.19) i. Part (Elsevier.356 Spinor Formulation ˜ Let P denote the representation of spatial inversion in the space of the wave functions Ψ.21) = etr( 2 z·σ) = 1 1 (11.e. is an element of SU(2). The proof of this important property is straightforward in case of hermitian M (see Chapter 5).. (11.G.22) which follows from the fact that for any complex. a(w + iϑ) for w = 0 is an element of SL(2. hence. 1990). However. for particles like the neutrinos which do not obey inversion symmetry two components of the wave function are suﬃcient. (11. P −1 = P.A. In fact.f. the transformations a(z) and b(z) become interchanged. 11.1) We have pointed out that a(iϑ). One can verify this by evaluating the determinant of a(z) det( a(z) ) = det e 2 z·σ 1 (11.A. C) deﬁned in (11. can be found in G. Amsterdam. ˜ ˜ P S(w + iϑ) P = S(−w + iϑ) = b(z) 0 0 a(z) . M is a complex 2 × 2–matrix. the space spanned by only two of the components of Ψ is not invariant under spatial inversion and.2 Relationship Between the Lie Groups SL(2. For the general case the proof. Exercise 1. 2 . Obviously. ϑ ∈ R3 . This implies ˜ Ψ1 ˜ Ψ2 P ˜ Ψ3 ˜ Ψ4 ˜ Ψ3 ˜ Ψ4 = Ψ1 ˜ ˜ Ψ2 .23) Exercise 11.1: Show that SL(2. (5. de Kerf Lie a Algebras. The transformation S(z) in the transformed space is then ˜ P 1.21) together with matrix multiplication as the binary operation forms a group. 3 .2. j = 1. 2. the Lorentz invariant equation for neutrinos is only 2–dimensional.20) ˜ Obviously.C) and SO(3.10.224)] tr( σj ) = 0 .e. based on the Jordan–Chevalley theorem. C) = { M. B¨uerle and E. which describes pure rotations. 2 = 1 i. does not suﬃce for particles like the electron which obey inversion symmetry. However.24) = etr(M ) (11.

27) −→ Straightforward transformation into another frame of reference would replace σµ by σµ . σ2 .25) can also be written M (Aµ ) = A0 + A3 A1 − iA2 A1 + iA2 A0 − A3 . In fact. The following important property holds for M (Aµ ) det ( M (Aµ ) ) = Aµ Aµ which follows directly from (11.30).. (11.11.1) Mapping Aµ onto matrices M (Aµ ) 357 We want to establish now the relationship between SL(2. Starting point is a bijective map between R4 and the set of two-dimensional hermitian matrices deﬁned through M (Aµ ) = σµ Aµ where (11. (11. (11. (11.76).31) Exercise 11. σ2 . i. Using Aµ = (L−1 )µ ν A ν one would expect in a transformed frame to hold Lρ ν σ (L−1 )µ ν A ν . (11. (11. σ3 are hermitian. the matrix M (Aµ ) is hermitian as can be seen from inspection of (11.28) σµ = 1 0 0 1 .2.29) where we used (10.135)] this transformation behaviour.28) and (11. in fact. one wishes that the deﬁnition (11. Demonstrate that (11.30) Since the components of Aµ are real.75) holds aν = Lν µ aµ one realizes that σµ transforms inversely to covariant 4–vectors. The function M (Aµ ) is bijective. 1 0 0 −1 .25) σ0 σ1 σ2 σ3 The quantity σµ thus deﬁned does not transform like a covariant 4–vector.2: Lie Groups SL(2.26) (11. −→ µ Consistency of (11. Noting that for covariant vectors according to (10.27) requires then σ µ Aµ Lν µ σµ = σν (11.C) and SO(3. Argue why M (Aµ ) = σµ Aµ provides a bijective map.2: Show that σ0 . σ3 provide a linear–independent basis for the space of hermitian 2 × 2–matrices. or+ thochronous Lorentz transformations. σ1 . M (Aµ ) according to (11.32) .25) of the matrix M (Aµ ) is independent of the frame of reference. one can provide a simple expression for the inverse of M (Aµ ) M = M (Aµ ) ↔ Aµ = 1 tr M σµ 2 . σ1 . 0 −i i 0 .e. in a transformed frame should hold Lρ ν σ µ Aµ σµ A µ .26) or from the fact that the matrices σ0 .31) holds. We will prove below [cf. 0 1 1 0 . C) and the group L↑ of proper.

37) which implies that (11.37) using (11.3: Show that L(a)µ ν deﬁned in (11. Due to det(a) = 1 the transformation (11. + . In fact.40) is an element of L↑ .35) holds for this transformation A µ Aµ = Aµ Aµ (11.36) The suitable A µ can readily be constructed using (11. (11. any a ∈ SL(2. C) deﬁnes the transformation [c. (11.358 Transforming the matrices M (Aµ ) Spinor Formulation We deﬁne now a transformation of the matrix M (Aµ ) in the space of hermitian 2 × 2–matrices a M −→ M = a M a† .25) Aµ = which allows us to express ﬁnally A µ = L(a)µ ν Aν . for any a ∈ SL(2.34) where we used the properties M † = M and (a† )† = a. 11.31)] 1 a Aµ −→ A µ = tr a M (Aµ ) a† σµ 2 Because of (11. The linear character of the transformation becomes apparent expressing A µ as given in (11. C) and for any Aµ there exists an A µ such that M (A µ ) = a M (Aµ ) a† . Accordingly.32.33) conserves the determinant of M .33) to M (Aµ ) describing the action of the transformation in terms of transformations of Aµ .31).33) det(M ) = = det( a M a† ) = det(a) det(a† ) det(M ) [det(a)]2 det(M ) = det(M ) . L(a)µ ν = 1 tr a σν a† σµ 2 . (11. (11.37) deﬁnes actually a Lorentz transformation.39) Exercise 11. (11. a ∈ SL(2.f.40) 1 tr a σν a† σµ 2 Aν (11. C) .2.35) We now apply the transformation (11. it holds for the matrix M deﬁned through (11.33) This transformation conserves the hermitian property of M since (M )† = ( a M a† )† = (a† )† M † a† = a M a† = M (11. In fact. (11.38) .

1) For this purpose one writes using tr(AB) = tr(BA) L(a1 )µ ν L(a2 )µ ρ = ν a1 a2 product in SL(2. i.δ (11.C) )µ ρ (11. C) and of SO(3.1).43) (σν )αβ Γβα Γγδ (σν )δγ = ν.C) and SO(3.44) (σν )αβ (σν )δγ .e.. 1 ¯ and using the deﬁnition of Lµ ρ in (11. β δ = δ = = δ = δ 1 2 = 2 = 1 Aαβγδ (11. β 2.δ 1 4 Aαβγδ Γβα Γγδ α.α. (11.11. 2 (11.41) results in 1 ¯ Lµ ρ = 4 where Aαβγδ = ν Γ = a2 σρ a† 2 (11.1) deﬁned through L(a)µ ν [cf.β γ. (11.40)] respects the group property of SL(2. (11.42) = ν Deﬁning Γ = a† σµ a1 .45) One can demonstrate through direct evaluation 2 α = 2 α = 2 α = = 2 α = 0 else β β γ γ = = = = γ = γ = 1. ¯ Lµ ρ = L(a1 )µ ν L(a2 )µ ρ = L( product in SO(3.46) which yields ¯ Lµ ρ = = = 1 1 Γ11 Γ11 + Γ22 Γ22 + Γ12 Γ21 + Γ21 Γ12 = tr ΓΓ 2 2 1 1 tr a† σµ a1 a2 σρ a† = tr σµ a1 a2 σρ a† a† 1 2 2 1 2 2 1 tr a1 a2 σρ (a1 a2 )† σµ = L(a1 a2 )µ ρ .β γ.47) This completes the proof of the homomorphic property of L(a)µ ν .1) L(a)µ ν provides a homomorphism 359 We want to demonstrate now that the map between SL(2.2: Lie Groups SL(2.41) 1 tr a1 σν a† σµ 1 2 1 tr σν a† σµ a1 1 2 1 tr a2 σρ a† σν 2 2 1 tr a2 σρ a† σν 2 2 . C) and SO(3. .

Aµ = (0. 0.58) . we write (11. To obtain the generator of a(z) corresponding to the generator K1 .48) We want to determine now the generators of the transformation a(z) as deﬁned in (11.54) This reads for the four cases Aµ = (1. of (11. 1 (11. J3 of the Lorentz transformations Lµ ν in the natural representations.55) (11.h.2) which correspond to the generators K1 .360 Generators for SL(2. C) as complex 2 × 2–matrices are deﬁned through four complex or. Aµ = (0.49).49) assuming (note that g µ ν is just the familiar Kronecker δµν ) Lµ ν a = = gµν + 1 + 1 (K1 )µ ν κ1 . 0. (11. J1 . 10. For the r. (11. Aµ = (0. K2 . 0). i.e. 0. 1. 0).s.52) where we employed (11. correspond to the generators given by (10.36) M (Lµ ν Aν ) = a M (Aµ ) a† (11. 1) − σ1 σ0 0 0 One can verify readily that 1 κ1 = − σ 1 2 (11.57) . eight real numbers. 0.47.48). correspondingly.51) Insertion of (K1 )µ ν as given in (10.49) we obtain using (11.50) (11.h. J2 . One expects then that six generators Gj and six real coordinates fj can be deﬁned which allow one to represent a in the form 6 a = exp j=1 fj Gj .53) results in the condition σ 0 A1 − σ 1 A0 = ( κ 1 σ µ + σ µ κ † ) Aµ .s. 1. Because of the condition det(a) = 1 only six independent real numbers actually suﬃce for the deﬁnition of a.. denoted below as κ1 . K3 . 0). 1 (11. 0. To this end we consider inﬁnitesimal transformations and keep only terms of zero order and ﬁrst order in the small variables. noting the linearity of M (Aµ ).59) = = = = κ1 σ 0 + σ 0 κ† = κ 1 + κ† 1 1 κ1 σ 1 + κ1 σ 2 + κ1 σ 3 + σ 1 κ† 1 σ 2 κ† 1 σ 3 κ† 1 (11. 0) ) σ0 A 1 − σ1 A 0 (11.51) ( 1 + κ1 ) M (Aµ ) ( 1 + κ† ) 1 1 1 = = M (Aµ ) + M (Aµ ) + ( κ1 M (Aµ ) + M (Aµ )κ† ) + O( 2 ) 1 ( κ1 σµ + σµ κ† ) Aµ + O( 2 ) . 0. of (11.56) (11. −A0 . (11.52) and (11.48) yields for the l. C) which correspond to K.53) Equating (11. M (Aµ + (K1 )µ ν Aν ) = = M (Aµ ) + M (Aµ ) − M ( (−A1 . 0.25) in the last step. J Spinor Formulation The transformations a ∈ SL(2.

61) This identiﬁes the transformations a(z = w − iϑ) as representations of Lorentz transformations.4: Show that the generators (11. (11. C) correspond to the generators K and J of Lµ ν . C) acts on spinors φα ∈ C2 (characterized below) 1 (11.2.60) of a ∈ SL(2. λ3 corresponding to J1 . hence. def α = 1. κ3 of a(z) corresponding to K2 .3 Spinors Deﬁnition of contravariant spinors We will now further characterize the states on which the transformation a(z) and its conjugate complex a∗ (z) act.63) where we extended the summation convention of 4-vectors to spinors. one can show that the generators κ2 . 2 2 φ1 φ2 ∈ C2 .64) . the so-called contravariant spinors. Similarly.1) acts on 4–vectors Aµ . + 1 2 2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 . 2 (11. 11.62) We denote the general (z = w + iϑ) transformation by φ α = aα β φβ = aα 1 φ1 + aα 2 φ2 . Exercise 11. 2 (11. λ2 . state that the following two transformations are equivalent L(w. J2 . 2 λ = i σ.3: Spinors 361 obeys these conditions.11. − 1 . Here a(z) = ( aα β ) = describes the matrix a(z). ϑ) = ew·K + ϑ·J ∈ SO(3. a(w − iϑ) = e− 2 (w − iϑ)·σ ∈ SL(2. K3 and λ1 .60) We can. We consider ﬁrst the transformation a(z) which acts on a 2-dimensional space of states denoted by φα = According to our earlier discussion holds φ1 φ2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 . w describing boosts and ϑ describing rotations. a1 1 a1 2 a2 1 a2 2 (11. J3 are given by 1 κ = − σ.

68) (11. exchange of φα and χβ alters the sign of the expression. is the singlet state. indeed. In this case spinors φα transform like spin– 1 states and an invariant.e. 0| 1 . exist and it plays a role for spinors which is as central as the role of the scalar product Aµ Aµ is for 4–vectors. should remain invariant under transformations a(iϑ).. (11. m 1 | .. 1 . Deﬁnition of covariant spinors Expression (11. 0 = 2 2 m=± 1 2 1 1 (0. as contravariant spinors and to φα . i.66) should constitute a singlet spin state.67) We will refer to φα . The relationship between the two can be expressed φ1 φ2 = = 3 φ1 φ2 0 1 −1 0 (11. To arrive at a suitable scalar product we consider ﬁrst only rotational transformations a(iϑ).e. Such a scalar product does. .. 0| 1 . However. 0. invariant and as such has the necessary properties of a scalar3 product. 0| .69) (11. The existence of a scalar product gives rise to the deﬁnition of a dual representation of the states φα denoted by φα . −m 2 2 2 2 1 1 1 1 2 (11. this scalar product is anti-symmetric. i. which 2 can be constructed from products φ1 χ2 .e. . m. ± 1 . . −m) | . those of “particle 2” with the spinor χβ one can state that the quantity 1 Σ = √ 2 φ1 χ2 − φ2 χ1 (11. i. etc. 1 . . invariant under transformations a(z). χβ . χβ .. . as we will demonstrate below such states are invariant under general Lorentz transformations a(z). Using (0. 1 ) stands for the Clebsch–Gordon coeﬃcient.66) is a bilinear form. . In the notation developed in Chapter 5 holds for the singlet state | . . .65) where | · · · 1 describes the spin state of “particle 1” and | · · · 2 describes the spin state of “particle 2” and (0.70) ‘Scalar’ implies invariance under rotations and is conventionally generalized to invariance under other symmetry trasnformations. In fact. as covariant spinors. . 1 ) = 2 2 2 2 2 2 2 2 √ ±1/ 2 and equating the spin states of “particle 1” with the spinor φα . ± 1 .362 Deﬁnition of a scalar product Spinor Formulation The question arises if for the states φα there exists a scalar product which is invariant under Lorentz transformations. The corresponding states are deﬁned through φ1 χ2 − φ2 χ1 = φ1 χ1 + φ2 χ2 It obviously holds χα = χ1 χ2 = χ2 −χ1 .

81) .79) = a −1 T a κβ . 11.72) in a notation analogous to that chosen for contravariant and covariant 4–vectors [cf. (10. −1 connecting contravariant and covariant spinors play the role of the metric tensors gµν .75) is given by φα = as can be readily veriﬁed.69. (11.70) and (11.69. 11.75) (11.11.73. 10.74)].74) = = 0 1 −1 0 0 −1 1 0 (11.77) (11.1: Show that for any non-singular complex 2 × 2–matrix M holds M −1 = det(M ) M −1 T The matrices . The inverse of (11.72) 0 −1 1 0 Exercise 11. (11.78) aβ γ γδ φδ (11.71) (11.3. Proof that φα χα is Lorentz invariant We want to verify now that the scalar product (11. we will express (11. 11.68). (11.72)] φα φ α = = = = αβ αβ φβ φβ 11 21 11 21 12 22 12 22 (11. 11. g µν of the Minkowski space [cf.70) is φ1 φ2 −1 363 = = −1 φ1 φ2 . (10.80) One can write in matrix notation aα β γ δκ αγ a δ αβ (11. In the transformed frame holds φ α χα = aα β αγ aγ δ δκ φβ χκ .67) will be expressed as φα χα = φ1 χ1 + φ2 χ2 = φ1 χ2 − φ2 χ1 .10.3: Spinors as can be veriﬁed from (11. For this scalar product holds φα χα = −χα φα .71. 11.76) αβ αβ The scalar product (11. Accordingly.73) (11.63. The transformation behaviour of φα according to (11.77) is Lorentz invariant.

˙ We also deﬁne covariant versions of φα φα = (φα )∗ . the ˙ conjugate complex spinors φα need to be considered separately from the spinors φα . i.83) Here we have employed the hermitian property of σ.e..80) φ α χα = φβ χβ .89) such that (11. The complex conjugate spinors We consider now the conjugate complex spinors (φ ) = α ∗ γ δκ αγ a δ = a −1 T a κβ = a−1 a κβ = δκβ (11. Obviously.87) which we will employ from now on. 2.63).14) one can write a −1 Spinor Formulation = e 2 z·σ 1 −1 = e 2 z· 1 σ −1 = e− 2 z·σ 1 ∗ (11. it holds φk = (φk )∗ .88) reads ˙ ˙ φ α = aα β φβ ˙ ˙ (11. from (11. hence.85) φ1 φ2 ∗ ∗ .364 Using (11. ˙ (11. σ † = (σ ∗ )T = σ. The transformation ˙ behaviour of φα is ˙ φ α = (aα β )∗ φβ ˙ (11. We denote ˙ (aα β )∗ = aα β ˙ (11.88) which one veriﬁes taking the conjugate complex of (11.83) into (11. k = 1.82) and with f (A)T = f (AT ) for polynomial f (A) a −1 T = e− 2 z·(σ 1 ∗ )T = e− 2 z·σ = a−1 1 (11.2) and (11. Insertion of (11. Hence.90) extending the summation convention to ‘dotted’ indices.91) .81) yields aα β and.86) A concise notation of the conjugate complex spinors is provided by ˙ (φα )∗ = φα = φ1 ˙ φ2 ˙ ˙ (11. As discussed above.84) (11. a∗ (z) provides a representation of the Lorentz group which is distinct from that provided by a(z). (11.

11.4 Spinor Tensors ˙ ˙ ˙ We generalize now our deﬁnition of spinors φα to tensors.18) ˜ 3 .92) (11.11.96) is Lorentz invariant. A tensor tα1 α2 ···αk β1 β2 ···β (11.95) ˙ where αβ and αβ are the real matrices deﬁned in (11. one can state ˜ and (11.73.18) of the bispinor wave function Ψ.75).99) Ψ3 (xµ ) χ ˙ (xµ ) ˜ χβ (xµ ) ˙ 1 ˜ χ2 (xµ ) Ψ4 (xµ ) ˙ 11. a∗ (z) −1 accounting for the ˜ transformation behaviour of the third and fourth spinor component of Ψ.100) . a property which is rather evident. 2 2 ˜ The transformation behaviour of Ψ (note that we do not include presently the space-time dependence of the wave function) ˜ a(z) Ψ1 ˜ Ψ2 ˜ ˜ ˜ Ψ = S(z) Ψ = (11. A comparision of (11. is governed by the operator a∗ (z) −1 which arises in ˜ the Lorentz transformation (11.76) φα ˙ ˙ φα αβ ˙ ˙ αβ ˙ ˙ = = = = αβ ˙ ˙ αβ ˙ ˙ αβ αβ φβ φβ ˙ ˙ (11. (11.75.4: Spinor Tensors 365 The relationship between contravariant and covariant conjugate complex spinors can be expressed in analogy to (11.94) (11. − 1 .97) The transformation. in matrix notation. (11. For the spinors φα and χα thus ˙ deﬁned holds that the scalar product ˙ φα χα = φ1 χ1 + φ2 χ2 ˙ ˙ ˙ ˙ ˙ (11. Ψ4 ..93) (11. The transformation behaviour of φα is ˙ φα = ˙ ˙ β γδ ˙˙ φδ ˙ ˙ αβ a γ ˙ ˙ .e. 11. + 1 2 2 2 1 transforms under rotations (z = iϑ) like a spin– 2 state | 1 .97) implies then that φα transforms like Ψ ˙ φ1 ˙ φ2 ˙ transforms under rotations (z = iϑ) like a spin– 1 state | 1 . i.98) ˜ ∗ (z) −1 Ψ3 a ˜ Ψ 4 obviously implies that the solution of the Dirac equation in the chiral representation can be written in spinor form ˜ 1 µ Ψ1 (xµ ) φ (x ) ˜ Ψ2 (xµ ) φ2 (xµ ) φα (xµ ) = = .

the elements of which are linear functions of Aµ .e. This tensor transforms according to t αβ = aα γ aβ δ tγ δ ˙ This reads in matrix notation.m aj akm t m (11.366 is a quantity which under Lorentz transformations behaves as t ˙ ˙ ˙ α1 α2 ···αk β1 β2 ···β k Spinor Formulation = m=1 aαm γm n=1 aβn δn tγ1 ···γk δ1 ···δ .. tjk = a t aT jk = . We had demonstrated that M (Aµ ) transforms according to M = M (Lµ ν Aν ) = a M (Aµ ) a† (11.107) which reads in spinor notation [cf. An example of a tensor is αβ and tions.104) Indices on tensors can also be lowered employing the formula tα β = and generalizations thereof. this transformation behaviour is in harmony with the tensor notation adopted. (11. we want to express Aµ through a spinor tensor. An obvious candidate is [cf. tjk = a t a† jk ˙ ˙ ˙ ˙ (11.105) This tensor is actually invariant under Lorentz transforma= αβ . k. . i.103) A αβ = aα γ aβ δ Aγ δ . with ˙ contravariant indices αβ. αβ = αβ (11.e.m aj a∗ t km m . i. (11.109) . (11. αβ ˙ αγ t ˙ γβ (11.25) the tensor is explicitly Aαβ = ˙ A1 1 A1 2 ˙ ˙ A2 1 A2 2 ˙ ˙ = A0 + A3 A1 − iA2 A1 + iA2 A0 − A3 .106) Exercise 11. 11.25] M (Aµ ) = σµ Aµ . The 4–vector Aµ in spinor form We want to provide now the spinor form of the 4-vector Aµ . According to (11.101) An example is the tensor tαβ which will play an important role in the spinor presentation of the Dirac equation.e.1: Prove equation (11. ˙ ˙ ˙ ˙ (11.108) Obviously.103) Similarly..102.106). m. it holds αβ . This task implies that we seek a tensor. i. the transformation bevaviour of a tensor tαβ reads in spinor and matrix notation t αβ = aαγ aβδ tγδ .4.102) = . ˙ ˙ ˙ ˙ (11.. using conventional matrix indices j. (11.

111) A11 A12 ˙ ˙ A21 A22 ˙ ˙ ˙ 0 −1 1 0 (11.113) Aα β = ˙ (11.2: Prove that (11.4.109. (11.114) yields ∂αβ = ˙ ∂0 + ∂3 ∂1 + i∂2 ∂1 − i∂2 ∂0 − ∂3 .112) A22 −A21 ˙ ˙ −A12 A1 1 . 11.110) The 4-vectors Aµ .4: Spinor Tensors One can express Aαβ also through Aµ Aα β = ˙ ˙ 367 A0 − A3 −A1 + iA2 −A1 − iA2 A0 + A3 .110) one can state ∂ αβ = Similarly. (11. ˙ 2 (11.115) is correct. Using (11. (11. (11. employing (11. ∂µ in spinor notation The relationship between 4–vectors Aµ . (11.115) Exercise 11.116) . (11. Hence.117) ˙ ˙ ∂0 − ∂3 −∂1 + i∂2 −∂1 − i∂2 ∂0 + ∂3 . Aµ can also be associated with tensors Aα β = ˙ This tensor reads in matrix notation A11 A12 ˙ ˙ A21 A22 ˙ ˙ = = 0 1 −1 0 ˙ ˙˙ αγ β δ A ˙ γδ .110) one obtains Aα β = ˙ A0 − A3 −A1 − iA2 −A1 + iA2 A0 + A3 A0 + A3 A1 + iA2 A1 − iA2 A0 − A3 . Aµ and tensors tαβ can be applied to the partial diﬀerential operator ∂µ .11.114) We ﬁnally note that the 4–vector scalar product Aµ Bµ reads in spinor notation Aµ B µ = 1 αβ ˙ A B αβ .

(11.. (11.f.243) in the chiral representation expressing S(Lη ξ ) by (11. one can conclude aσ µ ∗ −1 Lν µ σν 0 = .h. hence.e. To obtain the transformation properties of σ µ we employ then (10.123) (a∗ )−1 Lν µ a L ν µ = = σν (σ ) ν ∗ −1 (11. (11. . γµ = ˜ 0 µ )∗ (σ −1 0 σµ σµ 0 (σ µ )∗ −1 . 0 1 1 0 .e. Equation (10.125) is equivalent to µ ∗ −1 −1 a∗ (σ µ )∗ −1 −1 a Lν µ = (σ ν )∗ (11.121) One expects then that the transformation properties of σ µ should follow from the transformation properties established already for γ µ [c. (11.s. . Hence. −1 0 0 1 (11.124).124) constitutes the essential transformation property of σ µ and will be considered further. Comparision of (10. (11.18) and γ µ by (11. 0 −i i 0 . σµ .124).15) one can write σµ = and. holds also in the chiral representation. (11.243)].125) is equivalent to (11. hence.119) (11.121). . 1 0 0 −1 .243) reads then a 0 0 a∗ −1 0 (σ µ )∗ −1 σµ 0 a−1 0 0 a∗ −1 0 (σ ν )∗ −1 The l.118) 0 −1 −1 0 0 i −i 0 For this purpose we consider ﬁrst the transformation behaviour of σ µ and σµ .118) yields ˜ γµ = ˜ Using (11.120) σµ 0 .125) a (σ ) Equation (11.243) holds independently of the representation chosen. i. We will obtain the transformation behaviour of σ µ .229) and (11. of this equation is 0 ∗ (σ µ )∗ a and.126) which is the complex conjugate of (11. This is ˜ possible since γ µ can be expressed through σ µ . (10. σµ building on the known transformation behaviour of γ µ .122) aσ µ −1 a−1 −1 (a∗ )−1 0 Lν µ (11.26) and its contravariant analogue σ µ σµ = σµ = 1 0 0 1 1 0 0 1 . i..124) . Note that (10.368 σµ in Tensor Notation Spinor Formulation We want to develop now the tensor notation for σµ (11.

therefore. It holds according to (11. 0 −i i 0 µ=2 . = . (σ ∗ )T = σ yields using (11. 1 0 0 −1 µ=3 .124) using (11.4) with w = 0.2) −1 369 (a∗ )−1 = −1 − 1 z ∗ ·σ ∗ 2 e = e− 2 z 1 ∗ · −1 σ ∗ = e2z 1 ∗ ·σ . (11. 11.128) One can express. state that σ µ in a new reference frame is σ µ = a σ µ a† (11. −1 0 0 1 µ=3 . e.103) ˙ identiﬁes σ µ as a tensor of type tαβ . (11.2) −1 (a∗ )−1 = e2z 1 ∗ ·(σ ∗ )T = e2z 1 ∗ ·σ ∗ T = [a∗ ]T .4: Spinor Tensors One can rewrite (11. under rotations the Pauli matrices transform like (j = 1.e. according to (11. 11. 0 −1 −1 0 µ=1 . (11..132) 1 0 0 1 µ=0 . transformations (11.127) Exploiting the hermitian property of σ. (11.108) that the same transformation behaviour applies then for general Lorentz transformations.g. 11.130) We argue in analogy to the logic applied in going from (11. (11.133) Combining (11.131) where a is given by (11.129) We want to demonstrate now that the expression aσ µ [a∗ ]T is to be interpreted as the transform of σ µ under Lorentz transformations.. 0 i −i 0 µ=2 .131) one can express the transformation behaviour of σ µ in the succinct form Lν µ σ µ = σ ν . .11.118) (σ µ )11 (σ µ )12 ˙ ˙ (σ µ )21 (σ µ )22 and (σµ )11 (σµ )12 (σ µ )21 (σ µ )22 ˙ ˙ 1 0 0 1 µ=0 ˙ ˙ ˙ ˙ = (11. In fact.134) . i. equation (11.102. One can.4).129. (11.16. 0 1 1 0 µ=1 .2. 11. 3) σj −→ a(iϑ) σj a(iϑ) † = a(iϑ) σj a∗ (iϑ) T .2.124) a σ µ [a∗ ]T Lν µ = σ ν . 2. 11. hence.107) to (11. This transformation behaviour.

account for the 4-vector µ.370 Inverting contravariant and covariant indices one can also state Lν µ σµ = σν . the submatrices correspond to tαβ spinors. This yields 1 0 0 1 0 0 0 0 σ0 + σ3 σ1 + iσ2 (11. in fact. account for the 4–vector index µ. i.133) is characterized through a 4-vector ˙ index µ. i.110) to express σ αβ in terms 0 −1 1 0 0 0 0 1 (11.224) ˙ Using (11.139) g −f g and.29)]. δ.e. µ = 0. matrices. hence. = σαβ ˙ ˙ ˙ ˙ ˙ (σ21 )11 (σ21 )12 (σ22 )11 (σ22 )12 ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ 21 (σ )22 21 (σ )22 (σ21 ) (σ22 ) ˙ ˙ ˙ ˙ 21 22 Equating this with the r. corresponds to the transformation (σµ )11 (σµ )12 ˙ ˙ (σµ )21 (σµ )22 ˙ ˙ = 0 1 −1 0 (σµ )11 (σµ )12 ˙ ˙ (σµ )21 (σµ )22 and −1 .e. We can summarize that σ µ and σµ transform like a 4–vector. We want to express now σ µ and σµ also with respect to the 4–vector index µ in spinor form employing (11. however. γ. (11.138) the ‘inner’ covariant spinor indices. i.e.137) σαβ ˙ = δαγ δβ δ . 1. α.138) Note that all elements of σαβ are real and that there are only four non-vanishing elements. We will now consider the representation of σ µ .s. whereas ˙ account for the elements of the individual Pauli the ‘outer’ contravariant spinor indices. ˙˙ (11.. Each of the 4 × 4 = 16 matrix elements in (11..12) for of σµ yields together with (5. The ˙ desired change of representation(σµ )αβ −→ (σµ )αβ corresponds to a transformation of the basis of ˙ spin states f g f −→ = (11.135) This is the property surmised already above [cf. β. We can.h. σµ in which the contravariant indices are moved ‘inside’.141) σ αβ = σ0 − σ3 σ1 + iσ2 −σ1 − iσ2 σ0 + σ3 = 2 0 0 0 1 0 −1 0 0 . state ˙ ˙ ˙ ˙ (σ11 )11 (σ11 )12 (σ12 )11 (σ12 )12 ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ γδ (σ11 )21 (σ11 )22 (σ12 )21 (σ12 )22 ˙ ˙ ˙ ˙ . ˙ ˙ In (11.136) results in the succinct expression 1 2 ˙ γδ ˙ (11. ˙ ˙ 0 −1 1 0 ˙ (11.140) where we employed the expressions (11.. 2. and the covariant indices are moved outside. of (11. the transformation is inverse to that of ordinary 4–vectors. 3 as well as through two spinor indices αβ.136) σαβ = = 2 ˙ σ1 − iσ2 σ0 − σ3 0 0 0 0 1 0 0 1 where on the rhs. Spinor Formulation (11.114).132) and (11.

142) = This can be expressed 1 2 σ αβ ˙ ˙ γδ = δαγ δβ δ .147) .137) are in a basis (· · ·)αβ to a basis (· · ·)αβ results in ˙ σ αβ ˙ ˙ γδ = 2 (σ 11 )11 (σ 11 )12 ˙ ˙ ˙ ˙ (σ 11 )21 (σ 11 )22 ˙ ˙ (σ 21 )11 (σ 21 )12 ˙ ˙ ˙ ˙ 21 ) 21 ) (σ 21 (σ 22 ˙ ˙ 1 0 0 1 0 0 0 0 0 0 0 0 1 0 0 1 ˙ ˙ ˙ ˙ (σ 12 )11 (σ 12 )12 ˙ ˙ ˙ ˙ (σ 12 )21 (σ 12 )22 ˙ ˙ (σ 22 )11 (σ 22 )12 ˙ ˙ ˙ ˙ 22 ) 22 ) (σ 21 (σ 22 ˙ ˙ . (11.4: Spinor Tensors 371 and. ˙˙ (11.146) Exploiting the property (11. This expression implies that the element of γ µ given by σ µ −1 is in the expression (11. For this purpose we start from the ˜ µ . (11. Accordingly.145) Let Aµ be a covariant 4–vector. (11.138) one can conclude that the following property holds 1 2 ˙ γδ σ αβ ˙ = 1 2 σ αβ ˙ ˙ γδ = δαγ δβ δ .144) results in the simple relationship γ µ Aµ = ˜ 0 Aα β ˙ Aα β 0 ˙ .144) The Dirac Matrices γ µ in spinor notation We want to express now the Dirac matrices γ µ in spinor form. ˙ ˙ ˙ ˙ (11.143) Combined with (11. which in ˙ (11. we write ˜ ˙ γµ = ˜ 0 (σ µ )αβ µ ) )∗ ((σ αβ 0 ˙ ˙ .115) γ µ Aµ ˜ = 0 ((σ µ )αβ )∗ Aµ ˙ 0 ˙ ∗ 1 γδ ˙ ˙ 2 ((σ )αβ ) Aγ δ (σ µ )αβ Aµ 0 ˙ ˙ αβ γ δ 1 ˙ 2 (σγ δ ) A ˙ = 0 . ˙˙ (11.11.140) to transform each of the four submatrices.121) of γ ˜ ˜ ˙ basis |αβ whereas the element of γ µ given by σ µ is in the basis |αβ . One can write then the scalar product using (11. employing then transformation (11.

150) .149) (11.147) allows us to rewrite the Dirac equation in the chiral representation (10. (11.148) ˜ Employing Ψ(xµ ) in the form (11.226) ˜ i γ µ ∂µ Ψ(xµ ) = 0 i ∂ αβ ˙ i ∂ αβ 0 ˙ ˜ ˜ Ψ(xµ ) = m Ψ(xµ ) . α ˙ = = m φα m χβ . ˙ (11.372 Spinor Formulation 11.5 Lorentz Invariant Field Equations in Spinor Form Dirac Equation (11.99) yields the Dirac equation in spinor form i ∂ αβ χβ ˙ i ∂ αβ φ ˙ The simplicity of this equation is striking.

1 Symmetry and Degeneracies The degeneracies of energy levels of a quantum mechanical system are related to its symmetries.2) . The organization of this chapter is as follows: In the next section we will discuss the relation between symmetries of a quantum mechanical system and the degeneracies between its energy levels. In the following section we will introduce the concept of isospin as an approximate SU (2) symmetry.. we will discuss the SU (3) symmetry of three quark ﬂavors. the so-called Eightfold Way. i. which identiﬁes the proton and the neutron as diﬀerent states of the same particle. Sk ] = 0. O = exp k αk Sk . (12. In the ﬁnal section. We will particularly use the example of spherically symmetric potentials.Chapter 12 Symmetries in Physics: Isospin and the Eightfold Way by Melih Sener and Klaus Schulten Symmetries and their consequences are central to physics. The algebraic structure and the representations of SU (3) will be discussed in parallel to SU (2) and particle families will be identiﬁed in terms of representations of the underlying symmetry group. Sk . These are the isospin symmetry of nuclear interactions and its natural extension. 12. We will apply these concepts to an analysis of nucleon-nucleon and nucleon-meson scattering. 373 (12. Let us assume a continuous symmetry obeyed by a quantum mechanical system.1) Then the generators. The action of the symmetry operations on quantum mechanical states are given by elements of a corresponding Lie group. We will also introduce the quark model as a natural framework to represent the observed symmetries. [H. In this chapter we will discuss a particular set of symmetries that have played a seminal role in the development of elementary particle and nuclear physics.e. will commute with the Hamiltonian of the system.

Therefore. l and the corresponding energy levels are independent of m.3) If the newly obtained state is linearly independent of the original one. generates a state which has the same energy as the original one. In chapter 7 the dynamics of a particle moving in three dimensions under the inﬂuence of a spherically symmetric potential.18).. m r (12. φ). (12.λ1 . 2m r (12. We will investigate this shortly in more detail in the case of systems with spherical symmetry. · · · ..4) The stationary Schr¨dinger equation.. we ﬁrst consider a particular example of the implications of symmetry.39).5) with m = −l. (12. k = 1.. The energy levels are mk 2 En = − 2 2 . Classically. (7.m (r) Y m (θ. as studied in chapter 7.7) 2 n where the orbital angular momentum.8) . is allowed to take values in 0. Sk . (12. .. n − 1. (7. the additional symmetry generators of the Coulomb problem are the three components of the so-called eccentricity vector discovered by Hamilton = 1 r p×J −k . This degeneracy follows from the fact that any rotation. Lie groups play an essential role in the discussion of mass degeneracies in particle physics. .λ1 .. The energy levels are totally independent of l. .λn .. For example. the states 3s. We want to understand this extra degeneracy in terms of the extra symmetry of the hydrogen atom given by an additional set of symmetry generators introduced below. .8) ˆ [H. Jk ] = 0 . Presence of additional symmetries may further increase the degeneracy of the system. can then be reduced to a one-dimensional radial o equation.6) From elementary quantum mechanics we know the spectrum of the hydrogen atom.λ1 . 3 . has been discussed. The spherical symmetry implies the commutation of the Hamiltonian with angular momentum operators (7. V (r). Sk . 3p and 3d all have the same energy. . which yields a set of eigenstates of the form ψE. As an example of this we will consider the Coulomb problem with the Hamiltonian H = p2 k − . as represented by an element of SO(3). 2.374 Spinor Formulation The action of any symmetry generator.. Jk .λn = E exp(iαk Sk ) ψE. can be identiﬁed with the angular momentum operators. this implies a degeneracy in the spectrum.. namely motion in a spherically symmetric potential described by the group SO(3) (or its double covering SU (2) as discussed in section 5.12). . ψE. where the symmetry generators. each energy level is (2l + 1)-fold degenerate.24).. In order to illustrate this. . on an energy eigenstate. l.λn .m (r) = vE.. leaves the energy of the state invariant H exp(iαk Sk ) ψE. (7. (12..

16) Through some algebra [4] the following commutation relations can be veriﬁed [Ki . We will also need the following identity [4] 2 = 2H J2 + m 2 + k2 .14) This follows from a · (a × b) = (a × b) · a = 0. (12. which is valid even when a and b do not commute. (12. For this purpose we ﬁrst note that J · = 0. we will compute the hydrogen spectrum using the additional symmetry. 2m r (12. 2m r 1 x2 (J1 p3 − J3 p1 + i p2 ) − k .19) (12. 2m r 1 x3 (J2 p1 − J1 p2 + i p3 ) − k . 2E (12. ijk Jk .9) (12. Now we scale the eccentricity vector as follows K= − m . Therefore. 2 1 B = (J − K).13) It can be veriﬁed explicitly that its components commute with the Hamiltonian.10) (12. 2 which can be shown to satisfy [Ai .20) (12. The corresponding quantum mechanical hermitian operator can be deﬁned by 1 1 2 3 = = = = x1 1 (p2 J3 − p3 J2 + J3 p2 − J2 p3 ) − k . We introduce the following new operators 1 A = (J + K).11. (12. Jj ] = i [Ki . (12.15) which can be proved after very considerable algebra. which have a negative energy E. Kj ] = i ijk Kk .5: Lorentz Invariant Field Equations 375 The vector points along the symmetry axis of the elliptical orbit and its length equals the eccentricity of the orbit.21) .18) which complement the familiar angular momentum algebra of section 5. The vector in (12. in the subspace of the Hilbert space corresponding to a certain energy we can replace H by E.11) (12.8) is not a hermitian operator. 2m r 1 x1 (J3 p2 − J2 p3 + i p1 ) − k .12) (12.17) (12.3. In the following we consider the bound states. Aj ] = i ijk Ak . In order to understand the aforementioned extra degeneracy.

a lack of symmetry implies a lack of degeneracy in the energy levels of a quantum mechanical system. 1. In contrast to the discussion above about extra symmetries. 1. . .25) So far we have shown that the symmetry generators form an algebra.22): A2 = a(a + 1) B 2 = b(b + 1) Following (12. a = 0. . 2 (2a + 1)2 2 2 (12. a = 0. Spinor Formulation (12. Noticing that A2 and B 2 have the same eigenvalues because of (12. 2 1 . . By comparing to the rotation algebra introduced in chapter 5. (12. the energy eigenvalues are found to be E=− mk 2 1 . Using this equation the energy eigenvalues can be written in terms of the eigenvalues of A2 and B 2 operators. . This implies that a = b. A.26) (12. In order to arrive at the spectrum a ﬁnal bit algebra is needed A2 + B 2 = J 2 + K 2 m 2 = J2 − 2E mk 2 1 = − − 4E 2 (12. H B. . which is identical to the the direct sum of the Lie algebra of of two SO(3) (or SU (2)) algebras. can be seen to follow from the triangle inequality as applied to J = A + B. namely that J J > < A − B =0 A + B =2 A (12.22) (12. 2 (12. b = 0. . .24) (12.28). .376 [Bi .33) (12.23) (12. = 0.7) tells us that (2a + 1) = n. . l.21) and (12.31) where we have used (12. . a + b = n − 1}. .34) It follows that l has to have values in {0 = |a − b|.28) .30) 2 2 2 1 . .29) (12. .32) A comparison with (12. 1. . Furthermore the bound on the orbital angular momentum. . Bj ] = i [Ai . This illustrates the eﬀect of additional symmetries to the degeneracy structure of a quantum mechanical system. ijk Bk . H = 0.15). .27) (12. . . Bj ] = 0. Chaos is described classically as exponential . The most extreme case of this is the quantum analogue of a classically chaotic system.14) we note that A2 − B 2 = J · = 0. 1. we can read oﬀ the eigenvalues of A2 and B 2 from (12.

who. However. This is known as level repulsion.2 system p= 1 0 .28M eV /c2 . for the . for example. Following the mass-energy equivalence of special relativity E = mc2 . If the chosen boundary is ‘irregular’ in a suitably deﬁned sense. (Weak interactions are responsible. A typical example of this so called quantum chaos is the quantum billiard problem. the isospin symmetry is also governed by an SU (2) group ‘rotating’ components in (12. (12. In the case of billiards and other examples of quantum chaos one common observation is the almost nonexistence of degeneracies and the fact that the energy levels are more evenly spaced.57M eV /c2 . It is important to place the isospin concept in its proper historical context. it is natural to represent them in terms of a two component vector. This (approximate) degeneracy led into the idea of the existence of an (approximate) symmetry obeyed by the underlying nuclear interactions. which is a particle in box problem in two dimensions with a boundary which can be chosen arbitrarily.36) into each other in abstract isospin space. which was discovered by observing degeneracies in the particle spectrum. suggested that the proton and the neutron can be regarded as two states of a single particle. after the discovery of the neutron in 1932. 12.2 Isospin and the SU (2) ﬂavor symmetry The concept of isospin goes back to Heisenberg.35) this mass equivalence can be viewed as an energy degeneracy of the underlying interactions. This was motivated by the observation that their masses are approximately equal: mp = 938.36) In analogy to the concept of spin regarding the rotations in 3-space as discussed in chapter 5. In the next section we will proceed with the discussion of a symmetry. (Strong interactions bind the atomic nucleus together. This enables us to utilize what we already know about the SU (2) symmetry group from the study of angular momentum.11. (12.5: Lorentz Invariant Field Equations 377 sensitivity to initial conditions. For a more detailed discussion of quantum chaos in billiard systems we refer the reader to [7] and the references therein. in the sense that nearby trajectories in the phase space diverge from each other over time. the classical trajectories will diverge from each other after successive bounces from the boundary. another manifestation of chaos is the lack of independent operators commuting with the Hamiltonian. n= 0 1 . mp = 939. we will be able to use the familiar Clebsch-Gordan coeﬃcients to combine the isospin of two particles the same way we added spin in chapter 6. namely. analogous to the 1 spin-up and spin-down states of a spin. that the proton and the neutron behave identically under the so-called strong interactions and that their diﬀerence is solely in their charge content. For example. Originally it was believed that isospin was an exact symmetry of strong interactions and that it was violated by electromagnetic and weak interactions.) If the proton and the neutron are to be viewed as two linearly independent states of the same particle.

as good quantum numbers. π − = |1. (12.− 2 2 . respectively. We can describe isospin multiplets the same way we have described the angular momentum and spin multiplets. including the proton and the neutron.− 2 2 . 2 2 1 2 2 2 (12. it can be seen as part of a larger (and more approximate) symmetry which is of great utility to classify observed particle families. −1 = = = 1 1 1 1 .37) As an example of a multiplet with I = 1 we have the three pions or π-mesons π + = |1.− 2 2 1 2 2 2 1 1 1 1 . are made up of these two quarks.11.− . They can be constructed with the same rules that have been used for angular momentum addition ¯ u in chapter 6. n = I = . (12. albeit it is a good approximate one.36) as a multiplet with I = 2 1 1 p = I = . 2 (12. . π 0 = |1.6M eV /c2 . In the framework of the quark of model. 2 2 1 2 2 2 1 1 1 1 1 √ .42) (12. I3 = − 2 2 . sec. I3 = 2 2 1 1 . the fundamental representation of the isospin symmetry corresponds to the doublet that contains the so-called up and down quarks u= 1 1 . Further than that.39) All other isomultiplets. For example. 0 π − = |1. the proton and the neutron can be written as totally symmetric uud and udd states. 2 2 1 1 1 . If the mass diﬀerence (or the energy diﬀerence) were to be to be purely electrostatic in nature. spelling disaster for the stability of matter. (mπ± = 139. I. Denoting the total isospin. .38) are ud. 1 we can re-write (12. 2. However.− . If it were otherwise the proton would be unstable by decaying into the neutron. Isospin symmetry is not an exact symmetry of strong interactions. d= 1 1 . .378 Spinor Formulation beta decay).40) + 2 1 1 . 2 2 . For a precise description of the two nucleons as composite states. as we shall see below. mπ0 = 135.0M eV /c2 ). the three pions in (12.41) (12. 0 . Therefore it remains a useful concept. 1 π 0 = |1. I3 . u¯ and d¯ states. The mass diﬀerence between the neutron and the proton could then be attributed to the charge content of the latter. including the spin and color quantum numbers of their constituent quarks. we refer the reader to [1]. −1 .38) which have all nearly identical masses. They u form an isotriplet: π + = |1. 1 . . (12. and its third component. Shortly we 1 will see how to describe both the pion and nucleon states as composites of more fundamental I = 2 states. the proton had to be heavier. the proton is the lighter of the two.43) Similarly.

but at the time the empirical concepts like isospin and strangeness quantum numbers were in use. strongly interacting particles) which is about a GeV /c2 . The value of the strangeness quantum number is taken. By convention ¯ baryons have baryon number 1. baryons generally are heavy. as some mesons discovered later are heavier than some baryons and so on.47) (12. a hydrogen isotope. baryon and meson.11. The possibilities are that of an isosinglet 1 |0.44) 2 where B is the baryon number and S is the strangeness. 0 ) we should have seen nn and pp bound states of comparable energy in nature (because of isospin symmetry). −1 = |p > |p >. Following (12. (12. I3 = 0. we shall setup some terminology: baryons are qqq states. . Before proceeding further. 0 |1. but such states do not exist. The relation between electric charge and isospin are given by the Gell-Mann–Nishijima relation which was ﬁrst derived empirically 1 Q = I3 + (B + S). (12. All other composite states have their strangeness given by the sum of the strangeness content of their constituents. If taken literally. 0 = √ (|p > |n > −|n > |p >) 2 and that of an isotriplet |1. where leptons (electron.) are light. Naturally. whereas mesons are q q states. Now let us consider another example of combining the isospins of two particles. originally refer to the relative weight of particles. mesons have baryon number 0. As it later turned out. 2 = |n > |n > . Therefore the deuteron has to be an isosinglet state (|0.e. 1 |1.5: Lorentz Invariant Field Equations 379 The mass of the up and down quarks are not identical but they are both of the order of a few M eV /c2 ’s which is minuscule compared to the typical energy scale of hadrons (i. In the late 1940’s and early 1950’s. muon.45) Is the deuteron an isosinglet state or an isotriplet? If it were an isotriplet (|1. We will now try to describe its wave function in terms of its constituent nucleons. the up and down quarks are not the only quark ‘species’ . 0 ). In the next section we will be able to view the Gell-Mann–Nishijima relation in the light of the representation theory for the ﬂavor SU (3) symmetry.46) (12. mesons have intermediate mass ranges. the pions being examples thereof. This is why isospin is such a good symmetry and why isomultiplets have nearly identical masses. this remains only an inaccurate naming convention today. 1 = √ (|p > |n > +|n > |p >). The names.43). Therefore it has to have isospin. the neutrinos etc. consists of a proton and a neutron.37) and in analogy to (12. this will be mathematically identical to adding two spins. by accidental convention. It shall be noted here that the quark model was not invented until 1960’s. such as proton and the neutron. a number strange particles have been found which presumably contained a third quark species: the strange quark.48) (12. All antiparticles have their 3 quantum numbers reversed. and quarks have baryon number 1 .or ﬂavors as they are commonly called. to be −1 for the strange quark. The up and down quarks have strangeness zero. The reader may know that the deuteron.

I3 .38) and the fact that the deuteron is an isosinglet we know that the isospins of the ﬁnal states in (I) and (II) are |1. √ According to (12. Exercise.55) . Show that I(1) ·I(2) is an ‘isotensor’ of rank zero. 0 . σ.53) 0 √ 1 2I (i) +1 I (f ) . I3 (f ) = 1. We will eventually be able to compute ratios of scattering cross-sections between diﬀerent processes.7 to the case of isospin. I3 = 1. 0 + |0. Now let us re-write more carefully the scattering amplitudes for the two processes in the light of what we have just learned MI = I (f ) = 1. Consider the generalization of tensor operators discussed in section 6. The dot product here refers to the abstract isospin space. Exercise. γ (i) is a reduced matrix element deﬁned in the same sense as in section 6.49) The only assumption that we put in will be that the interaction is of the form V = α I(1) · I(2) . The initial and ﬁnal states can be denoted in more detail as | initial | ﬁnal = = I (i) . γ (f ) |T00 |I (i) I3 . where M is the scattering amplitude given by M = ﬁnal | α I(1) · I(2) | initial . (12.48) the initial states in (I) and (II) have isospin values |1. γ (i) I (f ) .8 to compute the ratio of the scattering amplitudes MI and MII . γ (f ) ||T0 ||I (i) . I3 . γ (f ) T00 I (i) = 1.380 Spinor Formulation As an exercise on the implications of isospin symmetry we will consider nucleon-nucleon scattering. 1 and |1. Refer to exercise 6. 0 ). (f ) (i) (I (f ) I3 |00I (i) I3 ) is a Clebsch-Gordon coeﬃcient and I (f ) . Show that the expectation of I(1) · I(2) is state.51) (12. such as the spatial dependence of the wave function and spin.7 and 6. . γ (f ) (f ) (i) (12.7. 1 and (1/ 2)(|1. γ (i) = (f ) (i) (I (f ) I3 |00I (i) I3 ) (f ) (i) (−1)I −I (f ) (i) (12. For example. The cross-section.8.54) Here T00 ≡ V = α I(1) · I(2) .52) where γ (i) and γ (f ) denote degrees of freedom other than isospin. 1 4 in an isotriplet state and − 3 in an isosinglet 4 Using (12. respectively. We will now employ the (isospin analogue) of the Wigner-Eckart theorem (6.5 for the spin-analogue of the same problem.45) and (12.259) discussed in detail in sections 6. γ (i) (i) (12.50) . which is an isoscalar. is proportional to |M|2 . γ (f ) ||T ||I (i) .259) in the present context: I (f ) I3 . consider (I) p + p → d + π + (II) p + n → d + π 0 (12. as discussed in the exercise above. γ (i) . (12. For completeness let us start by restating the Wigner-Eckart theorem (6.

σII which is in approximate agreement with the observed ratio. particles with strangeness. σ( π − + p → anything ) The possibilities are (a) π + + p → π + + p . 0 1 1 3 3 2. γ (i) 3 +0. γ (f ) ||T00 || I (i) = 1. for example.63) where common dynamical factors (which would not be as easy to compute) have dropped out thanks to the Wigner-Eckart theorem. (12. −2 . (12. I3 = 0. 2 .5: Lorentz Invariant Field Equations 1 (11|0011) √ I (f ) = 1.65) (12.11. = MII (1/ 2) (12. which are obtained by a standard Clebsch-Gordan expansion π + + p : |1. which is also manifested by a vanishing Clebsch-Gordon prefactor. (c) π − + p → π 0 + n .64) There are more exotic possibilities. γ (f ) T00 I (i) = 1.62) We can now write the ratio of the scattering amplitudes: MI 1 √ . 2 = 3 2. I3 = 0. γ (i) = √ 2 1 (f ) (i) +√ I (f ) = 1. γ (f ) T00 I (i) = 0. −2 1 1 2. 1 π − + p : |1. involving. We want to compute the ratio of total cross-sections assuming a similar interaction as in the previous example σ( π + + p → anything ) . = 381 (12. γ (f ) ||T00 || I (i) = 1.56) (12.57) (12. 2 = 2. γ (i) 3 1 (f ) (i) I (f ) = 1. I3 = 0.61) MII Note that the second term in MII vanishes due to the isospin conservation. .58) (12. but these are not dominant at relatively low energies. 1 1 1 1 3 √ 2. I3 = 0. γ (i) 2 1 = (10|0010) √ I (f ) = 1. (12.60) (12. The relevant Clebsch-Gordon coeﬃcients are easily evaluated: (11|0011) = (10|0010) = 1.59) (12.66) = 2 3 1 3 2. [2] As a further example. we will consider pion-nucleon scattering. (b) π − + p → π − + p . −2 1 1 √ . −1 π0 + n : |1. −2 2 1 1 3 2. −1 2 3 2 − + . Once again we need the isospins for the initial and ﬁnal states. It follows σI = 2.

S. it appears an obvious generalization to add another component for an extra quark to the isospin vector space 1 0 0 (12. m V V 3 2.71) u = 0 . such that S(K + ) = 1. Hence Gell-Mann and independently Nishijima postulated the existence of a separate quantum number. 0 0 1 . S(Σ− ) = −1 and S(π) = S(N ) = 0. m 1 2. For instance. for example.70) (12.68) − 2 3 M1 2 Guided by empirical data we will further assume that M 3 >> M 1 . As the total cross-section is the sum of individual processes we obtain σ( π + + p ) σa = =3 σ( π − + p ) σb + σc again in approximate agreement with the observed value. while weak interactions did not. K + . which can be produced by π − + p → K + + Σ− . however. m . Hence the strangeness changing strong decays of K + (or Σ− ) were forbidden. giving it a higher than usual lifetime.65) Ma = M 3 2 Mb = 1 M 3 + 2 M 1 3 3 √ 2 Mc = 2 3 M3 2 √ 2 (12.3 The Eightfold Way and the ﬂavor SU (3) symmetry The discovery of the concept of strangeness. etc. are up to 40% heavier. m 1 2. [2] (12. A computation similar to the previous example of nucleon-nucleon scattering yields (apart from common prefactors) the following amplitudes for the reactions in (12.382 Spinor Formulation As in the example of nucleon-meson scattering we deﬁne the relevant matrix elements M3 = 2 M1 = 2 3 2. called strangeness. In the light of the quark model. which leads to the following 2 2 ratios for the cross-sections σa : σb : σc = 9 : 1 : 2. Strange partners of the familiar nucleons. (12.69) 12.s = 0 . has a lifetime which is comparable to that of π + albeit being more than three times heavier. was motivated by the existence of particles that are produced strongly but decay only weakly. making an identiﬁcation of the underlying symmetry and the multiplet structure less straightforward. mentioned in the previous section.d = 1 . It was assumed that strong interactions conserved S (at least approximately). The classiﬁcation of the newly found particles as members of some higher multiplet structure was less obvious then the case of isospin.67) which are independent of m. .

It is seen that such a matrix U has to satisfy the following k U † = U −1 . The unitarity relation imposes 9 constraints on the total of 18 real degrees of freedom of a 3 × 3 complex matrix. The bottom and top are even heavier. Given a complex vector. it is the group . as in (12. called quantum chromodynamics. we want to ﬁnd those transformations ak → Ukl al that preserve the norm. Hence the group U (3) has 9 dimensions. Therefore U (3) can be decomposed into a direct product U (1) × SU (3) where SU (3) consists of 3 × 3 unitary matrices of unit determinant. As quarks were never directly observed. Flavor SU (Nf ) symmetry on the other hand.3 for a brief discussion of gauge transformations). of three dimensions. In many respects it will resemble the more familiar group SU (2) discussed in some detail in chapter 5. Lie algebras and related concepts. we will establish some mathematical preliminaries about the group SU (3). called color. SU (N ).1 whenever necessary.5: Lorentz Invariant Field Equations 383 In this case the transformations that ‘rotate’ the components of (12. unitary relation k (12. Then came the question of what the fundamental representation. Because of this additional constraint SU (3) has 8 dimensions. The reader shall note that most of what is being said trivially generalizes to other unitary groups. (12. This is believed to be an exact symmetry of strong interactions. bottom (beauty) and top (truth) are signiﬁcantly heavier than the hadronic energy scale.72) a∗ ak . eiφ . in reference to Lie groups. However SU (3) symmetry appears in another and much more fundamental context in strong interaction physics. The reason is that the other known quarks. have to be elements of the group SU (3) (which we will investigate closely very soon). This group of 3 × 3 unitary matrices is denoted by U (3). but we will stick to N = 3 in the following.71).74) for any two unitary matrices U and V . still leaves the norm invariant. (The reader is referred to section 8. history followed the reverse of this path. while preserving the norm. giving rise to the quark model. where Nf is the number quark ﬂavors.73) To verify that all such matrices form a group. namely charm. which set the hadronic energy scale. in fact modern theory of strong interactions is a ‘gauge theory’ of this color group. we observe that (U V )† = V † U † = V −1 U −1 = (U V )−1 . Multiplying U by a phase.71) into each other. becomes increasingly inaccurate for Nf > 3. but there are a number of subtle diﬀerences.11. In this perspective the ﬂavor SU (3) symmetry may appear to be mainly of historical interest. ak . should correspond to. of a. First particle multiplets were identiﬁed as representations of the SU (3) group. (12. for a period they were considered as useful bookkeeping devices without physical content. (The ‘bare’ mass of the charm quark is already heavier than the two nucleons. Since arbitrary phase factors are of no physical interest. The quarks posses another quantum number. However. which again form representations of an SU (3) group. The reader is also invited to revisit section 5. [2] ) Before discussing the signiﬁcance and the physical implications of the quark model. the same way nucleons and pions were identiﬁed as representations of the isospin SU (2) symmetry.

regardless of the dimension of the representation. which can be veriﬁed explicitly as matrix identities. The Lie algebra structure is given by the commutators of λk [λj .80) This is the fundamental or deﬁning representation of SU (3). (12. An explicit basis is constructed in analogy to the Pauli algebra of spin operators 1 0 = 0 0 0 = 0 0 0 = 1 1 1 = √3 0 0 0 1 0 0 0 1 0 0 0 0 0 0 1 0 0 0 1 0 0 1 0 0 i 0 0 0 i 0 0 0 −i 0 0 0 0 0 0 −i 0 0 0 0 0 0 0 −i i 0 1 0 0 . As discussed in section 5. since det(eA ) = etrA . Therefore we will express elements of SU (3) as U = ei k (12.384 Spinor Formulation SU (3) and not U (3) that is of main interest.32). λ2 = . −2 λ4 λ6 λ8 The generators. (We shall at times refer to the Lie algebra with the name of the group. λk . λ3 = 0 −1 0 . the meaning being apparent from the context. λ5 = . 0 0 0 .79) (12. via the anticommutator relation 4 [λj .75) αk λk (12.77) λ1 .7. obey the following relation tr(λj λk ) = 2δjk . λk ]+ = δjk + 2δjkl λl .) The unit determinant condition requires that all λk are traceless.1.79). . λk ] = 2ifjkl λl .76) where λk are 8 linearly independent matrices forming the basis of the Lie algebra of SU (3). .78) where fjkl are the antisymmetric structure constants similar to the jkl of SU (2) given in (5. (12. The fundamental relation to be preserved is (12. U . any unitary matrix. λ7 = 0 0 . We can also introduce the constants. The reader is invited to compare the structure of SU (3) to that of SU (2) discussed in section 5. As in the case of SU (2) higher dimensional representations obeying the same structure can be found. can be written in the form U = eiH where H is a hermitian matrix. δjkl . 3 (12.

94) .82) With another change of basis we arrive at the ‘standard’ form of the generators of the Lie algebra of SU (3) T± = F1 ± iF2 .91) [T0 . I3 . V± ] = ± V± . T± ] = ±T± . We ﬁrst introduce the F-spin operators 1 Fi = λ i .84) (12. (Recall that the strange quark has S = −1 by convention).44). (12.88) which deﬁnes (not uniquely) a maximal set of mutually commuting operators {Y.93) (12.89) (12. 3 (12. In the basis (12.92) (12.90) (12. J− } proved useful. Exercise. T 0 = F3 . (12. [Y. Derive the Gell-Mann–Nishijima relation (12. T± ] = 0. U± ] = U± . 2 Y = √ F8 . The generators of SU (3) can be arranged into a very similar form to that of SU (2). [Y. U± = F6 ± iF7 .11. J± ] = ± J± .87) the commutation relations between the generators can be expressed in a succinct manner.85) (12. Using the convention in (12.77) is not the most useful basis in the study of SU (3). starting with the observation that Y = B + S.87) Exercise. First. U± ] = ±U± . 2 (12. (12. namely spherical harmonics. we have [Y.71) show that T0 is the isospin operator. 2 1 [T0 . these relations have been used to construct the angular momentum spectrum as well as the function space representation of the rotation algebra.81) In chapter 5. which satisﬁed an ‘eigenvalue equation’ [J0 . T0 } and [Y. Y is called the hypercharge. 2 (12. T0 ] = 0.83) (12.5: Lorentz Invariant Field Equations 385 As it turns out the set of generators in (12.86) (12. In the case of SU (2) the identiﬁcation of ‘ladder’ operators {J+ . V± ] = ±V± . 1 [T0 . V± = F4 ± iF5 .

98) (12.104) (12.88).105) C2 = jkl djkl λj λk λl In general the number of independent Casimir operators of a Lie group is equal to its rank.100) (12. Another important property of SU (2) is the existence of an operator (namely the total angular momentum. 2 3 Y + T0 = 2V0 . 3 [U+ . The maximum number of mutually commuting generators of a Lie algebra is called its rank. V+ ] = 0. Finally. In the case of SU (2) the representations lay on a line on the J0 axis. V− ] = −U− . [T+ .5. U+ ] = V+ . [V+ .102) (12. Note that. Thus. We can construct two such independent Casimir operators for the group SU (3). while SU (3) has rank 2. An operator which commutes with all generators of a Lie group is called a Casimir operator.95) (12. T0 }. When the basis of a Lie algebra is expressed in such a way to satisfy the form of the eigenvalue relations as given above.103) Any remaining commutators follow from hermiticity. U± } to this maximal set by ‘eigenvalue equations’ and [T+ . as the relation between the states in a given representation can be conveniently expressed in terms of ladder operators. [U+ . since there are two mutually commuting generators in SU (3) as given in (12.101) (12. jkl (12. However. This form is essential for easy labeling of the representations of the group. T− ] = 2T0 . the representations will now lie in a T0 − Y -plane. C1 = k λ2 = − k 2i 3 fjkl λj λk λl . Casimir operators can be used to label irreducible representations of the Lie algebra. (12. V− ] = 2 (12.99) (12. The interested reader is instead referred to a very readable account given in chapter 12 of [4]. A formal deﬁnition and a detailed discussion of the Cartan-Weyl form is beyond the scope of this chapter.97) which relate commutators of generators with opposite eigenvalues to the maximal set {Y.386 Spinor Formulation which relate the remaining generators {T± . [T+ . [T+ . we have [T+ . The same way the angular momentum ladder operators have been used to construct the representations of SU (2). V± . similar to the way it was done in section 5. V+ ] = 0. [U+ . V− ] = T− .96) (12. As in the case of J 2 and SU (2). . U− ] = Y − T0 = 2U0 . U0 and V0 are linear combinations of T0 and Y . it is said to be in Cartan-Weyl form. J 2 ) which commutes with all of the generators. we will use these commutation relations to construct representations of SU (3). SU (2) has rank 1. U− ] = 0.

2 (12. y U± |t3 .2) shows the representation D(2.112) The eﬀect of these operators to the states in the y − t3 plane have been outlined in Fig. Now we will construct explicit representations of SU (3). y .) All other representations can be constructed by combining these two conjugate representations. y = t3 |t3 . 1) as an example. (12. For example the pion octet (or any other meson octet) is therefore realized as states of a quark .antiquark pair. (See Fig. we have T± |t3 . The representations for hexagons with sides of length p and q in the T0 − Y -plane. The details of this procedure is beyond the scope of this chapter. V± . 2 1 = γ t3 .11. Such a representation is labeled as 1 D(p. y V± |t3 . The representations of SU (3) are constructed analogous to those of SU (2) by identifying the ‘boundary’ states annihilated by raising (or lowering) operators.109) The same way that J± |m is proportional to |m ± 1 in the case of the angular momentum algebra. respectively.1). y .107) From the commutation relations we have presented above (the Cartan-Weyl form) we can write down the eﬀect of various generators on the state |t3 .110) (12. The interested reader is referred to [4]. 2 1 = β t3 ± .3).5 how to label the irreducible representations of the angular momentum algebra SU (2) in terms of the value of the total angular momentum. y .111) (12. Figure (12. y . |t3 . All other states of the representation are then constructed by successive application of ladder operators T± .106) (12.y ± 1 . y V0 |t3 . y Y |t3 . For example. (12. = y |t3 . 1) representation. y . y : T0 |t3 . This is because the states in a given representation are connected by the action of the generators of the Lie algebra and such generators commute with the Casimir operators. The representations D(1. 2 (12. y 1 = α t3 ± . This property can be used to label representations in terms of the values of the Casimir operators. U± . 1) correspond to the triplets of quarks and antiquarks. For example. 2 1 t3 ) |t3 . (12. especially chapters 7 and 8. y . y 3 = ( y− 4 3 = ( y+ 4 1 t3 ) |t3 . Because of (12.5: Lorentz Invariant Field Equations 387 The utility of Casimir operators arises from the fact that all states in a given representation assume the same value for a Casimir operator. y ± 1 . . A notation suggestive of the dimensionality of the representation can be used to identify representations. the pion family forms part of an octet corresponding to the D(1. we have U0 |t3 . As another example for the representations of SU (3). it was shown in section 5.108) (12. q) and it has a dimensionality of 2 (p + 1)(q + 1)(p + q + 2).88) we can label states by the eigenvalues of T0 and Y operators. 0) and D(0. For example.

1) representation of SU (3). The states in the inner triangle are doubly degenerate. £ Figure 12.388 £ Spinor Formulation Figure 12. ¡ ¢ ¡ ¢ 8 6 2 '745301) ¥¦¤ ¤ #" ! (& % '$ © ¨ ¦ §¥ ©§ ¨ .2: D(2.1: The eﬀect of SU (3) ladder operators on the y − t3 plane.

2 3 1 1 − . .11. in the case of SU (2).5: Lorentz Invariant Field Equations 389 D(1. £ (12. 1) is written as [8] etc. 1) are denoted by [3] and [¯ respectively. This expansion can be compared. 3]. 2 3 2 0. where we would write [3] ⊗ [3] = [1] ⊕ [3] ⊕ [5]. for example.118) ¡ ¢ ¨ ©§ ¨¥¦¤ " ! . The three quarks in the fundamental representation can now be written as u = d = s = 1 1 .3: D(1.114) Figure 12.113) The additional singlet state corresponds to the η meson.115) (12. 2 (12. to the case of adding two spin triplet states. The octet D(1.116) (12. . 0) and D(0. 0) or the fundamental representation of ﬂavor SU (3) symmetry. This way the quark-antiquark states can be represented as follows [3] ⊗ [¯ = [8] ⊕ [1] 3] (12. − .117) The Gell-Mann–Nishijima relation can then be succinctly expressed as 1 Q = y + t3 . 3 (12.

Griﬃths. Wiley. W. 3 1 Qd = − . 3 1 Qs = − .120) (12. and spectroscopy. Kosztin. (1999). 1993. D. Wiley. I. Am. Eisenberg. Kaufman. Phys. 1984. p. Schulten. 1972. Greiner. 1) representation of SU (3). D. References [1] Quarks and leptons: an introductory course in particle physics. [4] Quantum mechanics: symmetries. A. J. 1987.4: The baryon octet as a D(1. 1989. [7] Expansion method for stationary states of quantum billiards. L. Springer-Verlag u [5] Principles of symmetry. M. Halzen. Harter. Martin. M¨ller. Oxford. W. Li. 67 (2). L. ¡ ¢ S - S 0 ¤ ¥ S + . dynamics. F. [3] Microscopic theory of the nucleus. Greiner. J. T. 133. [6] Gauge theory of elementary particle physics. G. Cheng. 1984.-P. D. B.390 from which the quark charges follow 2 Qu = . W.121) L 0 X - X 0 Figure 12. [2] Introduction to elementary particles. North holland. 3 £ Spinor Formulation (12.119) (12. K.-F. Wiley.

- Read and print without ads
- Download to keep your version
- Edit, email or read offline

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

CANCEL

OK

You've been reading!

NO, THANKS

OK

scribd

/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->